diff --git a/CHANGELOG.md b/CHANGELOG.md index be30f774c..9b4317f71 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -1,276 +1,24 @@ -# v2.14.4 -- [#992](https://github.com/xmrig/xmrig/pull/992) Fixed compilation with Clang 3.5. -- [#1012](https://github.com/xmrig/xmrig/pull/1012) Fixed compilation with Clang 9.0. -- In HTTP API for unknown hashrate now used `null` instead of `0.0`. -- Fixed MSVC 2019 version detection. -- Removed obsolete automatic variants. +# v3.0.0 +- **[#1111](https://github.com/xmrig/xmrig/pull/1111) Added RandomX (`rx/test`) algorithm for testing and benchmarking.** +- **[#1036](https://github.com/xmrig/xmrig/pull/1036) Added RandomWOW (`rx/wow`) algorithm for [Wownero](http://wownero.org/).** +- **[#1050](https://github.com/xmrig/xmrig/pull/1050) Added RandomXL (`rx/loki`) algorithm for [Loki](https://loki.network/).** +- **[#1077](https://github.com/xmrig/xmrig/issues/1077) Added NUMA support via hwloc**. +- **Added flexible [multi algorithm](doc/CPU.md) configuration.** +- **Added unlimited switching between incompatible algorithms, all mining options can be changed in runtime.** +- [#257](https://github.com/xmrig/xmrig-nvidia/pull/257) New logging subsystem, file and syslog now always without colors. +- [#314](https://github.com/xmrig/xmrig-proxy/issues/314) Added donate over proxy feature. +- [#1007](https://github.com/xmrig/xmrig/issues/1007) Old HTTP API backend based on libmicrohttpd, replaced to custom HTTP server (libuv + http_parser). +- [#1010](https://github.com/xmrig/xmrig/pull/1010#issuecomment-482632107) Added daemon support (solo mining). +- [#1066](https://github.com/xmrig/xmrig/issues/1066#issuecomment-518080529) Added error message if pool not ready for RandomX. +- [#1105](https://github.com/xmrig/xmrig/issues/1105) Improved auto configuration for `cn-pico` algorithm. +- Added command line option `--export-topology` for export hwloc topology to a XML file. +- Breaked backward compatibility with previous configs and command line, `variant` option replaced to `algo`, global option `algo` removed, all CPU related settings moved to `cpu` object. +- Options `av`, `safe` and `max-cpu-usage` removed. +- Algorithm `cn/msr` renamed to `cn/fast`. +- Algorithm `cn/xtl` removed. +- API endpoint `GET /1/threads` replaced to `GET /2/backends`. +- Added global uptime and extended connection information in API. +- API now return current algorithm. -# v2.14.1 -* [#975](https://github.com/xmrig/xmrig/issues/975) Fixed crash on Linux if double thread mode used. - -# v2.14.0 -- **[#969](https://github.com/xmrig/xmrig/pull/969) Added new algorithm `cryptonight/rwz`, short alias `cn/rwz` (also known as CryptoNight ReverseWaltz), for upcoming [Graft](https://www.graft.network/) fork.** -- **[#931](https://github.com/xmrig/xmrig/issues/931) Added new algorithm `cryptonight/zls`, short alias `cn/zls` for [Zelerius Network](https://zelerius.org) fork.** -- **[#940](https://github.com/xmrig/xmrig/issues/940) Added new algorithm `cryptonight/double`, short alias `cn/double` (also known as CryptoNight HeavyX), for [X-CASH](https://x-cash.org/).** -- [#951](https://github.com/xmrig/xmrig/issues/951#issuecomment-469581529) Fixed crash if AVX was disabled on OS level. -- [#952](https://github.com/xmrig/xmrig/issues/952) Fixed compile error on some Linux. -- [#957](https://github.com/xmrig/xmrig/issues/957#issuecomment-468890667) Added support for embedded config. -- [#958](https://github.com/xmrig/xmrig/pull/958) Fixed incorrect user agent on ARM platforms. -- [#968](https://github.com/xmrig/xmrig/pull/968) Optimized `cn/r` algorithm performance. - -# v2.13.1 -- [#946](https://github.com/xmrig/xmrig/pull/946) Optimized software AES implementations for CPUs without hardware AES support. `cn/r`, `cn/wow` up to 2.6 times faster, 4-9% improvements for other algorithms. - -# v2.13.0 -- **[#938](https://github.com/xmrig/xmrig/issues/938) Added support for new algorithm `cryptonight/r`, short alias `cn/r` (also known as CryptoNightR or CryptoNight variant 4), for upcoming [Monero](https://www.getmonero.org/) fork on March 9, thanks [@SChernykh](https://github.com/SChernykh).** -- [#939](https://github.com/xmrig/xmrig/issues/939) Added support for dynamic (runtime) pools reload. -- [#932](https://github.com/xmrig/xmrig/issues/932) Fixed `cn-pico` hashrate drop, regression since v2.11.0. - -# v2.12.0 -- [#929](https://github.com/xmrig/xmrig/pull/929) Added support for new algorithm `cryptonight/wow`, short alias `cn/wow` (also known as CryptonightR), for upcoming [Wownero](http://wownero.org) fork on February 14. - -# v2.11.0 -- [#928](https://github.com/xmrig/xmrig/issues/928) Added support for new algorithm `cryptonight/gpu`, short alias `cn/gpu` (original name `cryptonight-gpu`), for upcoming [Ryo currency](https://ryo-currency.com) fork on February 14. -- [#749](https://github.com/xmrig/xmrig/issues/749) Added support for detect hardware AES in runtime on ARMv8 platforms. -- [#292](https://github.com/xmrig/xmrig/issues/292) Fixed build on ARMv8 platforms if compiler not support hardware AES. - -# v2.10.0 -- [#904](https://github.com/xmrig/xmrig/issues/904) Added new algorithm `cn-pico/trtl` (aliases `cryptonight-turtle`, `cn-trtl`) for upcoming TurtleCoin (TRTL) fork. -- Default value for option `max-cpu-usage` changed to `100` also this option now deprecated. - -# v2.9.4 -- [#913](https://github.com/xmrig/xmrig/issues/913) Fixed Masari (MSR) support (this update required for upcoming fork). -- [#915](https://github.com/xmrig/xmrig/pull/915) Improved security, JIT memory now read-only after patching. - -# v2.9.3 -- [#909](https://github.com/xmrig/xmrig/issues/909) Fixed compile errors on FreeBSD. -- [#912](https://github.com/xmrig/xmrig/pull/912) Fixed, C++ implementation of `cn/half` was produce up to 13% of invalid hashes. - -# v2.9.2 -- [#907](https://github.com/xmrig/xmrig/pull/907) Fixed crash on Linux. - -# v2.9.1 -- Restored compatibility with https://stellite.hashvault.pro. - -# v2.9.0 -- [#899](https://github.com/xmrig/xmrig/issues/899) Added support for new algorithm `cn/half` for Masari and Stellite forks. -- [#834](https://github.com/xmrig/xmrig/pull/834) Added ASM optimized code for AMD Bulldozer. -- [#839](https://github.com/xmrig/xmrig/issues/839) Fixed FreeBSD compile. -- [#857](https://github.com/xmrig/xmrig/pull/857) Fixed impossible to build for macOS without clang. - -# v2.8.3 -- [#813](https://github.com/xmrig/xmrig/issues/813) Fixed critical bug with Minergate pool and variant 2. - -# v2.8.1 -- [#768](https://github.com/xmrig/xmrig/issues/768) Fixed build with Visual Studio 2015. -- [#769](https://github.com/xmrig/xmrig/issues/769) Fixed regression, some ANSI escape sequences was in log with disabled colors. -- [#777](https://github.com/xmrig/xmrig/issues/777) Better report about pool connection issues. -- Simplified checks for ASM auto detection, only AES support necessary. -- Added missing options to `--help` output. - -# v2.8.0 -- **[#753](https://github.com/xmrig/xmrig/issues/753) Added new algorithm [CryptoNight variant 2](https://github.com/xmrig/xmrig/issues/753) for Monero fork, thanks [@SChernykh](https://github.com/SChernykh).** - - Added global and per thread option `"asm"` and and command line equivalent. -- **[#758](https://github.com/xmrig/xmrig/issues/758) Added SSL/TLS support for secure connections to pools.** - - Added per pool options `"tls"` and `"tls-fingerprint"` and command line equivalents. -- [#767](https://github.com/xmrig/xmrig/issues/767) Added config autosave feature, same with GPU miners. -- [#245](https://github.com/xmrig/xmrig-proxy/issues/245) Fixed API ID collision when run multiple miners on same machine. -- [#757](https://github.com/xmrig/xmrig/issues/757) Fixed send buffer overflow. - -# v2.6.4 -- [#700](https://github.com/xmrig/xmrig/issues/700) `cryptonight-lite/ipbc` replaced to `cryptonight-heavy/tube` for **Bittube (TUBE)**. -- Added `cryptonight/rto` (cryptonight variant 1 with IPBC/TUBE mod) variant for **Arto (RTO)** coin. -- Added `cryptonight/xao` (original cryptonight with bigger iteration count) variant for **Alloy (XAO)** coin. -- Better variant detection for **nicehash.com** and **minergate.com**. -- [#692](https://github.com/xmrig/xmrig/issues/692) Added support for specify both algorithm and variant via single `algo` option. - -# v2.6.3 -- **Added support for new cryptonight-heavy variant xhv** (`cn-heavy/xhv`) for upcoming Haven Protocol fork. -- **Added support for new cryptonight variant msr** (`cn/msr`) also known as `cryptonight-fast` for upcoming Masari fork. -- Added new detailed hashrate report. -- [#446](https://github.com/xmrig/xmrig/issues/446) Likely fixed SIGBUS error on 32 bit ARM CPUs. -- [#551](https://github.com/xmrig/xmrig/issues/551) Fixed `cn-heavy` algorithm on ARMv8. -- [#614](https://github.com/xmrig/xmrig/issues/614) Fixed display issue with huge pages percentage when colors disabled. -- [#615](https://github.com/xmrig/xmrig/issues/615) Fixed build without libcpuid. -- [#629](https://github.com/xmrig/xmrig/pull/629) Fixed file logging with non-seekable files. -- [#672](https://github.com/xmrig/xmrig/pull/672) Reverted back `cryptonight-light` and exit if no valid algorithm specified. - -# v2.6.2 - - [#607](https://github.com/xmrig/xmrig/issues/607) Fixed donation bug. - - [#610](https://github.com/xmrig/xmrig/issues/610) Fixed ARM build. - -# v2.6.1 - - [#168](https://github.com/xmrig/xmrig-proxy/issues/168) Added support for [mining algorithm negotiation](https://github.com/xmrig/xmrig-proxy/blob/dev/doc/STRATUM_EXT.md#1-mining-algorithm-negotiation). - - Added IPBC coin support, base algorithm `cn-lite` variant `ipbc`. - - [#581](https://github.com/xmrig/xmrig/issues/581) Added support for upcoming Stellite (XTL) fork, base algorithm `cn` variant `xtl`, variant can set now, no need do it after fork. - - Added support for **rig-id** stratum protocol extensions, compatible with xmr-stak. - - Changed behavior for option `variant=-1` for `cryptonight`, now variant is `1` by default, if you mine old coins need change `variant` to `0`. - - A lot of small fixes and better unification with proxy code. - -# v2.6.0-beta3 -- [#563](https://github.com/xmrig/xmrig/issues/563) **Added [advanced threads mode](https://github.com/xmrig/xmrig/issues/563), now possible configure each thread individually.** -- [#255](https://github.com/xmrig/xmrig/issues/563) Low power mode extended to **triple**, **quard** and **penta** modes. -- [#519](https://github.com/xmrig/xmrig/issues/519) Fixed high donation levels, improved donation start time randomization. -- [#554](https://github.com/xmrig/xmrig/issues/554) Fixed regression with `print-time` option. - -# v2.6.0-beta2 -- Improved performance for `cryptonight v7` especially in double hash mode. -- [#499](https://github.com/xmrig/xmrig/issues/499) IPv6 disabled for internal HTTP API by default, was causing issues on some systems. -- Added short aliases for algorithm names: `cn`, `cn-lite` and `cn-heavy`. -- Fixed regressions (v2.6.0-beta1 affected) - - [#494](https://github.com/xmrig/xmrig/issues/494) Command line option `--donate-level` was broken. - - [#502](https://github.com/xmrig/xmrig/issues/502) Build without libmicrohttpd was broken. - - Fixed nonce calculation for `--av 4` (software AES, double hash) was causing reduction of effective hashrate and rejected shares on nicehash. - -# v2.6.0-beta1 - - [#476](https://github.com/xmrig/xmrig/issues/476) **Added Cryptonight-Heavy support for Sumokoin ASIC resistance fork.** - - HTTP server now runs in main loop, it make possible easy extend API without worry about thread synchronization. - - Added initial graceful reload support, miner will reload configuration if config file changed, disabled by default until it will be fully implemented and tested. - - Added API endpoint `PUT /1/config` to update current config. - - Added API endpoint `GET /1/config` to get current active config. - - Added API endpoint `GET /1/threads` to get current active threads configuration. - - API endpoint `GET /` now deprecated, use `GET /1/summary` instead. - - Added `--api-no-ipv6` and similar config option to disable IPv6 support for HTTP API. - - Added `--api-no-restricted` to enable full access to api, this option has no effect if `--api-access-token` not specified. - -# v2.5.3 -- Fixed critical bug, in some cases miner was can't recovery connection and switch to failover pool, version 2.5.2 affected. If you use v2.6.0-beta3 this issue doesn't concern you. -- [#499](https://github.com/xmrig/xmrig/issues/499) IPv6 support disabled for internal HTTP API. -- Added workaround for nicehash.com if you use `cryptonightv7..nicehash.com` option `variant=1` will be set automatically. - -# v2.5.2 -- [#448](https://github.com/xmrig/xmrig/issues/478) Fixed broken reconnect. - -# v2.5.1 -- [#454](https://github.com/xmrig/xmrig/issues/454) Fixed build with libmicrohttpd version below v0.9.35. -- [#456](https://github.com/xmrig/xmrig/issues/459) Verbose errors related to donation pool was not fully silenced. -- [#459](https://github.com/xmrig/xmrig/issues/459) Fixed regression (version 2.5.0 affected) with connection to **xmr.f2pool.com**. - -# v2.5.0 -- [#434](https://github.com/xmrig/xmrig/issues/434) **Added support for Monero v7 PoW, scheduled on April 6.** -- Added full IPv6 support. -- Added protocol extension, when use the miner with xmrig-proxy 2.5+ no more need manually specify `nicehash` option. -- [#123](https://github.com/xmrig/xmrig-proxy/issues/123) Fixed regression (all versions since 2.4 affected) fragmented responses from pool/proxy was parsed incorrectly. -- [#428](https://github.com/xmrig/xmrig/issues/428) Fixed regression (version 2.4.5 affected) with CPU cache size detection. - -# v2.4.5 -- [#324](https://github.com/xmrig/xmrig/pull/324) Fixed build without libmicrohttpd (CMake cache issue). -- [#341](https://github.com/xmrig/xmrig/issues/341) Fixed wrong exit code and added command line option `--dry-run`. -- [#385](https://github.com/xmrig/xmrig/pull/385) Up to 20% performance increase for non-AES CPU and fixed Intel Core 2 cache detection. - -# v2.4.4 - - Added libmicrohttpd version to --version output. - - Fixed bug in singal handler, in some cases miner wasn't shutdown properly. - - Fixed recent MSVC 2017 version detection. - - [#279](https://github.com/xmrig/xmrig/pull/279) Fixed build on some macOS versions. - -# v2.4.3 - - [#94](https://github.com/xmrig/xmrig/issues/94#issuecomment-342019257) [#216](https://github.com/xmrig/xmrig/issues/216) Added **ARMv8** and **ARMv7** support. Hardware AES supported, thanks [Imran Yusuff](https://github.com/imranyusuff). - - [#157](https://github.com/xmrig/xmrig/issues/157) [#196](https://github.com/xmrig/xmrig/issues/196) Fixed Linux compile issues. - - [#184](https://github.com/xmrig/xmrig/issues/184) Fixed cache size detection for CPUs with disabled Hyper-Threading. - - [#200](https://github.com/xmrig/xmrig/issues/200) In some cases miner was doesn't write log to stdout. - -# v2.4.2 - - [#60](https://github.com/xmrig/xmrig/issues/60) Added FreeBSD support, thanks [vcambur](https://github.com/vcambur). - - [#153](https://github.com/xmrig/xmrig/issues/153) Fixed issues with dwarfpool.com. - -# v2.4.1 - - [#147](https://github.com/xmrig/xmrig/issues/147) Fixed comparability with monero-stratum. - -# v2.4.0 - - Added [HTTP API](https://github.com/xmrig/xmrig/wiki/API). - - Added comments support in config file. - - libjansson replaced to rapidjson. - - [#98](https://github.com/xmrig/xmrig/issues/98) Ignore `keepalive` option with minergate.com and nicehash.com. - - [#101](https://github.com/xmrig/xmrig/issues/101) Fixed MSVC 2017 (15.3) compile time version detection. - - [#108](https://github.com/xmrig/xmrig/issues/108) Silently ignore invalid values for `donate-level` option. - - [#111](https://github.com/xmrig/xmrig/issues/111) Fixed build without AEON support. - -# v2.3.1 -- [#68](https://github.com/xmrig/xmrig/issues/68) Fixed compatibility with Docker containers, was nothing print on console. - -# v2.3.0 -- Added `--cpu-priority` option (0 idle, 2 normal to 5 highest). -- Added `--user-agent` option, to set custom user-agent string for pool. For example `cpuminer-multi/0.1`. -- Added `--no-huge-pages` option, to disable huge pages support. -- [#62](https://github.com/xmrig/xmrig/issues/62) Don't send the login to the dev pool. -- Force reconnect if pool block miner IP address. helps switch to backup pool. -- Fixed: failed open default config file if path contains non English characters. -- Fixed: error occurred if try use unavailable stdin or stdout, regression since version 2.2.0. -- Fixed: message about huge pages support successfully enabled on Windows was not shown in release builds. - -# v2.2.1 -- Fixed [terminal issues](https://github.com/xmrig/xmrig-proxy/issues/2#issuecomment-319914085) after exit on Linux and OS X. - -# v2.2.0 -- [#46](https://github.com/xmrig/xmrig/issues/46) Restored config file support. Now possible use multiple config files and combine with command line options also added support for default config. -- Improved colors support on Windows, now used uv_tty, legacy code removed. -- QuickEdit Mode now disabled on Windows. -- Added interactive commands in console window:: **h**ashrate, **p**ause, **r**esume. -- Fixed autoconf mode for AMD FX CPUs. - -# v2.1.0 -- [#40](https://github.com/xmrig/xmrig/issues/40) -Improved miner shutdown, fixed crash on exit for Linux and OS X. -- Fixed, login request was contain malformed JSON if username or password has some special characters for example `\`. -- [#220](https://github.com/fireice-uk/xmr-stak-cpu/pull/220) Better support for Round Robin DNS, IP address now always chosen randomly instead of stuck on first one. -- Changed donation address, new [xmrig-proxy](https://github.com/xmrig/xmrig-proxy) is coming soon. - -# v2.0.2 -- Better deal with possible duplicate jobs from pool, show warning and ignore duplicates. -- For Windows builds libuv updated to version 1.13.1 and gcc to 7.1.0. - -# v2.0.1 - - [#27](https://github.com/xmrig/xmrig/issues/27) Fixed possibility crash on 32bit systems. - -# v2.0.0 - - Option `--backup-url` removed, instead now possibility specify multiple pools for example: `-o example1.com:3333 -u user1 -p password1 -k -o example2.com:5555 -u user2 -o example3.com:4444 -u user3` - - [#15](https://github.com/xmrig/xmrig/issues/15) Added option `-l, --log-file=FILE` to write log to file. - - [#15](https://github.com/xmrig/xmrig/issues/15) Added option `-S, --syslog` to use syslog for logging, Linux only. - - [#18](https://github.com/xmrig/xmrig/issues/18) Added nice messages for accepted/rejected shares with diff and network latency. - - [#20](https://github.com/xmrig/xmrig/issues/20) Fixed `--cpu-affinity` for more than 32 threads. - - Fixed Windows XP support. - - Fixed regression, option `--no-color` was not fully disable colored output. - - Show resolved pool IP address in miner output. - -# v1.0.1 -- Fix broken software AES implementation, app has crashed if CPU not support AES-NI, only version 1.0.0 affected. - -# v1.0.0 -- Miner complete rewritten in C++ with libuv. -- This version should be fully compatible (except config file) with previos versions, many new nice features will come in next versions. -- This is still beta. If you found regression, stability or perfomance issues or have an idea for new feature please fell free to open new [issue](https://github.com/xmrig/xmrig/issues/new). -- Added new option `--print-time=N`, print hashrate report every N seconds. -- New hashrate reports, by default every 60 secons. -- Added Microsoft Visual C++ 2015 and 2017 support. -- Removed dependency on libcurl. -- To compile this version from source please switch to [dev](https://github.com/xmrig/xmrig/tree/dev) branch. - -# v0.8.2 -- Fixed L2 cache size detection for AMD CPUs (Bulldozer/Piledriver/Steamroller/Excavator architecture). - -# v0.8.2 -- Fixed L2 cache size detection for AMD CPUs (Bulldozer/Piledriver/Steamroller/Excavator architecture). -- Fixed gcc 7.1 support. - -# v0.8.1 -- Added nicehash support, detects automaticaly by pool URL, for example `cryptonight.eu.nicehash.com:3355` or manually via option `--nicehash`. - -# v0.8.0 -- Added double hash mode, also known as lower power mode. `--av=2` and `--av=4`. -- Added smart automatic CPU configuration. Default threads count now depends on size of the L3 cache of CPU. -- Added CryptoNight-Lite support for AEON `-a cryptonight-lite`. -- Added `--max-cpu-usage` option for auto CPU configuration mode. -- Added `--safe` option for adjust threads and algorithm variations to current CPU. -- No more manual steps to enable huge pages on Windows. XMRig will do it automatically. -- Removed BMI2 algorithm variation. -- Removed default pool URL. - -# v0.6.0 -- Added automatic cryptonight self test. -- New software AES algorithm variation. Will be automatically selected if cpu not support AES-NI. -- Added 32 bit builds. -- Documented [algorithm variations](https://github.com/xmrig/xmrig#algorithm-variations). - -# v0.5.0 -- Initial public release. +# Previous versions +[doc/CHANGELOG_OLD.md](doc/CHANGELOG_OLD.md) diff --git a/CMakeLists.txt b/CMakeLists.txt index 14dcc931a..7600d556e 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -2,186 +2,135 @@ cmake_minimum_required(VERSION 2.8) project(xmrig) option(WITH_LIBCPUID "Use Libcpuid" ON) -option(WITH_AEON "CryptoNight-Lite support" ON) -option(WITH_SUMO "CryptoNight-Heavy support" ON) +option(WITH_HWLOC "Use hwloc" ON) +option(WITH_CN_LITE "CryptoNight-Lite support" ON) +option(WITH_CN_HEAVY "CryptoNight-Heavy support" ON) option(WITH_CN_PICO "CryptoNight-Pico support" ON) option(WITH_CN_GPU "CryptoNight-GPU support" ON) -option(WITH_HTTPD "HTTP REST API" ON) +option(WITH_RANDOMX "RandomX support" ON) +option(WITH_HTTP "HTTP protocol support (client/server)" ON) option(WITH_DEBUG_LOG "Enable debug log output" OFF) option(WITH_TLS "Enable OpenSSL support" ON) option(WITH_ASM "Enable ASM PoW implementations" ON) +option(WITH_EMBEDDED_CONFIG "Enable internal embedded JSON config" OFF) + option(BUILD_STATIC "Build static binary" OFF) option(ARM_TARGET "Force use specific ARM target 8 or 7" 0) -option(WITH_EMBEDDED_CONFIG "Enable internal embedded JSON config" OFF) +option(HWLOC_DEBUG "Enable hwloc debug helpers and log" OFF) + + +set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} "${CMAKE_SOURCE_DIR}/cmake") + include (CheckIncludeFile) include (cmake/cpu.cmake) +include (src/base/base.cmake) +include (src/backend/backend.cmake) set(HEADERS - src/api/NetworkState.h + "${HEADERS_BASE}" + "${HEADERS_BASE_HTTP}" + "${HEADERS_BACKEND}" + src/api/interfaces/IApiListener.h src/App.h - src/base/io/Json.h - src/base/io/Watcher.h - src/base/kernel/interfaces/IConfigListener.h - src/base/kernel/interfaces/ISignalListener.h - src/base/kernel/interfaces/IWatcherListener.h - src/base/kernel/Entry.h - src/base/kernel/Process.h - src/base/kernel/Signals.h - src/base/net/Pool.h - src/base/net/Pools.h - src/base/tools/Arguments.h - src/base/tools/Handle.h - src/base/tools/String.h - src/common/config/CommonConfig.h - src/common/config/ConfigLoader.h - src/common/config/ConfigWatcher.h - src/common/Console.h - src/common/cpu/Cpu.h - src/common/crypto/Algorithm.h - src/common/crypto/keccak.h - src/common/interfaces/IClientListener.h - src/common/interfaces/IConfig.h - src/common/interfaces/IConfigCreator.h - src/common/interfaces/IConsoleListener.h - src/common/interfaces/IControllerListener.h - src/common/interfaces/ICpuInfo.h - src/common/interfaces/ILogBackend.h - src/common/interfaces/IStrategy.h - src/common/interfaces/IStrategyListener.h - src/common/log/BasicLog.h - src/common/log/ConsoleLog.h - src/common/log/FileLog.h - src/common/log/Log.h - src/common/net/Client.h - src/common/net/Id.h - src/common/net/Job.h - src/common/net/Storage.h - src/common/net/strategies/FailoverStrategy.h - src/common/net/strategies/SinglePoolStrategy.h - src/common/net/SubmitResult.h - src/common/Platform.h - src/common/utils/c_str.h - src/common/utils/mm_malloc.h - src/common/xmrig.h - src/core/ConfigLoader_platform.h - src/core/ConfigLoader_default.h + src/core/config/Config_default.h + src/core/config/Config_platform.h + src/core/config/Config.h + src/core/config/ConfigTransform.h + src/core/config/usage.h src/core/Controller.h - src/interfaces/IJobResultListener.h - src/interfaces/IThread.h - src/interfaces/IWorker.h - src/Mem.h + src/core/Miner.h + src/net/interfaces/IJobResultListener.h src/net/JobResult.h + src/net/JobResults.h src/net/Network.h + src/net/NetworkState.h src/net/strategies/DonateStrategy.h src/Summary.h src/version.h - src/workers/CpuThread.h - src/workers/Handle.h - src/workers/Hashrate.h - src/workers/MultiWorker.h - src/workers/Worker.h - src/workers/Workers.h ) set(HEADERS_CRYPTO - src/crypto/c_blake256.h - src/crypto/c_groestl.h - src/crypto/c_jh.h - src/crypto/c_skein.h - src/crypto/CryptoNight.h - src/crypto/CryptoNight_constants.h - src/crypto/CryptoNight_monero.h - src/crypto/CryptoNight_test.h - src/crypto/groestl_tables.h - src/crypto/hash.h - src/crypto/skein_port.h - src/crypto/soft_aes.h - src/crypto/asm/CryptonightR_template.h + src/crypto/cn/asm/CryptonightR_template.h + src/crypto/cn/c_blake256.h + src/crypto/cn/c_groestl.h + src/crypto/cn/c_jh.h + src/crypto/cn/c_skein.h + src/crypto/cn/CnAlgo.h + src/crypto/cn/CnCtx.h + src/crypto/cn/CnHash.h + src/crypto/cn/CryptoNight_monero.h + src/crypto/cn/CryptoNight_test.h + src/crypto/cn/CryptoNight.h + src/crypto/cn/groestl_tables.h + src/crypto/cn/hash.h + src/crypto/cn/skein_port.h + src/crypto/cn/soft_aes.h + src/crypto/common/Algorithm.h + src/crypto/common/keccak.h + src/crypto/common/Nonce.h + src/crypto/common/portable/mm_malloc.h + src/crypto/common/VirtualMemory.h ) if (XMRIG_ARM) - set(HEADERS_CRYPTO "${HEADERS_CRYPTO}" src/crypto/CryptoNight_arm.h) + set(HEADERS_CRYPTO "${HEADERS_CRYPTO}" src/crypto/cn/CryptoNight_arm.h) else() - set(HEADERS_CRYPTO "${HEADERS_CRYPTO}" src/crypto/CryptoNight_x86.h) + set(HEADERS_CRYPTO "${HEADERS_CRYPTO}" src/crypto/cn/CryptoNight_x86.h) endif() set(SOURCES - src/api/NetworkState.cpp + "${SOURCES_BASE}" + "${SOURCES_BASE_HTTP}" + "${SOURCES_BACKEND}" src/App.cpp - src/base/io/Json.cpp - src/base/io/Watcher.cpp - src/base/kernel/Entry.cpp - src/base/kernel/Process.cpp - src/base/kernel/Signals.cpp - src/base/net/Pool.cpp - src/base/net/Pools.cpp - src/base/tools/Arguments.cpp - src/base/tools/Handle.cpp - src/base/tools/String.cpp - src/common/config/CommonConfig.cpp - src/common/config/ConfigLoader.cpp - src/common/config/ConfigWatcher.cpp - src/common/Console.cpp - src/common/crypto/Algorithm.cpp - src/common/crypto/keccak.cpp - src/common/log/BasicLog.cpp - src/common/log/ConsoleLog.cpp - src/common/log/FileLog.cpp - src/common/log/Log.cpp - src/common/net/Client.cpp - src/common/net/Job.cpp - src/common/net/strategies/FailoverStrategy.cpp - src/common/net/strategies/SinglePoolStrategy.cpp - src/common/net/SubmitResult.cpp - src/common/Platform.cpp - src/core/Config.cpp + src/core/config/Config.cpp + src/core/config/ConfigTransform.cpp src/core/Controller.cpp - src/Mem.cpp + src/core/Miner.cpp + src/net/JobResults.cpp src/net/Network.cpp + src/net/NetworkState.cpp src/net/strategies/DonateStrategy.cpp src/Summary.cpp - src/workers/CpuThread.cpp - src/workers/Handle.cpp - src/workers/Hashrate.cpp - src/workers/MultiWorker.cpp - src/workers/Worker.cpp - src/workers/Workers.cpp src/xmrig.cpp ) set(SOURCES_CRYPTO - src/crypto/c_groestl.c - src/crypto/c_blake256.c - src/crypto/c_jh.c - src/crypto/c_skein.c + src/crypto/cn/c_blake256.c + src/crypto/cn/c_groestl.c + src/crypto/cn/c_jh.c + src/crypto/cn/c_skein.c + src/crypto/cn/CnCtx.cpp + src/crypto/cn/CnHash.cpp + src/crypto/common/Algorithm.cpp + src/crypto/common/keccak.cpp + src/crypto/common/Nonce.cpp + src/crypto/common/VirtualMemory.cpp ) if (WIN32) set(SOURCES_OS + "${SOURCES_OS}" res/app.rc src/App_win.cpp - src/base/io/Json_win.cpp - src/common/Platform_win.cpp - src/Mem_win.cpp + src/crypto/common/VirtualMemory_win.cpp ) add_definitions(/DWIN32) set(EXTRA_LIBS ws2_32 psapi iphlpapi userenv) elseif (APPLE) set(SOURCES_OS + "${SOURCES_OS}" src/App_unix.cpp - src/base/io/Json_unix.cpp - src/common/Platform_mac.cpp - src/Mem_unix.cpp + src/crypto/common/VirtualMemory_unix.cpp ) else() set(SOURCES_OS + "${SOURCES_OS}" src/App_unix.cpp - src/base/io/Json_unix.cpp - src/common/Platform_unix.cpp - src/Mem_unix.cpp + src/crypto/common/VirtualMemory_unix.cpp ) if (CMAKE_SYSTEM_NAME STREQUAL FreeBSD) @@ -201,84 +150,102 @@ endif() add_definitions(/D__STDC_FORMAT_MACROS) add_definitions(/DUNICODE) -set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} "${CMAKE_SOURCE_DIR}/cmake") - find_package(UV REQUIRED) -include(cmake/flags.cmake) - -if (WITH_LIBCPUID) - add_subdirectory(src/3rdparty/libcpuid) - - include_directories(src/3rdparty/libcpuid) - set(CPUID_LIB cpuid) - set(SOURCES_CPUID src/core/cpu/AdvancedCpuInfo.h src/core/cpu/AdvancedCpuInfo.cpp src/core/cpu/Cpu.cpp) -else() - add_definitions(/DXMRIG_NO_LIBCPUID) - set(SOURCES_CPUID src/common/cpu/BasicCpuInfo.h src/common/cpu/Cpu.cpp) - - if (XMRIG_ARM) - set(SOURCES_CPUID ${SOURCES_CPUID} src/common/cpu/BasicCpuInfo_arm.cpp) - else() - set(SOURCES_CPUID ${SOURCES_CPUID} src/common/cpu/BasicCpuInfo.cpp) +if (WITH_RANDOMX) + include_directories(src/crypto/randomx) + add_definitions(/DXMRIG_ALGO_RANDOMX) + set(SOURCES_CRYPTO + "${SOURCES_CRYPTO}" + src/crypto/randomx/aes_hash.cpp + src/crypto/randomx/allocator.cpp + src/crypto/randomx/argon2_core.c + src/crypto/randomx/argon2_ref.c + src/crypto/randomx/blake2_generator.cpp + src/crypto/randomx/blake2/blake2b.c + src/crypto/randomx/bytecode_machine.cpp + src/crypto/randomx/dataset.cpp + src/crypto/randomx/instructions_portable.cpp + src/crypto/randomx/randomx.cpp + src/crypto/randomx/reciprocal.c + src/crypto/randomx/soft_aes.cpp + src/crypto/randomx/superscalar.cpp + src/crypto/randomx/virtual_machine.cpp + src/crypto/randomx/virtual_memory.cpp + src/crypto/randomx/vm_compiled_light.cpp + src/crypto/randomx/vm_compiled.cpp + src/crypto/randomx/vm_interpreted_light.cpp + src/crypto/randomx/vm_interpreted.cpp + src/crypto/rx/Rx.cpp + src/crypto/rx/Rx.h + src/crypto/rx/RxAlgo.cpp + src/crypto/rx/RxAlgo.h + src/crypto/rx/RxCache.cpp + src/crypto/rx/RxCache.h + src/crypto/rx/RxConfig.cpp + src/crypto/rx/RxConfig.h + src/crypto/rx/RxDataset.cpp + src/crypto/rx/RxDataset.h + src/crypto/rx/RxVm.cpp + src/crypto/rx/RxVm.h + ) + if (NOT ARCH_ID) + set(ARCH_ID ${CMAKE_HOST_SYSTEM_PROCESSOR}) endif() + if (CMAKE_C_COMPILER_ID MATCHES MSVC) + enable_language(ASM_MASM) + list(APPEND SOURCES_CRYPTO + src/crypto/randomx/jit_compiler_x86_static.asm + src/crypto/randomx/jit_compiler_x86.cpp + ) + elseif (NOT XMRIG_ARM AND CMAKE_SIZEOF_VOID_P EQUAL 8) + list(APPEND SOURCES_CRYPTO + src/crypto/randomx/jit_compiler_x86_static.S + src/crypto/randomx/jit_compiler_x86.cpp + ) + # cheat because cmake and ccache hate each other + set_property(SOURCE src/crypto/randomx/jit_compiler_x86_static.S PROPERTY LANGUAGE C) + endif() +else() + remove_definitions(/DXMRIG_ALGO_RANDOMX) endif() +include(cmake/flags.cmake) + include(cmake/OpenSSL.cmake) include(cmake/asm.cmake) include(cmake/cn-gpu.cmake) -CHECK_INCLUDE_FILE (syslog.h HAVE_SYSLOG_H) -if (HAVE_SYSLOG_H) - add_definitions(/DHAVE_SYSLOG_H) - set(SOURCES_SYSLOG src/common/log/SysLog.h src/common/log/SysLog.cpp) +if (WITH_CN_LITE) + add_definitions(/DXMRIG_ALGO_CN_LITE) endif() -if (NOT WITH_AEON) - add_definitions(/DXMRIG_NO_AEON) +if (WITH_CN_HEAVY) + add_definitions(/DXMRIG_ALGO_CN_HEAVY) endif() -if (NOT WITH_SUMO) - add_definitions(/DXMRIG_NO_SUMO) -endif() - -if (NOT WITH_IPBC) - add_definitions(/DXMRIG_NO_IPBC) -endif() - -if (NOT WITH_CN_PICO) - add_definitions(/DXMRIG_NO_CN_PICO) +if (WITH_CN_PICO) + add_definitions(/DXMRIG_ALGO_CN_PICO) endif() if (WITH_EMBEDDED_CONFIG) add_definitions(/DXMRIG_FEATURE_EMBEDDED_CONFIG) endif() -if (WITH_HTTPD) - find_package(MHD) - - if (MHD_FOUND) - include_directories(${MHD_INCLUDE_DIRS}) - set(HTTPD_SOURCES - src/api/Api.h - src/api/ApiRouter.h - src/common/api/HttpBody.h - src/common/api/Httpd.h - src/common/api/HttpReply.h - src/common/api/HttpRequest.h - src/api/Api.cpp - src/api/ApiRouter.cpp - src/common/api/Httpd.cpp - src/common/api/HttpRequest.cpp - ) - else() - message(FATAL_ERROR "microhttpd NOT found: use `-DWITH_HTTPD=OFF` to build without http deamon support") - endif() +if (WITH_HTTP) + set(HTTP_SOURCES + src/api/Api.cpp + src/api/Api.h + src/api/Httpd.cpp + src/api/Httpd.h + src/api/interfaces/IApiRequest.h + src/api/requests/ApiRequest.cpp + src/api/requests/ApiRequest.h + src/api/requests/HttpApiRequest.cpp + src/api/requests/HttpApiRequest.h + ) else() - set(HTTPD_SOURCES "") - set(MHD_LIBRARY "") - add_definitions(/DXMRIG_NO_HTTPD) - add_definitions(/DXMRIG_NO_API) + set(HTTP_SOURCES "") endif() include_directories(src) @@ -293,5 +260,5 @@ if (WITH_DEBUG_LOG) add_definitions(/DAPP_DEBUG) endif() -add_executable(${CMAKE_PROJECT_NAME} ${HEADERS} ${SOURCES} ${SOURCES_OS} ${SOURCES_CPUID} ${HEADERS_CRYPTO} ${SOURCES_CRYPTO} ${SOURCES_SYSLOG} ${HTTPD_SOURCES} ${TLS_SOURCES} ${XMRIG_ASM_SOURCES} ${CN_GPU_SOURCES}) -target_link_libraries(${CMAKE_PROJECT_NAME} ${XMRIG_ASM_LIBRARY} ${OPENSSL_LIBRARIES} ${UV_LIBRARIES} ${MHD_LIBRARY} ${EXTRA_LIBS} ${CPUID_LIB}) +add_executable(${CMAKE_PROJECT_NAME} ${HEADERS} ${SOURCES} ${SOURCES_OS} ${SOURCES_CPUID} ${HEADERS_CRYPTO} ${SOURCES_CRYPTO} ${SOURCES_SYSLOG} ${HTTP_SOURCES} ${TLS_SOURCES} ${XMRIG_ASM_SOURCES} ${CN_GPU_SOURCES}) +target_link_libraries(${CMAKE_PROJECT_NAME} ${XMRIG_ASM_LIBRARY} ${OPENSSL_LIBRARIES} ${UV_LIBRARIES} ${EXTRA_LIBS} ${CPUID_LIB}) diff --git a/README.md b/README.md index 5b6dc7a03..ee499d475 100644 --- a/README.md +++ b/README.md @@ -7,136 +7,79 @@ [![GitHub stars](https://img.shields.io/github/stars/xmrig/xmrig.svg)](https://github.com/xmrig/xmrig/stargazers) [![GitHub forks](https://img.shields.io/github/forks/xmrig/xmrig.svg)](https://github.com/xmrig/xmrig/network) -XMRig is a high performance Monero (XMR) CPU miner, with official support for Windows. -Originally based on cpuminer-multi with heavy optimizations/rewrites and removing a lot of legacy code, since version 1.0.0 completely rewritten from scratch on C++. +XMRig is a high performance RandomX and CryptoNight CPU miner, with official support for Windows. * This is the **CPU-mining** version, there is also a [NVIDIA GPU version](https://github.com/xmrig/xmrig-nvidia) and [AMD GPU version]( https://github.com/xmrig/xmrig-amd). -* [Roadmap](https://github.com/xmrig/xmrig/issues/106) for next releases. - + #### Table of contents -* [Features](#features) * [Download](#download) * [Usage](#usage) -* [Algorithm variations](#algorithm-variations) * [Build](https://github.com/xmrig/xmrig/wiki/Build) -* [Common Issues](#common-issues) -* [Other information](#other-information) * [Donations](#donations) -* [Release checksums](#release-checksums) * [Contacts](#contacts) -## Features -* High performance. -* Official Windows support. -* Small Windows executable, without dependencies. -* x86/x64 support. -* Support for backup (failover) mining server. -* keepalived support. -* Command line options compatible with cpuminer. -* CryptoNight-Lite support for AEON. -* Smart automatic [CPU configuration](https://github.com/xmrig/xmrig/wiki/Threads). -* Nicehash support -* It's open source software. - ## Download * Binary releases: https://github.com/xmrig/xmrig/releases * Git tree: https://github.com/xmrig/xmrig.git * Clone with `git clone https://github.com/xmrig/xmrig.git` :hammer: [Build instructions](https://github.com/xmrig/xmrig/wiki/Build). ## Usage -Use [config.xmrig.com](https://config.xmrig.com/xmrig) to generate, edit or share configurations. +Preferend way to configure miner is [JSON config file](src/config.json) as more flexible and human frendly, command line interface not cover all features, for example mining profiles for different algorithms. Most impotant options can be changed in runtime without miner restart by editing config or via API. ### Options ``` - -a, --algo=ALGO specify the algorithm to use - cryptonight - cryptonight-lite - cryptonight-heavy - -o, --url=URL URL of mining server - -O, --userpass=U:P username:password pair for mining server - -u, --user=USERNAME username for mining server - -p, --pass=PASSWORD password for mining server - --rig-id=ID rig identifier for pool-side statistics (needs pool support) - -t, --threads=N number of miner threads - -v, --av=N algorithm variation, 0 auto select - -k, --keepalive send keepalived packet for prevent timeout (needs pool support) - --nicehash enable nicehash.com support - --tls enable SSL/TLS support (needs pool support) - --tls-fingerprint=F pool TLS certificate fingerprint, if set enable strict certificate pinning - -r, --retries=N number of times to retry before switch to backup server (default: 5) - -R, --retry-pause=N time to pause between retries (default: 5) - --cpu-affinity set process affinity to CPU core(s), mask 0x3 for cores 0 and 1 - --cpu-priority set process priority (0 idle, 2 normal to 5 highest) - --no-huge-pages disable huge pages support - --no-color disable colored output - --variant algorithm PoW variant - --donate-level=N donate level, default 5% (5 minutes in 100 minutes) - --user-agent set custom user-agent string for pool - -B, --background run the miner in the background - -c, --config=FILE load a JSON-format configuration file - -l, --log-file=FILE log all output to a file - -S, --syslog use system log for output messages - --max-cpu-usage=N maximum CPU usage for automatic threads mode (default 75) - --safe safe adjust threads and av settings for current CPU - --asm=ASM ASM code for cn/2, possible values: auto, none, intel, ryzen. - --print-time=N print hashrate report every N seconds - --api-port=N port for the miner API - --api-access-token=T access token for API - --api-worker-id=ID custom worker-id for API - --api-id=ID custom instance ID for API - --api-ipv6 enable IPv6 support for API - --api-no-restricted enable full remote access (only if API token set) - --dry-run test configuration and exit - -h, --help display this help and exit - -V, --version output version information and exit + -a, --algo=ALGO specify the algorithm to use + cn/r, cn/2, cn/1, cn/0, cn/double, cn/half, cn/fast, + cn/rwz, cn/zls, cn/xao, cn/rto, cn/gpu, + cn-lite/1, + cn-heavy/xhv, cn-heavy/tube, cn-heavy/0, + cn-pico, + rx/wow, rx/loki + -o, --url=URL URL of mining server + -O, --userpass=U:P username:password pair for mining server + -u, --user=USERNAME username for mining server + -p, --pass=PASSWORD password for mining server + --rig-id=ID rig identifier for pool-side statistics (needs pool support) + -t, --threads=N number of miner threads + -v, --av=N algorithm variation, 0 auto select + -k, --keepalive send keepalived packet for prevent timeout (needs pool support) + --nicehash enable nicehash.com support + --tls enable SSL/TLS support (needs pool support) + --tls-fingerprint=F pool TLS certificate fingerprint, if set enable strict certificate pinning + --daemon use daemon RPC instead of pool for solo mining + --daemon-poll-interval=N daemon poll interval in milliseconds (default: 1000) + -r, --retries=N number of times to retry before switch to backup server (default: 5) + -R, --retry-pause=N time to pause between retries (default: 5) + --cpu-affinity set process affinity to CPU core(s), mask 0x3 for cores 0 and 1 + --cpu-priority set process priority (0 idle, 2 normal to 5 highest) + --no-huge-pages disable huge pages support + --no-color disable colored output + --donate-level=N donate level, default 5% (5 minutes in 100 minutes) + --user-agent set custom user-agent string for pool + -B, --background run the miner in the background + -c, --config=FILE load a JSON-format configuration file + -l, --log-file=FILE log all output to a file + --asm=ASM ASM optimizations, possible values: auto, none, intel, ryzen, bulldozer. + --print-time=N print hashrate report every N seconds + --api-worker-id=ID custom worker-id for API + --api-id=ID custom instance ID for API + --http-enabled enable HTTP API + --http-host=HOST bind host for HTTP API (default: 127.0.0.1) + --http-port=N bind port for HTTP API + --http-access-token=T access token for HTTP API + --http-no-restricted enable full remote access to HTTP API (only if access token set) + --randomx-init=N threads count to initialize RandomX dataset + --randomx-no-numa disable NUMA support for RandomX + --export-topology export hwloc topology to a XML file and exit + --dry-run test configuration and exit + -h, --help display this help and exit + -V, --version output version information and exit ``` -Also you can use configuration via config file, default name **config.json**. Some options available only via config file: [`autosave`](https://github.com/xmrig/xmrig/issues/767), [`hw-aes`](https://github.com/xmrig/xmrig/issues/563). `watch` option currently not implemented in miners only in proxy. - -## Algorithm variations - -- `av` option used for automatic and simple threads mode (when you specify only threads count). -- For [advanced threads mode](https://github.com/xmrig/xmrig/issues/563) each thread configured individually and `av` option not used. - -| av | Hashes per round | Hardware AES | -|----|------------------|--------------| -| 1 | 1 (Single) | yes | -| 2 | 2 (Double) | yes | -| 3 | 1 (Single) | no | -| 4 | 2 (Double) | no | -| 5 | 3 (Triple) | yes | -| 6 | 4 (Quard) | yes | -| 7 | 5 (Penta) | yes | -| 8 | 3 (Triple) | no | -| 9 | 4 (Quard) | no | -| 10 | 5 (Penta) | no | - -## Common Issues -### HUGE PAGES unavailable -* Run XMRig as Administrator. -* Since version 0.8.0 XMRig automatically enables SeLockMemoryPrivilege for current user, but reboot or sign out still required. [Manual instruction](https://msdn.microsoft.com/en-gb/library/ms190730.aspx). - -## Other information -* No HTTP support, only stratum protocol support. -* Default donation 5% (5 minutes in 100 minutes) can be reduced to 1% via option `donate-level`. - - -### CPU mining performance -* **Intel i7-7700** - 307 H/s (4 threads) -* **AMD Ryzen 7 1700X** - 560 H/s (8 threads) - -Please note performance is highly dependent on system load. The numbers above are obtained on an idle system. Tasks heavily using a processor cache, such as video playback, can greatly degrade hashrate. Optimal number of threads depends on the size of the L3 cache of a processor, 1 thread requires 2 MB of cache. - -### Maximum performance checklist -* Idle operating system. -* Do not exceed optimal thread count. -* Use modern CPUs with AES-NI instruction set. -* Try setup optimal cpu affinity. -* Enable fast memory (Large/Huge pages). - ## Donations +* Default donation 5% (5 minutes in 100 minutes) can be reduced to 1% via option `donate-level` or disabled in source code. * XMR: `48edfHu7V9Z84YzzMa6fUueoELZ9ZRXq9VetWzYGzKt52XU5xvqgzYnDK9URnRoJMk1j8nLwEVsaSWJ4fhdUyZijBGUicoD` * BTC: `1P7ujsXeX7GxQwHNnJsRMgAdNkFZmNVqJT` diff --git a/cmake/FindHWLOC.cmake b/cmake/FindHWLOC.cmake new file mode 100644 index 000000000..55309d3e5 --- /dev/null +++ b/cmake/FindHWLOC.cmake @@ -0,0 +1,25 @@ +find_path( + HWLOC_INCLUDE_DIR + NAMES hwloc.h + PATHS "${XMRIG_DEPS}" ENV "XMRIG_DEPS" + PATH_SUFFIXES "include" + NO_DEFAULT_PATH +) + +find_path(HWLOC_INCLUDE_DIR NAMES hwloc.h) + +find_library( + HWLOC_LIBRARY + NAMES hwloc.a hwloc libhwloc + PATHS "${XMRIG_DEPS}" ENV "XMRIG_DEPS" + PATH_SUFFIXES "lib" + NO_DEFAULT_PATH +) + +find_library(HWLOC_LIBRARY NAMES hwloc.a hwloc libhwloc) + +set(HWLOC_LIBRARIES ${HWLOC_LIBRARY}) +set(HWLOC_INCLUDE_DIRS ${HWLOC_INCLUDE_DIR}) + +include(FindPackageHandleStandardArgs) +find_package_handle_standard_args(HWLOC DEFAULT_MSG HWLOC_LIBRARY HWLOC_INCLUDE_DIR) diff --git a/cmake/FindMHD.cmake b/cmake/FindMHD.cmake deleted file mode 100644 index 7a598e026..000000000 --- a/cmake/FindMHD.cmake +++ /dev/null @@ -1,49 +0,0 @@ -# - Try to find MHD -# Once done this will define -# -# MHD_FOUND - system has MHD -# MHD_INCLUDE_DIRS - the MHD include directory -# MHD_LIBRARY - Link these to use MHD - -find_path( - MHD_INCLUDE_DIR - NAMES microhttpd.h - PATHS "${XMRIG_DEPS}" ENV "XMRIG_DEPS" - PATH_SUFFIXES "include" - DOC "microhttpd include dir" - NO_DEFAULT_PATH -) - -find_path(MHD_INCLUDE_DIR NAMES microhttpd.h) - -find_library( - MHD_LIBRARY - NAMES libmicrohttpd.a microhttpd libmicrohttpd - PATHS "${XMRIG_DEPS}" ENV "XMRIG_DEPS" - PATH_SUFFIXES "lib" - DOC "microhttpd library" - NO_DEFAULT_PATH -) - -find_library(MHD_LIBRARY NAMES microhttpd libmicrohttpd) - -set(MHD_INCLUDE_DIRS ${MHD_INCLUDE_DIR}) -set(MHD_LIBRARIES ${MHD_LIBRARY}) - -# debug library on windows -# same naming convention as in qt (appending debug library with d) -# boost is using the same "hack" as us with "optimized" and "debug" -# official MHD project actually uses _d suffix -if (${CMAKE_CXX_COMPILER_ID} STREQUAL MSVC) - find_library( - MHD_LIBRARY_DEBUG - NAMES microhttpd_d microhttpd-10_d libmicrohttpd_d libmicrohttpd-dll_d - DOC "mhd debug library" - ) - set(MHD_LIBRARIES optimized ${MHD_LIBRARIES} debug ${MHD_LIBRARY_DEBUG}) -endif() - -include(FindPackageHandleStandardArgs) -find_package_handle_standard_args(MHD DEFAULT_MSG MHD_LIBRARY MHD_INCLUDE_DIR) -mark_as_advanced(MHD_INCLUDE_DIR MHD_LIBRARY) - diff --git a/cmake/OpenSSL.cmake b/cmake/OpenSSL.cmake index 79731d6f7..ea46081cf 100644 --- a/cmake/OpenSSL.cmake +++ b/cmake/OpenSSL.cmake @@ -11,15 +11,21 @@ if (WITH_TLS) find_package(OpenSSL) if (OPENSSL_FOUND) - set(TLS_SOURCES src/common/net/Tls.h src/common/net/Tls.cpp) + set(TLS_SOURCES src/base/net/stratum/Tls.h src/base/net/stratum/Tls.cpp) include_directories(${OPENSSL_INCLUDE_DIR}) + + if (WITH_HTTP) + set(TLS_SOURCES ${TLS_SOURCES} src/base/net/http/HttpsClient.h src/base/net/http/HttpsClient.cpp) + endif() else() message(FATAL_ERROR "OpenSSL NOT found: use `-DWITH_TLS=OFF` to build without TLS support") endif() + + add_definitions(/DXMRIG_FEATURE_TLS) else() set(TLS_SOURCES "") set(OPENSSL_LIBRARIES "") - add_definitions(/DXMRIG_NO_TLS) + remove_definitions(/DXMRIG_FEATURE_TLS) set(CMAKE_PROJECT_NAME "${CMAKE_PROJECT_NAME}-notls") endif() diff --git a/cmake/asm.cmake b/cmake/asm.cmake index 389f67231..e445defde 100644 --- a/cmake/asm.cmake +++ b/cmake/asm.cmake @@ -6,13 +6,13 @@ if (WITH_ASM AND NOT XMRIG_ARM AND CMAKE_SIZEOF_VOID_P EQUAL 8) if (MSVC_TOOLSET_VERSION GREATER_EQUAL 141) set(XMRIG_ASM_FILES - "src/crypto/asm/cn_main_loop.asm" - "src/crypto/asm/CryptonightR_template.asm" + "src/crypto/cn/asm/cn_main_loop.asm" + "src/crypto/cn/asm/CryptonightR_template.asm" ) else() set(XMRIG_ASM_FILES - "src/crypto/asm/win64/cn_main_loop.asm" - "src/crypto/asm/win64/CryptonightR_template.asm" + "src/crypto/cn/asm/win64/cn_main_loop.asm" + "src/crypto/cn/asm/win64/CryptonightR_template.asm" ) endif() @@ -22,13 +22,13 @@ if (WITH_ASM AND NOT XMRIG_ARM AND CMAKE_SIZEOF_VOID_P EQUAL 8) if (WIN32 AND CMAKE_C_COMPILER_ID MATCHES GNU) set(XMRIG_ASM_FILES - "src/crypto/asm/win64/cn_main_loop.S" - "src/crypto/asm/CryptonightR_template.S" + "src/crypto/cn/asm/win64/cn_main_loop.S" + "src/crypto/cn/asm/CryptonightR_template.S" ) else() set(XMRIG_ASM_FILES - "src/crypto/asm/cn_main_loop.S" - "src/crypto/asm/CryptonightR_template.S" + "src/crypto/cn/asm/cn_main_loop.S" + "src/crypto/cn/asm/CryptonightR_template.S" ) endif() @@ -36,10 +36,17 @@ if (WITH_ASM AND NOT XMRIG_ARM AND CMAKE_SIZEOF_VOID_P EQUAL 8) endif() add_library(${XMRIG_ASM_LIBRARY} STATIC ${XMRIG_ASM_FILES}) - set(XMRIG_ASM_SOURCES src/crypto/Asm.h src/crypto/Asm.cpp src/crypto/CryptonightR_gen.cpp) + set(XMRIG_ASM_SOURCES + src/crypto/common/Assembly.h + src/crypto/common/Assembly.cpp + src/crypto/cn/r/CryptonightR_gen.cpp + ) set_property(TARGET ${XMRIG_ASM_LIBRARY} PROPERTY LINKER_LANGUAGE C) + + add_definitions(/DXMRIG_FEATURE_ASM) else() set(XMRIG_ASM_SOURCES "") set(XMRIG_ASM_LIBRARY "") - add_definitions(/DXMRIG_NO_ASM) + + remove_definitions(/DXMRIG_FEATURE_ASM) endif() diff --git a/cmake/cn-gpu.cmake b/cmake/cn-gpu.cmake index b529f0b2d..7aa489e6f 100644 --- a/cmake/cn-gpu.cmake +++ b/cmake/cn-gpu.cmake @@ -1,23 +1,25 @@ if (WITH_CN_GPU AND CMAKE_SIZEOF_VOID_P EQUAL 8) if (XMRIG_ARM) - set(CN_GPU_SOURCES src/crypto/cn_gpu_arm.cpp) + set(CN_GPU_SOURCES src/crypto/cn/gpu/cn_gpu_arm.cpp) if (CMAKE_CXX_COMPILER_ID MATCHES GNU OR CMAKE_CXX_COMPILER_ID MATCHES Clang) - set_source_files_properties(src/crypto/cn_gpu_arm.cpp PROPERTIES COMPILE_FLAGS "-O3") + set_source_files_properties(src/crypto/cn/gpu/cn_gpu_arm.cpp PROPERTIES COMPILE_FLAGS "-O3") endif() else() - set(CN_GPU_SOURCES src/crypto/cn_gpu_avx.cpp src/crypto/cn_gpu_ssse3.cpp) + set(CN_GPU_SOURCES src/crypto/cn/gpu/cn_gpu_avx.cpp src/crypto/cn/gpu/cn_gpu_ssse3.cpp) if (CMAKE_CXX_COMPILER_ID MATCHES GNU OR CMAKE_CXX_COMPILER_ID MATCHES Clang) - set_source_files_properties(src/crypto/cn_gpu_avx.cpp PROPERTIES COMPILE_FLAGS "-O3 -mavx2") - set_source_files_properties(src/crypto/cn_gpu_ssse3.cpp PROPERTIES COMPILE_FLAGS "-O3") + set_source_files_properties(src/crypto/cn/gpu/cn_gpu_avx.cpp PROPERTIES COMPILE_FLAGS "-O3 -mavx2") + set_source_files_properties(src/crypto/cn/gpu/cn_gpu_ssse3.cpp PROPERTIES COMPILE_FLAGS "-O3") elseif (CMAKE_CXX_COMPILER_ID MATCHES MSVC) - set_source_files_properties(src/crypto/cn_gpu_avx.cpp PROPERTIES COMPILE_FLAGS "/arch:AVX") + set_source_files_properties(src/crypto/cn/gpu/cn_gpu_avx.cpp PROPERTIES COMPILE_FLAGS "/arch:AVX") endif() endif() + + add_definitions(/DXMRIG_ALGO_CN_GPU) else() set(CN_GPU_SOURCES "") - add_definitions(/DXMRIG_NO_CN_GPU) + remove_definitions(/DXMRIG_ALGO_CN_GPU) endif() diff --git a/cmake/flags.cmake b/cmake/flags.cmake index d50b5c84b..bc441dd03 100644 --- a/cmake/flags.cmake +++ b/cmake/flags.cmake @@ -17,7 +17,7 @@ if (CMAKE_CXX_COMPILER_ID MATCHES GNU) set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Wall -Wno-strict-aliasing") set(CMAKE_C_FLAGS_RELEASE "${CMAKE_C_FLAGS_RELEASE} -Ofast") - set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wall -fno-exceptions -fno-rtti -Wno-class-memaccess") + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wall -fexceptions -fno-rtti -Wno-class-memaccess") set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} -Ofast -s") if (XMRIG_ARMv8) @@ -34,7 +34,11 @@ if (CMAKE_CXX_COMPILER_ID MATCHES GNU) endif() if (WIN32) - set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -static") + if (CMAKE_SIZEOF_VOID_P EQUAL 8) + set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -static") + else() + set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -static -Wl,--large-address-aware") + endif() else() set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -static-libgcc -static-libstdc++") endif() @@ -61,7 +65,7 @@ elseif (CMAKE_CXX_COMPILER_ID MATCHES Clang) set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Wall") set(CMAKE_C_FLAGS_RELEASE "${CMAKE_C_FLAGS_RELEASE} -Ofast -funroll-loops -fmerge-all-constants") - set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wall -fno-exceptions -fno-rtti -Wno-missing-braces") + set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wall -fexceptions -fno-rtti -Wno-missing-braces") set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} -Ofast -funroll-loops -fmerge-all-constants") if (XMRIG_ARMv8) @@ -81,3 +85,10 @@ elseif (CMAKE_CXX_COMPILER_ID MATCHES Clang) endif() endif() + +if (NOT WIN32) + check_symbol_exists("__builtin___clear_cache" "stdlib.h" HAVE_BUILTIN_CLEAR_CACHE) + if (HAVE_BUILTIN_CLEAR_CACHE) + add_definitions(/DHAVE_BUILTIN_CLEAR_CACHE) + endif() +endif() diff --git a/doc/ALGORITHMS.md b/doc/ALGORITHMS.md index 835a1d491..ab55bb743 100644 --- a/doc/ALGORITHMS.md +++ b/doc/ALGORITHMS.md @@ -1,22 +1,14 @@ # Algorithms -XMRig uses a different way to specify algorithms, compared to other miners. - -Algorithm selection splitted to 2 parts: - - * Global base algorithm per miner or proxy instance, `algo` option. Possible values: `cryptonight`, `cryptonight-lite`, `cryptonight-heavy`. - * Algorithm variant specified separately for each pool, `variant` option. - * [Full table for supported algorithm and variants.](https://github.com/xmrig/xmrig-proxy/blob/master/doc/STRATUM_EXT.md#14-algorithm-names-and-variants) +Since version 3 mining [algorithm](#algorithm-names) should specified for each pool separately (`algo` option), earlier versions was use one global `algo` option and per pool `variant` option (this option was removed in v3). If your pool support [mining algorithm negotiation](https://github.com/xmrig/xmrig-proxy/issues/168) you may not specify this option at all. #### Example ```json { - "algo": "cryptonight", - ... "pools": [ { "url": "...", - "variant": 1, + "algo": "cn/r", ... } ], @@ -24,8 +16,11 @@ Algorithm selection splitted to 2 parts: } ``` -## Mining algorithm negotiation -If your pool support [mining algorithm negotiation](https://github.com/xmrig/xmrig-proxy/issues/168) miner will choice proper variant automaticaly and if you choice wrong base algorithm you will see error message. +#### Pools with mining algorithm negotiation support. -Pools with mining algorithm negotiation support. * [www.hashvault.pro](https://www.hashvault.pro/) + * [moneroocean.stream](https://moneroocean.stream) + + ## Algorithm names + +* https://xmrig.com/docs/algorithms diff --git a/doc/API.md b/doc/API.md index 3357eabb5..2cd0fbbe1 100644 --- a/doc/API.md +++ b/doc/API.md @@ -1,26 +1,39 @@ # HTTP API -If you want use API you need choice a port where is internal HTTP server will listen for incoming connections. API will not available if miner built without `libmicrohttpd`. +If you want use HTTP API you need enable it (`"enabled": true,`) then choice `port` and optionaly `host`. API not available if miner built without HTTP support (`-DWITH_HTTP=OFF`). + +Offical HTTP client for API: http://workers.xmrig.info/ Example configuration: ```json "api": { - "port": 44444, - "access-token": "TOKEN", - "worker-id": null, - "ipv6": false, - "restricted": false + "id": null, + "worker-id": null, }, +"http": { + "enabled": false, + "host": "127.0.0.1", + "port": 0, + "access-token": null, + "restricted": true +} ``` -* **port** Port for incoming connections `http://:`. -* **access-token** [Bearer](https://gist.github.com/xmrig/c75fdd1f8e0f3bac05500be2ab718f8e#file-api-html-L54) access token to secure access to API. +#### Global API options +* **id** Miner ID, if not set created automatically. * **worker-id** Optional worker name, if not set will be detected automatically. -* **ipv6** Enable (`true`) or disable (`false`) IPv6 for API. + +#### HTTP API options, +* **enabled** Enable (`true`) or disable (`false`) HTTP API. +* **host** Host for incoming connections `http://:`, to allow connections from all interfaces use `0.0.0.0` (IPv4) or `::` (IPv4+IPv6). +* **port** Port for incoming connections `http://:`, zero port is valid option and means random port. +* **access-token** [Bearer](https://gist.github.com/xmrig/c75fdd1f8e0f3bac05500be2ab718f8e#file-api-html-L54) access token to secure access to API. Miner support this token only via `Authorization` header. * **restricted** Use `false` to allow remote configuration. -If you prefer use command line options instead of config file, you can use options: `--api-port`, `--api-access-token`, `--api-worker-id`, `--api-ipv6` and `api-no-restricted`. +If you prefer use command line options instead of config file, you can use options: `--api-id`, `--api-worker-id`, `--http-enabled`, `--http-host`, `--http-access-token`, `--http-port`, `--http-no-restricted`. + +Versions before 2.15 was use another options for API https://github.com/xmrig/xmrig/issues/1007 ## Endpoints @@ -50,4 +63,4 @@ Curl example: ``` curl -v --data-binary @config.json -X PUT -H "Content-Type: application/json" -H "Authorization: Bearer SECRET" http://127.0.0.1:44444/1/config -``` \ No newline at end of file +``` diff --git a/doc/CHANGELOG_OLD.md b/doc/CHANGELOG_OLD.md new file mode 100644 index 000000000..15ae35a9c --- /dev/null +++ b/doc/CHANGELOG_OLD.md @@ -0,0 +1,343 @@ +# v2.99.5-beta +- [#1066](https://github.com/xmrig/xmrig/issues/1066#issuecomment-518080529) Fixed crash and added error message if pool not ready for RandomX. +- [#1092](https://github.com/xmrig/xmrig/issues/1092) Fixed crash if wrong CPU affinity used. +- [#1103](https://github.com/xmrig/xmrig/issues/1103) Improved auto configuration for RandomX for CPUs where L2 cache is limiting factor. +- [#1105](https://github.com/xmrig/xmrig/issues/1105) Improved auto configuration for `cn-pico` algorithm. +- [#1106](https://github.com/xmrig/xmrig/issues/1106) Fixed `hugepages` field in summary API. +- Added alternative short format for CPU threads. +- Changed format for CPU threads with intensity above 1. +- Name for reference RandomX configuration changed to `rx/test` to avoid potential conflicts in future. + +# v2.99.4-beta +- [#1062](https://github.com/xmrig/xmrig/issues/1062) Fixed 32 bit support. **32 bit is slow and deprecated**. +- [#1088](https://github.com/xmrig/xmrig/pull/1088) Fixed macOS compilation. +- [#1095](https://github.com/xmrig/xmrig/pull/1095) Fixed compatibility with hwloc 1.10.x. +- Optimized RandomX initialization and switching, fixed rare crash when re-initialize dataset. +- Fixed ARM build with hwloc. + +# v2.99.3-beta +- [#1082](https://github.com/xmrig/xmrig/issues/1082) Fixed hwloc auto configuration on AMD FX CPUs. +- Added command line option `--export-topology` for export hwloc topology to a XML file. + +# v2.99.2-beta +- [#1077](https://github.com/xmrig/xmrig/issues/1077) Added NUMA support via **hwloc**. +- Fixed miner freeze when switch between RandomX variants. +- Fixed dataset initialization speed on Linux if thread affinity was used. + +# v2.99.1-beta +- [#1072](https://github.com/xmrig/xmrig/issues/1072) Fixed RandomX `seed_hash` re-initialization. + +# v2.99.0-beta +- [#1050](https://github.com/xmrig/xmrig/pull/1050) Added RandomXL algorithm for [Loki](https://loki.network/), algorithm name used by miner is `randomx/loki` or `rx/loki`. +- Added [flexible](https://github.com/xmrig/xmrig/blob/evo/doc/CPU.md) multi algorithm configuration. +- Added unlimited switching between incompatible algorithms, all mining options can be changed in runtime. +- Breaked backward compatibility with previous configs and command line, `variant` option replaced to `algo`, global option `algo` removed, all CPU related settings moved to `cpu` object. +- Options `av`, `safe` and `max-cpu-usage` removed. +- Algorithm `cn/msr` renamed to `cn/fast`. +- Algorithm `cn/xtl` removed. +- API endpoint `GET /1/threads` replaced to `GET /2/backends`. + +# v2.16.0-beta +- [#1036](https://github.com/xmrig/xmrig/pull/1036) Added RandomWOW (RandomX with different preferences) algorithm support for [Wownero](http://wownero.org/). + - Algorithm name used by miner is `randomx/wow` or `rx/wow`. + - Currently runtime algorithm switching NOT supported with other algorithms. + +# v2.15.4-beta +- Added global uptime and extended connection information in API. +- API now return current algorithm instead of global algorithm specified in config. +- This version also include all changes from stable version v2.14.4. + +# v2.15.3-beta +- [#1014](https://github.com/xmrig/xmrig/issues/1014) Fixed regression, default value for `algo` option was not applied. + +# v2.15.2-beta +- [#1010](https://github.com/xmrig/xmrig/pull/1010#issuecomment-482632107) Added daemon support (solo mining). +- [#1012](https://github.com/xmrig/xmrig/pull/1012) Fixed compatibility with clang 9. +- Config subsystem was rewritten, internally JSON is primary format now. +- Fixed regression, big HTTP responses was truncated. + +# v2.15.1-beta +- [#1007](https://github.com/xmrig/xmrig/issues/1007) Old HTTP API backend based on libmicrohttpd, replaced to custom HTTP server (libuv + http_parser). +- [#257](https://github.com/xmrig/xmrig-nvidia/pull/257) New logging subsystem, file and syslog now always without colors. + +# v2.15.0-beta +- [#314](https://github.com/xmrig/xmrig-proxy/issues/314) Added donate over proxy feature. + - Added new option `donate-over-proxy`. + - Added real graceful exit. + +# v2.14.4 +- [#992](https://github.com/xmrig/xmrig/pull/992) Fixed compilation with Clang 3.5. +- [#1012](https://github.com/xmrig/xmrig/pull/1012) Fixed compilation with Clang 9.0. +- In HTTP API for unknown hashrate now used `null` instead of `0.0`. +- Fixed MSVC 2019 version detection. +- Removed obsolete automatic variants. + +# v2.14.1 +* [#975](https://github.com/xmrig/xmrig/issues/975) Fixed crash on Linux if double thread mode used. + +# v2.14.0 +- **[#969](https://github.com/xmrig/xmrig/pull/969) Added new algorithm `cryptonight/rwz`, short alias `cn/rwz` (also known as CryptoNight ReverseWaltz), for upcoming [Graft](https://www.graft.network/) fork.** +- **[#931](https://github.com/xmrig/xmrig/issues/931) Added new algorithm `cryptonight/zls`, short alias `cn/zls` for [Zelerius Network](https://zelerius.org) fork.** +- **[#940](https://github.com/xmrig/xmrig/issues/940) Added new algorithm `cryptonight/double`, short alias `cn/double` (also known as CryptoNight HeavyX), for [X-CASH](https://x-cash.org/).** +- [#951](https://github.com/xmrig/xmrig/issues/951#issuecomment-469581529) Fixed crash if AVX was disabled on OS level. +- [#952](https://github.com/xmrig/xmrig/issues/952) Fixed compile error on some Linux. +- [#957](https://github.com/xmrig/xmrig/issues/957#issuecomment-468890667) Added support for embedded config. +- [#958](https://github.com/xmrig/xmrig/pull/958) Fixed incorrect user agent on ARM platforms. +- [#968](https://github.com/xmrig/xmrig/pull/968) Optimized `cn/r` algorithm performance. + +# v2.13.1 +- [#946](https://github.com/xmrig/xmrig/pull/946) Optimized software AES implementations for CPUs without hardware AES support. `cn/r`, `cn/wow` up to 2.6 times faster, 4-9% improvements for other algorithms. + +# v2.13.0 +- **[#938](https://github.com/xmrig/xmrig/issues/938) Added support for new algorithm `cryptonight/r`, short alias `cn/r` (also known as CryptoNightR or CryptoNight variant 4), for upcoming [Monero](https://www.getmonero.org/) fork on March 9, thanks [@SChernykh](https://github.com/SChernykh).** +- [#939](https://github.com/xmrig/xmrig/issues/939) Added support for dynamic (runtime) pools reload. +- [#932](https://github.com/xmrig/xmrig/issues/932) Fixed `cn-pico` hashrate drop, regression since v2.11.0. + +# v2.12.0 +- [#929](https://github.com/xmrig/xmrig/pull/929) Added support for new algorithm `cryptonight/wow`, short alias `cn/wow` (also known as CryptonightR), for upcoming [Wownero](http://wownero.org) fork on February 14. + +# v2.11.0 +- [#928](https://github.com/xmrig/xmrig/issues/928) Added support for new algorithm `cryptonight/gpu`, short alias `cn/gpu` (original name `cryptonight-gpu`), for upcoming [Ryo currency](https://ryo-currency.com) fork on February 14. +- [#749](https://github.com/xmrig/xmrig/issues/749) Added support for detect hardware AES in runtime on ARMv8 platforms. +- [#292](https://github.com/xmrig/xmrig/issues/292) Fixed build on ARMv8 platforms if compiler not support hardware AES. + +# v2.10.0 +- [#904](https://github.com/xmrig/xmrig/issues/904) Added new algorithm `cn-pico/trtl` (aliases `cryptonight-turtle`, `cn-trtl`) for upcoming TurtleCoin (TRTL) fork. +- Default value for option `max-cpu-usage` changed to `100` also this option now deprecated. + +# v2.9.4 +- [#913](https://github.com/xmrig/xmrig/issues/913) Fixed Masari (MSR) support (this update required for upcoming fork). +- [#915](https://github.com/xmrig/xmrig/pull/915) Improved security, JIT memory now read-only after patching. + +# v2.9.3 +- [#909](https://github.com/xmrig/xmrig/issues/909) Fixed compile errors on FreeBSD. +- [#912](https://github.com/xmrig/xmrig/pull/912) Fixed, C++ implementation of `cn/half` was produce up to 13% of invalid hashes. + +# v2.9.2 +- [#907](https://github.com/xmrig/xmrig/pull/907) Fixed crash on Linux. + +# v2.9.1 +- Restored compatibility with https://stellite.hashvault.pro. + +# v2.9.0 +- [#899](https://github.com/xmrig/xmrig/issues/899) Added support for new algorithm `cn/half` for Masari and Stellite forks. +- [#834](https://github.com/xmrig/xmrig/pull/834) Added ASM optimized code for AMD Bulldozer. +- [#839](https://github.com/xmrig/xmrig/issues/839) Fixed FreeBSD compile. +- [#857](https://github.com/xmrig/xmrig/pull/857) Fixed impossible to build for macOS without clang. + +# v2.8.3 +- [#813](https://github.com/xmrig/xmrig/issues/813) Fixed critical bug with Minergate pool and variant 2. + +# v2.8.1 +- [#768](https://github.com/xmrig/xmrig/issues/768) Fixed build with Visual Studio 2015. +- [#769](https://github.com/xmrig/xmrig/issues/769) Fixed regression, some ANSI escape sequences was in log with disabled colors. +- [#777](https://github.com/xmrig/xmrig/issues/777) Better report about pool connection issues. +- Simplified checks for ASM auto detection, only AES support necessary. +- Added missing options to `--help` output. + +# v2.8.0 +- **[#753](https://github.com/xmrig/xmrig/issues/753) Added new algorithm [CryptoNight variant 2](https://github.com/xmrig/xmrig/issues/753) for Monero fork, thanks [@SChernykh](https://github.com/SChernykh).** + - Added global and per thread option `"asm"` and and command line equivalent. +- **[#758](https://github.com/xmrig/xmrig/issues/758) Added SSL/TLS support for secure connections to pools.** + - Added per pool options `"tls"` and `"tls-fingerprint"` and command line equivalents. +- [#767](https://github.com/xmrig/xmrig/issues/767) Added config autosave feature, same with GPU miners. +- [#245](https://github.com/xmrig/xmrig-proxy/issues/245) Fixed API ID collision when run multiple miners on same machine. +- [#757](https://github.com/xmrig/xmrig/issues/757) Fixed send buffer overflow. + +# v2.6.4 +- [#700](https://github.com/xmrig/xmrig/issues/700) `cryptonight-lite/ipbc` replaced to `cryptonight-heavy/tube` for **Bittube (TUBE)**. +- Added `cryptonight/rto` (cryptonight variant 1 with IPBC/TUBE mod) variant for **Arto (RTO)** coin. +- Added `cryptonight/xao` (original cryptonight with bigger iteration count) variant for **Alloy (XAO)** coin. +- Better variant detection for **nicehash.com** and **minergate.com**. +- [#692](https://github.com/xmrig/xmrig/issues/692) Added support for specify both algorithm and variant via single `algo` option. + +# v2.6.3 +- **Added support for new cryptonight-heavy variant xhv** (`cn-heavy/xhv`) for upcoming Haven Protocol fork. +- **Added support for new cryptonight variant msr** (`cn/msr`) also known as `cryptonight-fast` for upcoming Masari fork. +- Added new detailed hashrate report. +- [#446](https://github.com/xmrig/xmrig/issues/446) Likely fixed SIGBUS error on 32 bit ARM CPUs. +- [#551](https://github.com/xmrig/xmrig/issues/551) Fixed `cn-heavy` algorithm on ARMv8. +- [#614](https://github.com/xmrig/xmrig/issues/614) Fixed display issue with huge pages percentage when colors disabled. +- [#615](https://github.com/xmrig/xmrig/issues/615) Fixed build without libcpuid. +- [#629](https://github.com/xmrig/xmrig/pull/629) Fixed file logging with non-seekable files. +- [#672](https://github.com/xmrig/xmrig/pull/672) Reverted back `cryptonight-light` and exit if no valid algorithm specified. + +# v2.6.2 + - [#607](https://github.com/xmrig/xmrig/issues/607) Fixed donation bug. + - [#610](https://github.com/xmrig/xmrig/issues/610) Fixed ARM build. + +# v2.6.1 + - [#168](https://github.com/xmrig/xmrig-proxy/issues/168) Added support for [mining algorithm negotiation](https://github.com/xmrig/xmrig-proxy/blob/dev/doc/STRATUM_EXT.md#1-mining-algorithm-negotiation). + - Added IPBC coin support, base algorithm `cn-lite` variant `ipbc`. + - [#581](https://github.com/xmrig/xmrig/issues/581) Added support for upcoming Stellite (XTL) fork, base algorithm `cn` variant `xtl`, variant can set now, no need do it after fork. + - Added support for **rig-id** stratum protocol extensions, compatible with xmr-stak. + - Changed behavior for option `variant=-1` for `cryptonight`, now variant is `1` by default, if you mine old coins need change `variant` to `0`. + - A lot of small fixes and better unification with proxy code. + +# v2.6.0-beta3 +- [#563](https://github.com/xmrig/xmrig/issues/563) **Added [advanced threads mode](https://github.com/xmrig/xmrig/issues/563), now possible configure each thread individually.** +- [#255](https://github.com/xmrig/xmrig/issues/563) Low power mode extended to **triple**, **quard** and **penta** modes. +- [#519](https://github.com/xmrig/xmrig/issues/519) Fixed high donation levels, improved donation start time randomization. +- [#554](https://github.com/xmrig/xmrig/issues/554) Fixed regression with `print-time` option. + +# v2.6.0-beta2 +- Improved performance for `cryptonight v7` especially in double hash mode. +- [#499](https://github.com/xmrig/xmrig/issues/499) IPv6 disabled for internal HTTP API by default, was causing issues on some systems. +- Added short aliases for algorithm names: `cn`, `cn-lite` and `cn-heavy`. +- Fixed regressions (v2.6.0-beta1 affected) + - [#494](https://github.com/xmrig/xmrig/issues/494) Command line option `--donate-level` was broken. + - [#502](https://github.com/xmrig/xmrig/issues/502) Build without libmicrohttpd was broken. + - Fixed nonce calculation for `--av 4` (software AES, double hash) was causing reduction of effective hashrate and rejected shares on nicehash. + +# v2.6.0-beta1 + - [#476](https://github.com/xmrig/xmrig/issues/476) **Added Cryptonight-Heavy support for Sumokoin ASIC resistance fork.** + - HTTP server now runs in main loop, it make possible easy extend API without worry about thread synchronization. + - Added initial graceful reload support, miner will reload configuration if config file changed, disabled by default until it will be fully implemented and tested. + - Added API endpoint `PUT /1/config` to update current config. + - Added API endpoint `GET /1/config` to get current active config. + - Added API endpoint `GET /1/threads` to get current active threads configuration. + - API endpoint `GET /` now deprecated, use `GET /1/summary` instead. + - Added `--api-no-ipv6` and similar config option to disable IPv6 support for HTTP API. + - Added `--api-no-restricted` to enable full access to api, this option has no effect if `--api-access-token` not specified. + +# v2.5.3 +- Fixed critical bug, in some cases miner was can't recovery connection and switch to failover pool, version 2.5.2 affected. If you use v2.6.0-beta3 this issue doesn't concern you. +- [#499](https://github.com/xmrig/xmrig/issues/499) IPv6 support disabled for internal HTTP API. +- Added workaround for nicehash.com if you use `cryptonightv7..nicehash.com` option `variant=1` will be set automatically. + +# v2.5.2 +- [#448](https://github.com/xmrig/xmrig/issues/478) Fixed broken reconnect. + +# v2.5.1 +- [#454](https://github.com/xmrig/xmrig/issues/454) Fixed build with libmicrohttpd version below v0.9.35. +- [#456](https://github.com/xmrig/xmrig/issues/459) Verbose errors related to donation pool was not fully silenced. +- [#459](https://github.com/xmrig/xmrig/issues/459) Fixed regression (version 2.5.0 affected) with connection to **xmr.f2pool.com**. + +# v2.5.0 +- [#434](https://github.com/xmrig/xmrig/issues/434) **Added support for Monero v7 PoW, scheduled on April 6.** +- Added full IPv6 support. +- Added protocol extension, when use the miner with xmrig-proxy 2.5+ no more need manually specify `nicehash` option. +- [#123](https://github.com/xmrig/xmrig-proxy/issues/123) Fixed regression (all versions since 2.4 affected) fragmented responses from pool/proxy was parsed incorrectly. +- [#428](https://github.com/xmrig/xmrig/issues/428) Fixed regression (version 2.4.5 affected) with CPU cache size detection. + +# v2.4.5 +- [#324](https://github.com/xmrig/xmrig/pull/324) Fixed build without libmicrohttpd (CMake cache issue). +- [#341](https://github.com/xmrig/xmrig/issues/341) Fixed wrong exit code and added command line option `--dry-run`. +- [#385](https://github.com/xmrig/xmrig/pull/385) Up to 20% performance increase for non-AES CPU and fixed Intel Core 2 cache detection. + +# v2.4.4 + - Added libmicrohttpd version to --version output. + - Fixed bug in singal handler, in some cases miner wasn't shutdown properly. + - Fixed recent MSVC 2017 version detection. + - [#279](https://github.com/xmrig/xmrig/pull/279) Fixed build on some macOS versions. + +# v2.4.3 + - [#94](https://github.com/xmrig/xmrig/issues/94#issuecomment-342019257) [#216](https://github.com/xmrig/xmrig/issues/216) Added **ARMv8** and **ARMv7** support. Hardware AES supported, thanks [Imran Yusuff](https://github.com/imranyusuff). + - [#157](https://github.com/xmrig/xmrig/issues/157) [#196](https://github.com/xmrig/xmrig/issues/196) Fixed Linux compile issues. + - [#184](https://github.com/xmrig/xmrig/issues/184) Fixed cache size detection for CPUs with disabled Hyper-Threading. + - [#200](https://github.com/xmrig/xmrig/issues/200) In some cases miner was doesn't write log to stdout. + +# v2.4.2 + - [#60](https://github.com/xmrig/xmrig/issues/60) Added FreeBSD support, thanks [vcambur](https://github.com/vcambur). + - [#153](https://github.com/xmrig/xmrig/issues/153) Fixed issues with dwarfpool.com. + +# v2.4.1 + - [#147](https://github.com/xmrig/xmrig/issues/147) Fixed comparability with monero-stratum. + +# v2.4.0 + - Added [HTTP API](https://github.com/xmrig/xmrig/wiki/API). + - Added comments support in config file. + - libjansson replaced to rapidjson. + - [#98](https://github.com/xmrig/xmrig/issues/98) Ignore `keepalive` option with minergate.com and nicehash.com. + - [#101](https://github.com/xmrig/xmrig/issues/101) Fixed MSVC 2017 (15.3) compile time version detection. + - [#108](https://github.com/xmrig/xmrig/issues/108) Silently ignore invalid values for `donate-level` option. + - [#111](https://github.com/xmrig/xmrig/issues/111) Fixed build without AEON support. + +# v2.3.1 +- [#68](https://github.com/xmrig/xmrig/issues/68) Fixed compatibility with Docker containers, was nothing print on console. + +# v2.3.0 +- Added `--cpu-priority` option (0 idle, 2 normal to 5 highest). +- Added `--user-agent` option, to set custom user-agent string for pool. For example `cpuminer-multi/0.1`. +- Added `--no-huge-pages` option, to disable huge pages support. +- [#62](https://github.com/xmrig/xmrig/issues/62) Don't send the login to the dev pool. +- Force reconnect if pool block miner IP address. helps switch to backup pool. +- Fixed: failed open default config file if path contains non English characters. +- Fixed: error occurred if try use unavailable stdin or stdout, regression since version 2.2.0. +- Fixed: message about huge pages support successfully enabled on Windows was not shown in release builds. + +# v2.2.1 +- Fixed [terminal issues](https://github.com/xmrig/xmrig-proxy/issues/2#issuecomment-319914085) after exit on Linux and OS X. + +# v2.2.0 +- [#46](https://github.com/xmrig/xmrig/issues/46) Restored config file support. Now possible use multiple config files and combine with command line options also added support for default config. +- Improved colors support on Windows, now used uv_tty, legacy code removed. +- QuickEdit Mode now disabled on Windows. +- Added interactive commands in console window:: **h**ashrate, **p**ause, **r**esume. +- Fixed autoconf mode for AMD FX CPUs. + +# v2.1.0 +- [#40](https://github.com/xmrig/xmrig/issues/40) +Improved miner shutdown, fixed crash on exit for Linux and OS X. +- Fixed, login request was contain malformed JSON if username or password has some special characters for example `\`. +- [#220](https://github.com/fireice-uk/xmr-stak-cpu/pull/220) Better support for Round Robin DNS, IP address now always chosen randomly instead of stuck on first one. +- Changed donation address, new [xmrig-proxy](https://github.com/xmrig/xmrig-proxy) is coming soon. + +# v2.0.2 +- Better deal with possible duplicate jobs from pool, show warning and ignore duplicates. +- For Windows builds libuv updated to version 1.13.1 and gcc to 7.1.0. + +# v2.0.1 + - [#27](https://github.com/xmrig/xmrig/issues/27) Fixed possibility crash on 32bit systems. + +# v2.0.0 + - Option `--backup-url` removed, instead now possibility specify multiple pools for example: `-o example1.com:3333 -u user1 -p password1 -k -o example2.com:5555 -u user2 -o example3.com:4444 -u user3` + - [#15](https://github.com/xmrig/xmrig/issues/15) Added option `-l, --log-file=FILE` to write log to file. + - [#15](https://github.com/xmrig/xmrig/issues/15) Added option `-S, --syslog` to use syslog for logging, Linux only. + - [#18](https://github.com/xmrig/xmrig/issues/18) Added nice messages for accepted/rejected shares with diff and network latency. + - [#20](https://github.com/xmrig/xmrig/issues/20) Fixed `--cpu-affinity` for more than 32 threads. + - Fixed Windows XP support. + - Fixed regression, option `--no-color` was not fully disable colored output. + - Show resolved pool IP address in miner output. + +# v1.0.1 +- Fix broken software AES implementation, app has crashed if CPU not support AES-NI, only version 1.0.0 affected. + +# v1.0.0 +- Miner complete rewritten in C++ with libuv. +- This version should be fully compatible (except config file) with previos versions, many new nice features will come in next versions. +- This is still beta. If you found regression, stability or perfomance issues or have an idea for new feature please fell free to open new [issue](https://github.com/xmrig/xmrig/issues/new). +- Added new option `--print-time=N`, print hashrate report every N seconds. +- New hashrate reports, by default every 60 secons. +- Added Microsoft Visual C++ 2015 and 2017 support. +- Removed dependency on libcurl. +- To compile this version from source please switch to [dev](https://github.com/xmrig/xmrig/tree/dev) branch. + +# v0.8.2 +- Fixed L2 cache size detection for AMD CPUs (Bulldozer/Piledriver/Steamroller/Excavator architecture). + +# v0.8.2 +- Fixed L2 cache size detection for AMD CPUs (Bulldozer/Piledriver/Steamroller/Excavator architecture). +- Fixed gcc 7.1 support. + +# v0.8.1 +- Added nicehash support, detects automaticaly by pool URL, for example `cryptonight.eu.nicehash.com:3355` or manually via option `--nicehash`. + +# v0.8.0 +- Added double hash mode, also known as lower power mode. `--av=2` and `--av=4`. +- Added smart automatic CPU configuration. Default threads count now depends on size of the L3 cache of CPU. +- Added CryptoNight-Lite support for AEON `-a cryptonight-lite`. +- Added `--max-cpu-usage` option for auto CPU configuration mode. +- Added `--safe` option for adjust threads and algorithm variations to current CPU. +- No more manual steps to enable huge pages on Windows. XMRig will do it automatically. +- Removed BMI2 algorithm variation. +- Removed default pool URL. + +# v0.6.0 +- Added automatic cryptonight self test. +- New software AES algorithm variation. Will be automatically selected if cpu not support AES-NI. +- Added 32 bit builds. +- Documented [algorithm variations](https://github.com/xmrig/xmrig#algorithm-variations). + +# v0.5.0 +- Initial public release. diff --git a/doc/CPU.md b/doc/CPU.md new file mode 100644 index 000000000..756d3a0ca --- /dev/null +++ b/doc/CPU.md @@ -0,0 +1,95 @@ +# CPU backend + +**Information in this document actual to version 2.99.5+** + +All CPU related settings contains in one `cpu` object in config file, CPU backend allow specify multiple profiles and allow switch between them without restrictions by pool request or config change. Default auto-configuration create reasonable minimum of profiles which cover all supported algorithms. + +### Example + +Example below demonstrate all primary ideas of flexible profiles configuration: + +* `"rx/wow"` Exact match to algorithm `rx/wow`, defined 4 threads without CPU affinity. +* `"cn"` Default failback profile for all `cn/*` algorithms, defined 2 threads with CPU affinity, another failback profiles is `cn-lite`, `cn-heavy` and `rx`. +* `"cn-lite"` Default failback profile for all `cn-lite/*` algorithms, defined 2 double threads with CPU affinity. +* `"cn-pico"` Alternative short object format. +* `"custom-profile"` Custom user defined profile. +* `"*"` Failback profile for all unhandled by other profiles algorithms. +* `"cn/r"` Exact match, alias to profile `custom-profile`. +* `"cn/0"` Exact match, disabled algorithm. + +```json +{ + "cpu": { + "enabled": true, + "huge-pages": true, + "hw-aes": null, + "priority": null, + "asm": true, + "rx/wow": [-1, -1, -1, -1], + "cn": [ + [1, 0], + [1, 2] + ], + "cn-lite": [ + [2, 0], + [2, 2] + ], + "cn-pico": { + "intensity": 2, + "threads": 8, + "affinity": -1 + }, + "custom-profile": [0, 2], + "*": [-1], + "cn/r": "custom-profile", + "cn/0": false + } +} +``` + +## Threads definition +Threads can be defined in 3 formats. + +#### Array format +```json +[ + [1, 0], + [1, 2], + [1, -1], + [2, -1] +] +``` +Each line represent one thread, first element is intensity, this option was known as `low_power_mode`, possible values is range from 1 to 5, second element is CPU affinity, special value `-1` means no affinity. + +#### Short array format +```json +[-1, -1, -1, -1] +``` +Each number represent one thread and means CPU affinity, this is default format for algorithm with maximum intensity 1, currently it all RandomX variants and cryptonight-gpu. + +#### Short object format +```json +{ + "intensity": 2, + "threads": 8, + "affinity": -1 +} +``` +Internal format, but can be user defined. + +## Shared options + +#### `enabled` +Enable (`true`) or disable (`false`) CPU backend, by default `true`. + +#### `huge-pages` +Enable (`true`) or disable (`false`) huge pages support, by default `true`. + +#### `hw-aes` +Force enable (`true`) or disable (`false`) hardware AES support. Default value `null` means miner autodetect this feature. Usually don't need change this option, this option useful for some rare cases when miner can't detect hardware AES, but it available. If you force enable this option, but your hardware not support it, miner will crash. + +#### `priority` +Mining threads priority, value from `1` (lowest priority) to `5` (highest possible priority). Default value `null` means miner don't change threads priority at all. + +#### `asm` +Enable/configure or disable ASM optimizations. Possible values: `true`, `false`, `"intel"`, `"ryzen"`, `"bulldozer"`. diff --git a/doc/data/algorithms.json b/doc/data/algorithms.json new file mode 100644 index 000000000..b42fa82f3 --- /dev/null +++ b/doc/data/algorithms.json @@ -0,0 +1,32 @@ +{ + "current": [ + ["rx/test", "2 MB", "2.99.5+", "RandomX (reference configuration)."], + ["rx/0", "2 MB", "2.99.0+", "RandomX (reference configuration), reserved for future use."], + ["rx/wow", "1 MB", "2.99.0+", "RandomWOW."], + ["rx/loki", "2 MB", "2.99.0+", "RandomXL."], + ["cn/fast", "2 MB", "2.99.0+", "CryptoNight variant 1 with half iterations."], + ["cn/rwz", "2 MB", "2.14.0+", "CryptoNight variant 2 with 3/4 iterations and reversed shuffle operation."], + ["cn/zls", "2 MB", "2.14.0+", "CryptoNight variant 2 with 3/4 iterations."], + ["cn/double", "2 MB", "2.14.0+", "CryptoNight variant 2 with double iterations."], + ["cn/r", "2 MB", "2.13.0+", "CryptoNightR (Monero's variant 4)."], + ["cn/wow", "2 MB", "2.12.0+", "CryptoNightR (Wownero)."], + ["cn/gpu", "2 MB", "2.11.0+", "CryptoNight-GPU."], + ["cn-pico", "256 KB", "2.10.0+", "CryptoNight-Pico."], + ["cn/half", "2 MB", "2.9.0+", "CryptoNight variant 2 with half iterations."], + ["cn/2", "2 MB", "2.8.0+", "CryptoNight variant 2."], + ["cn/xao", "2 MB", "2.6.4+", "CryptoNight variant 0 (modified)."], + ["cn/rto", "2 MB", "2.6.4+", "CryptoNight variant 1 (modified)."], + ["cn-heavy/tube", "4 MB", "2.6.4+", "CryptoNight-Heavy (modified)."], + ["cn-heavy/xhv", "4 MB", "2.6.3+", "CryptoNight-Heavy (modified)."], + ["cn-heavy/0", "4 MB", "2.6.0+", "CryptoNight-Heavy."], + ["cn/1", "2 MB", "2.5.0+", "CryptoNight variant 1."], + ["cn-lite/1", "1 MB", "2.5.0+", "CryptoNight-Lite variant 1."], + ["cn-lite/0", "1 MB", "0.8.0+", "CryptoNight-Lite variant 0."], + ["cn/0", "2 MB", "0.5.0+", "CryptoNight (original)."] + ], + "removed": [ + ["cn/msr", "2 MB", "2.6.3+", "Renamed to cn/fast, still supported as alias."], + ["cn/xtl", "2 MB", "2.6.1-2.16.0", "Coin forked to cn/half."], + ["cn-lite/ipbc", "1 MB", "2.6.1-2.6.3", "Coin forked to cn-heavy/tube."] + ] +} \ No newline at end of file diff --git a/doc/screenshot.png b/doc/screenshot.png new file mode 100644 index 000000000..28e1e3a1c Binary files /dev/null and b/doc/screenshot.png differ diff --git a/doc/topology/AMD_FX_8320_windows_2_0_4.xml b/doc/topology/AMD_FX_8320_windows_2_0_4.xml new file mode 100644 index 000000000..55fe6b5d0 --- /dev/null +++ b/doc/topology/AMD_FX_8320_windows_2_0_4.xml @@ -0,0 +1,86 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/doc/topology/AMD_Opteron_6272_x4_N8_linux_2_0_4_LXC.xml b/doc/topology/AMD_Opteron_6272_x4_N8_linux_2_0_4_LXC.xml new file mode 100644 index 000000000..39576bb46 --- /dev/null +++ b/doc/topology/AMD_Opteron_6272_x4_N8_linux_2_0_4_LXC.xml @@ -0,0 +1,234 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + 0 1 2 3 4 5 6 7 + 10 16 16 22 16 22 16 22 16 10 + 22 16 22 16 22 16 16 22 10 16 + 16 22 16 22 22 16 16 10 22 16 + 22 16 16 22 16 22 10 16 16 22 + 22 16 22 16 16 10 22 16 16 22 + 16 22 16 22 10 16 22 16 22 16 + 22 16 16 10 + + \ No newline at end of file diff --git a/doc/topology/AMD_Opteron_6278_x2_UMA_windows_2_0_4.xml b/doc/topology/AMD_Opteron_6278_x2_UMA_windows_2_0_4.xml new file mode 100644 index 000000000..b59f773cd --- /dev/null +++ b/doc/topology/AMD_Opteron_6278_x2_UMA_windows_2_0_4.xml @@ -0,0 +1,294 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/doc/topology/AMD_Opteron_6348_x4_N8_linux_1_11_2.xml b/doc/topology/AMD_Opteron_6348_x4_N8_linux_1_11_2.xml new file mode 100644 index 000000000..2d889819f --- /dev/null +++ b/doc/topology/AMD_Opteron_6348_x4_N8_linux_1_11_2.xml @@ -0,0 +1,550 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/doc/topology/AMD_Opteron_6380_x4_N8_linux_1_11_5.xml b/doc/topology/AMD_Opteron_6380_x4_N8_linux_1_11_5.xml new file mode 100644 index 000000000..2ecbe3cbb --- /dev/null +++ b/doc/topology/AMD_Opteron_6380_x4_N8_linux_1_11_5.xml @@ -0,0 +1,670 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/doc/topology/AMD_Ryzen_7_2700X_windows_2_0_4.xml b/doc/topology/AMD_Ryzen_7_2700X_windows_2_0_4.xml new file mode 100644 index 000000000..e3ecb6fda --- /dev/null +++ b/doc/topology/AMD_Ryzen_7_2700X_windows_2_0_4.xml @@ -0,0 +1,105 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/doc/topology/AMD_Ryzen_7_3700X_windows_2_0_4.xml b/doc/topology/AMD_Ryzen_7_3700X_windows_2_0_4.xml new file mode 100644 index 000000000..1f2d0ee49 --- /dev/null +++ b/doc/topology/AMD_Ryzen_7_3700X_windows_2_0_4.xml @@ -0,0 +1,104 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/doc/topology/AMD_Ryzen_Threadripper_2950X_N2_linux_2_0_4.xml b/doc/topology/AMD_Ryzen_Threadripper_2950X_N2_linux_2_0_4.xml new file mode 100644 index 000000000..c168e2a0f --- /dev/null +++ b/doc/topology/AMD_Ryzen_Threadripper_2950X_N2_linux_2_0_4.xml @@ -0,0 +1,226 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + 0 1 + 10 16 16 10 + + diff --git a/doc/topology/AMD_Ryzen_Threadripper_2950X_UMA_linux_1_11_9.xml b/doc/topology/AMD_Ryzen_Threadripper_2950X_UMA_linux_1_11_9.xml new file mode 100644 index 000000000..ed3776c00 --- /dev/null +++ b/doc/topology/AMD_Ryzen_Threadripper_2950X_UMA_linux_1_11_9.xml @@ -0,0 +1,328 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/doc/topology/Intel_Core_i7-3770_linux_2_0_4.xml b/doc/topology/Intel_Core_i7-3770_linux_2_0_4.xml new file mode 100644 index 000000000..18c802103 --- /dev/null +++ b/doc/topology/Intel_Core_i7-3770_linux_2_0_4.xml @@ -0,0 +1,87 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/doc/topology/Intel_Core_i7-6700_linux_2_0_4.xml b/doc/topology/Intel_Core_i7-6700_linux_2_0_4.xml new file mode 100644 index 000000000..c80c74031 --- /dev/null +++ b/doc/topology/Intel_Core_i7-6700_linux_2_0_4.xml @@ -0,0 +1,88 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/doc/topology/Intel_Core_i7-6700_windows_2_0_4.xml b/doc/topology/Intel_Core_i7-6700_windows_2_0_4.xml new file mode 100644 index 000000000..dd3c201c5 --- /dev/null +++ b/doc/topology/Intel_Core_i7-6700_windows_2_0_4.xml @@ -0,0 +1,61 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/doc/topology/Intel_Core_i7-7660U_windows_2_0_4.xml b/doc/topology/Intel_Core_i7-7660U_windows_2_0_4.xml new file mode 100644 index 000000000..01a29e86a --- /dev/null +++ b/doc/topology/Intel_Core_i7-7660U_windows_2_0_4.xml @@ -0,0 +1,41 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/doc/topology/Intel_Xeon_E5-4650_0_x4_N4_windows_2_0_4.xml b/doc/topology/Intel_Xeon_E5-4650_0_x4_N4_windows_2_0_4.xml new file mode 100644 index 000000000..7811a49fc --- /dev/null +++ b/doc/topology/Intel_Xeon_E5-4650_0_x4_N4_windows_2_0_4.xml @@ -0,0 +1,477 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/doc/topology/Intel_Xeon_E5620_x2_UMA_windows_2_0_4.xml b/doc/topology/Intel_Xeon_E5620_x2_UMA_windows_2_0_4.xml new file mode 100644 index 000000000..9dad397ab --- /dev/null +++ b/doc/topology/Intel_Xeon_E5620_x2_UMA_windows_2_0_4.xml @@ -0,0 +1,111 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/doc/topology/Intel_Xeon_E7-4870_x4_N4_windows_2_0_4.xml b/doc/topology/Intel_Xeon_E7-4870_x4_N4_windows_2_0_4.xml new file mode 100644 index 000000000..3d0a67367 --- /dev/null +++ b/doc/topology/Intel_Xeon_E7-4870_x4_N4_windows_2_0_4.xml @@ -0,0 +1,541 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/doc/topology/Intel_Xeon_Gold_6146_x2_UMA_linux_2_0_4.xml b/doc/topology/Intel_Xeon_Gold_6146_x2_UMA_linux_2_0_4.xml new file mode 100644 index 000000000..fe94194c4 --- /dev/null +++ b/doc/topology/Intel_Xeon_Gold_6146_x2_UMA_linux_2_0_4.xml @@ -0,0 +1,246 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/doc/topology/Intel_Xeon_Silver_4114_x2_N2_linux_1_11_9.xml b/doc/topology/Intel_Xeon_Silver_4114_x2_N2_linux_1_11_9.xml new file mode 100644 index 000000000..0dadfed22 --- /dev/null +++ b/doc/topology/Intel_Xeon_Silver_4114_x2_N2_linux_1_11_9.xml @@ -0,0 +1,403 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/doc/topology/Intel_Xeon_Silver_4114_x2_N2_linux_2_0_4.xml b/doc/topology/Intel_Xeon_Silver_4114_x2_N2_linux_2_0_4.xml new file mode 100644 index 000000000..fd56a10c4 --- /dev/null +++ b/doc/topology/Intel_Xeon_Silver_4114_x2_N2_linux_2_0_4.xml @@ -0,0 +1,263 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + 0 1 + 10 21 21 10 + + diff --git a/res/app.rc b/res/app.rc index 037d842ae..fb41b8b1d 100644 --- a/res/app.rc +++ b/res/app.rc @@ -1,7 +1,7 @@ #include #include "../src/version.h" -IDI_ICON1 ICON DISCARDABLE "app.ico" +101 ICON "app.ico" VS_VERSION_INFO VERSIONINFO FILEVERSION APP_VER_MAJOR,APP_VER_MINOR,APP_VER_PATCH,0 diff --git a/src/3rdparty/aligned_malloc.h b/src/3rdparty/aligned_malloc.h deleted file mode 100644 index 0b74b17e0..000000000 --- a/src/3rdparty/aligned_malloc.h +++ /dev/null @@ -1,65 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2016-2017 XMRig - * - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef __ALIGNED_MALLOC_H__ -#define __ALIGNED_MALLOC_H__ - - -#include - - -#ifndef __cplusplus -extern int posix_memalign(void **__memptr, size_t __alignment, size_t __size); -#else -// Some systems (e.g. those with GNU libc) declare posix_memalign with an -// exception specifier. Via an "egregious workaround" in -// Sema::CheckEquivalentExceptionSpec, Clang accepts the following as a valid -// redeclaration of glibc's declaration. -extern "C" int posix_memalign(void **__memptr, size_t __alignment, size_t __size); -#endif - - -static __inline__ void *__attribute__((__always_inline__, __malloc__)) _mm_malloc(size_t __size, size_t __align) -{ - if (__align == 1) { - return malloc(__size); - } - - if (!(__align & (__align - 1)) && __align < sizeof(void *)) - __align = sizeof(void *); - - void *__mallocedMemory; - if (posix_memalign(&__mallocedMemory, __align, __size)) { - return 0; - } - - return __mallocedMemory; -} - - -static __inline__ void __attribute__((__always_inline__)) _mm_free(void *__p) -{ - free(__p); -} - -#endif /* __ALIGNED_MALLOC_H__ */ diff --git a/src/3rdparty/http-parser/AUTHORS b/src/3rdparty/http-parser/AUTHORS new file mode 100644 index 000000000..5323b685c --- /dev/null +++ b/src/3rdparty/http-parser/AUTHORS @@ -0,0 +1,68 @@ +# Authors ordered by first contribution. +Ryan Dahl +Jeremy Hinegardner +Sergey Shepelev +Joe Damato +tomika +Phoenix Sol +Cliff Frey +Ewen Cheslack-Postava +Santiago Gala +Tim Becker +Jeff Terrace +Ben Noordhuis +Nathan Rajlich +Mark Nottingham +Aman Gupta +Tim Becker +Sean Cunningham +Peter Griess +Salman Haq +Cliff Frey +Jon Kolb +Fouad Mardini +Paul Querna +Felix Geisendörfer +koichik +Andre Caron +Ivo Raisr +James McLaughlin +David Gwynne +Thomas LE ROUX +Randy Rizun +Andre Louis Caron +Simon Zimmermann +Erik Dubbelboer +Martell Malone +Bertrand Paquet +BogDan Vatra +Peter Faiman +Corey Richardson +Tóth Tamás +Cam Swords +Chris Dickinson +Uli Köhler +Charlie Somerville +Patrik Stutz +Fedor Indutny +runner +Alexis Campailla +David Wragg +Vinnie Falco +Alex Butum +Rex Feng +Alex Kocharin +Mark Koopman +Helge Heß +Alexis La Goutte +George Miroshnykov +Maciej Małecki +Marc O'Morain +Jeff Pinner +Timothy J Fontaine +Akagi201 +Romain Giraud +Jay Satiro +Arne Steen +Kjell Schubert +Olivier Mengué diff --git a/src/3rdparty/http-parser/LICENSE-MIT b/src/3rdparty/http-parser/LICENSE-MIT new file mode 100644 index 000000000..1ec0ab4e1 --- /dev/null +++ b/src/3rdparty/http-parser/LICENSE-MIT @@ -0,0 +1,19 @@ +Copyright Joyent, Inc. and other Node contributors. + +Permission is hereby granted, free of charge, to any person obtaining a copy +of this software and associated documentation files (the "Software"), to +deal in the Software without restriction, including without limitation the +rights to use, copy, modify, merge, publish, distribute, sublicense, and/or +sell copies of the Software, and to permit persons to whom the Software is +furnished to do so, subject to the following conditions: + +The above copyright notice and this permission notice shall be included in +all copies or substantial portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING +FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS +IN THE SOFTWARE. diff --git a/src/3rdparty/http-parser/README.md b/src/3rdparty/http-parser/README.md new file mode 100644 index 000000000..b265d7171 --- /dev/null +++ b/src/3rdparty/http-parser/README.md @@ -0,0 +1,246 @@ +HTTP Parser +=========== + +[![Build Status](https://api.travis-ci.org/nodejs/http-parser.svg?branch=master)](https://travis-ci.org/nodejs/http-parser) + +This is a parser for HTTP messages written in C. It parses both requests and +responses. The parser is designed to be used in performance HTTP +applications. It does not make any syscalls nor allocations, it does not +buffer data, it can be interrupted at anytime. Depending on your +architecture, it only requires about 40 bytes of data per message +stream (in a web server that is per connection). + +Features: + + * No dependencies + * Handles persistent streams (keep-alive). + * Decodes chunked encoding. + * Upgrade support + * Defends against buffer overflow attacks. + +The parser extracts the following information from HTTP messages: + + * Header fields and values + * Content-Length + * Request method + * Response status code + * Transfer-Encoding + * HTTP version + * Request URL + * Message body + + +Usage +----- + +One `http_parser` object is used per TCP connection. Initialize the struct +using `http_parser_init()` and set the callbacks. That might look something +like this for a request parser: +```c +http_parser_settings settings; +settings.on_url = my_url_callback; +settings.on_header_field = my_header_field_callback; +/* ... */ + +http_parser *parser = malloc(sizeof(http_parser)); +http_parser_init(parser, HTTP_REQUEST); +parser->data = my_socket; +``` + +When data is received on the socket execute the parser and check for errors. + +```c +size_t len = 80*1024, nparsed; +char buf[len]; +ssize_t recved; + +recved = recv(fd, buf, len, 0); + +if (recved < 0) { + /* Handle error. */ +} + +/* Start up / continue the parser. + * Note we pass recved==0 to signal that EOF has been received. + */ +nparsed = http_parser_execute(parser, &settings, buf, recved); + +if (parser->upgrade) { + /* handle new protocol */ +} else if (nparsed != recved) { + /* Handle error. Usually just close the connection. */ +} +``` + +`http_parser` needs to know where the end of the stream is. For example, sometimes +servers send responses without Content-Length and expect the client to +consume input (for the body) until EOF. To tell `http_parser` about EOF, give +`0` as the fourth parameter to `http_parser_execute()`. Callbacks and errors +can still be encountered during an EOF, so one must still be prepared +to receive them. + +Scalar valued message information such as `status_code`, `method`, and the +HTTP version are stored in the parser structure. This data is only +temporally stored in `http_parser` and gets reset on each new message. If +this information is needed later, copy it out of the structure during the +`headers_complete` callback. + +The parser decodes the transfer-encoding for both requests and responses +transparently. That is, a chunked encoding is decoded before being sent to +the on_body callback. + + +The Special Problem of Upgrade +------------------------------ + +`http_parser` supports upgrading the connection to a different protocol. An +increasingly common example of this is the WebSocket protocol which sends +a request like + + GET /demo HTTP/1.1 + Upgrade: WebSocket + Connection: Upgrade + Host: example.com + Origin: http://example.com + WebSocket-Protocol: sample + +followed by non-HTTP data. + +(See [RFC6455](https://tools.ietf.org/html/rfc6455) for more information the +WebSocket protocol.) + +To support this, the parser will treat this as a normal HTTP message without a +body, issuing both on_headers_complete and on_message_complete callbacks. However +http_parser_execute() will stop parsing at the end of the headers and return. + +The user is expected to check if `parser->upgrade` has been set to 1 after +`http_parser_execute()` returns. Non-HTTP data begins at the buffer supplied +offset by the return value of `http_parser_execute()`. + + +Callbacks +--------- + +During the `http_parser_execute()` call, the callbacks set in +`http_parser_settings` will be executed. The parser maintains state and +never looks behind, so buffering the data is not necessary. If you need to +save certain data for later usage, you can do that from the callbacks. + +There are two types of callbacks: + +* notification `typedef int (*http_cb) (http_parser*);` + Callbacks: on_message_begin, on_headers_complete, on_message_complete. +* data `typedef int (*http_data_cb) (http_parser*, const char *at, size_t length);` + Callbacks: (requests only) on_url, + (common) on_header_field, on_header_value, on_body; + +Callbacks must return 0 on success. Returning a non-zero value indicates +error to the parser, making it exit immediately. + +For cases where it is necessary to pass local information to/from a callback, +the `http_parser` object's `data` field can be used. +An example of such a case is when using threads to handle a socket connection, +parse a request, and then give a response over that socket. By instantiation +of a thread-local struct containing relevant data (e.g. accepted socket, +allocated memory for callbacks to write into, etc), a parser's callbacks are +able to communicate data between the scope of the thread and the scope of the +callback in a threadsafe manner. This allows `http_parser` to be used in +multi-threaded contexts. + +Example: +```c + typedef struct { + socket_t sock; + void* buffer; + int buf_len; + } custom_data_t; + + +int my_url_callback(http_parser* parser, const char *at, size_t length) { + /* access to thread local custom_data_t struct. + Use this access save parsed data for later use into thread local + buffer, or communicate over socket + */ + parser->data; + ... + return 0; +} + +... + +void http_parser_thread(socket_t sock) { + int nparsed = 0; + /* allocate memory for user data */ + custom_data_t *my_data = malloc(sizeof(custom_data_t)); + + /* some information for use by callbacks. + * achieves thread -> callback information flow */ + my_data->sock = sock; + + /* instantiate a thread-local parser */ + http_parser *parser = malloc(sizeof(http_parser)); + http_parser_init(parser, HTTP_REQUEST); /* initialise parser */ + /* this custom data reference is accessible through the reference to the + parser supplied to callback functions */ + parser->data = my_data; + + http_parser_settings settings; /* set up callbacks */ + settings.on_url = my_url_callback; + + /* execute parser */ + nparsed = http_parser_execute(parser, &settings, buf, recved); + + ... + /* parsed information copied from callback. + can now perform action on data copied into thread-local memory from callbacks. + achieves callback -> thread information flow */ + my_data->buffer; + ... +} + +``` + +In case you parse HTTP message in chunks (i.e. `read()` request line +from socket, parse, read half headers, parse, etc) your data callbacks +may be called more than once. `http_parser` guarantees that data pointer is only +valid for the lifetime of callback. You can also `read()` into a heap allocated +buffer to avoid copying memory around if this fits your application. + +Reading headers may be a tricky task if you read/parse headers partially. +Basically, you need to remember whether last header callback was field or value +and apply the following logic: + + (on_header_field and on_header_value shortened to on_h_*) + ------------------------ ------------ -------------------------------------------- + | State (prev. callback) | Callback | Description/action | + ------------------------ ------------ -------------------------------------------- + | nothing (first call) | on_h_field | Allocate new buffer and copy callback data | + | | | into it | + ------------------------ ------------ -------------------------------------------- + | value | on_h_field | New header started. | + | | | Copy current name,value buffers to headers | + | | | list and allocate new buffer for new name | + ------------------------ ------------ -------------------------------------------- + | field | on_h_field | Previous name continues. Reallocate name | + | | | buffer and append callback data to it | + ------------------------ ------------ -------------------------------------------- + | field | on_h_value | Value for current header started. Allocate | + | | | new buffer and copy callback data to it | + ------------------------ ------------ -------------------------------------------- + | value | on_h_value | Value continues. Reallocate value buffer | + | | | and append callback data to it | + ------------------------ ------------ -------------------------------------------- + + +Parsing URLs +------------ + +A simplistic zero-copy URL parser is provided as `http_parser_parse_url()`. +Users of this library may wish to use it to parse URLs constructed from +consecutive `on_url` callbacks. + +See examples of reading in headers: + +* [partial example](http://gist.github.com/155877) in C +* [from http-parser tests](http://github.com/joyent/http-parser/blob/37a0ff8/test.c#L403) in C +* [from Node library](http://github.com/joyent/node/blob/842eaf4/src/http.js#L284) in Javascript diff --git a/src/3rdparty/http-parser/http_parser.c b/src/3rdparty/http-parser/http_parser.c new file mode 100644 index 000000000..e2fc5d2ee --- /dev/null +++ b/src/3rdparty/http-parser/http_parser.c @@ -0,0 +1,2501 @@ +/* Copyright Joyent, Inc. and other Node contributors. + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to + * deal in the Software without restriction, including without limitation the + * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or + * sell copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + */ +#include "http_parser.h" +#include +#include +#include +#include +#include + +static uint32_t max_header_size = HTTP_MAX_HEADER_SIZE; + +#ifndef ULLONG_MAX +# define ULLONG_MAX ((uint64_t) -1) /* 2^64-1 */ +#endif + +#ifndef MIN +# define MIN(a,b) ((a) < (b) ? (a) : (b)) +#endif + +#ifndef ARRAY_SIZE +# define ARRAY_SIZE(a) (sizeof(a) / sizeof((a)[0])) +#endif + +#ifndef BIT_AT +# define BIT_AT(a, i) \ + (!!((unsigned int) (a)[(unsigned int) (i) >> 3] & \ + (1 << ((unsigned int) (i) & 7)))) +#endif + +#ifndef ELEM_AT +# define ELEM_AT(a, i, v) ((unsigned int) (i) < ARRAY_SIZE(a) ? (a)[(i)] : (v)) +#endif + +#define SET_ERRNO(e) \ +do { \ + parser->nread = nread; \ + parser->http_errno = (e); \ +} while(0) + +#define CURRENT_STATE() p_state +#define UPDATE_STATE(V) p_state = (enum state) (V); +#define RETURN(V) \ +do { \ + parser->nread = nread; \ + parser->state = CURRENT_STATE(); \ + return (V); \ +} while (0); +#define REEXECUTE() \ + goto reexecute; \ + + +#ifdef __GNUC__ +# define LIKELY(X) __builtin_expect(!!(X), 1) +# define UNLIKELY(X) __builtin_expect(!!(X), 0) +#else +# define LIKELY(X) (X) +# define UNLIKELY(X) (X) +#endif + + +/* Run the notify callback FOR, returning ER if it fails */ +#define CALLBACK_NOTIFY_(FOR, ER) \ +do { \ + assert(HTTP_PARSER_ERRNO(parser) == HPE_OK); \ + \ + if (LIKELY(settings->on_##FOR)) { \ + parser->state = CURRENT_STATE(); \ + if (UNLIKELY(0 != settings->on_##FOR(parser))) { \ + SET_ERRNO(HPE_CB_##FOR); \ + } \ + UPDATE_STATE(parser->state); \ + \ + /* We either errored above or got paused; get out */ \ + if (UNLIKELY(HTTP_PARSER_ERRNO(parser) != HPE_OK)) { \ + return (ER); \ + } \ + } \ +} while (0) + +/* Run the notify callback FOR and consume the current byte */ +#define CALLBACK_NOTIFY(FOR) CALLBACK_NOTIFY_(FOR, p - data + 1) + +/* Run the notify callback FOR and don't consume the current byte */ +#define CALLBACK_NOTIFY_NOADVANCE(FOR) CALLBACK_NOTIFY_(FOR, p - data) + +/* Run data callback FOR with LEN bytes, returning ER if it fails */ +#define CALLBACK_DATA_(FOR, LEN, ER) \ +do { \ + assert(HTTP_PARSER_ERRNO(parser) == HPE_OK); \ + \ + if (FOR##_mark) { \ + if (LIKELY(settings->on_##FOR)) { \ + parser->state = CURRENT_STATE(); \ + if (UNLIKELY(0 != \ + settings->on_##FOR(parser, FOR##_mark, (LEN)))) { \ + SET_ERRNO(HPE_CB_##FOR); \ + } \ + UPDATE_STATE(parser->state); \ + \ + /* We either errored above or got paused; get out */ \ + if (UNLIKELY(HTTP_PARSER_ERRNO(parser) != HPE_OK)) { \ + return (ER); \ + } \ + } \ + FOR##_mark = NULL; \ + } \ +} while (0) + +/* Run the data callback FOR and consume the current byte */ +#define CALLBACK_DATA(FOR) \ + CALLBACK_DATA_(FOR, p - FOR##_mark, p - data + 1) + +/* Run the data callback FOR and don't consume the current byte */ +#define CALLBACK_DATA_NOADVANCE(FOR) \ + CALLBACK_DATA_(FOR, p - FOR##_mark, p - data) + +/* Set the mark FOR; non-destructive if mark is already set */ +#define MARK(FOR) \ +do { \ + if (!FOR##_mark) { \ + FOR##_mark = p; \ + } \ +} while (0) + +/* Don't allow the total size of the HTTP headers (including the status + * line) to exceed max_header_size. This check is here to protect + * embedders against denial-of-service attacks where the attacker feeds + * us a never-ending header that the embedder keeps buffering. + * + * This check is arguably the responsibility of embedders but we're doing + * it on the embedder's behalf because most won't bother and this way we + * make the web a little safer. max_header_size is still far bigger + * than any reasonable request or response so this should never affect + * day-to-day operation. + */ +#define COUNT_HEADER_SIZE(V) \ +do { \ + nread += (uint32_t)(V); \ + if (UNLIKELY(nread > max_header_size)) { \ + SET_ERRNO(HPE_HEADER_OVERFLOW); \ + goto error; \ + } \ +} while (0) + + +#define PROXY_CONNECTION "proxy-connection" +#define CONNECTION "connection" +#define CONTENT_LENGTH "content-length" +#define TRANSFER_ENCODING "transfer-encoding" +#define UPGRADE "upgrade" +#define CHUNKED "chunked" +#define KEEP_ALIVE "keep-alive" +#define CLOSE "close" + + +static const char *method_strings[] = + { +#define XX(num, name, string) #string, + HTTP_METHOD_MAP(XX) +#undef XX + }; + + +/* Tokens as defined by rfc 2616. Also lowercases them. + * token = 1* + * separators = "(" | ")" | "<" | ">" | "@" + * | "," | ";" | ":" | "\" | <"> + * | "/" | "[" | "]" | "?" | "=" + * | "{" | "}" | SP | HT + */ +static const char tokens[256] = { +/* 0 nul 1 soh 2 stx 3 etx 4 eot 5 enq 6 ack 7 bel */ + 0, 0, 0, 0, 0, 0, 0, 0, +/* 8 bs 9 ht 10 nl 11 vt 12 np 13 cr 14 so 15 si */ + 0, 0, 0, 0, 0, 0, 0, 0, +/* 16 dle 17 dc1 18 dc2 19 dc3 20 dc4 21 nak 22 syn 23 etb */ + 0, 0, 0, 0, 0, 0, 0, 0, +/* 24 can 25 em 26 sub 27 esc 28 fs 29 gs 30 rs 31 us */ + 0, 0, 0, 0, 0, 0, 0, 0, +/* 32 sp 33 ! 34 " 35 # 36 $ 37 % 38 & 39 ' */ + ' ', '!', 0, '#', '$', '%', '&', '\'', +/* 40 ( 41 ) 42 * 43 + 44 , 45 - 46 . 47 / */ + 0, 0, '*', '+', 0, '-', '.', 0, +/* 48 0 49 1 50 2 51 3 52 4 53 5 54 6 55 7 */ + '0', '1', '2', '3', '4', '5', '6', '7', +/* 56 8 57 9 58 : 59 ; 60 < 61 = 62 > 63 ? */ + '8', '9', 0, 0, 0, 0, 0, 0, +/* 64 @ 65 A 66 B 67 C 68 D 69 E 70 F 71 G */ + 0, 'a', 'b', 'c', 'd', 'e', 'f', 'g', +/* 72 H 73 I 74 J 75 K 76 L 77 M 78 N 79 O */ + 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', +/* 80 P 81 Q 82 R 83 S 84 T 85 U 86 V 87 W */ + 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', +/* 88 X 89 Y 90 Z 91 [ 92 \ 93 ] 94 ^ 95 _ */ + 'x', 'y', 'z', 0, 0, 0, '^', '_', +/* 96 ` 97 a 98 b 99 c 100 d 101 e 102 f 103 g */ + '`', 'a', 'b', 'c', 'd', 'e', 'f', 'g', +/* 104 h 105 i 106 j 107 k 108 l 109 m 110 n 111 o */ + 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', +/* 112 p 113 q 114 r 115 s 116 t 117 u 118 v 119 w */ + 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', +/* 120 x 121 y 122 z 123 { 124 | 125 } 126 ~ 127 del */ + 'x', 'y', 'z', 0, '|', 0, '~', 0 }; + + +static const int8_t unhex[256] = + {-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1 + ,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1 + ,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1 + , 0, 1, 2, 3, 4, 5, 6, 7, 8, 9,-1,-1,-1,-1,-1,-1 + ,-1,10,11,12,13,14,15,-1,-1,-1,-1,-1,-1,-1,-1,-1 + ,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1 + ,-1,10,11,12,13,14,15,-1,-1,-1,-1,-1,-1,-1,-1,-1 + ,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1 + }; + + +#if HTTP_PARSER_STRICT +# define T(v) 0 +#else +# define T(v) v +#endif + + +static const uint8_t normal_url_char[32] = { +/* 0 nul 1 soh 2 stx 3 etx 4 eot 5 enq 6 ack 7 bel */ + 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0, +/* 8 bs 9 ht 10 nl 11 vt 12 np 13 cr 14 so 15 si */ + 0 | T(2) | 0 | 0 | T(16) | 0 | 0 | 0, +/* 16 dle 17 dc1 18 dc2 19 dc3 20 dc4 21 nak 22 syn 23 etb */ + 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0, +/* 24 can 25 em 26 sub 27 esc 28 fs 29 gs 30 rs 31 us */ + 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0, +/* 32 sp 33 ! 34 " 35 # 36 $ 37 % 38 & 39 ' */ + 0 | 2 | 4 | 0 | 16 | 32 | 64 | 128, +/* 40 ( 41 ) 42 * 43 + 44 , 45 - 46 . 47 / */ + 1 | 2 | 4 | 8 | 16 | 32 | 64 | 128, +/* 48 0 49 1 50 2 51 3 52 4 53 5 54 6 55 7 */ + 1 | 2 | 4 | 8 | 16 | 32 | 64 | 128, +/* 56 8 57 9 58 : 59 ; 60 < 61 = 62 > 63 ? */ + 1 | 2 | 4 | 8 | 16 | 32 | 64 | 0, +/* 64 @ 65 A 66 B 67 C 68 D 69 E 70 F 71 G */ + 1 | 2 | 4 | 8 | 16 | 32 | 64 | 128, +/* 72 H 73 I 74 J 75 K 76 L 77 M 78 N 79 O */ + 1 | 2 | 4 | 8 | 16 | 32 | 64 | 128, +/* 80 P 81 Q 82 R 83 S 84 T 85 U 86 V 87 W */ + 1 | 2 | 4 | 8 | 16 | 32 | 64 | 128, +/* 88 X 89 Y 90 Z 91 [ 92 \ 93 ] 94 ^ 95 _ */ + 1 | 2 | 4 | 8 | 16 | 32 | 64 | 128, +/* 96 ` 97 a 98 b 99 c 100 d 101 e 102 f 103 g */ + 1 | 2 | 4 | 8 | 16 | 32 | 64 | 128, +/* 104 h 105 i 106 j 107 k 108 l 109 m 110 n 111 o */ + 1 | 2 | 4 | 8 | 16 | 32 | 64 | 128, +/* 112 p 113 q 114 r 115 s 116 t 117 u 118 v 119 w */ + 1 | 2 | 4 | 8 | 16 | 32 | 64 | 128, +/* 120 x 121 y 122 z 123 { 124 | 125 } 126 ~ 127 del */ + 1 | 2 | 4 | 8 | 16 | 32 | 64 | 0, }; + +#undef T + +enum state + { s_dead = 1 /* important that this is > 0 */ + + , s_start_req_or_res + , s_res_or_resp_H + , s_start_res + , s_res_H + , s_res_HT + , s_res_HTT + , s_res_HTTP + , s_res_http_major + , s_res_http_dot + , s_res_http_minor + , s_res_http_end + , s_res_first_status_code + , s_res_status_code + , s_res_status_start + , s_res_status + , s_res_line_almost_done + + , s_start_req + + , s_req_method + , s_req_spaces_before_url + , s_req_schema + , s_req_schema_slash + , s_req_schema_slash_slash + , s_req_server_start + , s_req_server + , s_req_server_with_at + , s_req_path + , s_req_query_string_start + , s_req_query_string + , s_req_fragment_start + , s_req_fragment + , s_req_http_start + , s_req_http_H + , s_req_http_HT + , s_req_http_HTT + , s_req_http_HTTP + , s_req_http_I + , s_req_http_IC + , s_req_http_major + , s_req_http_dot + , s_req_http_minor + , s_req_http_end + , s_req_line_almost_done + + , s_header_field_start + , s_header_field + , s_header_value_discard_ws + , s_header_value_discard_ws_almost_done + , s_header_value_discard_lws + , s_header_value_start + , s_header_value + , s_header_value_lws + + , s_header_almost_done + + , s_chunk_size_start + , s_chunk_size + , s_chunk_parameters + , s_chunk_size_almost_done + + , s_headers_almost_done + , s_headers_done + + /* Important: 's_headers_done' must be the last 'header' state. All + * states beyond this must be 'body' states. It is used for overflow + * checking. See the PARSING_HEADER() macro. + */ + + , s_chunk_data + , s_chunk_data_almost_done + , s_chunk_data_done + + , s_body_identity + , s_body_identity_eof + + , s_message_done + }; + + +#define PARSING_HEADER(state) (state <= s_headers_done) + + +enum header_states + { h_general = 0 + , h_C + , h_CO + , h_CON + + , h_matching_connection + , h_matching_proxy_connection + , h_matching_content_length + , h_matching_transfer_encoding + , h_matching_upgrade + + , h_connection + , h_content_length + , h_content_length_num + , h_content_length_ws + , h_transfer_encoding + , h_upgrade + + , h_matching_transfer_encoding_chunked + , h_matching_connection_token_start + , h_matching_connection_keep_alive + , h_matching_connection_close + , h_matching_connection_upgrade + , h_matching_connection_token + + , h_transfer_encoding_chunked + , h_connection_keep_alive + , h_connection_close + , h_connection_upgrade + }; + +enum http_host_state + { + s_http_host_dead = 1 + , s_http_userinfo_start + , s_http_userinfo + , s_http_host_start + , s_http_host_v6_start + , s_http_host + , s_http_host_v6 + , s_http_host_v6_end + , s_http_host_v6_zone_start + , s_http_host_v6_zone + , s_http_host_port_start + , s_http_host_port +}; + +/* Macros for character classes; depends on strict-mode */ +#define CR '\r' +#define LF '\n' +#define LOWER(c) (unsigned char)(c | 0x20) +#define IS_ALPHA(c) (LOWER(c) >= 'a' && LOWER(c) <= 'z') +#define IS_NUM(c) ((c) >= '0' && (c) <= '9') +#define IS_ALPHANUM(c) (IS_ALPHA(c) || IS_NUM(c)) +#define IS_HEX(c) (IS_NUM(c) || (LOWER(c) >= 'a' && LOWER(c) <= 'f')) +#define IS_MARK(c) ((c) == '-' || (c) == '_' || (c) == '.' || \ + (c) == '!' || (c) == '~' || (c) == '*' || (c) == '\'' || (c) == '(' || \ + (c) == ')') +#define IS_USERINFO_CHAR(c) (IS_ALPHANUM(c) || IS_MARK(c) || (c) == '%' || \ + (c) == ';' || (c) == ':' || (c) == '&' || (c) == '=' || (c) == '+' || \ + (c) == '$' || (c) == ',') + +#define STRICT_TOKEN(c) ((c == ' ') ? 0 : tokens[(unsigned char)c]) + +#if HTTP_PARSER_STRICT +#define TOKEN(c) STRICT_TOKEN(c) +#define IS_URL_CHAR(c) (BIT_AT(normal_url_char, (unsigned char)c)) +#define IS_HOST_CHAR(c) (IS_ALPHANUM(c) || (c) == '.' || (c) == '-') +#else +#define TOKEN(c) tokens[(unsigned char)c] +#define IS_URL_CHAR(c) \ + (BIT_AT(normal_url_char, (unsigned char)c) || ((c) & 0x80)) +#define IS_HOST_CHAR(c) \ + (IS_ALPHANUM(c) || (c) == '.' || (c) == '-' || (c) == '_') +#endif + +/** + * Verify that a char is a valid visible (printable) US-ASCII + * character or %x80-FF + **/ +#define IS_HEADER_CHAR(ch) \ + (ch == CR || ch == LF || ch == 9 || ((unsigned char)ch > 31 && ch != 127)) + +#define start_state (parser->type == HTTP_REQUEST ? s_start_req : s_start_res) + + +#if HTTP_PARSER_STRICT +# define STRICT_CHECK(cond) \ +do { \ + if (cond) { \ + SET_ERRNO(HPE_STRICT); \ + goto error; \ + } \ +} while (0) +# define NEW_MESSAGE() (http_should_keep_alive(parser) ? start_state : s_dead) +#else +# define STRICT_CHECK(cond) +# define NEW_MESSAGE() start_state +#endif + + +/* Map errno values to strings for human-readable output */ +#define HTTP_STRERROR_GEN(n, s) { "HPE_" #n, s }, +static struct { + const char *name; + const char *description; +} http_strerror_tab[] = { + HTTP_ERRNO_MAP(HTTP_STRERROR_GEN) +}; +#undef HTTP_STRERROR_GEN + +int http_message_needs_eof(const http_parser *parser); + +/* Our URL parser. + * + * This is designed to be shared by http_parser_execute() for URL validation, + * hence it has a state transition + byte-for-byte interface. In addition, it + * is meant to be embedded in http_parser_parse_url(), which does the dirty + * work of turning state transitions URL components for its API. + * + * This function should only be invoked with non-space characters. It is + * assumed that the caller cares about (and can detect) the transition between + * URL and non-URL states by looking for these. + */ +static enum state +parse_url_char(enum state s, const char ch) +{ + if (ch == ' ' || ch == '\r' || ch == '\n') { + return s_dead; + } + +#if HTTP_PARSER_STRICT + if (ch == '\t' || ch == '\f') { + return s_dead; + } +#endif + + switch (s) { + case s_req_spaces_before_url: + /* Proxied requests are followed by scheme of an absolute URI (alpha). + * All methods except CONNECT are followed by '/' or '*'. + */ + + if (ch == '/' || ch == '*') { + return s_req_path; + } + + if (IS_ALPHA(ch)) { + return s_req_schema; + } + + break; + + case s_req_schema: + if (IS_ALPHA(ch)) { + return s; + } + + if (ch == ':') { + return s_req_schema_slash; + } + + break; + + case s_req_schema_slash: + if (ch == '/') { + return s_req_schema_slash_slash; + } + + break; + + case s_req_schema_slash_slash: + if (ch == '/') { + return s_req_server_start; + } + + break; + + case s_req_server_with_at: + if (ch == '@') { + return s_dead; + } + + /* fall through */ + case s_req_server_start: + case s_req_server: + if (ch == '/') { + return s_req_path; + } + + if (ch == '?') { + return s_req_query_string_start; + } + + if (ch == '@') { + return s_req_server_with_at; + } + + if (IS_USERINFO_CHAR(ch) || ch == '[' || ch == ']') { + return s_req_server; + } + + break; + + case s_req_path: + if (IS_URL_CHAR(ch)) { + return s; + } + + switch (ch) { + case '?': + return s_req_query_string_start; + + case '#': + return s_req_fragment_start; + } + + break; + + case s_req_query_string_start: + case s_req_query_string: + if (IS_URL_CHAR(ch)) { + return s_req_query_string; + } + + switch (ch) { + case '?': + /* allow extra '?' in query string */ + return s_req_query_string; + + case '#': + return s_req_fragment_start; + } + + break; + + case s_req_fragment_start: + if (IS_URL_CHAR(ch)) { + return s_req_fragment; + } + + switch (ch) { + case '?': + return s_req_fragment; + + case '#': + return s; + } + + break; + + case s_req_fragment: + if (IS_URL_CHAR(ch)) { + return s; + } + + switch (ch) { + case '?': + case '#': + return s; + } + + break; + + default: + break; + } + + /* We should never fall out of the switch above unless there's an error */ + return s_dead; +} + +size_t http_parser_execute (http_parser *parser, + const http_parser_settings *settings, + const char *data, + size_t len) +{ + char c, ch; + int8_t unhex_val; + const char *p = data; + const char *header_field_mark = 0; + const char *header_value_mark = 0; + const char *url_mark = 0; + const char *body_mark = 0; + const char *status_mark = 0; + enum state p_state = (enum state) parser->state; + const unsigned int lenient = parser->lenient_http_headers; + uint32_t nread = parser->nread; + + /* We're in an error state. Don't bother doing anything. */ + if (HTTP_PARSER_ERRNO(parser) != HPE_OK) { + return 0; + } + + if (len == 0) { + switch (CURRENT_STATE()) { + case s_body_identity_eof: + /* Use of CALLBACK_NOTIFY() here would erroneously return 1 byte read if + * we got paused. + */ + CALLBACK_NOTIFY_NOADVANCE(message_complete); + return 0; + + case s_dead: + case s_start_req_or_res: + case s_start_res: + case s_start_req: + return 0; + + default: + SET_ERRNO(HPE_INVALID_EOF_STATE); + return 1; + } + } + + + if (CURRENT_STATE() == s_header_field) + header_field_mark = data; + if (CURRENT_STATE() == s_header_value) + header_value_mark = data; + switch (CURRENT_STATE()) { + case s_req_path: + case s_req_schema: + case s_req_schema_slash: + case s_req_schema_slash_slash: + case s_req_server_start: + case s_req_server: + case s_req_server_with_at: + case s_req_query_string_start: + case s_req_query_string: + case s_req_fragment_start: + case s_req_fragment: + url_mark = data; + break; + case s_res_status: + status_mark = data; + break; + default: + break; + } + + for (p=data; p != data + len; p++) { + ch = *p; + + if (PARSING_HEADER(CURRENT_STATE())) + COUNT_HEADER_SIZE(1); + +reexecute: + switch (CURRENT_STATE()) { + + case s_dead: + /* this state is used after a 'Connection: close' message + * the parser will error out if it reads another message + */ + if (LIKELY(ch == CR || ch == LF)) + break; + + SET_ERRNO(HPE_CLOSED_CONNECTION); + goto error; + + case s_start_req_or_res: + { + if (ch == CR || ch == LF) + break; + parser->flags = 0; + parser->content_length = ULLONG_MAX; + + if (ch == 'H') { + UPDATE_STATE(s_res_or_resp_H); + + CALLBACK_NOTIFY(message_begin); + } else { + parser->type = HTTP_REQUEST; + UPDATE_STATE(s_start_req); + REEXECUTE(); + } + + break; + } + + case s_res_or_resp_H: + if (ch == 'T') { + parser->type = HTTP_RESPONSE; + UPDATE_STATE(s_res_HT); + } else { + if (UNLIKELY(ch != 'E')) { + SET_ERRNO(HPE_INVALID_CONSTANT); + goto error; + } + + parser->type = HTTP_REQUEST; + parser->method = HTTP_HEAD; + parser->index = 2; + UPDATE_STATE(s_req_method); + } + break; + + case s_start_res: + { + if (ch == CR || ch == LF) + break; + parser->flags = 0; + parser->content_length = ULLONG_MAX; + + if (ch == 'H') { + UPDATE_STATE(s_res_H); + } else { + SET_ERRNO(HPE_INVALID_CONSTANT); + goto error; + } + + CALLBACK_NOTIFY(message_begin); + break; + } + + case s_res_H: + STRICT_CHECK(ch != 'T'); + UPDATE_STATE(s_res_HT); + break; + + case s_res_HT: + STRICT_CHECK(ch != 'T'); + UPDATE_STATE(s_res_HTT); + break; + + case s_res_HTT: + STRICT_CHECK(ch != 'P'); + UPDATE_STATE(s_res_HTTP); + break; + + case s_res_HTTP: + STRICT_CHECK(ch != '/'); + UPDATE_STATE(s_res_http_major); + break; + + case s_res_http_major: + if (UNLIKELY(!IS_NUM(ch))) { + SET_ERRNO(HPE_INVALID_VERSION); + goto error; + } + + parser->http_major = ch - '0'; + UPDATE_STATE(s_res_http_dot); + break; + + case s_res_http_dot: + { + if (UNLIKELY(ch != '.')) { + SET_ERRNO(HPE_INVALID_VERSION); + goto error; + } + + UPDATE_STATE(s_res_http_minor); + break; + } + + case s_res_http_minor: + if (UNLIKELY(!IS_NUM(ch))) { + SET_ERRNO(HPE_INVALID_VERSION); + goto error; + } + + parser->http_minor = ch - '0'; + UPDATE_STATE(s_res_http_end); + break; + + case s_res_http_end: + { + if (UNLIKELY(ch != ' ')) { + SET_ERRNO(HPE_INVALID_VERSION); + goto error; + } + + UPDATE_STATE(s_res_first_status_code); + break; + } + + case s_res_first_status_code: + { + if (!IS_NUM(ch)) { + if (ch == ' ') { + break; + } + + SET_ERRNO(HPE_INVALID_STATUS); + goto error; + } + parser->status_code = ch - '0'; + UPDATE_STATE(s_res_status_code); + break; + } + + case s_res_status_code: + { + if (!IS_NUM(ch)) { + switch (ch) { + case ' ': + UPDATE_STATE(s_res_status_start); + break; + case CR: + case LF: + UPDATE_STATE(s_res_status_start); + REEXECUTE(); + break; + default: + SET_ERRNO(HPE_INVALID_STATUS); + goto error; + } + break; + } + + parser->status_code *= 10; + parser->status_code += ch - '0'; + + if (UNLIKELY(parser->status_code > 999)) { + SET_ERRNO(HPE_INVALID_STATUS); + goto error; + } + + break; + } + + case s_res_status_start: + { + MARK(status); + UPDATE_STATE(s_res_status); + parser->index = 0; + + if (ch == CR || ch == LF) + REEXECUTE(); + + break; + } + + case s_res_status: + if (ch == CR) { + UPDATE_STATE(s_res_line_almost_done); + CALLBACK_DATA(status); + break; + } + + if (ch == LF) { + UPDATE_STATE(s_header_field_start); + CALLBACK_DATA(status); + break; + } + + break; + + case s_res_line_almost_done: + STRICT_CHECK(ch != LF); + UPDATE_STATE(s_header_field_start); + break; + + case s_start_req: + { + if (ch == CR || ch == LF) + break; + parser->flags = 0; + parser->content_length = ULLONG_MAX; + + if (UNLIKELY(!IS_ALPHA(ch))) { + SET_ERRNO(HPE_INVALID_METHOD); + goto error; + } + + parser->method = (enum http_method) 0; + parser->index = 1; + switch (ch) { + case 'A': parser->method = HTTP_ACL; break; + case 'B': parser->method = HTTP_BIND; break; + case 'C': parser->method = HTTP_CONNECT; /* or COPY, CHECKOUT */ break; + case 'D': parser->method = HTTP_DELETE; break; + case 'G': parser->method = HTTP_GET; break; + case 'H': parser->method = HTTP_HEAD; break; + case 'L': parser->method = HTTP_LOCK; /* or LINK */ break; + case 'M': parser->method = HTTP_MKCOL; /* or MOVE, MKACTIVITY, MERGE, M-SEARCH, MKCALENDAR */ break; + case 'N': parser->method = HTTP_NOTIFY; break; + case 'O': parser->method = HTTP_OPTIONS; break; + case 'P': parser->method = HTTP_POST; + /* or PROPFIND|PROPPATCH|PUT|PATCH|PURGE */ + break; + case 'R': parser->method = HTTP_REPORT; /* or REBIND */ break; + case 'S': parser->method = HTTP_SUBSCRIBE; /* or SEARCH, SOURCE */ break; + case 'T': parser->method = HTTP_TRACE; break; + case 'U': parser->method = HTTP_UNLOCK; /* or UNSUBSCRIBE, UNBIND, UNLINK */ break; + default: + SET_ERRNO(HPE_INVALID_METHOD); + goto error; + } + UPDATE_STATE(s_req_method); + + CALLBACK_NOTIFY(message_begin); + + break; + } + + case s_req_method: + { + const char *matcher; + if (UNLIKELY(ch == '\0')) { + SET_ERRNO(HPE_INVALID_METHOD); + goto error; + } + + matcher = method_strings[parser->method]; + if (ch == ' ' && matcher[parser->index] == '\0') { + UPDATE_STATE(s_req_spaces_before_url); + } else if (ch == matcher[parser->index]) { + ; /* nada */ + } else if ((ch >= 'A' && ch <= 'Z') || ch == '-') { + + switch (parser->method << 16 | parser->index << 8 | ch) { +#define XX(meth, pos, ch, new_meth) \ + case (HTTP_##meth << 16 | pos << 8 | ch): \ + parser->method = HTTP_##new_meth; break; + + XX(POST, 1, 'U', PUT) + XX(POST, 1, 'A', PATCH) + XX(POST, 1, 'R', PROPFIND) + XX(PUT, 2, 'R', PURGE) + XX(CONNECT, 1, 'H', CHECKOUT) + XX(CONNECT, 2, 'P', COPY) + XX(MKCOL, 1, 'O', MOVE) + XX(MKCOL, 1, 'E', MERGE) + XX(MKCOL, 1, '-', MSEARCH) + XX(MKCOL, 2, 'A', MKACTIVITY) + XX(MKCOL, 3, 'A', MKCALENDAR) + XX(SUBSCRIBE, 1, 'E', SEARCH) + XX(SUBSCRIBE, 1, 'O', SOURCE) + XX(REPORT, 2, 'B', REBIND) + XX(PROPFIND, 4, 'P', PROPPATCH) + XX(LOCK, 1, 'I', LINK) + XX(UNLOCK, 2, 'S', UNSUBSCRIBE) + XX(UNLOCK, 2, 'B', UNBIND) + XX(UNLOCK, 3, 'I', UNLINK) +#undef XX + default: + SET_ERRNO(HPE_INVALID_METHOD); + goto error; + } + } else { + SET_ERRNO(HPE_INVALID_METHOD); + goto error; + } + + ++parser->index; + break; + } + + case s_req_spaces_before_url: + { + if (ch == ' ') break; + + MARK(url); + if (parser->method == HTTP_CONNECT) { + UPDATE_STATE(s_req_server_start); + } + + UPDATE_STATE(parse_url_char(CURRENT_STATE(), ch)); + if (UNLIKELY(CURRENT_STATE() == s_dead)) { + SET_ERRNO(HPE_INVALID_URL); + goto error; + } + + break; + } + + case s_req_schema: + case s_req_schema_slash: + case s_req_schema_slash_slash: + case s_req_server_start: + { + switch (ch) { + /* No whitespace allowed here */ + case ' ': + case CR: + case LF: + SET_ERRNO(HPE_INVALID_URL); + goto error; + default: + UPDATE_STATE(parse_url_char(CURRENT_STATE(), ch)); + if (UNLIKELY(CURRENT_STATE() == s_dead)) { + SET_ERRNO(HPE_INVALID_URL); + goto error; + } + } + + break; + } + + case s_req_server: + case s_req_server_with_at: + case s_req_path: + case s_req_query_string_start: + case s_req_query_string: + case s_req_fragment_start: + case s_req_fragment: + { + switch (ch) { + case ' ': + UPDATE_STATE(s_req_http_start); + CALLBACK_DATA(url); + break; + case CR: + case LF: + parser->http_major = 0; + parser->http_minor = 9; + UPDATE_STATE((ch == CR) ? + s_req_line_almost_done : + s_header_field_start); + CALLBACK_DATA(url); + break; + default: + UPDATE_STATE(parse_url_char(CURRENT_STATE(), ch)); + if (UNLIKELY(CURRENT_STATE() == s_dead)) { + SET_ERRNO(HPE_INVALID_URL); + goto error; + } + } + break; + } + + case s_req_http_start: + switch (ch) { + case ' ': + break; + case 'H': + UPDATE_STATE(s_req_http_H); + break; + case 'I': + if (parser->method == HTTP_SOURCE) { + UPDATE_STATE(s_req_http_I); + break; + } + /* fall through */ + default: + SET_ERRNO(HPE_INVALID_CONSTANT); + goto error; + } + break; + + case s_req_http_H: + STRICT_CHECK(ch != 'T'); + UPDATE_STATE(s_req_http_HT); + break; + + case s_req_http_HT: + STRICT_CHECK(ch != 'T'); + UPDATE_STATE(s_req_http_HTT); + break; + + case s_req_http_HTT: + STRICT_CHECK(ch != 'P'); + UPDATE_STATE(s_req_http_HTTP); + break; + + case s_req_http_I: + STRICT_CHECK(ch != 'C'); + UPDATE_STATE(s_req_http_IC); + break; + + case s_req_http_IC: + STRICT_CHECK(ch != 'E'); + UPDATE_STATE(s_req_http_HTTP); /* Treat "ICE" as "HTTP". */ + break; + + case s_req_http_HTTP: + STRICT_CHECK(ch != '/'); + UPDATE_STATE(s_req_http_major); + break; + + case s_req_http_major: + if (UNLIKELY(!IS_NUM(ch))) { + SET_ERRNO(HPE_INVALID_VERSION); + goto error; + } + + parser->http_major = ch - '0'; + UPDATE_STATE(s_req_http_dot); + break; + + case s_req_http_dot: + { + if (UNLIKELY(ch != '.')) { + SET_ERRNO(HPE_INVALID_VERSION); + goto error; + } + + UPDATE_STATE(s_req_http_minor); + break; + } + + case s_req_http_minor: + if (UNLIKELY(!IS_NUM(ch))) { + SET_ERRNO(HPE_INVALID_VERSION); + goto error; + } + + parser->http_minor = ch - '0'; + UPDATE_STATE(s_req_http_end); + break; + + case s_req_http_end: + { + if (ch == CR) { + UPDATE_STATE(s_req_line_almost_done); + break; + } + + if (ch == LF) { + UPDATE_STATE(s_header_field_start); + break; + } + + SET_ERRNO(HPE_INVALID_VERSION); + goto error; + break; + } + + /* end of request line */ + case s_req_line_almost_done: + { + if (UNLIKELY(ch != LF)) { + SET_ERRNO(HPE_LF_EXPECTED); + goto error; + } + + UPDATE_STATE(s_header_field_start); + break; + } + + case s_header_field_start: + { + if (ch == CR) { + UPDATE_STATE(s_headers_almost_done); + break; + } + + if (ch == LF) { + /* they might be just sending \n instead of \r\n so this would be + * the second \n to denote the end of headers*/ + UPDATE_STATE(s_headers_almost_done); + REEXECUTE(); + } + + c = TOKEN(ch); + + if (UNLIKELY(!c)) { + SET_ERRNO(HPE_INVALID_HEADER_TOKEN); + goto error; + } + + MARK(header_field); + + parser->index = 0; + UPDATE_STATE(s_header_field); + + switch (c) { + case 'c': + parser->header_state = h_C; + break; + + case 'p': + parser->header_state = h_matching_proxy_connection; + break; + + case 't': + parser->header_state = h_matching_transfer_encoding; + break; + + case 'u': + parser->header_state = h_matching_upgrade; + break; + + default: + parser->header_state = h_general; + break; + } + break; + } + + case s_header_field: + { + const char* start = p; + for (; p != data + len; p++) { + ch = *p; + c = TOKEN(ch); + + if (!c) + break; + + switch (parser->header_state) { + case h_general: { + size_t limit = data + len - p; + limit = MIN(limit, max_header_size); + while (p+1 < data + limit && TOKEN(p[1])) { + p++; + } + break; + } + + case h_C: + parser->index++; + parser->header_state = (c == 'o' ? h_CO : h_general); + break; + + case h_CO: + parser->index++; + parser->header_state = (c == 'n' ? h_CON : h_general); + break; + + case h_CON: + parser->index++; + switch (c) { + case 'n': + parser->header_state = h_matching_connection; + break; + case 't': + parser->header_state = h_matching_content_length; + break; + default: + parser->header_state = h_general; + break; + } + break; + + /* connection */ + + case h_matching_connection: + parser->index++; + if (parser->index > sizeof(CONNECTION)-1 + || c != CONNECTION[parser->index]) { + parser->header_state = h_general; + } else if (parser->index == sizeof(CONNECTION)-2) { + parser->header_state = h_connection; + } + break; + + /* proxy-connection */ + + case h_matching_proxy_connection: + parser->index++; + if (parser->index > sizeof(PROXY_CONNECTION)-1 + || c != PROXY_CONNECTION[parser->index]) { + parser->header_state = h_general; + } else if (parser->index == sizeof(PROXY_CONNECTION)-2) { + parser->header_state = h_connection; + } + break; + + /* content-length */ + + case h_matching_content_length: + parser->index++; + if (parser->index > sizeof(CONTENT_LENGTH)-1 + || c != CONTENT_LENGTH[parser->index]) { + parser->header_state = h_general; + } else if (parser->index == sizeof(CONTENT_LENGTH)-2) { + parser->header_state = h_content_length; + } + break; + + /* transfer-encoding */ + + case h_matching_transfer_encoding: + parser->index++; + if (parser->index > sizeof(TRANSFER_ENCODING)-1 + || c != TRANSFER_ENCODING[parser->index]) { + parser->header_state = h_general; + } else if (parser->index == sizeof(TRANSFER_ENCODING)-2) { + parser->header_state = h_transfer_encoding; + } + break; + + /* upgrade */ + + case h_matching_upgrade: + parser->index++; + if (parser->index > sizeof(UPGRADE)-1 + || c != UPGRADE[parser->index]) { + parser->header_state = h_general; + } else if (parser->index == sizeof(UPGRADE)-2) { + parser->header_state = h_upgrade; + } + break; + + case h_connection: + case h_content_length: + case h_transfer_encoding: + case h_upgrade: + if (ch != ' ') parser->header_state = h_general; + break; + + default: + assert(0 && "Unknown header_state"); + break; + } + } + + if (p == data + len) { + --p; + COUNT_HEADER_SIZE(p - start); + break; + } + + COUNT_HEADER_SIZE(p - start); + + if (ch == ':') { + UPDATE_STATE(s_header_value_discard_ws); + CALLBACK_DATA(header_field); + break; + } + + SET_ERRNO(HPE_INVALID_HEADER_TOKEN); + goto error; + } + + case s_header_value_discard_ws: + if (ch == ' ' || ch == '\t') break; + + if (ch == CR) { + UPDATE_STATE(s_header_value_discard_ws_almost_done); + break; + } + + if (ch == LF) { + UPDATE_STATE(s_header_value_discard_lws); + break; + } + + /* fall through */ + + case s_header_value_start: + { + MARK(header_value); + + UPDATE_STATE(s_header_value); + parser->index = 0; + + c = LOWER(ch); + + switch (parser->header_state) { + case h_upgrade: + parser->flags |= F_UPGRADE; + parser->header_state = h_general; + break; + + case h_transfer_encoding: + /* looking for 'Transfer-Encoding: chunked' */ + if ('c' == c) { + parser->header_state = h_matching_transfer_encoding_chunked; + } else { + parser->header_state = h_general; + } + break; + + case h_content_length: + if (UNLIKELY(!IS_NUM(ch))) { + SET_ERRNO(HPE_INVALID_CONTENT_LENGTH); + goto error; + } + + if (parser->flags & F_CONTENTLENGTH) { + SET_ERRNO(HPE_UNEXPECTED_CONTENT_LENGTH); + goto error; + } + + parser->flags |= F_CONTENTLENGTH; + parser->content_length = ch - '0'; + parser->header_state = h_content_length_num; + break; + + /* when obsolete line folding is encountered for content length + * continue to the s_header_value state */ + case h_content_length_ws: + break; + + case h_connection: + /* looking for 'Connection: keep-alive' */ + if (c == 'k') { + parser->header_state = h_matching_connection_keep_alive; + /* looking for 'Connection: close' */ + } else if (c == 'c') { + parser->header_state = h_matching_connection_close; + } else if (c == 'u') { + parser->header_state = h_matching_connection_upgrade; + } else { + parser->header_state = h_matching_connection_token; + } + break; + + /* Multi-value `Connection` header */ + case h_matching_connection_token_start: + break; + + default: + parser->header_state = h_general; + break; + } + break; + } + + case s_header_value: + { + const char* start = p; + enum header_states h_state = (enum header_states) parser->header_state; + for (; p != data + len; p++) { + ch = *p; + if (ch == CR) { + UPDATE_STATE(s_header_almost_done); + parser->header_state = h_state; + CALLBACK_DATA(header_value); + break; + } + + if (ch == LF) { + UPDATE_STATE(s_header_almost_done); + COUNT_HEADER_SIZE(p - start); + parser->header_state = h_state; + CALLBACK_DATA_NOADVANCE(header_value); + REEXECUTE(); + } + + if (!lenient && !IS_HEADER_CHAR(ch)) { + SET_ERRNO(HPE_INVALID_HEADER_TOKEN); + goto error; + } + + c = LOWER(ch); + + switch (h_state) { + case h_general: + { + const char* p_cr; + const char* p_lf; + size_t limit = data + len - p; + + limit = MIN(limit, max_header_size); + + p_cr = (const char*) memchr(p, CR, limit); + p_lf = (const char*) memchr(p, LF, limit); + if (p_cr != NULL) { + if (p_lf != NULL && p_cr >= p_lf) + p = p_lf; + else + p = p_cr; + } else if (UNLIKELY(p_lf != NULL)) { + p = p_lf; + } else { + p = data + len; + } + --p; + break; + } + + case h_connection: + case h_transfer_encoding: + assert(0 && "Shouldn't get here."); + break; + + case h_content_length: + if (ch == ' ') break; + h_state = h_content_length_num; + /* fall through */ + + case h_content_length_num: + { + uint64_t t; + + if (ch == ' ') { + h_state = h_content_length_ws; + break; + } + + if (UNLIKELY(!IS_NUM(ch))) { + SET_ERRNO(HPE_INVALID_CONTENT_LENGTH); + parser->header_state = h_state; + goto error; + } + + t = parser->content_length; + t *= 10; + t += ch - '0'; + + /* Overflow? Test against a conservative limit for simplicity. */ + if (UNLIKELY((ULLONG_MAX - 10) / 10 < parser->content_length)) { + SET_ERRNO(HPE_INVALID_CONTENT_LENGTH); + parser->header_state = h_state; + goto error; + } + + parser->content_length = t; + break; + } + + case h_content_length_ws: + if (ch == ' ') break; + SET_ERRNO(HPE_INVALID_CONTENT_LENGTH); + parser->header_state = h_state; + goto error; + + /* Transfer-Encoding: chunked */ + case h_matching_transfer_encoding_chunked: + parser->index++; + if (parser->index > sizeof(CHUNKED)-1 + || c != CHUNKED[parser->index]) { + h_state = h_general; + } else if (parser->index == sizeof(CHUNKED)-2) { + h_state = h_transfer_encoding_chunked; + } + break; + + case h_matching_connection_token_start: + /* looking for 'Connection: keep-alive' */ + if (c == 'k') { + h_state = h_matching_connection_keep_alive; + /* looking for 'Connection: close' */ + } else if (c == 'c') { + h_state = h_matching_connection_close; + } else if (c == 'u') { + h_state = h_matching_connection_upgrade; + } else if (STRICT_TOKEN(c)) { + h_state = h_matching_connection_token; + } else if (c == ' ' || c == '\t') { + /* Skip lws */ + } else { + h_state = h_general; + } + break; + + /* looking for 'Connection: keep-alive' */ + case h_matching_connection_keep_alive: + parser->index++; + if (parser->index > sizeof(KEEP_ALIVE)-1 + || c != KEEP_ALIVE[parser->index]) { + h_state = h_matching_connection_token; + } else if (parser->index == sizeof(KEEP_ALIVE)-2) { + h_state = h_connection_keep_alive; + } + break; + + /* looking for 'Connection: close' */ + case h_matching_connection_close: + parser->index++; + if (parser->index > sizeof(CLOSE)-1 || c != CLOSE[parser->index]) { + h_state = h_matching_connection_token; + } else if (parser->index == sizeof(CLOSE)-2) { + h_state = h_connection_close; + } + break; + + /* looking for 'Connection: upgrade' */ + case h_matching_connection_upgrade: + parser->index++; + if (parser->index > sizeof(UPGRADE) - 1 || + c != UPGRADE[parser->index]) { + h_state = h_matching_connection_token; + } else if (parser->index == sizeof(UPGRADE)-2) { + h_state = h_connection_upgrade; + } + break; + + case h_matching_connection_token: + if (ch == ',') { + h_state = h_matching_connection_token_start; + parser->index = 0; + } + break; + + case h_transfer_encoding_chunked: + if (ch != ' ') h_state = h_general; + break; + + case h_connection_keep_alive: + case h_connection_close: + case h_connection_upgrade: + if (ch == ',') { + if (h_state == h_connection_keep_alive) { + parser->flags |= F_CONNECTION_KEEP_ALIVE; + } else if (h_state == h_connection_close) { + parser->flags |= F_CONNECTION_CLOSE; + } else if (h_state == h_connection_upgrade) { + parser->flags |= F_CONNECTION_UPGRADE; + } + h_state = h_matching_connection_token_start; + parser->index = 0; + } else if (ch != ' ') { + h_state = h_matching_connection_token; + } + break; + + default: + UPDATE_STATE(s_header_value); + h_state = h_general; + break; + } + } + parser->header_state = h_state; + + if (p == data + len) + --p; + + COUNT_HEADER_SIZE(p - start); + break; + } + + case s_header_almost_done: + { + if (UNLIKELY(ch != LF)) { + SET_ERRNO(HPE_LF_EXPECTED); + goto error; + } + + UPDATE_STATE(s_header_value_lws); + break; + } + + case s_header_value_lws: + { + if (ch == ' ' || ch == '\t') { + if (parser->header_state == h_content_length_num) { + /* treat obsolete line folding as space */ + parser->header_state = h_content_length_ws; + } + UPDATE_STATE(s_header_value_start); + REEXECUTE(); + } + + /* finished the header */ + switch (parser->header_state) { + case h_connection_keep_alive: + parser->flags |= F_CONNECTION_KEEP_ALIVE; + break; + case h_connection_close: + parser->flags |= F_CONNECTION_CLOSE; + break; + case h_transfer_encoding_chunked: + parser->flags |= F_CHUNKED; + break; + case h_connection_upgrade: + parser->flags |= F_CONNECTION_UPGRADE; + break; + default: + break; + } + + UPDATE_STATE(s_header_field_start); + REEXECUTE(); + } + + case s_header_value_discard_ws_almost_done: + { + STRICT_CHECK(ch != LF); + UPDATE_STATE(s_header_value_discard_lws); + break; + } + + case s_header_value_discard_lws: + { + if (ch == ' ' || ch == '\t') { + UPDATE_STATE(s_header_value_discard_ws); + break; + } else { + switch (parser->header_state) { + case h_connection_keep_alive: + parser->flags |= F_CONNECTION_KEEP_ALIVE; + break; + case h_connection_close: + parser->flags |= F_CONNECTION_CLOSE; + break; + case h_connection_upgrade: + parser->flags |= F_CONNECTION_UPGRADE; + break; + case h_transfer_encoding_chunked: + parser->flags |= F_CHUNKED; + break; + case h_content_length: + /* do not allow empty content length */ + SET_ERRNO(HPE_INVALID_CONTENT_LENGTH); + goto error; + break; + default: + break; + } + + /* header value was empty */ + MARK(header_value); + UPDATE_STATE(s_header_field_start); + CALLBACK_DATA_NOADVANCE(header_value); + REEXECUTE(); + } + } + + case s_headers_almost_done: + { + STRICT_CHECK(ch != LF); + + if (parser->flags & F_TRAILING) { + /* End of a chunked request */ + UPDATE_STATE(s_message_done); + CALLBACK_NOTIFY_NOADVANCE(chunk_complete); + REEXECUTE(); + } + + /* Cannot use chunked encoding and a content-length header together + per the HTTP specification. */ + if ((parser->flags & F_CHUNKED) && + (parser->flags & F_CONTENTLENGTH)) { + SET_ERRNO(HPE_UNEXPECTED_CONTENT_LENGTH); + goto error; + } + + UPDATE_STATE(s_headers_done); + + /* Set this here so that on_headers_complete() callbacks can see it */ + if ((parser->flags & F_UPGRADE) && + (parser->flags & F_CONNECTION_UPGRADE)) { + /* For responses, "Upgrade: foo" and "Connection: upgrade" are + * mandatory only when it is a 101 Switching Protocols response, + * otherwise it is purely informational, to announce support. + */ + parser->upgrade = + (parser->type == HTTP_REQUEST || parser->status_code == 101); + } else { + parser->upgrade = (parser->method == HTTP_CONNECT); + } + + /* Here we call the headers_complete callback. This is somewhat + * different than other callbacks because if the user returns 1, we + * will interpret that as saying that this message has no body. This + * is needed for the annoying case of recieving a response to a HEAD + * request. + * + * We'd like to use CALLBACK_NOTIFY_NOADVANCE() here but we cannot, so + * we have to simulate it by handling a change in errno below. + */ + if (settings->on_headers_complete) { + switch (settings->on_headers_complete(parser)) { + case 0: + break; + + case 2: + parser->upgrade = 1; + + /* fall through */ + case 1: + parser->flags |= F_SKIPBODY; + break; + + default: + SET_ERRNO(HPE_CB_headers_complete); + RETURN(p - data); /* Error */ + } + } + + if (HTTP_PARSER_ERRNO(parser) != HPE_OK) { + RETURN(p - data); + } + + REEXECUTE(); + } + + case s_headers_done: + { + int hasBody; + STRICT_CHECK(ch != LF); + + parser->nread = 0; + nread = 0; + + hasBody = parser->flags & F_CHUNKED || + (parser->content_length > 0 && parser->content_length != ULLONG_MAX); + if (parser->upgrade && (parser->method == HTTP_CONNECT || + (parser->flags & F_SKIPBODY) || !hasBody)) { + /* Exit, the rest of the message is in a different protocol. */ + UPDATE_STATE(NEW_MESSAGE()); + CALLBACK_NOTIFY(message_complete); + RETURN((p - data) + 1); + } + + if (parser->flags & F_SKIPBODY) { + UPDATE_STATE(NEW_MESSAGE()); + CALLBACK_NOTIFY(message_complete); + } else if (parser->flags & F_CHUNKED) { + /* chunked encoding - ignore Content-Length header */ + UPDATE_STATE(s_chunk_size_start); + } else { + if (parser->content_length == 0) { + /* Content-Length header given but zero: Content-Length: 0\r\n */ + UPDATE_STATE(NEW_MESSAGE()); + CALLBACK_NOTIFY(message_complete); + } else if (parser->content_length != ULLONG_MAX) { + /* Content-Length header given and non-zero */ + UPDATE_STATE(s_body_identity); + } else { + if (!http_message_needs_eof(parser)) { + /* Assume content-length 0 - read the next */ + UPDATE_STATE(NEW_MESSAGE()); + CALLBACK_NOTIFY(message_complete); + } else { + /* Read body until EOF */ + UPDATE_STATE(s_body_identity_eof); + } + } + } + + break; + } + + case s_body_identity: + { + uint64_t to_read = MIN(parser->content_length, + (uint64_t) ((data + len) - p)); + + assert(parser->content_length != 0 + && parser->content_length != ULLONG_MAX); + + /* The difference between advancing content_length and p is because + * the latter will automaticaly advance on the next loop iteration. + * Further, if content_length ends up at 0, we want to see the last + * byte again for our message complete callback. + */ + MARK(body); + parser->content_length -= to_read; + p += to_read - 1; + + if (parser->content_length == 0) { + UPDATE_STATE(s_message_done); + + /* Mimic CALLBACK_DATA_NOADVANCE() but with one extra byte. + * + * The alternative to doing this is to wait for the next byte to + * trigger the data callback, just as in every other case. The + * problem with this is that this makes it difficult for the test + * harness to distinguish between complete-on-EOF and + * complete-on-length. It's not clear that this distinction is + * important for applications, but let's keep it for now. + */ + CALLBACK_DATA_(body, p - body_mark + 1, p - data); + REEXECUTE(); + } + + break; + } + + /* read until EOF */ + case s_body_identity_eof: + MARK(body); + p = data + len - 1; + + break; + + case s_message_done: + UPDATE_STATE(NEW_MESSAGE()); + CALLBACK_NOTIFY(message_complete); + if (parser->upgrade) { + /* Exit, the rest of the message is in a different protocol. */ + RETURN((p - data) + 1); + } + break; + + case s_chunk_size_start: + { + assert(nread == 1); + assert(parser->flags & F_CHUNKED); + + unhex_val = unhex[(unsigned char)ch]; + if (UNLIKELY(unhex_val == -1)) { + SET_ERRNO(HPE_INVALID_CHUNK_SIZE); + goto error; + } + + parser->content_length = unhex_val; + UPDATE_STATE(s_chunk_size); + break; + } + + case s_chunk_size: + { + uint64_t t; + + assert(parser->flags & F_CHUNKED); + + if (ch == CR) { + UPDATE_STATE(s_chunk_size_almost_done); + break; + } + + unhex_val = unhex[(unsigned char)ch]; + + if (unhex_val == -1) { + if (ch == ';' || ch == ' ') { + UPDATE_STATE(s_chunk_parameters); + break; + } + + SET_ERRNO(HPE_INVALID_CHUNK_SIZE); + goto error; + } + + t = parser->content_length; + t *= 16; + t += unhex_val; + + /* Overflow? Test against a conservative limit for simplicity. */ + if (UNLIKELY((ULLONG_MAX - 16) / 16 < parser->content_length)) { + SET_ERRNO(HPE_INVALID_CONTENT_LENGTH); + goto error; + } + + parser->content_length = t; + break; + } + + case s_chunk_parameters: + { + assert(parser->flags & F_CHUNKED); + /* just ignore this shit. TODO check for overflow */ + if (ch == CR) { + UPDATE_STATE(s_chunk_size_almost_done); + break; + } + break; + } + + case s_chunk_size_almost_done: + { + assert(parser->flags & F_CHUNKED); + STRICT_CHECK(ch != LF); + + parser->nread = 0; + nread = 0; + + if (parser->content_length == 0) { + parser->flags |= F_TRAILING; + UPDATE_STATE(s_header_field_start); + } else { + UPDATE_STATE(s_chunk_data); + } + CALLBACK_NOTIFY(chunk_header); + break; + } + + case s_chunk_data: + { + uint64_t to_read = MIN(parser->content_length, + (uint64_t) ((data + len) - p)); + + assert(parser->flags & F_CHUNKED); + assert(parser->content_length != 0 + && parser->content_length != ULLONG_MAX); + + /* See the explanation in s_body_identity for why the content + * length and data pointers are managed this way. + */ + MARK(body); + parser->content_length -= to_read; + p += to_read - 1; + + if (parser->content_length == 0) { + UPDATE_STATE(s_chunk_data_almost_done); + } + + break; + } + + case s_chunk_data_almost_done: + assert(parser->flags & F_CHUNKED); + assert(parser->content_length == 0); + STRICT_CHECK(ch != CR); + UPDATE_STATE(s_chunk_data_done); + CALLBACK_DATA(body); + break; + + case s_chunk_data_done: + assert(parser->flags & F_CHUNKED); + STRICT_CHECK(ch != LF); + parser->nread = 0; + nread = 0; + UPDATE_STATE(s_chunk_size_start); + CALLBACK_NOTIFY(chunk_complete); + break; + + default: + assert(0 && "unhandled state"); + SET_ERRNO(HPE_INVALID_INTERNAL_STATE); + goto error; + } + } + + /* Run callbacks for any marks that we have leftover after we ran out of + * bytes. There should be at most one of these set, so it's OK to invoke + * them in series (unset marks will not result in callbacks). + * + * We use the NOADVANCE() variety of callbacks here because 'p' has already + * overflowed 'data' and this allows us to correct for the off-by-one that + * we'd otherwise have (since CALLBACK_DATA() is meant to be run with a 'p' + * value that's in-bounds). + */ + + assert(((header_field_mark ? 1 : 0) + + (header_value_mark ? 1 : 0) + + (url_mark ? 1 : 0) + + (body_mark ? 1 : 0) + + (status_mark ? 1 : 0)) <= 1); + + CALLBACK_DATA_NOADVANCE(header_field); + CALLBACK_DATA_NOADVANCE(header_value); + CALLBACK_DATA_NOADVANCE(url); + CALLBACK_DATA_NOADVANCE(body); + CALLBACK_DATA_NOADVANCE(status); + + RETURN(len); + +error: + if (HTTP_PARSER_ERRNO(parser) == HPE_OK) { + SET_ERRNO(HPE_UNKNOWN); + } + + RETURN(p - data); +} + + +/* Does the parser need to see an EOF to find the end of the message? */ +int +http_message_needs_eof (const http_parser *parser) +{ + if (parser->type == HTTP_REQUEST) { + return 0; + } + + /* See RFC 2616 section 4.4 */ + if (parser->status_code / 100 == 1 || /* 1xx e.g. Continue */ + parser->status_code == 204 || /* No Content */ + parser->status_code == 304 || /* Not Modified */ + parser->flags & F_SKIPBODY) { /* response to a HEAD request */ + return 0; + } + + if ((parser->flags & F_CHUNKED) || parser->content_length != ULLONG_MAX) { + return 0; + } + + return 1; +} + + +int +http_should_keep_alive (const http_parser *parser) +{ + if (parser->http_major > 0 && parser->http_minor > 0) { + /* HTTP/1.1 */ + if (parser->flags & F_CONNECTION_CLOSE) { + return 0; + } + } else { + /* HTTP/1.0 or earlier */ + if (!(parser->flags & F_CONNECTION_KEEP_ALIVE)) { + return 0; + } + } + + return !http_message_needs_eof(parser); +} + + +const char * +http_method_str (enum http_method m) +{ + return ELEM_AT(method_strings, m, ""); +} + +const char * +http_status_str (enum http_status s) +{ + switch (s) { +#define XX(num, name, string) case HTTP_STATUS_##name: return #string; + HTTP_STATUS_MAP(XX) +#undef XX + default: return ""; + } +} + +void +http_parser_init (http_parser *parser, enum http_parser_type t) +{ + void *data = parser->data; /* preserve application data */ + memset(parser, 0, sizeof(*parser)); + parser->data = data; + parser->type = t; + parser->state = (t == HTTP_REQUEST ? s_start_req : (t == HTTP_RESPONSE ? s_start_res : s_start_req_or_res)); + parser->http_errno = HPE_OK; +} + +void +http_parser_settings_init(http_parser_settings *settings) +{ + memset(settings, 0, sizeof(*settings)); +} + +const char * +http_errno_name(enum http_errno err) { + assert(((size_t) err) < ARRAY_SIZE(http_strerror_tab)); + return http_strerror_tab[err].name; +} + +const char * +http_errno_description(enum http_errno err) { + assert(((size_t) err) < ARRAY_SIZE(http_strerror_tab)); + return http_strerror_tab[err].description; +} + +static enum http_host_state +http_parse_host_char(enum http_host_state s, const char ch) { + switch(s) { + case s_http_userinfo: + case s_http_userinfo_start: + if (ch == '@') { + return s_http_host_start; + } + + if (IS_USERINFO_CHAR(ch)) { + return s_http_userinfo; + } + break; + + case s_http_host_start: + if (ch == '[') { + return s_http_host_v6_start; + } + + if (IS_HOST_CHAR(ch)) { + return s_http_host; + } + + break; + + case s_http_host: + if (IS_HOST_CHAR(ch)) { + return s_http_host; + } + + /* fall through */ + case s_http_host_v6_end: + if (ch == ':') { + return s_http_host_port_start; + } + + break; + + case s_http_host_v6: + if (ch == ']') { + return s_http_host_v6_end; + } + + /* fall through */ + case s_http_host_v6_start: + if (IS_HEX(ch) || ch == ':' || ch == '.') { + return s_http_host_v6; + } + + if (s == s_http_host_v6 && ch == '%') { + return s_http_host_v6_zone_start; + } + break; + + case s_http_host_v6_zone: + if (ch == ']') { + return s_http_host_v6_end; + } + + /* fall through */ + case s_http_host_v6_zone_start: + /* RFC 6874 Zone ID consists of 1*( unreserved / pct-encoded) */ + if (IS_ALPHANUM(ch) || ch == '%' || ch == '.' || ch == '-' || ch == '_' || + ch == '~') { + return s_http_host_v6_zone; + } + break; + + case s_http_host_port: + case s_http_host_port_start: + if (IS_NUM(ch)) { + return s_http_host_port; + } + + break; + + default: + break; + } + return s_http_host_dead; +} + +static int +http_parse_host(const char * buf, struct http_parser_url *u, int found_at) { + enum http_host_state s; + + const char *p; + size_t buflen = u->field_data[UF_HOST].off + u->field_data[UF_HOST].len; + + assert(u->field_set & (1 << UF_HOST)); + + u->field_data[UF_HOST].len = 0; + + s = found_at ? s_http_userinfo_start : s_http_host_start; + + for (p = buf + u->field_data[UF_HOST].off; p < buf + buflen; p++) { + enum http_host_state new_s = http_parse_host_char(s, *p); + + if (new_s == s_http_host_dead) { + return 1; + } + + switch(new_s) { + case s_http_host: + if (s != s_http_host) { + u->field_data[UF_HOST].off = (uint16_t)(p - buf); + } + u->field_data[UF_HOST].len++; + break; + + case s_http_host_v6: + if (s != s_http_host_v6) { + u->field_data[UF_HOST].off = (uint16_t)(p - buf); + } + u->field_data[UF_HOST].len++; + break; + + case s_http_host_v6_zone_start: + case s_http_host_v6_zone: + u->field_data[UF_HOST].len++; + break; + + case s_http_host_port: + if (s != s_http_host_port) { + u->field_data[UF_PORT].off = (uint16_t)(p - buf); + u->field_data[UF_PORT].len = 0; + u->field_set |= (1 << UF_PORT); + } + u->field_data[UF_PORT].len++; + break; + + case s_http_userinfo: + if (s != s_http_userinfo) { + u->field_data[UF_USERINFO].off = (uint16_t)(p - buf); + u->field_data[UF_USERINFO].len = 0; + u->field_set |= (1 << UF_USERINFO); + } + u->field_data[UF_USERINFO].len++; + break; + + default: + break; + } + s = new_s; + } + + /* Make sure we don't end somewhere unexpected */ + switch (s) { + case s_http_host_start: + case s_http_host_v6_start: + case s_http_host_v6: + case s_http_host_v6_zone_start: + case s_http_host_v6_zone: + case s_http_host_port_start: + case s_http_userinfo: + case s_http_userinfo_start: + return 1; + default: + break; + } + + return 0; +} + +void +http_parser_url_init(struct http_parser_url *u) { + memset(u, 0, sizeof(*u)); +} + +int +http_parser_parse_url(const char *buf, size_t buflen, int is_connect, + struct http_parser_url *u) +{ + enum state s; + const char *p; + enum http_parser_url_fields uf, old_uf; + int found_at = 0; + + if (buflen == 0) { + return 1; + } + + u->port = u->field_set = 0; + s = is_connect ? s_req_server_start : s_req_spaces_before_url; + old_uf = UF_MAX; + + for (p = buf; p < buf + buflen; p++) { + s = parse_url_char(s, *p); + + /* Figure out the next field that we're operating on */ + switch (s) { + case s_dead: + return 1; + + /* Skip delimeters */ + case s_req_schema_slash: + case s_req_schema_slash_slash: + case s_req_server_start: + case s_req_query_string_start: + case s_req_fragment_start: + continue; + + case s_req_schema: + uf = UF_SCHEMA; + break; + + case s_req_server_with_at: + found_at = 1; + + /* fall through */ + case s_req_server: + uf = UF_HOST; + break; + + case s_req_path: + uf = UF_PATH; + break; + + case s_req_query_string: + uf = UF_QUERY; + break; + + case s_req_fragment: + uf = UF_FRAGMENT; + break; + + default: + assert(!"Unexpected state"); + return 1; + } + + /* Nothing's changed; soldier on */ + if (uf == old_uf) { + u->field_data[uf].len++; + continue; + } + + u->field_data[uf].off = (uint16_t)(p - buf); + u->field_data[uf].len = 1; + + u->field_set |= (1 << uf); + old_uf = uf; + } + + /* host must be present if there is a schema */ + /* parsing http:///toto will fail */ + if ((u->field_set & (1 << UF_SCHEMA)) && + (u->field_set & (1 << UF_HOST)) == 0) { + return 1; + } + + if (u->field_set & (1 << UF_HOST)) { + if (http_parse_host(buf, u, found_at) != 0) { + return 1; + } + } + + /* CONNECT requests can only contain "hostname:port" */ + if (is_connect && u->field_set != ((1 << UF_HOST)|(1 << UF_PORT))) { + return 1; + } + + if (u->field_set & (1 << UF_PORT)) { + uint16_t off; + uint16_t len; + const char* p; + const char* end; + unsigned long v; + + off = u->field_data[UF_PORT].off; + len = u->field_data[UF_PORT].len; + end = buf + off + len; + + /* NOTE: The characters are already validated and are in the [0-9] range */ + assert(off + len <= buflen && "Port number overflow"); + v = 0; + for (p = buf + off; p < end; p++) { + v *= 10; + v += *p - '0'; + + /* Ports have a max value of 2^16 */ + if (v > 0xffff) { + return 1; + } + } + + u->port = (uint16_t) v; + } + + return 0; +} + +void +http_parser_pause(http_parser *parser, int paused) { + /* Users should only be pausing/unpausing a parser that is not in an error + * state. In non-debug builds, there's not much that we can do about this + * other than ignore it. + */ + if (HTTP_PARSER_ERRNO(parser) == HPE_OK || + HTTP_PARSER_ERRNO(parser) == HPE_PAUSED) { + uint32_t nread = parser->nread; /* used by the SET_ERRNO macro */ + SET_ERRNO((paused) ? HPE_PAUSED : HPE_OK); + } else { + assert(0 && "Attempting to pause parser in error state"); + } +} + +int +http_body_is_final(const struct http_parser *parser) { + return parser->state == s_message_done; +} + +unsigned long +http_parser_version(void) { + return HTTP_PARSER_VERSION_MAJOR * 0x10000 | + HTTP_PARSER_VERSION_MINOR * 0x00100 | + HTTP_PARSER_VERSION_PATCH * 0x00001; +} + +void +http_parser_set_max_header_size(uint32_t size) { + max_header_size = size; +} diff --git a/src/3rdparty/http-parser/http_parser.h b/src/3rdparty/http-parser/http_parser.h new file mode 100644 index 000000000..880ed278e --- /dev/null +++ b/src/3rdparty/http-parser/http_parser.h @@ -0,0 +1,439 @@ +/* Copyright Joyent, Inc. and other Node contributors. All rights reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to + * deal in the Software without restriction, including without limitation the + * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or + * sell copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS + * IN THE SOFTWARE. + */ +#ifndef http_parser_h +#define http_parser_h +#ifdef __cplusplus +extern "C" { +#endif + +/* Also update SONAME in the Makefile whenever you change these. */ +#define HTTP_PARSER_VERSION_MAJOR 2 +#define HTTP_PARSER_VERSION_MINOR 9 +#define HTTP_PARSER_VERSION_PATCH 0 + +#include +#if defined(_WIN32) && !defined(__MINGW32__) && \ + (!defined(_MSC_VER) || _MSC_VER<1600) && !defined(__WINE__) +#include +typedef __int8 int8_t; +typedef unsigned __int8 uint8_t; +typedef __int16 int16_t; +typedef unsigned __int16 uint16_t; +typedef __int32 int32_t; +typedef unsigned __int32 uint32_t; +typedef __int64 int64_t; +typedef unsigned __int64 uint64_t; +#else +#include +#endif + +/* Compile with -DHTTP_PARSER_STRICT=0 to make less checks, but run + * faster + */ +#ifndef HTTP_PARSER_STRICT +# define HTTP_PARSER_STRICT 1 +#endif + +/* Maximium header size allowed. If the macro is not defined + * before including this header then the default is used. To + * change the maximum header size, define the macro in the build + * environment (e.g. -DHTTP_MAX_HEADER_SIZE=). To remove + * the effective limit on the size of the header, define the macro + * to a very large number (e.g. -DHTTP_MAX_HEADER_SIZE=0x7fffffff) + */ +#ifndef HTTP_MAX_HEADER_SIZE +# define HTTP_MAX_HEADER_SIZE (80*1024) +#endif + +typedef struct http_parser http_parser; +typedef struct http_parser_settings http_parser_settings; + + +/* Callbacks should return non-zero to indicate an error. The parser will + * then halt execution. + * + * The one exception is on_headers_complete. In a HTTP_RESPONSE parser + * returning '1' from on_headers_complete will tell the parser that it + * should not expect a body. This is used when receiving a response to a + * HEAD request which may contain 'Content-Length' or 'Transfer-Encoding: + * chunked' headers that indicate the presence of a body. + * + * Returning `2` from on_headers_complete will tell parser that it should not + * expect neither a body nor any futher responses on this connection. This is + * useful for handling responses to a CONNECT request which may not contain + * `Upgrade` or `Connection: upgrade` headers. + * + * http_data_cb does not return data chunks. It will be called arbitrarily + * many times for each string. E.G. you might get 10 callbacks for "on_url" + * each providing just a few characters more data. + */ +typedef int (*http_data_cb) (http_parser*, const char *at, size_t length); +typedef int (*http_cb) (http_parser*); + + +/* Status Codes */ +#define HTTP_STATUS_MAP(XX) \ + XX(100, CONTINUE, Continue) \ + XX(101, SWITCHING_PROTOCOLS, Switching Protocols) \ + XX(102, PROCESSING, Processing) \ + XX(200, OK, OK) \ + XX(201, CREATED, Created) \ + XX(202, ACCEPTED, Accepted) \ + XX(203, NON_AUTHORITATIVE_INFORMATION, Non-Authoritative Information) \ + XX(204, NO_CONTENT, No Content) \ + XX(205, RESET_CONTENT, Reset Content) \ + XX(206, PARTIAL_CONTENT, Partial Content) \ + XX(207, MULTI_STATUS, Multi-Status) \ + XX(208, ALREADY_REPORTED, Already Reported) \ + XX(226, IM_USED, IM Used) \ + XX(300, MULTIPLE_CHOICES, Multiple Choices) \ + XX(301, MOVED_PERMANENTLY, Moved Permanently) \ + XX(302, FOUND, Found) \ + XX(303, SEE_OTHER, See Other) \ + XX(304, NOT_MODIFIED, Not Modified) \ + XX(305, USE_PROXY, Use Proxy) \ + XX(307, TEMPORARY_REDIRECT, Temporary Redirect) \ + XX(308, PERMANENT_REDIRECT, Permanent Redirect) \ + XX(400, BAD_REQUEST, Bad Request) \ + XX(401, UNAUTHORIZED, Unauthorized) \ + XX(402, PAYMENT_REQUIRED, Payment Required) \ + XX(403, FORBIDDEN, Forbidden) \ + XX(404, NOT_FOUND, Not Found) \ + XX(405, METHOD_NOT_ALLOWED, Method Not Allowed) \ + XX(406, NOT_ACCEPTABLE, Not Acceptable) \ + XX(407, PROXY_AUTHENTICATION_REQUIRED, Proxy Authentication Required) \ + XX(408, REQUEST_TIMEOUT, Request Timeout) \ + XX(409, CONFLICT, Conflict) \ + XX(410, GONE, Gone) \ + XX(411, LENGTH_REQUIRED, Length Required) \ + XX(412, PRECONDITION_FAILED, Precondition Failed) \ + XX(413, PAYLOAD_TOO_LARGE, Payload Too Large) \ + XX(414, URI_TOO_LONG, URI Too Long) \ + XX(415, UNSUPPORTED_MEDIA_TYPE, Unsupported Media Type) \ + XX(416, RANGE_NOT_SATISFIABLE, Range Not Satisfiable) \ + XX(417, EXPECTATION_FAILED, Expectation Failed) \ + XX(421, MISDIRECTED_REQUEST, Misdirected Request) \ + XX(422, UNPROCESSABLE_ENTITY, Unprocessable Entity) \ + XX(423, LOCKED, Locked) \ + XX(424, FAILED_DEPENDENCY, Failed Dependency) \ + XX(426, UPGRADE_REQUIRED, Upgrade Required) \ + XX(428, PRECONDITION_REQUIRED, Precondition Required) \ + XX(429, TOO_MANY_REQUESTS, Too Many Requests) \ + XX(431, REQUEST_HEADER_FIELDS_TOO_LARGE, Request Header Fields Too Large) \ + XX(451, UNAVAILABLE_FOR_LEGAL_REASONS, Unavailable For Legal Reasons) \ + XX(500, INTERNAL_SERVER_ERROR, Internal Server Error) \ + XX(501, NOT_IMPLEMENTED, Not Implemented) \ + XX(502, BAD_GATEWAY, Bad Gateway) \ + XX(503, SERVICE_UNAVAILABLE, Service Unavailable) \ + XX(504, GATEWAY_TIMEOUT, Gateway Timeout) \ + XX(505, HTTP_VERSION_NOT_SUPPORTED, HTTP Version Not Supported) \ + XX(506, VARIANT_ALSO_NEGOTIATES, Variant Also Negotiates) \ + XX(507, INSUFFICIENT_STORAGE, Insufficient Storage) \ + XX(508, LOOP_DETECTED, Loop Detected) \ + XX(510, NOT_EXTENDED, Not Extended) \ + XX(511, NETWORK_AUTHENTICATION_REQUIRED, Network Authentication Required) \ + +enum http_status + { +#define XX(num, name, string) HTTP_STATUS_##name = num, + HTTP_STATUS_MAP(XX) +#undef XX + }; + + +/* Request Methods */ +#define HTTP_METHOD_MAP(XX) \ + XX(0, DELETE, DELETE) \ + XX(1, GET, GET) \ + XX(2, HEAD, HEAD) \ + XX(3, POST, POST) \ + XX(4, PUT, PUT) \ + /* pathological */ \ + XX(5, CONNECT, CONNECT) \ + XX(6, OPTIONS, OPTIONS) \ + XX(7, TRACE, TRACE) \ + /* WebDAV */ \ + XX(8, COPY, COPY) \ + XX(9, LOCK, LOCK) \ + XX(10, MKCOL, MKCOL) \ + XX(11, MOVE, MOVE) \ + XX(12, PROPFIND, PROPFIND) \ + XX(13, PROPPATCH, PROPPATCH) \ + XX(14, SEARCH, SEARCH) \ + XX(15, UNLOCK, UNLOCK) \ + XX(16, BIND, BIND) \ + XX(17, REBIND, REBIND) \ + XX(18, UNBIND, UNBIND) \ + XX(19, ACL, ACL) \ + /* subversion */ \ + XX(20, REPORT, REPORT) \ + XX(21, MKACTIVITY, MKACTIVITY) \ + XX(22, CHECKOUT, CHECKOUT) \ + XX(23, MERGE, MERGE) \ + /* upnp */ \ + XX(24, MSEARCH, M-SEARCH) \ + XX(25, NOTIFY, NOTIFY) \ + XX(26, SUBSCRIBE, SUBSCRIBE) \ + XX(27, UNSUBSCRIBE, UNSUBSCRIBE) \ + /* RFC-5789 */ \ + XX(28, PATCH, PATCH) \ + XX(29, PURGE, PURGE) \ + /* CalDAV */ \ + XX(30, MKCALENDAR, MKCALENDAR) \ + /* RFC-2068, section 19.6.1.2 */ \ + XX(31, LINK, LINK) \ + XX(32, UNLINK, UNLINK) \ + /* icecast */ \ + XX(33, SOURCE, SOURCE) \ + +enum http_method + { +#define XX(num, name, string) HTTP_##name = num, + HTTP_METHOD_MAP(XX) +#undef XX + }; + + +enum http_parser_type { HTTP_REQUEST, HTTP_RESPONSE, HTTP_BOTH }; + + +/* Flag values for http_parser.flags field */ +enum flags + { F_CHUNKED = 1 << 0 + , F_CONNECTION_KEEP_ALIVE = 1 << 1 + , F_CONNECTION_CLOSE = 1 << 2 + , F_CONNECTION_UPGRADE = 1 << 3 + , F_TRAILING = 1 << 4 + , F_UPGRADE = 1 << 5 + , F_SKIPBODY = 1 << 6 + , F_CONTENTLENGTH = 1 << 7 + }; + + +/* Map for errno-related constants + * + * The provided argument should be a macro that takes 2 arguments. + */ +#define HTTP_ERRNO_MAP(XX) \ + /* No error */ \ + XX(OK, "success") \ + \ + /* Callback-related errors */ \ + XX(CB_message_begin, "the on_message_begin callback failed") \ + XX(CB_url, "the on_url callback failed") \ + XX(CB_header_field, "the on_header_field callback failed") \ + XX(CB_header_value, "the on_header_value callback failed") \ + XX(CB_headers_complete, "the on_headers_complete callback failed") \ + XX(CB_body, "the on_body callback failed") \ + XX(CB_message_complete, "the on_message_complete callback failed") \ + XX(CB_status, "the on_status callback failed") \ + XX(CB_chunk_header, "the on_chunk_header callback failed") \ + XX(CB_chunk_complete, "the on_chunk_complete callback failed") \ + \ + /* Parsing-related errors */ \ + XX(INVALID_EOF_STATE, "stream ended at an unexpected time") \ + XX(HEADER_OVERFLOW, \ + "too many header bytes seen; overflow detected") \ + XX(CLOSED_CONNECTION, \ + "data received after completed connection: close message") \ + XX(INVALID_VERSION, "invalid HTTP version") \ + XX(INVALID_STATUS, "invalid HTTP status code") \ + XX(INVALID_METHOD, "invalid HTTP method") \ + XX(INVALID_URL, "invalid URL") \ + XX(INVALID_HOST, "invalid host") \ + XX(INVALID_PORT, "invalid port") \ + XX(INVALID_PATH, "invalid path") \ + XX(INVALID_QUERY_STRING, "invalid query string") \ + XX(INVALID_FRAGMENT, "invalid fragment") \ + XX(LF_EXPECTED, "LF character expected") \ + XX(INVALID_HEADER_TOKEN, "invalid character in header") \ + XX(INVALID_CONTENT_LENGTH, \ + "invalid character in content-length header") \ + XX(UNEXPECTED_CONTENT_LENGTH, \ + "unexpected content-length header") \ + XX(INVALID_CHUNK_SIZE, \ + "invalid character in chunk size header") \ + XX(INVALID_CONSTANT, "invalid constant string") \ + XX(INVALID_INTERNAL_STATE, "encountered unexpected internal state")\ + XX(STRICT, "strict mode assertion failed") \ + XX(PAUSED, "parser is paused") \ + XX(UNKNOWN, "an unknown error occurred") + + +/* Define HPE_* values for each errno value above */ +#define HTTP_ERRNO_GEN(n, s) HPE_##n, +enum http_errno { + HTTP_ERRNO_MAP(HTTP_ERRNO_GEN) +}; +#undef HTTP_ERRNO_GEN + + +/* Get an http_errno value from an http_parser */ +#define HTTP_PARSER_ERRNO(p) ((enum http_errno) (p)->http_errno) + + +struct http_parser { + /** PRIVATE **/ + unsigned int type : 2; /* enum http_parser_type */ + unsigned int flags : 8; /* F_* values from 'flags' enum; semi-public */ + unsigned int state : 7; /* enum state from http_parser.c */ + unsigned int header_state : 7; /* enum header_state from http_parser.c */ + unsigned int index : 7; /* index into current matcher */ + unsigned int lenient_http_headers : 1; + + uint32_t nread; /* # bytes read in various scenarios */ + uint64_t content_length; /* # bytes in body (0 if no Content-Length header) */ + + /** READ-ONLY **/ + unsigned short http_major; + unsigned short http_minor; + unsigned int status_code : 16; /* responses only */ + unsigned int method : 8; /* requests only */ + unsigned int http_errno : 7; + + /* 1 = Upgrade header was present and the parser has exited because of that. + * 0 = No upgrade header present. + * Should be checked when http_parser_execute() returns in addition to + * error checking. + */ + unsigned int upgrade : 1; + + /** PUBLIC **/ + void *data; /* A pointer to get hook to the "connection" or "socket" object */ +}; + + +struct http_parser_settings { + http_cb on_message_begin; + http_data_cb on_url; + http_data_cb on_status; + http_data_cb on_header_field; + http_data_cb on_header_value; + http_cb on_headers_complete; + http_data_cb on_body; + http_cb on_message_complete; + /* When on_chunk_header is called, the current chunk length is stored + * in parser->content_length. + */ + http_cb on_chunk_header; + http_cb on_chunk_complete; +}; + + +enum http_parser_url_fields + { UF_SCHEMA = 0 + , UF_HOST = 1 + , UF_PORT = 2 + , UF_PATH = 3 + , UF_QUERY = 4 + , UF_FRAGMENT = 5 + , UF_USERINFO = 6 + , UF_MAX = 7 + }; + + +/* Result structure for http_parser_parse_url(). + * + * Callers should index into field_data[] with UF_* values iff field_set + * has the relevant (1 << UF_*) bit set. As a courtesy to clients (and + * because we probably have padding left over), we convert any port to + * a uint16_t. + */ +struct http_parser_url { + uint16_t field_set; /* Bitmask of (1 << UF_*) values */ + uint16_t port; /* Converted UF_PORT string */ + + struct { + uint16_t off; /* Offset into buffer in which field starts */ + uint16_t len; /* Length of run in buffer */ + } field_data[UF_MAX]; +}; + + +/* Returns the library version. Bits 16-23 contain the major version number, + * bits 8-15 the minor version number and bits 0-7 the patch level. + * Usage example: + * + * unsigned long version = http_parser_version(); + * unsigned major = (version >> 16) & 255; + * unsigned minor = (version >> 8) & 255; + * unsigned patch = version & 255; + * printf("http_parser v%u.%u.%u\n", major, minor, patch); + */ +unsigned long http_parser_version(void); + +void http_parser_init(http_parser *parser, enum http_parser_type type); + + +/* Initialize http_parser_settings members to 0 + */ +void http_parser_settings_init(http_parser_settings *settings); + + +/* Executes the parser. Returns number of parsed bytes. Sets + * `parser->http_errno` on error. */ +size_t http_parser_execute(http_parser *parser, + const http_parser_settings *settings, + const char *data, + size_t len); + + +/* If http_should_keep_alive() in the on_headers_complete or + * on_message_complete callback returns 0, then this should be + * the last message on the connection. + * If you are the server, respond with the "Connection: close" header. + * If you are the client, close the connection. + */ +int http_should_keep_alive(const http_parser *parser); + +/* Returns a string version of the HTTP method. */ +const char *http_method_str(enum http_method m); + +/* Returns a string version of the HTTP status code. */ +const char *http_status_str(enum http_status s); + +/* Return a string name of the given error */ +const char *http_errno_name(enum http_errno err); + +/* Return a string description of the given error */ +const char *http_errno_description(enum http_errno err); + +/* Initialize all http_parser_url members to 0 */ +void http_parser_url_init(struct http_parser_url *u); + +/* Parse a URL; return nonzero on failure */ +int http_parser_parse_url(const char *buf, size_t buflen, + int is_connect, + struct http_parser_url *u); + +/* Pause or un-pause the parser; a nonzero value pauses */ +void http_parser_pause(http_parser *parser, int paused); + +/* Checks if this is the final chunk of the body. */ +int http_body_is_final(const http_parser *parser); + +/* Change the maximum header size provided at compile time. */ +void http_parser_set_max_header_size(uint32_t size); + +#ifdef __cplusplus +} +#endif +#endif diff --git a/src/3rdparty/hwloc/AUTHORS b/src/3rdparty/hwloc/AUTHORS new file mode 100644 index 000000000..7187a723d --- /dev/null +++ b/src/3rdparty/hwloc/AUTHORS @@ -0,0 +1,44 @@ +hwloc Authors +============= + +The following cumulative list contains the names of most individuals +who have committed code to the hwloc repository +(either directly or through a third party). + +Name Affiliation(s) +--------------------------- -------------------- +Grzegorz Andrejczuk Intel +Cédric Augonnet University of Bordeaux +Guillaume Beauchamp Inria +Ahmad Boissetri Binzagr Inria +Cyril Bordage Inria +Nicholas Buroker UWL +Christopher M. Cantalupo Intel +Jérôme Clet-Ortega University of Bordeaux +Ludovic Courtès Inria +Clément Foyer Inria +Nathalie Furmento CNRS +Bryon Gloden +Brice Goglin Inria +Gilles Gouaillardet RIST +Joshua Hursey UWL +Alexey Kardashevskiy IBM +Rob Latham ANL +Douglas MacFarland UWL +Marc Marí BSC +Jonathan L Peyton Intel +Piotr Luc Intel +Antoine Rougier intern from University of Bordeaux +Jeff Squyres Cisco +Samuel Thibault University of Bordeaux +Jean-Yves VET DDN +Benjamin Worpitz +Jeff Zhao Zhaoxin + +Affiliaion abbreviations: +------------------------- +ANL = Argonne National Lab +BSC = Barcelona Supercomputing Center +Cisco = Cisco Systems, Inc. +CNRS = Centre national de la recherche scientifique (France) +UWL = University of Wisconsin-La Crosse diff --git a/src/3rdparty/hwloc/CMakeLists.txt b/src/3rdparty/hwloc/CMakeLists.txt new file mode 100644 index 000000000..431c11eb3 --- /dev/null +++ b/src/3rdparty/hwloc/CMakeLists.txt @@ -0,0 +1,38 @@ +cmake_minimum_required (VERSION 2.8) +project (hwloc C) + +include_directories(include) +include_directories(src) + +add_definitions(/D_CRT_SECURE_NO_WARNINGS) +set(CMAKE_C_FLAGS_RELEASE "${CMAKE_C_FLAGS_RELEASE} /MT") + +set(HEADERS + include/hwloc.h + src/static-components.h + ) + +set(SOURCES + src/base64.c + src/bind.c + src/bitmap.c + src/components.c + src/diff.c + src/distances.c + src/misc.c + src/pci-common.c + src/shmem.c + src/topology.c + src/topology-noos.c + src/topology-synthetic.c + src/topology-windows.c + src/topology-x86.c + src/topology-xml.c + src/topology-xml-nolibxml.c + src/traversal.c + ) + +add_library(hwloc STATIC + ${HEADERS} + ${SOURCES} + ) diff --git a/src/3rdparty/hwloc/COPYING b/src/3rdparty/hwloc/COPYING new file mode 100644 index 000000000..e77516e18 --- /dev/null +++ b/src/3rdparty/hwloc/COPYING @@ -0,0 +1,39 @@ +Copyright © 2004-2006 The Trustees of Indiana University and Indiana University Research and Technology Corporation. All rights reserved. +Copyright © 2004-2005 The University of Tennessee and The University of Tennessee Research Foundation. All rights reserved. +Copyright © 2004-2005 High Performance Computing Center Stuttgart, University of Stuttgart. All rights reserved. +Copyright © 2004-2005 The Regents of the University of California. All rights reserved. +Copyright © 2009 CNRS +Copyright © 2009-2016 Inria. All rights reserved. +Copyright © 2009-2015 Université Bordeaux +Copyright © 2009-2015 Cisco Systems, Inc. All rights reserved. +Copyright © 2009-2012 Oracle and/or its affiliates. All rights reserved. +Copyright © 2010 IBM +Copyright © 2010 Jirka Hladky +Copyright © 2012 Aleksej Saushev, The NetBSD Foundation +Copyright © 2012 Blue Brain Project, EPFL. All rights reserved. +Copyright © 2013-2014 University of Wisconsin-La Crosse. All rights reserved. +Copyright © 2015 Research Organization for Information Science and Technology (RIST). All rights reserved. +Copyright © 2015-2016 Intel, Inc. All rights reserved. +See COPYING in top-level directory. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions +are met: +1. Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. +2. Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. +3. The name of the author may not be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR +IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES +OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. +IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, +INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT +NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, +DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY +THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT +(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF +THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. diff --git a/src/3rdparty/hwloc/NEWS b/src/3rdparty/hwloc/NEWS new file mode 100644 index 000000000..664c8d55c --- /dev/null +++ b/src/3rdparty/hwloc/NEWS @@ -0,0 +1,1599 @@ +Copyright © 2009 CNRS +Copyright © 2009-2019 Inria. All rights reserved. +Copyright © 2009-2013 Université Bordeaux +Copyright © 2009-2011 Cisco Systems, Inc. All rights reserved. + +$COPYRIGHT$ + +Additional copyrights may follow + +$HEADER$ + +=========================================================================== + +This file contains the main features as well as overviews of specific +bug fixes (and other actions) for each version of hwloc since version +0.9 (as initially released as "libtopology", then re-branded to "hwloc" +in v0.9.1). + + +Version 2.0.4 (also included in 1.11.13 when appropriate) +------------- +* Add support for Linux 5.3 new sysfs cpu topology files with Die information. +* Add support for Intel v2 Extended Topology Enumeration in the x86 backend. +* Tiles, Modules and Dies are exposed as Groups for now. + + HWLOC_DONT_MERGE_DIE_GROUPS=1 may be set in the environment to prevent + Die groups from being automatically merged with identical parent or children. +* Ignore NUMA node information from AMD topoext in the x86 backend, + unless HWLOC_X86_TOPOEXT_NUMANODES=1 is set in the environment. +* Group objects have a new "dont_merge" attribute to prevent them from + being automatically merged with identical parent or children. + + +Version 2.0.3 (also included in 1.11.12 when appropriate) +------------- +* Fix build on Cygwin, thanks to Marco Atzeri for the patches. +* Fix a corner case of hwloc_topology_restrict() where children would + become out-of-order. +* Fix the return length of export_xmlbuffer() functions to always + include the ending \0. +* Fix lstopo --children-order argument parsing. + + +Version 2.0.2 (also included in 1.11.11 when appropriate) +------------- +* Add support for Hygon Dhyana processors in the x86 backend, + thanks to Pu Wen for the patch. +* Fix symbol renaming to also rename internal components, + thanks to Evan Ramos for the patch. +* Fix build on HP-UX, thanks to Richard Lloyd for reporting the issues. +* Detect PCI link speed without being root on Linux >= 4.13. +* Add HWLOC_VERSION* macros to the public headers, + thanks to Gilles Gouaillardet for the suggestion. + + +Version 2.0.1 (also included in 1.11.10 when relevant) +------------- +* Bump the library soname to 15:0:0 to avoid conflicts with hwloc 1.11.x + releases. The hwloc 2.0.0 soname was buggy (12:0:0), applications will + have to be recompiled. +* Serialize pciaccess discovery to fix concurrent topology loads in + multiple threads. +* Fix hwloc-dump-hwdata to only process SMBIOS information that correspond + to the KNL and KNM configuration. +* Add a heuristic for guessing KNL/KNM memory and cluster modes when + hwloc-dump-hwdata could not run as root earlier. +* Add --no-text lstopo option to remove text from some boxes in the + graphical output. Mostly useful for removing Group labels. +* Some minor fixes to memory binding. + + +Version 2.0.0 +------------- +*** The ABI of the library has changed. *** + For instance some hwloc_obj fields were reordered, added or removed, see below. + + HWLOC_API_VERSION and hwloc_get_api_version() now give 0x00020000. + + See "How do I handle ABI breaks and API upgrades ?" in the FAQ + and "Upgrading to hwloc 2.0 API" in the documentation. +* Major API changes + + Memory, I/O and Misc objects are now stored in dedicated children lists, + not in the usual children list that is now only used for CPU-side objects. + - hwloc_get_next_child() may still be used to iterate over these 4 lists + of children at once. + - hwloc_obj_type_is_normal(), _memory() and _io() may be used to check + the kind of a given object type. + + Topologies always have at least one NUMA object. On non-NUMA machines, + a single NUMA object is added to describe the entire machine memory. + The NUMA level cannot be ignored anymore. + + The NUMA level is special since NUMA nodes are not in the main hierarchy + of objects anymore. Its depth is a fake negative depth that should not be + compared with normal levels. + - If all memory objects are attached to parents at the same depth, + it may be retrieved with hwloc_get_memory_parents_depth(). + + The HWLOC_OBJ_CACHE type is replaced with 8 types HWLOC_OBJ_L[1-5]CACHE + and HWLOC_OBJ_L[1-3]ICACHE that remove the need to disambiguate levels + when looking for caches with _by_type() functions. + - New hwloc_obj_type_is_{,d,i}cache() functions may be used to check whether + a given type is a cache. + + Reworked ignoring/filtering API + - Replace hwloc_topology_ignore*() functions with hwloc_topology_set_type_filter() + and hwloc_topology_set_all_types_filter(). + . Contrary to hwloc_topology_ignore_{type,all}_keep_structure() which + removed individual objects, HWLOC_TYPE_FILTER_KEEP_STRUCTURE only removes + entire levels (so that topology do not become too asymmetric). + - Remove HWLOC_TOPOLOGY_FLAG_ICACHES in favor of hwloc_topology_set_icache_types_filter() + with HWLOC_TYPE_FILTER_KEEP_ALL. + - Remove HWLOC_TOPOLOGY_FLAG_IO_DEVICES, _IO_BRIDGES and _WHOLE_IO in favor of + hwloc_topology_set_io_types_filter() with HWLOC_TYPE_FILTER_KEEP_ALL or + HWLOC_TYPE_FILTER_KEEP_IMPORTANT. + + The distance API has been completely reworked. It is now described + in hwloc/distances.h. + + Return values + - Most functions in hwloc/bitmap.h now return an int that may be negative + in case of failure to realloc/extend the internal storage of a bitmap. + - hwloc_obj_add_info() also returns an int in case allocations fail. +* Minor API changes + + Object attributes + - obj->memory is removed. + . local_memory and page_types attributes are now in obj->attr->numanode + . total_memory moves obj->total_memory. + - Objects do not have allowed_cpuset and allowed_nodeset anymore. + They are only available for the entire topology using + hwloc_topology_get_allowed_cpuset() and hwloc_topology_get_allowed_nodeset(). + - Objects now have a "subtype" field that supersedes former "Type" and + "CoProcType" info attributes. + + Object and level depths are now signed ints. + + Object string printing and parsing + - hwloc_type_sscanf() deprecates the old hwloc_obj_type_sscanf(). + - hwloc_type_sscanf_as_depth() is added to convert a type name into + a level depth. + - hwloc_obj_cpuset_snprintf() is deprecated in favor of hwloc_bitmap_snprintf(). + + Misc objects + - Replace hwloc_topology_insert_misc_object_by_cpuset() with + hwloc_topology_insert_group_object() to precisely specify the location + of an additional hierarchy level in the topology. + - Misc objects have their own level and depth to iterate over all of them. + - Misc objects may now only be inserted as a leaf object with + hwloc_topology_insert_misc_object() which deprecates + hwloc_topology_insert_misc_object_by_parent(). + + hwloc_topology_restrict() doesn't remove objects that contain memory + by default anymore. + - The list of existing restrict flags was modified. + + The discovery support array now contains some NUMA specific bits. + + XML export functions take an additional flags argument, + for instance for exporting XMLs that are compatible with hwloc 1.x. + + Functions diff_load_xml*(), diff_export_xml*() and diff_destroy() in + hwloc/diff.h do not need a topology as first parameter anymore. + + hwloc_parse_cpumap_file () superseded by hwloc_linux_read_path_as_cpumask() + in hwloc/linux.h. + + HWLOC_MEMBIND_DEFAULT and HWLOC_MEMBIND_FIRSTTOUCH were clarified. +* New APIs and Features + + Add hwloc/shmem.h for sharing topologies between processes running on + the same machine (for reducing the memory footprint). + + Add the experimental netloc subproject. It is disabled by default + and can be enabled with --enable-netloc. + It currently brings command-line tools to gather and visualize the + topology of InfiniBand fabrics, and an API to convert such topologies + into Scotch architectures for process mapping. + See the documentation for details. +* Removed APIs and features + + Remove the online_cpuset from struct hwloc_obj. Offline PUs get unknown + topologies on Linux nowadays, and wrong topology on Solaris. Other OS + do not support them. And one cannot do much about them anyway. Just keep + them in complete_cpuset. + + Remove the now-unused "System" object type HWLOC_OBJ_SYSTEM, + defined to MACHINE for backward compatibility. + + The almost-unused "os_level" attribute has been removed from the + hwloc_obj structure. + + Remove the custom interface for assembling the topologies of different + nodes as well as the hwloc-assembler tools. + + hwloc_topology_set_fsroot() is removed, the environment variable + HWLOC_FSROOT may be used for the same remote testing/debugging purpose. + + Remove the deprecated hwloc_obj_snprintf(), hwloc_obj_type_of_string(), + hwloc_distribute[v](). + * Remove Myrinet Express interoperability (hwloc/myriexpress.h). + + Remove Kerrighed support from the Linux backend. + + Remove Tru64 (OSF/1) support. + - Remove HWLOC_MEMBIND_REPLICATE which wasn't available anywhere else. +* Backend improvements + + Linux + - OS devices do not have to be attached through PCI anymore, + for instance enabling the discovery of NVDIMM block devices. + - Remove the dependency on libnuma. + - Add a SectorSize attribute to block OS devices. + + Mac OS X + - Fix detection of cores and hyperthreads. + - Add CPUVendor, Model, ... attributes. + + Windows + - Add get_area_memlocation(). +* Tools + + lstopo and hwloc-info have a new --filter option matching the new filtering API. + + lstopo can be given --children-order=plain to force a basic displaying + of memory and normal children together below their parent. + + hwloc-distances was removed and replaced with lstopo --distances. +* Misc + + Exports + - Exporting to synthetic now ignores I/O and Misc objects. + + PCI discovery + - Separate OS device discovery from PCI discovery. Only the latter is disabled + with --disable-pci at configure time. Both may be disabled with --disable-io. + - The `linuxpci' component is now renamed into `linuxio'. + - The old `libpci' component name from hwloc 1.6 is not supported anymore, + only the `pci' name from hwloc 1.7 is now recognized. + - The HWLOC_PCI___LOCALCPUS environment variables are superseded + with a single HWLOC_PCI_LOCALITY where bus ranges may be specified. + - Do not set PCI devices and bridges name automatically. Vendor and device + names are already in info attributes. + + Components and discovery + - Add HWLOC_SYNTHETIC environment variable to enforce a synthetic topology + as if hwloc_topology_set_synthetic() had been called. + - HWLOC_COMPONENTS doesn't support xml or synthetic component attributes + anymore, they should be passed in HWLOC_XMLFILE or HWLOC_SYNTHETIC instead. + - HWLOC_COMPONENTS takes precedence over other environment variables + for selecting components. + + hwloc now requires a C99 compliant compiler. + + +Version 1.11.9 +-------------- +* Add support for Zhaoxin ZX-C and ZX-D processors in the x86 backend, + thanks to Jeff Zhao for the patch. +* Fix AMD Epyc 24-core L3 cache locality in the x86 backend. +* Don't crash in the x86 backend when the CPUID vendor string is unknown. +* Fix the missing pu discovery support bit on some OS. +* Fix the management of the lstopoStyle info attribute for custom colors. +* Add verbose warnings when failing to load hwloc v2.0+ XMLs. + + +Version 1.11.8 +-------------- +* Multiple Solaris improvements, thanks to Maureen Chew for the help: + + Detect caches on Sparc. + + Properly detect allowed/disallowed PUs and NUMA nodes with processor sets. + + Add hwloc_get_last_cpu_location() support for the current thread. +* Add support for CUDA compute capability 7.0 and fix support for 6.[12]. +* Tools improvements + + Fix search for objects by physical index in command-line tools. + + Add missing "cpubind:get_thisthread_last_cpu_location" in the output + of hwloc-info --support. + + Add --pid and --name to specify target processes in hwloc-ps. + + Display thread names in lstopo and hwloc-ps on Linux. +* Doc improvements + + Add a FAQ entry about building on Windows. + + Install missing sub-manpage for hwloc_obj_add_info() and + hwloc_obj_get_info_by_name(). + + +Version 1.11.7 +-------------- +* Fix hwloc-bind --membind for CPU-less NUMA nodes (again). + Thanks to Gilles Gouaillardet for reporting the issue. +* Fix a memory leak on IBM S/390 platforms running Linux. +* Fix a memory leak when forcing the x86 backend first on amd64/topoext + platforms running Linux. +* Command-line tools now support "hbm" instead "numanode" for filtering + only high-bandwidth memory nodes when selecting locations. + + hwloc-bind also support --hbm and --no-hbm for filtering only or + no HBM nodes. + Thanks to Nicolas Denoyelle for the suggestion. +* Add --children and --descendants to hwloc-info for listing object + children or object descendants of a specific type. +* Add --no-index, --index, --no-attrs, --attrs to disable/enable display + of index numbers or attributes in the graphical lstopo output. +* Try to gather hwloc-dump-hwdata output from all possible locations + in hwloc-gather-topology. +* Updates to the documentation of locations in hwloc(7) and + command-line tools manpages. + + +Version 1.11.6 +-------------- +* Make the Linux discovery about twice faster, especially on the CPU side, + by trying to avoid sysfs file accesses as much as possible. +* Add support for AMD Family 17h processors (Zen) SMT cores in the Linux + and x86 backends. +* Add the HWLOC_TOPOLOGY_FLAG_THISSYSTEM_ALLOWED_RESOURCES flag (and the + HWLOC_THISSYSTEM_ALLOWED_RESOURCES environment variable) for reading the + set of allowed resources from the local operating system even if the + topology was loaded from XML or synthetic. +* Fix hwloc_bitmap_set/clr_range() for infinite ranges that do not + overlap currently defined ranges in the bitmap. +* Don't reset the lstopo zoom scale when moving the X11 window. +* lstopo now has --flags for manually setting topology flags. +* hwloc_get_depth_type() returns HWLOC_TYPE_DEPTH_UNKNOWN for Misc objects. + + +Version 1.11.5 +-------------- +* Add support for Knights Mill Xeon Phi, thanks to Piotr Luc for the patch. +* Reenable distance gathering on Solaris, disabled by mistake since v1.0. + Thanks to TU Wien for the help. +* Fix hwloc_get_*obj*_inside_cpuset() functions to ignore objects with + empty CPU sets, for instance, CPU-less NUMA nodes such as KNL MCDRAM. + Thanks to Nicolas Denoyelle for the report. +* Fix XML import of multiple distance matrices. +* Add a FAQ entry about "hwloc is only a structural model, it ignores + performance models, memory bandwidth, etc.?" + + +Version 1.11.4 +-------------- +* Add MemoryMode and ClusterMode attributes in the Machine object on KNL. + Add doc/examples/get-knl-modes.c for an example of retrieving them. + Thanks to Grzegorz Andrejczuk. +* Fix Linux build with -m32 with respect to libudev. + Thanks to Paul Hargrove for reporting the issue. +* Fix build with Visual Studio 2015, thanks to Eloi Gaudry for reporting + the issue and providing the patch. +* Don't forget to display OS device children in the graphical lstopo. +* Fix a memory leak on Solaris, thanks to Bryon Gloden for the patch. +* Properly handle realloc() failures, thanks to Bryon Gloden for reporting + the issue. +* Fix lstopo crash in ascii/fig/windows outputs when some objects have a + lstopoStyle info attribute. + + +Version 1.11.3 +-------------- +* Bug fixes + + Fix a memory leak on Linux S/390 hosts with books. + + Fix /proc/mounts parsing on Linux by using mntent.h. + Thanks to Nathan Hjelm for reporting the issue. + + Fix a x86 infinite loop on VMware due to the x2APIC feature being + advertised without actually being fully supported. + Thanks to Jianjun Wen for reporting the problem and testing the patch. + + Fix the return value of hwloc_alloc() on mmap() failure. + Thanks to Hugo Brunie for reporting the issue. + + Fix the return value of command-line tools in some error cases. + + Do not break individual thread bindings during x86 backend discovery in a + multithreaded process. Thanks to Farouk Mansouri for the report. + + Fix hwloc-bind --membind for CPU-less NUMA nodes. + + Fix some corner cases in the XML export/import of application userdata. +* API Improvements + + Add HWLOC_MEMBIND_BYNODESET flag so that membind() functions accept + either cpusets or nodesets. + + Add hwloc_get_area_memlocation() to check where pages are actually + allocated. Only implemented on Linux for now. + - There's no _nodeset() variant, but the new flag HWLOC_MEMBIND_BYNODESET + is supported. + + Make hwloc_obj_type_sscanf() parse back everything that may be outputted + by hwloc_obj_type_snprintf(). +* Detection Improvements + + Allow the x86 backend to add missing cache levels, so that it completes + what the Solaris backend lacks. + Thanks to Ryan Zezeski for reporting the issue. + + Do not filter-out FibreChannel PCI adapters by default anymore. + Thanks to Matt Muggeridge for the report. + + Add support for CUDA compute capability 6.x. +* Tools + + Add --support to hwloc-info to list supported features, just like with + hwloc_topology_get_support(). + - Also add --objects and --topology to explicitly switch between the + default modes. + + Add --tid to let hwloc-bind operate on individual threads on Linux. + + Add --nodeset to let hwloc-bind report memory binding as NUMA node sets. + + hwloc-annotate and lstopo don't drop application userdata from XMLs anymore. + - Add --cu to hwloc-annotate to drop these application userdata. + + Make the hwloc-dump-hwdata dump directory configurable through configure + options such as --runstatedir or --localstatedir. +* Misc Improvements + + Add systemd service template contrib/systemd/hwloc-dump-hwdata.service + for launching hwloc-dump-hwdata at boot on Linux. + Thanks to Grzegorz Andrejczuk. + + Add HWLOC_PLUGINS_BLACKLIST environment variable to prevent some plugins + from being loaded. Thanks to Alexandre Denis for the suggestion. + + Small improvements for various Windows build systems, + thanks to Jonathan L Peyton and Marco Atzeri. + + +Version 1.11.2 +-------------- +* Improve support for Intel Knights Landing Xeon Phi on Linux: + + Group local NUMA nodes of normal memory (DDR) and high-bandwidth memory + (MCDRAM) together through "Cluster" groups so that the local MCDRAM is + easy to find. + - See "How do I find the local MCDRAM NUMA node on Intel Knights + Landing Xeon Phi?" in the documentation. + - For uniformity across all KNL configurations, always have a NUMA node + object even if the host is UMA. + + Fix the detection of the memory-side cache: + - Add the hwloc-dump-hwdata superuser utility to dump SMBIOS information + into /var/run/hwloc/ as root during boot, and load this dumped + information from the hwloc library at runtime. + - See "Why do I need hwloc-dump-hwdata for caches on Intel Knights + Landing Xeon Phi?" in the documentation. + Thanks to Grzegorz Andrejczuk for the patches and for the help. +* The x86 and linux backends may now be combined for discovering CPUs + through x86 CPUID and memory from the Linux kernel. + This is useful for working around buggy CPU information reported by Linux + (for instance the AMD Bulldozer/Piledriver bug below). + Combination is enabled by passing HWLOC_COMPONENTS=x86 in the environment. +* Fix L3 cache sharing on AMD Opteron 63xx (Piledriver) and 62xx (Bulldozer) + in the x86 backend. Thanks to many users who helped. +* Fix the overzealous L3 cache sharing fix added to the x86 backend in 1.11.1 + for AMD Opteron 61xx (Magny-Cours) processors. +* The x86 backend may now add the info attribute Inclusive=0 or 1 to caches + it discovers, or to caches discovered by other backends earlier. + Thanks to Guillaume Beauchamp for the patch. +* Fix the management on alloc_membind() allocation failures on AIX, HP-UX + and OSF/Tru64. +* Fix spurious failures to load with ENOMEM on AIX in case of Misc objects + below PUs. +* lstopo improvements in X11 and Windows graphical mode: + + Add + - f 1 shortcuts to manually zoom-in, zoom-out, reset the scale, + or fit the entire window. + + Display all keyboard shortcuts in the console. +* Debug messages may be disabled at runtime by passing HWLOC_DEBUG_VERBOSE=0 + in the environment when --enable-debug was passed to configure. +* Add a FAQ entry "What are these Group objects in my topology?". + + +Version 1.11.1 +-------------- +* Detection fixes + + Hardwire the topology of Fujitsu K-computer, FX10, FX100 servers to + workaround buggy Linux kernels. + Thanks to Takahiro Kawashima and Gilles Gouaillardet. + + Fix L3 cache information on AMD Opteron 61xx Magny-Cours processors + in the x86 backend. Thanks to Guillaume Beauchamp for the patch. + + Detect block devices directly attached to PCI without a controller, + for instance NVMe disks. Thanks to Barry M. Tannenbaum. + + Add the PCISlot attribute to all PCI functions instead of only the + first one. +* Miscellaneous internal fixes + + Ignore PCI bridges that could fail assertions by reporting buggy + secondary-subordinate bus numbers + Thanks to George Bosilca for reporting the issue. + + Fix an overzealous assertion when inserting an intermediate Group object + while Groups are totally ignored. + + Fix a memory leak on Linux on AMD processors with dual-core compute units. + Thanks to Bob Benner. + + Fix a memory leak on failure to load a xml diff file. + + Fix some segfaults when inputting an invalid synthetic description. + + Fix a segfault when plugins fail to find core symbols. + Thanks to Guy Streeter. +* Many fixes and improvements in the Windows backend: + + Fix the discovery of more than 32 processors and multiple processor + groups. Thanks to Barry M. Tannenbaum for the help. + + Add thread binding set support in case of multiple process groups. + + Add thread binding get support. + + Add get_last_cpu_location() support for the current thread. + + Disable the unsupported process binding in case of multiple processor + groups. + + Fix/update the Visual Studio support under contrib/windows. + Thanks to Eloi Gaudry for the help. +* Tools fixes + + Fix a segfault when displaying logical indexes in the graphical lstopo. + Thanks to Guillaume Mercier for reporting the issue. + + Fix lstopo linking with X11 libraries, for instance on Mac OS X. + Thanks to Scott Atchley and Pierre Ramet for reporting the issue. + + hwloc-annotate, hwloc-diff and hwloc-patch do not drop unavailable + resources from the output anymore and those may be annotated as well. + + Command-line tools may now import XML from the standard input with -i -.xml + + Add missing documentation for the hwloc-info --no-icaches option. + + +Version 1.11.0 +-------------- +* API + + Socket objects are renamed into Package to align with the terminology + used by processor vendors. The old HWLOC_OBJ_SOCKET type and "Socket" + name are still supported for backward compatibility. + + HWLOC_OBJ_NODE is replaced with HWLOC_OBJ_NUMANODE for clarification. + HWLOC_OBJ_NODE is still supported for backward compatibility. + "Node" and "NUMANode" strings are supported as in earlier releases. +* Detection improvements + + Add support for Intel Knights Landing Xeon Phi. + Thanks to Grzegorz Andrejczuk and Lukasz Anaczkowski. + + Add Vendor, Model, Revision, SerialNumber, Type and LinuxDeviceID + info attributes to Block OS devices on Linux. Thanks to Vineet Pedaballe + for the help. + - Add --disable-libudev to avoid dependency on the libudev library. + + Add "MemoryModule" Misc objects with information about DIMMs, on Linux + when privileged and when I/O is enabled. + Thanks to Vineet Pedaballe for the help. + + Add a PCISlot attribute to PCI devices on Linux when supported to + identify the physical PCI slot where the board is plugged. + + Add CPUStepping info attribute on x86 processors, + thanks to Thomas Röhl for the suggestion. + + Ignore the device-tree on non-Power architectures to avoid buggy + detection on ARM. Thanks to Orion Poplawski for reporting the issue. + + Work-around buggy Xeon E5v3 BIOS reporting invalid PCI-NUMA affinity + for the PCI links on the second processor. + + Add support for CUDA compute capability 5.x, thanks Benjamin Worpitz. + + Many fixes to the x86 backend + - Add L1i and fix L2/L3 type on old AMD processors without topoext support. + - Fix Intel CPU family and model numbers when basic family isn't 6 or 15. + - Fix package IDs on recent AMD processors. + - Fix misc issues due to incomplete APIC IDs on x2APIC processors. + - Avoid buggy discovery on old SGI Altix UVs with non-unique APIC IDs. + + Gather total machine memory on NetBSD. +* Tools + + lstopo + - Collapse identical PCI devices unless --no-collapse is given. + This avoids gigantic outputs when a PCI device contains dozens of + identical virtual functions. + - The ASCII art output is now called "ascii", for instance in + "lstopo -.ascii". + The former "txt" extension is retained for backward compatibility. + - Automatically scales graphical box width to the inner text in Cairo, + ASCII and Windows outputs. + - Add --rect to lstopo to force rectangular layout even for NUMA nodes. + - Add --restrict-flags to configure the behavior of --restrict. + - Objects may have a "Type" info attribute to specify a better type name + and display it in lstopo. + - Really export all verbose information to the given output file. + + hwloc-annotate + - May now operate on all types of objects, including I/O. + - May now insert Misc objects in the topology. + - Do not drop instruction caches and I/O devices from the output anymore. + + Fix lstopo path in hwloc-gather-topology after install. +* Misc + + Fix hwloc/cudart.h for machines with multiple PCI domains, + thanks to Imre Kerr for reporting the problem. + + Fix PCI Bridge-specific depth attribute. + + Fix hwloc_bitmap_intersect() for two infinite bitmaps. + + Fix some corner cases in the building of levels on large NUMA machines + with non-uniform NUMA groups and I/Os. + + Improve the performance of object insertion by cpuset for large + topologies. + + Prefix verbose XML import errors with the source name. + + Improve pkg-config checks and error messages. + + Fix excluding after a component with an argument in the HWLOC_COMPONENTS + environment variable. +* Documentation + + Fix the recommended way in documentation and examples to allocate memory + on some node, it should use HWLOC_MEMBIND_BIND. + Thanks to Nicolas Bouzat for reporting the issue. + + Add a "Miscellaneous objects" section in the documentation. + + Add a FAQ entry "What happens to my topology if I disable symmetric + multithreading, hyper-threading, etc. ?" to the documentation. + + +Version 1.10.1 +-------------- +* Actually remove disallowed NUMA nodes from nodesets when the whole-system + flag isn't enabled. +* Fix the gathering of PCI domains. Thanks to James Custer for reporting + the issue and providing a patch. +* Fix the merging of identical parent and child in presence of Misc objects. + Thanks to Dave Love for reporting the issue. +* Fix some misordering of children when merging with ignore_keep_structure() + in partially allowed topologies. +* Fix an overzealous assertion in the debug code when running on a single-PU + host with I/O. Thanks to Thomas Van Doren for reporting the issue. +* Don't forget to setup NUMA node object nodesets in x86 backend (for BSDs) + and OSF/Tru64 backend. +* Fix cpuid-x86 build error with gcc -O3 on x86-32. Thanks to Thomas Van Doren + for reporting the issue. +* Fix support for future very large caches in the x86 backend. +* Fix vendor/device names for SR-IOV PCI devices on Linux. +* Fix an unlikely crash in case of buggy hierarchical distance matrix. +* Fix PU os_index on some AIX releases. Thanks to Hendryk Bockelmann and + Erik Schnetter for helping debugging. +* Fix hwloc_bitmap_isincluded() in case of infinite sets. +* Change hwloc-ls.desktop into a lstopo.desktop and only install it if + lstopo is built with Cairo/X11 support. It cannot work with a non-graphical + lstopo or hwloc-ls. +* Add support for the renaming of Socket into Package in future releases. +* Add support for the replacement of HWLOC_OBJ_NODE with HWLOC_OBJ_NUMANODE + in future releases. +* Clarify the documentation of distance matrices in hwloc.h and in the manpage + of the hwloc-distances. Thanks to Dave Love for the suggestion. +* Improve some error messages by displaying more information about the + hwloc library in use. +* Document how to deal with the ABI break when upgrading to the upcoming 2.0 + See "How do I handle ABI breaks and API upgrades ?" in the FAQ. + + +Version 1.10.0 +-------------- +* API + + Add hwloc_topology_export_synthetic() to export a topology to a + synthetic string without using lstopo. See the Synthetic topologies + section in the documentation. + + Add hwloc_topology_set/get_userdata() to let the application save + a private pointer in the topology whenever it needs a way to find + its own object corresponding to a topology. + + Add hwloc_get_numanode_obj_by_os_index() and document that this function + as well as hwloc_get_pu_obj_by_os_index() are good at converting + nodesets and cpusets into objects. + + hwloc_distrib() does not ignore any objects anymore when there are + too many of them. They get merged with others instead. + Thanks to Tim Creech for reporting the issue. +* Tools + + hwloc-bind --get now executes the command after displaying + the binding instead of ignoring the command entirely. + Thanks to John Donners for the suggestion. + + Clarify that memory sizes shown in lstopo are local by default + unless specified (total memory added in the root object). +* Synthetic topologies + + Synthetic topology descriptions may now specify attributes such as + memory sizes and OS indexes. See the Synthetic topologies section + in the documentation. + + lstopo now exports in this fully-detailed format by default. + The new option --export-synthetic-flags may be used to revert + back the old format. +* Documentation + + Add the doc/examples/ subdirectory with several real-life examples, + including the already existing hwloc-hello.C for basics. + Thanks to Rob Aulwes for the suggestion. + + Improve the documentation of CPU and memory binding in the API. + + Add a FAQ entry about operating system errors, especially on AMD + platforms with buggy cache information. + + Add a FAQ entry about loading many topologies in a single program. +* Misc + + Work around buggy Linux kernels reporting 2 sockets instead + 1 socket with 2 NUMA nodes for each Xeon E5 v3 (Haswell) processor. + + pciutils/libpci support is now removed since libpciaccess works + well and there's also a Linux-specific PCI backend. For the record, + pciutils was GPL and therefore disabled by default since v1.6.2. + + Add --disable-cpuid configure flag to work around buggy processor + simulators reporting invalid CPUID information. + Thanks for Andrew Friedley for reporting the issue. + + Fix a racy use of libltdl when manipulating multiple topologies in + different threads. + Thanks to Andra Hugo for reporting the issue and testing patches. + + Fix some build failures in private/misc.h. + Thanks to Pavan Balaji and Ralph Castain for the reports. + + Fix failures to detect X11/Xutil.h on some Solaris platforms. + Thanks to Siegmar Gross for reporting the failure. + + The plugin ABI has changed, this release will not load plugins + built against previous hwloc releases. + + +Version 1.9.1 +------------- +* Fix a crash when the PCI locality is invalid. Attach to the root object + instead. Thanks to Nicolas Denoyelle for reporting the issue. +* Fix -f in lstopo manpage. Thanks to Jirka Hladky for reporting the issue. +* Fix hwloc_obj_type_sscanf() and others when strncasecmp() is not properly + available. Thanks to Nick Papior Andersen for reporting the problem. +* Mark Linux file descriptors as close-on-exec to avoid leaks on exec. +* Fix some minor memory leaks. + + +Version 1.9.0 +------------- +* API + + Add hwloc_obj_type_sscanf() to extend hwloc_obj_type_of_string() with + type-specific attributes such as Cache/Group depth and Cache type. + hwloc_obj_type_of_string() is moved to hwloc/deprecated.h. + + Add hwloc_linux_get_tid_last_cpu_location() for retrieving the + last CPU where a Linux thread given by TID ran. + + Add hwloc_distrib() to extend the old hwloc_distribute[v]() functions. + hwloc_distribute[v]() is moved to hwloc/deprecated.h. + + Don't mix total and local memory when displaying verbose object attributes + with hwloc_obj_attr_snprintf() or in lstopo. +* Backends + + Add CPUVendor, CPUModelNumber and CPUFamilyNumber info attributes for + x86, ia64 and Xeon Phi sockets on Linux, to extend the x86-specific + support added in v1.8.1. Requested by Ralph Castain. + + Add many CPU- and Platform-related info attributes on ARM and POWER + platforms, in the Machine and Socket objects. + + Add CUDA info attributes describing the number of multiprocessors and + cores and the size of the global, shared and L2 cache memories in CUDA + OS devices. + + Add OpenCL info attributes describing the number of compute units and + the global memory size in OpenCL OS devices. + + The synthetic backend now accepts extended types such as L2Cache, L1i or + Group3. lstopo also exports synthetic strings using these extended types. +* Tools + + lstopo + - Do not overwrite output files by default anymore. + Pass -f or --force to enforce it. + - Display OpenCL, CUDA and Xeon Phi numbers of cores and memory sizes + in the graphical output. + - Fix export to stdout when specifying a Cairo-based output type + with --of. + + hwloc-ps + - Add -e or --get-last-cpu-location to report where processes/threads + run instead of where they are bound. + - Report locations as likely-more-useful objects such as Cores or Sockets + instead of Caches when possible. + + hwloc-bind + - Fix failure on Windows when not using --pid. + - Add -e as a synonym to --get-last-cpu-location. + + hwloc-distrib + - Add --reverse to distribute using last objects first and singlify + into last bits first. Thanks to Jirka Hladky for the suggestion. + + hwloc-info + - Report unified caches when looking for data or instruction cache + ancestor objects. +* Misc + + Add experimental Visual Studio support under contrib/windows. + Thanks to Eloi Gaudry for his help and for providing the first draft. + + Fix some overzealous assertions and warnings about the ordering of + objects on a level with respect to cpusets. The ordering is only + guaranteed for complete cpusets (based on the first bit in sets). + + Fix some memory leaks when importing xml diffs and when exporting a + "too complex" entry. + + +Version 1.8.1 +------------- +* Fix the cpuid code on Windows 64bits so that the x86 backend gets + enabled as expected and can populate CPU information. + Thanks to Robin Scher for reporting the problem. +* Add CPUVendor/CPUModelNumber/CPUFamilyNumber attributes when running + on x86 architecture. Thanks to Ralph Castain for the suggestion. +* Work around buggy BIOS reporting duplicate NUMA nodes on Linux. + Thanks to Jeff Becker for reporting the problem and testing the patch. +* Add a name to the lstopo graphical window. Thanks to Michael Prokop + for reporting the issue. + + +Version 1.8.0 +------------- +* New components + + Add the "linuxpci" component that always works on Linux even when + libpciaccess and libpci aren't available (and even with a modified + file-system root). By default the old "pci" component runs first + because "linuxpci" lacks device names (obj->name is always NULL). +* API + + Add the topology difference API in hwloc/diff.h for manipulating + many similar topologies. + + Add hwloc_topology_dup() for duplicating an entire topology. + + hwloc.h and hwloc/helper.h have been reorganized to clarify the + documentation sections. The actual inline code has moved out of hwloc.h + into the new hwloc/inlines.h. + + Deprecated functions are now in hwloc/deprecated.h, and not in the + official documentation anymore. +* Tools + + Add hwloc-diff and hwloc-patch tools together with the new diff API. + + Add hwloc-compress-dir to (de)compress an entire directory of XML files + using hwloc-diff and hwloc-patch. + + Object colors in the graphical output of lstopo may be changed by adding + a "lstopoStyle" info attribute. See CUSTOM COLORS in the lstopo(1) manpage + for details. Thanks to Jirka Hladky for discussing the idea. + + hwloc-gather-topology may now gather I/O-related files on Linux when + --io is given. Only the linuxpci component supports discovering I/O + objects from these extended tarballs. + + hwloc-annotate now supports --ri to remove/replace info attributes with + a given name. + + hwloc-info supports "root" and "all" special locations for dumping + information about the root object. + + lstopo now supports --append-legend to append custom lines of text + to the legend in the graphical output. Thanks to Jirka Hladky for + discussing the idea. + + hwloc-calc and friends have a more robust parsing of locations given + on the command-line and they report useful error messages about it. + + Add --whole-system to hwloc-bind, hwloc-calc, hwloc-distances and + hwloc-distrib, and add --restrict to hwloc-bind for uniformity among + tools. +* Misc + + Calling hwloc_topology_load() or hwloc_topology_set_*() on an already + loaded topology now returns an error (deprecated since release 1.6.1). + + Fix the initialisation of cpusets and nodesets in Group objects added + when inserting PCI hostbridges. + + Never merge Group objects that were added explicitly by the user with + hwloc_custom_insert_group_object_by_parent(). + + Add a sanity check during dynamic plugin loading to prevent some + crashes when hwloc is dynamically loaded by another plugin mechanisms. + + Add --with-hwloc-plugins-path to specify the install/load directories + of plugins. + + Add the MICSerialNumber info attribute to the root object when running + hwloc inside a Xeon Phi to match the same attribute in the MIC OS device + when running in the host. + + +Version 1.7.2 +------------- +* Do not create invalid block OS devices on very old Linux kernel such + as RHEL4 2.6.9. +* Fix PCI subvendor/device IDs. +* Fix the management of Misc objects inserted by parent. + Thanks to Jirka Hladky for reporting the problem. +* Add a PortState into attribute to OpenFabrics OS devices. +* Add a MICSerialNumber info attribute to Xeon PHI/MIC OS devices. +* Improve verbose error messages when failing to load from XML. + + +Version 1.7.1 +------------- +* Fix a failed assertion in the distance grouping code when loading a XML + file that already contains some groups. + Thanks to Laercio Lima Pilla for reporting the problem. +* Remove unexpected Group objects when loading XML topologies with I/O + objects and NUMA distances. + Thanks to Elena Elkina for reporting the problem and testing patches. +* Fix PCI link speed discovery when using libpciaccess. +* Fix invalid libpciaccess virtual function device/vendor IDs when using + SR-IOV PCI devices on Linux. +* Fix GL component build with old NVCtrl releases. + Thanks to Jirka Hladky for reporting the problem. +* Fix embedding breakage caused by libltdl. + Thanks to Pavan Balaji for reporting the problem. +* Always use the system-wide libltdl instead of shipping one inside hwloc. +* Document issues when enabling plugins while embedding hwloc in another + project, in the documentation section Embedding hwloc in Other Software. +* Add a FAQ entry "How to get useful topology information on NetBSD?" + in the documentation. +* Somes fixes in the renaming code for embedding. +* Miscellaneous minor build fixes. + + +Version 1.7.0 +------------- +* New operating system backends + + Add BlueGene/Q compute node kernel (CNK) support. See the FAQ in the + documentation for details. Thanks to Jeff Hammond, Christopher Samuel + and Erik Schnetter for their help. + + Add NetBSD support, thanks to Aleksej Saushev. +* New I/O device discovery + + Add co-processor OS devices such as "mic0" for Intel Xeon Phi (MIC) + on Linux. Thanks to Jerome Vienne for helping. + + Add co-processor OS devices such as "cuda0" for NVIDIA CUDA-capable GPUs. + + Add co-processor OS devices such as "opencl0d0" for OpenCL GPU devices + on the AMD OpenCL implementation. + + Add GPU OS devices such as ":0.0" for NVIDIA X11 displays. + + Add GPU OS devices such as "nvml0" for NVIDIA GPUs. + Thanks to Marwan Abdellah and Stefan Eilemann for helping. + These new OS devices have some string info attributes such as CoProcType, + GPUModel, etc. to better identify them. + See the I/O Devices and Attributes documentation sections for details. +* New components + + Add the "opencl", "cuda", "nvml" and "gl" components for I/O device + discovery. + + "nvml" also improves the discovery of NVIDIA GPU PCIe link speed. + All of these new components may be built as plugins. They may also be + disabled entirely by passing --disable-opencl/cuda/nvml/gl to configure. + See the I/O Devices, Components and Plugins, and FAQ documentation + sections for details. +* API + + Add hwloc_topology_get_flags(). + + Add hwloc/plugins.h for building external plugins. + See the Adding new discovery components and plugins section. +* Interoperability + + Add hwloc/opencl.h, hwloc/nvml.h, hwloc/gl.h and hwloc/intel-mic.h + to retrieve the locality of OS devices that correspond to AMD OpenCL + GPU devices or indexes, to NVML devices or indexes, to NVIDIA X11 + displays, or to Intel Xeon Phi (MIC) device indexes. + + Add new helpers in hwloc/cuda.h and hwloc/cudart.h to convert + between CUDA devices or indexes and hwloc OS devices. + + Add hwloc_ibv_get_device_osdev() and clarify the requirements + of the OpenFabrics Verbs helpers in hwloc/openfabrics-verbs.h. +* Tools + + hwloc-info is not only a synonym of lstopo -s anymore, it also + dumps information about objects given on the command-line. +* Documentation + + Add a section "Existing components and plugins". + + Add a list of common OS devices in section "Software devices". + + Add a new FAQ entry "Why is lstopo slow?" about lstopo slowness + issues because of GPUs. + + Clarify the documentation of inline helpers in hwloc/myriexpress.h + and hwloc/openfabrics-verbs.h. +* Misc + + Improve cache detection on AIX. + + The HWLOC_COMPONENTS variable now excludes the components whose + names are prefixed with '-'. + + lstopo --ignore PU now works when displaying the topology in + graphical and textual mode (not when exporting to XML). + + Make sure I/O options always appear in lstopo usage, not only when + using pciutils/libpci. + + Remove some unneeded Linux specific includes from some interoperability + headers. + + Fix some inconsistencies in hwloc-distrib and hwloc-assembler-remote + manpages. Thanks to Guy Streeter for the report. + + Fix a memory leak on AIX when getting memory binding. + + Fix many small memory leaks on Linux. + + The `libpci' component is now called `pci' but the old name is still + accepted in the HWLOC_COMPONENTS variable for backward compatibility. + + +Version 1.6.2 +------------- +* Use libpciaccess instead of pciutils/libpci by default for I/O discovery. + pciutils/libpci is only used if --enable-libpci is given to configure + because its GPL license may taint hwloc. See the Installation section + in the documentation for details. +* Fix get_cpubind on Solaris when bound to a single PU with + processor_bind(). Thanks to Eugene Loh for reporting the problem + and providing a patch. + + +Version 1.6.1 +------------- +* Fix some crash or buggy detection in the x86 backend when Linux + cgroups/cpusets restrict the available CPUs. +* Fix the pkg-config output with --libs --static. + Thanks to Erik Schnetter for reporting one of the problems. +* Fix the output of hwloc-calc -H --hierarchical when using logical + indexes in the output. +* Calling hwloc_topology_load() multiple times on the same topology + is officially deprecated. hwloc will warn in such cases. +* Add some documentation about existing plugins/components, package + dependencies, and I/O devices specification on the command-line. + + +Version 1.6.0 +------------- +* Major changes + + Reorganize the backend infrastructure to support dynamic selection + of components and dynamic loading of plugins. For details, see the + new documentation section Components and plugins. + - The HWLOC_COMPONENTS variable lets one replace the default discovery + components. + - Dynamic loading of plugins may be enabled with --enable-plugins + (except on AIX and Windows). It will build libxml2 and libpci + support as separated modules. This helps reducing the dependencies + of the core hwloc library when distributed as a binary package. +* Backends + + Add CPUModel detection on Darwin and x86/FreeBSD. + Thanks to Robin Scher for providing ways to implement this. + + The x86 backend now adds CPUModel info attributes to socket objects + created by other backends that do not natively support this attribute. + + Fix detection on FreeBSD in case of cpuset restriction. Thanks to + Sebastian Kuzminsky for reporting the problem. +* XML + + Add hwloc_topology_set_userdata_import/export_callback(), + hwloc_export_obj_userdata() and _userdata_base64() to let + applications specify how to save/restore the custom data they placed + in the userdata private pointer field of hwloc objects. +* Tools + + Add hwloc-annotate program to add string info attributes to XML + topologies. + + Add --pid-cmd to hwloc-ps to append the output of a command to each + PID line. May be used for showing Open MPI process ranks, see the + hwloc-ps(1) manpage for details. + + hwloc-bind now exits with an error if binding fails; the executable + is not launched unless binding suceeeded or --force was given. + + Add --quiet to hwloc-calc and hwloc-bind to hide non-fatal error + messages. + + Fix command-line pid support in windows tools. + + All programs accept --verbose as a synonym to -v. +* Misc + + Fix some DIR descriptor leaks on Linux. + + Fix I/O device lists when some were filtered out after a XML import. + + Fix the removal of I/O objects when importing a I/O-enabled XML topology + without any I/O topology flag. + + When merging objects with HWLOC_IGNORE_TYPE_KEEP_STRUCTURE or + lstopo --merge, compare object types before deciding which one of two + identical object to remove (e.g. keep sockets in favor of caches). + + Add some GUID- and LID-related info attributes to OpenFabrics + OS devices. + + Only add CPUType socket attributes on Solaris/Sparc. Other cases + don't report reliable information (Solaris/x86), and a replacement + is available as the Architecture string info in the Machine object. + + Add missing Backend string info on Solaris in most cases. + + Document object attributes and string infos in a new Attributes + section in the documentation. + + Add a section about Synthetic topologies in the documentation. + + +Version 1.5.2 (some of these changes are in v1.6.2 but not in v1.6) +------------- +* Use libpciaccess instead of pciutils/libpci by default for I/O discovery. + pciutils/libpci is only used if --enable-libpci is given to configure + because its GPL license may taint hwloc. See the Installation section + in the documentation for details. +* Fix get_cpubind on Solaris when bound to a single PU with + processor_bind(). Thanks to Eugene Loh for reporting the problem + and providing a patch. +* Fix some DIR descriptor leaks on Linux. +* Fix I/O device lists when some were filtered out after a XML import. +* Add missing Backend string info on Solaris in most cases. +* Fix the removal of I/O objects when importing a I/O-enabled XML topology + without any I/O topology flag. +* Fix the output of hwloc-calc -H --hierarchical when using logical + indexes in the output. +* Fix the pkg-config output with --libs --static. + Thanks to Erik Schnetter for reporting one of the problems. + + +Version 1.5.1 +------------- +* Fix block OS device detection on Linux kernel 3.3 and later. + Thanks to Guy Streeter for reporting the problem and testing the fix. +* Fix the cpuid code in the x86 backend (for FreeBSD). Thanks to + Sebastian Kuzminsky for reporting problems and testing patches. +* Fix 64bit detection on FreeBSD. +* Fix some corner cases in the management of the thissystem flag with + respect to topology flags and environment variables. +* Fix some corner cases in command-line parsing checks in hwloc-distrib + and hwloc-distances. +* Make sure we do not miss some block OS devices on old Linux kernels + when a single PCI device has multiple IDE hosts/devices behind it. +* Do not disable I/O devices or instruction caches in hwloc-assembler output. + + +Version 1.5.0 +------------- +* Backends + + Do not limit the number of processors to 1024 on Solaris anymore. + + Gather total machine memory on FreeBSD. Thanks to Cyril Roelandt. + + XML topology files do not depend on the locale anymore. Float numbers + such as NUMA distances or PCI link speeds now always use a dot as a + decimal separator. + + Add instruction caches detection on Linux, AIX, Windows and Darwin. + + Add get_last_cpu_location() support for the current thread on AIX. + + Support binding on AIX when threads or processes were bound with + bindprocessor(). Thanks to Hendryk Bockelmann for reporting the issue + and testing patches, and to Farid Parpia for explaining the binding + interfaces. + + Improve AMD topology detection in the x86 backend (for FreeBSD) using + the topoext feature. +* API + + Increase HWLOC_API_VERSION to 0x00010500 so that API changes may be + detected at build-time. + + Add a cache type attribute describind Data, Instruction and Unified + caches. Caches with different types but same depth (for instance L1d + and L1i) are placed on different levels. + + Add hwloc_get_cache_type_depth() to retrieve the hwloc level depth of + of the given cache depth and type, for instance L1i or L2. + It helps disambiguating the case where hwloc_get_type_depth() returns + HWLOC_TYPE_DEPTH_MULTIPLE. + + Instruction caches are ignored unless HWLOC_TOPOLOGY_FLAG_ICACHES is + passed to hwloc_topology_set_flags() before load. + + Add hwloc_ibv_get_device_osdev_by_name() OpenFabrics helper in + openfabrics-verbs.h to find the hwloc OS device object corresponding to + an OpenFabrics device. +* Tools + + Add lstopo-no-graphics, a lstopo built without graphical support to + avoid dependencies on external libraries such as Cairo and X11. When + supported, graphical outputs are only available in the original lstopo + program. + - Packagers splitting lstopo and lstopo-no-graphics into different + packages are advised to use the alternatives system so that lstopo + points to the best available binary. + + Instruction caches are enabled in lstopo by default. Use --no-icaches + to disable them. + + Add -t/--threads to show threads in hwloc-ps. +* Removal of obsolete components + + Remove the old cpuset interface (hwloc/cpuset.h) which is deprecated and + superseded by the bitmap API (hwloc/bitmap.h) since v1.1. + hwloc_cpuset and nodeset types are still defined, but all hwloc_cpuset_* + compatibility wrappers are now gone. + + Remove Linux libnuma conversion helpers for the deprecated and + broken nodemask_t interface. + + Remove support for "Proc" type name, it was superseded by "PU" in v1.0. + + Remove hwloc-mask symlinks, it was replaced by hwloc-calc in v1.0. +* Misc + + Fix PCIe 3.0 link speed computation. + + Non-printable characters are dropped from strings during XML export. + + Fix importing of escaped characters with the minimalistic XML backend. + + Assert hwloc_is_thissystem() in several I/O related helpers. + + Fix some memory leaks in the x86 backend for FreeBSD. + + Minor fixes to ease native builds on Windows. + + Limit the number of retries when operating on all threads within a + process on Linux if the list of threads is heavily getting modified. + + +Version 1.4.3 +------------- +* This release is only meant to fix the pciutils license issue when upgrading + to hwloc v1.5 or later is not possible. It contains several other minor + fixes but ignores many of them that are only in v1.5 or later. +* Use libpciaccess instead of pciutils/libpci by default for I/O discovery. + pciutils/libpci is only used if --enable-libpci is given to configure + because its GPL license may taint hwloc. See the Installation section + in the documentation for details. +* Fix PCIe 3.0 link speed computation. +* Fix importing of escaped characters with the minimalistic XML backend. +* Fix a memory leak in the x86 backend. + + +Version 1.4.2 +------------- +* Fix build on Solaris 9 and earlier when fabsf() is not a compiler + built-in. Thanks to Igor Galić for reporting the problem. +* Fix support for more than 32 processors on Windows. Thanks to Hartmut + Kaiser for reporting the problem. +* Fix process-wide binding and cpulocation routines on Linux when some + threads disappear in the meantime. Thanks to Vlad Roubtsov for reporting + the issue. +* Make installed scripts executable. Thanks to Jirka Hladky for reporting + the problem. +* Fix libtool revision management when building for Windows. This fix was + also released as hwloc v1.4.1.1 Windows builds. Thanks to Hartmut Kaiser + for reporting the problem. +* Fix the __hwloc_inline keyword in public headers when compiling with a + C++ compiler. +* Add Port info attribute to network OS devices inside OpenFabrics PCI + devices so as to identify which interface corresponds to which port. +* Document requirements for interoperability helpers: I/O devices discovery + is required for some of them; the topology must match the current host + for most of them. + + +Version 1.4.1 +------------- +* This release contains all changes from v1.3.2. +* Fix hwloc_alloc_membind, thanks Karl Napf for reporting the issue. +* Fix memory leaks in some get_membind() functions. +* Fix helpers converting from Linux libnuma to hwloc (hwloc/linux-libnuma.h) + in case of out-of-order NUMA node ids. +* Fix some overzealous assertions in the distance grouping code. +* Workaround BIOS reporting empty I/O locality in CUDA and OpenFabrics + helpers on Linux. Thanks to Albert Solernou for reporting the problem. +* Install a valgrind suppressions file hwloc-valgrind.supp (see the FAQ). +* Fix memory binding documentation. Thanks to Karl Napf for reporting the + issues. + + +Version 1.4.0 (does not contain all v1.3.2 changes) +------------- +* Major features + + Add "custom" interface and "assembler" tools to build multi-node + topology. See the Multi-node Topologies section in the documentation + for details. +* Interface improvements + + Add symmetric_subtree object attribute to ease assumptions when consulting + regular symmetric topologies. + + Add a CPUModel and CPUType info attribute to Socket objects on Linux + and Solaris. + + Add hwloc_get_obj_index_inside_cpuset() to retrieve the "logical" index + of an object within a subtree of the topology. + + Add more NVIDIA CUDA helpers in cuda.h and cudart.h to find hwloc objects + corresponding to CUDA devices. +* Discovery improvements + + Add a group object above partial distance matrices to make sure + the matrices are available in the final topology, except when this + new object would contradict the existing hierarchy. + + Grouping by distances now also works when loading from XML. + + Fix some corner cases in object insertion, for instance when dealing + with NUMA nodes without any CPU. +* Backends + + Implement hwloc_get_area_membind() on Linux. + + Honor I/O topology flags when importing from XML. + + Further improve XML-related error checking and reporting. + + Hide synthetic topology error messages unless HWLOC_SYNTHETIC_VERBOSE=1. +* Tools + + Add synthetic exporting of symmetric topologies to lstopo. + + lstopo --horiz and --vert can now be applied to some specific object types. + + lstopo -v -p now displays distance matrices with physical indexes. + + Add hwloc-distances utility to list distances. +* Documentation + + Fix and/or document the behavior of most inline functions in hwloc/helper.h + when the topology contains some I/O or Misc objects. + + Backend documentation enhancements. +* Bug fixes + + Fix missing last bit in hwloc_linux_get_thread_cpubind(). + Thanks to Carolina Gómez-Tostón Gutiérrez for reporting the issue. + + Fix FreeBSD build without cpuid support. + + Fix several Windows build issues. + + Fix inline keyword definition in public headers. + + Fix dependencies in the embedded library. + + Improve visibility support detection. Thanks to Dave Love for providing + the patch. + + Remove references to internal symbols in the tools. + + +Version 1.3.3 +------------- +* This release is only meant to fix the pciutils license issue when upgrading + to hwloc v1.4 or later is not possible. It contains several other minor + fixes but ignores many of them that are only in v1.4 or later. +* Use libpciaccess instead of pciutils/libpci by default for I/O discovery. + pciutils/libpci is only used if --enable-libpci is given to configure + because its GPL license may taint hwloc. See the Installation section + in the documentation for details. + + +Version 1.3.2 +------------- +* Fix missing last bit in hwloc_linux_get_thread_cpubind(). + Thanks to Carolina Gómez-Tostón Gutiérrez for reporting the issue. +* Fix build with -mcmodel=medium. Thanks to Devendar Bureddy for reporting + the issue. +* Fix build with Solaris Studio 12 compiler when XML is disabled. + Thanks to Paul H. Hargrove for reporting the problem. +* Fix installation with old GNU sed, for instance on Red Hat 8. + Thanks to Paul H. Hargrove for reporting the problem. +* Fix PCI locality when Linux cgroups restrict the available CPUs. +* Fix floating point issue when grouping by distance on mips64 architecture. + Thanks to Paul H. Hargrove for reporting the problem. +* Fix conversion from/to Linux libnuma when some NUMA nodes have no memory. +* Fix support for gccfss compilers with broken ffs() support. Thanks to + Paul H. Hargrove for reporting the problem and providing a patch. +* Fix FreeBSD build without cpuid support. +* Fix several Windows build issues. +* Fix inline keyword definition in public headers. +* Fix dependencies in the embedded library. +* Detect when a compiler such as xlc may not report compile errors + properly, causing some configure checks to be wrong. Thanks to + Paul H. Hargrove for reporting the problem and providing a patch. +* Improve visibility support detection. Thanks to Dave Love for providing + the patch. +* Remove references to internal symbols in the tools. +* Fix installation on systems with limited command-line size. + Thanks to Paul H. Hargrove for reporting the problem. +* Further improve XML-related error checking and reporting. + + +Version 1.3.1 +------------- +* Fix pciutils detection with pkg-config when not installed in standard + directories. +* Fix visibility options detection with the Solaris Studio compiler. + Thanks to Igor Galić and Terry Dontje for reporting the problems. +* Fix support for old Linux sched.h headers such as those found + on Red Hat 8. Thanks to Paul H. Hargrove for reporting the problems. +* Fix inline and attribute support for Solaris compilers. Thanks to + Dave Love for reporting the problems. +* Print a short summary at the end of the configure output. Thanks to + Stefan Eilemann for the suggestion. +* Add --disable-libnuma configure option to disable libnuma-based + memory binding support on Linux. Thanks to Rayson Ho for the + suggestion. +* Make hwloc's configure script properly obey $PKG_CONFIG. Thanks to + Nathan Phillip Brink for raising the issue. +* Silence some harmless pciutils warnings, thanks to Paul H. Hargrove + for reporting the problem. +* Fix the documentation with respect to hwloc_pid_t and hwloc_thread_t + being either pid_t and pthread_t on Unix, or HANDLE on Windows. + + +Version 1.3.0 +------------- +* Major features + + Add I/O devices and bridges to the topology using the pciutils + library. Only enabled after setting the relevant flag with + hwloc_topology_set_flags() before hwloc_topology_load(). See the + I/O Devices section in the documentation for details. +* Discovery improvements + + Add associativity to the cache attributes. + + Add support for s390/z11 "books" on Linux. + + Add the HWLOC_GROUPING_ACCURACY environment variable to relax + distance-based grouping constraints. See the Environment Variables + section in the documentation for details about grouping behavior + and configuration. + + Allow user-given distance matrices to remove or replace those + discovered by the OS backend. +* XML improvements + + XML is now always supported: a minimalistic custom import/export + code is used when libxml2 is not available. It is only guaranteed + to read XML files generated by hwloc. + + hwloc_topology_export_xml() and export_xmlbuffer() now return an + integer. + + Add hwloc_free_xmlbuffer() to free the buffer allocated by + hwloc_topology_export_xmlbuffer(). + + Hide XML topology error messages unless HWLOC_XML_VERBOSE=1. +* Minor API updates + + Add hwloc_obj_add_info to customize object info attributes. +* Tools + + lstopo now displays I/O devices by default. Several options are + added to configure the I/O discovery. + + hwloc-calc and hwloc-bind now accept I/O devices as input. + + Add --restrict option to hwloc-calc and hwloc-distribute. + + Add --sep option to change the output field separator in hwloc-calc. + + Add --whole-system option to hwloc-ps. + + +Version 1.2.2 +------------- +* Fix build on AIX 5.2, thanks Utpal Kumar Ray for the report. +* Fix XML import of very large page sizes or counts on 32bits platform, + thanks to Karsten Hopp for the RedHat ticket. +* Fix crash when administrator limitations such as Linux cgroup require + to restrict distance matrices. Thanks to Ake Sandgren for reporting the + problem. +* Fix the removal of objects such as AMD Magny-Cours dual-node sockets + in case of administrator restrictions. +* Improve error reporting and messages in case of wrong synthetic topology + description. +* Several other minor internal fixes and documentation improvements. + + +Version 1.2.1 +------------- +* Improve support of AMD Bulldozer "Compute-Unit" modules by detecting + logical processors with different core IDs on Linux. +* Fix hwloc-ps crash when listing processes from another Linux cpuset. + Thanks to Carl Smith for reporting the problem. +* Fix build on AIX and Solaris. Thanks to Carl Smith and Andreas Kupries + for reporting the problems. +* Fix cache size detection on Darwin. Thanks to Erkcan Özcan for reporting + the problem. +* Make configure fail if --enable-xml or --enable-cairo is given and + proper support cannot be found. Thanks to Andreas Kupries for reporting + the XML problem. +* Fix spurious L1 cache detection on AIX. Thanks to Hendryk Bockelmann + for reporting the problem. +* Fix hwloc_get_last_cpu_location(THREAD) on Linux. Thanks to Gabriele + Fatigati for reporting the problem. +* Fix object distance detection on Solaris. +* Add pthread_self weak symbol to ease static linking. +* Minor documentation fixes. + + +Version 1.2.0 +------------- +* Major features + + Expose latency matrices in the API as an array of distance structures + within objects. Add several helpers to find distances. + + Add hwloc_topology_set_distance_matrix() and environment variables + to provide a matrix of distances between a given set of objects. + + Add hwloc_get_last_cpu_location() and hwloc_get_proc_last_cpu_location() + to retrieve the processors where a process or thread recently ran. + - Add the corresponding --get-last-cpu-location option to hwloc-bind. + + Add hwloc_topology_restrict() to restrict an existing topology to a + given cpuset. + - Add the corresponding --restrict option to lstopo. +* Minor API updates + + Add hwloc_bitmap_list_sscanf/snprintf/asprintf to convert between bitmaps + and strings such as 4-5,7-9,12,15- + + hwloc_bitmap_set/clr_range() now support infinite ranges. + + Clarify the difference between inserting Misc objects by cpuset or by + parent. + + hwloc_insert_misc_object_by_cpuset() now returns NULL in case of error. +* Discovery improvements + + x86 backend (for freebsd): add x2APIC support + + Support standard device-tree phandle, to get better support on e.g. ARM + systems providing it. + + Detect cache size on AIX. Thanks Christopher and IBM. + + Improve grouping to support asymmetric topologies. +* Tools + + Command-line tools now support "all" and "root" special locations + consisting in the entire topology, as well as type names with depth + attributes such as L2 or Group4. + + hwloc-calc improvements: + - Add --number-of/-N option to report the number of objects of a given + type or depth. + - -I is now equivalent to --intersect for listing the indexes of + objects of a given type or depth that intersects the input. + - Add -H to report the output as a hierarchical combination of types + and depths. + + Add --thissystem to lstopo. + + Add lstopo-win, a console-less lstopo variant on Windows. +* Miscellaneous + + Remove C99 usage from code base. + + Rename hwloc-gather-topology.sh into hwloc-gather-topology + + Fix AMD cache discovery on freebsd when there is no L3 cache, thanks + Andriy Gapon for the fix. + + +Version 1.1.2 +------------- +* Fix a segfault in the distance-based grouping code when some objects + are not placed in any group. Thanks to Bernd Kallies for reporting + the problem and providing a patch. +* Fix the command-line parsing of hwloc-bind --mempolicy interleave. + Thanks to Guy Streeter for reporting the problem. +* Stop truncating the output in hwloc_obj_attr_snprintf() and in the + corresponding lstopo output. Thanks to Guy Streeter for reporting the + problem. +* Fix object levels ordering in synthetic topologies. +* Fix potential incoherency between device tree and kernel information, + when SMT is disabled on Power machines. +* Fix and document the behavior of hwloc_topology_set_synthetic() in case + of invalid argument. Thanks to Guy Streeter for reporting the problem. +* Add some verbose error message reporting when it looks like the OS + gives erroneous information. +* Do not include unistd.h and stdint.h in public headers on Windows. +* Move config.h files into their own subdirectories to avoid name + conflicts when AC_CONFIG_HEADERS adds -I's for them. +* Remove the use of declaring variables inside "for" loops. +* Some other minor fixes. +* Many minor documentation fixes. + + +Version 1.1.1 +------------- +* Add hwloc_get_api_version() which returns the version of hwloc used + at runtime. Thanks to Guy Streeter for the suggestion. +* Fix the number of hugepages reported for NUMA nodes on Linux. +* Fix hwloc_bitmap_to_ulong() right after allocating the bitmap. + Thanks to Bernd Kallies for reporting the problem. +* Fix hwloc_bitmap_from_ith_ulong() to properly zero the first ulong. + Thanks to Guy Streeter for reporting the problem. +* Fix hwloc_get_membind_nodeset() on Linux. + Thanks to Bernd Kallies for reporting the problem and providing a patch. +* Fix some file descriptor leaks in the Linux discovery. +* Fix the minimum width of NUMA nodes, caches and the legend in the graphical + lstopo output. Thanks to Jirka Hladky for reporting the problem. +* Various fixes to bitmap conversion from/to taskset-strings. +* Fix and document snprintf functions behavior when the buffer size is too + small or zero. Thanks to Guy Streeter for reporting the problem. +* Fix configure to avoid spurious enabling of the cpuid backend. + Thanks to Tim Anderson for reporting the problem. +* Cleanup error management in hwloc-gather-topology.sh. + Thanks to Jirka Hladky for reporting the problem and providing a patch. +* Add a manpage and usage for hwloc-gather-topology.sh on Linux. + Thanks to Jirka Hladky for providing a patch. +* Memory binding documentation enhancements. + + +Version 1.1.0 +------------- + +* API + + Increase HWLOC_API_VERSION to 0x00010100 so that API changes may be + detected at build-time. + + Add a memory binding interface. + + The cpuset API (hwloc/cpuset.h) is now deprecated. It is replaced by + the bitmap API (hwloc/bitmap.h) which offers the same features with more + generic names since it applies to CPU sets, node sets and more. + Backward compatibility with the cpuset API and ABI is still provided but + it will be removed in a future release. + Old types (hwloc_cpuset_t, ...) are still available as a way to clarify + what kind of hwloc_bitmap_t each API function manipulates. + Upgrading to the new API only requires to replace hwloc_cpuset_ function + calls with the corresponding hwloc_bitmap_ calls, with the following + renaming exceptions: + - hwloc_cpuset_cpu -> hwloc_bitmap_only + - hwloc_cpuset_all_but_cpu -> hwloc_bitmap_allbut + - hwloc_cpuset_from_string -> hwloc_bitmap_sscanf + + Add an `infos' array in each object to store couples of info names and + values. It enables generic storage of things like the old dmi board infos + that were previously stored in machine specific attributes. + + Add linesize cache attribute. +* Features + + Bitmaps (and thus CPU sets and node sets) are dynamically (re-)allocated, + the maximal number of CPUs (HWLOC_NBMAXCPUS) has been removed. + + Improve the distance-based grouping code to better support irregular + distance matrices. + + Add support for device-tree to get cache information (useful on Power + architectures). +* Helpers + + Add NVIDIA CUDA helpers in cuda.h and cudart.h to ease interoperability + with CUDA Runtime and Driver APIs. + + Add Myrinet Express helper in myriexpress.h to ease interoperability. +* Tools + + lstopo now displays physical/OS indexes by default in graphical mode + (use -l to switch back to logical indexes). The textual output still uses + logical by default (use -p to switch to physical indexes). + + lstopo prefixes logical indexes with `L#' and physical indexes with `P#'. + Physical indexes are also printed as `P#N' instead of `phys=N' within + object attributes (in parentheses). + + Add a legend at the bottom of the lstopo graphical output, use --no-legend + to remove it. + + Add hwloc-ps to list process' bindings. + + Add --membind and --mempolicy options to hwloc-bind. + + Improve tools command-line options by adding a generic --input option + (and more) which replaces the old --xml, --synthetic and --fsys-root. + + Cleanup lstopo output configuration by adding --output-format. + + Add --intersect in hwloc-calc, and replace --objects with --largest. + + Add the ability to work on standard input in hwloc-calc. + + Add --from, --to and --at in hwloc-distrib. + + Add taskset-specific functions and command-line tools options to + manipulate CPU set strings in the format of the taskset program. + + Install hwloc-gather-topology.sh on Linux. + + +Version 1.0.3 +------------- + +* Fix support for Linux cpuset when emulated by a cgroup mount point. +* Remove unneeded runtime dependency on libibverbs.so in the library and + all utils programs. +* Fix hwloc_cpuset_to_linux_libnuma_ulongs in case of non-linear OS-indexes + for NUMA nodes. +* lstopo now displays physical/OS indexes by default in graphical mode + (use -l to switch back to logical indexes). The textual output still uses + logical by default (use -p to switch to physical indexes). + + +Version 1.0.2 +------------- + +* Public headers can now be included directly from C++ programs. +* Solaris fix for non-contiguous cpu numbers. Thanks to Rolf vandeVaart for + reporting the issue. +* Darwin 10.4 fix. Thanks to Olivier Cessenat for reporting the issue. +* Revert 1.0.1 patch that ignored sockets with unknown ID values since it + only slightly helped POWER7 machines with old Linux kernels while it + prevents recent kernels from getting the complete POWER7 topology. +* Fix hwloc_get_common_ancestor_obj(). +* Remove arch-specific bits in public headers. +* Some fixes in the lstopo graphical output. +* Various man page clarifications and minor updates. + + +Version 1.0.1 +------------- + +* Various Solaris fixes. Thanks to Yannick Martin for reporting the issue. +* Fix "non-native" builds on x86 platforms (e.g., when building 32 + bit executables with compilers that natively build 64 bit). +* Ignore sockets with unknown ID values (which fixes issues on POWER7 + machines). Thanks to Greg Bauer for reporting the issue. +* Various man page clarifications and minor updates. +* Fixed memory leaks in hwloc_setup_group_from_min_distance_clique(). +* Fix cache type filtering on MS Windows 7. Thanks to Αλέξανδρος + Παπαδογιαννάκ for reporting the issue. +* Fixed warnings when compiling with -DNDEBUG. + + +Version 1.0.0 +------------- + +* The ABI of the library has changed. +* Backend updates + + Add FreeBSD support. + + Add x86 cpuid based backend. + + Add Linux cgroup support to the Linux cpuset code. + + Support binding of entire multithreaded process on Linux. + + Fix and enable Group support in Windows. + + Cleanup XML export/import. +* Objects + + HWLOC_OBJ_PROC is renamed into HWLOC_OBJ_PU for "Processing Unit", + its stringified type name is now "PU". + + Use new HWLOC_OBJ_GROUP objects instead of MISC when grouping + objects according to NUMA distances or arbitrary OS aggregation. + + Rework memory attributes. + + Add different cpusets in each object to specify processors that + are offline, unavailable, ... + + Cleanup the storage of object names and DMI infos. +* Features + + Add support for looking up specific PID topology information. + + Add hwloc_topology_export_xml() to export the topology in a XML file. + + Add hwloc_topology_get_support() to retrieve the supported features + for the current topology context. + + Support non-SYSTEM object as the root of the tree, use MACHINE in + most common cases. + + Add hwloc_get_*cpubind() routines to retrieve the current binding + of processes and threads. +* API + + Add HWLOC_API_VERSION to help detect the currently used API version. + + Add missing ending "e" to *compare* functions. + + Add several routines to emulate PLPA functions. + + Rename and rework the cpuset and/or/xor/not/clear operators to output + their result in a dedicated argument instead of modifying one input. + + Deprecate hwloc_obj_snprintf() in favor of hwloc_obj_type/attr_snprintf(). + + Clarify the use of parent and ancestor in the API, do not use father. + + Replace hwloc_get_system_obj() with hwloc_get_root_obj(). + + Return -1 instead of HWLOC_OBJ_TYPE_MAX in the API since the latter + isn't public. + + Relax constraints in hwloc_obj_type_of_string(). + + Improve displaying of memory sizes. + + Add 0x prefix to cpuset strings. +* Tools + + lstopo now displays logical indexes by default, use --physical to + revert back to OS/physical indexes. + + Add colors in the lstopo graphical outputs to distinguish between online, + offline, reserved, ... objects. + + Extend lstopo to show cpusets, filter objects by type, ... + + Renamed hwloc-mask into hwloc-calc which supports many new options. +* Documentation + + Add a hwloc(7) manpage containing general information. + + Add documentation about how to switch from PLPA to hwloc. + + Cleanup the distributed documentation files. +* Miscellaneous + + Many compilers warning fixes. + + Cleanup the ABI by using the visibility attribute. + + Add project embedding support. + + +Version 0.9.4 (unreleased) +-------------------------- + +* Fix reseting colors to normal in lstopo -.txt output. +* Fix Linux pthread_t binding error report. + + +Version 0.9.3 +------------- + +* Fix autogen.sh to work with Autoconf 2.63. +* Fix various crashes in particular conditions: + - xml files with root attributes + - offline CPUs + - partial sysfs support + - unparseable /proc/cpuinfo + - ignoring NUMA level while Misc level have been generated +* Tweak documentation a bit +* Do not require the pthread library for binding the current thread on Linux +* Do not erroneously consider the sched_setaffinity prototype is the old version + when there is actually none. +* Fix _syscall3 compilation on archs for which we do not have the + sched_setaffinity system call number. +* Fix AIX binding. +* Fix libraries dependencies: now only lstopo depends on libtermcap, fix + binutils-gold link +* Have make check always build and run hwloc-hello.c +* Do not limit size of a cpuset. + + +Version 0.9.2 +------------- + +* Trivial documentation changes. + + +Version 0.9.1 +------------- + +* Re-branded to "hwloc" and moved to the Open MPI project, relicensed under the + BSD license. +* The prefix of all functions and tools is now hwloc, and some public + functions were also renamed for real. +* Group NUMA nodes into Misc objects according to their physical distance + that may be reported by the OS/BIOS. + May be ignored by setting HWLOC_IGNORE_DISTANCES=1 in the environment. +* Ignore offline CPUs on Solaris. +* Improved binding support on AIX. +* Add HP-UX support. +* CPU sets are now allocated/freed dynamically. +* Add command line options to tune the lstopo graphical output, add + semi-graphical textual output +* Extend topobind to support multiple cpusets or objects on the command + line as topomask does. +* Add an Infiniband-specific helper hwloc/openfabrics-verbs.h to retrieve + the physical location of IB devices. + + +Version 0.9 (libtopology) +------------------------- + +* First release. diff --git a/src/3rdparty/hwloc/README b/src/3rdparty/hwloc/README new file mode 100644 index 000000000..5567b4d14 --- /dev/null +++ b/src/3rdparty/hwloc/README @@ -0,0 +1,85 @@ +Introduction + +The Hardware Locality (hwloc) software project aims at easing the process of +discovering hardware resources in parallel architectures. It offers +command-line tools and a C API for consulting these resources, their locality, +attributes, and interconnection. hwloc primarily aims at helping +high-performance computing (HPC) applications, but is also applicable to any +project seeking to exploit code and/or data locality on modern computing +platforms. + +hwloc is actually made of two subprojects distributed together: + + * The original hwloc project for describing the internals of computing nodes. + It is described in details starting at section Hardware Locality (hwloc) + Introduction. + * The network-oriented companion called netloc (Network Locality), described + in details starting with section Network Locality (netloc). + +See also the Related pages tab above for links to other sections. + +Netloc may be disabled, but the original hwloc cannot. Both hwloc and netloc +APIs are documented after these sections. + +Installation + +hwloc (http://www.open-mpi.org/projects/hwloc/) is available under the BSD +license. It is hosted as a sub-project of the overall Open MPI project (http:// +www.open-mpi.org/). Note that hwloc does not require any functionality from +Open MPI -- it is a wholly separate (and much smaller!) project and code base. +It just happens to be hosted as part of the overall Open MPI project. + +Basic Installation + +Installation is the fairly common GNU-based process: + +shell$ ./configure --prefix=... +shell$ make +shell$ make install + +hwloc- and netloc-specific configure options and requirements are documented in +sections hwloc Installation and Netloc Installation respectively. + +Also note that if you install supplemental libraries in non-standard locations, +hwloc's configure script may not be able to find them without some help. You +may need to specify additional CPPFLAGS, LDFLAGS, or PKG_CONFIG_PATH values on +the configure command line. + +For example, if libpciaccess was installed into /opt/pciaccess, hwloc's +configure script may not find it be default. Try adding PKG_CONFIG_PATH to the +./configure command line, like this: + +./configure PKG_CONFIG_PATH=/opt/pciaccess/lib/pkgconfig ... + +Running the "lstopo" tool is a good way to check as a graphical output whether +hwloc properly detected the architecture of your node. Netloc command-line +tools can be used to display the network topology interconnecting your nodes. + +Installing from a Git clone + +Additionally, the code can be directly cloned from Git: + +shell$ git clone https://github.com/open-mpi/hwloc.git +shell$ cd hwloc +shell$ ./autogen.sh + +Note that GNU Autoconf >=2.63, Automake >=1.11 and Libtool >=2.2.6 are required +when building from a Git clone. + +Nightly development snapshots are available on the web site, they can be +configured and built without any need for Git or GNU Autotools. + +Questions and Bugs + +Bugs should be reported in the tracker (https://github.com/open-mpi/hwloc/ +issues). Opening a new issue automatically displays lots of hints about how to +debug and report issues. + +Questions may be sent to the users or developers mailing lists (http:// +www.open-mpi.org/community/lists/hwloc.php). + +There is also a #hwloc IRC channel on Freenode (irc.freenode.net). + + + +See https://www.open-mpi.org/projects/hwloc/doc/ for more hwloc documentation. diff --git a/src/3rdparty/hwloc/VERSION b/src/3rdparty/hwloc/VERSION new file mode 100644 index 000000000..5ebc6bb47 --- /dev/null +++ b/src/3rdparty/hwloc/VERSION @@ -0,0 +1,47 @@ +# This is the VERSION file for hwloc, describing the precise version +# of hwloc in this distribution. The various components of the version +# number below are combined to form a single version number string. + +# major, minor, and release are generally combined in the form +# ... If release is zero, then it is omitted. + +# Please update HWLOC_VERSION* in contrib/windows/hwloc_config.h too. + +major=2 +minor=0 +release=4 + +# greek is used for alpha or beta release tags. If it is non-empty, +# it will be appended to the version number. It does not have to be +# numeric. Common examples include a1 (alpha release 1), b1 (beta +# release 1), sc2005 (Super Computing 2005 release). The only +# requirement is that it must be entirely printable ASCII characters +# and have no white space. + +greek= + +# The date when this release was created + +date="Jun 03, 2019" + +# If snapshot=1, then use the value from snapshot_version as the +# entire hwloc version (i.e., ignore major, minor, release, and +# greek). This is only set to 1 when making snapshot tarballs. +snapshot=0 +snapshot_version=${major}.${minor}.${release}${greek}-git + +# The shared library version of hwloc's public library. This version +# is maintained in accordance with the "Library Interface Versions" +# chapter from the GNU Libtool documentation. Notes: + +# 1. Since version numbers are associated with *releases*, the version +# number maintained on the hwloc git master (and developer branches) +# is always 0:0:0. + +# 2. Version numbers are described in the Libtool current:revision:age +# format. + +libhwloc_so_version=15:3:0 +libnetloc_so_version=0:0:0 + +# Please also update the lines in contrib/windows/libhwloc.vcxproj diff --git a/src/3rdparty/hwloc/include/hwloc.h b/src/3rdparty/hwloc/include/hwloc.h new file mode 100644 index 000000000..ee6da6fd1 --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc.h @@ -0,0 +1,2270 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2019 Inria. All rights reserved. + * Copyright © 2009-2012 Université Bordeaux + * Copyright © 2009-2011 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +/*===================================================================== + * PLEASE GO READ THE DOCUMENTATION! + * ------------------------------------------------ + * $tarball_directory/doc/doxygen-doc/ + * or + * http://www.open-mpi.org/projects/hwloc/doc/ + *===================================================================== + * + * FAIR WARNING: Do NOT expect to be able to figure out all the + * subtleties of hwloc by simply reading function prototypes and + * constant descrptions here in this file. + * + * Hwloc has wonderful documentation in both PDF and HTML formats for + * your reading pleasure. The formal documentation explains a LOT of + * hwloc-specific concepts, provides definitions, and discusses the + * "big picture" for many of the things that you'll find here in this + * header file. + * + * The PDF/HTML documentation was generated via Doxygen; much of what + * you'll see in there is also here in this file. BUT THERE IS A LOT + * THAT IS IN THE PDF/HTML THAT IS ***NOT*** IN hwloc.h! + * + * There are entire paragraph-length descriptions, discussions, and + * pretty prictures to explain subtle corner cases, provide concrete + * examples, etc. + * + * Please, go read the documentation. :-) + * + * Moreover there are several examples of hwloc use under doc/examples + * in the source tree. + * + *=====================================================================*/ + +/** \file + * \brief The hwloc API. + * + * See hwloc/bitmap.h for bitmap specific macros. + * See hwloc/helper.h for high-level topology traversal helpers. + * See hwloc/inlines.h for the actual inline code of some functions below. + * See hwloc/export.h for exporting topologies to XML or to synthetic descriptions. + * See hwloc/distances.h for querying and modifying distances between objects. + * See hwloc/diff.h for manipulating differences between similar topologies. + */ + +#ifndef HWLOC_H +#define HWLOC_H + +#include +#include +#include +#include +#include + +/* + * Symbol transforms + */ +#include + +/* + * Bitmap definitions + */ + +#include + + +#ifdef __cplusplus +extern "C" { +#endif + + +/** \defgroup hwlocality_api_version API version + * @{ + */ + +/** \brief Indicate at build time which hwloc API version is being used. + * + * This number is updated to (X<<16)+(Y<<8)+Z when a new release X.Y.Z + * actually modifies the API. + * + * Users may check for available features at build time using this number + * (see \ref faq_upgrade). + * + * \note This should not be confused with HWLOC_VERSION, the library version. + * Two stable releases of the same series usually have the same ::HWLOC_API_VERSION + * even if their HWLOC_VERSION are different. + */ +#define HWLOC_API_VERSION 0x00020000 + +/** \brief Indicate at runtime which hwloc API version was used at build time. + * + * Should be ::HWLOC_API_VERSION if running on the same version. + */ +HWLOC_DECLSPEC unsigned hwloc_get_api_version(void); + +/** \brief Current component and plugin ABI version (see hwloc/plugins.h) */ +#define HWLOC_COMPONENT_ABI 5 + +/** @} */ + + + +/** \defgroup hwlocality_object_sets Object Sets (hwloc_cpuset_t and hwloc_nodeset_t) + * + * Hwloc uses bitmaps to represent two distinct kinds of object sets: + * CPU sets (::hwloc_cpuset_t) and NUMA node sets (::hwloc_nodeset_t). + * These types are both typedefs to a common back end type + * (::hwloc_bitmap_t), and therefore all the hwloc bitmap functions + * are applicable to both ::hwloc_cpuset_t and ::hwloc_nodeset_t (see + * \ref hwlocality_bitmap). + * + * The rationale for having two different types is that even though + * the actions one wants to perform on these types are the same (e.g., + * enable and disable individual items in the set/mask), they're used + * in very different contexts: one for specifying which processors to + * use and one for specifying which NUMA nodes to use. Hence, the + * name difference is really just to reflect the intent of where the + * type is used. + * + * @{ + */ + +/** \brief A CPU set is a bitmap whose bits are set according to CPU + * physical OS indexes. + * + * It may be consulted and modified with the bitmap API as any + * ::hwloc_bitmap_t (see hwloc/bitmap.h). + * + * Each bit may be converted into a PU object using + * hwloc_get_pu_obj_by_os_index(). + */ +typedef hwloc_bitmap_t hwloc_cpuset_t; +/** \brief A non-modifiable ::hwloc_cpuset_t. */ +typedef hwloc_const_bitmap_t hwloc_const_cpuset_t; + +/** \brief A node set is a bitmap whose bits are set according to NUMA + * memory node physical OS indexes. + * + * It may be consulted and modified with the bitmap API as any + * ::hwloc_bitmap_t (see hwloc/bitmap.h). + * Each bit may be converted into a NUMA node object using + * hwloc_get_numanode_obj_by_os_index(). + * + * When binding memory on a system without any NUMA node, + * the single main memory bank is considered as NUMA node #0. + * + * See also \ref hwlocality_helper_nodeset_convert. + */ +typedef hwloc_bitmap_t hwloc_nodeset_t; +/** \brief A non-modifiable ::hwloc_nodeset_t. + */ +typedef hwloc_const_bitmap_t hwloc_const_nodeset_t; + +/** @} */ + + + +/** \defgroup hwlocality_object_types Object Types + * @{ + */ + +/** \brief Type of topology object. + * + * \note Do not rely on the ordering or completeness of the values as new ones + * may be defined in the future! If you need to compare types, use + * hwloc_compare_types() instead. + */ +#define HWLOC_OBJ_TYPE_MIN HWLOC_OBJ_MACHINE /**< \private Sentinel value */ +typedef enum { + HWLOC_OBJ_MACHINE, /**< \brief Machine. + * A set of processors and memory with cache + * coherency. + * + * This type is always used for the root object of a topology, + * and never used anywhere else. + * Hence its parent is always \c NULL. + */ + + HWLOC_OBJ_PACKAGE, /**< \brief Physical package. + * The physical package that usually gets inserted + * into a socket on the motherboard. + * A processor package usually contains multiple cores. + */ + HWLOC_OBJ_CORE, /**< \brief Core. + * A computation unit (may be shared by several + * logical processors). + */ + HWLOC_OBJ_PU, /**< \brief Processing Unit, or (Logical) Processor. + * An execution unit (may share a core with some + * other logical processors, e.g. in the case of + * an SMT core). + * + * This is the smallest object representing CPU resources, + * it cannot have any child except Misc objects. + * + * Objects of this kind are always reported and can + * thus be used as fallback when others are not. + */ + + HWLOC_OBJ_L1CACHE, /**< \brief Level 1 Data (or Unified) Cache. */ + HWLOC_OBJ_L2CACHE, /**< \brief Level 2 Data (or Unified) Cache. */ + HWLOC_OBJ_L3CACHE, /**< \brief Level 3 Data (or Unified) Cache. */ + HWLOC_OBJ_L4CACHE, /**< \brief Level 4 Data (or Unified) Cache. */ + HWLOC_OBJ_L5CACHE, /**< \brief Level 5 Data (or Unified) Cache. */ + + HWLOC_OBJ_L1ICACHE, /**< \brief Level 1 instruction Cache (filtered out by default). */ + HWLOC_OBJ_L2ICACHE, /**< \brief Level 2 instruction Cache (filtered out by default). */ + HWLOC_OBJ_L3ICACHE, /**< \brief Level 3 instruction Cache (filtered out by default). */ + + HWLOC_OBJ_GROUP, /**< \brief Group objects. + * Objects which do not fit in the above but are + * detected by hwloc and are useful to take into + * account for affinity. For instance, some operating systems + * expose their arbitrary processors aggregation this + * way. And hwloc may insert such objects to group + * NUMA nodes according to their distances. + * See also \ref faq_groups. + * + * These objects are removed when they do not bring + * any structure (see ::HWLOC_TYPE_FILTER_KEEP_STRUCTURE). + */ + + HWLOC_OBJ_NUMANODE, /**< \brief NUMA node. + * An object that contains memory that is directly + * and byte-accessible to the host processors. + * It is usually close to some cores (the corresponding objects + * are descendants of the NUMA node object in the hwloc tree). + * + * There is always at least one such object in the topology + * even if the machine is not NUMA. + * + * Memory objects are not listed in the main children list, + * but rather in the dedicated Memory children list. + * + * NUMA nodes have a special depth ::HWLOC_TYPE_DEPTH_NUMANODE + * instead of a normal depth just like other objects in the + * main tree. + */ + + HWLOC_OBJ_BRIDGE, /**< \brief Bridge (filtered out by default). + * Any bridge that connects the host or an I/O bus, + * to another I/O bus. + * They are not added to the topology unless I/O discovery + * is enabled with hwloc_topology_set_flags(). + * I/O objects are not listed in the main children list, + * but rather in the dedicated io children list. + * I/O objects have NULL CPU and node sets. + */ + HWLOC_OBJ_PCI_DEVICE, /**< \brief PCI device (filtered out by default). + * They are not added to the topology unless I/O discovery + * is enabled with hwloc_topology_set_flags(). + * I/O objects are not listed in the main children list, + * but rather in the dedicated io children list. + * I/O objects have NULL CPU and node sets. + */ + HWLOC_OBJ_OS_DEVICE, /**< \brief Operating system device (filtered out by default). + * They are not added to the topology unless I/O discovery + * is enabled with hwloc_topology_set_flags(). + * I/O objects are not listed in the main children list, + * but rather in the dedicated io children list. + * I/O objects have NULL CPU and node sets. + */ + + HWLOC_OBJ_MISC, /**< \brief Miscellaneous objects (filtered out by default). + * Objects without particular meaning, that can e.g. be + * added by the application for its own use, or by hwloc + * for miscellaneous objects such as MemoryModule (DIMMs). + * These objects are not listed in the main children list, + * but rather in the dedicated misc children list. + * Misc objects may only have Misc objects as children, + * and those are in the dedicated misc children list as well. + * Misc objects have NULL CPU and node sets. + */ + + HWLOC_OBJ_TYPE_MAX /**< \private Sentinel value */ +} hwloc_obj_type_t; + +/** \brief Cache type. */ +typedef enum hwloc_obj_cache_type_e { + HWLOC_OBJ_CACHE_UNIFIED, /**< \brief Unified cache. */ + HWLOC_OBJ_CACHE_DATA, /**< \brief Data cache. */ + HWLOC_OBJ_CACHE_INSTRUCTION /**< \brief Instruction cache (filtered out by default). */ +} hwloc_obj_cache_type_t; + +/** \brief Type of one side (upstream or downstream) of an I/O bridge. */ +typedef enum hwloc_obj_bridge_type_e { + HWLOC_OBJ_BRIDGE_HOST, /**< \brief Host-side of a bridge, only possible upstream. */ + HWLOC_OBJ_BRIDGE_PCI /**< \brief PCI-side of a bridge. */ +} hwloc_obj_bridge_type_t; + +/** \brief Type of a OS device. */ +typedef enum hwloc_obj_osdev_type_e { + HWLOC_OBJ_OSDEV_BLOCK, /**< \brief Operating system block device. + * For instance "sda" on Linux. */ + HWLOC_OBJ_OSDEV_GPU, /**< \brief Operating system GPU device. + * For instance ":0.0" for a GL display, + * "card0" for a Linux DRM device. */ + HWLOC_OBJ_OSDEV_NETWORK, /**< \brief Operating system network device. + * For instance the "eth0" interface on Linux. */ + HWLOC_OBJ_OSDEV_OPENFABRICS, /**< \brief Operating system openfabrics device. + * For instance the "mlx4_0" InfiniBand HCA, + * or "hfi1_0" Omni-Path interface on Linux. */ + HWLOC_OBJ_OSDEV_DMA, /**< \brief Operating system dma engine device. + * For instance the "dma0chan0" DMA channel on Linux. */ + HWLOC_OBJ_OSDEV_COPROC /**< \brief Operating system co-processor device. + * For instance "mic0" for a Xeon Phi (MIC) on Linux, + * "opencl0d0" for a OpenCL device, + * "cuda0" for a CUDA device. */ +} hwloc_obj_osdev_type_t; + +/** \brief Compare the depth of two object types + * + * Types shouldn't be compared as they are, since newer ones may be added in + * the future. This function returns less than, equal to, or greater than zero + * respectively if \p type1 objects usually include \p type2 objects, are the + * same as \p type2 objects, or are included in \p type2 objects. If the types + * can not be compared (because neither is usually contained in the other), + * ::HWLOC_TYPE_UNORDERED is returned. Object types containing CPUs can always + * be compared (usually, a system contains machines which contain nodes which + * contain packages which contain caches, which contain cores, which contain + * processors). + * + * \note ::HWLOC_OBJ_PU will always be the deepest, + * while ::HWLOC_OBJ_MACHINE is always the highest. + * + * \note This does not mean that the actual topology will respect that order: + * e.g. as of today cores may also contain caches, and packages may also contain + * nodes. This is thus just to be seen as a fallback comparison method. + */ +HWLOC_DECLSPEC int hwloc_compare_types (hwloc_obj_type_t type1, hwloc_obj_type_t type2) __hwloc_attribute_const; + +enum hwloc_compare_types_e { + HWLOC_TYPE_UNORDERED = INT_MAX /**< \brief Value returned by hwloc_compare_types() when types can not be compared. \hideinitializer */ +}; + +/** @} */ + + + +/** \defgroup hwlocality_objects Object Structure and Attributes + * @{ + */ + +union hwloc_obj_attr_u; + +/** \brief Structure of a topology object + * + * Applications must not modify any field except \p hwloc_obj.userdata. + */ +struct hwloc_obj { + /* physical information */ + hwloc_obj_type_t type; /**< \brief Type of object */ + char *subtype; /**< \brief Subtype string to better describe the type field. */ + + unsigned os_index; /**< \brief OS-provided physical index number. + * It is not guaranteed unique across the entire machine, + * except for PUs and NUMA nodes. + * Set to HWLOC_UNKNOWN_INDEX if unknown or irrelevant for this object. + */ +#define HWLOC_UNKNOWN_INDEX (unsigned)-1 + + char *name; /**< \brief Object-specific name if any. + * Mostly used for identifying OS devices and Misc objects where + * a name string is more useful than numerical indexes. + */ + + hwloc_uint64_t total_memory; /**< \brief Total memory (in bytes) in NUMA nodes below this object. */ + + union hwloc_obj_attr_u *attr; /**< \brief Object type-specific Attributes, + * may be \c NULL if no attribute value was found */ + + /* global position */ + int depth; /**< \brief Vertical index in the hierarchy. + * + * For normal objects, this is the depth of the horizontal level + * that contains this object and its cousins of the same type. + * If the topology is symmetric, this is equal to the parent depth + * plus one, and also equal to the number of parent/child links + * from the root object to here. + * + * For special objects (NUMA nodes, I/O and Misc) that are not + * in the main tree, this is a special negative value that + * corresponds to their dedicated level, + * see hwloc_get_type_depth() and ::hwloc_get_type_depth_e. + * Those special values can be passed to hwloc functions such + * hwloc_get_nbobjs_by_depth() as usual. + */ + unsigned logical_index; /**< \brief Horizontal index in the whole list of similar objects, + * hence guaranteed unique across the entire machine. + * Could be a "cousin_rank" since it's the rank within the "cousin" list below + * Note that this index may change when restricting the topology + * or when inserting a group. + */ + + /* cousins are all objects of the same type (and depth) across the entire topology */ + struct hwloc_obj *next_cousin; /**< \brief Next object of same type and depth */ + struct hwloc_obj *prev_cousin; /**< \brief Previous object of same type and depth */ + + /* children of the same parent are siblings, even if they may have different type and depth */ + struct hwloc_obj *parent; /**< \brief Parent, \c NULL if root (Machine object) */ + unsigned sibling_rank; /**< \brief Index in parent's \c children[] array. Or the index in parent's Memory, I/O or Misc children list. */ + struct hwloc_obj *next_sibling; /**< \brief Next object below the same parent (inside the same list of children). */ + struct hwloc_obj *prev_sibling; /**< \brief Previous object below the same parent (inside the same list of children). */ + /** @name List and array of normal children below this object (except Memory, I/O and Misc children). */ + /**@{*/ + unsigned arity; /**< \brief Number of normal children. + * Memory, Misc and I/O children are not listed here + * but rather in their dedicated children list. + */ + struct hwloc_obj **children; /**< \brief Normal children, \c children[0 .. arity -1] */ + struct hwloc_obj *first_child; /**< \brief First normal child */ + struct hwloc_obj *last_child; /**< \brief Last normal child */ + /**@}*/ + + int symmetric_subtree; /**< \brief Set if the subtree of normal objects below this object is symmetric, + * which means all normal children and their children have identical subtrees. + * + * Memory, I/O and Misc children are ignored. + * + * If set in the topology root object, lstopo may export the topology + * as a synthetic string. + */ + + /** @name List of Memory children below this object. */ + /**@{*/ + unsigned memory_arity; /**< \brief Number of Memory children. + * These children are listed in \p memory_first_child. + */ + struct hwloc_obj *memory_first_child; /**< \brief First Memory child. + * NUMA nodes are listed here (\p memory_arity and \p memory_first_child) + * instead of in the normal children list. + * See also hwloc_obj_type_is_memory(). + */ + /**@}*/ + + /** @name List of I/O children below this object. */ + /**@{*/ + unsigned io_arity; /**< \brief Number of I/O children. + * These children are listed in \p io_first_child. + */ + struct hwloc_obj *io_first_child; /**< \brief First I/O child. + * Bridges, PCI and OS devices are listed here (\p io_arity and \p io_first_child) + * instead of in the normal children list. + * See also hwloc_obj_type_is_io(). + */ + /**@}*/ + + /** @name List of Misc children below this object. */ + /**@{*/ + unsigned misc_arity; /**< \brief Number of Misc children. + * These children are listed in \p misc_first_child. + */ + struct hwloc_obj *misc_first_child; /**< \brief First Misc child. + * Misc objects are listed here (\p misc_arity and \p misc_first_child) + * instead of in the normal children list. + */ + /**@}*/ + + /* cpusets and nodesets */ + hwloc_cpuset_t cpuset; /**< \brief CPUs covered by this object + * + * This is the set of CPUs for which there are PU objects in the topology + * under this object, i.e. which are known to be physically contained in this + * object and known how (the children path between this object and the PU + * objects). + * + * If the ::HWLOC_TOPOLOGY_FLAG_WHOLE_SYSTEM configuration flag is set, + * some of these CPUs may not be allowed for binding, + * see hwloc_topology_get_allowed_cpuset(). + * + * \note All objects have non-NULL CPU and node sets except Misc and I/O objects. + * + * \note Its value must not be changed, hwloc_bitmap_dup() must be used instead. + */ + hwloc_cpuset_t complete_cpuset; /**< \brief The complete CPU set of logical processors of this object, + * + * This may include not only the same as the cpuset field, but also some CPUs for + * which topology information is unknown or incomplete, some offlines CPUs, and + * the CPUs that are ignored when the ::HWLOC_TOPOLOGY_FLAG_WHOLE_SYSTEM flag + * is not set. + * Thus no corresponding PU object may be found in the topology, because the + * precise position is undefined. It is however known that it would be somewhere + * under this object. + * + * \note Its value must not be changed, hwloc_bitmap_dup() must be used instead. + */ + + hwloc_nodeset_t nodeset; /**< \brief NUMA nodes covered by this object or containing this object + * + * This is the set of NUMA nodes for which there are NUMA node objects in the + * topology under or above this object, i.e. which are known to be physically + * contained in this object or containing it and known how (the children path + * between this object and the NUMA node objects). + * + * In the end, these nodes are those that are close to the current object. + * + * If the ::HWLOC_TOPOLOGY_FLAG_WHOLE_SYSTEM configuration flag is set, + * some of these nodes may not be allowed for allocation, + * see hwloc_topology_get_allowed_nodeset(). + * + * If there are no NUMA nodes in the machine, all the memory is close to this + * object, so only the first bit may be set in \p nodeset. + * + * \note All objects have non-NULL CPU and node sets except Misc and I/O objects. + * + * \note Its value must not be changed, hwloc_bitmap_dup() must be used instead. + */ + hwloc_nodeset_t complete_nodeset; /**< \brief The complete NUMA node set of this object, + * + * This may include not only the same as the nodeset field, but also some NUMA + * nodes for which topology information is unknown or incomplete, some offlines + * nodes, and the nodes that are ignored when the ::HWLOC_TOPOLOGY_FLAG_WHOLE_SYSTEM + * flag is not set. + * Thus no corresponding NUMA node object may be found in the topology, because the + * precise position is undefined. It is however known that it would be + * somewhere under this object. + * + * If there are no NUMA nodes in the machine, all the memory is close to this + * object, so only the first bit is set in \p complete_nodeset. + * + * \note Its value must not be changed, hwloc_bitmap_dup() must be used instead. + */ + + struct hwloc_info_s *infos; /**< \brief Array of stringified info type=name. */ + unsigned infos_count; /**< \brief Size of infos array. */ + + /* misc */ + void *userdata; /**< \brief Application-given private data pointer, + * initialized to \c NULL, use it as you wish. + * See hwloc_topology_set_userdata_export_callback() in hwloc/export.h + * if you wish to export this field to XML. */ + + hwloc_uint64_t gp_index; /**< \brief Global persistent index. + * Generated by hwloc, unique across the topology (contrary to os_index) + * and persistent across topology changes (contrary to logical_index). + * Mostly used internally, but could also be used by application to identify objects. + */ +}; +/** + * \brief Convenience typedef; a pointer to a struct hwloc_obj. + */ +typedef struct hwloc_obj * hwloc_obj_t; + +/** \brief Object type-specific Attributes */ +union hwloc_obj_attr_u { + /** \brief NUMA node-specific Object Attributes */ + struct hwloc_numanode_attr_s { + hwloc_uint64_t local_memory; /**< \brief Local memory (in bytes) */ + unsigned page_types_len; /**< \brief Size of array \p page_types */ + /** \brief Array of local memory page types, \c NULL if no local memory and \p page_types is 0. + * + * The array is sorted by increasing \p size fields. + * It contains \p page_types_len slots. + */ + struct hwloc_memory_page_type_s { + hwloc_uint64_t size; /**< \brief Size of pages */ + hwloc_uint64_t count; /**< \brief Number of pages of this size */ + } * page_types; + } numanode; + + /** \brief Cache-specific Object Attributes */ + struct hwloc_cache_attr_s { + hwloc_uint64_t size; /**< \brief Size of cache in bytes */ + unsigned depth; /**< \brief Depth of cache (e.g., L1, L2, ...etc.) */ + unsigned linesize; /**< \brief Cache-line size in bytes. 0 if unknown */ + int associativity; /**< \brief Ways of associativity, + * -1 if fully associative, 0 if unknown */ + hwloc_obj_cache_type_t type; /**< \brief Cache type */ + } cache; + /** \brief Group-specific Object Attributes */ + struct hwloc_group_attr_s { + unsigned depth; /**< \brief Depth of group object. + * It may change if intermediate Group objects are added. */ + unsigned kind; /**< \brief Internally-used kind of group. */ + unsigned subkind; /**< \brief Internally-used subkind to distinguish different levels of groups with same kind */ + unsigned char dont_merge; /**< \brief Flag preventing groups from being automatically merged with identical parent or children. */ + } group; + /** \brief PCI Device specific Object Attributes */ + struct hwloc_pcidev_attr_s { + unsigned short domain; + unsigned char bus, dev, func; + unsigned short class_id; + unsigned short vendor_id, device_id, subvendor_id, subdevice_id; + unsigned char revision; + float linkspeed; /* in GB/s */ + } pcidev; + /** \brief Bridge specific Object Attribues */ + struct hwloc_bridge_attr_s { + union { + struct hwloc_pcidev_attr_s pci; + } upstream; + hwloc_obj_bridge_type_t upstream_type; + union { + struct { + unsigned short domain; + unsigned char secondary_bus, subordinate_bus; + } pci; + } downstream; + hwloc_obj_bridge_type_t downstream_type; + unsigned depth; + } bridge; + /** \brief OS Device specific Object Attributes */ + struct hwloc_osdev_attr_s { + hwloc_obj_osdev_type_t type; + } osdev; +}; + +/** \brief Object info + * + * \sa hwlocality_info_attr + */ +struct hwloc_info_s { + char *name; /**< \brief Info name */ + char *value; /**< \brief Info value */ +}; + +/** @} */ + + + +/** \defgroup hwlocality_creation Topology Creation and Destruction + * @{ + */ + +struct hwloc_topology; +/** \brief Topology context + * + * To be initialized with hwloc_topology_init() and built with hwloc_topology_load(). + */ +typedef struct hwloc_topology * hwloc_topology_t; + +/** \brief Allocate a topology context. + * + * \param[out] topologyp is assigned a pointer to the new allocated context. + * + * \return 0 on success, -1 on error. + */ +HWLOC_DECLSPEC int hwloc_topology_init (hwloc_topology_t *topologyp); + +/** \brief Build the actual topology + * + * Build the actual topology once initialized with hwloc_topology_init() and + * tuned with \ref hwlocality_configuration and \ref hwlocality_setsource routines. + * No other routine may be called earlier using this topology context. + * + * \param topology is the topology to be loaded with objects. + * + * \return 0 on success, -1 on error. + * + * \note On failure, the topology is reinitialized. It should be either + * destroyed with hwloc_topology_destroy() or configured and loaded again. + * + * \note This function may be called only once per topology. + * + * \note The binding of the current thread or process may temporarily change + * during this call but it will be restored before it returns. + * + * \sa hwlocality_configuration and hwlocality_setsource + */ +HWLOC_DECLSPEC int hwloc_topology_load(hwloc_topology_t topology); + +/** \brief Terminate and free a topology context + * + * \param topology is the topology to be freed + */ +HWLOC_DECLSPEC void hwloc_topology_destroy (hwloc_topology_t topology); + +/** \brief Duplicate a topology. + * + * The entire topology structure as well as its objects + * are duplicated into a new one. + * + * This is useful for keeping a backup while modifying a topology. + * + * \note Object userdata is not duplicated since hwloc does not know what it point to. + * The objects of both old and new topologies will point to the same userdata. + */ +HWLOC_DECLSPEC int hwloc_topology_dup(hwloc_topology_t *newtopology, hwloc_topology_t oldtopology); + +/** \brief Verify that the topology is compatible with the current hwloc library. + * + * This is useful when using the same topology structure (in memory) + * in different libraries that may use different hwloc installations + * (for instance if one library embeds a specific version of hwloc, + * while another library uses a default system-wide hwloc installation). + * + * If all libraries/programs use the same hwloc installation, this function + * always returns success. + * + * \return \c 0 on success. + * + * \return \c -1 with \p errno set to \c EINVAL if incompatible. + * + * \note If sharing between processes with hwloc_shmem_topology_write(), + * the relevant check is already performed inside hwloc_shmem_topology_adopt(). + */ +HWLOC_DECLSPEC int hwloc_topology_abi_check(hwloc_topology_t topology); + +/** \brief Run internal checks on a topology structure + * + * The program aborts if an inconsistency is detected in the given topology. + * + * \param topology is the topology to be checked + * + * \note This routine is only useful to developers. + * + * \note The input topology should have been previously loaded with + * hwloc_topology_load(). + */ +HWLOC_DECLSPEC void hwloc_topology_check(hwloc_topology_t topology); + +/** @} */ + + + +/** \defgroup hwlocality_levels Object levels, depths and types + * @{ + * + * Be sure to see the figure in \ref termsanddefs that shows a + * complete topology tree, including depths, child/sibling/cousin + * relationships, and an example of an asymmetric topology where one + * package has fewer caches than its peers. + */ + +/** \brief Get the depth of the hierarchical tree of objects. + * + * This is the depth of ::HWLOC_OBJ_PU objects plus one. + * + * \note NUMA nodes, I/O and Misc objects are ignored when computing + * the depth of the tree (they are placed on special levels). + */ +HWLOC_DECLSPEC int hwloc_topology_get_depth(hwloc_topology_t __hwloc_restrict topology) __hwloc_attribute_pure; + +/** \brief Returns the depth of objects of type \p type. + * + * If no object of this type is present on the underlying architecture, or if + * the OS doesn't provide this kind of information, the function returns + * ::HWLOC_TYPE_DEPTH_UNKNOWN. + * + * If type is absent but a similar type is acceptable, see also + * hwloc_get_type_or_below_depth() and hwloc_get_type_or_above_depth(). + * + * If ::HWLOC_OBJ_GROUP is given, the function may return ::HWLOC_TYPE_DEPTH_MULTIPLE + * if multiple levels of Groups exist. + * + * If a NUMA node, I/O or Misc object type is given, the function returns a virtual + * value because these objects are stored in special levels that are not CPU-related. + * This virtual depth may be passed to other hwloc functions such as + * hwloc_get_obj_by_depth() but it should not be considered as an actual + * depth by the application. In particular, it should not be compared with + * any other object depth or with the entire topology depth. + * \sa hwloc_get_memory_parents_depth(). + * + * \sa hwloc_type_sscanf_as_depth() for returning the depth of objects + * whose type is given as a string. + */ +HWLOC_DECLSPEC int hwloc_get_type_depth (hwloc_topology_t topology, hwloc_obj_type_t type); + +enum hwloc_get_type_depth_e { + HWLOC_TYPE_DEPTH_UNKNOWN = -1, /**< \brief No object of given type exists in the topology. \hideinitializer */ + HWLOC_TYPE_DEPTH_MULTIPLE = -2, /**< \brief Objects of given type exist at different depth in the topology (only for Groups). \hideinitializer */ + HWLOC_TYPE_DEPTH_NUMANODE = -3, /**< \brief Virtual depth for NUMA nodes. \hideinitializer */ + HWLOC_TYPE_DEPTH_BRIDGE = -4, /**< \brief Virtual depth for bridge object level. \hideinitializer */ + HWLOC_TYPE_DEPTH_PCI_DEVICE = -5, /**< \brief Virtual depth for PCI device object level. \hideinitializer */ + HWLOC_TYPE_DEPTH_OS_DEVICE = -6, /**< \brief Virtual depth for software device object level. \hideinitializer */ + HWLOC_TYPE_DEPTH_MISC = -7 /**< \brief Virtual depth for Misc object. \hideinitializer */ +}; + +/** \brief Return the depth of parents where memory objects are attached. + * + * Memory objects have virtual negative depths because they are not part of + * the main CPU-side hierarchy of objects. This depth should not be compared + * with other level depths. + * + * If all Memory objects are attached to Normal parents at the same depth, + * this parent depth may be compared to other as usual, for instance + * for knowing whether NUMA nodes is attached above or below Packages. + * + * \return The depth of Normal parents of all memory children + * if all these parents have the same depth. For instance the depth of + * the Package level if all NUMA nodes are attached to Package objects. + * + * \return ::HWLOC_TYPE_DEPTH_MULTIPLE if Normal parents of all + * memory children do not have the same depth. For instance if some + * NUMA nodes are attached to Packages while others are attached to + * Groups. + */ +HWLOC_DECLSPEC int hwloc_get_memory_parents_depth (hwloc_topology_t topology); + +/** \brief Returns the depth of objects of type \p type or below + * + * If no object of this type is present on the underlying architecture, the + * function returns the depth of the first "present" object typically found + * inside \p type. + * + * This function is only meaningful for normal object types. + * If a memory, I/O or Misc object type is given, the corresponding virtual + * depth is always returned (see hwloc_get_type_depth()). + * + * May return ::HWLOC_TYPE_DEPTH_MULTIPLE for ::HWLOC_OBJ_GROUP just like + * hwloc_get_type_depth(). + */ +static __hwloc_inline int +hwloc_get_type_or_below_depth (hwloc_topology_t topology, hwloc_obj_type_t type) __hwloc_attribute_pure; + +/** \brief Returns the depth of objects of type \p type or above + * + * If no object of this type is present on the underlying architecture, the + * function returns the depth of the first "present" object typically + * containing \p type. + * + * This function is only meaningful for normal object types. + * If a memory, I/O or Misc object type is given, the corresponding virtual + * depth is always returned (see hwloc_get_type_depth()). + * + * May return ::HWLOC_TYPE_DEPTH_MULTIPLE for ::HWLOC_OBJ_GROUP just like + * hwloc_get_type_depth(). + */ +static __hwloc_inline int +hwloc_get_type_or_above_depth (hwloc_topology_t topology, hwloc_obj_type_t type) __hwloc_attribute_pure; + +/** \brief Returns the type of objects at depth \p depth. + * + * \p depth should between 0 and hwloc_topology_get_depth()-1. + * + * \return (hwloc_obj_type_t)-1 if depth \p depth does not exist. + */ +HWLOC_DECLSPEC hwloc_obj_type_t hwloc_get_depth_type (hwloc_topology_t topology, int depth) __hwloc_attribute_pure; + +/** \brief Returns the width of level at depth \p depth. + */ +HWLOC_DECLSPEC unsigned hwloc_get_nbobjs_by_depth (hwloc_topology_t topology, int depth) __hwloc_attribute_pure; + +/** \brief Returns the width of level type \p type + * + * If no object for that type exists, 0 is returned. + * If there are several levels with objects of that type, -1 is returned. + */ +static __hwloc_inline int +hwloc_get_nbobjs_by_type (hwloc_topology_t topology, hwloc_obj_type_t type) __hwloc_attribute_pure; + +/** \brief Returns the top-object of the topology-tree. + * + * Its type is ::HWLOC_OBJ_MACHINE. + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_root_obj (hwloc_topology_t topology) __hwloc_attribute_pure; + +/** \brief Returns the topology object at logical index \p idx from depth \p depth */ +HWLOC_DECLSPEC hwloc_obj_t hwloc_get_obj_by_depth (hwloc_topology_t topology, int depth, unsigned idx) __hwloc_attribute_pure; + +/** \brief Returns the topology object at logical index \p idx with type \p type + * + * If no object for that type exists, \c NULL is returned. + * If there are several levels with objects of that type (::HWLOC_OBJ_GROUP), + * \c NULL is returned and the caller may fallback to hwloc_get_obj_by_depth(). + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_obj_by_type (hwloc_topology_t topology, hwloc_obj_type_t type, unsigned idx) __hwloc_attribute_pure; + +/** \brief Returns the next object at depth \p depth. + * + * If \p prev is \c NULL, return the first object at depth \p depth. + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_next_obj_by_depth (hwloc_topology_t topology, int depth, hwloc_obj_t prev); + +/** \brief Returns the next object of type \p type. + * + * If \p prev is \c NULL, return the first object at type \p type. If + * there are multiple or no depth for given type, return \c NULL and + * let the caller fallback to hwloc_get_next_obj_by_depth(). + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_next_obj_by_type (hwloc_topology_t topology, hwloc_obj_type_t type, + hwloc_obj_t prev); + +/** @} */ + + + +/** \defgroup hwlocality_object_strings Converting between Object Types and Attributes, and Strings + * @{ + */ + +/** \brief Return a constant stringified object type. + * + * This function is the basic way to convert a generic type into a string. + * The output string may be parsed back by hwloc_type_sscanf(). + * + * hwloc_obj_type_snprintf() may return a more precise output for a specific + * object, but it requires the caller to provide the output buffer. + */ +HWLOC_DECLSPEC const char * hwloc_obj_type_string (hwloc_obj_type_t type) __hwloc_attribute_const; + +/** \brief Stringify the type of a given topology object into a human-readable form. + * + * Contrary to hwloc_obj_type_string(), this function includes object-specific + * attributes (such as the Group depth, the Bridge type, or OS device type) + * in the output, and it requires the caller to provide the output buffer. + * + * The output is guaranteed to be the same for all objects of a same topology level. + * + * If \p verbose is 1, longer type names are used, e.g. L1Cache instead of L1. + * + * The output string may be parsed back by hwloc_type_sscanf(). + * + * If \p size is 0, \p string may safely be \c NULL. + * + * \return the number of character that were actually written if not truncating, + * or that would have been written (not including the ending \\0). + */ +HWLOC_DECLSPEC int hwloc_obj_type_snprintf(char * __hwloc_restrict string, size_t size, + hwloc_obj_t obj, + int verbose); + +/** \brief Stringify the attributes of a given topology object into a human-readable form. + * + * Attribute values are separated by \p separator. + * + * Only the major attributes are printed in non-verbose mode. + * + * If \p size is 0, \p string may safely be \c NULL. + * + * \return the number of character that were actually written if not truncating, + * or that would have been written (not including the ending \\0). + */ +HWLOC_DECLSPEC int hwloc_obj_attr_snprintf(char * __hwloc_restrict string, size_t size, + hwloc_obj_t obj, const char * __hwloc_restrict separator, + int verbose); + +/** \brief Return an object type and attributes from a type string. + * + * Convert strings such as "Package" or "L1iCache" into the corresponding types. + * Matching is case-insensitive, and only the first letters are actually + * required to match. + * + * The matched object type is set in \p typep (which cannot be \c NULL). + * + * Type-specific attributes, for instance Cache type, Cache depth, Group depth, + * Bridge type or OS Device type may be returned in \p attrp. + * Attributes that are not specified in the string (for instance "Group" + * without a depth, or "L2Cache" without a cache type) are set to -1. + * + * \p attrp is only filled if not \c NULL and if its size specified in \p attrsize + * is large enough. It should be at least as large as union hwloc_obj_attr_u. + * + * \return 0 if a type was correctly identified, otherwise -1. + * + * \note This function is guaranteed to match any string returned by + * hwloc_obj_type_string() or hwloc_obj_type_snprintf(). + * + * \note This is an extended version of the now deprecated hwloc_obj_type_sscanf(). + */ +HWLOC_DECLSPEC int hwloc_type_sscanf(const char *string, + hwloc_obj_type_t *typep, + union hwloc_obj_attr_u *attrp, size_t attrsize); + +/** \brief Return an object type and its level depth from a type string. + * + * Convert strings such as "Package" or "L1iCache" into the corresponding types + * and return in \p depthp the depth of the corresponding level in the + * topology \p topology. + * + * If no object of this type is present on the underlying architecture, + * ::HWLOC_TYPE_DEPTH_UNKNOWN is returned. + * + * If multiple such levels exist (for instance if giving Group without any depth), + * the function may return ::HWLOC_TYPE_DEPTH_MULTIPLE instead. + * + * The matched object type is set in \p typep if \p typep is non \c NULL. + * + * \note This function is similar to hwloc_type_sscanf() followed + * by hwloc_get_type_depth() but it also automatically disambiguates + * multiple group levels etc. + * + * \note This function is guaranteed to match any string returned by + * hwloc_obj_type_string() or hwloc_obj_type_snprintf(). + */ +HWLOC_DECLSPEC int hwloc_type_sscanf_as_depth(const char *string, + hwloc_obj_type_t *typep, + hwloc_topology_t topology, int *depthp); + +/** @} */ + + + +/** \defgroup hwlocality_info_attr Consulting and Adding Key-Value Info Attributes + * + * @{ + */ + +/** \brief Search the given key name in object infos and return the corresponding value. + * + * If multiple keys match the given name, only the first one is returned. + * + * \return \c NULL if no such key exists. + */ +static __hwloc_inline const char * +hwloc_obj_get_info_by_name(hwloc_obj_t obj, const char *name) __hwloc_attribute_pure; + +/** \brief Add the given info name and value pair to the given object. + * + * The info is appended to the existing info array even if another key + * with the same name already exists. + * + * The input strings are copied before being added in the object infos. + * + * \return \c 0 on success, \c -1 on error. + * + * \note This function may be used to enforce object colors in the lstopo + * graphical output by using "lstopoStyle" as a name and "Background=#rrggbb" + * as a value. See CUSTOM COLORS in the lstopo(1) manpage for details. + * + * \note If \p value contains some non-printable characters, they will + * be dropped when exporting to XML, see hwloc_topology_export_xml() in hwloc/export.h. + */ +HWLOC_DECLSPEC int hwloc_obj_add_info(hwloc_obj_t obj, const char *name, const char *value); + +/** @} */ + + + +/** \defgroup hwlocality_cpubinding CPU binding + * + * Some operating systems only support binding threads or processes to a single PU. + * Others allow binding to larger sets such as entire Cores or Packages or + * even random sets of invididual PUs. In such operating system, the scheduler + * is free to run the task on one of these PU, then migrate it to another PU, etc. + * It is often useful to call hwloc_bitmap_singlify() on the target CPU set before + * passing it to the binding function to avoid these expensive migrations. + * See the documentation of hwloc_bitmap_singlify() for details. + * + * Some operating systems do not provide all hwloc-supported + * mechanisms to bind processes, threads, etc. + * hwloc_topology_get_support() may be used to query about the actual CPU + * binding support in the currently used operating system. + * + * When the requested binding operation is not available and the + * ::HWLOC_CPUBIND_STRICT flag was passed, the function returns -1. + * \p errno is set to \c ENOSYS when it is not possible to bind the requested kind of object + * processes/threads. errno is set to \c EXDEV when the requested cpuset + * can not be enforced (e.g. some systems only allow one CPU, and some + * other systems only allow one NUMA node). + * + * If ::HWLOC_CPUBIND_STRICT was not passed, the function may fail as well, + * or the operating system may use a slightly different operation + * (with side-effects, smaller binding set, etc.) + * when the requested operation is not exactly supported. + * + * The most portable version that should be preferred over the others, + * whenever possible, is the following one which just binds the current program, + * assuming it is single-threaded: + * + * \code + * hwloc_set_cpubind(topology, set, 0), + * \endcode + * + * If the program may be multithreaded, the following one should be preferred + * to only bind the current thread: + * + * \code + * hwloc_set_cpubind(topology, set, HWLOC_CPUBIND_THREAD), + * \endcode + * + * \sa Some example codes are available under doc/examples/ in the source tree. + * + * \note To unbind, just call the binding function with either a full cpuset or + * a cpuset equal to the system cpuset. + * + * \note On some operating systems, CPU binding may have effects on memory binding, see + * ::HWLOC_CPUBIND_NOMEMBIND + * + * \note Running lstopo \--top or hwloc-ps can be a very convenient tool to check + * how binding actually happened. + * @{ + */ + +/** \brief Process/Thread binding flags. + * + * These bit flags can be used to refine the binding policy. + * + * The default (0) is to bind the current process, assumed to be + * single-threaded, in a non-strict way. This is the most portable + * way to bind as all operating systems usually provide it. + * + * \note Not all systems support all kinds of binding. See the + * "Detailed Description" section of \ref hwlocality_cpubinding for a + * description of errors that can occur. + */ +typedef enum { + /** \brief Bind all threads of the current (possibly) multithreaded process. + * \hideinitializer */ + HWLOC_CPUBIND_PROCESS = (1<<0), + + /** \brief Bind current thread of current process. + * \hideinitializer */ + HWLOC_CPUBIND_THREAD = (1<<1), + + /** \brief Request for strict binding from the OS. + * + * By default, when the designated CPUs are all busy while other + * CPUs are idle, operating systems may execute the thread/process + * on those other CPUs instead of the designated CPUs, to let them + * progress anyway. Strict binding means that the thread/process + * will _never_ execute on other cpus than the designated CPUs, even + * when those are busy with other tasks and other CPUs are idle. + * + * \note Depending on the operating system, strict binding may not + * be possible (e.g., the OS does not implement it) or not allowed + * (e.g., for an administrative reasons), and the function will fail + * in that case. + * + * When retrieving the binding of a process, this flag checks + * whether all its threads actually have the same binding. If the + * flag is not given, the binding of each thread will be + * accumulated. + * + * \note This flag is meaningless when retrieving the binding of a + * thread. + * \hideinitializer + */ + HWLOC_CPUBIND_STRICT = (1<<2), + + /** \brief Avoid any effect on memory binding + * + * On some operating systems, some CPU binding function would also + * bind the memory on the corresponding NUMA node. It is often not + * a problem for the application, but if it is, setting this flag + * will make hwloc avoid using OS functions that would also bind + * memory. This will however reduce the support of CPU bindings, + * i.e. potentially return -1 with errno set to ENOSYS in some + * cases. + * + * This flag is only meaningful when used with functions that set + * the CPU binding. It is ignored when used with functions that get + * CPU binding information. + * \hideinitializer + */ + HWLOC_CPUBIND_NOMEMBIND = (1<<3) +} hwloc_cpubind_flags_t; + +/** \brief Bind current process or thread on cpus given in physical bitmap \p set. + * + * \return -1 with errno set to ENOSYS if the action is not supported + * \return -1 with errno set to EXDEV if the binding cannot be enforced + */ +HWLOC_DECLSPEC int hwloc_set_cpubind(hwloc_topology_t topology, hwloc_const_cpuset_t set, int flags); + +/** \brief Get current process or thread binding. + * + * Writes into \p set the physical cpuset which the process or thread (according to \e + * flags) was last bound to. + */ +HWLOC_DECLSPEC int hwloc_get_cpubind(hwloc_topology_t topology, hwloc_cpuset_t set, int flags); + +/** \brief Bind a process \p pid on cpus given in physical bitmap \p set. + * + * \note \p hwloc_pid_t is \p pid_t on Unix platforms, + * and \p HANDLE on native Windows platforms. + * + * \note As a special case on Linux, if a tid (thread ID) is supplied + * instead of a pid (process ID) and ::HWLOC_CPUBIND_THREAD is passed in flags, + * the binding is applied to that specific thread. + * + * \note On non-Linux systems, ::HWLOC_CPUBIND_THREAD can not be used in \p flags. + */ +HWLOC_DECLSPEC int hwloc_set_proc_cpubind(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_const_cpuset_t set, int flags); + +/** \brief Get the current physical binding of process \p pid. + * + * \note \p hwloc_pid_t is \p pid_t on Unix platforms, + * and \p HANDLE on native Windows platforms. + * + * \note As a special case on Linux, if a tid (thread ID) is supplied + * instead of a pid (process ID) and HWLOC_CPUBIND_THREAD is passed in flags, + * the binding for that specific thread is returned. + * + * \note On non-Linux systems, HWLOC_CPUBIND_THREAD can not be used in \p flags. + */ +HWLOC_DECLSPEC int hwloc_get_proc_cpubind(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_cpuset_t set, int flags); + +#ifdef hwloc_thread_t +/** \brief Bind a thread \p thread on cpus given in physical bitmap \p set. + * + * \note \p hwloc_thread_t is \p pthread_t on Unix platforms, + * and \p HANDLE on native Windows platforms. + * + * \note ::HWLOC_CPUBIND_PROCESS can not be used in \p flags. + */ +HWLOC_DECLSPEC int hwloc_set_thread_cpubind(hwloc_topology_t topology, hwloc_thread_t thread, hwloc_const_cpuset_t set, int flags); +#endif + +#ifdef hwloc_thread_t +/** \brief Get the current physical binding of thread \p tid. + * + * \note \p hwloc_thread_t is \p pthread_t on Unix platforms, + * and \p HANDLE on native Windows platforms. + * + * \note ::HWLOC_CPUBIND_PROCESS can not be used in \p flags. + */ +HWLOC_DECLSPEC int hwloc_get_thread_cpubind(hwloc_topology_t topology, hwloc_thread_t thread, hwloc_cpuset_t set, int flags); +#endif + +/** \brief Get the last physical CPU where the current process or thread ran. + * + * The operating system may move some tasks from one processor + * to another at any time according to their binding, + * so this function may return something that is already + * outdated. + * + * \p flags can include either ::HWLOC_CPUBIND_PROCESS or ::HWLOC_CPUBIND_THREAD to + * specify whether the query should be for the whole process (union of all CPUs + * on which all threads are running), or only the current thread. If the + * process is single-threaded, flags can be set to zero to let hwloc use + * whichever method is available on the underlying OS. + */ +HWLOC_DECLSPEC int hwloc_get_last_cpu_location(hwloc_topology_t topology, hwloc_cpuset_t set, int flags); + +/** \brief Get the last physical CPU where a process ran. + * + * The operating system may move some tasks from one processor + * to another at any time according to their binding, + * so this function may return something that is already + * outdated. + * + * \note \p hwloc_pid_t is \p pid_t on Unix platforms, + * and \p HANDLE on native Windows platforms. + * + * \note As a special case on Linux, if a tid (thread ID) is supplied + * instead of a pid (process ID) and ::HWLOC_CPUBIND_THREAD is passed in flags, + * the last CPU location of that specific thread is returned. + * + * \note On non-Linux systems, ::HWLOC_CPUBIND_THREAD can not be used in \p flags. + */ +HWLOC_DECLSPEC int hwloc_get_proc_last_cpu_location(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_cpuset_t set, int flags); + +/** @} */ + + + +/** \defgroup hwlocality_membinding Memory binding + * + * Memory binding can be done three ways: + * + * - explicit memory allocation thanks to hwloc_alloc_membind() and friends: + * the binding will have effect on the memory allocated by these functions. + * - implicit memory binding through binding policy: hwloc_set_membind() and + * friends only define the current policy of the process, which will be + * applied to the subsequent calls to malloc() and friends. + * - migration of existing memory ranges, thanks to hwloc_set_area_membind() + * and friends, which move already-allocated data. + * + * Not all operating systems support all three ways. + * hwloc_topology_get_support() may be used to query about the actual memory + * binding support in the currently used operating system. + * + * When the requested binding operation is not available and the + * ::HWLOC_MEMBIND_STRICT flag was passed, the function returns -1. + * \p errno will be set to \c ENOSYS when the system does support + * the specified action or policy + * (e.g., some systems only allow binding memory on a per-thread + * basis, whereas other systems only allow binding memory for all + * threads in a process). + * \p errno will be set to EXDEV when the requested set can not be enforced + * (e.g., some systems only allow binding memory to a single NUMA node). + * + * If ::HWLOC_MEMBIND_STRICT was not passed, the function may fail as well, + * or the operating system may use a slightly different operation + * (with side-effects, smaller binding set, etc.) + * when the requested operation is not exactly supported. + * + * The most portable form that should be preferred over the others + * whenever possible is as follows. + * It allocates some memory hopefully bound to the specified set. + * To do so, hwloc will possibly have to change the current memory + * binding policy in order to actually get the memory bound, if the OS + * does not provide any other way to simply allocate bound memory + * without changing the policy for all allocations. That is the + * difference with hwloc_alloc_membind(), which will never change the + * current memory binding policy. + * + * \code + * hwloc_alloc_membind_policy(topology, size, set, + * HWLOC_MEMBIND_BIND, 0); + * \endcode + * + * Each hwloc memory binding function takes a bitmap argument that + * is a CPU set by default, or a NUMA memory node set if the flag + * ::HWLOC_MEMBIND_BYNODESET is specified. + * See \ref hwlocality_object_sets and \ref hwlocality_bitmap for a + * discussion of CPU sets and NUMA memory node sets. + * It is also possible to convert between CPU set and node set using + * hwloc_cpuset_to_nodeset() or hwloc_cpuset_from_nodeset(). + * + * Memory binding by CPU set cannot work for CPU-less NUMA memory nodes. + * Binding by nodeset should therefore be preferred whenever possible. + * + * \sa Some example codes are available under doc/examples/ in the source tree. + * + * \note On some operating systems, memory binding affects the CPU + * binding; see ::HWLOC_MEMBIND_NOCPUBIND + * @{ + */ + +/** \brief Memory binding policy. + * + * These constants can be used to choose the binding policy. Only one policy can + * be used at a time (i.e., the values cannot be OR'ed together). + * + * Not all systems support all kinds of binding. + * hwloc_topology_get_support() may be used to query about the actual memory + * binding policy support in the currently used operating system. + * See the "Detailed Description" section of \ref hwlocality_membinding + * for a description of errors that can occur. + */ +typedef enum { + /** \brief Reset the memory allocation policy to the system default. + * Depending on the operating system, this may correspond to + * ::HWLOC_MEMBIND_FIRSTTOUCH (Linux), + * or ::HWLOC_MEMBIND_BIND (AIX, HP-UX, Solaris, Windows). + * This policy is never returned by get membind functions. + * The nodeset argument is ignored. + * \hideinitializer */ + HWLOC_MEMBIND_DEFAULT = 0, + + /** \brief Allocate each memory page individually on the local NUMA + * node of the thread that touches it. + * + * The given nodeset should usually be hwloc_topology_get_topology_nodeset() + * so that the touching thread may run and allocate on any node in the system. + * + * On AIX, if the nodeset is smaller, pages are allocated locally (if the local + * node is in the nodeset) or from a random non-local node (otherwise). + * \hideinitializer */ + HWLOC_MEMBIND_FIRSTTOUCH = 1, + + /** \brief Allocate memory on the specified nodes. + * \hideinitializer */ + HWLOC_MEMBIND_BIND = 2, + + /** \brief Allocate memory on the given nodes in an interleaved + * / round-robin manner. The precise layout of the memory across + * multiple NUMA nodes is OS/system specific. Interleaving can be + * useful when threads distributed across the specified NUMA nodes + * will all be accessing the whole memory range concurrently, since + * the interleave will then balance the memory references. + * \hideinitializer */ + HWLOC_MEMBIND_INTERLEAVE = 3, + + /** \brief For each page bound with this policy, by next time + * it is touched (and next time only), it is moved from its current + * location to the local NUMA node of the thread where the memory + * reference occurred (if it needs to be moved at all). + * \hideinitializer */ + HWLOC_MEMBIND_NEXTTOUCH = 4, + + /** \brief Returned by get_membind() functions when multiple + * threads or parts of a memory area have differing memory binding + * policies. + * Also returned when binding is unknown because binding hooks are empty + * when the topology is loaded from XML without HWLOC_THISSYSTEM=1, etc. + * \hideinitializer */ + HWLOC_MEMBIND_MIXED = -1 +} hwloc_membind_policy_t; + +/** \brief Memory binding flags. + * + * These flags can be used to refine the binding policy. + * All flags can be logically OR'ed together with the exception of + * ::HWLOC_MEMBIND_PROCESS and ::HWLOC_MEMBIND_THREAD; + * these two flags are mutually exclusive. + * + * Not all systems support all kinds of binding. + * hwloc_topology_get_support() may be used to query about the actual memory + * binding support in the currently used operating system. + * See the "Detailed Description" section of \ref hwlocality_membinding + * for a description of errors that can occur. + */ +typedef enum { + /** \brief Set policy for all threads of the specified (possibly + * multithreaded) process. This flag is mutually exclusive with + * ::HWLOC_MEMBIND_THREAD. + * \hideinitializer */ + HWLOC_MEMBIND_PROCESS = (1<<0), + + /** \brief Set policy for a specific thread of the current process. + * This flag is mutually exclusive with ::HWLOC_MEMBIND_PROCESS. + * \hideinitializer */ + HWLOC_MEMBIND_THREAD = (1<<1), + + /** Request strict binding from the OS. The function will fail if + * the binding can not be guaranteed / completely enforced. + * + * This flag has slightly different meanings depending on which + * function it is used with. + * \hideinitializer */ + HWLOC_MEMBIND_STRICT = (1<<2), + + /** \brief Migrate existing allocated memory. If the memory cannot + * be migrated and the ::HWLOC_MEMBIND_STRICT flag is passed, an error + * will be returned. + * \hideinitializer */ + HWLOC_MEMBIND_MIGRATE = (1<<3), + + /** \brief Avoid any effect on CPU binding. + * + * On some operating systems, some underlying memory binding + * functions also bind the application to the corresponding CPU(s). + * Using this flag will cause hwloc to avoid using OS functions that + * could potentially affect CPU bindings. Note, however, that using + * NOCPUBIND may reduce hwloc's overall memory binding + * support. Specifically: some of hwloc's memory binding functions + * may fail with errno set to ENOSYS when used with NOCPUBIND. + * \hideinitializer + */ + HWLOC_MEMBIND_NOCPUBIND = (1<<4), + + /** \brief Consider the bitmap argument as a nodeset. + * + * The bitmap argument is considered a nodeset if this flag is given, + * or a cpuset otherwise by default. + * + * Memory binding by CPU set cannot work for CPU-less NUMA memory nodes. + * Binding by nodeset should therefore be preferred whenever possible. + * \hideinitializer + */ + HWLOC_MEMBIND_BYNODESET = (1<<5) +} hwloc_membind_flags_t; + +/** \brief Set the default memory binding policy of the current + * process or thread to prefer the NUMA node(s) specified by \p set + * + * If neither ::HWLOC_MEMBIND_PROCESS nor ::HWLOC_MEMBIND_THREAD is + * specified, the current process is assumed to be single-threaded. + * This is the most portable form as it permits hwloc to use either + * process-based OS functions or thread-based OS functions, depending + * on which are available. + * + * If ::HWLOC_MEMBIND_BYNODESET is specified, set is considered a nodeset. + * Otherwise it's a cpuset. + * + * \return -1 with errno set to ENOSYS if the action is not supported + * \return -1 with errno set to EXDEV if the binding cannot be enforced + */ +HWLOC_DECLSPEC int hwloc_set_membind(hwloc_topology_t topology, hwloc_const_bitmap_t set, hwloc_membind_policy_t policy, int flags); + +/** \brief Query the default memory binding policy and physical locality of the + * current process or thread. + * + * This function has two output parameters: \p set and \p policy. + * The values returned in these parameters depend on both the \p flags + * passed in and the current memory binding policies and nodesets in + * the queried target. + * + * Passing the ::HWLOC_MEMBIND_PROCESS flag specifies that the query + * target is the current policies and nodesets for all the threads in + * the current process. Passing ::HWLOC_MEMBIND_THREAD specifies that + * the query target is the current policy and nodeset for only the + * thread invoking this function. + * + * If neither of these flags are passed (which is the most portable + * method), the process is assumed to be single threaded. This allows + * hwloc to use either process-based OS functions or thread-based OS + * functions, depending on which are available. + * + * ::HWLOC_MEMBIND_STRICT is only meaningful when ::HWLOC_MEMBIND_PROCESS + * is also specified. In this case, hwloc will check the default + * memory policies and nodesets for all threads in the process. If + * they are not identical, -1 is returned and errno is set to EXDEV. + * If they are identical, the values are returned in \p set and \p + * policy. + * + * Otherwise, if ::HWLOC_MEMBIND_PROCESS is specified (and + * ::HWLOC_MEMBIND_STRICT is \em not specified), the default set + * from each thread is logically OR'ed together. + * If all threads' default policies are the same, \p policy is set to + * that policy. If they are different, \p policy is set to + * ::HWLOC_MEMBIND_MIXED. + * + * In the ::HWLOC_MEMBIND_THREAD case (or when neither + * ::HWLOC_MEMBIND_PROCESS or ::HWLOC_MEMBIND_THREAD is specified), there + * is only one set and policy; they are returned in \p set and + * \p policy, respectively. + * + * If ::HWLOC_MEMBIND_BYNODESET is specified, set is considered a nodeset. + * Otherwise it's a cpuset. + * + * If any other flags are specified, -1 is returned and errno is set + * to EINVAL. + */ +HWLOC_DECLSPEC int hwloc_get_membind(hwloc_topology_t topology, hwloc_bitmap_t set, hwloc_membind_policy_t * policy, int flags); + +/** \brief Set the default memory binding policy of the specified + * process to prefer the NUMA node(s) specified by \p set + * + * If ::HWLOC_MEMBIND_BYNODESET is specified, set is considered a nodeset. + * Otherwise it's a cpuset. + * + * \return -1 with errno set to ENOSYS if the action is not supported + * \return -1 with errno set to EXDEV if the binding cannot be enforced + * + * \note \p hwloc_pid_t is \p pid_t on Unix platforms, + * and \p HANDLE on native Windows platforms. + */ +HWLOC_DECLSPEC int hwloc_set_proc_membind(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_const_bitmap_t set, hwloc_membind_policy_t policy, int flags); + +/** \brief Query the default memory binding policy and physical locality of the + * specified process. + * + * This function has two output parameters: \p set and \p policy. + * The values returned in these parameters depend on both the \p flags + * passed in and the current memory binding policies and nodesets in + * the queried target. + * + * Passing the ::HWLOC_MEMBIND_PROCESS flag specifies that the query + * target is the current policies and nodesets for all the threads in + * the specified process. If ::HWLOC_MEMBIND_PROCESS is not specified + * (which is the most portable method), the process is assumed to be + * single threaded. This allows hwloc to use either process-based OS + * functions or thread-based OS functions, depending on which are + * available. + * + * Note that it does not make sense to pass ::HWLOC_MEMBIND_THREAD to + * this function. + * + * If ::HWLOC_MEMBIND_STRICT is specified, hwloc will check the default + * memory policies and nodesets for all threads in the specified + * process. If they are not identical, -1 is returned and errno is + * set to EXDEV. If they are identical, the values are returned in \p + * set and \p policy. + * + * Otherwise, \p set is set to the logical OR of all threads' + * default set. If all threads' default policies + * are the same, \p policy is set to that policy. If they are + * different, \p policy is set to ::HWLOC_MEMBIND_MIXED. + * + * If ::HWLOC_MEMBIND_BYNODESET is specified, set is considered a nodeset. + * Otherwise it's a cpuset. + * + * If any other flags are specified, -1 is returned and errno is set + * to EINVAL. + * + * \note \p hwloc_pid_t is \p pid_t on Unix platforms, + * and \p HANDLE on native Windows platforms. + */ +HWLOC_DECLSPEC int hwloc_get_proc_membind(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_bitmap_t set, hwloc_membind_policy_t * policy, int flags); + +/** \brief Bind the already-allocated memory identified by (addr, len) + * to the NUMA node(s) specified by \p set. + * + * If ::HWLOC_MEMBIND_BYNODESET is specified, set is considered a nodeset. + * Otherwise it's a cpuset. + * + * \return 0 if \p len is 0. + * \return -1 with errno set to ENOSYS if the action is not supported + * \return -1 with errno set to EXDEV if the binding cannot be enforced + */ +HWLOC_DECLSPEC int hwloc_set_area_membind(hwloc_topology_t topology, const void *addr, size_t len, hwloc_const_bitmap_t set, hwloc_membind_policy_t policy, int flags); + +/** \brief Query the CPUs near the physical NUMA node(s) and binding policy of + * the memory identified by (\p addr, \p len ). + * + * This function has two output parameters: \p set and \p policy. + * The values returned in these parameters depend on both the \p flags + * passed in and the memory binding policies and nodesets of the pages + * in the address range. + * + * If ::HWLOC_MEMBIND_STRICT is specified, the target pages are first + * checked to see if they all have the same memory binding policy and + * nodeset. If they do not, -1 is returned and errno is set to EXDEV. + * If they are identical across all pages, the set and policy are + * returned in \p set and \p policy, respectively. + * + * If ::HWLOC_MEMBIND_STRICT is not specified, the union of all NUMA + * node(s) containing pages in the address range is calculated. + * If all pages in the target have the same policy, it is returned in + * \p policy. Otherwise, \p policy is set to ::HWLOC_MEMBIND_MIXED. + * + * If ::HWLOC_MEMBIND_BYNODESET is specified, set is considered a nodeset. + * Otherwise it's a cpuset. + * + * If any other flags are specified, -1 is returned and errno is set + * to EINVAL. + * + * If \p len is 0, -1 is returned and errno is set to EINVAL. + */ +HWLOC_DECLSPEC int hwloc_get_area_membind(hwloc_topology_t topology, const void *addr, size_t len, hwloc_bitmap_t set, hwloc_membind_policy_t * policy, int flags); + +/** \brief Get the NUMA nodes where memory identified by (\p addr, \p len ) is physically allocated. + * + * Fills \p set according to the NUMA nodes where the memory area pages + * are physically allocated. If no page is actually allocated yet, + * \p set may be empty. + * + * If pages spread to multiple nodes, it is not specified whether they spread + * equitably, or whether most of them are on a single node, etc. + * + * The operating system may move memory pages from one processor + * to another at any time according to their binding, + * so this function may return something that is already + * outdated. + * + * If ::HWLOC_MEMBIND_BYNODESET is specified in \p flags, set is + * considered a nodeset. Otherwise it's a cpuset. + * + * If \p len is 0, \p set is emptied. + */ +HWLOC_DECLSPEC int hwloc_get_area_memlocation(hwloc_topology_t topology, const void *addr, size_t len, hwloc_bitmap_t set, int flags); + +/** \brief Allocate some memory + * + * This is equivalent to malloc(), except that it tries to allocate + * page-aligned memory from the OS. + * + * \note The allocated memory should be freed with hwloc_free(). + */ +HWLOC_DECLSPEC void *hwloc_alloc(hwloc_topology_t topology, size_t len); + +/** \brief Allocate some memory on NUMA memory nodes specified by \p set + * + * \return NULL with errno set to ENOSYS if the action is not supported + * and ::HWLOC_MEMBIND_STRICT is given + * \return NULL with errno set to EXDEV if the binding cannot be enforced + * and ::HWLOC_MEMBIND_STRICT is given + * \return NULL with errno set to ENOMEM if the memory allocation failed + * even before trying to bind. + * + * If ::HWLOC_MEMBIND_BYNODESET is specified, set is considered a nodeset. + * Otherwise it's a cpuset. + * + * \note The allocated memory should be freed with hwloc_free(). + */ +HWLOC_DECLSPEC void *hwloc_alloc_membind(hwloc_topology_t topology, size_t len, hwloc_const_bitmap_t set, hwloc_membind_policy_t policy, int flags) __hwloc_attribute_malloc; + +/** \brief Allocate some memory on NUMA memory nodes specified by \p set + * + * This is similar to hwloc_alloc_membind_nodeset() except that it is allowed to change + * the current memory binding policy, thus providing more binding support, at + * the expense of changing the current state. + * + * If ::HWLOC_MEMBIND_BYNODESET is specified, set is considered a nodeset. + * Otherwise it's a cpuset. + */ +static __hwloc_inline void * +hwloc_alloc_membind_policy(hwloc_topology_t topology, size_t len, hwloc_const_bitmap_t set, hwloc_membind_policy_t policy, int flags) __hwloc_attribute_malloc; + +/** \brief Free memory that was previously allocated by hwloc_alloc() + * or hwloc_alloc_membind(). + */ +HWLOC_DECLSPEC int hwloc_free(hwloc_topology_t topology, void *addr, size_t len); + +/** @} */ + + + +/** \defgroup hwlocality_setsource Changing the Source of Topology Discovery + * + * If none of the functions below is called, the default is to detect all the objects + * of the machine that the caller is allowed to access. + * + * This default behavior may also be modified through environment variables + * if the application did not modify it already. + * Setting HWLOC_XMLFILE in the environment enforces the discovery from a XML + * file as if hwloc_topology_set_xml() had been called. + * Setting HWLOC_SYNTHETIC enforces a synthetic topology as if + * hwloc_topology_set_synthetic() had been called. + * + * Finally, HWLOC_THISSYSTEM enforces the return value of + * hwloc_topology_is_thissystem(). + * + * @{ + */ + +/** \brief Change which process the topology is viewed from. + * + * On some systems, processes may have different views of the machine, for + * instance the set of allowed CPUs. By default, hwloc exposes the view from + * the current process. Calling hwloc_topology_set_pid() permits to make it + * expose the topology of the machine from the point of view of another + * process. + * + * \note \p hwloc_pid_t is \p pid_t on Unix platforms, + * and \p HANDLE on native Windows platforms. + * + * \note -1 is returned and errno is set to ENOSYS on platforms that do not + * support this feature. + */ +HWLOC_DECLSPEC int hwloc_topology_set_pid(hwloc_topology_t __hwloc_restrict topology, hwloc_pid_t pid); + +/** \brief Enable synthetic topology. + * + * Gather topology information from the given \p description, + * a space-separated string of describing + * the object type and arity at each level. + * All types may be omitted (space-separated string of numbers) so that + * hwloc chooses all types according to usual topologies. + * See also the \ref synthetic. + * + * Setting the environment variable HWLOC_SYNTHETIC + * may also result in this behavior. + * + * If \p description was properly parsed and describes a valid topology + * configuration, this function returns 0. + * Otherwise -1 is returned and errno is set to EINVAL. + * + * Note that this function does not actually load topology + * information; it just tells hwloc where to load it from. You'll + * still need to invoke hwloc_topology_load() to actually load the + * topology information. + * + * \note For convenience, this backend provides empty binding hooks which just + * return success. + * + * \note On success, the synthetic component replaces the previously enabled + * component (if any), but the topology is not actually modified until + * hwloc_topology_load(). + */ +HWLOC_DECLSPEC int hwloc_topology_set_synthetic(hwloc_topology_t __hwloc_restrict topology, const char * __hwloc_restrict description); + +/** \brief Enable XML-file based topology. + * + * Gather topology information from the XML file given at \p xmlpath. + * Setting the environment variable HWLOC_XMLFILE may also result in this behavior. + * This file may have been generated earlier with hwloc_topology_export_xml() in hwloc/export.h, + * or lstopo file.xml. + * + * Note that this function does not actually load topology + * information; it just tells hwloc where to load it from. You'll + * still need to invoke hwloc_topology_load() to actually load the + * topology information. + * + * \return -1 with errno set to EINVAL on failure to read the XML file. + * + * \note See also hwloc_topology_set_userdata_import_callback() + * for importing application-specific object userdata. + * + * \note For convenience, this backend provides empty binding hooks which just + * return success. To have hwloc still actually call OS-specific hooks, the + * ::HWLOC_TOPOLOGY_FLAG_IS_THISSYSTEM has to be set to assert that the loaded + * file is really the underlying system. + * + * \note On success, the XML component replaces the previously enabled + * component (if any), but the topology is not actually modified until + * hwloc_topology_load(). + */ +HWLOC_DECLSPEC int hwloc_topology_set_xml(hwloc_topology_t __hwloc_restrict topology, const char * __hwloc_restrict xmlpath); + +/** \brief Enable XML based topology using a memory buffer (instead of + * a file, as with hwloc_topology_set_xml()). + * + * Gather topology information from the XML memory buffer given at \p + * buffer and of length \p size. This buffer may have been filled + * earlier with hwloc_topology_export_xmlbuffer() in hwloc/export.h. + * + * Note that this function does not actually load topology + * information; it just tells hwloc where to load it from. You'll + * still need to invoke hwloc_topology_load() to actually load the + * topology information. + * + * \return -1 with errno set to EINVAL on failure to read the XML buffer. + * + * \note See also hwloc_topology_set_userdata_import_callback() + * for importing application-specific object userdata. + * + * \note For convenience, this backend provides empty binding hooks which just + * return success. To have hwloc still actually call OS-specific hooks, the + * ::HWLOC_TOPOLOGY_FLAG_IS_THISSYSTEM has to be set to assert that the loaded + * file is really the underlying system. + * + * \note On success, the XML component replaces the previously enabled + * component (if any), but the topology is not actually modified until + * hwloc_topology_load(). + */ +HWLOC_DECLSPEC int hwloc_topology_set_xmlbuffer(hwloc_topology_t __hwloc_restrict topology, const char * __hwloc_restrict buffer, int size); + +/** @} */ + + + +/** \defgroup hwlocality_configuration Topology Detection Configuration and Query + * + * Several functions can optionally be called between hwloc_topology_init() and + * hwloc_topology_load() to configure how the detection should be performed, + * e.g. to ignore some objects types, define a synthetic topology, etc. + * + * @{ + */ + +/** \brief Flags to be set onto a topology context before load. + * + * Flags should be given to hwloc_topology_set_flags(). + * They may also be returned by hwloc_topology_get_flags(). + */ +enum hwloc_topology_flags_e { + /** \brief Detect the whole system, ignore reservations. + * + * Gather all resources, even if some were disabled by the administrator. + * For instance, ignore Linux Cgroup/Cpusets and gather all processors and memory nodes. + * + * When this flag is not set, PUs and NUMA nodes that are disallowed are not added to the topology. + * Parent objects (package, core, cache, etc.) are added only if some of their children are allowed. + * + * When this flag is set, the actual sets of allowed PUs and NUMA nodes are given + * by hwloc_topology_get_allowed_cpuset() and hwloc_topology_get_allowed_nodeset(). + * They may be smaller than the root object cpuset and nodeset. + * + * When this flag is not set, all existing PUs and NUMA nodes in the topology + * are allowed. hwloc_topology_get_allowed_cpuset() and hwloc_topology_get_allowed_nodeset() + * are equal to the root object cpuset and nodeset. + * + * If the current topology is exported to XML and reimported later, this flag + * should be set again in the reimported topology so that disallowed resources + * are reimported as well. + * \hideinitializer + */ + HWLOC_TOPOLOGY_FLAG_WHOLE_SYSTEM = (1UL<<0), + + /** \brief Assume that the selected backend provides the topology for the + * system on which we are running. + * + * This forces hwloc_topology_is_thissystem() to return 1, i.e. makes hwloc assume that + * the selected backend provides the topology for the system on which we are running, + * even if it is not the OS-specific backend but the XML backend for instance. + * This means making the binding functions actually call the OS-specific + * system calls and really do binding, while the XML backend would otherwise + * provide empty hooks just returning success. + * + * Setting the environment variable HWLOC_THISSYSTEM may also result in the + * same behavior. + * + * This can be used for efficiency reasons to first detect the topology once, + * save it to an XML file, and quickly reload it later through the XML + * backend, but still having binding functions actually do bind. + * \hideinitializer + */ + HWLOC_TOPOLOGY_FLAG_IS_THISSYSTEM = (1UL<<1), + + /** \brief Get the set of allowed resources from the local operating system even if the topology was loaded from XML or synthetic description. + * + * If the topology was loaded from XML or from a synthetic string, + * restrict it by applying the current process restrictions such as + * Linux Cgroup/Cpuset. + * + * This is useful when the topology is not loaded directly from + * the local machine (e.g. for performance reason) and it comes + * with all resources, while the running process is restricted + * to only parts of the machine. + * + * This flag is ignored unless ::HWLOC_TOPOLOGY_FLAG_IS_THISSYSTEM is + * also set since the loaded topology must match the underlying machine + * where restrictions will be gathered from. + * + * Setting the environment variable HWLOC_THISSYSTEM_ALLOWED_RESOURCES + * would result in the same behavior. + * \hideinitializer + */ + HWLOC_TOPOLOGY_FLAG_THISSYSTEM_ALLOWED_RESOURCES = (1UL<<2) +}; + +/** \brief Set OR'ed flags to non-yet-loaded topology. + * + * Set a OR'ed set of ::hwloc_topology_flags_e onto a topology that was not yet loaded. + * + * If this function is called multiple times, the last invokation will erase + * and replace the set of flags that was previously set. + * + * The flags set in a topology may be retrieved with hwloc_topology_get_flags() + */ +HWLOC_DECLSPEC int hwloc_topology_set_flags (hwloc_topology_t topology, unsigned long flags); + +/** \brief Get OR'ed flags of a topology. + * + * Get the OR'ed set of ::hwloc_topology_flags_e of a topology. + * + * \return the flags previously set with hwloc_topology_set_flags(). + */ +HWLOC_DECLSPEC unsigned long hwloc_topology_get_flags (hwloc_topology_t topology); + +/** \brief Does the topology context come from this system? + * + * \return 1 if this topology context was built using the system + * running this program. + * \return 0 instead (for instance if using another file-system root, + * a XML topology file, or a synthetic topology). + */ +HWLOC_DECLSPEC int hwloc_topology_is_thissystem(hwloc_topology_t __hwloc_restrict topology) __hwloc_attribute_pure; + +/** \brief Flags describing actual discovery support for this topology. */ +struct hwloc_topology_discovery_support { + /** \brief Detecting the number of PU objects is supported. */ + unsigned char pu; + /** \brief Detecting the number of NUMA nodes is supported. */ + unsigned char numa; + /** \brief Detecting the amount of memory in NUMA nodes is supported. */ + unsigned char numa_memory; +}; + +/** \brief Flags describing actual PU binding support for this topology. + * + * A flag may be set even if the feature isn't supported in all cases + * (e.g. binding to random sets of non-contiguous objects). + */ +struct hwloc_topology_cpubind_support { + /** Binding the whole current process is supported. */ + unsigned char set_thisproc_cpubind; + /** Getting the binding of the whole current process is supported. */ + unsigned char get_thisproc_cpubind; + /** Binding a whole given process is supported. */ + unsigned char set_proc_cpubind; + /** Getting the binding of a whole given process is supported. */ + unsigned char get_proc_cpubind; + /** Binding the current thread only is supported. */ + unsigned char set_thisthread_cpubind; + /** Getting the binding of the current thread only is supported. */ + unsigned char get_thisthread_cpubind; + /** Binding a given thread only is supported. */ + unsigned char set_thread_cpubind; + /** Getting the binding of a given thread only is supported. */ + unsigned char get_thread_cpubind; + /** Getting the last processors where the whole current process ran is supported */ + unsigned char get_thisproc_last_cpu_location; + /** Getting the last processors where a whole process ran is supported */ + unsigned char get_proc_last_cpu_location; + /** Getting the last processors where the current thread ran is supported */ + unsigned char get_thisthread_last_cpu_location; +}; + +/** \brief Flags describing actual memory binding support for this topology. + * + * A flag may be set even if the feature isn't supported in all cases + * (e.g. binding to random sets of non-contiguous objects). + */ +struct hwloc_topology_membind_support { + /** Binding the whole current process is supported. */ + unsigned char set_thisproc_membind; + /** Getting the binding of the whole current process is supported. */ + unsigned char get_thisproc_membind; + /** Binding a whole given process is supported. */ + unsigned char set_proc_membind; + /** Getting the binding of a whole given process is supported. */ + unsigned char get_proc_membind; + /** Binding the current thread only is supported. */ + unsigned char set_thisthread_membind; + /** Getting the binding of the current thread only is supported. */ + unsigned char get_thisthread_membind; + /** Binding a given memory area is supported. */ + unsigned char set_area_membind; + /** Getting the binding of a given memory area is supported. */ + unsigned char get_area_membind; + /** Allocating a bound memory area is supported. */ + unsigned char alloc_membind; + /** First-touch policy is supported. */ + unsigned char firsttouch_membind; + /** Bind policy is supported. */ + unsigned char bind_membind; + /** Interleave policy is supported. */ + unsigned char interleave_membind; + /** Next-touch migration policy is supported. */ + unsigned char nexttouch_membind; + /** Migration flags is supported. */ + unsigned char migrate_membind; + /** Getting the last NUMA nodes where a memory area was allocated is supported */ + unsigned char get_area_memlocation; +}; + +/** \brief Set of flags describing actual support for this topology. + * + * This is retrieved with hwloc_topology_get_support() and will be valid until + * the topology object is destroyed. Note: the values are correct only after + * discovery. + */ +struct hwloc_topology_support { + struct hwloc_topology_discovery_support *discovery; + struct hwloc_topology_cpubind_support *cpubind; + struct hwloc_topology_membind_support *membind; +}; + +/** \brief Retrieve the topology support. + * + * Each flag indicates whether a feature is supported. + * If set to 0, the feature is not supported. + * If set to 1, the feature is supported, but the corresponding + * call may still fail in some corner cases. + * + * These features are also listed by hwloc-info \--support + */ +HWLOC_DECLSPEC const struct hwloc_topology_support *hwloc_topology_get_support(hwloc_topology_t __hwloc_restrict topology); + +/** \brief Type filtering flags. + * + * By default, most objects are kept (::HWLOC_TYPE_FILTER_KEEP_ALL). + * Instruction caches, I/O and Misc objects are ignored by default (::HWLOC_TYPE_FILTER_KEEP_NONE). + * Group levels are ignored unless they bring structure (::HWLOC_TYPE_FILTER_KEEP_STRUCTURE). + * + * Note that group objects are also ignored individually (without the entire level) + * when they do not bring structure. + */ +enum hwloc_type_filter_e { + /** \brief Keep all objects of this type. + * + * Cannot be set for ::HWLOC_OBJ_GROUP (groups are designed only to add more structure to the topology). + * \hideinitializer + */ + HWLOC_TYPE_FILTER_KEEP_ALL = 0, + + /** \brief Ignore all objects of this type. + * + * The bottom-level type ::HWLOC_OBJ_PU, the ::HWLOC_OBJ_NUMANODE type, and + * the top-level type ::HWLOC_OBJ_MACHINE may not be ignored. + * \hideinitializer + */ + HWLOC_TYPE_FILTER_KEEP_NONE = 1, + + /** \brief Only ignore objects if their entire level does not bring any structure. + * + * Keep the entire level of objects if at least one of these objects adds + * structure to the topology. An object brings structure when it has multiple + * children and it is not the only child of its parent. + * + * If all objects in the level are the only child of their parent, and if none + * of them has multiple children, the entire level is removed. + * + * Cannot be set for I/O and Misc objects since the topology structure does not matter there. + * \hideinitializer + */ + HWLOC_TYPE_FILTER_KEEP_STRUCTURE = 2, + + /** \brief Only keep likely-important objects of the given type. + * + * It is only useful for I/O object types. + * For ::HWLOC_OBJ_PCI_DEVICE and ::HWLOC_OBJ_OS_DEVICE, it means that only objects + * of major/common kinds are kept (storage, network, OpenFabrics, Intel MICs, CUDA, + * OpenCL, NVML, and displays). + * Also, only OS devices directly attached on PCI (e.g. no USB) are reported. + * For ::HWLOC_OBJ_BRIDGE, it means that bridges are kept only if they have children. + * + * This flag equivalent to ::HWLOC_TYPE_FILTER_KEEP_ALL for Normal, Memory and Misc types + * since they are likely important. + * \hideinitializer + */ + HWLOC_TYPE_FILTER_KEEP_IMPORTANT = 3 +}; + +/** \brief Set the filtering for the given object type. + */ +HWLOC_DECLSPEC int hwloc_topology_set_type_filter(hwloc_topology_t topology, hwloc_obj_type_t type, enum hwloc_type_filter_e filter); + +/** \brief Get the current filtering for the given object type. + */ +HWLOC_DECLSPEC int hwloc_topology_get_type_filter(hwloc_topology_t topology, hwloc_obj_type_t type, enum hwloc_type_filter_e *filter); + +/** \brief Set the filtering for all object types. + * + * If some types do not support this filtering, they are silently ignored. + */ +HWLOC_DECLSPEC int hwloc_topology_set_all_types_filter(hwloc_topology_t topology, enum hwloc_type_filter_e filter); + +/** \brief Set the filtering for all cache object types. + */ +HWLOC_DECLSPEC int hwloc_topology_set_cache_types_filter(hwloc_topology_t topology, enum hwloc_type_filter_e filter); + +/** \brief Set the filtering for all instruction cache object types. + */ +HWLOC_DECLSPEC int hwloc_topology_set_icache_types_filter(hwloc_topology_t topology, enum hwloc_type_filter_e filter); + +/** \brief Set the filtering for all I/O object types. + */ +HWLOC_DECLSPEC int hwloc_topology_set_io_types_filter(hwloc_topology_t topology, enum hwloc_type_filter_e filter); + +/** \brief Set the topology-specific userdata pointer. + * + * Each topology may store one application-given private data pointer. + * It is initialized to \c NULL. + * hwloc will never modify it. + * + * Use it as you wish, after hwloc_topology_init() and until hwloc_topolog_destroy(). + * + * This pointer is not exported to XML. + */ +HWLOC_DECLSPEC void hwloc_topology_set_userdata(hwloc_topology_t topology, const void *userdata); + +/** \brief Retrieve the topology-specific userdata pointer. + * + * Retrieve the application-given private data pointer that was + * previously set with hwloc_topology_set_userdata(). + */ +HWLOC_DECLSPEC void * hwloc_topology_get_userdata(hwloc_topology_t topology); + +/** @} */ + + + +/** \defgroup hwlocality_tinker Modifying a loaded Topology + * @{ + */ + +/** \brief Flags to be given to hwloc_topology_restrict(). */ +enum hwloc_restrict_flags_e { + /** \brief Remove all objects that became CPU-less. + * By default, only objects that contain no PU and no memory are removed. + * \hideinitializer + */ + HWLOC_RESTRICT_FLAG_REMOVE_CPULESS = (1UL<<0), + + /** \brief Move Misc objects to ancestors if their parents are removed during restriction. + * If this flag is not set, Misc objects are removed when their parents are removed. + * \hideinitializer + */ + HWLOC_RESTRICT_FLAG_ADAPT_MISC = (1UL<<1), + + /** \brief Move I/O objects to ancestors if their parents are removed during restriction. + * If this flag is not set, I/O devices and bridges are removed when their parents are removed. + * \hideinitializer + */ + HWLOC_RESTRICT_FLAG_ADAPT_IO = (1UL<<2) +}; + +/** \brief Restrict the topology to the given CPU set. + * + * Topology \p topology is modified so as to remove all objects that + * are not included (or partially included) in the CPU set \p cpuset. + * All objects CPU and node sets are restricted accordingly. + * + * \p flags is a OR'ed set of ::hwloc_restrict_flags_e. + * + * \note This call may not be reverted by restricting back to a larger + * cpuset. Once dropped during restriction, objects may not be brought + * back, except by loading another topology with hwloc_topology_load(). + * + * \return 0 on success. + * + * \return -1 with errno set to EINVAL if the input cpuset is invalid. + * The topology is not modified in this case. + * + * \return -1 with errno set to ENOMEM on failure to allocate internal data. + * The topology is reinitialized in this case. It should be either + * destroyed with hwloc_topology_destroy() or configured and loaded again. + */ +HWLOC_DECLSPEC int hwloc_topology_restrict(hwloc_topology_t __hwloc_restrict topology, hwloc_const_cpuset_t cpuset, unsigned long flags); + +/** \brief Add a MISC object as a leaf of the topology + * + * A new MISC object will be created and inserted into the topology at the + * position given by parent. It is appended to the list of existing Misc children, + * without ever adding any intermediate hierarchy level. This is useful for + * annotating the topology without actually changing the hierarchy. + * + * \p name is supposed to be unique across all Misc objects in the topology. + * It will be duplicated to setup the new object attributes. + * + * The new leaf object will not have any \p cpuset. + * + * \return the newly-created object + * + * \return \c NULL on error. + * + * \return \c NULL if Misc objects are filtered-out of the topology (::HWLOC_TYPE_FILTER_KEEP_NONE). + * + * \note If \p name contains some non-printable characters, they will + * be dropped when exporting to XML, see hwloc_topology_export_xml() in hwloc/export.h. + */ +HWLOC_DECLSPEC hwloc_obj_t hwloc_topology_insert_misc_object(hwloc_topology_t topology, hwloc_obj_t parent, const char *name); + +/** \brief Allocate a Group object to insert later with hwloc_topology_insert_group_object(). + * + * This function returns a new Group object. + * The caller should (at least) initialize its sets before inserting the object. + * See hwloc_topology_insert_group_object(). + * + * The \p subtype object attribute may be set to display something else + * than "Group" as the type name for this object in lstopo. + * Custom name/value info pairs may be added with hwloc_obj_add_info() after + * insertion. + * + * The \p kind group attribute should be 0. The \p subkind group attribute may + * be set to identify multiple Groups of the same level. + * + * It is recommended not to set any other object attribute before insertion, + * since the Group may get discarded during insertion. + * + * The object will be destroyed if passed to hwloc_topology_insert_group_object() + * without any set defined. + */ +HWLOC_DECLSPEC hwloc_obj_t hwloc_topology_alloc_group_object(hwloc_topology_t topology); + +/** \brief Add more structure to the topology by adding an intermediate Group + * + * The caller should first allocate a new Group object with hwloc_topology_alloc_group_object(). + * Then it must setup at least one of its CPU or node sets to specify + * the final location of the Group in the topology. + * Then the object can be passed to this function for actual insertion in the topology. + * + * The group \p dont_merge attribute may be set to prevent the core from + * ever merging this object with another object hierarchically-identical. + * + * Either the cpuset or nodeset field (or both, if compatible) must be set + * to a non-empty bitmap. The complete_cpuset or complete_nodeset may be set + * instead if inserting with respect to the complete topology + * (including disallowed, offline or unknown objects). + * + * It grouping several objects, hwloc_obj_add_other_obj_sets() is an easy way + * to build the Group sets iteratively. + * + * These sets cannot be larger than the current topology, or they would get + * restricted silently. + * + * The core will setup the other sets after actual insertion. + * + * \return The inserted object if it was properly inserted. + * + * \return An existing object if the Group was discarded because the topology already + * contained an object at the same location (the Group did not add any locality information). + * Any name/info key pair set before inserting is appended to the existing object. + * + * \return \c NULL if the insertion failed because of conflicting sets in topology tree. + * + * \return \c NULL if Group objects are filtered-out of the topology (::HWLOC_TYPE_FILTER_KEEP_NONE). + * + * \return \c NULL if the object was discarded because no set was initialized in the Group + * before insert, or all of them were empty. + */ +HWLOC_DECLSPEC hwloc_obj_t hwloc_topology_insert_group_object(hwloc_topology_t topology, hwloc_obj_t group); + +/** \brief Setup object cpusets/nodesets by OR'ing another object's sets. + * + * For each defined cpuset or nodeset in \p src, allocate the corresponding set + * in \p dst and add \p src to it by OR'ing sets. + * + * This function is convenient between hwloc_topology_alloc_group_object() + * and hwloc_topology_insert_group_object(). It builds the sets of the new Group + * that will be inserted as a new intermediate parent of several objects. + */ +HWLOC_DECLSPEC int hwloc_obj_add_other_obj_sets(hwloc_obj_t dst, hwloc_obj_t src); + +/** @} */ + + + +#ifdef __cplusplus +} /* extern "C" */ +#endif + + +/* high-level helpers */ +#include + +/* inline code of some functions above */ +#include + +/* exporting to XML or synthetic */ +#include + +/* distances */ +#include + +/* topology diffs */ +#include + +/* deprecated headers */ +#include + +#endif /* HWLOC_H */ diff --git a/src/3rdparty/hwloc/include/hwloc/autogen/config.h b/src/3rdparty/hwloc/include/hwloc/autogen/config.h new file mode 100644 index 000000000..14d4481d2 --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc/autogen/config.h @@ -0,0 +1,59 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2018 Inria. All rights reserved. + * Copyright © 2009-2012 Université Bordeaux + * Copyright © 2009-2011 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +/* The configuration file */ + +#ifndef HWLOC_CONFIG_H +#define HWLOC_CONFIG_H + +#define HWLOC_VERSION "2.0.4" +#define HWLOC_VERSION_MAJOR 2 +#define HWLOC_VERSION_MINOR 0 +#define HWLOC_VERSION_RELEASE 4 +#define HWLOC_VERSION_GREEK "" + +#define __hwloc_restrict +#define __hwloc_inline __inline + +#define __hwloc_attribute_unused +#define __hwloc_attribute_malloc +#define __hwloc_attribute_const +#define __hwloc_attribute_pure +#define __hwloc_attribute_deprecated +#define __hwloc_attribute_may_alias +#define __hwloc_attribute_warn_unused_result + +/* Defined to 1 if you have the `windows.h' header. */ +#define HWLOC_HAVE_WINDOWS_H 1 +#define hwloc_pid_t HANDLE +#define hwloc_thread_t HANDLE + +#include +#include +typedef DWORDLONG hwloc_uint64_t; + +#if defined( _USRDLL ) /* dynamic linkage */ +#if defined( DECLSPEC_EXPORTS ) +#define HWLOC_DECLSPEC __declspec(dllexport) +#else +#define HWLOC_DECLSPEC __declspec(dllimport) +#endif +#else /* static linkage */ +#define HWLOC_DECLSPEC +#endif + +/* Whether we need to re-define all the hwloc public symbols or not */ +#define HWLOC_SYM_TRANSFORM 0 + +/* The hwloc symbol prefix */ +#define HWLOC_SYM_PREFIX hwloc_ + +/* The hwloc symbol prefix in all caps */ +#define HWLOC_SYM_PREFIX_CAPS HWLOC_ + +#endif /* HWLOC_CONFIG_H */ diff --git a/src/3rdparty/hwloc/include/hwloc/bitmap.h b/src/3rdparty/hwloc/include/hwloc/bitmap.h new file mode 100644 index 000000000..bae623c8c --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc/bitmap.h @@ -0,0 +1,467 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2018 Inria. All rights reserved. + * Copyright © 2009-2012 Université Bordeaux + * Copyright © 2009-2011 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +/** \file + * \brief The bitmap API, for use in hwloc itself. + */ + +#ifndef HWLOC_BITMAP_H +#define HWLOC_BITMAP_H + +#include +#include + + +#ifdef __cplusplus +extern "C" { +#endif + + +/** \defgroup hwlocality_bitmap The bitmap API + * + * The ::hwloc_bitmap_t type represents a set of integers (positive or null). + * A bitmap may be of infinite size (all bits are set after some point). + * A bitmap may even be full if all bits are set. + * + * Bitmaps are used by hwloc for sets of OS processors + * (which may actually be hardware threads) as by ::hwloc_cpuset_t + * (a typedef for ::hwloc_bitmap_t), or sets of NUMA memory nodes + * as ::hwloc_nodeset_t (also a typedef for ::hwloc_bitmap_t). + * Those are used for cpuset and nodeset fields in the ::hwloc_obj structure, + * see \ref hwlocality_object_sets. + * + * Both CPU and node sets are always indexed by OS physical number. + * However users should usually not build CPU and node sets manually + * (e.g. with hwloc_bitmap_set()). + * One should rather use existing object sets and combine them with + * hwloc_bitmap_or(), etc. + * For instance, binding the current thread on a pair of cores may be performed with: + * \code + * hwloc_obj_t core1 = ... , core2 = ... ; + * hwloc_bitmap_t set = hwloc_bitmap_alloc(); + * hwloc_bitmap_or(set, core1->cpuset, core2->cpuset); + * hwloc_set_cpubind(topology, set, HWLOC_CPUBIND_THREAD); + * hwloc_bitmap_free(set); + * \endcode + * + * \note Most functions below return an int that may be negative in case of + * error. The usual error case would be an internal failure to realloc/extend + * the storage of the bitmap (\p errno would be set to \c ENOMEM). + * + * \note Several examples of using the bitmap API are available under the + * doc/examples/ directory in the source tree. + * Regression tests such as tests/hwloc/hwloc_bitmap*.c also make intensive use + * of this API. + * @{ + */ + + +/** \brief + * Set of bits represented as an opaque pointer to an internal bitmap. + */ +typedef struct hwloc_bitmap_s * hwloc_bitmap_t; +/** \brief a non-modifiable ::hwloc_bitmap_t */ +typedef const struct hwloc_bitmap_s * hwloc_const_bitmap_t; + + +/* + * Bitmap allocation, freeing and copying. + */ + +/** \brief Allocate a new empty bitmap. + * + * \returns A valid bitmap or \c NULL. + * + * The bitmap should be freed by a corresponding call to + * hwloc_bitmap_free(). + */ +HWLOC_DECLSPEC hwloc_bitmap_t hwloc_bitmap_alloc(void) __hwloc_attribute_malloc; + +/** \brief Allocate a new full bitmap. */ +HWLOC_DECLSPEC hwloc_bitmap_t hwloc_bitmap_alloc_full(void) __hwloc_attribute_malloc; + +/** \brief Free bitmap \p bitmap. + * + * If \p bitmap is \c NULL, no operation is performed. + */ +HWLOC_DECLSPEC void hwloc_bitmap_free(hwloc_bitmap_t bitmap); + +/** \brief Duplicate bitmap \p bitmap by allocating a new bitmap and copying \p bitmap contents. + * + * If \p bitmap is \c NULL, \c NULL is returned. + */ +HWLOC_DECLSPEC hwloc_bitmap_t hwloc_bitmap_dup(hwloc_const_bitmap_t bitmap) __hwloc_attribute_malloc; + +/** \brief Copy the contents of bitmap \p src into the already allocated bitmap \p dst */ +HWLOC_DECLSPEC int hwloc_bitmap_copy(hwloc_bitmap_t dst, hwloc_const_bitmap_t src); + + +/* + * Bitmap/String Conversion + */ + +/** \brief Stringify a bitmap. + * + * Up to \p buflen characters may be written in buffer \p buf. + * + * If \p buflen is 0, \p buf may safely be \c NULL. + * + * \return the number of character that were actually written if not truncating, + * or that would have been written (not including the ending \\0). + */ +HWLOC_DECLSPEC int hwloc_bitmap_snprintf(char * __hwloc_restrict buf, size_t buflen, hwloc_const_bitmap_t bitmap); + +/** \brief Stringify a bitmap into a newly allocated string. + * + * \return -1 on error. + */ +HWLOC_DECLSPEC int hwloc_bitmap_asprintf(char ** strp, hwloc_const_bitmap_t bitmap); + +/** \brief Parse a bitmap string and stores it in bitmap \p bitmap. + */ +HWLOC_DECLSPEC int hwloc_bitmap_sscanf(hwloc_bitmap_t bitmap, const char * __hwloc_restrict string); + +/** \brief Stringify a bitmap in the list format. + * + * Lists are comma-separated indexes or ranges. + * Ranges are dash separated indexes. + * The last range may not have an ending indexes if the bitmap is infinitely set. + * + * Up to \p buflen characters may be written in buffer \p buf. + * + * If \p buflen is 0, \p buf may safely be \c NULL. + * + * \return the number of character that were actually written if not truncating, + * or that would have been written (not including the ending \\0). + */ +HWLOC_DECLSPEC int hwloc_bitmap_list_snprintf(char * __hwloc_restrict buf, size_t buflen, hwloc_const_bitmap_t bitmap); + +/** \brief Stringify a bitmap into a newly allocated list string. + * + * \return -1 on error. + */ +HWLOC_DECLSPEC int hwloc_bitmap_list_asprintf(char ** strp, hwloc_const_bitmap_t bitmap); + +/** \brief Parse a list string and stores it in bitmap \p bitmap. + */ +HWLOC_DECLSPEC int hwloc_bitmap_list_sscanf(hwloc_bitmap_t bitmap, const char * __hwloc_restrict string); + +/** \brief Stringify a bitmap in the taskset-specific format. + * + * The taskset command manipulates bitmap strings that contain a single + * (possible very long) hexadecimal number starting with 0x. + * + * Up to \p buflen characters may be written in buffer \p buf. + * + * If \p buflen is 0, \p buf may safely be \c NULL. + * + * \return the number of character that were actually written if not truncating, + * or that would have been written (not including the ending \\0). + */ +HWLOC_DECLSPEC int hwloc_bitmap_taskset_snprintf(char * __hwloc_restrict buf, size_t buflen, hwloc_const_bitmap_t bitmap); + +/** \brief Stringify a bitmap into a newly allocated taskset-specific string. + * + * \return -1 on error. + */ +HWLOC_DECLSPEC int hwloc_bitmap_taskset_asprintf(char ** strp, hwloc_const_bitmap_t bitmap); + +/** \brief Parse a taskset-specific bitmap string and stores it in bitmap \p bitmap. + */ +HWLOC_DECLSPEC int hwloc_bitmap_taskset_sscanf(hwloc_bitmap_t bitmap, const char * __hwloc_restrict string); + + +/* + * Building bitmaps. + */ + +/** \brief Empty the bitmap \p bitmap */ +HWLOC_DECLSPEC void hwloc_bitmap_zero(hwloc_bitmap_t bitmap); + +/** \brief Fill bitmap \p bitmap with all possible indexes (even if those objects don't exist or are otherwise unavailable) */ +HWLOC_DECLSPEC void hwloc_bitmap_fill(hwloc_bitmap_t bitmap); + +/** \brief Empty the bitmap \p bitmap and add bit \p id */ +HWLOC_DECLSPEC int hwloc_bitmap_only(hwloc_bitmap_t bitmap, unsigned id); + +/** \brief Fill the bitmap \p and clear the index \p id */ +HWLOC_DECLSPEC int hwloc_bitmap_allbut(hwloc_bitmap_t bitmap, unsigned id); + +/** \brief Setup bitmap \p bitmap from unsigned long \p mask */ +HWLOC_DECLSPEC int hwloc_bitmap_from_ulong(hwloc_bitmap_t bitmap, unsigned long mask); + +/** \brief Setup bitmap \p bitmap from unsigned long \p mask used as \p i -th subset */ +HWLOC_DECLSPEC int hwloc_bitmap_from_ith_ulong(hwloc_bitmap_t bitmap, unsigned i, unsigned long mask); + + +/* + * Modifying bitmaps. + */ + +/** \brief Add index \p id in bitmap \p bitmap */ +HWLOC_DECLSPEC int hwloc_bitmap_set(hwloc_bitmap_t bitmap, unsigned id); + +/** \brief Add indexes from \p begin to \p end in bitmap \p bitmap. + * + * If \p end is \c -1, the range is infinite. + */ +HWLOC_DECLSPEC int hwloc_bitmap_set_range(hwloc_bitmap_t bitmap, unsigned begin, int end); + +/** \brief Replace \p i -th subset of bitmap \p bitmap with unsigned long \p mask */ +HWLOC_DECLSPEC int hwloc_bitmap_set_ith_ulong(hwloc_bitmap_t bitmap, unsigned i, unsigned long mask); + +/** \brief Remove index \p id from bitmap \p bitmap */ +HWLOC_DECLSPEC int hwloc_bitmap_clr(hwloc_bitmap_t bitmap, unsigned id); + +/** \brief Remove indexes from \p begin to \p end in bitmap \p bitmap. + * + * If \p end is \c -1, the range is infinite. + */ +HWLOC_DECLSPEC int hwloc_bitmap_clr_range(hwloc_bitmap_t bitmap, unsigned begin, int end); + +/** \brief Keep a single index among those set in bitmap \p bitmap + * + * May be useful before binding so that the process does not + * have a chance of migrating between multiple logical CPUs + * in the original mask. + * Instead of running the task on any PU inside the given CPU set, + * the operating system scheduler will be forced to run it on a single + * of these PUs. + * It avoids a migration overhead and cache-line ping-pongs between PUs. + * + * \note This function is NOT meant to distribute multiple processes + * within a single CPU set. It always return the same single bit when + * called multiple times on the same input set. hwloc_distrib() may + * be used for generating CPU sets to distribute multiple tasks below + * a single multi-PU object. + * + * \note This function cannot be applied to an object set directly. It + * should be applied to a copy (which may be obtained with hwloc_bitmap_dup()). + */ +HWLOC_DECLSPEC int hwloc_bitmap_singlify(hwloc_bitmap_t bitmap); + + +/* + * Consulting bitmaps. + */ + +/** \brief Convert the beginning part of bitmap \p bitmap into unsigned long \p mask */ +HWLOC_DECLSPEC unsigned long hwloc_bitmap_to_ulong(hwloc_const_bitmap_t bitmap) __hwloc_attribute_pure; + +/** \brief Convert the \p i -th subset of bitmap \p bitmap into unsigned long mask */ +HWLOC_DECLSPEC unsigned long hwloc_bitmap_to_ith_ulong(hwloc_const_bitmap_t bitmap, unsigned i) __hwloc_attribute_pure; + +/** \brief Test whether index \p id is part of bitmap \p bitmap. + * + * \return 1 if the bit at index \p id is set in bitmap \p bitmap, 0 otherwise. + */ +HWLOC_DECLSPEC int hwloc_bitmap_isset(hwloc_const_bitmap_t bitmap, unsigned id) __hwloc_attribute_pure; + +/** \brief Test whether bitmap \p bitmap is empty + * + * \return 1 if bitmap is empty, 0 otherwise. + */ +HWLOC_DECLSPEC int hwloc_bitmap_iszero(hwloc_const_bitmap_t bitmap) __hwloc_attribute_pure; + +/** \brief Test whether bitmap \p bitmap is completely full + * + * \return 1 if bitmap is full, 0 otherwise. + * + * \note A full bitmap is always infinitely set. + */ +HWLOC_DECLSPEC int hwloc_bitmap_isfull(hwloc_const_bitmap_t bitmap) __hwloc_attribute_pure; + +/** \brief Compute the first index (least significant bit) in bitmap \p bitmap + * + * \return -1 if no index is set in \p bitmap. + */ +HWLOC_DECLSPEC int hwloc_bitmap_first(hwloc_const_bitmap_t bitmap) __hwloc_attribute_pure; + +/** \brief Compute the next index in bitmap \p bitmap which is after index \p prev + * + * If \p prev is -1, the first index is returned. + * + * \return -1 if no index with higher index is set in \p bitmap. + */ +HWLOC_DECLSPEC int hwloc_bitmap_next(hwloc_const_bitmap_t bitmap, int prev) __hwloc_attribute_pure; + +/** \brief Compute the last index (most significant bit) in bitmap \p bitmap + * + * \return -1 if no index is set in \p bitmap, or if \p bitmap is infinitely set. + */ +HWLOC_DECLSPEC int hwloc_bitmap_last(hwloc_const_bitmap_t bitmap) __hwloc_attribute_pure; + +/** \brief Compute the "weight" of bitmap \p bitmap (i.e., number of + * indexes that are in the bitmap). + * + * \return the number of indexes that are in the bitmap. + * + * \return -1 if \p bitmap is infinitely set. + */ +HWLOC_DECLSPEC int hwloc_bitmap_weight(hwloc_const_bitmap_t bitmap) __hwloc_attribute_pure; + +/** \brief Compute the first unset index (least significant bit) in bitmap \p bitmap + * + * \return -1 if no index is unset in \p bitmap. + */ +HWLOC_DECLSPEC int hwloc_bitmap_first_unset(hwloc_const_bitmap_t bitmap) __hwloc_attribute_pure; + +/** \brief Compute the next unset index in bitmap \p bitmap which is after index \p prev + * + * If \p prev is -1, the first unset index is returned. + * + * \return -1 if no index with higher index is unset in \p bitmap. + */ +HWLOC_DECLSPEC int hwloc_bitmap_next_unset(hwloc_const_bitmap_t bitmap, int prev) __hwloc_attribute_pure; + +/** \brief Compute the last unset index (most significant bit) in bitmap \p bitmap + * + * \return -1 if no index is unset in \p bitmap, or if \p bitmap is infinitely set. + */ +HWLOC_DECLSPEC int hwloc_bitmap_last_unset(hwloc_const_bitmap_t bitmap) __hwloc_attribute_pure; + +/** \brief Loop macro iterating on bitmap \p bitmap + * + * The loop must start with hwloc_bitmap_foreach_begin() and end + * with hwloc_bitmap_foreach_end() followed by a terminating ';'. + * + * \p index is the loop variable; it should be an unsigned int. The + * first iteration will set \p index to the lowest index in the bitmap. + * Successive iterations will iterate through, in order, all remaining + * indexes set in the bitmap. To be specific: each iteration will return a + * value for \p index such that hwloc_bitmap_isset(bitmap, index) is true. + * + * The assert prevents the loop from being infinite if the bitmap is infinitely set. + * + * \hideinitializer + */ +#define hwloc_bitmap_foreach_begin(id, bitmap) \ +do { \ + assert(hwloc_bitmap_weight(bitmap) != -1); \ + for (id = hwloc_bitmap_first(bitmap); \ + (unsigned) id != (unsigned) -1; \ + id = hwloc_bitmap_next(bitmap, id)) { + +/** \brief End of loop macro iterating on a bitmap. + * + * Needs a terminating ';'. + * + * \sa hwloc_bitmap_foreach_begin() + * \hideinitializer + */ +#define hwloc_bitmap_foreach_end() \ + } \ +} while (0) + + +/* + * Combining bitmaps. + */ + +/** \brief Or bitmaps \p bitmap1 and \p bitmap2 and store the result in bitmap \p res + * + * \p res can be the same as \p bitmap1 or \p bitmap2 + */ +HWLOC_DECLSPEC int hwloc_bitmap_or (hwloc_bitmap_t res, hwloc_const_bitmap_t bitmap1, hwloc_const_bitmap_t bitmap2); + +/** \brief And bitmaps \p bitmap1 and \p bitmap2 and store the result in bitmap \p res + * + * \p res can be the same as \p bitmap1 or \p bitmap2 + */ +HWLOC_DECLSPEC int hwloc_bitmap_and (hwloc_bitmap_t res, hwloc_const_bitmap_t bitmap1, hwloc_const_bitmap_t bitmap2); + +/** \brief And bitmap \p bitmap1 and the negation of \p bitmap2 and store the result in bitmap \p res + * + * \p res can be the same as \p bitmap1 or \p bitmap2 + */ +HWLOC_DECLSPEC int hwloc_bitmap_andnot (hwloc_bitmap_t res, hwloc_const_bitmap_t bitmap1, hwloc_const_bitmap_t bitmap2); + +/** \brief Xor bitmaps \p bitmap1 and \p bitmap2 and store the result in bitmap \p res + * + * \p res can be the same as \p bitmap1 or \p bitmap2 + */ +HWLOC_DECLSPEC int hwloc_bitmap_xor (hwloc_bitmap_t res, hwloc_const_bitmap_t bitmap1, hwloc_const_bitmap_t bitmap2); + +/** \brief Negate bitmap \p bitmap and store the result in bitmap \p res + * + * \p res can be the same as \p bitmap + */ +HWLOC_DECLSPEC int hwloc_bitmap_not (hwloc_bitmap_t res, hwloc_const_bitmap_t bitmap); + + +/* + * Comparing bitmaps. + */ + +/** \brief Test whether bitmaps \p bitmap1 and \p bitmap2 intersects. + * + * \return 1 if bitmaps intersect, 0 otherwise. + */ +HWLOC_DECLSPEC int hwloc_bitmap_intersects (hwloc_const_bitmap_t bitmap1, hwloc_const_bitmap_t bitmap2) __hwloc_attribute_pure; + +/** \brief Test whether bitmap \p sub_bitmap is part of bitmap \p super_bitmap. + * + * \return 1 if \p sub_bitmap is included in \p super_bitmap, 0 otherwise. + * + * \note The empty bitmap is considered included in any other bitmap. + */ +HWLOC_DECLSPEC int hwloc_bitmap_isincluded (hwloc_const_bitmap_t sub_bitmap, hwloc_const_bitmap_t super_bitmap) __hwloc_attribute_pure; + +/** \brief Test whether bitmap \p bitmap1 is equal to bitmap \p bitmap2. + * + * \return 1 if bitmaps are equal, 0 otherwise. + */ +HWLOC_DECLSPEC int hwloc_bitmap_isequal (hwloc_const_bitmap_t bitmap1, hwloc_const_bitmap_t bitmap2) __hwloc_attribute_pure; + +/** \brief Compare bitmaps \p bitmap1 and \p bitmap2 using their lowest index. + * + * A bitmap is considered smaller if its least significant bit is smaller. + * The empty bitmap is considered higher than anything (because its least significant bit does not exist). + * + * \return -1 if \p bitmap1 is considered smaller than \p bitmap2. + * \return 1 if \p bitmap1 is considered larger than \p bitmap2. + * + * For instance comparing binary bitmaps 0011 and 0110 returns -1 + * (hence 0011 is considered smaller than 0110) + * because least significant bit of 0011 (0001) is smaller than least significant bit of 0110 (0010). + * Comparing 01001 and 00110 would also return -1 for the same reason. + * + * \return 0 if bitmaps are considered equal, even if they are not strictly equal. + * They just need to have the same least significant bit. + * For instance, comparing binary bitmaps 0010 and 0110 returns 0 because they have the same least significant bit. + */ +HWLOC_DECLSPEC int hwloc_bitmap_compare_first(hwloc_const_bitmap_t bitmap1, hwloc_const_bitmap_t bitmap2) __hwloc_attribute_pure; + +/** \brief Compare bitmaps \p bitmap1 and \p bitmap2 in lexicographic order. + * + * Lexicographic comparison of bitmaps, starting for their highest indexes. + * Compare last indexes first, then second, etc. + * The empty bitmap is considered lower than anything. + * + * \return -1 if \p bitmap1 is considered smaller than \p bitmap2. + * \return 1 if \p bitmap1 is considered larger than \p bitmap2. + * \return 0 if bitmaps are equal (contrary to hwloc_bitmap_compare_first()). + * + * For instance comparing binary bitmaps 0011 and 0110 returns -1 + * (hence 0011 is considered smaller than 0110). + * Comparing 00101 and 01010 returns -1 too. + * + * \note This is different from the non-existing hwloc_bitmap_compare_last() + * which would only compare the highest index of each bitmap. + */ +HWLOC_DECLSPEC int hwloc_bitmap_compare(hwloc_const_bitmap_t bitmap1, hwloc_const_bitmap_t bitmap2) __hwloc_attribute_pure; + +/** @} */ + + +#ifdef __cplusplus +} /* extern "C" */ +#endif + + +#endif /* HWLOC_BITMAP_H */ diff --git a/src/3rdparty/hwloc/include/hwloc/cuda.h b/src/3rdparty/hwloc/include/hwloc/cuda.h new file mode 100644 index 000000000..77c8473e6 --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc/cuda.h @@ -0,0 +1,220 @@ +/* + * Copyright © 2010-2017 Inria. All rights reserved. + * Copyright © 2010-2011 Université Bordeaux + * Copyright © 2011 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +/** \file + * \brief Macros to help interaction between hwloc and the CUDA Driver API. + * + * Applications that use both hwloc and the CUDA Driver API may want to + * include this file so as to get topology information for CUDA devices. + * + */ + +#ifndef HWLOC_CUDA_H +#define HWLOC_CUDA_H + +#include +#include +#include +#ifdef HWLOC_LINUX_SYS +#include +#endif + +#include + + +#ifdef __cplusplus +extern "C" { +#endif + + +/** \defgroup hwlocality_cuda Interoperability with the CUDA Driver API + * + * This interface offers ways to retrieve topology information about + * CUDA devices when using the CUDA Driver API. + * + * @{ + */ + +/** \brief Return the domain, bus and device IDs of the CUDA device \p cudevice. + * + * Device \p cudevice must match the local machine. + */ +static __hwloc_inline int +hwloc_cuda_get_device_pci_ids(hwloc_topology_t topology __hwloc_attribute_unused, + CUdevice cudevice, int *domain, int *bus, int *dev) +{ + CUresult cres; + +#if CUDA_VERSION >= 4000 + cres = cuDeviceGetAttribute(domain, CU_DEVICE_ATTRIBUTE_PCI_DOMAIN_ID, cudevice); + if (cres != CUDA_SUCCESS) { + errno = ENOSYS; + return -1; + } +#else + *domain = 0; +#endif + cres = cuDeviceGetAttribute(bus, CU_DEVICE_ATTRIBUTE_PCI_BUS_ID, cudevice); + if (cres != CUDA_SUCCESS) { + errno = ENOSYS; + return -1; + } + cres = cuDeviceGetAttribute(dev, CU_DEVICE_ATTRIBUTE_PCI_DEVICE_ID, cudevice); + if (cres != CUDA_SUCCESS) { + errno = ENOSYS; + return -1; + } + + return 0; +} + +/** \brief Get the CPU set of logical processors that are physically + * close to device \p cudevice. + * + * Return the CPU set describing the locality of the CUDA device \p cudevice. + * + * Topology \p topology and device \p cudevice must match the local machine. + * I/O devices detection and the CUDA component are not needed in the topology. + * + * The function only returns the locality of the device. + * If more information about the device is needed, OS objects should + * be used instead, see hwloc_cuda_get_device_osdev() + * and hwloc_cuda_get_device_osdev_by_index(). + * + * This function is currently only implemented in a meaningful way for + * Linux; other systems will simply get a full cpuset. + */ +static __hwloc_inline int +hwloc_cuda_get_device_cpuset(hwloc_topology_t topology __hwloc_attribute_unused, + CUdevice cudevice, hwloc_cpuset_t set) +{ +#ifdef HWLOC_LINUX_SYS + /* If we're on Linux, use the sysfs mechanism to get the local cpus */ +#define HWLOC_CUDA_DEVICE_SYSFS_PATH_MAX 128 + char path[HWLOC_CUDA_DEVICE_SYSFS_PATH_MAX]; + int domainid, busid, deviceid; + + if (hwloc_cuda_get_device_pci_ids(topology, cudevice, &domainid, &busid, &deviceid)) + return -1; + + if (!hwloc_topology_is_thissystem(topology)) { + errno = EINVAL; + return -1; + } + + sprintf(path, "/sys/bus/pci/devices/%04x:%02x:%02x.0/local_cpus", domainid, busid, deviceid); + if (hwloc_linux_read_path_as_cpumask(path, set) < 0 + || hwloc_bitmap_iszero(set)) + hwloc_bitmap_copy(set, hwloc_topology_get_complete_cpuset(topology)); +#else + /* Non-Linux systems simply get a full cpuset */ + hwloc_bitmap_copy(set, hwloc_topology_get_complete_cpuset(topology)); +#endif + return 0; +} + +/** \brief Get the hwloc PCI device object corresponding to the + * CUDA device \p cudevice. + * + * Return the PCI device object describing the CUDA device \p cudevice. + * Return NULL if there is none. + * + * Topology \p topology and device \p cudevice must match the local machine. + * I/O devices detection must be enabled in topology \p topology. + * The CUDA component is not needed in the topology. + */ +static __hwloc_inline hwloc_obj_t +hwloc_cuda_get_device_pcidev(hwloc_topology_t topology, CUdevice cudevice) +{ + int domain, bus, dev; + + if (hwloc_cuda_get_device_pci_ids(topology, cudevice, &domain, &bus, &dev)) + return NULL; + + return hwloc_get_pcidev_by_busid(topology, domain, bus, dev, 0); +} + +/** \brief Get the hwloc OS device object corresponding to CUDA device \p cudevice. + * + * Return the hwloc OS device object that describes the given + * CUDA device \p cudevice. Return NULL if there is none. + * + * Topology \p topology and device \p cudevice must match the local machine. + * I/O devices detection and the CUDA component must be enabled in the topology. + * If not, the locality of the object may still be found using + * hwloc_cuda_get_device_cpuset(). + * + * \note This function cannot work if PCI devices are filtered out. + * + * \note The corresponding hwloc PCI device may be found by looking + * at the result parent pointer (unless PCI devices are filtered out). + */ +static __hwloc_inline hwloc_obj_t +hwloc_cuda_get_device_osdev(hwloc_topology_t topology, CUdevice cudevice) +{ + hwloc_obj_t osdev = NULL; + int domain, bus, dev; + + if (hwloc_cuda_get_device_pci_ids(topology, cudevice, &domain, &bus, &dev)) + return NULL; + + osdev = NULL; + while ((osdev = hwloc_get_next_osdev(topology, osdev)) != NULL) { + hwloc_obj_t pcidev = osdev->parent; + if (strncmp(osdev->name, "cuda", 4)) + continue; + if (pcidev + && pcidev->type == HWLOC_OBJ_PCI_DEVICE + && (int) pcidev->attr->pcidev.domain == domain + && (int) pcidev->attr->pcidev.bus == bus + && (int) pcidev->attr->pcidev.dev == dev + && pcidev->attr->pcidev.func == 0) + return osdev; + /* if PCI are filtered out, we need a info attr to match on */ + } + + return NULL; +} + +/** \brief Get the hwloc OS device object corresponding to the + * CUDA device whose index is \p idx. + * + * Return the OS device object describing the CUDA device whose + * index is \p idx. Return NULL if there is none. + * + * The topology \p topology does not necessarily have to match the current + * machine. For instance the topology may be an XML import of a remote host. + * I/O devices detection and the CUDA component must be enabled in the topology. + * + * \note The corresponding PCI device object can be obtained by looking + * at the OS device parent object (unless PCI devices are filtered out). + * + * \note This function is identical to hwloc_cudart_get_device_osdev_by_index(). + */ +static __hwloc_inline hwloc_obj_t +hwloc_cuda_get_device_osdev_by_index(hwloc_topology_t topology, unsigned idx) +{ + hwloc_obj_t osdev = NULL; + while ((osdev = hwloc_get_next_osdev(topology, osdev)) != NULL) { + if (HWLOC_OBJ_OSDEV_COPROC == osdev->attr->osdev.type + && osdev->name + && !strncmp("cuda", osdev->name, 4) + && atoi(osdev->name + 4) == (int) idx) + return osdev; + } + return NULL; +} + +/** @} */ + + +#ifdef __cplusplus +} /* extern "C" */ +#endif + + +#endif /* HWLOC_CUDA_H */ diff --git a/src/3rdparty/hwloc/include/hwloc/cudart.h b/src/3rdparty/hwloc/include/hwloc/cudart.h new file mode 100644 index 000000000..63c7f59c6 --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc/cudart.h @@ -0,0 +1,177 @@ +/* + * Copyright © 2010-2017 Inria. All rights reserved. + * Copyright © 2010-2011 Université Bordeaux + * Copyright © 2011 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +/** \file + * \brief Macros to help interaction between hwloc and the CUDA Runtime API. + * + * Applications that use both hwloc and the CUDA Runtime API may want to + * include this file so as to get topology information for CUDA devices. + * + */ + +#ifndef HWLOC_CUDART_H +#define HWLOC_CUDART_H + +#include +#include +#include +#ifdef HWLOC_LINUX_SYS +#include +#endif + +#include /* for CUDA_VERSION */ +#include + + +#ifdef __cplusplus +extern "C" { +#endif + + +/** \defgroup hwlocality_cudart Interoperability with the CUDA Runtime API + * + * This interface offers ways to retrieve topology information about + * CUDA devices when using the CUDA Runtime API. + * + * @{ + */ + +/** \brief Return the domain, bus and device IDs of the CUDA device whose index is \p idx. + * + * Device index \p idx must match the local machine. + */ +static __hwloc_inline int +hwloc_cudart_get_device_pci_ids(hwloc_topology_t topology __hwloc_attribute_unused, + int idx, int *domain, int *bus, int *dev) +{ + cudaError_t cerr; + struct cudaDeviceProp prop; + + cerr = cudaGetDeviceProperties(&prop, idx); + if (cerr) { + errno = ENOSYS; + return -1; + } + +#if CUDA_VERSION >= 4000 + *domain = prop.pciDomainID; +#else + *domain = 0; +#endif + + *bus = prop.pciBusID; + *dev = prop.pciDeviceID; + + return 0; +} + +/** \brief Get the CPU set of logical processors that are physically + * close to device \p idx. + * + * Return the CPU set describing the locality of the CUDA device + * whose index is \p idx. + * + * Topology \p topology and device \p idx must match the local machine. + * I/O devices detection and the CUDA component are not needed in the topology. + * + * The function only returns the locality of the device. + * If more information about the device is needed, OS objects should + * be used instead, see hwloc_cudart_get_device_osdev_by_index(). + * + * This function is currently only implemented in a meaningful way for + * Linux; other systems will simply get a full cpuset. + */ +static __hwloc_inline int +hwloc_cudart_get_device_cpuset(hwloc_topology_t topology __hwloc_attribute_unused, + int idx, hwloc_cpuset_t set) +{ +#ifdef HWLOC_LINUX_SYS + /* If we're on Linux, use the sysfs mechanism to get the local cpus */ +#define HWLOC_CUDART_DEVICE_SYSFS_PATH_MAX 128 + char path[HWLOC_CUDART_DEVICE_SYSFS_PATH_MAX]; + int domain, bus, dev; + + if (hwloc_cudart_get_device_pci_ids(topology, idx, &domain, &bus, &dev)) + return -1; + + if (!hwloc_topology_is_thissystem(topology)) { + errno = EINVAL; + return -1; + } + + sprintf(path, "/sys/bus/pci/devices/%04x:%02x:%02x.0/local_cpus", (unsigned) domain, (unsigned) bus, (unsigned) dev); + if (hwloc_linux_read_path_as_cpumask(path, set) < 0 + || hwloc_bitmap_iszero(set)) + hwloc_bitmap_copy(set, hwloc_topology_get_complete_cpuset(topology)); +#else + /* Non-Linux systems simply get a full cpuset */ + hwloc_bitmap_copy(set, hwloc_topology_get_complete_cpuset(topology)); +#endif + return 0; +} + +/** \brief Get the hwloc PCI device object corresponding to the + * CUDA device whose index is \p idx. + * + * Return the PCI device object describing the CUDA device whose + * index is \p idx. Return NULL if there is none. + * + * Topology \p topology and device \p idx must match the local machine. + * I/O devices detection must be enabled in topology \p topology. + * The CUDA component is not needed in the topology. + */ +static __hwloc_inline hwloc_obj_t +hwloc_cudart_get_device_pcidev(hwloc_topology_t topology, int idx) +{ + int domain, bus, dev; + + if (hwloc_cudart_get_device_pci_ids(topology, idx, &domain, &bus, &dev)) + return NULL; + + return hwloc_get_pcidev_by_busid(topology, domain, bus, dev, 0); +} + +/** \brief Get the hwloc OS device object corresponding to the + * CUDA device whose index is \p idx. + * + * Return the OS device object describing the CUDA device whose + * index is \p idx. Return NULL if there is none. + * + * The topology \p topology does not necessarily have to match the current + * machine. For instance the topology may be an XML import of a remote host. + * I/O devices detection and the CUDA component must be enabled in the topology. + * If not, the locality of the object may still be found using + * hwloc_cudart_get_device_cpuset(). + * + * \note The corresponding PCI device object can be obtained by looking + * at the OS device parent object (unless PCI devices are filtered out). + * + * \note This function is identical to hwloc_cuda_get_device_osdev_by_index(). + */ +static __hwloc_inline hwloc_obj_t +hwloc_cudart_get_device_osdev_by_index(hwloc_topology_t topology, unsigned idx) +{ + hwloc_obj_t osdev = NULL; + while ((osdev = hwloc_get_next_osdev(topology, osdev)) != NULL) { + if (HWLOC_OBJ_OSDEV_COPROC == osdev->attr->osdev.type + && osdev->name + && !strncmp("cuda", osdev->name, 4) + && atoi(osdev->name + 4) == (int) idx) + return osdev; + } + return NULL; +} + +/** @} */ + + +#ifdef __cplusplus +} /* extern "C" */ +#endif + + +#endif /* HWLOC_CUDART_H */ diff --git a/src/3rdparty/hwloc/include/hwloc/deprecated.h b/src/3rdparty/hwloc/include/hwloc/deprecated.h new file mode 100644 index 000000000..8f3b1459a --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc/deprecated.h @@ -0,0 +1,206 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2017 Inria. All rights reserved. + * Copyright © 2009-2012 Université Bordeaux + * Copyright © 2009-2010 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +/** + * This file contains the inline code of functions declared in hwloc.h + */ + +#ifndef HWLOC_DEPRECATED_H +#define HWLOC_DEPRECATED_H + +#ifndef HWLOC_H +#error Please include the main hwloc.h instead +#endif + +#ifdef __cplusplus +extern "C" { +#endif + +/* backward compat with v1.11 before System removal */ +#define HWLOC_OBJ_SYSTEM HWLOC_OBJ_MACHINE +/* backward compat with v1.10 before Socket->Package renaming */ +#define HWLOC_OBJ_SOCKET HWLOC_OBJ_PACKAGE +/* backward compat with v1.10 before Node->NUMANode clarification */ +#define HWLOC_OBJ_NODE HWLOC_OBJ_NUMANODE + +/** \brief Insert a misc object by parent. + * + * Identical to hwloc_topology_insert_misc_object(). + */ +static __hwloc_inline hwloc_obj_t +hwloc_topology_insert_misc_object_by_parent(hwloc_topology_t topology, hwloc_obj_t parent, const char *name) __hwloc_attribute_deprecated; +static __hwloc_inline hwloc_obj_t +hwloc_topology_insert_misc_object_by_parent(hwloc_topology_t topology, hwloc_obj_t parent, const char *name) +{ + return hwloc_topology_insert_misc_object(topology, parent, name); +} + +/** \brief Stringify the cpuset containing a set of objects. + * + * If \p size is 0, \p string may safely be \c NULL. + * + * \return the number of character that were actually written if not truncating, + * or that would have been written (not including the ending \\0). + */ +static __hwloc_inline int +hwloc_obj_cpuset_snprintf(char *str, size_t size, size_t nobj, struct hwloc_obj * const *objs) __hwloc_attribute_deprecated; +static __hwloc_inline int +hwloc_obj_cpuset_snprintf(char *str, size_t size, size_t nobj, struct hwloc_obj * const *objs) +{ + hwloc_bitmap_t set = hwloc_bitmap_alloc(); + int res; + unsigned i; + + hwloc_bitmap_zero(set); + for(i=0; icpuset) + hwloc_bitmap_or(set, set, objs[i]->cpuset); + + res = hwloc_bitmap_snprintf(str, size, set); + hwloc_bitmap_free(set); + return res; +} + +/** \brief Convert a type string into a type and some attributes. + * + * Deprecated by hwloc_type_sscanf() + */ +static __hwloc_inline int +hwloc_obj_type_sscanf(const char *string, hwloc_obj_type_t *typep, int *depthattrp, void *typeattrp, size_t typeattrsize) __hwloc_attribute_deprecated; +static __hwloc_inline int +hwloc_obj_type_sscanf(const char *string, hwloc_obj_type_t *typep, int *depthattrp, void *typeattrp, size_t typeattrsize) +{ + union hwloc_obj_attr_u attr; + int err = hwloc_type_sscanf(string, typep, &attr, sizeof(attr)); + if (err < 0) + return err; + if (hwloc_obj_type_is_cache(*typep)) { + if (depthattrp) + *depthattrp = (int) attr.cache.depth; + if (typeattrp && typeattrsize >= sizeof(hwloc_obj_cache_type_t)) + memcpy(typeattrp, &attr.cache.type, sizeof(hwloc_obj_cache_type_t)); + } else if (*typep == HWLOC_OBJ_GROUP) { + if (depthattrp) + *depthattrp = (int) attr.group.depth; + } + return 0; +} + +/** \brief Set the default memory binding policy of the current + * process or thread to prefer the NUMA node(s) specified by physical \p nodeset + */ +static __hwloc_inline int +hwloc_set_membind_nodeset(hwloc_topology_t topology, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags) __hwloc_attribute_deprecated; +static __hwloc_inline int +hwloc_set_membind_nodeset(hwloc_topology_t topology, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags) +{ + return hwloc_set_membind(topology, nodeset, policy, flags | HWLOC_MEMBIND_BYNODESET); +} + +/** \brief Query the default memory binding policy and physical locality of the + * current process or thread. + */ +static __hwloc_inline int +hwloc_get_membind_nodeset(hwloc_topology_t topology, hwloc_nodeset_t nodeset, hwloc_membind_policy_t * policy, int flags) __hwloc_attribute_deprecated; +static __hwloc_inline int +hwloc_get_membind_nodeset(hwloc_topology_t topology, hwloc_nodeset_t nodeset, hwloc_membind_policy_t * policy, int flags) +{ + return hwloc_get_membind(topology, nodeset, policy, flags | HWLOC_MEMBIND_BYNODESET); +} + +/** \brief Set the default memory binding policy of the specified + * process to prefer the NUMA node(s) specified by physical \p nodeset + */ +static __hwloc_inline int +hwloc_set_proc_membind_nodeset(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags) __hwloc_attribute_deprecated; +static __hwloc_inline int +hwloc_set_proc_membind_nodeset(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags) +{ + return hwloc_set_proc_membind(topology, pid, nodeset, policy, flags | HWLOC_MEMBIND_BYNODESET); +} + +/** \brief Query the default memory binding policy and physical locality of the + * specified process. + */ +static __hwloc_inline int +hwloc_get_proc_membind_nodeset(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_nodeset_t nodeset, hwloc_membind_policy_t * policy, int flags) __hwloc_attribute_deprecated; +static __hwloc_inline int +hwloc_get_proc_membind_nodeset(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_nodeset_t nodeset, hwloc_membind_policy_t * policy, int flags) +{ + return hwloc_get_proc_membind(topology, pid, nodeset, policy, flags | HWLOC_MEMBIND_BYNODESET); +} + +/** \brief Bind the already-allocated memory identified by (addr, len) + * to the NUMA node(s) in physical \p nodeset. + */ +static __hwloc_inline int +hwloc_set_area_membind_nodeset(hwloc_topology_t topology, const void *addr, size_t len, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags) __hwloc_attribute_deprecated; +static __hwloc_inline int +hwloc_set_area_membind_nodeset(hwloc_topology_t topology, const void *addr, size_t len, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags) +{ + return hwloc_set_area_membind(topology, addr, len, nodeset, policy, flags | HWLOC_MEMBIND_BYNODESET); +} + +/** \brief Query the physical NUMA node(s) and binding policy of the memory + * identified by (\p addr, \p len ). + */ +static __hwloc_inline int +hwloc_get_area_membind_nodeset(hwloc_topology_t topology, const void *addr, size_t len, hwloc_nodeset_t nodeset, hwloc_membind_policy_t * policy, int flags) __hwloc_attribute_deprecated; +static __hwloc_inline int +hwloc_get_area_membind_nodeset(hwloc_topology_t topology, const void *addr, size_t len, hwloc_nodeset_t nodeset, hwloc_membind_policy_t * policy, int flags) +{ + return hwloc_get_area_membind(topology, addr, len, nodeset, policy, flags | HWLOC_MEMBIND_BYNODESET); +} + +/** \brief Allocate some memory on the given physical nodeset \p nodeset + */ +static __hwloc_inline void * +hwloc_alloc_membind_nodeset(hwloc_topology_t topology, size_t len, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags) __hwloc_attribute_malloc __hwloc_attribute_deprecated; +static __hwloc_inline void * +hwloc_alloc_membind_nodeset(hwloc_topology_t topology, size_t len, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags) +{ + return hwloc_alloc_membind(topology, len, nodeset, policy, flags | HWLOC_MEMBIND_BYNODESET); +} + +/** \brief Allocate some memory on the given nodeset \p nodeset. + */ +static __hwloc_inline void * +hwloc_alloc_membind_policy_nodeset(hwloc_topology_t topology, size_t len, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags) __hwloc_attribute_malloc __hwloc_attribute_deprecated; +static __hwloc_inline void * +hwloc_alloc_membind_policy_nodeset(hwloc_topology_t topology, size_t len, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags) +{ + return hwloc_alloc_membind_policy(topology, len, nodeset, policy, flags | HWLOC_MEMBIND_BYNODESET); +} + +/** \brief Convert a CPU set into a NUMA node set and handle non-NUMA cases + */ +static __hwloc_inline void +hwloc_cpuset_to_nodeset_strict(hwloc_topology_t topology, hwloc_const_cpuset_t _cpuset, hwloc_nodeset_t nodeset) __hwloc_attribute_deprecated; +static __hwloc_inline void +hwloc_cpuset_to_nodeset_strict(hwloc_topology_t topology, hwloc_const_cpuset_t _cpuset, hwloc_nodeset_t nodeset) +{ + hwloc_cpuset_to_nodeset(topology, _cpuset, nodeset); +} + +/** \brief Convert a NUMA node set into a CPU set and handle non-NUMA cases + */ +static __hwloc_inline void +hwloc_cpuset_from_nodeset_strict(hwloc_topology_t topology, hwloc_cpuset_t _cpuset, hwloc_const_nodeset_t nodeset) __hwloc_attribute_deprecated; +static __hwloc_inline void +hwloc_cpuset_from_nodeset_strict(hwloc_topology_t topology, hwloc_cpuset_t _cpuset, hwloc_const_nodeset_t nodeset) +{ + hwloc_cpuset_from_nodeset(topology, _cpuset, nodeset); +} + + +#ifdef __cplusplus +} /* extern "C" */ +#endif + + +#endif /* HWLOC_DEPRECATED_H */ diff --git a/src/3rdparty/hwloc/include/hwloc/diff.h b/src/3rdparty/hwloc/include/hwloc/diff.h new file mode 100644 index 000000000..79f2df3de --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc/diff.h @@ -0,0 +1,289 @@ +/* + * Copyright © 2013-2018 Inria. All rights reserved. + * See COPYING in top-level directory. + */ + +/** \file + * \brief Topology differences. + */ + +#ifndef HWLOC_DIFF_H +#define HWLOC_DIFF_H + +#ifndef HWLOC_H +#error Please include the main hwloc.h instead +#endif + + +#ifdef __cplusplus +extern "C" { +#elif 0 +} +#endif + + +/** \defgroup hwlocality_diff Topology differences + * + * Applications that manipulate many similar topologies, for instance + * one for each node of a homogeneous cluster, may want to compress + * topologies to reduce the memory footprint. + * + * This file offers a way to manipulate the difference between topologies + * and export/import it to/from XML. + * Compression may therefore be achieved by storing one topology + * entirely while the others are only described by their differences + * with the former. + * The actual topology can be reconstructed when actually needed by + * applying the precomputed difference to the reference topology. + * + * This interface targets very similar nodes. + * Only very simple differences between topologies are actually + * supported, for instance a change in the memory size, the name + * of the object, or some info attribute. + * More complex differences such as adding or removing objects cannot + * be represented in the difference structures and therefore return + * errors. + * Differences between object sets or topology-wide allowed sets, + * cannot be represented either. + * + * It means that there is no need to apply the difference when + * looking at the tree organization (how many levels, how many + * objects per level, what kind of objects, CPU and node sets, etc) + * and when binding to objects. + * However the difference must be applied when looking at object + * attributes such as the name, the memory size or info attributes. + * + * @{ + */ + + +/** \brief Type of one object attribute difference. + */ +typedef enum hwloc_topology_diff_obj_attr_type_e { + /** \brief The object local memory is modified. + * The union is a hwloc_topology_diff_obj_attr_u::hwloc_topology_diff_obj_attr_uint64_s + * (and the index field is ignored). + */ + HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_SIZE, + + /** \brief The object name is modified. + * The union is a hwloc_topology_diff_obj_attr_u::hwloc_topology_diff_obj_attr_string_s + * (and the name field is ignored). + */ + + HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_NAME, + /** \brief the value of an info attribute is modified. + * The union is a hwloc_topology_diff_obj_attr_u::hwloc_topology_diff_obj_attr_string_s. + */ + HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_INFO +} hwloc_topology_diff_obj_attr_type_t; + +/** \brief One object attribute difference. + */ +union hwloc_topology_diff_obj_attr_u { + struct hwloc_topology_diff_obj_attr_generic_s { + /* each part of the union must start with these */ + hwloc_topology_diff_obj_attr_type_t type; + } generic; + + /** \brief Integer attribute modification with an optional index. */ + struct hwloc_topology_diff_obj_attr_uint64_s { + /* used for storing integer attributes */ + hwloc_topology_diff_obj_attr_type_t type; + hwloc_uint64_t index; /* not used for SIZE */ + hwloc_uint64_t oldvalue; + hwloc_uint64_t newvalue; + } uint64; + + /** \brief String attribute modification with an optional name */ + struct hwloc_topology_diff_obj_attr_string_s { + /* used for storing name and info pairs */ + hwloc_topology_diff_obj_attr_type_t type; + char *name; /* not used for NAME */ + char *oldvalue; + char *newvalue; + } string; +}; + + +/** \brief Type of one element of a difference list. + */ +typedef enum hwloc_topology_diff_type_e { + /** \brief An object attribute was changed. + * The union is a hwloc_topology_diff_obj_attr_u::hwloc_topology_diff_obj_attr_s. + */ + HWLOC_TOPOLOGY_DIFF_OBJ_ATTR, + + /** \brief The difference is too complex, + * it cannot be represented. The difference below + * this object has not been checked. + * hwloc_topology_diff_build() will return 1. + * + * The union is a hwloc_topology_diff_obj_attr_u::hwloc_topology_diff_too_complex_s. + */ + HWLOC_TOPOLOGY_DIFF_TOO_COMPLEX +} hwloc_topology_diff_type_t; + +/** \brief One element of a difference list between two topologies. + */ +typedef union hwloc_topology_diff_u { + struct hwloc_topology_diff_generic_s { + /* each part of the union must start with these */ + hwloc_topology_diff_type_t type; + union hwloc_topology_diff_u * next; /* pointer to the next element of the list, or NULL */ + } generic; + + /* A difference in an object attribute. */ + struct hwloc_topology_diff_obj_attr_s { + hwloc_topology_diff_type_t type; /* must be ::HWLOC_TOPOLOGY_DIFF_OBJ_ATTR */ + union hwloc_topology_diff_u * next; + /* List of attribute differences for a single object */ + int obj_depth; + unsigned obj_index; + union hwloc_topology_diff_obj_attr_u diff; + } obj_attr; + + /* A difference that is too complex. */ + struct hwloc_topology_diff_too_complex_s { + hwloc_topology_diff_type_t type; /* must be ::HWLOC_TOPOLOGY_DIFF_TOO_COMPLEX */ + union hwloc_topology_diff_u * next; + /* Where we had to stop computing the diff in the first topology */ + int obj_depth; + unsigned obj_index; + } too_complex; +} * hwloc_topology_diff_t; + + +/** \brief Compute the difference between 2 topologies. + * + * The difference is stored as a list of ::hwloc_topology_diff_t entries + * starting at \p diff. + * It is computed by doing a depth-first traversal of both topology trees + * simultaneously. + * + * If the difference between 2 objects is too complex to be represented + * (for instance if some objects have different types, or different numbers + * of children), a special diff entry of type ::HWLOC_TOPOLOGY_DIFF_TOO_COMPLEX + * is queued. + * The computation of the diff does not continue below these objects. + * So each such diff entry means that the difference between two subtrees + * could not be computed. + * + * \return 0 if the difference can be represented properly. + * + * \return 0 with \p diff pointing to NULL if there is no difference + * between the topologies. + * + * \return 1 if the difference is too complex (see above). Some entries in + * the list will be of type ::HWLOC_TOPOLOGY_DIFF_TOO_COMPLEX. + * + * \return -1 on any other error. + * + * \note \p flags is currently not used. It should be 0. + * + * \note The output diff has to be freed with hwloc_topology_diff_destroy(). + * + * \note The output diff can only be exported to XML or passed to + * hwloc_topology_diff_apply() if 0 was returned, i.e. if no entry of type + * ::HWLOC_TOPOLOGY_DIFF_TOO_COMPLEX is listed. + * + * \note The output diff may be modified by removing some entries from + * the list. The removed entries should be freed by passing them to + * to hwloc_topology_diff_destroy() (possible as another list). +*/ +HWLOC_DECLSPEC int hwloc_topology_diff_build(hwloc_topology_t topology, hwloc_topology_t newtopology, unsigned long flags, hwloc_topology_diff_t *diff); + +/** \brief Flags to be given to hwloc_topology_diff_apply(). + */ +enum hwloc_topology_diff_apply_flags_e { + /** \brief Apply topology diff in reverse direction. + * \hideinitializer + */ + HWLOC_TOPOLOGY_DIFF_APPLY_REVERSE = (1UL<<0) +}; + +/** \brief Apply a topology diff to an existing topology. + * + * \p flags is an OR'ed set of ::hwloc_topology_diff_apply_flags_e. + * + * The new topology is modified in place. hwloc_topology_dup() + * may be used to duplicate it before patching. + * + * If the difference cannot be applied entirely, all previous applied + * elements are unapplied before returning. + * + * \return 0 on success. + * + * \return -N if applying the difference failed while trying + * to apply the N-th part of the difference. For instance -1 + * is returned if the very first difference element could not + * be applied. + */ +HWLOC_DECLSPEC int hwloc_topology_diff_apply(hwloc_topology_t topology, hwloc_topology_diff_t diff, unsigned long flags); + +/** \brief Destroy a list of topology differences. + */ +HWLOC_DECLSPEC int hwloc_topology_diff_destroy(hwloc_topology_diff_t diff); + +/** \brief Load a list of topology differences from a XML file. + * + * If not \c NULL, \p refname will be filled with the identifier + * string of the reference topology for the difference file, + * if any was specified in the XML file. + * This identifier is usually the name of the other XML file + * that contains the reference topology. + * + * \note the pointer returned in refname should later be freed + * by the caller. + */ +HWLOC_DECLSPEC int hwloc_topology_diff_load_xml(const char *xmlpath, hwloc_topology_diff_t *diff, char **refname); + +/** \brief Export a list of topology differences to a XML file. + * + * If not \c NULL, \p refname defines an identifier string + * for the reference topology which was used as a base when + * computing this difference. + * This identifier is usually the name of the other XML file + * that contains the reference topology. + * This attribute is given back when reading the diff from XML. + */ +HWLOC_DECLSPEC int hwloc_topology_diff_export_xml(hwloc_topology_diff_t diff, const char *refname, const char *xmlpath); + +/** \brief Load a list of topology differences from a XML buffer. + * + * If not \c NULL, \p refname will be filled with the identifier + * string of the reference topology for the difference file, + * if any was specified in the XML file. + * This identifier is usually the name of the other XML file + * that contains the reference topology. + * + * \note the pointer returned in refname should later be freed + * by the caller. + */ +HWLOC_DECLSPEC int hwloc_topology_diff_load_xmlbuffer(const char *xmlbuffer, int buflen, hwloc_topology_diff_t *diff, char **refname); + +/** \brief Export a list of topology differences to a XML buffer. + * + * If not \c NULL, \p refname defines an identifier string + * for the reference topology which was used as a base when + * computing this difference. + * This identifier is usually the name of the other XML file + * that contains the reference topology. + * This attribute is given back when reading the diff from XML. + * + * The returned buffer ends with a \0 that is included in the returned + * length. + * + * \note The XML buffer should later be freed with hwloc_free_xmlbuffer(). + */ +HWLOC_DECLSPEC int hwloc_topology_diff_export_xmlbuffer(hwloc_topology_diff_t diff, const char *refname, char **xmlbuffer, int *buflen); + +/** @} */ + + +#ifdef __cplusplus +} /* extern "C" */ +#endif + + +#endif /* HWLOC_DIFF_H */ diff --git a/src/3rdparty/hwloc/include/hwloc/distances.h b/src/3rdparty/hwloc/include/hwloc/distances.h new file mode 100644 index 000000000..d523f29fc --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc/distances.h @@ -0,0 +1,271 @@ +/* + * Copyright © 2010-2019 Inria. All rights reserved. + * See COPYING in top-level directory. + */ + +/** \file + * \brief Object distances. + */ + +#ifndef HWLOC_DISTANCES_H +#define HWLOC_DISTANCES_H + +#ifndef HWLOC_H +#error Please include the main hwloc.h instead +#endif + + +#ifdef __cplusplus +extern "C" { +#elif 0 +} +#endif + + +/** \defgroup hwlocality_distances_get Retrieve distances between objects + * @{ + */ + +/** \brief Matrix of distances between a set of objects. + * + * This matrix often contains latencies between NUMA nodes + * (as reported in the System Locality Distance Information Table (SLIT) + * in the ACPI specification), which may or may not be physically accurate. + * It corresponds to the latency for accessing the memory of one node + * from a core in another node. + * The corresponding kind is ::HWLOC_DISTANCES_KIND_FROM_OS | ::HWLOC_DISTANCES_KIND_FROM_USER. + * + * The matrix may also contain bandwidths between random sets of objects, + * possibly provided by the user, as specified in the \p kind attribute. + */ +struct hwloc_distances_s { + unsigned nbobjs; /**< \brief Number of objects described by the distance matrix. */ + hwloc_obj_t *objs; /**< \brief Array of objects described by the distance matrix. + * These objects are not in any particular order, + * see hwloc_distances_obj_index() and hwloc_distances_obj_pair_values() + * for easy ways to find objects in this array and their corresponding values. + */ + unsigned long kind; /**< \brief OR'ed set of ::hwloc_distances_kind_e. */ + hwloc_uint64_t *values; /**< \brief Matrix of distances between objects, stored as a one-dimension array. + * + * Distance from i-th to j-th object is stored in slot i*nbobjs+j. + * The meaning of the value depends on the \p kind attribute. + */ +}; + +/** \brief Kinds of distance matrices. + * + * The \p kind attribute of struct hwloc_distances_s is a OR'ed set + * of kinds. + * + * A kind of format HWLOC_DISTANCES_KIND_FROM_* specifies where the + * distance information comes from, if known. + * + * A kind of format HWLOC_DISTANCES_KIND_MEANS_* specifies whether + * values are latencies or bandwidths, if applicable. + */ +enum hwloc_distances_kind_e { + /** \brief These distances were obtained from the operating system or hardware. + * \hideinitializer + */ + HWLOC_DISTANCES_KIND_FROM_OS = (1UL<<0), + /** \brief These distances were provided by the user. + * \hideinitializer + */ + HWLOC_DISTANCES_KIND_FROM_USER = (1UL<<1), + + /** \brief Distance values are similar to latencies between objects. + * Values are smaller for closer objects, hence minimal on the diagonal + * of the matrix (distance between an object and itself). + * It could also be the number of network hops between objects, etc. + * \hideinitializer + */ + HWLOC_DISTANCES_KIND_MEANS_LATENCY = (1UL<<2), + /** \brief Distance values are similar to bandwidths between objects. + * Values are higher for closer objects, hence maximal on the diagonal + * of the matrix (distance between an object and itself). + * Such values are currently ignored for distance-based grouping. + * \hideinitializer + */ + HWLOC_DISTANCES_KIND_MEANS_BANDWIDTH = (1UL<<3) +}; + +/** \brief Retrieve distance matrices. + * + * Retrieve distance matrices from the topology into the \p distances array. + * + * \p flags is currently unused, should be \c 0. + * + * \p kind serves as a filter. If \c 0, all distance matrices are returned. + * If it contains some HWLOC_DISTANCES_KIND_FROM_*, only distance matrices + * whose kind matches one of these are returned. + * If it contains some HWLOC_DISTANCES_KIND_MEANS_*, only distance matrices + * whose kind matches one of these are returned. + * + * On input, \p nr points to the number of distance matrices that may be stored + * in \p distances. + * On output, \p nr points to the number of distance matrices that were actually + * found, even if some of them couldn't be stored in \p distances. + * Distance matrices that couldn't be stored are ignored, but the function still + * returns success (\c 0). The caller may find out by comparing the value pointed + * by \p nr before and after the function call. + * + * Each distance matrix returned in the \p distances array should be released + * by the caller using hwloc_distances_release(). + */ +HWLOC_DECLSPEC int +hwloc_distances_get(hwloc_topology_t topology, + unsigned *nr, struct hwloc_distances_s **distances, + unsigned long kind, unsigned long flags); + +/** \brief Retrieve distance matrices for object at a specific depth in the topology. + * + * Identical to hwloc_distances_get() with the additional \p depth filter. + */ +HWLOC_DECLSPEC int +hwloc_distances_get_by_depth(hwloc_topology_t topology, int depth, + unsigned *nr, struct hwloc_distances_s **distances, + unsigned long kind, unsigned long flags); + +/** \brief Retrieve distance matrices for object of a specific type. + * + * Identical to hwloc_distances_get() with the additional \p type filter. + */ +static __hwloc_inline int +hwloc_distances_get_by_type(hwloc_topology_t topology, hwloc_obj_type_t type, + unsigned *nr, struct hwloc_distances_s **distances, + unsigned long kind, unsigned long flags) +{ + int depth = hwloc_get_type_depth(topology, type); + if (depth == HWLOC_TYPE_DEPTH_UNKNOWN || depth == HWLOC_TYPE_DEPTH_MULTIPLE) { + *nr = 0; + return 0; + } + return hwloc_distances_get_by_depth(topology, depth, nr, distances, kind, flags); +} + +/** \brief Release a distance matrix structure previously returned by hwloc_distances_get(). */ +HWLOC_DECLSPEC void +hwloc_distances_release(hwloc_topology_t topology, struct hwloc_distances_s *distances); + +/** @} */ + + + +/** \defgroup hwlocality_distances_consult Helpers for consulting distance matrices + * @{ + */ + +/** \brief Find the index of an object in a distances structure. + * + * \return -1 if object \p obj is not involved in structure \p distances. + */ +static __hwloc_inline int +hwloc_distances_obj_index(struct hwloc_distances_s *distances, hwloc_obj_t obj) +{ + unsigned i; + for(i=0; inbobjs; i++) + if (distances->objs[i] == obj) + return (int)i; + return -1; +} + +/** \brief Find the values between two objects in a distance matrices. + * + * The distance from \p obj1 to \p obj2 is stored in the value pointed by + * \p value1to2 and reciprocally. + * + * \return -1 if object \p obj1 or \p obj2 is not involved in structure \p distances. + */ +static __hwloc_inline int +hwloc_distances_obj_pair_values(struct hwloc_distances_s *distances, + hwloc_obj_t obj1, hwloc_obj_t obj2, + hwloc_uint64_t *value1to2, hwloc_uint64_t *value2to1) +{ + int i1 = hwloc_distances_obj_index(distances, obj1); + int i2 = hwloc_distances_obj_index(distances, obj2); + if (i1 < 0 || i2 < 0) + return -1; + *value1to2 = distances->values[i1 * distances->nbobjs + i2]; + *value2to1 = distances->values[i2 * distances->nbobjs + i1]; + return 0; +} + +/** @} */ + + + +/** \defgroup hwlocality_distances_add Add or remove distances between objects + * @{ + */ + +/** \brief Flags for adding a new distances to a topology. */ +enum hwloc_distances_add_flag_e { + /** \brief Try to group objects based on the newly provided distance information. + * \hideinitializer + */ + HWLOC_DISTANCES_ADD_FLAG_GROUP = (1UL<<0), + /** \brief If grouping, consider the distance values as inaccurate and relax the + * comparisons during the grouping algorithms. The actual accuracy may be modified + * through the HWLOC_GROUPING_ACCURACY environment variable (see \ref envvar). + * \hideinitializer + */ + HWLOC_DISTANCES_ADD_FLAG_GROUP_INACCURATE = (1UL<<1) +}; + +/** \brief Provide a new distance matrix. + * + * Provide the matrix of distances between a set of objects given by \p nbobjs + * and the \p objs array. \p nbobjs must be at least 2. + * The distances are stored as a one-dimension array in \p values. + * The distance from object i to object j is in slot i*nbobjs+j. + * + * \p kind specifies the kind of distance as a OR'ed set of ::hwloc_distances_kind_e. + * + * \p flags configures the behavior of the function using an optional OR'ed set of + * ::hwloc_distances_add_flag_e. + * + * Objects must be of the same type. They cannot be of type Group. + */ +HWLOC_DECLSPEC int hwloc_distances_add(hwloc_topology_t topology, + unsigned nbobjs, hwloc_obj_t *objs, hwloc_uint64_t *values, + unsigned long kind, unsigned long flags); + +/** \brief Remove all distance matrices from a topology. + * + * Remove all distance matrices, either provided by the user or + * gathered through the OS. + * + * If these distances were used to group objects, these additional + *Group objects are not removed from the topology. + */ +HWLOC_DECLSPEC int hwloc_distances_remove(hwloc_topology_t topology); + +/** \brief Remove distance matrices for objects at a specific depth in the topology. + * + * Identical to hwloc_distances_remove() but only applies to one level of the topology. + */ +HWLOC_DECLSPEC int hwloc_distances_remove_by_depth(hwloc_topology_t topology, int depth); + +/** \brief Remove distance matrices for objects of a specific type in the topology. + * + * Identical to hwloc_distances_remove() but only applies to one level of the topology. + */ +static __hwloc_inline int +hwloc_distances_remove_by_type(hwloc_topology_t topology, hwloc_obj_type_t type) +{ + int depth = hwloc_get_type_depth(topology, type); + if (depth == HWLOC_TYPE_DEPTH_UNKNOWN || depth == HWLOC_TYPE_DEPTH_MULTIPLE) + return 0; + return hwloc_distances_remove_by_depth(topology, depth); +} + +/** @} */ + + +#ifdef __cplusplus +} /* extern "C" */ +#endif + + +#endif /* HWLOC_DISTANCES_H */ diff --git a/src/3rdparty/hwloc/include/hwloc/export.h b/src/3rdparty/hwloc/include/hwloc/export.h new file mode 100644 index 000000000..b178b77e5 --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc/export.h @@ -0,0 +1,278 @@ +/* + * Copyright © 2009-2018 Inria. All rights reserved. + * Copyright © 2009-2012 Université Bordeaux + * Copyright © 2009-2011 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +/** \file + * \brief Exporting Topologies to XML or to Synthetic strings. + */ + +#ifndef HWLOC_EXPORT_H +#define HWLOC_EXPORT_H + +#ifndef HWLOC_H +#error Please include the main hwloc.h instead +#endif + + +#ifdef __cplusplus +extern "C" { +#elif 0 +} +#endif + + +/** \defgroup hwlocality_xmlexport Exporting Topologies to XML + * @{ + */ + +/** \brief Flags for exporting XML topologies. + * + * Flags to be given as a OR'ed set to hwloc_topology_export_xml(). + */ +enum hwloc_topology_export_xml_flags_e { + /** \brief Export XML that is loadable by hwloc v1.x. + * However, the export may miss some details about the topology. + * \hideinitializer + */ + HWLOC_TOPOLOGY_EXPORT_XML_FLAG_V1 = (1UL<<0) +}; + +/** \brief Export the topology into an XML file. + * + * This file may be loaded later through hwloc_topology_set_xml(). + * + * By default, the latest export format is used, which means older hwloc + * releases (e.g. v1.x) will not be able to import it. + * Exporting to v1.x specific XML format is possible using flag + * ::HWLOC_TOPOLOGY_EXPORT_XML_FLAG_V1 but it may miss some details + * about the topology. + * If there is any chance that the exported file may ever be imported + * back by a process using hwloc 1.x, one should consider detecting + * it at runtime and using the corresponding export format. + * + * \p flags is a OR'ed set of ::hwloc_topology_export_xml_flags_e. + * + * \return -1 if a failure occured. + * + * \note See also hwloc_topology_set_userdata_export_callback() + * for exporting application-specific object userdata. + * + * \note The topology-specific userdata pointer is ignored when exporting to XML. + * + * \note Only printable characters may be exported to XML string attributes. + * Any other character, especially any non-ASCII character, will be silently + * dropped. + * + * \note If \p name is "-", the XML output is sent to the standard output. + */ +HWLOC_DECLSPEC int hwloc_topology_export_xml(hwloc_topology_t topology, const char *xmlpath, unsigned long flags); + +/** \brief Export the topology into a newly-allocated XML memory buffer. + * + * \p xmlbuffer is allocated by the callee and should be freed with + * hwloc_free_xmlbuffer() later in the caller. + * + * This memory buffer may be loaded later through hwloc_topology_set_xmlbuffer(). + * + * By default, the latest export format is used, which means older hwloc + * releases (e.g. v1.x) will not be able to import it. + * Exporting to v1.x specific XML format is possible using flag + * ::HWLOC_TOPOLOGY_EXPORT_XML_FLAG_V1 but it may miss some details + * about the topology. + * If there is any chance that the exported buffer may ever be imported + * back by a process using hwloc 1.x, one should consider detecting + * it at runtime and using the corresponding export format. + * + * The returned buffer ends with a \0 that is included in the returned + * length. + * + * \p flags is a OR'ed set of ::hwloc_topology_export_xml_flags_e. + * + * \return -1 if a failure occured. + * + * \note See also hwloc_topology_set_userdata_export_callback() + * for exporting application-specific object userdata. + * + * \note The topology-specific userdata pointer is ignored when exporting to XML. + * + * \note Only printable characters may be exported to XML string attributes. + * Any other character, especially any non-ASCII character, will be silently + * dropped. + */ +HWLOC_DECLSPEC int hwloc_topology_export_xmlbuffer(hwloc_topology_t topology, char **xmlbuffer, int *buflen, unsigned long flags); + +/** \brief Free a buffer allocated by hwloc_topology_export_xmlbuffer() */ +HWLOC_DECLSPEC void hwloc_free_xmlbuffer(hwloc_topology_t topology, char *xmlbuffer); + +/** \brief Set the application-specific callback for exporting object userdata + * + * The object userdata pointer is not exported to XML by default because hwloc + * does not know what it contains. + * + * This function lets applications set \p export_cb to a callback function + * that converts this opaque userdata into an exportable string. + * + * \p export_cb is invoked during XML export for each object whose + * \p userdata pointer is not \c NULL. + * The callback should use hwloc_export_obj_userdata() or + * hwloc_export_obj_userdata_base64() to actually export + * something to XML (possibly multiple times per object). + * + * \p export_cb may be set to \c NULL if userdata should not be exported to XML. + * + * \note The topology-specific userdata pointer is ignored when exporting to XML. + */ +HWLOC_DECLSPEC void hwloc_topology_set_userdata_export_callback(hwloc_topology_t topology, + void (*export_cb)(void *reserved, hwloc_topology_t topology, hwloc_obj_t obj)); + +/** \brief Export some object userdata to XML + * + * This function may only be called from within the export() callback passed + * to hwloc_topology_set_userdata_export_callback(). + * It may be invoked one of multiple times to export some userdata to XML. + * The \p buffer content of length \p length is stored with optional name + * \p name. + * + * When importing this XML file, the import() callback (if set) will be + * called exactly as many times as hwloc_export_obj_userdata() was called + * during export(). It will receive the corresponding \p name, \p buffer + * and \p length arguments. + * + * \p reserved, \p topology and \p obj must be the first three parameters + * that were given to the export callback. + * + * Only printable characters may be exported to XML string attributes. + * If a non-printable character is passed in \p name or \p buffer, + * the function returns -1 with errno set to EINVAL. + * + * If exporting binary data, the application should first encode into + * printable characters only (or use hwloc_export_obj_userdata_base64()). + * It should also take care of portability issues if the export may + * be reimported on a different architecture. + */ +HWLOC_DECLSPEC int hwloc_export_obj_userdata(void *reserved, hwloc_topology_t topology, hwloc_obj_t obj, const char *name, const void *buffer, size_t length); + +/** \brief Encode and export some object userdata to XML + * + * This function is similar to hwloc_export_obj_userdata() but it encodes + * the input buffer into printable characters before exporting. + * On import, decoding is automatically performed before the data is given + * to the import() callback if any. + * + * This function may only be called from within the export() callback passed + * to hwloc_topology_set_userdata_export_callback(). + * + * The function does not take care of portability issues if the export + * may be reimported on a different architecture. + */ +HWLOC_DECLSPEC int hwloc_export_obj_userdata_base64(void *reserved, hwloc_topology_t topology, hwloc_obj_t obj, const char *name, const void *buffer, size_t length); + +/** \brief Set the application-specific callback for importing userdata + * + * On XML import, userdata is ignored by default because hwloc does not know + * how to store it in memory. + * + * This function lets applications set \p import_cb to a callback function + * that will get the XML-stored userdata and store it in the object as expected + * by the application. + * + * \p import_cb is called during hwloc_topology_load() as many times as + * hwloc_export_obj_userdata() was called during export. The topology + * is not entirely setup yet. Object attributes are ready to consult, + * but links between objects are not. + * + * \p import_cb may be \c NULL if userdata should be ignored during import. + * + * \note \p buffer contains \p length characters followed by a null byte ('\0'). + * + * \note This function should be called before hwloc_topology_load(). + * + * \note The topology-specific userdata pointer is ignored when importing from XML. + */ +HWLOC_DECLSPEC void hwloc_topology_set_userdata_import_callback(hwloc_topology_t topology, + void (*import_cb)(hwloc_topology_t topology, hwloc_obj_t obj, const char *name, const void *buffer, size_t length)); + +/** @} */ + + +/** \defgroup hwlocality_syntheticexport Exporting Topologies to Synthetic + * @{ + */ + +/** \brief Flags for exporting synthetic topologies. + * + * Flags to be given as a OR'ed set to hwloc_topology_export_synthetic(). + */ +enum hwloc_topology_export_synthetic_flags_e { + /** \brief Export extended types such as L2dcache as basic types such as Cache. + * + * This is required if loading the synthetic description with hwloc < 1.9. + * \hideinitializer + */ + HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_NO_EXTENDED_TYPES = (1UL<<0), + + /** \brief Do not export level attributes. + * + * Ignore level attributes such as memory/cache sizes or PU indexes. + * This is required if loading the synthetic description with hwloc < 1.10. + * \hideinitializer + */ + HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_NO_ATTRS = (1UL<<1), + + /** \brief Export the memory hierarchy as expected in hwloc 1.x. + * + * Instead of attaching memory children to levels, export single NUMA node child + * as normal intermediate levels, when possible. + * This is required if loading the synthetic description with hwloc 1.x. + * However this may fail if some objects have multiple local NUMA nodes. + * \hideinitializer + */ + HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_V1 = (1UL<<2), + + /** \brief Do not export memory information. + * + * Only export the actual hierarchy of normal CPU-side objects and ignore + * where memory is attached. + * This is useful for when the hierarchy of CPUs is what really matters, + * but it behaves as if there was a single machine-wide NUMA node. + * \hideinitializer + */ + HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_IGNORE_MEMORY = (1UL<<3) +}; + +/** \brief Export the topology as a synthetic string. + * + * At most \p buflen characters will be written in \p buffer, + * including the terminating \0. + * + * This exported string may be given back to hwloc_topology_set_synthetic(). + * + * \p flags is a OR'ed set of ::hwloc_topology_export_synthetic_flags_e. + * + * \return The number of characters that were written, + * not including the terminating \0. + * + * \return -1 if the topology could not be exported, + * for instance if it is not symmetric. + * + * \note I/O and Misc children are ignored, the synthetic string only + * describes normal children. + * + * \note A 1024-byte buffer should be large enough for exporting + * topologies in the vast majority of cases. + */ + HWLOC_DECLSPEC int hwloc_topology_export_synthetic(hwloc_topology_t topology, char *buffer, size_t buflen, unsigned long flags); + +/** @} */ + + + +#ifdef __cplusplus +} /* extern "C" */ +#endif + + +#endif /* HWLOC_EXPORT_H */ diff --git a/src/3rdparty/hwloc/include/hwloc/gl.h b/src/3rdparty/hwloc/include/hwloc/gl.h new file mode 100644 index 000000000..3e643fa9a --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc/gl.h @@ -0,0 +1,135 @@ +/* + * Copyright © 2012 Blue Brain Project, EPFL. All rights reserved. + * Copyright © 2012-2013 Inria. All rights reserved. + * See COPYING in top-level directory. + */ + +/** \file + * \brief Macros to help interaction between hwloc and OpenGL displays. + * + * Applications that use both hwloc and OpenGL may want to include + * this file so as to get topology information for OpenGL displays. + */ + +#ifndef HWLOC_GL_H +#define HWLOC_GL_H + +#include + +#include +#include + + +#ifdef __cplusplus +extern "C" { +#endif + + +/** \defgroup hwlocality_gl Interoperability with OpenGL displays + * + * This interface offers ways to retrieve topology information about + * OpenGL displays. + * + * Only the NVIDIA display locality information is currently available, + * using the NV-CONTROL X11 extension and the NVCtrl library. + * + * @{ + */ + +/** \brief Get the hwloc OS device object corresponding to the + * OpenGL display given by port and device index. + * + * Return the OS device object describing the OpenGL display + * whose port (server) is \p port and device (screen) is \p device. + * Return NULL if there is none. + * + * The topology \p topology does not necessarily have to match the current + * machine. For instance the topology may be an XML import of a remote host. + * I/O devices detection and the GL component must be enabled in the topology. + * + * \note The corresponding PCI device object can be obtained by looking + * at the OS device parent object (unless PCI devices are filtered out). + */ +static __hwloc_inline hwloc_obj_t +hwloc_gl_get_display_osdev_by_port_device(hwloc_topology_t topology, + unsigned port, unsigned device) +{ + unsigned x = (unsigned) -1, y = (unsigned) -1; + hwloc_obj_t osdev = NULL; + while ((osdev = hwloc_get_next_osdev(topology, osdev)) != NULL) { + if (HWLOC_OBJ_OSDEV_GPU == osdev->attr->osdev.type + && osdev->name + && sscanf(osdev->name, ":%u.%u", &x, &y) == 2 + && port == x && device == y) + return osdev; + } + errno = EINVAL; + return NULL; +} + +/** \brief Get the hwloc OS device object corresponding to the + * OpenGL display given by name. + * + * Return the OS device object describing the OpenGL display + * whose name is \p name, built as ":port.device" such as ":0.0" . + * Return NULL if there is none. + * + * The topology \p topology does not necessarily have to match the current + * machine. For instance the topology may be an XML import of a remote host. + * I/O devices detection and the GL component must be enabled in the topology. + * + * \note The corresponding PCI device object can be obtained by looking + * at the OS device parent object (unless PCI devices are filtered out). + */ +static __hwloc_inline hwloc_obj_t +hwloc_gl_get_display_osdev_by_name(hwloc_topology_t topology, + const char *name) +{ + hwloc_obj_t osdev = NULL; + while ((osdev = hwloc_get_next_osdev(topology, osdev)) != NULL) { + if (HWLOC_OBJ_OSDEV_GPU == osdev->attr->osdev.type + && osdev->name + && !strcmp(name, osdev->name)) + return osdev; + } + errno = EINVAL; + return NULL; +} + +/** \brief Get the OpenGL display port and device corresponding + * to the given hwloc OS object. + * + * Return the OpenGL display port (server) in \p port and device (screen) + * in \p screen that correspond to the given hwloc OS device object. + * Return \c -1 if there is none. + * + * The topology \p topology does not necessarily have to match the current + * machine. For instance the topology may be an XML import of a remote host. + * I/O devices detection and the GL component must be enabled in the topology. + */ +static __hwloc_inline int +hwloc_gl_get_display_by_osdev(hwloc_topology_t topology __hwloc_attribute_unused, + hwloc_obj_t osdev, + unsigned *port, unsigned *device) +{ + unsigned x = -1, y = -1; + if (HWLOC_OBJ_OSDEV_GPU == osdev->attr->osdev.type + && sscanf(osdev->name, ":%u.%u", &x, &y) == 2) { + *port = x; + *device = y; + return 0; + } + errno = EINVAL; + return -1; +} + +/** @} */ + + +#ifdef __cplusplus +} /* extern "C" */ +#endif + + +#endif /* HWLOC_GL_H */ + diff --git a/src/3rdparty/hwloc/include/hwloc/glibc-sched.h b/src/3rdparty/hwloc/include/hwloc/glibc-sched.h new file mode 100644 index 000000000..1f9ba7cdd --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc/glibc-sched.h @@ -0,0 +1,125 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2013 inria. All rights reserved. + * Copyright © 2009-2011 Université Bordeaux + * Copyright © 2011 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +/** \file + * \brief Macros to help interaction between hwloc and glibc scheduling routines. + * + * Applications that use both hwloc and glibc scheduling routines such as + * sched_getaffinity() or pthread_attr_setaffinity_np() may want to include + * this file so as to ease conversion between their respective types. + */ + +#ifndef HWLOC_GLIBC_SCHED_H +#define HWLOC_GLIBC_SCHED_H + +#include +#include +#include + +#if !defined _GNU_SOURCE || !defined _SCHED_H || (!defined CPU_SETSIZE && !defined sched_priority) +#error Please make sure to include sched.h before including glibc-sched.h, and define _GNU_SOURCE before any inclusion of sched.h +#endif + + +#ifdef __cplusplus +extern "C" { +#endif + + +#ifdef HWLOC_HAVE_CPU_SET + + +/** \defgroup hwlocality_glibc_sched Interoperability with glibc sched affinity + * + * This interface offers ways to convert between hwloc cpusets and glibc cpusets + * such as those manipulated by sched_getaffinity() or pthread_attr_setaffinity_np(). + * + * \note Topology \p topology must match the current machine. + * + * @{ + */ + + +/** \brief Convert hwloc CPU set \p toposet into glibc sched affinity CPU set \p schedset + * + * This function may be used before calling sched_setaffinity or any other function + * that takes a cpu_set_t as input parameter. + * + * \p schedsetsize should be sizeof(cpu_set_t) unless \p schedset was dynamically allocated with CPU_ALLOC + */ +static __hwloc_inline int +hwloc_cpuset_to_glibc_sched_affinity(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_const_cpuset_t hwlocset, + cpu_set_t *schedset, size_t schedsetsize) +{ +#ifdef CPU_ZERO_S + unsigned cpu; + CPU_ZERO_S(schedsetsize, schedset); + hwloc_bitmap_foreach_begin(cpu, hwlocset) + CPU_SET_S(cpu, schedsetsize, schedset); + hwloc_bitmap_foreach_end(); +#else /* !CPU_ZERO_S */ + unsigned cpu; + CPU_ZERO(schedset); + assert(schedsetsize == sizeof(cpu_set_t)); + hwloc_bitmap_foreach_begin(cpu, hwlocset) + CPU_SET(cpu, schedset); + hwloc_bitmap_foreach_end(); +#endif /* !CPU_ZERO_S */ + return 0; +} + +/** \brief Convert glibc sched affinity CPU set \p schedset into hwloc CPU set + * + * This function may be used before calling sched_setaffinity or any other function + * that takes a cpu_set_t as input parameter. + * + * \p schedsetsize should be sizeof(cpu_set_t) unless \p schedset was dynamically allocated with CPU_ALLOC + */ +static __hwloc_inline int +hwloc_cpuset_from_glibc_sched_affinity(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_cpuset_t hwlocset, + const cpu_set_t *schedset, size_t schedsetsize) +{ + int cpu; +#ifdef CPU_ZERO_S + int count; +#endif + hwloc_bitmap_zero(hwlocset); +#ifdef CPU_ZERO_S + count = CPU_COUNT_S(schedsetsize, schedset); + cpu = 0; + while (count) { + if (CPU_ISSET_S(cpu, schedsetsize, schedset)) { + hwloc_bitmap_set(hwlocset, cpu); + count--; + } + cpu++; + } +#else /* !CPU_ZERO_S */ + /* sched.h does not support dynamic cpu_set_t (introduced in glibc 2.7), + * assume we have a very old interface without CPU_COUNT (added in 2.6) + */ + assert(schedsetsize == sizeof(cpu_set_t)); + for(cpu=0; cpu +#include + + +#ifdef __cplusplus +extern "C" { +#endif + + +/** \defgroup hwlocality_helper_find_inside Finding Objects inside a CPU set + * @{ + */ + +/** \brief Get the first largest object included in the given cpuset \p set. + * + * \return the first object that is included in \p set and whose parent is not. + * + * This is convenient for iterating over all largest objects within a CPU set + * by doing a loop getting the first largest object and clearing its CPU set + * from the remaining CPU set. + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_first_largest_obj_inside_cpuset(hwloc_topology_t topology, hwloc_const_cpuset_t set) +{ + hwloc_obj_t obj = hwloc_get_root_obj(topology); + if (!hwloc_bitmap_intersects(obj->cpuset, set)) + return NULL; + while (!hwloc_bitmap_isincluded(obj->cpuset, set)) { + /* while the object intersects without being included, look at its children */ + hwloc_obj_t child = obj->first_child; + while (child) { + if (hwloc_bitmap_intersects(child->cpuset, set)) + break; + child = child->next_sibling; + } + if (!child) + /* no child intersects, return their father */ + return obj; + /* found one intersecting child, look at its children */ + obj = child; + } + /* obj is included, return it */ + return obj; +} + +/** \brief Get the set of largest objects covering exactly a given cpuset \p set + * + * \return the number of objects returned in \p objs. + */ +HWLOC_DECLSPEC int hwloc_get_largest_objs_inside_cpuset (hwloc_topology_t topology, hwloc_const_cpuset_t set, + hwloc_obj_t * __hwloc_restrict objs, int max); + +/** \brief Return the next object at depth \p depth included in CPU set \p set. + * + * If \p prev is \c NULL, return the first object at depth \p depth + * included in \p set. The next invokation should pass the previous + * return value in \p prev so as to obtain the next object in \p set. + * + * \note Objects with empty CPU sets are ignored + * (otherwise they would be considered included in any given set). + * + * \note This function cannot work if objects at the given depth do + * not have CPU sets (I/O or Misc objects). + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_next_obj_inside_cpuset_by_depth (hwloc_topology_t topology, hwloc_const_cpuset_t set, + int depth, hwloc_obj_t prev) +{ + hwloc_obj_t next = hwloc_get_next_obj_by_depth(topology, depth, prev); + if (!next) + return NULL; + while (next && (hwloc_bitmap_iszero(next->cpuset) || !hwloc_bitmap_isincluded(next->cpuset, set))) + next = next->next_cousin; + return next; +} + +/** \brief Return the next object of type \p type included in CPU set \p set. + * + * If there are multiple or no depth for given type, return \c NULL + * and let the caller fallback to + * hwloc_get_next_obj_inside_cpuset_by_depth(). + * + * \note Objects with empty CPU sets are ignored + * (otherwise they would be considered included in any given set). + * + * \note This function cannot work if objects of the given type do + * not have CPU sets (I/O or Misc objects). + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_next_obj_inside_cpuset_by_type (hwloc_topology_t topology, hwloc_const_cpuset_t set, + hwloc_obj_type_t type, hwloc_obj_t prev) +{ + int depth = hwloc_get_type_depth(topology, type); + if (depth == HWLOC_TYPE_DEPTH_UNKNOWN || depth == HWLOC_TYPE_DEPTH_MULTIPLE) + return NULL; + return hwloc_get_next_obj_inside_cpuset_by_depth(topology, set, depth, prev); +} + +/** \brief Return the (logically) \p idx -th object at depth \p depth included in CPU set \p set. + * + * \note Objects with empty CPU sets are ignored + * (otherwise they would be considered included in any given set). + * + * \note This function cannot work if objects at the given depth do + * not have CPU sets (I/O or Misc objects). + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_obj_inside_cpuset_by_depth (hwloc_topology_t topology, hwloc_const_cpuset_t set, + int depth, unsigned idx) __hwloc_attribute_pure; +static __hwloc_inline hwloc_obj_t +hwloc_get_obj_inside_cpuset_by_depth (hwloc_topology_t topology, hwloc_const_cpuset_t set, + int depth, unsigned idx) +{ + hwloc_obj_t obj = hwloc_get_obj_by_depth (topology, depth, 0); + unsigned count = 0; + if (!obj) + return NULL; + while (obj) { + if (!hwloc_bitmap_iszero(obj->cpuset) && hwloc_bitmap_isincluded(obj->cpuset, set)) { + if (count == idx) + return obj; + count++; + } + obj = obj->next_cousin; + } + return NULL; +} + +/** \brief Return the \p idx -th object of type \p type included in CPU set \p set. + * + * If there are multiple or no depth for given type, return \c NULL + * and let the caller fallback to + * hwloc_get_obj_inside_cpuset_by_depth(). + * + * \note Objects with empty CPU sets are ignored + * (otherwise they would be considered included in any given set). + * + * \note This function cannot work if objects of the given type do + * not have CPU sets (I/O or Misc objects). + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_obj_inside_cpuset_by_type (hwloc_topology_t topology, hwloc_const_cpuset_t set, + hwloc_obj_type_t type, unsigned idx) __hwloc_attribute_pure; +static __hwloc_inline hwloc_obj_t +hwloc_get_obj_inside_cpuset_by_type (hwloc_topology_t topology, hwloc_const_cpuset_t set, + hwloc_obj_type_t type, unsigned idx) +{ + int depth = hwloc_get_type_depth(topology, type); + if (depth == HWLOC_TYPE_DEPTH_UNKNOWN || depth == HWLOC_TYPE_DEPTH_MULTIPLE) + return NULL; + return hwloc_get_obj_inside_cpuset_by_depth(topology, set, depth, idx); +} + +/** \brief Return the number of objects at depth \p depth included in CPU set \p set. + * + * \note Objects with empty CPU sets are ignored + * (otherwise they would be considered included in any given set). + * + * \note This function cannot work if objects at the given depth do + * not have CPU sets (I/O or Misc objects). + */ +static __hwloc_inline unsigned +hwloc_get_nbobjs_inside_cpuset_by_depth (hwloc_topology_t topology, hwloc_const_cpuset_t set, + int depth) __hwloc_attribute_pure; +static __hwloc_inline unsigned +hwloc_get_nbobjs_inside_cpuset_by_depth (hwloc_topology_t topology, hwloc_const_cpuset_t set, + int depth) +{ + hwloc_obj_t obj = hwloc_get_obj_by_depth (topology, depth, 0); + unsigned count = 0; + if (!obj) + return 0; + while (obj) { + if (!hwloc_bitmap_iszero(obj->cpuset) && hwloc_bitmap_isincluded(obj->cpuset, set)) + count++; + obj = obj->next_cousin; + } + return count; +} + +/** \brief Return the number of objects of type \p type included in CPU set \p set. + * + * If no object for that type exists inside CPU set \p set, 0 is + * returned. If there are several levels with objects of that type + * inside CPU set \p set, -1 is returned. + * + * \note Objects with empty CPU sets are ignored + * (otherwise they would be considered included in any given set). + * + * \note This function cannot work if objects of the given type do + * not have CPU sets (I/O objects). + */ +static __hwloc_inline int +hwloc_get_nbobjs_inside_cpuset_by_type (hwloc_topology_t topology, hwloc_const_cpuset_t set, + hwloc_obj_type_t type) __hwloc_attribute_pure; +static __hwloc_inline int +hwloc_get_nbobjs_inside_cpuset_by_type (hwloc_topology_t topology, hwloc_const_cpuset_t set, + hwloc_obj_type_t type) +{ + int depth = hwloc_get_type_depth(topology, type); + if (depth == HWLOC_TYPE_DEPTH_UNKNOWN) + return 0; + if (depth == HWLOC_TYPE_DEPTH_MULTIPLE) + return -1; /* FIXME: agregate nbobjs from different levels? */ + return (int) hwloc_get_nbobjs_inside_cpuset_by_depth(topology, set, depth); +} + +/** \brief Return the logical index among the objects included in CPU set \p set. + * + * Consult all objects in the same level as \p obj and inside CPU set \p set + * in the logical order, and return the index of \p obj within them. + * If \p set covers the entire topology, this is the logical index of \p obj. + * Otherwise, this is similar to a logical index within the part of the topology + * defined by CPU set \p set. + * + * \note Objects with empty CPU sets are ignored + * (otherwise they would be considered included in any given set). + * + * \note This function cannot work if obj does not have CPU sets (I/O objects). + */ +static __hwloc_inline int +hwloc_get_obj_index_inside_cpuset (hwloc_topology_t topology __hwloc_attribute_unused, hwloc_const_cpuset_t set, + hwloc_obj_t obj) __hwloc_attribute_pure; +static __hwloc_inline int +hwloc_get_obj_index_inside_cpuset (hwloc_topology_t topology __hwloc_attribute_unused, hwloc_const_cpuset_t set, + hwloc_obj_t obj) +{ + int idx = 0; + if (!hwloc_bitmap_isincluded(obj->cpuset, set)) + return -1; + /* count how many objects are inside the cpuset on the way from us to the beginning of the level */ + while ((obj = obj->prev_cousin) != NULL) + if (!hwloc_bitmap_iszero(obj->cpuset) && hwloc_bitmap_isincluded(obj->cpuset, set)) + idx++; + return idx; +} + +/** @} */ + + + +/** \defgroup hwlocality_helper_find_covering Finding Objects covering at least CPU set + * @{ + */ + +/** \brief Get the child covering at least CPU set \p set. + * + * \return \c NULL if no child matches or if \p set is empty. + * + * \note This function cannot work if parent does not have a CPU set (I/O or Misc objects). + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_child_covering_cpuset (hwloc_topology_t topology __hwloc_attribute_unused, hwloc_const_cpuset_t set, + hwloc_obj_t parent) __hwloc_attribute_pure; +static __hwloc_inline hwloc_obj_t +hwloc_get_child_covering_cpuset (hwloc_topology_t topology __hwloc_attribute_unused, hwloc_const_cpuset_t set, + hwloc_obj_t parent) +{ + hwloc_obj_t child; + if (hwloc_bitmap_iszero(set)) + return NULL; + child = parent->first_child; + while (child) { + if (child->cpuset && hwloc_bitmap_isincluded(set, child->cpuset)) + return child; + child = child->next_sibling; + } + return NULL; +} + +/** \brief Get the lowest object covering at least CPU set \p set + * + * \return \c NULL if no object matches or if \p set is empty. + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_obj_covering_cpuset (hwloc_topology_t topology, hwloc_const_cpuset_t set) __hwloc_attribute_pure; +static __hwloc_inline hwloc_obj_t +hwloc_get_obj_covering_cpuset (hwloc_topology_t topology, hwloc_const_cpuset_t set) +{ + struct hwloc_obj *current = hwloc_get_root_obj(topology); + if (hwloc_bitmap_iszero(set) || !hwloc_bitmap_isincluded(set, current->cpuset)) + return NULL; + while (1) { + hwloc_obj_t child = hwloc_get_child_covering_cpuset(topology, set, current); + if (!child) + return current; + current = child; + } +} + +/** \brief Iterate through same-depth objects covering at least CPU set \p set + * + * If object \p prev is \c NULL, return the first object at depth \p + * depth covering at least part of CPU set \p set. The next + * invokation should pass the previous return value in \p prev so as + * to obtain the next object covering at least another part of \p set. + * + * \note This function cannot work if objects at the given depth do + * not have CPU sets (I/O or Misc objects). + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_next_obj_covering_cpuset_by_depth(hwloc_topology_t topology, hwloc_const_cpuset_t set, + int depth, hwloc_obj_t prev) +{ + hwloc_obj_t next = hwloc_get_next_obj_by_depth(topology, depth, prev); + if (!next) + return NULL; + while (next && !hwloc_bitmap_intersects(set, next->cpuset)) + next = next->next_cousin; + return next; +} + +/** \brief Iterate through same-type objects covering at least CPU set \p set + * + * If object \p prev is \c NULL, return the first object of type \p + * type covering at least part of CPU set \p set. The next invokation + * should pass the previous return value in \p prev so as to obtain + * the next object of type \p type covering at least another part of + * \p set. + * + * If there are no or multiple depths for type \p type, \c NULL is returned. + * The caller may fallback to hwloc_get_next_obj_covering_cpuset_by_depth() + * for each depth. + * + * \note This function cannot work if objects of the given type do + * not have CPU sets (I/O or Misc objects). + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_next_obj_covering_cpuset_by_type(hwloc_topology_t topology, hwloc_const_cpuset_t set, + hwloc_obj_type_t type, hwloc_obj_t prev) +{ + int depth = hwloc_get_type_depth(topology, type); + if (depth == HWLOC_TYPE_DEPTH_UNKNOWN || depth == HWLOC_TYPE_DEPTH_MULTIPLE) + return NULL; + return hwloc_get_next_obj_covering_cpuset_by_depth(topology, set, depth, prev); +} + +/** @} */ + + + +/** \defgroup hwlocality_helper_ancestors Looking at Ancestor and Child Objects + * @{ + * + * Be sure to see the figure in \ref termsanddefs that shows a + * complete topology tree, including depths, child/sibling/cousin + * relationships, and an example of an asymmetric topology where one + * package has fewer caches than its peers. + */ + +/** \brief Returns the ancestor object of \p obj at depth \p depth. + * + * \note \p depth should not be the depth of PU or NUMA objects + * since they are ancestors of no objects (except Misc or I/O). + * This function rather expects an intermediate level depth, + * such as the depth of Packages, Cores, or Caches. + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_ancestor_obj_by_depth (hwloc_topology_t topology __hwloc_attribute_unused, int depth, hwloc_obj_t obj) __hwloc_attribute_pure; +static __hwloc_inline hwloc_obj_t +hwloc_get_ancestor_obj_by_depth (hwloc_topology_t topology __hwloc_attribute_unused, int depth, hwloc_obj_t obj) +{ + hwloc_obj_t ancestor = obj; + if (obj->depth < depth) + return NULL; + while (ancestor && ancestor->depth > depth) + ancestor = ancestor->parent; + return ancestor; +} + +/** \brief Returns the ancestor object of \p obj with type \p type. + * + * \note \p type should not be ::HWLOC_OBJ_PU or ::HWLOC_OBJ_NUMANODE + * since these objects are ancestors of no objects (except Misc or I/O). + * This function rather expects an intermediate object type, + * such as ::HWLOC_OBJ_PACKAGE, ::HWLOC_OBJ_CORE, etc. + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_ancestor_obj_by_type (hwloc_topology_t topology __hwloc_attribute_unused, hwloc_obj_type_t type, hwloc_obj_t obj) __hwloc_attribute_pure; +static __hwloc_inline hwloc_obj_t +hwloc_get_ancestor_obj_by_type (hwloc_topology_t topology __hwloc_attribute_unused, hwloc_obj_type_t type, hwloc_obj_t obj) +{ + hwloc_obj_t ancestor = obj->parent; + while (ancestor && ancestor->type != type) + ancestor = ancestor->parent; + return ancestor; +} + +/** \brief Returns the common parent object to objects \p obj1 and \p obj2 */ +static __hwloc_inline hwloc_obj_t +hwloc_get_common_ancestor_obj (hwloc_topology_t topology __hwloc_attribute_unused, hwloc_obj_t obj1, hwloc_obj_t obj2) __hwloc_attribute_pure; +static __hwloc_inline hwloc_obj_t +hwloc_get_common_ancestor_obj (hwloc_topology_t topology __hwloc_attribute_unused, hwloc_obj_t obj1, hwloc_obj_t obj2) +{ + /* the loop isn't so easy since intermediate ancestors may have + * different depth, causing us to alternate between using obj1->parent + * and obj2->parent. Also, even if at some point we find ancestors of + * of the same depth, their ancestors may have different depth again. + */ + while (obj1 != obj2) { + while (obj1->depth > obj2->depth) + obj1 = obj1->parent; + while (obj2->depth > obj1->depth) + obj2 = obj2->parent; + if (obj1 != obj2 && obj1->depth == obj2->depth) { + obj1 = obj1->parent; + obj2 = obj2->parent; + } + } + return obj1; +} + +/** \brief Returns true if \p obj is inside the subtree beginning with ancestor object \p subtree_root. + * + * \note This function cannot work if \p obj and \p subtree_root objects do + * not have CPU sets (I/O or Misc objects). + */ +static __hwloc_inline int +hwloc_obj_is_in_subtree (hwloc_topology_t topology __hwloc_attribute_unused, hwloc_obj_t obj, hwloc_obj_t subtree_root) __hwloc_attribute_pure; +static __hwloc_inline int +hwloc_obj_is_in_subtree (hwloc_topology_t topology __hwloc_attribute_unused, hwloc_obj_t obj, hwloc_obj_t subtree_root) +{ + return obj->cpuset && subtree_root->cpuset && hwloc_bitmap_isincluded(obj->cpuset, subtree_root->cpuset); +} + +/** \brief Return the next child. + * + * Return the next child among the normal children list, + * then among the memory children list, then among the I/O + * children list, then among the Misc children list. + * + * If \p prev is \c NULL, return the first child. + * + * Return \c NULL when there is no next child. + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_next_child (hwloc_topology_t topology __hwloc_attribute_unused, hwloc_obj_t parent, hwloc_obj_t prev) +{ + hwloc_obj_t obj; + int state = 0; + if (prev) { + if (prev->type == HWLOC_OBJ_MISC) + state = 3; + else if (prev->type == HWLOC_OBJ_BRIDGE || prev->type == HWLOC_OBJ_PCI_DEVICE || prev->type == HWLOC_OBJ_OS_DEVICE) + state = 2; + else if (prev->type == HWLOC_OBJ_NUMANODE) + state = 1; + obj = prev->next_sibling; + } else { + obj = parent->first_child; + } + if (!obj && state == 0) { + obj = parent->memory_first_child; + state = 1; + } + if (!obj && state == 1) { + obj = parent->io_first_child; + state = 2; + } + if (!obj && state == 2) { + obj = parent->misc_first_child; + state = 3; + } + return obj; +} + +/** @} */ + + + +/** \defgroup hwlocality_helper_types Kinds of object Type + * @{ + * + * Each object type is + * either Normal (i.e. hwloc_obj_type_is_normal() returns 1), + * or Memory (i.e. hwloc_obj_type_is_memory() returns 1) + * or I/O (i.e. hwloc_obj_type_is_io() returns 1) + * or Misc (i.e. equal to ::HWLOC_OBJ_MISC). + * It cannot be of more than one of these kinds. + */ + +/** \brief Check whether an object type is Normal. + * + * Normal objects are objects of the main CPU hierarchy + * (Machine, Package, Core, PU, CPU caches, etc.), + * but they are not NUMA nodes, I/O devices or Misc objects. + * + * They are attached to parent as Normal children, + * not as Memory, I/O or Misc children. + * + * \return 1 if an object of type \p type is a Normal object, 0 otherwise. + */ +HWLOC_DECLSPEC int +hwloc_obj_type_is_normal(hwloc_obj_type_t type); + +/** \brief Check whether an object type is I/O. + * + * I/O objects are objects attached to their parents + * in the I/O children list. + * This current includes Bridges, PCI and OS devices. + * + * \return 1 if an object of type \p type is a I/O object, 0 otherwise. + */ +HWLOC_DECLSPEC int +hwloc_obj_type_is_io(hwloc_obj_type_t type); + +/** \brief Check whether an object type is Memory. + * + * Memory objects are objects attached to their parents + * in the Memory children list. + * This current only includes NUMA nodes. + * + * \return 1 if an object of type \p type is a Memory object, 0 otherwise. + */ +HWLOC_DECLSPEC int +hwloc_obj_type_is_memory(hwloc_obj_type_t type); + +/** \brief Check whether an object type is a Cache (Data, Unified or Instruction). + * + * \return 1 if an object of type \p type is a Cache, 0 otherwise. + */ +HWLOC_DECLSPEC int +hwloc_obj_type_is_cache(hwloc_obj_type_t type); + +/** \brief Check whether an object type is a Data or Unified Cache. + * + * \return 1 if an object of type \p type is a Data or Unified Cache, 0 otherwise. + */ +HWLOC_DECLSPEC int +hwloc_obj_type_is_dcache(hwloc_obj_type_t type); + +/** \brief Check whether an object type is a Instruction Cache, + * + * \return 1 if an object of type \p type is a Instruction Cache, 0 otherwise. + */ +HWLOC_DECLSPEC int +hwloc_obj_type_is_icache(hwloc_obj_type_t type); + +/** @} */ + + + +/** \defgroup hwlocality_helper_find_cache Looking at Cache Objects + * @{ + */ + +/** \brief Find the depth of cache objects matching cache level and type. + * + * Return the depth of the topology level that contains cache objects + * whose attributes match \p cachelevel and \p cachetype. + + * This function is identical to calling hwloc_get_type_depth() with the + * corresponding type such as ::HWLOC_OBJ_L1ICACHE, except that it may + * also return a Unified cache when looking for an instruction cache. + * + * If no cache level matches, ::HWLOC_TYPE_DEPTH_UNKNOWN is returned. + * + * If \p cachetype is ::HWLOC_OBJ_CACHE_UNIFIED, the depth of the + * unique matching unified cache level is returned. + * + * If \p cachetype is ::HWLOC_OBJ_CACHE_DATA or ::HWLOC_OBJ_CACHE_INSTRUCTION, + * either a matching cache, or a unified cache is returned. + * + * If \p cachetype is \c -1, it is ignored and multiple levels may + * match. The function returns either the depth of a uniquely matching + * level or ::HWLOC_TYPE_DEPTH_MULTIPLE. + */ +static __hwloc_inline int +hwloc_get_cache_type_depth (hwloc_topology_t topology, + unsigned cachelevel, hwloc_obj_cache_type_t cachetype) +{ + int depth; + int found = HWLOC_TYPE_DEPTH_UNKNOWN; + for (depth=0; ; depth++) { + hwloc_obj_t obj = hwloc_get_obj_by_depth(topology, depth, 0); + if (!obj) + break; + if (!hwloc_obj_type_is_dcache(obj->type) || obj->attr->cache.depth != cachelevel) + /* doesn't match, try next depth */ + continue; + if (cachetype == (hwloc_obj_cache_type_t) -1) { + if (found != HWLOC_TYPE_DEPTH_UNKNOWN) { + /* second match, return MULTIPLE */ + return HWLOC_TYPE_DEPTH_MULTIPLE; + } + /* first match, mark it as found */ + found = depth; + continue; + } + if (obj->attr->cache.type == cachetype || obj->attr->cache.type == HWLOC_OBJ_CACHE_UNIFIED) + /* exact match (either unified is alone, or we match instruction or data), return immediately */ + return depth; + } + /* went to the bottom, return what we found */ + return found; +} + +/** \brief Get the first data (or unified) cache covering a cpuset \p set + * + * \return \c NULL if no cache matches. + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_cache_covering_cpuset (hwloc_topology_t topology, hwloc_const_cpuset_t set) __hwloc_attribute_pure; +static __hwloc_inline hwloc_obj_t +hwloc_get_cache_covering_cpuset (hwloc_topology_t topology, hwloc_const_cpuset_t set) +{ + hwloc_obj_t current = hwloc_get_obj_covering_cpuset(topology, set); + while (current) { + if (hwloc_obj_type_is_dcache(current->type)) + return current; + current = current->parent; + } + return NULL; +} + +/** \brief Get the first data (or unified) cache shared between an object and somebody else. + * + * \return \c NULL if no cache matches or if an invalid object is given. + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_shared_cache_covering_obj (hwloc_topology_t topology __hwloc_attribute_unused, hwloc_obj_t obj) __hwloc_attribute_pure; +static __hwloc_inline hwloc_obj_t +hwloc_get_shared_cache_covering_obj (hwloc_topology_t topology __hwloc_attribute_unused, hwloc_obj_t obj) +{ + hwloc_obj_t current = obj->parent; + if (!obj->cpuset) + return NULL; + while (current) { + if (!hwloc_bitmap_isequal(current->cpuset, obj->cpuset) + && hwloc_obj_type_is_dcache(current->type)) + return current; + current = current->parent; + } + return NULL; +} + +/** @} */ + + + +/** \defgroup hwlocality_helper_find_misc Finding objects, miscellaneous helpers + * @{ + * + * Be sure to see the figure in \ref termsanddefs that shows a + * complete topology tree, including depths, child/sibling/cousin + * relationships, and an example of an asymmetric topology where one + * package has fewer caches than its peers. + */ + +/** \brief Returns the object of type ::HWLOC_OBJ_PU with \p os_index. + * + * This function is useful for converting a CPU set into the PU + * objects it contains. + * When retrieving the current binding (e.g. with hwloc_get_cpubind()), + * one may iterate over the bits of the resulting CPU set with + * hwloc_bitmap_foreach_begin(), and find the corresponding PUs + * with this function. + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_pu_obj_by_os_index(hwloc_topology_t topology, unsigned os_index) __hwloc_attribute_pure; +static __hwloc_inline hwloc_obj_t +hwloc_get_pu_obj_by_os_index(hwloc_topology_t topology, unsigned os_index) +{ + hwloc_obj_t obj = NULL; + while ((obj = hwloc_get_next_obj_by_type(topology, HWLOC_OBJ_PU, obj)) != NULL) + if (obj->os_index == os_index) + return obj; + return NULL; +} + +/** \brief Returns the object of type ::HWLOC_OBJ_NUMANODE with \p os_index. + * + * This function is useful for converting a nodeset into the NUMA node + * objects it contains. + * When retrieving the current binding (e.g. with hwloc_get_membind() with HWLOC_MEMBIND_BYNODESET), + * one may iterate over the bits of the resulting nodeset with + * hwloc_bitmap_foreach_begin(), and find the corresponding NUMA nodes + * with this function. + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_numanode_obj_by_os_index(hwloc_topology_t topology, unsigned os_index) __hwloc_attribute_pure; +static __hwloc_inline hwloc_obj_t +hwloc_get_numanode_obj_by_os_index(hwloc_topology_t topology, unsigned os_index) +{ + hwloc_obj_t obj = NULL; + while ((obj = hwloc_get_next_obj_by_type(topology, HWLOC_OBJ_NUMANODE, obj)) != NULL) + if (obj->os_index == os_index) + return obj; + return NULL; +} + +/** \brief Do a depth-first traversal of the topology to find and sort + * + * all objects that are at the same depth than \p src. + * Report in \p objs up to \p max physically closest ones to \p src. + * + * \return the number of objects returned in \p objs. + * + * \return 0 if \p src is an I/O object. + * + * \note This function requires the \p src object to have a CPU set. + */ +/* TODO: rather provide an iterator? Provide a way to know how much should be allocated? By returning the total number of objects instead? */ +HWLOC_DECLSPEC unsigned hwloc_get_closest_objs (hwloc_topology_t topology, hwloc_obj_t src, hwloc_obj_t * __hwloc_restrict objs, unsigned max); + +/** \brief Find an object below another object, both specified by types and indexes. + * + * Start from the top system object and find object of type \p type1 + * and logical index \p idx1. Then look below this object and find another + * object of type \p type2 and logical index \p idx2. Indexes are specified + * within the parent, not withing the entire system. + * + * For instance, if type1 is PACKAGE, idx1 is 2, type2 is CORE and idx2 + * is 3, return the fourth core object below the third package. + * + * \note This function requires these objects to have a CPU set. + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_obj_below_by_type (hwloc_topology_t topology, + hwloc_obj_type_t type1, unsigned idx1, + hwloc_obj_type_t type2, unsigned idx2) __hwloc_attribute_pure; +static __hwloc_inline hwloc_obj_t +hwloc_get_obj_below_by_type (hwloc_topology_t topology, + hwloc_obj_type_t type1, unsigned idx1, + hwloc_obj_type_t type2, unsigned idx2) +{ + hwloc_obj_t obj; + obj = hwloc_get_obj_by_type (topology, type1, idx1); + if (!obj) + return NULL; + return hwloc_get_obj_inside_cpuset_by_type(topology, obj->cpuset, type2, idx2); +} + +/** \brief Find an object below a chain of objects specified by types and indexes. + * + * This is a generalized version of hwloc_get_obj_below_by_type(). + * + * Arrays \p typev and \p idxv must contain \p nr types and indexes. + * + * Start from the top system object and walk the arrays \p typev and \p idxv. + * For each type and logical index couple in the arrays, look under the previously found + * object to find the index-th object of the given type. + * Indexes are specified within the parent, not withing the entire system. + * + * For instance, if nr is 3, typev contains NODE, PACKAGE and CORE, + * and idxv contains 0, 1 and 2, return the third core object below + * the second package below the first NUMA node. + * + * \note This function requires all these objects and the root object + * to have a CPU set. + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_obj_below_array_by_type (hwloc_topology_t topology, int nr, hwloc_obj_type_t *typev, unsigned *idxv) __hwloc_attribute_pure; +static __hwloc_inline hwloc_obj_t +hwloc_get_obj_below_array_by_type (hwloc_topology_t topology, int nr, hwloc_obj_type_t *typev, unsigned *idxv) +{ + hwloc_obj_t obj = hwloc_get_root_obj(topology); + int i; + for(i=0; icpuset, typev[i], idxv[i]); + } + return obj; +} + +/** @} */ + + + +/** \defgroup hwlocality_helper_distribute Distributing items over a topology + * @{ + */ + +/** \brief Flags to be given to hwloc_distrib(). + */ +enum hwloc_distrib_flags_e { + /** \brief Distrib in reverse order, starting from the last objects. + * \hideinitializer + */ + HWLOC_DISTRIB_FLAG_REVERSE = (1UL<<0) +}; + +/** \brief Distribute \p n items over the topology under \p roots + * + * Array \p set will be filled with \p n cpusets recursively distributed + * linearly over the topology under objects \p roots, down to depth \p until + * (which can be INT_MAX to distribute down to the finest level). + * + * \p n_roots is usually 1 and \p roots only contains the topology root object + * so as to distribute over the entire topology. + * + * This is typically useful when an application wants to distribute \p n + * threads over a machine, giving each of them as much private cache as + * possible and keeping them locally in number order. + * + * The caller may typically want to also call hwloc_bitmap_singlify() + * before binding a thread so that it does not move at all. + * + * \p flags should be 0 or a OR'ed set of ::hwloc_distrib_flags_e. + * + * \note This function requires the \p roots objects to have a CPU set. + * + * \note This function replaces the now deprecated hwloc_distribute() + * and hwloc_distributev() functions. + */ +static __hwloc_inline int +hwloc_distrib(hwloc_topology_t topology, + hwloc_obj_t *roots, unsigned n_roots, + hwloc_cpuset_t *set, + unsigned n, + int until, unsigned long flags) +{ + unsigned i; + unsigned tot_weight; + unsigned given, givenweight; + hwloc_cpuset_t *cpusetp = set; + + if (flags & ~HWLOC_DISTRIB_FLAG_REVERSE) { + errno = EINVAL; + return -1; + } + + tot_weight = 0; + for (i = 0; i < n_roots; i++) + tot_weight += (unsigned) hwloc_bitmap_weight(roots[i]->cpuset); + + for (i = 0, given = 0, givenweight = 0; i < n_roots; i++) { + unsigned chunk, weight; + hwloc_obj_t root = roots[flags & HWLOC_DISTRIB_FLAG_REVERSE ? n_roots-1-i : i]; + hwloc_cpuset_t cpuset = root->cpuset; + if (root->type == HWLOC_OBJ_NUMANODE) + /* NUMANodes have same cpuset as their parent, but we need normal objects below */ + root = root->parent; + weight = (unsigned) hwloc_bitmap_weight(cpuset); + if (!weight) + continue; + /* Give to root a chunk proportional to its weight. + * If previous chunks got rounded-up, we may get a bit less. */ + chunk = (( (givenweight+weight) * n + tot_weight-1) / tot_weight) + - (( givenweight * n + tot_weight-1) / tot_weight); + if (!root->arity || chunk <= 1 || root->depth >= until) { + /* We can't split any more, put everything there. */ + if (chunk) { + /* Fill cpusets with ours */ + unsigned j; + for (j=0; j < chunk; j++) + cpusetp[j] = hwloc_bitmap_dup(cpuset); + } else { + /* We got no chunk, just merge our cpuset to a previous one + * (the first chunk cannot be empty) + * so that this root doesn't get ignored. + */ + assert(given); + hwloc_bitmap_or(cpusetp[-1], cpusetp[-1], cpuset); + } + } else { + /* Still more to distribute, recurse into children */ + hwloc_distrib(topology, root->children, root->arity, cpusetp, chunk, until, flags); + } + cpusetp += chunk; + given += chunk; + givenweight += weight; + } + + return 0; +} + +/** @} */ + + + +/** \defgroup hwlocality_helper_topology_sets CPU and node sets of entire topologies + * @{ + */ + +/** \brief Get complete CPU set + * + * \return the complete CPU set of logical processors of the system. + * + * \note The returned cpuset is not newly allocated and should thus not be + * changed or freed; hwloc_bitmap_dup() must be used to obtain a local copy. + * + * \note This is equivalent to retrieving the root object complete CPU-set. + */ +HWLOC_DECLSPEC hwloc_const_cpuset_t +hwloc_topology_get_complete_cpuset(hwloc_topology_t topology) __hwloc_attribute_pure; + +/** \brief Get topology CPU set + * + * \return the CPU set of logical processors of the system for which hwloc + * provides topology information. This is equivalent to the cpuset of the + * system object. + * + * \note The returned cpuset is not newly allocated and should thus not be + * changed or freed; hwloc_bitmap_dup() must be used to obtain a local copy. + * + * \note This is equivalent to retrieving the root object complete CPU-set. + */ +HWLOC_DECLSPEC hwloc_const_cpuset_t +hwloc_topology_get_topology_cpuset(hwloc_topology_t topology) __hwloc_attribute_pure; + +/** \brief Get allowed CPU set + * + * \return the CPU set of allowed logical processors of the system. + * + * \note If the topology flag ::HWLOC_TOPOLOGY_FLAG_WHOLE_SYSTEM was not set, + * this is identical to hwloc_topology_get_topology_cpuset(), which means + * all PUs are allowed. + * + * \note If ::HWLOC_TOPOLOGY_FLAG_WHOLE_SYSTEM was set, applying + * hwloc_bitmap_intersects() on the result of this function and on an object + * cpuset checks whether there are allowed PUs inside that object. + * Applying hwloc_bitmap_and() returns the list of these allowed PUs. + * + * \note The returned cpuset is not newly allocated and should thus not be + * changed or freed, hwloc_bitmap_dup() must be used to obtain a local copy. + */ +HWLOC_DECLSPEC hwloc_const_cpuset_t +hwloc_topology_get_allowed_cpuset(hwloc_topology_t topology) __hwloc_attribute_pure; + +/** \brief Get complete node set + * + * \return the complete node set of memory of the system. + * + * \note The returned nodeset is not newly allocated and should thus not be + * changed or freed; hwloc_bitmap_dup() must be used to obtain a local copy. + * + * \note This is equivalent to retrieving the root object complete CPU-set. + */ +HWLOC_DECLSPEC hwloc_const_nodeset_t +hwloc_topology_get_complete_nodeset(hwloc_topology_t topology) __hwloc_attribute_pure; + +/** \brief Get topology node set + * + * \return the node set of memory of the system for which hwloc + * provides topology information. This is equivalent to the nodeset of the + * system object. + * + * \note The returned nodeset is not newly allocated and should thus not be + * changed or freed; hwloc_bitmap_dup() must be used to obtain a local copy. + * + * \note This is equivalent to retrieving the root object complete CPU-set. + */ +HWLOC_DECLSPEC hwloc_const_nodeset_t +hwloc_topology_get_topology_nodeset(hwloc_topology_t topology) __hwloc_attribute_pure; + +/** \brief Get allowed node set + * + * \return the node set of allowed memory of the system. + * + * \note If the topology flag ::HWLOC_TOPOLOGY_FLAG_WHOLE_SYSTEM was not set, + * this is identical to hwloc_topology_get_topology_nodeset(), which means + * all NUMA nodes are allowed. + * + * \note If ::HWLOC_TOPOLOGY_FLAG_WHOLE_SYSTEM was set, applying + * hwloc_bitmap_intersects() on the result of this function and on an object + * nodeset checks whether there are allowed NUMA nodes inside that object. + * Applying hwloc_bitmap_and() returns the list of these allowed NUMA nodes. + * + * \note The returned nodeset is not newly allocated and should thus not be + * changed or freed, hwloc_bitmap_dup() must be used to obtain a local copy. + */ +HWLOC_DECLSPEC hwloc_const_nodeset_t +hwloc_topology_get_allowed_nodeset(hwloc_topology_t topology) __hwloc_attribute_pure; + +/** @} */ + + + +/** \defgroup hwlocality_helper_nodeset_convert Converting between CPU sets and node sets + * + * @{ + */ + +/** \brief Convert a CPU set into a NUMA node set and handle non-NUMA cases + * + * If some NUMA nodes have no CPUs at all, this function never sets their + * indexes in the output node set, even if a full CPU set is given in input. + * + * If the topology contains no NUMA nodes, the machine is considered + * as a single memory node, and the following behavior is used: + * If \p cpuset is empty, \p nodeset will be emptied as well. + * Otherwise \p nodeset will be entirely filled. + */ +static __hwloc_inline int +hwloc_cpuset_to_nodeset(hwloc_topology_t topology, hwloc_const_cpuset_t _cpuset, hwloc_nodeset_t nodeset) +{ + int depth = hwloc_get_type_depth(topology, HWLOC_OBJ_NUMANODE); + hwloc_obj_t obj = NULL; + assert(depth != HWLOC_TYPE_DEPTH_UNKNOWN); + hwloc_bitmap_zero(nodeset); + while ((obj = hwloc_get_next_obj_covering_cpuset_by_depth(topology, _cpuset, depth, obj)) != NULL) + if (hwloc_bitmap_set(nodeset, obj->os_index) < 0) + return -1; + return 0; +} + +/** \brief Convert a NUMA node set into a CPU set and handle non-NUMA cases + * + * If the topology contains no NUMA nodes, the machine is considered + * as a single memory node, and the following behavior is used: + * If \p nodeset is empty, \p cpuset will be emptied as well. + * Otherwise \p cpuset will be entirely filled. + * This is useful for manipulating memory binding sets. + */ +static __hwloc_inline int +hwloc_cpuset_from_nodeset(hwloc_topology_t topology, hwloc_cpuset_t _cpuset, hwloc_const_nodeset_t nodeset) +{ + int depth = hwloc_get_type_depth(topology, HWLOC_OBJ_NUMANODE); + hwloc_obj_t obj = NULL; + assert(depth != HWLOC_TYPE_DEPTH_UNKNOWN); + hwloc_bitmap_zero(_cpuset); + while ((obj = hwloc_get_next_obj_by_depth(topology, depth, obj)) != NULL) { + if (hwloc_bitmap_isset(nodeset, obj->os_index)) + /* no need to check obj->cpuset because objects in levels always have a cpuset */ + if (hwloc_bitmap_or(_cpuset, _cpuset, obj->cpuset) < 0) + return -1; + } + return 0; +} + +/** @} */ + + + +/** \defgroup hwlocality_advanced_io Finding I/O objects + * @{ + */ + +/** \brief Get the first non-I/O ancestor object. + * + * Given the I/O object \p ioobj, find the smallest non-I/O ancestor + * object. This object (normal or memory) may then be used for binding + * because it has non-NULL CPU and node sets + * and because its locality is the same as \p ioobj. + * + * \note The resulting object is usually a normal object but it could also + * be a memory object (e.g. NUMA node) in future platforms if I/O objects + * ever get attached to memory instead of CPUs. + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_non_io_ancestor_obj(hwloc_topology_t topology __hwloc_attribute_unused, + hwloc_obj_t ioobj) +{ + hwloc_obj_t obj = ioobj; + while (obj && !obj->cpuset) { + obj = obj->parent; + } + return obj; +} + +/** \brief Get the next PCI device in the system. + * + * \return the first PCI device if \p prev is \c NULL. + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_next_pcidev(hwloc_topology_t topology, hwloc_obj_t prev) +{ + return hwloc_get_next_obj_by_type(topology, HWLOC_OBJ_PCI_DEVICE, prev); +} + +/** \brief Find the PCI device object matching the PCI bus id + * given domain, bus device and function PCI bus id. + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_pcidev_by_busid(hwloc_topology_t topology, + unsigned domain, unsigned bus, unsigned dev, unsigned func) +{ + hwloc_obj_t obj = NULL; + while ((obj = hwloc_get_next_pcidev(topology, obj)) != NULL) { + if (obj->attr->pcidev.domain == domain + && obj->attr->pcidev.bus == bus + && obj->attr->pcidev.dev == dev + && obj->attr->pcidev.func == func) + return obj; + } + return NULL; +} + +/** \brief Find the PCI device object matching the PCI bus id + * given as a string xxxx:yy:zz.t or yy:zz.t. + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_pcidev_by_busidstring(hwloc_topology_t topology, const char *busid) +{ + unsigned domain = 0; /* default */ + unsigned bus, dev, func; + + if (sscanf(busid, "%x:%x.%x", &bus, &dev, &func) != 3 + && sscanf(busid, "%x:%x:%x.%x", &domain, &bus, &dev, &func) != 4) { + errno = EINVAL; + return NULL; + } + + return hwloc_get_pcidev_by_busid(topology, domain, bus, dev, func); +} + +/** \brief Get the next OS device in the system. + * + * \return the first OS device if \p prev is \c NULL. + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_next_osdev(hwloc_topology_t topology, hwloc_obj_t prev) +{ + return hwloc_get_next_obj_by_type(topology, HWLOC_OBJ_OS_DEVICE, prev); +} + +/** \brief Get the next bridge in the system. + * + * \return the first bridge if \p prev is \c NULL. + */ +static __hwloc_inline hwloc_obj_t +hwloc_get_next_bridge(hwloc_topology_t topology, hwloc_obj_t prev) +{ + return hwloc_get_next_obj_by_type(topology, HWLOC_OBJ_BRIDGE, prev); +} + +/* \brief Checks whether a given bridge covers a given PCI bus. + */ +static __hwloc_inline int +hwloc_bridge_covers_pcibus(hwloc_obj_t bridge, + unsigned domain, unsigned bus) +{ + return bridge->type == HWLOC_OBJ_BRIDGE + && bridge->attr->bridge.downstream_type == HWLOC_OBJ_BRIDGE_PCI + && bridge->attr->bridge.downstream.pci.domain == domain + && bridge->attr->bridge.downstream.pci.secondary_bus <= bus + && bridge->attr->bridge.downstream.pci.subordinate_bus >= bus; +} + +/** @} */ + + + +#ifdef __cplusplus +} /* extern "C" */ +#endif + + +#endif /* HWLOC_HELPER_H */ diff --git a/src/3rdparty/hwloc/include/hwloc/inlines.h b/src/3rdparty/hwloc/include/hwloc/inlines.h new file mode 100644 index 000000000..494209ea6 --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc/inlines.h @@ -0,0 +1,146 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2018 Inria. All rights reserved. + * Copyright © 2009-2012 Université Bordeaux + * Copyright © 2009-2010 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +/** + * This file contains the inline code of functions declared in hwloc.h + */ + +#ifndef HWLOC_INLINES_H +#define HWLOC_INLINES_H + +#ifndef HWLOC_H +#error Please include the main hwloc.h instead +#endif + +#include +#include + + +#ifdef __cplusplus +extern "C" { +#endif + +static __hwloc_inline int +hwloc_get_type_or_below_depth (hwloc_topology_t topology, hwloc_obj_type_t type) +{ + int depth = hwloc_get_type_depth(topology, type); + + if (depth != HWLOC_TYPE_DEPTH_UNKNOWN) + return depth; + + /* find the highest existing level with type order >= */ + for(depth = hwloc_get_type_depth(topology, HWLOC_OBJ_PU); ; depth--) + if (hwloc_compare_types(hwloc_get_depth_type(topology, depth), type) < 0) + return depth+1; + + /* Shouldn't ever happen, as there is always a Machine level with lower order and known depth. */ + /* abort(); */ +} + +static __hwloc_inline int +hwloc_get_type_or_above_depth (hwloc_topology_t topology, hwloc_obj_type_t type) +{ + int depth = hwloc_get_type_depth(topology, type); + + if (depth != HWLOC_TYPE_DEPTH_UNKNOWN) + return depth; + + /* find the lowest existing level with type order <= */ + for(depth = 0; ; depth++) + if (hwloc_compare_types(hwloc_get_depth_type(topology, depth), type) > 0) + return depth-1; + + /* Shouldn't ever happen, as there is always a PU level with higher order and known depth. */ + /* abort(); */ +} + +static __hwloc_inline int +hwloc_get_nbobjs_by_type (hwloc_topology_t topology, hwloc_obj_type_t type) +{ + int depth = hwloc_get_type_depth(topology, type); + if (depth == HWLOC_TYPE_DEPTH_UNKNOWN) + return 0; + if (depth == HWLOC_TYPE_DEPTH_MULTIPLE) + return -1; /* FIXME: agregate nbobjs from different levels? */ + return (int) hwloc_get_nbobjs_by_depth(topology, depth); +} + +static __hwloc_inline hwloc_obj_t +hwloc_get_obj_by_type (hwloc_topology_t topology, hwloc_obj_type_t type, unsigned idx) +{ + int depth = hwloc_get_type_depth(topology, type); + if (depth == HWLOC_TYPE_DEPTH_UNKNOWN) + return NULL; + if (depth == HWLOC_TYPE_DEPTH_MULTIPLE) + return NULL; + return hwloc_get_obj_by_depth(topology, depth, idx); +} + +static __hwloc_inline hwloc_obj_t +hwloc_get_next_obj_by_depth (hwloc_topology_t topology, int depth, hwloc_obj_t prev) +{ + if (!prev) + return hwloc_get_obj_by_depth (topology, depth, 0); + if (prev->depth != depth) + return NULL; + return prev->next_cousin; +} + +static __hwloc_inline hwloc_obj_t +hwloc_get_next_obj_by_type (hwloc_topology_t topology, hwloc_obj_type_t type, + hwloc_obj_t prev) +{ + int depth = hwloc_get_type_depth(topology, type); + if (depth == HWLOC_TYPE_DEPTH_UNKNOWN || depth == HWLOC_TYPE_DEPTH_MULTIPLE) + return NULL; + return hwloc_get_next_obj_by_depth (topology, depth, prev); +} + +static __hwloc_inline hwloc_obj_t +hwloc_get_root_obj (hwloc_topology_t topology) +{ + return hwloc_get_obj_by_depth (topology, 0, 0); +} + +static __hwloc_inline const char * +hwloc_obj_get_info_by_name(hwloc_obj_t obj, const char *name) +{ + unsigned i; + for(i=0; iinfos_count; i++) { + struct hwloc_info_s *info = &obj->infos[i]; + if (!strcmp(info->name, name)) + return info->value; + } + return NULL; +} + +static __hwloc_inline void * +hwloc_alloc_membind_policy(hwloc_topology_t topology, size_t len, hwloc_const_cpuset_t set, hwloc_membind_policy_t policy, int flags) +{ + void *p = hwloc_alloc_membind(topology, len, set, policy, flags); + if (p) + return p; + + if (hwloc_set_membind(topology, set, policy, flags) < 0) + /* hwloc_set_membind() takes care of ignoring errors if non-STRICT */ + return NULL; + + p = hwloc_alloc(topology, len); + if (p && policy != HWLOC_MEMBIND_FIRSTTOUCH) + /* Enforce the binding by touching the data */ + memset(p, 0, len); + return p; +} + + +#ifdef __cplusplus +} /* extern "C" */ +#endif + + +#endif /* HWLOC_INLINES_H */ diff --git a/src/3rdparty/hwloc/include/hwloc/intel-mic.h b/src/3rdparty/hwloc/include/hwloc/intel-mic.h new file mode 100644 index 000000000..6f6f9d1b3 --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc/intel-mic.h @@ -0,0 +1,134 @@ +/* + * Copyright © 2013-2016 Inria. All rights reserved. + * See COPYING in top-level directory. + */ + +/** \file + * \brief Macros to help interaction between hwloc and Intel Xeon Phi (MIC). + * + * Applications that use both hwloc and Intel Xeon Phi (MIC) may want to + * include this file so as to get topology information for MIC devices. + */ + +#ifndef HWLOC_INTEL_MIC_H +#define HWLOC_INTEL_MIC_H + +#include +#include +#include +#ifdef HWLOC_LINUX_SYS +#include +#include +#include +#endif + +#include +#include + + +#ifdef __cplusplus +extern "C" { +#endif + + +/** \defgroup hwlocality_intel_mic Interoperability with Intel Xeon Phi (MIC) + * + * This interface offers ways to retrieve topology information about + * Intel Xeon Phi (MIC) devices. + * + * @{ + */ + +/** \brief Get the CPU set of logical processors that are physically + * close to MIC device whose index is \p idx. + * + * Return the CPU set describing the locality of the MIC device whose index is \p idx. + * + * Topology \p topology and device index \p idx must match the local machine. + * I/O devices detection is not needed in the topology. + * + * The function only returns the locality of the device. + * If more information about the device is needed, OS objects should + * be used instead, see hwloc_intel_mic_get_device_osdev_by_index(). + * + * This function is currently only implemented in a meaningful way for + * Linux; other systems will simply get a full cpuset. + */ +static __hwloc_inline int +hwloc_intel_mic_get_device_cpuset(hwloc_topology_t topology __hwloc_attribute_unused, + int idx __hwloc_attribute_unused, + hwloc_cpuset_t set) +{ +#ifdef HWLOC_LINUX_SYS + /* If we're on Linux, use the sysfs mechanism to get the local cpus */ +#define HWLOC_INTEL_MIC_DEVICE_SYSFS_PATH_MAX 128 + char path[HWLOC_INTEL_MIC_DEVICE_SYSFS_PATH_MAX]; + DIR *sysdir = NULL; + struct dirent *dirent; + unsigned pcibus, pcidev, pcifunc; + + if (!hwloc_topology_is_thissystem(topology)) { + errno = EINVAL; + return -1; + } + + sprintf(path, "/sys/class/mic/mic%d", idx); + sysdir = opendir(path); + if (!sysdir) + return -1; + + while ((dirent = readdir(sysdir)) != NULL) { + if (sscanf(dirent->d_name, "pci_%02x:%02x.%02x", &pcibus, &pcidev, &pcifunc) == 3) { + sprintf(path, "/sys/class/mic/mic%d/pci_%02x:%02x.%02x/local_cpus", idx, pcibus, pcidev, pcifunc); + if (hwloc_linux_read_path_as_cpumask(path, set) < 0 + || hwloc_bitmap_iszero(set)) + hwloc_bitmap_copy(set, hwloc_topology_get_complete_cpuset(topology)); + break; + } + } + + closedir(sysdir); +#else + /* Non-Linux systems simply get a full cpuset */ + hwloc_bitmap_copy(set, hwloc_topology_get_complete_cpuset(topology)); +#endif + return 0; +} + +/** \brief Get the hwloc OS device object corresponding to the + * MIC device for the given index. + * + * Return the OS device object describing the MIC device whose index is \p idx. + * Return NULL if there is none. + * + * The topology \p topology does not necessarily have to match the current + * machine. For instance the topology may be an XML import of a remote host. + * I/O devices detection must be enabled in the topology. + * + * \note The corresponding PCI device object can be obtained by looking + * at the OS device parent object. + */ +static __hwloc_inline hwloc_obj_t +hwloc_intel_mic_get_device_osdev_by_index(hwloc_topology_t topology, + unsigned idx) +{ + hwloc_obj_t osdev = NULL; + while ((osdev = hwloc_get_next_osdev(topology, osdev)) != NULL) { + if (HWLOC_OBJ_OSDEV_COPROC == osdev->attr->osdev.type + && osdev->name + && !strncmp("mic", osdev->name, 3) + && atoi(osdev->name + 3) == (int) idx) + return osdev; + } + return NULL; +} + +/** @} */ + + +#ifdef __cplusplus +} /* extern "C" */ +#endif + + +#endif /* HWLOC_INTEL_MIC_H */ diff --git a/src/3rdparty/hwloc/include/hwloc/linux-libnuma.h b/src/3rdparty/hwloc/include/hwloc/linux-libnuma.h new file mode 100644 index 000000000..7cea4166b --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc/linux-libnuma.h @@ -0,0 +1,273 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2017 Inria. All rights reserved. + * Copyright © 2009-2010, 2012 Université Bordeaux + * See COPYING in top-level directory. + */ + +/** \file + * \brief Macros to help interaction between hwloc and Linux libnuma. + * + * Applications that use both Linux libnuma and hwloc may want to + * include this file so as to ease conversion between their respective types. +*/ + +#ifndef HWLOC_LINUX_LIBNUMA_H +#define HWLOC_LINUX_LIBNUMA_H + +#include +#include + + +#ifdef __cplusplus +extern "C" { +#endif + + +/** \defgroup hwlocality_linux_libnuma_ulongs Interoperability with Linux libnuma unsigned long masks + * + * This interface helps converting between Linux libnuma unsigned long masks + * and hwloc cpusets and nodesets. + * + * \note Topology \p topology must match the current machine. + * + * \note The behavior of libnuma is undefined if the kernel is not NUMA-aware. + * (when CONFIG_NUMA is not set in the kernel configuration). + * This helper and libnuma may thus not be strictly compatible in this case, + * which may be detected by checking whether numa_available() returns -1. + * + * @{ + */ + + +/** \brief Convert hwloc CPU set \p cpuset into the array of unsigned long \p mask + * + * \p mask is the array of unsigned long that will be filled. + * \p maxnode contains the maximal node number that may be stored in \p mask. + * \p maxnode will be set to the maximal node number that was found, plus one. + * + * This function may be used before calling set_mempolicy, mbind, migrate_pages + * or any other function that takes an array of unsigned long and a maximal + * node number as input parameter. + */ +static __hwloc_inline int +hwloc_cpuset_to_linux_libnuma_ulongs(hwloc_topology_t topology, hwloc_const_cpuset_t cpuset, + unsigned long *mask, unsigned long *maxnode) +{ + int depth = hwloc_get_type_depth(topology, HWLOC_OBJ_NUMANODE); + unsigned long outmaxnode = -1; + hwloc_obj_t node = NULL; + + /* round-up to the next ulong and clear all bytes */ + *maxnode = (*maxnode + 8*sizeof(*mask) - 1) & ~(8*sizeof(*mask) - 1); + memset(mask, 0, *maxnode/8); + + while ((node = hwloc_get_next_obj_covering_cpuset_by_depth(topology, cpuset, depth, node)) != NULL) { + if (node->os_index >= *maxnode) + continue; + mask[node->os_index/sizeof(*mask)/8] |= 1UL << (node->os_index % (sizeof(*mask)*8)); + if (outmaxnode == (unsigned long) -1 || outmaxnode < node->os_index) + outmaxnode = node->os_index; + } + + *maxnode = outmaxnode+1; + return 0; +} + +/** \brief Convert hwloc NUMA node set \p nodeset into the array of unsigned long \p mask + * + * \p mask is the array of unsigned long that will be filled. + * \p maxnode contains the maximal node number that may be stored in \p mask. + * \p maxnode will be set to the maximal node number that was found, plus one. + * + * This function may be used before calling set_mempolicy, mbind, migrate_pages + * or any other function that takes an array of unsigned long and a maximal + * node number as input parameter. + */ +static __hwloc_inline int +hwloc_nodeset_to_linux_libnuma_ulongs(hwloc_topology_t topology, hwloc_const_nodeset_t nodeset, + unsigned long *mask, unsigned long *maxnode) +{ + int depth = hwloc_get_type_depth(topology, HWLOC_OBJ_NUMANODE); + unsigned long outmaxnode = -1; + hwloc_obj_t node = NULL; + + /* round-up to the next ulong and clear all bytes */ + *maxnode = (*maxnode + 8*sizeof(*mask) - 1) & ~(8*sizeof(*mask) - 1); + memset(mask, 0, *maxnode/8); + + while ((node = hwloc_get_next_obj_by_depth(topology, depth, node)) != NULL) { + if (node->os_index >= *maxnode) + continue; + if (!hwloc_bitmap_isset(nodeset, node->os_index)) + continue; + mask[node->os_index/sizeof(*mask)/8] |= 1UL << (node->os_index % (sizeof(*mask)*8)); + if (outmaxnode == (unsigned long) -1 || outmaxnode < node->os_index) + outmaxnode = node->os_index; + } + + *maxnode = outmaxnode+1; + return 0; +} + +/** \brief Convert the array of unsigned long \p mask into hwloc CPU set + * + * \p mask is a array of unsigned long that will be read. + * \p maxnode contains the maximal node number that may be read in \p mask. + * + * This function may be used after calling get_mempolicy or any other function + * that takes an array of unsigned long as output parameter (and possibly + * a maximal node number as input parameter). + */ +static __hwloc_inline int +hwloc_cpuset_from_linux_libnuma_ulongs(hwloc_topology_t topology, hwloc_cpuset_t cpuset, + const unsigned long *mask, unsigned long maxnode) +{ + int depth = hwloc_get_type_depth(topology, HWLOC_OBJ_NUMANODE); + hwloc_obj_t node = NULL; + hwloc_bitmap_zero(cpuset); + while ((node = hwloc_get_next_obj_by_depth(topology, depth, node)) != NULL) + if (node->os_index < maxnode + && (mask[node->os_index/sizeof(*mask)/8] & (1UL << (node->os_index % (sizeof(*mask)*8))))) + hwloc_bitmap_or(cpuset, cpuset, node->cpuset); + return 0; +} + +/** \brief Convert the array of unsigned long \p mask into hwloc NUMA node set + * + * \p mask is a array of unsigned long that will be read. + * \p maxnode contains the maximal node number that may be read in \p mask. + * + * This function may be used after calling get_mempolicy or any other function + * that takes an array of unsigned long as output parameter (and possibly + * a maximal node number as input parameter). + */ +static __hwloc_inline int +hwloc_nodeset_from_linux_libnuma_ulongs(hwloc_topology_t topology, hwloc_nodeset_t nodeset, + const unsigned long *mask, unsigned long maxnode) +{ + int depth = hwloc_get_type_depth(topology, HWLOC_OBJ_NUMANODE); + hwloc_obj_t node = NULL; + hwloc_bitmap_zero(nodeset); + while ((node = hwloc_get_next_obj_by_depth(topology, depth, node)) != NULL) + if (node->os_index < maxnode + && (mask[node->os_index/sizeof(*mask)/8] & (1UL << (node->os_index % (sizeof(*mask)*8))))) + hwloc_bitmap_set(nodeset, node->os_index); + return 0; +} + +/** @} */ + + + +/** \defgroup hwlocality_linux_libnuma_bitmask Interoperability with Linux libnuma bitmask + * + * This interface helps converting between Linux libnuma bitmasks + * and hwloc cpusets and nodesets. + * + * \note Topology \p topology must match the current machine. + * + * \note The behavior of libnuma is undefined if the kernel is not NUMA-aware. + * (when CONFIG_NUMA is not set in the kernel configuration). + * This helper and libnuma may thus not be strictly compatible in this case, + * which may be detected by checking whether numa_available() returns -1. + * + * @{ + */ + + +/** \brief Convert hwloc CPU set \p cpuset into the returned libnuma bitmask + * + * The returned bitmask should later be freed with numa_bitmask_free. + * + * This function may be used before calling many numa_ functions + * that use a struct bitmask as an input parameter. + * + * \return newly allocated struct bitmask. + */ +static __hwloc_inline struct bitmask * +hwloc_cpuset_to_linux_libnuma_bitmask(hwloc_topology_t topology, hwloc_const_cpuset_t cpuset) __hwloc_attribute_malloc; +static __hwloc_inline struct bitmask * +hwloc_cpuset_to_linux_libnuma_bitmask(hwloc_topology_t topology, hwloc_const_cpuset_t cpuset) +{ + int depth = hwloc_get_type_depth(topology, HWLOC_OBJ_NUMANODE); + hwloc_obj_t node = NULL; + struct bitmask *bitmask = numa_allocate_cpumask(); + if (!bitmask) + return NULL; + while ((node = hwloc_get_next_obj_covering_cpuset_by_depth(topology, cpuset, depth, node)) != NULL) + if (node->attr->numanode.local_memory) + numa_bitmask_setbit(bitmask, node->os_index); + return bitmask; +} + +/** \brief Convert hwloc NUMA node set \p nodeset into the returned libnuma bitmask + * + * The returned bitmask should later be freed with numa_bitmask_free. + * + * This function may be used before calling many numa_ functions + * that use a struct bitmask as an input parameter. + * + * \return newly allocated struct bitmask. + */ +static __hwloc_inline struct bitmask * +hwloc_nodeset_to_linux_libnuma_bitmask(hwloc_topology_t topology, hwloc_const_nodeset_t nodeset) __hwloc_attribute_malloc; +static __hwloc_inline struct bitmask * +hwloc_nodeset_to_linux_libnuma_bitmask(hwloc_topology_t topology, hwloc_const_nodeset_t nodeset) +{ + int depth = hwloc_get_type_depth(topology, HWLOC_OBJ_NUMANODE); + hwloc_obj_t node = NULL; + struct bitmask *bitmask = numa_allocate_cpumask(); + if (!bitmask) + return NULL; + while ((node = hwloc_get_next_obj_by_depth(topology, depth, node)) != NULL) + if (hwloc_bitmap_isset(nodeset, node->os_index) && node->attr->numanode.local_memory) + numa_bitmask_setbit(bitmask, node->os_index); + return bitmask; +} + +/** \brief Convert libnuma bitmask \p bitmask into hwloc CPU set \p cpuset + * + * This function may be used after calling many numa_ functions + * that use a struct bitmask as an output parameter. + */ +static __hwloc_inline int +hwloc_cpuset_from_linux_libnuma_bitmask(hwloc_topology_t topology, hwloc_cpuset_t cpuset, + const struct bitmask *bitmask) +{ + int depth = hwloc_get_type_depth(topology, HWLOC_OBJ_NUMANODE); + hwloc_obj_t node = NULL; + hwloc_bitmap_zero(cpuset); + while ((node = hwloc_get_next_obj_by_depth(topology, depth, node)) != NULL) + if (numa_bitmask_isbitset(bitmask, node->os_index)) + hwloc_bitmap_or(cpuset, cpuset, node->cpuset); + return 0; +} + +/** \brief Convert libnuma bitmask \p bitmask into hwloc NUMA node set \p nodeset + * + * This function may be used after calling many numa_ functions + * that use a struct bitmask as an output parameter. + */ +static __hwloc_inline int +hwloc_nodeset_from_linux_libnuma_bitmask(hwloc_topology_t topology, hwloc_nodeset_t nodeset, + const struct bitmask *bitmask) +{ + int depth = hwloc_get_type_depth(topology, HWLOC_OBJ_NUMANODE); + hwloc_obj_t node = NULL; + hwloc_bitmap_zero(nodeset); + while ((node = hwloc_get_next_obj_by_depth(topology, depth, node)) != NULL) + if (numa_bitmask_isbitset(bitmask, node->os_index)) + hwloc_bitmap_set(nodeset, node->os_index); + return 0; +} + +/** @} */ + + +#ifdef __cplusplus +} /* extern "C" */ +#endif + + +#endif /* HWLOC_LINUX_NUMA_H */ diff --git a/src/3rdparty/hwloc/include/hwloc/linux.h b/src/3rdparty/hwloc/include/hwloc/linux.h new file mode 100644 index 000000000..c409e1c2a --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc/linux.h @@ -0,0 +1,79 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2016 Inria. All rights reserved. + * Copyright © 2009-2011 Université Bordeaux + * See COPYING in top-level directory. + */ + +/** \file + * \brief Macros to help interaction between hwloc and Linux. + * + * Applications that use hwloc on Linux may want to include this file + * if using some low-level Linux features. + */ + +#ifndef HWLOC_LINUX_H +#define HWLOC_LINUX_H + +#include +#include + + +#ifdef __cplusplus +extern "C" { +#endif + + +/** \defgroup hwlocality_linux Linux-specific helpers + * + * This includes helpers for manipulating Linux kernel cpumap files, and hwloc + * equivalents of the Linux sched_setaffinity and sched_getaffinity system calls. + * + * @{ + */ + +/** \brief Bind a thread \p tid on cpus given in cpuset \p set + * + * The behavior is exactly the same as the Linux sched_setaffinity system call, + * but uses a hwloc cpuset. + * + * \note This is equivalent to calling hwloc_set_proc_cpubind() with + * HWLOC_CPUBIND_THREAD as flags. + */ +HWLOC_DECLSPEC int hwloc_linux_set_tid_cpubind(hwloc_topology_t topology, pid_t tid, hwloc_const_cpuset_t set); + +/** \brief Get the current binding of thread \p tid + * + * The behavior is exactly the same as the Linux sched_getaffinity system call, + * but uses a hwloc cpuset. + * + * \note This is equivalent to calling hwloc_get_proc_cpubind() with + * ::HWLOC_CPUBIND_THREAD as flags. + */ +HWLOC_DECLSPEC int hwloc_linux_get_tid_cpubind(hwloc_topology_t topology, pid_t tid, hwloc_cpuset_t set); + +/** \brief Get the last physical CPU where thread \p tid ran. + * + * \note This is equivalent to calling hwloc_get_proc_last_cpu_location() with + * ::HWLOC_CPUBIND_THREAD as flags. + */ +HWLOC_DECLSPEC int hwloc_linux_get_tid_last_cpu_location(hwloc_topology_t topology, pid_t tid, hwloc_bitmap_t set); + +/** \brief Convert a linux kernel cpumask file \p path into a hwloc bitmap \p set. + * + * Might be used when reading CPU set from sysfs attributes such as topology + * and caches for processors, or local_cpus for devices. + * + * \note This function ignores the HWLOC_FSROOT environment variable. + */ +HWLOC_DECLSPEC int hwloc_linux_read_path_as_cpumask(const char *path, hwloc_bitmap_t set); + +/** @} */ + + +#ifdef __cplusplus +} /* extern "C" */ +#endif + + +#endif /* HWLOC_LINUX_H */ diff --git a/src/3rdparty/hwloc/include/hwloc/nvml.h b/src/3rdparty/hwloc/include/hwloc/nvml.h new file mode 100644 index 000000000..197108660 --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc/nvml.h @@ -0,0 +1,181 @@ +/* + * Copyright © 2012-2016 Inria. All rights reserved. + * See COPYING in top-level directory. + */ + +/** \file + * \brief Macros to help interaction between hwloc and the NVIDIA Management Library. + * + * Applications that use both hwloc and the NVIDIA Management Library may want to + * include this file so as to get topology information for NVML devices. + */ + +#ifndef HWLOC_NVML_H +#define HWLOC_NVML_H + +#include +#include +#include +#ifdef HWLOC_LINUX_SYS +#include +#endif + +#include + + +#ifdef __cplusplus +extern "C" { +#endif + + +/** \defgroup hwlocality_nvml Interoperability with the NVIDIA Management Library + * + * This interface offers ways to retrieve topology information about + * devices managed by the NVIDIA Management Library (NVML). + * + * @{ + */ + +/** \brief Get the CPU set of logical processors that are physically + * close to NVML device \p device. + * + * Return the CPU set describing the locality of the NVML device \p device. + * + * Topology \p topology and device \p device must match the local machine. + * I/O devices detection and the NVML component are not needed in the topology. + * + * The function only returns the locality of the device. + * If more information about the device is needed, OS objects should + * be used instead, see hwloc_nvml_get_device_osdev() + * and hwloc_nvml_get_device_osdev_by_index(). + * + * This function is currently only implemented in a meaningful way for + * Linux; other systems will simply get a full cpuset. + */ +static __hwloc_inline int +hwloc_nvml_get_device_cpuset(hwloc_topology_t topology __hwloc_attribute_unused, + nvmlDevice_t device, hwloc_cpuset_t set) +{ +#ifdef HWLOC_LINUX_SYS + /* If we're on Linux, use the sysfs mechanism to get the local cpus */ +#define HWLOC_NVML_DEVICE_SYSFS_PATH_MAX 128 + char path[HWLOC_NVML_DEVICE_SYSFS_PATH_MAX]; + nvmlReturn_t nvres; + nvmlPciInfo_t pci; + + if (!hwloc_topology_is_thissystem(topology)) { + errno = EINVAL; + return -1; + } + + nvres = nvmlDeviceGetPciInfo(device, &pci); + if (NVML_SUCCESS != nvres) { + errno = EINVAL; + return -1; + } + + sprintf(path, "/sys/bus/pci/devices/%04x:%02x:%02x.0/local_cpus", pci.domain, pci.bus, pci.device); + if (hwloc_linux_read_path_as_cpumask(path, set) < 0 + || hwloc_bitmap_iszero(set)) + hwloc_bitmap_copy(set, hwloc_topology_get_complete_cpuset(topology)); +#else + /* Non-Linux systems simply get a full cpuset */ + hwloc_bitmap_copy(set, hwloc_topology_get_complete_cpuset(topology)); +#endif + return 0; +} + +/** \brief Get the hwloc OS device object corresponding to the + * NVML device whose index is \p idx. + * + * Return the OS device object describing the NVML device whose + * index is \p idx. Returns NULL if there is none. + * + * The topology \p topology does not necessarily have to match the current + * machine. For instance the topology may be an XML import of a remote host. + * I/O devices detection and the NVML component must be enabled in the topology. + * + * \note The corresponding PCI device object can be obtained by looking + * at the OS device parent object (unless PCI devices are filtered out). + */ +static __hwloc_inline hwloc_obj_t +hwloc_nvml_get_device_osdev_by_index(hwloc_topology_t topology, unsigned idx) +{ + hwloc_obj_t osdev = NULL; + while ((osdev = hwloc_get_next_osdev(topology, osdev)) != NULL) { + if (HWLOC_OBJ_OSDEV_GPU == osdev->attr->osdev.type + && osdev->name + && !strncmp("nvml", osdev->name, 4) + && atoi(osdev->name + 4) == (int) idx) + return osdev; + } + return NULL; +} + +/** \brief Get the hwloc OS device object corresponding to NVML device \p device. + * + * Return the hwloc OS device object that describes the given + * NVML device \p device. Return NULL if there is none. + * + * Topology \p topology and device \p device must match the local machine. + * I/O devices detection and the NVML component must be enabled in the topology. + * If not, the locality of the object may still be found using + * hwloc_nvml_get_device_cpuset(). + * + * \note The corresponding hwloc PCI device may be found by looking + * at the result parent pointer (unless PCI devices are filtered out). + */ +static __hwloc_inline hwloc_obj_t +hwloc_nvml_get_device_osdev(hwloc_topology_t topology, nvmlDevice_t device) +{ + hwloc_obj_t osdev; + nvmlReturn_t nvres; + nvmlPciInfo_t pci; + char uuid[64]; + + if (!hwloc_topology_is_thissystem(topology)) { + errno = EINVAL; + return NULL; + } + + nvres = nvmlDeviceGetPciInfo(device, &pci); + if (NVML_SUCCESS != nvres) + return NULL; + + nvres = nvmlDeviceGetUUID(device, uuid, sizeof(uuid)); + if (NVML_SUCCESS != nvres) + uuid[0] = '\0'; + + osdev = NULL; + while ((osdev = hwloc_get_next_osdev(topology, osdev)) != NULL) { + hwloc_obj_t pcidev = osdev->parent; + const char *info; + + if (strncmp(osdev->name, "nvml", 4)) + continue; + + if (pcidev + && pcidev->type == HWLOC_OBJ_PCI_DEVICE + && pcidev->attr->pcidev.domain == pci.domain + && pcidev->attr->pcidev.bus == pci.bus + && pcidev->attr->pcidev.dev == pci.device + && pcidev->attr->pcidev.func == 0) + return osdev; + + info = hwloc_obj_get_info_by_name(osdev, "NVIDIAUUID"); + if (info && !strcmp(info, uuid)) + return osdev; + } + + return NULL; +} + +/** @} */ + + +#ifdef __cplusplus +} /* extern "C" */ +#endif + + +#endif /* HWLOC_NVML_H */ diff --git a/src/3rdparty/hwloc/include/hwloc/opencl.h b/src/3rdparty/hwloc/include/hwloc/opencl.h new file mode 100644 index 000000000..058968d74 --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc/opencl.h @@ -0,0 +1,206 @@ +/* + * Copyright © 2012-2018 Inria. All rights reserved. + * Copyright © 2013, 2018 Université Bordeaux. All right reserved. + * See COPYING in top-level directory. + */ + +/** \file + * \brief Macros to help interaction between hwloc and the OpenCL interface. + * + * Applications that use both hwloc and OpenCL may want to + * include this file so as to get topology information for OpenCL devices. + */ + +#ifndef HWLOC_OPENCL_H +#define HWLOC_OPENCL_H + +#include +#include +#include +#ifdef HWLOC_LINUX_SYS +#include +#endif + +#ifdef __APPLE__ +#include +#include +#else +#include +#include +#endif + +#include + + +#ifdef __cplusplus +extern "C" { +#endif + + +/** \defgroup hwlocality_opencl Interoperability with OpenCL + * + * This interface offers ways to retrieve topology information about + * OpenCL devices. + * + * Only the AMD OpenCL interface currently offers useful locality information + * about its devices. + * + * @{ + */ + +/** \brief Get the CPU set of logical processors that are physically + * close to OpenCL device \p device. + * + * Return the CPU set describing the locality of the OpenCL device \p device. + * + * Topology \p topology and device \p device must match the local machine. + * I/O devices detection and the OpenCL component are not needed in the topology. + * + * The function only returns the locality of the device. + * If more information about the device is needed, OS objects should + * be used instead, see hwloc_opencl_get_device_osdev() + * and hwloc_opencl_get_device_osdev_by_index(). + * + * This function is currently only implemented in a meaningful way for + * Linux with the AMD OpenCL implementation; other systems will simply + * get a full cpuset. + */ +static __hwloc_inline int +hwloc_opencl_get_device_cpuset(hwloc_topology_t topology __hwloc_attribute_unused, + cl_device_id device __hwloc_attribute_unused, + hwloc_cpuset_t set) +{ +#if (defined HWLOC_LINUX_SYS) && (defined CL_DEVICE_TOPOLOGY_AMD) + /* If we're on Linux + AMD OpenCL, use the AMD extension + the sysfs mechanism to get the local cpus */ +#define HWLOC_OPENCL_DEVICE_SYSFS_PATH_MAX 128 + char path[HWLOC_OPENCL_DEVICE_SYSFS_PATH_MAX]; + cl_device_topology_amd amdtopo; + cl_int clret; + + if (!hwloc_topology_is_thissystem(topology)) { + errno = EINVAL; + return -1; + } + + clret = clGetDeviceInfo(device, CL_DEVICE_TOPOLOGY_AMD, sizeof(amdtopo), &amdtopo, NULL); + if (CL_SUCCESS != clret) { + hwloc_bitmap_copy(set, hwloc_topology_get_complete_cpuset(topology)); + return 0; + } + if (CL_DEVICE_TOPOLOGY_TYPE_PCIE_AMD != amdtopo.raw.type) { + hwloc_bitmap_copy(set, hwloc_topology_get_complete_cpuset(topology)); + return 0; + } + + sprintf(path, "/sys/bus/pci/devices/0000:%02x:%02x.%01x/local_cpus", + (unsigned) amdtopo.pcie.bus, (unsigned) amdtopo.pcie.device, (unsigned) amdtopo.pcie.function); + if (hwloc_linux_read_path_as_cpumask(path, set) < 0 + || hwloc_bitmap_iszero(set)) + hwloc_bitmap_copy(set, hwloc_topology_get_complete_cpuset(topology)); +#else + /* Non-Linux + AMD OpenCL systems simply get a full cpuset */ + hwloc_bitmap_copy(set, hwloc_topology_get_complete_cpuset(topology)); +#endif + return 0; +} + +/** \brief Get the hwloc OS device object corresponding to the + * OpenCL device for the given indexes. + * + * Return the OS device object describing the OpenCL device + * whose platform index is \p platform_index, + * and whose device index within this platform if \p device_index. + * Return NULL if there is none. + * + * The topology \p topology does not necessarily have to match the current + * machine. For instance the topology may be an XML import of a remote host. + * I/O devices detection and the OpenCL component must be enabled in the topology. + * + * \note The corresponding PCI device object can be obtained by looking + * at the OS device parent object (unless PCI devices are filtered out). + */ +static __hwloc_inline hwloc_obj_t +hwloc_opencl_get_device_osdev_by_index(hwloc_topology_t topology, + unsigned platform_index, unsigned device_index) +{ + unsigned x = (unsigned) -1, y = (unsigned) -1; + hwloc_obj_t osdev = NULL; + while ((osdev = hwloc_get_next_osdev(topology, osdev)) != NULL) { + if (HWLOC_OBJ_OSDEV_COPROC == osdev->attr->osdev.type + && osdev->name + && sscanf(osdev->name, "opencl%ud%u", &x, &y) == 2 + && platform_index == x && device_index == y) + return osdev; + } + return NULL; +} + +/** \brief Get the hwloc OS device object corresponding to OpenCL device \p deviceX. + * + * Use OpenCL device attributes to find the corresponding hwloc OS device object. + * Return NULL if there is none or if useful attributes are not available. + * + * This function currently only works on AMD OpenCL devices that support + * the CL_DEVICE_TOPOLOGY_AMD extension. hwloc_opencl_get_device_osdev_by_index() + * should be preferred whenever possible, i.e. when platform and device index + * are known. + * + * Topology \p topology and device \p device must match the local machine. + * I/O devices detection and the OpenCL component must be enabled in the topology. + * If not, the locality of the object may still be found using + * hwloc_opencl_get_device_cpuset(). + * + * \note This function cannot work if PCI devices are filtered out. + * + * \note The corresponding hwloc PCI device may be found by looking + * at the result parent pointer (unless PCI devices are filtered out). + */ +static __hwloc_inline hwloc_obj_t +hwloc_opencl_get_device_osdev(hwloc_topology_t topology __hwloc_attribute_unused, + cl_device_id device __hwloc_attribute_unused) +{ +#ifdef CL_DEVICE_TOPOLOGY_AMD + hwloc_obj_t osdev; + cl_device_topology_amd amdtopo; + cl_int clret; + + clret = clGetDeviceInfo(device, CL_DEVICE_TOPOLOGY_AMD, sizeof(amdtopo), &amdtopo, NULL); + if (CL_SUCCESS != clret) { + errno = EINVAL; + return NULL; + } + if (CL_DEVICE_TOPOLOGY_TYPE_PCIE_AMD != amdtopo.raw.type) { + errno = EINVAL; + return NULL; + } + + osdev = NULL; + while ((osdev = hwloc_get_next_osdev(topology, osdev)) != NULL) { + hwloc_obj_t pcidev = osdev->parent; + if (strncmp(osdev->name, "opencl", 6)) + continue; + if (pcidev + && pcidev->type == HWLOC_OBJ_PCI_DEVICE + && pcidev->attr->pcidev.domain == 0 + && pcidev->attr->pcidev.bus == amdtopo.pcie.bus + && pcidev->attr->pcidev.dev == amdtopo.pcie.device + && pcidev->attr->pcidev.func == amdtopo.pcie.function) + return osdev; + /* if PCI are filtered out, we need a info attr to match on */ + } + + return NULL; +#else + return NULL; +#endif +} + +/** @} */ + + +#ifdef __cplusplus +} /* extern "C" */ +#endif + + +#endif /* HWLOC_OPENCL_H */ diff --git a/src/3rdparty/hwloc/include/hwloc/openfabrics-verbs.h b/src/3rdparty/hwloc/include/hwloc/openfabrics-verbs.h new file mode 100644 index 000000000..174ab4a57 --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc/openfabrics-verbs.h @@ -0,0 +1,150 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2016 Inria. All rights reserved. + * Copyright © 2009-2010 Université Bordeaux + * Copyright © 2009-2011 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +/** \file + * \brief Macros to help interaction between hwloc and OpenFabrics + * verbs. + * + * Applications that use both hwloc and OpenFabrics verbs may want to + * include this file so as to get topology information for OpenFabrics + * hardware (InfiniBand, etc). + * + */ + +#ifndef HWLOC_OPENFABRICS_VERBS_H +#define HWLOC_OPENFABRICS_VERBS_H + +#include +#include +#ifdef HWLOC_LINUX_SYS +#include +#endif + +#include + + +#ifdef __cplusplus +extern "C" { +#endif + + +/** \defgroup hwlocality_openfabrics Interoperability with OpenFabrics + * + * This interface offers ways to retrieve topology information about + * OpenFabrics devices (InfiniBand, Omni-Path, usNIC, etc). + * + * @{ + */ + +/** \brief Get the CPU set of logical processors that are physically + * close to device \p ibdev. + * + * Return the CPU set describing the locality of the OpenFabrics + * device \p ibdev (InfiniBand, etc). + * + * Topology \p topology and device \p ibdev must match the local machine. + * I/O devices detection is not needed in the topology. + * + * The function only returns the locality of the device. + * If more information about the device is needed, OS objects should + * be used instead, see hwloc_ibv_get_device_osdev() + * and hwloc_ibv_get_device_osdev_by_name(). + * + * This function is currently only implemented in a meaningful way for + * Linux; other systems will simply get a full cpuset. + */ +static __hwloc_inline int +hwloc_ibv_get_device_cpuset(hwloc_topology_t topology __hwloc_attribute_unused, + struct ibv_device *ibdev, hwloc_cpuset_t set) +{ +#ifdef HWLOC_LINUX_SYS + /* If we're on Linux, use the verbs-provided sysfs mechanism to + get the local cpus */ +#define HWLOC_OPENFABRICS_VERBS_SYSFS_PATH_MAX 128 + char path[HWLOC_OPENFABRICS_VERBS_SYSFS_PATH_MAX]; + + if (!hwloc_topology_is_thissystem(topology)) { + errno = EINVAL; + return -1; + } + + sprintf(path, "/sys/class/infiniband/%s/device/local_cpus", + ibv_get_device_name(ibdev)); + if (hwloc_linux_read_path_as_cpumask(path, set) < 0 + || hwloc_bitmap_iszero(set)) + hwloc_bitmap_copy(set, hwloc_topology_get_complete_cpuset(topology)); +#else + /* Non-Linux systems simply get a full cpuset */ + hwloc_bitmap_copy(set, hwloc_topology_get_complete_cpuset(topology)); +#endif + return 0; +} + +/** \brief Get the hwloc OS device object corresponding to the OpenFabrics + * device named \p ibname. + * + * Return the OS device object describing the OpenFabrics device + * (InfiniBand, Omni-Path, usNIC, etc) whose name is \p ibname + * (mlx5_0, hfi1_0, usnic_0, qib0, etc). + * Returns NULL if there is none. + * The name \p ibname is usually obtained from ibv_get_device_name(). + * + * The topology \p topology does not necessarily have to match the current + * machine. For instance the topology may be an XML import of a remote host. + * I/O devices detection must be enabled in the topology. + * + * \note The corresponding PCI device object can be obtained by looking + * at the OS device parent object. + */ +static __hwloc_inline hwloc_obj_t +hwloc_ibv_get_device_osdev_by_name(hwloc_topology_t topology, + const char *ibname) +{ + hwloc_obj_t osdev = NULL; + while ((osdev = hwloc_get_next_osdev(topology, osdev)) != NULL) { + if (HWLOC_OBJ_OSDEV_OPENFABRICS == osdev->attr->osdev.type + && osdev->name && !strcmp(ibname, osdev->name)) + return osdev; + } + return NULL; +} + +/** \brief Get the hwloc OS device object corresponding to the OpenFabrics + * device \p ibdev. + * + * Return the OS device object describing the OpenFabrics device \p ibdev + * (InfiniBand, etc). Returns NULL if there is none. + * + * Topology \p topology and device \p ibdev must match the local machine. + * I/O devices detection must be enabled in the topology. + * If not, the locality of the object may still be found using + * hwloc_ibv_get_device_cpuset(). + * + * \note The corresponding PCI device object can be obtained by looking + * at the OS device parent object. + */ +static __hwloc_inline hwloc_obj_t +hwloc_ibv_get_device_osdev(hwloc_topology_t topology, + struct ibv_device *ibdev) +{ + if (!hwloc_topology_is_thissystem(topology)) { + errno = EINVAL; + return NULL; + } + return hwloc_ibv_get_device_osdev_by_name(topology, ibv_get_device_name(ibdev)); +} + +/** @} */ + + +#ifdef __cplusplus +} /* extern "C" */ +#endif + + +#endif /* HWLOC_OPENFABRICS_VERBS_H */ diff --git a/src/3rdparty/hwloc/include/hwloc/plugins.h b/src/3rdparty/hwloc/include/hwloc/plugins.h new file mode 100644 index 000000000..cb22000d4 --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc/plugins.h @@ -0,0 +1,542 @@ +/* + * Copyright © 2013-2017 Inria. All rights reserved. + * Copyright © 2016 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +#ifndef HWLOC_PLUGINS_H +#define HWLOC_PLUGINS_H + +/** \file + * \brief Public interface for building hwloc plugins. + */ + +struct hwloc_backend; + +#include +#ifdef HWLOC_INSIDE_PLUGIN +/* needed for hwloc_plugin_check_namespace() */ +#include +#endif + + + +/** \defgroup hwlocality_disc_components Components and Plugins: Discovery components + * @{ + */ + +/** \brief Discovery component type */ +typedef enum hwloc_disc_component_type_e { + /** \brief CPU-only discovery through the OS, or generic no-OS support. + * \hideinitializer */ + HWLOC_DISC_COMPONENT_TYPE_CPU = (1<<0), + + /** \brief xml or synthetic, + * platform-specific components such as bgq. + * Anything the discovers CPU and everything else. + * No misc backend is expected to complement a global component. + * \hideinitializer */ + HWLOC_DISC_COMPONENT_TYPE_GLOBAL = (1<<1), + + /** \brief OpenCL, Cuda, etc. + * \hideinitializer */ + HWLOC_DISC_COMPONENT_TYPE_MISC = (1<<2) +} hwloc_disc_component_type_t; + +/** \brief Discovery component structure + * + * This is the major kind of components, taking care of the discovery. + * They are registered by generic components, either statically-built or as plugins. + */ +struct hwloc_disc_component { + /** \brief Discovery component type */ + hwloc_disc_component_type_t type; + + /** \brief Name. + * If this component is built as a plugin, this name does not have to match the plugin filename. + */ + const char *name; + + /** \brief Component types to exclude, as an OR'ed set of ::hwloc_disc_component_type_e. + * + * For a GLOBAL component, this usually includes all other types (~0). + * + * Other components only exclude types that may bring conflicting + * topology information. MISC components should likely not be excluded + * since they usually bring non-primary additional information. + */ + unsigned excludes; + + /** \brief Instantiate callback to create a backend from the component. + * Parameters data1, data2, data3 are NULL except for components + * that have special enabling routines such as hwloc_topology_set_xml(). */ + struct hwloc_backend * (*instantiate)(struct hwloc_disc_component *component, const void *data1, const void *data2, const void *data3); + + /** \brief Component priority. + * Used to sort topology->components, higher priority first. + * Also used to decide between two components with the same name. + * + * Usual values are + * 50 for native OS (or platform) components, + * 45 for x86, + * 40 for no-OS fallback, + * 30 for global components (xml, synthetic), + * 20 for pci, + * 10 for other misc components (opencl etc.). + */ + unsigned priority; + + /** \brief Enabled by default. + * If unset, if will be disabled unless explicitly requested. + */ + unsigned enabled_by_default; + + /** \private Used internally to list components by priority on topology->components + * (the component structure is usually read-only, + * the core copies it before using this field for queueing) + */ + struct hwloc_disc_component * next; +}; + +/** @} */ + + + + +/** \defgroup hwlocality_disc_backends Components and Plugins: Discovery backends + * @{ + */ + +/** \brief Discovery backend structure + * + * A backend is the instantiation of a discovery component. + * When a component gets enabled for a topology, + * its instantiate() callback creates a backend. + * + * hwloc_backend_alloc() initializes all fields to default values + * that the component may change (except "component" and "next") + * before enabling the backend with hwloc_backend_enable(). + */ +struct hwloc_backend { + /** \private Reserved for the core, set by hwloc_backend_alloc() */ + struct hwloc_disc_component * component; + /** \private Reserved for the core, set by hwloc_backend_enable() */ + struct hwloc_topology * topology; + /** \private Reserved for the core. Set to 1 if forced through envvar, 0 otherwise. */ + int envvar_forced; + /** \private Reserved for the core. Used internally to list backends topology->backends. */ + struct hwloc_backend * next; + + /** \brief Backend flags, currently always 0. */ + unsigned long flags; + + /** \brief Backend-specific 'is_thissystem' property. + * Set to 0 or 1 if the backend should enforce the thissystem flag when it gets enabled. + * Set to -1 if the backend doesn't care (default). */ + int is_thissystem; + + /** \brief Backend private data, or NULL if none. */ + void * private_data; + /** \brief Callback for freeing the private_data. + * May be NULL. + */ + void (*disable)(struct hwloc_backend *backend); + + /** \brief Main discovery callback. + * returns -1 on error, either because it couldn't add its objects ot the existing topology, + * or because of an actual discovery/gathering failure. + * May be NULL. + */ + int (*discover)(struct hwloc_backend *backend); + + /** \brief Callback used by the PCI backend to retrieve the locality of a PCI object from the OS/cpu backend. + * May be NULL. */ + int (*get_pci_busid_cpuset)(struct hwloc_backend *backend, struct hwloc_pcidev_attr_s *busid, hwloc_bitmap_t cpuset); +}; + +/** \brief Allocate a backend structure, set good default values, initialize backend->component and topology, etc. + * The caller will then modify whatever needed, and call hwloc_backend_enable(). + */ +HWLOC_DECLSPEC struct hwloc_backend * hwloc_backend_alloc(struct hwloc_disc_component *component); + +/** \brief Enable a previously allocated and setup backend. */ +HWLOC_DECLSPEC int hwloc_backend_enable(struct hwloc_topology *topology, struct hwloc_backend *backend); + +/** @} */ + + + + +/** \defgroup hwlocality_generic_components Components and Plugins: Generic components + * @{ + */ + +/** \brief Generic component type */ +typedef enum hwloc_component_type_e { + /** \brief The data field must point to a struct hwloc_disc_component. */ + HWLOC_COMPONENT_TYPE_DISC, + + /** \brief The data field must point to a struct hwloc_xml_component. */ + HWLOC_COMPONENT_TYPE_XML +} hwloc_component_type_t; + +/** \brief Generic component structure + * + * Generic components structure, either statically listed by configure in static-components.h + * or dynamically loaded as a plugin. + */ +struct hwloc_component { + /** \brief Component ABI version, set to ::HWLOC_COMPONENT_ABI */ + unsigned abi; + + /** \brief Process-wide component initialization callback. + * + * This optional callback is called when the component is registered + * to the hwloc core (after loading the plugin). + * + * When the component is built as a plugin, this callback + * should call hwloc_check_plugin_namespace() + * and return an negative error code on error. + * + * \p flags is always 0 for now. + * + * \return 0 on success, or a negative code on error. + * + * \note If the component uses ltdl for loading its own plugins, + * it should load/unload them only in init() and finalize(), + * to avoid race conditions with hwloc's use of ltdl. + */ + int (*init)(unsigned long flags); + + /** \brief Process-wide component termination callback. + * + * This optional callback is called after unregistering the component + * from the hwloc core (before unloading the plugin). + * + * \p flags is always 0 for now. + * + * \note If the component uses ltdl for loading its own plugins, + * it should load/unload them only in init() and finalize(), + * to avoid race conditions with hwloc's use of ltdl. + */ + void (*finalize)(unsigned long flags); + + /** \brief Component type */ + hwloc_component_type_t type; + + /** \brief Component flags, unused for now */ + unsigned long flags; + + /** \brief Component data, pointing to a struct hwloc_disc_component or struct hwloc_xml_component. */ + void * data; +}; + +/** @} */ + + + + +/** \defgroup hwlocality_components_core_funcs Components and Plugins: Core functions to be used by components + * @{ + */ + +/** \brief Add an object to the topology. + * + * It is sorted along the tree of other objects according to the inclusion of + * cpusets, to eventually be added as a child of the smallest object including + * this object. + * + * If the cpuset is empty, the type of the object (and maybe some attributes) + * must be enough to find where to insert the object. This is especially true + * for NUMA nodes with memory and no CPUs. + * + * The given object should not have children. + * + * This shall only be called before levels are built. + * + * In case of error, hwloc_report_os_error() is called. + * + * The caller should check whether the object type is filtered-out before calling this function. + * + * The topology cpuset/nodesets will be enlarged to include the object sets. + * + * Returns the object on success. + * Returns NULL and frees obj on error. + * Returns another object and frees obj if it was merged with an identical pre-existing object. + */ +HWLOC_DECLSPEC struct hwloc_obj *hwloc_insert_object_by_cpuset(struct hwloc_topology *topology, hwloc_obj_t obj); + +/** \brief Type of error callbacks during object insertion */ +typedef void (*hwloc_report_error_t)(const char * msg, int line); +/** \brief Report an insertion error from a backend */ +HWLOC_DECLSPEC void hwloc_report_os_error(const char * msg, int line); +/** \brief Check whether insertion errors are hidden */ +HWLOC_DECLSPEC int hwloc_hide_errors(void); + +/** \brief Add an object to the topology and specify which error callback to use. + * + * This function is similar to hwloc_insert_object_by_cpuset() but it allows specifying + * where to start insertion from (if \p root is NULL, the topology root object is used), + * and specifying the error callback. + */ +HWLOC_DECLSPEC struct hwloc_obj *hwloc__insert_object_by_cpuset(struct hwloc_topology *topology, hwloc_obj_t root, hwloc_obj_t obj, hwloc_report_error_t report_error); + +/** \brief Insert an object somewhere in the topology. + * + * It is added as the last child of the given parent. + * The cpuset is completely ignored, so strange objects such as I/O devices should + * preferably be inserted with this. + * + * When used for "normal" children with cpusets (when importing from XML + * when duplicating a topology), the caller should make sure that: + * - children are inserted in order, + * - children cpusets do not intersect. + * + * The given object may have normal, I/O or Misc children, as long as they are in order as well. + * These children must have valid parent and next_sibling pointers. + * + * The caller should check whether the object type is filtered-out before calling this function. + */ +HWLOC_DECLSPEC void hwloc_insert_object_by_parent(struct hwloc_topology *topology, hwloc_obj_t parent, hwloc_obj_t obj); + +/** \brief Allocate and initialize an object of the given type and physical index. + * + * If \p os_index is unknown or irrelevant, use \c HWLOC_UNKNOWN_INDEX. + */ +HWLOC_DECLSPEC hwloc_obj_t hwloc_alloc_setup_object(hwloc_topology_t topology, hwloc_obj_type_t type, unsigned os_index); + +/** \brief Setup object cpusets/nodesets by OR'ing its children. + * + * Used when adding an object late in the topology. + * Will update the new object by OR'ing all its new children sets. + * + * Used when PCI backend adds a hostbridge parent, when distances + * add a new Group, etc. + */ +HWLOC_DECLSPEC int hwloc_obj_add_children_sets(hwloc_obj_t obj); + +/** \brief Request a reconnection of children and levels in the topology. + * + * May be used by backends during discovery if they need arrays or lists + * of object within levels or children to be fully connected. + * + * \p flags is currently unused, must 0. + */ +HWLOC_DECLSPEC int hwloc_topology_reconnect(hwloc_topology_t topology, unsigned long flags __hwloc_attribute_unused); + +/** \brief Make sure that plugins can lookup core symbols. + * + * This is a sanity check to avoid lazy-lookup failures when libhwloc + * is loaded within a plugin, and later tries to load its own plugins. + * This may fail (and abort the program) if libhwloc symbols are in a + * private namespace. + * + * \return 0 on success. + * \return -1 if the plugin cannot be successfully loaded. The caller + * plugin init() callback should return a negative error code as well. + * + * Plugins should call this function in their init() callback to avoid + * later crashes if lazy symbol resolution is used by the upper layer that + * loaded hwloc (e.g. OpenCL implementations using dlopen with RTLD_LAZY). + * + * \note The build system must define HWLOC_INSIDE_PLUGIN if and only if + * building the caller as a plugin. + * + * \note This function should remain inline so plugins can call it even + * when they cannot find libhwloc symbols. + */ +static __hwloc_inline int +hwloc_plugin_check_namespace(const char *pluginname __hwloc_attribute_unused, const char *symbol __hwloc_attribute_unused) +{ +#ifdef HWLOC_INSIDE_PLUGIN + lt_dlhandle handle; + void *sym; + handle = lt_dlopen(NULL); + if (!handle) + /* cannot check, assume things will work */ + return 0; + sym = lt_dlsym(handle, symbol); + lt_dlclose(handle); + if (!sym) { + static int verboseenv_checked = 0; + static int verboseenv_value = 0; + if (!verboseenv_checked) { + const char *verboseenv = getenv("HWLOC_PLUGINS_VERBOSE"); + verboseenv_value = verboseenv ? atoi(verboseenv) : 0; + verboseenv_checked = 1; + } + if (verboseenv_value) + fprintf(stderr, "Plugin `%s' disabling itself because it cannot find the `%s' core symbol.\n", + pluginname, symbol); + return -1; + } +#endif /* HWLOC_INSIDE_PLUGIN */ + return 0; +} + +/** @} */ + + + + +/** \defgroup hwlocality_components_filtering Components and Plugins: Filtering objects + * @{ + */ + +/** \brief Check whether the given PCI device classid is important. + * + * \return 1 if important, 0 otherwise. + */ +static __hwloc_inline int +hwloc_filter_check_pcidev_subtype_important(unsigned classid) +{ + unsigned baseclass = classid >> 8; + return (baseclass == 0x03 /* PCI_BASE_CLASS_DISPLAY */ + || baseclass == 0x02 /* PCI_BASE_CLASS_NETWORK */ + || baseclass == 0x01 /* PCI_BASE_CLASS_STORAGE */ + || baseclass == 0x0b /* PCI_BASE_CLASS_PROCESSOR */ + || classid == 0x0c04 /* PCI_CLASS_SERIAL_FIBER */ + || classid == 0x0c06 /* PCI_CLASS_SERIAL_INFINIBAND */ + || baseclass == 0x12 /* Processing Accelerators */); +} + +/** \brief Check whether the given OS device subtype is important. + * + * \return 1 if important, 0 otherwise. + */ +static __hwloc_inline int +hwloc_filter_check_osdev_subtype_important(hwloc_obj_osdev_type_t subtype) +{ + return (subtype != HWLOC_OBJ_OSDEV_DMA); +} + +/** \brief Check whether a non-I/O object type should be filtered-out. + * + * Cannot be used for I/O objects. + * + * \return 1 if the object type should be kept, 0 otherwise. + */ +static __hwloc_inline int +hwloc_filter_check_keep_object_type(hwloc_topology_t topology, hwloc_obj_type_t type) +{ + enum hwloc_type_filter_e filter = HWLOC_TYPE_FILTER_KEEP_NONE; + hwloc_topology_get_type_filter(topology, type, &filter); + assert(filter != HWLOC_TYPE_FILTER_KEEP_IMPORTANT); /* IMPORTANT only used for I/O */ + return filter == HWLOC_TYPE_FILTER_KEEP_NONE ? 0 : 1; +} + +/** \brief Check whether the given object should be filtered-out. + * + * \return 1 if the object type should be kept, 0 otherwise. + */ +static __hwloc_inline int +hwloc_filter_check_keep_object(hwloc_topology_t topology, hwloc_obj_t obj) +{ + hwloc_obj_type_t type = obj->type; + enum hwloc_type_filter_e filter = HWLOC_TYPE_FILTER_KEEP_NONE; + hwloc_topology_get_type_filter(topology, type, &filter); + if (filter == HWLOC_TYPE_FILTER_KEEP_NONE) + return 0; + if (filter == HWLOC_TYPE_FILTER_KEEP_IMPORTANT) { + if (type == HWLOC_OBJ_PCI_DEVICE) + return hwloc_filter_check_pcidev_subtype_important(obj->attr->pcidev.class_id); + if (type == HWLOC_OBJ_OS_DEVICE) + return hwloc_filter_check_osdev_subtype_important(obj->attr->osdev.type); + } + return 1; +} + +/** @} */ + + + + +/** \defgroup hwlocality_components_pcidisc Components and Plugins: helpers for PCI discovery + * @{ + */ + +/** \brief Return the offset of the given capability in the PCI config space buffer + * + * This function requires a 256-bytes config space. Unknown/unavailable bytes should be set to 0xff. + */ +HWLOC_DECLSPEC unsigned hwloc_pcidisc_find_cap(const unsigned char *config, unsigned cap); + +/** \brief Fill linkspeed by reading the PCI config space where PCI_CAP_ID_EXP is at position offset. + * + * Needs 20 bytes of EXP capability block starting at offset in the config space + * for registers up to link status. + */ +HWLOC_DECLSPEC int hwloc_pcidisc_find_linkspeed(const unsigned char *config, unsigned offset, float *linkspeed); + +/** \brief Return the hwloc object type (PCI device or Bridge) for the given class and configuration space. + * + * This function requires 16 bytes of common configuration header at the beginning of config. + */ +HWLOC_DECLSPEC hwloc_obj_type_t hwloc_pcidisc_check_bridge_type(unsigned device_class, const unsigned char *config); + +/** \brief Fills the attributes of the given PCI bridge using the given PCI config space. + * + * This function requires 32 bytes of common configuration header at the beginning of config. + * + * Returns -1 and destroys /p obj if bridge fields are invalid. + */ +HWLOC_DECLSPEC int hwloc_pcidisc_setup_bridge_attr(hwloc_obj_t obj, const unsigned char *config); + +/** \brief Insert a PCI object in the given PCI tree by looking at PCI bus IDs. + * + * If \p treep points to \c NULL, the new object is inserted there. + */ +HWLOC_DECLSPEC void hwloc_pcidisc_tree_insert_by_busid(struct hwloc_obj **treep, struct hwloc_obj *obj); + +/** \brief Add some hostbridges on top of the given tree of PCI objects and attach them to the topology. + * + * For now, they will be attached to the root object. The core will move them to their actual PCI + * locality using hwloc_pci_belowroot_apply_locality() at the end of the discovery. + * + * In the meantime, other backends lookup PCI objects or localities (for instance to attach OS devices) + * by using hwloc_pcidisc_find_by_busid() or hwloc_pcidisc_find_busid_parent(). + */ +HWLOC_DECLSPEC int hwloc_pcidisc_tree_attach(struct hwloc_topology *topology, struct hwloc_obj *tree); + +/** @} */ + + + + +/** \defgroup hwlocality_components_pcifind Components and Plugins: finding PCI objects during other discoveries + * @{ + */ + +/** \brief Find the PCI object that matches the bus ID. + * + * To be used after a PCI backend added PCI devices with hwloc_pcidisc_tree_attach() + * and before the core moves them to their actual location with hwloc_pci_belowroot_apply_locality(). + * + * If no exactly matching object is found, return the container bridge if any, or NULL. + * + * On failure, it may be possible to find the PCI locality (instead of the PCI device) + * by calling hwloc_pcidisc_find_busid_parent(). + * + * \note This is semantically identical to hwloc_get_pcidev_by_busid() which only works + * after the topology is fully loaded. + */ +HWLOC_DECLSPEC struct hwloc_obj * hwloc_pcidisc_find_by_busid(struct hwloc_topology *topology, unsigned domain, unsigned bus, unsigned dev, unsigned func); + +/** \brief Find the normal parent of a PCI bus ID. + * + * Look at PCI affinity to find out where the given PCI bus ID should be attached. + * + * This function should be used to attach an I/O device directly under a normal + * (non-I/O) object, instead of below a PCI object. + * It is usually used by backends when hwloc_pcidisc_find_by_busid() failed + * to find the hwloc object corresponding to this bus ID, for instance because + * PCI discovery is not supported on this platform. + */ +HWLOC_DECLSPEC struct hwloc_obj * hwloc_pcidisc_find_busid_parent(struct hwloc_topology *topology, unsigned domain, unsigned bus, unsigned dev, unsigned func); + +/** @} */ + + + + +#endif /* HWLOC_PLUGINS_H */ diff --git a/src/3rdparty/hwloc/include/hwloc/rename.h b/src/3rdparty/hwloc/include/hwloc/rename.h new file mode 100644 index 000000000..7cef1b2e8 --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc/rename.h @@ -0,0 +1,765 @@ +/* + * Copyright © 2009-2011 Cisco Systems, Inc. All rights reserved. + * Copyright © 2010-2018 Inria. All rights reserved. + * See COPYING in top-level directory. + */ + +#ifndef HWLOC_RENAME_H +#define HWLOC_RENAME_H + +#include + + +#ifdef __cplusplus +extern "C" { +#endif + + +/* Only enact these defines if we're actually renaming the symbols + (i.e., avoid trying to have no-op defines if we're *not* + renaming). */ + +#if HWLOC_SYM_TRANSFORM + +/* Use a preprocessor two-step in order to get the prefixing right. + Make 2 macros: HWLOC_NAME and HWLOC_NAME_CAPS for renaming + things. */ + +#define HWLOC_MUNGE_NAME(a, b) HWLOC_MUNGE_NAME2(a, b) +#define HWLOC_MUNGE_NAME2(a, b) a ## b +#define HWLOC_NAME(name) HWLOC_MUNGE_NAME(HWLOC_SYM_PREFIX, hwloc_ ## name) +#define HWLOC_NAME_CAPS(name) HWLOC_MUNGE_NAME(HWLOC_SYM_PREFIX_CAPS, hwloc_ ## name) + +/* Now define all the "real" names to be the prefixed names. This + allows us to use the real names throughout the code base (i.e., + "hwloc_"); the preprocessor will adjust to have the prefixed + name under the covers. */ + +/* Names from hwloc.h */ + +#define hwloc_get_api_version HWLOC_NAME(get_api_version) + +#define hwloc_topology HWLOC_NAME(topology) +#define hwloc_topology_t HWLOC_NAME(topology_t) + +#define hwloc_cpuset_t HWLOC_NAME(cpuset_t) +#define hwloc_const_cpuset_t HWLOC_NAME(const_cpuset_t) +#define hwloc_nodeset_t HWLOC_NAME(nodeset_t) +#define hwloc_const_nodeset_t HWLOC_NAME(const_nodeset_t) + +#define HWLOC_OBJ_MACHINE HWLOC_NAME_CAPS(OBJ_MACHINE) +#define HWLOC_OBJ_NUMANODE HWLOC_NAME_CAPS(OBJ_NUMANODE) +#define HWLOC_OBJ_PACKAGE HWLOC_NAME_CAPS(OBJ_PACKAGE) +#define HWLOC_OBJ_CORE HWLOC_NAME_CAPS(OBJ_CORE) +#define HWLOC_OBJ_PU HWLOC_NAME_CAPS(OBJ_PU) +#define HWLOC_OBJ_L1CACHE HWLOC_NAME_CAPS(OBJ_L1CACHE) +#define HWLOC_OBJ_L2CACHE HWLOC_NAME_CAPS(OBJ_L2CACHE) +#define HWLOC_OBJ_L3CACHE HWLOC_NAME_CAPS(OBJ_L3CACHE) +#define HWLOC_OBJ_L4CACHE HWLOC_NAME_CAPS(OBJ_L4CACHE) +#define HWLOC_OBJ_L5CACHE HWLOC_NAME_CAPS(OBJ_L5CACHE) +#define HWLOC_OBJ_L1ICACHE HWLOC_NAME_CAPS(OBJ_L1ICACHE) +#define HWLOC_OBJ_L2ICACHE HWLOC_NAME_CAPS(OBJ_L2ICACHE) +#define HWLOC_OBJ_L3ICACHE HWLOC_NAME_CAPS(OBJ_L3ICACHE) +#define HWLOC_OBJ_MISC HWLOC_NAME_CAPS(OBJ_MISC) +#define HWLOC_OBJ_GROUP HWLOC_NAME_CAPS(OBJ_GROUP) +#define HWLOC_OBJ_BRIDGE HWLOC_NAME_CAPS(OBJ_BRIDGE) +#define HWLOC_OBJ_PCI_DEVICE HWLOC_NAME_CAPS(OBJ_PCI_DEVICE) +#define HWLOC_OBJ_OS_DEVICE HWLOC_NAME_CAPS(OBJ_OS_DEVICE) +#define HWLOC_OBJ_TYPE_MAX HWLOC_NAME_CAPS(OBJ_TYPE_MAX) +#define hwloc_obj_type_t HWLOC_NAME(obj_type_t) + +#define hwloc_obj_cache_type_e HWLOC_NAME(obj_cache_type_e) +#define hwloc_obj_cache_type_t HWLOC_NAME(obj_cache_type_t) +#define HWLOC_OBJ_CACHE_UNIFIED HWLOC_NAME_CAPS(OBJ_CACHE_UNIFIED) +#define HWLOC_OBJ_CACHE_DATA HWLOC_NAME_CAPS(OBJ_CACHE_DATA) +#define HWLOC_OBJ_CACHE_INSTRUCTION HWLOC_NAME_CAPS(OBJ_CACHE_INSTRUCTION) + +#define hwloc_obj_bridge_type_e HWLOC_NAME(obj_bridge_type_e) +#define hwloc_obj_bridge_type_t HWLOC_NAME(obj_bridge_type_t) +#define HWLOC_OBJ_BRIDGE_HOST HWLOC_NAME_CAPS(OBJ_BRIDGE_HOST) +#define HWLOC_OBJ_BRIDGE_PCI HWLOC_NAME_CAPS(OBJ_BRIDGE_PCI) + +#define hwloc_obj_osdev_type_e HWLOC_NAME(obj_osdev_type_e) +#define hwloc_obj_osdev_type_t HWLOC_NAME(obj_osdev_type_t) +#define HWLOC_OBJ_OSDEV_BLOCK HWLOC_NAME_CAPS(OBJ_OSDEV_BLOCK) +#define HWLOC_OBJ_OSDEV_GPU HWLOC_NAME_CAPS(OBJ_OSDEV_GPU) +#define HWLOC_OBJ_OSDEV_NETWORK HWLOC_NAME_CAPS(OBJ_OSDEV_NETWORK) +#define HWLOC_OBJ_OSDEV_OPENFABRICS HWLOC_NAME_CAPS(OBJ_OSDEV_OPENFABRICS) +#define HWLOC_OBJ_OSDEV_DMA HWLOC_NAME_CAPS(OBJ_OSDEV_DMA) +#define HWLOC_OBJ_OSDEV_COPROC HWLOC_NAME_CAPS(OBJ_OSDEV_COPROC) + +#define hwloc_compare_types HWLOC_NAME(compare_types) + +#define hwloc_compare_types_e HWLOC_NAME(compare_types_e) +#define HWLOC_TYPE_UNORDERED HWLOC_NAME_CAPS(TYPE_UNORDERED) + +#define hwloc_obj HWLOC_NAME(obj) +#define hwloc_obj_t HWLOC_NAME(obj_t) + +#define hwloc_info_s HWLOC_NAME(info_s) + +#define hwloc_obj_attr_u HWLOC_NAME(obj_attr_u) +#define hwloc_numanode_attr_s HWLOC_NAME(numanode_attr_s) +#define hwloc_memory_page_type_s HWLOC_NAME(memory_page_type_s) +#define hwloc_cache_attr_s HWLOC_NAME(cache_attr_s) +#define hwloc_group_attr_s HWLOC_NAME(group_attr_s) +#define hwloc_pcidev_attr_s HWLOC_NAME(pcidev_attr_s) +#define hwloc_bridge_attr_s HWLOC_NAME(bridge_attr_s) +#define hwloc_osdev_attr_s HWLOC_NAME(osdev_attr_s) + +#define hwloc_topology_init HWLOC_NAME(topology_init) +#define hwloc_topology_load HWLOC_NAME(topology_load) +#define hwloc_topology_destroy HWLOC_NAME(topology_destroy) +#define hwloc_topology_dup HWLOC_NAME(topology_dup) +#define hwloc_topology_abi_check HWLOC_NAME(topology_abi_check) +#define hwloc_topology_check HWLOC_NAME(topology_check) + +#define hwloc_topology_flags_e HWLOC_NAME(topology_flags_e) + +#define HWLOC_TOPOLOGY_FLAG_WHOLE_SYSTEM HWLOC_NAME_CAPS(TOPOLOGY_FLAG_WHOLE_SYSTEM) +#define HWLOC_TOPOLOGY_FLAG_IS_THISSYSTEM HWLOC_NAME_CAPS(TOPOLOGY_FLAG_IS_THISSYSTEM) +#define HWLOC_TOPOLOGY_FLAG_THISSYSTEM_ALLOWED_RESOURCES HWLOC_NAME_CAPS(TOPOLOGY_FLAG_THISSYSTEM_ALLOWED_RESOURCES) + +#define hwloc_topology_set_pid HWLOC_NAME(topology_set_pid) +#define hwloc_topology_set_synthetic HWLOC_NAME(topology_set_synthetic) +#define hwloc_topology_set_xml HWLOC_NAME(topology_set_xml) +#define hwloc_topology_set_xmlbuffer HWLOC_NAME(topology_set_xmlbuffer) + +#define hwloc_topology_set_flags HWLOC_NAME(topology_set_flags) +#define hwloc_topology_is_thissystem HWLOC_NAME(topology_is_thissystem) +#define hwloc_topology_get_flags HWLOC_NAME(topology_get_flags) +#define hwloc_topology_discovery_support HWLOC_NAME(topology_discovery_support) +#define hwloc_topology_cpubind_support HWLOC_NAME(topology_cpubind_support) +#define hwloc_topology_membind_support HWLOC_NAME(topology_membind_support) +#define hwloc_topology_support HWLOC_NAME(topology_support) +#define hwloc_topology_get_support HWLOC_NAME(topology_get_support) + +#define hwloc_type_filter_e HWLOC_NAME(type_filter_e) +#define HWLOC_TYPE_FILTER_KEEP_ALL HWLOC_NAME_CAPS(TYPE_FILTER_KEEP_ALL) +#define HWLOC_TYPE_FILTER_KEEP_NONE HWLOC_NAME_CAPS(TYPE_FILTER_KEEP_NONE) +#define HWLOC_TYPE_FILTER_KEEP_STRUCTURE HWLOC_NAME_CAPS(TYPE_FILTER_KEEP_STRUCTURE) +#define HWLOC_TYPE_FILTER_KEEP_IMPORTANT HWLOC_NAME_CAPS(TYPE_FILTER_KEEP_IMPORTANT) +#define hwloc_topology_set_type_filter HWLOC_NAME(topology_set_type_filter) +#define hwloc_topology_get_type_filter HWLOC_NAME(topology_get_type_filter) +#define hwloc_topology_set_all_types_filter HWLOC_NAME(topology_set_all_types_filter) +#define hwloc_topology_set_cache_types_filter HWLOC_NAME(topology_set_cache_types_filter) +#define hwloc_topology_set_icache_types_filter HWLOC_NAME(topology_set_icache_types_filter) +#define hwloc_topology_set_io_types_filter HWLOC_NAME(topology_set_io_types_filter) + +#define hwloc_topology_set_userdata HWLOC_NAME(topology_set_userdata) +#define hwloc_topology_get_userdata HWLOC_NAME(topology_get_userdata) + +#define hwloc_restrict_flags_e HWLOC_NAME(restrict_flags_e) +#define HWLOC_RESTRICT_FLAG_REMOVE_CPULESS HWLOC_NAME_CAPS(RESTRICT_FLAG_REMOVE_CPULESS) +#define HWLOC_RESTRICT_FLAG_ADAPT_MISC HWLOC_NAME_CAPS(RESTRICT_FLAG_ADAPT_MISC) +#define HWLOC_RESTRICT_FLAG_ADAPT_IO HWLOC_NAME_CAPS(RESTRICT_FLAG_ADAPT_IO) +#define hwloc_topology_restrict HWLOC_NAME(topology_restrict) + +#define hwloc_topology_insert_misc_object HWLOC_NAME(topology_insert_misc_object) +#define hwloc_topology_alloc_group_object HWLOC_NAME(topology_alloc_group_object) +#define hwloc_topology_insert_group_object HWLOC_NAME(topology_insert_group_object) +#define hwloc_obj_add_other_obj_sets HWLOC_NAME(obj_add_other_obj_sets) + +#define hwloc_topology_get_depth HWLOC_NAME(topology_get_depth) +#define hwloc_get_type_depth HWLOC_NAME(get_type_depth) +#define hwloc_get_memory_parents_depth HWLOC_NAME(get_memory_parents_depth) + +#define hwloc_get_type_depth_e HWLOC_NAME(get_type_depth_e) +#define HWLOC_TYPE_DEPTH_UNKNOWN HWLOC_NAME_CAPS(TYPE_DEPTH_UNKNOWN) +#define HWLOC_TYPE_DEPTH_MULTIPLE HWLOC_NAME_CAPS(TYPE_DEPTH_MULTIPLE) +#define HWLOC_TYPE_DEPTH_BRIDGE HWLOC_NAME_CAPS(TYPE_DEPTH_BRIDGE) +#define HWLOC_TYPE_DEPTH_PCI_DEVICE HWLOC_NAME_CAPS(TYPE_DEPTH_PCI_DEVICE) +#define HWLOC_TYPE_DEPTH_OS_DEVICE HWLOC_NAME_CAPS(TYPE_DEPTH_OS_DEVICE) +#define HWLOC_TYPE_DEPTH_MISC HWLOC_NAME_CAPS(TYPE_DEPTH_MISC) +#define HWLOC_TYPE_DEPTH_NUMANODE HWLOC_NAME_CAPS(TYPE_DEPTH_NUMANODE) + +#define hwloc_get_depth_type HWLOC_NAME(get_depth_type) +#define hwloc_get_nbobjs_by_depth HWLOC_NAME(get_nbobjs_by_depth) +#define hwloc_get_nbobjs_by_type HWLOC_NAME(get_nbobjs_by_type) + +#define hwloc_get_obj_by_depth HWLOC_NAME(get_obj_by_depth ) +#define hwloc_get_obj_by_type HWLOC_NAME(get_obj_by_type ) + +#define hwloc_obj_type_string HWLOC_NAME(obj_type_string ) +#define hwloc_obj_type_snprintf HWLOC_NAME(obj_type_snprintf ) +#define hwloc_obj_attr_snprintf HWLOC_NAME(obj_attr_snprintf ) +#define hwloc_type_sscanf HWLOC_NAME(type_sscanf) +#define hwloc_type_sscanf_as_depth HWLOC_NAME(type_sscanf_as_depth) + +#define hwloc_obj_get_info_by_name HWLOC_NAME(obj_get_info_by_name) +#define hwloc_obj_add_info HWLOC_NAME(obj_add_info) + +#define HWLOC_CPUBIND_PROCESS HWLOC_NAME_CAPS(CPUBIND_PROCESS) +#define HWLOC_CPUBIND_THREAD HWLOC_NAME_CAPS(CPUBIND_THREAD) +#define HWLOC_CPUBIND_STRICT HWLOC_NAME_CAPS(CPUBIND_STRICT) +#define HWLOC_CPUBIND_NOMEMBIND HWLOC_NAME_CAPS(CPUBIND_NOMEMBIND) + +#define hwloc_cpubind_flags_t HWLOC_NAME(cpubind_flags_t) + +#define hwloc_set_cpubind HWLOC_NAME(set_cpubind) +#define hwloc_get_cpubind HWLOC_NAME(get_cpubind) +#define hwloc_set_proc_cpubind HWLOC_NAME(set_proc_cpubind) +#define hwloc_get_proc_cpubind HWLOC_NAME(get_proc_cpubind) +#define hwloc_set_thread_cpubind HWLOC_NAME(set_thread_cpubind) +#define hwloc_get_thread_cpubind HWLOC_NAME(get_thread_cpubind) + +#define hwloc_get_last_cpu_location HWLOC_NAME(get_last_cpu_location) +#define hwloc_get_proc_last_cpu_location HWLOC_NAME(get_proc_last_cpu_location) + +#define HWLOC_MEMBIND_DEFAULT HWLOC_NAME_CAPS(MEMBIND_DEFAULT) +#define HWLOC_MEMBIND_FIRSTTOUCH HWLOC_NAME_CAPS(MEMBIND_FIRSTTOUCH) +#define HWLOC_MEMBIND_BIND HWLOC_NAME_CAPS(MEMBIND_BIND) +#define HWLOC_MEMBIND_INTERLEAVE HWLOC_NAME_CAPS(MEMBIND_INTERLEAVE) +#define HWLOC_MEMBIND_NEXTTOUCH HWLOC_NAME_CAPS(MEMBIND_NEXTTOUCH) +#define HWLOC_MEMBIND_MIXED HWLOC_NAME_CAPS(MEMBIND_MIXED) + +#define hwloc_membind_policy_t HWLOC_NAME(membind_policy_t) + +#define HWLOC_MEMBIND_PROCESS HWLOC_NAME_CAPS(MEMBIND_PROCESS) +#define HWLOC_MEMBIND_THREAD HWLOC_NAME_CAPS(MEMBIND_THREAD) +#define HWLOC_MEMBIND_STRICT HWLOC_NAME_CAPS(MEMBIND_STRICT) +#define HWLOC_MEMBIND_MIGRATE HWLOC_NAME_CAPS(MEMBIND_MIGRATE) +#define HWLOC_MEMBIND_NOCPUBIND HWLOC_NAME_CAPS(MEMBIND_NOCPUBIND) +#define HWLOC_MEMBIND_BYNODESET HWLOC_NAME_CAPS(MEMBIND_BYNODESET) + +#define hwloc_membind_flags_t HWLOC_NAME(membind_flags_t) + +#define hwloc_set_membind HWLOC_NAME(set_membind) +#define hwloc_get_membind HWLOC_NAME(get_membind) +#define hwloc_set_proc_membind HWLOC_NAME(set_proc_membind) +#define hwloc_get_proc_membind HWLOC_NAME(get_proc_membind) +#define hwloc_set_area_membind HWLOC_NAME(set_area_membind) +#define hwloc_get_area_membind HWLOC_NAME(get_area_membind) +#define hwloc_get_area_memlocation HWLOC_NAME(get_area_memlocation) +#define hwloc_alloc_membind HWLOC_NAME(alloc_membind) +#define hwloc_alloc HWLOC_NAME(alloc) +#define hwloc_free HWLOC_NAME(free) + +#define hwloc_get_non_io_ancestor_obj HWLOC_NAME(get_non_io_ancestor_obj) +#define hwloc_get_next_pcidev HWLOC_NAME(get_next_pcidev) +#define hwloc_get_pcidev_by_busid HWLOC_NAME(get_pcidev_by_busid) +#define hwloc_get_pcidev_by_busidstring HWLOC_NAME(get_pcidev_by_busidstring) +#define hwloc_get_next_osdev HWLOC_NAME(get_next_osdev) +#define hwloc_get_next_bridge HWLOC_NAME(get_next_bridge) +#define hwloc_bridge_covers_pcibus HWLOC_NAME(bridge_covers_pcibus) + +/* hwloc/bitmap.h */ + +#define hwloc_bitmap_s HWLOC_NAME(bitmap_s) +#define hwloc_bitmap_t HWLOC_NAME(bitmap_t) +#define hwloc_const_bitmap_t HWLOC_NAME(const_bitmap_t) + +#define hwloc_bitmap_alloc HWLOC_NAME(bitmap_alloc) +#define hwloc_bitmap_alloc_full HWLOC_NAME(bitmap_alloc_full) +#define hwloc_bitmap_free HWLOC_NAME(bitmap_free) +#define hwloc_bitmap_dup HWLOC_NAME(bitmap_dup) +#define hwloc_bitmap_copy HWLOC_NAME(bitmap_copy) +#define hwloc_bitmap_snprintf HWLOC_NAME(bitmap_snprintf) +#define hwloc_bitmap_asprintf HWLOC_NAME(bitmap_asprintf) +#define hwloc_bitmap_sscanf HWLOC_NAME(bitmap_sscanf) +#define hwloc_bitmap_list_snprintf HWLOC_NAME(bitmap_list_snprintf) +#define hwloc_bitmap_list_asprintf HWLOC_NAME(bitmap_list_asprintf) +#define hwloc_bitmap_list_sscanf HWLOC_NAME(bitmap_list_sscanf) +#define hwloc_bitmap_taskset_snprintf HWLOC_NAME(bitmap_taskset_snprintf) +#define hwloc_bitmap_taskset_asprintf HWLOC_NAME(bitmap_taskset_asprintf) +#define hwloc_bitmap_taskset_sscanf HWLOC_NAME(bitmap_taskset_sscanf) +#define hwloc_bitmap_zero HWLOC_NAME(bitmap_zero) +#define hwloc_bitmap_fill HWLOC_NAME(bitmap_fill) +#define hwloc_bitmap_from_ulong HWLOC_NAME(bitmap_from_ulong) + +#define hwloc_bitmap_from_ith_ulong HWLOC_NAME(bitmap_from_ith_ulong) +#define hwloc_bitmap_to_ulong HWLOC_NAME(bitmap_to_ulong) +#define hwloc_bitmap_to_ith_ulong HWLOC_NAME(bitmap_to_ith_ulong) +#define hwloc_bitmap_only HWLOC_NAME(bitmap_only) +#define hwloc_bitmap_allbut HWLOC_NAME(bitmap_allbut) +#define hwloc_bitmap_set HWLOC_NAME(bitmap_set) +#define hwloc_bitmap_set_range HWLOC_NAME(bitmap_set_range) +#define hwloc_bitmap_set_ith_ulong HWLOC_NAME(bitmap_set_ith_ulong) +#define hwloc_bitmap_clr HWLOC_NAME(bitmap_clr) +#define hwloc_bitmap_clr_range HWLOC_NAME(bitmap_clr_range) +#define hwloc_bitmap_isset HWLOC_NAME(bitmap_isset) +#define hwloc_bitmap_iszero HWLOC_NAME(bitmap_iszero) +#define hwloc_bitmap_isfull HWLOC_NAME(bitmap_isfull) +#define hwloc_bitmap_isequal HWLOC_NAME(bitmap_isequal) +#define hwloc_bitmap_intersects HWLOC_NAME(bitmap_intersects) +#define hwloc_bitmap_isincluded HWLOC_NAME(bitmap_isincluded) +#define hwloc_bitmap_or HWLOC_NAME(bitmap_or) +#define hwloc_bitmap_and HWLOC_NAME(bitmap_and) +#define hwloc_bitmap_andnot HWLOC_NAME(bitmap_andnot) +#define hwloc_bitmap_xor HWLOC_NAME(bitmap_xor) +#define hwloc_bitmap_not HWLOC_NAME(bitmap_not) +#define hwloc_bitmap_first HWLOC_NAME(bitmap_first) +#define hwloc_bitmap_last HWLOC_NAME(bitmap_last) +#define hwloc_bitmap_next HWLOC_NAME(bitmap_next) +#define hwloc_bitmap_first_unset HWLOC_NAME(bitmap_first_unset) +#define hwloc_bitmap_last_unset HWLOC_NAME(bitmap_last_unset) +#define hwloc_bitmap_next_unset HWLOC_NAME(bitmap_next_unset) +#define hwloc_bitmap_singlify HWLOC_NAME(bitmap_singlify) +#define hwloc_bitmap_compare_first HWLOC_NAME(bitmap_compare_first) +#define hwloc_bitmap_compare HWLOC_NAME(bitmap_compare) +#define hwloc_bitmap_weight HWLOC_NAME(bitmap_weight) + +/* hwloc/helper.h */ + +#define hwloc_get_type_or_below_depth HWLOC_NAME(get_type_or_below_depth) +#define hwloc_get_type_or_above_depth HWLOC_NAME(get_type_or_above_depth) +#define hwloc_get_root_obj HWLOC_NAME(get_root_obj) +#define hwloc_get_ancestor_obj_by_depth HWLOC_NAME(get_ancestor_obj_by_depth) +#define hwloc_get_ancestor_obj_by_type HWLOC_NAME(get_ancestor_obj_by_type) +#define hwloc_get_next_obj_by_depth HWLOC_NAME(get_next_obj_by_depth) +#define hwloc_get_next_obj_by_type HWLOC_NAME(get_next_obj_by_type) +#define hwloc_get_pu_obj_by_os_index HWLOC_NAME(get_pu_obj_by_os_index) +#define hwloc_get_numanode_obj_by_os_index HWLOC_NAME(get_numanode_obj_by_os_index) +#define hwloc_get_next_child HWLOC_NAME(get_next_child) +#define hwloc_get_common_ancestor_obj HWLOC_NAME(get_common_ancestor_obj) +#define hwloc_obj_is_in_subtree HWLOC_NAME(obj_is_in_subtree) +#define hwloc_get_first_largest_obj_inside_cpuset HWLOC_NAME(get_first_largest_obj_inside_cpuset) +#define hwloc_get_largest_objs_inside_cpuset HWLOC_NAME(get_largest_objs_inside_cpuset) +#define hwloc_get_next_obj_inside_cpuset_by_depth HWLOC_NAME(get_next_obj_inside_cpuset_by_depth) +#define hwloc_get_next_obj_inside_cpuset_by_type HWLOC_NAME(get_next_obj_inside_cpuset_by_type) +#define hwloc_get_obj_inside_cpuset_by_depth HWLOC_NAME(get_obj_inside_cpuset_by_depth) +#define hwloc_get_obj_inside_cpuset_by_type HWLOC_NAME(get_obj_inside_cpuset_by_type) +#define hwloc_get_nbobjs_inside_cpuset_by_depth HWLOC_NAME(get_nbobjs_inside_cpuset_by_depth) +#define hwloc_get_nbobjs_inside_cpuset_by_type HWLOC_NAME(get_nbobjs_inside_cpuset_by_type) +#define hwloc_get_obj_index_inside_cpuset HWLOC_NAME(get_obj_index_inside_cpuset) +#define hwloc_get_child_covering_cpuset HWLOC_NAME(get_child_covering_cpuset) +#define hwloc_get_obj_covering_cpuset HWLOC_NAME(get_obj_covering_cpuset) +#define hwloc_get_next_obj_covering_cpuset_by_depth HWLOC_NAME(get_next_obj_covering_cpuset_by_depth) +#define hwloc_get_next_obj_covering_cpuset_by_type HWLOC_NAME(get_next_obj_covering_cpuset_by_type) +#define hwloc_obj_type_is_normal HWLOC_NAME(obj_type_is_normal) +#define hwloc_obj_type_is_memory HWLOC_NAME(obj_type_is_memory) +#define hwloc_obj_type_is_io HWLOC_NAME(obj_type_is_io) +#define hwloc_obj_type_is_cache HWLOC_NAME(obj_type_is_cache) +#define hwloc_obj_type_is_dcache HWLOC_NAME(obj_type_is_dcache) +#define hwloc_obj_type_is_icache HWLOC_NAME(obj_type_is_icache) +#define hwloc_get_cache_type_depth HWLOC_NAME(get_cache_type_depth) +#define hwloc_get_cache_covering_cpuset HWLOC_NAME(get_cache_covering_cpuset) +#define hwloc_get_shared_cache_covering_obj HWLOC_NAME(get_shared_cache_covering_obj) +#define hwloc_get_closest_objs HWLOC_NAME(get_closest_objs) +#define hwloc_get_obj_below_by_type HWLOC_NAME(get_obj_below_by_type) +#define hwloc_get_obj_below_array_by_type HWLOC_NAME(get_obj_below_array_by_type) +#define hwloc_distrib_flags_e HWLOC_NAME(distrib_flags_e) +#define HWLOC_DISTRIB_FLAG_REVERSE HWLOC_NAME_CAPS(DISTRIB_FLAG_REVERSE) +#define hwloc_distrib HWLOC_NAME(distrib) +#define hwloc_alloc_membind_policy HWLOC_NAME(alloc_membind_policy) +#define hwloc_alloc_membind_policy_nodeset HWLOC_NAME(alloc_membind_policy_nodeset) +#define hwloc_topology_get_complete_cpuset HWLOC_NAME(topology_get_complete_cpuset) +#define hwloc_topology_get_topology_cpuset HWLOC_NAME(topology_get_topology_cpuset) +#define hwloc_topology_get_allowed_cpuset HWLOC_NAME(topology_get_allowed_cpuset) +#define hwloc_topology_get_complete_nodeset HWLOC_NAME(topology_get_complete_nodeset) +#define hwloc_topology_get_topology_nodeset HWLOC_NAME(topology_get_topology_nodeset) +#define hwloc_topology_get_allowed_nodeset HWLOC_NAME(topology_get_allowed_nodeset) +#define hwloc_cpuset_to_nodeset HWLOC_NAME(cpuset_to_nodeset) +#define hwloc_cpuset_from_nodeset HWLOC_NAME(cpuset_from_nodeset) + +/* export.h */ + +#define hwloc_topology_export_xml_flags_e HWLOC_NAME(topology_export_xml_flags_e) +#define HWLOC_TOPOLOGY_EXPORT_XML_FLAG_V1 HWLOC_NAME_CAPS(TOPOLOGY_EXPORT_XML_FLAG_V1) +#define hwloc_topology_export_xml HWLOC_NAME(topology_export_xml) +#define hwloc_topology_export_xmlbuffer HWLOC_NAME(topology_export_xmlbuffer) +#define hwloc_free_xmlbuffer HWLOC_NAME(free_xmlbuffer) +#define hwloc_topology_set_userdata_export_callback HWLOC_NAME(topology_set_userdata_export_callback) +#define hwloc_export_obj_userdata HWLOC_NAME(export_obj_userdata) +#define hwloc_export_obj_userdata_base64 HWLOC_NAME(export_obj_userdata_base64) +#define hwloc_topology_set_userdata_import_callback HWLOC_NAME(topology_set_userdata_import_callback) + +#define hwloc_topology_export_synthetic_flags_e HWLOC_NAME(topology_export_synthetic_flags_e) +#define HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_NO_EXTENDED_TYPES HWLOC_NAME_CAPS(TOPOLOGY_EXPORT_SYNTHETIC_FLAG_NO_EXTENDED_TYPES) +#define HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_NO_ATTRS HWLOC_NAME_CAPS(TOPOLOGY_EXPORT_SYNTHETIC_FLAG_NO_ATTRS) +#define HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_V1 HWLOC_NAME_CAPS(TOPOLOGY_EXPORT_SYNTHETIC_FLAG_V1) +#define HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_IGNORE_MEMORY HWLOC_NAME_CAPS(TOPOLOGY_EXPORT_SYNTHETIC_FLAG_IGNORE_MEMORY) +#define hwloc_topology_export_synthetic HWLOC_NAME(topology_export_synthetic) + +/* distances.h */ + +#define hwloc_distances_s HWLOC_NAME(distances_s) + +#define hwloc_distances_kind_e HWLOC_NAME(distances_kind_e) +#define HWLOC_DISTANCES_KIND_FROM_OS HWLOC_NAME_CAPS(DISTANCES_KIND_FROM_OS) +#define HWLOC_DISTANCES_KIND_FROM_USER HWLOC_NAME_CAPS(DISTANCES_KIND_FROM_USER) +#define HWLOC_DISTANCES_KIND_MEANS_LATENCY HWLOC_NAME_CAPS(DISTANCES_KIND_MEANS_LATENCY) +#define HWLOC_DISTANCES_KIND_MEANS_BANDWIDTH HWLOC_NAME_CAPS(DISTANCES_KIND_MEANS_BANDWIDTH) + +#define hwloc_distances_get HWLOC_NAME(distances_get) +#define hwloc_distances_get_by_depth HWLOC_NAME(distances_get_by_depth) +#define hwloc_distances_get_by_type HWLOC_NAME(distances_get_by_type) +#define hwloc_distances_release HWLOC_NAME(distances_release) +#define hwloc_distances_obj_index HWLOC_NAME(distances_obj_index) +#define hwloc_distances_obj_pair_values HWLOC_NAME(distances_pair_values) + +#define hwloc_distances_add_flag_e HWLOC_NAME(distances_add_flag_e) +#define HWLOC_DISTANCES_ADD_FLAG_GROUP HWLOC_NAME_CAPS(DISTANCES_ADD_FLAG_GROUP) +#define HWLOC_DISTANCES_ADD_FLAG_GROUP_INACCURATE HWLOC_NAME_CAPS(DISTANCES_ADD_FLAG_GROUP_INACCURATE) + +#define hwloc_distances_add HWLOC_NAME(distances_add) +#define hwloc_distances_remove HWLOC_NAME(distances_remove) +#define hwloc_distances_remove_by_depth HWLOC_NAME(distances_remove_by_depth) +#define hwloc_distances_remove_by_type HWLOC_NAME(distances_remove_by_type) + +/* diff.h */ + +#define hwloc_topology_diff_obj_attr_type_e HWLOC_NAME(topology_diff_obj_attr_type_e) +#define hwloc_topology_diff_obj_attr_type_t HWLOC_NAME(topology_diff_obj_attr_type_t) +#define HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_SIZE HWLOC_NAME_CAPS(TOPOLOGY_DIFF_OBJ_ATTR_SIZE) +#define HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_NAME HWLOC_NAME_CAPS(TOPOLOGY_DIFF_OBJ_ATTR_NAME) +#define HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_INFO HWLOC_NAME_CAPS(TOPOLOGY_DIFF_OBJ_ATTR_INFO) +#define hwloc_topology_diff_obj_attr_u HWLOC_NAME(topology_diff_obj_attr_u) +#define hwloc_topology_diff_obj_attr_generic_s HWLOC_NAME(topology_diff_obj_attr_generic_s) +#define hwloc_topology_diff_obj_attr_uint64_s HWLOC_NAME(topology_diff_obj_attr_uint64_s) +#define hwloc_topology_diff_obj_attr_string_s HWLOC_NAME(topology_diff_obj_attr_string_s) +#define hwloc_topology_diff_type_e HWLOC_NAME(topology_diff_type_e) +#define hwloc_topology_diff_type_t HWLOC_NAME(topology_diff_type_t) +#define HWLOC_TOPOLOGY_DIFF_OBJ_ATTR HWLOC_NAME_CAPS(TOPOLOGY_DIFF_OBJ_ATTR) +#define HWLOC_TOPOLOGY_DIFF_TOO_COMPLEX HWLOC_NAME_CAPS(TOPOLOGY_DIFF_TOO_COMPLEX) +#define hwloc_topology_diff_u HWLOC_NAME(topology_diff_u) +#define hwloc_topology_diff_t HWLOC_NAME(topology_diff_t) +#define hwloc_topology_diff_generic_s HWLOC_NAME(topology_diff_generic_s) +#define hwloc_topology_diff_obj_attr_s HWLOC_NAME(topology_diff_obj_attr_s) +#define hwloc_topology_diff_too_complex_s HWLOC_NAME(topology_diff_too_complex_s) +#define hwloc_topology_diff_build HWLOC_NAME(topology_diff_build) +#define hwloc_topology_diff_apply_flags_e HWLOC_NAME(topology_diff_apply_flags_e) +#define HWLOC_TOPOLOGY_DIFF_APPLY_REVERSE HWLOC_NAME_CAPS(TOPOLOGY_DIFF_APPLY_REVERSE) +#define hwloc_topology_diff_apply HWLOC_NAME(topology_diff_apply) +#define hwloc_topology_diff_destroy HWLOC_NAME(topology_diff_destroy) +#define hwloc_topology_diff_load_xml HWLOC_NAME(topology_diff_load_xml) +#define hwloc_topology_diff_export_xml HWLOC_NAME(topology_diff_export_xml) +#define hwloc_topology_diff_load_xmlbuffer HWLOC_NAME(topology_diff_load_xmlbuffer) +#define hwloc_topology_diff_export_xmlbuffer HWLOC_NAME(topology_diff_export_xmlbuffer) + +/* shmem.h */ + +#define hwloc_shmem_topology_get_length HWLOC_NAME(shmem_topology_get_length) +#define hwloc_shmem_topology_write HWLOC_NAME(shmem_topology_write) +#define hwloc_shmem_topology_adopt HWLOC_NAME(shmem_topology_adopt) + +/* glibc-sched.h */ + +#define hwloc_cpuset_to_glibc_sched_affinity HWLOC_NAME(cpuset_to_glibc_sched_affinity) +#define hwloc_cpuset_from_glibc_sched_affinity HWLOC_NAME(cpuset_from_glibc_sched_affinity) + +/* linux-libnuma.h */ + +#define hwloc_cpuset_to_linux_libnuma_ulongs HWLOC_NAME(cpuset_to_linux_libnuma_ulongs) +#define hwloc_nodeset_to_linux_libnuma_ulongs HWLOC_NAME(nodeset_to_linux_libnuma_ulongs) +#define hwloc_cpuset_from_linux_libnuma_ulongs HWLOC_NAME(cpuset_from_linux_libnuma_ulongs) +#define hwloc_nodeset_from_linux_libnuma_ulongs HWLOC_NAME(nodeset_from_linux_libnuma_ulongs) +#define hwloc_cpuset_to_linux_libnuma_bitmask HWLOC_NAME(cpuset_to_linux_libnuma_bitmask) +#define hwloc_nodeset_to_linux_libnuma_bitmask HWLOC_NAME(nodeset_to_linux_libnuma_bitmask) +#define hwloc_cpuset_from_linux_libnuma_bitmask HWLOC_NAME(cpuset_from_linux_libnuma_bitmask) +#define hwloc_nodeset_from_linux_libnuma_bitmask HWLOC_NAME(nodeset_from_linux_libnuma_bitmask) + +/* linux.h */ + +#define hwloc_linux_set_tid_cpubind HWLOC_NAME(linux_set_tid_cpubind) +#define hwloc_linux_get_tid_cpubind HWLOC_NAME(linux_get_tid_cpubind) +#define hwloc_linux_get_tid_last_cpu_location HWLOC_NAME(linux_get_tid_last_cpu_location) +#define hwloc_linux_read_path_as_cpumask HWLOC_NAME(linux_read_file_cpumask) + +/* openfabrics-verbs.h */ + +#define hwloc_ibv_get_device_cpuset HWLOC_NAME(ibv_get_device_cpuset) +#define hwloc_ibv_get_device_osdev HWLOC_NAME(ibv_get_device_osdev) +#define hwloc_ibv_get_device_osdev_by_name HWLOC_NAME(ibv_get_device_osdev_by_name) + +/* intel-mic.h */ + +#define hwloc_intel_mic_get_device_cpuset HWLOC_NAME(intel_mic_get_device_cpuset) +#define hwloc_intel_mic_get_device_osdev_by_index HWLOC_NAME(intel_mic_get_device_osdev_by_index) + +/* opencl.h */ + +#define hwloc_opencl_get_device_cpuset HWLOC_NAME(opencl_get_device_cpuset) +#define hwloc_opencl_get_device_osdev HWLOC_NAME(opencl_get_device_osdev) +#define hwloc_opencl_get_device_osdev_by_index HWLOC_NAME(opencl_get_device_osdev_by_index) + +/* cuda.h */ + +#define hwloc_cuda_get_device_pci_ids HWLOC_NAME(cuda_get_device_pci_ids) +#define hwloc_cuda_get_device_cpuset HWLOC_NAME(cuda_get_device_cpuset) +#define hwloc_cuda_get_device_pcidev HWLOC_NAME(cuda_get_device_pcidev) +#define hwloc_cuda_get_device_osdev HWLOC_NAME(cuda_get_device_osdev) +#define hwloc_cuda_get_device_osdev_by_index HWLOC_NAME(cuda_get_device_osdev_by_index) + +/* cudart.h */ + +#define hwloc_cudart_get_device_pci_ids HWLOC_NAME(cudart_get_device_pci_ids) +#define hwloc_cudart_get_device_cpuset HWLOC_NAME(cudart_get_device_cpuset) +#define hwloc_cudart_get_device_pcidev HWLOC_NAME(cudart_get_device_pcidev) +#define hwloc_cudart_get_device_osdev_by_index HWLOC_NAME(cudart_get_device_osdev_by_index) + +/* nvml.h */ + +#define hwloc_nvml_get_device_cpuset HWLOC_NAME(nvml_get_device_cpuset) +#define hwloc_nvml_get_device_osdev HWLOC_NAME(nvml_get_device_osdev) +#define hwloc_nvml_get_device_osdev_by_index HWLOC_NAME(nvml_get_device_osdev_by_index) + +/* gl.h */ + +#define hwloc_gl_get_display_osdev_by_port_device HWLOC_NAME(gl_get_display_osdev_by_port_device) +#define hwloc_gl_get_display_osdev_by_name HWLOC_NAME(gl_get_display_osdev_by_name) +#define hwloc_gl_get_display_by_osdev HWLOC_NAME(gl_get_display_by_osdev) + +/* hwloc/plugins.h */ + +#define hwloc_disc_component_type_e HWLOC_NAME(disc_component_type_e) +#define HWLOC_DISC_COMPONENT_TYPE_CPU HWLOC_NAME_CAPS(DISC_COMPONENT_TYPE_CPU) +#define HWLOC_DISC_COMPONENT_TYPE_GLOBAL HWLOC_NAME_CAPS(DISC_COMPONENT_TYPE_GLOBAL) +#define HWLOC_DISC_COMPONENT_TYPE_MISC HWLOC_NAME_CAPS(DISC_COMPONENT_TYPE_MISC) +#define hwloc_disc_component_type_t HWLOC_NAME(disc_component_type_t) +#define hwloc_disc_component HWLOC_NAME(disc_component) + +#define hwloc_backend HWLOC_NAME(backend) + +#define hwloc_backend_alloc HWLOC_NAME(backend_alloc) +#define hwloc_backend_enable HWLOC_NAME(backend_enable) + +#define hwloc_component_type_e HWLOC_NAME(component_type_e) +#define HWLOC_COMPONENT_TYPE_DISC HWLOC_NAME_CAPS(COMPONENT_TYPE_DISC) +#define HWLOC_COMPONENT_TYPE_XML HWLOC_NAME_CAPS(COMPONENT_TYPE_XML) +#define hwloc_component_type_t HWLOC_NAME(component_type_t) +#define hwloc_component HWLOC_NAME(component) + +#define hwloc_plugin_check_namespace HWLOC_NAME(plugin_check_namespace) + +#define hwloc_insert_object_by_cpuset HWLOC_NAME(insert_object_by_cpuset) +#define hwloc_report_error_t HWLOC_NAME(report_error_t) +#define hwloc_report_os_error HWLOC_NAME(report_os_error) +#define hwloc_hide_errors HWLOC_NAME(hide_errors) +#define hwloc__insert_object_by_cpuset HWLOC_NAME(_insert_object_by_cpuset) +#define hwloc_insert_object_by_parent HWLOC_NAME(insert_object_by_parent) +#define hwloc_alloc_setup_object HWLOC_NAME(alloc_setup_object) +#define hwloc_obj_add_children_sets HWLOC_NAME(add_children_sets) +#define hwloc_topology_reconnect HWLOC_NAME(topology_reconnect) + +#define hwloc_filter_check_pcidev_subtype_important HWLOC_NAME(filter_check_pcidev_subtype_important) +#define hwloc_filter_check_osdev_subtype_important HWLOC_NAME(filter_check_osdev_subtype_important) +#define hwloc_filter_check_keep_object_type HWLOC_NAME(filter_check_keep_object_type) +#define hwloc_filter_check_keep_object HWLOC_NAME(filter_check_keep_object) + +#define hwloc_pcidisc_find_cap HWLOC_NAME(pcidisc_find_cap) +#define hwloc_pcidisc_find_linkspeed HWLOC_NAME(pcidisc_find_linkspeed) +#define hwloc_pcidisc_check_bridge_type HWLOC_NAME(pcidisc_check_bridge_type) +#define hwloc_pcidisc_setup_bridge_attr HWLOC_NAME(pcidisc_setup_bridge_attr) +#define hwloc_pcidisc_tree_insert_by_busid HWLOC_NAME(pcidisc_tree_insert_by_busid) +#define hwloc_pcidisc_tree_attach HWLOC_NAME(pcidisc_tree_attach) + +#define hwloc_pcidisc_find_by_busid HWLOC_NAME(pcidisc_find_by_busid) +#define hwloc_pcidisc_find_busid_parent HWLOC_NAME(pcidisc_find_busid_parent) + +/* hwloc/deprecated.h */ + +#define hwloc_topology_insert_misc_object_by_parent HWLOC_NAME(topology_insert_misc_object_by_parent) +#define hwloc_obj_cpuset_snprintf HWLOC_NAME(obj_cpuset_snprintf) +#define hwloc_obj_type_sscanf HWLOC_NAME(obj_type_sscanf) + +#define hwloc_set_membind_nodeset HWLOC_NAME(set_membind_nodeset) +#define hwloc_get_membind_nodeset HWLOC_NAME(get_membind_nodeset) +#define hwloc_set_proc_membind_nodeset HWLOC_NAME(set_proc_membind_nodeset) +#define hwloc_get_proc_membind_nodeset HWLOC_NAME(get_proc_membind_nodeset) +#define hwloc_set_area_membind_nodeset HWLOC_NAME(set_area_membind_nodeset) +#define hwloc_get_area_membind_nodeset HWLOC_NAME(get_area_membind_nodeset) +#define hwloc_alloc_membind_nodeset HWLOC_NAME(alloc_membind_nodeset) + +#define hwloc_cpuset_to_nodeset_strict HWLOC_NAME(cpuset_to_nodeset_strict) +#define hwloc_cpuset_from_nodeset_strict HWLOC_NAME(cpuset_from_nodeset_strict) + +/* private/debug.h */ + +#define hwloc_debug_enabled HWLOC_NAME(debug_enabled) +#define hwloc_debug HWLOC_NAME(debug) + +/* private/misc.h */ + +#define hwloc_snprintf HWLOC_NAME(snprintf) +#define hwloc_namecoloncmp HWLOC_NAME(namecoloncmp) +#define hwloc_ffsl_manual HWLOC_NAME(ffsl_manual) +#define hwloc_ffs32 HWLOC_NAME(ffs32) +#define hwloc_ffsl_from_ffs32 HWLOC_NAME(ffsl_from_ffs32) +#define hwloc_flsl_manual HWLOC_NAME(flsl_manual) +#define hwloc_fls32 HWLOC_NAME(fls32) +#define hwloc_flsl_from_fls32 HWLOC_NAME(flsl_from_fls32) +#define hwloc_weight_long HWLOC_NAME(weight_long) +#define hwloc_strncasecmp HWLOC_NAME(strncasecmp) + +#define hwloc_bitmap_compare_inclusion HWLOC_NAME(bitmap_compare_inclusion) + +#define hwloc_pci_class_string HWLOC_NAME(pci_class_string) +#define hwloc_linux_pci_link_speed_from_string HWLOC_NAME(linux_pci_link_speed_from_string) + +#define hwloc_cache_type_by_depth_type HWLOC_NAME(cache_type_by_depth_type) +#define hwloc__obj_type_is_normal HWLOC_NAME(_obj_type_is_normal) +#define hwloc__obj_type_is_memory HWLOC_NAME(_obj_type_is_memory) +#define hwloc__obj_type_is_io HWLOC_NAME(_obj_type_is_io) +#define hwloc__obj_type_is_special HWLOC_NAME(_obj_type_is_special) + +#define hwloc__obj_type_is_cache HWLOC_NAME(_obj_type_is_cache) +#define hwloc__obj_type_is_dcache HWLOC_NAME(_obj_type_is_dcache) +#define hwloc__obj_type_is_icache HWLOC_NAME(_obj_type_is_icache) + +/* private/cpuid-x86.h */ + +#define hwloc_have_x86_cpuid HWLOC_NAME(have_x86_cpuid) +#define hwloc_x86_cpuid HWLOC_NAME(x86_cpuid) + +/* private/xml.h */ + +#define hwloc__xml_verbose HWLOC_NAME(_xml_verbose) + +#define hwloc__xml_import_state_s HWLOC_NAME(_xml_import_state_s) +#define hwloc__xml_import_state_t HWLOC_NAME(_xml_import_state_t) +#define hwloc__xml_import_diff HWLOC_NAME(_xml_import_diff) +#define hwloc_xml_backend_data_s HWLOC_NAME(xml_backend_data_s) +#define hwloc__xml_export_state_s HWLOC_NAME(_xml_export_state_s) +#define hwloc__xml_export_state_t HWLOC_NAME(_xml_export_state_t) +#define hwloc__xml_export_data_s HWLOC_NAME(_xml_export_data_s) +#define hwloc__xml_export_topology HWLOC_NAME(_xml_export_topology) +#define hwloc__xml_export_diff HWLOC_NAME(_xml_export_diff) + +#define hwloc_xml_callbacks HWLOC_NAME(xml_callbacks) +#define hwloc_xml_component HWLOC_NAME(xml_component) +#define hwloc_xml_callbacks_register HWLOC_NAME(xml_callbacks_register) +#define hwloc_xml_callbacks_reset HWLOC_NAME(xml_callbacks_reset) + +#define hwloc__xml_imported_v1distances_s HWLOC_NAME(_xml_imported_v1distances_s) + +/* private/components.h */ + +#define hwloc_disc_component_force_enable HWLOC_NAME(disc_component_force_enable) +#define hwloc_disc_components_enable_others HWLOC_NAME(disc_components_instantiate_others) + +#define hwloc_backends_is_thissystem HWLOC_NAME(backends_is_thissystem) +#define hwloc_backends_find_callbacks HWLOC_NAME(backends_find_callbacks) + +#define hwloc_backends_init HWLOC_NAME(backends_init) +#define hwloc_backends_disable_all HWLOC_NAME(backends_disable_all) + +#define hwloc_components_init HWLOC_NAME(components_init) +#define hwloc_components_fini HWLOC_NAME(components_fini) + +/* private/internal-private.h */ + +#define hwloc_xml_component HWLOC_NAME(xml_component) +#define hwloc_synthetic_component HWLOC_NAME(synthetic_component) + +#define hwloc_aix_component HWLOC_NAME(aix_component) +#define hwloc_bgq_component HWLOC_NAME(bgq_component) +#define hwloc_darwin_component HWLOC_NAME(darwin_component) +#define hwloc_freebsd_component HWLOC_NAME(freebsd_component) +#define hwloc_hpux_component HWLOC_NAME(hpux_component) +#define hwloc_linux_component HWLOC_NAME(linux_component) +#define hwloc_netbsd_component HWLOC_NAME(netbsd_component) +#define hwloc_noos_component HWLOC_NAME(noos_component) +#define hwloc_solaris_component HWLOC_NAME(solaris_component) +#define hwloc_windows_component HWLOC_NAME(windows_component) +#define hwloc_x86_component HWLOC_NAME(x86_component) + +#define hwloc_cuda_component HWLOC_NAME(cuda_component) +#define hwloc_gl_component HWLOC_NAME(gl_component) +#define hwloc_linuxio_component HWLOC_NAME(linuxio_component) +#define hwloc_nvml_component HWLOC_NAME(nvml_component) +#define hwloc_opencl_component HWLOC_NAME(opencl_component) +#define hwloc_pci_component HWLOC_NAME(pci_component) + +#define hwloc_xml_libxml_component HWLOC_NAME(xml_libxml_component) +#define hwloc_xml_nolibxml_component HWLOC_NAME(xml_nolibxml_component) + +/* private/private.h */ + +#define hwloc_special_level_s HWLOC_NAME(special_level_s) + +#define hwloc_pci_forced_locality_s HWLOC_NAME(pci_forced_locality_s) + +#define hwloc_alloc_root_sets HWLOC_NAME(alloc_root_sets) +#define hwloc_setup_pu_level HWLOC_NAME(setup_pu_level) +#define hwloc_get_sysctlbyname HWLOC_NAME(get_sysctlbyname) +#define hwloc_get_sysctl HWLOC_NAME(get_sysctl) +#define hwloc_fallback_nbprocessors HWLOC_NAME(fallback_nbprocessors) + +#define hwloc__object_cpusets_compare_first HWLOC_NAME(_object_cpusets_compare_first) +#define hwloc__reorder_children HWLOC_NAME(_reorder_children) + +#define hwloc_topology_setup_defaults HWLOC_NAME(topology_setup_defaults) +#define hwloc_topology_clear HWLOC_NAME(topology_clear) + +#define hwloc__attach_memory_object HWLOC_NAME(insert_memory_object) + +#define hwloc_pci_discovery_init HWLOC_NAME(pci_discovery_init) +#define hwloc_pci_discovery_prepare HWLOC_NAME(pci_discovery_prepare) +#define hwloc_pci_discovery_exit HWLOC_NAME(pci_discovery_exit) +#define hwloc_find_insert_io_parent_by_complete_cpuset HWLOC_NAME(hwloc_find_insert_io_parent_by_complete_cpuset) +#define hwloc_pci_belowroot_apply_locality HWLOC_NAME(pci_belowroot_apply_locality) + +#define hwloc__add_info HWLOC_NAME(_add_info) +#define hwloc__add_info_nodup HWLOC_NAME(_add_info_nodup) +#define hwloc__move_infos HWLOC_NAME(_move_infos) +#define hwloc__free_infos HWLOC_NAME(_free_infos) + +#define hwloc_binding_hooks HWLOC_NAME(binding_hooks) +#define hwloc_set_native_binding_hooks HWLOC_NAME(set_native_binding_hooks) +#define hwloc_set_binding_hooks HWLOC_NAME(set_binding_hooks) + +#define hwloc_set_linuxfs_hooks HWLOC_NAME(set_linuxfs_hooks) +#define hwloc_set_bgq_hooks HWLOC_NAME(set_bgq_hooks) +#define hwloc_set_solaris_hooks HWLOC_NAME(set_solaris_hooks) +#define hwloc_set_aix_hooks HWLOC_NAME(set_aix_hooks) +#define hwloc_set_windows_hooks HWLOC_NAME(set_windows_hooks) +#define hwloc_set_darwin_hooks HWLOC_NAME(set_darwin_hooks) +#define hwloc_set_freebsd_hooks HWLOC_NAME(set_freebsd_hooks) +#define hwloc_set_netbsd_hooks HWLOC_NAME(set_netbsd_hooks) +#define hwloc_set_hpux_hooks HWLOC_NAME(set_hpux_hooks) + +#define hwloc_look_hardwired_fujitsu_k HWLOC_NAME(look_hardwired_fujitsu_k) +#define hwloc_look_hardwired_fujitsu_fx10 HWLOC_NAME(look_hardwired_fujitsu_fx10) +#define hwloc_look_hardwired_fujitsu_fx100 HWLOC_NAME(look_hardwired_fujitsu_fx100) + +#define hwloc_add_uname_info HWLOC_NAME(add_uname_info) +#define hwloc_free_unlinked_object HWLOC_NAME(free_unlinked_object) +#define hwloc_free_object_and_children HWLOC_NAME(free_object_and_children) +#define hwloc_free_object_siblings_and_children HWLOC_NAME(free_object_siblings_and_children) + +#define hwloc_alloc_heap HWLOC_NAME(alloc_heap) +#define hwloc_alloc_mmap HWLOC_NAME(alloc_mmap) +#define hwloc_free_heap HWLOC_NAME(free_heap) +#define hwloc_free_mmap HWLOC_NAME(free_mmap) +#define hwloc_alloc_or_fail HWLOC_NAME(alloc_or_fail) + +#define hwloc_internal_distances_s HWLOC_NAME(internal_distances_s) +#define hwloc_internal_distances_init HWLOC_NAME(internal_distances_init) +#define hwloc_internal_distances_prepare HWLOC_NAME(internal_distances_prepare) +#define hwloc_internal_distances_dup HWLOC_NAME(internal_distances_dup) +#define hwloc_internal_distances_refresh HWLOC_NAME(internal_distances_refresh) +#define hwloc_internal_distances_destroy HWLOC_NAME(internal_distances_destroy) + +#define hwloc_internal_distances_add HWLOC_NAME(internal_distances_add) +#define hwloc_internal_distances_add_by_index HWLOC_NAME(internal_distances_add_by_index) +#define hwloc_internal_distances_invalidate_cached_objs HWLOC_NAME(hwloc_internal_distances_invalidate_cached_objs) + +#define hwloc_encode_to_base64 HWLOC_NAME(encode_to_base64) +#define hwloc_decode_from_base64 HWLOC_NAME(decode_from_base64) + +#define hwloc_progname HWLOC_NAME(progname) + +#define hwloc__topology_disadopt HWLOC_NAME(_topology_disadopt) +#define hwloc__topology_dup HWLOC_NAME(_topology_dup) + +#define hwloc_tma HWLOC_NAME(tma) +#define hwloc_tma_malloc HWLOC_NAME(tma_malloc) +#define hwloc_tma_calloc HWLOC_NAME(tma_calloc) +#define hwloc_tma_strdup HWLOC_NAME(tma_strdup) +#define hwloc_bitmap_tma_dup HWLOC_NAME(bitmap_tma_dup) + +/* private/solaris-chiptype.h */ + +#define hwloc_solaris_chip_info_s HWLOC_NAME(solaris_chip_info_s) +#define hwloc_solaris_get_chip_info HWLOC_NAME(solaris_get_chip_info) + +#endif /* HWLOC_SYM_TRANSFORM */ + + +#ifdef __cplusplus +} /* extern "C" */ +#endif + + +#endif /* HWLOC_RENAME_H */ diff --git a/src/3rdparty/hwloc/include/hwloc/shmem.h b/src/3rdparty/hwloc/include/hwloc/shmem.h new file mode 100644 index 000000000..222494630 --- /dev/null +++ b/src/3rdparty/hwloc/include/hwloc/shmem.h @@ -0,0 +1,137 @@ +/* + * Copyright © 2013-2018 Inria. All rights reserved. + * See COPYING in top-level directory. + */ + +/** \file + * \brief Sharing topologies between processes + */ + +#ifndef HWLOC_SHMEM_H +#define HWLOC_SHMEM_H + +#include + +#ifdef __cplusplus +extern "C" { +#elif 0 +} +#endif + + +/** \defgroup hwlocality_shmem Sharing topologies between processes + * + * These functions are used to share a topology between processes by + * duplicating it into a file-backed shared-memory buffer. + * + * The master process must first get the required shared-memory size + * for storing this topology with hwloc_shmem_topology_get_length(). + * + * Then it must find a virtual memory area of that size that is available + * in all processes (identical virtual addresses in all processes). + * On Linux, this can be done by comparing holes found in /proc/\/maps + * for each process. + * + * Once found, it must open a destination file for storing the buffer, + * and pass it to hwloc_shmem_topology_write() together with + * virtual memory address and length obtained above. + * + * Other processes may then adopt this shared topology by opening the + * same file and passing it to hwloc_shmem_topology_adopt() with the + * exact same virtual memory address and length. + * + * @{ + */ + +/** \brief Get the required shared memory length for storing a topology. + * + * This length (in bytes) must be used in hwloc_shmem_topology_write() + * and hwloc_shmem_topology_adopt() later. + * + * \note Flags \p flags are currently unused, must be 0. + */ +HWLOC_DECLSPEC int hwloc_shmem_topology_get_length(hwloc_topology_t topology, + size_t *lengthp, + unsigned long flags); + +/** \brief Duplicate a topology to a shared memory file. + * + * Temporarily map a file in virtual memory and duplicate the + * topology \p topology by allocating duplicates in there. + * + * The segment of the file pointed by descriptor \p fd, + * starting at offset \p fileoffset, and of length \p length (in bytes), + * will be temporarily mapped at virtual address \p mmap_address + * during the duplication. + * + * The mapping length \p length must have been previously obtained with + * hwloc_shmem_topology_get_length() + * and the topology must not have been modified in the meantime. + * + * \note Flags \p flags are currently unused, must be 0. + * + * \note The object userdata pointer is duplicated but the pointed buffer + * is not. However the caller may also allocate it manually in shared memory + * to share it as well. + * + * \return -1 with errno set to EBUSY if the virtual memory mapping defined + * by \p mmap_address and \p length isn't available in the process. + * \return -1 with errno set to EINVAL if \p fileoffset, \p mmap_address + * or \p length aren't page-aligned. + */ +HWLOC_DECLSPEC int hwloc_shmem_topology_write(hwloc_topology_t topology, + int fd, hwloc_uint64_t fileoffset, + void *mmap_address, size_t length, + unsigned long flags); + +/** \brief Adopt a shared memory topology stored in a file. + * + * Map a file in virtual memory and adopt the topology that was previously + * stored there with hwloc_shmem_topology_write(). + * + * The returned adopted topology in \p topologyp can be used just like any + * topology. And it must be destroyed with hwloc_topology_destroy() as usual. + * + * However the topology is read-only. + * For instance, it cannot be modified with hwloc_topology_restrict() + * and object userdata pointers cannot be changed. + * + * The segment of the file pointed by descriptor \p fd, + * starting at offset \p fileoffset, and of length \p length (in bytes), + * will be mapped at virtual address \p mmap_address. + * + * The file pointed by descriptor \p fd, the offset \p fileoffset, + * the requested mapping virtual address \p mmap_address and the length \p length + * must be identical to what was given to hwloc_shmem_topology_write() earlier. + * + * \note Flags \p flags are currently unused, must be 0. + * + * \note The object userdata pointer should not be used unless the process + * that created the shared topology also placed userdata-pointed buffers + * in shared memory. + * + * \note This function takes care of calling hwloc_topology_abi_check(). + * + * \return -1 with errno set to EBUSY if the virtual memory mapping defined + * by \p mmap_address and \p length isn't available in the process. + * + * \return -1 with errno set to EINVAL if \p fileoffset, \p mmap_address + * or \p length aren't page-aligned, or do not match what was given to + * hwloc_shmem_topology_write() earlier. + * + * \return -1 with errno set to EINVAL if the layout of the topology structure + * is different between the writer process and the adopter process. + */ +HWLOC_DECLSPEC int hwloc_shmem_topology_adopt(hwloc_topology_t *topologyp, + int fd, hwloc_uint64_t fileoffset, + void *mmap_address, size_t length, + unsigned long flags); +/** @} */ + + +#ifdef __cplusplus +} /* extern "C" */ +#endif + + +#endif /* HWLOC_SHMEM_H */ diff --git a/src/3rdparty/hwloc/include/private/autogen/config.h b/src/3rdparty/hwloc/include/private/autogen/config.h new file mode 100644 index 000000000..a97bdfea2 --- /dev/null +++ b/src/3rdparty/hwloc/include/private/autogen/config.h @@ -0,0 +1,672 @@ +/* + * Copyright © 2009, 2011, 2012 CNRS. All rights reserved. + * Copyright © 2009-2018 Inria. All rights reserved. + * Copyright © 2009, 2011, 2012, 2015 Université Bordeaux. All rights reserved. + * Copyright © 2009 Cisco Systems, Inc. All rights reserved. + * $COPYRIGHT$ + * + * Additional copyrights may follow + * + * $HEADER$ + */ + +#ifndef HWLOC_CONFIGURE_H +#define HWLOC_CONFIGURE_H + +#define DECLSPEC_EXPORTS + +#define HWLOC_HAVE_MSVC_CPUIDEX 1 + +/* Define to 1 if the system has the type `CACHE_DESCRIPTOR'. */ +#define HAVE_CACHE_DESCRIPTOR 0 + +/* Define to 1 if the system has the type `CACHE_RELATIONSHIP'. */ +#define HAVE_CACHE_RELATIONSHIP 0 + +/* Define to 1 if you have the `clz' function. */ +/* #undef HAVE_CLZ */ + +/* Define to 1 if you have the `clzl' function. */ +/* #undef HAVE_CLZL */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_CL_CL_EXT_H */ + +/* Define to 1 if you have the `cpuset_setaffinity' function. */ +/* #undef HAVE_CPUSET_SETAFFINITY */ + +/* Define to 1 if you have the `cpuset_setid' function. */ +/* #undef HAVE_CPUSET_SETID */ + +/* Define to 1 if we have -lcuda */ +/* #undef HAVE_CUDA */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_CUDA_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_CUDA_RUNTIME_API_H */ + +/* Define to 1 if you have the declaration of `CL_DEVICE_TOPOLOGY_AMD', and to + 0 if you don't. */ +/* #undef HAVE_DECL_CL_DEVICE_TOPOLOGY_AMD */ + +/* Define to 1 if you have the declaration of `CTL_HW', and to 0 if you don't. + */ +/* #undef HAVE_DECL_CTL_HW */ + +/* Define to 1 if you have the declaration of `fabsf', and to 0 if you don't. + */ +#define HAVE_DECL_FABSF 1 + +/* Define to 1 if you have the declaration of `modff', and to 0 if you don't. + */ +#define HAVE_DECL_MODFF 1 + +/* Define to 1 if you have the declaration of `HW_NCPU', and to 0 if you + don't. */ +/* #undef HAVE_DECL_HW_NCPU */ + +/* Define to 1 if you have the declaration of + `nvmlDeviceGetMaxPcieLinkGeneration', and to 0 if you don't. */ +/* #undef HAVE_DECL_NVMLDEVICEGETMAXPCIELINKGENERATION */ + +/* Define to 1 if you have the declaration of `pthread_getaffinity_np', and to + 0 if you don't. */ +#define HAVE_DECL_PTHREAD_GETAFFINITY_NP 0 + +/* Define to 1 if you have the declaration of `pthread_setaffinity_np', and to + 0 if you don't. */ +#define HAVE_DECL_PTHREAD_SETAFFINITY_NP 0 + +/* Define to 1 if you have the declaration of `strtoull', and to 0 if you + don't. */ +#define HAVE_DECL_STRTOULL 0 + +/* Define to 1 if you have the declaration of `strcasecmp', and to 0 if you + don't. */ +/* #undef HWLOC_HAVE_DECL_STRCASECMP */ + +/* Define to 1 if you have the declaration of `snprintf', and to 0 if you + don't. */ +#define HAVE_DECL_SNPRINTF 0 + +/* Define to 1 if you have the declaration of `_strdup', and to 0 if you + don't. */ +#define HAVE_DECL__STRDUP 1 + +/* Define to 1 if you have the declaration of `_putenv', and to 0 if you + don't. */ +#define HAVE_DECL__PUTENV 1 + +/* Define to 1 if you have the declaration of `_SC_LARGE_PAGESIZE', and to 0 + if you don't. */ +#define HAVE_DECL__SC_LARGE_PAGESIZE 0 + +/* Define to 1 if you have the declaration of `_SC_NPROCESSORS_CONF', and to 0 + if you don't. */ +#define HAVE_DECL__SC_NPROCESSORS_CONF 0 + +/* Define to 1 if you have the declaration of `_SC_NPROCESSORS_ONLN', and to 0 + if you don't. */ +#define HAVE_DECL__SC_NPROCESSORS_ONLN 0 + +/* Define to 1 if you have the declaration of `_SC_NPROC_CONF', and to 0 if + you don't. */ +#define HAVE_DECL__SC_NPROC_CONF 0 + +/* Define to 1 if you have the declaration of `_SC_NPROC_ONLN', and to 0 if + you don't. */ +#define HAVE_DECL__SC_NPROC_ONLN 0 + +/* Define to 1 if you have the declaration of `_SC_PAGESIZE', and to 0 if you + don't. */ +#define HAVE_DECL__SC_PAGESIZE 0 + +/* Define to 1 if you have the declaration of `_SC_PAGE_SIZE', and to 0 if you + don't. */ +#define HAVE_DECL__SC_PAGE_SIZE 0 + +/* Define to 1 if you have the header file. */ +/* #define HAVE_DIRENT_H 1 */ +#undef HAVE_DIRENT_H + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_DLFCN_H */ + +/* Define to 1 if you have the `ffs' function. */ +/* #undef HAVE_FFS */ + +/* Define to 1 if you have the `ffsl' function. */ +/* #undef HAVE_FFSL */ + +/* Define to 1 if you have the `fls' function. */ +/* #undef HAVE_FLS */ + +/* Define to 1 if you have the `flsl' function. */ +/* #undef HAVE_FLSL */ + +/* Define to 1 if you have the `getpagesize' function. */ +#define HAVE_GETPAGESIZE 1 + +/* Define to 1 if the system has the type `GROUP_AFFINITY'. */ +#define HAVE_GROUP_AFFINITY 1 + +/* Define to 1 if the system has the type `GROUP_RELATIONSHIP'. */ +#define HAVE_GROUP_RELATIONSHIP 1 + +/* Define to 1 if you have the `host_info' function. */ +/* #undef HAVE_HOST_INFO */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_INFINIBAND_VERBS_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_INTTYPES_H 1 + +/* Define to 1 if the system has the type `KAFFINITY'. */ +#define HAVE_KAFFINITY 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_KSTAT_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_LANGINFO_H */ + +/* Define to 1 if we have -lgdi32 */ +#define HAVE_LIBGDI32 1 + +/* Define to 1 if we have -libverbs */ +/* #undef HAVE_LIBIBVERBS */ + +/* Define to 1 if we have -lkstat */ +/* #undef HAVE_LIBKSTAT */ + +/* Define to 1 if we have -llgrp */ +/* #undef HAVE_LIBLGRP */ + +/* Define to 1 if you have the header file. */ +#define HAVE_LOCALE_H 1 + +/* Define to 1 if the system has the type `LOGICAL_PROCESSOR_RELATIONSHIP'. */ +#define HAVE_LOGICAL_PROCESSOR_RELATIONSHIP 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_MACH_MACH_HOST_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_MACH_MACH_INIT_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_MALLOC_H 1 + +/* Define to 1 if you have the `memalign' function. */ +/* #undef HAVE_MEMALIGN */ + +/* Define to 1 if you have the header file. */ +#define HAVE_MEMORY_H 1 + +/* Define to 1 if you have the `nl_langinfo' function. */ +/* #undef HAVE_NL_LANGINFO */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NUMAIF_H */ + +/* Define to 1 if the system has the type `NUMA_NODE_RELATIONSHIP'. */ +#define HAVE_NUMA_NODE_RELATIONSHIP 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NVCTRL_NVCTRL_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_NVML_H */ + +/* Define to 1 if you have the `openat' function. */ +/* #undef HAVE_OPENAT */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_PICL_H */ + +/* Define to 1 if you have the `posix_memalign' function. */ +/* #undef HAVE_POSIX_MEMALIGN */ + +/* Define to 1 if the system has the type `PROCESSOR_CACHE_TYPE'. */ +#define HAVE_PROCESSOR_CACHE_TYPE 1 + +/* Define to 1 if the system has the type `PROCESSOR_GROUP_INFO'. */ +#define HAVE_PROCESSOR_GROUP_INFO 1 + +/* Define to 1 if the system has the type `PROCESSOR_RELATIONSHIP'. */ +#define HAVE_PROCESSOR_RELATIONSHIP 1 + +/* Define to 1 if the system has the type `PSAPI_WORKING_SET_EX_BLOCK'. */ +/* #undef HAVE_PSAPI_WORKING_SET_EX_BLOCK */ + +/* Define to 1 if the system has the type `PSAPI_WORKING_SET_EX_INFORMATION'. + */ +/* #undef HAVE_PSAPI_WORKING_SET_EX_INFORMATION */ + +/* Define to 1 if the system has the type `PROCESSOR_NUMBER'. */ +#define HAVE_PROCESSOR_NUMBER 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_PTHREAD_NP_H */ + +/* Define to 1 if the system has the type `pthread_t'. */ +/* #undef HAVE_PTHREAD_T */ +#undef HAVE_PTHREAD_T + +/* Define to 1 if you have the `putwc' function. */ +#define HAVE_PUTWC 1 + +/* Define to 1 if the system has the type `RelationProcessorPackage'. */ +/* #undef HAVE_RELATIONPROCESSORPACKAGE */ + +/* Define to 1 if you have the `setlocale' function. */ +#define HAVE_SETLOCALE 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_STDINT_H 1 + +/* Define to 1 if you have the header file. */ +#define HAVE_STDLIB_H 1 + +/* Define to 1 if you have the `strftime' function. */ +#define HAVE_STRFTIME 1 + +/* Define to 1 if you have the header file. */ +/* #define HAVE_STRINGS_H 1*/ +#undef HAVE_STRINGS_H + +/* Define to 1 if you have the header file. */ +#define HAVE_STRING_H 1 + +/* Define to 1 if you have the `strncasecmp' function. */ +#define HAVE_STRNCASECMP 1 + +/* Define to '1' if sysctl is present and usable */ +/* #undef HAVE_SYSCTL */ + +/* Define to '1' if sysctlbyname is present and usable */ +/* #undef HAVE_SYSCTLBYNAME */ + +/* Define to 1 if the system has the type + `SYSTEM_LOGICAL_PROCESSOR_INFORMATION'. */ +#define HAVE_SYSTEM_LOGICAL_PROCESSOR_INFORMATION 1 + +/* Define to 1 if the system has the type + `SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX'. */ +#define HAVE_SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_CPUSET_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_LGRP_USER_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_MMAN_H */ + +/* Define to 1 if you have the header file. */ +/* #define HAVE_SYS_PARAM_H 1 */ +#undef HAVE_SYS_PARAM_H + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_STAT_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_SYSCTL_H */ + +/* Define to 1 if you have the header file. */ +#define HAVE_SYS_TYPES_H 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_SYS_UTSNAME_H */ + +/* Define to 1 if you have the `uname' function. */ +/* #undef HAVE_UNAME */ + +/* Define to 1 if you have the header file. */ +/* #define HAVE_UNISTD_H 1 */ +#undef HAVE_UNISTD_H + +/* Define to 1 if you have the `uselocale' function. */ +/* #undef HAVE_USELOCALE */ + +/* Define to 1 if the system has the type `wchar_t'. */ +#define HAVE_WCHAR_T 1 + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_X11_KEYSYM_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_X11_XLIB_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_X11_XUTIL_H */ + +/* Define to 1 if you have the header file. */ +/* #undef HAVE_XLOCALE_H */ + +/* Define to 1 on AIX */ +/* #undef HWLOC_AIX_SYS */ + +/* Define to 1 on BlueGene/Q */ +/* #undef HWLOC_BGQ_SYS */ + +/* Whether C compiler supports symbol visibility or not */ +#define HWLOC_C_HAVE_VISIBILITY 0 + +/* Define to 1 on Darwin */ +/* #undef HWLOC_DARWIN_SYS */ + +/* Whether we are in debugging mode or not */ +/* #undef HWLOC_DEBUG */ + +/* Define to 1 on *FREEBSD */ +/* #undef HWLOC_FREEBSD_SYS */ + +/* Whether your compiler has __attribute__ or not */ +/* #define HWLOC_HAVE_ATTRIBUTE 1 */ +#undef HWLOC_HAVE_ATTRIBUTE + +/* Whether your compiler has __attribute__ aligned or not */ +/* #define HWLOC_HAVE_ATTRIBUTE_ALIGNED 1 */ + +/* Whether your compiler has __attribute__ always_inline or not */ +/* #define HWLOC_HAVE_ATTRIBUTE_ALWAYS_INLINE 1 */ + +/* Whether your compiler has __attribute__ cold or not */ +/* #define HWLOC_HAVE_ATTRIBUTE_COLD 1 */ + +/* Whether your compiler has __attribute__ const or not */ +/* #define HWLOC_HAVE_ATTRIBUTE_CONST 1 */ + +/* Whether your compiler has __attribute__ deprecated or not */ +/* #define HWLOC_HAVE_ATTRIBUTE_DEPRECATED 1 */ + +/* Whether your compiler has __attribute__ format or not */ +/* #define HWLOC_HAVE_ATTRIBUTE_FORMAT 1 */ + +/* Whether your compiler has __attribute__ hot or not */ +/* #define HWLOC_HAVE_ATTRIBUTE_HOT 1 */ + +/* Whether your compiler has __attribute__ malloc or not */ +/* #define HWLOC_HAVE_ATTRIBUTE_MALLOC 1 */ + +/* Whether your compiler has __attribute__ may_alias or not */ +/* #define HWLOC_HAVE_ATTRIBUTE_MAY_ALIAS 1 */ + +/* Whether your compiler has __attribute__ nonnull or not */ +/* #define HWLOC_HAVE_ATTRIBUTE_NONNULL 1 */ + +/* Whether your compiler has __attribute__ noreturn or not */ +/* #define HWLOC_HAVE_ATTRIBUTE_NORETURN 1 */ + +/* Whether your compiler has __attribute__ no_instrument_function or not */ +/* #define HWLOC_HAVE_ATTRIBUTE_NO_INSTRUMENT_FUNCTION 1 */ + +/* Whether your compiler has __attribute__ packed or not */ +/* #define HWLOC_HAVE_ATTRIBUTE_PACKED 1 */ + +/* Whether your compiler has __attribute__ pure or not */ +/* #define HWLOC_HAVE_ATTRIBUTE_PURE 1 */ + +/* Whether your compiler has __attribute__ sentinel or not */ +/* #define HWLOC_HAVE_ATTRIBUTE_SENTINEL 1 */ + +/* Whether your compiler has __attribute__ unused or not */ +/* #define HWLOC_HAVE_ATTRIBUTE_UNUSED 1 */ + +/* Whether your compiler has __attribute__ warn unused result or not */ +/* #define HWLOC_HAVE_ATTRIBUTE_WARN_UNUSED_RESULT 1 */ + +/* Whether your compiler has __attribute__ weak alias or not */ +/* #define HWLOC_HAVE_ATTRIBUTE_WEAK_ALIAS 1 */ + +/* Define to 1 if your `ffs' function is known to be broken. */ +/* #undef HWLOC_HAVE_BROKEN_FFS */ + +/* Define to 1 if you have the `cairo' library. */ +/* #undef HWLOC_HAVE_CAIRO */ + +/* Define to 1 if you have the `clz' function. */ +/* #undef HWLOC_HAVE_CLZ */ + +/* Define to 1 if you have the `clzl' function. */ +/* #undef HWLOC_HAVE_CLZL */ + +/* Define to 1 if you have cpuid */ +/* #undef HWLOC_HAVE_CPUID */ + +/* Define to 1 if the CPU_SET macro works */ +/* #undef HWLOC_HAVE_CPU_SET */ + +/* Define to 1 if the CPU_SET_S macro works */ +/* #undef HWLOC_HAVE_CPU_SET_S */ + +/* Define to 1 if you have the `cudart' SDK. */ +/* #undef HWLOC_HAVE_CUDART */ + +/* Define to 1 if function `clz' is declared by system headers */ +/* #undef HWLOC_HAVE_DECL_CLZ */ + +/* Define to 1 if function `clzl' is declared by system headers */ +/* #undef HWLOC_HAVE_DECL_CLZL */ + +/* Define to 1 if function `ffs' is declared by system headers */ +/* #undef HWLOC_HAVE_DECL_FFS */ + +/* Define to 1 if function `ffsl' is declared by system headers */ +/* #undef HWLOC_HAVE_DECL_FFSL */ + +/* Define to 1 if function `fls' is declared by system headers */ +/* #undef HWLOC_HAVE_DECL_FLS */ + +/* Define to 1 if function `flsl' is declared by system headers */ +/* #undef HWLOC_HAVE_DECL_FLSL */ + +/* Define to 1 if you have the `ffs' function. */ +/* #undef HWLOC_HAVE_FFS */ + +/* Define to 1 if you have the `ffsl' function. */ +/* #undef HWLOC_HAVE_FFSL */ + +/* Define to 1 if you have the `fls' function. */ +/* #undef HWLOC_HAVE_FLS */ + +/* Define to 1 if you have the `flsl' function. */ +/* #undef HWLOC_HAVE_FLSL */ + +/* Define to 1 if you have the GL module components. */ +/* #undef HWLOC_HAVE_GL */ + +/* Define to 1 if you have a library providing the termcap interface */ +/* #undef HWLOC_HAVE_LIBTERMCAP */ + +/* Define to 1 if you have the `libxml2' library. */ +/* #undef HWLOC_HAVE_LIBXML2 */ + +/* Define to 1 if building the Linux PCI component */ +/* #undef HWLOC_HAVE_LINUXPCI */ + +/* Define to 1 if you have the `NVML' library. */ +/* #undef HWLOC_HAVE_NVML */ + +/* Define to 1 if glibc provides the old prototype (without length) of + sched_setaffinity() */ +/* #undef HWLOC_HAVE_OLD_SCHED_SETAFFINITY */ + +/* Define to 1 if you have the `OpenCL' library. */ +/* #undef HWLOC_HAVE_OPENCL */ + +/* Define to 1 if the hwloc library should support dynamically-loaded plugins + */ +/* #undef HWLOC_HAVE_PLUGINS */ + +/* `Define to 1 if you have pthread_getthrds_np' */ +/* #undef HWLOC_HAVE_PTHREAD_GETTHRDS_NP */ + +/* Define to 1 if pthread mutexes are available */ +/* #undef HWLOC_HAVE_PTHREAD_MUTEX */ + +/* Define to 1 if glibc provides a prototype of sched_setaffinity() */ +#define HWLOC_HAVE_SCHED_SETAFFINITY 1 + +/* Define to 1 if you have the header file. */ +#define HWLOC_HAVE_STDINT_H 1 + +/* Define to 1 if you have the `windows.h' header. */ +#define HWLOC_HAVE_WINDOWS_H 1 + +/* Define to 1 if X11 headers including Xutil.h and keysym.h are available. */ +/* #undef HWLOC_HAVE_X11_KEYSYM */ + +/* Define to 1 if function `syscall' is available */ +/* #undef HWLOC_HAVE_SYSCALL */ + +/* Define to 1 on HP-UX */ +/* #undef HWLOC_HPUX_SYS */ + +/* Define to 1 on Linux */ +/* #undef HWLOC_LINUX_SYS */ + +/* Define to 1 on *NETBSD */ +/* #undef HWLOC_NETBSD_SYS */ + +/* The size of `unsigned int', as computed by sizeof */ +#define HWLOC_SIZEOF_UNSIGNED_INT 4 + +/* The size of `unsigned long', as computed by sizeof */ +#define HWLOC_SIZEOF_UNSIGNED_LONG 4 + +/* Define to 1 on Solaris */ +/* #undef HWLOC_SOLARIS_SYS */ + +/* The hwloc symbol prefix */ +#define HWLOC_SYM_PREFIX hwloc_ + +/* The hwloc symbol prefix in all caps */ +#define HWLOC_SYM_PREFIX_CAPS HWLOC_ + +/* Whether we need to re-define all the hwloc public symbols or not */ +#define HWLOC_SYM_TRANSFORM 0 + +/* Define to 1 on unsupported systems */ +/* #undef HWLOC_UNSUPPORTED_SYS */ + +/* Define to 1 if ncurses works, preferred over curses */ +/* #undef HWLOC_USE_NCURSES */ + +/* Define to 1 on WINDOWS */ +#define HWLOC_WIN_SYS 1 + +/* Define to 1 on x86_32 */ +/* #undef HWLOC_X86_32_ARCH */ + +/* Define to 1 on x86_64 */ +#define HWLOC_X86_64_ARCH 1 + +/* Define to the sub-directory in which libtool stores uninstalled libraries. + */ +#define LT_OBJDIR ".libs/" + +/* Name of package */ +#define PACKAGE "hwloc" + +/* Define to the address where bug reports for this package should be sent. */ +#define PACKAGE_BUGREPORT "http://www.open-mpi.org/projects/hwloc/" + +/* Define to the full name of this package. */ +#define PACKAGE_NAME "hwloc" + +/* Define to the full name and version of this package. */ +#define PACKAGE_STRING "hwloc" + +/* Define to the one symbol short name of this package. */ +#define PACKAGE_TARNAME "hwloc" + +/* Define to the home page for this package. */ +#define PACKAGE_URL "" + +/* Define to the version of this package. */ +#define PACKAGE_VERSION HWLOC_VERSION + +/* The size of `unsigned int', as computed by sizeof. */ +#define SIZEOF_UNSIGNED_INT 4 + +/* The size of `unsigned long', as computed by sizeof. */ +#define SIZEOF_UNSIGNED_LONG 4 + +/* The size of `void *', as computed by sizeof. */ +#define SIZEOF_VOID_P 8 + +/* Define to 1 if you have the ANSI C header files. */ +#define STDC_HEADERS 1 + +/* Enable extensions on HP-UX. */ +#ifndef _HPUX_SOURCE +# define _HPUX_SOURCE 1 +#endif + + +/* Enable extensions on AIX 3, Interix. */ +/* +#ifndef _ALL_SOURCE +# define _ALL_SOURCE 1 +#endif +*/ + +/* Enable GNU extensions on systems that have them. */ +/* +#ifndef _GNU_SOURCE +# define _GNU_SOURCE 1 +#endif +*/ +/* Enable threading extensions on Solaris. */ +/* +#ifndef _POSIX_PTHREAD_SEMANTICS +# define _POSIX_PTHREAD_SEMANTICS 1 +#endif +*/ +/* Enable extensions on HP NonStop. */ +/* +#ifndef _TANDEM_SOURCE +# define _TANDEM_SOURCE 1 +#endif +*/ +/* Enable general extensions on Solaris. */ +/* +#ifndef __EXTENSIONS__ +# define __EXTENSIONS__ 1 +#endif +*/ + + +/* Version number of package */ +#define VERSION HWLOC_VERSION + +/* Define to 1 if the X Window System is missing or not being used. */ +#define X_DISPLAY_MISSING 1 + +/* Define to 1 if on MINIX. */ +/* #undef _MINIX */ + +/* Define to 2 if the system does not provide POSIX.1 features except with + this defined. */ +/* #undef _POSIX_1_SOURCE */ + +/* Define to 1 if you need to in order for `stat' and other things to work. */ +/* #undef _POSIX_SOURCE */ + +/* Define this to the process ID type */ +#define hwloc_pid_t HANDLE + +/* Define this to either strncasecmp or strncmp */ +#define hwloc_strncasecmp strncasecmp + +/* Define this to the thread ID type */ +#define hwloc_thread_t HANDLE + + +#endif /* HWLOC_CONFIGURE_H */ diff --git a/src/3rdparty/hwloc/include/private/components.h b/src/3rdparty/hwloc/include/private/components.h new file mode 100644 index 000000000..8525bbe46 --- /dev/null +++ b/src/3rdparty/hwloc/include/private/components.h @@ -0,0 +1,43 @@ +/* + * Copyright © 2012-2015 Inria. All rights reserved. + * See COPYING in top-level directory. + */ + + +#ifdef HWLOC_INSIDE_PLUGIN +/* + * these declarations are internal only, they are not available to plugins + * (many functions below are internal static symbols). + */ +#error This file should not be used in plugins +#endif + + +#ifndef PRIVATE_COMPONENTS_H +#define PRIVATE_COMPONENTS_H 1 + +#include + +struct hwloc_topology; + +extern int hwloc_disc_component_force_enable(struct hwloc_topology *topology, + int envvar_forced, /* 1 if forced through envvar, 0 if forced through API */ + int type, const char *name, + const void *data1, const void *data2, const void *data3); +extern void hwloc_disc_components_enable_others(struct hwloc_topology *topology); + +/* Compute the topology is_thissystem flag and find some callbacks based on enabled backends */ +extern void hwloc_backends_is_thissystem(struct hwloc_topology *topology); +extern void hwloc_backends_find_callbacks(struct hwloc_topology *topology); + +/* Initialize the list of backends used by a topology */ +extern void hwloc_backends_init(struct hwloc_topology *topology); +/* Disable and destroy all backends used by a topology */ +extern void hwloc_backends_disable_all(struct hwloc_topology *topology); + +/* Used by the core to setup/destroy the list of components */ +extern void hwloc_components_init(void); /* increases components refcount, should be called exactly once per topology (during init) */ +extern void hwloc_components_fini(void); /* decreases components refcount, should be called exactly once per topology (during destroy) */ + +#endif /* PRIVATE_COMPONENTS_H */ + diff --git a/src/3rdparty/hwloc/include/private/cpuid-x86.h b/src/3rdparty/hwloc/include/private/cpuid-x86.h new file mode 100644 index 000000000..2758afe04 --- /dev/null +++ b/src/3rdparty/hwloc/include/private/cpuid-x86.h @@ -0,0 +1,86 @@ +/* + * Copyright © 2010-2012, 2014 Université Bordeaux + * Copyright © 2010 Cisco Systems, Inc. All rights reserved. + * Copyright © 2014 Inria. All rights reserved. + * + * See COPYING in top-level directory. + */ + +/* Internals for x86's cpuid. */ + +#ifndef HWLOC_PRIVATE_CPUID_X86_H +#define HWLOC_PRIVATE_CPUID_X86_H + +#if (defined HWLOC_X86_32_ARCH) && (!defined HWLOC_HAVE_MSVC_CPUIDEX) +static __hwloc_inline int hwloc_have_x86_cpuid(void) +{ + int ret; + unsigned tmp, tmp2; + __asm__( + "mov $0,%0\n\t" /* Not supported a priori */ + + "pushfl \n\t" /* Save flags */ + + "pushfl \n\t" \ + "pop %1 \n\t" /* Get flags */ \ + +#define TRY_TOGGLE \ + "xor $0x00200000,%1\n\t" /* Try to toggle ID */ \ + "mov %1,%2\n\t" /* Save expected value */ \ + "push %1 \n\t" \ + "popfl \n\t" /* Try to toggle */ \ + "pushfl \n\t" \ + "pop %1 \n\t" \ + "cmp %1,%2\n\t" /* Compare with expected value */ \ + "jnz 0f\n\t" /* Unexpected, failure */ \ + + TRY_TOGGLE /* Try to set/clear */ + TRY_TOGGLE /* Try to clear/set */ + + "mov $1,%0\n\t" /* Passed the test! */ + + "0: \n\t" + "popfl \n\t" /* Restore flags */ + + : "=r" (ret), "=&r" (tmp), "=&r" (tmp2)); + return ret; +} +#endif /* !defined HWLOC_X86_32_ARCH && !defined HWLOC_HAVE_MSVC_CPUIDEX*/ +#if (defined HWLOC_X86_64_ARCH) || (defined HWLOC_HAVE_MSVC_CPUIDEX) +static __hwloc_inline int hwloc_have_x86_cpuid(void) { return 1; } +#endif /* HWLOC_X86_64_ARCH */ + +static __hwloc_inline void hwloc_x86_cpuid(unsigned *eax, unsigned *ebx, unsigned *ecx, unsigned *edx) +{ +#ifdef HWLOC_HAVE_MSVC_CPUIDEX + int regs[4]; + __cpuidex(regs, *eax, *ecx); + *eax = regs[0]; + *ebx = regs[1]; + *ecx = regs[2]; + *edx = regs[3]; +#else /* HWLOC_HAVE_MSVC_CPUIDEX */ + /* Note: gcc might want to use bx or the stack for %1 addressing, so we can't + * use them :/ */ +#ifdef HWLOC_X86_64_ARCH + hwloc_uint64_t sav_rbx; + __asm__( + "mov %%rbx,%2\n\t" + "cpuid\n\t" + "xchg %2,%%rbx\n\t" + "movl %k2,%1\n\t" + : "+a" (*eax), "=m" (*ebx), "=&r"(sav_rbx), + "+c" (*ecx), "=&d" (*edx)); +#elif defined(HWLOC_X86_32_ARCH) + __asm__( + "mov %%ebx,%1\n\t" + "cpuid\n\t" + "xchg %%ebx,%1\n\t" + : "+a" (*eax), "=&SD" (*ebx), "+c" (*ecx), "=&d" (*edx)); +#else +#error unknown architecture +#endif +#endif /* HWLOC_HAVE_MSVC_CPUIDEX */ +} + +#endif /* HWLOC_PRIVATE_X86_CPUID_H */ diff --git a/src/3rdparty/hwloc/include/private/debug.h b/src/3rdparty/hwloc/include/private/debug.h new file mode 100644 index 000000000..74b697db4 --- /dev/null +++ b/src/3rdparty/hwloc/include/private/debug.h @@ -0,0 +1,83 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2017 Inria. All rights reserved. + * Copyright © 2009, 2011 Université Bordeaux + * Copyright © 2011 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +/* The configuration file */ + +#ifndef HWLOC_DEBUG_H +#define HWLOC_DEBUG_H + +#include +#include + +#ifdef HWLOC_DEBUG +#include +#include +#endif + +/* Compile-time assertion */ +#define HWLOC_BUILD_ASSERT(condition) ((void)sizeof(char[1 - 2*!(condition)])) + +#ifdef HWLOC_DEBUG +static __hwloc_inline int hwloc_debug_enabled(void) +{ + static int checked = 0; + static int enabled = 1; + if (!checked) { + const char *env = getenv("HWLOC_DEBUG_VERBOSE"); + if (env) + enabled = atoi(env); + if (enabled) + fprintf(stderr, "hwloc verbose debug enabled, may be disabled with HWLOC_DEBUG_VERBOSE=0 in the environment.\n"); + checked = 1; + } + return enabled; +} +#endif + +static __hwloc_inline void hwloc_debug(const char *s __hwloc_attribute_unused, ...) __hwloc_attribute_format(printf, 1, 2); +static __hwloc_inline void hwloc_debug(const char *s __hwloc_attribute_unused, ...) +{ +#ifdef HWLOC_DEBUG + if (hwloc_debug_enabled()) { + va_list ap; + va_start(ap, s); + vfprintf(stderr, s, ap); + va_end(ap); + } +#endif +} + +#ifdef HWLOC_DEBUG +#define hwloc_debug_bitmap(fmt, bitmap) do { \ +if (hwloc_debug_enabled()) { \ + char *s; \ + hwloc_bitmap_asprintf(&s, bitmap); \ + fprintf(stderr, fmt, s); \ + free(s); \ +} } while (0) +#define hwloc_debug_1arg_bitmap(fmt, arg1, bitmap) do { \ +if (hwloc_debug_enabled()) { \ + char *s; \ + hwloc_bitmap_asprintf(&s, bitmap); \ + fprintf(stderr, fmt, arg1, s); \ + free(s); \ +} } while (0) +#define hwloc_debug_2args_bitmap(fmt, arg1, arg2, bitmap) do { \ +if (hwloc_debug_enabled()) { \ + char *s; \ + hwloc_bitmap_asprintf(&s, bitmap); \ + fprintf(stderr, fmt, arg1, arg2, s); \ + free(s); \ +} } while (0) +#else +#define hwloc_debug_bitmap(s, bitmap) do { } while(0) +#define hwloc_debug_1arg_bitmap(s, arg1, bitmap) do { } while(0) +#define hwloc_debug_2args_bitmap(s, arg1, arg2, bitmap) do { } while(0) +#endif + +#endif /* HWLOC_DEBUG_H */ diff --git a/src/3rdparty/hwloc/include/private/internal-components.h b/src/3rdparty/hwloc/include/private/internal-components.h new file mode 100644 index 000000000..b138a0eb9 --- /dev/null +++ b/src/3rdparty/hwloc/include/private/internal-components.h @@ -0,0 +1,41 @@ +/* + * Copyright © 2018 Inria. All rights reserved. + * + * See COPYING in top-level directory. + */ + +/* List of components defined inside hwloc */ + +#ifndef PRIVATE_INTERNAL_COMPONENTS_H +#define PRIVATE_INTERNAL_COMPONENTS_H + +/* global discovery */ +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_xml_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_synthetic_component; + +/* CPU discovery */ +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_aix_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_bgq_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_darwin_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_freebsd_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_hpux_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_linux_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_netbsd_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_noos_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_solaris_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_windows_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_x86_component; + +/* I/O discovery */ +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_cuda_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_gl_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_linuxio_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_nvml_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_opencl_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_pci_component; + +/* XML backend */ +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_xml_nolibxml_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_xml_libxml_component; + +#endif /* PRIVATE_INTERNAL_COMPONENTS_H */ diff --git a/src/3rdparty/hwloc/include/private/misc.h b/src/3rdparty/hwloc/include/private/misc.h new file mode 100644 index 000000000..66608bc79 --- /dev/null +++ b/src/3rdparty/hwloc/include/private/misc.h @@ -0,0 +1,583 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2018 Inria. All rights reserved. + * Copyright © 2009-2012 Université Bordeaux + * Copyright © 2011 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +/* Misc macros and inlines. */ + +#ifndef HWLOC_PRIVATE_MISC_H +#define HWLOC_PRIVATE_MISC_H + +#include +#include +#include + +#ifdef HWLOC_HAVE_DECL_STRNCASECMP +#ifdef HAVE_STRINGS_H +#include +#endif +#else +#ifdef HAVE_CTYPE_H +#include +#endif +#endif + +#define HWLOC_BITS_PER_LONG (HWLOC_SIZEOF_UNSIGNED_LONG * 8) +#define HWLOC_BITS_PER_INT (HWLOC_SIZEOF_UNSIGNED_INT * 8) + +#if (HWLOC_BITS_PER_LONG != 32) && (HWLOC_BITS_PER_LONG != 64) +#error "unknown size for unsigned long." +#endif + +#if (HWLOC_BITS_PER_INT != 16) && (HWLOC_BITS_PER_INT != 32) && (HWLOC_BITS_PER_INT != 64) +#error "unknown size for unsigned int." +#endif + +/* internal-use-only value for when we don't know the type or don't have any value */ +#define HWLOC_OBJ_TYPE_NONE ((hwloc_obj_type_t) -1) + +/** + * ffsl helpers. + */ + +#if defined(HWLOC_HAVE_BROKEN_FFS) + +/* System has a broken ffs(). + * We must check the before __GNUC__ or HWLOC_HAVE_FFSL + */ +# define HWLOC_NO_FFS + +#elif defined(__GNUC__) + +# if (__GNUC__ >= 4) || ((__GNUC__ == 3) && (__GNUC_MINOR__ >= 4)) + /* Starting from 3.4, gcc has a long variant. */ +# define hwloc_ffsl(x) __builtin_ffsl(x) +# else +# define hwloc_ffs(x) __builtin_ffs(x) +# define HWLOC_NEED_FFSL +# endif + +#elif defined(HWLOC_HAVE_FFSL) + +# ifndef HWLOC_HAVE_DECL_FFSL +extern int ffsl(long) __hwloc_attribute_const; +# endif + +# define hwloc_ffsl(x) ffsl(x) + +#elif defined(HWLOC_HAVE_FFS) + +# ifndef HWLOC_HAVE_DECL_FFS +extern int ffs(int) __hwloc_attribute_const; +# endif + +# define hwloc_ffs(x) ffs(x) +# define HWLOC_NEED_FFSL + +#else /* no ffs implementation */ + +# define HWLOC_NO_FFS + +#endif + +#ifdef HWLOC_NO_FFS + +/* no ffs or it is known to be broken */ +static __hwloc_inline int +hwloc_ffsl_manual(unsigned long x) __hwloc_attribute_const; +static __hwloc_inline int +hwloc_ffsl_manual(unsigned long x) +{ + int i; + + if (!x) + return 0; + + i = 1; +#if HWLOC_BITS_PER_LONG >= 64 + if (!(x & 0xfffffffful)) { + x >>= 32; + i += 32; + } +#endif + if (!(x & 0xffffu)) { + x >>= 16; + i += 16; + } + if (!(x & 0xff)) { + x >>= 8; + i += 8; + } + if (!(x & 0xf)) { + x >>= 4; + i += 4; + } + if (!(x & 0x3)) { + x >>= 2; + i += 2; + } + if (!(x & 0x1)) { + x >>= 1; + i += 1; + } + + return i; +} +/* always define hwloc_ffsl as a macro, to avoid renaming breakage */ +#define hwloc_ffsl hwloc_ffsl_manual + +#elif defined(HWLOC_NEED_FFSL) + +/* We only have an int ffs(int) implementation, build a long one. */ + +/* First make it 32 bits if it was only 16. */ +static __hwloc_inline int +hwloc_ffs32(unsigned long x) __hwloc_attribute_const; +static __hwloc_inline int +hwloc_ffs32(unsigned long x) +{ +#if HWLOC_BITS_PER_INT == 16 + int low_ffs, hi_ffs; + + low_ffs = hwloc_ffs(x & 0xfffful); + if (low_ffs) + return low_ffs; + + hi_ffs = hwloc_ffs(x >> 16); + if (hi_ffs) + return hi_ffs + 16; + + return 0; +#else + return hwloc_ffs(x); +#endif +} + +/* Then make it 64 bit if longs are. */ +static __hwloc_inline int +hwloc_ffsl_from_ffs32(unsigned long x) __hwloc_attribute_const; +static __hwloc_inline int +hwloc_ffsl_from_ffs32(unsigned long x) +{ +#if HWLOC_BITS_PER_LONG == 64 + int low_ffs, hi_ffs; + + low_ffs = hwloc_ffs32(x & 0xfffffffful); + if (low_ffs) + return low_ffs; + + hi_ffs = hwloc_ffs32(x >> 32); + if (hi_ffs) + return hi_ffs + 32; + + return 0; +#else + return hwloc_ffs32(x); +#endif +} +/* always define hwloc_ffsl as a macro, to avoid renaming breakage */ +#define hwloc_ffsl hwloc_ffsl_from_ffs32 + +#endif + +/** + * flsl helpers. + */ +#ifdef __GNUC_____ + +# if (__GNUC__ >= 4) || ((__GNUC__ == 3) && (__GNUC_MINOR__ >= 4)) +# define hwloc_flsl(x) ((x) ? (8*sizeof(long) - __builtin_clzl(x)) : 0) +# else +# define hwloc_fls(x) ((x) ? (8*sizeof(int) - __builtin_clz(x)) : 0) +# define HWLOC_NEED_FLSL +# endif + +#elif defined(HWLOC_HAVE_FLSL) + +# ifndef HWLOC_HAVE_DECL_FLSL +extern int flsl(long) __hwloc_attribute_const; +# endif + +# define hwloc_flsl(x) flsl(x) + +#elif defined(HWLOC_HAVE_CLZL) + +# ifndef HWLOC_HAVE_DECL_CLZL +extern int clzl(long) __hwloc_attribute_const; +# endif + +# define hwloc_flsl(x) ((x) ? (8*sizeof(long) - clzl(x)) : 0) + +#elif defined(HWLOC_HAVE_FLS) + +# ifndef HWLOC_HAVE_DECL_FLS +extern int fls(int) __hwloc_attribute_const; +# endif + +# define hwloc_fls(x) fls(x) +# define HWLOC_NEED_FLSL + +#elif defined(HWLOC_HAVE_CLZ) + +# ifndef HWLOC_HAVE_DECL_CLZ +extern int clz(int) __hwloc_attribute_const; +# endif + +# define hwloc_fls(x) ((x) ? (8*sizeof(int) - clz(x)) : 0) +# define HWLOC_NEED_FLSL + +#else /* no fls implementation */ + +static __hwloc_inline int +hwloc_flsl_manual(unsigned long x) __hwloc_attribute_const; +static __hwloc_inline int +hwloc_flsl_manual(unsigned long x) +{ + int i = 0; + + if (!x) + return 0; + + i = 1; +#if HWLOC_BITS_PER_LONG >= 64 + if ((x & 0xffffffff00000000ul)) { + x >>= 32; + i += 32; + } +#endif + if ((x & 0xffff0000u)) { + x >>= 16; + i += 16; + } + if ((x & 0xff00)) { + x >>= 8; + i += 8; + } + if ((x & 0xf0)) { + x >>= 4; + i += 4; + } + if ((x & 0xc)) { + x >>= 2; + i += 2; + } + if ((x & 0x2)) { + x >>= 1; + i += 1; + } + + return i; +} +/* always define hwloc_flsl as a macro, to avoid renaming breakage */ +#define hwloc_flsl hwloc_flsl_manual + +#endif + +#ifdef HWLOC_NEED_FLSL + +/* We only have an int fls(int) implementation, build a long one. */ + +/* First make it 32 bits if it was only 16. */ +static __hwloc_inline int +hwloc_fls32(unsigned long x) __hwloc_attribute_const; +static __hwloc_inline int +hwloc_fls32(unsigned long x) +{ +#if HWLOC_BITS_PER_INT == 16 + int low_fls, hi_fls; + + hi_fls = hwloc_fls(x >> 16); + if (hi_fls) + return hi_fls + 16; + + low_fls = hwloc_fls(x & 0xfffful); + if (low_fls) + return low_fls; + + return 0; +#else + return hwloc_fls(x); +#endif +} + +/* Then make it 64 bit if longs are. */ +static __hwloc_inline int +hwloc_flsl_from_fls32(unsigned long x) __hwloc_attribute_const; +static __hwloc_inline int +hwloc_flsl_from_fls32(unsigned long x) +{ +#if HWLOC_BITS_PER_LONG == 64 + int low_fls, hi_fls; + + hi_fls = hwloc_fls32(x >> 32); + if (hi_fls) + return hi_fls + 32; + + low_fls = hwloc_fls32(x & 0xfffffffful); + if (low_fls) + return low_fls; + + return 0; +#else + return hwloc_fls32(x); +#endif +} +/* always define hwloc_flsl as a macro, to avoid renaming breakage */ +#define hwloc_flsl hwloc_flsl_from_fls32 + +#endif + +static __hwloc_inline int +hwloc_weight_long(unsigned long w) __hwloc_attribute_const; +static __hwloc_inline int +hwloc_weight_long(unsigned long w) +{ +#if HWLOC_BITS_PER_LONG == 32 +#if (__GNUC__ >= 4) || ((__GNUC__ == 3) && (__GNUC_MINOR__) >= 4) + return __builtin_popcount(w); +#else + unsigned int res = (w & 0x55555555) + ((w >> 1) & 0x55555555); + res = (res & 0x33333333) + ((res >> 2) & 0x33333333); + res = (res & 0x0F0F0F0F) + ((res >> 4) & 0x0F0F0F0F); + res = (res & 0x00FF00FF) + ((res >> 8) & 0x00FF00FF); + return (res & 0x0000FFFF) + ((res >> 16) & 0x0000FFFF); +#endif +#else /* HWLOC_BITS_PER_LONG == 32 */ +#if (__GNUC__ >= 4) || ((__GNUC__ == 3) && (__GNUC_MINOR__) >= 4) + return __builtin_popcountll(w); +#else + unsigned long res; + res = (w & 0x5555555555555555ul) + ((w >> 1) & 0x5555555555555555ul); + res = (res & 0x3333333333333333ul) + ((res >> 2) & 0x3333333333333333ul); + res = (res & 0x0F0F0F0F0F0F0F0Ful) + ((res >> 4) & 0x0F0F0F0F0F0F0F0Ful); + res = (res & 0x00FF00FF00FF00FFul) + ((res >> 8) & 0x00FF00FF00FF00FFul); + res = (res & 0x0000FFFF0000FFFFul) + ((res >> 16) & 0x0000FFFF0000FFFFul); + return (res & 0x00000000FFFFFFFFul) + ((res >> 32) & 0x00000000FFFFFFFFul); +#endif +#endif /* HWLOC_BITS_PER_LONG == 64 */ +} + +#if !HAVE_DECL_STRTOULL && defined(HAVE_STRTOULL) +unsigned long long int strtoull(const char *nptr, char **endptr, int base); +#endif + +static __hwloc_inline int hwloc_strncasecmp(const char *s1, const char *s2, size_t n) +{ +#ifdef HWLOC_HAVE_DECL_STRNCASECMP + return strncasecmp(s1, s2, n); +#else + while (n) { + char c1 = tolower(*s1), c2 = tolower(*s2); + if (!c1 || !c2 || c1 != c2) + return c1-c2; + n--; s1++; s2++; + } + return 0; +#endif +} + +static __hwloc_inline hwloc_obj_type_t hwloc_cache_type_by_depth_type(unsigned depth, hwloc_obj_cache_type_t type) +{ + if (type == HWLOC_OBJ_CACHE_INSTRUCTION) { + if (depth >= 1 && depth <= 3) + return HWLOC_OBJ_L1ICACHE + depth-1; + else + return HWLOC_OBJ_TYPE_NONE; + } else { + if (depth >= 1 && depth <= 5) + return HWLOC_OBJ_L1CACHE + depth-1; + else + return HWLOC_OBJ_TYPE_NONE; + } +} + +#define HWLOC_BITMAP_EQUAL 0 /* Bitmaps are equal */ +#define HWLOC_BITMAP_INCLUDED 1 /* First bitmap included in second */ +#define HWLOC_BITMAP_CONTAINS 2 /* First bitmap contains second */ +#define HWLOC_BITMAP_INTERSECTS 3 /* Bitmaps intersect without any inclusion */ +#define HWLOC_BITMAP_DIFFERENT 4 /* Bitmaps do not intersect */ + +/* Compare bitmaps \p bitmap1 and \p bitmap2 from an inclusion point of view. */ +HWLOC_DECLSPEC int hwloc_bitmap_compare_inclusion(hwloc_const_bitmap_t bitmap1, hwloc_const_bitmap_t bitmap2) __hwloc_attribute_pure; + +/* Return a stringified PCI class. */ +HWLOC_DECLSPEC extern const char * hwloc_pci_class_string(unsigned short class_id); + +/* Parse a PCI link speed (GT/s) string from Linux sysfs */ +#ifdef HWLOC_LINUX_SYS +#include /* for atof() */ +static __hwloc_inline float +hwloc_linux_pci_link_speed_from_string(const char *string) +{ + /* don't parse Gen1 with atof() since it expects a localized string + * while the kernel sysfs files aren't. + */ + if (!strncmp(string, "2.5 ", 4)) + /* "2.5 GT/s" is Gen1 with 8/10 encoding */ + return 2.5 * .8; + + /* also hardwire Gen2 since it also has a specific encoding */ + if (!strncmp(string, "5 ", 2)) + /* "5 GT/s" is Gen2 with 8/10 encoding */ + return 5 * .8; + + /* handle Gen3+ in a generic way */ + return atof(string) * 128./130; /* Gen3+ encoding is 128/130 */ +} +#endif + +/* Traverse children of a parent */ +#define for_each_child(child, parent) for(child = parent->first_child; child; child = child->next_sibling) +#define for_each_memory_child(child, parent) for(child = parent->memory_first_child; child; child = child->next_sibling) +#define for_each_io_child(child, parent) for(child = parent->io_first_child; child; child = child->next_sibling) +#define for_each_misc_child(child, parent) for(child = parent->misc_first_child; child; child = child->next_sibling) + +/* Any object attached to normal children */ +static __hwloc_inline int hwloc__obj_type_is_normal (hwloc_obj_type_t type) +{ + /* type contiguity is asserted in topology_check() */ + return type <= HWLOC_OBJ_GROUP; +} + +/* Any object attached to memory children, currently only NUMA nodes */ +static __hwloc_inline int hwloc__obj_type_is_memory (hwloc_obj_type_t type) +{ + /* type contiguity is asserted in topology_check() */ + return type == HWLOC_OBJ_NUMANODE; +} + +/* I/O or Misc object, without cpusets or nodesets. */ +static __hwloc_inline int hwloc__obj_type_is_special (hwloc_obj_type_t type) +{ + /* type contiguity is asserted in topology_check() */ + return type >= HWLOC_OBJ_BRIDGE && type <= HWLOC_OBJ_MISC; +} + +/* Any object attached to io children */ +static __hwloc_inline int hwloc__obj_type_is_io (hwloc_obj_type_t type) +{ + /* type contiguity is asserted in topology_check() */ + return type >= HWLOC_OBJ_BRIDGE && type <= HWLOC_OBJ_OS_DEVICE; +} + +static __hwloc_inline int +hwloc__obj_type_is_cache(hwloc_obj_type_t type) +{ + /* type contiguity is asserted in topology_check() */ + return (type >= HWLOC_OBJ_L1CACHE && type <= HWLOC_OBJ_L3ICACHE); +} + +static __hwloc_inline int +hwloc__obj_type_is_dcache(hwloc_obj_type_t type) +{ + /* type contiguity is asserted in topology_check() */ + return (type >= HWLOC_OBJ_L1CACHE && type <= HWLOC_OBJ_L5CACHE); +} + +/** \brief Check whether an object is a Instruction Cache. */ +static __hwloc_inline int +hwloc__obj_type_is_icache(hwloc_obj_type_t type) +{ + /* type contiguity is asserted in topology_check() */ + return (type >= HWLOC_OBJ_L1ICACHE && type <= HWLOC_OBJ_L3ICACHE); +} + +#ifdef HAVE_USELOCALE +#include "locale.h" +#ifdef HAVE_XLOCALE_H +#include "xlocale.h" +#endif +#define hwloc_localeswitch_declare locale_t __old_locale = (locale_t)0, __new_locale +#define hwloc_localeswitch_init() do { \ + __new_locale = newlocale(LC_ALL_MASK, "C", (locale_t)0); \ + if (__new_locale != (locale_t)0) \ + __old_locale = uselocale(__new_locale); \ +} while (0) +#define hwloc_localeswitch_fini() do { \ + if (__new_locale != (locale_t)0) { \ + uselocale(__old_locale); \ + freelocale(__new_locale); \ + } \ +} while(0) +#else /* HAVE_USELOCALE */ +#if __HWLOC_HAVE_ATTRIBUTE_UNUSED +#define hwloc_localeswitch_declare int __dummy_nolocale __hwloc_attribute_unused +#define hwloc_localeswitch_init() +#else +#define hwloc_localeswitch_declare int __dummy_nolocale +#define hwloc_localeswitch_init() (void)__dummy_nolocale +#endif +#define hwloc_localeswitch_fini() +#endif /* HAVE_USELOCALE */ + +#if !HAVE_DECL_FABSF +#define fabsf(f) fabs((double)(f)) +#endif + +#if !HAVE_DECL_MODFF +#define modff(x,iptr) (float)modf((double)x,(double *)iptr) +#endif + +#if HAVE_DECL__SC_PAGE_SIZE +#define hwloc_getpagesize() sysconf(_SC_PAGE_SIZE) +#elif HAVE_DECL__SC_PAGESIZE +#define hwloc_getpagesize() sysconf(_SC_PAGESIZE) +#elif defined HAVE_GETPAGESIZE +#define hwloc_getpagesize() getpagesize() +#else +#undef hwloc_getpagesize +#endif + +#if HWLOC_HAVE_ATTRIBUTE_FORMAT +# define __hwloc_attribute_format(type, str, arg) __attribute__((__format__(type, str, arg))) +#else +# define __hwloc_attribute_format(type, str, arg) +#endif + +#define hwloc_memory_size_printf_value(_size, _verbose) \ + ((_size) < (10ULL<<20) || (_verbose) ? (((_size)>>9)+1)>>1 : (_size) < (10ULL<<30) ? (((_size)>>19)+1)>>1 : (_size) < (10ULL<<40) ? (((_size)>>29)+1)>>1 : (((_size)>>39)+1)>>1) +#define hwloc_memory_size_printf_unit(_size, _verbose) \ + ((_size) < (10ULL<<20) || (_verbose) ? "KB" : (_size) < (10ULL<<30) ? "MB" : (_size) < (10ULL<<40) ? "GB" : "TB") + +#ifdef HWLOC_WIN_SYS +# ifndef HAVE_SSIZE_T +typedef SSIZE_T ssize_t; +# endif +# if !HAVE_DECL_STRTOULL && !defined(HAVE_STRTOULL) +# define strtoull _strtoui64 +# endif +# ifndef S_ISREG +# define S_ISREG(m) ((m) & S_IFREG) +# endif +# ifndef S_ISDIR +# define S_ISDIR(m) (((m) & S_IFMT) == S_IFDIR) +# endif +# ifndef S_IRWXU +# define S_IRWXU 00700 +# endif +# ifndef HWLOC_HAVE_DECL_STRCASECMP +# define strcasecmp _stricmp +# endif +# if !HAVE_DECL_SNPRINTF +# define snprintf _snprintf +# endif +# if HAVE_DECL__STRDUP +# define strdup _strdup +# endif +# if HAVE_DECL__PUTENV +# define putenv _putenv +# endif +#endif + +#if defined HWLOC_WIN_SYS && !defined __MINGW32__ && !defined(__CYGWIN__) +/* MSVC doesn't support C99 variable-length array */ +#include +#define HWLOC_VLA(_type, _name, _nb) _type *_name = (_type*) _alloca((_nb)*sizeof(_type)) +#else +#define HWLOC_VLA(_type, _name, _nb) _type _name[_nb] +#endif + +#endif /* HWLOC_PRIVATE_MISC_H */ diff --git a/src/3rdparty/hwloc/include/private/netloc.h b/src/3rdparty/hwloc/include/private/netloc.h new file mode 100644 index 000000000..c070c54cc --- /dev/null +++ b/src/3rdparty/hwloc/include/private/netloc.h @@ -0,0 +1,578 @@ +/* + * Copyright © 2014 Cisco Systems, Inc. All rights reserved. + * Copyright © 2013-2014 University of Wisconsin-La Crosse. + * All rights reserved. + * Copyright © 2015-2017 Inria. All rights reserved. + * + * $COPYRIGHT$ + * + * Additional copyrights may follow + * See COPYING in top-level directory. + * + * $HEADER$ + */ + +#ifndef _NETLOC_PRIVATE_H_ +#define _NETLOC_PRIVATE_H_ + +#include +#include +#include +#include +#include + +#define NETLOCFILE_VERSION 1 + +#ifdef NETLOC_SCOTCH +#include +#include +#define NETLOC_int SCOTCH_Num +#else +#define NETLOC_int int +#endif + +/* + * "Import" a few things from hwloc + */ +#define __netloc_attribute_unused __hwloc_attribute_unused +#define __netloc_attribute_malloc __hwloc_attribute_malloc +#define __netloc_attribute_const __hwloc_attribute_const +#define __netloc_attribute_pure __hwloc_attribute_pure +#define __netloc_attribute_deprecated __hwloc_attribute_deprecated +#define __netloc_attribute_may_alias __hwloc_attribute_may_alias +#define NETLOC_DECLSPEC HWLOC_DECLSPEC + + +/********************************************************************** + * Types + **********************************************************************/ + +/** + * Definitions for Comparators + * \sa These are the return values from the following functions: + * netloc_network_compare, netloc_dt_edge_t_compare, netloc_dt_node_t_compare + */ +typedef enum { + NETLOC_CMP_SAME = 0, /**< Compared as the Same */ + NETLOC_CMP_SIMILAR = -1, /**< Compared as Similar, but not the Same */ + NETLOC_CMP_DIFF = -2 /**< Compared as Different */ +} netloc_compare_type_t; + +/** + * Enumerated type for the various types of supported networks + */ +typedef enum { + NETLOC_NETWORK_TYPE_ETHERNET = 1, /**< Ethernet network */ + NETLOC_NETWORK_TYPE_INFINIBAND = 2, /**< InfiniBand network */ + NETLOC_NETWORK_TYPE_INVALID = 3 /**< Invalid network */ +} netloc_network_type_t; + +/** + * Enumerated type for the various types of supported topologies + */ +typedef enum { + NETLOC_TOPOLOGY_TYPE_INVALID = -1, /**< Invalid */ + NETLOC_TOPOLOGY_TYPE_TREE = 1, /**< Tree */ +} netloc_topology_type_t; + +/** + * Enumerated type for the various types of nodes + */ +typedef enum { + NETLOC_NODE_TYPE_HOST = 0, /**< Host (a.k.a., network addressable endpoint - e.g., MAC Address) node */ + NETLOC_NODE_TYPE_SWITCH = 1, /**< Switch node */ + NETLOC_NODE_TYPE_INVALID = 2 /**< Invalid node */ +} netloc_node_type_t; + +typedef enum { + NETLOC_ARCH_TREE = 0, /* Fat tree */ +} netloc_arch_type_t; + + +/* Pre declarations to avoid inter dependency problems */ +/** \cond IGNORE */ +struct netloc_topology_t; +typedef struct netloc_topology_t netloc_topology_t; +struct netloc_node_t; +typedef struct netloc_node_t netloc_node_t; +struct netloc_edge_t; +typedef struct netloc_edge_t netloc_edge_t; +struct netloc_physical_link_t; +typedef struct netloc_physical_link_t netloc_physical_link_t; +struct netloc_path_t; +typedef struct netloc_path_t netloc_path_t; + +struct netloc_arch_tree_t; +typedef struct netloc_arch_tree_t netloc_arch_tree_t; +struct netloc_arch_node_t; +typedef struct netloc_arch_node_t netloc_arch_node_t; +struct netloc_arch_node_slot_t; +typedef struct netloc_arch_node_slot_t netloc_arch_node_slot_t; +struct netloc_arch_t; +typedef struct netloc_arch_t netloc_arch_t; +/** \endcond */ + +/** + * \struct netloc_topology_t + * \brief Netloc Topology Context + * + * An opaque data structure used to reference a network topology. + * + * \note Must be initialized with \ref netloc_topology_construct() + */ +struct netloc_topology_t { + /** Topology path */ + char *topopath; + /** Subnet ID */ + char *subnet_id; + + /** Node List */ + netloc_node_t *nodes; /* Hash table of nodes by physical_id */ + netloc_node_t *nodesByHostname; /* Hash table of nodes by hostname */ + + netloc_physical_link_t *physical_links; /* Hash table with physcial links */ + + /** Partition List */ + UT_array *partitions; + + /** Hwloc topology List */ + char *hwlocpath; + UT_array *topos; + hwloc_topology_t *hwloc_topos; + + /** Type of the graph */ + netloc_topology_type_t type; +}; + +/** + * \brief Netloc Node Type + * + * Represents the concept of a node (a.k.a., vertex, endpoint) within a network + * graph. This could be a server or a network switch. The \ref node_type parameter + * will distinguish the exact type of node this represents in the graph. + */ +struct netloc_node_t { + UT_hash_handle hh; /* makes this structure hashable with physical_id */ + UT_hash_handle hh2; /* makes this structure hashable with hostname */ + + /** Physical ID of the node */ + char physical_id[20]; + + /** Logical ID of the node (if any) */ + int logical_id; + + /** Type of the node */ + netloc_node_type_t type; + + /* Pointer to physical_links */ + UT_array *physical_links; + + /** Description information from discovery (if any) */ + char *description; + + /** + * Application-given private data pointer. + * Initialized to NULL, and not used by the netloc library. + */ + void * userdata; + + /** Outgoing edges from this node */ + netloc_edge_t *edges; + + UT_array *subnodes; /* the group of nodes for the virtual nodes */ + + netloc_path_t *paths; + + char *hostname; + + UT_array *partitions; /* index in the list from the topology */ + + hwloc_topology_t hwlocTopo; + int hwlocTopoIdx; +}; + +/** + * \brief Netloc Edge Type + * + * Represents the concept of a directed edge within a network graph. + * + * \note We do not point to the netloc_node_t structure directly to + * simplify the representation, and allow the information to more easily + * be entered into the data store without circular references. + * \todo JJH Is the note above still true? + */ +struct netloc_edge_t { + UT_hash_handle hh; /* makes this structure hashable */ + + netloc_node_t *dest; + + int id; + + /** Pointers to the parent node */ + netloc_node_t *node; + + /* Pointer to physical_links */ + UT_array *physical_links; + + /** total gbits of the links */ + float total_gbits; + + UT_array *partitions; /* index in the list from the topology */ + + UT_array *subnode_edges; /* for edges going to virtual nodes */ + + struct netloc_edge_t *other_way; + + /** + * Application-given private data pointer. + * Initialized to NULL, and not used by the netloc library. + */ + void * userdata; +}; + + +struct netloc_physical_link_t { + UT_hash_handle hh; /* makes this structure hashable */ + + int id; // TODO long long + netloc_node_t *src; + netloc_node_t *dest; + int ports[2]; + char *width; + char *speed; + + netloc_edge_t *edge; + + int other_way_id; + struct netloc_physical_link_t *other_way; + + UT_array *partitions; /* index in the list from the topology */ + + /** gbits of the link from speed and width */ + float gbits; + + /** Description information from discovery (if any) */ + char *description; +}; + +struct netloc_path_t { + UT_hash_handle hh; /* makes this structure hashable */ + char dest_id[20]; + UT_array *links; +}; + + +/********************************************************************** + * Architecture structures + **********************************************************************/ +struct netloc_arch_tree_t { + NETLOC_int num_levels; + NETLOC_int *degrees; + NETLOC_int *cost; +}; + +struct netloc_arch_node_t { + UT_hash_handle hh; /* makes this structure hashable */ + char *name; /* Hash key */ + netloc_node_t *node; /* Corresponding node */ + int idx_in_topo; /* idx with ghost hosts to have complete topo */ + int num_slots; /* it is not the real number of slots but the maximum slot idx */ + int *slot_idx; /* corresponding idx in slot_tree */ + int *slot_os_idx; /* corresponding os index for each leaf in tree */ + netloc_arch_tree_t *slot_tree; /* Tree built from hwloc */ + int num_current_slots; /* Number of PUs */ + NETLOC_int *current_slots; /* indices in the complete tree */ + int *slot_ranks; /* corresponding MPI rank for each leaf in tree */ +}; + +struct netloc_arch_node_slot_t { + netloc_arch_node_t *node; + int slot; +}; + +struct netloc_arch_t { + netloc_topology_t *topology; + int has_slots; /* if slots are included in the architecture */ + netloc_arch_type_t type; + union { + netloc_arch_tree_t *node_tree; + netloc_arch_tree_t *global_tree; + } arch; + netloc_arch_node_t *nodes_by_name; + netloc_arch_node_slot_t *node_slot_by_idx; /* node_slot by index in complete topo */ + NETLOC_int num_current_hosts; /* if has_slots, host is a slot, else host is a node */ + NETLOC_int *current_hosts; /* indices in the complete topology */ +}; + +/********************************************************************** + * Topology Functions + **********************************************************************/ +/** + * Allocate a topology handle. + * + * User is responsible for calling \ref netloc_detach on the topology handle. + * The network parameter information is deep copied into the topology handle, so the + * user may destruct the network handle after calling this function and/or reuse + * the network handle. + * + * \returns NETLOC_SUCCESS on success + * \returns NETLOC_ERROR upon an error. + */ +netloc_topology_t *netloc_topology_construct(char *path); + +/** + * Destruct a topology handle + * + * \param topology A valid pointer to a \ref netloc_topology_t handle created + * from a prior call to \ref netloc_topology_construct. + * + * \returns NETLOC_SUCCESS on success + * \returns NETLOC_ERROR upon an error. + */ +int netloc_topology_destruct(netloc_topology_t *topology); + +int netloc_topology_find_partition_idx(netloc_topology_t *topology, char *partition_name); + +int netloc_topology_read_hwloc(netloc_topology_t *topology, int num_nodes, + netloc_node_t **node_list); + +#define netloc_topology_iter_partitions(topology,partition) \ + for ((partition) = (char **)utarray_front(topology->partitions); \ + (partition) != NULL; \ + (partition) = (char **)utarray_next(topology->partitions, partition)) + +#define netloc_topology_iter_hwloctopos(topology,hwloctopo) \ + for ((hwloctopo) = (char **)utarray_front(topology->topos); \ + (hwloctopo) != NULL; \ + (hwloctopo) = (char **)utarray_next(topology->topos, hwloctopo)) + +#define netloc_topology_find_node(topology,node_id,node) \ + HASH_FIND_STR(topology->nodes, node_id, node) + +#define netloc_topology_iter_nodes(topology,node,_tmp) \ + HASH_ITER(hh, topology->nodes, node, _tmp) + +#define netloc_topology_num_nodes(topology) \ + HASH_COUNT(topology->nodes) + +/*************************************************/ + + +/** + * Constructor for netloc_node_t + * + * User is responsible for calling the destructor on the handle. + * + * Returns + * A newly allocated pointer to the network information. + */ +netloc_node_t *netloc_node_construct(void); + +/** + * Destructor for netloc_node_t + * + * \param node A valid node handle + * + * Returns + * NETLOC_SUCCESS on success + * NETLOC_ERROR on error + */ +int netloc_node_destruct(netloc_node_t *node); + +char *netloc_node_pretty_print(netloc_node_t* node); + +#define netloc_node_get_num_subnodes(node) \ + utarray_len((node)->subnodes) + +#define netloc_node_get_subnode(node,i) \ + (*(netloc_node_t **)utarray_eltptr((node)->subnodes, (i))) + +#define netloc_node_get_num_edges(node) \ + utarray_len((node)->edges) + +#define netloc_node_get_edge(node,i) \ + (*(netloc_edge_t **)utarray_eltptr((node)->edges, (i))) + +#define netloc_node_iter_edges(node,edge,_tmp) \ + HASH_ITER(hh, node->edges, edge, _tmp) + +#define netloc_node_iter_paths(node,path,_tmp) \ + HASH_ITER(hh, node->paths, path, _tmp) + +#define netloc_node_is_host(node) \ + (node->type == NETLOC_NODE_TYPE_HOST) + +#define netloc_node_is_switch(node) \ + (node->type == NETLOC_NODE_TYPE_SWITCH) + +#define netloc_node_iter_paths(node, path,_tmp) \ + HASH_ITER(hh, node->paths, path, _tmp) + +int netloc_node_is_in_partition(netloc_node_t *node, int partition); + +/*************************************************/ + + +/** + * Constructor for netloc_edge_t + * + * User is responsible for calling the destructor on the handle. + * + * Returns + * A newly allocated pointer to the edge information. + */ +netloc_edge_t *netloc_edge_construct(void); + +/** + * Destructor for netloc_edge_t + * + * \param edge A valid edge handle + * + * Returns + * NETLOC_SUCCESS on success + * NETLOC_ERROR on error + */ +int netloc_edge_destruct(netloc_edge_t *edge); + +char * netloc_edge_pretty_print(netloc_edge_t* edge); + +void netloc_edge_reset_uid(void); + +int netloc_edge_is_in_partition(netloc_edge_t *edge, int partition); + +#define netloc_edge_get_num_links(edge) \ + utarray_len((edge)->physical_links) + +#define netloc_edge_get_link(edge,i) \ + (*(netloc_physical_link_t **)utarray_eltptr((edge)->physical_links, (i))) + +#define netloc_edge_get_num_subedges(edge) \ + utarray_len((edge)->subnode_edges) + +#define netloc_edge_get_subedge(edge,i) \ + (*(netloc_edge_t **)utarray_eltptr((edge)->subnode_edges, (i))) + +/*************************************************/ + + +/** + * Constructor for netloc_physical_link_t + * + * User is responsible for calling the destructor on the handle. + * + * Returns + * A newly allocated pointer to the physical link information. + */ +netloc_physical_link_t * netloc_physical_link_construct(void); + +/** + * Destructor for netloc_physical_link_t + * + * Returns + * NETLOC_SUCCESS on success + * NETLOC_ERROR on error + */ +int netloc_physical_link_destruct(netloc_physical_link_t *link); + +char * netloc_link_pretty_print(netloc_physical_link_t* link); + +/*************************************************/ + + +netloc_path_t *netloc_path_construct(void); +int netloc_path_destruct(netloc_path_t *path); + + +/********************************************************************** + * Architecture functions + **********************************************************************/ + +netloc_arch_t * netloc_arch_construct(void); + +int netloc_arch_destruct(netloc_arch_t *arch); + +int netloc_arch_build(netloc_arch_t *arch, int add_slots); + +int netloc_arch_set_current_resources(netloc_arch_t *arch); + +int netloc_arch_set_global_resources(netloc_arch_t *arch); + +int netloc_arch_node_get_hwloc_info(netloc_arch_node_t *arch); + +void netloc_arch_tree_complete(netloc_arch_tree_t *tree, UT_array **down_degrees_by_level, + int num_hosts, int **parch_idx); + +NETLOC_int netloc_arch_tree_num_leaves(netloc_arch_tree_t *tree); + + +/********************************************************************** + * Access functions of various elements of the topology + **********************************************************************/ + +#define netloc_get_num_partitions(object) \ + utarray_len((object)->partitions) + +#define netloc_get_partition(object,i) \ + (*(int *)utarray_eltptr((object)->partitions, (i))) + + +#define netloc_path_iter_links(path,link) \ + for ((link) = (netloc_physical_link_t **)utarray_front(path->links); \ + (link) != NULL; \ + (link) = (netloc_physical_link_t **)utarray_next(path->links, link)) + +/********************************************************************** + * Misc functions + **********************************************************************/ + +/** + * Decode the network type + * + * \param net_type A valid member of the \ref netloc_network_type_t type + * + * \returns NULL if the type is invalid + * \returns A string for that \ref netloc_network_type_t type + */ +static inline const char * netloc_network_type_decode(netloc_network_type_t net_type) { + if( NETLOC_NETWORK_TYPE_ETHERNET == net_type ) { + return "ETH"; + } + else if( NETLOC_NETWORK_TYPE_INFINIBAND == net_type ) { + return "IB"; + } + else { + return NULL; + } +} + +/** + * Decode the node type + * + * \param node_type A valid member of the \ref netloc_node_type_t type + * + * \returns NULL if the type is invalid + * \returns A string for that \ref netloc_node_type_t type + */ +static inline const char * netloc_node_type_decode(netloc_node_type_t node_type) { + if( NETLOC_NODE_TYPE_SWITCH == node_type ) { + return "SW"; + } + else if( NETLOC_NODE_TYPE_HOST == node_type ) { + return "CA"; + } + else { + return NULL; + } +} + +ssize_t netloc_line_get(char **lineptr, size_t *n, FILE *stream); + +char *netloc_line_get_next_token(char **string, char c); + +int netloc_build_comm_mat(char *filename, int *pn, double ***pmat); + +#define STRDUP_IF_NOT_NULL(str) (NULL == str ? NULL : strdup(str)) +#define STR_EMPTY_IF_NULL(str) (NULL == str ? "" : str) + + +#endif // _NETLOC_PRIVATE_H_ diff --git a/src/3rdparty/hwloc/include/private/private.h b/src/3rdparty/hwloc/include/private/private.h new file mode 100644 index 000000000..8e3964ab2 --- /dev/null +++ b/src/3rdparty/hwloc/include/private/private.h @@ -0,0 +1,417 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2019 Inria. All rights reserved. + * Copyright © 2009-2012 Université Bordeaux + * Copyright © 2009-2011 Cisco Systems, Inc. All rights reserved. + * + * See COPYING in top-level directory. + */ + +/* Internal types and helpers. */ + + +#ifdef HWLOC_INSIDE_PLUGIN +/* + * these declarations are internal only, they are not available to plugins + * (many functions below are internal static symbols). + */ +#error This file should not be used in plugins +#endif + + +#ifndef HWLOC_PRIVATE_H +#define HWLOC_PRIVATE_H + +#include +#include +#include +#include +#include +#include +#ifdef HAVE_UNISTD_H +#include +#endif +#ifdef HAVE_STDINT_H +#include +#endif +#ifdef HAVE_SYS_UTSNAME_H +#include +#endif +#include + +#define HWLOC_TOPOLOGY_ABI 0x20000 /* version of the layout of struct topology */ + +/***************************************************** + * WARNING: + * changes below in this structure (and its children) + * should cause a bump of HWLOC_TOPOLOGY_ABI. + *****************************************************/ + +struct hwloc_topology { + unsigned topology_abi; + + unsigned nb_levels; /* Number of horizontal levels */ + unsigned nb_levels_allocated; /* Number of levels allocated and zeroed in level_nbobjects and levels below */ + unsigned *level_nbobjects; /* Number of objects on each horizontal level */ + struct hwloc_obj ***levels; /* Direct access to levels, levels[l = 0 .. nblevels-1][0..level_nbobjects[l]] */ + unsigned long flags; + int type_depth[HWLOC_OBJ_TYPE_MAX]; + enum hwloc_type_filter_e type_filter[HWLOC_OBJ_TYPE_MAX]; + int is_thissystem; + int is_loaded; + int modified; /* >0 if objects were added/removed recently, which means a reconnect is needed */ + hwloc_pid_t pid; /* Process ID the topology is view from, 0 for self */ + void *userdata; + uint64_t next_gp_index; + + void *adopted_shmem_addr; + size_t adopted_shmem_length; + +#define HWLOC_NR_SLEVELS 5 +#define HWLOC_SLEVEL_NUMANODE 0 +#define HWLOC_SLEVEL_BRIDGE 1 +#define HWLOC_SLEVEL_PCIDEV 2 +#define HWLOC_SLEVEL_OSDEV 3 +#define HWLOC_SLEVEL_MISC 4 + /* order must match negative depth, it's asserted in setup_defaults() */ +#define HWLOC_SLEVEL_FROM_DEPTH(x) (HWLOC_TYPE_DEPTH_NUMANODE-(x)) +#define HWLOC_SLEVEL_TO_DEPTH(x) (HWLOC_TYPE_DEPTH_NUMANODE-(x)) + struct hwloc_special_level_s { + unsigned nbobjs; + struct hwloc_obj **objs; + struct hwloc_obj *first, *last; /* Temporarily used while listing object before building the objs array */ + } slevels[HWLOC_NR_SLEVELS]; + + hwloc_bitmap_t allowed_cpuset; + hwloc_bitmap_t allowed_nodeset; + + struct hwloc_binding_hooks { + int (*set_thisproc_cpubind)(hwloc_topology_t topology, hwloc_const_cpuset_t set, int flags); + int (*get_thisproc_cpubind)(hwloc_topology_t topology, hwloc_cpuset_t set, int flags); + int (*set_thisthread_cpubind)(hwloc_topology_t topology, hwloc_const_cpuset_t set, int flags); + int (*get_thisthread_cpubind)(hwloc_topology_t topology, hwloc_cpuset_t set, int flags); + int (*set_proc_cpubind)(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_const_cpuset_t set, int flags); + int (*get_proc_cpubind)(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_cpuset_t set, int flags); +#ifdef hwloc_thread_t + int (*set_thread_cpubind)(hwloc_topology_t topology, hwloc_thread_t tid, hwloc_const_cpuset_t set, int flags); + int (*get_thread_cpubind)(hwloc_topology_t topology, hwloc_thread_t tid, hwloc_cpuset_t set, int flags); +#endif + + int (*get_thisproc_last_cpu_location)(hwloc_topology_t topology, hwloc_cpuset_t set, int flags); + int (*get_thisthread_last_cpu_location)(hwloc_topology_t topology, hwloc_cpuset_t set, int flags); + int (*get_proc_last_cpu_location)(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_cpuset_t set, int flags); + + int (*set_thisproc_membind)(hwloc_topology_t topology, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags); + int (*get_thisproc_membind)(hwloc_topology_t topology, hwloc_nodeset_t nodeset, hwloc_membind_policy_t * policy, int flags); + int (*set_thisthread_membind)(hwloc_topology_t topology, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags); + int (*get_thisthread_membind)(hwloc_topology_t topology, hwloc_nodeset_t nodeset, hwloc_membind_policy_t * policy, int flags); + int (*set_proc_membind)(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags); + int (*get_proc_membind)(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_nodeset_t nodeset, hwloc_membind_policy_t * policy, int flags); + int (*set_area_membind)(hwloc_topology_t topology, const void *addr, size_t len, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags); + int (*get_area_membind)(hwloc_topology_t topology, const void *addr, size_t len, hwloc_nodeset_t nodeset, hwloc_membind_policy_t * policy, int flags); + int (*get_area_memlocation)(hwloc_topology_t topology, const void *addr, size_t len, hwloc_nodeset_t nodeset, int flags); + /* This has to return the same kind of pointer as alloc_membind, so that free_membind can be used on it */ + void *(*alloc)(hwloc_topology_t topology, size_t len); + /* alloc_membind has to always succeed if !(flags & HWLOC_MEMBIND_STRICT). + * see hwloc_alloc_or_fail which is convenient for that. */ + void *(*alloc_membind)(hwloc_topology_t topology, size_t len, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags); + int (*free_membind)(hwloc_topology_t topology, void *addr, size_t len); + + int (*get_allowed_resources)(hwloc_topology_t topology); + } binding_hooks; + + struct hwloc_topology_support support; + + void (*userdata_export_cb)(void *reserved, struct hwloc_topology *topology, struct hwloc_obj *obj); + void (*userdata_import_cb)(struct hwloc_topology *topology, struct hwloc_obj *obj, const char *name, const void *buffer, size_t length); + int userdata_not_decoded; + + struct hwloc_internal_distances_s { + hwloc_obj_type_t type; + /* add union hwloc_obj_attr_u if we ever support groups */ + unsigned nbobjs; + uint64_t *indexes; /* array of OS or GP indexes before we can convert them into objs. */ + uint64_t *values; /* distance matrices, ordered according to the above indexes/objs array. + * distance from i to j is stored in slot i*nbnodes+j. + */ + unsigned long kind; + + /* objects are currently stored in physical_index order */ + hwloc_obj_t *objs; /* array of objects */ + int objs_are_valid; /* set to 1 if the array objs is still valid, 0 if needs refresh */ + + unsigned id; /* to match the container id field of public distances structure */ + struct hwloc_internal_distances_s *prev, *next; + } *first_dist, *last_dist; + unsigned next_dist_id; + + int grouping; + int grouping_verbose; + unsigned grouping_nbaccuracies; + float grouping_accuracies[5]; + unsigned grouping_next_subkind; + + /* list of enabled backends. */ + struct hwloc_backend * backends; + struct hwloc_backend * get_pci_busid_cpuset_backend; + unsigned backend_excludes; + + /* memory allocator for topology objects */ + struct hwloc_tma * tma; + +/***************************************************** + * WARNING: + * changes above in this structure (and its children) + * should cause a bump of HWLOC_TOPOLOGY_ABI. + *****************************************************/ + + /* + * temporary variables during discovery + */ + + /* machine-wide memory. + * temporarily stored there by OSes that only provide this without NUMA information, + * and actually used later by the core. + */ + struct hwloc_numanode_attr_s machine_memory; + + /* pci stuff */ + int need_pci_belowroot_apply_locality; + int pci_has_forced_locality; + unsigned pci_forced_locality_nr; + struct hwloc_pci_forced_locality_s { + unsigned domain; + unsigned bus_first, bus_last; + hwloc_bitmap_t cpuset; + } * pci_forced_locality; + +}; + +extern void hwloc_alloc_root_sets(hwloc_obj_t root); +extern void hwloc_setup_pu_level(struct hwloc_topology *topology, unsigned nb_pus); +extern int hwloc_get_sysctlbyname(const char *name, int64_t *n); +extern int hwloc_get_sysctl(int name[], unsigned namelen, int *n); +extern int hwloc_fallback_nbprocessors(struct hwloc_topology *topology); + +extern int hwloc__object_cpusets_compare_first(hwloc_obj_t obj1, hwloc_obj_t obj2); +extern void hwloc__reorder_children(hwloc_obj_t parent); + +extern void hwloc_topology_setup_defaults(struct hwloc_topology *topology); +extern void hwloc_topology_clear(struct hwloc_topology *topology); + +/* insert memory object as memory child of normal parent */ +extern struct hwloc_obj * hwloc__attach_memory_object(struct hwloc_topology *topology, hwloc_obj_t parent, + hwloc_obj_t obj, + hwloc_report_error_t report_error); + +extern void hwloc_pci_discovery_init(struct hwloc_topology *topology); +extern void hwloc_pci_discovery_prepare(struct hwloc_topology *topology); +extern void hwloc_pci_discovery_exit(struct hwloc_topology *topology); + +/* Look for an object matching complete cpuset exactly, or insert one. + * Return NULL on failure. + * Return a good fallback (object above) on failure to insert. + */ +extern hwloc_obj_t hwloc_find_insert_io_parent_by_complete_cpuset(struct hwloc_topology *topology, hwloc_cpuset_t cpuset); + +/* Move PCI objects currently attached to the root object ot their actual location. + * Called by the core at the end of hwloc_topology_load(). + * Prior to this call, all PCI objects may be found below the root object. + * After this call and a reconnect of levels, all PCI objects are available through levels. + */ +extern int hwloc_pci_belowroot_apply_locality(struct hwloc_topology *topology); + +extern int hwloc__add_info(struct hwloc_info_s **infosp, unsigned *countp, const char *name, const char *value); +extern int hwloc__add_info_nodup(struct hwloc_info_s **infosp, unsigned *countp, const char *name, const char *value, int replace); +extern int hwloc__move_infos(struct hwloc_info_s **dst_infosp, unsigned *dst_countp, struct hwloc_info_s **src_infosp, unsigned *src_countp); +extern void hwloc__free_infos(struct hwloc_info_s *infos, unsigned count); + +/* set native OS binding hooks */ +extern void hwloc_set_native_binding_hooks(struct hwloc_binding_hooks *hooks, struct hwloc_topology_support *support); +/* set either native OS binding hooks (if thissystem), or dummy ones */ +extern void hwloc_set_binding_hooks(struct hwloc_topology *topology); + +#if defined(HWLOC_LINUX_SYS) +extern void hwloc_set_linuxfs_hooks(struct hwloc_binding_hooks *binding_hooks, struct hwloc_topology_support *support); +#endif /* HWLOC_LINUX_SYS */ + +#if defined(HWLOC_BGQ_SYS) +extern void hwloc_set_bgq_hooks(struct hwloc_binding_hooks *binding_hooks, struct hwloc_topology_support *support); +#endif /* HWLOC_BGQ_SYS */ + +#ifdef HWLOC_SOLARIS_SYS +extern void hwloc_set_solaris_hooks(struct hwloc_binding_hooks *binding_hooks, struct hwloc_topology_support *support); +#endif /* HWLOC_SOLARIS_SYS */ + +#ifdef HWLOC_AIX_SYS +extern void hwloc_set_aix_hooks(struct hwloc_binding_hooks *binding_hooks, struct hwloc_topology_support *support); +#endif /* HWLOC_AIX_SYS */ + +#ifdef HWLOC_WIN_SYS +extern void hwloc_set_windows_hooks(struct hwloc_binding_hooks *binding_hooks, struct hwloc_topology_support *support); +#endif /* HWLOC_WIN_SYS */ + +#ifdef HWLOC_DARWIN_SYS +extern void hwloc_set_darwin_hooks(struct hwloc_binding_hooks *binding_hooks, struct hwloc_topology_support *support); +#endif /* HWLOC_DARWIN_SYS */ + +#ifdef HWLOC_FREEBSD_SYS +extern void hwloc_set_freebsd_hooks(struct hwloc_binding_hooks *binding_hooks, struct hwloc_topology_support *support); +#endif /* HWLOC_FREEBSD_SYS */ + +#ifdef HWLOC_NETBSD_SYS +extern void hwloc_set_netbsd_hooks(struct hwloc_binding_hooks *binding_hooks, struct hwloc_topology_support *support); +#endif /* HWLOC_NETBSD_SYS */ + +#ifdef HWLOC_HPUX_SYS +extern void hwloc_set_hpux_hooks(struct hwloc_binding_hooks *binding_hooks, struct hwloc_topology_support *support); +#endif /* HWLOC_HPUX_SYS */ + +extern int hwloc_look_hardwired_fujitsu_k(struct hwloc_topology *topology); +extern int hwloc_look_hardwired_fujitsu_fx10(struct hwloc_topology *topology); +extern int hwloc_look_hardwired_fujitsu_fx100(struct hwloc_topology *topology); + +/* Insert uname-specific names/values in the object infos array. + * If cached_uname isn't NULL, it is used as a struct utsname instead of recalling uname. + * Any field that starts with \0 is ignored. + */ +extern void hwloc_add_uname_info(struct hwloc_topology *topology, void *cached_uname); + +/* Free obj and its attributes assuming it's not linked to a parent and doesn't have any child */ +extern void hwloc_free_unlinked_object(hwloc_obj_t obj); + +/* Free obj and its children, assuming it's not linked to a parent */ +extern void hwloc_free_object_and_children(hwloc_obj_t obj); + +/* Free obj, its next siblings, and their children, assuming they're not linked to a parent */ +extern void hwloc_free_object_siblings_and_children(hwloc_obj_t obj); + +/* This can be used for the alloc field to get allocated data that can be freed by free() */ +void *hwloc_alloc_heap(hwloc_topology_t topology, size_t len); + +/* This can be used for the alloc field to get allocated data that can be freed by munmap() */ +void *hwloc_alloc_mmap(hwloc_topology_t topology, size_t len); + +/* This can be used for the free_membind field to free data using free() */ +int hwloc_free_heap(hwloc_topology_t topology, void *addr, size_t len); + +/* This can be used for the free_membind field to free data using munmap() */ +int hwloc_free_mmap(hwloc_topology_t topology, void *addr, size_t len); + +/* Allocates unbound memory or fail, depending on whether STRICT is requested + * or not */ +static __hwloc_inline void * +hwloc_alloc_or_fail(hwloc_topology_t topology, size_t len, int flags) +{ + if (flags & HWLOC_MEMBIND_STRICT) + return NULL; + return hwloc_alloc(topology, len); +} + +extern void hwloc_internal_distances_init(hwloc_topology_t topology); +extern void hwloc_internal_distances_prepare(hwloc_topology_t topology); +extern void hwloc_internal_distances_destroy(hwloc_topology_t topology); +extern int hwloc_internal_distances_dup(hwloc_topology_t new, hwloc_topology_t old); +extern void hwloc_internal_distances_refresh(hwloc_topology_t topology); +extern int hwloc_internal_distances_add(hwloc_topology_t topology, unsigned nbobjs, hwloc_obj_t *objs, uint64_t *values, unsigned long kind, unsigned long flags); +extern int hwloc_internal_distances_add_by_index(hwloc_topology_t topology, hwloc_obj_type_t type, unsigned nbobjs, uint64_t *indexes, uint64_t *values, unsigned long kind, unsigned long flags); +extern void hwloc_internal_distances_invalidate_cached_objs(hwloc_topology_t topology); + +/* encode src buffer into target buffer. + * targsize must be at least 4*((srclength+2)/3)+1. + * target will be 0-terminated. + */ +extern int hwloc_encode_to_base64(const char *src, size_t srclength, char *target, size_t targsize); +/* decode src buffer into target buffer. + * src is 0-terminated. + * targsize must be at least srclength*3/4+1 (srclength not including \0) + * but only srclength*3/4 characters will be meaningful + * (the next one may be partially written during decoding, but it should be ignored). + */ +extern int hwloc_decode_from_base64(char const *src, char *target, size_t targsize); + +/* Check whether needle matches the beginning of haystack, at least n, and up + * to a colon or \0 */ +extern int hwloc_namecoloncmp(const char *haystack, const char *needle, size_t n); + +/* On some systems, snprintf returns the size of written data, not the actually + * required size. hwloc_snprintf always report the actually required size. */ +extern int hwloc_snprintf(char *str, size_t size, const char *format, ...) __hwloc_attribute_format(printf, 3, 4); + +/* Return the name of the currently running program, if supported. + * If not NULL, must be freed by the caller. + */ +extern char * hwloc_progname(struct hwloc_topology *topology); + +/* obj->attr->group.kind internal values. + * the core will keep the smallest ones when merging two groups, + * that's why user-given kinds are first. + */ +/* first, user-given groups, should remain as long as possible */ +#define HWLOC_GROUP_KIND_USER 0 /* user-given, user may use subkind too */ +#define HWLOC_GROUP_KIND_SYNTHETIC 10 /* subkind is group depth within synthetic description */ +/* then, hardware-specific groups */ +#define HWLOC_GROUP_KIND_INTEL_KNL_SUBNUMA_CLUSTER 100 /* no subkind */ +#define HWLOC_GROUP_KIND_INTEL_EXTTOPOENUM_UNKNOWN 101 /* subkind is unknown level */ +#define HWLOC_GROUP_KIND_INTEL_MODULE 102 /* no subkind */ +#define HWLOC_GROUP_KIND_INTEL_TILE 103 /* no subkind */ +#define HWLOC_GROUP_KIND_INTEL_DIE 104 /* no subkind */ +#define HWLOC_GROUP_KIND_S390_BOOK 110 /* no subkind */ +#define HWLOC_GROUP_KIND_AMD_COMPUTE_UNIT 120 /* no subkind */ +/* then, OS-specific groups */ +#define HWLOC_GROUP_KIND_SOLARIS_PG_HW_PERF 200 /* subkind is group width */ +#define HWLOC_GROUP_KIND_AIX_SDL_UNKNOWN 210 /* subkind is SDL level */ +#define HWLOC_GROUP_KIND_WINDOWS_PROCESSOR_GROUP 220 /* no subkind */ +#define HWLOC_GROUP_KIND_WINDOWS_RELATIONSHIP_UNKNOWN 221 /* no subkind */ +/* distance groups */ +#define HWLOC_GROUP_KIND_DISTANCE 900 /* subkind is round of adding these groups during distance based grouping */ +/* finally, hwloc-specific groups required to insert something else, should disappear as soon as possible */ +#define HWLOC_GROUP_KIND_IO 1000 /* no subkind */ +#define HWLOC_GROUP_KIND_MEMORY 1001 /* no subkind */ + +/* memory allocator for topology objects */ +struct hwloc_tma { + void * (*malloc)(struct hwloc_tma *, size_t); + void *data; + int dontfree; /* when set, free() or realloc() cannot be used, and tma->malloc() cannot fail */ +}; + +static __hwloc_inline void * +hwloc_tma_malloc(struct hwloc_tma *tma, + size_t size) +{ + if (tma) { + return tma->malloc(tma, size); + } else { + return malloc(size); + } +} + +static __hwloc_inline void * +hwloc_tma_calloc(struct hwloc_tma *tma, + size_t size) +{ + char *ptr = hwloc_tma_malloc(tma, size); + if (ptr) + memset(ptr, 0, size); + return ptr; +} + +static __hwloc_inline char * +hwloc_tma_strdup(struct hwloc_tma *tma, + const char *src) +{ + size_t len = strlen(src); + char *ptr = hwloc_tma_malloc(tma, len+1); + if (ptr) + memcpy(ptr, src, len+1); + return ptr; +} + +/* bitmap allocator to be used inside hwloc */ +extern hwloc_bitmap_t hwloc_bitmap_tma_dup(struct hwloc_tma *tma, hwloc_const_bitmap_t old); + +extern int hwloc__topology_dup(hwloc_topology_t *newp, hwloc_topology_t old, struct hwloc_tma *tma); +extern void hwloc__topology_disadopt(hwloc_topology_t topology); + +#endif /* HWLOC_PRIVATE_H */ diff --git a/src/3rdparty/hwloc/include/private/solaris-chiptype.h b/src/3rdparty/hwloc/include/private/solaris-chiptype.h new file mode 100644 index 000000000..4ad2130a0 --- /dev/null +++ b/src/3rdparty/hwloc/include/private/solaris-chiptype.h @@ -0,0 +1,43 @@ +/* + * Copyright © 2009-2010 Oracle and/or its affiliates. All rights reserved. + * + * Copyright © 2017 Inria. All rights reserved. + * $COPYRIGHT$ + * + * Additional copyrights may follow + * + * $HEADER$ + */ + + +#ifdef HWLOC_INSIDE_PLUGIN +/* + * these declarations are internal only, they are not available to plugins + * (functions below are internal static symbols). + */ +#error This file should not be used in plugins +#endif + + +#ifndef HWLOC_PRIVATE_SOLARIS_CHIPTYPE_H +#define HWLOC_PRIVATE_SOLARIS_CHIPTYPE_H + +struct hwloc_solaris_chip_info_s { + char *model; + char *type; + /* L1i, L1d, L2, L3 */ +#define HWLOC_SOLARIS_CHIP_INFO_L1I 0 +#define HWLOC_SOLARIS_CHIP_INFO_L1D 1 +#define HWLOC_SOLARIS_CHIP_INFO_L2I 2 +#define HWLOC_SOLARIS_CHIP_INFO_L2D 3 +#define HWLOC_SOLARIS_CHIP_INFO_L3 4 + long cache_size[5]; /* cleared to -1 if we don't want of that cache */ + unsigned cache_linesize[5]; + unsigned cache_associativity[5]; + int l2_unified; +}; + +/* fills the structure with 0 on error */ +extern void hwloc_solaris_get_chip_info(struct hwloc_solaris_chip_info_s *info); + +#endif /* HWLOC_PRIVATE_SOLARIS_CHIPTYPE_H */ diff --git a/src/3rdparty/hwloc/include/private/xml.h b/src/3rdparty/hwloc/include/private/xml.h new file mode 100644 index 000000000..7c73384d9 --- /dev/null +++ b/src/3rdparty/hwloc/include/private/xml.h @@ -0,0 +1,108 @@ +/* + * Copyright © 2009-2019 Inria. All rights reserved. + * See COPYING in top-level directory. + */ + +#ifndef PRIVATE_XML_H +#define PRIVATE_XML_H 1 + +#include + +#include + +HWLOC_DECLSPEC int hwloc__xml_verbose(void); + +/************** + * XML import * + **************/ + +typedef struct hwloc__xml_import_state_s { + struct hwloc__xml_import_state_s *parent; + + /* globals shared because the entire stack of states during import */ + struct hwloc_xml_backend_data_s *global; + + /* opaque data used to store backend-specific data. + * statically allocated to allow stack-allocation by the common code without knowing actual backend needs. + */ + char data[32]; +} * hwloc__xml_import_state_t; + +struct hwloc__xml_imported_v1distances_s { + unsigned long kind; + unsigned nbobjs; + float *floats; + struct hwloc__xml_imported_v1distances_s *prev, *next; +}; + +HWLOC_DECLSPEC int hwloc__xml_import_diff(hwloc__xml_import_state_t state, hwloc_topology_diff_t *firstdiffp); + +struct hwloc_xml_backend_data_s { + /* xml backend parameters */ + int (*look_init)(struct hwloc_xml_backend_data_s *bdata, struct hwloc__xml_import_state_s *state); + void (*look_done)(struct hwloc_xml_backend_data_s *bdata, int result); + void (*backend_exit)(struct hwloc_xml_backend_data_s *bdata); + int (*next_attr)(struct hwloc__xml_import_state_s * state, char **namep, char **valuep); + int (*find_child)(struct hwloc__xml_import_state_s * state, struct hwloc__xml_import_state_s * childstate, char **tagp); + int (*close_tag)(struct hwloc__xml_import_state_s * state); /* look for an explicit closing tag */ + void (*close_child)(struct hwloc__xml_import_state_s * state); + int (*get_content)(struct hwloc__xml_import_state_s * state, char **beginp, size_t expected_length); /* return 0 on empty content (and sets beginp to empty string), 1 on actual content, -1 on error or unexpected content length */ + void (*close_content)(struct hwloc__xml_import_state_s * state); + char * msgprefix; + void *data; /* libxml2 doc, or nolibxml buffer */ + unsigned version_major, version_minor; + unsigned nbnumanodes; + hwloc_obj_t first_numanode, last_numanode; /* temporary cousin-list for handling v1distances */ + struct hwloc__xml_imported_v1distances_s *first_v1dist, *last_v1dist; + int dont_merge_die_groups; +}; + +/************** + * XML export * + **************/ + +typedef struct hwloc__xml_export_state_s { + struct hwloc__xml_export_state_s *parent; + + void (*new_child)(struct hwloc__xml_export_state_s *parentstate, struct hwloc__xml_export_state_s *state, const char *name); + void (*new_prop)(struct hwloc__xml_export_state_s *state, const char *name, const char *value); + void (*add_content)(struct hwloc__xml_export_state_s *state, const char *buffer, size_t length); + void (*end_object)(struct hwloc__xml_export_state_s *state, const char *name); + + struct hwloc__xml_export_data_s { + hwloc_obj_t v1_memory_group; /* if we need to insert intermediate group above memory children when exporting to v1 */ + } *global; + + /* opaque data used to store backend-specific data. + * statically allocated to allow stack-allocation by the common code without knowing actual backend needs. + */ + char data[40]; +} * hwloc__xml_export_state_t; + +HWLOC_DECLSPEC void hwloc__xml_export_topology(hwloc__xml_export_state_t parentstate, hwloc_topology_t topology, unsigned long flags); + +HWLOC_DECLSPEC void hwloc__xml_export_diff(hwloc__xml_export_state_t parentstate, hwloc_topology_diff_t diff); + +/****************** + * XML components * + ******************/ + +struct hwloc_xml_callbacks { + int (*backend_init)(struct hwloc_xml_backend_data_s *bdata, const char *xmlpath, const char *xmlbuffer, int xmlbuflen); + int (*export_file)(struct hwloc_topology *topology, struct hwloc__xml_export_data_s *edata, const char *filename, unsigned long flags); + int (*export_buffer)(struct hwloc_topology *topology, struct hwloc__xml_export_data_s *edata, char **xmlbuffer, int *buflen, unsigned long flags); + void (*free_buffer)(void *xmlbuffer); + int (*import_diff)(struct hwloc__xml_import_state_s *state, const char *xmlpath, const char *xmlbuffer, int xmlbuflen, hwloc_topology_diff_t *diff, char **refnamep); + int (*export_diff_file)(union hwloc_topology_diff_u *diff, const char *refname, const char *filename); + int (*export_diff_buffer)(union hwloc_topology_diff_u *diff, const char *refname, char **xmlbuffer, int *buflen); +}; + +struct hwloc_xml_component { + struct hwloc_xml_callbacks *nolibxml_callbacks; + struct hwloc_xml_callbacks *libxml_callbacks; +}; + +HWLOC_DECLSPEC void hwloc_xml_callbacks_register(struct hwloc_xml_component *component); +HWLOC_DECLSPEC void hwloc_xml_callbacks_reset(void); + +#endif /* PRIVATE_XML_H */ diff --git a/src/3rdparty/hwloc/src/base64.c b/src/3rdparty/hwloc/src/base64.c new file mode 100644 index 000000000..7b3e12101 --- /dev/null +++ b/src/3rdparty/hwloc/src/base64.c @@ -0,0 +1,309 @@ +/* + * Copyright © 2012-2018 Inria. All rights reserved. + * See COPYING in top-level directory. + * + * Modifications after import: + * - removed all #if + * - updated prototypes + * - updated #include + */ + +/* include hwloc's config before anything else + * so that extensions and features are properly enabled + */ +#include + +/* $OpenBSD: base64.c,v 1.5 2006/10/21 09:55:03 otto Exp $ */ + +/* + * Copyright (c) 1996 by Internet Software Consortium. + * + * Permission to use, copy, modify, and distribute this software for any + * purpose with or without fee is hereby granted, provided that the above + * copyright notice and this permission notice appear in all copies. + * + * THE SOFTWARE IS PROVIDED "AS IS" AND INTERNET SOFTWARE CONSORTIUM DISCLAIMS + * ALL WARRANTIES WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL INTERNET SOFTWARE + * CONSORTIUM BE LIABLE FOR ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL + * DAMAGES OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR + * PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS + * ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS + * SOFTWARE. + */ + +/* + * Portions Copyright (c) 1995 by International Business Machines, Inc. + * + * International Business Machines, Inc. (hereinafter called IBM) grants + * permission under its copyrights to use, copy, modify, and distribute this + * Software with or without fee, provided that the above copyright notice and + * all paragraphs of this notice appear in all copies, and that the name of IBM + * not be used in connection with the marketing of any product incorporating + * the Software or modifications thereof, without specific, written prior + * permission. + * + * To the extent it has a right to do so, IBM grants an immunity from suit + * under its patents, if any, for the use, sale or manufacture of products to + * the extent that such products are used for performing Domain Name System + * dynamic updates in TCP/IP networks by means of the Software. No immunity is + * granted for any product per se or for any other function of any product. + * + * THE SOFTWARE IS PROVIDED "AS IS", AND IBM DISCLAIMS ALL WARRANTIES, + * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A + * PARTICULAR PURPOSE. IN NO EVENT SHALL IBM BE LIABLE FOR ANY SPECIAL, + * DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES WHATSOEVER ARISING + * OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE, EVEN + * IF IBM IS APPRISED OF THE POSSIBILITY OF SUCH DAMAGES. + */ + +/* OPENBSD ORIGINAL: lib/libc/net/base64.c */ + +static const char Base64[] = + "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/"; +static const char Pad64 = '='; + +/* (From RFC1521 and draft-ietf-dnssec-secext-03.txt) + The following encoding technique is taken from RFC 1521 by Borenstein + and Freed. It is reproduced here in a slightly edited form for + convenience. + + A 65-character subset of US-ASCII is used, enabling 6 bits to be + represented per printable character. (The extra 65th character, "=", + is used to signify a special processing function.) + + The encoding process represents 24-bit groups of input bits as output + strings of 4 encoded characters. Proceeding from left to right, a + 24-bit input group is formed by concatenating 3 8-bit input groups. + These 24 bits are then treated as 4 concatenated 6-bit groups, each + of which is translated into a single digit in the base64 alphabet. + + Each 6-bit group is used as an index into an array of 64 printable + characters. The character referenced by the index is placed in the + output string. + + Table 1: The Base64 Alphabet + + Value Encoding Value Encoding Value Encoding Value Encoding + 0 A 17 R 34 i 51 z + 1 B 18 S 35 j 52 0 + 2 C 19 T 36 k 53 1 + 3 D 20 U 37 l 54 2 + 4 E 21 V 38 m 55 3 + 5 F 22 W 39 n 56 4 + 6 G 23 X 40 o 57 5 + 7 H 24 Y 41 p 58 6 + 8 I 25 Z 42 q 59 7 + 9 J 26 a 43 r 60 8 + 10 K 27 b 44 s 61 9 + 11 L 28 c 45 t 62 + + 12 M 29 d 46 u 63 / + 13 N 30 e 47 v + 14 O 31 f 48 w (pad) = + 15 P 32 g 49 x + 16 Q 33 h 50 y + + Special processing is performed if fewer than 24 bits are available + at the end of the data being encoded. A full encoding quantum is + always completed at the end of a quantity. When fewer than 24 input + bits are available in an input group, zero bits are added (on the + right) to form an integral number of 6-bit groups. Padding at the + end of the data is performed using the '=' character. + + Since all base64 input is an integral number of octets, only the + ------------------------------------------------- + following cases can arise: + + (1) the final quantum of encoding input is an integral + multiple of 24 bits; here, the final unit of encoded + output will be an integral multiple of 4 characters + with no "=" padding, + (2) the final quantum of encoding input is exactly 8 bits; + here, the final unit of encoded output will be two + characters followed by two "=" padding characters, or + (3) the final quantum of encoding input is exactly 16 bits; + here, the final unit of encoded output will be three + characters followed by one "=" padding character. + */ + +#include +#include +#include + +int +hwloc_encode_to_base64(const char *src, size_t srclength, char *target, size_t targsize) +{ + size_t datalength = 0; + unsigned char input[3]; + unsigned char output[4]; + unsigned int i; + + while (2 < srclength) { + input[0] = *src++; + input[1] = *src++; + input[2] = *src++; + srclength -= 3; + + output[0] = input[0] >> 2; + output[1] = ((input[0] & 0x03) << 4) + (input[1] >> 4); + output[2] = ((input[1] & 0x0f) << 2) + (input[2] >> 6); + output[3] = input[2] & 0x3f; + + if (datalength + 4 > targsize) + return (-1); + target[datalength++] = Base64[output[0]]; + target[datalength++] = Base64[output[1]]; + target[datalength++] = Base64[output[2]]; + target[datalength++] = Base64[output[3]]; + } + + /* Now we worry about padding. */ + if (0 != srclength) { + /* Get what's left. */ + input[0] = input[1] = input[2] = '\0'; + for (i = 0; i < srclength; i++) + input[i] = *src++; + + output[0] = input[0] >> 2; + output[1] = ((input[0] & 0x03) << 4) + (input[1] >> 4); + output[2] = ((input[1] & 0x0f) << 2) + (input[2] >> 6); + + if (datalength + 4 > targsize) + return (-1); + target[datalength++] = Base64[output[0]]; + target[datalength++] = Base64[output[1]]; + if (srclength == 1) + target[datalength++] = Pad64; + else + target[datalength++] = Base64[output[2]]; + target[datalength++] = Pad64; + } + if (datalength >= targsize) + return (-1); + target[datalength] = '\0'; /* Returned value doesn't count \0. */ + return (int)(datalength); +} + +/* skips all whitespace anywhere. + converts characters, four at a time, starting at (or after) + src from base - 64 numbers into three 8 bit bytes in the target area. + it returns the number of data bytes stored at the target, or -1 on error. + */ + +int +hwloc_decode_from_base64(char const *src, char *target, size_t targsize) +{ + unsigned int tarindex, state; + int ch; + char *pos; + + state = 0; + tarindex = 0; + + while ((ch = *src++) != '\0') { + if (isspace(ch)) /* Skip whitespace anywhere. */ + continue; + + if (ch == Pad64) + break; + + pos = strchr(Base64, ch); + if (pos == 0) /* A non-base64 character. */ + return (-1); + + switch (state) { + case 0: + if (target) { + if (tarindex >= targsize) + return (-1); + target[tarindex] = (char)(pos - Base64) << 2; + } + state = 1; + break; + case 1: + if (target) { + if (tarindex + 1 >= targsize) + return (-1); + target[tarindex] |= (pos - Base64) >> 4; + target[tarindex+1] = ((pos - Base64) & 0x0f) + << 4 ; + } + tarindex++; + state = 2; + break; + case 2: + if (target) { + if (tarindex + 1 >= targsize) + return (-1); + target[tarindex] |= (pos - Base64) >> 2; + target[tarindex+1] = ((pos - Base64) & 0x03) + << 6; + } + tarindex++; + state = 3; + break; + case 3: + if (target) { + if (tarindex >= targsize) + return (-1); + target[tarindex] |= (pos - Base64); + } + tarindex++; + state = 0; + break; + } + } + + /* + * We are done decoding Base-64 chars. Let's see if we ended + * on a byte boundary, and/or with erroneous trailing characters. + */ + + if (ch == Pad64) { /* We got a pad char. */ + ch = *src++; /* Skip it, get next. */ + switch (state) { + case 0: /* Invalid = in first position */ + case 1: /* Invalid = in second position */ + return (-1); + + case 2: /* Valid, means one byte of info */ + /* Skip any number of spaces. */ + for (; ch != '\0'; ch = *src++) + if (!isspace(ch)) + break; + /* Make sure there is another trailing = sign. */ + if (ch != Pad64) + return (-1); + ch = *src++; /* Skip the = */ + /* Fall through to "single trailing =" case. */ + /* FALLTHROUGH */ + + case 3: /* Valid, means two bytes of info */ + /* + * We know this char is an =. Is there anything but + * whitespace after it? + */ + for (; ch != '\0'; ch = *src++) + if (!isspace(ch)) + return (-1); + + /* + * Now make sure for cases 2 and 3 that the "extra" + * bits that slopped past the last full byte were + * zeros. If we don't check them, they become a + * subliminal channel. + */ + if (target && target[tarindex] != 0) + return (-1); + } + } else { + /* + * We ended by seeing the end of the string. Make sure we + * have no partial bytes lying around. + */ + if (state != 0) + return (-1); + } + + return (tarindex); +} diff --git a/src/3rdparty/hwloc/src/bind.c b/src/3rdparty/hwloc/src/bind.c new file mode 100644 index 000000000..b3457bc76 --- /dev/null +++ b/src/3rdparty/hwloc/src/bind.c @@ -0,0 +1,922 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2018 Inria. All rights reserved. + * Copyright © 2009-2010, 2012 Université Bordeaux + * Copyright © 2011-2015 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +#include +#include +#include +#include +#ifdef HAVE_SYS_MMAN_H +# include +#endif +/* is only needed if we don't have posix_memalign() */ +#if defined(hwloc_getpagesize) && !defined(HAVE_POSIX_MEMALIGN) && defined(HAVE_MEMALIGN) && defined(HAVE_MALLOC_H) +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif +#include +#include + +/* TODO: HWLOC_GNU_SYS, + * + * We could use glibc's sched_setaffinity generically when it is available + * + * Darwin and OpenBSD don't seem to have binding facilities. + */ + +#define HWLOC_CPUBIND_ALLFLAGS (HWLOC_CPUBIND_PROCESS|HWLOC_CPUBIND_THREAD|HWLOC_CPUBIND_STRICT|HWLOC_CPUBIND_NOMEMBIND) + +static hwloc_const_bitmap_t +hwloc_fix_cpubind(hwloc_topology_t topology, hwloc_const_bitmap_t set) +{ + hwloc_const_bitmap_t topology_set = hwloc_topology_get_topology_cpuset(topology); + hwloc_const_bitmap_t complete_set = hwloc_topology_get_complete_cpuset(topology); + + if (hwloc_bitmap_iszero(set)) { + errno = EINVAL; + return NULL; + } + + if (!hwloc_bitmap_isincluded(set, complete_set)) { + errno = EINVAL; + return NULL; + } + + if (hwloc_bitmap_isincluded(topology_set, set)) + set = complete_set; + + return set; +} + +int +hwloc_set_cpubind(hwloc_topology_t topology, hwloc_const_bitmap_t set, int flags) +{ + if (flags & ~HWLOC_CPUBIND_ALLFLAGS) { + errno = EINVAL; + return -1; + } + + set = hwloc_fix_cpubind(topology, set); + if (!set) + return -1; + + if (flags & HWLOC_CPUBIND_PROCESS) { + if (topology->binding_hooks.set_thisproc_cpubind) + return topology->binding_hooks.set_thisproc_cpubind(topology, set, flags); + } else if (flags & HWLOC_CPUBIND_THREAD) { + if (topology->binding_hooks.set_thisthread_cpubind) + return topology->binding_hooks.set_thisthread_cpubind(topology, set, flags); + } else { + if (topology->binding_hooks.set_thisproc_cpubind) { + int err = topology->binding_hooks.set_thisproc_cpubind(topology, set, flags); + if (err >= 0 || errno != ENOSYS) + return err; + /* ENOSYS, fallback */ + } + if (topology->binding_hooks.set_thisthread_cpubind) + return topology->binding_hooks.set_thisthread_cpubind(topology, set, flags); + } + + errno = ENOSYS; + return -1; +} + +int +hwloc_get_cpubind(hwloc_topology_t topology, hwloc_bitmap_t set, int flags) +{ + if (flags & ~HWLOC_CPUBIND_ALLFLAGS) { + errno = EINVAL; + return -1; + } + + if (flags & HWLOC_CPUBIND_PROCESS) { + if (topology->binding_hooks.get_thisproc_cpubind) + return topology->binding_hooks.get_thisproc_cpubind(topology, set, flags); + } else if (flags & HWLOC_CPUBIND_THREAD) { + if (topology->binding_hooks.get_thisthread_cpubind) + return topology->binding_hooks.get_thisthread_cpubind(topology, set, flags); + } else { + if (topology->binding_hooks.get_thisproc_cpubind) { + int err = topology->binding_hooks.get_thisproc_cpubind(topology, set, flags); + if (err >= 0 || errno != ENOSYS) + return err; + /* ENOSYS, fallback */ + } + if (topology->binding_hooks.get_thisthread_cpubind) + return topology->binding_hooks.get_thisthread_cpubind(topology, set, flags); + } + + errno = ENOSYS; + return -1; +} + +int +hwloc_set_proc_cpubind(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_const_bitmap_t set, int flags) +{ + if (flags & ~HWLOC_CPUBIND_ALLFLAGS) { + errno = EINVAL; + return -1; + } + + set = hwloc_fix_cpubind(topology, set); + if (!set) + return -1; + + if (topology->binding_hooks.set_proc_cpubind) + return topology->binding_hooks.set_proc_cpubind(topology, pid, set, flags); + + errno = ENOSYS; + return -1; +} + +int +hwloc_get_proc_cpubind(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_bitmap_t set, int flags) +{ + if (flags & ~HWLOC_CPUBIND_ALLFLAGS) { + errno = EINVAL; + return -1; + } + + if (topology->binding_hooks.get_proc_cpubind) + return topology->binding_hooks.get_proc_cpubind(topology, pid, set, flags); + + errno = ENOSYS; + return -1; +} + +#ifdef hwloc_thread_t +int +hwloc_set_thread_cpubind(hwloc_topology_t topology, hwloc_thread_t tid, hwloc_const_bitmap_t set, int flags) +{ + if (flags & ~HWLOC_CPUBIND_ALLFLAGS) { + errno = EINVAL; + return -1; + } + + set = hwloc_fix_cpubind(topology, set); + if (!set) + return -1; + + if (topology->binding_hooks.set_thread_cpubind) + return topology->binding_hooks.set_thread_cpubind(topology, tid, set, flags); + + errno = ENOSYS; + return -1; +} + +int +hwloc_get_thread_cpubind(hwloc_topology_t topology, hwloc_thread_t tid, hwloc_bitmap_t set, int flags) +{ + if (flags & ~HWLOC_CPUBIND_ALLFLAGS) { + errno = EINVAL; + return -1; + } + + if (topology->binding_hooks.get_thread_cpubind) + return topology->binding_hooks.get_thread_cpubind(topology, tid, set, flags); + + errno = ENOSYS; + return -1; +} +#endif + +int +hwloc_get_last_cpu_location(hwloc_topology_t topology, hwloc_bitmap_t set, int flags) +{ + if (flags & ~HWLOC_CPUBIND_ALLFLAGS) { + errno = EINVAL; + return -1; + } + + if (flags & HWLOC_CPUBIND_PROCESS) { + if (topology->binding_hooks.get_thisproc_last_cpu_location) + return topology->binding_hooks.get_thisproc_last_cpu_location(topology, set, flags); + } else if (flags & HWLOC_CPUBIND_THREAD) { + if (topology->binding_hooks.get_thisthread_last_cpu_location) + return topology->binding_hooks.get_thisthread_last_cpu_location(topology, set, flags); + } else { + if (topology->binding_hooks.get_thisproc_last_cpu_location) { + int err = topology->binding_hooks.get_thisproc_last_cpu_location(topology, set, flags); + if (err >= 0 || errno != ENOSYS) + return err; + /* ENOSYS, fallback */ + } + if (topology->binding_hooks.get_thisthread_last_cpu_location) + return topology->binding_hooks.get_thisthread_last_cpu_location(topology, set, flags); + } + + errno = ENOSYS; + return -1; +} + +int +hwloc_get_proc_last_cpu_location(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_bitmap_t set, int flags) +{ + if (flags & ~HWLOC_CPUBIND_ALLFLAGS) { + errno = EINVAL; + return -1; + } + + if (topology->binding_hooks.get_proc_last_cpu_location) + return topology->binding_hooks.get_proc_last_cpu_location(topology, pid, set, flags); + + errno = ENOSYS; + return -1; +} + +#define HWLOC_MEMBIND_ALLFLAGS (HWLOC_MEMBIND_PROCESS|HWLOC_MEMBIND_THREAD|HWLOC_MEMBIND_STRICT|HWLOC_MEMBIND_MIGRATE|HWLOC_MEMBIND_NOCPUBIND|HWLOC_MEMBIND_BYNODESET) + +static hwloc_const_nodeset_t +hwloc_fix_membind(hwloc_topology_t topology, hwloc_const_nodeset_t nodeset) +{ + hwloc_const_bitmap_t topology_nodeset = hwloc_topology_get_topology_nodeset(topology); + hwloc_const_bitmap_t complete_nodeset = hwloc_topology_get_complete_nodeset(topology); + + if (hwloc_bitmap_iszero(nodeset)) { + errno = EINVAL; + return NULL; + } + + if (!hwloc_bitmap_isincluded(nodeset, complete_nodeset)) { + errno = EINVAL; + return NULL; + } + + if (hwloc_bitmap_isincluded(topology_nodeset, nodeset)) + return complete_nodeset; + + return nodeset; +} + +static int +hwloc_fix_membind_cpuset(hwloc_topology_t topology, hwloc_nodeset_t nodeset, hwloc_const_cpuset_t cpuset) +{ + hwloc_const_bitmap_t topology_set = hwloc_topology_get_topology_cpuset(topology); + hwloc_const_bitmap_t complete_set = hwloc_topology_get_complete_cpuset(topology); + hwloc_const_bitmap_t complete_nodeset = hwloc_topology_get_complete_nodeset(topology); + + if (hwloc_bitmap_iszero(cpuset)) { + errno = EINVAL; + return -1; + } + + if (!hwloc_bitmap_isincluded(cpuset, complete_set)) { + errno = EINVAL; + return -1; + } + + if (hwloc_bitmap_isincluded(topology_set, cpuset)) { + hwloc_bitmap_copy(nodeset, complete_nodeset); + return 0; + } + + hwloc_cpuset_to_nodeset(topology, cpuset, nodeset); + return 0; +} + +static __hwloc_inline int hwloc__check_membind_policy(hwloc_membind_policy_t policy) +{ + if (policy == HWLOC_MEMBIND_DEFAULT + || policy == HWLOC_MEMBIND_FIRSTTOUCH + || policy == HWLOC_MEMBIND_BIND + || policy == HWLOC_MEMBIND_INTERLEAVE + || policy == HWLOC_MEMBIND_NEXTTOUCH) + return 0; + return -1; +} + +static int +hwloc_set_membind_by_nodeset(hwloc_topology_t topology, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags) +{ + if ((flags & ~HWLOC_MEMBIND_ALLFLAGS) || hwloc__check_membind_policy(policy) < 0) { + errno = EINVAL; + return -1; + } + + nodeset = hwloc_fix_membind(topology, nodeset); + if (!nodeset) + return -1; + + if (flags & HWLOC_MEMBIND_PROCESS) { + if (topology->binding_hooks.set_thisproc_membind) + return topology->binding_hooks.set_thisproc_membind(topology, nodeset, policy, flags); + } else if (flags & HWLOC_MEMBIND_THREAD) { + if (topology->binding_hooks.set_thisthread_membind) + return topology->binding_hooks.set_thisthread_membind(topology, nodeset, policy, flags); + } else { + if (topology->binding_hooks.set_thisproc_membind) { + int err = topology->binding_hooks.set_thisproc_membind(topology, nodeset, policy, flags); + if (err >= 0 || errno != ENOSYS) + return err; + /* ENOSYS, fallback */ + } + if (topology->binding_hooks.set_thisthread_membind) + return topology->binding_hooks.set_thisthread_membind(topology, nodeset, policy, flags); + } + + errno = ENOSYS; + return -1; +} + +int +hwloc_set_membind(hwloc_topology_t topology, hwloc_const_bitmap_t set, hwloc_membind_policy_t policy, int flags) +{ + int ret; + + if (flags & HWLOC_MEMBIND_BYNODESET) { + ret = hwloc_set_membind_by_nodeset(topology, set, policy, flags); + } else { + hwloc_nodeset_t nodeset = hwloc_bitmap_alloc(); + if (hwloc_fix_membind_cpuset(topology, nodeset, set)) + ret = -1; + else + ret = hwloc_set_membind_by_nodeset(topology, nodeset, policy, flags); + hwloc_bitmap_free(nodeset); + } + return ret; +} + +static int +hwloc_get_membind_by_nodeset(hwloc_topology_t topology, hwloc_nodeset_t nodeset, hwloc_membind_policy_t * policy, int flags) +{ + if (flags & ~HWLOC_MEMBIND_ALLFLAGS) { + errno = EINVAL; + return -1; + } + + if (flags & HWLOC_MEMBIND_PROCESS) { + if (topology->binding_hooks.get_thisproc_membind) + return topology->binding_hooks.get_thisproc_membind(topology, nodeset, policy, flags); + } else if (flags & HWLOC_MEMBIND_THREAD) { + if (topology->binding_hooks.get_thisthread_membind) + return topology->binding_hooks.get_thisthread_membind(topology, nodeset, policy, flags); + } else { + if (topology->binding_hooks.get_thisproc_membind) { + int err = topology->binding_hooks.get_thisproc_membind(topology, nodeset, policy, flags); + if (err >= 0 || errno != ENOSYS) + return err; + /* ENOSYS, fallback */ + } + if (topology->binding_hooks.get_thisthread_membind) + return topology->binding_hooks.get_thisthread_membind(topology, nodeset, policy, flags); + } + + errno = ENOSYS; + return -1; +} + +int +hwloc_get_membind(hwloc_topology_t topology, hwloc_bitmap_t set, hwloc_membind_policy_t * policy, int flags) +{ + int ret; + + if (flags & HWLOC_MEMBIND_BYNODESET) { + ret = hwloc_get_membind_by_nodeset(topology, set, policy, flags); + } else { + hwloc_nodeset_t nodeset = hwloc_bitmap_alloc(); + ret = hwloc_get_membind_by_nodeset(topology, nodeset, policy, flags); + if (!ret) + hwloc_cpuset_from_nodeset(topology, set, nodeset); + hwloc_bitmap_free(nodeset); + } + + return ret; +} + +static int +hwloc_set_proc_membind_by_nodeset(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags) +{ + if ((flags & ~HWLOC_MEMBIND_ALLFLAGS) || hwloc__check_membind_policy(policy) < 0) { + errno = EINVAL; + return -1; + } + + nodeset = hwloc_fix_membind(topology, nodeset); + if (!nodeset) + return -1; + + if (topology->binding_hooks.set_proc_membind) + return topology->binding_hooks.set_proc_membind(topology, pid, nodeset, policy, flags); + + errno = ENOSYS; + return -1; +} + + +int +hwloc_set_proc_membind(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_const_bitmap_t set, hwloc_membind_policy_t policy, int flags) +{ + int ret; + + if (flags & HWLOC_MEMBIND_BYNODESET) { + ret = hwloc_set_proc_membind_by_nodeset(topology, pid, set, policy, flags); + } else { + hwloc_nodeset_t nodeset = hwloc_bitmap_alloc(); + if (hwloc_fix_membind_cpuset(topology, nodeset, set)) + ret = -1; + else + ret = hwloc_set_proc_membind_by_nodeset(topology, pid, nodeset, policy, flags); + hwloc_bitmap_free(nodeset); + } + + return ret; +} + +static int +hwloc_get_proc_membind_by_nodeset(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_nodeset_t nodeset, hwloc_membind_policy_t * policy, int flags) +{ + if (flags & ~HWLOC_MEMBIND_ALLFLAGS) { + errno = EINVAL; + return -1; + } + + if (topology->binding_hooks.get_proc_membind) + return topology->binding_hooks.get_proc_membind(topology, pid, nodeset, policy, flags); + + errno = ENOSYS; + return -1; +} + +int +hwloc_get_proc_membind(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_bitmap_t set, hwloc_membind_policy_t * policy, int flags) +{ + int ret; + + if (flags & HWLOC_MEMBIND_BYNODESET) { + ret = hwloc_get_proc_membind_by_nodeset(topology, pid, set, policy, flags); + } else { + hwloc_nodeset_t nodeset = hwloc_bitmap_alloc(); + ret = hwloc_get_proc_membind_by_nodeset(topology, pid, nodeset, policy, flags); + if (!ret) + hwloc_cpuset_from_nodeset(topology, set, nodeset); + hwloc_bitmap_free(nodeset); + } + + return ret; +} + +static int +hwloc_set_area_membind_by_nodeset(hwloc_topology_t topology, const void *addr, size_t len, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags) +{ + if ((flags & ~HWLOC_MEMBIND_ALLFLAGS) || hwloc__check_membind_policy(policy) < 0) { + errno = EINVAL; + return -1; + } + + if (!len) + /* nothing to do */ + return 0; + + nodeset = hwloc_fix_membind(topology, nodeset); + if (!nodeset) + return -1; + + if (topology->binding_hooks.set_area_membind) + return topology->binding_hooks.set_area_membind(topology, addr, len, nodeset, policy, flags); + + errno = ENOSYS; + return -1; +} + +int +hwloc_set_area_membind(hwloc_topology_t topology, const void *addr, size_t len, hwloc_const_bitmap_t set, hwloc_membind_policy_t policy, int flags) +{ + int ret; + + if (flags & HWLOC_MEMBIND_BYNODESET) { + ret = hwloc_set_area_membind_by_nodeset(topology, addr, len, set, policy, flags); + } else { + hwloc_nodeset_t nodeset = hwloc_bitmap_alloc(); + if (hwloc_fix_membind_cpuset(topology, nodeset, set)) + ret = -1; + else + ret = hwloc_set_area_membind_by_nodeset(topology, addr, len, nodeset, policy, flags); + hwloc_bitmap_free(nodeset); + } + + return ret; +} + +static int +hwloc_get_area_membind_by_nodeset(hwloc_topology_t topology, const void *addr, size_t len, hwloc_nodeset_t nodeset, hwloc_membind_policy_t * policy, int flags) +{ + if (flags & ~HWLOC_MEMBIND_ALLFLAGS) { + errno = EINVAL; + return -1; + } + + if (!len) { + /* nothing to query */ + errno = EINVAL; + return -1; + } + + if (topology->binding_hooks.get_area_membind) + return topology->binding_hooks.get_area_membind(topology, addr, len, nodeset, policy, flags); + + errno = ENOSYS; + return -1; +} + +int +hwloc_get_area_membind(hwloc_topology_t topology, const void *addr, size_t len, hwloc_bitmap_t set, hwloc_membind_policy_t * policy, int flags) +{ + int ret; + + if (flags & HWLOC_MEMBIND_BYNODESET) { + ret = hwloc_get_area_membind_by_nodeset(topology, addr, len, set, policy, flags); + } else { + hwloc_nodeset_t nodeset = hwloc_bitmap_alloc(); + ret = hwloc_get_area_membind_by_nodeset(topology, addr, len, nodeset, policy, flags); + if (!ret) + hwloc_cpuset_from_nodeset(topology, set, nodeset); + hwloc_bitmap_free(nodeset); + } + + return ret; +} + +static int +hwloc_get_area_memlocation_by_nodeset(hwloc_topology_t topology, const void *addr, size_t len, hwloc_nodeset_t nodeset, int flags) +{ + if (flags & ~HWLOC_MEMBIND_ALLFLAGS) { + errno = EINVAL; + return -1; + } + + if (!len) + /* nothing to do */ + return 0; + + if (topology->binding_hooks.get_area_memlocation) + return topology->binding_hooks.get_area_memlocation(topology, addr, len, nodeset, flags); + + errno = ENOSYS; + return -1; +} + +int +hwloc_get_area_memlocation(hwloc_topology_t topology, const void *addr, size_t len, hwloc_cpuset_t set, int flags) +{ + int ret; + + if (flags & HWLOC_MEMBIND_BYNODESET) { + ret = hwloc_get_area_memlocation_by_nodeset(topology, addr, len, set, flags); + } else { + hwloc_nodeset_t nodeset = hwloc_bitmap_alloc(); + ret = hwloc_get_area_memlocation_by_nodeset(topology, addr, len, nodeset, flags); + if (!ret) + hwloc_cpuset_from_nodeset(topology, set, nodeset); + hwloc_bitmap_free(nodeset); + } + + return ret; +} + +void * +hwloc_alloc_heap(hwloc_topology_t topology __hwloc_attribute_unused, size_t len) +{ + void *p = NULL; +#if defined(hwloc_getpagesize) && defined(HAVE_POSIX_MEMALIGN) + errno = posix_memalign(&p, hwloc_getpagesize(), len); + if (errno) + p = NULL; +#elif defined(hwloc_getpagesize) && defined(HAVE_MEMALIGN) + p = memalign(hwloc_getpagesize(), len); +#else + p = malloc(len); +#endif + return p; +} + +#ifdef MAP_ANONYMOUS +void * +hwloc_alloc_mmap(hwloc_topology_t topology __hwloc_attribute_unused, size_t len) +{ + void * buffer = mmap(NULL, len, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0); + return buffer == MAP_FAILED ? NULL : buffer; +} +#endif + +int +hwloc_free_heap(hwloc_topology_t topology __hwloc_attribute_unused, void *addr, size_t len __hwloc_attribute_unused) +{ + free(addr); + return 0; +} + +#ifdef MAP_ANONYMOUS +int +hwloc_free_mmap(hwloc_topology_t topology __hwloc_attribute_unused, void *addr, size_t len) +{ + if (!addr) + return 0; + return munmap(addr, len); +} +#endif + +void * +hwloc_alloc(hwloc_topology_t topology, size_t len) +{ + if (topology->binding_hooks.alloc) + return topology->binding_hooks.alloc(topology, len); + return hwloc_alloc_heap(topology, len); +} + +static void * +hwloc_alloc_membind_by_nodeset(hwloc_topology_t topology, size_t len, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags) +{ + void *p; + + if ((flags & ~HWLOC_MEMBIND_ALLFLAGS) || hwloc__check_membind_policy(policy) < 0) { + errno = EINVAL; + return NULL; + } + + nodeset = hwloc_fix_membind(topology, nodeset); + if (!nodeset) + goto fallback; + if (flags & HWLOC_MEMBIND_MIGRATE) { + errno = EINVAL; + goto fallback; + } + + if (topology->binding_hooks.alloc_membind) + return topology->binding_hooks.alloc_membind(topology, len, nodeset, policy, flags); + else if (topology->binding_hooks.set_area_membind) { + p = hwloc_alloc(topology, len); + if (!p) + return NULL; + if (topology->binding_hooks.set_area_membind(topology, p, len, nodeset, policy, flags) && flags & HWLOC_MEMBIND_STRICT) { + int error = errno; + free(p); + errno = error; + return NULL; + } + return p; + } else { + errno = ENOSYS; + } + +fallback: + if (flags & HWLOC_MEMBIND_STRICT) + /* Report error */ + return NULL; + /* Never mind, allocate anyway */ + return hwloc_alloc(topology, len); +} + +void * +hwloc_alloc_membind(hwloc_topology_t topology, size_t len, hwloc_const_bitmap_t set, hwloc_membind_policy_t policy, int flags) +{ + void *ret; + + if (flags & HWLOC_MEMBIND_BYNODESET) { + ret = hwloc_alloc_membind_by_nodeset(topology, len, set, policy, flags); + } else { + hwloc_nodeset_t nodeset = hwloc_bitmap_alloc(); + if (hwloc_fix_membind_cpuset(topology, nodeset, set)) { + if (flags & HWLOC_MEMBIND_STRICT) + ret = NULL; + else + ret = hwloc_alloc(topology, len); + } else + ret = hwloc_alloc_membind_by_nodeset(topology, len, nodeset, policy, flags); + hwloc_bitmap_free(nodeset); + } + + return ret; +} + +int +hwloc_free(hwloc_topology_t topology, void *addr, size_t len) +{ + if (topology->binding_hooks.free_membind) + return topology->binding_hooks.free_membind(topology, addr, len); + return hwloc_free_heap(topology, addr, len); +} + +/* + * Empty binding hooks always returning success + */ + +static int dontset_return_complete_cpuset(hwloc_topology_t topology, hwloc_cpuset_t set) +{ + hwloc_bitmap_copy(set, hwloc_topology_get_complete_cpuset(topology)); + return 0; +} + +static int dontset_thisthread_cpubind(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_const_bitmap_t set __hwloc_attribute_unused, int flags __hwloc_attribute_unused) +{ + return 0; +} +static int dontget_thisthread_cpubind(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_bitmap_t set, int flags __hwloc_attribute_unused) +{ + return dontset_return_complete_cpuset(topology, set); +} +static int dontset_thisproc_cpubind(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_const_bitmap_t set __hwloc_attribute_unused, int flags __hwloc_attribute_unused) +{ + return 0; +} +static int dontget_thisproc_cpubind(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_bitmap_t set, int flags __hwloc_attribute_unused) +{ + return dontset_return_complete_cpuset(topology, set); +} +static int dontset_proc_cpubind(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_pid_t pid __hwloc_attribute_unused, hwloc_const_bitmap_t set __hwloc_attribute_unused, int flags __hwloc_attribute_unused) +{ + return 0; +} +static int dontget_proc_cpubind(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_pid_t pid __hwloc_attribute_unused, hwloc_bitmap_t cpuset, int flags __hwloc_attribute_unused) +{ + return dontset_return_complete_cpuset(topology, cpuset); +} +#ifdef hwloc_thread_t +static int dontset_thread_cpubind(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_thread_t tid __hwloc_attribute_unused, hwloc_const_bitmap_t set __hwloc_attribute_unused, int flags __hwloc_attribute_unused) +{ + return 0; +} +static int dontget_thread_cpubind(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_thread_t tid __hwloc_attribute_unused, hwloc_bitmap_t cpuset, int flags __hwloc_attribute_unused) +{ + return dontset_return_complete_cpuset(topology, cpuset); +} +#endif + +static int dontset_return_complete_nodeset(hwloc_topology_t topology, hwloc_nodeset_t set, hwloc_membind_policy_t *policy) +{ + hwloc_bitmap_copy(set, hwloc_topology_get_complete_nodeset(topology)); + *policy = HWLOC_MEMBIND_MIXED; + return 0; +} + +static int dontset_thisproc_membind(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_const_bitmap_t set __hwloc_attribute_unused, hwloc_membind_policy_t policy __hwloc_attribute_unused, int flags __hwloc_attribute_unused) +{ + return 0; +} +static int dontget_thisproc_membind(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_bitmap_t set, hwloc_membind_policy_t * policy, int flags __hwloc_attribute_unused) +{ + return dontset_return_complete_nodeset(topology, set, policy); +} + +static int dontset_thisthread_membind(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_const_bitmap_t set __hwloc_attribute_unused, hwloc_membind_policy_t policy __hwloc_attribute_unused, int flags __hwloc_attribute_unused) +{ + return 0; +} +static int dontget_thisthread_membind(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_bitmap_t set, hwloc_membind_policy_t * policy, int flags __hwloc_attribute_unused) +{ + return dontset_return_complete_nodeset(topology, set, policy); +} + +static int dontset_proc_membind(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_pid_t pid __hwloc_attribute_unused, hwloc_const_bitmap_t set __hwloc_attribute_unused, hwloc_membind_policy_t policy __hwloc_attribute_unused, int flags __hwloc_attribute_unused) +{ + return 0; +} +static int dontget_proc_membind(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_pid_t pid __hwloc_attribute_unused, hwloc_bitmap_t set, hwloc_membind_policy_t * policy, int flags __hwloc_attribute_unused) +{ + return dontset_return_complete_nodeset(topology, set, policy); +} + +static int dontset_area_membind(hwloc_topology_t topology __hwloc_attribute_unused, const void *addr __hwloc_attribute_unused, size_t size __hwloc_attribute_unused, hwloc_const_bitmap_t set __hwloc_attribute_unused, hwloc_membind_policy_t policy __hwloc_attribute_unused, int flags __hwloc_attribute_unused) +{ + return 0; +} +static int dontget_area_membind(hwloc_topology_t topology __hwloc_attribute_unused, const void *addr __hwloc_attribute_unused, size_t size __hwloc_attribute_unused, hwloc_bitmap_t set, hwloc_membind_policy_t * policy, int flags __hwloc_attribute_unused) +{ + return dontset_return_complete_nodeset(topology, set, policy); +} +static int dontget_area_memlocation(hwloc_topology_t topology __hwloc_attribute_unused, const void *addr __hwloc_attribute_unused, size_t size __hwloc_attribute_unused, hwloc_bitmap_t set, int flags __hwloc_attribute_unused) +{ + hwloc_membind_policy_t policy; + return dontset_return_complete_nodeset(topology, set, &policy); +} + +static void * dontalloc_membind(hwloc_topology_t topology __hwloc_attribute_unused, size_t size __hwloc_attribute_unused, hwloc_const_bitmap_t set __hwloc_attribute_unused, hwloc_membind_policy_t policy __hwloc_attribute_unused, int flags __hwloc_attribute_unused) +{ + return malloc(size); +} +static int dontfree_membind(hwloc_topology_t topology __hwloc_attribute_unused, void *addr __hwloc_attribute_unused, size_t size __hwloc_attribute_unused) +{ + free(addr); + return 0; +} + +static void hwloc_set_dummy_hooks(struct hwloc_binding_hooks *hooks, + struct hwloc_topology_support *support __hwloc_attribute_unused) +{ + hooks->set_thisproc_cpubind = dontset_thisproc_cpubind; + hooks->get_thisproc_cpubind = dontget_thisproc_cpubind; + hooks->set_thisthread_cpubind = dontset_thisthread_cpubind; + hooks->get_thisthread_cpubind = dontget_thisthread_cpubind; + hooks->set_proc_cpubind = dontset_proc_cpubind; + hooks->get_proc_cpubind = dontget_proc_cpubind; +#ifdef hwloc_thread_t + hooks->set_thread_cpubind = dontset_thread_cpubind; + hooks->get_thread_cpubind = dontget_thread_cpubind; +#endif + hooks->get_thisproc_last_cpu_location = dontget_thisproc_cpubind; /* cpubind instead of last_cpu_location is ok */ + hooks->get_thisthread_last_cpu_location = dontget_thisthread_cpubind; /* cpubind instead of last_cpu_location is ok */ + hooks->get_proc_last_cpu_location = dontget_proc_cpubind; /* cpubind instead of last_cpu_location is ok */ + /* TODO: get_thread_last_cpu_location */ + hooks->set_thisproc_membind = dontset_thisproc_membind; + hooks->get_thisproc_membind = dontget_thisproc_membind; + hooks->set_thisthread_membind = dontset_thisthread_membind; + hooks->get_thisthread_membind = dontget_thisthread_membind; + hooks->set_proc_membind = dontset_proc_membind; + hooks->get_proc_membind = dontget_proc_membind; + hooks->set_area_membind = dontset_area_membind; + hooks->get_area_membind = dontget_area_membind; + hooks->get_area_memlocation = dontget_area_memlocation; + hooks->alloc_membind = dontalloc_membind; + hooks->free_membind = dontfree_membind; +} + +void +hwloc_set_native_binding_hooks(struct hwloc_binding_hooks *hooks, struct hwloc_topology_support *support) +{ +# ifdef HWLOC_LINUX_SYS + hwloc_set_linuxfs_hooks(hooks, support); +# endif /* HWLOC_LINUX_SYS */ + +# ifdef HWLOC_BGQ_SYS + hwloc_set_bgq_hooks(hooks, support); +# endif /* HWLOC_BGQ_SYS */ + +# ifdef HWLOC_AIX_SYS + hwloc_set_aix_hooks(hooks, support); +# endif /* HWLOC_AIX_SYS */ + +# ifdef HWLOC_SOLARIS_SYS + hwloc_set_solaris_hooks(hooks, support); +# endif /* HWLOC_SOLARIS_SYS */ + +# ifdef HWLOC_WIN_SYS + hwloc_set_windows_hooks(hooks, support); +# endif /* HWLOC_WIN_SYS */ + +# ifdef HWLOC_DARWIN_SYS + hwloc_set_darwin_hooks(hooks, support); +# endif /* HWLOC_DARWIN_SYS */ + +# ifdef HWLOC_FREEBSD_SYS + hwloc_set_freebsd_hooks(hooks, support); +# endif /* HWLOC_FREEBSD_SYS */ + +# ifdef HWLOC_NETBSD_SYS + hwloc_set_netbsd_hooks(hooks, support); +# endif /* HWLOC_NETBSD_SYS */ + +# ifdef HWLOC_HPUX_SYS + hwloc_set_hpux_hooks(hooks, support); +# endif /* HWLOC_HPUX_SYS */ +} + +/* If the represented system is actually not this system, use dummy binding hooks. */ +void +hwloc_set_binding_hooks(struct hwloc_topology *topology) +{ + if (topology->is_thissystem) { + hwloc_set_native_binding_hooks(&topology->binding_hooks, &topology->support); + /* every hook not set above will return ENOSYS */ + } else { + /* not this system, use dummy binding hooks that do nothing (but don't return ENOSYS) */ + hwloc_set_dummy_hooks(&topology->binding_hooks, &topology->support); + } + + /* if not is_thissystem, set_cpubind is fake + * and get_cpubind returns the whole system cpuset, + * so don't report that set/get_cpubind as supported + */ + if (topology->is_thissystem) { +#define DO(which,kind) \ + if (topology->binding_hooks.kind) \ + topology->support.which##bind->kind = 1; + DO(cpu,set_thisproc_cpubind); + DO(cpu,get_thisproc_cpubind); + DO(cpu,set_proc_cpubind); + DO(cpu,get_proc_cpubind); + DO(cpu,set_thisthread_cpubind); + DO(cpu,get_thisthread_cpubind); +#ifdef hwloc_thread_t + DO(cpu,set_thread_cpubind); + DO(cpu,get_thread_cpubind); +#endif + DO(cpu,get_thisproc_last_cpu_location); + DO(cpu,get_proc_last_cpu_location); + DO(cpu,get_thisthread_last_cpu_location); + DO(mem,set_thisproc_membind); + DO(mem,get_thisproc_membind); + DO(mem,set_thisthread_membind); + DO(mem,get_thisthread_membind); + DO(mem,set_proc_membind); + DO(mem,get_proc_membind); + DO(mem,set_area_membind); + DO(mem,get_area_membind); + DO(mem,get_area_memlocation); + DO(mem,alloc_membind); + } +} diff --git a/src/3rdparty/hwloc/src/bitmap.c b/src/3rdparty/hwloc/src/bitmap.c new file mode 100644 index 000000000..ea1264afc --- /dev/null +++ b/src/3rdparty/hwloc/src/bitmap.c @@ -0,0 +1,1676 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2017 Inria. All rights reserved. + * Copyright © 2009-2011 Université Bordeaux + * Copyright © 2009-2011 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include +#include + +/* + * possible improvements: + * - have a way to change the initial allocation size: + * add hwloc_bitmap_set_foo() to changes a global here, + * and make the hwloc core call based on the early number of PUs + * - make HWLOC_BITMAP_PREALLOC_BITS configurable, and detectable + * by parsing /proc/cpuinfo during configure on Linux. + * - preallocate inside the bitmap structure (so that the whole structure is a cacheline for instance) + * and allocate a dedicated array only later when reallocating larger + * - add a bitmap->ulongs_empty_first which guarantees that some first ulongs are empty, + * making tests much faster for big bitmaps since there's no need to look at first ulongs. + * no need for ulongs_empty_first to be exactly the max number of empty ulongs, + * clearing bits that were set earlier isn't very common. + */ + +/* magic number */ +#define HWLOC_BITMAP_MAGIC 0x20091007 + +/* preallocated bits in every bitmap */ +#define HWLOC_BITMAP_PREALLOC_BITS 512 +#define HWLOC_BITMAP_PREALLOC_ULONGS (HWLOC_BITMAP_PREALLOC_BITS/HWLOC_BITS_PER_LONG) + +/* actual opaque type internals */ +struct hwloc_bitmap_s { + unsigned ulongs_count; /* how many ulong bitmasks are valid, >= 1 */ + unsigned ulongs_allocated; /* how many ulong bitmasks are allocated, >= ulongs_count */ + unsigned long *ulongs; + int infinite; /* set to 1 if all bits beyond ulongs are set */ +#ifdef HWLOC_DEBUG + int magic; +#endif +}; + +/* overzealous check in debug-mode, not as powerful as valgrind but still useful */ +#ifdef HWLOC_DEBUG +#define HWLOC__BITMAP_CHECK(set) do { \ + assert((set)->magic == HWLOC_BITMAP_MAGIC); \ + assert((set)->ulongs_count >= 1); \ + assert((set)->ulongs_allocated >= (set)->ulongs_count); \ +} while (0) +#else +#define HWLOC__BITMAP_CHECK(set) +#endif + +/* extract a subset from a set using an index or a cpu */ +#define HWLOC_SUBBITMAP_INDEX(cpu) ((cpu)/(HWLOC_BITS_PER_LONG)) +#define HWLOC_SUBBITMAP_CPU_ULBIT(cpu) ((cpu)%(HWLOC_BITS_PER_LONG)) +/* Read from a bitmap ulong without knowing whether x is valid. + * Writers should make sure that x is valid and modify set->ulongs[x] directly. + */ +#define HWLOC_SUBBITMAP_READULONG(set,x) ((x) < (set)->ulongs_count ? (set)->ulongs[x] : (set)->infinite ? HWLOC_SUBBITMAP_FULL : HWLOC_SUBBITMAP_ZERO) + +/* predefined subset values */ +#define HWLOC_SUBBITMAP_ZERO 0UL +#define HWLOC_SUBBITMAP_FULL (~0UL) +#define HWLOC_SUBBITMAP_ULBIT(bit) (1UL<<(bit)) +#define HWLOC_SUBBITMAP_CPU(cpu) HWLOC_SUBBITMAP_ULBIT(HWLOC_SUBBITMAP_CPU_ULBIT(cpu)) +#define HWLOC_SUBBITMAP_ULBIT_TO(bit) (HWLOC_SUBBITMAP_FULL>>(HWLOC_BITS_PER_LONG-1-(bit))) +#define HWLOC_SUBBITMAP_ULBIT_FROM(bit) (HWLOC_SUBBITMAP_FULL<<(bit)) +#define HWLOC_SUBBITMAP_ULBIT_FROMTO(begin,end) (HWLOC_SUBBITMAP_ULBIT_TO(end) & HWLOC_SUBBITMAP_ULBIT_FROM(begin)) + +struct hwloc_bitmap_s * hwloc_bitmap_alloc(void) +{ + struct hwloc_bitmap_s * set; + + set = malloc(sizeof(struct hwloc_bitmap_s)); + if (!set) + return NULL; + + set->ulongs_count = 1; + set->ulongs_allocated = HWLOC_BITMAP_PREALLOC_ULONGS; + set->ulongs = malloc(HWLOC_BITMAP_PREALLOC_ULONGS * sizeof(unsigned long)); + if (!set->ulongs) { + free(set); + return NULL; + } + + set->ulongs[0] = HWLOC_SUBBITMAP_ZERO; + set->infinite = 0; +#ifdef HWLOC_DEBUG + set->magic = HWLOC_BITMAP_MAGIC; +#endif + return set; +} + +struct hwloc_bitmap_s * hwloc_bitmap_alloc_full(void) +{ + struct hwloc_bitmap_s * set = hwloc_bitmap_alloc(); + if (set) { + set->infinite = 1; + set->ulongs[0] = HWLOC_SUBBITMAP_FULL; + } + return set; +} + +void hwloc_bitmap_free(struct hwloc_bitmap_s * set) +{ + if (!set) + return; + + HWLOC__BITMAP_CHECK(set); +#ifdef HWLOC_DEBUG + set->magic = 0; +#endif + + free(set->ulongs); + free(set); +} + +/* enlarge until it contains at least needed_count ulongs. + */ +static int +hwloc_bitmap_enlarge_by_ulongs(struct hwloc_bitmap_s * set, unsigned needed_count) __hwloc_attribute_warn_unused_result; +static int +hwloc_bitmap_enlarge_by_ulongs(struct hwloc_bitmap_s * set, unsigned needed_count) +{ + unsigned tmp = 1U << hwloc_flsl((unsigned long) needed_count - 1); + if (tmp > set->ulongs_allocated) { + unsigned long *tmpulongs; + tmpulongs = realloc(set->ulongs, tmp * sizeof(unsigned long)); + if (!tmpulongs) + return -1; + set->ulongs = tmpulongs; + set->ulongs_allocated = tmp; + } + return 0; +} + +/* enlarge until it contains at least needed_count ulongs, + * and update new ulongs according to the infinite field. + */ +static int +hwloc_bitmap_realloc_by_ulongs(struct hwloc_bitmap_s * set, unsigned needed_count) __hwloc_attribute_warn_unused_result; +static int +hwloc_bitmap_realloc_by_ulongs(struct hwloc_bitmap_s * set, unsigned needed_count) +{ + unsigned i; + + HWLOC__BITMAP_CHECK(set); + + if (needed_count <= set->ulongs_count) + return 0; + + /* realloc larger if needed */ + if (hwloc_bitmap_enlarge_by_ulongs(set, needed_count) < 0) + return -1; + + /* fill the newly allocated subset depending on the infinite flag */ + for(i=set->ulongs_count; iulongs[i] = set->infinite ? HWLOC_SUBBITMAP_FULL : HWLOC_SUBBITMAP_ZERO; + set->ulongs_count = needed_count; + return 0; +} + +/* realloc until it contains at least cpu+1 bits */ +#define hwloc_bitmap_realloc_by_cpu_index(set, cpu) hwloc_bitmap_realloc_by_ulongs(set, ((cpu)/HWLOC_BITS_PER_LONG)+1) + +/* reset a bitmap to exactely the needed size. + * the caller must reinitialize all ulongs and the infinite flag later. + */ +static int +hwloc_bitmap_reset_by_ulongs(struct hwloc_bitmap_s * set, unsigned needed_count) __hwloc_attribute_warn_unused_result; +static int +hwloc_bitmap_reset_by_ulongs(struct hwloc_bitmap_s * set, unsigned needed_count) +{ + if (hwloc_bitmap_enlarge_by_ulongs(set, needed_count)) + return -1; + set->ulongs_count = needed_count; + return 0; +} + +/* reset until it contains exactly cpu+1 bits (roundup to a ulong). + * the caller must reinitialize all ulongs and the infinite flag later. + */ +#define hwloc_bitmap_reset_by_cpu_index(set, cpu) hwloc_bitmap_reset_by_ulongs(set, ((cpu)/HWLOC_BITS_PER_LONG)+1) + +struct hwloc_bitmap_s * hwloc_bitmap_tma_dup(struct hwloc_tma *tma, const struct hwloc_bitmap_s * old) +{ + struct hwloc_bitmap_s * new; + + if (!old) + return NULL; + + HWLOC__BITMAP_CHECK(old); + + new = hwloc_tma_malloc(tma, sizeof(struct hwloc_bitmap_s)); + if (!new) + return NULL; + + new->ulongs = hwloc_tma_malloc(tma, old->ulongs_allocated * sizeof(unsigned long)); + if (!new->ulongs) { + free(new); + return NULL; + } + new->ulongs_allocated = old->ulongs_allocated; + new->ulongs_count = old->ulongs_count; + memcpy(new->ulongs, old->ulongs, new->ulongs_count * sizeof(unsigned long)); + new->infinite = old->infinite; +#ifdef HWLOC_DEBUG + new->magic = HWLOC_BITMAP_MAGIC; +#endif + return new; +} + +struct hwloc_bitmap_s * hwloc_bitmap_dup(const struct hwloc_bitmap_s * old) +{ + return hwloc_bitmap_tma_dup(NULL, old); +} + +int hwloc_bitmap_copy(struct hwloc_bitmap_s * dst, const struct hwloc_bitmap_s * src) +{ + HWLOC__BITMAP_CHECK(dst); + HWLOC__BITMAP_CHECK(src); + + if (hwloc_bitmap_reset_by_ulongs(dst, src->ulongs_count) < 0) + return -1; + + memcpy(dst->ulongs, src->ulongs, src->ulongs_count * sizeof(unsigned long)); + dst->infinite = src->infinite; + return 0; +} + +/* Strings always use 32bit groups */ +#define HWLOC_PRIxSUBBITMAP "%08lx" +#define HWLOC_BITMAP_SUBSTRING_SIZE 32 +#define HWLOC_BITMAP_SUBSTRING_LENGTH (HWLOC_BITMAP_SUBSTRING_SIZE/4) +#define HWLOC_BITMAP_STRING_PER_LONG (HWLOC_BITS_PER_LONG/HWLOC_BITMAP_SUBSTRING_SIZE) + +int hwloc_bitmap_snprintf(char * __hwloc_restrict buf, size_t buflen, const struct hwloc_bitmap_s * __hwloc_restrict set) +{ + ssize_t size = buflen; + char *tmp = buf; + int res, ret = 0; + int needcomma = 0; + int i; + unsigned long accum = 0; + int accumed = 0; +#if HWLOC_BITS_PER_LONG == HWLOC_BITMAP_SUBSTRING_SIZE + const unsigned long accum_mask = ~0UL; +#else /* HWLOC_BITS_PER_LONG != HWLOC_BITMAP_SUBSTRING_SIZE */ + const unsigned long accum_mask = ((1UL << HWLOC_BITMAP_SUBSTRING_SIZE) - 1) << (HWLOC_BITS_PER_LONG - HWLOC_BITMAP_SUBSTRING_SIZE); +#endif /* HWLOC_BITS_PER_LONG != HWLOC_BITMAP_SUBSTRING_SIZE */ + + HWLOC__BITMAP_CHECK(set); + + /* mark the end in case we do nothing later */ + if (buflen > 0) + tmp[0] = '\0'; + + if (set->infinite) { + res = hwloc_snprintf(tmp, size, "0xf...f"); + needcomma = 1; + if (res < 0) + return -1; + ret += res; + if (res >= size) + res = size>0 ? (int)size - 1 : 0; + tmp += res; + size -= res; + } + + i=(int) set->ulongs_count-1; + + if (set->infinite) { + /* ignore starting FULL since we have 0xf...f already */ + while (i>=0 && set->ulongs[i] == HWLOC_SUBBITMAP_FULL) + i--; + } else { + /* ignore starting ZERO except the last one */ + while (i>=0 && set->ulongs[i] == HWLOC_SUBBITMAP_ZERO) + i--; + } + + while (i>=0 || accumed) { + /* Refill accumulator */ + if (!accumed) { + accum = set->ulongs[i--]; + accumed = HWLOC_BITS_PER_LONG; + } + + if (accum & accum_mask) { + /* print the whole subset if not empty */ + res = hwloc_snprintf(tmp, size, needcomma ? ",0x" HWLOC_PRIxSUBBITMAP : "0x" HWLOC_PRIxSUBBITMAP, + (accum & accum_mask) >> (HWLOC_BITS_PER_LONG - HWLOC_BITMAP_SUBSTRING_SIZE)); + needcomma = 1; + } else if (i == -1 && accumed == HWLOC_BITMAP_SUBSTRING_SIZE) { + /* print a single 0 to mark the last subset */ + res = hwloc_snprintf(tmp, size, needcomma ? ",0x0" : "0x0"); + } else if (needcomma) { + res = hwloc_snprintf(tmp, size, ","); + } else { + res = 0; + } + if (res < 0) + return -1; + ret += res; + +#if HWLOC_BITS_PER_LONG == HWLOC_BITMAP_SUBSTRING_SIZE + accum = 0; + accumed = 0; +#else + accum <<= HWLOC_BITMAP_SUBSTRING_SIZE; + accumed -= HWLOC_BITMAP_SUBSTRING_SIZE; +#endif + + if (res >= size) + res = size>0 ? (int)size - 1 : 0; + + tmp += res; + size -= res; + } + + /* if didn't display anything, display 0x0 */ + if (!ret) { + res = hwloc_snprintf(tmp, size, "0x0"); + if (res < 0) + return -1; + ret += res; + } + + return ret; +} + +int hwloc_bitmap_asprintf(char ** strp, const struct hwloc_bitmap_s * __hwloc_restrict set) +{ + int len; + char *buf; + + HWLOC__BITMAP_CHECK(set); + + len = hwloc_bitmap_snprintf(NULL, 0, set); + buf = malloc(len+1); + if (!buf) + return -1; + *strp = buf; + return hwloc_bitmap_snprintf(buf, len+1, set); +} + +int hwloc_bitmap_sscanf(struct hwloc_bitmap_s *set, const char * __hwloc_restrict string) +{ + const char * current = string; + unsigned long accum = 0; + int count=0; + int infinite = 0; + + /* count how many substrings there are */ + count++; + while ((current = strchr(current+1, ',')) != NULL) + count++; + + current = string; + if (!strncmp("0xf...f", current, 7)) { + current += 7; + if (*current != ',') { + /* special case for infinite/full bitmap */ + hwloc_bitmap_fill(set); + return 0; + } + current++; + infinite = 1; + count--; + } + + if (hwloc_bitmap_reset_by_ulongs(set, (count + HWLOC_BITMAP_STRING_PER_LONG - 1) / HWLOC_BITMAP_STRING_PER_LONG) < 0) + return -1; + set->infinite = 0; + + while (*current != '\0') { + unsigned long val; + char *next; + val = strtoul(current, &next, 16); + + assert(count > 0); + count--; + + accum |= (val << ((count * HWLOC_BITMAP_SUBSTRING_SIZE) % HWLOC_BITS_PER_LONG)); + if (!(count % HWLOC_BITMAP_STRING_PER_LONG)) { + set->ulongs[count / HWLOC_BITMAP_STRING_PER_LONG] = accum; + accum = 0; + } + + if (*next != ',') { + if (*next || count > 0) + goto failed; + else + break; + } + current = (const char*) next+1; + } + + set->infinite = infinite; /* set at the end, to avoid spurious realloc with filled new ulongs */ + + return 0; + + failed: + /* failure to parse */ + hwloc_bitmap_zero(set); + return -1; +} + +int hwloc_bitmap_list_snprintf(char * __hwloc_restrict buf, size_t buflen, const struct hwloc_bitmap_s * __hwloc_restrict set) +{ + int prev = -1; + ssize_t size = buflen; + char *tmp = buf; + int res, ret = 0; + int needcomma = 0; + + HWLOC__BITMAP_CHECK(set); + + /* mark the end in case we do nothing later */ + if (buflen > 0) + tmp[0] = '\0'; + + while (1) { + int begin, end; + + begin = hwloc_bitmap_next(set, prev); + if (begin == -1) + break; + end = hwloc_bitmap_next_unset(set, begin); + + if (end == begin+1) { + res = hwloc_snprintf(tmp, size, needcomma ? ",%d" : "%d", begin); + } else if (end == -1) { + res = hwloc_snprintf(tmp, size, needcomma ? ",%d-" : "%d-", begin); + } else { + res = hwloc_snprintf(tmp, size, needcomma ? ",%d-%d" : "%d-%d", begin, end-1); + } + if (res < 0) + return -1; + ret += res; + + if (res >= size) + res = size>0 ? (int)size - 1 : 0; + + tmp += res; + size -= res; + needcomma = 1; + + if (end == -1) + break; + else + prev = end - 1; + } + + return ret; +} + +int hwloc_bitmap_list_asprintf(char ** strp, const struct hwloc_bitmap_s * __hwloc_restrict set) +{ + int len; + char *buf; + + HWLOC__BITMAP_CHECK(set); + + len = hwloc_bitmap_list_snprintf(NULL, 0, set); + buf = malloc(len+1); + if (!buf) + return -1; + *strp = buf; + return hwloc_bitmap_list_snprintf(buf, len+1, set); +} + +int hwloc_bitmap_list_sscanf(struct hwloc_bitmap_s *set, const char * __hwloc_restrict string) +{ + const char * current = string; + char *next; + long begin = -1, val; + + hwloc_bitmap_zero(set); + + while (*current != '\0') { + + /* ignore empty ranges */ + while (*current == ',' || *current == ' ') + current++; + + val = strtoul(current, &next, 0); + /* make sure we got at least one digit */ + if (next == current) + goto failed; + + if (begin != -1) { + /* finishing a range */ + hwloc_bitmap_set_range(set, begin, val); + begin = -1; + + } else if (*next == '-') { + /* starting a new range */ + if (*(next+1) == '\0') { + /* infinite range */ + hwloc_bitmap_set_range(set, val, -1); + break; + } else { + /* normal range */ + begin = val; + } + + } else if (*next == ',' || *next == ' ' || *next == '\0') { + /* single digit */ + hwloc_bitmap_set(set, val); + } + + if (*next == '\0') + break; + current = next+1; + } + + return 0; + + failed: + /* failure to parse */ + hwloc_bitmap_zero(set); + return -1; +} + +int hwloc_bitmap_taskset_snprintf(char * __hwloc_restrict buf, size_t buflen, const struct hwloc_bitmap_s * __hwloc_restrict set) +{ + ssize_t size = buflen; + char *tmp = buf; + int res, ret = 0; + int started = 0; + int i; + + HWLOC__BITMAP_CHECK(set); + + /* mark the end in case we do nothing later */ + if (buflen > 0) + tmp[0] = '\0'; + + if (set->infinite) { + res = hwloc_snprintf(tmp, size, "0xf...f"); + started = 1; + if (res < 0) + return -1; + ret += res; + if (res >= size) + res = size>0 ? (int)size - 1 : 0; + tmp += res; + size -= res; + } + + i=set->ulongs_count-1; + + if (set->infinite) { + /* ignore starting FULL since we have 0xf...f already */ + while (i>=0 && set->ulongs[i] == HWLOC_SUBBITMAP_FULL) + i--; + } else { + /* ignore starting ZERO except the last one */ + while (i>=1 && set->ulongs[i] == HWLOC_SUBBITMAP_ZERO) + i--; + } + + while (i>=0) { + unsigned long val = set->ulongs[i--]; + if (started) { + /* print the whole subset */ +#if HWLOC_BITS_PER_LONG == 64 + res = hwloc_snprintf(tmp, size, "%016lx", val); +#else + res = hwloc_snprintf(tmp, size, "%08lx", val); +#endif + } else if (val || i == -1) { + res = hwloc_snprintf(tmp, size, "0x%lx", val); + started = 1; + } else { + res = 0; + } + if (res < 0) + return -1; + ret += res; + if (res >= size) + res = size>0 ? (int)size - 1 : 0; + tmp += res; + size -= res; + } + + /* if didn't display anything, display 0x0 */ + if (!ret) { + res = hwloc_snprintf(tmp, size, "0x0"); + if (res < 0) + return -1; + ret += res; + } + + return ret; +} + +int hwloc_bitmap_taskset_asprintf(char ** strp, const struct hwloc_bitmap_s * __hwloc_restrict set) +{ + int len; + char *buf; + + HWLOC__BITMAP_CHECK(set); + + len = hwloc_bitmap_taskset_snprintf(NULL, 0, set); + buf = malloc(len+1); + if (!buf) + return -1; + *strp = buf; + return hwloc_bitmap_taskset_snprintf(buf, len+1, set); +} + +int hwloc_bitmap_taskset_sscanf(struct hwloc_bitmap_s *set, const char * __hwloc_restrict string) +{ + const char * current = string; + int chars; + int count; + int infinite = 0; + + if (!strncmp("0xf...f", current, 7)) { + /* infinite bitmap */ + infinite = 1; + current += 7; + if (*current == '\0') { + /* special case for infinite/full bitmap */ + hwloc_bitmap_fill(set); + return 0; + } + } else { + /* finite bitmap */ + if (!strncmp("0x", current, 2)) + current += 2; + if (*current == '\0') { + /* special case for empty bitmap */ + hwloc_bitmap_zero(set); + return 0; + } + } + /* we know there are other characters now */ + + chars = (int)strlen(current); + count = (chars * 4 + HWLOC_BITS_PER_LONG - 1) / HWLOC_BITS_PER_LONG; + + if (hwloc_bitmap_reset_by_ulongs(set, count) < 0) + return -1; + set->infinite = 0; + + while (*current != '\0') { + int tmpchars; + char ustr[17]; + unsigned long val; + char *next; + + tmpchars = chars % (HWLOC_BITS_PER_LONG/4); + if (!tmpchars) + tmpchars = (HWLOC_BITS_PER_LONG/4); + + memcpy(ustr, current, tmpchars); + ustr[tmpchars] = '\0'; + val = strtoul(ustr, &next, 16); + if (*next != '\0') + goto failed; + + set->ulongs[count-1] = val; + + current += tmpchars; + chars -= tmpchars; + count--; + } + + set->infinite = infinite; /* set at the end, to avoid spurious realloc with filled new ulongs */ + + return 0; + + failed: + /* failure to parse */ + hwloc_bitmap_zero(set); + return -1; +} + +static void hwloc_bitmap__zero(struct hwloc_bitmap_s *set) +{ + unsigned i; + for(i=0; iulongs_count; i++) + set->ulongs[i] = HWLOC_SUBBITMAP_ZERO; + set->infinite = 0; +} + +void hwloc_bitmap_zero(struct hwloc_bitmap_s * set) +{ + HWLOC__BITMAP_CHECK(set); + + HWLOC_BUILD_ASSERT(HWLOC_BITMAP_PREALLOC_ULONGS >= 1); + if (hwloc_bitmap_reset_by_ulongs(set, 1) < 0) { + /* cannot fail since we preallocate some ulongs. + * if we ever preallocate nothing, we'll reset to 0 ulongs. + */ + } + hwloc_bitmap__zero(set); +} + +static void hwloc_bitmap__fill(struct hwloc_bitmap_s * set) +{ + unsigned i; + for(i=0; iulongs_count; i++) + set->ulongs[i] = HWLOC_SUBBITMAP_FULL; + set->infinite = 1; +} + +void hwloc_bitmap_fill(struct hwloc_bitmap_s * set) +{ + HWLOC__BITMAP_CHECK(set); + + HWLOC_BUILD_ASSERT(HWLOC_BITMAP_PREALLOC_ULONGS >= 1); + if (hwloc_bitmap_reset_by_ulongs(set, 1) < 0) { + /* cannot fail since we pre-allocate some ulongs. + * if we ever pre-allocate nothing, we'll reset to 0 ulongs. + */ + } + hwloc_bitmap__fill(set); +} + +int hwloc_bitmap_from_ulong(struct hwloc_bitmap_s *set, unsigned long mask) +{ + HWLOC__BITMAP_CHECK(set); + + HWLOC_BUILD_ASSERT(HWLOC_BITMAP_PREALLOC_ULONGS >= 1); + if (hwloc_bitmap_reset_by_ulongs(set, 1) < 0) { + /* cannot fail since we pre-allocate some ulongs. + * if ever pre-allocate nothing, we may have to return a failure. + */ + } + set->ulongs[0] = mask; /* there's always at least one ulong allocated */ + set->infinite = 0; + return 0; +} + +int hwloc_bitmap_from_ith_ulong(struct hwloc_bitmap_s *set, unsigned i, unsigned long mask) +{ + unsigned j; + + HWLOC__BITMAP_CHECK(set); + + if (hwloc_bitmap_reset_by_ulongs(set, i+1) < 0) + return -1; + + set->ulongs[i] = mask; + for(j=0; julongs[j] = HWLOC_SUBBITMAP_ZERO; + set->infinite = 0; + return 0; +} + +unsigned long hwloc_bitmap_to_ulong(const struct hwloc_bitmap_s *set) +{ + HWLOC__BITMAP_CHECK(set); + + return set->ulongs[0]; /* there's always at least one ulong allocated */ +} + +unsigned long hwloc_bitmap_to_ith_ulong(const struct hwloc_bitmap_s *set, unsigned i) +{ + HWLOC__BITMAP_CHECK(set); + + return HWLOC_SUBBITMAP_READULONG(set, i); +} + +int hwloc_bitmap_only(struct hwloc_bitmap_s * set, unsigned cpu) +{ + unsigned index_ = HWLOC_SUBBITMAP_INDEX(cpu); + + HWLOC__BITMAP_CHECK(set); + + if (hwloc_bitmap_reset_by_cpu_index(set, cpu) < 0) + return -1; + + hwloc_bitmap__zero(set); + set->ulongs[index_] |= HWLOC_SUBBITMAP_CPU(cpu); + return 0; +} + +int hwloc_bitmap_allbut(struct hwloc_bitmap_s * set, unsigned cpu) +{ + unsigned index_ = HWLOC_SUBBITMAP_INDEX(cpu); + + HWLOC__BITMAP_CHECK(set); + + if (hwloc_bitmap_reset_by_cpu_index(set, cpu) < 0) + return -1; + + hwloc_bitmap__fill(set); + set->ulongs[index_] &= ~HWLOC_SUBBITMAP_CPU(cpu); + return 0; +} + +int hwloc_bitmap_set(struct hwloc_bitmap_s * set, unsigned cpu) +{ + unsigned index_ = HWLOC_SUBBITMAP_INDEX(cpu); + + HWLOC__BITMAP_CHECK(set); + + /* nothing to do if setting inside the infinite part of the bitmap */ + if (set->infinite && cpu >= set->ulongs_count * HWLOC_BITS_PER_LONG) + return 0; + + if (hwloc_bitmap_realloc_by_cpu_index(set, cpu) < 0) + return -1; + + set->ulongs[index_] |= HWLOC_SUBBITMAP_CPU(cpu); + return 0; +} + +int hwloc_bitmap_set_range(struct hwloc_bitmap_s * set, unsigned begincpu, int _endcpu) +{ + unsigned i; + unsigned beginset,endset; + unsigned endcpu = (unsigned) _endcpu; + + HWLOC__BITMAP_CHECK(set); + + if (endcpu < begincpu) + return 0; + if (set->infinite && begincpu >= set->ulongs_count * HWLOC_BITS_PER_LONG) + /* setting only in the already-set infinite part, nothing to do */ + return 0; + + if (_endcpu == -1) { + /* infinite range */ + + /* make sure we can play with the ulong that contains begincpu */ + if (hwloc_bitmap_realloc_by_cpu_index(set, begincpu) < 0) + return -1; + + /* update the ulong that contains begincpu */ + beginset = HWLOC_SUBBITMAP_INDEX(begincpu); + set->ulongs[beginset] |= HWLOC_SUBBITMAP_ULBIT_FROM(HWLOC_SUBBITMAP_CPU_ULBIT(begincpu)); + /* set ulongs after begincpu if any already allocated */ + for(i=beginset+1; iulongs_count; i++) + set->ulongs[i] = HWLOC_SUBBITMAP_FULL; + /* mark the infinity as set */ + set->infinite = 1; + } else { + /* finite range */ + + /* ignore the part of the range that overlaps with the already-set infinite part */ + if (set->infinite && endcpu >= set->ulongs_count * HWLOC_BITS_PER_LONG) + endcpu = set->ulongs_count * HWLOC_BITS_PER_LONG - 1; + /* make sure we can play with the ulongs that contain begincpu and endcpu */ + if (hwloc_bitmap_realloc_by_cpu_index(set, endcpu) < 0) + return -1; + + /* update first and last ulongs */ + beginset = HWLOC_SUBBITMAP_INDEX(begincpu); + endset = HWLOC_SUBBITMAP_INDEX(endcpu); + if (beginset == endset) { + set->ulongs[beginset] |= HWLOC_SUBBITMAP_ULBIT_FROMTO(HWLOC_SUBBITMAP_CPU_ULBIT(begincpu), HWLOC_SUBBITMAP_CPU_ULBIT(endcpu)); + } else { + set->ulongs[beginset] |= HWLOC_SUBBITMAP_ULBIT_FROM(HWLOC_SUBBITMAP_CPU_ULBIT(begincpu)); + set->ulongs[endset] |= HWLOC_SUBBITMAP_ULBIT_TO(HWLOC_SUBBITMAP_CPU_ULBIT(endcpu)); + } + /* set ulongs in the middle of the range */ + for(i=beginset+1; iulongs[i] = HWLOC_SUBBITMAP_FULL; + } + + return 0; +} + +int hwloc_bitmap_set_ith_ulong(struct hwloc_bitmap_s *set, unsigned i, unsigned long mask) +{ + HWLOC__BITMAP_CHECK(set); + + if (hwloc_bitmap_realloc_by_ulongs(set, i+1) < 0) + return -1; + + set->ulongs[i] = mask; + return 0; +} + +int hwloc_bitmap_clr(struct hwloc_bitmap_s * set, unsigned cpu) +{ + unsigned index_ = HWLOC_SUBBITMAP_INDEX(cpu); + + HWLOC__BITMAP_CHECK(set); + + /* nothing to do if clearing inside the infinitely-unset part of the bitmap */ + if (!set->infinite && cpu >= set->ulongs_count * HWLOC_BITS_PER_LONG) + return 0; + + if (hwloc_bitmap_realloc_by_cpu_index(set, cpu) < 0) + return -1; + + set->ulongs[index_] &= ~HWLOC_SUBBITMAP_CPU(cpu); + return 0; +} + +int hwloc_bitmap_clr_range(struct hwloc_bitmap_s * set, unsigned begincpu, int _endcpu) +{ + unsigned i; + unsigned beginset,endset; + unsigned endcpu = (unsigned) _endcpu; + + HWLOC__BITMAP_CHECK(set); + + if (endcpu < begincpu) + return 0; + + if (!set->infinite && begincpu >= set->ulongs_count * HWLOC_BITS_PER_LONG) + /* clearing only in the already-unset infinite part, nothing to do */ + return 0; + + if (_endcpu == -1) { + /* infinite range */ + + /* make sure we can play with the ulong that contains begincpu */ + if (hwloc_bitmap_realloc_by_cpu_index(set, begincpu) < 0) + return -1; + + /* update the ulong that contains begincpu */ + beginset = HWLOC_SUBBITMAP_INDEX(begincpu); + set->ulongs[beginset] &= ~HWLOC_SUBBITMAP_ULBIT_FROM(HWLOC_SUBBITMAP_CPU_ULBIT(begincpu)); + /* clear ulong after begincpu if any already allocated */ + for(i=beginset+1; iulongs_count; i++) + set->ulongs[i] = HWLOC_SUBBITMAP_ZERO; + /* mark the infinity as unset */ + set->infinite = 0; + } else { + /* finite range */ + + /* ignore the part of the range that overlaps with the already-unset infinite part */ + if (!set->infinite && endcpu >= set->ulongs_count * HWLOC_BITS_PER_LONG) + endcpu = set->ulongs_count * HWLOC_BITS_PER_LONG - 1; + /* make sure we can play with the ulongs that contain begincpu and endcpu */ + if (hwloc_bitmap_realloc_by_cpu_index(set, endcpu) < 0) + return -1; + + /* update first and last ulongs */ + beginset = HWLOC_SUBBITMAP_INDEX(begincpu); + endset = HWLOC_SUBBITMAP_INDEX(endcpu); + if (beginset == endset) { + set->ulongs[beginset] &= ~HWLOC_SUBBITMAP_ULBIT_FROMTO(HWLOC_SUBBITMAP_CPU_ULBIT(begincpu), HWLOC_SUBBITMAP_CPU_ULBIT(endcpu)); + } else { + set->ulongs[beginset] &= ~HWLOC_SUBBITMAP_ULBIT_FROM(HWLOC_SUBBITMAP_CPU_ULBIT(begincpu)); + set->ulongs[endset] &= ~HWLOC_SUBBITMAP_ULBIT_TO(HWLOC_SUBBITMAP_CPU_ULBIT(endcpu)); + } + /* clear ulongs in the middle of the range */ + for(i=beginset+1; iulongs[i] = HWLOC_SUBBITMAP_ZERO; + } + + return 0; +} + +int hwloc_bitmap_isset(const struct hwloc_bitmap_s * set, unsigned cpu) +{ + unsigned index_ = HWLOC_SUBBITMAP_INDEX(cpu); + + HWLOC__BITMAP_CHECK(set); + + return (HWLOC_SUBBITMAP_READULONG(set, index_) & HWLOC_SUBBITMAP_CPU(cpu)) != 0; +} + +int hwloc_bitmap_iszero(const struct hwloc_bitmap_s *set) +{ + unsigned i; + + HWLOC__BITMAP_CHECK(set); + + if (set->infinite) + return 0; + for(i=0; iulongs_count; i++) + if (set->ulongs[i] != HWLOC_SUBBITMAP_ZERO) + return 0; + return 1; +} + +int hwloc_bitmap_isfull(const struct hwloc_bitmap_s *set) +{ + unsigned i; + + HWLOC__BITMAP_CHECK(set); + + if (!set->infinite) + return 0; + for(i=0; iulongs_count; i++) + if (set->ulongs[i] != HWLOC_SUBBITMAP_FULL) + return 0; + return 1; +} + +int hwloc_bitmap_isequal (const struct hwloc_bitmap_s *set1, const struct hwloc_bitmap_s *set2) +{ + unsigned count1 = set1->ulongs_count; + unsigned count2 = set2->ulongs_count; + unsigned min_count = count1 < count2 ? count1 : count2; + unsigned i; + + HWLOC__BITMAP_CHECK(set1); + HWLOC__BITMAP_CHECK(set2); + + for(i=0; iulongs[i] != set2->ulongs[i]) + return 0; + + if (count1 != count2) { + unsigned long w1 = set1->infinite ? HWLOC_SUBBITMAP_FULL : HWLOC_SUBBITMAP_ZERO; + unsigned long w2 = set2->infinite ? HWLOC_SUBBITMAP_FULL : HWLOC_SUBBITMAP_ZERO; + for(i=min_count; iulongs[i] != w2) + return 0; + } + for(i=min_count; iulongs[i] != w1) + return 0; + } + } + + if (set1->infinite != set2->infinite) + return 0; + + return 1; +} + +int hwloc_bitmap_intersects (const struct hwloc_bitmap_s *set1, const struct hwloc_bitmap_s *set2) +{ + unsigned count1 = set1->ulongs_count; + unsigned count2 = set2->ulongs_count; + unsigned min_count = count1 < count2 ? count1 : count2; + unsigned i; + + HWLOC__BITMAP_CHECK(set1); + HWLOC__BITMAP_CHECK(set2); + + for(i=0; iulongs[i] & set2->ulongs[i]) + return 1; + + if (count1 != count2) { + if (set2->infinite) { + for(i=min_count; iulongs_count; i++) + if (set1->ulongs[i]) + return 1; + } + if (set1->infinite) { + for(i=min_count; iulongs_count; i++) + if (set2->ulongs[i]) + return 1; + } + } + + if (set1->infinite && set2->infinite) + return 1; + + return 0; +} + +int hwloc_bitmap_isincluded (const struct hwloc_bitmap_s *sub_set, const struct hwloc_bitmap_s *super_set) +{ + unsigned super_count = super_set->ulongs_count; + unsigned sub_count = sub_set->ulongs_count; + unsigned min_count = super_count < sub_count ? super_count : sub_count; + unsigned i; + + HWLOC__BITMAP_CHECK(sub_set); + HWLOC__BITMAP_CHECK(super_set); + + for(i=0; iulongs[i] != (super_set->ulongs[i] | sub_set->ulongs[i])) + return 0; + + if (super_count != sub_count) { + if (!super_set->infinite) + for(i=min_count; iulongs[i]) + return 0; + if (sub_set->infinite) + for(i=min_count; iulongs[i] != HWLOC_SUBBITMAP_FULL) + return 0; + } + + if (sub_set->infinite && !super_set->infinite) + return 0; + + return 1; +} + +int hwloc_bitmap_or (struct hwloc_bitmap_s *res, const struct hwloc_bitmap_s *set1, const struct hwloc_bitmap_s *set2) +{ + /* cache counts so that we can reset res even if it's also set1 or set2 */ + unsigned count1 = set1->ulongs_count; + unsigned count2 = set2->ulongs_count; + unsigned max_count = count1 > count2 ? count1 : count2; + unsigned min_count = count1 + count2 - max_count; + unsigned i; + + HWLOC__BITMAP_CHECK(res); + HWLOC__BITMAP_CHECK(set1); + HWLOC__BITMAP_CHECK(set2); + + if (hwloc_bitmap_reset_by_ulongs(res, max_count) < 0) + return -1; + + for(i=0; iulongs[i] = set1->ulongs[i] | set2->ulongs[i]; + + if (count1 != count2) { + if (min_count < count1) { + if (set2->infinite) { + res->ulongs_count = min_count; + } else { + for(i=min_count; iulongs[i] = set1->ulongs[i]; + } + } else { + if (set1->infinite) { + res->ulongs_count = min_count; + } else { + for(i=min_count; iulongs[i] = set2->ulongs[i]; + } + } + } + + res->infinite = set1->infinite || set2->infinite; + return 0; +} + +int hwloc_bitmap_and (struct hwloc_bitmap_s *res, const struct hwloc_bitmap_s *set1, const struct hwloc_bitmap_s *set2) +{ + /* cache counts so that we can reset res even if it's also set1 or set2 */ + unsigned count1 = set1->ulongs_count; + unsigned count2 = set2->ulongs_count; + unsigned max_count = count1 > count2 ? count1 : count2; + unsigned min_count = count1 + count2 - max_count; + unsigned i; + + HWLOC__BITMAP_CHECK(res); + HWLOC__BITMAP_CHECK(set1); + HWLOC__BITMAP_CHECK(set2); + + if (hwloc_bitmap_reset_by_ulongs(res, max_count) < 0) + return -1; + + for(i=0; iulongs[i] = set1->ulongs[i] & set2->ulongs[i]; + + if (count1 != count2) { + if (min_count < count1) { + if (set2->infinite) { + for(i=min_count; iulongs[i] = set1->ulongs[i]; + } else { + res->ulongs_count = min_count; + } + } else { + if (set1->infinite) { + for(i=min_count; iulongs[i] = set2->ulongs[i]; + } else { + res->ulongs_count = min_count; + } + } + } + + res->infinite = set1->infinite && set2->infinite; + return 0; +} + +int hwloc_bitmap_andnot (struct hwloc_bitmap_s *res, const struct hwloc_bitmap_s *set1, const struct hwloc_bitmap_s *set2) +{ + /* cache counts so that we can reset res even if it's also set1 or set2 */ + unsigned count1 = set1->ulongs_count; + unsigned count2 = set2->ulongs_count; + unsigned max_count = count1 > count2 ? count1 : count2; + unsigned min_count = count1 + count2 - max_count; + unsigned i; + + HWLOC__BITMAP_CHECK(res); + HWLOC__BITMAP_CHECK(set1); + HWLOC__BITMAP_CHECK(set2); + + if (hwloc_bitmap_reset_by_ulongs(res, max_count) < 0) + return -1; + + for(i=0; iulongs[i] = set1->ulongs[i] & ~set2->ulongs[i]; + + if (count1 != count2) { + if (min_count < count1) { + if (!set2->infinite) { + for(i=min_count; iulongs[i] = set1->ulongs[i]; + } else { + res->ulongs_count = min_count; + } + } else { + if (set1->infinite) { + for(i=min_count; iulongs[i] = ~set2->ulongs[i]; + } else { + res->ulongs_count = min_count; + } + } + } + + res->infinite = set1->infinite && !set2->infinite; + return 0; +} + +int hwloc_bitmap_xor (struct hwloc_bitmap_s *res, const struct hwloc_bitmap_s *set1, const struct hwloc_bitmap_s *set2) +{ + /* cache counts so that we can reset res even if it's also set1 or set2 */ + unsigned count1 = set1->ulongs_count; + unsigned count2 = set2->ulongs_count; + unsigned max_count = count1 > count2 ? count1 : count2; + unsigned min_count = count1 + count2 - max_count; + unsigned i; + + HWLOC__BITMAP_CHECK(res); + HWLOC__BITMAP_CHECK(set1); + HWLOC__BITMAP_CHECK(set2); + + if (hwloc_bitmap_reset_by_ulongs(res, max_count) < 0) + return -1; + + for(i=0; iulongs[i] = set1->ulongs[i] ^ set2->ulongs[i]; + + if (count1 != count2) { + if (min_count < count1) { + unsigned long w2 = set2->infinite ? HWLOC_SUBBITMAP_FULL : HWLOC_SUBBITMAP_ZERO; + for(i=min_count; iulongs[i] = set1->ulongs[i] ^ w2; + } else { + unsigned long w1 = set1->infinite ? HWLOC_SUBBITMAP_FULL : HWLOC_SUBBITMAP_ZERO; + for(i=min_count; iulongs[i] = set2->ulongs[i] ^ w1; + } + } + + res->infinite = (!set1->infinite) != (!set2->infinite); + return 0; +} + +int hwloc_bitmap_not (struct hwloc_bitmap_s *res, const struct hwloc_bitmap_s *set) +{ + unsigned count = set->ulongs_count; + unsigned i; + + HWLOC__BITMAP_CHECK(res); + HWLOC__BITMAP_CHECK(set); + + if (hwloc_bitmap_reset_by_ulongs(res, count) < 0) + return -1; + + for(i=0; iulongs[i] = ~set->ulongs[i]; + + res->infinite = !set->infinite; + return 0; +} + +int hwloc_bitmap_first(const struct hwloc_bitmap_s * set) +{ + unsigned i; + + HWLOC__BITMAP_CHECK(set); + + for(i=0; iulongs_count; i++) { + /* subsets are unsigned longs, use ffsl */ + unsigned long w = set->ulongs[i]; + if (w) + return hwloc_ffsl(w) - 1 + HWLOC_BITS_PER_LONG*i; + } + + if (set->infinite) + return set->ulongs_count * HWLOC_BITS_PER_LONG; + + return -1; +} + +int hwloc_bitmap_first_unset(const struct hwloc_bitmap_s * set) +{ + unsigned i; + + HWLOC__BITMAP_CHECK(set); + + for(i=0; iulongs_count; i++) { + /* subsets are unsigned longs, use ffsl */ + unsigned long w = ~set->ulongs[i]; + if (w) + return hwloc_ffsl(w) - 1 + HWLOC_BITS_PER_LONG*i; + } + + if (!set->infinite) + return set->ulongs_count * HWLOC_BITS_PER_LONG; + + return -1; +} + +int hwloc_bitmap_last(const struct hwloc_bitmap_s * set) +{ + int i; + + HWLOC__BITMAP_CHECK(set); + + if (set->infinite) + return -1; + + for(i=(int)set->ulongs_count-1; i>=0; i--) { + /* subsets are unsigned longs, use flsl */ + unsigned long w = set->ulongs[i]; + if (w) + return hwloc_flsl(w) - 1 + HWLOC_BITS_PER_LONG*i; + } + + return -1; +} + +int hwloc_bitmap_last_unset(const struct hwloc_bitmap_s * set) +{ + int i; + + HWLOC__BITMAP_CHECK(set); + + if (!set->infinite) + return -1; + + for(i=(int)set->ulongs_count-1; i>=0; i--) { + /* subsets are unsigned longs, use flsl */ + unsigned long w = ~set->ulongs[i]; + if (w) + return hwloc_flsl(w) - 1 + HWLOC_BITS_PER_LONG*i; + } + + return -1; +} + +int hwloc_bitmap_next(const struct hwloc_bitmap_s * set, int prev_cpu) +{ + unsigned i = HWLOC_SUBBITMAP_INDEX(prev_cpu + 1); + + HWLOC__BITMAP_CHECK(set); + + if (i >= set->ulongs_count) { + if (set->infinite) + return prev_cpu + 1; + else + return -1; + } + + for(; iulongs_count; i++) { + /* subsets are unsigned longs, use ffsl */ + unsigned long w = set->ulongs[i]; + + /* if the prev cpu is in the same word as the possible next one, + we need to mask out previous cpus */ + if (prev_cpu >= 0 && HWLOC_SUBBITMAP_INDEX((unsigned) prev_cpu) == i) + w &= ~HWLOC_SUBBITMAP_ULBIT_TO(HWLOC_SUBBITMAP_CPU_ULBIT(prev_cpu)); + + if (w) + return hwloc_ffsl(w) - 1 + HWLOC_BITS_PER_LONG*i; + } + + if (set->infinite) + return set->ulongs_count * HWLOC_BITS_PER_LONG; + + return -1; +} + +int hwloc_bitmap_next_unset(const struct hwloc_bitmap_s * set, int prev_cpu) +{ + unsigned i = HWLOC_SUBBITMAP_INDEX(prev_cpu + 1); + + HWLOC__BITMAP_CHECK(set); + + if (i >= set->ulongs_count) { + if (!set->infinite) + return prev_cpu + 1; + else + return -1; + } + + for(; iulongs_count; i++) { + /* subsets are unsigned longs, use ffsl */ + unsigned long w = ~set->ulongs[i]; + + /* if the prev cpu is in the same word as the possible next one, + we need to mask out previous cpus */ + if (prev_cpu >= 0 && HWLOC_SUBBITMAP_INDEX((unsigned) prev_cpu) == i) + w &= ~HWLOC_SUBBITMAP_ULBIT_TO(HWLOC_SUBBITMAP_CPU_ULBIT(prev_cpu)); + + if (w) + return hwloc_ffsl(w) - 1 + HWLOC_BITS_PER_LONG*i; + } + + if (!set->infinite) + return set->ulongs_count * HWLOC_BITS_PER_LONG; + + return -1; +} + +int hwloc_bitmap_singlify(struct hwloc_bitmap_s * set) +{ + unsigned i; + int found = 0; + + HWLOC__BITMAP_CHECK(set); + + for(i=0; iulongs_count; i++) { + if (found) { + set->ulongs[i] = HWLOC_SUBBITMAP_ZERO; + continue; + } else { + /* subsets are unsigned longs, use ffsl */ + unsigned long w = set->ulongs[i]; + if (w) { + int _ffs = hwloc_ffsl(w); + set->ulongs[i] = HWLOC_SUBBITMAP_CPU(_ffs-1); + found = 1; + } + } + } + + if (set->infinite) { + if (found) { + set->infinite = 0; + } else { + /* set the first non allocated bit */ + unsigned first = set->ulongs_count * HWLOC_BITS_PER_LONG; + set->infinite = 0; /* do not let realloc fill the newly allocated sets */ + return hwloc_bitmap_set(set, first); + } + } + + return 0; +} + +int hwloc_bitmap_compare_first(const struct hwloc_bitmap_s * set1, const struct hwloc_bitmap_s * set2) +{ + unsigned count1 = set1->ulongs_count; + unsigned count2 = set2->ulongs_count; + unsigned max_count = count1 > count2 ? count1 : count2; + unsigned min_count = count1 + count2 - max_count; + unsigned i; + + HWLOC__BITMAP_CHECK(set1); + HWLOC__BITMAP_CHECK(set2); + + for(i=0; iulongs[i]; + unsigned long w2 = set2->ulongs[i]; + if (w1 || w2) { + int _ffs1 = hwloc_ffsl(w1); + int _ffs2 = hwloc_ffsl(w2); + /* if both have a bit set, compare for real */ + if (_ffs1 && _ffs2) + return _ffs1-_ffs2; + /* one is empty, and it is considered higher, so reverse-compare them */ + return _ffs2-_ffs1; + } + } + + if (count1 != count2) { + if (min_count < count2) { + for(i=min_count; iulongs[i]; + if (set1->infinite) + return -!(w2 & 1); + else if (w2) + return 1; + } + } else { + for(i=min_count; iulongs[i]; + if (set2->infinite) + return !(w1 & 1); + else if (w1) + return -1; + } + } + } + + return !!set1->infinite - !!set2->infinite; +} + +int hwloc_bitmap_compare(const struct hwloc_bitmap_s * set1, const struct hwloc_bitmap_s * set2) +{ + unsigned count1 = set1->ulongs_count; + unsigned count2 = set2->ulongs_count; + unsigned max_count = count1 > count2 ? count1 : count2; + unsigned min_count = count1 + count2 - max_count; + int i; + + HWLOC__BITMAP_CHECK(set1); + HWLOC__BITMAP_CHECK(set2); + + if ((!set1->infinite) != (!set2->infinite)) + return !!set1->infinite - !!set2->infinite; + + if (count1 != count2) { + if (min_count < count2) { + unsigned long val1 = set1->infinite ? HWLOC_SUBBITMAP_FULL : HWLOC_SUBBITMAP_ZERO; + for(i=(int)max_count-1; i>=(int) min_count; i--) { + unsigned long val2 = set2->ulongs[i]; + if (val1 == val2) + continue; + return val1 < val2 ? -1 : 1; + } + } else { + unsigned long val2 = set2->infinite ? HWLOC_SUBBITMAP_FULL : HWLOC_SUBBITMAP_ZERO; + for(i=(int)max_count-1; i>=(int) min_count; i--) { + unsigned long val1 = set1->ulongs[i]; + if (val1 == val2) + continue; + return val1 < val2 ? -1 : 1; + } + } + } + + for(i=(int)min_count-1; i>=0; i--) { + unsigned long val1 = set1->ulongs[i]; + unsigned long val2 = set2->ulongs[i]; + if (val1 == val2) + continue; + return val1 < val2 ? -1 : 1; + } + + return 0; +} + +int hwloc_bitmap_weight(const struct hwloc_bitmap_s * set) +{ + int weight = 0; + unsigned i; + + HWLOC__BITMAP_CHECK(set); + + if (set->infinite) + return -1; + + for(i=0; iulongs_count; i++) + weight += hwloc_weight_long(set->ulongs[i]); + return weight; +} + +int hwloc_bitmap_compare_inclusion(const struct hwloc_bitmap_s * set1, const struct hwloc_bitmap_s * set2) +{ + unsigned max_count = set1->ulongs_count > set2->ulongs_count ? set1->ulongs_count : set2->ulongs_count; + int result = HWLOC_BITMAP_EQUAL; /* means empty sets return equal */ + int empty1 = 1; + int empty2 = 1; + unsigned i; + + HWLOC__BITMAP_CHECK(set1); + HWLOC__BITMAP_CHECK(set2); + + for(i=0; iinfinite) { + if (set2->infinite) { + /* set2 infinite only */ + if (result == HWLOC_BITMAP_CONTAINS) { + if (!empty2) + return HWLOC_BITMAP_INTERSECTS; + result = HWLOC_BITMAP_DIFFERENT; + } else if (result == HWLOC_BITMAP_EQUAL) { + result = HWLOC_BITMAP_INCLUDED; + } + /* no change otherwise */ + } + } else if (!set2->infinite) { + /* set1 infinite only */ + if (result == HWLOC_BITMAP_INCLUDED) { + if (!empty1) + return HWLOC_BITMAP_INTERSECTS; + result = HWLOC_BITMAP_DIFFERENT; + } else if (result == HWLOC_BITMAP_EQUAL) { + result = HWLOC_BITMAP_CONTAINS; + } + /* no change otherwise */ + } else { + /* both infinite */ + if (result == HWLOC_BITMAP_DIFFERENT) + return HWLOC_BITMAP_INTERSECTS; + /* equal/contains/included unchanged */ + } + + return result; +} diff --git a/src/3rdparty/hwloc/src/components.c b/src/3rdparty/hwloc/src/components.c new file mode 100644 index 000000000..bd7c00e36 --- /dev/null +++ b/src/3rdparty/hwloc/src/components.c @@ -0,0 +1,785 @@ +/* + * Copyright © 2009-2017 Inria. All rights reserved. + * Copyright © 2012 Université Bordeaux + * See COPYING in top-level directory. + */ + +#include +#include +#include +#include +#include + +#define HWLOC_COMPONENT_STOP_NAME "stop" +#define HWLOC_COMPONENT_EXCLUDE_CHAR '-' +#define HWLOC_COMPONENT_SEPS "," + +/* list of all registered discovery components, sorted by priority, higher priority first. + * noos is last because its priority is 0. + * others' priority is 10. + */ +static struct hwloc_disc_component * hwloc_disc_components = NULL; + +static unsigned hwloc_components_users = 0; /* first one initializes, last ones destroys */ + +static int hwloc_components_verbose = 0; +#ifdef HWLOC_HAVE_PLUGINS +static int hwloc_plugins_verbose = 0; +static const char * hwloc_plugins_blacklist = NULL; +#endif + +/* hwloc_components_mutex serializes: + * - loading/unloading plugins, and modifications of the hwloc_plugins list + * - calls to ltdl, including in hwloc_check_plugin_namespace() + * - registration of components with hwloc_disc_component_register() + * and hwloc_xml_callbacks_register() + */ +#ifdef HWLOC_WIN_SYS +/* Basic mutex on top of InterlockedCompareExchange() on windows, + * Far from perfect, but easy to maintain, and way enough given that this code will never be needed for real. */ +#include +static LONG hwloc_components_mutex = 0; +#define HWLOC_COMPONENTS_LOCK() do { \ + while (InterlockedCompareExchange(&hwloc_components_mutex, 1, 0) != 0) \ + SwitchToThread(); \ +} while (0) +#define HWLOC_COMPONENTS_UNLOCK() do { \ + assert(hwloc_components_mutex == 1); \ + hwloc_components_mutex = 0; \ +} while (0) + +#elif defined HWLOC_HAVE_PTHREAD_MUTEX +/* pthread mutex if available (except on windows) */ +#include +static pthread_mutex_t hwloc_components_mutex = PTHREAD_MUTEX_INITIALIZER; +#define HWLOC_COMPONENTS_LOCK() pthread_mutex_lock(&hwloc_components_mutex) +#define HWLOC_COMPONENTS_UNLOCK() pthread_mutex_unlock(&hwloc_components_mutex) + +#else /* HWLOC_WIN_SYS || HWLOC_HAVE_PTHREAD_MUTEX */ +#error No mutex implementation available +#endif + + +#ifdef HWLOC_HAVE_PLUGINS + +#include + +/* array of pointers to dynamically loaded plugins */ +static struct hwloc__plugin_desc { + char *name; + struct hwloc_component *component; + char *filename; + lt_dlhandle handle; + struct hwloc__plugin_desc *next; +} *hwloc_plugins = NULL; + +static int +hwloc__dlforeach_cb(const char *filename, void *_data __hwloc_attribute_unused) +{ + const char *basename; + lt_dlhandle handle; + struct hwloc_component *component; + struct hwloc__plugin_desc *desc, **prevdesc; + + if (hwloc_plugins_verbose) + fprintf(stderr, "Plugin dlforeach found `%s'\n", filename); + + basename = strrchr(filename, '/'); + if (!basename) + basename = filename; + else + basename++; + + if (hwloc_plugins_blacklist && strstr(hwloc_plugins_blacklist, basename)) { + if (hwloc_plugins_verbose) + fprintf(stderr, "Plugin `%s' is blacklisted in the environment\n", basename); + goto out; + } + + /* dlopen and get the component structure */ + handle = lt_dlopenext(filename); + if (!handle) { + if (hwloc_plugins_verbose) + fprintf(stderr, "Failed to load plugin: %s\n", lt_dlerror()); + goto out; + } + +{ + char componentsymbolname[strlen(basename)+10+1]; + sprintf(componentsymbolname, "%s_component", basename); + component = lt_dlsym(handle, componentsymbolname); + if (!component) { + if (hwloc_plugins_verbose) + fprintf(stderr, "Failed to find component symbol `%s'\n", + componentsymbolname); + goto out_with_handle; + } + if (component->abi != HWLOC_COMPONENT_ABI) { + if (hwloc_plugins_verbose) + fprintf(stderr, "Plugin symbol ABI %u instead of %d\n", + component->abi, HWLOC_COMPONENT_ABI); + goto out_with_handle; + } + if (hwloc_plugins_verbose) + fprintf(stderr, "Plugin contains expected symbol `%s'\n", + componentsymbolname); +} + + if (HWLOC_COMPONENT_TYPE_DISC == component->type) { + if (strncmp(basename, "hwloc_", 6)) { + if (hwloc_plugins_verbose) + fprintf(stderr, "Plugin name `%s' doesn't match its type DISCOVERY\n", basename); + goto out_with_handle; + } + } else if (HWLOC_COMPONENT_TYPE_XML == component->type) { + if (strncmp(basename, "hwloc_xml_", 10)) { + if (hwloc_plugins_verbose) + fprintf(stderr, "Plugin name `%s' doesn't match its type XML\n", basename); + goto out_with_handle; + } + } else { + if (hwloc_plugins_verbose) + fprintf(stderr, "Plugin name `%s' has invalid type %u\n", + basename, (unsigned) component->type); + goto out_with_handle; + } + + /* allocate a plugin_desc and queue it */ + desc = malloc(sizeof(*desc)); + if (!desc) + goto out_with_handle; + desc->name = strdup(basename); + desc->filename = strdup(filename); + desc->component = component; + desc->handle = handle; + desc->next = NULL; + if (hwloc_plugins_verbose) + fprintf(stderr, "Plugin descriptor `%s' ready\n", basename); + + /* append to the list */ + prevdesc = &hwloc_plugins; + while (*prevdesc) + prevdesc = &((*prevdesc)->next); + *prevdesc = desc; + if (hwloc_plugins_verbose) + fprintf(stderr, "Plugin descriptor `%s' queued\n", basename); + return 0; + + out_with_handle: + lt_dlclose(handle); + out: + return 0; +} + +static void +hwloc_plugins_exit(void) +{ + struct hwloc__plugin_desc *desc, *next; + + if (hwloc_plugins_verbose) + fprintf(stderr, "Closing all plugins\n"); + + desc = hwloc_plugins; + while (desc) { + next = desc->next; + lt_dlclose(desc->handle); + free(desc->name); + free(desc->filename); + free(desc); + desc = next; + } + hwloc_plugins = NULL; + + lt_dlexit(); +} + +static int +hwloc_plugins_init(void) +{ + const char *verboseenv; + const char *path = HWLOC_PLUGINS_PATH; + const char *env; + int err; + + verboseenv = getenv("HWLOC_PLUGINS_VERBOSE"); + hwloc_plugins_verbose = verboseenv ? atoi(verboseenv) : 0; + + hwloc_plugins_blacklist = getenv("HWLOC_PLUGINS_BLACKLIST"); + + err = lt_dlinit(); + if (err) + goto out; + + env = getenv("HWLOC_PLUGINS_PATH"); + if (env) + path = env; + + hwloc_plugins = NULL; + + if (hwloc_plugins_verbose) + fprintf(stderr, "Starting plugin dlforeach in %s\n", path); + err = lt_dlforeachfile(path, hwloc__dlforeach_cb, NULL); + if (err) + goto out_with_init; + + return 0; + + out_with_init: + hwloc_plugins_exit(); + out: + return -1; +} + +#endif /* HWLOC_HAVE_PLUGINS */ + +static const char * +hwloc_disc_component_type_string(hwloc_disc_component_type_t type) +{ + switch (type) { + case HWLOC_DISC_COMPONENT_TYPE_CPU: return "cpu"; + case HWLOC_DISC_COMPONENT_TYPE_GLOBAL: return "global"; + case HWLOC_DISC_COMPONENT_TYPE_MISC: return "misc"; + default: return "**unknown**"; + } +} + +static int +hwloc_disc_component_register(struct hwloc_disc_component *component, + const char *filename) +{ + struct hwloc_disc_component **prev; + + /* check that the component name is valid */ + if (!strcmp(component->name, HWLOC_COMPONENT_STOP_NAME)) { + if (hwloc_components_verbose) + fprintf(stderr, "Cannot register discovery component with reserved name `" HWLOC_COMPONENT_STOP_NAME "'\n"); + return -1; + } + if (strchr(component->name, HWLOC_COMPONENT_EXCLUDE_CHAR) + || strcspn(component->name, HWLOC_COMPONENT_SEPS) != strlen(component->name)) { + if (hwloc_components_verbose) + fprintf(stderr, "Cannot register discovery component with name `%s' containing reserved characters `%c" HWLOC_COMPONENT_SEPS "'\n", + component->name, HWLOC_COMPONENT_EXCLUDE_CHAR); + return -1; + } + /* check that the component type is valid */ + switch ((unsigned) component->type) { + case HWLOC_DISC_COMPONENT_TYPE_CPU: + case HWLOC_DISC_COMPONENT_TYPE_GLOBAL: + case HWLOC_DISC_COMPONENT_TYPE_MISC: + break; + default: + fprintf(stderr, "Cannot register discovery component `%s' with unknown type %u\n", + component->name, (unsigned) component->type); + return -1; + } + + prev = &hwloc_disc_components; + while (NULL != *prev) { + if (!strcmp((*prev)->name, component->name)) { + /* if two components have the same name, only keep the highest priority one */ + if ((*prev)->priority < component->priority) { + /* drop the existing component */ + if (hwloc_components_verbose) + fprintf(stderr, "Dropping previously registered discovery component `%s', priority %u lower than new one %u\n", + (*prev)->name, (*prev)->priority, component->priority); + *prev = (*prev)->next; + } else { + /* drop the new one */ + if (hwloc_components_verbose) + fprintf(stderr, "Ignoring new discovery component `%s', priority %u lower than previously registered one %u\n", + component->name, component->priority, (*prev)->priority); + return -1; + } + } + prev = &((*prev)->next); + } + if (hwloc_components_verbose) + fprintf(stderr, "Registered %s discovery component `%s' with priority %u (%s%s)\n", + hwloc_disc_component_type_string(component->type), component->name, component->priority, + filename ? "from plugin " : "statically build", filename ? filename : ""); + + prev = &hwloc_disc_components; + while (NULL != *prev) { + if ((*prev)->priority < component->priority) + break; + prev = &((*prev)->next); + } + component->next = *prev; + *prev = component; + return 0; +} + +#include + +static void (**hwloc_component_finalize_cbs)(unsigned long); +static unsigned hwloc_component_finalize_cb_count; + +void +hwloc_components_init(void) +{ +#ifdef HWLOC_HAVE_PLUGINS + struct hwloc__plugin_desc *desc; +#endif + const char *verboseenv; + unsigned i; + + HWLOC_COMPONENTS_LOCK(); + assert((unsigned) -1 != hwloc_components_users); + if (0 != hwloc_components_users++) { + HWLOC_COMPONENTS_UNLOCK(); + return; + } + + verboseenv = getenv("HWLOC_COMPONENTS_VERBOSE"); + hwloc_components_verbose = verboseenv ? atoi(verboseenv) : 0; + +#ifdef HWLOC_HAVE_PLUGINS + hwloc_plugins_init(); +#endif + + hwloc_component_finalize_cbs = NULL; + hwloc_component_finalize_cb_count = 0; + /* count the max number of finalize callbacks */ + for(i=0; NULL != hwloc_static_components[i]; i++) + hwloc_component_finalize_cb_count++; +#ifdef HWLOC_HAVE_PLUGINS + for(desc = hwloc_plugins; NULL != desc; desc = desc->next) + hwloc_component_finalize_cb_count++; +#endif + if (hwloc_component_finalize_cb_count) { + hwloc_component_finalize_cbs = calloc(hwloc_component_finalize_cb_count, + sizeof(*hwloc_component_finalize_cbs)); + assert(hwloc_component_finalize_cbs); + /* forget that max number and recompute the real one below */ + hwloc_component_finalize_cb_count = 0; + } + + /* hwloc_static_components is created by configure in static-components.h */ + for(i=0; NULL != hwloc_static_components[i]; i++) { + if (hwloc_static_components[i]->flags) { + fprintf(stderr, "Ignoring static component with invalid flags %lx\n", + hwloc_static_components[i]->flags); + continue; + } + + /* initialize the component */ + if (hwloc_static_components[i]->init && hwloc_static_components[i]->init(0) < 0) { + if (hwloc_components_verbose) + fprintf(stderr, "Ignoring static component, failed to initialize\n"); + continue; + } + /* queue ->finalize() callback if any */ + if (hwloc_static_components[i]->finalize) + hwloc_component_finalize_cbs[hwloc_component_finalize_cb_count++] = hwloc_static_components[i]->finalize; + + /* register for real now */ + if (HWLOC_COMPONENT_TYPE_DISC == hwloc_static_components[i]->type) + hwloc_disc_component_register(hwloc_static_components[i]->data, NULL); + else if (HWLOC_COMPONENT_TYPE_XML == hwloc_static_components[i]->type) + hwloc_xml_callbacks_register(hwloc_static_components[i]->data); + else + assert(0); + } + + /* dynamic plugins */ +#ifdef HWLOC_HAVE_PLUGINS + for(desc = hwloc_plugins; NULL != desc; desc = desc->next) { + if (desc->component->flags) { + fprintf(stderr, "Ignoring plugin `%s' component with invalid flags %lx\n", + desc->name, desc->component->flags); + continue; + } + + /* initialize the component */ + if (desc->component->init && desc->component->init(0) < 0) { + if (hwloc_components_verbose) + fprintf(stderr, "Ignoring plugin `%s', failed to initialize\n", desc->name); + continue; + } + /* queue ->finalize() callback if any */ + if (desc->component->finalize) + hwloc_component_finalize_cbs[hwloc_component_finalize_cb_count++] = desc->component->finalize; + + /* register for real now */ + if (HWLOC_COMPONENT_TYPE_DISC == desc->component->type) + hwloc_disc_component_register(desc->component->data, desc->filename); + else if (HWLOC_COMPONENT_TYPE_XML == desc->component->type) + hwloc_xml_callbacks_register(desc->component->data); + else + assert(0); + } +#endif + + HWLOC_COMPONENTS_UNLOCK(); +} + +void +hwloc_backends_init(struct hwloc_topology *topology) +{ + topology->backends = NULL; + topology->backend_excludes = 0; +} + +static struct hwloc_disc_component * +hwloc_disc_component_find(int type /* hwloc_disc_component_type_t or -1 if any */, + const char *name /* name of NULL if any */) +{ + struct hwloc_disc_component *comp = hwloc_disc_components; + while (NULL != comp) { + if ((-1 == type || type == (int) comp->type) + && (NULL == name || !strcmp(name, comp->name))) + return comp; + comp = comp->next; + } + return NULL; +} + +/* used by set_xml(), set_synthetic(), ... environment variables, ... to force the first backend */ +int +hwloc_disc_component_force_enable(struct hwloc_topology *topology, + int envvar_forced, + int type, const char *name, + const void *data1, const void *data2, const void *data3) +{ + struct hwloc_disc_component *comp; + struct hwloc_backend *backend; + + if (topology->is_loaded) { + errno = EBUSY; + return -1; + } + + comp = hwloc_disc_component_find(type, name); + if (!comp) { + errno = ENOSYS; + return -1; + } + + backend = comp->instantiate(comp, data1, data2, data3); + if (backend) { + backend->envvar_forced = envvar_forced; + if (topology->backends) + hwloc_backends_disable_all(topology); + return hwloc_backend_enable(topology, backend); + } else + return -1; +} + +static int +hwloc_disc_component_try_enable(struct hwloc_topology *topology, + struct hwloc_disc_component *comp, + const char *comparg, + int envvar_forced) +{ + struct hwloc_backend *backend; + + if (topology->backend_excludes & comp->type) { + if (hwloc_components_verbose) + /* do not warn if envvar_forced since system-wide HWLOC_COMPONENTS must be silently ignored after set_xml() etc. + */ + fprintf(stderr, "Excluding %s discovery component `%s', conflicts with excludes 0x%x\n", + hwloc_disc_component_type_string(comp->type), comp->name, topology->backend_excludes); + return -1; + } + + backend = comp->instantiate(comp, comparg, NULL, NULL); + if (!backend) { + if (hwloc_components_verbose || envvar_forced) + fprintf(stderr, "Failed to instantiate discovery component `%s'\n", comp->name); + return -1; + } + + backend->envvar_forced = envvar_forced; + return hwloc_backend_enable(topology, backend); +} + +void +hwloc_disc_components_enable_others(struct hwloc_topology *topology) +{ + struct hwloc_disc_component *comp; + struct hwloc_backend *backend; + int tryall = 1; + const char *_env; + char *env; /* we'll to modify the env value, so duplicate it */ + + _env = getenv("HWLOC_COMPONENTS"); + env = _env ? strdup(_env) : NULL; + + /* enable explicitly listed components */ + if (env) { + char *curenv = env; + size_t s; + + while (*curenv) { + s = strcspn(curenv, HWLOC_COMPONENT_SEPS); + if (s) { + char c; + + /* replace linuxpci with linuxio for backward compatibility with pre-v2.0 */ + if (!strncmp(curenv, "linuxpci", 8) && s == 8) { + curenv[5] = 'i'; + curenv[6] = 'o'; + curenv[7] = *HWLOC_COMPONENT_SEPS; + } else if (curenv[0] == HWLOC_COMPONENT_EXCLUDE_CHAR && !strncmp(curenv+1, "linuxpci", 8) && s == 9) { + curenv[6] = 'i'; + curenv[7] = 'o'; + curenv[8] = *HWLOC_COMPONENT_SEPS; + /* skip this name, it's a negated one */ + goto nextname; + } + + if (curenv[0] == HWLOC_COMPONENT_EXCLUDE_CHAR) + goto nextname; + + if (!strncmp(curenv, HWLOC_COMPONENT_STOP_NAME, s)) { + tryall = 0; + break; + } + + /* save the last char and replace with \0 */ + c = curenv[s]; + curenv[s] = '\0'; + + comp = hwloc_disc_component_find(-1, curenv); + if (comp) { + hwloc_disc_component_try_enable(topology, comp, NULL, 1 /* envvar forced */); + } else { + fprintf(stderr, "Cannot find discovery component `%s'\n", curenv); + } + + /* restore chars (the second loop below needs env to be unmodified) */ + curenv[s] = c; + } + +nextname: + curenv += s; + if (*curenv) + /* Skip comma */ + curenv++; + } + } + + /* env is still the same, the above loop didn't modify it */ + + /* now enable remaining components (except the explicitly '-'-listed ones) */ + if (tryall) { + comp = hwloc_disc_components; + while (NULL != comp) { + if (!comp->enabled_by_default) + goto nextcomp; + /* check if this component was explicitly excluded in env */ + if (env) { + char *curenv = env; + while (*curenv) { + size_t s = strcspn(curenv, HWLOC_COMPONENT_SEPS); + if (curenv[0] == HWLOC_COMPONENT_EXCLUDE_CHAR && !strncmp(curenv+1, comp->name, s-1) && strlen(comp->name) == s-1) { + if (hwloc_components_verbose) + fprintf(stderr, "Excluding %s discovery component `%s' because of HWLOC_COMPONENTS environment variable\n", + hwloc_disc_component_type_string(comp->type), comp->name); + goto nextcomp; + } + curenv += s; + if (*curenv) + /* Skip comma */ + curenv++; + } + } + hwloc_disc_component_try_enable(topology, comp, NULL, 0 /* defaults, not envvar forced */); +nextcomp: + comp = comp->next; + } + } + + if (hwloc_components_verbose) { + /* print a summary */ + int first = 1; + backend = topology->backends; + fprintf(stderr, "Final list of enabled discovery components: "); + while (backend != NULL) { + fprintf(stderr, "%s%s", first ? "" : ",", backend->component->name); + backend = backend->next; + first = 0; + } + fprintf(stderr, "\n"); + } + + free(env); +} + +void +hwloc_components_fini(void) +{ + unsigned i; + + HWLOC_COMPONENTS_LOCK(); + assert(0 != hwloc_components_users); + if (0 != --hwloc_components_users) { + HWLOC_COMPONENTS_UNLOCK(); + return; + } + + for(i=0; icomponent = component; + backend->flags = 0; + backend->discover = NULL; + backend->get_pci_busid_cpuset = NULL; + backend->disable = NULL; + backend->is_thissystem = -1; + backend->next = NULL; + backend->envvar_forced = 0; + return backend; +} + +static void +hwloc_backend_disable(struct hwloc_backend *backend) +{ + if (backend->disable) + backend->disable(backend); + free(backend); +} + +int +hwloc_backend_enable(struct hwloc_topology *topology, struct hwloc_backend *backend) +{ + struct hwloc_backend **pprev; + + /* check backend flags */ + if (backend->flags) { + fprintf(stderr, "Cannot enable %s discovery component `%s' with unknown flags %lx\n", + hwloc_disc_component_type_string(backend->component->type), backend->component->name, backend->flags); + return -1; + } + + /* make sure we didn't already enable this backend, we don't want duplicates */ + pprev = &topology->backends; + while (NULL != *pprev) { + if ((*pprev)->component == backend->component) { + if (hwloc_components_verbose) + fprintf(stderr, "Cannot enable %s discovery component `%s' twice\n", + hwloc_disc_component_type_string(backend->component->type), backend->component->name); + hwloc_backend_disable(backend); + errno = EBUSY; + return -1; + } + pprev = &((*pprev)->next); + } + + if (hwloc_components_verbose) + fprintf(stderr, "Enabling %s discovery component `%s'\n", + hwloc_disc_component_type_string(backend->component->type), backend->component->name); + + /* enqueue at the end */ + pprev = &topology->backends; + while (NULL != *pprev) + pprev = &((*pprev)->next); + backend->next = *pprev; + *pprev = backend; + + backend->topology = topology; + topology->backend_excludes |= backend->component->excludes; + return 0; +} + +void +hwloc_backends_is_thissystem(struct hwloc_topology *topology) +{ + struct hwloc_backend *backend; + const char *local_env; + + /* Apply is_thissystem topology flag before we enforce envvar backends. + * If the application changed the backend with set_foo(), + * it may use set_flags() update the is_thissystem flag here. + * If it changes the backend with environment variables below, + * it may use HWLOC_THISSYSTEM envvar below as well. + */ + + topology->is_thissystem = 1; + + /* apply thissystem from normally-given backends (envvar_forced=0, either set_foo() or defaults) */ + backend = topology->backends; + while (backend != NULL) { + if (backend->envvar_forced == 0 && backend->is_thissystem != -1) { + assert(backend->is_thissystem == 0); + topology->is_thissystem = 0; + } + backend = backend->next; + } + + /* override set_foo() with flags */ + if (topology->flags & HWLOC_TOPOLOGY_FLAG_IS_THISSYSTEM) + topology->is_thissystem = 1; + + /* now apply envvar-forced backend (envvar_forced=1) */ + backend = topology->backends; + while (backend != NULL) { + if (backend->envvar_forced == 1 && backend->is_thissystem != -1) { + assert(backend->is_thissystem == 0); + topology->is_thissystem = 0; + } + backend = backend->next; + } + + /* override with envvar-given flag */ + local_env = getenv("HWLOC_THISSYSTEM"); + if (local_env) + topology->is_thissystem = atoi(local_env); +} + +void +hwloc_backends_find_callbacks(struct hwloc_topology *topology) +{ + struct hwloc_backend *backend = topology->backends; + /* use the first backend's get_pci_busid_cpuset callback */ + topology->get_pci_busid_cpuset_backend = NULL; + while (backend != NULL) { + if (backend->get_pci_busid_cpuset) { + topology->get_pci_busid_cpuset_backend = backend; + return; + } + backend = backend->next; + } + return; +} + +void +hwloc_backends_disable_all(struct hwloc_topology *topology) +{ + struct hwloc_backend *backend; + + while (NULL != (backend = topology->backends)) { + struct hwloc_backend *next = backend->next; + if (hwloc_components_verbose) + fprintf(stderr, "Disabling %s discovery component `%s'\n", + hwloc_disc_component_type_string(backend->component->type), backend->component->name); + hwloc_backend_disable(backend); + topology->backends = next; + } + topology->backends = NULL; + topology->backend_excludes = 0; +} diff --git a/src/3rdparty/hwloc/src/diff.c b/src/3rdparty/hwloc/src/diff.c new file mode 100644 index 000000000..00811a7b5 --- /dev/null +++ b/src/3rdparty/hwloc/src/diff.c @@ -0,0 +1,492 @@ +/* + * Copyright © 2013-2018 Inria. All rights reserved. + * See COPYING in top-level directory. + */ + +#include +#include +#include + +int hwloc_topology_diff_destroy(hwloc_topology_diff_t diff) +{ + hwloc_topology_diff_t next; + while (diff) { + next = diff->generic.next; + switch (diff->generic.type) { + default: + break; + case HWLOC_TOPOLOGY_DIFF_OBJ_ATTR: + switch (diff->obj_attr.diff.generic.type) { + default: + break; + case HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_NAME: + case HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_INFO: + free(diff->obj_attr.diff.string.name); + free(diff->obj_attr.diff.string.oldvalue); + free(diff->obj_attr.diff.string.newvalue); + break; + } + break; + } + free(diff); + diff = next; + } + return 0; +} + +/************************ + * Computing diffs + */ + +static void hwloc_append_diff(hwloc_topology_diff_t newdiff, + hwloc_topology_diff_t *firstdiffp, + hwloc_topology_diff_t *lastdiffp) +{ + if (*firstdiffp) + (*lastdiffp)->generic.next = newdiff; + else + *firstdiffp = newdiff; + *lastdiffp = newdiff; + newdiff->generic.next = NULL; +} + +static int hwloc_append_diff_too_complex(hwloc_obj_t obj1, + hwloc_topology_diff_t *firstdiffp, + hwloc_topology_diff_t *lastdiffp) +{ + hwloc_topology_diff_t newdiff; + newdiff = malloc(sizeof(*newdiff)); + if (!newdiff) + return -1; + + newdiff->too_complex.type = HWLOC_TOPOLOGY_DIFF_TOO_COMPLEX; + newdiff->too_complex.obj_depth = obj1->depth; + newdiff->too_complex.obj_index = obj1->logical_index; + hwloc_append_diff(newdiff, firstdiffp, lastdiffp); + return 0; +} + +static int hwloc_append_diff_obj_attr_string(hwloc_obj_t obj, + hwloc_topology_diff_obj_attr_type_t type, + const char *name, + const char *oldvalue, + const char *newvalue, + hwloc_topology_diff_t *firstdiffp, + hwloc_topology_diff_t *lastdiffp) +{ + hwloc_topology_diff_t newdiff; + newdiff = malloc(sizeof(*newdiff)); + if (!newdiff) + return -1; + + newdiff->obj_attr.type = HWLOC_TOPOLOGY_DIFF_OBJ_ATTR; + newdiff->obj_attr.obj_depth = obj->depth; + newdiff->obj_attr.obj_index = obj->logical_index; + newdiff->obj_attr.diff.string.type = type; + newdiff->obj_attr.diff.string.name = name ? strdup(name) : NULL; + newdiff->obj_attr.diff.string.oldvalue = oldvalue ? strdup(oldvalue) : NULL; + newdiff->obj_attr.diff.string.newvalue = newvalue ? strdup(newvalue) : NULL; + hwloc_append_diff(newdiff, firstdiffp, lastdiffp); + return 0; +} + +static int hwloc_append_diff_obj_attr_uint64(hwloc_obj_t obj, + hwloc_topology_diff_obj_attr_type_t type, + hwloc_uint64_t idx, + hwloc_uint64_t oldvalue, + hwloc_uint64_t newvalue, + hwloc_topology_diff_t *firstdiffp, + hwloc_topology_diff_t *lastdiffp) +{ + hwloc_topology_diff_t newdiff; + newdiff = malloc(sizeof(*newdiff)); + if (!newdiff) + return -1; + + newdiff->obj_attr.type = HWLOC_TOPOLOGY_DIFF_OBJ_ATTR; + newdiff->obj_attr.obj_depth = obj->depth; + newdiff->obj_attr.obj_index = obj->logical_index; + newdiff->obj_attr.diff.uint64.type = type; + newdiff->obj_attr.diff.uint64.index = idx; + newdiff->obj_attr.diff.uint64.oldvalue = oldvalue; + newdiff->obj_attr.diff.uint64.newvalue = newvalue; + hwloc_append_diff(newdiff, firstdiffp, lastdiffp); + return 0; +} + +static int +hwloc_diff_trees(hwloc_topology_t topo1, hwloc_obj_t obj1, + hwloc_topology_t topo2, hwloc_obj_t obj2, + unsigned flags, + hwloc_topology_diff_t *firstdiffp, hwloc_topology_diff_t *lastdiffp) +{ + unsigned i; + int err; + hwloc_obj_t child1, child2; + + if (obj1->depth != obj2->depth) + goto out_too_complex; + + if (obj1->type != obj2->type) + goto out_too_complex; + if ((!obj1->subtype) != (!obj2->subtype) + || (obj1->subtype && strcmp(obj1->subtype, obj2->subtype))) + goto out_too_complex; + + if (obj1->os_index != obj2->os_index) + /* we could allow different os_index for non-PU non-NUMAnode objects + * but it's likely useless anyway */ + goto out_too_complex; + +#define _SETS_DIFFERENT(_set1, _set2) \ + ( ( !(_set1) != !(_set2) ) \ + || ( (_set1) && !hwloc_bitmap_isequal(_set1, _set2) ) ) +#define SETS_DIFFERENT(_set, _obj1, _obj2) _SETS_DIFFERENT((_obj1)->_set, (_obj2)->_set) + if (SETS_DIFFERENT(cpuset, obj1, obj2) + || SETS_DIFFERENT(complete_cpuset, obj1, obj2) + || SETS_DIFFERENT(nodeset, obj1, obj2) + || SETS_DIFFERENT(complete_nodeset, obj1, obj2)) + goto out_too_complex; + + /* no need to check logical_index, sibling_rank, symmetric_subtree, + * the parents did it */ + + /* gp_index don't have to be strictly identical */ + + if ((!obj1->name) != (!obj2->name) + || (obj1->name && strcmp(obj1->name, obj2->name))) { + err = hwloc_append_diff_obj_attr_string(obj1, + HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_NAME, + NULL, + obj1->name, + obj2->name, + firstdiffp, lastdiffp); + if (err < 0) + return err; + } + + /* type-specific attrs */ + switch (obj1->type) { + default: + break; + case HWLOC_OBJ_NUMANODE: + if (obj1->attr->numanode.local_memory != obj2->attr->numanode.local_memory) { + err = hwloc_append_diff_obj_attr_uint64(obj1, + HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_SIZE, + 0, + obj1->attr->numanode.local_memory, + obj2->attr->numanode.local_memory, + firstdiffp, lastdiffp); + if (err < 0) + return err; + } + /* ignore memory page_types */ + break; + case HWLOC_OBJ_L1CACHE: + case HWLOC_OBJ_L2CACHE: + case HWLOC_OBJ_L3CACHE: + case HWLOC_OBJ_L4CACHE: + case HWLOC_OBJ_L5CACHE: + case HWLOC_OBJ_L1ICACHE: + case HWLOC_OBJ_L2ICACHE: + case HWLOC_OBJ_L3ICACHE: + if (memcmp(obj1->attr, obj2->attr, sizeof(obj1->attr->cache))) + goto out_too_complex; + break; + case HWLOC_OBJ_GROUP: + if (memcmp(obj1->attr, obj2->attr, sizeof(obj1->attr->group))) + goto out_too_complex; + break; + case HWLOC_OBJ_PCI_DEVICE: + if (memcmp(obj1->attr, obj2->attr, sizeof(obj1->attr->pcidev))) + goto out_too_complex; + break; + case HWLOC_OBJ_BRIDGE: + if (memcmp(obj1->attr, obj2->attr, sizeof(obj1->attr->bridge))) + goto out_too_complex; + break; + case HWLOC_OBJ_OS_DEVICE: + if (memcmp(obj1->attr, obj2->attr, sizeof(obj1->attr->osdev))) + goto out_too_complex; + break; + } + + /* infos */ + if (obj1->infos_count != obj2->infos_count) + goto out_too_complex; + for(i=0; iinfos_count; i++) { + struct hwloc_info_s *info1 = &obj1->infos[i], *info2 = &obj2->infos[i]; + if (strcmp(info1->name, info2->name)) + goto out_too_complex; + if (strcmp(obj1->infos[i].value, obj2->infos[i].value)) { + err = hwloc_append_diff_obj_attr_string(obj1, + HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_INFO, + info1->name, + info1->value, + info2->value, + firstdiffp, lastdiffp); + if (err < 0) + return err; + } + } + + /* ignore userdata */ + + /* children */ + for(child1 = obj1->first_child, child2 = obj2->first_child; + child1 != NULL && child2 != NULL; + child1 = child1->next_sibling, child2 = child2->next_sibling) { + err = hwloc_diff_trees(topo1, child1, + topo2, child2, + flags, + firstdiffp, lastdiffp); + if (err < 0) + return err; + } + if (child1 || child2) + goto out_too_complex; + + /* memory children */ + for(child1 = obj1->memory_first_child, child2 = obj2->memory_first_child; + child1 != NULL && child2 != NULL; + child1 = child1->next_sibling, child2 = child2->next_sibling) { + err = hwloc_diff_trees(topo1, child1, + topo2, child2, + flags, + firstdiffp, lastdiffp); + if (err < 0) + return err; + } + if (child1 || child2) + goto out_too_complex; + + /* I/O children */ + for(child1 = obj1->io_first_child, child2 = obj2->io_first_child; + child1 != NULL && child2 != NULL; + child1 = child1->next_sibling, child2 = child2->next_sibling) { + err = hwloc_diff_trees(topo1, child1, + topo2, child2, + flags, + firstdiffp, lastdiffp); + if (err < 0) + return err; + } + if (child1 || child2) + goto out_too_complex; + + /* misc children */ + for(child1 = obj1->misc_first_child, child2 = obj2->misc_first_child; + child1 != NULL && child2 != NULL; + child1 = child1->next_sibling, child2 = child2->next_sibling) { + err = hwloc_diff_trees(topo1, child1, + topo2, child2, + flags, + firstdiffp, lastdiffp); + if (err < 0) + return err; + } + if (child1 || child2) + goto out_too_complex; + + return 0; + +out_too_complex: + hwloc_append_diff_too_complex(obj1, firstdiffp, lastdiffp); + return 0; +} + +int hwloc_topology_diff_build(hwloc_topology_t topo1, + hwloc_topology_t topo2, + unsigned long flags, + hwloc_topology_diff_t *diffp) +{ + hwloc_topology_diff_t lastdiff, tmpdiff; + struct hwloc_internal_distances_s *dist1, *dist2; + unsigned i; + int err; + + if (!topo1->is_loaded || !topo2->is_loaded) { + errno = EINVAL; + return -1; + } + + if (flags != 0) { + errno = EINVAL; + return -1; + } + + *diffp = NULL; + err = hwloc_diff_trees(topo1, hwloc_get_root_obj(topo1), + topo2, hwloc_get_root_obj(topo2), + flags, + diffp, &lastdiff); + if (!err) { + tmpdiff = *diffp; + while (tmpdiff) { + if (tmpdiff->generic.type == HWLOC_TOPOLOGY_DIFF_TOO_COMPLEX) { + err = 1; + break; + } + tmpdiff = tmpdiff->generic.next; + } + } + + if (!err) { + if (SETS_DIFFERENT(allowed_cpuset, topo1, topo2) + || SETS_DIFFERENT(allowed_nodeset, topo1, topo2)) { + hwloc_append_diff_too_complex(hwloc_get_root_obj(topo1), diffp, &lastdiff); + err = 1; + } + } + + if (!err) { + /* distances */ + hwloc_internal_distances_refresh(topo1); + hwloc_internal_distances_refresh(topo2); + dist1 = topo1->first_dist; + dist2 = topo2->first_dist; + while (dist1 || dist2) { + if (!!dist1 != !!dist2) { + hwloc_append_diff_too_complex(hwloc_get_root_obj(topo1), diffp, &lastdiff); + err = 1; + break; + } + if (dist1->type != dist2->type + || dist1->nbobjs != dist2->nbobjs + || dist1->kind != dist2->kind + || memcmp(dist1->values, dist2->values, dist1->nbobjs * dist1->nbobjs * sizeof(*dist1->values))) { + hwloc_append_diff_too_complex(hwloc_get_root_obj(topo1), diffp, &lastdiff); + err = 1; + break; + } + for(i=0; inbobjs; i++) + /* gp_index isn't enforced above. so compare logical_index instead, which is enforced. requires distances refresh() above */ + if (dist1->objs[i]->logical_index != dist2->objs[i]->logical_index) { + hwloc_append_diff_too_complex(hwloc_get_root_obj(topo1), diffp, &lastdiff); + err = 1; + break; + } + dist1 = dist1->next; + dist2 = dist2->next; + } + } + + return err; +} + +/******************** + * Applying diffs + */ + +static int +hwloc_apply_diff_one(hwloc_topology_t topology, + hwloc_topology_diff_t diff, + unsigned long flags) +{ + int reverse = !!(flags & HWLOC_TOPOLOGY_DIFF_APPLY_REVERSE); + + switch (diff->generic.type) { + case HWLOC_TOPOLOGY_DIFF_OBJ_ATTR: { + struct hwloc_topology_diff_obj_attr_s *obj_attr = &diff->obj_attr; + hwloc_obj_t obj = hwloc_get_obj_by_depth(topology, obj_attr->obj_depth, obj_attr->obj_index); + if (!obj) + return -1; + + switch (obj_attr->diff.generic.type) { + case HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_SIZE: { + hwloc_obj_t tmpobj; + hwloc_uint64_t oldvalue = reverse ? obj_attr->diff.uint64.newvalue : obj_attr->diff.uint64.oldvalue; + hwloc_uint64_t newvalue = reverse ? obj_attr->diff.uint64.oldvalue : obj_attr->diff.uint64.newvalue; + hwloc_uint64_t valuediff = newvalue - oldvalue; + if (obj->type != HWLOC_OBJ_NUMANODE) + return -1; + if (obj->attr->numanode.local_memory != oldvalue) + return -1; + obj->attr->numanode.local_memory = newvalue; + tmpobj = obj; + while (tmpobj) { + tmpobj->total_memory += valuediff; + tmpobj = tmpobj->parent; + } + break; + } + case HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_NAME: { + const char *oldvalue = reverse ? obj_attr->diff.string.newvalue : obj_attr->diff.string.oldvalue; + const char *newvalue = reverse ? obj_attr->diff.string.oldvalue : obj_attr->diff.string.newvalue; + if (!obj->name || strcmp(obj->name, oldvalue)) + return -1; + free(obj->name); + obj->name = strdup(newvalue); + break; + } + case HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_INFO: { + const char *name = obj_attr->diff.string.name; + const char *oldvalue = reverse ? obj_attr->diff.string.newvalue : obj_attr->diff.string.oldvalue; + const char *newvalue = reverse ? obj_attr->diff.string.oldvalue : obj_attr->diff.string.newvalue; + unsigned i; + int found = 0; + for(i=0; iinfos_count; i++) { + struct hwloc_info_s *info = &obj->infos[i]; + if (!strcmp(info->name, name) + && !strcmp(info->value, oldvalue)) { + free(info->value); + info->value = strdup(newvalue); + found = 1; + break; + } + } + if (!found) + return -1; + break; + } + default: + return -1; + } + + break; + } + default: + return -1; + } + + return 0; +} + +int hwloc_topology_diff_apply(hwloc_topology_t topology, + hwloc_topology_diff_t diff, + unsigned long flags) +{ + hwloc_topology_diff_t tmpdiff, tmpdiff2; + int err, nr; + + if (!topology->is_loaded) { + errno = EINVAL; + return -1; + } + + if (flags & ~HWLOC_TOPOLOGY_DIFF_APPLY_REVERSE) { + errno = EINVAL; + return -1; + } + + tmpdiff = diff; + nr = 0; + while (tmpdiff) { + nr++; + err = hwloc_apply_diff_one(topology, tmpdiff, flags); + if (err < 0) + goto cancel; + tmpdiff = tmpdiff->generic.next; + } + return 0; + +cancel: + tmpdiff2 = tmpdiff; + tmpdiff = diff; + while (tmpdiff != tmpdiff2) { + hwloc_apply_diff_one(topology, tmpdiff, flags ^ HWLOC_TOPOLOGY_DIFF_APPLY_REVERSE); + tmpdiff = tmpdiff->generic.next; + } + errno = EINVAL; + return -nr; /* return the index (starting at 1) of the first element that couldn't be applied */ +} diff --git a/src/3rdparty/hwloc/src/distances.c b/src/3rdparty/hwloc/src/distances.c new file mode 100644 index 000000000..f0b91f019 --- /dev/null +++ b/src/3rdparty/hwloc/src/distances.c @@ -0,0 +1,920 @@ +/* + * Copyright © 2010-2018 Inria. All rights reserved. + * Copyright © 2011-2012 Université Bordeaux + * Copyright © 2011 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +#include +#include +#include +#include +#include + +#include +#include + +/****************************************************** + * Global init, prepare, destroy, dup + */ + +/* called during topology init() */ +void hwloc_internal_distances_init(struct hwloc_topology *topology) +{ + topology->first_dist = topology->last_dist = NULL; + topology->next_dist_id = 0; +} + +/* called at the beginning of load() */ +void hwloc_internal_distances_prepare(struct hwloc_topology *topology) +{ + char *env; + hwloc_localeswitch_declare; + + topology->grouping = 1; + if (topology->type_filter[HWLOC_OBJ_GROUP] == HWLOC_TYPE_FILTER_KEEP_NONE) + topology->grouping = 0; + env = getenv("HWLOC_GROUPING"); + if (env && !atoi(env)) + topology->grouping = 0; + + if (topology->grouping) { + topology->grouping_next_subkind = 0; + + HWLOC_BUILD_ASSERT(sizeof(topology->grouping_accuracies)/sizeof(*topology->grouping_accuracies) == 5); + topology->grouping_accuracies[0] = 0.0f; + topology->grouping_accuracies[1] = 0.01f; + topology->grouping_accuracies[2] = 0.02f; + topology->grouping_accuracies[3] = 0.05f; + topology->grouping_accuracies[4] = 0.1f; + topology->grouping_nbaccuracies = 5; + + hwloc_localeswitch_init(); + env = getenv("HWLOC_GROUPING_ACCURACY"); + if (!env) { + /* only use 0.0 */ + topology->grouping_nbaccuracies = 1; + } else if (strcmp(env, "try")) { + /* use the given value */ + topology->grouping_nbaccuracies = 1; + topology->grouping_accuracies[0] = (float) atof(env); + } /* otherwise try all values */ + hwloc_localeswitch_fini(); + + topology->grouping_verbose = 0; + env = getenv("HWLOC_GROUPING_VERBOSE"); + if (env) + topology->grouping_verbose = atoi(env); + } +} + +static void hwloc_internal_distances_free(struct hwloc_internal_distances_s *dist) +{ + free(dist->indexes); + free(dist->objs); + free(dist->values); + free(dist); +} + +/* called during topology destroy */ +void hwloc_internal_distances_destroy(struct hwloc_topology * topology) +{ + struct hwloc_internal_distances_s *dist, *next = topology->first_dist; + while ((dist = next) != NULL) { + next = dist->next; + hwloc_internal_distances_free(dist); + } + topology->first_dist = topology->last_dist = NULL; +} + +static int hwloc_internal_distances_dup_one(struct hwloc_topology *new, struct hwloc_internal_distances_s *olddist) +{ + struct hwloc_tma *tma = new->tma; + struct hwloc_internal_distances_s *newdist; + unsigned nbobjs = olddist->nbobjs; + + newdist = hwloc_tma_malloc(tma, sizeof(*newdist)); + if (!newdist) + return -1; + + newdist->type = olddist->type; + newdist->nbobjs = nbobjs; + newdist->kind = olddist->kind; + newdist->id = olddist->id; + + newdist->indexes = hwloc_tma_malloc(tma, nbobjs * sizeof(*newdist->indexes)); + newdist->objs = hwloc_tma_calloc(tma, nbobjs * sizeof(*newdist->objs)); + newdist->objs_are_valid = 0; + newdist->values = hwloc_tma_malloc(tma, nbobjs*nbobjs * sizeof(*newdist->values)); + if (!newdist->indexes || !newdist->objs || !newdist->values) { + assert(!tma || !tma->dontfree); /* this tma cannot fail to allocate */ + hwloc_internal_distances_free(newdist); + return -1; + } + + memcpy(newdist->indexes, olddist->indexes, nbobjs * sizeof(*newdist->indexes)); + memcpy(newdist->values, olddist->values, nbobjs*nbobjs * sizeof(*newdist->values)); + + newdist->next = NULL; + newdist->prev = new->last_dist; + if (new->last_dist) + new->last_dist->next = newdist; + else + new->first_dist = newdist; + new->last_dist = newdist; + + return 0; +} + +/* This function may be called with topology->tma set, it cannot free() or realloc() */ +int hwloc_internal_distances_dup(struct hwloc_topology *new, struct hwloc_topology *old) +{ + struct hwloc_internal_distances_s *olddist; + int err; + new->next_dist_id = old->next_dist_id; + for(olddist = old->first_dist; olddist; olddist = olddist->next) { + err = hwloc_internal_distances_dup_one(new, olddist); + if (err < 0) + return err; + } + return 0; +} + +/****************************************************** + * Remove distances from the topology + */ + +int hwloc_distances_remove(hwloc_topology_t topology) +{ + if (!topology->is_loaded) { + errno = EINVAL; + return -1; + } + hwloc_internal_distances_destroy(topology); + return 0; +} + +int hwloc_distances_remove_by_depth(hwloc_topology_t topology, int depth) +{ + struct hwloc_internal_distances_s *dist, *next; + hwloc_obj_type_t type; + + if (!topology->is_loaded) { + errno = EINVAL; + return -1; + } + + /* switch back to types since we don't support groups for now */ + type = hwloc_get_depth_type(topology, depth); + if (type == (hwloc_obj_type_t)-1) { + errno = EINVAL; + return -1; + } + + next = topology->first_dist; + while ((dist = next) != NULL) { + next = dist->next; + if (dist->type == type) { + if (next) + next->prev = dist->prev; + else + topology->last_dist = dist->prev; + if (dist->prev) + dist->prev->next = dist->next; + else + topology->first_dist = dist->next; + hwloc_internal_distances_free(dist); + } + } + + return 0; +} + +/****************************************************** + * Add distances to the topology + */ + +static void +hwloc__groups_by_distances(struct hwloc_topology *topology, unsigned nbobjs, struct hwloc_obj **objs, uint64_t *values, unsigned long kind, unsigned nbaccuracies, float *accuracies, int needcheck); + +/* insert a distance matrix in the topology. + * the caller gives us the distances and objs pointers, we'll free them later. + */ +static int +hwloc_internal_distances__add(hwloc_topology_t topology, + hwloc_obj_type_t type, unsigned nbobjs, hwloc_obj_t *objs, uint64_t *indexes, uint64_t *values, + unsigned long kind) +{ + struct hwloc_internal_distances_s *dist = calloc(1, sizeof(*dist)); + if (!dist) + goto err; + + dist->type = type; + dist->nbobjs = nbobjs; + dist->kind = kind; + + if (!objs) { + assert(indexes); + /* we only have indexes, we'll refresh objs from there */ + dist->indexes = indexes; + dist->objs = calloc(nbobjs, sizeof(hwloc_obj_t)); + if (!dist->objs) + goto err_with_dist; + dist->objs_are_valid = 0; + + } else { + unsigned i; + assert(!indexes); + /* we only have objs, generate the indexes arrays so that we can refresh objs later */ + dist->objs = objs; + dist->objs_are_valid = 1; + dist->indexes = malloc(nbobjs * sizeof(*dist->indexes)); + if (!dist->indexes) + goto err_with_dist; + if (dist->type == HWLOC_OBJ_PU || dist->type == HWLOC_OBJ_NUMANODE) { + for(i=0; iindexes[i] = objs[i]->os_index; + } else { + for(i=0; iindexes[i] = objs[i]->gp_index; + } + } + + dist->values = values; + + dist->id = topology->next_dist_id++; + + if (topology->last_dist) + topology->last_dist->next = dist; + else + topology->first_dist = dist; + dist->prev = topology->last_dist; + dist->next = NULL; + topology->last_dist = dist; + return 0; + + err_with_dist: + free(dist); + err: + free(objs); + free(indexes); + free(values); + return -1; +} + +int hwloc_internal_distances_add_by_index(hwloc_topology_t topology, + hwloc_obj_type_t type, unsigned nbobjs, uint64_t *indexes, uint64_t *values, + unsigned long kind, unsigned long flags) +{ + if (nbobjs < 2) { + errno = EINVAL; + goto err; + } + + /* cannot group without objects, + * and we don't group from XML anyway since the hwloc that generated the XML should have grouped already. + */ + if (flags & HWLOC_DISTANCES_ADD_FLAG_GROUP) { + errno = EINVAL; + goto err; + } + + return hwloc_internal_distances__add(topology, type, nbobjs, NULL, indexes, values, kind); + + err: + free(indexes); + free(values); + return -1; +} + +int hwloc_internal_distances_add(hwloc_topology_t topology, + unsigned nbobjs, hwloc_obj_t *objs, uint64_t *values, + unsigned long kind, unsigned long flags) +{ + if (nbobjs < 2) { + errno = EINVAL; + goto err; + } + + if (topology->grouping && (flags & HWLOC_DISTANCES_ADD_FLAG_GROUP)) { + float full_accuracy = 0.f; + float *accuracies; + unsigned nbaccuracies; + + if (flags & HWLOC_DISTANCES_ADD_FLAG_GROUP_INACCURATE) { + accuracies = topology->grouping_accuracies; + nbaccuracies = topology->grouping_nbaccuracies; + } else { + accuracies = &full_accuracy; + nbaccuracies = 1; + } + + if (topology->grouping_verbose) { + unsigned i, j; + int gp = (objs[0]->type != HWLOC_OBJ_NUMANODE && objs[0]->type != HWLOC_OBJ_PU); + fprintf(stderr, "Trying to group objects using distance matrix:\n"); + fprintf(stderr, "%s", gp ? "gp_index" : "os_index"); + for(j=0; jgp_index : objs[j]->os_index)); + fprintf(stderr, "\n"); + for(i=0; igp_index : objs[i]->os_index)); + for(j=0; jtype, nbobjs, objs, NULL, values, kind); + + err: + free(objs); + free(values); + return -1; +} + +#define HWLOC_DISTANCES_KIND_FROM_ALL (HWLOC_DISTANCES_KIND_FROM_OS|HWLOC_DISTANCES_KIND_FROM_USER) +#define HWLOC_DISTANCES_KIND_MEANS_ALL (HWLOC_DISTANCES_KIND_MEANS_LATENCY|HWLOC_DISTANCES_KIND_MEANS_BANDWIDTH) +#define HWLOC_DISTANCES_KIND_ALL (HWLOC_DISTANCES_KIND_FROM_ALL|HWLOC_DISTANCES_KIND_MEANS_ALL) +#define HWLOC_DISTANCES_ADD_FLAG_ALL (HWLOC_DISTANCES_ADD_FLAG_GROUP|HWLOC_DISTANCES_ADD_FLAG_GROUP_INACCURATE) + +/* The actual function exported to the user + */ +int hwloc_distances_add(hwloc_topology_t topology, + unsigned nbobjs, hwloc_obj_t *objs, hwloc_uint64_t *values, + unsigned long kind, unsigned long flags) +{ + hwloc_obj_type_t type; + unsigned i; + uint64_t *_values; + hwloc_obj_t *_objs; + int err; + + if (nbobjs < 2 || !objs || !values || !topology->is_loaded) { + errno = EINVAL; + return -1; + } + if ((kind & ~HWLOC_DISTANCES_KIND_ALL) + || hwloc_weight_long(kind & HWLOC_DISTANCES_KIND_FROM_ALL) != 1 + || hwloc_weight_long(kind & HWLOC_DISTANCES_KIND_MEANS_ALL) != 1 + || (flags & ~HWLOC_DISTANCES_ADD_FLAG_ALL)) { + errno = EINVAL; + return -1; + } + + /* no strict need to check for duplicates, things shouldn't break */ + + type = objs[0]->type; + if (type == HWLOC_OBJ_GROUP) { + /* not supported yet, would require we save the subkind together with the type. */ + errno = EINVAL; + return -1; + } + + for(i=1; itype != type) { + errno = EINVAL; + return -1; + } + + /* copy the input arrays and give them to the topology */ + _objs = malloc(nbobjs*sizeof(hwloc_obj_t)); + _values = malloc(nbobjs*nbobjs*sizeof(*_values)); + if (!_objs || !_values) + goto out_with_arrays; + + memcpy(_objs, objs, nbobjs*sizeof(hwloc_obj_t)); + memcpy(_values, values, nbobjs*nbobjs*sizeof(*_values)); + err = hwloc_internal_distances_add(topology, nbobjs, _objs, _values, kind, flags); + if (err < 0) + goto out; /* _objs and _values freed in hwloc_internal_distances_add() */ + + /* in case we added some groups, see if we need to reconnect */ + hwloc_topology_reconnect(topology, 0); + + return 0; + + out_with_arrays: + free(_values); + free(_objs); + out: + return -1; +} + +/****************************************************** + * Refresh objects in distances + */ + +static hwloc_obj_t hwloc_find_obj_by_type_and_gp_index(hwloc_topology_t topology, hwloc_obj_type_t type, uint64_t gp_index) +{ + hwloc_obj_t obj = hwloc_get_obj_by_type(topology, type, 0); + while (obj) { + if (obj->gp_index == gp_index) + return obj; + obj = obj->next_cousin; + } + return NULL; +} + +static void +hwloc_internal_distances_restrict(struct hwloc_internal_distances_s *dist, + hwloc_obj_t *objs, + unsigned disappeared) +{ + unsigned nbobjs = dist->nbobjs; + unsigned i, newi; + unsigned j, newj; + + for(i=0, newi=0; ivalues[newi*(nbobjs-disappeared)+newj] = dist->values[i*nbobjs+j]; + newj++; + } + newi++; + } + + for(i=0, newi=0; iindexes[newi] = dist->indexes[i]; + newi++; + } + + dist->nbobjs -= disappeared; +} + +static int +hwloc_internal_distances_refresh_one(hwloc_topology_t topology, + struct hwloc_internal_distances_s *dist) +{ + hwloc_obj_type_t type = dist->type; + unsigned nbobjs = dist->nbobjs; + hwloc_obj_t *objs = dist->objs; + uint64_t *indexes = dist->indexes; + unsigned disappeared = 0; + unsigned i; + + if (dist->objs_are_valid) + return 0; + + for(i=0; iobjs_are_valid = 1; + return 0; +} + +/* This function may be called with topology->tma set, it cannot free() or realloc() */ +void +hwloc_internal_distances_refresh(hwloc_topology_t topology) +{ + struct hwloc_internal_distances_s *dist, *next; + + for(dist = topology->first_dist; dist; dist = next) { + next = dist->next; + + if (hwloc_internal_distances_refresh_one(topology, dist) < 0) { + assert(!topology->tma || !topology->tma->dontfree); /* this tma cannot fail to allocate */ + if (dist->prev) + dist->prev->next = next; + else + topology->first_dist = next; + if (next) + next->prev = dist->prev; + else + topology->last_dist = dist->prev; + hwloc_internal_distances_free(dist); + continue; + } + } +} + +void +hwloc_internal_distances_invalidate_cached_objs(hwloc_topology_t topology) +{ + struct hwloc_internal_distances_s *dist; + for(dist = topology->first_dist; dist; dist = dist->next) + dist->objs_are_valid = 0; +} + +/****************************************************** + * User API for getting distances + */ + +void +hwloc_distances_release(hwloc_topology_t topology __hwloc_attribute_unused, + struct hwloc_distances_s *distances) +{ + free(distances->values); + free(distances->objs); + free(distances); +} + +static struct hwloc_distances_s * +hwloc_distances_get_one(hwloc_topology_t topology __hwloc_attribute_unused, + struct hwloc_internal_distances_s *dist) +{ + struct hwloc_distances_s *distances; + unsigned nbobjs; + + distances = malloc(sizeof(*distances)); + if (!distances) + return NULL; + + nbobjs = distances->nbobjs = dist->nbobjs; + + distances->objs = malloc(nbobjs * sizeof(hwloc_obj_t)); + if (!distances->objs) + goto out; + memcpy(distances->objs, dist->objs, nbobjs * sizeof(hwloc_obj_t)); + + distances->values = malloc(nbobjs * nbobjs * sizeof(*distances->values)); + if (!distances->values) + goto out_with_objs; + memcpy(distances->values, dist->values, nbobjs*nbobjs*sizeof(*distances->values)); + + distances->kind = dist->kind; + return distances; + + out_with_objs: + free(distances->objs); + out: + free(distances); + return NULL; +} + +static int +hwloc__distances_get(hwloc_topology_t topology, + hwloc_obj_type_t type, + unsigned *nrp, struct hwloc_distances_s **distancesp, + unsigned long kind, unsigned long flags __hwloc_attribute_unused) +{ + struct hwloc_internal_distances_s *dist; + unsigned nr = 0, i; + + /* We could return the internal arrays (as const), + * but it would require to prevent removing distances between get() and free(). + * Not performance critical anyway. + */ + + if (flags) { + errno = EINVAL; + return -1; + } + + /* we could refresh only the distances that match, but we won't have many distances anyway, + * so performance is totally negligible. + * + * This is also useful in multithreaded apps that modify the topology. + * They can call any valid hwloc_distances_get() to force a refresh after + * changing the topology, so that future concurrent get() won't cause + * concurrent refresh(). + */ + hwloc_internal_distances_refresh(topology); + + for(dist = topology->first_dist; dist; dist = dist->next) { + unsigned long kind_from = kind & HWLOC_DISTANCES_KIND_FROM_ALL; + unsigned long kind_means = kind & HWLOC_DISTANCES_KIND_MEANS_ALL; + + if (type != HWLOC_OBJ_TYPE_NONE && type != dist->type) + continue; + + if (kind_from && !(kind_from & dist->kind)) + continue; + if (kind_means && !(kind_means & dist->kind)) + continue; + + if (nr < *nrp) { + struct hwloc_distances_s *distances = hwloc_distances_get_one(topology, dist); + if (!distances) + goto error; + distancesp[nr] = distances; + } + nr++; + } + + for(i=nr; i<*nrp; i++) + distancesp[i] = NULL; + *nrp = nr; + return 0; + + error: + for(i=0; iis_loaded) { + errno = EINVAL; + return -1; + } + + return hwloc__distances_get(topology, HWLOC_OBJ_TYPE_NONE, nrp, distancesp, kind, flags); +} + +int +hwloc_distances_get_by_depth(hwloc_topology_t topology, int depth, + unsigned *nrp, struct hwloc_distances_s **distancesp, + unsigned long kind, unsigned long flags) +{ + hwloc_obj_type_t type; + + if (flags || !topology->is_loaded) { + errno = EINVAL; + return -1; + } + + /* switch back to types since we don't support groups for now */ + type = hwloc_get_depth_type(topology, depth); + if (type == (hwloc_obj_type_t)-1) { + errno = EINVAL; + return -1; + } + + return hwloc__distances_get(topology, type, nrp, distancesp, kind, flags); +} + +/****************************************************** + * Grouping objects according to distances + */ + +static void hwloc_report_user_distance_error(const char *msg, int line) +{ + static int reported = 0; + + if (!reported && !hwloc_hide_errors()) { + fprintf(stderr, "****************************************************************************\n"); + fprintf(stderr, "* hwloc %s was given invalid distances by the user.\n", HWLOC_VERSION); + fprintf(stderr, "*\n"); + fprintf(stderr, "* %s\n", msg); + fprintf(stderr, "* Error occurred in topology.c line %d\n", line); + fprintf(stderr, "*\n"); + fprintf(stderr, "* Please make sure that distances given through the programming API\n"); + fprintf(stderr, "* do not contradict any other topology information.\n"); + fprintf(stderr, "* \n"); + fprintf(stderr, "* hwloc will now ignore this invalid topology information and continue.\n"); + fprintf(stderr, "****************************************************************************\n"); + reported = 1; + } +} + +static int hwloc_compare_values(uint64_t a, uint64_t b, float accuracy) +{ + if (accuracy != 0.0f && fabsf((float)a-(float)b) < (float)a * accuracy) + return 0; + return a < b ? -1 : a == b ? 0 : 1; +} + +/* + * Place objects in groups if they are in a transitive graph of minimal values. + * Return how many groups were created, or 0 if some incomplete distance graphs were found. + */ +static unsigned +hwloc__find_groups_by_min_distance(unsigned nbobjs, + uint64_t *_values, + float accuracy, + unsigned *groupids, + int verbose) +{ + uint64_t min_distance = UINT64_MAX; + unsigned groupid = 1; + unsigned i,j,k; + unsigned skipped = 0; + +#define VALUE(i, j) _values[(i) * nbobjs + (j)] + + memset(groupids, 0, nbobjs*sizeof(*groupids)); + + /* find the minimal distance */ + for(i=0; igrouping_verbose; + + if (nbobjs <= 2) + return; + + if (!(kind & HWLOC_DISTANCES_KIND_MEANS_LATENCY)) + /* don't know use to use those for grouping */ + /* TODO hwloc__find_groups_by_max_distance() for bandwidth */ + return; + + for(i=0; itype), accuracies[i]); + if (needcheck && hwloc__check_grouping_matrix(nbobjs, _values, accuracies[i], verbose) < 0) + continue; + nbgroups = hwloc__find_groups_by_min_distance(nbobjs, _values, accuracies[i], groupids, verbose); + if (nbgroups) + break; + } + if (!nbgroups) + return; + + { + HWLOC_VLA(hwloc_obj_t, groupobjs, nbgroups); + HWLOC_VLA(unsigned, groupsizes, nbgroups); + HWLOC_VLA(uint64_t, groupvalues, nbgroups*nbgroups); + unsigned failed = 0; + + /* create new Group objects and record their size */ + memset(&(groupsizes[0]), 0, sizeof(groupsizes[0]) * nbgroups); + for(i=0; icpuset = hwloc_bitmap_alloc(); + group_obj->attr->group.kind = HWLOC_GROUP_KIND_DISTANCE; + group_obj->attr->group.subkind = topology->grouping_next_subkind; + for (j=0; jcpuset); + res_obj = hwloc__insert_object_by_cpuset(topology, NULL, group_obj, + (kind & HWLOC_DISTANCES_KIND_FROM_USER) ? hwloc_report_user_distance_error : hwloc_report_os_error); + /* res_obj may be NULL on failure to insert. */ + if (!res_obj) + failed++; + /* or it may be different from groupobjs if we got groups from XML import before grouping */ + groupobjs[i] = res_obj; + } + topology->grouping_next_subkind++; + + if (failed) + /* don't try to group above if we got a NULL group here, just keep this incomplete level */ + return; + + /* factorize values */ + memset(&(groupvalues[0]), 0, sizeof(groupvalues[0]) * nbgroups * nbgroups); +#undef VALUE +#define VALUE(i, j) _values[(i) * nbobjs + (j)] +#define GROUP_VALUE(i, j) groupvalues[(i) * nbgroups + (j)] + for(i=0; i +#include +#include + +#include +#ifdef HAVE_SYS_UTSNAME_H +#include +#endif +#include +#include +#include +#include +#include + +#ifdef HAVE_PROGRAM_INVOCATION_NAME +#include +extern char *program_invocation_name; +#endif +#ifdef HAVE___PROGNAME +extern char *__progname; +#endif + +int hwloc_snprintf(char *str, size_t size, const char *format, ...) +{ + int ret; + va_list ap; + static char bin; + size_t fakesize; + char *fakestr; + + /* Some systems crash on str == NULL */ + if (!size) { + str = &bin; + size = 1; + } + + va_start(ap, format); + ret = vsnprintf(str, size, format, ap); + va_end(ap); + + if (ret >= 0 && (size_t) ret != size-1) + return ret; + + /* vsnprintf returned size-1 or -1. That could be a system which reports the + * written data and not the actually required room. Try increasing buffer + * size to get the latter. */ + + fakesize = size; + fakestr = NULL; + do { + fakesize *= 2; + free(fakestr); + fakestr = malloc(fakesize); + if (NULL == fakestr) + return -1; + va_start(ap, format); + errno = 0; + ret = vsnprintf(fakestr, fakesize, format, ap); + va_end(ap); + } while ((size_t) ret == fakesize-1 || (ret < 0 && (!errno || errno == ERANGE))); + + if (ret >= 0 && size) { + if (size > (size_t) ret+1) + size = ret+1; + memcpy(str, fakestr, size-1); + str[size-1] = 0; + } + free(fakestr); + + return ret; +} + +int hwloc_namecoloncmp(const char *haystack, const char *needle, size_t n) +{ + size_t i = 0; + while (*haystack && *haystack != ':') { + int ha = *haystack++; + int low_h = tolower(ha); + int ne = *needle++; + int low_n = tolower(ne); + if (low_h != low_n) + return 1; + i++; + } + return i < n; +} + +void hwloc_add_uname_info(struct hwloc_topology *topology __hwloc_attribute_unused, + void *cached_uname __hwloc_attribute_unused) +{ +#ifdef HAVE_UNAME + struct utsname _utsname, *utsname; + + if (hwloc_obj_get_info_by_name(topology->levels[0][0], "OSName")) + /* don't annotate twice */ + return; + + if (cached_uname) + utsname = (struct utsname *) cached_uname; + else { + utsname = &_utsname; + if (uname(utsname) < 0) + return; + } + + if (*utsname->sysname) + hwloc_obj_add_info(topology->levels[0][0], "OSName", utsname->sysname); + if (*utsname->release) + hwloc_obj_add_info(topology->levels[0][0], "OSRelease", utsname->release); + if (*utsname->version) + hwloc_obj_add_info(topology->levels[0][0], "OSVersion", utsname->version); + if (*utsname->nodename) + hwloc_obj_add_info(topology->levels[0][0], "HostName", utsname->nodename); + if (*utsname->machine) + hwloc_obj_add_info(topology->levels[0][0], "Architecture", utsname->machine); +#endif /* HAVE_UNAME */ +} + +char * +hwloc_progname(struct hwloc_topology *topology __hwloc_attribute_unused) +{ +#if HAVE_DECL_GETMODULEFILENAME + char name[256], *local_basename; + unsigned res = GetModuleFileName(NULL, name, sizeof(name)); + if (res == sizeof(name) || !res) + return NULL; + local_basename = strrchr(name, '\\'); + if (!local_basename) + local_basename = name; + else + local_basename++; + return strdup(local_basename); +#else /* !HAVE_GETMODULEFILENAME */ + const char *name, *local_basename; +#if HAVE_DECL_GETPROGNAME + name = getprogname(); /* FreeBSD, NetBSD, some Solaris */ +#elif HAVE_DECL_GETEXECNAME + name = getexecname(); /* Solaris */ +#elif defined HAVE_PROGRAM_INVOCATION_NAME + name = program_invocation_name; /* Glibc. BGQ CNK. */ + /* could use program_invocation_short_name directly, but we have the code to remove the path below anyway */ +#elif defined HAVE___PROGNAME + name = __progname; /* fallback for most unix, used for OpenBSD */ +#else + /* TODO: _NSGetExecutablePath(path, &size) on Darwin */ + /* TODO: AIX, HPUX */ + name = NULL; +#endif + if (!name) + return NULL; + local_basename = strrchr(name, '/'); + if (!local_basename) + local_basename = name; + else + local_basename++; + return strdup(local_basename); +#endif /* !HAVE_GETMODULEFILENAME */ +} diff --git a/src/3rdparty/hwloc/src/pci-common.c b/src/3rdparty/hwloc/src/pci-common.c new file mode 100644 index 000000000..00f08a9e7 --- /dev/null +++ b/src/3rdparty/hwloc/src/pci-common.c @@ -0,0 +1,941 @@ +/* + * Copyright © 2009-2018 Inria. All rights reserved. + * See COPYING in top-level directory. + */ + +#include +#include +#include +#include +#include +#include + +#include +#ifdef HAVE_UNISTD_H +#include +#endif +#include + +#if defined(HWLOC_WIN_SYS) && !defined(__CYGWIN__) +#include +#define open _open +#define read _read +#define close _close +#endif + +static void +hwloc_pci_forced_locality_parse_one(struct hwloc_topology *topology, + const char *string /* must contain a ' ' */, + unsigned *allocated) +{ + unsigned nr = topology->pci_forced_locality_nr; + unsigned domain, bus_first, bus_last, dummy; + hwloc_bitmap_t set; + char *tmp; + + if (sscanf(string, "%x:%x-%x %x", &domain, &bus_first, &bus_last, &dummy) == 4) { + /* fine */ + } else if (sscanf(string, "%x:%x %x", &domain, &bus_first, &dummy) == 3) { + bus_last = bus_first; + } else if (sscanf(string, "%x %x", &domain, &dummy) == 2) { + bus_first = 0; + bus_last = 255; + } else + return; + + tmp = strchr(string, ' '); + if (!tmp) + return; + tmp++; + + set = hwloc_bitmap_alloc(); + hwloc_bitmap_sscanf(set, tmp); + + if (!*allocated) { + topology->pci_forced_locality = malloc(sizeof(*topology->pci_forced_locality)); + if (!topology->pci_forced_locality) + goto out_with_set; /* failed to allocate, ignore this forced locality */ + *allocated = 1; + } else if (nr >= *allocated) { + struct hwloc_pci_forced_locality_s *tmplocs; + tmplocs = realloc(topology->pci_forced_locality, + 2 * *allocated * sizeof(*topology->pci_forced_locality)); + if (!tmplocs) + goto out_with_set; /* failed to allocate, ignore this forced locality */ + topology->pci_forced_locality = tmplocs; + *allocated *= 2; + } + + topology->pci_forced_locality[nr].domain = domain; + topology->pci_forced_locality[nr].bus_first = bus_first; + topology->pci_forced_locality[nr].bus_last = bus_last; + topology->pci_forced_locality[nr].cpuset = set; + topology->pci_forced_locality_nr++; + return; + + out_with_set: + hwloc_bitmap_free(set); + return; +} + +static void +hwloc_pci_forced_locality_parse(struct hwloc_topology *topology, const char *_env) +{ + char *env = strdup(_env); + unsigned allocated = 0; + char *tmp = env; + + while (1) { + size_t len = strcspn(tmp, ";\r\n"); + char *next = NULL; + + if (tmp[len] != '\0') { + tmp[len] = '\0'; + if (tmp[len+1] != '\0') + next = &tmp[len]+1; + } + + hwloc_pci_forced_locality_parse_one(topology, tmp, &allocated); + + if (next) + tmp = next; + else + break; + } + + free(env); +} + +void +hwloc_pci_discovery_init(struct hwloc_topology *topology) +{ + topology->need_pci_belowroot_apply_locality = 0; + + topology->pci_has_forced_locality = 0; + topology->pci_forced_locality_nr = 0; + topology->pci_forced_locality = NULL; +} + +void +hwloc_pci_discovery_prepare(struct hwloc_topology *topology) +{ + char *env; + + env = getenv("HWLOC_PCI_LOCALITY"); + if (env) { + int fd; + + topology->pci_has_forced_locality = 1; + + fd = open(env, O_RDONLY); + if (fd >= 0) { + struct stat st; + char *buffer; + int err = fstat(fd, &st); + if (!err) { + if (st.st_size <= 64*1024) { /* random limit large enough to store multiple cpusets for thousands of PUs */ + buffer = malloc(st.st_size+1); + if (read(fd, buffer, st.st_size) == st.st_size) { + buffer[st.st_size] = '\0'; + hwloc_pci_forced_locality_parse(topology, buffer); + } + free(buffer); + } else { + fprintf(stderr, "Ignoring HWLOC_PCI_LOCALITY file `%s' too large (%lu bytes)\n", + env, (unsigned long) st.st_size); + } + } + close(fd); + } else + hwloc_pci_forced_locality_parse(topology, env); + } +} + +void +hwloc_pci_discovery_exit(struct hwloc_topology *topology __hwloc_attribute_unused) +{ + unsigned i; + for(i=0; ipci_forced_locality_nr; i++) + hwloc_bitmap_free(topology->pci_forced_locality[i].cpuset); + free(topology->pci_forced_locality); + + hwloc_pci_discovery_init(topology); +} + +#ifdef HWLOC_DEBUG +static void +hwloc_pci_traverse_print_cb(void * cbdata __hwloc_attribute_unused, + struct hwloc_obj *pcidev) +{ + char busid[14]; + hwloc_obj_t parent; + + /* indent */ + parent = pcidev->parent; + while (parent) { + hwloc_debug("%s", " "); + parent = parent->parent; + } + + snprintf(busid, sizeof(busid), "%04x:%02x:%02x.%01x", + pcidev->attr->pcidev.domain, pcidev->attr->pcidev.bus, pcidev->attr->pcidev.dev, pcidev->attr->pcidev.func); + + if (pcidev->type == HWLOC_OBJ_BRIDGE) { + if (pcidev->attr->bridge.upstream_type == HWLOC_OBJ_BRIDGE_HOST) + hwloc_debug("HostBridge"); + else + hwloc_debug("%s Bridge [%04x:%04x]", busid, + pcidev->attr->pcidev.vendor_id, pcidev->attr->pcidev.device_id); + hwloc_debug(" to %04x:[%02x:%02x]\n", + pcidev->attr->bridge.downstream.pci.domain, pcidev->attr->bridge.downstream.pci.secondary_bus, pcidev->attr->bridge.downstream.pci.subordinate_bus); + } else + hwloc_debug("%s Device [%04x:%04x (%04x:%04x) rev=%02x class=%04x]\n", busid, + pcidev->attr->pcidev.vendor_id, pcidev->attr->pcidev.device_id, + pcidev->attr->pcidev.subvendor_id, pcidev->attr->pcidev.subdevice_id, + pcidev->attr->pcidev.revision, pcidev->attr->pcidev.class_id); +} + +static void +hwloc_pci_traverse(void * cbdata, struct hwloc_obj *tree, + void (*cb)(void * cbdata, struct hwloc_obj *)) +{ + hwloc_obj_t child; + cb(cbdata, tree); + for_each_io_child(child, tree) { + if (child->type == HWLOC_OBJ_BRIDGE) + hwloc_pci_traverse(cbdata, child, cb); + } +} +#endif /* HWLOC_DEBUG */ + +enum hwloc_pci_busid_comparison_e { + HWLOC_PCI_BUSID_LOWER, + HWLOC_PCI_BUSID_HIGHER, + HWLOC_PCI_BUSID_INCLUDED, + HWLOC_PCI_BUSID_SUPERSET +}; + +static enum hwloc_pci_busid_comparison_e +hwloc_pci_compare_busids(struct hwloc_obj *a, struct hwloc_obj *b) +{ +#ifdef HWLOC_DEBUG + if (a->type == HWLOC_OBJ_BRIDGE) + assert(a->attr->bridge.upstream_type == HWLOC_OBJ_BRIDGE_PCI); + if (b->type == HWLOC_OBJ_BRIDGE) + assert(b->attr->bridge.upstream_type == HWLOC_OBJ_BRIDGE_PCI); +#endif + + if (a->attr->pcidev.domain < b->attr->pcidev.domain) + return HWLOC_PCI_BUSID_LOWER; + if (a->attr->pcidev.domain > b->attr->pcidev.domain) + return HWLOC_PCI_BUSID_HIGHER; + + if (a->type == HWLOC_OBJ_BRIDGE + && b->attr->pcidev.bus >= a->attr->bridge.downstream.pci.secondary_bus + && b->attr->pcidev.bus <= a->attr->bridge.downstream.pci.subordinate_bus) + return HWLOC_PCI_BUSID_SUPERSET; + if (b->type == HWLOC_OBJ_BRIDGE + && a->attr->pcidev.bus >= b->attr->bridge.downstream.pci.secondary_bus + && a->attr->pcidev.bus <= b->attr->bridge.downstream.pci.subordinate_bus) + return HWLOC_PCI_BUSID_INCLUDED; + + if (a->attr->pcidev.bus < b->attr->pcidev.bus) + return HWLOC_PCI_BUSID_LOWER; + if (a->attr->pcidev.bus > b->attr->pcidev.bus) + return HWLOC_PCI_BUSID_HIGHER; + + if (a->attr->pcidev.dev < b->attr->pcidev.dev) + return HWLOC_PCI_BUSID_LOWER; + if (a->attr->pcidev.dev > b->attr->pcidev.dev) + return HWLOC_PCI_BUSID_HIGHER; + + if (a->attr->pcidev.func < b->attr->pcidev.func) + return HWLOC_PCI_BUSID_LOWER; + if (a->attr->pcidev.func > b->attr->pcidev.func) + return HWLOC_PCI_BUSID_HIGHER; + + /* Should never reach here. Abort on both debug builds and + non-debug builds */ + assert(0); + fprintf(stderr, "Bad assertion in hwloc %s:%d (aborting)\n", __FILE__, __LINE__); + exit(1); +} + +static void +hwloc_pci_add_object(struct hwloc_obj *parent, struct hwloc_obj **parent_io_first_child_p, struct hwloc_obj *new) +{ + struct hwloc_obj **curp, **childp; + + curp = parent_io_first_child_p; + while (*curp) { + enum hwloc_pci_busid_comparison_e comp = hwloc_pci_compare_busids(new, *curp); + switch (comp) { + case HWLOC_PCI_BUSID_HIGHER: + /* go further */ + curp = &(*curp)->next_sibling; + continue; + case HWLOC_PCI_BUSID_INCLUDED: + /* insert new below current bridge */ + hwloc_pci_add_object(*curp, &(*curp)->io_first_child, new); + return; + case HWLOC_PCI_BUSID_LOWER: + case HWLOC_PCI_BUSID_SUPERSET: { + /* insert new before current */ + new->next_sibling = *curp; + *curp = new; + new->parent = parent; + if (new->type == HWLOC_OBJ_BRIDGE) { + /* look at remaining siblings and move some below new */ + childp = &new->io_first_child; + curp = &new->next_sibling; + while (*curp) { + hwloc_obj_t cur = *curp; + if (hwloc_pci_compare_busids(new, cur) == HWLOC_PCI_BUSID_LOWER) { + /* this sibling remains under root, after new. */ + if (cur->attr->pcidev.domain > new->attr->pcidev.domain + || cur->attr->pcidev.bus > new->attr->bridge.downstream.pci.subordinate_bus) + /* this sibling is even above new's subordinate bus, no other sibling could go below new */ + return; + curp = &cur->next_sibling; + } else { + /* this sibling goes under new */ + *childp = cur; + *curp = cur->next_sibling; + (*childp)->parent = new; + (*childp)->next_sibling = NULL; + childp = &(*childp)->next_sibling; + } + } + } + return; + } + } + } + /* add to the end of the list if higher than everybody */ + new->parent = parent; + new->next_sibling = NULL; + *curp = new; +} + +void +hwloc_pcidisc_tree_insert_by_busid(struct hwloc_obj **treep, + struct hwloc_obj *obj) +{ + hwloc_pci_add_object(NULL /* no parent on top of tree */, treep, obj); +} + +int +hwloc_pcidisc_tree_attach(struct hwloc_topology *topology, struct hwloc_obj *old_tree) +{ + struct hwloc_obj **next_hb_p; + enum hwloc_type_filter_e bfilter; + + if (!old_tree) + /* found nothing, exit */ + return 0; + +#ifdef HWLOC_DEBUG + hwloc_debug("%s", "\nPCI hierarchy:\n"); + hwloc_pci_traverse(NULL, old_tree, hwloc_pci_traverse_print_cb); + hwloc_debug("%s", "\n"); +#endif + + next_hb_p = &hwloc_get_root_obj(topology)->io_first_child; + while (*next_hb_p) + next_hb_p = &((*next_hb_p)->next_sibling); + + bfilter = topology->type_filter[HWLOC_OBJ_BRIDGE]; + if (bfilter == HWLOC_TYPE_FILTER_KEEP_NONE) { + *next_hb_p = old_tree; + topology->modified = 1; + goto done; + } + + /* + * tree points to all objects connected to any upstream bus in the machine. + * We now create one real hostbridge object per upstream bus. + * It's not actually a PCI device so we have to create it. + */ + while (old_tree) { + /* start a new host bridge */ + struct hwloc_obj *hostbridge = hwloc_alloc_setup_object(topology, HWLOC_OBJ_BRIDGE, HWLOC_UNKNOWN_INDEX); + struct hwloc_obj **dstnextp = &hostbridge->io_first_child; + struct hwloc_obj **srcnextp = &old_tree; + struct hwloc_obj *child = *srcnextp; + unsigned short current_domain = child->attr->pcidev.domain; + unsigned char current_bus = child->attr->pcidev.bus; + unsigned char current_subordinate = current_bus; + + hwloc_debug("Starting new PCI hostbridge %04x:%02x\n", current_domain, current_bus); + + next_child: + /* remove next child from tree */ + *srcnextp = child->next_sibling; + /* append it to hostbridge */ + *dstnextp = child; + child->parent = hostbridge; + child->next_sibling = NULL; + dstnextp = &child->next_sibling; + + /* compute hostbridge secondary/subordinate buses */ + if (child->type == HWLOC_OBJ_BRIDGE + && child->attr->bridge.downstream.pci.subordinate_bus > current_subordinate) + current_subordinate = child->attr->bridge.downstream.pci.subordinate_bus; + + /* use next child if it has the same domains/bus */ + child = *srcnextp; + if (child + && child->attr->pcidev.domain == current_domain + && child->attr->pcidev.bus == current_bus) + goto next_child; + + /* finish setting up this hostbridge */ + hostbridge->attr->bridge.upstream_type = HWLOC_OBJ_BRIDGE_HOST; + hostbridge->attr->bridge.downstream_type = HWLOC_OBJ_BRIDGE_PCI; + hostbridge->attr->bridge.downstream.pci.domain = current_domain; + hostbridge->attr->bridge.downstream.pci.secondary_bus = current_bus; + hostbridge->attr->bridge.downstream.pci.subordinate_bus = current_subordinate; + hwloc_debug("New PCI hostbridge %04x:[%02x-%02x]\n", + current_domain, current_bus, current_subordinate); + + *next_hb_p = hostbridge; + next_hb_p = &hostbridge->next_sibling; + topology->modified = 1; /* needed in case somebody reconnects levels before the core calls hwloc_pci_belowroot_apply_locality() + * or if hwloc_pci_belowroot_apply_locality() keeps hostbridges below root. + */ + } + + done: + topology->need_pci_belowroot_apply_locality = 1; + return 0; +} + +static struct hwloc_obj * +hwloc_pci_fixup_busid_parent(struct hwloc_topology *topology __hwloc_attribute_unused, + struct hwloc_pcidev_attr_s *busid, + struct hwloc_obj *parent) +{ + /* Xeon E5v3 in cluster-on-die mode only have PCI on the first NUMA node of each package. + * but many dual-processor host report the second PCI hierarchy on 2nd NUMA of first package. + */ + if (parent->depth >= 2 + && parent->type == HWLOC_OBJ_NUMANODE + && parent->sibling_rank == 1 && parent->parent->arity == 2 + && parent->parent->type == HWLOC_OBJ_PACKAGE + && parent->parent->sibling_rank == 0 && parent->parent->parent->arity == 2) { + const char *cpumodel = hwloc_obj_get_info_by_name(parent->parent, "CPUModel"); + if (cpumodel && strstr(cpumodel, "Xeon")) { + if (!hwloc_hide_errors()) { + fprintf(stderr, "****************************************************************************\n"); + fprintf(stderr, "* hwloc %s has encountered an incorrect PCI locality information.\n", HWLOC_VERSION); + fprintf(stderr, "* PCI bus %04x:%02x is supposedly close to 2nd NUMA node of 1st package,\n", + busid->domain, busid->bus); + fprintf(stderr, "* however hwloc believes this is impossible on this architecture.\n"); + fprintf(stderr, "* Therefore the PCI bus will be moved to 1st NUMA node of 2nd package.\n"); + fprintf(stderr, "*\n"); + fprintf(stderr, "* If you feel this fixup is wrong, disable it by setting in your environment\n"); + fprintf(stderr, "* HWLOC_PCI_%04x_%02x_LOCALCPUS= (empty value), and report the problem\n", + busid->domain, busid->bus); + fprintf(stderr, "* to the hwloc's user mailing list together with the XML output of lstopo.\n"); + fprintf(stderr, "*\n"); + fprintf(stderr, "* You may silence this message by setting HWLOC_HIDE_ERRORS=1 in your environment.\n"); + fprintf(stderr, "****************************************************************************\n"); + } + return parent->parent->next_sibling->first_child; + } + } + + return parent; +} + +static struct hwloc_obj * +hwloc__pci_find_busid_parent(struct hwloc_topology *topology, struct hwloc_pcidev_attr_s *busid) +{ + hwloc_bitmap_t cpuset = hwloc_bitmap_alloc(); + hwloc_obj_t parent; + int forced = 0; + int noquirks = 0; + unsigned i; + int err; + + /* try to match a forced locality */ + if (topology->pci_has_forced_locality) { + for(i=0; ipci_forced_locality_nr; i++) { + if (busid->domain == topology->pci_forced_locality[i].domain + && busid->bus >= topology->pci_forced_locality[i].bus_first + && busid->bus <= topology->pci_forced_locality[i].bus_last) { + hwloc_bitmap_copy(cpuset, topology->pci_forced_locality[i].cpuset); + forced = 1; + break; + } + } + /* if pci locality was forced, even empty, don't let quirks change what the OS reports */ + noquirks = 1; + } + + /* deprecated force locality variables */ + if (!forced) { + const char *env; + char envname[256]; + /* override the cpuset with the environment if given */ + snprintf(envname, sizeof(envname), "HWLOC_PCI_%04x_%02x_LOCALCPUS", + busid->domain, busid->bus); + env = getenv(envname); + if (env) { + static int reported = 0; + if (!topology->pci_has_forced_locality && !reported) { + fprintf(stderr, "Environment variable %s is deprecated, please use HWLOC_PCI_LOCALITY instead.\n", env); + reported = 1; + } + if (*env) { + /* force the cpuset */ + hwloc_debug("Overriding localcpus using %s in the environment\n", envname); + hwloc_bitmap_sscanf(cpuset, env); + forced = 1; + } + /* if env exists, even empty, don't let quirks change what the OS reports */ + noquirks = 1; + } + } + + if (!forced) { + /* get the cpuset by asking the OS backend. */ + struct hwloc_backend *backend = topology->get_pci_busid_cpuset_backend; + if (backend) + err = backend->get_pci_busid_cpuset(backend, busid, cpuset); + else + err = -1; + if (err < 0) + /* if we got nothing, assume this PCI bus is attached to the top of hierarchy */ + hwloc_bitmap_copy(cpuset, hwloc_topology_get_topology_cpuset(topology)); + } + + hwloc_debug_bitmap("Attaching PCI tree to cpuset %s\n", cpuset); + + parent = hwloc_find_insert_io_parent_by_complete_cpuset(topology, cpuset); + if (parent) { + if (!noquirks) + /* We found a valid parent. Check that the OS didn't report invalid locality */ + parent = hwloc_pci_fixup_busid_parent(topology, busid, parent); + } else { + /* Fallback to root */ + parent = hwloc_get_root_obj(topology); + } + + hwloc_bitmap_free(cpuset); + return parent; +} + +struct hwloc_obj * +hwloc_pcidisc_find_busid_parent(struct hwloc_topology *topology, + unsigned domain, unsigned bus, unsigned dev, unsigned func) +{ + struct hwloc_pcidev_attr_s busid; + busid.domain = domain; + busid.bus = bus; + busid.dev = dev; + busid.func = func; + return hwloc__pci_find_busid_parent(topology, &busid); +} + +int +hwloc_pci_belowroot_apply_locality(struct hwloc_topology *topology) +{ + struct hwloc_obj *root = hwloc_get_root_obj(topology); + struct hwloc_obj **listp, *obj; + + if (!topology->need_pci_belowroot_apply_locality) + return 0; + topology->need_pci_belowroot_apply_locality = 0; + + /* root->io_first_child contains some PCI hierarchies, any maybe some non-PCI things. + * insert the PCI trees according to their PCI-locality. + */ + listp = &root->io_first_child; + while ((obj = *listp) != NULL) { + struct hwloc_pcidev_attr_s *busid; + struct hwloc_obj *parent; + + /* skip non-PCI objects */ + if (obj->type != HWLOC_OBJ_PCI_DEVICE + && !(obj->type == HWLOC_OBJ_BRIDGE && obj->attr->bridge.downstream_type == HWLOC_OBJ_BRIDGE_PCI) + && !(obj->type == HWLOC_OBJ_BRIDGE && obj->attr->bridge.upstream_type == HWLOC_OBJ_BRIDGE_PCI)) { + listp = &obj->next_sibling; + continue; + } + + if (obj->type == HWLOC_OBJ_PCI_DEVICE + || (obj->type == HWLOC_OBJ_BRIDGE + && obj->attr->bridge.upstream_type == HWLOC_OBJ_BRIDGE_PCI)) + busid = &obj->attr->pcidev; + else { + /* hostbridges don't have a PCI busid for looking up locality, use their first child if PCI */ + hwloc_obj_t child = obj->io_first_child; + if (child && (child->type == HWLOC_OBJ_PCI_DEVICE + || (child->type == HWLOC_OBJ_BRIDGE + && child->attr->bridge.upstream_type == HWLOC_OBJ_BRIDGE_PCI))) + busid = &obj->io_first_child->attr->pcidev; + else + continue; + } + + /* attach the object (and children) where it belongs */ + parent = hwloc__pci_find_busid_parent(topology, busid); + if (parent == root) { + /* keep this object here */ + listp = &obj->next_sibling; + } else { + /* dequeue this object */ + *listp = obj->next_sibling; + obj->next_sibling = NULL; + hwloc_insert_object_by_parent(topology, parent, obj); + } + } + + return 0; +} + +static struct hwloc_obj * +hwloc__pci_belowroot_find_by_busid(hwloc_obj_t parent, + unsigned domain, unsigned bus, unsigned dev, unsigned func) +{ + hwloc_obj_t child; + + for_each_io_child(child, parent) { + if (child->type == HWLOC_OBJ_PCI_DEVICE + || (child->type == HWLOC_OBJ_BRIDGE + && child->attr->bridge.upstream_type == HWLOC_OBJ_BRIDGE_PCI)) { + if (child->attr->pcidev.domain == domain + && child->attr->pcidev.bus == bus + && child->attr->pcidev.dev == dev + && child->attr->pcidev.func == func) + /* that's the right bus id */ + return child; + if (child->attr->pcidev.domain > domain + || (child->attr->pcidev.domain == domain + && child->attr->pcidev.bus > bus)) + /* bus id too high, won't find anything later, return parent */ + return parent; + if (child->type == HWLOC_OBJ_BRIDGE + && child->attr->bridge.downstream_type == HWLOC_OBJ_BRIDGE_PCI + && child->attr->bridge.downstream.pci.domain == domain + && child->attr->bridge.downstream.pci.secondary_bus <= bus + && child->attr->bridge.downstream.pci.subordinate_bus >= bus) + /* not the right bus id, but it's included in the bus below that bridge */ + return hwloc__pci_belowroot_find_by_busid(child, domain, bus, dev, func); + + } else if (child->type == HWLOC_OBJ_BRIDGE + && child->attr->bridge.upstream_type != HWLOC_OBJ_BRIDGE_PCI + && child->attr->bridge.downstream_type == HWLOC_OBJ_BRIDGE_PCI + /* non-PCI to PCI bridge, just look at the subordinate bus */ + && child->attr->bridge.downstream.pci.domain == domain + && child->attr->bridge.downstream.pci.secondary_bus <= bus + && child->attr->bridge.downstream.pci.subordinate_bus >= bus) { + /* contains our bus, recurse */ + return hwloc__pci_belowroot_find_by_busid(child, domain, bus, dev, func); + } + } + /* didn't find anything, return parent */ + return parent; +} + +struct hwloc_obj * +hwloc_pcidisc_find_by_busid(struct hwloc_topology *topology, + unsigned domain, unsigned bus, unsigned dev, unsigned func) +{ + hwloc_obj_t root = hwloc_get_root_obj(topology); + hwloc_obj_t parent = hwloc__pci_belowroot_find_by_busid(root, domain, bus, dev, func); + if (parent == root) + return NULL; + else + return parent; +} + +#define HWLOC_PCI_STATUS 0x06 +#define HWLOC_PCI_STATUS_CAP_LIST 0x10 +#define HWLOC_PCI_CAPABILITY_LIST 0x34 +#define HWLOC_PCI_CAP_LIST_ID 0 +#define HWLOC_PCI_CAP_LIST_NEXT 1 + +unsigned +hwloc_pcidisc_find_cap(const unsigned char *config, unsigned cap) +{ + unsigned char seen[256] = { 0 }; + unsigned char ptr; /* unsigned char to make sure we stay within the 256-byte config space */ + + if (!(config[HWLOC_PCI_STATUS] & HWLOC_PCI_STATUS_CAP_LIST)) + return 0; + + for (ptr = config[HWLOC_PCI_CAPABILITY_LIST] & ~3; + ptr; /* exit if next is 0 */ + ptr = config[ptr + HWLOC_PCI_CAP_LIST_NEXT] & ~3) { + unsigned char id; + + /* Looped around! */ + if (seen[ptr]) + break; + seen[ptr] = 1; + + id = config[ptr + HWLOC_PCI_CAP_LIST_ID]; + if (id == cap) + return ptr; + if (id == 0xff) /* exit if id is 0 or 0xff */ + break; + } + return 0; +} + +#define HWLOC_PCI_EXP_LNKSTA 0x12 +#define HWLOC_PCI_EXP_LNKSTA_SPEED 0x000f +#define HWLOC_PCI_EXP_LNKSTA_WIDTH 0x03f0 + +int +hwloc_pcidisc_find_linkspeed(const unsigned char *config, + unsigned offset, float *linkspeed) +{ + unsigned linksta, speed, width; + float lanespeed; + + memcpy(&linksta, &config[offset + HWLOC_PCI_EXP_LNKSTA], 4); + speed = linksta & HWLOC_PCI_EXP_LNKSTA_SPEED; /* PCIe generation */ + width = (linksta & HWLOC_PCI_EXP_LNKSTA_WIDTH) >> 4; /* how many lanes */ + /* PCIe Gen1 = 2.5GT/s signal-rate per lane with 8/10 encoding = 0.25GB/s data-rate per lane + * PCIe Gen2 = 5 GT/s signal-rate per lane with 8/10 encoding = 0.5 GB/s data-rate per lane + * PCIe Gen3 = 8 GT/s signal-rate per lane with 128/130 encoding = 1 GB/s data-rate per lane + * PCIe Gen4 = 16 GT/s signal-rate per lane with 128/130 encoding = 2 GB/s data-rate per lane + */ + + /* lanespeed in Gbit/s */ + if (speed <= 2) + lanespeed = 2.5f * speed * 0.8f; + else + lanespeed = 8.0f * (1<<(speed-3)) * 128/130; /* assume Gen5 will be 32 GT/s and so on */ + + /* linkspeed in GB/s */ + *linkspeed = lanespeed * width / 8; + return 0; +} + +#define HWLOC_PCI_HEADER_TYPE 0x0e +#define HWLOC_PCI_HEADER_TYPE_BRIDGE 1 +#define HWLOC_PCI_CLASS_BRIDGE_PCI 0x0604 + +hwloc_obj_type_t +hwloc_pcidisc_check_bridge_type(unsigned device_class, const unsigned char *config) +{ + unsigned char headertype; + + if (device_class != HWLOC_PCI_CLASS_BRIDGE_PCI) + return HWLOC_OBJ_PCI_DEVICE; + + headertype = config[HWLOC_PCI_HEADER_TYPE] & 0x7f; + return (headertype == HWLOC_PCI_HEADER_TYPE_BRIDGE) + ? HWLOC_OBJ_BRIDGE : HWLOC_OBJ_PCI_DEVICE; +} + +#define HWLOC_PCI_PRIMARY_BUS 0x18 +#define HWLOC_PCI_SECONDARY_BUS 0x19 +#define HWLOC_PCI_SUBORDINATE_BUS 0x1a + +int +hwloc_pcidisc_setup_bridge_attr(hwloc_obj_t obj, + const unsigned char *config) +{ + struct hwloc_bridge_attr_s *battr = &obj->attr->bridge; + struct hwloc_pcidev_attr_s *pattr = &battr->upstream.pci; + + if (config[HWLOC_PCI_PRIMARY_BUS] != pattr->bus) { + /* Sometimes the config space contains 00 instead of the actual primary bus number. + * Always trust the bus ID because it was built by the system which has more information + * to workaround such problems (e.g. ACPI information about PCI parent/children). + */ + hwloc_debug(" %04x:%02x:%02x.%01x bridge with (ignored) invalid PCI_PRIMARY_BUS %02x\n", + pattr->domain, pattr->bus, pattr->dev, pattr->func, config[HWLOC_PCI_PRIMARY_BUS]); + } + + battr->upstream_type = HWLOC_OBJ_BRIDGE_PCI; + battr->downstream_type = HWLOC_OBJ_BRIDGE_PCI; + battr->downstream.pci.domain = pattr->domain; + battr->downstream.pci.secondary_bus = config[HWLOC_PCI_SECONDARY_BUS]; + battr->downstream.pci.subordinate_bus = config[HWLOC_PCI_SUBORDINATE_BUS]; + + if (battr->downstream.pci.secondary_bus <= pattr->bus + || battr->downstream.pci.subordinate_bus <= pattr->bus + || battr->downstream.pci.secondary_bus > battr->downstream.pci.subordinate_bus) { + /* This should catch most cases of invalid bridge information + * (e.g. 00 for secondary and subordinate). + * Ideally we would also check that [secondary-subordinate] is included + * in the parent bridge [secondary+1:subordinate]. But that's hard to do + * because objects may be discovered out of order (especially in the fsroot case). + */ + hwloc_debug(" %04x:%02x:%02x.%01x bridge has invalid secondary-subordinate buses [%02x-%02x]\n", + pattr->domain, pattr->bus, pattr->dev, pattr->func, + battr->downstream.pci.secondary_bus, battr->downstream.pci.subordinate_bus); + hwloc_free_unlinked_object(obj); + return -1; + } + + return 0; +} + +const char * +hwloc_pci_class_string(unsigned short class_id) +{ + /* See https://pci-ids.ucw.cz/read/PD/ */ + switch ((class_id & 0xff00) >> 8) { + case 0x00: + switch (class_id) { + case 0x0001: return "VGA"; + } + break; + case 0x01: + switch (class_id) { + case 0x0100: return "SCSI"; + case 0x0101: return "IDE"; + case 0x0102: return "Floppy"; + case 0x0103: return "IPI"; + case 0x0104: return "RAID"; + case 0x0105: return "ATA"; + case 0x0106: return "SATA"; + case 0x0107: return "SAS"; + case 0x0108: return "NVMExp"; + } + return "Storage"; + case 0x02: + switch (class_id) { + case 0x0200: return "Ethernet"; + case 0x0201: return "TokenRing"; + case 0x0202: return "FDDI"; + case 0x0203: return "ATM"; + case 0x0204: return "ISDN"; + case 0x0205: return "WorldFip"; + case 0x0206: return "PICMG"; + case 0x0207: return "InfiniBand"; + case 0x0208: return "Fabric"; + } + return "Network"; + case 0x03: + switch (class_id) { + case 0x0300: return "VGA"; + case 0x0301: return "XGA"; + case 0x0302: return "3D"; + } + return "Display"; + case 0x04: + switch (class_id) { + case 0x0400: return "MultimediaVideo"; + case 0x0401: return "MultimediaAudio"; + case 0x0402: return "Telephony"; + case 0x0403: return "AudioDevice"; + } + return "Multimedia"; + case 0x05: + switch (class_id) { + case 0x0500: return "RAM"; + case 0x0501: return "Flash"; + } + return "Memory"; + case 0x06: + switch (class_id) { + case 0x0600: return "HostBridge"; + case 0x0601: return "ISABridge"; + case 0x0602: return "EISABridge"; + case 0x0603: return "MicroChannelBridge"; + case 0x0604: return "PCIBridge"; + case 0x0605: return "PCMCIABridge"; + case 0x0606: return "NubusBridge"; + case 0x0607: return "CardBusBridge"; + case 0x0608: return "RACEwayBridge"; + case 0x0609: return "SemiTransparentPCIBridge"; + case 0x060a: return "InfiniBandPCIHostBridge"; + } + return "Bridge"; + case 0x07: + switch (class_id) { + case 0x0700: return "Serial"; + case 0x0701: return "Parallel"; + case 0x0702: return "MultiportSerial"; + case 0x0703: return "Model"; + case 0x0704: return "GPIB"; + case 0x0705: return "SmartCard"; + } + return "Communication"; + case 0x08: + switch (class_id) { + case 0x0800: return "PIC"; + case 0x0801: return "DMA"; + case 0x0802: return "Timer"; + case 0x0803: return "RTC"; + case 0x0804: return "PCIHotPlug"; + case 0x0805: return "SDHost"; + case 0x0806: return "IOMMU"; + } + return "SystemPeripheral"; + case 0x09: + switch (class_id) { + case 0x0900: return "Keyboard"; + case 0x0901: return "DigitizerPen"; + case 0x0902: return "Mouse"; + case 0x0903: return "Scanern"; + case 0x0904: return "Gameport"; + } + return "Input"; + case 0x0a: + return "DockingStation"; + case 0x0b: + switch (class_id) { + case 0x0b00: return "386"; + case 0x0b01: return "486"; + case 0x0b02: return "Pentium"; +/* 0x0b03 and 0x0b04 might be Pentium and P6 ? */ + case 0x0b10: return "Alpha"; + case 0x0b20: return "PowerPC"; + case 0x0b30: return "MIPS"; + case 0x0b40: return "Co-Processor"; + } + return "Processor"; + case 0x0c: + switch (class_id) { + case 0x0c00: return "FireWire"; + case 0x0c01: return "ACCESS"; + case 0x0c02: return "SSA"; + case 0x0c03: return "USB"; + case 0x0c04: return "FibreChannel"; + case 0x0c05: return "SMBus"; + case 0x0c06: return "InfiniBand"; + case 0x0c07: return "IPMI-SMIC"; + case 0x0c08: return "SERCOS"; + case 0x0c09: return "CANBUS"; + } + return "SerialBus"; + case 0x0d: + switch (class_id) { + case 0x0d00: return "IRDA"; + case 0x0d01: return "ConsumerIR"; + case 0x0d10: return "RF"; + case 0x0d11: return "Bluetooth"; + case 0x0d12: return "Broadband"; + case 0x0d20: return "802.1a"; + case 0x0d21: return "802.1b"; + } + return "Wireless"; + case 0x0e: + switch (class_id) { + case 0x0e00: return "I2O"; + } + return "Intelligent"; + case 0x0f: + return "Satellite"; + case 0x10: + return "Encryption"; + case 0x11: + return "SignalProcessing"; + case 0x12: + return "ProcessingAccelerator"; + case 0x13: + return "Instrumentation"; + case 0x40: + return "Co-Processor"; + } + return "Other"; +} diff --git a/src/3rdparty/hwloc/src/shmem.c b/src/3rdparty/hwloc/src/shmem.c new file mode 100644 index 000000000..6c507f522 --- /dev/null +++ b/src/3rdparty/hwloc/src/shmem.c @@ -0,0 +1,287 @@ +/* + * Copyright © 2017-2018 Inria. All rights reserved. + * See COPYING in top-level directory. + */ + +#include +#include +#include +#include + +#ifndef HWLOC_WIN_SYS + +#include +#ifdef HAVE_UNISTD_H +#include +#endif +#include + +#define HWLOC_SHMEM_HEADER_VERSION 1 + +struct hwloc_shmem_header { + uint32_t header_version; /* sanity check */ + uint32_t header_length; /* where the actual topology starts in the file/mapping */ + uint64_t mmap_address; /* virtual address to pass to mmap */ + uint64_t mmap_length; /* length to pass to mmap (includes the header) */ +}; + +#define HWLOC_SHMEM_MALLOC_ALIGN 8UL + +static void * +tma_shmem_malloc(struct hwloc_tma * tma, + size_t length) +{ + void *current = tma->data; + tma->data = (char*)tma->data + ((length + HWLOC_SHMEM_MALLOC_ALIGN - 1) & ~(HWLOC_SHMEM_MALLOC_ALIGN - 1)); + return current; + +} + +static void * +tma_get_length_malloc(struct hwloc_tma * tma, + size_t length) +{ + size_t *tma_length = tma->data; + *tma_length += (length + HWLOC_SHMEM_MALLOC_ALIGN - 1) & ~(HWLOC_SHMEM_MALLOC_ALIGN - 1); + return malloc(length); + +} + +int +hwloc_shmem_topology_get_length(hwloc_topology_t topology, + size_t *lengthp, + unsigned long flags) +{ + hwloc_topology_t new; + struct hwloc_tma tma; + size_t length = 0; + unsigned long pagesize = hwloc_getpagesize(); /* round-up to full page for mmap() */ + int err; + + if (flags) { + errno = EINVAL; + return -1; + } + + tma.malloc = tma_get_length_malloc; + tma.dontfree = 0; + tma.data = &length; + + err = hwloc__topology_dup(&new, topology, &tma); + if (err < 0) + return err; + hwloc_topology_destroy(new); + + *lengthp = (sizeof(struct hwloc_shmem_header) + length + pagesize - 1) & ~(pagesize - 1); + return 0; +} + +int +hwloc_shmem_topology_write(hwloc_topology_t topology, + int fd, hwloc_uint64_t fileoffset, + void *mmap_address, size_t length, + unsigned long flags) +{ + hwloc_topology_t new; + struct hwloc_tma tma; + struct hwloc_shmem_header header; + void *mmap_res; + int err; + + if (flags) { + errno = EINVAL; + return -1; + } + + /* refresh old topology distances so that we don't uselessly duplicate invalid distances + * without being able to free() them. + */ + hwloc_internal_distances_refresh(topology); + + header.header_version = HWLOC_SHMEM_HEADER_VERSION; + header.header_length = sizeof(header); + header.mmap_address = (uintptr_t) mmap_address; + header.mmap_length = length; + + err = lseek(fd, fileoffset, SEEK_SET); + if (err < 0) + return -1; + + err = write(fd, &header, sizeof(header)); + if (err != sizeof(header)) + return -1; + + err = ftruncate(fd, fileoffset + length); + if (err < 0) + return -1; + + mmap_res = mmap(mmap_address, length, PROT_READ|PROT_WRITE, MAP_SHARED, fd, fileoffset); + if (mmap_res == MAP_FAILED) + return -1; + if (mmap_res != mmap_address) { + munmap(mmap_res, length); + errno = EBUSY; + return -1; + } + + tma.malloc = tma_shmem_malloc; + tma.dontfree = 1; + tma.data = (char *)mmap_res + sizeof(header); + err = hwloc__topology_dup(&new, topology, &tma); + if (err < 0) + return err; + assert((char*)new == (char*)mmap_address + sizeof(header)); + + assert((char *)mmap_res <= (char *)mmap_address + length); + + /* now refresh the new distances so that adopters can use them without refreshing the R/O shmem mapping */ + hwloc_internal_distances_refresh(new); + + /* topology is saved, release resources now */ + munmap(mmap_address, length); + hwloc_components_fini(); + + return 0; +} + +int +hwloc_shmem_topology_adopt(hwloc_topology_t *topologyp, + int fd, hwloc_uint64_t fileoffset, + void *mmap_address, size_t length, + unsigned long flags) +{ + hwloc_topology_t new, old; + struct hwloc_shmem_header header; + void *mmap_res; + int err; + + if (flags) { + errno = EINVAL; + return -1; + } + + err = lseek(fd, fileoffset, SEEK_SET); + if (err < 0) + return -1; + + err = read(fd, &header, sizeof(header)); + if (err != sizeof(header)) + return -1; + + if (header.header_version != HWLOC_SHMEM_HEADER_VERSION + || header.header_length != sizeof(header) + || header.mmap_address != (uintptr_t) mmap_address + || header.mmap_length != length) { + errno = EINVAL; + return -1; + } + + mmap_res = mmap(mmap_address, length, PROT_READ, MAP_SHARED, fd, fileoffset); + if (mmap_res == MAP_FAILED) + return -1; + if (mmap_res != mmap_address) { + errno = EBUSY; + goto out_with_mmap; + } + + old = (hwloc_topology_t)((char*)mmap_address + sizeof(header)); + if (hwloc_topology_abi_check(old) < 0) { + errno = EINVAL; + goto out_with_mmap; + } + + /* enforced by dup() inside shmem_topology_write() */ + assert(old->is_loaded); + assert(old->backends == NULL); + assert(old->get_pci_busid_cpuset_backend == NULL); + + hwloc_components_init(); + + /* duplicate the topology object so that we ca change use local binding_hooks + * (those are likely not mapped at the same location in both processes). + */ + new = malloc(sizeof(struct hwloc_topology)); + if (!new) + goto out_with_components; + memcpy(new, old, sizeof(*old)); + new->tma = NULL; + new->adopted_shmem_addr = mmap_address; + new->adopted_shmem_length = length; + new->topology_abi = HWLOC_TOPOLOGY_ABI; + /* setting binding hooks will touch support arrays, so duplicate them too. + * could avoid that by requesting a R/W mmap + */ + new->support.discovery = malloc(sizeof(*new->support.discovery)); + new->support.cpubind = malloc(sizeof(*new->support.cpubind)); + new->support.membind = malloc(sizeof(*new->support.membind)); + memcpy(new->support.discovery, old->support.discovery, sizeof(*new->support.discovery)); + memcpy(new->support.cpubind, old->support.cpubind, sizeof(*new->support.cpubind)); + memcpy(new->support.membind, old->support.membind, sizeof(*new->support.membind)); + hwloc_set_binding_hooks(new); + /* clear userdata callbacks pointing to the writer process' functions */ + new->userdata_export_cb = NULL; + new->userdata_import_cb = NULL; + +#ifndef HWLOC_DEBUG + if (getenv("HWLOC_DEBUG_CHECK")) +#endif + hwloc_topology_check(new); + + *topologyp = new; + return 0; + + out_with_components: + hwloc_components_fini(); + out_with_mmap: + munmap(mmap_res, length); + return -1; +} + +void +hwloc__topology_disadopt(hwloc_topology_t topology) +{ + hwloc_components_fini(); + munmap(topology->adopted_shmem_addr, topology->adopted_shmem_length); + free(topology->support.discovery); + free(topology->support.cpubind); + free(topology->support.membind); + free(topology); +} + +#else /* HWLOC_WIN_SYS */ + +int +hwloc_shmem_topology_get_length(hwloc_topology_t topology __hwloc_attribute_unused, + size_t *lengthp __hwloc_attribute_unused, + unsigned long flags __hwloc_attribute_unused) +{ + errno = ENOSYS; + return -1; +} + +int +hwloc_shmem_topology_write(hwloc_topology_t topology __hwloc_attribute_unused, + int fd __hwloc_attribute_unused, hwloc_uint64_t fileoffset __hwloc_attribute_unused, + void *mmap_address __hwloc_attribute_unused, size_t length __hwloc_attribute_unused, + unsigned long flags __hwloc_attribute_unused) +{ + errno = ENOSYS; + return -1; +} + +int +hwloc_shmem_topology_adopt(hwloc_topology_t *topologyp __hwloc_attribute_unused, + int fd __hwloc_attribute_unused, hwloc_uint64_t fileoffset __hwloc_attribute_unused, + void *mmap_address __hwloc_attribute_unused, size_t length __hwloc_attribute_unused, + unsigned long flags __hwloc_attribute_unused) +{ + errno = ENOSYS; + return -1; +} + +void +hwloc__topology_disadopt(hwloc_topology_t topology __hwloc_attribute_unused) +{ +} + +#endif /* HWLOC_WIN_SYS */ diff --git a/src/3rdparty/hwloc/src/static-components.h b/src/3rdparty/hwloc/src/static-components.h new file mode 100644 index 000000000..dac227a60 --- /dev/null +++ b/src/3rdparty/hwloc/src/static-components.h @@ -0,0 +1,15 @@ +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_noos_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_xml_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_synthetic_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_xml_nolibxml_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_windows_component; +HWLOC_DECLSPEC extern const struct hwloc_component hwloc_x86_component; +static const struct hwloc_component * hwloc_static_components[] = { + &hwloc_noos_component, + &hwloc_xml_component, + &hwloc_synthetic_component, + &hwloc_xml_nolibxml_component, + &hwloc_windows_component, + &hwloc_x86_component, + NULL +}; diff --git a/src/3rdparty/hwloc/src/topology-noos.c b/src/3rdparty/hwloc/src/topology-noos.c new file mode 100644 index 000000000..77871eb17 --- /dev/null +++ b/src/3rdparty/hwloc/src/topology-noos.c @@ -0,0 +1,65 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2017 Inria. All rights reserved. + * Copyright © 2009-2012 Université Bordeaux + * Copyright © 2009-2011 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +#include +#include +#include + +static int +hwloc_look_noos(struct hwloc_backend *backend) +{ + struct hwloc_topology *topology = backend->topology; + int nbprocs; + + if (topology->levels[0][0]->cpuset) + /* somebody discovered things */ + return -1; + + nbprocs = hwloc_fallback_nbprocessors(topology); + if (nbprocs >= 1) + topology->support.discovery->pu = 1; + else + nbprocs = 1; + + hwloc_alloc_root_sets(topology->levels[0][0]); + hwloc_setup_pu_level(topology, nbprocs); + hwloc_add_uname_info(topology, NULL); + return 0; +} + +static struct hwloc_backend * +hwloc_noos_component_instantiate(struct hwloc_disc_component *component, + const void *_data1 __hwloc_attribute_unused, + const void *_data2 __hwloc_attribute_unused, + const void *_data3 __hwloc_attribute_unused) +{ + struct hwloc_backend *backend; + backend = hwloc_backend_alloc(component); + if (!backend) + return NULL; + backend->discover = hwloc_look_noos; + return backend; +} + +static struct hwloc_disc_component hwloc_noos_disc_component = { + HWLOC_DISC_COMPONENT_TYPE_CPU, + "no_os", + HWLOC_DISC_COMPONENT_TYPE_GLOBAL, + hwloc_noos_component_instantiate, + 40, /* lower than native OS component, higher than globals */ + 1, + NULL +}; + +const struct hwloc_component hwloc_noos_component = { + HWLOC_COMPONENT_ABI, + NULL, NULL, + HWLOC_COMPONENT_TYPE_DISC, + 0, + &hwloc_noos_disc_component +}; diff --git a/src/3rdparty/hwloc/src/topology-synthetic.c b/src/3rdparty/hwloc/src/topology-synthetic.c new file mode 100644 index 000000000..1fe334d1c --- /dev/null +++ b/src/3rdparty/hwloc/src/topology-synthetic.c @@ -0,0 +1,1521 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2019 Inria. All rights reserved. + * Copyright © 2009-2010 Université Bordeaux + * Copyright © 2009-2011 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +#include +#include +#include +#include +#include + +#include +#include +#ifdef HAVE_STRINGS_H +#include +#endif + +struct hwloc_synthetic_attr_s { + hwloc_obj_type_t type; + unsigned depth; /* For caches/groups */ + hwloc_obj_cache_type_t cachetype; /* For caches */ + hwloc_uint64_t memorysize; /* For caches/memory */ +}; + +struct hwloc_synthetic_indexes_s { + /* the indexes= attribute before parsing */ + const char *string; + unsigned long string_length; + /* the array of explicit indexes after parsing */ + unsigned *array; + + /* used while filling the topology */ + unsigned next; /* id of the next object for that level */ +}; + +struct hwloc_synthetic_level_data_s { + unsigned arity; + unsigned long totalwidth; + + struct hwloc_synthetic_attr_s attr; + struct hwloc_synthetic_indexes_s indexes; + + struct hwloc_synthetic_attached_s { + struct hwloc_synthetic_attr_s attr; + + struct hwloc_synthetic_attached_s *next; + } *attached; +}; + +struct hwloc_synthetic_backend_data_s { + /* synthetic backend parameters */ + char *string; + + unsigned long numa_attached_nr; + struct hwloc_synthetic_indexes_s numa_attached_indexes; + +#define HWLOC_SYNTHETIC_MAX_DEPTH 128 + struct hwloc_synthetic_level_data_s level[HWLOC_SYNTHETIC_MAX_DEPTH]; +}; + +struct hwloc_synthetic_intlv_loop_s { + unsigned step; + unsigned nb; + unsigned level_depth; +}; + +static void +hwloc_synthetic_process_indexes(struct hwloc_synthetic_backend_data_s *data, + struct hwloc_synthetic_indexes_s *indexes, + unsigned long total, + int verbose) +{ + const char *attr = indexes->string; + unsigned long length = indexes->string_length; + unsigned *array = NULL; + size_t i; + + if (!attr) + return; + + array = calloc(total, sizeof(*array)); + if (!array) { + if (verbose) + fprintf(stderr, "Failed to allocate synthetic index array of size %lu\n", total); + goto out; + } + + i = strspn(attr, "0123456789,"); + if (i == length) { + /* explicit array of indexes */ + + for(i=0; iarray = array; + + } else { + /* interleaving */ + unsigned nr_loops = 1, cur_loop; + unsigned minstep = total; + unsigned long nbs = 1; + unsigned j, mul; + const char *tmp; + + tmp = attr; + while (tmp) { + tmp = strchr(tmp, ':'); + if (!tmp || tmp >= attr+length) + break; + nr_loops++; + tmp++; + } + + { + /* nr_loops colon-separated fields, but we may need one more at the end */ + HWLOC_VLA(struct hwloc_synthetic_intlv_loop_s, loops, nr_loops+1); + + if (*attr >= '0' && *attr <= '9') { + /* interleaving as x*y:z*t:... */ + unsigned step, nb; + + tmp = attr; + cur_loop = 0; + while (tmp) { + char *tmp2, *tmp3; + step = (unsigned) strtol(tmp, &tmp2, 0); + if (tmp2 == tmp || *tmp2 != '*') { + if (verbose) + fprintf(stderr, "Failed to read synthetic index interleaving loop '%s' without number before '*'\n", tmp); + goto out_with_array; + } + if (!step) { + if (verbose) + fprintf(stderr, "Invalid interleaving loop with step 0 at '%s'\n", tmp); + goto out_with_array; + } + tmp2++; + nb = (unsigned) strtol(tmp2, &tmp3, 0); + if (tmp3 == tmp2 || (*tmp3 && *tmp3 != ':' && *tmp3 != ')' && *tmp3 != ' ')) { + if (verbose) + fprintf(stderr, "Failed to read synthetic index interleaving loop '%s' without number between '*' and ':'\n", tmp); + goto out_with_array; + } + if (!nb) { + if (verbose) + fprintf(stderr, "Invalid interleaving loop with number 0 at '%s'\n", tmp2); + goto out_with_array; + } + loops[cur_loop].step = step; + loops[cur_loop].nb = nb; + if (step < minstep) + minstep = step; + nbs *= nb; + cur_loop++; + if (*tmp3 == ')' || *tmp3 == ' ') + break; + tmp = (const char*) (tmp3+1); + } + + } else { + /* interleaving as type1:type2:... */ + hwloc_obj_type_t type; + union hwloc_obj_attr_u attrs; + int err; + + /* find level depths for each interleaving loop */ + tmp = attr; + cur_loop = 0; + while (tmp) { + err = hwloc_type_sscanf(tmp, &type, &attrs, sizeof(attrs)); + if (err < 0) { + if (verbose) + fprintf(stderr, "Failed to read synthetic index interleaving loop type '%s'\n", tmp); + goto out_with_array; + } + if (type == HWLOC_OBJ_MISC || type == HWLOC_OBJ_BRIDGE || type == HWLOC_OBJ_PCI_DEVICE || type == HWLOC_OBJ_OS_DEVICE) { + if (verbose) + fprintf(stderr, "Misc object type disallowed in synthetic index interleaving loop type '%s'\n", tmp); + goto out_with_array; + } + for(i=0; ; i++) { + if (!data->level[i].arity) { + loops[cur_loop].level_depth = (unsigned)-1; + break; + } + if (type != data->level[i].attr.type) + continue; + if (type == HWLOC_OBJ_GROUP + && attrs.group.depth != (unsigned) -1 + && attrs.group.depth != data->level[i].attr.depth) + continue; + loops[cur_loop].level_depth = (unsigned)i; + break; + } + if (loops[cur_loop].level_depth == (unsigned)-1) { + if (verbose) + fprintf(stderr, "Failed to find level for synthetic index interleaving loop type '%s'\n", + tmp); + goto out_with_array; + } + tmp = strchr(tmp, ':'); + if (!tmp || tmp > attr+length) + break; + tmp++; + cur_loop++; + } + + /* compute actual loop step/nb */ + for(cur_loop=0; cur_loop prevdepth) + prevdepth = loops[i].level_depth; + } + step = total / data->level[mydepth].totalwidth; /* number of objects below us */ + nb = data->level[mydepth].totalwidth / data->level[prevdepth].totalwidth; /* number of us within parent */ + + loops[cur_loop].step = step; + loops[cur_loop].nb = nb; + assert(nb); + assert(step); + if (step < minstep) + minstep = step; + nbs *= nb; + } + } + assert(nbs); + + if (nbs != total) { + /* one loop of total/nbs steps is missing, add it if it's just the smallest one */ + if (minstep == total/nbs) { + loops[nr_loops].step = 1; + loops[nr_loops].nb = total/nbs; + nr_loops++; + } else { + if (verbose) + fprintf(stderr, "Invalid index interleaving total width %lu instead of %lu\n", nbs, total); + goto out_with_array; + } + } + + /* generate the array of indexes */ + mul = 1; + for(i=0; i= total) { + if (verbose) + fprintf(stderr, "Invalid index interleaving generates out-of-range index %u\n", array[j]); + goto out_with_array; + } + if (!array[j] && j) { + if (verbose) + fprintf(stderr, "Invalid index interleaving generates duplicate index values\n"); + goto out_with_array; + } + } + + indexes->array = array; + } + } + + return; + + out_with_array: + free(array); + out: + return; +} + +static hwloc_uint64_t +hwloc_synthetic_parse_memory_attr(const char *attr, const char **endp) +{ + const char *endptr; + hwloc_uint64_t size; + size = strtoull(attr, (char **) &endptr, 0); + if (!hwloc_strncasecmp(endptr, "TB", 2)) { + size <<= 40; + endptr += 2; + } else if (!hwloc_strncasecmp(endptr, "GB", 2)) { + size <<= 30; + endptr += 2; + } else if (!hwloc_strncasecmp(endptr, "MB", 2)) { + size <<= 20; + endptr += 2; + } else if (!hwloc_strncasecmp(endptr, "kB", 2)) { + size <<= 10; + endptr += 2; + } + *endp = endptr; + return size; +} + +static int +hwloc_synthetic_parse_attrs(const char *attrs, const char **next_posp, + struct hwloc_synthetic_attr_s *sattr, + struct hwloc_synthetic_indexes_s *sind, + int verbose) +{ + hwloc_obj_type_t type = sattr->type; + const char *next_pos; + hwloc_uint64_t memorysize = 0; + const char *index_string = NULL; + size_t index_string_length = 0; + + next_pos = (const char *) strchr(attrs, ')'); + if (!next_pos) { + if (verbose) + fprintf(stderr, "Missing attribute closing bracket in synthetic string doesn't have a number of objects at '%s'\n", attrs); + errno = EINVAL; + return -1; + } + + while (')' != *attrs) { + int iscache = hwloc__obj_type_is_cache(type); + + if (iscache && !strncmp("size=", attrs, 5)) { + memorysize = hwloc_synthetic_parse_memory_attr(attrs+5, &attrs); + + } else if (!iscache && !strncmp("memory=", attrs, 7)) { + memorysize = hwloc_synthetic_parse_memory_attr(attrs+7, &attrs); + + } else if (!strncmp("indexes=", attrs, 8)) { + index_string = attrs+8; + attrs += 8; + index_string_length = strcspn(attrs, " )"); + attrs += index_string_length; + + } else { + if (verbose) + fprintf(stderr, "Unknown attribute at '%s'\n", attrs); + errno = EINVAL; + return -1; + } + + if (' ' == *attrs) + attrs++; + else if (')' != *attrs) { + if (verbose) + fprintf(stderr, "Missing parameter separator at '%s'\n", attrs); + errno = EINVAL; + return -1; + } + } + + sattr->memorysize = memorysize; + + if (index_string) { + if (sind->string && verbose) + fprintf(stderr, "Overwriting duplicate indexes attribute with last occurence\n"); + sind->string = index_string; + sind->string_length = (unsigned long)index_string_length; + } + + *next_posp = next_pos+1; + return 0; +} + +/* frees level until arity = 0 */ +static void +hwloc_synthetic_free_levels(struct hwloc_synthetic_backend_data_s *data) +{ + unsigned i; + for(i=0; ilevel[i]; + struct hwloc_synthetic_attached_s **pprev = &curlevel->attached; + while (*pprev) { + struct hwloc_synthetic_attached_s *cur = *pprev; + *pprev = cur->next; + free(cur); + } + free(curlevel->indexes.array); + if (!curlevel->arity) + break; + } + free(data->numa_attached_indexes.array); +} + +/* Read from description a series of integers describing a symmetrical + topology and update the hwloc_synthetic_backend_data_s accordingly. On + success, return zero. */ +static int +hwloc_backend_synthetic_init(struct hwloc_synthetic_backend_data_s *data, + const char *description) +{ + const char *pos, *next_pos; + unsigned long item, count; + unsigned i; + int type_count[HWLOC_OBJ_TYPE_MAX]; + unsigned unset; + int verbose = 0; + const char *env = getenv("HWLOC_SYNTHETIC_VERBOSE"); + int err; + unsigned long totalarity = 1; + + if (env) + verbose = atoi(env); + + data->numa_attached_nr = 0; + data->numa_attached_indexes.array = NULL; + + /* default values before we add root attributes */ + data->level[0].totalwidth = 1; + data->level[0].attr.type = HWLOC_OBJ_MACHINE; + data->level[0].indexes.string = NULL; + data->level[0].indexes.array = NULL; + data->level[0].attr.memorysize = 0; + data->level[0].attached = NULL; + type_count[HWLOC_OBJ_MACHINE] = 1; + if (*description == '(') { + err = hwloc_synthetic_parse_attrs(description+1, &description, &data->level[0].attr, &data->level[0].indexes, verbose); + if (err < 0) + return err; + } + + data->numa_attached_indexes.string = NULL; + data->numa_attached_indexes.array = NULL; + + for (pos = description, count = 1; *pos; pos = next_pos) { + hwloc_obj_type_t type = HWLOC_OBJ_TYPE_NONE; + union hwloc_obj_attr_u attrs; + + /* initialize parent arity to 0 so that the levels are not infinite */ + data->level[count-1].arity = 0; + + while (*pos == ' ') + pos++; + + if (!*pos) + break; + + if (*pos == '[') { + /* attached */ + struct hwloc_synthetic_attached_s *attached, **pprev; + char *attr; + + pos++; + + if (hwloc_type_sscanf(pos, &type, &attrs, sizeof(attrs)) < 0) { + if (verbose) + fprintf(stderr, "Synthetic string with unknown attached object type at '%s'\n", pos); + errno = EINVAL; + goto error; + } + if (type != HWLOC_OBJ_NUMANODE) { + if (verbose) + fprintf(stderr, "Synthetic string with disallowed attached object type at '%s'\n", pos); + errno = EINVAL; + goto error; + } + data->numa_attached_nr += data->level[count-1].totalwidth; + + attached = malloc(sizeof(*attached)); + if (attached) { + attached->attr.type = type; + attached->attr.memorysize = 0; + /* attached->attr.depth and .cachetype unused */ + attached->next = NULL; + pprev = &data->level[count-1].attached; + while (*pprev) + pprev = &((*pprev)->next); + *pprev = attached; + } + + next_pos = strchr(pos, ']'); + if (!next_pos) { + if (verbose) + fprintf(stderr,"Synthetic string doesn't have a closing `]' after attached object type at '%s'\n", pos); + errno = EINVAL; + goto error; + } + + attr = strchr(pos, '('); + if (attr && attr < next_pos && attached) { + const char *dummy; + err = hwloc_synthetic_parse_attrs(attr+1, &dummy, &attached->attr, &data->numa_attached_indexes, verbose); + if (err < 0) + goto error; + } + + next_pos++; + continue; + } + + /* normal level */ + + /* reset defaults */ + data->level[count].indexes.string = NULL; + data->level[count].indexes.array = NULL; + data->level[count].attached = NULL; + + if (*pos < '0' || *pos > '9') { + if (hwloc_type_sscanf(pos, &type, &attrs, sizeof(attrs)) < 0) { + if (!strncmp(pos, "Die", 3) || !strncmp(pos, "Tile", 4) || !strncmp(pos, "Module", 6)) { + type = HWLOC_OBJ_GROUP; + } else { + /* FIXME: allow generic "Cache" string? would require to deal with possibly duplicate cache levels */ + if (verbose) + fprintf(stderr, "Synthetic string with unknown object type at '%s'\n", pos); + errno = EINVAL; + goto error; + } + } + if (type == HWLOC_OBJ_MACHINE || type == HWLOC_OBJ_MISC || type == HWLOC_OBJ_BRIDGE || type == HWLOC_OBJ_PCI_DEVICE || type == HWLOC_OBJ_OS_DEVICE) { + if (verbose) + fprintf(stderr, "Synthetic string with disallowed object type at '%s'\n", pos); + errno = EINVAL; + goto error; + } + + next_pos = strchr(pos, ':'); + if (!next_pos) { + if (verbose) + fprintf(stderr,"Synthetic string doesn't have a `:' after object type at '%s'\n", pos); + errno = EINVAL; + goto error; + } + pos = next_pos + 1; + } + + data->level[count].attr.type = type; + data->level[count].attr.depth = (unsigned) -1; + data->level[count].attr.cachetype = (hwloc_obj_cache_type_t) -1; + if (hwloc__obj_type_is_cache(type)) { + /* these are always initialized */ + data->level[count].attr.depth = attrs.cache.depth; + data->level[count].attr.cachetype = attrs.cache.type; + } else if (type == HWLOC_OBJ_GROUP) { + /* could be -1 but will be set below */ + data->level[count].attr.depth = attrs.group.depth; + } + + /* number of normal children */ + item = strtoul(pos, (char **)&next_pos, 0); + if (next_pos == pos) { + if (verbose) + fprintf(stderr,"Synthetic string doesn't have a number of objects at '%s'\n", pos); + errno = EINVAL; + goto error; + } + if (!item) { + if (verbose) + fprintf(stderr,"Synthetic string with disallow 0 number of objects at '%s'\n", pos); + errno = EINVAL; + goto error; + } + + totalarity *= item; + data->level[count].totalwidth = totalarity; + data->level[count].indexes.string = NULL; + data->level[count].indexes.array = NULL; + data->level[count].attr.memorysize = 0; + if (*next_pos == '(') { + err = hwloc_synthetic_parse_attrs(next_pos+1, &next_pos, &data->level[count].attr, &data->level[count].indexes, verbose); + if (err < 0) + goto error; + } + + if (count + 1 >= HWLOC_SYNTHETIC_MAX_DEPTH) { + if (verbose) + fprintf(stderr,"Too many synthetic levels, max %d\n", HWLOC_SYNTHETIC_MAX_DEPTH); + errno = EINVAL; + goto error; + } + if (item > UINT_MAX) { + if (verbose) + fprintf(stderr,"Too big arity, max %u\n", UINT_MAX); + errno = EINVAL; + goto error; + } + + data->level[count-1].arity = (unsigned)item; + count++; + } + + if (data->level[count-1].attr.type != HWLOC_OBJ_TYPE_NONE && data->level[count-1].attr.type != HWLOC_OBJ_PU) { + if (verbose) + fprintf(stderr, "Synthetic string cannot use non-PU type for last level\n"); + errno = EINVAL; + return -1; + } + data->level[count-1].attr.type = HWLOC_OBJ_PU; + + for(i=HWLOC_OBJ_TYPE_MIN; i0; i--) { + hwloc_obj_type_t type = data->level[i].attr.type; + if (type != HWLOC_OBJ_TYPE_NONE) { + type_count[type]++; + } + } + + /* sanity checks */ + if (!type_count[HWLOC_OBJ_PU]) { + if (verbose) + fprintf(stderr, "Synthetic string missing ending number of PUs\n"); + errno = EINVAL; + return -1; + } else if (type_count[HWLOC_OBJ_PU] > 1) { + if (verbose) + fprintf(stderr, "Synthetic string cannot have several PU levels\n"); + errno = EINVAL; + return -1; + } + if (type_count[HWLOC_OBJ_PACKAGE] > 1) { + if (verbose) + fprintf(stderr, "Synthetic string cannot have several package levels\n"); + errno = EINVAL; + return -1; + } + if (type_count[HWLOC_OBJ_NUMANODE] > 1) { + if (verbose) + fprintf(stderr, "Synthetic string cannot have several NUMA node levels\n"); + errno = EINVAL; + return -1; + } + if (type_count[HWLOC_OBJ_NUMANODE] && data->numa_attached_nr) { + if (verbose) + fprintf(stderr,"Synthetic string cannot have NUMA nodes both as a level and attached\n"); + errno = EINVAL; + return -1; + } + if (type_count[HWLOC_OBJ_CORE] > 1) { + if (verbose) + fprintf(stderr, "Synthetic string cannot have several core levels\n"); + errno = EINVAL; + return -1; + } + + /* deal with missing intermediate levels */ + unset = 0; + for(i=1; ilevel[i].attr.type == HWLOC_OBJ_TYPE_NONE) + unset++; + } + if (unset && unset != count-2) { + if (verbose) + fprintf(stderr, "Synthetic string cannot mix unspecified and specified types for levels\n"); + errno = EINVAL; + return -1; + } + if (unset) { + /* we want in priority: numa, package, core, up to 3 caches, groups */ + unsigned _count = count; + unsigned neednuma = 0; + unsigned needpack = 0; + unsigned needcore = 0; + unsigned needcaches = 0; + unsigned needgroups = 0; + /* 2 levels for machine and PU */ + _count -= 2; + + neednuma = (_count >= 1 && !data->numa_attached_nr); + _count -= neednuma; + + needpack = (_count >= 1); + _count -= needpack; + + needcore = (_count >= 1); + _count -= needcore; + + needcaches = (_count > 4 ? 4 : _count); + _count -= needcaches; + + needgroups = _count; + + /* we place them in order: groups, package, numa, caches, core */ + for(i = 0; i < needgroups; i++) { + unsigned depth = 1 + i; + data->level[depth].attr.type = HWLOC_OBJ_GROUP; + type_count[HWLOC_OBJ_GROUP]++; + } + if (needpack) { + unsigned depth = 1 + needgroups; + data->level[depth].attr.type = HWLOC_OBJ_PACKAGE; + type_count[HWLOC_OBJ_PACKAGE] = 1; + } + if (neednuma) { + unsigned depth = 1 + needgroups + needpack; + data->level[depth].attr.type = HWLOC_OBJ_NUMANODE; + type_count[HWLOC_OBJ_NUMANODE] = 1; + } + if (needcaches) { + /* priority: l2, l1, l3, l1i */ + /* order: l3, l2, l1, l1i */ + unsigned l3depth = 1 + needgroups + needpack + neednuma; + unsigned l2depth = l3depth + (needcaches >= 3); + unsigned l1depth = l2depth + 1; + unsigned l1idepth = l1depth + 1; + if (needcaches >= 3) { + data->level[l3depth].attr.type = HWLOC_OBJ_L3CACHE; + data->level[l3depth].attr.depth = 3; + data->level[l3depth].attr.cachetype = HWLOC_OBJ_CACHE_UNIFIED; + type_count[HWLOC_OBJ_L3CACHE] = 1; + } + data->level[l2depth].attr.type = HWLOC_OBJ_L2CACHE; + data->level[l2depth].attr.depth = 2; + data->level[l2depth].attr.cachetype = HWLOC_OBJ_CACHE_UNIFIED; + type_count[HWLOC_OBJ_L2CACHE] = 1; + if (needcaches >= 2) { + data->level[l1depth].attr.type = HWLOC_OBJ_L1CACHE; + data->level[l1depth].attr.depth = 1; + data->level[l1depth].attr.cachetype = HWLOC_OBJ_CACHE_DATA; + type_count[HWLOC_OBJ_L1CACHE] = 1; + } + if (needcaches >= 4) { + data->level[l1idepth].attr.type = HWLOC_OBJ_L1ICACHE; + data->level[l1idepth].attr.depth = 1; + data->level[l1idepth].attr.cachetype = HWLOC_OBJ_CACHE_INSTRUCTION; + type_count[HWLOC_OBJ_L1ICACHE] = 1; + } + } + if (needcore) { + unsigned depth = 1 + needgroups + needpack + neednuma + needcaches; + data->level[depth].attr.type = HWLOC_OBJ_CORE; + type_count[HWLOC_OBJ_CORE] = 1; + } + } + + /* enforce a NUMA level */ + if (!type_count[HWLOC_OBJ_NUMANODE] && !data->numa_attached_nr) { + /* insert a NUMA level below the automatic machine root */ + if (verbose) + fprintf(stderr, "Inserting a NUMA level with a single object at depth 1\n"); + /* move existing levels by one */ + memmove(&data->level[2], &data->level[1], count*sizeof(struct hwloc_synthetic_level_data_s)); + data->level[1].attr.type = HWLOC_OBJ_NUMANODE; + data->level[1].indexes.string = NULL; + data->level[1].indexes.array = NULL; + data->level[1].attr.memorysize = 0; + data->level[1].totalwidth = data->level[0].totalwidth; + /* update arity to insert a single NUMA node per parent */ + data->level[1].arity = data->level[0].arity; + data->level[0].arity = 1; + count++; + } + + for (i=0; ilevel[i]; + hwloc_obj_type_t type = curlevel->attr.type; + + if (type == HWLOC_OBJ_GROUP) { + if (curlevel->attr.depth == (unsigned)-1) + curlevel->attr.depth = type_count[HWLOC_OBJ_GROUP]--; + + } else if (hwloc__obj_type_is_cache(type)) { + if (!curlevel->attr.memorysize) { + if (1 == curlevel->attr.depth) + /* 32Kb in L1 */ + curlevel->attr.memorysize = 32*1024; + else + /* *4 at each level, starting from 1MB for L2, unified */ + curlevel->attr.memorysize = 256ULL*1024 << (2*curlevel->attr.depth); + } + + } else if (type == HWLOC_OBJ_NUMANODE && !curlevel->attr.memorysize) { + /* 1GB in memory nodes. */ + curlevel->attr.memorysize = 1024*1024*1024; + } + + hwloc_synthetic_process_indexes(data, &data->level[i].indexes, data->level[i].totalwidth, verbose); + } + + hwloc_synthetic_process_indexes(data, &data->numa_attached_indexes, data->numa_attached_nr, verbose); + + data->string = strdup(description); + data->level[count-1].arity = 0; + return 0; + + error: + hwloc_synthetic_free_levels(data); + return -1; +} + +static void +hwloc_synthetic_set_attr(struct hwloc_synthetic_attr_s *sattr, + hwloc_obj_t obj) +{ + switch (obj->type) { + case HWLOC_OBJ_GROUP: + obj->attr->group.kind = HWLOC_GROUP_KIND_SYNTHETIC; + obj->attr->group.subkind = sattr->depth-1; + break; + case HWLOC_OBJ_MACHINE: + break; + case HWLOC_OBJ_NUMANODE: + obj->attr->numanode.local_memory = sattr->memorysize; + obj->attr->numanode.page_types_len = 1; + obj->attr->numanode.page_types = malloc(sizeof(*obj->attr->numanode.page_types)); + memset(obj->attr->numanode.page_types, 0, sizeof(*obj->attr->numanode.page_types)); + obj->attr->numanode.page_types[0].size = 4096; + obj->attr->numanode.page_types[0].count = sattr->memorysize / 4096; + break; + case HWLOC_OBJ_PACKAGE: + break; + case HWLOC_OBJ_L1CACHE: + case HWLOC_OBJ_L2CACHE: + case HWLOC_OBJ_L3CACHE: + case HWLOC_OBJ_L4CACHE: + case HWLOC_OBJ_L5CACHE: + case HWLOC_OBJ_L1ICACHE: + case HWLOC_OBJ_L2ICACHE: + case HWLOC_OBJ_L3ICACHE: + obj->attr->cache.depth = sattr->depth; + obj->attr->cache.linesize = 64; + obj->attr->cache.type = sattr->cachetype; + obj->attr->cache.size = sattr->memorysize; + break; + case HWLOC_OBJ_CORE: + break; + case HWLOC_OBJ_PU: + break; + default: + /* Should never happen */ + assert(0); + break; + } +} + +static unsigned +hwloc_synthetic_next_index(struct hwloc_synthetic_indexes_s *indexes, hwloc_obj_type_t type) +{ + unsigned os_index = indexes->next++; + + if (indexes->array) + os_index = indexes->array[os_index]; + else if (hwloc__obj_type_is_cache(type) || type == HWLOC_OBJ_GROUP) + /* don't enforce useless os_indexes for Caches and Groups */ + os_index = HWLOC_UNKNOWN_INDEX; + + return os_index; +} + +static void +hwloc_synthetic_insert_attached(struct hwloc_topology *topology, + struct hwloc_synthetic_backend_data_s *data, + struct hwloc_synthetic_attached_s *attached, + hwloc_bitmap_t set) +{ + hwloc_obj_t child; + unsigned attached_os_index; + + if (!attached) + return; + + assert(attached->attr.type == HWLOC_OBJ_NUMANODE); + + attached_os_index = hwloc_synthetic_next_index(&data->numa_attached_indexes, HWLOC_OBJ_NUMANODE); + + child = hwloc_alloc_setup_object(topology, attached->attr.type, attached_os_index); + child->cpuset = hwloc_bitmap_dup(set); + + child->nodeset = hwloc_bitmap_alloc(); + hwloc_bitmap_set(child->nodeset, attached_os_index); + + hwloc_synthetic_set_attr(&attached->attr, child); + + hwloc_insert_object_by_cpuset(topology, child); + + hwloc_synthetic_insert_attached(topology, data, attached->next, set); +} + +/* + * Recursively build objects whose cpu start at first_cpu + * - level gives where to look in the type, arity and id arrays + * - the id array is used as a variable to get unique IDs for a given level. + * - generated memory should be added to *memory_kB. + * - generated cpus should be added to parent_cpuset. + * - next cpu number to be used should be returned. + */ +static void +hwloc__look_synthetic(struct hwloc_topology *topology, + struct hwloc_synthetic_backend_data_s *data, + int level, + hwloc_bitmap_t parent_cpuset) +{ + hwloc_obj_t obj; + unsigned i; + struct hwloc_synthetic_level_data_s *curlevel = &data->level[level]; + hwloc_obj_type_t type = curlevel->attr.type; + hwloc_bitmap_t set; + unsigned os_index; + + assert(hwloc__obj_type_is_normal(type) || type == HWLOC_OBJ_NUMANODE); + assert(type != HWLOC_OBJ_MACHINE); + + os_index = hwloc_synthetic_next_index(&curlevel->indexes, type); + + set = hwloc_bitmap_alloc(); + if (!curlevel->arity) { + hwloc_bitmap_set(set, os_index); + } else { + for (i = 0; i < curlevel->arity; i++) + hwloc__look_synthetic(topology, data, level + 1, set); + } + + hwloc_bitmap_or(parent_cpuset, parent_cpuset, set); + + if (hwloc_filter_check_keep_object_type(topology, type)) { + obj = hwloc_alloc_setup_object(topology, type, os_index); + obj->cpuset = hwloc_bitmap_dup(set); + + if (type == HWLOC_OBJ_NUMANODE) { + obj->nodeset = hwloc_bitmap_alloc(); + hwloc_bitmap_set(obj->nodeset, os_index); + } + + hwloc_synthetic_set_attr(&curlevel->attr, obj); + + hwloc_insert_object_by_cpuset(topology, obj); + } + + hwloc_synthetic_insert_attached(topology, data, curlevel->attached, set); + + hwloc_bitmap_free(set); +} + +static int +hwloc_look_synthetic(struct hwloc_backend *backend) +{ + struct hwloc_topology *topology = backend->topology; + struct hwloc_synthetic_backend_data_s *data = backend->private_data; + hwloc_bitmap_t cpuset = hwloc_bitmap_alloc(); + unsigned i; + + assert(!topology->levels[0][0]->cpuset); + + hwloc_alloc_root_sets(topology->levels[0][0]); + + topology->support.discovery->pu = 1; + topology->support.discovery->numa = 1; /* we add a single NUMA node if none is given */ + topology->support.discovery->numa_memory = 1; /* specified or default size */ + + /* start with os_index 0 for each level */ + for (i = 0; data->level[i].arity > 0; i++) + data->level[i].indexes.next = 0; + data->numa_attached_indexes.next = 0; + /* ... including the last one */ + data->level[i].indexes.next = 0; + + /* update first level type according to the synthetic type array */ + topology->levels[0][0]->type = data->level[0].attr.type; + hwloc_synthetic_set_attr(&data->level[0].attr, topology->levels[0][0]); + + for (i = 0; i < data->level[0].arity; i++) + hwloc__look_synthetic(topology, data, 1, cpuset); + + hwloc_synthetic_insert_attached(topology, data, data->level[0].attached, cpuset); + + hwloc_bitmap_free(cpuset); + + hwloc_obj_add_info(topology->levels[0][0], "Backend", "Synthetic"); + hwloc_obj_add_info(topology->levels[0][0], "SyntheticDescription", data->string); + return 0; +} + +static void +hwloc_synthetic_backend_disable(struct hwloc_backend *backend) +{ + struct hwloc_synthetic_backend_data_s *data = backend->private_data; + hwloc_synthetic_free_levels(data); + free(data->string); + free(data); +} + +static struct hwloc_backend * +hwloc_synthetic_component_instantiate(struct hwloc_disc_component *component, + const void *_data1, + const void *_data2 __hwloc_attribute_unused, + const void *_data3 __hwloc_attribute_unused) +{ + struct hwloc_backend *backend; + struct hwloc_synthetic_backend_data_s *data; + int err; + + if (!_data1) { + const char *env = getenv("HWLOC_SYNTHETIC"); + if (env) { + /* 'synthetic' was given in HWLOC_COMPONENTS without a description */ + _data1 = env; + } else { + errno = EINVAL; + goto out; + } + } + + backend = hwloc_backend_alloc(component); + if (!backend) + goto out; + + data = malloc(sizeof(*data)); + if (!data) { + errno = ENOMEM; + goto out_with_backend; + } + + err = hwloc_backend_synthetic_init(data, (const char *) _data1); + if (err < 0) + goto out_with_data; + + backend->private_data = data; + backend->discover = hwloc_look_synthetic; + backend->disable = hwloc_synthetic_backend_disable; + backend->is_thissystem = 0; + + return backend; + + out_with_data: + free(data); + out_with_backend: + free(backend); + out: + return NULL; +} + +static struct hwloc_disc_component hwloc_synthetic_disc_component = { + HWLOC_DISC_COMPONENT_TYPE_GLOBAL, + "synthetic", + ~0, + hwloc_synthetic_component_instantiate, + 30, + 1, + NULL +}; + +const struct hwloc_component hwloc_synthetic_component = { + HWLOC_COMPONENT_ABI, + NULL, NULL, + HWLOC_COMPONENT_TYPE_DISC, + 0, + &hwloc_synthetic_disc_component +}; + +static __hwloc_inline int +hwloc__export_synthetic_update_status(int *ret, char **tmp, ssize_t *tmplen, int res) +{ + if (res < 0) + return -1; + *ret += res; + if (res >= *tmplen) + res = *tmplen>0 ? (int)(*tmplen) - 1 : 0; + *tmp += res; + *tmplen -= res; + return 0; +} + +static __hwloc_inline void +hwloc__export_synthetic_add_char(int *ret, char **tmp, ssize_t *tmplen, char c) +{ + if (*tmplen > 1) { + (*tmp)[0] = c; + (*tmp)[1] = '\0'; + (*tmp)++; + (*tmplen)--; + } + (*ret)++; +} + +static int +hwloc__export_synthetic_indexes(hwloc_obj_t *level, unsigned total, + char *buffer, size_t buflen) +{ + unsigned step = 1; + unsigned nr_loops = 0; + struct hwloc_synthetic_intlv_loop_s *loops = NULL, *tmploops; + hwloc_obj_t cur; + unsigned i, j; + ssize_t tmplen = buflen; + char *tmp = buffer; + int res, ret = 0; + + /* must start with 0 */ + if (level[0]->os_index) + goto exportall; + + while (step != total) { + /* must be a divider of the total */ + if (total % step) + goto exportall; + + /* look for os_index == step */ + for(i=1; ios_index == step) + break; + if (i == total) + goto exportall; + for(j=2; jos_index != step*j) + break; + + nr_loops++; + tmploops = realloc(loops, nr_loops*sizeof(*loops)); + if (!tmploops) + goto exportall; + loops = tmploops; + loops[nr_loops-1].step = i; + loops[nr_loops-1].nb = j; + step *= j; + } + + /* check this interleaving */ + for(i=0; ios_index != ind) + goto exportall; + } + + /* success, print it */ + for(j=0; jos_index, + cur->next_cousin ? "," : ")"); + if (hwloc__export_synthetic_update_status(&ret, &tmp, &tmplen, res) < 0) + return -1; + cur = cur->next_cousin; + } + return ret; +} + +static int +hwloc__export_synthetic_obj_attr(struct hwloc_topology * topology, + hwloc_obj_t obj, + char *buffer, size_t buflen) +{ + const char * separator = " "; + const char * prefix = "("; + char cachesize[64] = ""; + char memsize[64] = ""; + int needindexes = 0; + + if (hwloc__obj_type_is_cache(obj->type) && obj->attr->cache.size) { + snprintf(cachesize, sizeof(cachesize), "%ssize=%llu", + prefix, (unsigned long long) obj->attr->cache.size); + prefix = separator; + } + if (obj->type == HWLOC_OBJ_NUMANODE && obj->attr->numanode.local_memory) { + snprintf(memsize, sizeof(memsize), "%smemory=%llu", + prefix, (unsigned long long) obj->attr->numanode.local_memory); + prefix = separator; + } + if (!obj->logical_index /* only display indexes once per level (not for non-first NUMA children, etc.) */ + && (obj->type == HWLOC_OBJ_PU || obj->type == HWLOC_OBJ_NUMANODE)) { + hwloc_obj_t cur = obj; + while (cur) { + if (cur->os_index != cur->logical_index) { + needindexes = 1; + break; + } + cur = cur->next_cousin; + } + } + if (*cachesize || *memsize || needindexes) { + ssize_t tmplen = buflen; + char *tmp = buffer; + int res, ret = 0; + + res = hwloc_snprintf(tmp, tmplen, "%s%s%s", cachesize, memsize, needindexes ? "" : ")"); + if (hwloc__export_synthetic_update_status(&ret, &tmp, &tmplen, res) < 0) + return -1; + + if (needindexes) { + unsigned total; + hwloc_obj_t *level; + + if (obj->depth < 0) { + assert(obj->depth == HWLOC_TYPE_DEPTH_NUMANODE); + total = topology->slevels[HWLOC_SLEVEL_NUMANODE].nbobjs; + level = topology->slevels[HWLOC_SLEVEL_NUMANODE].objs; + } else { + total = topology->level_nbobjects[obj->depth]; + level = topology->levels[obj->depth]; + } + + res = hwloc_snprintf(tmp, tmplen, "%sindexes=", prefix); + if (hwloc__export_synthetic_update_status(&ret, &tmp, &tmplen, res) < 0) + return -1; + + res = hwloc__export_synthetic_indexes(level, total, tmp, tmplen); + if (hwloc__export_synthetic_update_status(&ret, &tmp, &tmplen, res) < 0) + return -1; + } + return ret; + } else { + return 0; + } +} + +static int +hwloc__export_synthetic_obj(struct hwloc_topology * topology, unsigned long flags, + hwloc_obj_t obj, unsigned arity, + char *buffer, size_t buflen) +{ + char aritys[12] = ""; + ssize_t tmplen = buflen; + char *tmp = buffer; + int res, ret = 0; + + /* :, except for root */ + if (arity != (unsigned)-1) + snprintf(aritys, sizeof(aritys), ":%u", arity); + if (hwloc__obj_type_is_cache(obj->type) + && (flags & HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_NO_EXTENDED_TYPES)) { + /* v1 uses generic "Cache" for non-extended type name */ + res = hwloc_snprintf(tmp, tmplen, "Cache%s", aritys); + + } else if (obj->type == HWLOC_OBJ_PACKAGE + && (flags & (HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_NO_EXTENDED_TYPES + |HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_V1))) { + /* if exporting to v1 or without extended-types, use all-v1-compatible Socket name */ + res = hwloc_snprintf(tmp, tmplen, "Socket%s", aritys); + + } else if (obj->type == HWLOC_OBJ_GROUP /* don't export group depth */ + || flags & HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_NO_EXTENDED_TYPES) { + res = hwloc_snprintf(tmp, tmplen, "%s%s", hwloc_obj_type_string(obj->type), aritys); + } else { + char types[64]; + hwloc_obj_type_snprintf(types, sizeof(types), obj, 1); + res = hwloc_snprintf(tmp, tmplen, "%s%s", types, aritys); + } + if (hwloc__export_synthetic_update_status(&ret, &tmp, &tmplen, res) < 0) + return -1; + + if (!(flags & HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_NO_ATTRS)) { + /* obj attributes */ + res = hwloc__export_synthetic_obj_attr(topology, obj, tmp, tmplen); + if (hwloc__export_synthetic_update_status(&ret, &tmp, &tmplen, res) < 0) + return -1; + } + + return ret; +} + +static int +hwloc__export_synthetic_memory_children(struct hwloc_topology * topology, unsigned long flags, + hwloc_obj_t parent, + char *buffer, size_t buflen, + int needprefix, int verbose) +{ + hwloc_obj_t mchild; + ssize_t tmplen = buflen; + char *tmp = buffer; + int res, ret = 0; + + mchild = parent->memory_first_child; + if (!mchild) + return 0; + + if (flags & HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_V1) { + /* v1: export a single NUMA child */ + if (parent->memory_arity > 1 || mchild->type != HWLOC_OBJ_NUMANODE) { + /* not supported */ + if (verbose) + fprintf(stderr, "Cannot export to synthetic v1 if multiple memory children are attached to the same location.\n"); + errno = EINVAL; + return -1; + } + + if (needprefix) + hwloc__export_synthetic_add_char(&ret, &tmp, &tmplen, ' '); + + res = hwloc__export_synthetic_obj(topology, flags, mchild, 1, tmp, tmplen); + if (hwloc__export_synthetic_update_status(&ret, &tmp, &tmplen, res) < 0) + return -1; + return ret; + } + + while (mchild) { + /* v2: export all NUMA children */ + + assert(mchild->type == HWLOC_OBJ_NUMANODE); /* only NUMA node memory children for now */ + + if (needprefix) + hwloc__export_synthetic_add_char(&ret, &tmp, &tmplen, ' '); + + hwloc__export_synthetic_add_char(&ret, &tmp, &tmplen, '['); + + res = hwloc__export_synthetic_obj(topology, flags, mchild, (unsigned)-1, tmp, tmplen); + if (hwloc__export_synthetic_update_status(&ret, &tmp, &tmplen, res) < 0) + return -1; + + hwloc__export_synthetic_add_char(&ret, &tmp, &tmplen, ']'); + + needprefix = 1; + mchild = mchild->next_sibling; + } + + return ret; +} + +static int +hwloc_check_memory_symmetric(struct hwloc_topology * topology) +{ + hwloc_bitmap_t remaining_nodes; + + remaining_nodes = hwloc_bitmap_dup(hwloc_get_root_obj(topology)->nodeset); + if (!remaining_nodes) + /* assume asymmetric */ + return -1; + + while (!hwloc_bitmap_iszero(remaining_nodes)) { + unsigned idx; + hwloc_obj_t node; + hwloc_obj_t first_parent; + unsigned i; + + idx = hwloc_bitmap_first(remaining_nodes); + node = hwloc_get_numanode_obj_by_os_index(topology, idx); + assert(node); + + first_parent = node->parent; + assert(hwloc__obj_type_is_normal(first_parent->type)); /* only depth-1 memory children for now */ + + /* check whether all object on parent's level have same number of NUMA children */ + for(i=0; idepth); i++) { + hwloc_obj_t parent, mchild; + + parent = hwloc_get_obj_by_depth(topology, first_parent->depth, i); + assert(parent); + + /* must have same memory arity */ + if (parent->memory_arity != first_parent->memory_arity) + goto out_with_bitmap; + + /* clear these NUMA children from remaining_nodes */ + mchild = parent->memory_first_child; + while (mchild) { + assert(mchild->type == HWLOC_OBJ_NUMANODE); /* only NUMA node memory children for now */ + hwloc_bitmap_clr(remaining_nodes, mchild->os_index); /* cannot use parent->nodeset, some normal children may have other NUMA nodes */ + mchild = mchild->next_sibling; + } + } + } + + hwloc_bitmap_free(remaining_nodes); + return 0; + + out_with_bitmap: + hwloc_bitmap_free(remaining_nodes); + return -1; +} + +int +hwloc_topology_export_synthetic(struct hwloc_topology * topology, + char *buffer, size_t buflen, + unsigned long flags) +{ + hwloc_obj_t obj = hwloc_get_root_obj(topology); + ssize_t tmplen = buflen; + char *tmp = buffer; + int res, ret = 0; + unsigned arity; + int needprefix = 0; + int verbose = 0; + const char *env = getenv("HWLOC_SYNTHETIC_VERBOSE"); + + if (env) + verbose = atoi(env); + + if (!topology->is_loaded) { + errno = EINVAL; + return -1; + } + + if (flags & ~(HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_NO_EXTENDED_TYPES + |HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_NO_ATTRS + |HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_V1 + |HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_IGNORE_MEMORY)) { + errno = EINVAL; + return -1; + } + + /* TODO: add a flag to ignore symmetric_subtree and I/Os. + * just assume things are symmetric with the left branches of the tree. + * but the number of objects per level may be wrong, what to do with OS index array in this case? + * only allow ignoring symmetric_subtree if the level width remains OK? + */ + + /* TODO: add a root object by default, with a prefix such as tree= + * so that we can backward-compatibly recognize whether there's a root or not. + * and add a flag to disable it. + */ + + /* TODO: flag to force all indexes, not only for PU and NUMA? */ + + if (!obj->symmetric_subtree) { + if (verbose) + fprintf(stderr, "Cannot export to synthetic unless topology is symmetric (root->symmetric_subtree must be set).\n"); + errno = EINVAL; + return -1; + } + + if (!(flags & HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_IGNORE_MEMORY) + && hwloc_check_memory_symmetric(topology) < 0) { + if (verbose) + fprintf(stderr, "Cannot export to synthetic unless memory is attached symmetrically.\n"); + errno = EINVAL; + return -1; + } + + if (flags & HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_V1) { + /* v1 requires all NUMA at the same level */ + hwloc_obj_t node; + signed pdepth; + + node = hwloc_get_obj_by_type(topology, HWLOC_OBJ_NUMANODE, 0); + assert(hwloc__obj_type_is_normal(node->parent->type)); /* only depth-1 memory children for now */ + pdepth = node->parent->depth; + + while ((node = node->next_cousin) != NULL) { + assert(hwloc__obj_type_is_normal(node->parent->type)); /* only depth-1 memory children for now */ + if (node->parent->depth != pdepth) { + if (verbose) + fprintf(stderr, "Cannot export to synthetic v1 if memory is attached to parents at different depths.\n"); + errno = EINVAL; + return -1; + } + } + } + + /* we're good, start exporting */ + + if (!(flags & HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_NO_ATTRS)) { + /* obj attributes */ + res = hwloc__export_synthetic_obj_attr(topology, obj, tmp, tmplen); + if (res > 0) + needprefix = 1; + if (hwloc__export_synthetic_update_status(&ret, &tmp, &tmplen, res) < 0) + return -1; + } + + if (!(flags & HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_IGNORE_MEMORY)) { + res = hwloc__export_synthetic_memory_children(topology, flags, obj, tmp, tmplen, needprefix, verbose); + if (res > 0) + needprefix = 1; + if (hwloc__export_synthetic_update_status(&ret, &tmp, &tmplen, res) < 0) + return -1; + } + + arity = obj->arity; + while (arity) { + /* for each level */ + obj = obj->first_child; + + if (needprefix) + hwloc__export_synthetic_add_char(&ret, &tmp, &tmplen, ' '); + + res = hwloc__export_synthetic_obj(topology, flags, obj, arity, tmp, tmplen); + if (hwloc__export_synthetic_update_status(&ret, &tmp, &tmplen, res) < 0) + return -1; + + if (!(flags & HWLOC_TOPOLOGY_EXPORT_SYNTHETIC_FLAG_IGNORE_MEMORY)) { + res = hwloc__export_synthetic_memory_children(topology, flags, obj, tmp, tmplen, 1, verbose); + if (hwloc__export_synthetic_update_status(&ret, &tmp, &tmplen, res) < 0) + return -1; + } + + /* next level */ + needprefix = 1; + arity = obj->arity; + } + + return ret; +} diff --git a/src/3rdparty/hwloc/src/topology-windows.c b/src/3rdparty/hwloc/src/topology-windows.c new file mode 100644 index 000000000..d03645c0f --- /dev/null +++ b/src/3rdparty/hwloc/src/topology-windows.c @@ -0,0 +1,1189 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2018 Inria. All rights reserved. + * Copyright © 2009-2012 Université Bordeaux + * Copyright © 2011 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +/* To try to get all declarations duplicated below. */ +#define _WIN32_WINNT 0x0601 + +#include +#include +#include +#include + +#include + +#ifndef HAVE_KAFFINITY +typedef ULONG_PTR KAFFINITY, *PKAFFINITY; +#endif + +#ifndef HAVE_PROCESSOR_CACHE_TYPE +typedef enum _PROCESSOR_CACHE_TYPE { + CacheUnified, + CacheInstruction, + CacheData, + CacheTrace +} PROCESSOR_CACHE_TYPE; +#endif + +#ifndef CACHE_FULLY_ASSOCIATIVE +#define CACHE_FULLY_ASSOCIATIVE 0xFF +#endif + +#ifndef MAXIMUM_PROC_PER_GROUP /* missing in MinGW */ +#define MAXIMUM_PROC_PER_GROUP 64 +#endif + +#ifndef HAVE_CACHE_DESCRIPTOR +typedef struct _CACHE_DESCRIPTOR { + BYTE Level; + BYTE Associativity; + WORD LineSize; + DWORD Size; /* in bytes */ + PROCESSOR_CACHE_TYPE Type; +} CACHE_DESCRIPTOR, *PCACHE_DESCRIPTOR; +#endif + +#ifndef HAVE_LOGICAL_PROCESSOR_RELATIONSHIP +typedef enum _LOGICAL_PROCESSOR_RELATIONSHIP { + RelationProcessorCore, + RelationNumaNode, + RelationCache, + RelationProcessorPackage, + RelationGroup, + RelationAll = 0xffff +} LOGICAL_PROCESSOR_RELATIONSHIP; +#else /* HAVE_LOGICAL_PROCESSOR_RELATIONSHIP */ +# ifndef HAVE_RELATIONPROCESSORPACKAGE +# define RelationProcessorPackage 3 +# define RelationGroup 4 +# define RelationAll 0xffff +# endif /* HAVE_RELATIONPROCESSORPACKAGE */ +#endif /* HAVE_LOGICAL_PROCESSOR_RELATIONSHIP */ + +#ifndef HAVE_SYSTEM_LOGICAL_PROCESSOR_INFORMATION +typedef struct _SYSTEM_LOGICAL_PROCESSOR_INFORMATION { + ULONG_PTR ProcessorMask; + LOGICAL_PROCESSOR_RELATIONSHIP Relationship; + _ANONYMOUS_UNION + union { + struct { + BYTE flags; + } ProcessorCore; + struct { + DWORD NodeNumber; + } NumaNode; + CACHE_DESCRIPTOR Cache; + ULONGLONG Reserved[2]; + } DUMMYUNIONNAME; +} SYSTEM_LOGICAL_PROCESSOR_INFORMATION, *PSYSTEM_LOGICAL_PROCESSOR_INFORMATION; +#endif + +/* Extended interface, for group support */ + +#ifndef HAVE_GROUP_AFFINITY +typedef struct _GROUP_AFFINITY { + KAFFINITY Mask; + WORD Group; + WORD Reserved[3]; +} GROUP_AFFINITY, *PGROUP_AFFINITY; +#endif + +#ifndef HAVE_PROCESSOR_RELATIONSHIP +typedef struct _PROCESSOR_RELATIONSHIP { + BYTE Flags; + BYTE Reserved[21]; + WORD GroupCount; + GROUP_AFFINITY GroupMask[ANYSIZE_ARRAY]; +} PROCESSOR_RELATIONSHIP, *PPROCESSOR_RELATIONSHIP; +#endif + +#ifndef HAVE_NUMA_NODE_RELATIONSHIP +typedef struct _NUMA_NODE_RELATIONSHIP { + DWORD NodeNumber; + BYTE Reserved[20]; + GROUP_AFFINITY GroupMask; +} NUMA_NODE_RELATIONSHIP, *PNUMA_NODE_RELATIONSHIP; +#endif + +#ifndef HAVE_CACHE_RELATIONSHIP +typedef struct _CACHE_RELATIONSHIP { + BYTE Level; + BYTE Associativity; + WORD LineSize; + DWORD CacheSize; + PROCESSOR_CACHE_TYPE Type; + BYTE Reserved[20]; + GROUP_AFFINITY GroupMask; +} CACHE_RELATIONSHIP, *PCACHE_RELATIONSHIP; +#endif + +#ifndef HAVE_PROCESSOR_GROUP_INFO +typedef struct _PROCESSOR_GROUP_INFO { + BYTE MaximumProcessorCount; + BYTE ActiveProcessorCount; + BYTE Reserved[38]; + KAFFINITY ActiveProcessorMask; +} PROCESSOR_GROUP_INFO, *PPROCESSOR_GROUP_INFO; +#endif + +#ifndef HAVE_GROUP_RELATIONSHIP +typedef struct _GROUP_RELATIONSHIP { + WORD MaximumGroupCount; + WORD ActiveGroupCount; + ULONGLONG Reserved[2]; + PROCESSOR_GROUP_INFO GroupInfo[ANYSIZE_ARRAY]; +} GROUP_RELATIONSHIP, *PGROUP_RELATIONSHIP; +#endif + +#ifndef HAVE_SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX +typedef struct _SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX { + LOGICAL_PROCESSOR_RELATIONSHIP Relationship; + DWORD Size; + _ANONYMOUS_UNION + union { + PROCESSOR_RELATIONSHIP Processor; + NUMA_NODE_RELATIONSHIP NumaNode; + CACHE_RELATIONSHIP Cache; + GROUP_RELATIONSHIP Group; + /* Odd: no member to tell the cpu mask of the package... */ + } DUMMYUNIONNAME; +} SYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX, *PSYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX; +#endif + +#ifndef HAVE_PSAPI_WORKING_SET_EX_BLOCK +typedef union _PSAPI_WORKING_SET_EX_BLOCK { + ULONG_PTR Flags; + struct { + unsigned Valid :1; + unsigned ShareCount :3; + unsigned Win32Protection :11; + unsigned Shared :1; + unsigned Node :6; + unsigned Locked :1; + unsigned LargePage :1; + }; +} PSAPI_WORKING_SET_EX_BLOCK; +#endif + +#ifndef HAVE_PSAPI_WORKING_SET_EX_INFORMATION +typedef struct _PSAPI_WORKING_SET_EX_INFORMATION { + PVOID VirtualAddress; + PSAPI_WORKING_SET_EX_BLOCK VirtualAttributes; +} PSAPI_WORKING_SET_EX_INFORMATION; +#endif + +#ifndef HAVE_PROCESSOR_NUMBER +typedef struct _PROCESSOR_NUMBER { + WORD Group; + BYTE Number; + BYTE Reserved; +} PROCESSOR_NUMBER, *PPROCESSOR_NUMBER; +#endif + +/* Function pointers */ + +typedef WORD (WINAPI *PFN_GETACTIVEPROCESSORGROUPCOUNT)(void); +static PFN_GETACTIVEPROCESSORGROUPCOUNT GetActiveProcessorGroupCountProc; + +static unsigned long nr_processor_groups = 1; +static unsigned long max_numanode_index = 0; + +typedef WORD (WINAPI *PFN_GETACTIVEPROCESSORCOUNT)(WORD); +static PFN_GETACTIVEPROCESSORCOUNT GetActiveProcessorCountProc; + +typedef DWORD (WINAPI *PFN_GETCURRENTPROCESSORNUMBER)(void); +static PFN_GETCURRENTPROCESSORNUMBER GetCurrentProcessorNumberProc; + +typedef VOID (WINAPI *PFN_GETCURRENTPROCESSORNUMBEREX)(PPROCESSOR_NUMBER); +static PFN_GETCURRENTPROCESSORNUMBEREX GetCurrentProcessorNumberExProc; + +typedef BOOL (WINAPI *PFN_GETLOGICALPROCESSORINFORMATION)(PSYSTEM_LOGICAL_PROCESSOR_INFORMATION Buffer, PDWORD ReturnLength); +static PFN_GETLOGICALPROCESSORINFORMATION GetLogicalProcessorInformationProc; + +typedef BOOL (WINAPI *PFN_GETLOGICALPROCESSORINFORMATIONEX)(LOGICAL_PROCESSOR_RELATIONSHIP relationship, PSYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX Buffer, PDWORD ReturnLength); +static PFN_GETLOGICALPROCESSORINFORMATIONEX GetLogicalProcessorInformationExProc; + +typedef BOOL (WINAPI *PFN_SETTHREADGROUPAFFINITY)(HANDLE hThread, const GROUP_AFFINITY *GroupAffinity, PGROUP_AFFINITY PreviousGroupAffinity); +static PFN_SETTHREADGROUPAFFINITY SetThreadGroupAffinityProc; + +typedef BOOL (WINAPI *PFN_GETTHREADGROUPAFFINITY)(HANDLE hThread, PGROUP_AFFINITY GroupAffinity); +static PFN_GETTHREADGROUPAFFINITY GetThreadGroupAffinityProc; + +typedef BOOL (WINAPI *PFN_GETNUMAAVAILABLEMEMORYNODE)(UCHAR Node, PULONGLONG AvailableBytes); +static PFN_GETNUMAAVAILABLEMEMORYNODE GetNumaAvailableMemoryNodeProc; + +typedef BOOL (WINAPI *PFN_GETNUMAAVAILABLEMEMORYNODEEX)(USHORT Node, PULONGLONG AvailableBytes); +static PFN_GETNUMAAVAILABLEMEMORYNODEEX GetNumaAvailableMemoryNodeExProc; + +typedef LPVOID (WINAPI *PFN_VIRTUALALLOCEXNUMA)(HANDLE hProcess, LPVOID lpAddress, SIZE_T dwSize, DWORD flAllocationType, DWORD flProtect, DWORD nndPreferred); +static PFN_VIRTUALALLOCEXNUMA VirtualAllocExNumaProc; + +typedef BOOL (WINAPI *PFN_VIRTUALFREEEX)(HANDLE hProcess, LPVOID lpAddress, SIZE_T dwSize, DWORD dwFreeType); +static PFN_VIRTUALFREEEX VirtualFreeExProc; + +typedef BOOL (WINAPI *PFN_QUERYWORKINGSETEX)(HANDLE hProcess, PVOID pv, DWORD cb); +static PFN_QUERYWORKINGSETEX QueryWorkingSetExProc; + +static void hwloc_win_get_function_ptrs(void) +{ + HMODULE kernel32; + + kernel32 = LoadLibrary("kernel32.dll"); + if (kernel32) { + GetActiveProcessorGroupCountProc = + (PFN_GETACTIVEPROCESSORGROUPCOUNT) GetProcAddress(kernel32, "GetActiveProcessorGroupCount"); + GetActiveProcessorCountProc = + (PFN_GETACTIVEPROCESSORCOUNT) GetProcAddress(kernel32, "GetActiveProcessorCount"); + GetLogicalProcessorInformationProc = + (PFN_GETLOGICALPROCESSORINFORMATION) GetProcAddress(kernel32, "GetLogicalProcessorInformation"); + GetCurrentProcessorNumberProc = + (PFN_GETCURRENTPROCESSORNUMBER) GetProcAddress(kernel32, "GetCurrentProcessorNumber"); + GetCurrentProcessorNumberExProc = + (PFN_GETCURRENTPROCESSORNUMBEREX) GetProcAddress(kernel32, "GetCurrentProcessorNumberEx"); + SetThreadGroupAffinityProc = + (PFN_SETTHREADGROUPAFFINITY) GetProcAddress(kernel32, "SetThreadGroupAffinity"); + GetThreadGroupAffinityProc = + (PFN_GETTHREADGROUPAFFINITY) GetProcAddress(kernel32, "GetThreadGroupAffinity"); + GetNumaAvailableMemoryNodeProc = + (PFN_GETNUMAAVAILABLEMEMORYNODE) GetProcAddress(kernel32, "GetNumaAvailableMemoryNode"); + GetNumaAvailableMemoryNodeExProc = + (PFN_GETNUMAAVAILABLEMEMORYNODEEX) GetProcAddress(kernel32, "GetNumaAvailableMemoryNodeEx"); + GetLogicalProcessorInformationExProc = + (PFN_GETLOGICALPROCESSORINFORMATIONEX)GetProcAddress(kernel32, "GetLogicalProcessorInformationEx"); + QueryWorkingSetExProc = + (PFN_QUERYWORKINGSETEX) GetProcAddress(kernel32, "K32QueryWorkingSetEx"); + VirtualAllocExNumaProc = + (PFN_VIRTUALALLOCEXNUMA) GetProcAddress(kernel32, "VirtualAllocExNuma"); + VirtualFreeExProc = + (PFN_VIRTUALFREEEX) GetProcAddress(kernel32, "VirtualFreeEx"); + } + + if (GetActiveProcessorGroupCountProc) + nr_processor_groups = GetActiveProcessorGroupCountProc(); + + if (!QueryWorkingSetExProc) { + HMODULE psapi = LoadLibrary("psapi.dll"); + if (psapi) + QueryWorkingSetExProc = (PFN_QUERYWORKINGSETEX) GetProcAddress(psapi, "QueryWorkingSetEx"); + } +} + +/* + * ULONG_PTR and DWORD_PTR are 64/32bits depending on the arch + * while bitmaps use unsigned long (always 32bits) + */ + +static void hwloc_bitmap_from_ULONG_PTR(hwloc_bitmap_t set, ULONG_PTR mask) +{ +#if SIZEOF_VOID_P == 8 + hwloc_bitmap_from_ulong(set, mask & 0xffffffff); + hwloc_bitmap_set_ith_ulong(set, 1, mask >> 32); +#else + hwloc_bitmap_from_ulong(set, mask); +#endif +} + +static void hwloc_bitmap_from_ith_ULONG_PTR(hwloc_bitmap_t set, unsigned i, ULONG_PTR mask) +{ +#if SIZEOF_VOID_P == 8 + hwloc_bitmap_from_ith_ulong(set, 2*i, mask & 0xffffffff); + hwloc_bitmap_set_ith_ulong(set, 2*i+1, mask >> 32); +#else + hwloc_bitmap_from_ith_ulong(set, i, mask); +#endif +} + +static void hwloc_bitmap_set_ith_ULONG_PTR(hwloc_bitmap_t set, unsigned i, ULONG_PTR mask) +{ +#if SIZEOF_VOID_P == 8 + hwloc_bitmap_set_ith_ulong(set, 2*i, mask & 0xffffffff); + hwloc_bitmap_set_ith_ulong(set, 2*i+1, mask >> 32); +#else + hwloc_bitmap_set_ith_ulong(set, i, mask); +#endif +} + +static ULONG_PTR hwloc_bitmap_to_ULONG_PTR(hwloc_const_bitmap_t set) +{ +#if SIZEOF_VOID_P == 8 + ULONG_PTR up = hwloc_bitmap_to_ith_ulong(set, 1); + up <<= 32; + up |= hwloc_bitmap_to_ulong(set); + return up; +#else + return hwloc_bitmap_to_ulong(set); +#endif +} + +static ULONG_PTR hwloc_bitmap_to_ith_ULONG_PTR(hwloc_const_bitmap_t set, unsigned i) +{ +#if SIZEOF_VOID_P == 8 + ULONG_PTR up = hwloc_bitmap_to_ith_ulong(set, 2*i+1); + up <<= 32; + up |= hwloc_bitmap_to_ith_ulong(set, 2*i); + return up; +#else + return hwloc_bitmap_to_ith_ulong(set, i); +#endif +} + +/* convert set into index+mask if all set bits are in the same ULONG. + * otherwise return -1. + */ +static int hwloc_bitmap_to_single_ULONG_PTR(hwloc_const_bitmap_t set, unsigned *index, ULONG_PTR *mask) +{ + unsigned first_ulp, last_ulp; + if (hwloc_bitmap_weight(set) == -1) + return -1; + first_ulp = hwloc_bitmap_first(set) / (sizeof(ULONG_PTR)*8); + last_ulp = hwloc_bitmap_last(set) / (sizeof(ULONG_PTR)*8); + if (first_ulp != last_ulp) + return -1; + *mask = hwloc_bitmap_to_ith_ULONG_PTR(set, first_ulp); + *index = first_ulp; + return 0; +} + +/************************************************************** + * hwloc PU numbering with respect to Windows processor groups + * + * Everywhere below we reserve 64 physical indexes per processor groups because that's + * the maximum (MAXIMUM_PROC_PER_GROUP). Windows may actually use less bits than that + * in some groups (either to avoid splitting NUMA nodes across groups, or because of OS + * tweaks such as "bcdedit /set groupsize 8") but we keep some unused indexes for simplicity. + * That means PU physical indexes and cpusets may be non-contigous. + * That also means hwloc_fallback_nbprocessors() below must return the last PU index + 1 + * instead the actual number of processors. + */ + +/******************** + * last_cpu_location + */ + +static int +hwloc_win_get_thisthread_last_cpu_location(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_cpuset_t set, int flags __hwloc_attribute_unused) +{ + assert(GetCurrentProcessorNumberExProc || (GetCurrentProcessorNumberProc && nr_processor_groups == 1)); + + if (nr_processor_groups > 1 || !GetCurrentProcessorNumberProc) { + PROCESSOR_NUMBER num; + GetCurrentProcessorNumberExProc(&num); + hwloc_bitmap_from_ith_ULONG_PTR(set, num.Group, ((ULONG_PTR)1) << num.Number); + return 0; + } + + hwloc_bitmap_from_ith_ULONG_PTR(set, 0, ((ULONG_PTR)1) << GetCurrentProcessorNumberProc()); + return 0; +} + +/* TODO: hwloc_win_get_thisproc_last_cpu_location() using + * CreateToolhelp32Snapshot(), Thread32First/Next() + * th.th32OwnerProcessID == GetCurrentProcessId() for filtering within process + * OpenThread(THREAD_SET_INFORMATION|THREAD_QUERY_INFORMATION, FALSE, te32.th32ThreadID) to get a handle. + */ + + +/****************************** + * set cpu/membind for threads + */ + +/* TODO: SetThreadIdealProcessor{,Ex} */ + +static int +hwloc_win_set_thread_cpubind(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_thread_t thread, hwloc_const_bitmap_t hwloc_set, int flags) +{ + DWORD_PTR mask; + unsigned group; + + if (flags & HWLOC_CPUBIND_NOMEMBIND) { + errno = ENOSYS; + return -1; + } + + if (hwloc_bitmap_to_single_ULONG_PTR(hwloc_set, &group, &mask) < 0) { + errno = ENOSYS; + return -1; + } + + assert(nr_processor_groups == 1 || SetThreadGroupAffinityProc); + + if (nr_processor_groups > 1) { + GROUP_AFFINITY aff; + memset(&aff, 0, sizeof(aff)); /* we get Invalid Parameter error if Reserved field isn't cleared */ + aff.Group = group; + aff.Mask = mask; + if (!SetThreadGroupAffinityProc(thread, &aff, NULL)) + return -1; + + } else { + /* SetThreadAffinityMask() only changes the mask inside the current processor group */ + /* The resulting binding is always strict */ + if (!SetThreadAffinityMask(thread, mask)) + return -1; + } + return 0; +} + +static int +hwloc_win_set_thisthread_cpubind(hwloc_topology_t topology, hwloc_const_bitmap_t hwloc_set, int flags) +{ + return hwloc_win_set_thread_cpubind(topology, GetCurrentThread(), hwloc_set, flags); +} + +static int +hwloc_win_set_thisthread_membind(hwloc_topology_t topology, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags) +{ + int ret; + hwloc_const_cpuset_t cpuset; + hwloc_cpuset_t _cpuset = NULL; + + if ((policy != HWLOC_MEMBIND_DEFAULT && policy != HWLOC_MEMBIND_BIND) + || flags & HWLOC_MEMBIND_NOCPUBIND) { + errno = ENOSYS; + return -1; + } + + if (policy == HWLOC_MEMBIND_DEFAULT) { + cpuset = hwloc_topology_get_complete_cpuset(topology); + } else { + cpuset = _cpuset = hwloc_bitmap_alloc(); + hwloc_cpuset_from_nodeset(topology, _cpuset, nodeset); + } + + ret = hwloc_win_set_thisthread_cpubind(topology, cpuset, + (flags & HWLOC_MEMBIND_STRICT) ? HWLOC_CPUBIND_STRICT : 0); + hwloc_bitmap_free(_cpuset); + return ret; +} + + +/****************************** + * get cpu/membind for threads + */ + +static int +hwloc_win_get_thread_cpubind(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_thread_t thread, hwloc_cpuset_t set, int flags __hwloc_attribute_unused) +{ + GROUP_AFFINITY aff; + + assert(GetThreadGroupAffinityProc); + + if (!GetThreadGroupAffinityProc(thread, &aff)) + return -1; + hwloc_bitmap_from_ith_ULONG_PTR(set, aff.Group, aff.Mask); + return 0; +} + +static int +hwloc_win_get_thisthread_cpubind(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_cpuset_t set, int flags __hwloc_attribute_unused) +{ + return hwloc_win_get_thread_cpubind(topology, GetCurrentThread(), set, flags); +} + +static int +hwloc_win_get_thisthread_membind(hwloc_topology_t topology, hwloc_nodeset_t nodeset, hwloc_membind_policy_t * policy, int flags) +{ + int ret; + hwloc_cpuset_t cpuset = hwloc_bitmap_alloc(); + ret = hwloc_win_get_thread_cpubind(topology, GetCurrentThread(), cpuset, flags); + if (!ret) { + *policy = HWLOC_MEMBIND_BIND; + hwloc_cpuset_to_nodeset(topology, cpuset, nodeset); + } + hwloc_bitmap_free(cpuset); + return ret; +} + + +/******************************** + * set cpu/membind for processes + */ + +static int +hwloc_win_set_proc_cpubind(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_pid_t proc, hwloc_const_bitmap_t hwloc_set, int flags) +{ + DWORD_PTR mask; + + assert(nr_processor_groups == 1); + + if (flags & HWLOC_CPUBIND_NOMEMBIND) { + errno = ENOSYS; + return -1; + } + + /* TODO: SetThreadGroupAffinity() for all threads doesn't enforce the whole process affinity, + * maybe because of process-specific resource locality */ + /* TODO: if we are in a single group (check with GetProcessGroupAffinity()), + * SetProcessAffinityMask() changes the binding within that same group. + */ + /* TODO: NtSetInformationProcess() works very well for binding to any mask in a single group, + * but it's an internal routine. + */ + /* TODO: checks whether hwloc-bind.c needs to pass INHERIT_PARENT_AFFINITY to CreateProcess() instead of execvp(). */ + + /* The resulting binding is always strict */ + mask = hwloc_bitmap_to_ULONG_PTR(hwloc_set); + if (!SetProcessAffinityMask(proc, mask)) + return -1; + return 0; +} + +static int +hwloc_win_set_thisproc_cpubind(hwloc_topology_t topology, hwloc_const_bitmap_t hwloc_set, int flags) +{ + return hwloc_win_set_proc_cpubind(topology, GetCurrentProcess(), hwloc_set, flags); +} + +static int +hwloc_win_set_proc_membind(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags) +{ + int ret; + hwloc_const_cpuset_t cpuset; + hwloc_cpuset_t _cpuset = NULL; + + if ((policy != HWLOC_MEMBIND_DEFAULT && policy != HWLOC_MEMBIND_BIND) + || flags & HWLOC_MEMBIND_NOCPUBIND) { + errno = ENOSYS; + return -1; + } + + if (policy == HWLOC_MEMBIND_DEFAULT) { + cpuset = hwloc_topology_get_complete_cpuset(topology); + } else { + cpuset = _cpuset = hwloc_bitmap_alloc(); + hwloc_cpuset_from_nodeset(topology, _cpuset, nodeset); + } + + ret = hwloc_win_set_proc_cpubind(topology, pid, cpuset, + (flags & HWLOC_MEMBIND_STRICT) ? HWLOC_CPUBIND_STRICT : 0); + hwloc_bitmap_free(_cpuset); + return ret; +} + +static int +hwloc_win_set_thisproc_membind(hwloc_topology_t topology, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags) +{ + return hwloc_win_set_proc_membind(topology, GetCurrentProcess(), nodeset, policy, flags); +} + + +/******************************** + * get cpu/membind for processes + */ + +static int +hwloc_win_get_proc_cpubind(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_pid_t proc, hwloc_bitmap_t hwloc_set, int flags) +{ + DWORD_PTR proc_mask, sys_mask; + + assert(nr_processor_groups == 1); + + if (flags & HWLOC_CPUBIND_NOMEMBIND) { + errno = ENOSYS; + return -1; + } + + /* TODO: if we are in a single group (check with GetProcessGroupAffinity()), + * GetProcessAffinityMask() gives the mask within that group. + */ + /* TODO: if we are in multiple groups, GetProcessGroupAffinity() gives their IDs, + * but we don't know their masks. + */ + /* TODO: GetThreadGroupAffinity() for all threads can be smaller than the whole process affinity, + * maybe because of process-specific resource locality. + */ + + if (!GetProcessAffinityMask(proc, &proc_mask, &sys_mask)) + return -1; + hwloc_bitmap_from_ULONG_PTR(hwloc_set, proc_mask); + return 0; +} + +static int +hwloc_win_get_proc_membind(hwloc_topology_t topology, hwloc_pid_t pid, hwloc_nodeset_t nodeset, hwloc_membind_policy_t * policy, int flags) +{ + int ret; + hwloc_cpuset_t cpuset = hwloc_bitmap_alloc(); + ret = hwloc_win_get_proc_cpubind(topology, pid, cpuset, + (flags & HWLOC_MEMBIND_STRICT) ? HWLOC_CPUBIND_STRICT : 0); + if (!ret) { + *policy = HWLOC_MEMBIND_BIND; + hwloc_cpuset_to_nodeset(topology, cpuset, nodeset); + } + hwloc_bitmap_free(cpuset); + return ret; +} + +static int +hwloc_win_get_thisproc_cpubind(hwloc_topology_t topology, hwloc_bitmap_t hwloc_cpuset, int flags) +{ + return hwloc_win_get_proc_cpubind(topology, GetCurrentProcess(), hwloc_cpuset, flags); +} + +static int +hwloc_win_get_thisproc_membind(hwloc_topology_t topology, hwloc_nodeset_t nodeset, hwloc_membind_policy_t * policy, int flags) +{ + return hwloc_win_get_proc_membind(topology, GetCurrentProcess(), nodeset, policy, flags); +} + + +/************************ + * membind alloc/free + */ + +static void * +hwloc_win_alloc(hwloc_topology_t topology __hwloc_attribute_unused, size_t len) { + return VirtualAlloc(NULL, len, MEM_COMMIT|MEM_RESERVE, PAGE_EXECUTE_READWRITE); +} + +static void * +hwloc_win_alloc_membind(hwloc_topology_t topology __hwloc_attribute_unused, size_t len, hwloc_const_nodeset_t nodeset, hwloc_membind_policy_t policy, int flags) { + int node; + + switch (policy) { + case HWLOC_MEMBIND_DEFAULT: + case HWLOC_MEMBIND_BIND: + break; + default: + errno = ENOSYS; + return hwloc_alloc_or_fail(topology, len, flags); + } + + if (flags & HWLOC_MEMBIND_STRICT) { + errno = ENOSYS; + return NULL; + } + + if (policy == HWLOC_MEMBIND_DEFAULT + || hwloc_bitmap_isequal(nodeset, hwloc_topology_get_complete_nodeset(topology))) + return hwloc_win_alloc(topology, len); + + if (hwloc_bitmap_weight(nodeset) != 1) { + /* Not a single node, can't do this */ + errno = EXDEV; + return hwloc_alloc_or_fail(topology, len, flags); + } + + node = hwloc_bitmap_first(nodeset); + return VirtualAllocExNumaProc(GetCurrentProcess(), NULL, len, MEM_COMMIT|MEM_RESERVE, PAGE_EXECUTE_READWRITE, node); +} + +static int +hwloc_win_free_membind(hwloc_topology_t topology __hwloc_attribute_unused, void *addr, size_t len __hwloc_attribute_unused) { + if (!addr) + return 0; + if (!VirtualFreeExProc(GetCurrentProcess(), addr, 0, MEM_RELEASE)) + return -1; + return 0; +} + + +/********************** + * membind for areas + */ + +static int +hwloc_win_get_area_memlocation(hwloc_topology_t topology __hwloc_attribute_unused, const void *addr, size_t len, hwloc_nodeset_t nodeset, int flags __hwloc_attribute_unused) +{ + SYSTEM_INFO SystemInfo; + DWORD page_size; + uintptr_t start; + unsigned nb; + PSAPI_WORKING_SET_EX_INFORMATION *pv; + unsigned i; + + GetSystemInfo(&SystemInfo); + page_size = SystemInfo.dwPageSize; + + start = (((uintptr_t) addr) / page_size) * page_size; + nb = (unsigned)((((uintptr_t) addr + len - start) + page_size - 1) / page_size); + + if (!nb) + nb = 1; + + pv = calloc(nb, sizeof(*pv)); + if (!pv) + return -1; + + for (i = 0; i < nb; i++) + pv[i].VirtualAddress = (void*) (start + i * page_size); + if (!QueryWorkingSetExProc(GetCurrentProcess(), pv, nb * sizeof(*pv))) { + free(pv); + return -1; + } + + for (i = 0; i < nb; i++) { + if (pv[i].VirtualAttributes.Valid) + hwloc_bitmap_set(nodeset, pv[i].VirtualAttributes.Node); + } + + free(pv); + return 0; +} + + +/************************* + * discovery + */ + +static int +hwloc_look_windows(struct hwloc_backend *backend) +{ + struct hwloc_topology *topology = backend->topology; + hwloc_bitmap_t groups_pu_set = NULL; + SYSTEM_INFO SystemInfo; + DWORD length; + int gotnuma = 0; + int gotnumamemory = 0; + + if (topology->levels[0][0]->cpuset) + /* somebody discovered things */ + return -1; + + hwloc_alloc_root_sets(topology->levels[0][0]); + + GetSystemInfo(&SystemInfo); + + if (!GetLogicalProcessorInformationExProc && GetLogicalProcessorInformationProc) { + PSYSTEM_LOGICAL_PROCESSOR_INFORMATION procInfo, tmpprocInfo; + unsigned id; + unsigned i; + struct hwloc_obj *obj; + hwloc_obj_type_t type; + + length = 0; + procInfo = NULL; + + while (1) { + if (GetLogicalProcessorInformationProc(procInfo, &length)) + break; + if (GetLastError() != ERROR_INSUFFICIENT_BUFFER) + return -1; + tmpprocInfo = realloc(procInfo, length); + if (!tmpprocInfo) { + free(procInfo); + goto out; + } + procInfo = tmpprocInfo; + } + + assert(!length || procInfo); + + for (i = 0; i < length / sizeof(*procInfo); i++) { + + /* Ignore unknown caches */ + if (procInfo->Relationship == RelationCache + && procInfo->Cache.Type != CacheUnified + && procInfo->Cache.Type != CacheData + && procInfo->Cache.Type != CacheInstruction) + continue; + + id = HWLOC_UNKNOWN_INDEX; + switch (procInfo[i].Relationship) { + case RelationNumaNode: + type = HWLOC_OBJ_NUMANODE; + id = procInfo[i].NumaNode.NodeNumber; + gotnuma++; + if (id > max_numanode_index) + max_numanode_index = id; + break; + case RelationProcessorPackage: + type = HWLOC_OBJ_PACKAGE; + break; + case RelationCache: + type = (procInfo[i].Cache.Type == CacheInstruction ? HWLOC_OBJ_L1ICACHE : HWLOC_OBJ_L1CACHE) + procInfo[i].Cache.Level - 1; + break; + case RelationProcessorCore: + type = HWLOC_OBJ_CORE; + break; + case RelationGroup: + default: + type = HWLOC_OBJ_GROUP; + break; + } + + if (!hwloc_filter_check_keep_object_type(topology, type)) + continue; + + obj = hwloc_alloc_setup_object(topology, type, id); + obj->cpuset = hwloc_bitmap_alloc(); + hwloc_debug("%s#%u mask %llx\n", hwloc_obj_type_string(type), id, (unsigned long long) procInfo[i].ProcessorMask); + /* ProcessorMask is a ULONG_PTR */ + hwloc_bitmap_set_ith_ULONG_PTR(obj->cpuset, 0, procInfo[i].ProcessorMask); + hwloc_debug_2args_bitmap("%s#%u bitmap %s\n", hwloc_obj_type_string(type), id, obj->cpuset); + + switch (type) { + case HWLOC_OBJ_NUMANODE: + { + ULONGLONG avail; + obj->nodeset = hwloc_bitmap_alloc(); + hwloc_bitmap_set(obj->nodeset, id); + if ((GetNumaAvailableMemoryNodeExProc && GetNumaAvailableMemoryNodeExProc(id, &avail)) + || (GetNumaAvailableMemoryNodeProc && GetNumaAvailableMemoryNodeProc(id, &avail))) { + obj->attr->numanode.local_memory = avail; + gotnumamemory++; + } + obj->attr->numanode.page_types_len = 2; + obj->attr->numanode.page_types = malloc(2 * sizeof(*obj->attr->numanode.page_types)); + memset(obj->attr->numanode.page_types, 0, 2 * sizeof(*obj->attr->numanode.page_types)); + obj->attr->numanode.page_types_len = 1; + obj->attr->numanode.page_types[0].size = SystemInfo.dwPageSize; +#if HAVE_DECL__SC_LARGE_PAGESIZE + obj->attr->numanode.page_types_len++; + obj->attr->numanode.page_types[1].size = sysconf(_SC_LARGE_PAGESIZE); +#endif + break; + } + case HWLOC_OBJ_L1CACHE: + case HWLOC_OBJ_L2CACHE: + case HWLOC_OBJ_L3CACHE: + case HWLOC_OBJ_L4CACHE: + case HWLOC_OBJ_L5CACHE: + case HWLOC_OBJ_L1ICACHE: + case HWLOC_OBJ_L2ICACHE: + case HWLOC_OBJ_L3ICACHE: + obj->attr->cache.size = procInfo[i].Cache.Size; + obj->attr->cache.associativity = procInfo[i].Cache.Associativity == CACHE_FULLY_ASSOCIATIVE ? -1 : procInfo[i].Cache.Associativity ; + obj->attr->cache.linesize = procInfo[i].Cache.LineSize; + obj->attr->cache.depth = procInfo[i].Cache.Level; + switch (procInfo->Cache.Type) { + case CacheUnified: + obj->attr->cache.type = HWLOC_OBJ_CACHE_UNIFIED; + break; + case CacheData: + obj->attr->cache.type = HWLOC_OBJ_CACHE_DATA; + break; + case CacheInstruction: + obj->attr->cache.type = HWLOC_OBJ_CACHE_INSTRUCTION; + break; + default: + hwloc_free_unlinked_object(obj); + continue; + } + break; + case HWLOC_OBJ_GROUP: + obj->attr->group.kind = procInfo[i].Relationship == RelationGroup ? HWLOC_GROUP_KIND_WINDOWS_PROCESSOR_GROUP : HWLOC_GROUP_KIND_WINDOWS_RELATIONSHIP_UNKNOWN; + break; + default: + break; + } + hwloc_insert_object_by_cpuset(topology, obj); + } + + free(procInfo); + } + + if (GetLogicalProcessorInformationExProc) { + PSYSTEM_LOGICAL_PROCESSOR_INFORMATION_EX procInfoTotal, tmpprocInfoTotal, procInfo; + unsigned id; + struct hwloc_obj *obj; + hwloc_obj_type_t type; + + length = 0; + procInfoTotal = NULL; + + while (1) { + if (GetLogicalProcessorInformationExProc(RelationAll, procInfoTotal, &length)) + break; + if (GetLastError() != ERROR_INSUFFICIENT_BUFFER) + return -1; + tmpprocInfoTotal = realloc(procInfoTotal, length); + if (!tmpprocInfoTotal) { + free(procInfoTotal); + goto out; + } + procInfoTotal = tmpprocInfoTotal; + } + + for (procInfo = procInfoTotal; + (void*) procInfo < (void*) ((uintptr_t) procInfoTotal + length); + procInfo = (void*) ((uintptr_t) procInfo + procInfo->Size)) { + unsigned num, i; + GROUP_AFFINITY *GroupMask; + + /* Ignore unknown caches */ + if (procInfo->Relationship == RelationCache + && procInfo->Cache.Type != CacheUnified + && procInfo->Cache.Type != CacheData + && procInfo->Cache.Type != CacheInstruction) + continue; + + id = HWLOC_UNKNOWN_INDEX; + switch (procInfo->Relationship) { + case RelationNumaNode: + type = HWLOC_OBJ_NUMANODE; + num = 1; + GroupMask = &procInfo->NumaNode.GroupMask; + id = procInfo->NumaNode.NodeNumber; + gotnuma++; + if (id > max_numanode_index) + max_numanode_index = id; + break; + case RelationProcessorPackage: + type = HWLOC_OBJ_PACKAGE; + num = procInfo->Processor.GroupCount; + GroupMask = procInfo->Processor.GroupMask; + break; + case RelationCache: + type = (procInfo->Cache.Type == CacheInstruction ? HWLOC_OBJ_L1ICACHE : HWLOC_OBJ_L1CACHE) + procInfo->Cache.Level - 1; + num = 1; + GroupMask = &procInfo->Cache.GroupMask; + break; + case RelationProcessorCore: + type = HWLOC_OBJ_CORE; + num = procInfo->Processor.GroupCount; + GroupMask = procInfo->Processor.GroupMask; + break; + case RelationGroup: + /* So strange an interface... */ + for (id = 0; id < procInfo->Group.ActiveGroupCount; id++) { + KAFFINITY mask; + hwloc_bitmap_t set; + + set = hwloc_bitmap_alloc(); + mask = procInfo->Group.GroupInfo[id].ActiveProcessorMask; + hwloc_debug("group %u %d cpus mask %lx\n", id, + procInfo->Group.GroupInfo[id].ActiveProcessorCount, mask); + /* KAFFINITY is ULONG_PTR */ + hwloc_bitmap_set_ith_ULONG_PTR(set, id, mask); + /* FIXME: what if running 32bits on a 64bits windows with 64-processor groups? + * ULONG_PTR is 32bits, so half the group is invisible? + * maybe scale id to id*8/sizeof(ULONG_PTR) so that groups are 64-PU aligned? + */ + hwloc_debug_2args_bitmap("group %u %d bitmap %s\n", id, procInfo->Group.GroupInfo[id].ActiveProcessorCount, set); + + /* save the set of PUs so that we can create them at the end */ + if (!groups_pu_set) + groups_pu_set = hwloc_bitmap_alloc(); + hwloc_bitmap_or(groups_pu_set, groups_pu_set, set); + + if (hwloc_filter_check_keep_object_type(topology, HWLOC_OBJ_GROUP)) { + obj = hwloc_alloc_setup_object(topology, HWLOC_OBJ_GROUP, id); + obj->cpuset = set; + obj->attr->group.kind = HWLOC_GROUP_KIND_WINDOWS_PROCESSOR_GROUP; + hwloc_insert_object_by_cpuset(topology, obj); + } else + hwloc_bitmap_free(set); + } + continue; + default: + /* Don't know how to get the mask. */ + hwloc_debug("unknown relation %d\n", procInfo->Relationship); + continue; + } + + if (!hwloc_filter_check_keep_object_type(topology, type)) + continue; + + obj = hwloc_alloc_setup_object(topology, type, id); + obj->cpuset = hwloc_bitmap_alloc(); + for (i = 0; i < num; i++) { + hwloc_debug("%s#%u %d: mask %d:%lx\n", hwloc_obj_type_string(type), id, i, GroupMask[i].Group, GroupMask[i].Mask); + /* GROUP_AFFINITY.Mask is KAFFINITY, which is ULONG_PTR */ + hwloc_bitmap_set_ith_ULONG_PTR(obj->cpuset, GroupMask[i].Group, GroupMask[i].Mask); + /* FIXME: scale id to id*8/sizeof(ULONG_PTR) as above? */ + } + hwloc_debug_2args_bitmap("%s#%u bitmap %s\n", hwloc_obj_type_string(type), id, obj->cpuset); + switch (type) { + case HWLOC_OBJ_NUMANODE: + { + ULONGLONG avail; + obj->nodeset = hwloc_bitmap_alloc(); + hwloc_bitmap_set(obj->nodeset, id); + if ((GetNumaAvailableMemoryNodeExProc && GetNumaAvailableMemoryNodeExProc(id, &avail)) + || (GetNumaAvailableMemoryNodeProc && GetNumaAvailableMemoryNodeProc(id, &avail))) { + obj->attr->numanode.local_memory = avail; + gotnumamemory++; + } + obj->attr->numanode.page_types = malloc(2 * sizeof(*obj->attr->numanode.page_types)); + memset(obj->attr->numanode.page_types, 0, 2 * sizeof(*obj->attr->numanode.page_types)); + obj->attr->numanode.page_types_len = 1; + obj->attr->numanode.page_types[0].size = SystemInfo.dwPageSize; +#if HAVE_DECL__SC_LARGE_PAGESIZE + obj->attr->numanode.page_types_len++; + obj->attr->numanode.page_types[1].size = sysconf(_SC_LARGE_PAGESIZE); +#endif + break; + } + case HWLOC_OBJ_L1CACHE: + case HWLOC_OBJ_L2CACHE: + case HWLOC_OBJ_L3CACHE: + case HWLOC_OBJ_L4CACHE: + case HWLOC_OBJ_L5CACHE: + case HWLOC_OBJ_L1ICACHE: + case HWLOC_OBJ_L2ICACHE: + case HWLOC_OBJ_L3ICACHE: + obj->attr->cache.size = procInfo->Cache.CacheSize; + obj->attr->cache.associativity = procInfo->Cache.Associativity == CACHE_FULLY_ASSOCIATIVE ? -1 : procInfo->Cache.Associativity ; + obj->attr->cache.linesize = procInfo->Cache.LineSize; + obj->attr->cache.depth = procInfo->Cache.Level; + switch (procInfo->Cache.Type) { + case CacheUnified: + obj->attr->cache.type = HWLOC_OBJ_CACHE_UNIFIED; + break; + case CacheData: + obj->attr->cache.type = HWLOC_OBJ_CACHE_DATA; + break; + case CacheInstruction: + obj->attr->cache.type = HWLOC_OBJ_CACHE_INSTRUCTION; + break; + default: + hwloc_free_unlinked_object(obj); + continue; + } + break; + default: + break; + } + hwloc_insert_object_by_cpuset(topology, obj); + } + free(procInfoTotal); + } + + topology->support.discovery->pu = 1; + topology->support.discovery->numa = gotnuma; + topology->support.discovery->numa_memory = gotnumamemory; + + if (groups_pu_set) { + /* the system supports multiple Groups. + * PU indexes may be discontiguous, especially if Groups contain less than 64 procs. + */ + hwloc_obj_t obj; + unsigned idx; + hwloc_bitmap_foreach_begin(idx, groups_pu_set) { + obj = hwloc_alloc_setup_object(topology, HWLOC_OBJ_PU, idx); + obj->cpuset = hwloc_bitmap_alloc(); + hwloc_bitmap_only(obj->cpuset, idx); + hwloc_debug_1arg_bitmap("cpu %u has cpuset %s\n", + idx, obj->cpuset); + hwloc_insert_object_by_cpuset(topology, obj); + } hwloc_bitmap_foreach_end(); + hwloc_bitmap_free(groups_pu_set); + } else { + /* no processor groups */ + SYSTEM_INFO sysinfo; + hwloc_obj_t obj; + unsigned idx; + GetSystemInfo(&sysinfo); + for(idx=0; idx<32; idx++) + if (sysinfo.dwActiveProcessorMask & (((DWORD_PTR)1)<cpuset = hwloc_bitmap_alloc(); + hwloc_bitmap_only(obj->cpuset, idx); + hwloc_debug_1arg_bitmap("cpu %u has cpuset %s\n", + idx, obj->cpuset); + hwloc_insert_object_by_cpuset(topology, obj); + } + } + + out: + hwloc_obj_add_info(topology->levels[0][0], "Backend", "Windows"); + hwloc_add_uname_info(topology, NULL); + return 0; +} + +void +hwloc_set_windows_hooks(struct hwloc_binding_hooks *hooks, + struct hwloc_topology_support *support) +{ + if (GetCurrentProcessorNumberExProc || (GetCurrentProcessorNumberProc && nr_processor_groups == 1)) + hooks->get_thisthread_last_cpu_location = hwloc_win_get_thisthread_last_cpu_location; + + if (nr_processor_groups == 1) { + hooks->set_proc_cpubind = hwloc_win_set_proc_cpubind; + hooks->get_proc_cpubind = hwloc_win_get_proc_cpubind; + hooks->set_thisproc_cpubind = hwloc_win_set_thisproc_cpubind; + hooks->get_thisproc_cpubind = hwloc_win_get_thisproc_cpubind; + hooks->set_proc_membind = hwloc_win_set_proc_membind; + hooks->get_proc_membind = hwloc_win_get_proc_membind; + hooks->set_thisproc_membind = hwloc_win_set_thisproc_membind; + hooks->get_thisproc_membind = hwloc_win_get_thisproc_membind; + } + if (nr_processor_groups == 1 || SetThreadGroupAffinityProc) { + hooks->set_thread_cpubind = hwloc_win_set_thread_cpubind; + hooks->set_thisthread_cpubind = hwloc_win_set_thisthread_cpubind; + hooks->set_thisthread_membind = hwloc_win_set_thisthread_membind; + } + if (GetThreadGroupAffinityProc) { + hooks->get_thread_cpubind = hwloc_win_get_thread_cpubind; + hooks->get_thisthread_cpubind = hwloc_win_get_thisthread_cpubind; + hooks->get_thisthread_membind = hwloc_win_get_thisthread_membind; + } + + if (VirtualAllocExNumaProc) { + hooks->alloc_membind = hwloc_win_alloc_membind; + hooks->alloc = hwloc_win_alloc; + hooks->free_membind = hwloc_win_free_membind; + support->membind->bind_membind = 1; + } + + if (QueryWorkingSetExProc && max_numanode_index <= 63 /* PSAPI_WORKING_SET_EX_BLOCK.Node is 6 bits only */) + hooks->get_area_memlocation = hwloc_win_get_area_memlocation; +} + +static int hwloc_windows_component_init(unsigned long flags __hwloc_attribute_unused) +{ + hwloc_win_get_function_ptrs(); + return 0; +} + +static void hwloc_windows_component_finalize(unsigned long flags __hwloc_attribute_unused) +{ +} + +static struct hwloc_backend * +hwloc_windows_component_instantiate(struct hwloc_disc_component *component, + const void *_data1 __hwloc_attribute_unused, + const void *_data2 __hwloc_attribute_unused, + const void *_data3 __hwloc_attribute_unused) +{ + struct hwloc_backend *backend; + backend = hwloc_backend_alloc(component); + if (!backend) + return NULL; + backend->discover = hwloc_look_windows; + return backend; +} + +static struct hwloc_disc_component hwloc_windows_disc_component = { + HWLOC_DISC_COMPONENT_TYPE_CPU, + "windows", + HWLOC_DISC_COMPONENT_TYPE_GLOBAL, + hwloc_windows_component_instantiate, + 50, + 1, + NULL +}; + +const struct hwloc_component hwloc_windows_component = { + HWLOC_COMPONENT_ABI, + hwloc_windows_component_init, hwloc_windows_component_finalize, + HWLOC_COMPONENT_TYPE_DISC, + 0, + &hwloc_windows_disc_component +}; + +int +hwloc_fallback_nbprocessors(struct hwloc_topology *topology __hwloc_attribute_unused) { + int n; + SYSTEM_INFO sysinfo; + + /* by default, ignore groups (return only the number in the current group) */ + GetSystemInfo(&sysinfo); + n = sysinfo.dwNumberOfProcessors; /* FIXME could be non-contigous, rather return a mask from dwActiveProcessorMask? */ + + if (nr_processor_groups > 1) { + /* assume n-1 groups are complete, since that's how we store things in cpusets */ + if (GetActiveProcessorCountProc) + n = MAXIMUM_PROC_PER_GROUP*(nr_processor_groups-1) + + GetActiveProcessorCountProc((WORD)nr_processor_groups-1); + else + n = MAXIMUM_PROC_PER_GROUP*nr_processor_groups; + } + + return n; +} diff --git a/src/3rdparty/hwloc/src/topology-x86.c b/src/3rdparty/hwloc/src/topology-x86.c new file mode 100644 index 000000000..4aefdcf1f --- /dev/null +++ b/src/3rdparty/hwloc/src/topology-x86.c @@ -0,0 +1,1583 @@ +/* + * Copyright © 2010-2019 Inria. All rights reserved. + * Copyright © 2010-2013 Université Bordeaux + * Copyright © 2010-2011 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + * + * + * This backend is only used when the operating system does not export + * the necessary hardware topology information to user-space applications. + * Currently, only the FreeBSD backend relies on this x86 backend. + * + * Other backends such as Linux have their own way to retrieve various + * pieces of hardware topology information from the operating system + * on various architectures, without having to use this x86-specific code. + */ + +#include +#include +#include +#include +#include + +#include + +#include +#ifdef HAVE_DIRENT_H +#include +#endif +#ifdef HAVE_VALGRIND_VALGRIND_H +#include +#endif + +struct hwloc_x86_backend_data_s { + unsigned nbprocs; + hwloc_bitmap_t apicid_set; + int apicid_unique; + char *src_cpuiddump_path; + int is_knl; +}; + +/************************************ + * Management of cpuid dump as input + */ + +struct cpuiddump { + unsigned nr; + struct cpuiddump_entry { + unsigned inmask; /* which of ine[abcd]x are set on input */ + unsigned ineax; + unsigned inebx; + unsigned inecx; + unsigned inedx; + unsigned outeax; + unsigned outebx; + unsigned outecx; + unsigned outedx; + } *entries; +}; + +static void +cpuiddump_free(struct cpuiddump *cpuiddump) +{ + if (cpuiddump->nr) + free(cpuiddump->entries); + free(cpuiddump); +} + +static struct cpuiddump * +cpuiddump_read(const char *dirpath, unsigned idx) +{ + struct cpuiddump *cpuiddump; + struct cpuiddump_entry *cur; + FILE *file; + char line[128]; + unsigned nr; + + cpuiddump = malloc(sizeof(*cpuiddump)); + if (!cpuiddump) { + fprintf(stderr, "Failed to allocate cpuiddump for PU #%u, ignoring cpuiddump.\n", idx); + goto out; + } + + { + size_t filenamelen = strlen(dirpath) + 15; + HWLOC_VLA(char, filename, filenamelen); + snprintf(filename, filenamelen, "%s/pu%u", dirpath, idx); + file = fopen(filename, "r"); + if (!file) { + fprintf(stderr, "Could not read dumped cpuid file %s, ignoring cpuiddump.\n", filename); + goto out_with_dump; + } + } + + nr = 0; + while (fgets(line, sizeof(line), file)) + nr++; + cpuiddump->entries = malloc(nr * sizeof(struct cpuiddump_entry)); + if (!cpuiddump->entries) { + fprintf(stderr, "Failed to allocate %u cpuiddump entries for PU #%u, ignoring cpuiddump.\n", nr, idx); + goto out_with_file; + } + + fseek(file, 0, SEEK_SET); + cur = &cpuiddump->entries[0]; + nr = 0; + while (fgets(line, sizeof(line), file)) { + if (*line == '#') + continue; + if (sscanf(line, "%x %x %x %x %x => %x %x %x %x", + &cur->inmask, + &cur->ineax, &cur->inebx, &cur->inecx, &cur->inedx, + &cur->outeax, &cur->outebx, &cur->outecx, &cur->outedx) == 9) { + cur++; + nr++; + } + } + + cpuiddump->nr = nr; + fclose(file); + return cpuiddump; + + out_with_file: + fclose(file); + out_with_dump: + free(cpuiddump); + out: + return NULL; +} + +static void +cpuiddump_find_by_input(unsigned *eax, unsigned *ebx, unsigned *ecx, unsigned *edx, struct cpuiddump *cpuiddump) +{ + unsigned i; + + for(i=0; inr; i++) { + struct cpuiddump_entry *entry = &cpuiddump->entries[i]; + if ((entry->inmask & 0x1) && *eax != entry->ineax) + continue; + if ((entry->inmask & 0x2) && *ebx != entry->inebx) + continue; + if ((entry->inmask & 0x4) && *ecx != entry->inecx) + continue; + if ((entry->inmask & 0x8) && *edx != entry->inedx) + continue; + *eax = entry->outeax; + *ebx = entry->outebx; + *ecx = entry->outecx; + *edx = entry->outedx; + return; + } + + fprintf(stderr, "Couldn't find %x,%x,%x,%x in dumped cpuid, returning 0s.\n", + *eax, *ebx, *ecx, *edx); + *eax = 0; + *ebx = 0; + *ecx = 0; + *edx = 0; +} + +static void cpuid_or_from_dump(unsigned *eax, unsigned *ebx, unsigned *ecx, unsigned *edx, struct cpuiddump *src_cpuiddump) +{ + if (src_cpuiddump) { + cpuiddump_find_by_input(eax, ebx, ecx, edx, src_cpuiddump); + } else { + hwloc_x86_cpuid(eax, ebx, ecx, edx); + } +} + +/******************************* + * Core detection routines and structures + */ + +#define has_topoext(features) ((features)[6] & (1 << 22)) +#define has_x2apic(features) ((features)[4] & (1 << 21)) + +struct cacheinfo { + hwloc_obj_cache_type_t type; + unsigned level; + unsigned nbthreads_sharing; + unsigned cacheid; + + unsigned linesize; + unsigned linepart; + int inclusive; + int ways; + unsigned sets; + unsigned long size; +}; + +struct procinfo { + unsigned present; + unsigned apicid; + unsigned packageid; + unsigned dieid; + unsigned nodeid; + unsigned unitid; + unsigned threadid; + unsigned coreid; + unsigned *otherids; + unsigned levels; + unsigned numcaches; + struct cacheinfo *cache; + char cpuvendor[13]; + char cpumodel[3*4*4+1]; + unsigned cpustepping; + unsigned cpumodelnumber; + unsigned cpufamilynumber; +}; + +enum cpuid_type { + intel, + amd, + zhaoxin, + hygon, + unknown +}; + +static void fill_amd_cache(struct procinfo *infos, unsigned level, hwloc_obj_cache_type_t type, unsigned nbthreads_sharing, unsigned cpuid) +{ + struct cacheinfo *cache, *tmpcaches; + unsigned cachenum; + unsigned long size = 0; + + if (level == 1) + size = ((cpuid >> 24)) << 10; + else if (level == 2) + size = ((cpuid >> 16)) << 10; + else if (level == 3) + size = ((cpuid >> 18)) << 19; + if (!size) + return; + + tmpcaches = realloc(infos->cache, (infos->numcaches+1)*sizeof(*infos->cache)); + if (!tmpcaches) + /* failed to allocated, ignore that cache */ + return; + infos->cache = tmpcaches; + cachenum = infos->numcaches++; + + cache = &infos->cache[cachenum]; + + cache->type = type; + cache->level = level; + cache->nbthreads_sharing = nbthreads_sharing; + cache->linesize = cpuid & 0xff; + cache->linepart = 0; + cache->inclusive = 0; /* old AMD (K8-K10) supposed to have exclusive caches */ + + if (level == 1) { + cache->ways = (cpuid >> 16) & 0xff; + if (cache->ways == 0xff) + /* Fully associative */ + cache->ways = -1; + } else { + static const unsigned ways_tab[] = { 0, 1, 2, 0, 4, 0, 8, 0, 16, 0, 32, 48, 64, 96, 128, -1 }; + unsigned ways = (cpuid >> 12) & 0xf; + cache->ways = ways_tab[ways]; + } + cache->size = size; + cache->sets = 0; + + hwloc_debug("cache L%u t%u linesize %u ways %d size %luKB\n", cache->level, cache->nbthreads_sharing, cache->linesize, cache->ways, cache->size >> 10); +} + +static void look_exttopoenum(struct procinfo *infos, unsigned leaf, struct cpuiddump *src_cpuiddump) +{ + unsigned level, apic_nextshift, apic_number, apic_type, apic_id = 0, apic_shift = 0, id; + unsigned threadid __hwloc_attribute_unused = 0; /* shut-up compiler */ + unsigned eax, ebx, ecx = 0, edx; + int apic_packageshift = 0; + + for (level = 0; ; level++) { + ecx = level; + eax = leaf; + cpuid_or_from_dump(&eax, &ebx, &ecx, &edx, src_cpuiddump); + if (!eax && !ebx) + break; + apic_packageshift = eax & 0x1f; + } + + if (level) { + infos->otherids = malloc(level * sizeof(*infos->otherids)); + if (infos->otherids) { + infos->levels = level; + for (level = 0; ; level++) { + ecx = level; + eax = leaf; + cpuid_or_from_dump(&eax, &ebx, &ecx, &edx, src_cpuiddump); + if (!eax && !ebx) + break; + apic_nextshift = eax & 0x1f; + apic_number = ebx & 0xffff; + apic_type = (ecx & 0xff00) >> 8; + apic_id = edx; + id = (apic_id >> apic_shift) & ((1 << (apic_packageshift - apic_shift)) - 1); + hwloc_debug("x2APIC %08x %u: nextshift %u num %2u type %u id %2u\n", apic_id, level, apic_nextshift, apic_number, apic_type, id); + infos->apicid = apic_id; + infos->otherids[level] = UINT_MAX; + switch (apic_type) { + case 1: + threadid = id; + /* apic_number is the actual number of threads per core */ + break; + case 2: + infos->coreid = id; + /* apic_number is the actual number of threads per module */ + break; + case 5: + infos->dieid = id; + /* apic_number is the actual number of threads per package */ + break; + default: + hwloc_debug("x2APIC %u: unknown type %u\n", level, apic_type); + infos->otherids[level] = apic_id >> apic_shift; + break; + } + apic_shift = apic_nextshift; + } + infos->apicid = apic_id; + infos->packageid = apic_id >> apic_shift; + hwloc_debug("x2APIC remainder: %u\n", infos->packageid); + hwloc_debug("this is thread %u of core %u\n", threadid, infos->coreid); + } + } +} + +/* Fetch information from the processor itself thanks to cpuid and store it in + * infos for summarize to analyze them globally */ +static void look_proc(struct hwloc_backend *backend, struct procinfo *infos, unsigned highest_cpuid, unsigned highest_ext_cpuid, unsigned *features, enum cpuid_type cpuid_type, struct cpuiddump *src_cpuiddump) +{ + struct hwloc_x86_backend_data_s *data = backend->private_data; + unsigned eax, ebx, ecx = 0, edx; + unsigned cachenum; + struct cacheinfo *cache; + unsigned regs[4]; + unsigned legacy_max_log_proc; /* not valid on Intel processors with > 256 threads, or when cpuid 0x80000008 is supported */ + unsigned legacy_log_proc_id; + unsigned _model, _extendedmodel, _family, _extendedfamily; + + infos->present = 1; + + /* Get apicid, legacy_max_log_proc, packageid, legacy_log_proc_id from cpuid 0x01 */ + eax = 0x01; + cpuid_or_from_dump(&eax, &ebx, &ecx, &edx, src_cpuiddump); + infos->apicid = ebx >> 24; + if (edx & (1 << 28)) + legacy_max_log_proc = 1 << hwloc_flsl(((ebx >> 16) & 0xff) - 1); + else + legacy_max_log_proc = 1; + hwloc_debug("APIC ID 0x%02x legacy_max_log_proc %u\n", infos->apicid, legacy_max_log_proc); + infos->packageid = infos->apicid / legacy_max_log_proc; + legacy_log_proc_id = infos->apicid % legacy_max_log_proc; + hwloc_debug("phys %u legacy thread %u\n", infos->packageid, legacy_log_proc_id); + + /* Get cpu model/family/stepping numbers from same cpuid */ + _model = (eax>>4) & 0xf; + _extendedmodel = (eax>>16) & 0xf; + _family = (eax>>8) & 0xf; + _extendedfamily = (eax>>20) & 0xff; + if ((cpuid_type == intel || cpuid_type == amd || cpuid_type == hygon) && _family == 0xf) { + infos->cpufamilynumber = _family + _extendedfamily; + } else { + infos->cpufamilynumber = _family; + } + if ((cpuid_type == intel && (_family == 0x6 || _family == 0xf)) + || ((cpuid_type == amd || cpuid_type == hygon) && _family == 0xf) + || (cpuid_type == zhaoxin && (_family == 0x6 || _family == 0x7))) { + infos->cpumodelnumber = _model + (_extendedmodel << 4); + } else { + infos->cpumodelnumber = _model; + } + infos->cpustepping = eax & 0xf; + + if (cpuid_type == intel && infos->cpufamilynumber == 0x6 && + (infos->cpumodelnumber == 0x57 || infos->cpumodelnumber == 0x85)) + data->is_knl = 1; /* KNM is the same as KNL */ + + /* Get cpu vendor string from cpuid 0x00 */ + memset(regs, 0, sizeof(regs)); + regs[0] = 0; + cpuid_or_from_dump(®s[0], ®s[1], ®s[3], ®s[2], src_cpuiddump); + memcpy(infos->cpuvendor, regs+1, 4*3); + /* infos was calloc'ed, already ends with \0 */ + + /* Get cpu model string from cpuid 0x80000002-4 */ + if (highest_ext_cpuid >= 0x80000004) { + memset(regs, 0, sizeof(regs)); + regs[0] = 0x80000002; + cpuid_or_from_dump(®s[0], ®s[1], ®s[2], ®s[3], src_cpuiddump); + memcpy(infos->cpumodel, regs, 4*4); + regs[0] = 0x80000003; + cpuid_or_from_dump(®s[0], ®s[1], ®s[2], ®s[3], src_cpuiddump); + memcpy(infos->cpumodel + 4*4, regs, 4*4); + regs[0] = 0x80000004; + cpuid_or_from_dump(®s[0], ®s[1], ®s[2], ®s[3], src_cpuiddump); + memcpy(infos->cpumodel + 4*4*2, regs, 4*4); + /* infos was calloc'ed, already ends with \0 */ + } + + /* Get core/thread information from cpuid 0x80000008 + * (not supported on Intel) + */ + if (cpuid_type != intel && cpuid_type != zhaoxin && highest_ext_cpuid >= 0x80000008) { + unsigned max_nbcores; + unsigned max_nbthreads; + unsigned coreidsize; + unsigned logprocid; + eax = 0x80000008; + cpuid_or_from_dump(&eax, &ebx, &ecx, &edx, src_cpuiddump); + coreidsize = (ecx >> 12) & 0xf; + hwloc_debug("core ID size: %u\n", coreidsize); + if (!coreidsize) { + max_nbcores = (ecx & 0xff) + 1; + } else + max_nbcores = 1 << coreidsize; + hwloc_debug("Thus max # of cores: %u\n", max_nbcores); + /* Still no multithreaded AMD */ + max_nbthreads = 1 ; + hwloc_debug("and max # of threads: %u\n", max_nbthreads); + /* legacy_max_log_proc is deprecated, it can be smaller than max_nbcores, + * which is the maximum number of cores that the processor could theoretically support + * (see "Multiple Core Calculation" in the AMD CPUID specification). + * Recompute packageid/threadid/coreid accordingly. + */ + infos->packageid = infos->apicid / max_nbcores; + logprocid = infos->apicid % max_nbcores; + infos->threadid = logprocid % max_nbthreads; + infos->coreid = logprocid / max_nbthreads; + hwloc_debug("this is thread %u of core %u\n", infos->threadid, infos->coreid); + } + + infos->numcaches = 0; + infos->cache = NULL; + + /* Get apicid, nodeid, unitid from cpuid 0x8000001e + * and cache information from cpuid 0x8000001d + * (AMD topology extension) + */ + if (cpuid_type != intel && cpuid_type != zhaoxin && has_topoext(features)) { + unsigned apic_id, node_id, nodes_per_proc; + + /* the code below doesn't want any other cache yet */ + assert(!infos->numcaches); + + eax = 0x8000001e; + cpuid_or_from_dump(&eax, &ebx, &ecx, &edx, src_cpuiddump); + infos->apicid = apic_id = eax; + + if (infos->cpufamilynumber == 0x16) { + /* ecx is reserved */ + node_id = 0; + nodes_per_proc = 1; + } else { + /* AMD other families or Hygon family 18h */ + node_id = ecx & 0xff; + nodes_per_proc = ((ecx >> 8) & 7) + 1; + } + infos->nodeid = node_id; + if ((infos->cpufamilynumber == 0x15 && nodes_per_proc > 2) + || ((infos->cpufamilynumber == 0x17 || infos->cpufamilynumber == 0x18) && nodes_per_proc > 4)) { + hwloc_debug("warning: undefined nodes_per_proc value %u, assuming it means %u\n", nodes_per_proc, nodes_per_proc); + } + + if (infos->cpufamilynumber <= 0x16) { /* topoext appeared in 0x15 and compute-units were only used in 0x15 and 0x16 */ + unsigned unit_id, cores_per_unit; + infos->unitid = unit_id = ebx & 0xff; + cores_per_unit = ((ebx >> 8) & 0xff) + 1; + hwloc_debug("topoext %08x, %u nodes, node %u, %u cores in unit %u\n", apic_id, nodes_per_proc, node_id, cores_per_unit, unit_id); + /* coreid and unitid are package-wide (core 0-15 and unit 0-7 on 16-core 2-NUMAnode processor). + * The Linux kernel reduces theses to NUMA-node-wide (by applying %core_per_node and %unit_per node respectively). + * It's not clear if we should do this as well. + */ + } else { + unsigned core_id, threads_per_core; + infos->coreid = core_id = ebx & 0xff; + threads_per_core = ((ebx >> 8) & 0xff) + 1; + hwloc_debug("topoext %08x, %u nodes, node %u, %u threads in core %u\n", apic_id, nodes_per_proc, node_id, threads_per_core, core_id); + } + + for (cachenum = 0; ; cachenum++) { + eax = 0x8000001d; + ecx = cachenum; + cpuid_or_from_dump(&eax, &ebx, &ecx, &edx, src_cpuiddump); + if ((eax & 0x1f) == 0) + break; + infos->numcaches++; + } + + cache = infos->cache = malloc(infos->numcaches * sizeof(*infos->cache)); + if (cache) { + for (cachenum = 0; ; cachenum++) { + unsigned long linesize, linepart, ways, sets; + eax = 0x8000001d; + ecx = cachenum; + cpuid_or_from_dump(&eax, &ebx, &ecx, &edx, src_cpuiddump); + + if ((eax & 0x1f) == 0) + break; + switch (eax & 0x1f) { + case 1: cache->type = HWLOC_OBJ_CACHE_DATA; break; + case 2: cache->type = HWLOC_OBJ_CACHE_INSTRUCTION; break; + default: cache->type = HWLOC_OBJ_CACHE_UNIFIED; break; + } + + cache->level = (eax >> 5) & 0x7; + /* Note: actually number of cores */ + cache->nbthreads_sharing = ((eax >> 14) & 0xfff) + 1; + + cache->linesize = linesize = (ebx & 0xfff) + 1; + cache->linepart = linepart = ((ebx >> 12) & 0x3ff) + 1; + ways = ((ebx >> 22) & 0x3ff) + 1; + + if (eax & (1 << 9)) + /* Fully associative */ + cache->ways = -1; + else + cache->ways = ways; + cache->sets = sets = ecx + 1; + cache->size = linesize * linepart * ways * sets; + cache->inclusive = edx & 0x2; + + hwloc_debug("cache %u L%u%c t%u linesize %lu linepart %lu ways %lu sets %lu, size %luKB\n", + cachenum, cache->level, + cache->type == HWLOC_OBJ_CACHE_DATA ? 'd' : cache->type == HWLOC_OBJ_CACHE_INSTRUCTION ? 'i' : 'u', + cache->nbthreads_sharing, linesize, linepart, ways, sets, cache->size >> 10); + + cache++; + } + } else { + infos->numcaches = 0; + } + } else { + /* If there's no topoext, + * get cache information from cpuid 0x80000005 and 0x80000006 + * (not supported on Intel) + */ + if (cpuid_type != intel && cpuid_type != zhaoxin && highest_ext_cpuid >= 0x80000005) { + eax = 0x80000005; + cpuid_or_from_dump(&eax, &ebx, &ecx, &edx, src_cpuiddump); + fill_amd_cache(infos, 1, HWLOC_OBJ_CACHE_DATA, 1, ecx); /* private L1d */ + fill_amd_cache(infos, 1, HWLOC_OBJ_CACHE_INSTRUCTION, 1, edx); /* private L1i */ + } + if (cpuid_type != intel && cpuid_type != zhaoxin && highest_ext_cpuid >= 0x80000006) { + eax = 0x80000006; + cpuid_or_from_dump(&eax, &ebx, &ecx, &edx, src_cpuiddump); + if (ecx & 0xf000) + /* This is actually supported on Intel but LinePerTag isn't returned in bits 8-11. + * Could be useful if some Intels (at least before Core micro-architecture) + * support this leaf without leaf 0x4. + */ + fill_amd_cache(infos, 2, HWLOC_OBJ_CACHE_UNIFIED, 1, ecx); /* private L2u */ + if (edx & 0xf000) + fill_amd_cache(infos, 3, HWLOC_OBJ_CACHE_UNIFIED, legacy_max_log_proc, edx); /* package-wide L3u */ + } + } + + /* Get thread/core + cache information from cpuid 0x04 + * (not supported on AMD) + */ + if ((cpuid_type != amd && cpuid_type != hygon) && highest_cpuid >= 0x04) { + unsigned max_nbcores; + unsigned max_nbthreads; + unsigned level; + struct cacheinfo *tmpcaches; + unsigned oldnumcaches = infos->numcaches; /* in case we got caches above */ + + for (cachenum = 0; ; cachenum++) { + eax = 0x04; + ecx = cachenum; + cpuid_or_from_dump(&eax, &ebx, &ecx, &edx, src_cpuiddump); + + hwloc_debug("cache %u type %u\n", cachenum, eax & 0x1f); + if ((eax & 0x1f) == 0) + break; + level = (eax >> 5) & 0x7; + if (data->is_knl && level == 3) + /* KNL reports wrong L3 information (size always 0, cpuset always the entire machine, ignore it */ + break; + infos->numcaches++; + + if (!cachenum) { + /* by the way, get thread/core information from the first cache */ + max_nbcores = ((eax >> 26) & 0x3f) + 1; + max_nbthreads = legacy_max_log_proc / max_nbcores; + hwloc_debug("thus %u threads\n", max_nbthreads); + infos->threadid = legacy_log_proc_id % max_nbthreads; + infos->coreid = legacy_log_proc_id / max_nbthreads; + hwloc_debug("this is thread %u of core %u\n", infos->threadid, infos->coreid); + } + } + + tmpcaches = realloc(infos->cache, infos->numcaches * sizeof(*infos->cache)); + if (!tmpcaches) { + infos->numcaches = oldnumcaches; + } else { + infos->cache = tmpcaches; + cache = &infos->cache[oldnumcaches]; + + for (cachenum = 0; ; cachenum++) { + unsigned long linesize, linepart, ways, sets; + eax = 0x04; + ecx = cachenum; + cpuid_or_from_dump(&eax, &ebx, &ecx, &edx, src_cpuiddump); + + if ((eax & 0x1f) == 0) + break; + level = (eax >> 5) & 0x7; + if (data->is_knl && level == 3) + /* KNL reports wrong L3 information (size always 0, cpuset always the entire machine, ignore it */ + break; + switch (eax & 0x1f) { + case 1: cache->type = HWLOC_OBJ_CACHE_DATA; break; + case 2: cache->type = HWLOC_OBJ_CACHE_INSTRUCTION; break; + default: cache->type = HWLOC_OBJ_CACHE_UNIFIED; break; + } + + cache->level = level; + cache->nbthreads_sharing = ((eax >> 14) & 0xfff) + 1; + + cache->linesize = linesize = (ebx & 0xfff) + 1; + cache->linepart = linepart = ((ebx >> 12) & 0x3ff) + 1; + ways = ((ebx >> 22) & 0x3ff) + 1; + if (eax & (1 << 9)) + /* Fully associative */ + cache->ways = -1; + else + cache->ways = ways; + cache->sets = sets = ecx + 1; + cache->size = linesize * linepart * ways * sets; + cache->inclusive = edx & 0x2; + + hwloc_debug("cache %u L%u%c t%u linesize %lu linepart %lu ways %lu sets %lu, size %luKB\n", + cachenum, cache->level, + cache->type == HWLOC_OBJ_CACHE_DATA ? 'd' : cache->type == HWLOC_OBJ_CACHE_INSTRUCTION ? 'i' : 'u', + cache->nbthreads_sharing, linesize, linepart, ways, sets, cache->size >> 10); + cache++; + } + } + } + + if ((cpuid_type == intel) && highest_cpuid >= 0x1f) { + /* Get package/die/module/tile/core/thread information from cpuid 0x1f + * (Intel v2 Extended Topology Enumeration) + */ + look_exttopoenum(infos, 0x1f, src_cpuiddump); + + } else if ((cpuid_type == intel || cpuid_type == zhaoxin) && highest_cpuid >= 0x0b && has_x2apic(features)) { + /* Get package/core/thread information from cpuid 0x0b + * (Intel v1 Extended Topology Enumeration) + */ + look_exttopoenum(infos, 0x0b, src_cpuiddump); + } + + /* Now that we have all info, compute cacheids and apply quirks */ + for (cachenum = 0; cachenum < infos->numcaches; cachenum++) { + cache = &infos->cache[cachenum]; + + /* default cacheid value */ + cache->cacheid = infos->apicid / cache->nbthreads_sharing; + + if (cpuid_type == amd) { + /* AMD quirks */ + if (infos->cpufamilynumber == 0x17 + && cache->level == 3 && cache->nbthreads_sharing == 6) { + /* AMD family 0x17 always shares L3 between 8 APIC ids, + * even when only 6 APIC ids are enabled and reported in nbthreads_sharing + * (on 24-core CPUs). + */ + cache->cacheid = infos->apicid / 8; + + } else if (infos->cpufamilynumber== 0x10 && infos->cpumodelnumber == 0x9 + && cache->level == 3 + && (cache->ways == -1 || (cache->ways % 2 == 0)) && cache->nbthreads_sharing >= 8) { + /* Fix AMD family 0x10 model 0x9 (Magny-Cours) with 8 or 12 cores. + * The L3 (and its associativity) is actually split into two halves). + */ + if (cache->nbthreads_sharing == 16) + cache->nbthreads_sharing = 12; /* nbthreads_sharing is a power of 2 but the processor actually has 8 or 12 cores */ + cache->nbthreads_sharing /= 2; + cache->size /= 2; + if (cache->ways != -1) + cache->ways /= 2; + /* AMD Magny-Cours 12-cores processor reserve APIC ids as AAAAAABBBBBB.... + * among first L3 (A), second L3 (B), and unexisting cores (.). + * On multi-socket servers, L3 in non-first sockets may have APIC id ranges + * such as [16-21] that are not aligned on multiple of nbthreads_sharing (6). + * That means, we can't just compare apicid/nbthreads_sharing to identify siblings. + */ + cache->cacheid = (infos->apicid % legacy_max_log_proc) / cache->nbthreads_sharing /* cacheid within the package */ + + 2 * (infos->apicid / legacy_max_log_proc); /* add 2 caches per previous package */ + + } else if (infos->cpufamilynumber == 0x15 + && (infos->cpumodelnumber == 0x1 /* Bulldozer */ || infos->cpumodelnumber == 0x2 /* Piledriver */) + && cache->level == 3 && cache->nbthreads_sharing == 6) { + /* AMD Bulldozer and Piledriver 12-core processors have same APIC ids as Magny-Cours below, + * but we can't merge the checks because the original nbthreads_sharing must be exactly 6 here. + */ + cache->cacheid = (infos->apicid % legacy_max_log_proc) / cache->nbthreads_sharing /* cacheid within the package */ + + 2 * (infos->apicid / legacy_max_log_proc); /* add 2 cache per previous package */ + } + } else if (cpuid_type == hygon) { + if (infos->cpufamilynumber == 0x18 + && cache->level == 3 && cache->nbthreads_sharing == 6) { + /* Hygon family 0x18 always shares L3 between 8 APIC ids, + * even when only 6 APIC ids are enabled and reported in nbthreads_sharing + * (on 24-core CPUs). + */ + cache->cacheid = infos->apicid / 8; + } + } + } + + if (hwloc_bitmap_isset(data->apicid_set, infos->apicid)) + data->apicid_unique = 0; + else + hwloc_bitmap_set(data->apicid_set, infos->apicid); +} + +static void +hwloc_x86_add_cpuinfos(hwloc_obj_t obj, struct procinfo *info, int replace) +{ + char number[12]; + if (info->cpuvendor[0]) + hwloc__add_info_nodup(&obj->infos, &obj->infos_count, "CPUVendor", info->cpuvendor, replace); + snprintf(number, sizeof(number), "%u", info->cpufamilynumber); + hwloc__add_info_nodup(&obj->infos, &obj->infos_count, "CPUFamilyNumber", number, replace); + snprintf(number, sizeof(number), "%u", info->cpumodelnumber); + hwloc__add_info_nodup(&obj->infos, &obj->infos_count, "CPUModelNumber", number, replace); + if (info->cpumodel[0]) { + const char *c = info->cpumodel; + while (*c == ' ') + c++; + hwloc__add_info_nodup(&obj->infos, &obj->infos_count, "CPUModel", c, replace); + } + snprintf(number, sizeof(number), "%u", info->cpustepping); + hwloc__add_info_nodup(&obj->infos, &obj->infos_count, "CPUStepping", number, replace); +} + +/* Analyse information stored in infos, and build/annotate topology levels accordingly */ +static void summarize(struct hwloc_backend *backend, struct procinfo *infos, int fulldiscovery) +{ + struct hwloc_topology *topology = backend->topology; + struct hwloc_x86_backend_data_s *data = backend->private_data; + unsigned nbprocs = data->nbprocs; + hwloc_bitmap_t complete_cpuset = hwloc_bitmap_alloc(); + unsigned i, j, l, level; + int one = -1; + hwloc_bitmap_t remaining_cpuset; + int gotnuma = 0; + + for (i = 0; i < nbprocs; i++) + if (infos[i].present) { + hwloc_bitmap_set(complete_cpuset, i); + one = i; + } + + if (one == -1) { + hwloc_bitmap_free(complete_cpuset); + return; + } + + remaining_cpuset = hwloc_bitmap_alloc(); + + /* Ideally, when fulldiscovery=0, we could add any object that doesn't exist yet. + * But what if the x86 and the native backends disagree because one is buggy? Which one to trust? + * We only add missing caches, and annotate other existing objects for now. + */ + + if (hwloc_filter_check_keep_object_type(topology, HWLOC_OBJ_PACKAGE)) { + /* Look for packages */ + hwloc_obj_t package; + + hwloc_bitmap_copy(remaining_cpuset, complete_cpuset); + while ((i = hwloc_bitmap_first(remaining_cpuset)) != (unsigned) -1) { + if (fulldiscovery) { + unsigned packageid = infos[i].packageid; + hwloc_bitmap_t package_cpuset = hwloc_bitmap_alloc(); + + for (j = i; j < nbprocs; j++) { + if (infos[j].packageid == packageid) { + hwloc_bitmap_set(package_cpuset, j); + hwloc_bitmap_clr(remaining_cpuset, j); + } + } + package = hwloc_alloc_setup_object(topology, HWLOC_OBJ_PACKAGE, packageid); + package->cpuset = package_cpuset; + + hwloc_x86_add_cpuinfos(package, &infos[i], 0); + + hwloc_debug_1arg_bitmap("os package %u has cpuset %s\n", + packageid, package_cpuset); + hwloc_insert_object_by_cpuset(topology, package); + + } else { + /* Annotate packages previously-existing packages */ + hwloc_bitmap_t set = hwloc_bitmap_alloc(); + hwloc_bitmap_set(set, i); + package = hwloc_get_next_obj_covering_cpuset_by_type(topology, set, HWLOC_OBJ_PACKAGE, NULL); + hwloc_bitmap_free(set); + if (package) { + /* Found package above that PU, annotate if no such attribute yet */ + hwloc_x86_add_cpuinfos(package, &infos[i], 1); + hwloc_bitmap_andnot(remaining_cpuset, remaining_cpuset, package->cpuset); + } else { + /* No package, annotate the root object */ + hwloc_x86_add_cpuinfos(hwloc_get_root_obj(topology), &infos[i], 1); + break; + } + } + } + } + + /* Look for Numa nodes inside packages (cannot be filtered-out) */ + if (fulldiscovery && getenv("HWLOC_X86_TOPOEXT_NUMANODES")) { + hwloc_bitmap_t node_cpuset; + hwloc_obj_t node; + + /* FIXME: if there's memory inside the root object, divide it into NUMA nodes? */ + + hwloc_bitmap_copy(remaining_cpuset, complete_cpuset); + while ((i = hwloc_bitmap_first(remaining_cpuset)) != (unsigned) -1) { + unsigned packageid = infos[i].packageid; + unsigned nodeid = infos[i].nodeid; + + if (nodeid == (unsigned)-1) { + hwloc_bitmap_clr(remaining_cpuset, i); + continue; + } + + node_cpuset = hwloc_bitmap_alloc(); + for (j = i; j < nbprocs; j++) { + if (infos[j].nodeid == (unsigned) -1) { + hwloc_bitmap_clr(remaining_cpuset, j); + continue; + } + + if (infos[j].packageid == packageid && infos[j].nodeid == nodeid) { + hwloc_bitmap_set(node_cpuset, j); + hwloc_bitmap_clr(remaining_cpuset, j); + } + } + node = hwloc_alloc_setup_object(topology, HWLOC_OBJ_NUMANODE, nodeid); + node->cpuset = node_cpuset; + node->nodeset = hwloc_bitmap_alloc(); + hwloc_bitmap_set(node->nodeset, nodeid); + hwloc_debug_1arg_bitmap("os node %u has cpuset %s\n", + nodeid, node_cpuset); + hwloc_insert_object_by_cpuset(topology, node); + gotnuma++; + } + } + + if (hwloc_filter_check_keep_object_type(topology, HWLOC_OBJ_GROUP)) { + if (fulldiscovery) { + char *env; + int dont_merge; + hwloc_bitmap_t unit_cpuset, die_cpuset; + hwloc_obj_t unit, die; + + /* Look for Compute units inside packages */ + hwloc_bitmap_copy(remaining_cpuset, complete_cpuset); + while ((i = hwloc_bitmap_first(remaining_cpuset)) != (unsigned) -1) { + unsigned packageid = infos[i].packageid; + unsigned unitid = infos[i].unitid; + + if (unitid == (unsigned)-1) { + hwloc_bitmap_clr(remaining_cpuset, i); + continue; + } + + unit_cpuset = hwloc_bitmap_alloc(); + for (j = i; j < nbprocs; j++) { + if (infos[j].unitid == (unsigned) -1) { + hwloc_bitmap_clr(remaining_cpuset, j); + continue; + } + + if (infos[j].packageid == packageid && infos[j].unitid == unitid) { + hwloc_bitmap_set(unit_cpuset, j); + hwloc_bitmap_clr(remaining_cpuset, j); + } + } + unit = hwloc_alloc_setup_object(topology, HWLOC_OBJ_GROUP, unitid); + unit->cpuset = unit_cpuset; + unit->subtype = strdup("ComputeUnit"); + unit->attr->group.kind = HWLOC_GROUP_KIND_AMD_COMPUTE_UNIT; + hwloc_debug_1arg_bitmap("os unit %u has cpuset %s\n", + unitid, unit_cpuset); + hwloc_insert_object_by_cpuset(topology, unit); + } + + /* Look for Dies inside packages */ + env = getenv("HWLOC_DONT_MERGE_DIE_GROUPS"); + dont_merge = env && atoi(env); + hwloc_bitmap_copy(remaining_cpuset, complete_cpuset); + while ((i = hwloc_bitmap_first(remaining_cpuset)) != (unsigned) -1) { + unsigned packageid = infos[i].packageid; + unsigned dieid = infos[i].dieid; + + if (dieid == (unsigned)-1) { + hwloc_bitmap_clr(remaining_cpuset, i); + continue; + } + + die_cpuset = hwloc_bitmap_alloc(); + for (j = i; j < nbprocs; j++) { + if (infos[j].dieid == (unsigned) -1) { + hwloc_bitmap_clr(remaining_cpuset, j); + continue; + } + + if (infos[j].packageid == packageid && infos[j].dieid == dieid) { + hwloc_bitmap_set(die_cpuset, j); + hwloc_bitmap_clr(remaining_cpuset, j); + } + } + die = hwloc_alloc_setup_object(topology, HWLOC_OBJ_GROUP, dieid); + die->cpuset = die_cpuset; + die->subtype = strdup("Die"); + die->attr->group.kind = HWLOC_GROUP_KIND_INTEL_DIE; + die->attr->group.dont_merge = dont_merge; + hwloc_debug_1arg_bitmap("os die %u has cpuset %s\n", + dieid, die_cpuset); + hwloc_insert_object_by_cpuset(topology, die); + } + + /* Look for unknown objects */ + if (infos[one].otherids) { + for (level = infos[one].levels-1; level <= infos[one].levels-1; level--) { + if (infos[one].otherids[level] != UINT_MAX) { + hwloc_bitmap_t unknown_cpuset; + hwloc_obj_t unknown_obj; + + hwloc_bitmap_copy(remaining_cpuset, complete_cpuset); + while ((i = hwloc_bitmap_first(remaining_cpuset)) != (unsigned) -1) { + unsigned unknownid = infos[i].otherids[level]; + + unknown_cpuset = hwloc_bitmap_alloc(); + for (j = i; j < nbprocs; j++) { + if (infos[j].otherids[level] == unknownid) { + hwloc_bitmap_set(unknown_cpuset, j); + hwloc_bitmap_clr(remaining_cpuset, j); + } + } + unknown_obj = hwloc_alloc_setup_object(topology, HWLOC_OBJ_GROUP, unknownid); + unknown_obj->cpuset = unknown_cpuset; + unknown_obj->attr->group.kind = HWLOC_GROUP_KIND_INTEL_EXTTOPOENUM_UNKNOWN; + unknown_obj->attr->group.subkind = level; + hwloc_debug_2args_bitmap("os unknown%u %u has cpuset %s\n", + level, unknownid, unknown_cpuset); + hwloc_insert_object_by_cpuset(topology, unknown_obj); + } + } + } + } + } + } + + if (hwloc_filter_check_keep_object_type(topology, HWLOC_OBJ_CORE)) { + /* Look for cores */ + if (fulldiscovery) { + hwloc_bitmap_t core_cpuset; + hwloc_obj_t core; + + hwloc_bitmap_copy(remaining_cpuset, complete_cpuset); + while ((i = hwloc_bitmap_first(remaining_cpuset)) != (unsigned) -1) { + unsigned packageid = infos[i].packageid; + unsigned nodeid = infos[i].nodeid; + unsigned coreid = infos[i].coreid; + + if (coreid == (unsigned) -1) { + hwloc_bitmap_clr(remaining_cpuset, i); + continue; + } + + core_cpuset = hwloc_bitmap_alloc(); + for (j = i; j < nbprocs; j++) { + if (infos[j].coreid == (unsigned) -1) { + hwloc_bitmap_clr(remaining_cpuset, j); + continue; + } + + if (infos[j].packageid == packageid && infos[j].nodeid == nodeid && infos[j].coreid == coreid) { + hwloc_bitmap_set(core_cpuset, j); + hwloc_bitmap_clr(remaining_cpuset, j); + } + } + core = hwloc_alloc_setup_object(topology, HWLOC_OBJ_CORE, coreid); + core->cpuset = core_cpuset; + hwloc_debug_1arg_bitmap("os core %u has cpuset %s\n", + coreid, core_cpuset); + hwloc_insert_object_by_cpuset(topology, core); + } + } + } + + /* Look for PUs (cannot be filtered-out) */ + if (fulldiscovery) { + hwloc_debug("%s", "\n\n * CPU cpusets *\n\n"); + for (i=0; icpuset = hwloc_bitmap_alloc(); + hwloc_bitmap_only(obj->cpuset, i); + hwloc_debug_1arg_bitmap("PU %u has cpuset %s\n", i, obj->cpuset); + hwloc_insert_object_by_cpuset(topology, obj); + } + } + + /* Look for caches */ + /* First find max level */ + level = 0; + for (i = 0; i < nbprocs; i++) + for (j = 0; j < infos[i].numcaches; j++) + if (infos[i].cache[j].level > level) + level = infos[i].cache[j].level; + while (level > 0) { + hwloc_obj_cache_type_t type; + HWLOC_BUILD_ASSERT(HWLOC_OBJ_CACHE_DATA == HWLOC_OBJ_CACHE_UNIFIED+1); + HWLOC_BUILD_ASSERT(HWLOC_OBJ_CACHE_INSTRUCTION == HWLOC_OBJ_CACHE_DATA+1); + for (type = HWLOC_OBJ_CACHE_UNIFIED; type <= HWLOC_OBJ_CACHE_INSTRUCTION; type++) { + /* Look for caches of that type at level level */ + hwloc_obj_type_t otype; + hwloc_obj_t cache; + + otype = hwloc_cache_type_by_depth_type(level, type); + if (otype == HWLOC_OBJ_TYPE_NONE) + continue; + if (!hwloc_filter_check_keep_object_type(topology, otype)) + continue; + + hwloc_bitmap_copy(remaining_cpuset, complete_cpuset); + while ((i = hwloc_bitmap_first(remaining_cpuset)) != (unsigned) -1) { + hwloc_bitmap_t puset; + + for (l = 0; l < infos[i].numcaches; l++) { + if (infos[i].cache[l].level == level && infos[i].cache[l].type == type) + break; + } + if (l == infos[i].numcaches) { + /* no cache Llevel of that type in i */ + hwloc_bitmap_clr(remaining_cpuset, i); + continue; + } + + puset = hwloc_bitmap_alloc(); + hwloc_bitmap_set(puset, i); + cache = hwloc_get_next_obj_covering_cpuset_by_type(topology, puset, otype, NULL); + hwloc_bitmap_free(puset); + + if (cache) { + /* Found cache above that PU, annotate if no such attribute yet */ + if (!hwloc_obj_get_info_by_name(cache, "Inclusive")) + hwloc_obj_add_info(cache, "Inclusive", infos[i].cache[l].inclusive ? "1" : "0"); + hwloc_bitmap_andnot(remaining_cpuset, remaining_cpuset, cache->cpuset); + } else { + /* Add the missing cache */ + hwloc_bitmap_t cache_cpuset; + unsigned packageid = infos[i].packageid; + unsigned cacheid = infos[i].cache[l].cacheid; + /* Now look for others sharing it */ + cache_cpuset = hwloc_bitmap_alloc(); + for (j = i; j < nbprocs; j++) { + unsigned l2; + for (l2 = 0; l2 < infos[j].numcaches; l2++) { + if (infos[j].cache[l2].level == level && infos[j].cache[l2].type == type) + break; + } + if (l2 == infos[j].numcaches) { + /* no cache Llevel of that type in j */ + hwloc_bitmap_clr(remaining_cpuset, j); + continue; + } + if (infos[j].packageid == packageid && infos[j].cache[l2].cacheid == cacheid) { + hwloc_bitmap_set(cache_cpuset, j); + hwloc_bitmap_clr(remaining_cpuset, j); + } + } + cache = hwloc_alloc_setup_object(topology, otype, HWLOC_UNKNOWN_INDEX); + cache->attr->cache.depth = level; + cache->attr->cache.size = infos[i].cache[l].size; + cache->attr->cache.linesize = infos[i].cache[l].linesize; + cache->attr->cache.associativity = infos[i].cache[l].ways; + cache->attr->cache.type = infos[i].cache[l].type; + cache->cpuset = cache_cpuset; + hwloc_obj_add_info(cache, "Inclusive", infos[i].cache[l].inclusive ? "1" : "0"); + hwloc_debug_2args_bitmap("os L%u cache %u has cpuset %s\n", + level, cacheid, cache_cpuset); + hwloc_insert_object_by_cpuset(topology, cache); + } + } + } + level--; + } + + /* FIXME: if KNL and L2 disabled, add tiles instead of L2 */ + + hwloc_bitmap_free(remaining_cpuset); + hwloc_bitmap_free(complete_cpuset); + + if (gotnuma) + topology->support.discovery->numa = 1; +} + +static int +look_procs(struct hwloc_backend *backend, struct procinfo *infos, int fulldiscovery, + unsigned highest_cpuid, unsigned highest_ext_cpuid, unsigned *features, enum cpuid_type cpuid_type, + int (*get_cpubind)(hwloc_topology_t topology, hwloc_cpuset_t set, int flags), + int (*set_cpubind)(hwloc_topology_t topology, hwloc_const_cpuset_t set, int flags)) +{ + struct hwloc_x86_backend_data_s *data = backend->private_data; + struct hwloc_topology *topology = backend->topology; + unsigned nbprocs = data->nbprocs; + hwloc_bitmap_t orig_cpuset = NULL; + hwloc_bitmap_t set = NULL; + unsigned i; + + if (!data->src_cpuiddump_path) { + orig_cpuset = hwloc_bitmap_alloc(); + if (get_cpubind(topology, orig_cpuset, HWLOC_CPUBIND_STRICT)) { + hwloc_bitmap_free(orig_cpuset); + return -1; + } + set = hwloc_bitmap_alloc(); + } + + for (i = 0; i < nbprocs; i++) { + struct cpuiddump *src_cpuiddump = NULL; + if (data->src_cpuiddump_path) { + src_cpuiddump = cpuiddump_read(data->src_cpuiddump_path, i); + if (!src_cpuiddump) + continue; + } else { + hwloc_bitmap_only(set, i); + hwloc_debug("binding to CPU%u\n", i); + if (set_cpubind(topology, set, HWLOC_CPUBIND_STRICT)) { + hwloc_debug("could not bind to CPU%u: %s\n", i, strerror(errno)); + continue; + } + } + + look_proc(backend, &infos[i], highest_cpuid, highest_ext_cpuid, features, cpuid_type, src_cpuiddump); + + if (data->src_cpuiddump_path) { + cpuiddump_free(src_cpuiddump); + } + } + + if (!data->src_cpuiddump_path) { + set_cpubind(topology, orig_cpuset, 0); + hwloc_bitmap_free(set); + hwloc_bitmap_free(orig_cpuset); + } + + if (!data->apicid_unique) + fulldiscovery = 0; + else + summarize(backend, infos, fulldiscovery); + return 0; +} + +#if defined HWLOC_FREEBSD_SYS && defined HAVE_CPUSET_SETID +#include +#include +typedef cpusetid_t hwloc_x86_os_state_t; +static void hwloc_x86_os_state_save(hwloc_x86_os_state_t *state, struct cpuiddump *src_cpuiddump) +{ + if (!src_cpuiddump) { + /* temporary make all cpus available during discovery */ + cpuset_getid(CPU_LEVEL_CPUSET, CPU_WHICH_PID, -1, state); + cpuset_setid(CPU_WHICH_PID, -1, 0); + } +} +static void hwloc_x86_os_state_restore(hwloc_x86_os_state_t *state, struct cpuiddump *src_cpuiddump) +{ + if (!src_cpuiddump) { + /* restore initial cpuset */ + cpuset_setid(CPU_WHICH_PID, -1, *state); + } +} +#else /* !defined HWLOC_FREEBSD_SYS || !defined HAVE_CPUSET_SETID */ +typedef void * hwloc_x86_os_state_t; +static void hwloc_x86_os_state_save(hwloc_x86_os_state_t *state __hwloc_attribute_unused, struct cpuiddump *src_cpuiddump __hwloc_attribute_unused) { } +static void hwloc_x86_os_state_restore(hwloc_x86_os_state_t *state __hwloc_attribute_unused, struct cpuiddump *src_cpuiddump __hwloc_attribute_unused) { } +#endif /* !defined HWLOC_FREEBSD_SYS || !defined HAVE_CPUSET_SETID */ + +/* GenuineIntel */ +#define INTEL_EBX ('G' | ('e'<<8) | ('n'<<16) | ('u'<<24)) +#define INTEL_EDX ('i' | ('n'<<8) | ('e'<<16) | ('I'<<24)) +#define INTEL_ECX ('n' | ('t'<<8) | ('e'<<16) | ('l'<<24)) + +/* AuthenticAMD */ +#define AMD_EBX ('A' | ('u'<<8) | ('t'<<16) | ('h'<<24)) +#define AMD_EDX ('e' | ('n'<<8) | ('t'<<16) | ('i'<<24)) +#define AMD_ECX ('c' | ('A'<<8) | ('M'<<16) | ('D'<<24)) + +/* HYGON "HygonGenuine" */ +#define HYGON_EBX ('H' | ('y'<<8) | ('g'<<16) | ('o'<<24)) +#define HYGON_EDX ('n' | ('G'<<8) | ('e'<<16) | ('n'<<24)) +#define HYGON_ECX ('u' | ('i'<<8) | ('n'<<16) | ('e'<<24)) + +/* (Zhaoxin) CentaurHauls */ +#define ZX_EBX ('C' | ('e'<<8) | ('n'<<16) | ('t'<<24)) +#define ZX_EDX ('a' | ('u'<<8) | ('r'<<16) | ('H'<<24)) +#define ZX_ECX ('a' | ('u'<<8) | ('l'<<16) | ('s'<<24)) +/* (Zhaoxin) Shanghai */ +#define SH_EBX (' ' | (' '<<8) | ('S'<<16) | ('h'<<24)) +#define SH_EDX ('a' | ('n'<<8) | ('g'<<16) | ('h'<<24)) +#define SH_ECX ('a' | ('i'<<8) | (' '<<16) | (' '<<24)) + +/* fake cpubind for when nbprocs=1 and no binding support */ +static int fake_get_cpubind(hwloc_topology_t topology __hwloc_attribute_unused, + hwloc_cpuset_t set __hwloc_attribute_unused, + int flags __hwloc_attribute_unused) +{ + return 0; +} +static int fake_set_cpubind(hwloc_topology_t topology __hwloc_attribute_unused, + hwloc_const_cpuset_t set __hwloc_attribute_unused, + int flags __hwloc_attribute_unused) +{ + return 0; +} + +static +int hwloc_look_x86(struct hwloc_backend *backend, int fulldiscovery) +{ + struct hwloc_x86_backend_data_s *data = backend->private_data; + unsigned nbprocs = data->nbprocs; + unsigned eax, ebx, ecx = 0, edx; + unsigned i; + unsigned highest_cpuid; + unsigned highest_ext_cpuid; + /* This stores cpuid features with the same indexing as Linux */ + unsigned features[10] = { 0 }; + struct procinfo *infos = NULL; + enum cpuid_type cpuid_type = unknown; + hwloc_x86_os_state_t os_state; + struct hwloc_binding_hooks hooks; + struct hwloc_topology_support support; + struct hwloc_topology_membind_support memsupport __hwloc_attribute_unused; + int (*get_cpubind)(hwloc_topology_t topology, hwloc_cpuset_t set, int flags) = NULL; + int (*set_cpubind)(hwloc_topology_t topology, hwloc_const_cpuset_t set, int flags) = NULL; + struct cpuiddump *src_cpuiddump = NULL; + int ret = -1; + + if (data->src_cpuiddump_path) { + /* just read cpuid from the dump */ + src_cpuiddump = cpuiddump_read(data->src_cpuiddump_path, 0); + if (!src_cpuiddump) + goto out; + + } else { + /* otherwise check if binding works */ + memset(&hooks, 0, sizeof(hooks)); + support.membind = &memsupport; + hwloc_set_native_binding_hooks(&hooks, &support); + if (hooks.get_thisthread_cpubind && hooks.set_thisthread_cpubind) { + get_cpubind = hooks.get_thisthread_cpubind; + set_cpubind = hooks.set_thisthread_cpubind; + } else if (hooks.get_thisproc_cpubind && hooks.set_thisproc_cpubind) { + /* FIXME: if called by a multithreaded program, we will restore the original process binding + * for each thread instead of their own original thread binding. + * See issue #158. + */ + get_cpubind = hooks.get_thisproc_cpubind; + set_cpubind = hooks.set_thisproc_cpubind; + } else { + /* we need binding support if there are multiple PUs */ + if (nbprocs > 1) + goto out; + get_cpubind = fake_get_cpubind; + set_cpubind = fake_set_cpubind; + } + } + + if (!src_cpuiddump && !hwloc_have_x86_cpuid()) + goto out; + + infos = calloc(nbprocs, sizeof(struct procinfo)); + if (NULL == infos) + goto out; + for (i = 0; i < nbprocs; i++) { + infos[i].nodeid = (unsigned) -1; + infos[i].packageid = (unsigned) -1; + infos[i].dieid = (unsigned) -1; + infos[i].unitid = (unsigned) -1; + infos[i].coreid = (unsigned) -1; + infos[i].threadid = (unsigned) -1; + } + + eax = 0x00; + cpuid_or_from_dump(&eax, &ebx, &ecx, &edx, src_cpuiddump); + highest_cpuid = eax; + if (ebx == INTEL_EBX && ecx == INTEL_ECX && edx == INTEL_EDX) + cpuid_type = intel; + else if (ebx == AMD_EBX && ecx == AMD_ECX && edx == AMD_EDX) + cpuid_type = amd; + else if ((ebx == ZX_EBX && ecx == ZX_ECX && edx == ZX_EDX) + || (ebx == SH_EBX && ecx == SH_ECX && edx == SH_EDX)) + cpuid_type = zhaoxin; + else if (ebx == HYGON_EBX && ecx == HYGON_ECX && edx == HYGON_EDX) + cpuid_type = hygon; + + hwloc_debug("highest cpuid %x, cpuid type %u\n", highest_cpuid, cpuid_type); + if (highest_cpuid < 0x01) { + goto out_with_infos; + } + + eax = 0x01; + cpuid_or_from_dump(&eax, &ebx, &ecx, &edx, src_cpuiddump); + features[0] = edx; + features[4] = ecx; + + eax = 0x80000000; + cpuid_or_from_dump(&eax, &ebx, &ecx, &edx, src_cpuiddump); + highest_ext_cpuid = eax; + + hwloc_debug("highest extended cpuid %x\n", highest_ext_cpuid); + + if (highest_cpuid >= 0x7) { + eax = 0x7; + ecx = 0; + cpuid_or_from_dump(&eax, &ebx, &ecx, &edx, src_cpuiddump); + features[9] = ebx; + } + + if (cpuid_type != intel && highest_ext_cpuid >= 0x80000001) { + eax = 0x80000001; + cpuid_or_from_dump(&eax, &ebx, &ecx, &edx, src_cpuiddump); + features[1] = edx; + features[6] = ecx; + } + + hwloc_x86_os_state_save(&os_state, src_cpuiddump); + + ret = look_procs(backend, infos, fulldiscovery, + highest_cpuid, highest_ext_cpuid, features, cpuid_type, + get_cpubind, set_cpubind); + if (!ret) + /* success, we're done */ + goto out_with_os_state; + + if (nbprocs == 1) { + /* only one processor, no need to bind */ + look_proc(backend, &infos[0], highest_cpuid, highest_ext_cpuid, features, cpuid_type, src_cpuiddump); + summarize(backend, infos, fulldiscovery); + ret = 0; + } + +out_with_os_state: + hwloc_x86_os_state_restore(&os_state, src_cpuiddump); + +out_with_infos: + if (NULL != infos) { + for (i = 0; i < nbprocs; i++) { + free(infos[i].cache); + free(infos[i].otherids); + } + free(infos); + } + +out: + if (src_cpuiddump) + cpuiddump_free(src_cpuiddump); + return ret; +} + +static int +hwloc_x86_discover(struct hwloc_backend *backend) +{ + struct hwloc_x86_backend_data_s *data = backend->private_data; + struct hwloc_topology *topology = backend->topology; + int alreadypus = 0; + int ret; + +#if HAVE_DECL_RUNNING_ON_VALGRIND + if (RUNNING_ON_VALGRIND && !data->src_cpuiddump_path) { + fprintf(stderr, "hwloc x86 backend cannot work under Valgrind, disabling.\n" + "May be reenabled by dumping CPUIDs with hwloc-gather-cpuid\n" + "and reloading them under Valgrind with HWLOC_CPUID_PATH.\n"); + return 0; + } +#endif + + if (data->src_cpuiddump_path) { + assert(data->nbprocs > 0); /* enforced by hwloc_x86_component_instantiate() */ + topology->support.discovery->pu = 1; + } else { + int nbprocs = hwloc_fallback_nbprocessors(topology); + if (nbprocs >= 1) + topology->support.discovery->pu = 1; + else + nbprocs = 1; + data->nbprocs = (unsigned) nbprocs; + } + + if (topology->levels[0][0]->cpuset) { + /* somebody else discovered things */ + if (topology->nb_levels == 2 && topology->level_nbobjects[1] == data->nbprocs) { + /* only PUs were discovered, as much as we would, complete the topology with everything else */ + alreadypus = 1; + goto fulldiscovery; + } + + /* several object types were added, we can't easily complete, just do partial discovery */ + hwloc_topology_reconnect(topology, 0); + ret = hwloc_look_x86(backend, 0); + if (ret) + hwloc_obj_add_info(topology->levels[0][0], "Backend", "x86"); + return 0; + } else { + /* topology is empty, initialize it */ + hwloc_alloc_root_sets(topology->levels[0][0]); + } + +fulldiscovery: + if (hwloc_look_x86(backend, 1) < 0) { + /* if failed, create PUs */ + if (!alreadypus) + hwloc_setup_pu_level(topology, data->nbprocs); + } + + hwloc_obj_add_info(topology->levels[0][0], "Backend", "x86"); + + if (!data->src_cpuiddump_path) { /* CPUID dump works for both x86 and x86_64 */ +#ifdef HAVE_UNAME + hwloc_add_uname_info(topology, NULL); /* we already know is_thissystem() is true */ +#else + /* uname isn't available, manually setup the "Architecture" info */ +#ifdef HWLOC_X86_64_ARCH + hwloc_obj_add_info(topology->levels[0][0], "Architecture", "x86_64"); +#else + hwloc_obj_add_info(topology->levels[0][0], "Architecture", "x86"); +#endif +#endif + } + + return 1; +} + +static int +hwloc_x86_check_cpuiddump_input(const char *src_cpuiddump_path, hwloc_bitmap_t set) +{ + +#if !(defined HWLOC_WIN_SYS && !defined __MINGW32__ && !defined __CYGWIN__) /* needs a lot of work */ + struct dirent *dirent; + DIR *dir; + FILE *file; + char line [32]; + + dir = opendir(src_cpuiddump_path); + if (!dir) + return -1; + + char path[strlen(src_cpuiddump_path) + strlen("/hwloc-cpuid-info") + 1]; + sprintf(path, "%s/hwloc-cpuid-info", src_cpuiddump_path); + file = fopen(path, "r"); + if (!file) { + fprintf(stderr, "Couldn't open dumped cpuid summary %s\n", path); + goto out_with_dir; + } + if (!fgets(line, sizeof(line), file)) { + fprintf(stderr, "Found read dumped cpuid summary in %s\n", path); + fclose(file); + goto out_with_dir; + } + fclose(file); + if (strcmp(line, "Architecture: x86\n")) { + fprintf(stderr, "Found non-x86 dumped cpuid summary in %s: %s\n", path, line); + goto out_with_dir; + } + + while ((dirent = readdir(dir)) != NULL) { + if (!strncmp(dirent->d_name, "pu", 2)) { + char *end; + unsigned long idx = strtoul(dirent->d_name+2, &end, 10); + if (!*end) + hwloc_bitmap_set(set, idx); + else + fprintf(stderr, "Ignoring invalid dirent `%s' in dumped cpuid directory `%s'\n", + dirent->d_name, src_cpuiddump_path); + } + } + closedir(dir); + + if (hwloc_bitmap_iszero(set)) { + fprintf(stderr, "Did not find any valid pu%%u entry in dumped cpuid directory `%s'\n", + src_cpuiddump_path); + return -1; + } else if (hwloc_bitmap_last(set) != hwloc_bitmap_weight(set) - 1) { + /* The x86 backends enforces contigous set of PUs starting at 0 so far */ + fprintf(stderr, "Found non-contigous pu%%u range in dumped cpuid directory `%s'\n", + src_cpuiddump_path); + return -1; + } + + return 0; + +out_with_dir: + closedir(dir); +#endif /* HWLOC_WIN_SYS & !__MINGW32__ needs a lot of work */ + return -1; +} + +static void +hwloc_x86_backend_disable(struct hwloc_backend *backend) +{ + struct hwloc_x86_backend_data_s *data = backend->private_data; + hwloc_bitmap_free(data->apicid_set); + free(data->src_cpuiddump_path); + free(data); +} + +static struct hwloc_backend * +hwloc_x86_component_instantiate(struct hwloc_disc_component *component, + const void *_data1 __hwloc_attribute_unused, + const void *_data2 __hwloc_attribute_unused, + const void *_data3 __hwloc_attribute_unused) +{ + struct hwloc_backend *backend; + struct hwloc_x86_backend_data_s *data; + const char *src_cpuiddump_path; + + backend = hwloc_backend_alloc(component); + if (!backend) + goto out; + + data = malloc(sizeof(*data)); + if (!data) { + errno = ENOMEM; + goto out_with_backend; + } + + backend->private_data = data; + backend->discover = hwloc_x86_discover; + backend->disable = hwloc_x86_backend_disable; + + /* default values */ + data->is_knl = 0; + data->apicid_set = hwloc_bitmap_alloc(); + data->apicid_unique = 1; + data->src_cpuiddump_path = NULL; + + src_cpuiddump_path = getenv("HWLOC_CPUID_PATH"); + if (src_cpuiddump_path) { + hwloc_bitmap_t set = hwloc_bitmap_alloc(); + if (!hwloc_x86_check_cpuiddump_input(src_cpuiddump_path, set)) { + backend->is_thissystem = 0; + data->src_cpuiddump_path = strdup(src_cpuiddump_path); + assert(!hwloc_bitmap_iszero(set)); /* enforced by hwloc_x86_check_cpuiddump_input() */ + data->nbprocs = hwloc_bitmap_weight(set); + } else { + fprintf(stderr, "Ignoring dumped cpuid directory.\n"); + } + hwloc_bitmap_free(set); + } + + return backend; + + out_with_backend: + free(backend); + out: + return NULL; +} + +static struct hwloc_disc_component hwloc_x86_disc_component = { + HWLOC_DISC_COMPONENT_TYPE_CPU, + "x86", + HWLOC_DISC_COMPONENT_TYPE_GLOBAL, + hwloc_x86_component_instantiate, + 45, /* between native and no_os */ + 1, + NULL +}; + +const struct hwloc_component hwloc_x86_component = { + HWLOC_COMPONENT_ABI, + NULL, NULL, + HWLOC_COMPONENT_TYPE_DISC, + 0, + &hwloc_x86_disc_component +}; diff --git a/src/3rdparty/hwloc/src/topology-xml-nolibxml.c b/src/3rdparty/hwloc/src/topology-xml-nolibxml.c new file mode 100644 index 000000000..5a0d02da4 --- /dev/null +++ b/src/3rdparty/hwloc/src/topology-xml-nolibxml.c @@ -0,0 +1,919 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2018 Inria. All rights reserved. + * Copyright © 2009-2011 Université Bordeaux + * Copyright © 2009-2011 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include +#ifdef HAVE_UNISTD_H +#include +#endif + +/******************* + * Import routines * + *******************/ + +struct hwloc__nolibxml_backend_data_s { + size_t buflen; /* size of both buffer and copy buffers, set during backend_init() */ + char *buffer; /* allocated and filled during backend_init() */ + char *copy; /* allocated during backend_init(), used later during actual parsing */ +}; + +typedef struct hwloc__nolibxml_import_state_data_s { + char *tagbuffer; /* buffer containing the next tag */ + char *attrbuffer; /* buffer containing the next attribute of the current node */ + char *tagname; /* tag name of the current node */ + int closed; /* set if the current node is auto-closing */ +} __hwloc_attribute_may_alias * hwloc__nolibxml_import_state_data_t; + +static char * +hwloc__nolibxml_import_ignore_spaces(char *buffer) +{ + return buffer + strspn(buffer, " \t\n"); +} + +static int +hwloc__nolibxml_import_next_attr(hwloc__xml_import_state_t state, char **namep, char **valuep) +{ + hwloc__nolibxml_import_state_data_t nstate = (void*) state->data; + size_t namelen; + size_t len, escaped; + char *buffer, *value, *end; + + if (!nstate->attrbuffer) + return -1; + + /* find the beginning of an attribute */ + buffer = hwloc__nolibxml_import_ignore_spaces(nstate->attrbuffer); + namelen = strspn(buffer, "abcdefghijklmnopqrstuvwxyz_"); + if (buffer[namelen] != '=' || buffer[namelen+1] != '\"') + return -1; + buffer[namelen] = '\0'; + *namep = buffer; + + /* find the beginning of its value, and unescape it */ + *valuep = value = buffer+namelen+2; + len = 0; escaped = 0; + while (value[len+escaped] != '\"') { + if (value[len+escaped] == '&') { + if (!strncmp(&value[1+len+escaped], "#10;", 4)) { + escaped += 4; + value[len] = '\n'; + } else if (!strncmp(&value[1+len+escaped], "#13;", 4)) { + escaped += 4; + value[len] = '\r'; + } else if (!strncmp(&value[1+len+escaped], "#9;", 3)) { + escaped += 3; + value[len] = '\t'; + } else if (!strncmp(&value[1+len+escaped], "quot;", 5)) { + escaped += 5; + value[len] = '\"'; + } else if (!strncmp(&value[1+len+escaped], "lt;", 3)) { + escaped += 3; + value[len] = '<'; + } else if (!strncmp(&value[1+len+escaped], "gt;", 3)) { + escaped += 3; + value[len] = '>'; + } else if (!strncmp(&value[1+len+escaped], "amp;", 4)) { + escaped += 4; + value[len] = '&'; + } else { + return -1; + } + } else { + value[len] = value[len+escaped]; + } + len++; + if (value[len+escaped] == '\0') + return -1; + } + value[len] = '\0'; + + /* find next attribute */ + end = &value[len+escaped+1]; /* skip the ending " */ + nstate->attrbuffer = hwloc__nolibxml_import_ignore_spaces(end); + return 0; +} + +static int +hwloc__nolibxml_import_find_child(hwloc__xml_import_state_t state, + hwloc__xml_import_state_t childstate, + char **tagp) +{ + hwloc__nolibxml_import_state_data_t nstate = (void*) state->data; + hwloc__nolibxml_import_state_data_t nchildstate = (void*) childstate->data; + char *buffer = nstate->tagbuffer; + char *end; + char *tag; + size_t namelen; + + childstate->parent = state; + childstate->global = state->global; + + /* auto-closed tags have no children */ + if (nstate->closed) + return 0; + + /* find the beginning of the tag */ + buffer = hwloc__nolibxml_import_ignore_spaces(buffer); + if (buffer[0] != '<') + return -1; + buffer++; + + /* if closing tag, return nothing and do not advance */ + if (buffer[0] == '/') + return 0; + + /* normal tag */ + tag = nchildstate->tagname = buffer; + + /* find the end, mark it and return it */ + end = strchr(buffer, '>'); + if (!end) + return -1; + end[0] = '\0'; + nchildstate->tagbuffer = end+1; + + /* handle auto-closing tags */ + if (end[-1] == '/') { + nchildstate->closed = 1; + end[-1] = '\0'; + } else + nchildstate->closed = 0; + + /* find attributes */ + namelen = strspn(buffer, "abcdefghijklmnopqrstuvwxyz1234567890_"); + + if (buffer[namelen] == '\0') { + /* no attributes */ + nchildstate->attrbuffer = NULL; + *tagp = tag; + return 1; + } + + if (buffer[namelen] != ' ') + return -1; + + /* found a space, likely starting attributes */ + buffer[namelen] = '\0'; + nchildstate->attrbuffer = buffer+namelen+1; + *tagp = tag; + return 1; +} + +static int +hwloc__nolibxml_import_close_tag(hwloc__xml_import_state_t state) +{ + hwloc__nolibxml_import_state_data_t nstate = (void*) state->data; + char *buffer = nstate->tagbuffer; + char *end; + + /* auto-closed tags need nothing */ + if (nstate->closed) + return 0; + + /* find the beginning of the tag */ + buffer = hwloc__nolibxml_import_ignore_spaces(buffer); + if (buffer[0] != '<') + return -1; + buffer++; + + /* find the end, mark it and return it to the parent */ + end = strchr(buffer, '>'); + if (!end) + return -1; + end[0] = '\0'; + nstate->tagbuffer = end+1; + + /* if closing tag, return nothing */ + if (buffer[0] != '/' || strcmp(buffer+1, nstate->tagname) ) + return -1; + return 0; +} + +static void +hwloc__nolibxml_import_close_child(hwloc__xml_import_state_t state) +{ + hwloc__nolibxml_import_state_data_t nstate = (void*) state->data; + hwloc__nolibxml_import_state_data_t nparent = (void*) state->parent->data; + nparent->tagbuffer = nstate->tagbuffer; +} + +static int +hwloc__nolibxml_import_get_content(hwloc__xml_import_state_t state, + char **beginp, size_t expected_length) +{ + hwloc__nolibxml_import_state_data_t nstate = (void*) state->data; + char *buffer = nstate->tagbuffer; + size_t length; + char *end; + + /* auto-closed tags have no content */ + if (nstate->closed) { + if (expected_length) + return -1; + *beginp = (char *) ""; + return 0; + } + + /* find the next tag, where the content ends */ + end = strchr(buffer, '<'); + if (!end) + return -1; + + length = (size_t) (end-buffer); + if (length != expected_length) + return -1; + nstate->tagbuffer = end; + *end = '\0'; /* mark as 0-terminated for now */ + *beginp = buffer; + return 1; +} + +static void +hwloc__nolibxml_import_close_content(hwloc__xml_import_state_t state) +{ + /* put back the '<' that we overwrote to 0-terminate the content */ + hwloc__nolibxml_import_state_data_t nstate = (void*) state->data; + if (!nstate->closed) + *nstate->tagbuffer = '<'; +} + +static int +hwloc_nolibxml_look_init(struct hwloc_xml_backend_data_s *bdata, + struct hwloc__xml_import_state_s *state) +{ + hwloc__nolibxml_import_state_data_t nstate = (void*) state->data; + struct hwloc__nolibxml_backend_data_s *nbdata = bdata->data; + unsigned major, minor; + char *end; + char *buffer; + + HWLOC_BUILD_ASSERT(sizeof(*nstate) <= sizeof(state->data)); + + /* use a copy in the temporary buffer, we may modify during parsing */ + buffer = nbdata->copy; + memcpy(buffer, nbdata->buffer, nbdata->buflen); + + /* skip headers */ + while (!strncmp(buffer, "", &major, &minor) == 2) { + bdata->version_major = major; + bdata->version_minor = minor; + end = strchr(buffer, '>') + 1; + } else if (!strncmp(buffer, "", 10)) { + bdata->version_major = 1; + bdata->version_minor = 0; + end = buffer + 10; + } else if (!strncmp(buffer, "", 6)) { + bdata->version_major = 0; + bdata->version_minor = 9; + end = buffer + 6; + } else + goto failed; + + state->global->next_attr = hwloc__nolibxml_import_next_attr; + state->global->find_child = hwloc__nolibxml_import_find_child; + state->global->close_tag = hwloc__nolibxml_import_close_tag; + state->global->close_child = hwloc__nolibxml_import_close_child; + state->global->get_content = hwloc__nolibxml_import_get_content; + state->global->close_content = hwloc__nolibxml_import_close_content; + state->parent = NULL; + nstate->closed = 0; + nstate->tagbuffer = end; + nstate->tagname = (char *) "topology"; + nstate->attrbuffer = NULL; + return 0; /* success */ + + failed: + return -1; /* failed */ +} + +/* can be called at the end of the import (to cleanup things early), + * or by backend_exit() if load failed for other reasons. + */ +static void +hwloc_nolibxml_free_buffers(struct hwloc_xml_backend_data_s *bdata) +{ + struct hwloc__nolibxml_backend_data_s *nbdata = bdata->data; + if (nbdata->buffer) { + free(nbdata->buffer); + nbdata->buffer = NULL; + } + if (nbdata->copy) { + free(nbdata->copy); + nbdata->copy = NULL; + } +} + +static void +hwloc_nolibxml_look_done(struct hwloc_xml_backend_data_s *bdata, int result) +{ + hwloc_nolibxml_free_buffers(bdata); + + if (result < 0 && hwloc__xml_verbose()) + fprintf(stderr, "Failed to parse XML input with the minimalistic parser. If it was not\n" + "generated by hwloc, try enabling full XML support with libxml2.\n"); +} + +/******************** + * Backend routines * + ********************/ + +static void +hwloc_nolibxml_backend_exit(struct hwloc_xml_backend_data_s *bdata) +{ + struct hwloc__nolibxml_backend_data_s *nbdata = bdata->data; + hwloc_nolibxml_free_buffers(bdata); + free(nbdata); +} + +static int +hwloc_nolibxml_read_file(const char *xmlpath, char **bufferp, size_t *buflenp) +{ + FILE * file; + size_t buflen, offset, readlen; + struct stat statbuf; + char *buffer, *tmp; + size_t ret; + + if (!strcmp(xmlpath, "-")) + xmlpath = "/dev/stdin"; + + file = fopen(xmlpath, "r"); + if (!file) + goto out; + + /* find the required buffer size for regular files, or use 4k when unknown, we'll realloc later if needed */ + buflen = 4096; + if (!stat(xmlpath, &statbuf)) + if (S_ISREG(statbuf.st_mode)) + buflen = statbuf.st_size+1; /* one additional byte so that the first fread() gets EOF too */ + + buffer = malloc(buflen+1); /* one more byte for the ending \0 */ + if (!buffer) + goto out_with_file; + + offset = 0; readlen = buflen; + while (1) { + ret = fread(buffer+offset, 1, readlen, file); + + offset += ret; + buffer[offset] = 0; + + if (ret != readlen) + break; + + buflen *= 2; + tmp = realloc(buffer, buflen+1); + if (!tmp) + goto out_with_buffer; + buffer = tmp; + readlen = buflen/2; + } + + fclose(file); + *bufferp = buffer; + *buflenp = offset+1; + return 0; + + out_with_buffer: + free(buffer); + out_with_file: + fclose(file); + out: + return -1; +} + +static int +hwloc_nolibxml_backend_init(struct hwloc_xml_backend_data_s *bdata, + const char *xmlpath, const char *xmlbuffer, int xmlbuflen) +{ + struct hwloc__nolibxml_backend_data_s *nbdata = malloc(sizeof(*nbdata)); + + if (!nbdata) + goto out; + bdata->data = nbdata; + + if (xmlbuffer) { + nbdata->buffer = malloc(xmlbuflen+1); + if (!nbdata->buffer) + goto out_with_nbdata; + nbdata->buflen = xmlbuflen+1; + memcpy(nbdata->buffer, xmlbuffer, xmlbuflen); + nbdata->buffer[xmlbuflen] = '\0'; + + } else { + int err = hwloc_nolibxml_read_file(xmlpath, &nbdata->buffer, &nbdata->buflen); + if (err < 0) + goto out_with_nbdata; + } + + /* allocate a temporary copy buffer that we may modify during parsing */ + nbdata->copy = malloc(nbdata->buflen+1); + if (!nbdata->copy) + goto out_with_buffer; + nbdata->copy[nbdata->buflen] = '\0'; + + bdata->look_init = hwloc_nolibxml_look_init; + bdata->look_done = hwloc_nolibxml_look_done; + bdata->backend_exit = hwloc_nolibxml_backend_exit; + return 0; + +out_with_buffer: + free(nbdata->buffer); +out_with_nbdata: + free(nbdata); +out: + return -1; +} + +static int +hwloc_nolibxml_import_diff(struct hwloc__xml_import_state_s *state, + const char *xmlpath, const char *xmlbuffer, int xmlbuflen, + hwloc_topology_diff_t *firstdiffp, char **refnamep) +{ + hwloc__nolibxml_import_state_data_t nstate = (void*) state->data; + struct hwloc__xml_import_state_s childstate; + char *refname = NULL; + char *buffer, *tmp, *tag; + size_t buflen; + int ret; + + HWLOC_BUILD_ASSERT(sizeof(*nstate) <= sizeof(state->data)); + + if (xmlbuffer) { + buffer = malloc(xmlbuflen); + if (!buffer) + goto out; + memcpy(buffer, xmlbuffer, xmlbuflen); + buflen = xmlbuflen; + + } else { + ret = hwloc_nolibxml_read_file(xmlpath, &buffer, &buflen); + if (ret < 0) + goto out; + } + + /* skip headers */ + tmp = buffer; + while (!strncmp(tmp, "global->next_attr = hwloc__nolibxml_import_next_attr; + state->global->find_child = hwloc__nolibxml_import_find_child; + state->global->close_tag = hwloc__nolibxml_import_close_tag; + state->global->close_child = hwloc__nolibxml_import_close_child; + state->global->get_content = hwloc__nolibxml_import_get_content; + state->global->close_content = hwloc__nolibxml_import_close_content; + state->parent = NULL; + nstate->closed = 0; + nstate->tagbuffer = tmp; + nstate->tagname = NULL; + nstate->attrbuffer = NULL; + + /* find root */ + ret = hwloc__nolibxml_import_find_child(state, &childstate, &tag); + if (ret < 0) + goto out_with_buffer; + if (!tag || strcmp(tag, "topologydiff")) + goto out_with_buffer; + + while (1) { + char *attrname, *attrvalue; + if (hwloc__nolibxml_import_next_attr(&childstate, &attrname, &attrvalue) < 0) + break; + if (!strcmp(attrname, "refname")) { + free(refname); + refname = strdup(attrvalue); + } else + goto out_with_buffer; + } + + ret = hwloc__xml_import_diff(&childstate, firstdiffp); + if (refnamep && !ret) + *refnamep = refname; + else + free(refname); + + free(buffer); + return ret; + +out_with_buffer: + free(buffer); + free(refname); +out: + return -1; +} + +/******************* + * Export routines * + *******************/ + +typedef struct hwloc__nolibxml_export_state_data_s { + char *buffer; /* (moving) buffer where to write */ + size_t written; /* how many bytes were written (or would have be written if not truncated) */ + size_t remaining; /* how many bytes are still available in the buffer */ + unsigned indent; /* indentation level for the next line */ + unsigned nr_children; + unsigned has_content; +} __hwloc_attribute_may_alias * hwloc__nolibxml_export_state_data_t; + +static void +hwloc__nolibxml_export_update_buffer(hwloc__nolibxml_export_state_data_t ndata, int res) +{ + if (res >= 0) { + ndata->written += res; + if (res >= (int) ndata->remaining) + res = ndata->remaining>0 ? (int)ndata->remaining-1 : 0; + ndata->buffer += res; + ndata->remaining -= res; + } +} + +static char * +hwloc__nolibxml_export_escape_string(const char *src) +{ + size_t fulllen, sublen; + char *escaped, *dst; + + fulllen = strlen(src); + + sublen = strcspn(src, "\n\r\t\"<>&"); + if (sublen == fulllen) + return NULL; /* nothing to escape */ + + escaped = malloc(fulllen*6+1); /* escaped chars are replaced by at most 6 char */ + dst = escaped; + + memcpy(dst, src, sublen); + src += sublen; + dst += sublen; + + while (*src) { + int replen; + switch (*src) { + case '\n': strcpy(dst, " "); replen=5; break; + case '\r': strcpy(dst, " "); replen=5; break; + case '\t': strcpy(dst, " "); replen=4; break; + case '\"': strcpy(dst, """); replen=6; break; + case '<': strcpy(dst, "<"); replen=4; break; + case '>': strcpy(dst, ">"); replen=4; break; + case '&': strcpy(dst, "&"); replen=5; break; + default: replen=0; break; + } + dst+=replen; src++; + + sublen = strcspn(src, "\n\r\t\"<>&"); + memcpy(dst, src, sublen); + src += sublen; + dst += sublen; + } + + *dst = 0; + return escaped; +} + +static void +hwloc__nolibxml_export_new_child(hwloc__xml_export_state_t parentstate, + hwloc__xml_export_state_t state, + const char *name) +{ + hwloc__nolibxml_export_state_data_t npdata = (void *) parentstate->data; + hwloc__nolibxml_export_state_data_t ndata = (void *) state->data; + int res; + + assert(!npdata->has_content); + if (!npdata->nr_children) { + res = hwloc_snprintf(npdata->buffer, npdata->remaining, ">\n"); + hwloc__nolibxml_export_update_buffer(npdata, res); + } + npdata->nr_children++; + + state->parent = parentstate; + state->new_child = parentstate->new_child; + state->new_prop = parentstate->new_prop; + state->add_content = parentstate->add_content; + state->end_object = parentstate->end_object; + state->global = parentstate->global; + + ndata->buffer = npdata->buffer; + ndata->written = npdata->written; + ndata->remaining = npdata->remaining; + ndata->indent = npdata->indent + 2; + + ndata->nr_children = 0; + ndata->has_content = 0; + + res = hwloc_snprintf(ndata->buffer, ndata->remaining, "%*s<%s", (int) npdata->indent, "", name); + hwloc__nolibxml_export_update_buffer(ndata, res); +} + +static void +hwloc__nolibxml_export_new_prop(hwloc__xml_export_state_t state, const char *name, const char *value) +{ + hwloc__nolibxml_export_state_data_t ndata = (void *) state->data; + char *escaped = hwloc__nolibxml_export_escape_string(value); + int res = hwloc_snprintf(ndata->buffer, ndata->remaining, " %s=\"%s\"", name, escaped ? (const char *) escaped : value); + hwloc__nolibxml_export_update_buffer(ndata, res); + free(escaped); +} + +static void +hwloc__nolibxml_export_end_object(hwloc__xml_export_state_t state, const char *name) +{ + hwloc__nolibxml_export_state_data_t ndata = (void *) state->data; + hwloc__nolibxml_export_state_data_t npdata = (void *) state->parent->data; + int res; + + assert (!(ndata->has_content && ndata->nr_children)); + if (ndata->has_content) { + res = hwloc_snprintf(ndata->buffer, ndata->remaining, "\n", name); + } else if (ndata->nr_children) { + res = hwloc_snprintf(ndata->buffer, ndata->remaining, "%*s\n", (int) npdata->indent, "", name); + } else { + res = hwloc_snprintf(ndata->buffer, ndata->remaining, "/>\n"); + } + hwloc__nolibxml_export_update_buffer(ndata, res); + + npdata->buffer = ndata->buffer; + npdata->written = ndata->written; + npdata->remaining = ndata->remaining; +} + +static void +hwloc__nolibxml_export_add_content(hwloc__xml_export_state_t state, const char *buffer, size_t length) +{ + hwloc__nolibxml_export_state_data_t ndata = (void *) state->data; + int res; + + assert(!ndata->nr_children); + if (!ndata->has_content) { + res = hwloc_snprintf(ndata->buffer, ndata->remaining, ">"); + hwloc__nolibxml_export_update_buffer(ndata, res); + } + ndata->has_content = 1; + + res = hwloc_snprintf(ndata->buffer, ndata->remaining, buffer, length); + hwloc__nolibxml_export_update_buffer(ndata, res); +} + +static size_t +hwloc___nolibxml_prepare_export(hwloc_topology_t topology, struct hwloc__xml_export_data_s *edata, + char *xmlbuffer, int buflen, unsigned long flags) +{ + struct hwloc__xml_export_state_s state, childstate; + hwloc__nolibxml_export_state_data_t ndata = (void *) &state.data; + int v1export = flags & HWLOC_TOPOLOGY_EXPORT_XML_FLAG_V1; + int res; + + HWLOC_BUILD_ASSERT(sizeof(*ndata) <= sizeof(state.data)); + + state.new_child = hwloc__nolibxml_export_new_child; + state.new_prop = hwloc__nolibxml_export_new_prop; + state.add_content = hwloc__nolibxml_export_add_content; + state.end_object = hwloc__nolibxml_export_end_object; + state.global = edata; + + ndata->indent = 0; + ndata->written = 0; + ndata->buffer = xmlbuffer; + ndata->remaining = buflen; + + ndata->nr_children = 1; /* don't close a non-existing previous tag when opening the topology tag */ + ndata->has_content = 0; + + res = hwloc_snprintf(ndata->buffer, ndata->remaining, + "\n" + "\n", v1export ? "hwloc.dtd" : "hwloc2.dtd"); + hwloc__nolibxml_export_update_buffer(ndata, res); + hwloc__nolibxml_export_new_child(&state, &childstate, "topology"); + if (!(flags & HWLOC_TOPOLOGY_EXPORT_XML_FLAG_V1)) + hwloc__nolibxml_export_new_prop(&childstate, "version", "2.0"); + hwloc__xml_export_topology (&childstate, topology, flags); + hwloc__nolibxml_export_end_object(&childstate, "topology"); + + return ndata->written+1; /* ending \0 */ +} + +static int +hwloc_nolibxml_export_buffer(hwloc_topology_t topology, struct hwloc__xml_export_data_s *edata, + char **bufferp, int *buflenp, unsigned long flags) +{ + char *buffer; + size_t bufferlen, res; + + bufferlen = 16384; /* random guess for large enough default */ + buffer = malloc(bufferlen); + if (!buffer) + return -1; + res = hwloc___nolibxml_prepare_export(topology, edata, buffer, (int)bufferlen, flags); + + if (res > bufferlen) { + char *tmp = realloc(buffer, res); + if (!tmp) { + free(buffer); + return -1; + } + buffer = tmp; + hwloc___nolibxml_prepare_export(topology, edata, buffer, (int)res, flags); + } + + *bufferp = buffer; + *buflenp = (int)res; + return 0; +} + +static int +hwloc_nolibxml_export_file(hwloc_topology_t topology, struct hwloc__xml_export_data_s *edata, + const char *filename, unsigned long flags) +{ + FILE *file; + char *buffer; + int bufferlen; + int ret; + + ret = hwloc_nolibxml_export_buffer(topology, edata, &buffer, &bufferlen, flags); + if (ret < 0) + return -1; + + if (!strcmp(filename, "-")) { + file = stdout; + } else { + file = fopen(filename, "w"); + if (!file) { + free(buffer); + return -1; + } + } + + ret = (int)fwrite(buffer, 1, bufferlen-1 /* don't write the ending \0 */, file); + if (ret == bufferlen-1) { + ret = 0; + } else { + errno = ferror(file); + ret = -1; + } + + free(buffer); + + if (file != stdout) + fclose(file); + return ret; +} + +static size_t +hwloc___nolibxml_prepare_export_diff(hwloc_topology_diff_t diff, const char *refname, char *xmlbuffer, int buflen) +{ + struct hwloc__xml_export_state_s state, childstate; + hwloc__nolibxml_export_state_data_t ndata = (void *) &state.data; + int res; + + HWLOC_BUILD_ASSERT(sizeof(*ndata) <= sizeof(state.data)); + + state.new_child = hwloc__nolibxml_export_new_child; + state.new_prop = hwloc__nolibxml_export_new_prop; + state.add_content = hwloc__nolibxml_export_add_content; + state.end_object = hwloc__nolibxml_export_end_object; + + ndata->indent = 0; + ndata->written = 0; + ndata->buffer = xmlbuffer; + ndata->remaining = buflen; + + ndata->nr_children = 1; /* don't close a non-existing previous tag when opening the topology tag */ + ndata->has_content = 0; + + res = hwloc_snprintf(ndata->buffer, ndata->remaining, + "\n" + "\n"); + hwloc__nolibxml_export_update_buffer(ndata, res); + hwloc__nolibxml_export_new_child(&state, &childstate, "topologydiff"); + if (refname) + hwloc__nolibxml_export_new_prop(&childstate, "refname", refname); + hwloc__xml_export_diff (&childstate, diff); + hwloc__nolibxml_export_end_object(&childstate, "topologydiff"); + + return ndata->written+1; +} + +static int +hwloc_nolibxml_export_diff_buffer(hwloc_topology_diff_t diff, const char *refname, char **bufferp, int *buflenp) +{ + char *buffer; + size_t bufferlen, res; + + bufferlen = 16384; /* random guess for large enough default */ + buffer = malloc(bufferlen); + if (!buffer) + return -1; + res = hwloc___nolibxml_prepare_export_diff(diff, refname, buffer, (int)bufferlen); + + if (res > bufferlen) { + char *tmp = realloc(buffer, res); + if (!tmp) { + free(buffer); + return -1; + } + buffer = tmp; + hwloc___nolibxml_prepare_export_diff(diff, refname, buffer, (int)res); + } + + *bufferp = buffer; + *buflenp = (int)res; + return 0; +} + +static int +hwloc_nolibxml_export_diff_file(hwloc_topology_diff_t diff, const char *refname, const char *filename) +{ + FILE *file; + char *buffer; + int bufferlen; + int ret; + + ret = hwloc_nolibxml_export_diff_buffer(diff, refname, &buffer, &bufferlen); + if (ret < 0) + return -1; + + if (!strcmp(filename, "-")) { + file = stdout; + } else { + file = fopen(filename, "w"); + if (!file) { + free(buffer); + return -1; + } + } + + ret = (int)fwrite(buffer, 1, bufferlen-1 /* don't write the ending \0 */, file); + if (ret == bufferlen-1) { + ret = 0; + } else { + errno = ferror(file); + ret = -1; + } + + free(buffer); + + if (file != stdout) + fclose(file); + return ret; +} + +static void +hwloc_nolibxml_free_buffer(void *xmlbuffer) +{ + free(xmlbuffer); +} + +/************* + * Callbacks * + *************/ + +static struct hwloc_xml_callbacks hwloc_xml_nolibxml_callbacks = { + hwloc_nolibxml_backend_init, + hwloc_nolibxml_export_file, + hwloc_nolibxml_export_buffer, + hwloc_nolibxml_free_buffer, + hwloc_nolibxml_import_diff, + hwloc_nolibxml_export_diff_file, + hwloc_nolibxml_export_diff_buffer +}; + +static struct hwloc_xml_component hwloc_nolibxml_xml_component = { + &hwloc_xml_nolibxml_callbacks, + NULL +}; + +const struct hwloc_component hwloc_xml_nolibxml_component = { + HWLOC_COMPONENT_ABI, + NULL, NULL, + HWLOC_COMPONENT_TYPE_XML, + 0, + &hwloc_nolibxml_xml_component +}; diff --git a/src/3rdparty/hwloc/src/topology-xml.c b/src/3rdparty/hwloc/src/topology-xml.c new file mode 100644 index 000000000..e7c5ef621 --- /dev/null +++ b/src/3rdparty/hwloc/src/topology-xml.c @@ -0,0 +1,2886 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2019 Inria. All rights reserved. + * Copyright © 2009-2011 Université Bordeaux + * Copyright © 2009-2018 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +#include +#include +#include +#include +#include +#include + +#include + +int +hwloc__xml_verbose(void) +{ + static int checked = 0; + static int verbose = 0; + if (!checked) { + const char *env = getenv("HWLOC_XML_VERBOSE"); + if (env) + verbose = atoi(env); + checked = 1; + } + return verbose; +} + +static int +hwloc_nolibxml_import(void) +{ + static int checked = 0; + static int nolibxml = 0; + if (!checked) { + const char *env = getenv("HWLOC_LIBXML"); + if (env) { + nolibxml = !atoi(env); + } else { + env = getenv("HWLOC_LIBXML_IMPORT"); + if (env) + nolibxml = !atoi(env); + } + checked = 1; + } + return nolibxml; +} + +static int +hwloc_nolibxml_export(void) +{ + static int checked = 0; + static int nolibxml = 0; + if (!checked) { + const char *env = getenv("HWLOC_LIBXML"); + if (env) { + nolibxml = !atoi(env); + } else { + env = getenv("HWLOC_LIBXML_EXPORT"); + if (env) + nolibxml = !atoi(env); + } + checked = 1; + } + return nolibxml; +} + +#define BASE64_ENCODED_LENGTH(length) (4*(((length)+2)/3)) + +/********************************* + ********* XML callbacks ********* + *********************************/ + +/* set when registering nolibxml and libxml components. + * modifications protected by the components mutex. + * read by the common XML code in topology-xml.c to jump to the right XML backend. + */ +static struct hwloc_xml_callbacks *hwloc_nolibxml_callbacks = NULL, *hwloc_libxml_callbacks = NULL; + +void +hwloc_xml_callbacks_register(struct hwloc_xml_component *comp) +{ + if (!hwloc_nolibxml_callbacks) + hwloc_nolibxml_callbacks = comp->nolibxml_callbacks; + if (!hwloc_libxml_callbacks) + hwloc_libxml_callbacks = comp->libxml_callbacks; +} + +void +hwloc_xml_callbacks_reset(void) +{ + hwloc_nolibxml_callbacks = NULL; + hwloc_libxml_callbacks = NULL; +} + +/************************************************ + ********* XML import (common routines) ********* + ************************************************/ + +#define _HWLOC_OBJ_CACHE_OLD (HWLOC_OBJ_TYPE_MAX+1) /* temporarily used when importing pre-v2.0 attribute-less cache types */ +#define _HWLOC_OBJ_FUTURE (HWLOC_OBJ_TYPE_MAX+2) /* temporarily used when ignoring future types */ + +static void +hwloc__xml_import_object_attr(struct hwloc_topology *topology, + struct hwloc_xml_backend_data_s *data, + struct hwloc_obj *obj, + const char *name, const char *value, + hwloc__xml_import_state_t state) +{ + if (!strcmp(name, "type")) { + /* already handled */ + return; + } + + else if (!strcmp(name, "os_index")) + obj->os_index = strtoul(value, NULL, 10); + else if (!strcmp(name, "gp_index")) { + obj->gp_index = strtoull(value, NULL, 10); + if (!obj->gp_index && hwloc__xml_verbose()) + fprintf(stderr, "%s: unexpected zero gp_index, topology may be invalid\n", state->global->msgprefix); + if (obj->gp_index >= topology->next_gp_index) + topology->next_gp_index = obj->gp_index + 1; + } else if (!strcmp(name, "cpuset")) { + if (!obj->cpuset) + obj->cpuset = hwloc_bitmap_alloc(); + hwloc_bitmap_sscanf(obj->cpuset, value); + } else if (!strcmp(name, "complete_cpuset")) { + if (!obj->complete_cpuset) + obj->complete_cpuset = hwloc_bitmap_alloc(); + hwloc_bitmap_sscanf(obj->complete_cpuset, value); + } else if (!strcmp(name, "allowed_cpuset")) { + /* ignored except for root */ + if (!obj->parent) + hwloc_bitmap_sscanf(topology->allowed_cpuset, value); + } else if (!strcmp(name, "nodeset")) { + if (!obj->nodeset) + obj->nodeset = hwloc_bitmap_alloc(); + hwloc_bitmap_sscanf(obj->nodeset, value); + } else if (!strcmp(name, "complete_nodeset")) { + if (!obj->complete_nodeset) + obj->complete_nodeset = hwloc_bitmap_alloc(); + hwloc_bitmap_sscanf(obj->complete_nodeset, value); + } else if (!strcmp(name, "allowed_nodeset")) { + /* ignored except for root */ + if (!obj->parent) + hwloc_bitmap_sscanf(topology->allowed_nodeset, value); + } else if (!strcmp(name, "name")) { + if (obj->name) + free(obj->name); + obj->name = strdup(value); + } else if (!strcmp(name, "subtype")) { + if (obj->subtype) + free(obj->subtype); + obj->subtype = strdup(value); + } + + else if (!strcmp(name, "cache_size")) { + unsigned long long lvalue = strtoull(value, NULL, 10); + if (hwloc__obj_type_is_cache(obj->type) || obj->type == _HWLOC_OBJ_CACHE_OLD) + obj->attr->cache.size = lvalue; + else if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring cache_size attribute for non-cache object type\n", + state->global->msgprefix); + } + + else if (!strcmp(name, "cache_linesize")) { + unsigned long lvalue = strtoul(value, NULL, 10); + if (hwloc__obj_type_is_cache(obj->type) || obj->type == _HWLOC_OBJ_CACHE_OLD) + obj->attr->cache.linesize = lvalue; + else if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring cache_linesize attribute for non-cache object type\n", + state->global->msgprefix); + } + + else if (!strcmp(name, "cache_associativity")) { + int lvalue = atoi(value); + if (hwloc__obj_type_is_cache(obj->type) || obj->type == _HWLOC_OBJ_CACHE_OLD) + obj->attr->cache.associativity = lvalue; + else if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring cache_associativity attribute for non-cache object type\n", + state->global->msgprefix); + } + + else if (!strcmp(name, "cache_type")) { + unsigned long lvalue = strtoul(value, NULL, 10); + if (hwloc__obj_type_is_cache(obj->type) || obj->type == _HWLOC_OBJ_CACHE_OLD) { + if (lvalue == HWLOC_OBJ_CACHE_UNIFIED + || lvalue == HWLOC_OBJ_CACHE_DATA + || lvalue == HWLOC_OBJ_CACHE_INSTRUCTION) + obj->attr->cache.type = (hwloc_obj_cache_type_t) lvalue; + else + fprintf(stderr, "%s: ignoring invalid cache_type attribute %lu\n", + state->global->msgprefix, lvalue); + } else if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring cache_type attribute for non-cache object type\n", + state->global->msgprefix); + } + + else if (!strcmp(name, "local_memory")) { + unsigned long long lvalue = strtoull(value, NULL, 10); + if (obj->type == HWLOC_OBJ_NUMANODE) + obj->attr->numanode.local_memory = lvalue; + else if (!obj->parent) + topology->machine_memory.local_memory = lvalue; + else if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring local_memory attribute for non-NUMAnode non-root object\n", + state->global->msgprefix); + } + + else if (!strcmp(name, "depth")) { + unsigned long lvalue = strtoul(value, NULL, 10); + if (hwloc__obj_type_is_cache(obj->type) || obj->type == _HWLOC_OBJ_CACHE_OLD) { + obj->attr->cache.depth = lvalue; + } else if (obj->type == HWLOC_OBJ_GROUP || obj->type == HWLOC_OBJ_BRIDGE) { + /* will be overwritten by the core */ + } else if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring depth attribute for object type without depth\n", + state->global->msgprefix); + } + + else if (!strcmp(name, "kind")) { + unsigned long lvalue = strtoul(value, NULL, 10); + if (obj->type == HWLOC_OBJ_GROUP) + obj->attr->group.kind = lvalue; + else if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring kind attribute for non-group object type\n", + state->global->msgprefix); + } + + else if (!strcmp(name, "subkind")) { + unsigned long lvalue = strtoul(value, NULL, 10); + if (obj->type == HWLOC_OBJ_GROUP) + obj->attr->group.subkind = lvalue; + else if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring subkind attribute for non-group object type\n", + state->global->msgprefix); + } + + else if (!strcmp(name, "dont_merge")) { + unsigned long lvalue = strtoul(value, NULL, 10); + if (obj->type == HWLOC_OBJ_GROUP) + obj->attr->group.dont_merge = lvalue; + else if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring dont_merge attribute for non-group object type\n", + state->global->msgprefix); + } + + else if (!strcmp(name, "pci_busid")) { + switch (obj->type) { + case HWLOC_OBJ_PCI_DEVICE: + case HWLOC_OBJ_BRIDGE: { + unsigned domain, bus, dev, func; + if (sscanf(value, "%04x:%02x:%02x.%01x", + &domain, &bus, &dev, &func) != 4) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring invalid pci_busid format string %s\n", + state->global->msgprefix, value); + } else { + obj->attr->pcidev.domain = domain; + obj->attr->pcidev.bus = bus; + obj->attr->pcidev.dev = dev; + obj->attr->pcidev.func = func; + } + break; + } + default: + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring pci_busid attribute for non-PCI object\n", + state->global->msgprefix); + break; + } + } + + else if (!strcmp(name, "pci_type")) { + switch (obj->type) { + case HWLOC_OBJ_PCI_DEVICE: + case HWLOC_OBJ_BRIDGE: { + unsigned classid, vendor, device, subvendor, subdevice, revision; + if (sscanf(value, "%04x [%04x:%04x] [%04x:%04x] %02x", + &classid, &vendor, &device, &subvendor, &subdevice, &revision) != 6) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring invalid pci_type format string %s\n", + state->global->msgprefix, value); + } else { + obj->attr->pcidev.class_id = classid; + obj->attr->pcidev.vendor_id = vendor; + obj->attr->pcidev.device_id = device; + obj->attr->pcidev.subvendor_id = subvendor; + obj->attr->pcidev.subdevice_id = subdevice; + obj->attr->pcidev.revision = revision; + } + break; + } + default: + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring pci_type attribute for non-PCI object\n", + state->global->msgprefix); + break; + } + } + + else if (!strcmp(name, "pci_link_speed")) { + switch (obj->type) { + case HWLOC_OBJ_PCI_DEVICE: + case HWLOC_OBJ_BRIDGE: { + obj->attr->pcidev.linkspeed = (float) atof(value); + break; + } + default: + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring pci_link_speed attribute for non-PCI object\n", + state->global->msgprefix); + break; + } + } + + else if (!strcmp(name, "bridge_type")) { + switch (obj->type) { + case HWLOC_OBJ_BRIDGE: { + unsigned upstream_type, downstream_type; + if (sscanf(value, "%u-%u", &upstream_type, &downstream_type) != 2) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring invalid bridge_type format string %s\n", + state->global->msgprefix, value); + } else { + obj->attr->bridge.upstream_type = (hwloc_obj_bridge_type_t) upstream_type; + obj->attr->bridge.downstream_type = (hwloc_obj_bridge_type_t) downstream_type; + }; + break; + } + default: + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring bridge_type attribute for non-bridge object\n", + state->global->msgprefix); + break; + } + } + + else if (!strcmp(name, "bridge_pci")) { + switch (obj->type) { + case HWLOC_OBJ_BRIDGE: { + unsigned domain, secbus, subbus; + if (sscanf(value, "%04x:[%02x-%02x]", + &domain, &secbus, &subbus) != 3) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring invalid bridge_pci format string %s\n", + state->global->msgprefix, value); + } else { + obj->attr->bridge.downstream.pci.domain = domain; + obj->attr->bridge.downstream.pci.secondary_bus = secbus; + obj->attr->bridge.downstream.pci.subordinate_bus = subbus; + } + break; + } + default: + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring bridge_pci attribute for non-bridge object\n", + state->global->msgprefix); + break; + } + } + + else if (!strcmp(name, "osdev_type")) { + switch (obj->type) { + case HWLOC_OBJ_OS_DEVICE: { + unsigned osdev_type; + if (sscanf(value, "%u", &osdev_type) != 1) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring invalid osdev_type format string %s\n", + state->global->msgprefix, value); + } else + obj->attr->osdev.type = (hwloc_obj_osdev_type_t) osdev_type; + break; + } + default: + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring osdev_type attribute for non-osdev object\n", + state->global->msgprefix); + break; + } + } + + else if (data->version_major < 2) { + /************************ + * deprecated from 1.x + */ + if (!strcmp(name, "os_level") + || !strcmp(name, "online_cpuset")) + { /* ignored */ } + + /************************* + * deprecated from 1.0 + */ + else if (!strcmp(name, "dmi_board_vendor")) { + if (value[0]) + hwloc_obj_add_info(obj, "DMIBoardVendor", value); + } + else if (!strcmp(name, "dmi_board_name")) { + if (value[0]) + hwloc_obj_add_info(obj, "DMIBoardName", value); + } + + else if (data->version_major < 1) { + /************************* + * deprecated from 0.9 + */ + if (!strcmp(name, "memory_kB")) { + unsigned long long lvalue = strtoull(value, NULL, 10); + if (obj->type == _HWLOC_OBJ_CACHE_OLD) + obj->attr->cache.size = lvalue << 10; + else if (obj->type == HWLOC_OBJ_NUMANODE) + obj->attr->numanode.local_memory = lvalue << 10; + else if (!obj->parent) + topology->machine_memory.local_memory = lvalue << 10; + else if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring memory_kB attribute for non-NUMAnode non-root object\n", + state->global->msgprefix); + } + else if (!strcmp(name, "huge_page_size_kB")) { + unsigned long lvalue = strtoul(value, NULL, 10); + if (obj->type == HWLOC_OBJ_NUMANODE || !obj->parent) { + struct hwloc_numanode_attr_s *memory = obj->type == HWLOC_OBJ_NUMANODE ? &obj->attr->numanode : &topology->machine_memory; + if (!memory->page_types) { + memory->page_types = malloc(sizeof(*memory->page_types)); + memory->page_types_len = 1; + } + memory->page_types[0].size = lvalue << 10; + } else if (hwloc__xml_verbose()) { + fprintf(stderr, "%s: ignoring huge_page_size_kB attribute for non-NUMAnode non-root object\n", + state->global->msgprefix); + } + } + else if (!strcmp(name, "huge_page_free")) { + unsigned long lvalue = strtoul(value, NULL, 10); + if (obj->type == HWLOC_OBJ_NUMANODE || !obj->parent) { + struct hwloc_numanode_attr_s *memory = obj->type == HWLOC_OBJ_NUMANODE ? &obj->attr->numanode : &topology->machine_memory; + if (!memory->page_types) { + memory->page_types = malloc(sizeof(*memory->page_types)); + memory->page_types_len = 1; + } + memory->page_types[0].count = lvalue; + } else if (hwloc__xml_verbose()) { + fprintf(stderr, "%s: ignoring huge_page_free attribute for non-NUMAnode non-root object\n", + state->global->msgprefix); + } + } + /* end of deprecated from 0.9 */ + else goto unknown; + } + /* end of deprecated from 1.0 */ + else goto unknown; + } + else { + unknown: + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring unknown object attribute %s\n", + state->global->msgprefix, name); + } +} + + +static int +hwloc__xml_import_info(struct hwloc_xml_backend_data_s *data, + hwloc_obj_t obj, + hwloc__xml_import_state_t state) +{ + char *infoname = NULL; + char *infovalue = NULL; + + while (1) { + char *attrname, *attrvalue; + if (state->global->next_attr(state, &attrname, &attrvalue) < 0) + break; + if (!strcmp(attrname, "name")) + infoname = attrvalue; + else if (!strcmp(attrname, "value")) + infovalue = attrvalue; + else + return -1; + } + + if (infoname) { + /* empty strings are ignored by libxml */ + if (data->version_major < 2 && + (!strcmp(infoname, "Type") || !strcmp(infoname, "CoProcType"))) { + /* 1.x stored subtype in Type or CoProcType */ + if (infovalue) { + if (obj->subtype) + free(obj->subtype); + obj->subtype = strdup(infovalue); + } + } else { + if (infovalue) + hwloc_obj_add_info(obj, infoname, infovalue); + } + } + + return state->global->close_tag(state); +} + +static int +hwloc__xml_import_pagetype(hwloc_topology_t topology __hwloc_attribute_unused, struct hwloc_numanode_attr_s *memory, + hwloc__xml_import_state_t state) +{ + uint64_t size = 0, count = 0; + + while (1) { + char *attrname, *attrvalue; + if (state->global->next_attr(state, &attrname, &attrvalue) < 0) + break; + if (!strcmp(attrname, "size")) + size = strtoull(attrvalue, NULL, 10); + else if (!strcmp(attrname, "count")) + count = strtoull(attrvalue, NULL, 10); + else + return -1; + } + + if (size) { + unsigned idx = memory->page_types_len; + struct hwloc_memory_page_type_s *tmp; + tmp = realloc(memory->page_types, (idx+1)*sizeof(*memory->page_types)); + if (tmp) { /* if failed to allocate, ignore this page_type entry */ + memory->page_types = tmp; + memory->page_types_len = idx+1; + memory->page_types[idx].size = size; + memory->page_types[idx].count = count; + } + } + + return state->global->close_tag(state); +} + +static int +hwloc__xml_v1import_distances(struct hwloc_xml_backend_data_s *data, + hwloc_obj_t obj, + hwloc__xml_import_state_t state) +{ + unsigned long reldepth = 0, nbobjs = 0; + float latbase = 0; + char *tag; + int ret; + + while (1) { + char *attrname, *attrvalue; + if (state->global->next_attr(state, &attrname, &attrvalue) < 0) + break; + if (!strcmp(attrname, "nbobjs")) + nbobjs = strtoul(attrvalue, NULL, 10); + else if (!strcmp(attrname, "relative_depth")) + reldepth = strtoul(attrvalue, NULL, 10); + else if (!strcmp(attrname, "latency_base")) + latbase = (float) atof(attrvalue); + else + return -1; + } + + if (nbobjs && reldepth && latbase) { + unsigned i; + float *matrix; + struct hwloc__xml_imported_v1distances_s *v1dist; + + matrix = malloc(nbobjs*nbobjs*sizeof(float)); + v1dist = malloc(sizeof(*v1dist)); + if (!matrix || !v1dist) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: failed to allocate v1distance matrix for %lu objects\n", + state->global->msgprefix, nbobjs); + free(v1dist); + free(matrix); + return -1; + } + + v1dist->kind = HWLOC_DISTANCES_KIND_FROM_OS|HWLOC_DISTANCES_KIND_MEANS_LATENCY; + /* TODO: we can't know for sure if it comes from the OS. + * On Linux/x86, it would be 10 on the diagonal. + * On Solaris/T5, 15 on the diagonal. + * Just check whether all values are integers, and that all values on the diagonal are minimal and identical? + */ + + v1dist->nbobjs = nbobjs; + v1dist->floats = matrix; + + for(i=0; iglobal->find_child(state, &childstate, &tag); + if (ret <= 0 || strcmp(tag, "latency")) { + /* a latency child is needed */ + free(matrix); + free(v1dist); + return -1; + } + + ret = state->global->next_attr(&childstate, &attrname, &attrvalue); + if (ret < 0 || strcmp(attrname, "value")) { + free(matrix); + free(v1dist); + return -1; + } + + val = (float) atof((char *) attrvalue); + matrix[i] = val * latbase; + + ret = state->global->close_tag(&childstate); + if (ret < 0) { + free(matrix); + free(v1dist); + return -1; + } + + state->global->close_child(&childstate); + } + + if (nbobjs < 2) { + /* distances with a single object are useless, even if the XML isn't invalid */ + assert(nbobjs == 1); + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring invalid distance matrix with only 1 object\n", + state->global->msgprefix); + free(matrix); + free(v1dist); + + } else if (obj->parent) { + /* we currently only import distances attached to root. + * we can't save obj in v1dist because obj could be dropped during insert if ignored. + * we could save its complete_cpu/nodeset instead to find it back later. + * but it doesn't matter much since only NUMA distances attached to root matter. + */ + free(matrix); + free(v1dist); + + } else { + /* queue the distance for real */ + v1dist->prev = data->last_v1dist; + v1dist->next = NULL; + if (data->last_v1dist) + data->last_v1dist->next = v1dist; + else + data->first_v1dist = v1dist; + data->last_v1dist = v1dist; + } + } + + return state->global->close_tag(state); +} + +static int +hwloc__xml_import_userdata(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_obj_t obj, + hwloc__xml_import_state_t state) +{ + size_t length = 0; + int encoded = 0; + char *name = NULL; /* optional */ + int ret; + + while (1) { + char *attrname, *attrvalue; + if (state->global->next_attr(state, &attrname, &attrvalue) < 0) + break; + if (!strcmp(attrname, "length")) + length = strtoul(attrvalue, NULL, 10); + else if (!strcmp(attrname, "encoding")) + encoded = !strcmp(attrvalue, "base64"); + else if (!strcmp(attrname, "name")) + name = attrvalue; + else + return -1; + } + + if (!topology->userdata_import_cb) { + char *buffer; + size_t reallength = encoded ? BASE64_ENCODED_LENGTH(length) : length; + ret = state->global->get_content(state, &buffer, reallength); + if (ret < 0) + return -1; + + } else if (topology->userdata_not_decoded) { + char *buffer, *fakename; + size_t reallength = encoded ? BASE64_ENCODED_LENGTH(length) : length; + ret = state->global->get_content(state, &buffer, reallength); + if (ret < 0) + return -1; + fakename = malloc(6 + 1 + (name ? strlen(name) : 4) + 1); + if (!fakename) + return -1; + sprintf(fakename, encoded ? "base64%c%s" : "normal%c%s", name ? ':' : '-', name ? name : "anon"); + topology->userdata_import_cb(topology, obj, fakename, buffer, length); + free(fakename); + + } else if (encoded && length) { + char *encoded_buffer; + size_t encoded_length = BASE64_ENCODED_LENGTH(length); + ret = state->global->get_content(state, &encoded_buffer, encoded_length); + if (ret < 0) + return -1; + if (ret) { + char *decoded_buffer = malloc(length+1); + if (!decoded_buffer) + return -1; + assert(encoded_buffer[encoded_length] == 0); + ret = hwloc_decode_from_base64(encoded_buffer, decoded_buffer, length+1); + if (ret != (int) length) { + free(decoded_buffer); + return -1; + } + topology->userdata_import_cb(topology, obj, name, decoded_buffer, length); + free(decoded_buffer); + } + + } else { /* always handle length==0 in the non-encoded case */ + char *buffer = (char *) ""; + if (length) { + ret = state->global->get_content(state, &buffer, length); + if (ret < 0) + return -1; + } + topology->userdata_import_cb(topology, obj, name, buffer, length); + } + + state->global->close_content(state); + return state->global->close_tag(state); +} + +static void hwloc__xml_import_report_outoforder(hwloc_topology_t topology, hwloc_obj_t new, hwloc_obj_t old) +{ + char *progname = hwloc_progname(topology); + const char *origversion = hwloc_obj_get_info_by_name(topology->levels[0][0], "hwlocVersion"); + const char *origprogname = hwloc_obj_get_info_by_name(topology->levels[0][0], "ProcessName"); + char *c1, *cc1, t1[64]; + char *c2 = NULL, *cc2 = NULL, t2[64]; + + hwloc_bitmap_asprintf(&c1, new->cpuset); + hwloc_bitmap_asprintf(&cc1, new->complete_cpuset); + hwloc_obj_type_snprintf(t1, sizeof(t1), new, 0); + + if (old->cpuset) + hwloc_bitmap_asprintf(&c2, old->cpuset); + if (old->complete_cpuset) + hwloc_bitmap_asprintf(&cc2, old->complete_cpuset); + hwloc_obj_type_snprintf(t2, sizeof(t2), old, 0); + + fprintf(stderr, "****************************************************************************\n"); + fprintf(stderr, "* hwloc has encountered an out-of-order XML topology load.\n"); + fprintf(stderr, "* Object %s cpuset %s complete %s\n", + t1, c1, cc1); + fprintf(stderr, "* was inserted after object %s with %s and %s.\n", + t2, c2 ? c2 : "none", cc2 ? cc2 : "none"); + fprintf(stderr, "* The error occured in hwloc %s inside process `%s', while\n", + HWLOC_VERSION, + progname ? progname : ""); + if (origversion || origprogname) + fprintf(stderr, "* the input XML was generated by hwloc %s inside process `%s'.\n", + origversion ? origversion : "(unknown version)", + origprogname ? origprogname : ""); + else + fprintf(stderr, "* the input XML was generated by an unspecified ancient hwloc release.\n"); + fprintf(stderr, "* Please check that your input topology XML file is valid.\n"); + fprintf(stderr, "* Set HWLOC_DEBUG_CHECK=1 in the environment to detect further issues.\n"); + fprintf(stderr, "****************************************************************************\n"); + + free(c1); + free(cc1); + free(c2); + free(cc2); + free(progname); +} + +static int +hwloc__xml_import_object(hwloc_topology_t topology, + struct hwloc_xml_backend_data_s *data, + hwloc_obj_t parent, hwloc_obj_t obj, int *gotignored, + hwloc__xml_import_state_t state) +{ + int ignored = 0; + int childrengotignored = 0; + int attribute_less_cache = 0; + int numa_was_root = 0; + char *tag; + struct hwloc__xml_import_state_s childstate; + + /* set parent now since it's used during import below or in subfunctions */ + obj->parent = parent; + + /* process attributes */ + while (1) { + char *attrname, *attrvalue; + if (state->global->next_attr(state, &attrname, &attrvalue) < 0) + break; + if (!strcmp(attrname, "type")) { + if (hwloc_type_sscanf(attrvalue, &obj->type, NULL, 0) < 0) { + if (!strcasecmp(attrvalue, "Cache")) { + obj->type = _HWLOC_OBJ_CACHE_OLD; /* will be fixed below */ + attribute_less_cache = 1; + } else if (!strcasecmp(attrvalue, "System")) { + if (!parent) + obj->type = HWLOC_OBJ_MACHINE; + else { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: obsolete System object only allowed at root\n", + state->global->msgprefix); + goto error_with_object; + } + } else if (!strcasecmp(attrvalue, "Die")) { + /* deal with possible future type */ + obj->type = HWLOC_OBJ_GROUP; + obj->subtype = strdup("Die"); + obj->attr->group.kind = HWLOC_GROUP_KIND_INTEL_DIE; + obj->attr->group.dont_merge = data->dont_merge_die_groups; + } else if (!strcasecmp(attrvalue, "Tile")) { + /* deal with possible future type */ + obj->type = HWLOC_OBJ_GROUP; + obj->subtype = strdup("Tile"); + obj->attr->group.kind = HWLOC_GROUP_KIND_INTEL_TILE; + } else if (!strcasecmp(attrvalue, "Module")) { + /* deal with possible future type */ + obj->type = HWLOC_OBJ_GROUP; + obj->subtype = strdup("Module"); + obj->attr->group.kind = HWLOC_GROUP_KIND_INTEL_MODULE; + } else if (!strcasecmp(attrvalue, "MemCache")) { + /* ignore possible future type */ + obj->type = _HWLOC_OBJ_FUTURE; + ignored = 1; + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: %s object not-supported, will be ignored\n", + state->global->msgprefix, attrvalue); + } else { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: unrecognized object type string %s\n", + state->global->msgprefix, attrvalue); + goto error_with_object; + } + } + } else { + /* type needed first */ + if (obj->type == HWLOC_OBJ_TYPE_NONE) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: object attribute %s found before type\n", + state->global->msgprefix, attrname); + goto error_with_object; + } + hwloc__xml_import_object_attr(topology, data, obj, attrname, attrvalue, state); + } + } + + /* process non-object subnodes to get info attrs (as well as page_types, etc) */ + while (1) { + int ret; + + tag = NULL; + ret = state->global->find_child(state, &childstate, &tag); + if (ret < 0) + goto error; + if (!ret) + break; + + if (!strcmp(tag, "object")) { + /* we'll handle children later */ + break; + + } else if (!strcmp(tag, "page_type")) { + if (obj->type == HWLOC_OBJ_NUMANODE) { + ret = hwloc__xml_import_pagetype(topology, &obj->attr->numanode, &childstate); + } else if (!parent) { + ret = hwloc__xml_import_pagetype(topology, &topology->machine_memory, &childstate); + } else { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: invalid non-NUMAnode object child %s\n", + state->global->msgprefix, tag); + ret = -1; + } + + } else if (!strcmp(tag, "info")) { + ret = hwloc__xml_import_info(data, obj, &childstate); + } else if (data->version_major < 2 && !strcmp(tag, "distances")) { + ret = hwloc__xml_v1import_distances(data, obj, &childstate); + } else if (!strcmp(tag, "userdata")) { + ret = hwloc__xml_import_userdata(topology, obj, &childstate); + } else { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: invalid special object child %s\n", + state->global->msgprefix, tag); + ret = -1; + } + + if (ret < 0) + goto error; + + state->global->close_child(&childstate); + } + + if (parent && obj->type == HWLOC_OBJ_MACHINE) { + /* replace non-root Machine with Groups */ + obj->type = HWLOC_OBJ_GROUP; + } + + if (parent && data->version_major >= 2) { + /* check parent/child types for 2.x */ + if (hwloc__obj_type_is_normal(obj->type)) { + if (!hwloc__obj_type_is_normal(parent->type)) { + if (hwloc__xml_verbose()) + fprintf(stderr, "normal object %s cannot be child of non-normal parent %s\n", + hwloc_obj_type_string(obj->type), hwloc_obj_type_string(parent->type)); + goto error_with_object; + } + } else if (hwloc__obj_type_is_memory(obj->type)) { + if (hwloc__obj_type_is_io(parent->type) || HWLOC_OBJ_MISC == parent->type) { + if (hwloc__xml_verbose()) + fprintf(stderr, "Memory object %s cannot be child of non-normal-or-memory parent %s\n", + hwloc_obj_type_string(obj->type), hwloc_obj_type_string(parent->type)); + goto error_with_object; + } + } else if (hwloc__obj_type_is_io(obj->type)) { + if (hwloc__obj_type_is_memory(parent->type) || HWLOC_OBJ_MISC == parent->type) { + if (hwloc__xml_verbose()) + fprintf(stderr, "I/O object %s cannot be child of non-normal-or-I/O parent %s\n", + hwloc_obj_type_string(obj->type), hwloc_obj_type_string(parent->type)); + goto error_with_object; + } + } + + } else if (parent && data->version_major < 2) { + /* check parent/child types for pre-v2.0 */ + if (hwloc__obj_type_is_normal(obj->type) || HWLOC_OBJ_NUMANODE == obj->type) { + if (hwloc__obj_type_is_special(parent->type)) { + if (hwloc__xml_verbose()) + fprintf(stderr, "v1.x normal v1.x object %s cannot be child of special parent %s\n", + hwloc_obj_type_string(obj->type), hwloc_obj_type_string(parent->type)); + goto error_with_object; + } + } else if (hwloc__obj_type_is_io(obj->type)) { + if (HWLOC_OBJ_MISC == parent->type) { + if (hwloc__xml_verbose()) + fprintf(stderr, "I/O object %s cannot be child of Misc parent\n", + hwloc_obj_type_string(obj->type)); + goto error_with_object; + } + } + } + + if (data->version_major < 2) { + /*************************** + * 1.x specific checks + */ + + /* attach pre-v2.0 children of NUMA nodes to normal parent */ + if (parent && parent->type == HWLOC_OBJ_NUMANODE) { + parent = parent->parent; + assert(parent); + } + + /* insert a group above pre-v2.0 NUMA nodes if needed */ + if (obj->type == HWLOC_OBJ_NUMANODE) { + if (!parent) { + /* crazy case of NUMA node root (only possible when filtering Machine keep_structure in v1.x), + * reinsert a Machine object + */ + hwloc_obj_t machine = hwloc_alloc_setup_object(topology, HWLOC_OBJ_MACHINE, HWLOC_UNKNOWN_INDEX); + machine->cpuset = hwloc_bitmap_dup(obj->cpuset); + machine->complete_cpuset = hwloc_bitmap_dup(obj->cpuset); + machine->nodeset = hwloc_bitmap_dup(obj->nodeset); + machine->complete_nodeset = hwloc_bitmap_dup(obj->complete_nodeset); + topology->levels[0][0] = machine; + parent = machine; + numa_was_root = 1; + + } else if (!hwloc_bitmap_isequal(obj->complete_cpuset, parent->complete_cpuset)) { + /* This NUMA node has a different locality from its parent. + * Don't attach it to this parent, or it well get its parent cpusets. + * Add an intermediate Group with the desired locality. + */ + int needgroup = 1; + hwloc_obj_t sibling; + + sibling = parent->memory_first_child; + if (sibling && !sibling->subtype + && !sibling->next_sibling + && obj->subtype && !strcmp(obj->subtype, "MCDRAM") + && hwloc_bitmap_iszero(obj->complete_cpuset)) { + /* this is KNL MCDRAM, we want to attach it near its DDR sibling */ + needgroup = 0; + } + /* Ideally we would also detect similar cases on future non-KNL platforms with multiple local NUMA nodes. + * That's unlikely to occur with v1.x. + * And we have no way to be sure if this CPU-less node is desired or not. + */ + + if (needgroup + && hwloc_filter_check_keep_object_type(topology, HWLOC_OBJ_GROUP)) { + hwloc_obj_t group = hwloc_alloc_setup_object(topology, HWLOC_OBJ_GROUP, HWLOC_UNKNOWN_INDEX); + group->gp_index = 0; /* will be initialized at the end of the discovery once we know the max */ + group->cpuset = hwloc_bitmap_dup(obj->cpuset); + group->complete_cpuset = hwloc_bitmap_dup(obj->cpuset); + group->nodeset = hwloc_bitmap_dup(obj->nodeset); + group->complete_nodeset = hwloc_bitmap_dup(obj->complete_nodeset); + group->attr->group.kind = HWLOC_GROUP_KIND_MEMORY; + hwloc_insert_object_by_parent(topology, parent, group); + parent = group; + } + } + } + + /* fixup attribute-less caches imported from pre-v2.0 XMLs */ + if (attribute_less_cache) { + assert(obj->type == _HWLOC_OBJ_CACHE_OLD); + obj->type = hwloc_cache_type_by_depth_type(obj->attr->cache.depth, obj->attr->cache.type); + } + + /* fixup Misc objects inserted by cpusets in pre-v2.0 XMLs */ + if (obj->type == HWLOC_OBJ_MISC && obj->cpuset) + obj->type = HWLOC_OBJ_GROUP; + + /* check set consistency. + * 1.7.2 and earlier reported I/O Groups with only a cpuset, we don't want to reject those XMLs yet. + * Ignore those Groups since fixing the missing sets is hard (would need to look at children sets which are not available yet). + * Just abort the XML for non-Groups. + */ + if (!obj->cpuset != !obj->complete_cpuset) { + /* has some cpuset without others */ + if (obj->type == HWLOC_OBJ_GROUP) { + ignored = 1; + } else { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: invalid object %s P#%u with some missing cpusets\n", + state->global->msgprefix, hwloc_obj_type_string(obj->type), obj->os_index); + goto error_with_object; + } + } else if (!obj->nodeset != !obj->complete_nodeset) { + /* has some nodeset without others */ + if (obj->type == HWLOC_OBJ_GROUP) { + ignored = 1; + } else { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: invalid object %s P#%u with some missing nodesets\n", + state->global->msgprefix, hwloc_obj_type_string(obj->type), obj->os_index); + goto error_with_object; + } + } else if (obj->nodeset && !obj->cpuset) { + /* has nodesets without cpusets (the contrary is allowed in pre-2.0) */ + if (obj->type == HWLOC_OBJ_GROUP) { + ignored = 1; + } else { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: invalid object %s P#%u with either cpuset or nodeset missing\n", + state->global->msgprefix, hwloc_obj_type_string(obj->type), obj->os_index); + goto error_with_object; + } + } + /* end of 1.x specific checks */ + } + + /* check that cache attributes are coherent with the actual type */ + if (hwloc__obj_type_is_cache(obj->type) + && obj->type != hwloc_cache_type_by_depth_type(obj->attr->cache.depth, obj->attr->cache.type)) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: invalid cache type %s with attribute depth %u and type %d\n", + state->global->msgprefix, hwloc_obj_type_string(obj->type), obj->attr->cache.depth, (int) obj->attr->cache.type); + goto error_with_object; + } + + /* check special types vs cpuset */ + if (!obj->cpuset && !hwloc__obj_type_is_special(obj->type)) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: invalid normal object %s P#%u without cpuset\n", + state->global->msgprefix, hwloc_obj_type_string(obj->type), obj->os_index); + goto error_with_object; + } + if (obj->cpuset && hwloc__obj_type_is_special(obj->type)) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: invalid special object %s with cpuset\n", + state->global->msgprefix, hwloc_obj_type_string(obj->type)); + goto error_with_object; + } + + /* check parent vs child sets */ + if (obj->cpuset && parent && !parent->cpuset) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: invalid object %s P#%u with cpuset while parent has none\n", + state->global->msgprefix, hwloc_obj_type_string(obj->type), obj->os_index); + goto error_with_object; + } + if (obj->nodeset && parent && !parent->nodeset) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: invalid object %s P#%u with nodeset while parent has none\n", + state->global->msgprefix, hwloc_obj_type_string(obj->type), obj->os_index); + goto error_with_object; + } + + /* check NUMA nodes */ + if (obj->type == HWLOC_OBJ_NUMANODE) { + if (!obj->nodeset) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: invalid NUMA node object P#%u without nodeset\n", + state->global->msgprefix, obj->os_index); + goto error_with_object; + } + data->nbnumanodes++; + obj->prev_cousin = data->last_numanode; + obj->next_cousin = NULL; + if (data->last_numanode) + data->last_numanode->next_cousin = obj; + else + data->first_numanode = obj; + data->last_numanode = obj; + } + + if (!hwloc_filter_check_keep_object(topology, obj)) { + /* Ignore this object instead of inserting it. + * + * Well, let the core ignore the root object later + * because we don't know yet if root has more than one child. + */ + if (parent) + ignored = 1; + } + + if (parent && !ignored) { + /* root->parent is NULL, and root is already inserted */ + hwloc_insert_object_by_parent(topology, parent, obj); + /* insert_object_by_parent() doesn't merge during insert, so obj is still valid */ + } + + /* process object subnodes, if we found one win the above loop */ + while (tag) { + int ret; + + if (!strcmp(tag, "object")) { + hwloc_obj_t childobj = hwloc_alloc_setup_object(topology, HWLOC_OBJ_TYPE_MAX, HWLOC_UNKNOWN_INDEX); + childobj->parent = ignored ? parent : obj; + ret = hwloc__xml_import_object(topology, data, ignored ? parent : obj, childobj, + &childrengotignored, + &childstate); + } else { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: invalid special object child %s while looking for objects\n", + state->global->msgprefix, tag); + ret = -1; + } + + if (ret < 0) + goto error; + + state->global->close_child(&childstate); + + tag = NULL; + ret = state->global->find_child(state, &childstate, &tag); + if (ret < 0) + goto error; + if (!ret) + break; + } + + if (numa_was_root) { + /* duplicate NUMA infos to root, most of them are likely root-specific */ + unsigned i; + for(i=0; iinfos_count; i++) { + struct hwloc_info_s *info = &obj->infos[i]; + hwloc_obj_add_info(parent, info->name, info->value); + } + /* TODO some infos are root-only (hwlocVersion, ProcessName, etc), remove them from obj? */ + } + + if (ignored) { + /* drop that object, and tell the parent that one child got ignored */ + hwloc_free_unlinked_object(obj); + *gotignored = 1; + + } else if (obj->first_child) { + /* now that all children are inserted, make sure they are in-order, + * so that the core doesn't have to deal with crappy children list. + */ + hwloc_obj_t cur, next; + for(cur = obj->first_child, next = cur->next_sibling; + next; + cur = next, next = next->next_sibling) { + /* If reordering is needed, at least one pair of consecutive children will be out-of-order. + * So just check pairs of consecutive children. + * + * We checked above that complete_cpuset is always set. + */ + if (hwloc_bitmap_compare_first(next->complete_cpuset, cur->complete_cpuset) < 0) { + /* next should be before cur */ + if (!childrengotignored) { + static int reported = 0; + if (!reported && !hwloc_hide_errors()) { + hwloc__xml_import_report_outoforder(topology, next, cur); + reported = 1; + } + } + hwloc__reorder_children(obj); + break; + } + } + /* no need to reorder memory children as long as there are no intermediate memory objects + * that could cause reordering when filtered-out. + */ + } + + return state->global->close_tag(state); + + error_with_object: + if (parent) + /* root->parent is NULL, and root is already inserted. the caller will cleanup that root. */ + hwloc_free_unlinked_object(obj); + error: + return -1; +} + +static int +hwloc__xml_v2import_distances(hwloc_topology_t topology, + hwloc__xml_import_state_t state) +{ + hwloc_obj_type_t type = HWLOC_OBJ_TYPE_NONE; + unsigned nbobjs = 0; + int indexing = 0; + int os_indexing = 0; + int gp_indexing = 0; + unsigned long kind = 0; + unsigned nr_indexes, nr_u64values; + uint64_t *indexes; + uint64_t *u64values; + int ret; + + /* process attributes */ + while (1) { + char *attrname, *attrvalue; + if (state->global->next_attr(state, &attrname, &attrvalue) < 0) + break; + if (!strcmp(attrname, "nbobjs")) + nbobjs = strtoul(attrvalue, NULL, 10); + else if (!strcmp(attrname, "type")) { + if (hwloc_type_sscanf(attrvalue, &type, NULL, 0) < 0) + goto out; + } + else if (!strcmp(attrname, "indexing")) { + indexing = 1; + if (!strcmp(attrvalue, "os")) + os_indexing = 1; + else if (!strcmp(attrvalue, "gp")) + gp_indexing = 1; + } + else if (!strcmp(attrname, "kind")) { + kind = strtoul(attrvalue, NULL, 10); + } + else { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring unknown distance attribute %s\n", + state->global->msgprefix, attrname); + } + } + + /* abort if missing attribute */ + if (!nbobjs || type == HWLOC_OBJ_TYPE_NONE || !indexing || !kind) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: distance2 missing some attributes\n", + state->global->msgprefix); + goto out; + } + + indexes = malloc(nbobjs*sizeof(*indexes)); + u64values = malloc(nbobjs*nbobjs*sizeof(*u64values)); + if (!indexes || !u64values) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: failed to allocate distances arrays for %u objects\n", + state->global->msgprefix, nbobjs); + goto out_with_arrays; + } + + /* process children */ + nr_indexes = 0; + nr_u64values = 0; + while (1) { + struct hwloc__xml_import_state_s childstate; + char *attrname, *attrvalue, *tag, *buffer; + int length; + int is_index = 0; + int is_u64values = 0; + + ret = state->global->find_child(state, &childstate, &tag); + if (ret <= 0) + break; + + if (!strcmp(tag, "indexes")) + is_index = 1; + else if (!strcmp(tag, "u64values")) + is_u64values = 1; + if (!is_index && !is_u64values) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: distance2 with unrecognized child %s\n", + state->global->msgprefix, tag); + goto out_with_arrays; + } + + if (state->global->next_attr(&childstate, &attrname, &attrvalue) < 0 + || strcmp(attrname, "length")) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: distance2 child must have length attribute\n", + state->global->msgprefix); + goto out_with_arrays; + } + length = atoi(attrvalue); + + ret = state->global->get_content(&childstate, &buffer, length); + if (ret < 0) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: distance2 child needs content of length %d\n", + state->global->msgprefix, length); + goto out_with_arrays; + } + + if (is_index) { + /* get indexes */ + char *tmp; + if (nr_indexes >= nbobjs) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: distance2 with more than %u indexes\n", + state->global->msgprefix, nbobjs); + goto out_with_arrays; + } + tmp = buffer; + while (1) { + char *next; + unsigned long long u = strtoull(tmp, &next, 0); + if (next == tmp) + break; + indexes[nr_indexes++] = u; + if (*next != ' ') + break; + if (nr_indexes == nbobjs) + break; + tmp = next+1; + } + + } else if (is_u64values) { + /* get uint64_t values */ + char *tmp; + if (nr_u64values >= nbobjs*nbobjs) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: distance2 with more than %u u64values\n", + state->global->msgprefix, nbobjs*nbobjs); + goto out_with_arrays; + } + tmp = buffer; + while (1) { + char *next; + unsigned long long u = strtoull(tmp, &next, 0); + if (next == tmp) + break; + u64values[nr_u64values++] = u; + if (*next != ' ') + break; + if (nr_u64values == nbobjs*nbobjs) + break; + tmp = next+1; + } + } + + state->global->close_content(&childstate); + + ret = state->global->close_tag(&childstate); + if (ret < 0) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: distance2 with more than %u indexes\n", + state->global->msgprefix, nbobjs); + goto out_with_arrays; + } + + state->global->close_child(&childstate); + } + + if (nr_indexes != nbobjs) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: distance2 with less than %u indexes\n", + state->global->msgprefix, nbobjs); + goto out_with_arrays; + } + if (nr_u64values != nbobjs*nbobjs) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: distance2 with less than %u u64values\n", + state->global->msgprefix, nbobjs*nbobjs); + goto out_with_arrays; + } + + if (nbobjs < 2) { + /* distances with a single object are useless, even if the XML isn't invalid */ + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring distances2 with only %u objects\n", + state->global->msgprefix, nbobjs); + goto out_ignore; + } + if (type == HWLOC_OBJ_PU || type == HWLOC_OBJ_NUMANODE) { + if (!os_indexing) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring PU or NUMA distances2 without os_indexing\n", + state->global->msgprefix); + goto out_ignore; + } + } else { + if (!gp_indexing) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring !PU or !NUMA distances2 without gp_indexing\n", + state->global->msgprefix); + goto out_ignore; + } + } + + hwloc_internal_distances_add_by_index(topology, type, nbobjs, indexes, u64values, kind, 0); + + /* prevent freeing below */ + indexes = NULL; + u64values = NULL; + + out_ignore: + free(indexes); + free(u64values); + return state->global->close_tag(state); + + out_with_arrays: + free(indexes); + free(u64values); + out: + return -1; +} + +static int +hwloc__xml_import_diff_one(hwloc__xml_import_state_t state, + hwloc_topology_diff_t *firstdiffp, + hwloc_topology_diff_t *lastdiffp) +{ + char *type_s = NULL; + char *obj_depth_s = NULL; + char *obj_index_s = NULL; + char *obj_attr_type_s = NULL; +/* char *obj_attr_index_s = NULL; unused for now */ + char *obj_attr_name_s = NULL; + char *obj_attr_oldvalue_s = NULL; + char *obj_attr_newvalue_s = NULL; + + while (1) { + char *attrname, *attrvalue; + if (state->global->next_attr(state, &attrname, &attrvalue) < 0) + break; + if (!strcmp(attrname, "type")) + type_s = attrvalue; + else if (!strcmp(attrname, "obj_depth")) + obj_depth_s = attrvalue; + else if (!strcmp(attrname, "obj_index")) + obj_index_s = attrvalue; + else if (!strcmp(attrname, "obj_attr_type")) + obj_attr_type_s = attrvalue; + else if (!strcmp(attrname, "obj_attr_index")) + { /* obj_attr_index_s = attrvalue; unused for now */ } + else if (!strcmp(attrname, "obj_attr_name")) + obj_attr_name_s = attrvalue; + else if (!strcmp(attrname, "obj_attr_oldvalue")) + obj_attr_oldvalue_s = attrvalue; + else if (!strcmp(attrname, "obj_attr_newvalue")) + obj_attr_newvalue_s = attrvalue; + else { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring unknown diff attribute %s\n", + state->global->msgprefix, attrname); + return -1; + } + } + + if (type_s) { + switch (atoi(type_s)) { + default: + break; + case HWLOC_TOPOLOGY_DIFF_OBJ_ATTR: { + /* object attribute diff */ + hwloc_topology_diff_obj_attr_type_t obj_attr_type; + hwloc_topology_diff_t diff; + + /* obj_attr mandatory generic attributes */ + if (!obj_depth_s || !obj_index_s || !obj_attr_type_s) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: missing mandatory obj attr generic attributes\n", + state->global->msgprefix); + break; + } + + /* obj_attr mandatory attributes common to all subtypes */ + if (!obj_attr_oldvalue_s || !obj_attr_newvalue_s) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: missing mandatory obj attr value attributes\n", + state->global->msgprefix); + break; + } + + /* mandatory attributes for obj_attr_info subtype */ + obj_attr_type = atoi(obj_attr_type_s); + if (obj_attr_type == HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_INFO && !obj_attr_name_s) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: missing mandatory obj attr info name attribute\n", + state->global->msgprefix); + break; + } + + /* now we know we have everything we need */ + diff = malloc(sizeof(*diff)); + if (!diff) + return -1; + diff->obj_attr.type = HWLOC_TOPOLOGY_DIFF_OBJ_ATTR; + diff->obj_attr.obj_depth = atoi(obj_depth_s); + diff->obj_attr.obj_index = atoi(obj_index_s); + memset(&diff->obj_attr.diff, 0, sizeof(diff->obj_attr.diff)); + diff->obj_attr.diff.generic.type = obj_attr_type; + + switch (atoi(obj_attr_type_s)) { + case HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_SIZE: + diff->obj_attr.diff.uint64.oldvalue = strtoull(obj_attr_oldvalue_s, NULL, 0); + diff->obj_attr.diff.uint64.newvalue = strtoull(obj_attr_newvalue_s, NULL, 0); + break; + case HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_INFO: + diff->obj_attr.diff.string.name = strdup(obj_attr_name_s); + /* FALLTHRU */ + case HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_NAME: + diff->obj_attr.diff.string.oldvalue = strdup(obj_attr_oldvalue_s); + diff->obj_attr.diff.string.newvalue = strdup(obj_attr_newvalue_s); + break; + } + + if (*firstdiffp) + (*lastdiffp)->generic.next = diff; + else + *firstdiffp = diff; + *lastdiffp = diff; + diff->generic.next = NULL; + } + } + } + + return state->global->close_tag(state); +} + +int +hwloc__xml_import_diff(hwloc__xml_import_state_t state, + hwloc_topology_diff_t *firstdiffp) +{ + hwloc_topology_diff_t firstdiff = NULL, lastdiff = NULL; + *firstdiffp = NULL; + + while (1) { + struct hwloc__xml_import_state_s childstate; + char *tag; + int ret; + + ret = state->global->find_child(state, &childstate, &tag); + if (ret < 0) + return -1; + if (!ret) + break; + + if (!strcmp(tag, "diff")) { + ret = hwloc__xml_import_diff_one(&childstate, &firstdiff, &lastdiff); + } else + ret = -1; + + if (ret < 0) + return ret; + + state->global->close_child(&childstate); + } + + *firstdiffp = firstdiff; + return 0; +} + +/*********************************** + ********* main XML import ********* + ***********************************/ + +static void +hwloc_convert_from_v1dist_floats(hwloc_topology_t topology, unsigned nbobjs, float *floats, uint64_t *u64s) +{ + unsigned i; + int is_uint; + char *env; + float scale = 1000.f; + char scalestring[20]; + + env = getenv("HWLOC_XML_V1DIST_SCALE"); + if (env) { + scale = (float) atof(env); + goto scale; + } + + is_uint = 1; + /* find out if all values are integers */ + for(i=0; i .001f && fptr < .999f) { + is_uint = 0; + break; + } + u64s[i] = (int)(f+.5f); + } + if (is_uint) + return; + + scale: + /* TODO heuristic to find a good scale */ + for(i=0; itopology; + struct hwloc_xml_backend_data_s *data = backend->private_data; + struct hwloc__xml_import_state_s state, childstate; + struct hwloc_obj *root = topology->levels[0][0]; + char *tag; + int gotignored = 0; + hwloc_localeswitch_declare; + char *env; + int ret; + + state.global = data; + + assert(!root->cpuset); + + hwloc_localeswitch_init(); + + data->nbnumanodes = 0; + data->first_numanode = data->last_numanode = NULL; + data->first_v1dist = data->last_v1dist = NULL; + + env = getenv("HWLOC_DONT_MERGE_DIE_GROUPS"); + data->dont_merge_die_groups = env && atoi(env); + + ret = data->look_init(data, &state); + if (ret < 0) + goto failed; + + if (data->version_major > 2) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: cannot import XML version %u.%u > 2\n", + data->msgprefix, data->version_major, data->version_minor); + goto err; + } + + /* find root object tag and import it */ + ret = state.global->find_child(&state, &childstate, &tag); + if (ret < 0 || !ret || strcmp(tag, "object")) + goto failed; + ret = hwloc__xml_import_object(topology, data, NULL /* no parent */, root, + &gotignored, + &childstate); + if (ret < 0) + goto failed; + state.global->close_child(&childstate); + assert(!gotignored); + + /* the root may have changed if we had to reinsert a Machine */ + root = topology->levels[0][0]; + + if (data->version_major >= 2) { + /* find v2 distances */ + while (1) { + ret = state.global->find_child(&state, &childstate, &tag); + if (ret < 0) + goto failed; + if (!ret) + break; + if (strcmp(tag, "distances2")) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: ignoring unknown tag `%s' after root object, expected `distances2'\n", + data->msgprefix, tag); + goto done; + } + ret = hwloc__xml_v2import_distances(topology, &childstate); + if (ret < 0) + goto failed; + state.global->close_child(&childstate); + } + } + + /* find end of topology tag */ + state.global->close_tag(&state); + +done: + if (!root->cpuset) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: invalid root object without cpuset\n", + data->msgprefix); + goto err; + } + + /* update pre-v2.0 memory group gp_index */ + if (data->version_major < 2 && data->first_numanode) { + hwloc_obj_t node = data->first_numanode; + do { + if (node->parent->type == HWLOC_OBJ_GROUP + && !node->parent->gp_index) + node->parent->gp_index = topology->next_gp_index++; + node = node->next_cousin; + } while (node); + } + + if (data->version_major < 2 && data->first_v1dist) { + /* handle v1 distances */ + struct hwloc__xml_imported_v1distances_s *v1dist, *v1next = data->first_v1dist; + while ((v1dist = v1next) != NULL) { + unsigned nbobjs = v1dist->nbobjs; + v1next = v1dist->next; + /* Handle distances as NUMA node distances if nbobjs matches. + * Otherwise drop, only NUMA distances really matter. + * + * We could also attach to a random level with the right nbobjs, + * but it would require to have those objects in the original XML order (like the first_numanode cousin-list). + * because the topology order can be different if some parents are ignored during load. + */ + if (nbobjs == data->nbnumanodes) { + hwloc_obj_t *objs = malloc(nbobjs*sizeof(hwloc_obj_t)); + uint64_t *values = malloc(nbobjs*nbobjs*sizeof(*values)); + if (objs && values) { + hwloc_obj_t node; + unsigned i; + for(i=0, node = data->first_numanode; + inext_cousin) + objs[i] = node; +hwloc_convert_from_v1dist_floats(topology, nbobjs, v1dist->floats, values); + hwloc_internal_distances_add(topology, nbobjs, objs, values, v1dist->kind, 0); + } else { + free(objs); + free(values); + } + } + free(v1dist->floats); + free(v1dist); + } + data->first_v1dist = data->last_v1dist = NULL; + } + + /* FIXME: + * We should check that the existing object sets are consistent: + * no intersection between objects of a same level, + * object sets included in parent sets. + * hwloc never generated such buggy XML, but users could create one. + * + * We want to add these checks to the existing core code that + * adds missing sets and propagates parent/children sets + * (in case another backend ever generates buggy object sets as well). + */ + + if (data->version_major >= 2) { + /* v2 must have non-empty nodesets since at least one NUMA node is required */ + if (!root->nodeset) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: invalid root object without nodeset\n", + data->msgprefix); + goto err; + } + if (hwloc_bitmap_iszero(root->nodeset)) { + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: invalid root object with empty nodeset\n", + data->msgprefix); + goto err; + } + } else { + /* if v1 without nodeset, the core will add a default NUMA node and nodesets */ + } + + /* allocate default cpusets and nodesets if missing, the core will restrict them */ + hwloc_alloc_root_sets(root); + + /* keep the "Backend" information intact */ + /* we could add "BackendSource=XML" to notify that XML was used between the actual backend and here */ + + topology->support.discovery->pu = 1; + if (data->nbnumanodes) { + topology->support.discovery->numa = 1; + topology->support.discovery->numa_memory = 1; // FIXME + } + + if (data->look_done) + data->look_done(data, 0); + + hwloc_localeswitch_fini(); + return 0; + + failed: + if (data->look_done) + data->look_done(data, -1); + if (hwloc__xml_verbose()) + fprintf(stderr, "%s: XML component discovery failed.\n", + data->msgprefix); + err: + hwloc_free_object_siblings_and_children(root->first_child); + root->first_child = NULL; + hwloc_free_object_siblings_and_children(root->memory_first_child); + root->memory_first_child = NULL; + hwloc_free_object_siblings_and_children(root->io_first_child); + root->io_first_child = NULL; + hwloc_free_object_siblings_and_children(root->misc_first_child); + root->misc_first_child = NULL; + + /* make sure the core will abort */ + if (root->cpuset) + hwloc_bitmap_zero(root->cpuset); + if (root->nodeset) + hwloc_bitmap_zero(root->nodeset); + + hwloc_localeswitch_fini(); + return -1; +} + +/* this can be the first XML call */ +int +hwloc_topology_diff_load_xml(const char *xmlpath, + hwloc_topology_diff_t *firstdiffp, char **refnamep) +{ + struct hwloc__xml_import_state_s state; + struct hwloc_xml_backend_data_s fakedata; /* only for storing global info during parsing */ + hwloc_localeswitch_declare; + const char *local_basename; + int force_nolibxml; + int ret; + + state.global = &fakedata; + + local_basename = strrchr(xmlpath, '/'); + if (local_basename) + local_basename++; + else + local_basename = xmlpath; + fakedata.msgprefix = strdup(local_basename); + + hwloc_components_init(); + assert(hwloc_nolibxml_callbacks); + + hwloc_localeswitch_init(); + + *firstdiffp = NULL; + + force_nolibxml = hwloc_nolibxml_import(); +retry: + if (!hwloc_libxml_callbacks || (hwloc_nolibxml_callbacks && force_nolibxml)) + ret = hwloc_nolibxml_callbacks->import_diff(&state, xmlpath, NULL, 0, firstdiffp, refnamep); + else { + ret = hwloc_libxml_callbacks->import_diff(&state, xmlpath, NULL, 0, firstdiffp, refnamep); + if (ret < 0 && errno == ENOSYS) { + hwloc_libxml_callbacks = NULL; + goto retry; + } + } + + hwloc_localeswitch_fini(); + hwloc_components_fini(); + free(fakedata.msgprefix); + return ret; +} + +/* this can be the first XML call */ +int +hwloc_topology_diff_load_xmlbuffer(const char *xmlbuffer, int buflen, + hwloc_topology_diff_t *firstdiffp, char **refnamep) +{ + struct hwloc__xml_import_state_s state; + struct hwloc_xml_backend_data_s fakedata; /* only for storing global info during parsing */ + hwloc_localeswitch_declare; + int force_nolibxml; + int ret; + + state.global = &fakedata; + fakedata.msgprefix = strdup("xmldiffbuffer"); + + hwloc_components_init(); + assert(hwloc_nolibxml_callbacks); + + hwloc_localeswitch_init(); + + *firstdiffp = NULL; + + force_nolibxml = hwloc_nolibxml_import(); + retry: + if (!hwloc_libxml_callbacks || (hwloc_nolibxml_callbacks && force_nolibxml)) + ret = hwloc_nolibxml_callbacks->import_diff(&state, NULL, xmlbuffer, buflen, firstdiffp, refnamep); + else { + ret = hwloc_libxml_callbacks->import_diff(&state, NULL, xmlbuffer, buflen, firstdiffp, refnamep); + if (ret < 0 && errno == ENOSYS) { + hwloc_libxml_callbacks = NULL; + goto retry; + } + } + + hwloc_localeswitch_fini(); + hwloc_components_fini(); + free(fakedata.msgprefix); + return ret; +} + +/************************************************ + ********* XML export (common routines) ********* + ************************************************/ + +#define HWLOC_XML_CHAR_VALID(c) (((c) >= 32 && (c) <= 126) || (c) == '\t' || (c) == '\n' || (c) == '\r') + +static int +hwloc__xml_export_check_buffer(const char *buf, size_t length) +{ + unsigned i; + for(i=0; itype == HWLOC_OBJ_PACKAGE) + state->new_prop(state, "type", "Socket"); + else if (v1export && hwloc__obj_type_is_cache(obj->type)) + state->new_prop(state, "type", "Cache"); + else + state->new_prop(state, "type", hwloc_obj_type_string(obj->type)); + + if (obj->os_index != HWLOC_UNKNOWN_INDEX) { + sprintf(tmp, "%u", obj->os_index); + state->new_prop(state, "os_index", tmp); + } + + if (obj->cpuset) { + if (v1export && obj->type == HWLOC_OBJ_NUMANODE && obj->sibling_rank > 0) { + /* v1 non-first NUMA nodes have empty cpusets */ + state->new_prop(state, "cpuset", "0x0"); + state->new_prop(state, "online_cpuset", "0x0"); + state->new_prop(state, "complete_cpuset", "0x0"); + state->new_prop(state, "allowed_cpuset", "0x0"); + + } else { + /* normal case */ + hwloc_bitmap_asprintf(&setstring, obj->cpuset); + state->new_prop(state, "cpuset", setstring); + + hwloc_bitmap_asprintf(&setstring2, obj->complete_cpuset); + state->new_prop(state, "complete_cpuset", setstring2); + free(setstring2); + + if (v1export) + state->new_prop(state, "online_cpuset", setstring); + free(setstring); + + if (v1export || !obj->parent) { + hwloc_bitmap_t allowed_cpuset = hwloc_bitmap_dup(obj->cpuset); + hwloc_bitmap_and(allowed_cpuset, allowed_cpuset, topology->allowed_cpuset); + hwloc_bitmap_asprintf(&setstring, allowed_cpuset); + state->new_prop(state, "allowed_cpuset", setstring); + free(setstring); + hwloc_bitmap_free(allowed_cpuset); + } + } + + /* If exporting v1, we should clear second local NUMA bits from nodeset, + * but the importer will clear them anyway. + */ + hwloc_bitmap_asprintf(&setstring, obj->nodeset); + state->new_prop(state, "nodeset", setstring); + free(setstring); + + hwloc_bitmap_asprintf(&setstring, obj->complete_nodeset); + state->new_prop(state, "complete_nodeset", setstring); + free(setstring); + + if (v1export || !obj->parent) { + hwloc_bitmap_t allowed_nodeset = hwloc_bitmap_dup(obj->nodeset); + hwloc_bitmap_and(allowed_nodeset, allowed_nodeset, topology->allowed_nodeset); + hwloc_bitmap_asprintf(&setstring, allowed_nodeset); + state->new_prop(state, "allowed_nodeset", setstring); + free(setstring); + hwloc_bitmap_free(allowed_nodeset); + } + } + + if (!v1export) { + sprintf(tmp, "%llu", (unsigned long long) obj->gp_index); + state->new_prop(state, "gp_index", tmp); + } + + if (obj->name) { + char *name = hwloc__xml_export_safestrdup(obj->name); + state->new_prop(state, "name", name); + free(name); + } + if (!v1export && obj->subtype) { + char *subtype = hwloc__xml_export_safestrdup(obj->subtype); + state->new_prop(state, "subtype", subtype); + free(subtype); + } + + switch (obj->type) { + case HWLOC_OBJ_NUMANODE: + if (obj->attr->numanode.local_memory) { + sprintf(tmp, "%llu", (unsigned long long) obj->attr->numanode.local_memory); + state->new_prop(state, "local_memory", tmp); + } + for(i=0; iattr->numanode.page_types_len; i++) { + struct hwloc__xml_export_state_s childstate; + state->new_child(state, &childstate, "page_type"); + sprintf(tmp, "%llu", (unsigned long long) obj->attr->numanode.page_types[i].size); + childstate.new_prop(&childstate, "size", tmp); + sprintf(tmp, "%llu", (unsigned long long) obj->attr->numanode.page_types[i].count); + childstate.new_prop(&childstate, "count", tmp); + childstate.end_object(&childstate, "page_type"); + } + break; + case HWLOC_OBJ_L1CACHE: + case HWLOC_OBJ_L2CACHE: + case HWLOC_OBJ_L3CACHE: + case HWLOC_OBJ_L4CACHE: + case HWLOC_OBJ_L5CACHE: + case HWLOC_OBJ_L1ICACHE: + case HWLOC_OBJ_L2ICACHE: + case HWLOC_OBJ_L3ICACHE: + sprintf(tmp, "%llu", (unsigned long long) obj->attr->cache.size); + state->new_prop(state, "cache_size", tmp); + sprintf(tmp, "%u", obj->attr->cache.depth); + state->new_prop(state, "depth", tmp); + sprintf(tmp, "%u", (unsigned) obj->attr->cache.linesize); + state->new_prop(state, "cache_linesize", tmp); + sprintf(tmp, "%d", obj->attr->cache.associativity); + state->new_prop(state, "cache_associativity", tmp); + sprintf(tmp, "%d", (int) obj->attr->cache.type); + state->new_prop(state, "cache_type", tmp); + break; + case HWLOC_OBJ_GROUP: + if (v1export) { + sprintf(tmp, "%u", obj->attr->group.depth); + state->new_prop(state, "depth", tmp); + if (obj->attr->group.dont_merge) + state->new_prop(state, "dont_merge", "1"); + } else { + sprintf(tmp, "%u", obj->attr->group.kind); + state->new_prop(state, "kind", tmp); + sprintf(tmp, "%u", obj->attr->group.subkind); + state->new_prop(state, "subkind", tmp); + if (obj->attr->group.dont_merge) + state->new_prop(state, "dont_merge", "1"); + } + break; + case HWLOC_OBJ_BRIDGE: + sprintf(tmp, "%d-%d", (int) obj->attr->bridge.upstream_type, (int) obj->attr->bridge.downstream_type); + state->new_prop(state, "bridge_type", tmp); + sprintf(tmp, "%u", obj->attr->bridge.depth); + state->new_prop(state, "depth", tmp); + if (obj->attr->bridge.downstream_type == HWLOC_OBJ_BRIDGE_PCI) { + sprintf(tmp, "%04x:[%02x-%02x]", + (unsigned) obj->attr->bridge.downstream.pci.domain, + (unsigned) obj->attr->bridge.downstream.pci.secondary_bus, + (unsigned) obj->attr->bridge.downstream.pci.subordinate_bus); + state->new_prop(state, "bridge_pci", tmp); + } + if (obj->attr->bridge.upstream_type != HWLOC_OBJ_BRIDGE_PCI) + break; + /* FALLTHRU */ + case HWLOC_OBJ_PCI_DEVICE: + sprintf(tmp, "%04x:%02x:%02x.%01x", + (unsigned) obj->attr->pcidev.domain, + (unsigned) obj->attr->pcidev.bus, + (unsigned) obj->attr->pcidev.dev, + (unsigned) obj->attr->pcidev.func); + state->new_prop(state, "pci_busid", tmp); + sprintf(tmp, "%04x [%04x:%04x] [%04x:%04x] %02x", + (unsigned) obj->attr->pcidev.class_id, + (unsigned) obj->attr->pcidev.vendor_id, (unsigned) obj->attr->pcidev.device_id, + (unsigned) obj->attr->pcidev.subvendor_id, (unsigned) obj->attr->pcidev.subdevice_id, + (unsigned) obj->attr->pcidev.revision); + state->new_prop(state, "pci_type", tmp); + sprintf(tmp, "%f", obj->attr->pcidev.linkspeed); + state->new_prop(state, "pci_link_speed", tmp); + break; + case HWLOC_OBJ_OS_DEVICE: + sprintf(tmp, "%d", (int) obj->attr->osdev.type); + state->new_prop(state, "osdev_type", tmp); + break; + default: + break; + } + + for(i=0; iinfos_count; i++) { + char *name = hwloc__xml_export_safestrdup(obj->infos[i].name); + char *value = hwloc__xml_export_safestrdup(obj->infos[i].value); + struct hwloc__xml_export_state_s childstate; + state->new_child(state, &childstate, "info"); + childstate.new_prop(&childstate, "name", name); + childstate.new_prop(&childstate, "value", value); + childstate.end_object(&childstate, "info"); + free(name); + free(value); + } + if (v1export && obj->subtype) { + char *subtype = hwloc__xml_export_safestrdup(obj->subtype); + struct hwloc__xml_export_state_s childstate; + int is_coproctype = (obj->type == HWLOC_OBJ_OS_DEVICE && obj->attr->osdev.type == HWLOC_OBJ_OSDEV_COPROC); + state->new_child(state, &childstate, "info"); + childstate.new_prop(&childstate, "name", is_coproctype ? "CoProcType" : "Type"); + childstate.new_prop(&childstate, "value", subtype); + childstate.end_object(&childstate, "info"); + free(subtype); + } + + if (v1export && !obj->parent) { + /* only latency matrices covering the entire machine can be exported to v1 */ + struct hwloc_internal_distances_s *dist; + /* refresh distances since we need objects below */ + hwloc_internal_distances_refresh(topology); + for(dist = topology->first_dist; dist; dist = dist->next) { + struct hwloc__xml_export_state_s childstate; + unsigned nbobjs = dist->nbobjs; + int depth; + + if (nbobjs != (unsigned) hwloc_get_nbobjs_by_type(topology, dist->type)) + continue; + if (!(dist->kind & HWLOC_DISTANCES_KIND_MEANS_LATENCY)) + continue; + { + HWLOC_VLA(unsigned, logical_to_v2array, nbobjs); + for(i=0; iobjs[i]->logical_index] = i; + + /* compute the relative depth */ + if (dist->type == HWLOC_OBJ_NUMANODE) { + /* for NUMA nodes, use the highest normal-parent depth + 1 */ + depth = -1; + for(i=0; iobjs[i]->parent; + while (hwloc__obj_type_is_memory(parent->type)) + parent = parent->parent; + if (parent->depth+1 > depth) + depth = parent->depth+1; + } + } else { + /* for non-NUMA nodes, increase the object depth if any of them has memory above */ + int parent_with_memory = 0; + for(i=0; iobjs[i]->parent; + while (parent) { + if (parent->memory_first_child) { + parent_with_memory = 1; + goto done; + } + parent = parent->parent; + } + } + done: + depth = hwloc_get_type_depth(topology, dist->type) + parent_with_memory; + } + + state->new_child(state, &childstate, "distances"); + sprintf(tmp, "%u", nbobjs); + childstate.new_prop(&childstate, "nbobjs", tmp); + sprintf(tmp, "%d", depth); + childstate.new_prop(&childstate, "relative_depth", tmp); + sprintf(tmp, "%f", 1.f); + childstate.new_prop(&childstate, "latency_base", tmp); + for(i=0; ivalues[k]); + greatchildstate.new_prop(&greatchildstate, "value", tmp); + greatchildstate.end_object(&greatchildstate, "latency"); + } + } + childstate.end_object(&childstate, "distances"); + } + } + } + + if (obj->userdata && topology->userdata_export_cb) + topology->userdata_export_cb((void*) state, topology, obj); +} + +static void +hwloc__xml_v2export_object (hwloc__xml_export_state_t parentstate, hwloc_topology_t topology, hwloc_obj_t obj, unsigned long flags) +{ + struct hwloc__xml_export_state_s state; + hwloc_obj_t child; + + parentstate->new_child(parentstate, &state, "object"); + + hwloc__xml_export_object_contents(&state, topology, obj, flags); + + for_each_memory_child(child, obj) + hwloc__xml_v2export_object (&state, topology, child, flags); + for_each_child(child, obj) + hwloc__xml_v2export_object (&state, topology, child, flags); + for_each_io_child(child, obj) + hwloc__xml_v2export_object (&state, topology, child, flags); + for_each_misc_child(child, obj) + hwloc__xml_v2export_object (&state, topology, child, flags); + + state.end_object(&state, "object"); +} + +static void +hwloc__xml_v1export_object (hwloc__xml_export_state_t parentstate, hwloc_topology_t topology, hwloc_obj_t obj, unsigned long flags); + +static void +hwloc__xml_v1export_object_with_memory(hwloc__xml_export_state_t parentstate, hwloc_topology_t topology, hwloc_obj_t obj, unsigned long flags) +{ + struct hwloc__xml_export_state_s gstate, mstate, ostate, *state = parentstate; + hwloc_obj_t child; + + if (obj->parent->arity > 1 && obj->memory_arity > 1 && parentstate->global->v1_memory_group) { + /* child has sibling, we must add a Group around those memory children */ + hwloc_obj_t group = parentstate->global->v1_memory_group; + parentstate->new_child(parentstate, &gstate, "object"); + group->cpuset = obj->cpuset; + group->complete_cpuset = obj->complete_cpuset; + group->nodeset = obj->nodeset; + group->complete_nodeset = obj->complete_nodeset; + hwloc__xml_export_object_contents (&gstate, topology, group, flags); + group->cpuset = NULL; + group->complete_cpuset = NULL; + group->nodeset = NULL; + group->complete_nodeset = NULL; + state = &gstate; + } + + /* export first memory child */ + child = obj->memory_first_child; + assert(child->type == HWLOC_OBJ_NUMANODE); + state->new_child(state, &mstate, "object"); + hwloc__xml_export_object_contents (&mstate, topology, child, flags); + + /* then the actual object */ + mstate.new_child(&mstate, &ostate, "object"); + hwloc__xml_export_object_contents (&ostate, topology, obj, flags); + + /* then its normal/io/misc children */ + for_each_child(child, obj) + hwloc__xml_v1export_object (&ostate, topology, child, flags); + for_each_io_child(child, obj) + hwloc__xml_v1export_object (&ostate, topology, child, flags); + for_each_misc_child(child, obj) + hwloc__xml_v1export_object (&ostate, topology, child, flags); + + /* close object and first memory child */ + ostate.end_object(&ostate, "object"); + mstate.end_object(&mstate, "object"); + + /* now other memory children */ + for_each_memory_child(child, obj) + if (child->sibling_rank > 0) + hwloc__xml_v1export_object (state, topology, child, flags); + + if (state == &gstate) { + /* close group if any */ + gstate.end_object(&gstate, "object"); + } +} + +static void +hwloc__xml_v1export_object (hwloc__xml_export_state_t parentstate, hwloc_topology_t topology, hwloc_obj_t obj, unsigned long flags) +{ + struct hwloc__xml_export_state_s state; + hwloc_obj_t child; + + parentstate->new_child(parentstate, &state, "object"); + + hwloc__xml_export_object_contents(&state, topology, obj, flags); + + for_each_child(child, obj) { + if (!child->memory_arity) { + /* no memory child, just export normally */ + hwloc__xml_v1export_object (&state, topology, child, flags); + } else { + hwloc__xml_v1export_object_with_memory(&state, topology, child, flags); + } + } + + for_each_io_child(child, obj) + hwloc__xml_v1export_object (&state, topology, child, flags); + for_each_misc_child(child, obj) + hwloc__xml_v1export_object (&state, topology, child, flags); + + state.end_object(&state, "object"); +} + +#define EXPORT_ARRAY(state, type, nr, values, tagname, format, maxperline) do { \ + unsigned _i = 0; \ + while (_i<(nr)) { \ + char _tmp[255]; /* enough for (snprintf(format)+space) x maxperline */ \ + char _tmp2[16]; \ + size_t _len = 0; \ + unsigned _j; \ + struct hwloc__xml_export_state_s _childstate; \ + (state)->new_child(state, &_childstate, tagname); \ + for(_j=0; \ + _i+_j<(nr) && _jfirst_dist; dist; dist = dist->next) { + char tmp[255]; + unsigned nbobjs = dist->nbobjs; + struct hwloc__xml_export_state_s state; + + parentstate->new_child(parentstate, &state, "distances2"); + + state.new_prop(&state, "type", hwloc_obj_type_string(dist->type)); + sprintf(tmp, "%u", nbobjs); + state.new_prop(&state, "nbobjs", tmp); + sprintf(tmp, "%lu", dist->kind); + state.new_prop(&state, "kind", tmp); + + state.new_prop(&state, "indexing", + (dist->type == HWLOC_OBJ_NUMANODE || dist->type == HWLOC_OBJ_PU) ? "os" : "gp"); + /* TODO don't hardwire 10 below. either snprintf the max to guess it, or just append until the end of the buffer */ + EXPORT_ARRAY(&state, unsigned long long, nbobjs, dist->indexes, "indexes", "%llu", 10); + EXPORT_ARRAY(&state, unsigned long long, nbobjs*nbobjs, dist->values, "u64values", "%llu", 10); + state.end_object(&state, "distances2"); + } +} + +void +hwloc__xml_export_topology(hwloc__xml_export_state_t state, hwloc_topology_t topology, unsigned long flags) +{ + hwloc_obj_t root = hwloc_get_root_obj(topology); + + if (flags & HWLOC_TOPOLOGY_EXPORT_XML_FLAG_V1) { + if (root->memory_first_child) { + /* we don't use hwloc__xml_v1export_object_with_memory() because we want/can keep root above the numa node */ + struct hwloc__xml_export_state_s rstate, mstate; + hwloc_obj_t child; + /* export the root */ + state->new_child(state, &rstate, "object"); + hwloc__xml_export_object_contents (&rstate, topology, root, flags); + /* export first memory child */ + child = root->memory_first_child; + assert(child->type == HWLOC_OBJ_NUMANODE); + rstate.new_child(&rstate, &mstate, "object"); + hwloc__xml_export_object_contents (&mstate, topology, child, flags); + /* then its normal/io/misc children */ + for_each_child(child, root) + hwloc__xml_v1export_object (&mstate, topology, child, flags); + for_each_io_child(child, root) + hwloc__xml_v1export_object (&mstate, topology, child, flags); + for_each_misc_child(child, root) + hwloc__xml_v1export_object (&mstate, topology, child, flags); + /* close first memory child */ + mstate.end_object(&mstate, "object"); + /* now other memory children */ + for_each_memory_child(child, root) + if (child->sibling_rank > 0) + hwloc__xml_v1export_object (&rstate, topology, child, flags); + /* close the root */ + rstate.end_object(&rstate, "object"); + } else { + hwloc__xml_v1export_object(state, topology, root, flags); + } + + } else { + hwloc__xml_v2export_object (state, topology, root, flags); + hwloc__xml_v2export_distances (state, topology); + } +} + +void +hwloc__xml_export_diff(hwloc__xml_export_state_t parentstate, hwloc_topology_diff_t diff) +{ + while (diff) { + struct hwloc__xml_export_state_s state; + char tmp[255]; + + parentstate->new_child(parentstate, &state, "diff"); + + sprintf(tmp, "%d", (int) diff->generic.type); + state.new_prop(&state, "type", tmp); + + switch (diff->generic.type) { + case HWLOC_TOPOLOGY_DIFF_OBJ_ATTR: + sprintf(tmp, "%d", diff->obj_attr.obj_depth); + state.new_prop(&state, "obj_depth", tmp); + sprintf(tmp, "%u", diff->obj_attr.obj_index); + state.new_prop(&state, "obj_index", tmp); + + sprintf(tmp, "%d", (int) diff->obj_attr.diff.generic.type); + state.new_prop(&state, "obj_attr_type", tmp); + + switch (diff->obj_attr.diff.generic.type) { + case HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_SIZE: + sprintf(tmp, "%llu", (unsigned long long) diff->obj_attr.diff.uint64.index); + state.new_prop(&state, "obj_attr_index", tmp); + sprintf(tmp, "%llu", (unsigned long long) diff->obj_attr.diff.uint64.oldvalue); + state.new_prop(&state, "obj_attr_oldvalue", tmp); + sprintf(tmp, "%llu", (unsigned long long) diff->obj_attr.diff.uint64.newvalue); + state.new_prop(&state, "obj_attr_newvalue", tmp); + break; + case HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_NAME: + case HWLOC_TOPOLOGY_DIFF_OBJ_ATTR_INFO: + if (diff->obj_attr.diff.string.name) + state.new_prop(&state, "obj_attr_name", diff->obj_attr.diff.string.name); + state.new_prop(&state, "obj_attr_oldvalue", diff->obj_attr.diff.string.oldvalue); + state.new_prop(&state, "obj_attr_newvalue", diff->obj_attr.diff.string.newvalue); + break; + } + + break; + default: + assert(0); + } + state.end_object(&state, "diff"); + + diff = diff->generic.next; + } +} + +/********************************** + ********* main XML export ******** + **********************************/ + +/* this can be the first XML call */ +int hwloc_topology_export_xml(hwloc_topology_t topology, const char *filename, unsigned long flags) +{ + hwloc_localeswitch_declare; + struct hwloc__xml_export_data_s edata; + int force_nolibxml; + int ret; + + if (!topology->is_loaded) { + errno = EINVAL; + return -1; + } + + assert(hwloc_nolibxml_callbacks); /* the core called components_init() for the topology */ + + if (flags & ~HWLOC_TOPOLOGY_EXPORT_XML_FLAG_V1) { + errno = EINVAL; + return -1; + } + + hwloc_internal_distances_refresh(topology); + + hwloc_localeswitch_init(); + + edata.v1_memory_group = NULL; + if (flags & HWLOC_TOPOLOGY_EXPORT_XML_FLAG_V1) + /* temporary group to be used during v1 export of memory children */ + edata.v1_memory_group = hwloc_alloc_setup_object(topology, HWLOC_OBJ_GROUP, HWLOC_UNKNOWN_INDEX); + + force_nolibxml = hwloc_nolibxml_export(); +retry: + if (!hwloc_libxml_callbacks || (hwloc_nolibxml_callbacks && force_nolibxml)) + ret = hwloc_nolibxml_callbacks->export_file(topology, &edata, filename, flags); + else { + ret = hwloc_libxml_callbacks->export_file(topology, &edata, filename, flags); + if (ret < 0 && errno == ENOSYS) { + hwloc_libxml_callbacks = NULL; + goto retry; + } + } + + if (edata.v1_memory_group) + hwloc_free_unlinked_object(edata.v1_memory_group); + + hwloc_localeswitch_fini(); + return ret; +} + +/* this can be the first XML call */ +int hwloc_topology_export_xmlbuffer(hwloc_topology_t topology, char **xmlbuffer, int *buflen, unsigned long flags) +{ + hwloc_localeswitch_declare; + struct hwloc__xml_export_data_s edata; + int force_nolibxml; + int ret; + + if (!topology->is_loaded) { + errno = EINVAL; + return -1; + } + + assert(hwloc_nolibxml_callbacks); /* the core called components_init() for the topology */ + + if (flags & ~HWLOC_TOPOLOGY_EXPORT_XML_FLAG_V1) { + errno = EINVAL; + return -1; + } + + hwloc_internal_distances_refresh(topology); + + hwloc_localeswitch_init(); + + edata.v1_memory_group = NULL; + if (flags & HWLOC_TOPOLOGY_EXPORT_XML_FLAG_V1) + /* temporary group to be used during v1 export of memory children */ + edata.v1_memory_group = hwloc_alloc_setup_object(topology, HWLOC_OBJ_GROUP, HWLOC_UNKNOWN_INDEX); + + force_nolibxml = hwloc_nolibxml_export(); +retry: + if (!hwloc_libxml_callbacks || (hwloc_nolibxml_callbacks && force_nolibxml)) + ret = hwloc_nolibxml_callbacks->export_buffer(topology, &edata, xmlbuffer, buflen, flags); + else { + ret = hwloc_libxml_callbacks->export_buffer(topology, &edata, xmlbuffer, buflen, flags); + if (ret < 0 && errno == ENOSYS) { + hwloc_libxml_callbacks = NULL; + goto retry; + } + } + + if (edata.v1_memory_group) + hwloc_free_unlinked_object(edata.v1_memory_group); + + hwloc_localeswitch_fini(); + return ret; +} + +/* this can be the first XML call */ +int +hwloc_topology_diff_export_xml(hwloc_topology_diff_t diff, const char *refname, + const char *filename) +{ + hwloc_localeswitch_declare; + hwloc_topology_diff_t tmpdiff; + int force_nolibxml; + int ret; + + tmpdiff = diff; + while (tmpdiff) { + if (tmpdiff->generic.type == HWLOC_TOPOLOGY_DIFF_TOO_COMPLEX) { + errno = EINVAL; + return -1; + } + tmpdiff = tmpdiff->generic.next; + } + + hwloc_components_init(); + assert(hwloc_nolibxml_callbacks); + + hwloc_localeswitch_init(); + + force_nolibxml = hwloc_nolibxml_export(); +retry: + if (!hwloc_libxml_callbacks || (hwloc_nolibxml_callbacks && force_nolibxml)) + ret = hwloc_nolibxml_callbacks->export_diff_file(diff, refname, filename); + else { + ret = hwloc_libxml_callbacks->export_diff_file(diff, refname, filename); + if (ret < 0 && errno == ENOSYS) { + hwloc_libxml_callbacks = NULL; + goto retry; + } + } + + hwloc_localeswitch_fini(); + hwloc_components_fini(); + return ret; +} + +/* this can be the first XML call */ +int +hwloc_topology_diff_export_xmlbuffer(hwloc_topology_diff_t diff, const char *refname, + char **xmlbuffer, int *buflen) +{ + hwloc_localeswitch_declare; + hwloc_topology_diff_t tmpdiff; + int force_nolibxml; + int ret; + + tmpdiff = diff; + while (tmpdiff) { + if (tmpdiff->generic.type == HWLOC_TOPOLOGY_DIFF_TOO_COMPLEX) { + errno = EINVAL; + return -1; + } + tmpdiff = tmpdiff->generic.next; + } + + hwloc_components_init(); + assert(hwloc_nolibxml_callbacks); + + hwloc_localeswitch_init(); + + force_nolibxml = hwloc_nolibxml_export(); +retry: + if (!hwloc_libxml_callbacks || (hwloc_nolibxml_callbacks && force_nolibxml)) + ret = hwloc_nolibxml_callbacks->export_diff_buffer(diff, refname, xmlbuffer, buflen); + else { + ret = hwloc_libxml_callbacks->export_diff_buffer(diff, refname, xmlbuffer, buflen); + if (ret < 0 && errno == ENOSYS) { + hwloc_libxml_callbacks = NULL; + goto retry; + } + } + + hwloc_localeswitch_fini(); + hwloc_components_fini(); + return ret; +} + +void hwloc_free_xmlbuffer(hwloc_topology_t topology __hwloc_attribute_unused, char *xmlbuffer) +{ + int force_nolibxml; + + assert(hwloc_nolibxml_callbacks); /* the core called components_init() for the topology */ + + force_nolibxml = hwloc_nolibxml_export(); + if (!hwloc_libxml_callbacks || (hwloc_nolibxml_callbacks && force_nolibxml)) + hwloc_nolibxml_callbacks->free_buffer(xmlbuffer); + else + hwloc_libxml_callbacks->free_buffer(xmlbuffer); +} + +void +hwloc_topology_set_userdata_export_callback(hwloc_topology_t topology, + void (*export)(void *reserved, struct hwloc_topology *topology, struct hwloc_obj *obj)) +{ + topology->userdata_export_cb = export; +} + +static void +hwloc__export_obj_userdata(hwloc__xml_export_state_t parentstate, int encoded, + const char *name, size_t length, const void *buffer, size_t encoded_length) +{ + struct hwloc__xml_export_state_s state; + char tmp[255]; + parentstate->new_child(parentstate, &state, "userdata"); + if (name) + state.new_prop(&state, "name", name); + sprintf(tmp, "%lu", (unsigned long) length); + state.new_prop(&state, "length", tmp); + if (encoded) + state.new_prop(&state, "encoding", "base64"); + if (encoded_length) + state.add_content(&state, buffer, encoded ? encoded_length : length); + state.end_object(&state, "userdata"); +} + +int +hwloc_export_obj_userdata(void *reserved, + struct hwloc_topology *topology, struct hwloc_obj *obj __hwloc_attribute_unused, + const char *name, const void *buffer, size_t length) +{ + hwloc__xml_export_state_t state = reserved; + + if (!buffer) { + errno = EINVAL; + return -1; + } + + if ((name && hwloc__xml_export_check_buffer(name, strlen(name)) < 0) + || hwloc__xml_export_check_buffer(buffer, length) < 0) { + errno = EINVAL; + return -1; + } + + if (topology->userdata_not_decoded) { + int encoded; + size_t encoded_length; + const char *realname; + if (!strncmp(name, "base64", 6)) { + encoded = 1; + encoded_length = BASE64_ENCODED_LENGTH(length); + } else { + assert(!strncmp(name, "normal", 6)); + encoded = 0; + encoded_length = length; + } + if (name[6] == ':') + realname = name+7; + else { + assert(!strcmp(name+6, "-anon")); + realname = NULL; + } + hwloc__export_obj_userdata(state, encoded, realname, length, buffer, encoded_length); + + } else + hwloc__export_obj_userdata(state, 0, name, length, buffer, length); + + return 0; +} + +int +hwloc_export_obj_userdata_base64(void *reserved, + struct hwloc_topology *topology __hwloc_attribute_unused, struct hwloc_obj *obj __hwloc_attribute_unused, + const char *name, const void *buffer, size_t length) +{ + hwloc__xml_export_state_t state = reserved; + size_t encoded_length; + char *encoded_buffer; + int ret __hwloc_attribute_unused; + + if (!buffer) { + errno = EINVAL; + return -1; + } + + assert(!topology->userdata_not_decoded); + + if (name && hwloc__xml_export_check_buffer(name, strlen(name)) < 0) { + errno = EINVAL; + return -1; + } + + encoded_length = BASE64_ENCODED_LENGTH(length); + encoded_buffer = malloc(encoded_length+1); + if (!encoded_buffer) { + errno = ENOMEM; + return -1; + } + + ret = hwloc_encode_to_base64(buffer, length, encoded_buffer, encoded_length+1); + assert(ret == (int) encoded_length); + + hwloc__export_obj_userdata(state, 1, name, length, encoded_buffer, encoded_length); + + free(encoded_buffer); + return 0; +} + +void +hwloc_topology_set_userdata_import_callback(hwloc_topology_t topology, + void (*import)(struct hwloc_topology *topology, struct hwloc_obj *obj, const char *name, const void *buffer, size_t length)) +{ + topology->userdata_import_cb = import; +} + +/*************************************** + ************ XML component ************ + ***************************************/ + +static void +hwloc_xml_backend_disable(struct hwloc_backend *backend) +{ + struct hwloc_xml_backend_data_s *data = backend->private_data; + data->backend_exit(data); + free(data->msgprefix); + free(data); +} + +static struct hwloc_backend * +hwloc_xml_component_instantiate(struct hwloc_disc_component *component, + const void *_data1, + const void *_data2, + const void *_data3) +{ + struct hwloc_xml_backend_data_s *data; + struct hwloc_backend *backend; + const char *env; + int force_nolibxml; + const char * xmlpath = (const char *) _data1; + const char * xmlbuffer = (const char *) _data2; + int xmlbuflen = (int)(uintptr_t) _data3; + const char *local_basename; + int err; + + assert(hwloc_nolibxml_callbacks); /* the core called components_init() for the component's topology */ + + if (!xmlpath && !xmlbuffer) { + env = getenv("HWLOC_XMLFILE"); + if (env) { + /* 'xml' was given in HWLOC_COMPONENTS without a filename */ + xmlpath = env; + } else { + errno = EINVAL; + goto out; + } + } + + backend = hwloc_backend_alloc(component); + if (!backend) + goto out; + + data = malloc(sizeof(*data)); + if (!data) { + errno = ENOMEM; + goto out_with_backend; + } + + backend->private_data = data; + backend->discover = hwloc_look_xml; + backend->disable = hwloc_xml_backend_disable; + backend->is_thissystem = 0; + + if (xmlpath) { + local_basename = strrchr(xmlpath, '/'); + if (local_basename) + local_basename++; + else + local_basename = xmlpath; + } else { + local_basename = "xmlbuffer"; + } + data->msgprefix = strdup(local_basename); + + force_nolibxml = hwloc_nolibxml_import(); +retry: + if (!hwloc_libxml_callbacks || (hwloc_nolibxml_callbacks && force_nolibxml)) + err = hwloc_nolibxml_callbacks->backend_init(data, xmlpath, xmlbuffer, xmlbuflen); + else { + err = hwloc_libxml_callbacks->backend_init(data, xmlpath, xmlbuffer, xmlbuflen); + if (err < 0 && errno == ENOSYS) { + hwloc_libxml_callbacks = NULL; + goto retry; + } + } + if (err < 0) + goto out_with_data; + + return backend; + + out_with_data: + free(data->msgprefix); + free(data); + out_with_backend: + free(backend); + out: + return NULL; +} + +static struct hwloc_disc_component hwloc_xml_disc_component = { + HWLOC_DISC_COMPONENT_TYPE_GLOBAL, + "xml", + ~0, + hwloc_xml_component_instantiate, + 30, + 1, + NULL +}; + +const struct hwloc_component hwloc_xml_component = { + HWLOC_COMPONENT_ABI, + NULL, NULL, + HWLOC_COMPONENT_TYPE_DISC, + 0, + &hwloc_xml_disc_component +}; diff --git a/src/3rdparty/hwloc/src/topology.c b/src/3rdparty/hwloc/src/topology.c new file mode 100644 index 000000000..55678a084 --- /dev/null +++ b/src/3rdparty/hwloc/src/topology.c @@ -0,0 +1,4484 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2018 Inria. All rights reserved. + * Copyright © 2009-2012 Université Bordeaux + * Copyright © 2009-2011 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +#include + +#define _ATFILE_SOURCE +#include +#include +#ifdef HAVE_DIRENT_H +#include +#endif +#ifdef HAVE_UNISTD_H +#include +#endif +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include +#include + +#ifdef HAVE_MACH_MACH_INIT_H +#include +#endif +#ifdef HAVE_MACH_MACH_HOST_H +#include +#endif + +#ifdef HAVE_SYS_PARAM_H +#include +#endif + +#ifdef HAVE_SYS_SYSCTL_H +#include +#endif + +#ifdef HWLOC_WIN_SYS +#include +#endif + +unsigned hwloc_get_api_version(void) +{ + return HWLOC_API_VERSION; +} + +int hwloc_topology_abi_check(hwloc_topology_t topology) +{ + return topology->topology_abi != HWLOC_TOPOLOGY_ABI ? -1 : 0; +} + +int hwloc_hide_errors(void) +{ + static int hide = 0; + static int checked = 0; + if (!checked) { + const char *envvar = getenv("HWLOC_HIDE_ERRORS"); + if (envvar) + hide = atoi(envvar); + checked = 1; + } + return hide; +} + +void hwloc_report_os_error(const char *msg, int line) +{ + static int reported = 0; + + if (!reported && !hwloc_hide_errors()) { + fprintf(stderr, "****************************************************************************\n"); + fprintf(stderr, "* hwloc %s received invalid information from the operating system.\n", HWLOC_VERSION); + fprintf(stderr, "*\n"); + fprintf(stderr, "* %s\n", msg); + fprintf(stderr, "* Error occurred in topology.c line %d\n", line); + fprintf(stderr, "*\n"); + fprintf(stderr, "* The following FAQ entry in the hwloc documentation may help:\n"); + fprintf(stderr, "* What should I do when hwloc reports \"operating system\" warnings?\n"); + fprintf(stderr, "* Otherwise please report this error message to the hwloc user's mailing list,\n"); +#ifdef HWLOC_LINUX_SYS + fprintf(stderr, "* along with the files generated by the hwloc-gather-topology script.\n"); +#else + fprintf(stderr, "* along with any relevant topology information from your platform.\n"); +#endif + fprintf(stderr, "* \n"); + fprintf(stderr, "* hwloc will now ignore this invalid topology information and continue.\n"); + fprintf(stderr, "****************************************************************************\n"); + reported = 1; + } +} + +#if defined(HAVE_SYSCTLBYNAME) +int hwloc_get_sysctlbyname(const char *name, int64_t *ret) +{ + union { + int32_t i32; + int64_t i64; + } n; + size_t size = sizeof(n); + if (sysctlbyname(name, &n, &size, NULL, 0)) + return -1; + switch (size) { + case sizeof(n.i32): + *ret = n.i32; + break; + case sizeof(n.i64): + *ret = n.i64; + break; + default: + return -1; + } + return 0; +} +#endif + +#if defined(HAVE_SYSCTL) +int hwloc_get_sysctl(int name[], unsigned namelen, int *ret) +{ + int n; + size_t size = sizeof(n); + if (sysctl(name, namelen, &n, &size, NULL, 0)) + return -1; + if (size != sizeof(n)) + return -1; + *ret = n; + return 0; +} +#endif + +/* Return the OS-provided number of processors. Unlike other methods such as + reading sysfs on Linux, this method is not virtualizable; thus it's only + used as a fall-back method, allowing virtual backends (FSROOT, etc) to + have the desired effect. */ +#ifndef HWLOC_WIN_SYS /* The windows implementation is in topology-windows.c */ +int +hwloc_fallback_nbprocessors(struct hwloc_topology *topology __hwloc_attribute_unused) { + int n; +#if HAVE_DECL__SC_NPROCESSORS_ONLN + n = sysconf(_SC_NPROCESSORS_ONLN); +#elif HAVE_DECL__SC_NPROC_ONLN + n = sysconf(_SC_NPROC_ONLN); +#elif HAVE_DECL__SC_NPROCESSORS_CONF + n = sysconf(_SC_NPROCESSORS_CONF); +#elif HAVE_DECL__SC_NPROC_CONF + n = sysconf(_SC_NPROC_CONF); +#elif defined(HAVE_HOST_INFO) && HAVE_HOST_INFO + struct host_basic_info info; + mach_msg_type_number_t count = HOST_BASIC_INFO_COUNT; + host_info(mach_host_self(), HOST_BASIC_INFO, (integer_t*) &info, &count); + n = info.avail_cpus; +#elif defined(HAVE_SYSCTLBYNAME) + int64_t nn; + if (hwloc_get_sysctlbyname("hw.ncpu", &nn)) + nn = -1; + n = nn; +#elif defined(HAVE_SYSCTL) && HAVE_DECL_CTL_HW && HAVE_DECL_HW_NCPU + static int name[2] = {CTL_HW, HW_NCPU}; + if (hwloc_get_sysctl(name, sizeof(name)/sizeof(*name), &n)) + n = -1; +#else +#ifdef __GNUC__ +#warning No known way to discover number of available processors on this system +#endif + n = -1; +#endif + return n; +} +#endif /* !HWLOC_WIN_SYS */ + +/* + * Use the given number of processors to set a PU level. + */ +void +hwloc_setup_pu_level(struct hwloc_topology *topology, + unsigned nb_pus) +{ + struct hwloc_obj *obj; + unsigned oscpu,cpu; + + hwloc_debug("%s", "\n\n * CPU cpusets *\n\n"); + for (cpu=0,oscpu=0; cpucpuset = hwloc_bitmap_alloc(); + hwloc_bitmap_only(obj->cpuset, oscpu); + + hwloc_debug_2args_bitmap("cpu %u (os %u) has cpuset %s\n", + cpu, oscpu, obj->cpuset); + hwloc_insert_object_by_cpuset(topology, obj); + + cpu++; + } +} + +/* Traverse children of a parent in a safe way: reread the next pointer as + * appropriate to prevent crash on child deletion: */ +#define for_each_child_safe(child, parent, pchild) \ + for (pchild = &(parent)->first_child, child = *pchild; \ + child; \ + /* Check whether the current child was not dropped. */ \ + (*pchild == child ? pchild = &(child->next_sibling) : NULL), \ + /* Get pointer to next child. */ \ + child = *pchild) +#define for_each_memory_child_safe(child, parent, pchild) \ + for (pchild = &(parent)->memory_first_child, child = *pchild; \ + child; \ + /* Check whether the current child was not dropped. */ \ + (*pchild == child ? pchild = &(child->next_sibling) : NULL), \ + /* Get pointer to next child. */ \ + child = *pchild) +#define for_each_io_child_safe(child, parent, pchild) \ + for (pchild = &(parent)->io_first_child, child = *pchild; \ + child; \ + /* Check whether the current child was not dropped. */ \ + (*pchild == child ? pchild = &(child->next_sibling) : NULL), \ + /* Get pointer to next child. */ \ + child = *pchild) +#define for_each_misc_child_safe(child, parent, pchild) \ + for (pchild = &(parent)->misc_first_child, child = *pchild; \ + child; \ + /* Check whether the current child was not dropped. */ \ + (*pchild == child ? pchild = &(child->next_sibling) : NULL), \ + /* Get pointer to next child. */ \ + child = *pchild) + +#ifdef HWLOC_DEBUG +/* Just for debugging. */ +static void +hwloc_debug_print_object(int indent __hwloc_attribute_unused, hwloc_obj_t obj) +{ + char type[64], idx[12], attr[1024], *cpuset = NULL; + hwloc_debug("%*s", 2*indent, ""); + hwloc_obj_type_snprintf(type, sizeof(type), obj, 1); + if (obj->os_index != HWLOC_UNKNOWN_INDEX) + snprintf(idx, sizeof(idx), "#%u", obj->os_index); + else + *idx = '\0'; + hwloc_obj_attr_snprintf(attr, sizeof(attr), obj, " ", 1); + hwloc_debug("%s%s%s%s%s", type, idx, *attr ? "(" : "", attr, *attr ? ")" : ""); + if (obj->name) + hwloc_debug(" name \"%s\"", obj->name); + if (obj->subtype) + hwloc_debug(" subtype \"%s\"", obj->subtype); + if (obj->cpuset) { + hwloc_bitmap_asprintf(&cpuset, obj->cpuset); + hwloc_debug(" cpuset %s", cpuset); + free(cpuset); + } + if (obj->complete_cpuset) { + hwloc_bitmap_asprintf(&cpuset, obj->complete_cpuset); + hwloc_debug(" complete %s", cpuset); + free(cpuset); + } + if (obj->nodeset) { + hwloc_bitmap_asprintf(&cpuset, obj->nodeset); + hwloc_debug(" nodeset %s", cpuset); + free(cpuset); + } + if (obj->complete_nodeset) { + hwloc_bitmap_asprintf(&cpuset, obj->complete_nodeset); + hwloc_debug(" completeN %s", cpuset); + free(cpuset); + } + if (obj->arity) + hwloc_debug(" arity %u", obj->arity); + hwloc_debug("%s", "\n"); +} + +static void +hwloc_debug_print_objects(int indent __hwloc_attribute_unused, hwloc_obj_t obj) +{ + hwloc_obj_t child; + hwloc_debug_print_object(indent, obj); + for_each_child (child, obj) + hwloc_debug_print_objects(indent + 1, child); + for_each_memory_child (child, obj) + hwloc_debug_print_objects(indent + 1, child); + for_each_io_child (child, obj) + hwloc_debug_print_objects(indent + 1, child); + for_each_misc_child (child, obj) + hwloc_debug_print_objects(indent + 1, child); +} +#else /* !HWLOC_DEBUG */ +#define hwloc_debug_print_object(indent, obj) do { /* nothing */ } while (0) +#define hwloc_debug_print_objects(indent, obj) do { /* nothing */ } while (0) +#endif /* !HWLOC_DEBUG */ + +void hwloc__free_infos(struct hwloc_info_s *infos, unsigned count) +{ + unsigned i; + for(i=0; iinfos, &obj->infos_count, name, value); +} + +/* This function may be called with topology->tma set, it cannot free() or realloc() */ +static int hwloc__tma_dup_infos(struct hwloc_tma *tma, hwloc_obj_t new, hwloc_obj_t src) +{ + unsigned i, j; + new->infos = hwloc_tma_calloc(tma, src->infos_count * sizeof(*src->infos)); + if (!new->infos) + return -1; + for(i=0; iinfos_count; i++) { + new->infos[i].name = hwloc_tma_strdup(tma, src->infos[i].name); + new->infos[i].value = hwloc_tma_strdup(tma, src->infos[i].value); + if (!new->infos[i].name || !new->infos[i].value) + goto failed; + } + new->infos_count = src->infos_count; + return 0; + + failed: + assert(!tma || !tma->dontfree); /* this tma cannot fail to allocate */ + for(j=0; j<=i; j++) { + free(new->infos[i].name); + free(new->infos[i].value); + } + free(new->infos); + new->infos = NULL; + return -1; +} + +static void +hwloc__free_object_contents(hwloc_obj_t obj) +{ + switch (obj->type) { + case HWLOC_OBJ_NUMANODE: + free(obj->attr->numanode.page_types); + break; + default: + break; + } + hwloc__free_infos(obj->infos, obj->infos_count); + free(obj->attr); + free(obj->children); + free(obj->subtype); + free(obj->name); + hwloc_bitmap_free(obj->cpuset); + hwloc_bitmap_free(obj->complete_cpuset); + hwloc_bitmap_free(obj->nodeset); + hwloc_bitmap_free(obj->complete_nodeset); +} + +/* Free an object and all its content. */ +void +hwloc_free_unlinked_object(hwloc_obj_t obj) +{ + hwloc__free_object_contents(obj); + free(obj); +} + +/* Replace old with contents of new object, and make new freeable by the caller. + * Only updates next_sibling/first_child pointers, + * so may only be used during early discovery. + */ +static void +hwloc_replace_linked_object(hwloc_obj_t old, hwloc_obj_t new) +{ + /* drop old fields */ + hwloc__free_object_contents(old); + /* copy old tree pointers to new */ + new->parent = old->parent; + new->next_sibling = old->next_sibling; + new->first_child = old->first_child; + new->memory_first_child = old->memory_first_child; + new->io_first_child = old->io_first_child; + new->misc_first_child = old->misc_first_child; + /* copy new contents to old now that tree pointers are OK */ + memcpy(old, new, sizeof(*old)); + /* clear new to that we may free it */ + memset(new, 0,sizeof(*new)); +} + +/* Remove an object and its children from its parent and free them. + * Only updates next_sibling/first_child pointers, + * so may only be used during early discovery or during destroy. + */ +static void +unlink_and_free_object_and_children(hwloc_obj_t *pobj) +{ + hwloc_obj_t obj = *pobj, child, *pchild; + + for_each_child_safe(child, obj, pchild) + unlink_and_free_object_and_children(pchild); + for_each_memory_child_safe(child, obj, pchild) + unlink_and_free_object_and_children(pchild); + for_each_io_child_safe(child, obj, pchild) + unlink_and_free_object_and_children(pchild); + for_each_misc_child_safe(child, obj, pchild) + unlink_and_free_object_and_children(pchild); + + *pobj = obj->next_sibling; + hwloc_free_unlinked_object(obj); +} + +/* Free an object and its children without unlinking from parent. + */ +void +hwloc_free_object_and_children(hwloc_obj_t obj) +{ + unlink_and_free_object_and_children(&obj); +} + +/* Free an object, its next siblings and their children without unlinking from parent. + */ +void +hwloc_free_object_siblings_and_children(hwloc_obj_t obj) +{ + while (obj) + unlink_and_free_object_and_children(&obj); +} + +/* insert the (non-empty) list of sibling starting at firstnew as new children of newparent, + * and return the address of the pointer to the next one + */ +static hwloc_obj_t * +insert_siblings_list(hwloc_obj_t *firstp, hwloc_obj_t firstnew, hwloc_obj_t newparent) +{ + hwloc_obj_t tmp; + assert(firstnew); + *firstp = tmp = firstnew; + tmp->parent = newparent; + while (tmp->next_sibling) { + tmp = tmp->next_sibling; + tmp->parent = newparent; + } + return &tmp->next_sibling; +} + +/* Take the new list starting at firstnew and prepend it to the old list starting at *firstp, + * and mark the new children as children of newparent. + * May be used during early or late discovery (updates prev_sibling and sibling_rank). + * List firstnew must be non-NULL. + */ +static void +prepend_siblings_list(hwloc_obj_t *firstp, hwloc_obj_t firstnew, hwloc_obj_t newparent) +{ + hwloc_obj_t *tmpp, tmp, last; + unsigned length; + + /* update parent pointers and find the length and end of the new list */ + for(length = 0, tmpp = &firstnew, last = NULL ; *tmpp; length++, last = *tmpp, tmpp = &((*tmpp)->next_sibling)) + (*tmpp)->parent = newparent; + + /* update sibling_rank */ + for(tmp = *firstp; tmp; tmp = tmp->next_sibling) + tmp->sibling_rank += length; /* if it wasn't initialized yet, it'll be overwritten later */ + + /* place the existing list at the end of the new one */ + *tmpp = *firstp; + if (*firstp) + (*firstp)->prev_sibling = last; + + /* use the beginning of the new list now */ + *firstp = firstnew; +} + +/* Take the new list starting at firstnew and append it to the old list starting at *firstp, + * and mark the new children as children of newparent. + * May be used during early or late discovery (updates prev_sibling and sibling_rank). + */ +static void +append_siblings_list(hwloc_obj_t *firstp, hwloc_obj_t firstnew, hwloc_obj_t newparent) +{ + hwloc_obj_t *tmpp, tmp, last; + unsigned length; + + /* find the length and end of the existing list */ + for(length = 0, tmpp = firstp, last = NULL ; *tmpp; length++, last = *tmpp, tmpp = &((*tmpp)->next_sibling)); + + /* update parent pointers and sibling_rank */ + for(tmp = firstnew; tmp; tmp = tmp->next_sibling) { + tmp->parent = newparent; + tmp->sibling_rank += length; /* if it wasn't set yet, it'll be overwritten later */ + } + + /* place new list at the end of the old one */ + *tmpp = firstnew; + if (firstnew) + firstnew->prev_sibling = last; +} + +/* Remove an object from its parent and free it. + * Only updates next_sibling/first_child pointers, + * so may only be used during early discovery. + * + * Children are inserted in the parent. + * If children should be inserted somewhere else (e.g. when merging with a child), + * the caller should move them before calling this function. + */ +static void +unlink_and_free_single_object(hwloc_obj_t *pparent) +{ + hwloc_obj_t old = *pparent; + hwloc_obj_t *lastp; + + if (old->type == HWLOC_OBJ_MISC) { + /* Misc object */ + + /* no normal children */ + assert(!old->first_child); + /* no memory children */ + assert(!old->memory_first_child); + /* no I/O children */ + assert(!old->io_first_child); + + if (old->misc_first_child) + /* insert old misc object children as new siblings below parent instead of old */ + lastp = insert_siblings_list(pparent, old->misc_first_child, old->parent); + else + lastp = pparent; + /* append old siblings back */ + *lastp = old->next_sibling; + + } else if (hwloc__obj_type_is_io(old->type)) { + /* I/O object */ + + /* no normal children */ + assert(!old->first_child); + /* no memory children */ + assert(!old->memory_first_child); + + if (old->io_first_child) + /* insert old I/O object children as new siblings below parent instead of old */ + lastp = insert_siblings_list(pparent, old->io_first_child, old->parent); + else + lastp = pparent; + /* append old siblings back */ + *lastp = old->next_sibling; + + /* append old Misc children to parent */ + if (old->misc_first_child) + append_siblings_list(&old->parent->misc_first_child, old->misc_first_child, old->parent); + + } else if (hwloc__obj_type_is_memory(old->type)) { + /* memory object */ + + /* no normal children */ + assert(!old->first_child); + /* no I/O children */ + assert(!old->io_first_child); + + if (old->memory_first_child) + /* insert old memory object children as new siblings below parent instead of old */ + lastp = insert_siblings_list(pparent, old->memory_first_child, old->parent); + else + lastp = pparent; + /* append old siblings back */ + *lastp = old->next_sibling; + + /* append old Misc children to parent */ + if (old->misc_first_child) + append_siblings_list(&old->parent->misc_first_child, old->misc_first_child, old->parent); + + } else { + /* Normal object */ + + if (old->first_child) + /* insert old object children as new siblings below parent instead of old */ + lastp = insert_siblings_list(pparent, old->first_child, old->parent); + else + lastp = pparent; + /* append old siblings back */ + *lastp = old->next_sibling; + + /* append old memory, I/O and Misc children to parent + * old->parent cannot be NULL (removing root), misc children should have been moved by the caller earlier. + */ + if (old->memory_first_child) + append_siblings_list(&old->parent->memory_first_child, old->memory_first_child, old->parent); + if (old->io_first_child) + append_siblings_list(&old->parent->io_first_child, old->io_first_child, old->parent); + if (old->misc_first_child) + append_siblings_list(&old->parent->misc_first_child, old->misc_first_child, old->parent); + } + + hwloc_free_unlinked_object(old); +} + +/* This function may use a tma, it cannot free() or realloc() */ +static int +hwloc__duplicate_object(struct hwloc_topology *newtopology, + struct hwloc_obj *newparent, + struct hwloc_obj *newobj, + struct hwloc_obj *src) +{ + struct hwloc_tma *tma = newtopology->tma; + hwloc_obj_t *level; + unsigned level_width; + size_t len; + unsigned i; + hwloc_obj_t child, prev; + int err = 0; + + /* either we're duplicating to an already allocated new root, which has no newparent, + * or we're duplicating to a non-yet allocated new non-root, which will have a newparent. + */ + assert(!newparent == !!newobj); + + if (!newobj) { + newobj = hwloc_alloc_setup_object(newtopology, src->type, src->os_index); + if (!newobj) + return -1; + } + + /* duplicate all non-object-pointer fields */ + newobj->logical_index = src->logical_index; + newobj->depth = src->depth; + newobj->sibling_rank = src->sibling_rank; + + newobj->type = src->type; + newobj->os_index = src->os_index; + newobj->gp_index = src->gp_index; + newobj->symmetric_subtree = src->symmetric_subtree; + + if (src->name) + newobj->name = hwloc_tma_strdup(tma, src->name); + if (src->subtype) + newobj->subtype = hwloc_tma_strdup(tma, src->subtype); + newobj->userdata = src->userdata; + + newobj->total_memory = src->total_memory; + + memcpy(newobj->attr, src->attr, sizeof(*newobj->attr)); + + if (src->type == HWLOC_OBJ_NUMANODE && src->attr->numanode.page_types_len) { + len = src->attr->numanode.page_types_len * sizeof(struct hwloc_memory_page_type_s); + newobj->attr->numanode.page_types = hwloc_tma_malloc(tma, len); + memcpy(newobj->attr->numanode.page_types, src->attr->numanode.page_types, len); + } + + newobj->cpuset = hwloc_bitmap_tma_dup(tma, src->cpuset); + newobj->complete_cpuset = hwloc_bitmap_tma_dup(tma, src->complete_cpuset); + newobj->nodeset = hwloc_bitmap_tma_dup(tma, src->nodeset); + newobj->complete_nodeset = hwloc_bitmap_tma_dup(tma, src->complete_nodeset); + + hwloc__tma_dup_infos(tma, newobj, src); + + /* find our level */ + if (src->depth < 0) { + i = HWLOC_SLEVEL_FROM_DEPTH(src->depth); + level = newtopology->slevels[i].objs; + level_width = newtopology->slevels[i].nbobjs; + /* deal with first/last pointers of special levels, even if not really needed */ + if (!newobj->logical_index) + newtopology->slevels[i].first = newobj; + if (newobj->logical_index == newtopology->slevels[i].nbobjs - 1) + newtopology->slevels[i].last = newobj; + } else { + level = newtopology->levels[src->depth]; + level_width = newtopology->level_nbobjects[src->depth]; + } + /* place us for real */ + assert(newobj->logical_index < level_width); + level[newobj->logical_index] = newobj; + /* link to already-inserted cousins + * (hwloc_pci_belowroot_apply_locality() can cause out-of-order logical indexes) + */ + if (newobj->logical_index > 0 && level[newobj->logical_index-1]) { + newobj->prev_cousin = level[newobj->logical_index-1]; + level[newobj->logical_index-1]->next_cousin = newobj; + } + if (newobj->logical_index < level_width-1 && level[newobj->logical_index+1]) { + newobj->next_cousin = level[newobj->logical_index+1]; + level[newobj->logical_index+1]->prev_cousin = newobj; + } + + /* prepare for children */ + if (src->arity) { + newobj->children = hwloc_tma_malloc(tma, src->arity * sizeof(*newobj->children)); + if (!newobj->children) + return -1; + } + newobj->arity = src->arity; + newobj->memory_arity = src->memory_arity; + newobj->io_arity = src->io_arity; + newobj->misc_arity = src->misc_arity; + + /* actually insert children now */ + for_each_child(child, src) { + err = hwloc__duplicate_object(newtopology, newobj, NULL, child); + if (err < 0) + goto out_with_children; + } + for_each_memory_child(child, src) { + err = hwloc__duplicate_object(newtopology, newobj, NULL, child); + if (err < 0) + return err; + } + for_each_io_child(child, src) { + err = hwloc__duplicate_object(newtopology, newobj, NULL, child); + if (err < 0) + goto out_with_children; + } + for_each_misc_child(child, src) { + err = hwloc__duplicate_object(newtopology, newobj, NULL, child); + if (err < 0) + goto out_with_children; + } + + out_with_children: + + /* link children if all of them where inserted */ + if (!err) { + /* only next_sibling is set by insert_by_parent(). + * sibling_rank was set above. + */ + if (newobj->arity) { + newobj->children[0]->prev_sibling = NULL; + for(i=1; iarity; i++) + newobj->children[i]->prev_sibling = newobj->children[i-1]; + newobj->last_child = newobj->children[newobj->arity-1]; + } + if (newobj->memory_arity) { + child = newobj->memory_first_child; + prev = NULL; + while (child) { + child->prev_sibling = prev; + prev = child; + child = child->next_sibling; + } + } + if (newobj->io_arity) { + child = newobj->io_first_child; + prev = NULL; + while (child) { + child->prev_sibling = prev; + prev = child; + child = child->next_sibling; + } + } + if (newobj->misc_arity) { + child = newobj->misc_first_child; + prev = NULL; + while (child) { + child->prev_sibling = prev; + prev = child; + child = child->next_sibling; + } + } + } + + /* some children insertion may have failed, but some children may have been inserted below us already. + * keep inserting ourself and let the caller clean the entire tree if we return an error. + */ + + if (newparent) { + /* no need to check the children insert order here, the source topology + * is supposed to be OK already, and we have debug asserts. + */ + hwloc_insert_object_by_parent(newtopology, newparent, newobj); + + /* place us inside our parent children array */ + if (hwloc__obj_type_is_normal(newobj->type)) + newparent->children[newobj->sibling_rank] = newobj; + } + + return err; +} + +static int +hwloc__topology_init (struct hwloc_topology **topologyp, unsigned nblevels, struct hwloc_tma *tma); + +/* This function may use a tma, it cannot free() or realloc() */ +int +hwloc__topology_dup(hwloc_topology_t *newp, + hwloc_topology_t old, + struct hwloc_tma *tma) +{ + hwloc_topology_t new; + hwloc_obj_t newroot; + hwloc_obj_t oldroot = hwloc_get_root_obj(old); + unsigned i; + int err; + + if (!old->is_loaded) { + errno = EINVAL; + return -1; + } + + err = hwloc__topology_init(&new, old->nb_levels_allocated, tma); + if (err < 0) + goto out; + + new->flags = old->flags; + memcpy(new->type_filter, old->type_filter, sizeof(old->type_filter)); + new->is_thissystem = old->is_thissystem; + new->is_loaded = 1; + new->pid = old->pid; + new->next_gp_index = old->next_gp_index; + + memcpy(&new->binding_hooks, &old->binding_hooks, sizeof(old->binding_hooks)); + + memcpy(new->support.discovery, old->support.discovery, sizeof(*old->support.discovery)); + memcpy(new->support.cpubind, old->support.cpubind, sizeof(*old->support.cpubind)); + memcpy(new->support.membind, old->support.membind, sizeof(*old->support.membind)); + + new->allowed_cpuset = hwloc_bitmap_tma_dup(tma, old->allowed_cpuset); + new->allowed_nodeset = hwloc_bitmap_tma_dup(tma, old->allowed_nodeset); + + new->userdata_export_cb = old->userdata_export_cb; + new->userdata_import_cb = old->userdata_import_cb; + new->userdata_not_decoded = old->userdata_not_decoded; + + assert(!old->machine_memory.local_memory); + assert(!old->machine_memory.page_types_len); + assert(!old->machine_memory.page_types); + + for(i = HWLOC_OBJ_TYPE_MIN; i < HWLOC_OBJ_TYPE_MAX; i++) + new->type_depth[i] = old->type_depth[i]; + + /* duplicate levels and we'll place objects there when duplicating objects */ + new->nb_levels = old->nb_levels; + assert(new->nb_levels_allocated >= new->nb_levels); + for(i=1 /* root level already allocated */ ; inb_levels; i++) { + new->level_nbobjects[i] = old->level_nbobjects[i]; + new->levels[i] = hwloc_tma_calloc(tma, new->level_nbobjects[i] * sizeof(*new->levels[i])); + } + for(i=0; islevels[i].nbobjs = old->slevels[i].nbobjs; + if (new->slevels[i].nbobjs) + new->slevels[i].objs = hwloc_tma_calloc(tma, new->slevels[i].nbobjs * sizeof(*new->slevels[i].objs)); + } + + /* recursively duplicate object children */ + newroot = hwloc_get_root_obj(new); + err = hwloc__duplicate_object(new, NULL, newroot, oldroot); + if (err < 0) + goto out_with_topology; + + err = hwloc_internal_distances_dup(new, old); + if (err < 0) + goto out_with_topology; + + /* we connected everything during duplication */ + new->modified = 0; + + /* no need to duplicate backends, topology is already loaded */ + new->backends = NULL; + new->get_pci_busid_cpuset_backend = NULL; + +#ifndef HWLOC_DEBUG + if (getenv("HWLOC_DEBUG_CHECK")) +#endif + hwloc_topology_check(new); + + *newp = new; + return 0; + + out_with_topology: + assert(!tma || !tma->dontfree); /* this tma cannot fail to allocate */ + hwloc_topology_destroy(new); + out: + return -1; +} + +int +hwloc_topology_dup(hwloc_topology_t *newp, + hwloc_topology_t old) +{ + return hwloc__topology_dup(newp, old, NULL); +} + +/* WARNING: The indexes of this array MUST match the ordering that of + the obj_order_type[] array, below. Specifically, the values must + be laid out such that: + + obj_order_type[obj_type_order[N]] = N + + for all HWLOC_OBJ_* values of N. Put differently: + + obj_type_order[A] = B + + where the A values are in order of the hwloc_obj_type_t enum, and + the B values are the corresponding indexes of obj_order_type. + + We can't use C99 syntax to initialize this in a little safer manner + -- bummer. :-( + + Correctness is asserted in hwloc_topology_init() when debug is enabled. + */ +/***** Make sure you update obj_type_priority[] below as well. *****/ +static const unsigned obj_type_order[] = { + /* first entry is HWLOC_OBJ_MACHINE */ 0, + /* next entry is HWLOC_OBJ_PACKAGE */ 3, + /* next entry is HWLOC_OBJ_CORE */ 12, + /* next entry is HWLOC_OBJ_PU */ 16, + /* next entry is HWLOC_OBJ_L1CACHE */ 10, + /* next entry is HWLOC_OBJ_L2CACHE */ 8, + /* next entry is HWLOC_OBJ_L3CACHE */ 6, + /* next entry is HWLOC_OBJ_L4CACHE */ 5, + /* next entry is HWLOC_OBJ_L5CACHE */ 4, + /* next entry is HWLOC_OBJ_L1ICACHE */ 11, + /* next entry is HWLOC_OBJ_L2ICACHE */ 9, + /* next entry is HWLOC_OBJ_L3ICACHE */ 7, + /* next entry is HWLOC_OBJ_GROUP */ 1, + /* next entry is HWLOC_OBJ_NUMANODE */ 2, + /* next entry is HWLOC_OBJ_BRIDGE */ 13, + /* next entry is HWLOC_OBJ_PCI_DEVICE */ 14, + /* next entry is HWLOC_OBJ_OS_DEVICE */ 15, + /* next entry is HWLOC_OBJ_MISC */ 17 +}; + +#ifndef NDEBUG /* only used in debug check assert if !NDEBUG */ +static const hwloc_obj_type_t obj_order_type[] = { + HWLOC_OBJ_MACHINE, + HWLOC_OBJ_GROUP, + HWLOC_OBJ_NUMANODE, + HWLOC_OBJ_PACKAGE, + HWLOC_OBJ_L5CACHE, + HWLOC_OBJ_L4CACHE, + HWLOC_OBJ_L3CACHE, + HWLOC_OBJ_L3ICACHE, + HWLOC_OBJ_L2CACHE, + HWLOC_OBJ_L2ICACHE, + HWLOC_OBJ_L1CACHE, + HWLOC_OBJ_L1ICACHE, + HWLOC_OBJ_CORE, + HWLOC_OBJ_BRIDGE, + HWLOC_OBJ_PCI_DEVICE, + HWLOC_OBJ_OS_DEVICE, + HWLOC_OBJ_PU, + HWLOC_OBJ_MISC /* Misc is always a leaf */ +}; +#endif +/***** Make sure you update obj_type_priority[] below as well. *****/ + +/* priority to be used when merging identical parent/children object + * (in merge_useless_child), keep the highest priority one. + * + * Always keep Machine/NUMANode/PU/PCIDev/OSDev + * then Core + * then Package + * then Cache, + * then Instruction Caches + * then always drop Group/Misc/Bridge. + * + * Some type won't actually ever be involved in such merging. + */ +/***** Make sure you update this array when changing the list of types. *****/ +static const int obj_type_priority[] = { + /* first entry is HWLOC_OBJ_MACHINE */ 90, + /* next entry is HWLOC_OBJ_PACKAGE */ 40, + /* next entry is HWLOC_OBJ_CORE */ 60, + /* next entry is HWLOC_OBJ_PU */ 100, + /* next entry is HWLOC_OBJ_L1CACHE */ 20, + /* next entry is HWLOC_OBJ_L2CACHE */ 20, + /* next entry is HWLOC_OBJ_L3CACHE */ 20, + /* next entry is HWLOC_OBJ_L4CACHE */ 20, + /* next entry is HWLOC_OBJ_L5CACHE */ 20, + /* next entry is HWLOC_OBJ_L1ICACHE */ 19, + /* next entry is HWLOC_OBJ_L2ICACHE */ 19, + /* next entry is HWLOC_OBJ_L3ICACHE */ 19, + /* next entry is HWLOC_OBJ_GROUP */ 0, + /* next entry is HWLOC_OBJ_NUMANODE */ 100, + /* next entry is HWLOC_OBJ_BRIDGE */ 0, + /* next entry is HWLOC_OBJ_PCI_DEVICE */ 100, + /* next entry is HWLOC_OBJ_OS_DEVICE */ 100, + /* next entry is HWLOC_OBJ_MISC */ 0 +}; + +int hwloc_compare_types (hwloc_obj_type_t type1, hwloc_obj_type_t type2) +{ + unsigned order1 = obj_type_order[type1]; + unsigned order2 = obj_type_order[type2]; + + /* only normal objects are comparable. others are only comparable with machine */ + if (!hwloc__obj_type_is_normal(type1) + && hwloc__obj_type_is_normal(type2) && type2 != HWLOC_OBJ_MACHINE) + return HWLOC_TYPE_UNORDERED; + if (!hwloc__obj_type_is_normal(type2) + && hwloc__obj_type_is_normal(type1) && type1 != HWLOC_OBJ_MACHINE) + return HWLOC_TYPE_UNORDERED; + + return order1 - order2; +} + +enum hwloc_obj_cmp_e { + HWLOC_OBJ_EQUAL = HWLOC_BITMAP_EQUAL, /**< \brief Equal */ + HWLOC_OBJ_INCLUDED = HWLOC_BITMAP_INCLUDED, /**< \brief Strictly included into */ + HWLOC_OBJ_CONTAINS = HWLOC_BITMAP_CONTAINS, /**< \brief Strictly contains */ + HWLOC_OBJ_INTERSECTS = HWLOC_BITMAP_INTERSECTS, /**< \brief Intersects, but no inclusion! */ + HWLOC_OBJ_DIFFERENT = HWLOC_BITMAP_DIFFERENT /**< \brief No intersection */ +}; + +static enum hwloc_obj_cmp_e +hwloc_type_cmp(hwloc_obj_t obj1, hwloc_obj_t obj2) +{ + hwloc_obj_type_t type1 = obj1->type; + hwloc_obj_type_t type2 = obj2->type; + int compare; + + compare = hwloc_compare_types(type1, type2); + if (compare == HWLOC_TYPE_UNORDERED) + return HWLOC_OBJ_DIFFERENT; /* we cannot do better */ + if (compare > 0) + return HWLOC_OBJ_INCLUDED; + if (compare < 0) + return HWLOC_OBJ_CONTAINS; + + if (obj1->type == HWLOC_OBJ_GROUP + && (obj1->attr->group.kind != obj2->attr->group.kind + || obj1->attr->group.subkind != obj2->attr->group.subkind)) + return HWLOC_OBJ_DIFFERENT; /* we cannot do better */ + + return HWLOC_OBJ_EQUAL; +} + +/* + * How to compare objects based on cpusets. + */ + +static int +hwloc_obj_cmp_sets(hwloc_obj_t obj1, hwloc_obj_t obj2) +{ + hwloc_bitmap_t set1, set2; + int res = HWLOC_OBJ_DIFFERENT; + + assert(!hwloc__obj_type_is_special(obj1->type)); + assert(!hwloc__obj_type_is_special(obj2->type)); + + /* compare cpusets first */ + if (obj1->complete_cpuset && obj2->complete_cpuset) { + set1 = obj1->complete_cpuset; + set2 = obj2->complete_cpuset; + } else { + set1 = obj1->cpuset; + set2 = obj2->cpuset; + } + if (set1 && set2 && !hwloc_bitmap_iszero(set1) && !hwloc_bitmap_iszero(set2)) { + res = hwloc_bitmap_compare_inclusion(set1, set2); + if (res == HWLOC_OBJ_INTERSECTS) + return HWLOC_OBJ_INTERSECTS; + } + + /* then compare nodesets, and combine the results */ + if (obj1->complete_nodeset && obj2->complete_nodeset) { + set1 = obj1->complete_nodeset; + set2 = obj2->complete_nodeset; + } else { + set1 = obj1->nodeset; + set2 = obj2->nodeset; + } + if (set1 && set2 && !hwloc_bitmap_iszero(set1) && !hwloc_bitmap_iszero(set2)) { + int noderes = hwloc_bitmap_compare_inclusion(set1, set2); + /* deal with conflicting cpusets/nodesets inclusions */ + if (noderes == HWLOC_OBJ_INCLUDED) { + if (res == HWLOC_OBJ_CONTAINS) + /* contradicting order for cpusets and nodesets */ + return HWLOC_OBJ_INTERSECTS; + res = HWLOC_OBJ_INCLUDED; + + } else if (noderes == HWLOC_OBJ_CONTAINS) { + if (res == HWLOC_OBJ_INCLUDED) + /* contradicting order for cpusets and nodesets */ + return HWLOC_OBJ_INTERSECTS; + res = HWLOC_OBJ_CONTAINS; + + } else if (noderes == HWLOC_OBJ_INTERSECTS) { + return HWLOC_OBJ_INTERSECTS; + + } else { + /* nodesets are different, keep the cpuset order */ + + } + } + + return res; +} + +/* Compare object cpusets based on complete_cpuset if defined (always correctly ordered), + * or fallback to the main cpusets (only correctly ordered during early insert before disallowed bits are cleared). + * + * This is the sane way to compare object among a horizontal level. + */ +int +hwloc__object_cpusets_compare_first(hwloc_obj_t obj1, hwloc_obj_t obj2) +{ + if (obj1->complete_cpuset && obj2->complete_cpuset) + return hwloc_bitmap_compare_first(obj1->complete_cpuset, obj2->complete_cpuset); + else if (obj1->cpuset && obj2->cpuset) + return hwloc_bitmap_compare_first(obj1->cpuset, obj2->cpuset); + else if (obj1->complete_nodeset && obj2->complete_nodeset) + return hwloc_bitmap_compare_first(obj1->complete_nodeset, obj2->complete_nodeset); + else if (obj1->nodeset && obj2->nodeset) + return hwloc_bitmap_compare_first(obj1->nodeset, obj2->nodeset); + return 0; +} + +/* format the obj info to print in error messages */ +static void +hwloc__report_error_format_obj(char *buf, size_t buflen, hwloc_obj_t obj) +{ + char typestr[64]; + char *cpusetstr; + char *nodesetstr = NULL; + hwloc_obj_type_snprintf(typestr, sizeof(typestr), obj, 0); + hwloc_bitmap_asprintf(&cpusetstr, obj->cpuset); + if (obj->nodeset) /* may be missing during insert */ + hwloc_bitmap_asprintf(&nodesetstr, obj->nodeset); + if (obj->os_index != HWLOC_UNKNOWN_INDEX) + snprintf(buf, buflen, "%s (P#%u cpuset %s%s%s)", + typestr, obj->os_index, cpusetstr, + nodesetstr ? " nodeset " : "", + nodesetstr ? nodesetstr : ""); + else + snprintf(buf, buflen, "%s (cpuset %s%s%s)", + typestr, cpusetstr, + nodesetstr ? " nodeset " : "", + nodesetstr ? nodesetstr : ""); + free(cpusetstr); + free(nodesetstr); +} + +/* + * How to insert objects into the topology. + * + * Note: during detection, only the first_child and next_sibling pointers are + * kept up to date. Others are computed only once topology detection is + * complete. + */ + +/* merge new object attributes in old. + * use old if defined, otherwise use new. + */ +static void +merge_insert_equal(hwloc_obj_t new, hwloc_obj_t old) +{ + if (old->os_index == HWLOC_UNKNOWN_INDEX) + old->os_index = new->os_index; + + if (new->infos_count) { + /* FIXME: dedup */ + hwloc__move_infos(&old->infos, &old->infos_count, + &new->infos, &new->infos_count); + } + + if (new->name && !old->name) { + old->name = new->name; + new->name = NULL; + } + if (new->subtype && !old->subtype) { + old->subtype = new->subtype; + new->subtype = NULL; + } + + /* Ignore userdata. It will be NULL before load(). + * It may be non-NULL if alloc+insert_group() after load(). + */ + + switch(new->type) { + case HWLOC_OBJ_NUMANODE: + if (new->attr->numanode.local_memory && !old->attr->numanode.local_memory) { + /* no memory in old, use new memory */ + old->attr->numanode.local_memory = new->attr->numanode.local_memory; + free(old->attr->numanode.page_types); + old->attr->numanode.page_types_len = new->attr->numanode.page_types_len; + old->attr->numanode.page_types = new->attr->numanode.page_types; + new->attr->numanode.page_types = NULL; + new->attr->numanode.page_types_len = 0; + } + /* old->attr->numanode.total_memory will be updated by propagate_total_memory() */ + break; + case HWLOC_OBJ_L1CACHE: + case HWLOC_OBJ_L2CACHE: + case HWLOC_OBJ_L3CACHE: + case HWLOC_OBJ_L4CACHE: + case HWLOC_OBJ_L5CACHE: + case HWLOC_OBJ_L1ICACHE: + case HWLOC_OBJ_L2ICACHE: + case HWLOC_OBJ_L3ICACHE: + if (!old->attr->cache.size) + old->attr->cache.size = new->attr->cache.size; + if (!old->attr->cache.linesize) + old->attr->cache.size = new->attr->cache.linesize; + if (!old->attr->cache.associativity) + old->attr->cache.size = new->attr->cache.linesize; + break; + default: + break; + } +} + +/* returns the result of merge, or NULL if not merged */ +static __hwloc_inline hwloc_obj_t +hwloc__insert_try_merge_group(hwloc_obj_t old, hwloc_obj_t new) +{ + if (new->type == HWLOC_OBJ_GROUP && old->type == HWLOC_OBJ_GROUP) { + /* which group do we keep? */ + if (new->attr->group.dont_merge) { + if (old->attr->group.dont_merge) + /* nobody wants to be merged */ + return NULL; + + /* keep the new one, it doesn't want to be merged */ + hwloc_replace_linked_object(old, new); + return new; + + } else { + if (old->attr->group.dont_merge) + /* keep the old one, it doesn't want to be merged */ + return old; + + /* compare subkinds to decice who to keep */ + if (new->attr->group.kind < old->attr->group.kind) + hwloc_replace_linked_object(old, new); + return old; + } + } + + if (new->type == HWLOC_OBJ_GROUP && !new->attr->group.dont_merge) { + + if (old->type == HWLOC_OBJ_PU && new->attr->group.kind == HWLOC_GROUP_KIND_MEMORY) + /* Never merge Memory groups with PU, we don't want to attach Memory under PU */ + return NULL; + + /* Remove the Group now. The normal ignore code path wouldn't tell us whether the Group was removed or not, + * while some callers need to know (at least hwloc_topology_insert_group()). + */ + return old; + + } else if (old->type == HWLOC_OBJ_GROUP && !old->attr->group.dont_merge) { + + if (new->type == HWLOC_OBJ_PU && old->attr->group.kind == HWLOC_GROUP_KIND_MEMORY) + /* Never merge Memory groups with PU, we don't want to attach Memory under PU */ + return NULL; + + /* Replace the Group with the new object contents + * and let the caller free the new object + */ + hwloc_replace_linked_object(old, new); + return old; + + } else { + /* cannot merge */ + return NULL; + } +} + +/* Try to insert OBJ in CUR, recurse if needed. + * Returns the object if it was inserted, + * the remaining object it was merged, + * NULL if failed to insert. + */ +static struct hwloc_obj * +hwloc___insert_object_by_cpuset(struct hwloc_topology *topology, hwloc_obj_t cur, hwloc_obj_t obj, + hwloc_report_error_t report_error) +{ + hwloc_obj_t child, next_child = NULL; + /* These will always point to the pointer to their next last child. */ + hwloc_obj_t *cur_children = &cur->first_child; + hwloc_obj_t *obj_children = &obj->first_child; + /* Pointer where OBJ should be put */ + hwloc_obj_t *putp = NULL; /* OBJ position isn't found yet */ + + assert(!hwloc__obj_type_is_memory(obj->type)); + + /* Iteration with prefetching to be completely safe against CHILD removal. + * The list is already sorted by cpuset, and there's no intersection between siblings. + */ + for (child = cur->first_child, child ? next_child = child->next_sibling : NULL; + child; + child = next_child, child ? next_child = child->next_sibling : NULL) { + + int res = hwloc_obj_cmp_sets(obj, child); + int setres = res; + + if (res == HWLOC_OBJ_EQUAL) { + hwloc_obj_t merged = hwloc__insert_try_merge_group(child, obj); + if (merged) + return merged; + /* otherwise compare actual types to decide of the inclusion */ + res = hwloc_type_cmp(obj, child); + } + + switch (res) { + case HWLOC_OBJ_EQUAL: + /* Two objects with same type. + * Groups are handled above. + */ + merge_insert_equal(obj, child); + /* Already present, no need to insert. */ + return child; + + case HWLOC_OBJ_INCLUDED: + /* OBJ is strictly contained is some child of CUR, go deeper. */ + return hwloc___insert_object_by_cpuset(topology, child, obj, report_error); + + case HWLOC_OBJ_INTERSECTS: + if (report_error) { + char childstr[512]; + char objstr[512]; + char msg[1100]; + hwloc__report_error_format_obj(objstr, sizeof(objstr), obj); + hwloc__report_error_format_obj(childstr, sizeof(childstr), child); + snprintf(msg, sizeof(msg), "%s intersects with %s without inclusion!", objstr, childstr); + report_error(msg, __LINE__); + } + goto putback; + + case HWLOC_OBJ_DIFFERENT: + /* OBJ should be a child of CUR before CHILD, mark its position if not found yet. */ + if (!putp && hwloc__object_cpusets_compare_first(obj, child) < 0) + /* Don't insert yet, there could be intersect errors later */ + putp = cur_children; + /* Advance cur_children. */ + cur_children = &child->next_sibling; + break; + + case HWLOC_OBJ_CONTAINS: + /* OBJ contains CHILD, remove CHILD from CUR */ + *cur_children = child->next_sibling; + child->next_sibling = NULL; + /* Put CHILD in OBJ */ + *obj_children = child; + obj_children = &child->next_sibling; + child->parent = obj; + if (setres == HWLOC_OBJ_EQUAL) { + obj->memory_first_child = child->memory_first_child; + child->memory_first_child = NULL; + } + break; + } + } + /* cur/obj_children points to last CUR/OBJ child next_sibling pointer, which must be NULL. */ + assert(!*obj_children); + assert(!*cur_children); + + /* Put OBJ where it belongs, or in last in CUR's children. */ + if (!putp) + putp = cur_children; + obj->next_sibling = *putp; + *putp = obj; + obj->parent = cur; + + topology->modified = 1; + return obj; + + putback: + /* Put-back OBJ children in CUR and return an error. */ + if (putp) + cur_children = putp; /* No need to try to insert before where OBJ was supposed to go */ + else + cur_children = &cur->first_child; /* Start from the beginning */ + /* We can insert in order, but there can be holes in the middle. */ + while ((child = obj->first_child) != NULL) { + /* Remove from OBJ */ + obj->first_child = child->next_sibling; + obj->parent = cur; + /* Find child position in CUR, and insert. */ + while (*cur_children && hwloc__object_cpusets_compare_first(*cur_children, child) < 0) + cur_children = &(*cur_children)->next_sibling; + child->next_sibling = *cur_children; + *cur_children = child; + } + return NULL; +} + +/* this differs from hwloc_get_obj_covering_cpuset() by: + * - not looking at the parent cpuset first, which means we can insert + * below root even if root PU bits are not set yet (PU are inserted later). + * - returning the first child that exactly matches instead of walking down in case + * of identical children. + */ +static struct hwloc_obj * +hwloc__find_obj_covering_memory_cpuset(struct hwloc_topology *topology, hwloc_obj_t parent, hwloc_bitmap_t cpuset) +{ + hwloc_obj_t child = hwloc_get_child_covering_cpuset(topology, cpuset, parent); + if (!child) + return parent; + if (child && hwloc_bitmap_isequal(child->cpuset, cpuset)) + return child; + return hwloc__find_obj_covering_memory_cpuset(topology, child, cpuset); +} + +static struct hwloc_obj * +hwloc__find_insert_memory_parent(struct hwloc_topology *topology, hwloc_obj_t obj, + hwloc_report_error_t report_error) +{ + hwloc_obj_t parent, group, result; + + if (hwloc_bitmap_iszero(obj->cpuset)) { + /* CPU-less go in dedicated group below root */ + parent = topology->levels[0][0]; + + } else { + /* find the highest obj covering the cpuset */ + parent = hwloc__find_obj_covering_memory_cpuset(topology, topology->levels[0][0], obj->cpuset); + if (!parent) { + /* fallback to root */ + parent = hwloc_get_root_obj(topology); + } + + if (parent->type == HWLOC_OBJ_PU) { + /* Never attach to PU, try parent */ + parent = parent->parent; + assert(parent); + } + + /* TODO: if root->cpuset was updated earlier, we would be sure whether the group will remain identical to root */ + if (parent != topology->levels[0][0] && hwloc_bitmap_isequal(parent->cpuset, obj->cpuset)) + /* that parent is fine */ + return parent; + } + + if (!hwloc_filter_check_keep_object_type(topology, HWLOC_OBJ_GROUP)) + /* even if parent isn't perfect, we don't want an intermediate group */ + return parent; + + /* need to insert an intermediate group for attaching the NUMA node */ + group = hwloc_alloc_setup_object(topology, HWLOC_OBJ_GROUP, HWLOC_UNKNOWN_INDEX); + if (!group) + /* failed to create the group, fallback to larger parent */ + return parent; + + group->attr->group.kind = HWLOC_GROUP_KIND_MEMORY; + group->cpuset = hwloc_bitmap_dup(obj->cpuset); + group->complete_cpuset = hwloc_bitmap_dup(obj->complete_cpuset); + /* we could duplicate nodesets too but hwloc__insert_object_by_cpuset() + * doesn't actually need it. and it could prevent future calls from reusing + * that groups for other NUMA nodes. + */ + if (!group->cpuset != !obj->cpuset + || !group->complete_cpuset != !obj->complete_cpuset) { + /* failed to create the group, fallback to larger parent */ + hwloc_free_unlinked_object(group); + return parent; + } + + result = hwloc__insert_object_by_cpuset(topology, parent, group, report_error); + if (!result) { + /* failed to insert, fallback to larger parent */ + return parent; + } + + assert(result == group); + return group; +} + +/*attach the given memory object below the given normal parent. */ +struct hwloc_obj * +hwloc__attach_memory_object(struct hwloc_topology *topology, hwloc_obj_t parent, + hwloc_obj_t obj, + hwloc_report_error_t report_error __hwloc_attribute_unused) +{ + hwloc_obj_t *cur_children; + + assert(parent); + assert(hwloc__obj_type_is_normal(parent->type)); + +#if 0 + /* TODO: enable this instead of hack in fixup_sets once NUMA nodes are inserted late */ + /* copy the parent cpuset in case it's larger than expected. + * we could also keep the cpuset smaller than the parent and say that a normal-parent + * can have multiple memory children with smaller cpusets. + * However, the user decided the ignore Groups, so hierarchy/locality loss is expected. + */ + hwloc_bitmap_copy(obj->cpuset, parent->cpuset); +#endif + + /* only NUMA nodes are memory for now, just append to the end of the list */ + assert(obj->type == HWLOC_OBJ_NUMANODE); + assert(obj->nodeset); + cur_children = &parent->memory_first_child; + while (*cur_children) { + /* TODO check that things are inserted in order. + * it's OK for KNL, the only user so far + */ + cur_children = &(*cur_children)->next_sibling; + } + *cur_children = obj; + obj->next_sibling = NULL; + + /* Initialize the complete nodeset if needed */ + if (!obj->complete_nodeset) { + obj->complete_nodeset = hwloc_bitmap_dup(obj->nodeset); + } + + /* Add the bit to the top sets, and to the parent CPU-side object */ + if (obj->type == HWLOC_OBJ_NUMANODE) { + if (hwloc_bitmap_isset(obj->nodeset, obj->os_index)) + hwloc_bitmap_set(topology->levels[0][0]->nodeset, obj->os_index); + hwloc_bitmap_set(topology->levels[0][0]->complete_nodeset, obj->os_index); + } + + topology->modified = 1; + return obj; +} + +/* insertion routine that lets you change the error reporting callback */ +struct hwloc_obj * +hwloc__insert_object_by_cpuset(struct hwloc_topology *topology, hwloc_obj_t root, + hwloc_obj_t obj, + hwloc_report_error_t report_error) +{ + struct hwloc_obj *result; + +#ifdef HWLOC_DEBUG + assert(!hwloc__obj_type_is_special(obj->type)); + + /* we need at least one non-NULL set (normal or complete, cpuset or nodeset) */ + assert(obj->cpuset || obj->complete_cpuset || obj->nodeset || obj->complete_nodeset); + /* we support the case where all of them are empty. + * it may happen when hwloc__find_insert_memory_parent() + * inserts a Group for a CPU-less NUMA-node. + */ +#endif + + if (hwloc__obj_type_is_memory(obj->type)) { + if (!root) { + root = hwloc__find_insert_memory_parent(topology, obj, report_error); + if (!root) { + hwloc_free_unlinked_object(obj); + return NULL; + } + } + return hwloc__attach_memory_object(topology, root, obj, report_error); + } + + if (!root) + /* Start at the top. */ + root = topology->levels[0][0]; + + result = hwloc___insert_object_by_cpuset(topology, root, obj, report_error); + if (result && result->type == HWLOC_OBJ_PU) { + /* Add the bit to the top sets */ + if (hwloc_bitmap_isset(result->cpuset, result->os_index)) + hwloc_bitmap_set(topology->levels[0][0]->cpuset, result->os_index); + hwloc_bitmap_set(topology->levels[0][0]->complete_cpuset, result->os_index); + } + if (result != obj) { + /* either failed to insert, or got merged, free the original object */ + hwloc_free_unlinked_object(obj); + } + return result; +} + +/* the default insertion routine warns in case of error. + * it's used by most backends */ +struct hwloc_obj * +hwloc_insert_object_by_cpuset(struct hwloc_topology *topology, hwloc_obj_t obj) +{ + return hwloc__insert_object_by_cpuset(topology, NULL, obj, hwloc_report_os_error); +} + +void +hwloc_insert_object_by_parent(struct hwloc_topology *topology, hwloc_obj_t parent, hwloc_obj_t obj) +{ + hwloc_obj_t *current; + + if (obj->type == HWLOC_OBJ_MISC) { + /* Append to the end of the Misc list */ + for (current = &parent->misc_first_child; *current; current = &(*current)->next_sibling); + } else if (hwloc__obj_type_is_io(obj->type)) { + /* Append to the end of the I/O list */ + for (current = &parent->io_first_child; *current; current = &(*current)->next_sibling); + } else if (hwloc__obj_type_is_memory(obj->type)) { + /* Append to the end of the memory list */ + for (current = &parent->memory_first_child; *current; current = &(*current)->next_sibling); + /* Add the bit to the top sets */ + if (obj->type == HWLOC_OBJ_NUMANODE) { + if (hwloc_bitmap_isset(obj->nodeset, obj->os_index)) + hwloc_bitmap_set(topology->levels[0][0]->nodeset, obj->os_index); + hwloc_bitmap_set(topology->levels[0][0]->complete_nodeset, obj->os_index); + } + } else { + /* Append to the end of the list. + * The caller takes care of inserting children in the right cpuset order, without intersection between them. + * Duplicating doesn't need to check the order since the source topology is supposed to be OK already. + * XML reorders if needed, and fails on intersecting siblings. + * Other callers just insert random objects such as I/O or Misc, no cpuset issue there. + */ + for (current = &parent->first_child; *current; current = &(*current)->next_sibling); + /* Add the bit to the top sets */ + if (obj->type == HWLOC_OBJ_PU) { + if (hwloc_bitmap_isset(obj->cpuset, obj->os_index)) + hwloc_bitmap_set(topology->levels[0][0]->cpuset, obj->os_index); + hwloc_bitmap_set(topology->levels[0][0]->complete_cpuset, obj->os_index); + } + } + + *current = obj; + obj->parent = parent; + obj->next_sibling = NULL; + topology->modified = 1; +} + +hwloc_obj_t +hwloc_alloc_setup_object(hwloc_topology_t topology, + hwloc_obj_type_t type, unsigned os_index) +{ + struct hwloc_obj *obj = hwloc_tma_malloc(topology->tma, sizeof(*obj)); + memset(obj, 0, sizeof(*obj)); + obj->type = type; + obj->os_index = os_index; + obj->gp_index = topology->next_gp_index++; + obj->attr = hwloc_tma_malloc(topology->tma, sizeof(*obj->attr)); + memset(obj->attr, 0, sizeof(*obj->attr)); + /* do not allocate the cpuset here, let the caller do it */ + return obj; +} + +hwloc_obj_t +hwloc_topology_alloc_group_object(struct hwloc_topology *topology) +{ + if (!topology->is_loaded) { + /* this could actually work, see insert() below */ + errno = EINVAL; + return NULL; + } + return hwloc_alloc_setup_object(topology, HWLOC_OBJ_GROUP, HWLOC_UNKNOWN_INDEX); +} + +static void hwloc_propagate_symmetric_subtree(hwloc_topology_t topology, hwloc_obj_t root); +static void propagate_total_memory(hwloc_obj_t obj); +static void hwloc_set_group_depth(hwloc_topology_t topology); + +hwloc_obj_t +hwloc_topology_insert_group_object(struct hwloc_topology *topology, hwloc_obj_t obj) +{ + hwloc_obj_t res, root; + int cmp; + + if (!topology->is_loaded) { + /* this could actually work, we would just need to disable connect_children/levels below */ + hwloc_free_unlinked_object(obj); + errno = EINVAL; + return NULL; + } + + if (topology->type_filter[HWLOC_OBJ_GROUP] == HWLOC_TYPE_FILTER_KEEP_NONE) { + hwloc_free_unlinked_object(obj); + errno = EINVAL; + return NULL; + } + + root = hwloc_get_root_obj(topology); + if (obj->cpuset) + hwloc_bitmap_and(obj->cpuset, obj->cpuset, root->cpuset); + if (obj->complete_cpuset) + hwloc_bitmap_and(obj->complete_cpuset, obj->complete_cpuset, root->complete_cpuset); + if (obj->nodeset) + hwloc_bitmap_and(obj->nodeset, obj->nodeset, root->nodeset); + if (obj->complete_nodeset) + hwloc_bitmap_and(obj->complete_nodeset, obj->complete_nodeset, root->complete_nodeset); + + if ((!obj->cpuset || hwloc_bitmap_iszero(obj->cpuset)) + && (!obj->complete_cpuset || hwloc_bitmap_iszero(obj->complete_cpuset)) + && (!obj->nodeset || hwloc_bitmap_iszero(obj->nodeset)) + && (!obj->complete_nodeset || hwloc_bitmap_iszero(obj->complete_nodeset))) { + hwloc_free_unlinked_object(obj); + errno = EINVAL; + return NULL; + } + + cmp = hwloc_obj_cmp_sets(obj, root); + if (cmp == HWLOC_OBJ_INCLUDED) { + res = hwloc__insert_object_by_cpuset(topology, NULL, obj, NULL /* do not show errors on stdout */); + } else { + /* just merge root */ + res = root; + } + + if (!res) + return NULL; + if (res != obj) + /* merged */ + return res; + + /* properly inserted */ + hwloc_obj_add_children_sets(obj); + if (hwloc_topology_reconnect(topology, 0) < 0) + return NULL; + + hwloc_propagate_symmetric_subtree(topology, topology->levels[0][0]); + hwloc_set_group_depth(topology); + +#ifndef HWLOC_DEBUG + if (getenv("HWLOC_DEBUG_CHECK")) +#endif + hwloc_topology_check(topology); + + return obj; +} + +hwloc_obj_t +hwloc_topology_insert_misc_object(struct hwloc_topology *topology, hwloc_obj_t parent, const char *name) +{ + hwloc_obj_t obj; + + if (topology->type_filter[HWLOC_OBJ_MISC] == HWLOC_TYPE_FILTER_KEEP_NONE) { + errno = EINVAL; + return NULL; + } + + if (!topology->is_loaded) { + errno = EINVAL; + return NULL; + } + + obj = hwloc_alloc_setup_object(topology, HWLOC_OBJ_MISC, HWLOC_UNKNOWN_INDEX); + if (name) + obj->name = strdup(name); + + hwloc_insert_object_by_parent(topology, parent, obj); + + /* FIXME: only connect misc parent children and misc level, + * but this API is likely not performance critical anyway + */ + hwloc_topology_reconnect(topology, 0); + +#ifndef HWLOC_DEBUG + if (getenv("HWLOC_DEBUG_CHECK")) +#endif + hwloc_topology_check(topology); + + return obj; +} + +/* assuming set is included in the topology complete_cpuset + * and all objects have a proper complete_cpuset, + * return the best one containing set. + * if some object are equivalent (same complete_cpuset), return the highest one. + */ +static hwloc_obj_t +hwloc_get_highest_obj_covering_complete_cpuset (hwloc_topology_t topology, hwloc_const_cpuset_t set) +{ + hwloc_obj_t current = hwloc_get_root_obj(topology); + hwloc_obj_t child; + + if (hwloc_bitmap_isequal(set, current->complete_cpuset)) + /* root cpuset is exactly what we want, no need to look at children, we want the highest */ + return current; + + recurse: + /* find the right child */ + for_each_child(child, current) { + if (hwloc_bitmap_isequal(set, child->complete_cpuset)) + /* child puset is exactly what we want, no need to look at children, we want the highest */ + return child; + if (!hwloc_bitmap_iszero(child->complete_cpuset) && hwloc_bitmap_isincluded(set, child->complete_cpuset)) + break; + } + + if (child) { + current = child; + goto recurse; + } + + /* no better child */ + return current; +} + +hwloc_obj_t +hwloc_find_insert_io_parent_by_complete_cpuset(struct hwloc_topology *topology, hwloc_cpuset_t cpuset) +{ + hwloc_obj_t group_obj, largeparent, parent; + + /* restrict to the existing complete cpuset to avoid errors later */ + hwloc_bitmap_and(cpuset, cpuset, hwloc_topology_get_complete_cpuset(topology)); + if (hwloc_bitmap_iszero(cpuset)) + /* remaining cpuset is empty, invalid */ + return NULL; + + largeparent = hwloc_get_highest_obj_covering_complete_cpuset(topology, cpuset); + if (hwloc_bitmap_isequal(largeparent->complete_cpuset, cpuset) + || !hwloc_filter_check_keep_object_type(topology, HWLOC_OBJ_GROUP)) + /* Found a valid object (normal case) */ + return largeparent; + + /* we need to insert an intermediate group */ + group_obj = hwloc_alloc_setup_object(topology, HWLOC_OBJ_GROUP, HWLOC_UNKNOWN_INDEX); + if (!group_obj) + /* Failed to insert the exact Group, fallback to largeparent */ + return largeparent; + + group_obj->complete_cpuset = hwloc_bitmap_dup(cpuset); + hwloc_bitmap_and(cpuset, cpuset, hwloc_topology_get_topology_cpuset(topology)); + group_obj->cpuset = hwloc_bitmap_dup(cpuset); + group_obj->attr->group.kind = HWLOC_GROUP_KIND_IO; + parent = hwloc__insert_object_by_cpuset(topology, largeparent, group_obj, hwloc_report_os_error); + if (!parent) + /* Failed to insert the Group, maybe a conflicting cpuset */ + return largeparent; + + /* Group couldn't get merged or we would have gotten the right largeparent earlier */ + assert(parent == group_obj); + + /* Group inserted without being merged, everything OK, setup its sets */ + hwloc_obj_add_children_sets(group_obj); + + return parent; +} + +static int hwloc_memory_page_type_compare(const void *_a, const void *_b) +{ + const struct hwloc_memory_page_type_s *a = _a; + const struct hwloc_memory_page_type_s *b = _b; + /* consider 0 as larger so that 0-size page_type go to the end */ + if (!b->size) + return -1; + /* don't cast a-b in int since those are ullongs */ + if (b->size == a->size) + return 0; + return a->size < b->size ? -1 : 1; +} + +/* Propagate memory counts */ +static void +propagate_total_memory(hwloc_obj_t obj) +{ + hwloc_obj_t child; + unsigned i; + + /* reset total before counting local and children memory */ + obj->total_memory = 0; + + /* Propagate memory up. */ + for_each_child(child, obj) { + propagate_total_memory(child); + obj->total_memory += child->total_memory; + } + for_each_memory_child(child, obj) { + propagate_total_memory(child); + obj->total_memory += child->total_memory; + } + /* No memory under I/O or Misc */ + + if (obj->type == HWLOC_OBJ_NUMANODE) { + obj->total_memory += obj->attr->numanode.local_memory; + + /* By the way, sort the page_type array. + * Cannot do it on insert since some backends (e.g. XML) add page_types after inserting the object. + */ + qsort(obj->attr->numanode.page_types, obj->attr->numanode.page_types_len, sizeof(*obj->attr->numanode.page_types), hwloc_memory_page_type_compare); + /* Ignore 0-size page_types, they are at the end */ + for(i=obj->attr->numanode.page_types_len; i>=1; i--) + if (obj->attr->numanode.page_types[i-1].size) + break; + obj->attr->numanode.page_types_len = i; + } +} + +/* Now that root sets are ready, propagate them to children + * by allocating missing sets and restricting existing ones. + */ +static void +fixup_sets(hwloc_obj_t obj) +{ + int in_memory_list; + hwloc_obj_t child; + + child = obj->first_child; + in_memory_list = 0; + /* iterate over normal children first, we'll come back for memory children later */ + + iterate: + while (child) { + /* our cpuset must be included in our parent's one */ + hwloc_bitmap_and(child->cpuset, child->cpuset, obj->cpuset); + hwloc_bitmap_and(child->nodeset, child->nodeset, obj->nodeset); + /* our complete_cpuset must be included in our parent's one, but can be larger than our cpuset */ + if (child->complete_cpuset) { + hwloc_bitmap_and(child->complete_cpuset, child->complete_cpuset, obj->complete_cpuset); + } else { + child->complete_cpuset = hwloc_bitmap_dup(child->cpuset); + } + if (child->complete_nodeset) { + hwloc_bitmap_and(child->complete_nodeset, child->complete_nodeset, obj->complete_nodeset); + } else { + child->complete_nodeset = hwloc_bitmap_dup(child->nodeset); + } + + fixup_sets(child); + child = child->next_sibling; + } + + /* switch to memory children list if any */ + if (!in_memory_list && obj->memory_first_child) { + child = obj->memory_first_child; + in_memory_list = 1; + goto iterate; + } + + /* No sets in I/O or Misc */ +} + +/* Setup object cpusets/nodesets by OR'ing its children. */ +int +hwloc_obj_add_other_obj_sets(hwloc_obj_t dst, hwloc_obj_t src) +{ +#define ADD_OTHER_OBJ_SET(_dst, _src, _set) \ + if ((_src)->_set) { \ + if (!(_dst)->_set) \ + (_dst)->_set = hwloc_bitmap_alloc(); \ + hwloc_bitmap_or((_dst)->_set, (_dst)->_set, (_src)->_set); \ + } + ADD_OTHER_OBJ_SET(dst, src, cpuset); + ADD_OTHER_OBJ_SET(dst, src, complete_cpuset); + ADD_OTHER_OBJ_SET(dst, src, nodeset); + ADD_OTHER_OBJ_SET(dst, src, complete_nodeset); + return 0; +} + +int +hwloc_obj_add_children_sets(hwloc_obj_t obj) +{ + hwloc_obj_t child; + for_each_child(child, obj) { + hwloc_obj_add_other_obj_sets(obj, child); + } + /* No need to look at Misc children, they contain no PU. */ + return 0; +} + +/* CPU objects are inserted by cpusets, we know their cpusets are properly included. + * We just need fixup_sets() to make sure they aren't too wide. + * + * Memory objects are inserted by cpusets to find their CPU parent, + * but nodesets are only used inside the memory hierarchy below that parent. + * Thus we need to propagate nodesets to CPU-side parents and children. + * + * A memory object nodeset consists of NUMA nodes below it. + * A normal object nodeset consists in NUMA nodes attached to any + * of its children or parents. + */ +static void +propagate_nodeset(hwloc_obj_t obj) +{ + hwloc_obj_t child; + + /* Start our nodeset from the parent one. + * It was emptied at root, and it's being filled with local nodes + * in that branch of the tree as we recurse down. + */ + if (!obj->nodeset) + obj->nodeset = hwloc_bitmap_alloc(); + if (obj->parent) + hwloc_bitmap_copy(obj->nodeset, obj->parent->nodeset); + else + hwloc_bitmap_zero(obj->nodeset); + + /* Don't clear complete_nodeset, just make sure it contains nodeset. + * We cannot clear the complete_nodeset at root and rebuild it down because + * some bits may correspond to offline/disallowed NUMA nodes missing in the topology. + */ + if (!obj->complete_nodeset) + obj->complete_nodeset = hwloc_bitmap_dup(obj->nodeset); + else + hwloc_bitmap_or(obj->complete_nodeset, obj->complete_nodeset, obj->nodeset); + + /* now add our local nodeset */ + for_each_memory_child(child, obj) { + /* FIXME rather recurse in the memory hierarchy */ + + /* first, update children complete_nodeset if needed */ + if (!child->complete_nodeset) + child->complete_nodeset = hwloc_bitmap_dup(child->nodeset); + else + hwloc_bitmap_or(child->complete_nodeset, child->complete_nodeset, child->nodeset); + + /* add memory children nodesets to ours */ + hwloc_bitmap_or(obj->nodeset, obj->nodeset, child->nodeset); + hwloc_bitmap_or(obj->complete_nodeset, obj->complete_nodeset, child->complete_nodeset); + + /* by the way, copy our cpusets to memory children */ + if (child->cpuset) + hwloc_bitmap_copy(child->cpuset, obj->cpuset); + else + child->cpuset = hwloc_bitmap_dup(obj->cpuset); + if (child->complete_cpuset) + hwloc_bitmap_copy(child->complete_cpuset, obj->complete_cpuset); + else + child->complete_cpuset = hwloc_bitmap_dup(obj->complete_cpuset); + } + + /* Propagate our nodeset to CPU children. */ + for_each_child(child, obj) { + propagate_nodeset(child); + } + + /* Propagate CPU children specific nodesets back to us. + * + * We cannot merge these two loops because we don't want to first child + * nodeset to be propagated back to us and then down to the second child. + * Each child may have its own local nodeset, + * each of them is propagated to us, but not to other children. + */ + for_each_child(child, obj) { + hwloc_bitmap_or(obj->nodeset, obj->nodeset, child->nodeset); + hwloc_bitmap_or(obj->complete_nodeset, obj->complete_nodeset, child->complete_nodeset); + } + + /* No nodeset under I/O or Misc */ + +} + +static void +remove_unused_sets(hwloc_topology_t topology, hwloc_obj_t obj) +{ + hwloc_obj_t child; + + hwloc_bitmap_and(obj->cpuset, obj->cpuset, topology->allowed_cpuset); + hwloc_bitmap_and(obj->nodeset, obj->nodeset, topology->allowed_nodeset); + + for_each_child(child, obj) + remove_unused_sets(topology, child); + for_each_memory_child(child, obj) + remove_unused_sets(topology, child); + /* No cpuset under I/O or Misc */ +} + +static void +hwloc__filter_bridges(hwloc_topology_t topology, hwloc_obj_t root, unsigned depth) +{ + hwloc_obj_t child, *pchild; + + /* filter I/O children and recurse */ + for_each_io_child_safe(child, root, pchild) { + enum hwloc_type_filter_e filter = topology->type_filter[child->type]; + + /* recurse into grand-children */ + hwloc__filter_bridges(topology, child, depth+1); + + child->attr->bridge.depth = depth; + + if (child->type == HWLOC_OBJ_BRIDGE + && filter == HWLOC_TYPE_FILTER_KEEP_IMPORTANT + && !child->io_first_child) { + unlink_and_free_single_object(pchild); + topology->modified = 1; + } + } +} + +static void +hwloc_filter_bridges(hwloc_topology_t topology, hwloc_obj_t parent) +{ + hwloc_obj_t child = parent->first_child; + while (child) { + hwloc_filter_bridges(topology, child); + child = child->next_sibling; + } + + hwloc__filter_bridges(topology, parent, 0); +} + +void +hwloc__reorder_children(hwloc_obj_t parent) +{ + /* move the children list on the side */ + hwloc_obj_t *prev, child, children = parent->first_child; + parent->first_child = NULL; + while (children) { + /* dequeue child */ + child = children; + children = child->next_sibling; + /* find where to enqueue it */ + prev = &parent->first_child; + while (*prev && hwloc__object_cpusets_compare_first(child, *prev) > 0) + prev = &((*prev)->next_sibling); + /* enqueue */ + child->next_sibling = *prev; + *prev = child; + } + /* No ordering to enforce for Misc or I/O children. */ +} + +/* Remove all normal children whose cpuset is empty, + * and memory children whose nodeset is empty. + * Also don't remove objects that have I/O children, but ignore Misc. + */ +static void +remove_empty(hwloc_topology_t topology, hwloc_obj_t *pobj) +{ + hwloc_obj_t obj = *pobj, child, *pchild; + + for_each_child_safe(child, obj, pchild) + remove_empty(topology, pchild); + for_each_memory_child_safe(child, obj, pchild) + remove_empty(topology, pchild); + /* No cpuset under I/O or Misc */ + + if (obj->first_child /* only remove if all children were removed above, so that we don't remove parents of NUMAnode */ + || obj->memory_first_child /* only remove if no memory attached there */ + || obj->io_first_child /* only remove if no I/O is attached there */) + /* ignore Misc */ + return; + + if (hwloc__obj_type_is_normal(obj->type)) { + if (!hwloc_bitmap_iszero(obj->cpuset)) + return; + } else { + assert(hwloc__obj_type_is_memory(obj->type)); + if (!hwloc_bitmap_iszero(obj->nodeset)) + return; + } + + hwloc_debug("%s", "\nRemoving empty object "); + hwloc_debug_print_object(0, obj); + unlink_and_free_single_object(pobj); + topology->modified = 1; +} + +/* reset type depth before modifying levels (either reconnecting or filtering/keep_structure) */ +static void +hwloc_reset_normal_type_depths(hwloc_topology_t topology) +{ + unsigned i; + for (i=HWLOC_OBJ_TYPE_MIN; i<=HWLOC_OBJ_GROUP; i++) + topology->type_depth[i] = HWLOC_TYPE_DEPTH_UNKNOWN; + /* type contiguity is asserted in topology_check() */ +} + +static int +hwloc_dont_merge_group_level(hwloc_topology_t topology, unsigned i) +{ + unsigned j; + + /* Don't merge some groups in that level? */ + for(j=0; jlevel_nbobjects[i]; j++) + if (topology->levels[i][j]->attr->group.dont_merge) + return 1; + + return 0; +} + +/* compare i-th and i-1-th levels structure */ +static int +hwloc_compare_levels_structure(hwloc_topology_t topology, unsigned i) +{ + int checkmemory = (topology->levels[i][0]->type == HWLOC_OBJ_PU); + unsigned j; + + if (topology->level_nbobjects[i-1] != topology->level_nbobjects[i]) + return -1; + + for(j=0; jlevel_nbobjects[i]; j++) { + if (topology->levels[i-1][j]->arity != 1) + return -1; + if (checkmemory && topology->levels[i-1][j]->memory_arity) + /* don't merge PUs if there's memory above */ + return -1; + } + /* same number of objects with arity 1 above, no problem */ + return 0; +} + +/* return > 0 if any level was removed, which means reconnect is needed */ +static void +hwloc_filter_levels_keep_structure(hwloc_topology_t topology) +{ + unsigned i, j; + int res = 0; + + /* start from the bottom since we'll remove intermediate levels */ + for(i=topology->nb_levels-1; i>0; i--) { + int replacechild = 0, replaceparent = 0; + hwloc_obj_t obj1 = topology->levels[i-1][0]; + hwloc_obj_t obj2 = topology->levels[i][0]; + hwloc_obj_type_t type1 = obj1->type; + hwloc_obj_type_t type2 = obj2->type; + + /* Check whether parents and/or children can be replaced */ + if (topology->type_filter[type1] == HWLOC_TYPE_FILTER_KEEP_STRUCTURE) { + /* Parents can be ignored in favor of children. */ + replaceparent = 1; + if (type1 == HWLOC_OBJ_GROUP && hwloc_dont_merge_group_level(topology, i-1)) + replaceparent = 0; + } + if (topology->type_filter[type2] == HWLOC_TYPE_FILTER_KEEP_STRUCTURE) { + /* Children can be ignored in favor of parents. */ + replacechild = 1; + if (type1 == HWLOC_OBJ_GROUP && hwloc_dont_merge_group_level(topology, i)) + replacechild = 0; + } + if (!replacechild && !replaceparent) + /* no ignoring */ + continue; + /* Decide which one to actually replace */ + if (replaceparent && replacechild) { + /* If both may be replaced, look at obj_type_priority */ + if (obj_type_priority[type1] >= obj_type_priority[type2]) + replaceparent = 0; + else + replacechild = 0; + } + /* Are these levels actually identical? */ + if (hwloc_compare_levels_structure(topology, i) < 0) + continue; + hwloc_debug("may merge levels #%u=%s and #%u=%s\n", + i-1, hwloc_obj_type_string(type1), i, hwloc_obj_type_string(type2)); + + /* OK, remove intermediate objects from the tree. */ + for(j=0; jlevel_nbobjects[i]; j++) { + hwloc_obj_t parent = topology->levels[i-1][j]; + hwloc_obj_t child = topology->levels[i][j]; + unsigned k; + if (replacechild) { + /* move child's children to parent */ + parent->first_child = child->first_child; + parent->last_child = child->last_child; + parent->arity = child->arity; + free(parent->children); + parent->children = child->children; + child->children = NULL; + /* update children parent */ + for(k=0; karity; k++) + parent->children[k]->parent = parent; + /* append child memory/io/misc children to parent */ + if (child->memory_first_child) { + append_siblings_list(&parent->memory_first_child, child->memory_first_child, parent); + parent->memory_arity += child->memory_arity; + } + if (child->io_first_child) { + append_siblings_list(&parent->io_first_child, child->io_first_child, parent); + parent->io_arity += child->io_arity; + } + if (child->misc_first_child) { + append_siblings_list(&parent->misc_first_child, child->misc_first_child, parent); + parent->misc_arity += child->misc_arity; + } + hwloc_free_unlinked_object(child); + } else { + /* replace parent with child in grand-parent */ + if (parent->parent) { + parent->parent->children[parent->sibling_rank] = child; + child->sibling_rank = parent->sibling_rank; + if (!parent->sibling_rank) { + parent->parent->first_child = child; + /* child->prev_sibling was already NULL, child was single */ + } else { + child->prev_sibling = parent->parent->children[parent->sibling_rank-1]; + child->prev_sibling->next_sibling = child; + } + if (parent->sibling_rank == parent->parent->arity-1) { + parent->parent->last_child = child; + /* child->next_sibling was already NULL, child was single */ + } else { + child->next_sibling = parent->parent->children[parent->sibling_rank+1]; + child->next_sibling->prev_sibling = child; + } + /* update child parent */ + child->parent = parent->parent; + } else { + /* make child the new root */ + topology->levels[0][0] = child; + child->parent = NULL; + } + /* prepend parent memory/io/misc children to child */ + if (parent->memory_first_child) { + prepend_siblings_list(&child->memory_first_child, parent->memory_first_child, child); + child->memory_arity += parent->memory_arity; + } + if (parent->io_first_child) { + prepend_siblings_list(&child->io_first_child, parent->io_first_child, child); + child->io_arity += parent->io_arity; + } + if (parent->misc_first_child) { + prepend_siblings_list(&child->misc_first_child, parent->misc_first_child, child); + child->misc_arity += parent->misc_arity; + } + hwloc_free_unlinked_object(parent); + /* prev/next_sibling will be updated below in another loop */ + } + } + if (replaceparent && i>1) { + /* Update sibling list within modified parent->parent arrays */ + for(j=0; jlevel_nbobjects[i]; j++) { + hwloc_obj_t child = topology->levels[i][j]; + unsigned rank = child->sibling_rank; + child->prev_sibling = rank > 0 ? child->parent->children[rank-1] : NULL; + child->next_sibling = rank < child->parent->arity-1 ? child->parent->children[rank+1] : NULL; + } + } + + /* Update levels so that the next reconnect isn't confused */ + if (replaceparent) { + /* Removing level i-1, so move levels [i..nb_levels-1] to [i-1..] */ + free(topology->levels[i-1]); + memmove(&topology->levels[i-1], + &topology->levels[i], + (topology->nb_levels-i)*sizeof(topology->levels[i])); + memmove(&topology->level_nbobjects[i-1], + &topology->level_nbobjects[i], + (topology->nb_levels-i)*sizeof(topology->level_nbobjects[i])); + hwloc_debug("removed parent level %s at depth %u\n", + hwloc_obj_type_string(type1), i-1); + } else { + /* Removing level i, so move levels [i+1..nb_levels-1] and later to [i..] */ + free(topology->levels[i]); + memmove(&topology->levels[i], + &topology->levels[i+1], + (topology->nb_levels-1-i)*sizeof(topology->levels[i])); + memmove(&topology->level_nbobjects[i], + &topology->level_nbobjects[i+1], + (topology->nb_levels-1-i)*sizeof(topology->level_nbobjects[i])); + hwloc_debug("removed child level %s at depth %u\n", + hwloc_obj_type_string(type2), i); + } + topology->level_nbobjects[topology->nb_levels-1] = 0; + topology->levels[topology->nb_levels-1] = NULL; + topology->nb_levels--; + + res++; + } + + if (res > 0) { + /* Update object and type depths if some levels were removed */ + hwloc_reset_normal_type_depths(topology); + for(i=0; inb_levels; i++) { + hwloc_obj_type_t type = topology->levels[i][0]->type; + for(j=0; jlevel_nbobjects[i]; j++) + topology->levels[i][j]->depth = (int)i; + if (topology->type_depth[type] == HWLOC_TYPE_DEPTH_UNKNOWN) + topology->type_depth[type] = (int)i; + else + topology->type_depth[type] = HWLOC_TYPE_DEPTH_MULTIPLE; + } + } +} + +static void +hwloc_propagate_symmetric_subtree(hwloc_topology_t topology, hwloc_obj_t root) +{ + hwloc_obj_t child; + unsigned arity = root->arity; + int ok; + + /* assume we're not symmetric by default */ + root->symmetric_subtree = 0; + + /* if no child, we are symmetric */ + if (!arity) + goto good; + + /* FIXME ignore memory just like I/O and Misc? */ + + /* look at normal children only, I/O and Misc are ignored. + * return if any child is not symmetric. + */ + ok = 1; + for_each_child(child, root) { + hwloc_propagate_symmetric_subtree(topology, child); + if (!child->symmetric_subtree) + ok = 0; + } + if (!ok) + return; + /* Misc and I/O children do not care about symmetric_subtree */ + + /* if single child is symmetric, we're good */ + if (arity == 1) + goto good; + + /* now check that children subtrees are identical. + * just walk down the first child in each tree and compare their depth and arities + */ +{ + HWLOC_VLA(hwloc_obj_t, array, arity); + memcpy(array, root->children, arity * sizeof(*array)); + while (1) { + unsigned i; + /* check current level arities and depth */ + for(i=1; idepth != array[0]->depth + || array[i]->arity != array[0]->arity) { + return; + } + if (!array[0]->arity) + /* no more children level, we're ok */ + break; + /* look at first child of each element now */ + for(i=0; ifirst_child; + } +} + + /* everything went fine, we're symmetric */ + good: + root->symmetric_subtree = 1; +} + +static void hwloc_set_group_depth(hwloc_topology_t topology) +{ + unsigned groupdepth = 0; + unsigned i, j; + for(i=0; inb_levels; i++) + if (topology->levels[i][0]->type == HWLOC_OBJ_GROUP) { + for (j = 0; j < topology->level_nbobjects[i]; j++) + topology->levels[i][j]->attr->group.depth = groupdepth; + groupdepth++; + } +} + +/* + * Initialize handy pointers in the whole topology. + * The topology only had first_child and next_sibling pointers. + * When this funtions return, all parent/children pointers are initialized. + * The remaining fields (levels, cousins, logical_index, depth, ...) will + * be setup later in hwloc_connect_levels(). + * + * Can be called several times, so may have to update the array. + */ +static void +hwloc_connect_children(hwloc_obj_t parent) +{ + unsigned n, oldn = parent->arity; + hwloc_obj_t child, prev_child; + int ok; + + /* Main children list */ + + ok = 1; + prev_child = NULL; + for (n = 0, child = parent->first_child; + child; + n++, prev_child = child, child = child->next_sibling) { + child->sibling_rank = n; + child->prev_sibling = prev_child; + /* already OK in the array? */ + if (n >= oldn || parent->children[n] != child) + ok = 0; + /* recurse */ + hwloc_connect_children(child); + } + parent->last_child = prev_child; + parent->arity = n; + if (!n) { + /* no need for an array anymore */ + free(parent->children); + parent->children = NULL; + goto memory; + } + if (ok) + /* array is already OK (even if too large) */ + goto memory; + + /* alloc a larger array if needed */ + if (oldn < n) { + free(parent->children); + parent->children = malloc(n * sizeof(*parent->children)); + } + /* refill */ + for (n = 0, child = parent->first_child; + child; + n++, child = child->next_sibling) { + parent->children[n] = child; + } + + + + memory: + /* Memory children list */ + + prev_child = NULL; + for (n = 0, child = parent->memory_first_child; + child; + n++, prev_child = child, child = child->next_sibling) { + child->parent = parent; + child->sibling_rank = n; + child->prev_sibling = prev_child; + hwloc_connect_children(child); + } + parent->memory_arity = n; + + /* I/O children list */ + + prev_child = NULL; + for (n = 0, child = parent->io_first_child; + child; + n++, prev_child = child, child = child->next_sibling) { + child->parent = parent; + child->sibling_rank = n; + child->prev_sibling = prev_child; + hwloc_connect_children(child); + } + parent->io_arity = n; + + /* Misc children list */ + + prev_child = NULL; + for (n = 0, child = parent->misc_first_child; + child; + n++, prev_child = child, child = child->next_sibling) { + child->parent = parent; + child->sibling_rank = n; + child->prev_sibling = prev_child; + hwloc_connect_children(child); + } + parent->misc_arity = n; +} + +/* + * Check whether there is an object below ROOT that has the same type as OBJ + */ +static int +find_same_type(hwloc_obj_t root, hwloc_obj_t obj) +{ + hwloc_obj_t child; + + if (hwloc_type_cmp(root, obj) == HWLOC_OBJ_EQUAL) + return 1; + + for_each_child (child, root) + if (find_same_type(child, obj)) + return 1; + + return 0; +} + +/* traverse the array of current object and compare them with top_obj. + * if equal, take the object and put its children into the remaining objs. + * if not equal, put the object into the remaining objs. + */ +static unsigned +hwloc_level_take_objects(hwloc_obj_t top_obj, + hwloc_obj_t *current_objs, unsigned n_current_objs, + hwloc_obj_t *taken_objs, unsigned n_taken_objs __hwloc_attribute_unused, + hwloc_obj_t *remaining_objs, unsigned n_remaining_objs __hwloc_attribute_unused) +{ + unsigned taken_i = 0; + unsigned new_i = 0; + unsigned i, j; + + for (i = 0; i < n_current_objs; i++) + if (hwloc_type_cmp(top_obj, current_objs[i]) == HWLOC_OBJ_EQUAL) { + /* Take it, add main children. */ + taken_objs[taken_i++] = current_objs[i]; + for (j = 0; j < current_objs[i]->arity; j++) + remaining_objs[new_i++] = current_objs[i]->children[j]; + } else { + /* Leave it. */ + remaining_objs[new_i++] = current_objs[i]; + } + +#ifdef HWLOC_DEBUG + /* Make sure we didn't mess up. */ + assert(taken_i == n_taken_objs); + assert(new_i == n_current_objs - n_taken_objs + n_remaining_objs); +#endif + + return new_i; +} + +static int +hwloc_build_level_from_list(struct hwloc_special_level_s *slevel) +{ + unsigned i, nb; + struct hwloc_obj * obj; + + /* count */ + obj = slevel->first; + i = 0; + while (obj) { + i++; + obj = obj->next_cousin; + } + nb = i; + + if (nb) { + /* allocate and fill level */ + slevel->objs = malloc(nb * sizeof(struct hwloc_obj *)); + obj = slevel->first; + i = 0; + while (obj) { + obj->logical_index = i; + slevel->objs[i] = obj; + i++; + obj = obj->next_cousin; + } + } + + slevel->nbobjs = nb; + return 0; +} + +static void +hwloc_append_special_object(struct hwloc_special_level_s *level, hwloc_obj_t obj) +{ + if (level->first) { + obj->prev_cousin = level->last; + obj->prev_cousin->next_cousin = obj; + level->last = obj; + } else { + obj->prev_cousin = NULL; + level->first = level->last = obj; + } +} + +/* Append special objects to their lists */ +static void +hwloc_list_special_objects(hwloc_topology_t topology, hwloc_obj_t obj) +{ + hwloc_obj_t child; + + if (obj->type == HWLOC_OBJ_NUMANODE) { + obj->next_cousin = NULL; + obj->depth = HWLOC_TYPE_DEPTH_NUMANODE; + /* Insert the main NUMA node list */ + hwloc_append_special_object(&topology->slevels[HWLOC_SLEVEL_NUMANODE], obj); + + /* Recurse */ + for_each_memory_child(child, obj) + hwloc_list_special_objects(topology, child); + for_each_misc_child(child, obj) + hwloc_list_special_objects(topology, child); + + } else if (obj->type == HWLOC_OBJ_MISC) { + obj->next_cousin = NULL; + obj->depth = HWLOC_TYPE_DEPTH_MISC; + /* Insert the main Misc list */ + hwloc_append_special_object(&topology->slevels[HWLOC_SLEVEL_MISC], obj); + /* Recurse, Misc only have Misc children */ + for_each_misc_child(child, obj) + hwloc_list_special_objects(topology, child); + + } else if (hwloc__obj_type_is_io(obj->type)) { + obj->next_cousin = NULL; + + if (obj->type == HWLOC_OBJ_BRIDGE) { + obj->depth = HWLOC_TYPE_DEPTH_BRIDGE; + /* Insert in the main bridge list */ + hwloc_append_special_object(&topology->slevels[HWLOC_SLEVEL_BRIDGE], obj); + + } else if (obj->type == HWLOC_OBJ_PCI_DEVICE) { + obj->depth = HWLOC_TYPE_DEPTH_PCI_DEVICE; + /* Insert in the main pcidev list */ + hwloc_append_special_object(&topology->slevels[HWLOC_SLEVEL_PCIDEV], obj); + + } else if (obj->type == HWLOC_OBJ_OS_DEVICE) { + obj->depth = HWLOC_TYPE_DEPTH_OS_DEVICE; + /* Insert in the main osdev list */ + hwloc_append_special_object(&topology->slevels[HWLOC_SLEVEL_OSDEV], obj); + } + /* Recurse, I/O only have I/O and Misc children */ + for_each_io_child(child, obj) + hwloc_list_special_objects(topology, child); + for_each_misc_child(child, obj) + hwloc_list_special_objects(topology, child); + + } else { + /* Recurse */ + for_each_child(child, obj) + hwloc_list_special_objects(topology, child); + for_each_memory_child(child, obj) + hwloc_list_special_objects(topology, child); + for_each_io_child(child, obj) + hwloc_list_special_objects(topology, child); + for_each_misc_child(child, obj) + hwloc_list_special_objects(topology, child); + } +} + +/* Build I/O levels */ +static void +hwloc_connect_io_misc_levels(hwloc_topology_t topology) +{ + unsigned i; + + for(i=0; islevels[i].objs); + memset(&topology->slevels, 0, sizeof(topology->slevels)); + + hwloc_list_special_objects(topology, topology->levels[0][0]); + + for(i=0; islevels[i]); +} + +/* + * Do the remaining work that hwloc_connect_children() did not do earlier. + * Requires object arity and children list to be properly initialized (by hwloc_connect_children()). + */ +static int +hwloc_connect_levels(hwloc_topology_t topology) +{ + unsigned l, i=0; + hwloc_obj_t *objs, *taken_objs, *new_objs, top_obj, root; + unsigned n_objs, n_taken_objs, n_new_objs; + + /* reset non-root levels (root was initialized during init and will not change here) */ + for(l=1; lnb_levels; l++) + free(topology->levels[l]); + memset(topology->levels+1, 0, (topology->nb_levels-1)*sizeof(*topology->levels)); + memset(topology->level_nbobjects+1, 0, (topology->nb_levels-1)*sizeof(*topology->level_nbobjects)); + topology->nb_levels = 1; + + /* initialize all non-IO/non-Misc depths to unknown */ + hwloc_reset_normal_type_depths(topology); + + /* initialize root type depth */ + root = topology->levels[0][0]; + root->depth = 0; + topology->type_depth[root->type] = 0; + /* root level */ + root->logical_index = 0; + root->prev_cousin = NULL; + root->next_cousin = NULL; + /* root as a child of nothing */ + root->parent = NULL; + root->sibling_rank = 0; + root->prev_sibling = NULL; + root->next_sibling = NULL; + + /* Start with children of the whole system. */ + n_objs = topology->levels[0][0]->arity; + objs = malloc(n_objs * sizeof(objs[0])); + if (!objs) { + errno = ENOMEM; + return -1; + } + memcpy(objs, topology->levels[0][0]->children, n_objs*sizeof(objs[0])); + + /* Keep building levels while there are objects left in OBJS. */ + while (n_objs) { + /* At this point, the objs array contains only objects that may go into levels */ + + /* First find which type of object is the topmost. + * Don't use PU if there are other types since we want to keep PU at the bottom. + */ + + /* Look for the first non-PU object, and use the first PU if we really find nothing else */ + for (i = 0; i < n_objs; i++) + if (objs[i]->type != HWLOC_OBJ_PU) + break; + top_obj = i == n_objs ? objs[0] : objs[i]; + + /* See if this is actually the topmost object */ + for (i = 0; i < n_objs; i++) { + if (hwloc_type_cmp(top_obj, objs[i]) != HWLOC_OBJ_EQUAL) { + if (find_same_type(objs[i], top_obj)) { + /* OBJS[i] is strictly above an object of the same type as TOP_OBJ, so it + * is above TOP_OBJ. */ + top_obj = objs[i]; + } + } + } + + /* Now peek all objects of the same type, build a level with that and + * replace them with their children. */ + + /* First count them. */ + n_taken_objs = 0; + n_new_objs = 0; + for (i = 0; i < n_objs; i++) + if (hwloc_type_cmp(top_obj, objs[i]) == HWLOC_OBJ_EQUAL) { + n_taken_objs++; + n_new_objs += objs[i]->arity; + } + + /* New level. */ + taken_objs = malloc((n_taken_objs + 1) * sizeof(taken_objs[0])); + /* New list of pending objects. */ + if (n_objs - n_taken_objs + n_new_objs) { + new_objs = malloc((n_objs - n_taken_objs + n_new_objs) * sizeof(new_objs[0])); + } else { +#ifdef HWLOC_DEBUG + assert(!n_new_objs); + assert(n_objs == n_taken_objs); +#endif + new_objs = NULL; + } + + n_new_objs = hwloc_level_take_objects(top_obj, + objs, n_objs, + taken_objs, n_taken_objs, + new_objs, n_new_objs); + + /* Ok, put numbers in the level and link cousins. */ + for (i = 0; i < n_taken_objs; i++) { + taken_objs[i]->depth = (int) topology->nb_levels; + taken_objs[i]->logical_index = i; + if (i) { + taken_objs[i]->prev_cousin = taken_objs[i-1]; + taken_objs[i-1]->next_cousin = taken_objs[i]; + } + } + taken_objs[0]->prev_cousin = NULL; + taken_objs[n_taken_objs-1]->next_cousin = NULL; + + /* One more level! */ + hwloc_debug("--- %s level", hwloc_obj_type_string(top_obj->type)); + hwloc_debug(" has number %u\n\n", topology->nb_levels); + + if (topology->type_depth[top_obj->type] == HWLOC_TYPE_DEPTH_UNKNOWN) + topology->type_depth[top_obj->type] = (int) topology->nb_levels; + else + topology->type_depth[top_obj->type] = HWLOC_TYPE_DEPTH_MULTIPLE; /* mark as unknown */ + + taken_objs[n_taken_objs] = NULL; + + if (topology->nb_levels == topology->nb_levels_allocated) { + /* extend the arrays of levels */ + void *tmplevels, *tmpnbobjs; + tmplevels = realloc(topology->levels, + 2 * topology->nb_levels_allocated * sizeof(*topology->levels)); + tmpnbobjs = realloc(topology->level_nbobjects, + 2 * topology->nb_levels_allocated * sizeof(*topology->level_nbobjects)); + if (!tmplevels || !tmpnbobjs) { + fprintf(stderr, "hwloc failed to realloc level arrays to %u\n", topology->nb_levels_allocated * 2); + + /* if one realloc succeeded, make sure the caller will free the new buffer */ + if (tmplevels) + topology->levels = tmplevels; + if (tmpnbobjs) + topology->level_nbobjects = tmpnbobjs; + /* the realloc that failed left topology->level_foo untouched, will be freed by the caller */ + + free(objs); + free(taken_objs); + free(new_objs); + errno = ENOMEM; + return -1; + } + topology->levels = tmplevels; + topology->level_nbobjects = tmpnbobjs; + memset(topology->levels + topology->nb_levels_allocated, + 0, topology->nb_levels_allocated * sizeof(*topology->levels)); + memset(topology->level_nbobjects + topology->nb_levels_allocated, + 0, topology->nb_levels_allocated * sizeof(*topology->level_nbobjects)); + topology->nb_levels_allocated *= 2; + } + /* add the new level */ + topology->level_nbobjects[topology->nb_levels] = n_taken_objs; + topology->levels[topology->nb_levels] = taken_objs; + + topology->nb_levels++; + + free(objs); + + /* Switch to new_objs */ + objs = new_objs; + n_objs = n_new_objs; + } + + /* It's empty now. */ + free(objs); + + return 0; +} + +int +hwloc_topology_reconnect(struct hwloc_topology *topology, unsigned long flags) +{ + if (flags) { + errno = EINVAL; + return -1; + } + if (!topology->modified) + return 0; + + hwloc_connect_children(topology->levels[0][0]); + + if (hwloc_connect_levels(topology) < 0) + return -1; + + hwloc_connect_io_misc_levels(topology); + + topology->modified = 0; + + return 0; +} + +void hwloc_alloc_root_sets(hwloc_obj_t root) +{ + /* + * All sets are initially NULL. + * + * At least one backend should call this function to initialize all sets at once. + * XML uses it lazily in case only some sets were given in the XML import. + * + * Other backends can check root->cpuset != NULL to see if somebody + * discovered things before them. + */ + if (!root->cpuset) + root->cpuset = hwloc_bitmap_alloc(); + if (!root->complete_cpuset) + root->complete_cpuset = hwloc_bitmap_alloc(); + if (!root->nodeset) + root->nodeset = hwloc_bitmap_alloc(); + if (!root->complete_nodeset) + root->complete_nodeset = hwloc_bitmap_alloc(); +} + +/* Main discovery loop */ +static int +hwloc_discover(struct hwloc_topology *topology) +{ + struct hwloc_backend *backend; + + topology->modified = 0; /* no need to reconnect yet */ + + topology->allowed_cpuset = hwloc_bitmap_alloc_full(); + topology->allowed_nodeset = hwloc_bitmap_alloc_full(); + + /* discover() callbacks should use hwloc_insert to add objects initialized + * through hwloc_alloc_setup_object. + * For node levels, nodeset and memory must be initialized. + * For cache levels, memory and type/depth must be initialized. + * For group levels, depth must be initialized. + */ + + /* There must be at least a PU object for each logical processor, at worse + * produced by hwloc_setup_pu_level() + */ + + /* To be able to just use hwloc_insert_object_by_cpuset to insert the object + * in the topology according to the cpuset, the cpuset field must be + * initialized. + */ + + /* A priori, All processors are visible in the topology, and allowed + * for the application. + * + * - If some processors exist but topology information is unknown for them + * (and thus the backend couldn't create objects for them), they should be + * added to the complete_cpuset field of the lowest object where the object + * could reside. + * + * - If some processors are not allowed for the application (e.g. for + * administration reasons), they should be dropped from the allowed_cpuset + * field. + * + * The same applies to the node sets complete_nodeset and allowed_cpuset. + * + * If such field doesn't exist yet, it can be allocated, and initialized to + * zero (for complete), or to full (for allowed). The values are + * automatically propagated to the whole tree after detection. + */ + + /* + * Discover CPUs first + */ + backend = topology->backends; + while (NULL != backend) { + if (backend->component->type != HWLOC_DISC_COMPONENT_TYPE_CPU + && backend->component->type != HWLOC_DISC_COMPONENT_TYPE_GLOBAL) + /* not yet */ + goto next_cpubackend; + if (!backend->discover) + goto next_cpubackend; + backend->discover(backend); + hwloc_debug_print_objects(0, topology->levels[0][0]); + +next_cpubackend: + backend = backend->next; + } + + /* One backend should have called hwloc_alloc_root_sets() + * and set bits during PU and NUMA insert. + */ + if (!topology->levels[0][0]->cpuset || hwloc_bitmap_iszero(topology->levels[0][0]->cpuset)) { + hwloc_debug("%s", "No PU added by any CPU and global backend\n"); + errno = EINVAL; + return -1; + } + + if (topology->binding_hooks.get_allowed_resources && topology->is_thissystem) { + const char *env = getenv("HWLOC_THISSYSTEM_ALLOWED_RESOURCES"); + if ((env && atoi(env)) + || (topology->flags & HWLOC_TOPOLOGY_FLAG_THISSYSTEM_ALLOWED_RESOURCES)) + topology->binding_hooks.get_allowed_resources(topology); + } + + /* If there's no NUMA node, add one with all the memory. + * root->complete_nodeset wouldn't be empty if any NUMA was ever added: + * - insert_by_cpuset() adds bits whe PU/NUMA are added. + * - XML takes care of sanitizing nodesets. + */ + if (hwloc_bitmap_iszero(topology->levels[0][0]->complete_nodeset)) { + hwloc_obj_t node; + hwloc_debug("%s", "\nAdd missing single NUMA node\n"); + node = hwloc_alloc_setup_object(topology, HWLOC_OBJ_NUMANODE, 0); + node->cpuset = hwloc_bitmap_dup(topology->levels[0][0]->cpuset); + node->nodeset = hwloc_bitmap_alloc(); + /* other nodesets will be filled below */ + hwloc_bitmap_set(node->nodeset, 0); + memcpy(&node->attr->numanode, &topology->machine_memory, sizeof(topology->machine_memory)); + memset(&topology->machine_memory, 0, sizeof(topology->machine_memory)); + hwloc_insert_object_by_cpuset(topology, node); + } else { + /* if we're sure we found all NUMA nodes without their sizes (x86 backend?), + * we could split topology->total_memory in all of them. + */ + free(topology->machine_memory.page_types); + memset(&topology->machine_memory, 0, sizeof(topology->machine_memory)); + } + + hwloc_debug("%s", "\nFixup root sets\n"); + hwloc_bitmap_and(topology->levels[0][0]->cpuset, topology->levels[0][0]->cpuset, topology->levels[0][0]->complete_cpuset); + hwloc_bitmap_and(topology->levels[0][0]->nodeset, topology->levels[0][0]->nodeset, topology->levels[0][0]->complete_nodeset); + + hwloc_bitmap_and(topology->allowed_cpuset, topology->allowed_cpuset, topology->levels[0][0]->cpuset); + hwloc_bitmap_and(topology->allowed_nodeset, topology->allowed_nodeset, topology->levels[0][0]->nodeset); + + hwloc_debug("%s", "\nPropagate sets\n"); + /* cpuset are already there thanks to the _by_cpuset insertion, + * but nodeset have to be propagated below and above NUMA nodes + */ + propagate_nodeset(topology->levels[0][0]); + /* now fixup parent/children sets */ + fixup_sets(topology->levels[0][0]); + + hwloc_debug_print_objects(0, topology->levels[0][0]); + + if (!(topology->flags & HWLOC_TOPOLOGY_FLAG_WHOLE_SYSTEM)) { + hwloc_debug("%s", "\nRemoving unauthorized sets from all sets\n"); + remove_unused_sets(topology, topology->levels[0][0]); + hwloc_debug_print_objects(0, topology->levels[0][0]); + } + + /* see if we should ignore the root now that we know how many children it has */ + if (!hwloc_filter_check_keep_object(topology, topology->levels[0][0]) + && topology->levels[0][0]->first_child && !topology->levels[0][0]->first_child->next_sibling) { + hwloc_obj_t oldroot = topology->levels[0][0]; + hwloc_obj_t newroot = oldroot->first_child; + /* switch to the new root */ + newroot->parent = NULL; + topology->levels[0][0] = newroot; + /* move oldroot memory/io/misc children before newroot children */ + if (oldroot->memory_first_child) + prepend_siblings_list(&newroot->memory_first_child, oldroot->memory_first_child, newroot); + if (oldroot->io_first_child) + prepend_siblings_list(&newroot->io_first_child, oldroot->io_first_child, newroot); + if (oldroot->misc_first_child) + prepend_siblings_list(&newroot->misc_first_child, oldroot->misc_first_child, newroot); + /* destroy oldroot and use the new one */ + hwloc_free_unlinked_object(oldroot); + } + + /* + * All object cpusets and nodesets are properly set now. + */ + + /* Now connect handy pointers to make remaining discovery easier. */ + hwloc_debug("%s", "\nOk, finished tweaking, now connect\n"); + if (hwloc_topology_reconnect(topology, 0) < 0) + return -1; + hwloc_debug_print_objects(0, topology->levels[0][0]); + + /* + * Additional discovery with other backends + */ + + backend = topology->backends; + while (NULL != backend) { + if (backend->component->type == HWLOC_DISC_COMPONENT_TYPE_CPU + || backend->component->type == HWLOC_DISC_COMPONENT_TYPE_GLOBAL) + /* already done above */ + goto next_noncpubackend; + if (!backend->discover) + goto next_noncpubackend; + backend->discover(backend); + hwloc_debug_print_objects(0, topology->levels[0][0]); + +next_noncpubackend: + backend = backend->next; + } + + hwloc_pci_belowroot_apply_locality(topology); + + hwloc_debug("%s", "\nNow reconnecting\n"); + hwloc_debug_print_objects(0, topology->levels[0][0]); + + /* Remove some stuff */ + + hwloc_debug("%s", "\nRemoving bridge objects if needed\n"); + hwloc_filter_bridges(topology, topology->levels[0][0]); + hwloc_debug_print_objects(0, topology->levels[0][0]); + + hwloc_debug("%s", "\nRemoving empty objects\n"); + remove_empty(topology, &topology->levels[0][0]); + if (!topology->levels[0][0]) { + fprintf(stderr, "Topology became empty, aborting!\n"); + return -1; + } + if (hwloc_bitmap_iszero(topology->levels[0][0]->cpuset)) { + fprintf(stderr, "Topology does not contain any PU, aborting!\n"); + return -1; + } + if (hwloc_bitmap_iszero(topology->levels[0][0]->nodeset)) { + fprintf(stderr, "Topology does not contain any NUMA node, aborting!\n"); + return -1; + } + hwloc_debug_print_objects(0, topology->levels[0][0]); + + /* Reconnect things after all these changes. + * Often needed because of Groups inserted for I/Os. + * And required for KEEP_STRUCTURE below. + */ + if (hwloc_topology_reconnect(topology, 0) < 0) + return -1; + + hwloc_debug("%s", "\nRemoving levels with HWLOC_TYPE_FILTER_KEEP_STRUCTURE\n"); + hwloc_filter_levels_keep_structure(topology); + hwloc_debug_print_objects(0, topology->levels[0][0]); + + /* accumulate children memory in total_memory fields (only once parent is set) */ + hwloc_debug("%s", "\nPropagate total memory up\n"); + propagate_total_memory(topology->levels[0][0]); + + /* setup the symmetric_subtree attribute */ + hwloc_propagate_symmetric_subtree(topology, topology->levels[0][0]); + + /* apply group depths */ + hwloc_set_group_depth(topology); + + /* add some identification attributes if not loading from XML */ + if (topology->backends + && strcmp(topology->backends->component->name, "xml")) { + char *value; + /* add a hwlocVersion */ + hwloc_obj_add_info(topology->levels[0][0], "hwlocVersion", HWLOC_VERSION); + /* add a ProcessName */ + value = hwloc_progname(topology); + if (value) { + hwloc_obj_add_info(topology->levels[0][0], "ProcessName", value); + free(value); + } + } + + return 0; +} + +/* To be called before discovery is actually launched, + * Resets everything in case a previous load initialized some stuff. + */ +void +hwloc_topology_setup_defaults(struct hwloc_topology *topology) +{ + struct hwloc_obj *root_obj; + + /* reset support */ + memset(&topology->binding_hooks, 0, sizeof(topology->binding_hooks)); + memset(topology->support.discovery, 0, sizeof(*topology->support.discovery)); + memset(topology->support.cpubind, 0, sizeof(*topology->support.cpubind)); + memset(topology->support.membind, 0, sizeof(*topology->support.membind)); + + /* Only the System object on top by default */ + topology->next_gp_index = 1; /* keep 0 as an invalid value */ + topology->nb_levels = 1; /* there's at least SYSTEM */ + topology->levels[0] = hwloc_tma_malloc (topology->tma, sizeof (hwloc_obj_t)); + topology->level_nbobjects[0] = 1; + + /* Machine-wide memory */ + topology->machine_memory.local_memory = 0; + topology->machine_memory.page_types_len = 0; + topology->machine_memory.page_types = NULL; + + /* Allowed stuff */ + topology->allowed_cpuset = NULL; + topology->allowed_nodeset = NULL; + + /* NULLify other special levels */ + memset(&topology->slevels, 0, sizeof(topology->slevels)); + /* assert the indexes of special levels */ + HWLOC_BUILD_ASSERT(HWLOC_SLEVEL_NUMANODE == HWLOC_SLEVEL_FROM_DEPTH(HWLOC_TYPE_DEPTH_NUMANODE)); + HWLOC_BUILD_ASSERT(HWLOC_SLEVEL_MISC == HWLOC_SLEVEL_FROM_DEPTH(HWLOC_TYPE_DEPTH_MISC)); + HWLOC_BUILD_ASSERT(HWLOC_SLEVEL_BRIDGE == HWLOC_SLEVEL_FROM_DEPTH(HWLOC_TYPE_DEPTH_BRIDGE)); + HWLOC_BUILD_ASSERT(HWLOC_SLEVEL_PCIDEV == HWLOC_SLEVEL_FROM_DEPTH(HWLOC_TYPE_DEPTH_PCI_DEVICE)); + HWLOC_BUILD_ASSERT(HWLOC_SLEVEL_OSDEV == HWLOC_SLEVEL_FROM_DEPTH(HWLOC_TYPE_DEPTH_OS_DEVICE)); + + /* sane values to type_depth */ + hwloc_reset_normal_type_depths(topology); + topology->type_depth[HWLOC_OBJ_NUMANODE] = HWLOC_TYPE_DEPTH_NUMANODE; + topology->type_depth[HWLOC_OBJ_MISC] = HWLOC_TYPE_DEPTH_MISC; + topology->type_depth[HWLOC_OBJ_BRIDGE] = HWLOC_TYPE_DEPTH_BRIDGE; + topology->type_depth[HWLOC_OBJ_PCI_DEVICE] = HWLOC_TYPE_DEPTH_PCI_DEVICE; + topology->type_depth[HWLOC_OBJ_OS_DEVICE] = HWLOC_TYPE_DEPTH_OS_DEVICE; + + /* Create the actual machine object, but don't touch its attributes yet + * since the OS backend may still change the object into something else + * (for instance System) + */ + root_obj = hwloc_alloc_setup_object(topology, HWLOC_OBJ_MACHINE, 0); + topology->levels[0][0] = root_obj; +} + +static void hwloc__topology_filter_init(struct hwloc_topology *topology); + +/* This function may use a tma, it cannot free() or realloc() */ +static int +hwloc__topology_init (struct hwloc_topology **topologyp, + unsigned nblevels, + struct hwloc_tma *tma) +{ + struct hwloc_topology *topology; + + topology = hwloc_tma_malloc (tma, sizeof (struct hwloc_topology)); + if(!topology) + return -1; + + topology->tma = tma; + + hwloc_components_init(); /* uses malloc without tma, but won't need it since dup() caller already took a reference */ + hwloc_backends_init(topology); + hwloc_pci_discovery_init(topology); /* make sure both dup() and load() get sane variables */ + + /* Setup topology context */ + topology->is_loaded = 0; + topology->flags = 0; + topology->is_thissystem = 1; + topology->pid = 0; + topology->userdata = NULL; + topology->topology_abi = HWLOC_TOPOLOGY_ABI; + topology->adopted_shmem_addr = NULL; + topology->adopted_shmem_length = 0; + + topology->support.discovery = hwloc_tma_malloc(tma, sizeof(*topology->support.discovery)); + topology->support.cpubind = hwloc_tma_malloc(tma, sizeof(*topology->support.cpubind)); + topology->support.membind = hwloc_tma_malloc(tma, sizeof(*topology->support.membind)); + + topology->nb_levels_allocated = nblevels; /* enough for default 9 levels = Mach+Pack+NUMA+L3+L2+L1d+L1i+Co+PU */ + topology->levels = hwloc_tma_calloc(tma, topology->nb_levels_allocated * sizeof(*topology->levels)); + topology->level_nbobjects = hwloc_tma_calloc(tma, topology->nb_levels_allocated * sizeof(*topology->level_nbobjects)); + + hwloc__topology_filter_init(topology); + + hwloc_internal_distances_init(topology); + + topology->userdata_export_cb = NULL; + topology->userdata_import_cb = NULL; + topology->userdata_not_decoded = 0; + + /* Make the topology look like something coherent but empty */ + hwloc_topology_setup_defaults(topology); + + *topologyp = topology; + return 0; +} + +int +hwloc_topology_init (struct hwloc_topology **topologyp) +{ + return hwloc__topology_init(topologyp, + 16, /* 16 is enough for default 9 levels = Mach+Pack+NUMA+L3+L2+L1d+L1i+Co+PU */ + NULL); /* no TMA for normal topologies, too many allocations to fix */ +} + +int +hwloc_topology_set_pid(struct hwloc_topology *topology __hwloc_attribute_unused, + hwloc_pid_t pid __hwloc_attribute_unused) +{ + if (topology->is_loaded) { + errno = EBUSY; + return -1; + } + + /* this does *not* change the backend */ +#ifdef HWLOC_LINUX_SYS + topology->pid = pid; + return 0; +#else /* HWLOC_LINUX_SYS */ + errno = ENOSYS; + return -1; +#endif /* HWLOC_LINUX_SYS */ +} + +int +hwloc_topology_set_synthetic(struct hwloc_topology *topology, const char *description) +{ + if (topology->is_loaded) { + errno = EBUSY; + return -1; + } + + return hwloc_disc_component_force_enable(topology, + 0 /* api */, + -1, "synthetic", + description, NULL, NULL); +} + +int +hwloc_topology_set_xml(struct hwloc_topology *topology, + const char *xmlpath) +{ + if (topology->is_loaded) { + errno = EBUSY; + return -1; + } + + return hwloc_disc_component_force_enable(topology, + 0 /* api */, + -1, "xml", + xmlpath, NULL, NULL); +} + +int +hwloc_topology_set_xmlbuffer(struct hwloc_topology *topology, + const char *xmlbuffer, + int size) +{ + if (topology->is_loaded) { + errno = EBUSY; + return -1; + } + + return hwloc_disc_component_force_enable(topology, + 0 /* api */, + -1, "xml", NULL, + xmlbuffer, (void*) (uintptr_t) size); +} + +int +hwloc_topology_set_flags (struct hwloc_topology *topology, unsigned long flags) +{ + if (topology->is_loaded) { + /* actually harmless */ + errno = EBUSY; + return -1; + } + + if (flags & ~(HWLOC_TOPOLOGY_FLAG_WHOLE_SYSTEM|HWLOC_TOPOLOGY_FLAG_IS_THISSYSTEM|HWLOC_TOPOLOGY_FLAG_THISSYSTEM_ALLOWED_RESOURCES)) { + errno = EINVAL; + return -1; + } + + topology->flags = flags; + return 0; +} + +unsigned long +hwloc_topology_get_flags (struct hwloc_topology *topology) +{ + return topology->flags; +} + +static void +hwloc__topology_filter_init(struct hwloc_topology *topology) +{ + hwloc_obj_type_t type; + /* Only ignore useless cruft by default */ + for(type = HWLOC_OBJ_TYPE_MIN; type < HWLOC_OBJ_TYPE_MAX; type++) + topology->type_filter[type] = HWLOC_TYPE_FILTER_KEEP_ALL; + topology->type_filter[HWLOC_OBJ_L1ICACHE] = HWLOC_TYPE_FILTER_KEEP_NONE; + topology->type_filter[HWLOC_OBJ_L2ICACHE] = HWLOC_TYPE_FILTER_KEEP_NONE; + topology->type_filter[HWLOC_OBJ_L3ICACHE] = HWLOC_TYPE_FILTER_KEEP_NONE; + topology->type_filter[HWLOC_OBJ_GROUP] = HWLOC_TYPE_FILTER_KEEP_STRUCTURE; + topology->type_filter[HWLOC_OBJ_MISC] = HWLOC_TYPE_FILTER_KEEP_NONE; + topology->type_filter[HWLOC_OBJ_BRIDGE] = HWLOC_TYPE_FILTER_KEEP_NONE; + topology->type_filter[HWLOC_OBJ_PCI_DEVICE] = HWLOC_TYPE_FILTER_KEEP_NONE; + topology->type_filter[HWLOC_OBJ_OS_DEVICE] = HWLOC_TYPE_FILTER_KEEP_NONE; +} + +static int +hwloc__topology_set_type_filter(struct hwloc_topology *topology, hwloc_obj_type_t type, enum hwloc_type_filter_e filter) +{ + if (type == HWLOC_OBJ_PU || type == HWLOC_OBJ_NUMANODE || type == HWLOC_OBJ_MACHINE) { + if (filter != HWLOC_TYPE_FILTER_KEEP_ALL) { + /* we need the Machine, PU and NUMA levels */ + errno = EINVAL; + return -1; + } + } else if (hwloc__obj_type_is_special(type)) { + if (filter == HWLOC_TYPE_FILTER_KEEP_STRUCTURE) { + /* I/O and Misc are outside of the main topology structure, makes no sense. */ + errno = EINVAL; + return -1; + } + } else if (type == HWLOC_OBJ_GROUP) { + if (filter == HWLOC_TYPE_FILTER_KEEP_ALL) { + /* Groups are always ignored, at least keep_structure */ + errno = EINVAL; + return -1; + } + } + + /* "important" just means "all" for non-I/O non-Misc */ + if (!hwloc__obj_type_is_special(type) && filter == HWLOC_TYPE_FILTER_KEEP_IMPORTANT) + filter = HWLOC_TYPE_FILTER_KEEP_ALL; + + topology->type_filter[type] = filter; + return 0; +} + +int +hwloc_topology_set_type_filter(struct hwloc_topology *topology, hwloc_obj_type_t type, enum hwloc_type_filter_e filter) +{ + HWLOC_BUILD_ASSERT(HWLOC_OBJ_TYPE_MIN == 0); + if ((unsigned) type >= HWLOC_OBJ_TYPE_MAX) { + errno = EINVAL; + return -1; + } + if (topology->is_loaded) { + errno = EBUSY; + return -1; + } + return hwloc__topology_set_type_filter(topology, type, filter); +} + +int +hwloc_topology_set_all_types_filter(struct hwloc_topology *topology, enum hwloc_type_filter_e filter) +{ + hwloc_obj_type_t type; + if (topology->is_loaded) { + errno = EBUSY; + return -1; + } + for(type = HWLOC_OBJ_TYPE_MIN; type < HWLOC_OBJ_TYPE_MAX; type++) + hwloc__topology_set_type_filter(topology, type, filter); + return 0; +} + +int +hwloc_topology_set_cache_types_filter(hwloc_topology_t topology, enum hwloc_type_filter_e filter) +{ + unsigned i; + for(i=HWLOC_OBJ_L1CACHE; i= HWLOC_OBJ_TYPE_MAX) { + errno = EINVAL; + return -1; + } + *filterp = topology->type_filter[type]; + return 0; +} + +void +hwloc_topology_clear (struct hwloc_topology *topology) +{ + /* no need to set to NULL after free() since callers will call setup_defaults() or just destroy the rest of the topology */ + unsigned l; + hwloc_internal_distances_destroy(topology); + hwloc_free_object_and_children(topology->levels[0][0]); + hwloc_bitmap_free(topology->allowed_cpuset); + hwloc_bitmap_free(topology->allowed_nodeset); + for (l=0; lnb_levels; l++) + free(topology->levels[l]); + for(l=0; lslevels[l].objs); + free(topology->machine_memory.page_types); +} + +void +hwloc_topology_destroy (struct hwloc_topology *topology) +{ + if (topology->adopted_shmem_addr) { + hwloc__topology_disadopt(topology); + return; + } + + hwloc_backends_disable_all(topology); + hwloc_components_fini(); + + hwloc_topology_clear(topology); + + free(topology->levels); + free(topology->level_nbobjects); + + free(topology->support.discovery); + free(topology->support.cpubind); + free(topology->support.membind); + free(topology); +} + +int +hwloc_topology_load (struct hwloc_topology *topology) +{ + int err; + + if (topology->is_loaded) { + errno = EBUSY; + return -1; + } + + hwloc_internal_distances_prepare(topology); + + if (getenv("HWLOC_XML_USERDATA_NOT_DECODED")) + topology->userdata_not_decoded = 1; + + /* Ignore variables if HWLOC_COMPONENTS is set. It will be processed later */ + if (!getenv("HWLOC_COMPONENTS")) { + /* Only apply variables if we have not changed the backend yet. + * Only the first one will be kept. + * Check for FSROOT first since it's for debugging so likely needs to override everything else. + * Check for XML last (that's the one that may be set system-wide by administrators) + * so that it's only used if other variables are not set, + * to allow users to override easily. + */ + if (!topology->backends) { + const char *fsroot_path_env = getenv("HWLOC_FSROOT"); + if (fsroot_path_env) + hwloc_disc_component_force_enable(topology, + 1 /* env force */, + HWLOC_DISC_COMPONENT_TYPE_CPU, "linux", + NULL /* backend will getenv again */, NULL, NULL); + } + if (!topology->backends) { + const char *cpuid_path_env = getenv("HWLOC_CPUID_PATH"); + if (cpuid_path_env) + hwloc_disc_component_force_enable(topology, + 1 /* env force */, + HWLOC_DISC_COMPONENT_TYPE_CPU, "x86", + NULL /* backend will getenv again */, NULL, NULL); + } + if (!topology->backends) { + const char *synthetic_env = getenv("HWLOC_SYNTHETIC"); + if (synthetic_env) + hwloc_disc_component_force_enable(topology, + 1 /* env force */, + -1, "synthetic", + synthetic_env, NULL, NULL); + } + if (!topology->backends) { + const char *xmlpath_env = getenv("HWLOC_XMLFILE"); + if (xmlpath_env) + hwloc_disc_component_force_enable(topology, + 1 /* env force */, + -1, "xml", + xmlpath_env, NULL, NULL); + } + } + + /* instantiate all possible other backends now */ + hwloc_disc_components_enable_others(topology); + /* now that backends are enabled, update the thissystem flag and some callbacks */ + hwloc_backends_is_thissystem(topology); + hwloc_backends_find_callbacks(topology); + /* + * Now set binding hooks according to topology->is_thissystem + * and what the native OS backend offers. + */ + hwloc_set_binding_hooks(topology); + + hwloc_pci_discovery_prepare(topology); + + /* actual topology discovery */ + err = hwloc_discover(topology); + if (err < 0) + goto out; + + hwloc_pci_discovery_exit(topology); + +#ifndef HWLOC_DEBUG + if (getenv("HWLOC_DEBUG_CHECK")) +#endif + hwloc_topology_check(topology); + + /* Mark distances objs arrays as invalid since we may have removed objects + * from the topology after adding the distances (remove_empty, etc). + * It would be hard to actually verify whether it's needed. + */ + hwloc_internal_distances_invalidate_cached_objs(topology); + /* And refresh distances so that multithreaded concurrent distances_get() + * don't refresh() concurrently (disallowed). + */ + hwloc_internal_distances_refresh(topology); + + topology->is_loaded = 1; + return 0; + + out: + hwloc_pci_discovery_exit(topology); + hwloc_topology_clear(topology); + hwloc_topology_setup_defaults(topology); + hwloc_backends_disable_all(topology); + return -1; +} + +/* adjust object cpusets according the given droppedcpuset, + * drop object whose cpuset becomes empty and that have no children, + * and propagate NUMA node removal as nodeset changes in parents. + */ +static void +restrict_object_by_cpuset(hwloc_topology_t topology, unsigned long flags, hwloc_obj_t *pobj, + hwloc_bitmap_t droppedcpuset, hwloc_bitmap_t droppednodeset) +{ + hwloc_obj_t obj = *pobj, child, *pchild; + int modified = 0; + + if (hwloc_bitmap_intersects(obj->complete_cpuset, droppedcpuset)) { + hwloc_bitmap_andnot(obj->cpuset, obj->cpuset, droppedcpuset); + hwloc_bitmap_andnot(obj->complete_cpuset, obj->complete_cpuset, droppedcpuset); + modified = 1; + } else { + if ((flags & HWLOC_RESTRICT_FLAG_REMOVE_CPULESS) + && hwloc_bitmap_iszero(obj->complete_cpuset)) { + /* we're empty, there's a NUMAnode below us, it'll be removed this time */ + modified = 1; + } + /* nodeset cannot intersect unless cpuset intersects or is empty */ + if (droppednodeset) + assert(!hwloc_bitmap_intersects(obj->complete_nodeset, droppednodeset) + || hwloc_bitmap_iszero(obj->complete_cpuset)); + } + if (droppednodeset) { + hwloc_bitmap_andnot(obj->nodeset, obj->nodeset, droppednodeset); + hwloc_bitmap_andnot(obj->complete_nodeset, obj->complete_nodeset, droppednodeset); + } + + if (modified) { + for_each_child_safe(child, obj, pchild) + restrict_object_by_cpuset(topology, flags, pchild, droppedcpuset, droppednodeset); + /* if some hwloc_bitmap_first(child->complete_cpuset) changed, children might need to be reordered */ + hwloc__reorder_children(obj); + + for_each_memory_child_safe(child, obj, pchild) + restrict_object_by_cpuset(topology, flags, pchild, droppedcpuset, droppednodeset); + /* local NUMA nodes have the same cpusets, no need to reorder them */ + + /* Nothing to restrict under I/O or Misc */ + } + + if (!obj->first_child && !obj->memory_first_child /* arity not updated before connect_children() */ + && hwloc_bitmap_iszero(obj->cpuset) + && (obj->type != HWLOC_OBJ_NUMANODE || (flags & HWLOC_RESTRICT_FLAG_REMOVE_CPULESS))) { + /* remove object */ + hwloc_debug("%s", "\nRemoving object during restrict"); + hwloc_debug_print_object(0, obj); + + if (!(flags & HWLOC_RESTRICT_FLAG_ADAPT_IO)) { + hwloc_free_object_siblings_and_children(obj->io_first_child); + obj->io_first_child = NULL; + } + if (!(flags & HWLOC_RESTRICT_FLAG_ADAPT_MISC)) { + hwloc_free_object_siblings_and_children(obj->misc_first_child); + obj->misc_first_child = NULL; + } + assert(!obj->first_child); + assert(!obj->memory_first_child); + unlink_and_free_single_object(pobj); + topology->modified = 1; + } +} + +int +hwloc_topology_restrict(struct hwloc_topology *topology, hwloc_const_cpuset_t cpuset, unsigned long flags) +{ + hwloc_bitmap_t droppedcpuset, droppednodeset; + + if (!topology->is_loaded) { + errno = EINVAL; + return -1; + } + + if (flags & ~(HWLOC_RESTRICT_FLAG_REMOVE_CPULESS + |HWLOC_RESTRICT_FLAG_ADAPT_MISC|HWLOC_RESTRICT_FLAG_ADAPT_IO)) { + errno = EINVAL; + return -1; + } + + /* make sure we'll keep something in the topology */ + if (!hwloc_bitmap_intersects(cpuset, topology->allowed_cpuset)) { + errno = EINVAL; /* easy failure, just don't touch the topology */ + return -1; + } + + droppedcpuset = hwloc_bitmap_alloc(); + droppednodeset = hwloc_bitmap_alloc(); + if (!droppedcpuset || !droppednodeset) { + hwloc_bitmap_free(droppedcpuset); + hwloc_bitmap_free(droppednodeset); + return -1; + } + + /* cpuset to clear */ + hwloc_bitmap_not(droppedcpuset, cpuset); + /* nodeset to clear */ + if (flags & HWLOC_RESTRICT_FLAG_REMOVE_CPULESS) { + hwloc_obj_t node = hwloc_get_obj_by_type(topology, HWLOC_OBJ_NUMANODE, 0); + do { + /* node will be removed if nodeset gets or was empty */ + if (hwloc_bitmap_iszero(node->cpuset) + || hwloc_bitmap_isincluded(node->cpuset, droppedcpuset)) + hwloc_bitmap_set(droppednodeset, node->os_index); + node = node->next_cousin; + } while (node); + + /* check we're not removing all NUMA nodes */ + if (hwloc_bitmap_isincluded(topology->allowed_nodeset, droppednodeset)) { + errno = EINVAL; /* easy failure, just don't touch the topology */ + hwloc_bitmap_free(droppedcpuset); + hwloc_bitmap_free(droppednodeset); + return -1; + } + } + /* remove nodeset if empty */ + if (!(flags & HWLOC_RESTRICT_FLAG_REMOVE_CPULESS) + || hwloc_bitmap_iszero(droppednodeset)) { + hwloc_bitmap_free(droppednodeset); + droppednodeset = NULL; + } + + /* now recurse to filter sets and drop things */ + restrict_object_by_cpuset(topology, flags, &topology->levels[0][0], droppedcpuset, droppednodeset); + hwloc_bitmap_andnot(topology->allowed_cpuset, topology->allowed_cpuset, droppedcpuset); + if (droppednodeset) + hwloc_bitmap_andnot(topology->allowed_nodeset, topology->allowed_nodeset, droppednodeset); + + hwloc_bitmap_free(droppedcpuset); + hwloc_bitmap_free(droppednodeset); + + if (hwloc_topology_reconnect(topology, 0) < 0) + goto out; + + /* some objects may have disappeared, we need to update distances objs arrays */ + hwloc_internal_distances_invalidate_cached_objs(topology); + + hwloc_filter_levels_keep_structure(topology); + hwloc_propagate_symmetric_subtree(topology, topology->levels[0][0]); + propagate_total_memory(topology->levels[0][0]); + +#ifndef HWLOC_DEBUG + if (getenv("HWLOC_DEBUG_CHECK")) +#endif + hwloc_topology_check(topology); + + return 0; + + out: + /* unrecoverable failure, re-init the topology */ + hwloc_topology_clear(topology); + hwloc_topology_setup_defaults(topology); + return -1; +} + +int +hwloc_topology_is_thissystem(struct hwloc_topology *topology) +{ + return topology->is_thissystem; +} + +int +hwloc_topology_get_depth(struct hwloc_topology *topology) +{ + return (int) topology->nb_levels; +} + +const struct hwloc_topology_support * +hwloc_topology_get_support(struct hwloc_topology * topology) +{ + return &topology->support; +} + +void hwloc_topology_set_userdata(struct hwloc_topology * topology, const void *userdata) +{ + topology->userdata = (void *) userdata; +} + +void * hwloc_topology_get_userdata(struct hwloc_topology * topology) +{ + return topology->userdata; +} + +hwloc_const_cpuset_t +hwloc_topology_get_complete_cpuset(hwloc_topology_t topology) +{ + return hwloc_get_root_obj(topology)->complete_cpuset; +} + +hwloc_const_cpuset_t +hwloc_topology_get_topology_cpuset(hwloc_topology_t topology) +{ + return hwloc_get_root_obj(topology)->cpuset; +} + +hwloc_const_cpuset_t +hwloc_topology_get_allowed_cpuset(hwloc_topology_t topology) +{ + return topology->allowed_cpuset; +} + +hwloc_const_nodeset_t +hwloc_topology_get_complete_nodeset(hwloc_topology_t topology) +{ + return hwloc_get_root_obj(topology)->complete_nodeset; +} + +hwloc_const_nodeset_t +hwloc_topology_get_topology_nodeset(hwloc_topology_t topology) +{ + return hwloc_get_root_obj(topology)->nodeset; +} + +hwloc_const_nodeset_t +hwloc_topology_get_allowed_nodeset(hwloc_topology_t topology) +{ + return topology->allowed_nodeset; +} + + +/**************** + * Debug Checks * + ****************/ + +#ifndef NDEBUG /* assert only enabled if !NDEBUG */ + +static void +hwloc__check_child_siblings(hwloc_obj_t parent, hwloc_obj_t *array, + unsigned arity, unsigned i, + hwloc_obj_t child, hwloc_obj_t prev) +{ + assert(child->parent == parent); + + assert(child->sibling_rank == i); + if (array) + assert(child == array[i]); + + if (prev) + assert(prev->next_sibling == child); + assert(child->prev_sibling == prev); + + if (!i) + assert(child->prev_sibling == NULL); + else + assert(child->prev_sibling != NULL); + + if (i == arity-1) + assert(child->next_sibling == NULL); + else + assert(child->next_sibling != NULL); +} + +static void +hwloc__check_object(hwloc_topology_t topology, hwloc_bitmap_t gp_indexes, hwloc_obj_t obj); + +/* check children between a parent object */ +static void +hwloc__check_normal_children(hwloc_topology_t topology, hwloc_bitmap_t gp_indexes, hwloc_obj_t parent) +{ + hwloc_obj_t child, prev; + unsigned j; + + if (!parent->arity) { + /* check whether that parent has no children for real */ + assert(!parent->children); + assert(!parent->first_child); + assert(!parent->last_child); + return; + } + /* check whether that parent has children for real */ + assert(parent->children); + assert(parent->first_child); + assert(parent->last_child); + + /* sibling checks */ + for(prev = NULL, child = parent->first_child, j = 0; + child; + prev = child, child = child->next_sibling, j++) { + /* normal child */ + assert(hwloc__obj_type_is_normal(child->type)); + /* check depth */ + assert(child->depth > parent->depth); + /* check siblings */ + hwloc__check_child_siblings(parent, parent->children, parent->arity, j, child, prev); + /* recurse */ + hwloc__check_object(topology, gp_indexes, child); + } + /* check arity */ + assert(j == parent->arity); + + assert(parent->first_child == parent->children[0]); + assert(parent->last_child == parent->children[parent->arity-1]); + + /* no normal children below a PU */ + if (parent->type == HWLOC_OBJ_PU) + assert(!parent->arity); +} + +static void +hwloc__check_children_cpusets(hwloc_topology_t topology __hwloc_attribute_unused, hwloc_obj_t obj) +{ + /* we already checked in the caller that objects have either all sets or none */ + hwloc_obj_t child; + int prev_first, prev_empty; + + if (obj->type == HWLOC_OBJ_PU) { + /* PU cpuset is just itself, with no normal children */ + assert(hwloc_bitmap_weight(obj->cpuset) == 1); + assert(hwloc_bitmap_first(obj->cpuset) == (int) obj->os_index); + assert(hwloc_bitmap_weight(obj->complete_cpuset) == 1); + assert(hwloc_bitmap_first(obj->complete_cpuset) == (int) obj->os_index); + if (!(topology->flags & HWLOC_TOPOLOGY_FLAG_WHOLE_SYSTEM)) { + assert(hwloc_bitmap_isset(topology->allowed_cpuset, (int) obj->os_index)); + } + assert(!obj->arity); + } else if (hwloc__obj_type_is_memory(obj->type)) { + /* memory object cpuset is equal to its parent */ + assert(hwloc_bitmap_isequal(obj->parent->cpuset, obj->cpuset)); + assert(!obj->arity); + } else if (!hwloc__obj_type_is_special(obj->type)) { + hwloc_bitmap_t set; + /* other obj cpuset is an exclusive OR of normal children, except for PUs */ + set = hwloc_bitmap_alloc(); + for_each_child(child, obj) { + assert(!hwloc_bitmap_intersects(set, child->cpuset)); + hwloc_bitmap_or(set, set, child->cpuset); + } + assert(hwloc_bitmap_isequal(set, obj->cpuset)); + hwloc_bitmap_free(set); + } + + /* check that memory children have same cpuset */ + for_each_memory_child(child, obj) + assert(hwloc_bitmap_isequal(obj->cpuset, child->cpuset)); + + /* check that children complete_cpusets are properly ordered, empty ones may be anywhere + * (can be wrong for main cpuset since removed PUs can break the ordering). + */ + prev_first = -1; /* -1 works fine with first comparisons below */ + prev_empty = 0; /* no empty cpuset in previous children */ + for_each_child(child, obj) { + int first = hwloc_bitmap_first(child->complete_cpuset); + if (first >= 0) { + assert(!prev_empty); /* no objects with CPU after objects without CPU */ + assert(prev_first < first); + } else { + prev_empty = 1; + } + prev_first = first; + } +} + +static void +hwloc__check_memory_children(hwloc_topology_t topology, hwloc_bitmap_t gp_indexes, hwloc_obj_t parent) +{ + unsigned j; + hwloc_obj_t child, prev; + + if (!parent->memory_arity) { + /* check whether that parent has no children for real */ + assert(!parent->memory_first_child); + return; + } + /* check whether that parent has children for real */ + assert(parent->memory_first_child); + + for(prev = NULL, child = parent->memory_first_child, j = 0; + child; + prev = child, child = child->next_sibling, j++) { + assert(hwloc__obj_type_is_memory(child->type)); + /* check siblings */ + hwloc__check_child_siblings(parent, NULL, parent->memory_arity, j, child, prev); + /* only Memory and Misc children, recurse */ + assert(!child->first_child); + assert(!child->io_first_child); + hwloc__check_object(topology, gp_indexes, child); + } + /* check arity */ + assert(j == parent->memory_arity); + + /* no memory children below a NUMA node */ + if (parent->type == HWLOC_OBJ_NUMANODE) + assert(!parent->memory_arity); +} + +static void +hwloc__check_io_children(hwloc_topology_t topology, hwloc_bitmap_t gp_indexes, hwloc_obj_t parent) +{ + unsigned j; + hwloc_obj_t child, prev; + + if (!parent->io_arity) { + /* check whether that parent has no children for real */ + assert(!parent->io_first_child); + return; + } + /* check whether that parent has children for real */ + assert(parent->io_first_child); + + for(prev = NULL, child = parent->io_first_child, j = 0; + child; + prev = child, child = child->next_sibling, j++) { + /* all children must be I/O */ + assert(hwloc__obj_type_is_io(child->type)); + /* check siblings */ + hwloc__check_child_siblings(parent, NULL, parent->io_arity, j, child, prev); + /* only I/O and Misc children, recurse */ + assert(!child->first_child); + assert(!child->memory_first_child); + hwloc__check_object(topology, gp_indexes, child); + } + /* check arity */ + assert(j == parent->io_arity); +} + +static void +hwloc__check_misc_children(hwloc_topology_t topology, hwloc_bitmap_t gp_indexes, hwloc_obj_t parent) +{ + unsigned j; + hwloc_obj_t child, prev; + + if (!parent->misc_arity) { + /* check whether that parent has no children for real */ + assert(!parent->misc_first_child); + return; + } + /* check whether that parent has children for real */ + assert(parent->misc_first_child); + + for(prev = NULL, child = parent->misc_first_child, j = 0; + child; + prev = child, child = child->next_sibling, j++) { + /* all children must be Misc */ + assert(child->type == HWLOC_OBJ_MISC); + /* check siblings */ + hwloc__check_child_siblings(parent, NULL, parent->misc_arity, j, child, prev); + /* only Misc children, recurse */ + assert(!child->first_child); + assert(!child->memory_first_child); + assert(!child->io_first_child); + hwloc__check_object(topology, gp_indexes, child); + } + /* check arity */ + assert(j == parent->misc_arity); +} + +static void +hwloc__check_object(hwloc_topology_t topology, hwloc_bitmap_t gp_indexes, hwloc_obj_t obj) +{ + assert(!hwloc_bitmap_isset(gp_indexes, obj->gp_index)); + hwloc_bitmap_set(gp_indexes, obj->gp_index); + + HWLOC_BUILD_ASSERT(HWLOC_OBJ_TYPE_MIN == 0); + assert((unsigned) obj->type < HWLOC_OBJ_TYPE_MAX); + + assert(hwloc_filter_check_keep_object(topology, obj)); + + /* check that sets and depth */ + if (hwloc__obj_type_is_special(obj->type)) { + assert(!obj->cpuset); + if (obj->type == HWLOC_OBJ_BRIDGE) + assert(obj->depth == HWLOC_TYPE_DEPTH_BRIDGE); + else if (obj->type == HWLOC_OBJ_PCI_DEVICE) + assert(obj->depth == HWLOC_TYPE_DEPTH_PCI_DEVICE); + else if (obj->type == HWLOC_OBJ_OS_DEVICE) + assert(obj->depth == HWLOC_TYPE_DEPTH_OS_DEVICE); + else if (obj->type == HWLOC_OBJ_MISC) + assert(obj->depth == HWLOC_TYPE_DEPTH_MISC); + } else { + assert(obj->cpuset); + if (obj->type == HWLOC_OBJ_NUMANODE) + assert(obj->depth == HWLOC_TYPE_DEPTH_NUMANODE); + else + assert(obj->depth >= 0); + } + + /* group depth cannot be -1 anymore in v2.0+ */ + if (obj->type == HWLOC_OBJ_GROUP) { + assert(obj->attr->group.depth != (unsigned) -1); + } + + /* there's other cpusets and nodesets if and only if there's a main cpuset */ + assert(!!obj->cpuset == !!obj->complete_cpuset); + assert(!!obj->cpuset == !!obj->nodeset); + assert(!!obj->nodeset == !!obj->complete_nodeset); + + /* check that complete/inline sets are larger than the main sets */ + if (obj->cpuset) { + assert(hwloc_bitmap_isincluded(obj->cpuset, obj->complete_cpuset)); + assert(hwloc_bitmap_isincluded(obj->nodeset, obj->complete_nodeset)); + } + + /* check cache type/depth vs type */ + if (hwloc__obj_type_is_cache(obj->type)) { + if (hwloc__obj_type_is_icache(obj->type)) + assert(obj->attr->cache.type == HWLOC_OBJ_CACHE_INSTRUCTION); + else if (hwloc__obj_type_is_dcache(obj->type)) + assert(obj->attr->cache.type == HWLOC_OBJ_CACHE_DATA + || obj->attr->cache.type == HWLOC_OBJ_CACHE_UNIFIED); + else + assert(0); + assert(hwloc_cache_type_by_depth_type(obj->attr->cache.depth, obj->attr->cache.type) == obj->type); + } + + /* check children */ + hwloc__check_normal_children(topology, gp_indexes, obj); + hwloc__check_memory_children(topology, gp_indexes, obj); + hwloc__check_io_children(topology, gp_indexes, obj); + hwloc__check_misc_children(topology, gp_indexes, obj); + hwloc__check_children_cpusets(topology, obj); + /* nodesets are checked during another recursion with state below */ +} + +static void +hwloc__check_nodesets(hwloc_topology_t topology, hwloc_obj_t obj, hwloc_bitmap_t parentset) +{ + hwloc_obj_t child; + int prev_first; + + if (obj->type == HWLOC_OBJ_NUMANODE) { + /* NUMANODE nodeset is just itself, with no memory/normal children */ + assert(hwloc_bitmap_weight(obj->nodeset) == 1); + assert(hwloc_bitmap_first(obj->nodeset) == (int) obj->os_index); + assert(hwloc_bitmap_weight(obj->complete_nodeset) == 1); + assert(hwloc_bitmap_first(obj->complete_nodeset) == (int) obj->os_index); + if (!(topology->flags & HWLOC_TOPOLOGY_FLAG_WHOLE_SYSTEM)) { + assert(hwloc_bitmap_isset(topology->allowed_nodeset, (int) obj->os_index)); + } + assert(!obj->arity); + assert(!obj->memory_arity); + assert(hwloc_bitmap_isincluded(obj->nodeset, parentset)); + } else { + hwloc_bitmap_t myset; + hwloc_bitmap_t childset; + + /* the local nodeset is an exclusive OR of memory children */ + myset = hwloc_bitmap_alloc(); + for_each_memory_child(child, obj) { + assert(!hwloc_bitmap_intersects(myset, child->nodeset)); + hwloc_bitmap_or(myset, myset, child->nodeset); + } + /* the local nodeset cannot intersect with parents' local nodeset */ + assert(!hwloc_bitmap_intersects(myset, parentset)); + hwloc_bitmap_or(parentset, parentset, myset); + hwloc_bitmap_free(myset); + /* parentset now contains parent+local contribution */ + + /* for each children, recurse to check/get its contribution */ + childset = hwloc_bitmap_alloc(); + for_each_child(child, obj) { + hwloc_bitmap_t set = hwloc_bitmap_dup(parentset); /* don't touch parentset, we don't want to propagate the first child contribution to other children */ + hwloc__check_nodesets(topology, child, set); + /* extract this child contribution */ + hwloc_bitmap_andnot(set, set, parentset); + /* save it */ + assert(!hwloc_bitmap_intersects(childset, set)); + hwloc_bitmap_or(childset, childset, set); + hwloc_bitmap_free(set); + } + /* combine child contribution into parentset */ + assert(!hwloc_bitmap_intersects(parentset, childset)); + hwloc_bitmap_or(parentset, parentset, childset); + hwloc_bitmap_free(childset); + /* now check that our nodeset is combination of parent, local and children */ + assert(hwloc_bitmap_isequal(obj->nodeset, parentset)); + } + + /* check that children complete_nodesets are properly ordered, empty ones may be anywhere + * (can be wrong for main nodeset since removed PUs can break the ordering). + */ + prev_first = -1; /* -1 works fine with first comparisons below */ + for_each_memory_child(child, obj) { + int first = hwloc_bitmap_first(child->complete_nodeset); + assert(prev_first < first); + prev_first = first; + } +} + +static void +hwloc__check_level(struct hwloc_topology *topology, int depth, + hwloc_obj_t first, hwloc_obj_t last) +{ + unsigned width = hwloc_get_nbobjs_by_depth(topology, depth); + struct hwloc_obj *prev = NULL; + hwloc_obj_t obj; + unsigned j; + + /* check each object of the level */ + for(j=0; jdepth == depth); + assert(obj->logical_index == j); + /* check that all objects in the level have the same type */ + if (prev) { + assert(hwloc_type_cmp(obj, prev) == HWLOC_OBJ_EQUAL); + assert(prev->next_cousin == obj); + } + assert(obj->prev_cousin == prev); + + /* check that PUs and NUMA nodes have correct cpuset/nodeset */ + if (obj->type == HWLOC_OBJ_NUMANODE) { + assert(hwloc_bitmap_weight(obj->complete_nodeset) == 1); + assert(hwloc_bitmap_first(obj->complete_nodeset) == (int) obj->os_index); + } + prev = obj; + } + if (prev) + assert(prev->next_cousin == NULL); + + if (width) { + /* check first object of the level */ + obj = hwloc_get_obj_by_depth(topology, depth, 0); + assert(obj); + assert(!obj->prev_cousin); + /* check type */ + assert(hwloc_get_depth_type(topology, depth) == obj->type); + assert(depth == hwloc_get_type_depth(topology, obj->type) + || HWLOC_TYPE_DEPTH_MULTIPLE == hwloc_get_type_depth(topology, obj->type)); + /* check last object of the level */ + obj = hwloc_get_obj_by_depth(topology, depth, width-1); + assert(obj); + assert(!obj->next_cousin); + } + + if (depth < 0) { + assert(first == hwloc_get_obj_by_depth(topology, depth, 0)); + assert(last == hwloc_get_obj_by_depth(topology, depth, width-1)); + } else { + assert(!first); + assert(!last); + } + + /* check last+1 object of the level */ + obj = hwloc_get_obj_by_depth(topology, depth, width); + assert(!obj); +} + +/* check a whole topology structure */ +void +hwloc_topology_check(struct hwloc_topology *topology) +{ + struct hwloc_obj *obj; + hwloc_bitmap_t gp_indexes, set; + hwloc_obj_type_t type; + unsigned i; + int j, depth; + + /* make sure we can use ranges to check types */ + + /* hwloc__obj_type_is_{,d,i}cache() want cache types to be ordered like this */ + HWLOC_BUILD_ASSERT(HWLOC_OBJ_L2CACHE == HWLOC_OBJ_L1CACHE + 1); + HWLOC_BUILD_ASSERT(HWLOC_OBJ_L3CACHE == HWLOC_OBJ_L2CACHE + 1); + HWLOC_BUILD_ASSERT(HWLOC_OBJ_L4CACHE == HWLOC_OBJ_L3CACHE + 1); + HWLOC_BUILD_ASSERT(HWLOC_OBJ_L5CACHE == HWLOC_OBJ_L4CACHE + 1); + HWLOC_BUILD_ASSERT(HWLOC_OBJ_L1ICACHE == HWLOC_OBJ_L5CACHE + 1); + HWLOC_BUILD_ASSERT(HWLOC_OBJ_L2ICACHE == HWLOC_OBJ_L1ICACHE + 1); + HWLOC_BUILD_ASSERT(HWLOC_OBJ_L3ICACHE == HWLOC_OBJ_L2ICACHE + 1); + + /* hwloc__obj_type_is_normal(), hwloc__obj_type_is_memory(), hwloc__obj_type_is_io(), hwloc__obj_type_is_special() + * and hwloc_reset_normal_type_depths() + * want special types to be ordered like this, after all normal types. + */ + HWLOC_BUILD_ASSERT(HWLOC_OBJ_NUMANODE + 1 == HWLOC_OBJ_BRIDGE); + HWLOC_BUILD_ASSERT(HWLOC_OBJ_BRIDGE + 1 == HWLOC_OBJ_PCI_DEVICE); + HWLOC_BUILD_ASSERT(HWLOC_OBJ_PCI_DEVICE + 1 == HWLOC_OBJ_OS_DEVICE); + HWLOC_BUILD_ASSERT(HWLOC_OBJ_OS_DEVICE + 1 == HWLOC_OBJ_MISC); + HWLOC_BUILD_ASSERT(HWLOC_OBJ_MISC + 1 == HWLOC_OBJ_TYPE_MAX); + + /* make sure order and priority arrays have the right size */ + HWLOC_BUILD_ASSERT(sizeof(obj_type_order)/sizeof(*obj_type_order) == HWLOC_OBJ_TYPE_MAX); + HWLOC_BUILD_ASSERT(sizeof(obj_order_type)/sizeof(*obj_order_type) == HWLOC_OBJ_TYPE_MAX); + HWLOC_BUILD_ASSERT(sizeof(obj_type_priority)/sizeof(*obj_type_priority) == HWLOC_OBJ_TYPE_MAX); + + /* make sure group are not entirely ignored */ + assert(topology->type_filter[HWLOC_OBJ_GROUP] != HWLOC_TYPE_FILTER_KEEP_ALL); + + /* make sure order arrays are coherent */ + for(type=HWLOC_OBJ_TYPE_MIN; typemodified); + + /* check that first level is Machine. + * Root object cannot be ignored. And Machine can only be merged into PU, + * but there must be a NUMA node below Machine, and it cannot be below PU. + */ + assert(hwloc_get_depth_type(topology, 0) == HWLOC_OBJ_MACHINE); + + /* check that last level is PU and that it doesn't have memory */ + assert(hwloc_get_depth_type(topology, depth-1) == HWLOC_OBJ_PU); + assert(hwloc_get_nbobjs_by_depth(topology, depth-1) > 0); + for(i=0; itype == HWLOC_OBJ_PU); + assert(!obj->memory_first_child); + } + /* check that other levels are not PU or Machine */ + for(j=1; j=0 || d == HWLOC_TYPE_DEPTH_UNKNOWN || d == HWLOC_TYPE_DEPTH_MULTIPLE); + } + } + + /* top-level specific checks */ + assert(hwloc_get_nbobjs_by_depth(topology, 0) == 1); + obj = hwloc_get_root_obj(topology); + assert(obj); + assert(!obj->parent); + assert(obj->cpuset); + assert(!obj->depth); + + /* check that allowed sets are larger than the main sets */ + if (topology->flags & HWLOC_TOPOLOGY_FLAG_WHOLE_SYSTEM) { + assert(hwloc_bitmap_isincluded(topology->allowed_cpuset, obj->cpuset)); + assert(hwloc_bitmap_isincluded(topology->allowed_nodeset, obj->nodeset)); + } else { + assert(hwloc_bitmap_isequal(topology->allowed_cpuset, obj->cpuset)); + assert(hwloc_bitmap_isequal(topology->allowed_nodeset, obj->nodeset)); + } + + /* check each level */ + for(j=0; jslevels[j].first, topology->slevels[j].last); + + /* recurse and check the tree of children, and type-specific checks */ + gp_indexes = hwloc_bitmap_alloc(); /* TODO prealloc to topology->next_gp_index */ + hwloc__check_object(topology, gp_indexes, obj); + hwloc_bitmap_free(gp_indexes); + + /* recurse and check the nodesets of children */ + set = hwloc_bitmap_alloc(); + hwloc__check_nodesets(topology, obj, set); + hwloc_bitmap_free(set); +} + +#else /* NDEBUG */ + +void +hwloc_topology_check(struct hwloc_topology *topology __hwloc_attribute_unused) +{ +} + +#endif /* NDEBUG */ diff --git a/src/3rdparty/hwloc/src/traversal.c b/src/3rdparty/hwloc/src/traversal.c new file mode 100644 index 000000000..9c5e6268c --- /dev/null +++ b/src/3rdparty/hwloc/src/traversal.c @@ -0,0 +1,616 @@ +/* + * Copyright © 2009 CNRS + * Copyright © 2009-2018 Inria. All rights reserved. + * Copyright © 2009-2010 Université Bordeaux + * Copyright © 2009-2011 Cisco Systems, Inc. All rights reserved. + * See COPYING in top-level directory. + */ + +#include +#include +#include +#include +#include +#ifdef HAVE_STRINGS_H +#include +#endif /* HAVE_STRINGS_H */ + +int +hwloc_get_type_depth (struct hwloc_topology *topology, hwloc_obj_type_t type) +{ + HWLOC_BUILD_ASSERT(HWLOC_OBJ_TYPE_MIN == 0); + if ((unsigned) type >= HWLOC_OBJ_TYPE_MAX) + return HWLOC_TYPE_DEPTH_UNKNOWN; + else + return topology->type_depth[type]; +} + +hwloc_obj_type_t +hwloc_get_depth_type (hwloc_topology_t topology, int depth) +{ + if ((unsigned)depth >= topology->nb_levels) + switch (depth) { + case HWLOC_TYPE_DEPTH_NUMANODE: + return HWLOC_OBJ_NUMANODE; + case HWLOC_TYPE_DEPTH_BRIDGE: + return HWLOC_OBJ_BRIDGE; + case HWLOC_TYPE_DEPTH_PCI_DEVICE: + return HWLOC_OBJ_PCI_DEVICE; + case HWLOC_TYPE_DEPTH_OS_DEVICE: + return HWLOC_OBJ_OS_DEVICE; + case HWLOC_TYPE_DEPTH_MISC: + return HWLOC_OBJ_MISC; + default: + return HWLOC_OBJ_TYPE_NONE; + } + return topology->levels[depth][0]->type; +} + +int +hwloc_get_memory_parents_depth (hwloc_topology_t topology) +{ + int depth = HWLOC_TYPE_DEPTH_UNKNOWN; + /* memory leaves are always NUMA nodes for now, no need to check parents of other memory types */ + hwloc_obj_t numa = hwloc_get_obj_by_depth(topology, HWLOC_TYPE_DEPTH_NUMANODE, 0); + assert(numa); + while (numa) { + hwloc_obj_t parent = numa->parent; + /* walk-up the memory hierarchy */ + while (hwloc__obj_type_is_memory(parent->type)) + parent = parent->parent; + + if (depth == HWLOC_TYPE_DEPTH_UNKNOWN) + depth = parent->depth; + else if (depth != parent->depth) + return HWLOC_TYPE_DEPTH_MULTIPLE; + + numa = numa->next_cousin; + } + + assert(depth >= 0); + return depth; +} + +unsigned +hwloc_get_nbobjs_by_depth (struct hwloc_topology *topology, int depth) +{ + if ((unsigned)depth >= topology->nb_levels) { + unsigned l = HWLOC_SLEVEL_FROM_DEPTH(depth); + if (l < HWLOC_NR_SLEVELS) + return topology->slevels[l].nbobjs; + else + return 0; + } + return topology->level_nbobjects[depth]; +} + +struct hwloc_obj * +hwloc_get_obj_by_depth (struct hwloc_topology *topology, int depth, unsigned idx) +{ + if ((unsigned)depth >= topology->nb_levels) { + unsigned l = HWLOC_SLEVEL_FROM_DEPTH(depth); + if (l < HWLOC_NR_SLEVELS) + return idx < topology->slevels[l].nbobjs ? topology->slevels[l].objs[idx] : NULL; + else + return NULL; + } + if (idx >= topology->level_nbobjects[depth]) + return NULL; + return topology->levels[depth][idx]; +} + +int +hwloc_obj_type_is_normal(hwloc_obj_type_t type) +{ + return hwloc__obj_type_is_normal(type); +} + +int +hwloc_obj_type_is_memory(hwloc_obj_type_t type) +{ + return hwloc__obj_type_is_memory(type); +} + +int +hwloc_obj_type_is_io(hwloc_obj_type_t type) +{ + return hwloc__obj_type_is_io(type); +} + +int +hwloc_obj_type_is_cache(hwloc_obj_type_t type) +{ + return hwloc__obj_type_is_cache(type); +} + +int +hwloc_obj_type_is_dcache(hwloc_obj_type_t type) +{ + return hwloc__obj_type_is_dcache(type); +} + +int +hwloc_obj_type_is_icache(hwloc_obj_type_t type) +{ + return hwloc__obj_type_is_icache(type); +} + +unsigned hwloc_get_closest_objs (struct hwloc_topology *topology, struct hwloc_obj *src, struct hwloc_obj **objs, unsigned max) +{ + struct hwloc_obj *parent, *nextparent, **src_objs; + unsigned i,src_nbobjects; + unsigned stored = 0; + + if (!src->cpuset) + return 0; + + src_nbobjects = topology->level_nbobjects[src->depth]; + src_objs = topology->levels[src->depth]; + + parent = src; + while (stored < max) { + while (1) { + nextparent = parent->parent; + if (!nextparent) + goto out; + if (!hwloc_bitmap_isequal(parent->cpuset, nextparent->cpuset)) + break; + parent = nextparent; + } + + /* traverse src's objects and find those that are in nextparent and were not in parent */ + for(i=0; icpuset, nextparent->cpuset) + && !hwloc_bitmap_isincluded(src_objs[i]->cpuset, parent->cpuset)) { + objs[stored++] = src_objs[i]; + if (stored == max) + goto out; + } + } + parent = nextparent; + } + + out: + return stored; +} + +static int +hwloc__get_largest_objs_inside_cpuset (struct hwloc_obj *current, hwloc_const_bitmap_t set, + struct hwloc_obj ***res, int *max) +{ + int gotten = 0; + unsigned i; + + /* the caller must ensure this */ + if (*max <= 0) + return 0; + + if (hwloc_bitmap_isequal(current->cpuset, set)) { + **res = current; + (*res)++; + (*max)--; + return 1; + } + + for (i=0; iarity; i++) { + hwloc_bitmap_t subset; + int ret; + + /* split out the cpuset part corresponding to this child and see if there's anything to do */ + if (!hwloc_bitmap_intersects(set,current->children[i]->cpuset)) + continue; + + subset = hwloc_bitmap_dup(set); + hwloc_bitmap_and(subset, subset, current->children[i]->cpuset); + ret = hwloc__get_largest_objs_inside_cpuset (current->children[i], subset, res, max); + gotten += ret; + hwloc_bitmap_free(subset); + + /* if no more room to store remaining objects, return what we got so far */ + if (!*max) + break; + } + + return gotten; +} + +int +hwloc_get_largest_objs_inside_cpuset (struct hwloc_topology *topology, hwloc_const_bitmap_t set, + struct hwloc_obj **objs, int max) +{ + struct hwloc_obj *current = topology->levels[0][0]; + + if (!hwloc_bitmap_isincluded(set, current->cpuset)) + return -1; + + if (max <= 0) + return 0; + + return hwloc__get_largest_objs_inside_cpuset (current, set, &objs, &max); +} + +const char * +hwloc_obj_type_string (hwloc_obj_type_t obj) +{ + switch (obj) + { + case HWLOC_OBJ_MACHINE: return "Machine"; + case HWLOC_OBJ_MISC: return "Misc"; + case HWLOC_OBJ_GROUP: return "Group"; + case HWLOC_OBJ_NUMANODE: return "NUMANode"; + case HWLOC_OBJ_PACKAGE: return "Package"; + case HWLOC_OBJ_L1CACHE: return "L1Cache"; + case HWLOC_OBJ_L2CACHE: return "L2Cache"; + case HWLOC_OBJ_L3CACHE: return "L3Cache"; + case HWLOC_OBJ_L4CACHE: return "L4Cache"; + case HWLOC_OBJ_L5CACHE: return "L5Cache"; + case HWLOC_OBJ_L1ICACHE: return "L1iCache"; + case HWLOC_OBJ_L2ICACHE: return "L2iCache"; + case HWLOC_OBJ_L3ICACHE: return "L3iCache"; + case HWLOC_OBJ_CORE: return "Core"; + case HWLOC_OBJ_BRIDGE: return "Bridge"; + case HWLOC_OBJ_PCI_DEVICE: return "PCIDev"; + case HWLOC_OBJ_OS_DEVICE: return "OSDev"; + case HWLOC_OBJ_PU: return "PU"; + default: return "Unknown"; + } +} + +int +hwloc_type_sscanf(const char *string, hwloc_obj_type_t *typep, + union hwloc_obj_attr_u *attrp, size_t attrsize) +{ + hwloc_obj_type_t type = (hwloc_obj_type_t) -1; + unsigned depthattr = (unsigned) -1; + hwloc_obj_cache_type_t cachetypeattr = (hwloc_obj_cache_type_t) -1; /* unspecified */ + hwloc_obj_bridge_type_t ubtype = (hwloc_obj_bridge_type_t) -1; + hwloc_obj_osdev_type_t ostype = (hwloc_obj_osdev_type_t) -1; + char *end; + + /* never match the ending \0 since we want to match things like core:2 too. + * just use hwloc_strncasecmp() everywhere. + */ + + /* types without a custom depth */ + + /* osdev subtype first to avoid conflicts coproc/core etc */ + if (!hwloc_strncasecmp(string, "os", 2)) { + type = HWLOC_OBJ_OS_DEVICE; + } else if (!hwloc_strncasecmp(string, "bloc", 4)) { + type = HWLOC_OBJ_OS_DEVICE; + ostype = HWLOC_OBJ_OSDEV_BLOCK; + } else if (!hwloc_strncasecmp(string, "net", 3)) { + type = HWLOC_OBJ_OS_DEVICE; + ostype = HWLOC_OBJ_OSDEV_NETWORK; + } else if (!hwloc_strncasecmp(string, "openfab", 7)) { + type = HWLOC_OBJ_OS_DEVICE; + ostype = HWLOC_OBJ_OSDEV_OPENFABRICS; + } else if (!hwloc_strncasecmp(string, "dma", 3)) { + type = HWLOC_OBJ_OS_DEVICE; + ostype = HWLOC_OBJ_OSDEV_DMA; + } else if (!hwloc_strncasecmp(string, "gpu", 3)) { + type = HWLOC_OBJ_OS_DEVICE; + ostype = HWLOC_OBJ_OSDEV_GPU; + } else if (!hwloc_strncasecmp(string, "copro", 5) + || !hwloc_strncasecmp(string, "co-pro", 6)) { + type = HWLOC_OBJ_OS_DEVICE; + ostype = HWLOC_OBJ_OSDEV_COPROC; + + } else if (!hwloc_strncasecmp(string, "machine", 2)) { + type = HWLOC_OBJ_MACHINE; + } else if (!hwloc_strncasecmp(string, "node", 2) + || !hwloc_strncasecmp(string, "numa", 2)) { /* matches node and numanode */ + type = HWLOC_OBJ_NUMANODE; + } else if (!hwloc_strncasecmp(string, "package", 2) + || !hwloc_strncasecmp(string, "socket", 2)) { /* backward compat with v1.10 */ + type = HWLOC_OBJ_PACKAGE; + } else if (!hwloc_strncasecmp(string, "core", 2)) { + type = HWLOC_OBJ_CORE; + } else if (!hwloc_strncasecmp(string, "pu", 2)) { + type = HWLOC_OBJ_PU; + } else if (!hwloc_strncasecmp(string, "misc", 4)) { + type = HWLOC_OBJ_MISC; + + } else if (!hwloc_strncasecmp(string, "bridge", 4)) { + type = HWLOC_OBJ_BRIDGE; + } else if (!hwloc_strncasecmp(string, "hostbridge", 6)) { + type = HWLOC_OBJ_BRIDGE; + ubtype = HWLOC_OBJ_BRIDGE_HOST; + } else if (!hwloc_strncasecmp(string, "pcibridge", 5)) { + type = HWLOC_OBJ_BRIDGE; + ubtype = HWLOC_OBJ_BRIDGE_PCI; + + } else if (!hwloc_strncasecmp(string, "pci", 3)) { + type = HWLOC_OBJ_PCI_DEVICE; + + /* types with depthattr */ + } else if ((string[0] == 'l' || string[0] == 'L') && string[1] >= '0' && string[1] <= '9') { + depthattr = strtol(string+1, &end, 10); + if (*end == 'i') { + if (depthattr >= 1 && depthattr <= 3) { + type = HWLOC_OBJ_L1ICACHE + depthattr-1; + cachetypeattr = HWLOC_OBJ_CACHE_INSTRUCTION; + } else + return -1; + } else { + if (depthattr >= 1 && depthattr <= 5) { + type = HWLOC_OBJ_L1CACHE + depthattr-1; + cachetypeattr = *end == 'd' ? HWLOC_OBJ_CACHE_DATA : HWLOC_OBJ_CACHE_UNIFIED; + } else + return -1; + } + + } else if (!hwloc_strncasecmp(string, "group", 2)) { + size_t length; + type = HWLOC_OBJ_GROUP; + length = strcspn(string, "0123456789"); + if (length <= 5 && !hwloc_strncasecmp(string, "group", length) + && string[length] >= '0' && string[length] <= '9') { + depthattr = strtol(string+length, &end, 10); + } + + } else + return -1; + + *typep = type; + if (attrp) { + if (hwloc__obj_type_is_cache(type) && attrsize >= sizeof(attrp->cache)) { + attrp->cache.depth = depthattr; + attrp->cache.type = cachetypeattr; + } else if (type == HWLOC_OBJ_GROUP && attrsize >= sizeof(attrp->group)) { + attrp->group.depth = depthattr; + } else if (type == HWLOC_OBJ_BRIDGE && attrsize >= sizeof(attrp->bridge)) { + attrp->bridge.upstream_type = ubtype; + attrp->bridge.downstream_type = HWLOC_OBJ_BRIDGE_PCI; /* nothing else so far */ + } else if (type == HWLOC_OBJ_OS_DEVICE && attrsize >= sizeof(attrp->osdev)) { + attrp->osdev.type = ostype; + } + } + return 0; +} + +int +hwloc_type_sscanf_as_depth(const char *string, hwloc_obj_type_t *typep, + hwloc_topology_t topology, int *depthp) +{ + union hwloc_obj_attr_u attr; + hwloc_obj_type_t type; + int depth; + int err; + + err = hwloc_type_sscanf(string, &type, &attr, sizeof(attr)); + if (err < 0) + return err; + + depth = hwloc_get_type_depth(topology, type); + if (type == HWLOC_OBJ_GROUP + && depth == HWLOC_TYPE_DEPTH_MULTIPLE + && attr.group.depth != (unsigned)-1) { + unsigned l; + depth = HWLOC_TYPE_DEPTH_UNKNOWN; + for(l=0; lnb_levels; l++) { + if (topology->levels[l][0]->type == HWLOC_OBJ_GROUP + && topology->levels[l][0]->attr->group.depth == attr.group.depth) { + depth = (int)l; + break; + } + } + } + + if (typep) + *typep = type; + *depthp = depth; + return 0; +} + +static const char* hwloc_obj_cache_type_letter(hwloc_obj_cache_type_t type) +{ + switch (type) { + case HWLOC_OBJ_CACHE_UNIFIED: return ""; + case HWLOC_OBJ_CACHE_DATA: return "d"; + case HWLOC_OBJ_CACHE_INSTRUCTION: return "i"; + default: return "unknown"; + } +} + +int +hwloc_obj_type_snprintf(char * __hwloc_restrict string, size_t size, hwloc_obj_t obj, int verbose) +{ + hwloc_obj_type_t type = obj->type; + switch (type) { + case HWLOC_OBJ_MISC: + case HWLOC_OBJ_MACHINE: + case HWLOC_OBJ_NUMANODE: + case HWLOC_OBJ_PACKAGE: + case HWLOC_OBJ_CORE: + case HWLOC_OBJ_PU: + return hwloc_snprintf(string, size, "%s", hwloc_obj_type_string(type)); + case HWLOC_OBJ_L1CACHE: + case HWLOC_OBJ_L2CACHE: + case HWLOC_OBJ_L3CACHE: + case HWLOC_OBJ_L4CACHE: + case HWLOC_OBJ_L5CACHE: + case HWLOC_OBJ_L1ICACHE: + case HWLOC_OBJ_L2ICACHE: + case HWLOC_OBJ_L3ICACHE: + return hwloc_snprintf(string, size, "L%u%s%s", obj->attr->cache.depth, + hwloc_obj_cache_type_letter(obj->attr->cache.type), + verbose ? "Cache" : ""); + case HWLOC_OBJ_GROUP: + if (obj->attr->group.depth != (unsigned) -1) + return hwloc_snprintf(string, size, "%s%u", hwloc_obj_type_string(type), obj->attr->group.depth); + else + return hwloc_snprintf(string, size, "%s", hwloc_obj_type_string(type)); + case HWLOC_OBJ_BRIDGE: + return hwloc_snprintf(string, size, obj->attr->bridge.upstream_type == HWLOC_OBJ_BRIDGE_PCI ? "PCIBridge" : "HostBridge"); + case HWLOC_OBJ_PCI_DEVICE: + return hwloc_snprintf(string, size, "PCI"); + case HWLOC_OBJ_OS_DEVICE: + switch (obj->attr->osdev.type) { + case HWLOC_OBJ_OSDEV_BLOCK: return hwloc_snprintf(string, size, "Block"); + case HWLOC_OBJ_OSDEV_NETWORK: return hwloc_snprintf(string, size, verbose ? "Network" : "Net"); + case HWLOC_OBJ_OSDEV_OPENFABRICS: return hwloc_snprintf(string, size, "OpenFabrics"); + case HWLOC_OBJ_OSDEV_DMA: return hwloc_snprintf(string, size, "DMA"); + case HWLOC_OBJ_OSDEV_GPU: return hwloc_snprintf(string, size, "GPU"); + case HWLOC_OBJ_OSDEV_COPROC: return hwloc_snprintf(string, size, verbose ? "Co-Processor" : "CoProc"); + default: + if (size > 0) + *string = '\0'; + return 0; + } + break; + default: + if (size > 0) + *string = '\0'; + return 0; + } +} + +int +hwloc_obj_attr_snprintf(char * __hwloc_restrict string, size_t size, hwloc_obj_t obj, const char * separator, int verbose) +{ + const char *prefix = ""; + char *tmp = string; + ssize_t tmplen = size; + int ret = 0; + int res; + + /* make sure we output at least an empty string */ + if (size) + *string = '\0'; + + /* print memory attributes */ + res = 0; + if (verbose) { + if (obj->type == HWLOC_OBJ_NUMANODE && obj->attr->numanode.local_memory) + res = hwloc_snprintf(tmp, tmplen, "%slocal=%lu%s%stotal=%lu%s", + prefix, + (unsigned long) hwloc_memory_size_printf_value(obj->attr->numanode.local_memory, verbose), + hwloc_memory_size_printf_unit(obj->attr->numanode.local_memory, verbose), + separator, + (unsigned long) hwloc_memory_size_printf_value(obj->total_memory, verbose), + hwloc_memory_size_printf_unit(obj->total_memory, verbose)); + else if (obj->total_memory) + res = hwloc_snprintf(tmp, tmplen, "%stotal=%lu%s", + prefix, + (unsigned long) hwloc_memory_size_printf_value(obj->total_memory, verbose), + hwloc_memory_size_printf_unit(obj->total_memory, verbose)); + } else { + if (obj->type == HWLOC_OBJ_NUMANODE && obj->attr->numanode.local_memory) + res = hwloc_snprintf(tmp, tmplen, "%s%lu%s", + prefix, + (unsigned long) hwloc_memory_size_printf_value(obj->attr->numanode.local_memory, verbose), + hwloc_memory_size_printf_unit(obj->attr->numanode.local_memory, verbose)); + } + if (res < 0) + return -1; + ret += res; + if (ret > 0) + prefix = separator; + if (res >= tmplen) + res = tmplen>0 ? (int)tmplen - 1 : 0; + tmp += res; + tmplen -= res; + + /* printf type-specific attributes */ + res = 0; + switch (obj->type) { + case HWLOC_OBJ_L1CACHE: + case HWLOC_OBJ_L2CACHE: + case HWLOC_OBJ_L3CACHE: + case HWLOC_OBJ_L4CACHE: + case HWLOC_OBJ_L5CACHE: + case HWLOC_OBJ_L1ICACHE: + case HWLOC_OBJ_L2ICACHE: + case HWLOC_OBJ_L3ICACHE: + if (verbose) { + char assoc[32]; + if (obj->attr->cache.associativity == -1) + snprintf(assoc, sizeof(assoc), "%sfully-associative", separator); + else if (obj->attr->cache.associativity == 0) + *assoc = '\0'; + else + snprintf(assoc, sizeof(assoc), "%sways=%d", separator, obj->attr->cache.associativity); + res = hwloc_snprintf(tmp, tmplen, "%ssize=%lu%s%slinesize=%u%s", + prefix, + (unsigned long) hwloc_memory_size_printf_value(obj->attr->cache.size, verbose), + hwloc_memory_size_printf_unit(obj->attr->cache.size, verbose), + separator, obj->attr->cache.linesize, + assoc); + } else + res = hwloc_snprintf(tmp, tmplen, "%s%lu%s", + prefix, + (unsigned long) hwloc_memory_size_printf_value(obj->attr->cache.size, verbose), + hwloc_memory_size_printf_unit(obj->attr->cache.size, verbose)); + break; + case HWLOC_OBJ_BRIDGE: + if (verbose) { + char up[128], down[64]; + /* upstream is PCI or HOST */ + if (obj->attr->bridge.upstream_type == HWLOC_OBJ_BRIDGE_PCI) { + char linkspeed[64]= ""; + if (obj->attr->pcidev.linkspeed) + snprintf(linkspeed, sizeof(linkspeed), "%slink=%.2fGB/s", separator, obj->attr->pcidev.linkspeed); + snprintf(up, sizeof(up), "busid=%04x:%02x:%02x.%01x%sid=%04x:%04x%sclass=%04x(%s)%s", + obj->attr->pcidev.domain, obj->attr->pcidev.bus, obj->attr->pcidev.dev, obj->attr->pcidev.func, separator, + obj->attr->pcidev.vendor_id, obj->attr->pcidev.device_id, separator, + obj->attr->pcidev.class_id, hwloc_pci_class_string(obj->attr->pcidev.class_id), linkspeed); + } else + *up = '\0'; + /* downstream is_PCI */ + snprintf(down, sizeof(down), "buses=%04x:[%02x-%02x]", + obj->attr->bridge.downstream.pci.domain, obj->attr->bridge.downstream.pci.secondary_bus, obj->attr->bridge.downstream.pci.subordinate_bus); + if (*up) + res = hwloc_snprintf(string, size, "%s%s%s", up, separator, down); + else + res = hwloc_snprintf(string, size, "%s", down); + } + break; + case HWLOC_OBJ_PCI_DEVICE: + if (verbose) { + char linkspeed[64]= ""; + if (obj->attr->pcidev.linkspeed) + snprintf(linkspeed, sizeof(linkspeed), "%slink=%.2fGB/s", separator, obj->attr->pcidev.linkspeed); + res = hwloc_snprintf(string, size, "busid=%04x:%02x:%02x.%01x%sid=%04x:%04x%sclass=%04x(%s)%s", + obj->attr->pcidev.domain, obj->attr->pcidev.bus, obj->attr->pcidev.dev, obj->attr->pcidev.func, separator, + obj->attr->pcidev.vendor_id, obj->attr->pcidev.device_id, separator, + obj->attr->pcidev.class_id, hwloc_pci_class_string(obj->attr->pcidev.class_id), linkspeed); + } + break; + default: + break; + } + if (res < 0) + return -1; + ret += res; + if (ret > 0) + prefix = separator; + if (res >= tmplen) + res = tmplen>0 ? (int)tmplen - 1 : 0; + tmp += res; + tmplen -= res; + + /* printf infos */ + if (verbose) { + unsigned i; + for(i=0; iinfos_count; i++) { + struct hwloc_info_s *info = &obj->infos[i]; + const char *quote = strchr(info->value, ' ') ? "\"" : ""; + res = hwloc_snprintf(tmp, tmplen, "%s%s=%s%s%s", + prefix, + info->name, + quote, info->value, quote); + if (res < 0) + return -1; + ret += res; + if (res >= tmplen) + res = tmplen>0 ? (int)tmplen - 1 : 0; + tmp += res; + tmplen -= res; + if (ret > 0) + prefix = separator; + } + } + + return ret; +} diff --git a/src/3rdparty/rapidjson/allocators.h b/src/3rdparty/rapidjson/allocators.h index 98affe03f..cc67c8971 100644 --- a/src/3rdparty/rapidjson/allocators.h +++ b/src/3rdparty/rapidjson/allocators.h @@ -52,6 +52,19 @@ concept Allocator { \endcode */ + +/*! \def RAPIDJSON_ALLOCATOR_DEFAULT_CHUNK_CAPACITY + \ingroup RAPIDJSON_CONFIG + \brief User-defined kDefaultChunkCapacity definition. + + User can define this as any \c size that is a power of 2. +*/ + +#ifndef RAPIDJSON_ALLOCATOR_DEFAULT_CHUNK_CAPACITY +#define RAPIDJSON_ALLOCATOR_DEFAULT_CHUNK_CAPACITY (64 * 1024) +#endif + + /////////////////////////////////////////////////////////////////////////////// // CrtAllocator @@ -236,7 +249,7 @@ private: */ bool AddChunk(size_t capacity) { if (!baseAllocator_) - ownBaseAllocator_ = baseAllocator_ = RAPIDJSON_NEW(BaseAllocator()); + ownBaseAllocator_ = baseAllocator_ = RAPIDJSON_NEW(BaseAllocator)(); if (ChunkHeader* chunk = reinterpret_cast(baseAllocator_->Malloc(RAPIDJSON_ALIGN(sizeof(ChunkHeader)) + capacity))) { chunk->capacity = capacity; chunk->size = 0; @@ -248,7 +261,7 @@ private: return false; } - static const int kDefaultChunkCapacity = 64 * 1024; //!< Default chunk capacity. + static const int kDefaultChunkCapacity = RAPIDJSON_ALLOCATOR_DEFAULT_CHUNK_CAPACITY; //!< Default chunk capacity. //! Chunk header for perpending to each chunk. /*! Chunks are stored as a singly linked list. diff --git a/src/3rdparty/rapidjson/cursorstreamwrapper.h b/src/3rdparty/rapidjson/cursorstreamwrapper.h new file mode 100644 index 000000000..52c11a7c0 --- /dev/null +++ b/src/3rdparty/rapidjson/cursorstreamwrapper.h @@ -0,0 +1,78 @@ +// Tencent is pleased to support the open source community by making RapidJSON available. +// +// Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved. +// +// Licensed under the MIT License (the "License"); you may not use this file except +// in compliance with the License. You may obtain a copy of the License at +// +// http://opensource.org/licenses/MIT +// +// Unless required by applicable law or agreed to in writing, software distributed +// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR +// CONDITIONS OF ANY KIND, either express or implied. See the License for the +// specific language governing permissions and limitations under the License. + +#ifndef RAPIDJSON_CURSORSTREAMWRAPPER_H_ +#define RAPIDJSON_CURSORSTREAMWRAPPER_H_ + +#include "stream.h" + +#if defined(__GNUC__) +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(effc++) +#endif + +#if defined(_MSC_VER) && _MSC_VER <= 1800 +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(4702) // unreachable code +RAPIDJSON_DIAG_OFF(4512) // assignment operator could not be generated +#endif + +RAPIDJSON_NAMESPACE_BEGIN + + +//! Cursor stream wrapper for counting line and column number if error exists. +/*! + \tparam InputStream Any stream that implements Stream Concept +*/ +template > +class CursorStreamWrapper : public GenericStreamWrapper { +public: + typedef typename Encoding::Ch Ch; + + CursorStreamWrapper(InputStream& is): + GenericStreamWrapper(is), line_(1), col_(0) {} + + // counting line and column number + Ch Take() { + Ch ch = this->is_.Take(); + if(ch == '\n') { + line_ ++; + col_ = 0; + } else { + col_ ++; + } + return ch; + } + + //! Get the error line number, if error exists. + size_t GetLine() const { return line_; } + //! Get the error column number, if error exists. + size_t GetColumn() const { return col_; } + +private: + size_t line_; //!< Current Line + size_t col_; //!< Current Column +}; + +#if defined(_MSC_VER) && _MSC_VER <= 1800 +RAPIDJSON_DIAG_POP +#endif + +#if defined(__GNUC__) +RAPIDJSON_DIAG_POP +#endif + +RAPIDJSON_NAMESPACE_END + +#endif // RAPIDJSON_CURSORSTREAMWRAPPER_H_ diff --git a/src/3rdparty/rapidjson/document.h b/src/3rdparty/rapidjson/document.h index e3e20dfbd..9783fe4ac 100644 --- a/src/3rdparty/rapidjson/document.h +++ b/src/3rdparty/rapidjson/document.h @@ -26,26 +26,21 @@ #include RAPIDJSON_DIAG_PUSH -#ifdef _MSC_VER -RAPIDJSON_DIAG_OFF(4127) // conditional expression is constant -RAPIDJSON_DIAG_OFF(4244) // conversion from kXxxFlags to 'uint16_t', possible loss of data -#endif - #ifdef __clang__ RAPIDJSON_DIAG_OFF(padded) RAPIDJSON_DIAG_OFF(switch-enum) RAPIDJSON_DIAG_OFF(c++98-compat) +#elif defined(_MSC_VER) +RAPIDJSON_DIAG_OFF(4127) // conditional expression is constant +RAPIDJSON_DIAG_OFF(4244) // conversion from kXxxFlags to 'uint16_t', possible loss of data #endif #ifdef __GNUC__ RAPIDJSON_DIAG_OFF(effc++) -#if __GNUC__ >= 6 -RAPIDJSON_DIAG_OFF(terminate) // ignore throwing RAPIDJSON_ASSERT in RAPIDJSON_NOEXCEPT functions -#endif #endif // __GNUC__ #ifndef RAPIDJSON_NOMEMBERITERATORCLASS -#include // std::iterator, std::random_access_iterator_tag +#include // std::random_access_iterator_tag #endif #if RAPIDJSON_HAS_CXX11_RVALUE_REFS @@ -71,6 +66,12 @@ template struct GenericMember { GenericValue name; //!< name of member (must be a string) GenericValue value; //!< value of member. + + // swap() for std::sort() and other potential use in STL. + friend inline void swap(GenericMember& a, GenericMember& b) RAPIDJSON_NOEXCEPT { + a.name.Swap(b.name); + a.value.Swap(b.value); + } }; /////////////////////////////////////////////////////////////////////////////// @@ -98,16 +99,13 @@ struct GenericMember { \see GenericMember, GenericValue::MemberIterator, GenericValue::ConstMemberIterator */ template -class GenericMemberIterator - : public std::iterator >::Type> { +class GenericMemberIterator { friend class GenericValue; template friend class GenericMemberIterator; typedef GenericMember PlainType; typedef typename internal::MaybeAddConst::Type ValueType; - typedef std::iterator BaseType; public: //! Iterator type itself @@ -117,12 +115,21 @@ public: //! Non-constant iterator type typedef GenericMemberIterator NonConstIterator; + /** \name std::iterator_traits support */ + //@{ + typedef ValueType value_type; + typedef ValueType * pointer; + typedef ValueType & reference; + typedef std::ptrdiff_t difference_type; + typedef std::random_access_iterator_tag iterator_category; + //@} + //! Pointer to (const) GenericMember - typedef typename BaseType::pointer Pointer; + typedef pointer Pointer; //! Reference to (const) GenericMember - typedef typename BaseType::reference Reference; + typedef reference Reference; //! Signed integer type (e.g. \c ptrdiff_t) - typedef typename BaseType::difference_type DifferenceType; + typedef difference_type DifferenceType; //! Default constructor (singular value) /*! Creates an iterator pointing to no element. @@ -198,17 +205,17 @@ private: // class-based member iterator implementation disabled, use plain pointers template -struct GenericMemberIterator; +class GenericMemberIterator; //! non-const GenericMemberIterator template -struct GenericMemberIterator { +class GenericMemberIterator { //! use plain pointer as iterator type typedef GenericMember* Iterator; }; //! const GenericMemberIterator template -struct GenericMemberIterator { +class GenericMemberIterator { //! use plain const pointer as iterator type typedef const GenericMember* Iterator; }; @@ -300,7 +307,7 @@ struct GenericStringRef { */ #endif explicit GenericStringRef(const CharType* str) - : s(str), length(internal::StrLen(str)){ RAPIDJSON_ASSERT(s != 0); } + : s(str), length(NotNullStrLen(str)) {} //! Create constant string reference from pointer and length #ifndef __clang__ // -Wdocumentation @@ -312,12 +319,10 @@ struct GenericStringRef { */ #endif GenericStringRef(const CharType* str, SizeType len) - : s(str), length(len) { RAPIDJSON_ASSERT(s != 0); } + : s(RAPIDJSON_LIKELY(str) ? str : emptyString), length(len) { RAPIDJSON_ASSERT(str != 0 || len == 0u); } GenericStringRef(const GenericStringRef& rhs) : s(rhs.s), length(rhs.length) {} - GenericStringRef& operator=(const GenericStringRef& rhs) { s = rhs.s; length = rhs.length; } - //! implicit conversion to plain CharType pointer operator const Ch *() const { return s; } @@ -325,11 +330,24 @@ struct GenericStringRef { const SizeType length; //!< length of the string (excluding the trailing NULL terminator) private: + SizeType NotNullStrLen(const CharType* str) { + RAPIDJSON_ASSERT(str != 0); + return internal::StrLen(str); + } + + /// Empty string - used when passing in a NULL pointer + static const Ch emptyString[]; + //! Disallow construction from non-const array template GenericStringRef(CharType (&str)[N]) /* = delete */; + //! Copy assignment operator not permitted - immutable type + GenericStringRef& operator=(const GenericStringRef& rhs) /* = delete */; }; +template +const CharType GenericStringRef::emptyString[] = { CharType() }; + //! Mark a character pointer as constant string /*! Mark a plain character pointer as a "string literal". This function can be used to avoid copying a character string to be referenced as a @@ -344,7 +362,7 @@ private: */ template inline GenericStringRef StringRef(const CharType* str) { - return GenericStringRef(str, internal::StrLen(str)); + return GenericStringRef(str); } //! Mark a character pointer as constant string @@ -434,6 +452,26 @@ struct TypeHelper { static ValueType& Set(ValueType& v, unsigned data, typename ValueType::AllocatorType&) { return v.SetUint(data); } }; +#ifdef _MSC_VER +RAPIDJSON_STATIC_ASSERT(sizeof(long) == sizeof(int)); +template +struct TypeHelper { + static bool Is(const ValueType& v) { return v.IsInt(); } + static long Get(const ValueType& v) { return v.GetInt(); } + static ValueType& Set(ValueType& v, long data) { return v.SetInt(data); } + static ValueType& Set(ValueType& v, long data, typename ValueType::AllocatorType&) { return v.SetInt(data); } +}; + +RAPIDJSON_STATIC_ASSERT(sizeof(unsigned long) == sizeof(unsigned)); +template +struct TypeHelper { + static bool Is(const ValueType& v) { return v.IsUint(); } + static unsigned long Get(const ValueType& v) { return v.GetUint(); } + static ValueType& Set(ValueType& v, unsigned long data) { return v.SetUint(data); } + static ValueType& Set(ValueType& v, unsigned long data, typename ValueType::AllocatorType&) { return v.SetUint(data); } +}; +#endif + template struct TypeHelper { static bool Is(const ValueType& v) { return v.IsInt64(); } @@ -507,7 +545,7 @@ struct TypeHelper { static bool Is(const ValueType& v) { return v.IsObject(); } static ObjectType Get(ValueType& v) { return v.GetObject(); } static ValueType& Set(ValueType& v, ObjectType data) { return v = data; } - static ValueType& Set(ValueType& v, ObjectType data, typename ValueType::AllocatorType&) { v = data; } + static ValueType& Set(ValueType& v, ObjectType data, typename ValueType::AllocatorType&) { return v = data; } }; template @@ -590,11 +628,11 @@ public: \note Default content for number is zero. */ explicit GenericValue(Type type) RAPIDJSON_NOEXCEPT : data_() { - static const uint16_t defaultFlags[7] = { + static const uint16_t defaultFlags[] = { kNullFlag, kFalseFlag, kTrueFlag, kObjectFlag, kArrayFlag, kShortStringFlag, kNumberAnyFlag }; - RAPIDJSON_ASSERT(type <= kNumberType); + RAPIDJSON_NOEXCEPT_ASSERT(type >= kNullType && type <= kNumberType); data_.f.flags = defaultFlags[type]; // Use ShortString to store empty string. @@ -607,10 +645,50 @@ public: \tparam SourceAllocator allocator of \c rhs \param rhs Value to copy from (read-only) \param allocator Allocator for allocating copied elements and buffers. Commonly use GenericDocument::GetAllocator(). + \param copyConstStrings Force copying of constant strings (e.g. referencing an in-situ buffer) \see CopyFrom() */ - template< typename SourceAllocator > - GenericValue(const GenericValue& rhs, Allocator & allocator); + template + GenericValue(const GenericValue& rhs, Allocator& allocator, bool copyConstStrings = false) { + switch (rhs.GetType()) { + case kObjectType: { + SizeType count = rhs.data_.o.size; + Member* lm = reinterpret_cast(allocator.Malloc(count * sizeof(Member))); + const typename GenericValue::Member* rm = rhs.GetMembersPointer(); + for (SizeType i = 0; i < count; i++) { + new (&lm[i].name) GenericValue(rm[i].name, allocator, copyConstStrings); + new (&lm[i].value) GenericValue(rm[i].value, allocator, copyConstStrings); + } + data_.f.flags = kObjectFlag; + data_.o.size = data_.o.capacity = count; + SetMembersPointer(lm); + } + break; + case kArrayType: { + SizeType count = rhs.data_.a.size; + GenericValue* le = reinterpret_cast(allocator.Malloc(count * sizeof(GenericValue))); + const GenericValue* re = rhs.GetElementsPointer(); + for (SizeType i = 0; i < count; i++) + new (&le[i]) GenericValue(re[i], allocator, copyConstStrings); + data_.f.flags = kArrayFlag; + data_.a.size = data_.a.capacity = count; + SetElementsPointer(le); + } + break; + case kStringType: + if (rhs.data_.f.flags == kConstStringFlag && !copyConstStrings) { + data_.f.flags = rhs.data_.f.flags; + data_ = *reinterpret_cast(&rhs.data_); + } + else + SetStringRaw(StringRef(rhs.GetString(), rhs.GetStringLength()), allocator); + break; + default: + data_.f.flags = rhs.data_.f.flags; + data_ = *reinterpret_cast(&rhs.data_); + break; + } + } //! Constructor for boolean value. /*! \param b Boolean value @@ -672,6 +750,9 @@ public: //! Constructor for double value. explicit GenericValue(double d) RAPIDJSON_NOEXCEPT : data_() { data_.n.d = d; data_.f.flags = kNumberDoubleFlag; } + //! Constructor for float value. + explicit GenericValue(float f) RAPIDJSON_NOEXCEPT : data_() { data_.n.d = static_cast(f); data_.f.flags = kNumberDoubleFlag; } + //! Constructor for constant string (i.e. do not make a copy of string) GenericValue(const Ch* s, SizeType length) RAPIDJSON_NOEXCEPT : data_() { SetStringRaw(StringRef(s, length)); } @@ -753,9 +834,10 @@ public: /*! \param rhs Source of the assignment. It will become a null value after assignment. */ GenericValue& operator=(GenericValue& rhs) RAPIDJSON_NOEXCEPT { - RAPIDJSON_ASSERT(this != &rhs); - this->~GenericValue(); - RawAssign(rhs); + if (RAPIDJSON_LIKELY(this != &rhs)) { + this->~GenericValue(); + RawAssign(rhs); + } return *this; } @@ -800,12 +882,13 @@ public: \tparam SourceAllocator Allocator type of \c rhs \param rhs Value to copy from (read-only) \param allocator Allocator to use for copying + \param copyConstStrings Force copying of constant strings (e.g. referencing an in-situ buffer) */ template - GenericValue& CopyFrom(const GenericValue& rhs, Allocator& allocator) { + GenericValue& CopyFrom(const GenericValue& rhs, Allocator& allocator, bool copyConstStrings = false) { RAPIDJSON_ASSERT(static_cast(this) != static_cast(&rhs)); this->~GenericValue(); - new (this) GenericValue(rhs, allocator); + new (this) GenericValue(rhs, allocator, copyConstStrings); return *this; } @@ -846,7 +929,7 @@ public: //! Equal-to operator /*! \note If an object contains duplicated named member, comparing equality with any object is always \c false. - \note Linear time complexity (number of all values in the subtree and total lengths of all strings). + \note Complexity is quadratic in Object's member number and linear for the rest (number of all values in the subtree and total lengths of all strings). */ template bool operator==(const GenericValue& rhs) const { @@ -955,14 +1038,14 @@ public: uint64_t u = GetUint64(); volatile double d = static_cast(u); return (d >= 0.0) - && (d < static_cast(std::numeric_limits::max())) + && (d < static_cast((std::numeric_limits::max)())) && (u == static_cast(d)); } if (IsInt64()) { int64_t i = GetInt64(); volatile double d = static_cast(i); - return (d >= static_cast(std::numeric_limits::min())) - && (d < static_cast(std::numeric_limits::max())) + return (d >= static_cast((std::numeric_limits::min)())) + && (d < static_cast((std::numeric_limits::max)())) && (i == static_cast(d)); } return true; // double, int, uint are always lossless @@ -979,8 +1062,8 @@ public: bool IsLosslessFloat() const { if (!IsNumber()) return false; double a = GetDouble(); - if (a < static_cast(-std::numeric_limits::max()) - || a > static_cast(std::numeric_limits::max())) + if (a < static_cast(-(std::numeric_limits::max)()) + || a > static_cast((std::numeric_limits::max)())) return false; double b = static_cast(static_cast(a)); return a >= b && a <= b; // Prevent -Wfloat-equal @@ -1015,6 +1098,9 @@ public: //! Get the number of members in the object. SizeType MemberCount() const { RAPIDJSON_ASSERT(IsObject()); return data_.o.size; } + //! Get the capacity of object. + SizeType MemberCapacity() const { RAPIDJSON_ASSERT(IsObject()); return data_.o.capacity; } + //! Check whether the object is empty. bool ObjectEmpty() const { RAPIDJSON_ASSERT(IsObject()); return data_.o.size == 0; } @@ -1083,6 +1169,21 @@ public: /*! \pre IsObject() == true */ MemberIterator MemberEnd() { RAPIDJSON_ASSERT(IsObject()); return MemberIterator(GetMembersPointer() + data_.o.size); } + //! Request the object to have enough capacity to store members. + /*! \param newCapacity The capacity that the object at least need to have. + \param allocator Allocator for reallocating memory. It must be the same one as used before. Commonly use GenericDocument::GetAllocator(). + \return The value itself for fluent API. + \note Linear time complexity. + */ + GenericValue& MemberReserve(SizeType newCapacity, Allocator &allocator) { + RAPIDJSON_ASSERT(IsObject()); + if (newCapacity > data_.o.capacity) { + SetMembersPointer(reinterpret_cast(allocator.Realloc(GetMembersPointer(), data_.o.capacity * sizeof(Member), newCapacity * sizeof(Member)))); + data_.o.capacity = newCapacity; + } + return *this; + } + //! Check whether a member exists in the object. /*! \param name Member name to be searched. @@ -1188,17 +1289,8 @@ public: RAPIDJSON_ASSERT(name.IsString()); ObjectData& o = data_.o; - if (o.size >= o.capacity) { - if (o.capacity == 0) { - o.capacity = kDefaultObjectCapacity; - SetMembersPointer(reinterpret_cast(allocator.Malloc(o.capacity * sizeof(Member)))); - } - else { - SizeType oldCapacity = o.capacity; - o.capacity += (oldCapacity + 1) / 2; // grow by factor 1.5 - SetMembersPointer(reinterpret_cast(allocator.Realloc(GetMembersPointer(), oldCapacity * sizeof(Member), o.capacity * sizeof(Member)))); - } - } + if (o.size >= o.capacity) + MemberReserve(o.capacity == 0 ? kDefaultObjectCapacity : (o.capacity + (o.capacity + 1) / 2), allocator); Member* members = GetMembersPointer(); members[o.size].name.RawAssign(name); members[o.size].value.RawAssign(value); @@ -1425,7 +1517,7 @@ public: MemberIterator pos = MemberBegin() + (first - MemberBegin()); for (MemberIterator itr = pos; itr != last; ++itr) itr->~Member(); - std::memmove(&*pos, &*last, static_cast(MemberEnd() - last) * sizeof(Member)); + std::memmove(static_cast(&*pos), &*last, static_cast(MemberEnd() - last) * sizeof(Member)); data_.o.size -= static_cast(last - first); return pos; } @@ -1628,8 +1720,8 @@ public: RAPIDJSON_ASSERT(last <= End()); ValueIterator pos = Begin() + (first - Begin()); for (ValueIterator itr = pos; itr != last; ++itr) - itr->~GenericValue(); - std::memmove(pos, last, static_cast(End() - last) * sizeof(GenericValue)); + itr->~GenericValue(); + std::memmove(static_cast(pos), last, static_cast(End() - last) * sizeof(GenericValue)); data_.a.size -= static_cast(last - first); return pos; } @@ -1671,7 +1763,7 @@ public: GenericValue& SetInt64(int64_t i64) { this->~GenericValue(); new (this) GenericValue(i64); return *this; } GenericValue& SetUint64(uint64_t u64) { this->~GenericValue(); new (this) GenericValue(u64); return *this; } GenericValue& SetDouble(double d) { this->~GenericValue(); new (this) GenericValue(d); return *this; } - GenericValue& SetFloat(float f) { this->~GenericValue(); new (this) GenericValue(f); return *this; } + GenericValue& SetFloat(float f) { this->~GenericValue(); new (this) GenericValue(static_cast(f)); return *this; } //@} @@ -1710,7 +1802,7 @@ public: \return The value itself for fluent API. \post IsString() == true && GetString() != s && strcmp(GetString(),s) == 0 && GetStringLength() == length */ - GenericValue& SetString(const Ch* s, SizeType length, Allocator& allocator) { this->~GenericValue(); SetStringRaw(StringRef(s, length), allocator); return *this; } + GenericValue& SetString(const Ch* s, SizeType length, Allocator& allocator) { return SetString(StringRef(s, length), allocator); } //! Set this value as a string by copying from source string. /*! \param s source string. @@ -1718,7 +1810,15 @@ public: \return The value itself for fluent API. \post IsString() == true && GetString() != s && strcmp(GetString(),s) == 0 && GetStringLength() == length */ - GenericValue& SetString(const Ch* s, Allocator& allocator) { return SetString(s, internal::StrLen(s), allocator); } + GenericValue& SetString(const Ch* s, Allocator& allocator) { return SetString(StringRef(s), allocator); } + + //! Set this value as a string by copying from source string. + /*! \param s source string reference + \param allocator Allocator for allocating copied buffer. Commonly use GenericDocument::GetAllocator(). + \return The value itself for fluent API. + \post IsString() == true && GetString() != s.s && strcmp(GetString(),s) == 0 && GetStringLength() == length + */ + GenericValue& SetString(StringRefType s, Allocator& allocator) { this->~GenericValue(); SetStringRaw(s, allocator); return *this; } #if RAPIDJSON_HAS_STDSTRING //! Set this value as a string by copying from source string. @@ -1728,7 +1828,7 @@ public: \post IsString() == true && GetString() != s.data() && strcmp(GetString(),s.data() == 0 && GetStringLength() == s.size() \note Requires the definition of the preprocessor symbol \ref RAPIDJSON_HAS_STDSTRING. */ - GenericValue& SetString(const std::basic_string& s, Allocator& allocator) { return SetString(s.data(), SizeType(s.size()), allocator); } + GenericValue& SetString(const std::basic_string& s, Allocator& allocator) { return SetString(StringRef(s), allocator); } #endif //@} @@ -1936,7 +2036,7 @@ private: if (count) { GenericValue* e = static_cast(allocator.Malloc(count * sizeof(GenericValue))); SetElementsPointer(e); - std::memcpy(e, values, count * sizeof(GenericValue)); + std::memcpy(static_cast(e), values, count * sizeof(GenericValue)); } else SetElementsPointer(0); @@ -1949,7 +2049,7 @@ private: if (count) { Member* m = static_cast(allocator.Malloc(count * sizeof(Member))); SetMembersPointer(m); - std::memcpy(m, members, count * sizeof(Member)); + std::memcpy(static_cast(m), members, count * sizeof(Member)); } else SetMembersPointer(0); @@ -2038,7 +2138,7 @@ public: GenericValue(type), allocator_(allocator), ownAllocator_(0), stack_(stackAllocator, stackCapacity), parseResult_() { if (!allocator_) - ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator()); + ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator)(); } //! Constructor @@ -2051,7 +2151,7 @@ public: allocator_(allocator), ownAllocator_(0), stack_(stackAllocator, stackCapacity), parseResult_() { if (!allocator_) - ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator()); + ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator)(); } #if RAPIDJSON_HAS_CXX11_RVALUE_REFS @@ -2112,6 +2212,10 @@ public: return *this; } + // Allow Swap with ValueType. + // Refer to Effective C++ 3rd Edition/Item 33: Avoid hiding inherited names. + using ValueType::Swap; + //! free-standing swap function helper /*! Helper function to enable support for common swap implementation pattern based on \c std::swap: @@ -2243,7 +2347,7 @@ public: template GenericDocument& Parse(const typename SourceEncoding::Ch* str, size_t length) { RAPIDJSON_ASSERT(!(parseFlags & kParseInsituFlag)); - MemoryStream ms(static_cast(str), length * sizeof(typename SourceEncoding::Ch)); + MemoryStream ms(reinterpret_cast(str), length * sizeof(typename SourceEncoding::Ch)); EncodedInputStream is(ms); ParseStream(is); return *this; @@ -2280,7 +2384,7 @@ public: //!@name Handling parse errors //!@{ - //! Whether a parse error has occured in the last parsing. + //! Whether a parse error has occurred in the last parsing. bool HasParseError() const { return parseResult_.IsError(); } //! Get the \ref ParseErrorCode of last parsing. @@ -2401,35 +2505,6 @@ private: //! GenericDocument with UTF8 encoding typedef GenericDocument > Document; -// defined here due to the dependency on GenericDocument -template -template -inline -GenericValue::GenericValue(const GenericValue& rhs, Allocator& allocator) -{ - switch (rhs.GetType()) { - case kObjectType: - case kArrayType: { // perform deep copy via SAX Handler - GenericDocument d(&allocator); - rhs.Accept(d); - RawAssign(*d.stack_.template Pop(1)); - } - break; - case kStringType: - if (rhs.data_.f.flags == kConstStringFlag) { - data_.f.flags = rhs.data_.f.flags; - data_ = *reinterpret_cast(&rhs.data_); - } else { - SetStringRaw(StringRef(rhs.GetString(), rhs.GetStringLength()), allocator); - } - break; - default: - data_.f.flags = rhs.data_.f.flags; - data_ = *reinterpret_cast(&rhs.data_); - break; - } -} - //! Helper class for accessing Value of array type. /*! Instance of this helper class is obtained by \c GenericValue::GetArray(). @@ -2510,6 +2585,7 @@ public: ~GenericObject() {} SizeType MemberCount() const { return value_.MemberCount(); } + SizeType MemberCapacity() const { return value_.MemberCapacity(); } bool ObjectEmpty() const { return value_.ObjectEmpty(); } template ValueType& operator[](T* name) const { return value_[name]; } template ValueType& operator[](const GenericValue& name) const { return value_[name]; } @@ -2518,6 +2594,7 @@ public: #endif MemberIterator MemberBegin() const { return value_.MemberBegin(); } MemberIterator MemberEnd() const { return value_.MemberEnd(); } + GenericObject MemberReserve(SizeType newCapacity, AllocatorType &allocator) const { value_.MemberReserve(newCapacity, allocator); return *this; } bool HasMember(const Ch* name) const { return value_.HasMember(name); } #if RAPIDJSON_HAS_STDSTRING bool HasMember(const std::basic_string& name) const { return value_.HasMember(name); } @@ -2543,7 +2620,7 @@ public: GenericObject AddMember(StringRefType name, ValueType& value, AllocatorType& allocator) const { value_.AddMember(name, value, allocator); return *this; } GenericObject AddMember(StringRefType name, StringRefType value, AllocatorType& allocator) const { value_.AddMember(name, value, allocator); return *this; } template RAPIDJSON_DISABLEIF_RETURN((internal::OrExpr, internal::IsGenericValue >), (GenericObject)) AddMember(StringRefType name, T value, AllocatorType& allocator) const { value_.AddMember(name, value, allocator); return *this; } - void RemoveAllMembers() { return value_.RemoveAllMembers(); } + void RemoveAllMembers() { value_.RemoveAllMembers(); } bool RemoveMember(const Ch* name) const { return value_.RemoveMember(name); } #if RAPIDJSON_HAS_STDSTRING bool RemoveMember(const std::basic_string& name) const { return value_.RemoveMember(name); } diff --git a/src/3rdparty/rapidjson/encodedstream.h b/src/3rdparty/rapidjson/encodedstream.h index 145068386..223601c05 100644 --- a/src/3rdparty/rapidjson/encodedstream.h +++ b/src/3rdparty/rapidjson/encodedstream.h @@ -200,7 +200,7 @@ private: // xx xx xx xx UTF-8 if (!hasBOM_) { - unsigned pattern = (c[0] ? 1 : 0) | (c[1] ? 2 : 0) | (c[2] ? 4 : 0) | (c[3] ? 8 : 0); + int pattern = (c[0] ? 1 : 0) | (c[1] ? 2 : 0) | (c[2] ? 4 : 0) | (c[3] ? 8 : 0); switch (pattern) { case 0x08: type_ = kUTF32BE; break; case 0x0A: type_ = kUTF16BE; break; diff --git a/src/3rdparty/rapidjson/encodings.h b/src/3rdparty/rapidjson/encodings.h index baa7c2b17..0b2446795 100644 --- a/src/3rdparty/rapidjson/encodings.h +++ b/src/3rdparty/rapidjson/encodings.h @@ -17,7 +17,7 @@ #include "rapidjson.h" -#ifdef _MSC_VER +#if defined(_MSC_VER) && !defined(__clang__) RAPIDJSON_DIAG_PUSH RAPIDJSON_DIAG_OFF(4244) // conversion from 'type1' to 'type2', possible loss of data RAPIDJSON_DIAG_OFF(4702) // unreachable code @@ -144,9 +144,9 @@ struct UTF8 { template static bool Decode(InputStream& is, unsigned* codepoint) { -#define COPY() c = is.Take(); *codepoint = (*codepoint << 6) | (static_cast(c) & 0x3Fu) -#define TRANS(mask) result &= ((GetRange(static_cast(c)) & mask) != 0) -#define TAIL() COPY(); TRANS(0x70) +#define RAPIDJSON_COPY() c = is.Take(); *codepoint = (*codepoint << 6) | (static_cast(c) & 0x3Fu) +#define RAPIDJSON_TRANS(mask) result &= ((GetRange(static_cast(c)) & mask) != 0) +#define RAPIDJSON_TAIL() RAPIDJSON_COPY(); RAPIDJSON_TRANS(0x70) typename InputStream::Ch c = is.Take(); if (!(c & 0x80)) { *codepoint = static_cast(c); @@ -157,48 +157,48 @@ struct UTF8 { if (type >= 32) { *codepoint = 0; } else { - *codepoint = (0xFF >> type) & static_cast(c); + *codepoint = (0xFFu >> type) & static_cast(c); } bool result = true; switch (type) { - case 2: TAIL(); return result; - case 3: TAIL(); TAIL(); return result; - case 4: COPY(); TRANS(0x50); TAIL(); return result; - case 5: COPY(); TRANS(0x10); TAIL(); TAIL(); return result; - case 6: TAIL(); TAIL(); TAIL(); return result; - case 10: COPY(); TRANS(0x20); TAIL(); return result; - case 11: COPY(); TRANS(0x60); TAIL(); TAIL(); return result; + case 2: RAPIDJSON_TAIL(); return result; + case 3: RAPIDJSON_TAIL(); RAPIDJSON_TAIL(); return result; + case 4: RAPIDJSON_COPY(); RAPIDJSON_TRANS(0x50); RAPIDJSON_TAIL(); return result; + case 5: RAPIDJSON_COPY(); RAPIDJSON_TRANS(0x10); RAPIDJSON_TAIL(); RAPIDJSON_TAIL(); return result; + case 6: RAPIDJSON_TAIL(); RAPIDJSON_TAIL(); RAPIDJSON_TAIL(); return result; + case 10: RAPIDJSON_COPY(); RAPIDJSON_TRANS(0x20); RAPIDJSON_TAIL(); return result; + case 11: RAPIDJSON_COPY(); RAPIDJSON_TRANS(0x60); RAPIDJSON_TAIL(); RAPIDJSON_TAIL(); return result; default: return false; } -#undef COPY -#undef TRANS -#undef TAIL +#undef RAPIDJSON_COPY +#undef RAPIDJSON_TRANS +#undef RAPIDJSON_TAIL } template static bool Validate(InputStream& is, OutputStream& os) { -#define COPY() os.Put(c = is.Take()) -#define TRANS(mask) result &= ((GetRange(static_cast(c)) & mask) != 0) -#define TAIL() COPY(); TRANS(0x70) +#define RAPIDJSON_COPY() os.Put(c = is.Take()) +#define RAPIDJSON_TRANS(mask) result &= ((GetRange(static_cast(c)) & mask) != 0) +#define RAPIDJSON_TAIL() RAPIDJSON_COPY(); RAPIDJSON_TRANS(0x70) Ch c; - COPY(); + RAPIDJSON_COPY(); if (!(c & 0x80)) return true; bool result = true; switch (GetRange(static_cast(c))) { - case 2: TAIL(); return result; - case 3: TAIL(); TAIL(); return result; - case 4: COPY(); TRANS(0x50); TAIL(); return result; - case 5: COPY(); TRANS(0x10); TAIL(); TAIL(); return result; - case 6: TAIL(); TAIL(); TAIL(); return result; - case 10: COPY(); TRANS(0x20); TAIL(); return result; - case 11: COPY(); TRANS(0x60); TAIL(); TAIL(); return result; + case 2: RAPIDJSON_TAIL(); return result; + case 3: RAPIDJSON_TAIL(); RAPIDJSON_TAIL(); return result; + case 4: RAPIDJSON_COPY(); RAPIDJSON_TRANS(0x50); RAPIDJSON_TAIL(); return result; + case 5: RAPIDJSON_COPY(); RAPIDJSON_TRANS(0x10); RAPIDJSON_TAIL(); RAPIDJSON_TAIL(); return result; + case 6: RAPIDJSON_TAIL(); RAPIDJSON_TAIL(); RAPIDJSON_TAIL(); return result; + case 10: RAPIDJSON_COPY(); RAPIDJSON_TRANS(0x20); RAPIDJSON_TAIL(); return result; + case 11: RAPIDJSON_COPY(); RAPIDJSON_TRANS(0x60); RAPIDJSON_TAIL(); RAPIDJSON_TAIL(); return result; default: return false; } -#undef COPY -#undef TRANS -#undef TAIL +#undef RAPIDJSON_COPY +#undef RAPIDJSON_TRANS +#undef RAPIDJSON_TAIL } static unsigned char GetRange(unsigned char c) { @@ -283,7 +283,7 @@ struct UTF16 { RAPIDJSON_ASSERT(codepoint <= 0x10FFFF); unsigned v = codepoint - 0x10000; os.Put(static_cast((v >> 10) | 0xD800)); - os.Put((v & 0x3FF) | 0xDC00); + os.Put(static_cast((v & 0x3FF) | 0xDC00)); } } @@ -299,7 +299,7 @@ struct UTF16 { RAPIDJSON_ASSERT(codepoint <= 0x10FFFF); unsigned v = codepoint - 0x10000; PutUnsafe(os, static_cast((v >> 10) | 0xD800)); - PutUnsafe(os, (v & 0x3FF) | 0xDC00); + PutUnsafe(os, static_cast((v & 0x3FF) | 0xDC00)); } } @@ -384,7 +384,7 @@ struct UTF16BE : UTF16 { static CharType Take(InputByteStream& is) { RAPIDJSON_STATIC_ASSERT(sizeof(typename InputByteStream::Ch) == 1); unsigned c = static_cast(static_cast(is.Take())) << 8; - c |= static_cast(is.Take()); + c |= static_cast(static_cast(is.Take())); return static_cast(c); } @@ -620,28 +620,28 @@ struct AutoUTF { #define RAPIDJSON_ENCODINGS_FUNC(x) UTF8::x, UTF16LE::x, UTF16BE::x, UTF32LE::x, UTF32BE::x template - RAPIDJSON_FORCEINLINE static void Encode(OutputStream& os, unsigned codepoint) { + static RAPIDJSON_FORCEINLINE void Encode(OutputStream& os, unsigned codepoint) { typedef void (*EncodeFunc)(OutputStream&, unsigned); static const EncodeFunc f[] = { RAPIDJSON_ENCODINGS_FUNC(Encode) }; (*f[os.GetType()])(os, codepoint); } template - RAPIDJSON_FORCEINLINE static void EncodeUnsafe(OutputStream& os, unsigned codepoint) { + static RAPIDJSON_FORCEINLINE void EncodeUnsafe(OutputStream& os, unsigned codepoint) { typedef void (*EncodeFunc)(OutputStream&, unsigned); static const EncodeFunc f[] = { RAPIDJSON_ENCODINGS_FUNC(EncodeUnsafe) }; (*f[os.GetType()])(os, codepoint); } template - RAPIDJSON_FORCEINLINE static bool Decode(InputStream& is, unsigned* codepoint) { + static RAPIDJSON_FORCEINLINE bool Decode(InputStream& is, unsigned* codepoint) { typedef bool (*DecodeFunc)(InputStream&, unsigned*); static const DecodeFunc f[] = { RAPIDJSON_ENCODINGS_FUNC(Decode) }; return (*f[is.GetType()])(is, codepoint); } template - RAPIDJSON_FORCEINLINE static bool Validate(InputStream& is, OutputStream& os) { + static RAPIDJSON_FORCEINLINE bool Validate(InputStream& is, OutputStream& os) { typedef bool (*ValidateFunc)(InputStream&, OutputStream&); static const ValidateFunc f[] = { RAPIDJSON_ENCODINGS_FUNC(Validate) }; return (*f[is.GetType()])(is, os); @@ -658,7 +658,7 @@ template struct Transcoder { //! Take one Unicode codepoint from source encoding, convert it to target encoding and put it to the output stream. template - RAPIDJSON_FORCEINLINE static bool Transcode(InputStream& is, OutputStream& os) { + static RAPIDJSON_FORCEINLINE bool Transcode(InputStream& is, OutputStream& os) { unsigned codepoint; if (!SourceEncoding::Decode(is, &codepoint)) return false; @@ -667,7 +667,7 @@ struct Transcoder { } template - RAPIDJSON_FORCEINLINE static bool TranscodeUnsafe(InputStream& is, OutputStream& os) { + static RAPIDJSON_FORCEINLINE bool TranscodeUnsafe(InputStream& is, OutputStream& os) { unsigned codepoint; if (!SourceEncoding::Decode(is, &codepoint)) return false; @@ -677,7 +677,7 @@ struct Transcoder { //! Validate one Unicode codepoint from an encoded stream. template - RAPIDJSON_FORCEINLINE static bool Validate(InputStream& is, OutputStream& os) { + static RAPIDJSON_FORCEINLINE bool Validate(InputStream& is, OutputStream& os) { return Transcode(is, os); // Since source/target encoding is different, must transcode. } }; @@ -690,26 +690,26 @@ inline void PutUnsafe(Stream& stream, typename Stream::Ch c); template struct Transcoder { template - RAPIDJSON_FORCEINLINE static bool Transcode(InputStream& is, OutputStream& os) { + static RAPIDJSON_FORCEINLINE bool Transcode(InputStream& is, OutputStream& os) { os.Put(is.Take()); // Just copy one code unit. This semantic is different from primary template class. return true; } template - RAPIDJSON_FORCEINLINE static bool TranscodeUnsafe(InputStream& is, OutputStream& os) { + static RAPIDJSON_FORCEINLINE bool TranscodeUnsafe(InputStream& is, OutputStream& os) { PutUnsafe(os, is.Take()); // Just copy one code unit. This semantic is different from primary template class. return true; } template - RAPIDJSON_FORCEINLINE static bool Validate(InputStream& is, OutputStream& os) { + static RAPIDJSON_FORCEINLINE bool Validate(InputStream& is, OutputStream& os) { return Encoding::Validate(is, os); // source/target encoding are the same } }; RAPIDJSON_NAMESPACE_END -#if defined(__GNUC__) || defined(_MSC_VER) +#if defined(__GNUC__) || (defined(_MSC_VER) && !defined(__clang__)) RAPIDJSON_DIAG_POP #endif diff --git a/src/3rdparty/rapidjson/error/error.h b/src/3rdparty/rapidjson/error/error.h index 95cb31a72..9311d2f03 100644 --- a/src/3rdparty/rapidjson/error/error.h +++ b/src/3rdparty/rapidjson/error/error.h @@ -104,6 +104,8 @@ enum ParseErrorCode { \see GenericReader::Parse, GenericDocument::Parse */ struct ParseResult { + //!! Unspecified boolean type + typedef bool (ParseResult::*BooleanType)() const; public: //! Default constructor, no error. ParseResult() : code_(kParseErrorNone), offset_(0) {} @@ -115,8 +117,8 @@ public: //! Get the error offset, if \ref IsError(), 0 otherwise. size_t Offset() const { return offset_; } - //! Conversion to \c bool, returns \c true, iff !\ref IsError(). - operator bool() const { return !IsError(); } + //! Explicit conversion to \c bool, returns \c true, iff !\ref IsError(). + operator BooleanType() const { return !IsError() ? &ParseResult::IsError : NULL; } //! Whether the result is an error. bool IsError() const { return code_ != kParseErrorNone; } @@ -124,6 +126,10 @@ public: bool operator==(ParseErrorCode code) const { return code_ == code; } friend bool operator==(ParseErrorCode code, const ParseResult & err) { return code == err.code_; } + bool operator!=(const ParseResult& that) const { return !(*this == that); } + bool operator!=(ParseErrorCode code) const { return !(*this == code); } + friend bool operator!=(ParseErrorCode code, const ParseResult & err) { return err != code; } + //! Reset error code. void Clear() { Set(kParseErrorNone); } //! Update error code and offset. diff --git a/src/3rdparty/rapidjson/filereadstream.h b/src/3rdparty/rapidjson/filereadstream.h index b56ea13b3..6b343707a 100644 --- a/src/3rdparty/rapidjson/filereadstream.h +++ b/src/3rdparty/rapidjson/filereadstream.h @@ -59,7 +59,7 @@ public: // For encoding detection only. const Ch* Peek4() const { - return (current_ + 4 <= bufferLast_) ? current_ : 0; + return (current_ + 4 - !eof_ <= bufferLast_) ? current_ : 0; } private: @@ -68,7 +68,7 @@ private: ++current_; else if (!eof_) { count_ += readCount_; - readCount_ = fread(buffer_, 1, bufferSize_, fp_); + readCount_ = std::fread(buffer_, 1, bufferSize_, fp_); bufferLast_ = buffer_ + readCount_ - 1; current_ = buffer_; diff --git a/src/3rdparty/rapidjson/filewritestream.h b/src/3rdparty/rapidjson/filewritestream.h index 6378dd60e..8b48fee19 100644 --- a/src/3rdparty/rapidjson/filewritestream.h +++ b/src/3rdparty/rapidjson/filewritestream.h @@ -25,7 +25,7 @@ RAPIDJSON_DIAG_OFF(unreachable-code) RAPIDJSON_NAMESPACE_BEGIN -//! Wrapper of C file stream for input using fread(). +//! Wrapper of C file stream for output using fwrite(). /*! \note implements Stream concept */ @@ -62,7 +62,7 @@ public: void Flush() { if (current_ != buffer_) { - size_t result = fwrite(buffer_, 1, static_cast(current_ - buffer_), fp_); + size_t result = std::fwrite(buffer_, 1, static_cast(current_ - buffer_), fp_); if (result < static_cast(current_ - buffer_)) { // failure deliberately ignored at this time // added to avoid warn_unused_result build errors diff --git a/src/3rdparty/rapidjson/internal/biginteger.h b/src/3rdparty/rapidjson/internal/biginteger.h index 9d3e88c99..a31c8a88d 100644 --- a/src/3rdparty/rapidjson/internal/biginteger.h +++ b/src/3rdparty/rapidjson/internal/biginteger.h @@ -17,7 +17,7 @@ #include "../rapidjson.h" -#if defined(_MSC_VER) && defined(_M_AMD64) +#if defined(_MSC_VER) && !__INTEL_COMPILER && defined(_M_AMD64) #include // for _umul128 #pragma intrinsic(_umul128) #endif @@ -133,7 +133,7 @@ public: RAPIDJSON_ASSERT(count_ + offset <= kCapacity); if (interShift == 0) { - std::memmove(&digits_[count_ - 1 + offset], &digits_[count_ - 1], count_ * sizeof(Type)); + std::memmove(digits_ + offset, digits_, count_ * sizeof(Type)); count_ += offset; } else { diff --git a/src/3rdparty/rapidjson/internal/diyfp.h b/src/3rdparty/rapidjson/internal/diyfp.h index c9fefdc61..b6c2cf561 100644 --- a/src/3rdparty/rapidjson/internal/diyfp.h +++ b/src/3rdparty/rapidjson/internal/diyfp.h @@ -1,5 +1,5 @@ // Tencent is pleased to support the open source community by making RapidJSON available. -// +// // Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved. // // Licensed under the MIT License (the "License"); you may not use this file except @@ -7,9 +7,9 @@ // // http://opensource.org/licenses/MIT // -// Unless required by applicable law or agreed to in writing, software distributed -// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR -// CONDITIONS OF ANY KIND, either express or implied. See the License for the +// Unless required by applicable law or agreed to in writing, software distributed +// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR +// CONDITIONS OF ANY KIND, either express or implied. See the License for the // specific language governing permissions and limitations under the License. // This is a C++ header-only implementation of Grisu2 algorithm from the publication: @@ -20,8 +20,9 @@ #define RAPIDJSON_DIYFP_H_ #include "../rapidjson.h" +#include -#if defined(_MSC_VER) && defined(_M_AMD64) +#if defined(_MSC_VER) && defined(_M_AMD64) && !defined(__INTEL_COMPILER) #include #pragma intrinsic(_BitScanReverse64) #pragma intrinsic(_umul128) @@ -56,7 +57,7 @@ struct DiyFp { if (biased_e != 0) { f = significand + kDpHiddenBit; e = biased_e - kDpExponentBias; - } + } else { f = significand; e = kDpMinExponent + 1; @@ -99,6 +100,7 @@ struct DiyFp { } DiyFp Normalize() const { + RAPIDJSON_ASSERT(f != 0); // https://stackoverflow.com/a/26809183/291737 #if defined(_MSC_VER) && defined(_M_AMD64) unsigned long index; _BitScanReverse64(&index, f); @@ -141,7 +143,16 @@ struct DiyFp { double d; uint64_t u64; }u; - const uint64_t be = (e == kDpDenormalExponent && (f & kDpHiddenBit) == 0) ? 0 : + RAPIDJSON_ASSERT(f <= kDpHiddenBit + kDpSignificandMask); + if (e < kDpDenormalExponent) { + // Underflow. + return 0.0; + } + if (e >= kDpMaxExponent) { + // Overflow. + return std::numeric_limits::infinity(); + } + const uint64_t be = (e == kDpDenormalExponent && (f & kDpHiddenBit) == 0) ? 0 : static_cast(e + kDpExponentBias); u.u64 = (f & kDpSignificandMask) | (be << kDpSignificandSize); return u.d; @@ -220,9 +231,10 @@ inline DiyFp GetCachedPowerByIndex(size_t index) { 641, 667, 694, 720, 747, 774, 800, 827, 853, 880, 907, 933, 960, 986, 1013, 1039, 1066 }; + RAPIDJSON_ASSERT(index < 87); return DiyFp(kCachedPowers_F[index], kCachedPowers_E[index]); } - + inline DiyFp GetCachedPower(int e, int* K) { //int k = static_cast(ceil((-61 - e) * 0.30102999566398114)) + 374; @@ -238,10 +250,11 @@ inline DiyFp GetCachedPower(int e, int* K) { } inline DiyFp GetCachedPower10(int exp, int *outExp) { - unsigned index = (static_cast(exp) + 348u) / 8u; - *outExp = -348 + static_cast(index) * 8; - return GetCachedPowerByIndex(index); - } + RAPIDJSON_ASSERT(exp >= -348); + unsigned index = static_cast(exp + 348) / 8u; + *outExp = -348 + static_cast(index) * 8; + return GetCachedPowerByIndex(index); +} #ifdef __GNUC__ RAPIDJSON_DIAG_POP diff --git a/src/3rdparty/rapidjson/internal/dtoa.h b/src/3rdparty/rapidjson/internal/dtoa.h index 8d6350e62..bf2e9b2e5 100644 --- a/src/3rdparty/rapidjson/internal/dtoa.h +++ b/src/3rdparty/rapidjson/internal/dtoa.h @@ -41,7 +41,7 @@ inline void GrisuRound(char* buffer, int len, uint64_t delta, uint64_t rest, uin } } -inline unsigned CountDecimalDigit32(uint32_t n) { +inline int CountDecimalDigit32(uint32_t n) { // Simple pure C++ implementation was faster than __builtin_clz version in this situation. if (n < 10) return 1; if (n < 100) return 2; @@ -63,7 +63,7 @@ inline void DigitGen(const DiyFp& W, const DiyFp& Mp, uint64_t delta, char* buff const DiyFp wp_w = Mp - W; uint32_t p1 = static_cast(Mp.f >> -one.e); uint64_t p2 = Mp.f & (one.f - 1); - unsigned kappa = CountDecimalDigit32(p1); // kappa in [0, 9] + int kappa = CountDecimalDigit32(p1); // kappa in [0, 9] *len = 0; while (kappa > 0) { @@ -102,8 +102,8 @@ inline void DigitGen(const DiyFp& W, const DiyFp& Mp, uint64_t delta, char* buff kappa--; if (p2 < delta) { *K += kappa; - int index = -static_cast(kappa); - GrisuRound(buffer, *len, delta, p2, one.f, wp_w.f * (index < 9 ? kPow10[-static_cast(kappa)] : 0)); + int index = -kappa; + GrisuRound(buffer, *len, delta, p2, one.f, wp_w.f * (index < 9 ? kPow10[index] : 0)); return; } } diff --git a/src/3rdparty/rapidjson/internal/ieee754.h b/src/3rdparty/rapidjson/internal/ieee754.h index 82bb0b99e..c2684ba2a 100644 --- a/src/3rdparty/rapidjson/internal/ieee754.h +++ b/src/3rdparty/rapidjson/internal/ieee754.h @@ -48,13 +48,13 @@ public: int IntegerExponent() const { return (IsNormal() ? Exponent() : kDenormalExponent) - kSignificandSize; } uint64_t ToBias() const { return (u_ & kSignMask) ? ~u_ + 1 : u_ | kSignMask; } - static unsigned EffectiveSignificandSize(int order) { + static int EffectiveSignificandSize(int order) { if (order >= -1021) return 53; else if (order <= -1074) return 0; else - return static_cast(order) + 1074; + return order + 1074; } private: diff --git a/src/3rdparty/rapidjson/internal/itoa.h b/src/3rdparty/rapidjson/internal/itoa.h index 01a4e7e72..9b1c45cc1 100644 --- a/src/3rdparty/rapidjson/internal/itoa.h +++ b/src/3rdparty/rapidjson/internal/itoa.h @@ -1,5 +1,5 @@ // Tencent is pleased to support the open source community by making RapidJSON available. -// +// // Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved. // // Licensed under the MIT License (the "License"); you may not use this file except @@ -7,9 +7,9 @@ // // http://opensource.org/licenses/MIT // -// Unless required by applicable law or agreed to in writing, software distributed -// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR -// CONDITIONS OF ANY KIND, either express or implied. See the License for the +// Unless required by applicable law or agreed to in writing, software distributed +// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR +// CONDITIONS OF ANY KIND, either express or implied. See the License for the // specific language governing permissions and limitations under the License. #ifndef RAPIDJSON_ITOA_ @@ -37,12 +37,14 @@ inline const char* GetDigitsLut() { } inline char* u32toa(uint32_t value, char* buffer) { + RAPIDJSON_ASSERT(buffer != 0); + const char* cDigitsLut = GetDigitsLut(); if (value < 10000) { const uint32_t d1 = (value / 100) << 1; const uint32_t d2 = (value % 100) << 1; - + if (value >= 1000) *buffer++ = cDigitsLut[d1]; if (value >= 100) @@ -55,13 +57,13 @@ inline char* u32toa(uint32_t value, char* buffer) { // value = bbbbcccc const uint32_t b = value / 10000; const uint32_t c = value % 10000; - + const uint32_t d1 = (b / 100) << 1; const uint32_t d2 = (b % 100) << 1; - + const uint32_t d3 = (c / 100) << 1; const uint32_t d4 = (c % 100) << 1; - + if (value >= 10000000) *buffer++ = cDigitsLut[d1]; if (value >= 1000000) @@ -69,7 +71,7 @@ inline char* u32toa(uint32_t value, char* buffer) { if (value >= 100000) *buffer++ = cDigitsLut[d2]; *buffer++ = cDigitsLut[d2 + 1]; - + *buffer++ = cDigitsLut[d3]; *buffer++ = cDigitsLut[d3 + 1]; *buffer++ = cDigitsLut[d4]; @@ -77,10 +79,10 @@ inline char* u32toa(uint32_t value, char* buffer) { } else { // value = aabbbbcccc in decimal - + const uint32_t a = value / 100000000; // 1 to 42 value %= 100000000; - + if (a >= 10) { const unsigned i = a << 1; *buffer++ = cDigitsLut[i]; @@ -91,13 +93,13 @@ inline char* u32toa(uint32_t value, char* buffer) { const uint32_t b = value / 10000; // 0 to 9999 const uint32_t c = value % 10000; // 0 to 9999 - + const uint32_t d1 = (b / 100) << 1; const uint32_t d2 = (b % 100) << 1; - + const uint32_t d3 = (c / 100) << 1; const uint32_t d4 = (c % 100) << 1; - + *buffer++ = cDigitsLut[d1]; *buffer++ = cDigitsLut[d1 + 1]; *buffer++ = cDigitsLut[d2]; @@ -111,6 +113,7 @@ inline char* u32toa(uint32_t value, char* buffer) { } inline char* i32toa(int32_t value, char* buffer) { + RAPIDJSON_ASSERT(buffer != 0); uint32_t u = static_cast(value); if (value < 0) { *buffer++ = '-'; @@ -121,6 +124,7 @@ inline char* i32toa(int32_t value, char* buffer) { } inline char* u64toa(uint64_t value, char* buffer) { + RAPIDJSON_ASSERT(buffer != 0); const char* cDigitsLut = GetDigitsLut(); const uint64_t kTen8 = 100000000; const uint64_t kTen9 = kTen8 * 10; @@ -131,13 +135,13 @@ inline char* u64toa(uint64_t value, char* buffer) { const uint64_t kTen14 = kTen8 * 1000000; const uint64_t kTen15 = kTen8 * 10000000; const uint64_t kTen16 = kTen8 * kTen8; - + if (value < kTen8) { uint32_t v = static_cast(value); if (v < 10000) { const uint32_t d1 = (v / 100) << 1; const uint32_t d2 = (v % 100) << 1; - + if (v >= 1000) *buffer++ = cDigitsLut[d1]; if (v >= 100) @@ -150,13 +154,13 @@ inline char* u64toa(uint64_t value, char* buffer) { // value = bbbbcccc const uint32_t b = v / 10000; const uint32_t c = v % 10000; - + const uint32_t d1 = (b / 100) << 1; const uint32_t d2 = (b % 100) << 1; - + const uint32_t d3 = (c / 100) << 1; const uint32_t d4 = (c % 100) << 1; - + if (value >= 10000000) *buffer++ = cDigitsLut[d1]; if (value >= 1000000) @@ -164,7 +168,7 @@ inline char* u64toa(uint64_t value, char* buffer) { if (value >= 100000) *buffer++ = cDigitsLut[d2]; *buffer++ = cDigitsLut[d2 + 1]; - + *buffer++ = cDigitsLut[d3]; *buffer++ = cDigitsLut[d3 + 1]; *buffer++ = cDigitsLut[d4]; @@ -174,22 +178,22 @@ inline char* u64toa(uint64_t value, char* buffer) { else if (value < kTen16) { const uint32_t v0 = static_cast(value / kTen8); const uint32_t v1 = static_cast(value % kTen8); - + const uint32_t b0 = v0 / 10000; const uint32_t c0 = v0 % 10000; - + const uint32_t d1 = (b0 / 100) << 1; const uint32_t d2 = (b0 % 100) << 1; - + const uint32_t d3 = (c0 / 100) << 1; const uint32_t d4 = (c0 % 100) << 1; const uint32_t b1 = v1 / 10000; const uint32_t c1 = v1 % 10000; - + const uint32_t d5 = (b1 / 100) << 1; const uint32_t d6 = (b1 % 100) << 1; - + const uint32_t d7 = (c1 / 100) << 1; const uint32_t d8 = (c1 % 100) << 1; @@ -207,9 +211,8 @@ inline char* u64toa(uint64_t value, char* buffer) { *buffer++ = cDigitsLut[d3 + 1]; if (value >= kTen9) *buffer++ = cDigitsLut[d4]; - if (value >= kTen8) - *buffer++ = cDigitsLut[d4 + 1]; - + + *buffer++ = cDigitsLut[d4 + 1]; *buffer++ = cDigitsLut[d5]; *buffer++ = cDigitsLut[d5 + 1]; *buffer++ = cDigitsLut[d6]; @@ -222,7 +225,7 @@ inline char* u64toa(uint64_t value, char* buffer) { else { const uint32_t a = static_cast(value / kTen16); // 1 to 1844 value %= kTen16; - + if (a < 10) *buffer++ = static_cast('0' + static_cast(a)); else if (a < 100) { @@ -232,7 +235,7 @@ inline char* u64toa(uint64_t value, char* buffer) { } else if (a < 1000) { *buffer++ = static_cast('0' + static_cast(a / 100)); - + const uint32_t i = (a % 100) << 1; *buffer++ = cDigitsLut[i]; *buffer++ = cDigitsLut[i + 1]; @@ -245,28 +248,28 @@ inline char* u64toa(uint64_t value, char* buffer) { *buffer++ = cDigitsLut[j]; *buffer++ = cDigitsLut[j + 1]; } - + const uint32_t v0 = static_cast(value / kTen8); const uint32_t v1 = static_cast(value % kTen8); - + const uint32_t b0 = v0 / 10000; const uint32_t c0 = v0 % 10000; - + const uint32_t d1 = (b0 / 100) << 1; const uint32_t d2 = (b0 % 100) << 1; - + const uint32_t d3 = (c0 / 100) << 1; const uint32_t d4 = (c0 % 100) << 1; - + const uint32_t b1 = v1 / 10000; const uint32_t c1 = v1 % 10000; - + const uint32_t d5 = (b1 / 100) << 1; const uint32_t d6 = (b1 % 100) << 1; - + const uint32_t d7 = (c1 / 100) << 1; const uint32_t d8 = (c1 % 100) << 1; - + *buffer++ = cDigitsLut[d1]; *buffer++ = cDigitsLut[d1 + 1]; *buffer++ = cDigitsLut[d2]; @@ -284,11 +287,12 @@ inline char* u64toa(uint64_t value, char* buffer) { *buffer++ = cDigitsLut[d8]; *buffer++ = cDigitsLut[d8 + 1]; } - + return buffer; } inline char* i64toa(int64_t value, char* buffer) { + RAPIDJSON_ASSERT(buffer != 0); uint64_t u = static_cast(value); if (value < 0) { *buffer++ = '-'; diff --git a/src/3rdparty/rapidjson/internal/meta.h b/src/3rdparty/rapidjson/internal/meta.h index 5a9aaa428..d401edf85 100644 --- a/src/3rdparty/rapidjson/internal/meta.h +++ b/src/3rdparty/rapidjson/internal/meta.h @@ -21,7 +21,8 @@ RAPIDJSON_DIAG_PUSH RAPIDJSON_DIAG_OFF(effc++) #endif -#if defined(_MSC_VER) + +#if defined(_MSC_VER) && !defined(__clang__) RAPIDJSON_DIAG_PUSH RAPIDJSON_DIAG_OFF(6334) #endif @@ -174,7 +175,11 @@ template struct RemoveSfinaeTag { typedef T Type; RAPIDJSON_NAMESPACE_END //@endcond -#if defined(__GNUC__) || defined(_MSC_VER) +#if defined(_MSC_VER) && !defined(__clang__) +RAPIDJSON_DIAG_POP +#endif + +#ifdef __GNUC__ RAPIDJSON_DIAG_POP #endif diff --git a/src/3rdparty/rapidjson/internal/regex.h b/src/3rdparty/rapidjson/internal/regex.h index 422a5240b..16e355921 100644 --- a/src/3rdparty/rapidjson/internal/regex.h +++ b/src/3rdparty/rapidjson/internal/regex.h @@ -24,16 +24,17 @@ RAPIDJSON_DIAG_PUSH RAPIDJSON_DIAG_OFF(padded) RAPIDJSON_DIAG_OFF(switch-enum) RAPIDJSON_DIAG_OFF(implicit-fallthrough) +#elif defined(_MSC_VER) +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(4512) // assignment operator could not be generated #endif #ifdef __GNUC__ RAPIDJSON_DIAG_PUSH RAPIDJSON_DIAG_OFF(effc++) +#if __GNUC__ >= 7 +RAPIDJSON_DIAG_OFF(implicit-fallthrough) #endif - -#ifdef _MSC_VER -RAPIDJSON_DIAG_PUSH -RAPIDJSON_DIAG_OFF(4512) // assignment operator could not be generated #endif #ifndef RAPIDJSON_REGEX_VERBOSE @@ -43,12 +44,40 @@ RAPIDJSON_DIAG_OFF(4512) // assignment operator could not be generated RAPIDJSON_NAMESPACE_BEGIN namespace internal { +/////////////////////////////////////////////////////////////////////////////// +// DecodedStream + +template +class DecodedStream { +public: + DecodedStream(SourceStream& ss) : ss_(ss), codepoint_() { Decode(); } + unsigned Peek() { return codepoint_; } + unsigned Take() { + unsigned c = codepoint_; + if (c) // No further decoding when '\0' + Decode(); + return c; + } + +private: + void Decode() { + if (!Encoding::Decode(ss_, &codepoint_)) + codepoint_ = 0; + } + + SourceStream& ss_; + unsigned codepoint_; +}; + /////////////////////////////////////////////////////////////////////////////// // GenericRegex static const SizeType kRegexInvalidState = ~SizeType(0); //!< Represents an invalid index in GenericRegex::State::out, out1 static const SizeType kRegexInvalidRange = ~SizeType(0); +template +class GenericRegexSearch; + //! Regular expression engine with subset of ECMAscript grammar. /*! Supported regular expression syntax: @@ -84,45 +113,29 @@ static const SizeType kRegexInvalidRange = ~SizeType(0); template class GenericRegex { public: + typedef Encoding EncodingType; typedef typename Encoding::Ch Ch; + template friend class GenericRegexSearch; GenericRegex(const Ch* source, Allocator* allocator = 0) : - states_(allocator, 256), ranges_(allocator, 256), root_(kRegexInvalidState), stateCount_(), rangeCount_(), - stateSet_(), state0_(allocator, 0), state1_(allocator, 0), anchorBegin_(), anchorEnd_() + ownAllocator_(allocator ? 0 : RAPIDJSON_NEW(Allocator)()), allocator_(allocator ? allocator : ownAllocator_), + states_(allocator_, 256), ranges_(allocator_, 256), root_(kRegexInvalidState), stateCount_(), rangeCount_(), + anchorBegin_(), anchorEnd_() { GenericStringStream ss(source); - DecodedStream > ds(ss); + DecodedStream, Encoding> ds(ss); Parse(ds); } - ~GenericRegex() { - Allocator::Free(stateSet_); + ~GenericRegex() + { + RAPIDJSON_DELETE(ownAllocator_); } bool IsValid() const { return root_ != kRegexInvalidState; } - template - bool Match(InputStream& is) const { - return SearchWithAnchoring(is, true, true); - } - - bool Match(const Ch* s) const { - GenericStringStream is(s); - return Match(is); - } - - template - bool Search(InputStream& is) const { - return SearchWithAnchoring(is, anchorBegin_, anchorEnd_); - } - - bool Search(const Ch* s) const { - GenericStringStream is(s); - return Search(is); - } - private: enum Operator { kZeroOrOne, @@ -157,28 +170,6 @@ private: SizeType minIndex; }; - template - class DecodedStream { - public: - DecodedStream(SourceStream& ss) : ss_(ss), codepoint_() { Decode(); } - unsigned Peek() { return codepoint_; } - unsigned Take() { - unsigned c = codepoint_; - if (c) // No further decoding when '\0' - Decode(); - return c; - } - - private: - void Decode() { - if (!Encoding::Decode(ss_, &codepoint_)) - codepoint_ = 0; - } - - SourceStream& ss_; - unsigned codepoint_; - }; - State& GetState(SizeType index) { RAPIDJSON_ASSERT(index < stateCount_); return states_.template Bottom()[index]; @@ -200,11 +191,10 @@ private: } template - void Parse(DecodedStream& ds) { - Allocator allocator; - Stack operandStack(&allocator, 256); // Frag - Stack operatorStack(&allocator, 256); // Operator - Stack atomCountStack(&allocator, 256); // unsigned (Atom per parenthesis) + void Parse(DecodedStream& ds) { + Stack operandStack(allocator_, 256); // Frag + Stack operatorStack(allocator_, 256); // Operator + Stack atomCountStack(allocator_, 256); // unsigned (Atom per parenthesis) *atomCountStack.template Push() = 0; @@ -327,14 +317,6 @@ private: printf("\n"); #endif } - - // Preallocate buffer for SearchWithAnchoring() - RAPIDJSON_ASSERT(stateSet_ == 0); - if (stateCount_ > 0) { - stateSet_ = static_cast(states_.GetAllocator().Malloc(GetStateSetSize())); - state0_.template Reserve(stateCount_); - state1_.template Reserve(stateCount_); - } } SizeType NewState(SizeType out, SizeType out1, unsigned codepoint) { @@ -413,8 +395,7 @@ private: } return false; - default: - RAPIDJSON_ASSERT(op == kOneOrMore); + case kOneOrMore: if (operandStack.GetSize() >= sizeof(Frag)) { Frag e = *operandStack.template Pop(1); SizeType s = NewState(kRegexInvalidState, e.start, 0); @@ -423,6 +404,10 @@ private: return true; } return false; + + default: + // syntax error (e.g. unclosed kLeftParenthesis) + return false; } } @@ -483,7 +468,7 @@ private: } template - bool ParseUnsigned(DecodedStream& ds, unsigned* u) { + bool ParseUnsigned(DecodedStream& ds, unsigned* u) { unsigned r = 0; if (ds.Peek() < '0' || ds.Peek() > '9') return false; @@ -497,7 +482,7 @@ private: } template - bool ParseRange(DecodedStream& ds, SizeType* range) { + bool ParseRange(DecodedStream& ds, SizeType* range) { bool isBegin = true; bool negate = false; int step = 0; @@ -575,7 +560,7 @@ private: } template - bool CharacterEscape(DecodedStream& ds, unsigned* escapedCodepoint) { + bool CharacterEscape(DecodedStream& ds, unsigned* escapedCodepoint) { unsigned codepoint; switch (codepoint = ds.Take()) { case '^': @@ -603,72 +588,8 @@ private: } } - template - bool SearchWithAnchoring(InputStream& is, bool anchorBegin, bool anchorEnd) const { - RAPIDJSON_ASSERT(IsValid()); - DecodedStream ds(is); - - state0_.Clear(); - Stack *current = &state0_, *next = &state1_; - const size_t stateSetSize = GetStateSetSize(); - std::memset(stateSet_, 0, stateSetSize); - - bool matched = AddState(*current, root_); - unsigned codepoint; - while (!current->Empty() && (codepoint = ds.Take()) != 0) { - std::memset(stateSet_, 0, stateSetSize); - next->Clear(); - matched = false; - for (const SizeType* s = current->template Bottom(); s != current->template End(); ++s) { - const State& sr = GetState(*s); - if (sr.codepoint == codepoint || - sr.codepoint == kAnyCharacterClass || - (sr.codepoint == kRangeCharacterClass && MatchRange(sr.rangeStart, codepoint))) - { - matched = AddState(*next, sr.out) || matched; - if (!anchorEnd && matched) - return true; - } - if (!anchorBegin) - AddState(*next, root_); - } - internal::Swap(current, next); - } - - return matched; - } - - size_t GetStateSetSize() const { - return (stateCount_ + 31) / 32 * 4; - } - - // Return whether the added states is a match state - bool AddState(Stack& l, SizeType index) const { - RAPIDJSON_ASSERT(index != kRegexInvalidState); - - const State& s = GetState(index); - if (s.out1 != kRegexInvalidState) { // Split - bool matched = AddState(l, s.out); - return AddState(l, s.out1) || matched; - } - else if (!(stateSet_[index >> 5] & (1 << (index & 31)))) { - stateSet_[index >> 5] |= (1 << (index & 31)); - *l.template PushUnsafe() = index; - } - return s.out == kRegexInvalidState; // by using PushUnsafe() above, we can ensure s is not validated due to reallocation. - } - - bool MatchRange(SizeType rangeIndex, unsigned codepoint) const { - bool yes = (GetRange(rangeIndex).start & kRangeNegationFlag) == 0; - while (rangeIndex != kRegexInvalidRange) { - const Range& r = GetRange(rangeIndex); - if (codepoint >= (r.start & ~kRangeNegationFlag) && codepoint <= r.end) - return yes; - rangeIndex = r.next; - } - return !yes; - } - + Allocator* ownAllocator_; + Allocator* allocator_; Stack states_; Stack ranges_; SizeType root_; @@ -678,23 +599,141 @@ private: static const unsigned kInfinityQuantifier = ~0u; // For SearchWithAnchoring() - uint32_t* stateSet_; // allocated by states_.GetAllocator() - mutable Stack state0_; - mutable Stack state1_; bool anchorBegin_; bool anchorEnd_; }; +template +class GenericRegexSearch { +public: + typedef typename RegexType::EncodingType Encoding; + typedef typename Encoding::Ch Ch; + + GenericRegexSearch(const RegexType& regex, Allocator* allocator = 0) : + regex_(regex), allocator_(allocator), ownAllocator_(0), + state0_(allocator, 0), state1_(allocator, 0), stateSet_() + { + RAPIDJSON_ASSERT(regex_.IsValid()); + if (!allocator_) + ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator)(); + stateSet_ = static_cast(allocator_->Malloc(GetStateSetSize())); + state0_.template Reserve(regex_.stateCount_); + state1_.template Reserve(regex_.stateCount_); + } + + ~GenericRegexSearch() { + Allocator::Free(stateSet_); + RAPIDJSON_DELETE(ownAllocator_); + } + + template + bool Match(InputStream& is) { + return SearchWithAnchoring(is, true, true); + } + + bool Match(const Ch* s) { + GenericStringStream is(s); + return Match(is); + } + + template + bool Search(InputStream& is) { + return SearchWithAnchoring(is, regex_.anchorBegin_, regex_.anchorEnd_); + } + + bool Search(const Ch* s) { + GenericStringStream is(s); + return Search(is); + } + +private: + typedef typename RegexType::State State; + typedef typename RegexType::Range Range; + + template + bool SearchWithAnchoring(InputStream& is, bool anchorBegin, bool anchorEnd) { + DecodedStream ds(is); + + state0_.Clear(); + Stack *current = &state0_, *next = &state1_; + const size_t stateSetSize = GetStateSetSize(); + std::memset(stateSet_, 0, stateSetSize); + + bool matched = AddState(*current, regex_.root_); + unsigned codepoint; + while (!current->Empty() && (codepoint = ds.Take()) != 0) { + std::memset(stateSet_, 0, stateSetSize); + next->Clear(); + matched = false; + for (const SizeType* s = current->template Bottom(); s != current->template End(); ++s) { + const State& sr = regex_.GetState(*s); + if (sr.codepoint == codepoint || + sr.codepoint == RegexType::kAnyCharacterClass || + (sr.codepoint == RegexType::kRangeCharacterClass && MatchRange(sr.rangeStart, codepoint))) + { + matched = AddState(*next, sr.out) || matched; + if (!anchorEnd && matched) + return true; + } + if (!anchorBegin) + AddState(*next, regex_.root_); + } + internal::Swap(current, next); + } + + return matched; + } + + size_t GetStateSetSize() const { + return (regex_.stateCount_ + 31) / 32 * 4; + } + + // Return whether the added states is a match state + bool AddState(Stack& l, SizeType index) { + RAPIDJSON_ASSERT(index != kRegexInvalidState); + + const State& s = regex_.GetState(index); + if (s.out1 != kRegexInvalidState) { // Split + bool matched = AddState(l, s.out); + return AddState(l, s.out1) || matched; + } + else if (!(stateSet_[index >> 5] & (1u << (index & 31)))) { + stateSet_[index >> 5] |= (1u << (index & 31)); + *l.template PushUnsafe() = index; + } + return s.out == kRegexInvalidState; // by using PushUnsafe() above, we can ensure s is not validated due to reallocation. + } + + bool MatchRange(SizeType rangeIndex, unsigned codepoint) const { + bool yes = (regex_.GetRange(rangeIndex).start & RegexType::kRangeNegationFlag) == 0; + while (rangeIndex != kRegexInvalidRange) { + const Range& r = regex_.GetRange(rangeIndex); + if (codepoint >= (r.start & ~RegexType::kRangeNegationFlag) && codepoint <= r.end) + return yes; + rangeIndex = r.next; + } + return !yes; + } + + const RegexType& regex_; + Allocator* allocator_; + Allocator* ownAllocator_; + Stack state0_; + Stack state1_; + uint32_t* stateSet_; +}; + typedef GenericRegex > Regex; +typedef GenericRegexSearch RegexSearch; } // namespace internal RAPIDJSON_NAMESPACE_END -#ifdef __clang__ +#ifdef __GNUC__ RAPIDJSON_DIAG_POP #endif -#ifdef _MSC_VER +#if defined(__clang__) || defined(_MSC_VER) RAPIDJSON_DIAG_POP #endif diff --git a/src/3rdparty/rapidjson/internal/stack.h b/src/3rdparty/rapidjson/internal/stack.h index 022c9aab4..45dca6a8b 100644 --- a/src/3rdparty/rapidjson/internal/stack.h +++ b/src/3rdparty/rapidjson/internal/stack.h @@ -17,6 +17,7 @@ #include "../allocators.h" #include "swap.h" +#include #if defined(__clang__) RAPIDJSON_DIAG_PUSH @@ -100,7 +101,7 @@ public: void ShrinkToFit() { if (Empty()) { // If the stack is empty, completely deallocate the memory. - Allocator::Free(stack_); + Allocator::Free(stack_); // NOLINT (+clang-analyzer-unix.Malloc) stack_ = 0; stackTop_ = 0; stackEnd_ = 0; @@ -114,7 +115,7 @@ public: template RAPIDJSON_FORCEINLINE void Reserve(size_t count = 1) { // Expand the stack if needed - if (RAPIDJSON_UNLIKELY(stackTop_ + sizeof(T) * count > stackEnd_)) + if (RAPIDJSON_UNLIKELY(static_cast(sizeof(T) * count) > (stackEnd_ - stackTop_))) Expand(count); } @@ -126,7 +127,8 @@ public: template RAPIDJSON_FORCEINLINE T* PushUnsafe(size_t count = 1) { - RAPIDJSON_ASSERT(stackTop_ + sizeof(T) * count <= stackEnd_); + RAPIDJSON_ASSERT(stackTop_); + RAPIDJSON_ASSERT(static_cast(sizeof(T) * count) <= (stackEnd_ - stackTop_)); T* ret = reinterpret_cast(stackTop_); stackTop_ += sizeof(T) * count; return ret; @@ -183,7 +185,7 @@ private: size_t newCapacity; if (stack_ == 0) { if (!allocator_) - ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator()); + ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator)(); newCapacity = initialCapacity_; } else { newCapacity = GetCapacity(); diff --git a/src/3rdparty/rapidjson/internal/strfunc.h b/src/3rdparty/rapidjson/internal/strfunc.h index 2edfae526..226439a76 100644 --- a/src/3rdparty/rapidjson/internal/strfunc.h +++ b/src/3rdparty/rapidjson/internal/strfunc.h @@ -16,6 +16,7 @@ #define RAPIDJSON_INTERNAL_STRFUNC_H_ #include "../stream.h" +#include RAPIDJSON_NAMESPACE_BEGIN namespace internal { @@ -28,14 +29,27 @@ namespace internal { */ template inline SizeType StrLen(const Ch* s) { + RAPIDJSON_ASSERT(s != 0); const Ch* p = s; while (*p) ++p; return SizeType(p - s); } +template <> +inline SizeType StrLen(const char* s) { + return SizeType(std::strlen(s)); +} + +template <> +inline SizeType StrLen(const wchar_t* s) { + return SizeType(std::wcslen(s)); +} + //! Returns number of code points in a encoded string. template bool CountStringCodePoint(const typename Encoding::Ch* s, SizeType length, SizeType* outCount) { + RAPIDJSON_ASSERT(s != 0); + RAPIDJSON_ASSERT(outCount != 0); GenericStringStream is(s); const typename Encoding::Ch* end = s + length; SizeType count = 0; diff --git a/src/3rdparty/rapidjson/internal/strtod.h b/src/3rdparty/rapidjson/internal/strtod.h index 289c413b0..dfca22b65 100644 --- a/src/3rdparty/rapidjson/internal/strtod.h +++ b/src/3rdparty/rapidjson/internal/strtod.h @@ -19,6 +19,8 @@ #include "biginteger.h" #include "diyfp.h" #include "pow10.h" +#include +#include RAPIDJSON_NAMESPACE_BEGIN namespace internal { @@ -126,46 +128,46 @@ inline bool StrtodFast(double d, int p, double* result) { } // Compute an approximation and see if it is within 1/2 ULP -inline bool StrtodDiyFp(const char* decimals, size_t length, size_t decimalPosition, int exp, double* result) { +inline bool StrtodDiyFp(const char* decimals, int dLen, int dExp, double* result) { uint64_t significand = 0; - size_t i = 0; // 2^64 - 1 = 18446744073709551615, 1844674407370955161 = 0x1999999999999999 - for (; i < length; i++) { + int i = 0; // 2^64 - 1 = 18446744073709551615, 1844674407370955161 = 0x1999999999999999 + for (; i < dLen; i++) { if (significand > RAPIDJSON_UINT64_C2(0x19999999, 0x99999999) || (significand == RAPIDJSON_UINT64_C2(0x19999999, 0x99999999) && decimals[i] > '5')) break; significand = significand * 10u + static_cast(decimals[i] - '0'); } - if (i < length && decimals[i] >= '5') // Rounding + if (i < dLen && decimals[i] >= '5') // Rounding significand++; - size_t remaining = length - i; - const unsigned kUlpShift = 3; - const unsigned kUlp = 1 << kUlpShift; + int remaining = dLen - i; + const int kUlpShift = 3; + const int kUlp = 1 << kUlpShift; int64_t error = (remaining == 0) ? 0 : kUlp / 2; DiyFp v(significand, 0); v = v.Normalize(); error <<= -v.e; - const int dExp = static_cast(decimalPosition) - static_cast(i) + exp; + dExp += remaining; int actualExp; DiyFp cachedPower = GetCachedPower10(dExp, &actualExp); if (actualExp != dExp) { static const DiyFp kPow10[] = { - DiyFp(RAPIDJSON_UINT64_C2(0xa0000000, 00000000), -60), // 10^1 - DiyFp(RAPIDJSON_UINT64_C2(0xc8000000, 00000000), -57), // 10^2 - DiyFp(RAPIDJSON_UINT64_C2(0xfa000000, 00000000), -54), // 10^3 - DiyFp(RAPIDJSON_UINT64_C2(0x9c400000, 00000000), -50), // 10^4 - DiyFp(RAPIDJSON_UINT64_C2(0xc3500000, 00000000), -47), // 10^5 - DiyFp(RAPIDJSON_UINT64_C2(0xf4240000, 00000000), -44), // 10^6 - DiyFp(RAPIDJSON_UINT64_C2(0x98968000, 00000000), -40) // 10^7 + DiyFp(RAPIDJSON_UINT64_C2(0xa0000000, 0x00000000), -60), // 10^1 + DiyFp(RAPIDJSON_UINT64_C2(0xc8000000, 0x00000000), -57), // 10^2 + DiyFp(RAPIDJSON_UINT64_C2(0xfa000000, 0x00000000), -54), // 10^3 + DiyFp(RAPIDJSON_UINT64_C2(0x9c400000, 0x00000000), -50), // 10^4 + DiyFp(RAPIDJSON_UINT64_C2(0xc3500000, 0x00000000), -47), // 10^5 + DiyFp(RAPIDJSON_UINT64_C2(0xf4240000, 0x00000000), -44), // 10^6 + DiyFp(RAPIDJSON_UINT64_C2(0x98968000, 0x00000000), -40) // 10^7 }; - int adjustment = dExp - actualExp - 1; - RAPIDJSON_ASSERT(adjustment >= 0 && adjustment < 7); - v = v * kPow10[adjustment]; - if (length + static_cast(adjustment)> 19u) // has more digits than decimal digits in 64-bit + int adjustment = dExp - actualExp; + RAPIDJSON_ASSERT(adjustment >= 1 && adjustment < 8); + v = v * kPow10[adjustment - 1]; + if (dLen + adjustment > 19) // has more digits than decimal digits in 64-bit error += kUlp / 2; } @@ -177,17 +179,17 @@ inline bool StrtodDiyFp(const char* decimals, size_t length, size_t decimalPosit v = v.Normalize(); error <<= oldExp - v.e; - const unsigned effectiveSignificandSize = Double::EffectiveSignificandSize(64 + v.e); - unsigned precisionSize = 64 - effectiveSignificandSize; + const int effectiveSignificandSize = Double::EffectiveSignificandSize(64 + v.e); + int precisionSize = 64 - effectiveSignificandSize; if (precisionSize + kUlpShift >= 64) { - unsigned scaleExp = (precisionSize + kUlpShift) - 63; + int scaleExp = (precisionSize + kUlpShift) - 63; v.f >>= scaleExp; v.e += scaleExp; - error = (error >> scaleExp) + 1 + static_cast(kUlp); + error = (error >> scaleExp) + 1 + kUlp; precisionSize -= scaleExp; } - DiyFp rounded(v.f >> precisionSize, v.e + static_cast(precisionSize)); + DiyFp rounded(v.f >> precisionSize, v.e + precisionSize); const uint64_t precisionBits = (v.f & ((uint64_t(1) << precisionSize) - 1)) * kUlp; const uint64_t halfWay = (uint64_t(1) << (precisionSize - 1)) * kUlp; if (precisionBits >= halfWay + static_cast(error)) { @@ -203,9 +205,9 @@ inline bool StrtodDiyFp(const char* decimals, size_t length, size_t decimalPosit return halfWay - static_cast(error) >= precisionBits || precisionBits >= halfWay + static_cast(error); } -inline double StrtodBigInteger(double approx, const char* decimals, size_t length, size_t decimalPosition, int exp) { - const BigInteger dInt(decimals, length); - const int dExp = static_cast(decimalPosition) - static_cast(length) + exp; +inline double StrtodBigInteger(double approx, const char* decimals, int dLen, int dExp) { + RAPIDJSON_ASSERT(dLen >= 0); + const BigInteger dInt(decimals, static_cast(dLen)); Double a(approx); int cmp = CheckWithinHalfULP(a.Value(), dInt, dExp); if (cmp < 0) @@ -225,42 +227,61 @@ inline double StrtodFullPrecision(double d, int p, const char* decimals, size_t RAPIDJSON_ASSERT(d >= 0.0); RAPIDJSON_ASSERT(length >= 1); - double result; + double result = 0.0; if (StrtodFast(d, p, &result)) return result; + RAPIDJSON_ASSERT(length <= INT_MAX); + int dLen = static_cast(length); + + RAPIDJSON_ASSERT(length >= decimalPosition); + RAPIDJSON_ASSERT(length - decimalPosition <= INT_MAX); + int dExpAdjust = static_cast(length - decimalPosition); + + RAPIDJSON_ASSERT(exp >= INT_MIN + dExpAdjust); + int dExp = exp - dExpAdjust; + + // Make sure length+dExp does not overflow + RAPIDJSON_ASSERT(dExp <= INT_MAX - dLen); + // Trim leading zeros - while (*decimals == '0' && length > 1) { - length--; + while (dLen > 0 && *decimals == '0') { + dLen--; decimals++; - decimalPosition--; } // Trim trailing zeros - while (decimals[length - 1] == '0' && length > 1) { - length--; - decimalPosition--; - exp++; + while (dLen > 0 && decimals[dLen - 1] == '0') { + dLen--; + dExp++; + } + + if (dLen == 0) { // Buffer only contains zeros. + return 0.0; } // Trim right-most digits - const int kMaxDecimalDigit = 780; - if (static_cast(length) > kMaxDecimalDigit) { - int delta = (static_cast(length) - kMaxDecimalDigit); - exp += delta; - decimalPosition -= static_cast(delta); - length = kMaxDecimalDigit; + const int kMaxDecimalDigit = 767 + 1; + if (dLen > kMaxDecimalDigit) { + dExp += dLen - kMaxDecimalDigit; + dLen = kMaxDecimalDigit; } - // If too small, underflow to zero - if (int(length) + exp < -324) + // If too small, underflow to zero. + // Any x <= 10^-324 is interpreted as zero. + if (dLen + dExp <= -324) return 0.0; - if (StrtodDiyFp(decimals, length, decimalPosition, exp, &result)) + // If too large, overflow to infinity. + // Any x >= 10^309 is interpreted as +infinity. + if (dLen + dExp > 309) + return std::numeric_limits::infinity(); + + if (StrtodDiyFp(decimals, dLen, dExp, &result)) return result; // Use approximation from StrtodDiyFp and make adjustment with BigInteger comparison - return StrtodBigInteger(result, decimals, length, decimalPosition, exp); + return StrtodBigInteger(result, decimals, dLen, dExp); } } // namespace internal diff --git a/src/3rdparty/rapidjson/istreamwrapper.h b/src/3rdparty/rapidjson/istreamwrapper.h index f5fe28977..c4950b9dc 100644 --- a/src/3rdparty/rapidjson/istreamwrapper.h +++ b/src/3rdparty/rapidjson/istreamwrapper.h @@ -17,13 +17,12 @@ #include "stream.h" #include +#include #ifdef __clang__ RAPIDJSON_DIAG_PUSH RAPIDJSON_DIAG_OFF(padded) -#endif - -#ifdef _MSC_VER +#elif defined(_MSC_VER) RAPIDJSON_DIAG_PUSH RAPIDJSON_DIAG_OFF(4351) // new behavior: elements of array 'array' will be default initialized #endif @@ -50,57 +49,71 @@ template class BasicIStreamWrapper { public: typedef typename StreamType::char_type Ch; - BasicIStreamWrapper(StreamType& stream) : stream_(stream), count_(), peekBuffer_() {} - Ch Peek() const { - typename StreamType::int_type c = stream_.peek(); - return RAPIDJSON_LIKELY(c != StreamType::traits_type::eof()) ? static_cast(c) : '\0'; + //! Constructor. + /*! + \param stream stream opened for read. + */ + BasicIStreamWrapper(StreamType &stream) : stream_(stream), buffer_(peekBuffer_), bufferSize_(4), bufferLast_(0), current_(buffer_), readCount_(0), count_(0), eof_(false) { + Read(); } - Ch Take() { - typename StreamType::int_type c = stream_.get(); - if (RAPIDJSON_LIKELY(c != StreamType::traits_type::eof())) { - count_++; - return static_cast(c); - } - else - return '\0'; + //! Constructor. + /*! + \param stream stream opened for read. + \param buffer user-supplied buffer. + \param bufferSize size of buffer in bytes. Must >=4 bytes. + */ + BasicIStreamWrapper(StreamType &stream, char* buffer, size_t bufferSize) : stream_(stream), buffer_(buffer), bufferSize_(bufferSize), bufferLast_(0), current_(buffer_), readCount_(0), count_(0), eof_(false) { + RAPIDJSON_ASSERT(bufferSize >= 4); + Read(); } - // tellg() may return -1 when failed. So we count by ourself. - size_t Tell() const { return count_; } + Ch Peek() const { return *current_; } + Ch Take() { Ch c = *current_; Read(); return c; } + size_t Tell() const { return count_ + static_cast(current_ - buffer_); } - Ch* PutBegin() { RAPIDJSON_ASSERT(false); return 0; } + // Not implemented void Put(Ch) { RAPIDJSON_ASSERT(false); } - void Flush() { RAPIDJSON_ASSERT(false); } + void Flush() { RAPIDJSON_ASSERT(false); } + Ch* PutBegin() { RAPIDJSON_ASSERT(false); return 0; } size_t PutEnd(Ch*) { RAPIDJSON_ASSERT(false); return 0; } // For encoding detection only. const Ch* Peek4() const { - RAPIDJSON_ASSERT(sizeof(Ch) == 1); // Only usable for byte stream. - int i; - bool hasError = false; - for (i = 0; i < 4; ++i) { - typename StreamType::int_type c = stream_.get(); - if (c == StreamType::traits_type::eof()) { - hasError = true; - stream_.clear(); - break; - } - peekBuffer_[i] = static_cast(c); - } - for (--i; i >= 0; --i) - stream_.putback(peekBuffer_[i]); - return !hasError ? peekBuffer_ : 0; + return (current_ + 4 - !eof_ <= bufferLast_) ? current_ : 0; } private: + BasicIStreamWrapper(); BasicIStreamWrapper(const BasicIStreamWrapper&); BasicIStreamWrapper& operator=(const BasicIStreamWrapper&); - StreamType& stream_; - size_t count_; //!< Number of characters read. Note: - mutable Ch peekBuffer_[4]; + void Read() { + if (current_ < bufferLast_) + ++current_; + else if (!eof_) { + count_ += readCount_; + readCount_ = bufferSize_; + bufferLast_ = buffer_ + readCount_ - 1; + current_ = buffer_; + + if (!stream_.read(buffer_, static_cast(bufferSize_))) { + readCount_ = static_cast(stream_.gcount()); + *(bufferLast_ = buffer_ + readCount_) = '\0'; + eof_ = true; + } + } + } + + StreamType &stream_; + Ch peekBuffer_[4], *buffer_; + size_t bufferSize_; + Ch *bufferLast_; + Ch *current_; + size_t readCount_; + size_t count_; //!< Number of characters read + bool eof_; }; typedef BasicIStreamWrapper IStreamWrapper; diff --git a/src/3rdparty/rapidjson/license.txt b/src/3rdparty/rapidjson/license.txt new file mode 100644 index 000000000..7ccc161c8 --- /dev/null +++ b/src/3rdparty/rapidjson/license.txt @@ -0,0 +1,57 @@ +Tencent is pleased to support the open source community by making RapidJSON available. + +Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved. + +If you have downloaded a copy of the RapidJSON binary from Tencent, please note that the RapidJSON binary is licensed under the MIT License. +If you have downloaded a copy of the RapidJSON source code from Tencent, please note that RapidJSON source code is licensed under the MIT License, except for the third-party components listed below which are subject to different license terms. Your integration of RapidJSON into your own projects may require compliance with the MIT License, as well as the other licenses applicable to the third-party components included within RapidJSON. To avoid the problematic JSON license in your own projects, it's sufficient to exclude the bin/jsonchecker/ directory, as it's the only code under the JSON license. +A copy of the MIT License is included in this file. + +Other dependencies and licenses: + +Open Source Software Licensed Under the BSD License: +-------------------------------------------------------------------- + +The msinttypes r29 +Copyright (c) 2006-2013 Alexander Chemeris +All rights reserved. + +Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met: + +* Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer. +* Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution. +* Neither the name of copyright holder nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS AND CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +Open Source Software Licensed Under the JSON License: +-------------------------------------------------------------------- + +json.org +Copyright (c) 2002 JSON.org +All Rights Reserved. + +JSON_checker +Copyright (c) 2002 JSON.org +All Rights Reserved. + + +Terms of the JSON License: +--------------------------------------------------- + +Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: + +The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. + +The Software shall be used for Good, not Evil. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + + +Terms of the MIT License: +-------------------------------------------------------------------- + +Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: + +The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. diff --git a/src/3rdparty/rapidjson/pointer.h b/src/3rdparty/rapidjson/pointer.h index 0206ac1c8..063abab9a 100644 --- a/src/3rdparty/rapidjson/pointer.h +++ b/src/3rdparty/rapidjson/pointer.h @@ -21,9 +21,7 @@ #ifdef __clang__ RAPIDJSON_DIAG_PUSH RAPIDJSON_DIAG_OFF(switch-enum) -#endif - -#ifdef _MSC_VER +#elif defined(_MSC_VER) RAPIDJSON_DIAG_PUSH RAPIDJSON_DIAG_OFF(4512) // assignment operator could not be generated #endif @@ -165,7 +163,12 @@ public: GenericPointer(const Token* tokens, size_t tokenCount) : allocator_(), ownAllocator_(), nameBuffer_(), tokens_(const_cast(tokens)), tokenCount_(tokenCount), parseErrorOffset_(), parseErrorCode_(kPointerParseErrorNone) {} //! Copy constructor. - GenericPointer(const GenericPointer& rhs, Allocator* allocator = 0) : allocator_(allocator), ownAllocator_(), nameBuffer_(), tokens_(), tokenCount_(), parseErrorOffset_(), parseErrorCode_(kPointerParseErrorNone) { + GenericPointer(const GenericPointer& rhs) : allocator_(rhs.allocator_), ownAllocator_(), nameBuffer_(), tokens_(), tokenCount_(), parseErrorOffset_(), parseErrorCode_(kPointerParseErrorNone) { + *this = rhs; + } + + //! Copy constructor. + GenericPointer(const GenericPointer& rhs, Allocator* allocator) : allocator_(allocator), ownAllocator_(), nameBuffer_(), tokens_(), tokenCount_(), parseErrorOffset_(), parseErrorCode_(kPointerParseErrorNone) { *this = rhs; } @@ -197,6 +200,36 @@ public: return *this; } + //! Swap the content of this pointer with an other. + /*! + \param other The pointer to swap with. + \note Constant complexity. + */ + GenericPointer& Swap(GenericPointer& other) RAPIDJSON_NOEXCEPT { + internal::Swap(allocator_, other.allocator_); + internal::Swap(ownAllocator_, other.ownAllocator_); + internal::Swap(nameBuffer_, other.nameBuffer_); + internal::Swap(tokens_, other.tokens_); + internal::Swap(tokenCount_, other.tokenCount_); + internal::Swap(parseErrorOffset_, other.parseErrorOffset_); + internal::Swap(parseErrorCode_, other.parseErrorCode_); + return *this; + } + + //! free-standing swap function helper + /*! + Helper function to enable support for common swap implementation pattern based on \c std::swap: + \code + void swap(MyClass& a, MyClass& b) { + using std::swap; + swap(a.pointer, b.pointer); + // ... + } + \endcode + \see Swap() + */ + friend inline void swap(GenericPointer& a, GenericPointer& b) RAPIDJSON_NOEXCEPT { a.Swap(b); } + //@} //!@name Append token @@ -240,7 +273,7 @@ public: template RAPIDJSON_DISABLEIF_RETURN((internal::NotExpr::Type, Ch> >), (GenericPointer)) Append(T* name, Allocator* allocator = 0) const { - return Append(name, StrLen(name), allocator); + return Append(name, internal::StrLen(name), allocator); } #if RAPIDJSON_HAS_STDSTRING @@ -274,7 +307,7 @@ public: else { Ch name[21]; for (size_t i = 0; i <= length; i++) - name[i] = buffer[i]; + name[i] = static_cast(buffer[i]); Token token = { name, length, index }; return Append(token, allocator); } @@ -353,6 +386,33 @@ public: */ bool operator!=(const GenericPointer& rhs) const { return !(*this == rhs); } + //! Less than operator. + /*! + \note Invalid pointers are always greater than valid ones. + */ + bool operator<(const GenericPointer& rhs) const { + if (!IsValid()) + return false; + if (!rhs.IsValid()) + return true; + + if (tokenCount_ != rhs.tokenCount_) + return tokenCount_ < rhs.tokenCount_; + + for (size_t i = 0; i < tokenCount_; i++) { + if (tokens_[i].index != rhs.tokens_[i].index) + return tokens_[i].index < rhs.tokens_[i].index; + + if (tokens_[i].length != rhs.tokens_[i].length) + return tokens_[i].length < rhs.tokens_[i].length; + + if (int cmp = std::memcmp(tokens_[i].name, rhs.tokens_[i].name, sizeof(Ch) * tokens_[i].length)) + return cmp < 0; + } + + return false; + } + //@} //!@name Stringify @@ -532,14 +592,14 @@ public: */ ValueType& GetWithDefault(ValueType& root, const ValueType& defaultValue, typename ValueType::AllocatorType& allocator) const { bool alreadyExist; - Value& v = Create(root, allocator, &alreadyExist); + ValueType& v = Create(root, allocator, &alreadyExist); return alreadyExist ? v : v.CopyFrom(defaultValue, allocator); } //! Query a value in a subtree with default null-terminated string. ValueType& GetWithDefault(ValueType& root, const Ch* defaultValue, typename ValueType::AllocatorType& allocator) const { bool alreadyExist; - Value& v = Create(root, allocator, &alreadyExist); + ValueType& v = Create(root, allocator, &alreadyExist); return alreadyExist ? v : v.SetString(defaultValue, allocator); } @@ -547,7 +607,7 @@ public: //! Query a value in a subtree with default std::basic_string. ValueType& GetWithDefault(ValueType& root, const std::basic_string& defaultValue, typename ValueType::AllocatorType& allocator) const { bool alreadyExist; - Value& v = Create(root, allocator, &alreadyExist); + ValueType& v = Create(root, allocator, &alreadyExist); return alreadyExist ? v : v.SetString(defaultValue, allocator); } #endif @@ -758,7 +818,7 @@ private: */ Ch* CopyFromRaw(const GenericPointer& rhs, size_t extraToken = 0, size_t extraNameBufferSize = 0) { if (!allocator_) // allocator is independently owned. - ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator()); + ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator)(); size_t nameBufferSize = rhs.tokenCount_; // null terminators for tokens for (Token *t = rhs.tokens_; t != rhs.tokens_ + rhs.tokenCount_; ++t) @@ -806,7 +866,7 @@ private: // Create own allocator if user did not supply. if (!allocator_) - ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator()); + ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator)(); // Count number of '/' as tokenCount tokenCount_ = 0; @@ -1029,8 +1089,8 @@ private: unsigned char u = static_cast(c); static const char hexDigits[16] = { '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'A', 'B', 'C', 'D', 'E', 'F' }; os_.Put('%'); - os_.Put(hexDigits[u >> 4]); - os_.Put(hexDigits[u & 15]); + os_.Put(static_cast(hexDigits[u >> 4])); + os_.Put(static_cast(hexDigits[u & 15])); } private: OutputStream& os_; @@ -1347,11 +1407,7 @@ bool EraseValueByPointer(T& root, const CharType(&source)[N]) { RAPIDJSON_NAMESPACE_END -#ifdef __clang__ -RAPIDJSON_DIAG_POP -#endif - -#ifdef _MSC_VER +#if defined(__clang__) || defined(_MSC_VER) RAPIDJSON_DIAG_POP #endif diff --git a/src/3rdparty/rapidjson/prettywriter.h b/src/3rdparty/rapidjson/prettywriter.h index 0dcb0fee9..c7c29b214 100644 --- a/src/3rdparty/rapidjson/prettywriter.h +++ b/src/3rdparty/rapidjson/prettywriter.h @@ -22,6 +22,11 @@ RAPIDJSON_DIAG_PUSH RAPIDJSON_DIAG_OFF(effc++) #endif +#if defined(__clang__) +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(c++98-compat) +#endif + RAPIDJSON_NAMESPACE_BEGIN //! Combination of PrettyWriter format flags. @@ -34,7 +39,7 @@ enum PrettyFormatOptions { //! Writer with indentation and spacing. /*! - \tparam OutputStream Type of ouptut os. + \tparam OutputStream Type of output os. \tparam SourceEncoding Encoding of source string. \tparam TargetEncoding Encoding of output stream. \tparam StackAllocator Type of allocator for allocating memory of stack. @@ -42,7 +47,7 @@ enum PrettyFormatOptions { template, typename TargetEncoding = UTF8<>, typename StackAllocator = CrtAllocator, unsigned writeFlags = kWriteDefaultFlags> class PrettyWriter : public Writer { public: - typedef Writer Base; + typedef Writer Base; typedef typename Base::Ch Ch; //! Constructor @@ -57,6 +62,11 @@ public: explicit PrettyWriter(StackAllocator* allocator = 0, size_t levelDepth = Base::kDefaultLevelDepth) : Base(allocator, levelDepth), indentChar_(' '), indentCharCount_(4) {} +#if RAPIDJSON_HAS_CXX11_RVALUE_REFS + PrettyWriter(PrettyWriter&& rhs) : + Base(std::forward(rhs)), indentChar_(rhs.indentChar_), indentCharCount_(rhs.indentCharCount_), formatOptions_(rhs.formatOptions_) {} +#endif + //! Set custom indentation. /*! \param indentChar Character for indentation. Must be whitespace character (' ', '\\t', '\\n', '\\r'). \param indentCharCount Number of indent characters for each indentation level. @@ -82,24 +92,26 @@ public: */ //@{ - bool Null() { PrettyPrefix(kNullType); return Base::WriteNull(); } - bool Bool(bool b) { PrettyPrefix(b ? kTrueType : kFalseType); return Base::WriteBool(b); } - bool Int(int i) { PrettyPrefix(kNumberType); return Base::WriteInt(i); } - bool Uint(unsigned u) { PrettyPrefix(kNumberType); return Base::WriteUint(u); } - bool Int64(int64_t i64) { PrettyPrefix(kNumberType); return Base::WriteInt64(i64); } - bool Uint64(uint64_t u64) { PrettyPrefix(kNumberType); return Base::WriteUint64(u64); } - bool Double(double d) { PrettyPrefix(kNumberType); return Base::WriteDouble(d); } + bool Null() { PrettyPrefix(kNullType); return Base::EndValue(Base::WriteNull()); } + bool Bool(bool b) { PrettyPrefix(b ? kTrueType : kFalseType); return Base::EndValue(Base::WriteBool(b)); } + bool Int(int i) { PrettyPrefix(kNumberType); return Base::EndValue(Base::WriteInt(i)); } + bool Uint(unsigned u) { PrettyPrefix(kNumberType); return Base::EndValue(Base::WriteUint(u)); } + bool Int64(int64_t i64) { PrettyPrefix(kNumberType); return Base::EndValue(Base::WriteInt64(i64)); } + bool Uint64(uint64_t u64) { PrettyPrefix(kNumberType); return Base::EndValue(Base::WriteUint64(u64)); } + bool Double(double d) { PrettyPrefix(kNumberType); return Base::EndValue(Base::WriteDouble(d)); } bool RawNumber(const Ch* str, SizeType length, bool copy = false) { + RAPIDJSON_ASSERT(str != 0); (void)copy; PrettyPrefix(kNumberType); - return Base::WriteString(str, length); + return Base::EndValue(Base::WriteString(str, length)); } bool String(const Ch* str, SizeType length, bool copy = false) { + RAPIDJSON_ASSERT(str != 0); (void)copy; PrettyPrefix(kStringType); - return Base::WriteString(str, length); + return Base::EndValue(Base::WriteString(str, length)); } #if RAPIDJSON_HAS_STDSTRING @@ -124,19 +136,21 @@ public: bool EndObject(SizeType memberCount = 0) { (void)memberCount; - RAPIDJSON_ASSERT(Base::level_stack_.GetSize() >= sizeof(typename Base::Level)); - RAPIDJSON_ASSERT(!Base::level_stack_.template Top()->inArray); + RAPIDJSON_ASSERT(Base::level_stack_.GetSize() >= sizeof(typename Base::Level)); // not inside an Object + RAPIDJSON_ASSERT(!Base::level_stack_.template Top()->inArray); // currently inside an Array, not Object + RAPIDJSON_ASSERT(0 == Base::level_stack_.template Top()->valueCount % 2); // Object has a Key without a Value + bool empty = Base::level_stack_.template Pop(1)->valueCount == 0; if (!empty) { Base::os_->Put('\n'); WriteIndent(); } - bool ret = Base::WriteEndObject(); + bool ret = Base::EndValue(Base::WriteEndObject()); (void)ret; RAPIDJSON_ASSERT(ret == true); if (Base::level_stack_.Empty()) // end of json text - Base::os_->Flush(); + Base::Flush(); return true; } @@ -150,17 +164,18 @@ public: (void)memberCount; RAPIDJSON_ASSERT(Base::level_stack_.GetSize() >= sizeof(typename Base::Level)); RAPIDJSON_ASSERT(Base::level_stack_.template Top()->inArray); - bool empty = Base::level_stack_.template Pop(1)->valueCount == 0; + typename Base::Level* level = Base::level_stack_.template Pop(1); + bool empty = level->valueCount == 0; - if (!empty && !(formatOptions_ & kFormatSingleLineArray)) { + if (!empty && !level->inLine) { Base::os_->Put('\n'); WriteIndent(); } - bool ret = Base::WriteEndArray(); + bool ret = Base::EndValue(Base::WriteEndArray()); (void)ret; RAPIDJSON_ASSERT(ret == true); if (Base::level_stack_.Empty()) // end of json text - Base::os_->Flush(); + Base::Flush(); return true; } @@ -184,7 +199,11 @@ public: \param type Type of the root of json. \note When using PrettyWriter::RawValue(), the result json may not be indented correctly. */ - bool RawValue(const Ch* json, size_t length, Type type) { PrettyPrefix(type); return Base::WriteRawValue(json, length); } + bool RawValue(const Ch* json, size_t length, Type type) { + RAPIDJSON_ASSERT(json != 0); + PrettyPrefix(type); + return Base::EndValue(Base::WriteRawValue(json, length)); + } protected: void PrettyPrefix(Type type) { @@ -193,13 +212,16 @@ protected: typename Base::Level* level = Base::level_stack_.template Top(); if (level->inArray) { + level->inLine = (formatOptions_ & kFormatSingleLineArray) && type != kObjectType && type != kArrayType; + if (level->valueCount > 0) { Base::os_->Put(','); // add comma if it is not the first element in array - if (formatOptions_ & kFormatSingleLineArray) + if (level->inLine) { Base::os_->Put(' '); + } } - if (!(formatOptions_ & kFormatSingleLineArray)) { + if (!level->inLine) { Base::os_->Put('\n'); WriteIndent(); } @@ -233,7 +255,7 @@ protected: void WriteIndent() { size_t count = (Base::level_stack_.GetSize() / sizeof(typename Base::Level)) * indentCharCount_; - PutN(*Base::os_, static_cast(indentChar_), count); + PutN(*Base::os_, static_cast(indentChar_), count); } Ch indentChar_; @@ -248,6 +270,10 @@ private: RAPIDJSON_NAMESPACE_END +#if defined(__clang__) +RAPIDJSON_DIAG_POP +#endif + #ifdef __GNUC__ RAPIDJSON_DIAG_POP #endif diff --git a/src/3rdparty/rapidjson/rapidjson.h b/src/3rdparty/rapidjson/rapidjson.h index 2ef9bc56c..78c8aae0c 100644 --- a/src/3rdparty/rapidjson/rapidjson.h +++ b/src/3rdparty/rapidjson/rapidjson.h @@ -26,7 +26,7 @@ Some RapidJSON features are configurable to adapt the library to a wide variety of platforms, environments and usage scenarios. Most of the - features can be configured in terms of overriden or predefined + features can be configured in terms of overridden or predefined preprocessor macros at compile-time. Some additional customization is available in the \ref RAPIDJSON_ERRORS APIs. @@ -49,6 +49,11 @@ // token stringification #define RAPIDJSON_STRINGIFY(x) RAPIDJSON_DO_STRINGIFY(x) #define RAPIDJSON_DO_STRINGIFY(x) #x + +// token concatenation +#define RAPIDJSON_JOIN(X, Y) RAPIDJSON_DO_JOIN(X, Y) +#define RAPIDJSON_DO_JOIN(X, Y) RAPIDJSON_DO_JOIN2(X, Y) +#define RAPIDJSON_DO_JOIN2(X, Y) X##Y //!@endcond /*! \def RAPIDJSON_MAJOR_VERSION @@ -214,7 +219,7 @@ # elif __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ # define RAPIDJSON_ENDIAN RAPIDJSON_BIGENDIAN # else -# error Unknown machine endianess detected. User needs to define RAPIDJSON_ENDIAN. +# error Unknown machine endianness detected. User needs to define RAPIDJSON_ENDIAN. # endif // __BYTE_ORDER__ // Detect with GLIBC's endian.h # elif defined(__GLIBC__) @@ -224,7 +229,7 @@ # elif (__BYTE_ORDER == __BIG_ENDIAN) # define RAPIDJSON_ENDIAN RAPIDJSON_BIGENDIAN # else -# error Unknown machine endianess detected. User needs to define RAPIDJSON_ENDIAN. +# error Unknown machine endianness detected. User needs to define RAPIDJSON_ENDIAN. # endif // __GLIBC__ // Detect with _LITTLE_ENDIAN and _BIG_ENDIAN macro # elif defined(_LITTLE_ENDIAN) && !defined(_BIG_ENDIAN) @@ -236,12 +241,12 @@ # define RAPIDJSON_ENDIAN RAPIDJSON_BIGENDIAN # elif defined(__i386__) || defined(__alpha__) || defined(__ia64) || defined(__ia64__) || defined(_M_IX86) || defined(_M_IA64) || defined(_M_ALPHA) || defined(__amd64) || defined(__amd64__) || defined(_M_AMD64) || defined(__x86_64) || defined(__x86_64__) || defined(_M_X64) || defined(__bfin__) # define RAPIDJSON_ENDIAN RAPIDJSON_LITTLEENDIAN -# elif defined(_MSC_VER) && defined(_M_ARM) +# elif defined(_MSC_VER) && (defined(_M_ARM) || defined(_M_ARM64)) # define RAPIDJSON_ENDIAN RAPIDJSON_LITTLEENDIAN # elif defined(RAPIDJSON_DOXYGEN_RUNNING) # define RAPIDJSON_ENDIAN # else -# error Unknown machine endianess detected. User needs to define RAPIDJSON_ENDIAN. +# error Unknown machine endianness detected. User needs to define RAPIDJSON_ENDIAN. # endif #endif // RAPIDJSON_ENDIAN @@ -264,16 +269,11 @@ /*! \ingroup RAPIDJSON_CONFIG \param x pointer to align - Some machines require strict data alignment. Currently the default uses 4 bytes - alignment on 32-bit platforms and 8 bytes alignment for 64-bit platforms. + Some machines require strict data alignment. The default is 8 bytes. User can customize by defining the RAPIDJSON_ALIGN function macro. */ #ifndef RAPIDJSON_ALIGN -#if RAPIDJSON_64BIT == 1 -#define RAPIDJSON_ALIGN(x) (((x) + static_cast(7u)) & ~static_cast(7u)) -#else -#define RAPIDJSON_ALIGN(x) (((x) + 3u) & ~3u) -#endif +#define RAPIDJSON_ALIGN(x) (((x) + static_cast(7u)) & ~static_cast(7u)) #endif /////////////////////////////////////////////////////////////////////////////// @@ -320,17 +320,17 @@ #endif /////////////////////////////////////////////////////////////////////////////// -// RAPIDJSON_SSE2/RAPIDJSON_SSE42/RAPIDJSON_SIMD +// RAPIDJSON_SSE2/RAPIDJSON_SSE42/RAPIDJSON_NEON/RAPIDJSON_SIMD /*! \def RAPIDJSON_SIMD \ingroup RAPIDJSON_CONFIG - \brief Enable SSE2/SSE4.2 optimization. + \brief Enable SSE2/SSE4.2/Neon optimization. RapidJSON supports optimized implementations for some parsing operations - based on the SSE2 or SSE4.2 SIMD extensions on modern Intel-compatible - processors. + based on the SSE2, SSE4.2 or NEon SIMD extensions on modern Intel + or ARM compatible processors. - To enable these optimizations, two different symbols can be defined; + To enable these optimizations, three different symbols can be defined; \code // Enable SSE2 optimization. #define RAPIDJSON_SSE2 @@ -339,13 +339,17 @@ #define RAPIDJSON_SSE42 \endcode - \c RAPIDJSON_SSE42 takes precedence, if both are defined. + // Enable ARM Neon optimization. + #define RAPIDJSON_NEON + \endcode + + \c RAPIDJSON_SSE42 takes precedence over SSE2, if both are defined. If any of these symbols is defined, RapidJSON defines the macro \c RAPIDJSON_SIMD to indicate the availability of the optimized code. */ #if defined(RAPIDJSON_SSE2) || defined(RAPIDJSON_SSE42) \ - || defined(RAPIDJSON_DOXYGEN_RUNNING) + || defined(RAPIDJSON_NEON) || defined(RAPIDJSON_DOXYGEN_RUNNING) #define RAPIDJSON_SIMD #endif @@ -398,13 +402,22 @@ RAPIDJSON_NAMESPACE_END \ref RAPIDJSON_ERRORS APIs. */ #ifndef RAPIDJSON_ASSERT +#include #define RAPIDJSON_ASSERT(x) #endif // RAPIDJSON_ASSERT /////////////////////////////////////////////////////////////////////////////// // RAPIDJSON_STATIC_ASSERT -// Adopt from boost +// Prefer C++11 static_assert, if available +#ifndef RAPIDJSON_STATIC_ASSERT +#if __cplusplus >= 201103L || ( defined(_MSC_VER) && _MSC_VER >= 1800 ) +#define RAPIDJSON_STATIC_ASSERT(x) \ + static_assert(x, RAPIDJSON_STRINGIFY(x)) +#endif // C++11 +#endif // RAPIDJSON_STATIC_ASSERT + +// Adopt C++03 implementation from boost #ifndef RAPIDJSON_STATIC_ASSERT #ifndef __clang__ //!@cond RAPIDJSON_HIDDEN_FROM_DOXYGEN @@ -412,14 +425,10 @@ RAPIDJSON_NAMESPACE_END RAPIDJSON_NAMESPACE_BEGIN template struct STATIC_ASSERTION_FAILURE; template <> struct STATIC_ASSERTION_FAILURE { enum { value = 1 }; }; -template struct StaticAssertTest {}; +template struct StaticAssertTest {}; RAPIDJSON_NAMESPACE_END -#define RAPIDJSON_JOIN(X, Y) RAPIDJSON_DO_JOIN(X, Y) -#define RAPIDJSON_DO_JOIN(X, Y) RAPIDJSON_DO_JOIN2(X, Y) -#define RAPIDJSON_DO_JOIN2(X, Y) X##Y - -#if defined(__GNUC__) +#if defined(__GNUC__) || defined(__clang__) #define RAPIDJSON_STATIC_ASSERT_UNUSED_ATTRIBUTE __attribute__((unused)) #else #define RAPIDJSON_STATIC_ASSERT_UNUSED_ATTRIBUTE @@ -437,7 +446,7 @@ RAPIDJSON_NAMESPACE_END typedef ::RAPIDJSON_NAMESPACE::StaticAssertTest< \ sizeof(::RAPIDJSON_NAMESPACE::STATIC_ASSERTION_FAILURE)> \ RAPIDJSON_JOIN(StaticAssertTypedef, __LINE__) RAPIDJSON_STATIC_ASSERT_UNUSED_ATTRIBUTE -#endif +#endif // RAPIDJSON_STATIC_ASSERT /////////////////////////////////////////////////////////////////////////////// // RAPIDJSON_LIKELY, RAPIDJSON_UNLIKELY @@ -529,13 +538,14 @@ RAPIDJSON_NAMESPACE_END #ifndef RAPIDJSON_HAS_CXX11_RVALUE_REFS #if defined(__clang__) #if __has_feature(cxx_rvalue_references) && \ - (defined(_LIBCPP_VERSION) || defined(__GLIBCXX__) && __GLIBCXX__ >= 20080306) + (defined(_MSC_VER) || defined(_LIBCPP_VERSION) || defined(__GLIBCXX__) && __GLIBCXX__ >= 20080306) #define RAPIDJSON_HAS_CXX11_RVALUE_REFS 1 #else #define RAPIDJSON_HAS_CXX11_RVALUE_REFS 0 #endif #elif (defined(RAPIDJSON_GNUC) && (RAPIDJSON_GNUC >= RAPIDJSON_VERSION_CODE(4,3,0)) && defined(__GXX_EXPERIMENTAL_CXX0X__)) || \ - (defined(_MSC_VER) && _MSC_VER >= 1600) + (defined(_MSC_VER) && _MSC_VER >= 1600) || \ + (defined(__SUNPRO_CC) && __SUNPRO_CC >= 0x5140 && defined(__GXX_EXPERIMENTAL_CXX0X__)) #define RAPIDJSON_HAS_CXX11_RVALUE_REFS 1 #else @@ -546,8 +556,9 @@ RAPIDJSON_NAMESPACE_END #ifndef RAPIDJSON_HAS_CXX11_NOEXCEPT #if defined(__clang__) #define RAPIDJSON_HAS_CXX11_NOEXCEPT __has_feature(cxx_noexcept) -#elif (defined(RAPIDJSON_GNUC) && (RAPIDJSON_GNUC >= RAPIDJSON_VERSION_CODE(4,6,0)) && defined(__GXX_EXPERIMENTAL_CXX0X__)) -// (defined(_MSC_VER) && _MSC_VER >= ????) // not yet supported +#elif (defined(RAPIDJSON_GNUC) && (RAPIDJSON_GNUC >= RAPIDJSON_VERSION_CODE(4,6,0)) && defined(__GXX_EXPERIMENTAL_CXX0X__)) || \ + (defined(_MSC_VER) && _MSC_VER >= 1900) || \ + (defined(__SUNPRO_CC) && __SUNPRO_CC >= 0x5140 && defined(__GXX_EXPERIMENTAL_CXX0X__)) #define RAPIDJSON_HAS_CXX11_NOEXCEPT 1 #else #define RAPIDJSON_HAS_CXX11_NOEXCEPT 0 @@ -561,14 +572,19 @@ RAPIDJSON_NAMESPACE_END // no automatic detection, yet #ifndef RAPIDJSON_HAS_CXX11_TYPETRAITS +#if (defined(_MSC_VER) && _MSC_VER >= 1700) +#define RAPIDJSON_HAS_CXX11_TYPETRAITS 1 +#else #define RAPIDJSON_HAS_CXX11_TYPETRAITS 0 #endif +#endif #ifndef RAPIDJSON_HAS_CXX11_RANGE_FOR #if defined(__clang__) #define RAPIDJSON_HAS_CXX11_RANGE_FOR __has_feature(cxx_range_for) -#elif (defined(RAPIDJSON_GNUC) && (RAPIDJSON_GNUC >= RAPIDJSON_VERSION_CODE(4,3,0)) && defined(__GXX_EXPERIMENTAL_CXX0X__)) || \ - (defined(_MSC_VER) && _MSC_VER >= 1700) +#elif (defined(RAPIDJSON_GNUC) && (RAPIDJSON_GNUC >= RAPIDJSON_VERSION_CODE(4,6,0)) && defined(__GXX_EXPERIMENTAL_CXX0X__)) || \ + (defined(_MSC_VER) && _MSC_VER >= 1700) || \ + (defined(__SUNPRO_CC) && __SUNPRO_CC >= 0x5140 && defined(__GXX_EXPERIMENTAL_CXX0X__)) #define RAPIDJSON_HAS_CXX11_RANGE_FOR 1 #else #define RAPIDJSON_HAS_CXX11_RANGE_FOR 0 @@ -577,12 +593,38 @@ RAPIDJSON_NAMESPACE_END //!@endcond +//! Assertion (in non-throwing contexts). + /*! \ingroup RAPIDJSON_CONFIG + Some functions provide a \c noexcept guarantee, if the compiler supports it. + In these cases, the \ref RAPIDJSON_ASSERT macro cannot be overridden to + throw an exception. This macro adds a separate customization point for + such cases. + + Defaults to C \c assert() (as \ref RAPIDJSON_ASSERT), if \c noexcept is + supported, and to \ref RAPIDJSON_ASSERT otherwise. + */ + +/////////////////////////////////////////////////////////////////////////////// +// RAPIDJSON_NOEXCEPT_ASSERT + +#ifndef RAPIDJSON_NOEXCEPT_ASSERT +#ifdef RAPIDJSON_ASSERT_THROWS +#if RAPIDJSON_HAS_CXX11_NOEXCEPT +#define RAPIDJSON_NOEXCEPT_ASSERT(x) +#else +#define RAPIDJSON_NOEXCEPT_ASSERT(x) RAPIDJSON_ASSERT(x) +#endif // RAPIDJSON_HAS_CXX11_NOEXCEPT +#else +#define RAPIDJSON_NOEXCEPT_ASSERT(x) RAPIDJSON_ASSERT(x) +#endif // RAPIDJSON_ASSERT_THROWS +#endif // RAPIDJSON_NOEXCEPT_ASSERT + /////////////////////////////////////////////////////////////////////////////// // new/delete #ifndef RAPIDJSON_NEW ///! customization point for global \c new -#define RAPIDJSON_NEW(x) new x +#define RAPIDJSON_NEW(TypeName) new TypeName #endif #ifndef RAPIDJSON_DELETE ///! customization point for global \c delete diff --git a/src/3rdparty/rapidjson/reader.h b/src/3rdparty/rapidjson/reader.h index 303aac2e3..44a6bcd30 100644 --- a/src/3rdparty/rapidjson/reader.h +++ b/src/3rdparty/rapidjson/reader.h @@ -33,12 +33,8 @@ #include #elif defined(RAPIDJSON_SSE2) #include -#endif - -#ifdef _MSC_VER -RAPIDJSON_DIAG_PUSH -RAPIDJSON_DIAG_OFF(4127) // conditional expression is constant -RAPIDJSON_DIAG_OFF(4702) // unreachable code +#elif defined(RAPIDJSON_NEON) +#include #endif #ifdef __clang__ @@ -46,6 +42,10 @@ RAPIDJSON_DIAG_PUSH RAPIDJSON_DIAG_OFF(old-style-cast) RAPIDJSON_DIAG_OFF(padded) RAPIDJSON_DIAG_OFF(switch-enum) +#elif defined(_MSC_VER) +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(4127) // conditional expression is constant +RAPIDJSON_DIAG_OFF(4702) // unreachable code #endif #ifdef __GNUC__ @@ -136,7 +136,7 @@ RAPIDJSON_NAMESPACE_BEGIN User can define this as any \c ParseFlag combinations. */ #ifndef RAPIDJSON_PARSE_DEFAULT_FLAGS -#define RAPIDJSON_PARSE_DEFAULT_FLAGS kParseCommentsFlag | kParseTrailingCommasFlag +#define RAPIDJSON_PARSE_DEFAULT_FLAGS kParseNoFlags #endif //! Combination of parseFlags @@ -299,16 +299,9 @@ inline const char *SkipWhitespace_SIMD(const char* p) { for (;; p += 16) { const __m128i s = _mm_load_si128(reinterpret_cast(p)); - const int r = _mm_cvtsi128_si32(_mm_cmpistrm(w, s, _SIDD_UBYTE_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_BIT_MASK | _SIDD_NEGATIVE_POLARITY)); - if (r != 0) { // some of characters is non-whitespace -#ifdef _MSC_VER // Find the index of first non-whitespace - unsigned long offset; - _BitScanForward(&offset, r); - return p + offset; -#else - return p + __builtin_ffs(r) - 1; -#endif - } + const int r = _mm_cmpistri(w, s, _SIDD_UBYTE_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_LEAST_SIGNIFICANT | _SIDD_NEGATIVE_POLARITY); + if (r != 16) // some of characters is non-whitespace + return p + r; } } @@ -325,16 +318,9 @@ inline const char *SkipWhitespace_SIMD(const char* p, const char* end) { for (; p <= end - 16; p += 16) { const __m128i s = _mm_loadu_si128(reinterpret_cast(p)); - const int r = _mm_cvtsi128_si32(_mm_cmpistrm(w, s, _SIDD_UBYTE_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_BIT_MASK | _SIDD_NEGATIVE_POLARITY)); - if (r != 0) { // some of characters is non-whitespace -#ifdef _MSC_VER // Find the index of first non-whitespace - unsigned long offset; - _BitScanForward(&offset, r); - return p + offset; -#else - return p + __builtin_ffs(r) - 1; -#endif - } + const int r = _mm_cmpistri(w, s, _SIDD_UBYTE_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_LEAST_SIGNIFICANT | _SIDD_NEGATIVE_POLARITY); + if (r != 16) // some of characters is non-whitespace + return p + r; } return SkipWhitespace(p, end); @@ -425,7 +411,92 @@ inline const char *SkipWhitespace_SIMD(const char* p, const char* end) { return SkipWhitespace(p, end); } -#endif // RAPIDJSON_SSE2 +#elif defined(RAPIDJSON_NEON) + +//! Skip whitespace with ARM Neon instructions, testing 16 8-byte characters at once. +inline const char *SkipWhitespace_SIMD(const char* p) { + // Fast return for single non-whitespace + if (*p == ' ' || *p == '\n' || *p == '\r' || *p == '\t') + ++p; + else + return p; + + // 16-byte align to the next boundary + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); + while (p != nextAligned) + if (*p == ' ' || *p == '\n' || *p == '\r' || *p == '\t') + ++p; + else + return p; + + const uint8x16_t w0 = vmovq_n_u8(' '); + const uint8x16_t w1 = vmovq_n_u8('\n'); + const uint8x16_t w2 = vmovq_n_u8('\r'); + const uint8x16_t w3 = vmovq_n_u8('\t'); + + for (;; p += 16) { + const uint8x16_t s = vld1q_u8(reinterpret_cast(p)); + uint8x16_t x = vceqq_u8(s, w0); + x = vorrq_u8(x, vceqq_u8(s, w1)); + x = vorrq_u8(x, vceqq_u8(s, w2)); + x = vorrq_u8(x, vceqq_u8(s, w3)); + + x = vmvnq_u8(x); // Negate + x = vrev64q_u8(x); // Rev in 64 + uint64_t low = vgetq_lane_u64(reinterpret_cast(x), 0); // extract + uint64_t high = vgetq_lane_u64(reinterpret_cast(x), 1); // extract + + if (low == 0) { + if (high != 0) { + int lz =__builtin_clzll(high);; + return p + 8 + (lz >> 3); + } + } else { + int lz = __builtin_clzll(low);; + return p + (lz >> 3); + } + } +} + +inline const char *SkipWhitespace_SIMD(const char* p, const char* end) { + // Fast return for single non-whitespace + if (p != end && (*p == ' ' || *p == '\n' || *p == '\r' || *p == '\t')) + ++p; + else + return p; + + const uint8x16_t w0 = vmovq_n_u8(' '); + const uint8x16_t w1 = vmovq_n_u8('\n'); + const uint8x16_t w2 = vmovq_n_u8('\r'); + const uint8x16_t w3 = vmovq_n_u8('\t'); + + for (; p <= end - 16; p += 16) { + const uint8x16_t s = vld1q_u8(reinterpret_cast(p)); + uint8x16_t x = vceqq_u8(s, w0); + x = vorrq_u8(x, vceqq_u8(s, w1)); + x = vorrq_u8(x, vceqq_u8(s, w2)); + x = vorrq_u8(x, vceqq_u8(s, w3)); + + x = vmvnq_u8(x); // Negate + x = vrev64q_u8(x); // Rev in 64 + uint64_t low = vgetq_lane_u64(reinterpret_cast(x), 0); // extract + uint64_t high = vgetq_lane_u64(reinterpret_cast(x), 1); // extract + + if (low == 0) { + if (high != 0) { + int lz = __builtin_clzll(high); + return p + 8 + (lz >> 3); + } + } else { + int lz = __builtin_clzll(low); + return p + (lz >> 3); + } + } + + return SkipWhitespace(p, end); +} + +#endif // RAPIDJSON_NEON #ifdef RAPIDJSON_SIMD //! Template function specialization for InsituStringStream @@ -471,7 +542,8 @@ public: /*! \param stackAllocator Optional allocator for allocating stack memory. (Only use for non-destructive parsing) \param stackCapacity stack capacity in bytes for storing a single decoded string. (Only use for non-destructive parsing) */ - GenericReader(StackAllocator* stackAllocator = 0, size_t stackCapacity = kDefaultStackCapacity) : stack_(stackAllocator, stackCapacity), parseResult_() {} + GenericReader(StackAllocator* stackAllocator = 0, size_t stackCapacity = kDefaultStackCapacity) : + stack_(stackAllocator, stackCapacity), parseResult_(), state_(IterativeParsingStartState) {} //! Parse JSON text. /*! \tparam parseFlags Combination of \ref ParseFlag. @@ -527,7 +599,84 @@ public: return Parse(is, handler); } - //! Whether a parse error has occured in the last parsing. + //! Initialize JSON text token-by-token parsing + /*! + */ + void IterativeParseInit() { + parseResult_.Clear(); + state_ = IterativeParsingStartState; + } + + //! Parse one token from JSON text + /*! \tparam InputStream Type of input stream, implementing Stream concept + \tparam Handler Type of handler, implementing Handler concept. + \param is Input stream to be parsed. + \param handler The handler to receive events. + \return Whether the parsing is successful. + */ + template + bool IterativeParseNext(InputStream& is, Handler& handler) { + while (RAPIDJSON_LIKELY(is.Peek() != '\0')) { + SkipWhitespaceAndComments(is); + + Token t = Tokenize(is.Peek()); + IterativeParsingState n = Predict(state_, t); + IterativeParsingState d = Transit(state_, t, n, is, handler); + + // If we've finished or hit an error... + if (RAPIDJSON_UNLIKELY(IsIterativeParsingCompleteState(d))) { + // Report errors. + if (d == IterativeParsingErrorState) { + HandleError(state_, is); + return false; + } + + // Transition to the finish state. + RAPIDJSON_ASSERT(d == IterativeParsingFinishState); + state_ = d; + + // If StopWhenDone is not set... + if (!(parseFlags & kParseStopWhenDoneFlag)) { + // ... and extra non-whitespace data is found... + SkipWhitespaceAndComments(is); + if (is.Peek() != '\0') { + // ... this is considered an error. + HandleError(state_, is); + return false; + } + } + + // Success! We are done! + return true; + } + + // Transition to the new state. + state_ = d; + + // If we parsed anything other than a delimiter, we invoked the handler, so we can return true now. + if (!IsIterativeParsingDelimiterState(n)) + return true; + } + + // We reached the end of file. + stack_.Clear(); + + if (state_ != IterativeParsingFinishState) { + HandleError(state_, is); + return false; + } + + return true; + } + + //! Check if token-by-token parsing JSON text is complete + /*! \return Whether the JSON has been fully decoded. + */ + RAPIDJSON_FORCEINLINE bool IterativeParseComplete() const { + return IsIterativeParsingCompleteState(state_); + } + + //! Whether a parse error has occurred in the last parsing. bool HasParseError() const { return parseResult_.IsError(); } //! Get the \ref ParseErrorCode of last parsing. @@ -575,7 +724,7 @@ private: } } else if (RAPIDJSON_LIKELY(Consume(is, '/'))) - while (is.Peek() != '\0' && is.Take() != '\n'); + while (is.Peek() != '\0' && is.Take() != '\n') {} else RAPIDJSON_PARSE_ERROR(kParseErrorUnspecificSyntaxError, is.Tell()); @@ -750,7 +899,7 @@ private: return false; } - // Helper function to parse four hexidecimal digits in \uXXXX in ParseString(). + // Helper function to parse four hexadecimal digits in \uXXXX in ParseString(). template unsigned ParseHex4(InputStream& is, size_t escapeOffset) { unsigned codepoint = 0; @@ -857,7 +1006,7 @@ private: Ch c = is.Peek(); if (RAPIDJSON_UNLIKELY(c == '\\')) { // Escape - size_t escapeOffset = is.Tell(); // For invalid escaping, report the inital '\\' as error offset + size_t escapeOffset = is.Tell(); // For invalid escaping, report the initial '\\' as error offset is.Take(); Ch e = is.Peek(); if ((sizeof(Ch) == 1 || unsigned(e) < 256) && RAPIDJSON_LIKELY(escape[static_cast(e)])) { @@ -892,7 +1041,7 @@ private: if (c == '\0') RAPIDJSON_PARSE_ERROR(kParseErrorStringMissQuotationMark, is.Tell()); else - RAPIDJSON_PARSE_ERROR(kParseErrorStringEscapeInvalid, is.Tell()); + RAPIDJSON_PARSE_ERROR(kParseErrorStringInvalidEncoding, is.Tell()); } else { size_t offset = is.Tell(); @@ -927,7 +1076,7 @@ private: // The rest of string using SIMD static const char dquote[16] = { '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"' }; static const char bslash[16] = { '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\' }; - static const char space[16] = { 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19 }; + static const char space[16] = { 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F }; const __m128i dq = _mm_loadu_si128(reinterpret_cast(&dquote[0])); const __m128i bs = _mm_loadu_si128(reinterpret_cast(&bslash[0])); const __m128i sp = _mm_loadu_si128(reinterpret_cast(&space[0])); @@ -936,7 +1085,7 @@ private: const __m128i s = _mm_load_si128(reinterpret_cast(p)); const __m128i t1 = _mm_cmpeq_epi8(s, dq); const __m128i t2 = _mm_cmpeq_epi8(s, bs); - const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x19) == 0x19 + const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x1F) == 0x1F const __m128i x = _mm_or_si128(_mm_or_si128(t1, t2), t3); unsigned short r = static_cast(_mm_movemask_epi8(x)); if (RAPIDJSON_UNLIKELY(r != 0)) { // some of characters is escaped @@ -948,11 +1097,13 @@ private: #else length = static_cast(__builtin_ffs(r) - 1); #endif - char* q = reinterpret_cast(os.Push(length)); - for (size_t i = 0; i < length; i++) - q[i] = p[i]; + if (length != 0) { + char* q = reinterpret_cast(os.Push(length)); + for (size_t i = 0; i < length; i++) + q[i] = p[i]; - p += length; + p += length; + } break; } _mm_storeu_si128(reinterpret_cast<__m128i *>(os.Push(16)), s); @@ -988,7 +1139,7 @@ private: // The rest of string using SIMD static const char dquote[16] = { '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"' }; static const char bslash[16] = { '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\' }; - static const char space[16] = { 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19 }; + static const char space[16] = { 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F }; const __m128i dq = _mm_loadu_si128(reinterpret_cast(&dquote[0])); const __m128i bs = _mm_loadu_si128(reinterpret_cast(&bslash[0])); const __m128i sp = _mm_loadu_si128(reinterpret_cast(&space[0])); @@ -997,7 +1148,7 @@ private: const __m128i s = _mm_load_si128(reinterpret_cast(p)); const __m128i t1 = _mm_cmpeq_epi8(s, dq); const __m128i t2 = _mm_cmpeq_epi8(s, bs); - const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x19) == 0x19 + const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x1F) == 0x1F const __m128i x = _mm_or_si128(_mm_or_si128(t1, t2), t3); unsigned short r = static_cast(_mm_movemask_epi8(x)); if (RAPIDJSON_UNLIKELY(r != 0)) { // some of characters is escaped @@ -1036,7 +1187,7 @@ private: // The rest of string using SIMD static const char dquote[16] = { '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"' }; static const char bslash[16] = { '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\' }; - static const char space[16] = { 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19 }; + static const char space[16] = { 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F }; const __m128i dq = _mm_loadu_si128(reinterpret_cast(&dquote[0])); const __m128i bs = _mm_loadu_si128(reinterpret_cast(&bslash[0])); const __m128i sp = _mm_loadu_si128(reinterpret_cast(&space[0])); @@ -1045,7 +1196,7 @@ private: const __m128i s = _mm_load_si128(reinterpret_cast(p)); const __m128i t1 = _mm_cmpeq_epi8(s, dq); const __m128i t2 = _mm_cmpeq_epi8(s, bs); - const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x19) == 0x19 + const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x1F) == 0x1F const __m128i x = _mm_or_si128(_mm_or_si128(t1, t2), t3); unsigned short r = static_cast(_mm_movemask_epi8(x)); if (RAPIDJSON_UNLIKELY(r != 0)) { // some of characters is escaped @@ -1064,7 +1215,180 @@ private: is.src_ = is.dst_ = p; } -#endif +#elif defined(RAPIDJSON_NEON) + // StringStream -> StackStream + static RAPIDJSON_FORCEINLINE void ScanCopyUnescapedString(StringStream& is, StackStream& os) { + const char* p = is.src_; + + // Scan one by one until alignment (unaligned load may cross page boundary and cause crash) + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); + while (p != nextAligned) + if (RAPIDJSON_UNLIKELY(*p == '\"') || RAPIDJSON_UNLIKELY(*p == '\\') || RAPIDJSON_UNLIKELY(static_cast(*p) < 0x20)) { + is.src_ = p; + return; + } + else + os.Put(*p++); + + // The rest of string using SIMD + const uint8x16_t s0 = vmovq_n_u8('"'); + const uint8x16_t s1 = vmovq_n_u8('\\'); + const uint8x16_t s2 = vmovq_n_u8('\b'); + const uint8x16_t s3 = vmovq_n_u8(32); + + for (;; p += 16) { + const uint8x16_t s = vld1q_u8(reinterpret_cast(p)); + uint8x16_t x = vceqq_u8(s, s0); + x = vorrq_u8(x, vceqq_u8(s, s1)); + x = vorrq_u8(x, vceqq_u8(s, s2)); + x = vorrq_u8(x, vcltq_u8(s, s3)); + + x = vrev64q_u8(x); // Rev in 64 + uint64_t low = vgetq_lane_u64(reinterpret_cast(x), 0); // extract + uint64_t high = vgetq_lane_u64(reinterpret_cast(x), 1); // extract + + SizeType length = 0; + bool escaped = false; + if (low == 0) { + if (high != 0) { + unsigned lz = (unsigned)__builtin_clzll(high);; + length = 8 + (lz >> 3); + escaped = true; + } + } else { + unsigned lz = (unsigned)__builtin_clzll(low);; + length = lz >> 3; + escaped = true; + } + if (RAPIDJSON_UNLIKELY(escaped)) { // some of characters is escaped + if (length != 0) { + char* q = reinterpret_cast(os.Push(length)); + for (size_t i = 0; i < length; i++) + q[i] = p[i]; + + p += length; + } + break; + } + vst1q_u8(reinterpret_cast(os.Push(16)), s); + } + + is.src_ = p; + } + + // InsituStringStream -> InsituStringStream + static RAPIDJSON_FORCEINLINE void ScanCopyUnescapedString(InsituStringStream& is, InsituStringStream& os) { + RAPIDJSON_ASSERT(&is == &os); + (void)os; + + if (is.src_ == is.dst_) { + SkipUnescapedString(is); + return; + } + + char* p = is.src_; + char *q = is.dst_; + + // Scan one by one until alignment (unaligned load may cross page boundary and cause crash) + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); + while (p != nextAligned) + if (RAPIDJSON_UNLIKELY(*p == '\"') || RAPIDJSON_UNLIKELY(*p == '\\') || RAPIDJSON_UNLIKELY(static_cast(*p) < 0x20)) { + is.src_ = p; + is.dst_ = q; + return; + } + else + *q++ = *p++; + + // The rest of string using SIMD + const uint8x16_t s0 = vmovq_n_u8('"'); + const uint8x16_t s1 = vmovq_n_u8('\\'); + const uint8x16_t s2 = vmovq_n_u8('\b'); + const uint8x16_t s3 = vmovq_n_u8(32); + + for (;; p += 16, q += 16) { + const uint8x16_t s = vld1q_u8(reinterpret_cast(p)); + uint8x16_t x = vceqq_u8(s, s0); + x = vorrq_u8(x, vceqq_u8(s, s1)); + x = vorrq_u8(x, vceqq_u8(s, s2)); + x = vorrq_u8(x, vcltq_u8(s, s3)); + + x = vrev64q_u8(x); // Rev in 64 + uint64_t low = vgetq_lane_u64(reinterpret_cast(x), 0); // extract + uint64_t high = vgetq_lane_u64(reinterpret_cast(x), 1); // extract + + SizeType length = 0; + bool escaped = false; + if (low == 0) { + if (high != 0) { + unsigned lz = (unsigned)__builtin_clzll(high); + length = 8 + (lz >> 3); + escaped = true; + } + } else { + unsigned lz = (unsigned)__builtin_clzll(low); + length = lz >> 3; + escaped = true; + } + if (RAPIDJSON_UNLIKELY(escaped)) { // some of characters is escaped + for (const char* pend = p + length; p != pend; ) { + *q++ = *p++; + } + break; + } + vst1q_u8(reinterpret_cast(q), s); + } + + is.src_ = p; + is.dst_ = q; + } + + // When read/write pointers are the same for insitu stream, just skip unescaped characters + static RAPIDJSON_FORCEINLINE void SkipUnescapedString(InsituStringStream& is) { + RAPIDJSON_ASSERT(is.src_ == is.dst_); + char* p = is.src_; + + // Scan one by one until alignment (unaligned load may cross page boundary and cause crash) + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); + for (; p != nextAligned; p++) + if (RAPIDJSON_UNLIKELY(*p == '\"') || RAPIDJSON_UNLIKELY(*p == '\\') || RAPIDJSON_UNLIKELY(static_cast(*p) < 0x20)) { + is.src_ = is.dst_ = p; + return; + } + + // The rest of string using SIMD + const uint8x16_t s0 = vmovq_n_u8('"'); + const uint8x16_t s1 = vmovq_n_u8('\\'); + const uint8x16_t s2 = vmovq_n_u8('\b'); + const uint8x16_t s3 = vmovq_n_u8(32); + + for (;; p += 16) { + const uint8x16_t s = vld1q_u8(reinterpret_cast(p)); + uint8x16_t x = vceqq_u8(s, s0); + x = vorrq_u8(x, vceqq_u8(s, s1)); + x = vorrq_u8(x, vceqq_u8(s, s2)); + x = vorrq_u8(x, vcltq_u8(s, s3)); + + x = vrev64q_u8(x); // Rev in 64 + uint64_t low = vgetq_lane_u64(reinterpret_cast(x), 0); // extract + uint64_t high = vgetq_lane_u64(reinterpret_cast(x), 1); // extract + + if (low == 0) { + if (high != 0) { + int lz = __builtin_clzll(high); + p += 8 + (lz >> 3); + break; + } + } else { + int lz = __builtin_clzll(low); + p += lz >> 3; + break; + } + } + + is.src_ = is.dst_ = p; + } +#endif // RAPIDJSON_NEON template class NumberStream; @@ -1075,7 +1399,6 @@ private: typedef typename InputStream::Ch Ch; NumberStream(GenericReader& reader, InputStream& s) : is(s) { (void)reader; } - ~NumberStream() {} RAPIDJSON_FORCEINLINE Ch Peek() const { return is.Peek(); } RAPIDJSON_FORCEINLINE Ch TakePush() { return is.Take(); } @@ -1097,7 +1420,6 @@ private: typedef NumberStream Base; public: NumberStream(GenericReader& reader, InputStream& is) : Base(reader, is), stackStream(reader.stack_) {} - ~NumberStream() {} RAPIDJSON_FORCEINLINE Ch TakePush() { stackStream.Put(static_cast(Base::is.Peek())); @@ -1124,7 +1446,6 @@ private: typedef NumberStream Base; public: NumberStream(GenericReader& reader, InputStream& is) : Base(reader, is) {} - ~NumberStream() {} RAPIDJSON_FORCEINLINE Ch Take() { return Base::TakePush(); } }; @@ -1185,18 +1506,27 @@ private: } // Parse NaN or Infinity here else if ((parseFlags & kParseNanAndInfFlag) && RAPIDJSON_LIKELY((s.Peek() == 'I' || s.Peek() == 'N'))) { - useNanOrInf = true; - if (RAPIDJSON_LIKELY(Consume(s, 'N') && Consume(s, 'a') && Consume(s, 'N'))) { - d = std::numeric_limits::quiet_NaN(); + if (Consume(s, 'N')) { + if (Consume(s, 'a') && Consume(s, 'N')) { + d = std::numeric_limits::quiet_NaN(); + useNanOrInf = true; + } } - else if (RAPIDJSON_LIKELY(Consume(s, 'I') && Consume(s, 'n') && Consume(s, 'f'))) { - d = (minus ? -std::numeric_limits::infinity() : std::numeric_limits::infinity()); - if (RAPIDJSON_UNLIKELY(s.Peek() == 'i' && !(Consume(s, 'i') && Consume(s, 'n') - && Consume(s, 'i') && Consume(s, 't') && Consume(s, 'y')))) - RAPIDJSON_PARSE_ERROR(kParseErrorValueInvalid, s.Tell()); + else if (RAPIDJSON_LIKELY(Consume(s, 'I'))) { + if (Consume(s, 'n') && Consume(s, 'f')) { + d = (minus ? -std::numeric_limits::infinity() : std::numeric_limits::infinity()); + useNanOrInf = true; + + if (RAPIDJSON_UNLIKELY(s.Peek() == 'i' && !(Consume(s, 'i') && Consume(s, 'n') + && Consume(s, 'i') && Consume(s, 't') && Consume(s, 'y')))) { + RAPIDJSON_PARSE_ERROR(kParseErrorValueInvalid, s.Tell()); + } + } } - else + + if (RAPIDJSON_UNLIKELY(!useNanOrInf)) { RAPIDJSON_PARSE_ERROR(kParseErrorValueInvalid, s.Tell()); + } } else RAPIDJSON_PARSE_ERROR(kParseErrorValueInvalid, s.Tell()); @@ -1231,8 +1561,6 @@ private: // Force double for big integer if (useDouble) { while (RAPIDJSON_LIKELY(s.Peek() >= '0' && s.Peek() <= '9')) { - if (RAPIDJSON_UNLIKELY(d >= 1.7976931348623157e307)) // DBL_MAX / 10.0 - RAPIDJSON_PARSE_ERROR(kParseErrorNumberTooBig, startOffset); d = d * 10 + (s.TakePush() - '0'); } } @@ -1302,9 +1630,18 @@ private: if (RAPIDJSON_LIKELY(s.Peek() >= '0' && s.Peek() <= '9')) { exp = static_cast(s.Take() - '0'); if (expMinus) { + // (exp + expFrac) must not underflow int => we're detecting when -exp gets + // dangerously close to INT_MIN (a pessimistic next digit 9 would push it into + // underflow territory): + // + // -(exp * 10 + 9) + expFrac >= INT_MIN + // <=> exp <= (expFrac - INT_MIN - 9) / 10 + RAPIDJSON_ASSERT(expFrac <= 0); + int maxExp = (expFrac + 2147483639) / 10; + while (RAPIDJSON_LIKELY(s.Peek() >= '0' && s.Peek() <= '9')) { exp = exp * 10 + static_cast(s.Take() - '0'); - if (exp >= 214748364) { // Issue #313: prevent overflow exponent + if (RAPIDJSON_UNLIKELY(exp > maxExp)) { while (RAPIDJSON_UNLIKELY(s.Peek() >= '0' && s.Peek() <= '9')) // Consume the rest of exponent s.Take(); } @@ -1363,6 +1700,13 @@ private: else d = internal::StrtodNormalPrecision(d, p); + // Use > max, instead of == inf, to fix bogus warning -Wfloat-equal + if (d > (std::numeric_limits::max)()) { + // Overflow + // TODO: internal::StrtodX should report overflow (or underflow) + RAPIDJSON_PARSE_ERROR(kParseErrorNumberTooBig, startOffset); + } + cont = handler.Double(minus ? -d : d); } else if (useNanOrInf) { @@ -1408,29 +1752,31 @@ private: // States enum IterativeParsingState { - IterativeParsingStartState = 0, - IterativeParsingFinishState, - IterativeParsingErrorState, + IterativeParsingFinishState = 0, // sink states at top + IterativeParsingErrorState, // sink states at top + IterativeParsingStartState, // Object states IterativeParsingObjectInitialState, IterativeParsingMemberKeyState, - IterativeParsingKeyValueDelimiterState, IterativeParsingMemberValueState, - IterativeParsingMemberDelimiterState, IterativeParsingObjectFinishState, // Array states IterativeParsingArrayInitialState, IterativeParsingElementState, - IterativeParsingElementDelimiterState, IterativeParsingArrayFinishState, // Single value state - IterativeParsingValueState - }; + IterativeParsingValueState, - enum { cIterativeParsingStateCount = IterativeParsingValueState + 1 }; + // Delimiter states (at bottom) + IterativeParsingElementDelimiterState, + IterativeParsingMemberDelimiterState, + IterativeParsingKeyValueDelimiterState, + + cIterativeParsingStateCount + }; // Tokens enum Token { @@ -1452,7 +1798,7 @@ private: kTokenCount }; - RAPIDJSON_FORCEINLINE Token Tokenize(Ch c) { + RAPIDJSON_FORCEINLINE Token Tokenize(Ch c) const { //!@cond RAPIDJSON_HIDDEN_FROM_DOXYGEN #define N NumberToken @@ -1479,9 +1825,21 @@ private: return NumberToken; } - RAPIDJSON_FORCEINLINE IterativeParsingState Predict(IterativeParsingState state, Token token) { + RAPIDJSON_FORCEINLINE IterativeParsingState Predict(IterativeParsingState state, Token token) const { // current state x one lookahead token -> new state static const char G[cIterativeParsingStateCount][kTokenCount] = { + // Finish(sink state) + { + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState + }, + // Error(sink state) + { + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState + }, // Start { IterativeParsingArrayInitialState, // Left bracket @@ -1496,18 +1854,6 @@ private: IterativeParsingValueState, // Null IterativeParsingValueState // Number }, - // Finish(sink state) - { - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState - }, - // Error(sink state) - { - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState - }, // ObjectInitial { IterativeParsingErrorState, // Left bracket @@ -1536,20 +1882,6 @@ private: IterativeParsingErrorState, // Null IterativeParsingErrorState // Number }, - // KeyValueDelimiter - { - IterativeParsingArrayInitialState, // Left bracket(push MemberValue state) - IterativeParsingErrorState, // Right bracket - IterativeParsingObjectInitialState, // Left curly bracket(push MemberValue state) - IterativeParsingErrorState, // Right curly bracket - IterativeParsingErrorState, // Comma - IterativeParsingErrorState, // Colon - IterativeParsingMemberValueState, // String - IterativeParsingMemberValueState, // False - IterativeParsingMemberValueState, // True - IterativeParsingMemberValueState, // Null - IterativeParsingMemberValueState // Number - }, // MemberValue { IterativeParsingErrorState, // Left bracket @@ -1564,20 +1896,6 @@ private: IterativeParsingErrorState, // Null IterativeParsingErrorState // Number }, - // MemberDelimiter - { - IterativeParsingErrorState, // Left bracket - IterativeParsingErrorState, // Right bracket - IterativeParsingErrorState, // Left curly bracket - IterativeParsingObjectFinishState, // Right curly bracket - IterativeParsingErrorState, // Comma - IterativeParsingErrorState, // Colon - IterativeParsingMemberKeyState, // String - IterativeParsingErrorState, // False - IterativeParsingErrorState, // True - IterativeParsingErrorState, // Null - IterativeParsingErrorState // Number - }, // ObjectFinish(sink state) { IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, @@ -1612,6 +1930,18 @@ private: IterativeParsingErrorState, // Null IterativeParsingErrorState // Number }, + // ArrayFinish(sink state) + { + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState + }, + // Single Value (sink state) + { + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, + IterativeParsingErrorState + }, // ElementDelimiter { IterativeParsingArrayInitialState, // Left bracket(push Element state) @@ -1626,18 +1956,34 @@ private: IterativeParsingElementState, // Null IterativeParsingElementState // Number }, - // ArrayFinish(sink state) + // MemberDelimiter { - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState + IterativeParsingErrorState, // Left bracket + IterativeParsingErrorState, // Right bracket + IterativeParsingErrorState, // Left curly bracket + IterativeParsingObjectFinishState, // Right curly bracket + IterativeParsingErrorState, // Comma + IterativeParsingErrorState, // Colon + IterativeParsingMemberKeyState, // String + IterativeParsingErrorState, // False + IterativeParsingErrorState, // True + IterativeParsingErrorState, // Null + IterativeParsingErrorState // Number }, - // Single Value (sink state) + // KeyValueDelimiter { - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, IterativeParsingErrorState, - IterativeParsingErrorState - } + IterativeParsingArrayInitialState, // Left bracket(push MemberValue state) + IterativeParsingErrorState, // Right bracket + IterativeParsingObjectInitialState, // Left curly bracket(push MemberValue state) + IterativeParsingErrorState, // Right curly bracket + IterativeParsingErrorState, // Comma + IterativeParsingErrorState, // Colon + IterativeParsingMemberValueState, // String + IterativeParsingMemberValueState, // False + IterativeParsingMemberValueState, // True + IterativeParsingMemberValueState, // Null + IterativeParsingMemberValueState // Number + }, }; // End of G return static_cast(G[state][token]); @@ -1818,6 +2164,14 @@ private: } } + RAPIDJSON_FORCEINLINE bool IsIterativeParsingDelimiterState(IterativeParsingState s) const { + return s >= IterativeParsingElementDelimiterState; + } + + RAPIDJSON_FORCEINLINE bool IsIterativeParsingCompleteState(IterativeParsingState s) const { + return s <= IterativeParsingErrorState; + } + template ParseResult IterativeParse(InputStream& is, Handler& handler) { parseResult_.Clear(); @@ -1856,6 +2210,7 @@ private: static const size_t kDefaultStackCapacity = 256; //!< Default stack capacity in bytes for storing a single decoded string. internal::Stack stack_; //!< A stack for storing decoded string temporarily during non-destructive parsing. ParseResult parseResult_; + IterativeParsingState state_; }; // class GenericReader //! Reader with UTF8 encoding and default allocator. @@ -1863,7 +2218,7 @@ typedef GenericReader, UTF8<> > Reader; RAPIDJSON_NAMESPACE_END -#ifdef __clang__ +#if defined(__clang__) || defined(_MSC_VER) RAPIDJSON_DIAG_POP #endif @@ -1872,8 +2227,4 @@ RAPIDJSON_DIAG_POP RAPIDJSON_DIAG_POP #endif -#ifdef _MSC_VER -RAPIDJSON_DIAG_POP -#endif - #endif // RAPIDJSON_READER_H_ diff --git a/src/3rdparty/rapidjson/schema.h b/src/3rdparty/rapidjson/schema.h index b182aa27f..26ae94748 100644 --- a/src/3rdparty/rapidjson/schema.h +++ b/src/3rdparty/rapidjson/schema.h @@ -17,6 +17,7 @@ #include "document.h" #include "pointer.h" +#include "stringbuffer.h" #include // abs, floor #if !defined(RAPIDJSON_SCHEMA_USE_INTERNALREGEX) @@ -25,7 +26,7 @@ #define RAPIDJSON_SCHEMA_USE_INTERNALREGEX 0 #endif -#if !RAPIDJSON_SCHEMA_USE_INTERNALREGEX && !defined(RAPIDJSON_SCHEMA_USE_STDREGEX) && (__cplusplus >=201103L || (defined(_MSC_VER) && _MSC_VER >= 1800)) +#if !RAPIDJSON_SCHEMA_USE_INTERNALREGEX && defined(RAPIDJSON_SCHEMA_USE_STDREGEX) && (__cplusplus >=201103L || (defined(_MSC_VER) && _MSC_VER >= 1800)) #define RAPIDJSON_SCHEMA_USE_STDREGEX 1 #else #define RAPIDJSON_SCHEMA_USE_STDREGEX 0 @@ -62,9 +63,7 @@ RAPIDJSON_DIAG_OFF(weak-vtables) RAPIDJSON_DIAG_OFF(exit-time-destructors) RAPIDJSON_DIAG_OFF(c++98-compat-pedantic) RAPIDJSON_DIAG_OFF(variadic-macros) -#endif - -#ifdef _MSC_VER +#elif defined(_MSC_VER) RAPIDJSON_DIAG_OFF(4512) // assignment operator could not be generated #endif @@ -157,6 +156,62 @@ public: virtual void FreeState(void* p) = 0; }; +/////////////////////////////////////////////////////////////////////////////// +// IValidationErrorHandler + +template +class IValidationErrorHandler { +public: + typedef typename SchemaType::Ch Ch; + typedef typename SchemaType::SValue SValue; + + virtual ~IValidationErrorHandler() {} + + virtual void NotMultipleOf(int64_t actual, const SValue& expected) = 0; + virtual void NotMultipleOf(uint64_t actual, const SValue& expected) = 0; + virtual void NotMultipleOf(double actual, const SValue& expected) = 0; + virtual void AboveMaximum(int64_t actual, const SValue& expected, bool exclusive) = 0; + virtual void AboveMaximum(uint64_t actual, const SValue& expected, bool exclusive) = 0; + virtual void AboveMaximum(double actual, const SValue& expected, bool exclusive) = 0; + virtual void BelowMinimum(int64_t actual, const SValue& expected, bool exclusive) = 0; + virtual void BelowMinimum(uint64_t actual, const SValue& expected, bool exclusive) = 0; + virtual void BelowMinimum(double actual, const SValue& expected, bool exclusive) = 0; + + virtual void TooLong(const Ch* str, SizeType length, SizeType expected) = 0; + virtual void TooShort(const Ch* str, SizeType length, SizeType expected) = 0; + virtual void DoesNotMatch(const Ch* str, SizeType length) = 0; + + virtual void DisallowedItem(SizeType index) = 0; + virtual void TooFewItems(SizeType actualCount, SizeType expectedCount) = 0; + virtual void TooManyItems(SizeType actualCount, SizeType expectedCount) = 0; + virtual void DuplicateItems(SizeType index1, SizeType index2) = 0; + + virtual void TooManyProperties(SizeType actualCount, SizeType expectedCount) = 0; + virtual void TooFewProperties(SizeType actualCount, SizeType expectedCount) = 0; + virtual void StartMissingProperties() = 0; + virtual void AddMissingProperty(const SValue& name) = 0; + virtual bool EndMissingProperties() = 0; + virtual void PropertyViolations(ISchemaValidator** subvalidators, SizeType count) = 0; + virtual void DisallowedProperty(const Ch* name, SizeType length) = 0; + + virtual void StartDependencyErrors() = 0; + virtual void StartMissingDependentProperties() = 0; + virtual void AddMissingDependentProperty(const SValue& targetName) = 0; + virtual void EndMissingDependentProperties(const SValue& sourceName) = 0; + virtual void AddDependencySchemaError(const SValue& souceName, ISchemaValidator* subvalidator) = 0; + virtual bool EndDependencyErrors() = 0; + + virtual void DisallowedValue() = 0; + virtual void StartDisallowedType() = 0; + virtual void AddExpectedType(const typename SchemaType::ValueType& expectedType) = 0; + virtual void EndDisallowedType(const typename SchemaType::ValueType& actualType) = 0; + virtual void NotAllOf(ISchemaValidator** subvalidators, SizeType count) = 0; + virtual void NoneOf(ISchemaValidator** subvalidators, SizeType count) = 0; + virtual void NotOneOf(ISchemaValidator** subvalidators, SizeType count) = 0; + virtual void Disallowed() = 0; +}; + + /////////////////////////////////////////////////////////////////////////////// // Hasher @@ -261,6 +316,7 @@ template struct SchemaValidationContext { typedef Schema SchemaType; typedef ISchemaStateFactory SchemaValidatorFactoryType; + typedef IValidationErrorHandler ErrorHandlerType; typedef typename SchemaType::ValueType ValueType; typedef typename ValueType::Ch Ch; @@ -270,8 +326,9 @@ struct SchemaValidationContext { kPatternValidatorWithAdditionalProperty }; - SchemaValidationContext(SchemaValidatorFactoryType& f, const SchemaType* s) : + SchemaValidationContext(SchemaValidatorFactoryType& f, ErrorHandlerType& eh, const SchemaType* s) : factory(f), + error_handler(eh), schema(s), valueSchema(), invalidKeyword(), @@ -311,6 +368,7 @@ struct SchemaValidationContext { } SchemaValidatorFactoryType& factory; + ErrorHandlerType& error_handler; const SchemaType* schema; const SchemaType* valueSchema; const Ch* invalidKeyword; @@ -345,15 +403,20 @@ public: typedef SchemaValidationContext Context; typedef Schema SchemaType; typedef GenericValue SValue; + typedef IValidationErrorHandler ErrorHandler; friend class GenericSchemaDocument; Schema(SchemaDocumentType* schemaDocument, const PointerType& p, const ValueType& value, const ValueType& document, AllocatorType* allocator) : allocator_(allocator), + uri_(schemaDocument->GetURI(), *allocator), + pointer_(p, allocator), + typeless_(schemaDocument->GetTypeless()), enum_(), enumCount_(), not_(), type_((1 << kTotalSchemaType) - 1), // typeless validatorCount_(), + notValidatorIndex_(), properties_(), additionalPropertiesSchema_(), patternProperties_(), @@ -377,7 +440,8 @@ public: minLength_(0), maxLength_(~SizeType(0)), exclusiveMinimum_(false), - exclusiveMaximum_(false) + exclusiveMaximum_(false), + defaultValueLength_(0) { typedef typename SchemaDocumentType::ValueType ValueType; typedef typename ValueType::ConstValueIterator ConstValueIterator; @@ -400,7 +464,7 @@ public: enum_ = static_cast(allocator_->Malloc(sizeof(uint64_t) * v->Size())); for (ConstValueIterator itr = v->Begin(); itr != v->End(); ++itr) { typedef Hasher > EnumHasherType; - char buffer[256 + 24]; + char buffer[256u + 24]; MemoryPoolAllocator<> hasherAllocator(buffer, sizeof(buffer)); EnumHasherType h(&hasherAllocator, 256); itr->Accept(h); @@ -453,7 +517,7 @@ public: for (SizeType i = 0; i < propertyCount_; i++) { new (&properties_[i]) Property(); properties_[i].name = allProperties[i]; - properties_[i].schema = GetTypeless(); + properties_[i].schema = typeless_; } } } @@ -572,12 +636,16 @@ public: if (const ValueType* v = GetMember(value, GetMultipleOfString())) if (v->IsNumber() && v->GetDouble() > 0.0) multipleOf_.CopyFrom(*v, *allocator_); + + // Default + if (const ValueType* v = GetMember(value, GetDefaultValueString())) + if (v->IsString()) + defaultValueLength_ = v->GetStringLength(); + } ~Schema() { - if (allocator_) { - allocator_->Free(enum_); - } + AllocatorType::Free(enum_); if (properties_) { for (SizeType i = 0; i < propertyCount_; i++) properties_[i].~Property(); @@ -592,11 +660,19 @@ public: #if RAPIDJSON_SCHEMA_HAS_REGEX if (pattern_) { pattern_->~RegexType(); - allocator_->Free(pattern_); + AllocatorType::Free(pattern_); } #endif } + const SValue& GetURI() const { + return uri_; + } + + const PointerType& GetPointer() const { + return pointer_; + } + bool BeginValue(Context& context) const { if (context.inArray) { if (uniqueItems_) @@ -610,12 +686,14 @@ public: else if (additionalItemsSchema_) context.valueSchema = additionalItemsSchema_; else if (additionalItems_) - context.valueSchema = GetTypeless(); - else + context.valueSchema = typeless_; + else { + context.error_handler.DisallowedItem(context.arrayElementIndex); RAPIDJSON_INVALID_KEYWORD_RETURN(GetItemsString()); + } } else - context.valueSchema = GetTypeless(); + context.valueSchema = typeless_; context.arrayElementIndex++; } @@ -637,15 +715,21 @@ public: } if (context.objectPatternValidatorType == Context::kPatternValidatorOnly) { - if (!patternValid) + if (!patternValid) { + context.error_handler.PropertyViolations(context.patternPropertiesValidators, count); RAPIDJSON_INVALID_KEYWORD_RETURN(GetPatternPropertiesString()); + } } else if (context.objectPatternValidatorType == Context::kPatternValidatorWithProperty) { - if (!patternValid || !otherValid) + if (!patternValid || !otherValid) { + context.error_handler.PropertyViolations(context.patternPropertiesValidators, count + 1); RAPIDJSON_INVALID_KEYWORD_RETURN(GetPatternPropertiesString()); + } } - else if (!patternValid && !otherValid) // kPatternValidatorWithAdditionalProperty) + else if (!patternValid && !otherValid) { // kPatternValidatorWithAdditionalProperty) + context.error_handler.PropertyViolations(context.patternPropertiesValidators, count + 1); RAPIDJSON_INVALID_KEYWORD_RETURN(GetPatternPropertiesString()); + } } if (enum_) { @@ -653,19 +737,23 @@ public: for (SizeType i = 0; i < enumCount_; i++) if (enum_[i] == h) goto foundEnum; + context.error_handler.DisallowedValue(); RAPIDJSON_INVALID_KEYWORD_RETURN(GetEnumString()); foundEnum:; } if (allOf_.schemas) for (SizeType i = allOf_.begin; i < allOf_.begin + allOf_.count; i++) - if (!context.validators[i]->IsValid()) + if (!context.validators[i]->IsValid()) { + context.error_handler.NotAllOf(&context.validators[allOf_.begin], allOf_.count); RAPIDJSON_INVALID_KEYWORD_RETURN(GetAllOfString()); + } if (anyOf_.schemas) { for (SizeType i = anyOf_.begin; i < anyOf_.begin + anyOf_.count; i++) if (context.validators[i]->IsValid()) goto foundAny; + context.error_handler.NoneOf(&context.validators[anyOf_.begin], anyOf_.count); RAPIDJSON_INVALID_KEYWORD_RETURN(GetAnyOfString()); foundAny:; } @@ -674,30 +762,39 @@ public: bool oneValid = false; for (SizeType i = oneOf_.begin; i < oneOf_.begin + oneOf_.count; i++) if (context.validators[i]->IsValid()) { - if (oneValid) + if (oneValid) { + context.error_handler.NotOneOf(&context.validators[oneOf_.begin], oneOf_.count); RAPIDJSON_INVALID_KEYWORD_RETURN(GetOneOfString()); - else + } else oneValid = true; } - if (!oneValid) + if (!oneValid) { + context.error_handler.NotOneOf(&context.validators[oneOf_.begin], oneOf_.count); RAPIDJSON_INVALID_KEYWORD_RETURN(GetOneOfString()); + } } - if (not_ && context.validators[notValidatorIndex_]->IsValid()) + if (not_ && context.validators[notValidatorIndex_]->IsValid()) { + context.error_handler.Disallowed(); RAPIDJSON_INVALID_KEYWORD_RETURN(GetNotString()); + } return true; } - bool Null(Context& context) const { - if (!(type_ & (1 << kNullSchemaType))) + bool Null(Context& context) const { + if (!(type_ & (1 << kNullSchemaType))) { + DisallowedType(context, GetNullString()); RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + } return CreateParallelValidator(context); } - bool Bool(Context& context, bool) const { - if (!(type_ & (1 << kBooleanSchemaType))) + bool Bool(Context& context, bool) const { + if (!(type_ & (1 << kBooleanSchemaType))) { + DisallowedType(context, GetBooleanString()); RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + } return CreateParallelValidator(context); } @@ -726,8 +823,10 @@ public: } bool Double(Context& context, double d) const { - if (!(type_ & (1 << kNumberSchemaType))) + if (!(type_ & (1 << kNumberSchemaType))) { + DisallowedType(context, GetNumberString()); RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + } if (!minimum_.IsNull() && !CheckDoubleMinimum(context, d)) return false; @@ -742,28 +841,38 @@ public: } bool String(Context& context, const Ch* str, SizeType length, bool) const { - if (!(type_ & (1 << kStringSchemaType))) + if (!(type_ & (1 << kStringSchemaType))) { + DisallowedType(context, GetStringString()); RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + } if (minLength_ != 0 || maxLength_ != SizeType(~0)) { SizeType count; if (internal::CountStringCodePoint(str, length, &count)) { - if (count < minLength_) + if (count < minLength_) { + context.error_handler.TooShort(str, length, minLength_); RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinLengthString()); - if (count > maxLength_) + } + if (count > maxLength_) { + context.error_handler.TooLong(str, length, maxLength_); RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaxLengthString()); + } } } - if (pattern_ && !IsPatternMatch(pattern_, str, length)) + if (pattern_ && !IsPatternMatch(pattern_, str, length)) { + context.error_handler.DoesNotMatch(str, length); RAPIDJSON_INVALID_KEYWORD_RETURN(GetPatternString()); + } return CreateParallelValidator(context); } - bool StartObject(Context& context) const { - if (!(type_ & (1 << kObjectSchemaType))) + bool StartObject(Context& context) const { + if (!(type_ & (1 << kObjectSchemaType))) { + DisallowedType(context, GetObjectString()); RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + } if (hasDependencies_ || hasRequired_) { context.propertyExist = static_cast(context.factory.MallocState(sizeof(bool) * propertyCount_)); @@ -784,15 +893,17 @@ public: if (patternProperties_) { context.patternPropertiesSchemaCount = 0; for (SizeType i = 0; i < patternPropertyCount_; i++) - if (patternProperties_[i].pattern && IsPatternMatch(patternProperties_[i].pattern, str, len)) + if (patternProperties_[i].pattern && IsPatternMatch(patternProperties_[i].pattern, str, len)) { context.patternPropertiesSchemas[context.patternPropertiesSchemaCount++] = patternProperties_[i].schema; + context.valueSchema = typeless_; + } } SizeType index; if (FindPropertyIndex(ValueType(str, len).Move(), &index)) { if (context.patternPropertiesSchemaCount > 0) { context.patternPropertiesSchemas[context.patternPropertiesSchemaCount++] = properties_[index].schema; - context.valueSchema = GetTypeless(); + context.valueSchema = typeless_; context.valuePatternValidatorType = Context::kPatternValidatorWithProperty; } else @@ -807,7 +918,7 @@ public: if (additionalPropertiesSchema_) { if (additionalPropertiesSchema_ && context.patternPropertiesSchemaCount > 0) { context.patternPropertiesSchemas[context.patternPropertiesSchemaCount++] = additionalPropertiesSchema_; - context.valueSchema = GetTypeless(); + context.valueSchema = typeless_; context.valuePatternValidatorType = Context::kPatternValidatorWithAdditionalProperty; } else @@ -815,49 +926,70 @@ public: return true; } else if (additionalProperties_) { - context.valueSchema = GetTypeless(); + context.valueSchema = typeless_; return true; } - if (context.patternPropertiesSchemaCount == 0) // patternProperties are not additional properties + if (context.patternPropertiesSchemaCount == 0) { // patternProperties are not additional properties + context.error_handler.DisallowedProperty(str, len); RAPIDJSON_INVALID_KEYWORD_RETURN(GetAdditionalPropertiesString()); + } return true; } bool EndObject(Context& context, SizeType memberCount) const { - if (hasRequired_) + if (hasRequired_) { + context.error_handler.StartMissingProperties(); for (SizeType index = 0; index < propertyCount_; index++) - if (properties_[index].required) - if (!context.propertyExist[index]) - RAPIDJSON_INVALID_KEYWORD_RETURN(GetRequiredString()); + if (properties_[index].required && !context.propertyExist[index]) + if (properties_[index].schema->defaultValueLength_ == 0 ) + context.error_handler.AddMissingProperty(properties_[index].name); + if (context.error_handler.EndMissingProperties()) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetRequiredString()); + } - if (memberCount < minProperties_) + if (memberCount < minProperties_) { + context.error_handler.TooFewProperties(memberCount, minProperties_); RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinPropertiesString()); + } - if (memberCount > maxProperties_) + if (memberCount > maxProperties_) { + context.error_handler.TooManyProperties(memberCount, maxProperties_); RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaxPropertiesString()); + } if (hasDependencies_) { - for (SizeType sourceIndex = 0; sourceIndex < propertyCount_; sourceIndex++) + context.error_handler.StartDependencyErrors(); + for (SizeType sourceIndex = 0; sourceIndex < propertyCount_; sourceIndex++) { + const Property& source = properties_[sourceIndex]; if (context.propertyExist[sourceIndex]) { - if (properties_[sourceIndex].dependencies) { + if (source.dependencies) { + context.error_handler.StartMissingDependentProperties(); for (SizeType targetIndex = 0; targetIndex < propertyCount_; targetIndex++) - if (properties_[sourceIndex].dependencies[targetIndex] && !context.propertyExist[targetIndex]) - RAPIDJSON_INVALID_KEYWORD_RETURN(GetDependenciesString()); + if (source.dependencies[targetIndex] && !context.propertyExist[targetIndex]) + context.error_handler.AddMissingDependentProperty(properties_[targetIndex].name); + context.error_handler.EndMissingDependentProperties(source.name); + } + else if (source.dependenciesSchema) { + ISchemaValidator* dependenciesValidator = context.validators[source.dependenciesValidatorIndex]; + if (!dependenciesValidator->IsValid()) + context.error_handler.AddDependencySchemaError(source.name, dependenciesValidator); } - else if (properties_[sourceIndex].dependenciesSchema) - if (!context.validators[properties_[sourceIndex].dependenciesValidatorIndex]->IsValid()) - RAPIDJSON_INVALID_KEYWORD_RETURN(GetDependenciesString()); } + } + if (context.error_handler.EndDependencyErrors()) + RAPIDJSON_INVALID_KEYWORD_RETURN(GetDependenciesString()); } return true; } - bool StartArray(Context& context) const { - if (!(type_ & (1 << kArraySchemaType))) + bool StartArray(Context& context) const { + if (!(type_ & (1 << kArraySchemaType))) { + DisallowedType(context, GetArrayString()); RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + } context.arrayElementIndex = 0; context.inArray = true; @@ -865,14 +997,18 @@ public: return CreateParallelValidator(context); } - bool EndArray(Context& context, SizeType elementCount) const { + bool EndArray(Context& context, SizeType elementCount) const { context.inArray = false; - if (elementCount < minItems_) + if (elementCount < minItems_) { + context.error_handler.TooFewItems(elementCount, minItems_); RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinItemsString()); + } - if (elementCount > maxItems_) + if (elementCount > maxItems_) { + context.error_handler.TooManyItems(elementCount, maxItems_); RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaxItemsString()); + } return true; } @@ -881,7 +1017,7 @@ public: #define RAPIDJSON_STRING_(name, ...) \ static const ValueType& Get##name##String() {\ static const Ch s[] = { __VA_ARGS__, '\0' };\ - static const ValueType v(s, sizeof(s) / sizeof(Ch) - 1);\ + static const ValueType v(s, static_cast(sizeof(s) / sizeof(Ch) - 1));\ return v;\ } @@ -918,6 +1054,7 @@ public: RAPIDJSON_STRING_(ExclusiveMinimum, 'e', 'x', 'c', 'l', 'u', 's', 'i', 'v', 'e', 'M', 'i', 'n', 'i', 'm', 'u', 'm') RAPIDJSON_STRING_(ExclusiveMaximum, 'e', 'x', 'c', 'l', 'u', 's', 'i', 'v', 'e', 'M', 'a', 'x', 'i', 'm', 'u', 'm') RAPIDJSON_STRING_(MultipleOf, 'm', 'u', 'l', 't', 'i', 'p', 'l', 'e', 'O', 'f') + RAPIDJSON_STRING_(DefaultValue, 'd', 'e', 'f', 'a', 'u', 'l', 't') #undef RAPIDJSON_STRING_ @@ -934,7 +1071,7 @@ private: }; #if RAPIDJSON_SCHEMA_USE_INTERNALREGEX - typedef internal::GenericRegex RegexType; + typedef internal::GenericRegex RegexType; #elif RAPIDJSON_SCHEMA_USE_STDREGEX typedef std::basic_regex RegexType; #else @@ -949,11 +1086,6 @@ private: SizeType count; }; - static const SchemaType* GetTypeless() { - static SchemaType typeless(0, PointerType(), ValueType(kObjectType).Move(), ValueType(kObjectType).Move(), 0); - return &typeless; - } - template void AddUniqueElement(V1& a, const V2& v) { for (typename V1::ConstValueIterator itr = a.Begin(); itr != a.End(); ++itr) @@ -999,7 +1131,7 @@ private: template RegexType* CreatePattern(const ValueType& value) { if (value.IsString()) { - RegexType* r = new (allocator_->Malloc(sizeof(RegexType))) RegexType(value.GetString()); + RegexType* r = new (allocator_->Malloc(sizeof(RegexType))) RegexType(value.GetString(), allocator_); if (!r->IsValid()) { r->~RegexType(); AllocatorType::Free(r); @@ -1011,17 +1143,21 @@ private: } static bool IsPatternMatch(const RegexType* pattern, const Ch *str, SizeType) { - return pattern->Search(str); + GenericRegexSearch rs(*pattern); + return rs.Search(str); } #elif RAPIDJSON_SCHEMA_USE_STDREGEX template RegexType* CreatePattern(const ValueType& value) { - if (value.IsString()) + if (value.IsString()) { + RegexType *r = static_cast(allocator_->Malloc(sizeof(RegexType))); try { - return new (allocator_->Malloc(sizeof(RegexType))) RegexType(value.GetString(), std::size_t(value.GetStringLength()), std::regex_constants::ECMAScript); + return new (r) RegexType(value.GetString(), std::size_t(value.GetStringLength()), std::regex_constants::ECMAScript); } catch (const std::regex_error&) { + AllocatorType::Free(r); } + } return 0; } @@ -1097,15 +1233,20 @@ private: } bool CheckInt(Context& context, int64_t i) const { - if (!(type_ & ((1 << kIntegerSchemaType) | (1 << kNumberSchemaType)))) + if (!(type_ & ((1 << kIntegerSchemaType) | (1 << kNumberSchemaType)))) { + DisallowedType(context, GetIntegerString()); RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + } if (!minimum_.IsNull()) { if (minimum_.IsInt64()) { - if (exclusiveMinimum_ ? i <= minimum_.GetInt64() : i < minimum_.GetInt64()) + if (exclusiveMinimum_ ? i <= minimum_.GetInt64() : i < minimum_.GetInt64()) { + context.error_handler.BelowMinimum(i, minimum_, exclusiveMinimum_); RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinimumString()); + } } else if (minimum_.IsUint64()) { + context.error_handler.BelowMinimum(i, minimum_, exclusiveMinimum_); RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinimumString()); // i <= max(int64_t) < minimum.GetUint64() } else if (!CheckDoubleMinimum(context, static_cast(i))) @@ -1114,19 +1255,23 @@ private: if (!maximum_.IsNull()) { if (maximum_.IsInt64()) { - if (exclusiveMaximum_ ? i >= maximum_.GetInt64() : i > maximum_.GetInt64()) + if (exclusiveMaximum_ ? i >= maximum_.GetInt64() : i > maximum_.GetInt64()) { + context.error_handler.AboveMaximum(i, maximum_, exclusiveMaximum_); RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaximumString()); + } } - else if (maximum_.IsUint64()) - /* do nothing */; // i <= max(int64_t) < maximum_.GetUint64() + else if (maximum_.IsUint64()) { } + /* do nothing */ // i <= max(int64_t) < maximum_.GetUint64() else if (!CheckDoubleMaximum(context, static_cast(i))) return false; } if (!multipleOf_.IsNull()) { if (multipleOf_.IsUint64()) { - if (static_cast(i >= 0 ? i : -i) % multipleOf_.GetUint64() != 0) + if (static_cast(i >= 0 ? i : -i) % multipleOf_.GetUint64() != 0) { + context.error_handler.NotMultipleOf(i, multipleOf_); RAPIDJSON_INVALID_KEYWORD_RETURN(GetMultipleOfString()); + } } else if (!CheckDoubleMultipleOf(context, static_cast(i))) return false; @@ -1136,13 +1281,17 @@ private: } bool CheckUint(Context& context, uint64_t i) const { - if (!(type_ & ((1 << kIntegerSchemaType) | (1 << kNumberSchemaType)))) + if (!(type_ & ((1 << kIntegerSchemaType) | (1 << kNumberSchemaType)))) { + DisallowedType(context, GetIntegerString()); RAPIDJSON_INVALID_KEYWORD_RETURN(GetTypeString()); + } if (!minimum_.IsNull()) { if (minimum_.IsUint64()) { - if (exclusiveMinimum_ ? i <= minimum_.GetUint64() : i < minimum_.GetUint64()) + if (exclusiveMinimum_ ? i <= minimum_.GetUint64() : i < minimum_.GetUint64()) { + context.error_handler.BelowMinimum(i, minimum_, exclusiveMinimum_); RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinimumString()); + } } else if (minimum_.IsInt64()) /* do nothing */; // i >= 0 > minimum.Getint64() @@ -1152,19 +1301,25 @@ private: if (!maximum_.IsNull()) { if (maximum_.IsUint64()) { - if (exclusiveMaximum_ ? i >= maximum_.GetUint64() : i > maximum_.GetUint64()) + if (exclusiveMaximum_ ? i >= maximum_.GetUint64() : i > maximum_.GetUint64()) { + context.error_handler.AboveMaximum(i, maximum_, exclusiveMaximum_); RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaximumString()); + } } - else if (maximum_.IsInt64()) + else if (maximum_.IsInt64()) { + context.error_handler.AboveMaximum(i, maximum_, exclusiveMaximum_); RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaximumString()); // i >= 0 > maximum_ + } else if (!CheckDoubleMaximum(context, static_cast(i))) return false; } if (!multipleOf_.IsNull()) { if (multipleOf_.IsUint64()) { - if (i % multipleOf_.GetUint64() != 0) + if (i % multipleOf_.GetUint64() != 0) { + context.error_handler.NotMultipleOf(i, multipleOf_); RAPIDJSON_INVALID_KEYWORD_RETURN(GetMultipleOfString()); + } } else if (!CheckDoubleMultipleOf(context, static_cast(i))) return false; @@ -1174,14 +1329,18 @@ private: } bool CheckDoubleMinimum(Context& context, double d) const { - if (exclusiveMinimum_ ? d <= minimum_.GetDouble() : d < minimum_.GetDouble()) + if (exclusiveMinimum_ ? d <= minimum_.GetDouble() : d < minimum_.GetDouble()) { + context.error_handler.BelowMinimum(d, minimum_, exclusiveMinimum_); RAPIDJSON_INVALID_KEYWORD_RETURN(GetMinimumString()); + } return true; } bool CheckDoubleMaximum(Context& context, double d) const { - if (exclusiveMaximum_ ? d >= maximum_.GetDouble() : d > maximum_.GetDouble()) + if (exclusiveMaximum_ ? d >= maximum_.GetDouble() : d > maximum_.GetDouble()) { + context.error_handler.AboveMaximum(d, maximum_, exclusiveMaximum_); RAPIDJSON_INVALID_KEYWORD_RETURN(GetMaximumString()); + } return true; } @@ -1189,11 +1348,29 @@ private: double a = std::abs(d), b = std::abs(multipleOf_.GetDouble()); double q = std::floor(a / b); double r = a - q * b; - if (r > 0.0) + if (r > 0.0) { + context.error_handler.NotMultipleOf(d, multipleOf_); RAPIDJSON_INVALID_KEYWORD_RETURN(GetMultipleOfString()); + } return true; } + void DisallowedType(Context& context, const ValueType& actualType) const { + ErrorHandler& eh = context.error_handler; + eh.StartDisallowedType(); + + if (type_ & (1 << kNullSchemaType)) eh.AddExpectedType(GetNullString()); + if (type_ & (1 << kBooleanSchemaType)) eh.AddExpectedType(GetBooleanString()); + if (type_ & (1 << kObjectSchemaType)) eh.AddExpectedType(GetObjectString()); + if (type_ & (1 << kArraySchemaType)) eh.AddExpectedType(GetArrayString()); + if (type_ & (1 << kStringSchemaType)) eh.AddExpectedType(GetStringString()); + + if (type_ & (1 << kNumberSchemaType)) eh.AddExpectedType(GetNumberString()); + else if (type_ & (1 << kIntegerSchemaType)) eh.AddExpectedType(GetIntegerString()); + + eh.EndDisallowedType(actualType); + } + struct Property { Property() : schema(), dependenciesSchema(), dependenciesValidatorIndex(), dependencies(), required(false) {} ~Property() { AllocatorType::Free(dependencies); } @@ -1218,6 +1395,9 @@ private: }; AllocatorType* allocator_; + SValue uri_; + PointerType pointer_; + const SchemaType* typeless_; uint64_t* enum_; SizeType enumCount_; SchemaArray allOf_; @@ -1258,6 +1438,8 @@ private: SValue multipleOf_; bool exclusiveMinimum_; bool exclusiveMaximum_; + + SizeType defaultValueLength_; }; template @@ -1267,7 +1449,7 @@ struct TokenHelper { char buffer[21]; size_t length = static_cast((sizeof(SizeType) == 4 ? u32toa(index, buffer) : u64toa(index, buffer)) - buffer); for (size_t i = 0; i < length; i++) - *documentStack.template Push() = buffer[i]; + *documentStack.template Push() = static_cast(buffer[i]); } }; @@ -1326,6 +1508,7 @@ public: typedef typename EncodingType::Ch Ch; typedef internal::Schema SchemaType; typedef GenericPointer PointerType; + typedef GenericValue URIType; friend class internal::Schema; template friend class GenericSchemaValidator; @@ -1335,19 +1518,29 @@ public: Compile a JSON document into schema document. \param document A JSON document as source. + \param uri The base URI of this schema document for purposes of violation reporting. + \param uriLength Length of \c name, in code points. \param remoteProvider An optional remote schema document provider for resolving remote reference. Can be null. \param allocator An optional allocator instance for allocating memory. Can be null. */ - explicit GenericSchemaDocument(const ValueType& document, IRemoteSchemaDocumentProviderType* remoteProvider = 0, Allocator* allocator = 0) : + explicit GenericSchemaDocument(const ValueType& document, const Ch* uri = 0, SizeType uriLength = 0, + IRemoteSchemaDocumentProviderType* remoteProvider = 0, Allocator* allocator = 0) : remoteProvider_(remoteProvider), allocator_(allocator), ownAllocator_(), root_(), + typeless_(), schemaMap_(allocator, kInitialSchemaMapSize), schemaRef_(allocator, kInitialSchemaRefSize) { if (!allocator_) - ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator()); + ownAllocator_ = allocator_ = RAPIDJSON_NEW(Allocator)(); + + Ch noUri[1] = {0}; + uri_.SetString(uri ? uri : noUri, uriLength, *allocator_); + + typeless_ = static_cast(allocator_->Malloc(sizeof(SchemaType))); + new (typeless_) SchemaType(this, PointerType(), ValueType(kObjectType).Move(), ValueType(kObjectType).Move(), allocator_); // Generate root schema, it will call CreateSchema() to create sub-schemas, // And call AddRefSchema() if there are $ref. @@ -1365,6 +1558,9 @@ public: new (schemaMap_.template Push()) SchemaEntry(refEntry->source, const_cast(s), false, allocator_); } } + else if (refEntry->schema) + *refEntry->schema = typeless_; + refEntry->~SchemaRefEntry(); } @@ -1380,12 +1576,15 @@ public: allocator_(rhs.allocator_), ownAllocator_(rhs.ownAllocator_), root_(rhs.root_), + typeless_(rhs.typeless_), schemaMap_(std::move(rhs.schemaMap_)), - schemaRef_(std::move(rhs.schemaRef_)) + schemaRef_(std::move(rhs.schemaRef_)), + uri_(std::move(rhs.uri_)) { rhs.remoteProvider_ = 0; rhs.allocator_ = 0; rhs.ownAllocator_ = 0; + rhs.typeless_ = 0; } #endif @@ -1394,9 +1593,16 @@ public: while (!schemaMap_.Empty()) schemaMap_.template Pop(1)->~SchemaEntry(); + if (typeless_) { + typeless_->~SchemaType(); + Allocator::Free(typeless_); + } + RAPIDJSON_DELETE(ownAllocator_); } + const URIType& GetURI() const { return uri_; } + //! Get the root schema. const SchemaType& GetRoot() const { return *root_; } @@ -1428,7 +1634,7 @@ private: void CreateSchemaRecursive(const SchemaType** schema, const PointerType& pointer, const ValueType& v, const ValueType& document) { if (schema) - *schema = SchemaType::GetTypeless(); + *schema = typeless_; if (v.GetType() == kObjectType) { const SchemaType* s = GetSchema(pointer); @@ -1473,12 +1679,13 @@ private: if (i > 0) { // Remote reference, resolve immediately if (remoteProvider_) { - if (const GenericSchemaDocument* remoteDocument = remoteProvider_->GetRemoteDocument(s, i - 1)) { + if (const GenericSchemaDocument* remoteDocument = remoteProvider_->GetRemoteDocument(s, i)) { PointerType pointer(&s[i], len - i, allocator_); if (pointer.IsValid()) { if (const SchemaType* sc = remoteDocument->GetSchema(pointer)) { if (schema) *schema = sc; + new (schemaMap_.template Push()) SchemaEntry(source, const_cast(sc), false, allocator_); return true; } } @@ -1515,6 +1722,8 @@ private: return PointerType(); } + const SchemaType* GetTypeless() const { return typeless_; } + static const size_t kInitialSchemaMapSize = 64; static const size_t kInitialSchemaRefSize = 64; @@ -1522,8 +1731,10 @@ private: Allocator *allocator_; Allocator *ownAllocator_; const SchemaType* root_; //!< Root schema. + SchemaType* typeless_; internal::Stack schemaMap_; // Stores created Pointer -> Schemas internal::Stack schemaRef_; // Stores Pointer from $ref and schema which holds the $ref + URIType uri_; }; //! GenericSchemaDocument using Value type. @@ -1552,13 +1763,17 @@ template < typename StateAllocator = CrtAllocator> class GenericSchemaValidator : public internal::ISchemaStateFactory, - public internal::ISchemaValidator + public internal::ISchemaValidator, + public internal::IValidationErrorHandler { public: typedef typename SchemaDocumentType::SchemaType SchemaType; typedef typename SchemaDocumentType::PointerType PointerType; typedef typename SchemaType::EncodingType EncodingType; + typedef typename SchemaType::SValue SValue; typedef typename EncodingType::Ch Ch; + typedef GenericStringRef StringRefType; + typedef GenericValue ValueType; //! Constructor without output handler. /*! @@ -1575,11 +1790,14 @@ public: : schemaDocument_(&schemaDocument), root_(schemaDocument.GetRoot()), - outputHandler_(GetNullHandler()), stateAllocator_(allocator), ownStateAllocator_(0), schemaStack_(allocator, schemaStackCapacity), documentStack_(allocator, documentStackCapacity), + outputHandler_(0), + error_(kObjectType), + currentError_(), + missingDependents_(), valid_(true) #if RAPIDJSON_SCHEMA_VERBOSE , depth_(0) @@ -1603,11 +1821,14 @@ public: : schemaDocument_(&schemaDocument), root_(schemaDocument.GetRoot()), - outputHandler_(outputHandler), stateAllocator_(allocator), ownStateAllocator_(0), schemaStack_(allocator, schemaStackCapacity), documentStack_(allocator, documentStackCapacity), + outputHandler_(&outputHandler), + error_(kObjectType), + currentError_(), + missingDependents_(), valid_(true) #if RAPIDJSON_SCHEMA_VERBOSE , depth_(0) @@ -1626,6 +1847,9 @@ public: while (!schemaStack_.Empty()) PopSchema(); documentStack_.Clear(); + error_.SetObject(); + currentError_.SetNull(); + missingDependents_.SetNull(); valid_ = true; } @@ -1633,9 +1857,13 @@ public: // Implementation of ISchemaValidator virtual bool IsValid() const { return valid_; } + //! Gets the error object. + ValueType& GetError() { return error_; } + const ValueType& GetError() const { return error_; } + //! Gets the JSON pointer pointed to the invalid schema. PointerType GetInvalidSchemaPointer() const { - return schemaStack_.Empty() ? PointerType() : schemaDocument_->GetPointer(&CurrentSchema()); + return schemaStack_.Empty() ? PointerType() : CurrentSchema().GetPointer(); } //! Gets the keyword of invalid schema. @@ -1645,9 +1873,196 @@ public: //! Gets the JSON pointer pointed to the invalid value. PointerType GetInvalidDocumentPointer() const { - return documentStack_.Empty() ? PointerType() : PointerType(documentStack_.template Bottom(), documentStack_.GetSize() / sizeof(Ch)); + if (documentStack_.Empty()) { + return PointerType(); + } + else { + return PointerType(documentStack_.template Bottom(), documentStack_.GetSize() / sizeof(Ch)); + } } + void NotMultipleOf(int64_t actual, const SValue& expected) { + AddNumberError(SchemaType::GetMultipleOfString(), ValueType(actual).Move(), expected); + } + void NotMultipleOf(uint64_t actual, const SValue& expected) { + AddNumberError(SchemaType::GetMultipleOfString(), ValueType(actual).Move(), expected); + } + void NotMultipleOf(double actual, const SValue& expected) { + AddNumberError(SchemaType::GetMultipleOfString(), ValueType(actual).Move(), expected); + } + void AboveMaximum(int64_t actual, const SValue& expected, bool exclusive) { + AddNumberError(SchemaType::GetMaximumString(), ValueType(actual).Move(), expected, + exclusive ? &SchemaType::GetExclusiveMaximumString : 0); + } + void AboveMaximum(uint64_t actual, const SValue& expected, bool exclusive) { + AddNumberError(SchemaType::GetMaximumString(), ValueType(actual).Move(), expected, + exclusive ? &SchemaType::GetExclusiveMaximumString : 0); + } + void AboveMaximum(double actual, const SValue& expected, bool exclusive) { + AddNumberError(SchemaType::GetMaximumString(), ValueType(actual).Move(), expected, + exclusive ? &SchemaType::GetExclusiveMaximumString : 0); + } + void BelowMinimum(int64_t actual, const SValue& expected, bool exclusive) { + AddNumberError(SchemaType::GetMinimumString(), ValueType(actual).Move(), expected, + exclusive ? &SchemaType::GetExclusiveMinimumString : 0); + } + void BelowMinimum(uint64_t actual, const SValue& expected, bool exclusive) { + AddNumberError(SchemaType::GetMinimumString(), ValueType(actual).Move(), expected, + exclusive ? &SchemaType::GetExclusiveMinimumString : 0); + } + void BelowMinimum(double actual, const SValue& expected, bool exclusive) { + AddNumberError(SchemaType::GetMinimumString(), ValueType(actual).Move(), expected, + exclusive ? &SchemaType::GetExclusiveMinimumString : 0); + } + + void TooLong(const Ch* str, SizeType length, SizeType expected) { + AddNumberError(SchemaType::GetMaxLengthString(), + ValueType(str, length, GetStateAllocator()).Move(), SValue(expected).Move()); + } + void TooShort(const Ch* str, SizeType length, SizeType expected) { + AddNumberError(SchemaType::GetMinLengthString(), + ValueType(str, length, GetStateAllocator()).Move(), SValue(expected).Move()); + } + void DoesNotMatch(const Ch* str, SizeType length) { + currentError_.SetObject(); + currentError_.AddMember(GetActualString(), ValueType(str, length, GetStateAllocator()).Move(), GetStateAllocator()); + AddCurrentError(SchemaType::GetPatternString()); + } + + void DisallowedItem(SizeType index) { + currentError_.SetObject(); + currentError_.AddMember(GetDisallowedString(), ValueType(index).Move(), GetStateAllocator()); + AddCurrentError(SchemaType::GetAdditionalItemsString(), true); + } + void TooFewItems(SizeType actualCount, SizeType expectedCount) { + AddNumberError(SchemaType::GetMinItemsString(), + ValueType(actualCount).Move(), SValue(expectedCount).Move()); + } + void TooManyItems(SizeType actualCount, SizeType expectedCount) { + AddNumberError(SchemaType::GetMaxItemsString(), + ValueType(actualCount).Move(), SValue(expectedCount).Move()); + } + void DuplicateItems(SizeType index1, SizeType index2) { + ValueType duplicates(kArrayType); + duplicates.PushBack(index1, GetStateAllocator()); + duplicates.PushBack(index2, GetStateAllocator()); + currentError_.SetObject(); + currentError_.AddMember(GetDuplicatesString(), duplicates, GetStateAllocator()); + AddCurrentError(SchemaType::GetUniqueItemsString(), true); + } + + void TooManyProperties(SizeType actualCount, SizeType expectedCount) { + AddNumberError(SchemaType::GetMaxPropertiesString(), + ValueType(actualCount).Move(), SValue(expectedCount).Move()); + } + void TooFewProperties(SizeType actualCount, SizeType expectedCount) { + AddNumberError(SchemaType::GetMinPropertiesString(), + ValueType(actualCount).Move(), SValue(expectedCount).Move()); + } + void StartMissingProperties() { + currentError_.SetArray(); + } + void AddMissingProperty(const SValue& name) { + currentError_.PushBack(ValueType(name, GetStateAllocator()).Move(), GetStateAllocator()); + } + bool EndMissingProperties() { + if (currentError_.Empty()) + return false; + ValueType error(kObjectType); + error.AddMember(GetMissingString(), currentError_, GetStateAllocator()); + currentError_ = error; + AddCurrentError(SchemaType::GetRequiredString()); + return true; + } + void PropertyViolations(ISchemaValidator** subvalidators, SizeType count) { + for (SizeType i = 0; i < count; ++i) + MergeError(static_cast(subvalidators[i])->GetError()); + } + void DisallowedProperty(const Ch* name, SizeType length) { + currentError_.SetObject(); + currentError_.AddMember(GetDisallowedString(), ValueType(name, length, GetStateAllocator()).Move(), GetStateAllocator()); + AddCurrentError(SchemaType::GetAdditionalPropertiesString(), true); + } + + void StartDependencyErrors() { + currentError_.SetObject(); + } + void StartMissingDependentProperties() { + missingDependents_.SetArray(); + } + void AddMissingDependentProperty(const SValue& targetName) { + missingDependents_.PushBack(ValueType(targetName, GetStateAllocator()).Move(), GetStateAllocator()); + } + void EndMissingDependentProperties(const SValue& sourceName) { + if (!missingDependents_.Empty()) + currentError_.AddMember(ValueType(sourceName, GetStateAllocator()).Move(), + missingDependents_, GetStateAllocator()); + } + void AddDependencySchemaError(const SValue& sourceName, ISchemaValidator* subvalidator) { + currentError_.AddMember(ValueType(sourceName, GetStateAllocator()).Move(), + static_cast(subvalidator)->GetError(), GetStateAllocator()); + } + bool EndDependencyErrors() { + if (currentError_.ObjectEmpty()) + return false; + ValueType error(kObjectType); + error.AddMember(GetErrorsString(), currentError_, GetStateAllocator()); + currentError_ = error; + AddCurrentError(SchemaType::GetDependenciesString()); + return true; + } + + void DisallowedValue() { + currentError_.SetObject(); + AddCurrentError(SchemaType::GetEnumString()); + } + void StartDisallowedType() { + currentError_.SetArray(); + } + void AddExpectedType(const typename SchemaType::ValueType& expectedType) { + currentError_.PushBack(ValueType(expectedType, GetStateAllocator()).Move(), GetStateAllocator()); + } + void EndDisallowedType(const typename SchemaType::ValueType& actualType) { + ValueType error(kObjectType); + error.AddMember(GetExpectedString(), currentError_, GetStateAllocator()); + error.AddMember(GetActualString(), ValueType(actualType, GetStateAllocator()).Move(), GetStateAllocator()); + currentError_ = error; + AddCurrentError(SchemaType::GetTypeString()); + } + void NotAllOf(ISchemaValidator** subvalidators, SizeType count) { + for (SizeType i = 0; i < count; ++i) { + MergeError(static_cast(subvalidators[i])->GetError()); + } + } + void NoneOf(ISchemaValidator** subvalidators, SizeType count) { + AddErrorArray(SchemaType::GetAnyOfString(), subvalidators, count); + } + void NotOneOf(ISchemaValidator** subvalidators, SizeType count) { + AddErrorArray(SchemaType::GetOneOfString(), subvalidators, count); + } + void Disallowed() { + currentError_.SetObject(); + AddCurrentError(SchemaType::GetNotString()); + } + +#define RAPIDJSON_STRING_(name, ...) \ + static const StringRefType& Get##name##String() {\ + static const Ch s[] = { __VA_ARGS__, '\0' };\ + static const StringRefType v(s, static_cast(sizeof(s) / sizeof(Ch) - 1)); \ + return v;\ + } + + RAPIDJSON_STRING_(InstanceRef, 'i', 'n', 's', 't', 'a', 'n', 'c', 'e', 'R', 'e', 'f') + RAPIDJSON_STRING_(SchemaRef, 's', 'c', 'h', 'e', 'm', 'a', 'R', 'e', 'f') + RAPIDJSON_STRING_(Expected, 'e', 'x', 'p', 'e', 'c', 't', 'e', 'd') + RAPIDJSON_STRING_(Actual, 'a', 'c', 't', 'u', 'a', 'l') + RAPIDJSON_STRING_(Disallowed, 'd', 'i', 's', 'a', 'l', 'l', 'o', 'w', 'e', 'd') + RAPIDJSON_STRING_(Missing, 'm', 'i', 's', 's', 'i', 'n', 'g') + RAPIDJSON_STRING_(Errors, 'e', 'r', 'r', 'o', 'r', 's') + RAPIDJSON_STRING_(Duplicates, 'd', 'u', 'p', 'l', 'i', 'c', 'a', 't', 'e', 's') + +#undef RAPIDJSON_STRING_ + #if RAPIDJSON_SCHEMA_VERBOSE #define RAPIDJSON_SCHEMA_HANDLE_BEGIN_VERBOSE_() \ RAPIDJSON_MULTILINEMACRO_BEGIN\ @@ -1679,14 +2094,14 @@ RAPIDJSON_MULTILINEMACRO_END } #define RAPIDJSON_SCHEMA_HANDLE_END_(method, arg2)\ - return valid_ = EndValue() && outputHandler_.method arg2 + return valid_ = EndValue() && (!outputHandler_ || outputHandler_->method arg2) #define RAPIDJSON_SCHEMA_HANDLE_VALUE_(method, arg1, arg2) \ RAPIDJSON_SCHEMA_HANDLE_BEGIN_ (method, arg1);\ RAPIDJSON_SCHEMA_HANDLE_PARALLEL_(method, arg2);\ RAPIDJSON_SCHEMA_HANDLE_END_ (method, arg2) - bool Null() { RAPIDJSON_SCHEMA_HANDLE_VALUE_(Null, (CurrentContext() ), ( )); } + bool Null() { RAPIDJSON_SCHEMA_HANDLE_VALUE_(Null, (CurrentContext()), ( )); } bool Bool(bool b) { RAPIDJSON_SCHEMA_HANDLE_VALUE_(Bool, (CurrentContext(), b), (b)); } bool Int(int i) { RAPIDJSON_SCHEMA_HANDLE_VALUE_(Int, (CurrentContext(), i), (i)); } bool Uint(unsigned u) { RAPIDJSON_SCHEMA_HANDLE_VALUE_(Uint, (CurrentContext(), u), (u)); } @@ -1701,7 +2116,7 @@ RAPIDJSON_MULTILINEMACRO_END bool StartObject() { RAPIDJSON_SCHEMA_HANDLE_BEGIN_(StartObject, (CurrentContext())); RAPIDJSON_SCHEMA_HANDLE_PARALLEL_(StartObject, ()); - return valid_ = outputHandler_.StartObject(); + return valid_ = !outputHandler_ || outputHandler_->StartObject(); } bool Key(const Ch* str, SizeType len, bool copy) { @@ -1709,7 +2124,7 @@ RAPIDJSON_MULTILINEMACRO_END AppendToken(str, len); if (!CurrentSchema().Key(CurrentContext(), str, len, copy)) return valid_ = false; RAPIDJSON_SCHEMA_HANDLE_PARALLEL_(Key, (str, len, copy)); - return valid_ = outputHandler_.Key(str, len, copy); + return valid_ = !outputHandler_ || outputHandler_->Key(str, len, copy); } bool EndObject(SizeType memberCount) { @@ -1722,7 +2137,7 @@ RAPIDJSON_MULTILINEMACRO_END bool StartArray() { RAPIDJSON_SCHEMA_HANDLE_BEGIN_(StartArray, (CurrentContext())); RAPIDJSON_SCHEMA_HANDLE_PARALLEL_(StartArray, ()); - return valid_ = outputHandler_.StartArray(); + return valid_ = !outputHandler_ || outputHandler_->StartArray(); } bool EndArray(SizeType elementCount) { @@ -1739,7 +2154,7 @@ RAPIDJSON_MULTILINEMACRO_END // Implementation of ISchemaStateFactory virtual ISchemaValidator* CreateSchemaValidator(const SchemaType& root) { - return new (GetStateAllocator().Malloc(sizeof(GenericSchemaValidator))) GenericSchemaValidator(*schemaDocument_, root, + return new (GetStateAllocator().Malloc(sizeof(GenericSchemaValidator))) GenericSchemaValidator(*schemaDocument_, root, documentStack_.template Bottom(), documentStack_.GetSize(), #if RAPIDJSON_SCHEMA_VERBOSE depth_ + 1, #endif @@ -1771,7 +2186,7 @@ RAPIDJSON_MULTILINEMACRO_END } virtual void FreeState(void* p) { - return StateAllocator::Free(p); + StateAllocator::Free(p); } private: @@ -1782,6 +2197,7 @@ private: GenericSchemaValidator( const SchemaDocumentType& schemaDocument, const SchemaType& root, + const char* basePath, size_t basePathSize, #if RAPIDJSON_SCHEMA_VERBOSE unsigned depth, #endif @@ -1791,21 +2207,26 @@ private: : schemaDocument_(&schemaDocument), root_(root), - outputHandler_(GetNullHandler()), stateAllocator_(allocator), ownStateAllocator_(0), schemaStack_(allocator, schemaStackCapacity), documentStack_(allocator, documentStackCapacity), + outputHandler_(0), + error_(kObjectType), + currentError_(), + missingDependents_(), valid_(true) #if RAPIDJSON_SCHEMA_VERBOSE , depth_(depth) #endif { + if (basePath && basePathSize) + memcpy(documentStack_.template Push(basePathSize), basePath, basePathSize); } StateAllocator& GetStateAllocator() { if (!stateAllocator_) - stateAllocator_ = ownStateAllocator_ = RAPIDJSON_NEW(StateAllocator()); + stateAllocator_ = ownStateAllocator_ = RAPIDJSON_NEW(StateAllocator)(); return *stateAllocator_; } @@ -1823,8 +2244,8 @@ private: const SchemaType** sa = CurrentContext().patternPropertiesSchemas; typename Context::PatternValidatorType patternValidatorType = CurrentContext().valuePatternValidatorType; bool valueUniqueness = CurrentContext().valueUniqueness; - if (CurrentContext().valueSchema) - PushSchema(*CurrentContext().valueSchema); + RAPIDJSON_ASSERT(CurrentContext().valueSchema); + PushSchema(*CurrentContext().valueSchema); if (count > 0) { CurrentContext().objectPatternValidatorType = patternValidatorType; @@ -1864,8 +2285,10 @@ private: if (!a) CurrentContext().arrayElementHashCodes = a = new (GetStateAllocator().Malloc(sizeof(HashCodeArray))) HashCodeArray(kArrayType); for (typename HashCodeArray::ConstValueIterator itr = a->Begin(); itr != a->End(); ++itr) - if (itr->GetUint64() == h) + if (itr->GetUint64() == h) { + DuplicateItems(static_cast(itr - a->Begin()), a->Size()); RAPIDJSON_INVALID_KEYWORD_RETURN(SchemaType::GetUniqueItemsString()); + } a->PushBack(h, GetStateAllocator()); } } @@ -1894,7 +2317,7 @@ private: } } - RAPIDJSON_FORCEINLINE void PushSchema(const SchemaType& schema) { new (schemaStack_.template Push()) Context(*this, &schema); } + RAPIDJSON_FORCEINLINE void PushSchema(const SchemaType& schema) { new (schemaStack_.template Push()) Context(*this, *this, &schema); } RAPIDJSON_FORCEINLINE void PopSchema() { Context* c = schemaStack_.template Pop(1); @@ -1905,24 +2328,86 @@ private: c->~Context(); } + void AddErrorLocation(ValueType& result, bool parent) { + GenericStringBuffer sb; + PointerType instancePointer = GetInvalidDocumentPointer(); + ((parent && instancePointer.GetTokenCount() > 0) + ? PointerType(instancePointer.GetTokens(), instancePointer.GetTokenCount() - 1) + : instancePointer).StringifyUriFragment(sb); + ValueType instanceRef(sb.GetString(), static_cast(sb.GetSize() / sizeof(Ch)), + GetStateAllocator()); + result.AddMember(GetInstanceRefString(), instanceRef, GetStateAllocator()); + sb.Clear(); + memcpy(sb.Push(CurrentSchema().GetURI().GetStringLength()), + CurrentSchema().GetURI().GetString(), + CurrentSchema().GetURI().GetStringLength() * sizeof(Ch)); + GetInvalidSchemaPointer().StringifyUriFragment(sb); + ValueType schemaRef(sb.GetString(), static_cast(sb.GetSize() / sizeof(Ch)), + GetStateAllocator()); + result.AddMember(GetSchemaRefString(), schemaRef, GetStateAllocator()); + } + + void AddError(ValueType& keyword, ValueType& error) { + typename ValueType::MemberIterator member = error_.FindMember(keyword); + if (member == error_.MemberEnd()) + error_.AddMember(keyword, error, GetStateAllocator()); + else { + if (member->value.IsObject()) { + ValueType errors(kArrayType); + errors.PushBack(member->value, GetStateAllocator()); + member->value = errors; + } + member->value.PushBack(error, GetStateAllocator()); + } + } + + void AddCurrentError(const typename SchemaType::ValueType& keyword, bool parent = false) { + AddErrorLocation(currentError_, parent); + AddError(ValueType(keyword, GetStateAllocator(), false).Move(), currentError_); + } + + void MergeError(ValueType& other) { + for (typename ValueType::MemberIterator it = other.MemberBegin(), end = other.MemberEnd(); it != end; ++it) { + AddError(it->name, it->value); + } + } + + void AddNumberError(const typename SchemaType::ValueType& keyword, ValueType& actual, const SValue& expected, + const typename SchemaType::ValueType& (*exclusive)() = 0) { + currentError_.SetObject(); + currentError_.AddMember(GetActualString(), actual, GetStateAllocator()); + currentError_.AddMember(GetExpectedString(), ValueType(expected, GetStateAllocator()).Move(), GetStateAllocator()); + if (exclusive) + currentError_.AddMember(ValueType(exclusive(), GetStateAllocator()).Move(), true, GetStateAllocator()); + AddCurrentError(keyword); + } + + void AddErrorArray(const typename SchemaType::ValueType& keyword, + ISchemaValidator** subvalidators, SizeType count) { + ValueType errors(kArrayType); + for (SizeType i = 0; i < count; ++i) + errors.PushBack(static_cast(subvalidators[i])->GetError(), GetStateAllocator()); + currentError_.SetObject(); + currentError_.AddMember(GetErrorsString(), errors, GetStateAllocator()); + AddCurrentError(keyword); + } + const SchemaType& CurrentSchema() const { return *schemaStack_.template Top()->schema; } Context& CurrentContext() { return *schemaStack_.template Top(); } const Context& CurrentContext() const { return *schemaStack_.template Top(); } - static OutputHandler& GetNullHandler() { - static OutputHandler nullHandler; - return nullHandler; - } - static const size_t kDefaultSchemaStackCapacity = 1024; static const size_t kDefaultDocumentStackCapacity = 256; const SchemaDocumentType* schemaDocument_; const SchemaType& root_; - OutputHandler& outputHandler_; StateAllocator* stateAllocator_; StateAllocator* ownStateAllocator_; internal::Stack schemaStack_; //!< stack to store the current path of schema (BaseSchemaType *) internal::Stack documentStack_; //!< stack to store the current path of validating document (Ch) + OutputHandler* outputHandler_; + ValueType error_; + ValueType currentError_; + ValueType missingDependents_; bool valid_; #if RAPIDJSON_SCHEMA_VERBOSE unsigned depth_; @@ -1954,13 +2439,14 @@ class SchemaValidatingReader { public: typedef typename SchemaDocumentType::PointerType PointerType; typedef typename InputStream::Ch Ch; + typedef GenericValue ValueType; //! Constructor /*! \param is Input stream. \param sd Schema document. */ - SchemaValidatingReader(InputStream& is, const SchemaDocumentType& sd) : is_(is), sd_(sd), invalidSchemaKeyword_(), isValid_(true) {} + SchemaValidatingReader(InputStream& is, const SchemaDocumentType& sd) : is_(is), sd_(sd), invalidSchemaKeyword_(), error_(kObjectType), isValid_(true) {} template bool operator()(Handler& handler) { @@ -1973,11 +2459,13 @@ public: invalidSchemaPointer_ = PointerType(); invalidSchemaKeyword_ = 0; invalidDocumentPointer_ = PointerType(); + error_.SetObject(); } else { invalidSchemaPointer_ = validator.GetInvalidSchemaPointer(); invalidSchemaKeyword_ = validator.GetInvalidSchemaKeyword(); invalidDocumentPointer_ = validator.GetInvalidDocumentPointer(); + error_.CopyFrom(validator.GetError(), allocator_); } return parseResult_; @@ -1988,6 +2476,7 @@ public: const PointerType& GetInvalidSchemaPointer() const { return invalidSchemaPointer_; } const Ch* GetInvalidSchemaKeyword() const { return invalidSchemaKeyword_; } const PointerType& GetInvalidDocumentPointer() const { return invalidDocumentPointer_; } + const ValueType& GetError() const { return error_; } private: InputStream& is_; @@ -1997,6 +2486,8 @@ private: PointerType invalidSchemaPointer_; const Ch* invalidSchemaKeyword_; PointerType invalidDocumentPointer_; + StackAllocator allocator_; + ValueType error_; bool isValid_; }; diff --git a/src/3rdparty/rapidjson/stream.h b/src/3rdparty/rapidjson/stream.h index fef82c252..7f2643e48 100644 --- a/src/3rdparty/rapidjson/stream.h +++ b/src/3rdparty/rapidjson/stream.h @@ -1,5 +1,5 @@ // Tencent is pleased to support the open source community by making RapidJSON available. -// +// // Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved. // // Licensed under the MIT License (the "License"); you may not use this file except @@ -7,9 +7,9 @@ // // http://opensource.org/licenses/MIT // -// Unless required by applicable law or agreed to in writing, software distributed -// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR -// CONDITIONS OF ANY KIND, either express or implied. See the License for the +// Unless required by applicable law or agreed to in writing, software distributed +// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR +// CONDITIONS OF ANY KIND, either express or implied. See the License for the // specific language governing permissions and limitations under the License. #include "rapidjson.h" @@ -100,6 +100,50 @@ inline void PutN(Stream& stream, Ch c, size_t n) { PutUnsafe(stream, c); } +/////////////////////////////////////////////////////////////////////////////// +// GenericStreamWrapper + +//! A Stream Wrapper +/*! \tThis string stream is a wrapper for any stream by just forwarding any + \treceived message to the origin stream. + \note implements Stream concept +*/ + +#if defined(_MSC_VER) && _MSC_VER <= 1800 +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(4702) // unreachable code +RAPIDJSON_DIAG_OFF(4512) // assignment operator could not be generated +#endif + +template > +class GenericStreamWrapper { +public: + typedef typename Encoding::Ch Ch; + GenericStreamWrapper(InputStream& is): is_(is) {} + + Ch Peek() const { return is_.Peek(); } + Ch Take() { return is_.Take(); } + size_t Tell() { return is_.Tell(); } + Ch* PutBegin() { return is_.PutBegin(); } + void Put(Ch ch) { is_.Put(ch); } + void Flush() { is_.Flush(); } + size_t PutEnd(Ch* ch) { return is_.PutEnd(ch); } + + // wrapper for MemoryStream + const Ch* Peek4() const { return is_.Peek4(); } + + // wrapper for AutoUTFInputStream + UTFType GetType() const { return is_.GetType(); } + bool HasBOM() const { return is_.HasBOM(); } + +protected: + InputStream& is_; +}; + +#if defined(_MSC_VER) && _MSC_VER <= 1800 +RAPIDJSON_DIAG_POP +#endif + /////////////////////////////////////////////////////////////////////////////// // StringStream diff --git a/src/3rdparty/rapidjson/stringbuffer.h b/src/3rdparty/rapidjson/stringbuffer.h index 78f34d209..4e38b82c3 100644 --- a/src/3rdparty/rapidjson/stringbuffer.h +++ b/src/3rdparty/rapidjson/stringbuffer.h @@ -78,8 +78,12 @@ public: return stack_.template Bottom(); } + //! Get the size of string in bytes in the string buffer. size_t GetSize() const { return stack_.GetSize(); } + //! Get the length of string in Ch in the string buffer. + size_t GetLength() const { return stack_.GetSize() / sizeof(Ch); } + static const size_t kDefaultCapacity = 256; mutable internal::Stack stack_; diff --git a/src/3rdparty/rapidjson/writer.h b/src/3rdparty/rapidjson/writer.h index 94f22dd5f..1d33b2f92 100644 --- a/src/3rdparty/rapidjson/writer.h +++ b/src/3rdparty/rapidjson/writer.h @@ -16,6 +16,7 @@ #define RAPIDJSON_WRITER_H_ #include "stream.h" +#include "internal/meta.h" #include "internal/stack.h" #include "internal/strfunc.h" #include "internal/dtoa.h" @@ -31,17 +32,18 @@ #include #elif defined(RAPIDJSON_SSE2) #include -#endif - -#ifdef _MSC_VER -RAPIDJSON_DIAG_PUSH -RAPIDJSON_DIAG_OFF(4127) // conditional expression is constant +#elif defined(RAPIDJSON_NEON) +#include #endif #ifdef __clang__ RAPIDJSON_DIAG_PUSH RAPIDJSON_DIAG_OFF(padded) RAPIDJSON_DIAG_OFF(unreachable-code) +RAPIDJSON_DIAG_OFF(c++98-compat) +#elif defined(_MSC_VER) +RAPIDJSON_DIAG_PUSH +RAPIDJSON_DIAG_OFF(4127) // conditional expression is constant #endif RAPIDJSON_NAMESPACE_BEGIN @@ -103,6 +105,13 @@ public: Writer(StackAllocator* allocator = 0, size_t levelDepth = kDefaultLevelDepth) : os_(0), level_stack_(allocator, levelDepth * sizeof(Level)), maxDecimalPlaces_(kDefaultMaxDecimalPlaces), hasRoot_(false) {} +#if RAPIDJSON_HAS_CXX11_RVALUE_REFS + Writer(Writer&& rhs) : + os_(rhs.os_), level_stack_(std::move(rhs.level_stack_)), maxDecimalPlaces_(rhs.maxDecimalPlaces_), hasRoot_(rhs.hasRoot_) { + rhs.os_ = 0; + } +#endif + //! Reset the writer with a new stream. /*! This function reset the writer with a new stream and default settings, @@ -184,12 +193,14 @@ public: bool Double(double d) { Prefix(kNumberType); return EndValue(WriteDouble(d)); } bool RawNumber(const Ch* str, SizeType length, bool copy = false) { + RAPIDJSON_ASSERT(str != 0); (void)copy; Prefix(kNumberType); return EndValue(WriteString(str, length)); } bool String(const Ch* str, SizeType length, bool copy = false) { + RAPIDJSON_ASSERT(str != 0); (void)copy; Prefix(kStringType); return EndValue(WriteString(str, length)); @@ -209,10 +220,18 @@ public: bool Key(const Ch* str, SizeType length, bool copy = false) { return String(str, length, copy); } +#if RAPIDJSON_HAS_STDSTRING + bool Key(const std::basic_string& str) + { + return Key(str.data(), SizeType(str.size())); + } +#endif + bool EndObject(SizeType memberCount = 0) { (void)memberCount; - RAPIDJSON_ASSERT(level_stack_.GetSize() >= sizeof(Level)); - RAPIDJSON_ASSERT(!level_stack_.template Top()->inArray); + RAPIDJSON_ASSERT(level_stack_.GetSize() >= sizeof(Level)); // not inside an Object + RAPIDJSON_ASSERT(!level_stack_.template Top()->inArray); // currently inside an Array, not Object + RAPIDJSON_ASSERT(0 == level_stack_.template Top()->valueCount % 2); // Object has a Key without a Value level_stack_.template Pop(1); return EndValue(WriteEndObject()); } @@ -236,9 +255,9 @@ public: //@{ //! Simpler but slower overload. - bool String(const Ch* str) { return String(str, internal::StrLen(str)); } - bool Key(const Ch* str) { return Key(str, internal::StrLen(str)); } - + bool String(const Ch* const& str) { return String(str, internal::StrLen(str)); } + bool Key(const Ch* const& str) { return Key(str, internal::StrLen(str)); } + //@} //! Write a raw JSON value. @@ -249,7 +268,19 @@ public: \param length Length of the json. \param type Type of the root of json. */ - bool RawValue(const Ch* json, size_t length, Type type) { Prefix(type); return EndValue(WriteRawValue(json, length)); } + bool RawValue(const Ch* json, size_t length, Type type) { + RAPIDJSON_ASSERT(json != 0); + Prefix(type); + return EndValue(WriteRawValue(json, length)); + } + + //! Flush the output stream. + /*! + Allows the user to flush the output stream immediately. + */ + void Flush() { + os_->Flush(); + } protected: //! Information for each nested level @@ -257,6 +288,7 @@ protected: Level(bool inArray_) : valueCount(0), inArray(inArray_) {} size_t valueCount; //!< number of values in this level bool inArray; //!< true if in array, otherwise in object + bool inLine = false; }; static const size_t kDefaultLevelDepth = 32; @@ -283,7 +315,7 @@ protected: const char* end = internal::i32toa(i, buffer); PutReserve(*os_, static_cast(end - buffer)); for (const char* p = buffer; p != end; ++p) - PutUnsafe(*os_, static_cast(*p)); + PutUnsafe(*os_, static_cast(*p)); return true; } @@ -292,7 +324,7 @@ protected: const char* end = internal::u32toa(u, buffer); PutReserve(*os_, static_cast(end - buffer)); for (const char* p = buffer; p != end; ++p) - PutUnsafe(*os_, static_cast(*p)); + PutUnsafe(*os_, static_cast(*p)); return true; } @@ -301,7 +333,7 @@ protected: const char* end = internal::i64toa(i64, buffer); PutReserve(*os_, static_cast(end - buffer)); for (const char* p = buffer; p != end; ++p) - PutUnsafe(*os_, static_cast(*p)); + PutUnsafe(*os_, static_cast(*p)); return true; } @@ -310,7 +342,7 @@ protected: char* end = internal::u64toa(u64, buffer); PutReserve(*os_, static_cast(end - buffer)); for (char* p = buffer; p != end; ++p) - PutUnsafe(*os_, static_cast(*p)); + PutUnsafe(*os_, static_cast(*p)); return true; } @@ -338,12 +370,12 @@ protected: char* end = internal::dtoa(d, buffer, maxDecimalPlaces_); PutReserve(*os_, static_cast(end - buffer)); for (char* p = buffer; p != end; ++p) - PutUnsafe(*os_, static_cast(*p)); + PutUnsafe(*os_, static_cast(*p)); return true; } bool WriteString(const Ch* str, SizeType length) { - static const typename TargetEncoding::Ch hexDigits[16] = { '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'A', 'B', 'C', 'D', 'E', 'F' }; + static const typename OutputStream::Ch hexDigits[16] = { '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', 'A', 'B', 'C', 'D', 'E', 'F' }; static const char escape[256] = { #define Z16 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 //0 1 2 3 4 5 6 7 8 9 A B C D E F @@ -399,7 +431,7 @@ protected: else if ((sizeof(Ch) == 1 || static_cast(c) < 256) && RAPIDJSON_UNLIKELY(escape[static_cast(c)])) { is.Take(); PutUnsafe(*os_, '\\'); - PutUnsafe(*os_, static_cast(escape[static_cast(c)])); + PutUnsafe(*os_, static_cast(escape[static_cast(c)])); if (escape[static_cast(c)] == 'u') { PutUnsafe(*os_, '0'); PutUnsafe(*os_, '0'); @@ -427,9 +459,13 @@ protected: bool WriteRawValue(const Ch* json, size_t length) { PutReserve(*os_, length); - for (size_t i = 0; i < length; i++) { - RAPIDJSON_ASSERT(json[i] != '\0'); - PutUnsafe(*os_, json[i]); + GenericStringStream is(json); + while (RAPIDJSON_LIKELY(is.Tell() < length)) { + RAPIDJSON_ASSERT(is.Peek() != '\0'); + if (RAPIDJSON_UNLIKELY(!(writeFlags & kWriteValidateEncodingFlag ? + Transcoder::Validate(is, *os_) : + Transcoder::TranscodeUnsafe(is, *os_)))) + return false; } return true; } @@ -457,7 +493,7 @@ protected: // Flush the value if it is the top level one. bool EndValue(bool ret) { if (RAPIDJSON_UNLIKELY(level_stack_.Empty())) // end of json text - os_->Flush(); + Flush(); return ret; } @@ -561,7 +597,7 @@ inline bool Writer::ScanWriteUnescapedString(StringStream& is, siz // The rest of string using SIMD static const char dquote[16] = { '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"', '\"' }; static const char bslash[16] = { '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\', '\\' }; - static const char space[16] = { 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19, 0x19 }; + static const char space[16] = { 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F, 0x1F }; const __m128i dq = _mm_loadu_si128(reinterpret_cast(&dquote[0])); const __m128i bs = _mm_loadu_si128(reinterpret_cast(&bslash[0])); const __m128i sp = _mm_loadu_si128(reinterpret_cast(&space[0])); @@ -570,7 +606,7 @@ inline bool Writer::ScanWriteUnescapedString(StringStream& is, siz const __m128i s = _mm_load_si128(reinterpret_cast(p)); const __m128i t1 = _mm_cmpeq_epi8(s, dq); const __m128i t2 = _mm_cmpeq_epi8(s, bs); - const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x19) == 0x19 + const __m128i t3 = _mm_cmpeq_epi8(_mm_max_epu8(s, sp), sp); // s < 0x20 <=> max(s, 0x1F) == 0x1F const __m128i x = _mm_or_si128(_mm_or_si128(t1, t2), t3); unsigned short r = static_cast(_mm_movemask_epi8(x)); if (RAPIDJSON_UNLIKELY(r != 0)) { // some of characters is escaped @@ -595,15 +631,79 @@ inline bool Writer::ScanWriteUnescapedString(StringStream& is, siz is.src_ = p; return RAPIDJSON_LIKELY(is.Tell() < length); } -#endif // defined(RAPIDJSON_SSE2) || defined(RAPIDJSON_SSE42) +#elif defined(RAPIDJSON_NEON) +template<> +inline bool Writer::ScanWriteUnescapedString(StringStream& is, size_t length) { + if (length < 16) + return RAPIDJSON_LIKELY(is.Tell() < length); + + if (!RAPIDJSON_LIKELY(is.Tell() < length)) + return false; + + const char* p = is.src_; + const char* end = is.head_ + length; + const char* nextAligned = reinterpret_cast((reinterpret_cast(p) + 15) & static_cast(~15)); + const char* endAligned = reinterpret_cast(reinterpret_cast(end) & static_cast(~15)); + if (nextAligned > end) + return true; + + while (p != nextAligned) + if (*p < 0x20 || *p == '\"' || *p == '\\') { + is.src_ = p; + return RAPIDJSON_LIKELY(is.Tell() < length); + } + else + os_->PutUnsafe(*p++); + + // The rest of string using SIMD + const uint8x16_t s0 = vmovq_n_u8('"'); + const uint8x16_t s1 = vmovq_n_u8('\\'); + const uint8x16_t s2 = vmovq_n_u8('\b'); + const uint8x16_t s3 = vmovq_n_u8(32); + + for (; p != endAligned; p += 16) { + const uint8x16_t s = vld1q_u8(reinterpret_cast(p)); + uint8x16_t x = vceqq_u8(s, s0); + x = vorrq_u8(x, vceqq_u8(s, s1)); + x = vorrq_u8(x, vceqq_u8(s, s2)); + x = vorrq_u8(x, vcltq_u8(s, s3)); + + x = vrev64q_u8(x); // Rev in 64 + uint64_t low = vgetq_lane_u64(reinterpret_cast(x), 0); // extract + uint64_t high = vgetq_lane_u64(reinterpret_cast(x), 1); // extract + + SizeType len = 0; + bool escaped = false; + if (low == 0) { + if (high != 0) { + unsigned lz = (unsigned)__builtin_clzll(high); + len = 8 + (lz >> 3); + escaped = true; + } + } else { + unsigned lz = (unsigned)__builtin_clzll(low); + len = lz >> 3; + escaped = true; + } + if (RAPIDJSON_UNLIKELY(escaped)) { // some of characters is escaped + char* q = reinterpret_cast(os_->PushUnsafe(len)); + for (size_t i = 0; i < len; i++) + q[i] = p[i]; + + p += len; + break; + } + vst1q_u8(reinterpret_cast(os_->PushUnsafe(16)), s); + } + + is.src_ = p; + return RAPIDJSON_LIKELY(is.Tell() < length); +} +#endif // RAPIDJSON_NEON RAPIDJSON_NAMESPACE_END -#ifdef _MSC_VER -RAPIDJSON_DIAG_POP -#endif - -#ifdef __clang__ +#if defined(_MSC_VER) || defined(__clang__) RAPIDJSON_DIAG_POP #endif diff --git a/src/App.cpp b/src/App.cpp index e75766acf..ccbaad4fc 100644 --- a/src/App.cpp +++ b/src/App.cpp @@ -6,7 +6,7 @@ * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , * Copyright 2018 Lee Clagett - * Copyright 2018 SChernykh + * Copyright 2018-2019 SChernykh * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify @@ -30,29 +30,21 @@ #include "api/Api.h" #include "App.h" +#include "backend/cpu/Cpu.h" +#include "base/io/Console.h" +#include "base/io/log/Log.h" #include "base/kernel/Signals.h" -#include "common/Console.h" -#include "common/cpu/Cpu.h" -#include "common/log/Log.h" -#include "common/Platform.h" -#include "core/Config.h" +#include "core/config/Config.h" #include "core/Controller.h" -#include "crypto/CryptoNight.h" -#include "Mem.h" +#include "core/Miner.h" +#include "crypto/common/VirtualMemory.h" #include "net/Network.h" #include "Summary.h" #include "version.h" -#include "workers/Workers.h" - - -#ifndef XMRIG_NO_HTTPD -# include "common/api/Httpd.h" -#endif xmrig::App::App(Process *process) : m_console(nullptr), - m_httpd(nullptr), m_signals(nullptr) { m_controller = new Controller(process); @@ -68,15 +60,9 @@ xmrig::App::App(Process *process) : xmrig::App::~App() { - uv_tty_reset_mode(); - delete m_signals; delete m_console; delete m_controller; - -# ifndef XMRIG_NO_HTTPD - delete m_httpd; -# endif } @@ -90,7 +76,7 @@ int xmrig::App::exec() background(); - Mem::init(m_controller->config()->isHugePages()); + VirtualMemory::init(m_controller->config()->cpu().isHugePages()); Summary::print(m_controller); @@ -100,24 +86,7 @@ int xmrig::App::exec() return 0; } -# ifndef XMRIG_NO_API - Api::start(m_controller); -# endif - -# ifndef XMRIG_NO_HTTPD - m_httpd = new Httpd( - m_controller->config()->apiPort(), - m_controller->config()->apiToken(), - m_controller->config()->isApiIPv6(), - m_controller->config()->isApiRestricted() - ); - - m_httpd->start(); -# endif - - Workers::start(m_controller); - - m_controller->network()->connect(); + m_controller->start(); const int r = uv_run(uv_default_loop(), UV_RUN_DEFAULT); uv_loop_close(uv_default_loop()); @@ -131,23 +100,17 @@ void xmrig::App::onConsoleCommand(char command) switch (command) { case 'h': case 'H': - Workers::printHashrate(true); + m_controller->miner()->printHashrate(true); break; case 'p': case 'P': - if (Workers::isEnabled()) { - LOG_INFO(m_controller->config()->isColors() ? "\x1B[01;33mpaused\x1B[0m, press \x1B[01;35mr\x1B[0m to resume" : "paused, press 'r' to resume"); - Workers::setEnabled(false); - } + m_controller->miner()->setEnabled(false); break; case 'r': case 'R': - if (!Workers::isEnabled()) { - LOG_INFO(m_controller->config()->isColors() ? "\x1B[01;32mresumed" : "resumed"); - Workers::setEnabled(true); - } + m_controller->miner()->setEnabled(true); break; case 3: @@ -187,8 +150,9 @@ void xmrig::App::onSignal(int signum) void xmrig::App::close() { - m_controller->network()->stop(); - Workers::stop(); + m_signals->stop(); + m_console->stop(); + m_controller->stop(); - uv_stop(uv_default_loop()); + Log::destroy(); } diff --git a/src/App.h b/src/App.h index fc9449673..aa534aad2 100644 --- a/src/App.h +++ b/src/App.h @@ -27,17 +27,14 @@ #define XMRIG_APP_H +#include "base/kernel/interfaces/IConsoleListener.h" #include "base/kernel/interfaces/ISignalListener.h" -#include "common/interfaces/IConsoleListener.h" - - -class Console; -class Httpd; namespace xmrig { +class Console; class Controller; class Network; class Process; @@ -62,7 +59,6 @@ private: Console *m_console; Controller *m_controller; - Httpd *m_httpd; Signals *m_signals; }; diff --git a/src/App_unix.cpp b/src/App_unix.cpp index 2b598c229..5149513c3 100644 --- a/src/App_unix.cpp +++ b/src/App_unix.cpp @@ -30,8 +30,8 @@ #include "App.h" -#include "common/log/Log.h" -#include "core/Config.h" +#include "base/io/log/Log.h" +#include "core/config/Config.h" #include "core/Controller.h" diff --git a/src/App_win.cpp b/src/App_win.cpp index 57f57a6df..a41ed505a 100644 --- a/src/App_win.cpp +++ b/src/App_win.cpp @@ -29,7 +29,7 @@ #include "App.h" #include "core/Controller.h" -#include "core/Config.h" +#include "core/config/Config.h" void xmrig::App::background() diff --git a/src/Mem_unix.cpp b/src/Mem_unix.cpp deleted file mode 100644 index 833c200c1..000000000 --- a/src/Mem_unix.cpp +++ /dev/null @@ -1,114 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2018 Lee Clagett - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - - -#include -#include - - -#include "common/log/Log.h" -#include "common/utils/mm_malloc.h" -#include "common/xmrig.h" -#include "crypto/CryptoNight.h" -#include "Mem.h" - - -void Mem::init(bool enabled) -{ - m_enabled = enabled; -} - - -void Mem::allocate(MemInfo &info, bool enabled) -{ - info.hugePages = 0; - - if (!enabled) { - info.memory = static_cast(_mm_malloc(info.size, 4096)); - - return; - } - -# if defined(__APPLE__) - info.memory = static_cast(mmap(0, info.size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANON, VM_FLAGS_SUPERPAGE_SIZE_2MB, 0)); -# elif defined(__FreeBSD__) - info.memory = static_cast(mmap(0, info.size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS | MAP_ALIGNED_SUPER | MAP_PREFAULT_READ, -1, 0)); -# else - info.memory = static_cast(mmap(0, info.size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS | MAP_HUGETLB | MAP_POPULATE, 0, 0)); -# endif - - if (info.memory == MAP_FAILED) { - return allocate(info, false);; - } - - info.hugePages = info.pages; - - if (madvise(info.memory, info.size, MADV_RANDOM | MADV_WILLNEED) != 0) { - LOG_ERR("madvise failed"); - } - - if (mlock(info.memory, info.size) == 0) { - m_flags |= Lock; - } -} - - -void Mem::release(MemInfo &info) -{ - if (info.hugePages) { - if (m_flags & Lock) { - munlock(info.memory, info.size); - } - - munmap(info.memory, info.size); - } - else { - _mm_free(info.memory); - } -} - - -void *Mem::allocateExecutableMemory(size_t size) -{ -# if defined(__APPLE__) - return mmap(0, size, PROT_READ | PROT_WRITE | PROT_EXEC, MAP_PRIVATE | MAP_ANON, -1, 0); -# else - return mmap(0, size, PROT_READ | PROT_WRITE | PROT_EXEC, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0); -# endif -} - - -void Mem::protectExecutableMemory(void *p, size_t size) -{ - mprotect(p, size, PROT_READ | PROT_EXEC); -} - - -void Mem::flushInstructionCache(void *p, size_t size) -{ -# ifndef __FreeBSD__ - __builtin___clear_cache(reinterpret_cast(p), reinterpret_cast(p) + size); -# endif -} diff --git a/src/Summary.cpp b/src/Summary.cpp index 60a9278f9..ab6b7b1ec 100644 --- a/src/Summary.cpp +++ b/src/Summary.cpp @@ -28,141 +28,125 @@ #include -#include "base/net/Pool.h" -#include "common/cpu/Cpu.h" -#include "common/log/Log.h" -#include "core/Config.h" +#include "backend/cpu/Cpu.h" +#include "base/io/log/Log.h" +#include "base/net/stratum/Pool.h" +#include "core/config/Config.h" #include "core/Controller.h" -#include "crypto/Asm.h" -#include "Mem.h" +#include "crypto/common/Assembly.h" +#include "crypto/common/VirtualMemory.h" #include "Summary.h" #include "version.h" -#ifndef XMRIG_NO_ASM +namespace xmrig { + + +#ifdef XMRIG_FEATURE_ASM static const char *coloredAsmNames[] = { - "\x1B[1;31mnone\x1B[0m", + RED_BOLD("none"), "auto", - "\x1B[1;32mintel\x1B[0m", - "\x1B[1;32mryzen\x1B[0m", - "\x1B[1;32mbulldozer\x1B[0m" + GREEN_BOLD("intel"), + GREEN_BOLD("ryzen"), + GREEN_BOLD("bulldozer") }; -inline static const char *asmName(xmrig::Assembly assembly, bool colors) +inline static const char *asmName(Assembly::Id assembly) { - return colors ? coloredAsmNames[assembly] : xmrig::Asm::toString(assembly); + return coloredAsmNames[assembly]; } #endif -static void print_memory(xmrig::Config *config) { +static void print_memory(Config *) { # ifdef _WIN32 - if (config->isColors()) { - Log::i()->text(GREEN_BOLD(" * ") WHITE_BOLD("%-13s") "%s", - "HUGE PAGES", Mem::isHugepagesAvailable() ? "\x1B[1;32mavailable" : "\x1B[01;31munavailable"); + Log::print(GREEN_BOLD(" * ") WHITE_BOLD("%-13s") "%s", + "HUGE PAGES", VirtualMemory::isHugepagesAvailable() ? GREEN_BOLD("permission granted") : RED_BOLD("unavailable")); +# endif +} + + +static void print_cpu(Config *) +{ + const ICpuInfo *info = Cpu::info(); + + Log::print(GREEN_BOLD(" * ") WHITE_BOLD("%-13s%s (%zu)") " %sx64 %sAES %sAVX2", + "CPU", + info->brand(), + info->packages(), + info->isX64() ? GREEN_BOLD_S : RED_BOLD_S "-", + info->hasAES() ? GREEN_BOLD_S : RED_BOLD_S "-", + info->hasAVX2() ? GREEN_BOLD_S : RED_BOLD_S "-" + ); +# if defined(XMRIG_FEATURE_LIBCPUID) || defined (XMRIG_FEATURE_HWLOC) + Log::print(WHITE_BOLD(" %-13s") BLACK_BOLD("L2:") WHITE_BOLD("%.1f MB") BLACK_BOLD(" L3:") WHITE_BOLD("%.1f MB") + CYAN_BOLD(" %zu") "C" BLACK_BOLD("/") CYAN_BOLD("%zu") "T" +# ifdef XMRIG_FEATURE_HWLOC + BLACK_BOLD(" NUMA:") CYAN_BOLD("%zu") +# endif + , "", + info->L2() / 1048576.0, + info->L3() / 1048576.0, + info->cores(), + info->threads() +# ifdef XMRIG_FEATURE_HWLOC + , info->nodes() +# endif + ); +# else + Log::print(WHITE_BOLD(" %-13s") BLACK_BOLD("threads:") CYAN_BOLD("%zu"), + "", + info->threads() + ); +# endif +} + + +static void print_threads(Config *config) +{ + Log::print(GREEN_BOLD(" * ") WHITE_BOLD("%-13s") WHITE_BOLD("%s%d%%"), + "DONATE", + config->pools().donateLevel() == 0 ? RED_BOLD_S : "", + config->pools().donateLevel() + ); + +# ifdef XMRIG_FEATURE_ASM + if (config->cpu().assembly() == Assembly::AUTO) { + const Assembly assembly = Cpu::info()->assembly(); + + Log::print(GREEN_BOLD(" * ") WHITE_BOLD("%-13sauto:%s"), "ASSEMBLY", asmName(assembly)); } else { - Log::i()->text(" * %-13s%s", "HUGE PAGES", Mem::isHugepagesAvailable() ? "available" : "unavailable"); + Log::print(GREEN_BOLD(" * ") WHITE_BOLD("%-13s%s"), "ASSEMBLY", asmName(config->cpu().assembly())); } # endif } -static void print_cpu(xmrig::Config *config) +static void print_commands(Config *) { - using namespace xmrig; - - if (config->isColors()) { - Log::i()->text(GREEN_BOLD(" * ") WHITE_BOLD("%-13s%s (%d)") " %sx64 %sAES %sAVX2", - "CPU", - Cpu::info()->brand(), - Cpu::info()->sockets(), - Cpu::info()->isX64() ? "\x1B[1;32m" : "\x1B[1;31m-", - Cpu::info()->hasAES() ? "\x1B[1;32m" : "\x1B[1;31m-", - Cpu::info()->hasAVX2() ? "\x1B[1;32m" : "\x1B[1;31m-"); -# ifndef XMRIG_NO_LIBCPUID - Log::i()->text(GREEN_BOLD(" * ") WHITE_BOLD("%-13s%.1f MB/%.1f MB"), "CPU L2/L3", Cpu::info()->L2() / 1024.0, Cpu::info()->L3() / 1024.0); -# endif - } - else { - Log::i()->text(" * %-13s%s (%d) %sx64 %sAES %sAVX2", - "CPU", - Cpu::info()->brand(), - Cpu::info()->sockets(), - Cpu::info()->isX64() ? "" : "-", - Cpu::info()->hasAES() ? "" : "-", - Cpu::info()->hasAVX2() ? "" : "-"); -# ifndef XMRIG_NO_LIBCPUID - Log::i()->text(" * %-13s%.1f MB/%.1f MB", "CPU L2/L3", Cpu::info()->L2() / 1024.0, Cpu::info()->L3() / 1024.0); -# endif - } -} - - -static void print_threads(xmrig::Config *config) -{ - if (config->threadsMode() != xmrig::Config::Advanced) { - char buf[32] = { 0 }; - if (config->affinity() != -1L) { - snprintf(buf, sizeof buf, ", affinity=0x%" PRIX64, config->affinity()); - } - - Log::i()->text(config->isColors() ? GREEN_BOLD(" * ") WHITE_BOLD("%-13s") CYAN_BOLD("%d") WHITE_BOLD(", %s, av=%d, %sdonate=%d%%") WHITE_BOLD("%s") - : " * %-13s%d, %s, av=%d, %sdonate=%d%%%s", - "THREADS", - config->threadsCount(), - config->algorithm().name(), - config->algoVariant(), - config->isColors() && config->donateLevel() == 0 ? "\x1B[1;31m" : "", - config->donateLevel(), - buf); - } - else { - Log::i()->text(config->isColors() ? GREEN_BOLD(" * ") WHITE_BOLD("%-13s") CYAN_BOLD("%d") WHITE_BOLD(", %s, %sdonate=%d%%") - : " * %-13s%d, %s, %sdonate=%d%%", - "THREADS", - config->threadsCount(), - config->algorithm().name(), - config->isColors() && config->donateLevel() == 0 ? "\x1B[1;31m" : "", - config->donateLevel()); - } - -# ifndef XMRIG_NO_ASM - if (config->assembly() == xmrig::ASM_AUTO) { - const xmrig::Assembly assembly = xmrig::Cpu::info()->assembly(); - - Log::i()->text(config->isColors() ? GREEN_BOLD(" * ") WHITE_BOLD("%-13sauto:%s") - : " * %-13sauto:%s", "ASSEMBLY", asmName(assembly, config->isColors())); - } - else { - Log::i()->text(config->isColors() ? GREEN_BOLD(" * ") WHITE_BOLD("%-13s%s") : " * %-13s%s", "ASSEMBLY", asmName(config->assembly(), config->isColors())); - } -# endif -} - - -static void print_commands(xmrig::Config *config) -{ - if (config->isColors()) { - Log::i()->text(GREEN_BOLD(" * ") WHITE_BOLD("COMMANDS ") MAGENTA_BOLD("h") WHITE_BOLD("ashrate, ") + if (Log::colors) { + Log::print(GREEN_BOLD(" * ") WHITE_BOLD("COMMANDS ") MAGENTA_BOLD("h") WHITE_BOLD("ashrate, ") MAGENTA_BOLD("p") WHITE_BOLD("ause, ") MAGENTA_BOLD("r") WHITE_BOLD("esume")); } else { - Log::i()->text(" * COMMANDS 'h' hashrate, 'p' pause, 'r' resume"); + Log::print(" * COMMANDS 'h' hashrate, 'p' pause, 'r' resume"); } } -void Summary::print(xmrig::Controller *controller) +} // namespace xmrig + + +void xmrig::Summary::print(Controller *controller) { controller->config()->printVersions(); print_memory(controller->config()); print_cpu(controller->config()); print_threads(controller->config()); - controller->config()->printPools(); - controller->config()->printAPI(); + controller->config()->pools().print(); print_commands(controller->config()); } diff --git a/src/Summary.h b/src/Summary.h index f07dba35f..4317d13e1 100644 --- a/src/Summary.h +++ b/src/Summary.h @@ -5,7 +5,8 @@ * Copyright 2014-2016 Wolf9466 * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -21,20 +22,24 @@ * along with this program. If not, see . */ -#ifndef __SUMMARY_H__ -#define __SUMMARY_H__ +#ifndef XMRIG_SUMMARY_H +#define XMRIG_SUMMARY_H namespace xmrig { - class Controller; -} + + +class Controller; class Summary { public: - static void print(xmrig::Controller *controller); + static void print(Controller *controller); }; -#endif /* __SUMMARY_H__ */ +} // namespace xmrig + + +#endif /* XMRIG_SUMMARY_H */ diff --git a/src/api/Api.cpp b/src/api/Api.cpp index 2abdf4be0..4c8e73230 100644 --- a/src/api/Api.cpp +++ b/src/api/Api.cpp @@ -22,52 +22,187 @@ * along with this program. If not, see . */ -#include + +#include +#ifndef _WIN32 +# include +#endif + + +#include "3rdparty/http-parser/http_parser.h" #include "api/Api.h" -#include "api/ApiRouter.h" -#include "common/api/HttpReply.h" -#include "common/api/HttpRequest.h" +#include "api/interfaces/IApiListener.h" +#include "api/requests/HttpApiRequest.h" +#include "base/kernel/Base.h" +#include "base/tools/Buffer.h" +#include "base/tools/Chrono.h" +#include "core/config/Config.h" +#include "core/Controller.h" +#include "crypto/common/keccak.h" +#include "version.h" -ApiRouter *Api::m_router = nullptr; +#ifdef XMRIG_FEATURE_HTTP +# include "api/Httpd.h" +#endif -bool Api::start(xmrig::Controller *controller) +xmrig::Api::Api(Base *base) : + m_base(base), + m_id(), + m_workerId(), + m_timestamp(Chrono::currentMSecsSinceEpoch()), + m_httpd(nullptr) { - m_router = new ApiRouter(controller); + base->addListener(this); - return true; + genId(base->config()->apiId()); } -void Api::release() +xmrig::Api::~Api() { - delete m_router; +# ifdef XMRIG_FEATURE_HTTP + delete m_httpd; +# endif } -void Api::exec(const xmrig::HttpRequest &req, xmrig::HttpReply &reply) +void xmrig::Api::request(const HttpData &req) { - if (!m_router) { - reply.status = 500; + HttpApiRequest request(req, m_base->config()->http().isRestricted()); + + exec(request); +} + + +void xmrig::Api::start() +{ + genWorkerId(m_base->config()->apiWorkerId()); + +# ifdef XMRIG_FEATURE_HTTP + m_httpd = new Httpd(m_base); + m_httpd->start(); +# endif +} + + +void xmrig::Api::stop() +{ +# ifdef XMRIG_FEATURE_HTTP + m_httpd->stop(); +# endif +} + + +void xmrig::Api::onConfigChanged(Config *config, Config *previousConfig) +{ + if (config->apiId() != previousConfig->apiId()) { + genId(config->apiId()); + } + + if (config->apiWorkerId() != previousConfig->apiWorkerId()) { + genWorkerId(config->apiWorkerId()); + } +} + + +void xmrig::Api::exec(IApiRequest &request) +{ + using namespace rapidjson; + + if (request.type() == IApiRequest::REQ_SUMMARY) { + auto &allocator = request.doc().GetAllocator(); + + request.accept(); + request.reply().AddMember("id", StringRef(m_id), allocator); + request.reply().AddMember("worker_id", StringRef(m_workerId), allocator); + request.reply().AddMember("uptime", (Chrono::currentMSecsSinceEpoch() - m_timestamp) / 1000, allocator); + + Value features(kArrayType); +# ifdef XMRIG_FEATURE_API + features.PushBack("api", allocator); +# endif +# ifdef XMRIG_FEATURE_ASM + features.PushBack("asm", allocator); +# endif +# ifdef XMRIG_FEATURE_HTTP + features.PushBack("http", allocator); +# endif +# ifdef XMRIG_FEATURE_LIBCPUID + features.PushBack("cpuid", allocator); +# endif +# ifdef XMRIG_FEATURE_HWLOC + features.PushBack("hwloc", allocator); +# endif +# ifdef XMRIG_FEATURE_TLS + features.PushBack("tls", allocator); +# endif + request.reply().AddMember("features", features, allocator); + } + + for (IApiListener *listener : m_listeners) { + listener->onRequest(request); + + if (request.isDone()) { + return; + } + } + + request.done(request.isNew() ? HTTP_STATUS_NOT_FOUND : HTTP_STATUS_OK); +} + + +void xmrig::Api::genId(const String &id) +{ + memset(m_id, 0, sizeof(m_id)); + + if (id.size() > 0) { + strncpy(m_id, id.data(), sizeof(m_id) - 1); return; } - if (req.method() == xmrig::HttpRequest::Get) { - return m_router->get(req, reply); - } + uv_interface_address_t *interfaces; + int count = 0; - m_router->exec(req, reply); -} - - -void Api::tick(const xmrig::NetworkState &network) -{ - if (!m_router) { + if (uv_interface_addresses(&interfaces, &count) < 0) { return; } - m_router->tick(network); + for (int i = 0; i < count; i++) { + if (!interfaces[i].is_internal && interfaces[i].address.address4.sin_family == AF_INET) { + uint8_t hash[200]; + const size_t addrSize = sizeof(interfaces[i].phys_addr); + const size_t inSize = strlen(APP_KIND) + addrSize + sizeof(uint16_t); + const uint16_t port = static_cast(m_base->config()->http().port()); + + uint8_t *input = new uint8_t[inSize](); + memcpy(input, &port, sizeof(uint16_t)); + memcpy(input + sizeof(uint16_t), interfaces[i].phys_addr, addrSize); + memcpy(input + sizeof(uint16_t) + addrSize, APP_KIND, strlen(APP_KIND)); + + keccak(input, inSize, hash); + Buffer::toHex(hash, 8, m_id); + + delete [] input; + break; + } + } + + uv_free_interface_addresses(interfaces, count); +} + + +void xmrig::Api::genWorkerId(const String &id) +{ + memset(m_workerId, 0, sizeof(m_workerId)); + + if (id.size() > 0) { + strncpy(m_workerId, id.data(), sizeof(m_workerId) - 1); + } + else { + gethostname(m_workerId, sizeof(m_workerId) - 1); + } } diff --git a/src/api/Api.h b/src/api/Api.h index caee887a6..334609c95 100644 --- a/src/api/Api.h +++ b/src/api/Api.h @@ -26,32 +26,56 @@ #define XMRIG_API_H -#include +#include +#include -class ApiRouter; -class Hashrate; +#include "base/kernel/interfaces/IBaseListener.h" namespace xmrig { - class Controller; - class HttpReply; - class HttpRequest; - class NetworkState; -} -class Api +class Base; +class Httpd; +class HttpData; +class IApiListener; +class IApiRequest; +class String; + + +class Api : public IBaseListener { public: - static bool start(xmrig::Controller *controller); - static void release(); + Api(Base *base); + ~Api() override; - static void exec(const xmrig::HttpRequest &req, xmrig::HttpReply &reply); - static void tick(const xmrig::NetworkState &results); + inline const char *id() const { return m_id; } + inline const char *workerId() const { return m_workerId; } + inline void addListener(IApiListener *listener) { m_listeners.push_back(listener); } + + void request(const HttpData &req); + void start(); + void stop(); + +protected: + void onConfigChanged(Config *config, Config *previousConfig) override; private: - static ApiRouter *m_router; + void exec(IApiRequest &request); + void genId(const String &id); + void genWorkerId(const String &id); + + Base *m_base; + char m_id[32]; + char m_workerId[128]; + const uint64_t m_timestamp; + Httpd *m_httpd; + std::vector m_listeners; }; + +} // namespace xmrig + + #endif /* XMRIG_API_H */ diff --git a/src/api/ApiRouter.cpp b/src/api/ApiRouter.cpp deleted file mode 100644 index beee8fd35..000000000 --- a/src/api/ApiRouter.cpp +++ /dev/null @@ -1,341 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#include -#include -#include - -#if _WIN32 -# include "winsock2.h" -#else -# include "unistd.h" -#endif - - -#include "api/ApiRouter.h" -#include "common/api/HttpReply.h" -#include "common/api/HttpRequest.h" -#include "common/cpu/Cpu.h" -#include "common/crypto/keccak.h" -#include "common/net/Job.h" -#include "common/Platform.h" -#include "core/Config.h" -#include "core/Controller.h" -#include "interfaces/IThread.h" -#include "rapidjson/document.h" -#include "rapidjson/prettywriter.h" -#include "rapidjson/stringbuffer.h" -#include "version.h" -#include "workers/Hashrate.h" -#include "workers/Workers.h" - - -static inline rapidjson::Value normalize(double d) -{ - using namespace rapidjson; - - if (!isnormal(d)) { - return Value(kNullType); - } - - return Value(floor(d * 100.0) / 100.0); -} - - -ApiRouter::ApiRouter(xmrig::Controller *controller) : - m_controller(controller) -{ - memset(m_workerId, 0, sizeof(m_workerId)); - - setWorkerId(controller->config()->apiWorkerId()); - genId(controller->config()->apiId()); -} - - -ApiRouter::~ApiRouter() -{ -} - - -void ApiRouter::ApiRouter::get(const xmrig::HttpRequest &req, xmrig::HttpReply &reply) const -{ - rapidjson::Document doc; - - if (req.match("/1/config")) { - if (req.isRestricted()) { - reply.status = 403; - return; - } - - m_controller->config()->getJSON(doc); - - return finalize(reply, doc); - } - - if (req.match("/1/threads")) { - getThreads(doc); - - return finalize(reply, doc); - } - - doc.SetObject(); - - getIdentify(doc); - getMiner(doc); - getHashrate(doc); - getResults(doc); - getConnection(doc); - - return finalize(reply, doc); -} - - -void ApiRouter::exec(const xmrig::HttpRequest &req, xmrig::HttpReply &reply) -{ - if (req.method() == xmrig::HttpRequest::Put && req.match("/1/config")) { - m_controller->config()->reload(req.body()); - return; - } - - reply.status = 404; -} - - -void ApiRouter::tick(const xmrig::NetworkState &network) -{ - m_network = network; -} - - -void ApiRouter::onConfigChanged(xmrig::Config *config, xmrig::Config *previousConfig) -{ - updateWorkerId(config->apiWorkerId(), previousConfig->apiWorkerId()); -} - - -void ApiRouter::finalize(xmrig::HttpReply &reply, rapidjson::Document &doc) const -{ - rapidjson::StringBuffer buffer(nullptr, 4096); - rapidjson::PrettyWriter writer(buffer); - writer.SetMaxDecimalPlaces(10); - doc.Accept(writer); - - reply.status = 200; - reply.buf = strdup(buffer.GetString()); - reply.size = buffer.GetSize(); -} - - -void ApiRouter::genId(const char *id) -{ - memset(m_id, 0, sizeof(m_id)); - - if (id && strlen(id) > 0) { - strncpy(m_id, id, sizeof(m_id) - 1); - return; - } - - uv_interface_address_t *interfaces; - int count = 0; - - if (uv_interface_addresses(&interfaces, &count) < 0) { - return; - } - - for (int i = 0; i < count; i++) { - if (!interfaces[i].is_internal && interfaces[i].address.address4.sin_family == AF_INET) { - uint8_t hash[200]; - const size_t addrSize = sizeof(interfaces[i].phys_addr); - const size_t inSize = strlen(APP_KIND) + addrSize + sizeof(uint16_t); - const uint16_t port = static_cast(m_controller->config()->apiPort()); - - uint8_t *input = new uint8_t[inSize](); - memcpy(input, &port, sizeof(uint16_t)); - memcpy(input + sizeof(uint16_t), interfaces[i].phys_addr, addrSize); - memcpy(input + sizeof(uint16_t) + addrSize, APP_KIND, strlen(APP_KIND)); - - xmrig::keccak(input, inSize, hash); - xmrig::Job::toHex(hash, 8, m_id); - - delete [] input; - break; - } - } - - uv_free_interface_addresses(interfaces, count); -} - - -void ApiRouter::getConnection(rapidjson::Document &doc) const -{ - auto &allocator = doc.GetAllocator(); - - rapidjson::Value connection(rapidjson::kObjectType); - connection.AddMember("pool", rapidjson::StringRef(m_network.pool), allocator); - connection.AddMember("uptime", m_network.connectionTime(), allocator); - connection.AddMember("ping", m_network.latency(), allocator); - connection.AddMember("failures", m_network.failures, allocator); - connection.AddMember("error_log", rapidjson::Value(rapidjson::kArrayType), allocator); - - doc.AddMember("connection", connection, allocator); -} - - -void ApiRouter::getHashrate(rapidjson::Document &doc) const -{ - auto &allocator = doc.GetAllocator(); - - rapidjson::Value hashrate(rapidjson::kObjectType); - rapidjson::Value total(rapidjson::kArrayType); - rapidjson::Value threads(rapidjson::kArrayType); - - const Hashrate *hr = Workers::hashrate(); - - total.PushBack(normalize(hr->calc(Hashrate::ShortInterval)), allocator); - total.PushBack(normalize(hr->calc(Hashrate::MediumInterval)), allocator); - total.PushBack(normalize(hr->calc(Hashrate::LargeInterval)), allocator); - - for (size_t i = 0; i < Workers::threads(); i++) { - rapidjson::Value thread(rapidjson::kArrayType); - thread.PushBack(normalize(hr->calc(i, Hashrate::ShortInterval)), allocator); - thread.PushBack(normalize(hr->calc(i, Hashrate::MediumInterval)), allocator); - thread.PushBack(normalize(hr->calc(i, Hashrate::LargeInterval)), allocator); - - threads.PushBack(thread, allocator); - } - - hashrate.AddMember("total", total, allocator); - hashrate.AddMember("highest", normalize(hr->highest()), allocator); - hashrate.AddMember("threads", threads, allocator); - doc.AddMember("hashrate", hashrate, allocator); -} - - -void ApiRouter::getIdentify(rapidjson::Document &doc) const -{ - doc.AddMember("id", rapidjson::StringRef(m_id), doc.GetAllocator()); - doc.AddMember("worker_id", rapidjson::StringRef(m_workerId), doc.GetAllocator()); -} - - -void ApiRouter::getMiner(rapidjson::Document &doc) const -{ - using namespace xmrig; - auto &allocator = doc.GetAllocator(); - - rapidjson::Value cpu(rapidjson::kObjectType); - cpu.AddMember("brand", rapidjson::StringRef(Cpu::info()->brand()), allocator); - cpu.AddMember("aes", Cpu::info()->hasAES(), allocator); - cpu.AddMember("x64", Cpu::info()->isX64(), allocator); - cpu.AddMember("sockets", Cpu::info()->sockets(), allocator); - - doc.AddMember("version", APP_VERSION, allocator); - doc.AddMember("kind", APP_KIND, allocator); - doc.AddMember("ua", rapidjson::StringRef(Platform::userAgent()), allocator); - doc.AddMember("cpu", cpu, allocator); - doc.AddMember("algo", rapidjson::StringRef(m_controller->config()->algorithm().name()), allocator); - doc.AddMember("hugepages", Workers::hugePages() > 0, allocator); - doc.AddMember("donate_level", m_controller->config()->donateLevel(), allocator); -} - - -void ApiRouter::getResults(rapidjson::Document &doc) const -{ - auto &allocator = doc.GetAllocator(); - - rapidjson::Value results(rapidjson::kObjectType); - - results.AddMember("diff_current", m_network.diff, allocator); - results.AddMember("shares_good", m_network.accepted, allocator); - results.AddMember("shares_total", m_network.accepted + m_network.rejected, allocator); - results.AddMember("avg_time", m_network.avgTime(), allocator); - results.AddMember("hashes_total", m_network.total, allocator); - - rapidjson::Value best(rapidjson::kArrayType); - for (size_t i = 0; i < m_network.topDiff.size(); ++i) { - best.PushBack(m_network.topDiff[i], allocator); - } - - results.AddMember("best", best, allocator); - results.AddMember("error_log", rapidjson::Value(rapidjson::kArrayType), allocator); - - doc.AddMember("results", results, allocator); -} - - -void ApiRouter::getThreads(rapidjson::Document &doc) const -{ - doc.SetObject(); - auto &allocator = doc.GetAllocator(); - const Hashrate *hr = Workers::hashrate(); - - Workers::threadsSummary(doc); - - const std::vector &threads = m_controller->config()->threads(); - rapidjson::Value list(rapidjson::kArrayType); - - size_t i = 0; - for (const xmrig::IThread *thread : threads) { - rapidjson::Value value = thread->toAPI(doc); - - rapidjson::Value hashrate(rapidjson::kArrayType); - hashrate.PushBack(normalize(hr->calc(i, Hashrate::ShortInterval)), allocator); - hashrate.PushBack(normalize(hr->calc(i, Hashrate::MediumInterval)), allocator); - hashrate.PushBack(normalize(hr->calc(i, Hashrate::LargeInterval)), allocator); - - i++; - - value.AddMember("hashrate", hashrate, allocator); - list.PushBack(value, allocator); - } - - doc.AddMember("threads", list, allocator); -} - - -void ApiRouter::setWorkerId(const char *id) -{ - memset(m_workerId, 0, sizeof(m_workerId)); - - if (id && strlen(id) > 0) { - strncpy(m_workerId, id, sizeof(m_workerId) - 1); - } - else { - gethostname(m_workerId, sizeof(m_workerId) - 1); - } -} - - -void ApiRouter::updateWorkerId(const char *id, const char *previousId) -{ - if (id == previousId) { - return; - } - - if (id != nullptr && previousId != nullptr && strcmp(id, previousId) == 0) { - return; - } - - setWorkerId(id); -} diff --git a/src/api/ApiRouter.h b/src/api/ApiRouter.h deleted file mode 100644 index a92173ce4..000000000 --- a/src/api/ApiRouter.h +++ /dev/null @@ -1,76 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef XMRIG_APIROUTER_H -#define XMRIG_APIROUTER_H - - -#include "api/NetworkState.h" -#include "common/interfaces/IControllerListener.h" -#include "rapidjson/fwd.h" - - -class Hashrate; - - -namespace xmrig { - class Controller; - class HttpReply; - class HttpRequest; -} - - -class ApiRouter : public xmrig::IControllerListener -{ -public: - ApiRouter(xmrig::Controller *controller); - ~ApiRouter() override; - - void get(const xmrig::HttpRequest &req, xmrig::HttpReply &reply) const; - void exec(const xmrig::HttpRequest &req, xmrig::HttpReply &reply); - - void tick(const xmrig::NetworkState &results); - -protected: - void onConfigChanged(xmrig::Config *config, xmrig::Config *previousConfig) override; - -private: - void finalize(xmrig::HttpReply &reply, rapidjson::Document &doc) const; - void genId(const char *id); - void getConnection(rapidjson::Document &doc) const; - void getHashrate(rapidjson::Document &doc) const; - void getIdentify(rapidjson::Document &doc) const; - void getMiner(rapidjson::Document &doc) const; - void getResults(rapidjson::Document &doc) const; - void getThreads(rapidjson::Document &doc) const; - void setWorkerId(const char *id); - void updateWorkerId(const char *id, const char *previousId); - - char m_id[32]; - char m_workerId[128]; - xmrig::NetworkState m_network; - xmrig::Controller *m_controller; -}; - -#endif /* XMRIG_APIROUTER_H */ diff --git a/src/api/Httpd.cpp b/src/api/Httpd.cpp new file mode 100644 index 000000000..57a112e94 --- /dev/null +++ b/src/api/Httpd.cpp @@ -0,0 +1,193 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "3rdparty/http-parser/http_parser.h" +#include "api/Api.h" +#include "api/Httpd.h" +#include "base/io/log/Log.h" +#include "base/net/http/HttpApiResponse.h" +#include "base/net/http/HttpData.h" +#include "base/net/http/HttpServer.h" +#include "base/net/tools/TcpServer.h" +#include "core/config/Config.h" +#include "core/Controller.h" + + +namespace xmrig { + +static const char *kAuthorization = "authorization"; +static const char *kContentType = "content-type"; + +#ifdef _WIN32 +static const char *favicon = nullptr; +static size_t faviconSize = 0; +#endif + +} // namespace xmrig + + +xmrig::Httpd::Httpd(Base *base) : + m_base(base), + m_http(nullptr), + m_server(nullptr), + m_port(0) +{ + base->addListener(this); +} + + +xmrig::Httpd::~Httpd() +{ +} + + +bool xmrig::Httpd::start() +{ + const Http &config = m_base->config()->http(); + + if (!config.isEnabled()) { + return true; + } + + m_http = new HttpServer(this); + m_server = new TcpServer(config.host(), config.port(), m_http); + + const int rc = m_server->bind(); + Log::print(GREEN_BOLD(" * ") WHITE_BOLD("%-13s") BLUE_BOLD("%s:%d") " " RED_BOLD("%s"), + "HTTP API", + config.host().data(), + rc < 0 ? config.port() : rc, + rc < 0 ? uv_strerror(rc) : "" + ); + + if (rc < 0) { + stop(); + + return false; + } + + m_port = static_cast(rc); + +# ifdef _WIN32 + HRSRC src = FindResource(nullptr, MAKEINTRESOURCE(1), RT_ICON); + if (src != nullptr) { + HGLOBAL res = LoadResource(nullptr, src); + if (res != nullptr) { + favicon = static_cast(LockResource(res)); + faviconSize = SizeofResource(nullptr, src); + } + } +# endif + + return true; +} + + +void xmrig::Httpd::stop() +{ + delete m_server; + delete m_http; + + m_server = nullptr; + m_http = nullptr; + m_port = 0; +} + + + +void xmrig::Httpd::onConfigChanged(Config *config, Config *previousConfig) +{ + if (config->http() == previousConfig->http()) { + return; + } + + stop(); + start(); +} + + +void xmrig::Httpd::onHttpData(const HttpData &data) +{ + if (data.method == HTTP_OPTIONS) { + return HttpApiResponse(data.id()).end(); + } + + if (data.method == HTTP_GET && data.url == "/favicon.ico") { +# ifdef _WIN32 + if (favicon != nullptr) { + HttpResponse response(data.id()); + response.setHeader("Content-Type", "image/x-icon"); + + return response.end(favicon, faviconSize); + } +# endif + + return HttpResponse(data.id(), 404).end(); + } + + if (data.method > 4) { + return HttpApiResponse(data.id(), HTTP_STATUS_METHOD_NOT_ALLOWED).end(); + } + + const int status = auth(data); + if (status != HTTP_STATUS_OK) { + return HttpApiResponse(data.id(), status).end(); + } + + if (data.method != HTTP_GET) { + if (m_base->config()->http().isRestricted()) { + return HttpApiResponse(data.id(), HTTP_STATUS_FORBIDDEN).end(); + } + + if (!data.headers.count(kContentType) || data.headers.at(kContentType) != "application/json") { + return HttpApiResponse(data.id(), HTTP_STATUS_UNSUPPORTED_MEDIA_TYPE).end(); + } + } + + m_base->api()->request(data); +} + + +int xmrig::Httpd::auth(const HttpData &req) const +{ + const Http &config = m_base->config()->http(); + + if (!req.headers.count(kAuthorization)) { + return config.isAuthRequired() ? HTTP_STATUS_UNAUTHORIZED : HTTP_STATUS_OK; + } + + if (config.token().isNull()) { + return HTTP_STATUS_UNAUTHORIZED; + } + + const std::string &token = req.headers.at(kAuthorization); + const size_t size = token.size(); + + if (token.size() < 8 || config.token().size() != size - 7 || memcmp("Bearer ", token.c_str(), 7) != 0) { + return HTTP_STATUS_FORBIDDEN; + } + + return strncmp(config.token().data(), token.c_str() + 7, config.token().size()) == 0 ? HTTP_STATUS_OK : HTTP_STATUS_FORBIDDEN; +} diff --git a/src/api/Httpd.h b/src/api/Httpd.h new file mode 100644 index 000000000..220bb7f5a --- /dev/null +++ b/src/api/Httpd.h @@ -0,0 +1,70 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_HTTPD_H +#define XMRIG_HTTPD_H + + +#include + + +#include "base/kernel/interfaces/IBaseListener.h" +#include "base/kernel/interfaces/IHttpListener.h" + + +namespace xmrig { + + +class Base; +class HttpServer; +class TcpServer; + + +class Httpd : public IBaseListener, public IHttpListener +{ +public: + Httpd(Base *base); + ~Httpd() override; + + bool start(); + void stop(); + +protected: + void onConfigChanged(Config *config, Config *previousConfig) override; + void onHttpData(const HttpData &data) override; + +private: + int auth(const HttpData &req) const; + + Base *m_base; + HttpServer *m_http; + TcpServer *m_server; + uint16_t m_port; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_HTTPD_H */ diff --git a/src/core/ConfigCreator.h b/src/api/interfaces/IApiListener.h similarity index 78% rename from src/core/ConfigCreator.h rename to src/api/interfaces/IApiListener.h index 054eb78c9..bbf153a69 100644 --- a/src/core/ConfigCreator.h +++ b/src/api/interfaces/IApiListener.h @@ -20,31 +20,28 @@ * along with this program. If not, see . */ -#ifndef __CONFIGCREATOR_H__ -#define __CONFIGCREATOR_H__ - - -#include "common/interfaces/IConfigCreator.h" -#include "core/Config.h" +#ifndef XMRIG_IAPILISTENER_H +#define XMRIG_IAPILISTENER_H namespace xmrig { -class IConfig; +class IApiRequest; -class ConfigCreator : public IConfigCreator +class IApiListener { public: - inline IConfig *create() const override - { - return new Config(); - } + virtual ~IApiListener() = default; + +# ifdef XMRIG_FEATURE_API + virtual void onRequest(IApiRequest &request) = 0; +# endif }; } /* namespace xmrig */ -#endif // __CONFIGCREATOR_H__ +#endif // XMRIG_IAPILISTENER_H diff --git a/src/api/interfaces/IApiRequest.h b/src/api/interfaces/IApiRequest.h new file mode 100644 index 000000000..8e65a9215 --- /dev/null +++ b/src/api/interfaces/IApiRequest.h @@ -0,0 +1,82 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_IAPIREQUEST_H +#define XMRIG_IAPIREQUEST_H + + +#include "rapidjson/fwd.h" + + +namespace xmrig { + + +class String; + + +class IApiRequest +{ +public: + enum Method { + METHOD_DELETE, + METHOD_GET, + METHOD_HEAD, + METHOD_POST, + METHOD_PUT + }; + + + enum Source { + SOURCE_HTTP + }; + + + enum RequestType { + REQ_UNKNOWN, + REQ_SUMMARY + }; + + + virtual ~IApiRequest() = default; + + virtual bool isDone() const = 0; + virtual bool isNew() const = 0; + virtual bool isRestricted() const = 0; + virtual const rapidjson::Value &json() const = 0; + virtual const String &url() const = 0; + virtual int version() const = 0; + virtual Method method() const = 0; + virtual rapidjson::Document &doc() = 0; + virtual rapidjson::Value &reply() = 0; + virtual RequestType type() const = 0; + virtual Source source() const = 0; + virtual void accept() = 0; + virtual void done(int status) = 0; +}; + + +} /* namespace xmrig */ + + +#endif // XMRIG_IAPIREQUEST_H diff --git a/src/common/net/SubmitResult.cpp b/src/api/requests/ApiRequest.cpp similarity index 78% rename from src/common/net/SubmitResult.cpp rename to src/api/requests/ApiRequest.cpp index d14bbb408..3812e4198 100644 --- a/src/common/net/SubmitResult.cpp +++ b/src/api/requests/ApiRequest.cpp @@ -23,24 +23,16 @@ */ -#include +#include "api/requests/ApiRequest.h" -#include "common/net/SubmitResult.h" - - -xmrig::SubmitResult::SubmitResult(int64_t seq, uint32_t diff, uint64_t actualDiff, int64_t reqId) : - reqId(reqId), - seq(seq), - diff(diff), - actualDiff(actualDiff), - elapsed(0) +xmrig::ApiRequest::ApiRequest(Source source, bool restricted) : + m_restricted(restricted), + m_source(source) { - start = uv_hrtime(); } -void xmrig::SubmitResult::done() +xmrig::ApiRequest::~ApiRequest() { - elapsed = (uv_hrtime() - start) / 1000000; } diff --git a/src/api/requests/ApiRequest.h b/src/api/requests/ApiRequest.h new file mode 100644 index 000000000..05716e29d --- /dev/null +++ b/src/api/requests/ApiRequest.h @@ -0,0 +1,72 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#ifndef XMRIG_APIREQUEST_H +#define XMRIG_APIREQUEST_H + + +#include "api/interfaces/IApiRequest.h" + + +namespace xmrig { + + +class ApiRequest : public IApiRequest +{ +public: + ApiRequest(Source source, bool restricted); + ~ApiRequest() override; + +protected: + inline bool isDone() const override { return m_state == STATE_DONE; } + inline bool isNew() const override { return m_state == STATE_NEW; } + inline bool isRestricted() const override { return m_restricted; } + inline int version() const override { return m_version; } + inline RequestType type() const override { return m_type; } + inline Source source() const override { return m_source; } + inline void accept() override { m_state = STATE_ACCEPTED; } + inline void done(int) override { m_state = STATE_DONE; } + + int m_version = 1; + RequestType m_type = REQ_UNKNOWN; + +private: + enum State { + STATE_NEW, + STATE_ACCEPTED, + STATE_DONE + }; + + bool m_restricted; + Source m_source; + State m_state = STATE_NEW; +}; + + +} // namespace xmrig + + +#endif // XMRIG_APIREQUEST_H + diff --git a/src/api/requests/HttpApiRequest.cpp b/src/api/requests/HttpApiRequest.cpp new file mode 100644 index 000000000..b4dc18107 --- /dev/null +++ b/src/api/requests/HttpApiRequest.cpp @@ -0,0 +1,87 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "api/requests/HttpApiRequest.h" +#include "base/net/http/HttpData.h" +#include "rapidjson/error/en.h" + + +xmrig::HttpApiRequest::HttpApiRequest(const HttpData &req, bool restricted) : + ApiRequest(SOURCE_HTTP, restricted), + m_parsed(false), + m_req(req), + m_res(req.id()), + m_url(req.url.c_str()) +{ + if (method() == METHOD_GET) { + if (url() == "/1/summary" || url() == "/2/summary" || url() == "/api.json") { + m_type = REQ_SUMMARY; + } + } + + if (url().size() > 4) { + if (memcmp(url().data(), "/2/", 3) == 0) { + m_version = 2; + } + } +} + + +const rapidjson::Value &xmrig::HttpApiRequest::json() const +{ + return m_body; +} + + +xmrig::IApiRequest::Method xmrig::HttpApiRequest::method() const +{ + return static_cast(m_req.method); +} + + +void xmrig::HttpApiRequest::accept() +{ + using namespace rapidjson; + + ApiRequest::accept(); + + if (!m_parsed && !m_req.body.empty()) { + m_parsed = true; + m_body.Parse(m_req.body.c_str()); + + if (m_body.HasParseError()) { + reply().AddMember("error", StringRef(GetParseError_En(m_body.GetParseError())), doc().GetAllocator());; + } + } +} + + +void xmrig::HttpApiRequest::done(int status) +{ + ApiRequest::done(status); + + m_res.setStatus(status); + m_res.end(); +} diff --git a/src/api/requests/HttpApiRequest.h b/src/api/requests/HttpApiRequest.h new file mode 100644 index 000000000..f34d4be5d --- /dev/null +++ b/src/api/requests/HttpApiRequest.h @@ -0,0 +1,69 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#ifndef XMRIG_HTTPAPIREQUEST_H +#define XMRIG_HTTPAPIREQUEST_H + + +#include "api/requests/ApiRequest.h" +#include "base/net/http/HttpApiResponse.h" +#include "base/tools/String.h" + + +namespace xmrig { + + +class HttpData; + + +class HttpApiRequest : public ApiRequest +{ +public: + HttpApiRequest(const HttpData &req, bool restricted); + +protected: + inline rapidjson::Document &doc() override { return m_res.doc(); } + inline rapidjson::Value &reply() override { return m_res.doc(); } + inline const String &url() const override { return m_url; } + + const rapidjson::Value &json() const override; + Method method() const override; + void accept() override; + void done(int status) override; + +private: + bool m_parsed; + const HttpData &m_req; + HttpApiResponse m_res; + rapidjson::Document m_body; + String m_url; +}; + + +} // namespace xmrig + + +#endif // XMRIG_HTTPAPIREQUEST_H + diff --git a/src/backend/backend.cmake b/src/backend/backend.cmake new file mode 100644 index 000000000..c37cf262a --- /dev/null +++ b/src/backend/backend.cmake @@ -0,0 +1,13 @@ +include (src/backend/cpu/cpu.cmake) +include (src/backend/common/common.cmake) + + +set(HEADERS_BACKEND + "${HEADERS_BACKEND_COMMON}" + "${HEADERS_BACKEND_CPU}" + ) + +set(SOURCES_BACKEND + "${SOURCES_BACKEND_COMMON}" + "${SOURCES_BACKEND_CPU}" + ) diff --git a/src/workers/Hashrate.cpp b/src/backend/common/Hashrate.cpp similarity index 59% rename from src/workers/Hashrate.cpp rename to src/backend/common/Hashrate.cpp index 2a7503180..99a9a9c5f 100644 --- a/src/workers/Hashrate.cpp +++ b/src/backend/common/Hashrate.cpp @@ -5,7 +5,8 @@ * Copyright 2014-2016 Wolf9466 * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -23,21 +24,20 @@ #include -#include -#include +#include #include #include -#include "common/log/Log.h" -#include "core/Config.h" -#include "core/Controller.h" -#include "workers/Hashrate.h" +#include "backend/common/Hashrate.h" +#include "base/tools/Chrono.h" +#include "base/tools/Handle.h" +#include "rapidjson/document.h" inline static const char *format(double h, char *buf, size_t size) { - if (isnormal(h)) { + if (std::isnormal(h)) { snprintf(buf, size, "%03.1f", h); return buf; } @@ -46,10 +46,9 @@ inline static const char *format(double h, char *buf, size_t size) } -Hashrate::Hashrate(size_t threads, xmrig::Controller *controller) : +xmrig::Hashrate::Hashrate(size_t threads) : m_highest(0.0), - m_threads(threads), - m_controller(controller) + m_threads(threads) { m_counts = new uint64_t*[threads]; m_timestamps = new uint64_t*[threads]; @@ -60,26 +59,30 @@ Hashrate::Hashrate(size_t threads, xmrig::Controller *controller) : m_timestamps[i] = new uint64_t[kBucketSize](); m_top[i] = 0; } - - const int printTime = controller->config()->printTime(); - - if (printTime > 0) { - uv_timer_init(uv_default_loop(), &m_timer); - m_timer.data = this; - - uv_timer_start(&m_timer, Hashrate::onReport, (printTime + 4) * 1000, printTime * 1000); - } } -double Hashrate::calc(size_t ms) const +xmrig::Hashrate::~Hashrate() +{ + for (size_t i = 0; i < m_threads; i++) { + delete [] m_counts[i]; + delete [] m_timestamps[i]; + } + + delete [] m_counts; + delete [] m_timestamps; + delete [] m_top; +} + + +double xmrig::Hashrate::calc(size_t ms) const { double result = 0.0; double data; for (size_t i = 0; i < m_threads; ++i) { data = calc(i, ms); - if (isnormal(data)) { + if (std::isnormal(data)) { result += data; } } @@ -88,16 +91,13 @@ double Hashrate::calc(size_t ms) const } -double Hashrate::calc(size_t threadId, size_t ms) const +double xmrig::Hashrate::calc(size_t threadId, size_t ms) const { assert(threadId < m_threads); if (threadId >= m_threads) { return nan(""); } - using namespace std::chrono; - const uint64_t now = time_point_cast(high_resolution_clock::now()).time_since_epoch().count(); - uint64_t earliestHashCount = 0; uint64_t earliestStamp = 0; uint64_t lastestStamp = 0; @@ -116,7 +116,7 @@ double Hashrate::calc(size_t threadId, size_t ms) const lastestHashCnt = m_counts[threadId][idx]; } - if (now - m_timestamps[threadId][idx] > ms) { + if (xmrig::Chrono::highResolutionMSecs() - m_timestamps[threadId][idx] > ms) { haveFullSet = true; break; } @@ -133,16 +133,14 @@ double Hashrate::calc(size_t threadId, size_t ms) const return nan(""); } - double hashes, time; - hashes = (double) lastestHashCnt - earliestHashCount; - time = (double) lastestStamp - earliestStamp; - time /= 1000.0; + const double hashes = static_cast(lastestHashCnt - earliestHashCount); + const double time = static_cast(lastestStamp - earliestStamp) / 1000.0; return hashes / time; } -void Hashrate::add(size_t threadId, uint64_t count, uint64_t timestamp) +void xmrig::Hashrate::add(size_t threadId, uint64_t count, uint64_t timestamp) { const size_t top = m_top[threadId]; m_counts[threadId][top] = count; @@ -152,45 +150,28 @@ void Hashrate::add(size_t threadId, uint64_t count, uint64_t timestamp) } -void Hashrate::print() const -{ - char num1[8] = { 0 }; - char num2[8] = { 0 }; - char num3[8] = { 0 }; - char num4[8] = { 0 }; - - LOG_INFO(m_controller->config()->isColors() ? WHITE_BOLD("speed") " 10s/60s/15m " CYAN_BOLD("%s") CYAN(" %s %s ") CYAN_BOLD("H/s") " max " CYAN_BOLD("%s H/s") - : "speed 10s/60s/15m %s %s %s H/s max %s H/s", - format(calc(ShortInterval), num1, sizeof(num1)), - format(calc(MediumInterval), num2, sizeof(num2)), - format(calc(LargeInterval), num3, sizeof(num3)), - format(m_highest, num4, sizeof(num4)) - ); -} - - -void Hashrate::stop() -{ - uv_timer_stop(&m_timer); -} - - -void Hashrate::updateHighest() +void xmrig::Hashrate::updateHighest() { double highest = calc(ShortInterval); - if (isnormal(highest) && highest > m_highest) { + if (std::isnormal(highest) && highest > m_highest) { m_highest = highest; } } -const char *Hashrate::format(double h, char *buf, size_t size) +const char *xmrig::Hashrate::format(double h, char *buf, size_t size) { return ::format(h, buf, size); } -void Hashrate::onReport(uv_timer_t *handle) +rapidjson::Value xmrig::Hashrate::normalize(double d) { - static_cast(handle->data)->print(); + using namespace rapidjson; + + if (!std::isnormal(d)) { + return Value(kNullType); + } + + return Value(floor(d * 100.0) / 100.0); } diff --git a/src/workers/Hashrate.h b/src/backend/common/Hashrate.h similarity index 83% rename from src/workers/Hashrate.h rename to src/backend/common/Hashrate.h index e766f1170..0674c6aba 100644 --- a/src/workers/Hashrate.h +++ b/src/backend/common/Hashrate.h @@ -5,7 +5,8 @@ * Copyright 2014-2016 Wolf9466 * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -21,17 +22,18 @@ * along with this program. If not, see . */ -#ifndef __HASHRATE_H__ -#define __HASHRATE_H__ +#ifndef XMRIG_HASHRATE_H +#define XMRIG_HASHRATE_H +#include #include -#include + + +#include "rapidjson/fwd.h" namespace xmrig { - class Controller; -} class Hashrate @@ -43,22 +45,20 @@ public: LargeInterval = 900000 }; - Hashrate(size_t threads, xmrig::Controller *controller); + Hashrate(size_t threads); + ~Hashrate(); double calc(size_t ms) const; double calc(size_t threadId, size_t ms) const; void add(size_t threadId, uint64_t count, uint64_t timestamp); - void print() const; - void stop(); void updateHighest(); inline double highest() const { return m_highest; } inline size_t threads() const { return m_threads; } static const char *format(double h, char *buf, size_t size); + static rapidjson::Value normalize(double d); private: - static void onReport(uv_timer_t *handle); - constexpr static size_t kBucketSize = 2 << 11; constexpr static size_t kBucketMask = kBucketSize - 1; @@ -67,9 +67,10 @@ private: uint32_t* m_top; uint64_t** m_counts; uint64_t** m_timestamps; - uv_timer_t m_timer; - xmrig::Controller *m_controller; }; -#endif /* __HASHRATE_H__ */ +} // namespace xmrig + + +#endif /* XMRIG_HASHRATE_H */ diff --git a/src/backend/common/Thread.h b/src/backend/common/Thread.h new file mode 100644 index 000000000..b71659151 --- /dev/null +++ b/src/backend/common/Thread.h @@ -0,0 +1,67 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_THREAD_H +#define XMRIG_THREAD_H + + +#include + + +#include "backend/common/interfaces/IWorker.h" + + +namespace xmrig { + + +class IBackend; + + +template +class Thread +{ +public: + inline Thread(IBackend *backend, size_t index, const T &config) : m_index(index), m_config(config), m_backend(backend) {} + inline ~Thread() { m_thread.join(); delete m_worker; } + + inline const T &config() const { return m_config; } + inline IBackend *backend() const { return m_backend; } + inline IWorker *worker() const { return m_worker; } + inline size_t index() const { return m_index; } + inline void setWorker(IWorker *worker) { m_worker = worker; } + inline void start(void (*callback) (void *)) { m_thread = std::thread(callback, this); } + +private: + const size_t m_index = 0; + const T m_config; + IBackend *m_backend; + IWorker *m_worker = nullptr; + std::thread m_thread; +}; + + +} // namespace xmrig + + +#endif /* XMRIG_THREAD_H */ diff --git a/src/backend/common/Threads.cpp b/src/backend/common/Threads.cpp new file mode 100644 index 000000000..17fc29511 --- /dev/null +++ b/src/backend/common/Threads.cpp @@ -0,0 +1,151 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "backend/common/Threads.h" +#include "backend/cpu/CpuThreads.h" +#include "rapidjson/document.h" + + +namespace xmrig { + + +static const char *kAsterisk = "*"; + + +} // namespace xmrig + + +template +const T &xmrig::Threads::get(const String &profileName) const +{ + static T empty; + if (profileName.isNull() || !has(profileName)) { + return empty; + } + + return m_profiles.at(profileName); +} + + +template +size_t xmrig::Threads::read(const rapidjson::Value &value) +{ + using namespace rapidjson; + + for (auto &member : value.GetObject()) { + if (member.value.IsArray() || member.value.IsObject()) { + T threads(member.value); + + if (!threads.isEmpty()) { + move(member.name.GetString(), std::move(threads)); + } + + continue; + } + + const Algorithm algo(member.name.GetString()); + if (!algo.isValid()) { + continue; + } + + if (member.value.IsBool() && member.value.IsFalse()) { + disable(algo); + continue; + } + + if (member.value.IsString()) { + if (has(member.value.GetString())) { + m_aliases.insert({ algo, member.value.GetString() }); + } + else { + m_disabled.insert(algo); + } + } + } + + return m_profiles.size(); +} + + +template +xmrig::String xmrig::Threads::profileName(const Algorithm &algorithm, bool strict) const +{ + if (isDisabled(algorithm)) { + return String(); + } + + const String name = algorithm.shortName(); + if (has(name)) { + return name; + } + + if (m_aliases.count(algorithm) > 0) { + return m_aliases.at(algorithm); + } + + if (strict) { + return String(); + } + + if (name.contains("/")) { + const String base = name.split('/').at(0); + if (has(base)) { + return base; + } + } + + if (has(kAsterisk)) { + return kAsterisk; + } + + return String(); +} + + +template +void xmrig::Threads::toJSON(rapidjson::Value &out, rapidjson::Document &doc) const +{ + using namespace rapidjson; + auto &allocator = doc.GetAllocator(); + + for (const auto &kv : m_profiles) { + out.AddMember(kv.first.toJSON(), kv.second.toJSON(doc), allocator); + } + + for (const Algorithm &algo : m_disabled) { + out.AddMember(StringRef(algo.shortName()), false, allocator); + } + + for (const auto &kv : m_aliases) { + out.AddMember(StringRef(kv.first.shortName()), kv.second.toJSON(), allocator); + } +} + + +namespace xmrig { + +template class Threads; + +} // namespace xmrig diff --git a/src/backend/common/Threads.h b/src/backend/common/Threads.h new file mode 100644 index 000000000..2cb333d6f --- /dev/null +++ b/src/backend/common/Threads.h @@ -0,0 +1,67 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_THREADS_H +#define XMRIG_THREADS_H + + +#include +#include + + +#include "base/tools/String.h" +#include "crypto/common/Algorithm.h" +#include "rapidjson/fwd.h" + + +namespace xmrig { + + +template +class Threads +{ +public: + inline bool has(const char *profile) const { return m_profiles.count(profile) > 0; } + inline bool isDisabled(const Algorithm &algo) const { return m_disabled.count(algo) > 0; } + inline bool isExist(const Algorithm &algo) const { return isDisabled(algo) || m_aliases.count(algo) > 0 || has(algo.shortName()); } + inline const T &get(const Algorithm &algo, bool strict = false) const { return get(profileName(algo, strict)); } + inline void disable(const Algorithm &algo) { m_disabled.insert(algo); } + inline void move(const char *profile, T &&threads) { m_profiles.insert({ profile, threads }); } + + const T &get(const String &profileName) const; + size_t read(const rapidjson::Value &value); + String profileName(const Algorithm &algorithm, bool strict = false) const; + void toJSON(rapidjson::Value &out, rapidjson::Document &doc) const; + +private: + std::map m_aliases; + std::map m_profiles; + std::set m_disabled; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_THREADS_H */ diff --git a/src/backend/common/Worker.cpp b/src/backend/common/Worker.cpp new file mode 100644 index 000000000..91ef0c7ad --- /dev/null +++ b/src/backend/common/Worker.cpp @@ -0,0 +1,51 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "backend/common/Worker.h" +#include "base/kernel/Platform.h" +#include "base/tools/Chrono.h" +#include "crypto/common/VirtualMemory.h" + + +xmrig::Worker::Worker(size_t id, int64_t affinity, int priority) : + m_affinity(affinity), + m_id(id), + m_hashCount(0), + m_timestamp(0), + m_count(0) +{ + m_node = VirtualMemory::bindToNUMANode(affinity); + + Platform::trySetThreadAffinity(affinity); + Platform::setThreadPriority(priority); +} + + +void xmrig::Worker::storeStats() +{ + m_hashCount.store(m_count, std::memory_order_relaxed); + m_timestamp.store(Chrono::highResolutionMSecs(), std::memory_order_relaxed); +} diff --git a/src/workers/Worker.h b/src/backend/common/Worker.h similarity index 66% rename from src/workers/Worker.h rename to src/backend/common/Worker.h index 73e250330..5f5df9250 100644 --- a/src/workers/Worker.h +++ b/src/backend/common/Worker.h @@ -5,7 +5,9 @@ * Copyright 2014-2016 Wolf9466 * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -29,41 +31,35 @@ #include -#include "interfaces/IWorker.h" -#include "Mem.h" - - -class Handle; +#include "backend/common/interfaces/IWorker.h" namespace xmrig { - class CpuThread; -} class Worker : public IWorker { public: - Worker(Handle *handle); + Worker(size_t id, int64_t affinity, int priority); - inline const MemInfo &memory() const { return m_memory; } - inline size_t id() const override { return m_id; } - inline uint64_t hashCount() const override { return m_hashCount.load(std::memory_order_relaxed); } - inline uint64_t timestamp() const override { return m_timestamp.load(std::memory_order_relaxed); } + inline const VirtualMemory *memory() const override { return nullptr; } + inline size_t id() const override { return m_id; } + inline uint64_t hashCount() const override { return m_hashCount.load(std::memory_order_relaxed); } + inline uint64_t timestamp() const override { return m_timestamp.load(std::memory_order_relaxed); } protected: void storeStats(); + const int64_t m_affinity; const size_t m_id; - const size_t m_totalWays; - const uint32_t m_offset; - MemInfo m_memory; std::atomic m_hashCount; std::atomic m_timestamp; + uint32_t m_node = 0; uint64_t m_count; - uint64_t m_sequence; - xmrig::CpuThread *m_thread; }; +} // namespace xmrig + + #endif /* XMRIG_WORKER_H */ diff --git a/src/backend/common/WorkerJob.h b/src/backend/common/WorkerJob.h new file mode 100644 index 000000000..c9a3d55ca --- /dev/null +++ b/src/backend/common/WorkerJob.h @@ -0,0 +1,142 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_WORKERJOB_H +#define XMRIG_WORKERJOB_H + + +#include + + +#include "base/net/stratum/Job.h" +#include "crypto/common/Nonce.h" + + +namespace xmrig { + + +template +class WorkerJob +{ +public: + inline const Job ¤tJob() const { return m_jobs[index()]; } + inline uint32_t *nonce(size_t i = 0) { return reinterpret_cast(blob() + (i * currentJob().size()) + 39); } + inline uint64_t sequence() const { return m_sequence; } + inline uint8_t *blob() { return m_blobs[index()]; } + inline uint8_t index() const { return m_index; } + + + inline void add(const Job &job, uint64_t sequence, uint32_t reserveCount) + { + m_sequence = sequence; + + if (currentJob() == job) { + return; + } + + if (index() == 1 && job.index() == 0 && job == m_jobs[0]) { + return; + } + + save(job, reserveCount); + } + + + inline void nextRound(uint32_t reserveCount) + { + m_rounds[index()]++; + + if ((m_rounds[index()] % reserveCount) == 0) { + for (size_t i = 0; i < N; ++i) { + *nonce(i) = Nonce::next(index(), *nonce(i), reserveCount, currentJob().isNicehash()); + } + } + else { + for (size_t i = 0; i < N; ++i) { + *nonce(i) += 1; + } + } + } + + +private: + inline void save(const Job &job, uint32_t reserveCount) + { + m_index = job.index(); + const size_t size = job.size(); + m_jobs[index()] = job; + m_rounds[index()] = 0; + + for (size_t i = 0; i < N; ++i) { + memcpy(m_blobs[index()] + (i * size), job.blob(), size); + *nonce(i) = Nonce::next(index(), *nonce(i), reserveCount, job.isNicehash()); + } + } + + + alignas(16) uint8_t m_blobs[2][Job::kMaxBlobSize * N]; + Job m_jobs[2]; + uint32_t m_rounds[2] = { 0, 0 }; + uint64_t m_sequence = 0; + uint8_t m_index = 0; +}; + + +template<> +inline uint32_t *xmrig::WorkerJob<1>::nonce(size_t) +{ + return reinterpret_cast(blob() + 39); +} + + +template<> +inline void xmrig::WorkerJob<1>::nextRound(uint32_t reserveCount) +{ + m_rounds[index()]++; + + if ((m_rounds[index()] % reserveCount) == 0) { + *nonce() = Nonce::next(index(), *nonce(), reserveCount, currentJob().isNicehash()); + } + else { + *nonce() += 1; + } +} + + +template<> +inline void xmrig::WorkerJob<1>::save(const Job &job, uint32_t reserveCount) +{ + m_index = job.index(); + m_jobs[index()] = job; + m_rounds[index()] = 0; + + memcpy(blob(), job.blob(), job.size()); + *nonce() = Nonce::next(index(), *nonce(), reserveCount, currentJob().isNicehash()); +} + + +} // namespace xmrig + + +#endif /* XMRIG_WORKERJOB_H */ diff --git a/src/backend/common/Workers.cpp b/src/backend/common/Workers.cpp new file mode 100644 index 000000000..036bc8b66 --- /dev/null +++ b/src/backend/common/Workers.cpp @@ -0,0 +1,204 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "backend/common/Hashrate.h" +#include "backend/common/interfaces/IBackend.h" +#include "backend/common/Workers.h" +#include "backend/cpu/CpuWorker.h" +#include "base/io/log/Log.h" + + +namespace xmrig { + + +class WorkersPrivate +{ +public: + inline WorkersPrivate() + { + } + + + inline ~WorkersPrivate() + { + delete hashrate; + } + + + Hashrate *hashrate = nullptr; + IBackend *backend = nullptr; +}; + + +} // namespace xmrig + + +template +xmrig::Workers::Workers() : + d_ptr(new WorkersPrivate()) +{ + +} + + +template +xmrig::Workers::~Workers() +{ + delete d_ptr; +} + + +template +const xmrig::Hashrate *xmrig::Workers::hashrate() const +{ + return d_ptr->hashrate; +} + + +template +void xmrig::Workers::setBackend(IBackend *backend) +{ + d_ptr->backend = backend; +} + + +template +void xmrig::Workers::start(const std::vector &data) +{ + for (const T &item : data) { + m_workers.push_back(new Thread(d_ptr->backend, m_workers.size(), item)); + } + + d_ptr->hashrate = new Hashrate(m_workers.size()); + + for (Thread *worker : m_workers) { + worker->start(Workers::onReady); + } +} + + +template +void xmrig::Workers::stop() +{ + Nonce::stop(T::backend()); + + for (Thread *worker : m_workers) { + delete worker; + } + + m_workers.clear(); + Nonce::touch(T::backend()); + + delete d_ptr->hashrate; + d_ptr->hashrate = nullptr; +} + + +template +void xmrig::Workers::tick(uint64_t) +{ + if (!d_ptr->hashrate) { + return; + } + + for (Thread *handle : m_workers) { + if (!handle->worker()) { + return; + } + + d_ptr->hashrate->add(handle->index(), handle->worker()->hashCount(), handle->worker()->timestamp()); + } + + d_ptr->hashrate->updateHighest(); +} + + +template +xmrig::IWorker *xmrig::Workers::create(Thread *) +{ + return nullptr; +} + + +template +void xmrig::Workers::onReady(void *arg) +{ + Thread *handle = static_cast* >(arg); + + IWorker *worker = create(handle); + if (!worker || !worker->selfTest()) { + LOG_ERR("thread %zu error: \"hash self-test failed\".", worker->id()); + + return; + } + + handle->setWorker(worker); + handle->backend()->start(worker); +} + + +namespace xmrig { + + +template<> +xmrig::IWorker *xmrig::Workers::create(Thread *handle) +{ + const int intensity = handle->config().intensity; + +# if defined(XMRIG_ALGO_RANDOMX) || defined(XMRIG_ALGO_CN_GPU) + if (intensity > handle->config().algorithm.maxIntensity()) { + LOG_WARN("CPU thread %zu warning: \"intensity %d not supported for %s algorithm\".", handle->index(), handle->config().intensity, handle->config().algorithm.shortName()); + + return new CpuWorker<1>(handle->index(), handle->config()); + } +# endif + + + switch (intensity) { + case 1: + return new CpuWorker<1>(handle->index(), handle->config()); + + case 2: + return new CpuWorker<2>(handle->index(), handle->config()); + + case 3: + return new CpuWorker<3>(handle->index(), handle->config()); + + case 4: + return new CpuWorker<4>(handle->index(), handle->config()); + + case 5: + return new CpuWorker<5>(handle->index(), handle->config()); + } + + return nullptr; +} + + +template class Workers; + + +} // namespace xmrig diff --git a/src/backend/common/Workers.h b/src/backend/common/Workers.h new file mode 100644 index 000000000..77dd434c2 --- /dev/null +++ b/src/backend/common/Workers.h @@ -0,0 +1,73 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_WORKERS_H +#define XMRIG_WORKERS_H + + +#include "backend/common/Thread.h" +#include "backend/cpu/CpuLaunchData.h" + + +namespace xmrig { + + +class Hashrate; +class WorkersPrivate; + + +template +class Workers +{ +public: + Workers(); + ~Workers(); + + const Hashrate *hashrate() const; + void setBackend(IBackend *backend); + void start(const std::vector &data); + void stop(); + void tick(uint64_t ticks); + +private: + static IWorker *create(Thread *handle); + static void onReady(void *arg); + + std::vector *> m_workers; + WorkersPrivate *d_ptr; +}; + + +template<> +IWorker *Workers::create(Thread *handle); + + +extern template class Workers; + + +} // namespace xmrig + + +#endif /* XMRIG_WORKERS_H */ diff --git a/src/backend/common/common.cmake b/src/backend/common/common.cmake new file mode 100644 index 000000000..c470ea507 --- /dev/null +++ b/src/backend/common/common.cmake @@ -0,0 +1,18 @@ +set(HEADERS_BACKEND_COMMON + src/backend/common/interfaces/IBackend.h + src/backend/common/interfaces/IThread.h + src/backend/common/interfaces/IWorker.h + src/backend/common/Hashrate.h + src/backend/common/Thread.h + src/backend/common/Threads.h + src/backend/common/Worker.h + src/backend/common/Workers.h + src/backend/common/WorkerJob.h + ) + +set(SOURCES_BACKEND_COMMON + src/backend/common/Hashrate.cpp + src/backend/common/Threads.cpp + src/backend/common/Worker.cpp + src/backend/common/Workers.cpp + ) diff --git a/src/backend/common/interfaces/IBackend.h b/src/backend/common/interfaces/IBackend.h new file mode 100644 index 000000000..2ec8bf045 --- /dev/null +++ b/src/backend/common/interfaces/IBackend.h @@ -0,0 +1,71 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_IBACKEND_H +#define XMRIG_IBACKEND_H + + +#include + + +#include "rapidjson/fwd.h" + + +namespace xmrig { + + +class Algorithm; +class Hashrate; +class IWorker; +class Job; +class String; + + +class IBackend +{ +public: + virtual ~IBackend() = default; + + virtual bool isEnabled() const = 0; + virtual bool isEnabled(const Algorithm &algorithm) const = 0; + virtual const Hashrate *hashrate() const = 0; + virtual const String &profileName() const = 0; + virtual const String &type() const = 0; + virtual void prepare(const Job &nextJob) = 0; + virtual void printHashrate(bool details) = 0; + virtual void setJob(const Job &job) = 0; + virtual void start(IWorker *worker) = 0; + virtual void stop() = 0; + virtual void tick(uint64_t ticks) = 0; + +# ifdef XMRIG_FEATURE_API + virtual rapidjson::Value toJSON(rapidjson::Document &doc) const = 0; +# endif +}; + + +} // namespace xmrig + + +#endif // XMRIG_IBACKEND_H diff --git a/src/interfaces/IThread.h b/src/backend/common/interfaces/IThread.h similarity index 92% rename from src/interfaces/IThread.h rename to src/backend/common/interfaces/IThread.h index 3a8708e63..3c0a7287a 100644 --- a/src/interfaces/IThread.h +++ b/src/backend/common/interfaces/IThread.h @@ -27,7 +27,7 @@ #include -#include "common/xmrig.h" +#include "crypto/common/Algorithm.h" #include "rapidjson/fwd.h" @@ -51,9 +51,9 @@ public: PentaWay }; - virtual ~IThread() {} + virtual ~IThread() = default; - virtual Algo algorithm() const = 0; + virtual Algorithm algorithm() const = 0; virtual int priority() const = 0; virtual int64_t affinity() const = 0; virtual Multiway multiway() const = 0; @@ -61,7 +61,7 @@ public: virtual size_t index() const = 0; virtual Type type() const = 0; -# ifndef XMRIG_NO_API +# ifdef XMRIG_FEATURE_API virtual rapidjson::Value toAPI(rapidjson::Document &doc) const = 0; # endif diff --git a/src/interfaces/IWorker.h b/src/backend/common/interfaces/IWorker.h similarity index 76% rename from src/interfaces/IWorker.h rename to src/backend/common/interfaces/IWorker.h index 83e9306ec..0d7fe1d26 100644 --- a/src/interfaces/IWorker.h +++ b/src/backend/common/interfaces/IWorker.h @@ -27,6 +27,13 @@ #include +#include + + +namespace xmrig { + + +class VirtualMemory; class IWorker @@ -34,12 +41,16 @@ class IWorker public: virtual ~IWorker() = default; - virtual bool selfTest() = 0; - virtual size_t id() const = 0; - virtual uint64_t hashCount() const = 0; - virtual uint64_t timestamp() const = 0; - virtual void start() = 0; + virtual bool selfTest() = 0; + virtual const VirtualMemory *memory() const = 0; + virtual size_t id() const = 0; + virtual uint64_t hashCount() const = 0; + virtual uint64_t timestamp() const = 0; + virtual void start() = 0; }; +} // namespace xmrig + + #endif // XMRIG_IWORKER_H diff --git a/src/backend/cpu/Cpu.cpp b/src/backend/cpu/Cpu.cpp new file mode 100644 index 000000000..4d9effce7 --- /dev/null +++ b/src/backend/cpu/Cpu.cpp @@ -0,0 +1,104 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include + + +#include "backend/cpu/Cpu.h" +#include "rapidjson/document.h" + + +#if defined(XMRIG_FEATURE_HWLOC) +# include "backend/cpu/platform/HwlocCpuInfo.h" +#elif defined(XMRIG_FEATURE_LIBCPUID) +# include "backend/cpu/platform/AdvancedCpuInfo.h" +#else +# include "backend/cpu/platform/BasicCpuInfo.h" +#endif + + +static xmrig::ICpuInfo *cpuInfo = nullptr; + + +xmrig::ICpuInfo *xmrig::Cpu::info() +{ + assert(cpuInfo != nullptr); + + return cpuInfo; +} + + +rapidjson::Value xmrig::Cpu::toJSON(rapidjson::Document &doc) +{ + using namespace rapidjson; + auto &allocator = doc.GetAllocator(); + + ICpuInfo *i = info(); + Value cpu(kObjectType); + Assembly assembly(i->assembly()); + + cpu.AddMember("brand", StringRef(i->brand()), allocator); + cpu.AddMember("aes", i->hasAES(), allocator); + cpu.AddMember("avx2", i->hasAVX2(), allocator); + cpu.AddMember("x64", i->isX64(), allocator); + cpu.AddMember("l2", static_cast(i->L2()), allocator); + cpu.AddMember("l3", static_cast(i->L3()), allocator); + cpu.AddMember("cores", static_cast(i->cores()), allocator); + cpu.AddMember("threads", static_cast(i->threads()), allocator); + cpu.AddMember("packages", static_cast(i->packages()), allocator); + cpu.AddMember("nodes", static_cast(i->nodes()), allocator); + cpu.AddMember("backend", StringRef(i->backend()), allocator); + +# ifdef XMRIG_FEATURE_ASM + cpu.AddMember("assembly", StringRef(assembly.toString()), allocator); +# else + cpu.AddMember("assembly", "none", allocator); +# endif + + return cpu; +} + + +void xmrig::Cpu::init() +{ + assert(cpuInfo == nullptr); + +# if defined(XMRIG_FEATURE_HWLOC) + cpuInfo = new HwlocCpuInfo(); +# elif defined(XMRIG_FEATURE_LIBCPUID) + cpuInfo = new AdvancedCpuInfo(); +# else + cpuInfo = new BasicCpuInfo(); +# endif +} + + +void xmrig::Cpu::release() +{ + assert(cpuInfo != nullptr); + + delete cpuInfo; + cpuInfo = nullptr; +} diff --git a/src/common/cpu/Cpu.h b/src/backend/cpu/Cpu.h similarity index 78% rename from src/common/cpu/Cpu.h rename to src/backend/cpu/Cpu.h index 1d5a9fb1d..bece97d3a 100644 --- a/src/common/cpu/Cpu.h +++ b/src/backend/cpu/Cpu.h @@ -5,7 +5,8 @@ * Copyright 2014-2016 Wolf9466 * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -25,7 +26,7 @@ #define XMRIG_CPU_H -#include "common/interfaces/ICpuInfo.h" +#include "backend/cpu/interfaces/ICpuInfo.h" namespace xmrig { @@ -35,8 +36,11 @@ class Cpu { public: static ICpuInfo *info(); + static rapidjson::Value toJSON(rapidjson::Document &doc); static void init(); static void release(); + + inline static Assembly::Id assembly(Assembly::Id hint) { return hint == Assembly::AUTO ? Cpu::info()->assembly() : hint; } }; diff --git a/src/backend/cpu/CpuBackend.cpp b/src/backend/cpu/CpuBackend.cpp new file mode 100644 index 000000000..77baeaf60 --- /dev/null +++ b/src/backend/cpu/CpuBackend.cpp @@ -0,0 +1,360 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include + + +#include "backend/common/Hashrate.h" +#include "backend/common/interfaces/IWorker.h" +#include "backend/common/Workers.h" +#include "backend/cpu/Cpu.h" +#include "backend/cpu/CpuBackend.h" +#include "base/io/log/Log.h" +#include "base/net/stratum/Job.h" +#include "base/tools/Chrono.h" +#include "base/tools/String.h" +#include "core/config/Config.h" +#include "core/Controller.h" +#include "crypto/common/VirtualMemory.h" +#include "crypto/rx/Rx.h" +#include "crypto/rx/RxDataset.h" +#include "rapidjson/document.h" + + +namespace xmrig { + + +extern template class Threads; + + +static const char *tag = CYAN_BG_BOLD(" cpu "); +static const String kType = "cpu"; + + +struct LaunchStatus +{ +public: + inline void reset() + { + hugePages = 0; + memory = 0; + pages = 0; + started = 0; + threads = 0; + ways = 0; + ts = Chrono::steadyMSecs(); + } + + size_t hugePages = 0; + size_t memory = 0; + size_t pages = 0; + size_t started = 0; + size_t threads = 0; + size_t ways = 0; + uint64_t ts = 0; +}; + + +class CpuBackendPrivate +{ +public: + inline CpuBackendPrivate(Controller *controller) : + controller(controller) + { + } + + + inline void start() + { + LOG_INFO("%s use profile " BLUE_BG(WHITE_BOLD_S " %s ") WHITE_BOLD_S " (" CYAN_BOLD("%zu") WHITE_BOLD(" threads)") " scratchpad " CYAN_BOLD("%zu KB"), + tag, + profileName.data(), + threads.size(), + algo.l3() / 1024 + ); + + workers.stop(); + + status.reset(); + status.memory = algo.l3(); + status.threads = threads.size(); + + for (const CpuLaunchData &data : threads) { + status.ways += static_cast(data.intensity); + } + + workers.start(threads); + } + + + size_t ways() + { + std::lock_guard lock(mutex); + + return status.ways; + } + + + Algorithm algo; + Controller *controller; + LaunchStatus status; + std::mutex mutex; + std::vector threads; + String profileName; + Workers workers; +}; + + +} // namespace xmrig + + +xmrig::CpuBackend::CpuBackend(Controller *controller) : + d_ptr(new CpuBackendPrivate(controller)) +{ + d_ptr->workers.setBackend(this); +} + + +xmrig::CpuBackend::~CpuBackend() +{ + delete d_ptr; +} + + +std::pair xmrig::CpuBackend::hugePages() const +{ + std::pair pages(0, 0); + +# ifdef XMRIG_ALGO_RANDOMX + if (d_ptr->algo.family() == Algorithm::RANDOM_X) { + pages = Rx::hugePages(); + } +# endif + + std::lock_guard lock(d_ptr->mutex); + + pages.first += d_ptr->status.hugePages; + pages.second += d_ptr->status.pages; + + return pages; +} + + +bool xmrig::CpuBackend::isEnabled() const +{ + return d_ptr->controller->config()->cpu().isEnabled(); +} + + +bool xmrig::CpuBackend::isEnabled(const Algorithm &algorithm) const +{ + return !d_ptr->controller->config()->cpu().threads().get(algorithm).isEmpty(); +} + + +const xmrig::Hashrate *xmrig::CpuBackend::hashrate() const +{ + return d_ptr->workers.hashrate(); +} + + +const xmrig::String &xmrig::CpuBackend::profileName() const +{ + return d_ptr->profileName; +} + + +const xmrig::String &xmrig::CpuBackend::type() const +{ + return kType; +} + + +void xmrig::CpuBackend::prepare(const Job &) +{ +} + + +void xmrig::CpuBackend::printHashrate(bool details) +{ + if (!details || !hashrate()) { + return; + } + + char num[8 * 3] = { 0 }; + + Log::print(WHITE_BOLD_S "| CPU THREAD | AFFINITY | 10s H/s | 60s H/s | 15m H/s |"); + + size_t i = 0; + for (const CpuLaunchData &data : d_ptr->threads) { + Log::print("| %13zu | %8" PRId64 " | %7s | %7s | %7s |", + i, + data.affinity, + Hashrate::format(hashrate()->calc(i, Hashrate::ShortInterval), num, sizeof num / 3), + Hashrate::format(hashrate()->calc(i, Hashrate::MediumInterval), num + 8, sizeof num / 3), + Hashrate::format(hashrate()->calc(i, Hashrate::LargeInterval), num + 8 * 2, sizeof num / 3) + ); + + i++; + } +} + + +void xmrig::CpuBackend::setJob(const Job &job) +{ + if (!isEnabled()) { + return stop(); + } + + const CpuConfig &cpu = d_ptr->controller->config()->cpu(); + + std::vector threads = cpu.get(d_ptr->controller->miner(), job.algorithm()); + if (d_ptr->threads.size() == threads.size() && std::equal(d_ptr->threads.begin(), d_ptr->threads.end(), threads.begin())) { + return; + } + + d_ptr->algo = job.algorithm(); + d_ptr->profileName = cpu.threads().profileName(job.algorithm()); + + if (d_ptr->profileName.isNull() || threads.empty()) { + d_ptr->workers.stop(); + + LOG_WARN(YELLOW_BOLD_S "CPU disabled, no suitable configuration for algo %s", job.algorithm().shortName()); + + return; + } + + d_ptr->threads = std::move(threads); + d_ptr->start(); +} + + +void xmrig::CpuBackend::start(IWorker *worker) +{ + d_ptr->mutex.lock(); + + const auto pages = worker->memory()->hugePages(); + + d_ptr->status.started++; + d_ptr->status.hugePages += pages.first; + d_ptr->status.pages += pages.second; + + if (d_ptr->status.started == d_ptr->status.threads) { + const double percent = d_ptr->status.hugePages == 0 ? 0.0 : static_cast(d_ptr->status.hugePages) / d_ptr->status.pages * 100.0; + const size_t memory = d_ptr->status.ways * d_ptr->status.memory / 1024; + + LOG_INFO("%s" GREEN_BOLD(" READY") " threads " CYAN_BOLD("%zu(%zu)") " huge pages %s%zu/%zu %1.0f%%\x1B[0m memory " CYAN_BOLD("%zu KB") BLACK_BOLD(" (%" PRIu64 " ms)"), + tag, + d_ptr->status.threads, d_ptr->status.ways, + (d_ptr->status.hugePages == d_ptr->status.pages ? GREEN_BOLD_S : (d_ptr->status.hugePages == 0 ? RED_BOLD_S : YELLOW_BOLD_S)), + d_ptr->status.hugePages, d_ptr->status.pages, percent, memory, + Chrono::steadyMSecs() - d_ptr->status.ts + ); + } + + d_ptr->mutex.unlock(); + + worker->start(); +} + + +void xmrig::CpuBackend::stop() +{ + const uint64_t ts = Chrono::steadyMSecs(); + + d_ptr->workers.stop(); + d_ptr->threads.clear(); + + LOG_INFO("%s" YELLOW(" stopped") BLACK_BOLD(" (%" PRIu64 " ms)"), tag, Chrono::steadyMSecs() - ts); +} + + +void xmrig::CpuBackend::tick(uint64_t ticks) +{ + d_ptr->workers.tick(ticks); +} + + +#ifdef XMRIG_FEATURE_API +rapidjson::Value xmrig::CpuBackend::toJSON(rapidjson::Document &doc) const +{ + using namespace rapidjson; + auto &allocator = doc.GetAllocator(); + const CpuConfig &cpu = d_ptr->controller->config()->cpu(); + + Value out(kObjectType); + out.AddMember("type", type().toJSON(), allocator); + out.AddMember("enabled", isEnabled(), allocator); + out.AddMember("algo", d_ptr->algo.toJSON(), allocator); + out.AddMember("profile", profileName().toJSON(), allocator); + out.AddMember("hw-aes", cpu.isHwAES(), allocator); + out.AddMember("priority", cpu.priority(), allocator); + +# ifdef XMRIG_FEATURE_ASM + const Assembly assembly = Cpu::assembly(cpu.assembly()); + out.AddMember("asm", assembly.toJSON(), allocator); +# else + out.AddMember("asm", false, allocator); +# endif + + const auto pages = hugePages(); + + rapidjson::Value hugepages(rapidjson::kArrayType); + hugepages.PushBack(pages.first, allocator); + hugepages.PushBack(pages.second, allocator); + + out.AddMember("hugepages", hugepages, allocator); + out.AddMember("memory", static_cast(d_ptr->algo.isValid() ? (d_ptr->ways() * d_ptr->algo.l3()) : 0), allocator); + + if (d_ptr->threads.empty() || !hashrate()) { + return out; + } + + Value threads(kArrayType); + const Hashrate *hr = hashrate(); + + size_t i = 0; + for (const CpuLaunchData &data : d_ptr->threads) { + Value thread(kObjectType); + thread.AddMember("intensity", data.intensity, allocator); + thread.AddMember("affinity", data.affinity, allocator); + thread.AddMember("av", data.av(), allocator); + + Value hashrate(kArrayType); + hashrate.PushBack(Hashrate::normalize(hr->calc(i, Hashrate::ShortInterval)), allocator); + hashrate.PushBack(Hashrate::normalize(hr->calc(i, Hashrate::MediumInterval)), allocator); + hashrate.PushBack(Hashrate::normalize(hr->calc(i, Hashrate::LargeInterval)), allocator); + + i++; + + thread.AddMember("hashrate", hashrate, allocator); + threads.PushBack(thread, allocator); + } + + out.AddMember("threads", threads, allocator); + + return out; +} +#endif diff --git a/src/backend/cpu/CpuBackend.h b/src/backend/cpu/CpuBackend.h new file mode 100644 index 000000000..2b907840f --- /dev/null +++ b/src/backend/cpu/CpuBackend.h @@ -0,0 +1,76 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_CPUBACKEND_H +#define XMRIG_CPUBACKEND_H + + +#include + + +#include "backend/common/interfaces/IBackend.h" + + +namespace xmrig { + + +class Controller; +class CpuBackendPrivate; +class Miner; + + +class CpuBackend : public IBackend +{ +public: + CpuBackend(Controller *controller); + ~CpuBackend() override; + + std::pair hugePages() const; + +protected: + bool isEnabled() const override; + bool isEnabled(const Algorithm &algorithm) const override; + const Hashrate *hashrate() const override; + const String &profileName() const override; + const String &type() const override; + void prepare(const Job &nextJob) override; + void printHashrate(bool details) override; + void setJob(const Job &job) override; + void start(IWorker *worker) override; + void stop() override; + void tick(uint64_t ticks) override; + +# ifdef XMRIG_FEATURE_API + rapidjson::Value toJSON(rapidjson::Document &doc) const override; +# endif + +private: + CpuBackendPrivate *d_ptr; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_CPUBACKEND_H */ diff --git a/src/backend/cpu/CpuConfig.cpp b/src/backend/cpu/CpuConfig.cpp new file mode 100644 index 000000000..5905b7b4b --- /dev/null +++ b/src/backend/cpu/CpuConfig.cpp @@ -0,0 +1,188 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "backend/cpu/Cpu.h" +#include "backend/cpu/CpuConfig.h" +#include "base/io/json/Json.h" +#include "rapidjson/document.h" + + +namespace xmrig { + +static const char *kCn = "cn"; +static const char *kEnabled = "enabled"; +static const char *kHugePages = "huge-pages"; +static const char *kHwAes = "hw-aes"; +static const char *kPriority = "priority"; + +#ifdef XMRIG_FEATURE_ASM +static const char *kAsm = "asm"; +#endif + +#ifdef XMRIG_ALGO_CN_GPU +static const char *kCnGPU = "cn/gpu"; +#endif + +#ifdef XMRIG_ALGO_CN_LITE +static const char *kCnLite = "cn-lite"; +#endif + +#ifdef XMRIG_ALGO_CN_HEAVY +static const char *kCnHeavy = "cn-heavy"; +#endif + +#ifdef XMRIG_ALGO_CN_PICO +static const char *kCnPico = "cn-pico"; +#endif + +#ifdef XMRIG_ALGO_RANDOMX +static const char *kRx = "rx"; +static const char *kRxWOW = "rx/wow"; +#endif + +extern template class Threads; + +} + + +xmrig::CpuConfig::CpuConfig() +{ +} + + +bool xmrig::CpuConfig::isHwAES() const +{ + return (m_aes == AES_AUTO ? (Cpu::info()->hasAES() ? AES_HW : AES_SOFT) : m_aes) == AES_HW; +} + + +rapidjson::Value xmrig::CpuConfig::toJSON(rapidjson::Document &doc) const +{ + using namespace rapidjson; + auto &allocator = doc.GetAllocator(); + + Value obj(kObjectType); + + obj.AddMember(StringRef(kEnabled), m_enabled, allocator); + obj.AddMember(StringRef(kHugePages), m_hugePages, allocator); + obj.AddMember(StringRef(kHwAes), m_aes == AES_AUTO ? Value(kNullType) : Value(m_aes == AES_HW), allocator); + obj.AddMember(StringRef(kPriority), priority() != -1 ? Value(priority()) : Value(kNullType), allocator); + +# ifdef XMRIG_FEATURE_ASM + obj.AddMember(StringRef(kAsm), m_assembly.toJSON(), allocator); +# endif + + m_threads.toJSON(obj, doc); + + return obj; +} + + +std::vector xmrig::CpuConfig::get(const Miner *miner, const Algorithm &algorithm) const +{ + std::vector out; + const CpuThreads &threads = m_threads.get(algorithm); + + if (threads.isEmpty()) { + return out; + } + + out.reserve(threads.count()); + + for (const CpuThread &thread : threads.data()) { + out.push_back(CpuLaunchData(miner, algorithm, *this, thread)); + } + + return out; +} + + +void xmrig::CpuConfig::read(const rapidjson::Value &value) +{ + if (value.IsObject()) { + m_enabled = Json::getBool(value, kEnabled, m_enabled); + m_hugePages = Json::getBool(value, kHugePages, m_hugePages); + + setAesMode(Json::getValue(value, kHwAes)); + setPriority(Json::getInt(value, kPriority, -1)); + +# ifdef XMRIG_FEATURE_ASM + m_assembly = Json::getValue(value, kAsm); +# endif + + if (!m_threads.read(value)) { + generate(); + } + } + else if (value.IsBool() && value.IsFalse()) { + m_enabled = false; + } + else { + generate(); + } +} + + +void xmrig::CpuConfig::generate() +{ + m_shouldSave = true; + ICpuInfo *cpu = Cpu::info(); + + m_threads.disable(Algorithm::CN_0); + m_threads.move(kCn, cpu->threads(Algorithm::CN_0)); + +# ifdef XMRIG_ALGO_CN_GPU + m_threads.move(kCnGPU, cpu->threads(Algorithm::CN_GPU)); +# endif + +# ifdef XMRIG_ALGO_CN_LITE + m_threads.disable(Algorithm::CN_LITE_0); + m_threads.move(kCnLite, cpu->threads(Algorithm::CN_LITE_1)); +# endif + +# ifdef XMRIG_ALGO_CN_HEAVY + m_threads.move(kCnHeavy, cpu->threads(Algorithm::CN_HEAVY_0)); +# endif + +# ifdef XMRIG_ALGO_CN_PICO + m_threads.move(kCnPico, cpu->threads(Algorithm::CN_PICO_0)); +# endif + +# ifdef XMRIG_ALGO_RANDOMX + m_threads.move(kRx, cpu->threads(Algorithm::RX_0)); + m_threads.move(kRxWOW, cpu->threads(Algorithm::RX_WOW)); +# endif +} + + +void xmrig::CpuConfig::setAesMode(const rapidjson::Value &aesMode) +{ + if (aesMode.IsBool()) { + m_aes = aesMode.GetBool() ? AES_HW : AES_SOFT; + } + else { + m_aes = AES_AUTO; + } +} diff --git a/src/backend/cpu/CpuConfig.h b/src/backend/cpu/CpuConfig.h new file mode 100644 index 000000000..5aca51884 --- /dev/null +++ b/src/backend/cpu/CpuConfig.h @@ -0,0 +1,80 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_CPUCONFIG_H +#define XMRIG_CPUCONFIG_H + + +#include "backend/common/Threads.h" +#include "backend/cpu/CpuLaunchData.h" +#include "backend/cpu/CpuThreads.h" +#include "crypto/common/Assembly.h" + + +namespace xmrig { + + +class CpuConfig +{ +public: + enum AesMode { + AES_AUTO, + AES_HW, + AES_SOFT + }; + + CpuConfig(); + + bool isHwAES() const; + rapidjson::Value toJSON(rapidjson::Document &doc) const; + std::vector get(const Miner *miner, const Algorithm &algorithm) const; + void read(const rapidjson::Value &value); + + inline bool isEnabled() const { return m_enabled; } + inline bool isHugePages() const { return m_hugePages; } + inline bool isShouldSave() const { return m_shouldSave; } + inline const Assembly &assembly() const { return m_assembly; } + inline const Threads &threads() const { return m_threads; } + inline int priority() const { return m_priority; } + +private: + void generate(); + void setAesMode(const rapidjson::Value &aesMode); + + inline void setPriority(int priority) { m_priority = (priority >= -1 && priority <= 5) ? priority : -1; } + + AesMode m_aes = AES_AUTO; + Assembly m_assembly; + bool m_enabled = true; + bool m_hugePages = true; + bool m_shouldSave = false; + int m_priority = -1; + Threads m_threads; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_CPUCONFIG_H */ diff --git a/src/backend/cpu/CpuLaunchData.cpp b/src/backend/cpu/CpuLaunchData.cpp new file mode 100644 index 000000000..a01f22a6a --- /dev/null +++ b/src/backend/cpu/CpuLaunchData.cpp @@ -0,0 +1,64 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "backend/cpu/CpuLaunchData.h" +#include "backend/cpu/CpuConfig.h" + + +xmrig::CpuLaunchData::CpuLaunchData(const Miner *miner, const Algorithm &algorithm, const CpuConfig &config, const CpuThread &thread) : + algorithm(algorithm), + assembly(config.assembly()), + hugePages(config.isHugePages()), + hwAES(config.isHwAES()), + intensity(thread.intensity()), + priority(config.priority()), + affinity(thread.affinity()), + miner(miner) +{ +} + + +bool xmrig::CpuLaunchData::isEqual(const CpuLaunchData &other) const +{ + return (algorithm.l3() == other.algorithm.l3() + && assembly == other.assembly + && hugePages == other.hugePages + && hwAES == other.hwAES + && intensity == other.intensity + && priority == other.priority + && affinity == other.affinity + ); +} + + +xmrig::CnHash::AlgoVariant xmrig::CpuLaunchData::av() const +{ + if (intensity <= 2) { + return static_cast(!hwAES ? (intensity + 2) : intensity); + } + + return static_cast(!hwAES ? (intensity + 5) : (intensity + 2)); +} diff --git a/src/workers/MultiWorker.h b/src/backend/cpu/CpuLaunchData.h similarity index 56% rename from src/workers/MultiWorker.h rename to src/backend/cpu/CpuLaunchData.h index b7e4c8cae..bb18816a5 100644 --- a/src/workers/MultiWorker.h +++ b/src/backend/cpu/CpuLaunchData.h @@ -23,54 +23,49 @@ * along with this program. If not, see . */ -#ifndef XMRIG_MULTIWORKER_H -#define XMRIG_MULTIWORKER_H +#ifndef XMRIG_CPULAUNCHDATA_H +#define XMRIG_CPULAUNCHDATA_H -#include "common/net/Job.h" -#include "Mem.h" -#include "net/JobResult.h" -#include "workers/Worker.h" +#include "crypto/cn/CnHash.h" +#include "crypto/common/Algorithm.h" +#include "crypto/common/Assembly.h" +#include "crypto/common/Nonce.h" -class Handle; +namespace xmrig { -template -class MultiWorker : public Worker +class CpuConfig; +class CpuThread; +class Miner; + + +class CpuLaunchData { public: - MultiWorker(Handle *handle); - ~MultiWorker(); + CpuLaunchData(const Miner *miner, const Algorithm &algorithm, const CpuConfig &config, const CpuThread &thread); -protected: - bool selfTest() override; - void start() override; + bool isEqual(const CpuLaunchData &other) const; + CnHash::AlgoVariant av() const; -private: - bool resume(const xmrig::Job &job); - bool verify(xmrig::Variant variant, const uint8_t *referenceValue); - bool verify2(xmrig::Variant variant, const uint8_t *referenceValue); - void consumeJob(); - void save(const xmrig::Job &job); + inline constexpr static Nonce::Backend backend() { return Nonce::CPU; } - inline uint32_t *nonce(size_t index) - { - return reinterpret_cast(m_state.blob + (index * m_state.job.size()) + 39); - } + inline bool operator!=(const CpuLaunchData &other) const { return !isEqual(other); } + inline bool operator==(const CpuLaunchData &other) const { return isEqual(other); } - struct State - { - alignas(16) uint8_t blob[xmrig::Job::kMaxBlobSize * N]; - xmrig::Job job; - }; - - - cryptonight_ctx *m_ctx[N]; - State m_pausedState; - State m_state; - uint8_t m_hash[N * 32]; + const Algorithm algorithm; + const Assembly assembly; + const bool hugePages; + const bool hwAES; + const int intensity; + const int priority; + const int64_t affinity; + const Miner *miner; }; -#endif /* XMRIG_MULTIWORKER_H */ +} // namespace xmrig + + +#endif /* XMRIG_CPULAUNCHDATA_H */ diff --git a/src/backend/cpu/CpuThread.cpp b/src/backend/cpu/CpuThread.cpp new file mode 100644 index 000000000..7d7a9e855 --- /dev/null +++ b/src/backend/cpu/CpuThread.cpp @@ -0,0 +1,58 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "backend/cpu/CpuThread.h" +#include "base/io/json/Json.h" +#include "rapidjson/document.h" + + +xmrig::CpuThread::CpuThread(const rapidjson::Value &value) +{ + if (value.IsArray() && value.Size() >= 2) { + m_intensity = value[0].GetInt(); + m_affinity = value[1].GetInt(); + } + else if (value.IsInt()) { + m_intensity = -1; + m_affinity = value.GetInt(); + } +} + + +rapidjson::Value xmrig::CpuThread::toJSON(rapidjson::Document &doc) const +{ + using namespace rapidjson; + if (m_intensity == -1) { + return Value(m_affinity); + } + + auto &allocator = doc.GetAllocator(); + + Value out(kArrayType); + out.PushBack(m_intensity, allocator); + out.PushBack(m_affinity, allocator); + + return out; +} diff --git a/src/common/config/ConfigLoader.h b/src/backend/cpu/CpuThread.h similarity index 56% rename from src/common/config/ConfigLoader.h rename to src/backend/cpu/CpuThread.h index 99d6143f6..691ee1146 100644 --- a/src/common/config/ConfigLoader.h +++ b/src/backend/cpu/CpuThread.h @@ -22,51 +22,41 @@ * along with this program. If not, see . */ -#ifndef XMRIG_CONFIGLOADER_H -#define XMRIG_CONFIGLOADER_H - - -#include +#ifndef XMRIG_CPUTHREAD_H +#define XMRIG_CPUTHREAD_H #include "rapidjson/fwd.h" -struct option; - - namespace xmrig { -class ConfigWatcher; -class IConfigCreator; -class IConfigListener; -class IConfig; -class Process; - - -class ConfigLoader +class CpuThread { public: - static bool loadFromFile(IConfig *config, const char *fileName); - static bool loadFromJSON(IConfig *config, const char *json); - static bool loadFromJSON(IConfig *config, const rapidjson::Document &doc); - static bool reload(IConfig *oldConfig, const char *json); - static bool watch(IConfig *config); - static IConfig *load(Process *process, IConfigCreator *creator, IConfigListener *listener); - static void release(); + inline constexpr CpuThread() {} + inline constexpr CpuThread(int64_t affinity, int intensity) : m_intensity(intensity), m_affinity(affinity) {} + + CpuThread(const rapidjson::Value &value); + + inline bool isEqual(const CpuThread &other) const { return other.m_affinity == m_affinity && other.m_intensity == m_intensity; } + inline bool isValid() const { return m_intensity == -1 || (m_intensity >= 1 && m_intensity <= 5); } + inline int intensity() const { return m_intensity == -1 ? 1 : m_intensity; } + inline int64_t affinity() const { return m_affinity; } + + inline bool operator!=(const CpuThread &other) const { return !isEqual(other); } + inline bool operator==(const CpuThread &other) const { return isEqual(other); } + + rapidjson::Value toJSON(rapidjson::Document &doc) const; private: - static bool getJSON(const char *fileName, rapidjson::Document &doc); - static bool parseArg(IConfig *config, int key, const char *arg); - static void parseJSON(IConfig *config, const struct option *option, const rapidjson::Value &object); - - static ConfigWatcher *m_watcher; - static IConfigCreator *m_creator; - static IConfigListener *m_listener; + int m_intensity = -1; + int64_t m_affinity = -1; }; } /* namespace xmrig */ -#endif /* XMRIG_CONFIGLOADER_H */ + +#endif /* XMRIG_CPUTHREAD_H */ diff --git a/src/backend/cpu/CpuThreads.cpp b/src/backend/cpu/CpuThreads.cpp new file mode 100644 index 000000000..07e8ca33f --- /dev/null +++ b/src/backend/cpu/CpuThreads.cpp @@ -0,0 +1,146 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include + + +#include "backend/cpu/CpuThreads.h" +#include "base/io/json/Json.h" +#include "rapidjson/document.h" + + +namespace xmrig { + + +static const char *kAffinity = "affinity"; +static const char *kIntensity = "intensity"; +static const char *kThreads = "threads"; + + +static inline int64_t getAffinityMask(const rapidjson::Value &value) +{ + if (value.IsInt64()) { + return value.GetInt64(); + } + + if (value.IsString()) { + const char *arg = value.GetString(); + const char *p = strstr(arg, "0x"); + + return p ? strtoll(p, nullptr, 16) : strtoll(arg, nullptr, 10); + } + + return -1L; +} + + +static inline int64_t getAffinity(uint64_t index, int64_t affinity) +{ + if (affinity == -1L) { + return -1L; + } + + size_t idx = 0; + + for (size_t i = 0; i < 64; i++) { + if (!(static_cast(affinity) & (1ULL << i))) { + continue; + } + + if (idx == index) { + return static_cast(i); + } + + idx++; + } + + return -1L; +} + + +} + + +xmrig::CpuThreads::CpuThreads(const rapidjson::Value &value) +{ + if (value.IsArray()) { + for (auto &v : value.GetArray()) { + CpuThread thread(v); + if (thread.isValid()) { + add(std::move(thread)); + } + } + } + else if (value.IsObject()) { + int intensity = Json::getInt(value, kIntensity, 1); + const size_t threads = std::min(Json::getUint(value, kThreads), 1024); + m_affinity = getAffinityMask(Json::getValue(value, kAffinity)); + m_format = ObjectFormat; + + if (intensity < 1 || intensity > 5) { + intensity = 1; + } + + for (size_t i = 0; i < threads; ++i) { + add(getAffinity(i, m_affinity), intensity); + } + } +} + + +xmrig::CpuThreads::CpuThreads(size_t count, int intensity) +{ + m_data.reserve(count); + + for (size_t i = 0; i < count; ++i) { + add(-1, intensity); + } +} + + +rapidjson::Value xmrig::CpuThreads::toJSON(rapidjson::Document &doc) const +{ + using namespace rapidjson; + auto &allocator = doc.GetAllocator(); + + Value out; + + if (m_format == ArrayFormat) { + out.SetArray(); + + for (const CpuThread &thread : m_data) { + out.PushBack(thread.toJSON(doc), allocator); + } + } + else { + out.SetObject(); + + out.AddMember(StringRef(kIntensity), m_data.empty() ? 1 : m_data.front().intensity(), allocator); + out.AddMember(StringRef(kThreads), static_cast(m_data.size()), allocator); + out.AddMember(StringRef(kAffinity), m_affinity, allocator); + } + + return out; +} diff --git a/src/backend/cpu/CpuThreads.h b/src/backend/cpu/CpuThreads.h new file mode 100644 index 000000000..f8ad64308 --- /dev/null +++ b/src/backend/cpu/CpuThreads.h @@ -0,0 +1,71 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_CPUTHREADS_H +#define XMRIG_CPUTHREADS_H + + +#include + + +#include "backend/cpu/CpuThread.h" + + +namespace xmrig { + + +class CpuThreads +{ +public: + inline CpuThreads() {} + inline CpuThreads(size_t count) : m_data(count) {} + + CpuThreads(const rapidjson::Value &value); + CpuThreads(size_t count, int intensity); + + inline bool isEmpty() const { return m_data.empty(); } + inline const std::vector &data() const { return m_data; } + inline size_t count() const { return m_data.size(); } + inline void add(CpuThread &&thread) { m_data.push_back(thread); } + inline void add(int64_t affinity, int intensity) { add(CpuThread(affinity, intensity)); } + inline void reserve(size_t capacity) { m_data.reserve(capacity); } + + rapidjson::Value toJSON(rapidjson::Document &doc) const; + +private: + enum Format { + ArrayFormat, + ObjectFormat + }; + + Format m_format = ArrayFormat; + int64_t m_affinity = -1; + std::vector m_data; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_CPUTHREADS_H */ diff --git a/src/backend/cpu/CpuWorker.cpp b/src/backend/cpu/CpuWorker.cpp new file mode 100644 index 000000000..14ef1797e --- /dev/null +++ b/src/backend/cpu/CpuWorker.cpp @@ -0,0 +1,317 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include +#include + + +#include "backend/cpu/CpuWorker.h" +#include "core/Miner.h" +#include "crypto/cn/CnCtx.h" +#include "crypto/cn/CryptoNight_test.h" +#include "crypto/common/Nonce.h" +#include "crypto/common/VirtualMemory.h" +#include "crypto/rx/Rx.h" +#include "crypto/rx/RxVm.h" +#include "net/JobResults.h" + + +#ifdef XMRIG_ALGO_RANDOMX +# include "crypto/randomx/randomx.h" +#endif + + +namespace xmrig { + +static constexpr uint32_t kReserveCount = 4096; + +} // namespace xmrig + + + +template +xmrig::CpuWorker::CpuWorker(size_t index, const CpuLaunchData &data) : + Worker(index, data.affinity, data.priority), + m_algorithm(data.algorithm), + m_assembly(data.assembly), + m_hwAES(data.hwAES), + m_av(data.av()), + m_miner(data.miner), + m_ctx() +{ + m_memory = new VirtualMemory(m_algorithm.l3() * N, data.hugePages); +} + + +template +xmrig::CpuWorker::~CpuWorker() +{ + CnCtx::release(m_ctx, N); + delete m_memory; + +# ifdef XMRIG_ALGO_RANDOMX + delete m_vm; +# endif +} + + +#ifdef XMRIG_ALGO_RANDOMX +template +void xmrig::CpuWorker::allocateRandomX_VM() +{ + RxDataset *dataset = Rx::dataset(m_job.currentJob(), m_node); + + while (dataset == nullptr) { + std::this_thread::sleep_for(std::chrono::milliseconds(200)); + + if (Nonce::sequence(Nonce::CPU) == 0) { + return; + } + } + + if (!m_vm) { + m_vm = new RxVm(dataset, m_memory->scratchpad(), !m_hwAES); + } +} +#endif + + +template +bool xmrig::CpuWorker::selfTest() +{ +# ifdef XMRIG_ALGO_RANDOMX + if (m_algorithm.family() == Algorithm::RANDOM_X) { + return N == 1; + } +# endif + + allocateCnCtx(); + + if (m_algorithm.family() == Algorithm::CN) { + const bool rc = verify(Algorithm::CN_0, test_output_v0) && + verify(Algorithm::CN_1, test_output_v1) && + verify(Algorithm::CN_2, test_output_v2) && + verify(Algorithm::CN_FAST, test_output_msr) && + verify(Algorithm::CN_XAO, test_output_xao) && + verify(Algorithm::CN_RTO, test_output_rto) && + verify(Algorithm::CN_HALF, test_output_half) && + verify2(Algorithm::CN_WOW, test_output_wow) && + verify2(Algorithm::CN_R, test_output_r) && + verify(Algorithm::CN_RWZ, test_output_rwz) && + verify(Algorithm::CN_ZLS, test_output_zls) && + verify(Algorithm::CN_DOUBLE, test_output_double); + +# ifdef XMRIG_ALGO_CN_GPU + if (!rc || N > 1) { + return rc; + } + + return verify(Algorithm::CN_GPU, test_output_gpu); +# else + return rc; +# endif + } + +# ifdef XMRIG_ALGO_CN_LITE + if (m_algorithm.family() == Algorithm::CN_LITE) { + return verify(Algorithm::CN_LITE_0, test_output_v0_lite) && + verify(Algorithm::CN_LITE_1, test_output_v1_lite); + } +# endif + +# ifdef XMRIG_ALGO_CN_HEAVY + if (m_algorithm.family() == Algorithm::CN_HEAVY) { + return verify(Algorithm::CN_HEAVY_0, test_output_v0_heavy) && + verify(Algorithm::CN_HEAVY_XHV, test_output_xhv_heavy) && + verify(Algorithm::CN_HEAVY_TUBE, test_output_tube_heavy); + } +# endif + +# ifdef XMRIG_ALGO_CN_PICO + if (m_algorithm.family() == Algorithm::CN_PICO) { + return verify(Algorithm::CN_PICO_0, test_output_pico_trtl); + } +# endif + + return false; +} + + +template +void xmrig::CpuWorker::start() +{ + while (Nonce::sequence(Nonce::CPU) > 0) { + if (Nonce::isPaused()) { + do { + std::this_thread::sleep_for(std::chrono::milliseconds(200)); + } + while (Nonce::isPaused() && Nonce::sequence(Nonce::CPU) > 0); + + if (Nonce::sequence(Nonce::CPU) == 0) { + break; + } + + consumeJob(); + } + + while (!Nonce::isOutdated(Nonce::CPU, m_job.sequence())) { + if ((m_count & 0x7) == 0) { + storeStats(); + } + + const Job &job = m_job.currentJob(); + + if (job.algorithm().l3() != m_algorithm.l3()) { + break; + } + +# ifdef XMRIG_ALGO_RANDOMX + if (job.algorithm().family() == Algorithm::RANDOM_X) { + randomx_calculate_hash(m_vm->get(), m_job.blob(), job.size(), m_hash); + } + else +# endif + { + fn(job.algorithm())(m_job.blob(), job.size(), m_hash, m_ctx, job.height()); + } + + for (size_t i = 0; i < N; ++i) { + if (*reinterpret_cast(m_hash + (i * 32) + 24) < job.target()) { + JobResults::submit(JobResult(job, *m_job.nonce(i), m_hash + (i * 32))); + } + } + + m_job.nextRound(kReserveCount); + m_count += N; + + std::this_thread::yield(); + } + + consumeJob(); + } +} + + +template +bool xmrig::CpuWorker::verify(const Algorithm &algorithm, const uint8_t *referenceValue) +{ + cn_hash_fun func = fn(algorithm); + if (!func) { + return false; + } + + func(test_input, 76, m_hash, m_ctx, 0); + return memcmp(m_hash, referenceValue, sizeof m_hash) == 0; +} + + +template +bool xmrig::CpuWorker::verify2(const Algorithm &algorithm, const uint8_t *referenceValue) +{ + cn_hash_fun func = fn(algorithm); + if (!func) { + return false; + } + + for (size_t i = 0; i < (sizeof(cn_r_test_input) / sizeof(cn_r_test_input[0])); ++i) { + const size_t size = cn_r_test_input[i].size; + for (size_t k = 0; k < N; ++k) { + memcpy(m_job.blob() + (k * size), cn_r_test_input[i].data, size); + } + + func(m_job.blob(), size, m_hash, m_ctx, cn_r_test_input[i].height); + + for (size_t k = 0; k < N; ++k) { + if (memcmp(m_hash + k * 32, referenceValue + i * 32, sizeof m_hash / N) != 0) { + return false; + } + } + } + + return true; +} + + +namespace xmrig { + +template<> +bool CpuWorker<1>::verify2(const Algorithm &algorithm, const uint8_t *referenceValue) +{ + cn_hash_fun func = fn(algorithm); + if (!func) { + return false; + } + + for (size_t i = 0; i < (sizeof(cn_r_test_input) / sizeof(cn_r_test_input[0])); ++i) { + func(cn_r_test_input[i].data, cn_r_test_input[i].size, m_hash, m_ctx, cn_r_test_input[i].height); + + if (memcmp(m_hash, referenceValue + i * 32, sizeof m_hash) != 0) { + return false; + } + } + + return true; +} + +} // namespace xmrig + + +template +void xmrig::CpuWorker::allocateCnCtx() +{ + if (m_ctx[0] == nullptr) { + CnCtx::create(m_ctx, m_memory->scratchpad(), m_algorithm.l3(), N); + } +} + + +template +void xmrig::CpuWorker::consumeJob() +{ + m_job.add(m_miner->job(), Nonce::sequence(Nonce::CPU), kReserveCount); + +# ifdef XMRIG_ALGO_RANDOMX + if (m_job.currentJob().algorithm().family() == Algorithm::RANDOM_X) { + allocateRandomX_VM(); + } + else +# endif + { + allocateCnCtx(); + } +} + + +namespace xmrig { + +template class CpuWorker<1>; +template class CpuWorker<2>; +template class CpuWorker<3>; +template class CpuWorker<4>; +template class CpuWorker<5>; + +} // namespace xmrig + diff --git a/src/backend/cpu/CpuWorker.h b/src/backend/cpu/CpuWorker.h new file mode 100644 index 000000000..4cdd10f8f --- /dev/null +++ b/src/backend/cpu/CpuWorker.h @@ -0,0 +1,98 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_CPUWORKER_H +#define XMRIG_CPUWORKER_H + + +#include "backend/common/Worker.h" +#include "backend/common/WorkerJob.h" +#include "backend/cpu/CpuLaunchData.h" +#include "base/net/stratum/Job.h" +#include "net/JobResult.h" + + +namespace xmrig { + + +class RxVm; + + +template +class CpuWorker : public Worker +{ +public: + CpuWorker(size_t index, const CpuLaunchData &data); + ~CpuWorker() override; + +protected: + bool selfTest() override; + void start() override; + + inline const VirtualMemory *memory() const override { return m_memory; } + +private: + inline cn_hash_fun fn(const Algorithm &algorithm) const { return CnHash::fn(algorithm, m_av, m_assembly); } + +# ifdef XMRIG_ALGO_RANDOMX + void allocateRandomX_VM(); +# endif + + bool verify(const Algorithm &algorithm, const uint8_t *referenceValue); + bool verify2(const Algorithm &algorithm, const uint8_t *referenceValue); + void allocateCnCtx(); + void consumeJob(); + + const Algorithm m_algorithm; + const Assembly m_assembly; + const bool m_hwAES; + const CnHash::AlgoVariant m_av; + const Miner *m_miner; + cryptonight_ctx *m_ctx[N]; + uint8_t m_hash[N * 32]; + VirtualMemory *m_memory = nullptr; + WorkerJob m_job; + +# ifdef XMRIG_ALGO_RANDOMX + RxVm *m_vm = nullptr; +# endif +}; + + +template<> +bool CpuWorker<1>::verify2(const Algorithm &algorithm, const uint8_t *referenceValue); + + +extern template class CpuWorker<1>; +extern template class CpuWorker<2>; +extern template class CpuWorker<3>; +extern template class CpuWorker<4>; +extern template class CpuWorker<5>; + + +} // namespace xmrig + + +#endif /* XMRIG_CPUWORKER_H */ diff --git a/src/backend/cpu/cpu.cmake b/src/backend/cpu/cpu.cmake new file mode 100644 index 000000000..b6c8915b5 --- /dev/null +++ b/src/backend/cpu/cpu.cmake @@ -0,0 +1,79 @@ +set(HEADERS_BACKEND_CPU + src/backend/cpu/Cpu.h + src/backend/cpu/CpuBackend.h + src/backend/cpu/CpuConfig.h + src/backend/cpu/CpuLaunchData.cpp + src/backend/cpu/CpuThread.h + src/backend/cpu/CpuThreads.h + src/backend/cpu/CpuWorker.h + src/backend/cpu/interfaces/ICpuInfo.h + ) + +set(SOURCES_BACKEND_CPU + src/backend/cpu/Cpu.cpp + src/backend/cpu/CpuBackend.cpp + src/backend/cpu/CpuConfig.cpp + src/backend/cpu/CpuLaunchData.h + src/backend/cpu/CpuThread.cpp + src/backend/cpu/CpuThreads.cpp + src/backend/cpu/CpuWorker.cpp + ) + + +if (WITH_HWLOC) + if (CMAKE_CXX_COMPILER_ID MATCHES MSVC) + add_subdirectory(src/3rdparty/hwloc) + include_directories(src/3rdparty/hwloc/include) + set(CPUID_LIB hwloc) + else() + find_package(HWLOC REQUIRED) + include_directories(${HWLOC_INCLUDE_DIR}) + set(CPUID_LIB ${HWLOC_LIBRARY}) + endif() + + set(WITH_LIBCPUID OFF) + + remove_definitions(/DXMRIG_FEATURE_LIBCPUID) + add_definitions(/DXMRIG_FEATURE_HWLOC) + + if (HWLOC_DEBUG) + add_definitions(/DXMRIG_HWLOC_DEBUG) + endif() + + set(SOURCES_CPUID + src/backend/cpu/platform/BasicCpuInfo.h + src/backend/cpu/platform/HwlocCpuInfo.cpp + src/backend/cpu/platform/HwlocCpuInfo.h + ) +elseif (WITH_LIBCPUID) + set(WITH_HWLOC OFF) + + add_subdirectory(src/3rdparty/libcpuid) + include_directories(src/3rdparty/libcpuid) + + add_definitions(/DXMRIG_FEATURE_LIBCPUID) + remove_definitions(/DXMRIG_FEATURE_HWLOC) + + set(CPUID_LIB cpuid) + set(SOURCES_CPUID + src/backend/cpu/platform/AdvancedCpuInfo.cpp + src/backend/cpu/platform/AdvancedCpuInfo.h + ) +else() + remove_definitions(/DXMRIG_FEATURE_LIBCPUID) + remove_definitions(/DXMRIG_FEATURE_HWLOC) + + set(CPUID_LIB "") + set(SOURCES_CPUID + src/backend/cpu/platform/BasicCpuInfo.h + ) +endif() + + +if (NOT WITH_LIBCPUID) + if (XMRIG_ARM) + set(SOURCES_CPUID ${SOURCES_CPUID} src/backend/cpu/platform/BasicCpuInfo_arm.cpp) + else() + set(SOURCES_CPUID ${SOURCES_CPUID} src/backend/cpu/platform/BasicCpuInfo.cpp) + endif() +endif() diff --git a/src/common/interfaces/ICpuInfo.h b/src/backend/cpu/interfaces/ICpuInfo.h similarity index 65% rename from src/common/interfaces/ICpuInfo.h rename to src/backend/cpu/interfaces/ICpuInfo.h index dd4034b33..9bc3b11aa 100644 --- a/src/common/interfaces/ICpuInfo.h +++ b/src/backend/cpu/interfaces/ICpuInfo.h @@ -26,11 +26,9 @@ #define XMRIG_CPUINFO_H -#include -#include - - -#include "common/xmrig.h" +#include "backend/cpu/CpuThreads.h" +#include "crypto/common/Assembly.h" +#include "crypto/common/Algorithm.h" namespace xmrig { @@ -39,21 +37,26 @@ namespace xmrig { class ICpuInfo { public: - virtual ~ICpuInfo() {} + virtual ~ICpuInfo() = default; +# if defined(__x86_64__) || defined(_M_AMD64) || defined (__arm64__) || defined (__aarch64__) + inline constexpr static bool isX64() { return true; } +# else + inline constexpr static bool isX64() { return false; } +# endif + + virtual Assembly::Id assembly() const = 0; virtual bool hasAES() const = 0; virtual bool hasAVX2() const = 0; - virtual bool isSupported() const = 0; - virtual bool isX64() const = 0; + virtual const char *backend() const = 0; virtual const char *brand() const = 0; - virtual int32_t cores() const = 0; - virtual int32_t L2() const = 0; - virtual int32_t L3() const = 0; - virtual int32_t nodes() const = 0; - virtual int32_t sockets() const = 0; - virtual int32_t threads() const = 0; - virtual size_t optimalThreadsCount(size_t memSize, int maxCpuUsage) const = 0; - virtual xmrig::Assembly assembly() const = 0; + virtual CpuThreads threads(const Algorithm &algorithm) const = 0; + virtual size_t cores() const = 0; + virtual size_t L2() const = 0; + virtual size_t L3() const = 0; + virtual size_t nodes() const = 0; + virtual size_t packages() const = 0; + virtual size_t threads() const = 0; }; diff --git a/src/core/cpu/AdvancedCpuInfo.cpp b/src/backend/cpu/platform/AdvancedCpuInfo.cpp similarity index 52% rename from src/core/cpu/AdvancedCpuInfo.cpp rename to src/backend/cpu/platform/AdvancedCpuInfo.cpp index df6a385ee..de8ff2721 100644 --- a/src/core/cpu/AdvancedCpuInfo.cpp +++ b/src/backend/cpu/platform/AdvancedCpuInfo.cpp @@ -22,66 +22,86 @@ * along with this program. If not, see . */ -#include +#include +#include #include +#include #include -#include "core/cpu/AdvancedCpuInfo.h" +#include "3rdparty/libcpuid/libcpuid.h" +#include "backend/cpu/platform/AdvancedCpuInfo.h" + + +namespace xmrig { + + +static inline void cpu_brand_string(char out[64], const char *in) { + size_t pos = 0; + const size_t size = strlen(in); + + for (size_t i = 0; i < size; ++i) { + if (in[i] == ' ' && ((pos > 0 && out[pos - 1] == ' ') || pos == 0)) { + continue; + } + + out[pos++] = in[i]; + } + + if (pos > 0 && out[pos - 1] == ' ') { + out[pos - 1] = '\0'; + } +} + + +} // namespace xmrig xmrig::AdvancedCpuInfo::AdvancedCpuInfo() : - m_assembly(ASM_NONE), - m_aes(false), - m_avx2(false), - m_L2_exclusive(false), - m_brand(), - m_cores(0), - m_L2(0), - m_L3(0), - m_sockets(1), - m_threads(0) + m_brand() { - struct cpu_raw_data_t raw = { 0 }; - struct cpu_id_t data = { 0 }; + struct cpu_raw_data_t raw = {}; + struct cpu_id_t data = {}; cpuid_get_raw_data(&raw); cpu_identify(&raw, &data); - strncpy(m_brand, data.brand_str, sizeof(m_brand)); + cpu_brand_string(m_brand, data.brand_str); + snprintf(m_backend, sizeof m_backend, "libcpuid/%s", cpuid_lib_version()); - m_threads = data.total_logical_cpus; - m_sockets = threads() / data.num_logical_cpus; - if (m_sockets == 0) { - m_sockets = 1; - } + m_threads = static_cast(data.total_logical_cpus); + m_packages = std::max(threads() / static_cast(data.num_logical_cpus), 1); + m_cores = static_cast(data.num_cores) * m_packages; + m_L3 = data.l3_cache > 0 ? static_cast(data.l3_cache) * m_packages : 0; - m_cores = data.num_cores * m_sockets; - m_L3 = data.l3_cache > 0 ? data.l3_cache * m_sockets : 0; + const size_t l2 = static_cast(data.l2_cache); // Workaround for AMD CPUs https://github.com/anrieff/libcpuid/issues/97 if (data.vendor == VENDOR_AMD && data.ext_family >= 0x15 && data.ext_family < 0x17) { - m_L2 = data.l2_cache * (cores() / 2) * m_sockets; + m_L2 = l2 * (cores() / 2) * m_packages; m_L2_exclusive = true; } // Workaround for Intel Pentium Dual-Core, Core Duo, Core 2 Duo, Core 2 Quad and their Xeon homologue // These processors have L2 cache shared by 2 cores. else if (data.vendor == VENDOR_INTEL && data.ext_family == 0x06 && (data.ext_model == 0x0E || data.ext_model == 0x0F || data.ext_model == 0x17)) { - int l2_count_per_socket = cores() > 1 ? cores() / 2 : 1; - m_L2 = data.l2_cache > 0 ? data.l2_cache * l2_count_per_socket * m_sockets : 0; + size_t l2_count_per_socket = cores() > 1 ? cores() / 2 : 1; + m_L2 = data.l2_cache > 0 ? l2 * l2_count_per_socket * m_packages : 0; } else{ - m_L2 = data.l2_cache > 0 ? data.l2_cache * cores() * m_sockets : 0; + m_L2 = data.l2_cache > 0 ? l2 * cores() * m_packages : 0; } + m_L2 *= 1024; + m_L3 *= 1024; + if (data.flags[CPU_FEATURE_AES]) { m_aes = true; if (data.vendor == VENDOR_AMD) { - m_assembly = (data.ext_family >= 23) ? ASM_RYZEN : ASM_BULLDOZER; + m_assembly = (data.ext_family >= 23) ? Assembly::RYZEN : Assembly::BULLDOZER; } else if (data.vendor == VENDOR_INTEL) { - m_assembly = ASM_INTEL; + m_assembly = Assembly::INTEL; } } @@ -89,13 +109,21 @@ xmrig::AdvancedCpuInfo::AdvancedCpuInfo() : } -size_t xmrig::AdvancedCpuInfo::optimalThreadsCount(size_t memSize, int maxCpuUsage) const +xmrig::CpuThreads xmrig::AdvancedCpuInfo::threads(const Algorithm &algorithm) const { if (threads() == 1) { return 1; } +# ifdef XMRIG_ALGO_CN_GPU + if (algorithm == Algorithm::CN_GPU) { + return CpuThreads(threads()); + } +# endif + size_t cache = 0; + size_t count = 0; + if (m_L3) { cache = m_L2_exclusive ? (m_L2 + m_L3) : m_L3; } @@ -103,12 +131,13 @@ size_t xmrig::AdvancedCpuInfo::optimalThreadsCount(size_t memSize, int maxCpuUsa cache = m_L2; } - size_t count = 0; - if (cache) { - count = cache / memSize; + const size_t memory = algorithm.l3(); + assert(memory > 0); - if (cache % memSize >= memSize / 2) { + count = cache / memory; + + if (cache % memory >= memory / 2) { count++; } } @@ -116,13 +145,13 @@ size_t xmrig::AdvancedCpuInfo::optimalThreadsCount(size_t memSize, int maxCpuUsa count = threads() / 2; } - if (count > (size_t) threads()) { - count = threads(); - } + int intensity = algorithm.maxIntensity() == 1 ? -1 : 1; - if (((float) count / threads() * 100) > maxCpuUsage) { - count = (int) ceil((float) threads() * (maxCpuUsage / 100.0)); +# ifdef XMRIG_ALGO_CN_PICO + if (algorithm == Algorithm::CN_PICO_0 && (count / cores()) >= 2) { + intensity = 2; } +# endif - return count < 1 ? 1 : count; + return CpuThreads(std::max(std::min(count, threads()), 1), intensity); } diff --git a/src/core/cpu/AdvancedCpuInfo.h b/src/backend/cpu/platform/AdvancedCpuInfo.h similarity index 63% rename from src/core/cpu/AdvancedCpuInfo.h rename to src/backend/cpu/platform/AdvancedCpuInfo.h index 0765da339..51b84c9fd 100644 --- a/src/core/cpu/AdvancedCpuInfo.h +++ b/src/backend/cpu/platform/AdvancedCpuInfo.h @@ -26,7 +26,7 @@ #define XMRIG_ADVANCEDCPUINFO_H -#include "common/interfaces/ICpuInfo.h" +#include "backend/cpu/interfaces/ICpuInfo.h" namespace xmrig { @@ -38,37 +38,32 @@ public: AdvancedCpuInfo(); protected: - size_t optimalThreadsCount(size_t memSize, int maxCpuUsage) const override; + CpuThreads threads(const Algorithm &algorithm) const override; - inline Assembly assembly() const override { return m_assembly; } + inline Assembly::Id assembly() const override { return m_assembly; } inline bool hasAES() const override { return m_aes; } inline bool hasAVX2() const override { return m_avx2; } - inline bool isSupported() const override { return true; } + inline const char *backend() const override { return m_backend; } inline const char *brand() const override { return m_brand; } - inline int32_t cores() const override { return m_cores; } - inline int32_t L2() const override { return m_L2; } - inline int32_t L3() const override { return m_L3; } - inline int32_t nodes() const override { return -1; } - inline int32_t sockets() const override { return m_sockets; } - inline int32_t threads() const override { return m_threads; } - -# if defined(__x86_64__) || defined(_M_AMD64) - inline bool isX64() const override { return true; } -# else - inline bool isX64() const override { return false; } -# endif + inline size_t cores() const override { return m_cores; } + inline size_t L2() const override { return m_L2; } + inline size_t L3() const override { return m_L3; } + inline size_t nodes() const override { return 0; } + inline size_t packages() const override { return m_packages; } + inline size_t threads() const override { return m_threads; } private: Assembly m_assembly; - bool m_aes; - bool m_avx2; - bool m_L2_exclusive; - char m_brand[64]; - int32_t m_cores; - int32_t m_L2; - int32_t m_L3; - int32_t m_sockets; - int32_t m_threads; + bool m_aes = false; + bool m_avx2 = false; + bool m_L2_exclusive = false; + char m_backend[32]; + char m_brand[64 + 5]; + size_t m_cores = 0; + size_t m_L2 = 0; + size_t m_L3 = 0; + size_t m_packages = 1; + size_t m_threads = 0; }; diff --git a/src/backend/cpu/platform/BasicCpuInfo.cpp b/src/backend/cpu/platform/BasicCpuInfo.cpp new file mode 100644 index 000000000..2f3177d2e --- /dev/null +++ b/src/backend/cpu/platform/BasicCpuInfo.cpp @@ -0,0 +1,226 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2019 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#include +#include +#include + + +#ifdef _MSC_VER +# include +#else +# include +#endif + +#ifndef bit_AES +# define bit_AES (1 << 25) +#endif + +#ifndef bit_OSXSAVE +# define bit_OSXSAVE (1 << 27) +#endif + +#ifndef bit_AVX2 +# define bit_AVX2 (1 << 5) +#endif + + +#include "backend/cpu/platform/BasicCpuInfo.h" +#include "crypto/common/Assembly.h" + + +#define VENDOR_ID (0) +#define PROCESSOR_INFO (1) +#define CACHE_TLB_DESCRIPTOR (2) +#define EXTENDED_FEATURES (7) +#define PROCESSOR_BRAND_STRING_1 (0x80000002) +#define PROCESSOR_BRAND_STRING_2 (0x80000003) +#define PROCESSOR_BRAND_STRING_3 (0x80000004) + +#define EAX_Reg (0) +#define EBX_Reg (1) +#define ECX_Reg (2) +#define EDX_Reg (3) + + +namespace xmrig { + + +static inline void cpuid(uint32_t level, int32_t output[4]) +{ + memset(output, 0, sizeof(int32_t) * 4); + +# ifdef _MSC_VER + __cpuid(output, static_cast(level)); +# else + __cpuid_count(level, 0, output[0], output[1], output[2], output[3]); +# endif +} + + +static void cpu_brand_string(char out[64 + 6]) { + int32_t cpu_info[4] = { 0 }; + char buf[64] = { 0 }; + + cpuid(VENDOR_ID, cpu_info); + + if (cpu_info[EAX_Reg] >= 4) { + for (uint32_t i = 0; i < 4; i++) { + cpuid(0x80000002 + i, cpu_info); + memcpy(buf + (i * 16), cpu_info, sizeof(cpu_info)); + } + } + + size_t pos = 0; + const size_t size = strlen(buf); + + for (size_t i = 0; i < size; ++i) { + if (buf[i] == ' ' && ((pos > 0 && out[pos - 1] == ' ') || pos == 0)) { + continue; + } + + out[pos++] = buf[i]; + } + + if (pos > 0 && out[pos - 1] == ' ') { + out[pos - 1] = '\0'; + } +} + + +static bool has_feature(uint32_t level, uint32_t reg, int32_t bit) +{ + int32_t cpu_info[4] = { 0 }; + cpuid(level, cpu_info); + + return (cpu_info[reg] & bit) != 0; +} + + +static inline int32_t get_masked(int32_t val, int32_t h, int32_t l) +{ + val &= (0x7FFFFFFF >> (31 - (h - l))) << l; + return val >> l; +} + + +static inline bool has_aes_ni() +{ + return has_feature(PROCESSOR_INFO, ECX_Reg, bit_AES); +} + + +static inline bool has_avx2() +{ + return has_feature(EXTENDED_FEATURES, EBX_Reg, bit_AVX2) && has_feature(PROCESSOR_INFO, ECX_Reg, bit_OSXSAVE); +} + + +} // namespace xmrig + + +xmrig::BasicCpuInfo::BasicCpuInfo() : + m_brand(), + m_threads(std::thread::hardware_concurrency()), + m_assembly(Assembly::NONE), + m_aes(has_aes_ni()), + m_avx2(has_avx2()) +{ + cpu_brand_string(m_brand); + +# ifdef XMRIG_FEATURE_ASM + if (hasAES()) { + char vendor[13] = { 0 }; + int32_t data[4] = { 0 }; + + cpuid(VENDOR_ID, data); + + memcpy(vendor + 0, &data[1], 4); + memcpy(vendor + 4, &data[3], 4); + memcpy(vendor + 8, &data[2], 4); + + if (memcmp(vendor, "AuthenticAMD", 12) == 0) { + cpuid(PROCESSOR_INFO, data); + const int32_t family = get_masked(data[EAX_Reg], 12, 8) + get_masked(data[EAX_Reg], 28, 20); + + m_assembly = family >= 23 ? Assembly::RYZEN : Assembly::BULLDOZER; + } + else { + m_assembly = Assembly::INTEL; + } + } +# endif +} + + +const char *xmrig::BasicCpuInfo::backend() const +{ + return "basic"; +} + + +xmrig::CpuThreads xmrig::BasicCpuInfo::threads(const Algorithm &algorithm) const +{ + const size_t count = std::thread::hardware_concurrency(); + + if (count == 1) { + return 1; + } + +# ifdef XMRIG_ALGO_CN_GPU + if (algorithm == Algorithm::CN_GPU) { + return count; + } +# endif + +# ifdef XMRIG_ALGO_CN_LITE + if (algorithm.family() == Algorithm::CN_LITE) { + return CpuThreads(count, 1); + } +# endif + +# ifdef XMRIG_ALGO_CN_PICO + if (algorithm.family() == Algorithm::CN_PICO) { + return CpuThreads(count, 2); + } +# endif + +# ifdef XMRIG_ALGO_CN_HEAVY + if (algorithm.family() == Algorithm::CN_HEAVY) { + return CpuThreads(std::max(count / 4, 1), 1); + } +# endif + +# ifdef XMRIG_ALGO_RANDOMX + if (algorithm.family() == Algorithm::RANDOM_X) { + if (algorithm == Algorithm::RX_WOW) { + return count; + } + + return std::max(count / 2, 1); + } +# endif + + return CpuThreads(std::max(count / 2, 1), 1); +} diff --git a/src/common/cpu/BasicCpuInfo.h b/src/backend/cpu/platform/BasicCpuInfo.h similarity index 64% rename from src/common/cpu/BasicCpuInfo.h rename to src/backend/cpu/platform/BasicCpuInfo.h index 95857ed27..6cf257146 100644 --- a/src/common/cpu/BasicCpuInfo.h +++ b/src/backend/cpu/platform/BasicCpuInfo.h @@ -26,7 +26,7 @@ #define XMRIG_BASICCPUINFO_H -#include "common/interfaces/ICpuInfo.h" +#include "backend/cpu/interfaces/ICpuInfo.h" namespace xmrig { @@ -38,32 +38,28 @@ public: BasicCpuInfo(); protected: - size_t optimalThreadsCount(size_t memSize, int maxCpuUsage) const override; + const char *backend() const override; + CpuThreads threads(const Algorithm &algorithm) const override; - inline Assembly assembly() const override { return m_assembly; } + inline Assembly::Id assembly() const override { return m_assembly; } inline bool hasAES() const override { return m_aes; } inline bool hasAVX2() const override { return m_avx2; } - inline bool isSupported() const override { return true; } inline const char *brand() const override { return m_brand; } - inline int32_t cores() const override { return -1; } - inline int32_t L2() const override { return -1; } - inline int32_t L3() const override { return -1; } - inline int32_t nodes() const override { return -1; } - inline int32_t sockets() const override { return 1; } - inline int32_t threads() const override { return m_threads; } + inline size_t cores() const override { return 0; } + inline size_t L2() const override { return 0; } + inline size_t L3() const override { return 0; } + inline size_t nodes() const override { return 0; } + inline size_t packages() const override { return 1; } + inline size_t threads() const override { return m_threads; } -# if defined(__x86_64__) || defined(_M_AMD64) || defined (__arm64__) || defined (__aarch64__) - inline bool isX64() const override { return true; } -# else - inline bool isX64() const override { return false; } -# endif +protected: + char m_brand[64 + 6]; + size_t m_threads; private: Assembly m_assembly; bool m_aes; - bool m_avx2; - char m_brand[64]; - int32_t m_threads; + const bool m_avx2; }; diff --git a/src/common/cpu/BasicCpuInfo_arm.cpp b/src/backend/cpu/platform/BasicCpuInfo_arm.cpp similarity index 79% rename from src/common/cpu/BasicCpuInfo_arm.cpp rename to src/backend/cpu/platform/BasicCpuInfo_arm.cpp index 339613466..b241e1970 100644 --- a/src/common/cpu/BasicCpuInfo_arm.cpp +++ b/src/backend/cpu/platform/BasicCpuInfo_arm.cpp @@ -25,20 +25,21 @@ #include #include -#if __ARM_FEATURE_CRYPTO + +#if __ARM_FEATURE_CRYPTO && !defined(__APPLE__) # include # include #endif -#include "common/cpu/BasicCpuInfo.h" +#include "backend/cpu/platform/BasicCpuInfo.h" xmrig::BasicCpuInfo::BasicCpuInfo() : - m_aes(false), - m_avx2(false), m_brand(), - m_threads(std::thread::hardware_concurrency()) + m_threads(std::thread::hardware_concurrency()), + m_aes(false), + m_avx2(false) { # ifdef XMRIG_ARMv8 memcpy(m_brand, "ARMv8", 5); @@ -47,12 +48,22 @@ xmrig::BasicCpuInfo::BasicCpuInfo() : # endif # if __ARM_FEATURE_CRYPTO +# if !defined(__APPLE__) m_aes = getauxval(AT_HWCAP) & HWCAP_AES; +# else + m_aes = true; +# endif # endif } -size_t xmrig::BasicCpuInfo::optimalThreadsCount(size_t memSize, int maxCpuUsage) const +const char *xmrig::BasicCpuInfo::backend() const { - return threads(); + return "basic_arm"; +} + + +xmrig::CpuThreads xmrig::BasicCpuInfo::threads(const Algorithm &) const +{ + return CpuThreads(threads()); } diff --git a/src/backend/cpu/platform/HwlocCpuInfo.cpp b/src/backend/cpu/platform/HwlocCpuInfo.cpp new file mode 100644 index 000000000..491305ec3 --- /dev/null +++ b/src/backend/cpu/platform/HwlocCpuInfo.cpp @@ -0,0 +1,336 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2019 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#ifdef XMRIG_HWLOC_DEBUG +# include +#endif + + +#include +#include + + +#if HWLOC_API_VERSION < 0x00010b00 +# define HWLOC_OBJ_PACKAGE HWLOC_OBJ_SOCKET +# define HWLOC_OBJ_NUMANODE HWLOC_OBJ_NODE +#endif + + +#include "backend/cpu/platform/HwlocCpuInfo.h" +#include "base/io/log/Log.h" + + +namespace xmrig { + + +std::vector HwlocCpuInfo::m_nodeIndexes; +uint32_t HwlocCpuInfo::m_features = 0; + + +static inline bool isCacheObject(hwloc_obj_t obj) +{ +# if HWLOC_API_VERSION >= 0x20000 + return hwloc_obj_type_is_cache(obj->type); +# else + return obj->type == HWLOC_OBJ_CACHE; +# endif +} + + +template +static inline void findCache(hwloc_obj_t obj, unsigned min, unsigned max, func lambda) +{ + for (size_t i = 0; i < obj->arity; i++) { + if (isCacheObject(obj->children[i])) { + const unsigned depth = obj->children[i]->attr->cache.depth; + if (depth < min || depth > max) { + continue; + } + + lambda(obj->children[i]); + } + + findCache(obj->children[i], min, max, lambda); + } +} + + +template +static inline void findByType(hwloc_obj_t obj, hwloc_obj_type_t type, func lambda) +{ + for (size_t i = 0; i < obj->arity; i++) { + if (obj->children[i]->type == type) { + lambda(obj->children[i]); + } + else { + findByType(obj->children[i], type, lambda); + } + } +} + + +static inline std::vector findByType(hwloc_obj_t obj, hwloc_obj_type_t type) +{ + std::vector out; + findByType(obj, type, [&out](hwloc_obj_t found) { out.emplace_back(found); }); + + return out; +} + + +static inline size_t countByType(hwloc_topology_t topology, hwloc_obj_type_t type) +{ + const int count = hwloc_get_nbobjs_by_type(topology, type); + + return count > 0 ? static_cast(count) : 0; +} + + +static inline size_t countByType(hwloc_obj_t obj, hwloc_obj_type_t type) +{ + size_t count = 0; + findByType(obj, type, [&count](hwloc_obj_t) { count++; }); + + return count; +} + + +static inline bool isCacheExclusive(hwloc_obj_t obj) +{ + const char *value = hwloc_obj_get_info_by_name(obj, "Inclusive"); + return value == nullptr || value[0] != '1'; +} + + +} // namespace xmrig + + +xmrig::HwlocCpuInfo::HwlocCpuInfo() : BasicCpuInfo(), + m_backend(), + m_cache() +{ + m_threads = 0; + + hwloc_topology_init(&m_topology); + hwloc_topology_load(m_topology); + +# ifdef XMRIG_HWLOC_DEBUG +# if defined(UV_VERSION_HEX) && UV_VERSION_HEX >= 0x010c00 + { + char env[520] = { 0 }; + size_t size = sizeof(env); + + if (uv_os_getenv("HWLOC_XMLFILE", env, &size) == 0) { + printf("use HWLOC XML file: \"%s\"\n", env); + } + } +# endif + + const std::vector packages = findByType(hwloc_get_root_obj(m_topology), HWLOC_OBJ_PACKAGE); + if (packages.size()) { + const char *value = hwloc_obj_get_info_by_name(packages[0], "CPUModel"); + if (value) { + strncpy(m_brand, value, 64); + } + } +# endif + + hwloc_obj_t root = hwloc_get_root_obj(m_topology); + +# if HWLOC_API_VERSION >= 0x00010b00 + const char *version = hwloc_obj_get_info_by_name(root, "hwlocVersion"); + if (version) { + snprintf(m_backend, sizeof m_backend, "hwloc/%s", version); + } + else +# endif + { + snprintf(m_backend, sizeof m_backend, "hwloc/%d.%d.%d", + (HWLOC_API_VERSION>>16)&0x000000ff, + (HWLOC_API_VERSION>>8 )&0x000000ff, + (HWLOC_API_VERSION )&0x000000ff + ); + } + + findCache(root, 2, 3, [this](hwloc_obj_t found) { this->m_cache[found->attr->cache.depth] += found->attr->cache.size; }); + + m_threads = countByType(m_topology, HWLOC_OBJ_PU); + m_cores = countByType(m_topology, HWLOC_OBJ_CORE); + m_nodes = std::max(countByType(m_topology, HWLOC_OBJ_NUMANODE), 1); + m_packages = countByType(m_topology, HWLOC_OBJ_PACKAGE); + + if (m_nodes > 1) { + if (hwloc_topology_get_support(m_topology)->membind->set_thisthread_membind) { + m_features |= SET_THISTHREAD_MEMBIND; + } + + m_nodeIndexes.reserve(m_nodes); + hwloc_obj_t node = nullptr; + + while ((node = hwloc_get_next_obj_by_type(m_topology, HWLOC_OBJ_NUMANODE, node)) != nullptr) { + m_nodeIndexes.emplace_back(node->os_index); + } + } +} + + +xmrig::HwlocCpuInfo::~HwlocCpuInfo() +{ + hwloc_topology_destroy(m_topology); +} + + +xmrig::CpuThreads xmrig::HwlocCpuInfo::threads(const Algorithm &algorithm) const +{ + if (L2() == 0 && L3() == 0) { + return BasicCpuInfo::threads(algorithm); + } + + const unsigned depth = L3() > 0 ? 3 : 2; + + CpuThreads threads; + threads.reserve(m_threads); + + std::vector caches; + caches.reserve(16); + + findCache(hwloc_get_root_obj(m_topology), depth, depth, [&caches](hwloc_obj_t found) { caches.emplace_back(found); }); + + for (hwloc_obj_t cache : caches) { + processTopLevelCache(cache, algorithm, threads); + } + + if (threads.isEmpty()) { + LOG_WARN("hwloc auto configuration for algorithm \"%s\" failed.", algorithm.shortName()); + + return BasicCpuInfo::threads(algorithm); + } + + return threads; +} + + +void xmrig::HwlocCpuInfo::processTopLevelCache(hwloc_obj_t cache, const Algorithm &algorithm, CpuThreads &threads) const +{ + constexpr size_t oneMiB = 1024u * 1024u; + + size_t PUs = countByType(cache, HWLOC_OBJ_PU); + if (PUs == 0) { + return; + } + + std::vector cores; + cores.reserve(m_cores); + findByType(cache, HWLOC_OBJ_CORE, [&cores](hwloc_obj_t found) { cores.emplace_back(found); }); + + size_t L3 = cache->attr->cache.size; + size_t L2 = 0; + int L2_associativity = 0; + size_t extra = 0; + const size_t scratchpad = algorithm.l3(); + int intensity = algorithm.maxIntensity() == 1 ? -1 : 1; + + if (cache->attr->cache.depth == 3 && isCacheExclusive(cache)) { + for (size_t i = 0; i < cache->arity; ++i) { + hwloc_obj_t l2 = cache->children[i]; + if (!isCacheObject(l2) || l2->attr == nullptr) { + continue; + } + + L2 += l2->attr->cache.size; + L2_associativity = l2->attr->cache.associativity; + + if (l2->attr->cache.size >= scratchpad) { + extra += scratchpad; + } + } + } + + if (scratchpad == 2 * oneMiB) { + if (L2 && (cores.size() * oneMiB) == L2 && L2_associativity == 16 && L3 >= L2) { + L3 = L2; + extra = L2; + } + } + + size_t cacheHashes = ((L3 + extra) + (scratchpad / 2)) / scratchpad; + +# ifdef XMRIG_ALGO_CN_PICO + if (algorithm == Algorithm::CN_PICO_0 && (cacheHashes / PUs) >= 2) { + intensity = 2; + } +# endif + +# ifdef XMRIG_ALGO_CN_GPU + if (algorithm == Algorithm::CN_GPU) { + cacheHashes = PUs; + } +# endif + +# ifdef XMRIG_ALGO_RANDOMX + if (extra == 0 && algorithm.l2() > 0) { + cacheHashes = std::min(std::max(L2 / algorithm.l2(), cores.size()), cacheHashes); + } +# endif + + if (cacheHashes >= PUs) { + for (hwloc_obj_t core : cores) { + const std::vector units = findByType(core, HWLOC_OBJ_PU); + for (hwloc_obj_t pu : units) { + threads.add(pu->os_index, intensity); + } + } + + return; + } + + size_t pu_id = 0; + while (cacheHashes > 0 && PUs > 0) { + bool allocated_pu = false; + + for (hwloc_obj_t core : cores) { + const std::vector units = findByType(core, HWLOC_OBJ_PU); + if (units.size() <= pu_id) { + continue; + } + + cacheHashes--; + PUs--; + + allocated_pu = true; + threads.add(units[pu_id]->os_index, intensity); + + if (cacheHashes == 0) { + break; + } + } + + if (!allocated_pu) { + break; + } + + pu_id++; + } +} diff --git a/src/backend/cpu/platform/HwlocCpuInfo.h b/src/backend/cpu/platform/HwlocCpuInfo.h new file mode 100644 index 000000000..340864f50 --- /dev/null +++ b/src/backend/cpu/platform/HwlocCpuInfo.h @@ -0,0 +1,81 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2019 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_HWLOCCPUINFO_H +#define XMRIG_HWLOCCPUINFO_H + + +#include "backend/cpu/platform/BasicCpuInfo.h" + + +typedef struct hwloc_obj *hwloc_obj_t; +typedef struct hwloc_topology *hwloc_topology_t; + + +namespace xmrig { + + +class HwlocCpuInfo : public BasicCpuInfo +{ +public: + enum Feature : uint32_t { + SET_THISTHREAD_MEMBIND = 1 + }; + + + HwlocCpuInfo(); + ~HwlocCpuInfo() override; + + static inline bool has(Feature feature) { return m_features & feature; } + static inline const std::vector &nodeIndexes() { return m_nodeIndexes; } + +protected: + CpuThreads threads(const Algorithm &algorithm) const override; + + inline const char *backend() const override { return m_backend; } + inline size_t cores() const override { return m_cores; } + inline size_t L2() const override { return m_cache[2]; } + inline size_t L3() const override { return m_cache[3]; } + inline size_t nodes() const override { return m_nodes; } + inline size_t packages() const override { return m_packages; } + +private: + void processTopLevelCache(hwloc_obj_t obj, const Algorithm &algorithm, CpuThreads &threads) const; + + static std::vector m_nodeIndexes; + static uint32_t m_features; + + char m_backend[20]; + hwloc_topology_t m_topology; + size_t m_cache[5]; + size_t m_cores = 0; + size_t m_nodes = 0; + size_t m_packages = 0; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_HWLOCCPUINFO_H */ diff --git a/src/base/base.cmake b/src/base/base.cmake new file mode 100644 index 000000000..ef4da1310 --- /dev/null +++ b/src/base/base.cmake @@ -0,0 +1,148 @@ +set(HEADERS_BASE + src/base/io/Console.h + src/base/io/json/Json.h + src/base/io/json/JsonChain.h + src/base/io/json/JsonRequest.h + src/base/io/log/backends/ConsoleLog.h + src/base/io/log/backends/FileLog.h + src/base/io/log/Log.h + src/base/io/Watcher.h + src/base/kernel/Base.h + src/base/kernel/config/BaseConfig.h + src/base/kernel/config/BaseTransform.h + src/base/kernel/Entry.h + src/base/kernel/interfaces/IBaseListener.h + src/base/kernel/interfaces/IClient.h + src/base/kernel/interfaces/IClientListener.h + src/base/kernel/interfaces/IConfig.h + src/base/kernel/interfaces/IConfigListener.h + src/base/kernel/interfaces/IConfigTransform.h + src/base/kernel/interfaces/IConsoleListener.h + src/base/kernel/interfaces/IDnsListener.h + src/base/kernel/interfaces/ILineListener.h + src/base/kernel/interfaces/ILogBackend.h + src/base/kernel/interfaces/ISignalListener.h + src/base/kernel/interfaces/IStrategy.h + src/base/kernel/interfaces/IStrategyListener.h + src/base/kernel/interfaces/ITimerListener.h + src/base/kernel/interfaces/IWatcherListener.h + src/base/kernel/Platform.h + src/base/kernel/Process.h + src/base/kernel/Signals.h + src/base/net/dns/Dns.h + src/base/net/dns/DnsRecord.h + src/base/net/http/Http.h + src/base/net/stratum/BaseClient.h + src/base/net/stratum/Client.h + src/base/net/stratum/Job.h + src/base/net/stratum/Pool.h + src/base/net/stratum/Pools.h + src/base/net/stratum/strategies/FailoverStrategy.h + src/base/net/stratum/strategies/SinglePoolStrategy.h + src/base/net/stratum/SubmitResult.h + src/base/net/tools/RecvBuf.h + src/base/net/tools/Storage.h + src/base/tools/Arguments.h + src/base/tools/Baton.h + src/base/tools/Buffer.h + src/base/tools/Chrono.h + src/base/tools/Handle.h + src/base/tools/String.h + src/base/tools/Timer.h + ) + +set(SOURCES_BASE + src/base/io/Console.cpp + src/base/io/json/Json.cpp + src/base/io/json/JsonChain.cpp + src/base/io/json/JsonRequest.cpp + src/base/io/log/backends/ConsoleLog.cpp + src/base/io/log/backends/FileLog.cpp + src/base/io/log/Log.cpp + src/base/io/Watcher.cpp + src/base/kernel/Base.cpp + src/base/kernel/config/BaseConfig.cpp + src/base/kernel/config/BaseTransform.cpp + src/base/kernel/Entry.cpp + src/base/kernel/Platform.cpp + src/base/kernel/Process.cpp + src/base/kernel/Signals.cpp + src/base/net/dns/Dns.cpp + src/base/net/dns/DnsRecord.cpp + src/base/net/http/Http.cpp + src/base/net/stratum/BaseClient.cpp + src/base/net/stratum/Client.cpp + src/base/net/stratum/Job.cpp + src/base/net/stratum/Pool.cpp + src/base/net/stratum/Pools.cpp + src/base/net/stratum/strategies/FailoverStrategy.cpp + src/base/net/stratum/strategies/SinglePoolStrategy.cpp + src/base/tools/Arguments.cpp + src/base/tools/Buffer.cpp + src/base/tools/String.cpp + src/base/tools/Timer.cpp + ) + + +if (WIN32) + set(SOURCES_OS + src/base/io/json/Json_win.cpp + src/base/kernel/Platform_win.cpp + ) +elseif (APPLE) + set(SOURCES_OS + src/base/io/json/Json_unix.cpp + src/base/kernel/Platform_mac.cpp + ) +else() + set(SOURCES_OS + src/base/io/json/Json_unix.cpp + src/base/kernel//Platform_unix.cpp + ) +endif() + + +if (NOT WIN32) + CHECK_INCLUDE_FILE (syslog.h HAVE_SYSLOG_H) + if (HAVE_SYSLOG_H) + add_definitions(/DHAVE_SYSLOG_H) + set(SOURCES_SYSLOG src/base/io/log/backends/SysLog.h src/base/io/log/backends/SysLog.cpp) + endif() +endif() + + +if (WITH_HTTP) + set(HEADERS_BASE_HTTP + src/3rdparty/http-parser/http_parser.h + src/base/kernel/interfaces/IHttpListener.h + src/base/kernel/interfaces/IJsonReader.h + src/base/kernel/interfaces/ITcpServerListener.h + src/base/net/http/HttpApiResponse.h + src/base/net/http/HttpClient.h + src/base/net/http/HttpContext.h + src/base/net/http/HttpData.h + src/base/net/http/HttpResponse.h + src/base/net/http/HttpServer.h + src/base/net/stratum/DaemonClient.h + src/base/net/tools/TcpServer.h + ) + + set(SOURCES_BASE_HTTP + src/3rdparty/http-parser/http_parser.c + src/base/net/http/HttpApiResponse.cpp + src/base/net/http/HttpClient.cpp + src/base/net/http/HttpContext.cpp + src/base/net/http/HttpResponse.cpp + src/base/net/http/HttpServer.cpp + src/base/net/stratum/DaemonClient.cpp + src/base/net/tools/TcpServer.cpp + ) + + add_definitions(/DXMRIG_FEATURE_HTTP) + add_definitions(/DXMRIG_FEATURE_API) +else() + set(HEADERS_BASE_HTTP "") + set(SOURCES_BASE_HTTP "") + remove_definitions(/DXMRIG_FEATURE_HTTP) + remove_definitions(/DXMRIG_FEATURE_API) +endif() diff --git a/src/common/Console.cpp b/src/base/io/Console.cpp similarity index 62% rename from src/common/Console.cpp rename to src/base/io/Console.cpp index 350fb1390..0e5cd2696 100644 --- a/src/common/Console.cpp +++ b/src/base/io/Console.cpp @@ -5,7 +5,8 @@ * Copyright 2014-2016 Wolf9466 * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -22,26 +23,44 @@ */ -#include "common/Console.h" -#include "interfaces/IConsoleListener.h" +#include "base/io/Console.h" +#include "base/kernel/interfaces/IConsoleListener.h" +#include "base/tools/Handle.h" -Console::Console(IConsoleListener *listener) +xmrig::Console::Console(IConsoleListener *listener) : m_listener(listener) { - m_tty.data = this; - uv_tty_init(uv_default_loop(), &m_tty, 0, 1); + m_tty = new uv_tty_t; - if (!uv_is_readable(reinterpret_cast(&m_tty))) { + m_tty->data = this; + uv_tty_init(uv_default_loop(), m_tty, 0, 1); + + if (!uv_is_readable(reinterpret_cast(m_tty))) { return; } - uv_tty_set_mode(&m_tty, UV_TTY_MODE_RAW); - uv_read_start(reinterpret_cast(&m_tty), Console::onAllocBuffer, Console::onRead); + uv_tty_set_mode(m_tty, UV_TTY_MODE_RAW); + uv_read_start(reinterpret_cast(m_tty), Console::onAllocBuffer, Console::onRead); } -void Console::onAllocBuffer(uv_handle_t *handle, size_t suggested_size, uv_buf_t *buf) +xmrig::Console::~Console() +{ + stop(); +} + + +void xmrig::Console::stop() +{ + uv_tty_reset_mode(); + + Handle::close(m_tty); + m_tty = nullptr; +} + + +void xmrig::Console::onAllocBuffer(uv_handle_t *handle, size_t, uv_buf_t *buf) { auto console = static_cast(handle->data); buf->len = 1; @@ -49,7 +68,7 @@ void Console::onAllocBuffer(uv_handle_t *handle, size_t suggested_size, uv_buf_t } -void Console::onRead(uv_stream_t *stream, ssize_t nread, const uv_buf_t *buf) +void xmrig::Console::onRead(uv_stream_t *stream, ssize_t nread, const uv_buf_t *buf) { if (nread < 0) { return uv_close(reinterpret_cast(stream), nullptr); diff --git a/src/common/Console.h b/src/base/io/Console.h similarity index 82% rename from src/common/Console.h rename to src/base/io/Console.h index 7f2e3cc90..c0a36ec4f 100644 --- a/src/common/Console.h +++ b/src/base/io/Console.h @@ -5,7 +5,8 @@ * Copyright 2014-2016 Wolf9466 * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -21,13 +22,17 @@ * along with this program. If not, see . */ -#ifndef __CONSOLE_H__ -#define __CONSOLE_H__ +#ifndef XMRIG_CONSOLE_H +#define XMRIG_CONSOLE_H #include + +namespace xmrig { + + class IConsoleListener; @@ -35,6 +40,9 @@ class Console { public: Console(IConsoleListener *listener); + ~Console(); + + void stop(); private: static void onAllocBuffer(uv_handle_t *handle, size_t suggested_size, uv_buf_t *buf); @@ -42,8 +50,11 @@ private: char m_buf[1]; IConsoleListener *m_listener; - uv_tty_t m_tty; + uv_tty_t *m_tty; }; -#endif /* __CONSOLE_H__ */ +} /* namespace xmrig */ + + +#endif /* XMRIG_CONSOLE_H */ diff --git a/src/base/io/Watcher.cpp b/src/base/io/Watcher.cpp index b8ea4c3b9..f5ce9647a 100644 --- a/src/base/io/Watcher.cpp +++ b/src/base/io/Watcher.cpp @@ -29,37 +29,31 @@ #include "base/kernel/interfaces/IWatcherListener.h" #include "base/io/Watcher.h" #include "base/tools/Handle.h" +#include "base/tools/Timer.h" xmrig::Watcher::Watcher(const String &path, IWatcherListener *listener) : m_listener(listener), m_path(path) { + m_timer = new Timer(this); + m_fsEvent = new uv_fs_event_t; + m_fsEvent->data = this; uv_fs_event_init(uv_default_loop(), m_fsEvent); - m_timer = new uv_timer_t; - uv_timer_init(uv_default_loop(), m_timer); - - m_fsEvent->data = m_timer->data = this; - start(); } xmrig::Watcher::~Watcher() { - Handle::close(m_timer); + delete m_timer; + Handle::close(m_fsEvent); } -void xmrig::Watcher::onTimer(uv_timer_t *handle) -{ - static_cast(handle->data)->reload(); -} - - void xmrig::Watcher::onFsEvent(uv_fs_event_t *handle, const char *filename, int, int) { if (!filename) { @@ -72,8 +66,8 @@ void xmrig::Watcher::onFsEvent(uv_fs_event_t *handle, const char *filename, int, void xmrig::Watcher::queueUpdate() { - uv_timer_stop(m_timer); - uv_timer_start(m_timer, xmrig::Watcher::onTimer, kDelay, 0); + m_timer->stop(); + m_timer->start(kDelay, 0); } diff --git a/src/base/io/Watcher.h b/src/base/io/Watcher.h index 4fec4c687..6438cb7a2 100644 --- a/src/base/io/Watcher.h +++ b/src/base/io/Watcher.h @@ -26,30 +26,33 @@ #define XMRIG_WATCHER_H +#include "base/kernel/interfaces/ITimerListener.h" #include "base/tools/String.h" typedef struct uv_fs_event_s uv_fs_event_t; -typedef struct uv_timer_s uv_timer_t; namespace xmrig { class IWatcherListener; +class Timer; -class Watcher +class Watcher : public ITimerListener { public: Watcher(const String &path, IWatcherListener *listener); - ~Watcher(); + ~Watcher() override; + +protected: + inline void onTimer(const Timer *) override { reload(); } private: constexpr static int kDelay = 500; static void onFsEvent(uv_fs_event_t *handle, const char *filename, int events, int status); - static void onTimer(uv_timer_t *handle); void queueUpdate(); void reload(); @@ -57,11 +60,12 @@ private: IWatcherListener *m_listener; String m_path; + Timer *m_timer; uv_fs_event_t *m_fsEvent; - uv_timer_t *m_timer; }; } /* namespace xmrig */ + #endif /* XMRIG_WATCHER_H */ diff --git a/src/base/io/Json.cpp b/src/base/io/json/Json.cpp similarity index 75% rename from src/base/io/Json.cpp rename to src/base/io/json/Json.cpp index b95994e4e..07986c4af 100644 --- a/src/base/io/Json.cpp +++ b/src/base/io/json/Json.cpp @@ -23,10 +23,17 @@ */ -#include "base/io/Json.h" +#include "base/io/json/Json.h" #include "rapidjson/document.h" +namespace xmrig { + +static const rapidjson::Value kNullValue; + +} + + bool xmrig::Json::getBool(const rapidjson::Value &obj, const char *key, bool defaultValue) { auto i = obj.FindMember(key); @@ -49,6 +56,39 @@ const char *xmrig::Json::getString(const rapidjson::Value &obj, const char *key, } +const rapidjson::Value &xmrig::Json::getArray(const rapidjson::Value &obj, const char *key) +{ + auto i = obj.FindMember(key); + if (i != obj.MemberEnd() && i->value.IsArray()) { + return i->value; + } + + return kNullValue; +} + + +const rapidjson::Value &xmrig::Json::getObject(const rapidjson::Value &obj, const char *key) +{ + auto i = obj.FindMember(key); + if (i != obj.MemberEnd() && i->value.IsObject()) { + return i->value; + } + + return kNullValue; +} + + +const rapidjson::Value &xmrig::Json::getValue(const rapidjson::Value &obj, const char *key) +{ + auto i = obj.FindMember(key); + if (i != obj.MemberEnd()) { + return i->value; + } + + return kNullValue; +} + + int xmrig::Json::getInt(const rapidjson::Value &obj, const char *key, int defaultValue) { auto i = obj.FindMember(key); @@ -91,3 +131,9 @@ unsigned xmrig::Json::getUint(const rapidjson::Value &obj, const char *key, unsi return defaultValue; } + + +bool xmrig::JsonReader::isEmpty() const +{ + return !m_obj.IsObject() || m_obj.ObjectEmpty(); +} diff --git a/src/base/io/json/Json.h b/src/base/io/json/Json.h new file mode 100644 index 000000000..80fe5dc26 --- /dev/null +++ b/src/base/io/json/Json.h @@ -0,0 +1,79 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_JSON_H +#define XMRIG_JSON_H + + +#include "base/kernel/interfaces/IJsonReader.h" +#include "rapidjson/fwd.h" + + +namespace xmrig { + + +class Json +{ +public: + static bool getBool(const rapidjson::Value &obj, const char *key, bool defaultValue = false); + static const char *getString(const rapidjson::Value &obj, const char *key, const char *defaultValue = nullptr); + static const rapidjson::Value &getArray(const rapidjson::Value &obj, const char *key); + static const rapidjson::Value &getObject(const rapidjson::Value &obj, const char *key); + static const rapidjson::Value &getValue(const rapidjson::Value &obj, const char *key); + static int getInt(const rapidjson::Value &obj, const char *key, int defaultValue = 0); + static int64_t getInt64(const rapidjson::Value &obj, const char *key, int64_t defaultValue = 0); + static uint64_t getUint64(const rapidjson::Value &obj, const char *key, uint64_t defaultValue = 0); + static unsigned getUint(const rapidjson::Value &obj, const char *key, unsigned defaultValue = 0); + + static bool get(const char *fileName, rapidjson::Document &doc); + static bool save(const char *fileName, const rapidjson::Document &doc); +}; + + +class JsonReader : public IJsonReader +{ +public: + inline JsonReader(const rapidjson::Value &obj) : m_obj(obj) {} + + inline bool getBool(const char *key, bool defaultValue = false) const override { return Json::getBool(m_obj, key, defaultValue); } + inline const char *getString(const char *key, const char *defaultValue = nullptr) const override { return Json::getString(m_obj, key, defaultValue); } + inline const rapidjson::Value &getArray(const char *key) const override { return Json::getArray(m_obj, key); } + inline const rapidjson::Value &getObject(const char *key) const override { return Json::getObject(m_obj, key); } + inline const rapidjson::Value &getValue(const char *key) const override { return Json::getValue(m_obj, key); } + inline int getInt(const char *key, int defaultValue = 0) const override { return Json::getInt(m_obj, key, defaultValue); } + inline int64_t getInt64(const char *key, int64_t defaultValue = 0) const override { return Json::getInt64(m_obj, key, defaultValue); } + inline uint64_t getUint64(const char *key, uint64_t defaultValue = 0) const override { return Json::getUint64(m_obj, key, defaultValue); } + inline unsigned getUint(const char *key, unsigned defaultValue = 0) const override { return Json::getUint(m_obj, key, defaultValue); } + + bool isEmpty() const override; + +private: + const rapidjson::Value &m_obj; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_JSON_H */ diff --git a/src/base/io/json/JsonChain.cpp b/src/base/io/json/JsonChain.cpp new file mode 100644 index 000000000..bbaabbde9 --- /dev/null +++ b/src/base/io/json/JsonChain.cpp @@ -0,0 +1,213 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "base/io/json/Json.h" +#include "base/io/json/JsonChain.h" +#include "base/io/log/Log.h" +#include "rapidjson/error/en.h" + + +namespace xmrig { + +static const rapidjson::Value kNullValue; + +} + + +xmrig::JsonChain::JsonChain() +{ +} + + +bool xmrig::JsonChain::add(rapidjson::Document &&doc) +{ + if (doc.HasParseError() || !doc.IsObject() || doc.ObjectEmpty()) { + return false; + } + + m_chain.push_back(std::move(doc)); + + return true; +} + + +bool xmrig::JsonChain::addFile(const char *fileName) +{ + using namespace rapidjson; + Document doc; + if (Json::get(fileName, doc)) { + m_fileName = fileName; + + return add(std::move(doc)); + } + + if (doc.HasParseError()) { + LOG_ERR("%s: \"%s\"", fileName, doc.GetErrorOffset(), GetParseError_En(doc.GetParseError())); + } + else { + LOG_ERR("unable to open \"%s\".", fileName); + } + + return false; +} + + +bool xmrig::JsonChain::addRaw(const char *json) +{ + using namespace rapidjson; + Document doc; + doc.Parse(json); + + return add(std::move(doc)); +} + + +void xmrig::JsonChain::dump(const char *fileName) +{ + rapidjson::Document doc(rapidjson::kArrayType); + + for (rapidjson::Document &value : m_chain) { + doc.PushBack(value, doc.GetAllocator()); + } + + Json::save(fileName, doc); +} + + +bool xmrig::JsonChain::getBool(const char *key, bool defaultValue) const +{ + for (auto it = m_chain.rbegin(); it != m_chain.rend(); ++it) { + auto i = it->FindMember(key); + if (i != it->MemberEnd() && i->value.IsBool()) { + return i->value.GetBool(); + } + } + + return defaultValue; +} + + +const char *xmrig::JsonChain::getString(const char *key, const char *defaultValue) const +{ + for (auto it = m_chain.rbegin(); it != m_chain.rend(); ++it) { + auto i = it->FindMember(key); + if (i != it->MemberEnd() && i->value.IsString()) { + return i->value.GetString(); + } + } + + return defaultValue; +} + + +const rapidjson::Value &xmrig::JsonChain::getArray(const char *key) const +{ + for (auto it = m_chain.rbegin(); it != m_chain.rend(); ++it) { + auto i = it->FindMember(key); + if (i != it->MemberEnd() && i->value.IsArray()) { + return i->value; + } + } + + return kNullValue; +} + + +const rapidjson::Value &xmrig::JsonChain::getObject(const char *key) const +{ + for (auto it = m_chain.rbegin(); it != m_chain.rend(); ++it) { + auto i = it->FindMember(key); + if (i != it->MemberEnd() && i->value.IsObject()) { + return i->value; + } + } + + return kNullValue; +} + + +const rapidjson::Value &xmrig::JsonChain::getValue(const char *key) const +{ + for (auto it = m_chain.rbegin(); it != m_chain.rend(); ++it) { + auto i = it->FindMember(key); + if (i != it->MemberEnd()) { + return i->value; + } + } + + return kNullValue; +} + + +int xmrig::JsonChain::getInt(const char *key, int defaultValue) const +{ + for (auto it = m_chain.rbegin(); it != m_chain.rend(); ++it) { + auto i = it->FindMember(key); + if (i != it->MemberEnd() && i->value.IsInt()) { + return i->value.GetInt(); + } + } + + return defaultValue; +} + + +int64_t xmrig::JsonChain::getInt64(const char *key, int64_t defaultValue) const +{ + for (auto it = m_chain.rbegin(); it != m_chain.rend(); ++it) { + auto i = it->FindMember(key); + if (i != it->MemberEnd() && i->value.IsInt64()) { + return i->value.GetInt64(); + } + } + + return defaultValue; +} + + +uint64_t xmrig::JsonChain::getUint64(const char *key, uint64_t defaultValue) const +{ + for (auto it = m_chain.rbegin(); it != m_chain.rend(); ++it) { + auto i = it->FindMember(key); + if (i != it->MemberEnd() && i->value.IsUint64()) { + return i->value.GetUint64(); + } + } + + return defaultValue; +} + + +unsigned xmrig::JsonChain::getUint(const char *key, unsigned defaultValue) const +{ + for (auto it = m_chain.rbegin(); it != m_chain.rend(); ++it) { + auto i = it->FindMember(key); + if (i != it->MemberEnd() && i->value.IsUint()) { + return i->value.GetUint(); + } + } + + return defaultValue; +} diff --git a/src/base/io/json/JsonChain.h b/src/base/io/json/JsonChain.h new file mode 100644 index 000000000..275f789ea --- /dev/null +++ b/src/base/io/json/JsonChain.h @@ -0,0 +1,76 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_JSONCHAIN_H +#define XMRIG_JSONCHAIN_H + + +#include + + +#include "base/kernel/interfaces/IJsonReader.h" +#include "base/tools/String.h" +#include "rapidjson/document.h" + + +namespace xmrig { + + +class JsonChain : public IJsonReader +{ +public: + JsonChain(); + + bool add(rapidjson::Document &&doc); + bool addFile(const char *fileName); + bool addRaw(const char *json); + + void dump(const char *fileName); + + inline const String &fileName() const { return m_fileName; } + inline size_t size() const { return m_chain.size(); } + +protected: + inline bool isEmpty() const override { return m_chain.empty(); } + + bool getBool(const char *key, bool defaultValue = false) const override; + const char *getString(const char *key, const char *defaultValue = nullptr) const override; + const rapidjson::Value &getArray(const char *key) const override; + const rapidjson::Value &getObject(const char *key) const override; + const rapidjson::Value &getValue(const char *key) const override; + int getInt(const char *key, int defaultValue = 0) const override; + int64_t getInt64(const char *key, int64_t defaultValue = 0) const override; + uint64_t getUint64(const char *key, uint64_t defaultValue = 0) const override; + unsigned getUint(const char *key, unsigned defaultValue = 0) const override; + +private: + std::vector m_chain; + String m_fileName; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_JSONCHAIN_H */ diff --git a/src/base/io/json/JsonRequest.cpp b/src/base/io/json/JsonRequest.cpp new file mode 100644 index 000000000..4556f3f08 --- /dev/null +++ b/src/base/io/json/JsonRequest.cpp @@ -0,0 +1,38 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "base/io/json/JsonRequest.h" +#include "rapidjson/document.h" + + +void xmrig::JsonRequest::create(rapidjson::Document &doc, int64_t id, const char *method, rapidjson::Value ¶ms) +{ + auto &allocator = doc.GetAllocator(); + + doc.AddMember("id", id, allocator); + doc.AddMember("jsonrpc", "2.0", allocator); + doc.AddMember("method", rapidjson::StringRef(method), allocator); + doc.AddMember("params", params, allocator); +} diff --git a/src/base/io/json/JsonRequest.h b/src/base/io/json/JsonRequest.h new file mode 100644 index 000000000..e98c0baec --- /dev/null +++ b/src/base/io/json/JsonRequest.h @@ -0,0 +1,45 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_JSONREQUEST_H +#define XMRIG_JSONREQUEST_H + + +#include "rapidjson/fwd.h" + + +namespace xmrig { + + +class JsonRequest +{ +public: + static void create(rapidjson::Document &doc, int64_t id, const char *method, rapidjson::Value ¶ms); +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_JSONREQUEST_H */ diff --git a/src/base/io/Json_unix.cpp b/src/base/io/json/Json_unix.cpp similarity index 95% rename from src/base/io/Json_unix.cpp rename to src/base/io/json/Json_unix.cpp index da3902d8d..dedea947d 100644 --- a/src/base/io/Json_unix.cpp +++ b/src/base/io/json/Json_unix.cpp @@ -26,7 +26,7 @@ #include -#include "base/io/Json.h" +#include "base/io/json/Json.h" #include "rapidjson/document.h" #include "rapidjson/istreamwrapper.h" #include "rapidjson/ostreamwrapper.h" @@ -56,6 +56,8 @@ bool xmrig::Json::save(const char *fileName, const rapidjson::Document &doc) rapidjson::OStreamWrapper osw(ofs); rapidjson::PrettyWriter writer(osw); + writer.SetFormatOptions(rapidjson::kFormatSingleLineArray); + doc.Accept(writer); return true; diff --git a/src/base/io/Json_win.cpp b/src/base/io/json/Json_win.cpp similarity index 96% rename from src/base/io/Json_win.cpp rename to src/base/io/json/Json_win.cpp index 4a1c52664..73aff2c59 100644 --- a/src/base/io/Json_win.cpp +++ b/src/base/io/json/Json_win.cpp @@ -35,7 +35,7 @@ #include -#include "base/io/Json.h" +#include "base/io/json/Json.h" #include "rapidjson/document.h" #include "rapidjson/istreamwrapper.h" #include "rapidjson/ostreamwrapper.h" @@ -102,7 +102,7 @@ bool xmrig::Json::save(const char *fileName, const rapidjson::Document &doc) return false; } # elif defined(__GNUC__) - const int fd = _wopen(toUtf16(fileName).c_str(), _O_WRONLY | _O_BINARY | _O_TRUNC); + const int fd = _wopen(toUtf16(fileName).c_str(), _O_WRONLY | _O_BINARY | _O_CREAT | _O_TRUNC); if (fd == -1) { return false; } @@ -118,6 +118,8 @@ bool xmrig::Json::save(const char *fileName, const rapidjson::Document &doc) OStreamWrapper osw(ofs); PrettyWriter writer(osw); + writer.SetFormatOptions(kFormatSingleLineArray); + doc.Accept(writer); return true; diff --git a/src/base/io/log/Log.cpp b/src/base/io/log/Log.cpp new file mode 100644 index 000000000..4e3bd5a5c --- /dev/null +++ b/src/base/io/log/Log.cpp @@ -0,0 +1,243 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2019 Spudz76 + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#ifdef WIN32 +# include +# include +#endif + + +#include +#include +#include +#include +#include +#include +#include + + +#include "base/io/log/Log.h" +#include "base/kernel/interfaces/ILogBackend.h" +#include "base/tools/Chrono.h" + + +namespace xmrig { + + +static const char *colors_map[] = { + RED_BOLD_S, // EMERG + RED_BOLD_S, // ALERT + RED_BOLD_S, // CRIT + RED_S, // ERR + YELLOW_S, // WARNING + WHITE_BOLD_S, // NOTICE + nullptr, // INFO +# ifdef WIN32 + BLACK_BOLD_S // DEBUG +# else + BRIGHT_BLACK_S // DEBUG +# endif +}; + + + +class LogPrivate +{ +public: + inline LogPrivate() : + m_buf() + { + } + + + inline ~LogPrivate() + { + for (ILogBackend *backend : m_backends) { + delete backend; + } + } + + + inline void add(ILogBackend *backend) { m_backends.push_back(backend); } + + + void print(Log::Level level, const char *fmt, va_list args) + { + size_t size = 0; + size_t offset = 0; + + std::lock_guard lock(m_mutex); + + timestamp(level, size, offset); + color(level, size); + + const int rc = vsnprintf(m_buf + size, sizeof (m_buf) - offset - 32, fmt, args); + if (rc < 0) { + return; + } + + size += std::min(static_cast(rc), sizeof (m_buf) - offset - 32); + endl(size); + + std::string txt(m_buf); + size_t i; + while ((i = txt.find(CSI)) != std::string::npos) { + txt.erase(i, txt.find('m', i) - i + 1); + } + + if (!m_backends.empty()) { + for (ILogBackend *backend : m_backends) { + backend->print(level, m_buf, offset, size, true); + backend->print(level, txt.c_str(), offset, txt.size(), false); + } + } + else { + fputs(txt.c_str(), stdout); + fflush(stdout); + } + } + + +private: + inline void timestamp(Log::Level level, size_t &size, size_t &offset) + { + if (level == Log::NONE) { + return; + } + + const uint64_t ms = Chrono::currentMSecsSinceEpoch(); + time_t now = ms / 1000; + tm stime; + +# ifdef _WIN32 + localtime_s(&stime, &now); +# else + localtime_r(&now, &stime); +# endif + + const int rc = snprintf(m_buf, sizeof(m_buf) - 1, "[%d-%02d-%02d %02d:%02d:%02d" BLACK_BOLD(".%03d") "] ", + stime.tm_year + 1900, + stime.tm_mon + 1, + stime.tm_mday, + stime.tm_hour, + stime.tm_min, + stime.tm_sec, + static_cast(ms % 1000) + ); + + if (rc > 0) { + size = offset = static_cast(rc); + } + } + + + inline void color(Log::Level level, size_t &size) + { + if (level == Log::NONE) { + return; + } + + const char *color = colors_map[level]; + if (color == nullptr) { + return; + } + + const size_t s = strlen(color); + memcpy(m_buf + size, color, s); + + size += s; + } + + + inline void endl(size_t &size) + { +# ifdef _WIN32 + memcpy(m_buf + size, CLEAR "\r\n", 7); + size += 6; +# else + memcpy(m_buf + size, CLEAR "\n", 6); + size += 5; +# endif + } + + + char m_buf[4096]; + std::mutex m_mutex; + std::vector m_backends; +}; + + +bool Log::colors = true; +LogPrivate *Log::d = new LogPrivate(); + + +} /* namespace xmrig */ + + + +void xmrig::Log::add(ILogBackend *backend) +{ + if (d) { + d->add(backend); + } +} + + +void xmrig::Log::destroy() +{ + delete d; + d = nullptr; +} + + +void xmrig::Log::print(const char *fmt, ...) +{ + if (!d) { + return; + } + + va_list args; + va_start(args, fmt); + + d->print(NONE, fmt, args); + + va_end(args); +} + + +void xmrig::Log::print(Level level, const char *fmt, ...) +{ + if (!d) { + return; + } + + va_list args; + va_start(args, fmt); + + d->print(level, fmt, args); + + va_end(args); +} diff --git a/src/base/io/log/Log.h b/src/base/io/log/Log.h new file mode 100644 index 000000000..d8bcb44a7 --- /dev/null +++ b/src/base/io/log/Log.h @@ -0,0 +1,143 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2019 Spudz76 + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_LOG_H +#define XMRIG_LOG_H + + +namespace xmrig { + + +class ILogBackend; +class LogPrivate; + + +class Log +{ +public: + enum Level : int { + NONE = -1, + EMERG, // system is unusable + ALERT, // action must be taken immediately + CRIT, // critical conditions + ERR, // error conditions + WARNING, // warning conditions + NOTICE, // normal but significant condition + INFO, // informational + DEBUG, // debug-level messages + }; + + static void add(ILogBackend *backend); + static void destroy(); + static void print(const char *fmt, ...); + static void print(Level level, const char *fmt, ...); + + static bool colors; + +private: + static LogPrivate *d; +}; + + +#define CSI "\x1B[" // Control Sequence Introducer (ANSI spec name) +#define CLEAR CSI "0m" // all attributes off +#define BRIGHT_BLACK_S CSI "0;90m" // somewhat MD.GRAY +#define BLACK_S CSI "0;30m" +#define BLACK_BOLD_S CSI "1;30m" // another name for GRAY +#define RED_S CSI "0;31m" +#define RED_BOLD_S CSI "1;31m" +#define GREEN_S CSI "0;32m" +#define GREEN_BOLD_S CSI "1;32m" +#define YELLOW_S CSI "0;33m" +#define YELLOW_BOLD_S CSI "1;33m" +#define BLUE_S CSI "0;34m" +#define BLUE_BOLD_S CSI "1;34m" +#define MAGENTA_S CSI "0;35m" +#define MAGENTA_BOLD_S CSI "1;35m" +#define CYAN_S CSI "0;36m" +#define CYAN_BOLD_S CSI "1;36m" +#define WHITE_S CSI "0;37m" // another name for LT.GRAY +#define WHITE_BOLD_S CSI "1;37m" // actually white + +#define BLUE_BG_S CSI "44m" +#define BLUE_BG_BOLD_S CSI "44;1m" +#define MAGENTA_BG_S CSI "45m" +#define MAGENTA_BG_BOLD_S CSI "45;1m" +#define CYAN_BG_S CSI "46m" +#define CYAN_BG_BOLD_S CSI "46;1m" + +//color wrappings +#define BLACK(x) BLACK_S x CLEAR +#define BLACK_BOLD(x) BLACK_BOLD_S x CLEAR +#define RED(x) RED_S x CLEAR +#define RED_BOLD(x) RED_BOLD_S x CLEAR +#define GREEN(x) GREEN_S x CLEAR +#define GREEN_BOLD(x) GREEN_BOLD_S x CLEAR +#define YELLOW(x) YELLOW_S x CLEAR +#define YELLOW_BOLD(x) YELLOW_BOLD_S x CLEAR +#define BLUE(x) BLUE_S x CLEAR +#define BLUE_BOLD(x) BLUE_BOLD_S x CLEAR +#define MAGENTA(x) MAGENTA_S x CLEAR +#define MAGENTA_BOLD(x) MAGENTA_BOLD_S x CLEAR +#define CYAN(x) CYAN_S x CLEAR +#define CYAN_BOLD(x) CYAN_BOLD_S x CLEAR +#define WHITE(x) WHITE_S x CLEAR +#define WHITE_BOLD(x) WHITE_BOLD_S x CLEAR + +#define BLUE_BG(x) BLUE_BG_S x CLEAR +#define BLUE_BG_BOLD(x) BLUE_BG_BOLD_S x CLEAR +#define MAGENTA_BG(x) MAGENTA_BG_S x CLEAR +#define MAGENTA_BG_BOLD(x) MAGENTA_BG_BOLD_S x CLEAR +#define CYAN_BG(x) CYAN_BG_S x CLEAR +#define CYAN_BG_BOLD(x) CYAN_BG_BOLD_S x CLEAR + + +#define LOG_EMERG(x, ...) xmrig::Log::print(xmrig::Log::EMERG, x, ##__VA_ARGS__) +#define LOG_ALERT(x, ...) xmrig::Log::print(xmrig::Log::ALERT, x, ##__VA_ARGS__) +#define LOG_CRIT(x, ...) xmrig::Log::print(xmrig::Log::CRIT, x, ##__VA_ARGS__) +#define LOG_ERR(x, ...) xmrig::Log::print(xmrig::Log::ERR, x, ##__VA_ARGS__) +#define LOG_WARN(x, ...) xmrig::Log::print(xmrig::Log::WARNING, x, ##__VA_ARGS__) +#define LOG_NOTICE(x, ...) xmrig::Log::print(xmrig::Log::NOTICE, x, ##__VA_ARGS__) +#define LOG_INFO(x, ...) xmrig::Log::print(xmrig::Log::INFO, x, ##__VA_ARGS__) + +#ifdef APP_DEBUG +# define LOG_DEBUG(x, ...) xmrig::Log::print(xmrig::Log::DEBUG, x, ##__VA_ARGS__) +#else +# define LOG_DEBUG(x, ...) +#endif + +#if defined(APP_DEBUG) || defined(APP_DEVEL) +# define LOG_DEBUG_ERR(x, ...) xmrig::Log::print(xmrig::Log::ERR, x, ##__VA_ARGS__) +# define LOG_DEBUG_WARN(x, ...) xmrig::Log::print(xmrig::Log::WARNING, x, ##__VA_ARGS__) +#else +# define LOG_DEBUG_ERR(x, ...) +# define LOG_DEBUG_WARN(x, ...) +#endif + + +} /* namespace xmrig */ + + +#endif /* XMRIG_LOG_H */ diff --git a/src/common/log/ConsoleLog.cpp b/src/base/io/log/backends/ConsoleLog.cpp similarity index 52% rename from src/common/log/ConsoleLog.cpp rename to src/base/io/log/backends/ConsoleLog.cpp index b10812a6d..a5b6c1a7d 100644 --- a/src/common/log/ConsoleLog.cpp +++ b/src/base/io/log/backends/ConsoleLog.cpp @@ -5,7 +5,9 @@ * Copyright 2014-2016 Wolf9466 * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , + * Copyright 2019 Spudz76 + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -22,36 +24,26 @@ */ -#include #include -#include -#include -#include - -#ifdef WIN32 -# include -# include -#endif -#include "common/log/ConsoleLog.h" -#include "common/log/Log.h" -#include "core/Config.h" -#include "core/Controller.h" +#include "base/tools/Handle.h" +#include "base/io/log/backends/ConsoleLog.h" +#include "base/io/log/Log.h" -ConsoleLog::ConsoleLog(xmrig::Controller *controller) : - m_stream(nullptr), - m_controller(controller) +xmrig::ConsoleLog::ConsoleLog() : + m_stream(nullptr) { - if (uv_tty_init(uv_default_loop(), &m_tty, 1, 0) < 0) { + m_tty = new uv_tty_t; + + if (uv_tty_init(uv_default_loop(), m_tty, 1, 0) < 0) { Log::colors = false; return; } - uv_tty_set_mode(&m_tty, UV_TTY_MODE_NORMAL); - m_uvBuf.base = m_buf; - m_stream = reinterpret_cast(&m_tty); + uv_tty_set_mode(m_tty, UV_TTY_MODE_NORMAL); + m_stream = reinterpret_cast(m_tty); # ifdef WIN32 HANDLE handle = GetStdHandle(STD_INPUT_HANDLE); @@ -66,44 +58,35 @@ ConsoleLog::ConsoleLog(xmrig::Controller *controller) : } -void ConsoleLog::message(Level level, const char* fmt, va_list args) +xmrig::ConsoleLog::~ConsoleLog() { - time_t now = time(nullptr); - tm stime; + Handle::close(m_tty); +} + + +void xmrig::ConsoleLog::print(int, const char *line, size_t, size_t size, bool colors) +{ + if (Log::colors != colors) { + return; + } # ifdef _WIN32 - localtime_s(&stime, &now); + uv_buf_t buf = uv_buf_init(const_cast(line), static_cast(size)); # else - localtime_r(&now, &stime); + uv_buf_t buf = uv_buf_init(const_cast(line), size); # endif - const bool isColors = m_controller->config()->isColors(); - - snprintf(m_fmt, sizeof(m_fmt) - 1, "[%d-%02d-%02d %02d:%02d:%02d]%s %s%s", - stime.tm_year + 1900, - stime.tm_mon + 1, - stime.tm_mday, - stime.tm_hour, - stime.tm_min, - stime.tm_sec, - Log::colorByLevel(level, isColors), - fmt, - Log::endl(isColors) - ); - - print(args); + if (!isWritable()) { + fputs(line, stdout); + fflush(stdout); + } + else { + uv_try_write(m_stream, &buf, 1); + } } -void ConsoleLog::text(const char* fmt, va_list args) -{ - snprintf(m_fmt, sizeof(m_fmt) - 1, "%s%s", fmt, Log::endl(m_controller->config()->isColors())); - - print(args); -} - - -bool ConsoleLog::isWritable() const +bool xmrig::ConsoleLog::isWritable() const { if (!m_stream || uv_is_writable(m_stream) != 1) { return false; @@ -112,20 +95,3 @@ bool ConsoleLog::isWritable() const const uv_handle_type type = uv_guess_handle(1); return type == UV_TTY || type == UV_NAMED_PIPE; } - - -void ConsoleLog::print(va_list args) -{ - m_uvBuf.len = vsnprintf(m_buf, sizeof(m_buf) - 1, m_fmt, args); - if (m_uvBuf.len <= 0) { - return; - } - - if (!isWritable()) { - fputs(m_buf, stdout); - fflush(stdout); - } - else { - uv_try_write(m_stream, &m_uvBuf, 1); - } -} diff --git a/src/base/io/log/backends/ConsoleLog.h b/src/base/io/log/backends/ConsoleLog.h new file mode 100644 index 000000000..90e4fa148 --- /dev/null +++ b/src/base/io/log/backends/ConsoleLog.h @@ -0,0 +1,60 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2019 Spudz76 + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_CONSOLELOG_H +#define XMRIG_CONSOLELOG_H + + +typedef struct uv_stream_s uv_stream_t; +typedef struct uv_tty_s uv_tty_t; + + +#include "base/kernel/interfaces/ILogBackend.h" + + +namespace xmrig { + + +class ConsoleLog : public ILogBackend +{ +public: + ConsoleLog(); + ~ConsoleLog() override; + +protected: + void print(int level, const char *line, size_t offset, size_t size, bool colors) override; + +private: + bool isWritable() const; + + uv_stream_t *m_stream; + uv_tty_t *m_tty; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_CONSOLELOG_H */ diff --git a/src/common/config/ConfigWatcher.cpp b/src/base/io/log/backends/FileLog.cpp similarity index 58% rename from src/common/config/ConfigWatcher.cpp rename to src/base/io/log/backends/FileLog.cpp index 1e35bc9b7..1ff016371 100644 --- a/src/common/config/ConfigWatcher.cpp +++ b/src/base/io/log/backends/FileLog.cpp @@ -5,6 +5,7 @@ * Copyright 2014-2016 Wolf9466 * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , + * Copyright 2019 Spudz76 * Copyright 2018-2019 SChernykh * Copyright 2016-2019 XMRig , * @@ -23,42 +24,44 @@ */ -#include "base/io/Watcher.h" -#include "base/kernel/interfaces/IConfigListener.h" -#include "common/config/ConfigLoader.h" -#include "common/config/ConfigWatcher.h" -#include "common/log/Log.h" -#include "core/ConfigCreator.h" +#include +#include -xmrig::ConfigWatcher::ConfigWatcher(const String &path, IConfigCreator *creator, IConfigListener *listener) : - m_creator(creator), - m_listener(listener) +#include "base/io/log/backends/FileLog.h" + + +xmrig::FileLog::FileLog(const char *fileName) { - m_watcher = new Watcher(path, this); + uv_fs_t req; + m_file = uv_fs_open(uv_default_loop(), &req, fileName, O_CREAT | O_APPEND | O_WRONLY, 0644, nullptr); + uv_fs_req_cleanup(&req); } -xmrig::ConfigWatcher::~ConfigWatcher() +void xmrig::FileLog::print(int, const char *line, size_t, size_t size, bool colors) { - delete m_watcher; -} - - - -void xmrig::ConfigWatcher::onFileChanged(const String &fileName) -{ - LOG_WARN("\"%s\" was changed, reloading configuration", fileName.data()); - - IConfig *config = m_creator->create(); - ConfigLoader::loadFromFile(config, fileName); - - if (!config->finalize()) { - LOG_ERR("reloading failed"); - - delete config; + if (m_file < 0 || colors) { return; } - m_listener->onNewConfig(config); +# ifdef _WIN32 + uv_buf_t buf = uv_buf_init(strdup(line), static_cast(size)); +# else + uv_buf_t buf = uv_buf_init(strdup(line), size); +# endif + + uv_fs_t *req = new uv_fs_t; + req->data = buf.base; + + uv_fs_write(uv_default_loop(), req, m_file, &buf, 1, -1, FileLog::onWrite); +} + + +void xmrig::FileLog::onWrite(uv_fs_t *req) +{ + delete [] static_cast(req->data); + + uv_fs_req_cleanup(req); + delete req; } diff --git a/src/common/log/FileLog.h b/src/base/io/log/backends/FileLog.h similarity index 70% rename from src/common/log/FileLog.h rename to src/base/io/log/backends/FileLog.h index 8a58d4e40..188a99aaa 100644 --- a/src/common/log/FileLog.h +++ b/src/base/io/log/backends/FileLog.h @@ -5,7 +5,9 @@ * Copyright 2014-2016 Wolf9466 * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , + * Copyright 2019 Spudz76 + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -21,37 +23,36 @@ * along with this program. If not, see . */ -#ifndef __FILELOG_H__ -#define __FILELOG_H__ +#ifndef XMRIG_FILELOG_H +#define XMRIG_FILELOG_H -#include +typedef struct uv_fs_s uv_fs_t; -#include "common/interfaces/ILogBackend.h" +#include "base/kernel/interfaces/ILogBackend.h" namespace xmrig { - class Controller; -} class FileLog : public ILogBackend { public: - FileLog(xmrig::Controller *controller, const char *fileName); + FileLog(const char *fileName); - void message(Level level, const char* fmt, va_list args) override; - void text(const char* fmt, va_list args) override; + +protected: + void print(int level, const char *line, size_t offset, size_t size, bool colors) override; private: static void onWrite(uv_fs_t *req); - void write(char *data, size_t size); - - char m_fmt[256]; int m_file; - xmrig::Controller *m_controller; }; -#endif /* __FILELOG_H__ */ + +} /* namespace xmrig */ + + +#endif /* XMRIG_FILELOG_H */ diff --git a/src/base/io/log/backends/SysLog.cpp b/src/base/io/log/backends/SysLog.cpp new file mode 100644 index 000000000..e66f2e35a --- /dev/null +++ b/src/base/io/log/backends/SysLog.cpp @@ -0,0 +1,53 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2019 Spudz76 + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include + + +#include "base/io/log/backends/SysLog.h" +#include "version.h" + + +xmrig::SysLog::SysLog() +{ + openlog(APP_ID, LOG_PID, LOG_USER); +} + + +xmrig::SysLog::~SysLog() +{ + closelog(); +} + + +void xmrig::SysLog::print(int level, const char *line, size_t offset, size_t, bool colors) +{ + if (colors) { + return; + } + + syslog(level == -1 ? LOG_INFO : level, "%s", line + offset); +} diff --git a/src/base/io/log/backends/SysLog.h b/src/base/io/log/backends/SysLog.h new file mode 100644 index 000000000..0e8d0f8ed --- /dev/null +++ b/src/base/io/log/backends/SysLog.h @@ -0,0 +1,50 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2019 Spudz76 + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_SYSLOG_H +#define XMRIG_SYSLOG_H + + +#include "base/kernel/interfaces/ILogBackend.h" + + +namespace xmrig { + + +class SysLog : public ILogBackend +{ +public: + SysLog(); + ~SysLog(); + +protected: + void print(int level, const char *line, size_t offset, size_t size, bool colors) override; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_SYSLOG_H */ diff --git a/src/base/kernel/Base.cpp b/src/base/kernel/Base.cpp new file mode 100644 index 000000000..03feef89f --- /dev/null +++ b/src/base/kernel/Base.cpp @@ -0,0 +1,328 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include +#include + + +#include "base/io/json/Json.h" +#include "base/io/json/JsonChain.h" +#include "base/io/log/backends/ConsoleLog.h" +#include "base/io/log/backends/FileLog.h" +#include "base/io/log/Log.h" +#include "base/io/Watcher.h" +#include "base/kernel/Base.h" +#include "base/kernel/interfaces/IBaseListener.h" +#include "base/kernel/Platform.h" +#include "base/kernel/Process.h" +#include "core/config/Config.h" +#include "core/config/ConfigTransform.h" + + +#ifdef HAVE_SYSLOG_H +# include "base/io/log/backends/SysLog.h" +#endif + + +#ifdef XMRIG_FEATURE_API +# include "api/Api.h" +# include "api/interfaces/IApiRequest.h" +#endif + + +#ifdef XMRIG_FEATURE_EMBEDDED_CONFIG +# include "core/config/Config_default.h" +#endif + + +namespace xmrig { + +static const char *kConfigPathV1 = "/1/config"; +static const char *kConfigPathV2 = "/2/config"; + +} // namespace xmrig + + +class xmrig::BasePrivate +{ +public: + inline BasePrivate(Process *process) : + api(nullptr), + config(nullptr), + process(process), + watcher(nullptr) + {} + + + inline ~BasePrivate() + { +# ifdef XMRIG_FEATURE_API + delete api; +# endif + + delete config; + delete watcher; + } + + + inline bool read(const JsonChain &chain, std::unique_ptr &config) + { + config = std::unique_ptr(new Config()); + + return config->read(chain, chain.fileName()); + } + + + inline Config *load() + { + JsonChain chain; + ConfigTransform transform; + std::unique_ptr config; + + transform.load(chain, process, transform); + + if (read(chain, config)) { + return config.release(); + } + + chain.addFile(process->location(Process::ExeLocation, "config.json")); + + if (read(chain, config)) { + return config.release(); + } + +# ifdef XMRIG_FEATURE_EMBEDDED_CONFIG + chain.addRaw(default_config); + + if (read(chain, config)) { + return config.release(); + } +# endif + + return nullptr; + } + + + inline void replace(Config *newConfig) + { + Config *previousConfig = config; + config = newConfig; + + for (IBaseListener *listener : listeners) { + listener->onConfigChanged(config, previousConfig); + } + + delete previousConfig; + } + + + Api *api; + Config *config; + Process *process; + std::vector listeners; + Watcher *watcher; +}; + + +xmrig::Base::Base(Process *process) + : d_ptr(new BasePrivate(process)) +{ +} + + +xmrig::Base::~Base() +{ + delete d_ptr; +} + + +bool xmrig::Base::isReady() const +{ + return d_ptr->config != nullptr; +} + + +int xmrig::Base::init() +{ + d_ptr->config = d_ptr->load(); + + if (!d_ptr->config) { + LOG_EMERG("No valid configuration found. Exiting."); + + return 1; + } + +# ifdef XMRIG_FEATURE_API + d_ptr->api = new Api(this); + d_ptr->api->addListener(this); +# endif + + Platform::init(config()->userAgent()); + +# ifndef XMRIG_PROXY_PROJECT + Platform::setProcessPriority(config()->cpu().priority()); +# endif + + if (!config()->isBackground()) { + Log::add(new ConsoleLog()); + } + + if (config()->logFile()) { + Log::add(new FileLog(config()->logFile())); + } + +# ifdef HAVE_SYSLOG_H + if (config()->isSyslog()) { + Log::add(new SysLog()); + } +# endif + + return 0; +} + + +void xmrig::Base::start() +{ +# ifdef XMRIG_FEATURE_API + api()->start(); +# endif + + if (config()->isShouldSave()) { + config()->save(); + } + + if (config()->isWatch()) { + d_ptr->watcher = new Watcher(config()->fileName(), this); + } +} + + +void xmrig::Base::stop() +{ +# ifdef XMRIG_FEATURE_API + api()->stop(); +# endif + + delete d_ptr->watcher; + d_ptr->watcher = nullptr; +} + + +xmrig::Api *xmrig::Base::api() const +{ + assert(d_ptr->api != nullptr); + + return d_ptr->api; +} + + +bool xmrig::Base::reload(const rapidjson::Value &json) +{ + JsonReader reader(json); + if (reader.isEmpty()) { + return false; + } + + Config *config = new Config(); + if (!config->read(reader, d_ptr->config->fileName())) { + delete config; + + return false; + } + + const bool saved = config->save(); + + if (config->isWatch() && d_ptr->watcher && saved) { + delete config; + + return true; + } + + d_ptr->replace(config); + + return true; +} + + +xmrig::Config *xmrig::Base::config() const +{ + assert(d_ptr->config != nullptr); + + return d_ptr->config; +} + + +void xmrig::Base::addListener(IBaseListener *listener) +{ + d_ptr->listeners.push_back(listener); +} + + +void xmrig::Base::onFileChanged(const String &fileName) +{ + LOG_WARN("\"%s\" was changed, reloading configuration", fileName.data()); + + JsonChain chain; + chain.addFile(fileName); + + Config *config = new Config(); + + if (!config->read(chain, chain.fileName())) { + LOG_ERR("reloading failed"); + + delete config; + return; + } + + d_ptr->replace(config); +} + + +#ifdef XMRIG_FEATURE_API +void xmrig::Base::onRequest(IApiRequest &request) +{ + if (request.method() == IApiRequest::METHOD_GET) { + if (request.url() == kConfigPathV1 || request.url() == kConfigPathV2) { + if (request.isRestricted()) { + return request.done(403); + } + + request.accept(); + config()->getJSON(request.doc()); + } + } + else if (request.method() == IApiRequest::METHOD_PUT || request.method() == IApiRequest::METHOD_POST) { + if (request.url() == kConfigPathV1 || request.url() == kConfigPathV2) { + request.accept(); + + if (!reload(request.json())) { + return request.done(400); + } + + request.done(204); + } + } +} +#endif diff --git a/src/base/io/Json.h b/src/base/kernel/Base.h similarity index 60% rename from src/base/io/Json.h rename to src/base/kernel/Base.h index 28dcf9a37..6a33a802a 100644 --- a/src/base/io/Json.h +++ b/src/base/kernel/Base.h @@ -22,32 +22,55 @@ * along with this program. If not, see . */ -#ifndef XMRIG_JSON_H -#define XMRIG_JSON_H +#ifndef XMRIG_BASE_H +#define XMRIG_BASE_H +#include "api/interfaces/IApiListener.h" +#include "base/kernel/interfaces/IConfigListener.h" +#include "base/kernel/interfaces/IWatcherListener.h" #include "rapidjson/fwd.h" namespace xmrig { -class Json +class Api; +class BasePrivate; +class Config; +class IBaseListener; +class Process; + + +class Base : public IWatcherListener, public IApiListener { public: - static bool getBool(const rapidjson::Value &obj, const char *key, bool defaultValue = false); - static const char *getString(const rapidjson::Value &obj, const char *key, const char *defaultValue = nullptr); - static int getInt(const rapidjson::Value &obj, const char *key, int defaultValue = 0); - static int64_t getInt64(const rapidjson::Value &obj, const char *key, int64_t defaultValue = 0); - static uint64_t getUint64(const rapidjson::Value &obj, const char *key, uint64_t defaultValue = 0); - static unsigned getUint(const rapidjson::Value &obj, const char *key, unsigned defaultValue = 0); + Base(Process *process); + ~Base() override; - static bool get(const char *fileName, rapidjson::Document &doc); - static bool save(const char *fileName, const rapidjson::Document &doc); + virtual bool isReady() const; + virtual int init(); + virtual void start(); + virtual void stop(); + + Api *api() const; + bool reload(const rapidjson::Value &json); + Config *config() const; + void addListener(IBaseListener *listener); + +protected: + void onFileChanged(const String &fileName) override; + +# ifdef XMRIG_FEATURE_API + void onRequest(IApiRequest &request) override; +# endif + +private: + BasePrivate *d_ptr; }; } /* namespace xmrig */ -#endif /* XMRIG_JSON_H */ +#endif /* XMRIG_BASE_H */ diff --git a/src/base/kernel/Entry.cpp b/src/base/kernel/Entry.cpp index 84c4c9718..f9e97c2d7 100644 --- a/src/base/kernel/Entry.cpp +++ b/src/base/kernel/Entry.cpp @@ -27,22 +27,23 @@ #include -#ifndef XMRIG_NO_HTTPD -# include -#endif - - -#ifndef XMRIG_NO_TLS +#ifdef XMRIG_FEATURE_TLS # include #endif +#ifdef XMRIG_FEATURE_HWLOC +# include +#endif #include "base/kernel/Entry.h" #include "base/kernel/Process.h" -#include "core/usage.h" +#include "core/config/usage.h" #include "version.h" +namespace xmrig { + + static int showVersion() { printf(APP_NAME " " APP_VERSION "\n built on " __DATE__ @@ -73,21 +74,57 @@ static int showVersion() printf("\nlibuv/%s\n", uv_version_string()); -# ifndef XMRIG_NO_HTTPD - printf("microhttpd/%s\n", MHD_get_version()); -# endif - -# if !defined(XMRIG_NO_TLS) && defined(OPENSSL_VERSION_TEXT) +# if defined(XMRIG_FEATURE_TLS) && defined(OPENSSL_VERSION_TEXT) { constexpr const char *v = OPENSSL_VERSION_TEXT + 8; printf("OpenSSL/%.*s\n", static_cast(strchr(v, ' ') - v), v); } # endif +# if defined(XMRIG_FEATURE_HWLOC) +# if defined(HWLOC_VERSION) + printf("hwloc/%s\n", HWLOC_VERSION); +# elif HWLOC_API_VERSION >= 0x20000 + printf("hwloc/2\n"); +# else + printf("hwloc/1\n"); +# endif +# endif + return 0; } +#ifdef XMRIG_FEATURE_HWLOC +static int exportTopology(const Process &process) +{ + const String path = process.location(Process::ExeLocation, "topology.xml"); + + hwloc_topology_t topology; + hwloc_topology_init(&topology); + hwloc_topology_load(topology); + +# if HWLOC_API_VERSION >= 0x20000 + if (hwloc_topology_export_xml(topology, path, 0) == -1) { +# else + if (hwloc_topology_export_xml(topology, path) == -1) { +# endif + printf("failed to export hwloc topology.\n"); + } + else { + printf("hwloc topology successfully exported to \"%s\"\n", path.data()); + } + + hwloc_topology_destroy(topology); + + return 0; +} +#endif + + +} // namespace xmrig + + xmrig::Entry::Id xmrig::Entry::get(const Process &process) { const Arguments &args = process.arguments(); @@ -99,11 +136,17 @@ xmrig::Entry::Id xmrig::Entry::get(const Process &process) return Version; } +# ifdef XMRIG_FEATURE_HWLOC + if (args.hasArg("--export-topology")) { + return Topo; + } +# endif + return Default; } -int xmrig::Entry::exec(const Process &, Id id) +int xmrig::Entry::exec(const Process &process, Id id) { switch (id) { case Usage: @@ -113,6 +156,11 @@ int xmrig::Entry::exec(const Process &, Id id) case Version: return showVersion(); +# ifdef XMRIG_FEATURE_HWLOC + case Topo: + return exportTopology(process); +# endif + default: break; } diff --git a/src/base/kernel/Entry.h b/src/base/kernel/Entry.h index 0208ecdb2..c0bde080d 100644 --- a/src/base/kernel/Entry.h +++ b/src/base/kernel/Entry.h @@ -38,7 +38,8 @@ public: enum Id { Default, Usage, - Version + Version, + Topo }; static Id get(const Process &process); diff --git a/src/common/Platform.cpp b/src/base/kernel/Platform.cpp similarity index 97% rename from src/common/Platform.cpp rename to src/base/kernel/Platform.cpp index 6bd0ac80d..a74f1978f 100644 --- a/src/common/Platform.cpp +++ b/src/base/kernel/Platform.cpp @@ -27,7 +27,7 @@ #include -#ifndef XMRIG_NO_TLS +#ifdef XMRIG_FEATURE_TLS # include # include #endif @@ -41,7 +41,7 @@ xmrig::String Platform::m_userAgent; void Platform::init(const char *userAgent) { -# ifndef XMRIG_NO_TLS +# ifdef XMRIG_FEATURE_TLS SSL_library_init(); SSL_load_error_strings(); ERR_load_BIO_strings(); diff --git a/src/common/Platform.h b/src/base/kernel/Platform.h similarity index 90% rename from src/common/Platform.h rename to src/base/kernel/Platform.h index 85f08a2eb..f3c2c719f 100644 --- a/src/common/Platform.h +++ b/src/base/kernel/Platform.h @@ -35,6 +35,15 @@ class Platform { public: + static inline bool trySetThreadAffinity(int64_t cpu_id) + { + if (cpu_id < 0) { + return false; + } + + return setThreadAffinity(static_cast(cpu_id)); + } + static bool setThreadAffinity(uint64_t cpu_id); static uint32_t setTimerResolution(uint32_t resolution); static void init(const char *userAgent); diff --git a/src/common/Platform_mac.cpp b/src/base/kernel/Platform_mac.cpp similarity index 100% rename from src/common/Platform_mac.cpp rename to src/base/kernel/Platform_mac.cpp diff --git a/src/common/Platform_unix.cpp b/src/base/kernel/Platform_unix.cpp similarity index 100% rename from src/common/Platform_unix.cpp rename to src/base/kernel/Platform_unix.cpp diff --git a/src/common/Platform_win.cpp b/src/base/kernel/Platform_win.cpp similarity index 99% rename from src/common/Platform_win.cpp rename to src/base/kernel/Platform_win.cpp index 9e9b772d2..f2363cd0d 100644 --- a/src/common/Platform_win.cpp +++ b/src/base/kernel/Platform_win.cpp @@ -29,7 +29,7 @@ #include -#include "log/Log.h" +#include "base/io/log/Log.h" #include "Platform.h" #include "version.h" diff --git a/src/base/kernel/Process.cpp b/src/base/kernel/Process.cpp index 0bc40e2da..9193d378e 100644 --- a/src/base/kernel/Process.cpp +++ b/src/base/kernel/Process.cpp @@ -28,6 +28,7 @@ #include "base/kernel/Process.h" +#include "base/tools/Chrono.h" static size_t location(xmrig::Process::Location location, char *buf, size_t max) @@ -50,7 +51,7 @@ static size_t location(xmrig::Process::Location location, char *buf, size_t max) xmrig::Process::Process(int argc, char **argv) : m_arguments(argc, argv) { - srand(static_cast(static_cast(time(nullptr)) ^ reinterpret_cast(this))); + srand(static_cast(Chrono::currentMSecsSinceEpoch() ^ reinterpret_cast(this))); } diff --git a/src/base/kernel/Signals.cpp b/src/base/kernel/Signals.cpp index a37c49c6f..87825b832 100644 --- a/src/base/kernel/Signals.cpp +++ b/src/base/kernel/Signals.cpp @@ -5,8 +5,8 @@ * Copyright 2014-2016 Wolf9466 * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , - * Copyright 2018 SChernykh - * Copyright 2016-2018 XMRig , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -44,20 +44,31 @@ xmrig::Signals::Signals(ISignalListener *listener) m_signals[i] = signal; uv_signal_init(uv_default_loop(), signal); - uv_signal_start(signal, xmrig::Signals::onSignal, signums[i]); + uv_signal_start(signal, Signals::onSignal, signums[i]); } } xmrig::Signals::~Signals() { + stop(); +} + + +void xmrig::Signals::stop() +{ + if (!m_signals[0]) { + return; + } + for (size_t i = 0; i < kSignalsCount; ++i) { Handle::close(m_signals[i]); + m_signals[i] = nullptr; } } void xmrig::Signals::onSignal(uv_signal_t *handle, int signum) { - static_cast(handle->data)->m_listener->onSignal(signum); + static_cast(handle->data)->m_listener->onSignal(signum); } diff --git a/src/base/kernel/Signals.h b/src/base/kernel/Signals.h index 8ef4a4bb4..9b4a870a4 100644 --- a/src/base/kernel/Signals.h +++ b/src/base/kernel/Signals.h @@ -46,6 +46,8 @@ public: Signals(ISignalListener *listener); ~Signals(); + void stop(); + private: void close(int signum); diff --git a/src/base/kernel/config/BaseConfig.cpp b/src/base/kernel/config/BaseConfig.cpp new file mode 100644 index 000000000..4600a1f4b --- /dev/null +++ b/src/base/kernel/config/BaseConfig.cpp @@ -0,0 +1,168 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include +#include +#include +#include +#include + + +#ifdef XMRIG_FEATURE_TLS +# include +#endif + +#ifdef XMRIG_FEATURE_HWLOC +# include "backend/cpu/Cpu.h" +#endif + +#ifdef XMRIG_AMD_PROJECT +# if defined(__APPLE__) +# include +# else +# include "3rdparty/CL/cl.h" +# endif +#endif + + +#ifdef XMRIG_NVIDIA_PROJECT +# include "nvidia/cryptonight.h" +#endif + + +#include "base/io/json/Json.h" +#include "base/io/log/Log.h" +#include "base/kernel/config/BaseConfig.h" +#include "base/kernel/interfaces/IJsonReader.h" +#include "donate.h" +#include "rapidjson/document.h" +#include "rapidjson/filewritestream.h" +#include "rapidjson/prettywriter.h" +#include "version.h" + + +xmrig::BaseConfig::BaseConfig() +{ +} + + +void xmrig::BaseConfig::printVersions() +{ + char buf[256] = { 0 }; + +# if defined(__clang__) + snprintf(buf, sizeof buf, "clang/%d.%d.%d", __clang_major__, __clang_minor__, __clang_patchlevel__); +# elif defined(__GNUC__) + snprintf(buf, sizeof buf, "gcc/%d.%d.%d", __GNUC__, __GNUC_MINOR__, __GNUC_PATCHLEVEL__); +# elif defined(_MSC_VER) + snprintf(buf, sizeof buf, "MSVC/%d", MSVC_VERSION); +# endif + + Log::print(GREEN_BOLD(" * ") WHITE_BOLD("%-13s") CYAN_BOLD("%s/%s") WHITE_BOLD(" %s"), "ABOUT", APP_NAME, APP_VERSION, buf); + +# if defined(XMRIG_AMD_PROJECT) +# if CL_VERSION_2_0 + const char *ocl = "2.0"; +# elif CL_VERSION_1_2 + const char *ocl = "1.2"; +# elif CL_VERSION_1_1 + const char *ocl = "1.1"; +# elif CL_VERSION_1_0 + const char *ocl = "1.0"; +# else + const char *ocl = "0.0"; +# endif + int length = snprintf(buf, sizeof buf, "OpenCL/%s ", ocl); +# elif defined(XMRIG_NVIDIA_PROJECT) + const int cudaVersion = cuda_get_runtime_version(); + int length = snprintf(buf, sizeof buf, "CUDA/%d.%d ", cudaVersion / 1000, cudaVersion % 100); +# endif + + std::string libs; + +# if defined(XMRIG_FEATURE_TLS) && defined(OPENSSL_VERSION_TEXT) + { + constexpr const char *v = OPENSSL_VERSION_TEXT + 8; + snprintf(buf, sizeof buf, "OpenSSL/%.*s ", static_cast(strchr(v, ' ') - v), v); + libs += buf; + } +# endif + +# if defined(XMRIG_FEATURE_HWLOC) + libs += Cpu::info()->backend(); +# endif + + Log::print(GREEN_BOLD(" * ") WHITE_BOLD("%-13slibuv/%s %s"), "LIBS", uv_version_string(), libs.c_str()); +} + + +bool xmrig::BaseConfig::read(const IJsonReader &reader, const char *fileName) +{ + m_fileName = fileName; + + if (reader.isEmpty()) { + return false; + } + + m_autoSave = reader.getBool("autosave", m_autoSave); + m_background = reader.getBool("background", m_background); + m_dryRun = reader.getBool("dry-run", m_dryRun); + m_syslog = reader.getBool("syslog", m_syslog); + m_watch = reader.getBool("watch", m_watch); + Log::colors = reader.getBool("colors", Log::colors); + m_logFile = reader.getString("log-file"); + m_userAgent = reader.getString("user-agent"); + + setPrintTime(reader.getUint("print-time", 60)); + + const rapidjson::Value &api = reader.getObject("api"); + if (api.IsObject()) { + m_apiId = Json::getString(api, "id"); + m_apiWorkerId = Json::getString(api, "worker-id"); + } + + m_http.load(reader.getObject("http")); + m_pools.load(reader); + + return m_pools.active() > 0; +} + + +bool xmrig::BaseConfig::save() +{ + if (m_fileName.isNull()) { + return false; + } + + rapidjson::Document doc; + getJSON(doc); + + if (Json::save(m_fileName, doc)) { + LOG_NOTICE("configuration saved to: \"%s\"", m_fileName.data()); + return true; + } + + return false; +} diff --git a/src/common/config/CommonConfig.h b/src/base/kernel/config/BaseConfig.h similarity index 50% rename from src/common/config/CommonConfig.h rename to src/base/kernel/config/BaseConfig.h index 103e69d24..c5cf29a36 100644 --- a/src/common/config/CommonConfig.h +++ b/src/base/kernel/config/BaseConfig.h @@ -22,91 +22,72 @@ * along with this program. If not, see . */ -#ifndef XMRIG_COMMONCONFIG_H -#define XMRIG_COMMONCONFIG_H +#ifndef XMRIG_BASECONFIG_H +#define XMRIG_BASECONFIG_H -#include "base/net/Pools.h" -#include "base/tools/String.h" -#include "common/interfaces/IConfig.h" -#include "common/xmrig.h" +#include "base/kernel/interfaces/IConfig.h" +#include "base/net/http/Http.h" +#include "base/net/stratum/Pools.h" + + +struct option; namespace xmrig { -class CommonConfig : public IConfig +class IJsonReader; + + +class BaseConfig : public IConfig { public: - CommonConfig(); + BaseConfig(); - inline bool isApiIPv6() const { return m_apiIPv6; } - inline bool isApiRestricted() const { return m_apiRestricted; } inline bool isAutoSave() const { return m_autoSave; } inline bool isBackground() const { return m_background; } inline bool isDryRun() const { return m_dryRun; } inline bool isSyslog() const { return m_syslog; } - inline const char *apiId() const { return m_apiId.data(); } - inline const char *apiToken() const { return m_apiToken.data(); } - inline const char *apiWorkerId() const { return m_apiWorkerId.data(); } inline const char *logFile() const { return m_logFile.data(); } inline const char *userAgent() const { return m_userAgent.data(); } + inline const Http &http() const { return m_http; } inline const Pools &pools() const { return m_pools; } - inline int apiPort() const { return m_apiPort; } - inline int donateLevel() const { return m_donateLevel; } - inline int printTime() const { return m_printTime; } + inline const String &apiId() const { return m_apiId; } + inline const String &apiWorkerId() const { return m_apiWorkerId; } + inline uint32_t printTime() const { return m_printTime; } - inline bool isWatch() const override { return m_watch && !m_fileName.isNull(); } - inline const Algorithm &algorithm() const override { return m_algorithm; } - inline const String &fileName() const override { return m_fileName; } + inline bool isWatch() const override { return m_watch && !m_fileName.isNull(); } + inline const String &fileName() const override { return m_fileName; } + inline void setFileName(const char *fileName) override { m_fileName = fileName; } + bool read(const IJsonReader &reader, const char *fileName) override; bool save() override; - bool isColors() const; - void printAPI(); - void printPools(); void printVersions(); protected: - enum State { - NoneState, - ReadyState, - ErrorState - }; - - bool finalize() override; - bool parseBoolean(int key, bool enable) override; - bool parseString(int key, const char *arg) override; - bool parseUint64(int key, uint64_t arg) override; - void parseJSON(const rapidjson::Document &doc) override; - void setFileName(const char *fileName) override; - - Algorithm m_algorithm; - bool m_adjusted; - bool m_apiIPv6; - bool m_apiRestricted; - bool m_autoSave; - bool m_background; - bool m_dryRun; - bool m_syslog; - bool m_watch; - int m_apiPort; - int m_donateLevel; - int m_printTime; + bool m_autoSave = true; + bool m_background = false; + bool m_dryRun = false; + bool m_syslog = false; + bool m_upgrade = false; + bool m_watch = true; + Http m_http; Pools m_pools; - State m_state; String m_apiId; - String m_apiToken; String m_apiWorkerId; String m_fileName; String m_logFile; String m_userAgent; + uint32_t m_printTime; private: - bool parseInt(int key, int arg); + inline void setPrintTime(uint32_t printTime) { if (printTime <= 3600) { m_printTime = printTime; } } }; -} /* namespace xmrig */ +} // namespace xmrig -#endif /* XMRIG_COMMONCONFIG_H */ + +#endif /* XMRIG_BASECONFIG_H */ diff --git a/src/base/kernel/config/BaseTransform.cpp b/src/base/kernel/config/BaseTransform.cpp new file mode 100644 index 000000000..8043c6e91 --- /dev/null +++ b/src/base/kernel/config/BaseTransform.cpp @@ -0,0 +1,269 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include + + +#ifdef _MSC_VER +# include "getopt/getopt.h" +#else +# include +#endif + + +#include "base/kernel/config/BaseTransform.h" +#include "base/kernel/Process.h" +#include "base/io/log/Log.h" +#include "base/kernel/interfaces/IConfig.h" +#include "base/io/json/JsonChain.h" +#include "core/config/Config_platform.h" + + +namespace xmrig +{ + +static const char *kAlgo = "algo"; +static const char *kApi = "api"; +static const char *kHttp = "http"; +static const char *kPools = "pools"; + +} + + +xmrig::BaseTransform::BaseTransform() +{ +} + + +void xmrig::BaseTransform::load(JsonChain &chain, Process *process, IConfigTransform &transform) +{ + using namespace rapidjson; + + int key; + int argc = process->arguments().argc(); + char **argv = process->arguments().argv(); + + Document doc(kObjectType); + + while (1) { + key = getopt_long(argc, argv, short_options, options, nullptr); + if (key < 0) { + break; + } + + if (key == IConfig::ConfigKey) { + chain.add(std::move(doc)); + chain.addFile(optarg); + + doc = Document(kObjectType); + } + else { + transform.transform(doc, key, optarg); + } + } + + if (optind < argc) { + LOG_WARN("%s: unsupported non-option argument '%s'", argv[0], argv[optind]); + } + + transform.finalize(doc); + chain.add(std::move(doc)); +} + + +void xmrig::BaseTransform::finalize(rapidjson::Document &doc) +{ + using namespace rapidjson; + auto &allocator = doc.GetAllocator(); + + if (m_algorithm.isValid() && doc.HasMember(kPools)) { + auto &pools = doc[kPools]; + for (Value &pool : pools.GetArray()) { + if (!pool.HasMember(kAlgo)) { + pool.AddMember(StringRef(kAlgo), m_algorithm.toJSON(), allocator); + } + } + } +} + + +void xmrig::BaseTransform::transform(rapidjson::Document &doc, int key, const char *arg) +{ + switch (key) { + case IConfig::AlgorithmKey: /* --algo */ + if (!doc.HasMember(kPools)) { + m_algorithm = arg; + } + else { + return add(doc, kPools, kAlgo, arg); + } + break; + + case IConfig::UserpassKey: /* --userpass */ + { + const char *p = strrchr(arg, ':'); + if (!p) { + return; + } + + char *user = new char[p - arg + 1](); + strncpy(user, arg, static_cast(p - arg)); + + add(doc, kPools, "user", user); + add(doc, kPools, "pass", p + 1); + delete [] user; + } + break; + + case IConfig::UrlKey: /* --url */ + return add(doc, kPools, "url", arg, true); + + case IConfig::UserKey: /* --user */ + return add(doc, kPools, "user", arg); + + case IConfig::PasswordKey: /* --pass */ + return add(doc, kPools, "pass", arg); + + case IConfig::RigIdKey: /* --rig-id */ + return add(doc, kPools, "rig-id", arg); + + case IConfig::FingerprintKey: /* --tls-fingerprint */ + return add(doc, kPools, "tls-fingerprint", arg); + + case IConfig::LogFileKey: /* --log-file */ + return set(doc, "log-file", arg); + + case IConfig::HttpAccessTokenKey: /* --http-access-token */ + return set(doc, kHttp, "access-token", arg); + + case IConfig::HttpHostKey: /* --http-host */ + return set(doc, kHttp, "host", arg); + + case IConfig::ApiWorkerIdKey: /* --api-worker-id */ + return set(doc, kApi, "worker-id", arg); + + case IConfig::ApiIdKey: /* --api-id */ + return set(doc, kApi, "id", arg); + + case IConfig::UserAgentKey: /* --user-agent */ + return set(doc, "user-agent", arg); + + case IConfig::RetriesKey: /* --retries */ + case IConfig::RetryPauseKey: /* --retry-pause */ + case IConfig::PrintTimeKey: /* --print-time */ + case IConfig::HttpPort: /* --http-port */ + case IConfig::DonateLevelKey: /* --donate-level */ + case IConfig::DaemonPollKey: /* --daemon-poll-interval */ + return transformUint64(doc, key, static_cast(strtol(arg, nullptr, 10))); + + case IConfig::BackgroundKey: /* --background */ + case IConfig::SyslogKey: /* --syslog */ + case IConfig::KeepAliveKey: /* --keepalive */ + case IConfig::NicehashKey: /* --nicehash */ + case IConfig::TlsKey: /* --tls */ + case IConfig::DryRunKey: /* --dry-run */ + case IConfig::HttpEnabledKey: /* --http-enabled */ + case IConfig::DaemonKey: /* --daemon */ + return transformBoolean(doc, key, true); + + case IConfig::ColorKey: /* --no-color */ + case IConfig::HttpRestrictedKey: /* --http-no-restricted */ + return transformBoolean(doc, key, false); + + default: + break; + } +} + + +void xmrig::BaseTransform::transformBoolean(rapidjson::Document &doc, int key, bool enable) +{ + switch (key) { + case IConfig::BackgroundKey: /* --background */ + return set(doc, "background", enable); + + case IConfig::SyslogKey: /* --syslog */ + return set(doc, "syslog", enable); + + case IConfig::KeepAliveKey: /* --keepalive */ + return add(doc, kPools, "keepalive", enable); + + case IConfig::TlsKey: /* --tls */ + return add(doc, kPools, "tls", enable); + + case IConfig::DaemonKey: /* --daemon */ + return add(doc, kPools, "daemon", enable); + +# ifndef XMRIG_PROXY_PROJECT + case IConfig::NicehashKey: /* --nicehash */ + return add(doc, kPools, "nicehash", enable); +# endif + + case IConfig::ColorKey: /* --no-color */ + return set(doc, "colors", enable); + + case IConfig::HttpRestrictedKey: /* --http-no-restricted */ + return set(doc, kHttp, "restricted", enable); + + case IConfig::HttpEnabledKey: /* --http-enabled */ + return set(doc, kHttp, "enabled", enable); + + case IConfig::DryRunKey: /* --dry-run */ + return set(doc, "dry-run", enable); + + default: + break; + } +} + + +void xmrig::BaseTransform::transformUint64(rapidjson::Document &doc, int key, uint64_t arg) +{ + switch (key) { + case IConfig::RetriesKey: /* --retries */ + return set(doc, "retries", arg); + + case IConfig::RetryPauseKey: /* --retry-pause */ + return set(doc, "retry-pause", arg); + + case IConfig::DonateLevelKey: /* --donate-level */ + return set(doc, "donate-level", arg); + + case IConfig::ProxyDonateKey: /* --donate-over-proxy */ + return set(doc, "donate-over-proxy", arg); + + case IConfig::HttpPort: /* --http-port */ + return set(doc, kHttp, "port", arg); + + case IConfig::PrintTimeKey: /* --print-time */ + return set(doc, "print-time", arg); + + case IConfig::DaemonPollKey: /* --daemon-poll-interval */ + return add(doc, kPools, "daemon-poll-interval", arg); + + default: + break; + } +} diff --git a/src/base/kernel/config/BaseTransform.h b/src/base/kernel/config/BaseTransform.h new file mode 100644 index 000000000..02b28c12c --- /dev/null +++ b/src/base/kernel/config/BaseTransform.h @@ -0,0 +1,127 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_BASETRANSFORM_H +#define XMRIG_BASETRANSFORM_H + + +#include "base/kernel/interfaces/IConfigTransform.h" +#include "rapidjson/document.h" + + +struct option; + + +namespace xmrig { + + +class IConfigTransform; +class JsonChain; +class Process; + + +class BaseTransform : public IConfigTransform +{ +public: + BaseTransform(); + + static void load(JsonChain &chain, Process *process, IConfigTransform &transform); + +protected: + void finalize(rapidjson::Document &doc) override; + void transform(rapidjson::Document &doc, int key, const char *arg) override; + + + template + inline void set(rapidjson::Document &doc, const char *key, T value) { set(doc, doc, key, value); } + + + template + inline void set(rapidjson::Document &doc, const char *objKey, const char *key, T value) + { + if (!doc.HasMember(objKey)) { + doc.AddMember(rapidjson::StringRef(objKey), rapidjson::kObjectType, doc.GetAllocator()); + } + + set(doc, doc[objKey], key, value); + } + + + template + inline void add(rapidjson::Document &doc, const char *arrayKey, const char *key, T value, bool force = false) + { + auto &allocator = doc.GetAllocator(); + + if (!doc.HasMember(arrayKey)) { + doc.AddMember(rapidjson::StringRef(arrayKey), rapidjson::kArrayType, allocator); + } + + rapidjson::Value &array = doc[arrayKey]; + if (force || array.Size() == 0) { + array.PushBack(rapidjson::kObjectType, allocator); + } + + set(doc, array[array.Size() - 1], key, value); + } + + + template + inline void set(rapidjson::Document &doc, rapidjson::Value &obj, const char *key, T value) + { + if (obj.HasMember(key)) { + obj[key] = value; + } + else { + obj.AddMember(rapidjson::StringRef(key), value, doc.GetAllocator()); + } + } + +protected: + Algorithm m_algorithm; + + +private: + void transformBoolean(rapidjson::Document &doc, int key, bool enable); + void transformUint64(rapidjson::Document &doc, int key, uint64_t arg); +}; + + +template<> +inline void BaseTransform::set(rapidjson::Document &doc, rapidjson::Value &obj, const char *key, const char *value) +{ + auto &allocator = doc.GetAllocator(); + + if (obj.HasMember(key)) { + obj[key] = rapidjson::Value(value, allocator); + } + else { + obj.AddMember(rapidjson::StringRef(key), rapidjson::Value(value, allocator), allocator); + } +} + + +} // namespace xmrig + + +#endif /* XMRIG_BASETRANSFORM_H */ diff --git a/src/common/interfaces/IControllerListener.h b/src/base/kernel/interfaces/IBaseListener.h similarity index 88% rename from src/common/interfaces/IControllerListener.h rename to src/base/kernel/interfaces/IBaseListener.h index c61574083..1f2123690 100644 --- a/src/common/interfaces/IControllerListener.h +++ b/src/base/kernel/interfaces/IBaseListener.h @@ -22,8 +22,8 @@ * along with this program. If not, see . */ -#ifndef XMRIG_ICONTROLLERLISTENER_H -#define XMRIG_ICONTROLLERLISTENER_H +#ifndef XMRIG_IBASELISTENER_H +#define XMRIG_IBASELISTENER_H namespace xmrig { @@ -32,10 +32,10 @@ namespace xmrig { class Config; -class IControllerListener +class IBaseListener { public: - virtual ~IControllerListener() = default; + virtual ~IBaseListener() = default; virtual void onConfigChanged(Config *config, Config *previousConfig) = 0; }; @@ -44,4 +44,4 @@ public: } /* namespace xmrig */ -#endif // XMRIG_ICONTROLLERLISTENER_H +#endif // XMRIG_IBASELISTENER_H diff --git a/src/base/kernel/interfaces/IClient.h b/src/base/kernel/interfaces/IClient.h new file mode 100644 index 000000000..c872a37ad --- /dev/null +++ b/src/base/kernel/interfaces/IClient.h @@ -0,0 +1,85 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_ICLIENT_H +#define XMRIG_ICLIENT_H + + +#include + + +namespace xmrig { + + +class Algorithm; +class Job; +class JobResult; +class Pool; +class String; + + +class IClient +{ +public: + enum Extension { + EXT_ALGO, + EXT_NICEHASH, + EXT_CONNECT, + EXT_TLS, + EXT_KEEPALIVE, + EXT_MAX + }; + + virtual ~IClient() = default; + + virtual bool disconnect() = 0; + virtual bool hasExtension(Extension extension) const noexcept = 0; + virtual bool isEnabled() const = 0; + virtual bool isTLS() const = 0; + virtual const char *mode() const = 0; + virtual const char *tlsFingerprint() const = 0; + virtual const char *tlsVersion() const = 0; + virtual const Job &job() const = 0; + virtual const Pool &pool() const = 0; + virtual const String &ip() const = 0; + virtual int id() const = 0; + virtual int64_t submit(const JobResult &result) = 0; + virtual void connect() = 0; + virtual void connect(const Pool &pool) = 0; + virtual void deleteLater() = 0; + virtual void setAlgo(const Algorithm &algo) = 0; + virtual void setEnabled(bool enabled) = 0; + virtual void setPool(const Pool &pool) = 0; + virtual void setQuiet(bool quiet) = 0; + virtual void setRetries(int retries) = 0; + virtual void setRetryPause(uint64_t ms) = 0; + virtual void tick(uint64_t now) = 0; + +}; + + +} /* namespace xmrig */ + + +#endif // XMRIG_ICLIENT_H diff --git a/src/common/interfaces/IClientListener.h b/src/base/kernel/interfaces/IClientListener.h similarity index 69% rename from src/common/interfaces/IClientListener.h rename to src/base/kernel/interfaces/IClientListener.h index 70a3d1d0f..3583be5ad 100644 --- a/src/common/interfaces/IClientListener.h +++ b/src/base/kernel/interfaces/IClientListener.h @@ -29,10 +29,14 @@ #include +#include "rapidjson/fwd.h" + + namespace xmrig { -class Client; +class Algorithm; +class IClient; class Job; class SubmitResult; @@ -42,10 +46,12 @@ class IClientListener public: virtual ~IClientListener() = default; - virtual void onClose(Client *client, int failures) = 0; - virtual void onJobReceived(Client *client, const Job &job) = 0; - virtual void onLoginSuccess(Client *client) = 0; - virtual void onResultAccepted(Client *client, const SubmitResult &result, const char *error) = 0; + virtual void onClose(IClient *client, int failures) = 0; + virtual void onJobReceived(IClient *client, const Job &job, const rapidjson::Value ¶ms) = 0; + virtual void onLogin(IClient *client, rapidjson::Document &doc, rapidjson::Value ¶ms) = 0; + virtual void onLoginSuccess(IClient *client) = 0; + virtual void onResultAccepted(IClient *client, const SubmitResult &result, const char *error) = 0; + virtual void onVerifyAlgorithm(const IClient *client, const Algorithm &algorithm, bool *ok) = 0; }; diff --git a/src/base/kernel/interfaces/IConfig.h b/src/base/kernel/interfaces/IConfig.h new file mode 100644 index 000000000..86b8067c6 --- /dev/null +++ b/src/base/kernel/interfaces/IConfig.h @@ -0,0 +1,147 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_ICONFIG_H +#define XMRIG_ICONFIG_H + + +#include "crypto/common/Algorithm.h" +#include "rapidjson/fwd.h" + + +namespace xmrig { + + +class IJsonReader; +class String; + + +class IConfig +{ +public: + enum Keys { + // common + AlgorithmKey = 'a', + ApiWorkerIdKey = 4002, + ApiIdKey = 4005, + HttpPort = 4100, + HttpAccessTokenKey = 4101, + HttpRestrictedKey = 4104, + HttpEnabledKey = 4106, + HttpHostKey = 4107, + BackgroundKey = 'B', + ColorKey = 1002, + ConfigKey = 'c', + DonateLevelKey = 1003, + KeepAliveKey = 'k', + LogFileKey = 'l', + PasswordKey = 'p', + RetriesKey = 'r', + RetryPauseKey = 'R', + RigIdKey = 1012, + SyslogKey = 'S', + UrlKey = 'o', + UserAgentKey = 1008, + UserKey = 'u', + UserpassKey = 'O', + VerboseKey = 1100, + TlsKey = 1013, + FingerprintKey = 1014, + ProxyDonateKey = 1017, + DaemonKey = 1018, + DaemonPollKey = 1019, + + // xmrig common + CPUPriorityKey = 1021, + NicehashKey = 1006, + PrintTimeKey = 1007, + + // xmrig cpu + AVKey = 'v', + CPUAffinityKey = 1020, + DryRunKey = 5000, + HugePagesKey = 1009, + ThreadsKey = 't', + AssemblyKey = 1015, + RandomXInitKey = 1022, + RandomXNumaKey = 1023, + + // xmrig amd + OclPlatformKey = 1400, + OclAffinityKey = 1401, + OclDevicesKey = 1402, + OclLaunchKey = 1403, + OclCacheKey = 1404, + OclPrintKey = 1405, + OclLoaderKey = 1406, + OclSridedIndexKey = 1407, + OclMemChunkKey = 1408, + OclUnrollKey = 1409, + OclCompModeKey = 1410, + + // xmrig-proxy + AccessLogFileKey = 'A', + BindKey = 'b', + CustomDiffKey = 1102, + DebugKey = 1101, + ModeKey = 'm', + PoolCoinKey = 'C', + ReuseTimeoutKey = 1106, + WorkersKey = 1103, + WorkersAdvKey = 1107, + TlsBindKey = 1108, + TlsCertKey = 1109, + TlsCertKeyKey = 1110, + TlsDHparamKey = 1111, + TlsCiphersKey = 1112, + TlsCipherSuitesKey = 1113, + TlsProtocolsKey = 1114, + AlgoExtKey = 1115, + ProxyPasswordKey = 1116, + + // xmrig nvidia + CudaMaxThreadsKey = 1200, + CudaBFactorKey = 1201, + CudaBSleepKey = 1202, + CudaDevicesKey = 1203, + CudaLaunchKey = 1204, + CudaAffinityKey = 1205, + CudaMaxUsageKey = 1206, + }; + + virtual ~IConfig() = default; + + virtual bool isWatch() const = 0; + virtual bool read(const IJsonReader &reader, const char *fileName) = 0; + virtual bool save() = 0; + virtual const String &fileName() const = 0; + virtual void getJSON(rapidjson::Document &doc) const = 0; + virtual void setFileName(const char *fileName) = 0; +}; + + +} /* namespace xmrig */ + + +#endif // XMRIG_ICONFIG_H diff --git a/src/base/kernel/interfaces/IConfigTransform.h b/src/base/kernel/interfaces/IConfigTransform.h new file mode 100644 index 000000000..8afe82217 --- /dev/null +++ b/src/base/kernel/interfaces/IConfigTransform.h @@ -0,0 +1,53 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_ICONFIGTRANSFORM_H +#define XMRIG_ICONFIGTRANSFORM_H + + +#include "crypto/common/Algorithm.h" +#include "rapidjson/fwd.h" + + +namespace xmrig { + + +class IJsonReader; +class String; + + +class IConfigTransform +{ +public: + virtual ~IConfigTransform() = default; + + virtual void finalize(rapidjson::Document &doc) = 0; + virtual void transform(rapidjson::Document &doc, int key, const char *arg) = 0; +}; + + +} /* namespace xmrig */ + + +#endif // XMRIG_ICONFIGTRANSFORM_H diff --git a/src/base/kernel/interfaces/IConsoleListener.h b/src/base/kernel/interfaces/IConsoleListener.h new file mode 100644 index 000000000..88972f058 --- /dev/null +++ b/src/base/kernel/interfaces/IConsoleListener.h @@ -0,0 +1,44 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_ICONSOLELISTENER_H +#define XMRIG_ICONSOLELISTENER_H + + +namespace xmrig { + + +class IConsoleListener +{ +public: + virtual ~IConsoleListener() = default; + + virtual void onConsoleCommand(char command) = 0; +}; + + +} /* namespace xmrig */ + + +#endif // XMRIG_ICONSOLELISTENER_H diff --git a/src/base/kernel/interfaces/IDnsListener.h b/src/base/kernel/interfaces/IDnsListener.h new file mode 100644 index 000000000..3683ea51d --- /dev/null +++ b/src/base/kernel/interfaces/IDnsListener.h @@ -0,0 +1,47 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_IDNSLISTENER_H +#define XMRIG_IDNSLISTENER_H + + +namespace xmrig { + + +class Dns; + + +class IDnsListener +{ +public: + virtual ~IDnsListener() = default; + + virtual void onResolved(const Dns &dns, int status) = 0; +}; + + +} /* namespace xmrig */ + + +#endif // XMRIG_IDNSLISTENER_H diff --git a/src/base/kernel/interfaces/IHttpListener.h b/src/base/kernel/interfaces/IHttpListener.h new file mode 100644 index 000000000..94fd1d185 --- /dev/null +++ b/src/base/kernel/interfaces/IHttpListener.h @@ -0,0 +1,48 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_IHTTPLISTENER_H +#define XMRIG_IHTTPLISTENER_H + + +namespace xmrig { + + +class HttpData; +class HttpResponse; + + +class IHttpListener +{ +public: + virtual ~IHttpListener() = default; + + virtual void onHttpData(const HttpData &data) = 0; +}; + + +} /* namespace xmrig */ + + +#endif // XMRIG_IHTTPLISTENER_H diff --git a/src/base/kernel/interfaces/IJsonReader.h b/src/base/kernel/interfaces/IJsonReader.h new file mode 100644 index 000000000..c0fe09cb5 --- /dev/null +++ b/src/base/kernel/interfaces/IJsonReader.h @@ -0,0 +1,56 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_IJSONREADER_H +#define XMRIG_IJSONREADER_H + + +#include "rapidjson/fwd.h" + + +namespace xmrig { + + +class IJsonReader +{ +public: + virtual ~IJsonReader() = default; + + virtual bool getBool(const char *key, bool defaultValue = false) const = 0; + virtual bool isEmpty() const = 0; + virtual const char *getString(const char *key, const char *defaultValue = nullptr) const = 0; + virtual const rapidjson::Value &getArray(const char *key) const = 0; + virtual const rapidjson::Value &getObject(const char *key) const = 0; + virtual const rapidjson::Value &getValue(const char *key) const = 0; + virtual int getInt(const char *key, int defaultValue = 0) const = 0; + virtual int64_t getInt64(const char *key, int64_t defaultValue = 0) const = 0; + virtual uint64_t getUint64(const char *key, uint64_t defaultValue = 0) const = 0; + virtual unsigned getUint(const char *key, unsigned defaultValue = 0) const = 0; +}; + + +} /* namespace xmrig */ + + +#endif // XMRIG_IJSONREADER_H diff --git a/src/common/api/HttpReply.h b/src/base/kernel/interfaces/ILineListener.h similarity index 77% rename from src/common/api/HttpReply.h rename to src/base/kernel/interfaces/ILineListener.h index 6a6cb8027..1a6d49142 100644 --- a/src/common/api/HttpReply.h +++ b/src/base/kernel/interfaces/ILineListener.h @@ -5,8 +5,8 @@ * Copyright 2014-2016 Wolf9466 * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -22,8 +22,8 @@ * along with this program. If not, see . */ -#ifndef __HTTPREPLY_H__ -#define __HTTPREPLY_H__ +#ifndef XMRIG_ILINELISTENER_H +#define XMRIG_ILINELISTENER_H #include @@ -32,22 +32,19 @@ namespace xmrig { -class HttpReply +class String; + + +class ILineListener { public: - HttpReply() : - buf(nullptr), - status(200), - size(0) - {} + virtual ~ILineListener() = default; - char *buf; - int status; - size_t size; + virtual void onLine(char *line, size_t size) = 0; }; } /* namespace xmrig */ -#endif /* __HTTPREPLY_H__ */ +#endif // XMRIG_ILINELISTENER_H diff --git a/src/base/kernel/interfaces/ILogBackend.h b/src/base/kernel/interfaces/ILogBackend.h new file mode 100644 index 000000000..ef18da884 --- /dev/null +++ b/src/base/kernel/interfaces/ILogBackend.h @@ -0,0 +1,49 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2019 Spudz76 + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_ILOGBACKEND_H +#define XMRIG_ILOGBACKEND_H + + +#include +#include + + +namespace xmrig { + + +class ILogBackend +{ +public: + virtual ~ILogBackend() = default; + + virtual void print(int level, const char *line, size_t offset, size_t size, bool colors) = 0; +}; + + +} /* namespace xmrig */ + + +#endif // XMRIG_ILOGBACKEND_H diff --git a/src/common/interfaces/IStrategy.h b/src/base/kernel/interfaces/IStrategy.h similarity index 96% rename from src/common/interfaces/IStrategy.h rename to src/base/kernel/interfaces/IStrategy.h index 1e9c19bc4..f2e584083 100644 --- a/src/common/interfaces/IStrategy.h +++ b/src/base/kernel/interfaces/IStrategy.h @@ -29,13 +29,12 @@ #include -class JobResult; - - namespace xmrig { class Algorithm; +class IClient; +class JobResult; class IStrategy @@ -44,6 +43,7 @@ public: virtual ~IStrategy() = default; virtual bool isActive() const = 0; + virtual IClient *client() const = 0; virtual int64_t submit(const JobResult &result) = 0; virtual void connect() = 0; virtual void resume() = 0; diff --git a/src/common/interfaces/IStrategyListener.h b/src/base/kernel/interfaces/IStrategyListener.h similarity index 68% rename from src/common/interfaces/IStrategyListener.h rename to src/base/kernel/interfaces/IStrategyListener.h index cbec7742e..8b88b5068 100644 --- a/src/common/interfaces/IStrategyListener.h +++ b/src/base/kernel/interfaces/IStrategyListener.h @@ -26,13 +26,14 @@ #define XMRIG_ISTRATEGYLISTENER_H -#include +#include "rapidjson/fwd.h" namespace xmrig { -class Client; +class Algorithm; +class IClient; class IStrategy; class Job; class SubmitResult; @@ -43,10 +44,12 @@ class IStrategyListener public: virtual ~IStrategyListener() = default; - virtual void onActive(IStrategy *strategy, Client *client) = 0; - virtual void onJob(IStrategy *strategy, Client *client, const Job &job) = 0; - virtual void onPause(IStrategy *strategy) = 0; - virtual void onResultAccepted(IStrategy *strategy, Client *client, const SubmitResult &result, const char *error) = 0; + virtual void onActive(IStrategy *strategy, IClient *client) = 0; + virtual void onJob(IStrategy *strategy, IClient *client, const Job &job) = 0; + virtual void onLogin(IStrategy *strategy, IClient *client, rapidjson::Document &doc, rapidjson::Value ¶ms) = 0; + virtual void onPause(IStrategy *strategy) = 0; + virtual void onResultAccepted(IStrategy *strategy, IClient *client, const SubmitResult &result, const char *error) = 0; + virtual void onVerifyAlgorithm(IStrategy *strategy, const IClient *client, const Algorithm &algorithm, bool *ok) = 0; }; diff --git a/src/base/kernel/interfaces/ITcpServerListener.h b/src/base/kernel/interfaces/ITcpServerListener.h new file mode 100644 index 000000000..3cb1576e9 --- /dev/null +++ b/src/base/kernel/interfaces/ITcpServerListener.h @@ -0,0 +1,53 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_ITCPSERVERLISTENER_H +#define XMRIG_ITCPSERVERLISTENER_H + + +#include + + +typedef struct uv_stream_s uv_stream_t; + + +namespace xmrig { + + +class String; + + +class ITcpServerListener +{ +public: + virtual ~ITcpServerListener() = default; + + virtual void onConnection(uv_stream_t *stream, uint16_t port) = 0; +}; + + +} /* namespace xmrig */ + + +#endif // XMRIG_ITCPSERVERLISTENER_H diff --git a/src/base/kernel/interfaces/ITimerListener.h b/src/base/kernel/interfaces/ITimerListener.h new file mode 100644 index 000000000..1b6337d63 --- /dev/null +++ b/src/base/kernel/interfaces/ITimerListener.h @@ -0,0 +1,47 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_ITIMERLISTENER_H +#define XMRIG_ITIMERLISTENER_H + + +namespace xmrig { + + +class Timer; + + +class ITimerListener +{ +public: + virtual ~ITimerListener() = default; + + virtual void onTimer(const Timer *timer) = 0; +}; + + +} /* namespace xmrig */ + + +#endif // XMRIG_ITIMERLISTENER_H diff --git a/src/base/net/Pool.cpp b/src/base/net/Pool.cpp deleted file mode 100644 index 9d4f2bde6..000000000 --- a/src/base/net/Pool.cpp +++ /dev/null @@ -1,505 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2019 XMR-Stak , - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - - -#include -#include -#include -#include - - -#include "base/io/Json.h" -#include "base/net/Pool.h" -#include "rapidjson/document.h" - - -#ifdef APP_DEBUG -# include "common/log/Log.h" -#endif - - -#ifdef _MSC_VER -# define strncasecmp _strnicmp -# define strcasecmp _stricmp -#endif - - -namespace xmrig { - -static const char *kEnabled = "enabled"; -static const char *kFingerprint = "tls-fingerprint"; -static const char *kKeepalive = "keepalive"; -static const char *kNicehash = "nicehash"; -static const char *kPass = "pass"; -static const char *kRigId = "rig-id"; -static const char *kTls = "tls"; -static const char *kUrl = "url"; -static const char *kUser = "user"; -static const char *kVariant = "variant"; - -} - - -xmrig::Pool::Pool() : - m_enabled(true), - m_nicehash(false), - m_tls(false), - m_keepAlive(0), - m_port(kDefaultPort) -{ -} - - -/** - * @brief Parse url. - * - * Valid urls: - * example.com - * example.com:3333 - * stratum+tcp://example.com - * stratum+tcp://example.com:3333 - * - * @param url - */ -xmrig::Pool::Pool(const char *url) : - m_enabled(true), - m_nicehash(false), - m_tls(false), - m_keepAlive(0), - m_port(kDefaultPort) -{ - parse(url); -} - - -xmrig::Pool::Pool(const rapidjson::Value &object) : - m_enabled(true), - m_nicehash(false), - m_tls(false), - m_keepAlive(0), - m_port(kDefaultPort) -{ - if (!parse(Json::getString(object, kUrl))) { - return; - } - - setUser(Json::getString(object, kUser)); - setPassword(Json::getString(object, kPass)); - setRigId(Json::getString(object, kRigId)); - setNicehash(Json::getBool(object, kNicehash)); - - const rapidjson::Value &keepalive = object[kKeepalive]; - if (keepalive.IsInt()) { - setKeepAlive(keepalive.GetInt()); - } - else if (keepalive.IsBool()) { - setKeepAlive(keepalive.GetBool()); - } - - const rapidjson::Value &variant = object[kVariant]; - if (variant.IsString()) { - algorithm().parseVariant(variant.GetString()); - } - else if (variant.IsInt()) { - algorithm().parseVariant(variant.GetInt()); - } - - m_enabled = Json::getBool(object, kEnabled, true); - m_tls = Json::getBool(object, kTls); - m_fingerprint = Json::getString(object, kFingerprint); -} - - -xmrig::Pool::Pool(const char *host, uint16_t port, const char *user, const char *password, int keepAlive, bool nicehash, bool tls) : - m_enabled(true), - m_nicehash(nicehash), - m_tls(tls), - m_keepAlive(keepAlive), - m_host(host), - m_password(password), - m_user(user), - m_port(port) -{ - const size_t size = m_host.size() + 8; - assert(size > 8); - - char *url = new char[size](); - snprintf(url, size - 1, "%s:%d", m_host.data(), m_port); - - m_url = url; -} - - -bool xmrig::Pool::isCompatible(const Algorithm &algorithm) const -{ - if (m_algorithms.empty()) { - return true; - } - - for (const auto &a : m_algorithms) { - if (algorithm == a) { - return true; - } - } - -# ifdef XMRIG_PROXY_PROJECT - if (m_algorithm.algo() == xmrig::CRYPTONIGHT && algorithm.algo() == xmrig::CRYPTONIGHT) { - return m_algorithm.variant() == xmrig::VARIANT_XTL || m_algorithm.variant() == xmrig::VARIANT_MSR; - } -# endif - - return false; -} - - -bool xmrig::Pool::isEnabled() const -{ -# ifdef XMRIG_NO_TLS - if (isTLS()) { - return false; - } -# endif - - return m_enabled && isValid() && algorithm().isValid(); -} - - -bool xmrig::Pool::isEqual(const Pool &other) const -{ - return (m_nicehash == other.m_nicehash - && m_enabled == other.m_enabled - && m_tls == other.m_tls - && m_keepAlive == other.m_keepAlive - && m_port == other.m_port - && m_algorithm == other.m_algorithm - && m_fingerprint == other.m_fingerprint - && m_host == other.m_host - && m_password == other.m_password - && m_rigId == other.m_rigId - && m_url == other.m_url - && m_user == other.m_user); -} - - -bool xmrig::Pool::parse(const char *url) -{ - assert(url != nullptr); - - const char *p = strstr(url, "://"); - const char *base = url; - - if (p) { - if (strncasecmp(url, "stratum+tcp://", 14) == 0) { - m_tls = false; - } - else if (strncasecmp(url, "stratum+ssl://", 14) == 0) { - m_tls = true; - } - else { - return false; - } - - base = url + 14; - } - - if (!strlen(base) || *base == '/') { - return false; - } - - m_url = url; - if (base[0] == '[') { - return parseIPv6(base); - } - - const char *port = strchr(base, ':'); - if (!port) { - m_host = base; - return true; - } - - const size_t size = static_cast(port++ - base + 1); - char *host = new char[size](); - memcpy(host, base, size - 1); - - m_host = host; - m_port = static_cast(strtol(port, nullptr, 10)); - - return true; -} - - -bool xmrig::Pool::setUserpass(const char *userpass) -{ - const char *p = strchr(userpass, ':'); - if (!p) { - return false; - } - - char *user = new char[p - userpass + 1](); - strncpy(user, userpass, static_cast(p - userpass)); - - m_user = user; - m_password = p + 1; - - return true; -} - - -rapidjson::Value xmrig::Pool::toJSON(rapidjson::Document &doc) const -{ - using namespace rapidjson; - - auto &allocator = doc.GetAllocator(); - - Value obj(kObjectType); - - obj.AddMember(StringRef(kUrl), m_url.toJSON(), allocator); - obj.AddMember(StringRef(kUser), m_user.toJSON(), allocator); - obj.AddMember(StringRef(kPass), m_password.toJSON(), allocator); - obj.AddMember(StringRef(kRigId), m_rigId.toJSON(), allocator); - -# ifndef XMRIG_PROXY_PROJECT - obj.AddMember(StringRef(kNicehash), isNicehash(), allocator); -# endif - - if (m_keepAlive == 0 || m_keepAlive == kKeepAliveTimeout) { - obj.AddMember(StringRef(kKeepalive), m_keepAlive > 0, allocator); - } - else { - obj.AddMember(StringRef(kKeepalive), m_keepAlive, allocator); - } - - switch (m_algorithm.variant()) { - case VARIANT_AUTO: - case VARIANT_0: - case VARIANT_1: - obj.AddMember(StringRef(kVariant), m_algorithm.variant(), allocator); - break; - - case VARIANT_2: - obj.AddMember(StringRef(kVariant), 2, allocator); - break; - - default: - obj.AddMember(StringRef(kVariant), StringRef(m_algorithm.variantName()), allocator); - break; - } - - obj.AddMember(StringRef(kEnabled), m_enabled, allocator); - obj.AddMember(StringRef(kTls), isTLS(), allocator); - obj.AddMember(StringRef(kFingerprint), m_fingerprint.toJSON(), allocator); - - return obj; -} - - -void xmrig::Pool::adjust(const Algorithm &algorithm) -{ - if (!isValid()) { - return; - } - - if (!m_algorithm.isValid()) { - m_algorithm.setAlgo(algorithm.algo()); - adjustVariant(algorithm.variant()); - } - - rebuild(); -} - - -void xmrig::Pool::setAlgo(const xmrig::Algorithm &algorithm) -{ - m_algorithm = algorithm; - - rebuild(); -} - - -#ifdef APP_DEBUG -void xmrig::Pool::print() const -{ - LOG_NOTICE("url: %s", m_url.data()); - LOG_DEBUG ("host: %s", m_host.data()); - LOG_DEBUG ("port: %d", static_cast(m_port)); - LOG_DEBUG ("user: %s", m_user.data()); - LOG_DEBUG ("pass: %s", m_password.data()); - LOG_DEBUG ("rig-id %s", m_rigId.data()); - LOG_DEBUG ("algo: %s", m_algorithm.name()); - LOG_DEBUG ("nicehash: %d", static_cast(m_nicehash)); - LOG_DEBUG ("keepAlive: %d", m_keepAlive); -} -#endif - - -bool xmrig::Pool::parseIPv6(const char *addr) -{ - const char *end = strchr(addr, ']'); - if (!end) { - return false; - } - - const char *port = strchr(end, ':'); - if (!port) { - return false; - } - - const size_t size = static_cast(end - addr); - char *host = new char[size](); - memcpy(host, addr + 1, size - 1); - - m_host = host; - m_port = static_cast(strtol(port + 1, nullptr, 10)); - - return true; -} - - -void xmrig::Pool::addVariant(xmrig::Variant variant) -{ - const xmrig::Algorithm algorithm(m_algorithm.algo(), variant); - if (!algorithm.isValid() || m_algorithm == algorithm) { - return; - } - - m_algorithms.push_back(algorithm); -} - - -void xmrig::Pool::adjustVariant(const xmrig::Variant variantHint) -{ -# ifndef XMRIG_PROXY_PROJECT - using namespace xmrig; - - if (m_host.contains(".nicehash.com")) { - m_keepAlive = false; - m_nicehash = true; - bool valid = true; - - switch (m_port) { - case 3355: - case 33355: - valid = m_algorithm.algo() == CRYPTONIGHT && m_host.contains("cryptonight."); - m_algorithm.setVariant(VARIANT_0); - break; - - case 3363: - case 33363: - valid = m_algorithm.algo() == CRYPTONIGHT && m_host.contains("cryptonightv7."); - m_algorithm.setVariant(VARIANT_1); - break; - - case 3364: - valid = m_algorithm.algo() == CRYPTONIGHT_HEAVY && m_host.contains("cryptonightheavy."); - m_algorithm.setVariant(VARIANT_0); - break; - - case 3367: - case 33367: - valid = m_algorithm.algo() == CRYPTONIGHT && m_host.contains("cryptonightv8."); - m_algorithm.setVariant(VARIANT_2); - break; - - default: - break; - } - - if (!valid) { - m_algorithm.setAlgo(INVALID_ALGO); - } - - m_tls = m_port > 33000; - return; - } - - if (m_host.contains(".minergate.com")) { - m_keepAlive = false; - bool valid = true; - m_algorithm.setVariant(VARIANT_1); - - if (m_host.contains("xmr.pool.")) { - valid = m_algorithm.algo() == CRYPTONIGHT; - m_algorithm.setVariant(m_port == 45700 ? VARIANT_AUTO : VARIANT_0); - } - else if (m_host.contains("aeon.pool.") && m_port == 45690) { - valid = m_algorithm.algo() == CRYPTONIGHT_LITE; - m_algorithm.setVariant(VARIANT_1); - } - - if (!valid) { - m_algorithm.setAlgo(INVALID_ALGO); - } - - return; - } - - if (variantHint != VARIANT_AUTO) { - m_algorithm.setVariant(variantHint); - return; - } - - if (m_algorithm.variant() != VARIANT_AUTO) { - return; - } - - if (m_algorithm.algo() == CRYPTONIGHT_HEAVY) { - m_algorithm.setVariant(VARIANT_0); - } - else if (m_algorithm.algo() == CRYPTONIGHT_LITE) { - m_algorithm.setVariant(VARIANT_1); - } -# endif -} - - -void xmrig::Pool::rebuild() -{ - m_algorithms.clear(); - - if (!m_algorithm.isValid()) { - return; - } - - m_algorithms.push_back(m_algorithm); - -# ifndef XMRIG_PROXY_PROJECT - addVariant(VARIANT_4); - addVariant(VARIANT_WOW); - addVariant(VARIANT_2); - addVariant(VARIANT_1); - addVariant(VARIANT_0); - addVariant(VARIANT_HALF); - addVariant(VARIANT_XTL); - addVariant(VARIANT_TUBE); - addVariant(VARIANT_MSR); - addVariant(VARIANT_XHV); - addVariant(VARIANT_XAO); - addVariant(VARIANT_RTO); - addVariant(VARIANT_GPU); - addVariant(VARIANT_RWZ); - addVariant(VARIANT_ZLS); - addVariant(VARIANT_DOUBLE); - addVariant(VARIANT_AUTO); -# endif -} diff --git a/src/base/net/dns/Dns.cpp b/src/base/net/dns/Dns.cpp new file mode 100644 index 000000000..40d2a6e33 --- /dev/null +++ b/src/base/net/dns/Dns.cpp @@ -0,0 +1,163 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "base/kernel/interfaces/IDnsListener.h" +#include "base/net/dns/Dns.h" +#include "base/tools/Handle.h" + + +namespace xmrig { + Storage Dns::m_storage; + static const DnsRecord defaultRecord; +} + + +xmrig::Dns::Dns(IDnsListener *listener) : + m_hints(), + m_listener(listener), + m_status(0), + m_resolver(nullptr) +{ + m_key = m_storage.add(this); + + m_resolver = new uv_getaddrinfo_t; + m_resolver->data = m_storage.ptr(m_key); + + m_hints.ai_family = AF_UNSPEC; + m_hints.ai_socktype = SOCK_STREAM; + m_hints.ai_protocol = IPPROTO_TCP; +} + + +xmrig::Dns::~Dns() +{ + m_storage.release(m_key); + + delete m_resolver; +} + + +bool xmrig::Dns::resolve(const String &host) +{ + if (m_host != host) { + m_host = host; + + clear(); + } + + m_status = uv_getaddrinfo(uv_default_loop(), m_resolver, Dns::onResolved, m_host.data(), nullptr, &m_hints); + + return m_status == 0; +} + + +const char *xmrig::Dns::error() const +{ + return uv_strerror(m_status); +} + + +const xmrig::DnsRecord &xmrig::Dns::get(DnsRecord::Type prefered) const +{ + if (count() == 0) { + return defaultRecord; + } + + const size_t ipv4 = m_ipv4.size(); + const size_t ipv6 = m_ipv6.size(); + + if (ipv6 && (prefered == DnsRecord::AAAA || !ipv4)) { + return m_ipv6[ipv6 == 1 ? 0 : static_cast(rand()) % ipv6]; + } + + if (ipv4) { + return m_ipv4[ipv4 == 1 ? 0 : static_cast(rand()) % ipv4]; + } + + return defaultRecord; +} + + +size_t xmrig::Dns::count(DnsRecord::Type type) const +{ + if (type == DnsRecord::A) { + return m_ipv4.size(); + } + + if (type == DnsRecord::AAAA) { + return m_ipv6.size(); + } + + return m_ipv4.size() + m_ipv6.size(); +} + + +void xmrig::Dns::clear() +{ + m_ipv4.clear(); + m_ipv6.clear(); +} + + +void xmrig::Dns::onResolved(int status, addrinfo *res) +{ + m_status = status; + + if (m_status < 0) { + return m_listener->onResolved(*this, status); + } + + clear(); + + addrinfo *ptr = res; + while (ptr != nullptr) { + if (ptr->ai_family == AF_INET) { + m_ipv4.push_back(ptr); + } + + if (ptr->ai_family == AF_INET6) { + m_ipv6.push_back(ptr); + } + + ptr = ptr->ai_next; + } + + if (isEmpty()) { + m_status = UV_EAI_NONAME; + } + + m_listener->onResolved(*this, m_status); +} + + +void xmrig::Dns::onResolved(uv_getaddrinfo_t *req, int status, addrinfo *res) +{ + Dns *dns = m_storage.get(req->data); + if (dns) { + dns->onResolved(status, res); + } + + uv_freeaddrinfo(res); +} diff --git a/src/base/net/dns/Dns.h b/src/base/net/dns/Dns.h new file mode 100644 index 000000000..11f5bf80a --- /dev/null +++ b/src/base/net/dns/Dns.h @@ -0,0 +1,81 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_DNS_H +#define XMRIG_DNS_H + + +#include +#include + + +#include "base/net/dns/DnsRecord.h" +#include "base/net/tools/Storage.h" +#include "base/tools/String.h" + + +namespace xmrig { + + +class IDnsListener; + + +class Dns +{ +public: + Dns(IDnsListener *listener); + ~Dns(); + + inline bool isEmpty() const { return m_ipv4.empty() && m_ipv6.empty(); } + inline const String &host() const { return m_host; } + inline int status() const { return m_status; } + + bool resolve(const String &host); + const char *error() const; + const DnsRecord &get(DnsRecord::Type prefered = DnsRecord::A) const; + size_t count(DnsRecord::Type type = DnsRecord::Unknown) const; + +private: + void clear(); + void onResolved(int status, addrinfo *res); + + static void onResolved(uv_getaddrinfo_t *req, int status, addrinfo *res); + + addrinfo m_hints; + IDnsListener *m_listener; + int m_status; + std::vector m_ipv4; + std::vector m_ipv6; + String m_host; + uintptr_t m_key; + uv_getaddrinfo_t *m_resolver; + + static Storage m_storage; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_DNS_H */ diff --git a/src/base/net/dns/DnsRecord.cpp b/src/base/net/dns/DnsRecord.cpp new file mode 100644 index 000000000..7a85799b2 --- /dev/null +++ b/src/base/net/dns/DnsRecord.cpp @@ -0,0 +1,66 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include + + +#include "base/net/dns/DnsRecord.h" + + +xmrig::DnsRecord::DnsRecord(const addrinfo *addr) : + m_type(addr->ai_family == AF_INET6 ? AAAA : A) +{ + char *buf = nullptr; + + if (m_type == AAAA) { + buf = new char[45](); + uv_ip6_name(reinterpret_cast(addr->ai_addr), buf, 45); + } + else { + buf = new char[16](); + uv_ip4_name(reinterpret_cast(addr->ai_addr), buf, 16); + } + + m_ip = buf; +} + + +sockaddr *xmrig::DnsRecord::addr(uint16_t port) const +{ + if (m_type == A) { + sockaddr_in *addr = new sockaddr_in(); + uv_ip4_addr(m_ip.data(), port, addr); + + return reinterpret_cast(addr); + } + else if (m_type == AAAA) { + sockaddr_in6 *addr = new sockaddr_in6(); + uv_ip6_addr(m_ip.data(), port, addr); + + return reinterpret_cast(addr); + } + + return nullptr; +} diff --git a/src/common/config/ConfigWatcher.h b/src/base/net/dns/DnsRecord.h similarity index 70% rename from src/common/config/ConfigWatcher.h rename to src/base/net/dns/DnsRecord.h index c2c3ee29b..71d1e9c84 100644 --- a/src/common/config/ConfigWatcher.h +++ b/src/base/net/dns/DnsRecord.h @@ -22,42 +22,45 @@ * along with this program. If not, see . */ -#ifndef XMRIG_CONFIGWATCHER_H -#define XMRIG_CONFIGWATCHER_H +#ifndef XMRIG_DNSRECORD_H +#define XMRIG_DNSRECORD_H + + +struct addrinfo; +struct sockaddr; -#include "base/kernel/interfaces/IWatcherListener.h" #include "base/tools/String.h" -#include "rapidjson/fwd.h" - - -struct option; namespace xmrig { -class IConfigCreator; -class IConfigListener; -class Watcher; - - -class ConfigWatcher : public IWatcherListener +class DnsRecord { public: - ConfigWatcher(const String &path, IConfigCreator *creator, IConfigListener *listener); - ~ConfigWatcher() override; + enum Type { + Unknown, + A, + AAAA + }; -protected: - void onFileChanged(const String &fileName) override; + inline DnsRecord() : m_type(Unknown) {} + DnsRecord(const addrinfo *addr); + + sockaddr *addr(uint16_t port = 0) const; + + inline bool isValid() const { return m_type != Unknown; } + inline const String &ip() const { return m_ip; } + inline Type type() const { return m_type; } private: - IConfigCreator *m_creator; - IConfigListener *m_listener; - Watcher *m_watcher; + Type m_type; + String m_ip; }; } /* namespace xmrig */ -#endif /* __CONFIGWATCHER_H__ */ + +#endif /* XMRIG_DNSRECORD_H */ diff --git a/src/base/net/http/Http.cpp b/src/base/net/http/Http.cpp new file mode 100644 index 000000000..3c275824d --- /dev/null +++ b/src/base/net/http/Http.cpp @@ -0,0 +1,99 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "3rdparty/rapidjson/document.h" +#include "base/io/json/Json.h" +#include "base/net/http/Http.h" + + +namespace xmrig { + +static const char *kEnabled = "enabled"; +static const char *kHost = "host"; +static const char *kLocalhost = "127.0.0.1"; +static const char *kPort = "port"; +static const char *kRestricted = "restricted"; +static const char *kToken = "access-token"; + +} + + +xmrig::Http::Http() : + m_enabled(false), + m_restricted(true), + m_host(kLocalhost), + m_port(0) +{ +} + + +bool xmrig::Http::isEqual(const Http &other) const +{ + return other.m_enabled == m_enabled && + other.m_restricted == m_restricted && + other.m_host == m_host && + other.m_token == m_token && + other.m_port == m_port; +} + + +rapidjson::Value xmrig::Http::toJSON(rapidjson::Document &doc) const +{ + using namespace rapidjson; + auto &allocator = doc.GetAllocator(); + + Value obj(kObjectType); + + obj.AddMember(StringRef(kEnabled), m_enabled, allocator); + obj.AddMember(StringRef(kHost), m_host.toJSON(), allocator); + obj.AddMember(StringRef(kPort), m_port, allocator); + obj.AddMember(StringRef(kToken), m_token.toJSON(), allocator); + obj.AddMember(StringRef(kRestricted), m_restricted, allocator); + + return obj; +} + + +void xmrig::Http::load(const rapidjson::Value &http) +{ + if (!http.IsObject()) { + return; + } + + m_enabled = Json::getBool(http, kEnabled); + m_restricted = Json::getBool(http, kRestricted, true); + m_host = Json::getString(http, kHost, kLocalhost); + m_token = Json::getString(http, kToken); + + setPort(Json::getInt(http, kPort)); +} + + +void xmrig::Http::setPort(int port) +{ + if (port >= 0 && port <= 65536) { + m_port = static_cast(port); + } +} diff --git a/src/base/net/http/Http.h b/src/base/net/http/Http.h new file mode 100644 index 000000000..21eb581a0 --- /dev/null +++ b/src/base/net/http/Http.h @@ -0,0 +1,73 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#ifndef XMRIG_HTTP_H +#define XMRIG_HTTP_H + + +#include "base/tools/String.h" + + +namespace xmrig { + + +class Http +{ +public: + Http(); + + inline bool isAuthRequired() const { return m_restricted == false || !m_token.isNull(); } + inline bool isEnabled() const { return m_enabled; } + inline bool isRestricted() const { return m_restricted; } + inline const String &host() const { return m_host; } + inline const String &token() const { return m_token; } + inline uint16_t port() const { return m_port; } + inline void setEnabled(bool enabled) { m_enabled = enabled; } + inline void setHost(const char *host) { m_host = host; } + inline void setRestricted(bool restricted) { m_restricted = restricted; } + inline void setToken(const char *token) { m_token = token; } + + inline bool operator!=(const Http &other) const { return !isEqual(other); } + inline bool operator==(const Http &other) const { return isEqual(other); } + + bool isEqual(const Http &other) const; + rapidjson::Value toJSON(rapidjson::Document &doc) const; + void load(const rapidjson::Value &http); + void setPort(int port); + +private: + bool m_enabled; + bool m_restricted; + String m_host; + String m_token; + uint16_t m_port; +}; + + +} // namespace xmrig + + +#endif // XMRIG_HTTP_H + diff --git a/src/base/net/http/HttpApiResponse.cpp b/src/base/net/http/HttpApiResponse.cpp new file mode 100644 index 000000000..5fe926368 --- /dev/null +++ b/src/base/net/http/HttpApiResponse.cpp @@ -0,0 +1,88 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2014-2019 heapwolf + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "3rdparty/http-parser/http_parser.h" +#include "base/net/http/HttpApiResponse.h" +#include "rapidjson/prettywriter.h" +#include "rapidjson/stringbuffer.h" + + +namespace xmrig { + +static const char *kError = "error"; +static const char *kStatus = "status"; + +} // namespace xmrig + + +xmrig::HttpApiResponse::HttpApiResponse(uint64_t id) : + HttpResponse(id), + m_doc(rapidjson::kObjectType) +{ +} + + +xmrig::HttpApiResponse::HttpApiResponse(uint64_t id, int status) : + HttpResponse(id), + m_doc(rapidjson::kObjectType) +{ + setStatus(status); +} + + +void xmrig::HttpApiResponse::end() +{ + using namespace rapidjson; + + setHeader("Access-Control-Allow-Origin", "*"); + setHeader("Access-Control-Allow-Methods", "GET, PUT, POST, DELETE"); + setHeader("Access-Control-Allow-Headers", "Authorization, Content-Type"); + + if (statusCode() >= 400) { + if (!m_doc.HasMember(kStatus)) { + m_doc.AddMember(StringRef(kStatus), statusCode(), m_doc.GetAllocator()); + } + + if (!m_doc.HasMember(kError)) { + m_doc.AddMember(StringRef(kError), StringRef(http_status_str(static_cast(statusCode()))), m_doc.GetAllocator()); + } + } + + if (!m_doc.MemberCount()) { + return HttpResponse::end(); + } + + setHeader("Content-Type", "application/json"); + + StringBuffer buffer(nullptr, 4096); + PrettyWriter writer(buffer); + writer.SetMaxDecimalPlaces(10); + writer.SetFormatOptions(kFormatSingleLineArray); + + m_doc.Accept(writer); + + HttpResponse::end(buffer.GetString(), buffer.GetSize()); +} diff --git a/src/base/net/http/HttpApiResponse.h b/src/base/net/http/HttpApiResponse.h new file mode 100644 index 000000000..bbaf132ed --- /dev/null +++ b/src/base/net/http/HttpApiResponse.h @@ -0,0 +1,57 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2014-2019 heapwolf + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#ifndef XMRIG_HTTPAPIRESPONSE_H +#define XMRIG_HTTPAPIRESPONSE_H + + +#include "base/net/http/HttpResponse.h" +#include "rapidjson/document.h" + + +namespace xmrig { + + +class HttpApiResponse : public HttpResponse +{ +public: + HttpApiResponse(uint64_t id); + HttpApiResponse(uint64_t id, int status); + + inline rapidjson::Document &doc() { return m_doc; } + + void end(); + +private: + rapidjson::Document m_doc; +}; + + +} // namespace xmrig + + +#endif // XMRIG_HTTPAPIRESPONSE_H + diff --git a/src/base/net/http/HttpClient.cpp b/src/base/net/http/HttpClient.cpp new file mode 100644 index 000000000..113e2f139 --- /dev/null +++ b/src/base/net/http/HttpClient.cpp @@ -0,0 +1,224 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2014-2019 heapwolf + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include + + +#include "3rdparty/http-parser/http_parser.h" +#include "base/io/log/Log.h" +#include "base/kernel/Platform.h" +#include "base/net/dns/Dns.h" +#include "base/net/http/HttpClient.h" +#include "base/tools/Baton.h" + + +namespace xmrig { + +static const char *kCRLF = "\r\n"; + + +class ClientWriteBaton : public Baton +{ +public: + inline ClientWriteBaton(const std::string &header, std::string &&body) : + m_body(std::move(body)), + m_header(header) + { + bufs[0].len = m_header.size(); + bufs[0].base = const_cast(m_header.c_str()); + + if (!m_body.empty()) { + bufs[1].len = m_body.size(); + bufs[1].base = const_cast(m_body.c_str()); + } + else { + bufs[1].base = nullptr; + bufs[1].len = 0; + } + } + + + inline size_t count() const { return bufs[1].base == nullptr ? 1 : 2; } + inline size_t size() const { return bufs[0].len + bufs[1].len; } + inline static void onWrite(uv_write_t *req, int) { delete reinterpret_cast(req->data); } + + + uv_buf_t bufs[2]; + +private: + std::string m_body; + std::string m_header; +}; + + +} // namespace xmrig + + +xmrig::HttpClient::HttpClient(int method, const String &url, IHttpListener *listener, const char *data, size_t size) : + HttpContext(HTTP_RESPONSE, listener), + m_quiet(false), + m_port(0) +{ + this->method = method; + this->url = url; + + if (data) { + body = size ? std::string(data, size) : data; + } + + m_dns = new Dns(this); +} + + +xmrig::HttpClient::~HttpClient() +{ + delete m_dns; +} + + +bool xmrig::HttpClient::connect(const String &host, uint16_t port) +{ + m_port = port; + + return m_dns->resolve(host); +} + + +const xmrig::String &xmrig::HttpClient::host() const +{ + return m_dns->host(); +} + + +void xmrig::HttpClient::onResolved(const Dns &dns, int status) +{ + this->status = status; + + if (status < 0 && dns.isEmpty()) { + if (!m_quiet) { + LOG_ERR("[%s:%d] DNS error: \"%s\"", dns.host().data(), m_port, uv_strerror(status)); + } + + return; + } + + sockaddr *addr = dns.get().addr(m_port); + + uv_connect_t *req = new uv_connect_t; + req->data = this; + + uv_tcp_connect(req, m_tcp, addr, onConnect); +} + + +void xmrig::HttpClient::handshake() +{ + headers.insert({ "Host", m_dns->host().data() }); + headers.insert({ "Connection", "close" }); + headers.insert({ "User-Agent", Platform::userAgent() }); + + if (body.size()) { + headers.insert({ "Content-Length", std::to_string(body.size()) }); + } + + std::stringstream ss; + ss << http_method_str(static_cast(method)) << " " << url << " HTTP/1.1" << kCRLF; + + for (auto &header : headers) { + ss << header.first << ": " << header.second << kCRLF; + } + + ss << kCRLF; + + headers.clear(); + + write(ss.str()); +} + + +void xmrig::HttpClient::read(const char *data, size_t size) +{ + if (parse(data, size) < size) { + close(UV_EPROTO); + } +} + + +void xmrig::HttpClient::write(const std::string &header) +{ + ClientWriteBaton *baton = new ClientWriteBaton(header, std::move(body)); + uv_write(&baton->req, stream(), baton->bufs, baton->count(), ClientWriteBaton::onWrite); +} + + +void xmrig::HttpClient::onConnect(uv_connect_t *req, int status) +{ + HttpClient *client = static_cast(req->data); + if (!client) { + delete req; + return; + } + + if (status < 0) { + if (!client->m_quiet) { + LOG_ERR("[%s:%d] connect error: \"%s\"", client->m_dns->host().data(), client->m_port, uv_strerror(status)); + } + + delete req; + client->close(status); + return; + } + + uv_read_start(client->stream(), + [](uv_handle_t *, size_t suggested_size, uv_buf_t *buf) + { + buf->base = new char[suggested_size]; + +# ifdef _WIN32 + buf->len = static_cast(suggested_size); +# else + buf->len = suggested_size; +# endif + }, + [](uv_stream_t *tcp, ssize_t nread, const uv_buf_t *buf) + { + HttpClient *client = static_cast(tcp->data); + + if (nread >= 0) { + client->read(buf->base, static_cast(nread)); + } else { + if (!client->m_quiet && nread != UV_EOF) { + LOG_ERR("[%s:%d] read error: \"%s\"", client->m_dns->host().data(), client->m_port, uv_strerror(static_cast(nread))); + } + + client->close(static_cast(nread)); + } + + delete [] buf->base; + }); + + client->handshake(); +} diff --git a/src/base/net/http/HttpClient.h b/src/base/net/http/HttpClient.h new file mode 100644 index 000000000..c5dfc43d9 --- /dev/null +++ b/src/base/net/http/HttpClient.h @@ -0,0 +1,74 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2014-2019 heapwolf + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#ifndef XMRIG_HTTPCLIENT_H +#define XMRIG_HTTPCLIENT_H + + +#include "base/net/http/HttpContext.h" +#include "base/kernel/interfaces/IDnsListener.h" + + +namespace xmrig { + + +class String; + + +class HttpClient : public HttpContext, public IDnsListener +{ +public: + HttpClient(int method, const String &url, IHttpListener *listener, const char *data = nullptr, size_t size = 0); + ~HttpClient() override; + + inline uint16_t port() const { return m_port; } + inline void setQuiet(bool quiet) { m_quiet = quiet; } + + bool connect(const String &host, uint16_t port); + const String &host() const; + +protected: + void onResolved(const Dns &dns, int status) override; + + virtual void handshake(); + virtual void read(const char *data, size_t size); + virtual void write(const std::string &header); + + bool m_quiet; + +private: + static void onConnect(uv_connect_t *req, int status); + + Dns *m_dns; + uint16_t m_port; +}; + + +} // namespace xmrig + + +#endif // XMRIG_HTTPCLIENT_H + diff --git a/src/base/net/http/HttpContext.cpp b/src/base/net/http/HttpContext.cpp new file mode 100644 index 000000000..e97f989b2 --- /dev/null +++ b/src/base/net/http/HttpContext.cpp @@ -0,0 +1,228 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2014-2019 heapwolf + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include +#include + + +#include "3rdparty/http-parser/http_parser.h" +#include "base/kernel/interfaces/IHttpListener.h" +#include "base/net/http/HttpContext.h" + + +namespace xmrig { + +static http_parser_settings http_settings; +static std::map storage; +static uint64_t SEQUENCE = 0; + +} // namespace xmrig + + +xmrig::HttpContext::HttpContext(int parser_type, IHttpListener *listener) : + HttpData(SEQUENCE++), + m_wasHeaderValue(false), + m_listener(listener) +{ + storage[id()] = this; + + m_parser = new http_parser; + m_tcp = new uv_tcp_t; + + uv_tcp_init(uv_default_loop(), m_tcp); + uv_tcp_nodelay(m_tcp, 1); + + http_parser_init(m_parser, static_cast(parser_type)); + + m_parser->data = m_tcp->data = this; + + if (http_settings.on_message_complete == nullptr) { + attach(&http_settings); + } +} + + +xmrig::HttpContext::~HttpContext() +{ + delete m_tcp; + delete m_parser; +} + + +size_t xmrig::HttpContext::parse(const char *data, size_t size) +{ + return http_parser_execute(m_parser, &http_settings, data, size); +} + + +std::string xmrig::HttpContext::ip() const +{ + char ip[46] = {}; + sockaddr_storage addr = {}; + int size = sizeof(addr); + + uv_tcp_getpeername(m_tcp, reinterpret_cast(&addr), &size); + if (reinterpret_cast(&addr)->sin_family == AF_INET6) { + uv_ip6_name(reinterpret_cast(&addr), ip, 45); + } + else { + uv_ip4_name(reinterpret_cast(&addr), ip, 16); + } + + return ip; +} + + +void xmrig::HttpContext::close(int status) +{ + if (status < 0 && m_listener) { + this->status = status; + m_listener->onHttpData(*this); + } + + auto it = storage.find(id()); + if (it != storage.end()) { + storage.erase(it); + } + + if (!uv_is_closing(handle())) { + uv_close(handle(), [](uv_handle_t *handle) -> void { delete reinterpret_cast(handle->data); }); + } +} + + +xmrig::HttpContext *xmrig::HttpContext::get(uint64_t id) +{ + if (storage.count(id) == 0) { + return nullptr; + } + + return storage[id]; +} + + +void xmrig::HttpContext::closeAll() +{ + for (auto kv : storage) { + if (!uv_is_closing(kv.second->handle())) { + uv_close(kv.second->handle(), [](uv_handle_t *handle) -> void { delete reinterpret_cast(handle->data); }); + } + } +} + + +int xmrig::HttpContext::onHeaderField(http_parser *parser, const char *at, size_t length) +{ + HttpContext *ctx = static_cast(parser->data); + + if (ctx->m_wasHeaderValue) { + if (!ctx->m_lastHeaderField.empty()) { + ctx->setHeader(); + } + + ctx->m_lastHeaderField = std::string(at, length); + ctx->m_wasHeaderValue = false; + } else { + ctx->m_lastHeaderField += std::string(at, length); + } + + return 0; +} + + +int xmrig::HttpContext::onHeaderValue(http_parser *parser, const char *at, size_t length) +{ + HttpContext *ctx = static_cast(parser->data); + + if (!ctx->m_wasHeaderValue) { + ctx->m_lastHeaderValue = std::string(at, length); + ctx->m_wasHeaderValue = true; + } else { + ctx->m_lastHeaderValue += std::string(at, length); + } + + return 0; +} + + +void xmrig::HttpContext::attach(http_parser_settings *settings) +{ + settings->on_message_begin = nullptr; + settings->on_status = nullptr; + settings->on_chunk_header = nullptr; + settings->on_chunk_complete = nullptr; + + settings->on_url = [](http_parser *parser, const char *at, size_t length) -> int + { + static_cast(parser->data)->url = std::string(at, length); + return 0; + }; + + settings->on_header_field = onHeaderField; + settings->on_header_value = onHeaderValue; + + settings->on_headers_complete = [](http_parser* parser) -> int { + HttpContext *ctx = static_cast(parser->data); + ctx->status = parser->status_code; + + if (parser->type == HTTP_REQUEST) { + ctx->method = parser->method; + } + + if (!ctx->m_lastHeaderField.empty()) { + ctx->setHeader(); + } + + return 0; + }; + + settings->on_body = [](http_parser *parser, const char *at, size_t len) -> int + { + static_cast(parser->data)->body += std::string(at, len); + + return 0; + }; + + settings->on_message_complete = [](http_parser *parser) -> int + { + HttpContext *ctx = static_cast(parser->data); + ctx->m_listener->onHttpData(*ctx); + ctx->m_listener = nullptr; + + return 0; + }; +} + + +void xmrig::HttpContext::setHeader() +{ + std::transform(m_lastHeaderField.begin(), m_lastHeaderField.end(), m_lastHeaderField.begin(), ::tolower); + headers.insert({ m_lastHeaderField, m_lastHeaderValue }); + + m_lastHeaderField.clear(); + m_lastHeaderValue.clear(); +} + diff --git a/src/base/net/http/HttpContext.h b/src/base/net/http/HttpContext.h new file mode 100644 index 000000000..fbb453aa9 --- /dev/null +++ b/src/base/net/http/HttpContext.h @@ -0,0 +1,86 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2014-2019 heapwolf + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#ifndef XMRIG_HTTPCONTEXT_H +#define XMRIG_HTTPCONTEXT_H + + +typedef struct http_parser http_parser; +typedef struct http_parser_settings http_parser_settings; +typedef struct uv_connect_s uv_connect_t; +typedef struct uv_handle_s uv_handle_t; +typedef struct uv_stream_s uv_stream_t; +typedef struct uv_tcp_s uv_tcp_t; + + +#include "base/net/http/HttpData.h" + + +namespace xmrig { + + +class IHttpListener; + + +class HttpContext : public HttpData +{ +public: + HttpContext(int parser_type, IHttpListener *listener); + virtual ~HttpContext(); + + inline uv_stream_t *stream() const { return reinterpret_cast(m_tcp); } + inline uv_handle_t *handle() const { return reinterpret_cast(m_tcp); } + + size_t parse(const char *data, size_t size); + std::string ip() const; + void close(int status = 0); + + static HttpContext *get(uint64_t id); + static void closeAll(); + +protected: + uv_tcp_t *m_tcp; + +private: + static int onHeaderField(http_parser *parser, const char *at, size_t length); + static int onHeaderValue(http_parser *parser, const char *at, size_t length); + static void attach(http_parser_settings *settings); + + void setHeader(); + + bool m_wasHeaderValue; + http_parser *m_parser; + IHttpListener *m_listener; + std::string m_lastHeaderField; + std::string m_lastHeaderValue; +}; + + +} // namespace xmrig + + +#endif // XMRIG_HTTPCONTEXT_H + diff --git a/src/base/net/http/HttpData.h b/src/base/net/http/HttpData.h new file mode 100644 index 000000000..ceb19b8e4 --- /dev/null +++ b/src/base/net/http/HttpData.h @@ -0,0 +1,60 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2014-2019 heapwolf + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#ifndef XMRIG_HTTPDATA_H +#define XMRIG_HTTPDATA_H + + +#include +#include + + +namespace xmrig { + + +class HttpData +{ +public: + inline HttpData(uint64_t id) : method(0), status(0), m_id(id) {} + + inline uint64_t id() const { return m_id; } + + int method; + int status; + std::map headers; + std::string body; + std::string url; + +private: + const uint64_t m_id; +}; + + +} // namespace xmrig + + +#endif // XMRIG_HTTPDATA_H + diff --git a/src/base/net/http/HttpResponse.cpp b/src/base/net/http/HttpResponse.cpp new file mode 100644 index 000000000..7a4af8387 --- /dev/null +++ b/src/base/net/http/HttpResponse.cpp @@ -0,0 +1,153 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2014-2019 heapwolf + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include +#include +#include + + +#include "3rdparty/http-parser/http_parser.h" +#include "base/io/log/Log.h" +#include "base/net/http/HttpContext.h" +#include "base/net/http/HttpResponse.h" +#include "base/tools/Baton.h" + + +namespace xmrig { + +static const char *kCRLF = "\r\n"; +static const char *kUserAgent = "user-agent"; + + +class WriteBaton : public Baton +{ +public: + inline WriteBaton(const std::stringstream &ss, const char *data, size_t size, HttpContext *ctx) : + m_ctx(ctx), + m_header(ss.str()) + { + bufs[0].len = m_header.size(); + bufs[0].base = const_cast(m_header.c_str()); + + if (data) { + bufs[1].len = size; + bufs[1].base = new char[size]; + memcpy(bufs[1].base, data, size); + } + else { + bufs[1].base = nullptr; + bufs[1].len = 0; + } + } + + + inline ~WriteBaton() + { + if (count() == 2) { + delete [] bufs[1].base; + } + + m_ctx->close(); + } + + + inline size_t count() const { return bufs[1].base == nullptr ? 1 : 2; } + inline size_t size() const { return bufs[0].len + bufs[1].len; } + inline static void onWrite(uv_write_t *req, int) { delete reinterpret_cast(req->data); } + + + uv_buf_t bufs[2]; + +private: + HttpContext *m_ctx; + std::string m_header; +}; + +} // namespace xmrig + + +xmrig::HttpResponse::HttpResponse(uint64_t id, int statusCode) : + m_id(id), + m_statusCode(statusCode) +{ +} + + +bool xmrig::HttpResponse::isAlive() const +{ + HttpContext *ctx = HttpContext::get(m_id); + + return ctx && uv_is_writable(ctx->stream()); +} + + +void xmrig::HttpResponse::end(const char *data, size_t size) +{ + if (!isAlive()) { + return; + } + + if (data && !size) { + size = strlen(data); + } + + if (size) { + setHeader("Content-Length", std::to_string(size)); + } + + setHeader("Connection", "close"); + + std::stringstream ss; + ss << "HTTP/1.1 " << statusCode() << " " << http_status_str(static_cast(statusCode())) << kCRLF; + + for (auto &header : m_headers) { + ss << header.first << ": " << header.second << kCRLF; + } + + ss << kCRLF; + + HttpContext *ctx = HttpContext::get(m_id); + WriteBaton *baton = new WriteBaton(ss, data, size, ctx); + +# ifndef APP_DEBUG + if (statusCode() >= 400) +# endif + { + const bool err = statusCode() >= 400; + + Log::print(err ? Log::ERR : Log::INFO, CYAN("%s ") CLEAR MAGENTA_BOLD("%s") WHITE_BOLD(" %s ") CSI "1;%dm%d " CLEAR WHITE_BOLD("%zu ") BLACK_BOLD("\"%s\""), + ctx->ip().c_str(), + http_method_str(static_cast(ctx->method)), + ctx->url.c_str(), + err ? 31 : 32, + statusCode(), + baton->size(), + ctx->headers.count(kUserAgent) ? ctx->headers.at(kUserAgent).c_str() : nullptr + ); + } + + uv_write(&baton->req, ctx->stream(), baton->bufs, baton->count(), WriteBaton::onWrite); +} diff --git a/src/base/net/http/HttpResponse.h b/src/base/net/http/HttpResponse.h new file mode 100644 index 000000000..c2bc90011 --- /dev/null +++ b/src/base/net/http/HttpResponse.h @@ -0,0 +1,61 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2014-2019 heapwolf + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#ifndef XMRIG_HTTPRESPONSE_H +#define XMRIG_HTTPRESPONSE_H + + +#include +#include + + +namespace xmrig { + + +class HttpResponse +{ +public: + HttpResponse(uint64_t id, int statusCode = 200); + + inline int statusCode() const { return m_statusCode; } + inline void setHeader(const std::string &key, const std::string &value) { m_headers.insert({ key, value }); } + inline void setStatus(int code) { m_statusCode = code; } + + bool isAlive() const; + void end(const char *data = nullptr, size_t size = 0); + +private: + const uint64_t m_id; + int m_statusCode; + std::map m_headers; +}; + + +} // namespace xmrig + + +#endif // XMRIG_HTTPRESPONSE_H + diff --git a/src/base/net/http/HttpServer.cpp b/src/base/net/http/HttpServer.cpp new file mode 100644 index 000000000..60db31f62 --- /dev/null +++ b/src/base/net/http/HttpServer.cpp @@ -0,0 +1,83 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2014-2019 heapwolf + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include +#include + + +#include "3rdparty/http-parser/http_parser.h" +#include "base/kernel/interfaces/IHttpListener.h" +#include "base/net/http/HttpContext.h" +#include "base/net/http/HttpResponse.h" +#include "base/net/http/HttpServer.h" + + +xmrig::HttpServer::HttpServer(IHttpListener *listener) : + m_listener(listener) +{ +} + + +xmrig::HttpServer::~HttpServer() +{ + HttpContext::closeAll(); +} + + +void xmrig::HttpServer::onConnection(uv_stream_t *stream, uint16_t) +{ + HttpContext *ctx = new HttpContext(HTTP_REQUEST, m_listener); + uv_accept(stream, ctx->stream()); + + uv_read_start(ctx->stream(), + [](uv_handle_t *, size_t suggested_size, uv_buf_t *buf) + { + buf->base = new char[suggested_size]; + +# ifdef _WIN32 + buf->len = static_cast(suggested_size); +# else + buf->len = suggested_size; +# endif + }, + [](uv_stream_t *tcp, ssize_t nread, const uv_buf_t *buf) + { + HttpContext *ctx = static_cast(tcp->data); + + if (nread >= 0) { + const size_t size = static_cast(nread); + const size_t parsed = ctx->parse(buf->base, size); + + if (parsed < size) { + ctx->close(); + } + } else { + ctx->close(); + } + + delete [] buf->base; + }); +} diff --git a/src/base/net/http/HttpServer.h b/src/base/net/http/HttpServer.h new file mode 100644 index 000000000..45f3843d1 --- /dev/null +++ b/src/base/net/http/HttpServer.h @@ -0,0 +1,62 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2014-2019 heapwolf + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#ifndef XMRIG_HTTPSERVER_H +#define XMRIG_HTTPSERVER_H + + +typedef struct http_parser http_parser; +typedef struct http_parser_settings http_parser_settings; + + +#include "base/kernel/interfaces/ITcpServerListener.h" + + +namespace xmrig { + + +class IHttpListener; + + +class HttpServer : public ITcpServerListener +{ +public: + HttpServer(IHttpListener *listener); + ~HttpServer() override; + +protected: + void onConnection(uv_stream_t *stream, uint16_t port) override; + +private: + IHttpListener *m_listener; +}; + + +} // namespace xmrig + + +#endif // XMRIG_HTTPSERVER_H + diff --git a/src/base/net/http/HttpsClient.cpp b/src/base/net/http/HttpsClient.cpp new file mode 100644 index 000000000..2c2873309 --- /dev/null +++ b/src/base/net/http/HttpsClient.cpp @@ -0,0 +1,208 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2014-2019 heapwolf + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include +#include +#include + + +#include "base/io/log/Log.h" +#include "base/net/http/HttpsClient.h" +#include "base/tools/Buffer.h" + + +#ifdef _MSC_VER +# define strncasecmp(x,y,z) _strnicmp(x,y,z) +#endif + + +xmrig::HttpsClient::HttpsClient(int method, const String &url, IHttpListener *listener, const char *data, size_t size, const String &fingerprint) : + HttpClient(method, url, listener, data, size), + m_ready(false), + m_buf(), + m_ssl(nullptr), + m_fp(fingerprint) +{ + m_ctx = SSL_CTX_new(SSLv23_method()); + assert(m_ctx != nullptr); + + if (!m_ctx) { + return; + } + + m_writeBio = BIO_new(BIO_s_mem()); + m_readBio = BIO_new(BIO_s_mem()); + SSL_CTX_set_options(m_ctx, SSL_OP_NO_SSLv2 | SSL_OP_NO_SSLv3); +} + + +xmrig::HttpsClient::~HttpsClient() +{ + if (m_ctx) { + SSL_CTX_free(m_ctx); + } + + if (m_ssl) { + SSL_free(m_ssl); + } +} + + +const char *xmrig::HttpsClient::fingerprint() const +{ + return m_ready ? m_fingerprint : nullptr; +} + + +const char *xmrig::HttpsClient::version() const +{ + return m_ready ? SSL_get_version(m_ssl) : nullptr; +} + + +void xmrig::HttpsClient::handshake() +{ + m_ssl = SSL_new(m_ctx); + assert(m_ssl != nullptr); + + if (!m_ssl) { + return; + } + + SSL_set_connect_state(m_ssl); + SSL_set_bio(m_ssl, m_readBio, m_writeBio); + SSL_set_tlsext_host_name(m_ssl, host().data()); + + SSL_do_handshake(m_ssl); + + flush(); +} + + +void xmrig::HttpsClient::read(const char *data, size_t size) +{ + BIO_write(m_readBio, data, size); + + if (!SSL_is_init_finished(m_ssl)) { + const int rc = SSL_connect(m_ssl); + + if (rc < 0 && SSL_get_error(m_ssl, rc) == SSL_ERROR_WANT_READ) { + flush(); + } else if (rc == 1) { + X509 *cert = SSL_get_peer_certificate(m_ssl); + if (!verify(cert)) { + X509_free(cert); + return close(UV_EPROTO); + } + + X509_free(cert); + m_ready = true; + + HttpClient::handshake(); + } + + return; + } + + int bytes_read = 0; + while ((bytes_read = SSL_read(m_ssl, m_buf, sizeof(m_buf))) > 0) { + HttpClient::read(m_buf, static_cast(bytes_read)); + } +} + + +void xmrig::HttpsClient::write(const std::string &header) +{ + SSL_write(m_ssl, (header + body).c_str(), header.size() + body.size()); + body.clear(); + + flush(); +} + + +bool xmrig::HttpsClient::verify(X509 *cert) +{ + if (cert == nullptr) { + return false; + } + + if (!verifyFingerprint(cert)) { + if (!m_quiet) { + LOG_ERR("[%s:%d] Failed to verify server certificate fingerprint", host().data(), port()); + + if (strlen(m_fingerprint) == 64 && !m_fp.isNull()) { + LOG_ERR("\"%s\" was given", m_fingerprint); + LOG_ERR("\"%s\" was configured", m_fp.data()); + } + } + + return false; + } + + return true; +} + + +bool xmrig::HttpsClient::verifyFingerprint(X509 *cert) +{ + const EVP_MD *digest = EVP_get_digestbyname("sha256"); + if (digest == nullptr) { + return false; + } + + unsigned char md[EVP_MAX_MD_SIZE]; + unsigned int dlen; + + if (X509_digest(cert, digest, md, &dlen) != 1) { + return false; + } + + Buffer::toHex(md, 32, m_fingerprint); + + return m_fp.isNull() || strncasecmp(m_fingerprint, m_fp.data(), 64) == 0; +} + + +void xmrig::HttpsClient::flush() +{ + uv_buf_t buf; + buf.len = BIO_get_mem_data(m_writeBio, &buf.base); + + if (buf.len == 0) { + return; + } + + bool result = false; + if (uv_is_writable(stream())) { + result = uv_try_write(stream(), &buf, 1) == static_cast(buf.len); + + if (!result) { + close(UV_EIO); + } + } + + (void) BIO_reset(m_writeBio); +} diff --git a/src/base/net/http/HttpsClient.h b/src/base/net/http/HttpsClient.h new file mode 100644 index 000000000..c6a228099 --- /dev/null +++ b/src/base/net/http/HttpsClient.h @@ -0,0 +1,77 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2014-2019 heapwolf + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#ifndef XMRIG_HTTPSCLIENT_H +#define XMRIG_HTTPSCLIENT_H + + +typedef struct bio_st BIO; +typedef struct ssl_ctx_st SSL_CTX; +typedef struct ssl_st SSL; +typedef struct x509_st X509; + + +#include "base/net/http/HttpClient.h" +#include "base/tools/String.h" + + +namespace xmrig { + + +class HttpsClient : public HttpClient +{ +public: + HttpsClient(int method, const String &url, IHttpListener *listener, const char *data, size_t size, const String &fingerprint); + ~HttpsClient() override; + + const char *fingerprint() const; + const char *version() const; + +protected: + void handshake() override; + void read(const char *data, size_t size) override; + void write(const std::string &header) override; + +private: + bool verify(X509 *cert); + bool verifyFingerprint(X509 *cert); + void flush(); + + BIO *m_readBio; + BIO *m_writeBio; + bool m_ready; + char m_buf[1024 * 2]; + char m_fingerprint[32 * 2 + 8]; + SSL *m_ssl; + SSL_CTX *m_ctx; + String m_fp; +}; + + +} // namespace xmrig + + +#endif // XMRIG_HTTPSCLIENT_H diff --git a/src/base/net/stratum/BaseClient.cpp b/src/base/net/stratum/BaseClient.cpp new file mode 100644 index 000000000..f44415d58 --- /dev/null +++ b/src/base/net/stratum/BaseClient.cpp @@ -0,0 +1,62 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "base/kernel/interfaces/IClientListener.h" +#include "base/net/stratum/BaseClient.h" +#include "base/net/stratum/SubmitResult.h" + + +namespace xmrig { + +int64_t BaseClient::m_sequence = 1; + +} /* namespace xmrig */ + + +xmrig::BaseClient::BaseClient(int id, IClientListener *listener) : + m_quiet(false), + m_listener(listener), + m_id(id), + m_retries(5), + m_failures(0), + m_state(UnconnectedState), + m_retryPause(5000) +{ +} + + +bool xmrig::BaseClient::handleSubmitResponse(int64_t id, const char *error) +{ + auto it = m_results.find(id); + if (it != m_results.end()) { + it->second.done(); + m_listener->onResultAccepted(this, it->second, error); + m_results.erase(it); + + return true; + } + + return false; +} diff --git a/src/base/net/stratum/BaseClient.h b/src/base/net/stratum/BaseClient.h new file mode 100644 index 000000000..9e1c7ffbc --- /dev/null +++ b/src/base/net/stratum/BaseClient.h @@ -0,0 +1,96 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_BASECLIENT_H +#define XMRIG_BASECLIENT_H + + +#include + + +#include "base/kernel/interfaces/IClient.h" +#include "base/net/stratum/Job.h" +#include "base/net/stratum/Pool.h" + + +namespace xmrig { + + +class IClientListener; +class SubmitResult; + + +class BaseClient : public IClient +{ +public: + BaseClient(int id, IClientListener *listener); + + inline bool isEnabled() const override { return m_enabled; } + inline const Job &job() const override { return m_job; } + inline const Pool &pool() const override { return m_pool; } + inline const String &ip() const override { return m_ip; } + inline int id() const override { return m_id; } + inline void setAlgo(const Algorithm &algo) override { m_pool.setAlgo(algo); } + inline void setEnabled(bool enabled) override { m_enabled = enabled; } + inline void setPool(const Pool &pool) override { if (pool.isValid()) { m_pool = pool; } } + inline void setQuiet(bool quiet) override { m_quiet = quiet; } + inline void setRetries(int retries) override { m_retries = retries; } + inline void setRetryPause(uint64_t ms) override { m_retryPause = ms; } + +protected: + enum SocketState { + UnconnectedState, + HostLookupState, + ConnectingState, + ConnectedState, + ClosingState + }; + + inline bool isQuiet() const { return m_quiet || m_failures >= m_retries; } + + bool handleSubmitResponse(int64_t id, const char *error = nullptr); + + bool m_quiet; + IClientListener *m_listener; + int m_id; + int m_retries; + int64_t m_failures; + Job m_job; + Pool m_pool; + SocketState m_state; + std::map m_results; + String m_ip; + uint64_t m_retryPause; + + static int64_t m_sequence; + +private: + bool m_enabled; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_BASECLIENT_H */ diff --git a/src/common/net/Client.cpp b/src/base/net/stratum/Client.cpp similarity index 55% rename from src/common/net/Client.cpp rename to src/base/net/stratum/Client.cpp index 7cd09d46b..fe1823369 100644 --- a/src/common/net/Client.cpp +++ b/src/base/net/stratum/Client.cpp @@ -30,16 +30,21 @@ #include -#ifndef XMRIG_NO_TLS +#ifdef XMRIG_FEATURE_TLS # include # include -# include "common/net/Tls.h" +# include "base/net/stratum/Tls.h" #endif -#include "common/interfaces/IClientListener.h" -#include "common/log/Log.h" -#include "common/net/Client.h" +#include "base/io/json/Json.h" +#include "base/io/json/JsonRequest.h" +#include "base/io/log/Log.h" +#include "base/kernel/interfaces/IClientListener.h" +#include "base/net/dns/Dns.h" +#include "base/net/stratum/Client.h" +#include "base/tools/Buffer.h" +#include "base/tools/Chrono.h" #include "net/JobResult.h" #include "rapidjson/document.h" #include "rapidjson/error/en.h" @@ -54,7 +59,6 @@ namespace xmrig { -int64_t Client::m_sequence = 1; Storage Client::m_storage; } /* namespace xmrig */ @@ -72,18 +76,8 @@ static const char *states[] = { xmrig::Client::Client(int id, const char *agent, IClientListener *listener) : - m_ipv6(false), - m_nicehash(false), - m_quiet(false), + BaseClient(id, listener), m_agent(agent), - m_listener(listener), - m_extensions(0), - m_id(id), - m_retries(5), - m_retryPause(5000), - m_failures(0), - m_recvBufPos(0), - m_state(UnconnectedState), m_tls(nullptr), m_expire(0), m_jobs(0), @@ -93,94 +87,17 @@ xmrig::Client::Client(int id, const char *agent, IClientListener *listener) : m_socket(nullptr) { m_key = m_storage.add(this); - - memset(m_ip, 0, sizeof(m_ip)); - memset(&m_hints, 0, sizeof(m_hints)); - - m_resolver.data = m_storage.ptr(m_key); - - m_hints.ai_family = AF_UNSPEC; - m_hints.ai_socktype = SOCK_STREAM; - m_hints.ai_protocol = IPPROTO_TCP; - - m_recvBuf.base = m_buf; - m_recvBuf.len = sizeof(m_buf); + m_dns = new Dns(this); } xmrig::Client::~Client() { + delete m_dns; delete m_socket; } -void xmrig::Client::connect() -{ -# ifndef XMRIG_NO_TLS - if (m_pool.isTLS()) { - m_tls = new Tls(this); - } -# endif - - resolve(m_pool.host()); -} - - -/** - * @brief Connect to server. - * - * @param url - */ -void xmrig::Client::connect(const Pool &url) -{ - setPool(url); - connect(); -} - - -void xmrig::Client::deleteLater() -{ - if (!m_listener) { - return; - } - - m_listener = nullptr; - - if (!disconnect()) { - m_storage.remove(m_key); - } -} - - - -void xmrig::Client::setPool(const Pool &pool) -{ - if (!pool.isValid()) { - return; - } - - m_pool = pool; -} - - -void xmrig::Client::tick(uint64_t now) -{ - if (m_state == ConnectedState) { - if (m_expire && now > m_expire) { - LOG_DEBUG_ERR("[%s] timeout", m_pool.url()); - close(); - } - else if (m_keepAlive && now > m_keepAlive) { - ping(); - } - } - - if (m_expire && now > m_expire && m_state == ConnectingState) { - connect(); - } -} - - bool xmrig::Client::disconnect() { m_keepAlive = 0; @@ -191,9 +108,19 @@ bool xmrig::Client::disconnect() } +bool xmrig::Client::isTLS() const +{ +# ifdef XMRIG_FEATURE_TLS + return m_pool.isTLS() && m_tls; +# else + return false; +# endif +} + + const char *xmrig::Client::tlsFingerprint() const { -# ifndef XMRIG_NO_TLS +# ifdef XMRIG_FEATURE_TLS if (isTLS() && m_pool.fingerprint() == nullptr) { return m_tls->fingerprint(); } @@ -205,7 +132,7 @@ const char *xmrig::Client::tlsFingerprint() const const char *xmrig::Client::tlsVersion() const { -# ifndef XMRIG_NO_TLS +# ifdef XMRIG_FEATURE_TLS if (isTLS()) { return m_tls->version(); } @@ -232,31 +159,27 @@ int64_t xmrig::Client::submit(const JobResult &result) char *nonce = m_sendBuf; char *data = m_sendBuf + 16; - Job::toHex(reinterpret_cast(&result.nonce), 4, nonce); + Buffer::toHex(reinterpret_cast(&result.nonce), 4, nonce); nonce[8] = '\0'; - Job::toHex(result.result, 32, data); + Buffer::toHex(result.result(), 32, data); data[64] = '\0'; # endif Document doc(kObjectType); auto &allocator = doc.GetAllocator(); - doc.AddMember("id", m_sequence, allocator); - doc.AddMember("jsonrpc", "2.0", allocator); - doc.AddMember("method", "submit", allocator); - Value params(kObjectType); params.AddMember("id", StringRef(m_rpcId.data()), allocator); params.AddMember("job_id", StringRef(result.jobId.data()), allocator); params.AddMember("nonce", StringRef(nonce), allocator); params.AddMember("result", StringRef(data), allocator); - if (m_extensions & AlgoExt) { + if (has() && result.algorithm.isValid()) { params.AddMember("algo", StringRef(result.algorithm.shortName()), allocator); } - doc.AddMember("params", params, allocator); + JsonRequest::create(doc, m_sequence, "submit", params); # ifdef XMRIG_PROXY_PROJECT m_results[m_sequence] = SubmitResult(m_sequence, result.diff, result.actualDiff(), result.id); @@ -268,6 +191,79 @@ int64_t xmrig::Client::submit(const JobResult &result) } +void xmrig::Client::connect() +{ +# ifdef XMRIG_FEATURE_TLS + if (m_pool.isTLS()) { + m_tls = new Tls(this); + } +# endif + + resolve(m_pool.host()); +} + + +void xmrig::Client::connect(const Pool &pool) +{ + setPool(pool); + connect(); +} + + +void xmrig::Client::deleteLater() +{ + if (!m_listener) { + return; + } + + m_listener = nullptr; + + if (!disconnect()) { + m_storage.remove(m_key); + } +} + + +void xmrig::Client::tick(uint64_t now) +{ + if (m_state == ConnectedState) { + if (m_expire && now > m_expire) { + LOG_DEBUG_ERR("[%s] timeout", url()); + close(); + } + else if (m_keepAlive && now > m_keepAlive) { + ping(); + } + } + + if (m_expire && now > m_expire && m_state == ConnectingState) { + connect(); + } +} + + +void xmrig::Client::onResolved(const Dns &dns, int status) +{ + assert(m_listener != nullptr); + if (!m_listener) { + return reconnect(); + } + + if (status < 0 && dns.isEmpty()) { + if (!isQuiet()) { + LOG_ERR("[%s] DNS error: \"%s\"", url(), uv_strerror(status)); + } + + return reconnect(); + } + + const DnsRecord &record = dns.get(); + m_ip = record.ip(); + + connect(record.addr(m_pool.port())); +} + + bool xmrig::Client::close() { if (m_state == ClosingState) { @@ -310,16 +306,6 @@ bool xmrig::Client::isCriticalError(const char *message) } -bool xmrig::Client::isTLS() const -{ -# ifndef XMRIG_NO_TLS - return m_pool.isTLS() && m_tls; -# else - return false; -# endif -} - - bool xmrig::Client::parseJob(const rapidjson::Value ¶ms, int *code) { if (!params.IsObject()) { @@ -327,7 +313,7 @@ bool xmrig::Client::parseJob(const rapidjson::Value ¶ms, int *code) return false; } - Job job(m_id, m_nicehash, m_pool.algorithm(), m_rpcId); + Job job(has(), m_pool.algorithm(), m_rpcId); if (!job.setId(params["job_id"].GetString())) { *code = 3; @@ -344,33 +330,24 @@ bool xmrig::Client::parseJob(const rapidjson::Value ¶ms, int *code) return false; } - if (params.HasMember("algo")) { - job.setAlgorithm(params["algo"].GetString()); + const char *algo = Json::getString(params, "algo"); + if (algo) { + job.setAlgorithm(algo); } - if (params.HasMember("variant")) { - const rapidjson::Value &variant = params["variant"]; + job.setHeight(Json::getUint64(params, "height")); - if (variant.IsInt()) { - job.setVariant(variant.GetInt()); - } - else if (variant.IsString()){ - job.setVariant(variant.GetString()); - } - } - - if (params.HasMember("height")) { - const rapidjson::Value &variant = params["height"]; - - if (variant.IsUint64()) { - job.setHeight(variant.GetUint64()); - } - } - - if (!verifyAlgorithm(job.algorithm())) { + if (!verifyAlgorithm(job.algorithm(), algo)) { *code = 6; + return false; + } - close(); + if (job.algorithm().family() == Algorithm::RANDOM_X && !job.setSeedHash(Json::getString(params, "seed_hash"))) { + if (!isQuiet()) { + LOG_ERR("[%s] failed to parse field \"seed_hash\" required by RandomX", url(), algo); + } + + *code = 7; return false; } @@ -387,7 +364,7 @@ bool xmrig::Client::parseJob(const rapidjson::Value ¶ms, int *code) } if (!isQuiet()) { - LOG_WARN("[%s] duplicate job received, reconnect", m_pool.url()); + LOG_WARN("[%s] duplicate job received, reconnect", url()); } close(); @@ -397,16 +374,13 @@ bool xmrig::Client::parseJob(const rapidjson::Value ¶ms, int *code) bool xmrig::Client::parseLogin(const rapidjson::Value &result, int *code) { - if (!m_rpcId.setId(result["id"].GetString())) { + m_rpcId = result["id"].GetString(); + if (m_rpcId.isNull()) { *code = 1; return false; } - m_nicehash = m_pool.isNicehash(); - - if (result.HasMember("extensions")) { - parseExtensions(result["extensions"]); - } + parseExtensions(result); const bool rc = parseJob(result["job"], code); m_jobs = 0; @@ -417,7 +391,7 @@ bool xmrig::Client::parseLogin(const rapidjson::Value &result, int *code) bool xmrig::Client::send(BIO *bio) { -# ifndef XMRIG_NO_TLS +# ifdef XMRIG_FEATURE_TLS uv_buf_t buf; buf.len = BIO_get_mem_data(bio, &buf.base); @@ -425,7 +399,7 @@ bool xmrig::Client::send(BIO *bio) return true; } - LOG_DEBUG("[%s] TLS send (%d bytes)", m_pool.url(), static_cast(buf.len)); + LOG_DEBUG("[%s] TLS send (%d bytes)", url(), static_cast(buf.len)); bool result = false; if (state() == ConnectedState && uv_is_writable(m_stream)) { @@ -436,7 +410,7 @@ bool xmrig::Client::send(BIO *bio) } } else { - LOG_DEBUG_ERR("[%s] send failed, invalid state: %d", m_pool.url(), m_state); + LOG_DEBUG_ERR("[%s] send failed, invalid state: %d", url(), m_state); } (void) BIO_reset(bio); @@ -448,49 +422,43 @@ bool xmrig::Client::send(BIO *bio) } -bool xmrig::Client::verifyAlgorithm(const Algorithm &algorithm) const +bool xmrig::Client::verifyAlgorithm(const Algorithm &algorithm, const char *algo) const { -# ifdef XMRIG_PROXY_PROJECT - if (m_pool.algorithm().variant() == VARIANT_AUTO || m_id == -1) { - return true; - } -# endif + if (!algorithm.isValid()) { + if (!isQuiet()) { + LOG_ERR("[%s] Unknown/unsupported algorithm \"%s\" detected, reconnect", url(), algo); + } - if (m_pool.isCompatible(algorithm)) { - return true; - } - - if (isQuiet()) { return false; } - if (algorithm.isValid()) { - LOG_ERR("Incompatible algorithm \"%s\" detected, reconnect", algorithm.name()); - } - else { - LOG_ERR("Unknown/unsupported algorithm detected, reconnect"); + bool ok = true; + m_listener->onVerifyAlgorithm(this, algorithm, &ok); + + if (!ok && !isQuiet()) { + LOG_ERR("[%s] Incompatible/disabled algorithm \"%s\" detected, reconnect", url(), algorithm.shortName()); } - return false; + return ok; } -int xmrig::Client::resolve(const char *host) +int xmrig::Client::resolve(const String &host) { setState(HostLookupState); - m_expire = 0; - m_recvBufPos = 0; + m_expire = 0; + m_recvBuf.reset(); if (m_failures == -1) { m_failures = 0; } - const int r = uv_getaddrinfo(uv_default_loop(), &m_resolver, Client::onResolved, host, nullptr, &m_hints); - if (r) { + if (!m_dns->resolve(host)) { if (!isQuiet()) { - LOG_ERR("[%s:%u] getaddrinfo error: \"%s\"", host, m_pool.port(), uv_strerror(r)); + LOG_ERR("[%s:%u] getaddrinfo error: \"%s\"", host.data(), m_pool.port(), uv_strerror(m_dns->status())); } + return 1; } @@ -502,13 +470,13 @@ int64_t xmrig::Client::send(const rapidjson::Document &doc) { using namespace rapidjson; - StringBuffer buffer(0, 512); + StringBuffer buffer(nullptr, 512); Writer writer(buffer); doc.Accept(writer); const size_t size = buffer.GetSize(); if (size > (sizeof(m_sendBuf) - 2)) { - LOG_ERR("[%s] send failed: \"send buffer overflow: %zu > %zu\"", m_pool.url(), size, (sizeof(m_sendBuf) - 2)); + LOG_ERR("[%s] send failed: \"send buffer overflow: %zu > %zu\"", url(), size, (sizeof(m_sendBuf) - 2)); close(); return -1; } @@ -523,9 +491,9 @@ int64_t xmrig::Client::send(const rapidjson::Document &doc) int64_t xmrig::Client::send(size_t size) { - LOG_DEBUG("[%s] send (%d bytes): \"%s\"", m_pool.url(), size, m_sendBuf); + LOG_DEBUG("[%s] send (%d bytes): \"%.*s\"", url(), size, static_cast(size) - 1, m_sendBuf); -# ifndef XMRIG_NO_TLS +# ifdef XMRIG_FEATURE_TLS if (isTLS()) { if (!m_tls->send(m_sendBuf, size)) { return -1; @@ -535,7 +503,7 @@ int64_t xmrig::Client::send(size_t size) # endif { if (state() != ConnectedState || !uv_is_writable(m_stream)) { - LOG_DEBUG_ERR("[%s] send failed, invalid state: %d", m_pool.url(), m_state); + LOG_DEBUG_ERR("[%s] send failed, invalid state: %d", url(), m_state); return -1; } @@ -547,35 +515,15 @@ int64_t xmrig::Client::send(size_t size) } } - m_expire = uv_now(uv_default_loop()) + kResponseTimeout; + m_expire = Chrono::steadyMSecs() + kResponseTimeout; return m_sequence++; } -void xmrig::Client::connect(const std::vector &ipv4, const std::vector &ipv6) -{ - addrinfo *addr = nullptr; - m_ipv6 = ipv4.empty() && !ipv6.empty(); - - if (m_ipv6) { - addr = ipv6[ipv6.size() == 1 ? 0 : rand() % ipv6.size()]; - uv_ip6_name(reinterpret_cast(addr->ai_addr), m_ip, 45); - } - else { - addr = ipv4[ipv4.size() == 1 ? 0 : rand() % ipv4.size()]; - uv_ip4_name(reinterpret_cast(addr->ai_addr), m_ip, 16); - } - - connect(addr->ai_addr); -} - - void xmrig::Client::connect(sockaddr *addr) { setState(ConnectingState); - reinterpret_cast(addr)->sin_port = htons(m_pool.port()); - uv_connect_t *req = new uv_connect_t; req->data = m_storage.ptr(m_key); @@ -589,15 +537,17 @@ void xmrig::Client::connect(sockaddr *addr) uv_tcp_keepalive(m_socket, 1, 60); # endif - uv_tcp_connect(req, m_socket, reinterpret_cast(addr), Client::onConnect); + uv_tcp_connect(req, m_socket, addr, onConnect); + + delete addr; } void xmrig::Client::handshake() { -# ifndef XMRIG_NO_TLS +# ifdef XMRIG_FEATURE_TLS if (isTLS()) { - m_expire = uv_now(uv_default_loop()) + kResponseTimeout; + m_expire = Chrono::steadyMSecs() + kResponseTimeout; m_tls->handshake(); } @@ -617,33 +567,18 @@ void xmrig::Client::login() Document doc(kObjectType); auto &allocator = doc.GetAllocator(); - doc.AddMember("id", 1, allocator); - doc.AddMember("jsonrpc", "2.0", allocator); - doc.AddMember("method", "login", allocator); - Value params(kObjectType); - params.AddMember("login", StringRef(m_pool.user()), allocator); - params.AddMember("pass", StringRef(m_pool.password()), allocator); - params.AddMember("agent", StringRef(m_agent), allocator); + params.AddMember("login", m_pool.user().toJSON(), allocator); + params.AddMember("pass", m_pool.password().toJSON(), allocator); + params.AddMember("agent", StringRef(m_agent), allocator); - if (m_pool.rigId()) { - params.AddMember("rigid", StringRef(m_pool.rigId()), allocator); + if (!m_pool.rigId().isNull()) { + params.AddMember("rigid", m_pool.rigId().toJSON(), allocator); } -# ifdef XMRIG_PROXY_PROJECT - if (m_pool.algorithm().variant() != xmrig::VARIANT_AUTO) -# endif - { - Value algo(kArrayType); + m_listener->onLogin(this, doc, params); - for (const auto &a : m_pool.algorithms()) { - algo.PushBack(StringRef(a.shortName()), allocator); - } - - params.AddMember("algo", algo, allocator); - } - - doc.AddMember("params", params, allocator); + JsonRequest::create(doc, 1, "login", params); send(doc); } @@ -657,7 +592,7 @@ void xmrig::Client::onClose() m_socket = nullptr; setState(UnconnectedState); -# ifndef XMRIG_NO_TLS +# ifdef XMRIG_FEATURE_TLS if (m_tls) { delete m_tls; m_tls = nullptr; @@ -672,13 +607,11 @@ void xmrig::Client::parse(char *line, size_t len) { startTimeout(); - line[len - 1] = '\0'; - - LOG_DEBUG("[%s] received (%d bytes): \"%s\"", m_pool.url(), len, line); + LOG_DEBUG("[%s] received (%d bytes): \"%.*s\"", url(), len, static_cast(len), line); if (len < 32 || line[0] != '{') { if (!isQuiet()) { - LOG_ERR("[%s] JSON decode failed", m_pool.url()); + LOG_ERR("[%s] JSON decode failed", url()); } return; @@ -687,7 +620,7 @@ void xmrig::Client::parse(char *line, size_t len) rapidjson::Document doc; if (doc.ParseInsitu(line).HasParseError()) { if (!isQuiet()) { - LOG_ERR("[%s] JSON decode failed: \"%s\"", m_pool.url(), rapidjson::GetParseError_En(doc.GetParseError())); + LOG_ERR("[%s] JSON decode failed: \"%s\"", url(), rapidjson::GetParseError_En(doc.GetParseError())); } return; @@ -707,29 +640,43 @@ void xmrig::Client::parse(char *line, size_t len) } -void xmrig::Client::parseExtensions(const rapidjson::Value &value) +void xmrig::Client::parseExtensions(const rapidjson::Value &result) { - m_extensions = 0; + m_extensions.reset(); - if (!value.IsArray()) { + if (!result.HasMember("extensions")) { return; } - for (const rapidjson::Value &ext : value.GetArray()) { + const rapidjson::Value &extensions = result["extensions"]; + if (!extensions.IsArray()) { + return; + } + + for (const rapidjson::Value &ext : extensions.GetArray()) { if (!ext.IsString()) { continue; } - if (strcmp(ext.GetString(), "algo") == 0) { - m_extensions |= AlgoExt; - continue; - } + const char *name = ext.GetString(); - if (strcmp(ext.GetString(), "nicehash") == 0) { - m_extensions |= NicehashExt; - m_nicehash = true; - continue; + if (strcmp(name, "algo") == 0) { + setExtension(EXT_ALGO, true); } + else if (strcmp(name, "nicehash") == 0) { + setExtension(EXT_NICEHASH, true); + } + else if (strcmp(name, "connect") == 0) { + setExtension(EXT_CONNECT, true); + } + else if (strcmp(name, "keepalive") == 0) { + setExtension(EXT_KEEPALIVE, true); + } +# ifdef XMRIG_FEATURE_TLS + else if (strcmp(name, "tls") == 0) { + setExtension(EXT_TLS, true); + } +# endif } } @@ -738,7 +685,7 @@ void xmrig::Client::parseNotification(const char *method, const rapidjson::Value { if (error.IsObject()) { if (!isQuiet()) { - LOG_ERR("[%s] error: \"%s\", code: %d", m_pool.url(), error["message"].GetString(), error["code"].GetInt()); + LOG_ERR("[%s] error: \"%s\", code: %d", url(), error["message"].GetString(), error["code"].GetInt()); } return; } @@ -750,13 +697,16 @@ void xmrig::Client::parseNotification(const char *method, const rapidjson::Value if (strcmp(method, "job") == 0) { int code = -1; if (parseJob(params, &code)) { - m_listener->onJobReceived(this, m_job); + m_listener->onJobReceived(this, m_job, params); + } + else { + close(); } return; } - LOG_WARN("[%s] unsupported method: \"%s\"", m_pool.url(), method); + LOG_WARN("[%s] unsupported method: \"%s\"", url(), method); } @@ -765,17 +715,11 @@ void xmrig::Client::parseResponse(int64_t id, const rapidjson::Value &result, co if (error.IsObject()) { const char *message = error["message"].GetString(); - auto it = m_results.find(id); - if (it != m_results.end()) { - it->second.done(); - m_listener->onResultAccepted(this, it->second, message); - m_results.erase(it); - } - else if (!isQuiet()) { - LOG_ERR("[%s] error: \"%s\", code: %d", m_pool.url(), message, error["code"].GetInt()); + if (!handleSubmitResponse(id, message) && !isQuiet()) { + LOG_ERR("[%s] error: " RED_BOLD("\"%s\"") RED_S ", code: %d", url(), message, error["code"].GetInt()); } - if (isCriticalError(message)) { + if (m_id == 1 || isCriticalError(message)) { close(); } @@ -790,7 +734,7 @@ void xmrig::Client::parseResponse(int64_t id, const rapidjson::Value &result, co int code = -1; if (!parseLogin(result, &code)) { if (!isQuiet()) { - LOG_ERR("[%s] login error code: %d", m_pool.url(), code); + LOG_ERR("[%s] login error code: %d", url(), code); } close(); @@ -799,16 +743,11 @@ void xmrig::Client::parseResponse(int64_t id, const rapidjson::Value &result, co m_failures = 0; m_listener->onLoginSuccess(this); - m_listener->onJobReceived(this, m_job); + m_listener->onJobReceived(this, m_job, result["job"]); return; } - auto it = m_results.find(id); - if (it != m_results.end()) { - it->second.done(); - m_listener->onResultAccepted(this, it->second, nullptr); - m_results.erase(it); - } + handleSubmitResponse(id); } @@ -818,32 +757,42 @@ void xmrig::Client::ping() } -void xmrig::Client::read() +void xmrig::Client::read(ssize_t nread) { - char* end; - char* start = m_recvBuf.base; - size_t remaining = m_recvBufPos; + const size_t size = static_cast(nread); - while ((end = static_cast(memchr(start, '\n', remaining))) != nullptr) { - end++; - size_t len = end - start; - parse(start, len); - - remaining -= len; - start = end; + if (nread > 0 && size > m_recvBuf.available()) { + nread = UV_ENOBUFS; } - if (remaining == 0) { - m_recvBufPos = 0; + if (nread < 0) { + if (!isQuiet()) { + LOG_ERR("[%s] read error: \"%s\"", url(), uv_strerror(static_cast(nread))); + } + + close(); return; } - if (start == m_recvBuf.base) { - return; + assert(m_listener != nullptr); + if (!m_listener) { + return reconnect(); } - memcpy(m_recvBuf.base, start, remaining); - m_recvBufPos = remaining; + m_recvBuf.nread(size); + +# ifdef XMRIG_FEATURE_TLS + if (isTLS()) { + LOG_DEBUG("[%s] TLS received (%d bytes)", url(), static_cast(nread)); + + m_tls->read(m_recvBuf.base(), m_recvBuf.pos()); + m_recvBuf.reset(); + } + else +# endif + { + m_recvBuf.getline(this); + } } @@ -864,15 +813,15 @@ void xmrig::Client::reconnect() setState(ConnectingState); m_failures++; - m_listener->onClose(this, (int) m_failures); + m_listener->onClose(this, static_cast(m_failures)); - m_expire = uv_now(uv_default_loop()) + m_retryPause; + m_expire = Chrono::steadyMSecs() + m_retryPause; } void xmrig::Client::setState(SocketState state) { - LOG_DEBUG("[%s] state: \"%s\"", m_pool.url(), states[state]); + LOG_DEBUG("[%s] state: \"%s\"", url(), states[state]); if (m_state == state) { return; @@ -886,21 +835,28 @@ void xmrig::Client::startTimeout() { m_expire = 0; - if (m_pool.keepAlive()) { - m_keepAlive = uv_now(uv_default_loop()) + (m_pool.keepAlive() * 1000); + if (has()) { + const uint64_t ms = static_cast(m_pool.keepAlive() > 0 ? m_pool.keepAlive() : Pool::kKeepAliveTimeout) * 1000; + + m_keepAlive = Chrono::steadyMSecs() + ms; } } -void xmrig::Client::onAllocBuffer(uv_handle_t *handle, size_t suggested_size, uv_buf_t *buf) +void xmrig::Client::onAllocBuffer(uv_handle_t *handle, size_t, uv_buf_t *buf) { auto client = getClient(handle->data); if (!client) { return; } - buf->base = &client->m_recvBuf.base[client->m_recvBufPos]; - buf->len = client->m_recvBuf.len - client->m_recvBufPos; + buf->base = client->m_recvBuf.current(); + +# ifdef _WIN32 + buf->len = static_cast(client->m_recvBuf.available()); +# else + buf->len = client->m_recvBuf.available(); +# endif } @@ -925,7 +881,7 @@ void xmrig::Client::onConnect(uv_connect_t *req, int status) if (status < 0) { if (!client->isQuiet()) { - LOG_ERR("[%s] connect error: \"%s\"", client->m_pool.url(), uv_strerror(status)); + LOG_ERR("[%s] connect error: \"%s\"", client->url(), uv_strerror(status)); } delete req; @@ -937,101 +893,17 @@ void xmrig::Client::onConnect(uv_connect_t *req, int status) client->m_stream->data = req->data; client->setState(ConnectedState); - uv_read_start(client->m_stream, Client::onAllocBuffer, Client::onRead); + uv_read_start(client->m_stream, onAllocBuffer, onRead); delete req; client->handshake(); } -void xmrig::Client::onRead(uv_stream_t *stream, ssize_t nread, const uv_buf_t *buf) +void xmrig::Client::onRead(uv_stream_t *stream, ssize_t nread, const uv_buf_t *) { auto client = getClient(stream->data); - if (!client) { - return; - } - - if (nread < 0) { - if (!client->isQuiet()) { - LOG_ERR("[%s] read error: \"%s\"", client->m_pool.url(), uv_strerror((int) nread)); - } - - client->close(); - return; - } - - if ((size_t) nread > (sizeof(m_buf) - 8 - client->m_recvBufPos)) { - client->close(); - return; - } - - assert(client->m_listener != nullptr); - if (!client->m_listener) { - return client->reconnect(); - } - - client->m_recvBufPos += nread; - -# ifndef XMRIG_NO_TLS - if (client->isTLS()) { - LOG_DEBUG("[%s] TLS received (%d bytes)", client->m_pool.url(), static_cast(nread)); - - client->m_tls->read(client->m_recvBuf.base, client->m_recvBufPos); - client->m_recvBufPos = 0; - } - else -# endif - { - client->read(); + if (client) { + client->read(nread); } } - - -void xmrig::Client::onResolved(uv_getaddrinfo_t *req, int status, struct addrinfo *res) -{ - auto client = getClient(req->data); - if (!client) { - return; - } - - assert(client->m_listener != nullptr); - if (!client->m_listener) { - return client->reconnect(); - } - - if (status < 0) { - if (!client->isQuiet()) { - LOG_ERR("[%s] DNS error: \"%s\"", client->m_pool.url(), uv_strerror(status)); - } - - return client->reconnect(); - } - - addrinfo *ptr = res; - std::vector ipv4; - std::vector ipv6; - - while (ptr != nullptr) { - if (ptr->ai_family == AF_INET) { - ipv4.push_back(ptr); - } - - if (ptr->ai_family == AF_INET6) { - ipv6.push_back(ptr); - } - - ptr = ptr->ai_next; - } - - if (ipv4.empty() && ipv6.empty()) { - if (!client->isQuiet()) { - LOG_ERR("[%s] DNS error: \"No IPv4 (A) or IPv6 (AAAA) records found\"", client->m_pool.url()); - } - - uv_freeaddrinfo(res); - return client->reconnect(); - } - - client->connect(ipv4, ipv6); - uv_freeaddrinfo(res); -} diff --git a/src/common/net/Client.h b/src/base/net/stratum/Client.h similarity index 54% rename from src/common/net/Client.h rename to src/base/net/stratum/Client.h index 9314739a2..46030abaf 100644 --- a/src/common/net/Client.h +++ b/src/base/net/stratum/Client.h @@ -26,18 +26,21 @@ #define XMRIG_CLIENT_H +#include #include #include #include -#include "base/net/Pool.h" -#include "common/crypto/Algorithm.h" -#include "common/net/Id.h" -#include "common/net/Job.h" -#include "common/net/Storage.h" -#include "common/net/SubmitResult.h" -#include "rapidjson/fwd.h" +#include "base/kernel/interfaces/IDnsListener.h" +#include "base/kernel/interfaces/ILineListener.h" +#include "base/net/stratum/BaseClient.h" +#include "base/net/stratum/Job.h" +#include "base/net/stratum/Pool.h" +#include "base/net/stratum/SubmitResult.h" +#include "base/net/tools/RecvBuf.h" +#include "base/net/tools/Storage.h" +#include "crypto/common/Algorithm.h" typedef struct bio_st BIO; @@ -50,129 +53,97 @@ class IClientListener; class JobResult; -class Client +class Client : public BaseClient, public IDnsListener, public ILineListener { public: - enum SocketState { - UnconnectedState, - HostLookupState, - ConnectingState, - ConnectedState, - ClosingState - }; - constexpr static int kResponseTimeout = 20 * 1000; -# ifndef XMRIG_NO_TLS +# ifdef XMRIG_FEATURE_TLS constexpr static int kInputBufferSize = 1024 * 16; # else constexpr static int kInputBufferSize = 1024 * 2; # endif Client(int id, const char *agent, IClientListener *listener); - ~Client(); + ~Client() override; - bool disconnect(); - const char *tlsFingerprint() const; - const char *tlsVersion() const; - int64_t submit(const JobResult &result); - void connect(); - void connect(const Pool &pool); - void deleteLater(); - void setPool(const Pool &pool); - void tick(uint64_t now); +protected: + bool disconnect() override; + bool isTLS() const override; + const char *tlsFingerprint() const override; + const char *tlsVersion() const override; + int64_t submit(const JobResult &result) override; + void connect() override; + void connect(const Pool &pool) override; + void deleteLater() override; + void tick(uint64_t now) override; - inline bool isReady() const { return m_state == ConnectedState && m_failures == 0; } - inline const char *host() const { return m_pool.host(); } - inline const char *ip() const { return m_ip; } - inline const Job &job() const { return m_job; } - inline int id() const { return m_id; } - inline SocketState state() const { return m_state; } - inline uint16_t port() const { return m_pool.port(); } - inline void setAlgo(const Algorithm &algo) { m_pool.setAlgo(algo); } - inline void setQuiet(bool quiet) { m_quiet = quiet; } - inline void setRetries(int retries) { m_retries = retries; } - inline void setRetryPause(int ms) { m_retryPause = ms; } + void onResolved(const Dns &dns, int status) override; + + inline bool hasExtension(Extension extension) const noexcept override { return m_extensions.test(extension); } + inline const char *mode() const override { return "pool"; } + inline void onLine(char *line, size_t size) override { parse(line, size); } private: class Tls; - - enum Extensions { - NicehashExt = 1, - AlgoExt = 2 - }; - bool close(); bool isCriticalError(const char *message); - bool isTLS() const; bool parseJob(const rapidjson::Value ¶ms, int *code); bool parseLogin(const rapidjson::Value &result, int *code); bool send(BIO *bio); - bool verifyAlgorithm(const Algorithm &algorithm) const; - int resolve(const char *host); + bool verifyAlgorithm(const Algorithm &algorithm, const char *algo) const; + int resolve(const String &host); int64_t send(const rapidjson::Document &doc); int64_t send(size_t size); - void connect(const std::vector &ipv4, const std::vector &ipv6); void connect(sockaddr *addr); void handshake(); void login(); void onClose(); void parse(char *line, size_t len); - void parseExtensions(const rapidjson::Value &value); + void parseExtensions(const rapidjson::Value &result); void parseNotification(const char *method, const rapidjson::Value ¶ms, const rapidjson::Value &error); void parseResponse(int64_t id, const rapidjson::Value &result, const rapidjson::Value &error); void ping(); - void read(); + void read(ssize_t nread); void reconnect(); void setState(SocketState state); void startTimeout(); - inline bool isQuiet() const { return m_quiet || m_failures >= m_retries; } + inline const char *url() const { return m_pool.url(); } + inline SocketState state() const { return m_state; } + inline void setExtension(Extension ext, bool enable) noexcept { m_extensions.set(ext, enable); } + template inline bool has() const noexcept { return m_extensions.test(ext); } static void onAllocBuffer(uv_handle_t *handle, size_t suggested_size, uv_buf_t *buf); static void onClose(uv_handle_t *handle); static void onConnect(uv_connect_t *req, int status); static void onRead(uv_stream_t *stream, ssize_t nread, const uv_buf_t *buf); - static void onResolved(uv_getaddrinfo_t *req, int status, struct addrinfo *res); static inline Client *getClient(void *data) { return m_storage.get(data); } - addrinfo m_hints; - bool m_ipv6; - bool m_nicehash; - bool m_quiet; - char m_buf[kInputBufferSize]; - char m_ip[46]; char m_sendBuf[2048]; const char *m_agent; - IClientListener *m_listener; - int m_extensions; - int m_id; - int m_retries; - int m_retryPause; - int64_t m_failures; - Job m_job; - Pool m_pool; - size_t m_recvBufPos; - SocketState m_state; - std::map m_results; + Dns *m_dns; + RecvBuf m_recvBuf; + std::bitset m_extensions; + String m_rpcId; Tls *m_tls; uint64_t m_expire; uint64_t m_jobs; uint64_t m_keepAlive; uintptr_t m_key; - uv_buf_t m_recvBuf; - uv_getaddrinfo_t m_resolver; uv_stream_t *m_stream; uv_tcp_t *m_socket; - Id m_rpcId; - static int64_t m_sequence; static Storage m_storage; }; +template<> inline bool Client::has() const noexcept { return m_extensions.test(EXT_NICEHASH) || m_pool.isNicehash(); } +template<> inline bool Client::has() const noexcept { return m_extensions.test(EXT_KEEPALIVE) || m_pool.keepAlive() > 0; } + + } /* namespace xmrig */ diff --git a/src/base/net/stratum/DaemonClient.cpp b/src/base/net/stratum/DaemonClient.cpp new file mode 100644 index 000000000..0c141c7d9 --- /dev/null +++ b/src/base/net/stratum/DaemonClient.cpp @@ -0,0 +1,379 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2019 Howard Chu + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include +#include + + +#include "3rdparty/http-parser/http_parser.h" +#include "base/io/json/Json.h" +#include "base/io/json/JsonRequest.h" +#include "base/io/log/Log.h" +#include "base/kernel/interfaces/IClientListener.h" +#include "base/net/http/HttpClient.h" +#include "base/net/stratum/DaemonClient.h" +#include "base/net/stratum/SubmitResult.h" +#include "base/tools/Buffer.h" +#include "base/tools/Timer.h" +#include "net/JobResult.h" +#include "rapidjson/document.h" +#include "rapidjson/error/en.h" +#include "rapidjson/stringbuffer.h" +#include "rapidjson/writer.h" + + +#ifdef XMRIG_FEATURE_TLS +# include "base/net/http/HttpsClient.h" +#endif + + +namespace xmrig { + +static const char *kBlocktemplateBlob = "blocktemplate_blob"; +static const char *kGetHeight = "/getheight"; +static const char *kGetInfo = "/getinfo"; +static const char *kHash = "hash"; +static const char *kHeight = "height"; +static const char *kJsonRPC = "/json_rpc"; + +} + + +xmrig::DaemonClient::DaemonClient(int id, IClientListener *listener) : + BaseClient(id, listener), + m_monero(true) +{ + m_timer = new Timer(this); +} + + +xmrig::DaemonClient::~DaemonClient() +{ + delete m_timer; +} + + +bool xmrig::DaemonClient::disconnect() +{ + if (m_state != UnconnectedState) { + setState(UnconnectedState); + } + + return true; +} + + +bool xmrig::DaemonClient::isTLS() const +{ +# ifdef XMRIG_FEATURE_TLS + return m_pool.isTLS(); +# else + return false; +# endif +} + + +int64_t xmrig::DaemonClient::submit(const JobResult &result) +{ + if (result.jobId != (m_blocktemplate.data() + m_blocktemplate.size() - 32)) { + return -1; + } + +# ifdef XMRIG_PROXY_PROJECT + memcpy(m_blocktemplate.data() + 78, result.nonce, 8); +# else + Buffer::toHex(reinterpret_cast(&result.nonce), 4, m_blocktemplate.data() + 78); +# endif + + using namespace rapidjson; + Document doc(kObjectType); + + Value params(kArrayType); + params.PushBack(m_blocktemplate.toJSON(), doc.GetAllocator()); + + JsonRequest::create(doc, m_sequence, "submitblock", params); + +# ifdef XMRIG_PROXY_PROJECT + m_results[m_sequence] = SubmitResult(m_sequence, result.diff, result.actualDiff(), result.id); +# else + m_results[m_sequence] = SubmitResult(m_sequence, result.diff, result.actualDiff()); +# endif + + send(HTTP_POST, kJsonRPC, doc); + + return m_sequence++; +} + + +void xmrig::DaemonClient::connect() +{ + setState(ConnectingState); + getBlockTemplate(); +} + + +void xmrig::DaemonClient::connect(const Pool &pool) +{ + setPool(pool); + connect(); +} + + +void xmrig::DaemonClient::onHttpData(const HttpData &data) +{ + if (data.status != HTTP_STATUS_OK) { + return retry(); + } + + LOG_DEBUG("[%s:%d] received (%d bytes): \"%.*s\"", m_pool.host().data(), m_pool.port(), static_cast(data.body.size()), static_cast(data.body.size()), data.body.c_str()); + + m_ip = static_cast(data).ip().c_str(); + +# ifdef XMRIG_FEATURE_TLS + if (isTLS()) { + m_tlsVersion = static_cast(data).version(); + m_tlsFingerprint = static_cast(data).fingerprint(); + } +# endif + + rapidjson::Document doc; + if (doc.Parse(data.body.c_str()).HasParseError()) { + if (!isQuiet()) { + LOG_ERR("[%s:%d] JSON decode failed: \"%s\"", m_pool.host().data(), m_pool.port(), rapidjson::GetParseError_En(doc.GetParseError())); + } + + return retry(); + } + + if (data.method == HTTP_GET) { + if (data.url == kGetHeight) { + if (!doc.HasMember(kHash)) { + m_monero = false; + + return send(HTTP_GET, kGetInfo); + } + + if (isOutdated(Json::getUint64(doc, kHeight), Json::getString(doc, kHash))) { + getBlockTemplate(); + } + } + else if (data.url == kGetInfo && isOutdated(Json::getUint64(doc, kHeight), Json::getString(doc, "top_block_hash"))) { + getBlockTemplate(); + } + + return; + } + + if (!parseResponse(Json::getInt64(doc, "id", -1), Json::getObject(doc, "result"), Json::getObject(doc, "error"))) { + retry(); + } +} + + +void xmrig::DaemonClient::onTimer(const Timer *) +{ + if (m_state == ConnectingState) { + getBlockTemplate(); + } + else if (m_state == ConnectedState) { + send(HTTP_GET, m_monero ? kGetHeight : kGetInfo); + } +} + + +bool xmrig::DaemonClient::isOutdated(uint64_t height, const char *hash) const +{ + return m_job.height() != height || m_prevHash != hash; +} + + +bool xmrig::DaemonClient::parseJob(const rapidjson::Value ¶ms, int *code) +{ + Job job(false, m_pool.algorithm(), String()); + + String blocktemplate = Json::getString(params, kBlocktemplateBlob); + if (blocktemplate.isNull() || !job.setBlob(Json::getString(params, "blockhashing_blob"))) { + *code = 4; + return false; + } + + job.setSeedHash(Json::getString(params, "seed_hash")); + job.setHeight(Json::getUint64(params, kHeight)); + job.setDiff(Json::getUint64(params, "difficulty")); + job.setId(blocktemplate.data() + blocktemplate.size() - 32); + + m_job = std::move(job); + m_blocktemplate = std::move(blocktemplate); + m_prevHash = Json::getString(params, "prev_hash"); + + if (m_state == ConnectingState) { + setState(ConnectedState); + } + + m_listener->onJobReceived(this, m_job, params); + return true; +} + + +bool xmrig::DaemonClient::parseResponse(int64_t id, const rapidjson::Value &result, const rapidjson::Value &error) +{ + if (id == -1) { + return false; + } + + if (error.IsObject()) { + const char *message = error["message"].GetString(); + + if (!handleSubmitResponse(id, message) && !isQuiet()) { + LOG_ERR("[%s:%d] error: " RED_BOLD("\"%s\"") RED_S ", code: %d", m_pool.host().data(), m_pool.port(), message, error["code"].GetInt()); + } + + return false; + } + + if (!result.IsObject()) { + return false; + } + + int code = -1; + if (result.HasMember(kBlocktemplateBlob) && parseJob(result, &code)) { + return true; + } + + if (handleSubmitResponse(id)) { + getBlockTemplate(); + return true; + } + + + return false; +} + + +int64_t xmrig::DaemonClient::getBlockTemplate() +{ + using namespace rapidjson; + Document doc(kObjectType); + auto &allocator = doc.GetAllocator(); + + Value params(kObjectType); + params.AddMember("wallet_address", m_pool.user().toJSON(), allocator); + params.AddMember("reserve_size", 8, allocator); + + JsonRequest::create(doc, m_sequence, "getblocktemplate", params); + + send(HTTP_POST, kJsonRPC, doc); + + return m_sequence++; +} + + +void xmrig::DaemonClient::retry() +{ + m_failures++; + m_listener->onClose(this, static_cast(m_failures)); + + if (m_failures == -1) { + return; + } + + if (m_state == ConnectedState) { + setState(ConnectingState); + } + + m_timer->stop(); + m_timer->start(m_retryPause, 0); +} + + +void xmrig::DaemonClient::send(int method, const char *url, const char *data, size_t size) +{ + LOG_DEBUG("[%s:%d] " MAGENTA_BOLD("\"%s %s\"") BLACK_BOLD_S " send (%zu bytes): \"%.*s\"", + m_pool.host().data(), + m_pool.port(), + http_method_str(static_cast(method)), + url, + size, + static_cast(size), + data); + + HttpClient *client; +# ifdef XMRIG_FEATURE_TLS + if (m_pool.isTLS()) { + client = new HttpsClient(method, url, this, data, size, m_pool.fingerprint()); + } + else +# endif + { + client = new HttpClient(method, url, this, data, size); + } + + client->setQuiet(isQuiet()); + client->connect(m_pool.host(), m_pool.port()); +} + + +void xmrig::DaemonClient::send(int method, const char *url, const rapidjson::Document &doc) +{ + using namespace rapidjson; + + StringBuffer buffer(nullptr, 512); + Writer writer(buffer); + doc.Accept(writer); + + send(method, url, buffer.GetString(), buffer.GetSize()); +} + + +void xmrig::DaemonClient::setState(SocketState state) +{ + assert(m_state != state); + if (m_state == state) { + return; + } + + m_state = state; + + switch (state) { + case ConnectedState: + { + m_failures = 0; + m_listener->onLoginSuccess(this); + + const uint64_t interval = std::max(20, m_pool.pollInterval()); + m_timer->start(interval, interval); + } + break; + + case UnconnectedState: + m_failures = -1; + m_timer->stop(); + break; + + default: + break; + } +} diff --git a/src/base/net/stratum/DaemonClient.h b/src/base/net/stratum/DaemonClient.h new file mode 100644 index 000000000..00b62e39a --- /dev/null +++ b/src/base/net/stratum/DaemonClient.h @@ -0,0 +1,83 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2019 Howard Chu + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_DAEMONCLIENT_H +#define XMRIG_DAEMONCLIENT_H + + +#include "base/net/stratum/BaseClient.h" +#include "base/kernel/interfaces/ITimerListener.h" +#include "base/kernel/interfaces/IHttpListener.h" + + +namespace xmrig { + + +class DaemonClient : public BaseClient, public ITimerListener, public IHttpListener +{ +public: + DaemonClient(int id, IClientListener *listener); + ~DaemonClient() override; + +protected: + bool disconnect() override; + bool isTLS() const override; + int64_t submit(const JobResult &result) override; + void connect() override; + void connect(const Pool &pool) override; + + void onHttpData(const HttpData &data) override; + void onTimer(const Timer *timer) override; + + inline bool hasExtension(Extension) const noexcept override { return false; } + inline const char *mode() const override { return "daemon"; } + inline const char *tlsFingerprint() const override { return m_tlsFingerprint; } + inline const char *tlsVersion() const override { return m_tlsVersion; } + inline void deleteLater() override { delete this; } + inline void tick(uint64_t) override {} + +private: + bool isOutdated(uint64_t height, const char *hash) const; + bool parseJob(const rapidjson::Value ¶ms, int *code); + bool parseResponse(int64_t id, const rapidjson::Value &result, const rapidjson::Value &error); + int64_t getBlockTemplate(); + void retry(); + void send(int method, const char *url, const char *data = nullptr, size_t size = 0); + void send(int method, const char *url, const rapidjson::Document &doc); + void setState(SocketState state); + + bool m_monero; + String m_blocktemplate; + String m_prevHash; + String m_tlsFingerprint; + String m_tlsVersion; + Timer *m_timer; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_DAEMONCLIENT_H */ diff --git a/src/common/net/Job.cpp b/src/base/net/stratum/Job.cpp similarity index 53% rename from src/common/net/Job.cpp rename to src/base/net/stratum/Job.cpp index cb6be4e69..04bd82d87 100644 --- a/src/common/net/Job.cpp +++ b/src/base/net/stratum/Job.cpp @@ -7,6 +7,7 @@ * Copyright 2017-2018 XMR-Stak , * Copyright 2018 Lee Clagett * Copyright 2018 SChernykh + * Copyright 2019 Howard Chu * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify @@ -28,62 +29,23 @@ #include -#include "common/net/Job.h" - - -unsigned char hf_hex2bin(char c, bool &err) -{ - if (c >= '0' && c <= '9') { - return c - '0'; - } - else if (c >= 'a' && c <= 'f') { - return c - 'a' + 0xA; - } - else if (c >= 'A' && c <= 'F') { - return c - 'A' + 0xA; - } - - err = true; - return 0; -} - - -char hf_bin2hex(unsigned char c) -{ - if (c <= 0x9) { - return '0' + c; - } - - return 'a' - 0xA + c; -} +#include "base/net/stratum/Job.h" +#include "base/tools/Buffer.h" xmrig::Job::Job() : - m_autoVariant(false), - m_nicehash(false), - m_poolId(-2), - m_threadId(-1), - m_size(0), - m_diff(0), - m_target(0), m_blob(), - m_height(0) + m_seedHash() { } -xmrig::Job::Job(int poolId, bool nicehash, const Algorithm &algorithm, const Id &clientId) : - m_autoVariant(algorithm.variant() == VARIANT_AUTO), - m_nicehash(nicehash), - m_poolId(poolId), - m_threadId(-1), - m_size(0), - m_diff(0), - m_target(0), - m_blob(), - m_height(0), +xmrig::Job::Job(bool nicehash, const Algorithm &algorithm, const String &clientId) : m_algorithm(algorithm), - m_clientId(clientId) + m_nicehash(nicehash), + m_clientId(clientId), + m_blob(), + m_seedHash() { } @@ -115,7 +77,7 @@ bool xmrig::Job::setBlob(const char *blob) return false; } - if (!fromHex(blob, (int) m_size * 2, m_blob)) { + if (!Buffer::fromHex(blob, m_size * 2, m_blob)) { return false; } @@ -123,10 +85,6 @@ bool xmrig::Job::setBlob(const char *blob) m_nicehash = true; } - if (m_autoVariant) { - m_algorithm.setVariant(variant()); - } - # ifdef XMRIG_PROXY_PROJECT memset(m_rawBlob, 0, sizeof(m_rawBlob)); memcpy(m_rawBlob, blob, m_size * 2); @@ -136,6 +94,20 @@ bool xmrig::Job::setBlob(const char *blob) } +bool xmrig::Job::setSeedHash(const char *hash) +{ + if (!hash || (strlen(hash) != sizeof(m_seedHash) * 2)) { + return false; + } + +# ifdef XMRIG_PROXY_PROJECT + m_rawSeedHash = hash; +# endif + + return Buffer::fromHex(hash, sizeof(m_seedHash) * 2, m_seedHash); +} + + bool xmrig::Job::setTarget(const char *target) { if (!target) { @@ -149,7 +121,7 @@ bool xmrig::Job::setTarget(const char *target) char str[8]; memcpy(str, target, len); - if (!fromHex(str, 8, reinterpret_cast(&tmp)) || tmp == 0) { + if (!Buffer::fromHex(str, 8, reinterpret_cast(&tmp)) || tmp == 0) { return false; } @@ -160,7 +132,7 @@ bool xmrig::Job::setTarget(const char *target) char str[16]; memcpy(str, target, len); - if (!fromHex(str, 16, reinterpret_cast(&m_target)) || m_target == 0) { + if (!Buffer::fromHex(str, 16, reinterpret_cast(&m_target)) || m_target == 0) { return false; } } @@ -178,71 +150,37 @@ bool xmrig::Job::setTarget(const char *target) } -void xmrig::Job::setAlgorithm(const char *algo) +void xmrig::Job::setDiff(uint64_t diff) { - m_algorithm.parseAlgorithm(algo); + m_diff = diff; + m_target = toDiff(diff); - if (m_algorithm.variant() == xmrig::VARIANT_AUTO) { - m_algorithm.setVariant(variant()); - } +# ifdef XMRIG_PROXY_PROJECT + Buffer::toHex(reinterpret_cast(&m_target), 8, m_rawTarget); + m_rawTarget[16] = '\0'; +# endif } -void xmrig::Job::setHeight(uint64_t height) +void xmrig::Job::copy(const Job &other) { - m_height = height; -} - - -bool xmrig::Job::fromHex(const char* in, unsigned int len, unsigned char* out) -{ - bool error = false; - for (unsigned int i = 0; i < len; i += 2) { - out[i / 2] = (hf_hex2bin(in[i], error) << 4) | hf_hex2bin(in[i + 1], error); - - if (error) { - return false; - } - } - return true; -} - - -void xmrig::Job::toHex(const unsigned char* in, unsigned int len, char* out) -{ - for (unsigned int i = 0; i < len; i++) { - out[i * 2] = hf_bin2hex((in[i] & 0xF0) >> 4); - out[i * 2 + 1] = hf_bin2hex(in[i] & 0x0F); - } -} - - -#ifdef APP_DEBUG -char *xmrig::Job::toHex(const unsigned char* in, unsigned int len) -{ - char *out = new char[len * 2 + 1](); - toHex(in, len, out); - - return out; -} -#endif - - -xmrig::Variant xmrig::Job::variant() const -{ - switch (m_algorithm.algo()) { - case CRYPTONIGHT: - return (m_blob[0] >= 10) ? VARIANT_4 : ((m_blob[0] >= 8) ? VARIANT_2 : VARIANT_1); - - case CRYPTONIGHT_LITE: - return VARIANT_1; - - case CRYPTONIGHT_HEAVY: - return VARIANT_0; - - default: - break; - } - - return m_algorithm.variant(); + m_algorithm = other.m_algorithm; + m_nicehash = other.m_nicehash; + m_size = other.m_size; + m_clientId = other.m_clientId; + m_id = other.m_id; + m_diff = other.m_diff; + m_height = other.m_height; + m_target = other.m_target; + m_index = other.m_index; + + memcpy(m_blob, other.m_blob, sizeof(m_blob)); + memcpy(m_seedHash, other.m_seedHash, sizeof(m_seedHash)); + +# ifdef XMRIG_PROXY_PROJECT + m_rawSeedHash = other.m_rawSeedHash; + + memcpy(m_rawBlob, other.m_rawBlob, sizeof(m_rawBlob)); + memcpy(m_rawTarget, other.m_rawTarget, sizeof(m_rawTarget)); +# endif } diff --git a/src/common/net/Job.h b/src/base/net/stratum/Job.h similarity index 66% rename from src/common/net/Job.h rename to src/base/net/stratum/Job.h index 49ddc7da3..2b256a12f 100644 --- a/src/common/net/Job.h +++ b/src/base/net/stratum/Job.h @@ -7,6 +7,7 @@ * Copyright 2017-2018 XMR-Stak , * Copyright 2018 Lee Clagett * Copyright 2018-2019 SChernykh + * Copyright 2019 Howard Chu * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify @@ -31,8 +32,8 @@ #include -#include "common/crypto/Algorithm.h" -#include "common/net/Id.h" +#include "base/tools/String.h" +#include "crypto/common/Algorithm.h" namespace xmrig { @@ -46,73 +47,70 @@ public: static constexpr const size_t kMaxBlobSize = 128; Job(); - Job(int poolId, bool nicehash, const Algorithm &algorithm, const Id &clientId); + Job(bool nicehash, const Algorithm &algorithm, const String &clientId); ~Job(); bool isEqual(const Job &other) const; bool setBlob(const char *blob); + bool setSeedHash(const char *hash); bool setTarget(const char *target); - void setAlgorithm(const char *algo); - void setHeight(uint64_t height); + void setDiff(uint64_t diff); inline bool isNicehash() const { return m_nicehash; } inline bool isValid() const { return m_size > 0 && m_diff > 0; } - inline bool setId(const char *id) { return m_id.setId(id); } + inline bool setId(const char *id) { return m_id = id; } + inline const Algorithm &algorithm() const { return m_algorithm; } + inline const String &clientId() const { return m_clientId; } + inline const String &id() const { return m_id; } inline const uint32_t *nonce() const { return reinterpret_cast(m_blob + 39); } inline const uint8_t *blob() const { return m_blob; } - inline const Algorithm &algorithm() const { return m_algorithm; } - inline const Id &clientId() const { return m_clientId; } - inline const Id &id() const { return m_id; } - inline int poolId() const { return m_poolId; } - inline int threadId() const { return m_threadId; } + inline const uint8_t *seedHash() const { return m_seedHash; } inline size_t size() const { return m_size; } inline uint32_t *nonce() { return reinterpret_cast(m_blob + 39); } - inline uint32_t diff() const { return static_cast(m_diff); } - inline uint64_t target() const { return m_target; } + inline uint64_t diff() const { return m_diff; } inline uint64_t height() const { return m_height; } + inline uint64_t target() const { return m_target; } + inline uint8_t fixedByte() const { return *(m_blob + 42); } + inline uint8_t index() const { return m_index; } inline void reset() { m_size = 0; m_diff = 0; } - inline void setClientId(const Id &id) { m_clientId = id; } - inline void setPoolId(int poolId) { m_poolId = poolId; } - inline void setThreadId(int threadId) { m_threadId = threadId; } - inline void setVariant(const char *variant) { m_algorithm.parseVariant(variant); } - inline void setVariant(int variant) { m_algorithm.parseVariant(variant); } + inline void setAlgorithm(const char *algo) { m_algorithm = algo; } + inline void setClientId(const String &id) { m_clientId = id; } + inline void setHeight(uint64_t height) { m_height = height; } + inline void setIndex(uint8_t index) { m_index = index; } # ifdef XMRIG_PROXY_PROJECT - inline char *rawBlob() { return m_rawBlob; } - inline const char *rawTarget() const { return m_rawTarget; } + inline char *rawBlob() { return m_rawBlob; } + inline const char *rawBlob() const { return m_rawBlob; } + inline const char *rawTarget() const { return m_rawTarget; } + inline const String &rawSeedHash() const { return m_rawSeedHash; } # endif - static bool fromHex(const char* in, unsigned int len, unsigned char* out); static inline uint32_t *nonce(uint8_t *blob) { return reinterpret_cast(blob + 39); } static inline uint64_t toDiff(uint64_t target) { return 0xFFFFFFFFFFFFFFFFULL / target; } - static void toHex(const unsigned char* in, unsigned int len, char* out); - -# ifdef APP_DEBUG - static char *toHex(const unsigned char* in, unsigned int len); -# endif inline bool operator==(const Job &other) const { return isEqual(other); } inline bool operator!=(const Job &other) const { return !isEqual(other); } + inline Job &operator=(const Job &other) { copy(other); return *this; } private: - Variant variant() const; + void copy(const Job &other); - bool m_autoVariant; - bool m_nicehash; - int m_poolId; - int m_threadId; - size_t m_size; - uint64_t m_diff; - uint64_t m_target; + Algorithm m_algorithm; + bool m_nicehash = false; + size_t m_size = 0; + String m_clientId; + String m_id; + uint64_t m_diff = 0; + uint64_t m_height = 0; + uint64_t m_target = 0; uint8_t m_blob[kMaxBlobSize]; - uint64_t m_height; - xmrig::Algorithm m_algorithm; - xmrig::Id m_clientId; - xmrig::Id m_id; + uint8_t m_index = 0; + uint8_t m_seedHash[32]; # ifdef XMRIG_PROXY_PROJECT char m_rawBlob[kMaxBlobSize * 2 + 8]; char m_rawTarget[24]; + String m_rawSeedHash; # endif }; diff --git a/src/base/net/stratum/Pool.cpp b/src/base/net/stratum/Pool.cpp new file mode 100644 index 000000000..4d15ea473 --- /dev/null +++ b/src/base/net/stratum/Pool.cpp @@ -0,0 +1,336 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2019 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2019 Howard Chu + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include +#include +#include +#include + + +#include "base/io/json/Json.h" +#include "base/net/stratum/Pool.h" +#include "rapidjson/document.h" + + +#ifdef APP_DEBUG +# include "base/io/log/Log.h" +#endif + + +#ifdef _MSC_VER +# define strncasecmp _strnicmp +#endif + + +namespace xmrig { + +static const char *kAlgo = "algo"; +static const char *kDaemon = "daemon"; +static const char *kDaemonPollInterval = "daemon-poll-interval"; +static const char *kEnabled = "enabled"; +static const char *kFingerprint = "tls-fingerprint"; +static const char *kKeepalive = "keepalive"; +static const char *kNicehash = "nicehash"; +static const char *kPass = "pass"; +static const char *kRigId = "rig-id"; +static const char *kTls = "tls"; +static const char *kUrl = "url"; +static const char *kUser = "user"; + +const String Pool::kDefaultPassword = "x"; +const String Pool::kDefaultUser = "x"; + +static const char kStratumTcp[] = "stratum+tcp://"; +static const char kStratumSsl[] = "stratum+ssl://"; + +#ifdef XMRIG_FEATURE_HTTP +static const char kDaemonHttp[] = "daemon+http://"; +static const char kDaemonHttps[] = "daemon+https://"; +#endif + +} + + +xmrig::Pool::Pool() : + m_keepAlive(0), + m_flags(0), + m_port(kDefaultPort), + m_pollInterval(kDefaultPollInterval) +{ +} + + +/** + * @brief Parse url. + * + * Valid urls: + * example.com + * example.com:3333 + * stratum+tcp://example.com + * stratum+tcp://example.com:3333 + * + * @param url + */ +xmrig::Pool::Pool(const char *url) : + m_keepAlive(0), + m_flags(1), + m_port(kDefaultPort), + m_pollInterval(kDefaultPollInterval) +{ + parse(url); +} + + +xmrig::Pool::Pool(const rapidjson::Value &object) : + m_keepAlive(0), + m_flags(1), + m_port(kDefaultPort), + m_pollInterval(kDefaultPollInterval) +{ + if (!parse(Json::getString(object, kUrl))) { + return; + } + + m_user = Json::getString(object, kUser); + m_password = Json::getString(object, kPass); + m_rigId = Json::getString(object, kRigId); + m_fingerprint = Json::getString(object, kFingerprint); + m_pollInterval = Json::getUint64(object, kDaemonPollInterval, kDefaultPollInterval); + m_algorithm = Json::getString(object, kAlgo); + + m_flags.set(FLAG_ENABLED, Json::getBool(object, kEnabled, true)); + m_flags.set(FLAG_NICEHASH, Json::getBool(object, kNicehash)); + m_flags.set(FLAG_TLS, Json::getBool(object, kTls, m_flags.test(FLAG_TLS))); + m_flags.set(FLAG_DAEMON, Json::getBool(object, kDaemon, m_flags.test(FLAG_DAEMON))); + + const rapidjson::Value &keepalive = Json::getValue(object, kKeepalive); + if (keepalive.IsInt()) { + setKeepAlive(keepalive.GetInt()); + } + else if (keepalive.IsBool()) { + setKeepAlive(keepalive.GetBool()); + } +} + + +xmrig::Pool::Pool(const char *host, uint16_t port, const char *user, const char *password, int keepAlive, bool nicehash, bool tls) : + m_keepAlive(keepAlive), + m_flags(1), + m_host(host), + m_password(password), + m_user(user), + m_port(port), + m_pollInterval(kDefaultPollInterval) +{ + const size_t size = m_host.size() + 8; + assert(size > 8); + + char *url = new char[size](); + snprintf(url, size - 1, "%s:%d", m_host.data(), m_port); + + m_url = url; + + m_flags.set(FLAG_NICEHASH, nicehash); + m_flags.set(FLAG_TLS, tls); +} + + +bool xmrig::Pool::isEnabled() const +{ +# ifndef XMRIG_FEATURE_TLS + if (isTLS()) { + return false; + } +# endif + +# ifndef XMRIG_FEATURE_HTTP + if (isDaemon()) { + return false; + } +# endif + + if (isDaemon() && !algorithm().isValid()) { + return false; + } + + return m_flags.test(FLAG_ENABLED) && isValid(); +} + + +bool xmrig::Pool::isEqual(const Pool &other) const +{ + return (m_flags == other.m_flags + && m_keepAlive == other.m_keepAlive + && m_port == other.m_port + && m_algorithm == other.m_algorithm + && m_fingerprint == other.m_fingerprint + && m_host == other.m_host + && m_password == other.m_password + && m_rigId == other.m_rigId + && m_url == other.m_url + && m_user == other.m_user + && m_pollInterval == other.m_pollInterval + ); +} + + +bool xmrig::Pool::parse(const char *url) +{ + assert(url != nullptr); + + const char *p = strstr(url, "://"); + const char *base = url; + + if (p) { + if (strncasecmp(url, kStratumTcp, sizeof(kStratumTcp) - 1) == 0) { + m_flags.set(FLAG_DAEMON, false); + m_flags.set(FLAG_TLS, false); + } + else if (strncasecmp(url, kStratumSsl, sizeof(kStratumSsl) - 1) == 0) { + m_flags.set(FLAG_DAEMON, false); + m_flags.set(FLAG_TLS, true); + } +# ifdef XMRIG_FEATURE_HTTP + else if (strncasecmp(url, kDaemonHttps, sizeof(kDaemonHttps) - 1) == 0) { + m_flags.set(FLAG_DAEMON, true); + m_flags.set(FLAG_TLS, true); + } + else if (strncasecmp(url, kDaemonHttp, sizeof(kDaemonHttp) - 1) == 0) { + m_flags.set(FLAG_DAEMON, true); + m_flags.set(FLAG_TLS, false); + } +# endif + else { + return false; + } + + base = p + 3; + } + + if (!strlen(base) || *base == '/') { + return false; + } + + m_url = url; + if (base[0] == '[') { + return parseIPv6(base); + } + + const char *port = strchr(base, ':'); + if (!port) { + m_host = base; + return true; + } + + const size_t size = static_cast(port++ - base + 1); + char *host = new char[size](); + memcpy(host, base, size - 1); + + m_host = host; + m_port = static_cast(strtol(port, nullptr, 10)); + + return true; +} + + +rapidjson::Value xmrig::Pool::toJSON(rapidjson::Document &doc) const +{ + using namespace rapidjson; + + auto &allocator = doc.GetAllocator(); + + Value obj(kObjectType); + + obj.AddMember(StringRef(kAlgo), m_algorithm.toJSON(), allocator); + obj.AddMember(StringRef(kUrl), m_url.toJSON(), allocator); + obj.AddMember(StringRef(kUser), m_user.toJSON(), allocator); + + if (!isDaemon()) { + obj.AddMember(StringRef(kPass), m_password.toJSON(), allocator); + obj.AddMember(StringRef(kRigId), m_rigId.toJSON(), allocator); + +# ifndef XMRIG_PROXY_PROJECT + obj.AddMember(StringRef(kNicehash), isNicehash(), allocator); +# endif + + if (m_keepAlive == 0 || m_keepAlive == kKeepAliveTimeout) { + obj.AddMember(StringRef(kKeepalive), m_keepAlive > 0, allocator); + } + else { + obj.AddMember(StringRef(kKeepalive), m_keepAlive, allocator); + } + } + + obj.AddMember(StringRef(kEnabled), m_flags.test(FLAG_ENABLED), allocator); + obj.AddMember(StringRef(kTls), isTLS(), allocator); + obj.AddMember(StringRef(kFingerprint), m_fingerprint.toJSON(), allocator); + obj.AddMember(StringRef(kDaemon), m_flags.test(FLAG_DAEMON), allocator); + + if (isDaemon()) { + obj.AddMember(StringRef(kDaemonPollInterval), m_pollInterval, allocator); + } + + return obj; +} + + +#ifdef APP_DEBUG +void xmrig::Pool::print() const +{ + LOG_NOTICE("url: %s", m_url.data()); + LOG_DEBUG ("host: %s", m_host.data()); + LOG_DEBUG ("port: %d", static_cast(m_port)); + LOG_DEBUG ("user: %s", m_user.data()); + LOG_DEBUG ("pass: %s", m_password.data()); + LOG_DEBUG ("rig-id %s", m_rigId.data()); + LOG_DEBUG ("algo: %s", m_algorithm.name()); + LOG_DEBUG ("nicehash: %d", static_cast(m_flags.test(FLAG_NICEHASH))); + LOG_DEBUG ("keepAlive: %d", m_keepAlive); +} +#endif + + +bool xmrig::Pool::parseIPv6(const char *addr) +{ + const char *end = strchr(addr, ']'); + if (!end) { + return false; + } + + const char *port = strchr(end, ':'); + if (!port) { + return false; + } + + const size_t size = static_cast(end - addr); + char *host = new char[size](); + memcpy(host, addr + 1, size - 1); + + m_host = host; + m_port = static_cast(strtol(port + 1, nullptr, 10)); + + return true; +} diff --git a/src/base/net/Pool.h b/src/base/net/stratum/Pool.h similarity index 59% rename from src/base/net/Pool.h rename to src/base/net/stratum/Pool.h index a16cc15b9..36c3ed1b5 100644 --- a/src/base/net/Pool.h +++ b/src/base/net/stratum/Pool.h @@ -6,6 +6,7 @@ * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , * Copyright 2018-2019 SChernykh + * Copyright 2019 Howard Chu * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify @@ -26,11 +27,12 @@ #define XMRIG_POOL_H +#include #include #include "base/tools/String.h" -#include "common/crypto/Algorithm.h" +#include "crypto/common/Algorithm.h" #include "rapidjson/fwd.h" @@ -40,10 +42,20 @@ namespace xmrig { class Pool { public: - constexpr static const char *kDefaultPassword = "x"; - constexpr static const char *kDefaultUser = "x"; - constexpr static uint16_t kDefaultPort = 3333; - constexpr static int kKeepAliveTimeout = 60; + enum Flags { + FLAG_ENABLED, + FLAG_NICEHASH, + FLAG_TLS, + FLAG_DAEMON, + FLAG_MAX + }; + + static const String kDefaultPassword; + static const String kDefaultUser; + + constexpr static int kKeepAliveTimeout = 60; + constexpr static uint16_t kDefaultPort = 3333; + constexpr static uint64_t kDefaultPollInterval = 1000; Pool(); Pool(const char *url); @@ -57,57 +69,46 @@ public: bool tls = false ); - inline bool isNicehash() const { return m_nicehash; } - inline bool isTLS() const { return m_tls; } + inline bool isDaemon() const { return m_flags.test(FLAG_DAEMON); } + inline bool isNicehash() const { return m_flags.test(FLAG_NICEHASH); } + inline bool isTLS() const { return m_flags.test(FLAG_TLS); } inline bool isValid() const { return !m_host.isNull() && m_port > 0; } - inline const char *fingerprint() const { return m_fingerprint.data(); } - inline const char *host() const { return m_host.data(); } - inline const char *password() const { return !m_password.isNull() ? m_password.data() : kDefaultPassword; } - inline const char *rigId() const { return m_rigId.data(); } - inline const char *url() const { return m_url.data(); } - inline const char *user() const { return !m_user.isNull() ? m_user.data() : kDefaultUser; } inline const Algorithm &algorithm() const { return m_algorithm; } - inline const Algorithms &algorithms() const { return m_algorithms; } + inline const String &fingerprint() const { return m_fingerprint; } + inline const String &host() const { return m_host; } + inline const String &password() const { return !m_password.isNull() ? m_password : kDefaultPassword; } + inline const String &rigId() const { return m_rigId; } + inline const String &url() const { return m_url; } + inline const String &user() const { return !m_user.isNull() ? m_user : kDefaultUser; } inline int keepAlive() const { return m_keepAlive; } inline uint16_t port() const { return m_port; } - inline void setFingerprint(const char *fingerprint) { m_fingerprint = fingerprint; } - inline void setKeepAlive(int keepAlive) { m_keepAlive = keepAlive >= 0 ? keepAlive : 0; } - inline void setKeepAlive(bool enable) { setKeepAlive(enable ? kKeepAliveTimeout : 0); } - inline void setNicehash(bool nicehash) { m_nicehash = nicehash; } - inline void setPassword(const char *password) { m_password = password; } - inline void setRigId(const char *rigId) { m_rigId = rigId; } - inline void setTLS(bool tls) { m_tls = tls; } - inline void setUser(const char *user) { m_user = user; } - inline Algorithm &algorithm() { return m_algorithm; } + inline uint64_t pollInterval() const { return m_pollInterval; } + inline void setAlgo(const Algorithm &algorithm) { m_algorithm = algorithm; } + inline void setPassword(const String &password) { m_password = password; } + inline void setRigId(const String &rigId) { m_rigId = rigId; } + inline void setUser(const String &user) { m_user = user; } - inline bool operator!=(const Pool &other) const { return !isEqual(other); } - inline bool operator==(const Pool &other) const { return isEqual(other); } + inline bool operator!=(const Pool &other) const { return !isEqual(other); } + inline bool operator==(const Pool &other) const { return isEqual(other); } - bool isCompatible(const Algorithm &algorithm) const; bool isEnabled() const; bool isEqual(const Pool &other) const; bool parse(const char *url); - bool setUserpass(const char *userpass); rapidjson::Value toJSON(rapidjson::Document &doc) const; - void adjust(const Algorithm &algorithm); - void setAlgo(const Algorithm &algorithm); # ifdef APP_DEBUG void print() const; # endif private: + inline void setKeepAlive(bool enable) { setKeepAlive(enable ? kKeepAliveTimeout : 0); } + inline void setKeepAlive(int keepAlive) { m_keepAlive = keepAlive >= 0 ? keepAlive : 0; } + bool parseIPv6(const char *addr); - void addVariant(Variant variant); - void adjustVariant(const Variant variantHint); - void rebuild(); Algorithm m_algorithm; - Algorithms m_algorithms; - bool m_enabled; - bool m_nicehash; - bool m_tls; int m_keepAlive; + std::bitset m_flags; String m_fingerprint; String m_host; String m_password; @@ -115,6 +116,7 @@ private: String m_url; String m_user; uint16_t m_port; + uint64_t m_pollInterval; }; diff --git a/src/base/net/Pools.cpp b/src/base/net/stratum/Pools.cpp similarity index 69% rename from src/base/net/Pools.cpp rename to src/base/net/stratum/Pools.cpp index e3b86acac..4641ecd4f 100644 --- a/src/base/net/Pools.cpp +++ b/src/base/net/stratum/Pools.cpp @@ -23,16 +23,20 @@ */ -#include "base/net/Pools.h" -#include "common/log/Log.h" -#include "common/net/strategies/FailoverStrategy.h" -#include "common/net/strategies/SinglePoolStrategy.h" +#include "base/io/log/Log.h" +#include "base/kernel/interfaces/IJsonReader.h" +#include "base/net/stratum/Pools.h" +#include "base/net/stratum/strategies/FailoverStrategy.h" +#include "base/net/stratum/strategies/SinglePoolStrategy.h" +#include "donate.h" #include "rapidjson/document.h" xmrig::Pools::Pools() : + m_donateLevel(kDefaultDonateLevel), m_retries(5), - m_retryPause(5) + m_retryPause(5), + m_proxyDonate(PROXY_DONATE_AUTO) { # ifdef XMRIG_PROXY_PROJECT m_retries = 2; @@ -41,16 +45,6 @@ xmrig::Pools::Pools() : } -xmrig::Pool &xmrig::Pools::current() -{ - if (m_data.empty()) { - m_data.push_back(Pool()); - } - - return m_data.back(); -} - - bool xmrig::Pools::isEqual(const Pools &other) const { if (m_data.size() != other.m_data.size() || m_retries != other.m_retries || m_retryPause != other.m_retryPause) { @@ -61,25 +55,6 @@ bool xmrig::Pools::isEqual(const Pools &other) const } -bool xmrig::Pools::setUrl(const char *url) -{ - if (m_data.empty() || m_data.back().isValid()) { - Pool pool(url); - - if (pool.isValid()) { - m_data.push_back(std::move(pool)); - return true; - } - - return false; - } - - current().parse(url); - - return m_data.back().isValid(); -} - - xmrig::IStrategy *xmrig::Pools::createStrategy(IStrategyListener *listener) const { if (active() == 1) { @@ -129,18 +104,15 @@ size_t xmrig::Pools::active() const } -void xmrig::Pools::adjust(const Algorithm &algorithm) -{ - for (Pool &pool : m_data) { - pool.adjust(algorithm); - } -} - - -void xmrig::Pools::load(const rapidjson::Value &pools) +void xmrig::Pools::load(const IJsonReader &reader) { m_data.clear(); + const rapidjson::Value &pools = reader.getArray("pools"); + if (!pools.IsArray()) { + return; + } + for (const rapidjson::Value &value : pools.GetArray()) { if (!value.IsObject()) { continue; @@ -151,6 +123,11 @@ void xmrig::Pools::load(const rapidjson::Value &pools) m_data.push_back(std::move(pool)); } } + + setDonateLevel(reader.getInt("donate-level", kDefaultDonateLevel)); + setProxyDonate(reader.getInt("donate-over-proxy", PROXY_DONATE_AUTO)); + setRetries(reader.getInt("retries")); + setRetryPause(reader.getInt("retry-pause")); } @@ -158,25 +135,12 @@ void xmrig::Pools::print() const { size_t i = 1; for (const Pool &pool : m_data) { - if (Log::colors) { - const int color = pool.isEnabled() ? (pool.isTLS() ? 32 : 36) : 31; - - Log::i()->text(GREEN_BOLD(" * ") WHITE_BOLD("POOL #%-7zu") "\x1B[1;%dm%s\x1B[0m variant " WHITE_BOLD("%s"), - i, - color, - pool.url(), - pool.algorithm().variantName() - ); - } - else { - Log::i()->text(" * POOL #%-7zu%s%s variant=%s %s", - i, - pool.isEnabled() ? "" : "-", - pool.url(), - pool.algorithm().variantName(), - pool.isTLS() ? "TLS" : "" - ); - } + Log::print(GREEN_BOLD(" * ") WHITE_BOLD("POOL #%-7zu") CSI "1;%dm%s" CLEAR " algo " WHITE_BOLD("%s"), + i, + (pool.isEnabled() ? (pool.isTLS() ? 32 : 36) : 31), + pool.url().data(), + pool.algorithm().isValid() ? pool.algorithm().shortName() : "auto" + ); i++; } @@ -191,6 +155,25 @@ void xmrig::Pools::print() const } +void xmrig::Pools::setDonateLevel(int level) +{ + if (level >= kMinimumDonateLevel && level <= 99) { + m_donateLevel = level; + } +} + + +void xmrig::Pools::setProxyDonate(int value) +{ + switch (value) { + case PROXY_DONATE_NONE: + case PROXY_DONATE_AUTO: + case PROXY_DONATE_ALWAYS: + m_proxyDonate = static_cast(value); + } +} + + void xmrig::Pools::setRetries(int retries) { if (retries > 0 && retries <= 1000) { diff --git a/src/base/net/Pools.h b/src/base/net/stratum/Pools.h similarity index 64% rename from src/base/net/Pools.h rename to src/base/net/stratum/Pools.h index a6038a044..70e172255 100644 --- a/src/base/net/Pools.h +++ b/src/base/net/stratum/Pools.h @@ -29,12 +29,13 @@ #include -#include "base/net/Pool.h" +#include "base/net/stratum/Pool.h" namespace xmrig { +class IJsonReader; class IStrategy; class IStrategyListener; @@ -42,42 +43,40 @@ class IStrategyListener; class Pools { public: + enum ProxyDonate { + PROXY_DONATE_NONE, + PROXY_DONATE_AUTO, + PROXY_DONATE_ALWAYS + }; + Pools(); - inline bool setUserpass(const char *userpass) { return current().setUserpass(userpass); } inline const std::vector &data() const { return m_data; } + inline int donateLevel() const { return m_donateLevel; } inline int retries() const { return m_retries; } inline int retryPause() const { return m_retryPause; } - inline void setFingerprint(const char *fingerprint) { current().setFingerprint(fingerprint); } - inline void setKeepAlive(bool enable) { current().setKeepAlive(enable); } - inline void setKeepAlive(int keepAlive) { current().setKeepAlive(keepAlive); } - inline void setNicehash(bool enable) { current().setNicehash(enable); } - inline void setPassword(const char *password) { current().setPassword(password); } - inline void setRigId(const char *rigId) { current().setRigId(rigId); } - inline void setTLS(bool enable) { current().setTLS(enable); } - inline void setUser(const char *user) { current().setUser(user); } - inline void setVariant(const char *variant) { current().algorithm().parseVariant(variant); } - inline void setVariant(int variant) { current().algorithm().parseVariant(variant); } + inline ProxyDonate proxyDonate() const { return m_proxyDonate; } inline bool operator!=(const Pools &other) const { return !isEqual(other); } inline bool operator==(const Pools &other) const { return isEqual(other); } bool isEqual(const Pools &other) const; - bool setUrl(const char *url); IStrategy *createStrategy(IStrategyListener *listener) const; rapidjson::Value toJSON(rapidjson::Document &doc) const; size_t active() const; - void adjust(const Algorithm &algorithm); - void load(const rapidjson::Value &pools); + void load(const IJsonReader &reader); void print() const; + +private: + void setDonateLevel(int level); + void setProxyDonate(int value); void setRetries(int retries); void setRetryPause(int retryPause); -private: - Pool ¤t(); - + int m_donateLevel; int m_retries; int m_retryPause; + ProxyDonate m_proxyDonate; std::vector m_data; }; diff --git a/src/base/net/stratum/SubmitResult.h b/src/base/net/stratum/SubmitResult.h new file mode 100644 index 000000000..5abd3e4bf --- /dev/null +++ b/src/base/net/stratum/SubmitResult.h @@ -0,0 +1,72 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_SUBMITRESULT_H +#define XMRIG_SUBMITRESULT_H + + +#include "base/tools/Chrono.h" + + +namespace xmrig { + + +class SubmitResult +{ +public: + inline SubmitResult() : + reqId(0), + seq(0), + actualDiff(0), + diff(0), + elapsed(0), + m_start(0) + {} + + inline SubmitResult(int64_t seq, uint64_t diff, uint64_t actualDiff, int64_t reqId = 0) : + reqId(reqId), + seq(seq), + actualDiff(actualDiff), + diff(diff), + elapsed(0), + m_start(Chrono::steadyMSecs()) + {} + + inline void done() { elapsed = Chrono::steadyMSecs() - m_start; } + + int64_t reqId; + int64_t seq; + uint64_t actualDiff; + uint64_t diff; + uint64_t elapsed; + +private: + uint64_t m_start; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_SUBMITRESULT_H */ diff --git a/src/common/net/Tls.cpp b/src/base/net/stratum/Tls.cpp similarity index 93% rename from src/common/net/Tls.cpp rename to src/base/net/stratum/Tls.cpp index a9b3fa7de..8916579b4 100644 --- a/src/common/net/Tls.cpp +++ b/src/base/net/stratum/Tls.cpp @@ -27,9 +27,10 @@ #include -#include "common/net/Client.h" -#include "common/net/Tls.h" -#include "common/log/Log.h" +#include "base/io/log/Log.h" +#include "base/net/stratum/Client.h" +#include "base/net/stratum/Tls.h" +#include "base/tools/Buffer.h" #ifdef _MSC_VER @@ -134,7 +135,8 @@ void xmrig::Client::Tls::read(const char *data, size_t size) int bytes_read = 0; while ((bytes_read = SSL_read(m_ssl, m_buf, sizeof(m_buf))) > 0) { - m_client->parse(m_buf, bytes_read); + m_buf[bytes_read - 1] = '\0'; + m_client->parse(m_buf, static_cast(bytes_read)); } } @@ -148,13 +150,13 @@ bool xmrig::Client::Tls::send() bool xmrig::Client::Tls::verify(X509 *cert) { if (cert == nullptr) { - LOG_ERR("[%s] Failed to get server certificate", m_client->m_pool.url()); + LOG_ERR("[%s] Failed to get server certificate", m_client->url()); return false; } if (!verifyFingerprint(cert)) { - LOG_ERR("[%s] Failed to verify server certificate fingerprint", m_client->m_pool.url()); + LOG_ERR("[%s] Failed to verify server certificate fingerprint", m_client->url()); const char *fingerprint = m_client->m_pool.fingerprint(); if (strlen(m_fingerprint) == 64 && fingerprint != nullptr) { @@ -183,7 +185,7 @@ bool xmrig::Client::Tls::verifyFingerprint(X509 *cert) return false; } - Job::toHex(md, 32, m_fingerprint); + Buffer::toHex(md, 32, m_fingerprint); const char *fingerprint = m_client->m_pool.fingerprint(); return fingerprint == nullptr || strncasecmp(m_fingerprint, fingerprint, 64) == 0; diff --git a/src/common/net/Tls.h b/src/base/net/stratum/Tls.h similarity index 98% rename from src/common/net/Tls.h rename to src/base/net/stratum/Tls.h index cce78eeb4..e070be526 100644 --- a/src/common/net/Tls.h +++ b/src/base/net/stratum/Tls.h @@ -29,7 +29,7 @@ #include -#include "common/net/Client.h" +#include "base/net/stratum/Client.h" namespace xmrig { diff --git a/src/common/net/strategies/FailoverStrategy.cpp b/src/base/net/stratum/strategies/FailoverStrategy.cpp similarity index 66% rename from src/common/net/strategies/FailoverStrategy.cpp rename to src/base/net/stratum/strategies/FailoverStrategy.cpp index c833f6e6e..48be2ba35 100644 --- a/src/common/net/strategies/FailoverStrategy.cpp +++ b/src/base/net/stratum/strategies/FailoverStrategy.cpp @@ -23,10 +23,15 @@ */ -#include "common/interfaces/IStrategyListener.h" -#include "common/net/Client.h" -#include "common/net/strategies/FailoverStrategy.h" -#include "common/Platform.h" +#include "base/kernel/interfaces/IStrategyListener.h" +#include "base/kernel/Platform.h" +#include "base/net/stratum/Client.h" +#include "base/net/stratum/strategies/FailoverStrategy.h" + + +#ifdef XMRIG_FEATURE_HTTP +# include "base/net/stratum/DaemonClient.h" +#endif xmrig::FailoverStrategy::FailoverStrategy(const std::vector &pools, int retryPause, int retries, IStrategyListener *listener, bool quiet) : @@ -34,8 +39,8 @@ xmrig::FailoverStrategy::FailoverStrategy(const std::vector &pools, int re m_retries(retries), m_retryPause(retryPause), m_active(-1), - m_index(0), - m_listener(listener) + m_listener(listener), + m_index(0) { for (const Pool &pool : pools) { add(pool); @@ -48,15 +53,15 @@ xmrig::FailoverStrategy::FailoverStrategy(int retryPause, int retries, IStrategy m_retries(retries), m_retryPause(retryPause), m_active(-1), - m_index(0), - m_listener(listener) + m_listener(listener), + m_index(0) { } xmrig::FailoverStrategy::~FailoverStrategy() { - for (Client *client : m_pools) { + for (IClient *client : m_pools) { client->deleteLater(); } } @@ -64,7 +69,15 @@ xmrig::FailoverStrategy::~FailoverStrategy() void xmrig::FailoverStrategy::add(const Pool &pool) { - Client *client = new Client(static_cast(m_pools.size()), Platform::userAgent(), this); + const int id = static_cast(m_pools.size()); + +# ifdef XMRIG_FEATURE_HTTP + IClient *client = !pool.isDaemon() ? static_cast(new Client(id, Platform::userAgent(), this)) + : static_cast(new DaemonClient(id, this)); +# else + IClient *client = new Client(id, Platform::userAgent(), this); +# endif + client->setPool(pool); client->setRetries(m_retries); client->setRetryPause(m_retryPause * 1000); @@ -76,7 +89,7 @@ void xmrig::FailoverStrategy::add(const Pool &pool) int64_t xmrig::FailoverStrategy::submit(const JobResult &result) { - if (m_active == -1) { + if (!isActive()) { return -1; } @@ -86,7 +99,7 @@ int64_t xmrig::FailoverStrategy::submit(const JobResult &result) void xmrig::FailoverStrategy::connect() { - m_pools[static_cast(m_index)]->connect(); + m_pools[m_index]->connect(); } @@ -100,9 +113,9 @@ void xmrig::FailoverStrategy::resume() } -void xmrig::FailoverStrategy::setAlgo(const xmrig::Algorithm &algo) +void xmrig::FailoverStrategy::setAlgo(const Algorithm &algo) { - for (Client *client : m_pools) { + for (IClient *client : m_pools) { client->setAlgo(algo); } } @@ -123,13 +136,13 @@ void xmrig::FailoverStrategy::stop() void xmrig::FailoverStrategy::tick(uint64_t now) { - for (Client *client : m_pools) { + for (IClient *client : m_pools) { client->tick(now); } } -void xmrig::FailoverStrategy::onClose(Client *client, int failures) +void xmrig::FailoverStrategy::onClose(IClient *client, int failures) { if (failures == -1) { return; @@ -144,13 +157,19 @@ void xmrig::FailoverStrategy::onClose(Client *client, int failures) return; } - if (m_index == client->id() && (m_pools.size() - static_cast(m_index)) > 1) { - m_pools[static_cast(++m_index)]->connect(); + if (m_index == static_cast(client->id()) && (m_pools.size() - m_index) > 1) { + m_pools[++m_index]->connect(); } } -void xmrig::FailoverStrategy::onJobReceived(Client *client, const Job &job) +void xmrig::FailoverStrategy::onLogin(IClient *client, rapidjson::Document &doc, rapidjson::Value ¶ms) +{ + m_listener->onLogin(this, client, doc, params); +} + + +void xmrig::FailoverStrategy::onJobReceived(IClient *client, const Job &job, const rapidjson::Value &) { if (m_active == client->id()) { m_listener->onJob(this, client, job); @@ -158,7 +177,7 @@ void xmrig::FailoverStrategy::onJobReceived(Client *client, const Job &job) } -void xmrig::FailoverStrategy::onLoginSuccess(Client *client) +void xmrig::FailoverStrategy::onLoginSuccess(IClient *client) { int active = m_active; @@ -179,7 +198,13 @@ void xmrig::FailoverStrategy::onLoginSuccess(Client *client) } -void xmrig::FailoverStrategy::onResultAccepted(Client *client, const SubmitResult &result, const char *error) +void xmrig::FailoverStrategy::onResultAccepted(IClient *client, const SubmitResult &result, const char *error) { m_listener->onResultAccepted(this, client, result, error); } + + +void xmrig::FailoverStrategy::onVerifyAlgorithm(const IClient *client, const Algorithm &algorithm, bool *ok) +{ + m_listener->onVerifyAlgorithm(this, client, algorithm, ok); +} diff --git a/src/common/net/strategies/FailoverStrategy.h b/src/base/net/stratum/strategies/FailoverStrategy.h similarity index 69% rename from src/common/net/strategies/FailoverStrategy.h rename to src/base/net/stratum/strategies/FailoverStrategy.h index d2231a30b..283d49165 100644 --- a/src/common/net/strategies/FailoverStrategy.h +++ b/src/base/net/stratum/strategies/FailoverStrategy.h @@ -29,9 +29,9 @@ #include -#include "base/net/Pool.h" -#include "common/interfaces/IClientListener.h" -#include "common/interfaces/IStrategy.h" +#include "base/kernel/interfaces/IClientListener.h" +#include "base/kernel/interfaces/IStrategy.h" +#include "base/net/stratum/Pool.h" namespace xmrig { @@ -50,8 +50,9 @@ public: void add(const Pool &pool); -public: - inline bool isActive() const override { return m_active >= 0; } +protected: + inline bool isActive() const override { return m_active >= 0; } + inline IClient *client() const override { return isActive() ? active() : m_pools[m_index]; } int64_t submit(const JobResult &result) override; void connect() override; @@ -60,22 +61,23 @@ public: void stop() override; void tick(uint64_t now) override; -protected: - void onClose(Client *client, int failures) override; - void onJobReceived(Client *client, const Job &job) override; - void onLoginSuccess(Client *client) override; - void onResultAccepted(Client *client, const SubmitResult &result, const char *error) override; + void onClose(IClient *client, int failures) override; + void onJobReceived(IClient *client, const Job &job, const rapidjson::Value ¶ms) override; + void onLogin(IClient *client, rapidjson::Document &doc, rapidjson::Value ¶ms) override; + void onLoginSuccess(IClient *client) override; + void onResultAccepted(IClient *client, const SubmitResult &result, const char *error) override; + void onVerifyAlgorithm(const IClient *client, const Algorithm &algorithm, bool *ok) override; private: - inline Client *active() const { return m_pools[static_cast(m_active)]; } + inline IClient *active() const { return m_pools[static_cast(m_active)]; } const bool m_quiet; const int m_retries; const int m_retryPause; int m_active; - int m_index; IStrategyListener *m_listener; - std::vector m_pools; + size_t m_index; + std::vector m_pools; }; diff --git a/src/common/net/strategies/SinglePoolStrategy.cpp b/src/base/net/stratum/strategies/SinglePoolStrategy.cpp similarity index 67% rename from src/common/net/strategies/SinglePoolStrategy.cpp rename to src/base/net/stratum/strategies/SinglePoolStrategy.cpp index d17a43da1..c923e1c2c 100644 --- a/src/common/net/strategies/SinglePoolStrategy.cpp +++ b/src/base/net/stratum/strategies/SinglePoolStrategy.cpp @@ -23,17 +23,32 @@ */ -#include "common/interfaces/IStrategyListener.h" -#include "common/net/Client.h" -#include "common/net/strategies/SinglePoolStrategy.h" -#include "common/Platform.h" +#include "base/kernel/interfaces/IStrategyListener.h" +#include "base/kernel/Platform.h" +#include "base/net/stratum/Client.h" +#include "base/net/stratum/strategies/SinglePoolStrategy.h" + + +#ifdef XMRIG_FEATURE_HTTP +# include "base/net/stratum/DaemonClient.h" +#endif xmrig::SinglePoolStrategy::SinglePoolStrategy(const Pool &pool, int retryPause, int retries, IStrategyListener *listener, bool quiet) : m_active(false), m_listener(listener) { +# ifdef XMRIG_FEATURE_HTTP + if (!pool.isDaemon()) { + m_client = new Client(0, Platform::userAgent(), this); + } + else { + m_client = new DaemonClient(0, this); + } +# else m_client = new Client(0, Platform::userAgent(), this); +# endif + m_client->setPool(pool); m_client->setRetries(retries); m_client->setRetryPause(retryPause * 1000); @@ -69,7 +84,7 @@ void xmrig::SinglePoolStrategy::resume() } -void xmrig::SinglePoolStrategy::setAlgo(const xmrig::Algorithm &algo) +void xmrig::SinglePoolStrategy::setAlgo(const Algorithm &algo) { m_client->setAlgo(algo); } @@ -87,7 +102,7 @@ void xmrig::SinglePoolStrategy::tick(uint64_t now) } -void xmrig::SinglePoolStrategy::onClose(Client *, int) +void xmrig::SinglePoolStrategy::onClose(IClient *, int) { if (!isActive()) { return; @@ -98,20 +113,32 @@ void xmrig::SinglePoolStrategy::onClose(Client *, int) } -void xmrig::SinglePoolStrategy::onJobReceived(Client *client, const Job &job) +void xmrig::SinglePoolStrategy::onJobReceived(IClient *client, const Job &job, const rapidjson::Value &) { m_listener->onJob(this, client, job); } -void xmrig::SinglePoolStrategy::onLoginSuccess(Client *client) +void xmrig::SinglePoolStrategy::onLogin(IClient *client, rapidjson::Document &doc, rapidjson::Value ¶ms) +{ + m_listener->onLogin(this, client, doc, params); +} + + +void xmrig::SinglePoolStrategy::onLoginSuccess(IClient *client) { m_active = true; m_listener->onActive(this, client); } -void xmrig::SinglePoolStrategy::onResultAccepted(Client *client, const SubmitResult &result, const char *error) +void xmrig::SinglePoolStrategy::onResultAccepted(IClient *client, const SubmitResult &result, const char *error) { m_listener->onResultAccepted(this, client, result, error); } + + +void xmrig::SinglePoolStrategy::onVerifyAlgorithm(const IClient *client, const Algorithm &algorithm, bool *ok) +{ + m_listener->onVerifyAlgorithm(this, client, algorithm, ok); +} diff --git a/src/common/net/strategies/SinglePoolStrategy.h b/src/base/net/stratum/strategies/SinglePoolStrategy.h similarity index 72% rename from src/common/net/strategies/SinglePoolStrategy.h rename to src/base/net/stratum/strategies/SinglePoolStrategy.h index eeb6903e8..ea808193c 100644 --- a/src/common/net/strategies/SinglePoolStrategy.h +++ b/src/base/net/stratum/strategies/SinglePoolStrategy.h @@ -26,8 +26,8 @@ #define XMRIG_SINGLEPOOLSTRATEGY_H -#include "common/interfaces/IClientListener.h" -#include "common/interfaces/IStrategy.h" +#include "base/kernel/interfaces/IClientListener.h" +#include "base/kernel/interfaces/IStrategy.h" namespace xmrig { @@ -44,8 +44,9 @@ public: SinglePoolStrategy(const Pool &pool, int retryPause, int retries, IStrategyListener *listener, bool quiet = false); ~SinglePoolStrategy() override; -public: - inline bool isActive() const override { return m_active; } +protected: + inline bool isActive() const override { return m_active; } + inline IClient *client() const override { return m_client; } int64_t submit(const JobResult &result) override; void connect() override; @@ -54,15 +55,16 @@ public: void stop() override; void tick(uint64_t now) override; -protected: - void onClose(Client *client, int failures) override; - void onJobReceived(Client *client, const Job &job) override; - void onLoginSuccess(Client *client) override; - void onResultAccepted(Client *client, const SubmitResult &result, const char *error) override; + void onClose(IClient *client, int failures) override; + void onJobReceived(IClient *client, const Job &job, const rapidjson::Value ¶ms) override; + void onLogin(IClient *client, rapidjson::Document &doc, rapidjson::Value ¶ms) override; + void onLoginSuccess(IClient *client) override; + void onResultAccepted(IClient *client, const SubmitResult &result, const char *error) override; + void onVerifyAlgorithm(const IClient *client, const Algorithm &algorithm, bool *ok) override; private: bool m_active; - Client *m_client; + IClient *m_client; IStrategyListener *m_listener; }; diff --git a/src/base/net/tools/RecvBuf.h b/src/base/net/tools/RecvBuf.h new file mode 100644 index 000000000..5122c9804 --- /dev/null +++ b/src/base/net/tools/RecvBuf.h @@ -0,0 +1,99 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_RECVBUF_H +#define XMRIG_RECVBUF_H + + +#include + + +#include "base/kernel/interfaces/ILineListener.h" + + +namespace xmrig { + + +template +class RecvBuf +{ +public: + inline RecvBuf() : + m_buf(), + m_pos(0) + { + } + + inline char *base() { return m_buf; } + inline char *current() { return m_buf + m_pos; } + inline const char *base() const { return m_buf; } + inline const char *current() const { return m_buf + m_pos; } + inline size_t available() const { return N - m_pos; } + inline size_t pos() const { return m_pos; } + inline void nread(size_t size) { m_pos += size; } + inline void reset() { m_pos = 0; } + + constexpr inline size_t size() const { return N; } + + inline void getline(ILineListener *listener) + { + char *end; + char *start = m_buf; + size_t remaining = m_pos; + + while ((end = static_cast(memchr(start, '\n', remaining))) != nullptr) { + *end = '\0'; + + end++; + const size_t len = static_cast(end - start); + + listener->onLine(start, len - 1); + + remaining -= len; + start = end; + } + + if (remaining == 0) { + m_pos = 0; + return; + } + + if (start == m_buf) { + return; + } + + memcpy(m_buf, start, remaining); + m_pos = remaining; + } + +private: + char m_buf[N]; + size_t m_pos; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_RECVBUF_H */ diff --git a/src/common/net/Storage.h b/src/base/net/tools/Storage.h similarity index 70% rename from src/common/net/Storage.h rename to src/base/net/tools/Storage.h index f36ce5946..aff624816 100644 --- a/src/common/net/Storage.h +++ b/src/base/net/tools/Storage.h @@ -5,7 +5,8 @@ * Copyright 2014-2016 Wolf9466 * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -21,8 +22,8 @@ * along with this program. If not, see . */ -#ifndef __STORAGE_H__ -#define __STORAGE_H__ +#ifndef XMRIG_STORAGE_H +#define XMRIG_STORAGE_H #include @@ -53,11 +54,10 @@ public: inline static void *ptr(uintptr_t id) { return reinterpret_cast(id); } - inline TYPE *get(void *id) const { return get(reinterpret_cast(id)); } + inline TYPE *get(const void *id) const { return get(reinterpret_cast(id)); } inline TYPE *get(uintptr_t id) const { assert(m_data.count(id) > 0); - if (m_data.count(id) == 0) { return nullptr; } @@ -66,20 +66,22 @@ public: } - inline void remove(void *id) { remove(reinterpret_cast(id)); } - inline void remove(uintptr_t id) + inline void remove(const void *id) { delete release(reinterpret_cast(id)); } + inline void remove(uintptr_t id) { delete release(id); } + + + inline TYPE *release(const void *id) { release(reinterpret_cast(id)); } + inline TYPE *release(uintptr_t id) { TYPE *obj = get(id); - if (obj == nullptr) { - return; + if (obj != nullptr) { + auto it = m_data.find(id); + if (it != m_data.end()) { + m_data.erase(it); + } } - auto it = m_data.find(id); - if (it != m_data.end()) { - m_data.erase(it); - } - - delete obj; + return obj; } @@ -92,4 +94,4 @@ private: } /* namespace xmrig */ -#endif /* __STORAGE_H__ */ +#endif /* XMRIG_STORAGE_H */ diff --git a/src/base/net/tools/TcpServer.cpp b/src/base/net/tools/TcpServer.cpp new file mode 100644 index 000000000..e5e6e4e18 --- /dev/null +++ b/src/base/net/tools/TcpServer.cpp @@ -0,0 +1,102 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "base/kernel/interfaces/ITcpServerListener.h" +#include "base/net/tools/TcpServer.h" +#include "base/tools/Handle.h" +#include "base/tools/String.h" + + +static const xmrig::String kLocalHost("127.0.0.1"); + + +xmrig::TcpServer::TcpServer(const String &host, uint16_t port, ITcpServerListener *listener) : + m_host(host.isNull() ? kLocalHost : host), + m_version(0), + m_listener(listener), + m_addr(), + m_port(port) +{ + m_tcp = new uv_tcp_t; + uv_tcp_init(uv_default_loop(), m_tcp); + m_tcp->data = this; + + uv_tcp_nodelay(m_tcp, 1); + + if (m_host.contains(":") && uv_ip6_addr(m_host.data(), m_port, reinterpret_cast(&m_addr)) == 0) { + m_version = 6; + } + else if (uv_ip4_addr(m_host.data(), m_port, reinterpret_cast(&m_addr)) == 0) { + m_version = 4; + } +} + + +xmrig::TcpServer::~TcpServer() +{ + Handle::close(m_tcp); +} + + +int xmrig::TcpServer::bind() +{ + if (!m_version) { + return UV_EAI_ADDRFAMILY; + } + + uv_tcp_bind(m_tcp, reinterpret_cast(&m_addr), 0); + + const int rc = uv_listen(reinterpret_cast(m_tcp), 511, TcpServer::onConnection); + if (rc != 0) { + return rc; + } + + if (!m_port) { + sockaddr_storage storage = {}; + int size = sizeof(storage); + + uv_tcp_getsockname(m_tcp, reinterpret_cast(&storage), &size); + + m_port = ntohs(reinterpret_cast(&storage)->sin_port); + } + + return m_port; +} + + +void xmrig::TcpServer::create(uv_stream_t *stream, int status) +{ + if (status < 0) { + return; + } + + m_listener->onConnection(stream, m_port); +} + + +void xmrig::TcpServer::onConnection(uv_stream_t *stream, int status) +{ + static_cast(stream->data)->create(stream, status); +} diff --git a/src/base/net/tools/TcpServer.h b/src/base/net/tools/TcpServer.h new file mode 100644 index 000000000..fc29fa2d0 --- /dev/null +++ b/src/base/net/tools/TcpServer.h @@ -0,0 +1,64 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_TCPSERVER_H +#define XMRIG_TCPSERVER_H + + +#include + + +namespace xmrig { + + +class ITcpServerListener; +class String; + + +class TcpServer +{ +public: + TcpServer(const String &host, uint16_t port, ITcpServerListener *listener); + ~TcpServer(); + + int bind(); + +private: + void create(uv_stream_t *stream, int status); + + static void onConnection(uv_stream_t *stream, int status); + + const String &m_host; + int m_version; + ITcpServerListener *m_listener; + sockaddr_storage m_addr; + uint16_t m_port; + uv_tcp_t *m_tcp; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_TCPSERVER_H */ diff --git a/src/common/interfaces/IConfigCreator.h b/src/base/tools/Baton.h similarity index 85% rename from src/common/interfaces/IConfigCreator.h rename to src/base/tools/Baton.h index c2e0e62ea..646dc0b87 100644 --- a/src/common/interfaces/IConfigCreator.h +++ b/src/base/tools/Baton.h @@ -22,26 +22,24 @@ * along with this program. If not, see . */ -#ifndef XMRIG_ICONFIGCREATOR_H -#define XMRIG_ICONFIGCREATOR_H +#ifndef XMRIG_BATON_H +#define XMRIG_BATON_H namespace xmrig { -class IConfig; - - -class IConfigCreator +template +class Baton { public: - virtual ~IConfigCreator() = default; + inline Baton() { req.data = this; } - virtual IConfig *create() const = 0; + REQ req; }; } /* namespace xmrig */ -#endif // XMRIG_ICONFIGCREATOR_H +#endif /* XMRIG_BATON_H */ diff --git a/src/base/tools/Buffer.cpp b/src/base/tools/Buffer.cpp new file mode 100644 index 000000000..05a68cb4c --- /dev/null +++ b/src/base/tools/Buffer.cpp @@ -0,0 +1,201 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "base/tools/Buffer.h" + + +static inline uint8_t hf_hex2bin(uint8_t c, bool &err) +{ + if (c >= '0' && c <= '9') { + return c - '0'; + } + else if (c >= 'a' && c <= 'f') { + return c - 'a' + 0xA; + } + else if (c >= 'A' && c <= 'F') { + return c - 'A' + 0xA; + } + + err = true; + return 0; +} + + +static inline uint8_t hf_bin2hex(uint8_t c) +{ + if (c <= 0x9) { + return '0' + c; + } + + return 'a' - 0xA + c; +} + + +xmrig::Buffer::Buffer() : + m_data(nullptr), + m_size(0) +{ +} + + +xmrig::Buffer::Buffer(Buffer &&other) : + m_data(other.m_data), + m_size(other.m_size) +{ + other.m_data = nullptr; + other.m_size = 0; +} + + +xmrig::Buffer::Buffer(const Buffer &other) +{ + copy(other.data(), other.size()); +} + + +xmrig::Buffer::Buffer(const char *data, size_t size) +{ + copy(data, size); +} + + +xmrig::Buffer::Buffer(size_t size) : + m_size(size) +{ + m_data = new char[size](); +} + + +xmrig::Buffer::~Buffer() +{ + delete [] m_data; +} + + +void xmrig::Buffer::from(const char *data, size_t size) +{ + if (m_size > 0) { + if (m_size == size) { + memcpy(m_data, data, m_size); + + return; + } + + delete [] m_data; + } + + copy(data, size); +} + + +xmrig::Buffer xmrig::Buffer::allocUnsafe(size_t size) +{ + Buffer buf; + buf.m_size = size; + buf.m_data = new char[size]; + + return buf; +} + + +bool xmrig::Buffer::fromHex(const uint8_t *in, size_t size, uint8_t *out) +{ + bool error = false; + for (size_t i = 0; i < size; i += 2) { + out[i / 2] = static_cast((hf_hex2bin(in[i], error) << 4) | hf_hex2bin(in[i + 1], error)); + + if (error) { + return false; + } + } + + return true; +} + + +xmrig::Buffer xmrig::Buffer::fromHex(const char *data, size_t size) +{ + if (data == nullptr || size % 2 != 0) { + return Buffer(); + } + + Buffer buf(size / 2); + fromHex(data, size, buf.data()); + + return buf; +} + + +void xmrig::Buffer::toHex(const uint8_t *in, size_t size, uint8_t *out) +{ + for (size_t i = 0; i < size; i++) { + out[i * 2] = hf_bin2hex((in[i] & 0xF0) >> 4); + out[i * 2 + 1] = hf_bin2hex(in[i] & 0x0F); + } +} + + +xmrig::String xmrig::Buffer::toHex(const uint8_t *in, size_t size) +{ + return Buffer(reinterpret_cast(in), size).toHex(); +} + + +xmrig::String xmrig::Buffer::toHex() const +{ + if (m_size == 0) { + return String(); + } + + char *buf = new char[m_size * 2 + 1]; + buf[m_size * 2] = '\0'; + + toHex(m_data, m_size, buf); + + return String(buf); +} + + +void xmrig::Buffer::copy(const char *data, size_t size) +{ + m_data = new char[size]; + m_size = size; + + memcpy(m_data, data, m_size); +} + + +void xmrig::Buffer::move(Buffer &&other) +{ + if (m_size > 0) { + delete [] m_data; + } + + m_data = other.m_data; + m_size = other.m_size; + + other.m_data = nullptr; + other.m_size = 0; +} diff --git a/src/base/tools/Buffer.h b/src/base/tools/Buffer.h new file mode 100644 index 000000000..28f92b9e1 --- /dev/null +++ b/src/base/tools/Buffer.h @@ -0,0 +1,91 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_BUFFER_H +#define XMRIG_BUFFER_H + + +#include "base/tools/String.h" + + +namespace xmrig { + + +class Buffer +{ +public: + Buffer(); + Buffer(Buffer &&other); + Buffer(const Buffer &other); + Buffer(const char *data, size_t size); + Buffer(size_t size); + ~Buffer(); + + + inline bool isEqual(const Buffer &other) const { return m_size == other.m_size && (m_size == 0 || memcmp(m_data, other.m_data, m_size) == 0); } + inline char *data() { return m_data; } + inline const char *data() const { return m_data; } + inline size_t size() const { return m_size; } + inline void from(const Buffer &other) { from(other.data(), other.size()); } + + + void from(const char *data, size_t size); + + + inline bool operator!=(const Buffer &other) const { return !isEqual(other); } + inline bool operator==(const Buffer &other) const { return isEqual(other); } + inline Buffer &operator=(Buffer &&other) { move(std::move(other)); return *this; } + inline Buffer &operator=(const Buffer &other) { from(other); return *this; } + + + static Buffer allocUnsafe(size_t size); + static inline Buffer alloc(size_t size) { return Buffer(size); } + + + inline static bool fromHex(const char *in, size_t size, char *out) { return fromHex(reinterpret_cast(in), size, reinterpret_cast(out)); } + inline static bool fromHex(const char *in, size_t size, uint8_t *out) { return fromHex(reinterpret_cast(in), size, out); } + inline static Buffer fromHex(const char *data) { return fromHex(data, strlen(data)); } + inline static Buffer fromHex(const String &str) { return fromHex(str.data(), str.size()); } + inline static void toHex(const char *in, size_t size, char *out) { return toHex(reinterpret_cast(in), size, reinterpret_cast(out)); } + inline static void toHex(const uint8_t *in, size_t size, char *out) { return toHex(in, size, reinterpret_cast(out)); } + + static bool fromHex(const uint8_t *in, size_t size, uint8_t *out); + static Buffer fromHex(const char *data, size_t size); + static String toHex(const uint8_t *in, size_t size); + static void toHex(const uint8_t *in, size_t size, uint8_t *out); + String toHex() const; + +private: + void copy(const char *data, size_t size); + void move(Buffer &&other); + + char *m_data; + size_t m_size; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_BUFFER_H */ diff --git a/src/base/tools/Chrono.h b/src/base/tools/Chrono.h new file mode 100644 index 000000000..e464f3613 --- /dev/null +++ b/src/base/tools/Chrono.h @@ -0,0 +1,68 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_CHRONO_H +#define XMRIG_CHRONO_H + + +#include + + +namespace xmrig { + + +class Chrono +{ +public: + static inline uint64_t highResolutionMSecs() + { + using namespace std::chrono; + + return static_cast(time_point_cast(high_resolution_clock::now()).time_since_epoch().count()); + } + + + static inline uint64_t steadyMSecs() + { + using namespace std::chrono; + if (high_resolution_clock::is_steady) { + return static_cast(time_point_cast(high_resolution_clock::now()).time_since_epoch().count()); + } + + return static_cast(time_point_cast(steady_clock::now()).time_since_epoch().count()); + } + + + static inline uint64_t currentMSecsSinceEpoch() + { + using namespace std::chrono; + + return static_cast(time_point_cast(system_clock::now()).time_since_epoch().count()); + } +}; + + +} /* namespace xmrig */ + +#endif /* XMRIG_CHRONO_H */ diff --git a/src/base/tools/Handle.h b/src/base/tools/Handle.h index 547c92026..1a7d08f1c 100644 --- a/src/base/tools/Handle.h +++ b/src/base/tools/Handle.h @@ -26,12 +26,7 @@ #define XMRIG_HANDLE_H -typedef struct uv_fs_event_s uv_fs_event_t; -typedef struct uv_getaddrinfo_s uv_getaddrinfo_t; -typedef struct uv_handle_s uv_handle_t; -typedef struct uv_signal_s uv_signal_t; -typedef struct uv_tcp_s uv_tcp_t; -typedef struct uv_timer_s uv_timer_t; +#include namespace xmrig { @@ -40,15 +35,57 @@ namespace xmrig { class Handle { public: - static void close(uv_fs_event_t *handle); - static void close(uv_getaddrinfo_t *handle); - static void close(uv_handle_t *handle); - static void close(uv_signal_t *handle); - static void close(uv_tcp_t *handle); - static void close(uv_timer_t *handle); + template + static inline void close(T handle) + { + if (handle) { + deleteLater(handle); + } + } + + + template + static inline void deleteLater(T handle) + { + if (uv_is_closing(reinterpret_cast(handle))) { + return; + } + + uv_close(reinterpret_cast(handle), [](uv_handle_t *handle) { delete handle; }); + } }; +template<> +inline void Handle::close(uv_timer_t *handle) +{ + if (handle) { + uv_timer_stop(handle); + deleteLater(handle); + } +} + + +template<> +inline void Handle::close(uv_signal_t *handle) +{ + if (handle) { + uv_signal_stop(handle); + deleteLater(handle); + } +} + + +template<> +inline void Handle::close(uv_fs_event_t *handle) +{ + if (handle) { + uv_fs_event_stop(handle); + deleteLater(handle); + } +} + + } /* namespace xmrig */ diff --git a/src/base/tools/String.cpp b/src/base/tools/String.cpp index 7ed61d01f..7778c6da6 100644 --- a/src/base/tools/String.cpp +++ b/src/base/tools/String.cpp @@ -5,6 +5,7 @@ * Copyright 2014-2016 Wolf9466 * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify @@ -22,6 +23,9 @@ */ +#include + + #include "base/tools/String.h" #include "rapidjson/document.h" @@ -128,6 +132,20 @@ std::vector xmrig::String::split(char sep) const } +xmrig::String &xmrig::String::toLower() +{ + if (isNull() || isEmpty()) { + return *this; + } + + for (size_t i = 0; i < size(); ++i) { + m_data[i] = static_cast(tolower(m_data[i])); + } + + return *this; +} + + xmrig::String xmrig::String::join(const std::vector &vec, char sep) { if (vec.empty()) { @@ -178,14 +196,10 @@ void xmrig::String::copy(const char *str) void xmrig::String::copy(const String &other) { - if (m_size > 0) { - if (m_size == other.m_size) { - memcpy(m_data, other.m_data, m_size + 1); + if (m_size > 0 && m_size == other.m_size) { + memcpy(m_data, other.m_data, m_size + 1); - return; - } - - delete [] m_data; + return; } delete [] m_data; diff --git a/src/base/tools/String.h b/src/base/tools/String.h index 0c191dfde..2c47d8501 100644 --- a/src/base/tools/String.h +++ b/src/base/tools/String.h @@ -5,7 +5,8 @@ * Copyright 2014-2016 Wolf9466 * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -79,11 +80,13 @@ public: inline String &operator=(char *str) { move(str); return *this; } inline String &operator=(const char *str) { copy(str); return *this; } inline String &operator=(const String &str) { copy(str); return *this; } + inline String &operator=(std::nullptr_t) { delete [] m_data; m_data = nullptr; m_size = 0; return *this; } inline String &operator=(String &&other) { move(std::move(other)); return *this; } rapidjson::Value toJSON() const; rapidjson::Value toJSON(rapidjson::Document &doc) const; std::vector split(char sep) const; + String &toLower(); static String join(const std::vector &vec, char sep); diff --git a/src/base/tools/Timer.cpp b/src/base/tools/Timer.cpp new file mode 100644 index 000000000..d06df1636 --- /dev/null +++ b/src/base/tools/Timer.cpp @@ -0,0 +1,91 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "base/kernel/interfaces/ITimerListener.h" +#include "base/tools/Handle.h" +#include "base/tools/Timer.h" + + +xmrig::Timer::Timer(ITimerListener *listener) : + m_listener(listener), + m_timer(nullptr) +{ + init(); +} + + +xmrig::Timer::Timer(ITimerListener *listener, uint64_t timeout, uint64_t repeat) : + m_listener(listener), + m_timer(nullptr) +{ + init(); + start(timeout, repeat); +} + + +xmrig::Timer::~Timer() +{ + Handle::close(m_timer); +} + + +uint64_t xmrig::Timer::repeat() const +{ + return uv_timer_get_repeat(m_timer); +} + + +void xmrig::Timer::setRepeat(uint64_t repeat) +{ + uv_timer_set_repeat(m_timer, repeat); +} + + +void xmrig::Timer::start(uint64_t timeout, uint64_t repeat) +{ + uv_timer_start(m_timer, onTimer, timeout, repeat); +} + + +void xmrig::Timer::stop() +{ + uv_timer_stop(m_timer); +} + + +void xmrig::Timer::init() +{ + m_timer = new uv_timer_t; + m_timer->data = this; + uv_timer_init(uv_default_loop(), m_timer); +} + + +void xmrig::Timer::onTimer(uv_timer_t *handle) +{ + const Timer *timer = static_cast(handle->data); + + timer->m_listener->onTimer(timer); +} diff --git a/src/base/tools/Timer.h b/src/base/tools/Timer.h new file mode 100644 index 000000000..e0e210f52 --- /dev/null +++ b/src/base/tools/Timer.h @@ -0,0 +1,66 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_TIMER_H +#define XMRIG_TIMER_H + + +#include + + +typedef struct uv_timer_s uv_timer_t; + + +namespace xmrig { + + +class ITimerListener; + + +class Timer +{ +public: + Timer(ITimerListener *listener); + Timer(ITimerListener *listener, uint64_t timeout, uint64_t repeat); + ~Timer(); + + uint64_t repeat() const; + void setRepeat(uint64_t repeat); + void start(uint64_t timeout, uint64_t repeat); + void stop(); + +private: + void init(); + + static void onTimer(uv_timer_t *handle); + + ITimerListener *m_listener; + uv_timer_t *m_timer; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_TIMER_H */ diff --git a/src/common/api/HttpBody.h b/src/common/api/HttpBody.h deleted file mode 100644 index 0b143fb74..000000000 --- a/src/common/api/HttpBody.h +++ /dev/null @@ -1,69 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef __HTTPBODY_H__ -#define __HTTPBODY_H__ - - -#include - - -namespace xmrig { - - -class HttpBody -{ -public: - inline HttpBody() : - m_pos(0) - {} - - - inline bool write(const char *data, size_t size) - { - if (size > (sizeof(m_data) - m_pos - 1)) { - return false; - } - - memcpy(m_data + m_pos, data, size); - - m_pos += size; - m_data[m_pos] = '\0'; - - return true; - } - - - inline const char *data() const { return m_data; } - -private: - char m_data[32768]; - size_t m_pos; -}; - - -} /* namespace xmrig */ - - -#endif /* __HTTPBODY_H__ */ diff --git a/src/common/api/HttpRequest.cpp b/src/common/api/HttpRequest.cpp deleted file mode 100644 index 6898a3851..000000000 --- a/src/common/api/HttpRequest.cpp +++ /dev/null @@ -1,175 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - - -#include -#include - -#include "common/api/HttpBody.h" -#include "common/api/HttpRequest.h" -#include "common/api/HttpReply.h" - - -#ifndef MHD_HTTP_PAYLOAD_TOO_LARGE -# define MHD_HTTP_PAYLOAD_TOO_LARGE 413 -#endif - - -xmrig::HttpRequest::HttpRequest(MHD_Connection *connection, const char *url, const char *method, const char *uploadData, size_t *uploadSize, void **cls) : - m_fulfilled(true), - m_restricted(true), - m_uploadData(uploadData), - m_url(url), - m_body(static_cast(*cls)), - m_method(Unsupported), - m_connection(connection), - m_uploadSize(uploadSize), - m_cls(cls) -{ - if (strcmp(method, MHD_HTTP_METHOD_OPTIONS) == 0) { - m_method = Options; - } - else if (strcmp(method, MHD_HTTP_METHOD_GET) == 0) { - m_method = Get; - } - else if (strcmp(method, MHD_HTTP_METHOD_PUT) == 0) { - m_method = Put; - } -} - - -xmrig::HttpRequest::~HttpRequest() -{ - if (m_fulfilled) { - delete m_body; - } -} - - -bool xmrig::HttpRequest::match(const char *path) const -{ - return strcmp(m_url, path) == 0; -} - - -bool xmrig::HttpRequest::process(const char *accessToken, bool restricted, xmrig::HttpReply &reply) -{ - m_restricted = restricted || !accessToken; - - if (m_body) { - if (*m_uploadSize != 0) { - if (!m_body->write(m_uploadData, *m_uploadSize)) { - *m_cls = nullptr; - m_fulfilled = true; - reply.status = MHD_HTTP_PAYLOAD_TOO_LARGE; - return false; - } - - *m_uploadSize = 0; - m_fulfilled = false; - return true; - } - - m_fulfilled = true; - return true; - } - - reply.status = auth(accessToken); - if (reply.status != MHD_HTTP_OK) { - return false; - } - - if (m_restricted && m_method != Get) { - reply.status = MHD_HTTP_FORBIDDEN; - return false; - } - - if (m_method == Get) { - return true; - } - - const char *contentType = MHD_lookup_connection_value(m_connection, MHD_HEADER_KIND, "Content-Type"); - if (!contentType || strcmp(contentType, "application/json") != 0) { - reply.status = MHD_HTTP_UNSUPPORTED_MEDIA_TYPE; - return false; - } - - m_body = new xmrig::HttpBody(); - m_fulfilled = false; - *m_cls = m_body; - - return true; -} - - -const char *xmrig::HttpRequest::body() const -{ - return m_body ? m_body->data() : nullptr; -} - - -int xmrig::HttpRequest::end(const HttpReply &reply) -{ - if (reply.buf) { - return end(reply.status, MHD_create_response_from_buffer(reply.size ? reply.size : strlen(reply.buf), (void*) reply.buf, MHD_RESPMEM_MUST_FREE)); - } - - return end(reply.status, nullptr); -} - - -int xmrig::HttpRequest::end(int status, MHD_Response *rsp) -{ - if (!rsp) { - rsp = MHD_create_response_from_buffer(0, nullptr, MHD_RESPMEM_PERSISTENT); - } - - MHD_add_response_header(rsp, "Content-Type", "application/json"); - MHD_add_response_header(rsp, "Access-Control-Allow-Origin", "*"); - MHD_add_response_header(rsp, "Access-Control-Allow-Methods", "GET, PUT"); - MHD_add_response_header(rsp, "Access-Control-Allow-Headers", "Authorization, Content-Type"); - - const int ret = MHD_queue_response(m_connection, status, rsp); - MHD_destroy_response(rsp); - return ret; -} - - -int xmrig::HttpRequest::auth(const char *accessToken) -{ - if (!accessToken) { - return MHD_HTTP_OK; - } - - const char *header = MHD_lookup_connection_value(m_connection, MHD_HEADER_KIND, "Authorization"); - if (accessToken && !header) { - return MHD_HTTP_UNAUTHORIZED; - } - - const size_t size = strlen(header); - if (size < 8 || strlen(accessToken) != size - 7 || memcmp("Bearer ", header, 7) != 0) { - return MHD_HTTP_FORBIDDEN; - } - - return strncmp(accessToken, header + 7, strlen(accessToken)) == 0 ? MHD_HTTP_OK : MHD_HTTP_FORBIDDEN; -} diff --git a/src/common/api/HttpRequest.h b/src/common/api/HttpRequest.h deleted file mode 100644 index f6ff9a402..000000000 --- a/src/common/api/HttpRequest.h +++ /dev/null @@ -1,84 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef __HTTPREQUEST_H__ -#define __HTTPREQUEST_H__ - - -#include - - -struct MHD_Connection; -struct MHD_Response; - - -namespace xmrig { - - -class HttpBody; -class HttpReply; - - -class HttpRequest -{ -public: - enum Method { - Unsupported, - Options, - Get, - Put - }; - - HttpRequest(MHD_Connection *connection, const char *url, const char *method, const char *uploadData, size_t *uploadSize, void **cls); - ~HttpRequest(); - - inline bool isFulfilled() const { return m_fulfilled; } - inline bool isRestricted() const { return m_restricted; } - inline Method method() const { return m_method; } - - bool match(const char *path) const; - bool process(const char *accessToken, bool restricted, xmrig::HttpReply &reply); - const char *body() const; - int end(const HttpReply &reply); - int end(int status, MHD_Response *rsp); - -private: - int auth(const char *accessToken); - - bool m_fulfilled; - bool m_restricted; - const char *m_uploadData; - const char *m_url; - HttpBody *m_body; - Method m_method; - MHD_Connection *m_connection; - size_t *m_uploadSize; - void **m_cls; -}; - - -} /* namespace xmrig */ - - -#endif /* __HTTPREQUEST_H__ */ diff --git a/src/common/api/Httpd.cpp b/src/common/api/Httpd.cpp deleted file mode 100644 index eb6a4ba61..000000000 --- a/src/common/api/Httpd.cpp +++ /dev/null @@ -1,148 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - - -#include -#include - - -#include "api/Api.h" -#include "common/api/Httpd.h" -#include "common/api/HttpReply.h" -#include "common/api/HttpRequest.h" -#include "common/log/Log.h" - - -Httpd::Httpd(int port, const char *accessToken, bool IPv6, bool restricted) : - m_idle(true), - m_IPv6(IPv6), - m_restricted(restricted), - m_accessToken(accessToken ? strdup(accessToken) : nullptr), - m_port(port), - m_daemon(nullptr) -{ - uv_timer_init(uv_default_loop(), &m_timer); - m_timer.data = this; -} - - -Httpd::~Httpd() -{ - uv_timer_stop(&m_timer); - - if (m_daemon) { - MHD_stop_daemon(m_daemon); - } - - delete m_accessToken; -} - - -bool Httpd::start() -{ - if (!m_port) { - return false; - } - - unsigned int flags = 0; -# if MHD_VERSION >= 0x00093500 - if (m_IPv6 && MHD_is_feature_supported(MHD_FEATURE_IPv6)) { - flags |= MHD_USE_DUAL_STACK; - } - - if (MHD_is_feature_supported(MHD_FEATURE_EPOLL)) { - flags |= MHD_USE_EPOLL_LINUX_ONLY; - } -# endif - - m_daemon = MHD_start_daemon(flags, m_port, nullptr, nullptr, &Httpd::handler, this, MHD_OPTION_END); - if (!m_daemon) { - LOG_ERR("HTTP Daemon failed to start."); - return false; - } - -# if MHD_VERSION >= 0x00093900 - uv_timer_start(&m_timer, Httpd::onTimer, kIdleInterval, kIdleInterval); -# else - uv_timer_start(&m_timer, Httpd::onTimer, kActiveInterval, kActiveInterval); -# endif - - return true; -} - - -int Httpd::process(xmrig::HttpRequest &req) -{ - xmrig::HttpReply reply; - if (!req.process(m_accessToken, m_restricted, reply)) { - return req.end(reply); - } - - if (!req.isFulfilled()) { - return MHD_YES; - } - - Api::exec(req, reply); - - return req.end(reply); -} - - -void Httpd::run() -{ - MHD_run(m_daemon); - -# if MHD_VERSION >= 0x00093900 - const MHD_DaemonInfo *info = MHD_get_daemon_info(m_daemon, MHD_DAEMON_INFO_CURRENT_CONNECTIONS); - if (m_idle && info->num_connections) { - uv_timer_set_repeat(&m_timer, kActiveInterval); - m_idle = false; - } - else if (!m_idle && !info->num_connections) { - uv_timer_set_repeat(&m_timer, kIdleInterval); - m_idle = true; - } -# endif -} - - -int Httpd::handler(void *cls, struct MHD_Connection *connection, const char *url, const char *method, const char *version, const char *uploadData, size_t *uploadSize, void **con_cls) -{ - xmrig::HttpRequest req(connection, url, method, uploadData, uploadSize, con_cls); - - if (req.method() == xmrig::HttpRequest::Options) { - return req.end(MHD_HTTP_OK, nullptr); - } - - if (req.method() == xmrig::HttpRequest::Unsupported) { - return req.end(MHD_HTTP_METHOD_NOT_ALLOWED, nullptr); - } - - return static_cast(cls)->process(req); -} - - -void Httpd::onTimer(uv_timer_t *handle) -{ - static_cast(handle->data)->run(); -} diff --git a/src/common/api/Httpd.h b/src/common/api/Httpd.h deleted file mode 100644 index adec1d716..000000000 --- a/src/common/api/Httpd.h +++ /dev/null @@ -1,70 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef __HTTPD_H__ -#define __HTTPD_H__ - - -#include - - -struct MHD_Connection; -struct MHD_Daemon; -struct MHD_Response; - - -class UploadCtx; - - -namespace xmrig { - class HttpRequest; -} - - -class Httpd -{ -public: - Httpd(int port, const char *accessToken, bool IPv6, bool restricted); - ~Httpd(); - bool start(); - -private: - constexpr static const int kIdleInterval = 200; - constexpr static const int kActiveInterval = 25; - - int process(xmrig::HttpRequest &req); - void run(); - - static int handler(void *cls, MHD_Connection *connection, const char *url, const char *method, const char *version, const char *uploadData, size_t *uploadSize, void **con_cls); - static void onTimer(uv_timer_t *handle); - - bool m_idle; - bool m_IPv6; - bool m_restricted; - const char *m_accessToken; - const int m_port; - MHD_Daemon *m_daemon; - uv_timer_t m_timer; -}; - -#endif /* __HTTPD_H__ */ diff --git a/src/common/config/CommonConfig.cpp b/src/common/config/CommonConfig.cpp deleted file mode 100644 index 36d156a31..000000000 --- a/src/common/config/CommonConfig.cpp +++ /dev/null @@ -1,422 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - - -#include -#include -#include -#include -#include - - -#ifndef XMRIG_NO_HTTPD -# include -#endif - - -#ifndef XMRIG_NO_TLS -# include -#endif - - -#ifdef XMRIG_AMD_PROJECT -# if defined(__APPLE__) -# include -# else -# include "3rdparty/CL/cl.h" -# endif -#endif - - -#ifdef XMRIG_NVIDIA_PROJECT -# include "nvidia/cryptonight.h" -#endif - - -#include "base/io/Json.h" -#include "common/config/CommonConfig.h" -#include "common/log/Log.h" -#include "donate.h" -#include "rapidjson/document.h" -#include "rapidjson/filewritestream.h" -#include "rapidjson/prettywriter.h" -#include "version.h" - - -xmrig::CommonConfig::CommonConfig() : - m_algorithm(CRYPTONIGHT, VARIANT_AUTO), - m_adjusted(false), - m_apiIPv6(false), - m_apiRestricted(true), - m_autoSave(true), - m_background(false), - m_dryRun(false), - m_syslog(false), - m_watch(true), - m_apiPort(0), - m_donateLevel(kDefaultDonateLevel), - m_printTime(60), - m_state(NoneState) -{ -} - - -bool xmrig::CommonConfig::isColors() const -{ - return Log::colors; -} - - -void xmrig::CommonConfig::printAPI() -{ -# ifndef XMRIG_NO_API - if (apiPort() == 0) { - return; - } - - Log::i()->text(isColors() ? GREEN_BOLD(" * ") WHITE_BOLD("%-13s") CYAN("%s:") CYAN_BOLD("%d") - : " * %-13s%s:%d", - "API BIND", isApiIPv6() ? "[::]" : "0.0.0.0", apiPort()); -# endif -} - - -void xmrig::CommonConfig::printPools() -{ - m_pools.print(); -} - - -void xmrig::CommonConfig::printVersions() -{ - char buf[256] = { 0 }; - -# if defined(__clang__) - snprintf(buf, sizeof buf, "clang/%d.%d.%d", __clang_major__, __clang_minor__, __clang_patchlevel__); -# elif defined(__GNUC__) - snprintf(buf, sizeof buf, "gcc/%d.%d.%d", __GNUC__, __GNUC_MINOR__, __GNUC_PATCHLEVEL__); -# elif defined(_MSC_VER) - snprintf(buf, sizeof buf, "MSVC/%d", MSVC_VERSION); -# endif - - Log::i()->text(isColors() ? GREEN_BOLD(" * ") WHITE_BOLD("%-13s") CYAN_BOLD("%s/%s") WHITE_BOLD(" %s") - : " * %-13s%s/%s %s", - "ABOUT", APP_NAME, APP_VERSION, buf); - -# if defined(XMRIG_AMD_PROJECT) -# if CL_VERSION_2_0 - const char *ocl = "2.0"; -# elif CL_VERSION_1_2 - const char *ocl = "1.2"; -# elif CL_VERSION_1_1 - const char *ocl = "1.1"; -# elif CL_VERSION_1_0 - const char *ocl = "1.0"; -# else - const char *ocl = "0.0"; -# endif - int length = snprintf(buf, sizeof buf, "OpenCL/%s ", ocl); -# elif defined(XMRIG_NVIDIA_PROJECT) - const int cudaVersion = cuda_get_runtime_version(); - int length = snprintf(buf, sizeof buf, "CUDA/%d.%d ", cudaVersion / 1000, cudaVersion % 100); -# else - memset(buf, 0, 16); - -# if !defined(XMRIG_NO_HTTPD) || !defined(XMRIG_NO_TLS) - int length = 0; -# endif -# endif - -# if !defined(XMRIG_NO_TLS) && defined(OPENSSL_VERSION_TEXT) - { - constexpr const char *v = OPENSSL_VERSION_TEXT + 8; - length += snprintf(buf + length, (sizeof buf) - length, "OpenSSL/%.*s ", static_cast(strchr(v, ' ') - v), v); - } -# endif - -# ifndef XMRIG_NO_HTTPD - length += snprintf(buf + length, (sizeof buf) - length, "microhttpd/%s ", MHD_get_version()); -# endif - - Log::i()->text(isColors() ? GREEN_BOLD(" * ") WHITE_BOLD("%-13slibuv/%s %s") - : " * %-13slibuv/%s %s", - "LIBS", uv_version_string(), buf); -} - - -bool xmrig::CommonConfig::save() -{ - if (m_fileName.isNull()) { - return false; - } - - rapidjson::Document doc; - getJSON(doc); - - if (Json::save(m_fileName, doc)) { - LOG_NOTICE("configuration saved to: \"%s\"", m_fileName.data()); - return true; - } - - return false; -} - - -bool xmrig::CommonConfig::finalize() -{ - if (m_state == ReadyState) { - return true; - } - - if (m_state == ErrorState) { - return false; - } - - if (!m_algorithm.isValid()) { - return false; - } - - m_pools.adjust(m_algorithm); - - if (!m_pools.active()) { - m_state = ErrorState; - return false; - } - - m_state = ReadyState; - return true; -} - - -bool xmrig::CommonConfig::parseBoolean(int key, bool enable) -{ - switch (key) { - case BackgroundKey: /* --background */ - m_background = enable; - break; - - case SyslogKey: /* --syslog */ - m_syslog = enable; - break; - - case KeepAliveKey: /* --keepalive */ - m_pools.setKeepAlive(enable); - break; - - case TlsKey: /* --tls */ - m_pools.setTLS(enable); - break; - -# ifndef XMRIG_PROXY_PROJECT - case NicehashKey: /* --nicehash */ - m_pools.setNicehash(enable); - break; -# endif - - case ColorKey: /* --no-color */ - Log::colors = enable; - break; - - case WatchKey: /* watch */ - m_watch = enable; - break; - - case ApiIPv6Key: /* ipv6 */ - m_apiIPv6 = enable; - break; - - case ApiRestrictedKey: /* restricted */ - m_apiRestricted = enable; - break; - - case DryRunKey: /* --dry-run */ - m_dryRun = enable; - break; - - case AutoSaveKey: - m_autoSave = enable; - break; - - default: - break; - } - - return true; -} - - -bool xmrig::CommonConfig::parseString(int key, const char *arg) -{ - switch (key) { - case AlgorithmKey: /* --algo */ - m_algorithm.parseAlgorithm(arg); - break; - - case UserpassKey: /* --userpass */ - return m_pools.setUserpass(arg); - - case UrlKey: /* --url */ - return m_pools.setUrl(arg); - - case UserKey: /* --user */ - m_pools.setUser(arg); - break; - - case PasswordKey: /* --pass */ - m_pools.setPassword(arg); - break; - - case RigIdKey: /* --rig-id */ - m_pools.setRigId(arg); - break; - - case FingerprintKey: /* --tls-fingerprint */ - m_pools.setFingerprint(arg); - break; - - case VariantKey: /* --variant */ - m_pools.setVariant(arg); - break; - - case LogFileKey: /* --log-file */ - m_logFile = arg; - break; - - case ApiAccessTokenKey: /* --api-access-token */ - m_apiToken = arg; - break; - - case ApiWorkerIdKey: /* --api-worker-id */ - m_apiWorkerId = arg; - break; - - case ApiIdKey: /* --api-id */ - m_apiId = arg; - break; - - case UserAgentKey: /* --user-agent */ - m_userAgent = arg; - break; - - case RetriesKey: /* --retries */ - case RetryPauseKey: /* --retry-pause */ - case ApiPort: /* --api-port */ - case PrintTimeKey: /* --print-time */ - return parseUint64(key, strtol(arg, nullptr, 10)); - - case BackgroundKey: /* --background */ - case SyslogKey: /* --syslog */ - case KeepAliveKey: /* --keepalive */ - case NicehashKey: /* --nicehash */ - case TlsKey: /* --tls */ - case ApiIPv6Key: /* --api-ipv6 */ - case DryRunKey: /* --dry-run */ - return parseBoolean(key, true); - - case ColorKey: /* --no-color */ - case WatchKey: /* --no-watch */ - case ApiRestrictedKey: /* --api-no-restricted */ - return parseBoolean(key, false); - - case DonateLevelKey: /* --donate-level */ -# ifdef XMRIG_PROXY_PROJECT - if (strncmp(arg, "minemonero.pro", 14) == 0) { - m_donateLevel = 0; - return true; - } -# endif - return parseUint64(key, strtol(arg, nullptr, 10)); - - default: - break; - } - - return true; -} - - -bool xmrig::CommonConfig::parseUint64(int key, uint64_t arg) -{ - return parseInt(key, static_cast(arg)); -} - - -void xmrig::CommonConfig::parseJSON(const rapidjson::Document &doc) -{ - const rapidjson::Value &pools = doc["pools"]; - if (pools.IsArray()) { - m_pools.load(pools); - } -} - - -void xmrig::CommonConfig::setFileName(const char *fileName) -{ - m_fileName = fileName; -} - - -bool xmrig::CommonConfig::parseInt(int key, int arg) -{ - switch (key) { - case RetriesKey: /* --retries */ - m_pools.setRetries(arg); - break; - - case RetryPauseKey: /* --retry-pause */ - m_pools.setRetryPause(arg); - break; - - case KeepAliveKey: /* --keepalive */ - m_pools.setKeepAlive(arg); - break; - - case VariantKey: /* --variant */ - m_pools.setVariant(arg); - break; - - case DonateLevelKey: /* --donate-level */ - if (arg >= kMinimumDonateLevel && arg <= 99) { - m_donateLevel = arg; - } - break; - - case ApiPort: /* --api-port */ - if (arg > 0 && arg <= 65536) { - m_apiPort = arg; - } - break; - - case PrintTimeKey: /* --print-time */ - if (arg >= 0 && arg <= 3600) { - m_printTime = arg; - } - break; - - default: - break; - } - - return true; -} diff --git a/src/common/config/ConfigLoader.cpp b/src/common/config/ConfigLoader.cpp deleted file mode 100644 index 26742e5f2..000000000 --- a/src/common/config/ConfigLoader.cpp +++ /dev/null @@ -1,272 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - - -#include -#include -#include -#include - - -#ifndef XMRIG_NO_HTTPD -# include -#endif - - -#ifndef XMRIG_NO_TLS -# include -#endif - - -#include "base/io/Json.h" -#include "base/kernel/interfaces/IConfigListener.h" -#include "base/kernel/Process.h" -#include "common/config/ConfigLoader.h" -#include "common/config/ConfigWatcher.h" -#include "common/interfaces/IConfig.h" -#include "common/Platform.h" -#include "core/ConfigCreator.h" -#include "core/ConfigLoader_platform.h" -#include "rapidjson/document.h" -#include "rapidjson/error/en.h" -#include "rapidjson/fwd.h" - - -#ifdef XMRIG_FEATURE_EMBEDDED_CONFIG -# include "core/ConfigLoader_default.h" -#endif - - -xmrig::ConfigWatcher *xmrig::ConfigLoader::m_watcher = nullptr; -xmrig::IConfigCreator *xmrig::ConfigLoader::m_creator = nullptr; -xmrig::IConfigListener *xmrig::ConfigLoader::m_listener = nullptr; - - -#ifndef ARRAY_SIZE -# define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0])) -#endif - - -bool xmrig::ConfigLoader::loadFromFile(xmrig::IConfig *config, const char *fileName) -{ - rapidjson::Document doc; - if (!getJSON(fileName, doc)) { - return false; - } - - config->setFileName(fileName); - - return loadFromJSON(config, doc); -} - - -bool xmrig::ConfigLoader::loadFromJSON(xmrig::IConfig *config, const char *json) -{ - using namespace rapidjson; - Document doc; - doc.Parse(json); - - if (doc.HasParseError() || !doc.IsObject()) { - return false; - } - - return loadFromJSON(config, doc); -} - - -bool xmrig::ConfigLoader::loadFromJSON(xmrig::IConfig *config, const rapidjson::Document &doc) -{ - for (size_t i = 0; i < ARRAY_SIZE(config_options); i++) { - parseJSON(config, &config_options[i], doc); - } - - const rapidjson::Value &api = doc["api"]; - if (api.IsObject()) { - for (size_t i = 0; i < ARRAY_SIZE(api_options); i++) { - parseJSON(config, &api_options[i], api); - } - } - - config->parseJSON(doc); - - return config->finalize(); -} - - -bool xmrig::ConfigLoader::reload(xmrig::IConfig *oldConfig, const char *json) -{ - xmrig::IConfig *config = m_creator->create(); - if (!loadFromJSON(config, json)) { - delete config; - - return false; - } - - config->setFileName(oldConfig->fileName()); - const bool saved = config->save(); - - if (config->isWatch() && m_watcher && saved) { - delete config; - - return true; - } - - m_listener->onNewConfig(config); - return true; -} - - -bool xmrig::ConfigLoader::watch(IConfig *config) -{ - if (!config->isWatch()) { - return false; - } - - assert(m_watcher == nullptr); - - m_watcher = new xmrig::ConfigWatcher(config->fileName(), m_creator, m_listener); - return true; -} - - -xmrig::IConfig *xmrig::ConfigLoader::load(Process *process, IConfigCreator *creator, IConfigListener *listener) -{ - m_creator = creator; - m_listener = listener; - - xmrig::IConfig *config = m_creator->create(); - int key; - int argc = process->arguments().argc(); - char **argv = process->arguments().argv(); - - while (1) { - key = getopt_long(argc, argv, short_options, options, nullptr); - if (key < 0) { - break; - } - - if (!parseArg(config, key, optarg)) { - delete config; - return nullptr; - } - } - - if (optind < argc) { - fprintf(stderr, "%s: unsupported non-option argument '%s'\n", argv[0], argv[optind]); - delete config; - return nullptr; - } - - if (!config->finalize()) { - delete config; - - config = m_creator->create(); - loadFromFile(config, process->location(Process::ExeLocation, "config.json")); - } - -# ifdef XMRIG_FEATURE_EMBEDDED_CONFIG - if (!config->finalize()) { - delete config; - - config = m_creator->create(); - loadFromJSON(config, default_config); - } -# endif - - if (!config->finalize()) { - if (!config->algorithm().isValid()) { - fprintf(stderr, "No valid algorithm specified. Exiting.\n"); - } - else { - fprintf(stderr, "No valid configuration found. Exiting.\n"); - } - - delete config; - return nullptr; - } - - return config; -} - - -void xmrig::ConfigLoader::release() -{ - delete m_watcher; - delete m_creator; - - m_watcher = nullptr; - m_creator = nullptr; -} - - -bool xmrig::ConfigLoader::getJSON(const char *fileName, rapidjson::Document &doc) -{ - if (Json::get(fileName, doc)) { - return true; - } - - if (doc.HasParseError()) { - printf("%s: \"%s\"\n", fileName, doc.GetErrorOffset(), rapidjson::GetParseError_En(doc.GetParseError())); - } - else { - fprintf(stderr, "unable to open \"%s\".\n", fileName); - } - - return false; -} - - -bool xmrig::ConfigLoader::parseArg(xmrig::IConfig *config, int key, const char *arg) -{ - if (key == xmrig::IConfig::ConfigKey) { - return loadFromFile(config, arg); - } - - return config->parseString(key, arg); -} - - -void xmrig::ConfigLoader::parseJSON(xmrig::IConfig *config, const struct option *option, const rapidjson::Value &object) -{ - if (!option->name || !object.HasMember(option->name)) { - return; - } - - const rapidjson::Value &value = object[option->name]; - - if (option->has_arg) { - if (value.IsString()) { - config->parseString(option->val, value.GetString()); - } - else if (value.IsInt64()) { - config->parseUint64(option->val, value.GetUint64()); - } - else if (value.IsBool()) { - config->parseBoolean(option->val, value.IsTrue()); - } - } - else if (value.IsBool()) { - config->parseBoolean(option->val, value.IsTrue()); - } -} diff --git a/src/common/cpu/BasicCpuInfo.cpp b/src/common/cpu/BasicCpuInfo.cpp deleted file mode 100644 index d7778bdde..000000000 --- a/src/common/cpu/BasicCpuInfo.cpp +++ /dev/null @@ -1,159 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2019 XMR-Stak , - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#include -#include - - -#ifdef _MSC_VER -# include -#else -# include -#endif - -#ifndef bit_AES -# define bit_AES (1 << 25) -#endif - -#ifndef bit_OSXSAVE -# define bit_OSXSAVE (1 << 27) -#endif - -#ifndef bit_AVX2 -# define bit_AVX2 (1 << 5) -#endif - - -#include "common/cpu/BasicCpuInfo.h" - - -#define VENDOR_ID (0) -#define PROCESSOR_INFO (1) -#define CACHE_TLB_DESCRIPTOR (2) -#define EXTENDED_FEATURES (7) -#define PROCESSOR_BRAND_STRING_1 (0x80000002) -#define PROCESSOR_BRAND_STRING_2 (0x80000003) -#define PROCESSOR_BRAND_STRING_3 (0x80000004) - -#define EAX_Reg (0) -#define EBX_Reg (1) -#define ECX_Reg (2) -#define EDX_Reg (3) - - -#ifdef _MSC_VER -static inline void cpuid(int level, int output[4]) { - __cpuid(output, level); -} -#else -static inline void cpuid(int level, int output[4]) { - int a, b, c, d; - __cpuid_count(level, 0, a, b, c, d); - - output[0] = a; - output[1] = b; - output[2] = c; - output[3] = d; -} -#endif - - -static inline void cpu_brand_string(char* s) { - int32_t cpu_info[4] = { 0 }; - cpuid(VENDOR_ID, cpu_info); - - if (cpu_info[EAX_Reg] >= 4) { - for (int i = 0; i < 4; i++) { - cpuid(0x80000002 + i, cpu_info); - memcpy(s, cpu_info, sizeof(cpu_info)); - s += 16; - } - } -} - - -static inline bool has_aes_ni() -{ - int32_t cpu_info[4] = { 0 }; - cpuid(PROCESSOR_INFO, cpu_info); - - return (cpu_info[ECX_Reg] & bit_AES) != 0; -} - - -static inline bool has_avx2() -{ - int32_t cpu_info[4] = { 0 }; - cpuid(EXTENDED_FEATURES, cpu_info); - - return (cpu_info[EBX_Reg] & bit_AVX2) != 0; -} - - -static inline bool has_ossave() -{ - int32_t cpu_info[4] = { 0 }; - cpuid(PROCESSOR_INFO, cpu_info); - - return (cpu_info[ECX_Reg] & bit_OSXSAVE) != 0; -} - - -xmrig::BasicCpuInfo::BasicCpuInfo() : - m_assembly(ASM_NONE), - m_aes(has_aes_ni()), - m_avx2(has_avx2() && has_ossave()), - m_brand(), - m_threads(std::thread::hardware_concurrency()) -{ - cpu_brand_string(m_brand); - -# ifndef XMRIG_NO_ASM - if (hasAES()) { - char vendor[13] = { 0 }; - int32_t data[4] = { 0 }; - - cpuid(0, data); - - memcpy(vendor + 0, &data[1], 4); - memcpy(vendor + 4, &data[3], 4); - memcpy(vendor + 8, &data[2], 4); - - if (memcmp(vendor, "GenuineIntel", 12) == 0) { - m_assembly = ASM_INTEL; - } - else if (memcmp(vendor, "AuthenticAMD", 12) == 0) { - m_assembly = ASM_RYZEN; - } - } -# endif -} - - -size_t xmrig::BasicCpuInfo::optimalThreadsCount(size_t memSize, int maxCpuUsage) const -{ - const size_t count = threads() / 2; - - return count < 1 ? 1 : count; -} diff --git a/src/common/cpu/Cpu.cpp b/src/common/cpu/Cpu.cpp deleted file mode 100644 index b1bb28ac7..000000000 --- a/src/common/cpu/Cpu.cpp +++ /dev/null @@ -1,57 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2016-2017 XMRig - * - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - - -#include - - -#include "common/cpu/BasicCpuInfo.h" -#include "common/cpu/Cpu.h" - - -static xmrig::ICpuInfo *cpuInfo = nullptr; - - -xmrig::ICpuInfo *xmrig::Cpu::info() -{ - assert(cpuInfo != nullptr); - - return cpuInfo; -} - - -void xmrig::Cpu::init() -{ - assert(cpuInfo == nullptr); - - cpuInfo = new BasicCpuInfo(); -} - - -void xmrig::Cpu::release() -{ - assert(cpuInfo != nullptr); - - delete cpuInfo; - cpuInfo = nullptr; -} diff --git a/src/common/crypto/Algorithm.cpp b/src/common/crypto/Algorithm.cpp deleted file mode 100644 index f14d034de..000000000 --- a/src/common/crypto/Algorithm.cpp +++ /dev/null @@ -1,297 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2018 Lee Clagett - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - - -#include -#include -#include -#include - - -#include "common/crypto/Algorithm.h" - - -#ifdef _MSC_VER -# define strncasecmp _strnicmp -# define strcasecmp _stricmp -#endif - - -#ifndef ARRAY_SIZE -# define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0])) -#endif - - -struct AlgoData -{ - const char *name; - const char *shortName; - xmrig::Algo algo; - xmrig::Variant variant; -}; - - -static AlgoData const algorithms[] = { - { "cryptonight", "cn", xmrig::CRYPTONIGHT, xmrig::VARIANT_AUTO }, - { "cryptonight/0", "cn/0", xmrig::CRYPTONIGHT, xmrig::VARIANT_0 }, - { "cryptonight/1", "cn/1", xmrig::CRYPTONIGHT, xmrig::VARIANT_1 }, - { "cryptonight/xtl", "cn/xtl", xmrig::CRYPTONIGHT, xmrig::VARIANT_XTL }, - { "cryptonight/msr", "cn/msr", xmrig::CRYPTONIGHT, xmrig::VARIANT_MSR }, - { "cryptonight/xao", "cn/xao", xmrig::CRYPTONIGHT, xmrig::VARIANT_XAO }, - { "cryptonight/rto", "cn/rto", xmrig::CRYPTONIGHT, xmrig::VARIANT_RTO }, - { "cryptonight/2", "cn/2", xmrig::CRYPTONIGHT, xmrig::VARIANT_2 }, - { "cryptonight/half", "cn/half", xmrig::CRYPTONIGHT, xmrig::VARIANT_HALF }, - { "cryptonight/xtlv9", "cn/xtlv9", xmrig::CRYPTONIGHT, xmrig::VARIANT_HALF }, - { "cryptonight/wow", "cn/wow", xmrig::CRYPTONIGHT, xmrig::VARIANT_WOW }, - { "cryptonight/r", "cn/r", xmrig::CRYPTONIGHT, xmrig::VARIANT_4 }, - { "cryptonight/rwz", "cn/rwz", xmrig::CRYPTONIGHT, xmrig::VARIANT_RWZ }, - { "cryptonight/zls", "cn/zls", xmrig::CRYPTONIGHT, xmrig::VARIANT_ZLS }, - { "cryptonight/double", "cn/double", xmrig::CRYPTONIGHT, xmrig::VARIANT_DOUBLE }, - -# ifndef XMRIG_NO_AEON - { "cryptonight-lite", "cn-lite", xmrig::CRYPTONIGHT_LITE, xmrig::VARIANT_AUTO }, - { "cryptonight-light", "cn-light", xmrig::CRYPTONIGHT_LITE, xmrig::VARIANT_AUTO }, - { "cryptonight-lite/0", "cn-lite/0", xmrig::CRYPTONIGHT_LITE, xmrig::VARIANT_0 }, - { "cryptonight-lite/1", "cn-lite/1", xmrig::CRYPTONIGHT_LITE, xmrig::VARIANT_1 }, -# endif - -# ifndef XMRIG_NO_SUMO - { "cryptonight-heavy", "cn-heavy", xmrig::CRYPTONIGHT_HEAVY, xmrig::VARIANT_AUTO }, - { "cryptonight-heavy/0", "cn-heavy/0", xmrig::CRYPTONIGHT_HEAVY, xmrig::VARIANT_0 }, - { "cryptonight-heavy/xhv", "cn-heavy/xhv", xmrig::CRYPTONIGHT_HEAVY, xmrig::VARIANT_XHV }, - { "cryptonight-heavy/tube", "cn-heavy/tube", xmrig::CRYPTONIGHT_HEAVY, xmrig::VARIANT_TUBE }, -# endif - -# ifndef XMRIG_NO_CN_PICO - { "cryptonight-pico/trtl", "cn-pico/trtl", xmrig::CRYPTONIGHT_PICO, xmrig::VARIANT_TRTL }, - { "cryptonight-pico", "cn-pico", xmrig::CRYPTONIGHT_PICO, xmrig::VARIANT_TRTL }, - { "cryptonight-turtle", "cn-trtl", xmrig::CRYPTONIGHT_PICO, xmrig::VARIANT_TRTL }, - { "cryptonight-ultralite", "cn-ultralite", xmrig::CRYPTONIGHT_PICO, xmrig::VARIANT_TRTL }, - { "cryptonight_turtle", "cn_turtle", xmrig::CRYPTONIGHT_PICO, xmrig::VARIANT_TRTL }, -# endif - -# ifndef XMRIG_NO_CN_GPU - { "cryptonight/gpu", "cn/gpu", xmrig::CRYPTONIGHT, xmrig::VARIANT_GPU }, -# endif -}; - - -#ifdef XMRIG_PROXY_PROJECT -static AlgoData const xmrStakAlgorithms[] = { - { "cryptonight-monerov7", nullptr, xmrig::CRYPTONIGHT, xmrig::VARIANT_1 }, - { "cryptonight_v7", nullptr, xmrig::CRYPTONIGHT, xmrig::VARIANT_1 }, - { "cryptonight-monerov8", nullptr, xmrig::CRYPTONIGHT, xmrig::VARIANT_2 }, - { "cryptonight_v8", nullptr, xmrig::CRYPTONIGHT, xmrig::VARIANT_2 }, - { "cryptonight_v7_stellite", nullptr, xmrig::CRYPTONIGHT, xmrig::VARIANT_XTL }, - { "cryptonight_lite", nullptr, xmrig::CRYPTONIGHT_LITE, xmrig::VARIANT_0 }, - { "cryptonight-aeonv7", nullptr, xmrig::CRYPTONIGHT_LITE, xmrig::VARIANT_1 }, - { "cryptonight_lite_v7", nullptr, xmrig::CRYPTONIGHT_LITE, xmrig::VARIANT_1 }, - { "cryptonight_heavy", nullptr, xmrig::CRYPTONIGHT_HEAVY, xmrig::VARIANT_0 }, - { "cryptonight_haven", nullptr, xmrig::CRYPTONIGHT_HEAVY, xmrig::VARIANT_XHV }, - { "cryptonight_masari", nullptr, xmrig::CRYPTONIGHT, xmrig::VARIANT_MSR }, - { "cryptonight_masari", nullptr, xmrig::CRYPTONIGHT, xmrig::VARIANT_MSR }, - { "cryptonight-bittube2", nullptr, xmrig::CRYPTONIGHT_HEAVY, xmrig::VARIANT_TUBE }, // bittube-miner - { "cryptonight_alloy", nullptr, xmrig::CRYPTONIGHT, xmrig::VARIANT_XAO }, // xmr-stak-alloy - { "cryptonight_turtle", nullptr, xmrig::CRYPTONIGHT_PICO, xmrig::VARIANT_TRTL }, - { "cryptonight_gpu", nullptr, xmrig::CRYPTONIGHT, xmrig::VARIANT_GPU }, - { "cryptonight_r", nullptr, xmrig::CRYPTONIGHT, xmrig::VARIANT_4 }, -}; -#endif - - -static const char *variants[] = { - "0", - "1", - "tube", - "xtl", - "msr", - "xhv", - "xao", - "rto", - "2", - "half", - "trtl", - "gpu", - "wow", - "r", - "rwz", - "zls", - "double" -}; - - -static_assert(xmrig::VARIANT_MAX == ARRAY_SIZE(variants), "variants size mismatch"); - - -bool xmrig::Algorithm::isValid() const -{ - if (m_algo == INVALID_ALGO) { - return false; - } - - for (size_t i = 0; i < ARRAY_SIZE(algorithms); i++) { - if (algorithms[i].algo == m_algo && algorithms[i].variant == m_variant) { - return true; - } - } - - return false; -} - - -const char *xmrig::Algorithm::variantName() const -{ - if (m_variant == VARIANT_AUTO) { - return "auto"; - } - - return variants[m_variant]; -} - - -void xmrig::Algorithm::parseAlgorithm(const char *algo) -{ - m_algo = INVALID_ALGO; - m_variant = VARIANT_AUTO; - - assert(algo != nullptr); - if (algo == nullptr || strlen(algo) < 1) { - return; - } - - if (*algo == '!') { - m_flags |= Forced; - - return parseAlgorithm(algo + 1); - } - - for (size_t i = 0; i < ARRAY_SIZE(algorithms); i++) { - if ((strcasecmp(algo, algorithms[i].name) == 0) || (strcasecmp(algo, algorithms[i].shortName) == 0)) { - m_algo = algorithms[i].algo; - m_variant = algorithms[i].variant; - break; - } - } - - if (m_algo == INVALID_ALGO) { - assert(false); - } -} - - -void xmrig::Algorithm::parseVariant(const char *variant) -{ - m_variant = VARIANT_AUTO; - - if (variant == nullptr || strlen(variant) < 1) { - return; - } - - if (*variant == '!') { - m_flags |= Forced; - - return parseVariant(variant + 1); - } - - for (size_t i = 0; i < ARRAY_SIZE(variants); i++) { - if (strcasecmp(variant, variants[i]) == 0) { - m_variant = static_cast(i); - return; - } - } - - if (strcasecmp(variant, "xtlv9") == 0) { - m_variant = VARIANT_HALF; - } -} - - -void xmrig::Algorithm::parseVariant(int variant) -{ - assert(variant >= -1 && variant <= 2); - - switch (variant) { - case -1: - case 0: - case 1: - m_variant = static_cast(variant); - break; - - case 2: - m_variant = VARIANT_2; - break; - - default: - break; - } -} - - -void xmrig::Algorithm::setAlgo(Algo algo) -{ - m_algo = algo; - - if (m_algo == CRYPTONIGHT_PICO && m_variant == VARIANT_AUTO) { - m_variant = xmrig::VARIANT_TRTL; - } -} - - -#ifdef XMRIG_PROXY_PROJECT -void xmrig::Algorithm::parseXmrStakAlgorithm(const char *algo) -{ - m_algo = INVALID_ALGO; - m_variant = VARIANT_AUTO; - - assert(algo != nullptr); - if (algo == nullptr) { - return; - } - - for (size_t i = 0; i < ARRAY_SIZE(xmrStakAlgorithms); i++) { - if (strcasecmp(algo, xmrStakAlgorithms[i].name) == 0) { - m_algo = xmrStakAlgorithms[i].algo; - m_variant = xmrStakAlgorithms[i].variant; - break; - } - } - - if (m_algo == INVALID_ALGO) { - assert(false); - } -} -#endif - - -const char *xmrig::Algorithm::name(bool shortName) const -{ - for (size_t i = 0; i < ARRAY_SIZE(algorithms); i++) { - if (algorithms[i].algo == m_algo && algorithms[i].variant == m_variant) { - return shortName ? algorithms[i].shortName : algorithms[i].name; - } - } - - return "invalid"; -} diff --git a/src/common/crypto/Algorithm.h b/src/common/crypto/Algorithm.h deleted file mode 100644 index 664552aa5..000000000 --- a/src/common/crypto/Algorithm.h +++ /dev/null @@ -1,103 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2018 Lee Clagett - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef XMRIG_ALGORITHM_H -#define XMRIG_ALGORITHM_H - - -#include - - -#include "common/xmrig.h" - - -namespace xmrig { - - -class Algorithm -{ -public: - enum Flags { - None = 0, - Forced = 1 - }; - - inline Algorithm() : - m_algo(INVALID_ALGO), - m_flags(0), - m_variant(VARIANT_AUTO) - {} - - inline Algorithm(Algo algo, Variant variant) : - m_flags(0), - m_variant(variant) - { - setAlgo(algo); - } - - inline Algorithm(const char *algo) : - m_flags(0) - { - parseAlgorithm(algo); - } - - inline Algo algo() const { return m_algo; } - inline bool isEqual(const Algorithm &other) const { return m_algo == other.m_algo && m_variant == other.m_variant; } - inline bool isForced() const { return m_flags & Forced; } - inline const char *name() const { return name(false); } - inline const char *shortName() const { return name(true); } - inline int flags() const { return m_flags; } - inline Variant variant() const { return m_variant; } - inline void setVariant(Variant variant) { m_variant = variant; } - - inline bool operator!=(const Algorithm &other) const { return !isEqual(other); } - inline bool operator==(const Algorithm &other) const { return isEqual(other); } - - bool isValid() const; - const char *variantName() const; - void parseAlgorithm(const char *algo); - void parseVariant(const char *variant); - void parseVariant(int variant); - void setAlgo(Algo algo); - -# ifdef XMRIG_PROXY_PROJECT - void parseXmrStakAlgorithm(const char *algo); -# endif - -private: - const char *name(bool shortName) const; - - Algo m_algo; - int m_flags; - Variant m_variant; -}; - - -typedef std::vector Algorithms; - - -} /* namespace xmrig */ - -#endif /* __ALGORITHM_H__ */ diff --git a/src/common/interfaces/IConfig.h b/src/common/interfaces/IConfig.h deleted file mode 100644 index 7e6931a89..000000000 --- a/src/common/interfaces/IConfig.h +++ /dev/null @@ -1,150 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef XMRIG_ICONFIG_H -#define XMRIG_ICONFIG_H - - -#include "common/crypto/Algorithm.h" -#include "rapidjson/fwd.h" - - -namespace xmrig { - - -class String; - - -class IConfig -{ -public: - enum Keys { - // common - AlgorithmKey = 'a', - ApiAccessTokenKey = 4001, - ApiIPv6Key = 4003, - ApiPort = 4000, - ApiRestrictedKey = 4004, - ApiWorkerIdKey = 4002, - ApiIdKey = 4005, - BackgroundKey = 'B', - ColorKey = 1002, - ConfigKey = 'c', - DonateLevelKey = 1003, - KeepAliveKey = 'k', - LogFileKey = 'l', - PasswordKey = 'p', - RetriesKey = 'r', - RetryPauseKey = 'R', - RigIdKey = 1012, - SyslogKey = 'S', - UrlKey = 'o', - UserAgentKey = 1008, - UserKey = 'u', - UserpassKey = 'O', - VariantKey = 1010, - VerboseKey = 1100, - WatchKey = 1105, - TlsKey = 1013, - FingerprintKey = 1014, - AutoSaveKey = 1016, - - // xmrig common - CPUPriorityKey = 1021, - NicehashKey = 1006, - PrintTimeKey = 1007, - - // xmrig cpu - AVKey = 'v', - CPUAffinityKey = 1020, - DryRunKey = 5000, - HugePagesKey = 1009, - MaxCPUUsageKey = 1004, - SafeKey = 1005, - ThreadsKey = 't', - HardwareAESKey = 1011, - AssemblyKey = 1015, - - // xmrig amd - OclPlatformKey = 1400, - OclAffinityKey = 1401, - OclDevicesKey = 1402, - OclLaunchKey = 1403, - OclCacheKey = 1404, - OclPrintKey = 1405, - OclLoaderKey = 1406, - OclSridedIndexKey = 1407, - OclMemChunkKey = 1408, - OclUnrollKey = 1409, - OclCompModeKey = 1410, - - // xmrig-proxy - AccessLogFileKey = 'A', - BindKey = 'b', - CoinKey = 1104, - CustomDiffKey = 1102, - DebugKey = 1101, - ModeKey = 'm', - PoolCoinKey = 'C', - ReuseTimeoutKey = 1106, - WorkersKey = 1103, - WorkersAdvKey = 1107, - TlsBindKey = 1108, - TlsCertKey = 1109, - TlsCertKeyKey = 1110, - TlsDHparamKey = 1111, - TlsCiphersKey = 1112, - TlsCipherSuitesKey = 1113, - TlsProtocolsKey = 1114, - - // xmrig nvidia - CudaMaxThreadsKey = 1200, - CudaBFactorKey = 1201, - CudaBSleepKey = 1202, - CudaDevicesKey = 1203, - CudaLaunchKey = 1204, - CudaAffinityKey = 1205, - CudaMaxUsageKey = 1206, - }; - - virtual ~IConfig() = default; - - virtual bool finalize() = 0; - virtual bool isWatch() const = 0; - virtual bool parseBoolean(int key, bool enable) = 0; - virtual bool parseString(int key, const char *arg) = 0; - virtual bool parseUint64(int key, uint64_t arg) = 0; - virtual bool save() = 0; - virtual const Algorithm &algorithm() const = 0; - virtual const String &fileName() const = 0; - virtual void getJSON(rapidjson::Document &doc) const = 0; - virtual void parseJSON(const rapidjson::Document &doc) = 0; - virtual void setFileName(const char *fileName) = 0; -}; - - -} /* namespace xmrig */ - - -#endif // XMRIG_ICONFIG_H diff --git a/src/common/interfaces/IConsoleListener.h b/src/common/interfaces/IConsoleListener.h deleted file mode 100644 index 723f87dff..000000000 --- a/src/common/interfaces/IConsoleListener.h +++ /dev/null @@ -1,37 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2016-2017 XMRig - * - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef __ICONSOLELISTENER_H__ -#define __ICONSOLELISTENER_H__ - - -class IConsoleListener -{ -public: - virtual ~IConsoleListener() {} - - virtual void onConsoleCommand(char command) = 0; -}; - - -#endif // __ICONSOLELISTENER_H__ diff --git a/src/common/interfaces/ILogBackend.h b/src/common/interfaces/ILogBackend.h deleted file mode 100644 index 85a04e935..000000000 --- a/src/common/interfaces/ILogBackend.h +++ /dev/null @@ -1,56 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef __ILOGBACKEND_H__ -#define __ILOGBACKEND_H__ - - -#include -#include - - -class ILogBackend -{ -public: - enum Level { - ERR, - WARNING, - NOTICE, - INFO, - DEBUG - }; - -# ifdef APP_DEBUG - constexpr static const size_t kBufferSize = 1024; -# else - constexpr static const size_t kBufferSize = 512; -# endif - - virtual ~ILogBackend() {} - - virtual void message(Level level, const char* fmt, va_list args) = 0; - virtual void text(const char* fmt, va_list args) = 0; -}; - - -#endif // __ILOGBACKEND_H__ diff --git a/src/common/log/BasicLog.cpp b/src/common/log/BasicLog.cpp deleted file mode 100644 index cb4defcde..000000000 --- a/src/common/log/BasicLog.cpp +++ /dev/null @@ -1,89 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - - -#include -#include -#include -#include -#include - -#ifdef WIN32 -# include -# include -#endif - - -#include "common/log/BasicLog.h" -#include "common/log/Log.h" - - -BasicLog::BasicLog() -{ -} - - -void BasicLog::message(Level level, const char* fmt, va_list args) -{ - time_t now = time(nullptr); - tm stime; - -# ifdef _WIN32 - localtime_s(&stime, &now); -# else - localtime_r(&now, &stime); -# endif - - snprintf(m_fmt, sizeof(m_fmt) - 1, "[%d-%02d-%02d %02d:%02d:%02d]%s %s%s", - stime.tm_year + 1900, - stime.tm_mon + 1, - stime.tm_mday, - stime.tm_hour, - stime.tm_min, - stime.tm_sec, - Log::colorByLevel(level, false), - fmt, - Log::endl(false) - ); - - print(args); -} - - -void BasicLog::text(const char* fmt, va_list args) -{ - snprintf(m_fmt, sizeof(m_fmt) - 1, "%s%s", fmt, Log::endl(false)); - - print(args); -} - - -void BasicLog::print(va_list args) -{ - if (vsnprintf(m_buf, sizeof(m_buf) - 1, m_fmt, args) <= 0) { - return; - } - - fputs(m_buf, stdout); - fflush(stdout); -} diff --git a/src/common/log/BasicLog.h b/src/common/log/BasicLog.h deleted file mode 100644 index 523538e95..000000000 --- a/src/common/log/BasicLog.h +++ /dev/null @@ -1,55 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef __BASICLOG_H__ -#define __BASICLOG_H__ - - -#include - - -#include "common/interfaces/ILogBackend.h" - - -namespace xmrig { - class Controller; -} - - -class BasicLog : public ILogBackend -{ -public: - BasicLog(); - - void message(Level level, const char *fmt, va_list args) override; - void text(const char *fmt, va_list args) override; - -private: - bool isWritable() const; - void print(va_list args); - - char m_buf[kBufferSize]; - char m_fmt[256]; -}; - -#endif /* __BASICLOG_H__ */ diff --git a/src/common/log/ConsoleLog.h b/src/common/log/ConsoleLog.h deleted file mode 100644 index bac09a535..000000000 --- a/src/common/log/ConsoleLog.h +++ /dev/null @@ -1,59 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef __CONSOLELOG_H__ -#define __CONSOLELOG_H__ - - -#include - - -#include "common/interfaces/ILogBackend.h" - - -namespace xmrig { - class Controller; -} - - -class ConsoleLog : public ILogBackend -{ -public: - ConsoleLog(xmrig::Controller *controller); - - void message(Level level, const char *fmt, va_list args) override; - void text(const char *fmt, va_list args) override; - -private: - bool isWritable() const; - void print(va_list args); - - char m_buf[kBufferSize]; - char m_fmt[256]; - uv_buf_t m_uvBuf; - uv_stream_t *m_stream; - uv_tty_t m_tty; - xmrig::Controller *m_controller; -}; - -#endif /* __CONSOLELOG_H__ */ diff --git a/src/common/log/FileLog.cpp b/src/common/log/FileLog.cpp deleted file mode 100644 index 9134c7c75..000000000 --- a/src/common/log/FileLog.cpp +++ /dev/null @@ -1,105 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - - -#include -#include -#include -#include -#include - - -#include "common/log/FileLog.h" -#include "common/log/Log.h" -#include "core/Config.h" -#include "core/Controller.h" - - -FileLog::FileLog(xmrig::Controller *controller, const char *fileName) : - m_controller(controller) -{ - uv_fs_t req; - m_file = uv_fs_open(uv_default_loop(), &req, fileName, O_CREAT | O_APPEND | O_WRONLY, 0644, nullptr); - uv_fs_req_cleanup(&req); -} - - -void FileLog::message(Level level, const char* fmt, va_list args) -{ - if (m_file < 0) { - return; - } - - time_t now = time(nullptr); - tm stime; - -# ifdef _WIN32 - localtime_s(&stime, &now); -# else - localtime_r(&now, &stime); -# endif - - const bool isColors = m_controller->config()->isColors(); - - snprintf(m_fmt, sizeof(m_fmt) - 1, "[%d-%02d-%02d %02d:%02d:%02d]%s %s%s", - stime.tm_year + 1900, - stime.tm_mon + 1, - stime.tm_mday, - stime.tm_hour, - stime.tm_min, - stime.tm_sec, - Log::colorByLevel(level, isColors), - fmt, - Log::endl(isColors) - ); - - char *buf = new char[kBufferSize]; - const int size = vsnprintf(buf, kBufferSize - 1, m_fmt, args); - - write(buf, size); -} - - -void FileLog::text(const char* fmt, va_list args) -{ - message(INFO, fmt, args); -} - - -void FileLog::onWrite(uv_fs_t *req) -{ - delete [] static_cast(req->data); - - uv_fs_req_cleanup(req); - delete req; -} - - -void FileLog::write(char *data, size_t size) -{ - uv_buf_t buf = uv_buf_init(data, (unsigned int) size); - uv_fs_t *req = new uv_fs_t; - req->data = buf.base; - - uv_fs_write(uv_default_loop(), req, m_file, &buf, 1, -1, FileLog::onWrite); -} diff --git a/src/common/log/Log.cpp b/src/common/log/Log.cpp deleted file mode 100644 index 07e4f2998..000000000 --- a/src/common/log/Log.cpp +++ /dev/null @@ -1,128 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - - -#include -#include -#include -#include -#include - - -#include "common/interfaces/ILogBackend.h" -#include "common/log/BasicLog.h" -#include "common/log/Log.h" - - -Log *Log::m_self = nullptr; -bool Log::colors = true; - - -static const char *color[5] = { - "\x1B[0;31m", /* ERR */ - "\x1B[0;33m", /* WARNING */ - "\x1B[1;37m", /* NOTICE */ - "", /* INFO */ -# ifdef WIN32 - "\x1B[1;30m" /* DEBUG */ -# else - "\x1B[90m" /* DEBUG */ -# endif -}; - - -void Log::message(ILogBackend::Level level, const char* fmt, ...) -{ - uv_mutex_lock(&m_mutex); - - va_list args; - va_list copy; - va_start(args, fmt); - - for (ILogBackend *backend : m_backends) { - va_copy(copy, args); - backend->message(level, fmt, copy); - va_end(copy); - } - - va_end(args); - - uv_mutex_unlock(&m_mutex); -} - - -void Log::text(const char* fmt, ...) -{ - uv_mutex_lock(&m_mutex); - - va_list args; - va_list copy; - va_start(args, fmt); - - for (ILogBackend *backend : m_backends) { - va_copy(copy, args); - backend->text(fmt, copy); - va_end(copy); - } - - va_end(args); - - uv_mutex_unlock(&m_mutex); -} - - -const char *Log::colorByLevel(ILogBackend::Level level, bool isColors) -{ - if (!isColors) { - return ""; - } - - return color[level]; -} - - -const char *Log::endl(bool isColors) -{ -# ifdef _WIN32 - return isColors ? "\x1B[0m\r\n" : "\r\n"; -# else - return isColors ? "\x1B[0m\n" : "\n"; -# endif -} - - -void Log::defaultInit() -{ - m_self = new Log(); - - add(new BasicLog()); -} - - -Log::~Log() -{ - for (auto backend : m_backends) { - delete backend; - } -} diff --git a/src/common/log/Log.h b/src/common/log/Log.h deleted file mode 100644 index c32edddd4..000000000 --- a/src/common/log/Log.h +++ /dev/null @@ -1,105 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef XMRIG_LOG_H -#define XMRIG_LOG_H - - -#include -#include -#include - - -#include "common/interfaces/ILogBackend.h" - - -class Log -{ -public: - static inline Log* i() { if (!m_self) { defaultInit(); } return m_self; } - static inline void add(ILogBackend *backend) { i()->m_backends.push_back(backend); } - static inline void init() { if (!m_self) { new Log(); } } - static inline void release() { delete m_self; } - - void message(ILogBackend::Level level, const char* fmt, ...); - void text(const char* fmt, ...); - - static const char *colorByLevel(ILogBackend::Level level, bool isColors = true); - static const char *endl(bool isColors = true); - static void defaultInit(); - - static bool colors; - -private: - inline Log() { - assert(m_self == nullptr); - - uv_mutex_init(&m_mutex); - - m_self = this; - } - - ~Log(); - - static Log *m_self; - std::vector m_backends; - uv_mutex_t m_mutex; -}; - - -#define RED_BOLD(x) "\x1B[1;31m" x "\x1B[0m" -#define RED(x) "\x1B[0;31m" x "\x1B[0m" -#define GREEN_BOLD(x) "\x1B[1;32m" x "\x1B[0m" -#define GREEN(x) "\x1B[0;32m" x "\x1B[0m" -#define YELLOW(x) "\x1B[0;33m" x "\x1B[0m" -#define YELLOW_BOLD(x) "\x1B[1;33m" x "\x1B[0m" -#define MAGENTA_BOLD(x) "\x1B[1;35m" x "\x1B[0m" -#define MAGENTA(x) "\x1B[0;35m" x "\x1B[0m" -#define CYAN_BOLD(x) "\x1B[1;36m" x "\x1B[0m" -#define CYAN(x) "\x1B[0;36m" x "\x1B[0m" -#define WHITE_BOLD(x) "\x1B[1;37m" x "\x1B[0m" -#define WHITE(x) "\x1B[0;37m" x "\x1B[0m" -#define GRAY(x) "\x1B[1;30m" x "\x1B[0m" - - -#define LOG_ERR(x, ...) Log::i()->message(ILogBackend::ERR, x, ##__VA_ARGS__) -#define LOG_WARN(x, ...) Log::i()->message(ILogBackend::WARNING, x, ##__VA_ARGS__) -#define LOG_NOTICE(x, ...) Log::i()->message(ILogBackend::NOTICE, x, ##__VA_ARGS__) -#define LOG_INFO(x, ...) Log::i()->message(ILogBackend::INFO, x, ##__VA_ARGS__) - -#ifdef APP_DEBUG -# define LOG_DEBUG(x, ...) Log::i()->message(ILogBackend::DEBUG, x, ##__VA_ARGS__) -#else -# define LOG_DEBUG(x, ...) -#endif - -#if defined(APP_DEBUG) || defined(APP_DEVEL) -# define LOG_DEBUG_ERR(x, ...) Log::i()->message(ILogBackend::ERR, x, ##__VA_ARGS__) -# define LOG_DEBUG_WARN(x, ...) Log::i()->message(ILogBackend::WARNING, x, ##__VA_ARGS__) -#else -# define LOG_DEBUG_ERR(x, ...) -# define LOG_DEBUG_WARN(x, ...) -#endif - -#endif /* XMRIG_LOG_H */ diff --git a/src/common/log/SysLog.cpp b/src/common/log/SysLog.cpp deleted file mode 100644 index bcb963943..000000000 --- a/src/common/log/SysLog.cpp +++ /dev/null @@ -1,47 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - - -#include - - -#include "common/log/SysLog.h" -#include "version.h" - - -SysLog::SysLog() -{ - openlog(APP_ID, LOG_PID, LOG_USER); -} - - -void SysLog::message(Level level, const char *fmt, va_list args) -{ - vsyslog(static_cast(level), fmt, args); -} - - -void SysLog::text(const char *fmt, va_list args) -{ - vsyslog(LOG_INFO, fmt, args); -} diff --git a/src/common/log/SysLog.h b/src/common/log/SysLog.h deleted file mode 100644 index 5cfeefcdb..000000000 --- a/src/common/log/SysLog.h +++ /dev/null @@ -1,40 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef __SYSLOG_H__ -#define __SYSLOG_H__ - - -#include "common/interfaces/ILogBackend.h" - - -class SysLog : public ILogBackend -{ -public: - SysLog(); - - void message(Level level, const char *fmt, va_list args) override; - void text(const char *fmt, va_list args) override; -}; - -#endif /* __SYSLOG_BACKEND_H__ */ diff --git a/src/common/net/Id.h b/src/common/net/Id.h deleted file mode 100644 index 999e7837a..000000000 --- a/src/common/net/Id.h +++ /dev/null @@ -1,98 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef XMRIG_ID_H -#define XMRIG_ID_H - - -#include - - -namespace xmrig { - - -class Id -{ -public: - inline Id() : - m_data() - { - } - - - inline Id(const char *id, size_t sizeFix = 0) - { - setId(id, sizeFix); - } - - - inline bool operator==(const Id &other) const - { - return memcmp(m_data, other.m_data, sizeof(m_data)) == 0; - } - - - inline bool operator!=(const Id &other) const - { - return memcmp(m_data, other.m_data, sizeof(m_data)) != 0; - } - - - Id &operator=(const Id &other) - { - memcpy(m_data, other.m_data, sizeof(m_data)); - - return *this; - } - - - inline bool setId(const char *id, size_t sizeFix = 0) - { - memset(m_data, 0, sizeof(m_data)); - if (!id) { - return false; - } - - const size_t size = strlen(id); - if (size >= sizeof(m_data)) { - return false; - } - - memcpy(m_data, id, size - sizeFix); - return true; - } - - - inline const char *data() const { return m_data; } - inline bool isValid() const { return *m_data != '\0'; } - - -private: - char m_data[64]; -}; - - -} /* namespace xmrig */ - - -#endif /* XMRIG_ID_H */ diff --git a/src/common/utils/c_str.h b/src/common/utils/c_str.h deleted file mode 100644 index fe0164b9e..000000000 --- a/src/common/utils/c_str.h +++ /dev/null @@ -1,39 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef XMRIG_C_STR_H -#define XMRIG_C_STR_H - - -#include "base/tools/String.h" - - -namespace xmrig { - - -typedef String c_str; - - -} /* namespace xmrig */ - -#endif /* XMRIG_C_STR_H */ diff --git a/src/common/utils/mm_malloc.h b/src/common/utils/mm_malloc.h deleted file mode 100644 index 30c721a34..000000000 --- a/src/common/utils/mm_malloc.h +++ /dev/null @@ -1,43 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef __MM_MALLOC_PORTABLE_H__ -#define __MM_MALLOC_PORTABLE_H__ - - -#ifdef _WIN32 -# ifdef __GNUC__ -# include -# else -# include -# endif -#else -# if defined(XMRIG_ARM) && !defined(__clang__) -# include "aligned_malloc.h" -# else -# include -# endif -#endif - - -#endif /* __MM_MALLOC_PORTABLE_H__ */ diff --git a/src/common/utils/timestamp.h b/src/common/utils/timestamp.h deleted file mode 100644 index 7fc4ab503..000000000 --- a/src/common/utils/timestamp.h +++ /dev/null @@ -1,55 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef XMRIG_TIMESTAMP_H -#define XMRIG_TIMESTAMP_H - - -#include - - -namespace xmrig { - - -static inline int64_t steadyTimestamp() -{ - using namespace std::chrono; - if (high_resolution_clock::is_steady) { - return time_point_cast(high_resolution_clock::now()).time_since_epoch().count(); - } - - return time_point_cast(steady_clock::now()).time_since_epoch().count(); -} - - -static inline int64_t currentMSecsSinceEpoch() -{ - using namespace std::chrono; - - return time_point_cast(high_resolution_clock::now()).time_since_epoch().count(); -} - - -} /* namespace xmrig */ - -#endif /* XMRIG_TIMESTAMP_H */ diff --git a/src/common/xmrig.h b/src/common/xmrig.h deleted file mode 100644 index e8ca8857e..000000000 --- a/src/common/xmrig.h +++ /dev/null @@ -1,121 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef XMRIG_XMRIG_H -#define XMRIG_XMRIG_H - - -namespace xmrig -{ - - -enum Algo { - INVALID_ALGO = -1, - CRYPTONIGHT, /* CryptoNight (2 MB) */ - CRYPTONIGHT_LITE, /* CryptoNight (1 MB) */ - CRYPTONIGHT_HEAVY, /* CryptoNight (4 MB) */ - CRYPTONIGHT_PICO, /* CryptoNight (256 KB) */ - ALGO_MAX -}; - - -//--av=1 For CPUs with hardware AES. -//--av=2 Lower power mode (double hash) of 1. -//--av=3 Software AES implementation. -//--av=4 Lower power mode (double hash) of 3. -enum AlgoVariant { - AV_AUTO, // --av=0 Automatic mode. - AV_SINGLE, // --av=1 Single hash mode - AV_DOUBLE, // --av=2 Double hash mode - AV_SINGLE_SOFT, // --av=3 Single hash mode (Software AES) - AV_DOUBLE_SOFT, // --av=4 Double hash mode (Software AES) - AV_TRIPLE, // --av=5 Triple hash mode - AV_QUAD, // --av=6 Quard hash mode - AV_PENTA, // --av=7 Penta hash mode - AV_TRIPLE_SOFT, // --av=8 Triple hash mode (Software AES) - AV_QUAD_SOFT, // --av=9 Quard hash mode (Software AES) - AV_PENTA_SOFT, // --av=10 Penta hash mode (Software AES) - AV_MAX -}; - - -enum Variant { - VARIANT_AUTO = -1, // Autodetect - VARIANT_0 = 0, // Original CryptoNight or CryptoNight-Heavy - VARIANT_1 = 1, // CryptoNight variant 1 also known as Monero7 and CryptoNightV7 - VARIANT_TUBE = 2, // Modified CryptoNight-Heavy (TUBE only) - VARIANT_XTL = 3, // Modified CryptoNight variant 1 (Stellite only) - VARIANT_MSR = 4, // Modified CryptoNight variant 1 (Masari only) - VARIANT_XHV = 5, // Modified CryptoNight-Heavy (Haven Protocol only) - VARIANT_XAO = 6, // Modified CryptoNight variant 0 (Alloy only) - VARIANT_RTO = 7, // Modified CryptoNight variant 1 (Arto only) - VARIANT_2 = 8, // CryptoNight variant 2 - VARIANT_HALF = 9, // CryptoNight variant 2 with half iterations (Masari/Stellite) - VARIANT_TRTL = 10, // CryptoNight Turtle (TRTL) - VARIANT_GPU = 11, // CryptoNight-GPU (Ryo) - VARIANT_WOW = 12, // CryptoNightR (Wownero) - VARIANT_4 = 13, // CryptoNightR (Monero's variant 4) - VARIANT_RWZ = 14, // CryptoNight variant 2 with 3/4 iterations and reversed shuffle operation (Graft) - VARIANT_ZLS = 15, // CryptoNight variant 2 with 3/4 iterations (Zelerius) - VARIANT_DOUBLE = 16, // CryptoNight variant 2 with double iterations (X-CASH) - VARIANT_MAX -}; - - -enum AlgoVerify { - VERIFY_HW_AES = 1, - VERIFY_SOFT_AES = 2 -}; - - -enum AesMode { - AES_AUTO, - AES_HW, - AES_SOFT -}; - - -enum OclVendor { - OCL_VENDOR_UNKNOWN = -2, - OCL_VENDOR_MANUAL = -1, - OCL_VENDOR_AMD = 0, - OCL_VENDOR_NVIDIA = 1, - OCL_VENDOR_INTEL = 2 -}; - - -enum Assembly { - ASM_NONE, - ASM_AUTO, - ASM_INTEL, - ASM_RYZEN, - ASM_BULLDOZER, - ASM_MAX -}; - - -} /* namespace xmrig */ - - -#endif /* XMRIG_XMRIG_H */ diff --git a/src/config.json b/src/config.json index 5018db514..af3bd32a5 100644 --- a/src/config.json +++ b/src/config.json @@ -1,43 +1,53 @@ { - "algo": "cryptonight", "api": { + "id": null, + "worker-id": null + }, + "http": { + "enabled": false, + "host": "127.0.0.1", "port": 0, "access-token": null, - "id": null, - "worker-id": null, - "ipv6": false, "restricted": true }, - "asm": true, "autosave": true, - "av": 0, "background": false, "colors": true, - "cpu-affinity": null, - "cpu-priority": null, + "randomx": { + "init": -1, + "numa": true + }, + "cpu": { + "enabled": true, + "huge-pages": true, + "hw-aes": null, + "priority": null, + "asm": true, + "cn/0": false, + "cn-lite/0": false + }, "donate-level": 5, - "huge-pages": true, - "hw-aes": null, + "donate-over-proxy": 1, "log-file": null, - "max-cpu-usage": 100, "pools": [ { + "algo": null, "url": "donate.v2.xmrig.com:3333", "user": "YOUR_WALLET_ADDRESS", "pass": "x", "rig-id": null, "nicehash": false, "keepalive": false, - "variant": -1, + "enabled": true, "tls": false, - "tls-fingerprint": null + "tls-fingerprint": null, + "daemon": false } ], "print-time": 60, "retries": 5, "retry-pause": 5, - "safe": false, - "threads": null, + "syslog": false, "user-agent": null, "watch": true } \ No newline at end of file diff --git a/src/core/Config.cpp b/src/core/Config.cpp deleted file mode 100644 index 9216027a5..000000000 --- a/src/core/Config.cpp +++ /dev/null @@ -1,379 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#include -#include -#include -#include - - -#include "common/config/ConfigLoader.h" -#include "common/cpu/Cpu.h" -#include "core/Config.h" -#include "core/ConfigCreator.h" -#include "crypto/Asm.h" -#include "crypto/CryptoNight_constants.h" -#include "rapidjson/document.h" -#include "rapidjson/filewritestream.h" -#include "rapidjson/prettywriter.h" -#include "workers/CpuThread.h" - - -static char affinity_tmp[20] = { 0 }; - - -xmrig::Config::Config() : xmrig::CommonConfig(), - m_aesMode(AES_AUTO), - m_algoVariant(AV_AUTO), - m_assembly(ASM_AUTO), - m_hugePages(true), - m_safe(false), - m_shouldSave(false), - m_maxCpuUsage(100), - m_priority(-1) -{ -} - - -bool xmrig::Config::reload(const char *json) -{ - return xmrig::ConfigLoader::reload(this, json); -} - - -void xmrig::Config::getJSON(rapidjson::Document &doc) const -{ - using namespace rapidjson; - - doc.SetObject(); - - auto &allocator = doc.GetAllocator(); - - doc.AddMember("algo", StringRef(algorithm().name()), allocator); - - Value api(kObjectType); - api.AddMember("port", apiPort(), allocator); - api.AddMember("access-token", apiToken() ? Value(StringRef(apiToken())).Move() : Value(kNullType).Move(), allocator); - api.AddMember("id", apiId() ? Value(StringRef(apiId())).Move() : Value(kNullType).Move(), allocator); - api.AddMember("worker-id", apiWorkerId() ? Value(StringRef(apiWorkerId())).Move() : Value(kNullType).Move(), allocator); - api.AddMember("ipv6", isApiIPv6(), allocator); - api.AddMember("restricted", isApiRestricted(), allocator); - doc.AddMember("api", api, allocator); - -# ifndef XMRIG_NO_ASM - doc.AddMember("asm", Asm::toJSON(m_assembly), allocator); -# endif - - doc.AddMember("autosave", isAutoSave(), allocator); - doc.AddMember("av", algoVariant(), allocator); - doc.AddMember("background", isBackground(), allocator); - doc.AddMember("colors", isColors(), allocator); - - if (affinity() != -1L) { - snprintf(affinity_tmp, sizeof(affinity_tmp) - 1, "0x%" PRIX64, affinity()); - doc.AddMember("cpu-affinity", StringRef(affinity_tmp), allocator); - } - else { - doc.AddMember("cpu-affinity", kNullType, allocator); - } - - doc.AddMember("cpu-priority", priority() != -1 ? Value(priority()) : Value(kNullType), allocator); - doc.AddMember("donate-level", donateLevel(), allocator); - doc.AddMember("huge-pages", isHugePages(), allocator); - doc.AddMember("hw-aes", m_aesMode == AES_AUTO ? Value(kNullType) : Value(m_aesMode == AES_HW), allocator); - doc.AddMember("log-file", logFile() ? Value(StringRef(logFile())).Move() : Value(kNullType).Move(), allocator); - doc.AddMember("max-cpu-usage", m_maxCpuUsage, allocator); - doc.AddMember("pools", m_pools.toJSON(doc), allocator); - doc.AddMember("print-time", printTime(), allocator); - doc.AddMember("retries", m_pools.retries(), allocator); - doc.AddMember("retry-pause", m_pools.retryPause(), allocator); - doc.AddMember("safe", m_safe, allocator); - - if (threadsMode() != Simple) { - Value threads(kArrayType); - - for (const IThread *thread : m_threads.list) { - threads.PushBack(thread->toConfig(doc), allocator); - } - - doc.AddMember("threads", threads, allocator); - } - else { - doc.AddMember("threads", threadsCount(), allocator); - } - - doc.AddMember("user-agent", userAgent() ? Value(StringRef(userAgent())).Move() : Value(kNullType).Move(), allocator); - -# ifdef HAVE_SYSLOG_H - doc.AddMember("syslog", isSyslog(), allocator); -# endif - - doc.AddMember("watch", m_watch, allocator); -} - - -xmrig::Config *xmrig::Config::load(Process *process, IConfigListener *listener) -{ - return static_cast(ConfigLoader::load(process, new ConfigCreator(), listener)); -} - - -bool xmrig::Config::finalize() -{ - if (m_state != NoneState) { - return CommonConfig::finalize(); - } - - if (!CommonConfig::finalize()) { - return false; - } - - if (!m_threads.cpu.empty()) { - m_threads.mode = Advanced; - const bool softAES = (m_aesMode == AES_AUTO ? (Cpu::info()->hasAES() ? AES_HW : AES_SOFT) : m_aesMode) == AES_SOFT; - - for (size_t i = 0; i < m_threads.cpu.size(); ++i) { - m_threads.list.push_back(CpuThread::createFromData(i, m_algorithm.algo(), m_threads.cpu[i], m_priority, softAES)); - } - - return true; - } - - const AlgoVariant av = getAlgoVariant(); - m_threads.mode = m_threads.count ? Simple : Automatic; - - const size_t size = CpuThread::multiway(av) * cn_select_memory(m_algorithm.algo()) / 1024; - - if (!m_threads.count) { - m_threads.count = Cpu::info()->optimalThreadsCount(size, m_maxCpuUsage); - } - else if (m_safe) { - const size_t count = Cpu::info()->optimalThreadsCount(size, m_maxCpuUsage); - if (m_threads.count > count) { - m_threads.count = count; - } - } - - for (size_t i = 0; i < m_threads.count; ++i) { - m_threads.list.push_back(CpuThread::createFromAV(i, m_algorithm.algo(), av, m_threads.mask, m_priority, m_assembly)); - } - - m_shouldSave = m_threads.mode == Automatic; - return true; -} - - -bool xmrig::Config::parseBoolean(int key, bool enable) -{ - if (!CommonConfig::parseBoolean(key, enable)) { - return false; - } - - switch (key) { - case SafeKey: /* --safe */ - m_safe = enable; - break; - - case HugePagesKey: /* --no-huge-pages */ - m_hugePages = enable; - break; - - case HardwareAESKey: /* hw-aes config only */ - m_aesMode = enable ? AES_HW : AES_SOFT; - break; - -# ifndef XMRIG_NO_ASM - case AssemblyKey: - m_assembly = Asm::parse(enable); - break; -# endif - - default: - break; - } - - return true; -} - - -bool xmrig::Config::parseString(int key, const char *arg) -{ - if (!CommonConfig::parseString(key, arg)) { - return false; - } - - switch (key) { - case AVKey: /* --av */ - case MaxCPUUsageKey: /* --max-cpu-usage */ - case CPUPriorityKey: /* --cpu-priority */ - return parseUint64(key, strtol(arg, nullptr, 10)); - - case SafeKey: /* --safe */ - return parseBoolean(key, true); - - case HugePagesKey: /* --no-huge-pages */ - return parseBoolean(key, false); - - case ThreadsKey: /* --threads */ - if (strncmp(arg, "all", 3) == 0) { - m_threads.count = Cpu::info()->threads(); - return true; - } - - return parseUint64(key, strtol(arg, nullptr, 10)); - - case CPUAffinityKey: /* --cpu-affinity */ - { - const char *p = strstr(arg, "0x"); - return parseUint64(key, p ? strtoull(p, nullptr, 16) : strtoull(arg, nullptr, 10)); - } - -# ifndef XMRIG_NO_ASM - case AssemblyKey: /* --asm */ - m_assembly = Asm::parse(arg); - break; -# endif - - default: - break; - } - - return true; -} - - -bool xmrig::Config::parseUint64(int key, uint64_t arg) -{ - if (!CommonConfig::parseUint64(key, arg)) { - return false; - } - - switch (key) { - case CPUAffinityKey: /* --cpu-affinity */ - if (arg) { - m_threads.mask = arg; - } - break; - - default: - return parseInt(key, static_cast(arg)); - } - - return true; -} - - -void xmrig::Config::parseJSON(const rapidjson::Document &doc) -{ - CommonConfig::parseJSON(doc); - - const rapidjson::Value &threads = doc["threads"]; - - if (threads.IsArray()) { - for (const rapidjson::Value &value : threads.GetArray()) { - if (!value.IsObject()) { - continue; - } - - if (value.HasMember("low_power_mode")) { - auto data = CpuThread::parse(value); - - if (data.valid) { - m_threads.cpu.push_back(std::move(data)); - } - } - } - } -} - - -bool xmrig::Config::parseInt(int key, int arg) -{ - switch (key) { - case ThreadsKey: /* --threads */ - if (arg >= 0 && arg < 1024) { - m_threads.count = arg; - } - break; - - case AVKey: /* --av */ - if (arg >= AV_AUTO && arg < AV_MAX) { - m_algoVariant = static_cast(arg); - } - break; - - case MaxCPUUsageKey: /* --max-cpu-usage */ - if (m_maxCpuUsage > 0 && arg <= 100) { - m_maxCpuUsage = arg; - } - break; - - case CPUPriorityKey: /* --cpu-priority */ - if (arg >= 0 && arg <= 5) { - m_priority = arg; - } - break; - - default: - break; - } - - return true; -} - - -xmrig::AlgoVariant xmrig::Config::getAlgoVariant() const -{ -# ifndef XMRIG_NO_AEON - if (m_algorithm.algo() == xmrig::CRYPTONIGHT_LITE) { - return getAlgoVariantLite(); - } -# endif - - if (m_algoVariant <= AV_AUTO || m_algoVariant >= AV_MAX) { - return Cpu::info()->hasAES() ? AV_SINGLE : AV_SINGLE_SOFT; - } - - if (m_safe && !Cpu::info()->hasAES() && m_algoVariant <= AV_DOUBLE) { - return static_cast(m_algoVariant + 2); - } - - return m_algoVariant; -} - - -#ifndef XMRIG_NO_AEON -xmrig::AlgoVariant xmrig::Config::getAlgoVariantLite() const -{ - if (m_algoVariant <= AV_AUTO || m_algoVariant >= AV_MAX) { - return Cpu::info()->hasAES() ? AV_DOUBLE : AV_DOUBLE_SOFT; - } - - if (m_safe && !Cpu::info()->hasAES() && m_algoVariant <= AV_DOUBLE) { - return static_cast(m_algoVariant + 2); - } - - return m_algoVariant; -} -#endif diff --git a/src/core/Config.h b/src/core/Config.h deleted file mode 100644 index d2e8c166e..000000000 --- a/src/core/Config.h +++ /dev/null @@ -1,129 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef XMRIG_CONFIG_H -#define XMRIG_CONFIG_H - - -#include -#include - - -#include "common/config/CommonConfig.h" -#include "common/xmrig.h" -#include "rapidjson/fwd.h" -#include "workers/CpuThread.h" - - -namespace xmrig { - - -class ConfigLoader; -class IThread; -class IConfigListener; -class Process; - - -/** - * @brief The Config class - * - * Options with dynamic reload: - * colors - * debug - * verbose - * custom-diff (only for new connections) - * api/worker-id - * pools/ - */ -class Config : public CommonConfig -{ -public: - enum ThreadsMode { - Automatic, - Simple, - Advanced - }; - - - Config(); - - bool reload(const char *json); - - void getJSON(rapidjson::Document &doc) const override; - - inline AesMode aesMode() const { return m_aesMode; } - inline AlgoVariant algoVariant() const { return m_algoVariant; } - inline Assembly assembly() const { return m_assembly; } - inline bool isHugePages() const { return m_hugePages; } - inline bool isShouldSave() const { return m_shouldSave && isAutoSave(); } - inline const std::vector &threads() const { return m_threads.list; } - inline int priority() const { return m_priority; } - inline int threadsCount() const { return m_threads.list.size(); } - inline int64_t affinity() const { return m_threads.mask; } - inline ThreadsMode threadsMode() const { return m_threads.mode; } - - static Config *load(Process *process, IConfigListener *listener); - -protected: - bool finalize() override; - bool parseBoolean(int key, bool enable) override; - bool parseString(int key, const char *arg) override; - bool parseUint64(int key, uint64_t arg) override; - void parseJSON(const rapidjson::Document &doc) override; - -private: - bool parseInt(int key, int arg); - - AlgoVariant getAlgoVariant() const; -# ifndef XMRIG_NO_AEON - AlgoVariant getAlgoVariantLite() const; -# endif - - - struct Threads - { - inline Threads() : mask(-1L), count(0), mode(Automatic) {} - - int64_t mask; - size_t count; - std::vector cpu; - std::vector list; - ThreadsMode mode; - }; - - - AesMode m_aesMode; - AlgoVariant m_algoVariant; - Assembly m_assembly; - bool m_hugePages; - bool m_safe; - bool m_shouldSave; - int m_maxCpuUsage; - int m_priority; - Threads m_threads; -}; - - -} /* namespace xmrig */ - -#endif /* XMRIG_CONFIG_H */ diff --git a/src/core/ConfigLoader_platform.h b/src/core/ConfigLoader_platform.h deleted file mode 100644 index 0b71c3fda..000000000 --- a/src/core/ConfigLoader_platform.h +++ /dev/null @@ -1,128 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef XMRIG_CONFIGLOADER_PLATFORM_H -#define XMRIG_CONFIGLOADER_PLATFORM_H - - -#ifdef _MSC_VER -# include "getopt/getopt.h" -#else -# include -#endif - - -#include "common/interfaces/IConfig.h" -#include "version.h" - - -namespace xmrig { - - -static char const short_options[] = "a:c:kBp:Px:r:R:s:t:T:o:u:O:v:l:S"; - - -static struct option const options[] = { - { "algo", 1, nullptr, xmrig::IConfig::AlgorithmKey }, - { "api-access-token", 1, nullptr, xmrig::IConfig::ApiAccessTokenKey }, - { "api-port", 1, nullptr, xmrig::IConfig::ApiPort }, - { "api-worker-id", 1, nullptr, xmrig::IConfig::ApiWorkerIdKey }, - { "api-id", 1, nullptr, xmrig::IConfig::ApiIdKey }, - { "api-ipv6", 0, nullptr, xmrig::IConfig::ApiIPv6Key }, - { "api-no-restricted", 0, nullptr, xmrig::IConfig::ApiRestrictedKey }, - { "av", 1, nullptr, xmrig::IConfig::AVKey }, - { "background", 0, nullptr, xmrig::IConfig::BackgroundKey }, - { "config", 1, nullptr, xmrig::IConfig::ConfigKey }, - { "cpu-affinity", 1, nullptr, xmrig::IConfig::CPUAffinityKey }, - { "cpu-priority", 1, nullptr, xmrig::IConfig::CPUPriorityKey }, - { "donate-level", 1, nullptr, xmrig::IConfig::DonateLevelKey }, - { "dry-run", 0, nullptr, xmrig::IConfig::DryRunKey }, - { "keepalive", 0, nullptr, xmrig::IConfig::KeepAliveKey }, - { "log-file", 1, nullptr, xmrig::IConfig::LogFileKey }, - { "max-cpu-usage", 1, nullptr, xmrig::IConfig::MaxCPUUsageKey }, - { "nicehash", 0, nullptr, xmrig::IConfig::NicehashKey }, - { "no-color", 0, nullptr, xmrig::IConfig::ColorKey }, - { "no-watch", 0, nullptr, xmrig::IConfig::WatchKey }, - { "no-huge-pages", 0, nullptr, xmrig::IConfig::HugePagesKey }, - { "variant", 1, nullptr, xmrig::IConfig::VariantKey }, - { "pass", 1, nullptr, xmrig::IConfig::PasswordKey }, - { "print-time", 1, nullptr, xmrig::IConfig::PrintTimeKey }, - { "retries", 1, nullptr, xmrig::IConfig::RetriesKey }, - { "retry-pause", 1, nullptr, xmrig::IConfig::RetryPauseKey }, - { "safe", 0, nullptr, xmrig::IConfig::SafeKey }, - { "syslog", 0, nullptr, xmrig::IConfig::SyslogKey }, - { "threads", 1, nullptr, xmrig::IConfig::ThreadsKey }, - { "url", 1, nullptr, xmrig::IConfig::UrlKey }, - { "user", 1, nullptr, xmrig::IConfig::UserKey }, - { "user-agent", 1, nullptr, xmrig::IConfig::UserAgentKey }, - { "userpass", 1, nullptr, xmrig::IConfig::UserpassKey }, - { "rig-id", 1, nullptr, xmrig::IConfig::RigIdKey }, - { "tls", 0, nullptr, xmrig::IConfig::TlsKey }, - { "tls-fingerprint", 1, nullptr, xmrig::IConfig::FingerprintKey }, - { "asm", 1, nullptr, xmrig::IConfig::AssemblyKey }, - { nullptr, 0, nullptr, 0 } -}; - - -static struct option const config_options[] = { - { "algo", 1, nullptr, xmrig::IConfig::AlgorithmKey }, - { "av", 1, nullptr, xmrig::IConfig::AVKey }, - { "background", 0, nullptr, xmrig::IConfig::BackgroundKey }, - { "colors", 0, nullptr, xmrig::IConfig::ColorKey }, - { "cpu-affinity", 1, nullptr, xmrig::IConfig::CPUAffinityKey }, - { "cpu-priority", 1, nullptr, xmrig::IConfig::CPUPriorityKey }, - { "donate-level", 1, nullptr, xmrig::IConfig::DonateLevelKey }, - { "dry-run", 0, nullptr, xmrig::IConfig::DryRunKey }, - { "huge-pages", 0, nullptr, xmrig::IConfig::HugePagesKey }, - { "log-file", 1, nullptr, xmrig::IConfig::LogFileKey }, - { "max-cpu-usage", 1, nullptr, xmrig::IConfig::MaxCPUUsageKey }, - { "print-time", 1, nullptr, xmrig::IConfig::PrintTimeKey }, - { "retries", 1, nullptr, xmrig::IConfig::RetriesKey }, - { "retry-pause", 1, nullptr, xmrig::IConfig::RetryPauseKey }, - { "safe", 0, nullptr, xmrig::IConfig::SafeKey }, - { "syslog", 0, nullptr, xmrig::IConfig::SyslogKey }, - { "threads", 1, nullptr, xmrig::IConfig::ThreadsKey }, - { "user-agent", 1, nullptr, xmrig::IConfig::UserAgentKey }, - { "watch", 0, nullptr, xmrig::IConfig::WatchKey }, - { "hw-aes", 0, nullptr, xmrig::IConfig::HardwareAESKey }, - { "asm", 1, nullptr, xmrig::IConfig::AssemblyKey }, - { "autosave", 0, nullptr, xmrig::IConfig::AutoSaveKey }, - { nullptr, 0, nullptr, 0 } -}; - - -static struct option const api_options[] = { - { "port", 1, nullptr, xmrig::IConfig::ApiPort }, - { "access-token", 1, nullptr, xmrig::IConfig::ApiAccessTokenKey }, - { "worker-id", 1, nullptr, xmrig::IConfig::ApiWorkerIdKey }, - { "ipv6", 0, nullptr, xmrig::IConfig::ApiIPv6Key }, - { "restricted", 0, nullptr, xmrig::IConfig::ApiRestrictedKey }, - { "id", 1, nullptr, xmrig::IConfig::ApiIdKey }, - { nullptr, 0, nullptr, 0 } -}; - - -} /* namespace xmrig */ - -#endif /* XMRIG_CONFIGLOADER_PLATFORM_H */ diff --git a/src/core/Controller.cpp b/src/core/Controller.cpp index 7639bfaae..54c9ee34b 100644 --- a/src/core/Controller.cpp +++ b/src/core/Controller.cpp @@ -26,72 +26,27 @@ #include -#include "common/config/ConfigLoader.h" -#include "common/cpu/Cpu.h" -#include "common/interfaces/IControllerListener.h" -#include "common/log/ConsoleLog.h" -#include "common/log/FileLog.h" -#include "common/log/Log.h" -#include "common/Platform.h" -#include "core/Config.h" +#include "backend/cpu/Cpu.h" #include "core/Controller.h" +#include "core/Miner.h" #include "net/Network.h" -#ifdef HAVE_SYSLOG_H -# include "common/log/SysLog.h" -#endif - - -class xmrig::ControllerPrivate -{ -public: - inline ControllerPrivate(Process *process) : - config(nullptr), - network(nullptr), - process(process) - {} - - - inline ~ControllerPrivate() - { - delete network; - delete config; - } - - - Config *config; - Network *network; - Process *process; - std::vector listeners; -}; - - -xmrig::Controller::Controller(Process *process) - : d_ptr(new ControllerPrivate(process)) +xmrig::Controller::Controller(Process *process) : + Base(process) { } xmrig::Controller::~Controller() { - ConfigLoader::release(); - - delete d_ptr; + delete m_network; } bool xmrig::Controller::isReady() const { - return d_ptr->config && d_ptr->network; -} - - -xmrig::Config *xmrig::Controller::config() const -{ - assert(d_ptr->config != nullptr); - - return d_ptr->config; + return Base::isReady() && m_network; } @@ -99,70 +54,51 @@ int xmrig::Controller::init() { Cpu::init(); - d_ptr->config = Config::load(d_ptr->process, this); - if (!d_ptr->config) { - return 1; + const int rc = Base::init(); + if (rc != 0) { + return rc; } - Log::init(); - Platform::init(config()->userAgent()); - Platform::setProcessPriority(d_ptr->config->priority()); - - if (!config()->isBackground()) { - Log::add(new ConsoleLog(this)); - } - - if (config()->logFile()) { - Log::add(new FileLog(this, config()->logFile())); - } - -# ifdef HAVE_SYSLOG_H - if (config()->isSyslog()) { - Log::add(new SysLog()); - } -# endif - - d_ptr->network = new Network(this); + m_network = new Network(this); return 0; } +void xmrig::Controller::start() +{ + Base::start(); + + m_miner = new Miner(this); + + network()->connect(); +} + + +void xmrig::Controller::stop() +{ + Base::stop(); + + delete m_network; + m_network = nullptr; + + m_miner->stop(); + + delete m_miner; + m_miner = nullptr; +} + + +xmrig::Miner *xmrig::Controller::miner() const +{ + assert(m_miner != nullptr); + + return m_miner; +} + + xmrig::Network *xmrig::Controller::network() const { - assert(d_ptr->network != nullptr); + assert(m_network != nullptr); - return d_ptr->network; -} - - -void xmrig::Controller::addListener(IControllerListener *listener) -{ - d_ptr->listeners.push_back(listener); -} - - -void xmrig::Controller::save() -{ - if (!config()) { - return; - } - - if (d_ptr->config->isShouldSave()) { - d_ptr->config->save(); - } - - ConfigLoader::watch(d_ptr->config); -} - - -void xmrig::Controller::onNewConfig(IConfig *config) -{ - Config *previousConfig = d_ptr->config; - d_ptr->config = static_cast(config); - - for (xmrig::IControllerListener *listener : d_ptr->listeners) { - listener->onConfigChanged(d_ptr->config, previousConfig); - } - - delete previousConfig; + return m_network; } diff --git a/src/core/Controller.h b/src/core/Controller.h index 8d2f26745..da7ba3686 100644 --- a/src/core/Controller.h +++ b/src/core/Controller.h @@ -26,44 +26,38 @@ #define XMRIG_CONTROLLER_H -#include "base/kernel/interfaces/IConfigListener.h" - - -class StatsData; +#include "base/kernel/Base.h" namespace xmrig { -class Config; -class ControllerPrivate; -class IControllerListener; +class Job; +class Miner; class Network; -class Process; -class Controller : public IConfigListener +class Controller : public Base { public: Controller(Process *process); ~Controller() override; - bool isReady() const; - Config *config() const; - int init(); - Network *network() const; - void addListener(IControllerListener *listener); - void save(); + bool isReady() const override; + int init() override; + void start() override; + void stop() override; -protected: - void onNewConfig(IConfig *config) override; + Miner *miner() const; + Network *network() const; private: - ControllerPrivate *d_ptr; + Miner *m_miner = nullptr; + Network *m_network = nullptr; }; -} /* namespace xmrig */ +} // namespace xmrig #endif /* XMRIG_CONTROLLER_H */ diff --git a/src/core/Miner.cpp b/src/core/Miner.cpp new file mode 100644 index 000000000..4406ce529 --- /dev/null +++ b/src/core/Miner.cpp @@ -0,0 +1,481 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include +#include +#include + + +#include "backend/common/Hashrate.h" +#include "backend/cpu/Cpu.h" +#include "backend/cpu/CpuBackend.h" +#include "base/io/log/Log.h" +#include "base/kernel/Platform.h" +#include "base/net/stratum/Job.h" +#include "base/tools/Timer.h" +#include "core/config/Config.h" +#include "core/Controller.h" +#include "core/Miner.h" +#include "crypto/common/Nonce.h" +#include "crypto/rx/Rx.h" +#include "rapidjson/document.h" +#include "version.h" + + +#ifdef XMRIG_FEATURE_API +# include "api/Api.h" +# include "api/interfaces/IApiRequest.h" +#endif + + +namespace xmrig { + + +class MinerPrivate +{ +public: + inline MinerPrivate(Controller *controller) : controller(controller) + { + uv_rwlock_init(&rwlock); + +# ifdef XMRIG_ALGO_RANDOMX + Rx::init(); +# endif + } + + + inline ~MinerPrivate() + { + uv_rwlock_destroy(&rwlock); + + delete timer; + + for (IBackend *backend : backends) { + delete backend; + } + +# ifdef XMRIG_ALGO_RANDOMX + Rx::destroy(); +# endif + } + + + bool isEnabled(const Algorithm &algorithm) const + { + for (IBackend *backend : backends) { + if (backend->isEnabled(algorithm)) { + return true; + } + } + + return false; + } + + + inline void rebuild() + { + algorithms.clear(); + + for (int i = 0; i < Algorithm::MAX; ++i) { + const Algorithm algo(static_cast(i)); + + if (isEnabled(algo)) { + algorithms.push_back(algo); + } + } + } + + + inline void handleJobChange(bool reset) + { + active = true; + + for (IBackend *backend : backends) { + backend->setJob(job); + } + + if (reset) { + Nonce::reset(job.index()); + } + else { + Nonce::touch(); + } + + if (enabled) { + Nonce::pause(false);; + } + + if (ticks == 0) { + ticks++; + timer->start(500, 500); + } + } + + +# ifdef XMRIG_FEATURE_API + void getMiner(rapidjson::Value &reply, rapidjson::Document &doc, int version) const + { + using namespace rapidjson; + auto &allocator = doc.GetAllocator(); + + reply.AddMember("version", APP_VERSION, allocator); + reply.AddMember("kind", APP_KIND, allocator); + reply.AddMember("ua", StringRef(Platform::userAgent()), allocator); + reply.AddMember("cpu", Cpu::toJSON(doc), allocator); + + Value hugepages; + + if (!backends.empty() && backends.front()->type() == "cpu") { + const auto pages = static_cast(backends.front())->hugePages(); + if (version > 1) { + hugepages.SetArray(); + hugepages.PushBack(pages.first, allocator); + hugepages.PushBack(pages.second, allocator); + } + else { + hugepages = pages.first == pages.second; + } + } + else { + hugepages = false; + } + + reply.AddMember("hugepages", hugepages, allocator); + reply.AddMember("donate_level", controller->config()->pools().donateLevel(), allocator); + + Value algo(kArrayType); + + for (const Algorithm &a : algorithms) { + algo.PushBack(StringRef(a.shortName()), allocator); + } + + reply.AddMember("algorithms", algo, allocator); + } + + + void getHashrate(rapidjson::Value &reply, rapidjson::Document &doc, int version) const + { + using namespace rapidjson; + auto &allocator = doc.GetAllocator(); + + Value hashrate(kObjectType); + Value total(kArrayType); + Value threads(kArrayType); + + double t[3] = { 0.0 }; + + for (IBackend *backend : backends) { + const Hashrate *hr = backend->hashrate(); + if (!hr) { + continue; + } + + t[0] += hr->calc(Hashrate::ShortInterval); + t[1] += hr->calc(Hashrate::MediumInterval); + t[2] += hr->calc(Hashrate::LargeInterval); + + if (version > 1) { + continue; + } + + for (size_t i = 0; i < hr->threads(); i++) { + Value thread(kArrayType); + thread.PushBack(Hashrate::normalize(hr->calc(i, Hashrate::ShortInterval)), allocator); + thread.PushBack(Hashrate::normalize(hr->calc(i, Hashrate::MediumInterval)), allocator); + thread.PushBack(Hashrate::normalize(hr->calc(i, Hashrate::LargeInterval)), allocator); + + threads.PushBack(thread, allocator); + } + } + + total.PushBack(Hashrate::normalize(t[0]), allocator); + total.PushBack(Hashrate::normalize(t[1]), allocator); + total.PushBack(Hashrate::normalize(t[2]), allocator); + + hashrate.AddMember("total", total, allocator); + hashrate.AddMember("highest", Hashrate::normalize(maxHashrate[algorithm]), allocator); + + if (version == 1) { + hashrate.AddMember("threads", threads, allocator); + } + + reply.AddMember("hashrate", hashrate, allocator); + } + + + void getBackends(rapidjson::Value &reply, rapidjson::Document &doc) const + { + using namespace rapidjson; + auto &allocator = doc.GetAllocator(); + + reply.SetArray(); + + for (IBackend *backend : backends) { + reply.PushBack(backend->toJSON(doc), allocator); + } + } +# endif + + + Algorithm algorithm; + Algorithms algorithms; + bool active = false; + bool enabled = true; + Controller *controller; + Job job; + mutable std::map maxHashrate; + std::vector backends; + String userJobId; + Timer *timer = nullptr; + uint64_t ticks = 0; + uv_rwlock_t rwlock; +}; + + +} // namespace xmrig + + + +xmrig::Miner::Miner(Controller *controller) + : d_ptr(new MinerPrivate(controller)) +{ + controller->addListener(this); + +# ifdef XMRIG_FEATURE_API + controller->api()->addListener(this); +# endif + + d_ptr->timer = new Timer(this); + + d_ptr->backends.push_back(new CpuBackend(controller)); + + d_ptr->rebuild(); +} + + +xmrig::Miner::~Miner() +{ + delete d_ptr; +} + + +bool xmrig::Miner::isEnabled() const +{ + return d_ptr->enabled; +} + + +bool xmrig::Miner::isEnabled(const Algorithm &algorithm) const +{ + return std::find(d_ptr->algorithms.begin(), d_ptr->algorithms.end(), algorithm) != d_ptr->algorithms.end(); +} + + +const xmrig::Algorithms &xmrig::Miner::algorithms() const +{ + return d_ptr->algorithms; +} + + +const std::vector &xmrig::Miner::backends() const +{ + return d_ptr->backends; +} + + +xmrig::Job xmrig::Miner::job() const +{ + uv_rwlock_rdlock(&d_ptr->rwlock); + Job job = d_ptr->job; + uv_rwlock_rdunlock(&d_ptr->rwlock); + + return job; +} + + +void xmrig::Miner::pause() +{ + d_ptr->active = false; + + Nonce::pause(true); + Nonce::touch(); +} + + +void xmrig::Miner::printHashrate(bool details) +{ + char num[8 * 4] = { 0 }; + double speed[3] = { 0.0 }; + + for (IBackend *backend : d_ptr->backends) { + const Hashrate *hashrate = backend->hashrate(); + if (hashrate) { + speed[0] += hashrate->calc(Hashrate::ShortInterval); + speed[1] += hashrate->calc(Hashrate::MediumInterval); + speed[2] += hashrate->calc(Hashrate::LargeInterval); + } + + backend->printHashrate(details); + } + + LOG_INFO(WHITE_BOLD("speed") " 10s/60s/15m " CYAN_BOLD("%s") CYAN(" %s %s ") CYAN_BOLD("H/s") " max " CYAN_BOLD("%s H/s"), + Hashrate::format(speed[0], num, sizeof(num) / 4), + Hashrate::format(speed[1], num + 8, sizeof(num) / 4), + Hashrate::format(speed[2], num + 8 * 2, sizeof(num) / 4 ), + Hashrate::format(d_ptr->maxHashrate[d_ptr->algorithm], num + 8 * 3, sizeof(num) / 4) + ); +} + + +void xmrig::Miner::setEnabled(bool enabled) +{ + if (d_ptr->enabled == enabled) { + return; + } + + d_ptr->enabled = enabled; + + if (enabled) { + LOG_INFO(GREEN_BOLD("resumed")); + } + else { + LOG_INFO(YELLOW_BOLD("paused") ", press " MAGENTA_BG_BOLD(" r ") " to resume"); + } + + if (!d_ptr->active) { + return; + } + + Nonce::pause(!enabled); + Nonce::touch(); +} + + +void xmrig::Miner::setJob(const Job &job, bool donate) +{ + for (IBackend *backend : d_ptr->backends) { + backend->prepare(job); + } + +# ifdef XMRIG_ALGO_RANDOMX + if (d_ptr->algorithm.family() == Algorithm::RANDOM_X && job.algorithm().family() == Algorithm::RANDOM_X && !Rx::isReady(job)) { + stop(); + } +# endif + + d_ptr->algorithm = job.algorithm(); + + uv_rwlock_wrlock(&d_ptr->rwlock); + + const uint8_t index = donate ? 1 : 0; + const bool reset = !(d_ptr->job.index() == 1 && index == 0 && d_ptr->userJobId == job.id()); + + d_ptr->job = job; + d_ptr->job.setIndex(index); + + if (index == 0) { + d_ptr->userJobId = job.id(); + } + +# ifdef XMRIG_ALGO_RANDOMX + Rx::init(d_ptr->job, + d_ptr->controller->config()->rx().threads(), + d_ptr->controller->config()->cpu().isHugePages(), + d_ptr->controller->config()->rx().isNUMA() + ); +# endif + + uv_rwlock_wrunlock(&d_ptr->rwlock); + + d_ptr->handleJobChange(reset); +} + + +void xmrig::Miner::stop() +{ + Nonce::stop(); + + for (IBackend *backend : d_ptr->backends) { + backend->stop(); + } +} + + +void xmrig::Miner::onConfigChanged(Config *config, Config *previousConfig) +{ + d_ptr->rebuild(); + + if (config->pools() != previousConfig->pools() && config->pools().active() > 0) { + return; + } + + const Job job = this->job(); + + for (IBackend *backend : d_ptr->backends) { + backend->setJob(job); + } +} + + +void xmrig::Miner::onTimer(const Timer *) +{ + double maxHashrate = 0.0; + + for (IBackend *backend : d_ptr->backends) { + backend->tick(d_ptr->ticks); + + if (backend->hashrate()) { + maxHashrate += backend->hashrate()->calc(Hashrate::ShortInterval); + } + } + + d_ptr->maxHashrate[d_ptr->algorithm] = std::max(d_ptr->maxHashrate[d_ptr->algorithm], maxHashrate); + + if ((d_ptr->ticks % (d_ptr->controller->config()->printTime() * 2)) == 0) { + printHashrate(false); + } + + d_ptr->ticks++; +} + + +#ifdef XMRIG_FEATURE_API +void xmrig::Miner::onRequest(IApiRequest &request) +{ + if (request.method() == IApiRequest::METHOD_GET) { + if (request.type() == IApiRequest::REQ_SUMMARY) { + request.accept(); + + d_ptr->getMiner(request.reply(), request.doc(), request.version()); + d_ptr->getHashrate(request.reply(), request.doc(), request.version()); + } + else if (request.url() == "/2/backends") { + request.accept(); + + d_ptr->getBackends(request.reply(), request.doc()); + } + } +} +#endif diff --git a/src/core/Miner.h b/src/core/Miner.h new file mode 100644 index 000000000..035c02058 --- /dev/null +++ b/src/core/Miner.h @@ -0,0 +1,80 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_MINER_H +#define XMRIG_MINER_H + + +#include + + +#include "api/interfaces/IApiListener.h" +#include "base/kernel/interfaces/IBaseListener.h" +#include "base/kernel/interfaces/ITimerListener.h" +#include "crypto/common/Algorithm.h" + + +namespace xmrig { + + +class Controller; +class Job; +class MinerPrivate; +class IBackend; + + +class Miner : public ITimerListener, public IBaseListener, public IApiListener +{ +public: + Miner(Controller *controller); + ~Miner() override; + + bool isEnabled() const; + bool isEnabled(const Algorithm &algorithm) const; + const Algorithms &algorithms() const; + const std::vector &backends() const; + Job job() const; + void pause(); + void printHashrate(bool details); + void setEnabled(bool enabled); + void setJob(const Job &job, bool donate); + void stop(); + +protected: + void onConfigChanged(Config *config, Config *previousConfig) override; + void onTimer(const Timer *timer) override; + +# ifdef XMRIG_FEATURE_API + void onRequest(IApiRequest &request) override; +# endif + +private: + MinerPrivate *d_ptr; +}; + + +} // namespace xmrig + + +#endif /* XMRIG_MINER_H */ diff --git a/src/core/config/Config.cpp b/src/core/config/Config.cpp new file mode 100644 index 000000000..44aa60299 --- /dev/null +++ b/src/core/config/Config.cpp @@ -0,0 +1,108 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#include +#include +#include +#include + + +#include "backend/cpu/Cpu.h" +#include "base/io/log/Log.h" +#include "base/kernel/interfaces/IJsonReader.h" +#include "core/config/Config.h" +#include "crypto/common/Assembly.h" +#include "rapidjson/document.h" +#include "rapidjson/filewritestream.h" +#include "rapidjson/prettywriter.h" + + +namespace xmrig { + +static const char *kCPU = "cpu"; + +#ifdef XMRIG_ALGO_RANDOMX +static const char *kRandomX = "randomx"; +#endif + +} + + +xmrig::Config::Config() : BaseConfig() +{ +} + + +bool xmrig::Config::read(const IJsonReader &reader, const char *fileName) +{ + if (!BaseConfig::read(reader, fileName)) { + return false; + } + + m_cpu.read(reader.getValue(kCPU)); + +# ifdef XMRIG_ALGO_RANDOMX + if (!m_rx.read(reader.getValue(kRandomX))) { + m_upgrade = true; + } +# endif + + return true; +} + + +void xmrig::Config::getJSON(rapidjson::Document &doc) const +{ + using namespace rapidjson; + + doc.SetObject(); + + auto &allocator = doc.GetAllocator(); + + Value api(kObjectType); + api.AddMember("id", m_apiId.toJSON(), allocator); + api.AddMember("worker-id", m_apiWorkerId.toJSON(), allocator); + + doc.AddMember("api", api, allocator); + doc.AddMember("http", m_http.toJSON(doc), allocator); + doc.AddMember("autosave", isAutoSave(), allocator); + doc.AddMember("background", isBackground(), allocator); + doc.AddMember("colors", Log::colors, allocator); + +# ifdef XMRIG_ALGO_RANDOMX + doc.AddMember(StringRef(kRandomX), m_rx.toJSON(doc), allocator); +# endif + + doc.AddMember(StringRef(kCPU), m_cpu.toJSON(doc), allocator); + doc.AddMember("donate-level", m_pools.donateLevel(), allocator); + doc.AddMember("donate-over-proxy", m_pools.proxyDonate(), allocator); + doc.AddMember("log-file", m_logFile.toJSON(), allocator); + doc.AddMember("pools", m_pools.toJSON(doc), allocator); + doc.AddMember("print-time", printTime(), allocator); + doc.AddMember("retries", m_pools.retries(), allocator); + doc.AddMember("retry-pause", m_pools.retryPause(), allocator); + doc.AddMember("syslog", isSyslog(), allocator); + doc.AddMember("user-agent", m_userAgent.toJSON(), allocator); + doc.AddMember("watch", m_watch, allocator); +} diff --git a/src/core/config/Config.h b/src/core/config/Config.h new file mode 100644 index 000000000..0014cb05a --- /dev/null +++ b/src/core/config/Config.h @@ -0,0 +1,76 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_CONFIG_H +#define XMRIG_CONFIG_H + + +#include + + +#include "backend/cpu/CpuConfig.h" +#include "base/kernel/config/BaseConfig.h" +#include "rapidjson/fwd.h" + + +#ifdef XMRIG_ALGO_RANDOMX +# include "crypto/rx/RxConfig.h" +#endif + + +namespace xmrig { + + +class IThread; + + +class Config : public BaseConfig +{ +public: + Config(); + + bool read(const IJsonReader &reader, const char *fileName) override; + void getJSON(rapidjson::Document &doc) const override; + + inline bool isShouldSave() const { return (m_shouldSave || m_upgrade || m_cpu.isShouldSave()) && isAutoSave(); } + inline const CpuConfig &cpu() const { return m_cpu; } + +# ifdef XMRIG_ALGO_RANDOMX + inline const RxConfig &rx() const { return m_rx; } +# endif + +private: + bool m_shouldSave = false; + CpuConfig m_cpu; + +# ifdef XMRIG_ALGO_RANDOMX + RxConfig m_rx; +# endif +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_CONFIG_H */ diff --git a/src/core/config/ConfigTransform.cpp b/src/core/config/ConfigTransform.cpp new file mode 100644 index 000000000..622855af6 --- /dev/null +++ b/src/core/config/ConfigTransform.cpp @@ -0,0 +1,188 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "base/kernel/interfaces/IConfig.h" +#include "core/config/ConfigTransform.h" +#include "crypto/cn/CnHash.h" + + +namespace xmrig +{ + + +static const char *kAffinity = "affinity"; +static const char *kAsterisk = "*"; +static const char *kCpu = "cpu"; +static const char *kIntensity = "intensity"; +static const char *kThreads = "threads"; + +#ifdef XMRIG_ALGO_RANDOMX +static const char *kRandomX = "randomx"; +#endif + + +static inline uint64_t intensity(uint64_t av) +{ + switch (av) { + case CnHash::AV_SINGLE: + case CnHash::AV_SINGLE_SOFT: + return 1; + + case CnHash::AV_DOUBLE_SOFT: + case CnHash::AV_DOUBLE: + return 2; + + case CnHash::AV_TRIPLE_SOFT: + case CnHash::AV_TRIPLE: + return 3; + + case CnHash::AV_QUAD_SOFT: + case CnHash::AV_QUAD: + return 4; + + case CnHash::AV_PENTA_SOFT: + case CnHash::AV_PENTA: + return 5; + + default: + break; + } + + return 1; +} + + +static inline bool isHwAes(uint64_t av) +{ + return av == CnHash::AV_SINGLE || av == CnHash::AV_DOUBLE || (av > CnHash::AV_DOUBLE_SOFT && av < CnHash::AV_TRIPLE_SOFT); +} + + +} + + +xmrig::ConfigTransform::ConfigTransform() : BaseTransform() +{ +} + + +void xmrig::ConfigTransform::finalize(rapidjson::Document &doc) +{ + using namespace rapidjson; + auto &allocator = doc.GetAllocator(); + + BaseTransform::finalize(doc); + + if (m_threads) { + if (!doc.HasMember(kCpu)) { + doc.AddMember(StringRef(kCpu), Value(kObjectType), allocator); + } + + Value profile(kObjectType); + profile.AddMember(StringRef(kIntensity), m_intensity, allocator); + profile.AddMember(StringRef(kThreads), m_threads, allocator); + profile.AddMember(StringRef(kAffinity), m_affinity, allocator); + + doc[kCpu].AddMember(StringRef(kAsterisk), profile, doc.GetAllocator()); + } +} + + +void xmrig::ConfigTransform::transform(rapidjson::Document &doc, int key, const char *arg) +{ + BaseTransform::transform(doc, key, arg); + + switch (key) { + case IConfig::AVKey: /* --av */ + case IConfig::CPUPriorityKey: /* --cpu-priority */ + case IConfig::ThreadsKey: /* --threads */ + return transformUint64(doc, key, static_cast(strtol(arg, nullptr, 10))); + + case IConfig::HugePagesKey: /* --no-huge-pages */ + return transformBoolean(doc, key, false); + + case IConfig::CPUAffinityKey: /* --cpu-affinity */ + { + const char *p = strstr(arg, "0x"); + return transformUint64(doc, key, p ? strtoull(p, nullptr, 16) : strtoull(arg, nullptr, 10)); + } + +# ifndef XMRIG_NO_ASM + case IConfig::AssemblyKey: /* --asm */ + return set(doc, kCpu, "asm", arg); +# endif + +# ifdef XMRIG_ALGO_RANDOMX + case IConfig::RandomXInitKey: /* --randomx-init */ + return set(doc, kRandomX, "init", static_cast(strtol(arg, nullptr, 10))); + + case IConfig::RandomXNumaKey: /* --randomx-no-numa */ + return set(doc, kRandomX, "numa", false); +# endif + + default: + break; + } +} + + +void xmrig::ConfigTransform::transformBoolean(rapidjson::Document &doc, int key, bool enable) +{ + switch (key) { + case IConfig::HugePagesKey: /* --no-huge-pages */ + return set(doc, kCpu, "huge-pages", enable); + + default: + break; + } +} + + +void xmrig::ConfigTransform::transformUint64(rapidjson::Document &doc, int key, uint64_t arg) +{ + using namespace rapidjson; + + switch (key) { + case IConfig::CPUAffinityKey: /* --cpu-affinity */ + m_affinity = static_cast(arg); + break; + + case IConfig::ThreadsKey: /* --threads */ + m_threads = arg; + break; + + case IConfig::AVKey: /* --av */ + m_intensity = intensity(arg); + set(doc, kCpu, "hw-aes", isHwAes(arg)); + break; + + case IConfig::CPUPriorityKey: /* --cpu-priority */ + return set(doc, kCpu, "priority", arg); + + default: + break; + } +} + diff --git a/src/core/config/ConfigTransform.h b/src/core/config/ConfigTransform.h new file mode 100644 index 000000000..440a71695 --- /dev/null +++ b/src/core/config/ConfigTransform.h @@ -0,0 +1,57 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_CONFIGTRANSFORM_H +#define XMRIG_CONFIGTRANSFORM_H + + +#include "base/kernel/config/BaseTransform.h" + + +namespace xmrig { + + +class ConfigTransform : public BaseTransform +{ +public: + ConfigTransform(); + +protected: + void finalize(rapidjson::Document &doc) override; + void transform(rapidjson::Document &doc, int key, const char *arg) override; + +private: + void transformBoolean(rapidjson::Document &doc, int key, bool enable); + void transformUint64(rapidjson::Document &doc, int key, uint64_t arg); + + int64_t m_affinity = -1; + uint64_t m_intensity = 1; + uint64_t m_threads = 0; +}; + + +} // namespace xmrig + + +#endif /* XMRIG_CONFIGTRANSFORM_H */ diff --git a/src/core/ConfigLoader_default.h b/src/core/config/Config_default.h similarity index 74% rename from src/core/ConfigLoader_default.h rename to src/core/config/Config_default.h index 8fd0502b7..2f8534461 100644 --- a/src/core/ConfigLoader_default.h +++ b/src/core/config/Config_default.h @@ -22,8 +22,8 @@ * along with this program. If not, see . */ -#ifndef XMRIG_CONFIGLOADER_DEFAULT_H -#define XMRIG_CONFIGLOADER_DEFAULT_H +#ifndef XMRIG_CONFIG_DEFAULT_H +#define XMRIG_CONFIG_DEFAULT_H namespace xmrig { @@ -33,52 +33,63 @@ namespace xmrig { const static char *default_config = R"===( { - "algo": "cryptonight", "api": { + "id": null, + "worker-id": null + }, + "http": { + "enabled": false, + "host": "127.0.0.1", "port": 0, "access-token": null, - "id": null, - "worker-id": null, - "ipv6": false, "restricted": true }, - "asm": true, "autosave": true, - "av": 0, "background": false, "colors": true, - "cpu-affinity": null, - "cpu-priority": null, + "randomx": { + "init": -1, + "numa": true + }, + "cpu": { + "enabled": true, + "huge-pages": true, + "hw-aes": null, + "priority": null, + "asm": true, + "cn/0": false, + "cn-lite/0": false + }, "donate-level": 5, - "huge-pages": true, - "hw-aes": null, + "donate-over-proxy": 1, "log-file": null, - "max-cpu-usage": 100, "pools": [ { + "algo": null, "url": "donate.v2.xmrig.com:3333", "user": "YOUR_WALLET_ADDRESS", "pass": "x", "rig-id": null, "nicehash": false, "keepalive": false, - "variant": -1, + "enabled": true, "tls": false, - "tls-fingerprint": null + "tls-fingerprint": null, + "daemon": false } ], "print-time": 60, "retries": 5, "retry-pause": 5, - "safe": false, - "threads": null, + "syslog": false, "user-agent": null, - "watch": false + "watch": true } )==="; #endif -} /* namespace xmrig */ +} // namespace xmrig -#endif /* XMRIG_CONFIGLOADER_DEFAULT_H */ + +#endif /* XMRIG_CONFIG_DEFAULT_H */ diff --git a/src/core/config/Config_platform.h b/src/core/config/Config_platform.h new file mode 100644 index 000000000..b7415f4dc --- /dev/null +++ b/src/core/config/Config_platform.h @@ -0,0 +1,93 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_CONFIG_PLATFORM_H +#define XMRIG_CONFIG_PLATFORM_H + + +#ifdef _MSC_VER +# include "getopt/getopt.h" +#else +# include +#endif + + +#include "base/kernel/interfaces/IConfig.h" +#include "version.h" + + +namespace xmrig { + + +static const char short_options[] = "a:c:kBp:Px:r:R:s:t:T:o:u:O:v:l:S"; + + +static const option options[] = { + { "algo", 1, nullptr, IConfig::AlgorithmKey }, + { "api-worker-id", 1, nullptr, IConfig::ApiWorkerIdKey }, + { "api-id", 1, nullptr, IConfig::ApiIdKey }, + { "http-enabled", 0, nullptr, IConfig::HttpEnabledKey }, + { "http-host", 1, nullptr, IConfig::HttpHostKey }, + { "http-access-token", 1, nullptr, IConfig::HttpAccessTokenKey }, + { "http-port", 1, nullptr, IConfig::HttpPort }, + { "http-no-restricted", 0, nullptr, IConfig::HttpRestrictedKey }, + { "av", 1, nullptr, IConfig::AVKey }, + { "background", 0, nullptr, IConfig::BackgroundKey }, + { "config", 1, nullptr, IConfig::ConfigKey }, + { "cpu-affinity", 1, nullptr, IConfig::CPUAffinityKey }, + { "cpu-priority", 1, nullptr, IConfig::CPUPriorityKey }, + { "donate-level", 1, nullptr, IConfig::DonateLevelKey }, + { "donate-over-proxy", 1, nullptr, IConfig::ProxyDonateKey }, + { "dry-run", 0, nullptr, IConfig::DryRunKey }, + { "keepalive", 0, nullptr, IConfig::KeepAliveKey }, + { "log-file", 1, nullptr, IConfig::LogFileKey }, + { "nicehash", 0, nullptr, IConfig::NicehashKey }, + { "no-color", 0, nullptr, IConfig::ColorKey }, + { "no-huge-pages", 0, nullptr, IConfig::HugePagesKey }, + { "pass", 1, nullptr, IConfig::PasswordKey }, + { "print-time", 1, nullptr, IConfig::PrintTimeKey }, + { "retries", 1, nullptr, IConfig::RetriesKey }, + { "retry-pause", 1, nullptr, IConfig::RetryPauseKey }, + { "syslog", 0, nullptr, IConfig::SyslogKey }, + { "threads", 1, nullptr, IConfig::ThreadsKey }, + { "url", 1, nullptr, IConfig::UrlKey }, + { "user", 1, nullptr, IConfig::UserKey }, + { "user-agent", 1, nullptr, IConfig::UserAgentKey }, + { "userpass", 1, nullptr, IConfig::UserpassKey }, + { "rig-id", 1, nullptr, IConfig::RigIdKey }, + { "tls", 0, nullptr, IConfig::TlsKey }, + { "tls-fingerprint", 1, nullptr, IConfig::FingerprintKey }, + { "asm", 1, nullptr, IConfig::AssemblyKey }, + { "daemon", 0, nullptr, IConfig::DaemonKey }, + { "daemon-poll-interval", 1, nullptr, IConfig::DaemonPollKey }, + { "randomx-init", 1, nullptr, IConfig::RandomXInitKey }, + { "randomx-no-numa", 0, nullptr, IConfig::RandomXNumaKey }, + { nullptr, 0, nullptr, 0 } +}; + + +} // namespace xmrig + + +#endif /* XMRIG_CONFIG_PLATFORM_H */ diff --git a/src/core/config/usage.h b/src/core/config/usage.h new file mode 100644 index 000000000..b41ec6db8 --- /dev/null +++ b/src/core/config/usage.h @@ -0,0 +1,129 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_USAGE_H +#define XMRIG_USAGE_H + + +#include "version.h" + + +namespace xmrig { + + +static char const usage[] = "\ +Usage: " APP_ID " [OPTIONS]\n\ +Options:\n\ + -a, --algo=ALGO specify the algorithm to use\n\ + cn/r, cn/2, cn/1, cn/0, cn/double, cn/half, cn/fast,\n\ + cn/rwz, cn/zls, cn/xao, cn/rto" +#ifdef XMRIG_ALGO_CN_GPU +", cn/gpu,\n" +#else +",\n" +#endif +#ifdef XMRIG_ALGO_CN_LITE +"\ + cn-lite/1,\n" +#endif +#ifdef XMRIG_ALGO_CN_HEAVY +"\ + cn-heavy/xhv, cn-heavy/tube, cn-heavy/0,\n" +#endif +#ifdef XMRIG_ALGO_CN_PICO +"\ + cn-pico,\n" +#endif +#ifdef XMRIG_ALGO_RANDOMX +"\ + rx/wow, rx/loki\n" +#endif +"\ + -o, --url=URL URL of mining server\n\ + -O, --userpass=U:P username:password pair for mining server\n\ + -u, --user=USERNAME username for mining server\n\ + -p, --pass=PASSWORD password for mining server\n\ + --rig-id=ID rig identifier for pool-side statistics (needs pool support)\n\ + -t, --threads=N number of miner threads\n\ + -v, --av=N algorithm variation, 0 auto select\n\ + -k, --keepalive send keepalived packet for prevent timeout (needs pool support)\n\ + --nicehash enable nicehash.com support\n" +#ifdef XMRIG_FEATURE_TLS +"\ + --tls enable SSL/TLS support (needs pool support)\n\ + --tls-fingerprint=F pool TLS certificate fingerprint, if set enable strict certificate pinning\n" +#endif +#ifdef XMRIG_FEATURE_HTTP +"\ + --daemon use daemon RPC instead of pool for solo mining\n\ + --daemon-poll-interval=N daemon poll interval in milliseconds (default: 1000)\n" +#endif +"\ + -r, --retries=N number of times to retry before switch to backup server (default: 5)\n\ + -R, --retry-pause=N time to pause between retries (default: 5)\n\ + --cpu-affinity set process affinity to CPU core(s), mask 0x3 for cores 0 and 1\n\ + --cpu-priority set process priority (0 idle, 2 normal to 5 highest)\n\ + --no-huge-pages disable huge pages support\n\ + --no-color disable colored output\n\ + --donate-level=N donate level, default 5%% (5 minutes in 100 minutes)\n\ + --user-agent set custom user-agent string for pool\n\ + -B, --background run the miner in the background\n\ + -c, --config=FILE load a JSON-format configuration file\n\ + -l, --log-file=FILE log all output to a file\n" +# ifdef HAVE_SYSLOG_H +"\ + -S, --syslog use system log for output messages\n" +# endif +"\ + --asm=ASM ASM optimizations, possible values: auto, none, intel, ryzen, bulldozer.\n\ + --print-time=N print hashrate report every N seconds\n" +#ifdef XMRIG_FEATURE_HTTP +"\ + --api-worker-id=ID custom worker-id for API\n\ + --api-id=ID custom instance ID for API\n\ + --http-enabled enable HTTP API\n\ + --http-host=HOST bind host for HTTP API (default: 127.0.0.1)\n\ + --http-port=N bind port for HTTP API\n\ + --http-access-token=T access token for HTTP API\n\ + --http-no-restricted enable full remote access to HTTP API (only if access token set)\n" +#endif +#ifdef XMRIG_ALGO_RANDOMX +"\ + --randomx-init=N threads count to initialize RandomX dataset\n\ + --randomx-no-numa disable NUMA support for RandomX\n" +#endif +#ifdef XMRIG_FEATURE_HWLOC +"\ + --export-topology export hwloc topology to a XML file and exit\n" +#endif +"\ + --dry-run test configuration and exit\n\ + -h, --help display this help and exit\n\ + -V, --version output version information and exit\n\ +"; + + +} /* namespace xmrig */ + +#endif /* XMRIG_USAGE_H */ diff --git a/src/core/cpu/Cpu.cpp b/src/core/cpu/Cpu.cpp deleted file mode 100644 index 773255d27..000000000 --- a/src/core/cpu/Cpu.cpp +++ /dev/null @@ -1,61 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2016-2017 XMRig - * - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - - -#include - - -#include "common/cpu/Cpu.h" - - -#ifndef XMRIG_NO_LIBCPUID -# include "core/cpu/AdvancedCpuInfo.h" -#endif - - -static xmrig::ICpuInfo *cpuInfo = nullptr; - - -xmrig::ICpuInfo *xmrig::Cpu::info() -{ - assert(cpuInfo != nullptr); - - return cpuInfo; -} - - -void xmrig::Cpu::init() -{ - assert(cpuInfo == nullptr); - - cpuInfo = new AdvancedCpuInfo(); -} - - -void xmrig::Cpu::release() -{ - assert(cpuInfo != nullptr); - - delete cpuInfo; - cpuInfo = nullptr; -} diff --git a/src/core/usage.h b/src/core/usage.h deleted file mode 100644 index 0d5c47816..000000000 --- a/src/core/usage.h +++ /dev/null @@ -1,95 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef XMRIG_USAGE_H -#define XMRIG_USAGE_H - - -#include "version.h" - - -namespace xmrig { - - -static char const usage[] = "\ -Usage: " APP_ID " [OPTIONS]\n\ -Options:\n\ - -a, --algo=ALGO specify the algorithm to use\n\ - cryptonight\n" -#ifndef XMRIG_NO_AEON -"\ - cryptonight-lite\n" -#endif -#ifndef XMRIG_NO_SUMO -"\ - cryptonight-heavy\n" -#endif -"\ - -o, --url=URL URL of mining server\n\ - -O, --userpass=U:P username:password pair for mining server\n\ - -u, --user=USERNAME username for mining server\n\ - -p, --pass=PASSWORD password for mining server\n\ - --rig-id=ID rig identifier for pool-side statistics (needs pool support)\n\ - -t, --threads=N number of miner threads\n\ - -v, --av=N algorithm variation, 0 auto select\n\ - -k, --keepalive send keepalived packet for prevent timeout (needs pool support)\n\ - --nicehash enable nicehash.com support\n\ - --tls enable SSL/TLS support (needs pool support)\n\ - --tls-fingerprint=F pool TLS certificate fingerprint, if set enable strict certificate pinning\n\ - -r, --retries=N number of times to retry before switch to backup server (default: 5)\n\ - -R, --retry-pause=N time to pause between retries (default: 5)\n\ - --cpu-affinity set process affinity to CPU core(s), mask 0x3 for cores 0 and 1\n\ - --cpu-priority set process priority (0 idle, 2 normal to 5 highest)\n\ - --no-huge-pages disable huge pages support\n\ - --no-color disable colored output\n\ - --variant algorithm PoW variant\n\ - --donate-level=N donate level, default 5%% (5 minutes in 100 minutes)\n\ - --user-agent set custom user-agent string for pool\n\ - -B, --background run the miner in the background\n\ - -c, --config=FILE load a JSON-format configuration file\n\ - -l, --log-file=FILE log all output to a file\n" -# ifdef HAVE_SYSLOG_H -"\ - -S, --syslog use system log for output messages\n" -# endif -"\ - --max-cpu-usage=N maximum CPU usage for automatic threads mode (default 75)\n\ - --safe safe adjust threads and av settings for current CPU\n\ - --asm=ASM ASM code for cn/2, possible values: auto, none, intel, ryzen, bulldozer.\n\ - --print-time=N print hashrate report every N seconds\n\ - --api-port=N port for the miner API\n\ - --api-access-token=T access token for API\n\ - --api-worker-id=ID custom worker-id for API\n\ - --api-id=ID custom instance ID for API\n\ - --api-ipv6 enable IPv6 support for API\n\ - --api-no-restricted enable full remote access (only if API token set)\n\ - --dry-run test configuration and exit\n\ - -h, --help display this help and exit\n\ - -V, --version output version information and exit\n\ -"; - - -} /* namespace xmrig */ - -#endif /* XMRIG_USAGE_H */ diff --git a/src/crypto/Asm.h b/src/crypto/Asm.h deleted file mode 100644 index 3b755fd64..000000000 --- a/src/crypto/Asm.h +++ /dev/null @@ -1,50 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef XMRIG_ASM_H -#define XMRIG_ASM_H - - -#include "common/xmrig.h" -#include "rapidjson/fwd.h" - - -namespace xmrig { - - -class Asm -{ -public: - static Assembly parse(const char *assembly, Assembly defaultValue = ASM_AUTO); - static Assembly parse(const rapidjson::Value &value, Assembly defaultValue = ASM_AUTO); - static const char *toString(Assembly assembly); - static rapidjson::Value toJSON(Assembly assembly); - - inline static Assembly parse(bool enable) { return enable ? ASM_AUTO : ASM_NONE; } -}; - - -} /* namespace xmrig */ - - -#endif /* XMRIG_ASM_H */ diff --git a/src/crypto/CryptoNight_constants.h b/src/crypto/CryptoNight_constants.h deleted file mode 100644 index 58a3915fc..000000000 --- a/src/crypto/CryptoNight_constants.h +++ /dev/null @@ -1,225 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2019 XMR-Stak , - * Copyright 2018 Lee Clagett - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef XMRIG_CRYPTONIGHT_CONSTANTS_H -#define XMRIG_CRYPTONIGHT_CONSTANTS_H - - -#include -#include - - -#include "common/xmrig.h" - - -namespace xmrig -{ - -constexpr const size_t CRYPTONIGHT_MEMORY = 2 * 1024 * 1024; -constexpr const uint32_t CRYPTONIGHT_MASK = 0x1FFFF0; -constexpr const uint32_t CRYPTONIGHT_ITER = 0x80000; -constexpr const uint32_t CRYPTONIGHT_HALF_ITER = 0x40000; -constexpr const uint32_t CRYPTONIGHT_XAO_ITER = 0x100000; -constexpr const uint32_t CRYPTONIGHT_DOUBLE_ITER = 0x100000; -constexpr const uint32_t CRYPTONIGHT_WALTZ_ITER = 0x60000; -constexpr const uint32_t CRYPTONIGHT_ZLS_ITER = 0x60000; - -constexpr const uint32_t CRYPTONIGHT_GPU_ITER = 0xC000; -constexpr const uint32_t CRYPTONIGHT_GPU_MASK = 0x1FFFC0; - -constexpr const size_t CRYPTONIGHT_LITE_MEMORY = 1 * 1024 * 1024; -constexpr const uint32_t CRYPTONIGHT_LITE_MASK = 0xFFFF0; -constexpr const uint32_t CRYPTONIGHT_LITE_ITER = 0x40000; - -constexpr const size_t CRYPTONIGHT_HEAVY_MEMORY = 4 * 1024 * 1024; -constexpr const uint32_t CRYPTONIGHT_HEAVY_MASK = 0x3FFFF0; -constexpr const uint32_t CRYPTONIGHT_HEAVY_ITER = 0x40000; - -constexpr const size_t CRYPTONIGHT_PICO_MEMORY = 256 * 1024; -constexpr const uint32_t CRYPTONIGHT_PICO_MASK = 0x1FFF0; -constexpr const uint32_t CRYPTONIGHT_PICO_ITER = 0x40000; -constexpr const uint32_t CRYPTONIGHT_TRTL_ITER = 0x10000; - - -template inline constexpr size_t cn_select_memory() { return 0; } -template<> inline constexpr size_t cn_select_memory() { return CRYPTONIGHT_MEMORY; } -template<> inline constexpr size_t cn_select_memory() { return CRYPTONIGHT_LITE_MEMORY; } -template<> inline constexpr size_t cn_select_memory() { return CRYPTONIGHT_HEAVY_MEMORY; } -template<> inline constexpr size_t cn_select_memory() { return CRYPTONIGHT_PICO_MEMORY; } - - -inline size_t cn_select_memory(Algo algorithm) -{ - switch(algorithm) - { - case CRYPTONIGHT: - return CRYPTONIGHT_MEMORY; - - case CRYPTONIGHT_LITE: - return CRYPTONIGHT_LITE_MEMORY; - - case CRYPTONIGHT_HEAVY: - return CRYPTONIGHT_HEAVY_MEMORY; - - case CRYPTONIGHT_PICO: - return CRYPTONIGHT_PICO_MEMORY; - - default: - break; - } - - return 0; -} - - -template inline constexpr uint32_t cn_select_mask() { return 0; } -template<> inline constexpr uint32_t cn_select_mask() { return CRYPTONIGHT_MASK; } -template<> inline constexpr uint32_t cn_select_mask() { return CRYPTONIGHT_LITE_MASK; } -template<> inline constexpr uint32_t cn_select_mask() { return CRYPTONIGHT_HEAVY_MASK; } -template<> inline constexpr uint32_t cn_select_mask() { return CRYPTONIGHT_PICO_MASK; } - - -inline uint32_t cn_select_mask(Algo algorithm) -{ - switch(algorithm) - { - case CRYPTONIGHT: - return CRYPTONIGHT_MASK; - - case CRYPTONIGHT_LITE: - return CRYPTONIGHT_LITE_MASK; - - case CRYPTONIGHT_HEAVY: - return CRYPTONIGHT_HEAVY_MASK; - - case CRYPTONIGHT_PICO: - return CRYPTONIGHT_PICO_MASK; - - default: - break; - } - - return 0; -} - - -template inline constexpr uint32_t cn_select_iter() { return 0; } -template<> inline constexpr uint32_t cn_select_iter() { return CRYPTONIGHT_ITER; } -template<> inline constexpr uint32_t cn_select_iter() { return CRYPTONIGHT_ITER; } -template<> inline constexpr uint32_t cn_select_iter() { return CRYPTONIGHT_ITER; } -template<> inline constexpr uint32_t cn_select_iter() { return CRYPTONIGHT_ITER; } -template<> inline constexpr uint32_t cn_select_iter() { return CRYPTONIGHT_ITER; } -template<> inline constexpr uint32_t cn_select_iter() { return CRYPTONIGHT_ITER; } -template<> inline constexpr uint32_t cn_select_iter() { return CRYPTONIGHT_HALF_ITER; } -template<> inline constexpr uint32_t cn_select_iter() { return CRYPTONIGHT_HALF_ITER; } -template<> inline constexpr uint32_t cn_select_iter() { return CRYPTONIGHT_XAO_ITER; } -template<> inline constexpr uint32_t cn_select_iter() { return CRYPTONIGHT_ITER; } -template<> inline constexpr uint32_t cn_select_iter() { return CRYPTONIGHT_GPU_ITER; } -template<> inline constexpr uint32_t cn_select_iter() { return CRYPTONIGHT_WALTZ_ITER; } -template<> inline constexpr uint32_t cn_select_iter() { return CRYPTONIGHT_ZLS_ITER; } -template<> inline constexpr uint32_t cn_select_iter() { return CRYPTONIGHT_DOUBLE_ITER; } -template<> inline constexpr uint32_t cn_select_iter() { return CRYPTONIGHT_LITE_ITER; } -template<> inline constexpr uint32_t cn_select_iter() { return CRYPTONIGHT_LITE_ITER; } -template<> inline constexpr uint32_t cn_select_iter() { return CRYPTONIGHT_HEAVY_ITER; } -template<> inline constexpr uint32_t cn_select_iter() { return CRYPTONIGHT_HEAVY_ITER; } -template<> inline constexpr uint32_t cn_select_iter() { return CRYPTONIGHT_HEAVY_ITER; } -template<> inline constexpr uint32_t cn_select_iter() { return CRYPTONIGHT_TRTL_ITER; } - - -inline uint32_t cn_select_iter(Algo algorithm, Variant variant) -{ - switch (variant) { - case VARIANT_MSR: - case VARIANT_HALF: - return CRYPTONIGHT_HALF_ITER; - - case VARIANT_GPU: - return CRYPTONIGHT_GPU_ITER; - - case VARIANT_RTO: - case VARIANT_DOUBLE: - return CRYPTONIGHT_XAO_ITER; - - case VARIANT_TRTL: - return CRYPTONIGHT_TRTL_ITER; - - case VARIANT_RWZ: - case VARIANT_ZLS: - return CRYPTONIGHT_WALTZ_ITER; - - default: - break; - } - - switch(algorithm) - { - case CRYPTONIGHT: - return CRYPTONIGHT_ITER; - - case CRYPTONIGHT_LITE: - return CRYPTONIGHT_LITE_ITER; - - case CRYPTONIGHT_HEAVY: - return CRYPTONIGHT_HEAVY_ITER; - - case CRYPTONIGHT_PICO: - return CRYPTONIGHT_TRTL_ITER; - - default: - break; - } - - return 0; -} - - -template inline constexpr Variant cn_base_variant() { return VARIANT_0; } -template<> inline constexpr Variant cn_base_variant() { return VARIANT_0; } -template<> inline constexpr Variant cn_base_variant() { return VARIANT_1; } -template<> inline constexpr Variant cn_base_variant() { return VARIANT_1; } -template<> inline constexpr Variant cn_base_variant() { return VARIANT_1; } -template<> inline constexpr Variant cn_base_variant() { return VARIANT_1; } -template<> inline constexpr Variant cn_base_variant() { return VARIANT_0; } -template<> inline constexpr Variant cn_base_variant() { return VARIANT_0; } -template<> inline constexpr Variant cn_base_variant() { return VARIANT_1; } -template<> inline constexpr Variant cn_base_variant() { return VARIANT_2; } -template<> inline constexpr Variant cn_base_variant() { return VARIANT_2; } -template<> inline constexpr Variant cn_base_variant() { return VARIANT_2; } -template<> inline constexpr Variant cn_base_variant() { return VARIANT_GPU; } -template<> inline constexpr Variant cn_base_variant() { return VARIANT_2; } -template<> inline constexpr Variant cn_base_variant() { return VARIANT_2; } -template<> inline constexpr Variant cn_base_variant() { return VARIANT_2; } -template<> inline constexpr Variant cn_base_variant() { return VARIANT_2; } -template<> inline constexpr Variant cn_base_variant() { return VARIANT_2; } - - -template inline constexpr bool cn_is_cryptonight_r() { return false; } -template<> inline constexpr bool cn_is_cryptonight_r() { return true; } -template<> inline constexpr bool cn_is_cryptonight_r() { return true; } - -} /* namespace xmrig */ - - -#endif /* XMRIG_CRYPTONIGHT_CONSTANTS_H */ diff --git a/src/crypto/cn/CnAlgo.h b/src/crypto/cn/CnAlgo.h new file mode 100644 index 000000000..ed64331a9 --- /dev/null +++ b/src/crypto/cn/CnAlgo.h @@ -0,0 +1,222 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2019 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_CN_ALGO_H +#define XMRIG_CN_ALGO_H + + +#include +#include + + +#include "crypto/common/Algorithm.h" + + +namespace xmrig +{ + + +template +class CnAlgo +{ +public: + constexpr inline CnAlgo() + { + static_assert(ALGO != Algorithm::INVALID && m_memory[ALGO] > 0, "invalid CRYPTONIGHT algorithm"); + static_assert(sizeof(m_memory) / sizeof(m_memory)[0] == Algorithm::MAX, "memory table size mismatch"); + static_assert(sizeof(m_iterations) / sizeof(m_iterations)[0] == Algorithm::MAX, "iterations table size mismatch"); + static_assert(sizeof(m_base) / sizeof(m_base)[0] == Algorithm::MAX, "iterations table size mismatch"); + } + + constexpr inline Algorithm::Id base() const { return m_base[ALGO]; } + constexpr inline bool isHeavy() const { return memory() == CN_MEMORY * 2; } + constexpr inline bool isR() const { return ALGO == Algorithm::CN_R || ALGO == Algorithm::CN_WOW; } + constexpr inline size_t memory() const { return m_memory[ALGO]; } + constexpr inline uint32_t iterations() const { return m_iterations[ALGO]; } + constexpr inline uint32_t mask() const { return ((memory() - 1) / 16) * 16; } + + inline static size_t memory(Algorithm::Id algo) + { + switch (Algorithm::family(algo)) { + case Algorithm::CN: + return CN_MEMORY; + + case Algorithm::CN_LITE: + return CN_MEMORY / 2; + + case Algorithm::CN_HEAVY: + return CN_MEMORY * 2; + + case Algorithm::CN_PICO: + return CN_MEMORY / 8; + + default: + break; + } + + return 0; + } + + inline static uint32_t mask(Algorithm::Id algo) + { +# ifdef XMRIG_ALGO_CN_GPU + if (algo == Algorithm::CN_GPU) { + return 0x1FFFC0; + } +# endif + +# ifdef XMRIG_ALGO_CN_PICO + if (algo == Algorithm::CN_PICO_0) { + return 0x1FFF0; + } +# endif + + return ((memory(algo) - 1) / 16) * 16; + } + +private: + constexpr const static size_t CN_MEMORY = 0x200000; + constexpr const static uint32_t CN_ITER = 0x80000; + + constexpr const static size_t m_memory[] = { + CN_MEMORY, // CN_0 + CN_MEMORY, // CN_1 + CN_MEMORY, // CN_2 + CN_MEMORY, // CN_R + CN_MEMORY, // CN_WOW + CN_MEMORY, // CN_FAST + CN_MEMORY, // CN_HALF + CN_MEMORY, // CN_XAO + CN_MEMORY, // CN_RTO + CN_MEMORY, // CN_RWZ + CN_MEMORY, // CN_ZLS + CN_MEMORY, // CN_DOUBLE +# ifdef XMRIG_ALGO_CN_GPU + CN_MEMORY, // CN_GPU +# endif +# ifdef XMRIG_ALGO_CN_LITE + CN_MEMORY / 2, // CN_LITE_0 + CN_MEMORY / 2, // CN_LITE_1 +# endif +# ifdef XMRIG_ALGO_CN_HEAVY + CN_MEMORY * 2, // CN_HEAVY_0 + CN_MEMORY * 2, // CN_HEAVY_TUBE + CN_MEMORY * 2, // CN_HEAVY_XHV +# endif +# ifdef XMRIG_ALGO_CN_PICO + CN_MEMORY / 8, // CN_PICO_0 +# endif +# ifdef XMRIG_ALGO_RANDOMX + 0, // RX_0 + 0, // RX_WOW + 0, // RX_LOKI +# endif + }; + + constexpr const static uint32_t m_iterations[] = { + CN_ITER, // CN_0 + CN_ITER, // CN_1 + CN_ITER, // CN_2 + CN_ITER, // CN_R + CN_ITER, // CN_WOW + CN_ITER / 2, // CN_FAST + CN_ITER / 2, // CN_HALF + CN_ITER * 2, // CN_XAO + CN_ITER, // CN_RTO + 0x60000, // CN_RWZ + 0x60000, // CN_ZLS + CN_ITER * 2, // CN_DOUBLE +# ifdef XMRIG_ALGO_CN_GPU + 0xC000, // CN_GPU +# endif +# ifdef XMRIG_ALGO_CN_LITE + CN_ITER / 2, // CN_LITE_0 + CN_ITER / 2, // CN_LITE_1 +# endif +# ifdef XMRIG_ALGO_CN_HEAVY + CN_ITER / 2, // CN_HEAVY_0 + CN_ITER / 2, // CN_HEAVY_TUBE + CN_ITER / 2, // CN_HEAVY_XHV +# endif +# ifdef XMRIG_ALGO_CN_PICO + CN_ITER / 8, // CN_PICO_0 +# endif +# ifdef XMRIG_ALGO_RANDOMX + 0, // RX_0 + 0, // RX_WOW + 0, // RX_LOKI +# endif + }; + + constexpr const static Algorithm::Id m_base[] = { + Algorithm::CN_0, // CN_0 + Algorithm::CN_1, // CN_1 + Algorithm::CN_2, // CN_2 + Algorithm::CN_2, // CN_R + Algorithm::CN_2, // CN_WOW + Algorithm::CN_1, // CN_FAST + Algorithm::CN_2, // CN_HALF + Algorithm::CN_0, // CN_XAO + Algorithm::CN_1, // CN_RTO + Algorithm::CN_2, // CN_RWZ + Algorithm::CN_2, // CN_ZLS + Algorithm::CN_2, // CN_DOUBLE +# ifdef XMRIG_ALGO_CN_GPU + Algorithm::CN_GPU, // CN_GPU +# endif +# ifdef XMRIG_ALGO_CN_LITE + Algorithm::CN_0, // CN_LITE_0 + Algorithm::CN_1, // CN_LITE_1 +# endif +# ifdef XMRIG_ALGO_CN_HEAVY + Algorithm::CN_0, // CN_HEAVY_0 + Algorithm::CN_1, // CN_HEAVY_TUBE + Algorithm::CN_0, // CN_HEAVY_XHV +# endif +# ifdef XMRIG_ALGO_CN_PICO + Algorithm::CN_2, // CN_PICO_0, +# endif +# ifdef XMRIG_ALGO_RANDOMX + Algorithm::INVALID, // RX_0 + Algorithm::INVALID, // RX_WOW + Algorithm::INVALID, // RX_LOKI +# endif + }; +}; + + +#ifdef XMRIG_ALGO_CN_GPU +template<> constexpr inline uint32_t CnAlgo::mask() const { return 0x1FFFC0; } +#endif + +#ifdef XMRIG_ALGO_CN_PICO +template<> constexpr inline uint32_t CnAlgo::mask() const { return 0x1FFF0; } +#endif + + +} /* namespace xmrig */ + + +#endif /* XMRIG_CN_ALGO_H */ diff --git a/src/Mem.cpp b/src/crypto/cn/CnCtx.cpp similarity index 54% rename from src/Mem.cpp rename to src/crypto/cn/CnCtx.cpp index 01a2157b3..5d41bca0d 100644 --- a/src/Mem.cpp +++ b/src/crypto/cn/CnCtx.cpp @@ -4,7 +4,7 @@ * Copyright 2014 Lucas Jones * Copyright 2014-2016 Wolf9466 * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , + * Copyright 2017-2019 XMR-Stak , * Copyright 2018 Lee Clagett * Copyright 2018-2019 SChernykh * Copyright 2016-2019 XMRig , @@ -23,55 +23,38 @@ * along with this program. If not, see . */ - -#include "common/utils/mm_malloc.h" -#include "crypto/CryptoNight.h" -#include "crypto/CryptoNight_constants.h" -#include "Mem.h" +#include -bool Mem::m_enabled = true; -int Mem::m_flags = 0; +#include "crypto/cn/CnCtx.h" +#include "crypto/cn/CryptoNight.h" +#include "crypto/common/Algorithm.h" +#include "crypto/common/portable/mm_malloc.h" +#include "crypto/common/VirtualMemory.h" -MemInfo Mem::create(cryptonight_ctx **ctx, xmrig::Algo algorithm, size_t count) +void xmrig::CnCtx::create(cryptonight_ctx **ctx, uint8_t *memory, size_t size, size_t count) { - using namespace xmrig; - - MemInfo info; - info.size = cn_select_memory(algorithm) * count; - - constexpr const size_t align_size = 2 * 1024 * 1024; - info.size = ((info.size + align_size - 1) / align_size) * align_size; - info.pages = info.size / align_size; - - allocate(info, m_enabled); - for (size_t i = 0; i < count; ++i) { cryptonight_ctx *c = static_cast(_mm_malloc(sizeof(cryptonight_ctx), 4096)); - c->memory = info.memory + (i * cn_select_memory(algorithm)); + c->memory = memory + (i * size); - uint8_t* p = reinterpret_cast(allocateExecutableMemory(0x4000)); - c->generated_code = reinterpret_cast(p); - c->generated_code_double = reinterpret_cast(p + 0x2000); - - c->generated_code_data.variant = xmrig::VARIANT_MAX; - c->generated_code_data.height = (uint64_t)(-1); - c->generated_code_double_data = c->generated_code_data; + c->generated_code = reinterpret_cast(VirtualMemory::allocateExecutableMemory(0x4000)); + c->generated_code_data.algo = Algorithm::INVALID; + c->generated_code_data.height = std::numeric_limits::max(); ctx[i] = c; } - - return info; } -void Mem::release(cryptonight_ctx **ctx, size_t count, MemInfo &info) +void xmrig::CnCtx::release(cryptonight_ctx **ctx, size_t count) { - release(info); + if (ctx[0] == nullptr) { + return; + } for (size_t i = 0; i < count; ++i) { _mm_free(ctx[i]); } } - diff --git a/src/crypto/cn/CnCtx.h b/src/crypto/cn/CnCtx.h new file mode 100644 index 000000000..7b0adbec2 --- /dev/null +++ b/src/crypto/cn/CnCtx.h @@ -0,0 +1,52 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2019 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_CN_CTX_H +#define XMRIG_CN_CTX_H + + +#include +#include + + +struct cryptonight_ctx; + + +namespace xmrig +{ + + +class CnCtx +{ +public: + static void create(cryptonight_ctx **ctx, uint8_t *memory, size_t size, size_t count); + static void release(cryptonight_ctx **ctx, size_t count); +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_CN_CTX_H */ diff --git a/src/crypto/cn/CnHash.cpp b/src/crypto/cn/CnHash.cpp new file mode 100644 index 000000000..a2f8880cf --- /dev/null +++ b/src/crypto/cn/CnHash.cpp @@ -0,0 +1,272 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2019 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#include + + +#include "backend/cpu/Cpu.h" +#include "crypto/cn/CnHash.h" +#include "crypto/common/VirtualMemory.h" + + +#if defined(XMRIG_ARM) +# include "crypto/cn/CryptoNight_arm.h" +#else +# include "crypto/cn/CryptoNight_x86.h" +#endif + + +#define ADD_FN(algo) \ + m_map[algo][AV_SINGLE][Assembly::NONE] = cryptonight_single_hash; \ + m_map[algo][AV_SINGLE_SOFT][Assembly::NONE] = cryptonight_single_hash; \ + m_map[algo][AV_DOUBLE][Assembly::NONE] = cryptonight_double_hash; \ + m_map[algo][AV_DOUBLE_SOFT][Assembly::NONE] = cryptonight_double_hash; \ + m_map[algo][AV_TRIPLE][Assembly::NONE] = cryptonight_triple_hash; \ + m_map[algo][AV_TRIPLE_SOFT][Assembly::NONE] = cryptonight_triple_hash; \ + m_map[algo][AV_QUAD][Assembly::NONE] = cryptonight_quad_hash; \ + m_map[algo][AV_QUAD_SOFT][Assembly::NONE] = cryptonight_quad_hash; \ + m_map[algo][AV_PENTA][Assembly::NONE] = cryptonight_penta_hash; \ + m_map[algo][AV_PENTA_SOFT][Assembly::NONE] = cryptonight_penta_hash; + + +#ifdef XMRIG_FEATURE_ASM +# define ADD_FN_ASM(algo) \ + m_map[algo][AV_SINGLE][Assembly::INTEL] = cryptonight_single_hash_asm; \ + m_map[algo][AV_SINGLE][Assembly::RYZEN] = cryptonight_single_hash_asm; \ + m_map[algo][AV_SINGLE][Assembly::BULLDOZER] = cryptonight_single_hash_asm; \ + m_map[algo][AV_DOUBLE][Assembly::INTEL] = cryptonight_double_hash_asm; \ + m_map[algo][AV_DOUBLE][Assembly::RYZEN] = cryptonight_double_hash_asm; \ + m_map[algo][AV_DOUBLE][Assembly::BULLDOZER] = cryptonight_double_hash_asm; + + +extern "C" void cnv2_mainloop_ivybridge_asm(cryptonight_ctx **ctx); +extern "C" void cnv2_mainloop_ryzen_asm(cryptonight_ctx **ctx); +extern "C" void cnv2_mainloop_bulldozer_asm(cryptonight_ctx **ctx); +extern "C" void cnv2_double_mainloop_sandybridge_asm(cryptonight_ctx **ctx); + + +namespace xmrig { + + +cn_mainloop_fun cn_half_mainloop_ivybridge_asm = nullptr; +cn_mainloop_fun cn_half_mainloop_ryzen_asm = nullptr; +cn_mainloop_fun cn_half_mainloop_bulldozer_asm = nullptr; +cn_mainloop_fun cn_half_double_mainloop_sandybridge_asm = nullptr; + +cn_mainloop_fun cn_trtl_mainloop_ivybridge_asm = nullptr; +cn_mainloop_fun cn_trtl_mainloop_ryzen_asm = nullptr; +cn_mainloop_fun cn_trtl_mainloop_bulldozer_asm = nullptr; +cn_mainloop_fun cn_trtl_double_mainloop_sandybridge_asm = nullptr; + +cn_mainloop_fun cn_zls_mainloop_ivybridge_asm = nullptr; +cn_mainloop_fun cn_zls_mainloop_ryzen_asm = nullptr; +cn_mainloop_fun cn_zls_mainloop_bulldozer_asm = nullptr; +cn_mainloop_fun cn_zls_double_mainloop_sandybridge_asm = nullptr; + +cn_mainloop_fun cn_double_mainloop_ivybridge_asm = nullptr; +cn_mainloop_fun cn_double_mainloop_ryzen_asm = nullptr; +cn_mainloop_fun cn_double_mainloop_bulldozer_asm = nullptr; +cn_mainloop_fun cn_double_double_mainloop_sandybridge_asm = nullptr; + + +template +static void patchCode(T dst, U src, const uint32_t iterations, const uint32_t mask = CnAlgo().mask()) +{ + const uint8_t* p = reinterpret_cast(src); + + // Workaround for Visual Studio placing trampoline in debug builds. +# if defined(_MSC_VER) + if (p[0] == 0xE9) { + p += *(int32_t*)(p + 1) + 5; + } +# endif + + size_t size = 0; + while (*(uint32_t*)(p + size) != 0xDEADC0DE) { + ++size; + } + + size += sizeof(uint32_t); + + memcpy((void*) dst, (const void*) src, size); + + uint8_t* patched_data = reinterpret_cast(dst); + for (size_t i = 0; i + sizeof(uint32_t) <= size; ++i) { + switch (*(uint32_t*)(patched_data + i)) { + case CnAlgo().iterations(): + *(uint32_t*)(patched_data + i) = iterations; + break; + + case CnAlgo().mask(): + *(uint32_t*)(patched_data + i) = mask; + break; + } + } +} + + +static void patchAsmVariants() +{ + const int allocation_size = 65536; + uint8_t *base = static_cast(VirtualMemory::allocateExecutableMemory(allocation_size)); + + cn_half_mainloop_ivybridge_asm = reinterpret_cast (base + 0x0000); + cn_half_mainloop_ryzen_asm = reinterpret_cast (base + 0x1000); + cn_half_mainloop_bulldozer_asm = reinterpret_cast (base + 0x2000); + cn_half_double_mainloop_sandybridge_asm = reinterpret_cast (base + 0x3000); + +# ifdef XMRIG_ALGO_CN_PICO + cn_trtl_mainloop_ivybridge_asm = reinterpret_cast (base + 0x4000); + cn_trtl_mainloop_ryzen_asm = reinterpret_cast (base + 0x5000); + cn_trtl_mainloop_bulldozer_asm = reinterpret_cast (base + 0x6000); + cn_trtl_double_mainloop_sandybridge_asm = reinterpret_cast (base + 0x7000); +# endif + + cn_zls_mainloop_ivybridge_asm = reinterpret_cast (base + 0x8000); + cn_zls_mainloop_ryzen_asm = reinterpret_cast (base + 0x9000); + cn_zls_mainloop_bulldozer_asm = reinterpret_cast (base + 0xA000); + cn_zls_double_mainloop_sandybridge_asm = reinterpret_cast (base + 0xB000); + + cn_double_mainloop_ivybridge_asm = reinterpret_cast (base + 0xC000); + cn_double_mainloop_ryzen_asm = reinterpret_cast (base + 0xD000); + cn_double_mainloop_bulldozer_asm = reinterpret_cast (base + 0xE000); + cn_double_double_mainloop_sandybridge_asm = reinterpret_cast (base + 0xF000); + + { + constexpr uint32_t ITER = CnAlgo().iterations(); + + patchCode(cn_half_mainloop_ivybridge_asm, cnv2_mainloop_ivybridge_asm, ITER); + patchCode(cn_half_mainloop_ryzen_asm, cnv2_mainloop_ryzen_asm, ITER); + patchCode(cn_half_mainloop_bulldozer_asm, cnv2_mainloop_bulldozer_asm, ITER); + patchCode(cn_half_double_mainloop_sandybridge_asm, cnv2_double_mainloop_sandybridge_asm, ITER); + } + +# ifdef XMRIG_ALGO_CN_PICO + { + constexpr uint32_t ITER = CnAlgo().iterations(); + constexpr uint32_t MASK = CnAlgo().mask(); + + patchCode(cn_trtl_mainloop_ivybridge_asm, cnv2_mainloop_ivybridge_asm, ITER, MASK); + patchCode(cn_trtl_mainloop_ryzen_asm, cnv2_mainloop_ryzen_asm, ITER, MASK); + patchCode(cn_trtl_mainloop_bulldozer_asm, cnv2_mainloop_bulldozer_asm, ITER, MASK); + patchCode(cn_trtl_double_mainloop_sandybridge_asm, cnv2_double_mainloop_sandybridge_asm, ITER, MASK); + } +# endif + + { + constexpr uint32_t ITER = CnAlgo().iterations(); + + patchCode(cn_zls_mainloop_ivybridge_asm, cnv2_mainloop_ivybridge_asm, ITER); + patchCode(cn_zls_mainloop_ryzen_asm, cnv2_mainloop_ryzen_asm, ITER); + patchCode(cn_zls_mainloop_bulldozer_asm, cnv2_mainloop_bulldozer_asm, ITER); + patchCode(cn_zls_double_mainloop_sandybridge_asm, cnv2_double_mainloop_sandybridge_asm, ITER); + } + + { + constexpr uint32_t ITER = CnAlgo().iterations(); + + patchCode(cn_double_mainloop_ivybridge_asm, cnv2_mainloop_ivybridge_asm, ITER); + patchCode(cn_double_mainloop_ryzen_asm, cnv2_mainloop_ryzen_asm, ITER); + patchCode(cn_double_mainloop_bulldozer_asm, cnv2_mainloop_bulldozer_asm, ITER); + patchCode(cn_double_double_mainloop_sandybridge_asm, cnv2_double_mainloop_sandybridge_asm, ITER); + } + + VirtualMemory::protectExecutableMemory(base, allocation_size); + VirtualMemory::flushInstructionCache(base, allocation_size); +} +} // namespace xmrig +#else +# define ADD_FN_ASM(algo) +#endif + + +static const xmrig::CnHash cnHash; + + +xmrig::CnHash::CnHash() +{ + ADD_FN(Algorithm::CN_0); + ADD_FN(Algorithm::CN_1); + ADD_FN(Algorithm::CN_2); + ADD_FN(Algorithm::CN_R); + ADD_FN(Algorithm::CN_WOW); + ADD_FN(Algorithm::CN_FAST); + ADD_FN(Algorithm::CN_HALF); + ADD_FN(Algorithm::CN_XAO); + ADD_FN(Algorithm::CN_RTO); + ADD_FN(Algorithm::CN_RWZ); + ADD_FN(Algorithm::CN_ZLS); + ADD_FN(Algorithm::CN_DOUBLE); + + ADD_FN_ASM(Algorithm::CN_2); + ADD_FN_ASM(Algorithm::CN_HALF); + ADD_FN_ASM(Algorithm::CN_R); + ADD_FN_ASM(Algorithm::CN_WOW); + ADD_FN_ASM(Algorithm::CN_RWZ); + ADD_FN_ASM(Algorithm::CN_ZLS); + ADD_FN_ASM(Algorithm::CN_DOUBLE); + +# ifdef XMRIG_ALGO_CN_GPU + m_map[Algorithm::CN_GPU][AV_SINGLE][Assembly::NONE] = cryptonight_single_hash_gpu; + m_map[Algorithm::CN_GPU][AV_SINGLE_SOFT][Assembly::NONE] = cryptonight_single_hash_gpu; +# endif + +# ifdef XMRIG_ALGO_CN_LITE + ADD_FN(Algorithm::CN_LITE_0); + ADD_FN(Algorithm::CN_LITE_1); +# endif + +# ifdef XMRIG_ALGO_CN_HEAVY + ADD_FN(Algorithm::CN_HEAVY_0); + ADD_FN(Algorithm::CN_HEAVY_TUBE); + ADD_FN(Algorithm::CN_HEAVY_XHV); +# endif + +# ifdef XMRIG_ALGO_CN_PICO + ADD_FN(Algorithm::CN_PICO_0); + ADD_FN_ASM(Algorithm::CN_PICO_0); +# endif + +# ifdef XMRIG_FEATURE_ASM + patchAsmVariants(); +# endif +} + + +xmrig::cn_hash_fun xmrig::CnHash::fn(const Algorithm &algorithm, AlgoVariant av, Assembly::Id assembly) +{ + if (!algorithm.isValid()) { + return nullptr; + } + +# ifdef XMRIG_FEATURE_ASM + cn_hash_fun fun = cnHash.m_map[algorithm][av][Cpu::assembly(assembly)]; + if (fun) { + return fun; + } +# endif + + return cnHash.m_map[algorithm][av][Assembly::NONE]; +} diff --git a/src/crypto/cn/CnHash.h b/src/crypto/cn/CnHash.h new file mode 100644 index 000000000..e4a7ebd22 --- /dev/null +++ b/src/crypto/cn/CnHash.h @@ -0,0 +1,78 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2019 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_CN_HASH_H +#define XMRIG_CN_HASH_H + + +#include +#include + + +#include "crypto/cn/CnAlgo.h" +#include "crypto/common/Assembly.h" + + +struct cryptonight_ctx; + + +namespace xmrig +{ + +typedef void (*cn_hash_fun)(const uint8_t *input, size_t size, uint8_t *output, cryptonight_ctx **ctx, uint64_t height); +typedef void (*cn_mainloop_fun)(cryptonight_ctx **ctx); + + +class CnHash +{ +public: + enum AlgoVariant { + AV_AUTO, // --av=0 Automatic mode. + AV_SINGLE, // --av=1 Single hash mode + AV_DOUBLE, // --av=2 Double hash mode + AV_SINGLE_SOFT, // --av=3 Single hash mode (Software AES) + AV_DOUBLE_SOFT, // --av=4 Double hash mode (Software AES) + AV_TRIPLE, // --av=5 Triple hash mode + AV_QUAD, // --av=6 Quard hash mode + AV_PENTA, // --av=7 Penta hash mode + AV_TRIPLE_SOFT, // --av=8 Triple hash mode (Software AES) + AV_QUAD_SOFT, // --av=9 Quard hash mode (Software AES) + AV_PENTA_SOFT, // --av=10 Penta hash mode (Software AES) + AV_MAX + }; + + CnHash(); + + static cn_hash_fun fn(const Algorithm &algorithm, AlgoVariant av, Assembly::Id assembly); + +private: + cn_hash_fun m_map[Algorithm::MAX][AV_MAX][Assembly::MAX] = {}; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_CN_HASH_H */ diff --git a/src/crypto/CryptoNight.h b/src/crypto/cn/CryptoNight.h similarity index 85% rename from src/crypto/CryptoNight.h rename to src/crypto/cn/CryptoNight.h index b1ec2371d..434c34f82 100644 --- a/src/crypto/CryptoNight.h +++ b/src/crypto/cn/CryptoNight.h @@ -6,6 +6,7 @@ * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , * Copyright 2018 Lee Clagett + * Copyright 2018-2019 SChernykh * Copyright 2016-2018 XMRig , * * This program is free software: you can redistribute it and/or modify @@ -30,32 +31,33 @@ #include #if defined _MSC_VER || defined XMRIG_ARM -#define ABI_ATTRIBUTE +# define ABI_ATTRIBUTE #else -#define ABI_ATTRIBUTE __attribute__((ms_abi)) +# define ABI_ATTRIBUTE __attribute__((ms_abi)) #endif + struct cryptonight_ctx; typedef void(*cn_mainloop_fun_ms_abi)(cryptonight_ctx**) ABI_ATTRIBUTE; + struct cryptonight_r_data { - int variant; + int algo; uint64_t height; - bool match(const int v, const uint64_t h) const { return (v == variant) && (h == height); } + bool match(const int a, const uint64_t h) const { return (a == algo) && (h == height); } }; + struct cryptonight_ctx { alignas(16) uint8_t state[224]; alignas(16) uint8_t *memory; uint8_t unused[40]; - const uint32_t* saes_table; + const uint32_t *saes_table; cn_mainloop_fun_ms_abi generated_code; - cn_mainloop_fun_ms_abi generated_code_double; cryptonight_r_data generated_code_data; - cryptonight_r_data generated_code_double_data; }; diff --git a/src/crypto/CryptoNight_arm.h b/src/crypto/cn/CryptoNight_arm.h similarity index 79% rename from src/crypto/CryptoNight_arm.h rename to src/crypto/cn/CryptoNight_arm.h index d762929c2..022666349 100644 --- a/src/crypto/CryptoNight_arm.h +++ b/src/crypto/cn/CryptoNight_arm.h @@ -28,20 +28,20 @@ #define XMRIG_CRYPTONIGHT_ARM_H -#include "common/crypto/keccak.h" -#include "common/utils/mm_malloc.h" -#include "crypto/CryptoNight.h" -#include "crypto/CryptoNight_constants.h" -#include "crypto/CryptoNight_monero.h" -#include "crypto/soft_aes.h" +#include "crypto/cn/CnAlgo.h" +#include "crypto/cn/CryptoNight_monero.h" +#include "crypto/cn/CryptoNight.h" +#include "crypto/cn/soft_aes.h" +#include "crypto/common/keccak.h" +#include "crypto/common/portable/mm_malloc.h" extern "C" { -#include "crypto/c_groestl.h" -#include "crypto/c_blake256.h" -#include "crypto/c_jh.h" -#include "crypto/c_skein.h" +#include "crypto/cn/c_groestl.h" +#include "crypto/cn/c_blake256.h" +#include "crypto/cn/c_jh.h" +#include "crypto/cn/c_skein.h" } @@ -226,9 +226,14 @@ inline void mix_and_propagate(__m128i& x0, __m128i& x1, __m128i& x2, __m128i& x3 } -template +namespace xmrig { + + +template static inline void cn_explode_scratchpad(const __m128i *input, __m128i *output) { + constexpr CnAlgo props; + __m128i xin0, xin1, xin2, xin3, xin4, xin5, xin6, xin7; __m128i k0, k1, k2, k3, k4, k5, k6, k7, k8, k9; @@ -243,7 +248,7 @@ static inline void cn_explode_scratchpad(const __m128i *input, __m128i *output) xin6 = _mm_load_si128(input + 10); xin7 = _mm_load_si128(input + 11); - if (ALGO == xmrig::CRYPTONIGHT_HEAVY) { + if (props.isHeavy()) { for (size_t i = 0; i < 16; i++) { aes_round(k0, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7); aes_round(k1, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7); @@ -260,7 +265,7 @@ static inline void cn_explode_scratchpad(const __m128i *input, __m128i *output) } } - for (size_t i = 0; i < MEM / sizeof(__m128i); i += 8) { + for (size_t i = 0; i < props.memory() / sizeof(__m128i); i += 8) { aes_round(k0, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7); aes_round(k1, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7); aes_round(k2, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7); @@ -284,37 +289,17 @@ static inline void cn_explode_scratchpad(const __m128i *input, __m128i *output) } -#ifndef XMRIG_NO_CN_GPU -template -void cn_explode_scratchpad_gpu(const uint8_t *input, uint8_t *output) -{ - constexpr size_t hash_size = 200; // 25x8 bytes - alignas(16) uint64_t hash[25]; - - for (uint64_t i = 0; i < MEM / 512; i++) - { - memcpy(hash, input, hash_size); - hash[0] ^= i; - - xmrig::keccakf(hash, 24); - memcpy(output, hash, 160); - output += 160; - - xmrig::keccakf(hash, 24); - memcpy(output, hash, 176); - output += 176; - - xmrig::keccakf(hash, 24); - memcpy(output, hash, 176); - output += 176; - } -} -#endif - - -template +template static inline void cn_implode_scratchpad(const __m128i *input, __m128i *output) { + constexpr CnAlgo props; + +# ifdef XMRIG_ALGO_CN_GPU + constexpr bool IS_HEAVY = props.isHeavy() || ALGO == Algorithm::CN_GPU; +# else + constexpr bool IS_HEAVY = props.isHeavy(); +# endif + __m128i xout0, xout1, xout2, xout3, xout4, xout5, xout6, xout7; __m128i k0, k1, k2, k3, k4, k5, k6, k7, k8, k9; @@ -329,8 +314,7 @@ static inline void cn_implode_scratchpad(const __m128i *input, __m128i *output) xout6 = _mm_load_si128(output + 10); xout7 = _mm_load_si128(output + 11); - for (size_t i = 0; i < MEM / sizeof(__m128i); i += 8) - { + for (size_t i = 0; i < props.memory() / sizeof(__m128i); i += 8) { xout0 = _mm_xor_si128(_mm_load_si128(input + i + 0), xout0); xout1 = _mm_xor_si128(_mm_load_si128(input + i + 1), xout1); xout2 = _mm_xor_si128(_mm_load_si128(input + i + 2), xout2); @@ -351,13 +335,13 @@ static inline void cn_implode_scratchpad(const __m128i *input, __m128i *output) aes_round(k8, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7); aes_round(k9, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7); - if (ALGO == xmrig::CRYPTONIGHT_HEAVY) { + if (IS_HEAVY) { mix_and_propagate(xout0, xout1, xout2, xout3, xout4, xout5, xout6, xout7); } } - if (ALGO == xmrig::CRYPTONIGHT_HEAVY) { - for (size_t i = 0; i < MEM / sizeof(__m128i); i += 8) { + if (IS_HEAVY) { + for (size_t i = 0; i < props.memory() / sizeof(__m128i); i += 8) { xout0 = _mm_xor_si128(_mm_load_si128(input + i + 0), xout0); xout1 = _mm_xor_si128(_mm_load_si128(input + i + 1), xout1); xout2 = _mm_xor_si128(_mm_load_si128(input + i + 2), xout2); @@ -408,6 +392,9 @@ static inline void cn_implode_scratchpad(const __m128i *input, __m128i *output) } +} /* namespace xmrig */ + + static inline __m128i aes_round_tweak_div(const __m128i &in, const __m128i &key) { alignas(16) uint32_t k[4]; @@ -430,13 +417,18 @@ static inline __m128i aes_round_tweak_div(const __m128i &in, const __m128i &key) } -template +namespace xmrig { + + +template static inline void cryptonight_monero_tweak(const uint8_t* l, uint64_t idx, __m128i ax0, __m128i bx0, __m128i bx1, __m128i& cx) { + constexpr CnAlgo props; + uint64_t* mem_out = (uint64_t*)&l[idx]; - if (BASE == xmrig::VARIANT_2) { - VARIANT2_SHUFFLE(l, idx, ax0, bx0, bx1, cx, (VARIANT == xmrig::VARIANT_RWZ ? 1 : 0)); + if (props.base() == Algorithm::CN_2) { + VARIANT2_SHUFFLE(l, idx, ax0, bx0, bx1, cx, (ALGO == Algorithm::CN_RWZ ? 1 : 0)); _mm_store_si128((__m128i *)mem_out, _mm_xor_si128(bx0, cx)); } else { __m128i tmp = _mm_xor_si128(bx0, cx); @@ -446,7 +438,7 @@ static inline void cryptonight_monero_tweak(const uint8_t* l, uint64_t idx, __m1 uint8_t x = vh >> 24; static const uint16_t table = 0x7531; - const uint8_t index = (((x >> (VARIANT == xmrig::VARIANT_XTL ? 4 : 3)) & 6) | (x & 1)) << 1; + const uint8_t index = (((x >> (3)) & 6) | (x & 1)) << 1; vh ^= ((table >> index) & 0x3) << 28; mem_out[1] = vh; @@ -454,24 +446,28 @@ static inline void cryptonight_monero_tweak(const uint8_t* l, uint64_t idx, __m1 } -template +template inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t size, uint8_t *__restrict__ output, cryptonight_ctx **__restrict__ ctx, uint64_t height) { - constexpr size_t MASK = xmrig::cn_select_mask(); - constexpr size_t ITERATIONS = xmrig::cn_select_iter(); - constexpr size_t MEM = xmrig::cn_select_memory(); - constexpr xmrig::Variant BASE = xmrig::cn_base_variant(); + constexpr CnAlgo props; + constexpr size_t MASK = props.mask(); + constexpr Algorithm::Id BASE = props.base(); - if (BASE == xmrig::VARIANT_1 && size < 43) { +# ifdef XMRIG_ALGO_CN_HEAVY + constexpr bool IS_CN_HEAVY_TUBE = ALGO == Algorithm::CN_HEAVY_TUBE; +# else + constexpr bool IS_CN_HEAVY_TUBE = false; +# endif + + if (BASE == Algorithm::CN_1 && size < 43) { memset(output, 0, 32); return; } - xmrig::keccak(input, size, ctx[0]->state); + keccak(input, size, ctx[0]->state); + cn_explode_scratchpad(reinterpret_cast(ctx[0]->state), reinterpret_cast<__m128i *>(ctx[0]->memory)); - cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory); - - const uint8_t* l0 = ctx[0]->memory; + uint8_t* l0 = ctx[0]->memory; uint64_t* h0 = reinterpret_cast(ctx[0]->state); VARIANT1_INIT(0); @@ -480,19 +476,19 @@ inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t si uint64_t al0 = h0[0] ^ h0[4]; uint64_t ah0 = h0[1] ^ h0[5]; - __m128i bx0 = _mm_set_epi64x(h0[3] ^ h0[7], h0[2] ^ h0[6]); - __m128i bx1 = _mm_set_epi64x(h0[9] ^ h0[11], h0[8] ^ h0[10]); + __m128i bx0 = _mm_set_epi64x(h0[3] ^ h0[7], h0[2] ^ h0[6]); + __m128i bx1 = _mm_set_epi64x(h0[9] ^ h0[11], h0[8] ^ h0[10]); uint64_t idx0 = al0; - for (size_t i = 0; i < ITERATIONS; i++) { + for (size_t i = 0; i < props.iterations(); i++) { __m128i cx; - if (VARIANT == xmrig::VARIANT_TUBE || !SOFT_AES) { - cx = _mm_load_si128((__m128i *) &l0[idx0 & MASK]); + if (IS_CN_HEAVY_TUBE || !SOFT_AES) { + cx = _mm_load_si128(reinterpret_cast(&l0[idx0 & MASK])); } const __m128i ax0 = _mm_set_epi64x(ah0, al0); - if (VARIANT == xmrig::VARIANT_TUBE) { + if (IS_CN_HEAVY_TUBE) { cx = aes_round_tweak_div(cx, ax0); } else if (SOFT_AES) { @@ -502,8 +498,8 @@ inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t si cx = _mm_aesenc_si128(cx, ax0); } - if (BASE == xmrig::VARIANT_1 || BASE == xmrig::VARIANT_2) { - cryptonight_monero_tweak(l0, idx0 & MASK, ax0, bx0, bx1, cx); + if (BASE == Algorithm::CN_1 || BASE == Algorithm::CN_2) { + cryptonight_monero_tweak(l0, idx0 & MASK, ax0, bx0, bx1, cx); } else { _mm_store_si128((__m128i *)&l0[idx0 & MASK], _mm_xor_si128(bx0, cx)); } @@ -514,10 +510,10 @@ inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t si cl = ((uint64_t*) &l0[idx0 & MASK])[0]; ch = ((uint64_t*) &l0[idx0 & MASK])[1]; - if (BASE == xmrig::VARIANT_2) { - if ((VARIANT == xmrig::VARIANT_WOW) || (VARIANT == xmrig::VARIANT_4)) { + if (BASE == Algorithm::CN_2) { + if (props.isR()) { VARIANT4_RANDOM_MATH(0, al0, ah0, cl, bx0, bx1); - if (VARIANT == xmrig::VARIANT_4) { + if (ALGO == Algorithm::CN_R) { al0 ^= r0[2] | ((uint64_t)(r0[3]) << 32); ah0 ^= r0[0] | ((uint64_t)(r0[1]) << 32); } @@ -528,11 +524,11 @@ inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t si lo = __umul128(idx0, cl, &hi); - if (BASE == xmrig::VARIANT_2) { - if (VARIANT == xmrig::VARIANT_4) { + if (BASE == Algorithm::CN_2) { + if (ALGO == Algorithm::CN_R) { VARIANT2_SHUFFLE(l0, idx0 & MASK, ax0, bx0, bx1, cx, 0); } else { - VARIANT2_SHUFFLE2(l0, idx0 & MASK, ax0, bx0, bx1, hi, lo, (VARIANT == xmrig::VARIANT_RWZ ? 1 : 0)); + VARIANT2_SHUFFLE2(l0, idx0 & MASK, ax0, bx0, bx1, hi, lo, (ALGO == Algorithm::CN_RWZ ? 1 : 0)); } } @@ -541,9 +537,9 @@ inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t si ((uint64_t*)&l0[idx0 & MASK])[0] = al0; - if (BASE == xmrig::VARIANT_1 && (VARIANT == xmrig::VARIANT_TUBE || VARIANT == xmrig::VARIANT_RTO)) { + if (IS_CN_HEAVY_TUBE || ALGO == Algorithm::CN_RTO) { ((uint64_t*)&l0[idx0 & MASK])[1] = ah0 ^ tweak1_2_0 ^ al0; - } else if (BASE == xmrig::VARIANT_1) { + } else if (BASE == Algorithm::CN_1) { ((uint64_t*)&l0[idx0 & MASK])[1] = ah0 ^ tweak1_2_0; } else { ((uint64_t*)&l0[idx0 & MASK])[1] = ah0; @@ -553,7 +549,8 @@ inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t si ah0 ^= ch; idx0 = al0; - if (ALGO == xmrig::CRYPTONIGHT_HEAVY) { +# ifdef XMRIG_ALGO_CN_HEAVY + if (props.isHeavy()) { const int64x2_t x = vld1q_s64(reinterpret_cast(&l0[idx0 & MASK])); const int64_t n = vgetq_lane_s64(x, 0); const int32_t d = vgetq_lane_s32(x, 2); @@ -561,77 +558,113 @@ inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t si ((int64_t*)&l0[idx0 & MASK])[0] = n ^ q; - if (VARIANT == xmrig::VARIANT_XHV) { + if (ALGO == Algorithm::CN_HEAVY_XHV) { idx0 = (~d) ^ q; } else { idx0 = d ^ q; } } +# endif - if (BASE == xmrig::VARIANT_2) { + if (BASE == Algorithm::CN_2) { bx1 = bx0; } bx0 = cx; } - cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state); - - xmrig::keccakf(h0, 24); + cn_implode_scratchpad(reinterpret_cast(ctx[0]->memory), reinterpret_cast<__m128i *>(ctx[0]->state)); + keccakf(h0, 24); extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output); } -#ifndef XMRIG_NO_CN_GPU +} /* namespace xmrig */ + + +#ifdef XMRIG_ALGO_CN_GPU template void cn_gpu_inner_arm(const uint8_t *spad, uint8_t *lpad); -template +namespace xmrig { + + +template +void cn_explode_scratchpad_gpu(const uint8_t *input, uint8_t *output) +{ + constexpr size_t hash_size = 200; // 25x8 bytes + alignas(16) uint64_t hash[25]; + + for (uint64_t i = 0; i < MEM / 512; i++) { + memcpy(hash, input, hash_size); + hash[0] ^= i; + + xmrig::keccakf(hash, 24); + memcpy(output, hash, 160); + output += 160; + + xmrig::keccakf(hash, 24); + memcpy(output, hash, 176); + output += 176; + + xmrig::keccakf(hash, 24); + memcpy(output, hash, 176); + output += 176; + } +} + + +template inline void cryptonight_single_hash_gpu(const uint8_t *__restrict__ input, size_t size, uint8_t *__restrict__ output, cryptonight_ctx **__restrict__ ctx, uint64_t height) { - constexpr size_t MASK = xmrig::CRYPTONIGHT_GPU_MASK; - constexpr size_t ITERATIONS = xmrig::cn_select_iter(); - constexpr size_t MEM = xmrig::cn_select_memory(); + constexpr CnAlgo props; - static_assert(MASK > 0 && ITERATIONS > 0 && MEM > 0, "unsupported algorithm/variant"); - - xmrig::keccak(input, size, ctx[0]->state); - cn_explode_scratchpad_gpu(ctx[0]->state, ctx[0]->memory); + keccak(input, size, ctx[0]->state); + cn_explode_scratchpad_gpu(ctx[0]->state, ctx[0]->memory); fesetround(FE_TONEAREST); - cn_gpu_inner_arm(ctx[0]->state, ctx[0]->memory); + cn_gpu_inner_arm(ctx[0]->state, ctx[0]->memory); - cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state); - - xmrig::keccakf((uint64_t*) ctx[0]->state, 24); + cn_implode_scratchpad(reinterpret_cast(ctx[0]->memory), reinterpret_cast<__m128i *>(ctx[0]->state)); + keccakf(reinterpret_cast(ctx[0]->state), 24); memcpy(output, ctx[0]->state, 32); } + +} /* namespace xmrig */ #endif -template +namespace xmrig { + + +template inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t size, uint8_t *__restrict__ output, struct cryptonight_ctx **__restrict__ ctx, uint64_t height) { - constexpr size_t MASK = xmrig::cn_select_mask(); - constexpr size_t ITERATIONS = xmrig::cn_select_iter(); - constexpr size_t MEM = xmrig::cn_select_memory(); - constexpr xmrig::Variant BASE = xmrig::cn_base_variant(); + constexpr CnAlgo props; + constexpr size_t MASK = props.mask(); + constexpr Algorithm::Id BASE = props.base(); - if (BASE == xmrig::VARIANT_1 && size < 43) { +# ifdef XMRIG_ALGO_CN_HEAVY + constexpr bool IS_CN_HEAVY_TUBE = ALGO == Algorithm::CN_HEAVY_TUBE; +# else + constexpr bool IS_CN_HEAVY_TUBE = false; +# endif + + if (BASE == Algorithm::CN_1 && size < 43) { memset(output, 0, 64); return; } - xmrig::keccak(input, size, ctx[0]->state); - xmrig::keccak(input + size, size, ctx[1]->state); + keccak(input, size, ctx[0]->state); + keccak(input + size, size, ctx[1]->state); - const uint8_t* l0 = ctx[0]->memory; - const uint8_t* l1 = ctx[1]->memory; - uint64_t* h0 = reinterpret_cast(ctx[0]->state); - uint64_t* h1 = reinterpret_cast(ctx[1]->state); + uint8_t *l0 = ctx[0]->memory; + uint8_t *l1 = ctx[1]->memory; + uint64_t *h0 = reinterpret_cast(ctx[0]->state); + uint64_t *h1 = reinterpret_cast(ctx[1]->state); VARIANT1_INIT(0); VARIANT1_INIT(1); @@ -640,8 +673,8 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si VARIANT4_RANDOM_MATH_INIT(0); VARIANT4_RANDOM_MATH_INIT(1); - cn_explode_scratchpad((__m128i*) h0, (__m128i*) l0); - cn_explode_scratchpad((__m128i*) h1, (__m128i*) l1); + cn_explode_scratchpad(reinterpret_cast(h0), reinterpret_cast<__m128i *>(l0)); + cn_explode_scratchpad(reinterpret_cast(h1), reinterpret_cast<__m128i *>(l1)); uint64_t al0 = h0[0] ^ h0[4]; uint64_t al1 = h1[0] ^ h1[4]; @@ -656,16 +689,16 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si uint64_t idx0 = al0; uint64_t idx1 = al1; - for (size_t i = 0; i < ITERATIONS; i++) { + for (size_t i = 0; i < props.iterations(); i++) { __m128i cx0, cx1; - if (VARIANT == xmrig::VARIANT_TUBE || !SOFT_AES) { + if (IS_CN_HEAVY_TUBE || !SOFT_AES) { cx0 = _mm_load_si128((__m128i *) &l0[idx0 & MASK]); cx1 = _mm_load_si128((__m128i *) &l1[idx1 & MASK]); } const __m128i ax0 = _mm_set_epi64x(ah0, al0); const __m128i ax1 = _mm_set_epi64x(ah1, al1); - if (VARIANT == xmrig::VARIANT_TUBE) { + if (IS_CN_HEAVY_TUBE) { cx0 = aes_round_tweak_div(cx0, ax0); cx1 = aes_round_tweak_div(cx1, ax1); } @@ -678,9 +711,9 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si cx1 = _mm_aesenc_si128(cx1, ax1); } - if (BASE == xmrig::VARIANT_1 || (BASE == xmrig::VARIANT_2)) { - cryptonight_monero_tweak(l0, idx0 & MASK, ax0, bx00, bx01, cx0); - cryptonight_monero_tweak(l1, idx1 & MASK, ax1, bx10, bx11, cx1); + if (BASE == Algorithm::CN_1 || BASE == Algorithm::CN_2) { + cryptonight_monero_tweak(l0, idx0 & MASK, ax0, bx00, bx01, cx0); + cryptonight_monero_tweak(l1, idx1 & MASK, ax1, bx10, bx11, cx1); } else { _mm_store_si128((__m128i *) &l0[idx0 & MASK], _mm_xor_si128(bx00, cx0)); _mm_store_si128((__m128i *) &l1[idx1 & MASK], _mm_xor_si128(bx10, cx1)); @@ -693,10 +726,10 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si cl = ((uint64_t*) &l0[idx0 & MASK])[0]; ch = ((uint64_t*) &l0[idx0 & MASK])[1]; - if (BASE == xmrig::VARIANT_2) { - if ((VARIANT == xmrig::VARIANT_WOW) || (VARIANT == xmrig::VARIANT_4)) { + if (BASE == Algorithm::CN_2) { + if (props.isR()) { VARIANT4_RANDOM_MATH(0, al0, ah0, cl, bx00, bx01); - if (VARIANT == xmrig::VARIANT_4) { + if (ALGO == Algorithm::CN_R) { al0 ^= r0[2] | ((uint64_t)(r0[3]) << 32); ah0 ^= r0[0] | ((uint64_t)(r0[1]) << 32); } @@ -707,11 +740,11 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si lo = __umul128(idx0, cl, &hi); - if (BASE == xmrig::VARIANT_2) { - if (VARIANT == xmrig::VARIANT_4) { + if (BASE == Algorithm::CN_2) { + if (ALGO == Algorithm::CN_R) { VARIANT2_SHUFFLE(l0, idx0 & MASK, ax0, bx00, bx01, cx0, 0); } else { - VARIANT2_SHUFFLE2(l0, idx0 & MASK, ax0, bx00, bx01, hi, lo, (VARIANT == xmrig::VARIANT_RWZ ? 1 : 0)); + VARIANT2_SHUFFLE2(l0, idx0 & MASK, ax0, bx00, bx01, hi, lo, (ALGO == Algorithm::CN_RWZ ? 1 : 0)); } } @@ -720,9 +753,9 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si ((uint64_t*)&l0[idx0 & MASK])[0] = al0; - if (BASE == xmrig::VARIANT_1 && (VARIANT == xmrig::VARIANT_TUBE || VARIANT == xmrig::VARIANT_RTO)) { + if (IS_CN_HEAVY_TUBE || ALGO == Algorithm::CN_RTO) { ((uint64_t*)&l0[idx0 & MASK])[1] = ah0 ^ tweak1_2_0 ^ al0; - } else if (BASE == xmrig::VARIANT_1) { + } else if (BASE == Algorithm::CN_1) { ((uint64_t*)&l0[idx0 & MASK])[1] = ah0 ^ tweak1_2_0; } else { ((uint64_t*)&l0[idx0 & MASK])[1] = ah0; @@ -732,7 +765,8 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si ah0 ^= ch; idx0 = al0; - if (ALGO == xmrig::CRYPTONIGHT_HEAVY) { +# ifdef XMRIG_ALGO_CN_HEAVY + if (props.isHeavy()) { const int64x2_t x = vld1q_s64(reinterpret_cast(&l0[idx0 & MASK])); const int64_t n = vgetq_lane_s64(x, 0); const int32_t d = vgetq_lane_s32(x, 2); @@ -740,21 +774,22 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si ((int64_t*)&l0[idx0 & MASK])[0] = n ^ q; - if (VARIANT == xmrig::VARIANT_XHV) { + if (ALGO == Algorithm::CN_HEAVY_XHV) { idx0 = (~d) ^ q; } else { idx0 = d ^ q; } } +# endif cl = ((uint64_t*) &l1[idx1 & MASK])[0]; ch = ((uint64_t*) &l1[idx1 & MASK])[1]; - if (BASE == xmrig::VARIANT_2) { - if ((VARIANT == xmrig::VARIANT_WOW) || (VARIANT == xmrig::VARIANT_4)) { + if (BASE == Algorithm::CN_2) { + if (props.isR()) { VARIANT4_RANDOM_MATH(1, al1, ah1, cl, bx10, bx11); - if (VARIANT == xmrig::VARIANT_4) { + if (ALGO == Algorithm::CN_R) { al1 ^= r1[2] | ((uint64_t)(r1[3]) << 32); ah1 ^= r1[0] | ((uint64_t)(r1[1]) << 32); } @@ -765,11 +800,11 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si lo = __umul128(idx1, cl, &hi); - if (BASE == xmrig::VARIANT_2) { - if (VARIANT == xmrig::VARIANT_4) { + if (BASE == Algorithm::CN_2) { + if (ALGO == Algorithm::CN_R) { VARIANT2_SHUFFLE(l1, idx1 & MASK, ax1, bx10, bx11, cx1, 0); } else { - VARIANT2_SHUFFLE2(l1, idx1 & MASK, ax1, bx10, bx11, hi, lo, (VARIANT == xmrig::VARIANT_RWZ ? 1 : 0)); + VARIANT2_SHUFFLE2(l1, idx1 & MASK, ax1, bx10, bx11, hi, lo, (ALGO == Algorithm::CN_RWZ ? 1 : 0)); } } @@ -778,9 +813,9 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si ((uint64_t*)&l1[idx1 & MASK])[0] = al1; - if (BASE == xmrig::VARIANT_1 && (VARIANT == xmrig::VARIANT_TUBE || VARIANT == xmrig::VARIANT_RTO)) { + if (IS_CN_HEAVY_TUBE || ALGO == Algorithm::CN_RTO) { ((uint64_t*)&l1[idx1 & MASK])[1] = ah1 ^ tweak1_2_1 ^ al1; - } else if (BASE == xmrig::VARIANT_1) { + } else if (BASE == Algorithm::CN_1) { ((uint64_t*)&l1[idx1 & MASK])[1] = ah1 ^ tweak1_2_1; } else { ((uint64_t*)&l1[idx1 & MASK])[1] = ah1; @@ -790,7 +825,8 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si ah1 ^= ch; idx1 = al1; - if (ALGO == xmrig::CRYPTONIGHT_HEAVY) { +# ifdef XMRIG_ALGO_CN_HEAVY + if (props.isHeavy()) { const int64x2_t x = vld1q_s64(reinterpret_cast(&l1[idx1 & MASK])); const int64_t n = vgetq_lane_s64(x, 0); const int32_t d = vgetq_lane_s32(x, 2); @@ -798,47 +834,54 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si ((int64_t*)&l1[idx1 & MASK])[0] = n ^ q; - if (VARIANT == xmrig::VARIANT_XHV) { + if (ALGO == Algorithm::CN_HEAVY_XHV) { idx1 = (~d) ^ q; } else { idx1 = d ^ q; } } - if (BASE == xmrig::VARIANT_2) { +# endif + + if (BASE == Algorithm::CN_2) { bx01 = bx00; bx11 = bx10; } + bx00 = cx0; bx10 = cx1; } - cn_implode_scratchpad((__m128i*) l0, (__m128i*) h0); - cn_implode_scratchpad((__m128i*) l1, (__m128i*) h1); + cn_implode_scratchpad(reinterpret_cast(l0), reinterpret_cast<__m128i *>(h0)); + cn_implode_scratchpad(reinterpret_cast(l1), reinterpret_cast<__m128i *>(h1)); - xmrig::keccakf(h0, 24); - xmrig::keccakf(h1, 24); + keccakf(h0, 24); + keccakf(h1, 24); extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output); extra_hashes[ctx[1]->state[0] & 3](ctx[1]->state, 200, output + 32); } -template +template inline void cryptonight_triple_hash(const uint8_t *__restrict__ input, size_t size, uint8_t *__restrict__ output, struct cryptonight_ctx **__restrict__ ctx, uint64_t height) { } -template +template inline void cryptonight_quad_hash(const uint8_t *__restrict__ input, size_t size, uint8_t *__restrict__ output, struct cryptonight_ctx **__restrict__ ctx, uint64_t height) { } -template +template inline void cryptonight_penta_hash(const uint8_t *__restrict__ input, size_t size, uint8_t *__restrict__ output, struct cryptonight_ctx **__restrict__ ctx, uint64_t height) { } -#endif /* __CRYPTONIGHT_ARM_H__ */ + +} /* namespace xmrig */ + + +#endif /* XMRIG_CRYPTONIGHT_ARM_H */ diff --git a/src/crypto/CryptoNight_monero.h b/src/crypto/cn/CryptoNight_monero.h similarity index 89% rename from src/crypto/CryptoNight_monero.h rename to src/crypto/cn/CryptoNight_monero.h index 4e84ac5d0..dc012bdc7 100644 --- a/src/crypto/CryptoNight_monero.h +++ b/src/crypto/cn/CryptoNight_monero.h @@ -33,21 +33,21 @@ #ifndef XMRIG_ARM # define VARIANT1_INIT(part) \ uint64_t tweak1_2_##part = 0; \ - if (BASE == xmrig::VARIANT_1) { \ + if (BASE == Algorithm::CN_1) { \ tweak1_2_##part = (*reinterpret_cast(input + 35 + part * size) ^ \ *(reinterpret_cast(ctx[part]->state) + 24)); \ } #else # define VARIANT1_INIT(part) \ uint64_t tweak1_2_##part = 0; \ - if (BASE == xmrig::VARIANT_1) { \ + if (BASE == Algorithm::CN_1) { \ memcpy(&tweak1_2_##part, input + 35 + part * size, sizeof tweak1_2_##part); \ tweak1_2_##part ^= *(reinterpret_cast(ctx[part]->state) + 24); \ } #endif #define VARIANT1_1(p) \ - if (BASE == xmrig::VARIANT_1) { \ + if (BASE == Algorithm::CN_1) { \ const uint8_t tmp = reinterpret_cast(p)[11]; \ static const uint32_t table = 0x75310; \ const uint8_t index = (((tmp >> 3) & 6) | (tmp & 1)) << 1; \ @@ -55,20 +55,20 @@ } #define VARIANT1_2(p, part) \ - if (BASE == xmrig::VARIANT_1) { \ + if (BASE == Algorithm::CN_1) { \ (p) ^= tweak1_2_##part; \ } #ifndef XMRIG_ARM # define VARIANT2_INIT(part) \ - __m128i division_result_xmm_##part = _mm_cvtsi64_si128(h##part[12]); \ - __m128i sqrt_result_xmm_##part = _mm_cvtsi64_si128(h##part[13]); + __m128i division_result_xmm_##part = _mm_cvtsi64_si128(static_cast(h##part[12])); \ + __m128i sqrt_result_xmm_##part = _mm_cvtsi64_si128(static_cast(h##part[13])); #ifdef _MSC_VER -# define VARIANT2_SET_ROUNDING_MODE() if (BASE == xmrig::VARIANT_2) { _control87(RC_DOWN, MCW_RC); } +# define VARIANT2_SET_ROUNDING_MODE() if (BASE == Algorithm::CN_2) { _control87(RC_DOWN, MCW_RC); } #else -# define VARIANT2_SET_ROUNDING_MODE() if (BASE == xmrig::VARIANT_2) { fesetround(FE_DOWNWARD); } +# define VARIANT2_SET_ROUNDING_MODE() if (BASE == Algorithm::CN_2) { fesetround(FE_DOWNWARD); } #endif # define VARIANT2_INTEGER_MATH(part, cl, cx) \ @@ -91,7 +91,7 @@ _mm_store_si128((__m128i *)((base_ptr) + ((offset) ^ 0x10)), _mm_add_epi64(chunk3, _b1)); \ _mm_store_si128((__m128i *)((base_ptr) + ((offset) ^ 0x20)), _mm_add_epi64(chunk1, _b)); \ _mm_store_si128((__m128i *)((base_ptr) + ((offset) ^ 0x30)), _mm_add_epi64(chunk2, _a)); \ - if (VARIANT == xmrig::VARIANT_4) { \ + if (ALGO == Algorithm::CN_R) { \ _c = _mm_xor_si128(_mm_xor_si128(_c, chunk3), _mm_xor_si128(chunk1, chunk2)); \ } \ } while (0) @@ -141,7 +141,7 @@ vst1q_u64((uint64_t*)((base_ptr) + ((offset) ^ 0x10)), vaddq_u64(chunk3, vreinterpretq_u64_u8(_b1))); \ vst1q_u64((uint64_t*)((base_ptr) + ((offset) ^ 0x20)), vaddq_u64(chunk1, vreinterpretq_u64_u8(_b))); \ vst1q_u64((uint64_t*)((base_ptr) + ((offset) ^ 0x30)), vaddq_u64(chunk2, vreinterpretq_u64_u8(_a))); \ - if (VARIANT == xmrig::VARIANT_4) { \ + if (ALGO == Algorithm::CN_R) { \ _c = veorq_u64(veorq_u64(_c, chunk3), veorq_u64(chunk1, chunk2)); \ } \ } while (0) @@ -178,23 +178,22 @@ #endif #endif -#include "common/xmrig.h" -#include "variant4_random_math.h" +#include "crypto/cn/r/variant4_random_math.h" #define VARIANT4_RANDOM_MATH_INIT(part) \ uint32_t r##part[9]; \ struct V4_Instruction code##part[256]; \ - if ((VARIANT == xmrig::VARIANT_WOW) || (VARIANT == xmrig::VARIANT_4)) { \ - r##part[0] = (uint32_t)(h##part[12]); \ - r##part[1] = (uint32_t)(h##part[12] >> 32); \ - r##part[2] = (uint32_t)(h##part[13]); \ - r##part[3] = (uint32_t)(h##part[13] >> 32); \ + if (props.isR()) { \ + r##part[0] = static_cast(h##part[12]); \ + r##part[1] = static_cast(h##part[12] >> 32); \ + r##part[2] = static_cast(h##part[13]); \ + r##part[3] = static_cast(h##part[13] >> 32); \ } \ - v4_random_math_init(code##part, height); + v4_random_math_init(code##part, height); #define VARIANT4_RANDOM_MATH(part, al, ah, cl, bx0, bx1) \ - if ((VARIANT == xmrig::VARIANT_WOW) || (VARIANT == xmrig::VARIANT_4)) { \ - cl ^= (r##part[0] + r##part[1]) | ((uint64_t)(r##part[2] + r##part[3]) << 32); \ + if (props.isR()) { \ + cl ^= (r##part[0] + r##part[1]) | (static_cast(r##part[2] + r##part[3]) << 32); \ r##part[4] = static_cast(al); \ r##part[5] = static_cast(ah); \ r##part[6] = static_cast(_mm_cvtsi128_si32(bx0)); \ diff --git a/src/crypto/CryptoNight_test.h b/src/crypto/cn/CryptoNight_test.h similarity index 96% rename from src/crypto/CryptoNight_test.h rename to src/crypto/cn/CryptoNight_test.h index 6fa9dd286..77cbbb8e1 100644 --- a/src/crypto/CryptoNight_test.h +++ b/src/crypto/cn/CryptoNight_test.h @@ -156,21 +156,6 @@ const static uint8_t test_output_v2[160] = { }; -// "cn/xtl" Stellite (XTL) -const static uint8_t test_output_xtl[160] = { - 0x8F, 0xE5, 0xF0, 0x5F, 0x02, 0x2A, 0x61, 0x7D, 0xE5, 0x3F, 0x79, 0x36, 0x4B, 0x25, 0xCB, 0xC3, - 0xC0, 0x8E, 0x0E, 0x1F, 0xE3, 0xBE, 0x48, 0x57, 0x07, 0x03, 0xFE, 0xE1, 0xEC, 0x0E, 0xB0, 0xB1, - 0x21, 0x26, 0xFF, 0x98, 0xE6, 0x86, 0x08, 0x5B, 0xC9, 0x96, 0x44, 0xA3, 0xB8, 0x4E, 0x28, 0x90, - 0x76, 0xED, 0xAD, 0xB9, 0xAA, 0xAC, 0x01, 0x94, 0x1D, 0xBE, 0x3E, 0xEA, 0xAD, 0xEE, 0xB2, 0xCF, - 0xB0, 0x43, 0x4B, 0x88, 0xFC, 0xB2, 0xF3, 0x82, 0x9D, 0xD7, 0xDF, 0x51, 0x97, 0x2C, 0x5A, 0xE3, - 0xC7, 0x16, 0x0B, 0xC8, 0x7C, 0xB7, 0x2F, 0x1C, 0x55, 0x33, 0xCA, 0xE1, 0xEE, 0x08, 0xA4, 0x86, - 0x60, 0xED, 0x6E, 0x9D, 0x2D, 0x05, 0x0D, 0x7D, 0x02, 0x49, 0x23, 0x39, 0x7C, 0xC3, 0x6D, 0x3D, - 0x05, 0x51, 0x28, 0xF1, 0x9B, 0x3C, 0xDF, 0xC4, 0xEA, 0x8A, 0xA6, 0x6A, 0x3C, 0x8B, 0xE2, 0xAF, - 0x47, 0x00, 0xFC, 0x36, 0xED, 0x50, 0xBB, 0xD2, 0x2E, 0x63, 0x4B, 0x93, 0x11, 0x0C, 0xA7, 0xBA, - 0x32, 0x6E, 0x47, 0x4D, 0xCE, 0xCC, 0x82, 0x54, 0x1D, 0x06, 0xF8, 0x06, 0x86, 0xBD, 0x22, 0x48 -}; - - // "cn/half" const static uint8_t test_output_half[160] = { 0x5D, 0x4F, 0xBC, 0x35, 0x60, 0x97, 0xEA, 0x64, 0x40, 0xB0, 0x88, 0x8E, 0xDE, 0xB6, 0x35, 0xDD, @@ -272,7 +257,7 @@ const static uint8_t test_output_double[160] = { 0x5E, 0x2E, 0xC1, 0x80, 0x89, 0x39, 0xB3, 0x54, 0x39, 0x52, 0x0E, 0x69, 0x3D, 0xF6, 0xC5, 0x4A }; -#ifndef XMRIG_NO_AEON +#ifdef XMRIG_ALGO_CN_LITE // "cn-lite/0" const static uint8_t test_output_v0_lite[160] = { 0x36, 0x95, 0xB4, 0xB5, 0x3B, 0xB0, 0x03, 0x58, 0xB0, 0xAD, 0x38, 0xDC, 0x16, 0x0F, 0xEB, 0x9E, @@ -304,7 +289,7 @@ const static uint8_t test_output_v1_lite[160] = { #endif -#ifndef XMRIG_NO_SUMO +#ifdef XMRIG_ALGO_CN_HEAVY // "cn-heavy/0" const static uint8_t test_output_v0_heavy[160] = { 0x99, 0x83, 0xF2, 0x1B, 0xDF, 0x20, 0x10, 0xA8, 0xD7, 0x07, 0xBB, 0x2F, 0x14, 0xD7, 0x86, 0x64, @@ -351,7 +336,7 @@ const static uint8_t test_output_tube_heavy[160] = { #endif -#ifndef XMRIG_NO_CN_PICO +#ifdef XMRIG_ALGO_CN_PICO // "cn-pico/trtl" const static uint8_t test_output_pico_trtl[160] = { 0x08, 0xF4, 0x21, 0xD7, 0x83, 0x31, 0x17, 0x30, 0x0E, 0xDA, 0x66, 0xE9, 0x8F, 0x4A, 0x25, 0x69, @@ -368,7 +353,7 @@ const static uint8_t test_output_pico_trtl[160] = { #endif -#ifndef XMRIG_NO_CN_GPU +#ifdef XMRIG_ALGO_CN_GPU // "cn/gpu" const static uint8_t test_output_gpu[160] = { 0xE5, 0x5C, 0xB2, 0x3E, 0x51, 0x64, 0x9A, 0x59, 0xB1, 0x27, 0xB9, 0x6B, 0x51, 0x5F, 0x2B, 0xF7, diff --git a/src/crypto/CryptoNight_x86.h b/src/crypto/cn/CryptoNight_x86.h similarity index 63% rename from src/crypto/CryptoNight_x86.h rename to src/crypto/cn/CryptoNight_x86.h index 202b662a2..ae51cd182 100644 --- a/src/crypto/CryptoNight_x86.h +++ b/src/crypto/cn/CryptoNight_x86.h @@ -35,20 +35,20 @@ #endif -#include "common/cpu/Cpu.h" -#include "common/crypto/keccak.h" -#include "crypto/CryptoNight.h" -#include "crypto/CryptoNight_constants.h" -#include "crypto/CryptoNight_monero.h" -#include "crypto/soft_aes.h" +#include "backend/cpu/Cpu.h" +#include "crypto/cn/CnAlgo.h" +#include "crypto/cn/CryptoNight_monero.h" +#include "crypto/cn/CryptoNight.h" +#include "crypto/cn/soft_aes.h" +#include "crypto/common/keccak.h" extern "C" { -#include "crypto/c_groestl.h" -#include "crypto/c_blake256.h" -#include "crypto/c_jh.h" -#include "crypto/c_skein.h" +#include "crypto/cn/c_groestl.h" +#include "crypto/cn/c_blake256.h" +#include "crypto/cn/c_jh.h" +#include "crypto/cn/c_skein.h" } @@ -303,9 +303,14 @@ inline void mix_and_propagate(__m128i& x0, __m128i& x1, __m128i& x2, __m128i& x3 } -template +namespace xmrig { + + +template static inline void cn_explode_scratchpad(const __m128i *input, __m128i *output) { + constexpr CnAlgo props; + __m128i xin0, xin1, xin2, xin3, xin4, xin5, xin6, xin7; __m128i k0, k1, k2, k3, k4, k5, k6, k7, k8, k9; @@ -320,7 +325,7 @@ static inline void cn_explode_scratchpad(const __m128i *input, __m128i *output) xin6 = _mm_load_si128(input + 10); xin7 = _mm_load_si128(input + 11); - if (ALGO == xmrig::CRYPTONIGHT_HEAVY) { + if (props.isHeavy()) { for (size_t i = 0; i < 16; i++) { aes_round(k0, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7); aes_round(k1, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7); @@ -337,7 +342,7 @@ static inline void cn_explode_scratchpad(const __m128i *input, __m128i *output) } } - for (size_t i = 0; i < MEM / sizeof(__m128i); i += 8) { + for (size_t i = 0; i < props.memory() / sizeof(__m128i); i += 8) { aes_round(k0, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7); aes_round(k1, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7); aes_round(k2, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7); @@ -361,37 +366,17 @@ static inline void cn_explode_scratchpad(const __m128i *input, __m128i *output) } -#ifndef XMRIG_NO_CN_GPU -template -void cn_explode_scratchpad_gpu(const uint8_t *input, uint8_t *output) -{ - constexpr size_t hash_size = 200; // 25x8 bytes - alignas(16) uint64_t hash[25]; - - for (uint64_t i = 0; i < MEM / 512; i++) - { - memcpy(hash, input, hash_size); - hash[0] ^= i; - - xmrig::keccakf(hash, 24); - memcpy(output, hash, 160); - output += 160; - - xmrig::keccakf(hash, 24); - memcpy(output, hash, 176); - output += 176; - - xmrig::keccakf(hash, 24); - memcpy(output, hash, 176); - output += 176; - } -} -#endif - - -template +template static inline void cn_implode_scratchpad(const __m128i *input, __m128i *output) { + constexpr CnAlgo props; + +# ifdef XMRIG_ALGO_CN_GPU + constexpr bool IS_HEAVY = props.isHeavy() || ALGO == Algorithm::CN_GPU; +# else + constexpr bool IS_HEAVY = props.isHeavy(); +# endif + __m128i xout0, xout1, xout2, xout3, xout4, xout5, xout6, xout7; __m128i k0, k1, k2, k3, k4, k5, k6, k7, k8, k9; @@ -406,8 +391,7 @@ static inline void cn_implode_scratchpad(const __m128i *input, __m128i *output) xout6 = _mm_load_si128(output + 10); xout7 = _mm_load_si128(output + 11); - for (size_t i = 0; i < MEM / sizeof(__m128i); i += 8) - { + for (size_t i = 0; i < props.memory() / sizeof(__m128i); i += 8) { xout0 = _mm_xor_si128(_mm_load_si128(input + i + 0), xout0); xout1 = _mm_xor_si128(_mm_load_si128(input + i + 1), xout1); xout2 = _mm_xor_si128(_mm_load_si128(input + i + 2), xout2); @@ -428,13 +412,13 @@ static inline void cn_implode_scratchpad(const __m128i *input, __m128i *output) aes_round(k8, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7); aes_round(k9, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7); - if (ALGO == xmrig::CRYPTONIGHT_HEAVY) { + if (IS_HEAVY) { mix_and_propagate(xout0, xout1, xout2, xout3, xout4, xout5, xout6, xout7); } } - if (ALGO == xmrig::CRYPTONIGHT_HEAVY) { - for (size_t i = 0; i < MEM / sizeof(__m128i); i += 8) { + if (IS_HEAVY) { + for (size_t i = 0; i < props.memory() / sizeof(__m128i); i += 8) { xout0 = _mm_xor_si128(_mm_load_si128(input + i + 0), xout0); xout1 = _mm_xor_si128(_mm_load_si128(input + i + 1), xout1); xout2 = _mm_xor_si128(_mm_load_si128(input + i + 2), xout2); @@ -485,6 +469,9 @@ static inline void cn_implode_scratchpad(const __m128i *input, __m128i *output) } +} /* namespace xmrig */ + + static inline __m128i aes_round_tweak_div(const __m128i &in, const __m128i &key) { alignas(16) uint32_t k[4]; @@ -527,12 +514,21 @@ static inline __m128i int_sqrt_v2(const uint64_t n0) } -template -static inline void cryptonight_monero_tweak(uint64_t* mem_out, const uint8_t* l, uint64_t idx, __m128i ax0, __m128i bx0, __m128i bx1, __m128i& cx) +void wow_soft_aes_compile_code(const V4_Instruction *code, int code_size, void *machine_code, xmrig::Assembly ASM); +void v4_soft_aes_compile_code(const V4_Instruction *code, int code_size, void *machine_code, xmrig::Assembly ASM); + + +namespace xmrig { + + +template +static inline void cryptonight_monero_tweak(uint64_t *mem_out, const uint8_t *l, uint64_t idx, __m128i ax0, __m128i bx0, __m128i bx1, __m128i& cx) { - if (BASE == xmrig::VARIANT_2) { - VARIANT2_SHUFFLE(l, idx, ax0, bx0, bx1, cx, (VARIANT == xmrig::VARIANT_RWZ ? 1 : 0)); - _mm_store_si128((__m128i *)mem_out, _mm_xor_si128(bx0, cx)); + constexpr CnAlgo props; + + if (props.base() == Algorithm::CN_2) { + VARIANT2_SHUFFLE(l, idx, ax0, bx0, bx1, cx, (ALGO == Algorithm::CN_RWZ ? 1 : 0)); + _mm_store_si128(reinterpret_cast<__m128i *>(mem_out), _mm_xor_si128(bx0, cx)); } else { __m128i tmp = _mm_xor_si128(bx0, cx); mem_out[0] = _mm_cvtsi128_si64(tmp); @@ -542,107 +538,105 @@ static inline void cryptonight_monero_tweak(uint64_t* mem_out, const uint8_t* l, uint8_t x = static_cast(vh >> 24); static const uint16_t table = 0x7531; - const uint8_t index = (((x >> (VARIANT == xmrig::VARIANT_XTL ? 4 : 3)) & 6) | (x & 1)) << 1; + const uint8_t index = (((x >> (3)) & 6) | (x & 1)) << 1; vh ^= ((table >> index) & 0x3) << 28; mem_out[1] = vh; } } -void wow_soft_aes_compile_code(const V4_Instruction* code, int code_size, void* machine_code, xmrig::Assembly ASM); -void v4_soft_aes_compile_code(const V4_Instruction* code, int code_size, void* machine_code, xmrig::Assembly ASM); -template +template inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t size, uint8_t *__restrict__ output, cryptonight_ctx **__restrict__ ctx, uint64_t height) { - constexpr size_t MASK = xmrig::cn_select_mask(); - constexpr size_t ITERATIONS = xmrig::cn_select_iter(); - constexpr size_t MEM = xmrig::cn_select_memory(); - constexpr xmrig::Variant BASE = xmrig::cn_base_variant(); + constexpr CnAlgo props; + constexpr size_t MASK = props.mask(); + constexpr Algorithm::Id BASE = props.base(); - static_assert(MASK > 0 && ITERATIONS > 0 && MEM > 0, "unsupported algorithm/variant"); +# ifdef XMRIG_ALGO_CN_HEAVY + constexpr bool IS_CN_HEAVY_TUBE = ALGO == Algorithm::CN_HEAVY_TUBE; +# else + constexpr bool IS_CN_HEAVY_TUBE = false; +# endif - if (BASE == xmrig::VARIANT_1 && size < 43) { + if (BASE == Algorithm::CN_1 && size < 43) { memset(output, 0, 32); return; } - xmrig::keccak(input, size, ctx[0]->state); + keccak(input, size, ctx[0]->state); + cn_explode_scratchpad(reinterpret_cast(ctx[0]->state), reinterpret_cast<__m128i *>(ctx[0]->memory)); - cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory); + uint64_t *h0 = reinterpret_cast(ctx[0]->state); + uint8_t *l0 = ctx[0]->memory; - uint64_t* h0 = reinterpret_cast(ctx[0]->state); - -#ifndef XMRIG_NO_ASM - if (SOFT_AES && xmrig::cn_is_cryptonight_r()) - { - if (!ctx[0]->generated_code_data.match(VARIANT, height)) { +# ifdef XMRIG_FEATURE_ASM + if (SOFT_AES && props.isR()) { + if (!ctx[0]->generated_code_data.match(ALGO, height)) { V4_Instruction code[256]; - const int code_size = v4_random_math_init(code, height); + const int code_size = v4_random_math_init(code, height); - if (VARIANT == xmrig::VARIANT_WOW) - wow_soft_aes_compile_code(code, code_size, reinterpret_cast(ctx[0]->generated_code), xmrig::ASM_NONE); - else if (VARIANT == xmrig::VARIANT_4) - v4_soft_aes_compile_code(code, code_size, reinterpret_cast(ctx[0]->generated_code), xmrig::ASM_NONE); + if (ALGO == Algorithm::CN_WOW) { + wow_soft_aes_compile_code(code, code_size, reinterpret_cast(ctx[0]->generated_code), Assembly::NONE); + } + else if (ALGO == Algorithm::CN_R) { + v4_soft_aes_compile_code(code, code_size, reinterpret_cast(ctx[0]->generated_code), Assembly::NONE); + } - ctx[0]->generated_code_data.variant = VARIANT; - ctx[0]->generated_code_data.height = height; + ctx[0]->generated_code_data = { ALGO, height }; } - ctx[0]->saes_table = (const uint32_t*)saes_table; + ctx[0]->saes_table = reinterpret_cast(saes_table); ctx[0]->generated_code(ctx); } else { -#endif - - const uint8_t* l0 = ctx[0]->memory; +# endif VARIANT1_INIT(0); VARIANT2_INIT(0); VARIANT2_SET_ROUNDING_MODE(); VARIANT4_RANDOM_MATH_INIT(0); - uint64_t al0 = h0[0] ^ h0[4]; - uint64_t ah0 = h0[1] ^ h0[5]; - __m128i bx0 = _mm_set_epi64x(h0[3] ^ h0[7], h0[2] ^ h0[6]); - __m128i bx1 = _mm_set_epi64x(h0[9] ^ h0[11], h0[8] ^ h0[10]); - + uint64_t al0 = h0[0] ^ h0[4]; + uint64_t ah0 = h0[1] ^ h0[5]; uint64_t idx0 = al0; + __m128i bx0 = _mm_set_epi64x(static_cast(h0[3] ^ h0[7]), static_cast(h0[2] ^ h0[6])); + __m128i bx1 = _mm_set_epi64x(static_cast(h0[9] ^ h0[11]), static_cast(h0[8] ^ h0[10])); - for (size_t i = 0; i < ITERATIONS; i++) { + for (size_t i = 0; i < props.iterations(); i++) { __m128i cx; - if (VARIANT == xmrig::VARIANT_TUBE || !SOFT_AES) { - cx = _mm_load_si128((__m128i *) &l0[idx0 & MASK]); + if (IS_CN_HEAVY_TUBE || !SOFT_AES) { + cx = _mm_load_si128(reinterpret_cast(&l0[idx0 & MASK])); } - const __m128i ax0 = _mm_set_epi64x(ah0, al0); - if (VARIANT == xmrig::VARIANT_TUBE) { + const __m128i ax0 = _mm_set_epi64x(static_cast(ah0), static_cast(al0)); + if (IS_CN_HEAVY_TUBE) { cx = aes_round_tweak_div(cx, ax0); } else if (SOFT_AES) { - cx = soft_aesenc((uint32_t*)&l0[idx0 & MASK], ax0, (const uint32_t*)saes_table); + cx = soft_aesenc(&l0[idx0 & MASK], ax0, reinterpret_cast(saes_table)); } else { cx = _mm_aesenc_si128(cx, ax0); } - if (BASE == xmrig::VARIANT_1 || BASE == xmrig::VARIANT_2) { - cryptonight_monero_tweak((uint64_t*)&l0[idx0 & MASK], l0, idx0 & MASK, ax0, bx0, bx1, cx); + if (BASE == Algorithm::CN_1 || BASE == Algorithm::CN_2) { + cryptonight_monero_tweak(reinterpret_cast(&l0[idx0 & MASK]), l0, idx0 & MASK, ax0, bx0, bx1, cx); } else { - _mm_store_si128((__m128i *)&l0[idx0 & MASK], _mm_xor_si128(bx0, cx)); + _mm_store_si128(reinterpret_cast<__m128i *>(&l0[idx0 & MASK]), _mm_xor_si128(bx0, cx)); } - idx0 = _mm_cvtsi128_si64(cx); + idx0 = static_cast(_mm_cvtsi128_si64(cx)); uint64_t hi, lo, cl, ch; - cl = ((uint64_t*) &l0[idx0 & MASK])[0]; - ch = ((uint64_t*) &l0[idx0 & MASK])[1]; + cl = (reinterpret_cast(&l0[idx0 & MASK]))[0]; + ch = (reinterpret_cast(&l0[idx0 & MASK]))[1]; - if (BASE == xmrig::VARIANT_2) { - if ((VARIANT == xmrig::VARIANT_WOW) || (VARIANT == xmrig::VARIANT_4)) { + if (BASE == Algorithm::CN_2) { + if (props.isR()) { VARIANT4_RANDOM_MATH(0, al0, ah0, cl, bx0, bx1); - if (VARIANT == xmrig::VARIANT_4) { - al0 ^= r0[2] | ((uint64_t)(r0[3]) << 32); - ah0 ^= r0[0] | ((uint64_t)(r0[1]) << 32); + if (ALGO == Algorithm::CN_R) { + al0 ^= r0[2] | (static_cast(r0[3]) << 32); + ah0 ^= r0[0] | (static_cast(r0[1]) << 32); } } else { VARIANT2_INTEGER_MATH(0, cl, cx); @@ -651,64 +645,68 @@ inline void cryptonight_single_hash(const uint8_t *__restrict__ input, size_t si lo = __umul128(idx0, cl, &hi); - if (BASE == xmrig::VARIANT_2) { - if (VARIANT == xmrig::VARIANT_4) { + if (BASE == Algorithm::CN_2) { + if (ALGO == Algorithm::CN_R) { VARIANT2_SHUFFLE(l0, idx0 & MASK, ax0, bx0, bx1, cx, 0); } else { - VARIANT2_SHUFFLE2(l0, idx0 & MASK, ax0, bx0, bx1, hi, lo, (VARIANT == xmrig::VARIANT_RWZ ? 1 : 0)); + VARIANT2_SHUFFLE2(l0, idx0 & MASK, ax0, bx0, bx1, hi, lo, (ALGO == Algorithm::CN_RWZ ? 1 : 0)); } } al0 += hi; ah0 += lo; - ((uint64_t*)&l0[idx0 & MASK])[0] = al0; + reinterpret_cast(&l0[idx0 & MASK])[0] = al0; - if (BASE == xmrig::VARIANT_1 && (VARIANT == xmrig::VARIANT_TUBE || VARIANT == xmrig::VARIANT_RTO)) { - ((uint64_t*)&l0[idx0 & MASK])[1] = ah0 ^ tweak1_2_0 ^ al0; - } else if (BASE == xmrig::VARIANT_1) { - ((uint64_t*)&l0[idx0 & MASK])[1] = ah0 ^ tweak1_2_0; + if (IS_CN_HEAVY_TUBE || ALGO == Algorithm::CN_RTO) { + reinterpret_cast(&l0[idx0 & MASK])[1] = ah0 ^ tweak1_2_0 ^ al0; + } else if (BASE == Algorithm::CN_1) { + reinterpret_cast(&l0[idx0 & MASK])[1] = ah0 ^ tweak1_2_0; } else { - ((uint64_t*)&l0[idx0 & MASK])[1] = ah0; + reinterpret_cast(&l0[idx0 & MASK])[1] = ah0; } al0 ^= cl; ah0 ^= ch; idx0 = al0; - if (ALGO == xmrig::CRYPTONIGHT_HEAVY) { +# ifdef XMRIG_ALGO_CN_HEAVY + if (props.isHeavy()) { int64_t n = ((int64_t*)&l0[idx0 & MASK])[0]; int32_t d = ((int32_t*)&l0[idx0 & MASK])[2]; int64_t q = n / (d | 0x5); ((int64_t*)&l0[idx0 & MASK])[0] = n ^ q; - if (VARIANT == xmrig::VARIANT_XHV) { + if (ALGO == Algorithm::CN_HEAVY_XHV) { d = ~d; } idx0 = d ^ q; } +# endif - if (BASE == xmrig::VARIANT_2) { + if (BASE == Algorithm::CN_2) { bx1 = bx0; } bx0 = cx; } -#ifndef XMRIG_NO_ASM +# ifdef XMRIG_FEATURE_ASM } -#endif - - cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state); +# endif - xmrig::keccakf(h0, 24); + cn_implode_scratchpad(reinterpret_cast(ctx[0]->memory), reinterpret_cast<__m128i *>(ctx[0]->state)); + keccakf(h0, 24); extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output); } -#ifndef XMRIG_NO_CN_GPU +} /* namespace xmrig */ + + +#ifdef XMRIG_ALGO_CN_GPU template void cn_gpu_inner_avx(const uint8_t *spad, uint8_t *lpad); @@ -717,17 +715,41 @@ template void cn_gpu_inner_ssse3(const uint8_t *spad, uint8_t *lpad); -template -inline void cryptonight_single_hash_gpu(const uint8_t *__restrict__ input, size_t size, uint8_t *__restrict__ output, cryptonight_ctx **__restrict__ ctx, uint64_t height) +namespace xmrig { + + +template +void cn_explode_scratchpad_gpu(const uint8_t *input, uint8_t *output) { - constexpr size_t MASK = xmrig::CRYPTONIGHT_GPU_MASK; - constexpr size_t ITERATIONS = xmrig::cn_select_iter(); - constexpr size_t MEM = xmrig::cn_select_memory(); + constexpr size_t hash_size = 200; // 25x8 bytes + alignas(16) uint64_t hash[25]; - static_assert(MASK > 0 && ITERATIONS > 0 && MEM > 0, "unsupported algorithm/variant"); + for (uint64_t i = 0; i < MEM / 512; i++) { + memcpy(hash, input, hash_size); + hash[0] ^= i; - xmrig::keccak(input, size, ctx[0]->state); - cn_explode_scratchpad_gpu(ctx[0]->state, ctx[0]->memory); + xmrig::keccakf(hash, 24); + memcpy(output, hash, 160); + output += 160; + + xmrig::keccakf(hash, 24); + memcpy(output, hash, 176); + output += 176; + + xmrig::keccakf(hash, 24); + memcpy(output, hash, 176); + output += 176; + } +} + + +template +inline void cryptonight_single_hash_gpu(const uint8_t *__restrict__ input, size_t size, uint8_t *__restrict__ output, cryptonight_ctx **__restrict__ ctx, uint64_t) +{ + constexpr CnAlgo props; + + keccak(input, size, ctx[0]->state); + cn_explode_scratchpad_gpu(ctx[0]->state, ctx[0]->memory); # ifdef _MSC_VER _control87(RC_NEAR, MCW_RC); @@ -736,20 +758,22 @@ inline void cryptonight_single_hash_gpu(const uint8_t *__restrict__ input, size_ # endif if (xmrig::Cpu::info()->hasAVX2()) { - cn_gpu_inner_avx(ctx[0]->state, ctx[0]->memory); + cn_gpu_inner_avx(ctx[0]->state, ctx[0]->memory); } else { - cn_gpu_inner_ssse3(ctx[0]->state, ctx[0]->memory); + cn_gpu_inner_ssse3(ctx[0]->state, ctx[0]->memory); } - cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state); - - xmrig::keccakf((uint64_t*) ctx[0]->state, 24); + cn_implode_scratchpad(reinterpret_cast(ctx[0]->memory), reinterpret_cast<__m128i *>(ctx[0]->state)); + keccakf(reinterpret_cast(ctx[0]->state), 24); memcpy(output, ctx[0]->state, 32); } + + +} /* namespace xmrig */ #endif -#ifndef XMRIG_NO_ASM +#ifdef XMRIG_FEATURE_ASM extern "C" void cnv2_mainloop_ivybridge_asm(cryptonight_ctx **ctx); extern "C" void cnv2_mainloop_ryzen_asm(cryptonight_ctx **ctx); extern "C" void cnv2_mainloop_bulldozer_asm(cryptonight_ctx **ctx); @@ -757,212 +781,246 @@ extern "C" void cnv2_double_mainloop_sandybridge_asm(cryptonight_ctx **ctx); extern "C" void cnv2_rwz_mainloop_asm(cryptonight_ctx **ctx); extern "C" void cnv2_rwz_double_mainloop_asm(cryptonight_ctx **ctx); -extern xmrig::CpuThread::cn_mainloop_fun cn_half_mainloop_ivybridge_asm; -extern xmrig::CpuThread::cn_mainloop_fun cn_half_mainloop_ryzen_asm; -extern xmrig::CpuThread::cn_mainloop_fun cn_half_mainloop_bulldozer_asm; -extern xmrig::CpuThread::cn_mainloop_fun cn_half_double_mainloop_sandybridge_asm; -extern xmrig::CpuThread::cn_mainloop_fun cn_trtl_mainloop_ivybridge_asm; -extern xmrig::CpuThread::cn_mainloop_fun cn_trtl_mainloop_ryzen_asm; -extern xmrig::CpuThread::cn_mainloop_fun cn_trtl_mainloop_bulldozer_asm; -extern xmrig::CpuThread::cn_mainloop_fun cn_trtl_double_mainloop_sandybridge_asm; +namespace xmrig { -extern xmrig::CpuThread::cn_mainloop_fun cn_zls_mainloop_ivybridge_asm; -extern xmrig::CpuThread::cn_mainloop_fun cn_zls_mainloop_ryzen_asm; -extern xmrig::CpuThread::cn_mainloop_fun cn_zls_mainloop_bulldozer_asm; -extern xmrig::CpuThread::cn_mainloop_fun cn_zls_double_mainloop_sandybridge_asm; -extern xmrig::CpuThread::cn_mainloop_fun cn_double_mainloop_ivybridge_asm; -extern xmrig::CpuThread::cn_mainloop_fun cn_double_mainloop_ryzen_asm; -extern xmrig::CpuThread::cn_mainloop_fun cn_double_mainloop_bulldozer_asm; -extern xmrig::CpuThread::cn_mainloop_fun cn_double_double_mainloop_sandybridge_asm; +typedef void (*cn_mainloop_fun)(cryptonight_ctx **ctx); + + +extern cn_mainloop_fun cn_half_mainloop_ivybridge_asm; +extern cn_mainloop_fun cn_half_mainloop_ryzen_asm; +extern cn_mainloop_fun cn_half_mainloop_bulldozer_asm; +extern cn_mainloop_fun cn_half_double_mainloop_sandybridge_asm; + +extern cn_mainloop_fun cn_trtl_mainloop_ivybridge_asm; +extern cn_mainloop_fun cn_trtl_mainloop_ryzen_asm; +extern cn_mainloop_fun cn_trtl_mainloop_bulldozer_asm; +extern cn_mainloop_fun cn_trtl_double_mainloop_sandybridge_asm; + +extern cn_mainloop_fun cn_zls_mainloop_ivybridge_asm; +extern cn_mainloop_fun cn_zls_mainloop_ryzen_asm; +extern cn_mainloop_fun cn_zls_mainloop_bulldozer_asm; +extern cn_mainloop_fun cn_zls_double_mainloop_sandybridge_asm; + +extern cn_mainloop_fun cn_double_mainloop_ivybridge_asm; +extern cn_mainloop_fun cn_double_mainloop_ryzen_asm; +extern cn_mainloop_fun cn_double_mainloop_bulldozer_asm; +extern cn_mainloop_fun cn_double_double_mainloop_sandybridge_asm; + + +} // namespace xmrig + void wow_compile_code(const V4_Instruction* code, int code_size, void* machine_code, xmrig::Assembly ASM); void v4_compile_code(const V4_Instruction* code, int code_size, void* machine_code, xmrig::Assembly ASM); void wow_compile_code_double(const V4_Instruction* code, int code_size, void* machine_code, xmrig::Assembly ASM); void v4_compile_code_double(const V4_Instruction* code, int code_size, void* machine_code, xmrig::Assembly ASM); -template + +template void cn_r_compile_code(const V4_Instruction* code, int code_size, void* machine_code, xmrig::Assembly ASM) { v4_compile_code(code, code_size, machine_code, ASM); } -template + +template void cn_r_compile_code_double(const V4_Instruction* code, int code_size, void* machine_code, xmrig::Assembly ASM) { v4_compile_code_double(code, code_size, machine_code, ASM); } + template<> -void cn_r_compile_code(const V4_Instruction* code, int code_size, void* machine_code, xmrig::Assembly ASM) +void cn_r_compile_code(const V4_Instruction* code, int code_size, void* machine_code, xmrig::Assembly ASM) { wow_compile_code(code, code_size, machine_code, ASM); } + template<> -void cn_r_compile_code_double(const V4_Instruction* code, int code_size, void* machine_code, xmrig::Assembly ASM) +void cn_r_compile_code_double(const V4_Instruction* code, int code_size, void* machine_code, xmrig::Assembly ASM) { wow_compile_code_double(code, code_size, machine_code, ASM); } -template + +namespace xmrig { + + +template inline void cryptonight_single_hash_asm(const uint8_t *__restrict__ input, size_t size, uint8_t *__restrict__ output, cryptonight_ctx **__restrict__ ctx, uint64_t height) { - constexpr size_t MEM = xmrig::cn_select_memory(); + constexpr CnAlgo props; - if (xmrig::cn_is_cryptonight_r() && !ctx[0]->generated_code_data.match(VARIANT, height)) { + if (props.isR() && !ctx[0]->generated_code_data.match(ALGO, height)) { V4_Instruction code[256]; - const int code_size = v4_random_math_init(code, height); - cn_r_compile_code(code, code_size, reinterpret_cast(ctx[0]->generated_code), ASM); - ctx[0]->generated_code_data.variant = VARIANT; - ctx[0]->generated_code_data.height = height; + const int code_size = v4_random_math_init(code, height); + cn_r_compile_code(code, code_size, reinterpret_cast(ctx[0]->generated_code), ASM); + + ctx[0]->generated_code_data = { ALGO, height }; } - xmrig::keccak(input, size, ctx[0]->state); - cn_explode_scratchpad(reinterpret_cast<__m128i*>(ctx[0]->state), reinterpret_cast<__m128i*>(ctx[0]->memory)); + keccak(input, size, ctx[0]->state); + cn_explode_scratchpad(reinterpret_cast(ctx[0]->state), reinterpret_cast<__m128i*>(ctx[0]->memory)); - if (VARIANT == xmrig::VARIANT_2) { - if (ASM == xmrig::ASM_INTEL) { + if (ALGO == Algorithm::CN_2) { + if (ASM == Assembly::INTEL) { cnv2_mainloop_ivybridge_asm(ctx); } - else if (ASM == xmrig::ASM_RYZEN) { + else if (ASM == Assembly::RYZEN) { cnv2_mainloop_ryzen_asm(ctx); } else { cnv2_mainloop_bulldozer_asm(ctx); } } - else if (VARIANT == xmrig::VARIANT_HALF) { - if (ASM == xmrig::ASM_INTEL) { + else if (ALGO == Algorithm::CN_HALF) { + if (ASM == Assembly::INTEL) { cn_half_mainloop_ivybridge_asm(ctx); } - else if (ASM == xmrig::ASM_RYZEN) { + else if (ASM == Assembly::RYZEN) { cn_half_mainloop_ryzen_asm(ctx); } else { cn_half_mainloop_bulldozer_asm(ctx); } } - else if (VARIANT == xmrig::VARIANT_TRTL) { - if (ASM == xmrig::ASM_INTEL) { +# ifdef XMRIG_ALGO_CN_PICO + else if (ALGO == Algorithm::CN_PICO_0) { + if (ASM == Assembly::INTEL) { cn_trtl_mainloop_ivybridge_asm(ctx); } - else if (ASM == xmrig::ASM_RYZEN) { + else if (ASM == Assembly::RYZEN) { cn_trtl_mainloop_ryzen_asm(ctx); } else { cn_trtl_mainloop_bulldozer_asm(ctx); } } - else if (VARIANT == xmrig::VARIANT_RWZ) { +# endif + else if (ALGO == Algorithm::CN_RWZ) { cnv2_rwz_mainloop_asm(ctx); } - else if (VARIANT == xmrig::VARIANT_ZLS) { - if (ASM == xmrig::ASM_INTEL) { + else if (ALGO == Algorithm::CN_ZLS) { + if (ASM == Assembly::INTEL) { cn_zls_mainloop_ivybridge_asm(ctx); } - else if (ASM == xmrig::ASM_RYZEN) { + else if (ASM == Assembly::RYZEN) { cn_zls_mainloop_ryzen_asm(ctx); } else { cn_zls_mainloop_bulldozer_asm(ctx); } } - else if (VARIANT == xmrig::VARIANT_DOUBLE) { - if (ASM == xmrig::ASM_INTEL) { + else if (ALGO == Algorithm::CN_DOUBLE) { + if (ASM == Assembly::INTEL) { cn_double_mainloop_ivybridge_asm(ctx); } - else if (ASM == xmrig::ASM_RYZEN) { + else if (ASM == Assembly::RYZEN) { cn_double_mainloop_ryzen_asm(ctx); } else { cn_double_mainloop_bulldozer_asm(ctx); } } - else if (xmrig::cn_is_cryptonight_r()) { + else if (props.isR()) { ctx[0]->generated_code(ctx); } - cn_implode_scratchpad(reinterpret_cast<__m128i*>(ctx[0]->memory), reinterpret_cast<__m128i*>(ctx[0]->state)); - xmrig::keccakf(reinterpret_cast(ctx[0]->state), 24); + cn_implode_scratchpad(reinterpret_cast(ctx[0]->memory), reinterpret_cast<__m128i*>(ctx[0]->state)); + keccakf(reinterpret_cast(ctx[0]->state), 24); extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output); } -template +template inline void cryptonight_double_hash_asm(const uint8_t *__restrict__ input, size_t size, uint8_t *__restrict__ output, cryptonight_ctx **__restrict__ ctx, uint64_t height) { - constexpr size_t MEM = xmrig::cn_select_memory(); + constexpr CnAlgo props; - if (xmrig::cn_is_cryptonight_r() && !ctx[0]->generated_code_double_data.match(VARIANT, height)) { + if (props.isR() && !ctx[0]->generated_code_data.match(ALGO, height)) { V4_Instruction code[256]; - const int code_size = v4_random_math_init(code, height); - cn_r_compile_code_double(code, code_size, reinterpret_cast(ctx[0]->generated_code_double), ASM); - ctx[0]->generated_code_double_data.variant = VARIANT; - ctx[0]->generated_code_double_data.height = height; + const int code_size = v4_random_math_init(code, height); + cn_r_compile_code_double(code, code_size, reinterpret_cast(ctx[0]->generated_code), ASM); + + ctx[0]->generated_code_data = { ALGO, height }; } - xmrig::keccak(input, size, ctx[0]->state); - xmrig::keccak(input + size, size, ctx[1]->state); + keccak(input, size, ctx[0]->state); + keccak(input + size, size, ctx[1]->state); - cn_explode_scratchpad(reinterpret_cast<__m128i*>(ctx[0]->state), reinterpret_cast<__m128i*>(ctx[0]->memory)); - cn_explode_scratchpad(reinterpret_cast<__m128i*>(ctx[1]->state), reinterpret_cast<__m128i*>(ctx[1]->memory)); + cn_explode_scratchpad(reinterpret_cast(ctx[0]->state), reinterpret_cast<__m128i*>(ctx[0]->memory)); + cn_explode_scratchpad(reinterpret_cast(ctx[1]->state), reinterpret_cast<__m128i*>(ctx[1]->memory)); - if (VARIANT == xmrig::VARIANT_2) { + if (ALGO == Algorithm::CN_2) { cnv2_double_mainloop_sandybridge_asm(ctx); } - else if (VARIANT == xmrig::VARIANT_HALF) { + else if (ALGO == Algorithm::CN_HALF) { cn_half_double_mainloop_sandybridge_asm(ctx); } - else if (VARIANT == xmrig::VARIANT_TRTL) { +# ifdef XMRIG_ALGO_CN_PICO + else if (ALGO == Algorithm::CN_PICO_0) { cn_trtl_double_mainloop_sandybridge_asm(ctx); } - else if (VARIANT == xmrig::VARIANT_RWZ) { +# endif + else if (ALGO == Algorithm::CN_RWZ) { cnv2_rwz_double_mainloop_asm(ctx); } - else if (VARIANT == xmrig::VARIANT_ZLS) { + else if (ALGO == Algorithm::CN_ZLS) { cn_zls_double_mainloop_sandybridge_asm(ctx); } - else if (VARIANT == xmrig::VARIANT_DOUBLE) { + else if (ALGO == Algorithm::CN_DOUBLE) { cn_double_double_mainloop_sandybridge_asm(ctx); } - else if (xmrig::cn_is_cryptonight_r()) { - ctx[0]->generated_code_double(ctx); + else if (props.isR()) { + ctx[0]->generated_code(ctx); } - cn_implode_scratchpad(reinterpret_cast<__m128i*>(ctx[0]->memory), reinterpret_cast<__m128i*>(ctx[0]->state)); - cn_implode_scratchpad(reinterpret_cast<__m128i*>(ctx[1]->memory), reinterpret_cast<__m128i*>(ctx[1]->state)); + cn_implode_scratchpad(reinterpret_cast(ctx[0]->memory), reinterpret_cast<__m128i*>(ctx[0]->state)); + cn_implode_scratchpad(reinterpret_cast(ctx[1]->memory), reinterpret_cast<__m128i*>(ctx[1]->state)); - xmrig::keccakf(reinterpret_cast(ctx[0]->state), 24); - xmrig::keccakf(reinterpret_cast(ctx[1]->state), 24); + keccakf(reinterpret_cast(ctx[0]->state), 24); + keccakf(reinterpret_cast(ctx[1]->state), 24); extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output); extra_hashes[ctx[1]->state[0] & 3](ctx[1]->state, 200, output + 32); } + + +} /* namespace xmrig */ #endif -template +namespace xmrig { + + +template inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t size, uint8_t *__restrict__ output, cryptonight_ctx **__restrict__ ctx, uint64_t height) { - constexpr size_t MASK = xmrig::cn_select_mask(); - constexpr size_t ITERATIONS = xmrig::cn_select_iter(); - constexpr size_t MEM = xmrig::cn_select_memory(); - constexpr xmrig::Variant BASE = xmrig::cn_base_variant(); + constexpr CnAlgo props; + constexpr size_t MASK = props.mask(); + constexpr Algorithm::Id BASE = props.base(); - if (BASE == xmrig::VARIANT_1 && size < 43) { +# ifdef XMRIG_ALGO_CN_HEAVY + constexpr bool IS_CN_HEAVY_TUBE = ALGO == Algorithm::CN_HEAVY_TUBE; +# else + constexpr bool IS_CN_HEAVY_TUBE = false; +# endif + + if (BASE == Algorithm::CN_1 && size < 43) { memset(output, 0, 64); return; } - xmrig::keccak(input, size, ctx[0]->state); - xmrig::keccak(input + size, size, ctx[1]->state); + keccak(input, size, ctx[0]->state); + keccak(input + size, size, ctx[1]->state); - const uint8_t* l0 = ctx[0]->memory; - const uint8_t* l1 = ctx[1]->memory; - uint64_t* h0 = reinterpret_cast(ctx[0]->state); - uint64_t* h1 = reinterpret_cast(ctx[1]->state); + uint8_t *l0 = ctx[0]->memory; + uint8_t *l1 = ctx[1]->memory; + uint64_t *h0 = reinterpret_cast(ctx[0]->state); + uint64_t *h1 = reinterpret_cast(ctx[1]->state); VARIANT1_INIT(0); VARIANT1_INIT(1); @@ -972,8 +1030,8 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si VARIANT4_RANDOM_MATH_INIT(0); VARIANT4_RANDOM_MATH_INIT(1); - cn_explode_scratchpad((__m128i*) h0, (__m128i*) l0); - cn_explode_scratchpad((__m128i*) h1, (__m128i*) l1); + cn_explode_scratchpad(reinterpret_cast(h0), reinterpret_cast<__m128i *>(l0)); + cn_explode_scratchpad(reinterpret_cast(h1), reinterpret_cast<__m128i *>(l1)); uint64_t al0 = h0[0] ^ h0[4]; uint64_t al1 = h1[0] ^ h1[4]; @@ -988,31 +1046,31 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si uint64_t idx0 = al0; uint64_t idx1 = al1; - for (size_t i = 0; i < ITERATIONS; i++) { + for (size_t i = 0; i < props.iterations(); i++) { __m128i cx0, cx1; - if (VARIANT == xmrig::VARIANT_TUBE || !SOFT_AES) { - cx0 = _mm_load_si128((__m128i *) &l0[idx0 & MASK]); - cx1 = _mm_load_si128((__m128i *) &l1[idx1 & MASK]); + if (IS_CN_HEAVY_TUBE || !SOFT_AES) { + cx0 = _mm_load_si128(reinterpret_cast(&l0[idx0 & MASK])); + cx1 = _mm_load_si128(reinterpret_cast(&l1[idx1 & MASK])); } const __m128i ax0 = _mm_set_epi64x(ah0, al0); const __m128i ax1 = _mm_set_epi64x(ah1, al1); - if (VARIANT == xmrig::VARIANT_TUBE) { + if (IS_CN_HEAVY_TUBE) { cx0 = aes_round_tweak_div(cx0, ax0); cx1 = aes_round_tweak_div(cx1, ax1); } else if (SOFT_AES) { - cx0 = soft_aesenc((uint32_t*)&l0[idx0 & MASK], ax0, (const uint32_t*)saes_table); - cx1 = soft_aesenc((uint32_t*)&l1[idx1 & MASK], ax1, (const uint32_t*)saes_table); + cx0 = soft_aesenc(&l0[idx0 & MASK], ax0, reinterpret_cast(saes_table)); + cx1 = soft_aesenc(&l1[idx1 & MASK], ax1, reinterpret_cast(saes_table)); } else { cx0 = _mm_aesenc_si128(cx0, ax0); cx1 = _mm_aesenc_si128(cx1, ax1); } - if (BASE == xmrig::VARIANT_1 || (BASE == xmrig::VARIANT_2)) { - cryptonight_monero_tweak((uint64_t*)&l0[idx0 & MASK], l0, idx0 & MASK, ax0, bx00, bx01, cx0); - cryptonight_monero_tweak((uint64_t*)&l1[idx1 & MASK], l1, idx1 & MASK, ax1, bx10, bx11, cx1); + if (BASE == Algorithm::CN_1 || BASE == Algorithm::CN_2) { + cryptonight_monero_tweak((uint64_t*)&l0[idx0 & MASK], l0, idx0 & MASK, ax0, bx00, bx01, cx0); + cryptonight_monero_tweak((uint64_t*)&l1[idx1 & MASK], l1, idx1 & MASK, ax1, bx10, bx11, cx1); } else { _mm_store_si128((__m128i *) &l0[idx0 & MASK], _mm_xor_si128(bx00, cx0)); _mm_store_si128((__m128i *) &l1[idx1 & MASK], _mm_xor_si128(bx10, cx1)); @@ -1025,10 +1083,10 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si cl = ((uint64_t*) &l0[idx0 & MASK])[0]; ch = ((uint64_t*) &l0[idx0 & MASK])[1]; - if (BASE == xmrig::VARIANT_2) { - if ((VARIANT == xmrig::VARIANT_WOW) || (VARIANT == xmrig::VARIANT_4)) { + if (BASE == Algorithm::CN_2) { + if (props.isR()) { VARIANT4_RANDOM_MATH(0, al0, ah0, cl, bx00, bx01); - if (VARIANT == xmrig::VARIANT_4) { + if (ALGO == Algorithm::CN_R) { al0 ^= r0[2] | ((uint64_t)(r0[3]) << 32); ah0 ^= r0[0] | ((uint64_t)(r0[1]) << 32); } @@ -1039,11 +1097,11 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si lo = __umul128(idx0, cl, &hi); - if (BASE == xmrig::VARIANT_2) { - if (VARIANT == xmrig::VARIANT_4) { + if (BASE == Algorithm::CN_2) { + if (ALGO == Algorithm::CN_R) { VARIANT2_SHUFFLE(l0, idx0 & MASK, ax0, bx00, bx01, cx0, 0); } else { - VARIANT2_SHUFFLE2(l0, idx0 & MASK, ax0, bx00, bx01, hi, lo, (VARIANT == xmrig::VARIANT_RWZ ? 1 : 0)); + VARIANT2_SHUFFLE2(l0, idx0 & MASK, ax0, bx00, bx01, hi, lo, (ALGO == Algorithm::CN_RWZ ? 1 : 0)); } } @@ -1052,9 +1110,9 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si ((uint64_t*)&l0[idx0 & MASK])[0] = al0; - if (BASE == xmrig::VARIANT_1 && (VARIANT == xmrig::VARIANT_TUBE || VARIANT == xmrig::VARIANT_RTO)) { + if (IS_CN_HEAVY_TUBE || ALGO == Algorithm::CN_RTO) { ((uint64_t*) &l0[idx0 & MASK])[1] = ah0 ^ tweak1_2_0 ^ al0; - } else if (BASE == xmrig::VARIANT_1) { + } else if (BASE == Algorithm::CN_1) { ((uint64_t*) &l0[idx0 & MASK])[1] = ah0 ^ tweak1_2_0; } else { ((uint64_t*) &l0[idx0 & MASK])[1] = ah0; @@ -1064,27 +1122,29 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si ah0 ^= ch; idx0 = al0; - if (ALGO == xmrig::CRYPTONIGHT_HEAVY) { +# ifdef XMRIG_ALGO_CN_HEAVY + if (props.isHeavy()) { int64_t n = ((int64_t*)&l0[idx0 & MASK])[0]; int32_t d = ((int32_t*)&l0[idx0 & MASK])[2]; int64_t q = n / (d | 0x5); ((int64_t*)&l0[idx0 & MASK])[0] = n ^ q; - if (VARIANT == xmrig::VARIANT_XHV) { + if (ALGO == Algorithm::CN_HEAVY_XHV) { d = ~d; } idx0 = d ^ q; } +# endif cl = ((uint64_t*) &l1[idx1 & MASK])[0]; ch = ((uint64_t*) &l1[idx1 & MASK])[1]; - if (BASE == xmrig::VARIANT_2) { - if ((VARIANT == xmrig::VARIANT_WOW) || (VARIANT == xmrig::VARIANT_4)) { + if (BASE == Algorithm::CN_2) { + if (props.isR()) { VARIANT4_RANDOM_MATH(1, al1, ah1, cl, bx10, bx11); - if (VARIANT == xmrig::VARIANT_4) { + if (ALGO == Algorithm::CN_R) { al1 ^= r1[2] | ((uint64_t)(r1[3]) << 32); ah1 ^= r1[0] | ((uint64_t)(r1[1]) << 32); } @@ -1095,11 +1155,11 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si lo = __umul128(idx1, cl, &hi); - if (BASE == xmrig::VARIANT_2) { - if (VARIANT == xmrig::VARIANT_4) { + if (BASE == Algorithm::CN_2) { + if (ALGO == Algorithm::CN_R) { VARIANT2_SHUFFLE(l1, idx1 & MASK, ax1, bx10, bx11, cx1, 0); } else { - VARIANT2_SHUFFLE2(l1, idx1 & MASK, ax1, bx10, bx11, hi, lo, (VARIANT == xmrig::VARIANT_RWZ ? 1 : 0)); + VARIANT2_SHUFFLE2(l1, idx1 & MASK, ax1, bx10, bx11, hi, lo, (ALGO == Algorithm::CN_RWZ ? 1 : 0)); } } @@ -1108,9 +1168,9 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si ((uint64_t*)&l1[idx1 & MASK])[0] = al1; - if (BASE == xmrig::VARIANT_1 && (VARIANT == xmrig::VARIANT_TUBE || VARIANT == xmrig::VARIANT_RTO)) { + if (IS_CN_HEAVY_TUBE || ALGO == Algorithm::CN_RTO) { ((uint64_t*)&l1[idx1 & MASK])[1] = ah1 ^ tweak1_2_1 ^ al1; - } else if (BASE == xmrig::VARIANT_1) { + } else if (BASE == Algorithm::CN_1) { ((uint64_t*)&l1[idx1 & MASK])[1] = ah1 ^ tweak1_2_1; } else { ((uint64_t*)&l1[idx1 & MASK])[1] = ah1; @@ -1120,21 +1180,23 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si ah1 ^= ch; idx1 = al1; - if (ALGO == xmrig::CRYPTONIGHT_HEAVY) { +# ifdef XMRIG_ALGO_CN_HEAVY + if (props.isHeavy()) { int64_t n = ((int64_t*)&l1[idx1 & MASK])[0]; int32_t d = ((int32_t*)&l1[idx1 & MASK])[2]; int64_t q = n / (d | 0x5); ((int64_t*)&l1[idx1 & MASK])[0] = n ^ q; - if (VARIANT == xmrig::VARIANT_XHV) { + if (ALGO == Algorithm::CN_HEAVY_XHV) { d = ~d; } idx1 = d ^ q; } +# endif - if (BASE == xmrig::VARIANT_2) { + if (BASE == Algorithm::CN_2) { bx01 = bx00; bx11 = bx10; } @@ -1143,11 +1205,11 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si bx10 = cx1; } - cn_implode_scratchpad((__m128i*) l0, (__m128i*) h0); - cn_implode_scratchpad((__m128i*) l1, (__m128i*) h1); + cn_implode_scratchpad(reinterpret_cast(l0), reinterpret_cast<__m128i *>(h0)); + cn_implode_scratchpad(reinterpret_cast(l1), reinterpret_cast<__m128i *>(h1)); - xmrig::keccakf(h0, 24); - xmrig::keccakf(h1, 24); + keccakf(h0, 24); + keccakf(h1, 24); extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output); extra_hashes[ctx[1]->state[0] & 3](ctx[1]->state, 200, output + 32); @@ -1159,20 +1221,20 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si c = _mm_load_si128(ptr); -#define CN_STEP2(a, b0, b1, c, l, ptr, idx) \ - if (VARIANT == xmrig::VARIANT_TUBE) { \ - c = aes_round_tweak_div(c, a); \ - } \ - else if (SOFT_AES) { \ - c = soft_aesenc(&c, a, (const uint32_t*)saes_table); \ - } else { \ - c = _mm_aesenc_si128(c, a); \ - } \ - \ - if (BASE == xmrig::VARIANT_1 || BASE == xmrig::VARIANT_2) { \ - cryptonight_monero_tweak((uint64_t*)ptr, l, idx & MASK, a, b0, b1, c); \ - } else { \ - _mm_store_si128(ptr, _mm_xor_si128(b0, c)); \ +#define CN_STEP2(a, b0, b1, c, l, ptr, idx) \ + if (IS_CN_HEAVY_TUBE) { \ + c = aes_round_tweak_div(c, a); \ + } \ + else if (SOFT_AES) { \ + c = soft_aesenc(&c, a, (const uint32_t*)saes_table); \ + } else { \ + c = _mm_aesenc_si128(c, a); \ + } \ + \ + if (BASE == Algorithm::CN_1 || BASE == Algorithm::CN_2) { \ + cryptonight_monero_tweak((uint64_t*)ptr, l, idx & MASK, a, b0, b1, c); \ + } else { \ + _mm_store_si128(ptr, _mm_xor_si128(b0, c)); \ } @@ -1183,62 +1245,60 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si uint64_t ch##part = ((uint64_t*)ptr)[1]; -#define CN_STEP4(part, a, b0, b1, c, l, mc, ptr, idx) \ - uint64_t al##part, ah##part; \ - if (BASE == xmrig::VARIANT_2) { \ - if ((VARIANT == xmrig::VARIANT_WOW) || (VARIANT == xmrig::VARIANT_4)) { \ - al##part = _mm_cvtsi128_si64(a); \ - ah##part = _mm_cvtsi128_si64(_mm_srli_si128(a, 8)); \ - VARIANT4_RANDOM_MATH(part, al##part, ah##part, cl##part, b0, b1); \ - if (VARIANT == xmrig::VARIANT_4) { \ - al##part ^= r##part[2] | ((uint64_t)(r##part[3]) << 32); \ - ah##part ^= r##part[0] | ((uint64_t)(r##part[1]) << 32); \ - } \ - } else { \ - VARIANT2_INTEGER_MATH(part, cl##part, c); \ - } \ - } \ - lo = __umul128(idx, cl##part, &hi); \ - if (BASE == xmrig::VARIANT_2) { \ - if (VARIANT == xmrig::VARIANT_4) { \ - VARIANT2_SHUFFLE(l, idx & MASK, a, b0, b1, c, 0); \ - } else { \ - VARIANT2_SHUFFLE2(l, idx & MASK, a, b0, b1, hi, lo, (VARIANT == xmrig::VARIANT_RWZ ? 1 : 0)); \ - } \ - } \ - if (VARIANT == xmrig::VARIANT_4) { \ - a = _mm_set_epi64x(ah##part, al##part); \ - } \ - a = _mm_add_epi64(a, _mm_set_epi64x(lo, hi)); \ - \ - if (BASE == xmrig::VARIANT_1) { \ - _mm_store_si128(ptr, _mm_xor_si128(a, mc)); \ - \ - if (VARIANT == xmrig::VARIANT_TUBE || \ - VARIANT == xmrig::VARIANT_RTO) { \ - ((uint64_t*)ptr)[1] ^= ((uint64_t*)ptr)[0]; \ - } \ - } else { \ - _mm_store_si128(ptr, a); \ - } \ - \ - a = _mm_xor_si128(a, _mm_set_epi64x(ch##part, cl##part)); \ - idx = _mm_cvtsi128_si64(a); \ - \ - if (ALGO == xmrig::CRYPTONIGHT_HEAVY) { \ - int64_t n = ((int64_t*)&l[idx & MASK])[0]; \ - int32_t d = ((int32_t*)&l[idx & MASK])[2]; \ - int64_t q = n / (d | 0x5); \ - ((int64_t*)&l[idx & MASK])[0] = n ^ q; \ - if (VARIANT == xmrig::VARIANT_XHV) { \ - d = ~d; \ - } \ - \ - idx = d ^ q; \ - } \ - if (BASE == xmrig::VARIANT_2) { \ - b1 = b0; \ - } \ +#define CN_STEP4(part, a, b0, b1, c, l, mc, ptr, idx) \ + uint64_t al##part, ah##part; \ + if (BASE == Algorithm::CN_2) { \ + if (props.isR()) { \ + al##part = _mm_cvtsi128_si64(a); \ + ah##part = _mm_cvtsi128_si64(_mm_srli_si128(a, 8)); \ + VARIANT4_RANDOM_MATH(part, al##part, ah##part, cl##part, b0, b1); \ + if (ALGO == Algorithm::CN_R) { \ + al##part ^= r##part[2] | ((uint64_t)(r##part[3]) << 32); \ + ah##part ^= r##part[0] | ((uint64_t)(r##part[1]) << 32); \ + } \ + } else { \ + VARIANT2_INTEGER_MATH(part, cl##part, c); \ + } \ + } \ + lo = __umul128(idx, cl##part, &hi); \ + if (BASE == Algorithm::CN_2) { \ + if (ALGO == Algorithm::CN_R) { \ + VARIANT2_SHUFFLE(l, idx & MASK, a, b0, b1, c, 0); \ + } else { \ + VARIANT2_SHUFFLE2(l, idx & MASK, a, b0, b1, hi, lo, (ALGO == Algorithm::CN_RWZ ? 1 : 0)); \ + } \ + } \ + if (ALGO == Algorithm::CN_R) { \ + a = _mm_set_epi64x(ah##part, al##part); \ + } \ + a = _mm_add_epi64(a, _mm_set_epi64x(lo, hi)); \ + \ + if (BASE == Algorithm::CN_1) { \ + _mm_store_si128(ptr, _mm_xor_si128(a, mc)); \ + \ + if (IS_CN_HEAVY_TUBE || ALGO == Algorithm::CN_RTO) { \ + ((uint64_t*)ptr)[1] ^= ((uint64_t*)ptr)[0]; \ + } \ + } else { \ + _mm_store_si128(ptr, a); \ + } \ + \ + a = _mm_xor_si128(a, _mm_set_epi64x(ch##part, cl##part)); \ + idx = _mm_cvtsi128_si64(a); \ + if (props.isHeavy()) { \ + int64_t n = ((int64_t*)&l[idx & MASK])[0]; \ + int32_t d = ((int32_t*)&l[idx & MASK])[2]; \ + int64_t q = n / (d | 0x5); \ + ((int64_t*)&l[idx & MASK])[0] = n ^ q; \ + if (IS_CN_HEAVY_XHV) { \ + d = ~d; \ + } \ + \ + idx = d ^ q; \ + } \ + if (BASE == Algorithm::CN_2) { \ + b1 = b0; \ + } \ b0 = c; @@ -1246,11 +1306,11 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si __m128i mc##n; \ __m128i division_result_xmm_##n; \ __m128i sqrt_result_xmm_##n; \ - if (BASE == xmrig::VARIANT_1) { \ + if (BASE == Algorithm::CN_1) { \ mc##n = _mm_set_epi64x(*reinterpret_cast(input + n * size + 35) ^ \ *(reinterpret_cast((ctx)->state) + 24), 0); \ } \ - if (BASE == xmrig::VARIANT_2) { \ + if (BASE == Algorithm::CN_2) { \ division_result_xmm_##n = _mm_cvtsi64_si128(h##n[12]); \ sqrt_result_xmm_##n = _mm_cvtsi64_si128(h##n[13]); \ } \ @@ -1261,22 +1321,29 @@ inline void cryptonight_double_hash(const uint8_t *__restrict__ input, size_t si VARIANT4_RANDOM_MATH_INIT(n); -template +template inline void cryptonight_triple_hash(const uint8_t *__restrict__ input, size_t size, uint8_t *__restrict__ output, cryptonight_ctx **__restrict__ ctx, uint64_t height) { - constexpr size_t MASK = xmrig::cn_select_mask(); - constexpr size_t ITERATIONS = xmrig::cn_select_iter(); - constexpr size_t MEM = xmrig::cn_select_memory(); - constexpr xmrig::Variant BASE = xmrig::cn_base_variant(); + constexpr CnAlgo props; + constexpr size_t MASK = props.mask(); + constexpr Algorithm::Id BASE = props.base(); - if (BASE == xmrig::VARIANT_1 && size < 43) { +# ifdef XMRIG_ALGO_CN_HEAVY + constexpr bool IS_CN_HEAVY_TUBE = ALGO == Algorithm::CN_HEAVY_TUBE; + constexpr bool IS_CN_HEAVY_XHV = ALGO == Algorithm::CN_HEAVY_XHV; +# else + constexpr bool IS_CN_HEAVY_TUBE = false; + constexpr bool IS_CN_HEAVY_XHV = false; +# endif + + if (BASE == Algorithm::CN_1 && size < 43) { memset(output, 0, 32 * 3); return; } for (size_t i = 0; i < 3; i++) { - xmrig::keccak(input + size * i, size, ctx[i]->state); - cn_explode_scratchpad(reinterpret_cast<__m128i*>(ctx[i]->state), reinterpret_cast<__m128i*>(ctx[i]->memory)); + keccak(input + size * i, size, ctx[i]->state); + cn_explode_scratchpad(reinterpret_cast(ctx[i]->state), reinterpret_cast<__m128i*>(ctx[i]->memory)); } uint8_t* l0 = ctx[0]->memory; @@ -1296,7 +1363,7 @@ inline void cryptonight_triple_hash(const uint8_t *__restrict__ input, size_t si idx1 = _mm_cvtsi128_si64(ax1); idx2 = _mm_cvtsi128_si64(ax2); - for (size_t i = 0; i < ITERATIONS; i++) { + for (size_t i = 0; i < props.iterations(); i++) { uint64_t hi, lo; __m128i *ptr0, *ptr1, *ptr2; @@ -1318,29 +1385,36 @@ inline void cryptonight_triple_hash(const uint8_t *__restrict__ input, size_t si } for (size_t i = 0; i < 3; i++) { - cn_implode_scratchpad(reinterpret_cast<__m128i*>(ctx[i]->memory), reinterpret_cast<__m128i*>(ctx[i]->state)); - xmrig::keccakf(reinterpret_cast(ctx[i]->state), 24); + cn_implode_scratchpad(reinterpret_cast(ctx[i]->memory), reinterpret_cast<__m128i*>(ctx[i]->state)); + keccakf(reinterpret_cast(ctx[i]->state), 24); extra_hashes[ctx[i]->state[0] & 3](ctx[i]->state, 200, output + 32 * i); } } -template +template inline void cryptonight_quad_hash(const uint8_t *__restrict__ input, size_t size, uint8_t *__restrict__ output, cryptonight_ctx **__restrict__ ctx, uint64_t height) { - constexpr size_t MASK = xmrig::cn_select_mask(); - constexpr size_t ITERATIONS = xmrig::cn_select_iter(); - constexpr size_t MEM = xmrig::cn_select_memory(); - constexpr xmrig::Variant BASE = xmrig::cn_base_variant(); + constexpr CnAlgo props; + constexpr size_t MASK = props.mask(); + constexpr Algorithm::Id BASE = props.base(); - if (BASE == xmrig::VARIANT_1 && size < 43) { +# ifdef XMRIG_ALGO_CN_HEAVY + constexpr bool IS_CN_HEAVY_TUBE = ALGO == Algorithm::CN_HEAVY_TUBE; + constexpr bool IS_CN_HEAVY_XHV = ALGO == Algorithm::CN_HEAVY_XHV; +# else + constexpr bool IS_CN_HEAVY_TUBE = false; + constexpr bool IS_CN_HEAVY_XHV = false; +# endif + + if (BASE == Algorithm::CN_1 && size < 43) { memset(output, 0, 32 * 4); return; } for (size_t i = 0; i < 4; i++) { - xmrig::keccak(input + size * i, size, ctx[i]->state); - cn_explode_scratchpad(reinterpret_cast<__m128i*>(ctx[i]->state), reinterpret_cast<__m128i*>(ctx[i]->memory)); + keccak(input + size * i, size, ctx[i]->state); + cn_explode_scratchpad(reinterpret_cast(ctx[i]->state), reinterpret_cast<__m128i*>(ctx[i]->memory)); } uint8_t* l0 = ctx[0]->memory; @@ -1364,8 +1438,7 @@ inline void cryptonight_quad_hash(const uint8_t *__restrict__ input, size_t size idx2 = _mm_cvtsi128_si64(ax2); idx3 = _mm_cvtsi128_si64(ax3); - for (size_t i = 0; i < ITERATIONS; i++) - { + for (size_t i = 0; i < props.iterations(); i++) { uint64_t hi, lo; __m128i *ptr0, *ptr1, *ptr2, *ptr3; @@ -1391,29 +1464,36 @@ inline void cryptonight_quad_hash(const uint8_t *__restrict__ input, size_t size } for (size_t i = 0; i < 4; i++) { - cn_implode_scratchpad(reinterpret_cast<__m128i*>(ctx[i]->memory), reinterpret_cast<__m128i*>(ctx[i]->state)); - xmrig::keccakf(reinterpret_cast(ctx[i]->state), 24); + cn_implode_scratchpad(reinterpret_cast(ctx[i]->memory), reinterpret_cast<__m128i*>(ctx[i]->state)); + keccakf(reinterpret_cast(ctx[i]->state), 24); extra_hashes[ctx[i]->state[0] & 3](ctx[i]->state, 200, output + 32 * i); } } -template +template inline void cryptonight_penta_hash(const uint8_t *__restrict__ input, size_t size, uint8_t *__restrict__ output, cryptonight_ctx **__restrict__ ctx, uint64_t height) { - constexpr size_t MASK = xmrig::cn_select_mask(); - constexpr size_t ITERATIONS = xmrig::cn_select_iter(); - constexpr size_t MEM = xmrig::cn_select_memory(); - constexpr xmrig::Variant BASE = xmrig::cn_base_variant(); + constexpr CnAlgo props; + constexpr size_t MASK = props.mask(); + constexpr Algorithm::Id BASE = props.base(); - if (BASE == xmrig::VARIANT_1 && size < 43) { +# ifdef XMRIG_ALGO_CN_HEAVY + constexpr bool IS_CN_HEAVY_TUBE = ALGO == Algorithm::CN_HEAVY_TUBE; + constexpr bool IS_CN_HEAVY_XHV = ALGO == Algorithm::CN_HEAVY_XHV; +# else + constexpr bool IS_CN_HEAVY_TUBE = false; + constexpr bool IS_CN_HEAVY_XHV = false; +# endif + + if (BASE == Algorithm::CN_1 && size < 43) { memset(output, 0, 32 * 5); return; } for (size_t i = 0; i < 5; i++) { - xmrig::keccak(input + size * i, size, ctx[i]->state); - cn_explode_scratchpad(reinterpret_cast<__m128i*>(ctx[i]->state), reinterpret_cast<__m128i*>(ctx[i]->memory)); + keccak(input + size * i, size, ctx[i]->state); + cn_explode_scratchpad(reinterpret_cast(ctx[i]->state), reinterpret_cast<__m128i*>(ctx[i]->memory)); } uint8_t* l0 = ctx[0]->memory; @@ -1441,8 +1521,7 @@ inline void cryptonight_penta_hash(const uint8_t *__restrict__ input, size_t siz idx3 = _mm_cvtsi128_si64(ax3); idx4 = _mm_cvtsi128_si64(ax4); - for (size_t i = 0; i < ITERATIONS; i++) - { + for (size_t i = 0; i < props.iterations(); i++) { uint64_t hi, lo; __m128i *ptr0, *ptr1, *ptr2, *ptr3, *ptr4; @@ -1472,10 +1551,14 @@ inline void cryptonight_penta_hash(const uint8_t *__restrict__ input, size_t siz } for (size_t i = 0; i < 5; i++) { - cn_implode_scratchpad(reinterpret_cast<__m128i*>(ctx[i]->memory), reinterpret_cast<__m128i*>(ctx[i]->state)); - xmrig::keccakf(reinterpret_cast(ctx[i]->state), 24); + cn_implode_scratchpad(reinterpret_cast(ctx[i]->memory), reinterpret_cast<__m128i*>(ctx[i]->state)); + keccakf(reinterpret_cast(ctx[i]->state), 24); extra_hashes[ctx[i]->state[0] & 3](ctx[i]->state, 200, output + 32 * i); } } + +} /* namespace xmrig */ + + #endif /* XMRIG_CRYPTONIGHT_X86_H */ diff --git a/src/crypto/SSE2NEON.h b/src/crypto/cn/SSE2NEON.h similarity index 100% rename from src/crypto/SSE2NEON.h rename to src/crypto/cn/SSE2NEON.h diff --git a/src/crypto/asm/CryptonightR_soft_aes_template.inc b/src/crypto/cn/asm/CryptonightR_soft_aes_template.inc similarity index 100% rename from src/crypto/asm/CryptonightR_soft_aes_template.inc rename to src/crypto/cn/asm/CryptonightR_soft_aes_template.inc diff --git a/src/crypto/asm/CryptonightR_soft_aes_template_win.inc b/src/crypto/cn/asm/CryptonightR_soft_aes_template_win.inc similarity index 100% rename from src/crypto/asm/CryptonightR_soft_aes_template_win.inc rename to src/crypto/cn/asm/CryptonightR_soft_aes_template_win.inc diff --git a/src/crypto/asm/CryptonightR_template.S b/src/crypto/cn/asm/CryptonightR_template.S similarity index 100% rename from src/crypto/asm/CryptonightR_template.S rename to src/crypto/cn/asm/CryptonightR_template.S diff --git a/src/crypto/asm/CryptonightR_template.asm b/src/crypto/cn/asm/CryptonightR_template.asm similarity index 100% rename from src/crypto/asm/CryptonightR_template.asm rename to src/crypto/cn/asm/CryptonightR_template.asm diff --git a/src/crypto/asm/CryptonightR_template.h b/src/crypto/cn/asm/CryptonightR_template.h similarity index 100% rename from src/crypto/asm/CryptonightR_template.h rename to src/crypto/cn/asm/CryptonightR_template.h diff --git a/src/crypto/asm/CryptonightR_template.inc b/src/crypto/cn/asm/CryptonightR_template.inc similarity index 100% rename from src/crypto/asm/CryptonightR_template.inc rename to src/crypto/cn/asm/CryptonightR_template.inc diff --git a/src/crypto/asm/CryptonightR_template_win.inc b/src/crypto/cn/asm/CryptonightR_template_win.inc similarity index 100% rename from src/crypto/asm/CryptonightR_template_win.inc rename to src/crypto/cn/asm/CryptonightR_template_win.inc diff --git a/src/crypto/asm/CryptonightWOW_soft_aes_template.inc b/src/crypto/cn/asm/CryptonightWOW_soft_aes_template.inc similarity index 100% rename from src/crypto/asm/CryptonightWOW_soft_aes_template.inc rename to src/crypto/cn/asm/CryptonightWOW_soft_aes_template.inc diff --git a/src/crypto/asm/CryptonightWOW_soft_aes_template_win.inc b/src/crypto/cn/asm/CryptonightWOW_soft_aes_template_win.inc similarity index 100% rename from src/crypto/asm/CryptonightWOW_soft_aes_template_win.inc rename to src/crypto/cn/asm/CryptonightWOW_soft_aes_template_win.inc diff --git a/src/crypto/asm/CryptonightWOW_template.inc b/src/crypto/cn/asm/CryptonightWOW_template.inc similarity index 100% rename from src/crypto/asm/CryptonightWOW_template.inc rename to src/crypto/cn/asm/CryptonightWOW_template.inc diff --git a/src/crypto/asm/CryptonightWOW_template_win.inc b/src/crypto/cn/asm/CryptonightWOW_template_win.inc similarity index 100% rename from src/crypto/asm/CryptonightWOW_template_win.inc rename to src/crypto/cn/asm/CryptonightWOW_template_win.inc diff --git a/src/crypto/asm/cn2/cnv2_double_main_loop_sandybridge.inc b/src/crypto/cn/asm/cn2/cnv2_double_main_loop_sandybridge.inc similarity index 100% rename from src/crypto/asm/cn2/cnv2_double_main_loop_sandybridge.inc rename to src/crypto/cn/asm/cn2/cnv2_double_main_loop_sandybridge.inc diff --git a/src/crypto/asm/cn2/cnv2_main_loop_bulldozer.inc b/src/crypto/cn/asm/cn2/cnv2_main_loop_bulldozer.inc similarity index 100% rename from src/crypto/asm/cn2/cnv2_main_loop_bulldozer.inc rename to src/crypto/cn/asm/cn2/cnv2_main_loop_bulldozer.inc diff --git a/src/crypto/asm/cn2/cnv2_main_loop_ivybridge.inc b/src/crypto/cn/asm/cn2/cnv2_main_loop_ivybridge.inc similarity index 100% rename from src/crypto/asm/cn2/cnv2_main_loop_ivybridge.inc rename to src/crypto/cn/asm/cn2/cnv2_main_loop_ivybridge.inc diff --git a/src/crypto/asm/cn2/cnv2_main_loop_ryzen.inc b/src/crypto/cn/asm/cn2/cnv2_main_loop_ryzen.inc similarity index 100% rename from src/crypto/asm/cn2/cnv2_main_loop_ryzen.inc rename to src/crypto/cn/asm/cn2/cnv2_main_loop_ryzen.inc diff --git a/src/crypto/asm/cn2/cnv2_rwz_double_main_loop.inc b/src/crypto/cn/asm/cn2/cnv2_rwz_double_main_loop.inc similarity index 100% rename from src/crypto/asm/cn2/cnv2_rwz_double_main_loop.inc rename to src/crypto/cn/asm/cn2/cnv2_rwz_double_main_loop.inc diff --git a/src/crypto/asm/cn2/cnv2_rwz_main_loop.inc b/src/crypto/cn/asm/cn2/cnv2_rwz_main_loop.inc similarity index 100% rename from src/crypto/asm/cn2/cnv2_rwz_main_loop.inc rename to src/crypto/cn/asm/cn2/cnv2_rwz_main_loop.inc diff --git a/src/crypto/asm/cn_main_loop.S b/src/crypto/cn/asm/cn_main_loop.S similarity index 100% rename from src/crypto/asm/cn_main_loop.S rename to src/crypto/cn/asm/cn_main_loop.S diff --git a/src/crypto/asm/cn_main_loop.asm b/src/crypto/cn/asm/cn_main_loop.asm similarity index 100% rename from src/crypto/asm/cn_main_loop.asm rename to src/crypto/cn/asm/cn_main_loop.asm diff --git a/src/crypto/asm/win64/CryptonightR_soft_aes_template_win.inc b/src/crypto/cn/asm/win64/CryptonightR_soft_aes_template_win.inc similarity index 100% rename from src/crypto/asm/win64/CryptonightR_soft_aes_template_win.inc rename to src/crypto/cn/asm/win64/CryptonightR_soft_aes_template_win.inc diff --git a/src/crypto/asm/win64/CryptonightR_template.asm b/src/crypto/cn/asm/win64/CryptonightR_template.asm similarity index 100% rename from src/crypto/asm/win64/CryptonightR_template.asm rename to src/crypto/cn/asm/win64/CryptonightR_template.asm diff --git a/src/crypto/asm/win64/CryptonightR_template_win.inc b/src/crypto/cn/asm/win64/CryptonightR_template_win.inc similarity index 100% rename from src/crypto/asm/win64/CryptonightR_template_win.inc rename to src/crypto/cn/asm/win64/CryptonightR_template_win.inc diff --git a/src/crypto/asm/win64/CryptonightWOW_soft_aes_template_win.inc b/src/crypto/cn/asm/win64/CryptonightWOW_soft_aes_template_win.inc similarity index 100% rename from src/crypto/asm/win64/CryptonightWOW_soft_aes_template_win.inc rename to src/crypto/cn/asm/win64/CryptonightWOW_soft_aes_template_win.inc diff --git a/src/crypto/asm/win64/CryptonightWOW_template_win.inc b/src/crypto/cn/asm/win64/CryptonightWOW_template_win.inc similarity index 100% rename from src/crypto/asm/win64/CryptonightWOW_template_win.inc rename to src/crypto/cn/asm/win64/CryptonightWOW_template_win.inc diff --git a/src/crypto/asm/win64/cn2/cnv2_double_main_loop_sandybridge.inc b/src/crypto/cn/asm/win64/cn2/cnv2_double_main_loop_sandybridge.inc similarity index 100% rename from src/crypto/asm/win64/cn2/cnv2_double_main_loop_sandybridge.inc rename to src/crypto/cn/asm/win64/cn2/cnv2_double_main_loop_sandybridge.inc diff --git a/src/crypto/asm/win64/cn2/cnv2_main_loop_bulldozer.inc b/src/crypto/cn/asm/win64/cn2/cnv2_main_loop_bulldozer.inc similarity index 100% rename from src/crypto/asm/win64/cn2/cnv2_main_loop_bulldozer.inc rename to src/crypto/cn/asm/win64/cn2/cnv2_main_loop_bulldozer.inc diff --git a/src/crypto/asm/win64/cn2/cnv2_main_loop_ivybridge.inc b/src/crypto/cn/asm/win64/cn2/cnv2_main_loop_ivybridge.inc similarity index 100% rename from src/crypto/asm/win64/cn2/cnv2_main_loop_ivybridge.inc rename to src/crypto/cn/asm/win64/cn2/cnv2_main_loop_ivybridge.inc diff --git a/src/crypto/asm/win64/cn2/cnv2_main_loop_ryzen.inc b/src/crypto/cn/asm/win64/cn2/cnv2_main_loop_ryzen.inc similarity index 100% rename from src/crypto/asm/win64/cn2/cnv2_main_loop_ryzen.inc rename to src/crypto/cn/asm/win64/cn2/cnv2_main_loop_ryzen.inc diff --git a/src/crypto/asm/win64/cn2/cnv2_rwz_double_main_loop.inc b/src/crypto/cn/asm/win64/cn2/cnv2_rwz_double_main_loop.inc similarity index 100% rename from src/crypto/asm/win64/cn2/cnv2_rwz_double_main_loop.inc rename to src/crypto/cn/asm/win64/cn2/cnv2_rwz_double_main_loop.inc diff --git a/src/crypto/asm/win64/cn2/cnv2_rwz_main_loop.inc b/src/crypto/cn/asm/win64/cn2/cnv2_rwz_main_loop.inc similarity index 100% rename from src/crypto/asm/win64/cn2/cnv2_rwz_main_loop.inc rename to src/crypto/cn/asm/win64/cn2/cnv2_rwz_main_loop.inc diff --git a/src/crypto/asm/win64/cn_main_loop.S b/src/crypto/cn/asm/win64/cn_main_loop.S similarity index 100% rename from src/crypto/asm/win64/cn_main_loop.S rename to src/crypto/cn/asm/win64/cn_main_loop.S diff --git a/src/crypto/asm/win64/cn_main_loop.asm b/src/crypto/cn/asm/win64/cn_main_loop.asm similarity index 100% rename from src/crypto/asm/win64/cn_main_loop.asm rename to src/crypto/cn/asm/win64/cn_main_loop.asm diff --git a/src/crypto/c_blake256.c b/src/crypto/cn/c_blake256.c similarity index 100% rename from src/crypto/c_blake256.c rename to src/crypto/cn/c_blake256.c diff --git a/src/crypto/c_blake256.h b/src/crypto/cn/c_blake256.h similarity index 100% rename from src/crypto/c_blake256.h rename to src/crypto/cn/c_blake256.h diff --git a/src/crypto/c_groestl.c b/src/crypto/cn/c_groestl.c similarity index 100% rename from src/crypto/c_groestl.c rename to src/crypto/cn/c_groestl.c diff --git a/src/crypto/c_groestl.h b/src/crypto/cn/c_groestl.h similarity index 100% rename from src/crypto/c_groestl.h rename to src/crypto/cn/c_groestl.h diff --git a/src/crypto/c_jh.c b/src/crypto/cn/c_jh.c similarity index 100% rename from src/crypto/c_jh.c rename to src/crypto/cn/c_jh.c diff --git a/src/crypto/c_jh.h b/src/crypto/cn/c_jh.h similarity index 100% rename from src/crypto/c_jh.h rename to src/crypto/cn/c_jh.h diff --git a/src/crypto/c_skein.c b/src/crypto/cn/c_skein.c similarity index 100% rename from src/crypto/c_skein.c rename to src/crypto/cn/c_skein.c diff --git a/src/crypto/c_skein.h b/src/crypto/cn/c_skein.h similarity index 100% rename from src/crypto/c_skein.h rename to src/crypto/cn/c_skein.h diff --git a/src/crypto/cn_gpu_arm.cpp b/src/crypto/cn/gpu/cn_gpu_arm.cpp similarity index 97% rename from src/crypto/cn_gpu_arm.cpp rename to src/crypto/cn/gpu/cn_gpu_arm.cpp index b463dd2ec..520d3fc8c 100644 --- a/src/crypto/cn_gpu_arm.cpp +++ b/src/crypto/cn/gpu/cn_gpu_arm.cpp @@ -26,7 +26,7 @@ #include -#include "crypto/CryptoNight_constants.h" +#include "crypto/cn/CnAlgo.h" inline void vandq_f32(float32x4_t &v, uint32_t v2) @@ -237,4 +237,4 @@ void cn_gpu_inner_arm(const uint8_t *spad, uint8_t *lpad) } } -template void cn_gpu_inner_arm(const uint8_t* spad, uint8_t* lpad); +template void cn_gpu_inner_arm().iterations(), xmrig::CnAlgo().mask()>(const uint8_t* spad, uint8_t* lpad); diff --git a/src/crypto/cn_gpu_avx.cpp b/src/crypto/cn/gpu/cn_gpu_avx.cpp similarity index 97% rename from src/crypto/cn_gpu_avx.cpp rename to src/crypto/cn/gpu/cn_gpu_avx.cpp index 9f801c80a..38da97149 100644 --- a/src/crypto/cn_gpu_avx.cpp +++ b/src/crypto/cn/gpu/cn_gpu_avx.cpp @@ -22,7 +22,9 @@ * along with this program. If not, see . */ -#include "crypto/CryptoNight_constants.h" + +#include "crypto/cn/CnAlgo.h" + #ifdef __GNUC__ # include @@ -206,4 +208,4 @@ void cn_gpu_inner_avx(const uint8_t* spad, uint8_t* lpad) } } -template void cn_gpu_inner_avx(const uint8_t* spad, uint8_t* lpad); +template void cn_gpu_inner_avx().iterations(), xmrig::CnAlgo().mask()>(const uint8_t* spad, uint8_t* lpad); diff --git a/src/crypto/cn_gpu_ssse3.cpp b/src/crypto/cn/gpu/cn_gpu_ssse3.cpp similarity index 97% rename from src/crypto/cn_gpu_ssse3.cpp rename to src/crypto/cn/gpu/cn_gpu_ssse3.cpp index ce3d19add..7cca096ef 100644 --- a/src/crypto/cn_gpu_ssse3.cpp +++ b/src/crypto/cn/gpu/cn_gpu_ssse3.cpp @@ -22,7 +22,9 @@ * along with this program. If not, see . */ -#include "crypto/CryptoNight_constants.h" + +#include "crypto/cn/CnAlgo.h" + #ifdef __GNUC__ # include @@ -207,4 +209,4 @@ void cn_gpu_inner_ssse3(const uint8_t* spad, uint8_t* lpad) } } -template void cn_gpu_inner_ssse3(const uint8_t* spad, uint8_t* lpad); +template void cn_gpu_inner_ssse3().iterations(), xmrig::CnAlgo().mask()>(const uint8_t* spad, uint8_t* lpad); diff --git a/src/crypto/groestl_tables.h b/src/crypto/cn/groestl_tables.h similarity index 100% rename from src/crypto/groestl_tables.h rename to src/crypto/cn/groestl_tables.h diff --git a/src/crypto/hash.h b/src/crypto/cn/hash.h similarity index 100% rename from src/crypto/hash.h rename to src/crypto/cn/hash.h diff --git a/src/crypto/CryptonightR_gen.cpp b/src/crypto/cn/r/CryptonightR_gen.cpp similarity index 91% rename from src/crypto/CryptonightR_gen.cpp rename to src/crypto/cn/r/CryptonightR_gen.cpp index 3fba49cd9..3b80f8051 100644 --- a/src/crypto/CryptonightR_gen.cpp +++ b/src/crypto/cn/r/CryptonightR_gen.cpp @@ -24,12 +24,13 @@ */ #include -#include "crypto/CryptoNight_monero.h" +#include "crypto/cn/CryptoNight_monero.h" typedef void(*void_func)(); -#include "crypto/asm/CryptonightR_template.h" -#include "Mem.h" +#include "crypto/cn/asm/CryptonightR_template.h" +#include "crypto/common/Assembly.h" +#include "crypto/common/VirtualMemory.h" static inline void add_code(uint8_t* &p, void (*p1)(), void (*p2)()) @@ -41,7 +42,7 @@ static inline void add_code(uint8_t* &p, void (*p1)(), void (*p2)()) } } -static inline void add_random_math(uint8_t* &p, const V4_Instruction* code, int code_size, const void_func* instructions, const void_func* instructions_mov, bool is_64_bit, xmrig::Assembly ASM) +static inline void add_random_math(uint8_t* &p, const V4_Instruction* code, int code_size, const void_func* instructions, const void_func* instructions_mov, bool is_64_bit, xmrig::Assembly::Id ASM) { uint32_t prev_rot_src = (uint32_t)(-1); @@ -75,7 +76,7 @@ static inline void add_random_math(uint8_t* &p, const V4_Instruction* code, int void_func begin = instructions[c]; - if ((ASM = xmrig::ASM_BULLDOZER) && (inst.opcode == MUL) && !is_64_bit) { + if ((ASM = xmrig::Assembly::BULLDOZER) && (inst.opcode == MUL) && !is_64_bit) { // AMD Bulldozer has latency 4 for 32-bit IMUL and 6 for 64-bit IMUL // Always use 32-bit IMUL for AMD Bulldozer in 32-bit mode - skip prefix 0x48 and change 0x49 to 0x41 uint8_t* prefix = reinterpret_cast(begin); @@ -109,7 +110,7 @@ void wow_compile_code(const V4_Instruction* code, int code_size, void* machine_c *(int*)(p - 4) = static_cast((((const uint8_t*)CryptonightWOW_template_mainloop) - ((const uint8_t*)CryptonightWOW_template_part1)) - (p - p0)); add_code(p, CryptonightWOW_template_part3, CryptonightWOW_template_end); - Mem::flushInstructionCache(machine_code, p - p0); + xmrig::VirtualMemory::flushInstructionCache(machine_code, p - p0); } void v4_compile_code(const V4_Instruction* code, int code_size, void* machine_code, xmrig::Assembly ASM) @@ -123,7 +124,7 @@ void v4_compile_code(const V4_Instruction* code, int code_size, void* machine_co *(int*)(p - 4) = static_cast((((const uint8_t*)CryptonightR_template_mainloop) - ((const uint8_t*)CryptonightR_template_part1)) - (p - p0)); add_code(p, CryptonightR_template_part3, CryptonightR_template_end); - Mem::flushInstructionCache(machine_code, p - p0); + xmrig::VirtualMemory::flushInstructionCache(machine_code, p - p0); } void wow_compile_code_double(const V4_Instruction* code, int code_size, void* machine_code, xmrig::Assembly ASM) @@ -139,7 +140,7 @@ void wow_compile_code_double(const V4_Instruction* code, int code_size, void* ma *(int*)(p - 4) = static_cast((((const uint8_t*)CryptonightWOW_template_double_mainloop) - ((const uint8_t*)CryptonightWOW_template_double_part1)) - (p - p0)); add_code(p, CryptonightWOW_template_double_part4, CryptonightWOW_template_double_end); - Mem::flushInstructionCache(machine_code, p - p0); + xmrig::VirtualMemory::flushInstructionCache(machine_code, p - p0); } void v4_compile_code_double(const V4_Instruction* code, int code_size, void* machine_code, xmrig::Assembly ASM) @@ -155,7 +156,7 @@ void v4_compile_code_double(const V4_Instruction* code, int code_size, void* mac *(int*)(p - 4) = static_cast((((const uint8_t*)CryptonightR_template_double_mainloop) - ((const uint8_t*)CryptonightR_template_double_part1)) - (p - p0)); add_code(p, CryptonightR_template_double_part4, CryptonightR_template_double_end); - Mem::flushInstructionCache(machine_code, p - p0); + xmrig::VirtualMemory::flushInstructionCache(machine_code, p - p0); } void wow_soft_aes_compile_code(const V4_Instruction* code, int code_size, void* machine_code, xmrig::Assembly ASM) @@ -169,7 +170,7 @@ void wow_soft_aes_compile_code(const V4_Instruction* code, int code_size, void* *(int*)(p - 4) = static_cast((((const uint8_t*)CryptonightWOW_soft_aes_template_mainloop) - ((const uint8_t*)CryptonightWOW_soft_aes_template_part1)) - (p - p0)); add_code(p, CryptonightWOW_soft_aes_template_part3, CryptonightWOW_soft_aes_template_end); - Mem::flushInstructionCache(machine_code, p - p0); + xmrig::VirtualMemory::flushInstructionCache(machine_code, p - p0); } void v4_soft_aes_compile_code(const V4_Instruction* code, int code_size, void* machine_code, xmrig::Assembly ASM) @@ -183,5 +184,5 @@ void v4_soft_aes_compile_code(const V4_Instruction* code, int code_size, void* m *(int*)(p - 4) = static_cast((((const uint8_t*)CryptonightR_soft_aes_template_mainloop) - ((const uint8_t*)CryptonightR_soft_aes_template_part1)) - (p - p0)); add_code(p, CryptonightR_soft_aes_template_part3, CryptonightR_soft_aes_template_end); - Mem::flushInstructionCache(machine_code, p - p0); + xmrig::VirtualMemory::flushInstructionCache(machine_code, p - p0); } diff --git a/src/crypto/variant4_random_math.h b/src/crypto/cn/r/variant4_random_math.h similarity index 97% rename from src/crypto/variant4_random_math.h rename to src/crypto/cn/r/variant4_random_math.h index 1f3ea0ac3..48f0f6ce0 100644 --- a/src/crypto/variant4_random_math.h +++ b/src/crypto/cn/r/variant4_random_math.h @@ -1,9 +1,16 @@ #ifndef VARIANT4_RANDOM_MATH_H #define VARIANT4_RANDOM_MATH_H + +#include + + +#include "crypto/common/Algorithm.h" + + extern "C" { - #include "c_blake256.h" + #include "crypto/cn/c_blake256.h" } enum V4_Settings @@ -182,7 +189,7 @@ static FORCEINLINE void check_data(size_t* data_index, const size_t bytes_needed // Generates as many random math operations as possible with given latency and ALU restrictions // "code" array must have space for NUM_INSTRUCTIONS_MAX+1 instructions -template +template static int v4_random_math_init(struct V4_Instruction* code, const uint64_t height) { // MUL is 3 cycles, 3-way addition and rotations are 2 cycles, SUB/XOR are 1 cycle @@ -204,8 +211,7 @@ static int v4_random_math_init(struct V4_Instruction* code, const uint64_t heigh memset(data, 0, sizeof(data)); uint64_t tmp = SWAP64LE(height); memcpy(data, &tmp, sizeof(uint64_t)); - if (VARIANT == xmrig::VARIANT_4) - { + if (ALGO == xmrig::Algorithm::CN_R) { data[20] = -38; } @@ -249,7 +255,7 @@ static int v4_random_math_init(struct V4_Instruction* code, const uint64_t heigh code_size = 0; int total_iterations = 0; - r8_used = (VARIANT == xmrig::VARIANT_WOW); + r8_used = (ALGO == xmrig::Algorithm::CN_WOW); // Generate random code to achieve minimal required latency for our abstract CPU // Try to get this latency for all 4 registers @@ -291,10 +297,9 @@ static int v4_random_math_init(struct V4_Instruction* code, const uint64_t heigh int b = src_index; // Don't do ADD/SUB/XOR with the same register - if (((opcode == ADD) || (opcode == SUB) || (opcode == XOR)) && (a == b)) - { + if (((opcode == ADD) || (opcode == SUB) || (opcode == XOR)) && (a == b)) { // a is always < 4, so we don't need to check bounds here - b = (VARIANT == xmrig::VARIANT_WOW) ? (a + 4) : 8; + b = (ALGO == xmrig::Algorithm::CN_WOW) ? (a + 4) : 8; src_index = b; } diff --git a/src/crypto/skein_port.h b/src/crypto/cn/skein_port.h similarity index 100% rename from src/crypto/skein_port.h rename to src/crypto/cn/skein_port.h diff --git a/src/crypto/soft_aes.h b/src/crypto/cn/soft_aes.h similarity index 99% rename from src/crypto/soft_aes.h rename to src/crypto/cn/soft_aes.h index 4ad9bdd93..fca31d1c1 100644 --- a/src/crypto/soft_aes.h +++ b/src/crypto/cn/soft_aes.h @@ -28,7 +28,7 @@ #if defined(XMRIG_ARM) -# include "crypto/SSE2NEON.h" +# include "crypto/cn/SSE2NEON.h" #elif defined(__GNUC__) # include #else diff --git a/src/crypto/common/Algorithm.cpp b/src/crypto/common/Algorithm.cpp new file mode 100644 index 000000000..db6cb234b --- /dev/null +++ b/src/crypto/common/Algorithm.cpp @@ -0,0 +1,281 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include +#include +#include +#include + + +#include "crypto/cn/CnAlgo.h" +#include "crypto/common/Algorithm.h" +#include "rapidjson/document.h" + + +#ifdef _MSC_VER +# define strcasecmp _stricmp +#endif + + +#ifndef ARRAY_SIZE +# define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0])) +#endif + + +namespace xmrig { + + +struct AlgoName +{ + const char *name; + const char *shortName; + const Algorithm::Id id; +}; + + +static AlgoName const algorithm_names[] = { + { "cryptonight/0", "cn/0", Algorithm::CN_0 }, + { "cryptonight", "cn", Algorithm::CN_0 }, + { "cryptonight/1", "cn/1", Algorithm::CN_1 }, + { "cryptonight-monerov7", nullptr, Algorithm::CN_1 }, + { "cryptonight_v7", nullptr, Algorithm::CN_1 }, + { "cryptonight/2", "cn/2", Algorithm::CN_2 }, + { "cryptonight-monerov8", nullptr, Algorithm::CN_2 }, + { "cryptonight_v8", nullptr, Algorithm::CN_2 }, + { "cryptonight/r", "cn/r", Algorithm::CN_R }, + { "cryptonight_r", nullptr, Algorithm::CN_R }, + { "cryptonight/wow", "cn/wow", Algorithm::CN_WOW }, + { "cryptonight/fast", "cn/fast", Algorithm::CN_FAST }, + { "cryptonight/msr", "cn/msr", Algorithm::CN_FAST }, + { "cryptonight/half", "cn/half", Algorithm::CN_HALF }, + { "cryptonight/xao", "cn/xao", Algorithm::CN_XAO }, + { "cryptonight_alloy", nullptr, Algorithm::CN_XAO }, + { "cryptonight/rto", "cn/rto", Algorithm::CN_RTO }, + { "cryptonight/rwz", "cn/rwz", Algorithm::CN_RWZ }, + { "cryptonight/zls", "cn/zls", Algorithm::CN_ZLS }, + { "cryptonight/double", "cn/double", Algorithm::CN_DOUBLE }, +# ifdef XMRIG_ALGO_CN_GPU + { "cryptonight/gpu", "cn/gpu", Algorithm::CN_GPU }, + { "cryptonight_gpu", nullptr, Algorithm::CN_GPU }, +# endif +# ifdef XMRIG_ALGO_CN_LITE + { "cryptonight-lite/0", "cn-lite/0", Algorithm::CN_LITE_0 }, + { "cryptonight-lite/1", "cn-lite/1", Algorithm::CN_LITE_1 }, + { "cryptonight-lite", "cn-lite", Algorithm::CN_LITE_1 }, + { "cryptonight-light", "cn-light", Algorithm::CN_LITE_1 }, + { "cryptonight_lite", nullptr, Algorithm::CN_LITE_1 }, + { "cryptonight-aeonv7", nullptr, Algorithm::CN_LITE_1 }, + { "cryptonight_lite_v7", nullptr, Algorithm::CN_LITE_1 }, +# endif +# ifdef XMRIG_ALGO_CN_HEAVY + { "cryptonight-heavy/0", "cn-heavy/0", Algorithm::CN_HEAVY_0 }, + { "cryptonight-heavy", "cn-heavy", Algorithm::CN_HEAVY_0 }, + { "cryptonight_heavy", nullptr, Algorithm::CN_HEAVY_0 }, + { "cryptonight-heavy/xhv", "cn-heavy/xhv", Algorithm::CN_HEAVY_XHV }, + { "cryptonight_haven", nullptr, Algorithm::CN_HEAVY_XHV }, + { "cryptonight-heavy/tube", "cn-heavy/tube", Algorithm::CN_HEAVY_TUBE }, + { "cryptonight-bittube2", nullptr, Algorithm::CN_HEAVY_TUBE }, +# endif +# ifdef XMRIG_ALGO_CN_PICO + { "cryptonight-pico", "cn-pico", Algorithm::CN_PICO_0 }, + { "cryptonight-pico/trtl", "cn-pico/trtl", Algorithm::CN_PICO_0 }, + { "cryptonight-turtle", "cn-trtl", Algorithm::CN_PICO_0 }, + { "cryptonight-ultralite", "cn-ultralite", Algorithm::CN_PICO_0 }, + { "cryptonight_turtle", "cn_turtle", Algorithm::CN_PICO_0 }, +# endif +# ifdef XMRIG_ALGO_RANDOMX + { "randomx/test", "rx/test", Algorithm::RX_0 }, + { "randomx/0", "rx/0", Algorithm::RX_0 }, + { "randomx/0", "rx/0", Algorithm::RX_0 }, + { "RandomX", "rx", Algorithm::RX_0 }, + { "randomx/wow", "rx/wow", Algorithm::RX_WOW }, + { "RandomWOW", nullptr, Algorithm::RX_WOW }, + { "randomx/loki", "rx/loki", Algorithm::RX_LOKI }, + { "RandomXL", nullptr, Algorithm::RX_LOKI }, +# endif +}; + + +} /* namespace xmrig */ + + +int xmrig::Algorithm::maxIntensity() const +{ +# ifdef XMRIG_ALGO_RANDOMX + if (family() == RANDOM_X) { + return 1; + } +# endif + +# ifdef XMRIG_ALGO_CN_GPU + if (m_id == CN_GPU) { + return 1; + } +# endif + + return 5; +} + + +rapidjson::Value xmrig::Algorithm::toJSON() const +{ + using namespace rapidjson; + + return isValid() ? Value(StringRef(shortName())) : Value(kNullType); +} + + +size_t xmrig::Algorithm::l2() const +{ +# ifdef XMRIG_ALGO_RANDOMX + switch (m_id) { + case RX_0: + case RX_LOKI: + return 0x40000; + + case RX_WOW: + return 0x20000; + + default: + break; + } +# endif + + return 0; +} + + +size_t xmrig::Algorithm::l3() const +{ + const Family f = family(); + assert(f != UNKNOWN); + + if (f < RANDOM_X) { + return CnAlgo<>::memory(m_id); + } + +# ifdef XMRIG_ALGO_RANDOMX + if (f == RANDOM_X) { + constexpr size_t oneMiB = 0x100000; + + switch (m_id) { + case RX_0: + case RX_LOKI: + return oneMiB * 2; + + case RX_WOW: + return oneMiB; + + default: + break; + } + } +# endif + + return 0; +} + + +xmrig::Algorithm::Family xmrig::Algorithm::family(Id id) +{ + switch (id) { + case CN_0: + case CN_1: + case CN_2: + case CN_R: + case CN_WOW: + case CN_FAST: + case CN_HALF: + case CN_XAO: + case CN_RTO: + case CN_RWZ: + case CN_ZLS: + case CN_DOUBLE: +# ifdef XMRIG_ALGO_CN_GPU + case CN_GPU: +# endif + return CN; + +# ifdef XMRIG_ALGO_CN_LITE + case CN_LITE_0: + case CN_LITE_1: + return CN_LITE; +# endif + +# ifdef XMRIG_ALGO_CN_HEAVY + case CN_HEAVY_0: + case CN_HEAVY_TUBE: + case CN_HEAVY_XHV: + return CN_HEAVY; +# endif + +# ifdef XMRIG_ALGO_CN_PICO + case CN_PICO_0: + return CN_PICO; +# endif + +# ifdef XMRIG_ALGO_RANDOMX + case RX_0: + case RX_WOW: + case RX_LOKI: + return RANDOM_X; +# endif + + case INVALID: + case MAX: + return UNKNOWN; + } + + return UNKNOWN; +} + + +xmrig::Algorithm::Id xmrig::Algorithm::parse(const char *name) +{ + if (name == nullptr || strlen(name) < 1) { + return INVALID; + } + + for (size_t i = 0; i < ARRAY_SIZE(algorithm_names); i++) { + if ((strcasecmp(name, algorithm_names[i].name) == 0) || (algorithm_names[i].shortName != nullptr && strcasecmp(name, algorithm_names[i].shortName) == 0)) { + return algorithm_names[i].id; + } + } + + return INVALID; +} + + +const char *xmrig::Algorithm::name(bool shortName) const +{ + for (size_t i = 0; i < ARRAY_SIZE(algorithm_names); i++) { + if (algorithm_names[i].id == m_id) { + return shortName ? algorithm_names[i].shortName : algorithm_names[i].name; + } + } + + return "invalid"; +} diff --git a/src/crypto/common/Algorithm.h b/src/crypto/common/Algorithm.h new file mode 100644 index 000000000..92fcc61e7 --- /dev/null +++ b/src/crypto/common/Algorithm.h @@ -0,0 +1,126 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_ALGORITHM_H +#define XMRIG_ALGORITHM_H + + +#include + + +#include "rapidjson/fwd.h" + + +namespace xmrig { + + +class Algorithm +{ +public: + enum Id : int { + INVALID = -1, + CN_0, // "cn/0" CryptoNight (original). + CN_1, // "cn/1" CryptoNight variant 1 also known as Monero7 and CryptoNightV7. + CN_2, // "cn/2" CryptoNight variant 2. + CN_R, // "cn/r" CryptoNightR (Monero's variant 4). + CN_WOW, // "cn/wow" CryptoNightR (Wownero). + CN_FAST, // "cn/fast" CryptoNight variant 1 with half iterations. + CN_HALF, // "cn/half" CryptoNight variant 2 with half iterations (Masari/Torque). + CN_XAO, // "cn/xao" CryptoNight variant 0 (modified, Alloy only). + CN_RTO, // "cn/rto" CryptoNight variant 1 (modified, Arto only). + CN_RWZ, // "cn/rwz" CryptoNight variant 2 with 3/4 iterations and reversed shuffle operation (Graft). + CN_ZLS, // "cn/zls" CryptoNight variant 2 with 3/4 iterations (Zelerius). + CN_DOUBLE, // "cn/double" CryptoNight variant 2 with double iterations (X-CASH). +# ifdef XMRIG_ALGO_CN_GPU + CN_GPU, // "cn/gpu" CryptoNight-GPU (Ryo). +# endif +# ifdef XMRIG_ALGO_CN_LITE + CN_LITE_0, // "cn-lite/0" CryptoNight-Lite variant 0. + CN_LITE_1, // "cn-lite/1" CryptoNight-Lite variant 1. +# endif +# ifdef XMRIG_ALGO_CN_HEAVY + CN_HEAVY_0, // "cn-heavy/0" CryptoNight-Heavy (4 MB). + CN_HEAVY_TUBE, // "cn-heavy/tube" CryptoNight-Heavy (modified, TUBE only). + CN_HEAVY_XHV, // "cn-heavy/xhv" CryptoNight-Heavy (modified, Haven Protocol only). +# endif +# ifdef XMRIG_ALGO_CN_PICO + CN_PICO_0, // "cn-pico" CryptoNight Turtle (TRTL) +# endif +# ifdef XMRIG_ALGO_RANDOMX + RX_0, // "rx/0" RandomX (reference configuration). + RX_WOW, // "rx/wow" RandomWOW (Wownero). + RX_LOKI, // "rx/loki" RandomXL (Loki). +# endif + MAX + }; + + enum Family : int { + UNKNOWN, + CN, + CN_LITE, + CN_HEAVY, + CN_PICO, + RANDOM_X + }; + + inline Algorithm() {} + inline Algorithm(const char *algo) : m_id(parse(algo)) {} + inline Algorithm(Id id) : m_id(id) {} + + inline bool isEqual(const Algorithm &other) const { return m_id == other.m_id; } + inline bool isValid() const { return m_id != INVALID; } + inline const char *name() const { return name(false); } + inline const char *shortName() const { return name(true); } + inline Family family() const { return family(m_id); } + inline Id id() const { return m_id; } + + inline bool operator!=(Algorithm::Id id) const { return m_id != id; } + inline bool operator!=(const Algorithm &other) const { return !isEqual(other); } + inline bool operator==(Algorithm::Id id) const { return m_id == id; } + inline bool operator==(const Algorithm &other) const { return isEqual(other); } + inline operator Algorithm::Id() const { return m_id; } + + int maxIntensity() const; + rapidjson::Value toJSON() const; + size_t l2() const; + size_t l3() const; + + static Family family(Id id); + static Id parse(const char *name); + +private: + const char *name(bool shortName) const; + + Id m_id = INVALID; +}; + + +typedef std::vector Algorithms; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_ALGORITHM_H */ diff --git a/src/crypto/Asm.cpp b/src/crypto/common/Assembly.cpp similarity index 72% rename from src/crypto/Asm.cpp rename to src/crypto/common/Assembly.cpp index 88812c6c1..44bf0a949 100644 --- a/src/crypto/Asm.cpp +++ b/src/crypto/common/Assembly.cpp @@ -6,7 +6,8 @@ * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , * Copyright 2018 SChernykh - * Copyright 2016-2018 XMRig , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -28,15 +29,17 @@ #ifdef _MSC_VER -# define strncasecmp _strnicmp # define strcasecmp _stricmp #endif -#include "crypto/Asm.h" +#include "crypto/common/Assembly.h" #include "rapidjson/document.h" +namespace xmrig { + + static const char *asmNames[] = { "none", "auto", @@ -46,11 +49,13 @@ static const char *asmNames[] = { }; -xmrig::Assembly xmrig::Asm::parse(const char *assembly, Assembly defaultValue) +} /* namespace xmrig */ + + +xmrig::Assembly::Id xmrig::Assembly::parse(const char *assembly, Id defaultValue) { constexpr size_t const size = sizeof(asmNames) / sizeof((asmNames)[0]); - assert(assembly != nullptr); - assert(ASM_MAX == size); + static_assert(size == MAX, "asmNames size mismatch"); if (assembly == nullptr) { return defaultValue; @@ -58,7 +63,7 @@ xmrig::Assembly xmrig::Asm::parse(const char *assembly, Assembly defaultValue) for (size_t i = 0; i < size; i++) { if (strcasecmp(assembly, asmNames[i]) == 0) { - return static_cast(i); + return static_cast(i); } } @@ -66,10 +71,10 @@ xmrig::Assembly xmrig::Asm::parse(const char *assembly, Assembly defaultValue) } -xmrig::Assembly xmrig::Asm::parse(const rapidjson::Value &value, Assembly defaultValue) +xmrig::Assembly::Id xmrig::Assembly::parse(const rapidjson::Value &value, Id defaultValue) { if (value.IsBool()) { - return parse(value.GetBool()); + return value.GetBool() ? AUTO : NONE; } if (value.IsString()) { @@ -80,23 +85,23 @@ xmrig::Assembly xmrig::Asm::parse(const rapidjson::Value &value, Assembly defaul } -const char *xmrig::Asm::toString(Assembly assembly) +const char *xmrig::Assembly::toString() const { - return asmNames[assembly]; + return asmNames[m_id]; } -rapidjson::Value xmrig::Asm::toJSON(Assembly assembly) +rapidjson::Value xmrig::Assembly::toJSON() const { using namespace rapidjson; - if (assembly == ASM_NONE) { + if (m_id == NONE) { return Value(false); } - if (assembly == ASM_AUTO) { + if (m_id == AUTO) { return Value(true); } - return Value(StringRef(toString(assembly))); + return Value(StringRef(toString())); } diff --git a/src/crypto/common/Assembly.h b/src/crypto/common/Assembly.h new file mode 100644 index 000000000..5ea29e11e --- /dev/null +++ b/src/crypto/common/Assembly.h @@ -0,0 +1,75 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_ASSEMBLY_H +#define XMRIG_ASSEMBLY_H + + +#include "rapidjson/fwd.h" + + +namespace xmrig { + + +class Assembly +{ +public: + enum Id : int { + NONE, + AUTO, + INTEL, + RYZEN, + BULLDOZER, + MAX + }; + + + inline Assembly() {} + inline Assembly(Id id) : m_id(id) {} + inline Assembly(const char *assembly) : m_id(parse(assembly)) {} + inline Assembly(const rapidjson::Value &value) : m_id(parse(value)) {} + + static Id parse(const char *assembly, Id defaultValue = AUTO); + static Id parse(const rapidjson::Value &value, Id defaultValue = AUTO); + + const char *toString() const; + rapidjson::Value toJSON() const; + + inline bool isEqual(const Assembly &other) const { return m_id == other.m_id; } + + inline bool operator!=(Assembly::Id id) const { return m_id != id; } + inline bool operator!=(const Assembly &other) const { return !isEqual(other); } + inline bool operator==(Assembly::Id id) const { return m_id == id; } + inline bool operator==(const Assembly &other) const { return isEqual(other); } + inline operator Assembly::Id() const { return m_id; } + +private: + Id m_id = AUTO; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_ASSEMBLY_H */ diff --git a/src/crypto/common/Nonce.cpp b/src/crypto/common/Nonce.cpp new file mode 100644 index 000000000..151819e0d --- /dev/null +++ b/src/crypto/common/Nonce.cpp @@ -0,0 +1,100 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include + + +#include "crypto/common/Nonce.h" + + +namespace xmrig { + + +std::atomic Nonce::m_paused; +std::atomic Nonce::m_sequence[Nonce::MAX]; +uint32_t Nonce::m_nonces[2] = { 0, 0 }; + + +static std::mutex mutex; +static Nonce nonce; + + +} // namespace xmrig + + +xmrig::Nonce::Nonce() +{ + m_paused = true; + + for (int i = 0; i < MAX; ++i) { + m_sequence[i] = 1; + } +} + + +uint32_t xmrig::Nonce::next(uint8_t index, uint32_t nonce, uint32_t reserveCount, bool nicehash) +{ + uint32_t next; + + std::lock_guard lock(mutex); + + if (nicehash) { + next = (nonce & 0xFF000000) | m_nonces[index]; + } + else { + next = m_nonces[index]; + } + + m_nonces[index] += reserveCount; + + return next; +} + + +void xmrig::Nonce::reset(uint8_t index) +{ + std::lock_guard lock(mutex); + + m_nonces[index] = 0; + touch(); +} + + +void xmrig::Nonce::stop() +{ + pause(false); + + for (int i = 0; i < MAX; ++i) { + m_sequence[i] = 0; + } +} + + +void xmrig::Nonce::touch() +{ + for (int i = 0; i < MAX; ++i) { + m_sequence[i]++; + } +} diff --git a/src/crypto/common/Nonce.h b/src/crypto/common/Nonce.h new file mode 100644 index 000000000..401139fd6 --- /dev/null +++ b/src/crypto/common/Nonce.h @@ -0,0 +1,70 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_NONCE_H +#define XMRIG_NONCE_H + + +#include + + +namespace xmrig { + + +class Nonce +{ +public: + enum Backend { + CPU, + OPENCL, + CUDA, + MAX + }; + + + Nonce(); + + static inline bool isOutdated(Backend backend, uint64_t sequence) { return m_sequence[backend].load(std::memory_order_relaxed) != sequence; } + static inline bool isPaused() { return m_paused.load(std::memory_order_relaxed); } + static inline uint64_t sequence(Backend backend) { return m_sequence[backend].load(std::memory_order_relaxed); } + static inline void pause(bool paused) { m_paused = paused; } + static inline void stop(Backend backend) { m_sequence[backend] = 0; } + static inline void touch(Backend backend) { m_sequence[backend]++; } + + static uint32_t next(uint8_t index, uint32_t nonce, uint32_t reserveCount, bool nicehash); + static void reset(uint8_t index); + static void stop(); + static void touch(); + +private: + static std::atomic m_paused; + static std::atomic m_sequence[MAX]; + static uint32_t m_nonces[2]; +}; + + +} // namespace xmrig + + +#endif /* XMRIG_NONCE_H */ diff --git a/src/crypto/common/VirtualMemory.cpp b/src/crypto/common/VirtualMemory.cpp new file mode 100644 index 000000000..081b6c0fa --- /dev/null +++ b/src/crypto/common/VirtualMemory.cpp @@ -0,0 +1,85 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 SChernykh + * Copyright 2018-2019 tevador + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#ifdef XMRIG_FEATURE_HWLOC +# include +# include "backend/cpu/platform/HwlocCpuInfo.h" +# +# if HWLOC_API_VERSION < 0x00010b00 +# define HWLOC_OBJ_NUMANODE HWLOC_OBJ_NODE +# endif +#endif + + +#include "base/io/log/Log.h" +#include "crypto/common/VirtualMemory.h" + + +uint32_t xmrig::VirtualMemory::bindToNUMANode(int64_t affinity) +{ +# ifdef XMRIG_FEATURE_HWLOC + if (affinity < 0 || !HwlocCpuInfo::has(HwlocCpuInfo::SET_THISTHREAD_MEMBIND)) { + return 0; + } + + hwloc_topology_t topology; + hwloc_topology_init(&topology); + hwloc_topology_load(topology); + + const unsigned puId = static_cast(affinity); + + hwloc_obj_t pu = hwloc_get_pu_obj_by_os_index(topology, puId); + +# if HWLOC_API_VERSION >= 0x20000 + if (pu == nullptr || hwloc_set_membind(topology, pu->nodeset, HWLOC_MEMBIND_BIND, HWLOC_MEMBIND_THREAD | HWLOC_MEMBIND_BYNODESET) < 0) { +# else + if (pu == nullptr || hwloc_set_membind_nodeset(topology, pu->nodeset, HWLOC_MEMBIND_BIND, HWLOC_MEMBIND_THREAD) < 0) { +# endif + LOG_WARN("CPU #%02u warning: \"can't bind memory\"", puId); + } + + uint32_t nodeId = 0; + + if (pu) { + hwloc_obj_t node = nullptr; + + while ((node = hwloc_get_next_obj_by_type(topology, HWLOC_OBJ_NUMANODE, node)) != nullptr) { + if (hwloc_bitmap_intersects(node->cpuset, pu->cpuset)) { + nodeId = node->os_index; + + break; + } + } + } + + hwloc_topology_destroy(topology); + + return nodeId; +# else + return 0; +# endif +} diff --git a/src/crypto/common/VirtualMemory.h b/src/crypto/common/VirtualMemory.h new file mode 100644 index 000000000..ac2f75dd8 --- /dev/null +++ b/src/crypto/common/VirtualMemory.h @@ -0,0 +1,86 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 SChernykh + * Copyright 2018-2019 tevador + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_VIRTUALMEMORY_H +#define XMRIG_VIRTUALMEMORY_H + + +#include +#include +#include + + +namespace xmrig { + + +class VirtualMemory +{ +public: + inline VirtualMemory() {} + VirtualMemory(size_t size, bool hugePages = true, size_t align = 64); + ~VirtualMemory(); + + inline bool isHugePages() const { return m_flags & HUGEPAGES; } + inline size_t size() const { return m_size; } + inline uint8_t *scratchpad() const { return m_scratchpad; } + + inline std::pair hugePages() const + { + return std::pair(isHugePages() ? (align(size()) / 2097152) : 0, align(size()) / 2097152); + } + + static uint32_t bindToNUMANode(int64_t affinity); + static void *allocateExecutableMemory(size_t size); + static void *allocateLargePagesMemory(size_t size); + static void flushInstructionCache(void *p, size_t size); + static void freeLargePagesMemory(void *p, size_t size); + static void init(bool hugePages); + static void protectExecutableMemory(void *p, size_t size); + static void unprotectExecutableMemory(void *p, size_t size); + + static inline bool isHugepagesAvailable() { return (m_globalFlags & HUGEPAGES_AVAILABLE) != 0; } + static inline constexpr size_t align(size_t pos, size_t align = 2097152) { return ((pos - 1) / align + 1) * align; } + +private: + enum Flags { + HUGEPAGES_AVAILABLE = 1, + HUGEPAGES = 2, + LOCK = 4 + }; + + static int m_globalFlags; + + int m_flags = 0; + size_t m_size = 0; + uint8_t *m_scratchpad = nullptr; +}; + + +} /* namespace xmrig */ + + + +#endif /* XMRIG_VIRTUALMEMORY_H */ diff --git a/src/crypto/common/VirtualMemory_unix.cpp b/src/crypto/common/VirtualMemory_unix.cpp new file mode 100644 index 000000000..310a043a0 --- /dev/null +++ b/src/crypto/common/VirtualMemory_unix.cpp @@ -0,0 +1,143 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 SChernykh + * Copyright 2018-2019 tevador + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include +#include + + +#include "crypto/common/portable/mm_malloc.h" +#include "crypto/common/VirtualMemory.h" + + +#if defined(__APPLE__) +# include +#endif + + +int xmrig::VirtualMemory::m_globalFlags = 0; + + +xmrig::VirtualMemory::VirtualMemory(size_t size, bool hugePages, size_t align) : + m_size(VirtualMemory::align(size)) +{ + if (hugePages) { + m_scratchpad = static_cast(allocateLargePagesMemory(m_size)); + if (m_scratchpad) { + m_flags |= HUGEPAGES; + + madvise(m_scratchpad, size, MADV_RANDOM | MADV_WILLNEED); + + if (mlock(m_scratchpad, m_size) == 0) { + m_flags |= LOCK; + } + + return; + } + } + + m_scratchpad = static_cast(_mm_malloc(m_size, align)); +} + + +xmrig::VirtualMemory::~VirtualMemory() +{ + if (!m_scratchpad) { + return; + } + + if (isHugePages()) { + if (m_flags & LOCK) { + munlock(m_scratchpad, m_size); + } + + freeLargePagesMemory(m_scratchpad, m_size); + } + else { + _mm_free(m_scratchpad); + } +} + + + +void *xmrig::VirtualMemory::allocateExecutableMemory(size_t size) +{ +# if defined(__APPLE__) + void *mem = mmap(0, size, PROT_READ | PROT_WRITE | PROT_EXEC, MAP_PRIVATE | MAP_ANON, -1, 0); +# else + void *mem = mmap(0, size, PROT_READ | PROT_WRITE | PROT_EXEC, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0); +# endif + + return mem == MAP_FAILED ? nullptr : mem; +} + + +void *xmrig::VirtualMemory::allocateLargePagesMemory(size_t size) +{ +# if defined(__APPLE__) + void *mem = mmap(0, size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANON, VM_FLAGS_SUPERPAGE_SIZE_2MB, 0); +# elif defined(__FreeBSD__) + void *mem = mmap(0, size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS | MAP_ALIGNED_SUPER | MAP_PREFAULT_READ, -1, 0); +# else + void *mem = mmap(0, size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS | MAP_HUGETLB | MAP_POPULATE, 0, 0); +# endif + + return mem == MAP_FAILED ? nullptr : mem; +} + + +void xmrig::VirtualMemory::flushInstructionCache(void *p, size_t size) +{ +# ifdef HAVE_BUILTIN_CLEAR_CACHE + __builtin___clear_cache(reinterpret_cast(p), reinterpret_cast(p) + size); +# endif +} + + +void xmrig::VirtualMemory::freeLargePagesMemory(void *p, size_t size) +{ + munmap(p, size); +} + + +void xmrig::VirtualMemory::init(bool hugePages) +{ + if (hugePages) { + m_globalFlags = HUGEPAGES | HUGEPAGES_AVAILABLE; + } +} + + +void xmrig::VirtualMemory::protectExecutableMemory(void *p, size_t size) +{ + mprotect(p, size, PROT_READ | PROT_EXEC); +} + + +void xmrig::VirtualMemory::unprotectExecutableMemory(void *p, size_t size) +{ + mprotect(p, size, PROT_WRITE | PROT_EXEC); +} diff --git a/src/Mem_win.cpp b/src/crypto/common/VirtualMemory_win.cpp similarity index 66% rename from src/Mem_win.cpp rename to src/crypto/common/VirtualMemory_win.cpp index 27c1348b0..7bdb6365a 100644 --- a/src/Mem_win.cpp +++ b/src/crypto/common/VirtualMemory_win.cpp @@ -7,6 +7,7 @@ * Copyright 2017-2018 XMR-Stak , * Copyright 2018 Lee Clagett * Copyright 2018-2019 SChernykh + * Copyright 2018-2019 tevador * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify @@ -30,12 +31,9 @@ #include -#include "common/log/Log.h" -#include "common/utils/mm_malloc.h" -#include "common/xmrig.h" -#include "crypto/CryptoNight.h" -#include "crypto/CryptoNight_constants.h" -#include "Mem.h" +#include "base/io/log/Log.h" +#include "crypto/common/portable/mm_malloc.h" +#include "crypto/common/VirtualMemory.h" /***************************************************************** @@ -67,11 +65,11 @@ static BOOL SetLockPagesPrivilege() { tp.PrivilegeCount = 1; tp.Privileges[0].Attributes = SE_PRIVILEGE_ENABLED; - if (LookupPrivilegeValue(NULL, SE_LOCK_MEMORY_NAME, &(tp.Privileges[0].Luid)) != TRUE) { + if (LookupPrivilegeValue(nullptr, SE_LOCK_MEMORY_NAME, &(tp.Privileges[0].Luid)) != TRUE) { return FALSE; } - BOOL rc = AdjustTokenPrivileges(token, FALSE, (PTOKEN_PRIVILEGES) &tp, 0, NULL, NULL); + BOOL rc = AdjustTokenPrivileges(token, FALSE, (PTOKEN_PRIVILEGES) &tp, 0, nullptr, nullptr); if (rc != TRUE || GetLastError() != ERROR_SUCCESS) { return FALSE; } @@ -95,12 +93,12 @@ static LSA_UNICODE_STRING StringToLsaUnicodeString(LPCTSTR string) { static BOOL ObtainLockPagesPrivilege() { HANDLE token; - PTOKEN_USER user = NULL; + PTOKEN_USER user = nullptr; if (OpenProcessToken(GetCurrentProcess(), TOKEN_QUERY, &token) == TRUE) { DWORD size = 0; - GetTokenInformation(token, TokenUser, NULL, 0, &size); + GetTokenInformation(token, TokenUser, nullptr, 0, &size); if (size) { user = (PTOKEN_USER) LocalAlloc(LPTR, size); } @@ -118,7 +116,7 @@ static BOOL ObtainLockPagesPrivilege() { ZeroMemory(&attributes, sizeof(attributes)); BOOL result = FALSE; - if (LsaOpenPolicy(NULL, &attributes, POLICY_ALL_ACCESS, &handle) == 0) { + if (LsaOpenPolicy(nullptr, &attributes, POLICY_ALL_ACCESS, &handle) == 0) { LSA_UNICODE_STRING str = StringToLsaUnicodeString(_T(SE_LOCK_MEMORY_NAME)); if (LsaAddAccountRights(handle, user->User.Sid, &str, 1) == 0) { @@ -143,62 +141,94 @@ static BOOL TrySetLockPagesPrivilege() { } -void Mem::init(bool enabled) -{ - m_enabled = enabled; +int xmrig::VirtualMemory::m_globalFlags = 0; - if (enabled && TrySetLockPagesPrivilege()) { - m_flags |= HugepagesAvailable; + +xmrig::VirtualMemory::VirtualMemory(size_t size, bool hugePages, size_t align) : + m_size(VirtualMemory::align(size)) +{ + if (hugePages) { + m_scratchpad = static_cast(allocateLargePagesMemory(m_size)); + if (m_scratchpad) { + m_flags |= HUGEPAGES; + + return; + } } + + m_scratchpad = static_cast(_mm_malloc(m_size, align)); } -void Mem::allocate(MemInfo &info, bool enabled) +xmrig::VirtualMemory::~VirtualMemory() { - info.hugePages = 0; - - if (!enabled) { - info.memory = static_cast(_mm_malloc(info.size, 4096)); - + if (!m_scratchpad) { return; } - info.memory = static_cast(VirtualAlloc(nullptr, info.size, MEM_COMMIT | MEM_RESERVE | MEM_LARGE_PAGES, PAGE_READWRITE)); - if (info.memory) { - info.hugePages = info.pages; - - return; - } - - allocate(info, false); -} - - -void Mem::release(MemInfo &info) -{ - if (info.hugePages) { - VirtualFree(info.memory, 0, MEM_RELEASE); + if (isHugePages()) { + freeLargePagesMemory(m_scratchpad, m_size); } else { - _mm_free(info.memory); + _mm_free(m_scratchpad); } } -void *Mem::allocateExecutableMemory(size_t size) +void *xmrig::VirtualMemory::allocateExecutableMemory(size_t size) { - return VirtualAlloc(0, size, MEM_COMMIT | MEM_RESERVE, PAGE_EXECUTE_READWRITE); + return VirtualAlloc(nullptr, size, MEM_COMMIT | MEM_RESERVE, PAGE_EXECUTE_READWRITE); } -void Mem::protectExecutableMemory(void *p, size_t size) +void *xmrig::VirtualMemory::allocateLargePagesMemory(size_t size) +{ + const size_t min = GetLargePageMinimum(); + void *mem = nullptr; + + if (min > 0) { + mem = VirtualAlloc(nullptr, align(size, min), MEM_COMMIT | MEM_RESERVE | MEM_LARGE_PAGES, PAGE_READWRITE); + } + + return mem; +} + + +void xmrig::VirtualMemory::flushInstructionCache(void *p, size_t size) +{ + ::FlushInstructionCache(GetCurrentProcess(), p, size); +} + + +void xmrig::VirtualMemory::freeLargePagesMemory(void *p, size_t) +{ + VirtualFree(p, 0, MEM_RELEASE); +} + + +void xmrig::VirtualMemory::init(bool hugePages) +{ + if (!hugePages) { + return; + } + + m_globalFlags = HUGEPAGES; + + if (TrySetLockPagesPrivilege()) { + m_globalFlags |= HUGEPAGES_AVAILABLE; + } +} + + +void xmrig::VirtualMemory::protectExecutableMemory(void *p, size_t size) { DWORD oldProtect; VirtualProtect(p, size, PAGE_EXECUTE_READ, &oldProtect); } -void Mem::flushInstructionCache(void *p, size_t size) +void xmrig::VirtualMemory::unprotectExecutableMemory(void *p, size_t size) { - ::FlushInstructionCache(GetCurrentProcess(), p, size); + DWORD oldProtect; + VirtualProtect(p, size, PAGE_EXECUTE_READWRITE, &oldProtect); } diff --git a/src/common/crypto/keccak.cpp b/src/crypto/common/keccak.cpp similarity index 99% rename from src/common/crypto/keccak.cpp rename to src/crypto/common/keccak.cpp index 0219ce366..132ae0a86 100644 --- a/src/common/crypto/keccak.cpp +++ b/src/crypto/common/keccak.cpp @@ -27,7 +27,7 @@ #include -#include "common/crypto/keccak.h" +#include "crypto/common/keccak.h" #define HASH_DATA_AREA 136 diff --git a/src/common/crypto/keccak.h b/src/crypto/common/keccak.h similarity index 100% rename from src/common/crypto/keccak.h rename to src/crypto/common/keccak.h diff --git a/src/base/tools/Handle.cpp b/src/crypto/common/portable/mm_malloc.h similarity index 57% rename from src/base/tools/Handle.cpp rename to src/crypto/common/portable/mm_malloc.h index d486ab39f..34ca7d48b 100644 --- a/src/base/tools/Handle.cpp +++ b/src/crypto/common/portable/mm_malloc.h @@ -22,58 +22,50 @@ * along with this program. If not, see . */ - -#include +#ifndef XMRIG_MM_MALLOC_PORTABLE_H +#define XMRIG_MM_MALLOC_PORTABLE_H -#include "base/tools/Handle.h" +#if defined(XMRIG_ARM) && !defined(__clang__) +#include -void xmrig::Handle::close(uv_fs_event_t *handle) +#ifndef __cplusplus +extern +#else +extern "C" +#endif +int posix_memalign(void **__memptr, size_t __alignment, size_t __size); + + +static __inline__ void *__attribute__((__always_inline__, __malloc__)) _mm_malloc(size_t __size, size_t __align) { - if (handle) { - uv_fs_event_stop(handle); - close(reinterpret_cast(handle)); + if (__align == 1) { + return malloc(__size); } -} - -void xmrig::Handle::close(uv_getaddrinfo_t *handle) -{ - if (handle) { - uv_cancel(reinterpret_cast(handle)); - close(reinterpret_cast(handle)); + if (!(__align & (__align - 1)) && __align < sizeof(void *)) { + __align = sizeof(void *); } -} - -void xmrig::Handle::close(uv_handle_t *handle) -{ - uv_close(handle, [](uv_handle_t *handle) { delete handle; }); -} - - -void xmrig::Handle::close(uv_signal_t *handle) -{ - if (handle) { - uv_signal_stop(handle); - close(reinterpret_cast(handle)); + void *__mallocedMemory; + if (posix_memalign(&__mallocedMemory, __align, __size)) { + return nullptr; } + + return __mallocedMemory; } -void xmrig::Handle::close(uv_tcp_t *handle) +static __inline__ void __attribute__((__always_inline__)) _mm_free(void *__p) { - if (handle) { - close(reinterpret_cast(handle)); - } + free(__p); } +#elif defined(_WIN32) && !defined(__GNUC__) +# include +#else +# include +#endif -void xmrig::Handle::close(uv_timer_s *handle) -{ - if (handle) { - uv_timer_stop(handle); - close(reinterpret_cast(handle)); - } -} +#endif /* XMRIG_MM_MALLOC_PORTABLE_H */ diff --git a/src/crypto/randomx/aes_hash.cpp b/src/crypto/randomx/aes_hash.cpp new file mode 100644 index 000000000..5d6cb7438 --- /dev/null +++ b/src/crypto/randomx/aes_hash.cpp @@ -0,0 +1,214 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include "soft_aes.h" +#include "randomx.h" + +#define AES_HASH_1R_STATE0 0xd7983aad, 0xcc82db47, 0x9fa856de, 0x92b52c0d +#define AES_HASH_1R_STATE1 0xace78057, 0xf59e125a, 0x15c7b798, 0x338d996e +#define AES_HASH_1R_STATE2 0xe8a07ce4, 0x5079506b, 0xae62c7d0, 0x6a770017 +#define AES_HASH_1R_STATE3 0x7e994948, 0x79a10005, 0x07ad828d, 0x630a240c + +#define AES_HASH_1R_XKEY0 0x06890201, 0x90dc56bf, 0x8b24949f, 0xf6fa8389 +#define AES_HASH_1R_XKEY1 0xed18f99b, 0xee1043c6, 0x51f4e03c, 0x61b263d1 + +/* + Calculate a 512-bit hash of 'input' using 4 lanes of AES. + The input is treated as a set of round keys for the encryption + of the initial state. + + 'inputSize' must be a multiple of 64. + + For a 2 MiB input, this has the same security as 32768-round + AES encryption. + + Hashing throughput: >20 GiB/s per CPU core with hardware AES +*/ +template +void hashAes1Rx4(const void *input, size_t inputSize, void *hash) { + const uint8_t* inptr = (uint8_t*)input; + const uint8_t* inputEnd = inptr + inputSize; + + rx_vec_i128 state0, state1, state2, state3; + rx_vec_i128 in0, in1, in2, in3; + + //intial state + state0 = rx_set_int_vec_i128(AES_HASH_1R_STATE0); + state1 = rx_set_int_vec_i128(AES_HASH_1R_STATE1); + state2 = rx_set_int_vec_i128(AES_HASH_1R_STATE2); + state3 = rx_set_int_vec_i128(AES_HASH_1R_STATE3); + + //process 64 bytes at a time in 4 lanes + while (inptr < inputEnd) { + in0 = rx_load_vec_i128((rx_vec_i128*)inptr + 0); + in1 = rx_load_vec_i128((rx_vec_i128*)inptr + 1); + in2 = rx_load_vec_i128((rx_vec_i128*)inptr + 2); + in3 = rx_load_vec_i128((rx_vec_i128*)inptr + 3); + + state0 = aesenc(state0, in0); + state1 = aesdec(state1, in1); + state2 = aesenc(state2, in2); + state3 = aesdec(state3, in3); + + inptr += 64; + } + + //two extra rounds to achieve full diffusion + rx_vec_i128 xkey0 = rx_set_int_vec_i128(AES_HASH_1R_XKEY0); + rx_vec_i128 xkey1 = rx_set_int_vec_i128(AES_HASH_1R_XKEY1); + + state0 = aesenc(state0, xkey0); + state1 = aesdec(state1, xkey0); + state2 = aesenc(state2, xkey0); + state3 = aesdec(state3, xkey0); + + state0 = aesenc(state0, xkey1); + state1 = aesdec(state1, xkey1); + state2 = aesenc(state2, xkey1); + state3 = aesdec(state3, xkey1); + + //output hash + rx_store_vec_i128((rx_vec_i128*)hash + 0, state0); + rx_store_vec_i128((rx_vec_i128*)hash + 1, state1); + rx_store_vec_i128((rx_vec_i128*)hash + 2, state2); + rx_store_vec_i128((rx_vec_i128*)hash + 3, state3); +} + +template void hashAes1Rx4(const void *input, size_t inputSize, void *hash); +template void hashAes1Rx4(const void *input, size_t inputSize, void *hash); + +#define AES_GEN_1R_KEY0 0xb4f44917, 0xdbb5552b, 0x62716609, 0x6daca553 +#define AES_GEN_1R_KEY1 0x0da1dc4e, 0x1725d378, 0x846a710d, 0x6d7caf07 +#define AES_GEN_1R_KEY2 0x3e20e345, 0xf4c0794f, 0x9f947ec6, 0x3f1262f1 +#define AES_GEN_1R_KEY3 0x49169154, 0x16314c88, 0xb1ba317c, 0x6aef8135 + +/* + Fill 'buffer' with pseudorandom data based on 512-bit 'state'. + The state is encrypted using a single AES round per 16 bytes of output + in 4 lanes. + + 'outputSize' must be a multiple of 64. + + The modified state is written back to 'state' to allow multiple + calls to this function. +*/ +template +void fillAes1Rx4(void *state, size_t outputSize, void *buffer) { + const uint8_t* outptr = (uint8_t*)buffer; + const uint8_t* outputEnd = outptr + outputSize; + + rx_vec_i128 state0, state1, state2, state3; + rx_vec_i128 key0, key1, key2, key3; + + key0 = rx_set_int_vec_i128(AES_GEN_1R_KEY0); + key1 = rx_set_int_vec_i128(AES_GEN_1R_KEY1); + key2 = rx_set_int_vec_i128(AES_GEN_1R_KEY2); + key3 = rx_set_int_vec_i128(AES_GEN_1R_KEY3); + + state0 = rx_load_vec_i128((rx_vec_i128*)state + 0); + state1 = rx_load_vec_i128((rx_vec_i128*)state + 1); + state2 = rx_load_vec_i128((rx_vec_i128*)state + 2); + state3 = rx_load_vec_i128((rx_vec_i128*)state + 3); + + while (outptr < outputEnd) { + state0 = aesdec(state0, key0); + state1 = aesenc(state1, key1); + state2 = aesdec(state2, key2); + state3 = aesenc(state3, key3); + + rx_store_vec_i128((rx_vec_i128*)outptr + 0, state0); + rx_store_vec_i128((rx_vec_i128*)outptr + 1, state1); + rx_store_vec_i128((rx_vec_i128*)outptr + 2, state2); + rx_store_vec_i128((rx_vec_i128*)outptr + 3, state3); + + outptr += 64; + } + + rx_store_vec_i128((rx_vec_i128*)state + 0, state0); + rx_store_vec_i128((rx_vec_i128*)state + 1, state1); + rx_store_vec_i128((rx_vec_i128*)state + 2, state2); + rx_store_vec_i128((rx_vec_i128*)state + 3, state3); +} + +template void fillAes1Rx4(void *state, size_t outputSize, void *buffer); +template void fillAes1Rx4(void *state, size_t outputSize, void *buffer); + +template +void fillAes4Rx4(void *state, size_t outputSize, void *buffer) { + const uint8_t* outptr = (uint8_t*)buffer; + const uint8_t* outputEnd = outptr + outputSize; + + rx_vec_i128 state0, state1, state2, state3; + rx_vec_i128 key0, key1, key2, key3, key4, key5, key6, key7; + + key0 = RandomX_CurrentConfig.fillAes4Rx4_Key[0]; + key1 = RandomX_CurrentConfig.fillAes4Rx4_Key[1]; + key2 = RandomX_CurrentConfig.fillAes4Rx4_Key[2]; + key3 = RandomX_CurrentConfig.fillAes4Rx4_Key[3]; + key4 = RandomX_CurrentConfig.fillAes4Rx4_Key[4]; + key5 = RandomX_CurrentConfig.fillAes4Rx4_Key[5]; + key6 = RandomX_CurrentConfig.fillAes4Rx4_Key[6]; + key7 = RandomX_CurrentConfig.fillAes4Rx4_Key[7]; + + state0 = rx_load_vec_i128((rx_vec_i128*)state + 0); + state1 = rx_load_vec_i128((rx_vec_i128*)state + 1); + state2 = rx_load_vec_i128((rx_vec_i128*)state + 2); + state3 = rx_load_vec_i128((rx_vec_i128*)state + 3); + + while (outptr < outputEnd) { + state0 = aesdec(state0, key0); + state1 = aesenc(state1, key0); + state2 = aesdec(state2, key4); + state3 = aesenc(state3, key4); + + state0 = aesdec(state0, key1); + state1 = aesenc(state1, key1); + state2 = aesdec(state2, key5); + state3 = aesenc(state3, key5); + + state0 = aesdec(state0, key2); + state1 = aesenc(state1, key2); + state2 = aesdec(state2, key6); + state3 = aesenc(state3, key6); + + state0 = aesdec(state0, key3); + state1 = aesenc(state1, key3); + state2 = aesdec(state2, key7); + state3 = aesenc(state3, key7); + + rx_store_vec_i128((rx_vec_i128*)outptr + 0, state0); + rx_store_vec_i128((rx_vec_i128*)outptr + 1, state1); + rx_store_vec_i128((rx_vec_i128*)outptr + 2, state2); + rx_store_vec_i128((rx_vec_i128*)outptr + 3, state3); + + outptr += 64; + } +} + +template void fillAes4Rx4(void *state, size_t outputSize, void *buffer); +template void fillAes4Rx4(void *state, size_t outputSize, void *buffer); diff --git a/src/crypto/randomx/aes_hash.hpp b/src/crypto/randomx/aes_hash.hpp new file mode 100644 index 000000000..b4d0e9405 --- /dev/null +++ b/src/crypto/randomx/aes_hash.hpp @@ -0,0 +1,40 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include + +template +void hashAes1Rx4(const void *input, size_t inputSize, void *hash); + +template +void fillAes1Rx4(void *state, size_t outputSize, void *buffer); + +template +void fillAes4Rx4(void *state, size_t outputSize, void *buffer); diff --git a/src/crypto/randomx/allocator.cpp b/src/crypto/randomx/allocator.cpp new file mode 100644 index 000000000..2ddbed98b --- /dev/null +++ b/src/crypto/randomx/allocator.cpp @@ -0,0 +1,60 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include +#include "allocator.hpp" +#include "intrin_portable.h" +#include "virtual_memory.hpp" +#include "common.hpp" + +namespace randomx { + + template + void* AlignedAllocator::allocMemory(size_t count) { + void *mem = rx_aligned_alloc(count, alignment); + if (mem == nullptr) + throw std::bad_alloc(); + return mem; + } + + template + void AlignedAllocator::freeMemory(void* ptr, size_t count) { + rx_aligned_free(ptr); + } + + template class AlignedAllocator; + + void* LargePageAllocator::allocMemory(size_t count) { + return allocLargePagesMemory(count); + } + + void LargePageAllocator::freeMemory(void* ptr, size_t count) { + freePagedMemory(ptr, count); + }; + +} \ No newline at end of file diff --git a/src/crypto/randomx/allocator.hpp b/src/crypto/randomx/allocator.hpp new file mode 100644 index 000000000..d7aa3f95d --- /dev/null +++ b/src/crypto/randomx/allocator.hpp @@ -0,0 +1,46 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include + +namespace randomx { + + template + struct AlignedAllocator { + static void* allocMemory(size_t); + static void freeMemory(void*, size_t); + }; + + struct LargePageAllocator { + static void* allocMemory(size_t); + static void freeMemory(void*, size_t); + }; + +} \ No newline at end of file diff --git a/src/crypto/randomx/argon2.h b/src/crypto/randomx/argon2.h new file mode 100644 index 000000000..9d4271595 --- /dev/null +++ b/src/crypto/randomx/argon2.h @@ -0,0 +1,229 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +/* Original code from Argon2 reference source code package used under CC0 Licence + * https://github.com/P-H-C/phc-winner-argon2 + * Copyright 2015 + * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves +*/ + +#pragma once + +#include +#include +#include + +/* + * Argon2 input parameter restrictions + */ + + /* Minimum and maximum number of lanes (degree of parallelism) */ +#define ARGON2_MIN_LANES UINT32_C(1) +#define ARGON2_MAX_LANES UINT32_C(0xFFFFFF) + +/* Minimum and maximum number of threads */ +#define ARGON2_MIN_THREADS UINT32_C(1) +#define ARGON2_MAX_THREADS UINT32_C(0xFFFFFF) + +/* Number of synchronization points between lanes per pass */ +#define ARGON2_SYNC_POINTS UINT32_C(4) + +/* Minimum and maximum digest size in bytes */ +#define ARGON2_MIN_OUTLEN UINT32_C(4) +#define ARGON2_MAX_OUTLEN UINT32_C(0xFFFFFFFF) + +/* Minimum and maximum number of memory blocks (each of BLOCK_SIZE bytes) */ +#define ARGON2_MIN_MEMORY (2 * ARGON2_SYNC_POINTS) /* 2 blocks per slice */ + +#define ARGON2_MIN(a, b) ((a) < (b) ? (a) : (b)) +/* Max memory size is addressing-space/2, topping at 2^32 blocks (4 TB) */ +#define ARGON2_MAX_MEMORY_BITS \ + ARGON2_MIN(UINT32_C(32), (sizeof(void *) * CHAR_BIT - 10 - 1)) +#define ARGON2_MAX_MEMORY \ + ARGON2_MIN(UINT32_C(0xFFFFFFFF), UINT64_C(1) << ARGON2_MAX_MEMORY_BITS) + +/* Minimum and maximum number of passes */ +#define ARGON2_MIN_TIME UINT32_C(1) +#define ARGON2_MAX_TIME UINT32_C(0xFFFFFFFF) + +/* Minimum and maximum password length in bytes */ +#define ARGON2_MIN_PWD_LENGTH UINT32_C(0) +#define ARGON2_MAX_PWD_LENGTH UINT32_C(0xFFFFFFFF) + +/* Minimum and maximum associated data length in bytes */ +#define ARGON2_MIN_AD_LENGTH UINT32_C(0) +#define ARGON2_MAX_AD_LENGTH UINT32_C(0xFFFFFFFF) + +/* Minimum and maximum salt length in bytes */ +#define ARGON2_MIN_SALT_LENGTH UINT32_C(8) +#define ARGON2_MAX_SALT_LENGTH UINT32_C(0xFFFFFFFF) + +/* Minimum and maximum key length in bytes */ +#define ARGON2_MIN_SECRET UINT32_C(0) +#define ARGON2_MAX_SECRET UINT32_C(0xFFFFFFFF) + +/* Flags to determine which fields are securely wiped (default = no wipe). */ +#define ARGON2_DEFAULT_FLAGS UINT32_C(0) +#define ARGON2_FLAG_CLEAR_PASSWORD (UINT32_C(1) << 0) +#define ARGON2_FLAG_CLEAR_SECRET (UINT32_C(1) << 1) + + +/* Error codes */ +typedef enum Argon2_ErrorCodes { + ARGON2_OK = 0, + + ARGON2_OUTPUT_PTR_NULL = -1, + + ARGON2_OUTPUT_TOO_SHORT = -2, + ARGON2_OUTPUT_TOO_LONG = -3, + + ARGON2_PWD_TOO_SHORT = -4, + ARGON2_PWD_TOO_LONG = -5, + + ARGON2_SALT_TOO_SHORT = -6, + ARGON2_SALT_TOO_LONG = -7, + + ARGON2_AD_TOO_SHORT = -8, + ARGON2_AD_TOO_LONG = -9, + + ARGON2_SECRET_TOO_SHORT = -10, + ARGON2_SECRET_TOO_LONG = -11, + + ARGON2_TIME_TOO_SMALL = -12, + ARGON2_TIME_TOO_LARGE = -13, + + ARGON2_MEMORY_TOO_LITTLE = -14, + ARGON2_MEMORY_TOO_MUCH = -15, + + ARGON2_LANES_TOO_FEW = -16, + ARGON2_LANES_TOO_MANY = -17, + + ARGON2_PWD_PTR_MISMATCH = -18, /* NULL ptr with non-zero length */ + ARGON2_SALT_PTR_MISMATCH = -19, /* NULL ptr with non-zero length */ + ARGON2_SECRET_PTR_MISMATCH = -20, /* NULL ptr with non-zero length */ + ARGON2_AD_PTR_MISMATCH = -21, /* NULL ptr with non-zero length */ + + ARGON2_MEMORY_ALLOCATION_ERROR = -22, + + ARGON2_FREE_MEMORY_CBK_NULL = -23, + ARGON2_ALLOCATE_MEMORY_CBK_NULL = -24, + + ARGON2_INCORRECT_PARAMETER = -25, + ARGON2_INCORRECT_TYPE = -26, + + ARGON2_OUT_PTR_MISMATCH = -27, + + ARGON2_THREADS_TOO_FEW = -28, + ARGON2_THREADS_TOO_MANY = -29, + + ARGON2_MISSING_ARGS = -30, + + ARGON2_ENCODING_FAIL = -31, + + ARGON2_DECODING_FAIL = -32, + + ARGON2_THREAD_FAIL = -33, + + ARGON2_DECODING_LENGTH_FAIL = -34, + + ARGON2_VERIFY_MISMATCH = -35 +} argon2_error_codes; + +/* Memory allocator types --- for external allocation */ +typedef int(*allocate_fptr)(uint8_t **memory, size_t bytes_to_allocate); +typedef void(*deallocate_fptr)(uint8_t *memory, size_t bytes_to_allocate); + +/* Argon2 external data structures */ + +/* + ***** + * Context: structure to hold Argon2 inputs: + * output array and its length, + * password and its length, + * salt and its length, + * secret and its length, + * associated data and its length, + * number of passes, amount of used memory (in KBytes, can be rounded up a bit) + * number of parallel threads that will be run. + * All the parameters above affect the output hash value. + * Additionally, two function pointers can be provided to allocate and + * deallocate the memory (if NULL, memory will be allocated internally). + * Also, three flags indicate whether to erase password, secret as soon as they + * are pre-hashed (and thus not needed anymore), and the entire memory + ***** + * Simplest situation: you have output array out[8], password is stored in + * pwd[32], salt is stored in salt[16], you do not have keys nor associated + * data. You need to spend 1 GB of RAM and you run 5 passes of Argon2d with + * 4 parallel lanes. + * You want to erase the password, but you're OK with last pass not being + * erased. You want to use the default memory allocator. + * Then you initialize: + Argon2_Context(out,8,pwd,32,salt,16,NULL,0,NULL,0,5,1<<20,4,4,NULL,NULL,true,false,false,false) + */ +typedef struct Argon2_Context { + uint8_t *out; /* output array */ + uint32_t outlen; /* digest length */ + + uint8_t *pwd; /* password array */ + uint32_t pwdlen; /* password length */ + + uint8_t *salt; /* salt array */ + uint32_t saltlen; /* salt length */ + + uint8_t *secret; /* key array */ + uint32_t secretlen; /* key length */ + + uint8_t *ad; /* associated data array */ + uint32_t adlen; /* associated data length */ + + uint32_t t_cost; /* number of passes */ + uint32_t m_cost; /* amount of memory requested (KB) */ + uint32_t lanes; /* number of lanes */ + uint32_t threads; /* maximum number of threads */ + + uint32_t version; /* version number */ + + allocate_fptr allocate_cbk; /* pointer to memory allocator */ + deallocate_fptr free_cbk; /* pointer to memory deallocator */ + + uint32_t flags; /* array of bool options */ +} argon2_context; + +/* Argon2 primitive type */ +typedef enum Argon2_type { + Argon2_d = 0, + Argon2_i = 1, + Argon2_id = 2 +} argon2_type; + +/* Version of the algorithm */ +typedef enum Argon2_version { + ARGON2_VERSION_10 = 0x10, + ARGON2_VERSION_13 = 0x13, + ARGON2_VERSION_NUMBER = ARGON2_VERSION_13 +} argon2_version; diff --git a/src/crypto/randomx/argon2_core.c b/src/crypto/randomx/argon2_core.c new file mode 100644 index 000000000..4b8fa43d6 --- /dev/null +++ b/src/crypto/randomx/argon2_core.c @@ -0,0 +1,515 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +/* Original code from Argon2 reference source code package used under CC0 Licence + * https://github.com/P-H-C/phc-winner-argon2 + * Copyright 2015 + * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves +*/ + + /*For memory wiping*/ +#ifdef _MSC_VER +#include +#include /* For SecureZeroMemory */ +#endif +#if defined __STDC_LIB_EXT1__ +#define __STDC_WANT_LIB_EXT1__ 1 +#endif +#define VC_GE_2005(version) (version >= 1400) + +#include +#include +#include + +#include "argon2_core.h" +#include "blake2/blake2.h" +#include "blake2/blake2-impl.h" + +#ifdef GENKAT +#include "genkat.h" +#endif + +#if defined(__clang__) +#if __has_attribute(optnone) +#define NOT_OPTIMIZED __attribute__((optnone)) +#endif +#elif defined(__GNUC__) +#define GCC_VERSION \ + (__GNUC__ * 10000 + __GNUC_MINOR__ * 100 + __GNUC_PATCHLEVEL__) +#if GCC_VERSION >= 40400 +#define NOT_OPTIMIZED __attribute__((optimize("O0"))) +#endif +#endif +#ifndef NOT_OPTIMIZED +#define NOT_OPTIMIZED +#endif + +/***************Instance and Position constructors**********/ +void rxa2_init_block_value(block *b, uint8_t in) { memset(b->v, in, sizeof(b->v)); } + +void rxa2_copy_block(block *dst, const block *src) { + memcpy(dst->v, src->v, sizeof(uint64_t) * ARGON2_QWORDS_IN_BLOCK); +} + +void rxa2_xor_block(block *dst, const block *src) { + int i; + for (i = 0; i < ARGON2_QWORDS_IN_BLOCK; ++i) { + dst->v[i] ^= src->v[i]; + } +} + +static void load_block(block *dst, const void *input) { + unsigned i; + for (i = 0; i < ARGON2_QWORDS_IN_BLOCK; ++i) { + dst->v[i] = load64((const uint8_t *)input + i * sizeof(dst->v[i])); + } +} + +//static void store_block(void *output, const block *src) { +// unsigned i; +// for (i = 0; i < ARGON2_QWORDS_IN_BLOCK; ++i) { +// store64((uint8_t *)output + i * sizeof(src->v[i]), src->v[i]); +// } +//} + +/***************Memory functions*****************/ + +int rxa2_allocate_memory(const argon2_context *context, uint8_t **memory, + size_t num, size_t size) { + size_t memory_size = num * size; + if (memory == NULL) { + return ARGON2_MEMORY_ALLOCATION_ERROR; + } + + /* 1. Check for multiplication overflow */ + if (size != 0 && memory_size / size != num) { + return ARGON2_MEMORY_ALLOCATION_ERROR; + } + + /* 2. Try to allocate with appropriate allocator */ + if (context->allocate_cbk) { + (context->allocate_cbk)(memory, memory_size); + } + else { + *memory = (uint8_t*)malloc(memory_size); + } + + if (*memory == NULL) { + return ARGON2_MEMORY_ALLOCATION_ERROR; + } + + return ARGON2_OK; +} + +void rxa2_free_memory(const argon2_context *context, uint8_t *memory, + size_t num, size_t size) { + size_t memory_size = num * size; + rxa2_clear_internal_memory(memory, memory_size); + if (context->free_cbk) { + (context->free_cbk)(memory, memory_size); + } + else { + free(memory); + } +} + +void NOT_OPTIMIZED rxa2_secure_wipe_memory(void *v, size_t n) { +#if defined(_MSC_VER) && VC_GE_2005(_MSC_VER) + SecureZeroMemory(v, n); +#elif defined memset_s + memset_s(v, n, 0, n); +#elif defined(__OpenBSD__) + explicit_bzero(v, n); +#else + static void *(*const volatile memset_sec)(void *, int, size_t) = &memset; + memset_sec(v, 0, n); +#endif +} + +/* Memory clear flag defaults to true. */ +#define FLAG_clear_internal_memory 0 +void rxa2_clear_internal_memory(void *v, size_t n) { + if (FLAG_clear_internal_memory && v) { + rxa2_secure_wipe_memory(v, n); + } +} + +uint32_t rxa2_index_alpha(const argon2_instance_t *instance, + const argon2_position_t *position, uint32_t pseudo_rand, + int same_lane) { + /* + * Pass 0: + * This lane : all already finished segments plus already constructed + * blocks in this segment + * Other lanes : all already finished segments + * Pass 1+: + * This lane : (SYNC_POINTS - 1) last segments plus already constructed + * blocks in this segment + * Other lanes : (SYNC_POINTS - 1) last segments + */ + uint32_t reference_area_size; + uint64_t relative_position; + uint32_t start_position, absolute_position; + + if (0 == position->pass) { + /* First pass */ + if (0 == position->slice) { + /* First slice */ + reference_area_size = + position->index - 1; /* all but the previous */ + } + else { + if (same_lane) { + /* The same lane => add current segment */ + reference_area_size = + position->slice * instance->segment_length + + position->index - 1; + } + else { + reference_area_size = + position->slice * instance->segment_length + + ((position->index == 0) ? (-1) : 0); + } + } + } + else { + /* Second pass */ + if (same_lane) { + reference_area_size = instance->lane_length - + instance->segment_length + position->index - + 1; + } + else { + reference_area_size = instance->lane_length - + instance->segment_length + + ((position->index == 0) ? (-1) : 0); + } + } + + /* 1.2.4. Mapping pseudo_rand to 0.. and produce + * relative position */ + relative_position = pseudo_rand; + relative_position = relative_position * relative_position >> 32; + relative_position = reference_area_size - 1 - + (reference_area_size * relative_position >> 32); + + /* 1.2.5 Computing starting position */ + start_position = 0; + + if (0 != position->pass) { + start_position = (position->slice == ARGON2_SYNC_POINTS - 1) + ? 0 + : (position->slice + 1) * instance->segment_length; + } + + /* 1.2.6. Computing absolute position */ + absolute_position = (start_position + relative_position) % + instance->lane_length; /* absolute position */ + return absolute_position; +} + +/* Single-threaded version for p=1 case */ +static int fill_memory_blocks_st(argon2_instance_t *instance) { + uint32_t r, s, l; + + for (r = 0; r < instance->passes; ++r) { + for (s = 0; s < ARGON2_SYNC_POINTS; ++s) { + for (l = 0; l < instance->lanes; ++l) { + argon2_position_t position = { r, l, (uint8_t)s, 0 }; + rxa2_fill_segment(instance, position); + } + } +#ifdef GENKAT + internal_kat(instance, r); /* Print all memory blocks */ +#endif + } + return ARGON2_OK; +} + +int rxa2_fill_memory_blocks(argon2_instance_t *instance) { + if (instance == NULL || instance->lanes == 0) { + return ARGON2_INCORRECT_PARAMETER; + } + return fill_memory_blocks_st(instance); +} + +int rxa2_validate_inputs(const argon2_context *context) { + if (NULL == context) { + return ARGON2_INCORRECT_PARAMETER; + } + + if (NULL == context->out) { + return ARGON2_OUTPUT_PTR_NULL; + } + + /* Validate output length */ + if (ARGON2_MIN_OUTLEN > context->outlen) { + return ARGON2_OUTPUT_TOO_SHORT; + } + + if (ARGON2_MAX_OUTLEN < context->outlen) { + return ARGON2_OUTPUT_TOO_LONG; + } + + /* Validate password (required param) */ + if (NULL == context->pwd) { + if (0 != context->pwdlen) { + return ARGON2_PWD_PTR_MISMATCH; + } + } + + if (ARGON2_MIN_PWD_LENGTH > context->pwdlen) { + return ARGON2_PWD_TOO_SHORT; + } + + if (ARGON2_MAX_PWD_LENGTH < context->pwdlen) { + return ARGON2_PWD_TOO_LONG; + } + + /* Validate salt (required param) */ + if (NULL == context->salt) { + if (0 != context->saltlen) { + return ARGON2_SALT_PTR_MISMATCH; + } + } + + if (ARGON2_MIN_SALT_LENGTH > context->saltlen) { + return ARGON2_SALT_TOO_SHORT; + } + + if (ARGON2_MAX_SALT_LENGTH < context->saltlen) { + return ARGON2_SALT_TOO_LONG; + } + + /* Validate secret (optional param) */ + if (NULL == context->secret) { + if (0 != context->secretlen) { + return ARGON2_SECRET_PTR_MISMATCH; + } + } + else { + if (ARGON2_MIN_SECRET > context->secretlen) { + return ARGON2_SECRET_TOO_SHORT; + } + if (ARGON2_MAX_SECRET < context->secretlen) { + return ARGON2_SECRET_TOO_LONG; + } + } + + /* Validate associated data (optional param) */ + if (NULL == context->ad) { + if (0 != context->adlen) { + return ARGON2_AD_PTR_MISMATCH; + } + } + else { + if (ARGON2_MIN_AD_LENGTH > context->adlen) { + return ARGON2_AD_TOO_SHORT; + } + if (ARGON2_MAX_AD_LENGTH < context->adlen) { + return ARGON2_AD_TOO_LONG; + } + } + + /* Validate memory cost */ + if (ARGON2_MIN_MEMORY > context->m_cost) { + return ARGON2_MEMORY_TOO_LITTLE; + } + + if (ARGON2_MAX_MEMORY < context->m_cost) { + return ARGON2_MEMORY_TOO_MUCH; + } + + if (context->m_cost < 8 * context->lanes) { + return ARGON2_MEMORY_TOO_LITTLE; + } + + /* Validate time cost */ + if (ARGON2_MIN_TIME > context->t_cost) { + return ARGON2_TIME_TOO_SMALL; + } + + if (ARGON2_MAX_TIME < context->t_cost) { + return ARGON2_TIME_TOO_LARGE; + } + + /* Validate lanes */ + if (ARGON2_MIN_LANES > context->lanes) { + return ARGON2_LANES_TOO_FEW; + } + + if (ARGON2_MAX_LANES < context->lanes) { + return ARGON2_LANES_TOO_MANY; + } + + /* Validate threads */ + if (ARGON2_MIN_THREADS > context->threads) { + return ARGON2_THREADS_TOO_FEW; + } + + if (ARGON2_MAX_THREADS < context->threads) { + return ARGON2_THREADS_TOO_MANY; + } + + if (NULL != context->allocate_cbk && NULL == context->free_cbk) { + return ARGON2_FREE_MEMORY_CBK_NULL; + } + + if (NULL == context->allocate_cbk && NULL != context->free_cbk) { + return ARGON2_ALLOCATE_MEMORY_CBK_NULL; + } + + return ARGON2_OK; +} + +void rxa2_fill_first_blocks(uint8_t *blockhash, const argon2_instance_t *instance) { + uint32_t l; + /* Make the first and second block in each lane as G(H0||0||i) or + G(H0||1||i) */ + uint8_t blockhash_bytes[ARGON2_BLOCK_SIZE]; + for (l = 0; l < instance->lanes; ++l) { + + store32(blockhash + ARGON2_PREHASH_DIGEST_LENGTH, 0); + store32(blockhash + ARGON2_PREHASH_DIGEST_LENGTH + 4, l); + rxa2_blake2b_long(blockhash_bytes, ARGON2_BLOCK_SIZE, blockhash, + ARGON2_PREHASH_SEED_LENGTH); + load_block(&instance->memory[l * instance->lane_length + 0], + blockhash_bytes); + + store32(blockhash + ARGON2_PREHASH_DIGEST_LENGTH, 1); + rxa2_blake2b_long(blockhash_bytes, ARGON2_BLOCK_SIZE, blockhash, + ARGON2_PREHASH_SEED_LENGTH); + load_block(&instance->memory[l * instance->lane_length + 1], + blockhash_bytes); + } + rxa2_clear_internal_memory(blockhash_bytes, ARGON2_BLOCK_SIZE); +} + +void rxa2_initial_hash(uint8_t *blockhash, argon2_context *context, argon2_type type) { + blake2b_state BlakeHash; + uint8_t value[sizeof(uint32_t)]; + + if (NULL == context || NULL == blockhash) { + return; + } + + blake2b_init(&BlakeHash, ARGON2_PREHASH_DIGEST_LENGTH); + + store32(&value, context->lanes); + blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value)); + + store32(&value, context->outlen); + blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value)); + + store32(&value, context->m_cost); + blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value)); + + store32(&value, context->t_cost); + blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value)); + + store32(&value, context->version); + blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value)); + + store32(&value, (uint32_t)type); + blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value)); + + store32(&value, context->pwdlen); + blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value)); + + if (context->pwd != NULL) { + blake2b_update(&BlakeHash, (const uint8_t *)context->pwd, + context->pwdlen); + + if (context->flags & ARGON2_FLAG_CLEAR_PASSWORD) { + rxa2_secure_wipe_memory(context->pwd, context->pwdlen); + context->pwdlen = 0; + } + } + + store32(&value, context->saltlen); + blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value)); + + if (context->salt != NULL) { + blake2b_update(&BlakeHash, (const uint8_t *)context->salt, context->saltlen); + } + + store32(&value, context->secretlen); + blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value)); + + if (context->secret != NULL) { + blake2b_update(&BlakeHash, (const uint8_t *)context->secret, + context->secretlen); + + if (context->flags & ARGON2_FLAG_CLEAR_SECRET) { + rxa2_secure_wipe_memory(context->secret, context->secretlen); + context->secretlen = 0; + } + } + + store32(&value, context->adlen); + blake2b_update(&BlakeHash, (const uint8_t *)&value, sizeof(value)); + + if (context->ad != NULL) { + blake2b_update(&BlakeHash, (const uint8_t *)context->ad, + context->adlen); + } + + blake2b_final(&BlakeHash, blockhash, ARGON2_PREHASH_DIGEST_LENGTH); +} + +int rxa2_argon_initialize(argon2_instance_t *instance, argon2_context *context) { + uint8_t blockhash[ARGON2_PREHASH_SEED_LENGTH]; + + if (instance == NULL || context == NULL) + return ARGON2_INCORRECT_PARAMETER; + instance->context_ptr = context; + + /* 1. Memory allocation */ + /*result = allocate_memory(context, (uint8_t **)&(instance->memory), instance->memory_blocks, sizeof(block)); + if (result != ARGON2_OK) { + return result; + }*/ + + /* 2. Initial hashing */ + /* H_0 + 8 extra bytes to produce the first blocks */ + /* uint8_t blockhash[ARGON2_PREHASH_SEED_LENGTH]; */ + /* Hashing all inputs */ + rxa2_initial_hash(blockhash, context, instance->type); + /* Zeroing 8 extra bytes */ + rxa2_clear_internal_memory(blockhash + ARGON2_PREHASH_DIGEST_LENGTH, + ARGON2_PREHASH_SEED_LENGTH - + ARGON2_PREHASH_DIGEST_LENGTH); + + /* 3. Creating first blocks, we always have at least two blocks in a slice + */ + rxa2_fill_first_blocks(blockhash, instance); + /* Clearing the hash */ + rxa2_clear_internal_memory(blockhash, ARGON2_PREHASH_SEED_LENGTH); + + return ARGON2_OK; +} diff --git a/src/crypto/randomx/argon2_core.h b/src/crypto/randomx/argon2_core.h new file mode 100644 index 000000000..efd56d991 --- /dev/null +++ b/src/crypto/randomx/argon2_core.h @@ -0,0 +1,254 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +/* Original code from Argon2 reference source code package used under CC0 Licence + * https://github.com/P-H-C/phc-winner-argon2 + * Copyright 2015 + * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves +*/ + +#ifndef ARGON2_CORE_H +#define ARGON2_CORE_H + +#include +#include "argon2.h" + +#if defined(__cplusplus) +extern "C" { +#endif + +#define CONST_CAST(x) (x)(uintptr_t) + + /**********************Argon2 internal constants*******************************/ + +enum argon2_core_constants { + /* Memory block size in bytes */ + ARGON2_BLOCK_SIZE = 1024, + ARGON2_QWORDS_IN_BLOCK = ARGON2_BLOCK_SIZE / 8, + ARGON2_OWORDS_IN_BLOCK = ARGON2_BLOCK_SIZE / 16, + ARGON2_HWORDS_IN_BLOCK = ARGON2_BLOCK_SIZE / 32, + ARGON2_512BIT_WORDS_IN_BLOCK = ARGON2_BLOCK_SIZE / 64, + + /* Number of pseudo-random values generated by one call to Blake in Argon2i + to + generate reference block positions */ + ARGON2_ADDRESSES_IN_BLOCK = 128, + + /* Pre-hashing digest length and its extension*/ + ARGON2_PREHASH_DIGEST_LENGTH = 64, + ARGON2_PREHASH_SEED_LENGTH = 72 +}; + +/*************************Argon2 internal data types***********************/ + +/* + * Structure for the (1KB) memory block implemented as 128 64-bit words. + * Memory blocks can be copied, XORed. Internal words can be accessed by [] (no + * bounds checking). + */ +typedef struct block_ { uint64_t v[ARGON2_QWORDS_IN_BLOCK]; } block; + +/*****************Functions that work with the block******************/ + +/* Initialize each byte of the block with @in */ +void rxa2_init_block_value(block *b, uint8_t in); + +/* Copy block @src to block @dst */ +void rxa2_copy_block(block *dst, const block *src); + +/* XOR @src onto @dst bytewise */ +void rxa2_xor_block(block *dst, const block *src); + +/* + * Argon2 instance: memory pointer, number of passes, amount of memory, type, + * and derived values. + * Used to evaluate the number and location of blocks to construct in each + * thread + */ +typedef struct Argon2_instance_t { + block *memory; /* Memory pointer */ + uint32_t version; + uint32_t passes; /* Number of passes */ + uint32_t memory_blocks; /* Number of blocks in memory */ + uint32_t segment_length; + uint32_t lane_length; + uint32_t lanes; + uint32_t threads; + argon2_type type; + int print_internals; /* whether to print the memory blocks */ + argon2_context *context_ptr; /* points back to original context */ +} argon2_instance_t; + +/* + * Argon2 position: where we construct the block right now. Used to distribute + * work between threads. + */ +typedef struct Argon2_position_t { + uint32_t pass; + uint32_t lane; + uint8_t slice; + uint32_t index; +} argon2_position_t; + +/*Struct that holds the inputs for thread handling FillSegment*/ +typedef struct Argon2_thread_data { + argon2_instance_t *instance_ptr; + argon2_position_t pos; +} argon2_thread_data; + +/*************************Argon2 core functions********************************/ + +/* Allocates memory to the given pointer, uses the appropriate allocator as + * specified in the context. Total allocated memory is num*size. + * @param context argon2_context which specifies the allocator + * @param memory pointer to the pointer to the memory + * @param size the size in bytes for each element to be allocated + * @param num the number of elements to be allocated + * @return ARGON2_OK if @memory is a valid pointer and memory is allocated + */ +int rxa2_allocate_memory(const argon2_context *context, uint8_t **memory, + size_t num, size_t size); + +/* + * Frees memory at the given pointer, uses the appropriate deallocator as + * specified in the context. Also cleans the memory using clear_internal_memory. + * @param context argon2_context which specifies the deallocator + * @param memory pointer to buffer to be freed + * @param size the size in bytes for each element to be deallocated + * @param num the number of elements to be deallocated + */ +void rxa2_free_memory(const argon2_context *context, uint8_t *memory, + size_t num, size_t size); + +/* Function that securely cleans the memory. This ignores any flags set + * regarding clearing memory. Usually one just calls clear_internal_memory. + * @param mem Pointer to the memory + * @param s Memory size in bytes + */ +void rxa2_secure_wipe_memory(void *v, size_t n); + +/* Function that securely clears the memory if FLAG_clear_internal_memory is + * set. If the flag isn't set, this function does nothing. + * @param mem Pointer to the memory + * @param s Memory size in bytes + */ +void rxa2_clear_internal_memory(void *v, size_t n); + +/* + * Computes absolute position of reference block in the lane following a skewed + * distribution and using a pseudo-random value as input + * @param instance Pointer to the current instance + * @param position Pointer to the current position + * @param pseudo_rand 32-bit pseudo-random value used to determine the position + * @param same_lane Indicates if the block will be taken from the current lane. + * If so we can reference the current segment + * @pre All pointers must be valid + */ +uint32_t rxa2_index_alpha(const argon2_instance_t *instance, + const argon2_position_t *position, uint32_t pseudo_rand, + int same_lane); + +/* + * Function that validates all inputs against predefined restrictions and return + * an error code + * @param context Pointer to current Argon2 context + * @return ARGON2_OK if everything is all right, otherwise one of error codes + * (all defined in + */ +int rxa2_validate_inputs(const argon2_context *context); + +/* + * Hashes all the inputs into @a blockhash[PREHASH_DIGEST_LENGTH], clears + * password and secret if needed + * @param context Pointer to the Argon2 internal structure containing memory + * pointer, and parameters for time and space requirements. + * @param blockhash Buffer for pre-hashing digest + * @param type Argon2 type + * @pre @a blockhash must have at least @a PREHASH_DIGEST_LENGTH bytes + * allocated + */ +void rxa2_initial_hash(uint8_t *blockhash, argon2_context *context, + argon2_type type); + +/* + * Function creates first 2 blocks per lane + * @param instance Pointer to the current instance + * @param blockhash Pointer to the pre-hashing digest + * @pre blockhash must point to @a PREHASH_SEED_LENGTH allocated values + */ +void rxa2_fill_first_blocks(uint8_t *blockhash, const argon2_instance_t *instance); + +/* + * Function allocates memory, hashes the inputs with Blake, and creates first + * two blocks. Returns the pointer to the main memory with 2 blocks per lane + * initialized + * @param context Pointer to the Argon2 internal structure containing memory + * pointer, and parameters for time and space requirements. + * @param instance Current Argon2 instance + * @return Zero if successful, -1 if memory failed to allocate. @context->state + * will be modified if successful. + */ +int rxa2_argon_initialize(argon2_instance_t *instance, argon2_context *context); + +/* + * XORing the last block of each lane, hashing it, making the tag. Deallocates + * the memory. + * @param context Pointer to current Argon2 context (use only the out parameters + * from it) + * @param instance Pointer to current instance of Argon2 + * @pre instance->state must point to necessary amount of memory + * @pre context->out must point to outlen bytes of memory + * @pre if context->free_cbk is not NULL, it should point to a function that + * deallocates memory + */ +void rxa2_finalize(const argon2_context *context, argon2_instance_t *instance); + +/* + * Function that fills the segment using previous segments also from other + * threads + * @param context current context + * @param instance Pointer to the current instance + * @param position Current position + * @pre all block pointers must be valid + */ +void rxa2_fill_segment(const argon2_instance_t *instance, + argon2_position_t position); + +/* + * Function that fills the entire memory t_cost times based on the first two + * blocks in each lane + * @param instance Pointer to the current instance + * @return ARGON2_OK if successful, @context->state + */ +int rxa2_fill_memory_blocks(argon2_instance_t *instance); + +#if defined(__cplusplus) +} +#endif + +#endif diff --git a/src/crypto/randomx/argon2_ref.c b/src/crypto/randomx/argon2_ref.c new file mode 100644 index 000000000..018b985b8 --- /dev/null +++ b/src/crypto/randomx/argon2_ref.c @@ -0,0 +1,214 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +/* Original code from Argon2 reference source code package used under CC0 Licence + * https://github.com/P-H-C/phc-winner-argon2 + * Copyright 2015 + * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves +*/ + +#include +#include +#include + +#include "argon2.h" +#include "argon2_core.h" + +#include "blake2/blamka-round-ref.h" +#include "blake2/blake2-impl.h" +#include "blake2/blake2.h" + + /* + * Function fills a new memory block and optionally XORs the old block over the new one. + * @next_block must be initialized. + * @param prev_block Pointer to the previous block + * @param ref_block Pointer to the reference block + * @param next_block Pointer to the block to be constructed + * @param with_xor Whether to XOR into the new block (1) or just overwrite (0) + * @pre all block pointers must be valid + */ +static void fill_block(const block *prev_block, const block *ref_block, + block *next_block, int with_xor) { + block blockR, block_tmp; + unsigned i; + + rxa2_copy_block(&blockR, ref_block); + rxa2_xor_block(&blockR, prev_block); + rxa2_copy_block(&block_tmp, &blockR); + /* Now blockR = ref_block + prev_block and block_tmp = ref_block + prev_block */ + if (with_xor) { + /* Saving the next block contents for XOR over: */ + rxa2_xor_block(&block_tmp, next_block); + /* Now blockR = ref_block + prev_block and + block_tmp = ref_block + prev_block + next_block */ + } + + /* Apply Blake2 on columns of 64-bit words: (0,1,...,15) , then + (16,17,..31)... finally (112,113,...127) */ + for (i = 0; i < 8; ++i) { + BLAKE2_ROUND_NOMSG( + blockR.v[16 * i], blockR.v[16 * i + 1], blockR.v[16 * i + 2], + blockR.v[16 * i + 3], blockR.v[16 * i + 4], blockR.v[16 * i + 5], + blockR.v[16 * i + 6], blockR.v[16 * i + 7], blockR.v[16 * i + 8], + blockR.v[16 * i + 9], blockR.v[16 * i + 10], blockR.v[16 * i + 11], + blockR.v[16 * i + 12], blockR.v[16 * i + 13], blockR.v[16 * i + 14], + blockR.v[16 * i + 15]); + } + + /* Apply Blake2 on rows of 64-bit words: (0,1,16,17,...112,113), then + (2,3,18,19,...,114,115).. finally (14,15,30,31,...,126,127) */ + for (i = 0; i < 8; i++) { + BLAKE2_ROUND_NOMSG( + blockR.v[2 * i], blockR.v[2 * i + 1], blockR.v[2 * i + 16], + blockR.v[2 * i + 17], blockR.v[2 * i + 32], blockR.v[2 * i + 33], + blockR.v[2 * i + 48], blockR.v[2 * i + 49], blockR.v[2 * i + 64], + blockR.v[2 * i + 65], blockR.v[2 * i + 80], blockR.v[2 * i + 81], + blockR.v[2 * i + 96], blockR.v[2 * i + 97], blockR.v[2 * i + 112], + blockR.v[2 * i + 113]); + } + + rxa2_copy_block(next_block, &block_tmp); + rxa2_xor_block(next_block, &blockR); +} + +static void next_addresses(block *address_block, block *input_block, + const block *zero_block) { + input_block->v[6]++; + fill_block(zero_block, input_block, address_block, 0); + fill_block(zero_block, address_block, address_block, 0); +} + +void rxa2_fill_segment(const argon2_instance_t *instance, + argon2_position_t position) { + block *ref_block = NULL, *curr_block = NULL; + block address_block, input_block, zero_block; + uint64_t pseudo_rand, ref_index, ref_lane; + uint32_t prev_offset, curr_offset; + uint32_t starting_index; + uint32_t i; + int data_independent_addressing; + + if (instance == NULL) { + return; + } + + data_independent_addressing = + (instance->type == Argon2_i) || + (instance->type == Argon2_id && (position.pass == 0) && + (position.slice < ARGON2_SYNC_POINTS / 2)); + + if (data_independent_addressing) { + rxa2_init_block_value(&zero_block, 0); + rxa2_init_block_value(&input_block, 0); + + input_block.v[0] = position.pass; + input_block.v[1] = position.lane; + input_block.v[2] = position.slice; + input_block.v[3] = instance->memory_blocks; + input_block.v[4] = instance->passes; + input_block.v[5] = instance->type; + } + + starting_index = 0; + + if ((0 == position.pass) && (0 == position.slice)) { + starting_index = 2; /* we have already generated the first two blocks */ + + /* Don't forget to generate the first block of addresses: */ + if (data_independent_addressing) { + next_addresses(&address_block, &input_block, &zero_block); + } + } + + /* Offset of the current block */ + curr_offset = position.lane * instance->lane_length + + position.slice * instance->segment_length + starting_index; + + if (0 == curr_offset % instance->lane_length) { + /* Last block in this lane */ + prev_offset = curr_offset + instance->lane_length - 1; + } + else { + /* Previous block */ + prev_offset = curr_offset - 1; + } + + for (i = starting_index; i < instance->segment_length; + ++i, ++curr_offset, ++prev_offset) { + /*1.1 Rotating prev_offset if needed */ + if (curr_offset % instance->lane_length == 1) { + prev_offset = curr_offset - 1; + } + + /* 1.2 Computing the index of the reference block */ + /* 1.2.1 Taking pseudo-random value from the previous block */ + if (data_independent_addressing) { + if (i % ARGON2_ADDRESSES_IN_BLOCK == 0) { + next_addresses(&address_block, &input_block, &zero_block); + } + pseudo_rand = address_block.v[i % ARGON2_ADDRESSES_IN_BLOCK]; + } + else { + pseudo_rand = instance->memory[prev_offset].v[0]; + } + + /* 1.2.2 Computing the lane of the reference block */ + ref_lane = ((pseudo_rand >> 32)) % instance->lanes; + + if ((position.pass == 0) && (position.slice == 0)) { + /* Can not reference other lanes yet */ + ref_lane = position.lane; + } + + /* 1.2.3 Computing the number of possible reference block within the + * lane. + */ + position.index = i; + ref_index = rxa2_index_alpha(instance, &position, pseudo_rand & 0xFFFFFFFF, + ref_lane == position.lane); + + /* 2 Creating a new block */ + ref_block = + instance->memory + instance->lane_length * ref_lane + ref_index; + curr_block = instance->memory + curr_offset; + if (ARGON2_VERSION_10 == instance->version) { + /* version 1.2.1 and earlier: overwrite, not XOR */ + fill_block(instance->memory + prev_offset, ref_block, curr_block, 0); + } + else { + if (0 == position.pass) { + fill_block(instance->memory + prev_offset, ref_block, + curr_block, 0); + } + else { + fill_block(instance->memory + prev_offset, ref_block, + curr_block, 1); + } + } + } +} diff --git a/src/crypto/randomx/asm/program_epilogue_linux.inc b/src/crypto/randomx/asm/program_epilogue_linux.inc new file mode 100644 index 000000000..eaacae547 --- /dev/null +++ b/src/crypto/randomx/asm/program_epilogue_linux.inc @@ -0,0 +1,10 @@ + ;# restore callee-saved registers - System V AMD64 ABI + pop r15 + pop r14 + pop r13 + pop r12 + pop rbp + pop rbx + + ;# program finished + ret 0 \ No newline at end of file diff --git a/src/crypto/randomx/asm/program_epilogue_store.inc b/src/crypto/randomx/asm/program_epilogue_store.inc new file mode 100644 index 000000000..b94fa4d99 --- /dev/null +++ b/src/crypto/randomx/asm/program_epilogue_store.inc @@ -0,0 +1,19 @@ + ;# save VM register values + pop rcx + mov qword ptr [rcx+0], r8 + mov qword ptr [rcx+8], r9 + mov qword ptr [rcx+16], r10 + mov qword ptr [rcx+24], r11 + mov qword ptr [rcx+32], r12 + mov qword ptr [rcx+40], r13 + mov qword ptr [rcx+48], r14 + mov qword ptr [rcx+56], r15 + movdqa xmmword ptr [rcx+64], xmm0 + movdqa xmmword ptr [rcx+80], xmm1 + movdqa xmmword ptr [rcx+96], xmm2 + movdqa xmmword ptr [rcx+112], xmm3 + lea rcx, [rcx+64] + movdqa xmmword ptr [rcx+64], xmm4 + movdqa xmmword ptr [rcx+80], xmm5 + movdqa xmmword ptr [rcx+96], xmm6 + movdqa xmmword ptr [rcx+112], xmm7 \ No newline at end of file diff --git a/src/crypto/randomx/asm/program_epilogue_win64.inc b/src/crypto/randomx/asm/program_epilogue_win64.inc new file mode 100644 index 000000000..8d70a0a3f --- /dev/null +++ b/src/crypto/randomx/asm/program_epilogue_win64.inc @@ -0,0 +1,24 @@ + ;# restore callee-saved registers - Microsoft x64 calling convention + movdqu xmm15, xmmword ptr [rsp] + movdqu xmm14, xmmword ptr [rsp+16] + movdqu xmm13, xmmword ptr [rsp+32] + movdqu xmm12, xmmword ptr [rsp+48] + movdqu xmm11, xmmword ptr [rsp+64] + add rsp, 80 + movdqu xmm10, xmmword ptr [rsp] + movdqu xmm9, xmmword ptr [rsp+16] + movdqu xmm8, xmmword ptr [rsp+32] + movdqu xmm7, xmmword ptr [rsp+48] + movdqu xmm6, xmmword ptr [rsp+64] + add rsp, 80 + pop r15 + pop r14 + pop r13 + pop r12 + pop rsi + pop rdi + pop rbp + pop rbx + + ;# program finished + ret diff --git a/src/crypto/randomx/asm/program_loop_load.inc b/src/crypto/randomx/asm/program_loop_load.inc new file mode 100644 index 000000000..374af66ad --- /dev/null +++ b/src/crypto/randomx/asm/program_loop_load.inc @@ -0,0 +1,32 @@ + mov rdx, rax + and eax, RANDOMX_SCRATCHPAD_MASK + lea rcx, [rsi+rax] + push rcx + xor r8, qword ptr [rcx+0] + xor r9, qword ptr [rcx+8] + xor r10, qword ptr [rcx+16] + xor r11, qword ptr [rcx+24] + xor r12, qword ptr [rcx+32] + xor r13, qword ptr [rcx+40] + xor r14, qword ptr [rcx+48] + xor r15, qword ptr [rcx+56] + ror rdx, 32 + and edx, RANDOMX_SCRATCHPAD_MASK + lea rcx, [rsi+rdx] + push rcx + cvtdq2pd xmm0, qword ptr [rcx+0] + cvtdq2pd xmm1, qword ptr [rcx+8] + cvtdq2pd xmm2, qword ptr [rcx+16] + cvtdq2pd xmm3, qword ptr [rcx+24] + cvtdq2pd xmm4, qword ptr [rcx+32] + cvtdq2pd xmm5, qword ptr [rcx+40] + cvtdq2pd xmm6, qword ptr [rcx+48] + cvtdq2pd xmm7, qword ptr [rcx+56] + andps xmm4, xmm13 + andps xmm5, xmm13 + andps xmm6, xmm13 + andps xmm7, xmm13 + orps xmm4, xmm14 + orps xmm5, xmm14 + orps xmm6, xmm14 + orps xmm7, xmm14 diff --git a/src/crypto/randomx/asm/program_loop_store.inc b/src/crypto/randomx/asm/program_loop_store.inc new file mode 100644 index 000000000..53164cb07 --- /dev/null +++ b/src/crypto/randomx/asm/program_loop_store.inc @@ -0,0 +1,19 @@ + xor eax, eax + pop rcx + mov qword ptr [rcx+0], r8 + mov qword ptr [rcx+8], r9 + mov qword ptr [rcx+16], r10 + mov qword ptr [rcx+24], r11 + mov qword ptr [rcx+32], r12 + mov qword ptr [rcx+40], r13 + mov qword ptr [rcx+48], r14 + mov qword ptr [rcx+56], r15 + pop rcx + xorpd xmm0, xmm4 + xorpd xmm1, xmm5 + xorpd xmm2, xmm6 + xorpd xmm3, xmm7 + movapd xmmword ptr [rcx+0], xmm0 + movapd xmmword ptr [rcx+16], xmm1 + movapd xmmword ptr [rcx+32], xmm2 + movapd xmmword ptr [rcx+48], xmm3 diff --git a/src/crypto/randomx/asm/program_prologue_linux.inc b/src/crypto/randomx/asm/program_prologue_linux.inc new file mode 100644 index 000000000..ffde152cb --- /dev/null +++ b/src/crypto/randomx/asm/program_prologue_linux.inc @@ -0,0 +1,34 @@ + ;# callee-saved registers - System V AMD64 ABI + push rbx + push rbp + push r12 + push r13 + push r14 + push r15 + + ;# function arguments + mov rbx, rcx ;# loop counter + push rdi ;# RegisterFile& registerFile + mov rcx, rdi + mov rbp, qword ptr [rsi] ;# "mx", "ma" + mov rdi, qword ptr [rsi+8] ;# uint8_t* dataset + mov rsi, rdx ;# uint8_t* scratchpad + + mov rax, rbp + + ;# zero integer registers + xor r8, r8 + xor r9, r9 + xor r10, r10 + xor r11, r11 + xor r12, r12 + xor r13, r13 + xor r14, r14 + xor r15, r15 + + ;# load constant registers + lea rcx, [rcx+120] + movapd xmm8, xmmword ptr [rcx+72] + movapd xmm9, xmmword ptr [rcx+88] + movapd xmm10, xmmword ptr [rcx+104] + movapd xmm11, xmmword ptr [rcx+120] diff --git a/src/crypto/randomx/asm/program_prologue_win64.inc b/src/crypto/randomx/asm/program_prologue_win64.inc new file mode 100644 index 000000000..590a98de9 --- /dev/null +++ b/src/crypto/randomx/asm/program_prologue_win64.inc @@ -0,0 +1,47 @@ + ;# callee-saved registers - Microsoft x64 calling convention + push rbx + push rbp + push rdi + push rsi + push r12 + push r13 + push r14 + push r15 + sub rsp, 80 + movdqu xmmword ptr [rsp+64], xmm6 + movdqu xmmword ptr [rsp+48], xmm7 + movdqu xmmword ptr [rsp+32], xmm8 + movdqu xmmword ptr [rsp+16], xmm9 + movdqu xmmword ptr [rsp+0], xmm10 + sub rsp, 80 + movdqu xmmword ptr [rsp+64], xmm11 + movdqu xmmword ptr [rsp+48], xmm12 + movdqu xmmword ptr [rsp+32], xmm13 + movdqu xmmword ptr [rsp+16], xmm14 + movdqu xmmword ptr [rsp+0], xmm15 + + ;# function arguments + push rcx ;# RegisterFile& registerFile + mov rbp, qword ptr [rdx] ;# "mx", "ma" + mov rdi, qword ptr [rdx+8] ;# uint8_t* dataset + mov rsi, r8 ;# uint8_t* scratchpad + mov rbx, r9 ;# loop counter + + mov rax, rbp + + ;# zero integer registers + xor r8, r8 + xor r9, r9 + xor r10, r10 + xor r11, r11 + xor r12, r12 + xor r13, r13 + xor r14, r14 + xor r15, r15 + + ;# load constant registers + lea rcx, [rcx+120] + movapd xmm8, xmmword ptr [rcx+72] + movapd xmm9, xmmword ptr [rcx+88] + movapd xmm10, xmmword ptr [rcx+104] + movapd xmm11, xmmword ptr [rcx+120] diff --git a/src/crypto/randomx/asm/program_read_dataset.inc b/src/crypto/randomx/asm/program_read_dataset.inc new file mode 100644 index 000000000..b81d0c321 --- /dev/null +++ b/src/crypto/randomx/asm/program_read_dataset.inc @@ -0,0 +1,17 @@ + xor rbp, rax ;# modify "mx" + mov edx, ebp ;# edx = mx + and edx, RANDOMX_DATASET_BASE_MASK + prefetchnta byte ptr [rdi+rdx] + ror rbp, 32 ;# swap "ma" and "mx" + mov edx, ebp ;# edx = ma + and edx, RANDOMX_DATASET_BASE_MASK + lea rcx, [rdi+rdx] ;# dataset cache line + xor r8, qword ptr [rcx+0] + xor r9, qword ptr [rcx+8] + xor r10, qword ptr [rcx+16] + xor r11, qword ptr [rcx+24] + xor r12, qword ptr [rcx+32] + xor r13, qword ptr [rcx+40] + xor r14, qword ptr [rcx+48] + xor r15, qword ptr [rcx+56] + \ No newline at end of file diff --git a/src/crypto/randomx/asm/program_read_dataset_sshash_fin.inc b/src/crypto/randomx/asm/program_read_dataset_sshash_fin.inc new file mode 100644 index 000000000..f5a067d2c --- /dev/null +++ b/src/crypto/randomx/asm/program_read_dataset_sshash_fin.inc @@ -0,0 +1,10 @@ + mov rbx, qword ptr [rsp+64] + xor r8, qword ptr [rsp+56] + xor r9, qword ptr [rsp+48] + xor r10, qword ptr [rsp+40] + xor r11, qword ptr [rsp+32] + xor r12, qword ptr [rsp+24] + xor r13, qword ptr [rsp+16] + xor r14, qword ptr [rsp+8] + xor r15, qword ptr [rsp+0] + add rsp, 72 \ No newline at end of file diff --git a/src/crypto/randomx/asm/program_read_dataset_sshash_init.inc b/src/crypto/randomx/asm/program_read_dataset_sshash_init.inc new file mode 100644 index 000000000..6fe9525de --- /dev/null +++ b/src/crypto/randomx/asm/program_read_dataset_sshash_init.inc @@ -0,0 +1,17 @@ + sub rsp, 72 + mov qword ptr [rsp+64], rbx + mov qword ptr [rsp+56], r8 + mov qword ptr [rsp+48], r9 + mov qword ptr [rsp+40], r10 + mov qword ptr [rsp+32], r11 + mov qword ptr [rsp+24], r12 + mov qword ptr [rsp+16], r13 + mov qword ptr [rsp+8], r14 + mov qword ptr [rsp+0], r15 + xor rbp, rax ;# modify "mx" + ror rbp, 32 ;# swap "ma" and "mx" + mov ebx, ebp ;# ecx = ma + and ebx, RANDOMX_DATASET_BASE_MASK + shr ebx, 6 ;# ebx = Dataset block number + ;# add ebx, datasetOffset / 64 + ;# call 32768 \ No newline at end of file diff --git a/src/crypto/randomx/asm/program_sshash_constants.inc b/src/crypto/randomx/asm/program_sshash_constants.inc new file mode 100644 index 000000000..53dc1755c --- /dev/null +++ b/src/crypto/randomx/asm/program_sshash_constants.inc @@ -0,0 +1,24 @@ +r0_mul: + ;#/ 6364136223846793005 + db 45, 127, 149, 76, 45, 244, 81, 88 +r1_add: + ;#/ 9298411001130361340 + db 252, 161, 245, 89, 138, 151, 10, 129 +r2_add: + ;#/ 12065312585734608966 + db 70, 216, 194, 56, 223, 153, 112, 167 +r3_add: + ;#/ 9306329213124626780 + db 92, 73, 34, 191, 28, 185, 38, 129 +r4_add: + ;#/ 5281919268842080866 + db 98, 138, 159, 23, 151, 37, 77, 73 +r5_add: + ;#/ 10536153434571861004 + db 12, 236, 170, 206, 185, 239, 55, 146 +r6_add: + ;#/ 3398623926847679864 + db 120, 45, 230, 108, 116, 86, 42, 47 +r7_add: + ;#/ 9549104520008361294 + db 78, 229, 44, 182, 247, 59, 133, 132 \ No newline at end of file diff --git a/src/crypto/randomx/asm/program_sshash_load.inc b/src/crypto/randomx/asm/program_sshash_load.inc new file mode 100644 index 000000000..535135691 --- /dev/null +++ b/src/crypto/randomx/asm/program_sshash_load.inc @@ -0,0 +1,8 @@ + xor r8, qword ptr [rbx+0] + xor r9, qword ptr [rbx+8] + xor r10, qword ptr [rbx+16] + xor r11, qword ptr [rbx+24] + xor r12, qword ptr [rbx+32] + xor r13, qword ptr [rbx+40] + xor r14, qword ptr [rbx+48] + xor r15, qword ptr [rbx+56] \ No newline at end of file diff --git a/src/crypto/randomx/asm/program_sshash_prefetch.inc b/src/crypto/randomx/asm/program_sshash_prefetch.inc new file mode 100644 index 000000000..26efb5159 --- /dev/null +++ b/src/crypto/randomx/asm/program_sshash_prefetch.inc @@ -0,0 +1,4 @@ + and rbx, RANDOMX_CACHE_MASK + shl rbx, 6 + add rbx, rdi + prefetchnta byte ptr [rbx] \ No newline at end of file diff --git a/src/crypto/randomx/asm/program_xmm_constants.inc b/src/crypto/randomx/asm/program_xmm_constants.inc new file mode 100644 index 000000000..296237a45 --- /dev/null +++ b/src/crypto/randomx/asm/program_xmm_constants.inc @@ -0,0 +1,6 @@ +mantissaMask: + db 255, 255, 255, 255, 255, 255, 255, 0, 255, 255, 255, 255, 255, 255, 255, 0 +exp240: + db 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 +scaleMask: + db 0, 0, 0, 0, 0, 0, 240, 128, 0, 0, 0, 0, 0, 0, 240, 128 \ No newline at end of file diff --git a/src/crypto/randomx/asm/randomx_reciprocal.inc b/src/crypto/randomx/asm/randomx_reciprocal.inc new file mode 100644 index 000000000..e1f22fdc6 --- /dev/null +++ b/src/crypto/randomx/asm/randomx_reciprocal.inc @@ -0,0 +1,7 @@ + mov edx, 1 + mov r8, rcx + xor eax, eax + bsr rcx, rcx + shl rdx, cl + div r8 + ret \ No newline at end of file diff --git a/src/crypto/randomx/blake2/blake2-impl.h b/src/crypto/randomx/blake2/blake2-impl.h new file mode 100644 index 000000000..617f7c8a3 --- /dev/null +++ b/src/crypto/randomx/blake2/blake2-impl.h @@ -0,0 +1,76 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +/* Original code from Argon2 reference source code package used under CC0 Licence + * https://github.com/P-H-C/phc-winner-argon2 + * Copyright 2015 + * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves +*/ + +#ifndef PORTABLE_BLAKE2_IMPL_H +#define PORTABLE_BLAKE2_IMPL_H + +#include + +#include "endian.h" + +static FORCE_INLINE uint64_t load48(const void *src) { + const uint8_t *p = (const uint8_t *)src; + uint64_t w = *p++; + w |= (uint64_t)(*p++) << 8; + w |= (uint64_t)(*p++) << 16; + w |= (uint64_t)(*p++) << 24; + w |= (uint64_t)(*p++) << 32; + w |= (uint64_t)(*p++) << 40; + return w; +} + +static FORCE_INLINE void store48(void *dst, uint64_t w) { + uint8_t *p = (uint8_t *)dst; + *p++ = (uint8_t)w; + w >>= 8; + *p++ = (uint8_t)w; + w >>= 8; + *p++ = (uint8_t)w; + w >>= 8; + *p++ = (uint8_t)w; + w >>= 8; + *p++ = (uint8_t)w; + w >>= 8; + *p++ = (uint8_t)w; +} + +static FORCE_INLINE uint32_t rotr32(const uint32_t w, const unsigned c) { + return (w >> c) | (w << (32 - c)); +} + +static FORCE_INLINE uint64_t rotr64(const uint64_t w, const unsigned c) { + return (w >> c) | (w << (64 - c)); +} + +#endif diff --git a/src/crypto/randomx/blake2/blake2.h b/src/crypto/randomx/blake2/blake2.h new file mode 100644 index 000000000..7ac301ce4 --- /dev/null +++ b/src/crypto/randomx/blake2/blake2.h @@ -0,0 +1,107 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +/* Original code from Argon2 reference source code package used under CC0 Licence + * https://github.com/P-H-C/phc-winner-argon2 + * Copyright 2015 + * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves +*/ + +#ifndef PORTABLE_BLAKE2_H +#define PORTABLE_BLAKE2_H + +#include +#include + +#if defined(__cplusplus) +extern "C" { +#endif + + enum blake2b_constant { + BLAKE2B_BLOCKBYTES = 128, + BLAKE2B_OUTBYTES = 64, + BLAKE2B_KEYBYTES = 64, + BLAKE2B_SALTBYTES = 16, + BLAKE2B_PERSONALBYTES = 16 + }; + +#pragma pack(push, 1) + typedef struct __blake2b_param { + uint8_t digest_length; /* 1 */ + uint8_t key_length; /* 2 */ + uint8_t fanout; /* 3 */ + uint8_t depth; /* 4 */ + uint32_t leaf_length; /* 8 */ + uint64_t node_offset; /* 16 */ + uint8_t node_depth; /* 17 */ + uint8_t inner_length; /* 18 */ + uint8_t reserved[14]; /* 32 */ + uint8_t salt[BLAKE2B_SALTBYTES]; /* 48 */ + uint8_t personal[BLAKE2B_PERSONALBYTES]; /* 64 */ + } blake2b_param; +#pragma pack(pop) + + typedef struct __blake2b_state { + uint64_t h[8]; + uint64_t t[2]; + uint64_t f[2]; + uint8_t buf[BLAKE2B_BLOCKBYTES]; + unsigned buflen; + unsigned outlen; + uint8_t last_node; + } blake2b_state; + + /* Ensure param structs have not been wrongly padded */ + /* Poor man's static_assert */ + enum { + blake2_size_check_0 = 1 / !!(CHAR_BIT == 8), + blake2_size_check_2 = + 1 / !!(sizeof(blake2b_param) == sizeof(uint64_t) * CHAR_BIT) + }; + + /* Streaming API */ + int blake2b_init(blake2b_state *S, size_t outlen); + int blake2b_init_key(blake2b_state *S, size_t outlen, const void *key, + size_t keylen); + int blake2b_init_param(blake2b_state *S, const blake2b_param *P); + int blake2b_update(blake2b_state *S, const void *in, size_t inlen); + int blake2b_final(blake2b_state *S, void *out, size_t outlen); + + /* Simple API */ + int blake2b(void *out, size_t outlen, const void *in, size_t inlen, + const void *key, size_t keylen); + + /* Argon2 Team - Begin Code */ + int rxa2_blake2b_long(void *out, size_t outlen, const void *in, size_t inlen); + /* Argon2 Team - End Code */ + +#if defined(__cplusplus) +} +#endif + +#endif diff --git a/src/crypto/randomx/blake2/blake2b.c b/src/crypto/randomx/blake2/blake2b.c new file mode 100644 index 000000000..5931ee5c2 --- /dev/null +++ b/src/crypto/randomx/blake2/blake2b.c @@ -0,0 +1,409 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +/* Original code from Argon2 reference source code package used under CC0 Licence + * https://github.com/P-H-C/phc-winner-argon2 + * Copyright 2015 + * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves +*/ + +#include +#include +#include + +#include "blake2.h" +#include "blake2-impl.h" + +static const uint64_t blake2b_IV[8] = { + UINT64_C(0x6a09e667f3bcc908), UINT64_C(0xbb67ae8584caa73b), + UINT64_C(0x3c6ef372fe94f82b), UINT64_C(0xa54ff53a5f1d36f1), + UINT64_C(0x510e527fade682d1), UINT64_C(0x9b05688c2b3e6c1f), + UINT64_C(0x1f83d9abfb41bd6b), UINT64_C(0x5be0cd19137e2179) }; + +static const unsigned int blake2b_sigma[12][16] = { + {0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15}, + {14, 10, 4, 8, 9, 15, 13, 6, 1, 12, 0, 2, 11, 7, 5, 3}, + {11, 8, 12, 0, 5, 2, 15, 13, 10, 14, 3, 6, 7, 1, 9, 4}, + {7, 9, 3, 1, 13, 12, 11, 14, 2, 6, 5, 10, 4, 0, 15, 8}, + {9, 0, 5, 7, 2, 4, 10, 15, 14, 1, 11, 12, 6, 8, 3, 13}, + {2, 12, 6, 10, 0, 11, 8, 3, 4, 13, 7, 5, 15, 14, 1, 9}, + {12, 5, 1, 15, 14, 13, 4, 10, 0, 7, 6, 3, 9, 2, 8, 11}, + {13, 11, 7, 14, 12, 1, 3, 9, 5, 0, 15, 4, 8, 6, 2, 10}, + {6, 15, 14, 9, 11, 3, 0, 8, 12, 2, 13, 7, 1, 4, 10, 5}, + {10, 2, 8, 4, 7, 6, 1, 5, 15, 11, 9, 14, 3, 12, 13, 0}, + {0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15}, + {14, 10, 4, 8, 9, 15, 13, 6, 1, 12, 0, 2, 11, 7, 5, 3}, +}; + +static FORCE_INLINE void blake2b_set_lastnode(blake2b_state *S) { + S->f[1] = (uint64_t)-1; +} + +static FORCE_INLINE void blake2b_set_lastblock(blake2b_state *S) { + if (S->last_node) { + blake2b_set_lastnode(S); + } + S->f[0] = (uint64_t)-1; +} + +static FORCE_INLINE void blake2b_increment_counter(blake2b_state *S, + uint64_t inc) { + S->t[0] += inc; + S->t[1] += (S->t[0] < inc); +} + +static FORCE_INLINE void blake2b_invalidate_state(blake2b_state *S) { + //clear_internal_memory(S, sizeof(*S)); /* wipe */ + blake2b_set_lastblock(S); /* invalidate for further use */ +} + +static FORCE_INLINE void blake2b_init0(blake2b_state *S) { + memset(S, 0, sizeof(*S)); + memcpy(S->h, blake2b_IV, sizeof(S->h)); +} + +int blake2b_init_param(blake2b_state *S, const blake2b_param *P) { + const unsigned char *p = (const unsigned char *)P; + unsigned int i; + + if (NULL == P || NULL == S) { + return -1; + } + + blake2b_init0(S); + /* IV XOR Parameter Block */ + for (i = 0; i < 8; ++i) { + S->h[i] ^= load64(&p[i * sizeof(S->h[i])]); + } + S->outlen = P->digest_length; + return 0; +} + +/* Sequential blake2b initialization */ +int blake2b_init(blake2b_state *S, size_t outlen) { + blake2b_param P; + + if (S == NULL) { + return -1; + } + + if ((outlen == 0) || (outlen > BLAKE2B_OUTBYTES)) { + blake2b_invalidate_state(S); + return -1; + } + + /* Setup Parameter Block for unkeyed BLAKE2 */ + P.digest_length = (uint8_t)outlen; + P.key_length = 0; + P.fanout = 1; + P.depth = 1; + P.leaf_length = 0; + P.node_offset = 0; + P.node_depth = 0; + P.inner_length = 0; + memset(P.reserved, 0, sizeof(P.reserved)); + memset(P.salt, 0, sizeof(P.salt)); + memset(P.personal, 0, sizeof(P.personal)); + + return blake2b_init_param(S, &P); +} + +int blake2b_init_key(blake2b_state *S, size_t outlen, const void *key, size_t keylen) { + blake2b_param P; + + if (S == NULL) { + return -1; + } + + if ((outlen == 0) || (outlen > BLAKE2B_OUTBYTES)) { + blake2b_invalidate_state(S); + return -1; + } + + if ((key == 0) || (keylen == 0) || (keylen > BLAKE2B_KEYBYTES)) { + blake2b_invalidate_state(S); + return -1; + } + + /* Setup Parameter Block for keyed BLAKE2 */ + P.digest_length = (uint8_t)outlen; + P.key_length = (uint8_t)keylen; + P.fanout = 1; + P.depth = 1; + P.leaf_length = 0; + P.node_offset = 0; + P.node_depth = 0; + P.inner_length = 0; + memset(P.reserved, 0, sizeof(P.reserved)); + memset(P.salt, 0, sizeof(P.salt)); + memset(P.personal, 0, sizeof(P.personal)); + + if (blake2b_init_param(S, &P) < 0) { + blake2b_invalidate_state(S); + return -1; + } + + { + uint8_t block[BLAKE2B_BLOCKBYTES]; + memset(block, 0, BLAKE2B_BLOCKBYTES); + memcpy(block, key, keylen); + blake2b_update(S, block, BLAKE2B_BLOCKBYTES); + /* Burn the key from stack */ + //clear_internal_memory(block, BLAKE2B_BLOCKBYTES); + } + return 0; +} + +static void blake2b_compress(blake2b_state *S, const uint8_t *block) { + uint64_t m[16]; + uint64_t v[16]; + unsigned int i, r; + + for (i = 0; i < 16; ++i) { + m[i] = load64(block + i * sizeof(m[i])); + } + + for (i = 0; i < 8; ++i) { + v[i] = S->h[i]; + } + + v[8] = blake2b_IV[0]; + v[9] = blake2b_IV[1]; + v[10] = blake2b_IV[2]; + v[11] = blake2b_IV[3]; + v[12] = blake2b_IV[4] ^ S->t[0]; + v[13] = blake2b_IV[5] ^ S->t[1]; + v[14] = blake2b_IV[6] ^ S->f[0]; + v[15] = blake2b_IV[7] ^ S->f[1]; + +#define G(r, i, a, b, c, d) \ + do { \ + a = a + b + m[blake2b_sigma[r][2 * i + 0]]; \ + d = rotr64(d ^ a, 32); \ + c = c + d; \ + b = rotr64(b ^ c, 24); \ + a = a + b + m[blake2b_sigma[r][2 * i + 1]]; \ + d = rotr64(d ^ a, 16); \ + c = c + d; \ + b = rotr64(b ^ c, 63); \ + } while ((void)0, 0) + +#define ROUND(r) \ + do { \ + G(r, 0, v[0], v[4], v[8], v[12]); \ + G(r, 1, v[1], v[5], v[9], v[13]); \ + G(r, 2, v[2], v[6], v[10], v[14]); \ + G(r, 3, v[3], v[7], v[11], v[15]); \ + G(r, 4, v[0], v[5], v[10], v[15]); \ + G(r, 5, v[1], v[6], v[11], v[12]); \ + G(r, 6, v[2], v[7], v[8], v[13]); \ + G(r, 7, v[3], v[4], v[9], v[14]); \ + } while ((void)0, 0) + + for (r = 0; r < 12; ++r) { + ROUND(r); + } + + for (i = 0; i < 8; ++i) { + S->h[i] = S->h[i] ^ v[i] ^ v[i + 8]; + } + +#undef G +#undef ROUND +} + +int blake2b_update(blake2b_state *S, const void *in, size_t inlen) { + const uint8_t *pin = (const uint8_t *)in; + + if (inlen == 0) { + return 0; + } + + /* Sanity check */ + if (S == NULL || in == NULL) { + return -1; + } + + /* Is this a reused state? */ + if (S->f[0] != 0) { + return -1; + } + + if (S->buflen + inlen > BLAKE2B_BLOCKBYTES) { + /* Complete current block */ + size_t left = S->buflen; + size_t fill = BLAKE2B_BLOCKBYTES - left; + memcpy(&S->buf[left], pin, fill); + blake2b_increment_counter(S, BLAKE2B_BLOCKBYTES); + blake2b_compress(S, S->buf); + S->buflen = 0; + inlen -= fill; + pin += fill; + /* Avoid buffer copies when possible */ + while (inlen > BLAKE2B_BLOCKBYTES) { + blake2b_increment_counter(S, BLAKE2B_BLOCKBYTES); + blake2b_compress(S, pin); + inlen -= BLAKE2B_BLOCKBYTES; + pin += BLAKE2B_BLOCKBYTES; + } + } + memcpy(&S->buf[S->buflen], pin, inlen); + S->buflen += (unsigned int)inlen; + return 0; +} + +int blake2b_final(blake2b_state *S, void *out, size_t outlen) { + uint8_t buffer[BLAKE2B_OUTBYTES] = { 0 }; + unsigned int i; + + /* Sanity checks */ + if (S == NULL || out == NULL || outlen < S->outlen) { + return -1; + } + + /* Is this a reused state? */ + if (S->f[0] != 0) { + return -1; + } + + blake2b_increment_counter(S, S->buflen); + blake2b_set_lastblock(S); + memset(&S->buf[S->buflen], 0, BLAKE2B_BLOCKBYTES - S->buflen); /* Padding */ + blake2b_compress(S, S->buf); + + for (i = 0; i < 8; ++i) { /* Output full hash to temp buffer */ + store64(buffer + sizeof(S->h[i]) * i, S->h[i]); + } + + memcpy(out, buffer, S->outlen); + //clear_internal_memory(buffer, sizeof(buffer)); + //clear_internal_memory(S->buf, sizeof(S->buf)); + //clear_internal_memory(S->h, sizeof(S->h)); + return 0; +} + +int blake2b(void *out, size_t outlen, const void *in, size_t inlen, + const void *key, size_t keylen) { + blake2b_state S; + int ret = -1; + + /* Verify parameters */ + if (NULL == in && inlen > 0) { + goto fail; + } + + if (NULL == out || outlen == 0 || outlen > BLAKE2B_OUTBYTES) { + goto fail; + } + + if ((NULL == key && keylen > 0) || keylen > BLAKE2B_KEYBYTES) { + goto fail; + } + + if (keylen > 0) { + if (blake2b_init_key(&S, outlen, key, keylen) < 0) { + goto fail; + } + } + else { + if (blake2b_init(&S, outlen) < 0) { + goto fail; + } + } + + if (blake2b_update(&S, in, inlen) < 0) { + goto fail; + } + ret = blake2b_final(&S, out, outlen); + +fail: + //clear_internal_memory(&S, sizeof(S)); + return ret; +} + +/* Argon2 Team - Begin Code */ +int rxa2_blake2b_long(void *pout, size_t outlen, const void *in, size_t inlen) { + uint8_t *out = (uint8_t *)pout; + blake2b_state blake_state; + uint8_t outlen_bytes[sizeof(uint32_t)] = { 0 }; + int ret = -1; + + if (outlen > UINT32_MAX) { + goto fail; + } + + /* Ensure little-endian byte order! */ + store32(outlen_bytes, (uint32_t)outlen); + +#define TRY(statement) \ + do { \ + ret = statement; \ + if (ret < 0) { \ + goto fail; \ + } \ + } while ((void)0, 0) + + if (outlen <= BLAKE2B_OUTBYTES) { + TRY(blake2b_init(&blake_state, outlen)); + TRY(blake2b_update(&blake_state, outlen_bytes, sizeof(outlen_bytes))); + TRY(blake2b_update(&blake_state, in, inlen)); + TRY(blake2b_final(&blake_state, out, outlen)); + } + else { + uint32_t toproduce; + uint8_t out_buffer[BLAKE2B_OUTBYTES]; + uint8_t in_buffer[BLAKE2B_OUTBYTES]; + TRY(blake2b_init(&blake_state, BLAKE2B_OUTBYTES)); + TRY(blake2b_update(&blake_state, outlen_bytes, sizeof(outlen_bytes))); + TRY(blake2b_update(&blake_state, in, inlen)); + TRY(blake2b_final(&blake_state, out_buffer, BLAKE2B_OUTBYTES)); + memcpy(out, out_buffer, BLAKE2B_OUTBYTES / 2); + out += BLAKE2B_OUTBYTES / 2; + toproduce = (uint32_t)outlen - BLAKE2B_OUTBYTES / 2; + + while (toproduce > BLAKE2B_OUTBYTES) { + memcpy(in_buffer, out_buffer, BLAKE2B_OUTBYTES); + TRY(blake2b(out_buffer, BLAKE2B_OUTBYTES, in_buffer, + BLAKE2B_OUTBYTES, NULL, 0)); + memcpy(out, out_buffer, BLAKE2B_OUTBYTES / 2); + out += BLAKE2B_OUTBYTES / 2; + toproduce -= BLAKE2B_OUTBYTES / 2; + } + + memcpy(in_buffer, out_buffer, BLAKE2B_OUTBYTES); + TRY(blake2b(out_buffer, toproduce, in_buffer, BLAKE2B_OUTBYTES, NULL, + 0)); + memcpy(out, out_buffer, toproduce); + } +fail: + //clear_internal_memory(&blake_state, sizeof(blake_state)); + return ret; +#undef TRY +} +/* Argon2 Team - End Code */ + diff --git a/src/crypto/randomx/blake2/blamka-round-ref.h b/src/crypto/randomx/blake2/blamka-round-ref.h new file mode 100644 index 000000000..f1fb50bf8 --- /dev/null +++ b/src/crypto/randomx/blake2/blamka-round-ref.h @@ -0,0 +1,73 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +/* Original code from Argon2 reference source code package used under CC0 Licence + * https://github.com/P-H-C/phc-winner-argon2 + * Copyright 2015 + * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves +*/ + +#ifndef BLAKE_ROUND_MKA_H +#define BLAKE_ROUND_MKA_H + +#include "blake2.h" +#include "blake2-impl.h" + + /* designed by the Lyra PHC team */ +static FORCE_INLINE uint64_t fBlaMka(uint64_t x, uint64_t y) { + const uint64_t m = UINT64_C(0xFFFFFFFF); + const uint64_t xy = (x & m) * (y & m); + return x + y + 2 * xy; +} + +#define G(a, b, c, d) \ + do { \ + a = fBlaMka(a, b); \ + d = rotr64(d ^ a, 32); \ + c = fBlaMka(c, d); \ + b = rotr64(b ^ c, 24); \ + a = fBlaMka(a, b); \ + d = rotr64(d ^ a, 16); \ + c = fBlaMka(c, d); \ + b = rotr64(b ^ c, 63); \ + } while ((void)0, 0) + +#define BLAKE2_ROUND_NOMSG(v0, v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, \ + v12, v13, v14, v15) \ + do { \ + G(v0, v4, v8, v12); \ + G(v1, v5, v9, v13); \ + G(v2, v6, v10, v14); \ + G(v3, v7, v11, v15); \ + G(v0, v5, v10, v15); \ + G(v1, v6, v11, v12); \ + G(v2, v7, v8, v13); \ + G(v3, v4, v9, v14); \ + } while ((void)0, 0) + +#endif diff --git a/src/crypto/randomx/blake2/endian.h b/src/crypto/randomx/blake2/endian.h new file mode 100644 index 000000000..c7afed261 --- /dev/null +++ b/src/crypto/randomx/blake2/endian.h @@ -0,0 +1,107 @@ +#pragma once +#include +#include + +#if defined(_MSC_VER) +#define FORCE_INLINE __inline +#elif defined(__GNUC__) || defined(__clang__) +#define FORCE_INLINE __inline__ +#else +#define FORCE_INLINE +#endif + + /* Argon2 Team - Begin Code */ + /* + Not an exhaustive list, but should cover the majority of modern platforms + Additionally, the code will always be correct---this is only a performance + tweak. + */ +#if (defined(__BYTE_ORDER__) && \ + (__BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__)) || \ + defined(__LITTLE_ENDIAN__) || defined(__ARMEL__) || defined(__MIPSEL__) || \ + defined(__AARCH64EL__) || defined(__amd64__) || defined(__i386__) || \ + defined(_M_IX86) || defined(_M_X64) || defined(_M_AMD64) || \ + defined(_M_ARM) +#define NATIVE_LITTLE_ENDIAN +#endif + /* Argon2 Team - End Code */ + +static FORCE_INLINE uint32_t load32(const void *src) { +#if defined(NATIVE_LITTLE_ENDIAN) + uint32_t w; + memcpy(&w, src, sizeof w); + return w; +#else + const uint8_t *p = (const uint8_t *)src; + uint32_t w = *p++; + w |= (uint32_t)(*p++) << 8; + w |= (uint32_t)(*p++) << 16; + w |= (uint32_t)(*p++) << 24; + return w; +#endif +} + +static FORCE_INLINE uint64_t load64_native(const void *src) { + uint64_t w; + memcpy(&w, src, sizeof w); + return w; +} + +static FORCE_INLINE uint64_t load64(const void *src) { +#if defined(NATIVE_LITTLE_ENDIAN) + return load64_native(src); +#else + const uint8_t *p = (const uint8_t *)src; + uint64_t w = *p++; + w |= (uint64_t)(*p++) << 8; + w |= (uint64_t)(*p++) << 16; + w |= (uint64_t)(*p++) << 24; + w |= (uint64_t)(*p++) << 32; + w |= (uint64_t)(*p++) << 40; + w |= (uint64_t)(*p++) << 48; + w |= (uint64_t)(*p++) << 56; + return w; +#endif +} + +static FORCE_INLINE void store32(void *dst, uint32_t w) { +#if defined(NATIVE_LITTLE_ENDIAN) + memcpy(dst, &w, sizeof w); +#else + uint8_t *p = (uint8_t *)dst; + *p++ = (uint8_t)w; + w >>= 8; + *p++ = (uint8_t)w; + w >>= 8; + *p++ = (uint8_t)w; + w >>= 8; + *p++ = (uint8_t)w; +#endif +} + +static FORCE_INLINE void store64_native(void *dst, uint64_t w) { + memcpy(dst, &w, sizeof w); +} + +static FORCE_INLINE void store64(void *dst, uint64_t w) { +#if defined(NATIVE_LITTLE_ENDIAN) + store64_native(dst, w); +#else + uint8_t *p = (uint8_t *)dst; + *p++ = (uint8_t)w; + w >>= 8; + *p++ = (uint8_t)w; + w >>= 8; + *p++ = (uint8_t)w; + w >>= 8; + *p++ = (uint8_t)w; + w >>= 8; + *p++ = (uint8_t)w; + w >>= 8; + *p++ = (uint8_t)w; + w >>= 8; + *p++ = (uint8_t)w; + w >>= 8; + *p++ = (uint8_t)w; +#endif +} diff --git a/src/crypto/randomx/blake2_generator.cpp b/src/crypto/randomx/blake2_generator.cpp new file mode 100644 index 000000000..dcf51cecc --- /dev/null +++ b/src/crypto/randomx/blake2_generator.cpp @@ -0,0 +1,62 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include +#include "blake2/blake2.h" +#include "blake2/endian.h" +#include "blake2_generator.hpp" + +namespace randomx { + + constexpr int maxSeedSize = 60; + + Blake2Generator::Blake2Generator(const void* seed, size_t seedSize, int nonce) : dataIndex(sizeof(data)) { + memset(data, 0, sizeof(data)); + memcpy(data, seed, seedSize > maxSeedSize ? maxSeedSize : seedSize); + store32(&data[maxSeedSize], nonce); + } + + uint8_t Blake2Generator::getByte() { + checkData(1); + return data[dataIndex++]; + } + + uint32_t Blake2Generator::getInt32() { + checkData(4); + auto ret = load32(&data[dataIndex]); + dataIndex += 4; + return ret; + } + + void Blake2Generator::checkData(const size_t bytesNeeded) { + if (dataIndex + bytesNeeded > sizeof(data)) { + blake2b(data, sizeof(data), data, sizeof(data), nullptr, 0); + dataIndex = 0; + } + } +} \ No newline at end of file diff --git a/src/crypto/randomx/blake2_generator.hpp b/src/crypto/randomx/blake2_generator.hpp new file mode 100644 index 000000000..b5ac0801c --- /dev/null +++ b/src/crypto/randomx/blake2_generator.hpp @@ -0,0 +1,46 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include + +namespace randomx { + + class Blake2Generator { + public: + Blake2Generator(const void* seed, size_t seedSize, int nonce = 0); + uint8_t getByte(); + uint32_t getInt32(); + private: + void checkData(const size_t); + + uint8_t data[64]; + size_t dataIndex; + }; +} \ No newline at end of file diff --git a/src/crypto/randomx/bytecode_machine.cpp b/src/crypto/randomx/bytecode_machine.cpp new file mode 100644 index 000000000..8447318e7 --- /dev/null +++ b/src/crypto/randomx/bytecode_machine.cpp @@ -0,0 +1,482 @@ +/* +Copyright (c) 2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include "bytecode_machine.hpp" +#include "reciprocal.h" + +namespace randomx { + + const int_reg_t BytecodeMachine::zero = 0; + +#define INSTR_CASE(x) case InstructionType::x: \ + exe_ ## x(ibc, pc, scratchpad, config); \ + break; + + void BytecodeMachine::executeInstruction(RANDOMX_EXE_ARGS) { + switch (ibc.type) + { + INSTR_CASE(IADD_RS) + INSTR_CASE(IADD_M) + INSTR_CASE(ISUB_R) + INSTR_CASE(ISUB_M) + INSTR_CASE(IMUL_R) + INSTR_CASE(IMUL_M) + INSTR_CASE(IMULH_R) + INSTR_CASE(IMULH_M) + INSTR_CASE(ISMULH_R) + INSTR_CASE(ISMULH_M) + INSTR_CASE(INEG_R) + INSTR_CASE(IXOR_R) + INSTR_CASE(IXOR_M) + INSTR_CASE(IROR_R) + INSTR_CASE(IROL_R) + INSTR_CASE(ISWAP_R) + INSTR_CASE(FSWAP_R) + INSTR_CASE(FADD_R) + INSTR_CASE(FADD_M) + INSTR_CASE(FSUB_R) + INSTR_CASE(FSUB_M) + INSTR_CASE(FSCAL_R) + INSTR_CASE(FMUL_R) + INSTR_CASE(FDIV_M) + INSTR_CASE(FSQRT_R) + INSTR_CASE(CBRANCH) + INSTR_CASE(CFROUND) + INSTR_CASE(ISTORE) + + case InstructionType::NOP: + break; + + case InstructionType::IMUL_RCP: //executed as IMUL_R + default: + UNREACHABLE; + } + } + + void BytecodeMachine::compileInstruction(RANDOMX_GEN_ARGS) { + int opcode = instr.opcode; + + if (opcode < RandomX_CurrentConfig.CEIL_IADD_RS) { + auto dst = instr.dst % RegistersCount; + auto src = instr.src % RegistersCount; + ibc.type = InstructionType::IADD_RS; + ibc.idst = &nreg->r[dst]; + if (dst != RegisterNeedsDisplacement) { + ibc.isrc = &nreg->r[src]; + ibc.shift = instr.getModShift(); + ibc.imm = 0; + } + else { + ibc.isrc = &nreg->r[src]; + ibc.shift = instr.getModShift(); + ibc.imm = signExtend2sCompl(instr.getImm32()); + } + registerUsage[dst] = i; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_IADD_M) { + auto dst = instr.dst % RegistersCount; + auto src = instr.src % RegistersCount; + ibc.type = InstructionType::IADD_M; + ibc.idst = &nreg->r[dst]; + ibc.imm = signExtend2sCompl(instr.getImm32()); + if (src != dst) { + ibc.isrc = &nreg->r[src]; + ibc.memMask = (instr.getModMem() ? ScratchpadL1Mask : ScratchpadL2Mask); + } + else { + ibc.isrc = &zero; + ibc.memMask = ScratchpadL3Mask; + } + registerUsage[dst] = i; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_ISUB_R) { + auto dst = instr.dst % RegistersCount; + auto src = instr.src % RegistersCount; + ibc.type = InstructionType::ISUB_R; + ibc.idst = &nreg->r[dst]; + if (src != dst) { + ibc.isrc = &nreg->r[src]; + } + else { + ibc.imm = signExtend2sCompl(instr.getImm32()); + ibc.isrc = &ibc.imm; + } + registerUsage[dst] = i; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_ISUB_M) { + auto dst = instr.dst % RegistersCount; + auto src = instr.src % RegistersCount; + ibc.type = InstructionType::ISUB_M; + ibc.idst = &nreg->r[dst]; + ibc.imm = signExtend2sCompl(instr.getImm32()); + if (src != dst) { + ibc.isrc = &nreg->r[src]; + ibc.memMask = (instr.getModMem() ? ScratchpadL1Mask : ScratchpadL2Mask); + } + else { + ibc.isrc = &zero; + ibc.memMask = ScratchpadL3Mask; + } + registerUsage[dst] = i; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_IMUL_R) { + auto dst = instr.dst % RegistersCount; + auto src = instr.src % RegistersCount; + ibc.type = InstructionType::IMUL_R; + ibc.idst = &nreg->r[dst]; + if (src != dst) { + ibc.isrc = &nreg->r[src]; + } + else { + ibc.imm = signExtend2sCompl(instr.getImm32()); + ibc.isrc = &ibc.imm; + } + registerUsage[dst] = i; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_IMUL_M) { + auto dst = instr.dst % RegistersCount; + auto src = instr.src % RegistersCount; + ibc.type = InstructionType::IMUL_M; + ibc.idst = &nreg->r[dst]; + ibc.imm = signExtend2sCompl(instr.getImm32()); + if (src != dst) { + ibc.isrc = &nreg->r[src]; + ibc.memMask = (instr.getModMem() ? ScratchpadL1Mask : ScratchpadL2Mask); + } + else { + ibc.isrc = &zero; + ibc.memMask = ScratchpadL3Mask; + } + registerUsage[dst] = i; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_IMULH_R) { + auto dst = instr.dst % RegistersCount; + auto src = instr.src % RegistersCount; + ibc.type = InstructionType::IMULH_R; + ibc.idst = &nreg->r[dst]; + ibc.isrc = &nreg->r[src]; + registerUsage[dst] = i; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_IMULH_M) { + auto dst = instr.dst % RegistersCount; + auto src = instr.src % RegistersCount; + ibc.type = InstructionType::IMULH_M; + ibc.idst = &nreg->r[dst]; + ibc.imm = signExtend2sCompl(instr.getImm32()); + if (src != dst) { + ibc.isrc = &nreg->r[src]; + ibc.memMask = (instr.getModMem() ? ScratchpadL1Mask : ScratchpadL2Mask); + } + else { + ibc.isrc = &zero; + ibc.memMask = ScratchpadL3Mask; + } + registerUsage[dst] = i; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_ISMULH_R) { + auto dst = instr.dst % RegistersCount; + auto src = instr.src % RegistersCount; + ibc.type = InstructionType::ISMULH_R; + ibc.idst = &nreg->r[dst]; + ibc.isrc = &nreg->r[src]; + registerUsage[dst] = i; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_ISMULH_M) { + auto dst = instr.dst % RegistersCount; + auto src = instr.src % RegistersCount; + ibc.type = InstructionType::ISMULH_M; + ibc.idst = &nreg->r[dst]; + ibc.imm = signExtend2sCompl(instr.getImm32()); + if (src != dst) { + ibc.isrc = &nreg->r[src]; + ibc.memMask = (instr.getModMem() ? ScratchpadL1Mask : ScratchpadL2Mask); + } + else { + ibc.isrc = &zero; + ibc.memMask = ScratchpadL3Mask; + } + registerUsage[dst] = i; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_IMUL_RCP) { + uint64_t divisor = instr.getImm32(); + if (!isPowerOf2(divisor)) { + auto dst = instr.dst % RegistersCount; + ibc.type = InstructionType::IMUL_R; + ibc.idst = &nreg->r[dst]; + ibc.imm = randomx_reciprocal(divisor); + ibc.isrc = &ibc.imm; + registerUsage[dst] = i; + } + else { + ibc.type = InstructionType::NOP; + } + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_INEG_R) { + auto dst = instr.dst % RegistersCount; + ibc.type = InstructionType::INEG_R; + ibc.idst = &nreg->r[dst]; + registerUsage[dst] = i; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_IXOR_R) { + auto dst = instr.dst % RegistersCount; + auto src = instr.src % RegistersCount; + ibc.type = InstructionType::IXOR_R; + ibc.idst = &nreg->r[dst]; + if (src != dst) { + ibc.isrc = &nreg->r[src]; + } + else { + ibc.imm = signExtend2sCompl(instr.getImm32()); + ibc.isrc = &ibc.imm; + } + registerUsage[dst] = i; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_IXOR_M) { + auto dst = instr.dst % RegistersCount; + auto src = instr.src % RegistersCount; + ibc.type = InstructionType::IXOR_M; + ibc.idst = &nreg->r[dst]; + ibc.imm = signExtend2sCompl(instr.getImm32()); + if (src != dst) { + ibc.isrc = &nreg->r[src]; + ibc.memMask = (instr.getModMem() ? ScratchpadL1Mask : ScratchpadL2Mask); + } + else { + ibc.isrc = &zero; + ibc.memMask = ScratchpadL3Mask; + } + registerUsage[dst] = i; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_IROR_R) { + auto dst = instr.dst % RegistersCount; + auto src = instr.src % RegistersCount; + ibc.type = InstructionType::IROR_R; + ibc.idst = &nreg->r[dst]; + if (src != dst) { + ibc.isrc = &nreg->r[src]; + } + else { + ibc.imm = instr.getImm32(); + ibc.isrc = &ibc.imm; + } + registerUsage[dst] = i; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_IROL_R) { + auto dst = instr.dst % RegistersCount; + auto src = instr.src % RegistersCount; + ibc.type = InstructionType::IROL_R; + ibc.idst = &nreg->r[dst]; + if (src != dst) { + ibc.isrc = &nreg->r[src]; + } + else { + ibc.imm = instr.getImm32(); + ibc.isrc = &ibc.imm; + } + registerUsage[dst] = i; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_ISWAP_R) { + auto dst = instr.dst % RegistersCount; + auto src = instr.src % RegistersCount; + if (src != dst) { + ibc.idst = &nreg->r[dst]; + ibc.isrc = &nreg->r[src]; + ibc.type = InstructionType::ISWAP_R; + registerUsage[dst] = i; + registerUsage[src] = i; + } + else { + ibc.type = InstructionType::NOP; + } + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_FSWAP_R) { + auto dst = instr.dst % RegistersCount; + ibc.type = InstructionType::FSWAP_R; + if (dst < RegisterCountFlt) + ibc.fdst = &nreg->f[dst]; + else + ibc.fdst = &nreg->e[dst - RegisterCountFlt]; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_FADD_R) { + auto dst = instr.dst % RegisterCountFlt; + auto src = instr.src % RegisterCountFlt; + ibc.type = InstructionType::FADD_R; + ibc.fdst = &nreg->f[dst]; + ibc.fsrc = &nreg->a[src]; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_FADD_M) { + auto dst = instr.dst % RegisterCountFlt; + auto src = instr.src % RegistersCount; + ibc.type = InstructionType::FADD_M; + ibc.fdst = &nreg->f[dst]; + ibc.isrc = &nreg->r[src]; + ibc.memMask = (instr.getModMem() ? ScratchpadL1Mask : ScratchpadL2Mask); + ibc.imm = signExtend2sCompl(instr.getImm32()); + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_FSUB_R) { + auto dst = instr.dst % RegisterCountFlt; + auto src = instr.src % RegisterCountFlt; + ibc.type = InstructionType::FSUB_R; + ibc.fdst = &nreg->f[dst]; + ibc.fsrc = &nreg->a[src]; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_FSUB_M) { + auto dst = instr.dst % RegisterCountFlt; + auto src = instr.src % RegistersCount; + ibc.type = InstructionType::FSUB_M; + ibc.fdst = &nreg->f[dst]; + ibc.isrc = &nreg->r[src]; + ibc.memMask = (instr.getModMem() ? ScratchpadL1Mask : ScratchpadL2Mask); + ibc.imm = signExtend2sCompl(instr.getImm32()); + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_FSCAL_R) { + auto dst = instr.dst % RegisterCountFlt; + ibc.fdst = &nreg->f[dst]; + ibc.type = InstructionType::FSCAL_R; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_FMUL_R) { + auto dst = instr.dst % RegisterCountFlt; + auto src = instr.src % RegisterCountFlt; + ibc.type = InstructionType::FMUL_R; + ibc.fdst = &nreg->e[dst]; + ibc.fsrc = &nreg->a[src]; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_FDIV_M) { + auto dst = instr.dst % RegisterCountFlt; + auto src = instr.src % RegistersCount; + ibc.type = InstructionType::FDIV_M; + ibc.fdst = &nreg->e[dst]; + ibc.isrc = &nreg->r[src]; + ibc.memMask = (instr.getModMem() ? ScratchpadL1Mask : ScratchpadL2Mask); + ibc.imm = signExtend2sCompl(instr.getImm32()); + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_FSQRT_R) { + auto dst = instr.dst % RegisterCountFlt; + ibc.type = InstructionType::FSQRT_R; + ibc.fdst = &nreg->e[dst]; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_CBRANCH) { + ibc.type = InstructionType::CBRANCH; + //jump condition + int creg = instr.dst % RegistersCount; + ibc.idst = &nreg->r[creg]; + ibc.target = registerUsage[creg]; + int shift = instr.getModCond() + RandomX_CurrentConfig.JumpOffset; + ibc.imm = signExtend2sCompl(instr.getImm32()) | (1ULL << shift); + if (RandomX_CurrentConfig.JumpOffset > 0 || shift > 0) //clear the bit below the condition mask - this limits the number of successive jumps to 2 + ibc.imm &= ~(1ULL << (shift - 1)); + ibc.memMask = RandomX_CurrentConfig.ConditionMask_Calculated << shift; + //mark all registers as used + for (unsigned j = 0; j < RegistersCount; ++j) { + registerUsage[j] = i; + } + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_CFROUND) { + auto src = instr.src % RegistersCount; + ibc.isrc = &nreg->r[src]; + ibc.type = InstructionType::CFROUND; + ibc.imm = instr.getImm32() & 63; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_ISTORE) { + auto dst = instr.dst % RegistersCount; + auto src = instr.src % RegistersCount; + ibc.type = InstructionType::ISTORE; + ibc.idst = &nreg->r[dst]; + ibc.isrc = &nreg->r[src]; + ibc.imm = signExtend2sCompl(instr.getImm32()); + if (instr.getModCond() < StoreL3Condition) + ibc.memMask = (instr.getModMem() ? ScratchpadL1Mask : ScratchpadL2Mask); + else + ibc.memMask = ScratchpadL3Mask; + return; + } + + if (opcode < RandomX_CurrentConfig.CEIL_NOP) { + ibc.type = InstructionType::NOP; + return; + } + + UNREACHABLE; + } +} diff --git a/src/crypto/randomx/bytecode_machine.hpp b/src/crypto/randomx/bytecode_machine.hpp new file mode 100644 index 000000000..810f854aa --- /dev/null +++ b/src/crypto/randomx/bytecode_machine.hpp @@ -0,0 +1,288 @@ +/* +Copyright (c) 2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include "common.hpp" +#include "intrin_portable.h" +#include "instruction.hpp" +#include "program.hpp" + +namespace randomx { + + //register file in machine byte order + struct NativeRegisterFile { + int_reg_t r[RegistersCount] = { 0 }; + rx_vec_f128 f[RegisterCountFlt]; + rx_vec_f128 e[RegisterCountFlt]; + rx_vec_f128 a[RegisterCountFlt]; + }; + + struct InstructionByteCode { + union { + int_reg_t* idst; + rx_vec_f128* fdst; + }; + union { + const int_reg_t* isrc; + const rx_vec_f128* fsrc; + }; + union { + uint64_t imm; + int64_t simm; + }; + InstructionType type; + union { + int16_t target; + uint16_t shift; + }; + uint32_t memMask; + }; + +#define RANDOMX_EXE_ARGS InstructionByteCode& ibc, int& pc, uint8_t* scratchpad, ProgramConfiguration& config +#define RANDOMX_GEN_ARGS Instruction& instr, int i, InstructionByteCode& ibc + + class BytecodeMachine; + + typedef void(BytecodeMachine::*InstructionGenBytecode)(RANDOMX_GEN_ARGS); + + class BytecodeMachine { + public: + void beginCompilation(NativeRegisterFile& regFile) { + for (unsigned i = 0; i < RegistersCount; ++i) { + registerUsage[i] = -1; + } + nreg = ®File; + } + + void compileProgram(Program& program, InstructionByteCode* bytecode, NativeRegisterFile& regFile) { + beginCompilation(regFile); + for (unsigned i = 0; i < RandomX_CurrentConfig.ProgramSize; ++i) { + auto& instr = program(i); + auto& ibc = bytecode[i]; + compileInstruction(instr, i, ibc); + } + } + + static void executeBytecode(InstructionByteCode* bytecode, uint8_t* scratchpad, ProgramConfiguration& config) { + for (int pc = 0; pc < static_cast(RandomX_CurrentConfig.ProgramSize); ++pc) { + auto& ibc = bytecode[pc]; + executeInstruction(ibc, pc, scratchpad, config); + } + } + + void compileInstruction(RANDOMX_GEN_ARGS) +#ifdef RANDOMX_GEN_TABLE + { + auto generator = genTable[instr.opcode]; + (this->*generator)(instr, i, ibc); + } +#else + ; +#endif + + static void executeInstruction(RANDOMX_EXE_ARGS); + + static void exe_IADD_RS(RANDOMX_EXE_ARGS) { + *ibc.idst += (*ibc.isrc << ibc.shift) + ibc.imm; + } + + static void exe_IADD_M(RANDOMX_EXE_ARGS) { + *ibc.idst += load64(getScratchpadAddress(ibc, scratchpad)); + } + + static void exe_ISUB_R(RANDOMX_EXE_ARGS) { + *ibc.idst -= *ibc.isrc; + } + + static void exe_ISUB_M(RANDOMX_EXE_ARGS) { + *ibc.idst -= load64(getScratchpadAddress(ibc, scratchpad)); + } + + static void exe_IMUL_R(RANDOMX_EXE_ARGS) { + *ibc.idst *= *ibc.isrc; + } + + static void exe_IMUL_M(RANDOMX_EXE_ARGS) { + *ibc.idst *= load64(getScratchpadAddress(ibc, scratchpad)); + } + + static void exe_IMULH_R(RANDOMX_EXE_ARGS) { + *ibc.idst = mulh(*ibc.idst, *ibc.isrc); + } + + static void exe_IMULH_M(RANDOMX_EXE_ARGS) { + *ibc.idst = mulh(*ibc.idst, load64(getScratchpadAddress(ibc, scratchpad))); + } + + static void exe_ISMULH_R(RANDOMX_EXE_ARGS) { + *ibc.idst = smulh(unsigned64ToSigned2sCompl(*ibc.idst), unsigned64ToSigned2sCompl(*ibc.isrc)); + } + + static void exe_ISMULH_M(RANDOMX_EXE_ARGS) { + *ibc.idst = smulh(unsigned64ToSigned2sCompl(*ibc.idst), unsigned64ToSigned2sCompl(load64(getScratchpadAddress(ibc, scratchpad)))); + } + + static void exe_INEG_R(RANDOMX_EXE_ARGS) { + *ibc.idst = ~(*ibc.idst) + 1; //two's complement negative + } + + static void exe_IXOR_R(RANDOMX_EXE_ARGS) { + *ibc.idst ^= *ibc.isrc; + } + + static void exe_IXOR_M(RANDOMX_EXE_ARGS) { + *ibc.idst ^= load64(getScratchpadAddress(ibc, scratchpad)); + } + + static void exe_IROR_R(RANDOMX_EXE_ARGS) { + *ibc.idst = rotr64(*ibc.idst, *ibc.isrc & 63); + } + + static void exe_IROL_R(RANDOMX_EXE_ARGS) { + *ibc.idst = rotl64(*ibc.idst, *ibc.isrc & 63); + } + + static void exe_ISWAP_R(RANDOMX_EXE_ARGS) { + int_reg_t temp = *ibc.isrc; + *(int_reg_t*)ibc.isrc = *ibc.idst; + *ibc.idst = temp; + } + + static void exe_FSWAP_R(RANDOMX_EXE_ARGS) { + *ibc.fdst = rx_swap_vec_f128(*ibc.fdst); + } + + static void exe_FADD_R(RANDOMX_EXE_ARGS) { + *ibc.fdst = rx_add_vec_f128(*ibc.fdst, *ibc.fsrc); + } + + static void exe_FADD_M(RANDOMX_EXE_ARGS) { + rx_vec_f128 fsrc = rx_cvt_packed_int_vec_f128(getScratchpadAddress(ibc, scratchpad)); + *ibc.fdst = rx_add_vec_f128(*ibc.fdst, fsrc); + } + + static void exe_FSUB_R(RANDOMX_EXE_ARGS) { + *ibc.fdst = rx_sub_vec_f128(*ibc.fdst, *ibc.fsrc); + } + + static void exe_FSUB_M(RANDOMX_EXE_ARGS) { + rx_vec_f128 fsrc = rx_cvt_packed_int_vec_f128(getScratchpadAddress(ibc, scratchpad)); + *ibc.fdst = rx_sub_vec_f128(*ibc.fdst, fsrc); + } + + static void exe_FSCAL_R(RANDOMX_EXE_ARGS) { + const rx_vec_f128 mask = rx_set1_vec_f128(0x80F0000000000000); + *ibc.fdst = rx_xor_vec_f128(*ibc.fdst, mask); + } + + static void exe_FMUL_R(RANDOMX_EXE_ARGS) { + *ibc.fdst = rx_mul_vec_f128(*ibc.fdst, *ibc.fsrc); + } + + static void exe_FDIV_M(RANDOMX_EXE_ARGS) { + rx_vec_f128 fsrc = maskRegisterExponentMantissa( + config, + rx_cvt_packed_int_vec_f128(getScratchpadAddress(ibc, scratchpad)) + ); + *ibc.fdst = rx_div_vec_f128(*ibc.fdst, fsrc); + } + + static void exe_FSQRT_R(RANDOMX_EXE_ARGS) { + *ibc.fdst = rx_sqrt_vec_f128(*ibc.fdst); + } + + static void exe_CBRANCH(RANDOMX_EXE_ARGS) { + *ibc.idst += ibc.imm; + if ((*ibc.idst & ibc.memMask) == 0) { + pc = ibc.target; + } + } + + static void exe_CFROUND(RANDOMX_EXE_ARGS) { + rx_set_rounding_mode(rotr64(*ibc.isrc, ibc.imm) % 4); + } + + static void exe_ISTORE(RANDOMX_EXE_ARGS) { + store64(scratchpad + ((*ibc.idst + ibc.imm) & ibc.memMask), *ibc.isrc); + } + protected: + static rx_vec_f128 maskRegisterExponentMantissa(ProgramConfiguration& config, rx_vec_f128 x) { + const rx_vec_f128 xmantissaMask = rx_set_vec_f128(dynamicMantissaMask, dynamicMantissaMask); + const rx_vec_f128 xexponentMask = rx_load_vec_f128((const double*)&config.eMask); + x = rx_and_vec_f128(x, xmantissaMask); + x = rx_or_vec_f128(x, xexponentMask); + return x; + } + + private: + static const int_reg_t zero; + int registerUsage[RegistersCount]; + NativeRegisterFile* nreg; + + static void* getScratchpadAddress(InstructionByteCode& ibc, uint8_t* scratchpad) { + uint32_t addr = (*ibc.isrc + ibc.imm) & ibc.memMask; + return scratchpad + addr; + } + +#ifdef RANDOMX_GEN_TABLE + static InstructionGenBytecode genTable[256]; + + void gen_IADD_RS(RANDOMX_GEN_ARGS); + void gen_IADD_M(RANDOMX_GEN_ARGS); + void gen_ISUB_R(RANDOMX_GEN_ARGS); + void gen_ISUB_M(RANDOMX_GEN_ARGS); + void gen_IMUL_R(RANDOMX_GEN_ARGS); + void gen_IMUL_M(RANDOMX_GEN_ARGS); + void gen_IMULH_R(RANDOMX_GEN_ARGS); + void gen_IMULH_M(RANDOMX_GEN_ARGS); + void gen_ISMULH_R(RANDOMX_GEN_ARGS); + void gen_ISMULH_M(RANDOMX_GEN_ARGS); + void gen_IMUL_RCP(RANDOMX_GEN_ARGS); + void gen_INEG_R(RANDOMX_GEN_ARGS); + void gen_IXOR_R(RANDOMX_GEN_ARGS); + void gen_IXOR_M(RANDOMX_GEN_ARGS); + void gen_IROR_R(RANDOMX_GEN_ARGS); + void gen_IROL_R(RANDOMX_GEN_ARGS); + void gen_ISWAP_R(RANDOMX_GEN_ARGS); + void gen_FSWAP_R(RANDOMX_GEN_ARGS); + void gen_FADD_R(RANDOMX_GEN_ARGS); + void gen_FADD_M(RANDOMX_GEN_ARGS); + void gen_FSUB_R(RANDOMX_GEN_ARGS); + void gen_FSUB_M(RANDOMX_GEN_ARGS); + void gen_FSCAL_R(RANDOMX_GEN_ARGS); + void gen_FMUL_R(RANDOMX_GEN_ARGS); + void gen_FDIV_M(RANDOMX_GEN_ARGS); + void gen_FSQRT_R(RANDOMX_GEN_ARGS); + void gen_CBRANCH(RANDOMX_GEN_ARGS); + void gen_CFROUND(RANDOMX_GEN_ARGS); + void gen_ISTORE(RANDOMX_GEN_ARGS); + void gen_NOP(RANDOMX_GEN_ARGS); +#endif + }; +} diff --git a/src/crypto/randomx/common.hpp b/src/crypto/randomx/common.hpp new file mode 100644 index 000000000..7f7ea0edb --- /dev/null +++ b/src/crypto/randomx/common.hpp @@ -0,0 +1,173 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include +#include +#include +#include "blake2/endian.h" +#include "configuration.h" +#include "randomx.h" + +namespace randomx { + + //static_assert(RANDOMX_ARGON_MEMORY > 0, "RANDOMX_ARGON_MEMORY must be greater than 0."); + //static_assert((RANDOMX_ARGON_MEMORY & (RANDOMX_ARGON_MEMORY - 1)) == 0, "RANDOMX_ARGON_MEMORY must be a power of 2."); + //static_assert(RANDOMX_DATASET_BASE_SIZE >= 64, "RANDOMX_DATASET_BASE_SIZE must be at least 64."); + //static_assert((RANDOMX_DATASET_BASE_SIZE & (RANDOMX_DATASET_BASE_SIZE - 1)) == 0, "RANDOMX_DATASET_BASE_SIZE must be a power of 2."); + //static_assert(RANDOMX_DATASET_BASE_SIZE <= 4294967296ULL, "RANDOMX_DATASET_BASE_SIZE must not exceed 4294967296."); + //static_assert(RANDOMX_DATASET_EXTRA_SIZE % 64 == 0, "RANDOMX_DATASET_EXTRA_SIZE must be divisible by 64."); + //static_assert((uint64_t)RANDOMX_DATASET_BASE_SIZE + RANDOMX_DATASET_EXTRA_SIZE <= 17179869184, "Dataset size must not exceed 16 GiB."); + //static_assert(RANDOMX_PROGRAM_SIZE > 0, "RANDOMX_PROGRAM_SIZE must be greater than 0"); + //static_assert(RANDOMX_PROGRAM_ITERATIONS > 0, "RANDOMX_PROGRAM_ITERATIONS must be greater than 0"); + //static_assert(RANDOMX_PROGRAM_COUNT > 0, "RANDOMX_PROGRAM_COUNT must be greater than 0"); + //static_assert((RANDOMX_SCRATCHPAD_L3 & (RANDOMX_SCRATCHPAD_L3 - 1)) == 0, "RANDOMX_SCRATCHPAD_L3 must be a power of 2."); + //static_assert(RANDOMX_SCRATCHPAD_L3 >= RANDOMX_SCRATCHPAD_L2, "RANDOMX_SCRATCHPAD_L3 must be greater than or equal to RANDOMX_SCRATCHPAD_L2."); + //static_assert((RANDOMX_SCRATCHPAD_L2 & (RANDOMX_SCRATCHPAD_L2 - 1)) == 0, "RANDOMX_SCRATCHPAD_L2 must be a power of 2."); + //static_assert(RANDOMX_SCRATCHPAD_L2 >= RANDOMX_SCRATCHPAD_L1, "RANDOMX_SCRATCHPAD_L2 must be greater than or equal to RANDOMX_SCRATCHPAD_L1."); + //static_assert(RANDOMX_SCRATCHPAD_L1 >= 64, "RANDOMX_SCRATCHPAD_L1 must be at least 64."); + //static_assert((RANDOMX_SCRATCHPAD_L1 & (RANDOMX_SCRATCHPAD_L1 - 1)) == 0, "RANDOMX_SCRATCHPAD_L1 must be a power of 2."); + //static_assert(RANDOMX_CACHE_ACCESSES > 1, "RANDOMX_CACHE_ACCESSES must be greater than 1"); + //static_assert(RANDOMX_SUPERSCALAR_LATENCY > 0, "RANDOMX_SUPERSCALAR_LATENCY must be greater than 0"); + //static_assert(RANDOMX_JUMP_BITS > 0, "RANDOMX_JUMP_BITS must be greater than 0."); + //static_assert(RANDOMX_JUMP_OFFSET >= 0, "RANDOMX_JUMP_OFFSET must be greater than or equal to 0."); + //static_assert(RANDOMX_JUMP_BITS + RANDOMX_JUMP_OFFSET <= 16, "RANDOMX_JUMP_BITS + RANDOMX_JUMP_OFFSET must not exceed 16."); + + /*constexpr int wtSum = RANDOMX_FREQ_IADD_RS + RANDOMX_FREQ_IADD_M + RANDOMX_FREQ_ISUB_R + \ + RANDOMX_FREQ_ISUB_M + RANDOMX_FREQ_IMUL_R + RANDOMX_FREQ_IMUL_M + RANDOMX_FREQ_IMULH_R + \ + RANDOMX_FREQ_IMULH_M + RANDOMX_FREQ_ISMULH_R + RANDOMX_FREQ_ISMULH_M + RANDOMX_FREQ_IMUL_RCP + \ + RANDOMX_FREQ_INEG_R + RANDOMX_FREQ_IXOR_R + RANDOMX_FREQ_IXOR_M + RANDOMX_FREQ_IROR_R + RANDOMX_FREQ_IROL_R + RANDOMX_FREQ_ISWAP_R + \ + RANDOMX_FREQ_FSWAP_R + RANDOMX_FREQ_FADD_R + RANDOMX_FREQ_FADD_M + RANDOMX_FREQ_FSUB_R + RANDOMX_FREQ_FSUB_M + \ + RANDOMX_FREQ_FSCAL_R + RANDOMX_FREQ_FMUL_R + RANDOMX_FREQ_FDIV_M + RANDOMX_FREQ_FSQRT_R + RANDOMX_FREQ_CBRANCH + \ + RANDOMX_FREQ_CFROUND + RANDOMX_FREQ_ISTORE + RANDOMX_FREQ_NOP;*/ + + //static_assert(wtSum == 256, "Sum of instruction frequencies must be 256."); + + + constexpr uint32_t ArgonBlockSize = 1024; + constexpr int SuperscalarMaxSize = 3 * RANDOMX_SUPERSCALAR_MAX_LATENCY + 2; + constexpr size_t CacheLineSize = RANDOMX_DATASET_ITEM_SIZE; + #define ScratchpadSize RandomX_CurrentConfig.ScratchpadL3_Size + #define CacheLineAlignMask RandomX_CurrentConfig.CacheLineAlignMask_Calculated + #define DatasetExtraItems RandomX_CurrentConfig.DatasetExtraItems_Calculated + constexpr int StoreL3Condition = 14; + + //Prevent some unsafe configurations. +#ifndef RANDOMX_UNSAFE + //static_assert((uint64_t)ArgonBlockSize * RANDOMX_CACHE_ACCESSES * RANDOMX_ARGON_MEMORY + 33554432 >= (uint64_t)RANDOMX_DATASET_BASE_SIZE + RANDOMX_DATASET_EXTRA_SIZE, "Unsafe configuration: Memory-time tradeoffs"); + //static_assert((128 + RANDOMX_PROGRAM_SIZE * RANDOMX_FREQ_ISTORE / 256) * (RANDOMX_PROGRAM_COUNT * RANDOMX_PROGRAM_ITERATIONS) >= RANDOMX_SCRATCHPAD_L3, "Unsafe configuration: Insufficient Scratchpad writes"); + //static_assert(RANDOMX_PROGRAM_COUNT > 1, "Unsafe configuration: Program filtering strategies"); + //static_assert(RANDOMX_PROGRAM_SIZE >= 64, "Unsafe configuration: Low program entropy"); + //static_assert(RANDOMX_PROGRAM_ITERATIONS >= 400, "Unsafe configuration: High compilation overhead"); +#endif + +#ifdef TRACE + constexpr bool trace = true; +#else + constexpr bool trace = false; +#endif + +#ifndef UNREACHABLE +#ifdef __GNUC__ +#define UNREACHABLE __builtin_unreachable() +#elif _MSC_VER +#define UNREACHABLE __assume(false) +#else +#define UNREACHABLE +#endif +#endif + +#if defined(_M_X64) || defined(__x86_64__) + #define RANDOMX_HAVE_COMPILER 1 + class JitCompilerX86; + using JitCompiler = JitCompilerX86; +#elif defined(__aarch64__) + #define RANDOMX_HAVE_COMPILER 0 + class JitCompilerA64; + using JitCompiler = JitCompilerA64; +#else + #define RANDOMX_HAVE_COMPILER 0 + class JitCompilerFallback; + using JitCompiler = JitCompilerFallback; +#endif + + using addr_t = uint32_t; + + using int_reg_t = uint64_t; + + struct fpu_reg_t { + double lo; + double hi; + }; + + #define ScratchpadL1Mask RandomX_CurrentConfig.ScratchpadL1Mask_Calculated + #define ScratchpadL1Mask16 RandomX_CurrentConfig.ScratchpadL1Mask16_Calculated + #define ScratchpadL2Mask RandomX_CurrentConfig.ScratchpadL2Mask_Calculated + #define ScratchpadL2Mask16 RandomX_CurrentConfig.ScratchpadL2Mask16_Calculated + #define ScratchpadL3Mask RandomX_CurrentConfig.ScratchpadL3Mask_Calculated + #define ScratchpadL3Mask64 RandomX_CurrentConfig.ScratchpadL3Mask64_Calculated + constexpr int RegistersCount = 8; + constexpr int RegisterCountFlt = RegistersCount / 2; + constexpr int RegisterNeedsDisplacement = 5; //x86 r13 register + constexpr int RegisterNeedsSib = 4; //x86 r12 register + + inline bool isPowerOf2(uint64_t x) { + return (x & (x - 1)) == 0; + } + + constexpr int mantissaSize = 52; + constexpr int exponentSize = 11; + constexpr uint64_t mantissaMask = (1ULL << mantissaSize) - 1; + constexpr uint64_t exponentMask = (1ULL << exponentSize) - 1; + constexpr int exponentBias = 1023; + constexpr int dynamicExponentBits = 4; + constexpr int staticExponentBits = 4; + constexpr uint64_t constExponentBits = 0x300; + constexpr uint64_t dynamicMantissaMask = (1ULL << (mantissaSize + dynamicExponentBits)) - 1; + + struct MemoryRegisters { + addr_t mx, ma; + uint8_t* memory = nullptr; + }; + + //register file in little-endian byte order + struct RegisterFile { + int_reg_t r[RegistersCount]; + fpu_reg_t f[RegisterCountFlt]; + fpu_reg_t e[RegisterCountFlt]; + fpu_reg_t a[RegisterCountFlt]; + }; + + typedef void(ProgramFunc)(RegisterFile&, MemoryRegisters&, uint8_t* /* scratchpad */, uint64_t); + typedef void(DatasetInitFunc)(randomx_cache* cache, uint8_t* dataset, uint32_t startBlock, uint32_t endBlock); + + typedef void(DatasetDeallocFunc)(randomx_dataset*); + typedef void(CacheDeallocFunc)(randomx_cache*); + typedef void(CacheInitializeFunc)(randomx_cache*, const void*, size_t); +} diff --git a/src/crypto/randomx/configuration.h b/src/crypto/randomx/configuration.h new file mode 100644 index 000000000..678cb2f8b --- /dev/null +++ b/src/crypto/randomx/configuration.h @@ -0,0 +1,47 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +// Increase it if some configs use more cache accesses +#define RANDOMX_CACHE_MAX_ACCESSES 16 + +// Increase it if some configs use larger superscalar latency +#define RANDOMX_SUPERSCALAR_MAX_LATENCY 256 + +// Increase it if some configs use larger cache +#define RANDOMX_CACHE_MAX_SIZE 268435456 + +// Increase it if some configs use larger dataset +#define RANDOMX_DATASET_MAX_SIZE 2181038080 + +// Increase it if some configs use larger programs +#define RANDOMX_PROGRAM_MAX_SIZE 512 + +// Increase it if some configs use larger scratchpad +#define RANDOMX_SCRATCHPAD_L3_MAX_SIZE 2097152 diff --git a/src/crypto/randomx/dataset.cpp b/src/crypto/randomx/dataset.cpp new file mode 100644 index 000000000..b094b1cb4 --- /dev/null +++ b/src/crypto/randomx/dataset.cpp @@ -0,0 +1,189 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +/* Original code from Argon2 reference source code package used under CC0 Licence + * https://github.com/P-H-C/phc-winner-argon2 + * Copyright 2015 + * Daniel Dinu, Dmitry Khovratovich, Jean-Philippe Aumasson, and Samuel Neves +*/ + +#include +#include +#include +#include +#include +#include + +#include "common.hpp" +#include "dataset.hpp" +#include "virtual_memory.hpp" +#include "superscalar.hpp" +#include "blake2_generator.hpp" +#include "reciprocal.h" +#include "blake2/endian.h" +#include "argon2.h" +#include "argon2_core.h" +#include "jit_compiler.hpp" +#include "intrin_portable.h" + +//static_assert(RANDOMX_ARGON_MEMORY % (RANDOMX_ARGON_LANES * ARGON2_SYNC_POINTS) == 0, "RANDOMX_ARGON_MEMORY - invalid value"); +static_assert(ARGON2_BLOCK_SIZE == randomx::ArgonBlockSize, "Unpexpected value of ARGON2_BLOCK_SIZE"); + +namespace randomx { + + template + void deallocCache(randomx_cache* cache) { + if (cache->memory != nullptr) + Allocator::freeMemory(cache->memory, RANDOMX_CACHE_MAX_SIZE); + if (cache->jit != nullptr) + delete cache->jit; + } + + template void deallocCache(randomx_cache* cache); + template void deallocCache(randomx_cache* cache); + + void initCache(randomx_cache* cache, const void* key, size_t keySize) { + uint32_t memory_blocks, segment_length; + argon2_instance_t instance; + argon2_context context; + + context.out = nullptr; + context.outlen = 0; + context.pwd = CONST_CAST(uint8_t *)key; + context.pwdlen = (uint32_t)keySize; + context.salt = CONST_CAST(uint8_t *)RandomX_CurrentConfig.ArgonSalt; + context.saltlen = (uint32_t)strlen(RandomX_CurrentConfig.ArgonSalt); + context.secret = NULL; + context.secretlen = 0; + context.ad = NULL; + context.adlen = 0; + context.t_cost = RandomX_CurrentConfig.ArgonIterations; + context.m_cost = RandomX_CurrentConfig.ArgonMemory; + context.lanes = RandomX_CurrentConfig.ArgonLanes; + context.threads = 1; + context.allocate_cbk = NULL; + context.free_cbk = NULL; + context.flags = ARGON2_DEFAULT_FLAGS; + context.version = ARGON2_VERSION_NUMBER; + + /* 2. Align memory size */ + /* Minimum memory_blocks = 8L blocks, where L is the number of lanes */ + memory_blocks = context.m_cost; + + segment_length = memory_blocks / (context.lanes * ARGON2_SYNC_POINTS); + + instance.version = context.version; + instance.memory = NULL; + instance.passes = context.t_cost; + instance.memory_blocks = memory_blocks; + instance.segment_length = segment_length; + instance.lane_length = segment_length * ARGON2_SYNC_POINTS; + instance.lanes = context.lanes; + instance.threads = context.threads; + instance.type = Argon2_d; + instance.memory = (block*)cache->memory; + + if (instance.threads > instance.lanes) { + instance.threads = instance.lanes; + } + + /* 3. Initialization: Hashing inputs, allocating memory, filling first + * blocks + */ + rxa2_argon_initialize(&instance, &context); + + rxa2_fill_memory_blocks(&instance); + + cache->reciprocalCache.clear(); + randomx::Blake2Generator gen(key, keySize); + for (uint32_t i = 0; i < RandomX_CurrentConfig.CacheAccesses; ++i) { + randomx::generateSuperscalar(cache->programs[i], gen); + for (unsigned j = 0; j < cache->programs[i].getSize(); ++j) { + auto& instr = cache->programs[i](j); + if ((SuperscalarInstructionType)instr.opcode == SuperscalarInstructionType::IMUL_RCP) { + auto rcp = randomx_reciprocal(instr.getImm32()); + instr.setImm32(cache->reciprocalCache.size()); + cache->reciprocalCache.push_back(rcp); + } + } + } + } + + void initCacheCompile(randomx_cache* cache, const void* key, size_t keySize) { + initCache(cache, key, keySize); + cache->jit->generateSuperscalarHash(cache->programs, cache->reciprocalCache); + cache->jit->generateDatasetInitCode(); + } + + constexpr uint64_t superscalarMul0 = 6364136223846793005ULL; + constexpr uint64_t superscalarAdd1 = 9298411001130361340ULL; + constexpr uint64_t superscalarAdd2 = 12065312585734608966ULL; + constexpr uint64_t superscalarAdd3 = 9306329213124626780ULL; + constexpr uint64_t superscalarAdd4 = 5281919268842080866ULL; + constexpr uint64_t superscalarAdd5 = 10536153434571861004ULL; + constexpr uint64_t superscalarAdd6 = 3398623926847679864ULL; + constexpr uint64_t superscalarAdd7 = 9549104520008361294ULL; + + static inline uint8_t* getMixBlock(uint64_t registerValue, uint8_t *memory) { + const uint32_t mask = (RandomX_CurrentConfig.ArgonMemory * randomx::ArgonBlockSize) / CacheLineSize - 1; + return memory + (registerValue & mask) * CacheLineSize; + } + + void initDatasetItem(randomx_cache* cache, uint8_t* out, uint64_t itemNumber) { + int_reg_t rl[8]; + uint8_t* mixBlock; + uint64_t registerValue = itemNumber; + rl[0] = (itemNumber + 1) * superscalarMul0; + rl[1] = rl[0] ^ superscalarAdd1; + rl[2] = rl[0] ^ superscalarAdd2; + rl[3] = rl[0] ^ superscalarAdd3; + rl[4] = rl[0] ^ superscalarAdd4; + rl[5] = rl[0] ^ superscalarAdd5; + rl[6] = rl[0] ^ superscalarAdd6; + rl[7] = rl[0] ^ superscalarAdd7; + for (unsigned i = 0; i < RandomX_CurrentConfig.CacheAccesses; ++i) { + mixBlock = getMixBlock(registerValue, cache->memory); + rx_prefetch_nta(mixBlock); + SuperscalarProgram& prog = cache->programs[i]; + + executeSuperscalar(rl, prog, &cache->reciprocalCache); + + for (unsigned q = 0; q < 8; ++q) + rl[q] ^= load64_native(mixBlock + 8 * q); + + registerValue = rl[prog.getAddressRegister()]; + } + + memcpy(out, &rl, CacheLineSize); + } + + void initDataset(randomx_cache* cache, uint8_t* dataset, uint32_t startItem, uint32_t endItem) { + for (uint32_t itemNumber = startItem; itemNumber < endItem; ++itemNumber, dataset += CacheLineSize) + initDatasetItem(cache, dataset, itemNumber); + } +} diff --git a/src/crypto/randomx/dataset.hpp b/src/crypto/randomx/dataset.hpp new file mode 100644 index 000000000..42d4f05eb --- /dev/null +++ b/src/crypto/randomx/dataset.hpp @@ -0,0 +1,80 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include +#include +#include +#include "common.hpp" +#include "superscalar_program.hpp" +#include "allocator.hpp" + +/* Global scope for C binding */ +struct randomx_dataset { + uint8_t* memory = nullptr; + randomx::DatasetDeallocFunc* dealloc; +}; + +/* Global scope for C binding */ +struct randomx_cache { + uint8_t* memory = nullptr; + randomx::CacheDeallocFunc* dealloc; + randomx::JitCompiler* jit; + randomx::CacheInitializeFunc* initialize; + randomx::DatasetInitFunc* datasetInit; + randomx::SuperscalarProgram programs[RANDOMX_CACHE_MAX_ACCESSES]; + std::vector reciprocalCache; + + bool isInitialized() { + return programs[0].getSize() != 0; + } +}; + +//A pointer to a standard-layout struct object points to its initial member +static_assert(std::is_standard_layout(), "randomx_dataset must be a standard-layout struct"); +static_assert(std::is_standard_layout(), "randomx_cache must be a standard-layout struct"); + +namespace randomx { + + using DefaultAllocator = AlignedAllocator; + + template + void deallocDataset(randomx_dataset* dataset) { + if (dataset->memory != nullptr) + Allocator::freeMemory(dataset->memory, RANDOMX_DATASET_MAX_SIZE); + } + + template + void deallocCache(randomx_cache* cache); + + void initCache(randomx_cache*, const void*, size_t); + void initCacheCompile(randomx_cache*, const void*, size_t); + void initDatasetItem(randomx_cache* cache, uint8_t* out, uint64_t blockNumber); + void initDataset(randomx_cache* cache, uint8_t* dataset, uint32_t startBlock, uint32_t endBlock); +} diff --git a/src/crypto/randomx/instruction.hpp b/src/crypto/randomx/instruction.hpp new file mode 100644 index 000000000..1904afc0c --- /dev/null +++ b/src/crypto/randomx/instruction.hpp @@ -0,0 +1,102 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include +#include +#include "blake2/endian.h" + +namespace randomx { + + class Instruction; + + enum class InstructionType : uint16_t { + IADD_RS = 0, + IADD_M = 1, + ISUB_R = 2, + ISUB_M = 3, + IMUL_R = 4, + IMUL_M = 5, + IMULH_R = 6, + IMULH_M = 7, + ISMULH_R = 8, + ISMULH_M = 9, + IMUL_RCP = 10, + INEG_R = 11, + IXOR_R = 12, + IXOR_M = 13, + IROR_R = 14, + IROL_R = 15, + ISWAP_R = 16, + FSWAP_R = 17, + FADD_R = 18, + FADD_M = 19, + FSUB_R = 20, + FSUB_M = 21, + FSCAL_R = 22, + FMUL_R = 23, + FDIV_M = 24, + FSQRT_R = 25, + CBRANCH = 26, + CFROUND = 27, + ISTORE = 28, + NOP = 29, + }; + + class Instruction { + public: + uint32_t getImm32() const { + return load32(&imm32); + } + void setImm32(uint32_t val) { + return store32(&imm32, val); + } + int getModMem() const { + return mod % 4; //bits 0-1 + } + int getModShift() const { + return (mod >> 2) % 4; //bits 2-3 + } + int getModCond() const { + return mod >> 4; //bits 4-7 + } + void setMod(uint8_t val) { + mod = val; + } + + uint8_t opcode; + uint8_t dst; + uint8_t src; + uint8_t mod; + uint32_t imm32; + }; + + static_assert(sizeof(Instruction) == 8, "Invalid size of struct randomx::Instruction"); + static_assert(std::is_standard_layout(), "randomx::Instruction must be a standard-layout struct"); +} diff --git a/src/crypto/randomx/instructions_portable.cpp b/src/crypto/randomx/instructions_portable.cpp new file mode 100644 index 000000000..8f0592dda --- /dev/null +++ b/src/crypto/randomx/instructions_portable.cpp @@ -0,0 +1,193 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include +#include +#include "common.hpp" +#include "intrin_portable.h" +#include "blake2/endian.h" + +#if defined(__SIZEOF_INT128__) + typedef unsigned __int128 uint128_t; + typedef __int128 int128_t; + uint64_t mulh(uint64_t a, uint64_t b) { + return ((uint128_t)a * b) >> 64; + } + int64_t smulh(int64_t a, int64_t b) { + return ((int128_t)a * b) >> 64; + } + #define HAVE_MULH + #define HAVE_SMULH +#endif + +#if defined(_MSC_VER) + #define HAS_VALUE(X) X ## 0 + #define EVAL_DEFINE(X) HAS_VALUE(X) + #include + #include + + uint64_t rotl64(uint64_t x, unsigned int c) { + return _rotl64(x, c); + } + uint64_t rotr64(uint64_t x, unsigned int c) { + return _rotr64(x, c); + } + #define HAVE_ROTL64 + #define HAVE_ROTR64 + + #if EVAL_DEFINE(__MACHINEARM64_X64(1)) + uint64_t mulh(uint64_t a, uint64_t b) { + return __umulh(a, b); + } + #define HAVE_MULH + #endif + + #if EVAL_DEFINE(__MACHINEX64(1)) + int64_t smulh(int64_t a, int64_t b) { + int64_t hi; + _mul128(a, b, &hi); + return hi; + } + #define HAVE_SMULH + #endif + + static void setRoundMode_(uint32_t mode) { + _controlfp(mode, _MCW_RC); + } + #define HAVE_SETROUNDMODE_IMPL +#endif + +#ifndef HAVE_ROTR64 + uint64_t rotr64(uint64_t a, unsigned int b) { + return (a >> b) | (a << (-b & 63)); + } + #define HAVE_ROTR64 +#endif + +#ifndef HAVE_ROTL64 + uint64_t rotl64(uint64_t a, unsigned int b) { + return (a << b) | (a >> (-b & 63)); + } + #define HAVE_ROTL64 +#endif + +#ifndef HAVE_MULH + #define LO(x) ((x)&0xffffffff) + #define HI(x) ((x)>>32) + uint64_t mulh(uint64_t a, uint64_t b) { + uint64_t ah = HI(a), al = LO(a); + uint64_t bh = HI(b), bl = LO(b); + uint64_t x00 = al * bl; + uint64_t x01 = al * bh; + uint64_t x10 = ah * bl; + uint64_t x11 = ah * bh; + uint64_t m1 = LO(x10) + LO(x01) + HI(x00); + uint64_t m2 = HI(x10) + HI(x01) + LO(x11) + HI(m1); + uint64_t m3 = HI(x11) + HI(m2); + + return (m3 << 32) + LO(m2); + } + #define HAVE_MULH +#endif + +#ifndef HAVE_SMULH + int64_t smulh(int64_t a, int64_t b) { + int64_t hi = mulh(a, b); + if (a < 0LL) hi -= b; + if (b < 0LL) hi -= a; + return hi; + } + #define HAVE_SMULH +#endif + +#ifdef RANDOMX_DEFAULT_FENV + +# ifndef HAVE_SETROUNDMODE_IMPL + static void setRoundMode_(uint32_t mode) { + fesetround(mode); + } +# endif + +void rx_reset_float_state() { + setRoundMode_(FE_TONEAREST); + rx_set_double_precision(); //set precision to 53 bits if needed by the platform +} + +void rx_set_rounding_mode(uint32_t mode) { + switch (mode & 3) { + case RoundDown: + setRoundMode_(FE_DOWNWARD); + break; + case RoundUp: + setRoundMode_(FE_UPWARD); + break; + case RoundToZero: + setRoundMode_(FE_TOWARDZERO); + break; + case RoundToNearest: + setRoundMode_(FE_TONEAREST); + break; + default: + UNREACHABLE; + } +} + +#endif + +#ifdef RANDOMX_USE_X87 + +#ifdef _M_IX86 + +void rx_set_double_precision() { + _control87(_PC_53, _MCW_PC); +} + +#elif defined(__i386) + +void rx_set_double_precision() { + uint16_t volatile x87cw; + asm volatile("fstcw %0" : "=m" (x87cw)); + x87cw &= ~0x300; + x87cw |= 0x200; + asm volatile("fldcw %0" : : "m" (x87cw)); +} + +#endif + +#endif //RANDOMX_USE_X87 + +union double_ser_t { + double f; + uint64_t i; +}; + +double loadDoublePortable(const void* addr) { + double_ser_t ds; + ds.i = load64(addr); + return ds.f; +} diff --git a/src/crypto/randomx/intrin_portable.h b/src/crypto/randomx/intrin_portable.h new file mode 100644 index 000000000..83acbe65d --- /dev/null +++ b/src/crypto/randomx/intrin_portable.h @@ -0,0 +1,605 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include +#include "blake2/endian.h" + +constexpr int32_t unsigned32ToSigned2sCompl(uint32_t x) { + return (-1 == ~0) ? (int32_t)x : (x > INT32_MAX ? (-(int32_t)(UINT32_MAX - x) - 1) : (int32_t)x); +} + +constexpr int64_t unsigned64ToSigned2sCompl(uint64_t x) { + return (-1 == ~0) ? (int64_t)x : (x > INT64_MAX ? (-(int64_t)(UINT64_MAX - x) - 1) : (int64_t)x); +} + +constexpr uint64_t signExtend2sCompl(uint32_t x) { + return (-1 == ~0) ? (int64_t)(int32_t)(x) : (x > INT32_MAX ? (x | 0xffffffff00000000ULL) : (uint64_t)x); +} + +constexpr int RoundToNearest = 0; +constexpr int RoundDown = 1; +constexpr int RoundUp = 2; +constexpr int RoundToZero = 3; + +//MSVC doesn't define __SSE2__, so we have to define it manually if SSE2 is available +#if !defined(__SSE2__) && (defined(_M_X64) || (defined(_M_IX86_FP) && _M_IX86_FP == 2)) +#define __SSE2__ 1 +#endif + +//MSVC doesn't define __AES__ +#if defined(_MSC_VER) && defined(__SSE2__) +#define __AES__ +#endif + +//the library "sqrt" function provided by MSVC for x86 targets doesn't give +//the correct results, so we have to use inline assembly to call x87 fsqrt directly +#if !defined(__SSE2__) +#if defined(_M_IX86) +inline double __cdecl rx_sqrt(double x) { + __asm { + fld x + fsqrt + } +} +#define rx_sqrt rx_sqrt + +void rx_set_double_precision(); +#define RANDOMX_USE_X87 + +#elif defined(__i386) + +void rx_set_double_precision(); +#define RANDOMX_USE_X87 + +#endif +#endif //__SSE2__ + +#if !defined(rx_sqrt) +#define rx_sqrt sqrt +#endif + +#if !defined(RANDOMX_USE_X87) +#define rx_set_double_precision(x) +#endif + +#ifdef __SSE2__ +#ifdef __GNUC__ +#include +#else +#include +#endif + +typedef __m128i rx_vec_i128; +typedef __m128d rx_vec_f128; + +#define rx_aligned_alloc(a, b) _mm_malloc(a,b) +#define rx_aligned_free(a) _mm_free(a) +#define rx_prefetch_nta(x) _mm_prefetch((const char *)(x), _MM_HINT_NTA) + +#define rx_load_vec_f128 _mm_load_pd +#define rx_store_vec_f128 _mm_store_pd +#define rx_add_vec_f128 _mm_add_pd +#define rx_sub_vec_f128 _mm_sub_pd +#define rx_mul_vec_f128 _mm_mul_pd +#define rx_div_vec_f128 _mm_div_pd +#define rx_sqrt_vec_f128 _mm_sqrt_pd + +FORCE_INLINE rx_vec_f128 rx_swap_vec_f128(rx_vec_f128 a) { + return _mm_shuffle_pd(a, a, 1); +} + +FORCE_INLINE rx_vec_f128 rx_set_vec_f128(uint64_t x1, uint64_t x0) { + return _mm_castsi128_pd(_mm_set_epi64x(x1, x0)); +} + +FORCE_INLINE rx_vec_f128 rx_set1_vec_f128(uint64_t x) { + return _mm_castsi128_pd(_mm_set1_epi64x(x)); +} + +#define rx_xor_vec_f128 _mm_xor_pd +#define rx_and_vec_f128 _mm_and_pd +#define rx_or_vec_f128 _mm_or_pd + +#ifdef __AES__ + +#define rx_aesenc_vec_i128 _mm_aesenc_si128 +#define rx_aesdec_vec_i128 _mm_aesdec_si128 + +#define HAVE_AES + +#endif //__AES__ + +FORCE_INLINE int rx_vec_i128_x(rx_vec_i128 a) { + return _mm_cvtsi128_si32(a); +} + +FORCE_INLINE int rx_vec_i128_y(rx_vec_i128 a) { + return _mm_cvtsi128_si32(_mm_shuffle_epi32(a, 0x55)); +} + +FORCE_INLINE int rx_vec_i128_z(rx_vec_i128 a) { + return _mm_cvtsi128_si32(_mm_shuffle_epi32(a, 0xaa)); +} + +FORCE_INLINE int rx_vec_i128_w(rx_vec_i128 a) { + return _mm_cvtsi128_si32(_mm_shuffle_epi32(a, 0xff)); +} + +#define rx_set_int_vec_i128 _mm_set_epi32 +#define rx_xor_vec_i128 _mm_xor_si128 +#define rx_load_vec_i128 _mm_load_si128 +#define rx_store_vec_i128 _mm_store_si128 + +FORCE_INLINE rx_vec_f128 rx_cvt_packed_int_vec_f128(const void* addr) { + __m128i ix = _mm_loadl_epi64((const __m128i*)addr); + return _mm_cvtepi32_pd(ix); +} + +constexpr uint32_t rx_mxcsr_default = 0x9FC0; //Flush to zero, denormals are zero, default rounding mode, all exceptions disabled + +FORCE_INLINE void rx_reset_float_state() { + _mm_setcsr(rx_mxcsr_default); +} + +FORCE_INLINE void rx_set_rounding_mode(uint32_t mode) { + _mm_setcsr(rx_mxcsr_default | (mode << 13)); +} + +#elif defined(__PPC64__) && defined(__ALTIVEC__) && defined(__VSX__) //sadly only POWER7 and newer will be able to use SIMD acceleration. Earlier processors cant use doubles or 64 bit integers with SIMD +#include +#include +#include +#include +#undef vector +#undef pixel +#undef bool + +typedef __vector uint8_t __m128i; +typedef __vector uint32_t __m128l; +typedef __vector int __m128li; +typedef __vector uint64_t __m128ll; +typedef __vector double __m128d; + +typedef __m128i rx_vec_i128; +typedef __m128d rx_vec_f128; +typedef union{ + rx_vec_i128 i; + rx_vec_f128 d; + uint64_t u64[2]; + double d64[2]; + uint32_t u32[4]; + int i32[4]; +} vec_u; + +#define rx_aligned_alloc(a, b) malloc(a) +#define rx_aligned_free(a) free(a) +#define rx_prefetch_nta(x) + +/* Splat 64-bit long long to 2 64-bit long longs */ +FORCE_INLINE __m128i vec_splat2sd (int64_t scalar) +{ return (__m128i) vec_splats (scalar); } + +FORCE_INLINE rx_vec_f128 rx_load_vec_f128(const double* pd) { +#if defined(NATIVE_LITTLE_ENDIAN) + return (rx_vec_f128)vec_vsx_ld(0,pd); +#else + vec_u t; + t.u64[0] = load64(pd + 0); + t.u64[1] = load64(pd + 1); + return (rx_vec_f128)t.d; +#endif +} + +FORCE_INLINE void rx_store_vec_f128(double* mem_addr, rx_vec_f128 a) { +#if defined(NATIVE_LITTLE_ENDIAN) + vec_vsx_st(a,0,(rx_vec_f128*)mem_addr); +#else + vec_u _a; + _a.d = a; + store64(mem_addr + 0, _a.u64[0]); + store64(mem_addr + 1, _a.u64[1]); +#endif +} + +FORCE_INLINE rx_vec_f128 rx_swap_vec_f128(rx_vec_f128 a) { + return (rx_vec_f128)vec_perm((__m128i)a,(__m128i)a,(__m128i){8,9,10,11,12,13,14,15,0,1,2,3,4,5,6,7}); +} + +FORCE_INLINE rx_vec_f128 rx_add_vec_f128(rx_vec_f128 a, rx_vec_f128 b) { + return (rx_vec_f128)vec_add(a,b); +} + +FORCE_INLINE rx_vec_f128 rx_sub_vec_f128(rx_vec_f128 a, rx_vec_f128 b) { + return (rx_vec_f128)vec_sub(a,b); +} + +FORCE_INLINE rx_vec_f128 rx_mul_vec_f128(rx_vec_f128 a, rx_vec_f128 b) { + return (rx_vec_f128)vec_mul(a,b); +} + +FORCE_INLINE rx_vec_f128 rx_div_vec_f128(rx_vec_f128 a, rx_vec_f128 b) { + return (rx_vec_f128)vec_div(a,b); +} + +FORCE_INLINE rx_vec_f128 rx_sqrt_vec_f128(rx_vec_f128 a) { + return (rx_vec_f128)vec_sqrt(a); +} + +FORCE_INLINE rx_vec_i128 rx_set1_long_vec_i128(uint64_t a) { + return (rx_vec_i128)vec_splat2sd(a); +} + +FORCE_INLINE rx_vec_f128 rx_vec_i128_vec_f128(rx_vec_i128 a) { + return (rx_vec_f128)a; +} + +FORCE_INLINE rx_vec_f128 rx_set_vec_f128(uint64_t x1, uint64_t x0) { + return (rx_vec_f128)(__m128ll){x0,x1}; +} + +FORCE_INLINE rx_vec_f128 rx_set1_vec_f128(uint64_t x) { + return (rx_vec_f128)vec_splat2sd(x); +} + +FORCE_INLINE rx_vec_f128 rx_xor_vec_f128(rx_vec_f128 a, rx_vec_f128 b) { + return (rx_vec_f128)vec_xor(a,b); +} + +FORCE_INLINE rx_vec_f128 rx_and_vec_f128(rx_vec_f128 a, rx_vec_f128 b) { + return (rx_vec_f128)vec_and(a,b); +} + +FORCE_INLINE rx_vec_f128 rx_or_vec_f128(rx_vec_f128 a, rx_vec_f128 b) { + return (rx_vec_f128)vec_or(a,b); +} + +#if defined(__CRYPTO__) + +FORCE_INLINE __m128ll vrev(__m128i v){ +#if defined(NATIVE_LITTLE_ENDIAN) + return (__m128ll)vec_perm((__m128i)v,(__m128i){0},(__m128i){15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0}); +#else + return (__m128ll)vec_perm((__m128i)v,(__m128i){0},(__m128i){3,2,1,0, 7,6,5,4, 11,10,9,8, 15,14,13,12}); +#endif +} + +FORCE_INLINE rx_vec_i128 rx_aesenc_vec_i128(rx_vec_i128 v, rx_vec_i128 rkey) { + __m128ll _v = vrev(v); + __m128ll _rkey = vrev(rkey); + __m128ll result = vrev((__m128i)__builtin_crypto_vcipher(_v,_rkey)); + return (rx_vec_i128)result; +} + +FORCE_INLINE rx_vec_i128 rx_aesdec_vec_i128(rx_vec_i128 v, rx_vec_i128 rkey) { + __m128ll _v = vrev(v); + __m128ll zero = (__m128ll){0}; + __m128ll out = vrev((__m128i)__builtin_crypto_vncipher(_v,zero)); + return (rx_vec_i128)vec_xor((__m128i)out,rkey); +} +#define HAVE_AES + +#endif //__CRYPTO__ + +FORCE_INLINE int rx_vec_i128_x(rx_vec_i128 a) { + vec_u _a; + _a.i = a; + return _a.i32[0]; +} + +FORCE_INLINE int rx_vec_i128_y(rx_vec_i128 a) { + vec_u _a; + _a.i = a; + return _a.i32[1]; +} + +FORCE_INLINE int rx_vec_i128_z(rx_vec_i128 a) { + vec_u _a; + _a.i = a; + return _a.i32[2]; +} + +FORCE_INLINE int rx_vec_i128_w(rx_vec_i128 a) { + vec_u _a; + _a.i = a; + return _a.i32[3]; +} + +FORCE_INLINE rx_vec_i128 rx_set_int_vec_i128(int _I3, int _I2, int _I1, int _I0) { + return (rx_vec_i128)((__m128li){_I0,_I1,_I2,_I3}); +}; + +FORCE_INLINE rx_vec_i128 rx_xor_vec_i128(rx_vec_i128 _A, rx_vec_i128 _B) { + return (rx_vec_i128)vec_xor(_A,_B); +} + +FORCE_INLINE rx_vec_i128 rx_load_vec_i128(rx_vec_i128 const *_P) { +#if defined(NATIVE_LITTLE_ENDIAN) + return *_P; +#else + uint32_t* ptr = (uint32_t*)_P; + vec_u c; + c.u32[0] = load32(ptr + 0); + c.u32[1] = load32(ptr + 1); + c.u32[2] = load32(ptr + 2); + c.u32[3] = load32(ptr + 3); + return (rx_vec_i128)c.i; +#endif +} + +FORCE_INLINE void rx_store_vec_i128(rx_vec_i128 *_P, rx_vec_i128 _B) { +#if defined(NATIVE_LITTLE_ENDIAN) + *_P = _B; +#else + uint32_t* ptr = (uint32_t*)_P; + vec_u B; + B.i = _B; + store32(ptr + 0, B.u32[0]); + store32(ptr + 1, B.u32[1]); + store32(ptr + 2, B.u32[2]); + store32(ptr + 3, B.u32[3]); +#endif +} + +FORCE_INLINE rx_vec_f128 rx_cvt_packed_int_vec_f128(const void* addr) { + vec_u x; + x.d64[0] = (double)unsigned32ToSigned2sCompl(load32((uint8_t*)addr + 0)); + x.d64[1] = (double)unsigned32ToSigned2sCompl(load32((uint8_t*)addr + 4)); + return (rx_vec_f128)x.d; +} + +#define RANDOMX_DEFAULT_FENV + +void rx_reset_float_state(); + +void rx_set_rounding_mode(uint32_t mode); + +#else //end altivec + +#include +#include +#include +#include + +typedef union { + uint64_t u64[2]; + uint32_t u32[4]; + uint16_t u16[8]; + uint8_t u8[16]; +} rx_vec_i128; + +typedef union { + struct { + double lo; + double hi; + }; + rx_vec_i128 i; +} rx_vec_f128; + +#define rx_aligned_alloc(a, b) malloc(a) +#define rx_aligned_free(a) free(a) +#define rx_prefetch_nta(x) + +FORCE_INLINE rx_vec_f128 rx_load_vec_f128(const double* pd) { + rx_vec_f128 x; + x.i.u64[0] = load64(pd + 0); + x.i.u64[1] = load64(pd + 1); + return x; +} + +FORCE_INLINE void rx_store_vec_f128(double* mem_addr, rx_vec_f128 a) { + store64(mem_addr + 0, a.i.u64[0]); + store64(mem_addr + 1, a.i.u64[1]); +} + +FORCE_INLINE rx_vec_f128 rx_swap_vec_f128(rx_vec_f128 a) { + double temp = a.hi; + a.hi = a.lo; + a.lo = temp; + return a; +} + +FORCE_INLINE rx_vec_f128 rx_add_vec_f128(rx_vec_f128 a, rx_vec_f128 b) { + rx_vec_f128 x; + x.lo = a.lo + b.lo; + x.hi = a.hi + b.hi; + return x; +} + +FORCE_INLINE rx_vec_f128 rx_sub_vec_f128(rx_vec_f128 a, rx_vec_f128 b) { + rx_vec_f128 x; + x.lo = a.lo - b.lo; + x.hi = a.hi - b.hi; + return x; +} + +FORCE_INLINE rx_vec_f128 rx_mul_vec_f128(rx_vec_f128 a, rx_vec_f128 b) { + rx_vec_f128 x; + x.lo = a.lo * b.lo; + x.hi = a.hi * b.hi; + return x; +} + +FORCE_INLINE rx_vec_f128 rx_div_vec_f128(rx_vec_f128 a, rx_vec_f128 b) { + rx_vec_f128 x; + x.lo = a.lo / b.lo; + x.hi = a.hi / b.hi; + return x; +} + +FORCE_INLINE rx_vec_f128 rx_sqrt_vec_f128(rx_vec_f128 a) { + rx_vec_f128 x; + x.lo = rx_sqrt(a.lo); + x.hi = rx_sqrt(a.hi); + return x; +} + +FORCE_INLINE rx_vec_i128 rx_set1_long_vec_i128(uint64_t a) { + rx_vec_i128 x; + x.u64[0] = a; + x.u64[1] = a; + return x; +} + +FORCE_INLINE rx_vec_f128 rx_vec_i128_vec_f128(rx_vec_i128 a) { + rx_vec_f128 x; + x.i = a; + return x; +} + +FORCE_INLINE rx_vec_f128 rx_set_vec_f128(uint64_t x1, uint64_t x0) { + rx_vec_f128 v; + v.i.u64[0] = x0; + v.i.u64[1] = x1; + return v; +} + +FORCE_INLINE rx_vec_f128 rx_set1_vec_f128(uint64_t x) { + rx_vec_f128 v; + v.i.u64[0] = x; + v.i.u64[1] = x; + return v; +} + + +FORCE_INLINE rx_vec_f128 rx_xor_vec_f128(rx_vec_f128 a, rx_vec_f128 b) { + rx_vec_f128 x; + x.i.u64[0] = a.i.u64[0] ^ b.i.u64[0]; + x.i.u64[1] = a.i.u64[1] ^ b.i.u64[1]; + return x; +} + +FORCE_INLINE rx_vec_f128 rx_and_vec_f128(rx_vec_f128 a, rx_vec_f128 b) { + rx_vec_f128 x; + x.i.u64[0] = a.i.u64[0] & b.i.u64[0]; + x.i.u64[1] = a.i.u64[1] & b.i.u64[1]; + return x; +} + +FORCE_INLINE rx_vec_f128 rx_or_vec_f128(rx_vec_f128 a, rx_vec_f128 b) { + rx_vec_f128 x; + x.i.u64[0] = a.i.u64[0] | b.i.u64[0]; + x.i.u64[1] = a.i.u64[1] | b.i.u64[1]; + return x; +} + +FORCE_INLINE int rx_vec_i128_x(rx_vec_i128 a) { + return a.u32[0]; +} + +FORCE_INLINE int rx_vec_i128_y(rx_vec_i128 a) { + return a.u32[1]; +} + +FORCE_INLINE int rx_vec_i128_z(rx_vec_i128 a) { + return a.u32[2]; +} + +FORCE_INLINE int rx_vec_i128_w(rx_vec_i128 a) { + return a.u32[3]; +} + +FORCE_INLINE rx_vec_i128 rx_set_int_vec_i128(int _I3, int _I2, int _I1, int _I0) { + rx_vec_i128 v; + v.u32[0] = _I0; + v.u32[1] = _I1; + v.u32[2] = _I2; + v.u32[3] = _I3; + return v; +}; + +FORCE_INLINE rx_vec_i128 rx_xor_vec_i128(rx_vec_i128 _A, rx_vec_i128 _B) { + rx_vec_i128 c; + c.u32[0] = _A.u32[0] ^ _B.u32[0]; + c.u32[1] = _A.u32[1] ^ _B.u32[1]; + c.u32[2] = _A.u32[2] ^ _B.u32[2]; + c.u32[3] = _A.u32[3] ^ _B.u32[3]; + return c; +} + +FORCE_INLINE rx_vec_i128 rx_load_vec_i128(rx_vec_i128 const*_P) { +#if defined(NATIVE_LITTLE_ENDIAN) + return *_P; +#else + uint32_t* ptr = (uint32_t*)_P; + rx_vec_i128 c; + c.u32[0] = load32(ptr + 0); + c.u32[1] = load32(ptr + 1); + c.u32[2] = load32(ptr + 2); + c.u32[3] = load32(ptr + 3); + return c; +#endif +} + +FORCE_INLINE void rx_store_vec_i128(rx_vec_i128 *_P, rx_vec_i128 _B) { +#if defined(NATIVE_LITTLE_ENDIAN) + *_P = _B; +#else + uint32_t* ptr = (uint32_t*)_P; + store32(ptr + 0, _B.u32[0]); + store32(ptr + 1, _B.u32[1]); + store32(ptr + 2, _B.u32[2]); + store32(ptr + 3, _B.u32[3]); +#endif +} + +FORCE_INLINE rx_vec_f128 rx_cvt_packed_int_vec_f128(const void* addr) { + rx_vec_f128 x; + x.lo = (double)unsigned32ToSigned2sCompl(load32((uint8_t*)addr + 0)); + x.hi = (double)unsigned32ToSigned2sCompl(load32((uint8_t*)addr + 4)); + return x; +} + +#define RANDOMX_DEFAULT_FENV + +void rx_reset_float_state(); + +void rx_set_rounding_mode(uint32_t mode); + +#endif + +#ifndef HAVE_AES +static const char* platformError = "Platform doesn't support hardware AES"; + +#include + +FORCE_INLINE rx_vec_i128 rx_aesenc_vec_i128(rx_vec_i128 v, rx_vec_i128 rkey) { + throw std::runtime_error(platformError); +} + +FORCE_INLINE rx_vec_i128 rx_aesdec_vec_i128(rx_vec_i128 v, rx_vec_i128 rkey) { + throw std::runtime_error(platformError); +} +#endif + +double loadDoublePortable(const void* addr); +uint64_t mulh(uint64_t, uint64_t); +int64_t smulh(int64_t, int64_t); +uint64_t rotl64(uint64_t, unsigned int); +uint64_t rotr64(uint64_t, unsigned int); diff --git a/src/crypto/randomx/jit_compiler.hpp b/src/crypto/randomx/jit_compiler.hpp new file mode 100644 index 000000000..bd9c2b0e4 --- /dev/null +++ b/src/crypto/randomx/jit_compiler.hpp @@ -0,0 +1,37 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#if defined(_M_X64) || defined(__x86_64__) +#include "jit_compiler_x86.hpp" +#elif defined(__aarch64__) +#include "jit_compiler_a64.hpp" +#else +#include "jit_compiler_fallback.hpp" +#endif diff --git a/src/crypto/randomx/jit_compiler_a64.hpp b/src/crypto/randomx/jit_compiler_a64.hpp new file mode 100644 index 000000000..58aa25c4d --- /dev/null +++ b/src/crypto/randomx/jit_compiler_a64.hpp @@ -0,0 +1,73 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include +#include +#include +#include "common.hpp" + +namespace randomx { + + class Program; + class ProgramConfiguration; + class SuperscalarProgram; + + class JitCompilerA64 { + public: + JitCompilerA64() { + throw std::runtime_error("ARM64 JIT compiler is not implemented yet."); + } + void generateProgram(Program&, ProgramConfiguration&) { + + } + void generateProgramLight(Program&, ProgramConfiguration&, uint32_t) { + + } + template + void generateSuperscalarHash(SuperscalarProgram(&programs)[N], std::vector &) { + + } + void generateDatasetInitCode() { + + } + ProgramFunc* getProgramFunc() { + return nullptr; + } + DatasetInitFunc* getDatasetInitFunc() { + return nullptr; + } + uint8_t* getCode() { + return nullptr; + } + size_t getCodeSize() { + return 0; + } + }; +} \ No newline at end of file diff --git a/src/crypto/randomx/jit_compiler_fallback.hpp b/src/crypto/randomx/jit_compiler_fallback.hpp new file mode 100644 index 000000000..8103a6321 --- /dev/null +++ b/src/crypto/randomx/jit_compiler_fallback.hpp @@ -0,0 +1,73 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include +#include +#include +#include "common.hpp" + +namespace randomx { + + class Program; + class ProgramConfiguration; + class SuperscalarProgram; + + class JitCompilerFallback { + public: + JitCompilerFallback() { + throw std::runtime_error("JIT compilation is not supported on this platform"); + } + void generateProgram(Program&, ProgramConfiguration&) { + + } + void generateProgramLight(Program&, ProgramConfiguration&, uint32_t) { + + } + template + void generateSuperscalarHash(SuperscalarProgram(&programs)[N], std::vector &) { + + } + void generateDatasetInitCode() { + + } + ProgramFunc* getProgramFunc() { + return nullptr; + } + DatasetInitFunc* getDatasetInitFunc() { + return nullptr; + } + uint8_t* getCode() { + return nullptr; + } + size_t getCodeSize() { + return 0; + } + }; +} \ No newline at end of file diff --git a/src/crypto/randomx/jit_compiler_x86.cpp b/src/crypto/randomx/jit_compiler_x86.cpp new file mode 100644 index 000000000..6f04e28a1 --- /dev/null +++ b/src/crypto/randomx/jit_compiler_x86.cpp @@ -0,0 +1,774 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include +#include +#include +#include "jit_compiler_x86.hpp" +#include "jit_compiler_x86_static.hpp" +#include "superscalar.hpp" +#include "program.hpp" +#include "reciprocal.h" +#include "virtual_memory.hpp" + +namespace randomx { + /* + + REGISTER ALLOCATION: + + ; rax -> temporary + ; rbx -> iteration counter "ic" + ; rcx -> temporary + ; rdx -> temporary + ; rsi -> scratchpad pointer + ; rdi -> dataset pointer + ; rbp -> memory registers "ma" (high 32 bits), "mx" (low 32 bits) + ; rsp -> stack pointer + ; r8 -> "r0" + ; r9 -> "r1" + ; r10 -> "r2" + ; r11 -> "r3" + ; r12 -> "r4" + ; r13 -> "r5" + ; r14 -> "r6" + ; r15 -> "r7" + ; xmm0 -> "f0" + ; xmm1 -> "f1" + ; xmm2 -> "f2" + ; xmm3 -> "f3" + ; xmm4 -> "e0" + ; xmm5 -> "e1" + ; xmm6 -> "e2" + ; xmm7 -> "e3" + ; xmm8 -> "a0" + ; xmm9 -> "a1" + ; xmm10 -> "a2" + ; xmm11 -> "a3" + ; xmm12 -> temporary + ; xmm13 -> E 'and' mask = 0x00ffffffffffffff00ffffffffffffff + ; xmm14 -> E 'or' mask = 0x3*00000000******3*00000000****** + ; xmm15 -> scale mask = 0x81f000000000000081f0000000000000 + + */ + + const uint8_t* codePrologue = (uint8_t*)&randomx_program_prologue; + const uint8_t* codeLoopBegin = (uint8_t*)&randomx_program_loop_begin; + const uint8_t* codeLoopLoad = (uint8_t*)&randomx_program_loop_load; + const uint8_t* codeProgamStart = (uint8_t*)&randomx_program_start; + const uint8_t* codeReadDataset = (uint8_t*)&randomx_program_read_dataset; + const uint8_t* codeReadDatasetLightSshInit = (uint8_t*)&randomx_program_read_dataset_sshash_init; + const uint8_t* codeReadDatasetLightSshFin = (uint8_t*)&randomx_program_read_dataset_sshash_fin; + const uint8_t* codeDatasetInit = (uint8_t*)&randomx_dataset_init; + const uint8_t* codeLoopStore = (uint8_t*)&randomx_program_loop_store; + const uint8_t* codeLoopEnd = (uint8_t*)&randomx_program_loop_end; + const uint8_t* codeEpilogue = (uint8_t*)&randomx_program_epilogue; + const uint8_t* codeProgramEnd = (uint8_t*)&randomx_program_end; + const uint8_t* codeShhLoad = (uint8_t*)&randomx_sshash_load; + const uint8_t* codeShhPrefetch = (uint8_t*)&randomx_sshash_prefetch; + const uint8_t* codeShhEnd = (uint8_t*)&randomx_sshash_end; + const uint8_t* codeShhInit = (uint8_t*)&randomx_sshash_init; + + const int32_t prologueSize = codeLoopBegin - codePrologue; + const int32_t loopLoadSize = codeProgamStart - codeLoopLoad; + const int32_t readDatasetSize = codeReadDatasetLightSshInit - codeReadDataset; + const int32_t readDatasetLightInitSize = codeReadDatasetLightSshFin - codeReadDatasetLightSshInit; + const int32_t readDatasetLightFinSize = codeLoopStore - codeReadDatasetLightSshFin; + const int32_t loopStoreSize = codeLoopEnd - codeLoopStore; + const int32_t datasetInitSize = codeEpilogue - codeDatasetInit; + const int32_t epilogueSize = codeShhLoad - codeEpilogue; + const int32_t codeSshLoadSize = codeShhPrefetch - codeShhLoad; + const int32_t codeSshPrefetchSize = codeShhEnd - codeShhPrefetch; + const int32_t codeSshInitSize = codeProgramEnd - codeShhInit; + + const int32_t epilogueOffset = CodeSize - epilogueSize; + constexpr int32_t superScalarHashOffset = 32768; + + static const uint8_t REX_ADD_RR[] = { 0x4d, 0x03 }; + static const uint8_t REX_ADD_RM[] = { 0x4c, 0x03 }; + static const uint8_t REX_SUB_RR[] = { 0x4d, 0x2b }; + static const uint8_t REX_SUB_RM[] = { 0x4c, 0x2b }; + static const uint8_t REX_MOV_RR[] = { 0x41, 0x8b }; + static const uint8_t REX_MOV_RR64[] = { 0x49, 0x8b }; + static const uint8_t REX_MOV_R64R[] = { 0x4c, 0x8b }; + static const uint8_t REX_IMUL_RR[] = { 0x4d, 0x0f, 0xaf }; + static const uint8_t REX_IMUL_RRI[] = { 0x4d, 0x69 }; + static const uint8_t REX_IMUL_RM[] = { 0x4c, 0x0f, 0xaf }; + static const uint8_t REX_MUL_R[] = { 0x49, 0xf7 }; + static const uint8_t REX_MUL_M[] = { 0x48, 0xf7 }; + static const uint8_t REX_81[] = { 0x49, 0x81 }; + static const uint8_t AND_EAX_I = 0x25; + static const uint8_t MOV_EAX_I = 0xb8; + static const uint8_t MOV_RAX_I[] = { 0x48, 0xb8 }; + static const uint8_t MOV_RCX_I[] = { 0x48, 0xb9 }; + static const uint8_t REX_LEA[] = { 0x4f, 0x8d }; + static const uint8_t REX_MUL_MEM[] = { 0x48, 0xf7, 0x24, 0x0e }; + static const uint8_t REX_IMUL_MEM[] = { 0x48, 0xf7, 0x2c, 0x0e }; + static const uint8_t REX_SHR_RAX[] = { 0x48, 0xc1, 0xe8 }; + static const uint8_t RAX_ADD_SBB_1[] = { 0x48, 0x83, 0xC0, 0x01, 0x48, 0x83, 0xD8, 0x00 }; + static const uint8_t MUL_RCX[] = { 0x48, 0xf7, 0xe1 }; + static const uint8_t REX_SHR_RDX[] = { 0x48, 0xc1, 0xea }; + static const uint8_t REX_SH[] = { 0x49, 0xc1 }; + static const uint8_t MOV_RCX_RAX_SAR_RCX_63[] = { 0x48, 0x89, 0xc1, 0x48, 0xc1, 0xf9, 0x3f }; + static const uint8_t AND_ECX_I[] = { 0x81, 0xe1 }; + static const uint8_t ADD_RAX_RCX[] = { 0x48, 0x01, 0xC8 }; + static const uint8_t SAR_RAX_I8[] = { 0x48, 0xC1, 0xF8 }; + static const uint8_t NEG_RAX[] = { 0x48, 0xF7, 0xD8 }; + static const uint8_t ADD_R_RAX[] = { 0x4C, 0x03 }; + static const uint8_t XOR_EAX_EAX[] = { 0x33, 0xC0 }; + static const uint8_t ADD_RDX_R[] = { 0x4c, 0x01 }; + static const uint8_t SUB_RDX_R[] = { 0x4c, 0x29 }; + static const uint8_t SAR_RDX_I8[] = { 0x48, 0xC1, 0xFA }; + static const uint8_t TEST_RDX_RDX[] = { 0x48, 0x85, 0xD2 }; + static const uint8_t SETS_AL_ADD_RDX_RAX[] = { 0x0F, 0x98, 0xC0, 0x48, 0x03, 0xD0 }; + static const uint8_t REX_NEG[] = { 0x49, 0xF7 }; + static const uint8_t REX_XOR_RR[] = { 0x4D, 0x33 }; + static const uint8_t REX_XOR_RI[] = { 0x49, 0x81 }; + static const uint8_t REX_XOR_RM[] = { 0x4c, 0x33 }; + static const uint8_t REX_ROT_CL[] = { 0x49, 0xd3 }; + static const uint8_t REX_ROT_I8[] = { 0x49, 0xc1 }; + static const uint8_t SHUFPD[] = { 0x66, 0x0f, 0xc6 }; + static const uint8_t REX_ADDPD[] = { 0x66, 0x41, 0x0f, 0x58 }; + static const uint8_t REX_CVTDQ2PD_XMM12[] = { 0xf3, 0x44, 0x0f, 0xe6, 0x24, 0x06 }; + static const uint8_t REX_SUBPD[] = { 0x66, 0x41, 0x0f, 0x5c }; + static const uint8_t REX_XORPS[] = { 0x41, 0x0f, 0x57 }; + static const uint8_t REX_MULPD[] = { 0x66, 0x41, 0x0f, 0x59 }; + static const uint8_t REX_MAXPD[] = { 0x66, 0x41, 0x0f, 0x5f }; + static const uint8_t REX_DIVPD[] = { 0x66, 0x41, 0x0f, 0x5e }; + static const uint8_t SQRTPD[] = { 0x66, 0x0f, 0x51 }; + static const uint8_t AND_OR_MOV_LDMXCSR[] = { 0x25, 0x00, 0x60, 0x00, 0x00, 0x0D, 0xC0, 0x9F, 0x00, 0x00, 0x50, 0x0F, 0xAE, 0x14, 0x24, 0x58 }; + static const uint8_t ROL_RAX[] = { 0x48, 0xc1, 0xc0 }; + static const uint8_t XOR_ECX_ECX[] = { 0x33, 0xC9 }; + static const uint8_t REX_CMP_R32I[] = { 0x41, 0x81 }; + static const uint8_t REX_CMP_M32I[] = { 0x81, 0x3c, 0x06 }; + static const uint8_t MOVAPD[] = { 0x66, 0x0f, 0x29 }; + static const uint8_t REX_MOV_MR[] = { 0x4c, 0x89 }; + static const uint8_t REX_XOR_EAX[] = { 0x41, 0x33 }; + static const uint8_t SUB_EBX[] = { 0x83, 0xEB, 0x01 }; + static const uint8_t JNZ[] = { 0x0f, 0x85 }; + static const uint8_t JMP = 0xe9; + static const uint8_t REX_XOR_RAX_R64[] = { 0x49, 0x33 }; + static const uint8_t REX_XCHG[] = { 0x4d, 0x87 }; + static const uint8_t REX_ANDPS_XMM12[] = { 0x45, 0x0F, 0x54, 0xE5, 0x45, 0x0F, 0x56, 0xE6 }; + static const uint8_t REX_PADD[] = { 0x66, 0x44, 0x0f }; + static const uint8_t PADD_OPCODES[] = { 0xfc, 0xfd, 0xfe, 0xd4 }; + static const uint8_t CALL = 0xe8; + static const uint8_t REX_ADD_I[] = { 0x49, 0x81 }; + static const uint8_t REX_TEST[] = { 0x49, 0xF7 }; + static const uint8_t JZ[] = { 0x0f, 0x84 }; + static const uint8_t RET = 0xc3; + static const uint8_t LEA_32[] = { 0x67, 0x41, 0x8d }; + static const uint8_t MOVNTI[] = { 0x4c, 0x0f, 0xc3 }; + static const uint8_t ADD_EBX_I[] = { 0x81, 0xc3 }; + + static const uint8_t NOP1[] = { 0x90 }; + static const uint8_t NOP2[] = { 0x66, 0x90 }; + static const uint8_t NOP3[] = { 0x66, 0x66, 0x90 }; + static const uint8_t NOP4[] = { 0x0F, 0x1F, 0x40, 0x00 }; + static const uint8_t NOP5[] = { 0x0F, 0x1F, 0x44, 0x00, 0x00 }; + static const uint8_t NOP6[] = { 0x66, 0x0F, 0x1F, 0x44, 0x00, 0x00 }; + static const uint8_t NOP7[] = { 0x0F, 0x1F, 0x80, 0x00, 0x00, 0x00, 0x00 }; + static const uint8_t NOP8[] = { 0x0F, 0x1F, 0x84, 0x00, 0x00, 0x00, 0x00, 0x00 }; + +// static const uint8_t* NOPX[] = { NOP1, NOP2, NOP3, NOP4, NOP5, NOP6, NOP7, NOP8 }; + + size_t JitCompilerX86::getCodeSize() { + return codePos - prologueSize; + } + + JitCompilerX86::JitCompilerX86() { + code = (uint8_t*)allocExecutableMemory(CodeSize); + memcpy(code, codePrologue, prologueSize); + memcpy(code + epilogueOffset, codeEpilogue, epilogueSize); + } + + JitCompilerX86::~JitCompilerX86() { + freePagedMemory(code, CodeSize); + } + + void JitCompilerX86::generateProgram(Program& prog, ProgramConfiguration& pcfg) { + generateProgramPrologue(prog, pcfg); + memcpy(code + codePos, RandomX_CurrentConfig.codeReadDatasetTweaked, readDatasetSize); + codePos += readDatasetSize; + generateProgramEpilogue(prog); + } + + void JitCompilerX86::generateProgramLight(Program& prog, ProgramConfiguration& pcfg, uint32_t datasetOffset) { + generateProgramPrologue(prog, pcfg); + emit(RandomX_CurrentConfig.codeReadDatasetLightSshInitTweaked, readDatasetLightInitSize); + emit(ADD_EBX_I); + emit32(datasetOffset / CacheLineSize); + emitByte(CALL); + emit32(superScalarHashOffset - (codePos + 4)); + emit(codeReadDatasetLightSshFin, readDatasetLightFinSize); + generateProgramEpilogue(prog); + } + + template + void JitCompilerX86::generateSuperscalarHash(SuperscalarProgram(&programs)[N], std::vector &reciprocalCache) { + memcpy(code + superScalarHashOffset, codeShhInit, codeSshInitSize); + codePos = superScalarHashOffset + codeSshInitSize; + for (unsigned j = 0; j < RandomX_CurrentConfig.CacheAccesses; ++j) { + SuperscalarProgram& prog = programs[j]; + for (unsigned i = 0; i < prog.getSize(); ++i) { + Instruction& instr = prog(i); + generateSuperscalarCode(instr, reciprocalCache); + } + emit(codeShhLoad, codeSshLoadSize); + if (j < RandomX_CurrentConfig.CacheAccesses - 1) { + emit(REX_MOV_RR64); + emitByte(0xd8 + prog.getAddressRegister()); + emit(RandomX_CurrentConfig.codeShhPrefetchTweaked, codeSshPrefetchSize); +#ifdef RANDOMX_ALIGN + int align = (codePos % 16); + while (align != 0) { + int nopSize = 16 - align; + if (nopSize > 8) nopSize = 8; + emit(NOPX[nopSize - 1], nopSize); + align = (codePos % 16); + } +#endif + } + } + emitByte(RET); + } + + template + void JitCompilerX86::generateSuperscalarHash(SuperscalarProgram(&programs)[RANDOMX_CACHE_MAX_ACCESSES], std::vector &reciprocalCache); + + void JitCompilerX86::generateDatasetInitCode() { + memcpy(code, codeDatasetInit, datasetInitSize); + } + + void JitCompilerX86::generateProgramPrologue(Program& prog, ProgramConfiguration& pcfg) { + instructionOffsets.clear(); + for (unsigned i = 0; i < 8; ++i) { + registerUsage[i] = -1; + } + codePos = prologueSize; + memcpy(code + codePos - 48, &pcfg.eMask, sizeof(pcfg.eMask)); + emit(REX_XOR_RAX_R64); + emitByte(0xc0 + pcfg.readReg0); + emit(REX_XOR_RAX_R64); + emitByte(0xc0 + pcfg.readReg1); + memcpy(code + codePos, RandomX_CurrentConfig.codeLoopLoadTweaked, loopLoadSize); + codePos += loopLoadSize; + for (unsigned i = 0; i < prog.getSize(); ++i) { + Instruction& instr = prog(i); + instr.src %= RegistersCount; + instr.dst %= RegistersCount; + generateCode(instr, i); + } + emit(REX_MOV_RR); + emitByte(0xc0 + pcfg.readReg2); + emit(REX_XOR_EAX); + emitByte(0xc0 + pcfg.readReg3); + } + + void JitCompilerX86::generateProgramEpilogue(Program& prog) { + memcpy(code + codePos, codeLoopStore, loopStoreSize); + codePos += loopStoreSize; + emit(SUB_EBX); + emit(JNZ); + emit32(prologueSize - codePos - 4); + emitByte(JMP); + emit32(epilogueOffset - codePos - 4); + } + + void JitCompilerX86::generateCode(Instruction& instr, int i) { + instructionOffsets.push_back(codePos); + auto generator = engine[instr.opcode]; + (this->*generator)(instr, i); + } + + void JitCompilerX86::generateSuperscalarCode(Instruction& instr, std::vector &reciprocalCache) { + switch ((SuperscalarInstructionType)instr.opcode) + { + case randomx::SuperscalarInstructionType::ISUB_R: + emit(REX_SUB_RR); + emitByte(0xc0 + 8 * instr.dst + instr.src); + break; + case randomx::SuperscalarInstructionType::IXOR_R: + emit(REX_XOR_RR); + emitByte(0xc0 + 8 * instr.dst + instr.src); + break; + case randomx::SuperscalarInstructionType::IADD_RS: + emit(REX_LEA); + emitByte(0x04 + 8 * instr.dst); + genSIB(instr.getModShift(), instr.src, instr.dst); + break; + case randomx::SuperscalarInstructionType::IMUL_R: + emit(REX_IMUL_RR); + emitByte(0xc0 + 8 * instr.dst + instr.src); + break; + case randomx::SuperscalarInstructionType::IROR_C: + emit(REX_ROT_I8); + emitByte(0xc8 + instr.dst); + emitByte(instr.getImm32() & 63); + break; + case randomx::SuperscalarInstructionType::IADD_C7: + emit(REX_81); + emitByte(0xc0 + instr.dst); + emit32(instr.getImm32()); + break; + case randomx::SuperscalarInstructionType::IXOR_C7: + emit(REX_XOR_RI); + emitByte(0xf0 + instr.dst); + emit32(instr.getImm32()); + break; + case randomx::SuperscalarInstructionType::IADD_C8: + emit(REX_81); + emitByte(0xc0 + instr.dst); + emit32(instr.getImm32()); +#ifdef RANDOMX_ALIGN + emit(NOP1); +#endif + break; + case randomx::SuperscalarInstructionType::IXOR_C8: + emit(REX_XOR_RI); + emitByte(0xf0 + instr.dst); + emit32(instr.getImm32()); +#ifdef RANDOMX_ALIGN + emit(NOP1); +#endif + break; + case randomx::SuperscalarInstructionType::IADD_C9: + emit(REX_81); + emitByte(0xc0 + instr.dst); + emit32(instr.getImm32()); +#ifdef RANDOMX_ALIGN + emit(NOP2); +#endif + break; + case randomx::SuperscalarInstructionType::IXOR_C9: + emit(REX_XOR_RI); + emitByte(0xf0 + instr.dst); + emit32(instr.getImm32()); +#ifdef RANDOMX_ALIGN + emit(NOP2); +#endif + break; + case randomx::SuperscalarInstructionType::IMULH_R: + emit(REX_MOV_RR64); + emitByte(0xc0 + instr.dst); + emit(REX_MUL_R); + emitByte(0xe0 + instr.src); + emit(REX_MOV_R64R); + emitByte(0xc2 + 8 * instr.dst); + break; + case randomx::SuperscalarInstructionType::ISMULH_R: + emit(REX_MOV_RR64); + emitByte(0xc0 + instr.dst); + emit(REX_MUL_R); + emitByte(0xe8 + instr.src); + emit(REX_MOV_R64R); + emitByte(0xc2 + 8 * instr.dst); + break; + case randomx::SuperscalarInstructionType::IMUL_RCP: + emit(MOV_RAX_I); + emit64(reciprocalCache[instr.getImm32()]); + emit(REX_IMUL_RM); + emitByte(0xc0 + 8 * instr.dst); + break; + default: + UNREACHABLE; + } + } + + void JitCompilerX86::genAddressReg(Instruction& instr, bool rax = true) { + emit(LEA_32); + emitByte(0x80 + instr.src + (rax ? 0 : 8)); + if (instr.src == RegisterNeedsSib) { + emitByte(0x24); + } + emit32(instr.getImm32()); + if (rax) + emitByte(AND_EAX_I); + else + emit(AND_ECX_I); + emit32(instr.getModMem() ? ScratchpadL1Mask : ScratchpadL2Mask); + } + + void JitCompilerX86::genAddressRegDst(Instruction& instr) { + emit(LEA_32); + emitByte(0x80 + instr.dst); + if (instr.dst == RegisterNeedsSib) { + emitByte(0x24); + } + emit32(instr.getImm32()); + emitByte(AND_EAX_I); + if (instr.getModCond() < StoreL3Condition) { + emit32(instr.getModMem() ? ScratchpadL1Mask : ScratchpadL2Mask); + } + else { + emit32(ScratchpadL3Mask); + } + } + + void JitCompilerX86::genAddressImm(Instruction& instr) { + emit32(instr.getImm32() & ScratchpadL3Mask); + } + + void JitCompilerX86::h_IADD_RS(Instruction& instr, int i) { + registerUsage[instr.dst] = i; + emit(REX_LEA); + if (instr.dst == RegisterNeedsDisplacement) + emitByte(0xac); + else + emitByte(0x04 + 8 * instr.dst); + genSIB(instr.getModShift(), instr.src, instr.dst); + if (instr.dst == RegisterNeedsDisplacement) + emit32(instr.getImm32()); + } + + void JitCompilerX86::h_IADD_M(Instruction& instr, int i) { + registerUsage[instr.dst] = i; + if (instr.src != instr.dst) { + genAddressReg(instr); + emit(REX_ADD_RM); + emitByte(0x04 + 8 * instr.dst); + emitByte(0x06); + } + else { + emit(REX_ADD_RM); + emitByte(0x86 + 8 * instr.dst); + genAddressImm(instr); + } + } + + void JitCompilerX86::genSIB(int scale, int index, int base) { + emitByte((scale << 6) | (index << 3) | base); + } + + void JitCompilerX86::h_ISUB_R(Instruction& instr, int i) { + registerUsage[instr.dst] = i; + if (instr.src != instr.dst) { + emit(REX_SUB_RR); + emitByte(0xc0 + 8 * instr.dst + instr.src); + } + else { + emit(REX_81); + emitByte(0xe8 + instr.dst); + emit32(instr.getImm32()); + } + } + + void JitCompilerX86::h_ISUB_M(Instruction& instr, int i) { + registerUsage[instr.dst] = i; + if (instr.src != instr.dst) { + genAddressReg(instr); + emit(REX_SUB_RM); + emitByte(0x04 + 8 * instr.dst); + emitByte(0x06); + } + else { + emit(REX_SUB_RM); + emitByte(0x86 + 8 * instr.dst); + genAddressImm(instr); + } + } + + void JitCompilerX86::h_IMUL_R(Instruction& instr, int i) { + registerUsage[instr.dst] = i; + if (instr.src != instr.dst) { + emit(REX_IMUL_RR); + emitByte(0xc0 + 8 * instr.dst + instr.src); + } + else { + emit(REX_IMUL_RRI); + emitByte(0xc0 + 9 * instr.dst); + emit32(instr.getImm32()); + } + } + + void JitCompilerX86::h_IMUL_M(Instruction& instr, int i) { + registerUsage[instr.dst] = i; + if (instr.src != instr.dst) { + genAddressReg(instr); + emit(REX_IMUL_RM); + emitByte(0x04 + 8 * instr.dst); + emitByte(0x06); + } + else { + emit(REX_IMUL_RM); + emitByte(0x86 + 8 * instr.dst); + genAddressImm(instr); + } + } + + void JitCompilerX86::h_IMULH_R(Instruction& instr, int i) { + registerUsage[instr.dst] = i; + emit(REX_MOV_RR64); + emitByte(0xc0 + instr.dst); + emit(REX_MUL_R); + emitByte(0xe0 + instr.src); + emit(REX_MOV_R64R); + emitByte(0xc2 + 8 * instr.dst); + } + + void JitCompilerX86::h_IMULH_M(Instruction& instr, int i) { + registerUsage[instr.dst] = i; + if (instr.src != instr.dst) { + genAddressReg(instr, false); + emit(REX_MOV_RR64); + emitByte(0xc0 + instr.dst); + emit(REX_MUL_MEM); + } + else { + emit(REX_MOV_RR64); + emitByte(0xc0 + instr.dst); + emit(REX_MUL_M); + emitByte(0xa6); + genAddressImm(instr); + } + emit(REX_MOV_R64R); + emitByte(0xc2 + 8 * instr.dst); + } + + void JitCompilerX86::h_ISMULH_R(Instruction& instr, int i) { + registerUsage[instr.dst] = i; + emit(REX_MOV_RR64); + emitByte(0xc0 + instr.dst); + emit(REX_MUL_R); + emitByte(0xe8 + instr.src); + emit(REX_MOV_R64R); + emitByte(0xc2 + 8 * instr.dst); + } + + void JitCompilerX86::h_ISMULH_M(Instruction& instr, int i) { + registerUsage[instr.dst] = i; + if (instr.src != instr.dst) { + genAddressReg(instr, false); + emit(REX_MOV_RR64); + emitByte(0xc0 + instr.dst); + emit(REX_IMUL_MEM); + } + else { + emit(REX_MOV_RR64); + emitByte(0xc0 + instr.dst); + emit(REX_MUL_M); + emitByte(0xae); + genAddressImm(instr); + } + emit(REX_MOV_R64R); + emitByte(0xc2 + 8 * instr.dst); + } + + void JitCompilerX86::h_IMUL_RCP(Instruction& instr, int i) { + uint64_t divisor = instr.getImm32(); + if (!isPowerOf2(divisor)) { + registerUsage[instr.dst] = i; + emit(MOV_RAX_I); + emit64(randomx_reciprocal_fast(divisor)); + emit(REX_IMUL_RM); + emitByte(0xc0 + 8 * instr.dst); + } + } + + void JitCompilerX86::h_INEG_R(Instruction& instr, int i) { + registerUsage[instr.dst] = i; + emit(REX_NEG); + emitByte(0xd8 + instr.dst); + } + + void JitCompilerX86::h_IXOR_R(Instruction& instr, int i) { + registerUsage[instr.dst] = i; + if (instr.src != instr.dst) { + emit(REX_XOR_RR); + emitByte(0xc0 + 8 * instr.dst + instr.src); + } + else { + emit(REX_XOR_RI); + emitByte(0xf0 + instr.dst); + emit32(instr.getImm32()); + } + } + + void JitCompilerX86::h_IXOR_M(Instruction& instr, int i) { + registerUsage[instr.dst] = i; + if (instr.src != instr.dst) { + genAddressReg(instr); + emit(REX_XOR_RM); + emitByte(0x04 + 8 * instr.dst); + emitByte(0x06); + } + else { + emit(REX_XOR_RM); + emitByte(0x86 + 8 * instr.dst); + genAddressImm(instr); + } + } + + void JitCompilerX86::h_IROR_R(Instruction& instr, int i) { + registerUsage[instr.dst] = i; + if (instr.src != instr.dst) { + emit(REX_MOV_RR); + emitByte(0xc8 + instr.src); + emit(REX_ROT_CL); + emitByte(0xc8 + instr.dst); + } + else { + emit(REX_ROT_I8); + emitByte(0xc8 + instr.dst); + emitByte(instr.getImm32() & 63); + } + } + + void JitCompilerX86::h_IROL_R(Instruction& instr, int i) { + registerUsage[instr.dst] = i; + if (instr.src != instr.dst) { + emit(REX_MOV_RR); + emitByte(0xc8 + instr.src); + emit(REX_ROT_CL); + emitByte(0xc0 + instr.dst); + } + else { + emit(REX_ROT_I8); + emitByte(0xc0 + instr.dst); + emitByte(instr.getImm32() & 63); + } + } + + void JitCompilerX86::h_ISWAP_R(Instruction& instr, int i) { + if (instr.src != instr.dst) { + registerUsage[instr.dst] = i; + registerUsage[instr.src] = i; + emit(REX_XCHG); + emitByte(0xc0 + instr.src + 8 * instr.dst); + } + } + + void JitCompilerX86::h_FSWAP_R(Instruction& instr, int i) { + emit(SHUFPD); + emitByte(0xc0 + 9 * instr.dst); + emitByte(1); + } + + void JitCompilerX86::h_FADD_R(Instruction& instr, int i) { + instr.dst %= RegisterCountFlt; + instr.src %= RegisterCountFlt; + emit(REX_ADDPD); + emitByte(0xc0 + instr.src + 8 * instr.dst); + } + + void JitCompilerX86::h_FADD_M(Instruction& instr, int i) { + instr.dst %= RegisterCountFlt; + genAddressReg(instr); + emit(REX_CVTDQ2PD_XMM12); + emit(REX_ADDPD); + emitByte(0xc4 + 8 * instr.dst); + } + + void JitCompilerX86::h_FSUB_R(Instruction& instr, int i) { + instr.dst %= RegisterCountFlt; + instr.src %= RegisterCountFlt; + emit(REX_SUBPD); + emitByte(0xc0 + instr.src + 8 * instr.dst); + } + + void JitCompilerX86::h_FSUB_M(Instruction& instr, int i) { + instr.dst %= RegisterCountFlt; + genAddressReg(instr); + emit(REX_CVTDQ2PD_XMM12); + emit(REX_SUBPD); + emitByte(0xc4 + 8 * instr.dst); + } + + void JitCompilerX86::h_FSCAL_R(Instruction& instr, int i) { + instr.dst %= RegisterCountFlt; + emit(REX_XORPS); + emitByte(0xc7 + 8 * instr.dst); + } + + void JitCompilerX86::h_FMUL_R(Instruction& instr, int i) { + instr.dst %= RegisterCountFlt; + instr.src %= RegisterCountFlt; + emit(REX_MULPD); + emitByte(0xe0 + instr.src + 8 * instr.dst); + } + + void JitCompilerX86::h_FDIV_M(Instruction& instr, int i) { + instr.dst %= RegisterCountFlt; + genAddressReg(instr); + emit(REX_CVTDQ2PD_XMM12); + emit(REX_ANDPS_XMM12); + emit(REX_DIVPD); + emitByte(0xe4 + 8 * instr.dst); + } + + void JitCompilerX86::h_FSQRT_R(Instruction& instr, int i) { + instr.dst %= RegisterCountFlt; + emit(SQRTPD); + emitByte(0xe4 + 9 * instr.dst); + } + + void JitCompilerX86::h_CFROUND(Instruction& instr, int i) { + emit(REX_MOV_RR64); + emitByte(0xc0 + instr.src); + int rotate = (13 - (instr.getImm32() & 63)) & 63; + if (rotate != 0) { + emit(ROL_RAX); + emitByte(rotate); + } + emit(AND_OR_MOV_LDMXCSR); + } + + void JitCompilerX86::h_CBRANCH(Instruction& instr, int i) { + int reg = instr.dst; + int target = registerUsage[reg] + 1; + emit(REX_ADD_I); + emitByte(0xc0 + reg); + int shift = instr.getModCond() + RandomX_CurrentConfig.JumpOffset; + uint32_t imm = instr.getImm32() | (1UL << shift); + if (RandomX_CurrentConfig.JumpOffset > 0 || shift > 0) + imm &= ~(1UL << (shift - 1)); + emit32(imm); + emit(REX_TEST); + emitByte(0xc0 + reg); + emit32(RandomX_CurrentConfig.ConditionMask_Calculated << shift); + emit(JZ); + emit32(instructionOffsets[target] - (codePos + 4)); + //mark all registers as used + for (unsigned j = 0; j < RegistersCount; ++j) { + registerUsage[j] = i; + } + } + + void JitCompilerX86::h_ISTORE(Instruction& instr, int i) { + genAddressRegDst(instr); + emit(REX_MOV_MR); + emitByte(0x04 + 8 * instr.src); + emitByte(0x06); + } + + void JitCompilerX86::h_NOP(Instruction& instr, int i) { + emit(NOP1); + } + + InstructionGeneratorX86 JitCompilerX86::engine[256] = {}; + +} diff --git a/src/crypto/randomx/jit_compiler_x86.hpp b/src/crypto/randomx/jit_compiler_x86.hpp new file mode 100644 index 000000000..a81c8f4e6 --- /dev/null +++ b/src/crypto/randomx/jit_compiler_x86.hpp @@ -0,0 +1,141 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include +#include +#include +#include "common.hpp" + +namespace randomx { + + class Program; + class ProgramConfiguration; + class SuperscalarProgram; + class JitCompilerX86; + class Instruction; + + typedef void(JitCompilerX86::*InstructionGeneratorX86)(Instruction&, int); + + constexpr uint32_t CodeSize = 64 * 1024; + + class JitCompilerX86 { + public: + JitCompilerX86(); + ~JitCompilerX86(); + void generateProgram(Program&, ProgramConfiguration&); + void generateProgramLight(Program&, ProgramConfiguration&, uint32_t); + template + void generateSuperscalarHash(SuperscalarProgram (&programs)[N], std::vector &); + void generateDatasetInitCode(); + ProgramFunc* getProgramFunc() { + return (ProgramFunc*)code; + } + DatasetInitFunc* getDatasetInitFunc() { + return (DatasetInitFunc*)code; + } + uint8_t* getCode() { + return code; + } + size_t getCodeSize(); + + static InstructionGeneratorX86 engine[256]; + std::vector instructionOffsets; + int registerUsage[RegistersCount]; + uint8_t* code; + int32_t codePos; + + void generateProgramPrologue(Program&, ProgramConfiguration&); + void generateProgramEpilogue(Program&); + void genAddressReg(Instruction&, bool); + void genAddressRegDst(Instruction&); + void genAddressImm(Instruction&); + void genSIB(int scale, int index, int base); + + void generateCode(Instruction&, int); + void generateSuperscalarCode(Instruction &, std::vector &); + + void emitByte(uint8_t val) { + code[codePos] = val; + codePos++; + } + + void emit32(uint32_t val) { + memcpy(code + codePos, &val, sizeof val); + codePos += sizeof val; + } + + void emit64(uint64_t val) { + memcpy(code + codePos, &val, sizeof val); + codePos += sizeof val; + } + + template + void emit(const uint8_t (&src)[N]) { + emit(src, N); + } + + void emit(const uint8_t* src, size_t count) { + memcpy(code + codePos, src, count); + codePos += count; + } + + void h_IADD_RS(Instruction&, int); + void h_IADD_M(Instruction&, int); + void h_ISUB_R(Instruction&, int); + void h_ISUB_M(Instruction&, int); + void h_IMUL_R(Instruction&, int); + void h_IMUL_M(Instruction&, int); + void h_IMULH_R(Instruction&, int); + void h_IMULH_M(Instruction&, int); + void h_ISMULH_R(Instruction&, int); + void h_ISMULH_M(Instruction&, int); + void h_IMUL_RCP(Instruction&, int); + void h_INEG_R(Instruction&, int); + void h_IXOR_R(Instruction&, int); + void h_IXOR_M(Instruction&, int); + void h_IROR_R(Instruction&, int); + void h_IROL_R(Instruction&, int); + void h_ISWAP_R(Instruction&, int); + void h_FSWAP_R(Instruction&, int); + void h_FADD_R(Instruction&, int); + void h_FADD_M(Instruction&, int); + void h_FSUB_R(Instruction&, int); + void h_FSUB_M(Instruction&, int); + void h_FSCAL_R(Instruction&, int); + void h_FMUL_R(Instruction&, int); + void h_FDIV_M(Instruction&, int); + void h_FSQRT_R(Instruction&, int); + void h_CBRANCH(Instruction&, int); + void h_CFROUND(Instruction&, int); + void h_ISTORE(Instruction&, int); + void h_NOP(Instruction&, int); + }; + +} \ No newline at end of file diff --git a/src/crypto/randomx/jit_compiler_x86_static.S b/src/crypto/randomx/jit_compiler_x86_static.S new file mode 100644 index 000000000..67d2bdbc8 --- /dev/null +++ b/src/crypto/randomx/jit_compiler_x86_static.S @@ -0,0 +1,208 @@ +# Copyright (c) 2018-2019, tevador +# +# All rights reserved. +# +# Redistribution and use in source and binary forms, with or without +# modification, are permitted provided that the following conditions are met: +# * Redistributions of source code must retain the above copyright +# notice, this list of conditions and the following disclaimer. +# * Redistributions in binary form must reproduce the above copyright +# notice, this list of conditions and the following disclaimer in the +# documentation and/or other materials provided with the distribution. +# * Neither the name of the copyright holder nor the +# names of its contributors may be used to endorse or promote products +# derived from this software without specific prior written permission. +# +# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +# ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +# WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +.intel_syntax noprefix +#if defined(__APPLE__) +.text +#define DECL(x) _##x +#else +.section .text +#define DECL(x) x +#endif + +#if defined(__WIN32__) || defined(__CYGWIN__) +#define WINABI +#endif + +.global DECL(randomx_program_prologue) +.global DECL(randomx_program_loop_begin) +.global DECL(randomx_program_loop_load) +.global DECL(randomx_program_start) +.global DECL(randomx_program_read_dataset) +.global DECL(randomx_program_read_dataset_sshash_init) +.global DECL(randomx_program_read_dataset_sshash_fin) +.global DECL(randomx_program_loop_store) +.global DECL(randomx_program_loop_end) +.global DECL(randomx_dataset_init) +.global DECL(randomx_program_epilogue) +.global DECL(randomx_sshash_load) +.global DECL(randomx_sshash_prefetch) +.global DECL(randomx_sshash_end) +.global DECL(randomx_sshash_init) +.global DECL(randomx_program_end) +.global DECL(randomx_reciprocal_fast) + +#define RANDOMX_SCRATCHPAD_MASK 2097088 +#define RANDOMX_DATASET_BASE_MASK 2147483584 +#define RANDOMX_CACHE_MASK 4194303 + +#define db .byte + +.balign 64 +DECL(randomx_program_prologue): +#if defined(WINABI) + #include "asm/program_prologue_win64.inc" +#else + #include "asm/program_prologue_linux.inc" +#endif + movapd xmm13, xmmword ptr [mantissaMask+rip] + movapd xmm14, xmmword ptr [exp240+rip] + movapd xmm15, xmmword ptr [scaleMask+rip] + jmp DECL(randomx_program_loop_begin) + +.balign 64 + #include "asm/program_xmm_constants.inc" + +.balign 64 +DECL(randomx_program_loop_begin): + nop + +DECL(randomx_program_loop_load): + #include "asm/program_loop_load.inc" + +DECL(randomx_program_start): + nop + +DECL(randomx_program_read_dataset): + #include "asm/program_read_dataset.inc" + +DECL(randomx_program_read_dataset_sshash_init): + #include "asm/program_read_dataset_sshash_init.inc" + +DECL(randomx_program_read_dataset_sshash_fin): + #include "asm/program_read_dataset_sshash_fin.inc" + +DECL(randomx_program_loop_store): + #include "asm/program_loop_store.inc" + +DECL(randomx_program_loop_end): + nop + +.balign 64 +DECL(randomx_dataset_init): + push rbx + push rbp + push r12 + push r13 + push r14 + push r15 +#if defined(WINABI) + push rdi + push rsi + mov rdi, qword ptr [rcx] ;# cache->memory + mov rsi, rdx ;# dataset + mov rbp, r8 ;# block index + push r9 ;# max. block index +#else + mov rdi, qword ptr [rdi] ;# cache->memory + ;# dataset in rsi + mov rbp, rdx ;# block index + push rcx ;# max. block index +#endif +init_block_loop: + prefetchw byte ptr [rsi] + mov rbx, rbp + .byte 232 ;# 0xE8 = call + ;# .set CALL_LOC, + .int 32768 - (call_offset - DECL(randomx_dataset_init)) +call_offset: + mov qword ptr [rsi+0], r8 + mov qword ptr [rsi+8], r9 + mov qword ptr [rsi+16], r10 + mov qword ptr [rsi+24], r11 + mov qword ptr [rsi+32], r12 + mov qword ptr [rsi+40], r13 + mov qword ptr [rsi+48], r14 + mov qword ptr [rsi+56], r15 + add rbp, 1 + add rsi, 64 + cmp rbp, qword ptr [rsp] + jb init_block_loop + pop rax +#if defined(WINABI) + pop rsi + pop rdi +#endif + pop r15 + pop r14 + pop r13 + pop r12 + pop rbp + pop rbx + ret + +.balign 64 +DECL(randomx_program_epilogue): + #include "asm/program_epilogue_store.inc" +#if defined(WINABI) + #include "asm/program_epilogue_win64.inc" +#else + #include "asm/program_epilogue_linux.inc" +#endif + +.balign 64 +DECL(randomx_sshash_load): + #include "asm/program_sshash_load.inc" + +DECL(randomx_sshash_prefetch): + #include "asm/program_sshash_prefetch.inc" + +DECL(randomx_sshash_end): + nop + +.balign 64 +DECL(randomx_sshash_init): + lea r8, [rbx+1] + #include "asm/program_sshash_prefetch.inc" + imul r8, qword ptr [r0_mul+rip] + mov r9, qword ptr [r1_add+rip] + xor r9, r8 + mov r10, qword ptr [r2_add+rip] + xor r10, r8 + mov r11, qword ptr [r3_add+rip] + xor r11, r8 + mov r12, qword ptr [r4_add+rip] + xor r12, r8 + mov r13, qword ptr [r5_add+rip] + xor r13, r8 + mov r14, qword ptr [r6_add+rip] + xor r14, r8 + mov r15, qword ptr [r7_add+rip] + xor r15, r8 + jmp DECL(randomx_program_end) + +.balign 64 + #include "asm/program_sshash_constants.inc" + +.balign 64 +DECL(randomx_program_end): + nop + +DECL(randomx_reciprocal_fast): +#if !defined(WINABI) + mov rcx, rdi +#endif + #include "asm/randomx_reciprocal.inc" diff --git a/src/crypto/randomx/jit_compiler_x86_static.asm b/src/crypto/randomx/jit_compiler_x86_static.asm new file mode 100644 index 000000000..5ecfb435a --- /dev/null +++ b/src/crypto/randomx/jit_compiler_x86_static.asm @@ -0,0 +1,199 @@ +; Copyright (c) 2018-2019, tevador +; +; All rights reserved. +; +; Redistribution and use in source and binary forms, with or without +; modification, are permitted provided that the following conditions are met: +; * Redistributions of source code must retain the above copyright +; notice, this list of conditions and the following disclaimer. +; * Redistributions in binary form must reproduce the above copyright +; notice, this list of conditions and the following disclaimer in the +; documentation and/or other materials provided with the distribution. +; * Neither the name of the copyright holder nor the +; names of its contributors may be used to endorse or promote products +; derived from this software without specific prior written permission. +; +; THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +; ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +; WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +; DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +; FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +; DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +; SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +; CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +; OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +; OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + +IFDEF RAX + +_RANDOMX_JITX86_STATIC SEGMENT PAGE READ EXECUTE + +PUBLIC randomx_program_prologue +PUBLIC randomx_program_loop_begin +PUBLIC randomx_program_loop_load +PUBLIC randomx_program_start +PUBLIC randomx_program_read_dataset +PUBLIC randomx_program_read_dataset_sshash_init +PUBLIC randomx_program_read_dataset_sshash_fin +PUBLIC randomx_dataset_init +PUBLIC randomx_program_loop_store +PUBLIC randomx_program_loop_end +PUBLIC randomx_program_epilogue +PUBLIC randomx_sshash_load +PUBLIC randomx_sshash_prefetch +PUBLIC randomx_sshash_end +PUBLIC randomx_sshash_init +PUBLIC randomx_program_end +PUBLIC randomx_reciprocal_fast + +RANDOMX_SCRATCHPAD_MASK EQU 2097088 +RANDOMX_DATASET_BASE_MASK EQU 2147483584 +RANDOMX_CACHE_MASK EQU 4194303 + +ALIGN 64 +randomx_program_prologue PROC + include asm/program_prologue_win64.inc + movapd xmm13, xmmword ptr [mantissaMask] + movapd xmm14, xmmword ptr [exp240] + movapd xmm15, xmmword ptr [scaleMask] + jmp randomx_program_loop_begin +randomx_program_prologue ENDP + +ALIGN 64 + include asm/program_xmm_constants.inc + +ALIGN 64 +randomx_program_loop_begin PROC + nop +randomx_program_loop_begin ENDP + +randomx_program_loop_load PROC + include asm/program_loop_load.inc +randomx_program_loop_load ENDP + +randomx_program_start PROC + nop +randomx_program_start ENDP + +randomx_program_read_dataset PROC + include asm/program_read_dataset.inc +randomx_program_read_dataset ENDP + +randomx_program_read_dataset_sshash_init PROC + include asm/program_read_dataset_sshash_init.inc +randomx_program_read_dataset_sshash_init ENDP + +randomx_program_read_dataset_sshash_fin PROC + include asm/program_read_dataset_sshash_fin.inc +randomx_program_read_dataset_sshash_fin ENDP + +randomx_program_loop_store PROC + include asm/program_loop_store.inc +randomx_program_loop_store ENDP + +randomx_program_loop_end PROC + nop +randomx_program_loop_end ENDP + +ALIGN 64 +randomx_dataset_init PROC + push rbx + push rbp + push rdi + push rsi + push r12 + push r13 + push r14 + push r15 + mov rdi, qword ptr [rcx] ;# cache->memory + mov rsi, rdx ;# dataset + mov rbp, r8 ;# block index + push r9 ;# max. block index +init_block_loop: + prefetchw byte ptr [rsi] + mov rbx, rbp + db 232 ;# 0xE8 = call + dd 32768 - distance + distance equ $ - offset randomx_dataset_init + mov qword ptr [rsi+0], r8 + mov qword ptr [rsi+8], r9 + mov qword ptr [rsi+16], r10 + mov qword ptr [rsi+24], r11 + mov qword ptr [rsi+32], r12 + mov qword ptr [rsi+40], r13 + mov qword ptr [rsi+48], r14 + mov qword ptr [rsi+56], r15 + add rbp, 1 + add rsi, 64 + cmp rbp, qword ptr [rsp] + jb init_block_loop + pop r9 + pop r15 + pop r14 + pop r13 + pop r12 + pop rsi + pop rdi + pop rbp + pop rbx + ret +randomx_dataset_init ENDP + +ALIGN 64 +randomx_program_epilogue PROC + include asm/program_epilogue_store.inc + include asm/program_epilogue_win64.inc +randomx_program_epilogue ENDP + +ALIGN 64 +randomx_sshash_load PROC + include asm/program_sshash_load.inc +randomx_sshash_load ENDP + +randomx_sshash_prefetch PROC + include asm/program_sshash_prefetch.inc +randomx_sshash_prefetch ENDP + +randomx_sshash_end PROC + nop +randomx_sshash_end ENDP + +ALIGN 64 +randomx_sshash_init PROC + lea r8, [rbx+1] + include asm/program_sshash_prefetch.inc + imul r8, qword ptr [r0_mul] + mov r9, qword ptr [r1_add] + xor r9, r8 + mov r10, qword ptr [r2_add] + xor r10, r8 + mov r11, qword ptr [r3_add] + xor r11, r8 + mov r12, qword ptr [r4_add] + xor r12, r8 + mov r13, qword ptr [r5_add] + xor r13, r8 + mov r14, qword ptr [r6_add] + xor r14, r8 + mov r15, qword ptr [r7_add] + xor r15, r8 + jmp randomx_program_end +randomx_sshash_init ENDP + +ALIGN 64 + include asm/program_sshash_constants.inc + +ALIGN 64 +randomx_program_end PROC + nop +randomx_program_end ENDP + +randomx_reciprocal_fast PROC + include asm/randomx_reciprocal.inc +randomx_reciprocal_fast ENDP + +_RANDOMX_JITX86_STATIC ENDS + +ENDIF + +END \ No newline at end of file diff --git a/src/crypto/randomx/jit_compiler_x86_static.hpp b/src/crypto/randomx/jit_compiler_x86_static.hpp new file mode 100644 index 000000000..ba1968627 --- /dev/null +++ b/src/crypto/randomx/jit_compiler_x86_static.hpp @@ -0,0 +1,48 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +extern "C" { + void randomx_program_prologue(); + void randomx_program_loop_begin(); + void randomx_program_loop_load(); + void randomx_program_start(); + void randomx_program_read_dataset(); + void randomx_program_read_dataset_sshash_init(); + void randomx_program_read_dataset_sshash_fin(); + void randomx_program_loop_store(); + void randomx_program_loop_end(); + void randomx_dataset_init(); + void randomx_program_epilogue(); + void randomx_sshash_load(); + void randomx_sshash_prefetch(); + void randomx_sshash_end(); + void randomx_sshash_init(); + void randomx_program_end(); +} diff --git a/src/crypto/randomx/program.hpp b/src/crypto/randomx/program.hpp new file mode 100644 index 000000000..ef8ac748a --- /dev/null +++ b/src/crypto/randomx/program.hpp @@ -0,0 +1,60 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include +#include "common.hpp" +#include "instruction.hpp" +#include "blake2/endian.h" + +namespace randomx { + + struct ProgramConfiguration { + uint64_t eMask[2]; + uint32_t readReg0, readReg1, readReg2, readReg3; + }; + + class Program { + public: + Instruction& operator()(int pc) { + return programBuffer[pc]; + } + uint64_t getEntropy(int i) { + return load64(&entropyBuffer[i]); + } + uint32_t getSize() { + return RandomX_CurrentConfig.ProgramSize; + } + private: + uint64_t entropyBuffer[16]; + Instruction programBuffer[RANDOMX_PROGRAM_MAX_SIZE]; + }; + + static_assert(sizeof(Program) % 64 == 0, "Invalid size of class randomx::Program"); +} diff --git a/src/crypto/randomx/randomx.cpp b/src/crypto/randomx/randomx.cpp new file mode 100644 index 000000000..9e88bc6db --- /dev/null +++ b/src/crypto/randomx/randomx.cpp @@ -0,0 +1,444 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include "randomx.h" +#include "dataset.hpp" +#include "vm_interpreted.hpp" +#include "vm_interpreted_light.hpp" +#include "vm_compiled.hpp" +#include "vm_compiled_light.hpp" +#include "blake2/blake2.h" +#include "jit_compiler_x86_static.hpp" +#include + +RandomX_ConfigurationWownero::RandomX_ConfigurationWownero() +{ + ArgonSalt = "RandomWOW\x01"; + ProgramIterations = 1024; + ProgramCount = 16; + ScratchpadL2_Size = 131072; + ScratchpadL3_Size = 1048576; + + RANDOMX_FREQ_IROR_R = 10; + RANDOMX_FREQ_IROL_R = 0; + RANDOMX_FREQ_FSWAP_R = 8; + RANDOMX_FREQ_FADD_R = 20; + RANDOMX_FREQ_FSUB_R = 20; + RANDOMX_FREQ_FMUL_R = 20; + + fillAes4Rx4_Key[0] = rx_set_int_vec_i128(0xcf359e95, 0x141f82b7, 0x7ffbe4a6, 0xf890465d); + fillAes4Rx4_Key[1] = rx_set_int_vec_i128(0x6741ffdc, 0xbd5c5ac3, 0xfee8278a, 0x6a55c450); + fillAes4Rx4_Key[2] = rx_set_int_vec_i128(0x3d324aac, 0xa7279ad2, 0xd524fde4, 0x114c47a4); + fillAes4Rx4_Key[3] = rx_set_int_vec_i128(0x76f6db08, 0x42d3dbd9, 0x99a9aeff, 0x810c3a2a); + fillAes4Rx4_Key[4] = fillAes4Rx4_Key[0]; + fillAes4Rx4_Key[5] = fillAes4Rx4_Key[1]; + fillAes4Rx4_Key[6] = fillAes4Rx4_Key[2]; + fillAes4Rx4_Key[7] = fillAes4Rx4_Key[3]; +} + +RandomX_ConfigurationLoki::RandomX_ConfigurationLoki() +{ + ArgonIterations = 4; + ArgonLanes = 2; + ArgonSalt = "RandomXL\x12"; + ProgramSize = 320; + ProgramCount = 7; +} + +RandomX_ConfigurationBase::RandomX_ConfigurationBase() + : ArgonMemory(262144) + , ArgonIterations(3) + , ArgonLanes(1) + , ArgonSalt("RandomX\x03") + , CacheAccesses(8) + , SuperscalarLatency(170) + , DatasetBaseSize(2147483648) + , DatasetExtraSize(33554368) + , ScratchpadL1_Size(16384) + , ScratchpadL2_Size(262144) + , ScratchpadL3_Size(2097152) + , ProgramSize(256) + , ProgramIterations(2048) + , ProgramCount(8) + , JumpBits(8) + , JumpOffset(8) + , RANDOMX_FREQ_IADD_RS(25) + , RANDOMX_FREQ_IADD_M(7) + , RANDOMX_FREQ_ISUB_R(16) + , RANDOMX_FREQ_ISUB_M(7) + , RANDOMX_FREQ_IMUL_R(16) + , RANDOMX_FREQ_IMUL_M(4) + , RANDOMX_FREQ_IMULH_R(4) + , RANDOMX_FREQ_IMULH_M(1) + , RANDOMX_FREQ_ISMULH_R(4) + , RANDOMX_FREQ_ISMULH_M(1) + , RANDOMX_FREQ_IMUL_RCP(8) + , RANDOMX_FREQ_INEG_R(2) + , RANDOMX_FREQ_IXOR_R(15) + , RANDOMX_FREQ_IXOR_M(5) + , RANDOMX_FREQ_IROR_R(8) + , RANDOMX_FREQ_IROL_R(2) + , RANDOMX_FREQ_ISWAP_R(4) + , RANDOMX_FREQ_FSWAP_R(4) + , RANDOMX_FREQ_FADD_R(16) + , RANDOMX_FREQ_FADD_M(5) + , RANDOMX_FREQ_FSUB_R(16) + , RANDOMX_FREQ_FSUB_M(5) + , RANDOMX_FREQ_FSCAL_R(6) + , RANDOMX_FREQ_FMUL_R(32) + , RANDOMX_FREQ_FDIV_M(4) + , RANDOMX_FREQ_FSQRT_R(6) + , RANDOMX_FREQ_CBRANCH(16) + , RANDOMX_FREQ_CFROUND(1) + , RANDOMX_FREQ_ISTORE(16) + , RANDOMX_FREQ_NOP(0) +{ + fillAes4Rx4_Key[0] = rx_set_int_vec_i128(0x99e5d23f, 0x2f546d2b, 0xd1833ddb, 0x6421aadd); + fillAes4Rx4_Key[1] = rx_set_int_vec_i128(0xa5dfcde5, 0x06f79d53, 0xb6913f55, 0xb20e3450); + fillAes4Rx4_Key[2] = rx_set_int_vec_i128(0x171c02bf, 0x0aa4679f, 0x515e7baf, 0x5c3ed904); + fillAes4Rx4_Key[3] = rx_set_int_vec_i128(0xd8ded291, 0xcd673785, 0xe78f5d08, 0x85623763); + fillAes4Rx4_Key[4] = rx_set_int_vec_i128(0x229effb4, 0x3d518b6d, 0xe3d6a7a6, 0xb5826f73); + fillAes4Rx4_Key[5] = rx_set_int_vec_i128(0xb272b7d2, 0xe9024d4e, 0x9c10b3d9, 0xc7566bf3); + fillAes4Rx4_Key[6] = rx_set_int_vec_i128(0xf63befa7, 0x2ba9660a, 0xf765a38b, 0xf273c9e7); + fillAes4Rx4_Key[7] = rx_set_int_vec_i128(0xc0b0762d, 0x0c06d1fd, 0x915839de, 0x7a7cd609); + +#if defined(_M_X64) || defined(__x86_64__) + { + const uint8_t* a = (const uint8_t*)&randomx_sshash_prefetch; + const uint8_t* b = (const uint8_t*)&randomx_sshash_end; + memcpy(codeShhPrefetchTweaked, a, b - a); + } + { + const uint8_t* a = (const uint8_t*)&randomx_program_read_dataset; + const uint8_t* b = (const uint8_t*)&randomx_program_read_dataset_sshash_init; + memcpy(codeReadDatasetTweaked, a, b - a); + } + { + const uint8_t* a = (const uint8_t*)&randomx_program_read_dataset_sshash_init; + const uint8_t* b = (const uint8_t*)&randomx_program_read_dataset_sshash_fin; + memcpy(codeReadDatasetLightSshInitTweaked, a, b - a); + } + { + const uint8_t* a = (const uint8_t*)&randomx_program_loop_load; + const uint8_t* b = (const uint8_t*)&randomx_program_start; + memcpy(codeLoopLoadTweaked, a, b - a); + } +#endif +} + +void RandomX_ConfigurationBase::Apply() +{ +#if defined(_M_X64) || defined(__x86_64__) + *(uint32_t*)(codeShhPrefetchTweaked + 3) = ArgonMemory * 16 - 1; + const uint32_t DatasetBaseMask = DatasetBaseSize - RANDOMX_DATASET_ITEM_SIZE; + *(uint32_t*)(codeReadDatasetTweaked + 7) = DatasetBaseMask; + *(uint32_t*)(codeReadDatasetTweaked + 23) = DatasetBaseMask; + *(uint32_t*)(codeReadDatasetLightSshInitTweaked + 59) = DatasetBaseMask; +#endif + + CacheLineAlignMask_Calculated = (DatasetBaseSize - 1) & ~(RANDOMX_DATASET_ITEM_SIZE - 1); + DatasetExtraItems_Calculated = DatasetExtraSize / RANDOMX_DATASET_ITEM_SIZE; + + ScratchpadL1Mask_Calculated = (ScratchpadL1_Size / sizeof(uint64_t) - 1) * 8; + ScratchpadL1Mask16_Calculated = (ScratchpadL1_Size / sizeof(uint64_t) / 2 - 1) * 16; + ScratchpadL2Mask_Calculated = (ScratchpadL2_Size / sizeof(uint64_t) - 1) * 8; + ScratchpadL2Mask16_Calculated = (ScratchpadL2_Size / sizeof(uint64_t) / 2 - 1) * 16; + ScratchpadL3Mask_Calculated = (((ScratchpadL3_Size / sizeof(uint64_t)) - 1) * 8); + ScratchpadL3Mask64_Calculated = ((ScratchpadL3_Size / sizeof(uint64_t)) / 8 - 1) * 64; + +#if defined(_M_X64) || defined(__x86_64__) + *(uint32_t*)(codeLoopLoadTweaked + 4) = ScratchpadL3Mask64_Calculated; + *(uint32_t*)(codeLoopLoadTweaked + 50) = ScratchpadL3Mask64_Calculated; +#endif + + ConditionMask_Calculated = (1 << JumpBits) - 1; + + constexpr int CEIL_NULL = 0; + int k = 0; + +#if defined(_M_X64) || defined(__x86_64__) +#define JIT_HANDLE(x, prev) randomx::JitCompilerX86::engine[k] = &randomx::JitCompilerX86::h_##x +#else +#define JIT_HANDLE(x, prev) +#endif + +#define INST_HANDLE(x, prev) \ + CEIL_##x = CEIL_##prev + RANDOMX_FREQ_##x; \ + for (; k < CEIL_##x; ++k) { JIT_HANDLE(x, prev); } + + INST_HANDLE(IADD_RS, NULL); + INST_HANDLE(IADD_M, IADD_RS); + INST_HANDLE(ISUB_R, IADD_M); + INST_HANDLE(ISUB_M, ISUB_R); + INST_HANDLE(IMUL_R, ISUB_M); + INST_HANDLE(IMUL_M, IMUL_R); + INST_HANDLE(IMULH_R, IMUL_M); + INST_HANDLE(IMULH_M, IMULH_R); + INST_HANDLE(ISMULH_R, IMULH_M); + INST_HANDLE(ISMULH_M, ISMULH_R); + INST_HANDLE(IMUL_RCP, ISMULH_M); + INST_HANDLE(INEG_R, IMUL_RCP); + INST_HANDLE(IXOR_R, INEG_R); + INST_HANDLE(IXOR_M, IXOR_R); + INST_HANDLE(IROR_R, IXOR_M); + INST_HANDLE(IROL_R, IROR_R); + INST_HANDLE(ISWAP_R, IROL_R); + INST_HANDLE(FSWAP_R, ISWAP_R); + INST_HANDLE(FADD_R, FSWAP_R); + INST_HANDLE(FADD_M, FADD_R); + INST_HANDLE(FSUB_R, FADD_M); + INST_HANDLE(FSUB_M, FSUB_R); + INST_HANDLE(FSCAL_R, FSUB_M); + INST_HANDLE(FMUL_R, FSCAL_R); + INST_HANDLE(FDIV_M, FMUL_R); + INST_HANDLE(FSQRT_R, FDIV_M); + INST_HANDLE(CBRANCH, FSQRT_R); + INST_HANDLE(CFROUND, CBRANCH); + INST_HANDLE(ISTORE, CFROUND); + INST_HANDLE(NOP, ISTORE); +#undef INST_HANDLE +} + +RandomX_ConfigurationMonero RandomX_MoneroConfig; +RandomX_ConfigurationWownero RandomX_WowneroConfig; +RandomX_ConfigurationLoki RandomX_LokiConfig; + +RandomX_ConfigurationBase RandomX_CurrentConfig; + +extern "C" { + + randomx_cache *randomx_alloc_cache(randomx_flags flags) { + randomx_cache *cache = nullptr; + + try { + cache = new randomx_cache(); + switch (flags & (RANDOMX_FLAG_JIT | RANDOMX_FLAG_LARGE_PAGES)) { + case RANDOMX_FLAG_DEFAULT: + cache->dealloc = &randomx::deallocCache; + cache->jit = nullptr; + cache->initialize = &randomx::initCache; + cache->datasetInit = &randomx::initDataset; + cache->memory = (uint8_t*)randomx::DefaultAllocator::allocMemory(RANDOMX_CACHE_MAX_SIZE); + break; + + case RANDOMX_FLAG_JIT: + cache->dealloc = &randomx::deallocCache; + cache->jit = new randomx::JitCompiler(); + cache->initialize = &randomx::initCacheCompile; + cache->datasetInit = cache->jit->getDatasetInitFunc(); + cache->memory = (uint8_t*)randomx::DefaultAllocator::allocMemory(RANDOMX_CACHE_MAX_SIZE); + break; + + case RANDOMX_FLAG_LARGE_PAGES: + cache->dealloc = &randomx::deallocCache; + cache->jit = nullptr; + cache->initialize = &randomx::initCache; + cache->datasetInit = &randomx::initDataset; + cache->memory = (uint8_t*)randomx::LargePageAllocator::allocMemory(RANDOMX_CACHE_MAX_SIZE); + break; + + case RANDOMX_FLAG_JIT | RANDOMX_FLAG_LARGE_PAGES: + cache->dealloc = &randomx::deallocCache; + cache->jit = new randomx::JitCompiler(); + cache->initialize = &randomx::initCacheCompile; + cache->datasetInit = cache->jit->getDatasetInitFunc(); + cache->memory = (uint8_t*)randomx::LargePageAllocator::allocMemory(RANDOMX_CACHE_MAX_SIZE); + break; + + default: + UNREACHABLE; + } + } + catch (std::exception &ex) { + if (cache != nullptr) { + randomx_release_cache(cache); + cache = nullptr; + } + } + + return cache; + } + + void randomx_init_cache(randomx_cache *cache, const void *key, size_t keySize) { + assert(cache != nullptr); + assert(keySize == 0 || key != nullptr); + cache->initialize(cache, key, keySize); + } + + void randomx_release_cache(randomx_cache* cache) { + assert(cache != nullptr); + cache->dealloc(cache); + delete cache; + } + + randomx_dataset *randomx_alloc_dataset(randomx_flags flags) { + randomx_dataset *dataset = nullptr; + + try { + dataset = new randomx_dataset(); + if (flags & RANDOMX_FLAG_LARGE_PAGES) { + dataset->dealloc = &randomx::deallocDataset; + dataset->memory = (uint8_t*)randomx::LargePageAllocator::allocMemory(RANDOMX_DATASET_MAX_SIZE); + } + else { + dataset->dealloc = &randomx::deallocDataset; + dataset->memory = (uint8_t*)randomx::DefaultAllocator::allocMemory(RANDOMX_DATASET_MAX_SIZE); + } + } + catch (std::exception &ex) { + if (dataset != nullptr) { + randomx_release_dataset(dataset); + dataset = nullptr; + } + } + + return dataset; + } + + #define DatasetItemCount ((RandomX_CurrentConfig.DatasetBaseSize + RandomX_CurrentConfig.DatasetExtraSize) / RANDOMX_DATASET_ITEM_SIZE) + + unsigned long randomx_dataset_item_count() { + return DatasetItemCount; + } + + void randomx_init_dataset(randomx_dataset *dataset, randomx_cache *cache, unsigned long startItem, unsigned long itemCount) { + assert(dataset != nullptr); + assert(cache != nullptr); + assert(startItem < DatasetItemCount && itemCount <= DatasetItemCount); + assert(startItem + itemCount <= DatasetItemCount); + cache->datasetInit(cache, dataset->memory + startItem * randomx::CacheLineSize, startItem, startItem + itemCount); + } + + void *randomx_get_dataset_memory(randomx_dataset *dataset) { + assert(dataset != nullptr); + return dataset->memory; + } + + void randomx_release_dataset(randomx_dataset *dataset) { + assert(dataset != nullptr); + dataset->dealloc(dataset); + delete dataset; + } + + randomx_vm *randomx_create_vm(randomx_flags flags, randomx_cache *cache, randomx_dataset *dataset, uint8_t *scratchpad) { + assert(cache != nullptr || (flags & RANDOMX_FLAG_FULL_MEM)); + assert(cache == nullptr || cache->isInitialized()); + assert(dataset != nullptr || !(flags & RANDOMX_FLAG_FULL_MEM)); + + randomx_vm *vm = nullptr; + + try { + switch (flags & (RANDOMX_FLAG_FULL_MEM | RANDOMX_FLAG_JIT | RANDOMX_FLAG_HARD_AES)) { + case RANDOMX_FLAG_DEFAULT: + vm = new randomx::InterpretedLightVmDefault(); + break; + + case RANDOMX_FLAG_FULL_MEM: + vm = new randomx::InterpretedVmDefault(); + break; + + case RANDOMX_FLAG_JIT: + vm = new randomx::CompiledLightVmDefault(); + break; + + case RANDOMX_FLAG_FULL_MEM | RANDOMX_FLAG_JIT: + vm = new randomx::CompiledVmDefault(); + break; + + case RANDOMX_FLAG_HARD_AES: + vm = new randomx::InterpretedLightVmHardAes(); + break; + + case RANDOMX_FLAG_FULL_MEM | RANDOMX_FLAG_HARD_AES: + vm = new randomx::InterpretedVmHardAes(); + break; + + case RANDOMX_FLAG_JIT | RANDOMX_FLAG_HARD_AES: + vm = new randomx::CompiledLightVmHardAes(); + break; + + case RANDOMX_FLAG_FULL_MEM | RANDOMX_FLAG_JIT | RANDOMX_FLAG_HARD_AES: + vm = new randomx::CompiledVmHardAes(); + break; + + default: + UNREACHABLE; + } + + if (cache != nullptr) { + vm->setCache(cache); + } + + if (dataset != nullptr) { + vm->setDataset(dataset); + } + + vm->setScratchpad(scratchpad); + } + catch (std::exception &ex) { + delete vm; + vm = nullptr; + } + + return vm; + } + + void randomx_vm_set_cache(randomx_vm *machine, randomx_cache* cache) { + assert(machine != nullptr); + assert(cache != nullptr && cache->isInitialized()); + machine->setCache(cache); + } + + void randomx_vm_set_dataset(randomx_vm *machine, randomx_dataset *dataset) { + assert(machine != nullptr); + assert(dataset != nullptr); + machine->setDataset(dataset); + } + + void randomx_destroy_vm(randomx_vm *machine) { + assert(machine != nullptr); + delete machine; + } + + void randomx_calculate_hash(randomx_vm *machine, const void *input, size_t inputSize, void *output) { + assert(machine != nullptr); + assert(inputSize == 0 || input != nullptr); + assert(output != nullptr); + alignas(16) uint64_t tempHash[8]; + blake2b(tempHash, sizeof(tempHash), input, inputSize, nullptr, 0); + machine->initScratchpad(&tempHash); + machine->resetRoundingMode(); + for (uint32_t chain = 0; chain < RandomX_CurrentConfig.ProgramCount - 1; ++chain) { + machine->run(&tempHash); + blake2b(tempHash, sizeof(tempHash), machine->getRegisterFile(), sizeof(randomx::RegisterFile), nullptr, 0); + } + machine->run(&tempHash); + machine->getFinalResult(output, RANDOMX_HASH_SIZE); + } + +} diff --git a/src/crypto/randomx/randomx.h b/src/crypto/randomx/randomx.h new file mode 100644 index 000000000..d688189fe --- /dev/null +++ b/src/crypto/randomx/randomx.h @@ -0,0 +1,332 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#ifndef RANDOMX_H +#define RANDOMX_H + +#include +#include +#include +#include "intrin_portable.h" + +#define RANDOMX_HASH_SIZE 32 +#define RANDOMX_DATASET_ITEM_SIZE 64 + +#ifndef RANDOMX_EXPORT +#define RANDOMX_EXPORT +#endif + +typedef enum { + RANDOMX_FLAG_DEFAULT = 0, + RANDOMX_FLAG_LARGE_PAGES = 1, + RANDOMX_FLAG_HARD_AES = 2, + RANDOMX_FLAG_FULL_MEM = 4, + RANDOMX_FLAG_JIT = 8, +} randomx_flags; + +typedef struct randomx_dataset randomx_dataset; +typedef struct randomx_cache randomx_cache; +typedef struct randomx_vm randomx_vm; + +struct RandomX_ConfigurationBase +{ + RandomX_ConfigurationBase(); + + void Apply(); + + uint32_t ArgonMemory; + uint32_t ArgonIterations; + uint32_t ArgonLanes; + const char* ArgonSalt; + uint32_t CacheAccesses; + uint32_t SuperscalarLatency; + + uint32_t DatasetBaseSize; + uint32_t DatasetExtraSize; + + uint32_t ScratchpadL1_Size; + uint32_t ScratchpadL2_Size; + uint32_t ScratchpadL3_Size; + + uint32_t ProgramSize; + uint32_t ProgramIterations; + uint32_t ProgramCount; + + uint32_t JumpBits; + uint32_t JumpOffset; + + uint32_t RANDOMX_FREQ_IADD_RS; + uint32_t RANDOMX_FREQ_IADD_M; + uint32_t RANDOMX_FREQ_ISUB_R; + uint32_t RANDOMX_FREQ_ISUB_M; + uint32_t RANDOMX_FREQ_IMUL_R; + uint32_t RANDOMX_FREQ_IMUL_M; + uint32_t RANDOMX_FREQ_IMULH_R; + uint32_t RANDOMX_FREQ_IMULH_M; + uint32_t RANDOMX_FREQ_ISMULH_R; + uint32_t RANDOMX_FREQ_ISMULH_M; + uint32_t RANDOMX_FREQ_IMUL_RCP; + uint32_t RANDOMX_FREQ_INEG_R; + uint32_t RANDOMX_FREQ_IXOR_R; + uint32_t RANDOMX_FREQ_IXOR_M; + uint32_t RANDOMX_FREQ_IROR_R; + uint32_t RANDOMX_FREQ_IROL_R; + uint32_t RANDOMX_FREQ_ISWAP_R; + uint32_t RANDOMX_FREQ_FSWAP_R; + uint32_t RANDOMX_FREQ_FADD_R; + uint32_t RANDOMX_FREQ_FADD_M; + uint32_t RANDOMX_FREQ_FSUB_R; + uint32_t RANDOMX_FREQ_FSUB_M; + uint32_t RANDOMX_FREQ_FSCAL_R; + uint32_t RANDOMX_FREQ_FMUL_R; + uint32_t RANDOMX_FREQ_FDIV_M; + uint32_t RANDOMX_FREQ_FSQRT_R; + uint32_t RANDOMX_FREQ_CBRANCH; + uint32_t RANDOMX_FREQ_CFROUND; + uint32_t RANDOMX_FREQ_ISTORE; + uint32_t RANDOMX_FREQ_NOP; + + rx_vec_i128 fillAes4Rx4_Key[8]; + + uint8_t codeShhPrefetchTweaked[20]; + uint8_t codeReadDatasetTweaked[64]; + uint8_t codeReadDatasetLightSshInitTweaked[68]; + uint8_t codeLoopLoadTweaked[140]; + + uint32_t CacheLineAlignMask_Calculated; + uint32_t DatasetExtraItems_Calculated; + + uint32_t ScratchpadL1Mask_Calculated; + uint32_t ScratchpadL1Mask16_Calculated; + uint32_t ScratchpadL2Mask_Calculated; + uint32_t ScratchpadL2Mask16_Calculated; + uint32_t ScratchpadL3Mask_Calculated; + uint32_t ScratchpadL3Mask64_Calculated; + + uint32_t ConditionMask_Calculated; + + int CEIL_IADD_RS; + int CEIL_IADD_M; + int CEIL_ISUB_R; + int CEIL_ISUB_M; + int CEIL_IMUL_R; + int CEIL_IMUL_M; + int CEIL_IMULH_R; + int CEIL_IMULH_M; + int CEIL_ISMULH_R; + int CEIL_ISMULH_M; + int CEIL_IMUL_RCP; + int CEIL_INEG_R; + int CEIL_IXOR_R; + int CEIL_IXOR_M; + int CEIL_IROR_R; + int CEIL_IROL_R; + int CEIL_ISWAP_R; + int CEIL_FSWAP_R; + int CEIL_FADD_R; + int CEIL_FADD_M; + int CEIL_FSUB_R; + int CEIL_FSUB_M; + int CEIL_FSCAL_R; + int CEIL_FMUL_R; + int CEIL_FDIV_M; + int CEIL_FSQRT_R; + int CEIL_CBRANCH; + int CEIL_CFROUND; + int CEIL_ISTORE; + int CEIL_NOP; +}; + +struct RandomX_ConfigurationMonero : public RandomX_ConfigurationBase {}; +struct RandomX_ConfigurationWownero : public RandomX_ConfigurationBase { RandomX_ConfigurationWownero(); }; +struct RandomX_ConfigurationLoki : public RandomX_ConfigurationBase { RandomX_ConfigurationLoki(); }; + +extern RandomX_ConfigurationMonero RandomX_MoneroConfig; +extern RandomX_ConfigurationWownero RandomX_WowneroConfig; +extern RandomX_ConfigurationLoki RandomX_LokiConfig; + +extern RandomX_ConfigurationBase RandomX_CurrentConfig; + +template +void randomx_apply_config(const T& config) +{ + static_assert(sizeof(T) == sizeof(RandomX_ConfigurationBase), "Invalid RandomX configuration struct size"); + static_assert(std::is_base_of::value, "Incompatible RandomX configuration struct"); + RandomX_CurrentConfig = config; + RandomX_CurrentConfig.Apply(); +} + +#if defined(__cplusplus) +extern "C" { +#endif + +/** + * Creates a randomx_cache structure and allocates memory for RandomX Cache. + * + * @param flags is any combination of these 2 flags (each flag can be set or not set): + * RANDOMX_FLAG_LARGE_PAGES - allocate memory in large pages + * RANDOMX_FLAG_JIT - create cache structure with JIT compilation support; this makes + * subsequent Dataset initialization faster + * + * @return Pointer to an allocated randomx_cache structure. + * NULL is returned if memory allocation fails or if the RANDOMX_FLAG_JIT + * is set and JIT compilation is not supported on the current platform. + */ +RANDOMX_EXPORT randomx_cache *randomx_alloc_cache(randomx_flags flags); + +/** + * Initializes the cache memory and SuperscalarHash using the provided key value. + * + * @param cache is a pointer to a previously allocated randomx_cache structure. Must not be NULL. + * @param key is a pointer to memory which contains the key value. Must not be NULL. + * @param keySize is the number of bytes of the key. +*/ +RANDOMX_EXPORT void randomx_init_cache(randomx_cache *cache, const void *key, size_t keySize); + +/** + * Releases all memory occupied by the randomx_cache structure. + * + * @param cache is a pointer to a previously allocated randomx_cache structure. +*/ +RANDOMX_EXPORT void randomx_release_cache(randomx_cache* cache); + +/** + * Creates a randomx_dataset structure and allocates memory for RandomX Dataset. + * + * @param flags is the initialization flags. Only one flag is supported (can be set or not set): + * RANDOMX_FLAG_LARGE_PAGES - allocate memory in large pages + * + * @return Pointer to an allocated randomx_dataset structure. + * NULL is returned if memory allocation fails. + */ +RANDOMX_EXPORT randomx_dataset *randomx_alloc_dataset(randomx_flags flags); + +/** + * Gets the number of items contained in the dataset. + * + * @return the number of items contained in the dataset. +*/ +RANDOMX_EXPORT unsigned long randomx_dataset_item_count(void); + +/** + * Initializes dataset items. + * + * Note: In order to use the Dataset, all items from 0 to (randomx_dataset_item_count() - 1) must be initialized. + * This may be done by several calls to this function using non-overlapping item sequences. + * + * @param dataset is a pointer to a previously allocated randomx_dataset structure. Must not be NULL. + * @param cache is a pointer to a previously allocated and initialized randomx_cache structure. Must not be NULL. + * @param startItem is the item number where intialization should start. + * @param itemCount is the number of items that should be initialized. +*/ +RANDOMX_EXPORT void randomx_init_dataset(randomx_dataset *dataset, randomx_cache *cache, unsigned long startItem, unsigned long itemCount); + +/** + * Returns a pointer to the internal memory buffer of the dataset structure. The size + * of the internal memory buffer is randomx_dataset_item_count() * RANDOMX_DATASET_ITEM_SIZE. + * + * @param dataset is dataset is a pointer to a previously allocated randomx_dataset structure. Must not be NULL. + * + * @return Pointer to the internal memory buffer of the dataset structure. +*/ +RANDOMX_EXPORT void *randomx_get_dataset_memory(randomx_dataset *dataset); + +/** + * Releases all memory occupied by the randomx_dataset structure. + * + * @param dataset is a pointer to a previously allocated randomx_dataset structure. +*/ +RANDOMX_EXPORT void randomx_release_dataset(randomx_dataset *dataset); + +/** + * Creates and initializes a RandomX virtual machine. + * + * @param flags is any combination of these 4 flags (each flag can be set or not set): + * RANDOMX_FLAG_LARGE_PAGES - allocate scratchpad memory in large pages + * RANDOMX_FLAG_HARD_AES - virtual machine will use hardware accelerated AES + * RANDOMX_FLAG_FULL_MEM - virtual machine will use the full dataset + * RANDOMX_FLAG_JIT - virtual machine will use a JIT compiler + * The numeric values of the flags are ordered so that a higher value will provide + * faster hash calculation and a lower numeric value will provide higher portability. + * Using RANDOMX_FLAG_DEFAULT (all flags not set) works on all platforms, but is the slowest. + * @param cache is a pointer to an initialized randomx_cache structure. Can be + * NULL if RANDOMX_FLAG_FULL_MEM is set. + * @param dataset is a pointer to a randomx_dataset structure. Can be NULL + * if RANDOMX_FLAG_FULL_MEM is not set. + * + * @return Pointer to an initialized randomx_vm structure. + * Returns NULL if: + * (1) Scratchpad memory allocation fails. + * (2) The requested initialization flags are not supported on the current platform. + * (3) cache parameter is NULL and RANDOMX_FLAG_FULL_MEM is not set + * (4) dataset parameter is NULL and RANDOMX_FLAG_FULL_MEM is set +*/ +RANDOMX_EXPORT randomx_vm *randomx_create_vm(randomx_flags flags, randomx_cache *cache, randomx_dataset *dataset, uint8_t *scratchpad); + +/** + * Reinitializes a virtual machine with a new Cache. This function should be called anytime + * the Cache is reinitialized with a new key. + * + * @param machine is a pointer to a randomx_vm structure that was initialized + * without RANDOMX_FLAG_FULL_MEM. Must not be NULL. + * @param cache is a pointer to an initialized randomx_cache structure. Must not be NULL. +*/ +RANDOMX_EXPORT void randomx_vm_set_cache(randomx_vm *machine, randomx_cache* cache); + +/** + * Reinitializes a virtual machine with a new Dataset. + * + * @param machine is a pointer to a randomx_vm structure that was initialized + * with RANDOMX_FLAG_FULL_MEM. Must not be NULL. + * @param dataset is a pointer to an initialized randomx_dataset structure. Must not be NULL. +*/ +RANDOMX_EXPORT void randomx_vm_set_dataset(randomx_vm *machine, randomx_dataset *dataset); + +/** + * Releases all memory occupied by the randomx_vm structure. + * + * @param machine is a pointer to a previously created randomx_vm structure. +*/ +RANDOMX_EXPORT void randomx_destroy_vm(randomx_vm *machine); + +/** + * Calculates a RandomX hash value. + * + * @param machine is a pointer to a randomx_vm structure. Must not be NULL. + * @param input is a pointer to memory to be hashed. Must not be NULL. + * @param inputSize is the number of bytes to be hashed. + * @param output is a pointer to memory where the hash will be stored. Must not + * be NULL and at least RANDOMX_HASH_SIZE bytes must be available for writing. +*/ +RANDOMX_EXPORT void randomx_calculate_hash(randomx_vm *machine, const void *input, size_t inputSize, void *output); + +#if defined(__cplusplus) +} +#endif + +#endif diff --git a/src/crypto/randomx/reciprocal.c b/src/crypto/randomx/reciprocal.c new file mode 100644 index 000000000..22620f53a --- /dev/null +++ b/src/crypto/randomx/reciprocal.c @@ -0,0 +1,80 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include +#include "reciprocal.h" + +/* + Calculates rcp = 2**x / divisor for highest integer x such that rcp < 2**64. + divisor must not be 0 or a power of 2 + + Equivalent x86 assembly (divisor in rcx): + + mov edx, 1 + mov r8, rcx + xor eax, eax + bsr rcx, rcx + shl rdx, cl + div r8 + ret + +*/ +uint64_t randomx_reciprocal(uint64_t divisor) { + + assert(divisor != 0); + + const uint64_t p2exp63 = 1ULL << 63; + + uint64_t quotient = p2exp63 / divisor, remainder = p2exp63 % divisor; + + unsigned bsr = 0; //highest set bit in divisor + + for (uint64_t bit = divisor; bit > 0; bit >>= 1) + bsr++; + + for (unsigned shift = 0; shift < bsr; shift++) { + if (remainder >= divisor - remainder) { + quotient = quotient * 2 + 1; + remainder = remainder * 2 - divisor; + } + else { + quotient = quotient * 2; + remainder = remainder * 2; + } + } + + return quotient; +} + +#if !RANDOMX_HAVE_FAST_RECIPROCAL + +uint64_t randomx_reciprocal_fast(uint64_t divisor) { + return randomx_reciprocal(divisor); +} + +#endif diff --git a/src/crypto/randomx/reciprocal.h b/src/crypto/randomx/reciprocal.h new file mode 100644 index 000000000..8858df2b8 --- /dev/null +++ b/src/crypto/randomx/reciprocal.h @@ -0,0 +1,48 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include + +#if defined(_M_X64) || defined(__x86_64__) +#define RANDOMX_HAVE_FAST_RECIPROCAL 1 +#else +#define RANDOMX_HAVE_FAST_RECIPROCAL 0 +#endif + +#if defined(__cplusplus) +extern "C" { +#endif + +uint64_t randomx_reciprocal(uint64_t); +uint64_t randomx_reciprocal_fast(uint64_t); + +#if defined(__cplusplus) +} +#endif diff --git a/src/crypto/randomx/soft_aes.cpp b/src/crypto/randomx/soft_aes.cpp new file mode 100644 index 000000000..3e82fa2ef --- /dev/null +++ b/src/crypto/randomx/soft_aes.cpp @@ -0,0 +1,364 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include "soft_aes.h" + +alignas(16) const uint8_t sbox[256] = { + 0x63, 0x7c, 0x77, 0x7b, 0xf2, 0x6b, 0x6f, 0xc5, 0x30, 0x01, 0x67, 0x2b, 0xfe, 0xd7, 0xab, 0x76, + 0xca, 0x82, 0xc9, 0x7d, 0xfa, 0x59, 0x47, 0xf0, 0xad, 0xd4, 0xa2, 0xaf, 0x9c, 0xa4, 0x72, 0xc0, + 0xb7, 0xfd, 0x93, 0x26, 0x36, 0x3f, 0xf7, 0xcc, 0x34, 0xa5, 0xe5, 0xf1, 0x71, 0xd8, 0x31, 0x15, + 0x04, 0xc7, 0x23, 0xc3, 0x18, 0x96, 0x05, 0x9a, 0x07, 0x12, 0x80, 0xe2, 0xeb, 0x27, 0xb2, 0x75, + 0x09, 0x83, 0x2c, 0x1a, 0x1b, 0x6e, 0x5a, 0xa0, 0x52, 0x3b, 0xd6, 0xb3, 0x29, 0xe3, 0x2f, 0x84, + 0x53, 0xd1, 0x00, 0xed, 0x20, 0xfc, 0xb1, 0x5b, 0x6a, 0xcb, 0xbe, 0x39, 0x4a, 0x4c, 0x58, 0xcf, + 0xd0, 0xef, 0xaa, 0xfb, 0x43, 0x4d, 0x33, 0x85, 0x45, 0xf9, 0x02, 0x7f, 0x50, 0x3c, 0x9f, 0xa8, + 0x51, 0xa3, 0x40, 0x8f, 0x92, 0x9d, 0x38, 0xf5, 0xbc, 0xb6, 0xda, 0x21, 0x10, 0xff, 0xf3, 0xd2, + 0xcd, 0x0c, 0x13, 0xec, 0x5f, 0x97, 0x44, 0x17, 0xc4, 0xa7, 0x7e, 0x3d, 0x64, 0x5d, 0x19, 0x73, + 0x60, 0x81, 0x4f, 0xdc, 0x22, 0x2a, 0x90, 0x88, 0x46, 0xee, 0xb8, 0x14, 0xde, 0x5e, 0x0b, 0xdb, + 0xe0, 0x32, 0x3a, 0x0a, 0x49, 0x06, 0x24, 0x5c, 0xc2, 0xd3, 0xac, 0x62, 0x91, 0x95, 0xe4, 0x79, + 0xe7, 0xc8, 0x37, 0x6d, 0x8d, 0xd5, 0x4e, 0xa9, 0x6c, 0x56, 0xf4, 0xea, 0x65, 0x7a, 0xae, 0x08, + 0xba, 0x78, 0x25, 0x2e, 0x1c, 0xa6, 0xb4, 0xc6, 0xe8, 0xdd, 0x74, 0x1f, 0x4b, 0xbd, 0x8b, 0x8a, + 0x70, 0x3e, 0xb5, 0x66, 0x48, 0x03, 0xf6, 0x0e, 0x61, 0x35, 0x57, 0xb9, 0x86, 0xc1, 0x1d, 0x9e, + 0xe1, 0xf8, 0x98, 0x11, 0x69, 0xd9, 0x8e, 0x94, 0x9b, 0x1e, 0x87, 0xe9, 0xce, 0x55, 0x28, 0xdf, + 0x8c, 0xa1, 0x89, 0x0d, 0xbf, 0xe6, 0x42, 0x68, 0x41, 0x99, 0x2d, 0x0f, 0xb0, 0x54, 0xbb, 0x16, +}; + +alignas(16) const uint32_t lutEnc0[256] = { + 0xa56363c6, 0x847c7cf8, 0x997777ee, 0x8d7b7bf6, 0x0df2f2ff, 0xbd6b6bd6, 0xb16f6fde, 0x54c5c591, + 0x50303060, 0x03010102, 0xa96767ce, 0x7d2b2b56, 0x19fefee7, 0x62d7d7b5, 0xe6abab4d, 0x9a7676ec, + 0x45caca8f, 0x9d82821f, 0x40c9c989, 0x877d7dfa, 0x15fafaef, 0xeb5959b2, 0xc947478e, 0x0bf0f0fb, + 0xecadad41, 0x67d4d4b3, 0xfda2a25f, 0xeaafaf45, 0xbf9c9c23, 0xf7a4a453, 0x967272e4, 0x5bc0c09b, + 0xc2b7b775, 0x1cfdfde1, 0xae93933d, 0x6a26264c, 0x5a36366c, 0x413f3f7e, 0x02f7f7f5, 0x4fcccc83, + 0x5c343468, 0xf4a5a551, 0x34e5e5d1, 0x08f1f1f9, 0x937171e2, 0x73d8d8ab, 0x53313162, 0x3f15152a, + 0x0c040408, 0x52c7c795, 0x65232346, 0x5ec3c39d, 0x28181830, 0xa1969637, 0x0f05050a, 0xb59a9a2f, + 0x0907070e, 0x36121224, 0x9b80801b, 0x3de2e2df, 0x26ebebcd, 0x6927274e, 0xcdb2b27f, 0x9f7575ea, + 0x1b090912, 0x9e83831d, 0x742c2c58, 0x2e1a1a34, 0x2d1b1b36, 0xb26e6edc, 0xee5a5ab4, 0xfba0a05b, + 0xf65252a4, 0x4d3b3b76, 0x61d6d6b7, 0xceb3b37d, 0x7b292952, 0x3ee3e3dd, 0x712f2f5e, 0x97848413, + 0xf55353a6, 0x68d1d1b9, 0x00000000, 0x2cededc1, 0x60202040, 0x1ffcfce3, 0xc8b1b179, 0xed5b5bb6, + 0xbe6a6ad4, 0x46cbcb8d, 0xd9bebe67, 0x4b393972, 0xde4a4a94, 0xd44c4c98, 0xe85858b0, 0x4acfcf85, + 0x6bd0d0bb, 0x2aefefc5, 0xe5aaaa4f, 0x16fbfbed, 0xc5434386, 0xd74d4d9a, 0x55333366, 0x94858511, + 0xcf45458a, 0x10f9f9e9, 0x06020204, 0x817f7ffe, 0xf05050a0, 0x443c3c78, 0xba9f9f25, 0xe3a8a84b, + 0xf35151a2, 0xfea3a35d, 0xc0404080, 0x8a8f8f05, 0xad92923f, 0xbc9d9d21, 0x48383870, 0x04f5f5f1, + 0xdfbcbc63, 0xc1b6b677, 0x75dadaaf, 0x63212142, 0x30101020, 0x1affffe5, 0x0ef3f3fd, 0x6dd2d2bf, + 0x4ccdcd81, 0x140c0c18, 0x35131326, 0x2fececc3, 0xe15f5fbe, 0xa2979735, 0xcc444488, 0x3917172e, + 0x57c4c493, 0xf2a7a755, 0x827e7efc, 0x473d3d7a, 0xac6464c8, 0xe75d5dba, 0x2b191932, 0x957373e6, + 0xa06060c0, 0x98818119, 0xd14f4f9e, 0x7fdcdca3, 0x66222244, 0x7e2a2a54, 0xab90903b, 0x8388880b, + 0xca46468c, 0x29eeeec7, 0xd3b8b86b, 0x3c141428, 0x79dedea7, 0xe25e5ebc, 0x1d0b0b16, 0x76dbdbad, + 0x3be0e0db, 0x56323264, 0x4e3a3a74, 0x1e0a0a14, 0xdb494992, 0x0a06060c, 0x6c242448, 0xe45c5cb8, + 0x5dc2c29f, 0x6ed3d3bd, 0xefacac43, 0xa66262c4, 0xa8919139, 0xa4959531, 0x37e4e4d3, 0x8b7979f2, + 0x32e7e7d5, 0x43c8c88b, 0x5937376e, 0xb76d6dda, 0x8c8d8d01, 0x64d5d5b1, 0xd24e4e9c, 0xe0a9a949, + 0xb46c6cd8, 0xfa5656ac, 0x07f4f4f3, 0x25eaeacf, 0xaf6565ca, 0x8e7a7af4, 0xe9aeae47, 0x18080810, + 0xd5baba6f, 0x887878f0, 0x6f25254a, 0x722e2e5c, 0x241c1c38, 0xf1a6a657, 0xc7b4b473, 0x51c6c697, + 0x23e8e8cb, 0x7cdddda1, 0x9c7474e8, 0x211f1f3e, 0xdd4b4b96, 0xdcbdbd61, 0x868b8b0d, 0x858a8a0f, + 0x907070e0, 0x423e3e7c, 0xc4b5b571, 0xaa6666cc, 0xd8484890, 0x05030306, 0x01f6f6f7, 0x120e0e1c, + 0xa36161c2, 0x5f35356a, 0xf95757ae, 0xd0b9b969, 0x91868617, 0x58c1c199, 0x271d1d3a, 0xb99e9e27, + 0x38e1e1d9, 0x13f8f8eb, 0xb398982b, 0x33111122, 0xbb6969d2, 0x70d9d9a9, 0x898e8e07, 0xa7949433, + 0xb69b9b2d, 0x221e1e3c, 0x92878715, 0x20e9e9c9, 0x49cece87, 0xff5555aa, 0x78282850, 0x7adfdfa5, + 0x8f8c8c03, 0xf8a1a159, 0x80898909, 0x170d0d1a, 0xdabfbf65, 0x31e6e6d7, 0xc6424284, 0xb86868d0, + 0xc3414182, 0xb0999929, 0x772d2d5a, 0x110f0f1e, 0xcbb0b07b, 0xfc5454a8, 0xd6bbbb6d, 0x3a16162c, +}; + +alignas(16) const uint32_t lutEnc1[256] = { + 0x6363c6a5, 0x7c7cf884, 0x7777ee99, 0x7b7bf68d, 0xf2f2ff0d, 0x6b6bd6bd, 0x6f6fdeb1, 0xc5c59154, + 0x30306050, 0x01010203, 0x6767cea9, 0x2b2b567d, 0xfefee719, 0xd7d7b562, 0xabab4de6, 0x7676ec9a, + 0xcaca8f45, 0x82821f9d, 0xc9c98940, 0x7d7dfa87, 0xfafaef15, 0x5959b2eb, 0x47478ec9, 0xf0f0fb0b, + 0xadad41ec, 0xd4d4b367, 0xa2a25ffd, 0xafaf45ea, 0x9c9c23bf, 0xa4a453f7, 0x7272e496, 0xc0c09b5b, + 0xb7b775c2, 0xfdfde11c, 0x93933dae, 0x26264c6a, 0x36366c5a, 0x3f3f7e41, 0xf7f7f502, 0xcccc834f, + 0x3434685c, 0xa5a551f4, 0xe5e5d134, 0xf1f1f908, 0x7171e293, 0xd8d8ab73, 0x31316253, 0x15152a3f, + 0x0404080c, 0xc7c79552, 0x23234665, 0xc3c39d5e, 0x18183028, 0x969637a1, 0x05050a0f, 0x9a9a2fb5, + 0x07070e09, 0x12122436, 0x80801b9b, 0xe2e2df3d, 0xebebcd26, 0x27274e69, 0xb2b27fcd, 0x7575ea9f, + 0x0909121b, 0x83831d9e, 0x2c2c5874, 0x1a1a342e, 0x1b1b362d, 0x6e6edcb2, 0x5a5ab4ee, 0xa0a05bfb, + 0x5252a4f6, 0x3b3b764d, 0xd6d6b761, 0xb3b37dce, 0x2929527b, 0xe3e3dd3e, 0x2f2f5e71, 0x84841397, + 0x5353a6f5, 0xd1d1b968, 0x00000000, 0xededc12c, 0x20204060, 0xfcfce31f, 0xb1b179c8, 0x5b5bb6ed, + 0x6a6ad4be, 0xcbcb8d46, 0xbebe67d9, 0x3939724b, 0x4a4a94de, 0x4c4c98d4, 0x5858b0e8, 0xcfcf854a, + 0xd0d0bb6b, 0xefefc52a, 0xaaaa4fe5, 0xfbfbed16, 0x434386c5, 0x4d4d9ad7, 0x33336655, 0x85851194, + 0x45458acf, 0xf9f9e910, 0x02020406, 0x7f7ffe81, 0x5050a0f0, 0x3c3c7844, 0x9f9f25ba, 0xa8a84be3, + 0x5151a2f3, 0xa3a35dfe, 0x404080c0, 0x8f8f058a, 0x92923fad, 0x9d9d21bc, 0x38387048, 0xf5f5f104, + 0xbcbc63df, 0xb6b677c1, 0xdadaaf75, 0x21214263, 0x10102030, 0xffffe51a, 0xf3f3fd0e, 0xd2d2bf6d, + 0xcdcd814c, 0x0c0c1814, 0x13132635, 0xececc32f, 0x5f5fbee1, 0x979735a2, 0x444488cc, 0x17172e39, + 0xc4c49357, 0xa7a755f2, 0x7e7efc82, 0x3d3d7a47, 0x6464c8ac, 0x5d5dbae7, 0x1919322b, 0x7373e695, + 0x6060c0a0, 0x81811998, 0x4f4f9ed1, 0xdcdca37f, 0x22224466, 0x2a2a547e, 0x90903bab, 0x88880b83, + 0x46468cca, 0xeeeec729, 0xb8b86bd3, 0x1414283c, 0xdedea779, 0x5e5ebce2, 0x0b0b161d, 0xdbdbad76, + 0xe0e0db3b, 0x32326456, 0x3a3a744e, 0x0a0a141e, 0x494992db, 0x06060c0a, 0x2424486c, 0x5c5cb8e4, + 0xc2c29f5d, 0xd3d3bd6e, 0xacac43ef, 0x6262c4a6, 0x919139a8, 0x959531a4, 0xe4e4d337, 0x7979f28b, + 0xe7e7d532, 0xc8c88b43, 0x37376e59, 0x6d6ddab7, 0x8d8d018c, 0xd5d5b164, 0x4e4e9cd2, 0xa9a949e0, + 0x6c6cd8b4, 0x5656acfa, 0xf4f4f307, 0xeaeacf25, 0x6565caaf, 0x7a7af48e, 0xaeae47e9, 0x08081018, + 0xbaba6fd5, 0x7878f088, 0x25254a6f, 0x2e2e5c72, 0x1c1c3824, 0xa6a657f1, 0xb4b473c7, 0xc6c69751, + 0xe8e8cb23, 0xdddda17c, 0x7474e89c, 0x1f1f3e21, 0x4b4b96dd, 0xbdbd61dc, 0x8b8b0d86, 0x8a8a0f85, + 0x7070e090, 0x3e3e7c42, 0xb5b571c4, 0x6666ccaa, 0x484890d8, 0x03030605, 0xf6f6f701, 0x0e0e1c12, + 0x6161c2a3, 0x35356a5f, 0x5757aef9, 0xb9b969d0, 0x86861791, 0xc1c19958, 0x1d1d3a27, 0x9e9e27b9, + 0xe1e1d938, 0xf8f8eb13, 0x98982bb3, 0x11112233, 0x6969d2bb, 0xd9d9a970, 0x8e8e0789, 0x949433a7, + 0x9b9b2db6, 0x1e1e3c22, 0x87871592, 0xe9e9c920, 0xcece8749, 0x5555aaff, 0x28285078, 0xdfdfa57a, + 0x8c8c038f, 0xa1a159f8, 0x89890980, 0x0d0d1a17, 0xbfbf65da, 0xe6e6d731, 0x424284c6, 0x6868d0b8, + 0x414182c3, 0x999929b0, 0x2d2d5a77, 0x0f0f1e11, 0xb0b07bcb, 0x5454a8fc, 0xbbbb6dd6, 0x16162c3a, +}; + +alignas(16) const uint32_t lutEnc2[256] = { + 0x63c6a563, 0x7cf8847c, 0x77ee9977, 0x7bf68d7b, 0xf2ff0df2, 0x6bd6bd6b, 0x6fdeb16f, 0xc59154c5, + 0x30605030, 0x01020301, 0x67cea967, 0x2b567d2b, 0xfee719fe, 0xd7b562d7, 0xab4de6ab, 0x76ec9a76, + 0xca8f45ca, 0x821f9d82, 0xc98940c9, 0x7dfa877d, 0xfaef15fa, 0x59b2eb59, 0x478ec947, 0xf0fb0bf0, + 0xad41ecad, 0xd4b367d4, 0xa25ffda2, 0xaf45eaaf, 0x9c23bf9c, 0xa453f7a4, 0x72e49672, 0xc09b5bc0, + 0xb775c2b7, 0xfde11cfd, 0x933dae93, 0x264c6a26, 0x366c5a36, 0x3f7e413f, 0xf7f502f7, 0xcc834fcc, + 0x34685c34, 0xa551f4a5, 0xe5d134e5, 0xf1f908f1, 0x71e29371, 0xd8ab73d8, 0x31625331, 0x152a3f15, + 0x04080c04, 0xc79552c7, 0x23466523, 0xc39d5ec3, 0x18302818, 0x9637a196, 0x050a0f05, 0x9a2fb59a, + 0x070e0907, 0x12243612, 0x801b9b80, 0xe2df3de2, 0xebcd26eb, 0x274e6927, 0xb27fcdb2, 0x75ea9f75, + 0x09121b09, 0x831d9e83, 0x2c58742c, 0x1a342e1a, 0x1b362d1b, 0x6edcb26e, 0x5ab4ee5a, 0xa05bfba0, + 0x52a4f652, 0x3b764d3b, 0xd6b761d6, 0xb37dceb3, 0x29527b29, 0xe3dd3ee3, 0x2f5e712f, 0x84139784, + 0x53a6f553, 0xd1b968d1, 0x00000000, 0xedc12ced, 0x20406020, 0xfce31ffc, 0xb179c8b1, 0x5bb6ed5b, + 0x6ad4be6a, 0xcb8d46cb, 0xbe67d9be, 0x39724b39, 0x4a94de4a, 0x4c98d44c, 0x58b0e858, 0xcf854acf, + 0xd0bb6bd0, 0xefc52aef, 0xaa4fe5aa, 0xfbed16fb, 0x4386c543, 0x4d9ad74d, 0x33665533, 0x85119485, + 0x458acf45, 0xf9e910f9, 0x02040602, 0x7ffe817f, 0x50a0f050, 0x3c78443c, 0x9f25ba9f, 0xa84be3a8, + 0x51a2f351, 0xa35dfea3, 0x4080c040, 0x8f058a8f, 0x923fad92, 0x9d21bc9d, 0x38704838, 0xf5f104f5, + 0xbc63dfbc, 0xb677c1b6, 0xdaaf75da, 0x21426321, 0x10203010, 0xffe51aff, 0xf3fd0ef3, 0xd2bf6dd2, + 0xcd814ccd, 0x0c18140c, 0x13263513, 0xecc32fec, 0x5fbee15f, 0x9735a297, 0x4488cc44, 0x172e3917, + 0xc49357c4, 0xa755f2a7, 0x7efc827e, 0x3d7a473d, 0x64c8ac64, 0x5dbae75d, 0x19322b19, 0x73e69573, + 0x60c0a060, 0x81199881, 0x4f9ed14f, 0xdca37fdc, 0x22446622, 0x2a547e2a, 0x903bab90, 0x880b8388, + 0x468cca46, 0xeec729ee, 0xb86bd3b8, 0x14283c14, 0xdea779de, 0x5ebce25e, 0x0b161d0b, 0xdbad76db, + 0xe0db3be0, 0x32645632, 0x3a744e3a, 0x0a141e0a, 0x4992db49, 0x060c0a06, 0x24486c24, 0x5cb8e45c, + 0xc29f5dc2, 0xd3bd6ed3, 0xac43efac, 0x62c4a662, 0x9139a891, 0x9531a495, 0xe4d337e4, 0x79f28b79, + 0xe7d532e7, 0xc88b43c8, 0x376e5937, 0x6ddab76d, 0x8d018c8d, 0xd5b164d5, 0x4e9cd24e, 0xa949e0a9, + 0x6cd8b46c, 0x56acfa56, 0xf4f307f4, 0xeacf25ea, 0x65caaf65, 0x7af48e7a, 0xae47e9ae, 0x08101808, + 0xba6fd5ba, 0x78f08878, 0x254a6f25, 0x2e5c722e, 0x1c38241c, 0xa657f1a6, 0xb473c7b4, 0xc69751c6, + 0xe8cb23e8, 0xdda17cdd, 0x74e89c74, 0x1f3e211f, 0x4b96dd4b, 0xbd61dcbd, 0x8b0d868b, 0x8a0f858a, + 0x70e09070, 0x3e7c423e, 0xb571c4b5, 0x66ccaa66, 0x4890d848, 0x03060503, 0xf6f701f6, 0x0e1c120e, + 0x61c2a361, 0x356a5f35, 0x57aef957, 0xb969d0b9, 0x86179186, 0xc19958c1, 0x1d3a271d, 0x9e27b99e, + 0xe1d938e1, 0xf8eb13f8, 0x982bb398, 0x11223311, 0x69d2bb69, 0xd9a970d9, 0x8e07898e, 0x9433a794, + 0x9b2db69b, 0x1e3c221e, 0x87159287, 0xe9c920e9, 0xce8749ce, 0x55aaff55, 0x28507828, 0xdfa57adf, + 0x8c038f8c, 0xa159f8a1, 0x89098089, 0x0d1a170d, 0xbf65dabf, 0xe6d731e6, 0x4284c642, 0x68d0b868, + 0x4182c341, 0x9929b099, 0x2d5a772d, 0x0f1e110f, 0xb07bcbb0, 0x54a8fc54, 0xbb6dd6bb, 0x162c3a16, +}; + +alignas(16) const uint32_t lutEnc3[256] = { + 0xc6a56363, 0xf8847c7c, 0xee997777, 0xf68d7b7b, 0xff0df2f2, 0xd6bd6b6b, 0xdeb16f6f, 0x9154c5c5, + 0x60503030, 0x02030101, 0xcea96767, 0x567d2b2b, 0xe719fefe, 0xb562d7d7, 0x4de6abab, 0xec9a7676, + 0x8f45caca, 0x1f9d8282, 0x8940c9c9, 0xfa877d7d, 0xef15fafa, 0xb2eb5959, 0x8ec94747, 0xfb0bf0f0, + 0x41ecadad, 0xb367d4d4, 0x5ffda2a2, 0x45eaafaf, 0x23bf9c9c, 0x53f7a4a4, 0xe4967272, 0x9b5bc0c0, + 0x75c2b7b7, 0xe11cfdfd, 0x3dae9393, 0x4c6a2626, 0x6c5a3636, 0x7e413f3f, 0xf502f7f7, 0x834fcccc, + 0x685c3434, 0x51f4a5a5, 0xd134e5e5, 0xf908f1f1, 0xe2937171, 0xab73d8d8, 0x62533131, 0x2a3f1515, + 0x080c0404, 0x9552c7c7, 0x46652323, 0x9d5ec3c3, 0x30281818, 0x37a19696, 0x0a0f0505, 0x2fb59a9a, + 0x0e090707, 0x24361212, 0x1b9b8080, 0xdf3de2e2, 0xcd26ebeb, 0x4e692727, 0x7fcdb2b2, 0xea9f7575, + 0x121b0909, 0x1d9e8383, 0x58742c2c, 0x342e1a1a, 0x362d1b1b, 0xdcb26e6e, 0xb4ee5a5a, 0x5bfba0a0, + 0xa4f65252, 0x764d3b3b, 0xb761d6d6, 0x7dceb3b3, 0x527b2929, 0xdd3ee3e3, 0x5e712f2f, 0x13978484, + 0xa6f55353, 0xb968d1d1, 0x00000000, 0xc12ceded, 0x40602020, 0xe31ffcfc, 0x79c8b1b1, 0xb6ed5b5b, + 0xd4be6a6a, 0x8d46cbcb, 0x67d9bebe, 0x724b3939, 0x94de4a4a, 0x98d44c4c, 0xb0e85858, 0x854acfcf, + 0xbb6bd0d0, 0xc52aefef, 0x4fe5aaaa, 0xed16fbfb, 0x86c54343, 0x9ad74d4d, 0x66553333, 0x11948585, + 0x8acf4545, 0xe910f9f9, 0x04060202, 0xfe817f7f, 0xa0f05050, 0x78443c3c, 0x25ba9f9f, 0x4be3a8a8, + 0xa2f35151, 0x5dfea3a3, 0x80c04040, 0x058a8f8f, 0x3fad9292, 0x21bc9d9d, 0x70483838, 0xf104f5f5, + 0x63dfbcbc, 0x77c1b6b6, 0xaf75dada, 0x42632121, 0x20301010, 0xe51affff, 0xfd0ef3f3, 0xbf6dd2d2, + 0x814ccdcd, 0x18140c0c, 0x26351313, 0xc32fecec, 0xbee15f5f, 0x35a29797, 0x88cc4444, 0x2e391717, + 0x9357c4c4, 0x55f2a7a7, 0xfc827e7e, 0x7a473d3d, 0xc8ac6464, 0xbae75d5d, 0x322b1919, 0xe6957373, + 0xc0a06060, 0x19988181, 0x9ed14f4f, 0xa37fdcdc, 0x44662222, 0x547e2a2a, 0x3bab9090, 0x0b838888, + 0x8cca4646, 0xc729eeee, 0x6bd3b8b8, 0x283c1414, 0xa779dede, 0xbce25e5e, 0x161d0b0b, 0xad76dbdb, + 0xdb3be0e0, 0x64563232, 0x744e3a3a, 0x141e0a0a, 0x92db4949, 0x0c0a0606, 0x486c2424, 0xb8e45c5c, + 0x9f5dc2c2, 0xbd6ed3d3, 0x43efacac, 0xc4a66262, 0x39a89191, 0x31a49595, 0xd337e4e4, 0xf28b7979, + 0xd532e7e7, 0x8b43c8c8, 0x6e593737, 0xdab76d6d, 0x018c8d8d, 0xb164d5d5, 0x9cd24e4e, 0x49e0a9a9, + 0xd8b46c6c, 0xacfa5656, 0xf307f4f4, 0xcf25eaea, 0xcaaf6565, 0xf48e7a7a, 0x47e9aeae, 0x10180808, + 0x6fd5baba, 0xf0887878, 0x4a6f2525, 0x5c722e2e, 0x38241c1c, 0x57f1a6a6, 0x73c7b4b4, 0x9751c6c6, + 0xcb23e8e8, 0xa17cdddd, 0xe89c7474, 0x3e211f1f, 0x96dd4b4b, 0x61dcbdbd, 0x0d868b8b, 0x0f858a8a, + 0xe0907070, 0x7c423e3e, 0x71c4b5b5, 0xccaa6666, 0x90d84848, 0x06050303, 0xf701f6f6, 0x1c120e0e, + 0xc2a36161, 0x6a5f3535, 0xaef95757, 0x69d0b9b9, 0x17918686, 0x9958c1c1, 0x3a271d1d, 0x27b99e9e, + 0xd938e1e1, 0xeb13f8f8, 0x2bb39898, 0x22331111, 0xd2bb6969, 0xa970d9d9, 0x07898e8e, 0x33a79494, + 0x2db69b9b, 0x3c221e1e, 0x15928787, 0xc920e9e9, 0x8749cece, 0xaaff5555, 0x50782828, 0xa57adfdf, + 0x038f8c8c, 0x59f8a1a1, 0x09808989, 0x1a170d0d, 0x65dabfbf, 0xd731e6e6, 0x84c64242, 0xd0b86868, + 0x82c34141, 0x29b09999, 0x5a772d2d, 0x1e110f0f, 0x7bcbb0b0, 0xa8fc5454, 0x6dd6bbbb, 0x2c3a1616, +}; + +alignas(16) const uint32_t lutDec0[256] = { + 0x50a7f451, 0x5365417e, 0xc3a4171a, 0x965e273a, 0xcb6bab3b, 0xf1459d1f, 0xab58faac, 0x9303e34b, + 0x55fa3020, 0xf66d76ad, 0x9176cc88, 0x254c02f5, 0xfcd7e54f, 0xd7cb2ac5, 0x80443526, 0x8fa362b5, + 0x495ab1de, 0x671bba25, 0x980eea45, 0xe1c0fe5d, 0x02752fc3, 0x12f04c81, 0xa397468d, 0xc6f9d36b, + 0xe75f8f03, 0x959c9215, 0xeb7a6dbf, 0xda595295, 0x2d83bed4, 0xd3217458, 0x2969e049, 0x44c8c98e, + 0x6a89c275, 0x78798ef4, 0x6b3e5899, 0xdd71b927, 0xb64fe1be, 0x17ad88f0, 0x66ac20c9, 0xb43ace7d, + 0x184adf63, 0x82311ae5, 0x60335197, 0x457f5362, 0xe07764b1, 0x84ae6bbb, 0x1ca081fe, 0x942b08f9, + 0x58684870, 0x19fd458f, 0x876cde94, 0xb7f87b52, 0x23d373ab, 0xe2024b72, 0x578f1fe3, 0x2aab5566, + 0x0728ebb2, 0x03c2b52f, 0x9a7bc586, 0xa50837d3, 0xf2872830, 0xb2a5bf23, 0xba6a0302, 0x5c8216ed, + 0x2b1ccf8a, 0x92b479a7, 0xf0f207f3, 0xa1e2694e, 0xcdf4da65, 0xd5be0506, 0x1f6234d1, 0x8afea6c4, + 0x9d532e34, 0xa055f3a2, 0x32e18a05, 0x75ebf6a4, 0x39ec830b, 0xaaef6040, 0x069f715e, 0x51106ebd, + 0xf98a213e, 0x3d06dd96, 0xae053edd, 0x46bde64d, 0xb58d5491, 0x055dc471, 0x6fd40604, 0xff155060, + 0x24fb9819, 0x97e9bdd6, 0xcc434089, 0x779ed967, 0xbd42e8b0, 0x888b8907, 0x385b19e7, 0xdbeec879, + 0x470a7ca1, 0xe90f427c, 0xc91e84f8, 0x00000000, 0x83868009, 0x48ed2b32, 0xac70111e, 0x4e725a6c, + 0xfbff0efd, 0x5638850f, 0x1ed5ae3d, 0x27392d36, 0x64d90f0a, 0x21a65c68, 0xd1545b9b, 0x3a2e3624, + 0xb1670a0c, 0x0fe75793, 0xd296eeb4, 0x9e919b1b, 0x4fc5c080, 0xa220dc61, 0x694b775a, 0x161a121c, + 0x0aba93e2, 0xe52aa0c0, 0x43e0223c, 0x1d171b12, 0x0b0d090e, 0xadc78bf2, 0xb9a8b62d, 0xc8a91e14, + 0x8519f157, 0x4c0775af, 0xbbdd99ee, 0xfd607fa3, 0x9f2601f7, 0xbcf5725c, 0xc53b6644, 0x347efb5b, + 0x7629438b, 0xdcc623cb, 0x68fcedb6, 0x63f1e4b8, 0xcadc31d7, 0x10856342, 0x40229713, 0x2011c684, + 0x7d244a85, 0xf83dbbd2, 0x1132f9ae, 0x6da129c7, 0x4b2f9e1d, 0xf330b2dc, 0xec52860d, 0xd0e3c177, + 0x6c16b32b, 0x99b970a9, 0xfa489411, 0x2264e947, 0xc48cfca8, 0x1a3ff0a0, 0xd82c7d56, 0xef903322, + 0xc74e4987, 0xc1d138d9, 0xfea2ca8c, 0x360bd498, 0xcf81f5a6, 0x28de7aa5, 0x268eb7da, 0xa4bfad3f, + 0xe49d3a2c, 0x0d927850, 0x9bcc5f6a, 0x62467e54, 0xc2138df6, 0xe8b8d890, 0x5ef7392e, 0xf5afc382, + 0xbe805d9f, 0x7c93d069, 0xa92dd56f, 0xb31225cf, 0x3b99acc8, 0xa77d1810, 0x6e639ce8, 0x7bbb3bdb, + 0x097826cd, 0xf418596e, 0x01b79aec, 0xa89a4f83, 0x656e95e6, 0x7ee6ffaa, 0x08cfbc21, 0xe6e815ef, + 0xd99be7ba, 0xce366f4a, 0xd4099fea, 0xd67cb029, 0xafb2a431, 0x31233f2a, 0x3094a5c6, 0xc066a235, + 0x37bc4e74, 0xa6ca82fc, 0xb0d090e0, 0x15d8a733, 0x4a9804f1, 0xf7daec41, 0x0e50cd7f, 0x2ff69117, + 0x8dd64d76, 0x4db0ef43, 0x544daacc, 0xdf0496e4, 0xe3b5d19e, 0x1b886a4c, 0xb81f2cc1, 0x7f516546, + 0x04ea5e9d, 0x5d358c01, 0x737487fa, 0x2e410bfb, 0x5a1d67b3, 0x52d2db92, 0x335610e9, 0x1347d66d, + 0x8c61d79a, 0x7a0ca137, 0x8e14f859, 0x893c13eb, 0xee27a9ce, 0x35c961b7, 0xede51ce1, 0x3cb1477a, + 0x59dfd29c, 0x3f73f255, 0x79ce1418, 0xbf37c773, 0xeacdf753, 0x5baafd5f, 0x146f3ddf, 0x86db4478, + 0x81f3afca, 0x3ec468b9, 0x2c342438, 0x5f40a3c2, 0x72c31d16, 0x0c25e2bc, 0x8b493c28, 0x41950dff, + 0x7101a839, 0xdeb30c08, 0x9ce4b4d8, 0x90c15664, 0x6184cb7b, 0x70b632d5, 0x745c6c48, 0x4257b8d0, +}; + +alignas(16) const uint32_t lutDec1[256] = { + 0xa7f45150, 0x65417e53, 0xa4171ac3, 0x5e273a96, 0x6bab3bcb, 0x459d1ff1, 0x58faacab, 0x03e34b93, + 0xfa302055, 0x6d76adf6, 0x76cc8891, 0x4c02f525, 0xd7e54ffc, 0xcb2ac5d7, 0x44352680, 0xa362b58f, + 0x5ab1de49, 0x1bba2567, 0x0eea4598, 0xc0fe5de1, 0x752fc302, 0xf04c8112, 0x97468da3, 0xf9d36bc6, + 0x5f8f03e7, 0x9c921595, 0x7a6dbfeb, 0x595295da, 0x83bed42d, 0x217458d3, 0x69e04929, 0xc8c98e44, + 0x89c2756a, 0x798ef478, 0x3e58996b, 0x71b927dd, 0x4fe1beb6, 0xad88f017, 0xac20c966, 0x3ace7db4, + 0x4adf6318, 0x311ae582, 0x33519760, 0x7f536245, 0x7764b1e0, 0xae6bbb84, 0xa081fe1c, 0x2b08f994, + 0x68487058, 0xfd458f19, 0x6cde9487, 0xf87b52b7, 0xd373ab23, 0x024b72e2, 0x8f1fe357, 0xab55662a, + 0x28ebb207, 0xc2b52f03, 0x7bc5869a, 0x0837d3a5, 0x872830f2, 0xa5bf23b2, 0x6a0302ba, 0x8216ed5c, + 0x1ccf8a2b, 0xb479a792, 0xf207f3f0, 0xe2694ea1, 0xf4da65cd, 0xbe0506d5, 0x6234d11f, 0xfea6c48a, + 0x532e349d, 0x55f3a2a0, 0xe18a0532, 0xebf6a475, 0xec830b39, 0xef6040aa, 0x9f715e06, 0x106ebd51, + 0x8a213ef9, 0x06dd963d, 0x053eddae, 0xbde64d46, 0x8d5491b5, 0x5dc47105, 0xd406046f, 0x155060ff, + 0xfb981924, 0xe9bdd697, 0x434089cc, 0x9ed96777, 0x42e8b0bd, 0x8b890788, 0x5b19e738, 0xeec879db, + 0x0a7ca147, 0x0f427ce9, 0x1e84f8c9, 0x00000000, 0x86800983, 0xed2b3248, 0x70111eac, 0x725a6c4e, + 0xff0efdfb, 0x38850f56, 0xd5ae3d1e, 0x392d3627, 0xd90f0a64, 0xa65c6821, 0x545b9bd1, 0x2e36243a, + 0x670a0cb1, 0xe757930f, 0x96eeb4d2, 0x919b1b9e, 0xc5c0804f, 0x20dc61a2, 0x4b775a69, 0x1a121c16, + 0xba93e20a, 0x2aa0c0e5, 0xe0223c43, 0x171b121d, 0x0d090e0b, 0xc78bf2ad, 0xa8b62db9, 0xa91e14c8, + 0x19f15785, 0x0775af4c, 0xdd99eebb, 0x607fa3fd, 0x2601f79f, 0xf5725cbc, 0x3b6644c5, 0x7efb5b34, + 0x29438b76, 0xc623cbdc, 0xfcedb668, 0xf1e4b863, 0xdc31d7ca, 0x85634210, 0x22971340, 0x11c68420, + 0x244a857d, 0x3dbbd2f8, 0x32f9ae11, 0xa129c76d, 0x2f9e1d4b, 0x30b2dcf3, 0x52860dec, 0xe3c177d0, + 0x16b32b6c, 0xb970a999, 0x489411fa, 0x64e94722, 0x8cfca8c4, 0x3ff0a01a, 0x2c7d56d8, 0x903322ef, + 0x4e4987c7, 0xd138d9c1, 0xa2ca8cfe, 0x0bd49836, 0x81f5a6cf, 0xde7aa528, 0x8eb7da26, 0xbfad3fa4, + 0x9d3a2ce4, 0x9278500d, 0xcc5f6a9b, 0x467e5462, 0x138df6c2, 0xb8d890e8, 0xf7392e5e, 0xafc382f5, + 0x805d9fbe, 0x93d0697c, 0x2dd56fa9, 0x1225cfb3, 0x99acc83b, 0x7d1810a7, 0x639ce86e, 0xbb3bdb7b, + 0x7826cd09, 0x18596ef4, 0xb79aec01, 0x9a4f83a8, 0x6e95e665, 0xe6ffaa7e, 0xcfbc2108, 0xe815efe6, + 0x9be7bad9, 0x366f4ace, 0x099fead4, 0x7cb029d6, 0xb2a431af, 0x233f2a31, 0x94a5c630, 0x66a235c0, + 0xbc4e7437, 0xca82fca6, 0xd090e0b0, 0xd8a73315, 0x9804f14a, 0xdaec41f7, 0x50cd7f0e, 0xf691172f, + 0xd64d768d, 0xb0ef434d, 0x4daacc54, 0x0496e4df, 0xb5d19ee3, 0x886a4c1b, 0x1f2cc1b8, 0x5165467f, + 0xea5e9d04, 0x358c015d, 0x7487fa73, 0x410bfb2e, 0x1d67b35a, 0xd2db9252, 0x5610e933, 0x47d66d13, + 0x61d79a8c, 0x0ca1377a, 0x14f8598e, 0x3c13eb89, 0x27a9ceee, 0xc961b735, 0xe51ce1ed, 0xb1477a3c, + 0xdfd29c59, 0x73f2553f, 0xce141879, 0x37c773bf, 0xcdf753ea, 0xaafd5f5b, 0x6f3ddf14, 0xdb447886, + 0xf3afca81, 0xc468b93e, 0x3424382c, 0x40a3c25f, 0xc31d1672, 0x25e2bc0c, 0x493c288b, 0x950dff41, + 0x01a83971, 0xb30c08de, 0xe4b4d89c, 0xc1566490, 0x84cb7b61, 0xb632d570, 0x5c6c4874, 0x57b8d042, +}; + +alignas(16) const uint32_t lutDec2[256] = { + 0xf45150a7, 0x417e5365, 0x171ac3a4, 0x273a965e, 0xab3bcb6b, 0x9d1ff145, 0xfaacab58, 0xe34b9303, + 0x302055fa, 0x76adf66d, 0xcc889176, 0x02f5254c, 0xe54ffcd7, 0x2ac5d7cb, 0x35268044, 0x62b58fa3, + 0xb1de495a, 0xba25671b, 0xea45980e, 0xfe5de1c0, 0x2fc30275, 0x4c8112f0, 0x468da397, 0xd36bc6f9, + 0x8f03e75f, 0x9215959c, 0x6dbfeb7a, 0x5295da59, 0xbed42d83, 0x7458d321, 0xe0492969, 0xc98e44c8, + 0xc2756a89, 0x8ef47879, 0x58996b3e, 0xb927dd71, 0xe1beb64f, 0x88f017ad, 0x20c966ac, 0xce7db43a, + 0xdf63184a, 0x1ae58231, 0x51976033, 0x5362457f, 0x64b1e077, 0x6bbb84ae, 0x81fe1ca0, 0x08f9942b, + 0x48705868, 0x458f19fd, 0xde94876c, 0x7b52b7f8, 0x73ab23d3, 0x4b72e202, 0x1fe3578f, 0x55662aab, + 0xebb20728, 0xb52f03c2, 0xc5869a7b, 0x37d3a508, 0x2830f287, 0xbf23b2a5, 0x0302ba6a, 0x16ed5c82, + 0xcf8a2b1c, 0x79a792b4, 0x07f3f0f2, 0x694ea1e2, 0xda65cdf4, 0x0506d5be, 0x34d11f62, 0xa6c48afe, + 0x2e349d53, 0xf3a2a055, 0x8a0532e1, 0xf6a475eb, 0x830b39ec, 0x6040aaef, 0x715e069f, 0x6ebd5110, + 0x213ef98a, 0xdd963d06, 0x3eddae05, 0xe64d46bd, 0x5491b58d, 0xc471055d, 0x06046fd4, 0x5060ff15, + 0x981924fb, 0xbdd697e9, 0x4089cc43, 0xd967779e, 0xe8b0bd42, 0x8907888b, 0x19e7385b, 0xc879dbee, + 0x7ca1470a, 0x427ce90f, 0x84f8c91e, 0x00000000, 0x80098386, 0x2b3248ed, 0x111eac70, 0x5a6c4e72, + 0x0efdfbff, 0x850f5638, 0xae3d1ed5, 0x2d362739, 0x0f0a64d9, 0x5c6821a6, 0x5b9bd154, 0x36243a2e, + 0x0a0cb167, 0x57930fe7, 0xeeb4d296, 0x9b1b9e91, 0xc0804fc5, 0xdc61a220, 0x775a694b, 0x121c161a, + 0x93e20aba, 0xa0c0e52a, 0x223c43e0, 0x1b121d17, 0x090e0b0d, 0x8bf2adc7, 0xb62db9a8, 0x1e14c8a9, + 0xf1578519, 0x75af4c07, 0x99eebbdd, 0x7fa3fd60, 0x01f79f26, 0x725cbcf5, 0x6644c53b, 0xfb5b347e, + 0x438b7629, 0x23cbdcc6, 0xedb668fc, 0xe4b863f1, 0x31d7cadc, 0x63421085, 0x97134022, 0xc6842011, + 0x4a857d24, 0xbbd2f83d, 0xf9ae1132, 0x29c76da1, 0x9e1d4b2f, 0xb2dcf330, 0x860dec52, 0xc177d0e3, + 0xb32b6c16, 0x70a999b9, 0x9411fa48, 0xe9472264, 0xfca8c48c, 0xf0a01a3f, 0x7d56d82c, 0x3322ef90, + 0x4987c74e, 0x38d9c1d1, 0xca8cfea2, 0xd498360b, 0xf5a6cf81, 0x7aa528de, 0xb7da268e, 0xad3fa4bf, + 0x3a2ce49d, 0x78500d92, 0x5f6a9bcc, 0x7e546246, 0x8df6c213, 0xd890e8b8, 0x392e5ef7, 0xc382f5af, + 0x5d9fbe80, 0xd0697c93, 0xd56fa92d, 0x25cfb312, 0xacc83b99, 0x1810a77d, 0x9ce86e63, 0x3bdb7bbb, + 0x26cd0978, 0x596ef418, 0x9aec01b7, 0x4f83a89a, 0x95e6656e, 0xffaa7ee6, 0xbc2108cf, 0x15efe6e8, + 0xe7bad99b, 0x6f4ace36, 0x9fead409, 0xb029d67c, 0xa431afb2, 0x3f2a3123, 0xa5c63094, 0xa235c066, + 0x4e7437bc, 0x82fca6ca, 0x90e0b0d0, 0xa73315d8, 0x04f14a98, 0xec41f7da, 0xcd7f0e50, 0x91172ff6, + 0x4d768dd6, 0xef434db0, 0xaacc544d, 0x96e4df04, 0xd19ee3b5, 0x6a4c1b88, 0x2cc1b81f, 0x65467f51, + 0x5e9d04ea, 0x8c015d35, 0x87fa7374, 0x0bfb2e41, 0x67b35a1d, 0xdb9252d2, 0x10e93356, 0xd66d1347, + 0xd79a8c61, 0xa1377a0c, 0xf8598e14, 0x13eb893c, 0xa9ceee27, 0x61b735c9, 0x1ce1ede5, 0x477a3cb1, + 0xd29c59df, 0xf2553f73, 0x141879ce, 0xc773bf37, 0xf753eacd, 0xfd5f5baa, 0x3ddf146f, 0x447886db, + 0xafca81f3, 0x68b93ec4, 0x24382c34, 0xa3c25f40, 0x1d1672c3, 0xe2bc0c25, 0x3c288b49, 0x0dff4195, + 0xa8397101, 0x0c08deb3, 0xb4d89ce4, 0x566490c1, 0xcb7b6184, 0x32d570b6, 0x6c48745c, 0xb8d04257, +}; + +alignas(16) const uint32_t lutDec3[256] = { + 0x5150a7f4, 0x7e536541, 0x1ac3a417, 0x3a965e27, 0x3bcb6bab, 0x1ff1459d, 0xacab58fa, 0x4b9303e3, + 0x2055fa30, 0xadf66d76, 0x889176cc, 0xf5254c02, 0x4ffcd7e5, 0xc5d7cb2a, 0x26804435, 0xb58fa362, + 0xde495ab1, 0x25671bba, 0x45980eea, 0x5de1c0fe, 0xc302752f, 0x8112f04c, 0x8da39746, 0x6bc6f9d3, + 0x03e75f8f, 0x15959c92, 0xbfeb7a6d, 0x95da5952, 0xd42d83be, 0x58d32174, 0x492969e0, 0x8e44c8c9, + 0x756a89c2, 0xf478798e, 0x996b3e58, 0x27dd71b9, 0xbeb64fe1, 0xf017ad88, 0xc966ac20, 0x7db43ace, + 0x63184adf, 0xe582311a, 0x97603351, 0x62457f53, 0xb1e07764, 0xbb84ae6b, 0xfe1ca081, 0xf9942b08, + 0x70586848, 0x8f19fd45, 0x94876cde, 0x52b7f87b, 0xab23d373, 0x72e2024b, 0xe3578f1f, 0x662aab55, + 0xb20728eb, 0x2f03c2b5, 0x869a7bc5, 0xd3a50837, 0x30f28728, 0x23b2a5bf, 0x02ba6a03, 0xed5c8216, + 0x8a2b1ccf, 0xa792b479, 0xf3f0f207, 0x4ea1e269, 0x65cdf4da, 0x06d5be05, 0xd11f6234, 0xc48afea6, + 0x349d532e, 0xa2a055f3, 0x0532e18a, 0xa475ebf6, 0x0b39ec83, 0x40aaef60, 0x5e069f71, 0xbd51106e, + 0x3ef98a21, 0x963d06dd, 0xddae053e, 0x4d46bde6, 0x91b58d54, 0x71055dc4, 0x046fd406, 0x60ff1550, + 0x1924fb98, 0xd697e9bd, 0x89cc4340, 0x67779ed9, 0xb0bd42e8, 0x07888b89, 0xe7385b19, 0x79dbeec8, + 0xa1470a7c, 0x7ce90f42, 0xf8c91e84, 0x00000000, 0x09838680, 0x3248ed2b, 0x1eac7011, 0x6c4e725a, + 0xfdfbff0e, 0x0f563885, 0x3d1ed5ae, 0x3627392d, 0x0a64d90f, 0x6821a65c, 0x9bd1545b, 0x243a2e36, + 0x0cb1670a, 0x930fe757, 0xb4d296ee, 0x1b9e919b, 0x804fc5c0, 0x61a220dc, 0x5a694b77, 0x1c161a12, + 0xe20aba93, 0xc0e52aa0, 0x3c43e022, 0x121d171b, 0x0e0b0d09, 0xf2adc78b, 0x2db9a8b6, 0x14c8a91e, + 0x578519f1, 0xaf4c0775, 0xeebbdd99, 0xa3fd607f, 0xf79f2601, 0x5cbcf572, 0x44c53b66, 0x5b347efb, + 0x8b762943, 0xcbdcc623, 0xb668fced, 0xb863f1e4, 0xd7cadc31, 0x42108563, 0x13402297, 0x842011c6, + 0x857d244a, 0xd2f83dbb, 0xae1132f9, 0xc76da129, 0x1d4b2f9e, 0xdcf330b2, 0x0dec5286, 0x77d0e3c1, + 0x2b6c16b3, 0xa999b970, 0x11fa4894, 0x472264e9, 0xa8c48cfc, 0xa01a3ff0, 0x56d82c7d, 0x22ef9033, + 0x87c74e49, 0xd9c1d138, 0x8cfea2ca, 0x98360bd4, 0xa6cf81f5, 0xa528de7a, 0xda268eb7, 0x3fa4bfad, + 0x2ce49d3a, 0x500d9278, 0x6a9bcc5f, 0x5462467e, 0xf6c2138d, 0x90e8b8d8, 0x2e5ef739, 0x82f5afc3, + 0x9fbe805d, 0x697c93d0, 0x6fa92dd5, 0xcfb31225, 0xc83b99ac, 0x10a77d18, 0xe86e639c, 0xdb7bbb3b, + 0xcd097826, 0x6ef41859, 0xec01b79a, 0x83a89a4f, 0xe6656e95, 0xaa7ee6ff, 0x2108cfbc, 0xefe6e815, + 0xbad99be7, 0x4ace366f, 0xead4099f, 0x29d67cb0, 0x31afb2a4, 0x2a31233f, 0xc63094a5, 0x35c066a2, + 0x7437bc4e, 0xfca6ca82, 0xe0b0d090, 0x3315d8a7, 0xf14a9804, 0x41f7daec, 0x7f0e50cd, 0x172ff691, + 0x768dd64d, 0x434db0ef, 0xcc544daa, 0xe4df0496, 0x9ee3b5d1, 0x4c1b886a, 0xc1b81f2c, 0x467f5165, + 0x9d04ea5e, 0x015d358c, 0xfa737487, 0xfb2e410b, 0xb35a1d67, 0x9252d2db, 0xe9335610, 0x6d1347d6, + 0x9a8c61d7, 0x377a0ca1, 0x598e14f8, 0xeb893c13, 0xceee27a9, 0xb735c961, 0xe1ede51c, 0x7a3cb147, + 0x9c59dfd2, 0x553f73f2, 0x1879ce14, 0x73bf37c7, 0x53eacdf7, 0x5f5baafd, 0xdf146f3d, 0x7886db44, + 0xca81f3af, 0xb93ec468, 0x382c3424, 0xc25f40a3, 0x1672c31d, 0xbc0c25e2, 0x288b493c, 0xff41950d, + 0x397101a8, 0x08deb30c, 0xd89ce4b4, 0x6490c156, 0x7b6184cb, 0xd570b632, 0x48745c6c, 0xd04257b8, +}; + +rx_vec_i128 soft_aesenc(rx_vec_i128 in, rx_vec_i128 key) { + uint32_t s0, s1, s2, s3; + + s0 = rx_vec_i128_w(in); + s1 = rx_vec_i128_z(in); + s2 = rx_vec_i128_y(in); + s3 = rx_vec_i128_x(in); + + rx_vec_i128 out = rx_set_int_vec_i128( + (lutEnc0[s0 & 0xff] ^ lutEnc1[(s3 >> 8) & 0xff] ^ lutEnc2[(s2 >> 16) & 0xff] ^ lutEnc3[s1 >> 24]), + (lutEnc0[s1 & 0xff] ^ lutEnc1[(s0 >> 8) & 0xff] ^ lutEnc2[(s3 >> 16) & 0xff] ^ lutEnc3[s2 >> 24]), + (lutEnc0[s2 & 0xff] ^ lutEnc1[(s1 >> 8) & 0xff] ^ lutEnc2[(s0 >> 16) & 0xff] ^ lutEnc3[s3 >> 24]), + (lutEnc0[s3 & 0xff] ^ lutEnc1[(s2 >> 8) & 0xff] ^ lutEnc2[(s1 >> 16) & 0xff] ^ lutEnc3[s0 >> 24]) + ); + + return rx_xor_vec_i128(out, key); +} + +rx_vec_i128 soft_aesdec(rx_vec_i128 in, rx_vec_i128 key) { + uint32_t s0, s1, s2, s3; + + s0 = rx_vec_i128_w(in); + s1 = rx_vec_i128_z(in); + s2 = rx_vec_i128_y(in); + s3 = rx_vec_i128_x(in); + + rx_vec_i128 out = rx_set_int_vec_i128( + (lutDec0[s0 & 0xff] ^ lutDec1[(s1 >> 8) & 0xff] ^ lutDec2[(s2 >> 16) & 0xff] ^ lutDec3[s3 >> 24]), + (lutDec0[s1 & 0xff] ^ lutDec1[(s2 >> 8) & 0xff] ^ lutDec2[(s3 >> 16) & 0xff] ^ lutDec3[s0 >> 24]), + (lutDec0[s2 & 0xff] ^ lutDec1[(s3 >> 8) & 0xff] ^ lutDec2[(s0 >> 16) & 0xff] ^ lutDec3[s1 >> 24]), + (lutDec0[s3 & 0xff] ^ lutDec1[(s0 >> 8) & 0xff] ^ lutDec2[(s1 >> 16) & 0xff] ^ lutDec3[s2 >> 24]) + ); + + return rx_xor_vec_i128(out, key); +} diff --git a/src/crypto/randomx/soft_aes.h b/src/crypto/randomx/soft_aes.h new file mode 100644 index 000000000..254f8d630 --- /dev/null +++ b/src/crypto/randomx/soft_aes.h @@ -0,0 +1,46 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include +#include "intrin_portable.h" + +rx_vec_i128 soft_aesenc(rx_vec_i128 in, rx_vec_i128 key); + +rx_vec_i128 soft_aesdec(rx_vec_i128 in, rx_vec_i128 key); + +template +inline rx_vec_i128 aesenc(rx_vec_i128 in, rx_vec_i128 key) { + return soft ? soft_aesenc(in, key) : rx_aesenc_vec_i128(in, key); +} + +template +inline rx_vec_i128 aesdec(rx_vec_i128 in, rx_vec_i128 key) { + return soft ? soft_aesdec(in, key) : rx_aesdec_vec_i128(in, key); +} \ No newline at end of file diff --git a/src/crypto/randomx/superscalar.cpp b/src/crypto/randomx/superscalar.cpp new file mode 100644 index 000000000..0ca1fe693 --- /dev/null +++ b/src/crypto/randomx/superscalar.cpp @@ -0,0 +1,891 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include "configuration.h" +#include "program.hpp" +#include "blake2/endian.h" +#include "superscalar.hpp" +#include "intrin_portable.h" +#include "reciprocal.h" + +namespace randomx { + + static bool isMultiplication(SuperscalarInstructionType type) { + return type == SuperscalarInstructionType::IMUL_R || type == SuperscalarInstructionType::IMULH_R || type == SuperscalarInstructionType::ISMULH_R || type == SuperscalarInstructionType::IMUL_RCP; + } + + //uOPs (micro-ops) are represented only by the execution port they can go to + namespace ExecutionPort { + using type = int; + constexpr type Null = 0; + constexpr type P0 = 1; + constexpr type P1 = 2; + constexpr type P5 = 4; + constexpr type P01 = P0 | P1; + constexpr type P05 = P0 | P5; + constexpr type P015 = P0 | P1 | P5; + } + + //Macro-operation as output of the x86 decoder + //Usually one macro-op = one x86 instruction, but 2 instructions are sometimes fused into 1 macro-op + //Macro-op can consist of 1 or 2 uOPs. + class MacroOp { + public: + MacroOp(const char* name, int size) + : name_(name), size_(size), latency_(0), uop1_(ExecutionPort::Null), uop2_(ExecutionPort::Null) {} + MacroOp(const char* name, int size, int latency, ExecutionPort::type uop) + : name_(name), size_(size), latency_(latency), uop1_(uop), uop2_(ExecutionPort::Null) {} + MacroOp(const char* name, int size, int latency, ExecutionPort::type uop1, ExecutionPort::type uop2) + : name_(name), size_(size), latency_(latency), uop1_(uop1), uop2_(uop2) {} + MacroOp(const MacroOp& parent, bool dependent) + : name_(parent.name_), size_(parent.size_), latency_(parent.latency_), uop1_(parent.uop1_), uop2_(parent.uop2_), dependent_(dependent) {} + const char* getName() const { + return name_; + } + int getSize() const { + return size_; + } + int getLatency() const { + return latency_; + } + ExecutionPort::type getUop1() const { + return uop1_; + } + ExecutionPort::type getUop2() const { + return uop2_; + } + bool isSimple() const { + return uop2_ == ExecutionPort::Null; + } + bool isEliminated() const { + return uop1_ == ExecutionPort::Null; + } + bool isDependent() const { + return dependent_; + } + static const MacroOp Add_rr; + static const MacroOp Add_ri; + static const MacroOp Lea_sib; + static const MacroOp Sub_rr; + static const MacroOp Imul_rr; + static const MacroOp Imul_r; + static const MacroOp Mul_r; + static const MacroOp Mov_rr; + static const MacroOp Mov_ri64; + static const MacroOp Xor_rr; + static const MacroOp Xor_ri; + static const MacroOp Ror_rcl; + static const MacroOp Ror_ri; + static const MacroOp TestJz_fused; + static const MacroOp Xor_self; + static const MacroOp Cmp_ri; + static const MacroOp Setcc_r; + private: + const char* name_; + int size_; + int latency_; + ExecutionPort::type uop1_; + ExecutionPort::type uop2_; + bool dependent_ = false; + }; + + //Size: 3 bytes + const MacroOp MacroOp::Add_rr = MacroOp("add r,r", 3, 1, ExecutionPort::P015); + const MacroOp MacroOp::Sub_rr = MacroOp("sub r,r", 3, 1, ExecutionPort::P015); + const MacroOp MacroOp::Xor_rr = MacroOp("xor r,r", 3, 1, ExecutionPort::P015); + const MacroOp MacroOp::Imul_r = MacroOp("imul r", 3, 4, ExecutionPort::P1, ExecutionPort::P5); + const MacroOp MacroOp::Mul_r = MacroOp("mul r", 3, 4, ExecutionPort::P1, ExecutionPort::P5); + const MacroOp MacroOp::Mov_rr = MacroOp("mov r,r", 3); + + //Size: 4 bytes + const MacroOp MacroOp::Lea_sib = MacroOp("lea r,r+r*s", 4, 1, ExecutionPort::P01); + const MacroOp MacroOp::Imul_rr = MacroOp("imul r,r", 4, 3, ExecutionPort::P1); + const MacroOp MacroOp::Ror_ri = MacroOp("ror r,i", 4, 1, ExecutionPort::P05); + + //Size: 7 bytes (can be optionally padded with nop to 8 or 9 bytes) + const MacroOp MacroOp::Add_ri = MacroOp("add r,i", 7, 1, ExecutionPort::P015); + const MacroOp MacroOp::Xor_ri = MacroOp("xor r,i", 7, 1, ExecutionPort::P015); + + //Size: 10 bytes + const MacroOp MacroOp::Mov_ri64 = MacroOp("mov rax,i64", 10, 1, ExecutionPort::P015); + + //Unused: + const MacroOp MacroOp::Ror_rcl = MacroOp("ror r,cl", 3, 1, ExecutionPort::P0, ExecutionPort::P5); + const MacroOp MacroOp::Xor_self = MacroOp("xor rcx,rcx", 3); + const MacroOp MacroOp::Cmp_ri = MacroOp("cmp r,i", 7, 1, ExecutionPort::P015); + const MacroOp MacroOp::Setcc_r = MacroOp("setcc cl", 3, 1, ExecutionPort::P05); + const MacroOp MacroOp::TestJz_fused = MacroOp("testjz r,i", 13, 0, ExecutionPort::P5); + + const MacroOp IMULH_R_ops_array[] = { MacroOp::Mov_rr, MacroOp::Mul_r, MacroOp::Mov_rr }; + const MacroOp ISMULH_R_ops_array[] = { MacroOp::Mov_rr, MacroOp::Imul_r, MacroOp::Mov_rr }; + const MacroOp IMUL_RCP_ops_array[] = { MacroOp::Mov_ri64, MacroOp(MacroOp::Imul_rr, true) }; + + class SuperscalarInstructionInfo { + public: + const char* getName() const { + return name_; + } + int getSize() const { + return ops_.size(); + } + bool isSimple() const { + return getSize() == 1; + } + int getLatency() const { + return latency_; + } + const MacroOp& getOp(int index) const { + return ops_[index]; + } + SuperscalarInstructionType getType() const { + return type_; + } + int getResultOp() const { + return resultOp_; + } + int getDstOp() const { + return dstOp_; + } + int getSrcOp() const { + return srcOp_; + } + static const SuperscalarInstructionInfo ISUB_R; + static const SuperscalarInstructionInfo IXOR_R; + static const SuperscalarInstructionInfo IADD_RS; + static const SuperscalarInstructionInfo IMUL_R; + static const SuperscalarInstructionInfo IROR_C; + static const SuperscalarInstructionInfo IADD_C7; + static const SuperscalarInstructionInfo IXOR_C7; + static const SuperscalarInstructionInfo IADD_C8; + static const SuperscalarInstructionInfo IXOR_C8; + static const SuperscalarInstructionInfo IADD_C9; + static const SuperscalarInstructionInfo IXOR_C9; + static const SuperscalarInstructionInfo IMULH_R; + static const SuperscalarInstructionInfo ISMULH_R; + static const SuperscalarInstructionInfo IMUL_RCP; + static const SuperscalarInstructionInfo NOP; + private: + const char* name_; + SuperscalarInstructionType type_; + std::vector ops_; + int latency_; + int resultOp_ = 0; + int dstOp_ = 0; + int srcOp_; + + SuperscalarInstructionInfo(const char* name) + : name_(name), type_(SuperscalarInstructionType::INVALID), latency_(0) {} + SuperscalarInstructionInfo(const char* name, SuperscalarInstructionType type, const MacroOp& op, int srcOp) + : name_(name), type_(type), latency_(op.getLatency()), srcOp_(srcOp) { + ops_.push_back(MacroOp(op)); + } + template + SuperscalarInstructionInfo(const char* name, SuperscalarInstructionType type, const MacroOp(&arr)[N], int resultOp, int dstOp, int srcOp) + : name_(name), type_(type), latency_(0), resultOp_(resultOp), dstOp_(dstOp), srcOp_(srcOp) { + for (unsigned i = 0; i < N; ++i) { + ops_.push_back(MacroOp(arr[i])); + latency_ += ops_.back().getLatency(); + } + static_assert(N > 1, "Invalid array size"); + } + }; + + const SuperscalarInstructionInfo SuperscalarInstructionInfo::ISUB_R = SuperscalarInstructionInfo("ISUB_R", SuperscalarInstructionType::ISUB_R, MacroOp::Sub_rr, 0); + const SuperscalarInstructionInfo SuperscalarInstructionInfo::IXOR_R = SuperscalarInstructionInfo("IXOR_R", SuperscalarInstructionType::IXOR_R, MacroOp::Xor_rr, 0); + const SuperscalarInstructionInfo SuperscalarInstructionInfo::IADD_RS = SuperscalarInstructionInfo("IADD_RS", SuperscalarInstructionType::IADD_RS, MacroOp::Lea_sib, 0); + const SuperscalarInstructionInfo SuperscalarInstructionInfo::IMUL_R = SuperscalarInstructionInfo("IMUL_R", SuperscalarInstructionType::IMUL_R, MacroOp::Imul_rr, 0); + const SuperscalarInstructionInfo SuperscalarInstructionInfo::IROR_C = SuperscalarInstructionInfo("IROR_C", SuperscalarInstructionType::IROR_C, MacroOp::Ror_ri, -1); + + const SuperscalarInstructionInfo SuperscalarInstructionInfo::IADD_C7 = SuperscalarInstructionInfo("IADD_C7", SuperscalarInstructionType::IADD_C7, MacroOp::Add_ri, -1); + const SuperscalarInstructionInfo SuperscalarInstructionInfo::IXOR_C7 = SuperscalarInstructionInfo("IXOR_C7", SuperscalarInstructionType::IXOR_C7, MacroOp::Xor_ri, -1); + const SuperscalarInstructionInfo SuperscalarInstructionInfo::IADD_C8 = SuperscalarInstructionInfo("IADD_C8", SuperscalarInstructionType::IADD_C8, MacroOp::Add_ri, -1); + const SuperscalarInstructionInfo SuperscalarInstructionInfo::IXOR_C8 = SuperscalarInstructionInfo("IXOR_C8", SuperscalarInstructionType::IXOR_C8, MacroOp::Xor_ri, -1); + const SuperscalarInstructionInfo SuperscalarInstructionInfo::IADD_C9 = SuperscalarInstructionInfo("IADD_C9", SuperscalarInstructionType::IADD_C9, MacroOp::Add_ri, -1); + const SuperscalarInstructionInfo SuperscalarInstructionInfo::IXOR_C9 = SuperscalarInstructionInfo("IXOR_C9", SuperscalarInstructionType::IXOR_C9, MacroOp::Xor_ri, -1); + + const SuperscalarInstructionInfo SuperscalarInstructionInfo::IMULH_R = SuperscalarInstructionInfo("IMULH_R", SuperscalarInstructionType::IMULH_R, IMULH_R_ops_array, 1, 0, 1); + const SuperscalarInstructionInfo SuperscalarInstructionInfo::ISMULH_R = SuperscalarInstructionInfo("ISMULH_R", SuperscalarInstructionType::ISMULH_R, ISMULH_R_ops_array, 1, 0, 1); + const SuperscalarInstructionInfo SuperscalarInstructionInfo::IMUL_RCP = SuperscalarInstructionInfo("IMUL_RCP", SuperscalarInstructionType::IMUL_RCP, IMUL_RCP_ops_array, 1, 1, -1); + + const SuperscalarInstructionInfo SuperscalarInstructionInfo::NOP = SuperscalarInstructionInfo("NOP"); + + //these are some of the options how to split a 16-byte window into 3 or 4 x86 instructions. + //RandomX uses instructions with a native size of 3 (sub, xor, mul, mov), 4 (lea, mul), 7 (xor, add immediate) or 10 bytes (mov 64-bit immediate). + //Slots with sizes of 8 or 9 bytes need to be padded with a nop instruction. + const int buffer0[] = { 4, 8, 4 }; + const int buffer1[] = { 7, 3, 3, 3 }; + const int buffer2[] = { 3, 7, 3, 3 }; + const int buffer3[] = { 4, 9, 3 }; + const int buffer4[] = { 4, 4, 4, 4 }; + const int buffer5[] = { 3, 3, 10 }; + + class DecoderBuffer { + public: + static const DecoderBuffer Default; + template + DecoderBuffer(const char* name, int index, const int(&arr)[N]) + : name_(name), index_(index), counts_(arr), opsCount_(N) {} + const int* getCounts() const { + return counts_; + } + int getSize() const { + return opsCount_; + } + int getIndex() const { + return index_; + } + const char* getName() const { + return name_; + } + const DecoderBuffer* fetchNext(SuperscalarInstructionType instrType, int cycle, int mulCount, Blake2Generator& gen) const { + //If the current RandomX instruction is "IMULH", the next fetch configuration must be 3-3-10 + //because the full 128-bit multiplication instruction is 3 bytes long and decodes to 2 uOPs on Intel CPUs. + //Intel CPUs can decode at most 4 uOPs per cycle, so this requires a 2-1-1 configuration for a total of 3 macro ops. + if (instrType == SuperscalarInstructionType::IMULH_R || instrType == SuperscalarInstructionType::ISMULH_R) + return &decodeBuffer3310; + + //To make sure that the multiplication port is saturated, a 4-4-4-4 configuration is generated if the number of multiplications + //is lower than the number of cycles. + if (mulCount < cycle + 1) + return &decodeBuffer4444; + + //If the current RandomX instruction is "IMUL_RCP", the next buffer must begin with a 4-byte slot for multiplication. + if(instrType == SuperscalarInstructionType::IMUL_RCP) + return (gen.getByte() & 1) ? &decodeBuffer484 : &decodeBuffer493; + + //Default: select a random fetch configuration. + return fetchNextDefault(gen); + } + private: + const char* name_; + int index_; + const int* counts_; + int opsCount_; + DecoderBuffer() : index_(-1) {} + static const DecoderBuffer decodeBuffer484; + static const DecoderBuffer decodeBuffer7333; + static const DecoderBuffer decodeBuffer3733; + static const DecoderBuffer decodeBuffer493; + static const DecoderBuffer decodeBuffer4444; + static const DecoderBuffer decodeBuffer3310; + static const DecoderBuffer* decodeBuffers[4]; + const DecoderBuffer* fetchNextDefault(Blake2Generator& gen) const { + return decodeBuffers[gen.getByte() & 3]; + } + }; + + const DecoderBuffer DecoderBuffer::decodeBuffer484 = DecoderBuffer("4,8,4", 0, buffer0); + const DecoderBuffer DecoderBuffer::decodeBuffer7333 = DecoderBuffer("7,3,3,3", 1, buffer1); + const DecoderBuffer DecoderBuffer::decodeBuffer3733 = DecoderBuffer("3,7,3,3", 2, buffer2); + const DecoderBuffer DecoderBuffer::decodeBuffer493 = DecoderBuffer("4,9,3", 3, buffer3); + const DecoderBuffer DecoderBuffer::decodeBuffer4444 = DecoderBuffer("4,4,4,4", 4, buffer4); + const DecoderBuffer DecoderBuffer::decodeBuffer3310 = DecoderBuffer("3,3,10", 5, buffer5); + + const DecoderBuffer* DecoderBuffer::decodeBuffers[4] = { + &DecoderBuffer::decodeBuffer484, + &DecoderBuffer::decodeBuffer7333, + &DecoderBuffer::decodeBuffer3733, + &DecoderBuffer::decodeBuffer493, + }; + + const DecoderBuffer DecoderBuffer::Default = DecoderBuffer(); + + const SuperscalarInstructionInfo* slot_3[] = { &SuperscalarInstructionInfo::ISUB_R, &SuperscalarInstructionInfo::IXOR_R }; + const SuperscalarInstructionInfo* slot_3L[] = { &SuperscalarInstructionInfo::ISUB_R, &SuperscalarInstructionInfo::IXOR_R, &SuperscalarInstructionInfo::IMULH_R, &SuperscalarInstructionInfo::ISMULH_R }; + const SuperscalarInstructionInfo* slot_4[] = { &SuperscalarInstructionInfo::IROR_C, &SuperscalarInstructionInfo::IADD_RS }; + const SuperscalarInstructionInfo* slot_7[] = { &SuperscalarInstructionInfo::IXOR_C7, &SuperscalarInstructionInfo::IADD_C7 }; + const SuperscalarInstructionInfo* slot_8[] = { &SuperscalarInstructionInfo::IXOR_C8, &SuperscalarInstructionInfo::IADD_C8 }; + const SuperscalarInstructionInfo* slot_9[] = { &SuperscalarInstructionInfo::IXOR_C9, &SuperscalarInstructionInfo::IADD_C9 }; + const SuperscalarInstructionInfo* slot_10 = &SuperscalarInstructionInfo::IMUL_RCP; + + static bool selectRegister(std::vector& availableRegisters, Blake2Generator& gen, int& reg) { + int index; + if (availableRegisters.size() == 0) + return false; + + if (availableRegisters.size() > 1) { + index = gen.getInt32() % availableRegisters.size(); + } + else { + index = 0; + } + reg = availableRegisters[index]; + return true; + } + + class RegisterInfo { + public: + RegisterInfo() : latency(0), lastOpGroup(SuperscalarInstructionType::INVALID), lastOpPar(-1), value(0) {} + int latency; + SuperscalarInstructionType lastOpGroup; + int lastOpPar; + int value; + }; + + //"SuperscalarInstruction" consists of one or more macro-ops + class SuperscalarInstruction { + public: + void toInstr(Instruction& instr) { //translate to a RandomX instruction format + instr.opcode = (int)getType(); + instr.dst = dst_; + instr.src = src_ >= 0 ? src_ : dst_; + instr.setMod(mod_); + instr.setImm32(imm32_); + } + + void createForSlot(Blake2Generator& gen, int slotSize, int fetchType, bool isLast, bool isFirst) { + switch (slotSize) + { + case 3: + //if this is the last slot, we can also select "IMULH" instructions + if (isLast) { + create(slot_3L[gen.getByte() & 3], gen); + } + else { + create(slot_3[gen.getByte() & 1], gen); + } + break; + case 4: + //if this is the 4-4-4-4 buffer, issue multiplications as the first 3 instructions + if (fetchType == 4 && !isLast) { + create(&SuperscalarInstructionInfo::IMUL_R, gen); + } + else { + create(slot_4[gen.getByte() & 1], gen); + } + break; + case 7: + create(slot_7[gen.getByte() & 1], gen); + break; + case 8: + create(slot_8[gen.getByte() & 1], gen); + break; + case 9: + create(slot_9[gen.getByte() & 1], gen); + break; + case 10: + create(slot_10, gen); + break; + default: + UNREACHABLE; + } + } + + void create(const SuperscalarInstructionInfo* info, Blake2Generator& gen) { + info_ = info; + reset(); + switch (info->getType()) + { + case SuperscalarInstructionType::ISUB_R: { + mod_ = 0; + imm32_ = 0; + opGroup_ = SuperscalarInstructionType::IADD_RS; + groupParIsSource_ = true; + } break; + + case SuperscalarInstructionType::IXOR_R: { + mod_ = 0; + imm32_ = 0; + opGroup_ = SuperscalarInstructionType::IXOR_R; + groupParIsSource_ = true; + } break; + + case SuperscalarInstructionType::IADD_RS: { + mod_ = gen.getByte(); + imm32_ = 0; + opGroup_ = SuperscalarInstructionType::IADD_RS; + groupParIsSource_ = true; + } break; + + case SuperscalarInstructionType::IMUL_R: { + mod_ = 0; + imm32_ = 0; + opGroup_ = SuperscalarInstructionType::IMUL_R; + groupParIsSource_ = true; + } break; + + case SuperscalarInstructionType::IROR_C: { + mod_ = 0; + do { + imm32_ = gen.getByte() & 63; + } while (imm32_ == 0); + opGroup_ = SuperscalarInstructionType::IROR_C; + opGroupPar_ = -1; + } break; + + case SuperscalarInstructionType::IADD_C7: + case SuperscalarInstructionType::IADD_C8: + case SuperscalarInstructionType::IADD_C9: { + mod_ = 0; + imm32_ = gen.getInt32(); + opGroup_ = SuperscalarInstructionType::IADD_C7; + opGroupPar_ = -1; + } break; + + case SuperscalarInstructionType::IXOR_C7: + case SuperscalarInstructionType::IXOR_C8: + case SuperscalarInstructionType::IXOR_C9: { + mod_ = 0; + imm32_ = gen.getInt32(); + opGroup_ = SuperscalarInstructionType::IXOR_C7; + opGroupPar_ = -1; + } break; + + case SuperscalarInstructionType::IMULH_R: { + canReuse_ = true; + mod_ = 0; + imm32_ = 0; + opGroup_ = SuperscalarInstructionType::IMULH_R; + opGroupPar_ = gen.getInt32(); + } break; + + case SuperscalarInstructionType::ISMULH_R: { + canReuse_ = true; + mod_ = 0; + imm32_ = 0; + opGroup_ = SuperscalarInstructionType::ISMULH_R; + opGroupPar_ = gen.getInt32(); + } break; + + case SuperscalarInstructionType::IMUL_RCP: { + mod_ = 0; + do { + imm32_ = gen.getInt32(); + } while ((imm32_ & (imm32_ - 1)) == 0); + opGroup_ = SuperscalarInstructionType::IMUL_RCP; + opGroupPar_ = -1; + } break; + + default: + break; + } + } + + bool selectDestination(int cycle, bool allowChainedMul, RegisterInfo (®isters)[8], Blake2Generator& gen) { + /*if (allowChainedMultiplication && opGroup_ == SuperscalarInstructionType::IMUL_R) + std::cout << "Selecting destination with chained MUL enabled" << std::endl;*/ + std::vector availableRegisters; + //Conditions for the destination register: + // * value must be ready at the required cycle + // * cannot be the same as the source register unless the instruction allows it + // - this avoids optimizable instructions such as "xor r, r" or "sub r, r" + // * register cannot be multiplied twice in a row unless allowChainedMul is true + // - this avoids accumulation of trailing zeroes in registers due to excessive multiplication + // - allowChainedMul is set to true if an attempt to find source/destination registers failed (this is quite rare, but prevents a catastrophic failure of the generator) + // * either the last instruction applied to the register or its source must be different than this instruction + // - this avoids optimizable instruction sequences such as "xor r1, r2; xor r1, r2" or "ror r, C1; ror r, C2" or "add r, C1; add r, C2" + // * register r5 cannot be the destination of the IADD_RS instruction (limitation of the x86 lea instruction) + for (int i = 0; i < 8; ++i) { + if (registers[i].latency <= cycle && (canReuse_ || i != src_) && (allowChainedMul || opGroup_ != SuperscalarInstructionType::IMUL_R || registers[i].lastOpGroup != SuperscalarInstructionType::IMUL_R) && (registers[i].lastOpGroup != opGroup_ || registers[i].lastOpPar != opGroupPar_) && (info_->getType() != SuperscalarInstructionType::IADD_RS || i != RegisterNeedsDisplacement)) + availableRegisters.push_back(i); + } + return selectRegister(availableRegisters, gen, dst_); + } + + bool selectSource(int cycle, RegisterInfo(®isters)[8], Blake2Generator& gen) { + std::vector availableRegisters; + //all registers that are ready at the cycle + for (unsigned i = 0; i < 8; ++i) { + if (registers[i].latency <= cycle) + availableRegisters.push_back(i); + } + //if there are only 2 available registers for IADD_RS and one of them is r5, select it as the source because it cannot be the destination + if (availableRegisters.size() == 2 && info_->getType() == SuperscalarInstructionType::IADD_RS) { + if (availableRegisters[0] == RegisterNeedsDisplacement || availableRegisters[1] == RegisterNeedsDisplacement) { + opGroupPar_ = src_ = RegisterNeedsDisplacement; + return true; + } + } + if (selectRegister(availableRegisters, gen, src_)) { + if (groupParIsSource_) + opGroupPar_ = src_; + return true; + } + return false; + } + + SuperscalarInstructionType getType() { + return info_->getType(); + } + int getSource() { + return src_; + } + int getDestination() { + return dst_; + } + SuperscalarInstructionType getGroup() { + return opGroup_; + } + int getGroupPar() { + return opGroupPar_; + } + + const SuperscalarInstructionInfo& getInfo() const { + return *info_; + } + + static const SuperscalarInstruction Null; + + private: + const SuperscalarInstructionInfo* info_; + int src_ = -1; + int dst_ = -1; + int mod_; + uint32_t imm32_; + SuperscalarInstructionType opGroup_; + int opGroupPar_; + bool canReuse_ = false; + bool groupParIsSource_ = false; + + void reset() { + src_ = dst_ = -1; + canReuse_ = groupParIsSource_ = false; + } + + SuperscalarInstruction(const SuperscalarInstructionInfo* info) : info_(info) { + } + }; + + const SuperscalarInstruction SuperscalarInstruction::Null = SuperscalarInstruction(&SuperscalarInstructionInfo::NOP); + + constexpr int CYCLE_MAP_SIZE = RANDOMX_SUPERSCALAR_MAX_LATENCY + 4; + constexpr int LOOK_FORWARD_CYCLES = 4; + constexpr int MAX_THROWAWAY_COUNT = 256; + + template + static int scheduleUop(ExecutionPort::type uop, ExecutionPort::type(&portBusy)[CYCLE_MAP_SIZE][3], int cycle) { + //The scheduling here is done optimistically by checking port availability in order P5 -> P0 -> P1 to not overload + //port P1 (multiplication) by instructions that can go to any port. + for (; cycle < static_cast(RandomX_CurrentConfig.SuperscalarLatency) + 4; ++cycle) { + if ((uop & ExecutionPort::P5) != 0 && !portBusy[cycle][2]) { + if (commit) { + if (trace) std::cout << "; P5 at cycle " << cycle << std::endl; + portBusy[cycle][2] = uop; + } + return cycle; + } + if ((uop & ExecutionPort::P0) != 0 && !portBusy[cycle][0]) { + if (commit) { + if (trace) std::cout << "; P0 at cycle " << cycle << std::endl; + portBusy[cycle][0] = uop; + } + return cycle; + } + if ((uop & ExecutionPort::P1) != 0 && !portBusy[cycle][1]) { + if (commit) { + if (trace) std::cout << "; P1 at cycle " << cycle << std::endl; + portBusy[cycle][1] = uop; + } + return cycle; + } + } + return -1; + } + + template + static int scheduleMop(const MacroOp& mop, ExecutionPort::type(&portBusy)[CYCLE_MAP_SIZE][3], int cycle, int depCycle) { + //if this macro-op depends on the previous one, increase the starting cycle if needed + //this handles an explicit dependency chain in IMUL_RCP + if (mop.isDependent()) { + cycle = (cycle > depCycle) ? cycle : depCycle; + } + //move instructions are eliminated and don't need an execution unit + if (mop.isEliminated()) { + if (commit) + if (trace) std::cout << "; (eliminated)" << std::endl; + return cycle; + } + else if (mop.isSimple()) { + //this macro-op has only one uOP + return scheduleUop(mop.getUop1(), portBusy, cycle); + } + else { + //macro-ops with 2 uOPs are scheduled conservatively by requiring both uOPs to execute in the same cycle + for (; cycle < static_cast(RandomX_CurrentConfig.SuperscalarLatency) + 4; ++cycle) { + + int cycle1 = scheduleUop(mop.getUop1(), portBusy, cycle); + int cycle2 = scheduleUop(mop.getUop2(), portBusy, cycle); + + if (cycle1 == cycle2) { + if (commit) { + scheduleUop(mop.getUop1(), portBusy, cycle1); + scheduleUop(mop.getUop2(), portBusy, cycle2); + } + return cycle1; + } + } + } + + return -1; + } + + void generateSuperscalar(SuperscalarProgram& prog, Blake2Generator& gen) { + + ExecutionPort::type portBusy[CYCLE_MAP_SIZE][3]; + memset(portBusy, 0, sizeof(portBusy)); + RegisterInfo registers[8]; + + const DecoderBuffer* decodeBuffer = &DecoderBuffer::Default; + SuperscalarInstruction currentInstruction = SuperscalarInstruction::Null; + int macroOpIndex = 0; + int codeSize = 0; + int macroOpCount = 0; + int cycle = 0; + int depCycle = 0; + int retireCycle = 0; + bool portsSaturated = false; + int programSize = 0; + int mulCount = 0; + int decodeCycle; + int throwAwayCount = 0; + + //decode instructions for RANDOMX_SUPERSCALAR_LATENCY cycles or until an execution port is saturated. + //Each decode cycle decodes 16 bytes of x86 code. + //Since a decode cycle produces on average 3.45 macro-ops and there are only 3 ALU ports, execution ports are always + //saturated first. The cycle limit is present only to guarantee loop termination. + //Program size is limited to SuperscalarMaxSize instructions. + for (decodeCycle = 0; decodeCycle < static_cast(RandomX_CurrentConfig.SuperscalarLatency) && !portsSaturated && programSize < 3 * static_cast(RandomX_CurrentConfig.SuperscalarLatency) + 2; ++decodeCycle) { + + //select a decode configuration + decodeBuffer = decodeBuffer->fetchNext(currentInstruction.getType(), decodeCycle, mulCount, gen); + if (trace) std::cout << "; ------------- fetch cycle " << cycle << " (" << decodeBuffer->getName() << ")" << std::endl; + + int bufferIndex = 0; + + //fill all instruction slots in the current decode buffer + while (bufferIndex < decodeBuffer->getSize()) { + int topCycle = cycle; + + //if we have issued all macro-ops for the current RandomX instruction, create a new instruction + if (macroOpIndex >= currentInstruction.getInfo().getSize()) { + if (portsSaturated || programSize >= 3 * static_cast(RandomX_CurrentConfig.SuperscalarLatency) + 2) + break; + //select an instruction so that the first macro-op fits into the current slot + currentInstruction.createForSlot(gen, decodeBuffer->getCounts()[bufferIndex], decodeBuffer->getIndex(), decodeBuffer->getSize() == bufferIndex + 1, bufferIndex == 0); + macroOpIndex = 0; + if (trace) std::cout << "; " << currentInstruction.getInfo().getName() << std::endl; + } + const MacroOp& mop = currentInstruction.getInfo().getOp(macroOpIndex); + if (trace) std::cout << mop.getName() << " "; + + //calculate the earliest cycle when this macro-op (all of its uOPs) can be scheduled for execution + int scheduleCycle = scheduleMop(mop, portBusy, cycle, depCycle); + if (scheduleCycle < 0) { + if (trace) std::cout << "Unable to map operation '" << mop.getName() << "' to execution port (cycle " << cycle << ")" << std::endl; + //__debugbreak(); + portsSaturated = true; + break; + } + + //find a source register (if applicable) that will be ready when this instruction executes + if (macroOpIndex == currentInstruction.getInfo().getSrcOp()) { + int forward; + //if no suitable operand is ready, look up to LOOK_FORWARD_CYCLES forward + for (forward = 0; forward < LOOK_FORWARD_CYCLES && !currentInstruction.selectSource(scheduleCycle, registers, gen); ++forward) { + if (trace) std::cout << "; src STALL at cycle " << cycle << std::endl; + ++scheduleCycle; + ++cycle; + } + //if no register was found, throw the instruction away and try another one + if (forward == LOOK_FORWARD_CYCLES) { + if (throwAwayCount < MAX_THROWAWAY_COUNT) { + throwAwayCount++; + macroOpIndex = currentInstruction.getInfo().getSize(); + if (trace) std::cout << "; THROW away " << currentInstruction.getInfo().getName() << std::endl; + //cycle = topCycle; + continue; + } + //abort this decode buffer + if (trace) std::cout << "Aborting at cycle " << cycle << " with decode buffer " << decodeBuffer->getName() << " - source registers not available for operation " << currentInstruction.getInfo().getName() << std::endl; + currentInstruction = SuperscalarInstruction::Null; + break; + } + if (trace) std::cout << "; src = r" << currentInstruction.getSource() << std::endl; + } + //find a destination register that will be ready when this instruction executes + if (macroOpIndex == currentInstruction.getInfo().getDstOp()) { + int forward; + for (forward = 0; forward < LOOK_FORWARD_CYCLES && !currentInstruction.selectDestination(scheduleCycle, throwAwayCount > 0, registers, gen); ++forward) { + if (trace) std::cout << "; dst STALL at cycle " << cycle << std::endl; + ++scheduleCycle; + ++cycle; + } + if (forward == LOOK_FORWARD_CYCLES) { //throw instruction away + if (throwAwayCount < MAX_THROWAWAY_COUNT) { + throwAwayCount++; + macroOpIndex = currentInstruction.getInfo().getSize(); + if (trace) std::cout << "; THROW away " << currentInstruction.getInfo().getName() << std::endl; + //cycle = topCycle; + continue; + } + //abort this decode buffer + if (trace) std::cout << "Aborting at cycle " << cycle << " with decode buffer " << decodeBuffer->getName() << " - destination registers not available" << std::endl; + currentInstruction = SuperscalarInstruction::Null; + break; + } + if (trace) std::cout << "; dst = r" << currentInstruction.getDestination() << std::endl; + } + throwAwayCount = 0; + + //recalculate when the instruction can be scheduled for execution based on operand availability + scheduleCycle = scheduleMop(mop, portBusy, scheduleCycle, scheduleCycle); + + //calculate when the result will be ready + depCycle = scheduleCycle + mop.getLatency(); + + //if this instruction writes the result, modify register information + // RegisterInfo.latency - which cycle the register will be ready + // RegisterInfo.lastOpGroup - the last operation that was applied to the register + // RegisterInfo.lastOpPar - the last operation source value (-1 = constant, 0-7 = register) + if (macroOpIndex == currentInstruction.getInfo().getResultOp()) { + int dst = currentInstruction.getDestination(); + RegisterInfo& ri = registers[dst]; + retireCycle = depCycle; + ri.latency = retireCycle; + ri.lastOpGroup = currentInstruction.getGroup(); + ri.lastOpPar = currentInstruction.getGroupPar(); + if (trace) std::cout << "; RETIRED at cycle " << retireCycle << std::endl; + } + codeSize += mop.getSize(); + bufferIndex++; + macroOpIndex++; + macroOpCount++; + + //terminating condition + if (scheduleCycle >= static_cast(RandomX_CurrentConfig.SuperscalarLatency)) { + portsSaturated = true; + } + cycle = topCycle; + + //when all macro-ops of the current instruction have been issued, add the instruction into the program + if (macroOpIndex >= currentInstruction.getInfo().getSize()) { + currentInstruction.toInstr(prog(programSize++)); + mulCount += isMultiplication(currentInstruction.getType()); + } + } + ++cycle; + } + + double ipc = (macroOpCount / (double)retireCycle); + + memset(prog.asicLatencies, 0, sizeof(prog.asicLatencies)); + + //Calculate ASIC latency: + //Assumes 1 cycle latency for all operations and unlimited parallelization. + for (int i = 0; i < programSize; ++i) { + Instruction& instr = prog(i); + int latDst = prog.asicLatencies[instr.dst] + 1; + int latSrc = instr.dst != instr.src ? prog.asicLatencies[instr.src] + 1 : 0; + prog.asicLatencies[instr.dst] = (latDst > latSrc) ? latDst : latSrc; + } + + //address register is the register with the highest ASIC latency + int asicLatencyMax = 0; + int addressReg = 0; + for (int i = 0; i < 8; ++i) { + if (prog.asicLatencies[i] > asicLatencyMax) { + asicLatencyMax = prog.asicLatencies[i]; + addressReg = i; + } + prog.cpuLatencies[i] = registers[i].latency; + } + + prog.setSize(programSize); + prog.setAddressRegister(addressReg); + + prog.cpuLatency = retireCycle; + prog.asicLatency = asicLatencyMax; + prog.codeSize = codeSize; + prog.macroOps = macroOpCount; + prog.decodeCycles = decodeCycle; + prog.ipc = ipc; + prog.mulCount = mulCount; + + + /*if(INFO) std::cout << "; ALU port utilization:" << std::endl; + if (INFO) std::cout << "; (* = in use, _ = idle)" << std::endl; + + int portCycles = 0; + for (int i = 0; i < RandomX_Config.SuperscalarLatency + 4; ++i) { + std::cout << "; " << std::setw(3) << i << " "; + for (int j = 0; j < 3; ++j) { + std::cout << (portBusy[i][j] ? '*' : '_'); + portCycles += !!portBusy[i][j]; + } + std::cout << std::endl; + }*/ + } + + void executeSuperscalar(int_reg_t(&r)[8], SuperscalarProgram& prog, std::vector *reciprocals) { + for (unsigned j = 0; j < prog.getSize(); ++j) { + Instruction& instr = prog(j); + switch ((SuperscalarInstructionType)instr.opcode) + { + case SuperscalarInstructionType::ISUB_R: + r[instr.dst] -= r[instr.src]; + break; + case SuperscalarInstructionType::IXOR_R: + r[instr.dst] ^= r[instr.src]; + break; + case SuperscalarInstructionType::IADD_RS: + r[instr.dst] += r[instr.src] << instr.getModShift(); + break; + case SuperscalarInstructionType::IMUL_R: + r[instr.dst] *= r[instr.src]; + break; + case SuperscalarInstructionType::IROR_C: + r[instr.dst] = rotr64(r[instr.dst], instr.getImm32()); + break; + case SuperscalarInstructionType::IADD_C7: + case SuperscalarInstructionType::IADD_C8: + case SuperscalarInstructionType::IADD_C9: + r[instr.dst] += signExtend2sCompl(instr.getImm32()); + break; + case SuperscalarInstructionType::IXOR_C7: + case SuperscalarInstructionType::IXOR_C8: + case SuperscalarInstructionType::IXOR_C9: + r[instr.dst] ^= signExtend2sCompl(instr.getImm32()); + break; + case SuperscalarInstructionType::IMULH_R: + r[instr.dst] = mulh(r[instr.dst], r[instr.src]); + break; + case SuperscalarInstructionType::ISMULH_R: + r[instr.dst] = smulh(r[instr.dst], r[instr.src]); + break; + case SuperscalarInstructionType::IMUL_RCP: + if (reciprocals != nullptr) + r[instr.dst] *= (*reciprocals)[instr.getImm32()]; + else + r[instr.dst] *= randomx_reciprocal(instr.getImm32()); + break; + default: + UNREACHABLE; + } + } + } +} diff --git a/src/crypto/randomx/superscalar.hpp b/src/crypto/randomx/superscalar.hpp new file mode 100644 index 000000000..bc101c453 --- /dev/null +++ b/src/crypto/randomx/superscalar.hpp @@ -0,0 +1,60 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include +#include +#include "superscalar_program.hpp" +#include "blake2_generator.hpp" + +namespace randomx { + // Intel Ivy Bridge reference + enum class SuperscalarInstructionType { //uOPs (decode) execution ports latency code size + ISUB_R = 0, //1 p015 1 3 (sub) + IXOR_R = 1, //1 p015 1 3 (xor) + IADD_RS = 2, //1 p01 1 4 (lea) + IMUL_R = 3, //1 p1 3 4 (imul) + IROR_C = 4, //1 p05 1 4 (ror) + IADD_C7 = 5, //1 p015 1 7 (add) + IXOR_C7 = 6, //1 p015 1 7 (xor) + IADD_C8 = 7, //1+0 p015 1 7+1 (add+nop) + IXOR_C8 = 8, //1+0 p015 1 7+1 (xor+nop) + IADD_C9 = 9, //1+0 p015 1 7+2 (add+nop) + IXOR_C9 = 10, //1+0 p015 1 7+2 (xor+nop) + IMULH_R = 11, //1+2+1 0+(p1,p5)+0 3 3+3+3 (mov+mul+mov) + ISMULH_R = 12, //1+2+1 0+(p1,p5)+0 3 3+3+3 (mov+imul+mov) + IMUL_RCP = 13, //1+1 p015+p1 4 10+4 (mov+imul) + + COUNT = 14, + INVALID = -1 + }; + + void generateSuperscalar(SuperscalarProgram& prog, Blake2Generator& gen); + void executeSuperscalar(uint64_t(&r)[8], SuperscalarProgram& prog, std::vector *reciprocals = nullptr); +} \ No newline at end of file diff --git a/src/crypto/randomx/superscalar_program.hpp b/src/crypto/randomx/superscalar_program.hpp new file mode 100644 index 000000000..dbb0b1635 --- /dev/null +++ b/src/crypto/randomx/superscalar_program.hpp @@ -0,0 +1,73 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include +#include "instruction.hpp" +#include "common.hpp" + +namespace randomx { + + class SuperscalarProgram { + public: + Instruction& operator()(int pc) { + return programBuffer[pc]; + } + uint32_t getSize() { + return size; + } + void setSize(uint32_t val) { + size = val; + } + int getAddressRegister() { + return addrReg; + } + void setAddressRegister(int val) { + addrReg = val; + } + + Instruction programBuffer[SuperscalarMaxSize]; + uint32_t size +#ifndef NDEBUG + = 0 +#endif + ; + int addrReg; + double ipc; + int codeSize; + int macroOps; + int decodeCycles; + int cpuLatency; + int asicLatency; + int mulCount; + int cpuLatencies[8]; + int asicLatencies[8]; + }; + +} \ No newline at end of file diff --git a/src/crypto/randomx/virtual_machine.cpp b/src/crypto/randomx/virtual_machine.cpp new file mode 100644 index 000000000..c216de697 --- /dev/null +++ b/src/crypto/randomx/virtual_machine.cpp @@ -0,0 +1,129 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include +#include +#include +#include "virtual_machine.hpp" +#include "common.hpp" +#include "aes_hash.hpp" +#include "blake2/blake2.h" +#include "intrin_portable.h" +#include "allocator.hpp" + +randomx_vm::~randomx_vm() { + +} + +void randomx_vm::resetRoundingMode() { + rx_reset_float_state(); +} + +namespace randomx { + + static inline uint64_t getSmallPositiveFloatBits(uint64_t entropy) { + auto exponent = entropy >> 59; //0..31 + auto mantissa = entropy & mantissaMask; + exponent += exponentBias; + exponent &= exponentMask; + exponent <<= mantissaSize; + return exponent | mantissa; + } + + static inline uint64_t getStaticExponent(uint64_t entropy) { + auto exponent = constExponentBits; + exponent |= (entropy >> (64 - staticExponentBits)) << dynamicExponentBits; + exponent <<= mantissaSize; + return exponent; + } + + static inline uint64_t getFloatMask(uint64_t entropy) { + constexpr uint64_t mask22bit = (1ULL << 22) - 1; + return (entropy & mask22bit) | getStaticExponent(entropy); + } + +} + +void randomx_vm::initialize() { + store64(®.a[0].lo, randomx::getSmallPositiveFloatBits(program.getEntropy(0))); + store64(®.a[0].hi, randomx::getSmallPositiveFloatBits(program.getEntropy(1))); + store64(®.a[1].lo, randomx::getSmallPositiveFloatBits(program.getEntropy(2))); + store64(®.a[1].hi, randomx::getSmallPositiveFloatBits(program.getEntropy(3))); + store64(®.a[2].lo, randomx::getSmallPositiveFloatBits(program.getEntropy(4))); + store64(®.a[2].hi, randomx::getSmallPositiveFloatBits(program.getEntropy(5))); + store64(®.a[3].lo, randomx::getSmallPositiveFloatBits(program.getEntropy(6))); + store64(®.a[3].hi, randomx::getSmallPositiveFloatBits(program.getEntropy(7))); + mem.ma = program.getEntropy(8) & CacheLineAlignMask; + mem.mx = program.getEntropy(10); + auto addressRegisters = program.getEntropy(12); + config.readReg0 = 0 + (addressRegisters & 1); + addressRegisters >>= 1; + config.readReg1 = 2 + (addressRegisters & 1); + addressRegisters >>= 1; + config.readReg2 = 4 + (addressRegisters & 1); + addressRegisters >>= 1; + config.readReg3 = 6 + (addressRegisters & 1); + datasetOffset = (program.getEntropy(13) % (DatasetExtraItems + 1)) * randomx::CacheLineSize; + store64(&config.eMask[0], randomx::getFloatMask(program.getEntropy(14))); + store64(&config.eMask[1], randomx::getFloatMask(program.getEntropy(15))); +} + +namespace randomx { + + template + VmBase::~VmBase() { + } + + template + void VmBase::setScratchpad(uint8_t *scratchpad) { + if (datasetPtr == nullptr) { + throw std::invalid_argument("Cache/Dataset not set"); + } + + this->scratchpad = scratchpad; + } + + template + void VmBase::getFinalResult(void* out, size_t outSize) { + hashAes1Rx4(scratchpad, ScratchpadSize, ®.a); + blake2b(out, outSize, ®, sizeof(RegisterFile), nullptr, 0); + } + + template + void VmBase::initScratchpad(void* seed) { + fillAes1Rx4(seed, ScratchpadSize, scratchpad); + } + + template + void VmBase::generateProgram(void* seed) { + fillAes4Rx4(seed, 128 + RandomX_CurrentConfig.ProgramSize * 8, &program); + } + + template class VmBase; + template class VmBase; +} diff --git a/src/crypto/randomx/virtual_machine.hpp b/src/crypto/randomx/virtual_machine.hpp new file mode 100644 index 000000000..cba79d724 --- /dev/null +++ b/src/crypto/randomx/virtual_machine.hpp @@ -0,0 +1,90 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include +#include "common.hpp" +#include "program.hpp" + +/* Global namespace for C binding */ +class randomx_vm +{ +public: + virtual ~randomx_vm() = 0; + virtual void setScratchpad(uint8_t *scratchpad) = 0; + virtual void getFinalResult(void* out, size_t outSize) = 0; + virtual void setDataset(randomx_dataset* dataset) { } + virtual void setCache(randomx_cache* cache) { } + virtual void initScratchpad(void* seed) = 0; + virtual void run(void* seed) = 0; + void resetRoundingMode(); + + randomx::RegisterFile *getRegisterFile() { + return ® + } + + const void* getScratchpad() { + return scratchpad; + } + + const randomx::Program& getProgram() + { + return program; + } + +protected: + void initialize(); + alignas(64) randomx::Program program; + alignas(64) randomx::RegisterFile reg; + alignas(16) randomx::ProgramConfiguration config; + randomx::MemoryRegisters mem; + uint8_t* scratchpad; + union { + randomx_cache* cachePtr = nullptr; + randomx_dataset* datasetPtr; + }; + uint64_t datasetOffset; +}; + +namespace randomx { + + template + class VmBase : public randomx_vm + { + public: + ~VmBase() override; + void setScratchpad(uint8_t *scratchpad) override; + void initScratchpad(void* seed) override; + void getFinalResult(void* out, size_t outSize) override; + + protected: + void generateProgram(void* seed); + }; + +} diff --git a/src/crypto/randomx/virtual_memory.cpp b/src/crypto/randomx/virtual_memory.cpp new file mode 100644 index 000000000..661740fc9 --- /dev/null +++ b/src/crypto/randomx/virtual_memory.cpp @@ -0,0 +1,58 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include + + +#include "crypto/common/VirtualMemory.h" +#include "virtual_memory.hpp" + + +void* allocExecutableMemory(std::size_t bytes) { + void *mem = xmrig::VirtualMemory::allocateExecutableMemory(bytes); + if (mem == nullptr) { + throw std::runtime_error("Failed to allocate executable memory"); + } + + return mem; +} + + +void* allocLargePagesMemory(std::size_t bytes) { + void *mem = xmrig::VirtualMemory::allocateLargePagesMemory(bytes); + if (mem == nullptr) { + throw std::runtime_error("Failed to allocate large pages memory"); + } + + return mem; +} + + +void freePagedMemory(void* ptr, std::size_t bytes) { + xmrig::VirtualMemory::freeLargePagesMemory(ptr, bytes); +} diff --git a/src/crypto/randomx/virtual_memory.hpp b/src/crypto/randomx/virtual_memory.hpp new file mode 100644 index 000000000..d3b31db12 --- /dev/null +++ b/src/crypto/randomx/virtual_memory.hpp @@ -0,0 +1,35 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include + +void* allocExecutableMemory(std::size_t); +void* allocLargePagesMemory(std::size_t); +void freePagedMemory(void*, std::size_t); diff --git a/src/crypto/randomx/vm_compiled.cpp b/src/crypto/randomx/vm_compiled.cpp new file mode 100644 index 000000000..4d14c7931 --- /dev/null +++ b/src/crypto/randomx/vm_compiled.cpp @@ -0,0 +1,58 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include "vm_compiled.hpp" +#include "common.hpp" + +namespace randomx { + + static_assert(sizeof(MemoryRegisters) == 2 * sizeof(addr_t) + sizeof(uintptr_t), "Invalid alignment of struct randomx::MemoryRegisters"); + static_assert(sizeof(RegisterFile) == 256, "Invalid alignment of struct randomx::RegisterFile"); + + template + void CompiledVm::setDataset(randomx_dataset* dataset) { + datasetPtr = dataset; + } + + template + void CompiledVm::run(void* seed) { + VmBase::generateProgram(seed); + randomx_vm::initialize(); + compiler.generateProgram(program, config); + mem.memory = datasetPtr->memory + datasetOffset; + execute(); + } + + template + void CompiledVm::execute() { + compiler.getProgramFunc()(reg, mem, scratchpad, RandomX_CurrentConfig.ProgramIterations); + } + + template class CompiledVm; + template class CompiledVm; +} diff --git a/src/crypto/randomx/vm_compiled.hpp b/src/crypto/randomx/vm_compiled.hpp new file mode 100644 index 000000000..05b34b9c7 --- /dev/null +++ b/src/crypto/randomx/vm_compiled.hpp @@ -0,0 +1,74 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include +#include +#include "virtual_machine.hpp" +#include "jit_compiler.hpp" +#include "allocator.hpp" +#include "dataset.hpp" + +namespace randomx { + + template + class CompiledVm : public VmBase + { + public: + void* operator new(size_t size) { + void* ptr = AlignedAllocator::allocMemory(size); + if (ptr == nullptr) + throw std::bad_alloc(); + return ptr; + } + + void operator delete(void* ptr) { + AlignedAllocator::freeMemory(ptr, sizeof(CompiledVm)); + } + + void setDataset(randomx_dataset* dataset) override; + void run(void* seed) override; + + using VmBase::mem; + using VmBase::program; + using VmBase::config; + using VmBase::reg; + using VmBase::scratchpad; + using VmBase::datasetPtr; + using VmBase::datasetOffset; + + protected: + void execute(); + + JitCompiler compiler; + }; + + using CompiledVmDefault = CompiledVm; + using CompiledVmHardAes = CompiledVm; +} diff --git a/src/crypto/randomx/vm_compiled_light.cpp b/src/crypto/randomx/vm_compiled_light.cpp new file mode 100644 index 000000000..6009216b3 --- /dev/null +++ b/src/crypto/randomx/vm_compiled_light.cpp @@ -0,0 +1,52 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include "vm_compiled_light.hpp" +#include "common.hpp" +#include + +namespace randomx { + + template + void CompiledLightVm::setCache(randomx_cache* cache) { + cachePtr = cache; + mem.memory = cache->memory; + compiler.generateSuperscalarHash(cache->programs, cache->reciprocalCache); + } + + template + void CompiledLightVm::run(void* seed) { + VmBase::generateProgram(seed); + randomx_vm::initialize(); + compiler.generateProgramLight(program, config, datasetOffset); + CompiledVm::execute(); + } + + template class CompiledLightVm; + template class CompiledLightVm; +} diff --git a/src/crypto/randomx/vm_compiled_light.hpp b/src/crypto/randomx/vm_compiled_light.hpp new file mode 100644 index 000000000..6cd3cb203 --- /dev/null +++ b/src/crypto/randomx/vm_compiled_light.hpp @@ -0,0 +1,65 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include +#include "vm_compiled.hpp" + +namespace randomx { + + template + class CompiledLightVm : public CompiledVm + { + public: + void* operator new(size_t size) { + void* ptr = AlignedAllocator::allocMemory(size); + if (ptr == nullptr) + throw std::bad_alloc(); + return ptr; + } + + void operator delete(void* ptr) { + AlignedAllocator::freeMemory(ptr, sizeof(CompiledLightVm)); + } + + void setCache(randomx_cache* cache) override; + void setDataset(randomx_dataset* dataset) override { } + void run(void* seed) override; + + using CompiledVm::mem; + using CompiledVm::compiler; + using CompiledVm::program; + using CompiledVm::config; + using CompiledVm::cachePtr; + using CompiledVm::datasetOffset; + }; + + using CompiledLightVmDefault = CompiledLightVm; + using CompiledLightVmHardAes = CompiledLightVm; +} diff --git a/src/crypto/randomx/vm_interpreted.cpp b/src/crypto/randomx/vm_interpreted.cpp new file mode 100644 index 000000000..f4c1e05c9 --- /dev/null +++ b/src/crypto/randomx/vm_interpreted.cpp @@ -0,0 +1,123 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include "vm_interpreted.hpp" +#include "dataset.hpp" +#include "intrin_portable.h" +#include "reciprocal.h" + +namespace randomx { + + template + void InterpretedVm::setDataset(randomx_dataset* dataset) { + datasetPtr = dataset; + mem.memory = dataset->memory; + } + + template + void InterpretedVm::run(void* seed) { + VmBase::generateProgram(seed); + randomx_vm::initialize(); + execute(); + } + + template + void InterpretedVm::execute() { + + NativeRegisterFile nreg; + + for(unsigned i = 0; i < RegisterCountFlt; ++i) + nreg.a[i] = rx_load_vec_f128(®.a[i].lo); + + compileProgram(program, bytecode, nreg); + + uint32_t spAddr0 = mem.mx; + uint32_t spAddr1 = mem.ma; + + for(unsigned ic = 0; ic < RandomX_CurrentConfig.ProgramIterations; ++ic) { + uint64_t spMix = nreg.r[config.readReg0] ^ nreg.r[config.readReg1]; + spAddr0 ^= spMix; + spAddr0 &= ScratchpadL3Mask64; + spAddr1 ^= spMix >> 32; + spAddr1 &= ScratchpadL3Mask64; + + for (unsigned i = 0; i < RegistersCount; ++i) + nreg.r[i] ^= load64(scratchpad + spAddr0 + 8 * i); + + for (unsigned i = 0; i < RegisterCountFlt; ++i) + nreg.f[i] = rx_cvt_packed_int_vec_f128(scratchpad + spAddr1 + 8 * i); + + for (unsigned i = 0; i < RegisterCountFlt; ++i) + nreg.e[i] = maskRegisterExponentMantissa(config, rx_cvt_packed_int_vec_f128(scratchpad + spAddr1 + 8 * (RegisterCountFlt + i))); + + executeBytecode(bytecode, scratchpad, config); + + mem.mx ^= nreg.r[config.readReg2] ^ nreg.r[config.readReg3]; + mem.mx &= CacheLineAlignMask; + datasetPrefetch(datasetOffset + mem.mx); + datasetRead(datasetOffset + mem.ma, nreg.r); + std::swap(mem.mx, mem.ma); + + for (unsigned i = 0; i < RegistersCount; ++i) + store64(scratchpad + spAddr1 + 8 * i, nreg.r[i]); + + for (unsigned i = 0; i < RegisterCountFlt; ++i) + nreg.f[i] = rx_xor_vec_f128(nreg.f[i], nreg.e[i]); + + for (unsigned i = 0; i < RegisterCountFlt; ++i) + rx_store_vec_f128((double*)(scratchpad + spAddr0 + 16 * i), nreg.f[i]); + + spAddr0 = 0; + spAddr1 = 0; + } + + for (unsigned i = 0; i < RegistersCount; ++i) + store64(®.r[i], nreg.r[i]); + + for (unsigned i = 0; i < RegisterCountFlt; ++i) + rx_store_vec_f128(®.f[i].lo, nreg.f[i]); + + for (unsigned i = 0; i < RegisterCountFlt; ++i) + rx_store_vec_f128(®.e[i].lo, nreg.e[i]); + } + + template + void InterpretedVm::datasetRead(uint64_t address, int_reg_t(&r)[RegistersCount]) { + uint64_t* datasetLine = (uint64_t*)(mem.memory + address); + for (int i = 0; i < RegistersCount; ++i) + r[i] ^= datasetLine[i]; + } + + template + void InterpretedVm::datasetPrefetch(uint64_t address) { + rx_prefetch_nta(mem.memory + address); + } + + template class InterpretedVm; + template class InterpretedVm; +} diff --git a/src/crypto/randomx/vm_interpreted.hpp b/src/crypto/randomx/vm_interpreted.hpp new file mode 100644 index 000000000..1dc9ab6dc --- /dev/null +++ b/src/crypto/randomx/vm_interpreted.hpp @@ -0,0 +1,78 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include +#include +#include "common.hpp" +#include "virtual_machine.hpp" +#include "bytecode_machine.hpp" +#include "intrin_portable.h" +#include "allocator.hpp" + +namespace randomx { + + template + class InterpretedVm : public VmBase, public BytecodeMachine { + public: + using VmBase::mem; + using VmBase::scratchpad; + using VmBase::program; + using VmBase::config; + using VmBase::reg; + using VmBase::datasetPtr; + using VmBase::datasetOffset; + + void* operator new(size_t size) { + void* ptr = AlignedAllocator::allocMemory(size); + if (ptr == nullptr) + throw std::bad_alloc(); + return ptr; + } + + void operator delete(void* ptr) { + AlignedAllocator::freeMemory(ptr, sizeof(InterpretedVm)); + } + + void run(void* seed) override; + void setDataset(randomx_dataset* dataset) override; + + protected: + virtual void datasetRead(uint64_t blockNumber, int_reg_t(&r)[RegistersCount]); + virtual void datasetPrefetch(uint64_t blockNumber); + + private: + void execute(); + + InstructionByteCode bytecode[RANDOMX_PROGRAM_MAX_SIZE]; + }; + + using InterpretedVmDefault = InterpretedVm; + using InterpretedVmHardAes = InterpretedVm; +} diff --git a/src/crypto/randomx/vm_interpreted_light.cpp b/src/crypto/randomx/vm_interpreted_light.cpp new file mode 100644 index 000000000..9c97187b2 --- /dev/null +++ b/src/crypto/randomx/vm_interpreted_light.cpp @@ -0,0 +1,53 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#include "vm_interpreted_light.hpp" +#include "dataset.hpp" + +namespace randomx { + + template + void InterpretedLightVm::setCache(randomx_cache* cache) { + cachePtr = cache; + mem.memory = cache->memory; + } + + template + void InterpretedLightVm::datasetRead(uint64_t address, int_reg_t(&r)[8]) { + uint32_t itemNumber = address / CacheLineSize; + int_reg_t rl[8]; + + initDatasetItem(cachePtr, (uint8_t*)rl, itemNumber); + + for (unsigned q = 0; q < 8; ++q) + r[q] ^= rl[q]; + } + + template class InterpretedLightVm; + template class InterpretedLightVm; +} diff --git a/src/crypto/randomx/vm_interpreted_light.hpp b/src/crypto/randomx/vm_interpreted_light.hpp new file mode 100644 index 000000000..1a35c5801 --- /dev/null +++ b/src/crypto/randomx/vm_interpreted_light.hpp @@ -0,0 +1,63 @@ +/* +Copyright (c) 2018-2019, tevador + +All rights reserved. + +Redistribution and use in source and binary forms, with or without +modification, are permitted provided that the following conditions are met: + * Redistributions of source code must retain the above copyright + notice, this list of conditions and the following disclaimer. + * Redistributions in binary form must reproduce the above copyright + notice, this list of conditions and the following disclaimer in the + documentation and/or other materials provided with the distribution. + * Neither the name of the copyright holder nor the + names of its contributors may be used to endorse or promote products + derived from this software without specific prior written permission. + +THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND +ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED +WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE +DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE +FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL +DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR +SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER +CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, +OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE +OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. +*/ + +#pragma once + +#include +#include "vm_interpreted.hpp" + +namespace randomx { + + template + class InterpretedLightVm : public InterpretedVm { + public: + using VmBase::mem; + using VmBase::cachePtr; + + void* operator new(size_t size) { + void* ptr = AlignedAllocator::allocMemory(size); + if (ptr == nullptr) + throw std::bad_alloc(); + return ptr; + } + + void operator delete(void* ptr) { + AlignedAllocator::freeMemory(ptr, sizeof(InterpretedLightVm)); + } + + void setDataset(randomx_dataset* dataset) override { } + void setCache(randomx_cache* cache) override; + + protected: + void datasetRead(uint64_t address, int_reg_t(&r)[8]) override; + void datasetPrefetch(uint64_t address) override { } + }; + + using InterpretedLightVmDefault = InterpretedLightVm; + using InterpretedLightVmHardAes = InterpretedLightVm; +} diff --git a/src/crypto/rx/Rx.cpp b/src/crypto/rx/Rx.cpp new file mode 100644 index 000000000..8e757ddf2 --- /dev/null +++ b/src/crypto/rx/Rx.cpp @@ -0,0 +1,323 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2019 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 tevador + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include +#include +#include + + +#ifdef XMRIG_FEATURE_HWLOC +# include +# include "backend/cpu/platform/HwlocCpuInfo.h" +#endif + + +#include "backend/cpu/Cpu.h" +#include "base/io/log/Log.h" +#include "base/kernel/Platform.h" +#include "base/net/stratum/Job.h" +#include "base/tools/Buffer.h" +#include "base/tools/Chrono.h" +#include "crypto/rx/Rx.h" +#include "crypto/rx/RxAlgo.h" +#include "crypto/rx/RxCache.h" +#include "crypto/rx/RxDataset.h" + + +namespace xmrig { + + +class RxPrivate; + + +static const char *tag = BLUE_BG(WHITE_BOLD_S " rx ") " "; +static RxPrivate *d_ptr = nullptr; + + +#ifdef XMRIG_FEATURE_HWLOC +static void bindToNUMANode(uint32_t nodeId) +{ + hwloc_topology_t topology; + hwloc_topology_init(&topology); + hwloc_topology_load(topology); + + hwloc_obj_t node = hwloc_get_numanode_obj_by_os_index(topology, nodeId); + if (node) { + if (HwlocCpuInfo::has(HwlocCpuInfo::SET_THISTHREAD_MEMBIND)) { +# if HWLOC_API_VERSION >= 0x20000 + hwloc_set_membind(topology, node->nodeset, HWLOC_MEMBIND_BIND, HWLOC_MEMBIND_THREAD | HWLOC_MEMBIND_BYNODESET); +# else + hwloc_set_membind_nodeset(topology, node->nodeset, HWLOC_MEMBIND_BIND, HWLOC_MEMBIND_THREAD); +# endif + } + + Platform::setThreadAffinity(static_cast(hwloc_bitmap_first(node->cpuset))); + } + + hwloc_topology_destroy(topology); +} +#else +inline static void bindToNUMANode(uint32_t) {} +#endif + + +class RxPrivate +{ +public: + inline RxPrivate() : + m_seed() + { +# ifdef XMRIG_FEATURE_HWLOC + if (Cpu::info()->nodes() > 1) { + for (uint32_t nodeId : HwlocCpuInfo::nodeIndexes()) { + datasets.insert({ nodeId, nullptr }); + } + } + else +# endif + { + datasets.insert({ 0, nullptr }); + } + } + + + inline ~RxPrivate() + { + for (auto const &item : datasets) { + delete item.second; + } + + datasets.clear(); + } + + + inline bool isNUMA() const { return m_numa; } + inline const Algorithm &algorithm() const { return m_algorithm; } + inline const uint8_t *seed() const { return m_seed; } + inline size_t count() const { return isNUMA() ? datasets.size() : 1; } + + + static void allocate(uint32_t nodeId) + { + const uint64_t ts = Chrono::steadyMSecs(); + + if (d_ptr->isNUMA()) { + bindToNUMANode(nodeId); + } + + LOG_INFO("%s" CYAN_BOLD("#%u") MAGENTA_BOLD(" allocate") CYAN_BOLD(" %zu MB") BLACK_BOLD(" (%zu+%zu) for RandomX dataset & cache"), + tag, + nodeId, + (RxDataset::size() + RxCache::size()) / 1024 / 1024, + RxDataset::size() / 1024 / 1024, + RxCache::size() / 1024 / 1024 + ); + + RxDataset *dataset = new RxDataset(d_ptr->m_hugePages); + d_ptr->datasets[nodeId] = dataset; + + if (dataset->get() != nullptr) { + const auto hugePages = dataset->hugePages(); + const double percent = hugePages.first == 0 ? 0.0 : static_cast(hugePages.first) / hugePages.second * 100.0; + + LOG_INFO("%s" CYAN_BOLD("#%u") GREEN(" allocate done") " huge pages %s%u/%u %1.0f%%" CLEAR " %sJIT" BLACK_BOLD(" (%" PRIu64 " ms)"), + tag, + nodeId, + (hugePages.first == hugePages.second ? GREEN_BOLD_S : (hugePages.first == 0 ? RED_BOLD_S : YELLOW_BOLD_S)), + hugePages.first, + hugePages.second, + percent, + dataset->cache()->isJIT() ? GREEN_BOLD_S "+" : RED_BOLD_S "-", + Chrono::steadyMSecs() - ts + ); + } + else { + LOG_WARN(CLEAR "%s" CYAN_BOLD("#%u") YELLOW_BOLD_S " failed to allocate RandomX dataset, switching to slow mode", tag, nodeId); + } + } + + + static void initDataset(uint32_t nodeId, uint32_t threads) + { + std::lock_guard lock(d_ptr->mutex); + + const uint64_t ts = Chrono::steadyMSecs(); + + d_ptr->getOrAllocate(nodeId)->init(d_ptr->seed(), threads); + d_ptr->m_ready++; + + LOG_INFO("%s" CYAN_BOLD("#%u") GREEN(" init done") BLACK_BOLD(" (%" PRIu64 " ms)"), tag, nodeId, Chrono::steadyMSecs() - ts); + } + + + inline RxDataset *getOrAllocate(uint32_t nodeId) + { + RxDataset *dataset = datasets.at(nodeId); + + if (dataset == nullptr) { + # ifdef XMRIG_FEATURE_HWLOC + if (d_ptr->isNUMA()) { + std::thread thread(allocate, nodeId); + thread.join(); + } else + # endif + { + allocate(nodeId); + } + + dataset = datasets.at(nodeId); + } + + return dataset; + } + + + inline void setState(const Job &job, bool hugePages, bool numa) + { + if (m_algorithm != job.algorithm()) { + m_algorithm = RxAlgo::apply(job.algorithm()); + } + + m_ready = 0; + m_numa = numa && Cpu::info()->nodes() > 1; + m_hugePages = hugePages; + + memcpy(m_seed, job.seedHash(), sizeof(m_seed)); + } + + + inline bool isReady(const Job &job) + { + return m_ready == count() && m_algorithm == job.algorithm() && memcmp(m_seed, job.seedHash(), sizeof(m_seed)) == 0; + } + + + std::map datasets; + std::mutex mutex; + +private: + bool m_hugePages = true; + bool m_numa = true; + Algorithm m_algorithm; + size_t m_ready = 0; + uint8_t m_seed[32]; +}; + + +} // namespace xmrig + + +bool xmrig::Rx::isReady(const Job &job) +{ + std::lock_guard lock(d_ptr->mutex); + + return d_ptr->isReady(job); +} + + +xmrig::RxDataset *xmrig::Rx::dataset(const Job &job, uint32_t nodeId) +{ + std::lock_guard lock(d_ptr->mutex); + if (!d_ptr->isReady(job)) { + return nullptr; + } + + return d_ptr->datasets.at(d_ptr->isNUMA() ? nodeId : 0); +} + + +std::pair xmrig::Rx::hugePages() +{ + std::pair pages(0, 0); + std::lock_guard lock(d_ptr->mutex); + + for (auto const &item : d_ptr->datasets) { + if (!item.second) { + continue; + } + + const auto p = item.second->hugePages(); + pages.first += p.first; + pages.second += p.second; + } + + return pages; +} + + +void xmrig::Rx::destroy() +{ + delete d_ptr; + + d_ptr = nullptr; +} + + +void xmrig::Rx::init() +{ + d_ptr = new RxPrivate(); +} + + +void xmrig::Rx::init(const Job &job, int initThreads, bool hugePages, bool numa) +{ + if (job.algorithm().family() != Algorithm::RANDOM_X) { + return; + } + + std::lock_guard lock(d_ptr->mutex); + + if (d_ptr->isReady(job)) { + return; + } + + d_ptr->setState(job, hugePages, numa); + const uint32_t threads = initThreads < 1 ? static_cast(Cpu::info()->threads()) : static_cast(initThreads); + const String buf = Buffer::toHex(job.seedHash(), 8); + + LOG_INFO("%s" MAGENTA_BOLD("init dataset%s") " algo " WHITE_BOLD("%s (") CYAN_BOLD("%u") WHITE_BOLD(" threads)") BLACK_BOLD(" seed %s..."), + tag, + d_ptr->count() > 1 ? "s" : "", + job.algorithm().shortName(), + threads, + buf.data() + ); + +# ifdef XMRIG_FEATURE_HWLOC + if (d_ptr->isNUMA()) { + for (auto const &item : d_ptr->datasets) { + std::thread thread(RxPrivate::initDataset, item.first, threads); + thread.detach(); + } + } + else +# endif + { + std::thread thread(RxPrivate::initDataset, 0, threads); + thread.detach(); + } +} diff --git a/src/crypto/rx/Rx.h b/src/crypto/rx/Rx.h new file mode 100644 index 000000000..1a383055a --- /dev/null +++ b/src/crypto/rx/Rx.h @@ -0,0 +1,59 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2019 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 tevador + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_RX_H +#define XMRIG_RX_H + + +#include +#include + + +namespace xmrig +{ + + +class Algorithm; +class RxDataset; +class Job; + + +class Rx +{ +public: + static bool isReady(const Job &job); + static RxDataset *dataset(const Job &job, uint32_t nodeId); + static std::pair hugePages(); + static void destroy(); + static void init(); + static void init(const Job &job, int initThreads, bool hugePages, bool numa); +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_RX_H */ diff --git a/src/crypto/rx/RxAlgo.cpp b/src/crypto/rx/RxAlgo.cpp new file mode 100644 index 000000000..b0e92e6e0 --- /dev/null +++ b/src/crypto/rx/RxAlgo.cpp @@ -0,0 +1,69 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2019 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 tevador + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "crypto/randomx/randomx.h" +#include "crypto/rx/RxAlgo.h" + + +xmrig::Algorithm::Id xmrig::RxAlgo::apply(Algorithm::Id algorithm) +{ + switch (algorithm) { + case Algorithm::RX_WOW: + randomx_apply_config(RandomX_WowneroConfig); + break; + + case Algorithm::RX_LOKI: + randomx_apply_config(RandomX_LokiConfig); + break; + + default: + randomx_apply_config(RandomX_MoneroConfig); + break; + } + + return algorithm; +} + + +size_t xmrig::RxAlgo::l3(Algorithm::Id algorithm) +{ + switch (algorithm) { + case Algorithm::RX_0: + return RandomX_MoneroConfig.ScratchpadL3_Size; + + case Algorithm::RX_WOW: + return RandomX_WowneroConfig.ScratchpadL3_Size; + + case Algorithm::RX_LOKI: + return RandomX_LokiConfig.ScratchpadL3_Size; + + default: + break; + } + + return 0; +} diff --git a/src/crypto/rx/RxAlgo.h b/src/crypto/rx/RxAlgo.h new file mode 100644 index 000000000..dd3f0aa73 --- /dev/null +++ b/src/crypto/rx/RxAlgo.h @@ -0,0 +1,56 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2019 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 tevador + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_RX_ALGO_H +#define XMRIG_RX_ALGO_H + + +#include +#include + + +#include "crypto/common/Algorithm.h" + + +struct RandomX_ConfigurationBase; + + +namespace xmrig +{ + + +class RxAlgo +{ +public: + static Algorithm::Id apply(Algorithm::Id algorithm); + static size_t l3(Algorithm::Id algorithm); +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_RX_ALGO_H */ diff --git a/src/crypto/rx/RxCache.cpp b/src/crypto/rx/RxCache.cpp new file mode 100644 index 000000000..92c366a2e --- /dev/null +++ b/src/crypto/rx/RxCache.cpp @@ -0,0 +1,83 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2019 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 tevador + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "crypto/randomx/randomx.h" +#include "crypto/rx/RxCache.h" + + +static_assert(RANDOMX_FLAG_JIT == 8, "RANDOMX_FLAG_JIT flag mismatch"); +static_assert(RANDOMX_FLAG_LARGE_PAGES == 1, "RANDOMX_FLAG_LARGE_PAGES flag mismatch"); + + + +xmrig::RxCache::RxCache(bool hugePages) : + m_seed() +{ + if (hugePages) { + m_flags = RANDOMX_FLAG_JIT | RANDOMX_FLAG_LARGE_PAGES; + m_cache = randomx_alloc_cache(static_cast(m_flags)); + } + + if (!m_cache) { + m_flags = RANDOMX_FLAG_JIT; + m_cache = randomx_alloc_cache(static_cast(m_flags)); + } + + if (!m_cache) { + m_flags = RANDOMX_FLAG_DEFAULT; + m_cache = randomx_alloc_cache(static_cast(m_flags)); + } +} + + +xmrig::RxCache::~RxCache() +{ + if (m_cache) { + randomx_release_cache(m_cache); + } +} + + +bool xmrig::RxCache::init(const uint8_t *seed) +{ + if (isReady(seed)) { + return false; + } + + memcpy(m_seed, seed, sizeof(m_seed)); + randomx_init_cache(m_cache, m_seed, sizeof(m_seed)); + + m_initCount++; + + return true; +} + + +bool xmrig::RxCache::isReady(const uint8_t *seed) const +{ + return m_initCount && memcmp(m_seed, seed, sizeof(m_seed)) == 0; +} diff --git a/src/Mem.h b/src/crypto/rx/RxCache.h similarity index 55% rename from src/Mem.h rename to src/crypto/rx/RxCache.h index 9e39e963c..e6b2397ca 100644 --- a/src/Mem.h +++ b/src/crypto/rx/RxCache.h @@ -4,8 +4,9 @@ * Copyright 2014 Lucas Jones * Copyright 2014-2016 Wolf9466 * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , + * Copyright 2017-2019 XMR-Stak , * Copyright 2018 Lee Clagett + * Copyright 2018-2019 tevador * Copyright 2018-2019 SChernykh * Copyright 2016-2019 XMRig , * @@ -23,56 +24,50 @@ * along with this program. If not, see . */ -#ifndef XMRIG_MEM_H -#define XMRIG_MEM_H +#ifndef XMRIG_RX_CACHE_H +#define XMRIG_RX_CACHE_H -#include #include -#include "common/xmrig.h" +#include "crypto/randomx/configuration.h" -struct cryptonight_ctx; +struct randomx_cache; -struct MemInfo +namespace xmrig { - alignas(16) uint8_t *memory; - - size_t hugePages; - size_t pages; - size_t size; -}; -class Mem +class RxCache { public: - enum Flags { - HugepagesAvailable = 1, - HugepagesEnabled = 2, - Lock = 4 - }; + RxCache(bool hugePages = true); + ~RxCache(); - static MemInfo create(cryptonight_ctx **ctx, xmrig::Algo algorithm, size_t count); - static void init(bool enabled); - static void release(cryptonight_ctx **ctx, size_t count, MemInfo &info); + inline bool isHugePages() const { return m_flags & 1; } + inline bool isJIT() const { return m_flags & 8; } + inline const uint8_t *seed() const { return m_seed; } + inline randomx_cache *get() const { return m_cache; } + inline uint64_t initCount() const { return m_initCount; } - static void *allocateExecutableMemory(size_t size); - static void protectExecutableMemory(void *p, size_t size); - static void flushInstructionCache(void *p, size_t size); + bool init(const uint8_t *seed); - static inline bool isHugepagesAvailable() { return (m_flags & HugepagesAvailable) != 0; } + static inline constexpr size_t size() { return RANDOMX_CACHE_MAX_SIZE; } private: - static void allocate(MemInfo &info, bool enabled); - static void release(MemInfo &info); + bool isReady(const uint8_t *seed) const; - static int m_flags; - static bool m_enabled; + int m_flags = 0; + randomx_cache *m_cache = nullptr; + uint64_t m_initCount = 0; + uint8_t m_seed[32]; }; -#endif /* XMRIG_MEM_H */ +} /* namespace xmrig */ + + +#endif /* XMRIG_RX_CACHE_H */ diff --git a/src/crypto/rx/RxConfig.cpp b/src/crypto/rx/RxConfig.cpp new file mode 100644 index 000000000..dc543fb43 --- /dev/null +++ b/src/crypto/rx/RxConfig.cpp @@ -0,0 +1,63 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "base/io/json/Json.h" +#include "crypto/rx/RxConfig.h" +#include "rapidjson/document.h" + + +namespace xmrig { + +static const char *kInit = "init"; +static const char *kNUMA = "numa"; + +} + + +rapidjson::Value xmrig::RxConfig::toJSON(rapidjson::Document &doc) const +{ + using namespace rapidjson; + auto &allocator = doc.GetAllocator(); + + Value obj(kObjectType); + + obj.AddMember(StringRef(kInit), m_threads, allocator); + obj.AddMember(StringRef(kNUMA), m_numa, allocator); + + return obj; +} + + +bool xmrig::RxConfig::read(const rapidjson::Value &value) +{ + if (value.IsObject()) { + m_numa = Json::getBool(value, kNUMA, m_numa); + m_threads = Json::getInt(value, kInit, m_threads); + + return true; + } + + return false; +} diff --git a/src/common/net/SubmitResult.h b/src/crypto/rx/RxConfig.h similarity index 74% rename from src/common/net/SubmitResult.h rename to src/crypto/rx/RxConfig.h index db97ac38a..e06c764cf 100644 --- a/src/common/net/SubmitResult.h +++ b/src/crypto/rx/RxConfig.h @@ -22,36 +22,32 @@ * along with this program. If not, see . */ -#ifndef XMRIG_SUBMITRESULT_H -#define XMRIG_SUBMITRESULT_H +#ifndef XMRIG_RXCONFIG_H +#define XMRIG_RXCONFIG_H -#include +#include "rapidjson/fwd.h" namespace xmrig { -class SubmitResult +class RxConfig { public: - inline SubmitResult() : reqId(0), seq(0), diff(0), actualDiff(0), elapsed(0), start(0) {} - SubmitResult(int64_t seq, uint32_t diff, uint64_t actualDiff, int64_t reqId = 0); + bool read(const rapidjson::Value &value); + rapidjson::Value toJSON(rapidjson::Document &doc) const; - void done(); - - int64_t reqId; - int64_t seq; - uint32_t diff; - uint64_t actualDiff; - uint64_t elapsed; + inline bool isNUMA() const { return m_numa; } + inline int threads() const { return m_threads; } private: - uint64_t start; + bool m_numa = true; + int m_threads = -1; }; } /* namespace xmrig */ -#endif /* XMRIG_SUBMITRESULT_H */ +#endif /* XMRIG_RXCONFIG_H */ diff --git a/src/crypto/rx/RxDataset.cpp b/src/crypto/rx/RxDataset.cpp new file mode 100644 index 000000000..50459a555 --- /dev/null +++ b/src/crypto/rx/RxDataset.cpp @@ -0,0 +1,114 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2019 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 tevador + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include + + +#include "crypto/common/VirtualMemory.h" +#include "crypto/randomx/randomx.h" +#include "crypto/rx/RxAlgo.h" +#include "crypto/rx/RxCache.h" +#include "crypto/rx/RxDataset.h" + + +static_assert(RANDOMX_FLAG_LARGE_PAGES == 1, "RANDOMX_FLAG_LARGE_PAGES flag mismatch"); + + +xmrig::RxDataset::RxDataset(bool hugePages) +{ + if (hugePages) { + m_flags = RANDOMX_FLAG_LARGE_PAGES; + m_dataset = randomx_alloc_dataset(static_cast(m_flags)); + } + + if (!m_dataset) { + m_flags = RANDOMX_FLAG_DEFAULT; + m_dataset = randomx_alloc_dataset(static_cast(m_flags)); + } + + m_cache = new RxCache(hugePages); +} + + +xmrig::RxDataset::~RxDataset() +{ + if (m_dataset) { + randomx_release_dataset(m_dataset); + } + + delete m_cache; +} + + +bool xmrig::RxDataset::init(const uint8_t *seed, uint32_t numThreads) +{ + cache()->init(seed); + + if (!get()) { + return true; + } + + const uint32_t datasetItemCount = randomx_dataset_item_count(); + + if (numThreads > 1) { + std::vector threads; + threads.reserve(numThreads); + + for (uint32_t i = 0; i < numThreads; ++i) { + const uint32_t a = (datasetItemCount * i) / numThreads; + const uint32_t b = (datasetItemCount * (i + 1)) / numThreads; + threads.emplace_back(randomx_init_dataset, m_dataset, m_cache->get(), a, b - a); + } + + for (uint32_t i = 0; i < numThreads; ++i) { + threads[i].join(); + } + } + else { + randomx_init_dataset(m_dataset, m_cache->get(), 0, datasetItemCount); + } + + return true; +} + + +std::pair xmrig::RxDataset::hugePages() const +{ + constexpr size_t twoMiB = 2u * 1024u * 1024u; + constexpr const size_t total = (VirtualMemory::align(size(), twoMiB) + VirtualMemory::align(RxCache::size(), twoMiB)) / twoMiB; + + size_t count = 0; + if (isHugePages()) { + count += VirtualMemory::align(size(), twoMiB) / twoMiB; + } + + if (m_cache->isHugePages()) { + count += VirtualMemory::align(RxCache::size(), twoMiB) / twoMiB; + } + + return std::pair(count, total); +} diff --git a/src/crypto/rx/RxDataset.h b/src/crypto/rx/RxDataset.h new file mode 100644 index 000000000..932f4ed94 --- /dev/null +++ b/src/crypto/rx/RxDataset.h @@ -0,0 +1,71 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2019 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 tevador + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_RX_DATASET_H +#define XMRIG_RX_DATASET_H + + +#include "crypto/common/Algorithm.h" +#include "crypto/randomx/configuration.h" + + +struct randomx_dataset; + + +namespace xmrig +{ + + +class RxCache; + + +class RxDataset +{ +public: + RxDataset(bool hugePages = true); + ~RxDataset(); + + inline bool isHugePages() const { return m_flags & 1; } + inline randomx_dataset *get() const { return m_dataset; } + inline RxCache *cache() const { return m_cache; } + + bool init(const uint8_t *seed, uint32_t numThreads); + std::pair hugePages() const; + + static inline constexpr size_t size() { return RANDOMX_DATASET_MAX_SIZE; } + +private: + Algorithm m_algorithm; + int m_flags = 0; + randomx_dataset *m_dataset = nullptr; + RxCache *m_cache = nullptr; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_RX_DATASET_H */ diff --git a/src/crypto/rx/RxVm.cpp b/src/crypto/rx/RxVm.cpp new file mode 100644 index 000000000..275f9558b --- /dev/null +++ b/src/crypto/rx/RxVm.cpp @@ -0,0 +1,59 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2019 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 tevador + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include "crypto/randomx/randomx.h" +#include "crypto/rx/RxCache.h" +#include "crypto/rx/RxDataset.h" +#include "crypto/rx/RxVm.h" + + +xmrig::RxVm::RxVm(RxDataset *dataset, uint8_t *scratchpad, bool softAes) +{ +# ifndef XMRIG_ARM + if (!softAes) { + m_flags |= RANDOMX_FLAG_HARD_AES; + } +# endif + + if (dataset->get()) { + m_flags |= RANDOMX_FLAG_FULL_MEM; + } + + if (dataset->cache()->isJIT()) { + m_flags |= RANDOMX_FLAG_JIT; + } + + m_vm = randomx_create_vm(static_cast(m_flags), dataset->cache()->get(), dataset->get(), scratchpad); +} + + +xmrig::RxVm::~RxVm() +{ + if (m_vm) { + randomx_destroy_vm(m_vm); + } +} diff --git a/src/crypto/rx/RxVm.h b/src/crypto/rx/RxVm.h new file mode 100644 index 000000000..d7e617e4d --- /dev/null +++ b/src/crypto/rx/RxVm.h @@ -0,0 +1,61 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2019 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 tevador + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_RX_VM_H +#define XMRIG_RX_VM_H + + +#include + + +struct randomx_vm; + + +namespace xmrig +{ + + +class RxDataset; + + +class RxVm +{ +public: + RxVm(RxDataset *dataset, uint8_t *scratchpad, bool softAes); + ~RxVm(); + + inline randomx_vm *get() const { return m_vm; } + +private: + int m_flags = 0; + randomx_vm *m_vm = nullptr; +}; + + +} /* namespace xmrig */ + + +#endif /* XMRIG_RX_CACHE_H */ diff --git a/src/net/JobResult.h b/src/net/JobResult.h index 5ab4d2942..2c2fded53 100644 --- a/src/net/JobResult.h +++ b/src/net/JobResult.h @@ -31,7 +31,8 @@ #include -#include "common/net/Job.h" +#include "base/tools/String.h" +#include "base/net/stratum/Job.h" namespace xmrig { @@ -40,43 +41,31 @@ namespace xmrig { class JobResult { public: - inline JobResult() : poolId(0), diff(0), nonce(0) {} - inline JobResult(int poolId, const Id &jobId, const Id &clientId, uint32_t nonce, const uint8_t *result, uint32_t diff, const Algorithm &algorithm) : - algorithm(algorithm), - clientId(clientId), - jobId(jobId), - poolId(poolId), - diff(diff), - nonce(nonce) + inline JobResult() {} + + inline JobResult(const Job &job, uint32_t nonce, const uint8_t *result) : + algorithm(job.algorithm()), + clientId(job.clientId()), + jobId(job.id()), + nonce(nonce), + diff(job.diff()), + index(job.index()) { - memcpy(this->result, result, sizeof(this->result)); + memcpy(m_result, result, sizeof(m_result)); } + inline const uint8_t *result() const { return m_result; } + inline uint64_t actualDiff() const { return Job::toDiff(reinterpret_cast(m_result)[3]); } - inline JobResult(const Job &job) : poolId(0), diff(0), nonce(0) - { - jobId = job.id(); - clientId = job.clientId(); - poolId = job.poolId(); - diff = job.diff(); - nonce = *job.nonce(); - algorithm = job.algorithm(); - } + const Algorithm algorithm; + const String clientId; + const String jobId; + const uint32_t nonce = 0; + const uint64_t diff = 0; + const uint8_t index = 0; - - inline uint64_t actualDiff() const - { - return Job::toDiff(reinterpret_cast(result)[3]); - } - - - Algorithm algorithm; - Id clientId; - Id jobId; - int poolId; - uint32_t diff; - uint32_t nonce; - uint8_t result[32]; +private: + uint8_t m_result[32]; }; diff --git a/src/net/JobResults.cpp b/src/net/JobResults.cpp new file mode 100644 index 000000000..40c2e50be --- /dev/null +++ b/src/net/JobResults.cpp @@ -0,0 +1,134 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + + +#include +#include +#include +#include + + +#include "base/tools/Handle.h" +#include "net/interfaces/IJobResultListener.h" +#include "net/JobResult.h" +#include "net/JobResults.h" + + +namespace xmrig { + + +class JobResultsPrivate +{ +public: + inline JobResultsPrivate(IJobResultListener *listener) : + listener(listener) + { + async = new uv_async_t; + async->data = this; + + uv_async_init(uv_default_loop(), async, JobResultsPrivate::onResult); + } + + + inline ~JobResultsPrivate() + { + Handle::close(async); + } + + + void submit(const JobResult &result) + { + mutex.lock(); + queue.push_back(result); + mutex.unlock(); + + uv_async_send(async); + } + + +private: + static void onResult(uv_async_t *handle) { static_cast(handle->data)->submit(); } + + + inline void submit() + { + std::list results; + + mutex.lock(); + + while (!queue.empty()) { + results.push_back(std::move(queue.front())); + queue.pop_front(); + } + + mutex.unlock(); + + for (auto result : results) { + listener->onJobResult(result); + } + + results.clear(); + } + + + IJobResultListener *listener; + std::list queue; + std::mutex mutex; + uv_async_t *async; +}; + + +static JobResultsPrivate *handler = nullptr; + + +} // namespace xmrig + + + +void xmrig::JobResults::setListener(IJobResultListener *listener) +{ + assert(handler == nullptr); + + handler = new JobResultsPrivate(listener); +} + + +void xmrig::JobResults::stop() +{ + assert(handler != nullptr); + + delete handler; + + handler = nullptr; +} + + +void xmrig::JobResults::submit(const JobResult &result) +{ + assert(handler != nullptr); + + if (handler) { + handler->submit(result); + } +} diff --git a/src/net/JobResults.h b/src/net/JobResults.h new file mode 100644 index 000000000..e7082acb8 --- /dev/null +++ b/src/net/JobResults.h @@ -0,0 +1,49 @@ +/* XMRig + * Copyright 2010 Jeff Garzik + * Copyright 2012-2014 pooler + * Copyright 2014 Lucas Jones + * Copyright 2014-2016 Wolf9466 + * Copyright 2016 Jay D Dee + * Copyright 2017-2018 XMR-Stak , + * Copyright 2018 Lee Clagett + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + */ + +#ifndef XMRIG_JOBRESULTS_H +#define XMRIG_JOBRESULTS_H + + +namespace xmrig { + + +class IJobResultListener; +class JobResult; + + +class JobResults +{ +public: + static void setListener(IJobResultListener *listener); + static void stop(); + static void submit(const JobResult &result); +}; + + +} // namespace xmrig + + +#endif /* XMRIG_JOBRESULTS_H */ diff --git a/src/net/Network.cpp b/src/net/Network.cpp index 34714c8ad..547a86387 100644 --- a/src/net/Network.cpp +++ b/src/net/Network.cpp @@ -6,6 +6,7 @@ * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , * Copyright 2018-2019 SChernykh + * Copyright 2019 Howard Chu * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify @@ -26,44 +27,68 @@ #pragma warning(disable:4244) #endif +#include #include +#include #include #include #include "api/Api.h" -#include "common/log/Log.h" -#include "common/net/Client.h" -#include "common/net/SubmitResult.h" -#include "core/Config.h" +#include "base/io/log/Log.h" +#include "base/net/stratum/Client.h" +#include "base/net/stratum/SubmitResult.h" +#include "base/tools/Chrono.h" +#include "base/tools/Timer.h" +#include "core/config/Config.h" #include "core/Controller.h" +#include "core/Miner.h" +#include "net/JobResult.h" +#include "net/JobResults.h" #include "net/Network.h" #include "net/strategies/DonateStrategy.h" -#include "workers/Workers.h" +#include "rapidjson/document.h" + + +#ifdef XMRIG_FEATURE_API +# include "api/Api.h" +# include "api/interfaces/IApiRequest.h" +#endif xmrig::Network::Network(Controller *controller) : - m_donate(nullptr) + m_controller(controller), + m_donate(nullptr), + m_timer(nullptr) { - Workers::setListener(this); + JobResults::setListener(this); controller->addListener(this); +# ifdef XMRIG_FEATURE_API + controller->api()->addListener(this); +# endif + const Pools &pools = controller->config()->pools(); m_strategy = pools.createStrategy(this); - if (controller->config()->donateLevel() > 0) { - m_donate = new DonateStrategy(controller->config()->donateLevel(), pools.data().front().user(), controller->config()->algorithm().algo(), this); + if (pools.donateLevel() > 0) { + m_donate = new DonateStrategy(controller, this); } - m_timer.data = this; - uv_timer_init(uv_default_loop(), &m_timer); - - uv_timer_start(&m_timer, Network::onTick, kTickInterval, kTickInterval); + m_timer = new Timer(this, kTickInterval, kTickInterval); } xmrig::Network::~Network() { + JobResults::stop(); + + delete m_timer; + + if (m_donate) { + delete m_donate; + } + delete m_strategy; } @@ -74,33 +99,22 @@ void xmrig::Network::connect() } -void xmrig::Network::stop() -{ - if (m_donate) { - m_donate->stop(); - } - - m_strategy->stop(); -} - - -void xmrig::Network::onActive(IStrategy *strategy, Client *client) +void xmrig::Network::onActive(IStrategy *strategy, IClient *client) { if (m_donate && m_donate == strategy) { LOG_NOTICE("dev donate started"); return; } - m_state.setPool(client->host(), client->port(), client->ip()); + m_state.onActive(client); const char *tlsVersion = client->tlsVersion(); - LOG_INFO(isColors() ? WHITE_BOLD("use pool ") CYAN_BOLD("%s:%d ") GREEN_BOLD("%s") " \x1B[1;30m%s " - : "use pool %s:%d %s %s", - client->host(), client->port(), tlsVersion ? tlsVersion : "", client->ip()); + LOG_INFO(WHITE_BOLD("use %s ") CYAN_BOLD("%s:%d ") GREEN_BOLD("%s") " " BLACK_BOLD("%s"), + client->mode(), client->pool().host().data(), client->pool().port(), tlsVersion ? tlsVersion : "", client->ip().data()); const char *fingerprint = client->tlsFingerprint(); if (fingerprint != nullptr) { - LOG_INFO("%sfingerprint (SHA-256): \"%s\"", isColors() ? "\x1B[1;30m" : "", fingerprint); + LOG_INFO(BLACK_BOLD("fingerprint (SHA-256): \"%s\""), fingerprint); } } @@ -121,7 +135,7 @@ void xmrig::Network::onConfigChanged(Config *config, Config *previousConfig) } -void xmrig::Network::onJob(IStrategy *strategy, Client *client, const Job &job) +void xmrig::Network::onJob(IStrategy *strategy, IClient *client, const Job &job) { if (m_donate && m_donate->isActive() && m_donate != strategy) { return; @@ -133,7 +147,7 @@ void xmrig::Network::onJob(IStrategy *strategy, Client *client, const Job &job) void xmrig::Network::onJobResult(const JobResult &result) { - if (result.poolId == -1 && m_donate) { + if (result.index == 1 && m_donate) { m_donate->submit(result); return; } @@ -142,6 +156,30 @@ void xmrig::Network::onJobResult(const JobResult &result) } +void xmrig::Network::onLogin(IStrategy *, IClient *client, rapidjson::Document &doc, rapidjson::Value ¶ms) +{ + using namespace rapidjson; + auto &allocator = doc.GetAllocator(); + + Algorithms algorithms = m_controller->miner()->algorithms(); + const Algorithm algorithm = client->pool().algorithm(); + if (algorithm.isValid()) { + const size_t index = static_cast(std::distance(algorithms.begin(), std::find(algorithms.begin(), algorithms.end(), algorithm))); + if (index > 0 && index < algorithms.size()) { + std::swap(algorithms[0], algorithms[index]); + } + } + + Value algo(kArrayType); + + for (const auto &a : algorithms) { + algo.PushBack(StringRef(a.shortName()), allocator); + } + + params.AddMember("algo", algo, allocator); +} + + void xmrig::Network::onPause(IStrategy *strategy) { if (m_donate && m_donate == strategy) { @@ -152,45 +190,59 @@ void xmrig::Network::onPause(IStrategy *strategy) if (!m_strategy->isActive()) { LOG_ERR("no active pools, stop mining"); m_state.stop(); - return Workers::pause(); + + return m_controller->miner()->pause(); } } -void xmrig::Network::onResultAccepted(IStrategy *, Client *, const SubmitResult &result, const char *error) +void xmrig::Network::onResultAccepted(IStrategy *, IClient *, const SubmitResult &result, const char *error) { m_state.add(result, error); if (error) { - LOG_INFO(isColors() ? "\x1B[1;31mrejected\x1B[0m (%" PRId64 "/%" PRId64 ") diff \x1B[1;37m%u\x1B[0m \x1B[31m\"%s\"\x1B[0m \x1B[1;30m(%" PRIu64 " ms)" - : "rejected (%" PRId64 "/%" PRId64 ") diff %u \"%s\" (%" PRIu64 " ms)", + LOG_INFO(RED_BOLD("rejected") " (%" PRId64 "/%" PRId64 ") diff " WHITE_BOLD("%" PRIu64) " " RED("\"%s\"") " " BLACK_BOLD("(%" PRIu64 " ms)"), m_state.accepted, m_state.rejected, result.diff, error, result.elapsed); } else { - LOG_INFO(isColors() ? "\x1B[1;32maccepted\x1B[0m (%" PRId64 "/%" PRId64 ") diff \x1B[1;37m%u\x1B[0m \x1B[1;30m(%" PRIu64 " ms)" - : "accepted (%" PRId64 "/%" PRId64 ") diff %u (%" PRIu64 " ms)", + LOG_INFO(GREEN_BOLD("accepted") " (%" PRId64 "/%" PRId64 ") diff " WHITE_BOLD("%" PRIu64) " " BLACK_BOLD("(%" PRIu64 " ms)"), m_state.accepted, m_state.rejected, result.diff, result.elapsed); } } -bool xmrig::Network::isColors() const +void xmrig::Network::onVerifyAlgorithm(IStrategy *, const IClient *, const Algorithm &algorithm, bool *ok) { - return Log::colors; + if (!m_controller->miner()->isEnabled(algorithm)) { + *ok = false; + + return; + } } -void xmrig::Network::setJob(Client *client, const Job &job, bool donate) +#ifdef XMRIG_FEATURE_API +void xmrig::Network::onRequest(IApiRequest &request) +{ + if (request.type() == IApiRequest::REQ_SUMMARY) { + request.accept(); + + getResults(request.reply(), request.doc(), request.version()); + getConnection(request.reply(), request.doc(), request.version()); + } +} +#endif + + +void xmrig::Network::setJob(IClient *client, const Job &job, bool donate) { if (job.height()) { - LOG_INFO(isColors() ? MAGENTA_BOLD("new job") " from " WHITE_BOLD("%s:%d") " diff " WHITE_BOLD("%d") " algo " WHITE_BOLD("%s") " height " WHITE_BOLD("%" PRIu64) - : "new job from %s:%d diff %d algo %s height %" PRIu64, - client->host(), client->port(), job.diff(), job.algorithm().shortName(), job.height()); + LOG_INFO(MAGENTA_BOLD("new job") " from " WHITE_BOLD("%s:%d") " diff " WHITE_BOLD("%" PRIu64) " algo " WHITE_BOLD("%s") " height " WHITE_BOLD("%" PRIu64), + client->pool().host().data(), client->pool().port(), job.diff(), job.algorithm().shortName(), job.height()); } else { - LOG_INFO(isColors() ? MAGENTA_BOLD("new job") " from " WHITE_BOLD("%s:%d") " diff " WHITE_BOLD("%d") " algo " WHITE_BOLD("%s") - : "new job from %s:%d diff %d algo %s", - client->host(), client->port(), job.diff(), job.algorithm().shortName()); + LOG_INFO(MAGENTA_BOLD("new job") " from " WHITE_BOLD("%s:%d") " diff " WHITE_BOLD("%" PRIu64) " algo " WHITE_BOLD("%s"), + client->pool().host().data(), client->pool().port(), job.diff(), job.algorithm().shortName()); } if (!donate && m_donate) { @@ -198,27 +250,71 @@ void xmrig::Network::setJob(Client *client, const Job &job, bool donate) } m_state.diff = job.diff(); - Workers::setJob(job, donate); + m_controller->miner()->setJob(job, donate); } void xmrig::Network::tick() { - const uint64_t now = uv_now(uv_default_loop()); + const uint64_t now = Chrono::steadyMSecs(); m_strategy->tick(now); if (m_donate) { m_donate->tick(now); } - -# ifndef XMRIG_NO_API - Api::tick(m_state); -# endif } -void xmrig::Network::onTick(uv_timer_t *handle) +#ifdef XMRIG_FEATURE_API +void xmrig::Network::getConnection(rapidjson::Value &reply, rapidjson::Document &doc, int version) const { - static_cast(handle->data)->tick(); + using namespace rapidjson; + auto &allocator = doc.GetAllocator(); + + reply.AddMember("algo", StringRef(m_strategy->client()->job().algorithm().shortName()), allocator); + + Value connection(kObjectType); + connection.AddMember("pool", StringRef(m_state.pool), allocator); + connection.AddMember("ip", m_state.ip().toJSON(), allocator); + connection.AddMember("uptime", m_state.connectionTime(), allocator); + connection.AddMember("ping", m_state.latency(), allocator); + connection.AddMember("failures", m_state.failures, allocator); + connection.AddMember("tls", m_state.tls().toJSON(), allocator); + connection.AddMember("tls-fingerprint", m_state.fingerprint().toJSON(), allocator); + + if (version == 1) { + connection.AddMember("error_log", Value(kArrayType), allocator); + } + + reply.AddMember("connection", connection, allocator); } + + +void xmrig::Network::getResults(rapidjson::Value &reply, rapidjson::Document &doc, int version) const +{ + using namespace rapidjson; + auto &allocator = doc.GetAllocator(); + + Value results(kObjectType); + + results.AddMember("diff_current", m_state.diff, allocator); + results.AddMember("shares_good", m_state.accepted, allocator); + results.AddMember("shares_total", m_state.accepted + m_state.rejected, allocator); + results.AddMember("avg_time", m_state.avgTime(), allocator); + results.AddMember("hashes_total", m_state.total, allocator); + + Value best(kArrayType); + for (size_t i = 0; i < m_state.topDiff.size(); ++i) { + best.PushBack(m_state.topDiff[i], allocator); + } + + results.AddMember("best", best, allocator); + + if (version == 1) { + results.AddMember("error_log", Value(kArrayType), allocator); + } + + reply.AddMember("results", results, allocator); +} +#endif diff --git a/src/net/Network.h b/src/net/Network.h index 5a5cbc3ef..ddf6d6f39 100644 --- a/src/net/Network.h +++ b/src/net/Network.h @@ -6,6 +6,7 @@ * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , * Copyright 2018-2019 SChernykh + * Copyright 2019 Howard Chu * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify @@ -27,13 +28,15 @@ #include -#include -#include "api/NetworkState.h" -#include "common/interfaces/IControllerListener.h" -#include "common/interfaces/IStrategyListener.h" +#include "api/interfaces/IApiListener.h" +#include "base/kernel/interfaces/IBaseListener.h" +#include "base/kernel/interfaces/IStrategyListener.h" +#include "base/kernel/interfaces/ITimerListener.h" #include "interfaces/IJobResultListener.h" +#include "net/NetworkState.h" +#include "rapidjson/fwd.h" namespace xmrig { @@ -43,36 +46,48 @@ class Controller; class IStrategy; -class Network : public IJobResultListener, public IStrategyListener, public IControllerListener +class Network : public IJobResultListener, public IStrategyListener, public IBaseListener, public ITimerListener, public IApiListener { public: Network(Controller *controller); ~Network() override; + inline IStrategy *strategy() const { return m_strategy; } + void connect(); - void stop(); protected: - void onActive(IStrategy *strategy, Client *client) override; + inline void onTimer(const Timer *) override { tick(); } + + void onActive(IStrategy *strategy, IClient *client) override; void onConfigChanged(Config *config, Config *previousConfig) override; - void onJob(IStrategy *strategy, Client *client, const Job &job) override; + void onJob(IStrategy *strategy, IClient *client, const Job &job) override; void onJobResult(const JobResult &result) override; + void onLogin(IStrategy *strategy, IClient *client, rapidjson::Document &doc, rapidjson::Value ¶ms) override; void onPause(IStrategy *strategy) override; - void onResultAccepted(IStrategy *strategy, Client *client, const SubmitResult &result, const char *error) override; + void onResultAccepted(IStrategy *strategy, IClient *client, const SubmitResult &result, const char *error) override; + void onVerifyAlgorithm(IStrategy *strategy, const IClient *client, const Algorithm &algorithm, bool *ok) override; + +# ifdef XMRIG_FEATURE_API + void onRequest(IApiRequest &request) override; +# endif private: constexpr static int kTickInterval = 1 * 1000; - bool isColors() const; - void setJob(Client *client, const Job &job, bool donate); + void setJob(IClient *client, const Job &job, bool donate); void tick(); - static void onTick(uv_timer_t *handle); +# ifdef XMRIG_FEATURE_API + void getConnection(rapidjson::Value &reply, rapidjson::Document &doc, int version) const; + void getResults(rapidjson::Value &reply, rapidjson::Document &doc, int version) const; +# endif + Controller *m_controller; IStrategy *m_donate; IStrategy *m_strategy; NetworkState m_state; - uv_timer_t m_timer; + Timer *m_timer; }; diff --git a/src/api/NetworkState.cpp b/src/net/NetworkState.cpp similarity index 67% rename from src/api/NetworkState.cpp rename to src/net/NetworkState.cpp index 0d60cbd2e..6868f57e2 100644 --- a/src/api/NetworkState.cpp +++ b/src/net/NetworkState.cpp @@ -5,7 +5,8 @@ * Copyright 2014-2016 Wolf9466 * Copyright 2016 Jay D Dee * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , + * Copyright 2018-2019 SChernykh + * Copyright 2016-2019 XMRig , * * This program is free software: you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by @@ -28,25 +29,22 @@ #include -#include "api/NetworkState.h" -#include "common/net/SubmitResult.h" +#include "base/kernel/interfaces/IClient.h" +#include "base/net/stratum/Pool.h" +#include "base/net/stratum/SubmitResult.h" +#include "base/tools/Chrono.h" +#include "net/NetworkState.h" xmrig::NetworkState::NetworkState() : - diff(0), + pool(), accepted(0), + diff(0), failures(0), rejected(0), total(0), m_active(false) { - memset(pool, 0, sizeof(pool)); -} - - -int xmrig::NetworkState::connectionTime() const -{ - return m_active ? (int)((uv_now(uv_default_loop()) - m_connectionTime) / 1000) : 0; } @@ -74,6 +72,12 @@ uint32_t xmrig::NetworkState::latency() const } +uint64_t xmrig::NetworkState::connectionTime() const +{ + return m_active ? ((Chrono::steadyMSecs() - m_connectionTime) / 1000) : 0; +} + + void xmrig::NetworkState::add(const SubmitResult &result, const char *error) { if (error) { @@ -90,23 +94,29 @@ void xmrig::NetworkState::add(const SubmitResult &result, const char *error) std::sort(topDiff.rbegin(), topDiff.rend()); } - m_latency.push_back(result.elapsed > 0xFFFF ? 0xFFFF : (uint16_t) result.elapsed); + m_latency.push_back(result.elapsed > 0xFFFF ? 0xFFFF : static_cast(result.elapsed)); } -void xmrig::NetworkState::setPool(const char *host, int port, const char *ip) +void xmrig::NetworkState::onActive(IClient *client) { - snprintf(pool, sizeof(pool) - 1, "%s:%d", host, port); + snprintf(pool, sizeof(pool) - 1, "%s:%d", client->pool().host().data(), client->pool().port()); - m_active = true; - m_connectionTime = uv_now(uv_default_loop()); + m_ip = client->ip(); + m_tls = client->tlsVersion(); + m_fingerprint = client->tlsFingerprint(); + m_active = true; + m_connectionTime = Chrono::steadyMSecs(); } void xmrig::NetworkState::stop() { - m_active = false; - diff = 0; + m_active = false; + diff = 0; + m_ip = nullptr; + m_tls = nullptr; + m_fingerprint = nullptr; failures++; m_latency.clear(); diff --git a/src/api/NetworkState.h b/src/net/NetworkState.h similarity index 82% rename from src/api/NetworkState.h rename to src/net/NetworkState.h index c45761fb2..ce56ec725 100644 --- a/src/api/NetworkState.h +++ b/src/net/NetworkState.h @@ -30,9 +30,13 @@ #include +#include "base/tools/String.h" + + namespace xmrig { +class IClient; class SubmitResult; @@ -41,17 +45,21 @@ class NetworkState public: NetworkState(); - int connectionTime() const; + inline const String &fingerprint() const { return m_fingerprint; } + inline const String &ip() const { return m_ip; } + inline const String &tls() const { return m_tls; } + uint32_t avgTime() const; uint32_t latency() const; + uint64_t connectionTime() const; void add(const SubmitResult &result, const char *error); - void setPool(const char *host, int port, const char *ip); + void onActive(IClient *client); void stop(); char pool[256]; std::array topDiff { { } }; - uint32_t diff; uint64_t accepted; + uint64_t diff; uint64_t failures; uint64_t rejected; uint64_t total; @@ -59,6 +67,9 @@ public: private: bool m_active; std::vector m_latency; + String m_fingerprint; + String m_ip; + String m_tls; uint64_t m_connectionTime; }; diff --git a/src/interfaces/IJobResultListener.h b/src/net/interfaces/IJobResultListener.h similarity index 100% rename from src/interfaces/IJobResultListener.h rename to src/net/interfaces/IJobResultListener.h diff --git a/src/net/strategies/DonateStrategy.cpp b/src/net/strategies/DonateStrategy.cpp index 9593dc9a8..4393cd463 100644 --- a/src/net/strategies/DonateStrategy.cpp +++ b/src/net/strategies/DonateStrategy.cpp @@ -23,46 +23,63 @@ */ -#include "common/crypto/keccak.h" -#include "common/interfaces/IStrategyListener.h" -#include "common/net/Client.h" -#include "common/net/Job.h" -#include "common/net/strategies/FailoverStrategy.h" -#include "common/net/strategies/SinglePoolStrategy.h" -#include "common/Platform.h" -#include "common/xmrig.h" +#include +#include +#include + + +#include "base/kernel/Platform.h" +#include "base/net/stratum/Client.h" +#include "base/net/stratum/Job.h" +#include "base/net/stratum/strategies/FailoverStrategy.h" +#include "base/net/stratum/strategies/SinglePoolStrategy.h" +#include "base/tools/Buffer.h" +#include "base/tools/Timer.h" +#include "core/config/Config.h" +#include "core/Controller.h" +#include "core/Miner.h" +#include "crypto/common/keccak.h" +#include "net/Network.h" #include "net/strategies/DonateStrategy.h" +#include "rapidjson/document.h" -static inline float randomf(float min, float max) { - return (max - min) * ((((float) rand()) / (float) RAND_MAX)) + min; -} +namespace xmrig { + +static inline double randomf(double min, double max) { return (max - min) * (((static_cast(rand())) / static_cast(RAND_MAX))) + min; } +static inline uint64_t random(uint64_t base, double min, double max) { return static_cast(base * randomf(min, max)); } + +static const char *kDonateHost = "donate.v2.xmrig.com"; +#ifdef XMRIG_FEATURE_TLS +static const char *kDonateHostTls = "donate.ssl.xmrig.com"; +#endif + +} /* namespace xmrig */ -xmrig::DonateStrategy::DonateStrategy(int level, const char *user, Algo algo, IStrategyListener *listener) : - m_active(false), - m_donateTime(level * 60 * 1000), - m_idleTime((100 - level) * 60 * 1000), +xmrig::DonateStrategy::DonateStrategy(Controller *controller, IStrategyListener *listener) : + m_tls(false), + m_userId(), + m_donateTime(static_cast(controller->config()->pools().donateLevel()) * 60 * 1000), + m_idleTime((100 - static_cast(controller->config()->pools().donateLevel())) * 60 * 1000), + m_controller(controller), + m_proxy(nullptr), m_strategy(nullptr), m_listener(listener), + m_state(STATE_NEW), m_now(0), - m_stop(0) + m_timestamp(0) { uint8_t hash[200]; - char userId[65] = { 0 }; - keccak(reinterpret_cast(user), strlen(user), hash); - Job::toHex(hash, 32, userId); + const String &user = controller->config()->pools().data().front().user(); + keccak(reinterpret_cast(user.data()), user.size(), hash); + Buffer::toHex(hash, 32, m_userId); -# ifndef XMRIG_NO_TLS - m_pools.push_back(Pool("donate.ssl.xmrig.com", 443, userId, nullptr, false, true, true)); +# ifdef XMRIG_FEATURE_TLS + m_pools.push_back(Pool(kDonateHostTls, 443, m_userId, nullptr, 0, true, true)); # endif - - m_pools.push_back(Pool("donate.v2.xmrig.com", 3333, userId, nullptr, false, true)); - - for (Pool &pool : m_pools) { - pool.adjust(Algorithm(algo, VARIANT_AUTO)); - } + m_pools.push_back(Pool(kDonateHost, 3333, m_userId, nullptr, 0, true)); if (m_pools.size() > 1) { m_strategy = new FailoverStrategy(m_pools, 1, 2, this, true); @@ -71,40 +88,55 @@ xmrig::DonateStrategy::DonateStrategy(int level, const char *user, Algo algo, IS m_strategy = new SinglePoolStrategy(m_pools.front(), 1, 2, this, true); } - m_timer.data = this; - uv_timer_init(uv_default_loop(), &m_timer); + m_timer = new Timer(this); - idle(m_idleTime * randomf(0.5, 1.5)); + setState(STATE_IDLE); } xmrig::DonateStrategy::~DonateStrategy() { + delete m_timer; delete m_strategy; + + if (m_proxy) { + m_proxy->deleteLater(); + } } int64_t xmrig::DonateStrategy::submit(const JobResult &result) { - return m_strategy->submit(result); + return m_proxy ? m_proxy->submit(result) : m_strategy->submit(result); } void xmrig::DonateStrategy::connect() { - m_strategy->connect(); + m_proxy = createProxy(); + if (m_proxy) { + m_proxy->connect(); + } + else if (m_controller->config()->pools().proxyDonate() == Pools::PROXY_DONATE_ALWAYS) { + setState(STATE_IDLE); + } + else { + m_strategy->connect(); + } } void xmrig::DonateStrategy::setAlgo(const xmrig::Algorithm &algo) { + m_algorithm = algo; + m_strategy->setAlgo(algo); } void xmrig::DonateStrategy::stop() { - uv_timer_stop(&m_timer); + m_timer->stop(); m_strategy->stop(); } @@ -115,25 +147,141 @@ void xmrig::DonateStrategy::tick(uint64_t now) m_strategy->tick(now); - if (m_stop && now > m_stop) { - m_strategy->stop(); - m_stop = 0; + if (m_proxy) { + m_proxy->tick(now); + } + + if (state() == STATE_WAIT && now > m_timestamp) { + setState(STATE_IDLE); } } -void xmrig::DonateStrategy::onActive(IStrategy *strategy, Client *client) +void xmrig::DonateStrategy::onActive(IStrategy *, IClient *client) { - if (!isActive()) { - uv_timer_start(&m_timer, DonateStrategy::onTimer, m_donateTime, 0); + if (isActive()) { + return; } - m_active = true; + setState(STATE_ACTIVE); m_listener->onActive(this, client); } -void xmrig::DonateStrategy::onJob(IStrategy *strategy, Client *client, const Job &job) +void xmrig::DonateStrategy::onPause(IStrategy *) +{ +} + + +void xmrig::DonateStrategy::onClose(IClient *, int failures) +{ + if (failures == 2 && m_controller->config()->pools().proxyDonate() == Pools::PROXY_DONATE_AUTO) { + m_proxy->deleteLater(); + m_proxy = nullptr; + + m_strategy->connect(); + } +} + + +void xmrig::DonateStrategy::onLogin(IClient *, rapidjson::Document &doc, rapidjson::Value ¶ms) +{ + using namespace rapidjson; + auto &allocator = doc.GetAllocator(); + +# ifdef XMRIG_FEATURE_TLS + if (m_tls) { + char buf[40] = { 0 }; + snprintf(buf, sizeof(buf), "stratum+ssl://%s", m_pools[0].url().data()); + params.AddMember("url", Value(buf, allocator), allocator); + } + else { + params.AddMember("url", m_pools[1].url().toJSON(), allocator); + } +# else + params.AddMember("url", m_pools[0].url().toJSON(), allocator); +# endif + + setAlgorithms(doc, params); +} + + +void xmrig::DonateStrategy::onLogin(IStrategy *, IClient *, rapidjson::Document &doc, rapidjson::Value ¶ms) +{ + setAlgorithms(doc, params); +} + + +void xmrig::DonateStrategy::onLoginSuccess(IClient *client) +{ + if (isActive()) { + return; + } + + setState(STATE_ACTIVE); + m_listener->onActive(this, client); +} + + +void xmrig::DonateStrategy::onTimer(const Timer *) +{ + setState(isActive() ? STATE_WAIT : STATE_CONNECT); +} + + +xmrig::Client *xmrig::DonateStrategy::createProxy() +{ + if (m_controller->config()->pools().proxyDonate() == Pools::PROXY_DONATE_NONE) { + return nullptr; + } + + IStrategy *strategy = m_controller->network()->strategy(); + if (!strategy->isActive() || !strategy->client()->hasExtension(IClient::EXT_CONNECT)) { + return nullptr; + } + + const IClient *client = strategy->client(); + m_tls = client->hasExtension(IClient::EXT_TLS); + + Pool pool(client->ip(), client->pool().port(), m_userId, client->pool().password(), 0, true, client->isTLS()); + pool.setAlgo(client->pool().algorithm()); + + Client *proxy = new Client(-1, Platform::userAgent(), this); + proxy->setPool(pool); + proxy->setQuiet(true); + + return proxy; +} + + +void xmrig::DonateStrategy::idle(double min, double max) +{ + m_timer->start(random(m_idleTime, min, max), 0); +} + + +void xmrig::DonateStrategy::setAlgorithms(rapidjson::Document &doc, rapidjson::Value ¶ms) +{ + using namespace rapidjson; + auto &allocator = doc.GetAllocator(); + + Algorithms algorithms = m_controller->miner()->algorithms(); + const size_t index = static_cast(std::distance(algorithms.begin(), std::find(algorithms.begin(), algorithms.end(), m_algorithm))); + if (index > 0 && index < algorithms.size()) { + std::swap(algorithms[0], algorithms[index]); + } + + Value algo(kArrayType); + + for (const auto &a : algorithms) { + algo.PushBack(StringRef(a.shortName()), allocator); + } + + params.AddMember("algo", algo, allocator); +} + + +void xmrig::DonateStrategy::setJob(IClient *client, const Job &job) { if (isActive()) { m_listener->onJob(this, client, job); @@ -141,45 +289,57 @@ void xmrig::DonateStrategy::onJob(IStrategy *strategy, Client *client, const Job } -void xmrig::DonateStrategy::onPause(IStrategy *strategy) -{ -} - - -void xmrig::DonateStrategy::onResultAccepted(IStrategy *strategy, Client *client, const SubmitResult &result, const char *error) +void xmrig::DonateStrategy::setResult(IClient *client, const SubmitResult &result, const char *error) { m_listener->onResultAccepted(this, client, result, error); } -void xmrig::DonateStrategy::idle(uint64_t timeout) +void xmrig::DonateStrategy::setState(State state) { - uv_timer_start(&m_timer, DonateStrategy::onTimer, timeout, 0); -} + constexpr const uint64_t waitTime = 3000; - -void xmrig::DonateStrategy::suspend() -{ -# if defined(XMRIG_AMD_PROJECT) || defined(XMRIG_NVIDIA_PROJECT) - m_stop = m_now + 5000; -# else - m_stop = m_now + 500; -# endif - - m_active = false; - m_listener->onPause(this); - - idle(m_idleTime); -} - - -void xmrig::DonateStrategy::onTimer(uv_timer_t *handle) -{ - auto strategy = static_cast(handle->data); - - if (!strategy->isActive()) { - return strategy->connect(); + assert(m_state != state && state != STATE_NEW); + if (m_state == state) { + return; } - strategy->suspend(); + const State prev = m_state; + m_state = state; + + switch (state) { + case STATE_NEW: + break; + + case STATE_IDLE: + if (prev == STATE_NEW) { + idle(0.5, 1.5); + } + else if (prev == STATE_CONNECT) { + m_timer->start(20000, 0); + } + else { + m_strategy->stop(); + if (m_proxy) { + m_proxy->deleteLater(); + m_proxy = nullptr; + } + + idle(0.8, 1.2); + } + break; + + case STATE_CONNECT: + connect(); + break; + + case STATE_ACTIVE: + m_timer->start(m_donateTime, 0); + break; + + case STATE_WAIT: + m_timestamp = m_now + waitTime; + m_listener->onPause(this); + break; + } } diff --git a/src/net/strategies/DonateStrategy.h b/src/net/strategies/DonateStrategy.h index 76702ef32..134127bf6 100644 --- a/src/net/strategies/DonateStrategy.h +++ b/src/net/strategies/DonateStrategy.h @@ -26,32 +26,40 @@ #define XMRIG_DONATESTRATEGY_H -#include #include -#include "base/net/Pool.h" -#include "common/interfaces/IClientListener.h" -#include "common/interfaces/IStrategy.h" -#include "common/interfaces/IStrategyListener.h" +#include "base/kernel/interfaces/IClientListener.h" +#include "base/kernel/interfaces/IStrategy.h" +#include "base/kernel/interfaces/IStrategyListener.h" +#include "base/kernel/interfaces/ITimerListener.h" +#include "base/net/stratum/Pool.h" namespace xmrig { class Client; +class Controller; class IStrategyListener; -class DonateStrategy : public IStrategy, public IStrategyListener +class DonateStrategy : public IStrategy, public IStrategyListener, public ITimerListener, public IClientListener { public: - DonateStrategy(int level, const char *user, Algo algo, IStrategyListener *listener); + DonateStrategy(Controller *controller, IStrategyListener *listener); ~DonateStrategy() override; -public: - inline bool isActive() const override { return m_active; } - inline void resume() override {} +protected: + inline bool isActive() const override { return state() == STATE_ACTIVE; } + inline IClient *client() const override { return m_proxy ? m_proxy : m_strategy->client(); } + inline void onJob(IStrategy *, IClient *client, const Job &job) override { setJob(client, job); } + inline void onJobReceived(IClient *client, const Job &job, const rapidjson::Value &) override { setJob(client, job); } + inline void onResultAccepted(IClient *client, const SubmitResult &result, const char *error) override { setResult(client, result, error); } + inline void onResultAccepted(IStrategy *, IClient *client, const SubmitResult &result, const char *error) override { setResult(client, result, error); } + inline void onVerifyAlgorithm(const IClient *, const Algorithm &, bool *) override {} + inline void onVerifyAlgorithm(IStrategy *, const IClient *, const Algorithm &, bool *) override {} + inline void resume() override {} int64_t submit(const JobResult &result) override; void connect() override; @@ -59,27 +67,48 @@ public: void stop() override; void tick(uint64_t now) override; -protected: - void onActive(IStrategy *strategy, Client *client) override; - void onJob(IStrategy *strategy, Client *client, const Job &job) override; + void onActive(IStrategy *strategy, IClient *client) override; void onPause(IStrategy *strategy) override; - void onResultAccepted(IStrategy *strategy, Client *client, const SubmitResult &result, const char *error) override; + + void onClose(IClient *client, int failures) override; + void onLogin(IClient *client, rapidjson::Document &doc, rapidjson::Value ¶ms) override; + void onLogin(IStrategy *strategy, IClient *client, rapidjson::Document &doc, rapidjson::Value ¶ms) override; + void onLoginSuccess(IClient *client) override; + + void onTimer(const Timer *timer) override; private: - void idle(uint64_t timeout); - void suspend(); + enum State { + STATE_NEW, + STATE_IDLE, + STATE_CONNECT, + STATE_ACTIVE, + STATE_WAIT + }; - static void onTimer(uv_timer_t *handle); + inline State state() const { return m_state; } - bool m_active; + Client *createProxy(); + void idle(double min, double max); + void setAlgorithms(rapidjson::Document &doc, rapidjson::Value ¶ms); + void setJob(IClient *client, const Job &job); + void setResult(IClient *client, const SubmitResult &result, const char *error); + void setState(State state); + + Algorithm m_algorithm; + bool m_tls; + char m_userId[65]; const uint64_t m_donateTime; const uint64_t m_idleTime; + Controller *m_controller; + IClient *m_proxy; IStrategy *m_strategy; IStrategyListener *m_listener; + State m_state; std::vector m_pools; + Timer *m_timer; uint64_t m_now; - uint64_t m_stop; - uv_timer_t m_timer; + uint64_t m_timestamp; }; diff --git a/src/version.h b/src/version.h index 7ba9c7767..a0018a790 100644 --- a/src/version.h +++ b/src/version.h @@ -28,15 +28,15 @@ #define APP_ID "xmrig" #define APP_NAME "XMRig" #define APP_DESC "XMRig CPU miner" -#define APP_VERSION "2.14.5-dev" +#define APP_VERSION "2.99.6-beta" #define APP_DOMAIN "xmrig.com" #define APP_SITE "www.xmrig.com" #define APP_COPYRIGHT "Copyright (C) 2016-2019 xmrig.com" #define APP_KIND "cpu" #define APP_VER_MAJOR 2 -#define APP_VER_MINOR 14 -#define APP_VER_PATCH 5 +#define APP_VER_MINOR 99 +#define APP_VER_PATCH 6 #ifdef _MSC_VER # if (_MSC_VER >= 1920) diff --git a/src/workers/CpuThread.cpp b/src/workers/CpuThread.cpp deleted file mode 100644 index 6548b4617..000000000 --- a/src/workers/CpuThread.cpp +++ /dev/null @@ -1,744 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#include - - -#include "common/cpu/Cpu.h" -#include "common/log/Log.h" -#include "crypto/Asm.h" -#include "Mem.h" -#include "rapidjson/document.h" -#include "workers/CpuThread.h" - - -#if defined(XMRIG_ARM) -# include "crypto/CryptoNight_arm.h" -#else -# include "crypto/CryptoNight_x86.h" -#endif - - -xmrig::CpuThread::CpuThread(size_t index, Algo algorithm, AlgoVariant av, Multiway multiway, int64_t affinity, int priority, bool softAES, bool prefetch, Assembly assembly) : - m_algorithm(algorithm), - m_av(av), - m_assembly(assembly), - m_prefetch(prefetch), - m_softAES(softAES), - m_priority(priority), - m_affinity(affinity), - m_multiway(multiway), - m_index(index) -{ -} - - -#ifndef XMRIG_NO_ASM -template -static void patchCode(T dst, U src, const uint32_t iterations, const uint32_t mask) -{ - const uint8_t* p = reinterpret_cast(src); - - // Workaround for Visual Studio placing trampoline in debug builds. -# if defined(_MSC_VER) - if (p[0] == 0xE9) { - p += *(int32_t*)(p + 1) + 5; - } -# endif - - size_t size = 0; - while (*(uint32_t*)(p + size) != 0xDEADC0DE) { - ++size; - } - size += sizeof(uint32_t); - - memcpy((void*) dst, (const void*) src, size); - - uint8_t* patched_data = reinterpret_cast(dst); - for (size_t i = 0; i + sizeof(uint32_t) <= size; ++i) { - switch (*(uint32_t*)(patched_data + i)) { - case xmrig::CRYPTONIGHT_ITER: - *(uint32_t*)(patched_data + i) = iterations; - break; - - case xmrig::CRYPTONIGHT_MASK: - *(uint32_t*)(patched_data + i) = mask; - break; - } - } -} - - -extern "C" void cnv2_mainloop_ivybridge_asm(cryptonight_ctx **ctx); -extern "C" void cnv2_mainloop_ryzen_asm(cryptonight_ctx **ctx); -extern "C" void cnv2_mainloop_bulldozer_asm(cryptonight_ctx **ctx); -extern "C" void cnv2_double_mainloop_sandybridge_asm(cryptonight_ctx **ctx); - - -xmrig::CpuThread::cn_mainloop_fun cn_half_mainloop_ivybridge_asm = nullptr; -xmrig::CpuThread::cn_mainloop_fun cn_half_mainloop_ryzen_asm = nullptr; -xmrig::CpuThread::cn_mainloop_fun cn_half_mainloop_bulldozer_asm = nullptr; -xmrig::CpuThread::cn_mainloop_fun cn_half_double_mainloop_sandybridge_asm = nullptr; - -xmrig::CpuThread::cn_mainloop_fun cn_trtl_mainloop_ivybridge_asm = nullptr; -xmrig::CpuThread::cn_mainloop_fun cn_trtl_mainloop_ryzen_asm = nullptr; -xmrig::CpuThread::cn_mainloop_fun cn_trtl_mainloop_bulldozer_asm = nullptr; -xmrig::CpuThread::cn_mainloop_fun cn_trtl_double_mainloop_sandybridge_asm = nullptr; - -xmrig::CpuThread::cn_mainloop_fun cn_zls_mainloop_ivybridge_asm = nullptr; -xmrig::CpuThread::cn_mainloop_fun cn_zls_mainloop_ryzen_asm = nullptr; -xmrig::CpuThread::cn_mainloop_fun cn_zls_mainloop_bulldozer_asm = nullptr; -xmrig::CpuThread::cn_mainloop_fun cn_zls_double_mainloop_sandybridge_asm = nullptr; - -xmrig::CpuThread::cn_mainloop_fun cn_double_mainloop_ivybridge_asm = nullptr; -xmrig::CpuThread::cn_mainloop_fun cn_double_mainloop_ryzen_asm = nullptr; -xmrig::CpuThread::cn_mainloop_fun cn_double_mainloop_bulldozer_asm = nullptr; -xmrig::CpuThread::cn_mainloop_fun cn_double_double_mainloop_sandybridge_asm = nullptr; - - -void xmrig::CpuThread::patchAsmVariants() -{ - const int allocation_size = 65536; - uint8_t *base = static_cast(Mem::allocateExecutableMemory(allocation_size)); - - cn_half_mainloop_ivybridge_asm = reinterpret_cast (base + 0x0000); - cn_half_mainloop_ryzen_asm = reinterpret_cast (base + 0x1000); - cn_half_mainloop_bulldozer_asm = reinterpret_cast (base + 0x2000); - cn_half_double_mainloop_sandybridge_asm = reinterpret_cast (base + 0x3000); - - cn_trtl_mainloop_ivybridge_asm = reinterpret_cast (base + 0x4000); - cn_trtl_mainloop_ryzen_asm = reinterpret_cast (base + 0x5000); - cn_trtl_mainloop_bulldozer_asm = reinterpret_cast (base + 0x6000); - cn_trtl_double_mainloop_sandybridge_asm = reinterpret_cast (base + 0x7000); - - cn_zls_mainloop_ivybridge_asm = reinterpret_cast (base + 0x8000); - cn_zls_mainloop_ryzen_asm = reinterpret_cast (base + 0x9000); - cn_zls_mainloop_bulldozer_asm = reinterpret_cast (base + 0xA000); - cn_zls_double_mainloop_sandybridge_asm = reinterpret_cast (base + 0xB000); - - cn_double_mainloop_ivybridge_asm = reinterpret_cast (base + 0xC000); - cn_double_mainloop_ryzen_asm = reinterpret_cast (base + 0xD000); - cn_double_mainloop_bulldozer_asm = reinterpret_cast (base + 0xE000); - cn_double_double_mainloop_sandybridge_asm = reinterpret_cast (base + 0xF000); - - patchCode(cn_half_mainloop_ivybridge_asm, cnv2_mainloop_ivybridge_asm, xmrig::CRYPTONIGHT_HALF_ITER, xmrig::CRYPTONIGHT_MASK); - patchCode(cn_half_mainloop_ryzen_asm, cnv2_mainloop_ryzen_asm, xmrig::CRYPTONIGHT_HALF_ITER, xmrig::CRYPTONIGHT_MASK); - patchCode(cn_half_mainloop_bulldozer_asm, cnv2_mainloop_bulldozer_asm, xmrig::CRYPTONIGHT_HALF_ITER, xmrig::CRYPTONIGHT_MASK); - patchCode(cn_half_double_mainloop_sandybridge_asm, cnv2_double_mainloop_sandybridge_asm, xmrig::CRYPTONIGHT_HALF_ITER, xmrig::CRYPTONIGHT_MASK); - - patchCode(cn_trtl_mainloop_ivybridge_asm, cnv2_mainloop_ivybridge_asm, xmrig::CRYPTONIGHT_TRTL_ITER, xmrig::CRYPTONIGHT_PICO_MASK); - patchCode(cn_trtl_mainloop_ryzen_asm, cnv2_mainloop_ryzen_asm, xmrig::CRYPTONIGHT_TRTL_ITER, xmrig::CRYPTONIGHT_PICO_MASK); - patchCode(cn_trtl_mainloop_bulldozer_asm, cnv2_mainloop_bulldozer_asm, xmrig::CRYPTONIGHT_TRTL_ITER, xmrig::CRYPTONIGHT_PICO_MASK); - patchCode(cn_trtl_double_mainloop_sandybridge_asm, cnv2_double_mainloop_sandybridge_asm, xmrig::CRYPTONIGHT_TRTL_ITER, xmrig::CRYPTONIGHT_PICO_MASK); - - patchCode(cn_zls_mainloop_ivybridge_asm, cnv2_mainloop_ivybridge_asm, xmrig::CRYPTONIGHT_ZLS_ITER, xmrig::CRYPTONIGHT_MASK); - patchCode(cn_zls_mainloop_ryzen_asm, cnv2_mainloop_ryzen_asm, xmrig::CRYPTONIGHT_ZLS_ITER, xmrig::CRYPTONIGHT_MASK); - patchCode(cn_zls_mainloop_bulldozer_asm, cnv2_mainloop_bulldozer_asm, xmrig::CRYPTONIGHT_ZLS_ITER, xmrig::CRYPTONIGHT_MASK); - patchCode(cn_zls_double_mainloop_sandybridge_asm, cnv2_double_mainloop_sandybridge_asm, xmrig::CRYPTONIGHT_ZLS_ITER, xmrig::CRYPTONIGHT_MASK); - - patchCode(cn_double_mainloop_ivybridge_asm, cnv2_mainloop_ivybridge_asm, xmrig::CRYPTONIGHT_DOUBLE_ITER, xmrig::CRYPTONIGHT_MASK); - patchCode(cn_double_mainloop_ryzen_asm, cnv2_mainloop_ryzen_asm, xmrig::CRYPTONIGHT_DOUBLE_ITER, xmrig::CRYPTONIGHT_MASK); - patchCode(cn_double_mainloop_bulldozer_asm, cnv2_mainloop_bulldozer_asm, xmrig::CRYPTONIGHT_DOUBLE_ITER, xmrig::CRYPTONIGHT_MASK); - patchCode(cn_double_double_mainloop_sandybridge_asm, cnv2_double_mainloop_sandybridge_asm, xmrig::CRYPTONIGHT_DOUBLE_ITER, xmrig::CRYPTONIGHT_MASK); - - Mem::protectExecutableMemory(base, allocation_size); - Mem::flushInstructionCache(base, allocation_size); -} -#endif - - -bool xmrig::CpuThread::isSoftAES(AlgoVariant av) -{ - return av == AV_SINGLE_SOFT || av == AV_DOUBLE_SOFT || av > AV_PENTA; -} - - -#ifndef XMRIG_NO_ASM -template -static inline void add_asm_func(xmrig::CpuThread::cn_hash_fun(&asm_func_map)[xmrig::ALGO_MAX][xmrig::AV_MAX][xmrig::VARIANT_MAX][xmrig::ASM_MAX]) -{ - asm_func_map[algo][xmrig::AV_SINGLE][variant][xmrig::ASM_INTEL] = cryptonight_single_hash_asm; - asm_func_map[algo][xmrig::AV_SINGLE][variant][xmrig::ASM_RYZEN] = cryptonight_single_hash_asm; - asm_func_map[algo][xmrig::AV_SINGLE][variant][xmrig::ASM_BULLDOZER] = cryptonight_single_hash_asm; - - asm_func_map[algo][xmrig::AV_DOUBLE][variant][xmrig::ASM_INTEL] = cryptonight_double_hash_asm; - asm_func_map[algo][xmrig::AV_DOUBLE][variant][xmrig::ASM_RYZEN] = cryptonight_double_hash_asm; - asm_func_map[algo][xmrig::AV_DOUBLE][variant][xmrig::ASM_BULLDOZER] = cryptonight_double_hash_asm; -} -#endif - -xmrig::CpuThread::cn_hash_fun xmrig::CpuThread::fn(Algo algorithm, AlgoVariant av, Variant variant, Assembly assembly) -{ - assert(variant >= VARIANT_0 && variant < VARIANT_MAX); - -# ifndef XMRIG_NO_ASM - if (assembly == ASM_AUTO) { - assembly = Cpu::info()->assembly(); - } - - static cn_hash_fun asm_func_map[ALGO_MAX][AV_MAX][VARIANT_MAX][ASM_MAX] = {}; - static bool asm_func_map_initialized = false; - - if (!asm_func_map_initialized) { - add_asm_func(asm_func_map); - add_asm_func(asm_func_map); - add_asm_func(asm_func_map); - add_asm_func(asm_func_map); - -# ifndef XMRIG_NO_CN_PICO - add_asm_func(asm_func_map); -# endif - - add_asm_func(asm_func_map); - add_asm_func(asm_func_map); - add_asm_func(asm_func_map); - - asm_func_map_initialized = true; - } - - cn_hash_fun fun = asm_func_map[algorithm][av][variant][assembly]; - if (fun) { - return fun; - } -# endif - - constexpr const size_t count = VARIANT_MAX * 10 * ALGO_MAX; - - static const cn_hash_fun func_table[] = { - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_TUBE - - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_XHV - - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_TRTL - -# ifndef XMRIG_NO_CN_GPU - cryptonight_single_hash_gpu, - nullptr, - cryptonight_single_hash_gpu, - nullptr, - nullptr, - nullptr, - nullptr, - nullptr, - nullptr, - nullptr, -# else - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_GPU -# endif - - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - -# ifndef XMRIG_NO_AEON - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_TUBE - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_XTL - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_MSR - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_XHV - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_XAO - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_RTO - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_2 - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_HALF - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_TRTL - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_GPU - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_WOW - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_4 - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_RWZ - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_ZLS - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_DOUBLE -# else - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_0 - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_1 - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_TUBE - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_XTL - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_MSR - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_XHV - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_XAO - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_RTO - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_2 - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_HALF - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_TRTL - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_GPU - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_WOW - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_4 - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_RWZ - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_ZLS - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_DOUBLE -# endif - -# ifndef XMRIG_NO_SUMO - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_1 - - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_XTL - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_MSR - - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_XAO - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_RTO - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_2 - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_HALF - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_TRTL - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_GPU - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_WOW - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_4 - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_RWZ - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_ZLS - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_DOUBLE -# else - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_0 - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_1 - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_TUBE - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_XTL - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_MSR - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_XHV - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_XAO - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_RTO - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_2 - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_HALF - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_TRTL - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_GPU - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_WOW - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_4 - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_RWZ - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_ZLS - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_DOUBLE -# endif - -# ifndef XMRIG_NO_CN_PICO - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_0 - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_1 - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_TUBE - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_XTL - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_MSR - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_XHV - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_XAO - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_RTO - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_2 - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_HALF - - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_single_hash, - cryptonight_double_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - cryptonight_triple_hash, - cryptonight_quad_hash, - cryptonight_penta_hash, - - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_GPU - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_WOW - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_4 - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_RWZ - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_ZLS - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_DOUBLE -# else - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_0 - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_1 - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_TUBE - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_XTL - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_MSR - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_XHV - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_XAO - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_RTO - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_2 - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_HALF - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_TRTL - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_GPU - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_WOW - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_4 - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_RWZ - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_ZLS - nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, nullptr, // VARIANT_DOUBLE -# endif - }; - - static_assert(count == sizeof(func_table) / sizeof(func_table[0]), "func_table size mismatch"); - - const size_t index = VARIANT_MAX * 10 * algorithm + 10 * variant + av - 1; - -# ifndef NDEBUG - cn_hash_fun func = func_table[index]; - - assert(index < sizeof(func_table) / sizeof(func_table[0])); - assert(func != nullptr); - - return func; -# else - return func_table[index]; -# endif -} - - -xmrig::CpuThread *xmrig::CpuThread::createFromAV(size_t index, Algo algorithm, AlgoVariant av, int64_t affinity, int priority, Assembly assembly) -{ - assert(av > AV_AUTO && av < AV_MAX); - - int64_t cpuId = -1L; - - if (affinity != -1L) { - size_t idx = 0; - - for (size_t i = 0; i < 64; i++) { - if (!(affinity & (1ULL << i))) { - continue; - } - - if (idx == index) { - cpuId = i; - break; - } - - idx++; - } - } - - return new CpuThread(index, algorithm, av, multiway(av), cpuId, priority, isSoftAES(av), false, assembly); -} - - -xmrig::CpuThread *xmrig::CpuThread::createFromData(size_t index, Algo algorithm, const CpuThread::Data &data, int priority, bool softAES) -{ - int av = AV_AUTO; - const Multiway multiway = data.multiway; - - if (multiway <= DoubleWay) { - av = softAES ? (multiway + 2) : multiway; - } - else { - av = softAES ? (multiway + 5) : (multiway + 2); - } - - assert(av > AV_AUTO && av < AV_MAX); - - return new CpuThread(index, algorithm, static_cast(av), multiway, data.affinity, priority, softAES, false, data.assembly); -} - - -xmrig::CpuThread::Data xmrig::CpuThread::parse(const rapidjson::Value &object) -{ - Data data; - - const auto &multiway = object["low_power_mode"]; - if (multiway.IsBool()) { - data.multiway = multiway.IsTrue() ? DoubleWay : SingleWay; - data.valid = true; - } - else if (multiway.IsUint()) { - data.setMultiway(multiway.GetInt()); - } - - if (!data.valid) { - return data; - } - - const auto &affinity = object["affine_to_cpu"]; - if (affinity.IsUint64()) { - data.affinity = affinity.GetInt64(); - } - -# ifndef XMRIG_NO_ASM - data.assembly = Asm::parse(object["asm"]); -# endif - - return data; -} - - -xmrig::IThread::Multiway xmrig::CpuThread::multiway(AlgoVariant av) -{ - switch (av) { - case AV_SINGLE: - case AV_SINGLE_SOFT: - return SingleWay; - - case AV_DOUBLE_SOFT: - case AV_DOUBLE: - return DoubleWay; - - case AV_TRIPLE_SOFT: - case AV_TRIPLE: - return TripleWay; - - case AV_QUAD_SOFT: - case AV_QUAD: - return QuadWay; - - case AV_PENTA_SOFT: - case AV_PENTA: - return PentaWay; - - default: - break; - } - - return SingleWay; -} - - -#ifdef APP_DEBUG -void xmrig::CpuThread::print() const -{ - LOG_DEBUG(GREEN_BOLD("CPU thread: ") " index " WHITE_BOLD("%zu") ", multiway " WHITE_BOLD("%d") ", av " WHITE_BOLD("%d") ",", - index(), static_cast(multiway()), static_cast(m_av)); - -# ifndef XMRIG_NO_ASM - LOG_DEBUG(" assembly: %s, affine_to_cpu: %" PRId64, Asm::toString(m_assembly), affinity()); -# else - LOG_DEBUG(" affine_to_cpu: %" PRId64, affinity()); -# endif -} -#endif - - -#ifndef XMRIG_NO_API -rapidjson::Value xmrig::CpuThread::toAPI(rapidjson::Document &doc) const -{ - using namespace rapidjson; - - Value obj(kObjectType); - auto &allocator = doc.GetAllocator(); - - obj.AddMember("type", "cpu", allocator); - obj.AddMember("av", m_av, allocator); - obj.AddMember("low_power_mode", multiway(), allocator); - obj.AddMember("affine_to_cpu", affinity(), allocator); - obj.AddMember("priority", priority(), allocator); - obj.AddMember("soft_aes", isSoftAES(), allocator); - - return obj; -} -#endif - - -rapidjson::Value xmrig::CpuThread::toConfig(rapidjson::Document &doc) const -{ - using namespace rapidjson; - - Value obj(kObjectType); - auto &allocator = doc.GetAllocator(); - - obj.AddMember("low_power_mode", multiway(), allocator); - obj.AddMember("affine_to_cpu", affinity() == -1L ? Value(kFalseType) : Value(affinity()), allocator); - -# ifndef XMRIG_NO_ASM - obj.AddMember("asm", Asm::toJSON(m_assembly), allocator); -# endif - - return obj; -} diff --git a/src/workers/CpuThread.h b/src/workers/CpuThread.h deleted file mode 100644 index 05d4a0661..000000000 --- a/src/workers/CpuThread.h +++ /dev/null @@ -1,115 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef XMRIG_CPUTHREAD_H -#define XMRIG_CPUTHREAD_H - - -#include "common/xmrig.h" -#include "interfaces/IThread.h" - - -struct cryptonight_ctx; - - -namespace xmrig { - - -class CpuThread : public IThread -{ -public: - struct Data - { - inline Data() : assembly(ASM_AUTO), valid(false), affinity(-1L), multiway(SingleWay) {} - - inline void setMultiway(int value) - { - if (value >= SingleWay && value <= PentaWay) { - multiway = static_cast(value); - valid = true; - } - } - - Assembly assembly; - bool valid; - int64_t affinity; - Multiway multiway; - }; - - - CpuThread(size_t index, Algo algorithm, AlgoVariant av, Multiway multiway, int64_t affinity, int priority, bool softAES, bool prefetch, Assembly assembly); - - typedef void (*cn_hash_fun)(const uint8_t *input, size_t size, uint8_t *output, cryptonight_ctx **ctx, uint64_t height); - typedef void (*cn_mainloop_fun)(cryptonight_ctx **ctx); - -# ifndef XMRIG_NO_ASM - static void patchAsmVariants(); -# endif - - static bool isSoftAES(AlgoVariant av); - static cn_hash_fun fn(Algo algorithm, AlgoVariant av, Variant variant, Assembly assembly); - static CpuThread *createFromAV(size_t index, Algo algorithm, AlgoVariant av, int64_t affinity, int priority, Assembly assembly); - static CpuThread *createFromData(size_t index, Algo algorithm, const CpuThread::Data &data, int priority, bool softAES); - static Data parse(const rapidjson::Value &object); - static Multiway multiway(AlgoVariant av); - - inline bool isPrefetch() const { return m_prefetch; } - inline bool isSoftAES() const { return m_softAES; } - inline cn_hash_fun fn(Variant variant) const { return fn(m_algorithm, m_av, variant, m_assembly); } - - inline Algo algorithm() const override { return m_algorithm; } - inline int priority() const override { return m_priority; } - inline int64_t affinity() const override { return m_affinity; } - inline Multiway multiway() const override { return m_multiway; } - inline size_t index() const override { return m_index; } - inline Type type() const override { return CPU; } - -protected: -# ifdef APP_DEBUG - void print() const override; -# endif - -# ifndef XMRIG_NO_API - rapidjson::Value toAPI(rapidjson::Document &doc) const override; -# endif - - rapidjson::Value toConfig(rapidjson::Document &doc) const override; - -private: - const Algo m_algorithm; - const AlgoVariant m_av; - const Assembly m_assembly; - const bool m_prefetch; - const bool m_softAES; - const int m_priority; - const int64_t m_affinity; - const Multiway m_multiway; - const size_t m_index; -}; - - -} /* namespace xmrig */ - - -#endif /* XMRIG_CPUTHREAD_H */ diff --git a/src/workers/Handle.cpp b/src/workers/Handle.cpp deleted file mode 100644 index d42ea3689..000000000 --- a/src/workers/Handle.cpp +++ /dev/null @@ -1,46 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - - -#include "workers/Handle.h" - - -Handle::Handle(xmrig::IThread *config, uint32_t offset, size_t totalWays) : - m_worker(nullptr), - m_totalWays(totalWays), - m_offset(offset), - m_config(config) -{ -} - - -void Handle::join() -{ - uv_thread_join(&m_thread); -} - - -void Handle::start(void (*callback) (void *)) -{ - uv_thread_create(&m_thread, callback, this); -} diff --git a/src/workers/Handle.h b/src/workers/Handle.h deleted file mode 100644 index 4bb899f9f..000000000 --- a/src/workers/Handle.h +++ /dev/null @@ -1,62 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef __HANDLE_H__ -#define __HANDLE_H__ - - -#include -#include -#include - - -#include "interfaces/IThread.h" - - -class IWorker; - - -class Handle -{ -public: - Handle(xmrig::IThread *config, uint32_t offset, size_t totalWays); - void join(); - void start(void (*callback) (void *)); - - inline IWorker *worker() const { return m_worker; } - inline size_t threadId() const { return m_config->index(); } - inline size_t totalWays() const { return m_totalWays; } - inline uint32_t offset() const { return m_offset; } - inline void setWorker(IWorker *worker) { assert(worker != nullptr); m_worker = worker; } - inline xmrig::IThread *config() const { return m_config; } - -private: - IWorker *m_worker; - size_t m_totalWays; - uint32_t m_offset; - uv_thread_t m_thread; - xmrig::IThread *m_config; -}; - - -#endif /* __HANDLE_H__ */ diff --git a/src/workers/MultiWorker.cpp b/src/workers/MultiWorker.cpp deleted file mode 100644 index 02eec378e..000000000 --- a/src/workers/MultiWorker.cpp +++ /dev/null @@ -1,273 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2018 Lee Clagett - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - - -#include - - -#include "crypto/CryptoNight_test.h" -#include "common/log/Log.h" -#include "workers/CpuThread.h" -#include "workers/MultiWorker.h" -#include "workers/Workers.h" - - -template -MultiWorker::MultiWorker(Handle *handle) - : Worker(handle) -{ - m_memory = Mem::create(m_ctx, m_thread->algorithm(), N); -} - - -template -MultiWorker::~MultiWorker() -{ - Mem::release(m_ctx, N, m_memory); -} - - -template -bool MultiWorker::selfTest() -{ - using namespace xmrig; - - if (m_thread->algorithm() == CRYPTONIGHT) { - const bool rc = verify(VARIANT_0, test_output_v0) && - verify(VARIANT_1, test_output_v1) && - verify(VARIANT_2, test_output_v2) && - verify(VARIANT_XTL, test_output_xtl) && - verify(VARIANT_MSR, test_output_msr) && - verify(VARIANT_XAO, test_output_xao) && - verify(VARIANT_RTO, test_output_rto) && - verify(VARIANT_HALF, test_output_half) && - verify2(VARIANT_WOW, test_output_wow) && - verify2(VARIANT_4, test_output_r) && - verify(VARIANT_RWZ, test_output_rwz) && - verify(VARIANT_ZLS, test_output_zls) && - verify(VARIANT_DOUBLE, test_output_double); - -# ifndef XMRIG_NO_CN_GPU - if (!rc || N > 1) { - return rc; - } - - return verify(VARIANT_GPU, test_output_gpu); -# else - return rc; -# endif - } - -# ifndef XMRIG_NO_AEON - if (m_thread->algorithm() == CRYPTONIGHT_LITE) { - return verify(VARIANT_0, test_output_v0_lite) && - verify(VARIANT_1, test_output_v1_lite); - } -# endif - -# ifndef XMRIG_NO_SUMO - if (m_thread->algorithm() == CRYPTONIGHT_HEAVY) { - return verify(VARIANT_0, test_output_v0_heavy) && - verify(VARIANT_XHV, test_output_xhv_heavy) && - verify(VARIANT_TUBE, test_output_tube_heavy); - } -# endif - -# ifndef XMRIG_NO_CN_PICO - if (m_thread->algorithm() == CRYPTONIGHT_PICO) { - return verify(VARIANT_TRTL, test_output_pico_trtl); - } -# endif - - return false; -} - - -template -void MultiWorker::start() -{ - while (Workers::sequence() > 0) { - if (Workers::isPaused()) { - do { - std::this_thread::sleep_for(std::chrono::milliseconds(200)); - } - while (Workers::isPaused()); - - if (Workers::sequence() == 0) { - break; - } - - consumeJob(); - } - - while (!Workers::isOutdated(m_sequence)) { - if ((m_count & 0x7) == 0) { - storeStats(); - } - - m_thread->fn(m_state.job.algorithm().variant())(m_state.blob, m_state.job.size(), m_hash, m_ctx, m_state.job.height()); - - for (size_t i = 0; i < N; ++i) { - if (*reinterpret_cast(m_hash + (i * 32) + 24) < m_state.job.target()) { - Workers::submit(xmrig::JobResult(m_state.job.poolId(), m_state.job.id(), m_state.job.clientId(), *nonce(i), m_hash + (i * 32), m_state.job.diff(), m_state.job.algorithm())); - } - - *nonce(i) += 1; - } - - m_count += N; - - std::this_thread::yield(); - } - - consumeJob(); - } -} - - -template -bool MultiWorker::resume(const xmrig::Job &job) -{ - if (m_state.job.poolId() == -1 && job.poolId() >= 0 && job.id() == m_pausedState.job.id()) { - m_state = m_pausedState; - return true; - } - - return false; -} - - -template -bool MultiWorker::verify(xmrig::Variant variant, const uint8_t *referenceValue) -{ - - xmrig::CpuThread::cn_hash_fun func = m_thread->fn(variant); - if (!func) { - return false; - } - - func(test_input, 76, m_hash, m_ctx, 0); - return memcmp(m_hash, referenceValue, sizeof m_hash) == 0; -} - - -template -bool MultiWorker::verify2(xmrig::Variant variant, const uint8_t *referenceValue) -{ - xmrig::CpuThread::cn_hash_fun func = m_thread->fn(variant); - if (!func) { - return false; - } - - for (size_t i = 0; i < (sizeof(cn_r_test_input) / sizeof(cn_r_test_input[0])); ++i) { - const size_t size = cn_r_test_input[i].size; - for (size_t k = 0; k < N; ++k) { - memcpy(m_state.blob + (k * size), cn_r_test_input[i].data, size); - } - - func(m_state.blob, size, m_hash, m_ctx, cn_r_test_input[i].height); - - for (size_t k = 0; k < N; ++k) { - if (memcmp(m_hash + k * 32, referenceValue + i * 32, sizeof m_hash / N) != 0) { - return false; - } - } - } - - return true; -} - - -template<> -bool MultiWorker<1>::verify2(xmrig::Variant variant, const uint8_t *referenceValue) -{ - xmrig::CpuThread::cn_hash_fun func = m_thread->fn(variant); - if (!func) { - return false; - } - - for (size_t i = 0; i < (sizeof(cn_r_test_input) / sizeof(cn_r_test_input[0])); ++i) { - func(cn_r_test_input[i].data, cn_r_test_input[i].size, m_hash, m_ctx, cn_r_test_input[i].height); - - if (memcmp(m_hash, referenceValue + i * 32, sizeof m_hash) != 0) { - return false; - } - } - - return true; -} - - -template -void MultiWorker::consumeJob() -{ - xmrig::Job job = Workers::job(); - m_sequence = Workers::sequence(); - if (m_state.job == job) { - return; - } - - save(job); - - if (resume(job)) { - return; - } - - m_state.job = job; - - const size_t size = m_state.job.size(); - memcpy(m_state.blob, m_state.job.blob(), m_state.job.size()); - - if (N > 1) { - for (size_t i = 1; i < N; ++i) { - memcpy(m_state.blob + (i * size), m_state.blob, size); - } - } - - for (size_t i = 0; i < N; ++i) { - if (m_state.job.isNicehash()) { - *nonce(i) = (*nonce(i) & 0xff000000U) + (0xffffffU / m_totalWays * (m_offset + i)); - } - else { - *nonce(i) = 0xffffffffU / m_totalWays * (m_offset + i); - } - } -} - - -template -void MultiWorker::save(const xmrig::Job &job) -{ - if (job.poolId() == -1 && m_state.job.poolId() >= 0) { - m_pausedState = m_state; - } -} - - -template class MultiWorker<1>; -template class MultiWorker<2>; -template class MultiWorker<3>; -template class MultiWorker<4>; -template class MultiWorker<5>; diff --git a/src/workers/Worker.cpp b/src/workers/Worker.cpp deleted file mode 100644 index c569908c6..000000000 --- a/src/workers/Worker.cpp +++ /dev/null @@ -1,59 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2016-2018 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#include - - -#include "common/cpu/Cpu.h" -#include "common/Platform.h" -#include "workers/CpuThread.h" -#include "workers/Handle.h" -#include "workers/Worker.h" - - -Worker::Worker(Handle *handle) : - m_id(handle->threadId()), - m_totalWays(handle->totalWays()), - m_offset(handle->offset()), - m_hashCount(0), - m_timestamp(0), - m_count(0), - m_sequence(0), - m_thread(static_cast(handle->config())) -{ - if (xmrig::Cpu::info()->threads() > 1 && m_thread->affinity() != -1L) { - Platform::setThreadAffinity(m_thread->affinity()); - } - - Platform::setThreadPriority(m_thread->priority()); -} - - -void Worker::storeStats() -{ - using namespace std::chrono; - - const uint64_t timestamp = time_point_cast(high_resolution_clock::now()).time_since_epoch().count(); - m_hashCount.store(m_count, std::memory_order_relaxed); - m_timestamp.store(timestamp, std::memory_order_relaxed); -} diff --git a/src/workers/Workers.cpp b/src/workers/Workers.cpp deleted file mode 100644 index f718a52c5..000000000 --- a/src/workers/Workers.cpp +++ /dev/null @@ -1,364 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#include -#include -#include - - -#include "api/Api.h" -#include "common/log/Log.h" -#include "core/Config.h" -#include "core/Controller.h" -#include "crypto/CryptoNight_constants.h" -#include "interfaces/IJobResultListener.h" -#include "interfaces/IThread.h" -#include "Mem.h" -#include "rapidjson/document.h" -#include "workers/Handle.h" -#include "workers/Hashrate.h" -#include "workers/MultiWorker.h" -#include "workers/Workers.h" - - -bool Workers::m_active = false; -bool Workers::m_enabled = true; -Hashrate *Workers::m_hashrate = nullptr; -xmrig::IJobResultListener *Workers::m_listener = nullptr; -xmrig::Job Workers::m_job; -Workers::LaunchStatus Workers::m_status; -std::atomic Workers::m_paused; -std::atomic Workers::m_sequence; -std::list Workers::m_queue; -std::vector Workers::m_workers; -uint64_t Workers::m_ticks = 0; -uv_async_t Workers::m_async; -uv_mutex_t Workers::m_mutex; -uv_rwlock_t Workers::m_rwlock; -uv_timer_t Workers::m_timer; -xmrig::Controller *Workers::m_controller = nullptr; - - -xmrig::Job Workers::job() -{ - uv_rwlock_rdlock(&m_rwlock); - xmrig::Job job = m_job; - uv_rwlock_rdunlock(&m_rwlock); - - return job; -} - - -size_t Workers::hugePages() -{ - uv_mutex_lock(&m_mutex); - const size_t hugePages = m_status.hugePages; - uv_mutex_unlock(&m_mutex); - - return hugePages; -} - - -size_t Workers::threads() -{ - uv_mutex_lock(&m_mutex); - const size_t threads = m_status.threads; - uv_mutex_unlock(&m_mutex); - - return threads; -} - - -void Workers::printHashrate(bool detail) -{ - assert(m_controller != nullptr); - if (!m_controller) { - return; - } - - if (detail) { - const bool isColors = m_controller->config()->isColors(); - char num1[8] = { 0 }; - char num2[8] = { 0 }; - char num3[8] = { 0 }; - - Log::i()->text("%s| THREAD | AFFINITY | 10s H/s | 60s H/s | 15m H/s |", isColors ? "\x1B[1;37m" : ""); - - size_t i = 0; - for (const xmrig::IThread *thread : m_controller->config()->threads()) { - Log::i()->text("| %6zu | %8" PRId64 " | %7s | %7s | %7s |", - thread->index(), - thread->affinity(), - Hashrate::format(m_hashrate->calc(thread->index(), Hashrate::ShortInterval), num1, sizeof num1), - Hashrate::format(m_hashrate->calc(thread->index(), Hashrate::MediumInterval), num2, sizeof num2), - Hashrate::format(m_hashrate->calc(thread->index(), Hashrate::LargeInterval), num3, sizeof num3) - ); - - i++; - } - } - - m_hashrate->print(); -} - - -void Workers::setEnabled(bool enabled) -{ - if (m_enabled == enabled) { - return; - } - - m_enabled = enabled; - if (!m_active) { - return; - } - - m_paused = enabled ? 0 : 1; - m_sequence++; -} - - -void Workers::setJob(const xmrig::Job &job, bool donate) -{ - uv_rwlock_wrlock(&m_rwlock); - m_job = job; - - if (donate) { - m_job.setPoolId(-1); - } - uv_rwlock_wrunlock(&m_rwlock); - - m_active = true; - if (!m_enabled) { - return; - } - - m_sequence++; - m_paused = 0; -} - - -void Workers::start(xmrig::Controller *controller) -{ -# ifdef APP_DEBUG - LOG_NOTICE("THREADS ------------------------------------------------------------------"); - for (const xmrig::IThread *thread : controller->config()->threads()) { - thread->print(); - } - LOG_NOTICE("--------------------------------------------------------------------------"); -# endif - -# ifndef XMRIG_NO_ASM - xmrig::CpuThread::patchAsmVariants(); -# endif - - m_controller = controller; - - const std::vector &threads = controller->config()->threads(); - m_status.algo = controller->config()->algorithm().algo(); - m_status.colors = controller->config()->isColors(); - m_status.threads = threads.size(); - - for (const xmrig::IThread *thread : threads) { - m_status.ways += thread->multiway(); - } - - m_hashrate = new Hashrate(threads.size(), controller); - - uv_mutex_init(&m_mutex); - uv_rwlock_init(&m_rwlock); - - m_sequence = 1; - m_paused = 1; - - uv_async_init(uv_default_loop(), &m_async, Workers::onResult); - uv_timer_init(uv_default_loop(), &m_timer); - uv_timer_start(&m_timer, Workers::onTick, 500, 500); - - uint32_t offset = 0; - - for (xmrig::IThread *thread : threads) { - Handle *handle = new Handle(thread, offset, m_status.ways); - offset += thread->multiway(); - - m_workers.push_back(handle); - handle->start(Workers::onReady); - } - - controller->save(); -} - - -void Workers::stop() -{ - uv_timer_stop(&m_timer); - m_hashrate->stop(); - - uv_close(reinterpret_cast(&m_async), nullptr); - m_paused = 0; - m_sequence = 0; - - for (size_t i = 0; i < m_workers.size(); ++i) { - m_workers[i]->join(); - } -} - - -void Workers::submit(const xmrig::JobResult &result) -{ - uv_mutex_lock(&m_mutex); - m_queue.push_back(result); - uv_mutex_unlock(&m_mutex); - - uv_async_send(&m_async); -} - - -#ifndef XMRIG_NO_API -void Workers::threadsSummary(rapidjson::Document &doc) -{ - uv_mutex_lock(&m_mutex); - const uint64_t pages[2] = { m_status.hugePages, m_status.pages }; - const uint64_t memory = m_status.ways * xmrig::cn_select_memory(m_status.algo); - uv_mutex_unlock(&m_mutex); - - auto &allocator = doc.GetAllocator(); - - rapidjson::Value hugepages(rapidjson::kArrayType); - hugepages.PushBack(pages[0], allocator); - hugepages.PushBack(pages[1], allocator); - - doc.AddMember("hugepages", hugepages, allocator); - doc.AddMember("memory", memory, allocator); -} -#endif - - -void Workers::onReady(void *arg) -{ - auto handle = static_cast(arg); - - IWorker *worker = nullptr; - - switch (handle->config()->multiway()) { - case 1: - worker = new MultiWorker<1>(handle); - break; - - case 2: - worker = new MultiWorker<2>(handle); - break; - - case 3: - worker = new MultiWorker<3>(handle); - break; - - case 4: - worker = new MultiWorker<4>(handle); - break; - - case 5: - worker = new MultiWorker<5>(handle); - break; - - default: - break; - } - - handle->setWorker(worker); - - if (!worker->selfTest()) { - LOG_ERR("thread %zu error: \"hash self-test failed\".", handle->worker()->id()); - - return; - } - - start(worker); -} - - -void Workers::onResult(uv_async_t *handle) -{ - std::list results; - - uv_mutex_lock(&m_mutex); - while (!m_queue.empty()) { - results.push_back(std::move(m_queue.front())); - m_queue.pop_front(); - } - uv_mutex_unlock(&m_mutex); - - for (auto result : results) { - m_listener->onJobResult(result); - } - - results.clear(); -} - - -void Workers::onTick(uv_timer_t *handle) -{ - for (Handle *handle : m_workers) { - if (!handle->worker()) { - return; - } - - m_hashrate->add(handle->threadId(), handle->worker()->hashCount(), handle->worker()->timestamp()); - } - - if ((m_ticks++ & 0xF) == 0) { - m_hashrate->updateHighest(); - } -} - - -void Workers::start(IWorker *worker) -{ - const Worker *w = static_cast(worker); - - uv_mutex_lock(&m_mutex); - m_status.started++; - m_status.pages += w->memory().pages; - m_status.hugePages += w->memory().hugePages; - - if (m_status.started == m_status.threads) { - const double percent = (double) m_status.hugePages / m_status.pages * 100.0; - const size_t memory = m_status.ways * xmrig::cn_select_memory(m_status.algo) / 1024; - - if (m_status.colors) { - LOG_INFO(GREEN_BOLD("READY (CPU)") " threads " CYAN_BOLD("%zu(%zu)") " huge pages %s%zu/%zu %1.0f%%\x1B[0m memory " CYAN_BOLD("%zu KB") "", - m_status.threads, m_status.ways, - (m_status.hugePages == m_status.pages ? "\x1B[1;32m" : (m_status.hugePages == 0 ? "\x1B[1;31m" : "\x1B[1;33m")), - m_status.hugePages, m_status.pages, percent, memory); - } - else { - LOG_INFO("READY (CPU) threads %zu(%zu) huge pages %zu/%zu %1.0f%% memory %zu KB", - m_status.threads, m_status.ways, m_status.hugePages, m_status.pages, percent, memory); - } - } - - uv_mutex_unlock(&m_mutex); - - worker->start(); -} diff --git a/src/workers/Workers.h b/src/workers/Workers.h deleted file mode 100644 index a9b8e6958..000000000 --- a/src/workers/Workers.h +++ /dev/null @@ -1,122 +0,0 @@ -/* XMRig - * Copyright 2010 Jeff Garzik - * Copyright 2012-2014 pooler - * Copyright 2014 Lucas Jones - * Copyright 2014-2016 Wolf9466 - * Copyright 2016 Jay D Dee - * Copyright 2017-2018 XMR-Stak , - * Copyright 2018-2019 SChernykh - * Copyright 2016-2019 XMRig , - * - * This program is free software: you can redistribute it and/or modify - * it under the terms of the GNU General Public License as published by - * the Free Software Foundation, either version 3 of the License, or - * (at your option) any later version. - * - * This program is distributed in the hope that it will be useful, - * but WITHOUT ANY WARRANTY; without even the implied warranty of - * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the - * GNU General Public License for more details. - * - * You should have received a copy of the GNU General Public License - * along with this program. If not, see . - */ - -#ifndef XMRIG_WORKERS_H -#define XMRIG_WORKERS_H - - -#include -#include -#include -#include - -#include "common/net/Job.h" -#include "net/JobResult.h" -#include "rapidjson/fwd.h" - - -class Handle; -class Hashrate; -class IWorker; - - -namespace xmrig { - class Controller; - class IJobResultListener; -} - - -class Workers -{ -public: - static xmrig::Job job(); - static size_t hugePages(); - static size_t threads(); - static void printHashrate(bool detail); - static void setEnabled(bool enabled); - static void setJob(const xmrig::Job &job, bool donate); - static void start(xmrig::Controller *controller); - static void stop(); - static void submit(const xmrig::JobResult &result); - - static inline bool isEnabled() { return m_enabled; } - static inline bool isOutdated(uint64_t sequence) { return m_sequence.load(std::memory_order_relaxed) != sequence; } - static inline bool isPaused() { return m_paused.load(std::memory_order_relaxed) == 1; } - static inline Hashrate *hashrate() { return m_hashrate; } - static inline uint64_t sequence() { return m_sequence.load(std::memory_order_relaxed); } - static inline void pause() { m_active = false; m_paused = 1; m_sequence++; } - static inline void setListener(xmrig::IJobResultListener *listener) { m_listener = listener; } - -# ifndef XMRIG_NO_API - static void threadsSummary(rapidjson::Document &doc); -# endif - -private: - static void onReady(void *arg); - static void onResult(uv_async_t *handle); - static void onTick(uv_timer_t *handle); - static void start(IWorker *worker); - - class LaunchStatus - { - public: - inline LaunchStatus() : - colors(true), - hugePages(0), - pages(0), - started(0), - threads(0), - ways(0), - algo(xmrig::CRYPTONIGHT) - {} - - bool colors; - size_t hugePages; - size_t pages; - size_t started; - size_t threads; - size_t ways; - xmrig::Algo algo; - }; - - static bool m_active; - static bool m_enabled; - static Hashrate *m_hashrate; - static xmrig::IJobResultListener *m_listener; - static xmrig::Job m_job; - static LaunchStatus m_status; - static std::atomic m_paused; - static std::atomic m_sequence; - static std::list m_queue; - static std::vector m_workers; - static uint64_t m_ticks; - static uv_async_t m_async; - static uv_mutex_t m_mutex; - static uv_rwlock_t m_rwlock; - static uv_timer_t m_timer; - static xmrig::Controller *m_controller; -}; - - -#endif /* XMRIG_WORKERS_H */