Fixed macOS build.

.gitignore

@@ -1,2 +1 @@
 /build
-/CMakeLists.txt.user

CHANGELOG.md

@@ -1,93 +1,11 @@
-# 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).
+# v0.9.0
+- **[#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 option `--asm`, possible values `--asm auto`, `--asm none`, `--asm intel` and `--asm ryzen`.
+- Added support for new style long and short algorithm names, possible values: `cryptonight`, `cryptonight/0`, `cryptonight/1`, `cryptonight/2`, `cryptonight-lite`, `cryptonight-lite/0`, `cryptonight-lite/1` and short equvalents `cn/2` etc. 
+- Added `--variant`, example `--algo cn --variant 2`, by default miner automaticaly detect proper variant for Monero by block version.  
+- Added CryptoNight-Lite variant 1.
+- Added xmrig-proxy autodetection, nicehash will be enabled automaticaly. 
+- Added workaround for xmrig-proxy [bug](https://github.com/xmrig/xmrig-proxy/commit/dfa1960fe3eeb13f80717b7dbfcc7c6e9f222d89).
 
 # v0.8.2
 - Fixed L2 cache size detection for AMD CPUs (Bulldozer/Piledriver/Steamroller/Excavator architecture).

CMakeLists.txt

@@ -1,207 +1,156 @@
-cmake_minimum_required(VERSION 2.8)
-project(xmrig)
+cmake_minimum_required(VERSION 3.0)
+project(xmrig C)
 
 option(WITH_LIBCPUID "Use Libcpuid" ON)
 option(WITH_AEON     "CryptoNight-Lite support" ON)
-option(WITH_HTTPD    "HTTP REST API" ON)
-
-include (CheckIncludeFile)
-include (cmake/cpu.cmake)
-
+option(WITH_ASM      "Enable ASM PoW implementations" ON)
 
 set(HEADERS
-    src/3rdparty/align.h
-    src/api/Api.h
-    src/api/ApiState.h
-    src/api/NetworkState.h
-    src/App.h
-    src/Console.h
-    src/Cpu.h
-    src/interfaces/IClientListener.h
-    src/interfaces/IConsoleListener.h
-    src/interfaces/IJobResultListener.h
-    src/interfaces/ILogBackend.h
-    src/interfaces/IStrategy.h
-    src/interfaces/IStrategyListener.h
-    src/interfaces/IWorker.h
-    src/log/ConsoleLog.h
-    src/log/FileLog.h
-    src/log/Log.h
-    src/Mem.h
-    src/net/Client.h
-    src/net/Job.h
-    src/net/JobId.h
-    src/net/JobResult.h
-    src/net/Network.h
-    src/net/strategies/DonateStrategy.h
-    src/net/strategies/FailoverStrategy.h
-    src/net/strategies/SinglePoolStrategy.h
-    src/net/SubmitResult.h
-    src/net/Url.h
-    src/Options.h
-    src/Platform.h
-    src/Summary.h
-    src/version.h
-    src/workers/DoubleWorker.h
-    src/workers/Handle.h
-    src/workers/Hashrate.h
-    src/workers/SingleWorker.h
-    src/workers/Worker.h
-    src/workers/Workers.h
+    algo/cryptonight/cryptonight.h
+    algo/cryptonight/cryptonight_aesni.h
+    algo/cryptonight/cryptonight_monero.h
+    algo/cryptonight/cryptonight_softaes.h
+    algo/cryptonight/cryptonight_test.h
+    algo/cryptonight/variant4_random_math.h
+    compat.h
+    cpu.h
+    donate.h
+    elist.h
+    options.h
+    persistent_memory.h
+    stats.h
+    stratum.h
+    util.h
+    version.h
+    xmrig.h
    )
 
 set(HEADERS_CRYPTO
-    src/crypto/c_blake256.h
-    src/crypto/c_groestl.h
-    src/crypto/c_jh.h
-    src/crypto/c_keccak.h
-    src/crypto/c_skein.h
-    src/crypto/CryptoNight.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
+    crypto/c_groestl.h
+    crypto/c_blake256.h
+    crypto/c_jh.h
+    crypto/c_skein.h
+    crypto/soft_aes.h
    )
 
-if (XMRIG_ARM)
-    set(HEADERS_CRYPTO "${HEADERS_CRYPTO}" src/crypto/CryptoNight_arm.h)
-else()
-    set(HEADERS_CRYPTO "${HEADERS_CRYPTO}" src/crypto/CryptoNight_x86.h)
-endif()
+set(HEADERS_COMPAT
+    compat/winansi.h
+   )
+
+set(HEADERS_UTILS
+    utils/applog.h
+    utils/threads.h
+    utils/summary.h
+   )
 
 set(SOURCES
-    src/api/Api.cpp
-    src/api/ApiState.cpp
-    src/api/NetworkState.cpp
-    src/App.cpp
-    src/Console.cpp
-    src/log/ConsoleLog.cpp
-    src/log/FileLog.cpp
-    src/log/Log.cpp
-    src/Mem.cpp
-    src/net/Client.cpp
-    src/net/Job.cpp
-    src/net/Network.cpp
-    src/net/strategies/DonateStrategy.cpp
-    src/net/strategies/FailoverStrategy.cpp
-    src/net/strategies/SinglePoolStrategy.cpp
-    src/net/SubmitResult.cpp
-    src/net/Url.cpp
-    src/Options.cpp
-    src/Platform.cpp
-    src/Summary.cpp
-    src/workers/DoubleWorker.cpp
-    src/workers/Handle.cpp
-    src/workers/Hashrate.cpp
-    src/workers/SingleWorker.cpp
-    src/workers/Worker.cpp
-    src/workers/Workers.cpp
-    src/xmrig.cpp
+    xmrig.c
+    algo/cryptonight/cryptonight.c
+    algo/cryptonight/cryptonight_av1.c
+    algo/cryptonight/cryptonight_av2.c
+    algo/cryptonight/cryptonight_av3.c
+    algo/cryptonight/cryptonight_av4.c
+    algo/cryptonight/cryptonight_r_av1.c
+    algo/cryptonight/cryptonight_r_av2.c
+    algo/cryptonight/cryptonight_r_av3.c
+    algo/cryptonight/cryptonight_r_av4.c
+    util.c
+    options.c
+    stratum.c
+    stats.c
+    memory.c
    )
 
 set(SOURCES_CRYPTO
-    src/crypto/c_keccak.c
-    src/crypto/c_groestl.c
-    src/crypto/c_blake256.c
-    src/crypto/c_jh.c
-    src/crypto/c_skein.c
-    src/crypto/CryptoNight.cpp
+    crypto/c_keccak.c
+    crypto/c_groestl.c
+    crypto/c_blake256.c
+    crypto/c_jh.c
+    crypto/c_skein.c
+   )
+
+set(SOURCES_UTILS
+    utils/applog.c
+    utils/summary.c
    )
 
 if (WIN32)
-    set(SOURCES_OS
-        res/app.rc
-        src/App_win.cpp
-        src/Cpu_win.cpp
-        src/Mem_win.cpp
-        src/Platform_win.cpp
-        )
-
-    add_definitions(/DWIN32)
-    set(EXTRA_LIBS ws2_32 psapi iphlpapi userenv)
+    set(SOURCES_OS win/cpu_win.c win/memory_win.c win/xmrig_win.c win/app.rc compat/winansi.c)
+    set(EXTRA_LIBS ws2_32 crypt32)
+    add_definitions(/D_WIN32_WINNT=0x600)
 elseif (APPLE)
-    set(SOURCES_OS
-        src/App_unix.cpp
-        src/Cpu_mac.cpp
-        src/Mem_unix.cpp
-        src/Platform_mac.cpp
-        )
+    set(SOURCES_OS mac/cpu_mac.c mac/memory_mac.c mac/xmrig_mac.c)
 else()
-    set(SOURCES_OS
-        src/App_unix.cpp
-        src/Cpu_unix.cpp
-        src/Mem_unix.cpp
-        src/Platform_unix.cpp
-        )
-
-    set(EXTRA_LIBS pthread rt)
+    set(SOURCES_OS unix/cpu_unix.c unix/memory_unix.c unix/xmrig_unix.c)
+    set(EXTRA_LIBS pthread rt m)
 endif()
 
-if (CMAKE_SYSTEM_NAME STREQUAL FreeBSD)
-    set(EXTRA_LIBS ${EXTRA_LIBS} kvm)
-endif()
-
-if (CMAKE_SYSTEM_NAME MATCHES "Linux")
-    EXECUTE_PROCESS(COMMAND uname -o COMMAND tr -d '\n' OUTPUT_VARIABLE OPERATING_SYSTEM)
-    if (OPERATING_SYSTEM MATCHES "Android")
-        set(EXTRA_LIBS ${EXTRA_LIBS} log)
-    endif()
-endif()
-
-add_definitions(/D__STDC_FORMAT_MACROS)
+include_directories(.)
+add_definitions(/DUSE_NATIVE_THREADS)
+add_definitions(/D_GNU_SOURCE)
 add_definitions(/DUNICODE)
-#add_definitions(/DAPP_DEBUG)
 
-set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} "${CMAKE_SOURCE_DIR}/cmake")
+if ("${CMAKE_BUILD_TYPE}" STREQUAL "")
+    set(CMAKE_BUILD_TYPE Release)
+endif()
 
-find_package(UV REQUIRED)
+set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -maes -Wno-pointer-to-int-cast")
 
-include(cmake/flags.cmake)
+if (CMAKE_C_COMPILER_ID MATCHES "Clang")
+    set(CMAKE_C_FLAGS_RELEASE "${CMAKE_C_FLAGS_RELEASE} -Ofast -s -funroll-loops -fvariable-expansion-in-unroller -fmerge-all-constants")
+else()
+    set(CMAKE_C_FLAGS_RELEASE "${CMAKE_C_FLAGS_RELEASE} -Ofast -s -funroll-loops -fvariable-expansion-in-unroller -ftree-loop-if-convert-stores -fmerge-all-constants -fbranch-target-load-optimize2")
+endif()
+
+#set(CMAKE_C_FLAGS_RELEASE "${CMAKE_C_FLAGS_RELEASE} -gdwarf-2")
+#set(CMAKE_C_FLAGS_RELEASE "${CMAKE_C_FLAGS_RELEASE} -fprofile-generate")
+#set(CMAKE_C_FLAGS_RELEASE "${CMAKE_C_FLAGS_RELEASE} -fprofile-use -fprofile-correction")
+
+if (WIN32)
+    set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -static")
+endif()
+
+include_directories(compat/jansson)
+add_subdirectory(compat/jansson)
+
+set(CMAKE_MODULE_PATH ${CMAKE_MODULE_PATH} "${CMAKE_SOURCE_DIR}/cmake/")
+
+find_package(CURL REQUIRED)
+include_directories(${CURL_INCLUDE_DIRS})
+add_definitions(/DCURL_STATICLIB)
+link_directories(${CURL_LIBRARIES})
 
 if (WITH_LIBCPUID)
-    add_subdirectory(src/3rdparty/libcpuid)
+    add_subdirectory(compat/libcpuid)
 
-    include_directories(src/3rdparty/libcpuid)
+    include_directories(compat/libcpuid)
     set(CPUID_LIB cpuid)
-    set(SOURCES_CPUID src/Cpu.cpp)
+    set(SOURCES_CPUID cpu.c)
 else()
     add_definitions(/DXMRIG_NO_LIBCPUID)
-
-    if (XMRIG_ARM)
-        set(SOURCES_CPUID src/Cpu_arm.cpp)
-    else()
-        set(SOURCES_CPUID src/Cpu_stub.cpp)
-    endif()
+    set(SOURCES_CPUID cpu_stub.c)
 endif()
 
-CHECK_INCLUDE_FILE (syslog.h HAVE_SYSLOG_H)
-if (HAVE_SYSLOG_H)
-    add_definitions(/DHAVE_SYSLOG_H)
-    set(SOURCES_SYSLOG src/log/SysLog.h src/log/SysLog.cpp)
-endif()
+include(cmake/asm.cmake)
 
-if (NOT WITH_AEON)
+if (WITH_AEON)
+    set(SOURCES_AEON
+    algo/cryptonight-lite/cryptonight_lite_av1.c
+    algo/cryptonight-lite/cryptonight_lite_av2.c
+    algo/cryptonight-lite/cryptonight_lite_av3.c
+    algo/cryptonight-lite/cryptonight_lite_av4.c
+    algo/cryptonight-lite/cryptonight_lite_aesni.h
+    algo/cryptonight-lite/cryptonight_lite_softaes.h
+    )
+else()
     add_definitions(/DXMRIG_NO_AEON)
 endif()
 
-if (WITH_HTTPD)
-    find_package(MHD)
-
-    if (MHD_FOUND)
-        include_directories(${MHD_INCLUDE_DIRS})
-        set(HTTPD_SOURCES src/api/Httpd.h src/api/Httpd.cpp)
-    else()
-        message(FATAL_ERROR "microhttpd NOT found: use `-DWITH_HTTPD=OFF` to build without http deamon support")
-    endif()
+if (CMAKE_SIZEOF_VOID_P EQUAL 8)
+    add_executable(xmrig ${HEADERS} ${HEADERS_CRYPTO} ${SOURCES} ${SOURCES_CRYPTO} ${HEADERS_UTILS} ${SOURCES_UTILS} ${HEADERS_COMPAT} ${SOURCES_COMPAT} ${SOURCES_OS} ${SOURCES_CPUID} ${SOURCES_AEON} ${XMRIG_ASM_SOURCES})
+    target_link_libraries(xmrig ${XMRIG_ASM_LIBRARY} jansson ${CURL_LIBRARY} ${CPUID_LIB} ${EXTRA_LIBS})
 else()
-    add_definitions(/DXMRIG_NO_HTTPD)
-    add_definitions(/DXMRIG_NO_API)
+    add_executable(xmrig32 ${HEADERS} ${HEADERS_CRYPTO} ${SOURCES} ${SOURCES_CRYPTO} ${HEADERS_UTILS} ${SOURCES_UTILS} ${HEADERS_COMPAT} ${SOURCES_COMPAT} ${SOURCES_OS} ${SOURCES_CPUID} ${SOURCES_AEON} ${XMRIG_ASM_SOURCES})
+    target_link_libraries(xmrig32 ${XMRIG_ASM_LIBRARY} jansson ${CURL_LIBRARY} ${CPUID_LIB} ${EXTRA_LIBS})
 endif()
 
-include_directories(src)
-include_directories(src/3rdparty)
-include_directories(${UV_INCLUDE_DIR})
-
-add_executable(xmrig ${HEADERS} ${SOURCES} ${SOURCES_OS} ${SOURCES_CPUID} ${HEADERS_CRYPTO} ${SOURCES_CRYPTO} ${SOURCES_SYSLOG} ${HTTPD_SOURCES})
-target_link_libraries(xmrig ${UV_LIBRARIES} ${MHD_LIBRARY} ${EXTRA_LIBS} ${CPUID_LIB})

README.md

@@ -1,28 +1,24 @@
 # XMRig
-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 high performance Monero (XMR) CPU miner, with the official full Windows support.
+Based on cpuminer-multi with heavy optimizations/rewrites and removing a lot of legacy code.
 
-* 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.
-
-<img src="http://i.imgur.com/OKZRVDh.png" width="619" >
+<img src="http://i.imgur.com/GdRDnAu.png" width="596" >
 
 #### Table of contents
 * [Features](#features)
 * [Download](#download)
 * [Usage](#usage)
 * [Algorithm variations](#algorithm-variations)
-* [Build](https://github.com/xmrig/xmrig/wiki/Build)
+* [Build](#build)
 * [Common Issues](#common-issues)
 * [Other information](#other-information)
 * [Donations](#donations)
 * [Contacts](#contacts)
 
 ## Features
-* High performance.
+* High performance (290+ H/s on i7 6700).
 * Official Windows support.
-* Small Windows executable, without dependencies.
-* x86/x64 support.
+* Small Windows executable, only 535 KB without dependencies.
 * Support for backup (failover) mining server.
 * keepalived support.
 * Command line options compatible with cpuminer.
@@ -34,54 +30,39 @@ Originally based on cpuminer-multi with heavy optimizations/rewrites and removin
 ## 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).
+  * Clone with `git clone https://github.com/xmrig/xmrig.git`
 
 ## Usage
 ### Basic example
 ```
-xmrig.exe -o pool.monero.hashvault.pro:5555 -u YOUR_WALLET -p x -k
+xmrig.exe -o xmr-eu.dwarfpool.com:8005 -u YOUR_WALLET -p x -k
 ```
 
-### Failover
-```
-xmrig.exe -o pool.monero.hashvault.pro:5555 -u YOUR_WALLET1 -p x -k -o pool.supportxmr.com:5555 -u YOUR_WALLET2 -p x -k
-```
-For failover you can add multiple pools, maximum count not limited.
-
 ### Options
 ```
-  -a, --algo=ALGO          cryptonight (default) or cryptonight-lite
-  -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
-  -t, --threads=N          number of miner threads
-  -v, --av=N               algorithm variation, 0 auto select
-  -k, --keepalive          send keepalived for prevent timeout (need pool support)
-  -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
-      --max-cpu-usage=N    maximum CPU usage for automatic threads mode (default 75)
-      --safe               safe adjust threads and av settings for current CPU
-      --nicehash           enable nicehash/xmrig-proxy support
-      --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
-  -h, --help               display this help and exit
-  -V, --version            output version information and exit
+  -a, --algo=ALGO       cryptonight (default) or cryptonight-lite
+  -o, --url=URL         URL of mining server
+  -b, --backup-url=URL  URL of backup 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
+  -t, --threads=N       number of miner threads
+  -v, --av=N            algorithm variation, 0 auto select
+  -k, --keepalive       send keepalived for prevent timeout (need pool support)
+  -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
+      --no-color        disable colored output
+      --donate-level=N  donate level, default 5% (5 minutes in 100 minutes)
+  -B, --background      run the miner in the background
+  -c, --config=FILE     load a JSON-format configuration file
+      --max-cpu-usage=N maximum cpu usage for automatic threads mode (default 75)
+      --safe            safe adjust threads and av settings for current cpu
+      --nicehash        enable nicehash support
+  -h, --help            display this help and exit
+  -V, --version         output version information and exit
 ```
 
-Also you can use configuration via config file, default **config.json**. You can load multiple config files and combine it with command line options.
-
 ## Algorithm variations
 Since version 0.8.0.
 * `--av=1` For CPUs with hardware AES.
@@ -89,10 +70,46 @@ Since version 0.8.0.
 * `--av=3` Software AES implementation.
 * `--av=4` Lower power mode (double hash) of `3`.
 
+## Build
+### Ubuntu (Debian-based distros)
+```
+sudo apt-get install git build-essential cmake libcurl4-openssl-dev
+git clone https://github.com/xmrig/xmrig.git
+cd xmrig
+mkdir build
+cd build
+cmake .. -DCMAKE_BUILD_TYPE=Release
+make
+```
+
+### Windows
+It's complicated, you need [MSYS2](http://www.msys2.org/), custom libcurl build, and of course CMake too.
+
+Necessary MSYS2 packages:
+```
+pacman -Sy
+pacman -S mingw-w64-x86_64-gcc
+pacman -S make
+pacman -S mingw-w64-x86_64-cmake
+pacman -S mingw-w64-x86_64-pkg-config
+```
+Configure options for libcurl:
+```
+./configure --disable-shared --enable-optimize --enable-threaded-resolver --disable-libcurl-option --disable-ares --disable-rt --disable-ftp --disable-file --disable-ldap --disable-ldaps --disable-rtsp --disable-dict --disable-telnet --disable-tftp --disable-pop3 --disable-imap --disable-smb --disable-smtp --disable-gopher --disable-manual --disable-ipv6 --disable-sspi --disable-crypto-auth --disable-ntlm-wb --disable-tls-srp --disable-unix-sockets --without-zlib --without-winssl --without-ssl --without-libssh2 --without-nghttp2 --disable-cookies --without-ca-bundle --without-librtmp
+```
+CMake options:
+```
+cmake .. -G "Unix Makefiles" -DCMAKE_BUILD_TYPE=Release -DCURL_INCLUDE_DIR="c:\xmrig-deps\gcc\x64\include" -DCURL_LIBRARY="c:\xmrig-deps\gcc\x64\lib\libcurl.a"
+```
+
+### Optional features
+`-DWITH_LIBCPUID=OFF` Disable libcpuid. Auto configuration of CPU after this will be very limited.
+`-DWITH_AEON=OFF` Disable CryptoNight-Lite support.
+
 ## 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).
+* Since version 0.8.0 XMRig automatically enable 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.
@@ -101,15 +118,15 @@ Since version 0.8.0.
 
 
 ### CPU mining performance
-* **Intel i7-7700** - 307 H/s (4 threads)
-* **AMD Ryzen 7 1700X** - 560 H/s (8 threads)
+* **i7-6700** - 290+ H/s (4 threads, cpu affinity 0xAA)
+* **Dual E5620** - 377 H/s (12 threads, cpu affinity 0xEEEE)
 
 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.
+* Use modern CPUs with AES-NI instructuon set.
 * Try setup optimal cpu affinity.
 * Enable fast memory (Large/Huge pages).
 

algo/cryptonight-lite/cryptonight_lite_aesni.h

@@ -0,0 +1,274 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017      fireice-uk  <https://github.com/fireice-uk>
+ * Copyright 2016-2017 XMRig       <support@xmrig.com>
+ *
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef XMRIG_CRYPTONIGHT_LITE_AESNI_H
+#define XMRIG_CRYPTONIGHT_LITE_AESNI_H
+
+
+#include <x86intrin.h>
+#include <stdint.h>
+
+
+#define aes_genkey_sub(imm8) \
+    __m128i xout1 = _mm_aeskeygenassist_si128(*xout2, (imm8)); \
+    xout1  = _mm_shuffle_epi32(xout1, 0xFF); \
+    *xout0 = sl_xor(*xout0); \
+    *xout0 = _mm_xor_si128(*xout0, xout1); \
+    xout1  = _mm_aeskeygenassist_si128(*xout0, 0x00);\
+    xout1  = _mm_shuffle_epi32(xout1, 0xAA); \
+    *xout2 = sl_xor(*xout2); \
+    *xout2 = _mm_xor_si128(*xout2, xout1); \
+
+
+// This will shift and xor tmp1 into itself as 4 32-bit vals such as
+// sl_xor(a1 a2 a3 a4) = a1 (a2^a1) (a3^a2^a1) (a4^a3^a2^a1)
+static inline __m128i sl_xor(__m128i tmp1)
+{
+    __m128i tmp4;
+    tmp4 = _mm_slli_si128(tmp1, 0x04);
+    tmp1 = _mm_xor_si128(tmp1, tmp4);
+    tmp4 = _mm_slli_si128(tmp4, 0x04);
+    tmp1 = _mm_xor_si128(tmp1, tmp4);
+    tmp4 = _mm_slli_si128(tmp4, 0x04);
+    tmp1 = _mm_xor_si128(tmp1, tmp4);
+    return tmp1;
+}
+
+
+static inline void aes_genkey_sub1(__m128i* xout0, __m128i* xout2)
+{
+   aes_genkey_sub(0x1)
+}
+
+
+static inline void aes_genkey_sub2(__m128i* xout0, __m128i* xout2)
+{
+   aes_genkey_sub(0x2)
+}
+
+
+static inline void aes_genkey_sub4(__m128i* xout0, __m128i* xout2)
+{
+   aes_genkey_sub(0x4)
+}
+
+
+static inline void aes_genkey_sub8(__m128i* xout0, __m128i* xout2)
+{
+   aes_genkey_sub(0x8)
+}
+
+
+static inline void aes_round(__m128i key, __m128i* x0, __m128i* x1, __m128i* x2, __m128i* x3, __m128i* x4, __m128i* x5, __m128i* x6, __m128i* x7)
+{
+    *x0 = _mm_aesenc_si128(*x0, key);
+    *x1 = _mm_aesenc_si128(*x1, key);
+    *x2 = _mm_aesenc_si128(*x2, key);
+    *x3 = _mm_aesenc_si128(*x3, key);
+    *x4 = _mm_aesenc_si128(*x4, key);
+    *x5 = _mm_aesenc_si128(*x5, key);
+    *x6 = _mm_aesenc_si128(*x6, key);
+    *x7 = _mm_aesenc_si128(*x7, key);
+}
+
+
+static inline void aes_genkey(const __m128i* memory, __m128i* k0, __m128i* k1, __m128i* k2, __m128i* k3, __m128i* k4, __m128i* k5, __m128i* k6, __m128i* k7, __m128i* k8, __m128i* k9)
+{
+    __m128i xout0 = _mm_load_si128(memory);
+    __m128i xout2 = _mm_load_si128(memory + 1);
+    *k0 = xout0;
+    *k1 = xout2;
+
+     aes_genkey_sub1(&xout0, &xout2);
+    *k2 = xout0;
+    *k3 = xout2;
+
+     aes_genkey_sub2(&xout0, &xout2);
+    *k4 = xout0;
+    *k5 = xout2;
+
+     aes_genkey_sub4(&xout0, &xout2);
+    *k6 = xout0;
+    *k7 = xout2;
+
+     aes_genkey_sub8(&xout0, &xout2);
+    *k8 = xout0;
+    *k9 = xout2;
+}
+
+
+static inline void cn_explode_scratchpad(const __m128i* input, __m128i* output)
+{
+    // This is more than we have registers, compiler will assign 2 keys on the stack
+    __m128i xin0, xin1, xin2, xin3, xin4, xin5, xin6, xin7;
+    __m128i k0, k1, k2, k3, k4, k5, k6, k7, k8, k9;
+
+    aes_genkey(input, &k0, &k1, &k2, &k3, &k4, &k5, &k6, &k7, &k8, &k9);
+
+    xin0 = _mm_load_si128(input + 4);
+    xin1 = _mm_load_si128(input + 5);
+    xin2 = _mm_load_si128(input + 6);
+    xin3 = _mm_load_si128(input + 7);
+    xin4 = _mm_load_si128(input + 8);
+    xin5 = _mm_load_si128(input + 9);
+    xin6 = _mm_load_si128(input + 10);
+    xin7 = _mm_load_si128(input + 11);
+
+    for (size_t i = 0; __builtin_expect(i < MEMORY_LITE / sizeof(__m128i), 1); 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);
+        aes_round(k3, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k4, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k5, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k6, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k7, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k8, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k9, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+
+        _mm_store_si128(output + i + 0, xin0);
+        _mm_store_si128(output + i + 1, xin1);
+        _mm_store_si128(output + i + 2, xin2);
+        _mm_store_si128(output + i + 3, xin3);
+        _mm_store_si128(output + i + 4, xin4);
+        _mm_store_si128(output + i + 5, xin5);
+        _mm_store_si128(output + i + 6, xin6);
+        _mm_store_si128(output + i + 7, xin7);
+    }
+}
+
+
+static inline void cn_implode_scratchpad(const __m128i* input, __m128i* output)
+{
+    // This is more than we have registers, compiler will assign 2 keys on the stack
+    __m128i xout0, xout1, xout2, xout3, xout4, xout5, xout6, xout7;
+    __m128i k0, k1, k2, k3, k4, k5, k6, k7, k8, k9;
+
+    aes_genkey(output + 2, &k0, &k1, &k2, &k3, &k4, &k5, &k6, &k7, &k8, &k9);
+
+    xout0 = _mm_load_si128(output + 4);
+    xout1 = _mm_load_si128(output + 5);
+    xout2 = _mm_load_si128(output + 6);
+    xout3 = _mm_load_si128(output + 7);
+    xout4 = _mm_load_si128(output + 8);
+    xout5 = _mm_load_si128(output + 9);
+    xout6 = _mm_load_si128(output + 10);
+    xout7 = _mm_load_si128(output + 11);
+
+    for (size_t i = 0; __builtin_expect(i < MEMORY_LITE / sizeof(__m128i), 1); 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);
+        xout3 = _mm_xor_si128(_mm_load_si128(input + i + 3), xout3);
+        xout4 = _mm_xor_si128(_mm_load_si128(input + i + 4), xout4);
+        xout5 = _mm_xor_si128(_mm_load_si128(input + i + 5), xout5);
+        xout6 = _mm_xor_si128(_mm_load_si128(input + i + 6), xout6);
+        xout7 = _mm_xor_si128(_mm_load_si128(input + i + 7), xout7);
+
+        aes_round(k0, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k1, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k2, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k3, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k4, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k5, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k6, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k7, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k8, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k9, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+    }
+
+    _mm_store_si128(output + 4, xout0);
+    _mm_store_si128(output + 5, xout1);
+    _mm_store_si128(output + 6, xout2);
+    _mm_store_si128(output + 7, xout3);
+    _mm_store_si128(output + 8, xout4);
+    _mm_store_si128(output + 9, xout5);
+    _mm_store_si128(output + 10, xout6);
+    _mm_store_si128(output + 11, xout7);
+}
+
+
+#if defined(__x86_64__)
+#   define EXTRACT64(X) _mm_cvtsi128_si64(X)
+
+static inline uint64_t _umul128(uint64_t a, uint64_t b, uint64_t* hi)
+{
+    unsigned __int128 r = (unsigned __int128) a * (unsigned __int128) b;
+    *hi = r >> 64;
+    return (uint64_t) r;
+}
+#elif defined(__i386__)
+#   define HI32(X) \
+    _mm_srli_si128((X), 4)
+
+
+#   define EXTRACT64(X) \
+    ((uint64_t)(uint32_t)_mm_cvtsi128_si32(X) | \
+    ((uint64_t)(uint32_t)_mm_cvtsi128_si32(HI32(X)) << 32))
+
+static inline uint64_t _umul128(uint64_t multiplier, uint64_t multiplicand, uint64_t *product_hi) {
+    // multiplier   = ab = a * 2^32 + b
+    // multiplicand = cd = c * 2^32 + d
+    // ab * cd = a * c * 2^64 + (a * d + b * c) * 2^32 + b * d
+    uint64_t a = multiplier >> 32;
+    uint64_t b = multiplier & 0xFFFFFFFF;
+    uint64_t c = multiplicand >> 32;
+    uint64_t d = multiplicand & 0xFFFFFFFF;
+
+    //uint64_t ac = a * c;
+    uint64_t ad = a * d;
+    //uint64_t bc = b * c;
+    uint64_t bd = b * d;
+
+    uint64_t adbc = ad + (b * c);
+    uint64_t adbc_carry = adbc < ad ? 1 : 0;
+
+    // multiplier * multiplicand = product_hi * 2^64 + product_lo
+    uint64_t product_lo = bd + (adbc << 32);
+    uint64_t product_lo_carry = product_lo < bd ? 1 : 0;
+    *product_hi = (a * c) + (adbc >> 32) + (adbc_carry << 32) + product_lo_carry;
+
+    return product_lo;
+}
+#endif
+
+
+static inline void cryptonight_monero_tweak(uint64_t* mem_out, __m128i tmp)
+{
+    mem_out[0] = EXTRACT64(tmp);
+
+    tmp = _mm_castps_si128(_mm_movehl_ps(_mm_castsi128_ps(tmp), _mm_castsi128_ps(tmp)));
+    uint64_t vh = EXTRACT64(tmp);
+
+    uint8_t x = vh >> 24;
+    static const uint16_t table = 0x7531;
+    const uint8_t index = (((x >> 3) & 6) | (x & 1)) << 1;
+    vh ^= ((table >> index) & 0x3) << 28;
+
+    mem_out[1] = vh;
+}
+
+
+#endif /* XMRIG_CRYPTONIGHT_LITE_AESNI_H */

algo/cryptonight-lite/cryptonight_lite_av1.c

@@ -0,0 +1,134 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017      fireice-uk  <https://github.com/fireice-uk>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018      Lee Clagett <https://github.com/vtnerd>
+ * Copyright 2016-2018 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#include <x86intrin.h>
+#include <string.h>
+
+#include "algo/cryptonight/cryptonight.h"
+#include "algo/cryptonight/cryptonight_monero.h"
+#include "crypto/c_keccak.h"
+#include "cryptonight_lite_aesni.h"
+
+
+void cryptonight_lite_av1_v0(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    keccak(input, size, ctx[0]->state, 200);
+
+    cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+
+    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]);
+
+    uint64_t idx0 = h0[0] ^ h0[4];
+
+    for (size_t i = 0; __builtin_expect(i < 0x40000, 1); i++) {
+        __m128i cx;
+        cx = _mm_load_si128((__m128i *) &l0[idx0 & 0xFFFF0]);
+        cx = _mm_aesenc_si128(cx, _mm_set_epi64x(ah0, al0));
+
+        _mm_store_si128((__m128i *) &l0[idx0 & 0xFFFF0], _mm_xor_si128(bx0, cx));
+        idx0 = EXTRACT64(cx);
+        bx0 = cx;
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0xFFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0xFFFF0])[1];
+        lo = _umul128(idx0, cl, &hi);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*)&l0[idx0 & 0xFFFF0])[0] = al0;
+        ((uint64_t*)&l0[idx0 & 0xFFFF0])[1] = ah0;
+
+        ah0 ^= ch;
+        al0 ^= cl;
+        idx0 = al0;
+    }
+
+    cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+
+    keccakf(h0, 24);
+    extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
+}
+
+
+void cryptonight_lite_av1_v1(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    if (size < 43) {
+        memset(output, 0, 32);
+        return;
+    }
+
+    keccak(input, size, ctx[0]->state, 200);
+
+    VARIANT1_INIT(0);
+
+    cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+
+    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]);
+
+    uint64_t idx0 = h0[0] ^ h0[4];
+
+    for (size_t i = 0; __builtin_expect(i < 0x40000, 1); i++) {
+        __m128i cx;
+        cx = _mm_load_si128((__m128i *) &l0[idx0 & 0xFFFF0]);
+        cx = _mm_aesenc_si128(cx, _mm_set_epi64x(ah0, al0));
+
+        cryptonight_monero_tweak((uint64_t*)&l0[idx0 & 0xFFFF0], _mm_xor_si128(bx0, cx));
+
+        idx0 = EXTRACT64(cx);
+        bx0 = cx;
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0xFFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0xFFFF0])[1];
+        lo = _umul128(idx0, cl, &hi);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*)&l0[idx0 & 0xFFFF0])[0] = al0;
+        ((uint64_t*)&l0[idx0 & 0xFFFF0])[1] = ah0 ^ tweak1_2_0;
+
+        ah0 ^= ch;
+        al0 ^= cl;
+        idx0 = al0;
+    }
+
+    cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+
+    keccakf(h0, 24);
+    extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
+}

algo/cryptonight-lite/cryptonight_lite_av2.c

@@ -0,0 +1,202 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017      fireice-uk  <https://github.com/fireice-uk>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018      Lee Clagett <https://github.com/vtnerd>
+ * Copyright 2016-2018 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#include <x86intrin.h>
+#include <string.h>
+
+#include "algo/cryptonight/cryptonight.h"
+#include "algo/cryptonight/cryptonight_monero.h"
+#include "cryptonight_lite_aesni.h"
+#include "crypto/c_keccak.h"
+
+
+void cryptonight_lite_av2_v0(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    keccak(input,        size, ctx[0]->state, 200);
+    keccak(input + size, size, ctx[1]->state, 200);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    const uint8_t* l1 = ctx[1]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+    uint64_t* h1 = (uint64_t*) ctx[1]->state;
+
+    cn_explode_scratchpad((__m128i*) h0, (__m128i*) l0);
+    cn_explode_scratchpad((__m128i*) h1, (__m128i*) l1);
+
+    uint64_t al0 = h0[0] ^ h0[4];
+    uint64_t al1 = h1[0] ^ h1[4];
+    uint64_t ah0 = h0[1] ^ h0[5];
+    uint64_t ah1 = h1[1] ^ h1[5];
+
+    __m128i bx0 = _mm_set_epi64x(h0[3] ^ h0[7], h0[2] ^ h0[6]);
+    __m128i bx1 = _mm_set_epi64x(h1[3] ^ h1[7], h1[2] ^ h1[6]);
+
+    uint64_t idx0 = h0[0] ^ h0[4];
+    uint64_t idx1 = h1[0] ^ h1[4];
+
+    for (size_t i = 0; __builtin_expect(i < 0x40000, 1); i++) {
+        __m128i cx0 = _mm_load_si128((__m128i *) &l0[idx0 & 0xFFFF0]);
+        __m128i cx1 = _mm_load_si128((__m128i *) &l1[idx1 & 0xFFFF0]);
+
+        cx0 = _mm_aesenc_si128(cx0, _mm_set_epi64x(ah0, al0));
+        cx1 = _mm_aesenc_si128(cx1, _mm_set_epi64x(ah1, al1));
+
+        _mm_store_si128((__m128i *) &l0[idx0 & 0xFFFF0], _mm_xor_si128(bx0, cx0));
+        _mm_store_si128((__m128i *) &l1[idx1 & 0xFFFF0], _mm_xor_si128(bx1, cx1));
+
+        idx0 = EXTRACT64(cx0);
+        idx1 = EXTRACT64(cx1);
+
+        bx0 = cx0;
+        bx1 = cx1;
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0xFFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0xFFFF0])[1];
+        lo = _umul128(idx0, cl, &hi);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*) &l0[idx0 & 0xFFFF0])[0] = al0;
+        ((uint64_t*) &l0[idx0 & 0xFFFF0])[1] = ah0;
+
+        ah0 ^= ch;
+        al0 ^= cl;
+        idx0 = al0;
+
+        cl = ((uint64_t*) &l1[idx1 & 0xFFFF0])[0];
+        ch = ((uint64_t*) &l1[idx1 & 0xFFFF0])[1];
+        lo = _umul128(idx1, cl, &hi);
+
+        al1 += hi;
+        ah1 += lo;
+
+        ((uint64_t*) &l1[idx1 & 0xFFFF0])[0] = al1;
+        ((uint64_t*) &l1[idx1 & 0xFFFF0])[1] = ah1;
+
+        ah1 ^= ch;
+        al1 ^= cl;
+        idx1 = al1;
+    }
+
+    cn_implode_scratchpad((__m128i*) l0, (__m128i*) h0);
+    cn_implode_scratchpad((__m128i*) l1, (__m128i*) h1);
+
+    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, (char*) output + 32);
+}
+
+
+void cryptonight_lite_av2_v1(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    if (size < 43) {
+        memset(output, 0, 64);
+        return;
+    }
+
+    keccak(input,        size, ctx[0]->state, 200);
+    keccak(input + size, size, ctx[1]->state, 200);
+
+    VARIANT1_INIT(0);
+    VARIANT1_INIT(1);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    const uint8_t* l1 = ctx[1]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+    uint64_t* h1 = (uint64_t*) ctx[1]->state;
+
+    cn_explode_scratchpad((__m128i*) h0, (__m128i*) l0);
+    cn_explode_scratchpad((__m128i*) h1, (__m128i*) l1);
+
+    uint64_t al0 = h0[0] ^ h0[4];
+    uint64_t al1 = h1[0] ^ h1[4];
+    uint64_t ah0 = h0[1] ^ h0[5];
+    uint64_t ah1 = h1[1] ^ h1[5];
+
+    __m128i bx0 = _mm_set_epi64x(h0[3] ^ h0[7], h0[2] ^ h0[6]);
+    __m128i bx1 = _mm_set_epi64x(h1[3] ^ h1[7], h1[2] ^ h1[6]);
+
+    uint64_t idx0 = h0[0] ^ h0[4];
+    uint64_t idx1 = h1[0] ^ h1[4];
+
+    for (size_t i = 0; __builtin_expect(i < 0x40000, 1); i++) {
+        __m128i cx0 = _mm_load_si128((__m128i *) &l0[idx0 & 0xFFFF0]);
+        __m128i cx1 = _mm_load_si128((__m128i *) &l1[idx1 & 0xFFFF0]);
+
+        cx0 = _mm_aesenc_si128(cx0, _mm_set_epi64x(ah0, al0));
+        cx1 = _mm_aesenc_si128(cx1, _mm_set_epi64x(ah1, al1));
+
+        cryptonight_monero_tweak((uint64_t*)&l0[idx0 & 0xFFFF0], _mm_xor_si128(bx0, cx0));
+        cryptonight_monero_tweak((uint64_t*)&l1[idx1 & 0xFFFF0], _mm_xor_si128(bx1, cx1));
+
+        idx0 = EXTRACT64(cx0);
+        idx1 = EXTRACT64(cx1);
+
+        bx0 = cx0;
+        bx1 = cx1;
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0xFFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0xFFFF0])[1];
+        lo = _umul128(idx0, cl, &hi);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*) &l0[idx0 & 0xFFFF0])[0] = al0;
+        ((uint64_t*) &l0[idx0 & 0xFFFF0])[1] = ah0 ^ tweak1_2_0;
+
+        ah0 ^= ch;
+        al0 ^= cl;
+        idx0 = al0;
+
+        cl = ((uint64_t*) &l1[idx1 & 0xFFFF0])[0];
+        ch = ((uint64_t*) &l1[idx1 & 0xFFFF0])[1];
+        lo = _umul128(idx1, cl, &hi);
+
+        al1 += hi;
+        ah1 += lo;
+
+        ((uint64_t*) &l1[idx1 & 0xFFFF0])[0] = al1;
+        ((uint64_t*) &l1[idx1 & 0xFFFF0])[1] = ah1 ^ tweak1_2_1;
+
+        ah1 ^= ch;
+        al1 ^= cl;
+        idx1 = al1;
+    }
+
+    cn_implode_scratchpad((__m128i*) l0, (__m128i*) h0);
+    cn_implode_scratchpad((__m128i*) l1, (__m128i*) h1);
+
+    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, (char*) output + 32);
+}

algo/cryptonight-lite/cryptonight_lite_av3.c

@@ -0,0 +1,134 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017      fireice-uk  <https://github.com/fireice-uk>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018      Lee Clagett <https://github.com/vtnerd>
+ * Copyright 2016-2018 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#include <x86intrin.h>
+#include <string.h>
+
+#include "algo/cryptonight/cryptonight.h"
+#include "algo/cryptonight/cryptonight_monero.h"
+#include "cryptonight_lite_softaes.h"
+#include "crypto/c_keccak.h"
+
+
+void cryptonight_lite_av3_v0(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    keccak(input, size, ctx[0]->state, 200);
+
+    cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+
+    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]);
+
+    uint64_t idx0 = h0[0] ^ h0[4];
+
+    for (size_t i = 0; __builtin_expect(i < 0x40000, 1); i++) {
+        __m128i cx;
+        cx = _mm_load_si128((__m128i *) &l0[idx0 & 0xFFFF0]);
+        cx = soft_aesenc(cx, _mm_set_epi64x(ah0, al0));
+
+        _mm_store_si128((__m128i *) &l0[idx0 & 0xFFFF0], _mm_xor_si128(bx0, cx));
+        idx0 = EXTRACT64(cx);
+        bx0 = cx;
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0xFFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0xFFFF0])[1];
+        lo = _umul128(idx0, cl, &hi);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*)&l0[idx0 & 0xFFFF0])[0] = al0;
+        ((uint64_t*)&l0[idx0 & 0xFFFF0])[1] = ah0;
+
+        ah0 ^= ch;
+        al0 ^= cl;
+        idx0 = al0;
+    }
+
+    cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+
+    keccakf(h0, 24);
+    extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
+}
+
+
+void cryptonight_lite_av3_v1(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    if (size < 43) {
+        memset(output, 0, 32);
+        return;
+    }
+
+    keccak(input, size, ctx[0]->state, 200);
+
+    VARIANT1_INIT(0);
+
+    cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+
+    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]);
+
+    uint64_t idx0 = h0[0] ^ h0[4];
+
+    for (size_t i = 0; __builtin_expect(i < 0x40000, 1); i++) {
+        __m128i cx;
+        cx = _mm_load_si128((__m128i *) &l0[idx0 & 0xFFFF0]);
+        cx = soft_aesenc(cx, _mm_set_epi64x(ah0, al0));
+
+        cryptonight_monero_tweak((uint64_t*)&l0[idx0 & 0xFFFF0], _mm_xor_si128(bx0, cx));
+
+        idx0 = EXTRACT64(cx);
+        bx0 = cx;
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0xFFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0xFFFF0])[1];
+        lo = _umul128(idx0, cl, &hi);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*)&l0[idx0 & 0xFFFF0])[0] = al0;
+        ((uint64_t*)&l0[idx0 & 0xFFFF0])[1] = ah0 ^ tweak1_2_0;
+
+        ah0 ^= ch;
+        al0 ^= cl;
+        idx0 = al0;
+    }
+
+    cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+
+    keccakf(h0, 24);
+    extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
+}

algo/cryptonight-lite/cryptonight_lite_av4.c

@@ -0,0 +1,202 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017      fireice-uk  <https://github.com/fireice-uk>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018      Lee Clagett <https://github.com/vtnerd>
+ * Copyright 2016-2018 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#include <x86intrin.h>
+#include <string.h>
+
+#include "algo/cryptonight/cryptonight.h"
+#include "algo/cryptonight/cryptonight_monero.h"
+#include "cryptonight_lite_softaes.h"
+#include "crypto/c_keccak.h"
+
+
+void cryptonight_lite_av4_v0(const void *restrict input, size_t size, void *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    keccak(input,        size, ctx[0]->state, 200);
+    keccak(input + size, size, ctx[1]->state, 200);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    const uint8_t* l1 = ctx[1]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+    uint64_t* h1 = (uint64_t*) ctx[1]->state;
+
+    cn_explode_scratchpad((__m128i*) h0, (__m128i*) l0);
+    cn_explode_scratchpad((__m128i*) h1, (__m128i*) l1);
+
+    uint64_t al0 = h0[0] ^ h0[4];
+    uint64_t al1 = h1[0] ^ h1[4];
+    uint64_t ah0 = h0[1] ^ h0[5];
+    uint64_t ah1 = h1[1] ^ h1[5];
+
+    __m128i bx0 = _mm_set_epi64x(h0[3] ^ h0[7], h0[2] ^ h0[6]);
+    __m128i bx1 = _mm_set_epi64x(h1[3] ^ h1[7], h1[2] ^ h1[6]);
+
+    uint64_t idx0 = h0[0] ^ h0[4];
+    uint64_t idx1 = h1[0] ^ h1[4];
+
+    for (size_t i = 0; __builtin_expect(i < 0x40000, 1); i++) {
+        __m128i cx0 = _mm_load_si128((__m128i *) &l0[idx0 & 0xFFFF0]);
+        __m128i cx1 = _mm_load_si128((__m128i *) &l1[idx1 & 0xFFFF0]);
+
+        cx0 = soft_aesenc(cx0, _mm_set_epi64x(ah0, al0));
+        cx1 = soft_aesenc(cx1, _mm_set_epi64x(ah1, al1));
+
+        _mm_store_si128((__m128i *) &l0[idx0 & 0xFFFF0], _mm_xor_si128(bx0, cx0));
+        _mm_store_si128((__m128i *) &l1[idx1 & 0xFFFF0], _mm_xor_si128(bx1, cx1));
+
+        idx0 = EXTRACT64(cx0);
+        idx1 = EXTRACT64(cx1);
+
+        bx0 = cx0;
+        bx1 = cx1;
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0xFFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0xFFFF0])[1];
+        lo = _umul128(idx0, cl, &hi);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*) &l0[idx0 & 0xFFFF0])[0] = al0;
+        ((uint64_t*) &l0[idx0 & 0xFFFF0])[1] = ah0;
+
+        ah0 ^= ch;
+        al0 ^= cl;
+        idx0 = al0;
+
+        cl = ((uint64_t*) &l1[idx1 & 0xFFFF0])[0];
+        ch = ((uint64_t*) &l1[idx1 & 0xFFFF0])[1];
+        lo = _umul128(idx1, cl, &hi);
+
+        al1 += hi;
+        ah1 += lo;
+
+        ((uint64_t*) &l1[idx1 & 0xFFFF0])[0] = al1;
+        ((uint64_t*) &l1[idx1 & 0xFFFF0])[1] = ah1;
+
+        ah1 ^= ch;
+        al1 ^= cl;
+        idx1 = al1;
+    }
+
+    cn_implode_scratchpad((__m128i*) l0, (__m128i*) h0);
+    cn_implode_scratchpad((__m128i*) l1, (__m128i*) h1);
+
+    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);
+}
+
+
+void cryptonight_lite_av4_v1(const void *restrict input, size_t size, void *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    if (size < 43) {
+        memset(output, 0, 64);
+        return;
+    }
+
+    keccak(input,        size, ctx[0]->state, 200);
+    keccak(input + size, size, ctx[1]->state, 200);
+
+    VARIANT1_INIT(0);
+    VARIANT1_INIT(1);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    const uint8_t* l1 = ctx[1]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+    uint64_t* h1 = (uint64_t*) ctx[1]->state;
+
+    cn_explode_scratchpad((__m128i*) h0, (__m128i*) l0);
+    cn_explode_scratchpad((__m128i*) h1, (__m128i*) l1);
+
+    uint64_t al0 = h0[0] ^ h0[4];
+    uint64_t al1 = h1[0] ^ h1[4];
+    uint64_t ah0 = h0[1] ^ h0[5];
+    uint64_t ah1 = h1[1] ^ h1[5];
+
+    __m128i bx0 = _mm_set_epi64x(h0[3] ^ h0[7], h0[2] ^ h0[6]);
+    __m128i bx1 = _mm_set_epi64x(h1[3] ^ h1[7], h1[2] ^ h1[6]);
+
+    uint64_t idx0 = h0[0] ^ h0[4];
+    uint64_t idx1 = h1[0] ^ h1[4];
+
+    for (size_t i = 0; __builtin_expect(i < 0x40000, 1); i++) {
+        __m128i cx0 = _mm_load_si128((__m128i *) &l0[idx0 & 0xFFFF0]);
+        __m128i cx1 = _mm_load_si128((__m128i *) &l1[idx1 & 0xFFFF0]);
+
+        cx0 = soft_aesenc(cx0, _mm_set_epi64x(ah0, al0));
+        cx1 = soft_aesenc(cx1, _mm_set_epi64x(ah1, al1));
+
+        cryptonight_monero_tweak((uint64_t*)&l0[idx0 & 0xFFFF0], _mm_xor_si128(bx0, cx0));
+        cryptonight_monero_tweak((uint64_t*)&l1[idx1 & 0xFFFF0], _mm_xor_si128(bx1, cx1));
+
+        idx0 = EXTRACT64(cx0);
+        idx1 = EXTRACT64(cx1);
+
+        bx0 = cx0;
+        bx1 = cx1;
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0xFFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0xFFFF0])[1];
+        lo = _umul128(idx0, cl, &hi);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*) &l0[idx0 & 0xFFFF0])[0] = al0;
+        ((uint64_t*) &l0[idx0 & 0xFFFF0])[1] = ah0 ^ tweak1_2_0;
+
+        ah0 ^= ch;
+        al0 ^= cl;
+        idx0 = al0;
+
+        cl = ((uint64_t*) &l1[idx1 & 0xFFFF0])[0];
+        ch = ((uint64_t*) &l1[idx1 & 0xFFFF0])[1];
+        lo = _umul128(idx1, cl, &hi);
+
+        al1 += hi;
+        ah1 += lo;
+
+        ((uint64_t*) &l1[idx1 & 0xFFFF0])[0] = al1;
+        ((uint64_t*) &l1[idx1 & 0xFFFF0])[1] = ah1 ^ tweak1_2_1;
+
+        ah1 ^= ch;
+        al1 ^= cl;
+        idx1 = al1;
+    }
+
+    cn_implode_scratchpad((__m128i*) l0, (__m128i*) h0);
+    cn_implode_scratchpad((__m128i*) l1, (__m128i*) h1);
+
+    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, (char*) output + 32);
+}

algo/cryptonight-lite/cryptonight_lite_softaes.h

@@ -0,0 +1,255 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018-2019 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2019 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef XMRIG_CRYPTONIGHT_LITE_SOFTAES_H
+#define XMRIG_CRYPTONIGHT_LITE_SOFTAES_H
+
+
+#include <x86intrin.h>
+#include <stdint.h>
+
+
+#include "crypto/soft_aes.h"
+
+
+// This will shift and xor tmp1 into itself as 4 32-bit vals such as
+// sl_xor(a1 a2 a3 a4) = a1 (a2^a1) (a3^a2^a1) (a4^a3^a2^a1)
+static inline __m128i sl_xor(__m128i tmp1)
+{
+    __m128i tmp4;
+    tmp4 = _mm_slli_si128(tmp1, 0x04);
+    tmp1 = _mm_xor_si128(tmp1, tmp4);
+    tmp4 = _mm_slli_si128(tmp4, 0x04);
+    tmp1 = _mm_xor_si128(tmp1, tmp4);
+    tmp4 = _mm_slli_si128(tmp4, 0x04);
+    tmp1 = _mm_xor_si128(tmp1, tmp4);
+    return tmp1;
+}
+
+
+static inline void aes_genkey_sub(__m128i* xout0, __m128i* xout2, uint8_t rcon)
+{
+    __m128i xout1 = soft_aeskeygenassist(*xout2, rcon);
+    xout1  = _mm_shuffle_epi32(xout1, 0xFF); // see PSHUFD, set all elems to 4th elem
+    *xout0 = sl_xor(*xout0);
+    *xout0 = _mm_xor_si128(*xout0, xout1);
+    xout1  = soft_aeskeygenassist(*xout0, 0x00);
+    xout1  = _mm_shuffle_epi32(xout1, 0xAA); // see PSHUFD, set all elems to 3rd elem
+    *xout2 = sl_xor(*xout2);
+    *xout2 = _mm_xor_si128(*xout2, xout1);
+}
+
+
+static inline void aes_round(__m128i key, __m128i* x0, __m128i* x1, __m128i* x2, __m128i* x3, __m128i* x4, __m128i* x5, __m128i* x6, __m128i* x7)
+{
+    *x0 = soft_aesenc(*x0, key);
+    *x1 = soft_aesenc(*x1, key);
+    *x2 = soft_aesenc(*x2, key);
+    *x3 = soft_aesenc(*x3, key);
+    *x4 = soft_aesenc(*x4, key);
+    *x5 = soft_aesenc(*x5, key);
+    *x6 = soft_aesenc(*x6, key);
+    *x7 = soft_aesenc(*x7, key);
+}
+
+
+static inline void aes_genkey(const __m128i* memory, __m128i* k0, __m128i* k1, __m128i* k2, __m128i* k3, __m128i* k4, __m128i* k5, __m128i* k6, __m128i* k7, __m128i* k8, __m128i* k9)
+{
+    __m128i xout0 = _mm_load_si128(memory);
+    __m128i xout2 = _mm_load_si128(memory + 1);
+    *k0 = xout0;
+    *k1 = xout2;
+
+    aes_genkey_sub(&xout0, &xout2, 0x1);
+    *k2 = xout0;
+    *k3 = xout2;
+
+    aes_genkey_sub(&xout0, &xout2, 0x2);
+    *k4 = xout0;
+    *k5 = xout2;
+
+    aes_genkey_sub(&xout0, &xout2, 0x4);
+    *k6 = xout0;
+    *k7 = xout2;
+
+    aes_genkey_sub(&xout0, &xout2, 0x8);
+    *k8 = xout0;
+    *k9 = xout2;
+}
+
+
+static inline void cn_explode_scratchpad(const __m128i* input, __m128i* output)
+{
+    // This is more than we have registers, compiler will assign 2 keys on the stack
+    __m128i xin0, xin1, xin2, xin3, xin4, xin5, xin6, xin7;
+    __m128i k0, k1, k2, k3, k4, k5, k6, k7, k8, k9;
+
+    aes_genkey(input, &k0, &k1, &k2, &k3, &k4, &k5, &k6, &k7, &k8, &k9);
+
+    xin0 = _mm_load_si128(input + 4);
+    xin1 = _mm_load_si128(input + 5);
+    xin2 = _mm_load_si128(input + 6);
+    xin3 = _mm_load_si128(input + 7);
+    xin4 = _mm_load_si128(input + 8);
+    xin5 = _mm_load_si128(input + 9);
+    xin6 = _mm_load_si128(input + 10);
+    xin7 = _mm_load_si128(input + 11);
+
+    for (size_t i = 0; i < MEMORY_LITE / 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);
+        aes_round(k3, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k4, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k5, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k6, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k7, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k8, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k9, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+
+        _mm_store_si128(output + i + 0, xin0);
+        _mm_store_si128(output + i + 1, xin1);
+        _mm_store_si128(output + i + 2, xin2);
+        _mm_store_si128(output + i + 3, xin3);
+        _mm_store_si128(output + i + 4, xin4);
+        _mm_store_si128(output + i + 5, xin5);
+        _mm_store_si128(output + i + 6, xin6);
+        _mm_store_si128(output + i + 7, xin7);
+    }
+}
+
+
+static inline void cn_implode_scratchpad(const __m128i* input, __m128i* output)
+{
+    // This is more than we have registers, compiler will assign 2 keys on the stack
+    __m128i xout0, xout1, xout2, xout3, xout4, xout5, xout6, xout7;
+    __m128i k0, k1, k2, k3, k4, k5, k6, k7, k8, k9;
+
+    aes_genkey(output + 2, &k0, &k1, &k2, &k3, &k4, &k5, &k6, &k7, &k8, &k9);
+
+    xout0 = _mm_load_si128(output + 4);
+    xout1 = _mm_load_si128(output + 5);
+    xout2 = _mm_load_si128(output + 6);
+    xout3 = _mm_load_si128(output + 7);
+    xout4 = _mm_load_si128(output + 8);
+    xout5 = _mm_load_si128(output + 9);
+    xout6 = _mm_load_si128(output + 10);
+    xout7 = _mm_load_si128(output + 11);
+
+    for (size_t i = 0; __builtin_expect(i < MEMORY_LITE / sizeof(__m128i), 1); 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);
+        xout3 = _mm_xor_si128(_mm_load_si128(input + i + 3), xout3);
+        xout4 = _mm_xor_si128(_mm_load_si128(input + i + 4), xout4);
+        xout5 = _mm_xor_si128(_mm_load_si128(input + i + 5), xout5);
+        xout6 = _mm_xor_si128(_mm_load_si128(input + i + 6), xout6);
+        xout7 = _mm_xor_si128(_mm_load_si128(input + i + 7), xout7);
+
+        aes_round(k0, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k1, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k2, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k3, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k4, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k5, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k6, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k7, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k8, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k9, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+    }
+
+    _mm_store_si128(output + 4, xout0);
+    _mm_store_si128(output + 5, xout1);
+    _mm_store_si128(output + 6, xout2);
+    _mm_store_si128(output + 7, xout3);
+    _mm_store_si128(output + 8, xout4);
+    _mm_store_si128(output + 9, xout5);
+    _mm_store_si128(output + 10, xout6);
+    _mm_store_si128(output + 11, xout7);
+}
+
+
+#if defined(__x86_64__)
+#   define EXTRACT64(X) _mm_cvtsi128_si64(X)
+
+static inline uint64_t _umul128(uint64_t a, uint64_t b, uint64_t* hi)
+{
+    unsigned __int128 r = (unsigned __int128) a * (unsigned __int128) b;
+    *hi = r >> 64;
+    return (uint64_t) r;
+}
+#elif defined(__i386__)
+#   define HI32(X) \
+    _mm_srli_si128((X), 4)
+
+
+#   define EXTRACT64(X) \
+    ((uint64_t)(uint32_t)_mm_cvtsi128_si32(X) | \
+    ((uint64_t)(uint32_t)_mm_cvtsi128_si32(HI32(X)) << 32))
+
+static inline uint64_t _umul128(uint64_t multiplier, uint64_t multiplicand, uint64_t *product_hi) {
+    // multiplier   = ab = a * 2^32 + b
+    // multiplicand = cd = c * 2^32 + d
+    // ab * cd = a * c * 2^64 + (a * d + b * c) * 2^32 + b * d
+    uint64_t a = multiplier >> 32;
+    uint64_t b = multiplier & 0xFFFFFFFF;
+    uint64_t c = multiplicand >> 32;
+    uint64_t d = multiplicand & 0xFFFFFFFF;
+
+    //uint64_t ac = a * c;
+    uint64_t ad = a * d;
+    //uint64_t bc = b * c;
+    uint64_t bd = b * d;
+
+    uint64_t adbc = ad + (b * c);
+    uint64_t adbc_carry = adbc < ad ? 1 : 0;
+
+    // multiplier * multiplicand = product_hi * 2^64 + product_lo
+    uint64_t product_lo = bd + (adbc << 32);
+    uint64_t product_lo_carry = product_lo < bd ? 1 : 0;
+    *product_hi = (a * c) + (adbc >> 32) + (adbc_carry << 32) + product_lo_carry;
+
+    return product_lo;
+}
+#endif
+
+
+static inline void cryptonight_monero_tweak(uint64_t* mem_out, __m128i tmp)
+{
+    mem_out[0] = EXTRACT64(tmp);
+
+    tmp = _mm_castps_si128(_mm_movehl_ps(_mm_castsi128_ps(tmp), _mm_castsi128_ps(tmp)));
+    uint64_t vh = EXTRACT64(tmp);
+
+    uint8_t x = vh >> 24;
+    static const uint16_t table = 0x7531;
+    const uint8_t index = (((x >> 3) & 6) | (x & 1)) << 1;
+    vh ^= ((table >> index) & 0x3) << 28;
+
+    mem_out[1] = vh;
+}
+
+
+#endif /* XMRIG_CRYPTONIGHT_LITE_SOFTAES_H */

algo/cryptonight/cryptonight.c

@@ -0,0 +1,407 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018      Lee Clagett <https://github.com/vtnerd>
+ * Copyright 2018-2019 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2019 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+
+#include <assert.h>
+#include <stdlib.h>
+#include <string.h>
+#include <mm_malloc.h>
+
+
+#ifndef BUILD_TEST
+#   include "xmrig.h"
+#endif
+
+#include "cpu.h"
+#include "crypto/c_blake256.h"
+#include "crypto/c_groestl.h"
+#include "crypto/c_jh.h"
+#include "crypto/c_skein.h"
+#include "cryptonight_test.h"
+#include "cryptonight.h"
+#include "options.h"
+#include "persistent_memory.h"
+
+
+static cn_hash_fun asm_func_map[AV_MAX][VARIANT_MAX][ASM_MAX] = {};
+
+
+void cryptonight_av1_v0(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_av1_v1(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_av1_v2(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_av2_v0(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_av2_v1(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_av2_v2(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_av3_v0(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_av3_v1(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_av3_v2(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_av4_v0(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_av4_v1(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_av4_v2(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+
+void cryptonight_r_av1(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_r_av2(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_r_av3(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_r_av4(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+
+
+#ifndef XMRIG_NO_AEON
+void cryptonight_lite_av1_v0(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_lite_av1_v1(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_lite_av2_v0(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_lite_av2_v1(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_lite_av3_v0(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_lite_av3_v1(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_lite_av4_v0(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_lite_av4_v1(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+#endif
+
+
+#ifndef XMRIG_NO_ASM
+void cryptonight_single_hash_asm_intel(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_single_hash_asm_ryzen(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_single_hash_asm_bulldozer(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_double_hash_asm(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+
+void cryptonight_r_av1_asm_intel(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_r_av1_asm_bulldozer(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_r_av2_asm_intel(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+void cryptonight_r_av2_asm_bulldozer(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+#endif
+
+
+static inline bool verify(enum Variant variant, uint8_t *output, struct cryptonight_ctx **ctx, const uint8_t *referenceValue)
+{
+    cn_hash_fun func = cryptonight_hash_fn(opt_algo, opt_av, variant);
+    if (func == NULL) {
+        return false;
+    }
+
+    func(test_input, 76, output, ctx);
+
+    return memcmp(output, referenceValue, opt_double_hash ? 64 : 32) == 0;
+}
+
+
+static inline bool verify2(enum Variant variant, uint8_t *output, struct cryptonight_ctx **ctx, const uint8_t *referenceValue)
+{
+    cn_hash_fun func = cryptonight_hash_fn(opt_algo, opt_av, variant);
+    if (func == NULL) {
+        return false;
+    }
+
+    if (opt_double_hash) {
+        uint8_t input[128];
+
+        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;
+            memcpy(input,        cn_r_test_input[i].data, size);
+            memcpy(input + size, cn_r_test_input[i].data, size);
+
+            ctx[0]->height = ctx[1]->height = cn_r_test_input[i].height;
+
+            func(input, size, output, ctx);
+
+            if (memcmp(output, referenceValue + i * 32, 32) != 0 || memcmp(output + 32, referenceValue + i * 32, 32) != 0) {
+                return false;
+            }
+        }
+    }
+    else {
+        for (size_t i = 0; i < (sizeof(cn_r_test_input) / sizeof(cn_r_test_input[0])); ++i) {
+            ctx[0]->height = cn_r_test_input[i].height;
+
+            func(cn_r_test_input[i].data, cn_r_test_input[i].size, output, ctx);
+
+            if (memcmp(output, referenceValue + i * 32, 32) != 0) {
+                return false;
+            }
+        }
+    }
+
+    return true;
+}
+
+
+static bool self_test() {
+    struct cryptonight_ctx *ctx[2];
+    uint8_t output[64];
+
+    const size_t count = opt_double_hash ? 2 : 1;
+    const size_t size  = opt_algo == ALGO_CRYPTONIGHT ? MEMORY : MEMORY_LITE;
+    bool result = false;
+
+    for (size_t i = 0; i < count; ++i) {
+        ctx[i]         = _mm_malloc(sizeof(struct cryptonight_ctx), 16);
+        ctx[i]->memory = _mm_malloc(size, 16);
+
+        init_cn_r(ctx[i]);
+    }
+
+    if (opt_algo == ALGO_CRYPTONIGHT) {
+        result = verify(VARIANT_0,  output, ctx, test_output_v0) &&
+                 verify(VARIANT_1,  output, ctx, test_output_v1) &&
+                 verify(VARIANT_2,  output, ctx, test_output_v2) &&
+                 verify2(VARIANT_4, output, ctx, test_output_r);
+    }
+#   ifndef XMRIG_NO_AEON
+    else {
+        result = verify(VARIANT_0, output, ctx, test_output_v0_lite) &&
+                 verify(VARIANT_1, output, ctx, test_output_v1_lite);
+    }
+#   endif
+
+
+    for (size_t i = 0; i < count; ++i) {
+        _mm_free(ctx[i]->memory);
+        _mm_free(ctx[i]);
+    }
+
+    return result;
+}
+
+
+#ifndef XMRIG_NO_ASM
+cn_hash_fun cryptonight_hash_asm_fn(enum AlgoVariant av, enum Variant variant, enum Assembly assembly)
+{
+    if (assembly == ASM_AUTO) {
+        assembly = (enum Assembly) cpu_info.assembly;
+    }
+
+    if (assembly == ASM_NONE) {
+        return NULL;
+    }
+
+    return asm_func_map[av][variant][assembly];
+}
+#endif
+
+
+cn_hash_fun cryptonight_hash_fn(enum Algo algorithm, enum AlgoVariant av, enum Variant variant)
+{
+    assert(av > AV_AUTO && av < AV_MAX);
+    assert(variant > VARIANT_AUTO && variant < VARIANT_MAX);
+
+#   ifndef XMRIG_NO_ASM
+    if (algorithm == ALGO_CRYPTONIGHT) {
+        cn_hash_fun fun = cryptonight_hash_asm_fn(av, variant, opt_assembly);
+        if (fun) {
+            return fun;
+        }
+    }
+#   endif
+
+    static const cn_hash_fun func_table[VARIANT_MAX * 4 * 2] = {
+        cryptonight_av1_v0,
+        cryptonight_av2_v0,
+        cryptonight_av3_v0,
+        cryptonight_av4_v0,
+        cryptonight_av1_v1,
+        cryptonight_av2_v1,
+        cryptonight_av3_v1,
+        cryptonight_av4_v1,
+        cryptonight_av1_v2,
+        cryptonight_av2_v2,
+        cryptonight_av3_v2,
+        cryptonight_av4_v2,
+
+        cryptonight_r_av1,
+        cryptonight_r_av2,
+        cryptonight_r_av3,
+        cryptonight_r_av4,
+
+#       ifndef XMRIG_NO_AEON
+        cryptonight_lite_av1_v0,
+        cryptonight_lite_av2_v0,
+        cryptonight_lite_av3_v0,
+        cryptonight_lite_av4_v0,
+        cryptonight_lite_av1_v1,
+        cryptonight_lite_av2_v1,
+        cryptonight_lite_av3_v1,
+        cryptonight_lite_av4_v1,
+        NULL,
+        NULL,
+        NULL,
+        NULL,
+        NULL,
+        NULL,
+        NULL,
+        NULL,
+#       else
+        NULL,
+        NULL,
+        NULL,
+        NULL,
+        NULL,
+        NULL,
+        NULL,
+        NULL,
+        NULL,
+        NULL,
+        NULL,
+        NULL,
+        NULL,
+        NULL,
+        NULL,
+        NULL,
+#       endif
+    };
+
+#   ifndef NDEBUG
+    const size_t index = VARIANT_MAX * 4 * algorithm + 4 * variant + av - 1;
+
+    cn_hash_fun func = func_table[index];
+
+    assert(index < sizeof(func_table) / sizeof(func_table[0]));
+    assert(func != NULL);
+
+    return func;
+#   else
+    return func_table[VARIANT_MAX * 4 * algorithm + 4 * variant + av - 1];
+#   endif
+}
+
+
+bool cryptonight_init(int av)
+{
+    opt_double_hash = av == AV_DOUBLE || av == AV_DOUBLE_SOFT;
+
+#   ifndef XMRIG_NO_ASM
+    asm_func_map[AV_SINGLE][VARIANT_2][ASM_INTEL]     = cryptonight_single_hash_asm_intel;
+    asm_func_map[AV_SINGLE][VARIANT_2][ASM_RYZEN]     = cryptonight_single_hash_asm_intel;
+    asm_func_map[AV_SINGLE][VARIANT_2][ASM_BULLDOZER] = cryptonight_single_hash_asm_bulldozer;
+
+    asm_func_map[AV_DOUBLE][VARIANT_2][ASM_INTEL]     = cryptonight_double_hash_asm;
+    asm_func_map[AV_DOUBLE][VARIANT_2][ASM_RYZEN]     = cryptonight_double_hash_asm;
+    asm_func_map[AV_DOUBLE][VARIANT_2][ASM_BULLDOZER] = cryptonight_double_hash_asm;
+
+    asm_func_map[AV_SINGLE][VARIANT_4][ASM_INTEL]     = cryptonight_r_av1_asm_intel;
+    asm_func_map[AV_SINGLE][VARIANT_4][ASM_RYZEN]     = cryptonight_r_av1_asm_intel;
+    asm_func_map[AV_SINGLE][VARIANT_4][ASM_BULLDOZER] = cryptonight_r_av1_asm_bulldozer;
+
+    asm_func_map[AV_DOUBLE][VARIANT_4][ASM_INTEL]     = cryptonight_r_av2_asm_intel;
+    asm_func_map[AV_DOUBLE][VARIANT_4][ASM_RYZEN]     = cryptonight_r_av2_asm_intel;
+    asm_func_map[AV_DOUBLE][VARIANT_4][ASM_BULLDOZER] = cryptonight_r_av2_asm_bulldozer;
+#   endif
+
+    return self_test();
+}
+
+
+static inline void do_blake_hash(const void* input, size_t len, char* output) {
+    blake256_hash((uint8_t*)output, input, len);
+}
+
+
+static inline void do_groestl_hash(const void* input, size_t len, char* output) {
+    groestl(input, len * 8, (uint8_t*)output);
+}
+
+
+static inline void do_jh_hash(const void* input, size_t len, char* output) {
+    jh_hash(32 * 8, input, 8 * len, (uint8_t*)output);
+}
+
+
+static inline void do_skein_hash(const void* input, size_t len, char* output) {
+    skein_hash(8 * 32, input, 8 * len, (uint8_t*)output);
+}
+
+
+void (* const extra_hashes[4])(const void *, size_t, char *) = {do_blake_hash, do_groestl_hash, do_jh_hash, do_skein_hash};
+
+
+static inline enum Variant cryptonight_variant(uint8_t version)
+{
+    if (opt_variant != VARIANT_AUTO) {
+        return opt_variant;
+    }
+
+    if (opt_algo == ALGO_CRYPTONIGHT_LITE) {
+        return VARIANT_1;
+    }
+
+    if (version >= 10) {
+        return VARIANT_4;
+    }
+
+    if (version >= 8) {
+        return VARIANT_2;
+    }
+
+    return version == 7 ? VARIANT_1 : VARIANT_0;
+}
+
+
+#ifndef BUILD_TEST
+int scanhash_cryptonight(int thr_id, uint32_t *hash, uint8_t *restrict blob, size_t blob_size, uint32_t target, uint32_t max_nonce, unsigned long *restrict hashes_done, struct cryptonight_ctx **restrict ctx) {
+    uint32_t *nonceptr   = (uint32_t*) (((char*) blob) + 39);
+    enum Variant variant = cryptonight_variant(blob[0]);
+
+    do {
+        cryptonight_hash_fn(opt_algo, opt_av, variant)(blob, blob_size, (uint8_t *) hash, ctx);
+
+        (*hashes_done)++;
+
+        if (unlikely(hash[7] < target)) {
+            return 1;
+        }
+
+        (*nonceptr)++;
+    } while (likely(((*nonceptr) < max_nonce && !work_restart[thr_id].restart)));
+
+    return 0;
+}
+
+
+int scanhash_cryptonight_double(int thr_id, uint32_t *hash, uint8_t *restrict blob, size_t blob_size, uint32_t target, uint32_t max_nonce, unsigned long *restrict hashes_done, struct cryptonight_ctx **restrict ctx) {
+    int rc               = 0;
+    uint32_t *nonceptr0  = (uint32_t*) (((char*) blob) + 39);
+    uint32_t *nonceptr1  = (uint32_t*) (((char*) blob) + 39 + blob_size);
+    enum Variant variant = cryptonight_variant(blob[0]);
+
+    do {
+        cryptonight_hash_fn(opt_algo, opt_av, variant)(blob, blob_size, (uint8_t *) hash, ctx);
+        (*hashes_done) += 2;
+
+        if (unlikely(hash[7] < target)) {
+            return rc |= 1;
+        }
+
+        if (unlikely(hash[15] < target)) {
+            return rc |= 2;
+        }
+
+        if (rc) {
+            break;
+        }
+
+        (*nonceptr0)++;
+        (*nonceptr1)++;
+    } while (likely(((*nonceptr0) < max_nonce && !work_restart[thr_id].restart)));
+
+    return rc;
+}
+#endif

algo/cryptonight/cryptonight.h

@@ -0,0 +1,81 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018      Lee Clagett <https://github.com/vtnerd>
+ * Copyright 2018-2019 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2019 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef XMRIG_CRYPTONIGHT_H
+#define XMRIG_CRYPTONIGHT_H
+
+
+#include <stddef.h>
+#include <stdint.h>
+#include <stdbool.h>
+
+
+#include "options.h"
+
+
+#define MEMORY      2097152 /* 2 MiB */
+#define MEMORY_LITE 1048576 /* 1 MiB */
+
+
+#if defined _MSC_VER || defined XMRIG_ARM
+#define ABI_ATTRIBUTE
+#else
+#define ABI_ATTRIBUTE __attribute__((ms_abi))
+#endif
+
+
+struct cryptonight_ctx;
+typedef void(*cn_mainloop_fun_ms_abi)(struct cryptonight_ctx*) ABI_ATTRIBUTE;
+typedef void(*cn_mainloop_double_fun_ms_abi)(struct cryptonight_ctx*, struct cryptonight_ctx*) ABI_ATTRIBUTE;
+
+
+struct cryptonight_ctx {
+    uint8_t state[224] __attribute__((aligned(16)));
+    uint8_t *memory    __attribute__((aligned(16)));
+
+    uint8_t unused[40];
+    const uint32_t *saes_table;
+
+    cn_mainloop_fun_ms_abi generated_code;
+    cn_mainloop_double_fun_ms_abi generated_code_double;
+    uint64_t generated_code_height;
+    uint64_t generated_code_double_height;
+    uint64_t height;
+};
+
+
+typedef void (*cn_hash_fun)(const uint8_t *input, size_t size, uint8_t *output, struct cryptonight_ctx **ctx);
+
+
+extern void (* const extra_hashes[4])(const void *, size_t, char *);
+
+cn_hash_fun cryptonight_hash_fn(enum Algo algorithm, enum AlgoVariant av, enum Variant variant);
+
+bool cryptonight_init(int av);
+int scanhash_cryptonight(int thr_id, uint32_t *hash, uint8_t *blob, size_t blob_size, uint32_t target, uint32_t max_nonce, unsigned long *hashes_done, struct cryptonight_ctx **ctx);
+int scanhash_cryptonight_double(int thr_id, uint32_t *hash, uint8_t *blob, size_t blob_size, uint32_t target, uint32_t max_nonce, unsigned long *hashes_done, struct cryptonight_ctx **ctx);
+
+
+#endif /* XMRIG_CRYPTONIGHT_H */

algo/cryptonight/cryptonight_aesni.h

@@ -0,0 +1,274 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017      fireice-uk  <https://github.com/fireice-uk>
+ * Copyright 2016-2017 XMRig       <support@xmrig.com>
+ *
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef XMRIG_CRYPTONIGHT_AESNI_H
+#define XMRIG_CRYPTONIGHT_AESNI_H
+
+
+#include <x86intrin.h>
+#include <stdint.h>
+
+
+#define aes_genkey_sub(imm8) \
+    __m128i xout1 = _mm_aeskeygenassist_si128(*xout2, (imm8)); \
+    xout1  = _mm_shuffle_epi32(xout1, 0xFF); \
+    *xout0 = sl_xor(*xout0); \
+    *xout0 = _mm_xor_si128(*xout0, xout1); \
+    xout1  = _mm_aeskeygenassist_si128(*xout0, 0x00);\
+    xout1  = _mm_shuffle_epi32(xout1, 0xAA); \
+    *xout2 = sl_xor(*xout2); \
+    *xout2 = _mm_xor_si128(*xout2, xout1); \
+
+
+// This will shift and xor tmp1 into itself as 4 32-bit vals such as
+// sl_xor(a1 a2 a3 a4) = a1 (a2^a1) (a3^a2^a1) (a4^a3^a2^a1)
+static inline __m128i sl_xor(__m128i tmp1)
+{
+    __m128i tmp4;
+    tmp4 = _mm_slli_si128(tmp1, 0x04);
+    tmp1 = _mm_xor_si128(tmp1, tmp4);
+    tmp4 = _mm_slli_si128(tmp4, 0x04);
+    tmp1 = _mm_xor_si128(tmp1, tmp4);
+    tmp4 = _mm_slli_si128(tmp4, 0x04);
+    tmp1 = _mm_xor_si128(tmp1, tmp4);
+    return tmp1;
+}
+
+
+static inline void aes_genkey_sub1(__m128i* xout0, __m128i* xout2)
+{
+   aes_genkey_sub(0x1)
+}
+
+
+static inline void aes_genkey_sub2(__m128i* xout0, __m128i* xout2)
+{
+   aes_genkey_sub(0x2)
+}
+
+
+static inline void aes_genkey_sub4(__m128i* xout0, __m128i* xout2)
+{
+   aes_genkey_sub(0x4)
+}
+
+
+static inline void aes_genkey_sub8(__m128i* xout0, __m128i* xout2)
+{
+   aes_genkey_sub(0x8)
+}
+
+
+static inline void aes_round(__m128i key, __m128i* x0, __m128i* x1, __m128i* x2, __m128i* x3, __m128i* x4, __m128i* x5, __m128i* x6, __m128i* x7)
+{
+    *x0 = _mm_aesenc_si128(*x0, key);
+    *x1 = _mm_aesenc_si128(*x1, key);
+    *x2 = _mm_aesenc_si128(*x2, key);
+    *x3 = _mm_aesenc_si128(*x3, key);
+    *x4 = _mm_aesenc_si128(*x4, key);
+    *x5 = _mm_aesenc_si128(*x5, key);
+    *x6 = _mm_aesenc_si128(*x6, key);
+    *x7 = _mm_aesenc_si128(*x7, key);
+}
+
+
+static inline void aes_genkey(const __m128i* memory, __m128i* k0, __m128i* k1, __m128i* k2, __m128i* k3, __m128i* k4, __m128i* k5, __m128i* k6, __m128i* k7, __m128i* k8, __m128i* k9)
+{
+    __m128i xout0 = _mm_load_si128(memory);
+    __m128i xout2 = _mm_load_si128(memory + 1);
+    *k0 = xout0;
+    *k1 = xout2;
+
+     aes_genkey_sub1(&xout0, &xout2);
+    *k2 = xout0;
+    *k3 = xout2;
+
+     aes_genkey_sub2(&xout0, &xout2);
+    *k4 = xout0;
+    *k5 = xout2;
+
+     aes_genkey_sub4(&xout0, &xout2);
+    *k6 = xout0;
+    *k7 = xout2;
+
+     aes_genkey_sub8(&xout0, &xout2);
+    *k8 = xout0;
+    *k9 = xout2;
+}
+
+
+static inline void cn_explode_scratchpad(const __m128i* input, __m128i* output)
+{
+    // This is more than we have registers, compiler will assign 2 keys on the stack
+    __m128i xin0, xin1, xin2, xin3, xin4, xin5, xin6, xin7;
+    __m128i k0, k1, k2, k3, k4, k5, k6, k7, k8, k9;
+
+    aes_genkey(input, &k0, &k1, &k2, &k3, &k4, &k5, &k6, &k7, &k8, &k9);
+
+    xin0 = _mm_load_si128(input + 4);
+    xin1 = _mm_load_si128(input + 5);
+    xin2 = _mm_load_si128(input + 6);
+    xin3 = _mm_load_si128(input + 7);
+    xin4 = _mm_load_si128(input + 8);
+    xin5 = _mm_load_si128(input + 9);
+    xin6 = _mm_load_si128(input + 10);
+    xin7 = _mm_load_si128(input + 11);
+
+    for (size_t i = 0; __builtin_expect(i < MEMORY / sizeof(__m128i), 1); 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);
+        aes_round(k3, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k4, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k5, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k6, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k7, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k8, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k9, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+
+        _mm_store_si128(output + i + 0, xin0);
+        _mm_store_si128(output + i + 1, xin1);
+        _mm_store_si128(output + i + 2, xin2);
+        _mm_store_si128(output + i + 3, xin3);
+        _mm_store_si128(output + i + 4, xin4);
+        _mm_store_si128(output + i + 5, xin5);
+        _mm_store_si128(output + i + 6, xin6);
+        _mm_store_si128(output + i + 7, xin7);
+    }
+}
+
+
+static inline void cn_implode_scratchpad(const __m128i* input, __m128i* output)
+{
+    // This is more than we have registers, compiler will assign 2 keys on the stack
+    __m128i xout0, xout1, xout2, xout3, xout4, xout5, xout6, xout7;
+    __m128i k0, k1, k2, k3, k4, k5, k6, k7, k8, k9;
+
+    aes_genkey(output + 2, &k0, &k1, &k2, &k3, &k4, &k5, &k6, &k7, &k8, &k9);
+
+    xout0 = _mm_load_si128(output + 4);
+    xout1 = _mm_load_si128(output + 5);
+    xout2 = _mm_load_si128(output + 6);
+    xout3 = _mm_load_si128(output + 7);
+    xout4 = _mm_load_si128(output + 8);
+    xout5 = _mm_load_si128(output + 9);
+    xout6 = _mm_load_si128(output + 10);
+    xout7 = _mm_load_si128(output + 11);
+
+    for (size_t i = 0; __builtin_expect(i < MEMORY / sizeof(__m128i), 1); 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);
+        xout3 = _mm_xor_si128(_mm_load_si128(input + i + 3), xout3);
+        xout4 = _mm_xor_si128(_mm_load_si128(input + i + 4), xout4);
+        xout5 = _mm_xor_si128(_mm_load_si128(input + i + 5), xout5);
+        xout6 = _mm_xor_si128(_mm_load_si128(input + i + 6), xout6);
+        xout7 = _mm_xor_si128(_mm_load_si128(input + i + 7), xout7);
+
+        aes_round(k0, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k1, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k2, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k3, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k4, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k5, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k6, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k7, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k8, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k9, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+    }
+
+    _mm_store_si128(output + 4, xout0);
+    _mm_store_si128(output + 5, xout1);
+    _mm_store_si128(output + 6, xout2);
+    _mm_store_si128(output + 7, xout3);
+    _mm_store_si128(output + 8, xout4);
+    _mm_store_si128(output + 9, xout5);
+    _mm_store_si128(output + 10, xout6);
+    _mm_store_si128(output + 11, xout7);
+}
+
+
+#if defined(__x86_64__)
+#   define EXTRACT64(X) _mm_cvtsi128_si64(X)
+
+static inline uint64_t _umul128(uint64_t a, uint64_t b, uint64_t* hi)
+{
+    unsigned __int128 r = (unsigned __int128) a * (unsigned __int128) b;
+    *hi = r >> 64;
+    return (uint64_t) r;
+}
+#elif defined(__i386__)
+#   define HI32(X) \
+    _mm_srli_si128((X), 4)
+
+
+#   define EXTRACT64(X) \
+    ((uint64_t)(uint32_t)_mm_cvtsi128_si32(X) | \
+    ((uint64_t)(uint32_t)_mm_cvtsi128_si32(HI32(X)) << 32))
+
+static inline uint64_t _umul128(uint64_t multiplier, uint64_t multiplicand, uint64_t *product_hi) {
+    // multiplier   = ab = a * 2^32 + b
+    // multiplicand = cd = c * 2^32 + d
+    // ab * cd = a * c * 2^64 + (a * d + b * c) * 2^32 + b * d
+    uint64_t a = multiplier >> 32;
+    uint64_t b = multiplier & 0xFFFFFFFF;
+    uint64_t c = multiplicand >> 32;
+    uint64_t d = multiplicand & 0xFFFFFFFF;
+
+    //uint64_t ac = a * c;
+    uint64_t ad = a * d;
+    //uint64_t bc = b * c;
+    uint64_t bd = b * d;
+
+    uint64_t adbc = ad + (b * c);
+    uint64_t adbc_carry = adbc < ad ? 1 : 0;
+
+    // multiplier * multiplicand = product_hi * 2^64 + product_lo
+    uint64_t product_lo = bd + (adbc << 32);
+    uint64_t product_lo_carry = product_lo < bd ? 1 : 0;
+    *product_hi = (a * c) + (adbc >> 32) + (adbc_carry << 32) + product_lo_carry;
+
+    return product_lo;
+}
+#endif
+
+
+static inline void cryptonight_monero_tweak(uint64_t* mem_out, __m128i tmp)
+{
+    mem_out[0] = EXTRACT64(tmp);
+
+    tmp = _mm_castps_si128(_mm_movehl_ps(_mm_castsi128_ps(tmp), _mm_castsi128_ps(tmp)));
+    uint64_t vh = EXTRACT64(tmp);
+
+    uint8_t x = vh >> 24;
+    static const uint16_t table = 0x7531;
+    const uint8_t index = (((x >> 3) & 6) | (x & 1)) << 1;
+    vh ^= ((table >> index) & 0x3) << 28;
+
+    mem_out[1] = vh;
+}
+
+
+#endif /* XMRIG_CRYPTONIGHT_AESNI_H */

algo/cryptonight/cryptonight_av1.c

@@ -0,0 +1,261 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017      fireice-uk  <https://github.com/fireice-uk>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018      Lee Clagett <https://github.com/vtnerd>
+ * Copyright 2016-2018 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#include <x86intrin.h>
+#include <string.h>
+
+#include "crypto/c_keccak.h"
+#include "cryptonight.h"
+#include "cryptonight_aesni.h"
+#include "cryptonight_monero.h"
+
+
+void cryptonight_av1_v0(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    keccak(input, size, ctx[0]->state, 200);
+
+    cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+
+    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]);
+
+    uint64_t idx0 = h0[0] ^ h0[4];
+
+    for (size_t i = 0; __builtin_expect(i < 0x80000, 1); i++) {
+        __m128i cx;
+        cx = _mm_load_si128((__m128i *) &l0[idx0 & 0x1FFFF0]);
+        cx = _mm_aesenc_si128(cx, _mm_set_epi64x(ah0, al0));
+
+        _mm_store_si128((__m128i *) &l0[idx0 & 0x1FFFF0], _mm_xor_si128(bx0, cx));
+        idx0 = EXTRACT64(cx);
+        bx0 = cx;
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[1];
+        lo = _umul128(idx0, cl, &hi);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[0] = al0;
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[1] = ah0;
+
+        ah0 ^= ch;
+        al0 ^= cl;
+        idx0 = al0;
+    }
+
+    cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+
+    keccakf(h0, 24);
+    extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
+}
+
+
+void cryptonight_av1_v1(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    if (size < 43) {
+        memset(output, 0, 32);
+        return;
+    }
+
+    keccak(input, size, ctx[0]->state, 200);
+
+    VARIANT1_INIT(0);
+
+    cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+
+    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]);
+
+    uint64_t idx0 = h0[0] ^ h0[4];
+
+    for (size_t i = 0; __builtin_expect(i < 0x80000, 1); i++) {
+        __m128i cx;
+        cx = _mm_load_si128((__m128i *) &l0[idx0 & 0x1FFFF0]);
+        cx = _mm_aesenc_si128(cx, _mm_set_epi64x(ah0, al0));
+
+        cryptonight_monero_tweak((uint64_t*)&l0[idx0 & 0x1FFFF0], _mm_xor_si128(bx0, cx));
+
+        idx0 = EXTRACT64(cx);
+        bx0 = cx;
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[1];
+        lo = _umul128(idx0, cl, &hi);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[0] = al0;
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[1] = ah0 ^ tweak1_2_0;
+
+        ah0 ^= ch;
+        al0 ^= cl;
+        idx0 = al0;
+    }
+
+    cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+
+    keccakf(h0, 24);
+    extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
+}
+
+
+void cryptonight_av1_v2(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    keccak(input, size, ctx[0]->state, 200);
+
+    cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+
+    VARIANT2_INIT(0);
+    VARIANT2_SET_ROUNDING_MODE();
+
+    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 idx0 = al0;
+
+    for (size_t i = 0; __builtin_expect(i < 0x80000, 1); i++) {
+        __m128i cx        = _mm_load_si128((__m128i *) &l0[idx0 & 0x1FFFF0]);
+        const __m128i ax0 = _mm_set_epi64x(ah0, al0);
+
+        cx = _mm_aesenc_si128(cx, ax0);
+
+        VARIANT2_SHUFFLE(l0, idx0 & 0x1FFFF0, ax0, bx0, bx1);
+        _mm_store_si128((__m128i *) &l0[idx0 & 0x1FFFF0], _mm_xor_si128(bx0, cx));
+
+        idx0 = _mm_cvtsi128_si64(cx);
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[1];
+
+        VARIANT2_INTEGER_MATH(0, cl, cx);
+        lo = _umul128(idx0, cl, &hi);
+        VARIANT2_SHUFFLE2(l0, idx0 & 0x1FFFF0, ax0, bx0, bx1, hi, lo);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[0] = al0;
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[1] = ah0;
+
+        al0 ^= cl;
+        ah0 ^= ch;
+        idx0 = al0;
+
+        bx1 = bx0;
+        bx0 = cx;
+    }
+
+    cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+
+    keccakf(h0, 24);
+    extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
+}
+
+
+#ifndef XMRIG_NO_ASM
+extern void cnv2_mainloop_ivybridge_asm(struct cryptonight_ctx *ctx);
+extern void cnv2_mainloop_ryzen_asm(struct cryptonight_ctx *ctx);
+extern void cnv2_mainloop_bulldozer_asm(struct cryptonight_ctx *ctx);
+extern void cnv2_double_mainloop_sandybridge_asm(struct cryptonight_ctx* ctx0, struct cryptonight_ctx* ctx1);
+
+
+void cryptonight_single_hash_asm_intel(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    keccak(input, size, ctx[0]->state, 200);
+    cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+
+    cnv2_mainloop_ivybridge_asm(ctx[0]);
+
+    cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+    keccakf((uint64_t*) ctx[0]->state, 24);
+    extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
+}
+
+
+void cryptonight_single_hash_asm_ryzen(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    keccak(input, size, ctx[0]->state, 200);
+    cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+
+    cnv2_mainloop_ryzen_asm(ctx[0]);
+
+    cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+    keccakf((uint64_t*) ctx[0]->state, 24);
+    extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
+}
+
+
+void cryptonight_single_hash_asm_bulldozer(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    keccak(input, size, ctx[0]->state, 200);
+    cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+
+    cnv2_mainloop_bulldozer_asm(ctx[0]);
+
+    cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+    keccakf((uint64_t*) ctx[0]->state, 24);
+    extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
+}
+
+
+void cryptonight_double_hash_asm(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    keccak(input,        size, ctx[0]->state, 200);
+    keccak(input + size, size, ctx[1]->state, 200);
+
+    cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+    cn_explode_scratchpad((__m128i*) ctx[1]->state, (__m128i*) ctx[1]->memory);
+
+    cnv2_double_mainloop_sandybridge_asm(ctx[0], ctx[1]);
+
+    cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+    cn_implode_scratchpad((__m128i*) ctx[1]->memory, (__m128i*) ctx[1]->state);
+
+    keccakf((uint64_t*) ctx[0]->state, 24);
+    keccakf((uint64_t*) 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);
+}
+#endif

algo/cryptonight/cryptonight_av2.c

@@ -0,0 +1,304 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017      fireice-uk  <https://github.com/fireice-uk>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018      Lee Clagett <https://github.com/vtnerd>
+ * Copyright 2016-2018 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#include <x86intrin.h>
+#include <string.h>
+
+#include "crypto/c_keccak.h"
+#include "cryptonight.h"
+#include "cryptonight_aesni.h"
+#include "cryptonight_monero.h"
+
+
+void cryptonight_av2_v0(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    keccak(input,        size, ctx[0]->state, 200);
+    keccak(input + size, size, ctx[1]->state, 200);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    const uint8_t* l1 = ctx[1]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+    uint64_t* h1 = (uint64_t*) ctx[1]->state;
+
+    cn_explode_scratchpad((__m128i*) h0, (__m128i*) l0);
+    cn_explode_scratchpad((__m128i*) h1, (__m128i*) l1);
+
+    uint64_t al0 = h0[0] ^ h0[4];
+    uint64_t al1 = h1[0] ^ h1[4];
+    uint64_t ah0 = h0[1] ^ h0[5];
+    uint64_t ah1 = h1[1] ^ h1[5];
+
+    __m128i bx0 = _mm_set_epi64x(h0[3] ^ h0[7], h0[2] ^ h0[6]);
+    __m128i bx1 = _mm_set_epi64x(h1[3] ^ h1[7], h1[2] ^ h1[6]);
+
+    uint64_t idx0 = h0[0] ^ h0[4];
+    uint64_t idx1 = h1[0] ^ h1[4];
+
+    for (size_t i = 0; __builtin_expect(i < 0x80000, 1); i++) {
+        __m128i cx0 = _mm_load_si128((__m128i *) &l0[idx0 & 0x1FFFF0]);
+        __m128i cx1 = _mm_load_si128((__m128i *) &l1[idx1 & 0x1FFFF0]);
+
+        cx0 = _mm_aesenc_si128(cx0, _mm_set_epi64x(ah0, al0));
+        cx1 = _mm_aesenc_si128(cx1, _mm_set_epi64x(ah1, al1));
+
+        _mm_store_si128((__m128i *) &l0[idx0 & 0x1FFFF0], _mm_xor_si128(bx0, cx0));
+        _mm_store_si128((__m128i *) &l1[idx1 & 0x1FFFF0], _mm_xor_si128(bx1, cx1));
+
+        idx0 = EXTRACT64(cx0);
+        idx1 = EXTRACT64(cx1);
+
+        bx0 = cx0;
+        bx1 = cx1;
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[1];
+        lo = _umul128(idx0, cl, &hi);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*) &l0[idx0 & 0x1FFFF0])[0] = al0;
+        ((uint64_t*) &l0[idx0 & 0x1FFFF0])[1] = ah0;
+
+        ah0 ^= ch;
+        al0 ^= cl;
+        idx0 = al0;
+
+        cl = ((uint64_t*) &l1[idx1 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l1[idx1 & 0x1FFFF0])[1];
+        lo = _umul128(idx1, cl, &hi);
+
+        al1 += hi;
+        ah1 += lo;
+
+        ((uint64_t*) &l1[idx1 & 0x1FFFF0])[0] = al1;
+        ((uint64_t*) &l1[idx1 & 0x1FFFF0])[1] = ah1;
+
+        ah1 ^= ch;
+        al1 ^= cl;
+        idx1 = al1;
+    }
+
+    cn_implode_scratchpad((__m128i*) l0, (__m128i*) h0);
+    cn_implode_scratchpad((__m128i*) l1, (__m128i*) h1);
+
+    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);
+}
+
+
+void cryptonight_av2_v1(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    if (size < 43) {
+        memset(output, 0, 64);
+        return;
+    }
+
+    keccak(input,        size, ctx[0]->state, 200);
+    keccak(input + size, size, ctx[1]->state, 200);
+
+    VARIANT1_INIT(0);
+    VARIANT1_INIT(1);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    const uint8_t* l1 = ctx[1]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+    uint64_t* h1 = (uint64_t*) ctx[1]->state;
+
+    cn_explode_scratchpad((__m128i*) h0, (__m128i*) l0);
+    cn_explode_scratchpad((__m128i*) h1, (__m128i*) l1);
+
+    uint64_t al0 = h0[0] ^ h0[4];
+    uint64_t al1 = h1[0] ^ h1[4];
+    uint64_t ah0 = h0[1] ^ h0[5];
+    uint64_t ah1 = h1[1] ^ h1[5];
+
+    __m128i bx0 = _mm_set_epi64x(h0[3] ^ h0[7], h0[2] ^ h0[6]);
+    __m128i bx1 = _mm_set_epi64x(h1[3] ^ h1[7], h1[2] ^ h1[6]);
+
+    uint64_t idx0 = h0[0] ^ h0[4];
+    uint64_t idx1 = h1[0] ^ h1[4];
+
+    for (size_t i = 0; __builtin_expect(i < 0x80000, 1); i++) {
+        __m128i cx0 = _mm_load_si128((__m128i *) &l0[idx0 & 0x1FFFF0]);
+        __m128i cx1 = _mm_load_si128((__m128i *) &l1[idx1 & 0x1FFFF0]);
+
+        cx0 = _mm_aesenc_si128(cx0, _mm_set_epi64x(ah0, al0));
+        cx1 = _mm_aesenc_si128(cx1, _mm_set_epi64x(ah1, al1));
+
+        cryptonight_monero_tweak((uint64_t*)&l0[idx0 & 0x1FFFF0], _mm_xor_si128(bx0, cx0));
+        cryptonight_monero_tweak((uint64_t*)&l1[idx1 & 0x1FFFF0], _mm_xor_si128(bx1, cx1));
+
+        idx0 = EXTRACT64(cx0);
+        idx1 = EXTRACT64(cx1);
+
+        bx0 = cx0;
+        bx1 = cx1;
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[1];
+        lo = _umul128(idx0, cl, &hi);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*) &l0[idx0 & 0x1FFFF0])[0] = al0;
+        ((uint64_t*) &l0[idx0 & 0x1FFFF0])[1] = ah0 ^ tweak1_2_0;
+
+        ah0 ^= ch;
+        al0 ^= cl;
+        idx0 = al0;
+
+        cl = ((uint64_t*) &l1[idx1 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l1[idx1 & 0x1FFFF0])[1];
+        lo = _umul128(idx1, cl, &hi);
+
+        al1 += hi;
+        ah1 += lo;
+
+        ((uint64_t*) &l1[idx1 & 0x1FFFF0])[0] = al1;
+        ((uint64_t*) &l1[idx1 & 0x1FFFF0])[1] = ah1 ^ tweak1_2_1;
+
+        ah1 ^= ch;
+        al1 ^= cl;
+        idx1 = al1;
+    }
+
+    cn_implode_scratchpad((__m128i*) l0, (__m128i*) h0);
+    cn_implode_scratchpad((__m128i*) l1, (__m128i*) h1);
+
+    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);
+}
+
+
+void cryptonight_av2_v2(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    keccak(input,        size, ctx[0]->state, 200);
+    keccak(input + size, size, ctx[1]->state, 200);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    const uint8_t* l1 = ctx[1]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+    uint64_t* h1 = (uint64_t*) ctx[1]->state;
+
+    VARIANT2_INIT(0);
+    VARIANT2_INIT(1);
+    VARIANT2_SET_ROUNDING_MODE();
+
+    cn_explode_scratchpad((__m128i*) h0, (__m128i*) l0);
+    cn_explode_scratchpad((__m128i*) h1, (__m128i*) l1);
+
+    uint64_t al0 = h0[0] ^ h0[4];
+    uint64_t al1 = h1[0] ^ h1[4];
+    uint64_t ah0 = h0[1] ^ h0[5];
+    uint64_t ah1 = h1[1] ^ h1[5];
+
+    __m128i bx00 = _mm_set_epi64x(h0[3] ^ h0[7], h0[2] ^ h0[6]);
+    __m128i bx01 = _mm_set_epi64x(h0[9] ^ h0[11], h0[8] ^ h0[10]);
+    __m128i bx10 = _mm_set_epi64x(h1[3] ^ h1[7], h1[2] ^ h1[6]);
+    __m128i bx11 = _mm_set_epi64x(h1[9] ^ h1[11], h1[8] ^ h1[10]);
+
+    uint64_t idx0 = al0;
+    uint64_t idx1 = al1;
+
+    for (size_t i = 0; __builtin_expect(i < 0x80000, 1); i++) {
+        __m128i cx0       = _mm_load_si128((__m128i *) &l0[idx0 & 0x1FFFF0]);
+        __m128i cx1       = _mm_load_si128((__m128i *) &l1[idx1 & 0x1FFFF0]);
+
+        const __m128i ax0 = _mm_set_epi64x(ah0, al0);
+        const __m128i ax1 = _mm_set_epi64x(ah1, al1);
+
+        cx0 = _mm_aesenc_si128(cx0, ax0);
+        cx1 = _mm_aesenc_si128(cx1, ax1);
+
+        VARIANT2_SHUFFLE(l0, idx0 & 0x1FFFF0, ax0, bx00, bx01);
+        _mm_store_si128((__m128i *) &l0[idx0 & 0x1FFFF0], _mm_xor_si128(bx00, cx0));
+
+        VARIANT2_SHUFFLE(l1, idx1 & 0x1FFFF0, ax1, bx10, bx11);
+        _mm_store_si128((__m128i *) &l1[idx1 & 0x1FFFF0], _mm_xor_si128(bx10, cx1));
+
+        idx0 = _mm_cvtsi128_si64(cx0);
+        idx1 = _mm_cvtsi128_si64(cx1);
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[1];
+
+        VARIANT2_INTEGER_MATH(0, cl, cx0);
+        lo = _umul128(idx0, cl, &hi);
+        VARIANT2_SHUFFLE2(l0, idx0 & 0x1FFFF0, ax0, bx00, bx01, hi, lo);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[0] = al0;
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[1] = ah0;
+
+        al0 ^= cl;
+        ah0 ^= ch;
+        idx0 = al0;
+
+        cl = ((uint64_t*) &l1[idx1 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l1[idx1 & 0x1FFFF0])[1];
+
+        VARIANT2_INTEGER_MATH(1, cl, cx1);
+        lo = _umul128(idx1, cl, &hi);
+        VARIANT2_SHUFFLE2(l1, idx1 & 0x1FFFF0, ax1, bx10, bx11, hi, lo);
+
+        al1 += hi;
+        ah1 += lo;
+
+        ((uint64_t*)&l1[idx1 & 0x1FFFF0])[0] = al1;
+        ((uint64_t*)&l1[idx1 & 0x1FFFF0])[1] = ah1;
+
+        al1 ^= cl;
+        ah1 ^= ch;
+        idx1 = al1;
+
+        bx01 = bx00;
+        bx11 = bx10;
+
+        bx00 = cx0;
+        bx10 = cx1;
+    }
+
+    cn_implode_scratchpad((__m128i*) l0, (__m128i*) h0);
+    cn_implode_scratchpad((__m128i*) l1, (__m128i*) h1);
+
+    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);
+}

algo/cryptonight/cryptonight_av3.c

@@ -0,0 +1,193 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017      fireice-uk  <https://github.com/fireice-uk>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018      Lee Clagett <https://github.com/vtnerd>
+ * Copyright 2016-2018 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#include <x86intrin.h>
+#include <string.h>
+
+#include "crypto/c_keccak.h"
+#include "cryptonight.h"
+#include "cryptonight_monero.h"
+#include "cryptonight_softaes.h"
+
+
+void cryptonight_av3_v0(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    keccak(input, size, ctx[0]->state, 200);
+
+    cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+
+    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]);
+
+    uint64_t idx0 = h0[0] ^ h0[4];
+
+    for (size_t i = 0; __builtin_expect(i < 0x80000, 1); i++) {
+        __m128i cx;
+        cx = _mm_load_si128((__m128i *) &l0[idx0 & 0x1FFFF0]);
+        cx = soft_aesenc(cx, _mm_set_epi64x(ah0, al0));
+
+        _mm_store_si128((__m128i *) &l0[idx0 & 0x1FFFF0], _mm_xor_si128(bx0, cx));
+        idx0 = EXTRACT64(cx);
+        bx0 = cx;
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[1];
+        lo = _umul128(idx0, cl, &hi);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[0] = al0;
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[1] = ah0;
+
+        ah0 ^= ch;
+        al0 ^= cl;
+        idx0 = al0;
+    }
+
+    cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+
+    keccakf(h0, 24);
+    extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
+}
+
+
+void cryptonight_av3_v1(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    if (size < 43) {
+        memset(output, 0, 32);
+        return;
+    }
+
+    keccak(input, size, ctx[0]->state, 200);
+
+    VARIANT1_INIT(0);
+
+    cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+
+    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]);
+
+    uint64_t idx0 = h0[0] ^ h0[4];
+
+    for (size_t i = 0; __builtin_expect(i < 0x80000, 1); i++) {
+        __m128i cx;
+        cx = _mm_load_si128((__m128i *) &l0[idx0 & 0x1FFFF0]);
+        cx = soft_aesenc(cx, _mm_set_epi64x(ah0, al0));
+
+        cryptonight_monero_tweak((uint64_t*)&l0[idx0 & 0x1FFFF0], _mm_xor_si128(bx0, cx));
+
+        idx0 = EXTRACT64(cx);
+        bx0 = cx;
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[1];
+        lo = _umul128(idx0, cl, &hi);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[0] = al0;
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[1] = ah0 ^ tweak1_2_0;
+
+        ah0 ^= ch;
+        al0 ^= cl;
+        idx0 = al0;
+    }
+
+    cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+
+    keccakf(h0, 24);
+    extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
+}
+
+
+void cryptonight_av3_v2(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    keccak(input, size, ctx[0]->state, 200);
+
+    cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+
+    VARIANT2_INIT(0);
+    VARIANT2_SET_ROUNDING_MODE();
+
+    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 idx0 = al0;
+
+    for (size_t i = 0; __builtin_expect(i < 0x80000, 1); i++) {
+        __m128i cx        = _mm_load_si128((__m128i *) &l0[idx0 & 0x1FFFF0]);
+        const __m128i ax0 = _mm_set_epi64x(ah0, al0);
+
+        cx = soft_aesenc(cx, ax0);
+
+        VARIANT2_SHUFFLE(l0, idx0 & 0x1FFFF0, ax0, bx0, bx1);
+        _mm_store_si128((__m128i *) &l0[idx0 & 0x1FFFF0], _mm_xor_si128(bx0, cx));
+
+        idx0 = _mm_cvtsi128_si64(cx);
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[1];
+
+        VARIANT2_INTEGER_MATH(0, cl, cx);
+        lo = _umul128(idx0, cl, &hi);
+        VARIANT2_SHUFFLE2(l0, idx0 & 0x1FFFF0, ax0, bx0, bx1, hi, lo);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[0] = al0;
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[1] = ah0;
+
+        al0 ^= cl;
+        ah0 ^= ch;
+        idx0 = al0;
+
+        bx1 = bx0;
+        bx0 = cx;
+    }
+
+    cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+
+    keccakf(h0, 24);
+    extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
+}

algo/cryptonight/cryptonight_av4.c

@@ -0,0 +1,304 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017      fireice-uk  <https://github.com/fireice-uk>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018      Lee Clagett <https://github.com/vtnerd>
+ * Copyright 2016-2018 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#include <x86intrin.h>
+#include <string.h>
+
+#include "crypto/c_keccak.h"
+#include "cryptonight.h"
+#include "cryptonight_monero.h"
+#include "cryptonight_softaes.h"
+
+
+void cryptonight_av4_v0(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    keccak(input,        size, ctx[0]->state, 200);
+    keccak(input + size, size, ctx[1]->state, 200);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    const uint8_t* l1 = ctx[1]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+    uint64_t* h1 = (uint64_t*) ctx[1]->state;
+
+    cn_explode_scratchpad((__m128i*) h0, (__m128i*) l0);
+    cn_explode_scratchpad((__m128i*) h1, (__m128i*) l1);
+
+    uint64_t al0 = h0[0] ^ h0[4];
+    uint64_t al1 = h1[0] ^ h1[4];
+    uint64_t ah0 = h0[1] ^ h0[5];
+    uint64_t ah1 = h1[1] ^ h1[5];
+
+    __m128i bx0 = _mm_set_epi64x(h0[3] ^ h0[7], h0[2] ^ h0[6]);
+    __m128i bx1 = _mm_set_epi64x(h1[3] ^ h1[7], h1[2] ^ h1[6]);
+
+    uint64_t idx0 = h0[0] ^ h0[4];
+    uint64_t idx1 = h1[0] ^ h1[4];
+
+    for (size_t i = 0; __builtin_expect(i < 0x80000, 1); i++) {
+        __m128i cx0 = _mm_load_si128((__m128i *) &l0[idx0 & 0x1FFFF0]);
+        __m128i cx1 = _mm_load_si128((__m128i *) &l1[idx1 & 0x1FFFF0]);
+
+        cx0 = soft_aesenc(cx0, _mm_set_epi64x(ah0, al0));
+        cx1 = soft_aesenc(cx1, _mm_set_epi64x(ah1, al1));
+
+        _mm_store_si128((__m128i *) &l0[idx0 & 0x1FFFF0], _mm_xor_si128(bx0, cx0));
+        _mm_store_si128((__m128i *) &l1[idx1 & 0x1FFFF0], _mm_xor_si128(bx1, cx1));
+
+        idx0 = EXTRACT64(cx0);
+        idx1 = EXTRACT64(cx1);
+
+        bx0 = cx0;
+        bx1 = cx1;
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[1];
+        lo = _umul128(idx0, cl, &hi);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*) &l0[idx0 & 0x1FFFF0])[0] = al0;
+        ((uint64_t*) &l0[idx0 & 0x1FFFF0])[1] = ah0;
+
+        ah0 ^= ch;
+        al0 ^= cl;
+        idx0 = al0;
+
+        cl = ((uint64_t*) &l1[idx1 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l1[idx1 & 0x1FFFF0])[1];
+        lo = _umul128(idx1, cl, &hi);
+
+        al1 += hi;
+        ah1 += lo;
+
+        ((uint64_t*) &l1[idx1 & 0x1FFFF0])[0] = al1;
+        ((uint64_t*) &l1[idx1 & 0x1FFFF0])[1] = ah1;
+
+        ah1 ^= ch;
+        al1 ^= cl;
+        idx1 = al1;
+    }
+
+    cn_implode_scratchpad((__m128i*) l0, (__m128i*) h0);
+    cn_implode_scratchpad((__m128i*) l1, (__m128i*) h1);
+
+    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);
+}
+
+
+void cryptonight_av4_v1(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    if (size < 43) {
+        memset(output, 0, 64);
+        return;
+    }
+
+    keccak(input,        size, ctx[0]->state, 200);
+    keccak(input + size, size, ctx[1]->state, 200);
+
+    VARIANT1_INIT(0);
+    VARIANT1_INIT(1);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    const uint8_t* l1 = ctx[1]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+    uint64_t* h1 = (uint64_t*) ctx[1]->state;
+
+    cn_explode_scratchpad((__m128i*) h0, (__m128i*) l0);
+    cn_explode_scratchpad((__m128i*) h1, (__m128i*) l1);
+
+    uint64_t al0 = h0[0] ^ h0[4];
+    uint64_t al1 = h1[0] ^ h1[4];
+    uint64_t ah0 = h0[1] ^ h0[5];
+    uint64_t ah1 = h1[1] ^ h1[5];
+
+    __m128i bx0 = _mm_set_epi64x(h0[3] ^ h0[7], h0[2] ^ h0[6]);
+    __m128i bx1 = _mm_set_epi64x(h1[3] ^ h1[7], h1[2] ^ h1[6]);
+
+    uint64_t idx0 = h0[0] ^ h0[4];
+    uint64_t idx1 = h1[0] ^ h1[4];
+
+    for (size_t i = 0; __builtin_expect(i < 0x80000, 1); i++) {
+        __m128i cx0 = _mm_load_si128((__m128i *) &l0[idx0 & 0x1FFFF0]);
+        __m128i cx1 = _mm_load_si128((__m128i *) &l1[idx1 & 0x1FFFF0]);
+
+        cx0 = soft_aesenc(cx0, _mm_set_epi64x(ah0, al0));
+        cx1 = soft_aesenc(cx1, _mm_set_epi64x(ah1, al1));
+
+        cryptonight_monero_tweak((uint64_t*)&l0[idx0 & 0x1FFFF0], _mm_xor_si128(bx0, cx0));
+        cryptonight_monero_tweak((uint64_t*)&l1[idx1 & 0x1FFFF0], _mm_xor_si128(bx1, cx1));
+
+        idx0 = EXTRACT64(cx0);
+        idx1 = EXTRACT64(cx1);
+
+        bx0 = cx0;
+        bx1 = cx1;
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[1];
+        lo = _umul128(idx0, cl, &hi);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*) &l0[idx0 & 0x1FFFF0])[0] = al0;
+        ((uint64_t*) &l0[idx0 & 0x1FFFF0])[1] = ah0 ^ tweak1_2_0;
+
+        ah0 ^= ch;
+        al0 ^= cl;
+        idx0 = al0;
+
+        cl = ((uint64_t*) &l1[idx1 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l1[idx1 & 0x1FFFF0])[1];
+        lo = _umul128(idx1, cl, &hi);
+
+        al1 += hi;
+        ah1 += lo;
+
+        ((uint64_t*) &l1[idx1 & 0x1FFFF0])[0] = al1;
+        ((uint64_t*) &l1[idx1 & 0x1FFFF0])[1] = ah1 ^ tweak1_2_1;
+
+        ah1 ^= ch;
+        al1 ^= cl;
+        idx1 = al1;
+    }
+
+    cn_implode_scratchpad((__m128i*) l0, (__m128i*) h0);
+    cn_implode_scratchpad((__m128i*) l1, (__m128i*) h1);
+
+    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);
+}
+
+
+void cryptonight_av4_v2(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    keccak(input,        size, ctx[0]->state, 200);
+    keccak(input + size, size, ctx[1]->state, 200);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    const uint8_t* l1 = ctx[1]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+    uint64_t* h1 = (uint64_t*) ctx[1]->state;
+
+    VARIANT2_INIT(0);
+    VARIANT2_INIT(1);
+    VARIANT2_SET_ROUNDING_MODE();
+
+    cn_explode_scratchpad((__m128i*) h0, (__m128i*) l0);
+    cn_explode_scratchpad((__m128i*) h1, (__m128i*) l1);
+
+    uint64_t al0 = h0[0] ^ h0[4];
+    uint64_t al1 = h1[0] ^ h1[4];
+    uint64_t ah0 = h0[1] ^ h0[5];
+    uint64_t ah1 = h1[1] ^ h1[5];
+
+    __m128i bx00 = _mm_set_epi64x(h0[3] ^ h0[7], h0[2] ^ h0[6]);
+    __m128i bx01 = _mm_set_epi64x(h0[9] ^ h0[11], h0[8] ^ h0[10]);
+    __m128i bx10 = _mm_set_epi64x(h1[3] ^ h1[7], h1[2] ^ h1[6]);
+    __m128i bx11 = _mm_set_epi64x(h1[9] ^ h1[11], h1[8] ^ h1[10]);
+
+    uint64_t idx0 = al0;
+    uint64_t idx1 = al1;
+
+    for (size_t i = 0; __builtin_expect(i < 0x80000, 1); i++) {
+        __m128i cx0       = _mm_load_si128((__m128i *) &l0[idx0 & 0x1FFFF0]);
+        __m128i cx1       = _mm_load_si128((__m128i *) &l1[idx1 & 0x1FFFF0]);
+
+        const __m128i ax0 = _mm_set_epi64x(ah0, al0);
+        const __m128i ax1 = _mm_set_epi64x(ah1, al1);
+
+        cx0 = soft_aesenc(cx0, ax0);
+        cx1 = soft_aesenc(cx1, ax1);
+
+        VARIANT2_SHUFFLE(l0, idx0 & 0x1FFFF0, ax0, bx00, bx01);
+        _mm_store_si128((__m128i *) &l0[idx0 & 0x1FFFF0], _mm_xor_si128(bx00, cx0));
+
+        VARIANT2_SHUFFLE(l1, idx1 & 0x1FFFF0, ax1, bx10, bx11);
+        _mm_store_si128((__m128i *) &l1[idx1 & 0x1FFFF0], _mm_xor_si128(bx10, cx1));
+
+        idx0 = _mm_cvtsi128_si64(cx0);
+        idx1 = _mm_cvtsi128_si64(cx1);
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[1];
+
+        VARIANT2_INTEGER_MATH(0, cl, cx0);
+        lo = _umul128(idx0, cl, &hi);
+        VARIANT2_SHUFFLE2(l0, idx0 & 0x1FFFF0, ax0, bx00, bx01, hi, lo);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[0] = al0;
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[1] = ah0;
+
+        al0 ^= cl;
+        ah0 ^= ch;
+        idx0 = al0;
+
+        cl = ((uint64_t*) &l1[idx1 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l1[idx1 & 0x1FFFF0])[1];
+
+        VARIANT2_INTEGER_MATH(1, cl, cx1);
+        lo = _umul128(idx1, cl, &hi);
+        VARIANT2_SHUFFLE2(l1, idx1 & 0x1FFFF0, ax1, bx10, bx11, hi, lo);
+
+        al1 += hi;
+        ah1 += lo;
+
+        ((uint64_t*)&l1[idx1 & 0x1FFFF0])[0] = al1;
+        ((uint64_t*)&l1[idx1 & 0x1FFFF0])[1] = ah1;
+
+        al1 ^= cl;
+        ah1 ^= ch;
+        idx1 = al1;
+
+        bx01 = bx00;
+        bx11 = bx10;
+
+        bx00 = cx0;
+        bx10 = cx1;
+    }
+
+    cn_implode_scratchpad((__m128i*) l0, (__m128i*) h0);
+    cn_implode_scratchpad((__m128i*) l1, (__m128i*) h1);
+
+    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);
+}

algo/cryptonight/cryptonight_monero.h

@@ -0,0 +1,150 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018      Lee Clagett <https://github.com/vtnerd>
+ * Copyright 2018-2019 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2019 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef XMRIG_CRYPTONIGHT_MONERO_H
+#define XMRIG_CRYPTONIGHT_MONERO_H
+
+
+#include <fenv.h>
+#include <math.h>
+#include <stdint.h>
+#include <x86intrin.h>
+
+
+static inline __m128i int_sqrt_v2(const uint64_t n0)
+{
+    __m128d x = _mm_castsi128_pd(_mm_add_epi64(_mm_cvtsi64_si128(n0 >> 12), _mm_set_epi64x(0, 1023ULL << 52)));
+    x = _mm_sqrt_sd(_mm_setzero_pd(), x);
+    uint64_t r = (uint64_t)(_mm_cvtsi128_si64(_mm_castpd_si128(x)));
+
+    const uint64_t s = r >> 20;
+    r >>= 19;
+
+    uint64_t x2 = (s - (1022ULL << 32)) * (r - s - (1022ULL << 32) + 1);
+#   if (defined(_MSC_VER) || __GNUC__ > 7 || (__GNUC__ == 7 && __GNUC_MINOR__ > 1)) && (defined(__x86_64__) || defined(_M_AMD64))
+    _addcarry_u64(_subborrow_u64(0, x2, n0, (unsigned long long int*)&x2), r, 0, (unsigned long long int*)&r);
+#   else
+    if (x2 < n0) ++r;
+#   endif
+
+    return _mm_cvtsi64_si128(r);
+}
+
+
+#   define VARIANT1_INIT(part) \
+    uint64_t tweak1_2_##part = (*(const uint64_t*)(input + 35 + part * size) ^ \
+                               *((const uint64_t*)(ctx[part]->state) + 24)); \
+
+#   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]);
+
+#ifdef _MSC_VER
+#   define VARIANT2_SET_ROUNDING_MODE() { _control87(RC_DOWN, MCW_RC); }
+#else
+#   define VARIANT2_SET_ROUNDING_MODE() { fesetround(FE_DOWNWARD); }
+#endif
+
+#   define VARIANT2_INTEGER_MATH(part, cl, cx) \
+    { \
+        const uint64_t sqrt_result = (uint64_t)(_mm_cvtsi128_si64(sqrt_result_xmm_##part)); \
+        const uint64_t cx_0 = _mm_cvtsi128_si64(cx); \
+        cl ^= (uint64_t)(_mm_cvtsi128_si64(division_result_xmm_##part)) ^ (sqrt_result << 32); \
+        const uint32_t d = (uint32_t)(cx_0 + (sqrt_result << 1)) | 0x80000001UL; \
+        const uint64_t cx_1 = _mm_cvtsi128_si64(_mm_srli_si128(cx, 8)); \
+        const uint64_t division_result = (uint32_t)(cx_1 / d) + ((cx_1 % d) << 32); \
+        division_result_xmm_##part = _mm_cvtsi64_si128((int64_t)(division_result)); \
+        sqrt_result_xmm_##part = int_sqrt_v2(cx_0 + division_result); \
+    }
+
+#   define VARIANT2_SHUFFLE(base_ptr, offset, _a, _b, _b1) \
+    { \
+        const __m128i chunk1 = _mm_load_si128((__m128i *)((base_ptr) + ((offset) ^ 0x10))); \
+        const __m128i chunk2 = _mm_load_si128((__m128i *)((base_ptr) + ((offset) ^ 0x20))); \
+        const __m128i chunk3 = _mm_load_si128((__m128i *)((base_ptr) + ((offset) ^ 0x30))); \
+        _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)); \
+    }
+
+#   define VARIANT4_SHUFFLE(base_ptr, offset, _a, _b, _b1, _c) \
+    { \
+        const __m128i chunk1 = _mm_load_si128((__m128i *)((base_ptr) + ((offset) ^ 0x10))); \
+        const __m128i chunk2 = _mm_load_si128((__m128i *)((base_ptr) + ((offset) ^ 0x20))); \
+        const __m128i chunk3 = _mm_load_si128((__m128i *)((base_ptr) + ((offset) ^ 0x30))); \
+        _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)); \
+        _c = _mm_xor_si128(_mm_xor_si128(_c, chunk3), _mm_xor_si128(chunk1, chunk2)); \
+    }
+
+#   define VARIANT2_SHUFFLE2(base_ptr, offset, _a, _b, _b1, hi, lo) \
+    { \
+        const __m128i chunk1 = _mm_xor_si128(_mm_load_si128((__m128i *)((base_ptr) + ((offset) ^ 0x10))), _mm_set_epi64x(lo, hi)); \
+        const __m128i chunk2 = _mm_load_si128((__m128i *)((base_ptr) + ((offset) ^ 0x20))); \
+        hi ^= ((uint64_t*)((base_ptr) + ((offset) ^ 0x20)))[0]; \
+        lo ^= ((uint64_t*)((base_ptr) + ((offset) ^ 0x20)))[1]; \
+        const __m128i chunk3 = _mm_load_si128((__m128i *)((base_ptr) + ((offset) ^ 0x30))); \
+        _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)); \
+    }
+
+
+#ifndef NOINLINE
+#ifdef __GNUC__
+#define NOINLINE __attribute__ ((noinline))
+#elif _MSC_VER
+#define NOINLINE __declspec(noinline)
+#else
+#define NOINLINE
+#endif
+#endif
+
+#include "variant4_random_math.h"
+
+#define VARIANT4_RANDOM_MATH_INIT(part) \
+  uint32_t r##part[9]; \
+  struct V4_Instruction code##part[256]; \
+  { \
+    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); \
+  } \
+  v4_random_math_init(code##part, ctx[part]->height);
+
+#define VARIANT4_RANDOM_MATH(part, al, ah, cl, bx0, bx1) \
+  { \
+    cl ^= (r##part[0] + r##part[1]) | ((uint64_t)(r##part[2] + r##part[3]) << 32); \
+    r##part[4] = (uint32_t)(al); \
+    r##part[5] = (uint32_t)(ah); \
+    r##part[6] = (uint32_t)(_mm_cvtsi128_si32(bx0)); \
+    r##part[7] = (uint32_t)(_mm_cvtsi128_si32(bx1)); \
+    r##part[8] = (uint32_t)(_mm_cvtsi128_si32(_mm_srli_si128(bx1, 8))); \
+    v4_random_math(code##part, r##part); \
+  }
+
+#endif /* XMRIG_CRYPTONIGHT_MONERO_H */

algo/cryptonight/cryptonight_r_av1.c

@@ -0,0 +1,143 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017      fireice-uk  <https://github.com/fireice-uk>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018      Lee Clagett <https://github.com/vtnerd>
+ * Copyright 2018-2019 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2019 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#include <x86intrin.h>
+#include <string.h>
+
+#include "crypto/c_keccak.h"
+#include "cryptonight.h"
+#include "cryptonight_aesni.h"
+#include "cryptonight_monero.h"
+
+
+void cryptonight_r_av1(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    keccak(input, size, ctx[0]->state, 200);
+
+    cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+
+    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 idx0 = al0;
+
+    for (size_t i = 0; __builtin_expect(i < 0x80000, 1); i++) {
+        __m128i cx        = _mm_load_si128((__m128i *) &l0[idx0 & 0x1FFFF0]);
+        const __m128i ax0 = _mm_set_epi64x(ah0, al0);
+
+        cx = _mm_aesenc_si128(cx, ax0);
+
+        VARIANT4_SHUFFLE(l0, idx0 & 0x1FFFF0, ax0, bx0, bx1, cx);
+        _mm_store_si128((__m128i *) &l0[idx0 & 0x1FFFF0], _mm_xor_si128(bx0, cx));
+
+        idx0 = _mm_cvtsi128_si64(cx);
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[1];
+
+        VARIANT4_RANDOM_MATH(0, al0, ah0, cl, bx0, bx1);
+        al0 ^= r0[2] | ((uint64_t)(r0[3]) << 32);
+        ah0 ^= r0[0] | ((uint64_t)(r0[1]) << 32);
+
+        lo = _umul128(idx0, cl, &hi);
+        VARIANT4_SHUFFLE(l0, idx0 & 0x1FFFF0, ax0, bx0, bx1, cx);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[0] = al0;
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[1] = ah0;
+
+        al0 ^= cl;
+        ah0 ^= ch;
+        idx0 = al0;
+
+        bx1 = bx0;
+        bx0 = cx;
+    }
+
+    cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+
+    keccakf(h0, 24);
+    extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
+}
+
+
+#ifndef XMRIG_NO_ASM
+void v4_compile_code(const struct V4_Instruction* code, int code_size, void* machine_code, enum Assembly ASM);
+
+
+void cryptonight_r_av1_asm_intel(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    if (ctx[0]->generated_code_height != ctx[0]->height) {
+        struct V4_Instruction code[256];
+        const int code_size = v4_random_math_init(code, ctx[0]->height);
+
+        v4_compile_code(code, code_size, (void*)(ctx[0]->generated_code), ASM_INTEL);
+        ctx[0]->generated_code_height = ctx[0]->height;
+    }
+
+    keccak(input, size, ctx[0]->state, 200);
+    cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+
+    ctx[0]->generated_code(ctx[0]);
+
+    cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+    keccakf((uint64_t*) ctx[0]->state, 24);
+    extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
+}
+
+
+void cryptonight_r_av1_asm_bulldozer(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    if (ctx[0]->generated_code_height != ctx[0]->height) {
+        struct V4_Instruction code[256];
+        const int code_size = v4_random_math_init(code, ctx[0]->height);
+
+        v4_compile_code(code, code_size, (void*)(ctx[0]->generated_code), ASM_BULLDOZER);
+        ctx[0]->generated_code_height = ctx[0]->height;
+    }
+
+    keccak(input, size, ctx[0]->state, 200);
+    cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+
+    ctx[0]->generated_code(ctx[0]);
+
+    cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+    keccakf((uint64_t*) ctx[0]->state, 24);
+    extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
+}
+#endif

algo/cryptonight/cryptonight_r_av2.c

@@ -0,0 +1,202 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017      fireice-uk  <https://github.com/fireice-uk>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018      Lee Clagett <https://github.com/vtnerd>
+ * Copyright 2018-2019 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2019 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#include <x86intrin.h>
+#include <string.h>
+
+#include "crypto/c_keccak.h"
+#include "cryptonight.h"
+#include "cryptonight_aesni.h"
+#include "cryptonight_monero.h"
+
+
+void cryptonight_r_av2(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    keccak(input,        size, ctx[0]->state, 200);
+    keccak(input + size, size, ctx[1]->state, 200);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    const uint8_t* l1 = ctx[1]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+    uint64_t* h1 = (uint64_t*) ctx[1]->state;
+
+    VARIANT2_INIT(0);
+    VARIANT2_INIT(1);
+    VARIANT2_SET_ROUNDING_MODE();
+    VARIANT4_RANDOM_MATH_INIT(0);
+    VARIANT4_RANDOM_MATH_INIT(1);
+
+    cn_explode_scratchpad((__m128i*) h0, (__m128i*) l0);
+    cn_explode_scratchpad((__m128i*) h1, (__m128i*) l1);
+
+    uint64_t al0 = h0[0] ^ h0[4];
+    uint64_t al1 = h1[0] ^ h1[4];
+    uint64_t ah0 = h0[1] ^ h0[5];
+    uint64_t ah1 = h1[1] ^ h1[5];
+
+    __m128i bx00 = _mm_set_epi64x(h0[3] ^ h0[7], h0[2] ^ h0[6]);
+    __m128i bx01 = _mm_set_epi64x(h0[9] ^ h0[11], h0[8] ^ h0[10]);
+    __m128i bx10 = _mm_set_epi64x(h1[3] ^ h1[7], h1[2] ^ h1[6]);
+    __m128i bx11 = _mm_set_epi64x(h1[9] ^ h1[11], h1[8] ^ h1[10]);
+
+    uint64_t idx0 = al0;
+    uint64_t idx1 = al1;
+
+    for (size_t i = 0; __builtin_expect(i < 0x80000, 1); i++) {
+        __m128i cx0       = _mm_load_si128((__m128i *) &l0[idx0 & 0x1FFFF0]);
+        __m128i cx1       = _mm_load_si128((__m128i *) &l1[idx1 & 0x1FFFF0]);
+
+        const __m128i ax0 = _mm_set_epi64x(ah0, al0);
+        const __m128i ax1 = _mm_set_epi64x(ah1, al1);
+
+        cx0 = _mm_aesenc_si128(cx0, ax0);
+        cx1 = _mm_aesenc_si128(cx1, ax1);
+
+        VARIANT4_SHUFFLE(l0, idx0 & 0x1FFFF0, ax0, bx00, bx01, cx0);
+        _mm_store_si128((__m128i *) &l0[idx0 & 0x1FFFF0], _mm_xor_si128(bx00, cx0));
+
+        VARIANT4_SHUFFLE(l1, idx1 & 0x1FFFF0, ax1, bx10, bx11, cx1);
+        _mm_store_si128((__m128i *) &l1[idx1 & 0x1FFFF0], _mm_xor_si128(bx10, cx1));
+
+        idx0 = _mm_cvtsi128_si64(cx0);
+        idx1 = _mm_cvtsi128_si64(cx1);
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[1];
+
+        VARIANT4_RANDOM_MATH(0, al0, ah0, cl, bx00, bx01);
+        al0 ^= r0[2] | ((uint64_t)(r0[3]) << 32);
+        ah0 ^= r0[0] | ((uint64_t)(r0[1]) << 32);
+
+        lo = _umul128(idx0, cl, &hi);
+        VARIANT4_SHUFFLE(l0, idx0 & 0x1FFFF0, ax0, bx00, bx01, cx0);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[0] = al0;
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[1] = ah0;
+
+        al0 ^= cl;
+        ah0 ^= ch;
+        idx0 = al0;
+
+        cl = ((uint64_t*) &l1[idx1 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l1[idx1 & 0x1FFFF0])[1];
+
+        VARIANT4_RANDOM_MATH(1, al1, ah1, cl, bx10, bx11);
+        al1 ^= r1[2] | ((uint64_t)(r1[3]) << 32);
+        ah1 ^= r1[0] | ((uint64_t)(r1[1]) << 32);
+
+        lo = _umul128(idx1, cl, &hi);
+        VARIANT4_SHUFFLE(l1, idx1 & 0x1FFFF0, ax1, bx10, bx11, cx1);
+
+        al1 += hi;
+        ah1 += lo;
+
+        ((uint64_t*)&l1[idx1 & 0x1FFFF0])[0] = al1;
+        ((uint64_t*)&l1[idx1 & 0x1FFFF0])[1] = ah1;
+
+        al1 ^= cl;
+        ah1 ^= ch;
+        idx1 = al1;
+
+        bx01 = bx00;
+        bx11 = bx10;
+
+        bx00 = cx0;
+        bx10 = cx1;
+    }
+
+    cn_implode_scratchpad((__m128i*) l0, (__m128i*) h0);
+    cn_implode_scratchpad((__m128i*) l1, (__m128i*) h1);
+
+    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);
+}
+
+
+#ifndef XMRIG_NO_ASM
+void v4_compile_code_double(const struct V4_Instruction* code, int code_size, void* machine_code, enum Assembly ASM);
+
+
+void cryptonight_r_av2_asm_intel(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    if (ctx[0]->generated_code_height != ctx[0]->height) {
+        struct V4_Instruction code[256];
+        const int code_size = v4_random_math_init(code, ctx[0]->height);
+        v4_compile_code_double(code, code_size, (void*)(ctx[0]->generated_code_double), ASM_INTEL);
+        ctx[0]->generated_code_height = ctx[0]->height;
+    }
+
+    keccak(input,        size, ctx[0]->state, 200);
+    keccak(input + size, size, ctx[1]->state, 200);
+    cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+    cn_explode_scratchpad((__m128i*) ctx[1]->state, (__m128i*) ctx[1]->memory);
+
+    ctx[0]->generated_code_double(ctx[0], ctx[1]);
+
+    cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+    cn_implode_scratchpad((__m128i*) ctx[1]->memory, (__m128i*) ctx[1]->state);
+
+    keccakf((uint64_t *) ctx[0]->state, 24);
+    keccakf((uint64_t *) 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);
+}
+
+
+void cryptonight_r_av2_asm_bulldozer(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    if (ctx[0]->generated_code_height != ctx[0]->height) {
+        struct V4_Instruction code[256];
+        const int code_size = v4_random_math_init(code, ctx[0]->height);
+        v4_compile_code_double(code, code_size, (void*)(ctx[0]->generated_code_double), ASM_BULLDOZER);
+        ctx[0]->generated_code_height = ctx[0]->height;
+    }
+
+    keccak(input,        size, ctx[0]->state, 200);
+    keccak(input + size, size, ctx[1]->state, 200);
+    cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+    cn_explode_scratchpad((__m128i*) ctx[1]->state, (__m128i*) ctx[1]->memory);
+
+    ctx[0]->generated_code_double(ctx[0], ctx[1]);
+
+    cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+    cn_implode_scratchpad((__m128i*) ctx[1]->memory, (__m128i*) ctx[1]->state);
+
+    keccakf((uint64_t *) ctx[0]->state, 24);
+    keccakf((uint64_t *) 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);
+}
+#endif

algo/cryptonight/cryptonight_r_av3.c

@@ -0,0 +1,112 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017      fireice-uk  <https://github.com/fireice-uk>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018      Lee Clagett <https://github.com/vtnerd>
+ * Copyright 2018-2019 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2019 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#include <x86intrin.h>
+#include <string.h>
+
+#include "crypto/c_keccak.h"
+#include "cryptonight.h"
+#include "cryptonight_monero.h"
+#include "cryptonight_softaes.h"
+
+
+#ifndef XMRIG_NO_ASM
+void v4_soft_aes_compile_code(const struct V4_Instruction* code, int code_size, void* machine_code, enum Assembly ASM);
+#endif
+
+
+void cryptonight_r_av3(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    keccak(input, size, ctx[0]->state, 200);
+    cn_explode_scratchpad((__m128i*) ctx[0]->state, (__m128i*) ctx[0]->memory);
+
+#   ifndef XMRIG_NO_ASM
+    if (ctx[0]->generated_code_height != ctx[0]->height) {
+        struct V4_Instruction code[256];
+        const int code_size = v4_random_math_init(code, ctx[0]->height);
+
+        v4_soft_aes_compile_code(code, code_size, (void*)(ctx[0]->generated_code), ASM_NONE);
+        ctx[0]->generated_code_height = ctx[0]->height;
+    }
+
+    ctx[0]->saes_table = (const uint32_t*)saes_table;
+    ctx[0]->generated_code(ctx[0]);
+#   else
+    const uint8_t* l0 = ctx[0]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+
+    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 idx0 = al0;
+
+    for (size_t i = 0; __builtin_expect(i < 0x80000, 1); i++) {
+        __m128i cx        = _mm_load_si128((__m128i *) &l0[idx0 & 0x1FFFF0]);
+        const __m128i ax0 = _mm_set_epi64x(ah0, al0);
+
+        cx = soft_aesenc(cx, ax0);
+
+        VARIANT4_SHUFFLE(l0, idx0 & 0x1FFFF0, ax0, bx0, bx1, cx);
+        _mm_store_si128((__m128i *) &l0[idx0 & 0x1FFFF0], _mm_xor_si128(bx0, cx));
+
+        idx0 = _mm_cvtsi128_si64(cx);
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[1];
+
+        VARIANT4_RANDOM_MATH(0, al0, ah0, cl, bx0, bx1);
+        al0 ^= r0[2] | ((uint64_t)(r0[3]) << 32);
+        ah0 ^= r0[0] | ((uint64_t)(r0[1]) << 32);
+
+        lo = _umul128(idx0, cl, &hi);
+        VARIANT4_SHUFFLE(l0, idx0 & 0x1FFFF0, ax0, bx0, bx1, cx);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[0] = al0;
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[1] = ah0;
+
+        al0 ^= cl;
+        ah0 ^= ch;
+        idx0 = al0;
+
+        bx1 = bx0;
+        bx0 = cx;
+    }
+#   endif
+
+    cn_implode_scratchpad((__m128i*) ctx[0]->memory, (__m128i*) ctx[0]->state);
+    keccakf((uint64_t *) ctx[0]->state, 24);
+    extra_hashes[ctx[0]->state[0] & 3](ctx[0]->state, 200, output);
+}

algo/cryptonight/cryptonight_r_av4.c

@@ -0,0 +1,143 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017      fireice-uk  <https://github.com/fireice-uk>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018      Lee Clagett <https://github.com/vtnerd>
+ * Copyright 2018-2019 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2019 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#include <x86intrin.h>
+#include <string.h>
+
+#include "crypto/c_keccak.h"
+#include "cryptonight.h"
+#include "cryptonight_monero.h"
+#include "cryptonight_softaes.h"
+
+
+void cryptonight_r_av4(const uint8_t *restrict input, size_t size, uint8_t *restrict output, struct cryptonight_ctx **restrict ctx)
+{
+    keccak(input,        size, ctx[0]->state, 200);
+    keccak(input + size, size, ctx[1]->state, 200);
+
+    const uint8_t* l0 = ctx[0]->memory;
+    const uint8_t* l1 = ctx[1]->memory;
+    uint64_t* h0 = (uint64_t*) ctx[0]->state;
+    uint64_t* h1 = (uint64_t*) ctx[1]->state;
+
+    VARIANT2_INIT(0);
+    VARIANT2_INIT(1);
+    VARIANT2_SET_ROUNDING_MODE();
+    VARIANT4_RANDOM_MATH_INIT(0);
+    VARIANT4_RANDOM_MATH_INIT(1);
+
+    cn_explode_scratchpad((__m128i*) h0, (__m128i*) l0);
+    cn_explode_scratchpad((__m128i*) h1, (__m128i*) l1);
+
+    uint64_t al0 = h0[0] ^ h0[4];
+    uint64_t al1 = h1[0] ^ h1[4];
+    uint64_t ah0 = h0[1] ^ h0[5];
+    uint64_t ah1 = h1[1] ^ h1[5];
+
+    __m128i bx00 = _mm_set_epi64x(h0[3] ^ h0[7], h0[2] ^ h0[6]);
+    __m128i bx01 = _mm_set_epi64x(h0[9] ^ h0[11], h0[8] ^ h0[10]);
+    __m128i bx10 = _mm_set_epi64x(h1[3] ^ h1[7], h1[2] ^ h1[6]);
+    __m128i bx11 = _mm_set_epi64x(h1[9] ^ h1[11], h1[8] ^ h1[10]);
+
+    uint64_t idx0 = al0;
+    uint64_t idx1 = al1;
+
+    for (size_t i = 0; __builtin_expect(i < 0x80000, 1); i++) {
+        __m128i cx0       = _mm_load_si128((__m128i *) &l0[idx0 & 0x1FFFF0]);
+        __m128i cx1       = _mm_load_si128((__m128i *) &l1[idx1 & 0x1FFFF0]);
+
+        const __m128i ax0 = _mm_set_epi64x(ah0, al0);
+        const __m128i ax1 = _mm_set_epi64x(ah1, al1);
+
+        cx0 = soft_aesenc(cx0, ax0);
+        cx1 = soft_aesenc(cx1, ax1);
+
+        VARIANT4_SHUFFLE(l0, idx0 & 0x1FFFF0, ax0, bx00, bx01, cx0);
+        _mm_store_si128((__m128i *) &l0[idx0 & 0x1FFFF0], _mm_xor_si128(bx00, cx0));
+
+        VARIANT4_SHUFFLE(l1, idx1 & 0x1FFFF0, ax1, bx10, bx11, cx1);
+        _mm_store_si128((__m128i *) &l1[idx1 & 0x1FFFF0], _mm_xor_si128(bx10, cx1));
+
+        idx0 = _mm_cvtsi128_si64(cx0);
+        idx1 = _mm_cvtsi128_si64(cx1);
+
+        uint64_t hi, lo, cl, ch;
+        cl = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l0[idx0 & 0x1FFFF0])[1];
+
+        VARIANT4_RANDOM_MATH(0, al0, ah0, cl, bx00, bx01);
+        al0 ^= r0[2] | ((uint64_t)(r0[3]) << 32);
+        ah0 ^= r0[0] | ((uint64_t)(r0[1]) << 32);
+
+        lo = _umul128(idx0, cl, &hi);
+        VARIANT4_SHUFFLE(l0, idx0 & 0x1FFFF0, ax0, bx00, bx01, cx0);
+
+        al0 += hi;
+        ah0 += lo;
+
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[0] = al0;
+        ((uint64_t*)&l0[idx0 & 0x1FFFF0])[1] = ah0;
+
+        al0 ^= cl;
+        ah0 ^= ch;
+        idx0 = al0;
+
+        cl = ((uint64_t*) &l1[idx1 & 0x1FFFF0])[0];
+        ch = ((uint64_t*) &l1[idx1 & 0x1FFFF0])[1];
+
+        VARIANT4_RANDOM_MATH(1, al1, ah1, cl, bx10, bx11);
+        al1 ^= r1[2] | ((uint64_t)(r1[3]) << 32);
+        ah1 ^= r1[0] | ((uint64_t)(r1[1]) << 32);
+
+        lo = _umul128(idx1, cl, &hi);
+        VARIANT4_SHUFFLE(l1, idx1 & 0x1FFFF0, ax1, bx10, bx11, cx1);
+
+        al1 += hi;
+        ah1 += lo;
+
+        ((uint64_t*)&l1[idx1 & 0x1FFFF0])[0] = al1;
+        ((uint64_t*)&l1[idx1 & 0x1FFFF0])[1] = ah1;
+
+        al1 ^= cl;
+        ah1 ^= ch;
+        idx1 = al1;
+
+        bx01 = bx00;
+        bx11 = bx10;
+
+        bx00 = cx0;
+        bx10 = cx1;
+    }
+
+    cn_implode_scratchpad((__m128i*) l0, (__m128i*) h0);
+    cn_implode_scratchpad((__m128i*) l1, (__m128i*) h1);
+
+    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);
+}

algo/cryptonight/cryptonight_softaes.h

@@ -0,0 +1,255 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018-2019 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2019 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef XMRIG_CRYPTONIGHT_SOFTAES_H
+#define XMRIG_CRYPTONIGHT_SOFTAES_H
+
+
+#include <x86intrin.h>
+#include <stdint.h>
+
+
+#include "crypto/soft_aes.h"
+
+
+// This will shift and xor tmp1 into itself as 4 32-bit vals such as
+// sl_xor(a1 a2 a3 a4) = a1 (a2^a1) (a3^a2^a1) (a4^a3^a2^a1)
+static inline __m128i sl_xor(__m128i tmp1)
+{
+    __m128i tmp4;
+    tmp4 = _mm_slli_si128(tmp1, 0x04);
+    tmp1 = _mm_xor_si128(tmp1, tmp4);
+    tmp4 = _mm_slli_si128(tmp4, 0x04);
+    tmp1 = _mm_xor_si128(tmp1, tmp4);
+    tmp4 = _mm_slli_si128(tmp4, 0x04);
+    tmp1 = _mm_xor_si128(tmp1, tmp4);
+    return tmp1;
+}
+
+
+static inline void aes_genkey_sub(__m128i* xout0, __m128i* xout2, uint8_t rcon)
+{
+    __m128i xout1 = soft_aeskeygenassist(*xout2, rcon);
+    xout1  = _mm_shuffle_epi32(xout1, 0xFF); // see PSHUFD, set all elems to 4th elem
+    *xout0 = sl_xor(*xout0);
+    *xout0 = _mm_xor_si128(*xout0, xout1);
+    xout1  = soft_aeskeygenassist(*xout0, 0x00);
+    xout1  = _mm_shuffle_epi32(xout1, 0xAA); // see PSHUFD, set all elems to 3rd elem
+    *xout2 = sl_xor(*xout2);
+    *xout2 = _mm_xor_si128(*xout2, xout1);
+}
+
+
+static inline void aes_round(__m128i key, __m128i* x0, __m128i* x1, __m128i* x2, __m128i* x3, __m128i* x4, __m128i* x5, __m128i* x6, __m128i* x7)
+{
+    *x0 = soft_aesenc(*x0, key);
+    *x1 = soft_aesenc(*x1, key);
+    *x2 = soft_aesenc(*x2, key);
+    *x3 = soft_aesenc(*x3, key);
+    *x4 = soft_aesenc(*x4, key);
+    *x5 = soft_aesenc(*x5, key);
+    *x6 = soft_aesenc(*x6, key);
+    *x7 = soft_aesenc(*x7, key);
+}
+
+
+static inline void aes_genkey(const __m128i* memory, __m128i* k0, __m128i* k1, __m128i* k2, __m128i* k3, __m128i* k4, __m128i* k5, __m128i* k6, __m128i* k7, __m128i* k8, __m128i* k9)
+{
+    __m128i xout0 = _mm_load_si128(memory);
+    __m128i xout2 = _mm_load_si128(memory + 1);
+    *k0 = xout0;
+    *k1 = xout2;
+
+    aes_genkey_sub(&xout0, &xout2, 0x1);
+    *k2 = xout0;
+    *k3 = xout2;
+
+    aes_genkey_sub(&xout0, &xout2, 0x2);
+    *k4 = xout0;
+    *k5 = xout2;
+
+    aes_genkey_sub(&xout0, &xout2, 0x4);
+    *k6 = xout0;
+    *k7 = xout2;
+
+    aes_genkey_sub(&xout0, &xout2, 0x8);
+    *k8 = xout0;
+    *k9 = xout2;
+}
+
+
+static inline void cn_explode_scratchpad(const __m128i* input, __m128i* output)
+{
+    // This is more than we have registers, compiler will assign 2 keys on the stack
+    __m128i xin0, xin1, xin2, xin3, xin4, xin5, xin6, xin7;
+    __m128i k0, k1, k2, k3, k4, k5, k6, k7, k8, k9;
+
+    aes_genkey(input, &k0, &k1, &k2, &k3, &k4, &k5, &k6, &k7, &k8, &k9);
+
+    xin0 = _mm_load_si128(input + 4);
+    xin1 = _mm_load_si128(input + 5);
+    xin2 = _mm_load_si128(input + 6);
+    xin3 = _mm_load_si128(input + 7);
+    xin4 = _mm_load_si128(input + 8);
+    xin5 = _mm_load_si128(input + 9);
+    xin6 = _mm_load_si128(input + 10);
+    xin7 = _mm_load_si128(input + 11);
+
+    for (size_t i = 0; i < 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);
+        aes_round(k3, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k4, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k5, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k6, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k7, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k8, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+        aes_round(k9, &xin0, &xin1, &xin2, &xin3, &xin4, &xin5, &xin6, &xin7);
+
+        _mm_store_si128(output + i + 0, xin0);
+        _mm_store_si128(output + i + 1, xin1);
+        _mm_store_si128(output + i + 2, xin2);
+        _mm_store_si128(output + i + 3, xin3);
+        _mm_store_si128(output + i + 4, xin4);
+        _mm_store_si128(output + i + 5, xin5);
+        _mm_store_si128(output + i + 6, xin6);
+        _mm_store_si128(output + i + 7, xin7);
+    }
+}
+
+
+static inline void cn_implode_scratchpad(const __m128i* input, __m128i* output)
+{
+    // This is more than we have registers, compiler will assign 2 keys on the stack
+    __m128i xout0, xout1, xout2, xout3, xout4, xout5, xout6, xout7;
+    __m128i k0, k1, k2, k3, k4, k5, k6, k7, k8, k9;
+
+    aes_genkey(output + 2, &k0, &k1, &k2, &k3, &k4, &k5, &k6, &k7, &k8, &k9);
+
+    xout0 = _mm_load_si128(output + 4);
+    xout1 = _mm_load_si128(output + 5);
+    xout2 = _mm_load_si128(output + 6);
+    xout3 = _mm_load_si128(output + 7);
+    xout4 = _mm_load_si128(output + 8);
+    xout5 = _mm_load_si128(output + 9);
+    xout6 = _mm_load_si128(output + 10);
+    xout7 = _mm_load_si128(output + 11);
+
+    for (size_t i = 0; __builtin_expect(i < MEMORY / sizeof(__m128i), 1); 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);
+        xout3 = _mm_xor_si128(_mm_load_si128(input + i + 3), xout3);
+        xout4 = _mm_xor_si128(_mm_load_si128(input + i + 4), xout4);
+        xout5 = _mm_xor_si128(_mm_load_si128(input + i + 5), xout5);
+        xout6 = _mm_xor_si128(_mm_load_si128(input + i + 6), xout6);
+        xout7 = _mm_xor_si128(_mm_load_si128(input + i + 7), xout7);
+
+        aes_round(k0, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k1, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k2, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k3, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k4, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k5, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k6, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k7, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k8, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+        aes_round(k9, &xout0, &xout1, &xout2, &xout3, &xout4, &xout5, &xout6, &xout7);
+    }
+
+    _mm_store_si128(output + 4, xout0);
+    _mm_store_si128(output + 5, xout1);
+    _mm_store_si128(output + 6, xout2);
+    _mm_store_si128(output + 7, xout3);
+    _mm_store_si128(output + 8, xout4);
+    _mm_store_si128(output + 9, xout5);
+    _mm_store_si128(output + 10, xout6);
+    _mm_store_si128(output + 11, xout7);
+}
+
+
+#if defined(__x86_64__)
+#   define EXTRACT64(X) _mm_cvtsi128_si64(X)
+
+static inline uint64_t _umul128(uint64_t a, uint64_t b, uint64_t* hi)
+{
+    unsigned __int128 r = (unsigned __int128) a * (unsigned __int128) b;
+    *hi = r >> 64;
+    return (uint64_t) r;
+}
+#elif defined(__i386__)
+#   define HI32(X) \
+    _mm_srli_si128((X), 4)
+
+
+#   define EXTRACT64(X) \
+    ((uint64_t)(uint32_t)_mm_cvtsi128_si32(X) | \
+    ((uint64_t)(uint32_t)_mm_cvtsi128_si32(HI32(X)) << 32))
+
+inline uint64_t _umul128(uint64_t multiplier, uint64_t multiplicand, uint64_t *product_hi) {
+    // multiplier   = ab = a * 2^32 + b
+    // multiplicand = cd = c * 2^32 + d
+    // ab * cd = a * c * 2^64 + (a * d + b * c) * 2^32 + b * d
+    uint64_t a = multiplier >> 32;
+    uint64_t b = multiplier & 0xFFFFFFFF;
+    uint64_t c = multiplicand >> 32;
+    uint64_t d = multiplicand & 0xFFFFFFFF;
+
+    //uint64_t ac = a * c;
+    uint64_t ad = a * d;
+    //uint64_t bc = b * c;
+    uint64_t bd = b * d;
+
+    uint64_t adbc = ad + (b * c);
+    uint64_t adbc_carry = adbc < ad ? 1 : 0;
+
+    // multiplier * multiplicand = product_hi * 2^64 + product_lo
+    uint64_t product_lo = bd + (adbc << 32);
+    uint64_t product_lo_carry = product_lo < bd ? 1 : 0;
+    *product_hi = (a * c) + (adbc >> 32) + (adbc_carry << 32) + product_lo_carry;
+
+    return product_lo;
+}
+#endif
+
+
+static inline void cryptonight_monero_tweak(uint64_t* mem_out, __m128i tmp)
+{
+    mem_out[0] = EXTRACT64(tmp);
+
+    tmp = _mm_castps_si128(_mm_movehl_ps(_mm_castsi128_ps(tmp), _mm_castsi128_ps(tmp)));
+    uint64_t vh = EXTRACT64(tmp);
+
+    uint8_t x = vh >> 24;
+    static const uint16_t table = 0x7531;
+    const uint8_t index = (((x >> 3) & 6) | (x & 1)) << 1;
+    vh ^= ((table >> index) & 0x3) << 28;
+
+    mem_out[1] = vh;
+}
+
+
+#endif /* XMRIG_CRYPTONIGHT_SOFTAES_H */

algo/cryptonight/cryptonight_test.h

@@ -0,0 +1,129 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018      Lee Clagett <https://github.com/vtnerd>
+ * Copyright 2018-2019 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2019 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef XMRIG_CRYPTONIGHT_TEST_H
+#define XMRIG_CRYPTONIGHT_TEST_H
+
+
+#include <stdint.h>
+
+
+const static uint8_t test_input[152] = {
+    0x03, 0x05, 0xA0, 0xDB, 0xD6, 0xBF, 0x05, 0xCF, 0x16, 0xE5, 0x03, 0xF3, 0xA6, 0x6F, 0x78, 0x00,
+    0x7C, 0xBF, 0x34, 0x14, 0x43, 0x32, 0xEC, 0xBF, 0xC2, 0x2E, 0xD9, 0x5C, 0x87, 0x00, 0x38, 0x3B,
+    0x30, 0x9A, 0xCE, 0x19, 0x23, 0xA0, 0x96, 0x4B, 0x00, 0x00, 0x00, 0x08, 0xBA, 0x93, 0x9A, 0x62,
+    0x72, 0x4C, 0x0D, 0x75, 0x81, 0xFC, 0xE5, 0x76, 0x1E, 0x9D, 0x8A, 0x0E, 0x6A, 0x1C, 0x3F, 0x92,
+    0x4F, 0xDD, 0x84, 0x93, 0xD1, 0x11, 0x56, 0x49, 0xC0, 0x5E, 0xB6, 0x01,
+    0x01, 0x00, 0xFB, 0x8E, 0x8A, 0xC8, 0x05, 0x89, 0x93, 0x23, 0x37, 0x1B, 0xB7, 0x90, 0xDB, 0x19,
+    0x21, 0x8A, 0xFD, 0x8D, 0xB8, 0xE3, 0x75, 0x5D, 0x8B, 0x90, 0xF3, 0x9B, 0x3D, 0x55, 0x06, 0xA9,
+    0xAB, 0xCE, 0x4F, 0xA9, 0x12, 0x24, 0x45, 0x00, 0x00, 0x00, 0x00, 0xEE, 0x81, 0x46, 0xD4, 0x9F,
+    0xA9, 0x3E, 0xE7, 0x24, 0xDE, 0xB5, 0x7D, 0x12, 0xCB, 0xC6, 0xC6, 0xF3, 0xB9, 0x24, 0xD9, 0x46,
+    0x12, 0x7C, 0x7A, 0x97, 0x41, 0x8F, 0x93, 0x48, 0x82, 0x8F, 0x0F, 0x02
+};
+
+
+const static uint8_t test_output_v0[64] = {
+    0x1A, 0x3F, 0xFB, 0xEE, 0x90, 0x9B, 0x42, 0x0D, 0x91, 0xF7, 0xBE, 0x6E, 0x5F, 0xB5, 0x6D, 0xB7,
+    0x1B, 0x31, 0x10, 0xD8, 0x86, 0x01, 0x1E, 0x87, 0x7E, 0xE5, 0x78, 0x6A, 0xFD, 0x08, 0x01, 0x00,
+    0x1B, 0x60, 0x6A, 0x3F, 0x4A, 0x07, 0xD6, 0x48, 0x9A, 0x1B, 0xCD, 0x07, 0x69, 0x7B, 0xD1, 0x66,
+    0x96, 0xB6, 0x1C, 0x8A, 0xE9, 0x82, 0xF6, 0x1A, 0x90, 0x16, 0x0F, 0x4E, 0x52, 0x82, 0x8A, 0x7F
+};
+
+
+// Cryptonight variant 1 (Monero v7)
+const static uint8_t test_output_v1[64] = {
+    0xF2, 0x2D, 0x3D, 0x62, 0x03, 0xD2, 0xA0, 0x8B, 0x41, 0xD9, 0x02, 0x72, 0x78, 0xD8, 0xBC, 0xC9,
+    0x83, 0xAC, 0xAD, 0xA9, 0xB6, 0x8E, 0x52, 0xE3, 0xC6, 0x89, 0x69, 0x2A, 0x50, 0xE9, 0x21, 0xD9,
+    0xC9, 0xFA, 0xE8, 0x42, 0x5D, 0x86, 0x88, 0xDC, 0x23, 0x6B, 0xCD, 0xBC, 0x42, 0xFD, 0xB4, 0x2D,
+    0x37, 0x6C, 0x6E, 0xC1, 0x90, 0x50, 0x1A, 0xA8, 0x4B, 0x04, 0xA4, 0xB4, 0xCF, 0x1E, 0xE1, 0x22
+};
+
+
+// Cryptonight variant 2 (Monero v8)
+const static uint8_t test_output_v2[64] = {
+    0x97, 0x37, 0x82, 0x82, 0xCF, 0x10, 0xE7, 0xAD, 0x03, 0x3F, 0x7B, 0x80, 0x74, 0xC4, 0x0E, 0x14,
+    0xD0, 0x6E, 0x7F, 0x60, 0x9D, 0xDD, 0xDA, 0x78, 0x76, 0x80, 0xB5, 0x8C, 0x05, 0xF4, 0x3D, 0x21,
+    0x87, 0x1F, 0xCD, 0x68, 0x23, 0xF6, 0xA8, 0x79, 0xBB, 0x3F, 0x33, 0x95, 0x1C, 0x8E, 0x8E, 0x89,
+    0x1D, 0x40, 0x43, 0x88, 0x0B, 0x02, 0xDF, 0xA1, 0xBB, 0x3B, 0xE4, 0x98, 0xB5, 0x0E, 0x75, 0x78
+};
+
+
+struct cn_r_test_input_data
+{
+    uint64_t height;
+    size_t size;
+    uint8_t data[64];
+};
+
+
+const static struct cn_r_test_input_data cn_r_test_input[] = {
+    { 1806260, 44, { 0x54, 0x68, 0x69, 0x73, 0x20, 0x69, 0x73, 0x20, 0x61, 0x20, 0x74, 0x65, 0x73, 0x74, 0x20, 0x54, 0x68, 0x69, 0x73, 0x20, 0x69, 0x73, 0x20, 0x61, 0x20, 0x74, 0x65, 0x73, 0x74, 0x20, 0x54, 0x68, 0x69, 0x73, 0x20, 0x69, 0x73, 0x20, 0x61, 0x20, 0x74, 0x65, 0x73, 0x74 } },
+    { 1806261, 50, { 0x4c, 0x6f, 0x72, 0x65, 0x6d, 0x20, 0x69, 0x70, 0x73, 0x75, 0x6d, 0x20, 0x64, 0x6f, 0x6c, 0x6f, 0x72, 0x20, 0x73, 0x69, 0x74, 0x20, 0x61, 0x6d, 0x65, 0x74, 0x2c, 0x20, 0x63, 0x6f, 0x6e, 0x73, 0x65, 0x63, 0x74, 0x65, 0x74, 0x75, 0x72, 0x20, 0x61, 0x64, 0x69, 0x70, 0x69, 0x73, 0x63, 0x69, 0x6e, 0x67 } },
+    { 1806262, 48, { 0x65, 0x6c, 0x69, 0x74, 0x2c, 0x20, 0x73, 0x65, 0x64, 0x20, 0x64, 0x6f, 0x20, 0x65, 0x69, 0x75, 0x73, 0x6d, 0x6f, 0x64, 0x20, 0x74, 0x65, 0x6d, 0x70, 0x6f, 0x72, 0x20, 0x69, 0x6e, 0x63, 0x69, 0x64, 0x69, 0x64, 0x75, 0x6e, 0x74, 0x20, 0x75, 0x74, 0x20, 0x6c, 0x61, 0x62, 0x6f, 0x72, 0x65 } },
+    { 1806263, 48, { 0x65, 0x74, 0x20, 0x64, 0x6f, 0x6c, 0x6f, 0x72, 0x65, 0x20, 0x6d, 0x61, 0x67, 0x6e, 0x61, 0x20, 0x61, 0x6c, 0x69, 0x71, 0x75, 0x61, 0x2e, 0x20, 0x55, 0x74, 0x20, 0x65, 0x6e, 0x69, 0x6d, 0x20, 0x61, 0x64, 0x20, 0x6d, 0x69, 0x6e, 0x69, 0x6d, 0x20, 0x76, 0x65, 0x6e, 0x69, 0x61, 0x6d, 0x2c } },
+    { 1806264, 46, { 0x71, 0x75, 0x69, 0x73, 0x20, 0x6e, 0x6f, 0x73, 0x74, 0x72, 0x75, 0x64, 0x20, 0x65, 0x78, 0x65, 0x72, 0x63, 0x69, 0x74, 0x61, 0x74, 0x69, 0x6f, 0x6e, 0x20, 0x75, 0x6c, 0x6c, 0x61, 0x6d, 0x63, 0x6f, 0x20, 0x6c, 0x61, 0x62, 0x6f, 0x72, 0x69, 0x73, 0x20, 0x6e, 0x69, 0x73, 0x69 } },
+    { 1806265, 45, { 0x75, 0x74, 0x20, 0x61, 0x6c, 0x69, 0x71, 0x75, 0x69, 0x70, 0x20, 0x65, 0x78, 0x20, 0x65, 0x61, 0x20, 0x63, 0x6f, 0x6d, 0x6d, 0x6f, 0x64, 0x6f, 0x20, 0x63, 0x6f, 0x6e, 0x73, 0x65, 0x71, 0x75, 0x61, 0x74, 0x2e, 0x20, 0x44, 0x75, 0x69, 0x73, 0x20, 0x61, 0x75, 0x74, 0x65 } },
+    { 1806266, 47, { 0x69, 0x72, 0x75, 0x72, 0x65, 0x20, 0x64, 0x6f, 0x6c, 0x6f, 0x72, 0x20, 0x69, 0x6e, 0x20, 0x72, 0x65, 0x70, 0x72, 0x65, 0x68, 0x65, 0x6e, 0x64, 0x65, 0x72, 0x69, 0x74, 0x20, 0x69, 0x6e, 0x20, 0x76, 0x6f, 0x6c, 0x75, 0x70, 0x74, 0x61, 0x74, 0x65, 0x20, 0x76, 0x65, 0x6c, 0x69, 0x74 } },
+    { 1806267, 44, { 0x65, 0x73, 0x73, 0x65, 0x20, 0x63, 0x69, 0x6c, 0x6c, 0x75, 0x6d, 0x20, 0x64, 0x6f, 0x6c, 0x6f, 0x72, 0x65, 0x20, 0x65, 0x75, 0x20, 0x66, 0x75, 0x67, 0x69, 0x61, 0x74, 0x20, 0x6e, 0x75, 0x6c, 0x6c, 0x61, 0x20, 0x70, 0x61, 0x72, 0x69, 0x61, 0x74, 0x75, 0x72, 0x2e } },
+    { 1806268, 47, { 0x45, 0x78, 0x63, 0x65, 0x70, 0x74, 0x65, 0x75, 0x72, 0x20, 0x73, 0x69, 0x6e, 0x74, 0x20, 0x6f, 0x63, 0x63, 0x61, 0x65, 0x63, 0x61, 0x74, 0x20, 0x63, 0x75, 0x70, 0x69, 0x64, 0x61, 0x74, 0x61, 0x74, 0x20, 0x6e, 0x6f, 0x6e, 0x20, 0x70, 0x72, 0x6f, 0x69, 0x64, 0x65, 0x6e, 0x74, 0x2c } },
+    { 1806269, 62, { 0x73, 0x75, 0x6e, 0x74, 0x20, 0x69, 0x6e, 0x20, 0x63, 0x75, 0x6c, 0x70, 0x61, 0x20, 0x71, 0x75, 0x69, 0x20, 0x6f, 0x66, 0x66, 0x69, 0x63, 0x69, 0x61, 0x20, 0x64, 0x65, 0x73, 0x65, 0x72, 0x75, 0x6e, 0x74, 0x20, 0x6d, 0x6f, 0x6c, 0x6c, 0x69, 0x74, 0x20, 0x61, 0x6e, 0x69, 0x6d, 0x20, 0x69, 0x64, 0x20, 0x65, 0x73, 0x74, 0x20, 0x6c, 0x61, 0x62, 0x6f, 0x72, 0x75, 0x6d, 0x2e } },
+};
+
+
+// "cn/r"
+const static uint8_t test_output_r[] = {
+    0xf7, 0x59, 0x58, 0x8a, 0xd5, 0x7e, 0x75, 0x84, 0x67, 0x29, 0x54, 0x43, 0xa9, 0xbd, 0x71, 0x49, 0x0a, 0xbf, 0xf8, 0xe9, 0xda, 0xd1, 0xb9, 0x5b, 0x6b, 0xf2, 0xf5, 0xd0, 0xd7, 0x83, 0x87, 0xbc,
+    0x5b, 0xb8, 0x33, 0xde, 0xca, 0x2b, 0xdd, 0x72, 0x52, 0xa9, 0xcc, 0xd7, 0xb4, 0xce, 0x0b, 0x6a, 0x48, 0x54, 0x51, 0x57, 0x94, 0xb5, 0x6c, 0x20, 0x72, 0x62, 0xf7, 0xa5, 0xb9, 0xbd, 0xb5, 0x66,
+    0x1e, 0xe6, 0x72, 0x8d, 0xa6, 0x0f, 0xbd, 0x8d, 0x7d, 0x55, 0xb2, 0xb1, 0xad, 0xe4, 0x87, 0xa3, 0xcf, 0x52, 0xa2, 0xc3, 0xac, 0x6f, 0x52, 0x0d, 0xb1, 0x2c, 0x27, 0xd8, 0x92, 0x1f, 0x6c, 0xab,
+    0x69, 0x69, 0xfe, 0x2d, 0xdf, 0xb7, 0x58, 0x43, 0x8d, 0x48, 0x04, 0x9f, 0x30, 0x2f, 0xc2, 0x10, 0x8a, 0x4f, 0xcc, 0x93, 0xe3, 0x76, 0x69, 0x17, 0x0e, 0x6d, 0xb4, 0xb0, 0xb9, 0xb4, 0xc4, 0xcb,
+    0x7f, 0x30, 0x48, 0xb4, 0xe9, 0x0d, 0x0c, 0xbe, 0x7a, 0x57, 0xc0, 0x39, 0x4f, 0x37, 0x33, 0x8a, 0x01, 0xfa, 0xe3, 0xad, 0xfd, 0xc0, 0xe5, 0x12, 0x6d, 0x86, 0x3a, 0x89, 0x5e, 0xb0, 0x4e, 0x02,
+    0x1d, 0x29, 0x04, 0x43, 0xa4, 0xb5, 0x42, 0xaf, 0x04, 0xa8, 0x2f, 0x6b, 0x24, 0x94, 0xa6, 0xee, 0x7f, 0x20, 0xf2, 0x75, 0x4c, 0x58, 0xe0, 0x84, 0x90, 0x32, 0x48, 0x3a, 0x56, 0xe8, 0xe2, 0xef,
+    0xc4, 0x3c, 0xc6, 0x56, 0x74, 0x36, 0xa8, 0x6a, 0xfb, 0xd6, 0xaa, 0x9e, 0xaa, 0x7c, 0x27, 0x6e, 0x98, 0x06, 0x83, 0x03, 0x34, 0xb6, 0x14, 0xb2, 0xbe, 0xe2, 0x3c, 0xc7, 0x66, 0x34, 0xf6, 0xfd,
+    0x87, 0xbe, 0x24, 0x79, 0xc0, 0xc4, 0xe8, 0xed, 0xfd, 0xfa, 0xa5, 0x60, 0x3e, 0x93, 0xf4, 0x26, 0x5b, 0x3f, 0x82, 0x24, 0xc1, 0xc5, 0x94, 0x6f, 0xeb, 0x42, 0x48, 0x19, 0xd1, 0x89, 0x90, 0xa4,
+    0xdd, 0x9d, 0x6a, 0x6d, 0x8e, 0x47, 0x46, 0x5c, 0xce, 0xac, 0x08, 0x77, 0xef, 0x88, 0x9b, 0x93, 0xe7, 0xeb, 0xa9, 0x79, 0x55, 0x7e, 0x39, 0x35, 0xd7, 0xf8, 0x6d, 0xce, 0x11, 0xb0, 0x70, 0xf3,
+    0x75, 0xc6, 0xf2, 0xae, 0x49, 0xa2, 0x05, 0x21, 0xde, 0x97, 0x28, 0x5b, 0x43, 0x1e, 0x71, 0x71, 0x25, 0x84, 0x7f, 0xb8, 0x93, 0x5e, 0xd8, 0x4a, 0x61, 0xe7, 0xf8, 0xd3, 0x6a, 0x2c, 0x3d, 0x8e,
+};
+
+
+#ifndef XMRIG_NO_AEON
+const static uint8_t test_output_v0_lite[64] = {
+    0x36, 0x95, 0xB4, 0xB5, 0x3B, 0xB0, 0x03, 0x58, 0xB0, 0xAD, 0x38, 0xDC, 0x16, 0x0F, 0xEB, 0x9E,
+    0x00, 0x4E, 0xEC, 0xE0, 0x9B, 0x83, 0xA7, 0x2E, 0xF6, 0xBA, 0x98, 0x64, 0xD3, 0x51, 0x0C, 0x88,
+    0x28, 0xA2, 0x2B, 0xAD, 0x3F, 0x93, 0xD1, 0x40, 0x8F, 0xCA, 0x47, 0x2E, 0xB5, 0xAD, 0x1C, 0xBE,
+    0x75, 0xF2, 0x1D, 0x05, 0x3C, 0x8C, 0xE5, 0xB3, 0xAF, 0x10, 0x5A, 0x57, 0x71, 0x3E, 0x21, 0xDD
+};
+
+
+// AEON v7
+const static uint8_t test_output_v1_lite[64] = {
+    0x6D, 0x8C, 0xDC, 0x44, 0x4E, 0x9B, 0xBB, 0xFD, 0x68, 0xFC, 0x43, 0xFC, 0xD4, 0x85, 0x5B, 0x22,
+    0x8C, 0x8A, 0x1B, 0xD9, 0x1D, 0x9D, 0x00, 0x28, 0x5B, 0xEC, 0x02, 0xB7, 0xCA, 0x2D, 0x67, 0x41,
+    0x87, 0xC4, 0xE5, 0x70, 0x65, 0x3E, 0xB4, 0xC2, 0xB4, 0x2B, 0x7A, 0x0D, 0x54, 0x65, 0x59, 0x45,
+    0x2D, 0xFA, 0xB5, 0x73, 0xB8, 0x2E, 0xC5, 0x2F, 0x15, 0x2B, 0x7F, 0xF9, 0x8E, 0x79, 0x44, 0x6F
+};
+#endif
+
+
+#endif /* XMRIG_CRYPTONIGHT_TEST_H */

algo/cryptonight/variant4_random_math.h

@@ -0,0 +1,449 @@
+#ifndef VARIANT4_RANDOM_MATH_H
+#define VARIANT4_RANDOM_MATH_H
+
+
+#include <stdint.h>
+#include <string.h>
+#include <stdbool.h>
+
+
+#include "crypto/c_blake256.h"
+
+
+enum V4_Settings
+{
+    // Generate code with minimal theoretical latency = 45 cycles, which is equivalent to 15 multiplications
+    TOTAL_LATENCY = 15 * 3,
+    
+    // Always generate at least 60 instructions
+    NUM_INSTRUCTIONS_MIN = 60,
+
+    // Never generate more than 70 instructions (final RET instruction doesn't count here)
+    NUM_INSTRUCTIONS_MAX = 70,
+
+    // Available ALUs for MUL
+    // Modern CPUs typically have only 1 ALU which can do multiplications
+    ALU_COUNT_MUL = 1,
+
+    // Total available ALUs
+    // Modern CPUs have 4 ALUs, but we use only 3 because random math executes together with other main loop code
+    ALU_COUNT = 3,
+};
+
+enum V4_InstructionList
+{
+    MUL,    // a*b
+    ADD,    // a+b + C, C is an unsigned 32-bit constant
+    SUB,    // a-b
+    ROR,    // rotate right "a" by "b & 31" bits
+    ROL,    // rotate left "a" by "b & 31" bits
+    XOR,    // a^b
+    RET,    // finish execution
+    V4_INSTRUCTION_COUNT = RET,
+};
+
+// V4_InstructionDefinition is used to generate code from random data
+// Every random sequence of bytes is a valid code
+//
+// There are 9 registers in total:
+// - 4 variable registers
+// - 5 constant registers initialized from loop variables
+// This is why dst_index is 2 bits
+enum V4_InstructionDefinition
+{
+    V4_OPCODE_BITS = 3,
+    V4_DST_INDEX_BITS = 2,
+    V4_SRC_INDEX_BITS = 3,
+};
+
+struct V4_Instruction
+{
+    uint8_t opcode;
+    uint8_t dst_index;
+    uint8_t src_index;
+    uint32_t C;
+};
+
+#ifndef FORCEINLINE
+#ifdef __GNUC__
+#define FORCEINLINE __attribute__((always_inline)) inline
+#elif _MSC_VER
+#define FORCEINLINE __forceinline
+#else
+#define FORCEINLINE inline
+#endif
+#endif
+
+#ifndef UNREACHABLE_CODE
+#ifdef __GNUC__
+#define UNREACHABLE_CODE __builtin_unreachable()
+#elif _MSC_VER
+#define UNREACHABLE_CODE __assume(false)
+#else
+#define UNREACHABLE_CODE
+#endif
+#endif
+
+#define SWAP32LE(x) x
+#define SWAP64LE(x) x
+#define hash_extra_blake(data, length, hash) blake256_hash((uint8_t*)(hash), (uint8_t*)(data), (length))
+
+// Random math interpreter's loop is fully unrolled and inlined to achieve 100% branch prediction on CPU:
+// every switch-case will point to the same destination on every iteration of Cryptonight main loop
+//
+// This is about as fast as it can get without using low-level machine code generation
+//template<typename v4_reg>
+static void v4_random_math(const struct V4_Instruction* code, uint32_t r[9])
+{
+#define REG_BITS 32
+#define V4_EXEC(i) \
+    { \
+        const struct V4_Instruction* op = code + i; \
+        const uint32_t src = r[op->src_index]; \
+        uint32_t *dst = r + op->dst_index; \
+        switch (op->opcode) \
+        { \
+        case MUL: \
+            *dst *= src; \
+            break; \
+        case ADD: \
+            *dst += src + op->C; \
+            break; \
+        case SUB: \
+            *dst -= src; \
+            break; \
+        case ROR: \
+            { \
+                const uint32_t shift = src % REG_BITS; \
+                *dst = (*dst >> shift) | (*dst << ((REG_BITS - shift) % REG_BITS)); \
+            } \
+            break; \
+        case ROL: \
+            { \
+                const uint32_t shift = src % REG_BITS; \
+                *dst = (*dst << shift) | (*dst >> ((REG_BITS - shift) % REG_BITS)); \
+            } \
+            break; \
+        case XOR: \
+            *dst ^= src; \
+            break; \
+        case RET: \
+            return; \
+        default: \
+            UNREACHABLE_CODE; \
+            break; \
+        } \
+    }
+
+#define V4_EXEC_10(j) \
+    V4_EXEC(j + 0) \
+    V4_EXEC(j + 1) \
+    V4_EXEC(j + 2) \
+    V4_EXEC(j + 3) \
+    V4_EXEC(j + 4) \
+    V4_EXEC(j + 5) \
+    V4_EXEC(j + 6) \
+    V4_EXEC(j + 7) \
+    V4_EXEC(j + 8) \
+    V4_EXEC(j + 9)
+
+    // Generated program can have 60 + a few more (usually 2-3) instructions to achieve required latency
+    // I've checked all block heights < 10,000,000 and here is the distribution of program sizes:
+    //
+    // 60      27960
+    // 61      105054
+    // 62      2452759
+    // 63      5115997
+    // 64      1022269
+    // 65      1109635
+    // 66      153145
+    // 67      8550
+    // 68      4529
+    // 69      102
+
+    // Unroll 70 instructions here
+    V4_EXEC_10(0);      // instructions 0-9
+    V4_EXEC_10(10);     // instructions 10-19
+    V4_EXEC_10(20);     // instructions 20-29
+    V4_EXEC_10(30);     // instructions 30-39
+    V4_EXEC_10(40);     // instructions 40-49
+    V4_EXEC_10(50);     // instructions 50-59
+    V4_EXEC_10(60);     // instructions 60-69
+
+#undef V4_EXEC_10
+#undef V4_EXEC
+#undef REG_BITS
+}
+
+// If we don't have enough data available, generate more
+static FORCEINLINE void check_data(size_t* data_index, const size_t bytes_needed, int8_t* data, const size_t data_size)
+{
+    if (*data_index + bytes_needed > data_size)
+    {
+        hash_extra_blake(data, data_size, (char*) data);
+        *data_index = 0;
+    }
+}
+
+// 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
+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
+    // These latencies match real-life instruction latencies for Intel CPUs starting from Sandy Bridge and up to Skylake/Coffee lake
+    //
+    // AMD Ryzen has the same latencies except 1-cycle ROR/ROL, so it'll be a bit faster than Intel Sandy Bridge and newer processors
+    // Surprisingly, Intel Nehalem also has 1-cycle ROR/ROL, so it'll also be faster than Intel Sandy Bridge and newer processors
+    // AMD Bulldozer has 4 cycles latency for MUL (slower than Intel) and 1 cycle for ROR/ROL (faster than Intel), so average performance will be the same
+    // Source: https://www.agner.org/optimize/instruction_tables.pdf
+    const int op_latency[V4_INSTRUCTION_COUNT] = { 3, 2, 1, 2, 2, 1 };
+
+    // Instruction latencies for theoretical ASIC implementation
+    const int asic_op_latency[V4_INSTRUCTION_COUNT] = { 3, 1, 1, 1, 1, 1 };
+
+    // Available ALUs for each instruction
+    const int op_ALUs[V4_INSTRUCTION_COUNT] = { ALU_COUNT_MUL, ALU_COUNT, ALU_COUNT, ALU_COUNT, ALU_COUNT, ALU_COUNT };
+
+    int8_t data[32];
+    memset(data, 0, sizeof(data));
+    uint64_t tmp = SWAP64LE(height);
+    memcpy(data, &tmp, sizeof(uint64_t));
+    data[20] = -38;
+
+    // Set data_index past the last byte in data
+    // to trigger full data update with blake hash
+    // before we start using it
+    size_t data_index = sizeof(data);
+
+    int code_size;
+
+    // There is a small chance (1.8%) that register R8 won't be used in the generated program
+    // So we keep track of it and try again if it's not used
+    bool r8_used;
+    do {
+        int latency[9];
+        int asic_latency[9];
+
+        // Tracks previous instruction and value of the source operand for registers R0-R3 throughout code execution
+        // byte 0: current value of the destination register
+        // byte 1: instruction opcode
+        // byte 2: current value of the source register
+        //
+        // Registers R4-R8 are constant and are treated as having the same value because when we do
+        // the same operation twice with two constant source registers, it can be optimized into a single operation
+        uint32_t inst_data[9] = { 0, 1, 2, 3, 0xFFFFFF, 0xFFFFFF, 0xFFFFFF, 0xFFFFFF, 0xFFFFFF };
+
+        bool alu_busy[TOTAL_LATENCY + 1][ALU_COUNT];
+        bool is_rotation[V4_INSTRUCTION_COUNT];
+        bool rotated[4];
+        int rotate_count = 0;
+
+        memset(latency, 0, sizeof(latency));
+        memset(asic_latency, 0, sizeof(asic_latency));
+        memset(alu_busy, 0, sizeof(alu_busy));
+        memset(is_rotation, 0, sizeof(is_rotation));
+        memset(rotated, 0, sizeof(rotated));
+        is_rotation[ROR] = true;
+        is_rotation[ROL] = true;
+
+        int num_retries = 0;
+        code_size = 0;
+
+        int total_iterations = 0;
+        r8_used = false;
+
+        // Generate random code to achieve minimal required latency for our abstract CPU
+        // Try to get this latency for all 4 registers
+        while (((latency[0] < TOTAL_LATENCY) || (latency[1] < TOTAL_LATENCY) || (latency[2] < TOTAL_LATENCY) || (latency[3] < TOTAL_LATENCY)) && (num_retries < 64))
+        {
+            // Fail-safe to guarantee loop termination
+            ++total_iterations;
+            if (total_iterations > 256)
+                break;
+
+            check_data(&data_index, 1, data, sizeof(data));
+
+            const uint8_t c = ((uint8_t*)data)[data_index++];
+
+            // MUL = opcodes 0-2
+            // ADD = opcode 3
+            // SUB = opcode 4
+            // ROR/ROL = opcode 5, shift direction is selected randomly
+            // XOR = opcodes 6-7
+            uint8_t opcode = c & ((1 << V4_OPCODE_BITS) - 1);
+            if (opcode == 5)
+            {
+                check_data(&data_index, 1, data, sizeof(data));
+                opcode = (data[data_index++] >= 0) ? ROR : ROL;
+            }
+            else if (opcode >= 6)
+            {
+                opcode = XOR;
+            }
+            else
+            {
+                opcode = (opcode <= 2) ? MUL : (opcode - 2);
+            }
+
+            uint8_t dst_index = (c >> V4_OPCODE_BITS) & ((1 << V4_DST_INDEX_BITS) - 1);
+            uint8_t src_index = (c >> (V4_OPCODE_BITS + V4_DST_INDEX_BITS)) & ((1 << V4_SRC_INDEX_BITS) - 1);
+
+            const int a = dst_index;
+            int b = src_index;
+
+            // Don't do ADD/SUB/XOR with the same register
+            if (((opcode == ADD) || (opcode == SUB) || (opcode == XOR)) && (a == b))
+            {
+                // a is always < 4, so we don't need to check bounds here
+                b = 8;
+                src_index = b;
+            }
+
+            // Don't do rotation with the same destination twice because it's equal to a single rotation
+            if (is_rotation[opcode] && rotated[a])
+            {
+                continue;
+            }
+
+            // Don't do the same instruction (except MUL) with the same source value twice because all other cases can be optimized:
+            // 2xADD(a, b, C) = ADD(a, b*2, C1+C2), same for SUB and rotations
+            // 2xXOR(a, b) = NOP
+            if ((opcode != MUL) && ((inst_data[a] & 0xFFFF00) == (opcode << 8) + ((inst_data[b] & 255) << 16)))
+            {
+                continue;
+            }
+
+            // Find which ALU is available (and when) for this instruction
+            int next_latency = (latency[a] > latency[b]) ? latency[a] : latency[b];
+            int alu_index = -1;
+            while (next_latency < TOTAL_LATENCY)
+            {
+                for (int i = op_ALUs[opcode] - 1; i >= 0; --i)
+                {
+                    if (!alu_busy[next_latency][i])
+                    {
+                        // ADD is implemented as two 1-cycle instructions on a real CPU, so do an additional availability check
+                        if ((opcode == ADD) && alu_busy[next_latency + 1][i])
+                        {
+                            continue;
+                        }
+
+                        // Rotation can only start when previous rotation is finished, so do an additional availability check
+                        if (is_rotation[opcode] && (next_latency < rotate_count * op_latency[opcode]))
+                        {
+                            continue;
+                        }
+
+                        alu_index = i;
+                        break;
+                    }
+                }
+                if (alu_index >= 0)
+                {
+                    break;
+                }
+                ++next_latency;
+            }
+
+            // Don't generate instructions that leave some register unchanged for more than 7 cycles
+            if (next_latency > latency[a] + 7)
+            {
+                continue;
+            }
+
+            next_latency += op_latency[opcode];
+
+            if (next_latency <= TOTAL_LATENCY)
+            {
+                if (is_rotation[opcode])
+                {
+                    ++rotate_count;
+                }
+
+                // Mark ALU as busy only for the first cycle when it starts executing the instruction because ALUs are fully pipelined
+                alu_busy[next_latency - op_latency[opcode]][alu_index] = true;
+                latency[a] = next_latency;
+
+                // ASIC is supposed to have enough ALUs to run as many independent instructions per cycle as possible, so latency calculation for ASIC is simple
+                asic_latency[a] = ((asic_latency[a] > asic_latency[b]) ? asic_latency[a] : asic_latency[b]) + asic_op_latency[opcode];
+
+                rotated[a] = is_rotation[opcode];
+
+                inst_data[a] = code_size + (opcode << 8) + ((inst_data[b] & 255) << 16);
+
+                code[code_size].opcode = opcode;
+                code[code_size].dst_index = dst_index;
+                code[code_size].src_index = src_index;
+                code[code_size].C = 0;
+
+                if (src_index == 8)
+                {
+                    r8_used = true;
+                }
+
+                if (opcode == ADD)
+                {
+                    // ADD instruction is implemented as two 1-cycle instructions on a real CPU, so mark ALU as busy for the next cycle too
+                    alu_busy[next_latency - op_latency[opcode] + 1][alu_index] = true;
+
+                    // ADD instruction requires 4 more random bytes for 32-bit constant "C" in "a = a + b + C"
+                    check_data(&data_index, sizeof(uint32_t), data, sizeof(data));
+                    uint32_t t;
+                    memcpy(&t, data + data_index, sizeof(uint32_t));
+                    code[code_size].C = SWAP32LE(t);
+                    data_index += sizeof(uint32_t);
+                }
+
+                ++code_size;
+                if (code_size >= NUM_INSTRUCTIONS_MIN)
+                {
+                    break;
+                }
+            }
+            else
+            {
+                ++num_retries;
+            }
+        }
+
+        // ASIC has more execution resources and can extract as much parallelism from the code as possible
+        // We need to add a few more MUL and ROR instructions to achieve minimal required latency for ASIC
+        // Get this latency for at least 1 of the 4 registers
+        const int prev_code_size = code_size;
+        while ((code_size < NUM_INSTRUCTIONS_MAX) && (asic_latency[0] < TOTAL_LATENCY) && (asic_latency[1] < TOTAL_LATENCY) && (asic_latency[2] < TOTAL_LATENCY) && (asic_latency[3] < TOTAL_LATENCY))
+        {
+            int min_idx = 0;
+            int max_idx = 0;
+            for (int i = 1; i < 4; ++i)
+            {
+                if (asic_latency[i] < asic_latency[min_idx]) min_idx = i;
+                if (asic_latency[i] > asic_latency[max_idx]) max_idx = i;
+            }
+
+            const uint8_t pattern[3] = { ROR, MUL, MUL };
+            const uint8_t opcode = pattern[(code_size - prev_code_size) % 3];
+            latency[min_idx] = latency[max_idx] + op_latency[opcode];
+            asic_latency[min_idx] = asic_latency[max_idx] + asic_op_latency[opcode];
+
+            code[code_size].opcode = opcode;
+            code[code_size].dst_index = min_idx;
+            code[code_size].src_index = max_idx;
+            code[code_size].C = 0;
+            ++code_size;
+        }
+
+    // There is ~98.15% chance that loop condition is false, so this loop will execute only 1 iteration most of the time
+    // It never does more than 4 iterations for all block heights < 10,000,000
+    }  while (!r8_used || (code_size < NUM_INSTRUCTIONS_MIN) || (code_size > NUM_INSTRUCTIONS_MAX));
+
+    // It's guaranteed that NUM_INSTRUCTIONS_MIN <= code_size <= NUM_INSTRUCTIONS_MAX here
+    // Add final instruction to stop the interpreter
+    code[code_size].opcode = RET;
+    code[code_size].dst_index = 0;
+    code[code_size].src_index = 0;
+    code[code_size].C = 0;
+
+    return code_size;
+}
+
+#endif

cmake/FindMHD.cmake

@@ -1,39 +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
-    DOC "microhttpd include dir"
-)
-
-find_library(
-    MHD_LIBRARY
-    NAMES microhttpd microhttpd-10 libmicrohttpd libmicrohttpd-dll
-    DOC "microhttpd library"
-)
-
-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_INCLUDE_DIR MHD_LIBRARY)
-mark_as_advanced(MHD_INCLUDE_DIR MHD_LIBRARY)
-

cmake/FindUV.cmake

@@ -1,8 +0,0 @@
-find_path(UV_INCLUDE_DIR NAMES uv.h)
-find_library(UV_LIBRARY NAMES uv libuv)
-
-set(UV_LIBRARIES ${UV_LIBRARY})
-set(UV_INCLUDE_DIRS ${UV_INCLUDE_DIR})
-
-include(FindPackageHandleStandardArgs)
-find_package_handle_standard_args(UV DEFAULT_MSG UV_LIBRARY UV_INCLUDE_DIR)

cmake/asm.cmake

@@ -0,0 +1,27 @@
+if (WITH_ASM AND NOT XMRIG_ARM AND CMAKE_SIZEOF_VOID_P EQUAL 8)
+    set(XMRIG_ASM_LIBRARY "xmrig-asm")
+
+    enable_language(ASM)
+
+    if (WIN32 AND CMAKE_C_COMPILER_ID MATCHES GNU)
+        set(XMRIG_ASM_FILES
+            "crypto/asm/win64/cn_main_loop.S"
+            "crypto/asm/CryptonightR_template.S"
+        )
+    else()
+        set(XMRIG_ASM_FILES
+            "crypto/asm/cn_main_loop.S"
+            "crypto/asm/CryptonightR_template.S"
+        )
+    endif()
+
+    set_property(SOURCE ${XMRIG_ASM_FILES} PROPERTY C)
+
+    add_library(${XMRIG_ASM_LIBRARY} STATIC ${XMRIG_ASM_FILES})
+    set(XMRIG_ASM_SOURCES "crypto/CryptonightR_gen.c")
+    set_property(TARGET ${XMRIG_ASM_LIBRARY} PROPERTY LINKER_LANGUAGE C)
+else()
+    set(XMRIG_ASM_SOURCES "")
+    set(XMRIG_ASM_LIBRARY "")
+    add_definitions(/DXMRIG_NO_ASM)
+endif()

cmake/cpu.cmake

@@ -1,25 +0,0 @@
-if (NOT CMAKE_SYSTEM_PROCESSOR)
-    message(WARNING "CMAKE_SYSTEM_PROCESSOR not defined")
-endif()
-
-
-if (CMAKE_SYSTEM_PROCESSOR MATCHES "^(x86_64|AMD64)$")
-    add_definitions(/DRAPIDJSON_SSE2)
-endif()
-
-
-if (CMAKE_SYSTEM_PROCESSOR MATCHES "^(aarch64)$")
-    set(XMRIG_ARM     ON)
-    set(XMRIG_ARMv8   ON)
-    set(WITH_LIBCPUID OFF)
-
-    add_definitions(/DXMRIG_ARM)
-    add_definitions(/DXMRIG_ARMv8)
-elseif (CMAKE_SYSTEM_PROCESSOR MATCHES "^(armv7|armv7f|armv7s|armv7k|armv7-a|armv7l)$")
-    set(XMRIG_ARM     ON)
-    set(XMRIG_ARMv7   ON)
-    set(WITH_LIBCPUID OFF)
-
-    add_definitions(/DXMRIG_ARM)
-    add_definitions(/DXMRIG_ARMv7)
-endif()

cmake/flags.cmake

@@ -1,69 +0,0 @@
-set(CMAKE_CXX_STANDARD_REQUIRED ON)
-set(CMAKE_CXX_EXTENSIONS OFF)
-set(CMAKE_CXX_STANDARD 11)
-
-if ("${CMAKE_BUILD_TYPE}" STREQUAL "")
-    set(CMAKE_BUILD_TYPE Release)
-endif()
-
-if (CMAKE_CXX_COMPILER_ID MATCHES GNU)
-
-    set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -Wall -Wno-strict-aliasing")
-    set(CMAKE_C_FLAGS_RELEASE "${CMAKE_C_FLAGS_RELEASE} -Ofast -funroll-loops -fvariable-expansion-in-unroller -ftree-loop-if-convert-stores -fmerge-all-constants -fbranch-target-load-optimize2")
-
-    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wall -fno-exceptions -fno-rtti")
-    set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} -Ofast -s -funroll-loops -fvariable-expansion-in-unroller -ftree-loop-if-convert-stores -fmerge-all-constants -fbranch-target-load-optimize2")
-
-    if (XMRIG_ARMv8)
-        set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -march=armv8-a+crypto")
-        set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=armv8-a+crypto -flax-vector-conversions")
-    elseif (XMRIG_ARMv7)
-        set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -mfpu=neon")
-        set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -mfpu=neon -flax-vector-conversions")
-    else()
-        set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -maes")
-        set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -maes")
-    endif()
-
-    if (WIN32)
-        set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -static")
-    else()
-        set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -static-libgcc -static-libstdc++")
-    endif()
-
-    add_definitions(/D_GNU_SOURCE)
-
-    if (${CMAKE_VERSION} VERSION_LESS "3.1.0")
-        set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")
-    endif()
-
-    #set(CMAKE_C_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} -gdwarf-2")
-
-elseif (CMAKE_CXX_COMPILER_ID MATCHES MSVC)
-
-    set(CMAKE_C_FLAGS_RELEASE "${CMAKE_C_FLAGS_RELEASE} /Ox /Ot /Oi /MT /GL")
-    set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} /Ox /Ot /Oi /MT /GL")
-    add_definitions(/D_CRT_SECURE_NO_WARNINGS)
-    add_definitions(/D_CRT_NONSTDC_NO_WARNINGS)
-    add_definitions(/DNOMINMAX)
-
-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 "${CMAKE_CXX_FLAGS} -Wall -fno-exceptions -fno-rtti -Wno-missing-braces")
-    set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} -Ofast -funroll-loops -fmerge-all-constants")
-
-    if (XMRIG_ARMv8)
-        set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -march=armv8-a+crypto")
-        set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=armv8-a+crypto")
-    elseif (XMRIG_ARMv7)
-        set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -mfpu=neon -march=${CMAKE_SYSTEM_PROCESSOR}")
-        set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -mfpu=neon -march=${CMAKE_SYSTEM_PROCESSOR}")
-    else()
-        set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -maes")
-        set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -maes")
-    endif()
-
-endif()

compat.h

@@ -21,28 +21,27 @@
  *   along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-#ifndef __SUBMITRESULT_H__
-#define __SUBMITRESULT_H__
+#ifndef __COMPAT_H__
+#define __COMPAT_H__
 
+#define unlikely(expr) (__builtin_expect(!!(expr), 0))
+#define likely(expr)   (__builtin_expect(!!(expr), 1))
 
-#include <uv.h>
+#ifdef WIN32
 
+#include <windows.h>
 
-class SubmitResult
-{
-public:
-    inline SubmitResult() : seq(0), diff(0), actualDiff(0), elapsed(0), start(0) {}
-    SubmitResult(int64_t seq, uint32_t diff, uint64_t actualDiff);
+#define sleep(secs) Sleep((secs) * 1000)
 
-    void done();
-
-    int64_t seq;
-    uint32_t diff;
-    uint64_t actualDiff;
-    uint64_t elapsed;
-
-private:
-    uint64_t start;
+enum {
+	PRIO_PROCESS		= 0,
 };
 
-#endif /* __SUBMITRESULT_H__ */
+static inline int setpriority(int which, int who, int prio)
+{
+	return -!SetThreadPriority(GetCurrentThread(), THREAD_PRIORITY_IDLE);
+}
+
+#endif /* WIN32 */
+
+#endif /* __COMPAT_H__ */

compat/jansson/CMakeLists.txt

@@ -0,0 +1,25 @@
+cmake_minimum_required (VERSION 2.8)
+project (jansson C)
+
+add_definitions(-DHAVE_CONFIG_H)
+
+# Add the lib sources.
+file(GLOB JANSSON_SRC *.c)
+
+set(CMAKE_C_FLAGS_RELEASE "${CMAKE_C_FLAGS_RELEASE} -Os")
+
+set(JANSSON_HDR_PRIVATE
+   ${CMAKE_CURRENT_SOURCE_DIR}/hashtable.h
+   ${CMAKE_CURRENT_SOURCE_DIR}/jansson_private.h
+   ${CMAKE_CURRENT_SOURCE_DIR}/strbuffer.h
+   ${CMAKE_CURRENT_SOURCE_DIR}/utf.h
+   ${CMAKE_CURRENT_SOURCE_DIR}/jansson_private_config.h)
+
+set(JANSSON_HDR_PUBLIC
+   ${CMAKE_CURRENT_SOURCE_DIR}/jansson_config.h
+   ${CMAKE_CURRENT_SOURCE_DIR}/jansson.h)
+
+add_library(jansson STATIC
+    ${JANSSON_SRC}
+    ${JANSSON_HDR_PRIVATE}
+    ${JANSSON_HDR_PUBLIC})

compat/jansson/LICENSE

@@ -0,0 +1,19 @@
+Copyright (c) 2009-2014 Petri Lehtinen <petri@digip.org>
+
+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.

compat/jansson/dump.c

@@ -0,0 +1,496 @@
+/*
+ * Copyright (c) 2009-2016 Petri Lehtinen <petri@digip.org>
+ *
+ * Jansson is free software; you can redistribute it and/or modify
+ * it under the terms of the MIT license. See LICENSE for details.
+ */
+
+#ifndef _GNU_SOURCE
+#define _GNU_SOURCE
+#endif
+
+#include "jansson_private.h"
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <assert.h>
+#ifdef HAVE_UNISTD_H
+#include <unistd.h>
+#endif
+
+#include "jansson.h"
+#include "strbuffer.h"
+#include "utf.h"
+
+#define MAX_INTEGER_STR_LENGTH  100
+#define MAX_REAL_STR_LENGTH     100
+
+#define FLAGS_TO_INDENT(f)      ((f) & 0x1F)
+#define FLAGS_TO_PRECISION(f)   (((f) >> 11) & 0x1F)
+
+struct buffer {
+    const size_t size;
+    size_t used;
+    char *data;
+};
+
+static int dump_to_strbuffer(const char *buffer, size_t size, void *data)
+{
+    return strbuffer_append_bytes((strbuffer_t *)data, buffer, size);
+}
+
+static int dump_to_buffer(const char *buffer, size_t size, void *data)
+{
+    struct buffer *buf = (struct buffer *)data;
+
+    if(buf->used + size <= buf->size)
+        memcpy(&buf->data[buf->used], buffer, size);
+
+    buf->used += size;
+    return 0;
+}
+
+static int dump_to_file(const char *buffer, size_t size, void *data)
+{
+    FILE *dest = (FILE *)data;
+    if(fwrite(buffer, size, 1, dest) != 1)
+        return -1;
+    return 0;
+}
+
+static int dump_to_fd(const char *buffer, size_t size, void *data)
+{
+    int *dest = (int *)data;
+#ifdef HAVE_UNISTD_H
+    if(write(*dest, buffer, size) == (ssize_t)size)
+        return 0;
+#endif
+    return -1;
+}
+
+/* 32 spaces (the maximum indentation size) */
+static const char whitespace[] = "                                ";
+
+static int dump_indent(size_t flags, int depth, int space, json_dump_callback_t dump, void *data)
+{
+    if(FLAGS_TO_INDENT(flags) > 0)
+    {
+        unsigned int ws_count = FLAGS_TO_INDENT(flags), n_spaces = depth * ws_count;
+
+        if(dump("\n", 1, data))
+            return -1;
+
+        while(n_spaces > 0)
+        {
+            int cur_n = n_spaces < sizeof whitespace - 1 ? n_spaces : sizeof whitespace - 1;
+
+            if(dump(whitespace, cur_n, data))
+                return -1;
+
+            n_spaces -= cur_n;
+        }
+    }
+    else if(space && !(flags & JSON_COMPACT))
+    {
+        return dump(" ", 1, data);
+    }
+    return 0;
+}
+
+static int dump_string(const char *str, size_t len, json_dump_callback_t dump, void *data, size_t flags)
+{
+    const char *pos, *end, *lim;
+    int32_t codepoint;
+
+    if(dump("\"", 1, data))
+        return -1;
+
+    end = pos = str;
+    lim = str + len;
+    while(1)
+    {
+        const char *text;
+        char seq[13];
+        int length;
+
+        while(end < lim)
+        {
+            end = utf8_iterate(pos, lim - pos, &codepoint);
+            if(!end)
+                return -1;
+
+            /* mandatory escape or control char */
+            if(codepoint == '\\' || codepoint == '"' || codepoint < 0x20)
+                break;
+
+            /* slash */
+            if((flags & JSON_ESCAPE_SLASH) && codepoint == '/')
+                break;
+
+            /* non-ASCII */
+            if((flags & JSON_ENSURE_ASCII) && codepoint > 0x7F)
+                break;
+
+            pos = end;
+        }
+
+        if(pos != str) {
+            if(dump(str, pos - str, data))
+                return -1;
+        }
+
+        if(end == pos)
+            break;
+
+        /* handle \, /, ", and control codes */
+        length = 2;
+        switch(codepoint)
+        {
+            case '\\': text = "\\\\"; break;
+            case '\"': text = "\\\""; break;
+            case '\b': text = "\\b"; break;
+            case '\f': text = "\\f"; break;
+            case '\n': text = "\\n"; break;
+            case '\r': text = "\\r"; break;
+            case '\t': text = "\\t"; break;
+            case '/':  text = "\\/"; break;
+            default:
+            {
+                /* codepoint is in BMP */
+                if(codepoint < 0x10000)
+                {
+                    snprintf(seq, sizeof(seq), "\\u%04X", (unsigned int)codepoint);
+                    length = 6;
+                }
+
+                /* not in BMP -> construct a UTF-16 surrogate pair */
+                else
+                {
+                    int32_t first, last;
+
+                    codepoint -= 0x10000;
+                    first = 0xD800 | ((codepoint & 0xffc00) >> 10);
+                    last = 0xDC00 | (codepoint & 0x003ff);
+
+                    snprintf(seq, sizeof(seq), "\\u%04X\\u%04X", (unsigned int)first, (unsigned int)last);
+                    length = 12;
+                }
+
+                text = seq;
+                break;
+            }
+        }
+
+        if(dump(text, length, data))
+            return -1;
+
+        str = pos = end;
+    }
+
+    return dump("\"", 1, data);
+}
+
+static int compare_keys(const void *key1, const void *key2)
+{
+    return strcmp(*(const char **)key1, *(const char **)key2);
+}
+
+static int do_dump(const json_t *json, size_t flags, int depth,
+                   json_dump_callback_t dump, void *data)
+{
+    int embed = flags & JSON_EMBED;
+
+    flags &= ~JSON_EMBED;
+
+    if(!json)
+        return -1;
+
+    switch(json_typeof(json)) {
+        case JSON_NULL:
+            return dump("null", 4, data);
+
+        case JSON_TRUE:
+            return dump("true", 4, data);
+
+        case JSON_FALSE:
+            return dump("false", 5, data);
+
+        case JSON_INTEGER:
+        {
+            char buffer[MAX_INTEGER_STR_LENGTH];
+            int size;
+
+            size = snprintf(buffer, MAX_INTEGER_STR_LENGTH,
+                            "%" JSON_INTEGER_FORMAT,
+                            json_integer_value(json));
+            if(size < 0 || size >= MAX_INTEGER_STR_LENGTH)
+                return -1;
+
+            return dump(buffer, size, data);
+        }
+
+        case JSON_REAL:
+        {
+            char buffer[MAX_REAL_STR_LENGTH];
+            int size;
+            double value = json_real_value(json);
+
+            size = jsonp_dtostr(buffer, MAX_REAL_STR_LENGTH, value,
+                                FLAGS_TO_PRECISION(flags));
+            if(size < 0)
+                return -1;
+
+            return dump(buffer, size, data);
+        }
+
+        case JSON_STRING:
+            return dump_string(json_string_value(json), json_string_length(json), dump, data, flags);
+
+        case JSON_ARRAY:
+        {
+            size_t n;
+            size_t i;
+
+            json_array_t *array;
+
+            /* detect circular references */
+            array = json_to_array(json);
+            if(array->visited)
+                goto array_error;
+            array->visited = 1;
+
+            n = json_array_size(json);
+
+            if(!embed && dump("[", 1, data))
+                goto array_error;
+            if(n == 0) {
+                array->visited = 0;
+                return embed ? 0 : dump("]", 1, data);
+            }
+            if(dump_indent(flags, depth + 1, 0, dump, data))
+                goto array_error;
+
+            for(i = 0; i < n; ++i) {
+                if(do_dump(json_array_get(json, i), flags, depth + 1,
+                           dump, data))
+                    goto array_error;
+
+                if(i < n - 1)
+                {
+                    if(dump(",", 1, data) ||
+                       dump_indent(flags, depth + 1, 1, dump, data))
+                        goto array_error;
+                }
+                else
+                {
+                    if(dump_indent(flags, depth, 0, dump, data))
+                        goto array_error;
+                }
+            }
+
+            array->visited = 0;
+            return embed ? 0 : dump("]", 1, data);
+
+        array_error:
+            array->visited = 0;
+            return -1;
+        }
+
+        case JSON_OBJECT:
+        {
+            json_object_t *object;
+            void *iter;
+            const char *separator;
+            int separator_length;
+
+            if(flags & JSON_COMPACT) {
+                separator = ":";
+                separator_length = 1;
+            }
+            else {
+                separator = ": ";
+                separator_length = 2;
+            }
+
+            /* detect circular references */
+            object = json_to_object(json);
+            if(object->visited)
+                goto object_error;
+            object->visited = 1;
+
+            iter = json_object_iter((json_t *)json);
+
+            if(!embed && dump("{", 1, data))
+                goto object_error;
+            if(!iter) {
+                object->visited = 0;
+                return embed ? 0 : dump("}", 1, data);
+            }
+            if(dump_indent(flags, depth + 1, 0, dump, data))
+                goto object_error;
+
+            if(flags & JSON_SORT_KEYS)
+            {
+                const char **keys;
+                size_t size, i;
+
+                size = json_object_size(json);
+                keys = jsonp_malloc(size * sizeof(const char *));
+                if(!keys)
+                    goto object_error;
+
+                i = 0;
+                while(iter)
+                {
+                    keys[i] = json_object_iter_key(iter);
+                    iter = json_object_iter_next((json_t *)json, iter);
+                    i++;
+                }
+                assert(i == size);
+
+                qsort(keys, size, sizeof(const char *), compare_keys);
+
+                for(i = 0; i < size; i++)
+                {
+                    const char *key;
+                    json_t *value;
+
+                    key = keys[i];
+                    value = json_object_get(json, key);
+                    assert(value);
+
+                    dump_string(key, strlen(key), dump, data, flags);
+                    if(dump(separator, separator_length, data) ||
+                       do_dump(value, flags, depth + 1, dump, data))
+                    {
+                        jsonp_free(keys);
+                        goto object_error;
+                    }
+
+                    if(i < size - 1)
+                    {
+                        if(dump(",", 1, data) ||
+                           dump_indent(flags, depth + 1, 1, dump, data))
+                        {
+                            jsonp_free(keys);
+                            goto object_error;
+                        }
+                    }
+                    else
+                    {
+                        if(dump_indent(flags, depth, 0, dump, data))
+                        {
+                            jsonp_free(keys);
+                            goto object_error;
+                        }
+                    }
+                }
+
+                jsonp_free(keys);
+            }
+            else
+            {
+                /* Don't sort keys */
+
+                while(iter)
+                {
+                    void *next = json_object_iter_next((json_t *)json, iter);
+                    const char *key = json_object_iter_key(iter);
+
+                    dump_string(key, strlen(key), dump, data, flags);
+                    if(dump(separator, separator_length, data) ||
+                       do_dump(json_object_iter_value(iter), flags, depth + 1,
+                               dump, data))
+                        goto object_error;
+
+                    if(next)
+                    {
+                        if(dump(",", 1, data) ||
+                           dump_indent(flags, depth + 1, 1, dump, data))
+                            goto object_error;
+                    }
+                    else
+                    {
+                        if(dump_indent(flags, depth, 0, dump, data))
+                            goto object_error;
+                    }
+
+                    iter = next;
+                }
+            }
+
+            object->visited = 0;
+            return embed ? 0 : dump("}", 1, data);
+
+        object_error:
+            object->visited = 0;
+            return -1;
+        }
+
+        default:
+            /* not reached */
+            return -1;
+    }
+}
+
+char *json_dumps(const json_t *json, size_t flags)
+{
+    strbuffer_t strbuff;
+    char *result;
+
+    if(strbuffer_init(&strbuff))
+        return NULL;
+
+    if(json_dump_callback(json, dump_to_strbuffer, (void *)&strbuff, flags))
+        result = NULL;
+    else
+        result = jsonp_strdup(strbuffer_value(&strbuff));
+
+    strbuffer_close(&strbuff);
+    return result;
+}
+
+size_t json_dumpb(const json_t *json, char *buffer, size_t size, size_t flags)
+{
+    struct buffer buf = { size, 0, buffer };
+
+    if(json_dump_callback(json, dump_to_buffer, (void *)&buf, flags))
+        return 0;
+
+    return buf.used;
+}
+
+int json_dumpf(const json_t *json, FILE *output, size_t flags)
+{
+    return json_dump_callback(json, dump_to_file, (void *)output, flags);
+}
+
+int json_dumpfd(const json_t *json, int output, size_t flags)
+{
+    return json_dump_callback(json, dump_to_fd, (void *)&output, flags);
+}
+
+int json_dump_file(const json_t *json, const char *path, size_t flags)
+{
+    int result;
+
+    FILE *output = fopen(path, "w");
+    if(!output)
+        return -1;
+
+    result = json_dumpf(json, output, flags);
+
+    fclose(output);
+    return result;
+}
+
+int json_dump_callback(const json_t *json, json_dump_callback_t callback, void *data, size_t flags)
+{
+    if(!(flags & JSON_ENCODE_ANY)) {
+        if(!json_is_array(json) && !json_is_object(json))
+           return -1;
+    }
+
+    return do_dump(json, flags, 0, callback, data);
+}

compat/jansson/error.c

@@ -0,0 +1,63 @@
+#include <string.h>
+#include "jansson_private.h"
+
+void jsonp_error_init(json_error_t *error, const char *source)
+{
+    if(error)
+    {
+        error->text[0] = '\0';
+        error->line = -1;
+        error->column = -1;
+        error->position = 0;
+        if(source)
+            jsonp_error_set_source(error, source);
+        else
+            error->source[0] = '\0';
+    }
+}
+
+void jsonp_error_set_source(json_error_t *error, const char *source)
+{
+    size_t length;
+
+    if(!error || !source)
+        return;
+
+    length = strlen(source);
+    if(length < JSON_ERROR_SOURCE_LENGTH)
+        strncpy(error->source, source, length + 1);
+    else {
+        size_t extra = length - JSON_ERROR_SOURCE_LENGTH + 4;
+        strncpy(error->source, "...", 3);
+        strncpy(error->source + 3, source + extra, length - extra + 1);
+    }
+}
+
+void jsonp_error_set(json_error_t *error, int line, int column,
+                     size_t position, const char *msg, ...)
+{
+    va_list ap;
+
+    va_start(ap, msg);
+    jsonp_error_vset(error, line, column, position, msg, ap);
+    va_end(ap);
+}
+
+void jsonp_error_vset(json_error_t *error, int line, int column,
+                      size_t position, const char *msg, va_list ap)
+{
+    if(!error)
+        return;
+
+    if(error->text[0] != '\0') {
+        /* error already set */
+        return;
+    }
+
+    error->line = line;
+    error->column = column;
+    error->position = (int)position;
+
+    vsnprintf(error->text, JSON_ERROR_TEXT_LENGTH, msg, ap);
+    error->text[JSON_ERROR_TEXT_LENGTH - 1] = '\0';
+}

compat/jansson/hashtable.c

@@ -0,0 +1,356 @@
+/*
+ * Copyright (c) 2009-2016 Petri Lehtinen <petri@digip.org>
+ *
+ * This library is free software; you can redistribute it and/or modify
+ * it under the terms of the MIT license. See LICENSE for details.
+ */
+
+#if HAVE_CONFIG_H
+#include <jansson_private_config.h>
+#endif
+
+#include <stdlib.h>
+#include <string.h>
+
+#if HAVE_STDINT_H
+#include <stdint.h>
+#endif
+
+#include <jansson_config.h>   /* for JSON_INLINE */
+#include "jansson_private.h"  /* for container_of() */
+#include "hashtable.h"
+
+#ifndef INITIAL_HASHTABLE_ORDER
+#define INITIAL_HASHTABLE_ORDER 3
+#endif
+
+typedef struct hashtable_list list_t;
+typedef struct hashtable_pair pair_t;
+typedef struct hashtable_bucket bucket_t;
+
+extern volatile uint32_t hashtable_seed;
+
+/* Implementation of the hash function */
+#include "lookup3.h"
+
+#define list_to_pair(list_)  container_of(list_, pair_t, list)
+#define ordered_list_to_pair(list_)  container_of(list_, pair_t, ordered_list)
+#define hash_str(key)        ((size_t)hashlittle((key), strlen(key), hashtable_seed))
+
+static JSON_INLINE void list_init(list_t *list)
+{
+    list->next = list;
+    list->prev = list;
+}
+
+static JSON_INLINE void list_insert(list_t *list, list_t *node)
+{
+    node->next = list;
+    node->prev = list->prev;
+    list->prev->next = node;
+    list->prev = node;
+}
+
+static JSON_INLINE void list_remove(list_t *list)
+{
+    list->prev->next = list->next;
+    list->next->prev = list->prev;
+}
+
+static JSON_INLINE int bucket_is_empty(hashtable_t *hashtable, bucket_t *bucket)
+{
+    return bucket->first == &hashtable->list && bucket->first == bucket->last;
+}
+
+static void insert_to_bucket(hashtable_t *hashtable, bucket_t *bucket,
+                             list_t *list)
+{
+    if(bucket_is_empty(hashtable, bucket))
+    {
+        list_insert(&hashtable->list, list);
+        bucket->first = bucket->last = list;
+    }
+    else
+    {
+        list_insert(bucket->first, list);
+        bucket->first = list;
+    }
+}
+
+static pair_t *hashtable_find_pair(hashtable_t *hashtable, bucket_t *bucket,
+                                   const char *key, size_t hash)
+{
+    list_t *list;
+    pair_t *pair;
+
+    if(bucket_is_empty(hashtable, bucket))
+        return NULL;
+
+    list = bucket->first;
+    while(1)
+    {
+        pair = list_to_pair(list);
+        if(pair->hash == hash && strcmp(pair->key, key) == 0)
+            return pair;
+
+        if(list == bucket->last)
+            break;
+
+        list = list->next;
+    }
+
+    return NULL;
+}
+
+/* returns 0 on success, -1 if key was not found */
+static int hashtable_do_del(hashtable_t *hashtable,
+                            const char *key, size_t hash)
+{
+    pair_t *pair;
+    bucket_t *bucket;
+    size_t index;
+
+    index = hash & hashmask(hashtable->order);
+    bucket = &hashtable->buckets[index];
+
+    pair = hashtable_find_pair(hashtable, bucket, key, hash);
+    if(!pair)
+        return -1;
+
+    if(&pair->list == bucket->first && &pair->list == bucket->last)
+        bucket->first = bucket->last = &hashtable->list;
+
+    else if(&pair->list == bucket->first)
+        bucket->first = pair->list.next;
+
+    else if(&pair->list == bucket->last)
+        bucket->last = pair->list.prev;
+
+    list_remove(&pair->list);
+    list_remove(&pair->ordered_list);
+    json_decref(pair->value);
+
+    jsonp_free(pair);
+    hashtable->size--;
+
+    return 0;
+}
+
+static void hashtable_do_clear(hashtable_t *hashtable)
+{
+    list_t *list, *next;
+    pair_t *pair;
+
+    for(list = hashtable->list.next; list != &hashtable->list; list = next)
+    {
+        next = list->next;
+        pair = list_to_pair(list);
+        json_decref(pair->value);
+        jsonp_free(pair);
+    }
+}
+
+static int hashtable_do_rehash(hashtable_t *hashtable)
+{
+    list_t *list, *next;
+    pair_t *pair;
+    size_t i, index, new_size, new_order;
+    struct hashtable_bucket *new_buckets;
+
+    new_order = hashtable->order + 1;
+    new_size = hashsize(new_order);
+
+    new_buckets = jsonp_malloc(new_size * sizeof(bucket_t));
+    if(!new_buckets)
+        return -1;
+
+    jsonp_free(hashtable->buckets);
+    hashtable->buckets = new_buckets;
+    hashtable->order = new_order;
+
+    for(i = 0; i < hashsize(hashtable->order); i++)
+    {
+        hashtable->buckets[i].first = hashtable->buckets[i].last =
+            &hashtable->list;
+    }
+
+    list = hashtable->list.next;
+    list_init(&hashtable->list);
+
+    for(; list != &hashtable->list; list = next) {
+        next = list->next;
+        pair = list_to_pair(list);
+        index = pair->hash % new_size;
+        insert_to_bucket(hashtable, &hashtable->buckets[index], &pair->list);
+    }
+
+    return 0;
+}
+
+
+int hashtable_init(hashtable_t *hashtable)
+{
+    size_t i;
+
+    hashtable->size = 0;
+    hashtable->order = INITIAL_HASHTABLE_ORDER;
+    hashtable->buckets = jsonp_malloc(hashsize(hashtable->order) * sizeof(bucket_t));
+    if(!hashtable->buckets)
+        return -1;
+
+    list_init(&hashtable->list);
+    list_init(&hashtable->ordered_list);
+
+    for(i = 0; i < hashsize(hashtable->order); i++)
+    {
+        hashtable->buckets[i].first = hashtable->buckets[i].last =
+            &hashtable->list;
+    }
+
+    return 0;
+}
+
+void hashtable_close(hashtable_t *hashtable)
+{
+    hashtable_do_clear(hashtable);
+    jsonp_free(hashtable->buckets);
+}
+
+int hashtable_set(hashtable_t *hashtable, const char *key, json_t *value)
+{
+    pair_t *pair;
+    bucket_t *bucket;
+    size_t hash, index;
+
+    /* rehash if the load ratio exceeds 1 */
+    if(hashtable->size >= hashsize(hashtable->order))
+        if(hashtable_do_rehash(hashtable))
+            return -1;
+
+    hash = hash_str(key);
+    index = hash & hashmask(hashtable->order);
+    bucket = &hashtable->buckets[index];
+    pair = hashtable_find_pair(hashtable, bucket, key, hash);
+
+    if(pair)
+    {
+        json_decref(pair->value);
+        pair->value = value;
+    }
+    else
+    {
+        /* offsetof(...) returns the size of pair_t without the last,
+           flexible member. This way, the correct amount is
+           allocated. */
+
+        size_t len = strlen(key);
+        if(len >= (size_t)-1 - offsetof(pair_t, key)) {
+            /* Avoid an overflow if the key is very long */
+            return -1;
+        }
+
+        pair = jsonp_malloc(offsetof(pair_t, key) + len + 1);
+        if(!pair)
+            return -1;
+
+        pair->hash = hash;
+        strncpy(pair->key, key, len + 1);
+        pair->value = value;
+        list_init(&pair->list);
+        list_init(&pair->ordered_list);
+
+        insert_to_bucket(hashtable, bucket, &pair->list);
+        list_insert(&hashtable->ordered_list, &pair->ordered_list);
+
+        hashtable->size++;
+    }
+    return 0;
+}
+
+void *hashtable_get(hashtable_t *hashtable, const char *key)
+{
+    pair_t *pair;
+    size_t hash;
+    bucket_t *bucket;
+
+    hash = hash_str(key);
+    bucket = &hashtable->buckets[hash & hashmask(hashtable->order)];
+
+    pair = hashtable_find_pair(hashtable, bucket, key, hash);
+    if(!pair)
+        return NULL;
+
+    return pair->value;
+}
+
+int hashtable_del(hashtable_t *hashtable, const char *key)
+{
+    size_t hash = hash_str(key);
+    return hashtable_do_del(hashtable, key, hash);
+}
+
+void hashtable_clear(hashtable_t *hashtable)
+{
+    size_t i;
+
+    hashtable_do_clear(hashtable);
+
+    for(i = 0; i < hashsize(hashtable->order); i++)
+    {
+        hashtable->buckets[i].first = hashtable->buckets[i].last =
+            &hashtable->list;
+    }
+
+    list_init(&hashtable->list);
+    list_init(&hashtable->ordered_list);
+    hashtable->size = 0;
+}
+
+void *hashtable_iter(hashtable_t *hashtable)
+{
+    return hashtable_iter_next(hashtable, &hashtable->ordered_list);
+}
+
+void *hashtable_iter_at(hashtable_t *hashtable, const char *key)
+{
+    pair_t *pair;
+    size_t hash;
+    bucket_t *bucket;
+
+    hash = hash_str(key);
+    bucket = &hashtable->buckets[hash & hashmask(hashtable->order)];
+
+    pair = hashtable_find_pair(hashtable, bucket, key, hash);
+    if(!pair)
+        return NULL;
+
+    return &pair->ordered_list;
+}
+
+void *hashtable_iter_next(hashtable_t *hashtable, void *iter)
+{
+    list_t *list = (list_t *)iter;
+    if(list->next == &hashtable->ordered_list)
+        return NULL;
+    return list->next;
+}
+
+void *hashtable_iter_key(void *iter)
+{
+    pair_t *pair = ordered_list_to_pair((list_t *)iter);
+    return pair->key;
+}
+
+void *hashtable_iter_value(void *iter)
+{
+    pair_t *pair = ordered_list_to_pair((list_t *)iter);
+    return pair->value;
+}
+
+void hashtable_iter_set(void *iter, json_t *value)
+{
+    pair_t *pair = ordered_list_to_pair((list_t *)iter);
+
+    json_decref(pair->value);
+    pair->value = value;
+}

compat/jansson/hashtable.h

@@ -0,0 +1,176 @@
+/*
+ * Copyright (c) 2009-2016 Petri Lehtinen <petri@digip.org>
+ *
+ * This library is free software; you can redistribute it and/or modify
+ * it under the terms of the MIT license. See LICENSE for details.
+ */
+
+#ifndef HASHTABLE_H
+#define HASHTABLE_H
+
+#include <stdlib.h>
+#include "jansson.h"
+
+struct hashtable_list {
+    struct hashtable_list *prev;
+    struct hashtable_list *next;
+};
+
+/* "pair" may be a bit confusing a name, but think of it as a
+   key-value pair. In this case, it just encodes some extra data,
+   too */
+struct hashtable_pair {
+    struct hashtable_list list;
+    struct hashtable_list ordered_list;
+    size_t hash;
+    json_t *value;
+    char key[1];
+};
+
+struct hashtable_bucket {
+    struct hashtable_list *first;
+    struct hashtable_list *last;
+};
+
+typedef struct hashtable {
+    size_t size;
+    struct hashtable_bucket *buckets;
+    size_t order;  /* hashtable has pow(2, order) buckets */
+    struct hashtable_list list;
+    struct hashtable_list ordered_list;
+} hashtable_t;
+
+
+#define hashtable_key_to_iter(key_) \
+    (&(container_of(key_, struct hashtable_pair, key)->ordered_list))
+
+
+/**
+ * hashtable_init - Initialize a hashtable object
+ *
+ * @hashtable: The (statically allocated) hashtable object
+ *
+ * Initializes a statically allocated hashtable object. The object
+ * should be cleared with hashtable_close when it's no longer used.
+ *
+ * Returns 0 on success, -1 on error (out of memory).
+ */
+int hashtable_init(hashtable_t *hashtable);
+
+/**
+ * hashtable_close - Release all resources used by a hashtable object
+ *
+ * @hashtable: The hashtable
+ *
+ * Destroys a statically allocated hashtable object.
+ */
+void hashtable_close(hashtable_t *hashtable);
+
+/**
+ * hashtable_set - Add/modify value in hashtable
+ *
+ * @hashtable: The hashtable object
+ * @key: The key
+ * @serial: For addition order of keys
+ * @value: The value
+ *
+ * If a value with the given key already exists, its value is replaced
+ * with the new value. Value is "stealed" in the sense that hashtable
+ * doesn't increment its refcount but decreases the refcount when the
+ * value is no longer needed.
+ *
+ * Returns 0 on success, -1 on failure (out of memory).
+ */
+int hashtable_set(hashtable_t *hashtable, const char *key, json_t *value);
+
+/**
+ * hashtable_get - Get a value associated with a key
+ *
+ * @hashtable: The hashtable object
+ * @key: The key
+ *
+ * Returns value if it is found, or NULL otherwise.
+ */
+void *hashtable_get(hashtable_t *hashtable, const char *key);
+
+/**
+ * hashtable_del - Remove a value from the hashtable
+ *
+ * @hashtable: The hashtable object
+ * @key: The key
+ *
+ * Returns 0 on success, or -1 if the key was not found.
+ */
+int hashtable_del(hashtable_t *hashtable, const char *key);
+
+/**
+ * hashtable_clear - Clear hashtable
+ *
+ * @hashtable: The hashtable object
+ *
+ * Removes all items from the hashtable.
+ */
+void hashtable_clear(hashtable_t *hashtable);
+
+/**
+ * hashtable_iter - Iterate over hashtable
+ *
+ * @hashtable: The hashtable object
+ *
+ * Returns an opaque iterator to the first element in the hashtable.
+ * The iterator should be passed to hashtable_iter_* functions.
+ * The hashtable items are not iterated over in any particular order.
+ *
+ * There's no need to free the iterator in any way. The iterator is
+ * valid as long as the item that is referenced by the iterator is not
+ * deleted. Other values may be added or deleted. In particular,
+ * hashtable_iter_next() may be called on an iterator, and after that
+ * the key/value pair pointed by the old iterator may be deleted.
+ */
+void *hashtable_iter(hashtable_t *hashtable);
+
+/**
+ * hashtable_iter_at - Return an iterator at a specific key
+ *
+ * @hashtable: The hashtable object
+ * @key: The key that the iterator should point to
+ *
+ * Like hashtable_iter() but returns an iterator pointing to a
+ * specific key.
+ */
+void *hashtable_iter_at(hashtable_t *hashtable, const char *key);
+
+/**
+ * hashtable_iter_next - Advance an iterator
+ *
+ * @hashtable: The hashtable object
+ * @iter: The iterator
+ *
+ * Returns a new iterator pointing to the next element in the
+ * hashtable or NULL if the whole hastable has been iterated over.
+ */
+void *hashtable_iter_next(hashtable_t *hashtable, void *iter);
+
+/**
+ * hashtable_iter_key - Retrieve the key pointed by an iterator
+ *
+ * @iter: The iterator
+ */
+void *hashtable_iter_key(void *iter);
+
+/**
+ * hashtable_iter_value - Retrieve the value pointed by an iterator
+ *
+ * @iter: The iterator
+ */
+void *hashtable_iter_value(void *iter);
+
+/**
+ * hashtable_iter_set - Set the value pointed by an iterator
+ *
+ * @iter: The iterator
+ * @value: The value to set
+ */
+void hashtable_iter_set(void *iter, json_t *value);
+
+#endif

compat/jansson/hashtable_seed.c

@@ -0,0 +1,277 @@
+/* Generate sizeof(uint32_t) bytes of as random data as possible to seed
+   the hash function.
+*/
+
+#ifdef HAVE_CONFIG_H
+#include <jansson_private_config.h>
+#endif
+
+#include <stdio.h>
+#include <time.h>
+
+#ifdef HAVE_STDINT_H
+#include <stdint.h>
+#endif
+
+#ifdef HAVE_FCNTL_H
+#include <fcntl.h>
+#endif
+
+#ifdef HAVE_SCHED_H
+#include <sched.h>
+#endif
+
+#ifdef HAVE_UNISTD_H
+#include <unistd.h>
+#endif
+
+#ifdef HAVE_SYS_STAT_H
+#include <sys/stat.h>
+#endif
+
+#ifdef HAVE_SYS_TIME_H
+#include <sys/time.h>
+#endif
+
+#ifdef HAVE_SYS_TYPES_H
+#include <sys/types.h>
+#endif
+
+#if defined(_WIN32)
+/* For GetModuleHandle(), GetProcAddress() and GetCurrentProcessId() */
+#include <windows.h>
+#endif
+
+#include "jansson.h"
+
+
+static uint32_t buf_to_uint32(char *data) {
+    size_t i;
+    uint32_t result = 0;
+
+    for (i = 0; i < sizeof(uint32_t); i++)
+        result = (result << 8) | (unsigned char)data[i];
+
+    return result;
+}
+
+
+
+/* /dev/urandom */
+#if !defined(_WIN32) && defined(USE_URANDOM)
+static int seed_from_urandom(uint32_t *seed) {
+    /* Use unbuffered I/O if we have open(), close() and read(). Otherwise
+       fall back to fopen() */
+
+    char data[sizeof(uint32_t)];
+    int ok;
+
+#if defined(HAVE_OPEN) && defined(HAVE_CLOSE) && defined(HAVE_READ)
+    int urandom;
+    urandom = open("/dev/urandom", O_RDONLY);
+    if (urandom == -1)
+        return 1;
+
+    ok = read(urandom, data, sizeof(uint32_t)) == sizeof(uint32_t);
+    close(urandom);
+#else
+    FILE *urandom;
+
+    urandom = fopen("/dev/urandom", "rb");
+    if (!urandom)
+        return 1;
+
+    ok = fread(data, 1, sizeof(uint32_t), urandom) == sizeof(uint32_t);
+    fclose(urandom);
+#endif
+
+    if (!ok)
+        return 1;
+
+    *seed = buf_to_uint32(data);
+    return 0;
+}
+#endif
+
+/* Windows Crypto API */
+#if defined(_WIN32) && defined(USE_WINDOWS_CRYPTOAPI)
+#include <wincrypt.h>
+
+typedef BOOL (WINAPI *CRYPTACQUIRECONTEXTA)(HCRYPTPROV *phProv, LPCSTR pszContainer, LPCSTR pszProvider, DWORD dwProvType, DWORD dwFlags);
+typedef BOOL (WINAPI *CRYPTGENRANDOM)(HCRYPTPROV hProv, DWORD dwLen, BYTE *pbBuffer);
+typedef BOOL (WINAPI *CRYPTRELEASECONTEXT)(HCRYPTPROV hProv, DWORD dwFlags);
+
+static int seed_from_windows_cryptoapi(uint32_t *seed)
+{
+    HINSTANCE hAdvAPI32 = NULL;
+    CRYPTACQUIRECONTEXTA pCryptAcquireContext = NULL;
+    CRYPTGENRANDOM pCryptGenRandom = NULL;
+    CRYPTRELEASECONTEXT pCryptReleaseContext = NULL;
+    HCRYPTPROV hCryptProv = 0;
+    BYTE data[sizeof(uint32_t)];
+    int ok;
+
+    hAdvAPI32 = GetModuleHandle(TEXT("advapi32.dll"));
+    if(hAdvAPI32 == NULL)
+        return 1;
+
+    pCryptAcquireContext = (CRYPTACQUIRECONTEXTA)GetProcAddress(hAdvAPI32, "CryptAcquireContextA");
+    if (!pCryptAcquireContext)
+        return 1;
+
+    pCryptGenRandom = (CRYPTGENRANDOM)GetProcAddress(hAdvAPI32, "CryptGenRandom");
+    if (!pCryptGenRandom)
+        return 1;
+
+    pCryptReleaseContext = (CRYPTRELEASECONTEXT)GetProcAddress(hAdvAPI32, "CryptReleaseContext");
+    if (!pCryptReleaseContext)
+        return 1;
+
+    if (!pCryptAcquireContext(&hCryptProv, NULL, NULL, PROV_RSA_FULL, CRYPT_VERIFYCONTEXT))
+        return 1;
+
+    ok = pCryptGenRandom(hCryptProv, sizeof(uint32_t), data);
+    pCryptReleaseContext(hCryptProv, 0);
+
+    if (!ok)
+        return 1;
+
+    *seed = buf_to_uint32((char *)data);
+    return 0;
+}
+#endif
+
+/* gettimeofday() and getpid() */
+static int seed_from_timestamp_and_pid(uint32_t *seed) {
+#ifdef HAVE_GETTIMEOFDAY
+    /* XOR of seconds and microseconds */
+    struct timeval tv;
+    gettimeofday(&tv, NULL);
+    *seed = (uint32_t)tv.tv_sec ^ (uint32_t)tv.tv_usec;
+#else
+    /* Seconds only */
+    *seed = (uint32_t)time(NULL);
+#endif
+
+    /* XOR with PID for more randomness */
+#if defined(_WIN32)
+    *seed ^= (uint32_t)GetCurrentProcessId();
+#elif defined(HAVE_GETPID)
+    *seed ^= (uint32_t)getpid();
+#endif
+
+    return 0;
+}
+
+static uint32_t generate_seed() {
+    uint32_t seed;
+    int done = 0;
+
+#if !defined(_WIN32) && defined(USE_URANDOM)
+    if (seed_from_urandom(&seed) == 0)
+        done = 1;
+#endif
+
+#if defined(_WIN32) && defined(USE_WINDOWS_CRYPTOAPI)
+    if (seed_from_windows_cryptoapi(&seed) == 0)
+        done = 1;
+#endif
+
+    if (!done) {
+        /* Fall back to timestamp and PID if no better randomness is
+           available */
+        seed_from_timestamp_and_pid(&seed);
+    }
+
+    /* Make sure the seed is never zero */
+    if (seed == 0)
+        seed = 1;
+
+    return seed;
+}
+
+
+volatile uint32_t hashtable_seed = 0;
+
+#if defined(HAVE_ATOMIC_BUILTINS) && (defined(HAVE_SCHED_YIELD) || !defined(_WIN32))
+static volatile char seed_initialized = 0;
+
+void json_object_seed(size_t seed) {
+    uint32_t new_seed = (uint32_t)seed;
+
+    if (hashtable_seed == 0) {
+        if (__atomic_test_and_set(&seed_initialized, __ATOMIC_RELAXED) == 0) {
+            /* Do the seeding ourselves */
+            if (new_seed == 0)
+                new_seed = generate_seed();
+
+            __atomic_store_n(&hashtable_seed, new_seed, __ATOMIC_RELEASE);
+        } else {
+            /* Wait for another thread to do the seeding */
+            do {
+#ifdef HAVE_SCHED_YIELD
+                sched_yield();
+#endif
+            } while(__atomic_load_n(&hashtable_seed, __ATOMIC_ACQUIRE) == 0);
+        }
+    }
+}
+#elif defined(HAVE_SYNC_BUILTINS) && (defined(HAVE_SCHED_YIELD) || !defined(_WIN32))
+void json_object_seed(size_t seed) {
+    uint32_t new_seed = (uint32_t)seed;
+
+    if (hashtable_seed == 0) {
+        if (new_seed == 0) {
+            /* Explicit synchronization fences are not supported by the
+               __sync builtins, so every thread getting here has to
+               generate the seed value.
+            */
+            new_seed = generate_seed();
+        }
+
+        do {
+            if (__sync_bool_compare_and_swap(&hashtable_seed, 0, new_seed)) {
+                /* We were the first to seed */
+                break;
+            } else {
+                /* Wait for another thread to do the seeding */
+#ifdef HAVE_SCHED_YIELD
+                sched_yield();
+#endif
+            }
+        } while(hashtable_seed == 0);
+    }
+}
+#elif defined(_WIN32)
+static long seed_initialized = 0;
+void json_object_seed(size_t seed) {
+    uint32_t new_seed = (uint32_t)seed;
+
+    if (hashtable_seed == 0) {
+        if (InterlockedIncrement(&seed_initialized) == 1) {
+            /* Do the seeding ourselves */
+            if (new_seed == 0)
+                new_seed = generate_seed();
+
+            hashtable_seed = new_seed;
+        } else {
+            /* Wait for another thread to do the seeding */
+            do {
+                SwitchToThread();
+            } while (hashtable_seed == 0);
+        }
+    }
+}
+#else
+/* Fall back to a thread-unsafe version */
+void json_object_seed(size_t seed) {
+    uint32_t new_seed = (uint32_t)seed;
+
+    if (hashtable_seed == 0) {
+        if (new_seed == 0)
+            new_seed = generate_seed();
+
+        hashtable_seed = new_seed;
+    }
+}
+#endif

compat/jansson/jansson.h

@@ -0,0 +1,315 @@
+/*
+ * Copyright (c) 2009-2016 Petri Lehtinen <petri@digip.org>
+ *
+ * Jansson is free software; you can redistribute it and/or modify
+ * it under the terms of the MIT license. See LICENSE for details.
+ */
+
+#ifndef JANSSON_H
+#define JANSSON_H
+
+#include <stdio.h>
+#include <stdlib.h>  /* for size_t */
+#include <stdarg.h>
+
+#include "jansson_config.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/* version */
+
+#define JANSSON_MAJOR_VERSION  2
+#define JANSSON_MINOR_VERSION  10
+#define JANSSON_MICRO_VERSION  0
+
+/* Micro version is omitted if it's 0 */
+#define JANSSON_VERSION  "2.10"
+
+/* Version as a 3-byte hex number, e.g. 0x010201 == 1.2.1. Use this
+   for numeric comparisons, e.g. #if JANSSON_VERSION_HEX >= ... */
+#define JANSSON_VERSION_HEX  ((JANSSON_MAJOR_VERSION << 16) |   \
+                              (JANSSON_MINOR_VERSION << 8)  |   \
+                              (JANSSON_MICRO_VERSION << 0))
+
+
+/* types */
+
+typedef enum {
+    JSON_OBJECT,
+    JSON_ARRAY,
+    JSON_STRING,
+    JSON_INTEGER,
+    JSON_REAL,
+    JSON_TRUE,
+    JSON_FALSE,
+    JSON_NULL
+} json_type;
+
+typedef struct json_t {
+    json_type type;
+    size_t refcount;
+} json_t;
+
+#ifndef JANSSON_USING_CMAKE /* disabled if using cmake */
+#if JSON_INTEGER_IS_LONG_LONG
+#ifdef _WIN32
+#define JSON_INTEGER_FORMAT "I64d"
+#else
+#define JSON_INTEGER_FORMAT "lld"
+#endif
+typedef long long json_int_t;
+#else
+#define JSON_INTEGER_FORMAT "ld"
+typedef long json_int_t;
+#endif /* JSON_INTEGER_IS_LONG_LONG */
+#endif
+
+#define json_typeof(json)      ((json)->type)
+#define json_is_object(json)   ((json) && json_typeof(json) == JSON_OBJECT)
+#define json_is_array(json)    ((json) && json_typeof(json) == JSON_ARRAY)
+#define json_is_string(json)   ((json) && json_typeof(json) == JSON_STRING)
+#define json_is_integer(json)  ((json) && json_typeof(json) == JSON_INTEGER)
+#define json_is_real(json)     ((json) && json_typeof(json) == JSON_REAL)
+#define json_is_number(json)   (json_is_integer(json) || json_is_real(json))
+#define json_is_true(json)     ((json) && json_typeof(json) == JSON_TRUE)
+#define json_is_false(json)    ((json) && json_typeof(json) == JSON_FALSE)
+#define json_boolean_value     json_is_true
+#define json_is_boolean(json)  (json_is_true(json) || json_is_false(json))
+#define json_is_null(json)     ((json) && json_typeof(json) == JSON_NULL)
+
+/* construction, destruction, reference counting */
+
+json_t *json_object(void);
+json_t *json_array(void);
+json_t *json_string(const char *value);
+json_t *json_stringn(const char *value, size_t len);
+json_t *json_string_nocheck(const char *value);
+json_t *json_stringn_nocheck(const char *value, size_t len);
+json_t *json_integer(json_int_t value);
+json_t *json_real(double value);
+json_t *json_true(void);
+json_t *json_false(void);
+#define json_boolean(val)      ((val) ? json_true() : json_false())
+json_t *json_null(void);
+
+static JSON_INLINE
+json_t *json_incref(json_t *json)
+{
+    if(json && json->refcount != (size_t)-1)
+        ++json->refcount;
+    return json;
+}
+
+/* do not call json_delete directly */
+void json_delete(json_t *json);
+
+static JSON_INLINE
+void json_decref(json_t *json)
+{
+    if(json && json->refcount != (size_t)-1 && --json->refcount == 0)
+        json_delete(json);
+}
+
+#if defined(__GNUC__) || defined(__clang__)
+static JSON_INLINE
+void json_decrefp(json_t **json)
+{
+    if(json) {
+        json_decref(*json);
+	*json = NULL;
+    }
+}
+
+#define json_auto_t json_t __attribute__((cleanup(json_decrefp)))
+#endif
+
+
+/* error reporting */
+
+#define JSON_ERROR_TEXT_LENGTH    160
+#define JSON_ERROR_SOURCE_LENGTH   80
+
+typedef struct {
+    int line;
+    int column;
+    int position;
+    char source[JSON_ERROR_SOURCE_LENGTH];
+    char text[JSON_ERROR_TEXT_LENGTH];
+} json_error_t;
+
+
+/* getters, setters, manipulation */
+
+void json_object_seed(size_t seed);
+size_t json_object_size(const json_t *object);
+json_t *json_object_get(const json_t *object, const char *key);
+int json_object_set_new(json_t *object, const char *key, json_t *value);
+int json_object_set_new_nocheck(json_t *object, const char *key, json_t *value);
+int json_object_del(json_t *object, const char *key);
+int json_object_clear(json_t *object);
+int json_object_update(json_t *object, json_t *other);
+int json_object_update_existing(json_t *object, json_t *other);
+int json_object_update_missing(json_t *object, json_t *other);
+void *json_object_iter(json_t *object);
+void *json_object_iter_at(json_t *object, const char *key);
+void *json_object_key_to_iter(const char *key);
+void *json_object_iter_next(json_t *object, void *iter);
+const char *json_object_iter_key(void *iter);
+json_t *json_object_iter_value(void *iter);
+int json_object_iter_set_new(json_t *object, void *iter, json_t *value);
+
+#define json_object_foreach(object, key, value) \
+    for(key = json_object_iter_key(json_object_iter(object)); \
+        key && (value = json_object_iter_value(json_object_key_to_iter(key))); \
+        key = json_object_iter_key(json_object_iter_next(object, json_object_key_to_iter(key))))
+
+#define json_object_foreach_safe(object, n, key, value)     \
+    for(key = json_object_iter_key(json_object_iter(object)), \
+            n = json_object_iter_next(object, json_object_key_to_iter(key)); \
+        key && (value = json_object_iter_value(json_object_key_to_iter(key))); \
+        key = json_object_iter_key(n), \
+            n = json_object_iter_next(object, json_object_key_to_iter(key)))
+
+#define json_array_foreach(array, index, value) \
+	for(index = 0; \
+		index < json_array_size(array) && (value = json_array_get(array, index)); \
+		index++)
+
+static JSON_INLINE
+int json_object_set(json_t *object, const char *key, json_t *value)
+{
+    return json_object_set_new(object, key, json_incref(value));
+}
+
+static JSON_INLINE
+int json_object_set_nocheck(json_t *object, const char *key, json_t *value)
+{
+    return json_object_set_new_nocheck(object, key, json_incref(value));
+}
+
+static JSON_INLINE
+int json_object_iter_set(json_t *object, void *iter, json_t *value)
+{
+    return json_object_iter_set_new(object, iter, json_incref(value));
+}
+
+size_t json_array_size(const json_t *array);
+json_t *json_array_get(const json_t *array, size_t index);
+int json_array_set_new(json_t *array, size_t index, json_t *value);
+int json_array_append_new(json_t *array, json_t *value);
+int json_array_insert_new(json_t *array, size_t index, json_t *value);
+int json_array_remove(json_t *array, size_t index);
+int json_array_clear(json_t *array);
+int json_array_extend(json_t *array, json_t *other);
+
+static JSON_INLINE
+int json_array_set(json_t *array, size_t ind, json_t *value)
+{
+    return json_array_set_new(array, ind, json_incref(value));
+}
+
+static JSON_INLINE
+int json_array_append(json_t *array, json_t *value)
+{
+    return json_array_append_new(array, json_incref(value));
+}
+
+static JSON_INLINE
+int json_array_insert(json_t *array, size_t ind, json_t *value)
+{
+    return json_array_insert_new(array, ind, json_incref(value));
+}
+
+const char *json_string_value(const json_t *string);
+size_t json_string_length(const json_t *string);
+json_int_t json_integer_value(const json_t *integer);
+double json_real_value(const json_t *real);
+double json_number_value(const json_t *json);
+
+int json_string_set(json_t *string, const char *value);
+int json_string_setn(json_t *string, const char *value, size_t len);
+int json_string_set_nocheck(json_t *string, const char *value);
+int json_string_setn_nocheck(json_t *string, const char *value, size_t len);
+int json_integer_set(json_t *integer, json_int_t value);
+int json_real_set(json_t *real, double value);
+
+/* pack, unpack */
+
+json_t *json_pack(const char *fmt, ...);
+json_t *json_pack_ex(json_error_t *error, size_t flags, const char *fmt, ...);
+json_t *json_vpack_ex(json_error_t *error, size_t flags, const char *fmt, va_list ap);
+
+#define JSON_VALIDATE_ONLY  0x1
+#define JSON_STRICT         0x2
+
+int json_unpack(json_t *root, const char *fmt, ...);
+int json_unpack_ex(json_t *root, json_error_t *error, size_t flags, const char *fmt, ...);
+int json_vunpack_ex(json_t *root, json_error_t *error, size_t flags, const char *fmt, va_list ap);
+
+
+/* equality */
+
+int json_equal(json_t *value1, json_t *value2);
+
+
+/* copying */
+
+json_t *json_copy(json_t *value);
+json_t *json_deep_copy(const json_t *value);
+
+
+/* decoding */
+
+#define JSON_REJECT_DUPLICATES  0x1
+#define JSON_DISABLE_EOF_CHECK  0x2
+#define JSON_DECODE_ANY         0x4
+#define JSON_DECODE_INT_AS_REAL 0x8
+#define JSON_ALLOW_NUL          0x10
+
+typedef size_t (*json_load_callback_t)(void *buffer, size_t buflen, void *data);
+
+json_t *json_loads(const char *input, size_t flags, json_error_t *error);
+json_t *json_loadb(const char *buffer, size_t buflen, size_t flags, json_error_t *error);
+json_t *json_loadf(FILE *input, size_t flags, json_error_t *error);
+json_t *json_loadfd(int input, size_t flags, json_error_t *error);
+json_t *json_load_file(const char *path, size_t flags, json_error_t *error);
+json_t *json_load_callback(json_load_callback_t callback, void *data, size_t flags, json_error_t *error);
+
+
+/* encoding */
+
+#define JSON_MAX_INDENT         0x1F
+#define JSON_INDENT(n)          ((n) & JSON_MAX_INDENT)
+#define JSON_COMPACT            0x20
+#define JSON_ENSURE_ASCII       0x40
+#define JSON_SORT_KEYS          0x80
+#define JSON_PRESERVE_ORDER     0x100
+#define JSON_ENCODE_ANY         0x200
+#define JSON_ESCAPE_SLASH       0x400
+#define JSON_REAL_PRECISION(n)  (((n) & 0x1F) << 11)
+#define JSON_EMBED              0x10000
+
+typedef int (*json_dump_callback_t)(const char *buffer, size_t size, void *data);
+
+char *json_dumps(const json_t *json, size_t flags);
+size_t json_dumpb(const json_t *json, char *buffer, size_t size, size_t flags);
+int json_dumpf(const json_t *json, FILE *output, size_t flags);
+int json_dumpfd(const json_t *json, int output, size_t flags);
+int json_dump_file(const json_t *json, const char *path, size_t flags);
+int json_dump_callback(const json_t *json, json_dump_callback_t callback, void *data, size_t flags);
+
+/* custom memory allocation */
+
+typedef void *(*json_malloc_t)(size_t);
+typedef void (*json_free_t)(void *);
+
+void json_set_alloc_funcs(json_malloc_t malloc_fn, json_free_t free_fn);
+void json_get_alloc_funcs(json_malloc_t *malloc_fn, json_free_t *free_fn);
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif

compat/jansson/jansson_config.h

@@ -0,0 +1,43 @@
+/*
+ * Copyright (c) 2010-2016 Petri Lehtinen <petri@digip.org>
+ *
+ * Jansson is free software; you can redistribute it and/or modify
+ * it under the terms of the MIT license. See LICENSE for details.
+ *
+ *
+ * This file specifies a part of the site-specific configuration for
+ * Jansson, namely those things that affect the public API in
+ * jansson.h.
+ *
+ * The configure script copies this file to jansson_config.h and
+ * replaces @var@ substitutions by values that fit your system. If you
+ * cannot run the configure script, you can do the value substitution
+ * by hand.
+ */
+
+#ifndef JANSSON_CONFIG_H
+#define JANSSON_CONFIG_H
+
+/* If your compiler supports the inline keyword in C, JSON_INLINE is
+   defined to `inline', otherwise empty. In C++, the inline is always
+   supported. */
+#ifdef __cplusplus
+#define JSON_INLINE inline
+#else
+#define JSON_INLINE inline
+#endif
+
+/* If your compiler supports the `long long` type and the strtoll()
+   library function, JSON_INTEGER_IS_LONG_LONG is defined to 1,
+   otherwise to 0. */
+#define JSON_INTEGER_IS_LONG_LONG 1
+
+/* If locale.h and localeconv() are available, define to 1,
+   otherwise to 0. */
+#define JSON_HAVE_LOCALECONV 1
+
+/* Maximum recursion depth for parsing JSON input.
+   This limits the depth of e.g. array-within-array constructions. */
+#define JSON_PARSER_MAX_DEPTH 2048
+
+#endif

compat/jansson/jansson_private.h

@@ -0,0 +1,109 @@
+/*
+ * Copyright (c) 2009-2016 Petri Lehtinen <petri@digip.org>
+ *
+ * Jansson is free software; you can redistribute it and/or modify
+ * it under the terms of the MIT license. See LICENSE for details.
+ */
+
+#ifndef JANSSON_PRIVATE_H
+#define JANSSON_PRIVATE_H
+
+#include "jansson_private_config.h"
+#include <stddef.h>
+#include "jansson.h"
+#include "hashtable.h"
+#include "strbuffer.h"
+
+#define container_of(ptr_, type_, member_)  \
+    ((type_ *)((char *)ptr_ - offsetof(type_, member_)))
+
+/* On some platforms, max() may already be defined */
+#ifndef max
+#define max(a, b)  ((a) > (b) ? (a) : (b))
+#endif
+
+/* va_copy is a C99 feature. In C89 implementations, it's sometimes
+   available as __va_copy. If not, memcpy() should do the trick. */
+#ifndef va_copy
+#ifdef __va_copy
+#define va_copy __va_copy
+#else
+#define va_copy(a, b)  memcpy(&(a), &(b), sizeof(va_list))
+#endif
+#endif
+
+typedef struct {
+    json_t json;
+    hashtable_t hashtable;
+    int visited;
+} json_object_t;
+
+typedef struct {
+    json_t json;
+    size_t size;
+    size_t entries;
+    json_t **table;
+    int visited;
+} json_array_t;
+
+typedef struct {
+    json_t json;
+    char *value;
+    size_t length;
+} json_string_t;
+
+typedef struct {
+    json_t json;
+    double value;
+} json_real_t;
+
+typedef struct {
+    json_t json;
+    json_int_t value;
+} json_integer_t;
+
+#define json_to_object(json_)  container_of(json_, json_object_t, json)
+#define json_to_array(json_)   container_of(json_, json_array_t, json)
+#define json_to_string(json_)  container_of(json_, json_string_t, json)
+#define json_to_real(json_)    container_of(json_, json_real_t, json)
+#define json_to_integer(json_) container_of(json_, json_integer_t, json)
+
+/* Create a string by taking ownership of an existing buffer */
+json_t *jsonp_stringn_nocheck_own(const char *value, size_t len);
+
+/* Error message formatting */
+void jsonp_error_init(json_error_t *error, const char *source);
+void jsonp_error_set_source(json_error_t *error, const char *source);
+void jsonp_error_set(json_error_t *error, int line, int column,
+                     size_t position, const char *msg, ...);
+void jsonp_error_vset(json_error_t *error, int line, int column,
+                      size_t position, const char *msg, va_list ap);
+
+/* Locale independent string<->double conversions */
+int jsonp_strtod(strbuffer_t *strbuffer, double *out);
+int jsonp_dtostr(char *buffer, size_t size, double value, int prec);
+
+/* Wrappers for custom memory functions */
+void* jsonp_malloc(size_t size);
+void jsonp_free(void *ptr);
+char *jsonp_strndup(const char *str, size_t length);
+char *jsonp_strdup(const char *str);
+char *jsonp_strndup(const char *str, size_t len);
+
+
+/* Windows compatibility */
+#if defined(_WIN32) || defined(WIN32)
+#  if defined(_MSC_VER)  /* MS compiller */
+#    if (_MSC_VER < 1900) && !defined(snprintf)  /* snprintf not defined yet & not introduced */
+#      define snprintf _snprintf
+#    endif
+#    if (_MSC_VER < 1500) && !defined(vsnprintf)  /* vsnprintf not defined yet & not introduced */
+#      define vsnprintf(b,c,f,a) _vsnprintf(b,c,f,a)
+#    endif
+#  else  /* Other Windows compiller, old definition */
+#    define snprintf _snprintf
+#    define vsnprintf _vsnprintf
+#  endif
+#endif
+
+#endif

compat/jansson/jansson_private_config.h

@@ -0,0 +1,159 @@
+/* jansson_private_config.h.  Generated from jansson_private_config.h.in by configure.  */
+/* jansson_private_config.h.in.  Generated from configure.ac by autoheader.  */
+
+/* Define to 1 if gcc's __atomic builtins are available */
+#define HAVE_ATOMIC_BUILTINS 1
+
+/* Define to 1 if you have the `close' function. */
+#define HAVE_CLOSE 1
+
+/* Define to 1 if you have the <dlfcn.h> header file. */
+/* #undef HAVE_DLFCN_H */
+
+/* Define to 1 if you have the <endian.h> header file. */
+/* #undef HAVE_ENDIAN_H */
+
+/* Define to 1 if you have the <fcntl.h> header file. */
+#define HAVE_FCNTL_H 1
+
+/* Define to 1 if you have the `getpid' function. */
+#define HAVE_GETPID 1
+
+/* Define to 1 if you have the `gettimeofday' function. */
+#define HAVE_GETTIMEOFDAY 1
+
+/* Define to 1 if you have the <inttypes.h> header file. */
+#define HAVE_INTTYPES_H 1
+
+/* Define to 1 if you have the `localeconv' function. */
+#define HAVE_LOCALECONV 1
+
+/* Define to 1 if you have the <locale.h> header file. */
+#define HAVE_LOCALE_H 1
+
+/* Define to 1 if the system has the type 'long long int'. */
+#define HAVE_LONG_LONG_INT 1
+
+/* Define to 1 if you have the <memory.h> header file. */
+#define HAVE_MEMORY_H 1
+
+/* Define to 1 if you have the `open' function. */
+#define HAVE_OPEN 1
+
+/* Define to 1 if you have the `read' function. */
+#define HAVE_READ 1
+
+/* Define to 1 if you have the <sched.h> header file. */
+#define HAVE_SCHED_H 1
+
+/* Define to 1 if you have the `sched_yield' function. */
+#define HAVE_SCHED_YIELD 1
+
+/* Define to 1 if you have the <stdint.h> header file. */
+#define HAVE_STDINT_H 1
+
+/* Define to 1 if you have the <stdlib.h> header file. */
+#define HAVE_STDLIB_H 1
+
+/* Define to 1 if you have the <strings.h> header file. */
+#define HAVE_STRINGS_H 1
+
+/* Define to 1 if you have the <string.h> header file. */
+#define HAVE_STRING_H 1
+
+/* Define to 1 if you have the `strtoll' function. */
+#define HAVE_STRTOLL 1
+
+/* Define to 1 if gcc's __sync builtins are available */
+#define HAVE_SYNC_BUILTINS 1
+
+/* Define to 1 if you have the <sys/param.h> header file. */
+#define HAVE_SYS_PARAM_H 1
+
+/* Define to 1 if you have the <sys/stat.h> header file. */
+#define HAVE_SYS_STAT_H 1
+
+/* Define to 1 if you have the <sys/time.h> header file. */
+#define HAVE_SYS_TIME_H 1
+
+/* Define to 1 if you have the <sys/types.h> header file. */
+#define HAVE_SYS_TYPES_H 1
+
+/* Define to 1 if you have the <unistd.h> header file. */
+#define HAVE_UNISTD_H 1
+
+/* Define to 1 if the system has the type 'unsigned long long int'. */
+#define HAVE_UNSIGNED_LONG_LONG_INT 1
+
+/* Number of buckets new object hashtables contain is 2 raised to this power.
+   E.g. 3 -> 2^3 = 8. */
+#define INITIAL_HASHTABLE_ORDER 3
+
+/* Define to the sub-directory where libtool stores uninstalled libraries. */
+#define LT_OBJDIR ".libs/"
+
+/* Name of package */
+#define PACKAGE "jansson"
+
+/* Define to the address where bug reports for this package should be sent. */
+#define PACKAGE_BUGREPORT "petri@digip.org"
+
+/* Define to the full name of this package. */
+#define PACKAGE_NAME "jansson"
+
+/* Define to the full name and version of this package. */
+#define PACKAGE_STRING "jansson 2.9"
+
+/* Define to the one symbol short name of this package. */
+#define PACKAGE_TARNAME "jansson"
+
+/* Define to the home page for this package. */
+#define PACKAGE_URL ""
+
+/* Define to the version of this package. */
+#define PACKAGE_VERSION "2.9"
+
+/* Define to 1 if you have the ANSI C header files. */
+#define STDC_HEADERS 1
+
+/* Define to 1 if /dev/urandom should be used for seeding the hash function */
+#define USE_URANDOM 1
+
+/* Define to 1 if CryptGenRandom should be used for seeding the hash function
+   */
+#define USE_WINDOWS_CRYPTOAPI 1
+
+/* Version number of package */
+#define VERSION "2.9"
+
+/* Define for Solaris 2.5.1 so the uint32_t typedef from <sys/synch.h>,
+   <pthread.h>, or <semaphore.h> is not used. If the typedef were allowed, the
+   #define below would cause a syntax error. */
+/* #undef _UINT32_T */
+
+/* Define for Solaris 2.5.1 so the uint8_t typedef from <sys/synch.h>,
+   <pthread.h>, or <semaphore.h> is not used. If the typedef were allowed, the
+   #define below would cause a syntax error. */
+/* #undef _UINT8_T */
+
+/* Define to `__inline__' or `__inline' if that's what the C compiler
+   calls it, or to nothing if 'inline' is not supported under any name.  */
+#ifndef __cplusplus
+/* #undef inline */
+#endif
+
+/* Define to the type of a signed integer type of width exactly 32 bits if
+   such a type exists and the standard includes do not define it. */
+/* #undef int32_t */
+
+/* Define to the type of an unsigned integer type of width exactly 16 bits if
+   such a type exists and the standard includes do not define it. */
+/* #undef uint16_t */
+
+/* Define to the type of an unsigned integer type of width exactly 32 bits if
+   such a type exists and the standard includes do not define it. */
+/* #undef uint32_t */
+
+/* Define to the type of an unsigned integer type of width exactly 8 bits if
+   such a type exists and the standard includes do not define it. */
+/* #undef uint8_t */

compat/jansson/lookup3.h

@@ -0,0 +1,381 @@
+/*
+-------------------------------------------------------------------------------
+lookup3.c, by Bob Jenkins, May 2006, Public Domain.
+
+These are functions for producing 32-bit hashes for hash table lookup.
+hashword(), hashlittle(), hashlittle2(), hashbig(), mix(), and final() 
+are externally useful functions.  Routines to test the hash are included 
+if SELF_TEST is defined.  You can use this free for any purpose.  It's in
+the public domain.  It has no warranty.
+
+You probably want to use hashlittle().  hashlittle() and hashbig()
+hash byte arrays.  hashlittle() is is faster than hashbig() on
+little-endian machines.  Intel and AMD are little-endian machines.
+On second thought, you probably want hashlittle2(), which is identical to
+hashlittle() except it returns two 32-bit hashes for the price of one.  
+You could implement hashbig2() if you wanted but I haven't bothered here.
+
+If you want to find a hash of, say, exactly 7 integers, do
+  a = i1;  b = i2;  c = i3;
+  mix(a,b,c);
+  a += i4; b += i5; c += i6;
+  mix(a,b,c);
+  a += i7;
+  final(a,b,c);
+then use c as the hash value.  If you have a variable length array of
+4-byte integers to hash, use hashword().  If you have a byte array (like
+a character string), use hashlittle().  If you have several byte arrays, or
+a mix of things, see the comments above hashlittle().  
+
+Why is this so big?  I read 12 bytes at a time into 3 4-byte integers, 
+then mix those integers.  This is fast (you can do a lot more thorough
+mixing with 12*3 instructions on 3 integers than you can with 3 instructions
+on 1 byte), but shoehorning those bytes into integers efficiently is messy.
+-------------------------------------------------------------------------------
+*/
+
+#include <stdlib.h>
+
+#ifdef HAVE_CONFIG_H
+#include <jansson_private_config.h>
+#endif
+
+#ifdef HAVE_STDINT_H
+#include <stdint.h>     /* defines uint32_t etc */
+#endif
+
+#ifdef HAVE_SYS_PARAM_H
+#include <sys/param.h>  /* attempt to define endianness */
+#endif
+
+#ifdef HAVE_ENDIAN_H
+# include <endian.h>    /* attempt to define endianness */
+#endif
+
+/*
+ * My best guess at if you are big-endian or little-endian.  This may
+ * need adjustment.
+ */
+#if (defined(__BYTE_ORDER) && defined(__LITTLE_ENDIAN) && \
+     __BYTE_ORDER == __LITTLE_ENDIAN) || \
+    (defined(i386) || defined(__i386__) || defined(__i486__) || \
+     defined(__i586__) || defined(__i686__) || defined(vax) || defined(MIPSEL))
+# define HASH_LITTLE_ENDIAN 1
+# define HASH_BIG_ENDIAN 0
+#elif (defined(__BYTE_ORDER) && defined(__BIG_ENDIAN) && \
+       __BYTE_ORDER == __BIG_ENDIAN) || \
+      (defined(sparc) || defined(POWERPC) || defined(mc68000) || defined(sel))
+# define HASH_LITTLE_ENDIAN 0
+# define HASH_BIG_ENDIAN 1
+#else
+# define HASH_LITTLE_ENDIAN 0
+# define HASH_BIG_ENDIAN 0
+#endif
+
+#define hashsize(n) ((uint32_t)1<<(n))
+#define hashmask(n) (hashsize(n)-1)
+#define rot(x,k) (((x)<<(k)) | ((x)>>(32-(k))))
+
+/*
+-------------------------------------------------------------------------------
+mix -- mix 3 32-bit values reversibly.
+
+This is reversible, so any information in (a,b,c) before mix() is
+still in (a,b,c) after mix().
+
+If four pairs of (a,b,c) inputs are run through mix(), or through
+mix() in reverse, there are at least 32 bits of the output that
+are sometimes the same for one pair and different for another pair.
+This was tested for:
+* pairs that differed by one bit, by two bits, in any combination
+  of top bits of (a,b,c), or in any combination of bottom bits of
+  (a,b,c).
+* "differ" is defined as +, -, ^, or ~^.  For + and -, I transformed
+  the output delta to a Gray code (a^(a>>1)) so a string of 1's (as
+  is commonly produced by subtraction) look like a single 1-bit
+  difference.
+* the base values were pseudorandom, all zero but one bit set, or 
+  all zero plus a counter that starts at zero.
+
+Some k values for my "a-=c; a^=rot(c,k); c+=b;" arrangement that
+satisfy this are
+    4  6  8 16 19  4
+    9 15  3 18 27 15
+   14  9  3  7 17  3
+Well, "9 15 3 18 27 15" didn't quite get 32 bits diffing
+for "differ" defined as + with a one-bit base and a two-bit delta.  I
+used http://burtleburtle.net/bob/hash/avalanche.html to choose 
+the operations, constants, and arrangements of the variables.
+
+This does not achieve avalanche.  There are input bits of (a,b,c)
+that fail to affect some output bits of (a,b,c), especially of a.  The
+most thoroughly mixed value is c, but it doesn't really even achieve
+avalanche in c.
+
+This allows some parallelism.  Read-after-writes are good at doubling
+the number of bits affected, so the goal of mixing pulls in the opposite
+direction as the goal of parallelism.  I did what I could.  Rotates
+seem to cost as much as shifts on every machine I could lay my hands
+on, and rotates are much kinder to the top and bottom bits, so I used
+rotates.
+-------------------------------------------------------------------------------
+*/
+#define mix(a,b,c) \
+{ \
+  a -= c;  a ^= rot(c, 4);  c += b; \
+  b -= a;  b ^= rot(a, 6);  a += c; \
+  c -= b;  c ^= rot(b, 8);  b += a; \
+  a -= c;  a ^= rot(c,16);  c += b; \
+  b -= a;  b ^= rot(a,19);  a += c; \
+  c -= b;  c ^= rot(b, 4);  b += a; \
+}
+
+/*
+-------------------------------------------------------------------------------
+final -- final mixing of 3 32-bit values (a,b,c) into c
+
+Pairs of (a,b,c) values differing in only a few bits will usually
+produce values of c that look totally different.  This was tested for
+* pairs that differed by one bit, by two bits, in any combination
+  of top bits of (a,b,c), or in any combination of bottom bits of
+  (a,b,c).
+* "differ" is defined as +, -, ^, or ~^.  For + and -, I transformed
+  the output delta to a Gray code (a^(a>>1)) so a string of 1's (as
+  is commonly produced by subtraction) look like a single 1-bit
+  difference.
+* the base values were pseudorandom, all zero but one bit set, or 
+  all zero plus a counter that starts at zero.
+
+These constants passed:
+ 14 11 25 16 4 14 24
+ 12 14 25 16 4 14 24
+and these came close:
+  4  8 15 26 3 22 24
+ 10  8 15 26 3 22 24
+ 11  8 15 26 3 22 24
+-------------------------------------------------------------------------------
+*/
+#define final(a,b,c) \
+{ \
+  c ^= b; c -= rot(b,14); \
+  a ^= c; a -= rot(c,11); \
+  b ^= a; b -= rot(a,25); \
+  c ^= b; c -= rot(b,16); \
+  a ^= c; a -= rot(c,4);  \
+  b ^= a; b -= rot(a,14); \
+  c ^= b; c -= rot(b,24); \
+}
+
+/*
+-------------------------------------------------------------------------------
+hashlittle() -- hash a variable-length key into a 32-bit value
+  k       : the key (the unaligned variable-length array of bytes)
+  length  : the length of the key, counting by bytes
+  initval : can be any 4-byte value
+Returns a 32-bit value.  Every bit of the key affects every bit of
+the return value.  Two keys differing by one or two bits will have
+totally different hash values.
+
+The best hash table sizes are powers of 2.  There is no need to do
+mod a prime (mod is sooo slow!).  If you need less than 32 bits,
+use a bitmask.  For example, if you need only 10 bits, do
+  h = (h & hashmask(10));
+In which case, the hash table should have hashsize(10) elements.
+
+If you are hashing n strings (uint8_t **)k, do it like this:
+  for (i=0, h=0; i<n; ++i) h = hashlittle( k[i], len[i], h);
+
+By Bob Jenkins, 2006.  bob_jenkins@burtleburtle.net.  You may use this
+code any way you wish, private, educational, or commercial.  It's free.
+
+Use for hash table lookup, or anything where one collision in 2^^32 is
+acceptable.  Do NOT use for cryptographic purposes.
+-------------------------------------------------------------------------------
+*/
+
+static uint32_t hashlittle(const void *key, size_t length, uint32_t initval)
+{
+  uint32_t a,b,c;                                          /* internal state */
+  union { const void *ptr; size_t i; } u;     /* needed for Mac Powerbook G4 */
+
+  /* Set up the internal state */
+  a = b = c = 0xdeadbeef + ((uint32_t)length) + initval;
+
+  u.ptr = key;
+  if (HASH_LITTLE_ENDIAN && ((u.i & 0x3) == 0)) {
+    const uint32_t *k = (const uint32_t *)key;         /* read 32-bit chunks */
+
+/* Detect Valgrind or AddressSanitizer */
+#ifdef VALGRIND
+# define NO_MASKING_TRICK 1
+#else
+# if defined(__has_feature)  /* Clang */
+#  if __has_feature(address_sanitizer)  /* is ASAN enabled? */
+#   define NO_MASKING_TRICK 1
+#  endif
+# else
+#  if defined(__SANITIZE_ADDRESS__)  /* GCC 4.8.x, is ASAN enabled? */
+#   define NO_MASKING_TRICK 1
+#  endif
+# endif
+#endif
+
+#ifdef NO_MASKING_TRICK
+    const uint8_t  *k8;
+#endif
+
+    /*------ all but last block: aligned reads and affect 32 bits of (a,b,c) */
+    while (length > 12)
+    {
+      a += k[0];
+      b += k[1];
+      c += k[2];
+      mix(a,b,c);
+      length -= 12;
+      k += 3;
+    }
+
+    /*----------------------------- handle the last (probably partial) block */
+    /* 
+     * "k[2]&0xffffff" actually reads beyond the end of the string, but
+     * then masks off the part it's not allowed to read.  Because the
+     * string is aligned, the masked-off tail is in the same word as the
+     * rest of the string.  Every machine with memory protection I've seen
+     * does it on word boundaries, so is OK with this.  But VALGRIND will
+     * still catch it and complain.  The masking trick does make the hash
+     * noticably faster for short strings (like English words).
+     */
+#ifndef NO_MASKING_TRICK
+
+    switch(length)
+    {
+    case 12: c+=k[2]; b+=k[1]; a+=k[0]; break;
+    case 11: c+=k[2]&0xffffff; b+=k[1]; a+=k[0]; break;
+    case 10: c+=k[2]&0xffff; b+=k[1]; a+=k[0]; break;
+    case 9 : c+=k[2]&0xff; b+=k[1]; a+=k[0]; break;
+    case 8 : b+=k[1]; a+=k[0]; break;
+    case 7 : b+=k[1]&0xffffff; a+=k[0]; break;
+    case 6 : b+=k[1]&0xffff; a+=k[0]; break;
+    case 5 : b+=k[1]&0xff; a+=k[0]; break;
+    case 4 : a+=k[0]; break;
+    case 3 : a+=k[0]&0xffffff; break;
+    case 2 : a+=k[0]&0xffff; break;
+    case 1 : a+=k[0]&0xff; break;
+    case 0 : return c;              /* zero length strings require no mixing */
+    }
+
+#else /* make valgrind happy */
+
+    k8 = (const uint8_t *)k;
+    switch(length)
+    {
+    case 12: c+=k[2]; b+=k[1]; a+=k[0]; break;
+    case 11: c+=((uint32_t)k8[10])<<16;  /* fall through */
+    case 10: c+=((uint32_t)k8[9])<<8;    /* fall through */
+    case 9 : c+=k8[8];                   /* fall through */
+    case 8 : b+=k[1]; a+=k[0]; break;
+    case 7 : b+=((uint32_t)k8[6])<<16;   /* fall through */
+    case 6 : b+=((uint32_t)k8[5])<<8;    /* fall through */
+    case 5 : b+=k8[4];                   /* fall through */
+    case 4 : a+=k[0]; break;
+    case 3 : a+=((uint32_t)k8[2])<<16;   /* fall through */
+    case 2 : a+=((uint32_t)k8[1])<<8;    /* fall through */
+    case 1 : a+=k8[0]; break;
+    case 0 : return c;
+    }
+
+#endif /* !valgrind */
+
+  } else if (HASH_LITTLE_ENDIAN && ((u.i & 0x1) == 0)) {
+    const uint16_t *k = (const uint16_t *)key;         /* read 16-bit chunks */
+    const uint8_t  *k8;
+
+    /*--------------- all but last block: aligned reads and different mixing */
+    while (length > 12)
+    {
+      a += k[0] + (((uint32_t)k[1])<<16);
+      b += k[2] + (((uint32_t)k[3])<<16);
+      c += k[4] + (((uint32_t)k[5])<<16);
+      mix(a,b,c);
+      length -= 12;
+      k += 6;
+    }
+
+    /*----------------------------- handle the last (probably partial) block */
+    k8 = (const uint8_t *)k;
+    switch(length)
+    {
+    case 12: c+=k[4]+(((uint32_t)k[5])<<16);
+             b+=k[2]+(((uint32_t)k[3])<<16);
+             a+=k[0]+(((uint32_t)k[1])<<16);
+             break;
+    case 11: c+=((uint32_t)k8[10])<<16;     /* fall through */
+    case 10: c+=k[4];
+             b+=k[2]+(((uint32_t)k[3])<<16);
+             a+=k[0]+(((uint32_t)k[1])<<16);
+             break;
+    case 9 : c+=k8[8];                      /* fall through */
+    case 8 : b+=k[2]+(((uint32_t)k[3])<<16);
+             a+=k[0]+(((uint32_t)k[1])<<16);
+             break;
+    case 7 : b+=((uint32_t)k8[6])<<16;      /* fall through */
+    case 6 : b+=k[2];
+             a+=k[0]+(((uint32_t)k[1])<<16);
+             break;
+    case 5 : b+=k8[4];                      /* fall through */
+    case 4 : a+=k[0]+(((uint32_t)k[1])<<16);
+             break;
+    case 3 : a+=((uint32_t)k8[2])<<16;      /* fall through */
+    case 2 : a+=k[0];
+             break;
+    case 1 : a+=k8[0];
+             break;
+    case 0 : return c;                     /* zero length requires no mixing */
+    }
+
+  } else {                        /* need to read the key one byte at a time */
+    const uint8_t *k = (const uint8_t *)key;
+
+    /*--------------- all but the last block: affect some 32 bits of (a,b,c) */
+    while (length > 12)
+    {
+      a += k[0];
+      a += ((uint32_t)k[1])<<8;
+      a += ((uint32_t)k[2])<<16;
+      a += ((uint32_t)k[3])<<24;
+      b += k[4];
+      b += ((uint32_t)k[5])<<8;
+      b += ((uint32_t)k[6])<<16;
+      b += ((uint32_t)k[7])<<24;
+      c += k[8];
+      c += ((uint32_t)k[9])<<8;
+      c += ((uint32_t)k[10])<<16;
+      c += ((uint32_t)k[11])<<24;
+      mix(a,b,c);
+      length -= 12;
+      k += 12;
+    }
+
+    /*-------------------------------- last block: affect all 32 bits of (c) */
+    switch(length)                   /* all the case statements fall through */
+    {
+    case 12: c+=((uint32_t)k[11])<<24;
+    case 11: c+=((uint32_t)k[10])<<16;
+    case 10: c+=((uint32_t)k[9])<<8;
+    case 9 : c+=k[8];
+    case 8 : b+=((uint32_t)k[7])<<24;
+    case 7 : b+=((uint32_t)k[6])<<16;
+    case 6 : b+=((uint32_t)k[5])<<8;
+    case 5 : b+=k[4];
+    case 4 : a+=((uint32_t)k[3])<<24;
+    case 3 : a+=((uint32_t)k[2])<<16;
+    case 2 : a+=((uint32_t)k[1])<<8;
+    case 1 : a+=k[0];
+             break;
+    case 0 : return c;
+    }
+  }
+
+  final(a,b,c);
+  return c;
+}

compat/jansson/memory.c

@@ -0,0 +1,69 @@
+/*
+ * Copyright (c) 2009-2016 Petri Lehtinen <petri@digip.org>
+ * Copyright (c) 2011-2012 Basile Starynkevitch <basile@starynkevitch.net>
+ *
+ * Jansson is free software; you can redistribute it and/or modify it
+ * under the terms of the MIT license. See LICENSE for details.
+ */
+
+#include <stdlib.h>
+#include <string.h>
+
+#include "jansson.h"
+#include "jansson_private.h"
+
+/* C89 allows these to be macros */
+#undef malloc
+#undef free
+
+/* memory function pointers */
+static json_malloc_t do_malloc = malloc;
+static json_free_t do_free = free;
+
+void *jsonp_malloc(size_t size)
+{
+    if(!size)
+        return NULL;
+
+    return (*do_malloc)(size);
+}
+
+void jsonp_free(void *ptr)
+{
+    if(!ptr)
+        return;
+
+    (*do_free)(ptr);
+}
+
+char *jsonp_strdup(const char *str)
+{
+    return jsonp_strndup(str, strlen(str));
+}
+
+char *jsonp_strndup(const char *str, size_t len)
+{
+    char *new_str;
+
+    new_str = jsonp_malloc(len + 1);
+    if(!new_str)
+        return NULL;
+
+    memcpy(new_str, str, len);
+    new_str[len] = '\0';
+    return new_str;
+}
+
+void json_set_alloc_funcs(json_malloc_t malloc_fn, json_free_t free_fn)
+{
+    do_malloc = malloc_fn;
+    do_free = free_fn;
+}
+
+void json_get_alloc_funcs(json_malloc_t *malloc_fn, json_free_t *free_fn)
+{
+    if (malloc_fn)
+        *malloc_fn = do_malloc;
+    if (free_fn)
+        *free_fn = do_free;
+}

compat/jansson/pack_unpack.c

@@ -0,0 +1,871 @@
+/*
+ * Copyright (c) 2009-2016 Petri Lehtinen <petri@digip.org>
+ * Copyright (c) 2011-2012 Graeme Smecher <graeme.smecher@mail.mcgill.ca>
+ *
+ * Jansson is free software; you can redistribute it and/or modify
+ * it under the terms of the MIT license. See LICENSE for details.
+ */
+
+#include <string.h>
+#include "jansson.h"
+#include "jansson_private.h"
+#include "utf.h"
+
+typedef struct {
+    int line;
+    int column;
+    size_t pos;
+    char token;
+} token_t;
+
+typedef struct {
+    const char *start;
+    const char *fmt;
+    token_t prev_token;
+    token_t token;
+    token_t next_token;
+    json_error_t *error;
+    size_t flags;
+    int line;
+    int column;
+    size_t pos;
+} scanner_t;
+
+#define token(scanner) ((scanner)->token.token)
+
+static const char * const type_names[] = {
+    "object",
+    "array",
+    "string",
+    "integer",
+    "real",
+    "true",
+    "false",
+    "null"
+};
+
+#define type_name(x) type_names[json_typeof(x)]
+
+static const char unpack_value_starters[] = "{[siIbfFOon";
+
+static void scanner_init(scanner_t *s, json_error_t *error,
+                         size_t flags, const char *fmt)
+{
+    s->error = error;
+    s->flags = flags;
+    s->fmt = s->start = fmt;
+    memset(&s->prev_token, 0, sizeof(token_t));
+    memset(&s->token, 0, sizeof(token_t));
+    memset(&s->next_token, 0, sizeof(token_t));
+    s->line = 1;
+    s->column = 0;
+    s->pos = 0;
+}
+
+static void next_token(scanner_t *s)
+{
+    const char *t;
+    s->prev_token = s->token;
+
+    if(s->next_token.line) {
+        s->token = s->next_token;
+        s->next_token.line = 0;
+        return;
+    }
+
+    t = s->fmt;
+    s->column++;
+    s->pos++;
+
+    /* skip space and ignored chars */
+    while(*t == ' ' || *t == '\t' || *t == '\n' || *t == ',' || *t == ':') {
+        if(*t == '\n') {
+            s->line++;
+            s->column = 1;
+        }
+        else
+            s->column++;
+
+        s->pos++;
+        t++;
+    }
+
+    s->token.token = *t;
+    s->token.line = s->line;
+    s->token.column = s->column;
+    s->token.pos = s->pos;
+
+    t++;
+    s->fmt = t;
+}
+
+static void prev_token(scanner_t *s)
+{
+    s->next_token = s->token;
+    s->token = s->prev_token;
+}
+
+static void set_error(scanner_t *s, const char *source, const char *fmt, ...)
+{
+    va_list ap;
+    va_start(ap, fmt);
+
+    jsonp_error_vset(s->error, s->token.line, s->token.column, s->token.pos,
+                     fmt, ap);
+
+    jsonp_error_set_source(s->error, source);
+
+    va_end(ap);
+}
+
+static json_t *pack(scanner_t *s, va_list *ap);
+
+
+/* ours will be set to 1 if jsonp_free() must be called for the result
+   afterwards */
+static char *read_string(scanner_t *s, va_list *ap,
+                         const char *purpose, size_t *out_len, int *ours)
+{
+    char t;
+    strbuffer_t strbuff;
+    const char *str;
+    size_t length;
+
+    next_token(s);
+    t = token(s);
+    prev_token(s);
+
+    if(t != '#' && t != '%' && t != '+') {
+        /* Optimize the simple case */
+        str = va_arg(*ap, const char *);
+
+        if(!str) {
+            set_error(s, "<args>", "NULL string argument");
+            return NULL;
+        }
+
+        length = strlen(str);
+
+        if(!utf8_check_string(str, length)) {
+            set_error(s, "<args>", "Invalid UTF-8 %s", purpose);
+            return NULL;
+        }
+
+        *out_len = length;
+        *ours = 0;
+        return (char *)str;
+    }
+
+    strbuffer_init(&strbuff);
+
+    while(1) {
+        str = va_arg(*ap, const char *);
+        if(!str) {
+            set_error(s, "<args>", "NULL string argument");
+            strbuffer_close(&strbuff);
+            return NULL;
+        }
+
+        next_token(s);
+
+        if(token(s) == '#') {
+            length = va_arg(*ap, int);
+        }
+        else if(token(s) == '%') {
+            length = va_arg(*ap, size_t);
+        }
+        else {
+            prev_token(s);
+            length = strlen(str);
+        }
+
+        if(strbuffer_append_bytes(&strbuff, str, length) == -1) {
+            set_error(s, "<internal>", "Out of memory");
+            strbuffer_close(&strbuff);
+            return NULL;
+        }
+
+        next_token(s);
+        if(token(s) != '+') {
+            prev_token(s);
+            break;
+        }
+    }
+
+    if(!utf8_check_string(strbuff.value, strbuff.length)) {
+        set_error(s, "<args>", "Invalid UTF-8 %s", purpose);
+        strbuffer_close(&strbuff);
+        return NULL;
+    }
+
+    *out_len = strbuff.length;
+    *ours = 1;
+    return strbuffer_steal_value(&strbuff);
+}
+
+static json_t *pack_object(scanner_t *s, va_list *ap)
+{
+    json_t *object = json_object();
+    next_token(s);
+
+    while(token(s) != '}') {
+        char *key;
+        size_t len;
+        int ours;
+        json_t *value;
+
+        if(!token(s)) {
+            set_error(s, "<format>", "Unexpected end of format string");
+            goto error;
+        }
+
+        if(token(s) != 's') {
+            set_error(s, "<format>", "Expected format 's', got '%c'", token(s));
+            goto error;
+        }
+
+        key = read_string(s, ap, "object key", &len, &ours);
+        if(!key)
+            goto error;
+
+        next_token(s);
+
+        value = pack(s, ap);
+        if(!value) {
+            if(ours)
+                jsonp_free(key);
+
+            goto error;
+        }
+
+        if(json_object_set_new_nocheck(object, key, value)) {
+            set_error(s, "<internal>", "Unable to add key \"%s\"", key);
+            if(ours)
+                jsonp_free(key);
+
+            goto error;
+        }
+
+        if(ours)
+            jsonp_free(key);
+
+        next_token(s);
+    }
+
+    return object;
+
+error:
+    json_decref(object);
+    return NULL;
+}
+
+static json_t *pack_array(scanner_t *s, va_list *ap)
+{
+    json_t *array = json_array();
+    next_token(s);
+
+    while(token(s) != ']') {
+        json_t *value;
+
+        if(!token(s)) {
+            set_error(s, "<format>", "Unexpected end of format string");
+            goto error;
+        }
+
+        value = pack(s, ap);
+        if(!value)
+            goto error;
+
+        if(json_array_append_new(array, value)) {
+            set_error(s, "<internal>", "Unable to append to array");
+            goto error;
+        }
+
+        next_token(s);
+    }
+    return array;
+
+error:
+    json_decref(array);
+    return NULL;
+}
+
+static json_t *pack_string(scanner_t *s, va_list *ap)
+{
+    char *str;
+    size_t len;
+    int ours;
+    int nullable;
+
+    next_token(s);
+    nullable = token(s) == '?';
+    if (!nullable)
+        prev_token(s);
+
+    str = read_string(s, ap, "string", &len, &ours);
+    if (!str) {
+        return nullable ? json_null() : NULL;
+    } else if (ours) {
+        return jsonp_stringn_nocheck_own(str, len);
+    } else {
+        return json_stringn_nocheck(str, len);
+    }
+}
+
+static json_t *pack(scanner_t *s, va_list *ap)
+{
+    switch(token(s)) {
+        case '{':
+            return pack_object(s, ap);
+
+        case '[':
+            return pack_array(s, ap);
+
+        case 's': /* string */
+            return pack_string(s, ap);
+
+        case 'n': /* null */
+            return json_null();
+
+        case 'b': /* boolean */
+            return va_arg(*ap, int) ? json_true() : json_false();
+
+        case 'i': /* integer from int */
+            return json_integer(va_arg(*ap, int));
+
+        case 'I': /* integer from json_int_t */
+            return json_integer(va_arg(*ap, json_int_t));
+
+        case 'f': /* real */
+            return json_real(va_arg(*ap, double));
+
+        case 'O': /* a json_t object; increments refcount */
+        {
+            int nullable;
+            json_t *json;
+
+            next_token(s);
+            nullable = token(s) == '?';
+            if (!nullable)
+                prev_token(s);
+
+            json = va_arg(*ap, json_t *);
+            if (!json && nullable) {
+                return json_null();
+            } else {
+                return json_incref(json);
+            }
+        }
+
+        case 'o': /* a json_t object; doesn't increment refcount */
+        {
+            int nullable;
+            json_t *json;
+
+            next_token(s);
+            nullable = token(s) == '?';
+            if (!nullable)
+                prev_token(s);
+
+            json = va_arg(*ap, json_t *);
+            if (!json && nullable) {
+                return json_null();
+            } else {
+                return json;
+            }
+        }
+
+        default:
+            set_error(s, "<format>", "Unexpected format character '%c'",
+                      token(s));
+            return NULL;
+    }
+}
+
+static int unpack(scanner_t *s, json_t *root, va_list *ap);
+
+static int unpack_object(scanner_t *s, json_t *root, va_list *ap)
+{
+    int ret = -1;
+    int strict = 0;
+    int gotopt = 0;
+
+    /* Use a set (emulated by a hashtable) to check that all object
+       keys are accessed. Checking that the correct number of keys
+       were accessed is not enough, as the same key can be unpacked
+       multiple times.
+    */
+    hashtable_t key_set;
+
+    if(hashtable_init(&key_set)) {
+        set_error(s, "<internal>", "Out of memory");
+        return -1;
+    }
+
+    if(root && !json_is_object(root)) {
+        set_error(s, "<validation>", "Expected object, got %s",
+                  type_name(root));
+        goto out;
+    }
+    next_token(s);
+
+    while(token(s) != '}') {
+        const char *key;
+        json_t *value;
+        int opt = 0;
+
+        if(strict != 0) {
+            set_error(s, "<format>", "Expected '}' after '%c', got '%c'",
+                      (strict == 1 ? '!' : '*'), token(s));
+            goto out;
+        }
+
+        if(!token(s)) {
+            set_error(s, "<format>", "Unexpected end of format string");
+            goto out;
+        }
+
+        if(token(s) == '!' || token(s) == '*') {
+            strict = (token(s) == '!' ? 1 : -1);
+            next_token(s);
+            continue;
+        }
+
+        if(token(s) != 's') {
+            set_error(s, "<format>", "Expected format 's', got '%c'", token(s));
+            goto out;
+        }
+
+        key = va_arg(*ap, const char *);
+        if(!key) {
+            set_error(s, "<args>", "NULL object key");
+            goto out;
+        }
+
+        next_token(s);
+
+        if(token(s) == '?') {
+            opt = gotopt = 1;
+            next_token(s);
+        }
+
+        if(!root) {
+            /* skipping */
+            value = NULL;
+        }
+        else {
+            value = json_object_get(root, key);
+            if(!value && !opt) {
+                set_error(s, "<validation>", "Object item not found: %s", key);
+                goto out;
+            }
+        }
+
+        if(unpack(s, value, ap))
+            goto out;
+
+        hashtable_set(&key_set, key, json_null());
+        next_token(s);
+    }
+
+    if(strict == 0 && (s->flags & JSON_STRICT))
+        strict = 1;
+
+    if(root && strict == 1) {
+        /* We need to check that all non optional items have been parsed */
+        const char *key;
+        int have_unrecognized_keys = 0;
+        strbuffer_t unrecognized_keys;
+        json_t *value;
+        long unpacked = 0;
+        if (gotopt) {
+            /* We have optional keys, we need to iter on each key */
+            json_object_foreach(root, key, value) {
+                if(!hashtable_get(&key_set, key)) {
+                    unpacked++;
+
+                    /* Save unrecognized keys for the error message */
+                    if (!have_unrecognized_keys) {
+                        strbuffer_init(&unrecognized_keys);
+                        have_unrecognized_keys = 1;
+                    } else {
+                        strbuffer_append_bytes(&unrecognized_keys, ", ", 2);
+                    }
+                    strbuffer_append_bytes(&unrecognized_keys, key, strlen(key));
+                }
+            }
+        } else {
+            /* No optional keys, we can just compare the number of items */
+            unpacked = (long)json_object_size(root) - (long)key_set.size;
+        }
+        if (unpacked) {
+            if (!gotopt) {
+                /* Save unrecognized keys for the error message */
+                json_object_foreach(root, key, value) {
+                    if(!hashtable_get(&key_set, key)) {
+                        if (!have_unrecognized_keys) {
+                            strbuffer_init(&unrecognized_keys);
+                            have_unrecognized_keys = 1;
+                        } else {
+                            strbuffer_append_bytes(&unrecognized_keys, ", ", 2);
+                        }
+                        strbuffer_append_bytes(&unrecognized_keys, key, strlen(key));
+                    }
+                }
+            }
+            set_error(s, "<validation>",
+                      "%li object item(s) left unpacked: %s",
+                      unpacked, strbuffer_value(&unrecognized_keys));
+            strbuffer_close(&unrecognized_keys);
+            goto out;
+        }
+    }
+
+    ret = 0;
+
+out:
+    hashtable_close(&key_set);
+    return ret;
+}
+
+static int unpack_array(scanner_t *s, json_t *root, va_list *ap)
+{
+    size_t i = 0;
+    int strict = 0;
+
+    if(root && !json_is_array(root)) {
+        set_error(s, "<validation>", "Expected array, got %s", type_name(root));
+        return -1;
+    }
+    next_token(s);
+
+    while(token(s) != ']') {
+        json_t *value;
+
+        if(strict != 0) {
+            set_error(s, "<format>", "Expected ']' after '%c', got '%c'",
+                      (strict == 1 ? '!' : '*'),
+                      token(s));
+            return -1;
+        }
+
+        if(!token(s)) {
+            set_error(s, "<format>", "Unexpected end of format string");
+            return -1;
+        }
+
+        if(token(s) == '!' || token(s) == '*') {
+            strict = (token(s) == '!' ? 1 : -1);
+            next_token(s);
+            continue;
+        }
+
+        if(!strchr(unpack_value_starters, token(s))) {
+            set_error(s, "<format>", "Unexpected format character '%c'",
+                      token(s));
+            return -1;
+        }
+
+        if(!root) {
+            /* skipping */
+            value = NULL;
+        }
+        else {
+            value = json_array_get(root, i);
+            if(!value) {
+                set_error(s, "<validation>", "Array index %lu out of range",
+                          (unsigned long)i);
+                return -1;
+            }
+        }
+
+        if(unpack(s, value, ap))
+            return -1;
+
+        next_token(s);
+        i++;
+    }
+
+    if(strict == 0 && (s->flags & JSON_STRICT))
+        strict = 1;
+
+    if(root && strict == 1 && i != json_array_size(root)) {
+        long diff = (long)json_array_size(root) - (long)i;
+        set_error(s, "<validation>", "%li array item(s) left unpacked", diff);
+        return -1;
+    }
+
+    return 0;
+}
+
+static int unpack(scanner_t *s, json_t *root, va_list *ap)
+{
+    switch(token(s))
+    {
+        case '{':
+            return unpack_object(s, root, ap);
+
+        case '[':
+            return unpack_array(s, root, ap);
+
+        case 's':
+            if(root && !json_is_string(root)) {
+                set_error(s, "<validation>", "Expected string, got %s",
+                          type_name(root));
+                return -1;
+            }
+
+            if(!(s->flags & JSON_VALIDATE_ONLY)) {
+                const char **str_target;
+                size_t *len_target = NULL;
+
+                str_target = va_arg(*ap, const char **);
+                if(!str_target) {
+                    set_error(s, "<args>", "NULL string argument");
+                    return -1;
+                }
+
+                next_token(s);
+
+                if(token(s) == '%') {
+                    len_target = va_arg(*ap, size_t *);
+                    if(!len_target) {
+                        set_error(s, "<args>", "NULL string length argument");
+                        return -1;
+                    }
+                }
+                else
+                    prev_token(s);
+
+                if(root) {
+                    *str_target = json_string_value(root);
+                    if(len_target)
+                        *len_target = json_string_length(root);
+                }
+            }
+            return 0;
+
+        case 'i':
+            if(root && !json_is_integer(root)) {
+                set_error(s, "<validation>", "Expected integer, got %s",
+                          type_name(root));
+                return -1;
+            }
+
+            if(!(s->flags & JSON_VALIDATE_ONLY)) {
+                int *target = va_arg(*ap, int*);
+                if(root)
+                    *target = (int)json_integer_value(root);
+            }
+
+            return 0;
+
+        case 'I':
+            if(root && !json_is_integer(root)) {
+                set_error(s, "<validation>", "Expected integer, got %s",
+                          type_name(root));
+                return -1;
+            }
+
+            if(!(s->flags & JSON_VALIDATE_ONLY)) {
+                json_int_t *target = va_arg(*ap, json_int_t*);
+                if(root)
+                    *target = json_integer_value(root);
+            }
+
+            return 0;
+
+        case 'b':
+            if(root && !json_is_boolean(root)) {
+                set_error(s, "<validation>", "Expected true or false, got %s",
+                          type_name(root));
+                return -1;
+            }
+
+            if(!(s->flags & JSON_VALIDATE_ONLY)) {
+                int *target = va_arg(*ap, int*);
+                if(root)
+                    *target = json_is_true(root);
+            }
+
+            return 0;
+
+        case 'f':
+            if(root && !json_is_real(root)) {
+                set_error(s, "<validation>", "Expected real, got %s",
+                          type_name(root));
+                return -1;
+            }
+
+            if(!(s->flags & JSON_VALIDATE_ONLY)) {
+                double *target = va_arg(*ap, double*);
+                if(root)
+                    *target = json_real_value(root);
+            }
+
+            return 0;
+
+        case 'F':
+            if(root && !json_is_number(root)) {
+                set_error(s, "<validation>", "Expected real or integer, got %s",
+                          type_name(root));
+                return -1;
+            }
+
+            if(!(s->flags & JSON_VALIDATE_ONLY)) {
+                double *target = va_arg(*ap, double*);
+                if(root)
+                    *target = json_number_value(root);
+            }
+
+            return 0;
+
+        case 'O':
+            if(root && !(s->flags & JSON_VALIDATE_ONLY))
+                json_incref(root);
+            /* Fall through */
+
+        case 'o':
+            if(!(s->flags & JSON_VALIDATE_ONLY)) {
+                json_t **target = va_arg(*ap, json_t**);
+                if(root)
+                    *target = root;
+            }
+
+            return 0;
+
+        case 'n':
+            /* Never assign, just validate */
+            if(root && !json_is_null(root)) {
+                set_error(s, "<validation>", "Expected null, got %s",
+                          type_name(root));
+                return -1;
+            }
+            return 0;
+
+        default:
+            set_error(s, "<format>", "Unexpected format character '%c'",
+                      token(s));
+            return -1;
+    }
+}
+
+json_t *json_vpack_ex(json_error_t *error, size_t flags,
+                      const char *fmt, va_list ap)
+{
+    scanner_t s;
+    va_list ap_copy;
+    json_t *value;
+
+    if(!fmt || !*fmt) {
+        jsonp_error_init(error, "<format>");
+        jsonp_error_set(error, -1, -1, 0, "NULL or empty format string");
+        return NULL;
+    }
+    jsonp_error_init(error, NULL);
+
+    scanner_init(&s, error, flags, fmt);
+    next_token(&s);
+
+    va_copy(ap_copy, ap);
+    value = pack(&s, &ap_copy);
+    va_end(ap_copy);
+
+    if(!value)
+        return NULL;
+
+    next_token(&s);
+    if(token(&s)) {
+        json_decref(value);
+        set_error(&s, "<format>", "Garbage after format string");
+        return NULL;
+    }
+
+    return value;
+}
+
+json_t *json_pack_ex(json_error_t *error, size_t flags, const char *fmt, ...)
+{
+    json_t *value;
+    va_list ap;
+
+    va_start(ap, fmt);
+    value = json_vpack_ex(error, flags, fmt, ap);
+    va_end(ap);
+
+    return value;
+}
+
+json_t *json_pack(const char *fmt, ...)
+{
+    json_t *value;
+    va_list ap;
+
+    va_start(ap, fmt);
+    value = json_vpack_ex(NULL, 0, fmt, ap);
+    va_end(ap);
+
+    return value;
+}
+
+int json_vunpack_ex(json_t *root, json_error_t *error, size_t flags,
+                    const char *fmt, va_list ap)
+{
+    scanner_t s;
+    va_list ap_copy;
+
+    if(!root) {
+        jsonp_error_init(error, "<root>");
+        jsonp_error_set(error, -1, -1, 0, "NULL root value");
+        return -1;
+    }
+
+    if(!fmt || !*fmt) {
+        jsonp_error_init(error, "<format>");
+        jsonp_error_set(error, -1, -1, 0, "NULL or empty format string");
+        return -1;
+    }
+    jsonp_error_init(error, NULL);
+
+    scanner_init(&s, error, flags, fmt);
+    next_token(&s);
+
+    va_copy(ap_copy, ap);
+    if(unpack(&s, root, &ap_copy)) {
+        va_end(ap_copy);
+        return -1;
+    }
+    va_end(ap_copy);
+
+    next_token(&s);
+    if(token(&s)) {
+        set_error(&s, "<format>", "Garbage after format string");
+        return -1;
+    }
+
+    return 0;
+}
+
+int json_unpack_ex(json_t *root, json_error_t *error, size_t flags, const char *fmt, ...)
+{
+    int ret;
+    va_list ap;
+
+    va_start(ap, fmt);
+    ret = json_vunpack_ex(root, error, flags, fmt, ap);
+    va_end(ap);
+
+    return ret;
+}
+
+int json_unpack(json_t *root, const char *fmt, ...)
+{
+    int ret;
+    va_list ap;
+
+    va_start(ap, fmt);
+    ret = json_vunpack_ex(root, NULL, 0, fmt, ap);
+    va_end(ap);
+
+    return ret;
+}

compat/jansson/strbuffer.c

@@ -0,0 +1,111 @@
+/*
+ * Copyright (c) 2009-2016 Petri Lehtinen <petri@digip.org>
+ *
+ * Jansson is free software; you can redistribute it and/or modify
+ * it under the terms of the MIT license. See LICENSE for details.
+ */
+
+#ifndef _GNU_SOURCE
+#define _GNU_SOURCE
+#endif
+
+#include <stdlib.h>
+#include <string.h>
+#include "jansson_private.h"
+#include "strbuffer.h"
+
+#define STRBUFFER_MIN_SIZE  16
+#define STRBUFFER_FACTOR    2
+#define STRBUFFER_SIZE_MAX  ((size_t)-1)
+
+int strbuffer_init(strbuffer_t *strbuff)
+{
+    strbuff->size = STRBUFFER_MIN_SIZE;
+    strbuff->length = 0;
+
+    strbuff->value = jsonp_malloc(strbuff->size);
+    if(!strbuff->value)
+        return -1;
+
+    /* initialize to empty */
+    strbuff->value[0] = '\0';
+    return 0;
+}
+
+void strbuffer_close(strbuffer_t *strbuff)
+{
+    if(strbuff->value)
+        jsonp_free(strbuff->value);
+
+    strbuff->size = 0;
+    strbuff->length = 0;
+    strbuff->value = NULL;
+}
+
+void strbuffer_clear(strbuffer_t *strbuff)
+{
+    strbuff->length = 0;
+    strbuff->value[0] = '\0';
+}
+
+const char *strbuffer_value(const strbuffer_t *strbuff)
+{
+    return strbuff->value;
+}
+
+char *strbuffer_steal_value(strbuffer_t *strbuff)
+{
+    char *result = strbuff->value;
+    strbuff->value = NULL;
+    return result;
+}
+
+int strbuffer_append_byte(strbuffer_t *strbuff, char byte)
+{
+    return strbuffer_append_bytes(strbuff, &byte, 1);
+}
+
+int strbuffer_append_bytes(strbuffer_t *strbuff, const char *data, size_t size)
+{
+    if(size >= strbuff->size - strbuff->length)
+    {
+        size_t new_size;
+        char *new_value;
+
+        /* avoid integer overflow */
+        if (strbuff->size > STRBUFFER_SIZE_MAX / STRBUFFER_FACTOR
+            || size > STRBUFFER_SIZE_MAX - 1
+            || strbuff->length > STRBUFFER_SIZE_MAX - 1 - size)
+            return -1;
+
+        new_size = max(strbuff->size * STRBUFFER_FACTOR,
+                       strbuff->length + size + 1);
+
+        new_value = jsonp_malloc(new_size);
+        if(!new_value)
+            return -1;
+
+        memcpy(new_value, strbuff->value, strbuff->length);
+
+        jsonp_free(strbuff->value);
+        strbuff->value = new_value;
+        strbuff->size = new_size;
+    }
+
+    memcpy(strbuff->value + strbuff->length, data, size);
+    strbuff->length += size;
+    strbuff->value[strbuff->length] = '\0';
+
+    return 0;
+}
+
+char strbuffer_pop(strbuffer_t *strbuff)
+{
+    if(strbuff->length > 0) {
+        char c = strbuff->value[--strbuff->length];
+        strbuff->value[strbuff->length] = '\0';
+        return c;
+    }
+    else
+        return '\0';
+}

compat/jansson/strbuffer.h

@@ -0,0 +1,34 @@
+/*
+ * Copyright (c) 2009-2016 Petri Lehtinen <petri@digip.org>
+ *
+ * Jansson is free software; you can redistribute it and/or modify
+ * it under the terms of the MIT license. See LICENSE for details.
+ */
+
+#ifndef STRBUFFER_H
+#define STRBUFFER_H
+
+#include <stdlib.h>
+
+typedef struct {
+    char *value;
+    size_t length;   /* bytes used */
+    size_t size;     /* bytes allocated */
+} strbuffer_t;
+
+int strbuffer_init(strbuffer_t *strbuff);
+void strbuffer_close(strbuffer_t *strbuff);
+
+void strbuffer_clear(strbuffer_t *strbuff);
+
+const char *strbuffer_value(const strbuffer_t *strbuff);
+
+/* Steal the value and close the strbuffer */
+char *strbuffer_steal_value(strbuffer_t *strbuff);
+
+int strbuffer_append_byte(strbuffer_t *strbuff, char byte);
+int strbuffer_append_bytes(strbuffer_t *strbuff, const char *data, size_t size);
+
+char strbuffer_pop(strbuffer_t *strbuff);
+
+#endif

compat/jansson/strconv.c

@@ -0,0 +1,145 @@
+#include <assert.h>
+#include <errno.h>
+#include <stdio.h>
+#include <string.h>
+#include <math.h>
+#ifdef __MINGW32__
+#undef __NO_ISOCEXT /* ensure stdlib.h will declare prototypes for mingw own 'strtod' replacement, called '__strtod' */
+#endif
+#include "jansson_private.h"
+#include "strbuffer.h"
+
+/* need jansson_private_config.h to get the correct snprintf */
+#ifdef HAVE_CONFIG_H
+#include <jansson_private_config.h>
+#endif
+
+#ifdef __MINGW32__
+#define strtod __strtod
+#endif
+
+#if JSON_HAVE_LOCALECONV
+#include <locale.h>
+
+/*
+  - This code assumes that the decimal separator is exactly one
+    character.
+
+  - If setlocale() is called by another thread between the call to
+    localeconv() and the call to sprintf() or strtod(), the result may
+    be wrong. setlocale() is not thread-safe and should not be used
+    this way. Multi-threaded programs should use uselocale() instead.
+*/
+
+static void to_locale(strbuffer_t *strbuffer)
+{
+    const char *point;
+    char *pos;
+
+    point = localeconv()->decimal_point;
+    if(*point == '.') {
+        /* No conversion needed */
+        return;
+    }
+
+    pos = strchr(strbuffer->value, '.');
+    if(pos)
+        *pos = *point;
+}
+
+static void from_locale(char *buffer)
+{
+    const char *point;
+    char *pos;
+
+    point = localeconv()->decimal_point;
+    if(*point == '.') {
+        /* No conversion needed */
+        return;
+    }
+
+    pos = strchr(buffer, *point);
+    if(pos)
+        *pos = '.';
+}
+#endif
+
+int jsonp_strtod(strbuffer_t *strbuffer, double *out)
+{
+    double value;
+    char *end;
+
+#if JSON_HAVE_LOCALECONV
+    to_locale(strbuffer);
+#endif
+
+    errno = 0;
+    value = strtod(strbuffer->value, &end);
+    assert(end == strbuffer->value + strbuffer->length);
+
+    if((value == HUGE_VAL || value == -HUGE_VAL) && errno == ERANGE) {
+        /* Overflow */
+        return -1;
+    }
+
+    *out = value;
+    return 0;
+}
+
+int jsonp_dtostr(char *buffer, size_t size, double value, int precision)
+{
+    int ret;
+    char *start, *end;
+    size_t length;
+
+    if (precision == 0)
+        precision = 17;
+
+    ret = snprintf(buffer, size, "%.*g", precision, value);
+    if(ret < 0)
+        return -1;
+
+    length = (size_t)ret;
+    if(length >= size)
+        return -1;
+
+#if JSON_HAVE_LOCALECONV
+    from_locale(buffer);
+#endif
+
+    /* Make sure there's a dot or 'e' in the output. Otherwise
+       a real is converted to an integer when decoding */
+    if(strchr(buffer, '.') == NULL &&
+       strchr(buffer, 'e') == NULL)
+    {
+        if(length + 3 >= size) {
+            /* No space to append ".0" */
+            return -1;
+        }
+        buffer[length] = '.';
+        buffer[length + 1] = '0';
+        buffer[length + 2] = '\0';
+        length += 2;
+    }
+
+    /* Remove leading '+' from positive exponent. Also remove leading
+       zeros from exponents (added by some printf() implementations) */
+    start = strchr(buffer, 'e');
+    if(start) {
+        start++;
+        end = start + 1;
+
+        if(*start == '-')
+            start++;
+
+        while(*end == '0')
+            end++;
+
+        if(end != start) {
+            memmove(start, end, length - (size_t)(end - buffer));
+            length -= (size_t)(end - start);
+        }
+    }
+
+    return (int)length;
+}

compat/jansson/utf.c

@@ -0,0 +1,187 @@
+/*
+ * Copyright (c) 2009-2016 Petri Lehtinen <petri@digip.org>
+ *
+ * Jansson is free software; you can redistribute it and/or modify
+ * it under the terms of the MIT license. See LICENSE for details.
+ */
+
+#include <string.h>
+#include "utf.h"
+
+int utf8_encode(int32_t codepoint, char *buffer, size_t *size)
+{
+    if(codepoint < 0)
+        return -1;
+    else if(codepoint < 0x80)
+    {
+        buffer[0] = (char)codepoint;
+        *size = 1;
+    }
+    else if(codepoint < 0x800)
+    {
+        buffer[0] = 0xC0 + ((codepoint & 0x7C0) >> 6);
+        buffer[1] = 0x80 + ((codepoint & 0x03F));
+        *size = 2;
+    }
+    else if(codepoint < 0x10000)
+    {
+        buffer[0] = 0xE0 + ((codepoint & 0xF000) >> 12);
+        buffer[1] = 0x80 + ((codepoint & 0x0FC0) >> 6);
+        buffer[2] = 0x80 + ((codepoint & 0x003F));
+        *size = 3;
+    }
+    else if(codepoint <= 0x10FFFF)
+    {
+        buffer[0] = 0xF0 + ((codepoint & 0x1C0000) >> 18);
+        buffer[1] = 0x80 + ((codepoint & 0x03F000) >> 12);
+        buffer[2] = 0x80 + ((codepoint & 0x000FC0) >> 6);
+        buffer[3] = 0x80 + ((codepoint & 0x00003F));
+        *size = 4;
+    }
+    else
+        return -1;
+
+    return 0;
+}
+
+size_t utf8_check_first(char byte)
+{
+    unsigned char u = (unsigned char)byte;
+
+    if(u < 0x80)
+        return 1;
+
+    if(0x80 <= u && u <= 0xBF) {
+        /* second, third or fourth byte of a multi-byte
+           sequence, i.e. a "continuation byte" */
+        return 0;
+    }
+    else if(u == 0xC0 || u == 0xC1) {
+        /* overlong encoding of an ASCII byte */
+        return 0;
+    }
+    else if(0xC2 <= u && u <= 0xDF) {
+        /* 2-byte sequence */
+        return 2;
+    }
+
+    else if(0xE0 <= u && u <= 0xEF) {
+        /* 3-byte sequence */
+        return 3;
+    }
+    else if(0xF0 <= u && u <= 0xF4) {
+        /* 4-byte sequence */
+        return 4;
+    }
+    else { /* u >= 0xF5 */
+        /* Restricted (start of 4-, 5- or 6-byte sequence) or invalid
+           UTF-8 */
+        return 0;
+    }
+}
+
+size_t utf8_check_full(const char *buffer, size_t size, int32_t *codepoint)
+{
+    size_t i;
+    int32_t value = 0;
+    unsigned char u = (unsigned char)buffer[0];
+
+    if(size == 2)
+    {
+        value = u & 0x1F;
+    }
+    else if(size == 3)
+    {
+        value = u & 0xF;
+    }
+    else if(size == 4)
+    {
+        value = u & 0x7;
+    }
+    else
+        return 0;
+
+    for(i = 1; i < size; i++)
+    {
+        u = (unsigned char)buffer[i];
+
+        if(u < 0x80 || u > 0xBF) {
+            /* not a continuation byte */
+            return 0;
+        }
+
+        value = (value << 6) + (u & 0x3F);
+    }
+
+    if(value > 0x10FFFF) {
+        /* not in Unicode range */
+        return 0;
+    }
+
+    else if(0xD800 <= value && value <= 0xDFFF) {
+        /* invalid code point (UTF-16 surrogate halves) */
+        return 0;
+    }
+
+    else if((size == 2 && value < 0x80) ||
+            (size == 3 && value < 0x800) ||
+            (size == 4 && value < 0x10000)) {
+        /* overlong encoding */
+        return 0;
+    }
+
+    if(codepoint)
+        *codepoint = value;
+
+    return 1;
+}
+
+const char *utf8_iterate(const char *buffer, size_t bufsize, int32_t *codepoint)
+{
+    size_t count;
+    int32_t value;
+
+    if(!bufsize)
+        return buffer;
+
+    count = utf8_check_first(buffer[0]);
+    if(count <= 0)
+        return NULL;
+
+    if(count == 1)
+        value = (unsigned char)buffer[0];
+    else
+    {
+        if(count > bufsize || !utf8_check_full(buffer, count, &value))
+            return NULL;
+    }
+
+    if(codepoint)
+        *codepoint = value;
+
+    return buffer + count;
+}
+
+int utf8_check_string(const char *string, size_t length)
+{
+    size_t i;
+
+    for(i = 0; i < length; i++)
+    {
+        size_t count = utf8_check_first(string[i]);
+        if(count == 0)
+            return 0;
+        else if(count > 1)
+        {
+            if(count > length - i)
+                return 0;
+
+            if(!utf8_check_full(&string[i], count, NULL))
+                return 0;
+
+            i += count - 1;
+        }
+    }
+
+    return 1;
+}

compat/jansson/utf.h

@@ -0,0 +1,27 @@
+/*
+ * Copyright (c) 2009-2016 Petri Lehtinen <petri@digip.org>
+ *
+ * Jansson is free software; you can redistribute it and/or modify
+ * it under the terms of the MIT license. See LICENSE for details.
+ */
+
+#ifndef UTF_H
+#define UTF_H
+
+#ifdef HAVE_CONFIG_H
+#include <jansson_private_config.h>
+#endif
+
+#ifdef HAVE_STDINT_H
+#include <stdint.h>
+#endif
+
+int utf8_encode(int32_t codepoint, char *buffer, size_t *size);
+
+size_t utf8_check_first(char byte);
+size_t utf8_check_full(const char *buffer, size_t size, int32_t *codepoint);
+const char *utf8_iterate(const char *buffer, size_t size, int32_t *codepoint);
+
+int utf8_check_string(const char *string, size_t length);
+
+#endif

compat/libcpuid/CMakeLists.txt

@@ -3,19 +3,17 @@ project (cpuid C)
 
 add_definitions(/DVERSION="0.4.0")
 
-set(CMAKE_C_FLAGS_RELEASE "${CMAKE_C_FLAGS_RELEASE} -Os")
-
 set(HEADERS
     libcpuid.h
     libcpuid_types.h
     libcpuid_constants.h
     libcpuid_internal.h
-    amd_code_t.h
-    intel_code_t.h
-    recog_amd.h
-    recog_intel.h
-    asm-bits.h
-    libcpuid_util.h
+amd_code_t.h
+intel_code_t.h
+recog_amd.h
+recog_intel.h
+asm-bits.h
+libcpuid_util.h
     )
 
 set(SOURCES
@@ -26,13 +24,7 @@ set(SOURCES
     libcpuid_util.c
    )
 
-if (CMAKE_CL_64)
-    enable_language(ASM_MASM)
-    set(SOURCES_ASM masm-x64.asm)
-endif()
-
 add_library(cpuid STATIC
     ${HEADERS}
     ${SOURCES}
-    ${SOURCES_ASM}
     )

compat/libcpuid/cpuid_main.c

@@ -46,6 +46,11 @@ int set_error(cpu_error_t err)
 	return (int) err;
 }
 
+static void raw_data_t_constructor(struct cpu_raw_data_t* raw)
+{
+	memset(raw, 0, sizeof(struct cpu_raw_data_t));
+}
+
 static void cpu_id_t_constructor(struct cpu_id_t* id)
 {
 	memset(id, 0, sizeof(struct cpu_id_t));
@@ -55,6 +60,29 @@ static void cpu_id_t_constructor(struct cpu_id_t* id)
 	id->sse_size = -1;
 }
 
+static int parse_token(const char* expected_token, const char *token,
+                        const char *value, uint32_t array[][4], int limit, int *recognized)
+{
+	char format[32];
+	int veax, vebx, vecx, vedx;
+	int index;
+
+	if (*recognized) return 1; /* already recognized */
+	if (strncmp(token, expected_token, strlen(expected_token))) return 1; /* not what we search for */
+	sprintf(format, "%s[%%d]", expected_token);
+	*recognized = 1;
+	if (1 == sscanf(token, format, &index) && index >=0 && index < limit) {
+		if (4 == sscanf(value, "%x%x%x%x", &veax, &vebx, &vecx, &vedx)) {
+			array[index][0] = veax;
+			array[index][1] = vebx;
+			array[index][2] = vecx;
+			array[index][3] = vedx;
+			return 1;
+		}
+	}
+	return 0;
+}
+
 /* get_total_cpus() system specific code: uses OS routines to determine total number of CPUs */
 #ifdef __APPLE__
 #include <unistd.h>
@@ -89,7 +117,7 @@ static int get_total_cpus(void)
 #if defined linux || defined __linux__ || defined __sun
 #include <sys/sysinfo.h>
 #include <unistd.h>
- 
+
 static int get_total_cpus(void)
 {
 	return sysconf(_SC_NPROCESSORS_ONLN);
@@ -221,42 +249,42 @@ static void load_features_common(struct cpu_raw_data_t* raw, struct cpu_id_t* da
 
 static cpu_vendor_t cpuid_vendor_identify(const uint32_t *raw_vendor, char *vendor_str)
 {
-    int i;
-    cpu_vendor_t vendor = VENDOR_UNKNOWN;
-    const struct { cpu_vendor_t vendor; char match[16]; }
-    matchtable[NUM_CPU_VENDORS] = {
-        /* source: http://www.sandpile.org/ia32/cpuid.htm */
-        { VENDOR_INTEL		, "GenuineIntel" },
-        { VENDOR_AMD		, "AuthenticAMD" },
-        { VENDOR_CYRIX		, "CyrixInstead" },
-        { VENDOR_NEXGEN		, "NexGenDriven" },
-        { VENDOR_TRANSMETA	, "GenuineTMx86" },
-        { VENDOR_UMC		, "UMC UMC UMC " },
-        { VENDOR_CENTAUR	, "CentaurHauls" },
-        { VENDOR_RISE		, "RiseRiseRise" },
-        { VENDOR_SIS		, "SiS SiS SiS " },
-        { VENDOR_NSC		, "Geode by NSC" },
-    };
+	int i;
+	cpu_vendor_t vendor = VENDOR_UNKNOWN;
+	const struct { cpu_vendor_t vendor; char match[16]; }
+	matchtable[NUM_CPU_VENDORS] = {
+		/* source: http://www.sandpile.org/ia32/cpuid.htm */
+		{ VENDOR_INTEL		, "GenuineIntel" },
+		{ VENDOR_AMD		, "AuthenticAMD" },
+		{ VENDOR_CYRIX		, "CyrixInstead" },
+		{ VENDOR_NEXGEN		, "NexGenDriven" },
+		{ VENDOR_TRANSMETA	, "GenuineTMx86" },
+		{ VENDOR_UMC		, "UMC UMC UMC " },
+		{ VENDOR_CENTAUR	, "CentaurHauls" },
+		{ VENDOR_RISE		, "RiseRiseRise" },
+		{ VENDOR_SIS		, "SiS SiS SiS " },
+		{ VENDOR_NSC		, "Geode by NSC" },
+	};
 
-    memcpy(vendor_str + 0, &raw_vendor[1], 4);
-    memcpy(vendor_str + 4, &raw_vendor[3], 4);
-    memcpy(vendor_str + 8, &raw_vendor[2], 4);
-    vendor_str[12] = 0;
+	memcpy(vendor_str + 0, &raw_vendor[1], 4);
+	memcpy(vendor_str + 4, &raw_vendor[3], 4);
+	memcpy(vendor_str + 8, &raw_vendor[2], 4);
+	vendor_str[12] = 0;
 
-    /* Determine vendor: */
-    for (i = 0; i < NUM_CPU_VENDORS; i++)
-        if (!strcmp(vendor_str, matchtable[i].match)) {
-            vendor = matchtable[i].vendor;
-            break;
-        }
-    return vendor;
+	/* Determine vendor: */
+	for (i = 0; i < NUM_CPU_VENDORS; i++)
+		if (!strcmp(vendor_str, matchtable[i].match)) {
+			vendor = matchtable[i].vendor;
+			break;
+		}
+	return vendor;
 }
 
 static int cpuid_basic_identify(struct cpu_raw_data_t* raw, struct cpu_id_t* data)
 {
 	int i, j, basic, xmodel, xfamily, ext;
 	char brandstr[64] = {0};
-    data->vendor = cpuid_vendor_identify(raw->basic_cpuid[0], data->vendor_str);
+	data->vendor = cpuid_vendor_identify(raw->basic_cpuid[0], data->vendor_str);
 
 	if (data->vendor == VENDOR_UNKNOWN)
 		return set_error(ERR_CPU_UNKN);
@@ -274,7 +302,7 @@ static int cpuid_basic_identify(struct cpu_raw_data_t* raw, struct cpu_id_t* dat
 		data->ext_model = data->model + (xmodel << 4);
 	}
 	ext = raw->ext_cpuid[0][0] - 0x8000000;
-	
+
 	/* obtain the brand string, if present: */
 	if (ext >= 4) {
 		for (i = 0; i < 3; i++)
@@ -292,6 +320,27 @@ static int cpuid_basic_identify(struct cpu_raw_data_t* raw, struct cpu_id_t* dat
 	return set_error(ERR_OK);
 }
 
+static void make_list_from_string(const char* csv, struct cpu_list_t* list)
+{
+	int i, n, l, last;
+	l = (int) strlen(csv);
+	n = 0;
+	for (i = 0; i < l; i++) if (csv[i] == ',') n++;
+	n++;
+	list->num_entries = n;
+	list->names = (char**) malloc(sizeof(char*) * n);
+	last = -1;
+	n = 0;
+	for (i = 0; i <= l; i++) if (i == l || csv[i] == ',') {
+		list->names[n] = (char*) malloc(i - last);
+		memcpy(list->names[n], &csv[last + 1], i - last - 1);
+		list->names[n][i - last - 1] = '\0';
+		n++;
+		last = i;
+	}
+}
+
+
 /* Interface: */
 
 int cpuid_get_total_cpus(void)

compat/libcpuid/libcpuid.h

@@ -610,6 +610,39 @@ void cpu_exec_cpuid_ext(uint32_t* regs);
  */
 int cpuid_get_raw_data(struct cpu_raw_data_t* data);
 
+/**
+ * @brief Writes the raw CPUID data to a text file
+ * @param data - a pointer to cpu_raw_data_t structure
+ * @param filename - the path of the file, where the serialized data should be
+ *                   written. If empty, stdout will be used.
+ * @note This is intended primarily for debugging. On some processor, which is
+ *       not currently supported or not completely recognized by cpu_identify,
+ *       one can still successfully get the raw data and write it to a file.
+ *       libcpuid developers can later import this file and debug the detection
+ *       code as if running on the actual hardware.
+ *       The file is simple text format of "something=value" pairs. Version info
+ *       is also written, but the format is not intended to be neither backward-
+ *       nor forward compatible.
+ * @returns zero if successful, and some negative number on error.
+ *          The error message can be obtained by calling \ref cpuid_error.
+ *          @see cpu_error_t
+ */
+int cpuid_serialize_raw_data(struct cpu_raw_data_t* data, const char* filename);
+
+/**
+ * @brief Reads raw CPUID data from file
+ * @param data - a pointer to cpu_raw_data_t structure. The deserialized data will
+ *               be written here.
+ * @param filename - the path of the file, containing the serialized raw data.
+ *                   If empty, stdin will be used.
+ * @note This function may fail, if the file is created by different version of
+ *       the library. Also, see the notes on cpuid_serialize_raw_data.
+ * @returns zero if successful, and some negative number on error.
+ *          The error message can be obtained by calling \ref cpuid_error.
+ *          @see cpu_error_t
+*/
+int cpuid_deserialize_raw_data(struct cpu_raw_data_t* data, const char* filename);
+
 /**
  * @brief Identifies the CPU
  * @param raw - Input - a pointer to the raw CPUID data, which is obtained
@@ -635,6 +668,222 @@ int cpuid_get_raw_data(struct cpu_raw_data_t* data);
  */
 int cpu_identify(struct cpu_raw_data_t* raw, struct cpu_id_t* data);
 
+/**
+ * @brief Returns the short textual representation of a CPU flag
+ * @param feature - the feature, whose textual representation is wanted.
+ * @returns a constant string like "fpu", "tsc", "sse2", etc.
+ * @note the names of the returned flags are compatible with those from
+ *       /proc/cpuinfo in Linux, with the exception of `tm_amd'
+ */
+const char* cpu_feature_str(cpu_feature_t feature);
+
+/**
+ * @brief Returns textual description of the last error
+ *
+ * libcpuid stores an `errno'-style error status, whose description
+ * can be obtained with this function.
+ * @note This function is not thread-safe
+ * @see cpu_error_t
+ */
+const char* cpuid_error(void);
+
+/**
+ * @brief Executes RDTSC
+ *
+ * The RDTSC (ReaD Time Stamp Counter) instruction gives access to an
+ * internal 64-bit counter, which usually increments at each clock cycle.
+ * This can be used for various timing routines, and as a very precise
+ * clock source. It is set to zero on system startup. Beware that may not
+ * increment at the same frequency as the CPU. Consecutive calls of RDTSC
+ * are, however, guaranteed to return monotonically-increasing values.
+ *
+ * @param result - a pointer to a 64-bit unsigned integer, where the TSC value
+ *                 will be stored
+ *
+ * @note  If 100% compatibility is a concern, you must first check if the
+ * RDTSC instruction is present (if it is not, your program will crash
+ * with "invalid opcode" exception). Only some very old processors (i486,
+ * early AMD K5 and some Cyrix CPUs) lack that instruction - they should
+ * have become exceedingly rare these days. To verify RDTSC presence,
+ * run cpu_identify() and check flags[CPU_FEATURE_TSC].
+ *
+ * @note The monotonically increasing nature of the TSC may be violated
+ * on SMP systems, if their TSC clocks run at different rate. If the OS
+ * doesn't account for that, the TSC drift may become arbitrary large.
+ */
+void cpu_rdtsc(uint64_t* result);
+
+/**
+ * @brief Store TSC and timing info
+ *
+ * This function stores the current TSC value and current
+ * time info from a precise OS-specific clock source in the cpu_mark_t
+ * structure. The sys_clock field contains time with microsecond resolution.
+ * The values can later be used to measure time intervals, number of clocks,
+ * FPU frequency, etc.
+ * @see cpu_rdtsc
+ *
+ * @param mark [out] - a pointer to a cpu_mark_t structure
+ */
+void cpu_tsc_mark(struct cpu_mark_t* mark);
+
+/**
+ * @brief Calculate TSC and timing difference
+ *
+ * @param mark - input/output: a pointer to a cpu_mark_t sturcture, which has
+ *               already been initialized by cpu_tsc_mark. The difference in
+ *               TSC and time will be written here.
+ *
+ * This function calculates the TSC and time difference, by obtaining the
+ * current TSC and timing values and subtracting the contents of the `mark'
+ * structure from them. Results are written in the same structure.
+ *
+ * Example:
+ * @code
+ * ...
+ * struct cpu_mark_t mark;
+ * cpu_tsc_mark(&mark);
+ * foo();
+ * cpu_tsc_unmark(&mark);
+ * printf("Foo finished. Executed in %llu cycles and %llu usecs\n",
+ *        mark.tsc, mark.sys_clock);
+ * ...
+ * @endcode
+ */
+void cpu_tsc_unmark(struct cpu_mark_t* mark);
+
+/**
+ * @brief Calculates the CPU clock
+ *
+ * @param mark - pointer to a cpu_mark_t structure, which has been initialized
+ *   with cpu_tsc_mark and later `stopped' with cpu_tsc_unmark.
+ *
+ * @note For reliable results, the marked time interval should be at least about
+ * 10 ms.
+ *
+ * @returns the CPU clock frequency, in MHz. Due to measurement error, it will
+ * differ from the true value in a few least-significant bits. Accuracy depends
+ * on the timing interval - the more, the better. If the timing interval is
+ * insufficient, the result is -1. Also, see the comment on cpu_clock_measure
+ * for additional issues and pitfalls in using RDTSC for CPU frequency
+ * measurements.
+ */
+int cpu_clock_by_mark(struct cpu_mark_t* mark);
+
+/**
+ * @brief Returns the CPU clock, as reported by the OS
+ *
+ * This function uses OS-specific functions to obtain the CPU clock. It may
+ * differ from the true clock for several reasons:<br><br>
+ *
+ * i) The CPU might be in some power saving state, while the OS reports its
+ *    full-power frequency, or vice-versa.<br>
+ * ii) In some cases you can raise or lower the CPU frequency with overclocking
+ *     utilities and the OS will not notice.
+ *
+ * @returns the CPU clock frequency in MHz. If the OS is not (yet) supported
+ * or lacks the necessary reporting machinery, the return value is -1
+ */
+int cpu_clock_by_os(void);
+
+/**
+ * @brief Measure the CPU clock frequency
+ *
+ * @param millis - How much time to waste in the busy-wait cycle. In millisecs.
+ *                 Useful values 10 - 1000
+ * @param quad_check - Do a more thorough measurement if nonzero
+ *                     (see the explanation).
+ *
+ * The function performs a busy-wait cycle for the given time and calculates
+ * the CPU frequency by the difference of the TSC values. The accuracy of the
+ * calculation depends on the length of the busy-wait cycle: more is better,
+ * but 100ms should be enough for most purposes.
+ *
+ * While this will calculate the CPU frequency correctly in most cases, there are
+ * several reasons why it might be incorrect:<br>
+ *
+ * i) RDTSC doesn't guarantee it will run at the same clock as the CPU.
+ *    Apparently there aren't CPUs at the moment, but still, there's no
+ *    guarantee.<br>
+ * ii) The CPU might be in a low-frequency power saving mode, and the CPU
+ *     might be switched to higher frequency at any time. If this happens
+ *     during the measurement, the result can be anywhere between the
+ *     low and high frequencies. Also, if you're interested in the
+ *     high frequency value only, this function might return the low one
+ *     instead.<br>
+ * iii) On SMP systems exhibiting TSC drift (see \ref cpu_rdtsc)
+ *
+ * the quad_check option will run four consecutive measurements and
+ * then return the average of the two most-consistent results. The total
+ * runtime of the function will still be `millis' - consider using
+ * a bit more time for the timing interval.
+ *
+ * Finally, for benchmarking / CPU intensive applications, the best strategy is
+ * to use the cpu_tsc_mark() / cpu_tsc_unmark() / cpu_clock_by_mark() method.
+ * Begin by mark()-ing about one second after application startup (allowing the
+ * power-saving manager to kick in and rise the frequency during that time),
+ * then unmark() just before application finishing. The result will most
+ * acurately represent at what frequency your app was running.
+ *
+ * @returns the CPU clock frequency in MHz (within some measurement error
+ * margin). If RDTSC is not supported, the result is -1.
+ */
+int cpu_clock_measure(int millis, int quad_check);
+
+/**
+ * @brief Measure the CPU clock frequency using instruction-counting
+ *
+ * @param millis - how much time to allocate for each run, in milliseconds
+ * @param runs - how many runs to perform
+ *
+ * The function performs a busy-wait cycle using a known number of "heavy" (SSE)
+ * instructions. These instructions run at (more or less guaranteed) 1 IPC rate,
+ * so by running a busy loop for a fixed amount of time, and measuring the
+ * amount of instructions done, the CPU clock is accurately measured.
+ *
+ * Of course, this function is still affected by the power-saving schemes, so
+ * the warnings as of cpu_clock_measure() still apply. However, this function is
+ * immune to problems with detection, related to the Intel Nehalem's "Turbo"
+ * mode, where the internal clock is raised, but the RDTSC rate is unaffected.
+ *
+ * The function will run for about (millis * runs) milliseconds.
+ * You can make only a single busy-wait run (runs == 1); however, this can
+ * be affected by task scheduling (which will break the counting), so allowing
+ * more than one run is recommended. As run length is not imperative for
+ * accurate readings (e.g., 50ms is sufficient), you can afford a lot of short
+ * runs, e.g. 10 runs of 50ms or 20 runs of 25ms.
+ *
+ * Recommended values - millis = 50, runs = 4. For more robustness,
+ * increase the number of runs.
+ * 
+ * NOTE: on Bulldozer and later CPUs, the busy-wait cycle runs at 1.4 IPC, thus
+ * the results are skewed. This is corrected internally by dividing the resulting
+ * value by 1.4.
+ * However, this only occurs if the thread is executed on a single CMT
+ * module - if there are other threads competing for resources, the results are
+ * unpredictable. Make sure you run cpu_clock_by_ic() on a CPU that is free from
+ * competing threads, or if there are such threads, they shouldn't exceed the
+ * number of modules. On a Bulldozer X8, that means 4 threads.
+ *
+ * @returns the CPU clock frequency in MHz (within some measurement error
+ * margin). If SSE is not supported, the result is -1. If the input parameters
+ * are incorrect, or some other internal fault is detected, the result is -2.
+ */
+int cpu_clock_by_ic(int millis, int runs);
+
+/**
+ * @brief Get the CPU clock frequency (all-in-one method)
+ *
+ * This is an all-in-one method for getting the CPU clock frequency.
+ * It tries to use the OS for that. If the OS doesn't have this info, it
+ * uses cpu_clock_measure with 200ms time interval and quadruple checking.
+ *
+ * @returns the CPU clock frequency in MHz. If every possible method fails,
+ * the result is -1.
+ */
+int cpu_clock(void);
+
+
 /**
  * @brief The return value of cpuid_get_epc().
  * @details
@@ -667,6 +916,230 @@ struct cpu_epc_t cpuid_get_epc(int index, const struct cpu_raw_data_t* raw);
  */
 const char* cpuid_lib_version(void);
 
+typedef void (*libcpuid_warn_fn_t) (const char *msg);
+/**
+ * @brief Sets the warning print function
+ *
+ * In some cases, the internal libcpuid machinery would like to emit useful
+ * debug warnings. By default, these warnings are written to stderr. However,
+ * you can set a custom function that will receive those warnings.
+ *
+ * @param warn_fun - the warning function you want to set. If NULL, warnings
+ *                   are disabled. The function takes const char* argument.
+ *
+ * @returns the current warning function. You can use the return value to
+ * keep the previous warning function and restore it at your discretion.
+ */
+libcpuid_warn_fn_t cpuid_set_warn_function(libcpuid_warn_fn_t warn_fun);
+
+/**
+ * @brief Sets the verbosiness level
+ *
+ * When the verbosiness level is above zero, some functions might print
+ * diagnostic information about what are they doing. The higher the level is,
+ * the more detail is printed. Level zero is guaranteed to omit all such
+ * output. The output is written using the same machinery as the warnings,
+ * @see cpuid_set_warn_function()
+ *
+ * @param level the desired verbosiness level. Useful values 0..2 inclusive
+ */
+void cpuid_set_verbosiness_level(int level);
+
+
+/**
+ * @brief Obtains the CPU vendor from CPUID from the current CPU
+ * @note The result is cached.
+ * @returns VENDOR_UNKNOWN if failed, otherwise the CPU vendor type.
+ *          @see cpu_vendor_t
+ */
+cpu_vendor_t cpuid_get_vendor(void);
+
+/**
+ * @brief a structure that holds a list of processor names
+ */
+struct cpu_list_t {
+	/** Number of entries in the list */
+	int num_entries;
+	/** Pointers to names. There will be num_entries of them */
+	char **names;
+};
+
+/**
+ * @brief Gets a list of all known CPU names from a specific vendor.
+ *
+ * This function compiles a list of all known CPU (code)names
+ * (i.e. the possible values of cpu_id_t::cpu_codename) for the given vendor.
+ *
+ * There are about 100 entries for Intel and AMD, and a few for the other
+ * vendors. The list is written out in approximate chronological introduction
+ * order of the parts.
+ *
+ * @param vendor the vendor to be queried
+ * @param list [out] the resulting list will be written here.
+ * NOTE: As the memory is dynamically allocated, be sure to call
+ *       cpuid_free_cpu_list() after you're done with the data
+ * @see cpu_list_t
+ */
+void cpuid_get_cpu_list(cpu_vendor_t vendor, struct cpu_list_t* list);
+
+/**
+ * @brief Frees a CPU list
+ *
+ * This function deletes all the memory associated with a CPU list, as obtained
+ * by cpuid_get_cpu_list()
+ *
+ * @param list - the list to be free()'d.
+ */
+void cpuid_free_cpu_list(struct cpu_list_t* list);
+
+struct msr_driver_t;
+/**
+ * @brief Starts/opens a driver, needed to read MSRs (Model Specific Registers)
+ *
+ * On systems that support it, this function will create a temporary
+ * system driver, that has privileges to execute the RDMSR instruction.
+ * After the driver is created, you can read MSRs by calling \ref cpu_rdmsr
+ *
+ * @returns a handle to the driver on success, and NULL on error.
+ *          The error message can be obtained by calling \ref cpuid_error.
+ *          @see cpu_error_t
+ */
+struct msr_driver_t* cpu_msr_driver_open(void);
+
+/**
+ * @brief Similar to \ref cpu_msr_driver_open, but accept one parameter
+ *
+ * This function works on certain operating systems (GNU/Linux, FreeBSD)
+ *
+ * @param core_num specify the core number for MSR.
+ *          The first core number is 0.
+ *          The last core number is \ref cpuid_get_total_cpus - 1.
+ *
+ * @returns a handle to the driver on success, and NULL on error.
+ *          The error message can be obtained by calling \ref cpuid_error.
+ *          @see cpu_error_t
+ */
+struct msr_driver_t* cpu_msr_driver_open_core(unsigned core_num);
+
+/**
+ * @brief Reads a Model-Specific Register (MSR)
+ *
+ * If the CPU has MSRs (as indicated by the CPU_FEATURE_MSR flag), you can
+ * read a MSR with the given index by calling this function.
+ *
+ * There are several prerequisites you must do before reading MSRs:
+ * 1) You must ensure the CPU has RDMSR. Check the CPU_FEATURE_MSR flag
+ *    in cpu_id_t::flags
+ * 2) You must ensure that the CPU implements the specific MSR you intend to
+ *    read.
+ * 3) You must open a MSR-reader driver. RDMSR is a privileged instruction and
+ *    needs ring-0 access in order to work. This temporary driver is created
+ *    by calling \ref cpu_msr_driver_open
+ *
+ * @param handle - a handle to the MSR reader driver, as created by
+ *                 cpu_msr_driver_open
+ * @param msr_index - the numeric ID of the MSR you want to read
+ * @param result - a pointer to a 64-bit integer, where the MSR value is stored
+ *
+ * @returns zero if successful, and some negative number on error.
+ *          The error message can be obtained by calling \ref cpuid_error.
+ *          @see cpu_error_t
+ */
+int cpu_rdmsr(struct msr_driver_t* handle, uint32_t msr_index, uint64_t* result);
+
+
+typedef enum {
+	INFO_MPERF,                /*!< Maximum performance frequency clock. This
+                                    is a counter, which increments as a
+                                    proportion of the actual processor speed. */
+	INFO_APERF,                /*!< Actual performance frequency clock. This
+                                    accumulates the core clock counts when the
+                                    core is active. */
+	INFO_MIN_MULTIPLIER,       /*!< Minimum CPU:FSB ratio for this CPU,
+                                    multiplied by 100. */
+	INFO_CUR_MULTIPLIER,       /*!< Current CPU:FSB ratio, multiplied by 100.
+                                    e.g., a CPU:FSB value of 18.5 reads as
+                                    "1850". */
+	INFO_MAX_MULTIPLIER,       /*!< Maximum CPU:FSB ratio for this CPU,
+                                    multiplied by 100. */
+	INFO_TEMPERATURE,          /*!< The current core temperature in Celsius. */
+	INFO_THROTTLING,           /*!< 1 if the current logical processor is
+                                    throttling. 0 if it is running normally. */
+	INFO_VOLTAGE,              /*!< The current core voltage in Volt,
+	                            multiplied by 100. */
+	INFO_BCLK,                 /*!< See \ref INFO_BUS_CLOCK. */
+	INFO_BUS_CLOCK,            /*!< The main bus clock in MHz,
+	                            e.g., FSB/QPI/DMI/HT base clock,
+	                            multiplied by 100. */
+} cpu_msrinfo_request_t;
+
+/**
+ * @brief Similar to \ref cpu_rdmsr, but extract a range of bits
+ *
+ * @param handle - a handle to the MSR reader driver, as created by
+ *                 cpu_msr_driver_open
+ * @param msr_index - the numeric ID of the MSR you want to read
+ * @param highbit - the high bit in range, must be inferior to 64
+ * @param lowbit - the low bit in range, must be equal or superior to 0
+ * @param result - a pointer to a 64-bit integer, where the MSR value is stored
+ *
+ * @returns zero if successful, and some negative number on error.
+ *          The error message can be obtained by calling \ref cpuid_error.
+ *          @see cpu_error_t
+ */
+int cpu_rdmsr_range(struct msr_driver_t* handle, uint32_t msr_index, uint8_t highbit,
+                    uint8_t lowbit, uint64_t* result);
+
+/**
+ * @brief Reads extended CPU information from Model-Specific Registers.
+ * @param handle - a handle to an open MSR driver, @see cpu_msr_driver_open
+ * @param which - which info field should be returned. A list of
+ *                available information entities is listed in the
+ *                cpu_msrinfo_request_t enum.
+ * @retval - if the requested information is available for the current
+ *           processor model, the respective value is returned.
+ *           if no information is available, or the CPU doesn't support
+ *           the query, the special value CPU_INVALID_VALUE is returned
+ * @note This function is not MT-safe. If you intend to call it from multiple
+ *       threads, guard it through a mutex or a similar primitive.
+ */
+int cpu_msrinfo(struct msr_driver_t* handle, cpu_msrinfo_request_t which);
+#define CPU_INVALID_VALUE 0x3fffffff
+
+/**
+ * @brief Writes the raw MSR data to a text file
+ * @param data - a pointer to msr_driver_t structure
+ * @param filename - the path of the file, where the serialized data should be
+ *                   written. If empty, stdout will be used.
+ * @note This is intended primarily for debugging. On some processor, which is
+ *       not currently supported or not completely recognized by cpu_identify,
+ *       one can still successfully get the raw data and write it to a file.
+ *       libcpuid developers can later import this file and debug the detection
+ *       code as if running on the actual hardware.
+ *       The file is simple text format of "something=value" pairs. Version info
+ *       is also written, but the format is not intended to be neither backward-
+ *       nor forward compatible.
+ * @returns zero if successful, and some negative number on error.
+ *          The error message can be obtained by calling \ref cpuid_error.
+ *          @see cpu_error_t
+ */
+int msr_serialize_raw_data(struct msr_driver_t* handle, const char* filename);
+
+/**
+ * @brief Closes an open MSR driver
+ *
+ * This function unloads the MSR driver opened by cpu_msr_driver_open and
+ * frees any resources associated with it.
+ *
+ * @param handle - a handle to the MSR reader driver, as created by
+ *                 cpu_msr_driver_open
+ *
+ * @returns zero if successful, and some negative number on error.
+ *          The error message can be obtained by calling \ref cpuid_error.
+ *          @see cpu_error_t
+ */
+int cpu_msr_driver_close(struct msr_driver_t* handle);
+
 #ifdef __cplusplus
 }; /* extern "C" */
 #endif

compat/libcpuid/libcpuid_internal.h

@@ -58,48 +58,6 @@ struct internal_id_info_t {
 	int score; // detection (matchtable) score
 };
 
-#define LBIT(x) (((long long) 1) << x)
-
-enum _common_bits_t {
-	_M_                     = LBIT(  0 ),
-	MOBILE_                 = LBIT(  1 ),
-	_MP_                    = LBIT(  2 ),
-};
-
-// additional detection bits for Intel CPUs:
-enum _intel_bits_t {
-	PENTIUM_                = LBIT( 10 ),
-	CELERON_                = LBIT( 11 ),
-	CORE_                   = LBIT( 12 ),
-	_I_                     = LBIT( 13 ),
-	_3                      = LBIT( 14 ),
-	_5                      = LBIT( 15 ),
-	_7                      = LBIT( 16 ),
-	XEON_                   = LBIT( 17 ),
-	ATOM_                   = LBIT( 18 ),
-};
-typedef enum _intel_bits_t intel_bits_t;
-
-enum _amd_bits_t {
-	ATHLON_      = LBIT( 10 ),
-	_XP_         = LBIT( 11 ),
-	DURON_       = LBIT( 12 ),
-	SEMPRON_     = LBIT( 13 ),
-	OPTERON_     = LBIT( 14 ),
-	TURION_      = LBIT( 15 ),
-	_LV_         = LBIT( 16 ),
-	_64_         = LBIT( 17 ),
-	_X2          = LBIT( 18 ),
-	_X3          = LBIT( 19 ),
-	_X4          = LBIT( 20 ),
-	_X6          = LBIT( 21 ),
-	_FX          = LBIT( 22 ),
-	_APU_        = LBIT( 23 ),
-};
-typedef enum _amd_bits_t amd_bits_t;
-
-
-
 int cpu_ident_internal(struct cpu_raw_data_t* raw, struct cpu_id_t* data, 
 		       struct internal_id_info_t* internal);
 

compat/libcpuid/libcpuid_util.c

@@ -0,0 +1,218 @@
+/*
+ * Copyright 2008  Veselin Georgiev,
+ * anrieffNOSPAM @ mgail_DOT.com (convert to gmail)
+ *
+ * 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.
+ *
+ * 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.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdarg.h>
+#include <string.h>
+#include <ctype.h>
+#include "libcpuid.h"
+#include "libcpuid_util.h"
+
+int _current_verboselevel;
+
+void match_features(const struct feature_map_t* matchtable, int count, uint32_t reg, struct cpu_id_t* data)
+{
+	int i;
+	for (i = 0; i < count; i++)
+		if (reg & (1u << matchtable[i].bit))
+			data->flags[matchtable[i].feature] = 1;
+}
+
+static void default_warn(const char *msg)
+{
+	fprintf(stderr, "%s", msg);
+}
+
+libcpuid_warn_fn_t _warn_fun = default_warn;
+
+#if defined(_MSC_VER)
+#	define vsnprintf _vsnprintf
+#endif
+void warnf(const char* format, ...)
+{
+	char buff[1024];
+	va_list va;
+	if (!_warn_fun) return;
+	va_start(va, format);
+	vsnprintf(buff, sizeof(buff), format, va);
+	va_end(va);
+	_warn_fun(buff);
+}
+
+void debugf(int verboselevel, const char* format, ...)
+{
+	char buff[1024];
+	va_list va;
+	if (verboselevel > _current_verboselevel) return;
+	va_start(va, format);
+	vsnprintf(buff, sizeof(buff), format, va);
+	va_end(va);
+	_warn_fun(buff);
+}
+
+static int popcount64(uint64_t mask)
+{
+	int num_set_bits = 0;
+	
+	while (mask) {
+		mask &= mask - 1;
+		num_set_bits++;
+	}
+	
+	return num_set_bits;
+}
+
+static int score(const struct match_entry_t* entry, const struct cpu_id_t* data,
+                 int brand_code, uint64_t bits, int model_code)
+{
+	int res = 0;
+	if (entry->family	== data->family    ) res += 2;
+	if (entry->model	== data->model     ) res += 2;
+	if (entry->stepping	== data->stepping  ) res += 2;
+	if (entry->ext_family	== data->ext_family) res += 2;
+	if (entry->ext_model	== data->ext_model ) res += 2;
+	if (entry->ncores	== data->num_cores ) res += 2;
+	if (entry->l2cache	== data->l2_cache  ) res += 1;
+	if (entry->l3cache	== data->l3_cache  ) res += 1;
+	if (entry->brand_code   == brand_code  ) res += 2;
+	if (entry->model_code   == model_code  ) res += 2;
+	
+	res += popcount64(entry->model_bits & bits) * 2;
+	return res;
+}
+
+int match_cpu_codename(const struct match_entry_t* matchtable, int count,
+                       struct cpu_id_t* data, int brand_code, uint64_t bits,
+                       int model_code)
+{
+	int bestscore = -1;
+	int bestindex = 0;
+	int i, t;
+	
+	debugf(3, "Matching cpu f:%d, m:%d, s:%d, xf:%d, xm:%d, ncore:%d, l2:%d, bcode:%d, bits:%llu, code:%d\n",
+		data->family, data->model, data->stepping, data->ext_family,
+		data->ext_model, data->num_cores, data->l2_cache, brand_code, (unsigned long long) bits, model_code);
+	
+	for (i = 0; i < count; i++) {
+		t = score(&matchtable[i], data, brand_code, bits, model_code);
+		debugf(3, "Entry %d, `%s', score %d\n", i, matchtable[i].name, t);
+		if (t > bestscore) {
+			debugf(2, "Entry `%s' selected - best score so far (%d)\n", matchtable[i].name, t);
+			bestscore = t;
+			bestindex = i;
+		}
+	}
+	strcpy(data->cpu_codename, matchtable[bestindex].name);
+	return bestscore;
+}
+
+void generic_get_cpu_list(const struct match_entry_t* matchtable, int count,
+                          struct cpu_list_t* list)
+{
+	int i, j, n, good;
+	n = 0;
+	list->names = (char**) malloc(sizeof(char*) * count);
+	for (i = 0; i < count; i++) {
+		if (strstr(matchtable[i].name, "Unknown")) continue;
+		good = 1;
+		for (j = n - 1; j >= 0; j--)
+			if (!strcmp(list->names[j], matchtable[i].name)) {
+				good = 0;
+				break;
+			}
+		if (!good) continue;
+#if defined(_MSC_VER)
+		list->names[n++] = _strdup(matchtable[i].name);
+#else
+		list->names[n++] = strdup(matchtable[i].name);
+#endif
+	}
+	list->num_entries = n;
+}
+
+static int xmatch_entry(char c, const char* p)
+{
+	int i, j;
+	if (c == 0) return -1;
+	if (c == p[0]) return 1;
+	if (p[0] == '.') return 1;
+	if (p[0] == '#' && isdigit(c)) return 1;
+	if (p[0] == '[') {
+		j = 1;
+		while (p[j] && p[j] != ']') j++;
+		if (!p[j]) return -1;
+		for (i = 1; i < j; i++)
+			if (p[i] == c) return j + 1;
+	}
+	return -1;
+}
+
+int match_pattern(const char* s, const char* p)
+{
+	int i, j, dj, k, n, m;
+	n = (int) strlen(s);
+	m = (int) strlen(p);
+	for (i = 0; i < n; i++) {
+		if (xmatch_entry(s[i], p) != -1) {
+			j = 0;
+			k = 0;
+			while (j < m && ((dj = xmatch_entry(s[i + k], p + j)) != -1)) {
+				k++;
+				j += dj;
+			}
+			if (j == m) return i + 1;
+		}
+	}
+	return 0;
+}
+
+struct cpu_id_t* get_cached_cpuid(void)
+{
+	static int initialized = 0;
+	static struct cpu_id_t id;
+	if (initialized) return &id;
+	if (cpu_identify(NULL, &id))
+		memset(&id, 0, sizeof(id));
+	initialized = 1;
+	return &id;
+}
+
+int match_all(uint64_t bits, uint64_t mask)
+{
+	return (bits & mask) == mask;
+}
+
+void debug_print_lbits(int debuglevel, uint64_t mask)
+{
+	int i, first = 0;
+	for (i = 0; i < 64; i++) if (mask & (((uint64_t) 1) << i)) {
+		if (first) first = 0;
+		else debugf(2, " + ");
+		debugf(2, "LBIT(%d)", i);
+	}
+	debugf(2, "\n");
+}

compat/libcpuid/libcpuid_util.h

@@ -28,6 +28,8 @@
 
 #define COUNT_OF(array) (sizeof(array) / sizeof(array[0]))
 
+#define LBIT(x) (((long long) 1) << x)
+
 struct feature_map_t {
 	unsigned bit;
 	cpu_feature_t feature;
@@ -48,6 +50,20 @@ struct match_entry_t {
 int match_cpu_codename(const struct match_entry_t* matchtable, int count,
                        struct cpu_id_t* data, int brand_code, uint64_t bits,
                        int model_code);
+
+void warnf(const char* format, ...)
+#ifdef __GNUC__
+__attribute__((format(printf, 1, 2)))
+#endif
+;
+void debugf(int verboselevel, const char* format, ...)
+#ifdef __GNUC__
+__attribute__((format(printf, 2, 3)))
+#endif
+;
+void generic_get_cpu_list(const struct match_entry_t* matchtable, int count,
+                          struct cpu_list_t* list);
+
 /*
  * Seek for a pattern in `haystack'.
  * Pattern may be an fixed string, or contain the special metacharacters
@@ -70,9 +86,15 @@ struct cpu_id_t* get_cached_cpuid(void);
 /* returns true if all bits of mask are present in `bits'. */
 int match_all(uint64_t bits, uint64_t mask);
 
+/* print what bits a mask consists of */
+void debug_print_lbits(int debuglevel, uint64_t mask);
+
 /*
  * Sets the current errno
  */
 int set_error(cpu_error_t err);
 
+extern libcpuid_warn_fn_t _warn_fun;
+extern int _current_verboselevel;
+
 #endif /* __LIBCPUID_UTIL_H__ */

compat/libcpuid/recog_amd.c

@@ -0,0 +1,549 @@
+/*
+ * Copyright 2008  Veselin Georgiev,
+ * anrieffNOSPAM @ mgail_DOT.com (convert to gmail)
+ *
+ * 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.
+ *
+ * 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.
+ */
+
+#include <stdio.h>
+#include <string.h>
+#include <ctype.h>
+#include "libcpuid.h"
+#include "libcpuid_util.h"
+#include "libcpuid_internal.h"
+#include "recog_amd.h"
+
+const struct amd_code_str { amd_code_t code; char *str; } amd_code_str[] = {
+	#define CODE(x) { x, #x }
+	#define CODE2(x, y) CODE(x)
+	#include "amd_code_t.h"
+	#undef CODE
+};
+
+struct amd_code_and_bits_t {
+	int code;
+	uint64_t bits;
+};
+
+enum _amd_bits_t {
+	ATHLON_      = LBIT(  0 ),
+	_XP_         = LBIT(  1 ),
+	_M_          = LBIT(  2 ),
+	_MP_         = LBIT(  3 ),
+	MOBILE_      = LBIT(  4 ),
+	DURON_       = LBIT(  5 ),
+	SEMPRON_     = LBIT(  6 ),
+	OPTERON_     = LBIT(  7 ),
+	TURION_      = LBIT(  8 ),
+	_LV_         = LBIT(  9 ),
+	_64_         = LBIT( 10 ),
+	_X2          = LBIT( 11 ),
+	_X3          = LBIT( 12 ),
+	_X4          = LBIT( 13 ),
+	_X6          = LBIT( 14 ),
+	_FX          = LBIT( 15 ),
+};
+typedef enum _amd_bits_t amd_bits_t;
+
+enum _amd_model_codes_t {
+	// Only for Ryzen CPUs:
+	_1400,
+	_1500,
+	_1600,
+};
+
+
+const struct match_entry_t cpudb_amd[] = {
+	{ -1, -1, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "Unknown AMD CPU"               },
+	
+	/* 486 and the likes */
+	{  4, -1, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "Unknown AMD 486"               },
+	{  4,  3, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "AMD 486DX2"                    },
+	{  4,  7, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "AMD 486DX2WB"                  },
+	{  4,  8, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "AMD 486DX4"                    },
+	{  4,  9, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "AMD 486DX4WB"                  },
+	
+	/* Pentia clones */
+	{  5, -1, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "Unknown AMD 586"               },
+	{  5,  0, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "K5"                            },
+	{  5,  1, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "K5"                            },
+	{  5,  2, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "K5"                            },
+	{  5,  3, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "K5"                            },
+	
+	/* The K6 */
+	{  5,  6, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "K6"                            },
+	{  5,  7, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "K6"                            },
+	
+	{  5,  8, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "K6-2"                          },
+	{  5,  9, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "K6-III"                        },
+	{  5, 10, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "Unknown K6"                    },
+	{  5, 11, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "Unknown K6"                    },
+	{  5, 12, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "Unknown K6"                    },
+	{  5, 13, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "K6-2+"                         },
+	
+	/* Athlon et al. */
+	{  6,  1, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "Athlon (Slot-A)"               },
+	{  6,  2, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "Athlon (Slot-A)"               },
+	{  6,  3, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "Duron (Spitfire)"              },
+	{  6,  4, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "Athlon (ThunderBird)"          },
+	
+	{  6,  6, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "Unknown Athlon"                },
+	{  6,  6, -1, -1,   -1,   1,    -1,    -1, NC, ATHLON_             ,     0, "Athlon (Palomino)"             },
+	{  6,  6, -1, -1,   -1,   1,    -1,    -1, NC, ATHLON_|_MP_        ,     0, "Athlon MP (Palomino)"          },
+	{  6,  6, -1, -1,   -1,   1,    -1,    -1, NC, DURON_              ,     0, "Duron (Palomino)"              },
+	{  6,  6, -1, -1,   -1,   1,    -1,    -1, NC, ATHLON_|_XP_        ,     0, "Athlon XP"                     },
+	
+	{  6,  7, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "Unknown Athlon XP"             },
+	{  6,  7, -1, -1,   -1,   1,    -1,    -1, NC, DURON_              ,     0, "Duron (Morgan)"                },
+	
+	{  6,  8, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "Athlon XP"                     },
+	{  6,  8, -1, -1,   -1,   1,    -1,    -1, NC, ATHLON_             ,     0, "Athlon XP (Thoroughbred)"      },
+	{  6,  8, -1, -1,   -1,   1,    -1,    -1, NC, ATHLON_|_XP_        ,     0, "Athlon XP (Thoroughbred)"      },
+	{  6,  8, -1, -1,   -1,   1,    -1,    -1, NC, DURON_              ,     0, "Duron (Applebred)"             },
+	{  6,  8, -1, -1,   -1,   1,    -1,    -1, NC, SEMPRON_            ,     0, "Sempron (Thoroughbred)"        },
+	{  6,  8, -1, -1,   -1,   1,   128,    -1, NC, SEMPRON_            ,     0, "Sempron (Thoroughbred)"        },
+	{  6,  8, -1, -1,   -1,   1,   256,    -1, NC, SEMPRON_            ,     0, "Sempron (Thoroughbred)"        },
+	{  6,  8, -1, -1,   -1,   1,    -1,    -1, NC, ATHLON_|_MP_        ,     0, "Athlon MP (Thoroughbred)"      },
+	{  6,  8, -1, -1,   -1,   1,    -1,    -1, NC, ATHLON_|_XP_|_M_    ,     0, "Mobile Athlon (T-Bred)"        },
+	{  6,  8, -1, -1,   -1,   1,    -1,    -1, NC, ATHLON_|_XP_|_M_|_LV_,    0, "Mobile Athlon (T-Bred)"        },
+	
+	{  6, 10, -1, -1,   -1,   1,    -1,    -1, NC, 0                   ,     0, "Athlon XP (Barton)"            },
+	{  6, 10, -1, -1,   -1,   1,   512,    -1, NC, ATHLON_|_XP_        ,     0, "Athlon XP (Barton)"            },
+	{  6, 10, -1, -1,   -1,   1,   512,    -1, NC, SEMPRON_            ,     0, "Sempron (Barton)"              },
+	{  6, 10, -1, -1,   -1,   1,   256,    -1, NC, SEMPRON_            ,     0, "Sempron (Thorton)"             },
+	{  6, 10, -1, -1,   -1,   1,   256,    -1, NC, ATHLON_|_XP_        ,     0, "Athlon XP (Thorton)"           },
+	{  6, 10, -1, -1,   -1,   1,    -1,    -1, NC, ATHLON_|_MP_        ,     0, "Athlon MP (Barton)"            },
+	{  6, 10, -1, -1,   -1,   1,    -1,    -1, NC, ATHLON_|_XP_|_M_    ,     0, "Mobile Athlon (Barton)"        },
+	{  6, 10, -1, -1,   -1,   1,    -1,    -1, NC, ATHLON_|_XP_|_M_|_LV_,    0, "Mobile Athlon (Barton)"        },
+	
+	/* K8 Architecture */
+	{ 15, -1, -1, 15,   -1,   1,    -1,    -1, NC, 0                   ,     0, "Unknown K8"                    },
+	{ 15, -1, -1, 16,   -1,   1,    -1,    -1, NC, 0                   ,     0, "Unknown K9"                    },
+	
+	{ 15, -1, -1, 15,   -1,   1,    -1,    -1, NC, 0                   ,     0, "Unknown A64"                   },
+	{ 15, -1, -1, 15,   -1,   1,    -1,    -1, NC, OPTERON_            ,     0, "Opteron"                       },
+	{ 15, -1, -1, 15,   -1,   2,    -1,    -1, NC, OPTERON_|_X2        ,     0, "Opteron (Dual Core)"           },
+	{ 15,  3, -1, 15,   -1,   1,    -1,    -1, NC, OPTERON_            ,     0, "Opteron"                       },
+	{ 15,  3, -1, 15,   -1,   2,    -1,    -1, NC, OPTERON_|_X2        ,     0, "Opteron (Dual Core)"           },
+	{ 15, -1, -1, 15,   -1,   1,   512,    -1, NC, ATHLON_|_64_        ,     0, "Athlon 64 (512K)"              },
+	{ 15, -1, -1, 15,   -1,   1,  1024,    -1, NC, ATHLON_|_64_        ,     0, "Athlon 64 (1024K)"             },
+	{ 15, -1, -1, 15,   -1,   1,    -1,    -1, NC, ATHLON_|_FX         ,     0, "Athlon FX"                     },
+	{ 15, -1, -1, 15,   -1,   1,    -1,    -1, NC, ATHLON_|_64_|_FX    ,     0, "Athlon 64 FX"                  },
+	{ 15,  3, -1, 15,   35,   2,    -1,    -1, NC, ATHLON_|_64_|_FX    ,     0, "Athlon 64 FX X2 (Toledo)"      },
+	{ 15, -1, -1, 15,   -1,   2,   512,    -1, NC, ATHLON_|_64_|_X2    ,     0, "Athlon 64 X2 (512K)"           },
+	{ 15, -1, -1, 15,   -1,   2,  1024,    -1, NC, ATHLON_|_64_|_X2    ,     0, "Athlon 64 X2 (1024K)"          },
+	{ 15, -1, -1, 15,   -1,   1,   512,    -1, NC, TURION_|_64_        ,     0, "Turion 64 (512K)"              },
+	{ 15, -1, -1, 15,   -1,   1,  1024,    -1, NC, TURION_|_64_        ,     0, "Turion 64 (1024K)"             },
+	{ 15, -1, -1, 15,   -1,   2,   512,    -1, NC, TURION_|_X2         ,     0, "Turion 64 X2 (512K)"           },
+	{ 15, -1, -1, 15,   -1,   2,  1024,    -1, NC, TURION_|_X2         ,     0, "Turion 64 X2 (1024K)"          },
+	{ 15, -1, -1, 15,   -1,   1,   128,    -1, NC, SEMPRON_            ,     0, "A64 Sempron (128K)"            },
+	{ 15, -1, -1, 15,   -1,   1,   256,    -1, NC, SEMPRON_            ,     0, "A64 Sempron (256K)"            },
+	{ 15, -1, -1, 15,   -1,   1,   512,    -1, NC, SEMPRON_            ,     0, "A64 Sempron (512K)"            },
+	{ 15, -1, -1, 15, 0x4f,   1,   512,    -1, NC, ATHLON_|_64_        ,     0, "Athlon 64 (Orleans/512K)"      },
+	{ 15, -1, -1, 15, 0x5f,   1,   512,    -1, NC, ATHLON_|_64_        ,     0, "Athlon 64 (Orleans/512K)"      },
+	{ 15, -1, -1, 15, 0x2f,   1,   512,    -1, NC, ATHLON_|_64_        ,     0, "Athlon 64 (Venice/512K)"       },
+	{ 15, -1, -1, 15, 0x2c,   1,   512,    -1, NC, ATHLON_|_64_        ,     0, "Athlon 64 (Venice/512K)"       },
+	{ 15, -1, -1, 15, 0x1f,   1,   512,    -1, NC, ATHLON_|_64_        ,     0, "Athlon 64 (Winchester/512K)"   },
+	{ 15, -1, -1, 15, 0x0c,   1,   512,    -1, NC, ATHLON_|_64_        ,     0, "Athlon 64 (Newcastle/512K)"    },
+	{ 15, -1, -1, 15, 0x27,   1,   512,    -1, NC, ATHLON_|_64_        ,     0, "Athlon 64 (San Diego/512K)"    },
+	{ 15, -1, -1, 15, 0x37,   1,   512,    -1, NC, ATHLON_|_64_        ,     0, "Athlon 64 (San Diego/512K)"    },
+	{ 15, -1, -1, 15, 0x04,   1,   512,    -1, NC, ATHLON_|_64_        ,     0, "Athlon 64 (ClawHammer/512K)"   },
+	
+	{ 15, -1, -1, 15, 0x5f,   1,  1024,    -1, NC, ATHLON_|_64_        ,     0, "Athlon 64 (Orleans/1024K)"     },
+	{ 15, -1, -1, 15, 0x27,   1,  1024,    -1, NC, ATHLON_|_64_        ,     0, "Athlon 64 (San Diego/1024K)"   },
+	{ 15, -1, -1, 15, 0x04,   1,  1024,    -1, NC, ATHLON_|_64_        ,     0, "Athlon 64 (ClawHammer/1024K)"  },
+	
+	{ 15, -1, -1, 15, 0x4b,   2,   256,    -1, NC, SEMPRON_            ,     0, "Athlon 64 X2 (Windsor/256K)"   },
+	
+	{ 15, -1, -1, 15, 0x23,   2,   512,    -1, NC, ATHLON_|_64_|_X2    ,     0, "Athlon 64 X2 (Toledo/512K)"    },
+	{ 15, -1, -1, 15, 0x4b,   2,   512,    -1, NC, ATHLON_|_64_|_X2    ,     0, "Athlon 64 X2 (Windsor/512K)"   },
+	{ 15, -1, -1, 15, 0x43,   2,   512,    -1, NC, ATHLON_|_64_|_X2    ,     0, "Athlon 64 X2 (Windsor/512K)"   },
+	{ 15, -1, -1, 15, 0x6b,   2,   512,    -1, NC, ATHLON_|_64_|_X2    ,     0, "Athlon 64 X2 (Brisbane/512K)"  },
+	{ 15, -1, -1, 15, 0x2b,   2,   512,    -1, NC, ATHLON_|_64_|_X2    ,     0, "Athlon 64 X2 (Manchester/512K)"},
+	
+	{ 15, -1, -1, 15, 0x23,   2,  1024,    -1, NC, ATHLON_|_64_|_X2    ,     0, "Athlon 64 X2 (Toledo/1024K)"   },
+	{ 15, -1, -1, 15, 0x43,   2,  1024,    -1, NC, ATHLON_|_64_|_X2    ,     0, "Athlon 64 X2 (Windsor/1024K)"  },
+	
+	{ 15, -1, -1, 15, 0x08,   1,   128,    -1, NC, MOBILE_|SEMPRON_    ,     0, "Mobile Sempron 64 (Dublin/128K)"},
+	{ 15, -1, -1, 15, 0x08,   1,   256,    -1, NC, MOBILE_|SEMPRON_    ,     0, "Mobile Sempron 64 (Dublin/256K)"},
+	{ 15, -1, -1, 15, 0x0c,   1,   256,    -1, NC, SEMPRON_            ,     0, "Sempron 64 (Paris)"            },
+	{ 15, -1, -1, 15, 0x1c,   1,   128,    -1, NC, SEMPRON_            ,     0, "Sempron 64 (Palermo/128K)"     },
+	{ 15, -1, -1, 15, 0x1c,   1,   256,    -1, NC, SEMPRON_            ,     0, "Sempron 64 (Palermo/256K)"     },
+	{ 15, -1, -1, 15, 0x1c,   1,   128,    -1, NC, MOBILE_| SEMPRON_   ,     0, "Mobile Sempron 64 (Sonora/128K)"},
+	{ 15, -1, -1, 15, 0x1c,   1,   256,    -1, NC, MOBILE_| SEMPRON_   ,     0, "Mobile Sempron 64 (Sonora/256K)"},
+	{ 15, -1, -1, 15, 0x2c,   1,   128,    -1, NC, SEMPRON_            ,     0, "Sempron 64 (Palermo/128K)"     },
+	{ 15, -1, -1, 15, 0x2c,   1,   256,    -1, NC, SEMPRON_            ,     0, "Sempron 64 (Palermo/256K)"     },
+	{ 15, -1, -1, 15, 0x2c,   1,   128,    -1, NC, MOBILE_| SEMPRON_   ,     0, "Mobile Sempron 64 (Albany/128K)"},
+	{ 15, -1, -1, 15, 0x2c,   1,   256,    -1, NC, MOBILE_| SEMPRON_   ,     0, "Mobile Sempron 64 (Albany/256K)"},
+	{ 15, -1, -1, 15, 0x2f,   1,   128,    -1, NC, SEMPRON_            ,     0, "Sempron 64 (Palermo/128K)"     },
+	{ 15, -1, -1, 15, 0x2f,   1,   256,    -1, NC, SEMPRON_            ,     0, "Sempron 64 (Palermo/256K)"     },
+	{ 15, -1, -1, 15, 0x4f,   1,   128,    -1, NC, SEMPRON_            ,     0, "Sempron 64 (Manila/128K)"      },
+	{ 15, -1, -1, 15, 0x4f,   1,   256,    -1, NC, SEMPRON_            ,     0, "Sempron 64 (Manila/256K)"      },
+	{ 15, -1, -1, 15, 0x5f,   1,   128,    -1, NC, SEMPRON_            ,     0, "Sempron 64 (Manila/128K)"      },
+	{ 15, -1, -1, 15, 0x5f,   1,   256,    -1, NC, SEMPRON_            ,     0, "Sempron 64 (Manila/256K)"      },
+	{ 15, -1, -1, 15, 0x6b,   2,   256,    -1, NC, SEMPRON_            ,     0, "Sempron 64 Dual (Sherman/256K)"},
+	{ 15, -1, -1, 15, 0x6b,   2,   512,    -1, NC, SEMPRON_            ,     0, "Sempron 64 Dual (Sherman/512K)"},
+	{ 15, -1, -1, 15, 0x7f,   1,   256,    -1, NC, SEMPRON_            ,     0, "Sempron 64 (Sparta/256K)"      },
+	{ 15, -1, -1, 15, 0x7f,   1,   512,    -1, NC, SEMPRON_            ,     0, "Sempron 64 (Sparta/512K)"      },
+	{ 15, -1, -1, 15, 0x4c,   1,   256,    -1, NC, MOBILE_| SEMPRON_   ,     0, "Mobile Sempron 64 (Keene/256K)"},
+	{ 15, -1, -1, 15, 0x4c,   1,   512,    -1, NC, MOBILE_| SEMPRON_   ,     0, "Mobile Sempron 64 (Keene/512K)"},
+	{ 15, -1, -1, 15,   -1,   2,    -1,    -1, NC, SEMPRON_            ,     0, "Sempron Dual Core"             },
+	
+	{ 15, -1, -1, 15, 0x24,   1,   512,    -1, NC, TURION_|_64_        ,     0, "Turion 64 (Lancaster/512K)"    },
+	{ 15, -1, -1, 15, 0x24,   1,  1024,    -1, NC, TURION_|_64_        ,     0, "Turion 64 (Lancaster/1024K)"   },
+	{ 15, -1, -1, 15, 0x48,   2,   256,    -1, NC, TURION_|_X2         ,     0, "Turion X2 (Taylor)"            },
+	{ 15, -1, -1, 15, 0x48,   2,   512,    -1, NC, TURION_|_X2         ,     0, "Turion X2 (Trinidad)"          },
+	{ 15, -1, -1, 15, 0x4c,   1,   512,    -1, NC, TURION_|_64_        ,     0, "Turion 64 (Richmond)"          },
+	{ 15, -1, -1, 15, 0x68,   2,   256,    -1, NC, TURION_|_X2         ,     0, "Turion X2 (Tyler/256K)"        },
+	{ 15, -1, -1, 15, 0x68,   2,   512,    -1, NC, TURION_|_X2         ,     0, "Turion X2 (Tyler/512K)"        },
+	{ 15, -1, -1, 17,    3,   2,   512,    -1, NC, TURION_|_X2         ,     0, "Turion X2 (Griffin/512K)"      },
+	{ 15, -1, -1, 17,    3,   2,  1024,    -1, NC, TURION_|_X2         ,     0, "Turion X2 (Griffin/1024K)"     },
+
+	/* K10 Architecture (2007) */
+	{ 15, -1, -1, 16,   -1,   1,    -1,    -1, PHENOM, 0               ,     0, "Unknown AMD Phenom"            },
+	{ 15,  2, -1, 16,   -1,   1,    -1,    -1, PHENOM, 0               ,     0, "Phenom"                        },
+	{ 15,  2, -1, 16,   -1,   3,    -1,    -1, PHENOM, 0               ,     0, "Phenom X3 (Toliman)"           },
+	{ 15,  2, -1, 16,   -1,   4,    -1,    -1, PHENOM, 0               ,     0, "Phenom X4 (Agena)"             },
+	{ 15,  2, -1, 16,   -1,   3,   512,    -1, PHENOM, 0               ,     0, "Phenom X3 (Toliman/256K)"      },
+	{ 15,  2, -1, 16,   -1,   3,   512,    -1, PHENOM, 0               ,     0, "Phenom X3 (Toliman/512K)"      },
+	{ 15,  2, -1, 16,   -1,   4,   128,    -1, PHENOM, 0               ,     0, "Phenom X4 (Agena/128K)"        },
+	{ 15,  2, -1, 16,   -1,   4,   256,    -1, PHENOM, 0               ,     0, "Phenom X4 (Agena/256K)"        },
+	{ 15,  2, -1, 16,   -1,   4,   512,    -1, PHENOM,  0              ,     0, "Phenom X4 (Agena/512K)"        },
+	{ 15,  2, -1, 16,   -1,   2,   512,    -1, NC, ATHLON_|_64_|_X2    ,     0, "Athlon X2 (Kuma)"              },
+	/* Phenom II derivates: */
+	{ 15,  4, -1, 16,   -1,   4,    -1,    -1, NC, 0                   ,     0, "Phenom (Deneb-based)"          },
+	{ 15,  4, -1, 16,   -1,   1,  1024,    -1, NC, SEMPRON_            ,     0, "Sempron (Sargas)"              },
+	{ 15,  4, -1, 16,   -1,   2,   512,    -1, PHENOM2, 0              ,     0, "Phenom II X2 (Callisto)"       },
+	{ 15,  4, -1, 16,   -1,   3,   512,    -1, PHENOM2, 0              ,     0, "Phenom II X3 (Heka)"           },
+	{ 15,  4, -1, 16,   -1,   4,   512,    -1, PHENOM2, 0              ,     0, "Phenom II X4"                  },
+	{ 15,  4, -1, 16,    4,   4,   512,    -1, PHENOM2, 0              ,     0, "Phenom II X4 (Deneb)"          },
+	{ 15,  5, -1, 16,    5,   4,   512,    -1, PHENOM2, 0              ,     0, "Phenom II X4 (Deneb)"          },
+	{ 15,  4, -1, 16,   10,   4,   512,    -1, PHENOM2, 0              ,     0, "Phenom II X4 (Zosma)"          },
+	{ 15,  4, -1, 16,   10,   6,   512,    -1, PHENOM2, 0              ,     0, "Phenom II X6 (Thuban)"         },
+	/* Athlon II derivates: */
+	{ 15,  6, -1, 16,    6,   2,   512,    -1, NC, ATHLON_|_X2         ,     0, "Athlon II (Champlain)"         },
+	{ 15,  6, -1, 16,    6,   2,   512,    -1, NC, ATHLON_|_64_|_X2    ,     0, "Athlon II X2 (Regor)"          },
+	{ 15,  6, -1, 16,    6,   2,  1024,    -1, NC, ATHLON_|_64_|_X2    ,     0, "Athlon II X2 (Regor)"          },
+	{ 15,  5, -1, 16,    5,   3,   512,    -1, NC, ATHLON_|_64_|_X3    ,     0, "Athlon II X3 (Rana)"           },
+	{ 15,  5, -1, 16,    5,   4,   512,    -1, NC, ATHLON_|_64_|_X4    ,     0, "Athlon II X4 (Propus)"         },
+	/* Llano APUs (2011): */
+	{ 15,  1, -1, 18,    1,   2,    -1,    -1, FUSION_EA, 0            ,     0, "Llano X2"                      },
+	{ 15,  1, -1, 18,    1,   3,    -1,    -1, FUSION_EA, 0            ,     0, "Llano X3"                      },
+	{ 15,  1, -1, 18,    1,   4,    -1,    -1, FUSION_EA, 0            ,     0, "Llano X4"                      },
+
+	/* Family 14h: Bobcat Architecture (2011) */
+	{ 15,  2, -1, 20,   -1,   1,    -1,    -1, FUSION_C, 0             ,     0, "Brazos Ontario"                },
+	{ 15,  2, -1, 20,   -1,   2,    -1,    -1, FUSION_C, 0             ,     0, "Brazos Ontario (Dual-core)"    },
+	{ 15,  1, -1, 20,   -1,   1,    -1,    -1, FUSION_E, 0             ,     0, "Brazos Zacate"                 },
+	{ 15,  1, -1, 20,   -1,   2,    -1,    -1, FUSION_E, 0             ,     0, "Brazos Zacate (Dual-core)"     },
+	{ 15,  2, -1, 20,   -1,   2,    -1,    -1, FUSION_Z, 0             ,     0, "Brazos Desna (Dual-core)"      },
+
+	/* Family 15h: Bulldozer Architecture (2011) */
+	{ 15, -1, -1, 21,    0,   4,    -1,    -1, NC, 0                   ,     0, "Bulldozer X2"                  },
+	{ 15, -1, -1, 21,    1,   4,    -1,    -1, NC, 0                   ,     0, "Bulldozer X2"                  },
+	{ 15, -1, -1, 21,    1,   6,    -1,    -1, NC, 0                   ,     0, "Bulldozer X3"                  },
+	{ 15, -1, -1, 21,    1,   8,    -1,    -1, NC, 0                   ,     0, "Bulldozer X4"                  },
+	/* 2nd-gen, Piledriver core (2012): */
+	{ 15, -1, -1, 21,    2,   4,    -1,    -1, NC, 0                   ,     0, "Vishera X2"                    },
+	{ 15, -1, -1, 21,    2,   6,    -1,    -1, NC, 0                   ,     0, "Vishera X3"                    },
+	{ 15, -1, -1, 21,    2,   8,    -1,    -1, NC, 0                   ,     0, "Vishera X4"                    },
+	{ 15,  0, -1, 21,   16,   2,    -1,    -1, FUSION_A, 0             ,     0, "Trinity X2"                    },
+	{ 15,  0, -1, 21,   16,   4,    -1,    -1, FUSION_A, 0             ,     0, "Trinity X4"                    },
+	{ 15,  3, -1, 21,   19,   2,    -1,    -1, FUSION_A, 0             ,     0, "Richland X2"                   },
+	{ 15,  3, -1, 21,   19,   4,    -1,    -1, FUSION_A, 0             ,     0, "Richland X4"                   },
+	/* 3rd-gen, Steamroller core (2014): */
+	{ 15,  0, -1, 21,   48,   2,    -1,    -1, FUSION_A, 0             ,     0, "Kaveri X2"                     },
+	{ 15,  0, -1, 21,   48,   4,    -1,    -1, FUSION_A, 0             ,     0, "Kaveri X4"                     },
+	{ 15,  8, -1, 21,   56,   4,    -1,    -1, FUSION_A, 0             ,     0, "Godavari X4"                   },
+	/* 4th-gen, Excavator core (2015): */
+	{ 15,  1, -1, 21,   96,   2,    -1,    -1, FUSION_A, 0             ,     0, "Carrizo X2"                    },
+	{ 15,  1, -1, 21,   96,   4,    -1,    -1, FUSION_A, 0             ,     0, "Carrizo X4"                    },
+	{ 15,  5, -1, 21,  101,   2,    -1,    -1, FUSION_A, 0             ,     0, "Bristol Ridge X2"              },
+	{ 15,  5, -1, 21,  101,   4,    -1,    -1, FUSION_A, 0             ,     0, "Bristol Ridge X4"              },
+	{ 15,  0, -1, 21,  112,   2,    -1,    -1, FUSION_A, 0             ,     0, "Stoney Ridge X2"               },
+	{ 15,  0, -1, 21,  112,   2,    -1,    -1, FUSION_E, 0             ,     0, "Stoney Ridge X2"               },
+
+	/* Family 16h: Jaguar Architecture (2013) */
+	{ 15,  0, -1, 22,    0,   2,    -1,    -1, FUSION_A, 0             ,     0, "Kabini X2"                     },
+	{ 15,  0, -1, 22,    0,   4,    -1,    -1, FUSION_A, 0             ,     0, "Kabini X4"                     },
+	/* 2nd-gen, Puma core (2013): */
+	{ 15,  0, -1, 22,   48,   2,    -1,    -1, FUSION_E, 0             ,     0, "Mullins X2"                    },
+	{ 15,  0, -1, 22,   48,   4,    -1,    -1, FUSION_A, 0             ,     0, "Mullins X4"                    },
+
+	/* Family 17h: Zen Architecture (2017) */
+	{ 15, -1, -1, 23,    1,   8,    -1,    -1, NC, 0                   ,     0, "Ryzen 7"                       },
+	{ 15, -1, -1, 23,    1,   6,    -1,    -1, NC, 0                   , _1600, "Ryzen 5"                       },
+	{ 15, -1, -1, 23,    1,   4,    -1,    -1, NC, 0                   , _1500, "Ryzen 5"                       },
+	{ 15, -1, -1, 23,    1,   4,    -1,    -1, NC, 0                   , _1400, "Ryzen 5"                       },
+	{ 15, -1, -1, 23,    1,   4,    -1,    -1, NC, 0                   ,     0, "Ryzen 3"                       },
+	//{ 15, -1, -1, 23,    1,   4,    -1,    -1, NC, 0                   ,     0, "Raven Ridge"                   }, //TBA
+
+	/* Newer Opterons: */
+	{ 15,  9, -1, 22,    9,   8,    -1,    -1, NC, OPTERON_            ,     0, "Magny-Cours Opteron"           },
+};
+
+
+static void load_amd_features(struct cpu_raw_data_t* raw, struct cpu_id_t* data)
+{
+	const struct feature_map_t matchtable_edx81[] = {
+		{ 20, CPU_FEATURE_NX },
+		{ 22, CPU_FEATURE_MMXEXT },
+		{ 25, CPU_FEATURE_FXSR_OPT },
+		{ 30, CPU_FEATURE_3DNOWEXT },
+		{ 31, CPU_FEATURE_3DNOW },
+	};
+	const struct feature_map_t matchtable_ecx81[] = {
+		{  1, CPU_FEATURE_CMP_LEGACY },
+		{  2, CPU_FEATURE_SVM },
+		{  5, CPU_FEATURE_ABM },
+		{  6, CPU_FEATURE_SSE4A },
+		{  7, CPU_FEATURE_MISALIGNSSE },
+		{  8, CPU_FEATURE_3DNOWPREFETCH },
+		{  9, CPU_FEATURE_OSVW },
+		{ 10, CPU_FEATURE_IBS },
+		{ 11, CPU_FEATURE_XOP },
+		{ 12, CPU_FEATURE_SKINIT },
+		{ 13, CPU_FEATURE_WDT },
+		{ 16, CPU_FEATURE_FMA4 },
+		{ 21, CPU_FEATURE_TBM },
+	};
+	const struct feature_map_t matchtable_edx87[] = {
+		{  0, CPU_FEATURE_TS },
+		{  1, CPU_FEATURE_FID },
+		{  2, CPU_FEATURE_VID },
+		{  3, CPU_FEATURE_TTP },
+		{  4, CPU_FEATURE_TM_AMD },
+		{  5, CPU_FEATURE_STC },
+		{  6, CPU_FEATURE_100MHZSTEPS },
+		{  7, CPU_FEATURE_HWPSTATE },
+		/* id 8 is handled in common */
+		{  9, CPU_FEATURE_CPB },
+		{ 10, CPU_FEATURE_APERFMPERF },
+		{ 11, CPU_FEATURE_PFI },
+		{ 12, CPU_FEATURE_PA },
+	};
+	if (raw->ext_cpuid[0][0] >= 0x80000001) {
+		match_features(matchtable_edx81, COUNT_OF(matchtable_edx81), raw->ext_cpuid[1][3], data);
+		match_features(matchtable_ecx81, COUNT_OF(matchtable_ecx81), raw->ext_cpuid[1][2], data);
+	}
+	if (raw->ext_cpuid[0][0] >= 0x80000007)
+		match_features(matchtable_edx87, COUNT_OF(matchtable_edx87), raw->ext_cpuid[7][3], data);
+	if (raw->ext_cpuid[0][0] >= 0x8000001a) {
+		/* We have the extended info about SSE unit size */
+		data->detection_hints[CPU_HINT_SSE_SIZE_AUTH] = 1;
+		data->sse_size = (raw->ext_cpuid[0x1a][0] & 1) ? 128 : 64;
+	}
+}
+
+static void decode_amd_cache_info(struct cpu_raw_data_t* raw, struct cpu_id_t* data)
+{
+	int l3_result;
+	const int assoc_table[16] = {
+		0, 1, 2, 0, 4, 0, 8, 0, 16, 0, 32, 48, 64, 96, 128, 255
+	};
+	unsigned n = raw->ext_cpuid[0][0];
+	
+	if (n >= 0x80000005) {
+		data->l1_data_cache = (raw->ext_cpuid[5][2] >> 24) & 0xff;
+		data->l1_assoc = (raw->ext_cpuid[5][2] >> 16) & 0xff;
+		data->l1_cacheline = (raw->ext_cpuid[5][2]) & 0xff;
+		data->l1_instruction_cache = (raw->ext_cpuid[5][3] >> 24) & 0xff;
+	}
+	if (n >= 0x80000006) {
+		data->l2_cache = (raw->ext_cpuid[6][2] >> 16) & 0xffff;
+		data->l2_assoc = assoc_table[(raw->ext_cpuid[6][2] >> 12) & 0xf];
+		data->l2_cacheline = (raw->ext_cpuid[6][2]) & 0xff;
+		
+		l3_result = (raw->ext_cpuid[6][3] >> 18);
+		if (l3_result > 0) {
+			l3_result = 512 * l3_result; /* AMD spec says it's a range,
+			                                but we take the lower bound */
+			data->l3_cache = l3_result;
+			data->l3_assoc = assoc_table[(raw->ext_cpuid[6][3] >> 12) & 0xf];
+			data->l3_cacheline = (raw->ext_cpuid[6][3]) & 0xff;
+		} else {
+			data->l3_cache = 0;
+		}
+	}
+}
+
+static void decode_amd_number_of_cores(struct cpu_raw_data_t* raw, struct cpu_id_t* data)
+{
+	int logical_cpus = -1, num_cores = -1;
+	
+	if (raw->basic_cpuid[0][0] >= 1) {
+		logical_cpus = (raw->basic_cpuid[1][1] >> 16) & 0xff;
+		if (raw->ext_cpuid[0][0] >= 8) {
+			num_cores = 1 + (raw->ext_cpuid[8][2] & 0xff);
+		}
+	}
+	if (data->flags[CPU_FEATURE_HT]) {
+		if (num_cores > 1) {
+			if (data->ext_family >= 23)
+				num_cores /= 2; // e.g., Ryzen 7 reports 16 "real" cores, but they are really just 8.
+			data->num_cores = num_cores;
+			data->num_logical_cpus = logical_cpus;
+		} else {
+			data->num_cores = 1;
+			data->num_logical_cpus = (logical_cpus >= 2 ? logical_cpus : 2);
+		}
+	} else {
+		data->num_cores = data->num_logical_cpus = 1;
+	}
+}
+
+static int amd_has_turion_modelname(const char *bs)
+{
+	/* We search for something like TL-60. Ahh, I miss regexes...*/
+	int i, l, k;
+	char code[3] = {0};
+	const char* codes[] = { "ML", "MT", "MK", "TK", "TL", "RM", "ZM", "" };
+	l = (int) strlen(bs);
+	for (i = 3; i < l - 2; i++) {
+		if (bs[i] == '-' &&
+		    isupper(bs[i-1]) && isupper(bs[i-2]) && !isupper(bs[i-3]) &&
+		    isdigit(bs[i+1]) && isdigit(bs[i+2]) && !isdigit(bs[i+3]))
+		{
+			code[0] = bs[i-2];
+			code[1] = bs[i-1];
+			for (k = 0; codes[k][0]; k++)
+				if (!strcmp(codes[k], code)) return 1;
+		}
+	}
+	return 0;
+}
+
+static struct amd_code_and_bits_t decode_amd_codename_part1(const char *bs)
+{
+	amd_code_t code = NC;
+	uint64_t bits = 0;
+	struct amd_code_and_bits_t result;
+
+	if (strstr(bs, "Dual Core") ||
+	    strstr(bs, "Dual-Core") ||
+	    strstr(bs, " X2 "))
+		bits |= _X2;
+	if (strstr(bs, " X4 ")) bits |= _X4;
+	if (strstr(bs, " X3 ")) bits |= _X3;
+	if (strstr(bs, "Opteron")) bits |= OPTERON_;
+	if (strstr(bs, "Phenom")) {
+		code = (strstr(bs, "II")) ? PHENOM2 : PHENOM;
+	}
+	if (amd_has_turion_modelname(bs)) {
+		bits |= TURION_;
+	}
+	if (strstr(bs, "Athlon(tm)")) bits |= ATHLON_;
+	if (strstr(bs, "Sempron(tm)")) bits |= SEMPRON_;
+	if (strstr(bs, "Duron")) bits |= DURON_;
+	if (strstr(bs, " 64 ")) bits |= _64_;
+	if (strstr(bs, " FX")) bits |= _FX;
+	if (strstr(bs, " MP")) bits |= _MP_;
+	if (strstr(bs, "Athlon(tm) 64") || strstr(bs, "Athlon(tm) II X") || match_pattern(bs, "Athlon(tm) X#")) {
+		bits |= ATHLON_ | _64_;
+	}
+	if (strstr(bs, "Turion")) bits |= TURION_;
+	
+	if (strstr(bs, "mobile") || strstr(bs, "Mobile")) {
+		bits |= MOBILE_;
+	}
+	
+	if (strstr(bs, "XP")) bits |= _XP_;
+	if (strstr(bs, "XP-M")) bits |= _M_;
+	if (strstr(bs, "(LV)")) bits |= _LV_;
+
+	if (match_pattern(bs, "C-##")) code = FUSION_C;
+	if (match_pattern(bs, "E-###")) code = FUSION_E;
+	if (match_pattern(bs, "Z-##")) code = FUSION_Z;
+	if (match_pattern(bs, "E#-####") || match_pattern(bs, "A#-####")) code = FUSION_EA;
+
+	result.code = code;
+	result.bits = bits;
+	return result;
+}
+
+static int decode_amd_ryzen_model_code(const char* bs)
+{
+	const struct {
+		int model_code;
+		const char* match_str;
+	} patterns[] = {
+		{ _1600, "1600" },
+		{ _1500, "1500" },
+		{ _1400, "1400" },
+	};
+	int i;
+
+	for (i = 0; i < COUNT_OF(patterns); i++)
+		if (strstr(bs, patterns[i].match_str))
+			return patterns[i].model_code;
+	//
+	return 0;
+}
+
+static void decode_amd_codename(struct cpu_raw_data_t* raw, struct cpu_id_t* data, struct internal_id_info_t* internal)
+{
+	struct amd_code_and_bits_t code_and_bits = decode_amd_codename_part1(data->brand_str);
+	int i = 0;
+	char* code_str = NULL;
+	int model_code;
+
+	for (i = 0; i < COUNT_OF(amd_code_str); i++) {
+		if (code_and_bits.code == amd_code_str[i].code) {
+			code_str = amd_code_str[i].str;
+			break;
+		}
+	}
+	if (/*code == ATHLON_64_X2*/ match_all(code_and_bits.bits, ATHLON_|_64_|_X2) && data->l2_cache < 512) {
+		code_and_bits.bits &= ~(ATHLON_ | _64_);
+		code_and_bits.bits |= SEMPRON_;
+	}
+	if (code_str)
+		debugf(2, "Detected AMD brand code: %d (%s)\n", code_and_bits.code, code_str);
+	else
+		debugf(2, "Detected AMD brand code: %d\n", code_and_bits.code);
+
+	if (code_and_bits.bits) {
+		debugf(2, "Detected AMD bits: ");
+		debug_print_lbits(2, code_and_bits.bits);
+	}
+	// is it Ryzen? if so, we need to detect discern between the four-core 1400/1500 (Ryzen 5) and the four-core Ryzen 3:
+	model_code = (data->ext_family == 23) ? decode_amd_ryzen_model_code(data->brand_str) : 0;
+
+	internal->code.amd = code_and_bits.code;
+	internal->bits = code_and_bits.bits;
+	internal->score = match_cpu_codename(cpudb_amd, COUNT_OF(cpudb_amd), data, code_and_bits.code,
+	                                     code_and_bits.bits, model_code);
+}
+
+int cpuid_identify_amd(struct cpu_raw_data_t* raw, struct cpu_id_t* data, struct internal_id_info_t* internal)
+{
+	load_amd_features(raw, data);
+	decode_amd_cache_info(raw, data);
+	decode_amd_number_of_cores(raw, data);
+	decode_amd_codename(raw, data, internal);
+	return 0;
+}
+
+void cpuid_get_list_amd(struct cpu_list_t* list)
+{
+	generic_get_cpu_list(cpudb_amd, COUNT_OF(cpudb_amd), list);
+}

compat/libcpuid/recog_amd.h

@@ -27,5 +27,6 @@
 #define __RECOG_AMD_H__
 
 int cpuid_identify_amd(struct cpu_raw_data_t* raw, struct cpu_id_t* data, struct internal_id_info_t* internal);
+void cpuid_get_list_amd(struct cpu_list_t* list);
 
 #endif /* __RECOG_AMD_H__ */

compat/libcpuid/recog_intel.c

@@ -0,0 +1,935 @@
+/*
+ * Copyright 2008  Veselin Georgiev,
+ * anrieffNOSPAM @ mgail_DOT.com (convert to gmail)
+ *
+ * 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.
+ *
+ * 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.
+ */
+#include <string.h>
+#include <ctype.h>
+#include "libcpuid.h"
+#include "libcpuid_util.h"
+#include "libcpuid_internal.h"
+#include "recog_intel.h"
+
+const struct intel_bcode_str { intel_code_t code; char *str; } intel_bcode_str[] = {
+	#define CODE(x) { x, #x }
+	#define CODE2(x, y) CODE(x)
+	#include "intel_code_t.h"
+	#undef CODE
+};
+
+typedef struct {
+	int code;
+	uint64_t bits;
+} intel_code_and_bits_t;
+
+enum _intel_model_t {
+	UNKNOWN = -1,
+	_3000 = 100,
+	_3100,
+	_3200,
+	X3200,
+	_3300,
+	X3300,
+	_5100,
+	_5200,
+	_5300,
+	_5400,
+	_2xxx, /* Core i[357] 2xxx */
+	_3xxx, /* Core i[357] 3xxx */
+};
+typedef enum _intel_model_t intel_model_t;
+
+enum _intel_bits_t {
+	PENTIUM_                = LBIT(  0 ),
+	CELERON_                = LBIT(  1 ),
+	MOBILE_                 = LBIT(  2 ),
+	CORE_                   = LBIT(  3 ),
+	_I_                     = LBIT(  4 ),
+	_M_                     = LBIT(  5 ),
+	_3                      = LBIT(  6 ),
+	_5                      = LBIT(  7 ),
+	_7                      = LBIT(  8 ),
+	XEON_                   = LBIT(  9 ),
+	_MP                     = LBIT( 10 ),
+	ATOM_                   = LBIT( 11 ),
+	
+};
+typedef enum _intel_bits_t intel_bits_t;
+
+const struct match_entry_t cpudb_intel[] = {
+	{ -1, -1, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Unknown Intel CPU"       },
+	
+	/* i486 */
+	{  4, -1, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Unknown i486"            },
+	{  4,  0, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "i486 DX-25/33"           },
+	{  4,  1, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "i486 DX-50"              },
+	{  4,  2, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "i486 SX"                 },
+	{  4,  3, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "i486 DX2"                },
+	{  4,  4, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "i486 SL"                 },
+	{  4,  5, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "i486 SX2"                },
+	{  4,  7, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "i486 DX2 WriteBack"      },
+	{  4,  8, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "i486 DX4"                },
+	{  4,  9, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "i486 DX4 WriteBack"      },
+	
+	/* All Pentia:
+	   Pentium 1 */
+	{  5, -1, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Unknown Pentium"         },
+	{  5,  0, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Pentium A-Step"          },
+	{  5,  1, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Pentium 1 (0.8u)"        },
+	{  5,  2, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Pentium 1 (0.35u)"       },
+	{  5,  3, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Pentium OverDrive"       },
+	{  5,  4, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Pentium 1 (0.35u)"       },
+	{  5,  7, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Pentium 1 (0.35u)"       },
+	{  5,  8, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Pentium MMX (0.25u)"     },
+	
+	/* Pentium 2 / 3 / M / Conroe / whatsnext - all P6 based. */
+	{  6, -1, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Unknown P6"              },
+	{  6,  0, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Pentium Pro"             },
+	{  6,  1, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Pentium Pro"             },
+	{  6,  3, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Pentium II (Klamath)"    },
+	{  6,  5, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Pentium II (Deschutes)"  },
+	{  6,  5, -1, -1, -1,   1,    -1,    -1, NC, MOBILE_|PENTIUM_,   0, "Mobile Pentium II (Tonga)"},
+	{  6,  6, -1, -1, -1,   1,    -1,    -1, NC,0              ,     0, "Pentium II (Dixon)"      },
+	
+	{  6,  3, -1, -1, -1,   1,    -1,    -1, NC, XEON_         ,     0, "P-II Xeon (Klamath)"     },
+	{  6,  5, -1, -1, -1,   1,    -1,    -1, NC, XEON_         ,     0, "P-II Xeon (Drake)"       },
+	{  6,  6, -1, -1, -1,   1,    -1,    -1, NC, XEON_         ,     0, "P-II Xeon (Dixon)"       },
+		
+	{  6,  5, -1, -1, -1,   1,    -1,    -1, NC, CELERON_      ,     0, "P-II Celeron (Covington)" },
+	{  6,  6, -1, -1, -1,   1,    -1,    -1, NC, CELERON_      ,     0, "P-II Celeron (Mendocino)" },
+	
+	/* -------------------------------------------------- */
+	
+	{  6,  7, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Pentium III (Katmai)"    },
+	{  6,  8, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Pentium III (Coppermine)"},
+	{  6, 10, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Pentium III (Coppermine)"},
+	{  6, 11, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Pentium III (Tualatin)"  },
+	
+	{  6,  7, -1, -1, -1,   1,    -1,    -1, NC, XEON_         ,     0, "P-III Xeon (Tanner)"     },
+	{  6,  8, -1, -1, -1,   1,    -1,    -1, NC, XEON_         ,     0, "P-III Xeon (Cascades)"   },
+	{  6, 10, -1, -1, -1,   1,    -1,    -1, NC, XEON_         ,     0, "P-III Xeon (Cascades)"   },
+	{  6, 11, -1, -1, -1,   1,    -1,    -1, NC, XEON_         ,     0, "P-III Xeon (Tualatin)"   },
+	
+	{  6,  7, -1, -1, -1,   1,    -1,    -1, NC, CELERON_      ,     0, "P-III Celeron (Katmai)"     },
+	{  6,  8, -1, -1, -1,   1,    -1,    -1, NC, CELERON_      ,     0, "P-III Celeron (Coppermine)" },
+	{  6, 10, -1, -1, -1,   1,    -1,    -1, NC, CELERON_      ,     0, "P-III Celeron (Coppermine)" },
+	{  6, 11, -1, -1, -1,   1,    -1,    -1, NC, CELERON_      ,     0, "P-III Celeron (Tualatin)"   },
+	
+	/* Netburst based (Pentium 4 and later)
+	   classic P4s */
+	{ 15, -1, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Unknown Pentium 4"       },
+	{ 15, -1, -1, 15, -1,   1,    -1,    -1, NC, CELERON_      ,     0, "Unknown P-4 Celeron"     },
+	{ 15, -1, -1, 15, -1,   1,    -1,    -1, NC, XEON_         ,     0, "Unknown Xeon"            },
+	
+	{ 15,  0, -1, 15, -1,   1,    -1,    -1, NC, PENTIUM_      ,     0, "Pentium 4 (Willamette)"  },
+	{ 15,  1, -1, 15, -1,   1,    -1,    -1, NC, PENTIUM_      ,     0, "Pentium 4 (Willamette)"  },
+	{ 15,  2, -1, 15, -1,   1,    -1,    -1, NC, PENTIUM_      ,     0, "Pentium 4 (Northwood)"   },
+	{ 15,  3, -1, 15, -1,   1,    -1,    -1, NC, PENTIUM_      ,     0, "Pentium 4 (Prescott)"    },
+	{ 15,  4, -1, 15, -1,   1,    -1,    -1, NC, PENTIUM_      ,     0, "Pentium 4 (Prescott)"    },
+	{ 15,  6, -1, 15, -1,   1,    -1,    -1, NC, PENTIUM_      ,     0, "Pentium 4 (Cedar Mill)"  },
+	{ 15,  0, -1, 15, -1,   1,    -1,    -1, NC, MOBILE_|PENTIUM_,   0, "Mobile P-4 (Willamette)" },
+	{ 15,  1, -1, 15, -1,   1,    -1,    -1, NC, MOBILE_|PENTIUM_,   0, "Mobile P-4 (Willamette)" },
+	{ 15,  2, -1, 15, -1,   1,    -1,    -1, NC, MOBILE_|PENTIUM_,   0, "Mobile P-4 (Northwood)"  },
+	{ 15,  3, -1, 15, -1,   1,    -1,    -1, NC, MOBILE_|PENTIUM_,   0, "Mobile P-4 (Prescott)"   },
+	{ 15,  4, -1, 15, -1,   1,    -1,    -1, NC, MOBILE_|PENTIUM_,   0, "Mobile P-4 (Prescott)"   },
+	{ 15,  6, -1, 15, -1,   1,    -1,    -1, NC, MOBILE_|PENTIUM_,   0, "Mobile P-4 (Cedar Mill)" },
+	
+	/* server CPUs */
+	{ 15,  0, -1, 15, -1,   1,    -1,    -1, NC, XEON_         ,     0, "Xeon (Foster)"           },
+	{ 15,  1, -1, 15, -1,   1,    -1,    -1, NC, XEON_         ,     0, "Xeon (Foster)"           },
+	{ 15,  2, -1, 15, -1,   1,    -1,    -1, NC, XEON_         ,     0, "Xeon (Prestonia)"        },
+	{ 15,  2, -1, 15, -1,   1,    -1,    -1, NC, XEON_|_MP     ,     0, "Xeon (Gallatin)"         },
+	{ 15,  3, -1, 15, -1,   1,    -1,    -1, NC, XEON_         ,     0, "Xeon (Nocona)"           },
+	{ 15,  4, -1, 15, -1,   1,    -1,    -1, NC, XEON_         ,     0, "Xeon (Nocona)"           },
+	{ 15,  4, -1, 15, -1,   1,    -1,    -1, IRWIN, XEON_      ,     0, "Xeon (Irwindale)"        },
+	{ 15,  4, -1, 15, -1,   1,    -1,    -1, NC, XEON_|_MP     ,     0, "Xeon (Cranford)"         },
+	{ 15,  4, -1, 15, -1,   1,    -1,    -1, POTOMAC, XEON_    ,     0, "Xeon (Potomac)"          },
+	{ 15,  6, -1, 15, -1,   1,    -1,    -1, NC, XEON_         ,     0, "Xeon (Dempsey)"          },
+	
+	/* Pentium Ds */
+	{ 15,  4,  4, 15, -1,   1,    -1,    -1, NC, 0             ,     0, "Pentium D (SmithField)"  },
+	{ 15,  4, -1, 15, -1,   1,    -1,    -1, PENTIUM_D, 0      ,     0, "Pentium D (SmithField)"  },
+	{ 15,  4,  7, 15, -1,   1,    -1,    -1, NC, 0             ,     0, "Pentium D (SmithField)"  },
+	{ 15,  6, -1, 15, -1,   1,    -1,    -1, PENTIUM_D, 0      ,     0, "Pentium D (Presler)"     },
+
+	/* Celeron and Celeron Ds */
+	{ 15,  1, -1, 15, -1,   1,    -1,    -1, NC, CELERON_      ,     0, "P-4 Celeron (Willamette)"   },
+	{ 15,  2, -1, 15, -1,   1,    -1,    -1, NC, CELERON_      ,     0, "P-4 Celeron (Northwood)"    },
+	{ 15,  3, -1, 15, -1,   1,    -1,    -1, NC, CELERON_      ,     0, "P-4 Celeron D (Prescott)"   },
+	{ 15,  4, -1, 15, -1,   1,    -1,    -1, NC, CELERON_      ,     0, "P-4 Celeron D (Prescott)"   },
+	{ 15,  6, -1, 15, -1,   1,    -1,    -1, NC, CELERON_      ,     0, "P-4 Celeron D (Cedar Mill)" },
+	
+	/* -------------------------------------------------- */
+	/* Intel Core microarchitecture - P6-based */
+	
+	{  6,  9, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Unknown Pentium M"          },
+	{  6,  9, -1, -1, -1,   1,    -1,    -1, PENTIUM_M, 0      ,     0, "Unknown Pentium M"          },
+	{  6,  9, -1, -1, -1,   1,    -1,    -1, NC, PENTIUM_      ,     0, "Pentium M (Banias)"         },
+	{  6,  9, -1, -1, -1,   1,    -1,    -1, PENTIUM_M, 0      ,     0, "Pentium M (Banias)"         },
+	{  6,  9, -1, -1, -1,   1,    -1,    -1, NC, CELERON_      ,     0, "Celeron M"                  },
+	{  6, 13, -1, -1, -1,   1,    -1,    -1, NC, PENTIUM_      ,     0, "Pentium M (Dothan)"         },
+	{  6, 13, -1, -1, -1,   1,    -1,    -1, PENTIUM_M, 0      ,     0, "Pentium M (Dothan)"         },
+	{  6, 13, -1, -1, -1,   1,    -1,    -1, NC, CELERON_      ,     0, "Celeron M"                  },
+	
+	{  6, 12, -1, -1, -1,  -1,    -1,    -1, NC, ATOM_         ,     0, "Unknown Atom"               },
+	{  6, 12, -1, -1, -1,  -1,    -1,    -1, DIAMONDVILLE,ATOM_,     0, "Atom (Diamondville)"        },
+	{  6, 12, -1, -1, -1,  -1,    -1,    -1, SILVERTHORNE,ATOM_,     0, "Atom (Silverthorne)"        },
+	{  6, 12, -1, -1, -1,  -1,    -1,    -1, CEDARVIEW, ATOM_  ,     0, "Atom (Cedarview)"           },
+	{  6,  6, -1, -1, -1,  -1,    -1,    -1, CEDARVIEW, ATOM_  ,     0, "Atom (Cedarview)"           },
+	{  6, 12, -1, -1, -1,  -1,    -1,    -1, PINEVIEW, ATOM_   ,     0, "Atom (Pineview)"            },
+	
+	/* -------------------------------------------------- */
+	
+	{  6, 14, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Unknown Yonah"             },
+	{  6, 14, -1, -1, -1,   1,    -1,    -1, CORE_SOLO, 0      ,     0, "Yonah (Core Solo)"         },
+	{  6, 14, -1, -1, -1,   2,    -1,    -1, CORE_DUO, 0       ,     0, "Yonah (Core Duo)"          },
+	{  6, 14, -1, -1, -1,   1,    -1,    -1, CORE_SOLO, MOBILE_,     0, "Yonah (Core Solo)"         },
+	{  6, 14, -1, -1, -1,   2,    -1,    -1, CORE_DUO , MOBILE_,     0, "Yonah (Core Duo)"          },
+	{  6, 14, -1, -1, -1,   1,    -1,    -1, CORE_SOLO, 0      ,     0, "Yonah (Core Solo)"         },
+	
+	{  6, 15, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Unknown Core 2"            },
+	{  6, 15, -1, -1, -1,   2,  4096,    -1, CORE_DUO, 0       ,     0, "Conroe (Core 2 Duo)"       },
+	{  6, 15, -1, -1, -1,   2,  1024,    -1, CORE_DUO, 0       ,     0, "Conroe (Core 2 Duo) 1024K" },
+	{  6, 15, -1, -1, -1,   2,   512,    -1, CORE_DUO, 0       ,     0, "Conroe (Core 2 Duo) 512K"  },
+	{  6, 15, -1, -1, -1,   4,    -1,    -1, QUAD_CORE, 0      ,     0, "Kentsfield (Core 2 Quad)"  },
+	{  6, 15, -1, -1, -1,   4,  4096,    -1, QUAD_CORE, 0      ,     0, "Kentsfield (Core 2 Quad)"  },
+	{  6, 15, -1, -1, -1, 400,    -1,    -1, MORE_THAN_QUADCORE, 0,  0, "More than quad-core"       },
+	{  6, 15, -1, -1, -1,   2,  2048,    -1, CORE_DUO, 0       ,     0, "Allendale (Core 2 Duo)"    },
+	{  6, 15, -1, -1, -1,   2,    -1,    -1, MOBILE_CORE_DUO, 0,     0, "Merom (Core 2 Duo)"        },
+	{  6, 15, -1, -1, -1,   2,  2048,    -1, MEROM, 0          ,     0, "Merom (Core 2 Duo) 2048K"  },
+	{  6, 15, -1, -1, -1,   2,  4096,    -1, MEROM, 0          ,     0, "Merom (Core 2 Duo) 4096K"  },
+	
+	{  6, 15, -1, -1, 15,   1,    -1,    -1, NC, CELERON_      ,     0, "Conroe-L (Celeron)"        },
+	{  6,  6, -1, -1, 22,   1,    -1,    -1, NC, CELERON_      ,     0, "Conroe-L (Celeron)"        },
+	{  6, 15, -1, -1, 15,   2,    -1,    -1, NC, CELERON_      ,     0, "Conroe-L (Allendale)"      },
+	{  6,  6, -1, -1, 22,   2,    -1,    -1, NC, CELERON_      ,     0, "Conroe-L (Allendale)"      },
+	
+	
+	{  6,  6, -1, -1, 22,   1,    -1,    -1, NC, 0             ,     0, "Unknown Core ?"           },
+	{  6,  7, -1, -1, 23,   1,    -1,    -1, NC, 0             ,     0, "Unknown Core ?"           },
+	{  6,  6, -1, -1, 22, 400,    -1,    -1, MORE_THAN_QUADCORE, 0,  0, "More than quad-core"      },
+	{  6,  7, -1, -1, 23, 400,    -1,    -1, MORE_THAN_QUADCORE, 0,  0, "More than quad-core"      },
+	
+	{  6,  7, -1, -1, 23,   1,    -1,    -1, CORE_SOLO         , 0,  0, "Unknown Core 45nm"        },
+	{  6,  7, -1, -1, 23,   1,    -1,    -1, CORE_DUO          , 0,  0, "Unknown Core 45nm"        },
+	{  6,  7, -1, -1, 23,   2,  1024,    -1, WOLFDALE          , 0,  0, "Celeron Wolfdale 1M"      },
+	{  6,  7, -1, -1, 23,   2,  2048,    -1, WOLFDALE          , 0,  0, "Wolfdale (Core 2 Duo) 2M" },
+	{  6,  7, -1, -1, 23,   2,  3072,    -1, WOLFDALE          , 0,  0, "Wolfdale (Core 2 Duo) 3M" },
+	{  6,  7, -1, -1, 23,   2,  6144,    -1, WOLFDALE          , 0,  0, "Wolfdale (Core 2 Duo) 6M" },
+	{  6,  7, -1, -1, 23,   1,    -1,    -1, MOBILE_CORE_DUO   , 0,  0, "Penryn (Core 2 Duo)"      },
+	{  6,  7, -1, -1, 23,   2,  1024,    -1, PENRYN            , 0,  0, "Penryn (Core 2 Duo)"      },
+	{  6,  7, -1, -1, 23,   2,  3072,    -1, PENRYN            , 0,  0, "Penryn (Core 2 Duo) 3M"   },
+	{  6,  7, -1, -1, 23,   2,  6144,    -1, PENRYN            , 0,  0, "Penryn (Core 2 Duo) 6M"   },
+	{  6,  7, -1, -1, 23,   4,  2048,    -1, NC                , 0,  0, "Yorkfield (Core 2 Quad) 2M"},
+	{  6,  7, -1, -1, 23,   4,  3072,    -1, NC                , 0,  0, "Yorkfield (Core 2 Quad) 3M"},
+	{  6,  7, -1, -1, 23,   4,  6144,    -1, NC                , 0,  0, "Yorkfield (Core 2 Quad) 6M"},
+	
+	/* Core microarchitecture-based Xeons: */
+	{  6, 14, -1, -1, 14,   1,    -1,    -1, NC, XEON_         ,     0, "Xeon LV"                  },
+	{  6, 15, -1, -1, 15,   2,  4096,    -1, NC, XEON_         , _5100, "Xeon (Woodcrest)"         },
+	{  6, 15, -1, -1, 15,   2,  2048,    -1, NC, XEON_         , _3000, "Xeon (Conroe/2M)"         },
+	{  6, 15, -1, -1, 15,   2,  4096,    -1, NC, XEON_         , _3000, "Xeon (Conroe/4M)"         },
+	{  6, 15, -1, -1, 15,   4,  4096,    -1, NC, XEON_         , X3200, "Xeon (Kentsfield)"        },
+	{  6, 15, -1, -1, 15,   4,  4096,    -1, NC, XEON_         , _5300, "Xeon (Clovertown)"        },
+	{  6,  7, -1, -1, 23,   2,  6144,    -1, NC, XEON_         , _3100, "Xeon (Wolfdale)"          },
+	{  6,  7, -1, -1, 23,   2,  6144,    -1, NC, XEON_         , _5200, "Xeon (Wolfdale DP)"       },
+	{  6,  7, -1, -1, 23,   4,  6144,    -1, NC, XEON_         , _5400, "Xeon (Harpertown)"        },
+	{  6,  7, -1, -1, 23,   4,  3072,    -1, NC, XEON_         , X3300, "Xeon (Yorkfield/3M)"      },
+	{  6,  7, -1, -1, 23,   4,  6144,    -1, NC, XEON_         , X3300, "Xeon (Yorkfield/6M)"      },
+
+	/* Nehalem CPUs (45nm): */
+	{  6, 10, -1, -1, 26,   4,    -1,    -1, GAINESTOWN, XEON_ ,     0, "Gainestown (Xeon)"        },
+	{  6, 10, -1, -1, 26,   4,    -1,  4096, GAINESTOWN, XEON_ ,     0, "Gainestown 4M (Xeon)"     },
+	{  6, 10, -1, -1, 26,   4,    -1,  8192, GAINESTOWN, XEON_ ,     0, "Gainestown 8M (Xeon)"     },
+	{  6, 10, -1, -1, 26,   4,    -1,    -1, NC, XEON_|_7      ,     0, "Bloomfield (Xeon)"        },
+	{  6, 10, -1, -1, 26,   4,    -1,    -1, NC, CORE_|_I_|_7  ,     0, "Bloomfield (Core i7)"     },
+	{  6, 10, -1, -1, 30,   4,    -1,    -1, NC, CORE_|_I_|_7  ,     0, "Lynnfield (Core i7)"      },
+	{  6,  5, -1, -1, 37,   4,    -1,  8192, NC, CORE_|_I_|_5  ,     0, "Lynnfield (Core i5)"      },
+
+	/* Westmere CPUs (32nm): */
+	{  6,  5, -1, -1, 37,   2,    -1,    -1, NC, 0             ,     0, "Unknown Core i3/i5"       },
+	{  6, 12, -1, -1, 44,  -1,    -1,    -1, WESTMERE, XEON_   ,     0, "Westmere (Xeon)"          },
+	{  6, 12, -1, -1, 44,  -1,    -1, 12288, WESTMERE, XEON_   ,     0, "Gulftown (Xeon)"          },
+	{  6, 12, -1, -1, 44,   4,    -1, 12288, NC, CORE_|_I_|_7  ,     0, "Gulftown (Core i7)"       },
+	{  6,  5, -1, -1, 37,   2,    -1,  4096, NC, CORE_|_I_|_5  ,     0, "Clarkdale (Core i5)"      },
+	{  6,  5, -1, -1, 37,   2,    -1,  4096, NC, CORE_|_I_|_3  ,     0, "Clarkdale (Core i3)"      },
+	{  6,  5, -1, -1, 37,   2,    -1,    -1, NC, PENTIUM_      ,     0, "Arrandale"                },
+	{  6,  5, -1, -1, 37,   2,    -1,  4096, NC, CORE_|_I_|_7  ,     0, "Arrandale (Core i7)"      },
+	{  6,  5, -1, -1, 37,   2,    -1,  3072, NC, CORE_|_I_|_5  ,     0, "Arrandale (Core i5)"      },
+	{  6,  5, -1, -1, 37,   2,    -1,  3072, NC, CORE_|_I_|_3  ,     0, "Arrandale (Core i3)"      },
+
+	/* Sandy Bridge CPUs (32nm): */
+	{  6, 10, -1, -1, 42,  -1,    -1,    -1, NC, 0             ,     0, "Unknown Sandy Bridge"     },
+	{  6, 10, -1, -1, 42,  -1,    -1,    -1, NC, XEON_         ,     0, "Sandy Bridge (Xeon)"      },
+	{  6, 10, -1, -1, 42,  -1,    -1,    -1, NC, CORE_|_I_|_7  ,     0, "Sandy Bridge (Core i7)"   },
+	{  6, 10, -1, -1, 42,   4,    -1,    -1, NC, CORE_|_I_|_7  ,     0, "Sandy Bridge (Core i7)"   },
+	{  6, 10, -1, -1, 42,   4,    -1,    -1, NC, CORE_|_I_|_5  ,     0, "Sandy Bridge (Core i5)"   },
+	{  6, 10, -1, -1, 42,   2,    -1,    -1, NC, CORE_|_I_|_3  ,     0, "Sandy Bridge (Core i3)"   },
+	{  6, 10, -1, -1, 42,   2,    -1,    -1, NC, PENTIUM_      ,     0, "Sandy Bridge (Pentium)"   },
+	{  6, 10, -1, -1, 42,   1,    -1,    -1, NC, CELERON_      ,     0, "Sandy Bridge (Celeron)"   },
+	{  6, 10, -1, -1, 42,   2,    -1,    -1, NC, CELERON_      ,     0, "Sandy Bridge (Celeron)"   },
+	{  6, 13, -1, -1, 45,  -1,    -1,    -1, NC, CORE_|_I_|_3  ,     0, "Sandy Bridge-E"           },
+	{  6, 13, -1, -1, 45,  -1,    -1,    -1, NC, XEON_         ,     0, "Sandy Bridge-E (Xeon)"    },
+
+	/* Ivy Bridge CPUs (22nm): */
+	{  6, 10, -1, -1, 58,  -1,    -1,    -1, NC, XEON_         ,     0, "Ivy Bridge (Xeon)"        },
+	{  6, 10, -1, -1, 58,   4,    -1,    -1, NC, CORE_|_I_|_7  ,     0, "Ivy Bridge (Core i7)"     },
+	{  6, 10, -1, -1, 58,   4,    -1,    -1, NC, CORE_|_I_|_5  ,     0, "Ivy Bridge (Core i5)"     },
+	{  6, 10, -1, -1, 58,   2,    -1,    -1, NC, CORE_|_I_|_3  ,     0, "Ivy Bridge (Core i3)"     },
+	{  6, 10, -1, -1, 58,   2,    -1,    -1, NC, PENTIUM_      ,     0, "Ivy Bridge (Pentium)"     },
+	{  6, 10, -1, -1, 58,   1,    -1,    -1, NC, CELERON_      ,     0, "Ivy Bridge (Celeron)"     },
+	{  6, 10, -1, -1, 58,   2,    -1,    -1, NC, CELERON_      ,     0, "Ivy Bridge (Celeron)"     },
+	{  6, 14, -1, -1, 62,  -1,    -1,    -1, NC, 0             ,     0, "Ivy Bridge-E"             },
+	
+	/* Haswell CPUs (22nm): */
+	{  6, 12, -1, -1, 60,  -1,    -1,    -1, NC, XEON_         ,     0, "Haswell (Xeon)"           },
+	{  6, 12, -1, -1, 60,   4,    -1,    -1, NC, CORE_|_I_|_7  ,     0, "Haswell (Core i7)"        },
+	{  6,  5, -1, -1, 69,   4,    -1,    -1, NC, CORE_|_I_|_7  ,     0, "Haswell (Core i7)"        },
+	{  6,  6, -1, -1, 70,   4,    -1,    -1, NC, CORE_|_I_|_7  ,     0, "Haswell (Core i7)"        },
+	{  6, 12, -1, -1, 60,   4,    -1,    -1, NC, CORE_|_I_|_5  ,     0, "Haswell (Core i5)"        },
+	{  6,  5, -1, -1, 69,   4,    -1,    -1, NC, CORE_|_I_|_5  ,     0, "Haswell (Core i5)"        },
+	{  6, 12, -1, -1, 60,   2,    -1,    -1, NC, CORE_|_I_|_5  ,     0, "Haswell (Core i5)"        },
+	{  6,  5, -1, -1, 69,   2,    -1,    -1, NC, CORE_|_I_|_5  ,     0, "Haswell (Core i5)"        },
+	{  6, 12, -1, -1, 60,   2,    -1,    -1, NC, CORE_|_I_|_3  ,     0, "Haswell (Core i3)"        },
+	{  6,  5, -1, -1, 69,   2,    -1,    -1, NC, CORE_|_I_|_3  ,     0, "Haswell (Core i3)"        },
+	{  6, 12, -1, -1, 60,   2,    -1,    -1, NC, PENTIUM_      ,     0, "Haswell (Pentium)"        },
+	{  6, 12, -1, -1, 60,   2,    -1,    -1, NC, CELERON_      ,     0, "Haswell (Celeron)"        },
+	{  6, 12, -1, -1, 60,   1,    -1,    -1, NC, CELERON_      ,     0, "Haswell (Celeron)"        },
+	{  6, 15, -1, -1, 63,  -1,    -1,    -1, NC, 0             ,     0, "Haswell-E"                },
+
+	/* Broadwell CPUs (14nm): */
+	{  6,  7, -1, -1, 71,   4,    -1,    -1, NC, CORE_|_I_|_7  ,     0, "Broadwell (Core i7)"      },
+	{  6,  7, -1, -1, 71,   4,    -1,    -1, NC, CORE_|_I_|_5  ,     0, "Broadwell (Core i5)"      },
+	{  6, 13, -1, -1, 61,   4,    -1,    -1, NC, CORE_|_I_|_7  ,     0, "Broadwell-U (Core i7)"    },
+	{  6, 13, -1, -1, 61,   2,    -1,    -1, NC, CORE_|_I_|_7  ,     0, "Broadwell-U (Core i7)"    },
+	{  6, 13, -1, -1, 61,   2,    -1,    -1, NC, CORE_|_I_|_5  ,     0, "Broadwell-U (Core i5)"    },
+	{  6, 13, -1, -1, 61,   2,    -1,    -1, NC, CORE_|_I_|_3  ,     0, "Broadwell-U (Core i3)"    },
+	{  6, 13, -1, -1, 61,   2,    -1,    -1, NC, PENTIUM_      ,     0, "Broadwell-U (Pentium)"    },
+	{  6, 13, -1, -1, 61,   2,    -1,    -1, NC, CELERON_      ,     0, "Broadwell-U (Celeron)"    },
+	{  6, 13, -1, -1, 61,   2,    -1,    -1, NA, 0             ,     0, "Broadwell-U (Core M)"     },
+	{  6, 15, -1, -1, 79,  -1,    -1,    -1, NC, XEON_         ,     0, "Broadwell-E (Xeon)"       },
+	{  6, 15, -1, -1, 79,   2,    -1,    -1, NC, CORE_|_I_|_3  ,     0, "Broadwell-E (Core i3)"    },
+	{  6, 15, -1, -1, 79,   2,    -1,    -1, NC, CORE_|_I_|_5  ,     0, "Broadwell-E (Core i5)"    },
+	{  6, 15, -1, -1, 79,   4,    -1,    -1, NC, CORE_|_I_|_5  ,     0, "Broadwell-E (Core i5)"    },
+	{  6, 15, -1, -1, 79,   2,    -1,    -1, NC, CORE_|_I_|_7  ,     0, "Broadwell-E (Core i7)"    },
+	{  6, 15, -1, -1, 79,   4,    -1,    -1, NC, CORE_|_I_|_7  ,     0, "Broadwell-E (Core i7)"    },
+
+	/* Skylake CPUs (14nm): */
+	{  6, 14, -1, -1, 94,  -1,    -1,    -1, NC, XEON_         ,     0, "Skylake (Xeon)"           },
+	{  6, 14, -1, -1, 94,   4,    -1,    -1, NC, CORE_|_I_|_7  ,     0, "Skylake (Core i7)"        },
+	{  6, 14, -1, -1, 94,   4,    -1,    -1, NC, CORE_|_I_|_5  ,     0, "Skylake (Core i5)"        },
+	{  6, 14, -1, -1, 94,   2,    -1,    -1, NC, CORE_|_I_|_3  ,     0, "Skylake (Core i3)"        },
+	{  6, 14, -1, -1, 94,   2,    -1,    -1, NC, PENTIUM_      ,     0, "Skylake (Pentium)"        },
+	{  6, 14, -1, -1, 78,   2,    -1,    -1, NC, PENTIUM_      ,     0, "Skylake (Pentium)"        },
+	{  6, 14, -1, -1, 94,   2,    -1,    -1, NC, CELERON_      ,     0, "Skylake (Celeron)"        },
+	{  6, 14, -1, -1, 78,   2,    -1,    -1, NC, CELERON_      ,     0, "Skylake (Celeron)"        },
+	{  6, 14, -1, -1, 78,   2,    -1,    -1, NC, CORE_|_M_|_7  ,     0, "Skylake (Core m7)"        },
+	{  6, 14, -1, -1, 78,   2,    -1,    -1, NC, CORE_|_M_|_5  ,     0, "Skylake (Core m5)"        },
+	{  6, 14, -1, -1, 78,   2,    -1,    -1, NC, CORE_|_M_|_3  ,     0, "Skylake (Core m3)"        },
+
+	/* Kaby Lake CPUs (14nm): */
+	{  6, 14, -1, -1, 158,  4,    -1,    -1, NC, CORE_|_I_|_7  ,     0, "Kaby Lake (Core i7)"      },
+	{  6, 14, -1, -1, 158,  4,    -1,    -1, NC, CORE_|_I_|_5  ,     0, "Kaby Lake (Core i5)"      },
+	{  6, 14, -1, -1, 158,  2,    -1,    -1, NC, CORE_|_I_|_3  ,     0, "Kaby Lake (Core i3)"      },
+	{  6, 14, -1, -1, 158,  2,    -1,    -1, NC, PENTIUM_      ,     0, "Kaby Lake (Pentium)"      },
+	{  6, 14, -1, -1, 158,  2,    -1,    -1, NC, CELERON_      ,     0, "Kaby Lake (Celeron)"      },
+	{  6, 14, -1, -1, 158,  2,    -1,    -1, NC, CORE_|_M_|_3  ,     0, "Kaby Lake (Core m3)"      },
+
+	/* Itaniums */
+	{  7, -1, -1, -1, -1,   1,    -1,    -1, NC, 0             ,     0, "Itanium"                  },
+	{ 15, -1, -1, 16, -1,   1,    -1,    -1, NC, 0             ,     0, "Itanium 2"                },
+};
+
+
+static void load_intel_features(struct cpu_raw_data_t* raw, struct cpu_id_t* data)
+{
+	const struct feature_map_t matchtable_edx1[] = {
+		{ 18, CPU_FEATURE_PN },
+		{ 21, CPU_FEATURE_DTS },
+		{ 22, CPU_FEATURE_ACPI },
+		{ 27, CPU_FEATURE_SS },
+		{ 29, CPU_FEATURE_TM },
+		{ 30, CPU_FEATURE_IA64 },
+		{ 31, CPU_FEATURE_PBE },
+	};
+	const struct feature_map_t matchtable_ecx1[] = {
+		{  2, CPU_FEATURE_DTS64 },
+		{  4, CPU_FEATURE_DS_CPL },
+		{  5, CPU_FEATURE_VMX },
+		{  6, CPU_FEATURE_SMX },
+		{  7, CPU_FEATURE_EST },
+		{  8, CPU_FEATURE_TM2 },
+		{ 10, CPU_FEATURE_CID },
+		{ 14, CPU_FEATURE_XTPR },
+		{ 15, CPU_FEATURE_PDCM },
+		{ 18, CPU_FEATURE_DCA },
+		{ 21, CPU_FEATURE_X2APIC },
+	};
+	const struct feature_map_t matchtable_edx81[] = {
+		{ 20, CPU_FEATURE_XD },
+	};
+	const struct feature_map_t matchtable_ebx7[] = {
+		{  2, CPU_FEATURE_SGX },
+		{  4, CPU_FEATURE_HLE },
+		{ 11, CPU_FEATURE_RTM },
+		{ 16, CPU_FEATURE_AVX512F },
+		{ 17, CPU_FEATURE_AVX512DQ },
+		{ 18, CPU_FEATURE_RDSEED },
+		{ 19, CPU_FEATURE_ADX },
+		{ 26, CPU_FEATURE_AVX512PF },
+		{ 27, CPU_FEATURE_AVX512ER },
+		{ 28, CPU_FEATURE_AVX512CD },
+		{ 29, CPU_FEATURE_SHA_NI },
+		{ 30, CPU_FEATURE_AVX512BW },
+		{ 31, CPU_FEATURE_AVX512VL },
+	};
+	if (raw->basic_cpuid[0][0] >= 1) {
+		match_features(matchtable_edx1, COUNT_OF(matchtable_edx1), raw->basic_cpuid[1][3], data);
+		match_features(matchtable_ecx1, COUNT_OF(matchtable_ecx1), raw->basic_cpuid[1][2], data);
+	}
+	if (raw->ext_cpuid[0][0] >= 1) {
+		match_features(matchtable_edx81, COUNT_OF(matchtable_edx81), raw->ext_cpuid[1][3], data);
+	}
+	// detect TSX/AVX512:
+	if (raw->basic_cpuid[0][0] >= 7) {
+		match_features(matchtable_ebx7, COUNT_OF(matchtable_ebx7), raw->basic_cpuid[7][1], data);
+	}
+}
+
+enum _cache_type_t {
+	L1I,
+	L1D,
+	L2,
+	L3,
+	L4
+};
+typedef enum _cache_type_t cache_type_t;
+
+static void check_case(uint8_t on, cache_type_t cache, int size, int assoc, int linesize, struct cpu_id_t* data)
+{
+	if (!on) return;
+	switch (cache) {
+		case L1I:
+			data->l1_instruction_cache = size;
+			break;
+		case L1D:
+			data->l1_data_cache = size;
+			data->l1_assoc = assoc;
+			data->l1_cacheline = linesize;
+			break;
+		case L2:
+			data->l2_cache = size;
+			data->l2_assoc = assoc;
+			data->l2_cacheline = linesize;
+			break;
+		case L3:
+			data->l3_cache = size;
+			data->l3_assoc = assoc;
+			data->l3_cacheline = linesize;
+			break;
+		case L4:
+			data->l4_cache = size;
+			data->l4_assoc = assoc;
+			data->l4_cacheline = linesize;
+			break;
+		default:
+			break;
+	}
+}
+
+static void decode_intel_oldstyle_cache_info(struct cpu_raw_data_t* raw, struct cpu_id_t* data)
+{
+	uint8_t f[256] = {0};
+	int reg, off;
+	uint32_t x;
+	for (reg = 0; reg < 4; reg++) {
+		x = raw->basic_cpuid[2][reg];
+		if (x & 0x80000000) continue;
+		for (off = 0; off < 4; off++) {
+			f[x & 0xff] = 1;
+			x >>= 8;
+		}
+	}
+	
+	check_case(f[0x06], L1I,      8,  4,  32, data);
+	check_case(f[0x08], L1I,     16,  4,  32, data);
+	check_case(f[0x0A], L1D,      8,  2,  32, data);
+	check_case(f[0x0C], L1D,     16,  4,  32, data);
+	check_case(f[0x22],  L3,    512,  4,  64, data);
+	check_case(f[0x23],  L3,   1024,  8,  64, data);
+	check_case(f[0x25],  L3,   2048,  8,  64, data);
+	check_case(f[0x29],  L3,   4096,  8,  64, data);
+	check_case(f[0x2C], L1D,     32,  8,  64, data);
+	check_case(f[0x30], L1I,     32,  8,  64, data);
+	check_case(f[0x39],  L2,    128,  4,  64, data);
+	check_case(f[0x3A],  L2,    192,  6,  64, data);
+	check_case(f[0x3B],  L2,    128,  2,  64, data);
+	check_case(f[0x3C],  L2,    256,  4,  64, data);
+	check_case(f[0x3D],  L2,    384,  6,  64, data);
+	check_case(f[0x3E],  L2,    512,  4,  64, data);
+	check_case(f[0x41],  L2,    128,  4,  32, data);
+	check_case(f[0x42],  L2,    256,  4,  32, data);
+	check_case(f[0x43],  L2,    512,  4,  32, data);
+	check_case(f[0x44],  L2,   1024,  4,  32, data);
+	check_case(f[0x45],  L2,   2048,  4,  32, data);
+	check_case(f[0x46],  L3,   4096,  4,  64, data);
+	check_case(f[0x47],  L3,   8192,  8,  64, data);
+	check_case(f[0x4A],  L3,   6144, 12,  64, data);
+	check_case(f[0x4B],  L3,   8192, 16,  64, data);
+	check_case(f[0x4C],  L3,  12288, 12,  64, data);
+	check_case(f[0x4D],  L3,  16384, 16,  64, data);
+	check_case(f[0x4E],  L2,   6144, 24,  64, data);
+	check_case(f[0x60], L1D,     16,  8,  64, data);
+	check_case(f[0x66], L1D,      8,  4,  64, data);
+	check_case(f[0x67], L1D,     16,  4,  64, data);
+	check_case(f[0x68], L1D,     32,  4,  64, data);
+	/* The following four entries are trace cache. Intel does not
+	 * specify a cache-line size, so we use -1 instead
+	 */
+	check_case(f[0x70], L1I,     12,  8,  -1, data);
+	check_case(f[0x71], L1I,     16,  8,  -1, data);
+	check_case(f[0x72], L1I,     32,  8,  -1, data);
+	check_case(f[0x73], L1I,     64,  8,  -1, data);
+	
+	check_case(f[0x78],  L2,   1024,  4,  64, data);
+	check_case(f[0x79],  L2,    128,  8,  64, data);
+	check_case(f[0x7A],  L2,    256,  8,  64, data);
+	check_case(f[0x7B],  L2,    512,  8,  64, data);
+	check_case(f[0x7C],  L2,   1024,  8,  64, data);
+	check_case(f[0x7D],  L2,   2048,  8,  64, data);
+	check_case(f[0x7F],  L2,    512,  2,  64, data);
+	check_case(f[0x82],  L2,    256,  8,  32, data);
+	check_case(f[0x83],  L2,    512,  8,  32, data);
+	check_case(f[0x84],  L2,   1024,  8,  32, data);
+	check_case(f[0x85],  L2,   2048,  8,  32, data);
+	check_case(f[0x86],  L2,    512,  4,  64, data);
+	check_case(f[0x87],  L2,   1024,  8,  64, data);
+	
+	if (f[0x49]) {
+		/* This flag is overloaded with two meanings. On Xeon MP
+		 * (family 0xf, model 0x6) this means L3 cache. On all other
+		 * CPUs (notably Conroe et al), this is L2 cache. In both cases
+		 * it means 4MB, 16-way associative, 64-byte line size.
+		 */
+		if (data->family == 0xf && data->model == 0x6) {
+			data->l3_cache = 4096;
+			data->l3_assoc = 16;
+			data->l3_cacheline = 64;
+		} else {
+			data->l2_cache = 4096;
+			data->l2_assoc = 16;
+			data->l2_cacheline = 64;
+		}
+	}
+	if (f[0x40]) {
+		/* Again, a special flag. It means:
+		 * 1) If no L2 is specified, then CPU is w/o L2 (0 KB)
+		 * 2) If L2 is specified by other flags, then, CPU is w/o L3.
+		 */
+		if (data->l2_cache == -1) {
+			data->l2_cache = 0;
+		} else {
+			data->l3_cache = 0;
+		}
+	}
+}
+
+static void decode_intel_deterministic_cache_info(struct cpu_raw_data_t* raw,
+                                                  struct cpu_id_t* data)
+{
+	int ecx;
+	int ways, partitions, linesize, sets, size, level, typenumber;
+	cache_type_t type;
+	for (ecx = 0; ecx < MAX_INTELFN4_LEVEL; ecx++) {
+		typenumber = raw->intel_fn4[ecx][0] & 0x1f;
+		if (typenumber == 0) break;
+		level = (raw->intel_fn4[ecx][0] >> 5) & 0x7;
+		if (level == 1 && typenumber == 1)
+			type = L1D;
+		else if (level == 1 && typenumber == 2)
+			type = L1I;
+		else if (level == 2 && typenumber == 3)
+			type = L2;
+		else if (level == 3 && typenumber == 3)
+			type = L3;
+		else if (level == 4 && typenumber == 3)
+			type = L4;
+		else {
+			warnf("deterministic_cache: unknown level/typenumber combo (%d/%d), cannot\n", level, typenumber);
+			warnf("deterministic_cache: recognize cache type\n");
+			continue;
+		}
+		ways = ((raw->intel_fn4[ecx][1] >> 22) & 0x3ff) + 1;
+		partitions = ((raw->intel_fn4[ecx][1] >> 12) & 0x3ff) + 1;
+		linesize = (raw->intel_fn4[ecx][1] & 0xfff) + 1;
+		sets = raw->intel_fn4[ecx][2] + 1;
+		size = ways * partitions * linesize * sets / 1024;
+		check_case(1, type, size, ways, linesize, data);
+	}
+}
+
+static int decode_intel_extended_topology(struct cpu_raw_data_t* raw,
+                                           struct cpu_id_t* data)
+{
+	int i, level_type, num_smt = -1, num_core = -1;
+	for (i = 0; i < MAX_INTELFN11_LEVEL; i++) {
+		level_type = (raw->intel_fn11[i][2] & 0xff00) >> 8;
+		switch (level_type) {
+			case 0x01:
+				num_smt = raw->intel_fn11[i][1] & 0xffff;
+				break;
+			case 0x02:
+				num_core = raw->intel_fn11[i][1] & 0xffff;
+				break;
+			default:
+				break;
+		}
+	}
+	if (num_smt == -1 || num_core == -1) return 0;
+	data->num_logical_cpus = num_core;
+	data->num_cores = num_core / num_smt;
+	// make sure num_cores is at least 1. In VMs, the CPUID instruction
+	// is rigged and may give nonsensical results, but we should at least
+	// avoid outputs like data->num_cores == 0.
+	if (data->num_cores <= 0) data->num_cores = 1;
+	return 1;
+}
+
+static void decode_intel_number_of_cores(struct cpu_raw_data_t* raw,
+                                         struct cpu_id_t* data)
+{
+	int logical_cpus = -1, num_cores = -1;
+	
+	if (raw->basic_cpuid[0][0] >= 11) {
+		if (decode_intel_extended_topology(raw, data)) return;
+	}
+	
+	if (raw->basic_cpuid[0][0] >= 1) {
+		logical_cpus = (raw->basic_cpuid[1][1] >> 16) & 0xff;
+		if (raw->basic_cpuid[0][0] >= 4) {
+			num_cores = 1 + ((raw->basic_cpuid[4][0] >> 26) & 0x3f);
+		}
+	}
+	if (data->flags[CPU_FEATURE_HT]) {
+		if (num_cores > 1) {
+			data->num_cores = num_cores;
+			data->num_logical_cpus = logical_cpus;
+		} else {
+			data->num_cores = 1;
+			data->num_logical_cpus = (logical_cpus >= 1 ? logical_cpus : 1);
+			if (data->num_logical_cpus == 1)
+				data->flags[CPU_FEATURE_HT] = 0;
+		}
+	} else {
+		data->num_cores = data->num_logical_cpus = 1;
+	}
+}
+
+static intel_code_and_bits_t get_brand_code_and_bits(struct cpu_id_t* data)
+{
+	intel_code_t code = (intel_code_t) NC;
+	intel_code_and_bits_t result;
+	uint64_t bits = 0;
+	int i = 0;
+	const char* bs = data->brand_str;
+	const char* s;
+	const struct { intel_code_t c; const char *search; } matchtable[] = {
+		{ PENTIUM_M, "Pentium(R) M" },
+		{ CORE_SOLO, "Pentium(R) Dual  CPU" },
+		{ CORE_SOLO, "Pentium(R) Dual-Core" },
+		{ PENTIUM_D, "Pentium(R) D" },
+		{ CORE_SOLO, "Genuine Intel(R) CPU" },
+		{ CORE_SOLO, "Intel(R) Core(TM)" },
+		{ DIAMONDVILLE, "CPU [N ][23]## " },
+		{ SILVERTHORNE, "CPU Z" },
+		{ PINEVIEW, "CPU [ND][45]## " },
+		{ CEDARVIEW, "CPU [ND]#### " },
+	};
+	
+	const struct { uint64_t bit; const char* search; } bit_matchtable[] = {
+		{ XEON_, "Xeon" },
+		{ _MP, " MP" },
+		{ ATOM_, "Atom(TM) CPU" },
+		{ MOBILE_, "Mobile" },
+		{ CELERON_, "Celeron" },
+		{ PENTIUM_, "Pentium" },
+	};
+	
+	for (i = 0; i < COUNT_OF(bit_matchtable); i++) {
+		if (match_pattern(bs, bit_matchtable[i].search))
+			bits |= bit_matchtable[i].bit;
+	}
+	
+	if ((i = match_pattern(bs, "Core(TM) [im][357]")) != 0) {
+		bits |= CORE_;
+		i--;
+		switch (bs[i + 9]) {
+			case 'i': bits |= _I_; break;
+			case 'm': bits |= _M_; break;
+		}
+		switch (bs[i + 10]) {
+			case '3': bits |= _3; break;
+			case '5': bits |= _5; break;
+			case '7': bits |= _7; break;
+		}
+	}
+	for (i = 0; i < COUNT_OF(matchtable); i++)
+		if (match_pattern(bs, matchtable[i].search)) {
+			code = matchtable[i].c;
+			break;
+		}
+	debugf(2, "intel matchtable result is %d\n", code);
+	if (bits & XEON_) {
+		if (match_pattern(bs, "W35##") || match_pattern(bs, "[ELXW]75##"))
+			bits |= _7;
+		else if (match_pattern(bs, "[ELXW]55##"))
+			code = GAINESTOWN;
+		else if (match_pattern(bs, "[ELXW]56##"))
+			code = WESTMERE;
+		else if (data->l3_cache > 0 && data->family == 16)
+			/* restrict by family, since later Xeons also have L3 ... */
+			code = IRWIN;
+	}
+	if (match_all(bits, XEON_ + _MP) && data->l3_cache > 0)
+		code = POTOMAC;
+	if (code == CORE_SOLO) {
+		s = strstr(bs, "CPU");
+		if (s) {
+			s += 3;
+			while (*s == ' ') s++;
+			if (*s == 'T')
+				bits |= MOBILE_;
+		}
+	}
+	if (code == CORE_SOLO) {
+		switch (data->num_cores) {
+			case 1: break;
+			case 2:
+			{
+				code = CORE_DUO;
+				if (data->num_logical_cpus > 2)
+					code = DUAL_CORE_HT;
+				break;
+			}
+			case 4:
+			{
+				code = QUAD_CORE;
+				if (data->num_logical_cpus > 4)
+					code = QUAD_CORE_HT;
+				break;
+			}
+			default:
+				code = MORE_THAN_QUADCORE; break;
+		}
+	}
+	
+	if (code == CORE_DUO && (bits & MOBILE_) && data->model != 14) {
+		if (data->ext_model < 23) {
+			code = MEROM;
+		} else {
+			code = PENRYN;
+		}
+	}
+	if (data->ext_model == 23 &&
+		(code == CORE_DUO || code == PENTIUM_D || (bits & CELERON_))) {
+		code = WOLFDALE;
+	}
+
+	result.code = code;
+	result.bits = bits;
+	return result;
+}
+
+static intel_model_t get_model_code(struct cpu_id_t* data)
+{
+	int i = 0;
+	int l = (int) strlen(data->brand_str);
+	const char *bs = data->brand_str;
+	int mod_flags = 0, model_no = 0, ndigs = 0;
+	/* If the CPU is a Core ix, then just return the model number generation: */
+	if ((i = match_pattern(bs, "Core(TM) i[357]")) != 0) {
+		i += 11;
+		if (i + 4 >= l) return UNKNOWN;
+		if (bs[i] == '2') return _2xxx;
+		if (bs[i] == '3') return _3xxx;
+		return UNKNOWN;
+	}
+	
+	/* For Core2-based Xeons: */
+	while (i < l - 3) {
+		if (bs[i] == 'C' && bs[i+1] == 'P' && bs[i+2] == 'U')
+			break;
+		i++;
+	}
+	if (i >= l - 3) return UNKNOWN;
+	i += 3;
+	while (i < l - 4 && bs[i] == ' ') i++;
+	if (i >= l - 4) return UNKNOWN;
+	while (i < l - 4 && !isdigit(bs[i])) {
+		if (bs[i] >= 'A' && bs[i] <= 'Z')
+			mod_flags |= (1 << (bs[i] - 'A'));
+		i++;
+	}
+	if (i >= l - 4) return UNKNOWN;
+	while (isdigit(bs[i])) {
+		ndigs++;
+		model_no = model_no * 10 + (int) (bs[i] - '0');
+		i++;
+	}
+	if (ndigs != 4) return UNKNOWN;
+#define HAVE(ch, flags) ((flags & (1 << ((int)(ch-'A')))) != 0)
+	switch (model_no / 100) {
+		case 30: return _3000;
+		case 31: return _3100;
+		case 32:
+		{
+			return (HAVE('X', mod_flags)) ? X3200 : _3200;
+		}
+		case 33:
+		{
+			return (HAVE('X', mod_flags)) ? X3300 : _3300;
+		}
+		case 51: return _5100;
+		case 52: return _5200;
+		case 53: return _5300;
+		case 54: return _5400;
+		default:
+			return UNKNOWN;
+	}
+#undef HAVE
+}
+
+static void decode_intel_sgx_features(const struct cpu_raw_data_t* raw, struct cpu_id_t* data)
+{
+	struct cpu_epc_t epc;
+	int i;
+	
+	if (raw->basic_cpuid[0][0] < 0x12) return; // no 12h leaf
+	if (raw->basic_cpuid[0x12][0] == 0) return; // no sub-leafs available, probably it's disabled by BIOS
+	
+	// decode sub-leaf 0:
+	if (raw->basic_cpuid[0x12][0] & 1) data->sgx.flags[INTEL_SGX1] = 1;
+	if (raw->basic_cpuid[0x12][0] & 2) data->sgx.flags[INTEL_SGX2] = 1;
+	if (data->sgx.flags[INTEL_SGX1] || data->sgx.flags[INTEL_SGX2])
+		data->sgx.present = 1;
+	data->sgx.misc_select = raw->basic_cpuid[0x12][1];
+	data->sgx.max_enclave_32bit = (raw->basic_cpuid[0x12][3]     ) & 0xff;
+	data->sgx.max_enclave_64bit = (raw->basic_cpuid[0x12][3] >> 8) & 0xff;
+	
+	// decode sub-leaf 1:
+	data->sgx.secs_attributes = raw->intel_fn12h[1][0] | (((uint64_t) raw->intel_fn12h[1][1]) << 32);
+	data->sgx.secs_xfrm       = raw->intel_fn12h[1][2] | (((uint64_t) raw->intel_fn12h[1][3]) << 32);
+	
+	// decode higher-order subleafs, whenever present:
+	data->sgx.num_epc_sections = -1;
+	for (i = 0; i < 1000000; i++) {
+		epc = cpuid_get_epc(i, raw);
+		if (epc.length == 0) {
+			debugf(2, "SGX: epc section request for %d returned null, no more EPC sections.\n", i);
+			data->sgx.num_epc_sections = i;
+			break;
+		}
+	}
+	if (data->sgx.num_epc_sections == -1) {
+		debugf(1, "SGX: warning: seems to be infinitude of EPC sections.\n");
+		data->sgx.num_epc_sections = 1000000;
+	}
+}
+
+struct cpu_epc_t cpuid_get_epc(int index, const struct cpu_raw_data_t* raw)
+{
+	uint32_t regs[4];
+	struct cpu_epc_t retval = {0, 0};
+	if (raw && index < MAX_INTELFN12H_LEVEL - 2) {
+		// this was queried already, use the data:
+		memcpy(regs, raw->intel_fn12h[2 + index], sizeof(regs));
+	} else {
+		// query this ourselves:
+		regs[0] = 0x12;
+		regs[2] = 2 + index;
+		regs[1] = regs[3] = 0;
+		cpu_exec_cpuid_ext(regs);
+	}
+	
+	// decode values:
+	if ((regs[0] & 0xf) == 0x1) {
+		retval.start_addr |= (regs[0] & 0xfffff000); // bits [12, 32) -> bits [12, 32)
+		retval.start_addr |= ((uint64_t) (regs[1] & 0x000fffff)) << 32; // bits [0, 20) -> bits [32, 52)
+		retval.length     |= (regs[2] & 0xfffff000); // bits [12, 32) -> bits [12, 32)
+		retval.length     |= ((uint64_t) (regs[3] & 0x000fffff)) << 32; // bits [0, 20) -> bits [32, 52)
+	}
+	return retval;
+}
+
+int cpuid_identify_intel(struct cpu_raw_data_t* raw, struct cpu_id_t* data, struct internal_id_info_t* internal)
+{
+	intel_code_and_bits_t brand;
+	intel_model_t model_code;
+	int i;
+	char* brand_code_str = NULL;
+
+	load_intel_features(raw, data);
+	if (raw->basic_cpuid[0][0] >= 4) {
+		/* Deterministic way is preferred, being more generic */
+		decode_intel_deterministic_cache_info(raw, data);
+	} else if (raw->basic_cpuid[0][0] >= 2) {
+		decode_intel_oldstyle_cache_info(raw, data);
+	}
+	decode_intel_number_of_cores(raw, data);
+
+	brand = get_brand_code_and_bits(data);
+	model_code = get_model_code(data);
+	for (i = 0; i < COUNT_OF(intel_bcode_str); i++) {
+		if (brand.code == intel_bcode_str[i].code) {
+			brand_code_str = intel_bcode_str[i].str;
+			break;
+		}
+	}
+	if (brand_code_str)
+		debugf(2, "Detected Intel brand code: %d (%s)\n", brand.code, brand_code_str);
+	else
+		debugf(2, "Detected Intel brand code: %d\n", brand.code);
+	if (brand.bits) {
+		debugf(2, "Detected Intel bits: ");
+		debug_print_lbits(2, brand.bits);
+	}
+	debugf(2, "Detected Intel model code: %d\n", model_code);
+	
+	internal->code.intel = brand.code;
+	internal->bits = brand.bits;
+	
+	if (data->flags[CPU_FEATURE_SGX]) {
+		debugf(2, "SGX seems to be present, decoding...\n");
+		// if SGX is indicated by the CPU, verify its presence:
+		decode_intel_sgx_features(raw, data);
+	}
+
+	internal->score = match_cpu_codename(cpudb_intel, COUNT_OF(cpudb_intel), data,
+		brand.code, brand.bits, model_code);
+	return 0;
+}
+
+void cpuid_get_list_intel(struct cpu_list_t* list)
+{
+	generic_get_cpu_list(cpudb_intel, COUNT_OF(cpudb_intel), list);
+}

compat/libcpuid/recog_intel.h

@@ -27,5 +27,6 @@
 #define __RECOG_INTEL_H__
 
 int cpuid_identify_intel(struct cpu_raw_data_t* raw, struct cpu_id_t* data, struct internal_id_info_t* internal);
+void cpuid_get_list_intel(struct cpu_list_t* list);
 
 #endif /*__RECOG_INTEL_H__*/

compat/winansi.c

@@ -0,0 +1,392 @@
+/**
+ * Old Git implementation of windows terminal colors (2009)
+ * before use of a threaded wrapper.
+ */
+
+#undef NOGDI
+#include <windows.h>
+#include <wingdi.h>
+#include <winreg.h>
+#include <malloc.h>
+#include <stdio.h>
+#include <io.h>
+
+#include "compat/winansi.h"
+/*
+* Copyright 2008 Peter Harris <git@peter.is-a-geek.org>
+*/
+
+/*
+Functions to be wrapped:
+*/
+#undef printf
+#undef fprintf
+#undef fputs
+#undef vfprintf
+/* TODO: write */
+
+/*
+ANSI codes used by git: m, K
+
+This file is git-specific. Therefore, this file does not attempt
+to implement any codes that are not used by git.
+*/
+
+static HANDLE console;
+static WORD plain_attr;
+static WORD attr;
+static int negative;
+
+static void init(void)
+{
+    CONSOLE_SCREEN_BUFFER_INFO sbi;
+
+    static int initialized = 0;
+    if (initialized)
+        return;
+
+    console = GetStdHandle(STD_OUTPUT_HANDLE);
+    if (console == INVALID_HANDLE_VALUE)
+        console = NULL;
+
+    if (!console)
+        return;
+
+    GetConsoleScreenBufferInfo(console, &sbi);
+    attr = plain_attr = sbi.wAttributes;
+    negative = 0;
+
+    initialized = 1;
+}
+
+static int write_console(const char *str, int len)
+{
+    /* convert utf-8 to utf-16, write directly to console */
+    int wlen = MultiByteToWideChar(CP_UTF8, 0, str, len, NULL, 0);
+    wchar_t *wbuf = (wchar_t *)alloca(wlen * sizeof(wchar_t));
+    MultiByteToWideChar(CP_UTF8, 0, str, len, wbuf, wlen);
+
+    WriteConsoleW(console, wbuf, wlen, NULL, NULL);
+
+    /* return original (utf-8 encoded) length */
+    return len;
+}
+
+#define FOREGROUND_ALL (FOREGROUND_RED | FOREGROUND_GREEN | FOREGROUND_BLUE)
+#define BACKGROUND_ALL (BACKGROUND_RED | BACKGROUND_GREEN | BACKGROUND_BLUE)
+
+static void set_console_attr(void)
+{
+    WORD attributes = attr;
+    if (negative) {
+        attributes &= ~FOREGROUND_ALL;
+        attributes &= ~BACKGROUND_ALL;
+
+        /* This could probably use a bitmask
+        instead of a series of ifs */
+        if (attr & FOREGROUND_RED)
+            attributes |= BACKGROUND_RED;
+        if (attr & FOREGROUND_GREEN)
+            attributes |= BACKGROUND_GREEN;
+        if (attr & FOREGROUND_BLUE)
+            attributes |= BACKGROUND_BLUE;
+
+        if (attr & BACKGROUND_RED)
+            attributes |= FOREGROUND_RED;
+        if (attr & BACKGROUND_GREEN)
+            attributes |= FOREGROUND_GREEN;
+        if (attr & BACKGROUND_BLUE)
+            attributes |= FOREGROUND_BLUE;
+    }
+    SetConsoleTextAttribute(console, attributes);
+}
+
+static void erase_in_line(void)
+{
+    CONSOLE_SCREEN_BUFFER_INFO sbi;
+    DWORD dummy; /* Needed for Windows 7 (or Vista) regression */
+
+    if (!console)
+        return;
+
+    GetConsoleScreenBufferInfo(console, &sbi);
+    FillConsoleOutputCharacterA(console, ' ',
+        sbi.dwSize.X - sbi.dwCursorPosition.X, sbi.dwCursorPosition,
+        &dummy);
+}
+
+
+static const char *set_attr(const char *str)
+{
+    const char *func;
+    size_t len = strspn(str, "0123456789;");
+    func = str + len;
+
+    switch (*func) {
+    case 'm':
+        do {
+            long val = strtol(str, (char **)&str, 10);
+            switch (val) {
+            case 0: /* reset */
+                attr = plain_attr;
+                negative = 0;
+                break;
+            case 1: /* bold */
+                attr |= FOREGROUND_INTENSITY;
+                break;
+            case 2:  /* faint */
+            case 22: /* normal */
+                attr &= ~FOREGROUND_INTENSITY;
+                break;
+            case 3:  /* italic */
+                /* Unsupported */
+                break;
+            case 4:  /* underline */
+            case 21: /* double underline */
+                /* Wikipedia says this flag does nothing */
+                /* Furthermore, mingw doesn't define this flag
+                attr |= COMMON_LVB_UNDERSCORE; */
+                break;
+            case 24: /* no underline */
+                /* attr &= ~COMMON_LVB_UNDERSCORE; */
+                break;
+            case 5:  /* slow blink */
+            case 6:  /* fast blink */
+                /* We don't have blink, but we do have
+                background intensity */
+                attr |= BACKGROUND_INTENSITY;
+                break;
+            case 25: /* no blink */
+                attr &= ~BACKGROUND_INTENSITY;
+                break;
+            case 7:  /* negative */
+                negative = 1;
+                break;
+            case 27: /* positive */
+                negative = 0;
+                break;
+            case 8:  /* conceal */
+            case 28: /* reveal */
+                /* Unsupported */
+                break;
+            case 30: /* Black */
+                attr &= ~FOREGROUND_ALL;
+                break;
+            case 31: /* Red */
+                attr &= ~FOREGROUND_ALL;
+                attr |= FOREGROUND_RED;
+                break;
+            case 32: /* Green */
+                attr &= ~FOREGROUND_ALL;
+                attr |= FOREGROUND_GREEN;
+                break;
+            case 33: /* Yellow */
+                attr &= ~FOREGROUND_ALL;
+                attr |= FOREGROUND_RED | FOREGROUND_GREEN;
+                break;
+            case 34: /* Blue */
+                attr &= ~FOREGROUND_ALL;
+                attr |= FOREGROUND_BLUE;
+                break;
+            case 35: /* Magenta */
+                attr &= ~FOREGROUND_ALL;
+                attr |= FOREGROUND_RED | FOREGROUND_BLUE;
+                break;
+            case 36: /* Cyan */
+                attr &= ~FOREGROUND_ALL;
+                attr |= FOREGROUND_GREEN | FOREGROUND_BLUE;
+                break;
+            case 37: /* White */
+                attr |= FOREGROUND_RED |
+                    FOREGROUND_GREEN |
+                    FOREGROUND_BLUE;
+                break;
+            case 38: /* Unknown */
+                break;
+            case 39: /* reset */
+                attr &= ~FOREGROUND_ALL;
+                attr |= (plain_attr & FOREGROUND_ALL);
+                break;
+            case 40: /* Black */
+                attr &= ~BACKGROUND_ALL;
+                break;
+            case 41: /* Red */
+                attr &= ~BACKGROUND_ALL;
+                attr |= BACKGROUND_RED;
+                break;
+            case 42: /* Green */
+                attr &= ~BACKGROUND_ALL;
+                attr |= BACKGROUND_GREEN;
+                break;
+            case 43: /* Yellow */
+                attr &= ~BACKGROUND_ALL;
+                attr |= BACKGROUND_RED | BACKGROUND_GREEN;
+                break;
+            case 44: /* Blue */
+                attr &= ~BACKGROUND_ALL;
+                attr |= BACKGROUND_BLUE;
+                break;
+            case 45: /* Magenta */
+                attr &= ~BACKGROUND_ALL;
+                attr |= BACKGROUND_RED | BACKGROUND_BLUE;
+                break;
+            case 46: /* Cyan */
+                attr &= ~BACKGROUND_ALL;
+                attr |= BACKGROUND_GREEN | BACKGROUND_BLUE;
+                break;
+            case 47: /* White */
+                attr |= BACKGROUND_RED |
+                    BACKGROUND_GREEN |
+                    BACKGROUND_BLUE;
+                break;
+            case 48: /* Unknown */
+                break;
+            case 49: /* reset */
+                attr &= ~BACKGROUND_ALL;
+                attr |= (plain_attr & BACKGROUND_ALL);
+                break;
+            default:
+                /* Unsupported code */
+                break;
+            }
+            str++;
+        } while (*(str - 1) == ';');
+
+        set_console_attr();
+        break;
+    case 'K':
+        erase_in_line();
+        break;
+    default:
+        /* Unsupported code */
+        break;
+    }
+
+    return func + 1;
+}
+
+static int ansi_emulate(const char *str, FILE *stream)
+{
+    int rv = 0;
+    const char *pos = str;
+
+    fflush(stream);
+
+    while (*pos) {
+        pos = strstr(str, "\033[");
+        if (pos) {
+            int len = (int) (pos - str);
+
+            if (len) {
+                int out_len = write_console(str, len);
+                rv += out_len;
+                if (out_len < len)
+                    return rv;
+            }
+
+            str = pos + 2;
+            rv += 2;
+
+            pos = set_attr(str);
+            rv += (int) (pos - str);
+            str = pos;
+        }
+        else {
+            int len = (int) strlen(str);
+            rv += write_console(str, len);
+            return rv;
+        }
+    }
+    return rv;
+}
+
+int winansi_fputs(const char *str, FILE *stream)
+{
+    int rv;
+
+    if (!isatty(fileno(stream)))
+        return fputs(str, stream);
+
+    init();
+
+    if (!console)
+        return fputs(str, stream);
+
+    rv = ansi_emulate(str, stream);
+
+    if (rv >= 0)
+        return 0;
+    else
+        return EOF;
+}
+
+int winansi_vfprintf(FILE *stream, const char *format, va_list list)
+{
+    int len, rv;
+    char small_buf[256] = { 0 };
+    char *buf = small_buf;
+    va_list cp;
+
+    if (!isatty(fileno(stream)))
+        goto abort;
+
+    init();
+
+    if (!console)
+        goto abort;
+
+    va_copy(cp, list);
+    len = vsnprintf(small_buf, sizeof(small_buf), format, cp);
+#ifdef WIN32
+    /* bug on long strings without that */
+    if (len == -1)
+        len = _vscprintf(format, cp);
+#endif
+    va_end(cp);
+
+    if (len > sizeof(small_buf) - 1) {
+        buf = malloc(len + 1);
+        if (!buf)
+            goto abort;
+
+        len = vsnprintf(buf, len + 1, format, list);
+#ifdef WIN32
+        if (len == -1)
+            len = _vscprintf(format, list);
+#endif
+    }
+
+    rv = ansi_emulate(buf, stream);
+
+    if (buf != small_buf)
+        free(buf);
+    return rv;
+
+abort:
+    rv = vfprintf(stream, format, list);
+    return rv;
+}
+
+int winansi_fprintf(FILE *stream, const char *format, ...)
+{
+    va_list list;
+    int rv;
+
+    va_start(list, format);
+    rv = winansi_vfprintf(stream, format, list);
+    va_end(list);
+
+    return rv;
+}
+
+int winansi_printf(const char *format, ...)
+{
+    va_list list;
+    int rv;
+
+    va_start(list, format);
+    rv = winansi_vfprintf(stdout, format, list);
+    va_end(list);
+
+    return rv;
+}

compat/winansi.h

@@ -0,0 +1,32 @@
+/*
+ * ANSI emulation wrappers
+ */
+#ifdef WIN32
+#include <windows.h>
+#include <stddef.h>
+#include <stdio.h>
+
+#define isatty(fd) _isatty(fd)
+#define fileno(fd) _fileno(fd)
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+    int winansi_fputs(const char *str, FILE *stream);
+    int winansi_printf(const char *format, ...);
+    int winansi_fprintf(FILE *stream, const char *format, ...);
+    int winansi_vfprintf(FILE *stream, const char *format, va_list list);
+#ifdef __cplusplus
+}
+#endif
+
+#undef fputs
+#undef fprintf
+#undef vfprintf
+
+#define fputs winansi_fputs
+#define printf winansi_printf
+#define fprintf winansi_fprintf
+#define vfprintf winansi_vfprintf
+
+#endif

cpu.c

@@ -0,0 +1,111 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018-2019 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2019 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#include <cpuid.h>
+#include <string.h>
+#include <stdbool.h>
+#include <math.h>
+
+#ifndef BUILD_TEST
+#   include <libcpuid.h>
+#endif
+
+#include "cpu.h"
+#include "options.h"
+
+
+#ifndef BUILD_TEST
+void cpu_init_common() {
+    struct cpu_raw_data_t raw = { 0 };
+    struct cpu_id_t data = { 0 };
+
+    cpuid_get_raw_data(&raw);
+    cpu_identify(&raw, &data);
+
+    strncpy(cpu_info.brand, data.brand_str, sizeof(cpu_info.brand) - 1);
+
+    cpu_info.total_logical_cpus = data.total_logical_cpus;
+    cpu_info.sockets            = data.total_logical_cpus / data.num_logical_cpus;
+    cpu_info.total_cores        = data.num_cores * cpu_info.sockets;
+    cpu_info.l3_cache           = data.l3_cache > 0 ? data.l3_cache * cpu_info.sockets : 0;
+
+    // Workaround for AMD CPUs https://github.com/anrieff/libcpuid/issues/97
+    if (data.vendor == VENDOR_AMD && data.l3_cache <= 0 && data.l2_assoc == 16 && data.ext_family >= 21) {
+        cpu_info.l2_cache = data.l2_cache * (cpu_info.total_cores / 2) * cpu_info.sockets;
+    }
+    else {
+        cpu_info.l2_cache = data.l2_cache > 0 ? data.l2_cache * cpu_info.total_cores * cpu_info.sockets : 0;
+    }
+
+
+#   ifdef __x86_64__
+    cpu_info.flags |= CPU_FLAG_X86_64;
+#   endif
+
+    if (data.flags[CPU_FEATURE_AES]) {
+        cpu_info.flags |= CPU_FLAG_AES;
+    }
+
+    if (data.flags[CPU_FEATURE_BMI2]) {
+        cpu_info.flags |= CPU_FLAG_BMI2;
+    }
+
+#   ifndef XMRIG_NO_ASM
+    if (data.vendor == VENDOR_AMD) {
+        cpu_info.assembly = (data.ext_family >= 23) ? ASM_RYZEN : ASM_BULLDOZER;
+    }
+    else if (data.vendor == VENDOR_INTEL) {
+        cpu_info.assembly = ASM_INTEL;
+    }
+#   endif
+}
+#endif
+
+
+int get_optimal_threads_count(int algo, bool double_hash, int max_cpu_usage) {
+    if (cpu_info.total_logical_cpus == 1) {
+        return 1;
+    }
+
+    int cache = cpu_info.l3_cache ? cpu_info.l3_cache : cpu_info.l2_cache;
+    int count = 0;
+    const int size = (algo ? 1024 : 2048) * (double_hash ? 2 : 1);
+
+    if (cache) {
+        count = cache / size;
+    }
+    else {
+        count = cpu_info.total_logical_cpus / 2;
+    }
+
+    if (count > cpu_info.total_logical_cpus) {
+        count = cpu_info.total_logical_cpus;
+    }
+
+    if (((float) count / cpu_info.total_logical_cpus * 100) > max_cpu_usage) {
+        count = ceil((float) cpu_info.total_logical_cpus * (max_cpu_usage / 100.0));
+    }
+
+    return count < 1 ? 1 : count;
+}

cpu.h

@@ -4,8 +4,9 @@
  * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
  * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
  * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
- * Copyright 2016-2017 XMRig       <support@xmrig.com>
- *
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018-2019 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2019 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
  *
  *   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
@@ -21,41 +22,34 @@
  *   along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-#ifndef __CRYPTONIGHT_H__
-#define __CRYPTONIGHT_H__
+#ifndef XMRIG_CPU_H
+#define XMRIG_CPU_H
+
+#include <stdbool.h>
+
+struct cpu_info {
+    int total_cores;
+    int total_logical_cpus;
+    int flags;
+    int sockets;
+    int l2_cache;
+    int l3_cache;
+    char brand[64];
+    int assembly;
+};
+
+extern struct cpu_info cpu_info;
 
 
-#include <stddef.h>
-#include <stdint.h>
-
-
-#include "align.h"
-
-
-#define MEMORY      2097152 /* 2 MiB */
-#define MEMORY_LITE 1048576 /* 1 MiB */
-
-
-struct cryptonight_ctx {
-    VAR_ALIGN(16, uint8_t state0[200]);
-    VAR_ALIGN(16, uint8_t state1[200]);
-    VAR_ALIGN(16, uint8_t* memory);
+enum cpu_flags {
+    CPU_FLAG_X86_64 = 1,
+    CPU_FLAG_AES    = 2,
+    CPU_FLAG_BMI2   = 4
 };
 
 
-class Job;
-class JobResult;
+void cpu_init();
+int get_optimal_threads_count(int algo, bool double_hash, int max_cpu_usage);
+int affine_to_cpu_mask(int id, unsigned long mask);
 
-
-class CryptoNight
-{
-public:
-    static bool hash(const Job &job, JobResult &result, cryptonight_ctx *ctx);
-    static bool init(int algo, int variant);
-    static void hash(const uint8_t *input, size_t size, uint8_t *output, cryptonight_ctx *ctx);
-
-private:
-    static bool selfTest(int algo);
-};
-
-#endif /* __CRYPTONIGHT_H__ */
+#endif /* XMRIG_CPU_H */

cpu_stub.c

@@ -21,20 +21,14 @@
  *   along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-
-#ifdef _MSC_VER
-#   include <intrin.h>
-
-#   define bit_AES  (1 << 25)
-#   define bit_BMI2 (1 << 8)
-#else
-#   include <cpuid.h>
-#endif
-
+#include <cpuid.h>
 #include <string.h>
+#include <stdbool.h>
+#include <stdint.h>
 
 
-#include "Cpu.h"
+#include "cpu.h"
+#include "options.h"
 
 
 #define VENDOR_ID                  (0)
@@ -51,11 +45,6 @@
 #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);
@@ -65,11 +54,10 @@ static inline void cpuid(int level, int output[4]) {
     output[2] = c;
     output[3] = d;
 }
-#endif
 
 
-static inline void cpu_brand_string(char* s) {
-    int cpu_info[4] = { 0 };
+static void cpu_brand_string(char* s) {
+    int32_t cpu_info[4] = { 0 };
     cpuid(VENDOR_ID, cpu_info);
 
     if (cpu_info[EAX_Reg] >= 4) {
@@ -82,52 +70,60 @@ static inline void cpu_brand_string(char* s) {
 }
 
 
-static inline bool has_aes_ni()
+static bool has_aes_ni()
 {
-    int cpu_info[4] = { 0 };
+    int32_t cpu_info[4] = { 0 };
     cpuid(PROCESSOR_INFO, cpu_info);
 
     return cpu_info[ECX_Reg] & bit_AES;
 }
 
 
-static inline bool has_bmi2() {
-    int cpu_info[4] = { 0 };
+static bool has_bmi2() {
+    int32_t cpu_info[4] = { 0 };
     cpuid(EXTENDED_FEATURES, cpu_info);
 
     return cpu_info[EBX_Reg] & bit_BMI2;
 }
 
 
-char Cpu::m_brand[64]   = { 0 };
-int Cpu::m_flags        = 0;
-int Cpu::m_l2_cache     = 0;
-int Cpu::m_l3_cache     = 0;
-int Cpu::m_sockets      = 1;
-int Cpu::m_totalCores   = 0;
-int Cpu::m_totalThreads = 0;
+void cpu_init_common() {
+    cpu_info.sockets = 1;
+    cpu_brand_string(cpu_info.brand);
 
-
-int Cpu::optimalThreadsCount(int algo, bool doubleHash, int maxCpuUsage)
-{
-    int count = m_totalThreads / 2;
-    return count < 1 ? 1 : count;
-}
-
-
-void Cpu::initCommon()
-{
-    cpu_brand_string(m_brand);
-
-#   if defined(__x86_64__) || defined(_M_AMD64)
-    m_flags |= X86_64;
+#   ifdef __x86_64__
+    cpu_info.flags |= CPU_FLAG_X86_64;
 #   endif
 
     if (has_aes_ni()) {
-        m_flags |= AES;
+        cpu_info.flags |= CPU_FLAG_AES;
+
+#       ifndef XMRIG_NO_ASM
+        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) {
+            cpu_info.assembly = ASM_INTEL;
+        }
+        else if (memcmp(vendor, "AuthenticAMD", 12) == 0) {
+            cpu_info.assembly = ASM_RYZEN;
+        }
+#       endif
     }
 
     if (has_bmi2()) {
-        m_flags |= BMI2;
+        cpu_info.flags |= CPU_FLAG_BMI2;
     }
 }
+
+
+int get_optimal_threads_count(int algo, bool double_hash, int max_cpu_usage) {
+    int count = cpu_info.total_logical_cpus / 2;
+    return count < 1 ? 1 : count;
+}

crypto/CryptonightR_gen.c

@@ -0,0 +1,146 @@
+/* XMRig
+ * Copyright 2010      Jeff Garzik <jgarzik@pobox.com>
+ * Copyright 2012-2014 pooler      <pooler@litecoinpool.org>
+ * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
+ * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
+ * Copyright 2016      Jay D Dee   <jayddee246@gmail.com>
+ * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
+ * Copyright 2018      Lee Clagett <https://github.com/vtnerd>
+ * Copyright 2018-2019 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2019 XMRig       <https://github.com/xmrig>, <support@xmrig.com>
+ *
+ *   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 <http://www.gnu.org/licenses/>.
+ */
+
+#include <string.h>
+
+#include "algo/cryptonight/cryptonight_monero.h"
+#include "crypto/asm/CryptonightR_template.h"
+#include "persistent_memory.h"
+
+
+static inline void add_code(uint8_t **p, void (*p1)(), void (*p2)())
+{
+    const ptrdiff_t size = (const uint8_t*)(p2) - (const uint8_t*)(p1);
+    if (size > 0) {
+        memcpy(*p, (const void *) p1, size);
+        *p += size;
+    }
+}
+
+
+static inline void add_random_math(uint8_t **p, const struct V4_Instruction* code, int code_size, const void_func* instructions, const void_func* instructions_mov, bool is_64_bit, enum Assembly ASM)
+{
+    uint32_t prev_rot_src = (uint32_t)(-1);
+
+    for (int i = 0;; ++i) {
+        const struct V4_Instruction inst = code[i];
+        if (inst.opcode == RET) {
+            break;
+        }
+
+        uint8_t opcode = (inst.opcode == MUL) ? inst.opcode : (inst.opcode + 2);
+        uint8_t dst_index = inst.dst_index;
+        uint8_t src_index = inst.src_index;
+
+        const uint32_t a = inst.dst_index;
+        const uint32_t b = inst.src_index;
+        const uint8_t c = opcode | (dst_index << V4_OPCODE_BITS) | (((src_index == 8) ? dst_index : src_index) << (V4_OPCODE_BITS + V4_DST_INDEX_BITS));
+
+        switch (inst.opcode) {
+        case ROR:
+        case ROL:
+            if (b != prev_rot_src) {
+                prev_rot_src = b;
+                add_code(p, instructions_mov[c], instructions_mov[c + 1]);
+            }
+            break;
+        }
+
+        if (a == prev_rot_src) {
+            prev_rot_src = (uint32_t)(-1);
+        }
+
+        void_func begin = instructions[c];
+
+        if ((ASM = ASM_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 = (uint8_t*) begin;
+
+            if (*prefix == 0x49) {
+                **p = 0x41;
+                *p += 1;
+            }
+
+            begin = (void_func)(prefix + 1);
+        }
+
+        add_code(p, begin, instructions[c + 1]);
+
+        if (inst.opcode == ADD) {
+            *(uint32_t*)(*p - sizeof(uint32_t) - (is_64_bit ? 3 : 0)) = inst.C;
+            if (is_64_bit) {
+                prev_rot_src = (uint32_t)(-1);
+            }
+        }
+    }
+}
+
+
+void v4_compile_code(const struct V4_Instruction* code, int code_size, void* machine_code, enum Assembly ASM)
+{
+    uint8_t* p0 = machine_code;
+    uint8_t* p  = p0;
+
+    add_code(&p, CryptonightR_template_part1, CryptonightR_template_part2);
+    add_random_math(&p, code, code_size, instructions, instructions_mov, false, ASM);
+    add_code(&p, CryptonightR_template_part2, CryptonightR_template_part3);
+    *(int*)(p - 4) = (int)((((const uint8_t*)CryptonightR_template_mainloop) - ((const uint8_t*)CryptonightR_template_part1)) - (p - p0));
+    add_code(&p, CryptonightR_template_part3, CryptonightR_template_end);
+
+    flush_instruction_cache(machine_code, p - p0);
+}
+
+
+void v4_compile_code_double(const struct V4_Instruction* code, int code_size, void* machine_code, enum Assembly ASM)
+{
+    uint8_t* p0 = (uint8_t*) machine_code;
+    uint8_t* p = p0;
+
+    add_code(&p, CryptonightR_template_double_part1, CryptonightR_template_double_part2);
+    add_random_math(&p, code, code_size, instructions, instructions_mov, false, ASM);
+    add_code(&p, CryptonightR_template_double_part2, CryptonightR_template_double_part3);
+    add_random_math(&p, code, code_size, instructions, instructions_mov, false, ASM);
+    add_code(&p, CryptonightR_template_double_part3, CryptonightR_template_double_part4);
+    *(int*)(p - 4) = (int)((((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);
+
+    flush_instruction_cache(machine_code, p - p0);
+}
+
+
+void v4_soft_aes_compile_code(const struct V4_Instruction* code, int code_size, void* machine_code, enum Assembly ASM)
+{
+    uint8_t* p0 = machine_code;
+    uint8_t* p  = p0;
+
+    add_code(&p, CryptonightR_soft_aes_template_part1, CryptonightR_soft_aes_template_part2);
+    add_random_math(&p, code, code_size, instructions, instructions_mov, false, ASM);
+    add_code(&p, CryptonightR_soft_aes_template_part2, CryptonightR_soft_aes_template_part3);
+    *(int*)(p - 4) = (int)((((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);
+
+    flush_instruction_cache(machine_code, p - p0);
+}

crypto/asm/CryptonightR_soft_aes_template.inc

@@ -0,0 +1,279 @@
+PUBLIC FN_PREFIX(CryptonightR_soft_aes_template_part1)
+PUBLIC FN_PREFIX(CryptonightR_soft_aes_template_mainloop)
+PUBLIC FN_PREFIX(CryptonightR_soft_aes_template_part2)
+PUBLIC FN_PREFIX(CryptonightR_soft_aes_template_part3)
+PUBLIC FN_PREFIX(CryptonightR_soft_aes_template_end)
+
+ALIGN(64)
+FN_PREFIX(CryptonightR_soft_aes_template_part1):
+	mov	QWORD PTR [rsp+8], rcx
+	push	rbx
+	push	rbp
+	push	rsi
+	push	rdi
+	push	r12
+	push	r13
+	push	r14
+	push	r15
+	sub	rsp, 232
+
+	mov	eax, [rcx+96]
+	mov	ebx, [rcx+100]
+	mov	esi, [rcx+104]
+	mov	edx, [rcx+108]
+	mov [rsp+144], eax
+	mov [rsp+148], ebx
+	mov [rsp+152], esi
+	mov [rsp+156], edx
+
+	mov	rax, QWORD PTR [rcx+48]
+	mov	r10, rcx
+	xor	rax, QWORD PTR [rcx+16]
+	mov	r8, QWORD PTR [rcx+32]
+	xor	r8, QWORD PTR [rcx]
+	mov	r9, QWORD PTR [rcx+40]
+	xor	r9, QWORD PTR [rcx+8]
+	movq	xmm4, rax
+	mov	rdx, QWORD PTR [rcx+56]
+	xor	rdx, QWORD PTR [rcx+24]
+	mov	r11, QWORD PTR [rcx+224]
+	mov	rcx, QWORD PTR [rcx+88]
+	xor	rcx, QWORD PTR [r10+72]
+	mov	rax, QWORD PTR [r10+80]
+	movq	xmm0, rdx
+	xor	rax, QWORD PTR [r10+64]
+
+	movaps	XMMWORD PTR [rsp+16], xmm6
+	movaps	XMMWORD PTR [rsp+32], xmm7
+	movaps	XMMWORD PTR [rsp+48], xmm8
+	movaps	XMMWORD PTR [rsp+64], xmm9
+	movaps	XMMWORD PTR [rsp+80], xmm10
+	movaps	XMMWORD PTR [rsp+96], xmm11
+	movaps	XMMWORD PTR [rsp+112], xmm12
+	movaps	XMMWORD PTR [rsp+128], xmm13
+
+	movq	xmm5, rax
+
+	mov	rax, r8
+	punpcklqdq xmm4, xmm0
+	and	eax, 2097136
+	movq	xmm10, QWORD PTR [r10+96]
+	movq	xmm0, rcx
+	mov	rcx, QWORD PTR [r10+104]
+	xorps	xmm9, xmm9
+	mov	QWORD PTR [rsp+328], rax
+	movq	xmm12, r11
+	mov	QWORD PTR [rsp+320], r9
+	punpcklqdq xmm5, xmm0
+	movq xmm13, rcx
+	mov r12d, 524288
+
+	ALIGN(64)
+FN_PREFIX(CryptonightR_soft_aes_template_mainloop):
+	movd xmm11, r12d
+	mov	r12, QWORD PTR [r10+272]
+	lea	r13, QWORD PTR [rax+r11]
+	mov	esi, DWORD PTR [r13]
+	movq	xmm0, r9
+	mov	r10d, DWORD PTR [r13+4]
+	movq	xmm7, r8
+	mov	ebp, DWORD PTR [r13+12]
+	mov	r14d, DWORD PTR [r13+8]
+	mov	rdx, QWORD PTR [rsp+328]
+	movzx	ecx, sil
+	shr	esi, 8
+	punpcklqdq xmm7, xmm0
+	mov	r15d, DWORD PTR [r12+rcx*4]
+	movzx	ecx, r10b
+	shr	r10d, 8
+	mov	edi, DWORD PTR [r12+rcx*4]
+	movzx	ecx, r14b
+	shr	r14d, 8
+	mov	ebx, DWORD PTR [r12+rcx*4]
+	movzx	ecx, bpl
+	shr	ebp, 8
+	mov	r9d, DWORD PTR [r12+rcx*4]
+	movzx	ecx, r10b
+	shr	r10d, 8
+	xor	r15d, DWORD PTR [r12+rcx*4+1024]
+	movzx	ecx, r14b
+	shr	r14d, 8
+	mov	eax, r14d
+	shr	eax, 8
+	xor	edi, DWORD PTR [r12+rcx*4+1024]
+	add	eax, 256
+	movzx	ecx, bpl
+	shr	ebp, 8
+	xor	ebx, DWORD PTR [r12+rcx*4+1024]
+	movzx	ecx, sil
+	shr	esi, 8
+	xor	r9d, DWORD PTR [r12+rcx*4+1024]
+	add	r12, 2048
+	movzx	ecx, r10b
+	shr	r10d, 8
+	add	r10d, 256
+	mov	r11d, DWORD PTR [r12+rax*4]
+	xor	r11d, DWORD PTR [r12+rcx*4]
+	xor	r11d, r9d
+	movzx	ecx, sil
+	mov	r10d, DWORD PTR [r12+r10*4]
+	shr	esi, 8
+	add	esi, 256
+	xor	r10d, DWORD PTR [r12+rcx*4]
+	movzx	ecx, bpl
+	xor	r10d, ebx
+	shr	ebp, 8
+	movd	xmm1, r11d
+	add	ebp, 256
+	movq	r11, xmm12
+	mov	r9d, DWORD PTR [r12+rcx*4]
+	xor	r9d, DWORD PTR [r12+rsi*4]
+	mov	eax, DWORD PTR [r12+rbp*4]
+	xor	r9d, edi
+	movzx	ecx, r14b
+	movd	xmm0, r10d
+	movd	xmm2, r9d
+	xor	eax, DWORD PTR [r12+rcx*4]
+	mov	rcx, rdx
+	xor	eax, r15d
+	punpckldq xmm2, xmm1
+	xor	rcx, 16
+	movd	xmm6, eax
+	mov	rax, rdx
+	punpckldq xmm6, xmm0
+	xor	rax, 32
+	punpckldq xmm6, xmm2
+	xor	rdx, 48
+	movdqu	xmm2, XMMWORD PTR [rcx+r11]
+	pxor xmm6, xmm2
+	pxor	xmm6, xmm7
+	paddq	xmm2, xmm4
+	movdqu	xmm1, XMMWORD PTR [rax+r11]
+	movdqu	xmm0, XMMWORD PTR [rdx+r11]
+	pxor xmm6, xmm1
+	pxor xmm6, xmm0
+	paddq	xmm0, xmm5
+	movdqu	XMMWORD PTR [rcx+r11], xmm0
+	movdqu	XMMWORD PTR [rax+r11], xmm2
+	movq rcx, xmm13
+	paddq	xmm1, xmm7
+	movdqu	XMMWORD PTR [rdx+r11], xmm1
+	movq	rdi, xmm6
+	mov	r10, rdi
+	and	r10d, 2097136
+	movdqa	xmm0, xmm6
+	pxor	xmm0, xmm4
+	movdqu	XMMWORD PTR [r13], xmm0
+
+	mov ebx, [rsp+144]
+	mov ebp, [rsp+152]
+	add ebx, [rsp+148]
+	add ebp, [rsp+156]
+	shl rbp, 32
+	or rbx, rbp
+
+	xor rbx, QWORD PTR [r10+r11]
+	lea	r14, QWORD PTR [r10+r11]
+	mov	rbp, QWORD PTR [r14+8]
+
+	mov [rsp+160], rbx
+	mov [rsp+168], rdi
+	mov [rsp+176], rbp
+	mov [rsp+184], r10
+	mov r10, rsp
+
+	mov ebx, [rsp+144]
+	mov esi, [rsp+148]
+	mov edi, [rsp+152]
+	mov ebp, [rsp+156]
+
+	movd esp, xmm7
+	movaps xmm0, xmm7
+	psrldq xmm0, 8
+	movd r15d, xmm0
+	movd eax, xmm4
+	movd edx, xmm5
+	movaps xmm0, xmm5
+	psrldq xmm0, 8
+	movd r9d, xmm0
+
+FN_PREFIX(CryptonightR_soft_aes_template_part2):
+	mov rsp, r10
+	mov [rsp+144], ebx
+	mov [rsp+148], esi
+	mov [rsp+152], edi
+	mov [rsp+156], ebp
+
+	mov edi, edi
+	shl rbp, 32
+	or rbp, rdi
+	xor r8, rbp
+
+	mov ebx, ebx
+	shl rsi, 32
+	or rsi, rbx
+	xor QWORD PTR [rsp+320], rsi
+
+	mov rbx, [rsp+160]
+	mov rdi, [rsp+168]
+	mov rbp, [rsp+176]
+	mov r10, [rsp+184]
+
+	mov	r9, r10
+	xor	r9, 16
+	mov	rcx, r10
+	xor	rcx, 32
+	xor	r10, 48
+	mov	rax, rbx
+	mul	rdi
+	movdqu	xmm2, XMMWORD PTR [r9+r11]
+	movdqu	xmm1, XMMWORD PTR [rcx+r11]
+	pxor xmm6, xmm2
+	pxor xmm6, xmm1
+	paddq	xmm1, xmm7
+	add	r8, rdx
+	movdqu	xmm0, XMMWORD PTR [r10+r11]
+	pxor xmm6, xmm0
+	paddq	xmm0, xmm5
+	paddq	xmm2, xmm4
+	movdqu	XMMWORD PTR [r9+r11], xmm0
+	movdqa	xmm5, xmm4
+	mov	r9, QWORD PTR [rsp+320]
+	movdqa	xmm4, xmm6
+	add	r9, rax
+	movdqu	XMMWORD PTR [rcx+r11], xmm2
+	movdqu	XMMWORD PTR [r10+r11], xmm1
+	mov	r10, QWORD PTR [rsp+304]
+	movd r12d, xmm11
+	mov	QWORD PTR [r14], r8
+	xor	r8, rbx
+	mov	rax, r8
+	mov	QWORD PTR [r14+8], r9
+	and	eax, 2097136
+	xor	r9, rbp
+	mov	QWORD PTR [rsp+320], r9
+	mov	QWORD PTR [rsp+328], rax
+	sub	r12d, 1
+	jne	FN_PREFIX(CryptonightR_soft_aes_template_mainloop)
+
+FN_PREFIX(CryptonightR_soft_aes_template_part3):
+	movaps	xmm6, XMMWORD PTR [rsp+16]
+	movaps	xmm7, XMMWORD PTR [rsp+32]
+	movaps	xmm8, XMMWORD PTR [rsp+48]
+	movaps	xmm9, XMMWORD PTR [rsp+64]
+	movaps	xmm10, XMMWORD PTR [rsp+80]
+	movaps	xmm11, XMMWORD PTR [rsp+96]
+	movaps	xmm12, XMMWORD PTR [rsp+112]
+	movaps	xmm13, XMMWORD PTR [rsp+128]
+
+	add	rsp, 232
+	pop	r15
+	pop	r14
+	pop	r13
+	pop	r12
+	pop	rdi
+	pop	rsi
+	pop	rbp
+	pop	rbx
+	ret
+FN_PREFIX(CryptonightR_soft_aes_template_end):

crypto/asm/CryptonightR_template.inc

@@ -0,0 +1,531 @@
+PUBLIC FN_PREFIX(CryptonightR_template_part1)
+PUBLIC FN_PREFIX(CryptonightR_template_mainloop)
+PUBLIC FN_PREFIX(CryptonightR_template_part2)
+PUBLIC FN_PREFIX(CryptonightR_template_part3)
+PUBLIC FN_PREFIX(CryptonightR_template_end)
+PUBLIC FN_PREFIX(CryptonightR_template_double_part1)
+PUBLIC FN_PREFIX(CryptonightR_template_double_mainloop)
+PUBLIC FN_PREFIX(CryptonightR_template_double_part2)
+PUBLIC FN_PREFIX(CryptonightR_template_double_part3)
+PUBLIC FN_PREFIX(CryptonightR_template_double_part4)
+PUBLIC FN_PREFIX(CryptonightR_template_double_end)
+
+ALIGN(64)
+FN_PREFIX(CryptonightR_template_part1):
+	mov	QWORD PTR [rsp+16], rbx
+	mov	QWORD PTR [rsp+24], rbp
+	mov	QWORD PTR [rsp+32], rsi
+	push	r10
+	push	r11
+	push	r12
+	push	r13
+	push	r14
+	push	r15
+	push	rdi
+	sub	rsp, 64
+	mov	r12, rcx
+	mov	r8, QWORD PTR [r12+32]
+	mov	rdx, r12
+	xor	r8, QWORD PTR [r12]
+	mov	r15, QWORD PTR [r12+40]
+	mov	r9, r8
+	xor	r15, QWORD PTR [r12+8]
+	mov	r11, QWORD PTR [r12+224]
+	mov	r12, QWORD PTR [r12+56]
+	xor	r12, QWORD PTR [rdx+24]
+	mov	rax, QWORD PTR [rdx+48]
+	xor	rax, QWORD PTR [rdx+16]
+	movaps	XMMWORD PTR [rsp+48], xmm6
+	movq	xmm0, r12
+	movaps	XMMWORD PTR [rsp+32], xmm7
+	movaps	XMMWORD PTR [rsp+16], xmm8
+	movaps	XMMWORD PTR [rsp], xmm9
+	mov	r12, QWORD PTR [rdx+88]
+	xor	r12, QWORD PTR [rdx+72]
+	movq	xmm6, rax
+	mov	rax, QWORD PTR [rdx+80]
+	xor	rax, QWORD PTR [rdx+64]
+	punpcklqdq xmm6, xmm0
+	and	r9d, 2097136
+	movq	xmm0, r12
+	movq	xmm7, rax
+	punpcklqdq xmm7, xmm0
+	mov r10d, r9d
+	movq	xmm9, rsp
+	mov rsp, r8
+	mov	r8d, 524288
+
+	mov	ebx, [rdx+96]
+	mov	esi, [rdx+100]
+	mov	edi, [rdx+104]
+	mov	ebp, [rdx+108]
+
+	ALIGN(64)
+FN_PREFIX(CryptonightR_template_mainloop):
+	movdqa	xmm5, XMMWORD PTR [r9+r11]
+	movq	xmm0, r15
+	movq	xmm4, rsp
+	punpcklqdq xmm4, xmm0
+	lea	rdx, QWORD PTR [r9+r11]
+
+	aesenc	xmm5, xmm4
+
+	mov	r13d, r9d
+	mov	eax, r9d
+	xor	r9d, 48
+	xor	r13d, 16
+	xor	eax, 32
+	movdqu	xmm0, XMMWORD PTR [r9+r11]
+	movaps xmm3, xmm0
+	movdqu	xmm2, XMMWORD PTR [r13+r11]
+	movdqu	xmm1, XMMWORD PTR [rax+r11]
+	pxor xmm0, xmm2
+	pxor xmm5, xmm1
+	pxor xmm5, xmm0
+
+	movq	r12, xmm5
+	movd	r10d, xmm5
+	and	r10d, 2097136
+
+	paddq	xmm3, xmm7
+	paddq	xmm2, xmm6
+	paddq	xmm1, xmm4
+	movdqu	XMMWORD PTR [r13+r11], xmm3
+	movdqu	XMMWORD PTR [rax+r11], xmm2
+	movdqu	XMMWORD PTR [r9+r11], xmm1
+
+	movdqa	xmm0, xmm5
+	pxor	xmm0, xmm6
+	movdqu	XMMWORD PTR [rdx], xmm0
+
+	lea	r13d, [ebx+esi]
+	lea	edx, [edi+ebp]
+	shl rdx, 32
+	or	r13, rdx
+
+	movd eax, xmm6
+	movd edx, xmm7
+	pextrd r9d, xmm7, 2
+
+	xor	r13, QWORD PTR [r10+r11]
+	mov	r14, QWORD PTR [r10+r11+8]
+
+FN_PREFIX(CryptonightR_template_part2):
+	lea	rcx, [r10+r11]
+
+	mov eax, edi
+	mov edx, ebp
+	shl rdx, 32
+	or rax, rdx
+	xor rsp, rax
+
+	mov eax, ebx
+	mov edx, esi
+	shl rdx, 32
+	or rax, rdx
+	xor r15, rax
+
+	mov	rax, r13
+	mul	r12
+	add	r15, rax
+	add	rsp, rdx
+
+	mov	r9d, r10d
+	mov	r12d, r10d
+	xor	r9d, 16
+	xor	r12d, 32
+	xor	r10d, 48
+	movdqa	xmm1, XMMWORD PTR [r12+r11]
+	movaps xmm3, xmm1
+	movdqa	xmm2, XMMWORD PTR [r9+r11]
+	movdqa	xmm0, XMMWORD PTR [r10+r11]
+	pxor xmm1, xmm2
+	pxor xmm5, xmm0
+	pxor xmm5, xmm1
+	paddq	xmm3, xmm4
+	paddq	xmm2, xmm6
+	paddq	xmm0, xmm7
+	movdqu	XMMWORD PTR [r9+r11], xmm0
+	movdqu	XMMWORD PTR [r12+r11], xmm2
+	movdqu	XMMWORD PTR [r10+r11], xmm3
+
+	movdqa	xmm7, xmm6
+	mov	QWORD PTR [rcx], rsp
+	xor	rsp, r13
+	mov	r9d, esp
+	mov	QWORD PTR [rcx+8], r15
+	and	r9d, 2097136
+	xor	r15, r14
+	movdqa	xmm6, xmm5
+	dec	r8d
+	jnz	FN_PREFIX(CryptonightR_template_mainloop)
+
+FN_PREFIX(CryptonightR_template_part3):
+	movq	rsp, xmm9
+
+	mov	rbx, QWORD PTR [rsp+136]
+	mov	rbp, QWORD PTR [rsp+144]
+	mov	rsi, QWORD PTR [rsp+152]
+	movaps	xmm6, XMMWORD PTR [rsp+48]
+	movaps	xmm7, XMMWORD PTR [rsp+32]
+	movaps	xmm8, XMMWORD PTR [rsp+16]
+	movaps	xmm9, XMMWORD PTR [rsp]
+	add	rsp, 64
+	pop	rdi
+	pop	r15
+	pop	r14
+	pop	r13
+	pop	r12
+	pop	r11
+	pop	r10
+	ret	0
+FN_PREFIX(CryptonightR_template_end):
+
+ALIGN(64)
+FN_PREFIX(CryptonightR_template_double_part1):
+	mov	QWORD PTR [rsp+24], rbx
+	push	rbp
+	push	rsi
+	push	rdi
+	push	r12
+	push	r13
+	push	r14
+	push	r15
+	sub	rsp, 320
+	mov	r14, QWORD PTR [rcx+32]
+	mov	r8, rcx
+	xor	r14, QWORD PTR [rcx]
+	mov	r12, QWORD PTR [rcx+40]
+	mov	ebx, r14d
+	mov	rsi, QWORD PTR [rcx+224]
+	and	ebx, 2097136
+	xor	r12, QWORD PTR [rcx+8]
+	mov	rcx, QWORD PTR [rcx+56]
+	xor	rcx, QWORD PTR [r8+24]
+	mov	rax, QWORD PTR [r8+48]
+	xor	rax, QWORD PTR [r8+16]
+	mov	r15, QWORD PTR [rdx+32]
+	xor	r15, QWORD PTR [rdx]
+	movq	xmm0, rcx
+	mov	rcx, QWORD PTR [r8+88]
+	xor	rcx, QWORD PTR [r8+72]
+	mov	r13, QWORD PTR [rdx+40]
+	mov	rdi, QWORD PTR [rdx+224]
+	xor	r13, QWORD PTR [rdx+8]
+	movaps	XMMWORD PTR [rsp+160], xmm6
+	movaps	XMMWORD PTR [rsp+176], xmm7
+	movaps	XMMWORD PTR [rsp+192], xmm8
+	movaps	XMMWORD PTR [rsp+208], xmm9
+	movaps	XMMWORD PTR [rsp+224], xmm10
+	movaps	XMMWORD PTR [rsp+240], xmm11
+	movaps	XMMWORD PTR [rsp+256], xmm12
+	movaps	XMMWORD PTR [rsp+272], xmm13
+	movaps	XMMWORD PTR [rsp+288], xmm14
+	movaps	XMMWORD PTR [rsp+304], xmm15
+	movq	xmm7, rax
+	mov	rax, QWORD PTR [r8+80]
+	xor	rax, QWORD PTR [r8+64]
+
+	movaps xmm1, XMMWORD PTR [rdx+96]
+	movaps xmm2, XMMWORD PTR [r8+96]
+	movaps XMMWORD PTR [rsp], xmm1
+	movaps XMMWORD PTR [rsp+16], xmm2
+
+	mov	r8d, r15d
+	punpcklqdq xmm7, xmm0
+	movq	xmm0, rcx
+	mov	rcx, QWORD PTR [rdx+56]
+	xor	rcx, QWORD PTR [rdx+24]
+	movq	xmm9, rax
+	mov	QWORD PTR [rsp+128], rsi
+	mov	rax, QWORD PTR [rdx+48]
+	xor	rax, QWORD PTR [rdx+16]
+	punpcklqdq xmm9, xmm0
+	movq	xmm0, rcx
+	mov	rcx, QWORD PTR [rdx+88]
+	xor	rcx, QWORD PTR [rdx+72]
+	movq	xmm8, rax
+	mov	QWORD PTR [rsp+136], rdi
+	mov	rax, QWORD PTR [rdx+80]
+	xor	rax, QWORD PTR [rdx+64]
+	punpcklqdq xmm8, xmm0
+	and	r8d, 2097136
+	movq	xmm0, rcx
+	mov	r11d, 524288
+	movq	xmm10, rax
+	punpcklqdq xmm10, xmm0
+	
+	movq xmm14, QWORD PTR [rsp+128]
+	movq xmm15, QWORD PTR [rsp+136]
+
+	ALIGN(64)
+FN_PREFIX(CryptonightR_template_double_mainloop):
+	movdqu	xmm6, XMMWORD PTR [rbx+rsi]
+	movq	xmm0, r12
+	mov	ecx, ebx
+	movq	xmm3, r14
+	punpcklqdq xmm3, xmm0
+	xor	ebx, 16
+	aesenc	xmm6, xmm3
+	movq	xmm4, r15
+	movdqu	xmm0, XMMWORD PTR [rbx+rsi]
+	pxor	xmm6, xmm0
+	xor	ebx, 48
+	paddq	xmm0, xmm7
+	movdqu	xmm1, XMMWORD PTR [rbx+rsi]
+	pxor	xmm6, xmm1
+	movdqu	XMMWORD PTR [rbx+rsi], xmm0
+	paddq	xmm1, xmm3
+	xor	ebx, 16
+	mov	eax, ebx
+	xor	rax, 32
+	movdqu	xmm0, XMMWORD PTR [rbx+rsi]
+	pxor	xmm6, xmm0
+	movq	rdx, xmm6
+	movdqu	XMMWORD PTR [rbx+rsi], xmm1
+	paddq	xmm0, xmm9
+	movdqu	XMMWORD PTR [rax+rsi], xmm0
+	movdqa	xmm0, xmm6
+	pxor	xmm0, xmm7
+	movdqu	XMMWORD PTR [rcx+rsi], xmm0
+	mov	esi, edx
+	movdqu	xmm5, XMMWORD PTR [r8+rdi]
+	and	esi, 2097136
+	mov	ecx, r8d
+	movq	xmm0, r13
+	punpcklqdq xmm4, xmm0
+	xor	r8d, 16
+	aesenc	xmm5, xmm4
+	movdqu	xmm0, XMMWORD PTR [r8+rdi]
+	pxor	xmm5, xmm0
+	xor	r8d, 48
+	paddq	xmm0, xmm8
+	movdqu	xmm1, XMMWORD PTR [r8+rdi]
+	pxor	xmm5, xmm1
+	movdqu	XMMWORD PTR [r8+rdi], xmm0
+	paddq	xmm1, xmm4
+	xor	r8d, 16
+	mov	eax, r8d
+	xor	rax, 32
+	movdqu	xmm0, XMMWORD PTR [r8+rdi]
+	pxor	xmm5, xmm0
+	movdqu	XMMWORD PTR [r8+rdi], xmm1
+	paddq	xmm0, xmm10
+	movdqu	XMMWORD PTR [rax+rdi], xmm0
+	movdqa	xmm0, xmm5
+	pxor	xmm0, xmm8
+	movdqu	XMMWORD PTR [rcx+rdi], xmm0
+	movq	rdi, xmm5
+	movq	rcx, xmm14
+	mov	ebp, edi
+	mov	r8, QWORD PTR [rcx+rsi]
+	mov	r10, QWORD PTR [rcx+rsi+8]
+	lea	r9, QWORD PTR [rcx+rsi]
+	xor	esi, 16
+
+	movq xmm0, rsp
+	movq xmm1, rsi
+	movq xmm2, rdi
+	movq xmm11, rbp
+	movq xmm12, r15
+	movq xmm13, rdx
+	mov [rsp+104], rcx
+	mov [rsp+112], r9
+
+	mov ebx, DWORD PTR [rsp+16]
+	mov esi, DWORD PTR [rsp+20]
+	mov edi, DWORD PTR [rsp+24]
+	mov ebp, DWORD PTR [rsp+28]
+
+	lea	eax, [ebx+esi]
+	lea	edx, [edi+ebp]
+	shl rdx, 32
+	or	rax, rdx
+	xor r8, rax
+
+	movd esp, xmm3
+	pextrd r15d, xmm3, 2
+	movd eax, xmm7
+	movd edx, xmm9
+	pextrd r9d, xmm9, 2
+
+FN_PREFIX(CryptonightR_template_double_part2):
+
+	mov eax, edi
+	mov edx, ebp
+	shl rdx, 32
+	or rax, rdx
+	xor r14, rax
+
+	mov eax, ebx
+	mov edx, esi
+	shl rdx, 32
+	or rax, rdx
+	xor r12, rax
+
+	movq rsp, xmm0
+	mov DWORD PTR [rsp+16], ebx
+	mov DWORD PTR [rsp+20], esi
+	mov DWORD PTR [rsp+24], edi
+	mov DWORD PTR [rsp+28], ebp
+
+	movq rsi, xmm1
+	movq rdi, xmm2
+	movq rbp, xmm11
+	movq r15, xmm12
+	movq rdx, xmm13
+	mov rcx, [rsp+104]
+	mov r9, [rsp+112]
+
+	mov rbx, r8
+	mov	rax, r8
+	mul	rdx
+	and	ebp, 2097136
+	mov	r8, rax
+	movdqu	xmm1, XMMWORD PTR [rcx+rsi]
+	pxor	xmm6, xmm1
+	xor	esi, 48
+	paddq	xmm1, xmm7
+	movdqu	xmm2, XMMWORD PTR [rsi+rcx]
+	pxor	xmm6, xmm2
+	paddq	xmm2, xmm3
+	movdqu	XMMWORD PTR [rsi+rcx], xmm1
+	xor	esi, 16
+	mov	eax, esi
+	mov	rsi, rcx
+	movdqu	xmm0, XMMWORD PTR [rax+rcx]
+	pxor	xmm6, xmm0
+	movdqu	XMMWORD PTR [rax+rcx], xmm2
+	paddq	xmm0, xmm9
+	add	r12, r8
+	xor	rax, 32
+	add	r14, rdx
+	movdqa	xmm9, xmm7
+	movdqa	xmm7, xmm6
+	movdqu	XMMWORD PTR [rax+rcx], xmm0
+	mov	QWORD PTR [r9+8], r12
+	xor	r12, r10
+	mov	QWORD PTR [r9], r14
+	movq rcx, xmm15
+	xor	r14, rbx
+	mov	r10d, ebp
+	mov	ebx, r14d
+	xor	ebp, 16
+	and	ebx, 2097136
+	mov	r8, QWORD PTR [r10+rcx]
+	mov	r9, QWORD PTR [r10+rcx+8]
+
+	movq xmm0, rsp
+	movq xmm1, rbx
+	movq xmm2, rsi
+	movq xmm11, rdi
+	movq xmm12, rbp
+	movq xmm13, r15
+	mov [rsp+104], rcx
+	mov [rsp+112], r9
+
+	mov ebx, DWORD PTR [rsp]
+	mov esi, DWORD PTR [rsp+4]
+	mov edi, DWORD PTR [rsp+8]
+	mov ebp, DWORD PTR [rsp+12]
+
+	lea	eax, [ebx+esi]
+	lea	edx, [edi+ebp]
+	shl rdx, 32
+	or	rax, rdx
+
+	xor r8, rax
+	movq xmm3, r8
+
+	movd esp, xmm4
+	pextrd r15d, xmm4, 2
+	movd eax, xmm8
+	movd edx, xmm10
+	pextrd r9d, xmm10, 2
+
+FN_PREFIX(CryptonightR_template_double_part3):
+
+	movq r15, xmm13
+
+	mov eax, edi
+	mov edx, ebp
+	shl rdx, 32
+	or rax, rdx
+	xor r15, rax
+
+	mov eax, ebx
+	mov edx, esi
+	shl rdx, 32
+	or rax, rdx
+	xor r13, rax
+
+	movq rsp, xmm0
+	mov DWORD PTR [rsp], ebx
+	mov DWORD PTR [rsp+4], esi
+	mov DWORD PTR [rsp+8], edi
+	mov DWORD PTR [rsp+12], ebp
+
+	movq rbx, xmm1
+	movq rsi, xmm2
+	movq rdi, xmm11
+	movq rbp, xmm12
+	mov rcx, [rsp+104]
+	mov r9, [rsp+112]
+
+	mov rax, r8
+	mul	rdi
+	mov	rdi, rcx
+	mov	r8, rax
+	movdqu	xmm1, XMMWORD PTR [rbp+rcx]
+	pxor xmm5, xmm1
+	xor	ebp, 48
+	paddq	xmm1, xmm8
+	add	r13, r8
+	movdqu	xmm2, XMMWORD PTR [rbp+rcx]
+	pxor xmm5, xmm2
+	add	r15, rdx
+	movdqu	XMMWORD PTR [rbp+rcx], xmm1
+	paddq	xmm2, xmm4
+	xor	ebp, 16
+	mov	eax, ebp
+	xor	rax, 32
+	movdqu	xmm0, XMMWORD PTR [rbp+rcx]
+	pxor xmm5, xmm0
+	movdqu	XMMWORD PTR [rbp+rcx], xmm2
+	paddq	xmm0, xmm10
+	movdqu	XMMWORD PTR [rax+rcx], xmm0
+	movq rax, xmm3
+	movdqa	xmm10, xmm8
+	mov	QWORD PTR [r10+rcx], r15
+	movdqa	xmm8, xmm5
+	xor	r15, rax
+	mov	QWORD PTR [r10+rcx+8], r13
+	mov	r8d, r15d
+	xor	r13, r9
+	and	r8d, 2097136
+	dec r11d
+	jnz	FN_PREFIX(CryptonightR_template_double_mainloop)
+
+FN_PREFIX(CryptonightR_template_double_part4):
+
+	mov	rbx, QWORD PTR [rsp+400]
+	movaps	xmm6, XMMWORD PTR [rsp+160]
+	movaps	xmm7, XMMWORD PTR [rsp+176]
+	movaps	xmm8, XMMWORD PTR [rsp+192]
+	movaps	xmm9, XMMWORD PTR [rsp+208]
+	movaps	xmm10, XMMWORD PTR [rsp+224]
+	movaps	xmm11, XMMWORD PTR [rsp+240]
+	movaps	xmm12, XMMWORD PTR [rsp+256]
+	movaps	xmm13, XMMWORD PTR [rsp+272]
+	movaps	xmm14, XMMWORD PTR [rsp+288]
+	movaps	xmm15, XMMWORD PTR [rsp+304]
+	add	rsp, 320
+	pop	r15
+	pop	r14
+	pop	r13
+	pop	r12
+	pop	rdi
+	pop	rsi
+	pop	rbp
+	ret	0
+FN_PREFIX(CryptonightR_template_double_end):

crypto/asm/cn2/cnv2_double_main_loop_sandybridge.inc

@@ -0,0 +1,410 @@
+	mov	rax, rsp
+	push	rbx
+	push	rbp
+	push	rsi
+	push	rdi
+	push	r12
+	push	r13
+	push	r14
+	push	r15
+	sub	rsp, 184
+
+	stmxcsr DWORD PTR [rsp+272]
+	mov DWORD PTR [rsp+276], 24448
+	ldmxcsr DWORD PTR [rsp+276]
+
+	mov	r13, QWORD PTR [rcx+224]
+	mov	r9, rdx
+	mov	r10, QWORD PTR [rcx+32]
+	mov	r8, rcx
+	xor	r10, QWORD PTR [rcx]
+	mov	r14d, 524288
+	mov	r11, QWORD PTR [rcx+40]
+	xor	r11, QWORD PTR [rcx+8]
+	mov	rsi, QWORD PTR [rdx+224]
+	mov	rdx, QWORD PTR [rcx+56]
+	xor	rdx, QWORD PTR [rcx+24]
+	mov	rdi, QWORD PTR [r9+32]
+	xor	rdi, QWORD PTR [r9]
+	mov	rbp, QWORD PTR [r9+40]
+	xor	rbp, QWORD PTR [r9+8]
+	movq	xmm0, rdx
+	movaps	XMMWORD PTR [rax-88], xmm6
+	movaps	XMMWORD PTR [rax-104], xmm7
+	movaps	XMMWORD PTR [rax-120], xmm8
+	movaps	XMMWORD PTR [rsp+112], xmm9
+	movaps	XMMWORD PTR [rsp+96], xmm10
+	movaps	XMMWORD PTR [rsp+80], xmm11
+	movaps	XMMWORD PTR [rsp+64], xmm12
+	movaps	XMMWORD PTR [rsp+48], xmm13
+	movaps	XMMWORD PTR [rsp+32], xmm14
+	movaps	XMMWORD PTR [rsp+16], xmm15
+	mov	rdx, r10
+	movq	xmm4, QWORD PTR [r8+96]
+	and	edx, 2097136
+	mov	rax, QWORD PTR [rcx+48]
+	xorps	xmm13, xmm13
+	xor	rax, QWORD PTR [rcx+16]
+	mov	rcx, QWORD PTR [rcx+88]
+	xor	rcx, QWORD PTR [r8+72]
+	movq	xmm5, QWORD PTR [r8+104]
+	movq	xmm7, rax
+
+	mov eax, 1
+	shl rax, 52
+	movq xmm14, rax
+	punpcklqdq xmm14, xmm14
+
+	mov eax, 1023
+	shl rax, 52
+	movq xmm12, rax
+	punpcklqdq xmm12, xmm12
+
+	mov	rax, QWORD PTR [r8+80]
+	xor	rax, QWORD PTR [r8+64]
+	punpcklqdq xmm7, xmm0
+	movq	xmm0, rcx
+	mov	rcx, QWORD PTR [r9+56]
+	xor	rcx, QWORD PTR [r9+24]
+	movq	xmm3, rax
+	mov	rax, QWORD PTR [r9+48]
+	xor	rax, QWORD PTR [r9+16]
+	punpcklqdq xmm3, xmm0
+	movq	xmm0, rcx
+	mov	QWORD PTR [rsp], r13
+	mov	rcx, QWORD PTR [r9+88]
+	xor	rcx, QWORD PTR [r9+72]
+	movq	xmm6, rax
+	mov	rax, QWORD PTR [r9+80]
+	xor	rax, QWORD PTR [r9+64]
+	punpcklqdq xmm6, xmm0
+	movq	xmm0, rcx
+	mov	QWORD PTR [rsp+256], r10
+	mov	rcx, rdi
+	mov	QWORD PTR [rsp+264], r11
+	movq	xmm8, rax
+	and	ecx, 2097136
+	punpcklqdq xmm8, xmm0
+	movq	xmm0, QWORD PTR [r9+96]
+	punpcklqdq xmm4, xmm0
+	movq	xmm0, QWORD PTR [r9+104]
+	lea	r8, QWORD PTR [rcx+rsi]
+	movdqu	xmm11, XMMWORD PTR [r8]
+	punpcklqdq xmm5, xmm0
+	lea	r9, QWORD PTR [rdx+r13]
+	movdqu	xmm15, XMMWORD PTR [r9]
+
+	ALIGN(64)
+main_loop_double_sandybridge:
+	movdqu	xmm9, xmm15
+	mov eax, edx
+	mov ebx, edx
+	xor eax, 16
+	xor ebx, 32
+	xor edx, 48
+
+	movq	xmm0, r11
+	movq	xmm2, r10
+	punpcklqdq xmm2, xmm0
+	aesenc	xmm9, xmm2
+
+	movdqu	xmm0, XMMWORD PTR [rax+r13]
+	movdqu	xmm1, XMMWORD PTR [rbx+r13]
+	paddq	xmm0, xmm7
+	paddq	xmm1, xmm2
+	movdqu	XMMWORD PTR [rbx+r13], xmm0
+	movdqu	xmm0, XMMWORD PTR [rdx+r13]
+	movdqu	XMMWORD PTR [rdx+r13], xmm1
+	paddq	xmm0, xmm3
+	movdqu	XMMWORD PTR [rax+r13], xmm0
+
+	movq	r11, xmm9
+	mov	edx, r11d
+	and	edx, 2097136
+	movdqa	xmm0, xmm9
+	pxor	xmm0, xmm7
+	movdqu	XMMWORD PTR [r9], xmm0
+
+	lea	rbx, QWORD PTR [rdx+r13]
+	mov	r10, QWORD PTR [rdx+r13]
+
+	movdqu	xmm10, xmm11
+	movq	xmm0, rbp
+	movq	xmm11, rdi
+	punpcklqdq xmm11, xmm0
+	aesenc	xmm10, xmm11
+
+	mov eax, ecx
+	mov r12d, ecx
+	xor eax, 16
+	xor r12d, 32
+	xor ecx, 48
+
+	movdqu	xmm0, XMMWORD PTR [rax+rsi]
+	paddq	xmm0, xmm6
+	movdqu	xmm1, XMMWORD PTR [r12+rsi]
+	movdqu	XMMWORD PTR [r12+rsi], xmm0
+	paddq	xmm1, xmm11
+	movdqu	xmm0, XMMWORD PTR [rcx+rsi]
+	movdqu	XMMWORD PTR [rcx+rsi], xmm1
+	paddq	xmm0, xmm8
+	movdqu	XMMWORD PTR [rax+rsi], xmm0
+
+	movq	rcx, xmm10
+	and	ecx, 2097136
+
+	movdqa	xmm0, xmm10
+	pxor	xmm0, xmm6
+	movdqu	XMMWORD PTR [r8], xmm0
+	mov r12, QWORD PTR [rcx+rsi]
+
+	mov	r9, QWORD PTR [rbx+8]
+
+	xor edx, 16
+	mov r8d, edx
+	mov r15d, edx
+
+	movq	rdx, xmm5
+	shl	rdx, 32
+	movq	rax, xmm4
+	xor	rdx, rax
+	xor	r10, rdx
+	mov	rax, r10
+	mul	r11
+	mov r11d, r8d
+	xor r11d, 48
+	movq xmm0, rdx
+	xor rdx, [r11+r13]
+	movq xmm1, rax
+	xor rax, [r11+r13+8]
+	punpcklqdq xmm0, xmm1
+
+	pxor xmm0, XMMWORD PTR [r8+r13]
+	xor	r8d, 32
+	movdqu	xmm1, XMMWORD PTR [r11+r13]
+	paddq	xmm0, xmm7
+	paddq	xmm1, xmm2
+	movdqu	XMMWORD PTR [r11+r13], xmm0
+	movdqu	xmm0, XMMWORD PTR [r8+r13]
+	movdqu	XMMWORD PTR [r8+r13], xmm1
+	paddq	xmm0, xmm3
+	movdqu	XMMWORD PTR [r15+r13], xmm0
+
+	mov	r11, QWORD PTR [rsp+256]
+	add	r11, rdx
+	mov	rdx, QWORD PTR [rsp+264]
+	add	rdx, rax
+	mov	QWORD PTR [rbx], r11
+	xor	r11, r10
+	mov	QWORD PTR [rbx+8], rdx
+	xor	rdx, r9
+	mov	QWORD PTR [rsp+256], r11
+	and	r11d, 2097136
+	mov	QWORD PTR [rsp+264], rdx
+	mov	QWORD PTR [rsp+8], r11
+	lea	r15, QWORD PTR [r11+r13]
+	movdqu xmm15, XMMWORD PTR [r11+r13]
+	lea	r13, QWORD PTR [rsi+rcx]
+	movdqa	xmm0, xmm5
+	psrldq	xmm0, 8
+	movaps	xmm2, xmm13
+	movq	r10, xmm0
+	psllq	xmm5, 1
+	shl	r10, 32
+	movdqa	xmm0, xmm9
+	psrldq	xmm0, 8
+	movdqa	xmm1, xmm10
+	movq	r11, xmm0
+	psrldq	xmm1, 8
+	movq	r8, xmm1
+	psrldq	xmm4, 8
+	movaps	xmm0, xmm13
+	movq	rax, xmm4
+	xor	r10, rax
+	movaps	xmm1, xmm13
+	xor	r10, r12
+	lea	rax, QWORD PTR [r11+1]
+	shr	rax, 1
+	movdqa	xmm3, xmm9
+	punpcklqdq xmm3, xmm10
+	paddq	xmm5, xmm3
+	movq	rdx, xmm5
+	psrldq	xmm5, 8
+	cvtsi2sd xmm2, rax
+	or	edx, -2147483647
+	lea	rax, QWORD PTR [r8+1]
+	shr	rax, 1
+	movq	r9, xmm5
+	cvtsi2sd xmm0, rax
+	or	r9d, -2147483647
+	cvtsi2sd xmm1, rdx
+	unpcklpd xmm2, xmm0
+	movaps	xmm0, xmm13
+	cvtsi2sd xmm0, r9
+	unpcklpd xmm1, xmm0
+	divpd	xmm2, xmm1
+	paddq	xmm2, xmm14
+	cvttsd2si rax, xmm2
+	psrldq	xmm2, 8
+	mov	rbx, rax
+	imul	rax, rdx
+	sub	r11, rax
+	js	div_fix_1_sandybridge
+div_fix_1_ret_sandybridge:
+
+	cvttsd2si rdx, xmm2
+	mov	rax, rdx
+	imul	rax, r9
+	movd	xmm2, r11d
+	movd	xmm4, ebx
+	sub	r8, rax
+	js	div_fix_2_sandybridge
+div_fix_2_ret_sandybridge:
+
+	movd	xmm1, r8d
+	movd	xmm0, edx
+	punpckldq xmm2, xmm1
+	punpckldq xmm4, xmm0
+	punpckldq xmm4, xmm2
+	paddq	xmm3, xmm4
+	movdqa	xmm0, xmm3
+	psrlq	xmm0, 12
+	paddq	xmm0, xmm12
+	sqrtpd	xmm1, xmm0
+	movq	r9, xmm1
+	movdqa xmm5, xmm1
+	psrlq xmm5, 19
+	test	r9, 524287
+	je	sqrt_fix_1_sandybridge
+sqrt_fix_1_ret_sandybridge:
+
+	movq r9, xmm10
+	psrldq	xmm1, 8
+	movq	r8, xmm1
+	test	r8, 524287
+	je	sqrt_fix_2_sandybridge
+sqrt_fix_2_ret_sandybridge:
+
+	mov r12d, ecx
+	mov r8d, ecx
+	xor r12d, 16
+	xor r8d, 32
+	xor ecx, 48
+	mov	rax, r10
+	mul	r9
+	movq xmm0, rax
+	movq xmm3, rdx
+	punpcklqdq xmm3, xmm0
+
+	movdqu	xmm0, XMMWORD PTR [r12+rsi]
+	pxor xmm0, xmm3
+	movdqu	xmm1, XMMWORD PTR [r8+rsi]
+	xor rdx, [r8+rsi]
+	xor rax, [r8+rsi+8]
+	movdqu	xmm3, XMMWORD PTR [rcx+rsi]
+	paddq	xmm0, xmm6
+	paddq	xmm1, xmm11
+	paddq	xmm3, xmm8
+	movdqu	XMMWORD PTR [r8+rsi], xmm0
+	movdqu	XMMWORD PTR [rcx+rsi], xmm1
+	movdqu	XMMWORD PTR [r12+rsi], xmm3
+
+	add	rdi, rdx
+	mov	QWORD PTR [r13], rdi
+	xor	rdi, r10
+	mov	ecx, edi
+	and	ecx, 2097136
+	lea	r8, QWORD PTR [rcx+rsi]
+
+	mov rdx, QWORD PTR [r13+8]	
+	add	rbp, rax
+	mov	QWORD PTR [r13+8], rbp
+	movdqu xmm11, XMMWORD PTR [rcx+rsi]
+	xor	rbp, rdx
+	mov	r13, QWORD PTR [rsp]
+	movdqa	xmm3, xmm7
+	mov	rdx, QWORD PTR [rsp+8]
+	movdqa	xmm8, xmm6
+	mov	r10, QWORD PTR [rsp+256]
+	movdqa	xmm7, xmm9
+	mov	r11, QWORD PTR [rsp+264]
+	movdqa	xmm6, xmm10
+	mov	r9, r15
+	dec r14d
+	jne	main_loop_double_sandybridge
+
+	ldmxcsr DWORD PTR [rsp+272]
+	movaps	xmm13, XMMWORD PTR [rsp+48]
+	lea	r11, QWORD PTR [rsp+184]
+	movaps	xmm6, XMMWORD PTR [r11-24]
+	movaps	xmm7, XMMWORD PTR [r11-40]
+	movaps	xmm8, XMMWORD PTR [r11-56]
+	movaps	xmm9, XMMWORD PTR [r11-72]
+	movaps	xmm10, XMMWORD PTR [r11-88]
+	movaps	xmm11, XMMWORD PTR [r11-104]
+	movaps	xmm12, XMMWORD PTR [r11-120]
+	movaps	xmm14, XMMWORD PTR [rsp+32]
+	movaps	xmm15, XMMWORD PTR [rsp+16]
+	mov	rsp, r11
+	pop	r15
+	pop	r14
+	pop	r13
+	pop	r12
+	pop	rdi
+	pop	rsi
+	pop	rbp
+	pop	rbx
+	jmp cnv2_double_mainloop_asm_sandybridge_endp
+
+div_fix_1_sandybridge:
+	dec	rbx
+	add	r11, rdx
+	jmp	div_fix_1_ret_sandybridge
+
+div_fix_2_sandybridge:
+	dec	rdx
+	add	r8, r9
+	jmp	div_fix_2_ret_sandybridge
+
+sqrt_fix_1_sandybridge:
+	movq	r8, xmm3
+	movdqa xmm0, xmm5
+	psrldq xmm0, 8
+	dec	r9
+	mov r11d, -1022
+	shl r11, 32
+	mov	rax, r9
+	shr	r9, 19
+	shr	rax, 20
+	mov	rdx, r9
+	sub	rdx, rax
+	lea	rdx, [rdx+r11+1]
+	add	rax, r11
+	imul	rdx, rax
+	sub	rdx, r8
+	adc	r9, 0
+	movq xmm5, r9
+	punpcklqdq xmm5, xmm0
+	jmp	sqrt_fix_1_ret_sandybridge
+
+sqrt_fix_2_sandybridge:
+	psrldq	xmm3, 8
+	movq	r11, xmm3
+	dec	r8
+	mov ebx, -1022
+	shl rbx, 32
+	mov	rax, r8
+	shr	r8, 19
+	shr	rax, 20
+	mov	rdx, r8
+	sub	rdx, rax
+	lea	rdx, [rdx+rbx+1]
+	add	rax, rbx
+	imul	rdx, rax
+	sub	rdx, r11
+	adc	r8, 0
+	movq xmm0, r8
+	punpcklqdq xmm5, xmm0
+	jmp	sqrt_fix_2_ret_sandybridge
+
+cnv2_double_mainloop_asm_sandybridge_endp:

crypto/asm/cn2/cnv2_main_loop_bulldozer.inc

@@ -0,0 +1,180 @@
+	mov	QWORD PTR [rsp+16], rbx
+	mov	QWORD PTR [rsp+24], rbp
+	mov	QWORD PTR [rsp+32], rsi
+	push	rdi
+	push	r12
+	push	r13
+	push	r14
+	push	r15
+	sub	rsp, 64
+
+	stmxcsr DWORD PTR [rsp]
+	mov DWORD PTR [rsp+4], 24448
+	ldmxcsr DWORD PTR [rsp+4]
+
+	mov	rax, QWORD PTR [rcx+48]
+	mov	r9, rcx
+	xor	rax, QWORD PTR [rcx+16]
+	mov	ebp, 524288
+	mov	r8, QWORD PTR [rcx+32]
+	xor	r8, QWORD PTR [rcx]
+	mov	r11, QWORD PTR [rcx+40]
+	mov	r10, r8
+	mov	rdx, QWORD PTR [rcx+56]
+	movq	xmm3, rax
+	xor	rdx, QWORD PTR [rcx+24]
+	xor	r11, QWORD PTR [rcx+8]
+	mov	rbx, QWORD PTR [rcx+224]
+	mov	rax, QWORD PTR [r9+80]
+	xor	rax, QWORD PTR [r9+64]
+	movq	xmm0, rdx
+	mov	rcx, QWORD PTR [rcx+88]
+	xor	rcx, QWORD PTR [r9+72]
+	mov	rdi, QWORD PTR [r9+104]
+	and	r10d, 2097136
+	movaps	XMMWORD PTR [rsp+48], xmm6
+	movq	xmm4, rax
+	movaps	XMMWORD PTR [rsp+32], xmm7
+	movaps	XMMWORD PTR [rsp+16], xmm8
+	xorps	xmm8, xmm8
+	mov ax, 1023
+	shl rax, 52
+	movq xmm7, rax
+	mov	r15, QWORD PTR [r9+96]
+	punpcklqdq xmm3, xmm0
+	movq	xmm0, rcx
+	punpcklqdq xmm4, xmm0
+
+	ALIGN(64)
+cnv2_main_loop_bulldozer:
+	movdqa	xmm5, XMMWORD PTR [r10+rbx]
+	movq xmm6, r8
+	pinsrq xmm6, r11, 1
+	lea	rdx, QWORD PTR [r10+rbx]
+	lea	r9, QWORD PTR [rdi+rdi]
+	shl	rdi, 32
+
+	mov	ecx, r10d
+	mov	eax, r10d
+	xor	ecx, 16
+	xor	eax, 32
+	xor	r10d, 48
+	aesenc	xmm5, xmm6
+	movdqa	xmm2, XMMWORD PTR [rcx+rbx]
+	movdqa	xmm1, XMMWORD PTR [rax+rbx]
+	movdqa	xmm0, XMMWORD PTR [r10+rbx]
+	paddq	xmm2, xmm3
+	paddq	xmm1, xmm6
+	paddq	xmm0, xmm4
+	movdqa	XMMWORD PTR [rcx+rbx], xmm0
+	movdqa	XMMWORD PTR [rax+rbx], xmm2
+	movdqa	XMMWORD PTR [r10+rbx], xmm1
+
+	movaps	xmm1, xmm8
+	mov	rsi, r15
+	xor	rsi, rdi
+
+	mov edi, 1023
+	shl rdi, 52
+
+	movq	r14, xmm5
+	pextrq rax, xmm5, 1
+
+	movdqa	xmm0, xmm5
+	pxor	xmm0, xmm3
+	mov	r10, r14
+	and	r10d, 2097136
+	movdqa	XMMWORD PTR [rdx], xmm0
+	xor	rsi, QWORD PTR [r10+rbx]
+	lea	r12, QWORD PTR [r10+rbx]
+	mov	r13, QWORD PTR [r10+rbx+8]
+
+	add	r9d, r14d
+	or	r9d, -2147483647
+	xor	edx, edx
+	div	r9
+	mov	eax, eax
+	shl	rdx, 32
+	lea	r15, [rax+rdx]
+	lea	rax, [r14+r15]
+	shr	rax, 12
+	add	rax, rdi
+	movq	xmm0, rax
+	sqrtsd	xmm1, xmm0
+	movq	rdi, xmm1
+	test	rdi, 524287
+	je	sqrt_fixup_bulldozer
+	shr	rdi, 19
+
+sqrt_fixup_bulldozer_ret:
+	mov	rax, rsi
+	mul	r14
+	movq xmm1, rax
+	movq xmm0, rdx
+	punpcklqdq xmm0, xmm1
+
+	mov	r9d, r10d
+	mov	ecx, r10d
+	xor	r9d, 16
+	xor	ecx, 32
+	xor	r10d, 48
+	movdqa	xmm1, XMMWORD PTR [rcx+rbx]
+	xor rdx, [rcx+rbx]
+	xor rax, [rcx+rbx+8]
+	movdqa	xmm2, XMMWORD PTR [r9+rbx]
+	pxor xmm2, xmm0
+	paddq xmm4, XMMWORD PTR [r10+rbx]
+	paddq	xmm2, xmm3
+	paddq	xmm1, xmm6
+	movdqa	XMMWORD PTR [r9+rbx], xmm4
+	movdqa	XMMWORD PTR [rcx+rbx], xmm2
+	movdqa	XMMWORD PTR [r10+rbx], xmm1
+
+	movdqa	xmm4, xmm3
+	add	r8, rdx
+	add	r11, rax
+	mov	QWORD PTR [r12], r8
+	xor	r8, rsi
+	mov	QWORD PTR [r12+8], r11
+	mov	r10, r8
+	xor	r11, r13
+	and	r10d, 2097136
+	movdqa	xmm3, xmm5
+	dec	ebp
+	jne	cnv2_main_loop_bulldozer
+
+	ldmxcsr DWORD PTR [rsp]
+	movaps	xmm6, XMMWORD PTR [rsp+48]
+	lea	r11, QWORD PTR [rsp+64]
+	mov	rbx, QWORD PTR [r11+56]
+	mov	rbp, QWORD PTR [r11+64]
+	mov	rsi, QWORD PTR [r11+72]
+	movaps	xmm8, XMMWORD PTR [r11-48]
+	movaps	xmm7, XMMWORD PTR [rsp+32]
+	mov	rsp, r11
+	pop	r15
+	pop	r14
+	pop	r13
+	pop	r12
+	pop	rdi
+	jmp cnv2_main_loop_bulldozer_endp
+
+sqrt_fixup_bulldozer:
+	movq r9, xmm5
+	add r9, r15
+	dec	rdi
+	mov edx, -1022
+	shl rdx, 32
+	mov	rax, rdi
+	shr	rdi, 19
+	shr	rax, 20
+	mov	rcx, rdi
+	sub	rcx, rax
+	lea	rcx, [rcx+rdx+1]
+	add	rax, rdx
+	imul	rcx, rax
+	sub	rcx, r9
+	adc	rdi, 0
+	jmp	sqrt_fixup_bulldozer_ret
+
+cnv2_main_loop_bulldozer_endp:

crypto/asm/cn2/cnv2_main_loop_ivybridge.inc

@@ -0,0 +1,186 @@
+	mov	 QWORD PTR [rsp+24], rbx
+	push	 rbp
+	push	 rsi
+	push	 rdi
+	push	 r12
+	push	 r13
+	push	 r14
+	push	 r15
+	sub	 rsp, 80
+
+	stmxcsr DWORD PTR [rsp]
+	mov DWORD PTR [rsp+4], 24448
+	ldmxcsr DWORD PTR [rsp+4]
+
+	mov	 rax, QWORD PTR [rcx+48]
+	mov	 r9, rcx
+	xor	 rax, QWORD PTR [rcx+16]
+	mov	 esi, 524288
+	mov	 r8, QWORD PTR [rcx+32]
+	mov	 r13d, -2147483647
+	xor	 r8, QWORD PTR [rcx]
+	mov	 r11, QWORD PTR [rcx+40]
+	mov	 r10, r8
+	mov	 rdx, QWORD PTR [rcx+56]
+	movq	 xmm4, rax
+	xor	 rdx, QWORD PTR [rcx+24]
+	xor	 r11, QWORD PTR [rcx+8]
+	mov	 rbx, QWORD PTR [rcx+224]
+	mov	 rax, QWORD PTR [r9+80]
+	xor	 rax, QWORD PTR [r9+64]
+	movq	 xmm0, rdx
+	mov	 rcx, QWORD PTR [rcx+88]
+	xor	 rcx, QWORD PTR [r9+72]
+	movq	 xmm3, QWORD PTR [r9+104]
+	movaps	 XMMWORD PTR [rsp+64], xmm6
+	movaps	 XMMWORD PTR [rsp+48], xmm7
+	movaps	 XMMWORD PTR [rsp+32], xmm8
+	and	 r10d, 2097136
+	movq	 xmm5, rax
+
+	xor eax, eax
+	mov QWORD PTR [rsp+16], rax
+
+	mov ax, 1023
+	shl rax, 52
+	movq xmm8, rax
+	mov r15, QWORD PTR [r9+96]
+	punpcklqdq xmm4, xmm0
+	movq	 xmm0, rcx
+	punpcklqdq xmm5, xmm0
+	movdqu	 xmm6, XMMWORD PTR [r10+rbx]
+
+	ALIGN(64)
+main_loop_ivybridge:
+	lea	 rdx, QWORD PTR [r10+rbx]
+	mov	 ecx, r10d
+	mov	 eax, r10d
+	mov rdi, r15
+	xor	 ecx, 16
+	xor	 eax, 32
+	xor	 r10d, 48
+	movq	 xmm0, r11
+	movq	 xmm7, r8
+	punpcklqdq xmm7, xmm0
+	aesenc	 xmm6, xmm7
+	movq	 rbp, xmm6
+	mov	 r9, rbp
+	and	 r9d, 2097136
+	movdqu	 xmm2, XMMWORD PTR [rcx+rbx]
+	movdqu	 xmm1, XMMWORD PTR [rax+rbx]
+	movdqu	 xmm0, XMMWORD PTR [r10+rbx]
+	paddq	 xmm1, xmm7
+	paddq	 xmm0, xmm5
+	paddq	 xmm2, xmm4
+	movdqu	 XMMWORD PTR [rcx+rbx], xmm0
+	movdqu	 XMMWORD PTR [rax+rbx], xmm2
+	movdqu	 XMMWORD PTR [r10+rbx], xmm1
+	mov r10, r9
+	xor r10d, 32
+	movq	 rcx, xmm3
+	mov	 rax, rcx
+	shl	 rax, 32
+	xor	 rdi, rax
+	movdqa	 xmm0, xmm6
+	pxor	 xmm0, xmm4
+	movdqu	 XMMWORD PTR [rdx], xmm0
+	xor	 rdi, QWORD PTR [r9+rbx]
+	lea	 r14, QWORD PTR [r9+rbx]
+	mov	 r12, QWORD PTR [r14+8]
+	xor	 edx, edx
+	lea	 r9d, DWORD PTR [ecx+ecx]
+	add	 r9d, ebp
+	movdqa	 xmm0, xmm6
+	psrldq	 xmm0, 8
+	or	 r9d, r13d
+	movq	 rax, xmm0
+	div	 r9
+	xorps xmm3, xmm3
+	mov	 eax, eax
+	shl	 rdx, 32
+	add	 rdx, rax
+	lea	 r9, QWORD PTR [rdx+rbp]
+	mov r15, rdx
+	mov	 rax, r9
+	shr	 rax, 12
+	movq	 xmm0, rax
+	paddq	 xmm0, xmm8
+	sqrtsd	 xmm3, xmm0
+	psubq	 xmm3, XMMWORD PTR [rsp+16]
+	movq	 rdx, xmm3
+	test	 edx, 524287
+	je	 sqrt_fixup_ivybridge
+	psrlq	 xmm3, 19
+sqrt_fixup_ivybridge_ret:
+
+	mov	 ecx, r10d
+	mov	 rax, rdi
+	mul	 rbp
+	movq xmm2, rdx
+	xor rdx, [rcx+rbx]
+	add	 r8, rdx
+	mov	 QWORD PTR [r14], r8
+	xor	 r8, rdi
+	mov edi, r8d
+	and edi, 2097136
+	movq xmm0, rax
+	xor rax, [rcx+rbx+8]
+	add	 r11, rax
+	mov	 QWORD PTR [r14+8], r11
+	punpcklqdq xmm2, xmm0
+
+	mov	 r9d, r10d
+	xor	 r9d, 48
+	xor	 r10d, 16
+	pxor	 xmm2, XMMWORD PTR [r9+rbx]
+	movdqu	 xmm0, XMMWORD PTR [r10+rbx]
+	paddq	 xmm0, xmm5
+	movdqu	 xmm1, XMMWORD PTR [rcx+rbx]
+	paddq	 xmm2, xmm4
+	paddq	 xmm1, xmm7
+	movdqa	 xmm5, xmm4
+	movdqu	 XMMWORD PTR [r9+rbx], xmm0
+	movdqa	 xmm4, xmm6
+	movdqu	 XMMWORD PTR [rcx+rbx], xmm2
+	movdqu	 XMMWORD PTR [r10+rbx], xmm1
+	movdqu xmm6, [rdi+rbx]
+	mov	 r10d, edi
+	xor	 r11, r12
+	dec rsi
+	jne	 main_loop_ivybridge
+
+	ldmxcsr DWORD PTR [rsp]
+	mov	 rbx, QWORD PTR [rsp+160]
+	movaps	 xmm6, XMMWORD PTR [rsp+64]
+	movaps	 xmm7, XMMWORD PTR [rsp+48]
+	movaps	 xmm8, XMMWORD PTR [rsp+32]
+	add	 rsp, 80
+	pop	 r15
+	pop	 r14
+	pop	 r13
+	pop	 r12
+	pop	 rdi
+	pop	 rsi
+	pop	 rbp
+	jmp cnv2_main_loop_ivybridge_endp
+
+sqrt_fixup_ivybridge:
+	dec	 rdx
+	mov r13d, -1022
+	shl r13, 32
+	mov	 rax, rdx
+	shr	 rdx, 19
+	shr	 rax, 20
+	mov	 rcx, rdx
+	sub	 rcx, rax
+	add	 rax, r13
+	not r13
+	sub	 rcx, r13
+	mov	 r13d, -2147483647
+	imul	 rcx, rax
+	sub	 rcx, r9
+	adc	 rdx, 0
+	movq	 xmm3, rdx
+	jmp	 sqrt_fixup_ivybridge_ret
+
+cnv2_main_loop_ivybridge_endp:

crypto/asm/cn2/cnv2_main_loop_ryzen.inc

@@ -0,0 +1,179 @@
+	mov	QWORD PTR [rsp+16], rbx
+	mov	QWORD PTR [rsp+24], rbp
+	mov	QWORD PTR [rsp+32], rsi
+	push	rdi
+	push	r12
+	push	r13
+	push	r14
+	push	r15
+	sub	rsp, 64
+
+	stmxcsr DWORD PTR [rsp]
+	mov DWORD PTR [rsp+4], 24448
+	ldmxcsr DWORD PTR [rsp+4]
+
+	mov	rax, QWORD PTR [rcx+48]
+	mov	r9, rcx
+	xor	rax, QWORD PTR [rcx+16]
+	mov	ebp, 524288
+	mov	r8, QWORD PTR [rcx+32]
+	xor	r8, QWORD PTR [rcx]
+	mov	r11, QWORD PTR [rcx+40]
+	mov	r10, r8
+	mov	rdx, QWORD PTR [rcx+56]
+	movq	xmm3, rax
+	xor	rdx, QWORD PTR [rcx+24]
+	xor	r11, QWORD PTR [rcx+8]
+	mov	rbx, QWORD PTR [rcx+224]
+	mov	rax, QWORD PTR [r9+80]
+	xor	rax, QWORD PTR [r9+64]
+	movq	xmm0, rdx
+	mov	rcx, QWORD PTR [rcx+88]
+	xor	rcx, QWORD PTR [r9+72]
+	mov	rdi, QWORD PTR [r9+104]
+	and	r10d, 2097136
+	movaps	XMMWORD PTR [rsp+48], xmm6
+	movq	xmm4, rax
+	movaps	XMMWORD PTR [rsp+32], xmm7
+	movaps	XMMWORD PTR [rsp+16], xmm8
+	xorps	xmm8, xmm8
+	mov ax, 1023
+	shl rax, 52
+	movq xmm7, rax
+	mov	r15, QWORD PTR [r9+96]
+	punpcklqdq xmm3, xmm0
+	movq	xmm0, rcx
+	punpcklqdq xmm4, xmm0
+
+	ALIGN(64)
+main_loop_ryzen:
+	movdqa	xmm5, XMMWORD PTR [r10+rbx]
+	movq	xmm0, r11
+	movq	xmm6, r8
+	punpcklqdq xmm6, xmm0
+	lea	rdx, QWORD PTR [r10+rbx]
+	lea	r9, QWORD PTR [rdi+rdi]
+	shl	rdi, 32
+
+	mov	ecx, r10d
+	mov	eax, r10d
+	xor	ecx, 16
+	xor	eax, 32
+	xor	r10d, 48
+	aesenc	xmm5, xmm6
+	movdqa	xmm2, XMMWORD PTR [rcx+rbx]
+	movdqa	xmm1, XMMWORD PTR [rax+rbx]
+	movdqa	xmm0, XMMWORD PTR [r10+rbx]
+	paddq	xmm2, xmm3
+	paddq	xmm1, xmm6
+	paddq	xmm0, xmm4
+	movdqa	XMMWORD PTR [rcx+rbx], xmm0
+	movdqa	XMMWORD PTR [rax+rbx], xmm2
+	movdqa	XMMWORD PTR [r10+rbx], xmm1
+
+	movaps	xmm1, xmm8
+	mov	rsi, r15
+	xor	rsi, rdi
+	movq	r14, xmm5
+	movdqa	xmm0, xmm5
+	pxor	xmm0, xmm3
+	mov	r10, r14
+	and	r10d, 2097136
+	movdqa	XMMWORD PTR [rdx], xmm0
+	xor	rsi, QWORD PTR [r10+rbx]
+	lea	r12, QWORD PTR [r10+rbx]
+	mov	r13, QWORD PTR [r10+rbx+8]
+
+	add	r9d, r14d
+	or	r9d, -2147483647
+	xor	edx, edx
+	movdqa	xmm0, xmm5
+	psrldq	xmm0, 8
+	movq	rax, xmm0
+
+	div	r9
+	movq xmm0, rax
+	movq xmm1, rdx
+	punpckldq xmm0, xmm1
+	movq r15, xmm0
+	paddq xmm0, xmm5
+	movdqa xmm2, xmm0
+	psrlq xmm0, 12
+	paddq	xmm0, xmm7
+	sqrtsd	xmm1, xmm0
+	movq	rdi, xmm1
+	test	rdi, 524287
+	je	sqrt_fixup_ryzen
+	shr	rdi, 19
+
+sqrt_fixup_ryzen_ret:
+	mov	rax, rsi
+	mul	r14
+	movq xmm1, rax
+	movq xmm0, rdx
+	punpcklqdq xmm0, xmm1
+
+	mov	r9d, r10d
+	mov	ecx, r10d
+	xor	r9d, 16
+	xor	ecx, 32
+	xor	r10d, 48
+	movdqa	xmm1, XMMWORD PTR [rcx+rbx]
+	xor rdx, [rcx+rbx]
+	xor rax, [rcx+rbx+8]
+	movdqa	xmm2, XMMWORD PTR [r9+rbx]
+	pxor xmm2, xmm0
+	paddq xmm4, XMMWORD PTR [r10+rbx]
+	paddq	xmm2, xmm3
+	paddq	xmm1, xmm6
+	movdqa	XMMWORD PTR [r9+rbx], xmm4
+	movdqa	XMMWORD PTR [rcx+rbx], xmm2
+	movdqa	XMMWORD PTR [r10+rbx], xmm1
+
+	movdqa	xmm4, xmm3
+	add	r8, rdx
+	add	r11, rax
+	mov	QWORD PTR [r12], r8
+	xor	r8, rsi
+	mov	QWORD PTR [r12+8], r11
+	mov	r10, r8
+	xor	r11, r13
+	and	r10d, 2097136
+	movdqa	xmm3, xmm5
+	dec	ebp
+	jne	main_loop_ryzen
+
+	ldmxcsr DWORD PTR [rsp]
+	movaps	xmm6, XMMWORD PTR [rsp+48]
+	lea	r11, QWORD PTR [rsp+64]
+	mov	rbx, QWORD PTR [r11+56]
+	mov	rbp, QWORD PTR [r11+64]
+	mov	rsi, QWORD PTR [r11+72]
+	movaps	xmm8, XMMWORD PTR [r11-48]
+	movaps	xmm7, XMMWORD PTR [rsp+32]
+	mov	rsp, r11
+	pop	r15
+	pop	r14
+	pop	r13
+	pop	r12
+	pop	rdi
+	jmp cnv2_main_loop_ryzen_endp
+
+sqrt_fixup_ryzen:
+	movq r9, xmm2
+	dec	rdi
+	mov edx, -1022
+	shl rdx, 32
+	mov	rax, rdi
+	shr	rdi, 19
+	shr	rax, 20
+	mov	rcx, rdi
+	sub	rcx, rax
+	lea	rcx, [rcx+rdx+1]
+	add	rax, rdx
+	imul	rcx, rax
+	sub	rcx, r9
+	adc	rdi, 0
+	jmp	sqrt_fixup_ryzen_ret
+
+cnv2_main_loop_ryzen_endp:

crypto/asm/cn_main_loop.S

@@ -0,0 +1,54 @@
+#ifdef __APPLE__
+#   define ALIGN(x) .align 6
+#else
+#   define ALIGN(x) .align 64
+#endif
+.intel_syntax noprefix
+#ifdef __APPLE__
+#   define FN_PREFIX(fn) _ ## fn
+.text
+#else
+#   define FN_PREFIX(fn) fn
+.section .text
+#endif
+.global FN_PREFIX(cnv2_mainloop_ivybridge_asm)
+.global FN_PREFIX(cnv2_mainloop_ryzen_asm)
+.global FN_PREFIX(cnv2_mainloop_bulldozer_asm)
+.global FN_PREFIX(cnv2_double_mainloop_sandybridge_asm)
+
+ALIGN(64)
+FN_PREFIX(cnv2_mainloop_ivybridge_asm):
+	sub rsp, 48
+	mov rcx, rdi
+	#include "cn2/cnv2_main_loop_ivybridge.inc"
+	add rsp, 48
+	ret 0
+	mov eax, 3735929054
+
+ALIGN(64)
+FN_PREFIX(cnv2_mainloop_ryzen_asm):
+	sub rsp, 48
+	mov rcx, rdi
+	#include "cn2/cnv2_main_loop_ryzen.inc"
+	add rsp, 48
+	ret 0
+	mov eax, 3735929054
+
+ALIGN(64)
+FN_PREFIX(cnv2_mainloop_bulldozer_asm):
+	sub rsp, 48
+	mov rcx, rdi
+	#include "cn2/cnv2_main_loop_bulldozer.inc"
+	add rsp, 48
+	ret 0
+	mov eax, 3735929054
+
+ALIGN(64)
+FN_PREFIX(cnv2_double_mainloop_sandybridge_asm):
+	sub rsp, 48
+	mov rcx, rdi
+	mov rdx, rsi
+	#include "cn2/cnv2_double_main_loop_sandybridge.inc"
+	add rsp, 48
+	ret 0
+	mov eax, 3735929054

crypto/asm/win64/cn_main_loop.S

@@ -0,0 +1,31 @@
+#define ALIGN(x) .align 64
+.intel_syntax noprefix
+.section .text
+.global cnv2_mainloop_ivybridge_asm
+.global cnv2_mainloop_ryzen_asm
+.global cnv2_mainloop_bulldozer_asm
+.global cnv2_double_mainloop_sandybridge_asm
+
+ALIGN(64)
+cnv2_mainloop_ivybridge_asm:
+	#include "../cn2/cnv2_main_loop_ivybridge.inc"
+	ret 0
+	mov eax, 3735929054
+
+ALIGN(64)
+cnv2_mainloop_ryzen_asm:
+	#include "../cn2/cnv2_main_loop_ryzen.inc"
+	ret 0
+	mov eax, 3735929054
+
+ALIGN(64)
+cnv2_mainloop_bulldozer_asm:
+	#include "../cn2/cnv2_main_loop_bulldozer.inc"
+	ret 0
+	mov eax, 3735929054
+
+ALIGN(64)
+cnv2_double_mainloop_sandybridge_asm:
+	#include "../cn2/cnv2_double_main_loop_sandybridge.inc"
+	ret 0
+	mov eax, 3735929054

crypto/c_blake256.c

@@ -148,7 +148,7 @@ void blake256_update(state *S, const uint8_t *data, uint64_t datalen) {
 
     if (datalen > 0) {
         memcpy((void *) (S->buf + left), (void *) data, datalen >> 3);
-        S->buflen = (left << 3) + (int) datalen;
+        S->buflen = (left << 3) + datalen;
     } else {
         S->buflen = 0;
     }

crypto/c_keccak.c

@@ -0,0 +1,123 @@
+// keccak.c
+// 19-Nov-11  Markku-Juhani O. Saarinen <mjos@iki.fi>
+// A baseline Keccak (3rd round) implementation.
+
+#include <stdint.h>
+#include <string.h>
+
+#define HASH_DATA_AREA 136
+#define KECCAK_ROUNDS 24
+
+#ifndef ROTL64
+#define ROTL64(x, y) (((x) << (y)) | ((x) >> (64 - (y))))
+#endif
+
+const uint64_t keccakf_rndc[24] = 
+{
+    0x0000000000000001, 0x0000000000008082, 0x800000000000808a,
+    0x8000000080008000, 0x000000000000808b, 0x0000000080000001,
+    0x8000000080008081, 0x8000000000008009, 0x000000000000008a,
+    0x0000000000000088, 0x0000000080008009, 0x000000008000000a,
+    0x000000008000808b, 0x800000000000008b, 0x8000000000008089,
+    0x8000000000008003, 0x8000000000008002, 0x8000000000000080, 
+    0x000000000000800a, 0x800000008000000a, 0x8000000080008081,
+    0x8000000000008080, 0x0000000080000001, 0x8000000080008008
+};
+
+const int keccakf_rotc[24] = 
+{
+    1,  3,  6,  10, 15, 21, 28, 36, 45, 55, 2,  14, 
+    27, 41, 56, 8,  25, 43, 62, 18, 39, 61, 20, 44
+};
+
+const int keccakf_piln[24] = 
+{
+    10, 7,  11, 17, 18, 3, 5,  16, 8,  21, 24, 4, 
+    15, 23, 19, 13, 12, 2, 20, 14, 22, 9,  6,  1 
+};
+
+// update the state with given number of rounds
+
+void keccakf(uint64_t st[25], int rounds)
+{
+    int i, j, round;
+    uint64_t t, bc[5];
+
+    for (round = 0; round < rounds; ++round) {
+
+        // Theta
+        bc[0] = st[0] ^ st[5] ^ st[10] ^ st[15] ^ st[20];
+        bc[1] = st[1] ^ st[6] ^ st[11] ^ st[16] ^ st[21];
+        bc[2] = st[2] ^ st[7] ^ st[12] ^ st[17] ^ st[22];
+        bc[3] = st[3] ^ st[8] ^ st[13] ^ st[18] ^ st[23];
+        bc[4] = st[4] ^ st[9] ^ st[14] ^ st[19] ^ st[24];
+
+        for (i = 0; i < 5; ++i) {
+            t = bc[(i + 4) % 5] ^ ROTL64(bc[(i + 1) % 5], 1);
+            st[i     ] ^= t;
+            st[i +  5] ^= t;
+            st[i + 10] ^= t;
+            st[i + 15] ^= t;
+            st[i + 20] ^= t;
+        }
+
+        // Rho Pi
+        t = st[1];
+        for (i = 0; i < 24; ++i) {
+            bc[0] = st[keccakf_piln[i]];
+            st[keccakf_piln[i]] = ROTL64(t, keccakf_rotc[i]);
+            t = bc[0];
+        }
+
+        //  Chi
+        for (j = 0; j < 25; j += 5) {
+            bc[0] = st[j    ];
+            bc[1] = st[j + 1];
+            bc[2] = st[j + 2];
+            bc[3] = st[j + 3];
+            bc[4] = st[j + 4];
+            st[j    ] ^= (~bc[1]) & bc[2];
+            st[j + 1] ^= (~bc[2]) & bc[3];
+            st[j + 2] ^= (~bc[3]) & bc[4];
+            st[j + 3] ^= (~bc[4]) & bc[0];
+            st[j + 4] ^= (~bc[0]) & bc[1];
+        }
+
+        //  Iota
+        st[0] ^= keccakf_rndc[round];
+    }
+}
+
+// compute a keccak hash (md) of given byte length from "in"
+typedef uint64_t state_t[25];
+
+void keccak(const uint8_t *in, int inlen, uint8_t *md, int mdlen)
+{
+    state_t st;
+    uint8_t temp[144];
+    int i, rsiz, rsizw;
+
+    rsiz = sizeof(state_t) == mdlen ? HASH_DATA_AREA : 200 - 2 * mdlen;
+    rsizw = rsiz / 8;
+    
+    memset(st, 0, sizeof(st));
+
+    for ( ; inlen >= rsiz; inlen -= rsiz, in += rsiz) {
+        for (i = 0; i < rsizw; i++)
+            st[i] ^= ((uint64_t *) in)[i];
+        keccakf(st, KECCAK_ROUNDS);
+    }
+    
+    // last block and padding
+    memcpy(temp, in, inlen);
+    temp[inlen++] = 1;
+    memset(temp + inlen, 0, rsiz - inlen);
+    temp[rsiz - 1] |= 0x80;
+
+    for (i = 0; i < rsizw; i++)
+        st[i] ^= ((uint64_t *) temp)[i];
+
+    keccakf(st, KECCAK_ROUNDS);
+
+    memcpy(md, st, mdlen);
+}

crypto/c_skein.h

@@ -44,6 +44,4 @@ typedef u08b_t   SkeinBitSequence;               /* bit stream type */
 SkeinHashReturn skein_hash(int hashbitlen,   const SkeinBitSequence *data,
         SkeinDataLength databitlen, SkeinBitSequence *hashval);
 
-void xmr_skein(const SkeinBitSequence *data, SkeinBitSequence *hashval);
-
 #endif  /* ifndef _SKEIN_H_ */

crypto/hash.c

@@ -0,0 +1,24 @@
+// Copyright (c) 2012-2013 The Cryptonote developers
+// Distributed under the MIT/X11 software license, see the accompanying
+// file COPYING or http://www.opensource.org/licenses/mit-license.php.
+
+#include <stddef.h>
+#include <stdint.h>
+#include <string.h>
+
+#include "hash-ops.h"
+#include "c_keccak.h"
+
+void hash_permutation(union hash_state *state) {
+  keccakf((uint64_t*)state, 24);
+}
+
+void hash_process(union hash_state *state, const uint8_t *buf, size_t count) {
+  keccak1600(buf, count, (uint8_t*)state);
+}
+
+void cn_fast_hash(const void *data, size_t length, char *hash) {
+  union hash_state state;
+  hash_process(&state, data, length);
+  memcpy(hash, &state, HASH_SIZE);
+}

crypto/soft_aes.h

@@ -86,42 +86,44 @@
 #define saes_u2(p)   saes_b2w(         p, saes_f3(p), saes_f2(p),          p)
 #define saes_u3(p)   saes_b2w(         p,          p, saes_f3(p), saes_f2(p))
 
-alignas(16) const uint32_t saes_table[4][256] = { saes_data(saes_u0), saes_data(saes_u1), saes_data(saes_u2), saes_data(saes_u3) };
-alignas(16) const uint8_t  saes_sbox[256] = saes_data(saes_h0);
+__attribute__((aligned(16))) const static uint32_t saes_table[4][256] = { saes_data(saes_u0), saes_data(saes_u1), saes_data(saes_u2), saes_data(saes_u3) };
+__attribute__((aligned(16))) const static uint8_t  saes_sbox[256] = saes_data(saes_h0);
+
 
 static inline __m128i soft_aesenc(__m128i in, __m128i key)
 {
-    const uint32_t x0 = _mm_cvtsi128_si32(in);
-    const uint32_t x1 = _mm_cvtsi128_si32(_mm_shuffle_epi32(in, 0x55));
-    const uint32_t x2 = _mm_cvtsi128_si32(_mm_shuffle_epi32(in, 0xAA));
-    const uint32_t x3 = _mm_cvtsi128_si32(_mm_shuffle_epi32(in, 0xFF));
+    uint32_t x0, x1, x2, x3;
+    x0 = _mm_cvtsi128_si32(in);
+    x1 = _mm_cvtsi128_si32(_mm_shuffle_epi32(in, 0x55));
+    x2 = _mm_cvtsi128_si32(_mm_shuffle_epi32(in, 0xAA));
+    x3 = _mm_cvtsi128_si32(_mm_shuffle_epi32(in, 0xFF));
 
-	__m128i out = _mm_set_epi32(
-		(saes_table[0][x3 & 0xff] ^ saes_table[1][(x0 >> 8) & 0xff] ^ saes_table[2][(x1 >> 16) & 0xff] ^ saes_table[3][x2 >> 24]),
-		(saes_table[0][x2 & 0xff] ^ saes_table[1][(x3 >> 8) & 0xff] ^ saes_table[2][(x0 >> 16) & 0xff] ^ saes_table[3][x1 >> 24]),
-		(saes_table[0][x1 & 0xff] ^ saes_table[1][(x2 >> 8) & 0xff] ^ saes_table[2][(x3 >> 16) & 0xff] ^ saes_table[3][x0 >> 24]),
-		(saes_table[0][x0 & 0xff] ^ saes_table[1][(x1 >> 8) & 0xff] ^ saes_table[2][(x2 >> 16) & 0xff] ^ saes_table[3][x3 >> 24]));
+    __m128i out = _mm_set_epi32(
+        (saes_table[0][x3 & 0xff] ^ saes_table[1][(x0 >> 8) & 0xff] ^ saes_table[2][(x1 >> 16) & 0xff] ^ saes_table[3][x2 >> 24]),
+        (saes_table[0][x2 & 0xff] ^ saes_table[1][(x3 >> 8) & 0xff] ^ saes_table[2][(x0 >> 16) & 0xff] ^ saes_table[3][x1 >> 24]),
+        (saes_table[0][x1 & 0xff] ^ saes_table[1][(x2 >> 8) & 0xff] ^ saes_table[2][(x3 >> 16) & 0xff] ^ saes_table[3][x0 >> 24]),
+        (saes_table[0][x0 & 0xff] ^ saes_table[1][(x1 >> 8) & 0xff] ^ saes_table[2][(x2 >> 16) & 0xff] ^ saes_table[3][x3 >> 24]));
 
-	return _mm_xor_si128(out, key);
+    return _mm_xor_si128(out, key);
 }
 
 static inline uint32_t sub_word(uint32_t key)
 {
-	return (saes_sbox[key >> 24 ] << 24)   | 
-		(saes_sbox[(key >> 16) & 0xff] << 16 ) | 
-		(saes_sbox[(key >> 8)  & 0xff] << 8  ) | 
-		 saes_sbox[key & 0xff];
+    return (saes_sbox[key >> 24 ] << 24)   | 
+        (saes_sbox[(key >> 16) & 0xff] << 16 ) | 
+        (saes_sbox[(key >> 8)  & 0xff] << 8  ) | 
+         saes_sbox[key & 0xff];
 }
 
 #if defined(__clang__) || defined(XMRIG_ARM)
 static inline uint32_t _rotr(uint32_t value, uint32_t amount)
 {
-	return (value >> amount) | (value << ((32 - amount) & 31));
+    return (value >> amount) | (value << ((32 - amount) & 31));
 }
 #endif
 
-template<uint8_t rcon>
-static inline __m128i soft_aeskeygenassist(__m128i key)
+
+static inline __m128i soft_aeskeygenassist(__m128i key, uint8_t rcon)
 {
     const uint32_t X1 = sub_word(_mm_cvtsi128_si32(_mm_shuffle_epi32(key, 0x55)));
     const uint32_t X3 = sub_word(_mm_cvtsi128_si32(_mm_shuffle_epi32(key, 0xFF)));

donate.h

@@ -21,11 +21,9 @@
  *   along with this program. If not, see <http://www.gnu.org/licenses/>.
  */
 
-#include "App.h"
+#ifndef __DONATE_H__
+#define __DONATE_H__
 
+#define DONATE_LEVEL 0
 
-int main(int argc, char **argv) {
-    App app(argc, argv);
-
-    return app.exec();
-}
+#endif /* __DONATE_H__ */