v4.6.2-beta

Fix function names for clang on Apple

CHANGELOG.md

@@ -1,3 +1,11 @@
+# v4.6.2-beta
+- [#1274](https://github.com/xmrig/xmrig/issues/1274) Added `--cuda-devices` command line option.
+- [#1277](https://github.com/xmrig/xmrig/pull/1277) Fixed function names for clang on Apple.
+
+# v4.6.1-beta
+- [#1272](https://github.com/xmrig/xmrig/pull/1272) Optimized hashrate calculation.
+- [#1273](https://github.com/xmrig/xmrig/issues/1273) Fixed crash when use `GET /2/backends` API endpoint with disabled CUDA.
+
 # v4.6.0-beta
 - [#1263](https://github.com/xmrig/xmrig/pull/1263) Added new option `dataset_host` for NVIDIA GPUs with less than 4 GB memory (RandomX only).
 

README.md

@@ -74,7 +74,7 @@ API:
 
 OpenCL backend:
       --opencl                  enable OpenCL mining backend
-      --opencl-devices=N        list of OpenCL devices to use
+      --opencl-devices=N        comma separated list of OpenCL devices to use
       --opencl-platform=N       OpenCL platform index or name
       --opencl-loader=PATH      path to OpenCL-ICD-Loader (OpenCL.dll or libOpenCL.so)
       --opencl-no-cache         disable OpenCL cache
@@ -83,6 +83,7 @@ OpenCL backend:
 CUDA backend:
       --cuda                    enable CUDA mining backend
       --cuda-loader=PATH        path to CUDA plugin (xmrig-cuda.dll or libxmrig-cuda.so)
+      --cuda-devices=N          comma separated list of CUDA devices to use
       --no-nvml                 disable NVML (NVIDIA Management Library) support
 
 Logging:

src/backend/common/Hashrate.cpp

@@ -47,7 +47,6 @@ inline static const char *format(double h, char *buf, size_t size)
 
 
 xmrig::Hashrate::Hashrate(size_t threads) :
-    m_highest(0.0),
     m_threads(threads)
 {
     m_counts     = new uint64_t*[threads];
@@ -100,30 +99,30 @@ double xmrig::Hashrate::calc(size_t threadId, size_t ms) const
 
     uint64_t earliestHashCount = 0;
     uint64_t earliestStamp     = 0;
-    uint64_t lastestStamp      = 0;
-    uint64_t lastestHashCnt    = 0;
     bool haveFullSet           = false;
 
-    for (size_t i = 1; i < kBucketSize; i++) {
+    const uint64_t timeStampLimit = xmrig::Chrono::highResolutionMSecs() - ms;
-        const size_t idx = (m_top[threadId] - i) & kBucketMask;
+    uint64_t* timestamps = m_timestamps[threadId];
+    uint64_t* counts = m_counts[threadId];
 
-        if (m_timestamps[threadId][idx] == 0) {
+    const size_t idx_start = (m_top[threadId] - 1) & kBucketMask;
+    size_t idx = idx_start;
+
+    uint64_t lastestStamp = timestamps[idx];
+    uint64_t lastestHashCnt = counts[idx];
+
+    do {
+        if (timestamps[idx] < timeStampLimit) {
+            haveFullSet = (timestamps[idx] != 0);
+            if (idx != idx_start) {
+                idx = (idx + 1) & kBucketMask;
+                earliestStamp = timestamps[idx];
+                earliestHashCount = counts[idx];
+            }
             break;
         }
-
+        idx = (idx - 1) & kBucketMask;
-        if (lastestStamp == 0) {
+    } while (idx != idx_start);
-            lastestStamp = m_timestamps[threadId][idx];
-            lastestHashCnt = m_counts[threadId][idx];
-        }
-
-        if (xmrig::Chrono::highResolutionMSecs() - m_timestamps[threadId][idx] > ms) {
-            haveFullSet = true;
-            break;
-        }
-
-        earliestStamp = m_timestamps[threadId][idx];
-        earliestHashCount = m_counts[threadId][idx];
-    }
 
     if (!haveFullSet || earliestStamp == 0 || lastestStamp == 0) {
         return nan("");
@@ -150,15 +149,6 @@ void xmrig::Hashrate::add(size_t threadId, uint64_t count, uint64_t timestamp)
 }
 
 
-void xmrig::Hashrate::updateHighest()
-{
-   double highest = calc(ShortInterval);
-   if (std::isnormal(highest) && highest > m_highest) {
-       m_highest = highest;
-   }
-}
-
-
 const char *xmrig::Hashrate::format(double h, char *buf, size_t size)
 {
     return ::format(h, buf, size);

src/backend/common/Hashrate.h

@@ -53,9 +53,7 @@ public:
     double calc(size_t ms) const;
     double calc(size_t threadId, size_t ms) const;
     void add(size_t threadId, uint64_t count, uint64_t timestamp);
-    void updateHighest();
 
-    inline double highest() const { return m_highest; }
     inline size_t threads() const { return m_threads; }
 
     static const char *format(double h, char *buf, size_t size);
@@ -70,7 +68,6 @@ private:
     constexpr static size_t kBucketSize = 2 << 11;
     constexpr static size_t kBucketMask = kBucketSize - 1;
 
-    double m_highest;
     size_t m_threads;
     uint32_t* m_top;
     uint64_t** m_counts;

src/backend/common/Workers.cpp

@@ -144,8 +144,6 @@ void xmrig::Workers<T>::tick(uint64_t)
 
         d_ptr->hashrate->add(handle->id(), handle->worker()->hashCount(), handle->worker()->timestamp());
     }
-
-    d_ptr->hashrate->updateHighest();
 }
 
 

src/backend/cuda/CudaBackend.cpp

@@ -155,11 +155,14 @@ public:
             return;
         }
 
+        devices = CudaLib::devices(cuda.bfactor(), cuda.bsleep(), cuda.devicesHint());
+        if (devices.empty()) {
+            return printDisabled(kLabel, RED_S " (no devices)");
+        }
+
         Log::print(GREEN_BOLD(" * ") WHITE_BOLD("%-13s") WHITE_BOLD("%s") "/" WHITE_BOLD("%s") BLACK_BOLD("/%s"), kLabel,
                    CudaLib::version(runtimeVersion).c_str(), CudaLib::version(driverVersion).c_str(), CudaLib::pluginVersion());
 
-        devices = CudaLib::devices(cuda.bfactor(), cuda.bsleep());
-
 #       ifdef XMRIG_FEATURE_NVML
         if (cuda.isNvmlEnabled()) {
             if (NvmlLib::init(cuda.nvmlLoader())) {
@@ -172,7 +175,7 @@ public:
                            );
             }
             else {
-                printDisabled(kLabel, RED_S " (failed to load NVML)");
+                printDisabled(kNvmlLabel, RED_S " (failed to load NVML)");
             }
         }
         else {
@@ -474,19 +477,21 @@ rapidjson::Value xmrig::CudaBackend::toJSON(rapidjson::Document &doc) const
     out.AddMember("algo",       d_ptr->algo.toJSON(), allocator);
     out.AddMember("profile",    profileName().toJSON(), allocator);
 
-    Value versions(kObjectType);
+    if (CudaLib::isReady()) {
-    versions.AddMember("cuda-runtime",   Value(CudaLib::version(d_ptr->runtimeVersion).c_str(), allocator), allocator);
+        Value versions(kObjectType);
-    versions.AddMember("cuda-driver",    Value(CudaLib::version(d_ptr->driverVersion).c_str(), allocator), allocator);
+        versions.AddMember("cuda-runtime",   Value(CudaLib::version(d_ptr->runtimeVersion).c_str(), allocator), allocator);
-    versions.AddMember("plugin",         String(CudaLib::pluginVersion()).toJSON(doc), allocator);
+        versions.AddMember("cuda-driver",    Value(CudaLib::version(d_ptr->driverVersion).c_str(), allocator), allocator);
+        versions.AddMember("plugin",         String(CudaLib::pluginVersion()).toJSON(doc), allocator);
 
-#   ifdef XMRIG_FEATURE_NVML
+#       ifdef XMRIG_FEATURE_NVML
-    if (NvmlLib::isReady()) {
+        if (NvmlLib::isReady()) {
-        versions.AddMember("nvml",       StringRef(NvmlLib::version()), allocator);
+            versions.AddMember("nvml",       StringRef(NvmlLib::version()), allocator);
-        versions.AddMember("driver",     StringRef(NvmlLib::driverVersion()), allocator);
+            versions.AddMember("driver",     StringRef(NvmlLib::driverVersion()), allocator);
+        }
+#       endif
+
+        out.AddMember("versions", versions, allocator);
     }
-#   endif
-
-    out.AddMember("versions", versions, allocator);
 
     if (d_ptr->threads.empty() || !hashrate()) {
         return out;

src/backend/cuda/CudaConfig.cpp

@@ -78,6 +78,16 @@ rapidjson::Value xmrig::CudaConfig::toJSON(rapidjson::Document &doc) const
 
 std::vector<xmrig::CudaLaunchData> xmrig::CudaConfig::get(const Miner *miner, const Algorithm &algorithm, const std::vector<CudaDevice> &devices) const
 {
+    auto deviceIndex = [&devices](uint32_t index) -> int {
+        for (uint32_t i = 0; i < devices.size(); ++i) {
+            if (devices[i].index() == index) {
+                return i;
+            }
+        }
+
+        return -1;
+    };
+
     std::vector<CudaLaunchData> out;
     const auto &threads = m_threads.get(algorithm);
 
@@ -85,15 +95,16 @@ std::vector<xmrig::CudaLaunchData> xmrig::CudaConfig::get(const Miner *miner, co
         return out;
     }
 
-    out.reserve(threads.count() * 2);
+    out.reserve(threads.count());
 
     for (const auto &thread : threads.data()) {
-        if (thread.index() >= devices.size()) {
+        const int index = deviceIndex(thread.index());
+        if (index == -1) {
             LOG_INFO("%s" YELLOW(" skip non-existing device with index ") YELLOW_BOLD("%u"), cuda_tag(), thread.index());
             continue;
         }
 
-        out.emplace_back(miner, algorithm, thread, devices[thread.index()]);
+        out.emplace_back(miner, algorithm, thread, devices[static_cast<size_t>(index)]);
     }
 
     return out;
@@ -153,7 +164,7 @@ void xmrig::CudaConfig::generate()
         return;
     }
 
-    const auto devices = CudaLib::devices(bfactor(), bsleep());
+    const auto devices = CudaLib::devices(bfactor(), bsleep(), m_devicesHint);
     if (devices.empty()) {
         return;
     }

src/backend/cuda/CudaConfig.h

@@ -43,16 +43,17 @@ public:
     std::vector<CudaLaunchData> get(const Miner *miner, const Algorithm &algorithm, const std::vector<CudaDevice> &devices) const;
     void read(const rapidjson::Value &value);
 
-    inline bool isEnabled() const                       { return m_enabled; }
+    inline bool isEnabled() const                               { return m_enabled; }
-    inline bool isShouldSave() const                    { return m_shouldSave; }
+    inline bool isShouldSave() const                            { return m_shouldSave; }
-    inline const String &loader() const                 { return m_loader; }
+    inline const std::vector<uint32_t> &devicesHint() const     { return m_devicesHint; }
-    inline const Threads<CudaThreads> &threads() const  { return m_threads; }
+    inline const String &loader() const                         { return m_loader; }
-    inline int32_t bfactor() const                      { return m_bfactor; }
+    inline const Threads<CudaThreads> &threads() const          { return m_threads; }
-    inline int32_t bsleep() const                       { return m_bsleep; }
+    inline int32_t bfactor() const                              { return m_bfactor; }
+    inline int32_t bsleep() const                               { return m_bsleep; }
 
 #   ifdef XMRIG_FEATURE_NVML
-    inline bool isNvmlEnabled() const                   { return m_nvml; }
+    inline bool isNvmlEnabled() const                           { return m_nvml; }
-    inline const String &nvmlLoader() const             { return m_nvmlLoader; }
+    inline const String &nvmlLoader() const                     { return m_nvmlLoader; }
 #   endif
 
 private:

src/backend/cuda/wrappers/CudaLib.cpp

@@ -209,7 +209,7 @@ std::string xmrig::CudaLib::version(uint32_t version)
 }
 
 
-std::vector<xmrig::CudaDevice> xmrig::CudaLib::devices(int32_t bfactor, int32_t bsleep) noexcept
+std::vector<xmrig::CudaDevice> xmrig::CudaLib::devices(int32_t bfactor, int32_t bsleep, const std::vector<uint32_t> &hints) noexcept
 {
     const uint32_t count = deviceCount();
     if (!count) {
@@ -219,10 +219,24 @@ std::vector<xmrig::CudaDevice> xmrig::CudaLib::devices(int32_t bfactor, int32_t
     std::vector<CudaDevice> out;
     out.reserve(count);
 
-    for (uint32_t i = 0; i < count; ++i) {
+    if (hints.empty()) {
-        CudaDevice device(i, bfactor, bsleep);
+        for (uint32_t i = 0; i < count; ++i) {
-        if (device.isValid()) {
+            CudaDevice device(i, bfactor, bsleep);
-            out.emplace_back(std::move(device));
+            if (device.isValid()) {
+                out.emplace_back(std::move(device));
+            }
+        }
+    }
+    else {
+        for (const uint32_t i : hints) {
+            if (i >= count) {
+                continue;
+            }
+
+            CudaDevice device(i, bfactor, bsleep);
+            if (device.isValid()) {
+                out.emplace_back(std::move(device));
+            }
         }
     }
 

src/backend/cuda/wrappers/CudaLib.h

@@ -69,8 +69,9 @@ public:
     static const char *lastError() noexcept;
     static void close();
 
-    static inline bool isInitialized()   { return m_initialized; }
+    static inline bool isInitialized()    { return m_initialized; }
-    static inline const String &loader() { return m_loader; }
+    static inline bool isReady() noexcept { return m_ready; }
+    static inline const String &loader()  { return m_loader; }
 
     static bool cnHash(nvid_ctx *ctx, uint32_t startNonce, uint64_t height, uint64_t target, uint32_t *rescount, uint32_t *resnonce);
     static bool deviceInit(nvid_ctx *ctx) noexcept;
@@ -84,7 +85,7 @@ public:
     static int32_t deviceInt(nvid_ctx *ctx, DeviceProperty property) noexcept;
     static nvid_ctx *alloc(uint32_t id, int32_t bfactor, int32_t bsleep) noexcept;
     static std::string version(uint32_t version);
-    static std::vector<CudaDevice> devices(int32_t bfactor, int32_t bsleep) noexcept;
+    static std::vector<CudaDevice> devices(int32_t bfactor, int32_t bsleep, const std::vector<uint32_t> &hints) noexcept;
     static uint32_t deviceCount() noexcept;
     static uint32_t deviceUint(nvid_ctx *ctx, DeviceProperty property) noexcept;
     static uint32_t driverVersion() noexcept;

src/base/io/json/Json.cpp

@@ -50,7 +50,7 @@ bool xmrig::Json::getBool(const rapidjson::Value &obj, const char *key, bool def
 }
 
 
-const char *xmrig::Json::getString(const rapidjson::Value &obj, const char *key,  const char *defaultValue)
+const char *xmrig::Json::getString(const rapidjson::Value &obj, const char *key, const char *defaultValue)
 {
     assert(obj.IsObject());
 

src/core/config/ConfigTransform.cpp

@@ -187,10 +187,14 @@ void xmrig::ConfigTransform::transform(rapidjson::Document &doc, int key, const
 
 #   ifdef XMRIG_FEATURE_CUDA
     case IConfig::CudaKey: /* --cuda */
-        return set(doc, kCuda, "enabled", true);
+        return set(doc, kCuda, kEnabled, true);
 
     case IConfig::CudaLoaderKey: /* --cuda-loader */
         return set(doc, kCuda, "loader", arg);
+
+    case IConfig::CudaDevicesKey: /* --cuda-devices */
+        set(doc, kCuda, kEnabled, true);
+        return set(doc, kCuda, "devices-hint", arg);
 #   endif
 
 #   ifdef XMRIG_FEATURE_NVML

src/core/config/Config_platform.h

@@ -107,6 +107,7 @@ static const option options[] = {
 #   ifdef XMRIG_FEATURE_CUDA
     { "cuda",                  0, nullptr, IConfig::CudaKey               },
     { "cuda-loader",           1, nullptr, IConfig::CudaLoaderKey         },
+    { "cuda-devices",          1, nullptr, IConfig::CudaDevicesKey        },
 #   endif
 #   ifdef XMRIG_FEATURE_NVML
     { "no-nvml",               0, nullptr, IConfig::NvmlKey               },

src/core/config/usage.h

@@ -101,7 +101,7 @@ static inline const std::string &usage()
 #   ifdef XMRIG_FEATURE_OPENCL
     u += "\nOpenCL backend:\n";
     u += "      --opencl                  enable OpenCL mining backend\n";
-    u += "      --opencl-devices=N        list of OpenCL devices to use\n";
+    u += "      --opencl-devices=N        comma separated list of OpenCL devices to use\n";
     u += "      --opencl-platform=N       OpenCL platform index or name\n";
     u += "      --opencl-loader=PATH      path to OpenCL-ICD-Loader (OpenCL.dll or libOpenCL.so)\n";
     u += "      --opencl-no-cache         disable OpenCL cache\n";
@@ -112,6 +112,7 @@ static inline const std::string &usage()
     u += "\nCUDA backend:\n";
     u += "      --cuda                    enable CUDA mining backend\n";
     u += "      --cuda-loader=PATH        path to CUDA plugin (xmrig-cuda.dll or libxmrig-cuda.so)\n";
+    u += "      --cuda-devices=N          comma separated list of CUDA devices to use\n";
 #   endif
 #   ifdef XMRIG_FEATURE_NVML
     u += "      --no-nvml                 disable NVML (NVIDIA Management Library) support\n";

src/crypto/randomx/jit_compiler_a64_static.S

@@ -25,26 +25,32 @@
 # 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.
 
+#if defined(__APPLE__)
+#define DECL(x) _##x
+#else
+#define DECL(x) x
+#endif
+
 	.arch armv8-a
 	.text
-	.global	randomx_program_aarch64
+	.global DECL(randomx_program_aarch64)
-	.global	randomx_program_aarch64_main_loop
+	.global DECL(randomx_program_aarch64_main_loop)
-	.global	randomx_program_aarch64_vm_instructions
+	.global DECL(randomx_program_aarch64_vm_instructions)
-	.global randomx_program_aarch64_imul_rcp_literals_end
+	.global DECL(randomx_program_aarch64_imul_rcp_literals_end)
-	.global	randomx_program_aarch64_vm_instructions_end
+	.global DECL(randomx_program_aarch64_vm_instructions_end)
-	.global randomx_program_aarch64_cacheline_align_mask1
+	.global DECL(randomx_program_aarch64_cacheline_align_mask1)
-	.global randomx_program_aarch64_cacheline_align_mask2
+	.global DECL(randomx_program_aarch64_cacheline_align_mask2)
-	.global randomx_program_aarch64_update_spMix1
+	.global DECL(randomx_program_aarch64_update_spMix1)
-	.global randomx_program_aarch64_vm_instructions_end_light
+	.global DECL(randomx_program_aarch64_vm_instructions_end_light)
-	.global randomx_program_aarch64_light_cacheline_align_mask
+	.global DECL(randomx_program_aarch64_light_cacheline_align_mask)
-	.global randomx_program_aarch64_light_dataset_offset
+	.global DECL(randomx_program_aarch64_light_dataset_offset)
-	.global randomx_init_dataset_aarch64
+	.global DECL(randomx_init_dataset_aarch64)
-	.global randomx_init_dataset_aarch64_end
+	.global DECL(randomx_init_dataset_aarch64_end)
-	.global randomx_calc_dataset_item_aarch64
+	.global DECL(randomx_calc_dataset_item_aarch64)
-	.global randomx_calc_dataset_item_aarch64_prefetch
+	.global DECL(randomx_calc_dataset_item_aarch64_prefetch)
-	.global randomx_calc_dataset_item_aarch64_mix
+	.global DECL(randomx_calc_dataset_item_aarch64_mix)
-	.global randomx_calc_dataset_item_aarch64_store_result
+	.global DECL(randomx_calc_dataset_item_aarch64_store_result)
-	.global randomx_calc_dataset_item_aarch64_end
+	.global DECL(randomx_calc_dataset_item_aarch64_end)
 
 # Register allocation
 
@@ -99,7 +105,7 @@
 # v31 -> scale mask   = 0x81f000000000000081f0000000000000
 
 	.balign 4
-randomx_program_aarch64:
+DECL(randomx_program_aarch64):
 	# Save callee-saved registers
 	sub	sp, sp, 192
 	stp	x16, x17, [sp]
@@ -187,7 +193,7 @@ randomx_program_aarch64:
 	ldr	q14, literal_v14
 	ldr	q15, literal_v15
 
-randomx_program_aarch64_main_loop:
+DECL(randomx_program_aarch64_main_loop):
 	# spAddr0 = spMix1 & ScratchpadL3Mask64;
 	# spAddr1 = (spMix1 >> 32) & ScratchpadL3Mask64;
 	lsr	x18, x10, 32
@@ -260,7 +266,7 @@ randomx_program_aarch64_main_loop:
 	orr	v23.16b, v23.16b, v30.16b
 
 	# Execute VM instructions
-randomx_program_aarch64_vm_instructions:
+DECL(randomx_program_aarch64_vm_instructions):
 
 	# 16 KB buffer for generated instructions
 	.fill 4096,4,0
@@ -278,7 +284,7 @@ literal_x27: .fill 1,8,0
 literal_x28: .fill 1,8,0
 literal_x29: .fill 1,8,0
 literal_x30: .fill 1,8,0
-randomx_program_aarch64_imul_rcp_literals_end:
+DECL(randomx_program_aarch64_imul_rcp_literals_end):
 
 literal_v0:  .fill 2,8,0
 literal_v1:  .fill 2,8,0
@@ -297,14 +303,14 @@ literal_v13: .fill 2,8,0
 literal_v14: .fill 2,8,0
 literal_v15: .fill 2,8,0
 
-randomx_program_aarch64_vm_instructions_end:
+DECL(randomx_program_aarch64_vm_instructions_end):
 
 	# mx ^= r[readReg2] ^ r[readReg3];
 	eor	x9, x9, x18
 
 	# Calculate dataset pointer for dataset prefetch
 	mov	w18, w9
-randomx_program_aarch64_cacheline_align_mask1:
+DECL(randomx_program_aarch64_cacheline_align_mask1):
 	# Actual mask will be inserted by JIT compiler
 	and	x18, x18, 1
 	add	x18, x18, x1
@@ -317,12 +323,12 @@ randomx_program_aarch64_cacheline_align_mask1:
 
 	# Calculate dataset pointer for dataset read
 	mov	w10, w9
-randomx_program_aarch64_cacheline_align_mask2:
+DECL(randomx_program_aarch64_cacheline_align_mask2):
 	# Actual mask will be inserted by JIT compiler
 	and	x10, x10, 1
 	add	x10, x10, x1
 
-randomx_program_aarch64_xor_with_dataset_line:
+DECL(randomx_program_aarch64_xor_with_dataset_line):
 	# xor integer registers with dataset data
 	ldp	x18, x19, [x10]
 	eor	x4, x4, x18
@@ -337,7 +343,7 @@ randomx_program_aarch64_xor_with_dataset_line:
 	eor	x14, x14, x18
 	eor	x15, x15, x19
 
-randomx_program_aarch64_update_spMix1:
+DECL(randomx_program_aarch64_update_spMix1):
 	# JIT compiler will replace it with "eor x10, config.readReg0, config.readReg1"
 	eor	x10, x0, x0
 
@@ -358,7 +364,7 @@ randomx_program_aarch64_update_spMix1:
 	stp	q18, q19, [x16, 32]
 
 	subs	x3, x3, 1
-	bne	randomx_program_aarch64_main_loop
+	bne	DECL(randomx_program_aarch64_main_loop)
 	
 	# Restore x0
 	ldr	x0, [sp], 16
@@ -392,7 +398,7 @@ randomx_program_aarch64_update_spMix1:
 
 	ret
 
-randomx_program_aarch64_vm_instructions_end_light:
+DECL(randomx_program_aarch64_vm_instructions_end_light):
 	sub	sp, sp, 96
 	stp	x0, x1, [sp, 64]
 	stp	x2, x30, [sp, 80]
@@ -409,26 +415,26 @@ randomx_program_aarch64_vm_instructions_end_light:
 	# x1 -> pointer to output
 	mov	x1, sp
 
-randomx_program_aarch64_light_cacheline_align_mask:
+DECL(randomx_program_aarch64_light_cacheline_align_mask):
 	# Actual mask will be inserted by JIT compiler
 	and	w2, w9, 1
 
 	# x2 -> item number
 	lsr	x2, x2, 6
 
-randomx_program_aarch64_light_dataset_offset:
+DECL(randomx_program_aarch64_light_dataset_offset):
 	# Apply dataset offset (filled in by JIT compiler)
 	add	x2, x2, 0
 	add	x2, x2, 0
 
-	bl	randomx_calc_dataset_item_aarch64
+	bl	DECL(randomx_calc_dataset_item_aarch64)
 
 	mov	x10, sp
 	ldp	x0, x1, [sp, 64]
 	ldp	x2, x30, [sp, 80]
 	add	sp, sp, 96
 
-	b	randomx_program_aarch64_xor_with_dataset_line
+	b	DECL(randomx_program_aarch64_xor_with_dataset_line)
 
 
 
@@ -439,26 +445,26 @@ randomx_program_aarch64_light_dataset_offset:
 # x2 -> start item
 # x3 -> end item
 
-randomx_init_dataset_aarch64:
+DECL(randomx_init_dataset_aarch64):
 	# Save x30 (return address)
 	str	x30, [sp, -16]!
 
 	# Load pointer to cache memory
 	ldr	x0, [x0]
 
-randomx_init_dataset_aarch64_main_loop:
+DECL(randomx_init_dataset_aarch64_main_loop):
-	bl	randomx_calc_dataset_item_aarch64
+	bl	DECL(randomx_calc_dataset_item_aarch64)
 	add	x1, x1, 64
 	add	x2, x2, 1
 	cmp	x2, x3
-	bne	randomx_init_dataset_aarch64_main_loop
+	bne	DECL(randomx_init_dataset_aarch64_main_loop)
 
 	# Restore x30 (return address)
 	ldr	x30, [sp], 16
 
 	ret
 
-randomx_init_dataset_aarch64_end:
+DECL(randomx_init_dataset_aarch64_end):
 
 # Input parameters
 #
@@ -476,7 +482,7 @@ randomx_init_dataset_aarch64_end:
 # x12 -> temporary
 # x13 -> temporary
 
-randomx_calc_dataset_item_aarch64:
+DECL(randomx_calc_dataset_item_aarch64):
 	sub	sp, sp, 112
 	stp	x0, x1, [sp]
 	stp	x2, x3, [sp, 16]
@@ -523,7 +529,7 @@ randomx_calc_dataset_item_aarch64:
 	ldr	x12, superscalarAdd7
 	eor	x7, x0, x12
 
-	b	randomx_calc_dataset_item_aarch64_prefetch
+	b	DECL(randomx_calc_dataset_item_aarch64_prefetch)
 
 superscalarMul0: .quad 6364136223846793005
 superscalarAdd1: .quad 9298411001130361340
@@ -536,7 +542,7 @@ superscalarAdd7: .quad 9549104520008361294
 
 # Prefetch -> SuperScalar hash -> Mix will be repeated N times
 
-randomx_calc_dataset_item_aarch64_prefetch:
+DECL(randomx_calc_dataset_item_aarch64_prefetch):
 	# Actual mask will be inserted by JIT compiler
 	and	x11, x10, 1
 	add	x11, x8, x11, lsl 6
@@ -544,7 +550,7 @@ randomx_calc_dataset_item_aarch64_prefetch:
 
 	# Generated SuperScalar hash program goes here
 
-randomx_calc_dataset_item_aarch64_mix:
+DECL(randomx_calc_dataset_item_aarch64_mix):
 	ldp	x12, x13, [x11]
 	eor	x0, x0, x12
 	eor	x1, x1, x13
@@ -558,7 +564,7 @@ randomx_calc_dataset_item_aarch64_mix:
 	eor	x6, x6, x12
 	eor	x7, x7, x13
 
-randomx_calc_dataset_item_aarch64_store_result:
+DECL(randomx_calc_dataset_item_aarch64_store_result):
 	stp	x0, x1, [x9]
 	stp	x2, x3, [x9, 16]
 	stp	x4, x5, [x9, 32]
@@ -575,4 +581,4 @@ randomx_calc_dataset_item_aarch64_store_result:
 
 	ret
 
-randomx_calc_dataset_item_aarch64_end:
+DECL(randomx_calc_dataset_item_aarch64_end):

src/version.h

@@ -28,7 +28,7 @@
 #define APP_ID        "xmrig"
 #define APP_NAME      "XMRig"
 #define APP_DESC      "XMRig miner"
-#define APP_VERSION   "4.6.0-beta"
+#define APP_VERSION   "4.6.2-beta"
 #define APP_DOMAIN    "xmrig.com"
 #define APP_SITE      "www.xmrig.com"
 #define APP_COPYRIGHT "Copyright (C) 2016-2019 xmrig.com"
@@ -36,7 +36,7 @@
 
 #define APP_VER_MAJOR  4
 #define APP_VER_MINOR  6
-#define APP_VER_PATCH  0
+#define APP_VER_PATCH  2
 
 #ifdef _MSC_VER
 #   if (_MSC_VER >= 1920)