Merge https://github.com/xmrig/xmrig

CHANGELOG.md

@@ -1,3 +1,13 @@
+# v5.5.3
+- [#1529](https://github.com/xmrig/xmrig/pull/1529) Fixed crash on Bulldozer CPUs.
+
+# v5.5.2
+- [#1500](https://github.com/xmrig/xmrig/pull/1500) Removed unnecessary code from RandomX JIT compiler.
+- [#1502](https://github.com/xmrig/xmrig/pull/1502) Optimizations for AMD Bulldozer.
+- [#1508](https://github.com/xmrig/xmrig/pull/1508) Added support for BMI2 instructions.
+- [#1510](https://github.com/xmrig/xmrig/pull/1510) Optimized `CFROUND` instruction for RandomX.
+- [#1520](https://github.com/xmrig/xmrig/pull/1520) Fixed thread affinity.
+
 # v5.5.1
 - [#1469](https://github.com/xmrig/xmrig/issues/1469) Fixed build with gcc 4.8.
 - [#1473](https://github.com/xmrig/xmrig/pull/1473) Added RandomX auto-config for mobile Ryzen APUs.

src/backend/cpu/interfaces/ICpuInfo.h

@@ -5,8 +5,8 @@
  * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
  * Copyright 2016      Jay D Dee   <[email protected]>
  * Copyright 2017-2019 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
- * Copyright 2018-2019 SChernykh   <https://github.com/SChernykh>
- * Copyright 2016-2019 XMRig       <[email protected]>
+ * Copyright 2018-2020 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2020 XMRig       <[email protected]>
  *
  *   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
@@ -62,6 +62,7 @@ class ICpuInfo
     virtual Assembly::Id assembly() const                                           = 0;
     virtual bool hasAES() const                                                     = 0;
     virtual bool hasAVX2() const                                                    = 0;
+    virtual bool hasBMI2() const                                                    = 0;
     virtual bool hasOneGbPages() const                                              = 0;
     virtual const char *backend() const                                             = 0;
     virtual const char *brand() const                                               = 0;

src/backend/cpu/platform/AdvancedCpuInfo.cpp

@@ -5,8 +5,8 @@
  * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
  * Copyright 2016      Jay D Dee   <[email protected]>
  * Copyright 2017-2019 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
- * Copyright 2018-2019 SChernykh   <https://github.com/SChernykh>
- * Copyright 2016-2019 XMRig       <[email protected]>
+ * Copyright 2018-2020 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2020 XMRig       <[email protected]>
  *
  *   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
@@ -153,6 +153,7 @@ xmrig::AdvancedCpuInfo::AdvancedCpuInfo() :
     }
 
     m_avx2 = data.flags[CPU_FEATURE_AVX2] && data.flags[CPU_FEATURE_OSXSAVE];
+    m_bmi2 = data.flags[CPU_FEATURE_BMI2];
 }
 
 

src/backend/cpu/platform/AdvancedCpuInfo.h

@@ -5,8 +5,8 @@
  * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
  * Copyright 2016      Jay D Dee   <[email protected]>
  * Copyright 2017-2019 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
- * Copyright 2018-2019 SChernykh   <https://github.com/SChernykh>
- * Copyright 2016-2019 XMRig       <[email protected]>
+ * Copyright 2018-2020 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2020 XMRig       <[email protected]>
  *
  *   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
@@ -43,6 +43,7 @@ class AdvancedCpuInfo : public ICpuInfo
     inline Assembly::Id assembly() const override   { return m_assembly; }
     inline bool hasAES() const override             { return m_aes; }
     inline bool hasAVX2() const override            { return m_avx2; }
+    inline bool hasBMI2() const override            { return m_bmi2; }
     inline bool hasOneGbPages() const override      { return m_pdpe1gb; }
     inline const char *backend() const override     { return m_backend; }
     inline const char *brand() const override       { return m_brand; }
@@ -59,6 +60,7 @@ class AdvancedCpuInfo : public ICpuInfo
     Assembly m_assembly;
     bool m_aes            = false;
     bool m_avx2           = false;
+    bool m_bmi2           = false;
     bool m_L2_exclusive   = false;
     char m_backend[32]{};
     char m_brand[64 + 5]{};

src/backend/cpu/platform/BasicCpuInfo.cpp

@@ -5,8 +5,8 @@
  * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
  * Copyright 2016      Jay D Dee   <[email protected]>
  * Copyright 2017-2019 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
- * Copyright 2018-2019 SChernykh   <https://github.com/SChernykh>
- * Copyright 2016-2019 XMRig       <[email protected]>
+ * Copyright 2018-2020 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2020 XMRig       <[email protected]>
  *
  *   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
@@ -45,6 +45,10 @@
 #   define bit_AVX2 (1 << 5)
 #endif
 
+#ifndef bit_BMI2
+#   define bit_BMI2 (1 << 8)
+#endif
+
 #ifndef bit_PDPE1GB
 #   define bit_PDPE1GB (1 << 26)
 #endif
@@ -141,6 +145,12 @@ static inline bool has_avx2()
 }
 
 
+static inline bool has_bmi2()
+{
+    return has_feature(EXTENDED_FEATURES, EBX_Reg, bit_BMI2);
+}
+
+
 static inline bool has_pdpe1gb()
 {
     return has_feature(PROCESSOR_EXT_INFO, EDX_Reg, bit_PDPE1GB);
@@ -154,6 +164,7 @@ xmrig::BasicCpuInfo::BasicCpuInfo() :
     m_threads(std::thread::hardware_concurrency()),
     m_aes(has_aes_ni()),
     m_avx2(has_avx2()),
+    m_bmi2(has_bmi2()),
     m_pdpe1gb(has_pdpe1gb())
 {
     cpu_brand_string(m_brand);

src/backend/cpu/platform/BasicCpuInfo.h

@@ -5,8 +5,8 @@
  * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
  * Copyright 2016      Jay D Dee   <[email protected]>
  * Copyright 2017-2019 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
- * Copyright 2018-2019 SChernykh   <https://github.com/SChernykh>
- * Copyright 2016-2019 XMRig       <[email protected]>
+ * Copyright 2018-2020 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2020 XMRig       <[email protected]>
  *
  *   This program is free software: you can redistribute it and/or modify
  *   it under the terms of the GNU General Public License as published by
@@ -44,6 +44,7 @@ class BasicCpuInfo : public ICpuInfo
     inline Assembly::Id assembly() const override   { return m_assembly; }
     inline bool hasAES() const override             { return m_aes; }
     inline bool hasAVX2() const override            { return m_avx2; }
+    inline bool hasBMI2() const override            { return m_bmi2; }
     inline bool hasOneGbPages() const override      { return m_pdpe1gb; }
     inline const char *brand() const override       { return m_brand; }
     inline MsrMod msrMod() const override           { return m_msrMod; }
@@ -63,6 +64,7 @@ class BasicCpuInfo : public ICpuInfo
     Assembly m_assembly     = Assembly::NONE;
     bool m_aes              = false;
     const bool m_avx2       = false;
+    const bool m_bmi2       = false;
     const bool m_pdpe1gb    = false;
     MsrMod m_msrMod         = MSR_MOD_NONE;
     Vendor m_vendor         = VENDOR_UNKNOWN;

src/base/kernel/Platform.h

@@ -5,8 +5,8 @@
  * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
  * Copyright 2016      Jay D Dee   <[email protected]>
  * 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>, <[email protected]>
+ * Copyright 2018-2020 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2020 XMRig       <https://github.com/xmrig>, <[email protected]>
  *
  *   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

src/base/kernel/Platform_hwloc.cpp

@@ -6,8 +6,8 @@
  * Copyright 2016      Jay D Dee   <[email protected]>
  * 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>, <[email protected]>
+ * Copyright 2018-2020 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2020 XMRig       <https://github.com/xmrig>, <[email protected]>
  *
  *   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
@@ -30,6 +30,7 @@
 
 
 #include <hwloc.h>
+#include <thread>
 
 
 bool xmrig::Platform::setThreadAffinity(uint64_t cpu_id)
@@ -42,8 +43,11 @@ bool xmrig::Platform::setThreadAffinity(uint64_t cpu_id)
     }
 
     if (hwloc_set_cpubind(cpu->topology(), pu->cpuset, HWLOC_CPUBIND_THREAD | HWLOC_CPUBIND_STRICT) >= 0) {
+        std::this_thread::sleep_for(std::chrono::milliseconds(1));
         return true;
     }
 
-    return hwloc_set_cpubind(cpu->topology(), pu->cpuset, HWLOC_CPUBIND_THREAD) >= 0;
+    const bool result = (hwloc_set_cpubind(cpu->topology(), pu->cpuset, HWLOC_CPUBIND_THREAD) >= 0);
+    std::this_thread::sleep_for(std::chrono::milliseconds(1));
+    return result;
 }

src/base/kernel/Platform_mac.cpp

@@ -4,8 +4,8 @@
  * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
  * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
  * Copyright 2016      Jay D Dee   <[email protected]>
- * Copyright 2016-2017 XMRig       <support@xmrig.com>
- *
+ * Copyright 2018-2020 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2020 XMRig       <https://github.com/xmrig>, <[email protected]>
  *
  *   This program is free software: you can redistribute it and/or modify
  *   it under the terms of the GNU General Public License as published by
@@ -28,6 +28,7 @@
 #include <stdlib.h>
 #include <sys/resource.h>
 #include <uv.h>
+#include <thread>
 
 
 #include "base/kernel/Platform.h"
@@ -67,7 +68,9 @@ bool xmrig::Platform::setThreadAffinity(uint64_t cpu_id)
     thread_affinity_policy_data_t policy = { static_cast<integer_t>(cpu_id) };
     mach_thread = pthread_mach_thread_np(pthread_self());
 
-    return thread_policy_set(mach_thread, THREAD_AFFINITY_POLICY, (thread_policy_t)&policy, 1) == KERN_SUCCESS;
+    const bool result = (thread_policy_set(mach_thread, THREAD_AFFINITY_POLICY, (thread_policy_t)&policy, 1) == KERN_SUCCESS);
+    std::this_thread::sleep_for(std::chrono::milliseconds(1));
+    return result;
 }
 #endif
 

src/base/kernel/Platform_unix.cpp

@@ -4,8 +4,8 @@
  * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
  * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
  * Copyright 2016      Jay D Dee   <[email protected]>
- * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
- * Copyright 2016-2018 XMRig       <https://github.com/xmrig>, <[email protected]>
+ * Copyright 2018-2020 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2020 XMRig       <https://github.com/xmrig>, <[email protected]>
  *
  *   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
@@ -37,6 +37,7 @@
 #include <sys/resource.h>
 #include <unistd.h>
 #include <uv.h>
+#include <thread>
 
 
 #include "base/kernel/Platform.h"
@@ -92,10 +93,13 @@ bool xmrig::Platform::setThreadAffinity(uint64_t cpu_id)
     CPU_SET(cpu_id, &mn);
 
 #   ifndef __ANDROID__
-    return pthread_setaffinity_np(pthread_self(), sizeof(cpu_set_t), &mn) == 0;
+    const bool result = (pthread_setaffinity_np(pthread_self(), sizeof(cpu_set_t), &mn) == 0);
 #   else
-    return sched_setaffinity(gettid(), sizeof(cpu_set_t), &mn) == 0;
+    const bool result = (sched_setaffinity(gettid(), sizeof(cpu_set_t), &mn) == 0);
 #   endif
+
+    std::this_thread::sleep_for(std::chrono::milliseconds(1));
+    return result;
 }
 #endif
 

src/base/kernel/Platform_win.cpp

@@ -4,9 +4,8 @@
  * Copyright 2014      Lucas Jones <https://github.com/lucasjones>
  * Copyright 2014-2016 Wolf9466    <https://github.com/OhGodAPet>
  * Copyright 2016      Jay D Dee   <[email protected]>
- * Copyright 2017-2018 XMR-Stak    <https://github.com/fireice-uk>, <https://github.com/psychocrypt>
- * Copyright 2018      SChernykh   <https://github.com/SChernykh>
- * Copyright 2016-2019 XMRig       <https://github.com/xmrig>, <[email protected]>
+ * Copyright 2018-2020 SChernykh   <https://github.com/SChernykh>
+ * Copyright 2016-2020 XMRig       <https://github.com/xmrig>, <[email protected]>
  *
  *   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
@@ -98,7 +97,9 @@ bool xmrig::Platform::setThreadAffinity(uint64_t cpu_id)
         LOG_ERR("Unable to set affinity. Windows supports only affinity up to 63.");
     }
 
-    return SetThreadAffinityMask(GetCurrentThread(), 1ULL << cpu_id) != 0;
+    const bool result = (SetThreadAffinityMask(GetCurrentThread(), 1ULL << cpu_id) != 0);
+    Sleep(1);
+    return result;
 }
 #endif
 

src/crypto/randomx/asm/program_epilogue_store.inc

@@ -1,5 +1,5 @@
 	;# save VM register values
-	add rsp, 24
+	add rsp, 40
 	pop rcx
 	mov qword ptr [rcx+0], r8
 	mov qword ptr [rcx+8], r9

src/crypto/randomx/asm/program_loop_load.inc

@@ -1,5 +1,5 @@
 	lea rcx, [rsi+rax]
-	mov [rsp+8], rcx
+	mov [rsp+16], rcx
 	xor r8,  qword ptr [rcx+0]
 	xor r9,  qword ptr [rcx+8]
 	xor r10, qword ptr [rcx+16]
@@ -9,7 +9,7 @@
 	xor r14, qword ptr [rcx+48]
 	xor r15, qword ptr [rcx+56]
 	lea rcx, [rsi+rdx]
-	mov [rsp+16], rcx
+	mov [rsp+24], rcx
 	cvtdq2pd xmm0, qword ptr [rcx+0]
 	cvtdq2pd xmm1, qword ptr [rcx+8]
 	cvtdq2pd xmm2, qword ptr [rcx+16]
@@ -18,11 +18,11 @@
 	cvtdq2pd xmm5, qword ptr [rcx+40]
 	cvtdq2pd xmm6, qword ptr [rcx+48]
 	cvtdq2pd xmm7, qword ptr [rcx+56]
-	andps xmm4, xmm13
-	andps xmm5, xmm13
-	andps xmm6, xmm13
-	andps xmm7, xmm13
-	orps xmm4, xmm14
-	orps xmm5, xmm14
-	orps xmm6, xmm14
-	orps xmm7, xmm14
+	andpd xmm4, xmm13
+	andpd xmm5, xmm13
+	andpd xmm6, xmm13
+	andpd xmm7, xmm13
+	orpd xmm4, xmm14
+	orpd xmm5, xmm14
+	orpd xmm6, xmm14
+	orpd xmm7, xmm14

src/crypto/randomx/asm/program_loop_load_xop.inc

@@ -0,0 +1,24 @@
+	lea rcx, [rsi+rax]
+	mov [rsp+16], rcx
+	xor r8,  qword ptr [rcx+0]
+	xor r9,  qword ptr [rcx+8]
+	xor r10, qword ptr [rcx+16]
+	xor r11, qword ptr [rcx+24]
+	xor r12, qword ptr [rcx+32]
+	xor r13, qword ptr [rcx+40]
+	xor r14, qword ptr [rcx+48]
+	xor r15, qword ptr [rcx+56]
+	lea rcx, [rsi+rdx]
+	mov [rsp+24], rcx
+	cvtdq2pd xmm0, qword ptr [rcx+0]
+	cvtdq2pd xmm1, qword ptr [rcx+8]
+	cvtdq2pd xmm2, qword ptr [rcx+16]
+	cvtdq2pd xmm3, qword ptr [rcx+24]
+	cvtdq2pd xmm4, qword ptr [rcx+32]
+	cvtdq2pd xmm5, qword ptr [rcx+40]
+	cvtdq2pd xmm6, qword ptr [rcx+48]
+	cvtdq2pd xmm7, qword ptr [rcx+56]
+	vpcmov xmm4, xmm4, xmm14, xmm13
+	vpcmov xmm5, xmm5, xmm14, xmm13
+	vpcmov xmm6, xmm6, xmm14, xmm13
+	vpcmov xmm7, xmm7, xmm14, xmm13

src/crypto/randomx/asm/program_loop_store.inc

@@ -1,4 +1,4 @@
-	mov rcx, [rsp+16]
+	mov rcx, [rsp+24]
 	mov qword ptr [rcx+0], r8
 	mov qword ptr [rcx+8], r9
 	mov qword ptr [rcx+16], r10
@@ -7,7 +7,7 @@
 	mov qword ptr [rcx+40], r13
 	mov qword ptr [rcx+48], r14
 	mov qword ptr [rcx+56], r15
-	mov rcx, [rsp+8]
+	mov rcx, [rsp+16]
 	xorpd xmm0, xmm4
 	xorpd xmm1, xmm5
 	xorpd xmm2, xmm6

src/crypto/randomx/asm/program_xmm_constants.inc

@@ -1,5 +1,5 @@
 mantissaMask:
-	db 255, 255, 255, 255, 255, 255, 255, 0, 255, 255, 255, 255, 255, 255, 255, 0
+	db 0, 0, 192, 255, 255, 255, 255, 0, 0, 0, 192, 255, 255, 255, 255, 0
 exp240:
 	db 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
 scaleMask: