diff options
| -rwxr-xr-x | cpu.cpp | 172 | ||||
| -rwxr-xr-x | cpu.h | 111 |
2 files changed, 283 insertions, 0 deletions
@@ -0,0 +1,172 @@ +// cpu.cpp - written and placed in the public domain by Wei Dai + +#include "pch.h" +#include "cpu.h" + +#include "misc.h" +#include <algorithm> + +#ifdef __GNUC__ +#include <signal.h> +#include <setjmp.h> +#endif + +NAMESPACE_BEGIN(CryptoPP) + +#ifdef CRYPTOPP_X86_ASM_AVAILABLE + +typedef void (*SigHandler)(int); + +static jmp_buf s_jmpNoCPUID; +static void SigIllHandlerCPUID(int) +{ + longjmp(s_jmpNoCPUID, 1); +} + +bool CpuId(word32 input, word32 *output) +{ +#ifdef _MSC_VER + __try + { + __asm + { + mov eax, input + cpuid + mov edi, output + mov [edi], eax + mov [edi+4], ebx + mov [edi+8], ecx + mov [edi+12], edx + } + } + __except (1) + { + return false; + } + return true; +#else + SigHandler oldHandler = signal(SIGILL, SigIllHandlerCPUID); + if (oldHandler == SIG_ERR) + return false; + + bool result = true; + if (setjmp(s_jmpNoCPUID)) + result = false; + else + { + __asm__ + ( + // save ebx in case -fPIC is being used + "push %%ebx; cpuid; mov %%ebx, %%edi; pop %%ebx" + : "=a" (output[0]), "=D" (output[1]), "=c" (output[2]), "=d" (output[3]) + : "a" (input) + ); + } + + signal(SIGILL, oldHandler); + return result; +#endif +} + +#ifndef _MSC_VER +static jmp_buf s_jmpNoSSE2; +static void SigIllHandlerSSE2(int) +{ + longjmp(s_jmpNoSSE2, 1); +} +#endif + +#elif _MSC_VER >= 1400 + +bool CpuId(word32 input, word32 *output) +{ + __cpuid((int *)output, input); + return true; +} + +inline bool TrySSE2() +{ + return true; +} + +#endif + +#ifdef CRYPTOPP_CPUID_AVAILABLE + +static bool TrySSE2() +{ +#ifdef _MSC_VER + __try + { +#if CRYPTOPP_BOOL_SSE2_ASM_AVAILABLE + __asm por xmm0, xmm0 // executing SSE2 instruction +#elif CRYPTOPP_BOOL_SSE2_INTRINSICS_AVAILABLE + __mm128i x = _mm_setzero_si128(); + return _mm_cvtsi128_si32(x) == 0; +#endif + } + __except (1) + { + return false; + } + return true; +#elif defined(__GNUC__) + SigHandler oldHandler = signal(SIGILL, SigIllHandlerSSE2); + if (oldHandler == SIG_ERR) + return false; + + bool result = true; + if (setjmp(s_jmpNoSSE2)) + result = false; + else + { +#if CRYPTOPP_BOOL_SSE2_ASM_AVAILABLE + __asm __volatile ("por %xmm0, %xmm0"); +#elif CRYPTOPP_BOOL_SSE2_INTRINSICS_AVAILABLE + __mm128i x = _mm_setzero_si128(); + result = _mm_cvtsi128_si32(x) == 0; +#endif + } + + signal(SIGILL, oldHandler); + return result; +#else + return false; +#endif +} + +bool g_x86DetectionDone = false; +bool g_hasSSE2 = false, g_hasSSSE3 = false, g_hasMMX = false, g_isP4 = false; +int g_cacheLineSize = CRYPTOPP_L1_CACHE_LINE_SIZE; + +void DetectX86Features() +{ + word32 cpuid[4], cpuid1[4]; + if (!CpuId(0, cpuid)) + return; + if (!CpuId(1, cpuid1)) + return; + + g_hasMMX = (cpuid1[3] & (1 << 23)) != 0; + if ((cpuid1[3] & (1 << 26)) != 0) + g_hasSSE2 = TrySSE2(); + g_hasSSSE3 = g_hasSSE2 && (cpuid1[2] & (1<<9)); + + std::swap(cpuid[2], cpuid[3]); + if (memcmp(cpuid+1, "GenuineIntel", 12) == 0) + { + g_isP4 = ((cpuid1[0] >> 8) & 0xf) == 0xf; + g_cacheLineSize = 8 * GETBYTE(cpuid1[1], 1); + } + else if (memcmp(cpuid+1, "AuthenticAMD", 12) == 0) + { + CpuId(0x80000005, cpuid); + g_cacheLineSize = GETBYTE(cpuid[2], 0); + } + + g_x86DetectionDone = true; +} + +#endif + +NAMESPACE_END @@ -0,0 +1,111 @@ +#ifndef CRYPTOPP_CPU_H +#define CRYPTOPP_CPU_H + +#include "config.h" + +NAMESPACE_BEGIN(CryptoPP) + +#if defined(CRYPTOPP_X86_ASM_AVAILABLE) || _MSC_VER >= 1400 + +#define CRYPTOPP_CPUID_AVAILABLE + +// these should not be used directly +extern bool g_x86DetectionDone; +extern bool g_hasSSE2, g_hasMMX, g_hasSSSE3, g_isP4; +extern int g_cacheLineSize; +void DetectX86Features(); + +bool CpuId(word32 input, word32 *output); + +#if !CRYPTOPP_BOOL_X64 + +inline bool HasSSE2() +{ + if (!g_x86DetectionDone) + DetectX86Features(); + return g_hasSSE2; +} + +inline bool HasMMX() +{ + if (!g_x86DetectionDone) + DetectX86Features(); + return g_hasMMX; +} + +#endif + +inline bool HasSSSE3() +{ + if (!g_x86DetectionDone) + DetectX86Features(); + return g_hasSSSE3; +} + +inline bool IsP4() +{ + if (!g_x86DetectionDone) + DetectX86Features(); + return g_isP4; +} + +inline int GetCacheLineSize() +{ + if (!g_x86DetectionDone) + DetectX86Features(); + return g_cacheLineSize; +} + +#else + +inline int GetCacheLineSize() +{ + return CRYPTOPP_L1_CACHE_LINE_SIZE; +} + +#endif // #ifdef CRYPTOPP_X86_ASM_AVAILABLE || _MSC_VER >= 1400 + +#if CRYPTOPP_BOOL_X64 + +inline bool HasSSE2() +{ + return true; +} + +inline bool HasMMX() +{ + return true; +} + +#endif + +#if defined(__GNUC__) + // define these in two steps to allow arguments to be expanded + #define GNU_AS1(x) #x ";" + #define GNU_AS2(x, y) #x ", " #y ";" + #define GNU_AS3(x, y, z) #x ", " #y ", " #z ";" + #define GNU_ASL(x) "\n" #x ":" + #define GNU_ASJ(x, y, z) #x " " #y #z ";" + #define AS1(x) GNU_AS1(x) + #define AS2(x, y) GNU_AS2(x, y) + #define AS3(x, y, z) GNU_AS3(x, y, z) + #define ASS(x, y, a, b, c, d) #x ", " #y ", " #a "*64+" #b "*16+" #c "*4+" #d ";" + #define ASL(x) GNU_ASL(x) + #define ASJ(x, y, z) GNU_ASJ(x, y, z) + #define ASC(x, y) #x " " #y ";" +#else + #define AS1(x) __asm {x} + #define AS2(x, y) __asm {x, y} + #define AS3(x, y, z) __asm {x, y, z} + #define ASS(x, y, a, b, c, d) __asm {x, y, _MM_SHUFFLE(a, b, c, d)} + #define ASL(x) __asm {label##x:} + #define ASJ(x, y, z) __asm {x label##y} + #define ASC(x, y) __asm {x label##y} +#endif + +// GNU assembler doesn't seem to have mod operator +#define ASM_MOD(x, y) ((x)-((x)/(y))*(y)) + +NAMESPACE_END + +#endif |