shadps4-emu/ext-cryptopp
1
0
Fork 0
mirror of https://github.com/shadps4-emu/ext-cryptopp.git synced 2024-11-23 09:59:42 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Move CPU_ProbePower7 and CPU_ProbePower8 into their own source files (GH #742)

Browse Source
This commit is contained in:
Jeffrey Walton 2018-11-18 16:35:02 -05:00
parent 1a06aadbf0
commit 3725c8411b
No known key found for this signature in database
GPG Key ID: B36AB348921B1838
5 changed files with 243 additions and 150 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
Filelist.txt
View File

@ -226,6 +226,8 @@ poly1305.cpp
poly1305.h
polynomi.cpp
polynomi.h
ppc_power7.cpp
ppc_power8.cpp
ppc_simd.h
ppc_simd.cpp
pssr.cpp

10
GNUmakefile
View File

@ -1440,10 +1440,18 @@ lea_simd.o : lea_simd.cpp
neon_simd.o : neon_simd.cpp
$(CXX) $(strip $(CXXFLAGS) $(NEON_FLAG) -c) $<
# AltiVec, Power7, Power8 available
# AltiVec available
ppc_simd.o : ppc_simd.cpp
$(CXX) $(strip $(CXXFLAGS) $(ALTIVEC_FLAG) -c) $<
# Power7 available
ppc_power7.o : ppc_power7.cpp
$(CXX) $(strip $(CXXFLAGS) $(POWER7_FLAG) -c) $<
# Power8 available
ppc_power8.o : ppc_power8.cpp
$(CXX) $(strip $(CXXFLAGS) $(POWER8_FLAG) -c) $<
# AESNI or ARMv7a/ARMv8a available
rijndael_simd.o : rijndael_simd.cpp
$(CXX) $(strip $(CXXFLAGS) $(AES_FLAG) -c) $<

88
ppc_power7.cpp Normal file
View File

@ -0,0 +1,88 @@
// ppc_power7.cpp - written and placed in the public domain by
// Jeffrey Walton, Uri Blumenthal and Marcel Raad.
//
// This source file uses intrinsics and built-ins to gain access to
// Power7 instructions. A separate source file is needed because
// additional CXXFLAGS are required to enable the appropriate
// instructions sets in some build configurations.
#include "pch.h"
#include "config.h"
#ifdef CRYPTOPP_GNU_STYLE_INLINE_ASSEMBLY
# include <signal.h>
# include <setjmp.h>
#endif
#if defined(_ARCH_PWR7)
# include "ppc_simd.h"
#endif
// Squash MS LNK4221 and libtool warnings
extern const char PPC_POWER7_FNAME[] = __FILE__;
NAMESPACE_BEGIN(CryptoPP)
// ************************* Feature Probes ************************* //
#ifdef CRYPTOPP_GNU_STYLE_INLINE_ASSEMBLY
extern "C" {
typedef void (*SigHandler)(int);
static jmp_buf s_jmpSIGILL;
static void SigIllHandler(int)
{
longjmp(s_jmpSIGILL, 1);
}
}
#endif // CRYPTOPP_MS_STYLE_INLINE_ASSEMBLY
#if (CRYPTOPP_BOOL_PPC32 || CRYPTOPP_BOOL_PPC64)
bool CPU_ProbePower7()
{
#if defined(CRYPTOPP_NO_CPU_FEATURE_PROBES)
return false;
#elif (_ARCH_PWR7)
# if defined(CRYPTOPP_GNU_STYLE_INLINE_ASSEMBLY)
// longjmp and clobber warnings. Volatile is required.
// http://github.com/weidai11/cryptopp/issues/24 and http://stackoverflow.com/q/7721854
volatile int result = false;
volatile SigHandler oldHandler = signal(SIGILL, SigIllHandler);
if (oldHandler == SIG_ERR)
return false;
volatile sigset_t oldMask;
if (sigprocmask(0, NULLPTR, (sigset_t*)&oldMask))
return false;
if (setjmp(s_jmpSIGILL))
result = false;
else
{
// POWER7 added unaligned loads and store operations
byte b1[19] = {255, 255, 255, 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, b2[17];
// Specifically call the VSX loads and stores
#if defined(__xlc__) || defined(__xlC__)
vec_xst(vec_xl(0, b1+3), 0, b2+1);
#else
vec_vsx_st(vec_vsx_ld(0, b1+3), 0, b2+1);
#endif
result = (0 == std::memcmp(b1+3, b2+1, 16));
}
sigprocmask(SIG_SETMASK, (sigset_t*)&oldMask, NULLPTR);
signal(SIGILL, oldHandler);
return result;
# endif
#else
return false;
#endif // _ARCH_PWR7
}
#endif // PPC32 or PPC64
NAMESPACE_END

144
ppc_power8.cpp Normal file
View File

@ -0,0 +1,144 @@
// ppc_power8.cpp - written and placed in the public domain by
// Jeffrey Walton, Uri Blumenthal and Marcel Raad.
//
// This source file uses intrinsics and built-ins to gain access to
// Power8 instructions. A separate source file is needed because
// additional CXXFLAGS are required to enable the appropriate
// instructions sets in some build configurations.
#include "pch.h"
#include "config.h"
#ifdef CRYPTOPP_GNU_STYLE_INLINE_ASSEMBLY
# include <signal.h>
# include <setjmp.h>
#endif
#if defined(_ARCH_PWR8) || defined(__CRYPTO__)
# include "ppc_simd.h"
#endif
// Squash MS LNK4221 and libtool warnings
extern const char PPC_POWER8_FNAME[] = __FILE__;
NAMESPACE_BEGIN(CryptoPP)
// ************************* Feature Probes ************************* //
#ifdef CRYPTOPP_GNU_STYLE_INLINE_ASSEMBLY
extern "C" {
typedef void (*SigHandler)(int);
static jmp_buf s_jmpSIGILL;
static void SigIllHandler(int)
{
longjmp(s_jmpSIGILL, 1);
}
}
#endif // CRYPTOPP_MS_STYLE_INLINE_ASSEMBLY
#if (CRYPTOPP_BOOL_PPC32 || CRYPTOPP_BOOL_PPC64)
bool CPU_ProbePower8()
{
#if defined(CRYPTOPP_NO_CPU_FEATURE_PROBES)
return false;
#elif (_ARCH_PWR8)
# if defined(CRYPTOPP_GNU_STYLE_INLINE_ASSEMBLY)
// longjmp and clobber warnings. Volatile is required.
// http://github.com/weidai11/cryptopp/issues/24 and http://stackoverflow.com/q/7721854
volatile int result = true;
volatile SigHandler oldHandler = signal(SIGILL, SigIllHandler);
if (oldHandler == SIG_ERR)
return false;
volatile sigset_t oldMask;
if (sigprocmask(0, NULLPTR, (sigset_t*)&oldMask))
return false;
if (setjmp(s_jmpSIGILL))
result = false;
else
{
// POWER8 added 64-bit SIMD operations
const word64 x = W64LIT(0xffffffffffffffff);
word64 w1[2] = {x, x}, w2[2] = {4, 6}, w3[2];
// Specifically call the VSX loads and stores
#if defined(__xlc__) || defined(__xlC__)
const uint64x2_p v1 = (uint64x2_p)vec_xl(0, (byte*)w1);
const uint64x2_p v2 = (uint64x2_p)vec_xl(0, (byte*)w2);
const uint64x2_p v3 = VecAdd(v1, v2); // 64-bit add
vec_xst((uint8x16_p)v3, 0, (byte*)w3);
#else
const uint64x2_p v1 = (uint64x2_p)vec_vsx_ld(0, (byte*)w1);
const uint64x2_p v2 = (uint64x2_p)vec_vsx_ld(0, (byte*)w2);
const uint64x2_p v3 = VecAdd(v1, v2); // 64-bit add
vec_vsx_st((uint8x16_p)v3, 0, (byte*)w3);
#endif
// Relies on integer wrap
result = (w3[0] == 3 && w3[1] == 5);
}
sigprocmask(SIG_SETMASK, (sigset_t*)&oldMask, NULLPTR);
signal(SIGILL, oldHandler);
return result;
# endif
#else
return false;
#endif // _ARCH_PWR8
}
bool CPU_ProbeAES()
{
#if defined(CRYPTOPP_NO_CPU_FEATURE_PROBES)
return false;
#elif (__CRYPTO__)
# if defined(CRYPTOPP_GNU_STYLE_INLINE_ASSEMBLY)
// longjmp and clobber warnings. Volatile is required.
// http://github.com/weidai11/cryptopp/issues/24 and http://stackoverflow.com/q/7721854
volatile int result = true;
volatile SigHandler oldHandler = signal(SIGILL, SigIllHandler);
if (oldHandler == SIG_ERR)
return false;
volatile sigset_t oldMask;
if (sigprocmask(0, NULLPTR, (sigset_t*)&oldMask))
return false;
if (setjmp(s_jmpSIGILL))
result = false;
else
{
byte key[16] = {0xA0, 0xFA, 0xFE, 0x17, 0x88, 0x54, 0x2c, 0xb1,
0x23, 0xa3, 0x39, 0x39, 0x2a, 0x6c, 0x76, 0x05};
byte state[16] = {0x19, 0x3d, 0xe3, 0xb3, 0xa0, 0xf4, 0xe2, 0x2b,
0x9a, 0xc6, 0x8d, 0x2a, 0xe9, 0xf8, 0x48, 0x08};
byte r[16] = {255}, z[16] = {};
uint8x16_p k = (uint8x16_p)VecLoad(0, key);
uint8x16_p s = (uint8x16_p)VecLoad(0, state);
s = VecEncrypt(s, k);
s = VecEncryptLast(s, k);
s = VecDecrypt(s, k);
s = VecDecryptLast(s, k);
VecStore(s, r);
result = (0 != std::memcmp(r, z, 16));
}
sigprocmask(SIG_SETMASK, (sigset_t*)&oldMask, NULLPTR);
signal(SIGILL, oldHandler);
return result;
# endif
#else
return false;
#endif // __CRYPTO__
}
#endif // PPC32 or PPC64
NAMESPACE_END

149
rijndael_simd.cpp
View File

@ -137,155 +137,6 @@ bool CPU_ProbeAES()
}
#endif // ARM32 or ARM64
#if (CRYPTOPP_BOOL_PPC32 || CRYPTOPP_BOOL_PPC64)
bool CPU_ProbePower7()
{
#if defined(CRYPTOPP_NO_CPU_FEATURE_PROBES)
return false;
#elif (CRYPTOPP_POWER7_AVAILABLE) || (CRYPTOPP_POWER8_AVAILABLE)
# if defined(CRYPTOPP_GNU_STYLE_INLINE_ASSEMBLY)
// longjmp and clobber warnings. Volatile is required.
// http://github.com/weidai11/cryptopp/issues/24 and http://stackoverflow.com/q/7721854
volatile int result = false;
volatile SigHandler oldHandler = signal(SIGILL, SigIllHandler);
if (oldHandler == SIG_ERR)
return false;
volatile sigset_t oldMask;
if (sigprocmask(0, NULLPTR, (sigset_t*)&oldMask))
return false;
if (setjmp(s_jmpSIGILL))
result = false;
else
{
// POWER7 added unaligned loads and store operations
byte b1[19] = {255, 255, 255, 1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1}, b2[17];
// Specifically call the VSX loads and stores
#if defined(__xlc__) || defined(__xlC__)
vec_xst(vec_xl(0, b1+3), 0, b2+1);
#else
vec_vsx_st(vec_vsx_ld(0, b1+3), 0, b2+1);
#endif
result = (0 == std::memcmp(b1+3, b2+1, 16));
}
sigprocmask(SIG_SETMASK, (sigset_t*)&oldMask, NULLPTR);
signal(SIGILL, oldHandler);
return result;
# endif
#else
return false;
#endif // CRYPTOPP_POWER7_AVAILABLE
}
bool CPU_ProbePower8()
{
#if defined(CRYPTOPP_NO_CPU_FEATURE_PROBES)
return false;
#elif (CRYPTOPP_POWER8_AVAILABLE)
# if defined(CRYPTOPP_GNU_STYLE_INLINE_ASSEMBLY)
// longjmp and clobber warnings. Volatile is required.
// http://github.com/weidai11/cryptopp/issues/24 and http://stackoverflow.com/q/7721854
volatile int result = true;
volatile SigHandler oldHandler = signal(SIGILL, SigIllHandler);
if (oldHandler == SIG_ERR)
return false;
volatile sigset_t oldMask;
if (sigprocmask(0, NULLPTR, (sigset_t*)&oldMask))
return false;
if (setjmp(s_jmpSIGILL))
result = false;
else
{
// POWER8 added 64-bit SIMD operations
const word64 x = W64LIT(0xffffffffffffffff);
word64 w1[2] = {x, x}, w2[2] = {4, 6}, w3[2];
// Specifically call the VSX loads and stores
#if defined(__xlc__) || defined(__xlC__)
const uint64x2_p v1 = (uint64x2_p)vec_xl(0, (byte*)w1);
const uint64x2_p v2 = (uint64x2_p)vec_xl(0, (byte*)w2);
const uint64x2_p v3 = VecAdd(v1, v2); // 64-bit add
vec_xst((uint8x16_p)v3, 0, (byte*)w3);
#else
const uint64x2_p v1 = (uint64x2_p)vec_vsx_ld(0, (byte*)w1);
const uint64x2_p v2 = (uint64x2_p)vec_vsx_ld(0, (byte*)w2);
const uint64x2_p v3 = VecAdd(v1, v2); // 64-bit add
vec_vsx_st((uint8x16_p)v3, 0, (byte*)w3);
#endif
// Relies on integer wrap
result = (w3[0] == 3 && w3[1] == 5);
}
sigprocmask(SIG_SETMASK, (sigset_t*)&oldMask, NULLPTR);
signal(SIGILL, oldHandler);
return result;
# endif
#else
return false;
#endif // CRYPTOPP_POWER8_AVAILABLE
}
bool CPU_ProbeAES()
{
#if defined(CRYPTOPP_NO_CPU_FEATURE_PROBES)
return false;
#elif (CRYPTOPP_POWER8_AES_AVAILABLE)
# if defined(CRYPTOPP_GNU_STYLE_INLINE_ASSEMBLY)
// longjmp and clobber warnings. Volatile is required.
// http://github.com/weidai11/cryptopp/issues/24 and http://stackoverflow.com/q/7721854
volatile int result = true;
volatile SigHandler oldHandler = signal(SIGILL, SigIllHandler);
if (oldHandler == SIG_ERR)
return false;
volatile sigset_t oldMask;
if (sigprocmask(0, NULLPTR, (sigset_t*)&oldMask))
return false;
if (setjmp(s_jmpSIGILL))
result = false;
else
{
byte key[16] = {0xA0, 0xFA, 0xFE, 0x17, 0x88, 0x54, 0x2c, 0xb1,
0x23, 0xa3, 0x39, 0x39, 0x2a, 0x6c, 0x76, 0x05};
byte state[16] = {0x19, 0x3d, 0xe3, 0xb3, 0xa0, 0xf4, 0xe2, 0x2b,
0x9a, 0xc6, 0x8d, 0x2a, 0xe9, 0xf8, 0x48, 0x08};
byte r[16] = {255}, z[16] = {};
uint8x16_p k = (uint8x16_p)VecLoad(0, key);
uint8x16_p s = (uint8x16_p)VecLoad(0, state);
s = VecEncrypt(s, k);
s = VecEncryptLast(s, k);
s = VecDecrypt(s, k);
s = VecDecryptLast(s, k);
VecStore(s, r);
result = (0 != std::memcmp(r, z, 16));
}
sigprocmask(SIG_SETMASK, (sigset_t*)&oldMask, NULLPTR);
signal(SIGILL, oldHandler);
return result;
# endif
#else
return false;
#endif // CRYPTOPP_POWER8_AES_AVAILABLE
}
#endif // PPC32 or PPC64
// ***************************** ARMv8 ***************************** //
#if (CRYPTOPP_ARM_AES_AVAILABLE)