shadps4-emu/ext-cryptopp
1
0
Fork 0
mirror of https://github.com/shadps4-emu/ext-cryptopp.git synced 2025-02-13 00:41:07 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Fix xorbuf compile when CRYPTOPP_DISABLE_ASM

Browse Source
This commit is contained in:
Jeffrey Walton 2021-03-18 12:05:36 -04:00
parent 315bfc6830
commit bb5b48d1dc
No known key found for this signature in database
GPG Key ID: B36AB348921B1838
1 changed files with 26 additions and 20 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

46
misc.cpp
View File

@ -19,15 +19,17 @@
#include "integer.h" #include "integer.h"
#include "secblock.h" #include "secblock.h"
#if defined(__AVX__) || defined(__SSE2__) #ifndef CRYPTOPP_DISABLE_ASM
# include <immintrin.h> # if defined(__AVX__) || defined(__SSE2__)
#endif # include <immintrin.h>
#if defined(__aarch64__) || defined(__aarch32__) || defined(_M_ARM64)
# if defined(CRYPTOPP_ARM_NEON_HEADER)
# include <arm_neon.h>
# endif # endif
#endif
# if defined(__aarch64__) || defined(__aarch32__) || defined(_M_ARM64)
# if defined(CRYPTOPP_ARM_NEON_HEADER)
# include <arm_neon.h>
# endif
# endif
#endif // CRYPTOPP_DISABLE_ASM
NAMESPACE_BEGIN(CryptoPP) NAMESPACE_BEGIN(CryptoPP)
@ -60,7 +62,8 @@ void xorbuf(byte *buf, const byte *mask, size_t count)
CRYPTOPP_ASSERT(mask != NULLPTR); CRYPTOPP_ASSERT(mask != NULLPTR);
CRYPTOPP_ASSERT(count > 0); CRYPTOPP_ASSERT(count > 0);
#if defined(__AVX__) #ifndef CRYPTOPP_DISABLE_ASM
# if defined(__AVX__)
while (count >= 32) while (count >= 32)
{ {
__m256i b = _mm256_loadu_si256(reinterpret_cast<const __m256i*>(buf)); __m256i b = _mm256_loadu_si256(reinterpret_cast<const __m256i*>(buf));
@ -71,8 +74,8 @@ void xorbuf(byte *buf, const byte *mask, size_t count)
} }
// https://software.intel.com/en-us/articles/avoiding-avx-sse-transition-penalties // https://software.intel.com/en-us/articles/avoiding-avx-sse-transition-penalties
_mm256_zeroupper(); _mm256_zeroupper();
#endif # endif
#if defined(__SSE2__) # if defined(__SSE2__)
while (count >= 16) while (count >= 16)
{ {
__m128i b = _mm_loadu_si128(reinterpret_cast<const __m128i*>(buf)); __m128i b = _mm_loadu_si128(reinterpret_cast<const __m128i*>(buf));
@ -83,8 +86,8 @@ void xorbuf(byte *buf, const byte *mask, size_t count)
} }
if (count == 0) return; if (count == 0) return;
#endif # endif
#if defined(__aarch64__) || defined(__aarch32__) || defined(_M_ARM64) # if defined(__aarch64__) || defined(__aarch32__) || defined(_M_ARM64)
while (count >= 16) while (count >= 16)
{ {
vst1q_u8(buf, veorq_u8(vld1q_u8(buf), vld1q_u8(mask))); vst1q_u8(buf, veorq_u8(vld1q_u8(buf), vld1q_u8(mask)));
@ -92,7 +95,8 @@ void xorbuf(byte *buf, const byte *mask, size_t count)
} }
if (count == 0) return; if (count == 0) return;
#endif # endif
#endif // CRYPTOPP_DISABLE_ASM
#if CRYPTOPP_BOOL_PPC32 || CRYPTOPP_BOOL_PPC64 #if CRYPTOPP_BOOL_PPC32 || CRYPTOPP_BOOL_PPC64
// word64 and stride of 8 slows things down on x86_64. // word64 and stride of 8 slows things down on x86_64.
@ -135,7 +139,8 @@ void xorbuf(byte *output, const byte *input, const byte *mask, size_t count)
CRYPTOPP_ASSERT(input != NULLPTR); CRYPTOPP_ASSERT(input != NULLPTR);
CRYPTOPP_ASSERT(count > 0); CRYPTOPP_ASSERT(count > 0);
#if defined(__AVX__) #ifndef CRYPTOPP_DISABLE_ASM
# if defined(__AVX__)
while (count >= 32) while (count >= 32)
{ {
__m256i b = _mm256_loadu_si256(reinterpret_cast<const __m256i*>(input)); __m256i b = _mm256_loadu_si256(reinterpret_cast<const __m256i*>(input));
@ -146,8 +151,8 @@ void xorbuf(byte *output, const byte *input, const byte *mask, size_t count)
} }
// https://software.intel.com/en-us/articles/avoiding-avx-sse-transition-penalties // https://software.intel.com/en-us/articles/avoiding-avx-sse-transition-penalties
_mm256_zeroupper(); _mm256_zeroupper();
#endif # endif
#if defined(__SSE2__) # if defined(__SSE2__)
while (count >= 16) while (count >= 16)
{ {
__m128i b = _mm_loadu_si128(reinterpret_cast<const __m128i*>(input)); __m128i b = _mm_loadu_si128(reinterpret_cast<const __m128i*>(input));
@ -158,8 +163,8 @@ void xorbuf(byte *output, const byte *input, const byte *mask, size_t count)
} }
if (count == 0) return; if (count == 0) return;
#endif # endif
#if defined(__aarch64__) || defined(__aarch32__) || defined(_M_ARM64) # if defined(__aarch64__) || defined(__aarch32__) || defined(_M_ARM64)
while (count >= 16) while (count >= 16)
{ {
vst1q_u8(output, veorq_u8(vld1q_u8(input), vld1q_u8(mask))); vst1q_u8(output, veorq_u8(vld1q_u8(input), vld1q_u8(mask)));
@ -167,7 +172,8 @@ void xorbuf(byte *output, const byte *input, const byte *mask, size_t count)
} }
if (count == 0) return; if (count == 0) return;
#endif # endif
#endif // CRYPTOPP_DISABLE_ASM
#if CRYPTOPP_BOOL_PPC32 || CRYPTOPP_BOOL_PPC64 #if CRYPTOPP_BOOL_PPC32 || CRYPTOPP_BOOL_PPC64
// word64 and stride of 8 slows things down on x86_64. // word64 and stride of 8 slows things down on x86_64.