shadps4-emu/ext-cryptopp
1
0
Fork 0
mirror of https://github.com/shadps4-emu/ext-cryptopp.git synced 2024-11-27 11:50:29 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Fix PPC64 compile with Clang 3.8 on Little Endian machines

Browse Source 
Lets see what this breaks. Clang is a bane in the PowerPC world. Its stupid preprocessor define tricks create more work than it is worth.
This commit is contained in:
Jeffrey Walton 2019-10-21 20:09:09 -04:00
parent d75bce40ec
commit 3a8d45b5e3
No known key found for this signature in database
GPG Key ID: B36AB348921B1838
1 changed files with 30 additions and 22 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

52
ppc_simd.h
View File

@ -95,10 +95,18 @@
// https://www-01.ibm.com/support/docview.wss?uid=swg21683541. // https://www-01.ibm.com/support/docview.wss?uid=swg21683541.
#if defined(__xlc__) && (__xlc__ < 0x0d01) #if defined(__xlc__) && (__xlc__ < 0x0d01)
# define __early_xlc__ 1 # define early_xlc 1
#endif #endif
#if defined(__xlC__) && (__xlC__ < 0x0d01) #if defined(__xlC__) && (__xlC__ < 0x0d01)
# define __early_xlC__ 1 # define early_xlC 1
#endif
// Clang did not add vec_xl for little endian until November 2016
// via https://reviews.llvm.org/D26519 . The release coincides
// with roughly Clang 4.0 released March 2017. There is a
// Clang 3.9.1 release, but it is not clear of vec_xl made it.
#if defined(__clang__) && (__clang_major__ >= 4)
# define later_clang 1
#endif #endif
// VecLoad_ALTIVEC and VecStore_ALTIVEC are // VecLoad_ALTIVEC and VecStore_ALTIVEC are
@ -253,9 +261,9 @@ inline uint32x4_p VecLoad_ALTIVEC(int off, const byte src[16])
inline uint32x4_p VecLoad(const byte src[16]) inline uint32x4_p VecLoad(const byte src[16])
{ {
#if defined(_ARCH_PWR8) #if defined(_ARCH_PWR8)
# if defined(__early_xlc__) || defined(__early_xlC__) # if defined(early_xlc) || defined(early_xlC)
return (uint32x4_p)vec_xlw4(0, (byte*)src); return (uint32x4_p)vec_xlw4(0, (byte*)src);
# elif defined(__xlc__) || defined(__xlC__) || defined(__clang__) # elif defined(__xlc__) || defined(__xlC__) || defined(later_clang)
return (uint32x4_p)vec_xl(0, (byte*)src); return (uint32x4_p)vec_xl(0, (byte*)src);
# else # else
return (uint32x4_p)vec_vsx_ld(0, (byte*)src); return (uint32x4_p)vec_vsx_ld(0, (byte*)src);
@ -281,9 +289,9 @@ inline uint32x4_p VecLoad(const byte src[16])
inline uint32x4_p VecLoad(int off, const byte src[16]) inline uint32x4_p VecLoad(int off, const byte src[16])
{ {
#if defined(_ARCH_PWR8) #if defined(_ARCH_PWR8)
# if defined(__early_xlc__) || defined(__early_xlC__) # if defined(early_xlc) || defined(early_xlC)
return (uint32x4_p)vec_xlw4(off, (byte*)src); return (uint32x4_p)vec_xlw4(off, (byte*)src);
# elif defined(__xlc__) || defined(__xlC__) || defined(__clang__) # elif defined(__xlc__) || defined(__xlC__) || defined(later_clang)
return (uint32x4_p)vec_xl(off, (byte*)src); return (uint32x4_p)vec_xl(off, (byte*)src);
# else # else
return (uint32x4_p)vec_vsx_ld(off, (byte*)src); return (uint32x4_p)vec_vsx_ld(off, (byte*)src);
@ -383,9 +391,9 @@ inline uint64x2_p VecLoad(int off, const word64 src[2])
inline uint32x4_p VecLoadAligned(const byte src[16]) inline uint32x4_p VecLoadAligned(const byte src[16])
{ {
#if defined(_ARCH_PWR8) #if defined(_ARCH_PWR8)
# if defined(__early_xlc__) || defined(__early_xlC__) # if defined(early_xlc) || defined(early_xlC)
return (uint32x4_p)vec_xlw4(0, (byte*)src); return (uint32x4_p)vec_xlw4(0, (byte*)src);
# elif defined(__xlc__) || defined(__xlC__) || defined(__clang__) # elif defined(__xlc__) || defined(__xlC__) || defined(later_clang)
return (uint32x4_p)vec_xl(0, (byte*)src); return (uint32x4_p)vec_xl(0, (byte*)src);
# else # else
return (uint32x4_p)vec_vsx_ld(0, (byte*)src); return (uint32x4_p)vec_vsx_ld(0, (byte*)src);
@ -411,9 +419,9 @@ inline uint32x4_p VecLoadAligned(const byte src[16])
inline uint32x4_p VecLoadAligned(int off, const byte src[16]) inline uint32x4_p VecLoadAligned(int off, const byte src[16])
{ {
#if defined(_ARCH_PWR8) #if defined(_ARCH_PWR8)
# if defined(__early_xlc__) || defined(__early_xlC__) # if defined(early_xlc) || defined(early_xlC)
return (uint32x4_p)vec_xlw4(off, (byte*)src); return (uint32x4_p)vec_xlw4(off, (byte*)src);
# elif defined(__xlc__) || defined(__xlC__) || defined(__clang__) # elif defined(__xlc__) || defined(__xlC__) || defined(later_clang)
return (uint32x4_p)vec_xl(off, (byte*)src); return (uint32x4_p)vec_xl(off, (byte*)src);
# else # else
return (uint32x4_p)vec_vsx_ld(off, (byte*)src); return (uint32x4_p)vec_vsx_ld(off, (byte*)src);
@ -440,13 +448,13 @@ inline uint32x4_p VecLoadAligned(int off, const byte src[16])
inline uint32x4_p VecLoadBE(const byte src[16]) inline uint32x4_p VecLoadBE(const byte src[16])
{ {
#if defined(_ARCH_PWR8) #if defined(_ARCH_PWR8)
# if defined(__early_xlc__) || defined(__early_xlC__) # if defined(early_xlc) || defined(early_xlC)
# if (CRYPTOPP_BIG_ENDIAN) # if (CRYPTOPP_BIG_ENDIAN)
return (uint32x4_p)vec_xlw4(0, (byte*)src); return (uint32x4_p)vec_xlw4(0, (byte*)src);
# else # else
return (uint32x4_p)VecReverse(vec_xlw4(0, (byte*)src)); return (uint32x4_p)VecReverse(vec_xlw4(0, (byte*)src));
# endif # endif
# elif defined(__xlc__) || defined(__xlC__) || defined(__clang__) # elif defined(__xlc__) || defined(__xlC__) || defined(later_clang)
return (uint32x4_p)vec_xl_be(0, (byte*)src); return (uint32x4_p)vec_xl_be(0, (byte*)src);
# else # else
# if (CRYPTOPP_BIG_ENDIAN) # if (CRYPTOPP_BIG_ENDIAN)
@ -481,13 +489,13 @@ inline uint32x4_p VecLoadBE(const byte src[16])
inline uint32x4_p VecLoadBE(int off, const byte src[16]) inline uint32x4_p VecLoadBE(int off, const byte src[16])
{ {
#if defined(_ARCH_PWR8) #if defined(_ARCH_PWR8)
# if defined(__early_xlc__) || defined(__early_xlC__) # if defined(early_xlc) || defined(early_xlC)
# if (CRYPTOPP_BIG_ENDIAN) # if (CRYPTOPP_BIG_ENDIAN)
return (uint32x4_p)vec_xlw4(off, (byte*)src); return (uint32x4_p)vec_xlw4(off, (byte*)src);
# else # else
return (uint32x4_p)VecReverse(vec_xlw4(off, (byte*)src)); return (uint32x4_p)VecReverse(vec_xlw4(off, (byte*)src));
# endif # endif
# elif defined(__xlc__) || defined(__xlC__) || defined(__clang__) # elif defined(__xlc__) || defined(__xlC__) || defined(later_clang)
return (uint32x4_p)vec_xl_be(off, (byte*)src); return (uint32x4_p)vec_xl_be(off, (byte*)src);
# else # else
# if (CRYPTOPP_BIG_ENDIAN) # if (CRYPTOPP_BIG_ENDIAN)
@ -605,9 +613,9 @@ template<class T>
inline void VecStore(const T data, byte dest[16]) inline void VecStore(const T data, byte dest[16])
{ {
#if defined(_ARCH_PWR8) #if defined(_ARCH_PWR8)
# if defined(__early_xlc__) || defined(__early_xlC__) # if defined(early_xlc) || defined(early_xlC)
vec_xstw4((uint8x16_p)data, 0, (byte*)dest); vec_xstw4((uint8x16_p)data, 0, (byte*)dest);
# elif defined(__xlc__) || defined(__xlC__) || defined(__clang__) # elif defined(__xlc__) || defined(__xlC__) || defined(later_clang)
vec_xst((uint8x16_p)data, 0, (byte*)dest); vec_xst((uint8x16_p)data, 0, (byte*)dest);
# else # else
vec_vsx_st((uint8x16_p)data, 0, (byte*)dest); vec_vsx_st((uint8x16_p)data, 0, (byte*)dest);
@ -636,9 +644,9 @@ template<class T>
inline void VecStore(const T data, int off, byte dest[16]) inline void VecStore(const T data, int off, byte dest[16])
{ {
#if defined(_ARCH_PWR8) #if defined(_ARCH_PWR8)
# if defined(__early_xlc__) || defined(__early_xlC__) # if defined(early_xlc) || defined(early_xlC)
vec_xstw4((uint8x16_p)data, off, (byte*)dest); vec_xstw4((uint8x16_p)data, off, (byte*)dest);
# elif defined(__xlc__) || defined(__xlC__) || defined(__clang__) # elif defined(__xlc__) || defined(__xlC__) || defined(later_clang)
vec_xst((uint8x16_p)data, off, (byte*)dest); vec_xst((uint8x16_p)data, off, (byte*)dest);
# else # else
vec_vsx_st((uint8x16_p)data, off, (byte*)dest); vec_vsx_st((uint8x16_p)data, off, (byte*)dest);
@ -751,13 +759,13 @@ template <class T>
inline void VecStoreBE(const T data, byte dest[16]) inline void VecStoreBE(const T data, byte dest[16])
{ {
#if defined(_ARCH_PWR8) #if defined(_ARCH_PWR8)
# if defined(__early_xlc__) || defined(__early_xlC__) # if defined(early_xlc) || defined(early_xlC)
# if (CRYPTOPP_BIG_ENDIAN) # if (CRYPTOPP_BIG_ENDIAN)
vec_xstw4((uint8x16_p)data, 0, (byte*)dest); vec_xstw4((uint8x16_p)data, 0, (byte*)dest);
# else # else
vec_xstw4((uint8x16_p)VecReverse(data), 0, (byte*)dest); vec_xstw4((uint8x16_p)VecReverse(data), 0, (byte*)dest);
# endif # endif
# elif defined(__xlc__) || defined(__xlC__) || defined(__clang__) # elif defined(__xlc__) || defined(__xlC__) || defined(later_clang)
vec_xst_be((uint8x16_p)data, 0, (byte*)dest); vec_xst_be((uint8x16_p)data, 0, (byte*)dest);
# else # else
# if (CRYPTOPP_BIG_ENDIAN) # if (CRYPTOPP_BIG_ENDIAN)
@ -795,13 +803,13 @@ template <class T>
inline void VecStoreBE(const T data, int off, byte dest[16]) inline void VecStoreBE(const T data, int off, byte dest[16])
{ {
#if defined(_ARCH_PWR8) #if defined(_ARCH_PWR8)
# if defined(__early_xlc__) || defined(__early_xlC__) # if defined(early_xlc) || defined(early_xlC)
# if (CRYPTOPP_BIG_ENDIAN) # if (CRYPTOPP_BIG_ENDIAN)
vec_xstw4((uint8x16_p)data, off, (byte*)dest); vec_xstw4((uint8x16_p)data, off, (byte*)dest);
# else # else
vec_xstw4((uint8x16_p)VecReverse(data), off, (byte*)dest); vec_xstw4((uint8x16_p)VecReverse(data), off, (byte*)dest);
# endif # endif
# elif defined(__xlc__) || defined(__xlC__) || defined(__clang__) # elif defined(__xlc__) || defined(__xlC__) || defined(later_clang)
vec_xst_be((uint8x16_p)data, off, (byte*)dest); vec_xst_be((uint8x16_p)data, off, (byte*)dest);
# else # else
# if (CRYPTOPP_BIG_ENDIAN) # if (CRYPTOPP_BIG_ENDIAN)