coffee/libyuv
1
0
Fork 0
mirror of https://chromium.googlesource.com/libyuv/libyuv synced 2025-12-06 16:56:55 +08:00
Code Issues Packages Projects Releases Wiki Activity

Remove code alignment declspec from Visual C versions for vs2014 compatibility.

Browse Source 
BUG=422
TESTED=local vs2013 build still passes.

Review URL: https://webrtc-codereview.appspot.com/45959004

git-svn-id: http://libyuv.googlecode.com/svn/trunk@1365 16f28f9a-4ce2-e073-06de-1de4eb20be90
This commit is contained in:
fbarchard@google.com 2015-04-12 23:54:26 +00:00
parent 1eb51bcf01
commit c7161d1c36
7 changed files with 162 additions and 166 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
README.chromium
View File

@ -1,6 +1,6 @@
Name: libyuv Name: libyuv
URL: http://code.google.com/p/libyuv/ URL: http://code.google.com/p/libyuv/
Version: 1364 Version: 1365
License: BSD License: BSD
License File: LICENSE License File: LICENSE

2
include/libyuv/version.h
View File

@ -11,6 +11,6 @@
#ifndef INCLUDE_LIBYUV_VERSION_H_ // NOLINT #ifndef INCLUDE_LIBYUV_VERSION_H_ // NOLINT
#define INCLUDE_LIBYUV_VERSION_H_ #define INCLUDE_LIBYUV_VERSION_H_
#define LIBYUV_VERSION 1364 #define LIBYUV_VERSION 1365
#endif // INCLUDE_LIBYUV_VERSION_H_ NOLINT #endif // INCLUDE_LIBYUV_VERSION_H_ NOLINT

8
source/compare_win.cc
View File

@ -19,7 +19,7 @@ extern "C" {
#if !defined(LIBYUV_DISABLE_X86) && defined(_M_IX86) && \ #if !defined(LIBYUV_DISABLE_X86) && defined(_M_IX86) && \
defined(_MSC_VER) && !defined(__clang__) defined(_MSC_VER) && !defined(__clang__)
__declspec(naked) __declspec(align(16)) __declspec(naked)
uint32 SumSquareError_SSE2(const uint8* src_a, const uint8* src_b, int count) { uint32 SumSquareError_SSE2(const uint8* src_a, const uint8* src_b, int count) {
__asm { __asm {
mov eax, [esp + 4] // src_a mov eax, [esp + 4] // src_a
@ -60,7 +60,7 @@ uint32 SumSquareError_SSE2(const uint8* src_a, const uint8* src_b, int count) {
#if _MSC_VER >= 1700 #if _MSC_VER >= 1700
// C4752: found Intel(R) Advanced Vector Extensions; consider using /arch:AVX. // C4752: found Intel(R) Advanced Vector Extensions; consider using /arch:AVX.
#pragma warning(disable: 4752) #pragma warning(disable: 4752)
__declspec(naked) __declspec(align(16)) __declspec(naked)
uint32 SumSquareError_AVX2(const uint8* src_a, const uint8* src_b, int count) { uint32 SumSquareError_AVX2(const uint8* src_a, const uint8* src_b, int count) {
__asm { __asm {
mov eax, [esp + 4] // src_a mov eax, [esp + 4] // src_a
@ -134,7 +134,7 @@ static uvec32 kHashMul3 = {
#define pmulld(reg) _asm _emit 0x66 _asm _emit 0x0F _asm _emit 0x38 \ #define pmulld(reg) _asm _emit 0x66 _asm _emit 0x0F _asm _emit 0x38 \
_asm _emit 0x40 _asm _emit reg _asm _emit 0x40 _asm _emit reg
__declspec(naked) __declspec(align(16)) __declspec(naked)
uint32 HashDjb2_SSE41(const uint8* src, int count, uint32 seed) { uint32 HashDjb2_SSE41(const uint8* src, int count, uint32 seed) {
__asm { __asm {
mov eax, [esp + 4] // src mov eax, [esp + 4] // src
@ -185,7 +185,7 @@ uint32 HashDjb2_SSE41(const uint8* src, int count, uint32 seed) {
// Visual C 2012 required for AVX2. // Visual C 2012 required for AVX2.
#if _MSC_VER >= 1700 #if _MSC_VER >= 1700
__declspec(naked) __declspec(align(16)) __declspec(naked)
uint32 HashDjb2_AVX2(const uint8* src, int count, uint32 seed) { uint32 HashDjb2_AVX2(const uint8* src, int count, uint32 seed) {
__asm { __asm {
mov eax, [esp + 4] // src mov eax, [esp + 4] // src

2
source/mjpeg_validate.cc
View File

@ -23,7 +23,7 @@ extern "C" {
#ifdef ENABLE_SCASB #ifdef ENABLE_SCASB
// Multiple of 1. // Multiple of 1.
__declspec(naked) __declspec(align(16)) __declspec(naked)
const uint8* ScanRow_ERMS(const uint8* src, uint32 val, int count) { const uint8* ScanRow_ERMS(const uint8* src, uint32 val, int count) {
__asm { __asm {
mov edx, edi mov edx, edi

4
source/rotate.cc
View File

@ -73,7 +73,7 @@ void TransposeUVWx8_MIPS_DSPR2(const uint8* src, int src_stride,
#if !defined(LIBYUV_DISABLE_X86) && \ #if !defined(LIBYUV_DISABLE_X86) && \
defined(_M_IX86) && defined(_MSC_VER) && !defined(__clang__) defined(_M_IX86) && defined(_MSC_VER) && !defined(__clang__)
#define HAS_TRANSPOSE_WX8_SSSE3 #define HAS_TRANSPOSE_WX8_SSSE3
__declspec(naked) __declspec(align(16)) __declspec(naked)
static void TransposeWx8_SSSE3(const uint8* src, int src_stride, static void TransposeWx8_SSSE3(const uint8* src, int src_stride,
uint8* dst, int dst_stride, int width) { uint8* dst, int dst_stride, int width) {
__asm { __asm {
@ -165,7 +165,7 @@ static void TransposeWx8_SSSE3(const uint8* src, int src_stride,
} }
#define HAS_TRANSPOSE_UVWX8_SSE2 #define HAS_TRANSPOSE_UVWX8_SSE2
__declspec(naked) __declspec(align(16)) __declspec(naked)
static void TransposeUVWx8_SSE2(const uint8* src, int src_stride, static void TransposeUVWx8_SSE2(const uint8* src, int src_stride,
uint8* dst_a, int dst_stride_a, uint8* dst_a, int dst_stride_a,
uint8* dst_b, int dst_stride_b, uint8* dst_b, int dst_stride_b,

256
source/row_win.cc
View File

File diff suppressed because it is too large Load Diff

54
source/scale_win.cc
View File

@ -95,7 +95,7 @@ static uvec16 kScaleAb2 =
{ 65536 / 3, 65536 / 3, 65536 / 2, 65536 / 3, 65536 / 3, 65536 / 2, 0, 0 }; { 65536 / 3, 65536 / 3, 65536 / 2, 65536 / 3, 65536 / 3, 65536 / 2, 0, 0 };
// Reads 32 pixels, throws half away and writes 16 pixels. // Reads 32 pixels, throws half away and writes 16 pixels.
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleRowDown2_SSE2(const uint8* src_ptr, ptrdiff_t src_stride, void ScaleRowDown2_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
uint8* dst_ptr, int dst_width) { uint8* dst_ptr, int dst_width) {
__asm { __asm {
@ -121,7 +121,7 @@ void ScaleRowDown2_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
} }
// Blends 32x1 rectangle to 16x1. // Blends 32x1 rectangle to 16x1.
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleRowDown2Linear_SSE2(const uint8* src_ptr, ptrdiff_t src_stride, void ScaleRowDown2Linear_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
uint8* dst_ptr, int dst_width) { uint8* dst_ptr, int dst_width) {
__asm { __asm {
@ -157,7 +157,7 @@ void ScaleRowDown2Linear_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
} }
// Blends 32x2 rectangle to 16x1. // Blends 32x2 rectangle to 16x1.
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleRowDown2Box_SSE2(const uint8* src_ptr, ptrdiff_t src_stride, void ScaleRowDown2Box_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
uint8* dst_ptr, int dst_width) { uint8* dst_ptr, int dst_width) {
__asm { __asm {
@ -200,7 +200,7 @@ void ScaleRowDown2Box_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
#ifdef HAS_SCALEROWDOWN2_AVX2 #ifdef HAS_SCALEROWDOWN2_AVX2
// Reads 64 pixels, throws half away and writes 32 pixels. // Reads 64 pixels, throws half away and writes 32 pixels.
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleRowDown2_AVX2(const uint8* src_ptr, ptrdiff_t src_stride, void ScaleRowDown2_AVX2(const uint8* src_ptr, ptrdiff_t src_stride,
uint8* dst_ptr, int dst_width) { uint8* dst_ptr, int dst_width) {
__asm { __asm {
@ -228,7 +228,7 @@ void ScaleRowDown2_AVX2(const uint8* src_ptr, ptrdiff_t src_stride,
} }
// Blends 64x1 rectangle to 32x1. // Blends 64x1 rectangle to 32x1.
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleRowDown2Linear_AVX2(const uint8* src_ptr, ptrdiff_t src_stride, void ScaleRowDown2Linear_AVX2(const uint8* src_ptr, ptrdiff_t src_stride,
uint8* dst_ptr, int dst_width) { uint8* dst_ptr, int dst_width) {
__asm { __asm {
@ -265,7 +265,7 @@ void ScaleRowDown2Linear_AVX2(const uint8* src_ptr, ptrdiff_t src_stride,
} }
// Blends 64x2 rectangle to 32x1. // Blends 64x2 rectangle to 32x1.
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleRowDown2Box_AVX2(const uint8* src_ptr, ptrdiff_t src_stride, void ScaleRowDown2Box_AVX2(const uint8* src_ptr, ptrdiff_t src_stride,
uint8* dst_ptr, int dst_width) { uint8* dst_ptr, int dst_width) {
__asm { __asm {
@ -307,7 +307,7 @@ void ScaleRowDown2Box_AVX2(const uint8* src_ptr, ptrdiff_t src_stride,
#endif // HAS_SCALEROWDOWN2_AVX2 #endif // HAS_SCALEROWDOWN2_AVX2
// Point samples 32 pixels to 8 pixels. // Point samples 32 pixels to 8 pixels.
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleRowDown4_SSE2(const uint8* src_ptr, ptrdiff_t src_stride, void ScaleRowDown4_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
uint8* dst_ptr, int dst_width) { uint8* dst_ptr, int dst_width) {
__asm { __asm {
@ -338,7 +338,7 @@ void ScaleRowDown4_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
} }
// Blends 32x4 rectangle to 8x1. // Blends 32x4 rectangle to 8x1.
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleRowDown4Box_SSE2(const uint8* src_ptr, ptrdiff_t src_stride, void ScaleRowDown4Box_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
uint8* dst_ptr, int dst_width) { uint8* dst_ptr, int dst_width) {
__asm { __asm {
@ -401,7 +401,7 @@ void ScaleRowDown4Box_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
// Then shuffled to do the scaling. // Then shuffled to do the scaling.
// Note that movdqa+palign may be better than movdqu. // Note that movdqa+palign may be better than movdqu.
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleRowDown34_SSSE3(const uint8* src_ptr, ptrdiff_t src_stride, void ScaleRowDown34_SSSE3(const uint8* src_ptr, ptrdiff_t src_stride,
uint8* dst_ptr, int dst_width) { uint8* dst_ptr, int dst_width) {
__asm { __asm {
@ -448,7 +448,7 @@ void ScaleRowDown34_SSSE3(const uint8* src_ptr, ptrdiff_t src_stride,
// xmm7 kRound34 // xmm7 kRound34
// Note that movdqa+palign may be better than movdqu. // Note that movdqa+palign may be better than movdqu.
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleRowDown34_1_Box_SSSE3(const uint8* src_ptr, void ScaleRowDown34_1_Box_SSSE3(const uint8* src_ptr,
ptrdiff_t src_stride, ptrdiff_t src_stride,
uint8* dst_ptr, int dst_width) { uint8* dst_ptr, int dst_width) {
@ -505,7 +505,7 @@ void ScaleRowDown34_1_Box_SSSE3(const uint8* src_ptr,
} }
// Note that movdqa+palign may be better than movdqu. // Note that movdqa+palign may be better than movdqu.
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleRowDown34_0_Box_SSSE3(const uint8* src_ptr, void ScaleRowDown34_0_Box_SSSE3(const uint8* src_ptr,
ptrdiff_t src_stride, ptrdiff_t src_stride,
uint8* dst_ptr, int dst_width) { uint8* dst_ptr, int dst_width) {
@ -567,7 +567,7 @@ void ScaleRowDown34_0_Box_SSSE3(const uint8* src_ptr,
// 3/8 point sampler // 3/8 point sampler
// Scale 32 pixels to 12 // Scale 32 pixels to 12
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleRowDown38_SSSE3(const uint8* src_ptr, ptrdiff_t src_stride, void ScaleRowDown38_SSSE3(const uint8* src_ptr, ptrdiff_t src_stride,
uint8* dst_ptr, int dst_width) { uint8* dst_ptr, int dst_width) {
__asm { __asm {
@ -598,7 +598,7 @@ void ScaleRowDown38_SSSE3(const uint8* src_ptr, ptrdiff_t src_stride,
} }
// Scale 16x3 pixels to 6x1 with interpolation // Scale 16x3 pixels to 6x1 with interpolation
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleRowDown38_3_Box_SSSE3(const uint8* src_ptr, void ScaleRowDown38_3_Box_SSSE3(const uint8* src_ptr,
ptrdiff_t src_stride, ptrdiff_t src_stride,
uint8* dst_ptr, int dst_width) { uint8* dst_ptr, int dst_width) {
@ -663,7 +663,7 @@ void ScaleRowDown38_3_Box_SSSE3(const uint8* src_ptr,
} }
// Scale 16x2 pixels to 6x1 with interpolation // Scale 16x2 pixels to 6x1 with interpolation
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleRowDown38_2_Box_SSSE3(const uint8* src_ptr, void ScaleRowDown38_2_Box_SSSE3(const uint8* src_ptr,
ptrdiff_t src_stride, ptrdiff_t src_stride,
uint8* dst_ptr, int dst_width) { uint8* dst_ptr, int dst_width) {
@ -709,7 +709,7 @@ void ScaleRowDown38_2_Box_SSSE3(const uint8* src_ptr,
// Reads 16xN bytes and produces 16 shorts at a time. // Reads 16xN bytes and produces 16 shorts at a time.
// TODO(fbarchard): Make this handle 4xN bytes for any width ARGB. // TODO(fbarchard): Make this handle 4xN bytes for any width ARGB.
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleAddRows_SSE2(const uint8* src_ptr, ptrdiff_t src_stride, void ScaleAddRows_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
uint16* dst_ptr, int src_width, uint16* dst_ptr, int src_width,
int src_height) { int src_height) {
@ -775,7 +775,7 @@ void ScaleAddRows_SSE2(const uint8* src_ptr, ptrdiff_t src_stride,
// when drmemory bug fixed. // when drmemory bug fixed.
// https://code.google.com/p/drmemory/issues/detail?id=1396 // https://code.google.com/p/drmemory/issues/detail?id=1396
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleFilterCols_SSSE3(uint8* dst_ptr, const uint8* src_ptr, void ScaleFilterCols_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
int dst_width, int x, int dx) { int dst_width, int x, int dx) {
__asm { __asm {
@ -852,7 +852,7 @@ void ScaleFilterCols_SSSE3(uint8* dst_ptr, const uint8* src_ptr,
} }
// Reads 16 pixels, duplicates them and writes 32 pixels. // Reads 16 pixels, duplicates them and writes 32 pixels.
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleColsUp2_SSE2(uint8* dst_ptr, const uint8* src_ptr, void ScaleColsUp2_SSE2(uint8* dst_ptr, const uint8* src_ptr,
int dst_width, int x, int dx) { int dst_width, int x, int dx) {
__asm { __asm {
@ -877,7 +877,7 @@ void ScaleColsUp2_SSE2(uint8* dst_ptr, const uint8* src_ptr,
} }
// Reads 8 pixels, throws half away and writes 4 even pixels (0, 2, 4, 6) // Reads 8 pixels, throws half away and writes 4 even pixels (0, 2, 4, 6)
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleARGBRowDown2_SSE2(const uint8* src_argb, void ScaleARGBRowDown2_SSE2(const uint8* src_argb,
ptrdiff_t src_stride, ptrdiff_t src_stride,
uint8* dst_argb, int dst_width) { uint8* dst_argb, int dst_width) {
@ -902,7 +902,7 @@ void ScaleARGBRowDown2_SSE2(const uint8* src_argb,
} }
// Blends 8x1 rectangle to 4x1. // Blends 8x1 rectangle to 4x1.
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleARGBRowDown2Linear_SSE2(const uint8* src_argb, void ScaleARGBRowDown2Linear_SSE2(const uint8* src_argb,
ptrdiff_t src_stride, ptrdiff_t src_stride,
uint8* dst_argb, int dst_width) { uint8* dst_argb, int dst_width) {
@ -930,7 +930,7 @@ void ScaleARGBRowDown2Linear_SSE2(const uint8* src_argb,
} }
// Blends 8x2 rectangle to 4x1. // Blends 8x2 rectangle to 4x1.
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleARGBRowDown2Box_SSE2(const uint8* src_argb, void ScaleARGBRowDown2Box_SSE2(const uint8* src_argb,
ptrdiff_t src_stride, ptrdiff_t src_stride,
uint8* dst_argb, int dst_width) { uint8* dst_argb, int dst_width) {
@ -964,7 +964,7 @@ void ScaleARGBRowDown2Box_SSE2(const uint8* src_argb,
} }
// Reads 4 pixels at a time. // Reads 4 pixels at a time.
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleARGBRowDownEven_SSE2(const uint8* src_argb, ptrdiff_t src_stride, void ScaleARGBRowDownEven_SSE2(const uint8* src_argb, ptrdiff_t src_stride,
int src_stepx, int src_stepx,
uint8* dst_argb, int dst_width) { uint8* dst_argb, int dst_width) {
@ -1000,7 +1000,7 @@ void ScaleARGBRowDownEven_SSE2(const uint8* src_argb, ptrdiff_t src_stride,
} }
// Blends four 2x2 to 4x1. // Blends four 2x2 to 4x1.
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleARGBRowDownEvenBox_SSE2(const uint8* src_argb, void ScaleARGBRowDownEvenBox_SSE2(const uint8* src_argb,
ptrdiff_t src_stride, ptrdiff_t src_stride,
int src_stepx, int src_stepx,
@ -1048,7 +1048,7 @@ void ScaleARGBRowDownEvenBox_SSE2(const uint8* src_argb,
} }
// Column scaling unfiltered. SSE2 version. // Column scaling unfiltered. SSE2 version.
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleARGBCols_SSE2(uint8* dst_argb, const uint8* src_argb, void ScaleARGBCols_SSE2(uint8* dst_argb, const uint8* src_argb,
int dst_width, int x, int dx) { int dst_width, int x, int dx) {
__asm { __asm {
@ -1139,7 +1139,7 @@ static uvec8 kShuffleFractions = {
0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u, 4u, 4u, 4u, 4u, 4u, 4u, 4u, 4u, 0u, 0u, 0u, 0u, 0u, 0u, 0u, 0u, 4u, 4u, 4u, 4u, 4u, 4u, 4u, 4u,
}; };
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleARGBFilterCols_SSSE3(uint8* dst_argb, const uint8* src_argb, void ScaleARGBFilterCols_SSSE3(uint8* dst_argb, const uint8* src_argb,
int dst_width, int x, int dx) { int dst_width, int x, int dx) {
__asm { __asm {
@ -1210,7 +1210,7 @@ void ScaleARGBFilterCols_SSSE3(uint8* dst_argb, const uint8* src_argb,
} }
// Reads 4 pixels, duplicates them and writes 8 pixels. // Reads 4 pixels, duplicates them and writes 8 pixels.
__declspec(naked) __declspec(align(16)) __declspec(naked)
void ScaleARGBColsUp2_SSE2(uint8* dst_argb, const uint8* src_argb, void ScaleARGBColsUp2_SSE2(uint8* dst_argb, const uint8* src_argb,
int dst_width, int x, int dx) { int dst_width, int x, int dx) {
__asm { __asm {
@ -1235,7 +1235,7 @@ void ScaleARGBColsUp2_SSE2(uint8* dst_argb, const uint8* src_argb,
} }
// Divide num by div and return as 16.16 fixed point result. // Divide num by div and return as 16.16 fixed point result.
__declspec(naked) __declspec(align(16)) __declspec(naked)
int FixedDiv_X86(int num, int div) { int FixedDiv_X86(int num, int div) {
__asm { __asm {
mov eax, [esp + 4] // num mov eax, [esp + 4] // num
@ -1248,7 +1248,7 @@ int FixedDiv_X86(int num, int div) {
} }
// Divide num by div and return as 16.16 fixed point result. // Divide num by div and return as 16.16 fixed point result.
__declspec(naked) __declspec(align(16)) __declspec(naked)
int FixedDiv1_X86(int num, int div) { int FixedDiv1_X86(int num, int div) {
__asm { __asm {
mov eax, [esp + 4] // num mov eax, [esp + 4] // num