From a65bdceb060628881578afb29df4eb222421381f Mon Sep 17 00:00:00 2001 From: Diego Biurrun Date: Sun, 8 Jul 2012 19:55:31 +0200 Subject: [PATCH] x86: mmx2 ---> mmxext in variable names --- libswscale/swscale_internal.h | 10 ++-- libswscale/utils.c | 80 ++++++++++++++++++------------- libswscale/x86/swscale_template.c | 11 ++--- 3 files changed, 56 insertions(+), 45 deletions(-) diff --git a/libswscale/swscale_internal.h b/libswscale/swscale_internal.h index c239c2a3ee..8752672536 100644 --- a/libswscale/swscale_internal.h +++ b/libswscale/swscale_internal.h @@ -307,12 +307,12 @@ typedef struct SwsContext { int vChrFilterSize; ///< Vertical filter size for chroma pixels. //@} - int lumMmx2FilterCodeSize; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code size for luma/alpha planes. - int chrMmx2FilterCodeSize; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code size for chroma planes. - uint8_t *lumMmx2FilterCode; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code for luma/alpha planes. - uint8_t *chrMmx2FilterCode; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code for chroma planes. + int lumMmxextFilterCodeSize; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code size for luma/alpha planes. + int chrMmxextFilterCodeSize; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code size for chroma planes. + uint8_t *lumMmxextFilterCode; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code for luma/alpha planes. + uint8_t *chrMmxextFilterCode; ///< Runtime-generated MMXEXT horizontal fast bilinear scaler code for chroma planes. - int canMMX2BeUsed; + int canMMXEXTBeUsed; int dstY; ///< Last destination vertical line output from last slice. int flags; ///< Flags passed by the user to select scaler algorithm, optimizations, subsampling, etc... diff --git a/libswscale/utils.c b/libswscale/utils.c index 9e5ebac72e..64a3a58067 100644 --- a/libswscale/utils.c +++ b/libswscale/utils.c @@ -1001,18 +1001,18 @@ av_cold int sws_init_context(SwsContext *c, SwsFilter *srcFilter, (FFALIGN(srcW, 16) * 2 * FFALIGN(c->srcBpc, 8) >> 3) + 16, fail); if (INLINE_MMXEXT(cpu_flags) && c->srcBpc == 8 && c->dstBpc <= 10) { - c->canMMX2BeUsed = (dstW >= srcW && (dstW & 31) == 0 && - (srcW & 15) == 0) ? 1 : 0; - if (!c->canMMX2BeUsed && dstW >= srcW && (srcW & 15) == 0 + c->canMMXEXTBeUsed = (dstW >= srcW && (dstW & 31) == 0 && + (srcW & 15) == 0) ? 1 : 0; + if (!c->canMMXEXTBeUsed && dstW >= srcW && (srcW & 15) == 0 && (flags & SWS_FAST_BILINEAR)) { if (flags & SWS_PRINT_INFO) av_log(c, AV_LOG_INFO, "output width is not a multiple of 32 -> no MMXEXT scaler\n"); } if (usesHFilter) - c->canMMX2BeUsed = 0; + c->canMMXEXTBeUsed = 0; } else - c->canMMX2BeUsed = 0; + c->canMMXEXTBeUsed = 0; c->chrXInc = (((int64_t)c->chrSrcW << 16) + (c->chrDstW >> 1)) / c->chrDstW; c->chrYInc = (((int64_t)c->chrSrcH << 16) + (c->chrDstH >> 1)) / c->chrDstH; @@ -1025,7 +1025,7 @@ av_cold int sws_init_context(SwsContext *c, SwsFilter *srcFilter, * correct variant would be like the vertical one, but that would require * some special code for the first and last pixel */ if (flags & SWS_FAST_BILINEAR) { - if (c->canMMX2BeUsed) { + if (c->canMMXEXTBeUsed) { c->lumXInc += 20; c->chrXInc += 20; } @@ -1042,38 +1042,50 @@ av_cold int sws_init_context(SwsContext *c, SwsFilter *srcFilter, { #if HAVE_MMXEXT_INLINE // can't downscale !!! - if (c->canMMX2BeUsed && (flags & SWS_FAST_BILINEAR)) { - c->lumMmx2FilterCodeSize = initMMX2HScaler(dstW, c->lumXInc, NULL, - NULL, NULL, 8); - c->chrMmx2FilterCodeSize = initMMX2HScaler(c->chrDstW, c->chrXInc, - NULL, NULL, NULL, 4); + if (c->canMMXEXTBeUsed && (flags & SWS_FAST_BILINEAR)) { + c->lumMmxextFilterCodeSize = initMMX2HScaler(dstW, c->lumXInc, NULL, + NULL, NULL, 8); + c->chrMmxextFilterCodeSize = initMMX2HScaler(c->chrDstW, c->chrXInc, + NULL, NULL, NULL, 4); #if USE_MMAP - c->lumMmx2FilterCode = mmap(NULL, c->lumMmx2FilterCodeSize, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0); - c->chrMmx2FilterCode = mmap(NULL, c->chrMmx2FilterCodeSize, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0); + c->lumMmxextFilterCode = mmap(NULL, c->lumMmxextFilterCodeSize, + PROT_READ | PROT_WRITE, + MAP_PRIVATE | MAP_ANONYMOUS, + -1, 0); + c->chrMmxextFilterCode = mmap(NULL, c->chrMmxextFilterCodeSize, + PROT_READ | PROT_WRITE, + MAP_PRIVATE | MAP_ANONYMOUS, + -1, 0); #elif HAVE_VIRTUALALLOC - c->lumMmx2FilterCode = VirtualAlloc(NULL, c->lumMmx2FilterCodeSize, MEM_COMMIT, PAGE_EXECUTE_READWRITE); - c->chrMmx2FilterCode = VirtualAlloc(NULL, c->chrMmx2FilterCodeSize, MEM_COMMIT, PAGE_EXECUTE_READWRITE); + c->lumMmxextFilterCode = VirtualAlloc(NULL, + c->lumMmxextFilterCodeSize, + MEM_COMMIT, + PAGE_EXECUTE_READWRITE); + c->chrMmxextFilterCode = VirtualAlloc(NULL, + c->chrMmxextFilterCodeSize, + MEM_COMMIT, + PAGE_EXECUTE_READWRITE); #else - c->lumMmx2FilterCode = av_malloc(c->lumMmx2FilterCodeSize); - c->chrMmx2FilterCode = av_malloc(c->chrMmx2FilterCodeSize); + c->lumMmxextFilterCode = av_malloc(c->lumMmxextFilterCodeSize); + c->chrMmxextFilterCode = av_malloc(c->chrMmxextFilterCodeSize); #endif - if (!c->lumMmx2FilterCode || !c->chrMmx2FilterCode) + if (!c->lumMmxextFilterCode || !c->chrMmxextFilterCode) return AVERROR(ENOMEM); FF_ALLOCZ_OR_GOTO(c, c->hLumFilter, (dstW / 8 + 8) * sizeof(int16_t), fail); FF_ALLOCZ_OR_GOTO(c, c->hChrFilter, (c->chrDstW / 4 + 8) * sizeof(int16_t), fail); FF_ALLOCZ_OR_GOTO(c, c->hLumFilterPos, (dstW / 2 / 8 + 8) * sizeof(int32_t), fail); FF_ALLOCZ_OR_GOTO(c, c->hChrFilterPos, (c->chrDstW / 2 / 4 + 8) * sizeof(int32_t), fail); - initMMX2HScaler(dstW, c->lumXInc, c->lumMmx2FilterCode, + initMMX2HScaler(dstW, c->lumXInc, c->lumMmxextFilterCode, c->hLumFilter, c->hLumFilterPos, 8); - initMMX2HScaler(c->chrDstW, c->chrXInc, c->chrMmx2FilterCode, + initMMX2HScaler(c->chrDstW, c->chrXInc, c->chrMmxextFilterCode, c->hChrFilter, c->hChrFilterPos, 4); #if USE_MMAP - mprotect(c->lumMmx2FilterCode, c->lumMmx2FilterCodeSize, PROT_EXEC | PROT_READ); - mprotect(c->chrMmx2FilterCode, c->chrMmx2FilterCodeSize, PROT_EXEC | PROT_READ); + mprotect(c->lumMmxextFilterCode, c->lumMmxextFilterCodeSize, PROT_EXEC | PROT_READ); + mprotect(c->chrMmxextFilterCode, c->chrMmxextFilterCodeSize, PROT_EXEC | PROT_READ); #endif } else #endif /* HAVE_MMXEXT_INLINE */ @@ -1651,21 +1663,21 @@ void sws_freeContext(SwsContext *c) #if HAVE_MMX_INLINE #if USE_MMAP - if (c->lumMmx2FilterCode) - munmap(c->lumMmx2FilterCode, c->lumMmx2FilterCodeSize); - if (c->chrMmx2FilterCode) - munmap(c->chrMmx2FilterCode, c->chrMmx2FilterCodeSize); + if (c->lumMmxextFilterCode) + munmap(c->lumMmxextFilterCode, c->lumMmxextFilterCodeSize); + if (c->chrMmxextFilterCode) + munmap(c->chrMmxextFilterCode, c->chrMmxextFilterCodeSize); #elif HAVE_VIRTUALALLOC - if (c->lumMmx2FilterCode) - VirtualFree(c->lumMmx2FilterCode, 0, MEM_RELEASE); - if (c->chrMmx2FilterCode) - VirtualFree(c->chrMmx2FilterCode, 0, MEM_RELEASE); + if (c->lumMmxextFilterCode) + VirtualFree(c->lumMmxextFilterCode, 0, MEM_RELEASE); + if (c->chrMmxextFilterCode) + VirtualFree(c->chrMmxextFilterCode, 0, MEM_RELEASE); #else - av_free(c->lumMmx2FilterCode); - av_free(c->chrMmx2FilterCode); + av_free(c->lumMmxextFilterCode); + av_free(c->chrMmxextFilterCode); #endif - c->lumMmx2FilterCode = NULL; - c->chrMmx2FilterCode = NULL; + c->lumMmxextFilterCode = NULL; + c->chrMmxextFilterCode = NULL; #endif /* HAVE_MMX_INLINE */ av_freep(&c->yuvTable); diff --git a/libswscale/x86/swscale_template.c b/libswscale/x86/swscale_template.c index 8be6398f59..d7a2fdbe18 100644 --- a/libswscale/x86/swscale_template.c +++ b/libswscale/x86/swscale_template.c @@ -1378,7 +1378,7 @@ static void RENAME(hyscale_fast)(SwsContext *c, int16_t *dst, { int32_t *filterPos = c->hLumFilterPos; int16_t *filter = c->hLumFilter; - void *mmx2FilterCode= c->lumMmx2FilterCode; + void *mmxextFilterCode = c->lumMmxextFilterCode; int i; #if defined(PIC) uint64_t ebxsave; @@ -1451,7 +1451,7 @@ static void RENAME(hyscale_fast)(SwsContext *c, int16_t *dst, #endif #endif :: "m" (src), "m" (dst), "m" (filter), "m" (filterPos), - "m" (mmx2FilterCode) + "m" (mmxextFilterCode) #if defined(PIC) ,"m" (ebxsave) #endif @@ -1474,7 +1474,7 @@ static void RENAME(hcscale_fast)(SwsContext *c, int16_t *dst1, int16_t *dst2, { int32_t *filterPos = c->hChrFilterPos; int16_t *filter = c->hChrFilter; - void *mmx2FilterCode= c->chrMmx2FilterCode; + void *mmxextFilterCode = c->chrMmxextFilterCode; int i; #if defined(PIC) DECLARE_ALIGNED(8, uint64_t, ebxsave); @@ -1535,7 +1535,7 @@ static void RENAME(hcscale_fast)(SwsContext *c, int16_t *dst1, int16_t *dst2, #endif #endif :: "m" (src1), "m" (dst1), "m" (filter), "m" (filterPos), - "m" (mmx2FilterCode), "m" (src2), "m"(dst2) + "m" (mmxextFilterCode), "m" (src2), "m"(dst2) #if defined(PIC) ,"m" (ebxsave) #endif @@ -1617,8 +1617,7 @@ static av_cold void RENAME(sws_init_swScale)(SwsContext *c) if (c->srcBpc == 8 && c->dstBpc <= 10) { // Use the new MMX scaler if the MMXEXT one can't be used (it is faster than the x86 ASM one). #if COMPILE_TEMPLATE_MMXEXT - if (c->flags & SWS_FAST_BILINEAR && c->canMMX2BeUsed) - { + if (c->flags & SWS_FAST_BILINEAR && c->canMMXEXTBeUsed) { c->hyscale_fast = RENAME(hyscale_fast); c->hcscale_fast = RENAME(hcscale_fast); } else {