virtualenv/FFmpeg
1
0
Fork 0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2025-08-15 14:13:16 +02:00
Code Issues Releases Activity

Turn ancient V offset numerical constants into named ones.

Browse Source 
Add a check that checks that the width is within the choosen constant.
This might have been exploitable.

Originally committed as revision 26165 to svn://svn.mplayerhq.hu/mplayer/trunk/libswscale
This commit is contained in:
Michael Niedermayer
2008-03-05 03:00:25 +00:00
parent fbd4293df9
commit 8b2fce0d3f
3 changed files with 81 additions and 72 deletions
Download Patch File Download Diff File Expand all files Collapse all files

26
libswscale/swscale.c
View File

@@ -397,7 +397,7 @@ static inline void yuv2yuvXinC(int16_t *lumFilter, int16_t **lumSrc, int lumFilt
for (j=0; j<chrFilterSize; j++) for (j=0; j<chrFilterSize; j++)
{ {
u += chrSrc[j][i] * chrFilter[j]; u += chrSrc[j][i] * chrFilter[j];
v += chrSrc[j][i + 2048] * chrFilter[j]; v += chrSrc[j][i + VOFW] * chrFilter[j];
} }
uDest[i]= av_clip_uint8(u>>19); uDest[i]= av_clip_uint8(u>>19);
@@ -433,7 +433,7 @@ static inline void yuv2nv12XinC(int16_t *lumFilter, int16_t **lumSrc, int lumFil
for (j=0; j<chrFilterSize; j++) for (j=0; j<chrFilterSize; j++)
{ {
u += chrSrc[j][i] * chrFilter[j]; u += chrSrc[j][i] * chrFilter[j];
v += chrSrc[j][i + 2048] * chrFilter[j]; v += chrSrc[j][i + VOFW] * chrFilter[j];
} }
uDest[2*i]= av_clip_uint8(u>>19); uDest[2*i]= av_clip_uint8(u>>19);
@@ -448,7 +448,7 @@ static inline void yuv2nv12XinC(int16_t *lumFilter, int16_t **lumSrc, int lumFil
for (j=0; j<chrFilterSize; j++) for (j=0; j<chrFilterSize; j++)
{ {
u += chrSrc[j][i] * chrFilter[j]; u += chrSrc[j][i] * chrFilter[j];
v += chrSrc[j][i + 2048] * chrFilter[j]; v += chrSrc[j][i + VOFW] * chrFilter[j];
} }
uDest[2*i]= av_clip_uint8(v>>19); uDest[2*i]= av_clip_uint8(v>>19);
@@ -474,7 +474,7 @@ static inline void yuv2nv12XinC(int16_t *lumFilter, int16_t **lumSrc, int lumFil
for (j=0; j<chrFilterSize; j++)\ for (j=0; j<chrFilterSize; j++)\
{\ {\
U += chrSrc[j][i] * chrFilter[j];\ U += chrSrc[j][i] * chrFilter[j];\
V += chrSrc[j][i+2048] * chrFilter[j];\ V += chrSrc[j][i+VOFW] * chrFilter[j];\
}\ }\
Y1>>=19;\ Y1>>=19;\
Y2>>=19;\ Y2>>=19;\
@@ -504,7 +504,7 @@ static inline void yuv2nv12XinC(int16_t *lumFilter, int16_t **lumSrc, int lumFil
int Y1= (buf0[i2 ]*yalpha1+buf1[i2 ]*yalpha)>>19; \ int Y1= (buf0[i2 ]*yalpha1+buf1[i2 ]*yalpha)>>19; \
int Y2= (buf0[i2+1]*yalpha1+buf1[i2+1]*yalpha)>>19; \ int Y2= (buf0[i2+1]*yalpha1+buf1[i2+1]*yalpha)>>19; \
int U= (uvbuf0[i ]*uvalpha1+uvbuf1[i ]*uvalpha)>>19; \ int U= (uvbuf0[i ]*uvalpha1+uvbuf1[i ]*uvalpha)>>19; \
int V= (uvbuf0[i+2048]*uvalpha1+uvbuf1[i+2048]*uvalpha)>>19; \ int V= (uvbuf0[i+VOFW]*uvalpha1+uvbuf1[i+VOFW]*uvalpha)>>19; \
#define YSCALE_YUV_2_RGB2_C(type) \ #define YSCALE_YUV_2_RGB2_C(type) \
YSCALE_YUV_2_PACKED2_C\ YSCALE_YUV_2_PACKED2_C\
@@ -519,7 +519,7 @@ static inline void yuv2nv12XinC(int16_t *lumFilter, int16_t **lumSrc, int lumFil
int Y1= buf0[i2 ]>>7;\ int Y1= buf0[i2 ]>>7;\
int Y2= buf0[i2+1]>>7;\ int Y2= buf0[i2+1]>>7;\
int U= (uvbuf1[i ])>>7;\ int U= (uvbuf1[i ])>>7;\
int V= (uvbuf1[i+2048])>>7;\ int V= (uvbuf1[i+VOFW])>>7;\
#define YSCALE_YUV_2_RGB1_C(type) \ #define YSCALE_YUV_2_RGB1_C(type) \
YSCALE_YUV_2_PACKED1_C\ YSCALE_YUV_2_PACKED1_C\
@@ -534,7 +534,7 @@ static inline void yuv2nv12XinC(int16_t *lumFilter, int16_t **lumSrc, int lumFil
int Y1= buf0[i2 ]>>7;\ int Y1= buf0[i2 ]>>7;\
int Y2= buf0[i2+1]>>7;\ int Y2= buf0[i2+1]>>7;\
int U= (uvbuf0[i ] + uvbuf1[i ])>>8;\ int U= (uvbuf0[i ] + uvbuf1[i ])>>8;\
int V= (uvbuf0[i+2048] + uvbuf1[i+2048])>>8;\ int V= (uvbuf0[i+VOFW] + uvbuf1[i+VOFW])>>8;\
#define YSCALE_YUV_2_RGB1B_C(type) \ #define YSCALE_YUV_2_RGB1B_C(type) \
YSCALE_YUV_2_PACKED1B_C\ YSCALE_YUV_2_PACKED1B_C\
@@ -2078,6 +2078,10 @@ SwsContext *sws_getContext(int srcW, int srcH, int srcFormat, int dstW, int dstH
srcW, srcH, dstW, dstH); srcW, srcH, dstW, dstH);
return NULL; return NULL;
} }
if(srcW > VOFW || dstW > VOFW){
av_log(NULL, AV_LOG_ERROR, "swScaler: Compile time max width is "AV_STRINGIFY(VOFW)" change VOF/VOFW and recompile\n");
return NULL;
}
if (!dstFilter) dstFilter= &dummyFilter; if (!dstFilter) dstFilter= &dummyFilter;
if (!srcFilter) srcFilter= &dummyFilter; if (!srcFilter) srcFilter= &dummyFilter;
@@ -2384,12 +2388,14 @@ SwsContext *sws_getContext(int srcW, int srcH, int srcFormat, int dstW, int dstH
//Note we need at least one pixel more at the end because of the mmx code (just in case someone wanna replace the 4000/8000) //Note we need at least one pixel more at the end because of the mmx code (just in case someone wanna replace the 4000/8000)
/* align at 16 bytes for AltiVec */ /* align at 16 bytes for AltiVec */
for (i=0; i<c->vLumBufSize; i++) for (i=0; i<c->vLumBufSize; i++)
c->lumPixBuf[i]= c->lumPixBuf[i+c->vLumBufSize]= av_mallocz(4000); c->lumPixBuf[i]= c->lumPixBuf[i+c->vLumBufSize]= av_mallocz(VOF+1);
for (i=0; i<c->vChrBufSize; i++) for (i=0; i<c->vChrBufSize; i++)
c->chrPixBuf[i]= c->chrPixBuf[i+c->vChrBufSize]= av_malloc(8000); c->chrPixBuf[i]= c->chrPixBuf[i+c->vChrBufSize]= av_malloc((VOF+1)*2);
//try to avoid drawing green stuff between the right end and the stride end //try to avoid drawing green stuff between the right end and the stride end
for (i=0; i<c->vChrBufSize; i++) memset(c->chrPixBuf[i], 64, 8000); for (i=0; i<c->vChrBufSize; i++) memset(c->chrPixBuf[i], 64, (VOF+1)*2);
assert(2*VOF == VOFW);
ASSERT(c->chrDstH <= dstH) ASSERT(c->chrDstH <= dstH)

5
libswscale/swscale_internal.h
View File

@@ -31,6 +31,9 @@
#define MAX_FILTER_SIZE 256 #define MAX_FILTER_SIZE 256
#define VOF 4096
#define VOFW 2048
typedef int (*SwsFunc)(struct SwsContext *context, uint8_t* src[], int srcStride[], int srcSliceY, typedef int (*SwsFunc)(struct SwsContext *context, uint8_t* src[], int srcStride[], int srcSliceY,
int srcSliceH, uint8_t* dst[], int dstStride[]); int srcSliceH, uint8_t* dst[], int dstStride[]);
@@ -70,7 +73,7 @@ typedef struct SwsContext{
int16_t *vChrFilter; int16_t *vChrFilter;
int16_t *vChrFilterPos; int16_t *vChrFilterPos;
uint8_t formatConvBuffer[4000]; //FIXME dynamic alloc, but we have to change a lot of code for this to be useful uint8_t formatConvBuffer[VOF]; //FIXME dynamic alloc, but we have to change a lot of code for this to be useful
int hLumFilterSize; int hLumFilterSize;
int hChrFilterSize; int hChrFilterSize;

122
libswscale/swscale_template.c
View File

@@ -80,8 +80,8 @@
ASMALIGN(4) /* FIXME Unroll? */\ ASMALIGN(4) /* FIXME Unroll? */\
"1: \n\t"\ "1: \n\t"\
"movq 8(%%"REG_d"), %%mm0 \n\t" /* filterCoeff */\ "movq 8(%%"REG_d"), %%mm0 \n\t" /* filterCoeff */\
"movq " #x "(%%"REG_S", %%"REG_a", 2), %%mm2 \n\t" /* srcData */\ "movq " x "(%%"REG_S", %%"REG_a", 2), %%mm2 \n\t" /* srcData */\
"movq 8+" #x "(%%"REG_S", %%"REG_a", 2), %%mm5 \n\t" /* srcData */\ "movq 8+" x "(%%"REG_S", %%"REG_a", 2), %%mm5 \n\t" /* srcData */\
"add $16, %%"REG_d" \n\t"\ "add $16, %%"REG_d" \n\t"\
"mov (%%"REG_d"), %%"REG_S" \n\t"\ "mov (%%"REG_d"), %%"REG_S" \n\t"\
"test %%"REG_S", %%"REG_S" \n\t"\ "test %%"REG_S", %%"REG_S" \n\t"\
@@ -117,10 +117,10 @@
"mov (%%"REG_d"), %%"REG_S" \n\t"\ "mov (%%"REG_d"), %%"REG_S" \n\t"\
ASMALIGN(4) \ ASMALIGN(4) \
"1: \n\t"\ "1: \n\t"\
"movq " #x "(%%"REG_S", %%"REG_a", 2), %%mm0 \n\t" /* srcData */\ "movq " x "(%%"REG_S", %%"REG_a", 2), %%mm0 \n\t" /* srcData */\
"movq 8+" #x "(%%"REG_S", %%"REG_a", 2), %%mm2 \n\t" /* srcData */\ "movq 8+" x "(%%"REG_S", %%"REG_a", 2), %%mm2 \n\t" /* srcData */\
"mov 4(%%"REG_d"), %%"REG_S" \n\t"\ "mov 4(%%"REG_d"), %%"REG_S" \n\t"\
"movq " #x "(%%"REG_S", %%"REG_a", 2), %%mm1 \n\t" /* srcData */\ "movq " x "(%%"REG_S", %%"REG_a", 2), %%mm1 \n\t" /* srcData */\
"movq %%mm0, %%mm3 \n\t"\ "movq %%mm0, %%mm3 \n\t"\
"punpcklwd %%mm1, %%mm0 \n\t"\ "punpcklwd %%mm1, %%mm0 \n\t"\
"punpckhwd %%mm1, %%mm3 \n\t"\ "punpckhwd %%mm1, %%mm3 \n\t"\
@@ -129,7 +129,7 @@
"pmaddwd %%mm1, %%mm3 \n\t"\ "pmaddwd %%mm1, %%mm3 \n\t"\
"paddd %%mm0, %%mm4 \n\t"\ "paddd %%mm0, %%mm4 \n\t"\
"paddd %%mm3, %%mm5 \n\t"\ "paddd %%mm3, %%mm5 \n\t"\
"movq 8+" #x "(%%"REG_S", %%"REG_a", 2), %%mm3 \n\t" /* srcData */\ "movq 8+" x "(%%"REG_S", %%"REG_a", 2), %%mm3 \n\t" /* srcData */\
"mov 16(%%"REG_d"), %%"REG_S" \n\t"\ "mov 16(%%"REG_d"), %%"REG_S" \n\t"\
"add $16, %%"REG_d" \n\t"\ "add $16, %%"REG_d" \n\t"\
"test %%"REG_S", %%"REG_S" \n\t"\ "test %%"REG_S", %%"REG_S" \n\t"\
@@ -202,7 +202,7 @@
"2: \n\t"\ "2: \n\t"\
"movq 8(%%"REG_d"), %%mm0 \n\t" /* filterCoeff */\ "movq 8(%%"REG_d"), %%mm0 \n\t" /* filterCoeff */\
"movq (%%"REG_S", %%"REG_a"), %%mm2 \n\t" /* UsrcData */\ "movq (%%"REG_S", %%"REG_a"), %%mm2 \n\t" /* UsrcData */\
"movq 4096(%%"REG_S", %%"REG_a"), %%mm5 \n\t" /* VsrcData */\ "movq "AV_STRINGIFY(VOF)"(%%"REG_S", %%"REG_a"), %%mm5 \n\t" /* VsrcData */\
"add $16, %%"REG_d" \n\t"\ "add $16, %%"REG_d" \n\t"\
"mov (%%"REG_d"), %%"REG_S" \n\t"\ "mov (%%"REG_d"), %%"REG_S" \n\t"\
"pmulhw %%mm0, %%mm2 \n\t"\ "pmulhw %%mm0, %%mm2 \n\t"\
@@ -252,7 +252,7 @@
ASMALIGN(4)\ ASMALIGN(4)\
"2: \n\t"\ "2: \n\t"\
"movq (%%"REG_S", %%"REG_a"), %%mm0 \n\t" /* UsrcData */\ "movq (%%"REG_S", %%"REG_a"), %%mm0 \n\t" /* UsrcData */\
"movq 4096(%%"REG_S", %%"REG_a"), %%mm2 \n\t" /* VsrcData */\ "movq "AV_STRINGIFY(VOF)"(%%"REG_S", %%"REG_a"), %%mm2 \n\t" /* VsrcData */\
"mov 4(%%"REG_d"), %%"REG_S" \n\t"\ "mov 4(%%"REG_d"), %%"REG_S" \n\t"\
"movq (%%"REG_S", %%"REG_a"), %%mm1 \n\t" /* UsrcData */\ "movq (%%"REG_S", %%"REG_a"), %%mm1 \n\t" /* UsrcData */\
"movq %%mm0, %%mm3 \n\t"\ "movq %%mm0, %%mm3 \n\t"\
@@ -263,7 +263,7 @@
"pmaddwd %%mm1, %%mm3 \n\t"\ "pmaddwd %%mm1, %%mm3 \n\t"\
"paddd %%mm0, %%mm4 \n\t"\ "paddd %%mm0, %%mm4 \n\t"\
"paddd %%mm3, %%mm5 \n\t"\ "paddd %%mm3, %%mm5 \n\t"\
"movq 4096(%%"REG_S", %%"REG_a"), %%mm3 \n\t" /* VsrcData */\ "movq "AV_STRINGIFY(VOF)"(%%"REG_S", %%"REG_a"), %%mm3 \n\t" /* VsrcData */\
"mov 16(%%"REG_d"), %%"REG_S" \n\t"\ "mov 16(%%"REG_d"), %%"REG_S" \n\t"\
"add $16, %%"REG_d" \n\t"\ "add $16, %%"REG_d" \n\t"\
"test %%"REG_S", %%"REG_S" \n\t"\ "test %%"REG_S", %%"REG_S" \n\t"\
@@ -388,10 +388,10 @@
"pmulhw %%mm6, %%mm0 \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\ "pmulhw %%mm6, %%mm0 \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\
"pmulhw %%mm5, %%mm2 \n\t" /* (uvbuf0[eax] - uvbuf1[eax])uvalpha1>>16*/\ "pmulhw %%mm5, %%mm2 \n\t" /* (uvbuf0[eax] - uvbuf1[eax])uvalpha1>>16*/\
"psraw $4, %%mm1 \n\t" /* buf0[eax] - buf1[eax] >>4*/\ "psraw $4, %%mm1 \n\t" /* buf0[eax] - buf1[eax] >>4*/\
"movq 4096(%2, %%"REG_a",2), %%mm4 \n\t" /* uvbuf0[eax+2048]*/\ "movq "AV_STRINGIFY(VOF)"(%2, %%"REG_a",2), %%mm4 \n\t" /* uvbuf0[eax+2048]*/\
"psraw $4, %%mm3 \n\t" /* uvbuf0[eax] - uvbuf1[eax] >>4*/\ "psraw $4, %%mm3 \n\t" /* uvbuf0[eax] - uvbuf1[eax] >>4*/\
"paddw %%mm0, %%mm1 \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\ "paddw %%mm0, %%mm1 \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\
"movq 4096(%3, %%"REG_a",2), %%mm0 \n\t" /* uvbuf1[eax+2048]*/\ "movq "AV_STRINGIFY(VOF)"(%3, %%"REG_a",2), %%mm0 \n\t" /* uvbuf1[eax+2048]*/\
"paddw %%mm2, %%mm3 \n\t" /* uvbuf0[eax]uvalpha1 - uvbuf1[eax](1-uvalpha1)*/\ "paddw %%mm2, %%mm3 \n\t" /* uvbuf0[eax]uvalpha1 - uvbuf1[eax](1-uvalpha1)*/\
"psubw %%mm0, %%mm4 \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048]*/\ "psubw %%mm0, %%mm4 \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048]*/\
"psubw "MANGLE(w80)", %%mm1 \n\t" /* 8(Y-16)*/\ "psubw "MANGLE(w80)", %%mm1 \n\t" /* 8(Y-16)*/\
@@ -434,8 +434,8 @@
"1: \n\t"\ "1: \n\t"\
"movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\ "movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\
"movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\ "movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\
"movq 4096(%2, "#index"), %%mm5 \n\t" /* uvbuf0[eax+2048]*/\ "movq "AV_STRINGIFY(VOF)"(%2, "#index"), %%mm5 \n\t" /* uvbuf0[eax+2048]*/\
"movq 4096(%3, "#index"), %%mm4 \n\t" /* uvbuf1[eax+2048]*/\ "movq "AV_STRINGIFY(VOF)"(%3, "#index"), %%mm4 \n\t" /* uvbuf1[eax+2048]*/\
"psubw %%mm3, %%mm2 \n\t" /* uvbuf0[eax] - uvbuf1[eax]*/\ "psubw %%mm3, %%mm2 \n\t" /* uvbuf0[eax] - uvbuf1[eax]*/\
"psubw %%mm4, %%mm5 \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048]*/\ "psubw %%mm4, %%mm5 \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048]*/\
"movq "CHR_MMX_FILTER_OFFSET"+8("#c"), %%mm0 \n\t"\ "movq "CHR_MMX_FILTER_OFFSET"+8("#c"), %%mm0 \n\t"\
@@ -466,8 +466,8 @@
"1: \n\t"\ "1: \n\t"\
"movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\ "movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\
"movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\ "movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\
"movq 4096(%2, "#index"), %%mm5 \n\t" /* uvbuf0[eax+2048]*/\ "movq "AV_STRINGIFY(VOF)"(%2, "#index"), %%mm5 \n\t" /* uvbuf0[eax+2048]*/\
"movq 4096(%3, "#index"), %%mm4 \n\t" /* uvbuf1[eax+2048]*/\ "movq "AV_STRINGIFY(VOF)"(%3, "#index"), %%mm4 \n\t" /* uvbuf1[eax+2048]*/\
"psubw %%mm3, %%mm2 \n\t" /* uvbuf0[eax] - uvbuf1[eax]*/\ "psubw %%mm3, %%mm2 \n\t" /* uvbuf0[eax] - uvbuf1[eax]*/\
"psubw %%mm4, %%mm5 \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048]*/\ "psubw %%mm4, %%mm5 \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048]*/\
"movq "CHR_MMX_FILTER_OFFSET"+8("#c"), %%mm0 \n\t"\ "movq "CHR_MMX_FILTER_OFFSET"+8("#c"), %%mm0 \n\t"\
@@ -531,7 +531,7 @@
ASMALIGN(4)\ ASMALIGN(4)\
"1: \n\t"\ "1: \n\t"\
"movq (%2, "#index"), %%mm3 \n\t" /* uvbuf0[eax]*/\ "movq (%2, "#index"), %%mm3 \n\t" /* uvbuf0[eax]*/\
"movq 4096(%2, "#index"), %%mm4 \n\t" /* uvbuf0[eax+2048]*/\ "movq "AV_STRINGIFY(VOF)"(%2, "#index"), %%mm4 \n\t" /* uvbuf0[eax+2048]*/\
"psraw $7, %%mm3 \n\t" \ "psraw $7, %%mm3 \n\t" \
"psraw $7, %%mm4 \n\t" \ "psraw $7, %%mm4 \n\t" \
"movq (%0, "#index", 2), %%mm1 \n\t" /*buf0[eax]*/\ "movq (%0, "#index", 2), %%mm1 \n\t" /*buf0[eax]*/\
@@ -546,7 +546,7 @@
ASMALIGN(4)\ ASMALIGN(4)\
"1: \n\t"\ "1: \n\t"\
"movq (%2, "#index"), %%mm3 \n\t" /* uvbuf0[eax]*/\ "movq (%2, "#index"), %%mm3 \n\t" /* uvbuf0[eax]*/\
"movq 4096(%2, "#index"), %%mm4 \n\t" /* uvbuf0[eax+2048]*/\ "movq "AV_STRINGIFY(VOF)"(%2, "#index"), %%mm4 \n\t" /* uvbuf0[eax+2048]*/\
"psraw $4, %%mm3 \n\t" /* uvbuf0[eax] - uvbuf1[eax] >>4*/\ "psraw $4, %%mm3 \n\t" /* uvbuf0[eax] - uvbuf1[eax] >>4*/\
"psraw $4, %%mm4 \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048] >>4*/\ "psraw $4, %%mm4 \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048] >>4*/\
"psubw "U_OFFSET"("#c"), %%mm3 \n\t" /* (U-128)8*/\ "psubw "U_OFFSET"("#c"), %%mm3 \n\t" /* (U-128)8*/\
@@ -596,8 +596,8 @@
"1: \n\t"\ "1: \n\t"\
"movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\ "movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\
"movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\ "movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\
"movq 4096(%2, "#index"), %%mm5 \n\t" /* uvbuf0[eax+2048]*/\ "movq "AV_STRINGIFY(VOF)"(%2, "#index"), %%mm5 \n\t" /* uvbuf0[eax+2048]*/\
"movq 4096(%3, "#index"), %%mm4 \n\t" /* uvbuf1[eax+2048]*/\ "movq "AV_STRINGIFY(VOF)"(%3, "#index"), %%mm4 \n\t" /* uvbuf1[eax+2048]*/\
"paddw %%mm2, %%mm3 \n\t" /* uvbuf0[eax] + uvbuf1[eax]*/\ "paddw %%mm2, %%mm3 \n\t" /* uvbuf0[eax] + uvbuf1[eax]*/\
"paddw %%mm5, %%mm4 \n\t" /* uvbuf0[eax+2048] + uvbuf1[eax+2048]*/\ "paddw %%mm5, %%mm4 \n\t" /* uvbuf0[eax+2048] + uvbuf1[eax+2048]*/\
"psrlw $8, %%mm3 \n\t" \ "psrlw $8, %%mm3 \n\t" \
@@ -615,8 +615,8 @@
"1: \n\t"\ "1: \n\t"\
"movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\ "movq (%2, "#index"), %%mm2 \n\t" /* uvbuf0[eax]*/\
"movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\ "movq (%3, "#index"), %%mm3 \n\t" /* uvbuf1[eax]*/\
"movq 4096(%2, "#index"), %%mm5 \n\t" /* uvbuf0[eax+2048]*/\ "movq "AV_STRINGIFY(VOF)"(%2, "#index"), %%mm5 \n\t" /* uvbuf0[eax+2048]*/\
"movq 4096(%3, "#index"), %%mm4 \n\t" /* uvbuf1[eax+2048]*/\ "movq "AV_STRINGIFY(VOF)"(%3, "#index"), %%mm4 \n\t" /* uvbuf1[eax+2048]*/\
"paddw %%mm2, %%mm3 \n\t" /* uvbuf0[eax] + uvbuf1[eax]*/\ "paddw %%mm2, %%mm3 \n\t" /* uvbuf0[eax] + uvbuf1[eax]*/\
"paddw %%mm5, %%mm4 \n\t" /* uvbuf0[eax+2048] + uvbuf1[eax+2048]*/\ "paddw %%mm5, %%mm4 \n\t" /* uvbuf0[eax+2048] + uvbuf1[eax+2048]*/\
"psrlw $5, %%mm3 \n\t" /*FIXME might overflow*/\ "psrlw $5, %%mm3 \n\t" /*FIXME might overflow*/\
@@ -934,18 +934,18 @@ static inline void RENAME(yuv2yuvX)(SwsContext *c, int16_t *lumFilter, int16_t *
#ifdef HAVE_MMX #ifdef HAVE_MMX
if (c->flags & SWS_ACCURATE_RND){ if (c->flags & SWS_ACCURATE_RND){
if (uDest){ if (uDest){
YSCALEYUV2YV12X_ACCURATE( 0, CHR_MMX_FILTER_OFFSET, uDest, chrDstW) YSCALEYUV2YV12X_ACCURATE( "0", CHR_MMX_FILTER_OFFSET, uDest, chrDstW)
YSCALEYUV2YV12X_ACCURATE(4096, CHR_MMX_FILTER_OFFSET, vDest, chrDstW) YSCALEYUV2YV12X_ACCURATE(AV_STRINGIFY(VOF), CHR_MMX_FILTER_OFFSET, vDest, chrDstW)
} }
YSCALEYUV2YV12X_ACCURATE(0, LUM_MMX_FILTER_OFFSET, dest, dstW) YSCALEYUV2YV12X_ACCURATE("0", LUM_MMX_FILTER_OFFSET, dest, dstW)
}else{ }else{
if (uDest){ if (uDest){
YSCALEYUV2YV12X( 0, CHR_MMX_FILTER_OFFSET, uDest, chrDstW) YSCALEYUV2YV12X( "0", CHR_MMX_FILTER_OFFSET, uDest, chrDstW)
YSCALEYUV2YV12X(4096, CHR_MMX_FILTER_OFFSET, vDest, chrDstW) YSCALEYUV2YV12X(AV_STRINGIFY(VOF), CHR_MMX_FILTER_OFFSET, vDest, chrDstW)
} }
YSCALEYUV2YV12X(0, LUM_MMX_FILTER_OFFSET, dest, dstW) YSCALEYUV2YV12X("0", LUM_MMX_FILTER_OFFSET, dest, dstW)
} }
#else #else
#ifdef HAVE_ALTIVEC #ifdef HAVE_ALTIVEC
@@ -984,7 +984,7 @@ static inline void RENAME(yuv2yuv1)(int16_t *lumSrc, int16_t *chrSrc,
asm volatile( asm volatile(
YSCALEYUV2YV121 YSCALEYUV2YV121
:: "r" (chrSrc + 2048 + chrDstW), "r" (vDest + chrDstW), :: "r" (chrSrc + VOFW + chrDstW), "r" (vDest + chrDstW),
"g" (-chrDstW) "g" (-chrDstW)
: "%"REG_a : "%"REG_a
); );
@@ -1014,7 +1014,7 @@ static inline void RENAME(yuv2yuv1)(int16_t *lumSrc, int16_t *chrSrc,
for (i=0; i<chrDstW; i++) for (i=0; i<chrDstW; i++)
{ {
int u=chrSrc[i]>>7; int u=chrSrc[i]>>7;
int v=chrSrc[i + 2048]>>7; int v=chrSrc[i + VOFW]>>7;
if ((u|v)&256){ if ((u|v)&256){
if (u<0) u=0; if (u<0) u=0;
@@ -1348,7 +1348,7 @@ FULL_YSCALEYUV2RGB
// vertical linear interpolation && yuv2rgb in a single step: // vertical linear interpolation && yuv2rgb in a single step:
int Y=yuvtab_2568[((buf0[i]*yalpha1+buf1[i]*yalpha)>>19)]; int Y=yuvtab_2568[((buf0[i]*yalpha1+buf1[i]*yalpha)>>19)];
int U=((uvbuf0[i]*uvalpha1+uvbuf1[i]*uvalpha)>>19); int U=((uvbuf0[i]*uvalpha1+uvbuf1[i]*uvalpha)>>19);
int V=((uvbuf0[i+2048]*uvalpha1+uvbuf1[i+2048]*uvalpha)>>19); int V=((uvbuf0[i+VOFW]*uvalpha1+uvbuf1[i+VOFW]*uvalpha)>>19);
dest[0]=clip_table[((Y + yuvtab_40cf[U]) >>13)]; dest[0]=clip_table[((Y + yuvtab_40cf[U]) >>13)];
dest[1]=clip_table[((Y + yuvtab_1a1e[V] + yuvtab_0c92[U]) >>13)]; dest[1]=clip_table[((Y + yuvtab_1a1e[V] + yuvtab_0c92[U]) >>13)];
dest[2]=clip_table[((Y + yuvtab_3343[V]) >>13)]; dest[2]=clip_table[((Y + yuvtab_3343[V]) >>13)];
@@ -1362,7 +1362,7 @@ FULL_YSCALEYUV2RGB
// vertical linear interpolation && yuv2rgb in a single step: // vertical linear interpolation && yuv2rgb in a single step:
int Y=yuvtab_2568[((buf0[i]*yalpha1+buf1[i]*yalpha)>>19)]; int Y=yuvtab_2568[((buf0[i]*yalpha1+buf1[i]*yalpha)>>19)];
int U=((uvbuf0[i]*uvalpha1+uvbuf1[i]*uvalpha)>>19); int U=((uvbuf0[i]*uvalpha1+uvbuf1[i]*uvalpha)>>19);
int V=((uvbuf0[i+2048]*uvalpha1+uvbuf1[i+2048]*uvalpha)>>19); int V=((uvbuf0[i+VOFW]*uvalpha1+uvbuf1[i+VOFW]*uvalpha)>>19);
dest[0]=clip_table[((Y + yuvtab_40cf[U]) >>13)]; dest[0]=clip_table[((Y + yuvtab_40cf[U]) >>13)];
dest[1]=clip_table[((Y + yuvtab_1a1e[V] + yuvtab_0c92[U]) >>13)]; dest[1]=clip_table[((Y + yuvtab_1a1e[V] + yuvtab_0c92[U]) >>13)];
dest[2]=clip_table[((Y + yuvtab_3343[V]) >>13)]; dest[2]=clip_table[((Y + yuvtab_3343[V]) >>13)];
@@ -1376,7 +1376,7 @@ FULL_YSCALEYUV2RGB
// vertical linear interpolation && yuv2rgb in a single step: // vertical linear interpolation && yuv2rgb in a single step:
int Y=yuvtab_2568[((buf0[i]*yalpha1+buf1[i]*yalpha)>>19)]; int Y=yuvtab_2568[((buf0[i]*yalpha1+buf1[i]*yalpha)>>19)];
int U=((uvbuf0[i]*uvalpha1+uvbuf1[i]*uvalpha)>>19); int U=((uvbuf0[i]*uvalpha1+uvbuf1[i]*uvalpha)>>19);
int V=((uvbuf0[i+2048]*uvalpha1+uvbuf1[i+2048]*uvalpha)>>19); int V=((uvbuf0[i+VOFW]*uvalpha1+uvbuf1[i+VOFW]*uvalpha)>>19);
((uint16_t*)dest)[i] = ((uint16_t*)dest)[i] =
clip_table16b[(Y + yuvtab_40cf[U]) >>13] | clip_table16b[(Y + yuvtab_40cf[U]) >>13] |
@@ -1391,7 +1391,7 @@ FULL_YSCALEYUV2RGB
// vertical linear interpolation && yuv2rgb in a single step: // vertical linear interpolation && yuv2rgb in a single step:
int Y=yuvtab_2568[((buf0[i]*yalpha1+buf1[i]*yalpha)>>19)]; int Y=yuvtab_2568[((buf0[i]*yalpha1+buf1[i]*yalpha)>>19)];
int U=((uvbuf0[i]*uvalpha1+uvbuf1[i]*uvalpha)>>19); int U=((uvbuf0[i]*uvalpha1+uvbuf1[i]*uvalpha)>>19);
int V=((uvbuf0[i+2048]*uvalpha1+uvbuf1[i+2048]*uvalpha)>>19); int V=((uvbuf0[i+VOFW]*uvalpha1+uvbuf1[i+VOFW]*uvalpha)>>19);
((uint16_t*)dest)[i] = ((uint16_t*)dest)[i] =
clip_table15b[(Y + yuvtab_40cf[U]) >>13] | clip_table15b[(Y + yuvtab_40cf[U]) >>13] |
@@ -2690,63 +2690,63 @@ inline static void RENAME(hcscale)(uint16_t *dst, long dstWidth, uint8_t *src1,
{ {
if (srcFormat==PIX_FMT_YUYV422) if (srcFormat==PIX_FMT_YUYV422)
{ {
RENAME(yuy2ToUV)(formatConvBuffer, formatConvBuffer+2048, src1, src2, srcW); RENAME(yuy2ToUV)(formatConvBuffer, formatConvBuffer+VOFW, src1, src2, srcW);
src1= formatConvBuffer; src1= formatConvBuffer;
src2= formatConvBuffer+2048; src2= formatConvBuffer+VOFW;
} }
else if (srcFormat==PIX_FMT_UYVY422) else if (srcFormat==PIX_FMT_UYVY422)
{ {
RENAME(uyvyToUV)(formatConvBuffer, formatConvBuffer+2048, src1, src2, srcW); RENAME(uyvyToUV)(formatConvBuffer, formatConvBuffer+VOFW, src1, src2, srcW);
src1= formatConvBuffer; src1= formatConvBuffer;
src2= formatConvBuffer+2048; src2= formatConvBuffer+VOFW;
} }
else if (srcFormat==PIX_FMT_RGB32) else if (srcFormat==PIX_FMT_RGB32)
{ {
RENAME(bgr32ToUV)(formatConvBuffer, formatConvBuffer+2048, src1, src2, srcW); RENAME(bgr32ToUV)(formatConvBuffer, formatConvBuffer+VOFW, src1, src2, srcW);
src1= formatConvBuffer; src1= formatConvBuffer;
src2= formatConvBuffer+2048; src2= formatConvBuffer+VOFW;
} }
else if (srcFormat==PIX_FMT_BGR24) else if (srcFormat==PIX_FMT_BGR24)
{ {
RENAME(bgr24ToUV)(formatConvBuffer, formatConvBuffer+2048, src1, src2, srcW); RENAME(bgr24ToUV)(formatConvBuffer, formatConvBuffer+VOFW, src1, src2, srcW);
src1= formatConvBuffer; src1= formatConvBuffer;
src2= formatConvBuffer+2048; src2= formatConvBuffer+VOFW;
} }
else if (srcFormat==PIX_FMT_BGR565) else if (srcFormat==PIX_FMT_BGR565)
{ {
RENAME(bgr16ToUV)(formatConvBuffer, formatConvBuffer+2048, src1, src2, srcW); RENAME(bgr16ToUV)(formatConvBuffer, formatConvBuffer+VOFW, src1, src2, srcW);
src1= formatConvBuffer; src1= formatConvBuffer;
src2= formatConvBuffer+2048; src2= formatConvBuffer+VOFW;
} }
else if (srcFormat==PIX_FMT_BGR555) else if (srcFormat==PIX_FMT_BGR555)
{ {
RENAME(bgr15ToUV)(formatConvBuffer, formatConvBuffer+2048, src1, src2, srcW); RENAME(bgr15ToUV)(formatConvBuffer, formatConvBuffer+VOFW, src1, src2, srcW);
src1= formatConvBuffer; src1= formatConvBuffer;
src2= formatConvBuffer+2048; src2= formatConvBuffer+VOFW;
} }
else if (srcFormat==PIX_FMT_BGR32) else if (srcFormat==PIX_FMT_BGR32)
{ {
RENAME(rgb32ToUV)(formatConvBuffer, formatConvBuffer+2048, src1, src2, srcW); RENAME(rgb32ToUV)(formatConvBuffer, formatConvBuffer+VOFW, src1, src2, srcW);
src1= formatConvBuffer; src1= formatConvBuffer;
src2= formatConvBuffer+2048; src2= formatConvBuffer+VOFW;
} }
else if (srcFormat==PIX_FMT_RGB24) else if (srcFormat==PIX_FMT_RGB24)
{ {
RENAME(rgb24ToUV)(formatConvBuffer, formatConvBuffer+2048, src1, src2, srcW); RENAME(rgb24ToUV)(formatConvBuffer, formatConvBuffer+VOFW, src1, src2, srcW);
src1= formatConvBuffer; src1= formatConvBuffer;
src2= formatConvBuffer+2048; src2= formatConvBuffer+VOFW;
} }
else if (srcFormat==PIX_FMT_RGB565) else if (srcFormat==PIX_FMT_RGB565)
{ {
RENAME(rgb16ToUV)(formatConvBuffer, formatConvBuffer+2048, src1, src2, srcW); RENAME(rgb16ToUV)(formatConvBuffer, formatConvBuffer+VOFW, src1, src2, srcW);
src1= formatConvBuffer; src1= formatConvBuffer;
src2= formatConvBuffer+2048; src2= formatConvBuffer+VOFW;
} }
else if (srcFormat==PIX_FMT_RGB555) else if (srcFormat==PIX_FMT_RGB555)
{ {
RENAME(rgb15ToUV)(formatConvBuffer, formatConvBuffer+2048, src1, src2, srcW); RENAME(rgb15ToUV)(formatConvBuffer, formatConvBuffer+VOFW, src1, src2, srcW);
src1= formatConvBuffer; src1= formatConvBuffer;
src2= formatConvBuffer+2048; src2= formatConvBuffer+VOFW;
} }
else if (isGray(srcFormat)) else if (isGray(srcFormat))
{ {
@@ -2754,9 +2754,9 @@ inline static void RENAME(hcscale)(uint16_t *dst, long dstWidth, uint8_t *src1,
} }
else if (srcFormat==PIX_FMT_RGB8 || srcFormat==PIX_FMT_BGR8 || srcFormat==PIX_FMT_PAL8 || srcFormat==PIX_FMT_BGR4_BYTE || srcFormat==PIX_FMT_RGB4_BYTE) else if (srcFormat==PIX_FMT_RGB8 || srcFormat==PIX_FMT_BGR8 || srcFormat==PIX_FMT_PAL8 || srcFormat==PIX_FMT_BGR4_BYTE || srcFormat==PIX_FMT_RGB4_BYTE)
{ {
RENAME(palToUV)(formatConvBuffer, formatConvBuffer+2048, src1, src2, srcW, pal); RENAME(palToUV)(formatConvBuffer, formatConvBuffer+VOFW, src1, src2, srcW, pal);
src1= formatConvBuffer; src1= formatConvBuffer;
src2= formatConvBuffer+2048; src2= formatConvBuffer+VOFW;
} }
#ifdef HAVE_MMX #ifdef HAVE_MMX
@@ -2767,7 +2767,7 @@ inline static void RENAME(hcscale)(uint16_t *dst, long dstWidth, uint8_t *src1,
#endif #endif
{ {
RENAME(hScale)(dst , dstWidth, src1, srcW, xInc, hChrFilter, hChrFilterPos, hChrFilterSize); RENAME(hScale)(dst , dstWidth, src1, srcW, xInc, hChrFilter, hChrFilterPos, hChrFilterSize);
RENAME(hScale)(dst+2048, dstWidth, src2, srcW, xInc, hChrFilter, hChrFilterPos, hChrFilterSize); RENAME(hScale)(dst+VOFW, dstWidth, src2, srcW, xInc, hChrFilter, hChrFilterPos, hChrFilterSize);
} }
else // Fast Bilinear upscale / crap downscale else // Fast Bilinear upscale / crap downscale
{ {
@@ -2821,7 +2821,7 @@ FUNNY_UV_CODE
"xor %%"REG_a", %%"REG_a" \n\t" // i "xor %%"REG_a", %%"REG_a" \n\t" // i
"mov %5, %%"REG_c" \n\t" // src "mov %5, %%"REG_c" \n\t" // src
"mov %1, %%"REG_D" \n\t" // buf1 "mov %1, %%"REG_D" \n\t" // buf1
"add $4096, %%"REG_D" \n\t" "add $"AV_STRINGIFY(VOF)", %%"REG_D" \n\t"
PREFETCH" (%%"REG_c") \n\t" PREFETCH" (%%"REG_c") \n\t"
PREFETCH" 32(%%"REG_c") \n\t" PREFETCH" 32(%%"REG_c") \n\t"
PREFETCH" 64(%%"REG_c") \n\t" PREFETCH" 64(%%"REG_c") \n\t"
@@ -2848,7 +2848,7 @@ FUNNY_UV_CODE
{ {
//printf("%d %d %d\n", dstWidth, i, srcW); //printf("%d %d %d\n", dstWidth, i, srcW);
dst[i] = src1[srcW-1]*128; dst[i] = src1[srcW-1]*128;
dst[i+2048] = src2[srcW-1]*128; dst[i+VOFW] = src2[srcW-1]*128;
} }
} }
else else
@@ -2881,7 +2881,7 @@ FUNNY_UV_CODE
"addl %%edi, %%esi \n\t" //src[xx+1]*2*xalpha + src[xx]*(1-2*xalpha) "addl %%edi, %%esi \n\t" //src[xx+1]*2*xalpha + src[xx]*(1-2*xalpha)
"mov %1, %%"REG_D" \n\t" "mov %1, %%"REG_D" \n\t"
"shrl $9, %%esi \n\t" "shrl $9, %%esi \n\t"
"movw %%si, 4096(%%"REG_D", %%"REG_a", 2) \n\t" "movw %%si, "AV_STRINGIFY(VOF)"(%%"REG_D", %%"REG_a", 2) \n\t"
"addw %4, %%cx \n\t" //2*xalpha += xInc&0xFF "addw %4, %%cx \n\t" //2*xalpha += xInc&0xFF
"adc %3, %%"REG_d" \n\t" //xx+= xInc>>8 + carry "adc %3, %%"REG_d" \n\t" //xx+= xInc>>8 + carry
@@ -2910,10 +2910,10 @@ FUNNY_UV_CODE
register unsigned int xx=xpos>>16; register unsigned int xx=xpos>>16;
register unsigned int xalpha=(xpos&0xFFFF)>>9; register unsigned int xalpha=(xpos&0xFFFF)>>9;
dst[i]=(src1[xx]*(xalpha^127)+src1[xx+1]*xalpha); dst[i]=(src1[xx]*(xalpha^127)+src1[xx+1]*xalpha);
dst[i+2048]=(src2[xx]*(xalpha^127)+src2[xx+1]*xalpha); dst[i+VOFW]=(src2[xx]*(xalpha^127)+src2[xx+1]*xalpha);
/* slower /* slower
dst[i]= (src1[xx]<<7) + (src1[xx+1] - src1[xx])*xalpha; dst[i]= (src1[xx]<<7) + (src1[xx+1] - src1[xx])*xalpha;
dst[i+2048]=(src2[xx]<<7) + (src2[xx+1] - src2[xx])*xalpha; dst[i+VOFW]=(src2[xx]<<7) + (src2[xx+1] - src2[xx])*xalpha;
*/ */
xpos+=xInc; xpos+=xInc;
} }