falkTX
/
FFmpeg
mirror of https://github.com/falkTX/FFmpeg.git
1
0
Fork 0
Code Issues 0 Releases 338 Wiki Activity
Browse Source

Do not misuse long as the size of a register in x86. 

typedef x86_reg as the appropriate size and use it instead.

Originally committed as revision 13081 to svn://svn.ffmpeg.org/ffmpeg/trunk
tags/v0.5
Ramiro Polla 17 years ago
parent
35027eddf3
commit
40d0e665d0
21 changed files with 187 additions and 175 deletions
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. +4
    -3
      libavcodec/i386/cavsdsp_mmx.c
  2. +1
    -1
      libavcodec/i386/cpuid.c
  3. +2
    -2
      libavcodec/i386/dsputil_h264_template_mmx.c
  4. +4
    -4
      libavcodec/i386/dsputil_h264_template_ssse3.c
  5. +32
    -32
      libavcodec/i386/dsputil_mmx.c
  6. +22
    -22
      libavcodec/i386/dsputil_mmx_avg.h
  7. +2
    -2
      libavcodec/i386/dsputil_mmx_qns.h
  8. +8
    -8
      libavcodec/i386/dsputil_mmx_rnd.h
  9. +20
    -20
      libavcodec/i386/dsputilenc_mmx.c
  10. +3
    -1
      libavcodec/i386/fft_3dn.c
  11. +5
    -2
      libavcodec/i386/fft_3dn2.c
  12. +5
    -2
      libavcodec/i386/fft_sse.c
  13. +5
    -4
      libavcodec/i386/flacdsp_mmx.c
  14. +1
    -1
      libavcodec/i386/h264_i386.h
  15. +28
    -28
      libavcodec/i386/h264dsp_mmx.c
  16. +11
    -11
      libavcodec/i386/motion_est_mmx.c
  17. +6
    -6
      libavcodec/i386/mpegvideo_mmx.c
  18. +1
    -1
      libavcodec/i386/mpegvideo_mmx_template.c
  19. +16
    -16
      libavcodec/i386/snowdsp_mmx.c
  20. +9
    -9
      libavcodec/i386/vc1dsp_mmx.c
  21. +2
    -0
      libavutil/x86_cpu.h

+ 4
- 3
libavcodec/i386/cavsdsp_mmx.c View File

@@ -25,6 +25,7 @@
#include "dsputil.h"
#include "dsputil_mmx.h"
#include "common.h"
#include "x86_cpu.h"

/*****************************************************************************
*
@@ -301,7 +302,7 @@ static void cavs_idct8_add_mmx(uint8_t *dst, int16_t *block, int stride)
VOP(%%mm1, %%mm2, %%mm3, %%mm4, %%mm5, %%mm0, OP)\
\
: "+a"(src), "+c"(dst)\
: "S"((long)srcStride), "D"((long)dstStride), "m"(ADD), "m"(MUL1), "m"(MUL2)\
: "S"((x86_reg)srcStride), "D"((x86_reg)dstStride), "m"(ADD), "m"(MUL1), "m"(MUL2)\
: "memory"\
);\
if(h==16){\
@@ -316,7 +317,7 @@ static void cavs_idct8_add_mmx(uint8_t *dst, int16_t *block, int stride)
VOP(%%mm3, %%mm4, %%mm5, %%mm0, %%mm1, %%mm2, OP)\
\
: "+a"(src), "+c"(dst)\
: "S"((long)srcStride), "D"((long)dstStride), "m"(ADD), "m"(MUL1), "m"(MUL2)\
: "S"((x86_reg)srcStride), "D"((x86_reg)dstStride), "m"(ADD), "m"(MUL1), "m"(MUL2)\
: "memory"\
);\
}\
@@ -367,7 +368,7 @@ static void OPNAME ## cavs_qpel8_h_ ## MMX(uint8_t *dst, uint8_t *src, int dstSt
"decl %2 \n\t"\
" jnz 1b \n\t"\
: "+a"(src), "+c"(dst), "+m"(h)\
: "d"((long)srcStride), "S"((long)dstStride), "m"(ff_pw_5), "m"(ff_pw_4)\
: "d"((x86_reg)srcStride), "S"((x86_reg)dstStride), "m"(ff_pw_5), "m"(ff_pw_4)\
: "memory"\
);\
}\


+ 1
- 1
libavcodec/i386/cpuid.c View File

@@ -42,7 +42,7 @@ int mm_support(void)
int rval = 0;
int eax, ebx, ecx, edx;
int max_std_level, max_ext_level, std_caps=0, ext_caps=0;
long a, c;
x86_reg a, c;

asm volatile (
/* See if CPUID instruction is supported ... */


+ 2
- 2
libavcodec/i386/dsputil_h264_template_mmx.c View File

@@ -249,7 +249,7 @@ static void H264_CHROMA_MC4_TMPL(uint8_t *dst/*align 4*/, uint8_t *src/*align 1*
"sub $2, %2 \n\t"
"jnz 1b \n\t"
: "+r"(dst), "+r"(src), "+r"(h)
: "r"((long)stride), "m"(ff_pw_32), "m"(x), "m"(y)
: "r"((x86_reg)stride), "m"(ff_pw_32), "m"(x), "m"(y)
);
}

@@ -300,7 +300,7 @@ static void H264_CHROMA_MC2_TMPL(uint8_t *dst/*align 2*/, uint8_t *src/*align 1*
"sub $1, %2\n\t"
"jnz 1b\n\t"
: "+r" (dst), "+r"(src), "+r"(h)
: "m" (ff_pw_32), "r"((long)stride)
: "m" (ff_pw_32), "r"((x86_reg)stride)
: "%esi");

}


+ 4
- 4
libavcodec/i386/dsputil_h264_template_ssse3.c View File

@@ -72,7 +72,7 @@ static void H264_CHROMA_MC8_TMPL(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*
"lea (%0,%3,2), %0 \n\t"
"jg 1b \n\t"
:"+r"(dst), "+r"(src), "+r"(h)
:"r"((long)stride)
:"r"((x86_reg)stride)
);
} else {
asm volatile(
@@ -100,7 +100,7 @@ static void H264_CHROMA_MC8_TMPL(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*
"lea (%0,%3,2), %0 \n\t"
"jg 1b \n\t"
:"+r"(dst), "+r"(src), "+r"(h)
:"r"((long)stride)
:"r"((x86_reg)stride)
);
}
return;
@@ -154,7 +154,7 @@ static void H264_CHROMA_MC8_TMPL(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*
"lea (%0,%3,2), %0 \n\t"
"jg 1b \n\t"
:"+r"(dst), "+r"(src), "+r"(h)
:"r"((long)stride)
:"r"((x86_reg)stride)
);
}

@@ -202,7 +202,7 @@ static void H264_CHROMA_MC4_TMPL(uint8_t *dst/*align 4*/, uint8_t *src/*align 1*
"lea (%0,%3,2), %0 \n\t"
"jg 1b \n\t"
:"+r"(dst), "+r"(src), "+r"(h)
:"r"((long)stride)
:"r"((x86_reg)stride)
);
}


+ 32
- 32
libavcodec/i386/dsputil_mmx.c View File

@@ -240,7 +240,7 @@ void put_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size
"movq %%mm2, (%0, %1) \n\t"
"movq %%mm4, (%0, %1, 2) \n\t"
"movq %%mm6, (%0, %2) \n\t"
::"r" (pix), "r" ((long)line_size), "r" ((long)line_size*3), "m"(*p)
::"r" (pix), "r" ((x86_reg)line_size), "r" ((x86_reg)line_size*3), "m"(*p)
:"memory");
pix += line_size*4;
p += 32;
@@ -265,7 +265,7 @@ void put_pixels_clamped_mmx(const DCTELEM *block, uint8_t *pixels, int line_size
"movq %%mm2, (%0, %1) \n\t"
"movq %%mm4, (%0, %1, 2) \n\t"
"movq %%mm6, (%0, %2) \n\t"
::"r" (pix), "r" ((long)line_size), "r" ((long)line_size*3), "r"(p)
::"r" (pix), "r" ((x86_reg)line_size), "r" ((x86_reg)line_size*3), "r"(p)
:"memory");
}

@@ -349,7 +349,7 @@ static void put_pixels4_mmx(uint8_t *block, const uint8_t *pixels, int line_size
"subl $4, %0 \n\t"
"jnz 1b \n\t"
: "+g"(h), "+r" (pixels), "+r" (block)
: "r"((long)line_size)
: "r"((x86_reg)line_size)
: "%"REG_a, "memory"
);
}
@@ -375,7 +375,7 @@ static void put_pixels8_mmx(uint8_t *block, const uint8_t *pixels, int line_size
"subl $4, %0 \n\t"
"jnz 1b \n\t"
: "+g"(h), "+r" (pixels), "+r" (block)
: "r"((long)line_size)
: "r"((x86_reg)line_size)
: "%"REG_a, "memory"
);
}
@@ -409,7 +409,7 @@ static void put_pixels16_mmx(uint8_t *block, const uint8_t *pixels, int line_siz
"subl $4, %0 \n\t"
"jnz 1b \n\t"
: "+g"(h), "+r" (pixels), "+r" (block)
: "r"((long)line_size)
: "r"((x86_reg)line_size)
: "%"REG_a, "memory"
);
}
@@ -431,7 +431,7 @@ static void put_pixels16_sse2(uint8_t *block, const uint8_t *pixels, int line_si
"lea (%2,%3,4), %2 \n\t"
"jnz 1b \n\t"
: "+g"(h), "+r" (pixels), "+r" (block)
: "r"((long)line_size), "r"(3L*line_size)
: "r"((x86_reg)line_size), "r"((x86_reg)3L*line_size)
: "memory"
);
}
@@ -457,7 +457,7 @@ static void avg_pixels16_sse2(uint8_t *block, const uint8_t *pixels, int line_si
"lea (%2,%3,4), %2 \n\t"
"jnz 1b \n\t"
: "+g"(h), "+r" (pixels), "+r" (block)
: "r"((long)line_size), "r"(3L*line_size)
: "r"((x86_reg)line_size), "r"((x86_reg)3L*line_size)
: "memory"
);
}
@@ -480,7 +480,7 @@ static void clear_blocks_mmx(DCTELEM *blocks)
}

static void add_bytes_mmx(uint8_t *dst, uint8_t *src, int w){
long i=0;
x86_reg i=0;
asm volatile(
"1: \n\t"
"movq (%1, %0), %%mm0 \n\t"
@@ -495,14 +495,14 @@ static void add_bytes_mmx(uint8_t *dst, uint8_t *src, int w){
"cmp %3, %0 \n\t"
" jb 1b \n\t"
: "+r" (i)
: "r"(src), "r"(dst), "r"((long)w-15)
: "r"(src), "r"(dst), "r"((x86_reg)w-15)
);
for(; i<w; i++)
dst[i+0] += src[i+0];
}

static void add_bytes_l2_mmx(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w){
long i=0;
x86_reg i=0;
asm volatile(
"1: \n\t"
"movq (%2, %0), %%mm0 \n\t"
@@ -515,7 +515,7 @@ static void add_bytes_l2_mmx(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w){
"cmp %4, %0 \n\t"
" jb 1b \n\t"
: "+r" (i)
: "r"(dst), "r"(src1), "r"(src2), "r"((long)w-15)
: "r"(dst), "r"(src1), "r"(src2), "r"((x86_reg)w-15)
);
for(; i<w; i++)
dst[i] = src1[i] + src2[i];
@@ -689,8 +689,8 @@ static void h263_h_loop_filter_mmx(uint8_t *src, int stride, int qscale){
"movd %%mm6, (%1,%3) \n\t"
:: "r" (src),
"r" (src + 4*stride),
"r" ((long) stride ),
"r" ((long)(3*stride))
"r" ((x86_reg) stride ),
"r" ((x86_reg)(3*stride))
);
}
}
@@ -723,7 +723,7 @@ static void draw_edges_mmx(uint8_t *buf, int wrap, int width, int height, int w)
"cmp %3, %0 \n\t"
" jb 1b \n\t"
: "+r" (ptr)
: "r" ((long)wrap), "r" ((long)width), "r" (ptr + wrap*height)
: "r" ((x86_reg)wrap), "r" ((x86_reg)width), "r" (ptr + wrap*height)
);
}
else
@@ -746,7 +746,7 @@ static void draw_edges_mmx(uint8_t *buf, int wrap, int width, int height, int w)
"cmp %3, %0 \n\t"
" jb 1b \n\t"
: "+r" (ptr)
: "r" ((long)wrap), "r" ((long)width), "r" (ptr + wrap*height)
: "r" ((x86_reg)wrap), "r" ((x86_reg)width), "r" (ptr + wrap*height)
);
}

@@ -764,7 +764,7 @@ static void draw_edges_mmx(uint8_t *buf, int wrap, int width, int height, int w)
"cmp %4, %0 \n\t"
" jb 1b \n\t"
: "+r" (ptr)
: "r" ((long)buf - (long)ptr - w), "r" ((long)-wrap), "r" ((long)-wrap*3), "r" (ptr+width+2*w)
: "r" ((x86_reg)buf - (x86_reg)ptr - w), "r" ((x86_reg)-wrap), "r" ((x86_reg)-wrap*3), "r" (ptr+width+2*w)
);
ptr= last_line + (i + 1) * wrap - w;
asm volatile(
@@ -778,7 +778,7 @@ static void draw_edges_mmx(uint8_t *buf, int wrap, int width, int height, int w)
"cmp %4, %0 \n\t"
" jb 1b \n\t"
: "+r" (ptr)
: "r" ((long)last_line - (long)ptr - w), "r" ((long)wrap), "r" ((long)wrap*3), "r" (ptr+width+2*w)
: "r" ((x86_reg)last_line - (x86_reg)ptr - w), "r" ((x86_reg)wrap), "r" ((x86_reg)wrap*3), "r" (ptr+width+2*w)
);
}
}
@@ -786,8 +786,8 @@ static void draw_edges_mmx(uint8_t *buf, int wrap, int width, int height, int w)
#define PAETH(cpu, abs3)\
void add_png_paeth_prediction_##cpu(uint8_t *dst, uint8_t *src, uint8_t *top, int w, int bpp)\
{\
long i = -bpp;\
long end = w-3;\
x86_reg i = -bpp;\
x86_reg end = w-3;\
asm volatile(\
"pxor %%mm7, %%mm7 \n"\
"movd (%1,%0), %%mm0 \n"\
@@ -830,7 +830,7 @@ void add_png_paeth_prediction_##cpu(uint8_t *dst, uint8_t *src, uint8_t *top, in
"cmp %5, %0 \n"\
"jle 1b \n"\
:"+r"(i)\
:"r"(dst), "r"(top), "r"(src), "r"((long)bpp), "g"(end),\
:"r"(dst), "r"(top), "r"(src), "r"((x86_reg)bpp), "g"(end),\
"m"(ff_pw_255)\
:"memory"\
);\
@@ -994,7 +994,7 @@ static void OPNAME ## mpeg4_qpel16_h_lowpass_mmx2(uint8_t *dst, uint8_t *src, in
"decl %2 \n\t"\
" jnz 1b \n\t"\
: "+a"(src), "+c"(dst), "+D"(h)\
: "d"((long)srcStride), "S"((long)dstStride), /*"m"(ff_pw_20), "m"(ff_pw_3),*/ "m"(temp), "m"(ROUNDER)\
: "d"((x86_reg)srcStride), "S"((x86_reg)dstStride), /*"m"(ff_pw_20), "m"(ff_pw_3),*/ "m"(temp), "m"(ROUNDER)\
: "memory"\
);\
}\
@@ -1105,7 +1105,7 @@ static void OPNAME ## mpeg4_qpel8_h_lowpass_mmx2(uint8_t *dst, uint8_t *src, int
"decl %2 \n\t"\
" jnz 1b \n\t"\
: "+a"(src), "+c"(dst), "+d"(h)\
: "S"((long)srcStride), "D"((long)dstStride), /*"m"(ff_pw_20), "m"(ff_pw_3),*/ "m"(ROUNDER)\
: "S"((x86_reg)srcStride), "D"((x86_reg)dstStride), /*"m"(ff_pw_20), "m"(ff_pw_3),*/ "m"(ROUNDER)\
: "memory"\
);\
}\
@@ -1169,7 +1169,7 @@ static void OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(uint8_t *dst, uint8_t *src,
"decl %2 \n\t"\
" jnz 1b \n\t"\
: "+r" (src), "+r" (temp_ptr), "+r"(count)\
: "r" ((long)srcStride)\
: "r" ((x86_reg)srcStride)\
: "memory"\
);\
\
@@ -1216,7 +1216,7 @@ static void OPNAME ## mpeg4_qpel16_v_lowpass_ ## MMX(uint8_t *dst, uint8_t *src,
" jnz 1b \n\t"\
\
: "+r"(temp_ptr), "+r"(dst), "+g"(count)\
: "r"((long)dstStride), "r"(2*(long)dstStride), /*"m"(ff_pw_20), "m"(ff_pw_3),*/ "m"(ROUNDER), "g"(4-14*(long)dstStride)\
: "r"((x86_reg)dstStride), "r"(2*(x86_reg)dstStride), /*"m"(ff_pw_20), "m"(ff_pw_3),*/ "m"(ROUNDER), "g"(4-14*(x86_reg)dstStride)\
:"memory"\
);\
}\
@@ -1241,7 +1241,7 @@ static void OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(uint8_t *dst, uint8_t *src,
"decl %2 \n\t"\
" jnz 1b \n\t"\
: "+r" (src), "+r" (temp_ptr), "+r"(count)\
: "r" ((long)srcStride)\
: "r" ((x86_reg)srcStride)\
: "memory"\
);\
\
@@ -1276,7 +1276,7 @@ static void OPNAME ## mpeg4_qpel8_v_lowpass_ ## MMX(uint8_t *dst, uint8_t *src,
" jnz 1b \n\t"\
\
: "+r"(temp_ptr), "+r"(dst), "+g"(count)\
: "r"((long)dstStride), "r"(2*(long)dstStride), /*"m"(ff_pw_20), "m"(ff_pw_3),*/ "m"(ROUNDER), "g"(4-6*(long)dstStride)\
: "r"((x86_reg)dstStride), "r"(2*(x86_reg)dstStride), /*"m"(ff_pw_20), "m"(ff_pw_3),*/ "m"(ROUNDER), "g"(4-6*(x86_reg)dstStride)\
: "memory"\
);\
}\
@@ -1839,7 +1839,7 @@ static void vorbis_inverse_coupling_sse(float *mag, float *ang, int blocksize)
}

static void vector_fmul_3dnow(float *dst, const float *src, int len){
long i = (len-4)*4;
x86_reg i = (len-4)*4;
asm volatile(
"1: \n\t"
"movq (%1,%0), %%mm0 \n\t"
@@ -1857,7 +1857,7 @@ static void vector_fmul_3dnow(float *dst, const float *src, int len){
);
}
static void vector_fmul_sse(float *dst, const float *src, int len){
long i = (len-8)*4;
x86_reg i = (len-8)*4;
asm volatile(
"1: \n\t"
"movaps (%1,%0), %%xmm0 \n\t"
@@ -1875,7 +1875,7 @@ static void vector_fmul_sse(float *dst, const float *src, int len){
}

static void vector_fmul_reverse_3dnow2(float *dst, const float *src0, const float *src1, int len){
long i = len*4-16;
x86_reg i = len*4-16;
asm volatile(
"1: \n\t"
"pswapd 8(%1), %%mm0 \n\t"
@@ -1893,7 +1893,7 @@ static void vector_fmul_reverse_3dnow2(float *dst, const float *src0, const floa
asm volatile("femms");
}
static void vector_fmul_reverse_sse(float *dst, const float *src0, const float *src1, int len){
long i = len*4-32;
x86_reg i = len*4-32;
asm volatile(
"1: \n\t"
"movaps 16(%1), %%xmm0 \n\t"
@@ -1914,7 +1914,7 @@ static void vector_fmul_reverse_sse(float *dst, const float *src0, const float *

static void vector_fmul_add_add_3dnow(float *dst, const float *src0, const float *src1,
const float *src2, int src3, int len, int step){
long i = (len-4)*4;
x86_reg i = (len-4)*4;
if(step == 2 && src3 == 0){
dst += (len-4)*2;
asm volatile(
@@ -1963,7 +1963,7 @@ static void vector_fmul_add_add_3dnow(float *dst, const float *src0, const float
}
static void vector_fmul_add_add_sse(float *dst, const float *src0, const float *src1,
const float *src2, int src3, int len, int step){
long i = (len-8)*4;
x86_reg i = (len-8)*4;
if(step == 2 && src3 == 0){
dst += (len-8)*2;
asm volatile(


+ 22
- 22
libavcodec/i386/dsputil_mmx_avg.h View File

@@ -55,7 +55,7 @@ static void DEF(put_pixels8_x2)(uint8_t *block, const uint8_t *pixels, int line_
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r" ((long)line_size)
:"r" ((x86_reg)line_size)
:"%"REG_a, "memory");
}

@@ -105,7 +105,7 @@ static void DEF(put_pixels4_l2)(uint8_t *dst, uint8_t *src1, uint8_t *src2, int
#else
:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
:"S"((long)src1Stride), "D"((long)dstStride)
:"S"((x86_reg)src1Stride), "D"((x86_reg)dstStride)
:"memory");
}

@@ -152,7 +152,7 @@ static void DEF(put_pixels8_l2)(uint8_t *dst, uint8_t *src1, uint8_t *src2, int
#else
:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
:"S"((long)src1Stride), "D"((long)dstStride)
:"S"((x86_reg)src1Stride), "D"((x86_reg)dstStride)
:"memory");
//the following should be used, though better not with gcc ...
/* :"+g"(h), "+r"(src1), "+r"(src2), "+r"(dst)
@@ -222,7 +222,7 @@ static void DEF(put_no_rnd_pixels8_l2)(uint8_t *dst, uint8_t *src1, uint8_t *src
#else
:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
:"S"((long)src1Stride), "D"((long)dstStride)
:"S"((x86_reg)src1Stride), "D"((x86_reg)dstStride)
:"memory");
//the following should be used, though better not with gcc ...
/* :"+g"(h), "+r"(src1), "+r"(src2), "+r"(dst)
@@ -277,7 +277,7 @@ static void DEF(avg_pixels4_l2)(uint8_t *dst, uint8_t *src1, uint8_t *src2, int
#else
:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
:"S"((long)src1Stride), "D"((long)dstStride)
:"S"((x86_reg)src1Stride), "D"((x86_reg)dstStride)
:"memory");
}

@@ -329,7 +329,7 @@ static void DEF(avg_pixels8_l2)(uint8_t *dst, uint8_t *src1, uint8_t *src2, int
#else
:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
:"S"((long)src1Stride), "D"((long)dstStride)
:"S"((x86_reg)src1Stride), "D"((x86_reg)dstStride)
:"memory");
//the following should be used, though better not with gcc ...
/* :"+g"(h), "+r"(src1), "+r"(src2), "+r"(dst)
@@ -373,7 +373,7 @@ static void DEF(put_pixels16_x2)(uint8_t *block, const uint8_t *pixels, int line
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r" ((long)line_size)
:"r" ((x86_reg)line_size)
:"%"REG_a, "memory");
}

@@ -417,7 +417,7 @@ static void DEF(put_pixels16_l2)(uint8_t *dst, uint8_t *src1, uint8_t *src2, int
#else
:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
:"S"((long)src1Stride), "D"((long)dstStride)
:"S"((x86_reg)src1Stride), "D"((x86_reg)dstStride)
:"memory");
//the following should be used, though better not with gcc ...
/* :"+g"(h), "+r"(src1), "+r"(src2), "+r"(dst)
@@ -471,7 +471,7 @@ static void DEF(avg_pixels16_l2)(uint8_t *dst, uint8_t *src1, uint8_t *src2, int
#else
:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
:"S"((long)src1Stride), "D"((long)dstStride)
:"S"((x86_reg)src1Stride), "D"((x86_reg)dstStride)
:"memory");
//the following should be used, though better not with gcc ...
/* :"+g"(h), "+r"(src1), "+r"(src2), "+r"(dst)
@@ -544,7 +544,7 @@ static void DEF(put_no_rnd_pixels16_l2)(uint8_t *dst, uint8_t *src1, uint8_t *sr
#else
:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
:"S"((long)src1Stride), "D"((long)dstStride)
:"S"((x86_reg)src1Stride), "D"((x86_reg)dstStride)
:"memory");
//the following should be used, though better not with gcc ...
/* :"+g"(h), "+r"(src1), "+r"(src2), "+r"(dst)
@@ -586,7 +586,7 @@ static void DEF(put_no_rnd_pixels8_x2)(uint8_t *block, const uint8_t *pixels, in
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r" ((long)line_size)
:"r" ((x86_reg)line_size)
:"%"REG_a, "memory");
}

@@ -616,7 +616,7 @@ static void DEF(put_pixels8_y2)(uint8_t *block, const uint8_t *pixels, int line_
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D" (block)
:"r" ((long)line_size)
:"r" ((x86_reg)line_size)
:"%"REG_a, "memory");
}

@@ -650,7 +650,7 @@ static void DEF(put_no_rnd_pixels8_y2)(uint8_t *block, const uint8_t *pixels, in
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D" (block)
:"r" ((long)line_size)
:"r" ((x86_reg)line_size)
:"%"REG_a, "memory");
}

@@ -678,7 +678,7 @@ static void DEF(avg_pixels8)(uint8_t *block, const uint8_t *pixels, int line_siz
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r" ((long)line_size)
:"r" ((x86_reg)line_size)
:"%"REG_a, "memory");
}

@@ -710,7 +710,7 @@ static void DEF(avg_pixels8_x2)(uint8_t *block, const uint8_t *pixels, int line_
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r" ((long)line_size)
:"r" ((x86_reg)line_size)
:"%"REG_a, "memory");
}

@@ -748,7 +748,7 @@ static void DEF(avg_pixels8_y2)(uint8_t *block, const uint8_t *pixels, int line_
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r" ((long)line_size)
:"r" ((x86_reg)line_size)
:"%"REG_a, "memory");
}

@@ -791,7 +791,7 @@ static void DEF(avg_pixels8_xy2)(uint8_t *block, const uint8_t *pixels, int line
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r" ((long)line_size)
:"r" ((x86_reg)line_size)
:"%"REG_a, "memory");
}

@@ -812,7 +812,7 @@ static void DEF(avg_pixels4)(uint8_t *block, const uint8_t *pixels, int line_siz
"movd %%mm2, (%1, %2, 2) \n\t"
"movd %%mm3, (%1, %3) \n\t"
::"S"(pixels), "D"(block),
"r" ((long)line_size), "r"(3L*line_size)
"r" ((x86_reg)line_size), "r"((x86_reg)3L*line_size)
:"memory");
block += 4*line_size;
pixels += 4*line_size;
@@ -868,8 +868,8 @@ static void DEF(OPNAME ## 2tap_qpel16_l3)(uint8_t *dst, uint8_t *src, int stride
"decl %0 \n\t"\
"jnz 1b \n\t"\
:"+g"(h), "+r"(src)\
:"r"((long)off1), "r"((long)off2),\
"r"((long)(dst-src)), "r"((long)stride)\
:"r"((x86_reg)off1), "r"((x86_reg)off2),\
"r"((x86_reg)(dst-src)), "r"((x86_reg)stride)\
:"memory"\
);\
}\
@@ -885,8 +885,8 @@ static void DEF(OPNAME ## 2tap_qpel8_l3)(uint8_t *dst, uint8_t *src, int stride,
"decl %0 \n\t"\
"jnz 1b \n\t"\
:"+g"(h), "+r"(src)\
:"r"((long)off1), "r"((long)off2),\
"r"((long)(dst-src)), "r"((long)stride)\
:"r"((x86_reg)off1), "r"((x86_reg)off2),\
"r"((x86_reg)(dst-src)), "r"((x86_reg)stride)\
:"memory"\
);\
}


+ 2
- 2
libavcodec/i386/dsputil_mmx_qns.h View File

@@ -30,7 +30,7 @@

static int DEF(try_8x8basis)(int16_t rem[64], int16_t weight[64], int16_t basis[64], int scale)
{
long i=0;
x86_reg i=0;

assert(FFABS(scale) < MAX_ABS);
scale<<= 16 + SCALE_OFFSET - BASIS_SHIFT + RECON_SHIFT;
@@ -72,7 +72,7 @@ static int DEF(try_8x8basis)(int16_t rem[64], int16_t weight[64], int16_t basis[

static void DEF(add_8x8basis)(int16_t rem[64], int16_t basis[64], int scale)
{
long i=0;
x86_reg i=0;

if(FFABS(scale) < MAX_ABS){
scale<<= 16 + SCALE_OFFSET - BASIS_SHIFT + RECON_SHIFT;


+ 8
- 8
libavcodec/i386/dsputil_mmx_rnd.h View File

@@ -57,7 +57,7 @@ static void DEF(put, pixels8_x2)(uint8_t *block, const uint8_t *pixels, int line
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r"((long)line_size)
:"r"((x86_reg)line_size)
:REG_a, "memory");
}

@@ -107,7 +107,7 @@ static void av_unused DEF(put, pixels8_l2)(uint8_t *dst, uint8_t *src1, uint8_t
#else
:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
:"S"((long)src1Stride), "D"((long)dstStride)
:"S"((x86_reg)src1Stride), "D"((x86_reg)dstStride)
:"memory");
}

@@ -153,7 +153,7 @@ static void DEF(put, pixels16_x2)(uint8_t *block, const uint8_t *pixels, int lin
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r"((long)line_size)
:"r"((x86_reg)line_size)
:REG_a, "memory");
}

@@ -202,7 +202,7 @@ static void av_unused DEF(put, pixels16_l2)(uint8_t *dst, uint8_t *src1, uint8_t
#else
:"+b"(h), "+a"(src1), "+c"(src2), "+d"(dst)
#endif
:"S"((long)src1Stride), "D"((long)dstStride)
:"S"((x86_reg)src1Stride), "D"((x86_reg)dstStride)
:"memory");
}

@@ -231,7 +231,7 @@ static void DEF(put, pixels8_y2)(uint8_t *block, const uint8_t *pixels, int line
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r"((long)line_size)
:"r"((x86_reg)line_size)
:REG_a, "memory");
}

@@ -297,7 +297,7 @@ static void DEF(put, pixels8_xy2)(uint8_t *block, const uint8_t *pixels, int lin
"subl $2, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels)
:"D"(block), "r"((long)line_size)
:"D"(block), "r"((x86_reg)line_size)
:REG_a, "memory");
}

@@ -493,7 +493,7 @@ static void DEF(avg, pixels8_y2)(uint8_t *block, const uint8_t *pixels, int line
"subl $4, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels), "+D"(block)
:"r"((long)line_size)
:"r"((x86_reg)line_size)
:REG_a, "memory");
}

@@ -568,7 +568,7 @@ static void DEF(avg, pixels8_xy2)(uint8_t *block, const uint8_t *pixels, int lin
"subl $2, %0 \n\t"
"jnz 1b \n\t"
:"+g"(h), "+S"(pixels)
:"D"(block), "r"((long)line_size)
:"D"(block), "r"((x86_reg)line_size)
:REG_a, "memory");
}



+ 20
- 20
libavcodec/i386/dsputilenc_mmx.c View File

@@ -51,7 +51,7 @@ static void get_pixels_mmx(DCTELEM *block, const uint8_t *pixels, int line_size)
"add $32, %%"REG_a" \n\t"
"js 1b \n\t"
: "+r" (pixels)
: "r" (block+64), "r" ((long)line_size), "r" ((long)line_size*2)
: "r" (block+64), "r" ((x86_reg)line_size), "r" ((x86_reg)line_size*2)
: "%"REG_a
);
}
@@ -80,7 +80,7 @@ static inline void diff_pixels_mmx(DCTELEM *block, const uint8_t *s1, const uint
"add $16, %%"REG_a" \n\t"
"jnz 1b \n\t"
: "+r" (s1), "+r" (s2)
: "r" (block+64), "r" ((long)stride)
: "r" (block+64), "r" ((x86_reg)stride)
: "%"REG_a
);
}
@@ -88,7 +88,7 @@ static inline void diff_pixels_mmx(DCTELEM *block, const uint8_t *s1, const uint
static int pix_sum16_mmx(uint8_t * pix, int line_size){
const int h=16;
int sum;
long index= -line_size*h;
x86_reg index= -line_size*h;

asm volatile(
"pxor %%mm7, %%mm7 \n\t"
@@ -117,7 +117,7 @@ static int pix_sum16_mmx(uint8_t * pix, int line_size){
"movd %%mm6, %0 \n\t"
"andl $0xFFFF, %0 \n\t"
: "=&r" (sum), "+r" (index)
: "r" (pix - index), "r" ((long)line_size)
: "r" (pix - index), "r" ((x86_reg)line_size)
);

return sum;
@@ -162,7 +162,7 @@ static int pix_norm1_mmx(uint8_t *pix, int line_size) {
"psrlq $32, %%mm7\n" /* shift hi dword to lo */
"paddd %%mm7,%%mm1\n"
"movd %%mm1,%1\n"
: "+r" (pix), "=r"(tmp) : "r" ((long)line_size) : "%ecx" );
: "+r" (pix), "=r"(tmp) : "r" ((x86_reg)line_size) : "%ecx" );
return tmp;
}

@@ -222,7 +222,7 @@ static int sse8_mmx(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int
"paddd %%mm7,%%mm1\n"
"movd %%mm1,%2\n"
: "+r" (pix1), "+r" (pix2), "=r"(tmp)
: "r" ((long)line_size) , "m" (h)
: "r" ((x86_reg)line_size) , "m" (h)
: "%ecx");
return tmp;
}
@@ -282,7 +282,7 @@ static int sse16_mmx(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int
"paddd %%mm7,%%mm1\n"
"movd %%mm1,%2\n"
: "+r" (pix1), "+r" (pix2), "=r"(tmp)
: "r" ((long)line_size) , "m" (h)
: "r" ((x86_reg)line_size) , "m" (h)
: "%ecx");
return tmp;
}
@@ -345,7 +345,7 @@ static int sse16_sse2(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, in
"paddd %%xmm1,%%xmm7\n"
"movd %%xmm7,%3\n"
: "+r" (pix1), "+r" (pix2), "+r"(h), "=r"(tmp)
: "r" ((long)line_size));
: "r" ((x86_reg)line_size));
return tmp;
}

@@ -469,7 +469,7 @@ static int hf_noise8_mmx(uint8_t * pix1, int line_size, int h) {
"paddd %%mm6,%%mm0\n"
"movd %%mm0,%1\n"
: "+r" (pix1), "=r"(tmp)
: "r" ((long)line_size) , "g" (h-2)
: "r" ((x86_reg)line_size) , "g" (h-2)
: "%ecx");
return tmp;
}
@@ -583,7 +583,7 @@ static int hf_noise16_mmx(uint8_t * pix1, int line_size, int h) {
"paddd %%mm6,%%mm0\n"
"movd %%mm0,%1\n"
: "+r" (pix1), "=r"(tmp)
: "r" ((long)line_size) , "g" (h-2)
: "r" ((x86_reg)line_size) , "g" (h-2)
: "%ecx");
return tmp + hf_noise8_mmx(pix+8, line_size, h);
}
@@ -665,7 +665,7 @@ static int vsad_intra16_mmx(void *v, uint8_t * pix, uint8_t * dummy, int line_si
"paddw %%mm6,%%mm0\n"
"movd %%mm0,%1\n"
: "+r" (pix), "=r"(tmp)
: "r" ((long)line_size) , "m" (h)
: "r" ((x86_reg)line_size) , "m" (h)
: "%ecx");
return tmp & 0xFFFF;
}
@@ -706,7 +706,7 @@ static int vsad_intra16_mmx2(void *v, uint8_t * pix, uint8_t * dummy, int line_s

"movd %%mm6,%1\n"
: "+r" (pix), "=r"(tmp)
: "r" ((long)line_size) , "m" (h)
: "r" ((x86_reg)line_size) , "m" (h)
: "%ecx");
return tmp;
}
@@ -785,7 +785,7 @@ static int vsad16_mmx(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, in
"paddw %%mm6,%%mm0\n"
"movd %%mm0,%2\n"
: "+r" (pix1), "+r" (pix2), "=r"(tmp)
: "r" ((long)line_size) , "m" (h)
: "r" ((x86_reg)line_size) , "m" (h)
: "%ecx");
return tmp & 0x7FFF;
}
@@ -843,14 +843,14 @@ static int vsad16_mmx2(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, i

"movd %%mm6,%2\n"
: "+r" (pix1), "+r" (pix2), "=r"(tmp)
: "r" ((long)line_size) , "m" (h)
: "r" ((x86_reg)line_size) , "m" (h)
: "%ecx");
return tmp;
}
#undef SUM

static void diff_bytes_mmx(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w){
long i=0;
x86_reg i=0;
asm volatile(
"1: \n\t"
"movq (%2, %0), %%mm0 \n\t"
@@ -865,14 +865,14 @@ static void diff_bytes_mmx(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w){
"cmp %4, %0 \n\t"
" jb 1b \n\t"
: "+r" (i)
: "r"(src1), "r"(src2), "r"(dst), "r"((long)w-15)
: "r"(src1), "r"(src2), "r"(dst), "r"((x86_reg)w-15)
);
for(; i<w; i++)
dst[i+0] = src1[i+0]-src2[i+0];
}

static void sub_hfyu_median_prediction_mmx2(uint8_t *dst, uint8_t *src1, uint8_t *src2, int w, int *left, int *left_top){
long i=0;
x86_reg i=0;
uint8_t l, lt;

asm volatile(
@@ -895,7 +895,7 @@ static void sub_hfyu_median_prediction_mmx2(uint8_t *dst, uint8_t *src1, uint8_t
"cmp %4, %0 \n\t"
" jb 1b \n\t"
: "+r" (i)
: "r"(src1), "r"(src2), "r"(dst), "r"((long)w)
: "r"(src1), "r"(src2), "r"(dst), "r"((x86_reg)w)
);

l= *left;
@@ -930,7 +930,7 @@ static void sub_hfyu_median_prediction_mmx2(uint8_t *dst, uint8_t *src1, uint8_t
DIFF_PIXELS_1(m0, mm##7, mm##0, (%1,%3,4), (%2,%3,4))\
"mov"#m1" %0, "#mm"0 \n\t"\
: "+m"(temp), "+r"(p1b), "+r"(p2b)\
: "r"((long)stride), "r"((long)stride*3)\
: "r"((x86_reg)stride), "r"((x86_reg)stride*3)\
);\
}
//the "+m"(temp) is needed as gcc 2.95 sometimes fails to compile "=m"(temp)
@@ -1237,7 +1237,7 @@ DCT_SAD_FUNC(ssse3)

static int ssd_int8_vs_int16_mmx(const int8_t *pix1, const int16_t *pix2, int size){
int sum;
long i=size;
x86_reg i=size;
asm volatile(
"pxor %%mm4, %%mm4 \n"
"1: \n"


+ 3
- 1
libavcodec/i386/fft_3dn.c View File

@@ -20,6 +20,7 @@
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include "dsputil.h"
#include "x86_cpu.h"

static const int p1m1[2] __attribute__((aligned(8))) =
{ 0, 1 << 31 };
@@ -30,7 +31,8 @@ static const int m1p1[2] __attribute__((aligned(8))) =
void ff_fft_calc_3dn(FFTContext *s, FFTComplex *z)
{
int ln = s->nbits;
long i, j;
long j;
x86_reg i;
long nblocks, nloops;
FFTComplex *p, *cptr;



+ 5
- 2
libavcodec/i386/fft_3dn2.c View File

@@ -20,6 +20,7 @@
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include "dsputil.h"
#include "x86_cpu.h"

static const int p1m1[2] __attribute__((aligned(8))) =
{ 0, 1 << 31 };
@@ -30,7 +31,8 @@ static const int m1p1[2] __attribute__((aligned(8))) =
void ff_fft_calc_3dn2(FFTContext *s, FFTComplex *z)
{
int ln = s->nbits;
long i, j;
long j;
x86_reg i;
long nblocks, nloops;
FFTComplex *p, *cptr;

@@ -124,7 +126,8 @@ void ff_fft_calc_3dn2(FFTContext *s, FFTComplex *z)
void ff_imdct_calc_3dn2(MDCTContext *s, FFTSample *output,
const FFTSample *input, FFTSample *tmp)
{
long k, n8, n4, n2, n;
long n8, n4, n2, n;
x86_reg k;
const uint16_t *revtab = s->fft.revtab;
const FFTSample *tcos = s->tcos;
const FFTSample *tsin = s->tsin;


+ 5
- 2
libavcodec/i386/fft_sse.c View File

@@ -19,6 +19,7 @@
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include "dsputil.h"
#include "x86_cpu.h"

static const int p1p1p1m1[4] __attribute__((aligned(16))) =
{ 0, 0, 0, 1 << 31 };
@@ -48,7 +49,8 @@ static void print_v4sf(const char *str, __m128 a)
void ff_fft_calc_sse(FFTContext *s, FFTComplex *z)
{
int ln = s->nbits;
long i, j;
x86_reg i;
long j;
long nblocks, nloops;
FFTComplex *p, *cptr;

@@ -142,7 +144,8 @@ void ff_fft_calc_sse(FFTContext *s, FFTComplex *z)
void ff_imdct_calc_sse(MDCTContext *s, FFTSample *output,
const FFTSample *input, FFTSample *tmp)
{
long k, n8, n4, n2, n;
x86_reg k;
long n8, n4, n2, n;
const uint16_t *revtab = s->fft.revtab;
const FFTSample *tcos = s->tcos;
const FFTSample *tsin = s->tsin;


+ 5
- 4
libavcodec/i386/flacdsp_mmx.c View File

@@ -20,13 +20,14 @@
*/

#include "dsputil_mmx.h"
#include "x86_cpu.h"

static void apply_welch_window_sse2(const int32_t *data, int len, double *w_data)
{
double c = 2.0 / (len-1.0);
int n2 = len>>1;
long i = -n2*sizeof(int32_t);
long j = n2*sizeof(int32_t);
x86_reg i = -n2*sizeof(int32_t);
x86_reg j = n2*sizeof(int32_t);
asm volatile(
"movsd %0, %%xmm7 \n\t"
"movapd %1, %%xmm6 \n\t"
@@ -71,7 +72,7 @@ void ff_flac_compute_autocorr_sse2(const int32_t *data, int len, int lag,
double *data1 = tmp + lag;
int j;

if((long)data1 & 15)
if((x86_reg)data1 & 15)
data1++;

apply_welch_window_sse2(data, len, data1);
@@ -81,7 +82,7 @@ void ff_flac_compute_autocorr_sse2(const int32_t *data, int len, int lag,
data1[len] = 0.0;

for(j=0; j<lag; j+=2){
long i = -len*sizeof(double);
x86_reg i = -len*sizeof(double);
if(j == lag-2) {
asm volatile(
"movsd %6, %%xmm0 \n\t"


+ 1
- 1
libavcodec/i386/h264_i386.h View File

@@ -96,7 +96,7 @@ static int decode_significance_8x8_x86(CABACContext *c,
int *index, const uint8_t *sig_off){
int minusindex= 4-(int)index;
int coeff_count;
long last=0;
x86_reg last=0;
asm volatile(
"movl "RANGE "(%3), %%esi \n\t"
"movl "LOW "(%3), %%ebx \n\t"


+ 28
- 28
libavcodec/i386/h264dsp_mmx.c View File

@@ -86,7 +86,7 @@ static void ff_h264_idct_add_mmx(uint8_t *dst, int16_t *block, int stride)
"add %1, %0 \n\t"
STORE_DIFF_4P( %%mm4, %%mm1, %%mm7)
: "+r"(dst)
: "r" ((long)stride)
: "r" ((x86_reg)stride)
);
}

@@ -294,7 +294,7 @@ static void ff_h264_idct8_add_sse2(uint8_t *dst, int16_t *block, int stride)
STORE_DIFF_8P(%%xmm0, (%0,%2,2), %%xmm6, %%xmm7)
STORE_DIFF_8P(%%xmm1, (%0,%3), %%xmm6, %%xmm7)
:"+r"(dst)
:"r"(block), "r"((long)stride), "r"(3L*stride), "m"(ff_pw_32)
:"r"(block), "r"((x86_reg)stride), "r"((x86_reg)3L*stride), "m"(ff_pw_32)
);
}

@@ -503,7 +503,7 @@ static inline void h264_loop_filter_luma_mmx2(uint8_t *pix, int stride, int alph
"movq %%mm2, (%2) \n\t"

: "=m"(*tmp0)
: "r"(pix-3*stride), "r"(pix), "r"((long)stride),
: "r"(pix-3*stride), "r"(pix), "r"((x86_reg)stride),
"m"(*tmp0/*unused*/), "m"(*(uint32_t*)tc0), "m"(alpha1), "m"(beta1),
"m"(ff_bone)
);
@@ -550,7 +550,7 @@ static inline void h264_loop_filter_chroma_mmx2(uint8_t *pix, int stride, int al
"movq %%mm1, (%0,%2) \n\t"
"movq %%mm2, (%1) \n\t"

:: "r"(pix-2*stride), "r"(pix), "r"((long)stride),
:: "r"(pix-2*stride), "r"(pix), "r"((x86_reg)stride),
"r"(*(uint32_t*)tc0),
"m"(alpha1), "m"(beta1), "m"(ff_bone), "m"(ff_pb_3F)
);
@@ -601,7 +601,7 @@ static inline void h264_loop_filter_chroma_intra_mmx2(uint8_t *pix, int stride,
"paddb %%mm6, %%mm2 \n\t"
"movq %%mm1, (%0,%2) \n\t"
"movq %%mm2, (%1) \n\t"
:: "r"(pix-2*stride), "r"(pix), "r"((long)stride),
:: "r"(pix-2*stride), "r"(pix), "r"((x86_reg)stride),
"m"(alpha1), "m"(beta1), "m"(ff_bone)
);
}
@@ -797,7 +797,7 @@ static av_noinline void OPNAME ## h264_qpel4_h_lowpass_ ## MMX(uint8_t *dst, uin
"decl %2 \n\t"\
" jnz 1b \n\t"\
: "+a"(src), "+c"(dst), "+g"(h)\
: "d"((long)srcStride), "S"((long)dstStride), "m"(ff_pw_5), "m"(ff_pw_16)\
: "d"((x86_reg)srcStride), "S"((x86_reg)dstStride), "m"(ff_pw_5), "m"(ff_pw_16)\
: "memory"\
);\
}\
@@ -840,7 +840,7 @@ static av_noinline void OPNAME ## h264_qpel4_h_lowpass_l2_ ## MMX(uint8_t *dst,
"add %4, %1 \n\t"\
"add %3, %2 \n\t"\
: "+a"(src), "+c"(dst), "+d"(src2)\
: "D"((long)src2Stride), "S"((long)dstStride)\
: "D"((x86_reg)src2Stride), "S"((x86_reg)dstStride)\
: "memory"\
);\
}while(--h);\
@@ -870,7 +870,7 @@ static av_noinline void OPNAME ## h264_qpel4_v_lowpass_ ## MMX(uint8_t *dst, uin
QPEL_H264V(%%mm3, %%mm4, %%mm5, %%mm0, %%mm1, %%mm2, OP)\
\
: "+a"(src), "+c"(dst)\
: "S"((long)srcStride), "D"((long)dstStride), "m"(ff_pw_5), "m"(ff_pw_16)\
: "S"((x86_reg)srcStride), "D"((x86_reg)dstStride), "m"(ff_pw_5), "m"(ff_pw_16)\
: "memory"\
);\
}\
@@ -902,7 +902,7 @@ static av_noinline void OPNAME ## h264_qpel4_hv_lowpass_ ## MMX(uint8_t *dst, in
QPEL_H264HV(%%mm3, %%mm4, %%mm5, %%mm0, %%mm1, %%mm2, 3*8*3)\
\
: "+a"(src)\
: "c"(tmp), "S"((long)srcStride), "m"(ff_pw_5), "m"(ff_pw_16)\
: "c"(tmp), "S"((x86_reg)srcStride), "m"(ff_pw_5), "m"(ff_pw_16)\
: "memory"\
);\
tmp += 4;\
@@ -931,7 +931,7 @@ static av_noinline void OPNAME ## h264_qpel4_hv_lowpass_ ## MMX(uint8_t *dst, in
"decl %2 \n\t"\
" jnz 1b \n\t"\
: "+a"(tmp), "+c"(dst), "+g"(h)\
: "S"((long)dstStride)\
: "S"((x86_reg)dstStride)\
: "memory"\
);\
}\
@@ -988,7 +988,7 @@ static av_noinline void OPNAME ## h264_qpel8_h_lowpass_ ## MMX(uint8_t *dst, uin
"decl %2 \n\t"\
" jnz 1b \n\t"\
: "+a"(src), "+c"(dst), "+g"(h)\
: "d"((long)srcStride), "S"((long)dstStride), "m"(ff_pw_5), "m"(ff_pw_16)\
: "d"((x86_reg)srcStride), "S"((x86_reg)dstStride), "m"(ff_pw_5), "m"(ff_pw_16)\
: "memory"\
);\
}\
@@ -1049,7 +1049,7 @@ static av_noinline void OPNAME ## h264_qpel8_h_lowpass_l2_ ## MMX(uint8_t *dst,
"add %4, %1 \n\t"\
"add %3, %2 \n\t"\
: "+a"(src), "+c"(dst), "+d"(src2)\
: "D"((long)src2Stride), "S"((long)dstStride),\
: "D"((x86_reg)src2Stride), "S"((x86_reg)dstStride),\
"m"(ff_pw_16)\
: "memory"\
);\
@@ -1088,7 +1088,7 @@ static av_noinline void OPNAME ## h264_qpel8or16_v_lowpass_ ## MMX(uint8_t *dst,
QPEL_H264V(%%mm1, %%mm2, %%mm3, %%mm4, %%mm5, %%mm0, OP)\
\
: "+a"(src), "+c"(dst)\
: "S"((long)srcStride), "D"((long)dstStride), "m"(ff_pw_5), "m"(ff_pw_16)\
: "S"((x86_reg)srcStride), "D"((x86_reg)dstStride), "m"(ff_pw_5), "m"(ff_pw_16)\
: "memory"\
);\
if(h==16){\
@@ -1103,7 +1103,7 @@ static av_noinline void OPNAME ## h264_qpel8or16_v_lowpass_ ## MMX(uint8_t *dst,
QPEL_H264V(%%mm3, %%mm4, %%mm5, %%mm0, %%mm1, %%mm2, OP)\
\
: "+a"(src), "+c"(dst)\
: "S"((long)srcStride), "D"((long)dstStride), "m"(ff_pw_5), "m"(ff_pw_16)\
: "S"((x86_reg)srcStride), "D"((x86_reg)dstStride), "m"(ff_pw_5), "m"(ff_pw_16)\
: "memory"\
);\
}\
@@ -1141,7 +1141,7 @@ static av_always_inline void OPNAME ## h264_qpel8or16_hv1_lowpass_ ## MMX(int16_
QPEL_H264HV(%%mm0, %%mm1, %%mm2, %%mm3, %%mm4, %%mm5, 6*48)\
QPEL_H264HV(%%mm1, %%mm2, %%mm3, %%mm4, %%mm5, %%mm0, 7*48)\
: "+a"(src)\
: "c"(tmp), "S"((long)srcStride), "m"(ff_pw_5), "m"(ff_pw_16)\
: "c"(tmp), "S"((x86_reg)srcStride), "m"(ff_pw_5), "m"(ff_pw_16)\
: "memory"\
);\
if(size==16){\
@@ -1155,7 +1155,7 @@ static av_always_inline void OPNAME ## h264_qpel8or16_hv1_lowpass_ ## MMX(int16_
QPEL_H264HV(%%mm2, %%mm3, %%mm4, %%mm5, %%mm0, %%mm1, 14*48)\
QPEL_H264HV(%%mm3, %%mm4, %%mm5, %%mm0, %%mm1, %%mm2, 15*48)\
: "+a"(src)\
: "c"(tmp), "S"((long)srcStride), "m"(ff_pw_5), "m"(ff_pw_16)\
: "c"(tmp), "S"((x86_reg)srcStride), "m"(ff_pw_5), "m"(ff_pw_16)\
: "memory"\
);\
}\
@@ -1202,7 +1202,7 @@ static av_always_inline void OPNAME ## h264_qpel8or16_hv2_lowpass_ ## MMX(uint8_
"decl %2 \n\t"\
" jnz 1b \n\t"\
: "+a"(tmp), "+c"(dst), "+g"(h)\
: "S"((long)dstStride)\
: "S"((x86_reg)dstStride)\
: "memory"\
);\
tmp += 8 - size*24;\
@@ -1275,7 +1275,7 @@ static av_noinline void OPNAME ## pixels4_l2_shift5_ ## MMX(uint8_t *dst, int16_
OP(%%mm0, (%2), %%mm4, d)\
OP(%%mm1, (%2,%4), %%mm5, d)\
:"+a"(src8), "+c"(src16), "+d"(dst)\
:"S"((long)src8Stride), "D"((long)dstStride)\
:"S"((x86_reg)src8Stride), "D"((x86_reg)dstStride)\
:"memory");\
}\
static av_noinline void OPNAME ## pixels8_l2_shift5_ ## MMX(uint8_t *dst, int16_t *src16, uint8_t *src8, int dstStride, int src8Stride, int h)\
@@ -1297,7 +1297,7 @@ static av_noinline void OPNAME ## pixels8_l2_shift5_ ## MMX(uint8_t *dst, int16_
OP(%%mm0, (%2), %%mm5, q)\
OP(%%mm2, (%2,%4), %%mm5, q)\
::"a"(src8), "c"(src16), "d"(dst),\
"r"((long)src8Stride), "r"((long)dstStride)\
"r"((x86_reg)src8Stride), "r"((x86_reg)dstStride)\
:"memory");\
src8 += 2L*src8Stride;\
src16 += 48;\
@@ -1372,7 +1372,7 @@ static av_noinline void OPNAME ## h264_qpel16_h_lowpass_l2_ ## MMX(uint8_t *dst,
"decl %3 \n\t"\
"jg 1b \n\t"\
: "+a"(src), "+c"(dst), "+d"(src2), "+g"(h)\
: "D"((long)src2Stride), "S"((long)dstStride),\
: "D"((x86_reg)src2Stride), "S"((x86_reg)dstStride),\
"m"(ff_pw_5), "m"(ff_pw_16)\
: "memory"\
);\
@@ -1430,7 +1430,7 @@ static av_noinline void OPNAME ## h264_qpel8_h_lowpass_l2_ ## MMX(uint8_t *dst,
"add %4, %1 \n\t"\
"add %3, %2 \n\t"\
: "+a"(src), "+c"(dst), "+d"(src2)\
: "D"((long)src2Stride), "S"((long)dstStride),\
: "D"((x86_reg)src2Stride), "S"((x86_reg)dstStride),\
"m"(ff_pw_16)\
: "memory"\
);\
@@ -1473,7 +1473,7 @@ static av_noinline void OPNAME ## h264_qpel8_h_lowpass_ ## MMX(uint8_t *dst, uin
"decl %2 \n\t"\
" jnz 1b \n\t"\
: "+a"(src), "+c"(dst), "+g"(h)\
: "D"((long)srcStride), "S"((long)dstStride),\
: "D"((x86_reg)srcStride), "S"((x86_reg)dstStride),\
"m"(ff_pw_5), "m"(ff_pw_16)\
: "memory"\
);\
@@ -1518,7 +1518,7 @@ static av_noinline void OPNAME ## h264_qpel8or16_v_lowpass_ ## MMX(uint8_t *dst,
QPEL_H264V_XMM(%%xmm1, %%xmm2, %%xmm3, %%xmm4, %%xmm5, %%xmm0, OP)\
\
: "+a"(src), "+c"(dst)\
: "S"((long)srcStride), "D"((long)dstStride), "m"(ff_pw_5), "m"(ff_pw_16)\
: "S"((x86_reg)srcStride), "D"((x86_reg)dstStride), "m"(ff_pw_5), "m"(ff_pw_16)\
: "memory"\
);\
if(h==16){\
@@ -1533,7 +1533,7 @@ static av_noinline void OPNAME ## h264_qpel8or16_v_lowpass_ ## MMX(uint8_t *dst,
QPEL_H264V_XMM(%%xmm3, %%xmm4, %%xmm5, %%xmm0, %%xmm1, %%xmm2, OP)\
\
: "+a"(src), "+c"(dst)\
: "S"((long)srcStride), "D"((long)dstStride), "m"(ff_pw_5), "m"(ff_pw_16)\
: "S"((x86_reg)srcStride), "D"((x86_reg)dstStride), "m"(ff_pw_5), "m"(ff_pw_16)\
: "memory"\
);\
}\
@@ -1576,7 +1576,7 @@ static av_always_inline void put_h264_qpel8or16_hv1_lowpass_sse2(int16_t *tmp, u
QPEL_H264HV_XMM(%%xmm0, %%xmm1, %%xmm2, %%xmm3, %%xmm4, %%xmm5, 6*48)
QPEL_H264HV_XMM(%%xmm1, %%xmm2, %%xmm3, %%xmm4, %%xmm5, %%xmm0, 7*48)
: "+a"(src)
: "c"(tmp), "S"((long)srcStride), "m"(ff_pw_5), "m"(ff_pw_16)
: "c"(tmp), "S"((x86_reg)srcStride), "m"(ff_pw_5), "m"(ff_pw_16)
: "memory"
);
if(size==16){
@@ -1590,7 +1590,7 @@ static av_always_inline void put_h264_qpel8or16_hv1_lowpass_sse2(int16_t *tmp, u
QPEL_H264HV_XMM(%%xmm2, %%xmm3, %%xmm4, %%xmm5, %%xmm0, %%xmm1, 14*48)
QPEL_H264HV_XMM(%%xmm3, %%xmm4, %%xmm5, %%xmm0, %%xmm1, %%xmm2, 15*48)
: "+a"(src)
: "c"(tmp), "S"((long)srcStride), "m"(ff_pw_5), "m"(ff_pw_16)
: "c"(tmp), "S"((x86_reg)srcStride), "m"(ff_pw_5), "m"(ff_pw_16)
: "memory"
);
}
@@ -1654,7 +1654,7 @@ static av_always_inline void OPNAME ## h264_qpel8or16_hv2_lowpass_ ## MMX(uint8_
"decl %2 \n\t"\
" jnz 1b \n\t"\
: "+a"(tmp), "+c"(dst), "+g"(h)\
: "S"((long)dstStride)\
: "S"((x86_reg)dstStride)\
: "memory"\
);\
}else{\
@@ -1688,7 +1688,7 @@ static av_always_inline void OPNAME ## h264_qpel8or16_hv2_lowpass_ ## MMX(uint8_
"decl %2 \n\t"\
" jnz 1b \n\t"\
: "+a"(tmp), "+c"(dst), "+g"(h)\
: "S"((long)dstStride)\
: "S"((x86_reg)dstStride)\
: "memory"\
);\
}\


+ 11
- 11
libavcodec/i386/motion_est_mmx.c View File

@@ -34,7 +34,7 @@ DECLARE_ASM_CONST(8, uint64_t, bone)= 0x0101010101010101LL;

static inline void sad8_1_mmx(uint8_t *blk1, uint8_t *blk2, int stride, int h)
{
long len= -(stride*h);
x86_reg len= -(stride*h);
asm volatile(
ASMALIGN(4)
"1: \n\t"
@@ -64,7 +64,7 @@ static inline void sad8_1_mmx(uint8_t *blk1, uint8_t *blk2, int stride, int h)
"add %3, %%"REG_a" \n\t"
" js 1b \n\t"
: "+a" (len)
: "r" (blk1 - len), "r" (blk2 - len), "r" ((long)stride)
: "r" (blk1 - len), "r" (blk2 - len), "r" ((x86_reg)stride)
);
}

@@ -84,7 +84,7 @@ static inline void sad8_1_mmx2(uint8_t *blk1, uint8_t *blk2, int stride, int h)
"sub $2, %0 \n\t"
" jg 1b \n\t"
: "+r" (h), "+r" (blk1), "+r" (blk2)
: "r" ((long)stride)
: "r" ((x86_reg)stride)
);
}

@@ -106,7 +106,7 @@ static int sad16_sse2(void *v, uint8_t *blk2, uint8_t *blk1, int stride, int h)
"sub $2, %0 \n\t"
" jg 1b \n\t"
: "+r" (h), "+r" (blk1), "+r" (blk2)
: "r" ((long)stride)
: "r" ((x86_reg)stride)
);
asm volatile(
"movhlps %%xmm6, %%xmm0 \n\t"
@@ -135,7 +135,7 @@ static inline void sad8_x2a_mmx2(uint8_t *blk1, uint8_t *blk2, int stride, int h
"sub $2, %0 \n\t"
" jg 1b \n\t"
: "+r" (h), "+r" (blk1), "+r" (blk2)
: "r" ((long)stride)
: "r" ((x86_reg)stride)
);
}

@@ -160,7 +160,7 @@ static inline void sad8_y2a_mmx2(uint8_t *blk1, uint8_t *blk2, int stride, int h
"sub $2, %0 \n\t"
" jg 1b \n\t"
: "+r" (h), "+r" (blk1), "+r" (blk2)
: "r" ((long)stride)
: "r" ((x86_reg)stride)
);
}

@@ -190,13 +190,13 @@ static inline void sad8_4_mmx2(uint8_t *blk1, uint8_t *blk2, int stride, int h)
"sub $2, %0 \n\t"
" jg 1b \n\t"
: "+r" (h), "+r" (blk1), "+r" (blk2)
: "r" ((long)stride)
: "r" ((x86_reg)stride)
);
}

static inline void sad8_2_mmx(uint8_t *blk1a, uint8_t *blk1b, uint8_t *blk2, int stride, int h)
{
long len= -(stride*h);
x86_reg len= -(stride*h);
asm volatile(
ASMALIGN(4)
"1: \n\t"
@@ -228,13 +228,13 @@ static inline void sad8_2_mmx(uint8_t *blk1a, uint8_t *blk1b, uint8_t *blk2, int
"add %4, %%"REG_a" \n\t"
" js 1b \n\t"
: "+a" (len)
: "r" (blk1a - len), "r" (blk1b -len), "r" (blk2 - len), "r" ((long)stride)
: "r" (blk1a - len), "r" (blk1b -len), "r" (blk2 - len), "r" ((x86_reg)stride)
);
}

static inline void sad8_4_mmx(uint8_t *blk1, uint8_t *blk2, int stride, int h)
{
long len= -(stride*h);
x86_reg len= -(stride*h);
asm volatile(
"movq (%1, %%"REG_a"), %%mm0 \n\t"
"movq 1(%1, %%"REG_a"), %%mm2 \n\t"
@@ -281,7 +281,7 @@ static inline void sad8_4_mmx(uint8_t *blk1, uint8_t *blk2, int stride, int h)
"add %4, %%"REG_a" \n\t"
" js 1b \n\t"
: "+a" (len)
: "r" (blk1 - len), "r" (blk1 -len + stride), "r" (blk2 - len), "r" ((long)stride)
: "r" (blk1 - len), "r" (blk1 -len + stride), "r" (blk2 - len), "r" ((x86_reg)stride)
);
}



+ 6
- 6
libavcodec/i386/mpegvideo_mmx.c View File

@@ -34,7 +34,7 @@ extern uint16_t inv_zigzag_direct16[64];
static void dct_unquantize_h263_intra_mmx(MpegEncContext *s,
DCTELEM *block, int n, int qscale)
{
long level, qmul, qadd, nCoeffs;
x86_reg level, qmul, qadd, nCoeffs;

qmul = qscale << 1;

@@ -109,7 +109,7 @@ asm volatile(
static void dct_unquantize_h263_inter_mmx(MpegEncContext *s,
DCTELEM *block, int n, int qscale)
{
long qmul, qadd, nCoeffs;
x86_reg qmul, qadd, nCoeffs;

qmul = qscale << 1;
qadd = (qscale - 1) | 1;
@@ -200,7 +200,7 @@ asm volatile(
static void dct_unquantize_mpeg1_intra_mmx(MpegEncContext *s,
DCTELEM *block, int n, int qscale)
{
long nCoeffs;
x86_reg nCoeffs;
const uint16_t *quant_matrix;
int block0;

@@ -269,7 +269,7 @@ asm volatile(
static void dct_unquantize_mpeg1_inter_mmx(MpegEncContext *s,
DCTELEM *block, int n, int qscale)
{
long nCoeffs;
x86_reg nCoeffs;
const uint16_t *quant_matrix;

assert(s->block_last_index[n]>=0);
@@ -335,7 +335,7 @@ asm volatile(
static void dct_unquantize_mpeg2_intra_mmx(MpegEncContext *s,
DCTELEM *block, int n, int qscale)
{
long nCoeffs;
x86_reg nCoeffs;
const uint16_t *quant_matrix;
int block0;

@@ -401,7 +401,7 @@ asm volatile(
static void dct_unquantize_mpeg2_inter_mmx(MpegEncContext *s,
DCTELEM *block, int n, int qscale)
{
long nCoeffs;
x86_reg nCoeffs;
const uint16_t *quant_matrix;

assert(s->block_last_index[n]>=0);


+ 1
- 1
libavcodec/i386/mpegvideo_mmx_template.c View File

@@ -95,7 +95,7 @@ static int RENAME(dct_quantize)(MpegEncContext *s,
DCTELEM *block, int n,
int qscale, int *overflow)
{
long last_non_zero_p1;
x86_reg last_non_zero_p1;
int level=0, q; //=0 is because gcc says uninitialized ...
const uint16_t *qmat, *bias;
DECLARE_ALIGNED_16(int16_t, temp_block[64]);


+ 16
- 16
libavcodec/i386/snowdsp_mmx.c View File

@@ -73,7 +73,7 @@ void ff_snow_horizontal_compose97i_sse2(IDWTELEM *b, int width){
IDWTELEM * const dst = b+w2;

i = 0;
for(; (((long)&dst[i]) & 0x1F) && i<w_r; i++){
for(; (((x86_reg)&dst[i]) & 0x1F) && i<w_r; i++){
dst[i] = dst[i] - (b[i] + b[i + 1]);
}
for(; i<w_r-15; i+=16){
@@ -146,7 +146,7 @@ void ff_snow_horizontal_compose97i_sse2(IDWTELEM *b, int width){
IDWTELEM * const src = b+w2;

i = 0;
for(; (((long)&temp[i]) & 0x1F) && i<w_r; i++){
for(; (((x86_reg)&temp[i]) & 0x1F) && i<w_r; i++){
temp[i] = src[i] - ((-W_AM*(b[i] + b[i+1]))>>W_AS);
}
for(; i<w_r-7; i+=8){
@@ -436,7 +436,7 @@ void ff_snow_horizontal_compose97i_mmx(IDWTELEM *b, int width){
"movdqa %%"s3", %%"t3" \n\t"

void ff_snow_vertical_compose97i_sse2(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, IDWTELEM *b3, IDWTELEM *b4, IDWTELEM *b5, int width){
long i = width;
x86_reg i = width;

while(i & 0x1F)
{
@@ -534,7 +534,7 @@ void ff_snow_vertical_compose97i_sse2(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2,


void ff_snow_vertical_compose97i_mmx(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, IDWTELEM *b3, IDWTELEM *b4, IDWTELEM *b5, int width){
long i = width;
x86_reg i = width;
while(i & 15)
{
i--;
@@ -605,7 +605,7 @@ void ff_snow_vertical_compose97i_mmx(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, I

#define snow_inner_add_yblock_sse2_header \
IDWTELEM * * dst_array = sb->line + src_y;\
long tmp;\
x86_reg tmp;\
asm volatile(\
"mov %7, %%"REG_c" \n\t"\
"mov %6, %2 \n\t"\
@@ -667,7 +667,7 @@ void ff_snow_vertical_compose97i_mmx(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, I
"jnz 1b \n\t"\
:"+m"(dst8),"+m"(dst_array),"=&r"(tmp)\
:\
"rm"((long)(src_x<<1)),"m"(obmc),"a"(block),"m"((long)b_h),"m"((long)src_stride):\
"rm"((x86_reg)(src_x<<1)),"m"(obmc),"a"(block),"m"((x86_reg)b_h),"m"((x86_reg)src_stride):\
"%"REG_c"","%"REG_S"","%"REG_D"","%"REG_d"");

#define snow_inner_add_yblock_sse2_end_8\
@@ -684,8 +684,8 @@ void ff_snow_vertical_compose97i_mmx(IDWTELEM *b0, IDWTELEM *b1, IDWTELEM *b2, I
"dec %2 \n\t"\
snow_inner_add_yblock_sse2_end_common2

static void inner_add_yblock_bw_8_obmc_16_bh_even_sse2(const uint8_t *obmc, const long obmc_stride, uint8_t * * block, int b_w, long b_h,
int src_x, int src_y, long src_stride, slice_buffer * sb, int add, uint8_t * dst8){
static void inner_add_yblock_bw_8_obmc_16_bh_even_sse2(const uint8_t *obmc, const x86_reg obmc_stride, uint8_t * * block, int b_w, x86_reg b_h,
int src_x, int src_y, x86_reg src_stride, slice_buffer * sb, int add, uint8_t * dst8){
snow_inner_add_yblock_sse2_header
snow_inner_add_yblock_sse2_start_8("xmm1", "xmm5", "3", "0")
snow_inner_add_yblock_sse2_accum_8("2", "8")
@@ -732,8 +732,8 @@ snow_inner_add_yblock_sse2_accum_8("0", "136")
snow_inner_add_yblock_sse2_end_8
}

static void inner_add_yblock_bw_16_obmc_32_sse2(const uint8_t *obmc, const long obmc_stride, uint8_t * * block, int b_w, long b_h,
int src_x, int src_y, long src_stride, slice_buffer * sb, int add, uint8_t * dst8){
static void inner_add_yblock_bw_16_obmc_32_sse2(const uint8_t *obmc, const x86_reg obmc_stride, uint8_t * * block, int b_w, x86_reg b_h,
int src_x, int src_y, x86_reg src_stride, slice_buffer * sb, int add, uint8_t * dst8){
snow_inner_add_yblock_sse2_header
snow_inner_add_yblock_sse2_start_16("xmm1", "xmm5", "3", "0")
snow_inner_add_yblock_sse2_accum_16("2", "16")
@@ -758,7 +758,7 @@ snow_inner_add_yblock_sse2_end_16

#define snow_inner_add_yblock_mmx_header \
IDWTELEM * * dst_array = sb->line + src_y;\
long tmp;\
x86_reg tmp;\
asm volatile(\
"mov %7, %%"REG_c" \n\t"\
"mov %6, %2 \n\t"\
@@ -815,11 +815,11 @@ snow_inner_add_yblock_sse2_end_16
"jnz 1b \n\t"\
:"+m"(dst8),"+m"(dst_array),"=&r"(tmp)\
:\
"rm"((long)(src_x<<1)),"m"(obmc),"a"(block),"m"((long)b_h),"m"((long)src_stride):\
"rm"((x86_reg)(src_x<<1)),"m"(obmc),"a"(block),"m"((x86_reg)b_h),"m"((x86_reg)src_stride):\
"%"REG_c"","%"REG_S"","%"REG_D"","%"REG_d"");

static void inner_add_yblock_bw_8_obmc_16_mmx(const uint8_t *obmc, const long obmc_stride, uint8_t * * block, int b_w, long b_h,
int src_x, int src_y, long src_stride, slice_buffer * sb, int add, uint8_t * dst8){
static void inner_add_yblock_bw_8_obmc_16_mmx(const uint8_t *obmc, const x86_reg obmc_stride, uint8_t * * block, int b_w, x86_reg b_h,
int src_x, int src_y, x86_reg src_stride, slice_buffer * sb, int add, uint8_t * dst8){
snow_inner_add_yblock_mmx_header
snow_inner_add_yblock_mmx_start("mm1", "mm5", "3", "0", "0")
snow_inner_add_yblock_mmx_accum("2", "8", "0")
@@ -829,8 +829,8 @@ snow_inner_add_yblock_mmx_mix("0", "0")
snow_inner_add_yblock_mmx_end("16")
}

static void inner_add_yblock_bw_16_obmc_32_mmx(const uint8_t *obmc, const long obmc_stride, uint8_t * * block, int b_w, long b_h,
int src_x, int src_y, long src_stride, slice_buffer * sb, int add, uint8_t * dst8){
static void inner_add_yblock_bw_16_obmc_32_mmx(const uint8_t *obmc, const x86_reg obmc_stride, uint8_t * * block, int b_w, x86_reg b_h,
int src_x, int src_y, x86_reg src_stride, slice_buffer * sb, int add, uint8_t * dst8){
snow_inner_add_yblock_mmx_header
snow_inner_add_yblock_mmx_start("mm1", "mm5", "3", "0", "0")
snow_inner_add_yblock_mmx_accum("2", "16", "0")


+ 9
- 9
libavcodec/i386/vc1dsp_mmx.c View File

@@ -71,7 +71,7 @@ DECLARE_ALIGNED_16(const uint64_t, ff_pw_9) = 0x0009000900090009ULL;

/** Sacrifying mm6 allows to pipeline loads from src */
static void vc1_put_ver_16b_shift2_mmx(int16_t *dst,
const uint8_t *src, long int stride,
const uint8_t *src, x86_reg stride,
int rnd, int64_t shift)
{
asm volatile(
@@ -107,7 +107,7 @@ static void vc1_put_ver_16b_shift2_mmx(int16_t *dst,
* Data is already unpacked, so some operations can directly be made from
* memory.
*/
static void vc1_put_hor_16b_shift2_mmx(uint8_t *dst, long int stride,
static void vc1_put_hor_16b_shift2_mmx(uint8_t *dst, x86_reg stride,
const int16_t *src, int rnd)
{
int h = 8;
@@ -152,7 +152,7 @@ static void vc1_put_hor_16b_shift2_mmx(uint8_t *dst, long int stride,
* Sacrify mm6 for *9 factor.
*/
static void vc1_put_shift2_mmx(uint8_t *dst, const uint8_t *src,
long int stride, int rnd, long int offset)
x86_reg stride, int rnd, x86_reg offset)
{
rnd = 8-rnd;
asm volatile(
@@ -259,7 +259,7 @@ DECLARE_ALIGNED_16(const uint64_t, ff_pw_18) = 0x0012001200120012ULL;
#define MSPEL_FILTER13_VER_16B(NAME, A1, A2, A3, A4) \
static void \
vc1_put_ver_16b_ ## NAME ## _mmx(int16_t *dst, const uint8_t *src, \
long int src_stride, \
x86_reg src_stride, \
int rnd, int64_t shift) \
{ \
int h = 8; \
@@ -314,7 +314,7 @@ vc1_put_ver_16b_ ## NAME ## _mmx(int16_t *dst, const uint8_t *src, \
*/
#define MSPEL_FILTER13_HOR_16B(NAME, A1, A2, A3, A4) \
static void \
vc1_put_hor_16b_ ## NAME ## _mmx(uint8_t *dst, long int stride, \
vc1_put_hor_16b_ ## NAME ## _mmx(uint8_t *dst, x86_reg stride, \
const int16_t *src, int rnd) \
{ \
int h = 8; \
@@ -353,7 +353,7 @@ vc1_put_hor_16b_ ## NAME ## _mmx(uint8_t *dst, long int stride, \
#define MSPEL_FILTER13_8B(NAME, A1, A2, A3, A4) \
static void \
vc1_put_## NAME ## _mmx(uint8_t *dst, const uint8_t *src, \
long int stride, int rnd, long int offset) \
x86_reg stride, int rnd, x86_reg offset) \
{ \
int h = 8; \
src -= offset; \
@@ -387,9 +387,9 @@ MSPEL_FILTER13_8B (shift3, "0(%1 )", "0(%1,%3 )", "0(%1,%3,2)", "0(%1,%
MSPEL_FILTER13_VER_16B(shift3, "0(%1 )", "0(%1,%3 )", "0(%1,%3,2)", "0(%1,%4 )")
MSPEL_FILTER13_HOR_16B(shift3, "2*0(%1)", "2*1(%1)", "2*2(%1)", "2*3(%1)")

typedef void (*vc1_mspel_mc_filter_ver_16bits)(int16_t *dst, const uint8_t *src, long int src_stride, int rnd, int64_t shift);
typedef void (*vc1_mspel_mc_filter_hor_16bits)(uint8_t *dst, long int dst_stride, const int16_t *src, int rnd);
typedef void (*vc1_mspel_mc_filter_8bits)(uint8_t *dst, const uint8_t *src, long int stride, int rnd, long int offset);
typedef void (*vc1_mspel_mc_filter_ver_16bits)(int16_t *dst, const uint8_t *src, x86_reg src_stride, int rnd, int64_t shift);
typedef void (*vc1_mspel_mc_filter_hor_16bits)(uint8_t *dst, x86_reg dst_stride, const int16_t *src, int rnd);
typedef void (*vc1_mspel_mc_filter_8bits)(uint8_t *dst, const uint8_t *src, x86_reg stride, int rnd, x86_reg offset);

/**
* Interpolates fractional pel values by applying proper vertical then


+ 2
- 0
libavutil/x86_cpu.h View File

@@ -31,6 +31,7 @@
# define REG_D "rdi"
# define REG_S "rsi"
# define PTR_SIZE "8"
typedef int64_t x86_reg;

# define REG_SP "rsp"
# define REG_BP "rbp"
@@ -50,6 +51,7 @@
# define REG_D "edi"
# define REG_S "esi"
# define PTR_SIZE "4"
typedef int32_t x86_reg;

# define REG_SP "esp"
# define REG_BP "ebp"


Write Preview
Loading…
Cancel
Save