Merge remote-tracking branch 'qatar/master'

* qatar/master:
  g723.1: simplify scale_vector()
  g723.1: simplify normalize_bits()
  vda: cosmetics: fix Doxygen comment formatting
  vda: better frame allocation
  vda: Merge implementation into one file
  vda: support synchronous decoding
  vda: Reuse the bitstream buffer and reallocate it only if needed
  build: Factor out mpegvideo encoding dependencies to CONFIG_MPEGVIDEOENC
  avprobe: Include libm.h for the log2 fallback
  proresenc: use the edge emulation buffer
  rtmp: handle bytes read reports
  configure: Fix typo in mpeg2video/svq1 decoder dependency declaration
  Use log2(x) instead of log(x) / log(2)
  x86: swscale: fix fragile memory accesses
  x86: swscale: remove disabled code
  x86: yadif: fix asm with suncc
  x86: cabac: allow building with suncc
  x86: mlpdsp: avoid taking address of void
  ARM: intmath: use native-size return types for clipping functions

Conflicts:
	configure
	ffprobe.c
	libavcodec/Makefile
	libavcodec/g723_1.c
	libavcodec/v210dec.h
	libavcodec/vda.h
	libavcodec/vda_h264.c
	libavcodec/x86/cabac.h
	libavfilter/x86/yadif_template.c
	libswscale/x86/rgb2rgb_template.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>

1 files changed, 137 insertions, 231 deletions

diff --git a/libswscale/x86/rgb2rgb_template.c b/libswscale/x86/rgb2rgb_template.c
index 594524d9ed..7e5ffdf8d1 100644
--- a/libswscale/x86/rgb2rgb_template.c
+++ b/libswscale/x86/rgb2rgb_template.c
@@ -73,25 +73,24 @@ static inline void RENAME(rgb24tobgr32)(const uint8_t *src, uint8_t *dst, int sr
     __asm__ volatile("movq        %0, %%mm7"::"m"(mask32a):"memory");
     while (s < mm_end) {
         __asm__ volatile(
-            PREFETCH"    32%1           \n\t"
-            "movd          %1, %%mm0    \n\t"
-            "punpckldq    3%1, %%mm0    \n\t"
-            "movd         6%1, %%mm1    \n\t"
-            "punpckldq    9%1, %%mm1    \n\t"
-            "movd        12%1, %%mm2    \n\t"
-            "punpckldq   15%1, %%mm2    \n\t"
-            "movd        18%1, %%mm3    \n\t"
-            "punpckldq   21%1, %%mm3    \n\t"
+            PREFETCH"  32(%1)           \n\t"
+            "movd        (%1), %%mm0    \n\t"
+            "punpckldq  3(%1), %%mm0    \n\t"
+            "movd       6(%1), %%mm1    \n\t"
+            "punpckldq  9(%1), %%mm1    \n\t"
+            "movd      12(%1), %%mm2    \n\t"
+            "punpckldq 15(%1), %%mm2    \n\t"
+            "movd      18(%1), %%mm3    \n\t"
+            "punpckldq 21(%1), %%mm3    \n\t"
             "por        %%mm7, %%mm0    \n\t"
             "por        %%mm7, %%mm1    \n\t"
             "por        %%mm7, %%mm2    \n\t"
             "por        %%mm7, %%mm3    \n\t"
-            MOVNTQ"     %%mm0,   %0     \n\t"
-            MOVNTQ"     %%mm1,  8%0     \n\t"
-            MOVNTQ"     %%mm2, 16%0     \n\t"
-            MOVNTQ"     %%mm3, 24%0"
-            :"=m"(*dest)
-            :"m"(*s)
+            MOVNTQ"     %%mm0,   (%0)   \n\t"
+            MOVNTQ"     %%mm1,  8(%0)   \n\t"
+            MOVNTQ"     %%mm2, 16(%0)   \n\t"
+            MOVNTQ"     %%mm3, 24(%0)"
+            :: "r"(dest), "r"(s)
             :"memory");
         dest += 32;
         s += 24;
@@ -138,9 +137,9 @@ static inline void RENAME(rgb24tobgr32)(const uint8_t *src, uint8_t *dst, int sr
             "pand  "MANGLE(mask24hhhh)", %%mm5\n\t" \
             "por        %%mm5, %%mm4    \n\t" \
  \
-            MOVNTQ"     %%mm0,   %0     \n\t" \
-            MOVNTQ"     %%mm1,  8%0     \n\t" \
-            MOVNTQ"     %%mm4, 16%0"
+            MOVNTQ"     %%mm0,   (%0)    \n\t" \
+            MOVNTQ"     %%mm1,  8(%0)    \n\t" \
+            MOVNTQ"     %%mm4, 16(%0)"
 
 
 static inline void RENAME(rgb32tobgr24)(const uint8_t *src, uint8_t *dst, int src_size)
@@ -154,18 +153,17 @@ static inline void RENAME(rgb32tobgr24)(const uint8_t *src, uint8_t *dst, int sr
     mm_end = end - 31;
     while (s < mm_end) {
         __asm__ volatile(
-            PREFETCH"    32%1           \n\t"
-            "movq          %1, %%mm0    \n\t"
-            "movq         8%1, %%mm1    \n\t"
-            "movq        16%1, %%mm4    \n\t"
-            "movq        24%1, %%mm5    \n\t"
+            PREFETCH"  32(%1)           \n\t"
+            "movq        (%1), %%mm0    \n\t"
+            "movq       8(%1), %%mm1    \n\t"
+            "movq      16(%1), %%mm4    \n\t"
+            "movq      24(%1), %%mm5    \n\t"
             "movq       %%mm0, %%mm2    \n\t"
             "movq       %%mm1, %%mm3    \n\t"
             "movq       %%mm4, %%mm6    \n\t"
             "movq       %%mm5, %%mm7    \n\t"
             STORE_BGR24_MMX
-            :"=m"(*dest)
-            :"m"(*s)
+            :: "r"(dest), "r"(s)
             :"memory");
         dest += 24;
         s += 32;
@@ -198,19 +196,18 @@ static inline void RENAME(rgb15to16)(const uint8_t *src, uint8_t *dst, int src_s
     mm_end = end - 15;
     while (s<mm_end) {
         __asm__ volatile(
-            PREFETCH"  32%1         \n\t"
-            "movq        %1, %%mm0  \n\t"
-            "movq       8%1, %%mm2  \n\t"
+            PREFETCH" 32(%1)        \n\t"
+            "movq      (%1), %%mm0  \n\t"
+            "movq     8(%1), %%mm2  \n\t"
             "movq     %%mm0, %%mm1  \n\t"
             "movq     %%mm2, %%mm3  \n\t"
             "pand     %%mm4, %%mm0  \n\t"
             "pand     %%mm4, %%mm2  \n\t"
             "paddw    %%mm1, %%mm0  \n\t"
             "paddw    %%mm3, %%mm2  \n\t"
-            MOVNTQ"   %%mm0,  %0    \n\t"
-            MOVNTQ"   %%mm2, 8%0"
-            :"=m"(*d)
-            :"m"(*s)
+            MOVNTQ"   %%mm0,  (%0)  \n\t"
+            MOVNTQ"   %%mm2, 8(%0)"
+            :: "r"(d), "r"(s)
         );
         d+=16;
         s+=16;
@@ -243,9 +240,9 @@ static inline void RENAME(rgb16to15)(const uint8_t *src, uint8_t *dst, int src_s
     mm_end = end - 15;
     while (s<mm_end) {
         __asm__ volatile(
-            PREFETCH"  32%1         \n\t"
-            "movq        %1, %%mm0  \n\t"
-            "movq       8%1, %%mm2  \n\t"
+            PREFETCH" 32(%1)        \n\t"
+            "movq      (%1), %%mm0  \n\t"
+            "movq     8(%1), %%mm2  \n\t"
             "movq     %%mm0, %%mm1  \n\t"
             "movq     %%mm2, %%mm3  \n\t"
             "psrlq       $1, %%mm0  \n\t"
@@ -256,10 +253,9 @@ static inline void RENAME(rgb16to15)(const uint8_t *src, uint8_t *dst, int src_s
             "pand     %%mm6, %%mm3  \n\t"
             "por      %%mm1, %%mm0  \n\t"
             "por      %%mm3, %%mm2  \n\t"
-            MOVNTQ"   %%mm0,  %0    \n\t"
-            MOVNTQ"   %%mm2, 8%0"
-            :"=m"(*d)
-            :"m"(*s)
+            MOVNTQ"   %%mm0,  (%0)  \n\t"
+            MOVNTQ"   %%mm2, 8(%0)"
+            :: "r"(d), "r"(s)
         );
         d+=16;
         s+=16;
@@ -287,7 +283,6 @@ static inline void RENAME(rgb32to16)(const uint8_t *src, uint8_t *dst, int src_s
     uint16_t *d = (uint16_t *)dst;
     end = s + src_size;
     mm_end = end - 15;
-#if 1 //is faster only if multiplies are reasonably fast (FIXME figure out on which CPUs this is faster, on Athlon it is slightly faster)
     __asm__ volatile(
         "movq           %3, %%mm5   \n\t"
         "movq           %4, %%mm6   \n\t"
@@ -322,47 +317,6 @@ static inline void RENAME(rgb32to16)(const uint8_t *src, uint8_t *dst, int src_s
         : "+r" (d), "+r"(s)
         : "r" (mm_end), "m" (mask3216g), "m" (mask3216br), "m" (mul3216)
     );
-#else
-    __asm__ volatile(PREFETCH"    %0"::"m"(*src):"memory");
-    __asm__ volatile(
-        "movq    %0, %%mm7    \n\t"
-        "movq    %1, %%mm6    \n\t"
-        ::"m"(red_16mask),"m"(green_16mask));
-    while (s < mm_end) {
-        __asm__ volatile(
-            PREFETCH"    32%1           \n\t"
-            "movd          %1, %%mm0    \n\t"
-            "movd         4%1, %%mm3    \n\t"
-            "punpckldq    8%1, %%mm0    \n\t"
-            "punpckldq   12%1, %%mm3    \n\t"
-            "movq       %%mm0, %%mm1    \n\t"
-            "movq       %%mm0, %%mm2    \n\t"
-            "movq       %%mm3, %%mm4    \n\t"
-            "movq       %%mm3, %%mm5    \n\t"
-            "psrlq         $3, %%mm0    \n\t"
-            "psrlq         $3, %%mm3    \n\t"
-            "pand          %2, %%mm0    \n\t"
-            "pand          %2, %%mm3    \n\t"
-            "psrlq         $5, %%mm1    \n\t"
-            "psrlq         $5, %%mm4    \n\t"
-            "pand       %%mm6, %%mm1    \n\t"
-            "pand       %%mm6, %%mm4    \n\t"
-            "psrlq         $8, %%mm2    \n\t"
-            "psrlq         $8, %%mm5    \n\t"
-            "pand       %%mm7, %%mm2    \n\t"
-            "pand       %%mm7, %%mm5    \n\t"
-            "por        %%mm1, %%mm0    \n\t"
-            "por        %%mm4, %%mm3    \n\t"
-            "por        %%mm2, %%mm0    \n\t"
-            "por        %%mm5, %%mm3    \n\t"
-            "psllq        $16, %%mm3    \n\t"
-            "por        %%mm3, %%mm0    \n\t"
-            MOVNTQ"     %%mm0, %0       \n\t"
-            :"=m"(*d):"m"(*s),"m"(blue_16mask):"memory");
-        d += 4;
-        s += 16;
-    }
-#endif
     __asm__ volatile(SFENCE:::"memory");
     __asm__ volatile(EMMS:::"memory");
     while (s < end) {
@@ -386,11 +340,11 @@ static inline void RENAME(rgb32tobgr16)(const uint8_t *src, uint8_t *dst, int sr
     mm_end = end - 15;
     while (s < mm_end) {
         __asm__ volatile(
-            PREFETCH"    32%1           \n\t"
-            "movd          %1, %%mm0    \n\t"
-            "movd         4%1, %%mm3    \n\t"
-            "punpckldq    8%1, %%mm0    \n\t"
-            "punpckldq   12%1, %%mm3    \n\t"
+            PREFETCH"  32(%1)           \n\t"
+            "movd        (%1), %%mm0    \n\t"
+            "movd       4(%1), %%mm3    \n\t"
+            "punpckldq  8(%1), %%mm0    \n\t"
+            "punpckldq 12(%1), %%mm3    \n\t"
             "movq       %%mm0, %%mm1    \n\t"
             "movq       %%mm0, %%mm2    \n\t"
             "movq       %%mm3, %%mm4    \n\t"
@@ -413,8 +367,8 @@ static inline void RENAME(rgb32tobgr16)(const uint8_t *src, uint8_t *dst, int sr
             "por        %%mm5, %%mm3    \n\t"
             "psllq        $16, %%mm3    \n\t"
             "por        %%mm3, %%mm0    \n\t"
-            MOVNTQ"     %%mm0, %0       \n\t"
-            :"=m"(*d):"m"(*s),"m"(blue_16mask):"memory");
+            MOVNTQ"     %%mm0, (%0)     \n\t"
+            :: "r"(d),"r"(s),"m"(blue_16mask):"memory");
         d += 4;
         s += 16;
     }
@@ -434,7 +388,6 @@ static inline void RENAME(rgb32to15)(const uint8_t *src, uint8_t *dst, int src_s
     uint16_t *d = (uint16_t *)dst;
     end = s + src_size;
     mm_end = end - 15;
-#if 1 //is faster only if multiplies are reasonably fast (FIXME figure out on which CPUs this is faster, on Athlon it is slightly faster)
     __asm__ volatile(
         "movq           %3, %%mm5   \n\t"
         "movq           %4, %%mm6   \n\t"
@@ -469,47 +422,6 @@ static inline void RENAME(rgb32to15)(const uint8_t *src, uint8_t *dst, int src_s
         : "+r" (d), "+r"(s)
         : "r" (mm_end), "m" (mask3215g), "m" (mask3216br), "m" (mul3215)
     );
-#else
-    __asm__ volatile(PREFETCH"    %0"::"m"(*src):"memory");
-    __asm__ volatile(
-        "movq          %0, %%mm7    \n\t"
-        "movq          %1, %%mm6    \n\t"
-        ::"m"(red_15mask),"m"(green_15mask));
-    while (s < mm_end) {
-        __asm__ volatile(
-            PREFETCH"    32%1           \n\t"
-            "movd          %1, %%mm0    \n\t"
-            "movd         4%1, %%mm3    \n\t"
-            "punpckldq    8%1, %%mm0    \n\t"
-            "punpckldq   12%1, %%mm3    \n\t"
-            "movq       %%mm0, %%mm1    \n\t"
-            "movq       %%mm0, %%mm2    \n\t"
-            "movq       %%mm3, %%mm4    \n\t"
-            "movq       %%mm3, %%mm5    \n\t"
-            "psrlq         $3, %%mm0    \n\t"
-            "psrlq         $3, %%mm3    \n\t"
-            "pand          %2, %%mm0    \n\t"
-            "pand          %2, %%mm3    \n\t"
-            "psrlq         $6, %%mm1    \n\t"
-            "psrlq         $6, %%mm4    \n\t"
-            "pand       %%mm6, %%mm1    \n\t"
-            "pand       %%mm6, %%mm4    \n\t"
-            "psrlq         $9, %%mm2    \n\t"
-            "psrlq         $9, %%mm5    \n\t"
-            "pand       %%mm7, %%mm2    \n\t"
-            "pand       %%mm7, %%mm5    \n\t"
-            "por        %%mm1, %%mm0    \n\t"
-            "por        %%mm4, %%mm3    \n\t"
-            "por        %%mm2, %%mm0    \n\t"
-            "por        %%mm5, %%mm3    \n\t"
-            "psllq        $16, %%mm3    \n\t"
-            "por        %%mm3, %%mm0    \n\t"
-            MOVNTQ"     %%mm0, %0       \n\t"
-            :"=m"(*d):"m"(*s),"m"(blue_15mask):"memory");
-        d += 4;
-        s += 16;
-    }
-#endif
     __asm__ volatile(SFENCE:::"memory");
     __asm__ volatile(EMMS:::"memory");
     while (s < end) {
@@ -533,11 +445,11 @@ static inline void RENAME(rgb32tobgr15)(const uint8_t *src, uint8_t *dst, int sr
     mm_end = end - 15;
     while (s < mm_end) {
         __asm__ volatile(
-            PREFETCH"    32%1           \n\t"
-            "movd          %1, %%mm0    \n\t"
-            "movd         4%1, %%mm3    \n\t"
-            "punpckldq    8%1, %%mm0    \n\t"
-            "punpckldq   12%1, %%mm3    \n\t"
+            PREFETCH"  32(%1)           \n\t"
+            "movd        (%1), %%mm0    \n\t"
+            "movd       4(%1), %%mm3    \n\t"
+            "punpckldq  8(%1), %%mm0    \n\t"
+            "punpckldq 12(%1), %%mm3    \n\t"
             "movq       %%mm0, %%mm1    \n\t"
             "movq       %%mm0, %%mm2    \n\t"
             "movq       %%mm3, %%mm4    \n\t"
@@ -560,8 +472,8 @@ static inline void RENAME(rgb32tobgr15)(const uint8_t *src, uint8_t *dst, int sr
             "por        %%mm5, %%mm3    \n\t"
             "psllq        $16, %%mm3    \n\t"
             "por        %%mm3, %%mm0    \n\t"
-            MOVNTQ"     %%mm0, %0       \n\t"
-            :"=m"(*d):"m"(*s),"m"(blue_15mask):"memory");
+            MOVNTQ"     %%mm0, (%0)     \n\t"
+            ::"r"(d),"r"(s),"m"(blue_15mask):"memory");
         d += 4;
         s += 16;
     }
@@ -588,11 +500,11 @@ static inline void RENAME(rgb24tobgr16)(const uint8_t *src, uint8_t *dst, int sr
     mm_end = end - 11;
     while (s < mm_end) {
         __asm__ volatile(
-            PREFETCH"    32%1           \n\t"
-            "movd          %1, %%mm0    \n\t"
-            "movd         3%1, %%mm3    \n\t"
-            "punpckldq    6%1, %%mm0    \n\t"
-            "punpckldq    9%1, %%mm3    \n\t"
+            PREFETCH"  32(%1)           \n\t"
+            "movd        (%1), %%mm0    \n\t"
+            "movd       3(%1), %%mm3    \n\t"
+            "punpckldq  6(%1), %%mm0    \n\t"
+            "punpckldq  9(%1), %%mm3    \n\t"
             "movq       %%mm0, %%mm1    \n\t"
             "movq       %%mm0, %%mm2    \n\t"
             "movq       %%mm3, %%mm4    \n\t"
@@ -615,8 +527,8 @@ static inline void RENAME(rgb24tobgr16)(const uint8_t *src, uint8_t *dst, int sr
             "por        %%mm5, %%mm3    \n\t"
             "psllq        $16, %%mm3    \n\t"
             "por        %%mm3, %%mm0    \n\t"
-            MOVNTQ"     %%mm0, %0       \n\t"
-            :"=m"(*d):"m"(*s),"m"(blue_16mask):"memory");
+            MOVNTQ"     %%mm0, (%0)     \n\t"
+            ::"r"(d),"r"(s),"m"(blue_16mask):"memory");
         d += 4;
         s += 12;
     }
@@ -645,11 +557,11 @@ static inline void RENAME(rgb24to16)(const uint8_t *src, uint8_t *dst, int src_s
     mm_end = end - 15;
     while (s < mm_end) {
         __asm__ volatile(
-            PREFETCH"    32%1           \n\t"
-            "movd          %1, %%mm0    \n\t"
-            "movd         3%1, %%mm3    \n\t"
-            "punpckldq    6%1, %%mm0    \n\t"
-            "punpckldq    9%1, %%mm3    \n\t"
+            PREFETCH"  32(%1)           \n\t"
+            "movd        (%1), %%mm0    \n\t"
+            "movd       3(%1), %%mm3    \n\t"
+            "punpckldq  6(%1), %%mm0    \n\t"
+            "punpckldq  9(%1), %%mm3    \n\t"
             "movq       %%mm0, %%mm1    \n\t"
             "movq       %%mm0, %%mm2    \n\t"
             "movq       %%mm3, %%mm4    \n\t"
@@ -672,8 +584,8 @@ static inline void RENAME(rgb24to16)(const uint8_t *src, uint8_t *dst, int src_s
             "por        %%mm5, %%mm3    \n\t"
             "psllq        $16, %%mm3    \n\t"
             "por        %%mm3, %%mm0    \n\t"
-            MOVNTQ"     %%mm0, %0       \n\t"
-            :"=m"(*d):"m"(*s),"m"(blue_16mask):"memory");
+            MOVNTQ"     %%mm0, (%0)     \n\t"
+            ::"r"(d),"r"(s),"m"(blue_16mask):"memory");
         d += 4;
         s += 12;
     }
@@ -702,11 +614,11 @@ static inline void RENAME(rgb24tobgr15)(const uint8_t *src, uint8_t *dst, int sr
     mm_end = end - 11;
     while (s < mm_end) {
         __asm__ volatile(
-            PREFETCH"    32%1           \n\t"
-            "movd          %1, %%mm0    \n\t"
-            "movd         3%1, %%mm3    \n\t"
-            "punpckldq    6%1, %%mm0    \n\t"
-            "punpckldq    9%1, %%mm3    \n\t"
+            PREFETCH"  32(%1)           \n\t"
+            "movd        (%1), %%mm0    \n\t"
+            "movd       3(%1), %%mm3    \n\t"
+            "punpckldq  6(%1), %%mm0    \n\t"
+            "punpckldq  9(%1), %%mm3    \n\t"
             "movq       %%mm0, %%mm1    \n\t"
             "movq       %%mm0, %%mm2    \n\t"
             "movq       %%mm3, %%mm4    \n\t"
@@ -729,8 +641,8 @@ static inline void RENAME(rgb24tobgr15)(const uint8_t *src, uint8_t *dst, int sr
             "por        %%mm5, %%mm3    \n\t"
             "psllq        $16, %%mm3    \n\t"
             "por        %%mm3, %%mm0    \n\t"
-            MOVNTQ"     %%mm0, %0       \n\t"
-            :"=m"(*d):"m"(*s),"m"(blue_15mask):"memory");
+            MOVNTQ"     %%mm0, (%0)     \n\t"
+            ::"r"(d),"r"(s),"m"(blue_15mask):"memory");
         d += 4;
         s += 12;
     }
@@ -759,11 +671,11 @@ static inline void RENAME(rgb24to15)(const uint8_t *src, uint8_t *dst, int src_s
     mm_end = end - 15;
     while (s < mm_end) {
         __asm__ volatile(
-            PREFETCH"   32%1            \n\t"
-            "movd         %1, %%mm0     \n\t"
-            "movd        3%1, %%mm3     \n\t"
-            "punpckldq   6%1, %%mm0     \n\t"
-            "punpckldq   9%1, %%mm3     \n\t"
+            PREFETCH" 32(%1)            \n\t"
+            "movd       (%1), %%mm0     \n\t"
+            "movd      3(%1), %%mm3     \n\t"
+            "punpckldq 6(%1), %%mm0     \n\t"
+            "punpckldq 9(%1), %%mm3     \n\t"
             "movq      %%mm0, %%mm1     \n\t"
             "movq      %%mm0, %%mm2     \n\t"
             "movq      %%mm3, %%mm4     \n\t"
@@ -786,8 +698,8 @@ static inline void RENAME(rgb24to15)(const uint8_t *src, uint8_t *dst, int src_s
             "por       %%mm5, %%mm3     \n\t"
             "psllq       $16, %%mm3     \n\t"
             "por       %%mm3, %%mm0     \n\t"
-            MOVNTQ"    %%mm0, %0        \n\t"
-            :"=m"(*d):"m"(*s),"m"(blue_15mask):"memory");
+            MOVNTQ"    %%mm0, (%0)      \n\t"
+            ::"r"(d),"r"(s),"m"(blue_15mask):"memory");
         d += 4;
         s += 12;
     }
@@ -812,10 +724,10 @@ static inline void RENAME(rgb15tobgr24)(const uint8_t *src, uint8_t *dst, int sr
     mm_end = end - 7;
     while (s < mm_end) {
         __asm__ volatile(
-            PREFETCH"    32%1           \n\t"
-            "movq          %1, %%mm0    \n\t"
-            "movq          %1, %%mm1    \n\t"
-            "movq          %1, %%mm2    \n\t"
+            PREFETCH"  32(%1)           \n\t"
+            "movq        (%1), %%mm0    \n\t"
+            "movq        (%1), %%mm1    \n\t"
+            "movq        (%1), %%mm2    \n\t"
             "pand          %2, %%mm0    \n\t"
             "pand          %3, %%mm1    \n\t"
             "pand          %4, %%mm2    \n\t"
@@ -844,9 +756,9 @@ static inline void RENAME(rgb15tobgr24)(const uint8_t *src, uint8_t *dst, int sr
             "movq       %%mm0, %%mm6    \n\t"
             "movq       %%mm3, %%mm7    \n\t"
 
-            "movq         8%1, %%mm0    \n\t"
-            "movq         8%1, %%mm1    \n\t"
-            "movq         8%1, %%mm2    \n\t"
+            "movq       8(%1), %%mm0    \n\t"
+            "movq       8(%1), %%mm1    \n\t"
+            "movq       8(%1), %%mm2    \n\t"
             "pand          %2, %%mm0    \n\t"
             "pand          %3, %%mm1    \n\t"
             "pand          %4, %%mm2    \n\t"
@@ -873,7 +785,7 @@ static inline void RENAME(rgb15tobgr24)(const uint8_t *src, uint8_t *dst, int sr
             "por        %%mm5, %%mm3    \n\t"
 
             :"=m"(*d)
-            :"m"(*s),"m"(mask15b),"m"(mask15g),"m"(mask15r),"m"(mmx_null)
+            :"r"(s),"m"(mask15b),"m"(mask15g),"m"(mask15r), "m"(mmx_null)
             :"memory");
         /* borrowed 32 to 24 */
         __asm__ volatile(
@@ -889,8 +801,7 @@ static inline void RENAME(rgb15tobgr24)(const uint8_t *src, uint8_t *dst, int sr
 
             STORE_BGR24_MMX
 
-            :"=m"(*d)
-            :"m"(*s)
+            :: "r"(d), "m"(*s)
             :"memory");
         d += 24;
         s += 8;
@@ -917,10 +828,10 @@ static inline void RENAME(rgb16tobgr24)(const uint8_t *src, uint8_t *dst, int sr
     mm_end = end - 7;
     while (s < mm_end) {
         __asm__ volatile(
-            PREFETCH"    32%1           \n\t"
-            "movq          %1, %%mm0    \n\t"
-            "movq          %1, %%mm1    \n\t"
-            "movq          %1, %%mm2    \n\t"
+            PREFETCH"  32(%1)           \n\t"
+            "movq        (%1), %%mm0    \n\t"
+            "movq        (%1), %%mm1    \n\t"
+            "movq        (%1), %%mm2    \n\t"
             "pand          %2, %%mm0    \n\t"
             "pand          %3, %%mm1    \n\t"
             "pand          %4, %%mm2    \n\t"
@@ -950,9 +861,9 @@ static inline void RENAME(rgb16tobgr24)(const uint8_t *src, uint8_t *dst, int sr
             "movq       %%mm0, %%mm6    \n\t"
             "movq       %%mm3, %%mm7    \n\t"
 
-            "movq         8%1, %%mm0    \n\t"
-            "movq         8%1, %%mm1    \n\t"
-            "movq         8%1, %%mm2    \n\t"
+            "movq       8(%1), %%mm0    \n\t"
+            "movq       8(%1), %%mm1    \n\t"
+            "movq       8(%1), %%mm2    \n\t"
             "pand          %2, %%mm0    \n\t"
             "pand          %3, %%mm1    \n\t"
             "pand          %4, %%mm2    \n\t"
@@ -979,7 +890,7 @@ static inline void RENAME(rgb16tobgr24)(const uint8_t *src, uint8_t *dst, int sr
             "por        %%mm4, %%mm3    \n\t"
             "por        %%mm5, %%mm3    \n\t"
             :"=m"(*d)
-            :"m"(*s),"m"(mask16b),"m"(mask16g),"m"(mask16r),"m"(mmx_null)
+            :"r"(s),"m"(mask16b),"m"(mask16g),"m"(mask16r),"m"(mmx_null)
             :"memory");
         /* borrowed 32 to 24 */
         __asm__ volatile(
@@ -995,8 +906,7 @@ static inline void RENAME(rgb16tobgr24)(const uint8_t *src, uint8_t *dst, int sr
 
             STORE_BGR24_MMX
 
-            :"=m"(*d)
-            :"m"(*s)
+            :: "r"(d), "m"(*s)
             :"memory");
         d += 24;
         s += 8;
@@ -1028,8 +938,8 @@ static inline void RENAME(rgb16tobgr24)(const uint8_t *src, uint8_t *dst, int sr
     "movq       %%mm0, %%mm3    \n\t"                               \
     "punpcklwd  %%mm2, %%mm0    \n\t" /* FF R1 G1 B1 FF R0 G0 B0 */ \
     "punpckhwd  %%mm2, %%mm3    \n\t" /* FF R3 G3 B3 FF R2 G2 B2 */ \
-    MOVNTQ"     %%mm0,  %0      \n\t"                               \
-    MOVNTQ"     %%mm3, 8%0      \n\t"                               \
+    MOVNTQ"     %%mm0,  (%0)    \n\t"                               \
+    MOVNTQ"     %%mm3, 8(%0)    \n\t"                               \
 
 static inline void RENAME(rgb15to32)(const uint8_t *src, uint8_t *dst, int src_size)
 {
@@ -1044,10 +954,10 @@ static inline void RENAME(rgb15to32)(const uint8_t *src, uint8_t *dst, int src_s
     mm_end = end - 3;
     while (s < mm_end) {
         __asm__ volatile(
-            PREFETCH"    32%1           \n\t"
-            "movq          %1, %%mm0    \n\t"
-            "movq          %1, %%mm1    \n\t"
-            "movq          %1, %%mm2    \n\t"
+            PREFETCH"  32(%1)           \n\t"
+            "movq        (%1), %%mm0    \n\t"
+            "movq        (%1), %%mm1    \n\t"
+            "movq        (%1), %%mm2    \n\t"
             "pand          %2, %%mm0    \n\t"
             "pand          %3, %%mm1    \n\t"
             "pand          %4, %%mm2    \n\t"
@@ -1056,8 +966,7 @@ static inline void RENAME(rgb15to32)(const uint8_t *src, uint8_t *dst, int src_s
             "pmulhw        %5, %%mm1    \n\t"
             "pmulhw        "MANGLE(mul15_hi)", %%mm2    \n\t"
             PACK_RGB32
-            :"=m"(*d)
-            :"m"(*s),"m"(mask15b),"m"(mask15g),"m"(mask15r),"m"(mul15_mid)
+            ::"r"(d),"r"(s),"m"(mask15b),"m"(mask15g),"m"(mask15r) ,"m"(mul15_mid)
             :"memory");
         d += 16;
         s += 4;
@@ -1087,10 +996,10 @@ static inline void RENAME(rgb16to32)(const uint8_t *src, uint8_t *dst, int src_s
     mm_end = end - 3;
     while (s < mm_end) {
         __asm__ volatile(
-            PREFETCH"    32%1           \n\t"
-            "movq          %1, %%mm0    \n\t"
-            "movq          %1, %%mm1    \n\t"
-            "movq          %1, %%mm2    \n\t"
+            PREFETCH"  32(%1)           \n\t"
+            "movq        (%1), %%mm0    \n\t"
+            "movq        (%1), %%mm1    \n\t"
+            "movq        (%1), %%mm2    \n\t"
             "pand          %2, %%mm0    \n\t"
             "pand          %3, %%mm1    \n\t"
             "pand          %4, %%mm2    \n\t"
@@ -1100,8 +1009,7 @@ static inline void RENAME(rgb16to32)(const uint8_t *src, uint8_t *dst, int src_s
             "pmulhw        "MANGLE(mul16_mid)", %%mm1    \n\t"
             "pmulhw        "MANGLE(mul15_hi)", %%mm2    \n\t"
             PACK_RGB32
-            :"=m"(*d)
-            :"m"(*s),"m"(mask16b),"m"(mask16g),"m"(mask16r),"m"(mul15_mid)
+            ::"r"(d),"r"(s),"m"(mask16b),"m"(mask16g),"m"(mask16r),"m"(mul15_mid)
             :"memory");
         d += 16;
         s += 4;
@@ -2029,8 +1937,8 @@ static inline void RENAME(vu9_to_vu12)(const uint8_t *src1, const uint8_t *src2,
                                        int srcStride1, int srcStride2,
                                        int dstStride1, int dstStride2)
 {
-    x86_reg y;
-    int x,w,h;
+    x86_reg x, y;
+    int w,h;
     w=width/2; h=height/2;
     __asm__ volatile(
         PREFETCH" %0    \n\t"
@@ -2042,11 +1950,11 @@ static inline void RENAME(vu9_to_vu12)(const uint8_t *src1, const uint8_t *src2,
         x=0;
         for (;x<w-31;x+=32) {
             __asm__ volatile(
-                PREFETCH"   32%1        \n\t"
-                "movq         %1, %%mm0 \n\t"
-                "movq        8%1, %%mm2 \n\t"
-                "movq       16%1, %%mm4 \n\t"
-                "movq       24%1, %%mm6 \n\t"
+                PREFETCH"   32(%1,%2)        \n\t"
+                "movq         (%1,%2), %%mm0 \n\t"
+                "movq        8(%1,%2), %%mm2 \n\t"
+                "movq       16(%1,%2), %%mm4 \n\t"
+                "movq       24(%1,%2), %%mm6 \n\t"
                 "movq      %%mm0, %%mm1 \n\t"
                 "movq      %%mm2, %%mm3 \n\t"
                 "movq      %%mm4, %%mm5 \n\t"
@@ -2059,16 +1967,15 @@ static inline void RENAME(vu9_to_vu12)(const uint8_t *src1, const uint8_t *src2,
                 "punpckhbw %%mm5, %%mm5 \n\t"
                 "punpcklbw %%mm6, %%mm6 \n\t"
                 "punpckhbw %%mm7, %%mm7 \n\t"
-                MOVNTQ"    %%mm0,   %0  \n\t"
-                MOVNTQ"    %%mm1,  8%0  \n\t"
-                MOVNTQ"    %%mm2, 16%0  \n\t"
-                MOVNTQ"    %%mm3, 24%0  \n\t"
-                MOVNTQ"    %%mm4, 32%0  \n\t"
-                MOVNTQ"    %%mm5, 40%0  \n\t"
-                MOVNTQ"    %%mm6, 48%0  \n\t"
-                MOVNTQ"    %%mm7, 56%0"
-                :"=m"(d[2*x])
-                :"m"(s1[x])
+                MOVNTQ"    %%mm0,   (%0,%2,2)  \n\t"
+                MOVNTQ"    %%mm1,  8(%0,%2,2)  \n\t"
+                MOVNTQ"    %%mm2, 16(%0,%2,2)  \n\t"
+                MOVNTQ"    %%mm3, 24(%0,%2,2)  \n\t"
+                MOVNTQ"    %%mm4, 32(%0,%2,2)  \n\t"
+                MOVNTQ"    %%mm5, 40(%0,%2,2)  \n\t"
+                MOVNTQ"    %%mm6, 48(%0,%2,2)  \n\t"
+                MOVNTQ"    %%mm7, 56(%0,%2,2)"
+                :: "r"(d), "r"(s1), "r"(x)
                 :"memory");
         }
         for (;x<w;x++) d[2*x]=d[2*x+1]=s1[x];
@@ -2079,11 +1986,11 @@ static inline void RENAME(vu9_to_vu12)(const uint8_t *src1, const uint8_t *src2,
         x=0;
         for (;x<w-31;x+=32) {
             __asm__ volatile(
-                PREFETCH"   32%1        \n\t"
-                "movq         %1, %%mm0 \n\t"
-                "movq        8%1, %%mm2 \n\t"
-                "movq       16%1, %%mm4 \n\t"
-                "movq       24%1, %%mm6 \n\t"
+                PREFETCH"   32(%1,%2)        \n\t"
+                "movq         (%1,%2), %%mm0 \n\t"
+                "movq        8(%1,%2), %%mm2 \n\t"
+                "movq       16(%1,%2), %%mm4 \n\t"
+                "movq       24(%1,%2), %%mm6 \n\t"
                 "movq      %%mm0, %%mm1 \n\t"
                 "movq      %%mm2, %%mm3 \n\t"
                 "movq      %%mm4, %%mm5 \n\t"
@@ -2096,16 +2003,15 @@ static inline void RENAME(vu9_to_vu12)(const uint8_t *src1, const uint8_t *src2,
                 "punpckhbw %%mm5, %%mm5 \n\t"
                 "punpcklbw %%mm6, %%mm6 \n\t"
                 "punpckhbw %%mm7, %%mm7 \n\t"
-                MOVNTQ"    %%mm0,   %0  \n\t"
-                MOVNTQ"    %%mm1,  8%0  \n\t"
-                MOVNTQ"    %%mm2, 16%0  \n\t"
-                MOVNTQ"    %%mm3, 24%0  \n\t"
-                MOVNTQ"    %%mm4, 32%0  \n\t"
-                MOVNTQ"    %%mm5, 40%0  \n\t"
-                MOVNTQ"    %%mm6, 48%0  \n\t"
-                MOVNTQ"    %%mm7, 56%0"
-                :"=m"(d[2*x])
-                :"m"(s2[x])
+                MOVNTQ"    %%mm0,   (%0,%2,2)  \n\t"
+                MOVNTQ"    %%mm1,  8(%0,%2,2)  \n\t"
+                MOVNTQ"    %%mm2, 16(%0,%2,2)  \n\t"
+                MOVNTQ"    %%mm3, 24(%0,%2,2)  \n\t"
+                MOVNTQ"    %%mm4, 32(%0,%2,2)  \n\t"
+                MOVNTQ"    %%mm5, 40(%0,%2,2)  \n\t"
+                MOVNTQ"    %%mm6, 48(%0,%2,2)  \n\t"
+                MOVNTQ"    %%mm7, 56(%0,%2,2)"
+                :: "r"(d), "r"(s2), "r"(x)
                 :"memory");
         }
         for (;x<w;x++) d[2*x]=d[2*x+1]=s2[x];