1 files changed, 107 insertions, 110 deletions

diff --git a/core/multimedia/opieplayer/libmpeg3/video/output.c b/core/multimedia/opieplayer/libmpeg3/video/output.c
index 919a0ff..bf0d6ed 100644
--- a/core/multimedia/opieplayer/libmpeg3/video/output.c
+++ b/core/multimedia/opieplayer/libmpeg3/video/output.c
@@ -93,422 +93,419 @@ inline void mpeg3video_rgb16_mmx(unsigned char *lum,
             "pand           mpeg3_MMX_redmask,      %%mm4\n"
             "psllw          $3,                     %%mm5\n"  /* GREEN       1 */
             "punpcklbw      %%mm6,                  %%mm6\n"
             "pand           mpeg3_MMX_grnmask,      %%mm5\n"
             "pand           mpeg3_MMX_redmask,      %%mm6\n"
             "por            %%mm5,                  %%mm4\n"  /* */
                     "psrlw          $11,                    %%mm6\n"  /* BLUE    1 */
             "movq           %%mm3,                  %%mm5\n"  /* lum2 */
                     "paddw          %%mm1,                  %%mm3\n"  /* lum2 +red 00 R7 00 R5 00 R3 00 R1 */
             "paddw          %%mm2,                  %%mm5\n"  /* lum2 +green 00 G7 00 G5 00 G3 00 G1 */
             "psraw          $6,                     %%mm3\n"  /* R2 */
             "por            %%mm6,                  %%mm4\n"  /* MM4 */
             "psraw          $6,                     %%mm5\n"  /* G2 */
             "movq           (%2, %3),               %%mm6\n"  /* L3 */
             "psraw          $6,                     %%mm7\n"
             "packuswb       %%mm3,                  %%mm3\n"
             "packuswb       %%mm5,                  %%mm5\n"
             "packuswb       %%mm7,                  %%mm7\n"
             "pand           mpeg3_MMX_00FFw,        %%mm6\n"  /* L3 */
             "punpcklbw      %%mm3,                  %%mm3\n"
             "punpcklbw      %%mm5,                  %%mm5\n"
             "pmullw         mpeg3_MMX_Ycoeff,       %%mm6\n"  /* lum3 */
             "punpcklbw      %%mm7,                  %%mm7\n"
             "psllw          $3,                     %%mm5\n"  /* GREEN 2 */
             "pand           mpeg3_MMX_redmask,      %%mm7\n"
             "pand           mpeg3_MMX_redmask,      %%mm3\n"
             "psrlw          $11,                    %%mm7\n"  /* BLUE  2 */
             "pand           mpeg3_MMX_grnmask,      %%mm5\n"
                     "por            %%mm7,                      %%mm3\n"
                     "movq           (%2,%3),                    %%mm7\n"  /* L4 */
                     "por            %%mm5,                              %%mm3\n" /* */
             "psrlw          $8,                     %%mm7\n"    /* L4 */
                     "movq           %%mm4,                      %%mm5\n"
             "punpcklwd      %%mm3,                  %%mm4\n"
             "pmullw         mpeg3_MMX_Ycoeff,       %%mm7\n"    /* lum4 */
             "punpckhwd      %%mm3,                  %%mm5\n"
 
                     "movq           %%mm4,                      (%4)\n"
                     "movq           %%mm5,                      8(%4)\n"
 
                     "movq           %%mm6,                              %%mm4\n"        /* Lum3 */
             "paddw          %%mm0,                  %%mm6\n"                /* Lum3 +blue */
 
                     "movq           %%mm4,                              %%mm5\n"                                        /* Lum3 */
             "paddw          %%mm1,                  %%mm4\n"       /* Lum3 +red */
             "paddw          %%mm2,                  %%mm5\n"                        /* Lum3 +green */
                     "psraw          $6,                             %%mm4\n"
                     "movq           %%mm7,                          %%mm3\n"/* Lum4 */
                     "psraw          $6,                             %%mm5\n"
             "paddw          %%mm0,                  %%mm7\n"                   /* Lum4 +blue */
             "psraw          $6,                     %%mm6\n"                        /* Lum3 +blue */
             "movq           %%mm3,                  %%mm0\n"  /* Lum4 */
             "packuswb       %%mm4,                  %%mm4\n"
             "paddw          %%mm1,                  %%mm3\n"  /* Lum4 +red */
             "packuswb       %%mm5,                  %%mm5\n"
             "paddw          %%mm2,                  %%mm0\n"         /* Lum4 +green */
             "packuswb       %%mm6,                  %%mm6\n"
             "punpcklbw      %%mm4,                  %%mm4\n"
             "punpcklbw      %%mm5,                  %%mm5\n"
             "punpcklbw      %%mm6,                  %%mm6\n"
             "psllw          $3,                     %%mm5\n" /* GREEN 3 */
             "pand           mpeg3_MMX_redmask,      %%mm4\n"
                     "psraw          $6,                         %%mm3\n" /* psr 6 */
                     "psraw          $6,                         %%mm0\n"
             "pand           mpeg3_MMX_redmask,      %%mm6\n" /* BLUE */
             "pand           mpeg3_MMX_grnmask,      %%mm5\n"
             "psrlw          $11,                    %%mm6\n"  /* BLUE  3 */
                     "por            %%mm5,                      %%mm4\n"
             "psraw          $6,                     %%mm7\n"
                     "por            %%mm6,                      %%mm4\n"
             "packuswb       %%mm3,                  %%mm3\n"
             "packuswb       %%mm0,                  %%mm0\n"
             "packuswb       %%mm7,                  %%mm7\n"
             "punpcklbw      %%mm3,                  %%mm3\n"
             "punpcklbw      %%mm0,                  %%mm0\n"
             "punpcklbw      %%mm7,                  %%mm7\n"
             "pand           mpeg3_MMX_redmask,      %%mm3\n"
             "pand           mpeg3_MMX_redmask,      %%mm7\n" /* BLUE */
             "psllw          $3,                     %%mm0\n" /* GREEN 4 */
             "psrlw          $11,                    %%mm7\n"
             "pand           mpeg3_MMX_grnmask,      %%mm0\n"
             "por            %%mm7,                  %%mm3\n"
             "addl           $8,                             %6\n"
             "por            %%mm0,                  %%mm3\n"
 
             "movq           %%mm4,                  %%mm5\n"
 
             "punpcklwd      %%mm3,                  %%mm4\n"
             "punpckhwd      %%mm3,                  %%mm5\n"
 
             "movq           %%mm4,                  (%4,%5,2)\n"
             "movq           %%mm5,                  8(%4,%5,2)\n"
 
             "addl           $8,                     %2\n"
             "addl           $4,                     %0\n"
             "addl           $4,                     %1\n"
             "cmpl           %3,                     %6\n"
             "leal           16(%4),                 %4\n"
         "jl             1b\n"
         "addl           %3,     %2\n"                   /* lum += cols */
         "addl           %7,     %4\n"                   /* row1 += mod */
         "movl           $0,     %6\n"
         "cmpl           %8,     %2\n"
         "jl             1b\n"
         : : "r" (cr), 
                         "r" (cb), 
                         "r" (lum), 
                         "r" (cols), 
                         "r" (row1) ,
                         "r" (col1), 
                         "m" (x), 
                         "m" (mod), 
                         "m" (y)
                 );
 }
 
 static unsigned LONGLONG  mpeg3_MMX_U_80 = 0x0000008000800000LL;
 static unsigned LONGLONG  mpeg3_MMX_V_80 = 0x0000000000800080LL;
 static LONGLONG  mpeg3_MMX_U_COEF        = 0x00000058ffd30000LL;
 static LONGLONG  mpeg3_MMX_V_COEF        = 0x00000000ffea006fLL;
 static LONGLONG  mpeg3_MMX_601_Y_COEF    = 0x0000004800480048LL;
 static LONGLONG  mpeg3_MMX_601_Y_DIFF    = 0x0000000000000010LL;
 
 inline void mpeg3_bgra32_mmx(unsigned long y, 
                 unsigned long u, 
                 unsigned long v, 
                 unsigned long *output)
 {
-asm("
+
+asm(    
 /* Output will be 0x00rrggbb with the 00 trailing so this can also be used */
 /* for bgr24. */
-        movd (%0), %%mm0;          /* Load y   0x00000000000000yy */
-        movd (%1), %%mm1;          /* Load u    0x00000000000000cr */
-        movq %%mm0, %%mm3;         /* Copy y to temp */
-        psllq $16, %%mm1;          /* Shift u   0x0000000000cr0000 */
-        movd (%2), %%mm2;          /* Load v    0x00000000000000cb */
-        psllq $16, %%mm3;          /* Shift y */
-        movq %%mm1, %%mm4;         /* Copy u to temp */
-        por %%mm3, %%mm0;          /* Overlay new y byte 0x0000000000yy00yy */
-        psllq $16, %%mm4;          /* Shift u */
-        movq %%mm2, %%mm5;         /* Copy v to temp */
-        psllq $16, %%mm3;          /* Shift y  */
-        por %%mm4, %%mm1;          /* Overlay new u byte 0x000000cr00cr0000 */
-        psllq $16, %%mm5;          /* Shift v  */
-        por %%mm3, %%mm0;          /* Overlay new y byte 0x000000yy00yy00yy */
-        por %%mm5, %%mm2;          /* Overlay new v byte 0x0000000000cb00cb */
+        "movd (%0), %%mm0;"          /* Load y   0x00000000000000yy */
+        "movd (%1), %%mm1;"          /* Load u    0x00000000000000cr */
+        "movq %%mm0, %%mm3;"         /* Copy y to temp */
+        "psllq $16, %%mm1;"          /* Shift u   0x0000000000cr0000 */
+        "movd (%2), %%mm2;"          /* Load v    0x00000000000000cb */
+        "psllq $16, %%mm3;"          /* Shift y */
+        "movq %%mm1, %%mm4;"         /* Copy u to temp */
+        "por %%mm3, %%mm0;"          /* Overlay new y byte 0x0000000000yy00yy */
+        "psllq $16, %%mm4;"          /* Shift u */
+        "movq %%mm2, %%mm5;"         /* Copy v to temp */
+        "psllq $16, %%mm3;"          /* Shift y  */
+        "por %%mm4, %%mm1;"          /* Overlay new u byte 0x000000cr00cr0000 */
+        "psllq $16, %%mm5;"          /* Shift v  */
+        "por %%mm3, %%mm0;"          /* Overlay new y byte 0x000000yy00yy00yy */
+        "por %%mm5, %%mm2;"          /* Overlay new v byte 0x0000000000cb00cb */
 
 /* mm0: 0x000000yy00yy00yy mm1: 0x000000uu00uu0000 mm2: 0x0000000000vv00vv */
-         psubw mpeg3_MMX_U_80, %%mm1;    /* Subtract 128 from u 0x000000uu00uu0000 */
-         pmullw mpeg3_MMX_U_COEF, %%mm1; /* Multiply u coeffs 0x0000uuuuuuuu0000 */
-         psllw $6, %%mm0;                /* Shift y coeffs 0x0000yyy0yyy0yyy0 */
-         psubw mpeg3_MMX_V_80, %%mm2;    /* Subtract 128 from v 0x0000000000cb00cb */
-         pmullw mpeg3_MMX_V_COEF, %%mm2; /* Multiply v coeffs 0x0000crcrcrcrcrcr */
+         "psubw mpeg3_MMX_U_80, %%mm1;"    /* Subtract 128 from u 0x000000uu00uu0000 */
+         "pmullw mpeg3_MMX_U_COEF, %%mm1;" /* Multiply u coeffs 0x0000uuuuuuuu0000 */
+         "psllw $6, %%mm0;"                /* Shift y coeffs 0x0000yyy0yyy0yyy0 */
+         "psubw mpeg3_MMX_V_80, %%mm2;"    /* Subtract 128 from v 0x0000000000cb00cb */
+         "pmullw mpeg3_MMX_V_COEF, %%mm2;" /* Multiply v coeffs 0x0000crcrcrcrcrcr */
 
 /* mm0: 0x000000yy00yy00yy mm1: 0x0000uuuuuuuu0000 mm2: 0x00000000vvvvvvvv */
-        paddsw %%mm1, %%mm0;        /* Add u to result */
-        paddsw %%mm2, %%mm0;        /* Add v to result 0x0000rrrrggggbbbb */
-        psraw $6, %%mm0;           /* Demote precision */
-        packuswb %%mm0, %%mm0;     /* Pack into ARGB 0x0000000000rrggbb */
-        movd %%mm0, (%3);          /* Store output */
-        "
+        "paddsw %%mm1, %%mm0;"        /* Add u to result */
+        "paddsw %%mm2, %%mm0;"        /* Add v to result 0x0000rrrrggggbbbb */
+        "psraw $6, %%mm0;"           /* Demote precision */
+        "packuswb %%mm0, %%mm0;"     /* Pack into ARGB 0x0000000000rrggbb */
+        "movd %%mm0, (%3);"          /* Store output */
 :
 : "r" (&y), "r" (&u), "r" (&v), "r" (output));
 }
 
 inline void mpeg3_601_bgra32_mmx(unsigned long y, 
                 unsigned long u, 
                 unsigned long v, 
                 unsigned long *output)
 {
-asm("
+asm(
 /* Output will be 0x00rrggbb with the 00 trailing so this can also be used */
 /* for bgr24. */
-        movd (%0), %%mm0;          /* Load y   0x00000000000000yy */
-        psubsw mpeg3_MMX_601_Y_DIFF, %%mm0;      /* Subtract 16 from y */
-        movd (%1), %%mm1;          /* Load u    0x00000000000000cr */
-        movq %%mm0, %%mm3;         /* Copy y to temp */
-        psllq $16, %%mm1;          /* Shift u   0x0000000000cr0000 */
-        movd (%2), %%mm2;          /* Load v    0x00000000000000cb */
-        psllq $16, %%mm3;          /* Shift y */
-        movq %%mm1, %%mm4;         /* Copy u to temp */
-        por %%mm3, %%mm0;          /* Overlay new y byte 0x0000000000yy00yy */
-        psllq $16, %%mm4;          /* Shift u */
-        movq %%mm2, %%mm5;         /* Copy v to temp */
-        psllq $16, %%mm3;          /* Shift y  */
-        por %%mm4, %%mm1;          /* Overlay new u byte 0x000000cr00cr0000 */
-        psllq $16, %%mm5;          /* Shift v  */
-        por %%mm3, %%mm0;          /* Overlay new y byte 0x000000yy00yy00yy */
-        por %%mm5, %%mm2;          /* Overlay new v byte 0x0000000000cb00cb */
+        "movd (%0), %%mm0;"          /* Load y   0x00000000000000yy */
+        "psubsw mpeg3_MMX_601_Y_DIFF, %%mm0;"      /* Subtract 16 from y */
+        "movd (%1), %%mm1;"          /* Load u    0x00000000000000cr */
+        "movq %%mm0, %%mm3;"         /* Copy y to temp */
+        "psllq $16, %%mm1;"          /* Shift u   0x0000000000cr0000 */
+        "movd (%2), %%mm2;"          /* Load v    0x00000000000000cb */
+        "psllq $16, %%mm3;"          /* Shift y */
+        "movq %%mm1, %%mm4;"         /* Copy u to temp */
+        "por %%mm3, %%mm0;"          /* Overlay new y byte 0x0000000000yy00yy */
+        "psllq $16, %%mm4;"          /* Shift u */
+        "movq %%mm2, %%mm5;"         /* Copy v to temp */
+        "psllq $16, %%mm3;"          /* Shift y  */
+        "por %%mm4, %%mm1;"          /* Overlay new u byte 0x000000cr00cr0000 */
+        "psllq $16, %%mm5;"          /* Shift v  */
+        "por %%mm3, %%mm0;"          /* Overlay new y byte 0x000000yy00yy00yy */
+        "por %%mm5, %%mm2;"          /* Overlay new v byte 0x0000000000cb00cb */
 
 /* mm0: 0x000000yy00yy00yy mm1: 0x000000uu00uu0000 mm2: 0x0000000000vv00vv */
-        pmullw mpeg3_MMX_601_Y_COEF, %%mm0; /* Scale and shift y coeffs */
-        psubw mpeg3_MMX_U_80, %%mm1;     /* Subtract 128 from u 0x000000uu00uu0000 */
-         pmullw mpeg3_MMX_U_COEF, %%mm1;  /* Multiply u coeffs 0x0000uuuuuuuu0000 */
-        psubw mpeg3_MMX_V_80, %%mm2;     /* Subtract 128 from v 0x0000000000cb00cb */
-         pmullw mpeg3_MMX_V_COEF, %%mm2;  /* Multiply v coeffs 0x0000crcrcrcrcrcr */
+        "pmullw mpeg3_MMX_601_Y_COEF, %%mm0;" /* Scale and shift y coeffs */
+        "psubw mpeg3_MMX_U_80, %%mm1;"     /* Subtract 128 from u 0x000000uu00uu0000 */
+         "pmullw mpeg3_MMX_U_COEF, %%mm1;"  /* Multiply u coeffs 0x0000uuuuuuuu0000 */
+        "psubw mpeg3_MMX_V_80, %%mm2;"     /* Subtract 128 from v 0x0000000000cb00cb */
+         "pmullw mpeg3_MMX_V_COEF, %%mm2;"  /* Multiply v coeffs 0x0000crcrcrcrcrcr */
 
 /* mm0: 0x000000yy00yy00yy mm1: 0x0000uuuuuuuu0000 mm2: 0x00000000vvvvvvvv */
-        paddsw %%mm1, %%mm0;        /* Add u to result */
-        paddsw %%mm2, %%mm0;        /* Add v to result 0x0000rrrrggggbbbb */
-        psraw $6, %%mm0;           /* Demote precision */
-        packuswb %%mm0, %%mm0;     /* Pack into ARGB 0x0000000000rrggbb */
-        movd %%mm0, (%3);          /* Store output */
-        "
+        "paddsw %%mm1, %%mm0;"        /* Add u to result */
+        "paddsw %%mm2, %%mm0;"        /* Add v to result 0x0000rrrrggggbbbb */
+        "psraw $6, %%mm0;"           /* Demote precision */
+        "packuswb %%mm0, %%mm0;"     /* Pack into ARGB 0x0000000000rrggbb */
+        "movd %%mm0, (%3);"          /* Store output */
 :
 : "r" (&y), "r" (&u), "r" (&v), "r" (output));
 }
 
 static unsigned LONGLONG  mpeg3_MMX_U_80_RGB    = 0x0000000000800080LL;
 static unsigned LONGLONG  mpeg3_MMX_V_80_RGB    = 0x0000008000800000LL;
 static LONGLONG  mpeg3_MMX_U_COEF_RGB    = 0x00000000ffd30058LL;
 static LONGLONG  mpeg3_MMX_V_COEF_RGB    = 0x0000006fffea0000LL;
 
 inline void mpeg3_rgba32_mmx(unsigned long y, 
                 unsigned long u, 
                 unsigned long v, 
                 unsigned long *output)
 {
-asm("
+asm(
 /* Output will be 0x00bbggrr with the 00 trailing so this can also be used */
 /* for rgb24. */
-        movd (%0), %%mm0;          /* Load y   0x00000000000000yy */
-        movd (%1), %%mm1;          /* Load v    0x00000000000000vv */
-        movq %%mm0, %%mm3;         /* Copy y to temp */
-        psllq $16, %%mm1;          /* Shift v   0x0000000000vv0000 */
-        movd (%2), %%mm2;          /* Load u    0x00000000000000uu */
-        psllq $16, %%mm3;          /* Shift y */
-        movq %%mm1, %%mm4;         /* Copy v to temp */
-        por %%mm3, %%mm0;          /* Overlay new y byte 0x0000000000yy00yy */
-        psllq $16, %%mm4;          /* Shift v */
-        movq %%mm2, %%mm5;         /* Copy u to temp */
-        psllq $16, %%mm3;          /* Shift y  */
-        por %%mm4, %%mm1;          /* Overlay new v byte 0x000000vv00vv0000 */
-        psllq $16, %%mm5;          /* Shift u  */
-        por %%mm3, %%mm0;          /* Overlay new y byte 0x000000yy00yy00yy */
-        por %%mm5, %%mm2;          /* Overlay new u byte 0x0000000000uu00uu */
+        "movd (%0), %%mm0;"          /* Load y   0x00000000000000yy */
+        "movd (%1), %%mm1;"          /* Load v    0x00000000000000vv */
+        "movq %%mm0, %%mm3;"         /* Copy y to temp */
+        "psllq $16, %%mm1;"          /* Shift v   0x0000000000vv0000 */
+        "movd (%2), %%mm2;"          /* Load u    0x00000000000000uu */
+        "psllq $16, %%mm3;"          /* Shift y */
+        "movq %%mm1, %%mm4;"         /* Copy v to temp */
+        "por %%mm3, %%mm0;"          /* Overlay new y byte 0x0000000000yy00yy */
+        "psllq $16, %%mm4;"          /* Shift v */
+        "movq %%mm2, %%mm5;"         /* Copy u to temp */
+        "psllq $16, %%mm3;"          /* Shift y  */
+        "por %%mm4, %%mm1;"          /* Overlay new v byte 0x000000vv00vv0000 */
+        "psllq $16, %%mm5;"          /* Shift u  */
+        "por %%mm3, %%mm0;"          /* Overlay new y byte 0x000000yy00yy00yy */
+        "por %%mm5, %%mm2;"          /* Overlay new u byte 0x0000000000uu00uu */
 
 /* mm0: 0x000000yy00yy00yy mm1: 0x000000vv00vv0000 mm2: 0x0000000000uu00uu */
-         psubw mpeg3_MMX_V_80_RGB, %%mm1;    /* Subtract 128 from v 0x000000vv00vv0000 */
-         pmullw mpeg3_MMX_V_COEF_RGB, %%mm1; /* Multiply v coeffs 0x0000vvvvvvvv0000 */
-         psllw $6, %%mm0;                /* Shift y coeffs 0x0000yyy0yyy0yyy0 */
-         psubw mpeg3_MMX_U_80_RGB, %%mm2;    /* Subtract 128 from u 0x0000000000uu00uu */
-         pmullw mpeg3_MMX_U_COEF_RGB, %%mm2; /* Multiply u coeffs 0x0000uuuuuuuuuuuu */
+         "psubw mpeg3_MMX_V_80_RGB, %%mm1;"    /* Subtract 128 from v 0x000000vv00vv0000 */
+         "pmullw mpeg3_MMX_V_COEF_RGB, %%mm1;" /* Multiply v coeffs 0x0000vvvvvvvv0000 */
+         "psllw $6, %%mm0;"                /* Shift y coeffs 0x0000yyy0yyy0yyy0 */
+         "psubw mpeg3_MMX_U_80_RGB, %%mm2;"    /* Subtract 128 from u 0x0000000000uu00uu */
+         "pmullw mpeg3_MMX_U_COEF_RGB, %%mm2;" /* Multiply u coeffs 0x0000uuuuuuuuuuuu */
 
 /* mm0: 0x000000yy00yy00yy mm1: 0x0000vvvvvvvv0000 mm2: 0x00000000uuuuuuuu */
-        paddsw %%mm1, %%mm0;        /* Add v to result */
-        paddsw %%mm2, %%mm0;        /* Add u to result 0x0000bbbbggggrrrr */
-        psraw $6, %%mm0;           /* Demote precision */
-        packuswb %%mm0, %%mm0;     /* Pack into RGBA 0x0000000000bbggrr */
-        movd %%mm0, (%3);          /* Store output */
-        "
+        "paddsw %%mm1, %%mm0;"        /* Add v to result */
+        "paddsw %%mm2, %%mm0;"        /* Add u to result 0x0000bbbbggggrrrr */
+        "psraw $6, %%mm0;"           /* Demote precision */
+        "packuswb %%mm0, %%mm0;"     /* Pack into RGBA 0x0000000000bbggrr */
+        "movd %%mm0, (%3);"          /* Store output */
 :
 : "r" (&y), "r" (&v), "r" (&u), "r" (output));
 }
 
 inline void mpeg3_601_rgba32_mmx(unsigned long y, 
                 unsigned long u, 
                 unsigned long v, 
                 unsigned long *output)
 {
-asm("
+asm(
 /* Output will be 0x00bbggrr with the 00 trailing so this can also be used */
 /* for rgb24. */
-        movd (%0), %%mm0;          /* Load y   0x00000000000000yy */
-        psubsw mpeg3_MMX_601_Y_DIFF, %%mm0;      /* Subtract 16 from y */
-        movd (%1), %%mm1;          /* Load v    0x00000000000000vv */
-        movq %%mm0, %%mm3;         /* Copy y to temp */
-        psllq $16, %%mm1;          /* Shift v   0x0000000000vv0000 */
-        movd (%2), %%mm2;          /* Load u    0x00000000000000uu */
-        psllq $16, %%mm3;          /* Shift y */
-        movq %%mm1, %%mm4;         /* Copy v to temp */
-        por %%mm3, %%mm0;          /* Overlay new y byte 0x0000000000yy00yy */
-        psllq $16, %%mm4;          /* Shift v */
-        movq %%mm2, %%mm5;         /* Copy u to temp */
-        psllq $16, %%mm3;          /* Shift y  */
-        por %%mm4, %%mm1;          /* Overlay new v byte 0x000000vv00vv0000 */
-        psllq $16, %%mm5;          /* Shift u  */
-        por %%mm3, %%mm0;          /* Overlay new y byte 0x000000yy00yy00yy */
-        por %%mm5, %%mm2;          /* Overlay new u byte 0x0000000000uu00uu */
+        "movd (%0), %%mm0;"          /* Load y   0x00000000000000yy */
+        "psubsw mpeg3_MMX_601_Y_DIFF, %%mm0;"      /* Subtract 16 from y */
+        "movd (%1), %%mm1;"          /* Load v    0x00000000000000vv */
+        "movq %%mm0, %%mm3;"         /* Copy y to temp */
+        "psllq $16, %%mm1;"          /* Shift v   0x0000000000vv0000 */
+        "movd (%2), %%mm2;"          /* Load u    0x00000000000000uu */
+        "psllq $16, %%mm3;"          /* Shift y */
+        "movq %%mm1, %%mm4;"         /* Copy v to temp */
+        "por %%mm3, %%mm0;"          /* Overlay new y byte 0x0000000000yy00yy */
+        "psllq $16, %%mm4;"          /* Shift v */
+        "movq %%mm2, %%mm5;"         /* Copy u to temp */
+        "psllq $16, %%mm3;"          /* Shift y  */
+        "por %%mm4, %%mm1;"          /* Overlay new v byte 0x000000vv00vv0000 */
+        "psllq $16, %%mm5;"          /* Shift u  */
+        "por %%mm3, %%mm0;"          /* Overlay new y byte 0x000000yy00yy00yy */
+        "por %%mm5, %%mm2;"          /* Overlay new u byte 0x0000000000uu00uu */
 
 /* mm0: 0x000000yy00yy00yy     mm1: 0x000000vv00vv0000     mm2: 0x0000000000uu00uu */
-        pmullw mpeg3_MMX_601_Y_COEF, %%mm0;     /* Scale y coeffs */
-         psubw mpeg3_MMX_V_80_RGB, %%mm1;    /* Subtract 128 from v 0x000000vv00vv0000 */
-         pmullw mpeg3_MMX_V_COEF_RGB, %%mm1; /* Multiply v coeffs 0x0000vvvvvvvv0000 */
-         psubw mpeg3_MMX_U_80_RGB, %%mm2;    /* Subtract 128 from u 0x0000000000uu00uu */
-         pmullw mpeg3_MMX_U_COEF_RGB, %%mm2; /* Multiply u coeffs 0x0000uuuuuuuuuuuu */
+        "pmullw mpeg3_MMX_601_Y_COEF, %%mm0;"     /* Scale y coeffs */
+         "psubw mpeg3_MMX_V_80_RGB, %%mm1;"    /* Subtract 128 from v 0x000000vv00vv0000 */
+         "pmullw mpeg3_MMX_V_COEF_RGB, %%mm1;" /* Multiply v coeffs 0x0000vvvvvvvv0000 */
+         "psubw mpeg3_MMX_U_80_RGB, %%mm2;"    /* Subtract 128 from u 0x0000000000uu00uu */
+         "pmullw mpeg3_MMX_U_COEF_RGB, %%mm2;" /* Multiply u coeffs 0x0000uuuuuuuuuuuu */
 
 /* mm0: 0x000000yy00yy00yy mm1: 0x0000vvvvvvvv0000 mm2: 0x00000000uuuuuuuu */
-        paddsw %%mm1, %%mm0;        /* Add v to result */
-        paddsw %%mm2, %%mm0;        /* Add u to result 0x0000bbbbggggrrrr */
-        psraw $6, %%mm0;           /* Demote precision */
-        packuswb %%mm0, %%mm0;     /* Pack into RGBA 0x0000000000bbggrr */
-        movd %%mm0, (%3);          /* Store output */
-        "
+        "paddsw %%mm1, %%mm0;"        /* Add v to result */
+        "paddsw %%mm2, %%mm0;"        /* Add u to result 0x0000bbbbggggrrrr */
+        "psraw $6, %%mm0;"           /* Demote precision */
+        "packuswb %%mm0, %%mm0;"     /* Pack into RGBA 0x0000000000bbggrr */
+        "movd %%mm0, (%3);"          /* Store output */
 :
 : "r" (&y), "r" (&v), "r" (&u), "r" (output));
 }
 
 #endif
 
 #define DITHER_ROW_HEAD \
         for(h = 0; h < video->out_h; h++) \
         { \
                 y_in = &src[0][(video->y_table[h] + video->in_y) * video->coded_picture_width] + video->in_x; \
                 cb_in = &src[1][((video->y_table[h] + video->in_y) >> 1) * video->chrom_width] + (video->in_x >> 2); \
                 cr_in = &src[2][((video->y_table[h] + video->in_y) >> 1) * video->chrom_width] + (video->in_x >> 1); \
                 data = output_rows[h];
 
 #define DITHER_ROW_TAIL \
         }
 
 #define DITHER_SCALE_HEAD \
         for(w = 0; w < video->out_w; w++) \
         { \
                 uv_subscript = video->x_table[w] / 2; \
                 y_l = y_in[video->x_table[w]]; \
                 y_l <<= 16; \
                 r_l = (y_l + video->cr_to_r[cr_in[uv_subscript]]) >> 16; \
                 g_l = (y_l + video->cr_to_g[cr_in[uv_subscript]] + video->cb_to_g[cb_in[uv_subscript]]) >> 16; \
                 b_l = (y_l + video->cb_to_b[cb_in[uv_subscript]]) >> 16;
 
 #define DITHER_SCALE_601_HEAD \
         for(w = 0; w < video->out_w; w++) \
         { \
                 uv_subscript = video->x_table[w] / 2; \
                 y_l = mpeg3_601_to_rgb[y_in[video->x_table[w]]]; \
                 y_l <<= 16; \
                 r_l = (y_l + video->cr_to_r[cr_in[uv_subscript]]) >> 16; \
                 g_l = (y_l + video->cr_to_g[cr_in[uv_subscript]] + video->cb_to_g[cb_in[uv_subscript]]) >> 16; \
                 b_l = (y_l + video->cb_to_b[cb_in[uv_subscript]]) >> 16;
 
 #define DITHER_SCALE_TAIL \
         }
 
 #define DITHER_MMX_SCALE_HEAD \
         for(w = 0; w < video->out_w; w++) \
         { \
                 uv_subscript = video->x_table[w] / 2;
 
 #define DITHER_MMX_SCALE_TAIL \
                 data += step; \
         }
 
 #define DITHER_MMX_HEAD \
         for(w = 0; w < video->out_w; w += 2) \
         {
 
 #define DITHER_MMX_TAIL \
                 data += step; \
         cr_in++; \
         cb_in++; \
         }
 
 #define DITHER_HEAD \
     for(w = 0; w < video->horizontal_size; w++) \
         { \
                 y_l = *y_in++; \
                 y_l <<= 16; \
                 r_l = (y_l + video->cr_to_r[*cr_in]) >> 16; \
                 g_l = (y_l + video->cr_to_g[*cr_in] + video->cb_to_g[*cb_in]) >> 16; \
                 b_l = (y_l + video->cb_to_b[*cb_in]) >> 16;
 
 #define DITHER_601_HEAD \
     for(w = 0; w < video->horizontal_size; w++) \
         { \
                 y_l = mpeg3_601_to_rgb[*y_in++]; \
                 y_l <<= 16; \
                 r_l = (y_l + video->cr_to_r[*cr_in]) >> 16; \
                 g_l = (y_l + video->cr_to_g[*cr_in] + video->cb_to_g[*cb_in]) >> 16; \
                 b_l = (y_l + video->cb_to_b[*cb_in]) >> 16;
 
 #define DITHER_TAIL \
             if(w & 1) \
                 { \
                 cr_in++; \
                 cb_in++; \
             } \
     }
 
 
 #define STORE_PIXEL_BGR888 \
         *data++ = CLIP(b_l); \
         *data++ = CLIP(g_l); \
         *data++ = CLIP(r_l);
 
 #define STORE_PIXEL_BGRA8888 \
         *data++ = CLIP(b_l); \
         *data++ = CLIP(g_l); \
         *data++ = CLIP(r_l); \
         *data++ = 0;
 
 #define STORE_PIXEL_RGB565 \
         *((unsigned short*)data)++ = \
                 ((CLIP(r_l) & 0xf8) << 8) | \
                 ((CLIP(g_l) & 0xfc) << 3) | \
                 ((CLIP(b_l) & 0xf8) >> 3);
 
 #define STORE_PIXEL_RGB888 \
         *data++ = CLIP(r_l); \
         *data++ = CLIP(g_l); \
         *data++ = CLIP(b_l);
 
 #define STORE_PIXEL_RGBA8888 \
         *data++ = CLIP(r_l); \
         *data++ = CLIP(g_l); \
         *data++ = CLIP(b_l); \
         *data++ = 0;
 
 #define STORE_PIXEL_RGBA16161616 \
         *data_s++ = CLIP(r_l); \
         *data_s++ = CLIP(g_l); \
         *data_s++ = CLIP(b_l); \
         *data_s++ = 0;
 
 
 
 /* Only good for YUV 4:2:0 */
 int mpeg3video_ditherframe(mpeg3video_t *video, unsigned char **src, unsigned char **output_rows)
 {
         int h = 0;
         register unsigned char *y_in, *cb_in, *cr_in;
         long y_l, r_l, b_l, g_l;