git.sesse.net Git - ffmpeg/blob - libswscale/swscale_template.c

   1 /*
   2  * Copyright (C) 2001-2003 Michael Niedermayer <michaelni@gmx.at>
   3  *
   4  * This file is part of FFmpeg.
   5  *
   6  * FFmpeg is free software; you can redistribute it and/or modify
   7  * it under the terms of the GNU General Public License as published by
   8  * the Free Software Foundation; either version 2 of the License, or
   9  * (at your option) any later version.
  10  *
  11  * FFmpeg is distributed in the hope that it will be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14  * GNU General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU General Public License
  17  * along with FFmpeg; if not, write to the Free Software
  18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  19  *
  20  * The C code (not assembly, MMX, ...) of this file can be used
  21  * under the LGPL license.
  22  */
  23
  24 #undef REAL_MOVNTQ
  25 #undef MOVNTQ
  26 #undef PAVGB
  27 #undef PREFETCH
  28
  29 #if COMPILE_TEMPLATE_AMD3DNOW
  30 #define PREFETCH  "prefetch"
  31 #elif COMPILE_TEMPLATE_MMX2
  32 #define PREFETCH "prefetchnta"
  33 #else
  34 #define PREFETCH  " # nop"
  35 #endif
  36
  37 #if COMPILE_TEMPLATE_MMX2
  38 #define PAVGB(a,b) "pavgb " #a ", " #b " \n\t"
  39 #elif COMPILE_TEMPLATE_AMD3DNOW
  40 #define PAVGB(a,b) "pavgusb " #a ", " #b " \n\t"
  41 #endif
  42
  43 #if COMPILE_TEMPLATE_MMX2
  44 #define REAL_MOVNTQ(a,b) "movntq " #a ", " #b " \n\t"
  45 #else
  46 #define REAL_MOVNTQ(a,b) "movq " #a ", " #b " \n\t"
  47 #endif
  48 #define MOVNTQ(a,b)  REAL_MOVNTQ(a,b)
  49
  50 #if COMPILE_TEMPLATE_ALTIVEC
  51 #include "ppc/swscale_altivec_template.c"
  52 #endif
  53
  54 #define YSCALEYUV2YV12X(x, offset, dest, width) \
  55     __asm__ volatile(\
  56         "xor                          %%"REG_a", %%"REG_a"  \n\t"\
  57         "movq             "VROUNDER_OFFSET"(%0), %%mm3      \n\t"\
  58         "movq                             %%mm3, %%mm4      \n\t"\
  59         "lea                     " offset "(%0), %%"REG_d"  \n\t"\
  60         "mov                        (%%"REG_d"), %%"REG_S"  \n\t"\
  61         ASMALIGN(4) /* FIXME Unroll? */\
  62         "1:                                                 \n\t"\
  63         "movq                      8(%%"REG_d"), %%mm0      \n\t" /* filterCoeff */\
  64         "movq   "  x "(%%"REG_S", %%"REG_a", 2), %%mm2      \n\t" /* srcData */\
  65         "movq 8+"  x "(%%"REG_S", %%"REG_a", 2), %%mm5      \n\t" /* srcData */\
  66         "add                                $16, %%"REG_d"  \n\t"\
  67         "mov                        (%%"REG_d"), %%"REG_S"  \n\t"\
  68         "test                         %%"REG_S", %%"REG_S"  \n\t"\
  69         "pmulhw                           %%mm0, %%mm2      \n\t"\
  70         "pmulhw                           %%mm0, %%mm5      \n\t"\
  71         "paddw                            %%mm2, %%mm3      \n\t"\
  72         "paddw                            %%mm5, %%mm4      \n\t"\
  73         " jnz                                1b             \n\t"\
  74         "psraw                               $3, %%mm3      \n\t"\
  75         "psraw                               $3, %%mm4      \n\t"\
  76         "packuswb                         %%mm4, %%mm3      \n\t"\
  77         MOVNTQ(%%mm3, (%1, %%REGa))\
  78         "add                                 $8, %%"REG_a"  \n\t"\
  79         "cmp                                 %2, %%"REG_a"  \n\t"\
  80         "movq             "VROUNDER_OFFSET"(%0), %%mm3      \n\t"\
  81         "movq                             %%mm3, %%mm4      \n\t"\
  82         "lea                     " offset "(%0), %%"REG_d"  \n\t"\
  83         "mov                        (%%"REG_d"), %%"REG_S"  \n\t"\
  84         "jb                                  1b             \n\t"\
  85         :: "r" (&c->redDither),\
  86         "r" (dest), "g" (width)\
  87         : "%"REG_a, "%"REG_d, "%"REG_S\
  88     );
  89
  90 #define YSCALEYUV2YV12X_ACCURATE(x, offset, dest, width) \
  91     __asm__ volatile(\
  92         "lea                     " offset "(%0), %%"REG_d"  \n\t"\
  93         "xor                          %%"REG_a", %%"REG_a"  \n\t"\
  94         "pxor                             %%mm4, %%mm4      \n\t"\
  95         "pxor                             %%mm5, %%mm5      \n\t"\
  96         "pxor                             %%mm6, %%mm6      \n\t"\
  97         "pxor                             %%mm7, %%mm7      \n\t"\
  98         "mov                        (%%"REG_d"), %%"REG_S"  \n\t"\
  99         ASMALIGN(4) \
 100         "1:                                                 \n\t"\
 101         "movq   "  x "(%%"REG_S", %%"REG_a", 2), %%mm0      \n\t" /* srcData */\
 102         "movq 8+"  x "(%%"REG_S", %%"REG_a", 2), %%mm2      \n\t" /* srcData */\
 103         "mov        "STR(APCK_PTR2)"(%%"REG_d"), %%"REG_S"  \n\t"\
 104         "movq   "  x "(%%"REG_S", %%"REG_a", 2), %%mm1      \n\t" /* srcData */\
 105         "movq                             %%mm0, %%mm3      \n\t"\
 106         "punpcklwd                        %%mm1, %%mm0      \n\t"\
 107         "punpckhwd                        %%mm1, %%mm3      \n\t"\
 108         "movq       "STR(APCK_COEF)"(%%"REG_d"), %%mm1      \n\t" /* filterCoeff */\
 109         "pmaddwd                          %%mm1, %%mm0      \n\t"\
 110         "pmaddwd                          %%mm1, %%mm3      \n\t"\
 111         "paddd                            %%mm0, %%mm4      \n\t"\
 112         "paddd                            %%mm3, %%mm5      \n\t"\
 113         "movq 8+"  x "(%%"REG_S", %%"REG_a", 2), %%mm3      \n\t" /* srcData */\
 114         "mov        "STR(APCK_SIZE)"(%%"REG_d"), %%"REG_S"  \n\t"\
 115         "add                  $"STR(APCK_SIZE)", %%"REG_d"  \n\t"\
 116         "test                         %%"REG_S", %%"REG_S"  \n\t"\
 117         "movq                             %%mm2, %%mm0      \n\t"\
 118         "punpcklwd                        %%mm3, %%mm2      \n\t"\
 119         "punpckhwd                        %%mm3, %%mm0      \n\t"\
 120         "pmaddwd                          %%mm1, %%mm2      \n\t"\
 121         "pmaddwd                          %%mm1, %%mm0      \n\t"\
 122         "paddd                            %%mm2, %%mm6      \n\t"\
 123         "paddd                            %%mm0, %%mm7      \n\t"\
 124         " jnz                                1b             \n\t"\
 125         "psrad                              $16, %%mm4      \n\t"\
 126         "psrad                              $16, %%mm5      \n\t"\
 127         "psrad                              $16, %%mm6      \n\t"\
 128         "psrad                              $16, %%mm7      \n\t"\
 129         "movq             "VROUNDER_OFFSET"(%0), %%mm0      \n\t"\
 130         "packssdw                         %%mm5, %%mm4      \n\t"\
 131         "packssdw                         %%mm7, %%mm6      \n\t"\
 132         "paddw                            %%mm0, %%mm4      \n\t"\
 133         "paddw                            %%mm0, %%mm6      \n\t"\
 134         "psraw                               $3, %%mm4      \n\t"\
 135         "psraw                               $3, %%mm6      \n\t"\
 136         "packuswb                         %%mm6, %%mm4      \n\t"\
 137         MOVNTQ(%%mm4, (%1, %%REGa))\
 138         "add                                 $8, %%"REG_a"  \n\t"\
 139         "cmp                                 %2, %%"REG_a"  \n\t"\
 140         "lea                     " offset "(%0), %%"REG_d"  \n\t"\
 141         "pxor                             %%mm4, %%mm4      \n\t"\
 142         "pxor                             %%mm5, %%mm5      \n\t"\
 143         "pxor                             %%mm6, %%mm6      \n\t"\
 144         "pxor                             %%mm7, %%mm7      \n\t"\
 145         "mov                        (%%"REG_d"), %%"REG_S"  \n\t"\
 146         "jb                                  1b             \n\t"\
 147         :: "r" (&c->redDither),\
 148         "r" (dest), "g" (width)\
 149         : "%"REG_a, "%"REG_d, "%"REG_S\
 150     );
 151
 152 #define YSCALEYUV2YV121 \
 153     "mov %2, %%"REG_a"                    \n\t"\
 154     ASMALIGN(4) /* FIXME Unroll? */\
 155     "1:                                   \n\t"\
 156     "movq  (%0, %%"REG_a", 2), %%mm0      \n\t"\
 157     "movq 8(%0, %%"REG_a", 2), %%mm1      \n\t"\
 158     "psraw                 $7, %%mm0      \n\t"\
 159     "psraw                 $7, %%mm1      \n\t"\
 160     "packuswb           %%mm1, %%mm0      \n\t"\
 161     MOVNTQ(%%mm0, (%1, %%REGa))\
 162     "add                   $8, %%"REG_a"  \n\t"\
 163     "jnc                   1b             \n\t"
 164
 165 #define YSCALEYUV2YV121_ACCURATE \
 166     "mov %2, %%"REG_a"                    \n\t"\
 167     "pcmpeqw %%mm7, %%mm7                 \n\t"\
 168     "psrlw                 $15, %%mm7     \n\t"\
 169     "psllw                  $6, %%mm7     \n\t"\
 170     ASMALIGN(4) /* FIXME Unroll? */\
 171     "1:                                   \n\t"\
 172     "movq  (%0, %%"REG_a", 2), %%mm0      \n\t"\
 173     "movq 8(%0, %%"REG_a", 2), %%mm1      \n\t"\
 174     "paddsw             %%mm7, %%mm0      \n\t"\
 175     "paddsw             %%mm7, %%mm1      \n\t"\
 176     "psraw                 $7, %%mm0      \n\t"\
 177     "psraw                 $7, %%mm1      \n\t"\
 178     "packuswb           %%mm1, %%mm0      \n\t"\
 179     MOVNTQ(%%mm0, (%1, %%REGa))\
 180     "add                   $8, %%"REG_a"  \n\t"\
 181     "jnc                   1b             \n\t"
 182
 183 /*
 184     :: "m" (-lumFilterSize), "m" (-chrFilterSize),
 185        "m" (lumMmxFilter+lumFilterSize*4), "m" (chrMmxFilter+chrFilterSize*4),
 186        "r" (dest), "m" (dstW),
 187        "m" (lumSrc+lumFilterSize), "m" (chrSrc+chrFilterSize)
 188     : "%eax", "%ebx", "%ecx", "%edx", "%esi"
 189 */
 190 #define YSCALEYUV2PACKEDX_UV \
 191     __asm__ volatile(\
 192         "xor                   %%"REG_a", %%"REG_a"     \n\t"\
 193         ASMALIGN(4)\
 194         "nop                                            \n\t"\
 195         "1:                                             \n\t"\
 196         "lea "CHR_MMX_FILTER_OFFSET"(%0), %%"REG_d"     \n\t"\
 197         "mov                 (%%"REG_d"), %%"REG_S"     \n\t"\
 198         "movq      "VROUNDER_OFFSET"(%0), %%mm3         \n\t"\
 199         "movq                      %%mm3, %%mm4         \n\t"\
 200         ASMALIGN(4)\
 201         "2:                                             \n\t"\
 202         "movq               8(%%"REG_d"), %%mm0         \n\t" /* filterCoeff */\
 203         "movq     (%%"REG_S", %%"REG_a"), %%mm2         \n\t" /* UsrcData */\
 204         "movq "AV_STRINGIFY(VOF)"(%%"REG_S", %%"REG_a"), %%mm5         \n\t" /* VsrcData */\
 205         "add                         $16, %%"REG_d"     \n\t"\
 206         "mov                 (%%"REG_d"), %%"REG_S"     \n\t"\
 207         "pmulhw                    %%mm0, %%mm2         \n\t"\
 208         "pmulhw                    %%mm0, %%mm5         \n\t"\
 209         "paddw                     %%mm2, %%mm3         \n\t"\
 210         "paddw                     %%mm5, %%mm4         \n\t"\
 211         "test                  %%"REG_S", %%"REG_S"     \n\t"\
 212         " jnz                         2b                \n\t"\
 213
 214 #define YSCALEYUV2PACKEDX_YA(offset,coeff,src1,src2,dst1,dst2) \
 215     "lea                "offset"(%0), %%"REG_d"     \n\t"\
 216     "mov                 (%%"REG_d"), %%"REG_S"     \n\t"\
 217     "movq      "VROUNDER_OFFSET"(%0), "#dst1"       \n\t"\
 218     "movq                    "#dst1", "#dst2"       \n\t"\
 219     ASMALIGN(4)\
 220     "2:                                             \n\t"\
 221     "movq               8(%%"REG_d"), "#coeff"      \n\t" /* filterCoeff */\
 222     "movq  (%%"REG_S", %%"REG_a", 2), "#src1"       \n\t" /* Y1srcData */\
 223     "movq 8(%%"REG_S", %%"REG_a", 2), "#src2"       \n\t" /* Y2srcData */\
 224     "add                         $16, %%"REG_d"            \n\t"\
 225     "mov                 (%%"REG_d"), %%"REG_S"     \n\t"\
 226     "pmulhw                 "#coeff", "#src1"       \n\t"\
 227     "pmulhw                 "#coeff", "#src2"       \n\t"\
 228     "paddw                   "#src1", "#dst1"       \n\t"\
 229     "paddw                   "#src2", "#dst2"       \n\t"\
 230     "test                  %%"REG_S", %%"REG_S"     \n\t"\
 231     " jnz                         2b                \n\t"\
 232
 233 #define YSCALEYUV2PACKEDX \
 234     YSCALEYUV2PACKEDX_UV \
 235     YSCALEYUV2PACKEDX_YA(LUM_MMX_FILTER_OFFSET,%%mm0,%%mm2,%%mm5,%%mm1,%%mm7) \
 236
 237 #define YSCALEYUV2PACKEDX_END                     \
 238         :: "r" (&c->redDither),                   \
 239             "m" (dummy), "m" (dummy), "m" (dummy),\
 240             "r" (dest), "m" (dstW)                \
 241         : "%"REG_a, "%"REG_d, "%"REG_S            \
 242     );
 243
 244 #define YSCALEYUV2PACKEDX_ACCURATE_UV \
 245     __asm__ volatile(\
 246         "xor %%"REG_a", %%"REG_a"                       \n\t"\
 247         ASMALIGN(4)\
 248         "nop                                            \n\t"\
 249         "1:                                             \n\t"\
 250         "lea "CHR_MMX_FILTER_OFFSET"(%0), %%"REG_d"     \n\t"\
 251         "mov                 (%%"REG_d"), %%"REG_S"     \n\t"\
 252         "pxor                      %%mm4, %%mm4         \n\t"\
 253         "pxor                      %%mm5, %%mm5         \n\t"\
 254         "pxor                      %%mm6, %%mm6         \n\t"\
 255         "pxor                      %%mm7, %%mm7         \n\t"\
 256         ASMALIGN(4)\
 257         "2:                                             \n\t"\
 258         "movq     (%%"REG_S", %%"REG_a"), %%mm0         \n\t" /* UsrcData */\
 259         "movq "AV_STRINGIFY(VOF)"(%%"REG_S", %%"REG_a"), %%mm2         \n\t" /* VsrcData */\
 260         "mov "STR(APCK_PTR2)"(%%"REG_d"), %%"REG_S"     \n\t"\
 261         "movq     (%%"REG_S", %%"REG_a"), %%mm1         \n\t" /* UsrcData */\
 262         "movq                      %%mm0, %%mm3         \n\t"\
 263         "punpcklwd                 %%mm1, %%mm0         \n\t"\
 264         "punpckhwd                 %%mm1, %%mm3         \n\t"\
 265         "movq "STR(APCK_COEF)"(%%"REG_d"),%%mm1         \n\t" /* filterCoeff */\
 266         "pmaddwd                   %%mm1, %%mm0         \n\t"\
 267         "pmaddwd                   %%mm1, %%mm3         \n\t"\
 268         "paddd                     %%mm0, %%mm4         \n\t"\
 269         "paddd                     %%mm3, %%mm5         \n\t"\
 270         "movq "AV_STRINGIFY(VOF)"(%%"REG_S", %%"REG_a"), %%mm3         \n\t" /* VsrcData */\
 271         "mov "STR(APCK_SIZE)"(%%"REG_d"), %%"REG_S"     \n\t"\
 272         "add           $"STR(APCK_SIZE)", %%"REG_d"     \n\t"\
 273         "test                  %%"REG_S", %%"REG_S"     \n\t"\
 274         "movq                      %%mm2, %%mm0         \n\t"\
 275         "punpcklwd                 %%mm3, %%mm2         \n\t"\
 276         "punpckhwd                 %%mm3, %%mm0         \n\t"\
 277         "pmaddwd                   %%mm1, %%mm2         \n\t"\
 278         "pmaddwd                   %%mm1, %%mm0         \n\t"\
 279         "paddd                     %%mm2, %%mm6         \n\t"\
 280         "paddd                     %%mm0, %%mm7         \n\t"\
 281         " jnz                         2b                \n\t"\
 282         "psrad                       $16, %%mm4         \n\t"\
 283         "psrad                       $16, %%mm5         \n\t"\
 284         "psrad                       $16, %%mm6         \n\t"\
 285         "psrad                       $16, %%mm7         \n\t"\
 286         "movq      "VROUNDER_OFFSET"(%0), %%mm0         \n\t"\
 287         "packssdw                  %%mm5, %%mm4         \n\t"\
 288         "packssdw                  %%mm7, %%mm6         \n\t"\
 289         "paddw                     %%mm0, %%mm4         \n\t"\
 290         "paddw                     %%mm0, %%mm6         \n\t"\
 291         "movq                      %%mm4, "U_TEMP"(%0)  \n\t"\
 292         "movq                      %%mm6, "V_TEMP"(%0)  \n\t"\
 293
 294 #define YSCALEYUV2PACKEDX_ACCURATE_YA(offset) \
 295     "lea                "offset"(%0), %%"REG_d"     \n\t"\
 296     "mov                 (%%"REG_d"), %%"REG_S"     \n\t"\
 297     "pxor                      %%mm1, %%mm1         \n\t"\
 298     "pxor                      %%mm5, %%mm5         \n\t"\
 299     "pxor                      %%mm7, %%mm7         \n\t"\
 300     "pxor                      %%mm6, %%mm6         \n\t"\
 301     ASMALIGN(4)\
 302     "2:                                             \n\t"\
 303     "movq  (%%"REG_S", %%"REG_a", 2), %%mm0         \n\t" /* Y1srcData */\
 304     "movq 8(%%"REG_S", %%"REG_a", 2), %%mm2         \n\t" /* Y2srcData */\
 305     "mov "STR(APCK_PTR2)"(%%"REG_d"), %%"REG_S"     \n\t"\
 306     "movq  (%%"REG_S", %%"REG_a", 2), %%mm4         \n\t" /* Y1srcData */\
 307     "movq                      %%mm0, %%mm3         \n\t"\
 308     "punpcklwd                 %%mm4, %%mm0         \n\t"\
 309     "punpckhwd                 %%mm4, %%mm3         \n\t"\
 310     "movq "STR(APCK_COEF)"(%%"REG_d"), %%mm4         \n\t" /* filterCoeff */\
 311     "pmaddwd                   %%mm4, %%mm0         \n\t"\
 312     "pmaddwd                   %%mm4, %%mm3         \n\t"\
 313     "paddd                     %%mm0, %%mm1         \n\t"\
 314     "paddd                     %%mm3, %%mm5         \n\t"\
 315     "movq 8(%%"REG_S", %%"REG_a", 2), %%mm3         \n\t" /* Y2srcData */\
 316     "mov "STR(APCK_SIZE)"(%%"REG_d"), %%"REG_S"     \n\t"\
 317     "add           $"STR(APCK_SIZE)", %%"REG_d"     \n\t"\
 318     "test                  %%"REG_S", %%"REG_S"     \n\t"\
 319     "movq                      %%mm2, %%mm0         \n\t"\
 320     "punpcklwd                 %%mm3, %%mm2         \n\t"\
 321     "punpckhwd                 %%mm3, %%mm0         \n\t"\
 322     "pmaddwd                   %%mm4, %%mm2         \n\t"\
 323     "pmaddwd                   %%mm4, %%mm0         \n\t"\
 324     "paddd                     %%mm2, %%mm7         \n\t"\
 325     "paddd                     %%mm0, %%mm6         \n\t"\
 326     " jnz                         2b                \n\t"\
 327     "psrad                       $16, %%mm1         \n\t"\
 328     "psrad                       $16, %%mm5         \n\t"\
 329     "psrad                       $16, %%mm7         \n\t"\
 330     "psrad                       $16, %%mm6         \n\t"\
 331     "movq      "VROUNDER_OFFSET"(%0), %%mm0         \n\t"\
 332     "packssdw                  %%mm5, %%mm1         \n\t"\
 333     "packssdw                  %%mm6, %%mm7         \n\t"\
 334     "paddw                     %%mm0, %%mm1         \n\t"\
 335     "paddw                     %%mm0, %%mm7         \n\t"\
 336     "movq               "U_TEMP"(%0), %%mm3         \n\t"\
 337     "movq               "V_TEMP"(%0), %%mm4         \n\t"\
 338
 339 #define YSCALEYUV2PACKEDX_ACCURATE \
 340     YSCALEYUV2PACKEDX_ACCURATE_UV \
 341     YSCALEYUV2PACKEDX_ACCURATE_YA(LUM_MMX_FILTER_OFFSET)
 342
 343 #define YSCALEYUV2RGBX \
 344     "psubw  "U_OFFSET"(%0), %%mm3       \n\t" /* (U-128)8*/\
 345     "psubw  "V_OFFSET"(%0), %%mm4       \n\t" /* (V-128)8*/\
 346     "movq            %%mm3, %%mm2       \n\t" /* (U-128)8*/\
 347     "movq            %%mm4, %%mm5       \n\t" /* (V-128)8*/\
 348     "pmulhw "UG_COEFF"(%0), %%mm3       \n\t"\
 349     "pmulhw "VG_COEFF"(%0), %%mm4       \n\t"\
 350     /* mm2=(U-128)8, mm3=ug, mm4=vg mm5=(V-128)8 */\
 351     "pmulhw "UB_COEFF"(%0), %%mm2       \n\t"\
 352     "pmulhw "VR_COEFF"(%0), %%mm5       \n\t"\
 353     "psubw  "Y_OFFSET"(%0), %%mm1       \n\t" /* 8(Y-16)*/\
 354     "psubw  "Y_OFFSET"(%0), %%mm7       \n\t" /* 8(Y-16)*/\
 355     "pmulhw  "Y_COEFF"(%0), %%mm1       \n\t"\
 356     "pmulhw  "Y_COEFF"(%0), %%mm7       \n\t"\
 357     /* mm1= Y1, mm2=ub, mm3=ug, mm4=vg mm5=vr, mm7=Y2 */\
 358     "paddw           %%mm3, %%mm4       \n\t"\
 359     "movq            %%mm2, %%mm0       \n\t"\
 360     "movq            %%mm5, %%mm6       \n\t"\
 361     "movq            %%mm4, %%mm3       \n\t"\
 362     "punpcklwd       %%mm2, %%mm2       \n\t"\
 363     "punpcklwd       %%mm5, %%mm5       \n\t"\
 364     "punpcklwd       %%mm4, %%mm4       \n\t"\
 365     "paddw           %%mm1, %%mm2       \n\t"\
 366     "paddw           %%mm1, %%mm5       \n\t"\
 367     "paddw           %%mm1, %%mm4       \n\t"\
 368     "punpckhwd       %%mm0, %%mm0       \n\t"\
 369     "punpckhwd       %%mm6, %%mm6       \n\t"\
 370     "punpckhwd       %%mm3, %%mm3       \n\t"\
 371     "paddw           %%mm7, %%mm0       \n\t"\
 372     "paddw           %%mm7, %%mm6       \n\t"\
 373     "paddw           %%mm7, %%mm3       \n\t"\
 374     /* mm0=B1, mm2=B2, mm3=G2, mm4=G1, mm5=R1, mm6=R2 */\
 375     "packuswb        %%mm0, %%mm2       \n\t"\
 376     "packuswb        %%mm6, %%mm5       \n\t"\
 377     "packuswb        %%mm3, %%mm4       \n\t"\
 378
 379 #define REAL_YSCALEYUV2PACKED(index, c) \
 380     "movq "CHR_MMX_FILTER_OFFSET"+8("#c"), %%mm0              \n\t"\
 381     "movq "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm1              \n\t"\
 382     "psraw                $3, %%mm0                           \n\t"\
 383     "psraw                $3, %%mm1                           \n\t"\
 384     "movq              %%mm0, "CHR_MMX_FILTER_OFFSET"+8("#c") \n\t"\
 385     "movq              %%mm1, "LUM_MMX_FILTER_OFFSET"+8("#c") \n\t"\
 386     "xor            "#index", "#index"                        \n\t"\
 387     ASMALIGN(4)\
 388     "1:                                 \n\t"\
 389     "movq     (%2, "#index"), %%mm2     \n\t" /* uvbuf0[eax]*/\
 390     "movq     (%3, "#index"), %%mm3     \n\t" /* uvbuf1[eax]*/\
 391     "movq "AV_STRINGIFY(VOF)"(%2, "#index"), %%mm5     \n\t" /* uvbuf0[eax+2048]*/\
 392     "movq "AV_STRINGIFY(VOF)"(%3, "#index"), %%mm4     \n\t" /* uvbuf1[eax+2048]*/\
 393     "psubw             %%mm3, %%mm2     \n\t" /* uvbuf0[eax] - uvbuf1[eax]*/\
 394     "psubw             %%mm4, %%mm5     \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048]*/\
 395     "movq "CHR_MMX_FILTER_OFFSET"+8("#c"), %%mm0    \n\t"\
 396     "pmulhw            %%mm0, %%mm2     \n\t" /* (uvbuf0[eax] - uvbuf1[eax])uvalpha1>>16*/\
 397     "pmulhw            %%mm0, %%mm5     \n\t" /* (uvbuf0[eax+2048] - uvbuf1[eax+2048])uvalpha1>>16*/\
 398     "psraw                $7, %%mm3     \n\t" /* uvbuf0[eax] - uvbuf1[eax] >>4*/\
 399     "psraw                $7, %%mm4     \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048] >>4*/\
 400     "paddw             %%mm2, %%mm3     \n\t" /* uvbuf0[eax]uvalpha1 - uvbuf1[eax](1-uvalpha1)*/\
 401     "paddw             %%mm5, %%mm4     \n\t" /* uvbuf0[eax+2048]uvalpha1 - uvbuf1[eax+2048](1-uvalpha1)*/\
 402     "movq  (%0, "#index", 2), %%mm0     \n\t" /*buf0[eax]*/\
 403     "movq  (%1, "#index", 2), %%mm1     \n\t" /*buf1[eax]*/\
 404     "movq 8(%0, "#index", 2), %%mm6     \n\t" /*buf0[eax]*/\
 405     "movq 8(%1, "#index", 2), %%mm7     \n\t" /*buf1[eax]*/\
 406     "psubw             %%mm1, %%mm0     \n\t" /* buf0[eax] - buf1[eax]*/\
 407     "psubw             %%mm7, %%mm6     \n\t" /* buf0[eax] - buf1[eax]*/\
 408     "pmulhw "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm0  \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\
 409     "pmulhw "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm6  \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\
 410     "psraw                $7, %%mm1     \n\t" /* buf0[eax] - buf1[eax] >>4*/\
 411     "psraw                $7, %%mm7     \n\t" /* buf0[eax] - buf1[eax] >>4*/\
 412     "paddw             %%mm0, %%mm1     \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\
 413     "paddw             %%mm6, %%mm7     \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\
 414
 415 #define YSCALEYUV2PACKED(index, c)  REAL_YSCALEYUV2PACKED(index, c)
 416
 417 #define REAL_YSCALEYUV2RGB_UV(index, c) \
 418     "xor            "#index", "#index"  \n\t"\
 419     ASMALIGN(4)\
 420     "1:                                 \n\t"\
 421     "movq     (%2, "#index"), %%mm2     \n\t" /* uvbuf0[eax]*/\
 422     "movq     (%3, "#index"), %%mm3     \n\t" /* uvbuf1[eax]*/\
 423     "movq "AV_STRINGIFY(VOF)"(%2, "#index"), %%mm5     \n\t" /* uvbuf0[eax+2048]*/\
 424     "movq "AV_STRINGIFY(VOF)"(%3, "#index"), %%mm4     \n\t" /* uvbuf1[eax+2048]*/\
 425     "psubw             %%mm3, %%mm2     \n\t" /* uvbuf0[eax] - uvbuf1[eax]*/\
 426     "psubw             %%mm4, %%mm5     \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048]*/\
 427     "movq "CHR_MMX_FILTER_OFFSET"+8("#c"), %%mm0    \n\t"\
 428     "pmulhw            %%mm0, %%mm2     \n\t" /* (uvbuf0[eax] - uvbuf1[eax])uvalpha1>>16*/\
 429     "pmulhw            %%mm0, %%mm5     \n\t" /* (uvbuf0[eax+2048] - uvbuf1[eax+2048])uvalpha1>>16*/\
 430     "psraw                $4, %%mm3     \n\t" /* uvbuf0[eax] - uvbuf1[eax] >>4*/\
 431     "psraw                $4, %%mm4     \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048] >>4*/\
 432     "paddw             %%mm2, %%mm3     \n\t" /* uvbuf0[eax]uvalpha1 - uvbuf1[eax](1-uvalpha1)*/\
 433     "paddw             %%mm5, %%mm4     \n\t" /* uvbuf0[eax+2048]uvalpha1 - uvbuf1[eax+2048](1-uvalpha1)*/\
 434     "psubw  "U_OFFSET"("#c"), %%mm3     \n\t" /* (U-128)8*/\
 435     "psubw  "V_OFFSET"("#c"), %%mm4     \n\t" /* (V-128)8*/\
 436     "movq              %%mm3, %%mm2     \n\t" /* (U-128)8*/\
 437     "movq              %%mm4, %%mm5     \n\t" /* (V-128)8*/\
 438     "pmulhw "UG_COEFF"("#c"), %%mm3     \n\t"\
 439     "pmulhw "VG_COEFF"("#c"), %%mm4     \n\t"\
 440     /* mm2=(U-128)8, mm3=ug, mm4=vg mm5=(V-128)8 */\
 441
 442 #define REAL_YSCALEYUV2RGB_YA(index, c, b1, b2) \
 443     "movq  ("#b1", "#index", 2), %%mm0     \n\t" /*buf0[eax]*/\
 444     "movq  ("#b2", "#index", 2), %%mm1     \n\t" /*buf1[eax]*/\
 445     "movq 8("#b1", "#index", 2), %%mm6     \n\t" /*buf0[eax]*/\
 446     "movq 8("#b2", "#index", 2), %%mm7     \n\t" /*buf1[eax]*/\
 447     "psubw             %%mm1, %%mm0     \n\t" /* buf0[eax] - buf1[eax]*/\
 448     "psubw             %%mm7, %%mm6     \n\t" /* buf0[eax] - buf1[eax]*/\
 449     "pmulhw "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm0  \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\
 450     "pmulhw "LUM_MMX_FILTER_OFFSET"+8("#c"), %%mm6  \n\t" /* (buf0[eax] - buf1[eax])yalpha1>>16*/\
 451     "psraw                $4, %%mm1     \n\t" /* buf0[eax] - buf1[eax] >>4*/\
 452     "psraw                $4, %%mm7     \n\t" /* buf0[eax] - buf1[eax] >>4*/\
 453     "paddw             %%mm0, %%mm1     \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\
 454     "paddw             %%mm6, %%mm7     \n\t" /* buf0[eax]yalpha1 + buf1[eax](1-yalpha1) >>16*/\
 455
 456 #define REAL_YSCALEYUV2RGB_COEFF(c) \
 457     "pmulhw "UB_COEFF"("#c"), %%mm2     \n\t"\
 458     "pmulhw "VR_COEFF"("#c"), %%mm5     \n\t"\
 459     "psubw  "Y_OFFSET"("#c"), %%mm1     \n\t" /* 8(Y-16)*/\
 460     "psubw  "Y_OFFSET"("#c"), %%mm7     \n\t" /* 8(Y-16)*/\
 461     "pmulhw  "Y_COEFF"("#c"), %%mm1     \n\t"\
 462     "pmulhw  "Y_COEFF"("#c"), %%mm7     \n\t"\
 463     /* mm1= Y1, mm2=ub, mm3=ug, mm4=vg mm5=vr, mm7=Y2 */\
 464     "paddw             %%mm3, %%mm4     \n\t"\
 465     "movq              %%mm2, %%mm0     \n\t"\
 466     "movq              %%mm5, %%mm6     \n\t"\
 467     "movq              %%mm4, %%mm3     \n\t"\
 468     "punpcklwd         %%mm2, %%mm2     \n\t"\
 469     "punpcklwd         %%mm5, %%mm5     \n\t"\
 470     "punpcklwd         %%mm4, %%mm4     \n\t"\
 471     "paddw             %%mm1, %%mm2     \n\t"\
 472     "paddw             %%mm1, %%mm5     \n\t"\
 473     "paddw             %%mm1, %%mm4     \n\t"\
 474     "punpckhwd         %%mm0, %%mm0     \n\t"\
 475     "punpckhwd         %%mm6, %%mm6     \n\t"\
 476     "punpckhwd         %%mm3, %%mm3     \n\t"\
 477     "paddw             %%mm7, %%mm0     \n\t"\
 478     "paddw             %%mm7, %%mm6     \n\t"\
 479     "paddw             %%mm7, %%mm3     \n\t"\
 480     /* mm0=B1, mm2=B2, mm3=G2, mm4=G1, mm5=R1, mm6=R2 */\
 481     "packuswb          %%mm0, %%mm2     \n\t"\
 482     "packuswb          %%mm6, %%mm5     \n\t"\
 483     "packuswb          %%mm3, %%mm4     \n\t"\
 484
 485 #define YSCALEYUV2RGB_YA(index, c, b1, b2) REAL_YSCALEYUV2RGB_YA(index, c, b1, b2)
 486
 487 #define YSCALEYUV2RGB(index, c) \
 488     REAL_YSCALEYUV2RGB_UV(index, c) \
 489     REAL_YSCALEYUV2RGB_YA(index, c, %0, %1) \
 490     REAL_YSCALEYUV2RGB_COEFF(c)
 491
 492 #define REAL_YSCALEYUV2PACKED1(index, c) \
 493     "xor            "#index", "#index"  \n\t"\
 494     ASMALIGN(4)\
 495     "1:                                 \n\t"\
 496     "movq     (%2, "#index"), %%mm3     \n\t" /* uvbuf0[eax]*/\
 497     "movq "AV_STRINGIFY(VOF)"(%2, "#index"), %%mm4     \n\t" /* uvbuf0[eax+2048]*/\
 498     "psraw                $7, %%mm3     \n\t" \
 499     "psraw                $7, %%mm4     \n\t" \
 500     "movq  (%0, "#index", 2), %%mm1     \n\t" /*buf0[eax]*/\
 501     "movq 8(%0, "#index", 2), %%mm7     \n\t" /*buf0[eax]*/\
 502     "psraw                $7, %%mm1     \n\t" \
 503     "psraw                $7, %%mm7     \n\t" \
 504
 505 #define YSCALEYUV2PACKED1(index, c)  REAL_YSCALEYUV2PACKED1(index, c)
 506
 507 #define REAL_YSCALEYUV2RGB1(index, c) \
 508     "xor            "#index", "#index"  \n\t"\
 509     ASMALIGN(4)\
 510     "1:                                 \n\t"\
 511     "movq     (%2, "#index"), %%mm3     \n\t" /* uvbuf0[eax]*/\
 512     "movq "AV_STRINGIFY(VOF)"(%2, "#index"), %%mm4     \n\t" /* uvbuf0[eax+2048]*/\
 513     "psraw                $4, %%mm3     \n\t" /* uvbuf0[eax] - uvbuf1[eax] >>4*/\
 514     "psraw                $4, %%mm4     \n\t" /* uvbuf0[eax+2048] - uvbuf1[eax+2048] >>4*/\
 515     "psubw  "U_OFFSET"("#c"), %%mm3     \n\t" /* (U-128)8*/\
 516     "psubw  "V_OFFSET"("#c"), %%mm4     \n\t" /* (V-128)8*/\
 517     "movq              %%mm3, %%mm2     \n\t" /* (U-128)8*/\
 518     "movq              %%mm4, %%mm5     \n\t" /* (V-128)8*/\
 519     "pmulhw "UG_COEFF"("#c"), %%mm3     \n\t"\
 520     "pmulhw "VG_COEFF"("#c"), %%mm4     \n\t"\
 521     /* mm2=(U-128)8, mm3=ug, mm4=vg mm5=(V-128)8 */\
 522     "movq  (%0, "#index", 2), %%mm1     \n\t" /*buf0[eax]*/\
 523     "movq 8(%0, "#index", 2), %%mm7     \n\t" /*buf0[eax]*/\
 524     "psraw                $4, %%mm1     \n\t" /* buf0[eax] - buf1[eax] >>4*/\
 525     "psraw                $4, %%mm7     \n\t" /* buf0[eax] - buf1[eax] >>4*/\
 526     "pmulhw "UB_COEFF"("#c"), %%mm2     \n\t"\
 527     "pmulhw "VR_COEFF"("#c"), %%mm5     \n\t"\
 528     "psubw  "Y_OFFSET"("#c"), %%mm1     \n\t" /* 8(Y-16)*/\
 529     "psubw  "Y_OFFSET"("#c"), %%mm7     \n\t" /* 8(Y-16)*/\
 530     "pmulhw  "Y_COEFF"("#c"), %%mm1     \n\t"\
 531     "pmulhw  "Y_COEFF"("#c"), %%mm7     \n\t"\
 532     /* mm1= Y1, mm2=ub, mm3=ug, mm4=vg mm5=vr, mm7=Y2 */\
 533     "paddw             %%mm3, %%mm4     \n\t"\
 534     "movq              %%mm2, %%mm0     \n\t"\
 535     "movq              %%mm5, %%mm6     \n\t"\
 536     "movq              %%mm4, %%mm3     \n\t"\
 537     "punpcklwd         %%mm2, %%mm2     \n\t"\
 538     "punpcklwd         %%mm5, %%mm5     \n\t"\
 539     "punpcklwd         %%mm4, %%mm4     \n\t"\
 540     "paddw             %%mm1, %%mm2     \n\t"\
 541     "paddw             %%mm1, %%mm5     \n\t"\
 542     "paddw             %%mm1, %%mm4     \n\t"\
 543     "punpckhwd         %%mm0, %%mm0     \n\t"\
 544     "punpckhwd         %%mm6, %%mm6     \n\t"\
 545     "punpckhwd         %%mm3, %%mm3     \n\t"\
 546     "paddw             %%mm7, %%mm0     \n\t"\
 547     "paddw             %%mm7, %%mm6     \n\t"\
 548     "paddw             %%mm7, %%mm3     \n\t"\
 549     /* mm0=B1, mm2=B2, mm3=G2, mm4=G1, mm5=R1, mm6=R2 */\
 550     "packuswb          %%mm0, %%mm2     \n\t"\
 551     "packuswb          %%mm6, %%mm5     \n\t"\
 552     "packuswb          %%mm3, %%mm4     \n\t"\
 553
 554 #define YSCALEYUV2RGB1(index, c)  REAL_YSCALEYUV2RGB1(index, c)
 555
 556 #define REAL_YSCALEYUV2PACKED1b(index, c) \
 557     "xor "#index", "#index"             \n\t"\
 558     ASMALIGN(4)\
 559     "1:                                 \n\t"\
 560     "movq     (%2, "#index"), %%mm2     \n\t" /* uvbuf0[eax]*/\
 561     "movq     (%3, "#index"), %%mm3     \n\t" /* uvbuf1[eax]*/\
 562     "movq "AV_STRINGIFY(VOF)"(%2, "#index"), %%mm5     \n\t" /* uvbuf0[eax+2048]*/\
 563     "movq "AV_STRINGIFY(VOF)"(%3, "#index"), %%mm4     \n\t" /* uvbuf1[eax+2048]*/\
 564     "paddw             %%mm2, %%mm3     \n\t" /* uvbuf0[eax] + uvbuf1[eax]*/\
 565     "paddw             %%mm5, %%mm4     \n\t" /* uvbuf0[eax+2048] + uvbuf1[eax+2048]*/\
 566     "psrlw                $8, %%mm3     \n\t" \
 567     "psrlw                $8, %%mm4     \n\t" \
 568     "movq  (%0, "#index", 2), %%mm1     \n\t" /*buf0[eax]*/\
 569     "movq 8(%0, "#index", 2), %%mm7     \n\t" /*buf0[eax]*/\
 570     "psraw                $7, %%mm1     \n\t" \
 571     "psraw                $7, %%mm7     \n\t"
 572 #define YSCALEYUV2PACKED1b(index, c)  REAL_YSCALEYUV2PACKED1b(index, c)
 573
 574 // do vertical chrominance interpolation
 575 #define REAL_YSCALEYUV2RGB1b(index, c) \
 576     "xor            "#index", "#index"  \n\t"\
 577     ASMALIGN(4)\
 578     "1:                                 \n\t"\
 579     "movq     (%2, "#index"), %%mm2     \n\t" /* uvbuf0[eax]*/\
 580     "movq     (%3, "#index"), %%mm3     \n\t" /* uvbuf1[eax]*/\
 581     "movq "AV_STRINGIFY(VOF)"(%2, "#index"), %%mm5     \n\t" /* uvbuf0[eax+2048]*/\
 582     "movq "AV_STRINGIFY(VOF)"(%3, "#index"), %%mm4     \n\t" /* uvbuf1[eax+2048]*/\
 583     "paddw             %%mm2, %%mm3     \n\t" /* uvbuf0[eax] + uvbuf1[eax]*/\
 584     "paddw             %%mm5, %%mm4     \n\t" /* uvbuf0[eax+2048] + uvbuf1[eax+2048]*/\
 585     "psrlw                $5, %%mm3     \n\t" /*FIXME might overflow*/\
 586     "psrlw                $5, %%mm4     \n\t" /*FIXME might overflow*/\
 587     "psubw  "U_OFFSET"("#c"), %%mm3     \n\t" /* (U-128)8*/\
 588     "psubw  "V_OFFSET"("#c"), %%mm4     \n\t" /* (V-128)8*/\
 589     "movq              %%mm3, %%mm2     \n\t" /* (U-128)8*/\
 590     "movq              %%mm4, %%mm5     \n\t" /* (V-128)8*/\
 591     "pmulhw "UG_COEFF"("#c"), %%mm3     \n\t"\
 592     "pmulhw "VG_COEFF"("#c"), %%mm4     \n\t"\
 593     /* mm2=(U-128)8, mm3=ug, mm4=vg mm5=(V-128)8 */\
 594     "movq  (%0, "#index", 2), %%mm1     \n\t" /*buf0[eax]*/\
 595     "movq 8(%0, "#index", 2), %%mm7     \n\t" /*buf0[eax]*/\
 596     "psraw                $4, %%mm1     \n\t" /* buf0[eax] - buf1[eax] >>4*/\
 597     "psraw                $4, %%mm7     \n\t" /* buf0[eax] - buf1[eax] >>4*/\
 598     "pmulhw "UB_COEFF"("#c"), %%mm2     \n\t"\
 599     "pmulhw "VR_COEFF"("#c"), %%mm5     \n\t"\
 600     "psubw  "Y_OFFSET"("#c"), %%mm1     \n\t" /* 8(Y-16)*/\
 601     "psubw  "Y_OFFSET"("#c"), %%mm7     \n\t" /* 8(Y-16)*/\
 602     "pmulhw  "Y_COEFF"("#c"), %%mm1     \n\t"\
 603     "pmulhw  "Y_COEFF"("#c"), %%mm7     \n\t"\
 604     /* mm1= Y1, mm2=ub, mm3=ug, mm4=vg mm5=vr, mm7=Y2 */\
 605     "paddw             %%mm3, %%mm4     \n\t"\
 606     "movq              %%mm2, %%mm0     \n\t"\
 607     "movq              %%mm5, %%mm6     \n\t"\
 608     "movq              %%mm4, %%mm3     \n\t"\
 609     "punpcklwd         %%mm2, %%mm2     \n\t"\
 610     "punpcklwd         %%mm5, %%mm5     \n\t"\
 611     "punpcklwd         %%mm4, %%mm4     \n\t"\
 612     "paddw             %%mm1, %%mm2     \n\t"\
 613     "paddw             %%mm1, %%mm5     \n\t"\
 614     "paddw             %%mm1, %%mm4     \n\t"\
 615     "punpckhwd         %%mm0, %%mm0     \n\t"\
 616     "punpckhwd         %%mm6, %%mm6     \n\t"\
 617     "punpckhwd         %%mm3, %%mm3     \n\t"\
 618     "paddw             %%mm7, %%mm0     \n\t"\
 619     "paddw             %%mm7, %%mm6     \n\t"\
 620     "paddw             %%mm7, %%mm3     \n\t"\
 621     /* mm0=B1, mm2=B2, mm3=G2, mm4=G1, mm5=R1, mm6=R2 */\
 622     "packuswb          %%mm0, %%mm2     \n\t"\
 623     "packuswb          %%mm6, %%mm5     \n\t"\
 624     "packuswb          %%mm3, %%mm4     \n\t"\
 625
 626 #define YSCALEYUV2RGB1b(index, c)  REAL_YSCALEYUV2RGB1b(index, c)
 627
 628 #define REAL_YSCALEYUV2RGB1_ALPHA(index) \
 629     "movq  (%1, "#index", 2), %%mm7     \n\t" /* abuf0[index  ]     */\
 630     "movq 8(%1, "#index", 2), %%mm1     \n\t" /* abuf0[index+4]     */\
 631     "psraw                $7, %%mm7     \n\t" /* abuf0[index  ] >>7 */\
 632     "psraw                $7, %%mm1     \n\t" /* abuf0[index+4] >>7 */\
 633     "packuswb          %%mm1, %%mm7     \n\t"
 634 #define YSCALEYUV2RGB1_ALPHA(index) REAL_YSCALEYUV2RGB1_ALPHA(index)
 635
 636 #define REAL_WRITEBGR32(dst, dstw, index, b, g, r, a, q0, q2, q3, t) \
 637     "movq       "#b", "#q2"     \n\t" /* B */\
 638     "movq       "#r", "#t"      \n\t" /* R */\
 639     "punpcklbw  "#g", "#b"      \n\t" /* GBGBGBGB 0 */\
 640     "punpcklbw  "#a", "#r"      \n\t" /* ARARARAR 0 */\
 641     "punpckhbw  "#g", "#q2"     \n\t" /* GBGBGBGB 2 */\
 642     "punpckhbw  "#a", "#t"      \n\t" /* ARARARAR 2 */\
 643     "movq       "#b", "#q0"     \n\t" /* GBGBGBGB 0 */\
 644     "movq      "#q2", "#q3"     \n\t" /* GBGBGBGB 2 */\
 645     "punpcklwd  "#r", "#q0"     \n\t" /* ARGBARGB 0 */\
 646     "punpckhwd  "#r", "#b"      \n\t" /* ARGBARGB 1 */\
 647     "punpcklwd  "#t", "#q2"     \n\t" /* ARGBARGB 2 */\
 648     "punpckhwd  "#t", "#q3"     \n\t" /* ARGBARGB 3 */\
 649 \
 650     MOVNTQ(   q0,   (dst, index, 4))\
 651     MOVNTQ(    b,  8(dst, index, 4))\
 652     MOVNTQ(   q2, 16(dst, index, 4))\
 653     MOVNTQ(   q3, 24(dst, index, 4))\
 654 \
 655     "add      $8, "#index"      \n\t"\
 656     "cmp "#dstw", "#index"      \n\t"\
 657     " jb      1b                \n\t"
 658 #define WRITEBGR32(dst, dstw, index, b, g, r, a, q0, q2, q3, t)  REAL_WRITEBGR32(dst, dstw, index, b, g, r, a, q0, q2, q3, t)
 659
 660 #define REAL_WRITERGB16(dst, dstw, index) \
 661     "pand "MANGLE(bF8)", %%mm2  \n\t" /* B */\
 662     "pand "MANGLE(bFC)", %%mm4  \n\t" /* G */\
 663     "pand "MANGLE(bF8)", %%mm5  \n\t" /* R */\
 664     "psrlq           $3, %%mm2  \n\t"\
 665 \
 666     "movq         %%mm2, %%mm1  \n\t"\
 667     "movq         %%mm4, %%mm3  \n\t"\
 668 \
 669     "punpcklbw    %%mm7, %%mm3  \n\t"\
 670     "punpcklbw    %%mm5, %%mm2  \n\t"\
 671     "punpckhbw    %%mm7, %%mm4  \n\t"\
 672     "punpckhbw    %%mm5, %%mm1  \n\t"\
 673 \
 674     "psllq           $3, %%mm3  \n\t"\
 675     "psllq           $3, %%mm4  \n\t"\
 676 \
 677     "por          %%mm3, %%mm2  \n\t"\
 678     "por          %%mm4, %%mm1  \n\t"\
 679 \
 680     MOVNTQ(%%mm2,  (dst, index, 2))\
 681     MOVNTQ(%%mm1, 8(dst, index, 2))\
 682 \
 683     "add             $8, "#index"   \n\t"\
 684     "cmp        "#dstw", "#index"   \n\t"\
 685     " jb             1b             \n\t"
 686 #define WRITERGB16(dst, dstw, index)  REAL_WRITERGB16(dst, dstw, index)
 687
 688 #define REAL_WRITERGB15(dst, dstw, index) \
 689     "pand "MANGLE(bF8)", %%mm2  \n\t" /* B */\
 690     "pand "MANGLE(bF8)", %%mm4  \n\t" /* G */\
 691     "pand "MANGLE(bF8)", %%mm5  \n\t" /* R */\
 692     "psrlq           $3, %%mm2  \n\t"\
 693     "psrlq           $1, %%mm5  \n\t"\
 694 \
 695     "movq         %%mm2, %%mm1  \n\t"\
 696     "movq         %%mm4, %%mm3  \n\t"\
 697 \
 698     "punpcklbw    %%mm7, %%mm3  \n\t"\
 699     "punpcklbw    %%mm5, %%mm2  \n\t"\
 700     "punpckhbw    %%mm7, %%mm4  \n\t"\
 701     "punpckhbw    %%mm5, %%mm1  \n\t"\
 702 \
 703     "psllq           $2, %%mm3  \n\t"\
 704     "psllq           $2, %%mm4  \n\t"\
 705 \
 706     "por          %%mm3, %%mm2  \n\t"\
 707     "por          %%mm4, %%mm1  \n\t"\
 708 \
 709     MOVNTQ(%%mm2,  (dst, index, 2))\
 710     MOVNTQ(%%mm1, 8(dst, index, 2))\
 711 \
 712     "add             $8, "#index"   \n\t"\
 713     "cmp        "#dstw", "#index"   \n\t"\
 714     " jb             1b             \n\t"
 715 #define WRITERGB15(dst, dstw, index)  REAL_WRITERGB15(dst, dstw, index)
 716
 717 #define WRITEBGR24OLD(dst, dstw, index) \
 718     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */\
 719     "movq      %%mm2, %%mm1             \n\t" /* B */\
 720     "movq      %%mm5, %%mm6             \n\t" /* R */\
 721     "punpcklbw %%mm4, %%mm2             \n\t" /* GBGBGBGB 0 */\
 722     "punpcklbw %%mm7, %%mm5             \n\t" /* 0R0R0R0R 0 */\
 723     "punpckhbw %%mm4, %%mm1             \n\t" /* GBGBGBGB 2 */\
 724     "punpckhbw %%mm7, %%mm6             \n\t" /* 0R0R0R0R 2 */\
 725     "movq      %%mm2, %%mm0             \n\t" /* GBGBGBGB 0 */\
 726     "movq      %%mm1, %%mm3             \n\t" /* GBGBGBGB 2 */\
 727     "punpcklwd %%mm5, %%mm0             \n\t" /* 0RGB0RGB 0 */\
 728     "punpckhwd %%mm5, %%mm2             \n\t" /* 0RGB0RGB 1 */\
 729     "punpcklwd %%mm6, %%mm1             \n\t" /* 0RGB0RGB 2 */\
 730     "punpckhwd %%mm6, %%mm3             \n\t" /* 0RGB0RGB 3 */\
 731 \
 732     "movq      %%mm0, %%mm4             \n\t" /* 0RGB0RGB 0 */\
 733     "psrlq        $8, %%mm0             \n\t" /* 00RGB0RG 0 */\
 734     "pand "MANGLE(bm00000111)", %%mm4   \n\t" /* 00000RGB 0 */\
 735     "pand "MANGLE(bm11111000)", %%mm0   \n\t" /* 00RGB000 0.5 */\
 736     "por       %%mm4, %%mm0             \n\t" /* 00RGBRGB 0 */\
 737     "movq      %%mm2, %%mm4             \n\t" /* 0RGB0RGB 1 */\
 738     "psllq       $48, %%mm2             \n\t" /* GB000000 1 */\
 739     "por       %%mm2, %%mm0             \n\t" /* GBRGBRGB 0 */\
 740 \
 741     "movq      %%mm4, %%mm2             \n\t" /* 0RGB0RGB 1 */\
 742     "psrld       $16, %%mm4             \n\t" /* 000R000R 1 */\
 743     "psrlq       $24, %%mm2             \n\t" /* 0000RGB0 1.5 */\
 744     "por       %%mm4, %%mm2             \n\t" /* 000RRGBR 1 */\
 745     "pand "MANGLE(bm00001111)", %%mm2   \n\t" /* 0000RGBR 1 */\
 746     "movq      %%mm1, %%mm4             \n\t" /* 0RGB0RGB 2 */\
 747     "psrlq        $8, %%mm1             \n\t" /* 00RGB0RG 2 */\
 748     "pand "MANGLE(bm00000111)", %%mm4   \n\t" /* 00000RGB 2 */\
 749     "pand "MANGLE(bm11111000)", %%mm1   \n\t" /* 00RGB000 2.5 */\
 750     "por       %%mm4, %%mm1             \n\t" /* 00RGBRGB 2 */\
 751     "movq      %%mm1, %%mm4             \n\t" /* 00RGBRGB 2 */\
 752     "psllq       $32, %%mm1             \n\t" /* BRGB0000 2 */\
 753     "por       %%mm1, %%mm2             \n\t" /* BRGBRGBR 1 */\
 754 \
 755     "psrlq       $32, %%mm4             \n\t" /* 000000RG 2.5 */\
 756     "movq      %%mm3, %%mm5             \n\t" /* 0RGB0RGB 3 */\
 757     "psrlq        $8, %%mm3             \n\t" /* 00RGB0RG 3 */\
 758     "pand "MANGLE(bm00000111)", %%mm5   \n\t" /* 00000RGB 3 */\
 759     "pand "MANGLE(bm11111000)", %%mm3   \n\t" /* 00RGB000 3.5 */\
 760     "por       %%mm5, %%mm3             \n\t" /* 00RGBRGB 3 */\
 761     "psllq       $16, %%mm3             \n\t" /* RGBRGB00 3 */\
 762     "por       %%mm4, %%mm3             \n\t" /* RGBRGBRG 2.5 */\
 763 \
 764     MOVNTQ(%%mm0,   (dst))\
 765     MOVNTQ(%%mm2,  8(dst))\
 766     MOVNTQ(%%mm3, 16(dst))\
 767     "add         $24, "#dst"            \n\t"\
 768 \
 769     "add          $8, "#index"          \n\t"\
 770     "cmp     "#dstw", "#index"          \n\t"\
 771     " jb          1b                    \n\t"
 772
 773 #define WRITEBGR24MMX(dst, dstw, index) \
 774     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */\
 775     "movq      %%mm2, %%mm1     \n\t" /* B */\
 776     "movq      %%mm5, %%mm6     \n\t" /* R */\
 777     "punpcklbw %%mm4, %%mm2     \n\t" /* GBGBGBGB 0 */\
 778     "punpcklbw %%mm7, %%mm5     \n\t" /* 0R0R0R0R 0 */\
 779     "punpckhbw %%mm4, %%mm1     \n\t" /* GBGBGBGB 2 */\
 780     "punpckhbw %%mm7, %%mm6     \n\t" /* 0R0R0R0R 2 */\
 781     "movq      %%mm2, %%mm0     \n\t" /* GBGBGBGB 0 */\
 782     "movq      %%mm1, %%mm3     \n\t" /* GBGBGBGB 2 */\
 783     "punpcklwd %%mm5, %%mm0     \n\t" /* 0RGB0RGB 0 */\
 784     "punpckhwd %%mm5, %%mm2     \n\t" /* 0RGB0RGB 1 */\
 785     "punpcklwd %%mm6, %%mm1     \n\t" /* 0RGB0RGB 2 */\
 786     "punpckhwd %%mm6, %%mm3     \n\t" /* 0RGB0RGB 3 */\
 787 \
 788     "movq      %%mm0, %%mm4     \n\t" /* 0RGB0RGB 0 */\
 789     "movq      %%mm2, %%mm6     \n\t" /* 0RGB0RGB 1 */\
 790     "movq      %%mm1, %%mm5     \n\t" /* 0RGB0RGB 2 */\
 791     "movq      %%mm3, %%mm7     \n\t" /* 0RGB0RGB 3 */\
 792 \
 793     "psllq       $40, %%mm0     \n\t" /* RGB00000 0 */\
 794     "psllq       $40, %%mm2     \n\t" /* RGB00000 1 */\
 795     "psllq       $40, %%mm1     \n\t" /* RGB00000 2 */\
 796     "psllq       $40, %%mm3     \n\t" /* RGB00000 3 */\
 797 \
 798     "punpckhdq %%mm4, %%mm0     \n\t" /* 0RGBRGB0 0 */\
 799     "punpckhdq %%mm6, %%mm2     \n\t" /* 0RGBRGB0 1 */\
 800     "punpckhdq %%mm5, %%mm1     \n\t" /* 0RGBRGB0 2 */\
 801     "punpckhdq %%mm7, %%mm3     \n\t" /* 0RGBRGB0 3 */\
 802 \
 803     "psrlq        $8, %%mm0     \n\t" /* 00RGBRGB 0 */\
 804     "movq      %%mm2, %%mm6     \n\t" /* 0RGBRGB0 1 */\
 805     "psllq       $40, %%mm2     \n\t" /* GB000000 1 */\
 806     "por       %%mm2, %%mm0     \n\t" /* GBRGBRGB 0 */\
 807     MOVNTQ(%%mm0, (dst))\
 808 \
 809     "psrlq       $24, %%mm6     \n\t" /* 0000RGBR 1 */\
 810     "movq      %%mm1, %%mm5     \n\t" /* 0RGBRGB0 2 */\
 811     "psllq       $24, %%mm1     \n\t" /* BRGB0000 2 */\
 812     "por       %%mm1, %%mm6     \n\t" /* BRGBRGBR 1 */\
 813     MOVNTQ(%%mm6, 8(dst))\
 814 \
 815     "psrlq       $40, %%mm5     \n\t" /* 000000RG 2 */\
 816     "psllq        $8, %%mm3     \n\t" /* RGBRGB00 3 */\
 817     "por       %%mm3, %%mm5     \n\t" /* RGBRGBRG 2 */\
 818     MOVNTQ(%%mm5, 16(dst))\
 819 \
 820     "add         $24, "#dst"    \n\t"\
 821 \
 822     "add          $8, "#index"  \n\t"\
 823     "cmp     "#dstw", "#index"  \n\t"\
 824     " jb          1b            \n\t"
 825
 826 #define WRITEBGR24MMX2(dst, dstw, index) \
 827     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */\
 828     "movq "MANGLE(ff_M24A)", %%mm0 \n\t"\
 829     "movq "MANGLE(ff_M24C)", %%mm7 \n\t"\
 830     "pshufw $0x50, %%mm2, %%mm1 \n\t" /* B3 B2 B3 B2  B1 B0 B1 B0 */\
 831     "pshufw $0x50, %%mm4, %%mm3 \n\t" /* G3 G2 G3 G2  G1 G0 G1 G0 */\
 832     "pshufw $0x00, %%mm5, %%mm6 \n\t" /* R1 R0 R1 R0  R1 R0 R1 R0 */\
 833 \
 834     "pand   %%mm0, %%mm1        \n\t" /*    B2        B1       B0 */\
 835     "pand   %%mm0, %%mm3        \n\t" /*    G2        G1       G0 */\
 836     "pand   %%mm7, %%mm6        \n\t" /*       R1        R0       */\
 837 \
 838     "psllq     $8, %%mm3        \n\t" /* G2        G1       G0    */\
 839     "por    %%mm1, %%mm6        \n\t"\
 840     "por    %%mm3, %%mm6        \n\t"\
 841     MOVNTQ(%%mm6, (dst))\
 842 \
 843     "psrlq     $8, %%mm4        \n\t" /* 00 G7 G6 G5  G4 G3 G2 G1 */\
 844     "pshufw $0xA5, %%mm2, %%mm1 \n\t" /* B5 B4 B5 B4  B3 B2 B3 B2 */\
 845     "pshufw $0x55, %%mm4, %%mm3 \n\t" /* G4 G3 G4 G3  G4 G3 G4 G3 */\
 846     "pshufw $0xA5, %%mm5, %%mm6 \n\t" /* R5 R4 R5 R4  R3 R2 R3 R2 */\
 847 \
 848     "pand "MANGLE(ff_M24B)", %%mm1 \n\t" /* B5       B4        B3    */\
 849     "pand   %%mm7, %%mm3        \n\t" /*       G4        G3       */\
 850     "pand   %%mm0, %%mm6        \n\t" /*    R4        R3       R2 */\
 851 \
 852     "por    %%mm1, %%mm3        \n\t" /* B5    G4 B4     G3 B3    */\
 853     "por    %%mm3, %%mm6        \n\t"\
 854     MOVNTQ(%%mm6, 8(dst))\
 855 \
 856     "pshufw $0xFF, %%mm2, %%mm1 \n\t" /* B7 B6 B7 B6  B7 B6 B6 B7 */\
 857     "pshufw $0xFA, %%mm4, %%mm3 \n\t" /* 00 G7 00 G7  G6 G5 G6 G5 */\
 858     "pshufw $0xFA, %%mm5, %%mm6 \n\t" /* R7 R6 R7 R6  R5 R4 R5 R4 */\
 859 \
 860     "pand   %%mm7, %%mm1        \n\t" /*       B7        B6       */\
 861     "pand   %%mm0, %%mm3        \n\t" /*    G7        G6       G5 */\
 862     "pand "MANGLE(ff_M24B)", %%mm6 \n\t" /* R7       R6        R5    */\
 863 \
 864     "por    %%mm1, %%mm3        \n\t"\
 865     "por    %%mm3, %%mm6        \n\t"\
 866     MOVNTQ(%%mm6, 16(dst))\
 867 \
 868     "add      $24, "#dst"       \n\t"\
 869 \
 870     "add       $8, "#index"     \n\t"\
 871     "cmp  "#dstw", "#index"     \n\t"\
 872     " jb       1b               \n\t"
 873
 874 #if COMPILE_TEMPLATE_MMX2
 875 #undef WRITEBGR24
 876 #define WRITEBGR24(dst, dstw, index)  WRITEBGR24MMX2(dst, dstw, index)
 877 #else
 878 #undef WRITEBGR24
 879 #define WRITEBGR24(dst, dstw, index)  WRITEBGR24MMX(dst, dstw, index)
 880 #endif
 881
 882 #define REAL_WRITEYUY2(dst, dstw, index) \
 883     "packuswb  %%mm3, %%mm3     \n\t"\
 884     "packuswb  %%mm4, %%mm4     \n\t"\
 885     "packuswb  %%mm7, %%mm1     \n\t"\
 886     "punpcklbw %%mm4, %%mm3     \n\t"\
 887     "movq      %%mm1, %%mm7     \n\t"\
 888     "punpcklbw %%mm3, %%mm1     \n\t"\
 889     "punpckhbw %%mm3, %%mm7     \n\t"\
 890 \
 891     MOVNTQ(%%mm1, (dst, index, 2))\
 892     MOVNTQ(%%mm7, 8(dst, index, 2))\
 893 \
 894     "add          $8, "#index"  \n\t"\
 895     "cmp     "#dstw", "#index"  \n\t"\
 896     " jb          1b            \n\t"
 897 #define WRITEYUY2(dst, dstw, index)  REAL_WRITEYUY2(dst, dstw, index)
 898
 899
 900 static inline void RENAME(yuv2yuvX)(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
 901                                     const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize, const int16_t **alpSrc,
 902                                     uint8_t *dest, uint8_t *uDest, uint8_t *vDest, uint8_t *aDest, long dstW, long chrDstW)
 903 {
 904 #if COMPILE_TEMPLATE_MMX
 905     if(!(c->flags & SWS_BITEXACT)) {
 906         if (c->flags & SWS_ACCURATE_RND) {
 907             if (uDest) {
 908                 YSCALEYUV2YV12X_ACCURATE(   "0", CHR_MMX_FILTER_OFFSET, uDest, chrDstW)
 909                 YSCALEYUV2YV12X_ACCURATE(AV_STRINGIFY(VOF), CHR_MMX_FILTER_OFFSET, vDest, chrDstW)
 910             }
 911             if (CONFIG_SWSCALE_ALPHA && aDest) {
 912                 YSCALEYUV2YV12X_ACCURATE(   "0", ALP_MMX_FILTER_OFFSET, aDest, dstW)
 913             }
 914
 915             YSCALEYUV2YV12X_ACCURATE("0", LUM_MMX_FILTER_OFFSET, dest, dstW)
 916         } else {
 917             if (uDest) {
 918                 YSCALEYUV2YV12X(   "0", CHR_MMX_FILTER_OFFSET, uDest, chrDstW)
 919                 YSCALEYUV2YV12X(AV_STRINGIFY(VOF), CHR_MMX_FILTER_OFFSET, vDest, chrDstW)
 920             }
 921             if (CONFIG_SWSCALE_ALPHA && aDest) {
 922                 YSCALEYUV2YV12X(   "0", ALP_MMX_FILTER_OFFSET, aDest, dstW)
 923             }
 924
 925             YSCALEYUV2YV12X("0", LUM_MMX_FILTER_OFFSET, dest, dstW)
 926         }
 927         return;
 928     }
 929 #endif
 930 #if COMPILE_TEMPLATE_ALTIVEC
 931     yuv2yuvX_altivec_real(lumFilter, lumSrc, lumFilterSize,
 932                           chrFilter, chrSrc, chrFilterSize,
 933                           dest, uDest, vDest, dstW, chrDstW);
 934 #else //COMPILE_TEMPLATE_ALTIVEC
 935     yuv2yuvXinC(lumFilter, lumSrc, lumFilterSize,
 936                 chrFilter, chrSrc, chrFilterSize,
 937                 alpSrc, dest, uDest, vDest, aDest, dstW, chrDstW);
 938 #endif //!COMPILE_TEMPLATE_ALTIVEC
 939 }
 940
 941 static inline void RENAME(yuv2nv12X)(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
 942                                      const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
 943                                      uint8_t *dest, uint8_t *uDest, int dstW, int chrDstW, enum PixelFormat dstFormat)
 944 {
 945     yuv2nv12XinC(lumFilter, lumSrc, lumFilterSize,
 946                  chrFilter, chrSrc, chrFilterSize,
 947                  dest, uDest, dstW, chrDstW, dstFormat);
 948 }
 949
 950 static inline void RENAME(yuv2yuv1)(SwsContext *c, const int16_t *lumSrc, const int16_t *chrSrc, const int16_t *alpSrc,
 951                                     uint8_t *dest, uint8_t *uDest, uint8_t *vDest, uint8_t *aDest, long dstW, long chrDstW)
 952 {
 953     int i;
 954 #if COMPILE_TEMPLATE_MMX
 955     if(!(c->flags & SWS_BITEXACT)) {
 956         long p= 4;
 957         const uint8_t *src[4]= {alpSrc + dstW, lumSrc + dstW, chrSrc + chrDstW, chrSrc + VOFW + chrDstW};
 958         uint8_t *dst[4]= {aDest, dest, uDest, vDest};
 959         x86_reg counter[4]= {dstW, dstW, chrDstW, chrDstW};
 960
 961         if (c->flags & SWS_ACCURATE_RND) {
 962             while(p--) {
 963                 if (dst[p]) {
 964                     __asm__ volatile(
 965                         YSCALEYUV2YV121_ACCURATE
 966                         :: "r" (src[p]), "r" (dst[p] + counter[p]),
 967                         "g" (-counter[p])
 968                         : "%"REG_a
 969                     );
 970                 }
 971             }
 972         } else {
 973             while(p--) {
 974                 if (dst[p]) {
 975                     __asm__ volatile(
 976                         YSCALEYUV2YV121
 977                         :: "r" (src[p]), "r" (dst[p] + counter[p]),
 978                         "g" (-counter[p])
 979                         : "%"REG_a
 980                     );
 981                 }
 982             }
 983         }
 984         return;
 985     }
 986 #endif
 987     for (i=0; i<dstW; i++) {
 988         int val= (lumSrc[i]+64)>>7;
 989
 990         if (val&256) {
 991             if (val<0) val=0;
 992             else       val=255;
 993         }
 994
 995         dest[i]= val;
 996     }
 997
 998     if (uDest)
 999         for (i=0; i<chrDstW; i++) {
1000             int u=(chrSrc[i       ]+64)>>7;
1001             int v=(chrSrc[i + VOFW]+64)>>7;
1002
1003             if ((u|v)&256) {
1004                 if (u<0)        u=0;
1005                 else if (u>255) u=255;
1006                 if (v<0)        v=0;
1007                 else if (v>255) v=255;
1008             }
1009
1010             uDest[i]= u;
1011             vDest[i]= v;
1012         }
1013
1014     if (CONFIG_SWSCALE_ALPHA && aDest)
1015         for (i=0; i<dstW; i++) {
1016             int val= (alpSrc[i]+64)>>7;
1017             aDest[i]= av_clip_uint8(val);
1018         }
1019 }
1020
1021
1022 /**
1023  * vertical scale YV12 to RGB
1024  */
1025 static inline void RENAME(yuv2packedX)(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize,
1026                                        const int16_t *chrFilter, const int16_t **chrSrc, int chrFilterSize,
1027                                        const int16_t **alpSrc, uint8_t *dest, long dstW, long dstY)
1028 {
1029 #if COMPILE_TEMPLATE_MMX
1030     x86_reg dummy=0;
1031     if(!(c->flags & SWS_BITEXACT)) {
1032         if (c->flags & SWS_ACCURATE_RND) {
1033             switch(c->dstFormat) {
1034             case PIX_FMT_RGB32:
1035                 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
1036                     YSCALEYUV2PACKEDX_ACCURATE
1037                     YSCALEYUV2RGBX
1038                     "movq                      %%mm2, "U_TEMP"(%0)  \n\t"
1039                     "movq                      %%mm4, "V_TEMP"(%0)  \n\t"
1040                     "movq                      %%mm5, "Y_TEMP"(%0)  \n\t"
1041                     YSCALEYUV2PACKEDX_ACCURATE_YA(ALP_MMX_FILTER_OFFSET)
1042                     "movq               "Y_TEMP"(%0), %%mm5         \n\t"
1043                     "psraw                        $3, %%mm1         \n\t"
1044                     "psraw                        $3, %%mm7         \n\t"
1045                     "packuswb                  %%mm7, %%mm1         \n\t"
1046                     WRITEBGR32(%4, %5, %%REGa, %%mm3, %%mm4, %%mm5, %%mm1, %%mm0, %%mm7, %%mm2, %%mm6)
1047
1048                     YSCALEYUV2PACKEDX_END
1049                 } else {
1050                     YSCALEYUV2PACKEDX_ACCURATE
1051                     YSCALEYUV2RGBX
1052                     "pcmpeqd %%mm7, %%mm7 \n\t"
1053                     WRITEBGR32(%4, %5, %%REGa, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
1054
1055                     YSCALEYUV2PACKEDX_END
1056                 }
1057                 return;
1058             case PIX_FMT_BGR24:
1059                 YSCALEYUV2PACKEDX_ACCURATE
1060                 YSCALEYUV2RGBX
1061                 "pxor %%mm7, %%mm7 \n\t"
1062                 "lea (%%"REG_a", %%"REG_a", 2), %%"REG_c"\n\t" //FIXME optimize
1063                 "add %4, %%"REG_c"                        \n\t"
1064                 WRITEBGR24(%%REGc, %5, %%REGa)
1065
1066
1067                 :: "r" (&c->redDither),
1068                 "m" (dummy), "m" (dummy), "m" (dummy),
1069                 "r" (dest), "m" (dstW)
1070                 : "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S
1071                 );
1072                 return;
1073             case PIX_FMT_RGB555:
1074                 YSCALEYUV2PACKEDX_ACCURATE
1075                 YSCALEYUV2RGBX
1076                 "pxor %%mm7, %%mm7 \n\t"
1077                 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1078 #ifdef DITHER1XBPP
1079                 "paddusb "BLUE_DITHER"(%0), %%mm2\n\t"
1080                 "paddusb "GREEN_DITHER"(%0), %%mm4\n\t"
1081                 "paddusb "RED_DITHER"(%0), %%mm5\n\t"
1082 #endif
1083
1084                 WRITERGB15(%4, %5, %%REGa)
1085                 YSCALEYUV2PACKEDX_END
1086                 return;
1087             case PIX_FMT_RGB565:
1088                 YSCALEYUV2PACKEDX_ACCURATE
1089                 YSCALEYUV2RGBX
1090                 "pxor %%mm7, %%mm7 \n\t"
1091                 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1092 #ifdef DITHER1XBPP
1093                 "paddusb "BLUE_DITHER"(%0), %%mm2\n\t"
1094                 "paddusb "GREEN_DITHER"(%0), %%mm4\n\t"
1095                 "paddusb "RED_DITHER"(%0), %%mm5\n\t"
1096 #endif
1097
1098                 WRITERGB16(%4, %5, %%REGa)
1099                 YSCALEYUV2PACKEDX_END
1100                 return;
1101             case PIX_FMT_YUYV422:
1102                 YSCALEYUV2PACKEDX_ACCURATE
1103                 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1104
1105                 "psraw $3, %%mm3    \n\t"
1106                 "psraw $3, %%mm4    \n\t"
1107                 "psraw $3, %%mm1    \n\t"
1108                 "psraw $3, %%mm7    \n\t"
1109                 WRITEYUY2(%4, %5, %%REGa)
1110                 YSCALEYUV2PACKEDX_END
1111                 return;
1112             }
1113         } else {
1114             switch(c->dstFormat) {
1115             case PIX_FMT_RGB32:
1116                 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
1117                     YSCALEYUV2PACKEDX
1118                     YSCALEYUV2RGBX
1119                     YSCALEYUV2PACKEDX_YA(ALP_MMX_FILTER_OFFSET, %%mm0, %%mm3, %%mm6, %%mm1, %%mm7)
1120                     "psraw                        $3, %%mm1         \n\t"
1121                     "psraw                        $3, %%mm7         \n\t"
1122                     "packuswb                  %%mm7, %%mm1         \n\t"
1123                     WRITEBGR32(%4, %5, %%REGa, %%mm2, %%mm4, %%mm5, %%mm1, %%mm0, %%mm7, %%mm3, %%mm6)
1124                     YSCALEYUV2PACKEDX_END
1125                 } else {
1126                     YSCALEYUV2PACKEDX
1127                     YSCALEYUV2RGBX
1128                     "pcmpeqd %%mm7, %%mm7 \n\t"
1129                     WRITEBGR32(%4, %5, %%REGa, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
1130                     YSCALEYUV2PACKEDX_END
1131                 }
1132                 return;
1133             case PIX_FMT_BGR24:
1134                 YSCALEYUV2PACKEDX
1135                 YSCALEYUV2RGBX
1136                 "pxor                    %%mm7, %%mm7       \n\t"
1137                 "lea (%%"REG_a", %%"REG_a", 2), %%"REG_c"   \n\t" //FIXME optimize
1138                 "add                        %4, %%"REG_c"   \n\t"
1139                 WRITEBGR24(%%REGc, %5, %%REGa)
1140
1141                 :: "r" (&c->redDither),
1142                 "m" (dummy), "m" (dummy), "m" (dummy),
1143                 "r" (dest),  "m" (dstW)
1144                 : "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S
1145                 );
1146                 return;
1147             case PIX_FMT_RGB555:
1148                 YSCALEYUV2PACKEDX
1149                 YSCALEYUV2RGBX
1150                 "pxor %%mm7, %%mm7 \n\t"
1151                 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1152 #ifdef DITHER1XBPP
1153                 "paddusb "BLUE_DITHER"(%0), %%mm2  \n\t"
1154                 "paddusb "GREEN_DITHER"(%0), %%mm4  \n\t"
1155                 "paddusb "RED_DITHER"(%0), %%mm5  \n\t"
1156 #endif
1157
1158                 WRITERGB15(%4, %5, %%REGa)
1159                 YSCALEYUV2PACKEDX_END
1160                 return;
1161             case PIX_FMT_RGB565:
1162                 YSCALEYUV2PACKEDX
1163                 YSCALEYUV2RGBX
1164                 "pxor %%mm7, %%mm7 \n\t"
1165                 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1166 #ifdef DITHER1XBPP
1167                 "paddusb "BLUE_DITHER"(%0), %%mm2  \n\t"
1168                 "paddusb "GREEN_DITHER"(%0), %%mm4  \n\t"
1169                 "paddusb "RED_DITHER"(%0), %%mm5  \n\t"
1170 #endif
1171
1172                 WRITERGB16(%4, %5, %%REGa)
1173                 YSCALEYUV2PACKEDX_END
1174                 return;
1175             case PIX_FMT_YUYV422:
1176                 YSCALEYUV2PACKEDX
1177                 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1178
1179                 "psraw $3, %%mm3    \n\t"
1180                 "psraw $3, %%mm4    \n\t"
1181                 "psraw $3, %%mm1    \n\t"
1182                 "psraw $3, %%mm7    \n\t"
1183                 WRITEYUY2(%4, %5, %%REGa)
1184                 YSCALEYUV2PACKEDX_END
1185                 return;
1186             }
1187         }
1188     }
1189 #endif /* COMPILE_TEMPLATE_MMX */
1190 #if COMPILE_TEMPLATE_ALTIVEC
1191     /* The following list of supported dstFormat values should
1192        match what's found in the body of ff_yuv2packedX_altivec() */
1193     if (!(c->flags & SWS_BITEXACT) && !c->alpPixBuf &&
1194          (c->dstFormat==PIX_FMT_ABGR  || c->dstFormat==PIX_FMT_BGRA  ||
1195           c->dstFormat==PIX_FMT_BGR24 || c->dstFormat==PIX_FMT_RGB24 ||
1196           c->dstFormat==PIX_FMT_RGBA  || c->dstFormat==PIX_FMT_ARGB))
1197             ff_yuv2packedX_altivec(c, lumFilter, lumSrc, lumFilterSize,
1198                                    chrFilter, chrSrc, chrFilterSize,
1199                                    dest, dstW, dstY);
1200     else
1201 #endif
1202         yuv2packedXinC(c, lumFilter, lumSrc, lumFilterSize,
1203                        chrFilter, chrSrc, chrFilterSize,
1204                        alpSrc, dest, dstW, dstY);
1205 }
1206
1207 /**
1208  * vertical bilinear scale YV12 to RGB
1209  */
1210 static inline void RENAME(yuv2packed2)(SwsContext *c, const uint16_t *buf0, const uint16_t *buf1, const uint16_t *uvbuf0, const uint16_t *uvbuf1,
1211                           const uint16_t *abuf0, const uint16_t *abuf1, uint8_t *dest, int dstW, int yalpha, int uvalpha, int y)
1212 {
1213     int  yalpha1=4095- yalpha;
1214     int uvalpha1=4095-uvalpha;
1215     int i;
1216
1217 #if COMPILE_TEMPLATE_MMX
1218     if(!(c->flags & SWS_BITEXACT)) {
1219         switch(c->dstFormat) {
1220         //Note 8280 == DSTW_OFFSET but the preprocessor can't handle that there :(
1221         case PIX_FMT_RGB32:
1222             if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
1223 #if ARCH_X86_64
1224                 __asm__ volatile(
1225                     YSCALEYUV2RGB(%%r8, %5)
1226                     YSCALEYUV2RGB_YA(%%r8, %5, %6, %7)
1227                     "psraw                  $3, %%mm1       \n\t" /* abuf0[eax] - abuf1[eax] >>7*/
1228                     "psraw                  $3, %%mm7       \n\t" /* abuf0[eax] - abuf1[eax] >>7*/
1229                     "packuswb            %%mm7, %%mm1       \n\t"
1230                     WRITEBGR32(%4, 8280(%5), %%r8, %%mm2, %%mm4, %%mm5, %%mm1, %%mm0, %%mm7, %%mm3, %%mm6)
1231
1232                     :: "c" (buf0), "d" (buf1), "S" (uvbuf0), "D" (uvbuf1), "r" (dest),
1233                     "a" (&c->redDither)
1234                     ,"r" (abuf0), "r" (abuf1)
1235                     : "%r8"
1236                 );
1237 #else
1238                 *(const uint16_t **)(&c->u_temp)=abuf0;
1239                 *(const uint16_t **)(&c->v_temp)=abuf1;
1240                 __asm__ volatile(
1241                     "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1242                     "mov        %4, %%"REG_b"               \n\t"
1243                     "push %%"REG_BP"                        \n\t"
1244                     YSCALEYUV2RGB(%%REGBP, %5)
1245                     "push                   %0              \n\t"
1246                     "push                   %1              \n\t"
1247                     "mov          "U_TEMP"(%5), %0          \n\t"
1248                     "mov          "V_TEMP"(%5), %1          \n\t"
1249                     YSCALEYUV2RGB_YA(%%REGBP, %5, %0, %1)
1250                     "psraw                  $3, %%mm1       \n\t" /* abuf0[eax] - abuf1[eax] >>7*/
1251                     "psraw                  $3, %%mm7       \n\t" /* abuf0[eax] - abuf1[eax] >>7*/
1252                     "packuswb            %%mm7, %%mm1       \n\t"
1253                     "pop                    %1              \n\t"
1254                     "pop                    %0              \n\t"
1255                     WRITEBGR32(%%REGb, 8280(%5), %%REGBP, %%mm2, %%mm4, %%mm5, %%mm1, %%mm0, %%mm7, %%mm3, %%mm6)
1256                     "pop %%"REG_BP"                         \n\t"
1257                     "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1258
1259                     :: "c" (buf0), "d" (buf1), "S" (uvbuf0), "D" (uvbuf1), "m" (dest),
1260                     "a" (&c->redDither)
1261                 );
1262 #endif
1263             } else {
1264                 __asm__ volatile(
1265                     "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1266                     "mov        %4, %%"REG_b"               \n\t"
1267                     "push %%"REG_BP"                        \n\t"
1268                     YSCALEYUV2RGB(%%REGBP, %5)
1269                     "pcmpeqd %%mm7, %%mm7                   \n\t"
1270                     WRITEBGR32(%%REGb, 8280(%5), %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
1271                     "pop %%"REG_BP"                         \n\t"
1272                     "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1273
1274                     :: "c" (buf0), "d" (buf1), "S" (uvbuf0), "D" (uvbuf1), "m" (dest),
1275                     "a" (&c->redDither)
1276                 );
1277             }
1278             return;
1279         case PIX_FMT_BGR24:
1280             __asm__ volatile(
1281                 "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1282                 "mov        %4, %%"REG_b"               \n\t"
1283                 "push %%"REG_BP"                        \n\t"
1284                 YSCALEYUV2RGB(%%REGBP, %5)
1285                 "pxor    %%mm7, %%mm7                   \n\t"
1286                 WRITEBGR24(%%REGb, 8280(%5), %%REGBP)
1287                 "pop %%"REG_BP"                         \n\t"
1288                 "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1289                 :: "c" (buf0), "d" (buf1), "S" (uvbuf0), "D" (uvbuf1), "m" (dest),
1290                 "a" (&c->redDither)
1291             );
1292             return;
1293         case PIX_FMT_RGB555:
1294             __asm__ volatile(
1295                 "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1296                 "mov        %4, %%"REG_b"               \n\t"
1297                 "push %%"REG_BP"                        \n\t"
1298                 YSCALEYUV2RGB(%%REGBP, %5)
1299                 "pxor    %%mm7, %%mm7                   \n\t"
1300                 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1301 #ifdef DITHER1XBPP
1302                 "paddusb "BLUE_DITHER"(%5), %%mm2      \n\t"
1303                 "paddusb "GREEN_DITHER"(%5), %%mm4      \n\t"
1304                 "paddusb "RED_DITHER"(%5), %%mm5      \n\t"
1305 #endif
1306
1307                 WRITERGB15(%%REGb, 8280(%5), %%REGBP)
1308                 "pop %%"REG_BP"                         \n\t"
1309                 "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1310
1311                 :: "c" (buf0), "d" (buf1), "S" (uvbuf0), "D" (uvbuf1), "m" (dest),
1312                 "a" (&c->redDither)
1313             );
1314             return;
1315         case PIX_FMT_RGB565:
1316             __asm__ volatile(
1317                 "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1318                 "mov        %4, %%"REG_b"               \n\t"
1319                 "push %%"REG_BP"                        \n\t"
1320                 YSCALEYUV2RGB(%%REGBP, %5)
1321                 "pxor    %%mm7, %%mm7                   \n\t"
1322                 /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1323 #ifdef DITHER1XBPP
1324                 "paddusb "BLUE_DITHER"(%5), %%mm2      \n\t"
1325                 "paddusb "GREEN_DITHER"(%5), %%mm4      \n\t"
1326                 "paddusb "RED_DITHER"(%5), %%mm5      \n\t"
1327 #endif
1328
1329                 WRITERGB16(%%REGb, 8280(%5), %%REGBP)
1330                 "pop %%"REG_BP"                         \n\t"
1331                 "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1332                 :: "c" (buf0), "d" (buf1), "S" (uvbuf0), "D" (uvbuf1), "m" (dest),
1333                 "a" (&c->redDither)
1334             );
1335             return;
1336         case PIX_FMT_YUYV422:
1337             __asm__ volatile(
1338                 "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1339                 "mov %4, %%"REG_b"                        \n\t"
1340                 "push %%"REG_BP"                        \n\t"
1341                 YSCALEYUV2PACKED(%%REGBP, %5)
1342                 WRITEYUY2(%%REGb, 8280(%5), %%REGBP)
1343                 "pop %%"REG_BP"                         \n\t"
1344                 "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1345                 :: "c" (buf0), "d" (buf1), "S" (uvbuf0), "D" (uvbuf1), "m" (dest),
1346                 "a" (&c->redDither)
1347             );
1348             return;
1349         default: break;
1350         }
1351     }
1352 #endif //COMPILE_TEMPLATE_MMX
1353     YSCALE_YUV_2_ANYRGB_C(YSCALE_YUV_2_RGB2_C, YSCALE_YUV_2_PACKED2_C(void,0), YSCALE_YUV_2_GRAY16_2_C, YSCALE_YUV_2_MONO2_C)
1354 }
1355
1356 /**
1357  * YV12 to RGB without scaling or interpolating
1358  */
1359 static inline void RENAME(yuv2packed1)(SwsContext *c, const uint16_t *buf0, const uint16_t *uvbuf0, const uint16_t *uvbuf1,
1360                           const uint16_t *abuf0, uint8_t *dest, int dstW, int uvalpha, enum PixelFormat dstFormat, int flags, int y)
1361 {
1362     const int yalpha1=0;
1363     int i;
1364
1365     const uint16_t *buf1= buf0; //FIXME needed for RGB1/BGR1
1366     const int yalpha= 4096; //FIXME ...
1367
1368     if (flags&SWS_FULL_CHR_H_INT) {
1369         c->yuv2packed2(c, buf0, buf0, uvbuf0, uvbuf1, abuf0, abuf0, dest, dstW, 0, uvalpha, y);
1370         return;
1371     }
1372
1373 #if COMPILE_TEMPLATE_MMX
1374     if(!(flags & SWS_BITEXACT)) {
1375         if (uvalpha < 2048) { // note this is not correct (shifts chrominance by 0.5 pixels) but it is a bit faster
1376             switch(dstFormat) {
1377             case PIX_FMT_RGB32:
1378                 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
1379                     __asm__ volatile(
1380                         "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1381                         "mov        %4, %%"REG_b"               \n\t"
1382                         "push %%"REG_BP"                        \n\t"
1383                         YSCALEYUV2RGB1(%%REGBP, %5)
1384                         YSCALEYUV2RGB1_ALPHA(%%REGBP)
1385                         WRITEBGR32(%%REGb, 8280(%5), %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
1386                         "pop %%"REG_BP"                         \n\t"
1387                         "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1388
1389                         :: "c" (buf0), "d" (abuf0), "S" (uvbuf0), "D" (uvbuf1), "m" (dest),
1390                         "a" (&c->redDither)
1391                     );
1392                 } else {
1393                     __asm__ volatile(
1394                         "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1395                         "mov        %4, %%"REG_b"               \n\t"
1396                         "push %%"REG_BP"                        \n\t"
1397                         YSCALEYUV2RGB1(%%REGBP, %5)
1398                         "pcmpeqd %%mm7, %%mm7                   \n\t"
1399                         WRITEBGR32(%%REGb, 8280(%5), %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
1400                         "pop %%"REG_BP"                         \n\t"
1401                         "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1402
1403                         :: "c" (buf0), "d" (buf1), "S" (uvbuf0), "D" (uvbuf1), "m" (dest),
1404                         "a" (&c->redDither)
1405                     );
1406                 }
1407                 return;
1408             case PIX_FMT_BGR24:
1409                 __asm__ volatile(
1410                     "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1411                     "mov        %4, %%"REG_b"               \n\t"
1412                     "push %%"REG_BP"                        \n\t"
1413                     YSCALEYUV2RGB1(%%REGBP, %5)
1414                     "pxor    %%mm7, %%mm7                   \n\t"
1415                     WRITEBGR24(%%REGb, 8280(%5), %%REGBP)
1416                     "pop %%"REG_BP"                         \n\t"
1417                     "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1418
1419                     :: "c" (buf0), "d" (buf1), "S" (uvbuf0), "D" (uvbuf1), "m" (dest),
1420                     "a" (&c->redDither)
1421                 );
1422                 return;
1423             case PIX_FMT_RGB555:
1424                 __asm__ volatile(
1425                     "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1426                     "mov        %4, %%"REG_b"               \n\t"
1427                     "push %%"REG_BP"                        \n\t"
1428                     YSCALEYUV2RGB1(%%REGBP, %5)
1429                     "pxor    %%mm7, %%mm7                   \n\t"
1430                     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1431 #ifdef DITHER1XBPP
1432                     "paddusb "BLUE_DITHER"(%5), %%mm2      \n\t"
1433                     "paddusb "GREEN_DITHER"(%5), %%mm4      \n\t"
1434                     "paddusb "RED_DITHER"(%5), %%mm5      \n\t"
1435 #endif
1436                     WRITERGB15(%%REGb, 8280(%5), %%REGBP)
1437                     "pop %%"REG_BP"                         \n\t"
1438                     "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1439
1440                     :: "c" (buf0), "d" (buf1), "S" (uvbuf0), "D" (uvbuf1), "m" (dest),
1441                     "a" (&c->redDither)
1442                 );
1443                 return;
1444             case PIX_FMT_RGB565:
1445                 __asm__ volatile(
1446                     "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1447                     "mov        %4, %%"REG_b"               \n\t"
1448                     "push %%"REG_BP"                        \n\t"
1449                     YSCALEYUV2RGB1(%%REGBP, %5)
1450                     "pxor    %%mm7, %%mm7                   \n\t"
1451                     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1452 #ifdef DITHER1XBPP
1453                     "paddusb "BLUE_DITHER"(%5), %%mm2      \n\t"
1454                     "paddusb "GREEN_DITHER"(%5), %%mm4      \n\t"
1455                     "paddusb "RED_DITHER"(%5), %%mm5      \n\t"
1456 #endif
1457
1458                     WRITERGB16(%%REGb, 8280(%5), %%REGBP)
1459                     "pop %%"REG_BP"                         \n\t"
1460                     "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1461
1462                     :: "c" (buf0), "d" (buf1), "S" (uvbuf0), "D" (uvbuf1), "m" (dest),
1463                     "a" (&c->redDither)
1464                 );
1465                 return;
1466             case PIX_FMT_YUYV422:
1467                 __asm__ volatile(
1468                     "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1469                     "mov        %4, %%"REG_b"               \n\t"
1470                     "push %%"REG_BP"                        \n\t"
1471                     YSCALEYUV2PACKED1(%%REGBP, %5)
1472                     WRITEYUY2(%%REGb, 8280(%5), %%REGBP)
1473                     "pop %%"REG_BP"                         \n\t"
1474                     "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1475
1476                     :: "c" (buf0), "d" (buf1), "S" (uvbuf0), "D" (uvbuf1), "m" (dest),
1477                     "a" (&c->redDither)
1478                 );
1479                 return;
1480             }
1481         } else {
1482             switch(dstFormat) {
1483             case PIX_FMT_RGB32:
1484                 if (CONFIG_SWSCALE_ALPHA && c->alpPixBuf) {
1485                     __asm__ volatile(
1486                         "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1487                         "mov        %4, %%"REG_b"               \n\t"
1488                         "push %%"REG_BP"                        \n\t"
1489                         YSCALEYUV2RGB1b(%%REGBP, %5)
1490                         YSCALEYUV2RGB1_ALPHA(%%REGBP)
1491                         WRITEBGR32(%%REGb, 8280(%5), %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
1492                         "pop %%"REG_BP"                         \n\t"
1493                         "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1494
1495                         :: "c" (buf0), "d" (abuf0), "S" (uvbuf0), "D" (uvbuf1), "m" (dest),
1496                         "a" (&c->redDither)
1497                     );
1498                 } else {
1499                     __asm__ volatile(
1500                         "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1501                         "mov        %4, %%"REG_b"               \n\t"
1502                         "push %%"REG_BP"                        \n\t"
1503                         YSCALEYUV2RGB1b(%%REGBP, %5)
1504                         "pcmpeqd %%mm7, %%mm7                   \n\t"
1505                         WRITEBGR32(%%REGb, 8280(%5), %%REGBP, %%mm2, %%mm4, %%mm5, %%mm7, %%mm0, %%mm1, %%mm3, %%mm6)
1506                         "pop %%"REG_BP"                         \n\t"
1507                         "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1508
1509                         :: "c" (buf0), "d" (buf1), "S" (uvbuf0), "D" (uvbuf1), "m" (dest),
1510                         "a" (&c->redDither)
1511                     );
1512                 }
1513                 return;
1514             case PIX_FMT_BGR24:
1515                 __asm__ volatile(
1516                     "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1517                     "mov        %4, %%"REG_b"               \n\t"
1518                     "push %%"REG_BP"                        \n\t"
1519                     YSCALEYUV2RGB1b(%%REGBP, %5)
1520                     "pxor    %%mm7, %%mm7                   \n\t"
1521                     WRITEBGR24(%%REGb, 8280(%5), %%REGBP)
1522                     "pop %%"REG_BP"                         \n\t"
1523                     "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1524
1525                     :: "c" (buf0), "d" (buf1), "S" (uvbuf0), "D" (uvbuf1), "m" (dest),
1526                     "a" (&c->redDither)
1527                 );
1528                 return;
1529             case PIX_FMT_RGB555:
1530                 __asm__ volatile(
1531                     "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1532                     "mov        %4, %%"REG_b"               \n\t"
1533                     "push %%"REG_BP"                        \n\t"
1534                     YSCALEYUV2RGB1b(%%REGBP, %5)
1535                     "pxor    %%mm7, %%mm7                   \n\t"
1536                     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1537 #ifdef DITHER1XBPP
1538                     "paddusb "BLUE_DITHER"(%5), %%mm2      \n\t"
1539                     "paddusb "GREEN_DITHER"(%5), %%mm4      \n\t"
1540                     "paddusb "RED_DITHER"(%5), %%mm5      \n\t"
1541 #endif
1542                     WRITERGB15(%%REGb, 8280(%5), %%REGBP)
1543                     "pop %%"REG_BP"                         \n\t"
1544                     "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1545
1546                     :: "c" (buf0), "d" (buf1), "S" (uvbuf0), "D" (uvbuf1), "m" (dest),
1547                     "a" (&c->redDither)
1548                 );
1549                 return;
1550             case PIX_FMT_RGB565:
1551                 __asm__ volatile(
1552                     "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1553                     "mov        %4, %%"REG_b"               \n\t"
1554                     "push %%"REG_BP"                        \n\t"
1555                     YSCALEYUV2RGB1b(%%REGBP, %5)
1556                     "pxor    %%mm7, %%mm7                   \n\t"
1557                     /* mm2=B, %%mm4=G, %%mm5=R, %%mm7=0 */
1558 #ifdef DITHER1XBPP
1559                     "paddusb "BLUE_DITHER"(%5), %%mm2      \n\t"
1560                     "paddusb "GREEN_DITHER"(%5), %%mm4      \n\t"
1561                     "paddusb "RED_DITHER"(%5), %%mm5      \n\t"
1562 #endif
1563
1564                     WRITERGB16(%%REGb, 8280(%5), %%REGBP)
1565                     "pop %%"REG_BP"                         \n\t"
1566                     "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1567
1568                     :: "c" (buf0), "d" (buf1), "S" (uvbuf0), "D" (uvbuf1), "m" (dest),
1569                     "a" (&c->redDither)
1570                 );
1571                 return;
1572             case PIX_FMT_YUYV422:
1573                 __asm__ volatile(
1574                     "mov %%"REG_b", "ESP_OFFSET"(%5)        \n\t"
1575                     "mov        %4, %%"REG_b"               \n\t"
1576                     "push %%"REG_BP"                        \n\t"
1577                     YSCALEYUV2PACKED1b(%%REGBP, %5)
1578                     WRITEYUY2(%%REGb, 8280(%5), %%REGBP)
1579                     "pop %%"REG_BP"                         \n\t"
1580                     "mov "ESP_OFFSET"(%5), %%"REG_b"        \n\t"
1581
1582                     :: "c" (buf0), "d" (buf1), "S" (uvbuf0), "D" (uvbuf1), "m" (dest),
1583                     "a" (&c->redDither)
1584                 );
1585                 return;
1586             }
1587         }
1588     }
1589 #endif /* COMPILE_TEMPLATE_MMX */
1590     if (uvalpha < 2048) {
1591         YSCALE_YUV_2_ANYRGB_C(YSCALE_YUV_2_RGB1_C, YSCALE_YUV_2_PACKED1_C(void,0), YSCALE_YUV_2_GRAY16_1_C, YSCALE_YUV_2_MONO2_C)
1592     } else {
1593         YSCALE_YUV_2_ANYRGB_C(YSCALE_YUV_2_RGB1B_C, YSCALE_YUV_2_PACKED1B_C(void,0), YSCALE_YUV_2_GRAY16_1_C, YSCALE_YUV_2_MONO2_C)
1594     }
1595 }
1596
1597 //FIXME yuy2* can read up to 7 samples too much
1598
1599 static inline void RENAME(yuy2ToY)(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)
1600 {
1601 #if COMPILE_TEMPLATE_MMX
1602     __asm__ volatile(
1603         "movq "MANGLE(bm01010101)", %%mm2           \n\t"
1604         "mov                    %0, %%"REG_a"       \n\t"
1605         "1:                                         \n\t"
1606         "movq    (%1, %%"REG_a",2), %%mm0           \n\t"
1607         "movq   8(%1, %%"REG_a",2), %%mm1           \n\t"
1608         "pand                %%mm2, %%mm0           \n\t"
1609         "pand                %%mm2, %%mm1           \n\t"
1610         "packuswb            %%mm1, %%mm0           \n\t"
1611         "movq                %%mm0, (%2, %%"REG_a") \n\t"
1612         "add                    $8, %%"REG_a"       \n\t"
1613         " js                    1b                  \n\t"
1614         : : "g" ((x86_reg)-width), "r" (src+width*2), "r" (dst+width)
1615         : "%"REG_a
1616     );
1617 #else
1618     int i;
1619     for (i=0; i<width; i++)
1620         dst[i]= src[2*i];
1621 #endif
1622 }
1623
1624 static inline void RENAME(yuy2ToUV)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src1, const uint8_t *src2, long width, uint32_t *unused)
1625 {
1626 #if COMPILE_TEMPLATE_MMX
1627     __asm__ volatile(
1628         "movq "MANGLE(bm01010101)", %%mm4           \n\t"
1629         "mov                    %0, %%"REG_a"       \n\t"
1630         "1:                                         \n\t"
1631         "movq    (%1, %%"REG_a",4), %%mm0           \n\t"
1632         "movq   8(%1, %%"REG_a",4), %%mm1           \n\t"
1633         "psrlw                  $8, %%mm0           \n\t"
1634         "psrlw                  $8, %%mm1           \n\t"
1635         "packuswb            %%mm1, %%mm0           \n\t"
1636         "movq                %%mm0, %%mm1           \n\t"
1637         "psrlw                  $8, %%mm0           \n\t"
1638         "pand                %%mm4, %%mm1           \n\t"
1639         "packuswb            %%mm0, %%mm0           \n\t"
1640         "packuswb            %%mm1, %%mm1           \n\t"
1641         "movd                %%mm0, (%3, %%"REG_a") \n\t"
1642         "movd                %%mm1, (%2, %%"REG_a") \n\t"
1643         "add                    $4, %%"REG_a"       \n\t"
1644         " js                    1b                  \n\t"
1645         : : "g" ((x86_reg)-width), "r" (src1+width*4), "r" (dstU+width), "r" (dstV+width)
1646         : "%"REG_a
1647     );
1648 #else
1649     int i;
1650     for (i=0; i<width; i++) {
1651         dstU[i]= src1[4*i + 1];
1652         dstV[i]= src1[4*i + 3];
1653     }
1654 #endif
1655     assert(src1 == src2);
1656 }
1657
1658 static inline void RENAME(LEToUV)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src1, const uint8_t *src2, long width, uint32_t *unused)
1659 {
1660 #if COMPILE_TEMPLATE_MMX
1661     __asm__ volatile(
1662         "mov                    %0, %%"REG_a"       \n\t"
1663         "1:                                         \n\t"
1664         "movq    (%1, %%"REG_a",2), %%mm0           \n\t"
1665         "movq   8(%1, %%"REG_a",2), %%mm1           \n\t"
1666         "movq    (%2, %%"REG_a",2), %%mm2           \n\t"
1667         "movq   8(%2, %%"REG_a",2), %%mm3           \n\t"
1668         "psrlw                  $8, %%mm0           \n\t"
1669         "psrlw                  $8, %%mm1           \n\t"
1670         "psrlw                  $8, %%mm2           \n\t"
1671         "psrlw                  $8, %%mm3           \n\t"
1672         "packuswb            %%mm1, %%mm0           \n\t"
1673         "packuswb            %%mm3, %%mm2           \n\t"
1674         "movq                %%mm0, (%3, %%"REG_a") \n\t"
1675         "movq                %%mm2, (%4, %%"REG_a") \n\t"
1676         "add                    $8, %%"REG_a"       \n\t"
1677         " js                    1b                  \n\t"
1678         : : "g" ((x86_reg)-width), "r" (src1+width*2), "r" (src2+width*2), "r" (dstU+width), "r" (dstV+width)
1679         : "%"REG_a
1680     );
1681 #else
1682     int i;
1683     for (i=0; i<width; i++) {
1684         dstU[i]= src1[2*i + 1];
1685         dstV[i]= src2[2*i + 1];
1686     }
1687 #endif
1688 }
1689
1690 /* This is almost identical to the previous, end exists only because
1691  * yuy2ToY/UV)(dst, src+1, ...) would have 100% unaligned accesses. */
1692 static inline void RENAME(uyvyToY)(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)
1693 {
1694 #if COMPILE_TEMPLATE_MMX
1695     __asm__ volatile(
1696         "mov                  %0, %%"REG_a"         \n\t"
1697         "1:                                         \n\t"
1698         "movq  (%1, %%"REG_a",2), %%mm0             \n\t"
1699         "movq 8(%1, %%"REG_a",2), %%mm1             \n\t"
1700         "psrlw                $8, %%mm0             \n\t"
1701         "psrlw                $8, %%mm1             \n\t"
1702         "packuswb          %%mm1, %%mm0             \n\t"
1703         "movq              %%mm0, (%2, %%"REG_a")   \n\t"
1704         "add                  $8, %%"REG_a"         \n\t"
1705         " js                  1b                    \n\t"
1706         : : "g" ((x86_reg)-width), "r" (src+width*2), "r" (dst+width)
1707         : "%"REG_a
1708     );
1709 #else
1710     int i;
1711     for (i=0; i<width; i++)
1712         dst[i]= src[2*i+1];
1713 #endif
1714 }
1715
1716 static inline void RENAME(uyvyToUV)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src1, const uint8_t *src2, long width, uint32_t *unused)
1717 {
1718 #if COMPILE_TEMPLATE_MMX
1719     __asm__ volatile(
1720         "movq "MANGLE(bm01010101)", %%mm4           \n\t"
1721         "mov                    %0, %%"REG_a"       \n\t"
1722         "1:                                         \n\t"
1723         "movq    (%1, %%"REG_a",4), %%mm0           \n\t"
1724         "movq   8(%1, %%"REG_a",4), %%mm1           \n\t"
1725         "pand                %%mm4, %%mm0           \n\t"
1726         "pand                %%mm4, %%mm1           \n\t"
1727         "packuswb            %%mm1, %%mm0           \n\t"
1728         "movq                %%mm0, %%mm1           \n\t"
1729         "psrlw                  $8, %%mm0           \n\t"
1730         "pand                %%mm4, %%mm1           \n\t"
1731         "packuswb            %%mm0, %%mm0           \n\t"
1732         "packuswb            %%mm1, %%mm1           \n\t"
1733         "movd                %%mm0, (%3, %%"REG_a") \n\t"
1734         "movd                %%mm1, (%2, %%"REG_a") \n\t"
1735         "add                    $4, %%"REG_a"       \n\t"
1736         " js                    1b                  \n\t"
1737         : : "g" ((x86_reg)-width), "r" (src1+width*4), "r" (dstU+width), "r" (dstV+width)
1738         : "%"REG_a
1739     );
1740 #else
1741     int i;
1742     for (i=0; i<width; i++) {
1743         dstU[i]= src1[4*i + 0];
1744         dstV[i]= src1[4*i + 2];
1745     }
1746 #endif
1747     assert(src1 == src2);
1748 }
1749
1750 static inline void RENAME(BEToUV)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src1, const uint8_t *src2, long width, uint32_t *unused)
1751 {
1752 #if COMPILE_TEMPLATE_MMX
1753     __asm__ volatile(
1754         "movq "MANGLE(bm01010101)", %%mm4           \n\t"
1755         "mov                    %0, %%"REG_a"       \n\t"
1756         "1:                                         \n\t"
1757         "movq    (%1, %%"REG_a",2), %%mm0           \n\t"
1758         "movq   8(%1, %%"REG_a",2), %%mm1           \n\t"
1759         "movq    (%2, %%"REG_a",2), %%mm2           \n\t"
1760         "movq   8(%2, %%"REG_a",2), %%mm3           \n\t"
1761         "pand                %%mm4, %%mm0           \n\t"
1762         "pand                %%mm4, %%mm1           \n\t"
1763         "pand                %%mm4, %%mm2           \n\t"
1764         "pand                %%mm4, %%mm3           \n\t"
1765         "packuswb            %%mm1, %%mm0           \n\t"
1766         "packuswb            %%mm3, %%mm2           \n\t"
1767         "movq                %%mm0, (%3, %%"REG_a") \n\t"
1768         "movq                %%mm2, (%4, %%"REG_a") \n\t"
1769         "add                    $8, %%"REG_a"       \n\t"
1770         " js                    1b                  \n\t"
1771         : : "g" ((x86_reg)-width), "r" (src1+width*2), "r" (src2+width*2), "r" (dstU+width), "r" (dstV+width)
1772         : "%"REG_a
1773     );
1774 #else
1775     int i;
1776     for (i=0; i<width; i++) {
1777         dstU[i]= src1[2*i];
1778         dstV[i]= src2[2*i];
1779     }
1780 #endif
1781 }
1782
1783 static inline void RENAME(nvXXtoUV)(uint8_t *dst1, uint8_t *dst2,
1784                                     const uint8_t *src, long width)
1785 {
1786 #if COMPILE_TEMPLATE_MMX
1787     __asm__ volatile(
1788         "movq "MANGLE(bm01010101)", %%mm4           \n\t"
1789         "mov                    %0, %%"REG_a"       \n\t"
1790         "1:                                         \n\t"
1791         "movq    (%1, %%"REG_a",2), %%mm0           \n\t"
1792         "movq   8(%1, %%"REG_a",2), %%mm1           \n\t"
1793         "movq                %%mm0, %%mm2           \n\t"
1794         "movq                %%mm1, %%mm3           \n\t"
1795         "pand                %%mm4, %%mm0           \n\t"
1796         "pand                %%mm4, %%mm1           \n\t"
1797         "psrlw                  $8, %%mm2           \n\t"
1798         "psrlw                  $8, %%mm3           \n\t"
1799         "packuswb            %%mm1, %%mm0           \n\t"
1800         "packuswb            %%mm3, %%mm2           \n\t"
1801         "movq                %%mm0, (%2, %%"REG_a") \n\t"
1802         "movq                %%mm2, (%3, %%"REG_a") \n\t"
1803         "add                    $8, %%"REG_a"       \n\t"
1804         " js                    1b                  \n\t"
1805         : : "g" ((x86_reg)-width), "r" (src+width*2), "r" (dst1+width), "r" (dst2+width)
1806         : "%"REG_a
1807     );
1808 #else
1809     int i;
1810     for (i = 0; i < width; i++) {
1811         dst1[i] = src[2*i+0];
1812         dst2[i] = src[2*i+1];
1813     }
1814 #endif
1815 }
1816
1817 static inline void RENAME(nv12ToUV)(uint8_t *dstU, uint8_t *dstV,
1818                                     const uint8_t *src1, const uint8_t *src2,
1819                                     long width, uint32_t *unused)
1820 {
1821     RENAME(nvXXtoUV)(dstU, dstV, src1, width);
1822 }
1823
1824 static inline void RENAME(nv21ToUV)(uint8_t *dstU, uint8_t *dstV,
1825                                     const uint8_t *src1, const uint8_t *src2,
1826                                     long width, uint32_t *unused)
1827 {
1828     RENAME(nvXXtoUV)(dstV, dstU, src1, width);
1829 }
1830
1831 #if COMPILE_TEMPLATE_MMX
1832 static inline void RENAME(bgr24ToY_mmx)(uint8_t *dst, const uint8_t *src, long width, enum PixelFormat srcFormat)
1833 {
1834
1835     if(srcFormat == PIX_FMT_BGR24) {
1836         __asm__ volatile(
1837             "movq  "MANGLE(ff_bgr24toY1Coeff)", %%mm5       \n\t"
1838             "movq  "MANGLE(ff_bgr24toY2Coeff)", %%mm6       \n\t"
1839             :
1840         );
1841     } else {
1842         __asm__ volatile(
1843             "movq  "MANGLE(ff_rgb24toY1Coeff)", %%mm5       \n\t"
1844             "movq  "MANGLE(ff_rgb24toY2Coeff)", %%mm6       \n\t"
1845             :
1846         );
1847     }
1848
1849     __asm__ volatile(
1850         "movq  "MANGLE(ff_bgr24toYOffset)", %%mm4   \n\t"
1851         "mov                        %2, %%"REG_a"   \n\t"
1852         "pxor                    %%mm7, %%mm7       \n\t"
1853         "1:                                         \n\t"
1854         PREFETCH"               64(%0)              \n\t"
1855         "movd                     (%0), %%mm0       \n\t"
1856         "movd                    2(%0), %%mm1       \n\t"
1857         "movd                    6(%0), %%mm2       \n\t"
1858         "movd                    8(%0), %%mm3       \n\t"
1859         "add                       $12, %0          \n\t"
1860         "punpcklbw               %%mm7, %%mm0       \n\t"
1861         "punpcklbw               %%mm7, %%mm1       \n\t"
1862         "punpcklbw               %%mm7, %%mm2       \n\t"
1863         "punpcklbw               %%mm7, %%mm3       \n\t"
1864         "pmaddwd                 %%mm5, %%mm0       \n\t"
1865         "pmaddwd                 %%mm6, %%mm1       \n\t"
1866         "pmaddwd                 %%mm5, %%mm2       \n\t"
1867         "pmaddwd                 %%mm6, %%mm3       \n\t"
1868         "paddd                   %%mm1, %%mm0       \n\t"
1869         "paddd                   %%mm3, %%mm2       \n\t"
1870         "paddd                   %%mm4, %%mm0       \n\t"
1871         "paddd                   %%mm4, %%mm2       \n\t"
1872         "psrad                     $15, %%mm0       \n\t"
1873         "psrad                     $15, %%mm2       \n\t"
1874         "packssdw                %%mm2, %%mm0       \n\t"
1875         "packuswb                %%mm0, %%mm0       \n\t"
1876         "movd                %%mm0, (%1, %%"REG_a") \n\t"
1877         "add                        $4, %%"REG_a"   \n\t"
1878         " js                        1b              \n\t"
1879     : "+r" (src)
1880     : "r" (dst+width), "g" ((x86_reg)-width)
1881     : "%"REG_a
1882     );
1883 }
1884
1885 static inline void RENAME(bgr24ToUV_mmx)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src, long width, enum PixelFormat srcFormat)
1886 {
1887     __asm__ volatile(
1888         "movq                    24+%4, %%mm6       \n\t"
1889         "mov                        %3, %%"REG_a"   \n\t"
1890         "pxor                    %%mm7, %%mm7       \n\t"
1891         "1:                                         \n\t"
1892         PREFETCH"               64(%0)              \n\t"
1893         "movd                     (%0), %%mm0       \n\t"
1894         "movd                    2(%0), %%mm1       \n\t"
1895         "punpcklbw               %%mm7, %%mm0       \n\t"
1896         "punpcklbw               %%mm7, %%mm1       \n\t"
1897         "movq                    %%mm0, %%mm2       \n\t"
1898         "movq                    %%mm1, %%mm3       \n\t"
1899         "pmaddwd                    %4, %%mm0       \n\t"
1900         "pmaddwd                  8+%4, %%mm1       \n\t"
1901         "pmaddwd                 16+%4, %%mm2       \n\t"
1902         "pmaddwd                 %%mm6, %%mm3       \n\t"
1903         "paddd                   %%mm1, %%mm0       \n\t"
1904         "paddd                   %%mm3, %%mm2       \n\t"
1905
1906         "movd                    6(%0), %%mm1       \n\t"
1907         "movd                    8(%0), %%mm3       \n\t"
1908         "add                       $12, %0          \n\t"
1909         "punpcklbw               %%mm7, %%mm1       \n\t"
1910         "punpcklbw               %%mm7, %%mm3       \n\t"
1911         "movq                    %%mm1, %%mm4       \n\t"
1912         "movq                    %%mm3, %%mm5       \n\t"
1913         "pmaddwd                    %4, %%mm1       \n\t"
1914         "pmaddwd                  8+%4, %%mm3       \n\t"
1915         "pmaddwd                 16+%4, %%mm4       \n\t"
1916         "pmaddwd                 %%mm6, %%mm5       \n\t"
1917         "paddd                   %%mm3, %%mm1       \n\t"
1918         "paddd                   %%mm5, %%mm4       \n\t"
1919
1920         "movq "MANGLE(ff_bgr24toUVOffset)", %%mm3       \n\t"
1921         "paddd                   %%mm3, %%mm0       \n\t"
1922         "paddd                   %%mm3, %%mm2       \n\t"
1923         "paddd                   %%mm3, %%mm1       \n\t"
1924         "paddd                   %%mm3, %%mm4       \n\t"
1925         "psrad                     $15, %%mm0       \n\t"
1926         "psrad                     $15, %%mm2       \n\t"
1927         "psrad                     $15, %%mm1       \n\t"
1928         "psrad                     $15, %%mm4       \n\t"
1929         "packssdw                %%mm1, %%mm0       \n\t"
1930         "packssdw                %%mm4, %%mm2       \n\t"
1931         "packuswb                %%mm0, %%mm0       \n\t"
1932         "packuswb                %%mm2, %%mm2       \n\t"
1933         "movd                %%mm0, (%1, %%"REG_a") \n\t"
1934         "movd                %%mm2, (%2, %%"REG_a") \n\t"
1935         "add                        $4, %%"REG_a"   \n\t"
1936         " js                        1b              \n\t"
1937     : "+r" (src)
1938     : "r" (dstU+width), "r" (dstV+width), "g" ((x86_reg)-width), "m"(ff_bgr24toUV[srcFormat == PIX_FMT_RGB24][0])
1939     : "%"REG_a
1940     );
1941 }
1942 #endif
1943
1944 static inline void RENAME(bgr24ToY)(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)
1945 {
1946 #if COMPILE_TEMPLATE_MMX
1947     RENAME(bgr24ToY_mmx)(dst, src, width, PIX_FMT_BGR24);
1948 #else
1949     int i;
1950     for (i=0; i<width; i++) {
1951         int b= src[i*3+0];
1952         int g= src[i*3+1];
1953         int r= src[i*3+2];
1954
1955         dst[i]= ((RY*r + GY*g + BY*b + (33<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);
1956     }
1957 #endif /* COMPILE_TEMPLATE_MMX */
1958 }
1959
1960 static inline void RENAME(bgr24ToUV)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src1, const uint8_t *src2, long width, uint32_t *unused)
1961 {
1962 #if COMPILE_TEMPLATE_MMX
1963     RENAME(bgr24ToUV_mmx)(dstU, dstV, src1, width, PIX_FMT_BGR24);
1964 #else
1965     int i;
1966     for (i=0; i<width; i++) {
1967         int b= src1[3*i + 0];
1968         int g= src1[3*i + 1];
1969         int r= src1[3*i + 2];
1970
1971         dstU[i]= (RU*r + GU*g + BU*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT;
1972         dstV[i]= (RV*r + GV*g + BV*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT;
1973     }
1974 #endif /* COMPILE_TEMPLATE_MMX */
1975     assert(src1 == src2);
1976 }
1977
1978 static inline void RENAME(bgr24ToUV_half)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src1, const uint8_t *src2, long width, uint32_t *unused)
1979 {
1980     int i;
1981     for (i=0; i<width; i++) {
1982         int b= src1[6*i + 0] + src1[6*i + 3];
1983         int g= src1[6*i + 1] + src1[6*i + 4];
1984         int r= src1[6*i + 2] + src1[6*i + 5];
1985
1986         dstU[i]= (RU*r + GU*g + BU*b + (257<<RGB2YUV_SHIFT))>>(RGB2YUV_SHIFT+1);
1987         dstV[i]= (RV*r + GV*g + BV*b + (257<<RGB2YUV_SHIFT))>>(RGB2YUV_SHIFT+1);
1988     }
1989     assert(src1 == src2);
1990 }
1991
1992 static inline void RENAME(rgb24ToY)(uint8_t *dst, const uint8_t *src, long width, uint32_t *unused)
1993 {
1994 #if COMPILE_TEMPLATE_MMX
1995     RENAME(bgr24ToY_mmx)(dst, src, width, PIX_FMT_RGB24);
1996 #else
1997     int i;
1998     for (i=0; i<width; i++) {
1999         int r= src[i*3+0];
2000         int g= src[i*3+1];
2001         int b= src[i*3+2];
2002
2003         dst[i]= ((RY*r + GY*g + BY*b + (33<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT);
2004     }
2005 #endif
2006 }
2007
2008 static inline void RENAME(rgb24ToUV)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src1, const uint8_t *src2, long width, uint32_t *unused)
2009 {
2010 #if COMPILE_TEMPLATE_MMX
2011     assert(src1==src2);
2012     RENAME(bgr24ToUV_mmx)(dstU, dstV, src1, width, PIX_FMT_RGB24);
2013 #else
2014     int i;
2015     assert(src1==src2);
2016     for (i=0; i<width; i++) {
2017         int r= src1[3*i + 0];
2018         int g= src1[3*i + 1];
2019         int b= src1[3*i + 2];
2020
2021         dstU[i]= (RU*r + GU*g + BU*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT;
2022         dstV[i]= (RV*r + GV*g + BV*b + (257<<(RGB2YUV_SHIFT-1)))>>RGB2YUV_SHIFT;
2023     }
2024 #endif
2025 }
2026
2027 static inline void RENAME(rgb24ToUV_half)(uint8_t *dstU, uint8_t *dstV, const uint8_t *src1, const uint8_t *src2, long width, uint32_t *unused)
2028 {
2029     int i;
2030     assert(src1==src2);
2031     for (i=0; i<width; i++) {
2032         int r= src1[6*i + 0] + src1[6*i + 3];
2033         int g= src1[6*i + 1] + src1[6*i + 4];
2034         int b= src1[6*i + 2] + src1[6*i + 5];
2035
2036         dstU[i]= (RU*r + GU*g + BU*b + (257<<RGB2YUV_SHIFT))>>(RGB2YUV_SHIFT+1);
2037         dstV[i]= (RV*r + GV*g + BV*b + (257<<RGB2YUV_SHIFT))>>(RGB2YUV_SHIFT+1);
2038     }
2039 }
2040
2041
2042 // bilinear / bicubic scaling
2043 static inline void RENAME(hScale)(int16_t *dst, int dstW, const uint8_t *src, int srcW, int xInc,
2044                                   const int16_t *filter, const int16_t *filterPos, long filterSize)
2045 {
2046 #if COMPILE_TEMPLATE_MMX
2047     assert(filterSize % 4 == 0 && filterSize>0);
2048     if (filterSize==4) { // Always true for upscaling, sometimes for down, too.
2049         x86_reg counter= -2*dstW;
2050         filter-= counter*2;
2051         filterPos-= counter/2;
2052         dst-= counter/2;
2053         __asm__ volatile(
2054 #if defined(PIC)
2055             "push            %%"REG_b"              \n\t"
2056 #endif
2057             "pxor                %%mm7, %%mm7       \n\t"
2058             "push           %%"REG_BP"              \n\t" // we use 7 regs here ...
2059             "mov             %%"REG_a", %%"REG_BP"  \n\t"
2060             ASMALIGN(4)
2061             "1:                                     \n\t"
2062             "movzwl   (%2, %%"REG_BP"), %%eax       \n\t"
2063             "movzwl  2(%2, %%"REG_BP"), %%ebx       \n\t"
2064             "movq  (%1, %%"REG_BP", 4), %%mm1       \n\t"
2065             "movq 8(%1, %%"REG_BP", 4), %%mm3       \n\t"
2066             "movd      (%3, %%"REG_a"), %%mm0       \n\t"
2067             "movd      (%3, %%"REG_b"), %%mm2       \n\t"
2068             "punpcklbw           %%mm7, %%mm0       \n\t"
2069             "punpcklbw           %%mm7, %%mm2       \n\t"
2070             "pmaddwd             %%mm1, %%mm0       \n\t"
2071             "pmaddwd             %%mm2, %%mm3       \n\t"
2072             "movq                %%mm0, %%mm4       \n\t"
2073             "punpckldq           %%mm3, %%mm0       \n\t"
2074             "punpckhdq           %%mm3, %%mm4       \n\t"
2075             "paddd               %%mm4, %%mm0       \n\t"
2076             "psrad                  $7, %%mm0       \n\t"
2077             "packssdw            %%mm0, %%mm0       \n\t"
2078             "movd                %%mm0, (%4, %%"REG_BP")    \n\t"
2079             "add                    $4, %%"REG_BP"  \n\t"
2080             " jnc                   1b              \n\t"
2081
2082             "pop            %%"REG_BP"              \n\t"
2083 #if defined(PIC)
2084             "pop             %%"REG_b"              \n\t"
2085 #endif
2086             : "+a" (counter)
2087             : "c" (filter), "d" (filterPos), "S" (src), "D" (dst)
2088 #if !defined(PIC)
2089             : "%"REG_b
2090 #endif
2091         );
2092     } else if (filterSize==8) {
2093         x86_reg counter= -2*dstW;
2094         filter-= counter*4;
2095         filterPos-= counter/2;
2096         dst-= counter/2;
2097         __asm__ volatile(
2098 #if defined(PIC)
2099             "push             %%"REG_b"             \n\t"
2100 #endif
2101             "pxor                 %%mm7, %%mm7      \n\t"
2102             "push            %%"REG_BP"             \n\t" // we use 7 regs here ...
2103             "mov              %%"REG_a", %%"REG_BP" \n\t"
2104             ASMALIGN(4)
2105             "1:                                     \n\t"
2106             "movzwl    (%2, %%"REG_BP"), %%eax      \n\t"
2107             "movzwl   2(%2, %%"REG_BP"), %%ebx      \n\t"
2108             "movq   (%1, %%"REG_BP", 8), %%mm1      \n\t"
2109             "movq 16(%1, %%"REG_BP", 8), %%mm3      \n\t"
2110             "movd       (%3, %%"REG_a"), %%mm0      \n\t"
2111             "movd       (%3, %%"REG_b"), %%mm2      \n\t"
2112             "punpcklbw            %%mm7, %%mm0      \n\t"
2113             "punpcklbw            %%mm7, %%mm2      \n\t"
2114             "pmaddwd              %%mm1, %%mm0      \n\t"
2115             "pmaddwd              %%mm2, %%mm3      \n\t"
2116
2117             "movq  8(%1, %%"REG_BP", 8), %%mm1      \n\t"
2118             "movq 24(%1, %%"REG_BP", 8), %%mm5      \n\t"
2119             "movd      4(%3, %%"REG_a"), %%mm4      \n\t"
2120             "movd      4(%3, %%"REG_b"), %%mm2      \n\t"
2121             "punpcklbw            %%mm7, %%mm4      \n\t"
2122             "punpcklbw            %%mm7, %%mm2      \n\t"
2123             "pmaddwd              %%mm1, %%mm4      \n\t"
2124             "pmaddwd              %%mm2, %%mm5      \n\t"
2125             "paddd                %%mm4, %%mm0      \n\t"
2126             "paddd                %%mm5, %%mm3      \n\t"
2127             "movq                 %%mm0, %%mm4      \n\t"
2128             "punpckldq            %%mm3, %%mm0      \n\t"
2129             "punpckhdq            %%mm3, %%mm4      \n\t"
2130             "paddd                %%mm4, %%mm0      \n\t"
2131             "psrad                   $7, %%mm0      \n\t"
2132             "packssdw             %%mm0, %%mm0      \n\t"
2133             "movd                 %%mm0, (%4, %%"REG_BP")   \n\t"
2134             "add                     $4, %%"REG_BP" \n\t"
2135             " jnc                    1b             \n\t"
2136
2137             "pop             %%"REG_BP"             \n\t"
2138 #if defined(PIC)
2139             "pop              %%"REG_b"             \n\t"
2140 #endif
2141             : "+a" (counter)
2142             : "c" (filter), "d" (filterPos), "S" (src), "D" (dst)
2143 #if !defined(PIC)
2144             : "%"REG_b
2145 #endif
2146         );
2147     } else {
2148         const uint8_t *offset = src+filterSize;
2149         x86_reg counter= -2*dstW;
2150         //filter-= counter*filterSize/2;
2151         filterPos-= counter/2;
2152         dst-= counter/2;
2153         __asm__ volatile(
2154             "pxor                  %%mm7, %%mm7     \n\t"
2155             ASMALIGN(4)
2156             "1:                                     \n\t"
2157             "mov                      %2, %%"REG_c" \n\t"
2158             "movzwl      (%%"REG_c", %0), %%eax     \n\t"
2159             "movzwl     2(%%"REG_c", %0), %%edx     \n\t"
2160             "mov                      %5, %%"REG_c" \n\t"
2161             "pxor                  %%mm4, %%mm4     \n\t"
2162             "pxor                  %%mm5, %%mm5     \n\t"
2163             "2:                                     \n\t"
2164             "movq                   (%1), %%mm1     \n\t"
2165             "movq               (%1, %6), %%mm3     \n\t"
2166             "movd (%%"REG_c", %%"REG_a"), %%mm0     \n\t"
2167             "movd (%%"REG_c", %%"REG_d"), %%mm2     \n\t"
2168             "punpcklbw             %%mm7, %%mm0     \n\t"
2169             "punpcklbw             %%mm7, %%mm2     \n\t"
2170             "pmaddwd               %%mm1, %%mm0     \n\t"
2171             "pmaddwd               %%mm2, %%mm3     \n\t"
2172             "paddd                 %%mm3, %%mm5     \n\t"
2173             "paddd                 %%mm0, %%mm4     \n\t"
2174             "add                      $8, %1        \n\t"
2175             "add                      $4, %%"REG_c" \n\t"
2176             "cmp                      %4, %%"REG_c" \n\t"
2177             " jb                      2b            \n\t"
2178             "add                      %6, %1        \n\t"
2179             "movq                  %%mm4, %%mm0     \n\t"
2180             "punpckldq             %%mm5, %%mm4     \n\t"
2181             "punpckhdq             %%mm5, %%mm0     \n\t"
2182             "paddd                 %%mm0, %%mm4     \n\t"
2183             "psrad                    $7, %%mm4     \n\t"
2184             "packssdw              %%mm4, %%mm4     \n\t"
2185             "mov                      %3, %%"REG_a" \n\t"
2186             "movd                  %%mm4, (%%"REG_a", %0)   \n\t"
2187             "add                      $4, %0        \n\t"
2188             " jnc                     1b            \n\t"
2189
2190             : "+r" (counter), "+r" (filter)
2191             : "m" (filterPos), "m" (dst), "m"(offset),
2192             "m" (src), "r" ((x86_reg)filterSize*2)
2193             : "%"REG_a, "%"REG_c, "%"REG_d
2194         );
2195     }
2196 #else
2197 #if COMPILE_TEMPLATE_ALTIVEC
2198     hScale_altivec_real(dst, dstW, src, srcW, xInc, filter, filterPos, filterSize);
2199 #else
2200     int i;
2201     for (i=0; i<dstW; i++) {
2202         int j;
2203         int srcPos= filterPos[i];
2204         int val=0;
2205         //printf("filterPos: %d\n", filterPos[i]);
2206         for (j=0; j<filterSize; j++) {
2207             //printf("filter: %d, src: %d\n", filter[i], src[srcPos + j]);
2208             val += ((int)src[srcPos + j])*filter[filterSize*i + j];
2209         }
2210         //filter += hFilterSize;
2211         dst[i] = FFMIN(val>>7, (1<<15)-1); // the cubic equation does overflow ...
2212         //dst[i] = val>>7;
2213     }
2214 #endif /* COMPILE_ALTIVEC */
2215 #endif /* COMPILE_MMX */
2216 }
2217
2218 //FIXME all pal and rgb srcFormats could do this convertion as well
2219 //FIXME all scalers more complex than bilinear could do half of this transform
2220 static void RENAME(chrRangeToJpeg)(uint16_t *dst, int width)
2221 {
2222     int i;
2223     for (i = 0; i < width; i++) {
2224         dst[i     ] = (FFMIN(dst[i     ],30775)*4663 - 9289992)>>12; //-264
2225         dst[i+VOFW] = (FFMIN(dst[i+VOFW],30775)*4663 - 9289992)>>12; //-264
2226     }
2227 }
2228 static void RENAME(chrRangeFromJpeg)(uint16_t *dst, int width)
2229 {
2230     int i;
2231     for (i = 0; i < width; i++) {
2232         dst[i     ] = (dst[i     ]*1799 + 4081085)>>11; //1469
2233         dst[i+VOFW] = (dst[i+VOFW]*1799 + 4081085)>>11; //1469
2234     }
2235 }
2236 static void RENAME(lumRangeToJpeg)(uint16_t *dst, int width)
2237 {
2238     int i;
2239     for (i = 0; i < width; i++)
2240         dst[i] = (FFMIN(dst[i],30189)*19077 - 39057361)>>14;
2241 }
2242 static void RENAME(lumRangeFromJpeg)(uint16_t *dst, int width)
2243 {
2244     int i;
2245     for (i = 0; i < width; i++)
2246         dst[i] = (dst[i]*14071 + 33561947)>>14;
2247 }
2248
2249 #define FAST_BILINEAR_X86 \
2250     "subl    %%edi, %%esi    \n\t" /*  src[xx+1] - src[xx] */                   \
2251     "imull   %%ecx, %%esi    \n\t" /* (src[xx+1] - src[xx])*xalpha */           \
2252     "shll      $16, %%edi    \n\t"                                              \
2253     "addl    %%edi, %%esi    \n\t" /* src[xx+1]*xalpha + src[xx]*(1-xalpha) */  \
2254     "mov        %1, %%"REG_D"\n\t"                                              \
2255     "shrl       $9, %%esi    \n\t"                                              \
2256
2257 static inline void RENAME(hyscale_fast)(SwsContext *c, int16_t *dst,
2258                                         long dstWidth, const uint8_t *src, int srcW,
2259                                         int xInc)
2260 {
2261 #if ARCH_X86 && CONFIG_GPL
2262 #if COMPILE_TEMPLATE_MMX2
2263     int32_t *filterPos = c->hLumFilterPos;
2264     int16_t *filter    = c->hLumFilter;
2265     int     canMMX2BeUsed  = c->canMMX2BeUsed;
2266     void    *mmx2FilterCode= c->lumMmx2FilterCode;
2267     int i;
2268 #if defined(PIC)
2269     DECLARE_ALIGNED(8, uint64_t, ebxsave);
2270 #endif
2271     if (canMMX2BeUsed) {
2272         __asm__ volatile(
2273 #if defined(PIC)
2274             "mov               %%"REG_b", %5        \n\t"
2275 #endif
2276             "pxor                  %%mm7, %%mm7     \n\t"
2277             "mov                      %0, %%"REG_c" \n\t"
2278             "mov                      %1, %%"REG_D" \n\t"
2279             "mov                      %2, %%"REG_d" \n\t"
2280             "mov                      %3, %%"REG_b" \n\t"
2281             "xor               %%"REG_a", %%"REG_a" \n\t" // i
2282             PREFETCH"        (%%"REG_c")            \n\t"
2283             PREFETCH"      32(%%"REG_c")            \n\t"
2284             PREFETCH"      64(%%"REG_c")            \n\t"
2285
2286 #if ARCH_X86_64
2287
2288 #define CALL_MMX2_FILTER_CODE \
2289             "movl            (%%"REG_b"), %%esi     \n\t"\
2290             "call                    *%4            \n\t"\
2291             "movl (%%"REG_b", %%"REG_a"), %%esi     \n\t"\
2292             "add               %%"REG_S", %%"REG_c" \n\t"\
2293             "add               %%"REG_a", %%"REG_D" \n\t"\
2294             "xor               %%"REG_a", %%"REG_a" \n\t"\
2295
2296 #else
2297
2298 #define CALL_MMX2_FILTER_CODE \
2299             "movl (%%"REG_b"), %%esi        \n\t"\
2300             "call         *%4                       \n\t"\
2301             "addl (%%"REG_b", %%"REG_a"), %%"REG_c" \n\t"\
2302             "add               %%"REG_a", %%"REG_D" \n\t"\
2303             "xor               %%"REG_a", %%"REG_a" \n\t"\
2304
2305 #endif /* ARCH_X86_64 */
2306
2307             CALL_MMX2_FILTER_CODE
2308             CALL_MMX2_FILTER_CODE
2309             CALL_MMX2_FILTER_CODE
2310             CALL_MMX2_FILTER_CODE
2311             CALL_MMX2_FILTER_CODE
2312             CALL_MMX2_FILTER_CODE
2313             CALL_MMX2_FILTER_CODE
2314             CALL_MMX2_FILTER_CODE
2315
2316 #if defined(PIC)
2317             "mov                      %5, %%"REG_b" \n\t"
2318 #endif
2319             :: "m" (src), "m" (dst), "m" (filter), "m" (filterPos),
2320             "m" (mmx2FilterCode)
2321 #if defined(PIC)
2322             ,"m" (ebxsave)
2323 #endif
2324             : "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S, "%"REG_D
2325 #if !defined(PIC)
2326             ,"%"REG_b
2327 #endif
2328         );
2329         for (i=dstWidth-1; (i*xInc)>>16 >=srcW-1; i--) dst[i] = src[srcW-1]*128;
2330     } else {
2331 #endif /* COMPILE_TEMPLATE_MMX2 */
2332     x86_reg xInc_shr16 = xInc >> 16;
2333     uint16_t xInc_mask = xInc & 0xffff;
2334     //NO MMX just normal asm ...
2335     __asm__ volatile(
2336         "xor %%"REG_a", %%"REG_a"            \n\t" // i
2337         "xor %%"REG_d", %%"REG_d"            \n\t" // xx
2338         "xorl    %%ecx, %%ecx                \n\t" // xalpha
2339         ASMALIGN(4)
2340         "1:                                  \n\t"
2341         "movzbl    (%0, %%"REG_d"), %%edi    \n\t" //src[xx]
2342         "movzbl   1(%0, %%"REG_d"), %%esi    \n\t" //src[xx+1]
2343         FAST_BILINEAR_X86
2344         "movw     %%si, (%%"REG_D", %%"REG_a", 2)   \n\t"
2345         "addw       %4, %%cx                 \n\t" //xalpha += xInc&0xFFFF
2346         "adc        %3, %%"REG_d"            \n\t" //xx+= xInc>>16 + carry
2347
2348         "movzbl    (%0, %%"REG_d"), %%edi    \n\t" //src[xx]
2349         "movzbl   1(%0, %%"REG_d"), %%esi    \n\t" //src[xx+1]
2350         FAST_BILINEAR_X86
2351         "movw     %%si, 2(%%"REG_D", %%"REG_a", 2)  \n\t"
2352         "addw       %4, %%cx                 \n\t" //xalpha += xInc&0xFFFF
2353         "adc        %3, %%"REG_d"            \n\t" //xx+= xInc>>16 + carry
2354
2355
2356         "add        $2, %%"REG_a"            \n\t"
2357         "cmp        %2, %%"REG_a"            \n\t"
2358         " jb        1b                       \n\t"
2359
2360
2361         :: "r" (src), "m" (dst), "m" (dstWidth), "m" (xInc_shr16), "m" (xInc_mask)
2362         : "%"REG_a, "%"REG_d, "%ecx", "%"REG_D, "%esi"
2363     );
2364 #if COMPILE_TEMPLATE_MMX2
2365     } //if MMX2 can't be used
2366 #endif
2367 #else
2368     int i;
2369     unsigned int xpos=0;
2370     for (i=0;i<dstWidth;i++) {
2371         register unsigned int xx=xpos>>16;
2372         register unsigned int xalpha=(xpos&0xFFFF)>>9;
2373         dst[i]= (src[xx]<<7) + (src[xx+1] - src[xx])*xalpha;
2374         xpos+=xInc;
2375     }
2376 #endif /* ARCH_X86 */
2377 }
2378
2379       // *** horizontal scale Y line to temp buffer
2380 static inline void RENAME(hyscale)(SwsContext *c, uint16_t *dst, long dstWidth, const uint8_t *src, int srcW, int xInc,
2381                                    const int16_t *hLumFilter,
2382                                    const int16_t *hLumFilterPos, int hLumFilterSize,
2383                                    uint8_t *formatConvBuffer,
2384                                    uint32_t *pal, int isAlpha)
2385 {
2386     void (*toYV12)(uint8_t *, const uint8_t *, long, uint32_t *) = isAlpha ? c->alpToYV12 : c->lumToYV12;
2387     void (*convertRange)(uint16_t *, int) = isAlpha ? NULL : c->lumConvertRange;
2388
2389     src += isAlpha ? c->alpSrcOffset : c->lumSrcOffset;
2390
2391     if (toYV12) {
2392         toYV12(formatConvBuffer, src, srcW, pal);
2393         src= formatConvBuffer;
2394     }
2395
2396     if (!c->hyscale_fast) {
2397         c->hScale(dst, dstWidth, src, srcW, xInc, hLumFilter, hLumFilterPos, hLumFilterSize);
2398     } else { // fast bilinear upscale / crap downscale
2399         c->hyscale_fast(c, dst, dstWidth, src, srcW, xInc);
2400     }
2401
2402     if (convertRange)
2403         convertRange(dst, dstWidth);
2404 }
2405
2406 static inline void RENAME(hcscale_fast)(SwsContext *c, int16_t *dst,
2407                                         long dstWidth, const uint8_t *src1,
2408                                         const uint8_t *src2, int srcW, int xInc)
2409 {
2410 #if ARCH_X86 && CONFIG_GPL
2411 #if COMPILE_TEMPLATE_MMX2
2412     int32_t *filterPos = c->hChrFilterPos;
2413     int16_t *filter    = c->hChrFilter;
2414     int     canMMX2BeUsed  = c->canMMX2BeUsed;
2415     void    *mmx2FilterCode= c->chrMmx2FilterCode;
2416     int i;
2417 #if defined(PIC)
2418     DECLARE_ALIGNED(8, uint64_t, ebxsave);
2419 #endif
2420     if (canMMX2BeUsed) {
2421         __asm__ volatile(
2422 #if defined(PIC)
2423             "mov          %%"REG_b", %6         \n\t"
2424 #endif
2425             "pxor             %%mm7, %%mm7      \n\t"
2426             "mov                 %0, %%"REG_c"  \n\t"
2427             "mov                 %1, %%"REG_D"  \n\t"
2428             "mov                 %2, %%"REG_d"  \n\t"
2429             "mov                 %3, %%"REG_b"  \n\t"
2430             "xor          %%"REG_a", %%"REG_a"  \n\t" // i
2431             PREFETCH"   (%%"REG_c")             \n\t"
2432             PREFETCH" 32(%%"REG_c")             \n\t"
2433             PREFETCH" 64(%%"REG_c")             \n\t"
2434
2435             CALL_MMX2_FILTER_CODE
2436             CALL_MMX2_FILTER_CODE
2437             CALL_MMX2_FILTER_CODE
2438             CALL_MMX2_FILTER_CODE
2439             "xor          %%"REG_a", %%"REG_a"  \n\t" // i
2440             "mov                 %5, %%"REG_c"  \n\t" // src
2441             "mov                 %1, %%"REG_D"  \n\t" // buf1
2442             "add              $"AV_STRINGIFY(VOF)", %%"REG_D"  \n\t"
2443             PREFETCH"   (%%"REG_c")             \n\t"
2444             PREFETCH" 32(%%"REG_c")             \n\t"
2445             PREFETCH" 64(%%"REG_c")             \n\t"
2446
2447             CALL_MMX2_FILTER_CODE
2448             CALL_MMX2_FILTER_CODE
2449             CALL_MMX2_FILTER_CODE
2450             CALL_MMX2_FILTER_CODE
2451
2452 #if defined(PIC)
2453             "mov %6, %%"REG_b"    \n\t"
2454 #endif
2455             :: "m" (src1), "m" (dst), "m" (filter), "m" (filterPos),
2456             "m" (mmx2FilterCode), "m" (src2)
2457 #if defined(PIC)
2458             ,"m" (ebxsave)
2459 #endif
2460             : "%"REG_a, "%"REG_c, "%"REG_d, "%"REG_S, "%"REG_D
2461 #if !defined(PIC)
2462             ,"%"REG_b
2463 #endif
2464         );
2465         for (i=dstWidth-1; (i*xInc)>>16 >=srcW-1; i--) {
2466             //printf("%d %d %d\n", dstWidth, i, srcW);
2467             dst[i] = src1[srcW-1]*128;
2468             dst[i+VOFW] = src2[srcW-1]*128;
2469         }
2470     } else {
2471 #endif /* COMPILE_TEMPLATE_MMX2 */
2472         x86_reg xInc_shr16 = (x86_reg) (xInc >> 16);
2473         uint16_t xInc_mask = xInc & 0xffff;
2474         __asm__ volatile(
2475             "xor %%"REG_a", %%"REG_a"               \n\t" // i
2476             "xor %%"REG_d", %%"REG_d"               \n\t" // xx
2477             "xorl    %%ecx, %%ecx                   \n\t" // xalpha
2478             ASMALIGN(4)
2479             "1:                                     \n\t"
2480             "mov        %0, %%"REG_S"               \n\t"
2481             "movzbl  (%%"REG_S", %%"REG_d"), %%edi  \n\t" //src[xx]
2482             "movzbl 1(%%"REG_S", %%"REG_d"), %%esi  \n\t" //src[xx+1]
2483             FAST_BILINEAR_X86
2484             "movw     %%si, (%%"REG_D", %%"REG_a", 2)   \n\t"
2485
2486             "movzbl    (%5, %%"REG_d"), %%edi       \n\t" //src[xx]
2487             "movzbl   1(%5, %%"REG_d"), %%esi       \n\t" //src[xx+1]
2488             FAST_BILINEAR_X86
2489             "movw     %%si, "AV_STRINGIFY(VOF)"(%%"REG_D", %%"REG_a", 2)   \n\t"
2490
2491             "addw       %4, %%cx                    \n\t" //xalpha += xInc&0xFFFF
2492             "adc        %3, %%"REG_d"               \n\t" //xx+= xInc>>16 + carry
2493             "add        $1, %%"REG_a"               \n\t"
2494             "cmp        %2, %%"REG_a"               \n\t"
2495             " jb        1b                          \n\t"
2496
2497 /* GCC 3.3 makes MPlayer crash on IA-32 machines when using "g" operand here,
2498 which is needed to support GCC 4.0. */
2499 #if ARCH_X86_64 && AV_GCC_VERSION_AT_LEAST(3,4)
2500             :: "m" (src1), "m" (dst), "g" (dstWidth), "m" (xInc_shr16), "m" (xInc_mask),
2501 #else
2502             :: "m" (src1), "m" (dst), "m" (dstWidth), "m" (xInc_shr16), "m" (xInc_mask),
2503 #endif
2504             "r" (src2)
2505             : "%"REG_a, "%"REG_d, "%ecx", "%"REG_D, "%esi"
2506         );
2507 #if COMPILE_TEMPLATE_MMX2
2508     } //if MMX2 can't be used
2509 #endif
2510 #else
2511     int i;
2512     unsigned int xpos=0;
2513     for (i=0;i<dstWidth;i++) {
2514         register unsigned int xx=xpos>>16;
2515         register unsigned int xalpha=(xpos&0xFFFF)>>9;
2516         dst[i]=(src1[xx]*(xalpha^127)+src1[xx+1]*xalpha);
2517         dst[i+VOFW]=(src2[xx]*(xalpha^127)+src2[xx+1]*xalpha);
2518         /* slower
2519         dst[i]= (src1[xx]<<7) + (src1[xx+1] - src1[xx])*xalpha;
2520         dst[i+VOFW]=(src2[xx]<<7) + (src2[xx+1] - src2[xx])*xalpha;
2521         */
2522         xpos+=xInc;
2523     }
2524 #endif /* ARCH_X86 */
2525 }
2526
2527 inline static void RENAME(hcscale)(SwsContext *c, uint16_t *dst, long dstWidth, const uint8_t *src1, const uint8_t *src2,
2528                                    int srcW, int xInc, const int16_t *hChrFilter,
2529                                    const int16_t *hChrFilterPos, int hChrFilterSize,
2530                                    uint8_t *formatConvBuffer,
2531                                    uint32_t *pal)
2532 {
2533
2534     src1 += c->chrSrcOffset;
2535     src2 += c->chrSrcOffset;
2536
2537     if (c->chrToYV12) {
2538         c->chrToYV12(formatConvBuffer, formatConvBuffer+VOFW, src1, src2, srcW, pal);
2539         src1= formatConvBuffer;
2540         src2= formatConvBuffer+VOFW;
2541     }
2542
2543     if (!c->hcscale_fast) {
2544         c->hScale(dst     , dstWidth, src1, srcW, xInc, hChrFilter, hChrFilterPos, hChrFilterSize);
2545         c->hScale(dst+VOFW, dstWidth, src2, srcW, xInc, hChrFilter, hChrFilterPos, hChrFilterSize);
2546     } else { // fast bilinear upscale / crap downscale
2547         c->hcscale_fast(c, dst, dstWidth, src1, src2, srcW, xInc);
2548     }
2549
2550     if (c->chrConvertRange)
2551         c->chrConvertRange(dst, dstWidth);
2552 }
2553
2554 #define DEBUG_SWSCALE_BUFFERS 0
2555 #define DEBUG_BUFFERS(...) if (DEBUG_SWSCALE_BUFFERS) av_log(c, AV_LOG_DEBUG, __VA_ARGS__)
2556
2557 static int RENAME(swScale)(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY,
2558                            int srcSliceH, uint8_t* dst[], int dstStride[])
2559 {
2560     /* load a few things into local vars to make the code more readable? and faster */
2561     const int srcW= c->srcW;
2562     const int dstW= c->dstW;
2563     const int dstH= c->dstH;
2564     const int chrDstW= c->chrDstW;
2565     const int chrSrcW= c->chrSrcW;
2566     const int lumXInc= c->lumXInc;
2567     const int chrXInc= c->chrXInc;
2568     const enum PixelFormat dstFormat= c->dstFormat;
2569     const int flags= c->flags;
2570     int16_t *vLumFilterPos= c->vLumFilterPos;
2571     int16_t *vChrFilterPos= c->vChrFilterPos;
2572     int16_t *hLumFilterPos= c->hLumFilterPos;
2573     int16_t *hChrFilterPos= c->hChrFilterPos;
2574     int16_t *vLumFilter= c->vLumFilter;
2575     int16_t *vChrFilter= c->vChrFilter;
2576     int16_t *hLumFilter= c->hLumFilter;
2577     int16_t *hChrFilter= c->hChrFilter;
2578     int32_t *lumMmxFilter= c->lumMmxFilter;
2579     int32_t *chrMmxFilter= c->chrMmxFilter;
2580     int32_t av_unused *alpMmxFilter= c->alpMmxFilter;
2581     const int vLumFilterSize= c->vLumFilterSize;
2582     const int vChrFilterSize= c->vChrFilterSize;
2583     const int hLumFilterSize= c->hLumFilterSize;
2584     const int hChrFilterSize= c->hChrFilterSize;
2585     int16_t **lumPixBuf= c->lumPixBuf;
2586     int16_t **chrPixBuf= c->chrPixBuf;
2587     int16_t **alpPixBuf= c->alpPixBuf;
2588     const int vLumBufSize= c->vLumBufSize;
2589     const int vChrBufSize= c->vChrBufSize;
2590     uint8_t *formatConvBuffer= c->formatConvBuffer;
2591     const int chrSrcSliceY= srcSliceY >> c->chrSrcVSubSample;
2592     const int chrSrcSliceH= -((-srcSliceH) >> c->chrSrcVSubSample);
2593     int lastDstY;
2594     uint32_t *pal=c->pal_yuv;
2595
2596     /* vars which will change and which we need to store back in the context */
2597     int dstY= c->dstY;
2598     int lumBufIndex= c->lumBufIndex;
2599     int chrBufIndex= c->chrBufIndex;
2600     int lastInLumBuf= c->lastInLumBuf;
2601     int lastInChrBuf= c->lastInChrBuf;
2602
2603     if (isPacked(c->srcFormat)) {
2604         src[0]=
2605         src[1]=
2606         src[2]=
2607         src[3]= src[0];
2608         srcStride[0]=
2609         srcStride[1]=
2610         srcStride[2]=
2611         srcStride[3]= srcStride[0];
2612     }
2613     srcStride[1]<<= c->vChrDrop;
2614     srcStride[2]<<= c->vChrDrop;
2615
2616     DEBUG_BUFFERS("swScale() %p[%d] %p[%d] %p[%d] %p[%d] -> %p[%d] %p[%d] %p[%d] %p[%d]\n",
2617                   src[0], srcStride[0], src[1], srcStride[1], src[2], srcStride[2], src[3], srcStride[3],
2618                   dst[0], dstStride[0], dst[1], dstStride[1], dst[2], dstStride[2], dst[3], dstStride[3]);
2619     DEBUG_BUFFERS("srcSliceY: %d srcSliceH: %d dstY: %d dstH: %d\n",
2620                    srcSliceY,    srcSliceH,    dstY,    dstH);
2621     DEBUG_BUFFERS("vLumFilterSize: %d vLumBufSize: %d vChrFilterSize: %d vChrBufSize: %d\n",
2622                    vLumFilterSize,    vLumBufSize,    vChrFilterSize,    vChrBufSize);
2623
2624     if (dstStride[0]%8 !=0 || dstStride[1]%8 !=0 || dstStride[2]%8 !=0 || dstStride[3]%8 != 0) {
2625         static int warnedAlready=0; //FIXME move this into the context perhaps
2626         if (flags & SWS_PRINT_INFO && !warnedAlready) {
2627             av_log(c, AV_LOG_WARNING, "Warning: dstStride is not aligned!\n"
2628                    "         ->cannot do aligned memory accesses anymore\n");
2629             warnedAlready=1;
2630         }
2631     }
2632
2633     /* Note the user might start scaling the picture in the middle so this
2634        will not get executed. This is not really intended but works
2635        currently, so people might do it. */
2636     if (srcSliceY ==0) {
2637         lumBufIndex=-1;
2638         chrBufIndex=-1;
2639         dstY=0;
2640         lastInLumBuf= -1;
2641         lastInChrBuf= -1;
2642     }
2643
2644     lastDstY= dstY;
2645
2646     for (;dstY < dstH; dstY++) {
2647         unsigned char *dest =dst[0]+dstStride[0]*dstY;
2648         const int chrDstY= dstY>>c->chrDstVSubSample;
2649         unsigned char *uDest=dst[1]+dstStride[1]*chrDstY;
2650         unsigned char *vDest=dst[2]+dstStride[2]*chrDstY;
2651         unsigned char *aDest=(CONFIG_SWSCALE_ALPHA && alpPixBuf) ? dst[3]+dstStride[3]*dstY : NULL;
2652
2653         const int firstLumSrcY= vLumFilterPos[dstY]; //First line needed as input
2654         const int firstChrSrcY= vChrFilterPos[chrDstY]; //First line needed as input
2655         int lastLumSrcY= firstLumSrcY + vLumFilterSize -1; // Last line needed as input
2656         int lastChrSrcY= firstChrSrcY + vChrFilterSize -1; // Last line needed as input
2657         int enough_lines;
2658
2659         //handle holes (FAST_BILINEAR & weird filters)
2660         if (firstLumSrcY > lastInLumBuf) lastInLumBuf= firstLumSrcY-1;
2661         if (firstChrSrcY > lastInChrBuf) lastInChrBuf= firstChrSrcY-1;
2662         assert(firstLumSrcY >= lastInLumBuf - vLumBufSize + 1);
2663         assert(firstChrSrcY >= lastInChrBuf - vChrBufSize + 1);
2664
2665         DEBUG_BUFFERS("dstY: %d\n", dstY);
2666         DEBUG_BUFFERS("\tfirstLumSrcY: %d lastLumSrcY: %d lastInLumBuf: %d\n",
2667                          firstLumSrcY,    lastLumSrcY,    lastInLumBuf);
2668         DEBUG_BUFFERS("\tfirstChrSrcY: %d lastChrSrcY: %d lastInChrBuf: %d\n",
2669                          firstChrSrcY,    lastChrSrcY,    lastInChrBuf);
2670
2671         // Do we have enough lines in this slice to output the dstY line
2672         enough_lines = lastLumSrcY < srcSliceY + srcSliceH && lastChrSrcY < -((-srcSliceY - srcSliceH)>>c->chrSrcVSubSample);
2673         if (!enough_lines) {
2674             lastLumSrcY = srcSliceY + srcSliceH - 1;
2675             lastChrSrcY = chrSrcSliceY + chrSrcSliceH - 1;
2676             DEBUG_BUFFERS("buffering slice: lastLumSrcY %d lastChrSrcY %d\n",
2677                                             lastLumSrcY, lastChrSrcY);
2678         }
2679
2680         //Do horizontal scaling
2681         while(lastInLumBuf < lastLumSrcY) {
2682             const uint8_t *src1= src[0]+(lastInLumBuf + 1 - srcSliceY)*srcStride[0];
2683             const uint8_t *src2= src[3]+(lastInLumBuf + 1 - srcSliceY)*srcStride[3];
2684             lumBufIndex++;
2685             assert(lumBufIndex < 2*vLumBufSize);
2686             assert(lastInLumBuf + 1 - srcSliceY < srcSliceH);
2687             assert(lastInLumBuf + 1 - srcSliceY >= 0);
2688             RENAME(hyscale)(c, lumPixBuf[ lumBufIndex ], dstW, src1, srcW, lumXInc,
2689                             hLumFilter, hLumFilterPos, hLumFilterSize,
2690                             formatConvBuffer,
2691                             pal, 0);
2692             if (CONFIG_SWSCALE_ALPHA && alpPixBuf)
2693                 RENAME(hyscale)(c, alpPixBuf[ lumBufIndex ], dstW, src2, srcW, lumXInc,
2694                                 hLumFilter, hLumFilterPos, hLumFilterSize,
2695                                 formatConvBuffer,
2696                                 pal, 1);
2697             lastInLumBuf++;
2698             DEBUG_BUFFERS("\t\tlumBufIndex %d: lastInLumBuf: %d\n",
2699                                lumBufIndex,    lastInLumBuf);
2700         }
2701         while(lastInChrBuf < lastChrSrcY) {
2702             const uint8_t *src1= src[1]+(lastInChrBuf + 1 - chrSrcSliceY)*srcStride[1];
2703             const uint8_t *src2= src[2]+(lastInChrBuf + 1 - chrSrcSliceY)*srcStride[2];
2704             chrBufIndex++;
2705             assert(chrBufIndex < 2*vChrBufSize);
2706             assert(lastInChrBuf + 1 - chrSrcSliceY < (chrSrcSliceH));
2707             assert(lastInChrBuf + 1 - chrSrcSliceY >= 0);
2708             //FIXME replace parameters through context struct (some at least)
2709
2710             if (c->needs_hcscale)
2711                 RENAME(hcscale)(c, chrPixBuf[ chrBufIndex ], chrDstW, src1, src2, chrSrcW, chrXInc,
2712                                 hChrFilter, hChrFilterPos, hChrFilterSize,
2713                                 formatConvBuffer,
2714                                 pal);
2715             lastInChrBuf++;
2716             DEBUG_BUFFERS("\t\tchrBufIndex %d: lastInChrBuf: %d\n",
2717                                chrBufIndex,    lastInChrBuf);
2718         }
2719         //wrap buf index around to stay inside the ring buffer
2720         if (lumBufIndex >= vLumBufSize) lumBufIndex-= vLumBufSize;
2721         if (chrBufIndex >= vChrBufSize) chrBufIndex-= vChrBufSize;
2722         if (!enough_lines)
2723             break; //we can't output a dstY line so let's try with the next slice
2724
2725 #if COMPILE_TEMPLATE_MMX
2726         c->blueDither= ff_dither8[dstY&1];
2727         if (c->dstFormat == PIX_FMT_RGB555 || c->dstFormat == PIX_FMT_BGR555)
2728             c->greenDither= ff_dither8[dstY&1];
2729         else
2730             c->greenDither= ff_dither4[dstY&1];
2731         c->redDither= ff_dither8[(dstY+1)&1];
2732 #endif
2733         if (dstY < dstH-2) {
2734             const int16_t **lumSrcPtr= (const int16_t **) lumPixBuf + lumBufIndex + firstLumSrcY - lastInLumBuf + vLumBufSize;
2735             const int16_t **chrSrcPtr= (const int16_t **) chrPixBuf + chrBufIndex + firstChrSrcY - lastInChrBuf + vChrBufSize;
2736             const int16_t **alpSrcPtr= (CONFIG_SWSCALE_ALPHA && alpPixBuf) ? (const int16_t **) alpPixBuf + lumBufIndex + firstLumSrcY - lastInLumBuf + vLumBufSize : NULL;
2737 #if COMPILE_TEMPLATE_MMX
2738             int i;
2739             if (flags & SWS_ACCURATE_RND) {
2740                 int s= APCK_SIZE / 8;
2741                 for (i=0; i<vLumFilterSize; i+=2) {
2742                     *(const void**)&lumMmxFilter[s*i              ]= lumSrcPtr[i  ];
2743                     *(const void**)&lumMmxFilter[s*i+APCK_PTR2/4  ]= lumSrcPtr[i+(vLumFilterSize>1)];
2744                               lumMmxFilter[s*i+APCK_COEF/4  ]=
2745                               lumMmxFilter[s*i+APCK_COEF/4+1]= vLumFilter[dstY*vLumFilterSize + i    ]
2746                         + (vLumFilterSize>1 ? vLumFilter[dstY*vLumFilterSize + i + 1]<<16 : 0);
2747                     if (CONFIG_SWSCALE_ALPHA && alpPixBuf) {
2748                         *(const void**)&alpMmxFilter[s*i              ]= alpSrcPtr[i  ];
2749                         *(const void**)&alpMmxFilter[s*i+APCK_PTR2/4  ]= alpSrcPtr[i+(vLumFilterSize>1)];
2750                                   alpMmxFilter[s*i+APCK_COEF/4  ]=
2751                                   alpMmxFilter[s*i+APCK_COEF/4+1]= lumMmxFilter[s*i+APCK_COEF/4  ];
2752                     }
2753                 }
2754                 for (i=0; i<vChrFilterSize; i+=2) {
2755                     *(const void**)&chrMmxFilter[s*i              ]= chrSrcPtr[i  ];
2756                     *(const void**)&chrMmxFilter[s*i+APCK_PTR2/4  ]= chrSrcPtr[i+(vChrFilterSize>1)];
2757                               chrMmxFilter[s*i+APCK_COEF/4  ]=
2758                               chrMmxFilter[s*i+APCK_COEF/4+1]= vChrFilter[chrDstY*vChrFilterSize + i    ]
2759                         + (vChrFilterSize>1 ? vChrFilter[chrDstY*vChrFilterSize + i + 1]<<16 : 0);
2760                 }
2761             } else {
2762                 for (i=0; i<vLumFilterSize; i++) {
2763                     lumMmxFilter[4*i+0]= (int32_t)lumSrcPtr[i];
2764                     lumMmxFilter[4*i+1]= (uint64_t)lumSrcPtr[i] >> 32;
2765                     lumMmxFilter[4*i+2]=
2766                     lumMmxFilter[4*i+3]=
2767                         ((uint16_t)vLumFilter[dstY*vLumFilterSize + i])*0x10001;
2768                     if (CONFIG_SWSCALE_ALPHA && alpPixBuf) {
2769                         alpMmxFilter[4*i+0]= (int32_t)alpSrcPtr[i];
2770                         alpMmxFilter[4*i+1]= (uint64_t)alpSrcPtr[i] >> 32;
2771                         alpMmxFilter[4*i+2]=
2772                         alpMmxFilter[4*i+3]= lumMmxFilter[4*i+2];
2773                     }
2774                 }
2775                 for (i=0; i<vChrFilterSize; i++) {
2776                     chrMmxFilter[4*i+0]= (int32_t)chrSrcPtr[i];
2777                     chrMmxFilter[4*i+1]= (uint64_t)chrSrcPtr[i] >> 32;
2778                     chrMmxFilter[4*i+2]=
2779                     chrMmxFilter[4*i+3]=
2780                         ((uint16_t)vChrFilter[chrDstY*vChrFilterSize + i])*0x10001;
2781                 }
2782             }
2783 #endif
2784             if (dstFormat == PIX_FMT_NV12 || dstFormat == PIX_FMT_NV21) {
2785                 const int chrSkipMask= (1<<c->chrDstVSubSample)-1;
2786                 if (dstY&chrSkipMask) uDest= NULL; //FIXME split functions in lumi / chromi
2787                 c->yuv2nv12X(c,
2788                              vLumFilter+dstY*vLumFilterSize   , lumSrcPtr, vLumFilterSize,
2789                              vChrFilter+chrDstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
2790                              dest, uDest, dstW, chrDstW, dstFormat);
2791             } else if (isPlanarYUV(dstFormat) || dstFormat==PIX_FMT_GRAY8) { //YV12 like
2792                 const int chrSkipMask= (1<<c->chrDstVSubSample)-1;
2793                 if ((dstY&chrSkipMask) || isGray(dstFormat)) uDest=vDest= NULL; //FIXME split functions in lumi / chromi
2794                 if (is16BPS(dstFormat)) {
2795                     yuv2yuvX16inC(
2796                                   vLumFilter+dstY*vLumFilterSize   , lumSrcPtr, vLumFilterSize,
2797                                   vChrFilter+chrDstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
2798                                   alpSrcPtr, (uint16_t *) dest, (uint16_t *) uDest, (uint16_t *) vDest, (uint16_t *) aDest, dstW, chrDstW,
2799                                   dstFormat);
2800                 } else if (vLumFilterSize == 1 && vChrFilterSize == 1) { // unscaled YV12
2801                     const int16_t *lumBuf = lumSrcPtr[0];
2802                     const int16_t *chrBuf= chrSrcPtr[0];
2803                     const int16_t *alpBuf= (CONFIG_SWSCALE_ALPHA && alpPixBuf) ? alpSrcPtr[0] : NULL;
2804                     c->yuv2yuv1(c, lumBuf, chrBuf, alpBuf, dest, uDest, vDest, aDest, dstW, chrDstW);
2805                 } else { //General YV12
2806                     c->yuv2yuvX(c,
2807                                 vLumFilter+dstY*vLumFilterSize   , lumSrcPtr, vLumFilterSize,
2808                                 vChrFilter+chrDstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
2809                                 alpSrcPtr, dest, uDest, vDest, aDest, dstW, chrDstW);
2810                 }
2811             } else {
2812                 assert(lumSrcPtr + vLumFilterSize - 1 < lumPixBuf + vLumBufSize*2);
2813                 assert(chrSrcPtr + vChrFilterSize - 1 < chrPixBuf + vChrBufSize*2);
2814                 if (vLumFilterSize == 1 && vChrFilterSize == 2) { //unscaled RGB
2815                     int chrAlpha= vChrFilter[2*dstY+1];
2816                     if(flags & SWS_FULL_CHR_H_INT) {
2817                         yuv2rgbXinC_full(c, //FIXME write a packed1_full function
2818                                          vLumFilter+dstY*vLumFilterSize, lumSrcPtr, vLumFilterSize,
2819                                          vChrFilter+dstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
2820                                          alpSrcPtr, dest, dstW, dstY);
2821                     } else {
2822                         c->yuv2packed1(c, *lumSrcPtr, *chrSrcPtr, *(chrSrcPtr+1),
2823                                        alpPixBuf ? *alpSrcPtr : NULL,
2824                                        dest, dstW, chrAlpha, dstFormat, flags, dstY);
2825                     }
2826                 } else if (vLumFilterSize == 2 && vChrFilterSize == 2) { //bilinear upscale RGB
2827                     int lumAlpha= vLumFilter[2*dstY+1];
2828                     int chrAlpha= vChrFilter[2*dstY+1];
2829                     lumMmxFilter[2]=
2830                     lumMmxFilter[3]= vLumFilter[2*dstY   ]*0x10001;
2831                     chrMmxFilter[2]=
2832                     chrMmxFilter[3]= vChrFilter[2*chrDstY]*0x10001;
2833                     if(flags & SWS_FULL_CHR_H_INT) {
2834                         yuv2rgbXinC_full(c, //FIXME write a packed2_full function
2835                                          vLumFilter+dstY*vLumFilterSize, lumSrcPtr, vLumFilterSize,
2836                                          vChrFilter+dstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
2837                                          alpSrcPtr, dest, dstW, dstY);
2838                     } else {
2839                         c->yuv2packed2(c, *lumSrcPtr, *(lumSrcPtr+1), *chrSrcPtr, *(chrSrcPtr+1),
2840                                        alpPixBuf ? *alpSrcPtr : NULL, alpPixBuf ? *(alpSrcPtr+1) : NULL,
2841                                        dest, dstW, lumAlpha, chrAlpha, dstY);
2842                     }
2843                 } else { //general RGB
2844                     if(flags & SWS_FULL_CHR_H_INT) {
2845                         yuv2rgbXinC_full(c,
2846                                          vLumFilter+dstY*vLumFilterSize, lumSrcPtr, vLumFilterSize,
2847                                          vChrFilter+dstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
2848                                          alpSrcPtr, dest, dstW, dstY);
2849                     } else {
2850                         c->yuv2packedX(c,
2851                                        vLumFilter+dstY*vLumFilterSize, lumSrcPtr, vLumFilterSize,
2852                                        vChrFilter+dstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
2853                                        alpSrcPtr, dest, dstW, dstY);
2854                     }
2855                 }
2856             }
2857         } else { // hmm looks like we can't use MMX here without overwriting this array's tail
2858             const int16_t **lumSrcPtr= (const int16_t **)lumPixBuf + lumBufIndex + firstLumSrcY - lastInLumBuf + vLumBufSize;
2859             const int16_t **chrSrcPtr= (const int16_t **)chrPixBuf + chrBufIndex + firstChrSrcY - lastInChrBuf + vChrBufSize;
2860             const int16_t **alpSrcPtr= (CONFIG_SWSCALE_ALPHA && alpPixBuf) ? (const int16_t **)alpPixBuf + lumBufIndex + firstLumSrcY - lastInLumBuf + vLumBufSize : NULL;
2861             if (dstFormat == PIX_FMT_NV12 || dstFormat == PIX_FMT_NV21) {
2862                 const int chrSkipMask= (1<<c->chrDstVSubSample)-1;
2863                 if (dstY&chrSkipMask) uDest= NULL; //FIXME split functions in lumi / chromi
2864                 yuv2nv12XinC(
2865                              vLumFilter+dstY*vLumFilterSize   , lumSrcPtr, vLumFilterSize,
2866                              vChrFilter+chrDstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
2867                              dest, uDest, dstW, chrDstW, dstFormat);
2868             } else if (isPlanarYUV(dstFormat) || dstFormat==PIX_FMT_GRAY8) { //YV12
2869                 const int chrSkipMask= (1<<c->chrDstVSubSample)-1;
2870                 if ((dstY&chrSkipMask) || isGray(dstFormat)) uDest=vDest= NULL; //FIXME split functions in lumi / chromi
2871                 if (is16BPS(dstFormat)) {
2872                     yuv2yuvX16inC(
2873                                   vLumFilter+dstY*vLumFilterSize   , lumSrcPtr, vLumFilterSize,
2874                                   vChrFilter+chrDstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
2875                                   alpSrcPtr, (uint16_t *) dest, (uint16_t *) uDest, (uint16_t *) vDest, (uint16_t *) aDest, dstW, chrDstW,
2876                                   dstFormat);
2877                 } else {
2878                     yuv2yuvXinC(
2879                                 vLumFilter+dstY*vLumFilterSize   , lumSrcPtr, vLumFilterSize,
2880                                 vChrFilter+chrDstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
2881                                 alpSrcPtr, dest, uDest, vDest, aDest, dstW, chrDstW);
2882                 }
2883             } else {
2884                 assert(lumSrcPtr + vLumFilterSize - 1 < lumPixBuf + vLumBufSize*2);
2885                 assert(chrSrcPtr + vChrFilterSize - 1 < chrPixBuf + vChrBufSize*2);
2886                 if(flags & SWS_FULL_CHR_H_INT) {
2887                     yuv2rgbXinC_full(c,
2888                                      vLumFilter+dstY*vLumFilterSize, lumSrcPtr, vLumFilterSize,
2889                                      vChrFilter+dstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
2890                                      alpSrcPtr, dest, dstW, dstY);
2891                 } else {
2892                     yuv2packedXinC(c,
2893                                    vLumFilter+dstY*vLumFilterSize, lumSrcPtr, vLumFilterSize,
2894                                    vChrFilter+dstY*vChrFilterSize, chrSrcPtr, vChrFilterSize,
2895                                    alpSrcPtr, dest, dstW, dstY);
2896                 }
2897             }
2898         }
2899     }
2900
2901     if ((dstFormat == PIX_FMT_YUVA420P) && !alpPixBuf)
2902         fillPlane(dst[3], dstStride[3], dstW, dstY-lastDstY, lastDstY, 255);
2903
2904 #if COMPILE_TEMPLATE_MMX
2905     if (flags & SWS_CPU_CAPS_MMX2 )  __asm__ volatile("sfence":::"memory");
2906     /* On K6 femms is faster than emms. On K7 femms is directly mapped to emms. */
2907     if (flags & SWS_CPU_CAPS_3DNOW)  __asm__ volatile("femms" :::"memory");
2908     else                             __asm__ volatile("emms"  :::"memory");
2909 #endif
2910     /* store changed local vars back in the context */
2911     c->dstY= dstY;
2912     c->lumBufIndex= lumBufIndex;
2913     c->chrBufIndex= chrBufIndex;
2914     c->lastInLumBuf= lastInLumBuf;
2915     c->lastInChrBuf= lastInChrBuf;
2916
2917     return dstY - lastDstY;
2918 }
2919
2920 static void RENAME(sws_init_swScale)(SwsContext *c)
2921 {
2922     enum PixelFormat srcFormat = c->srcFormat;
2923
2924     c->yuv2nv12X    = RENAME(yuv2nv12X   );
2925     c->yuv2yuv1     = RENAME(yuv2yuv1    );
2926     c->yuv2yuvX     = RENAME(yuv2yuvX    );
2927     c->yuv2packed1  = RENAME(yuv2packed1 );
2928     c->yuv2packed2  = RENAME(yuv2packed2 );
2929     c->yuv2packedX  = RENAME(yuv2packedX );
2930
2931     c->hScale       = RENAME(hScale      );
2932
2933 #if COMPILE_TEMPLATE_MMX
2934     // Use the new MMX scaler if the MMX2 one can't be used (it is faster than the x86 ASM one).
2935     if (c->flags & SWS_FAST_BILINEAR && c->canMMX2BeUsed)
2936 #else
2937     if (c->flags & SWS_FAST_BILINEAR)
2938 #endif
2939     {
2940         c->hyscale_fast = RENAME(hyscale_fast);
2941         c->hcscale_fast = RENAME(hcscale_fast);
2942     }
2943
2944     c->chrToYV12 = NULL;
2945     switch(srcFormat) {
2946         case PIX_FMT_YUYV422  : c->chrToYV12 = RENAME(yuy2ToUV); break;
2947         case PIX_FMT_UYVY422  : c->chrToYV12 = RENAME(uyvyToUV); break;
2948         case PIX_FMT_NV12     : c->chrToYV12 = RENAME(nv12ToUV); break;
2949         case PIX_FMT_NV21     : c->chrToYV12 = RENAME(nv21ToUV); break;
2950         case PIX_FMT_RGB8     :
2951         case PIX_FMT_BGR8     :
2952         case PIX_FMT_PAL8     :
2953         case PIX_FMT_BGR4_BYTE:
2954         case PIX_FMT_RGB4_BYTE: c->chrToYV12 = palToUV; break;
2955         case PIX_FMT_YUV420P16BE:
2956         case PIX_FMT_YUV422P16BE:
2957         case PIX_FMT_YUV444P16BE: c->chrToYV12 = RENAME(BEToUV); break;
2958         case PIX_FMT_YUV420P16LE:
2959         case PIX_FMT_YUV422P16LE:
2960         case PIX_FMT_YUV444P16LE: c->chrToYV12 = RENAME(LEToUV); break;
2961     }
2962     if (c->chrSrcHSubSample) {
2963         switch(srcFormat) {
2964         case PIX_FMT_RGB48BE:
2965         case PIX_FMT_RGB48LE: c->chrToYV12 = rgb48ToUV_half; break;
2966         case PIX_FMT_RGB32  :
2967         case PIX_FMT_RGB32_1: c->chrToYV12 = bgr32ToUV_half; break;
2968         case PIX_FMT_BGR24  : c->chrToYV12 = RENAME(bgr24ToUV_half); break;
2969         case PIX_FMT_BGR565 : c->chrToYV12 = bgr16ToUV_half; break;
2970         case PIX_FMT_BGR555 : c->chrToYV12 = bgr15ToUV_half; break;
2971         case PIX_FMT_BGR32  :
2972         case PIX_FMT_BGR32_1: c->chrToYV12 = rgb32ToUV_half; break;
2973         case PIX_FMT_RGB24  : c->chrToYV12 = RENAME(rgb24ToUV_half); break;
2974         case PIX_FMT_RGB565 : c->chrToYV12 = rgb16ToUV_half; break;
2975         case PIX_FMT_RGB555 : c->chrToYV12 = rgb15ToUV_half; break;
2976         }
2977     } else {
2978         switch(srcFormat) {
2979         case PIX_FMT_RGB48BE:
2980         case PIX_FMT_RGB48LE: c->chrToYV12 = rgb48ToUV; break;
2981         case PIX_FMT_RGB32  :
2982         case PIX_FMT_RGB32_1: c->chrToYV12 = bgr32ToUV; break;
2983         case PIX_FMT_BGR24  : c->chrToYV12 = RENAME(bgr24ToUV); break;
2984         case PIX_FMT_BGR565 : c->chrToYV12 = bgr16ToUV; break;
2985         case PIX_FMT_BGR555 : c->chrToYV12 = bgr15ToUV; break;
2986         case PIX_FMT_BGR32  :
2987         case PIX_FMT_BGR32_1: c->chrToYV12 = rgb32ToUV; break;
2988         case PIX_FMT_RGB24  : c->chrToYV12 = RENAME(rgb24ToUV); break;
2989         case PIX_FMT_RGB565 : c->chrToYV12 = rgb16ToUV; break;
2990         case PIX_FMT_RGB555 : c->chrToYV12 = rgb15ToUV; break;
2991         }
2992     }
2993
2994     c->lumToYV12 = NULL;
2995     c->alpToYV12 = NULL;
2996     switch (srcFormat) {
2997     case PIX_FMT_YUYV422  :
2998     case PIX_FMT_YUV420P16BE:
2999     case PIX_FMT_YUV422P16BE:
3000     case PIX_FMT_YUV444P16BE:
3001     case PIX_FMT_GRAY16BE : c->lumToYV12 = RENAME(yuy2ToY); break;
3002     case PIX_FMT_UYVY422  :
3003     case PIX_FMT_YUV420P16LE:
3004     case PIX_FMT_YUV422P16LE:
3005     case PIX_FMT_YUV444P16LE:
3006     case PIX_FMT_GRAY16LE : c->lumToYV12 = RENAME(uyvyToY); break;
3007     case PIX_FMT_BGR24    : c->lumToYV12 = RENAME(bgr24ToY); break;
3008     case PIX_FMT_BGR565   : c->lumToYV12 = bgr16ToY; break;
3009     case PIX_FMT_BGR555   : c->lumToYV12 = bgr15ToY; break;
3010     case PIX_FMT_RGB24    : c->lumToYV12 = RENAME(rgb24ToY); break;
3011     case PIX_FMT_RGB565   : c->lumToYV12 = rgb16ToY; break;
3012     case PIX_FMT_RGB555   : c->lumToYV12 = rgb15ToY; break;
3013     case PIX_FMT_RGB8     :
3014     case PIX_FMT_BGR8     :
3015     case PIX_FMT_PAL8     :
3016     case PIX_FMT_BGR4_BYTE:
3017     case PIX_FMT_RGB4_BYTE: c->lumToYV12 = palToY; break;
3018     case PIX_FMT_MONOBLACK: c->lumToYV12 = monoblack2Y; break;
3019     case PIX_FMT_MONOWHITE: c->lumToYV12 = monowhite2Y; break;
3020     case PIX_FMT_RGB32  :
3021     case PIX_FMT_RGB32_1: c->lumToYV12 = bgr32ToY; break;
3022     case PIX_FMT_BGR32  :
3023     case PIX_FMT_BGR32_1: c->lumToYV12 = rgb32ToY; break;
3024     case PIX_FMT_RGB48BE:
3025     case PIX_FMT_RGB48LE: c->lumToYV12 = rgb48ToY; break;
3026     }
3027     if (c->alpPixBuf) {
3028         switch (srcFormat) {
3029         case PIX_FMT_RGB32  :
3030         case PIX_FMT_RGB32_1:
3031         case PIX_FMT_BGR32  :
3032         case PIX_FMT_BGR32_1: c->alpToYV12 = abgrToA; break;
3033         }
3034     }
3035
3036     switch (srcFormat) {
3037     case PIX_FMT_RGB32  :
3038     case PIX_FMT_BGR32  :
3039         c->alpSrcOffset = 3;
3040         break;
3041     case PIX_FMT_RGB32_1:
3042     case PIX_FMT_BGR32_1:
3043         c->lumSrcOffset = ALT32_CORR;
3044         c->chrSrcOffset = ALT32_CORR;
3045         break;
3046     case PIX_FMT_RGB48LE:
3047         c->lumSrcOffset = 1;
3048         c->chrSrcOffset = 1;
3049         c->alpSrcOffset = 1;
3050         break;
3051     }
3052
3053     if (c->srcRange != c->dstRange && !isAnyRGB(c->dstFormat)) {
3054         if (c->srcRange) {
3055             c->lumConvertRange = RENAME(lumRangeFromJpeg);
3056             c->chrConvertRange = RENAME(chrRangeFromJpeg);
3057         } else {
3058             c->lumConvertRange = RENAME(lumRangeToJpeg);
3059             c->chrConvertRange = RENAME(chrRangeToJpeg);
3060         }
3061     }
3062
3063     if (!(isGray(srcFormat) || isGray(c->dstFormat) ||
3064           srcFormat == PIX_FMT_MONOBLACK || srcFormat == PIX_FMT_MONOWHITE))
3065         c->needs_hcscale = 1;
3066 }