ffmpeg roll of source to mar 9 version...

patched-ffmpeg-mt/libavcodec/x86/vc1dsp_mmx.c

 /*
  * VC-1 and WMV3 - DSP functions MMX-optimized
  * Copyright (c) 2007 Christophe GISQUET <christophe.gisquet@free.fr>
  *
  * Permission is hereby granted, free of charge, to any person
  * obtaining a copy of this software and associated documentation
  * files (the "Software"), to deal in the Software without
  * restriction, including without limitation the rights to use,
  * copy, modify, merge, publish, distribute, sublicense, and/or sell
  * copies of the Software, and to permit persons to whom the
@@ skipping 55 matching lines @@
     "punpcklbw %%mm0, %%mm"#R0"        \n\t"    \
     "movd      (%0,%2), %%mm"#R3"      \n\t"    \
     "psubw     %%mm"#R0", %%mm"#R1"    \n\t"    \
     "punpcklbw %%mm0, %%mm"#R3"        \n\t"    \
     "paddw     %%mm7, %%mm"#R1"        \n\t"    \
     "psubw     %%mm"#R3", %%mm"#R1"    \n\t"    \
     "psraw     %4, %%mm"#R1"           \n\t"    \
     "movq      %%mm"#R1", "#OFF"(%1)   \n\t"    \
     "add       %2, %0                  \n\t"
 
-DECLARE_ALIGNED_16(const uint64_t, ff_pw_9) = 0x0009000900090009ULL;
+DECLARE_ALIGNED(16, const uint64_t, ff_pw_9) = 0x0009000900090009ULL;
 
 /** Sacrifying mm6 allows to pipeline loads from src */
 static void vc1_put_ver_16b_shift2_mmx(int16_t *dst,
                                        const uint8_t *src, x86_reg stride,
                                        int rnd, int64_t shift)
 {
     __asm__ volatile(
         "mov       $3, %%"REG_c"           \n\t"
         LOAD_ROUNDER_MMX("%5")
         "movq      "MANGLE(ff_pw_9)", %%mm6 \n\t"
@@ skipping 348 matching lines @@
     __asm__ volatile(\
         "pxor %%mm0, %%mm0         \n\t"\
         ::: "memory"\
     );\
 \
     if (vmode) { /* Vertical filter to apply */\
         if (hmode) { /* Horizontal filter to apply, output to tmp */\
             static const int shift_value[] = { 0, 5, 1, 5 };\
             int              shift = (shift_value[hmode]+shift_value[vmode])>>1;\
             int              r;\
-            DECLARE_ALIGNED_16(int16_t, tmp)[12*8];\
+            DECLARE_ALIGNED(16, int16_t, tmp)[12*8];\
 \
             r = (1<<(shift-1)) + rnd-1;\
             vc1_put_shift_ver_16bits[vmode](tmp, src-1, stride, r, shift);\
 \
             vc1_put_shift_hor_16bits[hmode](dst, stride, tmp+1, 64-rnd);\
             return;\
         }\
         else { /* No horizontal filter, output 8 lines to dst */\
             vc1_put_shift_8bits[vmode](dst, src, stride, 1-rnd, stride);\
             return;\
         }\
     }\
 \
     /* Horizontal mode with no vertical mode */\
     vc1_put_shift_8bits[hmode](dst, src, stride, rnd, 1);\
 }
 
 VC1_MSPEL_MC(put_)
 VC1_MSPEL_MC(avg_)
 
-void ff_put_vc1_mspel_mc00_mmx(uint8_t *dst, const uint8_t *src, int stride, int rnd);
-void ff_avg_vc1_mspel_mc00_mmx2(uint8_t *dst, const uint8_t *src, int stride, int rnd);
-
 /** Macro to ease bicubic filter interpolation functions declarations */
 #define DECLARE_FUNCTION(a, b)                                          \
 static void put_vc1_mspel_mc ## a ## b ## _mmx(uint8_t *dst, const uint8_t *src, int stride, int rnd) { \
      put_vc1_mspel_mc(dst, src, stride, a, b, rnd);                     \
 }\
 static void avg_vc1_mspel_mc ## a ## b ## _mmx2(uint8_t *dst, const uint8_t *src, int stride, int rnd) { \
      avg_vc1_mspel_mc(dst, src, stride, a, b, rnd);                     \
 }
 
 DECLARE_FUNCTION(0, 1)
@@ skipping 256 matching lines @@
         dsp->avg_vc1_mspel_pixels_tab[ 7] = avg_vc1_mspel_mc31_mmx2;
         dsp->avg_vc1_mspel_pixels_tab[11] = avg_vc1_mspel_mc32_mmx2;
         dsp->avg_vc1_mspel_pixels_tab[15] = avg_vc1_mspel_mc33_mmx2;
 
         dsp->vc1_inv_trans_8x8_dc = vc1_inv_trans_8x8_dc_mmx2;
         dsp->vc1_inv_trans_4x8_dc = vc1_inv_trans_4x8_dc_mmx2;
         dsp->vc1_inv_trans_8x4_dc = vc1_inv_trans_8x4_dc_mmx2;
         dsp->vc1_inv_trans_4x4_dc = vc1_inv_trans_4x4_dc_mmx2;
     }
 }