sources/x264/common-ppc-dct.c.patch

   1 --- common/ppc/dct.c.orig       2007-02-01 20:50:54.000000000 -0800
   2 +++ common/ppc/dct.c    2007-02-02 11:29:45.000000000 -0800
   3 @@ -294,9 +294,9 @@
   4      vec_u8_t hv = vec_ld( 0, dest );                           \
   5      vec_u8_t lv = vec_ld( 7, dest );                           \
   6      vec_u8_t dstv   = vec_perm( hv, lv, (vec_u8_t)perm_ldv );  \
   7 -    vec_s16_t idct_sh6 = vec_sra(idctv, sixv);                 \
   8 -    vec_u16_t dst16 = vec_mergeh(zero_u8v, dstv);              \
   9 -    vec_s16_t idstsum = vec_adds(idct_sh6, (vec_s16_t)dst16);  \
  10 +    vec_s16_t idct_sh6 = (vec_s16_t) vec_sra(idctv, sixv);                 \
  11 +    vec_u16_t dst16 = (vec_u16_t) vec_mergeh(zero_u8v, dstv);              \
  12 +    vec_s16_t idstsum = (vec_s16_t) vec_adds(idct_sh6, (vec_s16_t)dst16);  \
  13      vec_u8_t idstsum8 = vec_packsu(zero_s16v, idstsum);        \
  14      /* unaligned store */                                      \
  15      vec_u8_t bodyv  = vec_perm( idstsum8, idstsum8, perm_stv );\
  16 @@ -311,8 +311,8 @@
  17
  18  void x264_add8x8_idct8_altivec( uint8_t *dst, int16_t dct[8][8] )
  19  {
  20 -    vec_u16_t onev = vec_splat_s16(1);
  21 -    vec_u16_t twov = vec_splat_s16(2);
  22 +    vec_u16_t onev = (vec_u16_t) vec_splat_s16(1);
  23 +    vec_u16_t twov = (vec_u16_t) vec_splat_s16(2);
  24
  25      dct[0][0] += 32; // rounding for the >>6 at the end
  26
  27 @@ -341,7 +341,7 @@
  28
  29      vec_u8_t perm_ldv = vec_lvsl(0, dst);
  30      vec_u8_t perm_stv = vec_lvsr(8, dst);
  31 -    vec_u16_t sixv = vec_splat_s16(6);
  32 +    vec_u16_t sixv = (vec_u16_t) vec_splat_s16(6);
  33      const vec_u8_t sel = (vec_u8_t) CV(0,0,0,0,0,0,0,0,-1,-1,-1,-1,-1,-1,-1,-1);
  34      LOAD_ZERO;
  35