diff options
| author | David Conrad <lessen42@gmail.com> | 2010-01-20 00:34:10 +0000 |
|---|---|---|
| committer | David Conrad <lessen42@gmail.com> | 2010-01-20 00:34:10 +0000 |
| commit | c4f2b6dce306933e6d8fa82022f24a11ea1d29f0 (patch) | |
| tree | e313c7798478f5fa5a6db4c4a80f717733945cb9 | |
| parent | 613f2dc4ae7970772040abd69ae30b4d0dab9da9 (diff) | |
| download | ffmpeg-c4f2b6dce306933e6d8fa82022f24a11ea1d29f0.tar.gz | |
Use constant offsets for memory operands since gcc is unable to
This fixes gcc failing to fit 6 memory locations into 7 registers on x86-32
Originally committed as revision 21337 to svn://svn.ffmpeg.org/ffmpeg/trunk
| -rw-r--r-- | libavcodec/x86/h264dsp_mmx.c | 20 |
1 files changed, 8 insertions, 12 deletions
diff --git a/libavcodec/x86/h264dsp_mmx.c b/libavcodec/x86/h264dsp_mmx.c index e26881beaf..724a6e6813 100644 --- a/libavcodec/x86/h264dsp_mmx.c +++ b/libavcodec/x86/h264dsp_mmx.c @@ -834,9 +834,9 @@ static void h264_loop_filter_strength_mmx2( int16_t bS[2][4][4], uint8_t nnz[40] "por %%mm1, %%mm0 \n\t" "movq %2, %%mm1 \n\t" - "movq %3, %%mm2 \n\t" - "psubw %4, %%mm1 \n\t" - "psubw %5, %%mm2 \n\t" + "movq 8+1*%2, %%mm2 \n\t" + "psubw %3, %%mm1 \n\t" + "psubw 8+1*%3, %%mm2 \n\t" "packsswb %%mm2, %%mm1 \n\t" "paddb %%mm5, %%mm1 \n\t" "pminub %%mm4, %%mm1 \n\t" @@ -845,9 +845,7 @@ static void h264_loop_filter_strength_mmx2( int16_t bS[2][4][4], uint8_t nnz[40] ::"m"(ref[l][b_idx]), "m"(ref[l][b_idx+d_idx]), "m"(mv[l][b_idx][0]), - "m"(mv[l][b_idx+2][0]), - "m"(mv[l][b_idx+d_idx][0]), - "m"(mv[l][b_idx+d_idx+2][0]) + "m"(mv[l][b_idx+d_idx][0]) ); } if(bidir==1){ @@ -863,9 +861,9 @@ static void h264_loop_filter_strength_mmx2( int16_t bS[2][4][4], uint8_t nnz[40] "por %%mm1, %%mm3 \n\t" "movq %2, %%mm1 \n\t" - "movq %3, %%mm2 \n\t" - "psubw %4, %%mm1 \n\t" - "psubw %5, %%mm2 \n\t" + "movq 8+1*%2, %%mm2 \n\t" + "psubw %3, %%mm1 \n\t" + "psubw 8+1*%3, %%mm2 \n\t" "packsswb %%mm2, %%mm1 \n\t" "paddb %%mm5, %%mm1 \n\t" "pminub %%mm4, %%mm1 \n\t" @@ -874,9 +872,7 @@ static void h264_loop_filter_strength_mmx2( int16_t bS[2][4][4], uint8_t nnz[40] ::"m"(ref[l][b_idx]), "m"(ref[1-l][b_idx+d_idx]), "m"(mv[l][b_idx][0]), - "m"(mv[l][b_idx+2][0]), - "m"(mv[1-l][b_idx+d_idx][0]), - "m"(mv[1-l][b_idx+d_idx+2][0]) + "m"(mv[1-l][b_idx+d_idx][0]) ); } __asm__ volatile( |