Merge remote-tracking branch 'qatar/master'

* qatar/master: (40 commits)
  H.264: template left MB handling
  H.264: faster fill_decode_caches
  H.264: faster write_back_*
  H.264: faster fill_filter_caches
  H.264: make filter_mb_fast support the case of unavailable top mb
  Do not include log.h in avutil.h
  Do not include pixfmt.h in avutil.h
  Do not include rational.h in avutil.h
  Do not include mathematics.h in avutil.h
  Do not include intfloat_readwrite.h in avutil.h
  Remove return statements following infinite loops without break
  RTSP: Doxygen comment cleanup
  doxygen: Escape '\' in Doxygen documentation.
  md5: cosmetics
  md5: use AV_WL32 to write result
  md5: add fate test
  md5: include correct headers
  md5: fix test program
  doxygen: Drop array size declarations from Doxygen parameter names.
  doxygen: Fix parameter names to match the function prototypes.
  ...

Conflicts:
	libavcodec/x86/dsputil_mmx.c
	libavformat/flvenc.c
	libavformat/oggenc.c
	libavformat/wtv.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>

1 files changed, 86 insertions, 50 deletions

diff --git a/libavcodec/x86/dsputil_mmx.c b/libavcodec/x86/dsputil_mmx.c
index 80bb6cd26c..13530418ff 100644
--- a/libavcodec/x86/dsputil_mmx.c
+++ b/libavcodec/x86/dsputil_mmx.c
@@ -2530,44 +2530,56 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
                 c->put_no_rnd_pixels_tab[1][2] = put_no_rnd_pixels8_y2_exact_mmx2;
             }
 
-#define SET_QPEL_FUNCS(PFX, IDX, SIZE, CPU) \
-            c->PFX ## _pixels_tab[IDX][ 0] = PFX ## SIZE ## _mc00_ ## CPU; \
-            c->PFX ## _pixels_tab[IDX][ 1] = PFX ## SIZE ## _mc10_ ## CPU; \
-            c->PFX ## _pixels_tab[IDX][ 2] = PFX ## SIZE ## _mc20_ ## CPU; \
-            c->PFX ## _pixels_tab[IDX][ 3] = PFX ## SIZE ## _mc30_ ## CPU; \
-            c->PFX ## _pixels_tab[IDX][ 4] = PFX ## SIZE ## _mc01_ ## CPU; \
-            c->PFX ## _pixels_tab[IDX][ 5] = PFX ## SIZE ## _mc11_ ## CPU; \
-            c->PFX ## _pixels_tab[IDX][ 6] = PFX ## SIZE ## _mc21_ ## CPU; \
-            c->PFX ## _pixels_tab[IDX][ 7] = PFX ## SIZE ## _mc31_ ## CPU; \
-            c->PFX ## _pixels_tab[IDX][ 8] = PFX ## SIZE ## _mc02_ ## CPU; \
-            c->PFX ## _pixels_tab[IDX][ 9] = PFX ## SIZE ## _mc12_ ## CPU; \
-            c->PFX ## _pixels_tab[IDX][10] = PFX ## SIZE ## _mc22_ ## CPU; \
-            c->PFX ## _pixels_tab[IDX][11] = PFX ## SIZE ## _mc32_ ## CPU; \
-            c->PFX ## _pixels_tab[IDX][12] = PFX ## SIZE ## _mc03_ ## CPU; \
-            c->PFX ## _pixels_tab[IDX][13] = PFX ## SIZE ## _mc13_ ## CPU; \
-            c->PFX ## _pixels_tab[IDX][14] = PFX ## SIZE ## _mc23_ ## CPU; \
-            c->PFX ## _pixels_tab[IDX][15] = PFX ## SIZE ## _mc33_ ## CPU
-
-            SET_QPEL_FUNCS(put_qpel, 0, 16, mmx2);
-            SET_QPEL_FUNCS(put_qpel, 1, 8, mmx2);
-            SET_QPEL_FUNCS(put_no_rnd_qpel, 0, 16, mmx2);
-            SET_QPEL_FUNCS(put_no_rnd_qpel, 1, 8, mmx2);
-            SET_QPEL_FUNCS(avg_qpel, 0, 16, mmx2);
-            SET_QPEL_FUNCS(avg_qpel, 1, 8, mmx2);
+#define SET_QPEL_FUNCS(PFX, IDX, SIZE, CPU, PREFIX) \
+            c->PFX ## _pixels_tab[IDX][ 0] = PREFIX ## PFX ## SIZE ## _mc00_ ## CPU; \
+            c->PFX ## _pixels_tab[IDX][ 1] = PREFIX ## PFX ## SIZE ## _mc10_ ## CPU; \
+            c->PFX ## _pixels_tab[IDX][ 2] = PREFIX ## PFX ## SIZE ## _mc20_ ## CPU; \
+            c->PFX ## _pixels_tab[IDX][ 3] = PREFIX ## PFX ## SIZE ## _mc30_ ## CPU; \
+            c->PFX ## _pixels_tab[IDX][ 4] = PREFIX ## PFX ## SIZE ## _mc01_ ## CPU; \
+            c->PFX ## _pixels_tab[IDX][ 5] = PREFIX ## PFX ## SIZE ## _mc11_ ## CPU; \
+            c->PFX ## _pixels_tab[IDX][ 6] = PREFIX ## PFX ## SIZE ## _mc21_ ## CPU; \
+            c->PFX ## _pixels_tab[IDX][ 7] = PREFIX ## PFX ## SIZE ## _mc31_ ## CPU; \
+            c->PFX ## _pixels_tab[IDX][ 8] = PREFIX ## PFX ## SIZE ## _mc02_ ## CPU; \
+            c->PFX ## _pixels_tab[IDX][ 9] = PREFIX ## PFX ## SIZE ## _mc12_ ## CPU; \
+            c->PFX ## _pixels_tab[IDX][10] = PREFIX ## PFX ## SIZE ## _mc22_ ## CPU; \
+            c->PFX ## _pixels_tab[IDX][11] = PREFIX ## PFX ## SIZE ## _mc32_ ## CPU; \
+            c->PFX ## _pixels_tab[IDX][12] = PREFIX ## PFX ## SIZE ## _mc03_ ## CPU; \
+            c->PFX ## _pixels_tab[IDX][13] = PREFIX ## PFX ## SIZE ## _mc13_ ## CPU; \
+            c->PFX ## _pixels_tab[IDX][14] = PREFIX ## PFX ## SIZE ## _mc23_ ## CPU; \
+            c->PFX ## _pixels_tab[IDX][15] = PREFIX ## PFX ## SIZE ## _mc33_ ## CPU
+
+            SET_QPEL_FUNCS(put_qpel, 0, 16, mmx2, );
+            SET_QPEL_FUNCS(put_qpel, 1, 8, mmx2, );
+            SET_QPEL_FUNCS(put_no_rnd_qpel, 0, 16, mmx2, );
+            SET_QPEL_FUNCS(put_no_rnd_qpel, 1, 8, mmx2, );
+            SET_QPEL_FUNCS(avg_qpel, 0, 16, mmx2, );
+            SET_QPEL_FUNCS(avg_qpel, 1, 8, mmx2, );
 
             if (!high_bit_depth) {
-            SET_QPEL_FUNCS(put_h264_qpel, 0, 16, mmx2);
-            SET_QPEL_FUNCS(put_h264_qpel, 1, 8, mmx2);
-            SET_QPEL_FUNCS(put_h264_qpel, 2, 4, mmx2);
-            SET_QPEL_FUNCS(avg_h264_qpel, 0, 16, mmx2);
-            SET_QPEL_FUNCS(avg_h264_qpel, 1, 8, mmx2);
-            SET_QPEL_FUNCS(avg_h264_qpel, 2, 4, mmx2);
+            SET_QPEL_FUNCS(put_h264_qpel, 0, 16, mmx2, );
+            SET_QPEL_FUNCS(put_h264_qpel, 1, 8, mmx2, );
+            SET_QPEL_FUNCS(put_h264_qpel, 2, 4, mmx2, );
+            SET_QPEL_FUNCS(avg_h264_qpel, 0, 16, mmx2, );
+            SET_QPEL_FUNCS(avg_h264_qpel, 1, 8, mmx2, );
+            SET_QPEL_FUNCS(avg_h264_qpel, 2, 4, mmx2, );
             }
+#if HAVE_YASM
+            else if (bit_depth == 10) {
+#if !ARCH_X86_64
+                SET_QPEL_FUNCS(avg_h264_qpel, 0, 16, 10_mmxext, ff_);
+                SET_QPEL_FUNCS(put_h264_qpel, 0, 16, 10_mmxext, ff_);
+                SET_QPEL_FUNCS(put_h264_qpel, 1, 8,  10_mmxext, ff_);
+                SET_QPEL_FUNCS(avg_h264_qpel, 1, 8,  10_mmxext, ff_);
+#endif
+                SET_QPEL_FUNCS(put_h264_qpel, 2, 4,  10_mmxext, ff_);
+                SET_QPEL_FUNCS(avg_h264_qpel, 2, 4,  10_mmxext, ff_);
+            }
+#endif
 
-            SET_QPEL_FUNCS(put_2tap_qpel, 0, 16, mmx2);
-            SET_QPEL_FUNCS(put_2tap_qpel, 1, 8, mmx2);
-            SET_QPEL_FUNCS(avg_2tap_qpel, 0, 16, mmx2);
-            SET_QPEL_FUNCS(avg_2tap_qpel, 1, 8, mmx2);
+            SET_QPEL_FUNCS(put_2tap_qpel, 0, 16, mmx2, );
+            SET_QPEL_FUNCS(put_2tap_qpel, 1, 8, mmx2, );
+            SET_QPEL_FUNCS(avg_2tap_qpel, 0, 16, mmx2, );
+            SET_QPEL_FUNCS(avg_2tap_qpel, 1, 8, mmx2, );
 
 #if HAVE_YASM
             c->avg_rv40_chroma_pixels_tab[0]= ff_avg_rv40_chroma_mc8_mmx2;
@@ -2627,26 +2639,26 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
                 c->put_no_rnd_pixels_tab[1][2] = put_no_rnd_pixels8_y2_exact_3dnow;
             }
 
-            SET_QPEL_FUNCS(put_qpel, 0, 16, 3dnow);
-            SET_QPEL_FUNCS(put_qpel, 1, 8, 3dnow);
-            SET_QPEL_FUNCS(put_no_rnd_qpel, 0, 16, 3dnow);
-            SET_QPEL_FUNCS(put_no_rnd_qpel, 1, 8, 3dnow);
-            SET_QPEL_FUNCS(avg_qpel, 0, 16, 3dnow);
-            SET_QPEL_FUNCS(avg_qpel, 1, 8, 3dnow);
+            SET_QPEL_FUNCS(put_qpel, 0, 16, 3dnow, );
+            SET_QPEL_FUNCS(put_qpel, 1, 8, 3dnow, );
+            SET_QPEL_FUNCS(put_no_rnd_qpel, 0, 16, 3dnow, );
+            SET_QPEL_FUNCS(put_no_rnd_qpel, 1, 8, 3dnow, );
+            SET_QPEL_FUNCS(avg_qpel, 0, 16, 3dnow, );
+            SET_QPEL_FUNCS(avg_qpel, 1, 8, 3dnow, );
 
             if (!high_bit_depth) {
-            SET_QPEL_FUNCS(put_h264_qpel, 0, 16, 3dnow);
-            SET_QPEL_FUNCS(put_h264_qpel, 1, 8, 3dnow);
-            SET_QPEL_FUNCS(put_h264_qpel, 2, 4, 3dnow);
-            SET_QPEL_FUNCS(avg_h264_qpel, 0, 16, 3dnow);
-            SET_QPEL_FUNCS(avg_h264_qpel, 1, 8, 3dnow);
-            SET_QPEL_FUNCS(avg_h264_qpel, 2, 4, 3dnow);
+            SET_QPEL_FUNCS(put_h264_qpel, 0, 16, 3dnow, );
+            SET_QPEL_FUNCS(put_h264_qpel, 1, 8, 3dnow, );
+            SET_QPEL_FUNCS(put_h264_qpel, 2, 4, 3dnow, );
+            SET_QPEL_FUNCS(avg_h264_qpel, 0, 16, 3dnow, );
+            SET_QPEL_FUNCS(avg_h264_qpel, 1, 8, 3dnow, );
+            SET_QPEL_FUNCS(avg_h264_qpel, 2, 4, 3dnow, );
             }
 
-            SET_QPEL_FUNCS(put_2tap_qpel, 0, 16, 3dnow);
-            SET_QPEL_FUNCS(put_2tap_qpel, 1, 8, 3dnow);
-            SET_QPEL_FUNCS(avg_2tap_qpel, 0, 16, 3dnow);
-            SET_QPEL_FUNCS(avg_2tap_qpel, 1, 8, 3dnow);
+            SET_QPEL_FUNCS(put_2tap_qpel, 0, 16, 3dnow, );
+            SET_QPEL_FUNCS(put_2tap_qpel, 1, 8, 3dnow, );
+            SET_QPEL_FUNCS(avg_2tap_qpel, 0, 16, 3dnow, );
+            SET_QPEL_FUNCS(avg_2tap_qpel, 1, 8, 3dnow, );
 
 #if HAVE_YASM
             if (!high_bit_depth) {
@@ -2690,7 +2702,20 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
             H264_QPEL_FUNCS(3, 3, sse2);
             }
 #if HAVE_YASM
+#define H264_QPEL_FUNCS_10(x, y, CPU)\
+            c->put_h264_qpel_pixels_tab[0][x+y*4] = ff_put_h264_qpel16_mc##x##y##_10_##CPU;\
+            c->put_h264_qpel_pixels_tab[1][x+y*4] = ff_put_h264_qpel8_mc##x##y##_10_##CPU;\
+            c->avg_h264_qpel_pixels_tab[0][x+y*4] = ff_avg_h264_qpel16_mc##x##y##_10_##CPU;\
+            c->avg_h264_qpel_pixels_tab[1][x+y*4] = ff_avg_h264_qpel8_mc##x##y##_10_##CPU;
             if (bit_depth == 10) {
+                SET_QPEL_FUNCS(put_h264_qpel, 0, 16, 10_sse2, ff_);
+                SET_QPEL_FUNCS(put_h264_qpel, 1, 8,  10_sse2, ff_);
+                SET_QPEL_FUNCS(avg_h264_qpel, 0, 16, 10_sse2, ff_);
+                SET_QPEL_FUNCS(avg_h264_qpel, 1, 8,  10_sse2, ff_);
+                H264_QPEL_FUNCS_10(1, 0, sse2_cache64)
+                H264_QPEL_FUNCS_10(2, 0, sse2_cache64)
+                H264_QPEL_FUNCS_10(3, 0, sse2_cache64)
+
                 c->put_h264_chroma_pixels_tab[0]= ff_put_h264_chroma_mc8_10_sse2;
                 c->avg_h264_chroma_pixels_tab[0]= ff_avg_h264_chroma_mc8_10_sse2;
             }
@@ -2712,6 +2737,11 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
             H264_QPEL_FUNCS(3, 2, ssse3);
             H264_QPEL_FUNCS(3, 3, ssse3);
             }
+            else if (bit_depth == 10) {
+                H264_QPEL_FUNCS_10(1, 0, ssse3_cache64)
+                H264_QPEL_FUNCS_10(2, 0, ssse3_cache64)
+                H264_QPEL_FUNCS_10(3, 0, ssse3_cache64)
+            }
 #if HAVE_YASM
             if (!high_bit_depth) {
             c->put_h264_chroma_pixels_tab[0]= ff_put_h264_chroma_mc8_ssse3_rnd;
@@ -2807,6 +2837,12 @@ void dsputil_init_mmx(DSPContext* c, AVCodecContext *avctx)
 #if HAVE_AVX && HAVE_YASM
         if (mm_flags & AV_CPU_FLAG_AVX) {
             if (bit_depth == 10) {
+                //AVX implies !cache64.
+                //TODO: Port cache(32|64) detection from x264.
+                H264_QPEL_FUNCS_10(1, 0, sse2)
+                H264_QPEL_FUNCS_10(2, 0, sse2)
+                H264_QPEL_FUNCS_10(3, 0, sse2)
+
                 c->put_h264_chroma_pixels_tab[0]= ff_put_h264_chroma_mc8_10_avx;
                 c->avg_h264_chroma_pixels_tab[0]= ff_avg_h264_chroma_mc8_10_avx;
             }