Choose h264 chroma dc dequant function dynamically.

Needed for high bit depth h264 decoding.
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>

libavcodec/dsputil.h

@@ -64,6 +64,7 @@ void ff_h264_idct_add16intra_c(uint8_t *dst, const int *blockoffset, DCTELEM *bl
 void ff_h264_idct8_add4_c(uint8_t *dst, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
 void ff_h264_idct_add8_c(uint8_t **dest, const int *blockoffset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
 
+void ff_h264_chroma_dc_dequant_idct_c(DCTELEM *block, int qmul);
 void ff_h264_luma_dc_dequant_idct_c(DCTELEM *output, DCTELEM *input, int qmul);
 void ff_svq3_luma_dc_dequant_idct_c(DCTELEM *output, DCTELEM *input, int qp);
 void ff_svq3_add_idct_c(uint8_t *dst, DCTELEM *block, int stride, int qp, int dc);

libavcodec/h264.c

@@ -288,27 +288,6 @@ static void h264_luma_dc_dct_c(DCTELEM *block/*, int qp*/){
 #undef xStride
 #undef stride
 
-static void chroma_dc_dequant_idct_c(DCTELEM *block, int qmul){
-    const int stride= 16*2;
-    const int xStride= 16;
-    int a,b,c,d,e;
-
-    a= block[stride*0 + xStride*0];
-    b= block[stride*0 + xStride*1];
-    c= block[stride*1 + xStride*0];
-    d= block[stride*1 + xStride*1];
-
-    e= a-b;
-    a= a+b;
-    b= c-d;
-    c= c+d;
-
-    block[stride*0 + xStride*0]= ((a+c)*qmul) >> 7;
-    block[stride*0 + xStride*1]= ((e+b)*qmul) >> 7;
-    block[stride*1 + xStride*0]= ((a-c)*qmul) >> 7;
-    block[stride*1 + xStride*1]= ((e-b)*qmul) >> 7;
-}
-
 #if 0
 static void chroma_dc_dct_c(DCTELEM *block){
     const int stride= 16*2;
@@ -1286,15 +1265,15 @@ static av_always_inline void hl_decode_mb_internal(H264Context *h, int simple){
             }else{
                 if(is_h264){
                     if(h->non_zero_count_cache[ scan8[CHROMA_DC_BLOCK_INDEX+0] ])
-                        chroma_dc_dequant_idct_c(h->mb + 16*16     , h->dequant4_coeff[IS_INTRA(mb_type) ? 1:4][h->chroma_qp[0]][0]);
+                        h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + 16*16     , h->dequant4_coeff[IS_INTRA(mb_type) ? 1:4][h->chroma_qp[0]][0]);
                     if(h->non_zero_count_cache[ scan8[CHROMA_DC_BLOCK_INDEX+1] ])
-                        chroma_dc_dequant_idct_c(h->mb + 16*16+4*16, h->dequant4_coeff[IS_INTRA(mb_type) ? 2:5][h->chroma_qp[1]][0]);
+                        h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + 16*16+4*16, h->dequant4_coeff[IS_INTRA(mb_type) ? 2:5][h->chroma_qp[1]][0]);
                     h->h264dsp.h264_idct_add8(dest, block_offset,
                                               h->mb, uvlinesize,
                                               h->non_zero_count_cache);
                 }else{
-                    chroma_dc_dequant_idct_c(h->mb + 16*16     , h->dequant4_coeff[IS_INTRA(mb_type) ? 1:4][h->chroma_qp[0]][0]);
-                    chroma_dc_dequant_idct_c(h->mb + 16*16+4*16, h->dequant4_coeff[IS_INTRA(mb_type) ? 2:5][h->chroma_qp[1]][0]);
+                    h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + 16*16     , h->dequant4_coeff[IS_INTRA(mb_type) ? 1:4][h->chroma_qp[0]][0]);
+                    h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + 16*16+4*16, h->dequant4_coeff[IS_INTRA(mb_type) ? 2:5][h->chroma_qp[1]][0]);
                     for(i=16; i<16+8; i++){
                         if(h->non_zero_count_cache[ scan8[i] ] || h->mb[i*16]){
                             uint8_t * const ptr= dest[(i&4)>>2] + block_offset[i];

libavcodec/h264dsp.c

@@ -299,6 +299,7 @@ void ff_h264dsp_init(H264DSPContext *c)
     c->h264_idct_add8      = ff_h264_idct_add8_c;
     c->h264_idct_add16intra= ff_h264_idct_add16intra_c;
     c->h264_luma_dc_dequant_idct= ff_h264_luma_dc_dequant_idct_c;
+    c->h264_chroma_dc_dequant_idct= ff_h264_chroma_dc_dequant_idct_c;
 
     c->weight_h264_pixels_tab[0]= weight_h264_pixels16x16_c;
     c->weight_h264_pixels_tab[1]= weight_h264_pixels16x8_c;

libavcodec/h264dsp.h

@@ -72,6 +72,7 @@ typedef struct H264DSPContext{
     void (*h264_idct_add8)(uint8_t **dst/*align 16*/, const int *blockoffset, DCTELEM *block/*align 16*/, int stride, const uint8_t nnzc[6*8]);
     void (*h264_idct_add16intra)(uint8_t *dst/*align 16*/, const int *blockoffset, DCTELEM *block/*align 16*/, int stride, const uint8_t nnzc[6*8]);
     void (*h264_luma_dc_dequant_idct)(DCTELEM *output, DCTELEM *input/*align 16*/, int qmul);
+    void (*h264_chroma_dc_dequant_idct)(DCTELEM *block, int qmul);
 }H264DSPContext;
 
 void ff_h264dsp_init(H264DSPContext *c);

libavcodec/h264idct.c

@@ -250,4 +250,26 @@ void ff_h264_luma_dc_dequant_idct_c(DCTELEM *output, DCTELEM *input, int qmul){
         output[stride* 4+offset]= ((((z1 - z2)*qmul + 128 ) >> 8));
         output[stride* 5+offset]= ((((z0 - z3)*qmul + 128 ) >> 8));
     }
+#undef stride
+}
+
+void ff_h264_chroma_dc_dequant_idct_c(DCTELEM *block, int qmul){
+    const int stride= 16*2;
+    const int xStride= 16;
+    int a,b,c,d,e;
+
+    a= block[stride*0 + xStride*0];
+    b= block[stride*0 + xStride*1];
+    c= block[stride*1 + xStride*0];
+    d= block[stride*1 + xStride*1];
+
+    e= a-b;
+    a= a+b;
+    b= c-d;
+    c= c+d;
+
+    block[stride*0 + xStride*0]= ((a+c)*qmul) >> 7;
+    block[stride*0 + xStride*1]= ((e+b)*qmul) >> 7;
+    block[stride*1 + xStride*0]= ((a-c)*qmul) >> 7;
+    block[stride*1 + xStride*1]= ((e-b)*qmul) >> 7;
 }