Preparatory patch for high bit depth h264 decoding support.

Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>

1 files changed, 75 insertions, 44 deletions

diff --git a/libavcodec/h264dsp_template.c b/libavcodec/h264dsp_template.c
index 2fe09ff56b..7190f4d9ca 100644
--- a/libavcodec/h264dsp_template.c
+++ b/libavcodec/h264dsp_template.c
@@ -25,12 +25,19 @@
  * @author Michael Niedermayer <michaelni@gmx.at>
  */
 
-#define op_scale1(x)  block[x] = av_clip_uint8( (block[x]*weight + offset) >> log2_denom )
-#define op_scale2(x)  dst[x] = av_clip_uint8( (src[x]*weights + dst[x]*weightd + offset) >> (log2_denom+1))
+#define BIT_DEPTH 8
+#define pixel uint8_t
+#define av_clip_pixel av_clip_uint8
+#define FUNCC(a) a ## _c
+
+#define op_scale1(x)  block[x] = av_clip_pixel( (block[x]*weight + offset) >> log2_denom )
+#define op_scale2(x)  dst[x] = av_clip_pixel( (src[x]*weights + dst[x]*weightd + offset) >> (log2_denom+1))
 #define H264_WEIGHT(W,H) \
-static void weight_h264_pixels ## W ## x ## H ## _c(uint8_t *block, int stride, int log2_denom, int weight, int offset){ \
+static void FUNCC(weight_h264_pixels ## W ## x ## H)(uint8_t *_block, int stride, int log2_denom, int weight, int offset){ \
     int y; \
-    offset <<= log2_denom; \
+    pixel *block = (pixel*)_block; \
+    stride /= sizeof(pixel); \
+    offset <<= (log2_denom + (BIT_DEPTH-8)); \
     if(log2_denom) offset += 1<<(log2_denom-1); \
     for(y=0; y<H; y++, block += stride){ \
         op_scale1(0); \
@@ -54,8 +61,11 @@ static void weight_h264_pixels ## W ## x ## H ## _c(uint8_t *block, int stride,
         op_scale1(15); \
     } \
 } \
-static void biweight_h264_pixels ## W ## x ## H ## _c(uint8_t *dst, uint8_t *src, int stride, int log2_denom, int weightd, int weights, int offset){ \
+static void FUNCC(biweight_h264_pixels ## W ## x ## H)(uint8_t *_dst, uint8_t *_src, int stride, int log2_denom, int weightd, int weights, int offset){ \
     int y; \
+    pixel *dst = (pixel*)_dst; \
+    pixel *src = (pixel*)_src; \
+    stride /= sizeof(pixel); \
     offset = ((offset + 1) | 1) << log2_denom; \
     for(y=0; y<H; y++, dst += stride, src += stride){ \
         op_scale2(0); \
@@ -95,11 +105,17 @@ H264_WEIGHT(2,2)
 #undef op_scale2
 #undef H264_WEIGHT
 
-static av_always_inline av_flatten void h264_loop_filter_luma_c(uint8_t *pix, int xstride, int ystride, int inner_iters, int alpha, int beta, int8_t *tc0)
+static av_always_inline av_flatten void FUNCC(h264_loop_filter_luma)(uint8_t *_pix, int xstride, int ystride, int inner_iters, int alpha, int beta, int8_t *tc0)
 {
+    pixel *pix = (pixel*)_pix;
     int i, d;
+    xstride /= sizeof(pixel);
+    ystride /= sizeof(pixel);
+    alpha <<= BIT_DEPTH - 8;
+    beta  <<= BIT_DEPTH - 8;
     for( i = 0; i < 4; i++ ) {
-        if( tc0[i] < 0 ) {
+        const int tc_orig = tc0[i] << (BIT_DEPTH - 8);
+        if( tc_orig < 0 ) {
             pix += inner_iters*ystride;
             continue;
         }
@@ -115,44 +131,49 @@ static av_always_inline av_flatten void h264_loop_filter_luma_c(uint8_t *pix, in
                 FFABS( p1 - p0 ) < beta &&
                 FFABS( q1 - q0 ) < beta ) {
 
-                int tc = tc0[i];
+                int tc = tc_orig;
                 int i_delta;
 
                 if( FFABS( p2 - p0 ) < beta ) {
-                    if(tc0[i])
-                    pix[-2*xstride] = p1 + av_clip( (( p2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - p1, -tc0[i], tc0[i] );
+                    if(tc_orig)
+                    pix[-2*xstride] = p1 + av_clip( (( p2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - p1, -tc_orig, tc_orig );
                     tc++;
                 }
                 if( FFABS( q2 - q0 ) < beta ) {
-                    if(tc0[i])
-                    pix[   xstride] = q1 + av_clip( (( q2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - q1, -tc0[i], tc0[i] );
+                    if(tc_orig)
+                    pix[   xstride] = q1 + av_clip( (( q2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - q1, -tc_orig, tc_orig );
                     tc++;
                 }
 
                 i_delta = av_clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
-                pix[-xstride] = av_clip_uint8( p0 + i_delta );    /* p0' */
-                pix[0]        = av_clip_uint8( q0 - i_delta );    /* q0' */
+                pix[-xstride] = av_clip_pixel( p0 + i_delta );    /* p0' */
+                pix[0]        = av_clip_pixel( q0 - i_delta );    /* q0' */
             }
             pix += ystride;
         }
     }
 }
-static void h264_v_loop_filter_luma_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
+static void FUNCC(h264_v_loop_filter_luma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
 {
-    h264_loop_filter_luma_c(pix, stride, 1, 4, alpha, beta, tc0);
+    FUNCC(h264_loop_filter_luma)(pix, stride, sizeof(pixel), 4, alpha, beta, tc0);
 }
-static void h264_h_loop_filter_luma_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
+static void FUNCC(h264_h_loop_filter_luma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
 {
-    h264_loop_filter_luma_c(pix, 1, stride, 4, alpha, beta, tc0);
+    FUNCC(h264_loop_filter_luma)(pix, sizeof(pixel), stride, 4, alpha, beta, tc0);
 }
-static void h264_h_loop_filter_luma_mbaff_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
+static void FUNCC(h264_h_loop_filter_luma_mbaff)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
 {
-    h264_loop_filter_luma_c(pix, 1, stride, 2, alpha, beta, tc0);
+    FUNCC(h264_loop_filter_luma)(pix, sizeof(pixel), stride, 2, alpha, beta, tc0);
 }
 
-static av_always_inline av_flatten void h264_loop_filter_luma_intra_c(uint8_t *pix, int xstride, int ystride, int inner_iters, int alpha, int beta)
+static av_always_inline av_flatten void FUNCC(h264_loop_filter_luma_intra)(uint8_t *_pix, int xstride, int ystride, int inner_iters, int alpha, int beta)
 {
+    pixel *pix = (pixel*)_pix;
     int d;
+    xstride /= sizeof(pixel);
+    ystride /= sizeof(pixel);
+    alpha <<= BIT_DEPTH - 8;
+    beta  <<= BIT_DEPTH - 8;
     for( d = 0; d < 4 * inner_iters; d++ ) {
         const int p2 = pix[-3*xstride];
         const int p1 = pix[-2*xstride];
@@ -198,24 +219,29 @@ static av_always_inline av_flatten void h264_loop_filter_luma_intra_c(uint8_t *p
         pix += ystride;
     }
 }
-static void h264_v_loop_filter_luma_intra_c(uint8_t *pix, int stride, int alpha, int beta)
+static void FUNCC(h264_v_loop_filter_luma_intra)(uint8_t *pix, int stride, int alpha, int beta)
 {
-    h264_loop_filter_luma_intra_c(pix, stride, 1, 4, alpha, beta);
+    FUNCC(h264_loop_filter_luma_intra)(pix, stride, sizeof(pixel), 4, alpha, beta);
 }
-static void h264_h_loop_filter_luma_intra_c(uint8_t *pix, int stride, int alpha, int beta)
+static void FUNCC(h264_h_loop_filter_luma_intra)(uint8_t *pix, int stride, int alpha, int beta)
 {
-    h264_loop_filter_luma_intra_c(pix, 1, stride, 4, alpha, beta);
+    FUNCC(h264_loop_filter_luma_intra)(pix, sizeof(pixel), stride, 4, alpha, beta);
 }
-static void h264_h_loop_filter_luma_mbaff_intra_c(uint8_t *pix, int stride, int alpha, int beta)
+static void FUNCC(h264_h_loop_filter_luma_mbaff_intra)(uint8_t *pix, int stride, int alpha, int beta)
 {
-    h264_loop_filter_luma_intra_c(pix, 1, stride, 2, alpha, beta);
+    FUNCC(h264_loop_filter_luma_intra)(pix, sizeof(pixel), stride, 2, alpha, beta);
 }
 
-static av_always_inline av_flatten void h264_loop_filter_chroma_c(uint8_t *pix, int xstride, int ystride, int inner_iters, int alpha, int beta, int8_t *tc0)
+static av_always_inline av_flatten void FUNCC(h264_loop_filter_chroma)(uint8_t *_pix, int xstride, int ystride, int inner_iters, int alpha, int beta, int8_t *tc0)
 {
+    pixel *pix = (pixel*)_pix;
     int i, d;
+    xstride /= sizeof(pixel);
+    ystride /= sizeof(pixel);
+    alpha <<= BIT_DEPTH - 8;
+    beta  <<= BIT_DEPTH - 8;
     for( i = 0; i < 4; i++ ) {
-        const int tc = tc0[i];
+        const int tc = ((tc0[i] - 1) << (BIT_DEPTH - 8)) + 1;
         if( tc <= 0 ) {
             pix += inner_iters*ystride;
             continue;
@@ -232,29 +258,34 @@ static av_always_inline av_flatten void h264_loop_filter_chroma_c(uint8_t *pix,
 
                 int delta = av_clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
 
-                pix[-xstride] = av_clip_uint8( p0 + delta );    /* p0' */
-                pix[0]        = av_clip_uint8( q0 - delta );    /* q0' */
+                pix[-xstride] = av_clip_pixel( p0 + delta );    /* p0' */
+                pix[0]        = av_clip_pixel( q0 - delta );    /* q0' */
             }
             pix += ystride;
         }
     }
 }
-static void h264_v_loop_filter_chroma_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
+static void FUNCC(h264_v_loop_filter_chroma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
 {
-    h264_loop_filter_chroma_c(pix, stride, 1, 2, alpha, beta, tc0);
+    FUNCC(h264_loop_filter_chroma)(pix, stride, sizeof(pixel), 2, alpha, beta, tc0);
 }
-static void h264_h_loop_filter_chroma_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
+static void FUNCC(h264_h_loop_filter_chroma)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
 {
-    h264_loop_filter_chroma_c(pix, 1, stride, 2, alpha, beta, tc0);
+    FUNCC(h264_loop_filter_chroma)(pix, sizeof(pixel), stride, 2, alpha, beta, tc0);
 }
-static void h264_h_loop_filter_chroma_mbaff_c(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
+static void FUNCC(h264_h_loop_filter_chroma_mbaff)(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
 {
-    h264_loop_filter_chroma_c(pix, 1, stride, 1, alpha, beta, tc0);
+    FUNCC(h264_loop_filter_chroma)(pix, sizeof(pixel), stride, 1, alpha, beta, tc0);
 }
 
-static av_always_inline av_flatten void h264_loop_filter_chroma_intra_c(uint8_t *pix, int xstride, int ystride, int inner_iters, int alpha, int beta)
+static av_always_inline av_flatten void FUNCC(h264_loop_filter_chroma_intra)(uint8_t *_pix, int xstride, int ystride, int inner_iters, int alpha, int beta)
 {
+    pixel *pix = (pixel*)_pix;
     int d;
+    xstride /= sizeof(pixel);
+    ystride /= sizeof(pixel);
+    alpha <<= BIT_DEPTH - 8;
+    beta  <<= BIT_DEPTH - 8;
     for( d = 0; d < 4 * inner_iters; d++ ) {
         const int p0 = pix[-1*xstride];
         const int p1 = pix[-2*xstride];
@@ -271,15 +302,15 @@ static av_always_inline av_flatten void h264_loop_filter_chroma_intra_c(uint8_t
         pix += ystride;
     }
 }
-static void h264_v_loop_filter_chroma_intra_c(uint8_t *pix, int stride, int alpha, int beta)
+static void FUNCC(h264_v_loop_filter_chroma_intra)(uint8_t *pix, int stride, int alpha, int beta)
 {
-    h264_loop_filter_chroma_intra_c(pix, stride, 1, 2, alpha, beta);
+    FUNCC(h264_loop_filter_chroma_intra)(pix, stride, sizeof(pixel), 2, alpha, beta);
 }
-static void h264_h_loop_filter_chroma_intra_c(uint8_t *pix, int stride, int alpha, int beta)
+static void FUNCC(h264_h_loop_filter_chroma_intra)(uint8_t *pix, int stride, int alpha, int beta)
 {
-    h264_loop_filter_chroma_intra_c(pix, 1, stride, 2, alpha, beta);
+    FUNCC(h264_loop_filter_chroma_intra)(pix, sizeof(pixel), stride, 2, alpha, beta);
 }
-static void h264_h_loop_filter_chroma_mbaff_intra_c(uint8_t *pix, int stride, int alpha, int beta)
+static void FUNCC(h264_h_loop_filter_chroma_mbaff_intra)(uint8_t *pix, int stride, int alpha, int beta)
 {
-    h264_loop_filter_chroma_intra_c(pix, 1, stride, 1, alpha, beta);
+    FUNCC(h264_loop_filter_chroma_intra)(pix, sizeof(pixel), stride, 1, alpha, beta);
 }