ppc: Rename H.264 optimization template file for consistency.
[ffmpeg.git] / libavcodec / cavs.c
index 4233504..42922e0 100644 (file)
@@ -2,25 +2,25 @@
  * Chinese AVS video (AVS1-P2, JiZhun profile) decoder.
  * Copyright (c) 2006  Stefan Gehrer <stefan.gehrer@gmx.de>
  *
  * Chinese AVS video (AVS1-P2, JiZhun profile) decoder.
  * Copyright (c) 2006  Stefan Gehrer <stefan.gehrer@gmx.de>
  *
- * This file is part of FFmpeg.
+ * This file is part of Libav.
  *
  *
- * FFmpeg is free software; you can redistribute it and/or
+ * Libav is free software; you can redistribute it and/or
  * modify it under the terms of the GNU Lesser General Public
  * License as published by the Free Software Foundation; either
  * version 2.1 of the License, or (at your option) any later version.
  *
  * modify it under the terms of the GNU Lesser General Public
  * License as published by the Free Software Foundation; either
  * version 2.1 of the License, or (at your option) any later version.
  *
- * FFmpeg is distributed in the hope that it will be useful,
+ * Libav is distributed in the hope that it will be useful,
  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  * Lesser General Public License for more details.
  *
  * You should have received a copy of the GNU Lesser General Public
  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  * Lesser General Public License for more details.
  *
  * You should have received a copy of the GNU Lesser General Public
- * License along with FFmpeg; if not, write to the Free Software
+ * License along with Libav; if not, write to the Free Software
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
 /**
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  */
 
 /**
- * @file libavcodec/cavs.c
+ * @file
  * Chinese AVS video (AVS1-P2, JiZhun profile) decoder
  * @author Stefan Gehrer <stefan.gehrer@gmx.de>
  */
  * Chinese AVS video (AVS1-P2, JiZhun profile) decoder
  * @author Stefan Gehrer <stefan.gehrer@gmx.de>
  */
@@ -113,22 +113,22 @@ void ff_cavs_filter(AVSContext *h, enum cavs_mb mb_type) {
             if(h->flags & A_AVAIL) {
                 qp_avg = (h->qp + h->left_qp + 1) >> 1;
                 SET_PARAMS;
             if(h->flags & A_AVAIL) {
                 qp_avg = (h->qp + h->left_qp + 1) >> 1;
                 SET_PARAMS;
-                h->s.dsp.cavs_filter_lv(h->cy,h->l_stride,alpha,beta,tc,bs[0],bs[1]);
-                h->s.dsp.cavs_filter_cv(h->cu,h->c_stride,alpha,beta,tc,bs[0],bs[1]);
-                h->s.dsp.cavs_filter_cv(h->cv,h->c_stride,alpha,beta,tc,bs[0],bs[1]);
+                h->cdsp.cavs_filter_lv(h->cy,h->l_stride,alpha,beta,tc,bs[0],bs[1]);
+                h->cdsp.cavs_filter_cv(h->cu,h->c_stride,alpha,beta,tc,bs[0],bs[1]);
+                h->cdsp.cavs_filter_cv(h->cv,h->c_stride,alpha,beta,tc,bs[0],bs[1]);
             }
             qp_avg = h->qp;
             SET_PARAMS;
             }
             qp_avg = h->qp;
             SET_PARAMS;
-            h->s.dsp.cavs_filter_lv(h->cy + 8,h->l_stride,alpha,beta,tc,bs[2],bs[3]);
-            h->s.dsp.cavs_filter_lh(h->cy + 8*h->l_stride,h->l_stride,alpha,beta,tc,
+            h->cdsp.cavs_filter_lv(h->cy + 8,h->l_stride,alpha,beta,tc,bs[2],bs[3]);
+            h->cdsp.cavs_filter_lh(h->cy + 8*h->l_stride,h->l_stride,alpha,beta,tc,
                            bs[6],bs[7]);
 
             if(h->flags & B_AVAIL) {
                 qp_avg = (h->qp + h->top_qp[h->mbx] + 1) >> 1;
                 SET_PARAMS;
                            bs[6],bs[7]);
 
             if(h->flags & B_AVAIL) {
                 qp_avg = (h->qp + h->top_qp[h->mbx] + 1) >> 1;
                 SET_PARAMS;
-                h->s.dsp.cavs_filter_lh(h->cy,h->l_stride,alpha,beta,tc,bs[4],bs[5]);
-                h->s.dsp.cavs_filter_ch(h->cu,h->c_stride,alpha,beta,tc,bs[4],bs[5]);
-                h->s.dsp.cavs_filter_ch(h->cv,h->c_stride,alpha,beta,tc,bs[4],bs[5]);
+                h->cdsp.cavs_filter_lh(h->cy,h->l_stride,alpha,beta,tc,bs[4],bs[5]);
+                h->cdsp.cavs_filter_ch(h->cu,h->c_stride,alpha,beta,tc,bs[4],bs[5]);
+                h->cdsp.cavs_filter_ch(h->cv,h->c_stride,alpha,beta,tc,bs[4],bs[5]);
             }
         }
     }
             }
         }
     }
@@ -324,18 +324,19 @@ void ff_cavs_modify_mb_i(AVSContext *h, int *pred_mode_uv) {
  *
  ****************************************************************************/
 
  *
  ****************************************************************************/
 
-static inline void mc_dir_part(AVSContext *h,Picture *pic,int square,
-                        int chroma_height,int delta,int list,uint8_t *dest_y,
-                        uint8_t *dest_cb,uint8_t *dest_cr,int src_x_offset,
-                        int src_y_offset,qpel_mc_func *qpix_op,
-                        h264_chroma_mc_func chroma_op,cavs_vector *mv){
+static inline void mc_dir_part(AVSContext *h,Picture *pic,
+                               int chroma_height,int delta,int list,uint8_t *dest_y,
+                               uint8_t *dest_cb,uint8_t *dest_cr,int src_x_offset,
+                               int src_y_offset,qpel_mc_func *qpix_op,
+                               h264_chroma_mc_func chroma_op,cavs_vector *mv)
+{
     MpegEncContext * const s = &h->s;
     const int mx= mv->x + src_x_offset*8;
     const int my= mv->y + src_y_offset*8;
     const int luma_xy= (mx&3) + ((my&3)<<2);
     MpegEncContext * const s = &h->s;
     const int mx= mv->x + src_x_offset*8;
     const int my= mv->y + src_y_offset*8;
     const int luma_xy= (mx&3) + ((my&3)<<2);
-    uint8_t * src_y = pic->data[0] + (mx>>2) + (my>>2)*h->l_stride;
-    uint8_t * src_cb= pic->data[1] + (mx>>3) + (my>>3)*h->c_stride;
-    uint8_t * src_cr= pic->data[2] + (mx>>3) + (my>>3)*h->c_stride;
+    uint8_t * src_y  = pic->f.data[0] + (mx >> 2) + (my >> 2) * h->l_stride;
+    uint8_t * src_cb = pic->f.data[1] + (mx >> 3) + (my >> 3) * h->c_stride;
+    uint8_t * src_cr = pic->f.data[2] + (mx >> 3) + (my >> 3) * h->c_stride;
     int extra_width= 0; //(s->flags&CODEC_FLAG_EMU_EDGE) ? 0 : 16;
     int extra_height= extra_width;
     int emu=0;
     int extra_width= 0; //(s->flags&CODEC_FLAG_EMU_EDGE) ? 0 : 16;
     int extra_height= extra_width;
     int emu=0;
@@ -344,7 +345,7 @@ static inline void mc_dir_part(AVSContext *h,Picture *pic,int square,
     const int pic_width  = 16*h->mb_width;
     const int pic_height = 16*h->mb_height;
 
     const int pic_width  = 16*h->mb_width;
     const int pic_height = 16*h->mb_height;
 
-    if(!pic->data[0])
+    if(!pic->f.data[0])
         return;
     if(mx&7) extra_width -= 3;
     if(my&7) extra_height -= 3;
         return;
     if(mx&7) extra_width -= 3;
     if(my&7) extra_height -= 3;
@@ -353,37 +354,35 @@ static inline void mc_dir_part(AVSContext *h,Picture *pic,int square,
           || full_my < 0-extra_height
           || full_mx + 16/*FIXME*/ > pic_width + extra_width
           || full_my + 16/*FIXME*/ > pic_height + extra_height){
           || full_my < 0-extra_height
           || full_mx + 16/*FIXME*/ > pic_width + extra_width
           || full_my + 16/*FIXME*/ > pic_height + extra_height){
-        ff_emulated_edge_mc(s->edge_emu_buffer, src_y - 2 - 2*h->l_stride, h->l_stride,
+        s->dsp.emulated_edge_mc(s->edge_emu_buffer, src_y - 2 - 2*h->l_stride, h->l_stride,
                             16+5, 16+5/*FIXME*/, full_mx-2, full_my-2, pic_width, pic_height);
         src_y= s->edge_emu_buffer + 2 + 2*h->l_stride;
         emu=1;
     }
 
     qpix_op[luma_xy](dest_y, src_y, h->l_stride); //FIXME try variable height perhaps?
                             16+5, 16+5/*FIXME*/, full_mx-2, full_my-2, pic_width, pic_height);
         src_y= s->edge_emu_buffer + 2 + 2*h->l_stride;
         emu=1;
     }
 
     qpix_op[luma_xy](dest_y, src_y, h->l_stride); //FIXME try variable height perhaps?
-    if(!square){
-        qpix_op[luma_xy](dest_y + delta, src_y + delta, h->l_stride);
-    }
 
     if(emu){
 
     if(emu){
-        ff_emulated_edge_mc(s->edge_emu_buffer, src_cb, h->c_stride,
+        s->dsp.emulated_edge_mc(s->edge_emu_buffer, src_cb, h->c_stride,
                             9, 9/*FIXME*/, (mx>>3), (my>>3), pic_width>>1, pic_height>>1);
         src_cb= s->edge_emu_buffer;
     }
     chroma_op(dest_cb, src_cb, h->c_stride, chroma_height, mx&7, my&7);
 
     if(emu){
                             9, 9/*FIXME*/, (mx>>3), (my>>3), pic_width>>1, pic_height>>1);
         src_cb= s->edge_emu_buffer;
     }
     chroma_op(dest_cb, src_cb, h->c_stride, chroma_height, mx&7, my&7);
 
     if(emu){
-        ff_emulated_edge_mc(s->edge_emu_buffer, src_cr, h->c_stride,
+        s->dsp.emulated_edge_mc(s->edge_emu_buffer, src_cr, h->c_stride,
                             9, 9/*FIXME*/, (mx>>3), (my>>3), pic_width>>1, pic_height>>1);
         src_cr= s->edge_emu_buffer;
     }
     chroma_op(dest_cr, src_cr, h->c_stride, chroma_height, mx&7, my&7);
 }
 
                             9, 9/*FIXME*/, (mx>>3), (my>>3), pic_width>>1, pic_height>>1);
         src_cr= s->edge_emu_buffer;
     }
     chroma_op(dest_cr, src_cr, h->c_stride, chroma_height, mx&7, my&7);
 }
 
-static inline void mc_part_std(AVSContext *h,int square,int chroma_height,int delta,
-                        uint8_t *dest_y,uint8_t *dest_cb,uint8_t *dest_cr,
-                        int x_offset, int y_offset,qpel_mc_func *qpix_put,
-                        h264_chroma_mc_func chroma_put,qpel_mc_func *qpix_avg,
-                        h264_chroma_mc_func chroma_avg, cavs_vector *mv){
+static inline void mc_part_std(AVSContext *h,int chroma_height,int delta,
+                               uint8_t *dest_y,uint8_t *dest_cb,uint8_t *dest_cr,
+                               int x_offset, int y_offset,qpel_mc_func *qpix_put,
+                               h264_chroma_mc_func chroma_put,qpel_mc_func *qpix_avg,
+                               h264_chroma_mc_func chroma_avg, cavs_vector *mv)
+{
     qpel_mc_func *qpix_op=  qpix_put;
     h264_chroma_mc_func chroma_op= chroma_put;
 
     qpel_mc_func *qpix_op=  qpix_put;
     h264_chroma_mc_func chroma_op= chroma_put;
 
@@ -395,7 +394,7 @@ static inline void mc_part_std(AVSContext *h,int square,int chroma_height,int de
 
     if(mv->ref >= 0){
         Picture *ref= &h->DPB[mv->ref];
 
     if(mv->ref >= 0){
         Picture *ref= &h->DPB[mv->ref];
-        mc_dir_part(h, ref, square, chroma_height, delta, 0,
+        mc_dir_part(h, ref, chroma_height, delta, 0,
                     dest_y, dest_cb, dest_cr, x_offset, y_offset,
                     qpix_op, chroma_op, mv);
 
                     dest_y, dest_cb, dest_cr, x_offset, y_offset,
                     qpix_op, chroma_op, mv);
 
@@ -405,7 +404,7 @@ static inline void mc_part_std(AVSContext *h,int square,int chroma_height,int de
 
     if((mv+MV_BWD_OFFS)->ref >= 0){
         Picture *ref= &h->DPB[0];
 
     if((mv+MV_BWD_OFFS)->ref >= 0){
         Picture *ref= &h->DPB[0];
-        mc_dir_part(h, ref, square, chroma_height, delta, 1,
+        mc_dir_part(h, ref, chroma_height, delta, 1,
                     dest_y, dest_cb, dest_cr, x_offset, y_offset,
                     qpix_op, chroma_op, mv+MV_BWD_OFFS);
     }
                     dest_y, dest_cb, dest_cr, x_offset, y_offset,
                     qpix_op, chroma_op, mv+MV_BWD_OFFS);
     }
@@ -413,31 +412,31 @@ static inline void mc_part_std(AVSContext *h,int square,int chroma_height,int de
 
 void ff_cavs_inter(AVSContext *h, enum cavs_mb mb_type) {
     if(ff_cavs_partition_flags[mb_type] == 0){ // 16x16
 
 void ff_cavs_inter(AVSContext *h, enum cavs_mb mb_type) {
     if(ff_cavs_partition_flags[mb_type] == 0){ // 16x16
-        mc_part_std(h, 1, 8, 0, h->cy, h->cu, h->cv, 0, 0,
-                h->s.dsp.put_cavs_qpel_pixels_tab[0],
+        mc_part_std(h, 8, 0, h->cy, h->cu, h->cv, 0, 0,
+                h->cdsp.put_cavs_qpel_pixels_tab[0],
                 h->s.dsp.put_h264_chroma_pixels_tab[0],
                 h->s.dsp.put_h264_chroma_pixels_tab[0],
-                h->s.dsp.avg_cavs_qpel_pixels_tab[0],
+                h->cdsp.avg_cavs_qpel_pixels_tab[0],
                 h->s.dsp.avg_h264_chroma_pixels_tab[0],&h->mv[MV_FWD_X0]);
     }else{
                 h->s.dsp.avg_h264_chroma_pixels_tab[0],&h->mv[MV_FWD_X0]);
     }else{
-        mc_part_std(h, 1, 4, 0, h->cy, h->cu, h->cv, 0, 0,
-                h->s.dsp.put_cavs_qpel_pixels_tab[1],
+        mc_part_std(h, 4, 0, h->cy, h->cu, h->cv, 0, 0,
+                h->cdsp.put_cavs_qpel_pixels_tab[1],
                 h->s.dsp.put_h264_chroma_pixels_tab[1],
                 h->s.dsp.put_h264_chroma_pixels_tab[1],
-                h->s.dsp.avg_cavs_qpel_pixels_tab[1],
+                h->cdsp.avg_cavs_qpel_pixels_tab[1],
                 h->s.dsp.avg_h264_chroma_pixels_tab[1],&h->mv[MV_FWD_X0]);
                 h->s.dsp.avg_h264_chroma_pixels_tab[1],&h->mv[MV_FWD_X0]);
-        mc_part_std(h, 1, 4, 0, h->cy, h->cu, h->cv, 4, 0,
-                h->s.dsp.put_cavs_qpel_pixels_tab[1],
+        mc_part_std(h, 4, 0, h->cy, h->cu, h->cv, 4, 0,
+                h->cdsp.put_cavs_qpel_pixels_tab[1],
                 h->s.dsp.put_h264_chroma_pixels_tab[1],
                 h->s.dsp.put_h264_chroma_pixels_tab[1],
-                h->s.dsp.avg_cavs_qpel_pixels_tab[1],
+                h->cdsp.avg_cavs_qpel_pixels_tab[1],
                 h->s.dsp.avg_h264_chroma_pixels_tab[1],&h->mv[MV_FWD_X1]);
                 h->s.dsp.avg_h264_chroma_pixels_tab[1],&h->mv[MV_FWD_X1]);
-        mc_part_std(h, 1, 4, 0, h->cy, h->cu, h->cv, 0, 4,
-                h->s.dsp.put_cavs_qpel_pixels_tab[1],
+        mc_part_std(h, 4, 0, h->cy, h->cu, h->cv, 0, 4,
+                h->cdsp.put_cavs_qpel_pixels_tab[1],
                 h->s.dsp.put_h264_chroma_pixels_tab[1],
                 h->s.dsp.put_h264_chroma_pixels_tab[1],
-                h->s.dsp.avg_cavs_qpel_pixels_tab[1],
+                h->cdsp.avg_cavs_qpel_pixels_tab[1],
                 h->s.dsp.avg_h264_chroma_pixels_tab[1],&h->mv[MV_FWD_X2]);
                 h->s.dsp.avg_h264_chroma_pixels_tab[1],&h->mv[MV_FWD_X2]);
-        mc_part_std(h, 1, 4, 0, h->cy, h->cu, h->cv, 4, 4,
-                h->s.dsp.put_cavs_qpel_pixels_tab[1],
+        mc_part_std(h, 4, 0, h->cy, h->cu, h->cv, 4, 4,
+                h->cdsp.put_cavs_qpel_pixels_tab[1],
                 h->s.dsp.put_h264_chroma_pixels_tab[1],
                 h->s.dsp.put_h264_chroma_pixels_tab[1],
-                h->s.dsp.avg_cavs_qpel_pixels_tab[1],
+                h->cdsp.avg_cavs_qpel_pixels_tab[1],
                 h->s.dsp.avg_h264_chroma_pixels_tab[1],&h->mv[MV_FWD_X3]);
     }
 }
                 h->s.dsp.avg_h264_chroma_pixels_tab[1],&h->mv[MV_FWD_X3]);
     }
 }
@@ -602,9 +601,9 @@ int ff_cavs_next_mb(AVSContext *h) {
         h->mbx = 0;
         h->mby++;
         /* re-calculate sample pointers */
         h->mbx = 0;
         h->mby++;
         /* re-calculate sample pointers */
-        h->cy = h->picture.data[0] + h->mby*16*h->l_stride;
-        h->cu = h->picture.data[1] + h->mby*8*h->c_stride;
-        h->cv = h->picture.data[2] + h->mby*8*h->c_stride;
+        h->cy = h->picture.f.data[0] + h->mby * 16 * h->l_stride;
+        h->cu = h->picture.f.data[1] + h->mby *  8 * h->c_stride;
+        h->cv = h->picture.f.data[2] + h->mby *  8 * h->c_stride;
         if(h->mby == h->mb_height) { //frame end
             return 0;
         }
         if(h->mby == h->mb_height) { //frame end
             return 0;
         }
@@ -629,11 +628,11 @@ void ff_cavs_init_pic(AVSContext *h) {
     h->mv[MV_FWD_X0] = ff_cavs_dir_mv;
     set_mvs(&h->mv[MV_FWD_X0], BLK_16X16);
     h->pred_mode_Y[3] = h->pred_mode_Y[6] = NOT_AVAIL;
     h->mv[MV_FWD_X0] = ff_cavs_dir_mv;
     set_mvs(&h->mv[MV_FWD_X0], BLK_16X16);
     h->pred_mode_Y[3] = h->pred_mode_Y[6] = NOT_AVAIL;
-    h->cy = h->picture.data[0];
-    h->cu = h->picture.data[1];
-    h->cv = h->picture.data[2];
-    h->l_stride = h->picture.linesize[0];
-    h->c_stride = h->picture.linesize[1];
+    h->cy           = h->picture.f.data[0];
+    h->cu           = h->picture.f.data[1];
+    h->cv           = h->picture.f.data[2];
+    h->l_stride     = h->picture.f.linesize[0];
+    h->c_stride     = h->picture.f.linesize[1];
     h->luma_scan[2] = 8*h->l_stride;
     h->luma_scan[3] = 8*h->l_stride+8;
     h->mbx = h->mby = h->mbidx = 0;
     h->luma_scan[2] = 8*h->l_stride;
     h->luma_scan[3] = 8*h->l_stride+8;
     h->mbx = h->mby = h->mbidx = 0;
@@ -658,8 +657,8 @@ void ff_cavs_init_top_lines(AVSContext *h) {
     h->top_mv[1]    = av_malloc((h->mb_width*2+1)*sizeof(cavs_vector));
     h->top_pred_Y   = av_malloc( h->mb_width*2*sizeof(*h->top_pred_Y));
     h->top_border_y = av_malloc((h->mb_width+1)*16);
     h->top_mv[1]    = av_malloc((h->mb_width*2+1)*sizeof(cavs_vector));
     h->top_pred_Y   = av_malloc( h->mb_width*2*sizeof(*h->top_pred_Y));
     h->top_border_y = av_malloc((h->mb_width+1)*16);
-    h->top_border_u = av_malloc((h->mb_width)*10);
-    h->top_border_v = av_malloc((h->mb_width)*10);
+    h->top_border_u = av_malloc( h->mb_width * 10);
+    h->top_border_v = av_malloc( h->mb_width * 10);
 
     /* alloc space for co-located MVs and types */
     h->col_mv       = av_malloc( h->mb_width*h->mb_height*4*sizeof(cavs_vector));
 
     /* alloc space for co-located MVs and types */
     h->col_mv       = av_malloc( h->mb_width*h->mb_height*4*sizeof(cavs_vector));
@@ -671,7 +670,8 @@ av_cold int ff_cavs_init(AVCodecContext *avctx) {
     AVSContext *h = avctx->priv_data;
     MpegEncContext * const s = &h->s;
 
     AVSContext *h = avctx->priv_data;
     MpegEncContext * const s = &h->s;
 
-    MPV_decode_defaults(s);
+    ff_MPV_decode_defaults(s);
+    ff_cavsdsp_init(&h->cdsp, avctx);
     s->avctx = avctx;
 
     avctx->pix_fmt= PIX_FMT_YUV420P;
     s->avctx = avctx;
 
     avctx->pix_fmt= PIX_FMT_YUV420P;