h264: use templates to avoid excessive inlining
authorMans Rullgard <mans@mansr.com>
Tue, 3 Jul 2012 22:16:11 +0000 (23:16 +0100)
committerMans Rullgard <mans@mansr.com>
Thu, 5 Jul 2012 10:50:18 +0000 (11:50 +0100)
Instead of inlining everything into ff_h264_hl_decode_mb(), use
explicit templating to create versions of the called functions
with constant parameters filled in.  This greatly speeds up
compilation of h264.c and reduces the code size without any
measurable impact on performance.

Compilation time for h264.c on an i7 goes from 30s to 5.5s.
Code size is reduced by 430kB.

Signed-off-by: Mans Rullgard <mans@mansr.com>
libavcodec/h264.c
libavcodec/h264_mb_template.c [new file with mode: 0644]
libavcodec/h264_mc_template.c [new file with mode: 0644]

index 2d6a08e..a4afcc8 100644 (file)
@@ -714,33 +714,6 @@ static av_always_inline void mc_part_weighted(H264Context *h, int n, int square,
     }
 }
 
-static av_always_inline void mc_part(H264Context *h, int n, int square,
-                                     int 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,
-                                     h264_weight_func *weight_op,
-                                     h264_biweight_func *weight_avg,
-                                     int list0, int list1,
-                                     int pixel_shift, int chroma_idc)
-{
-    if ((h->use_weight == 2 && list0 && list1 &&
-         (h->implicit_weight[h->ref_cache[0][scan8[n]]][h->ref_cache[1][scan8[n]]][h->s.mb_y & 1] != 32)) ||
-        h->use_weight == 1)
-        mc_part_weighted(h, n, square, height, delta, dest_y, dest_cb, dest_cr,
-                         x_offset, y_offset, qpix_put, chroma_put,
-                         weight_op[0], weight_op[1], weight_avg[0],
-                         weight_avg[1], list0, list1, pixel_shift, chroma_idc);
-    else
-        mc_part_std(h, n, square, height, delta, dest_y, dest_cb, dest_cr,
-                    x_offset, y_offset, qpix_put, chroma_put, qpix_avg,
-                    chroma_avg, list0, list1, pixel_shift, chroma_idc);
-}
-
 static av_always_inline void prefetch_motion(H264Context *h, int list,
                                              int pixel_shift, int chroma_idc)
 {
@@ -768,146 +741,6 @@ static av_always_inline void prefetch_motion(H264Context *h, int list,
     }
 }
 
-static av_always_inline void hl_motion(H264Context *h, uint8_t *dest_y,
-                                       uint8_t *dest_cb, uint8_t *dest_cr,
-                                       qpel_mc_func(*qpix_put)[16],
-                                       h264_chroma_mc_func(*chroma_put),
-                                       qpel_mc_func(*qpix_avg)[16],
-                                       h264_chroma_mc_func(*chroma_avg),
-                                       h264_weight_func *weight_op,
-                                       h264_biweight_func *weight_avg,
-                                       int pixel_shift, int chroma_idc)
-{
-    MpegEncContext *const s = &h->s;
-    const int mb_xy   = h->mb_xy;
-    const int mb_type = s->current_picture.f.mb_type[mb_xy];
-
-    assert(IS_INTER(mb_type));
-
-    if (HAVE_THREADS && (s->avctx->active_thread_type & FF_THREAD_FRAME))
-        await_references(h);
-    prefetch_motion(h, 0, pixel_shift, chroma_idc);
-
-    if (IS_16X16(mb_type)) {
-        mc_part(h, 0, 1, 16, 0, dest_y, dest_cb, dest_cr, 0, 0,
-                qpix_put[0], chroma_put[0], qpix_avg[0], chroma_avg[0],
-                weight_op, weight_avg,
-                IS_DIR(mb_type, 0, 0), IS_DIR(mb_type, 0, 1),
-                pixel_shift, chroma_idc);
-    } else if (IS_16X8(mb_type)) {
-        mc_part(h, 0, 0, 8, 8 << pixel_shift, dest_y, dest_cb, dest_cr, 0, 0,
-                qpix_put[1], chroma_put[0], qpix_avg[1], chroma_avg[0],
-                weight_op, weight_avg,
-                IS_DIR(mb_type, 0, 0), IS_DIR(mb_type, 0, 1),
-                pixel_shift, chroma_idc);
-        mc_part(h, 8, 0, 8, 8 << pixel_shift, dest_y, dest_cb, dest_cr, 0, 4,
-                qpix_put[1], chroma_put[0], qpix_avg[1], chroma_avg[0],
-                weight_op, weight_avg,
-                IS_DIR(mb_type, 1, 0), IS_DIR(mb_type, 1, 1),
-                pixel_shift, chroma_idc);
-    } else if (IS_8X16(mb_type)) {
-        mc_part(h, 0, 0, 16, 8 * h->mb_linesize, dest_y, dest_cb, dest_cr, 0, 0,
-                qpix_put[1], chroma_put[1], qpix_avg[1], chroma_avg[1],
-                &weight_op[1], &weight_avg[1],
-                IS_DIR(mb_type, 0, 0), IS_DIR(mb_type, 0, 1),
-                pixel_shift, chroma_idc);
-        mc_part(h, 4, 0, 16, 8 * h->mb_linesize, dest_y, dest_cb, dest_cr, 4, 0,
-                qpix_put[1], chroma_put[1], qpix_avg[1], chroma_avg[1],
-                &weight_op[1], &weight_avg[1],
-                IS_DIR(mb_type, 1, 0), IS_DIR(mb_type, 1, 1),
-                pixel_shift, chroma_idc);
-    } else {
-        int i;
-
-        assert(IS_8X8(mb_type));
-
-        for (i = 0; i < 4; i++) {
-            const int sub_mb_type = h->sub_mb_type[i];
-            const int n  = 4 * i;
-            int x_offset = (i & 1) << 2;
-            int y_offset = (i & 2) << 1;
-
-            if (IS_SUB_8X8(sub_mb_type)) {
-                mc_part(h, n, 1, 8, 0, dest_y, dest_cb, dest_cr,
-                        x_offset, y_offset,
-                        qpix_put[1], chroma_put[1], qpix_avg[1], chroma_avg[1],
-                        &weight_op[1], &weight_avg[1],
-                        IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1),
-                        pixel_shift, chroma_idc);
-            } else if (IS_SUB_8X4(sub_mb_type)) {
-                mc_part(h, n, 0, 4, 4 << pixel_shift, dest_y, dest_cb, dest_cr,
-                        x_offset, y_offset,
-                        qpix_put[2], chroma_put[1], qpix_avg[2], chroma_avg[1],
-                        &weight_op[1], &weight_avg[1],
-                        IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1),
-                        pixel_shift, chroma_idc);
-                mc_part(h, n + 2, 0, 4, 4 << pixel_shift,
-                        dest_y, dest_cb, dest_cr, x_offset, y_offset + 2,
-                        qpix_put[2], chroma_put[1], qpix_avg[2], chroma_avg[1],
-                        &weight_op[1], &weight_avg[1],
-                        IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1),
-                        pixel_shift, chroma_idc);
-            } else if (IS_SUB_4X8(sub_mb_type)) {
-                mc_part(h, n, 0, 8, 4 * h->mb_linesize,
-                        dest_y, dest_cb, dest_cr, x_offset, y_offset,
-                        qpix_put[2], chroma_put[2], qpix_avg[2], chroma_avg[2],
-                        &weight_op[2], &weight_avg[2],
-                        IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1),
-                        pixel_shift, chroma_idc);
-                mc_part(h, n + 1, 0, 8, 4 * h->mb_linesize,
-                        dest_y, dest_cb, dest_cr, x_offset + 2, y_offset,
-                        qpix_put[2], chroma_put[2], qpix_avg[2], chroma_avg[2],
-                        &weight_op[2], &weight_avg[2],
-                        IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1),
-                        pixel_shift, chroma_idc);
-            } else {
-                int j;
-                assert(IS_SUB_4X4(sub_mb_type));
-                for (j = 0; j < 4; j++) {
-                    int sub_x_offset = x_offset + 2 * (j & 1);
-                    int sub_y_offset = y_offset + (j & 2);
-                    mc_part(h, n + j, 1, 4, 0,
-                            dest_y, dest_cb, dest_cr, sub_x_offset, sub_y_offset,
-                            qpix_put[2], chroma_put[2], qpix_avg[2], chroma_avg[2],
-                            &weight_op[2], &weight_avg[2],
-                            IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1),
-                            pixel_shift, chroma_idc);
-                }
-            }
-        }
-    }
-
-    prefetch_motion(h, 1, pixel_shift, chroma_idc);
-}
-
-static av_always_inline void hl_motion_420(H264Context *h, uint8_t *dest_y,
-                                           uint8_t *dest_cb, uint8_t *dest_cr,
-                                           qpel_mc_func(*qpix_put)[16],
-                                           h264_chroma_mc_func(*chroma_put),
-                                           qpel_mc_func(*qpix_avg)[16],
-                                           h264_chroma_mc_func(*chroma_avg),
-                                           h264_weight_func *weight_op,
-                                           h264_biweight_func *weight_avg,
-                                           int pixel_shift)
-{
-    hl_motion(h, dest_y, dest_cb, dest_cr, qpix_put, chroma_put,
-              qpix_avg, chroma_avg, weight_op, weight_avg, pixel_shift, 1);
-}
-
-static av_always_inline void hl_motion_422(H264Context *h, uint8_t *dest_y,
-                                           uint8_t *dest_cb, uint8_t *dest_cr,
-                                           qpel_mc_func(*qpix_put)[16],
-                                           h264_chroma_mc_func(*chroma_put),
-                                           qpel_mc_func(*qpix_avg)[16],
-                                           h264_chroma_mc_func(*chroma_avg),
-                                           h264_weight_func *weight_op,
-                                           h264_biweight_func *weight_avg,
-                                           int pixel_shift)
-{
-    hl_motion(h, dest_y, dest_cb, dest_cr, qpix_put, chroma_put,
-              qpix_avg, chroma_avg, weight_op, weight_avg, pixel_shift, 2);
-}
-
 static void free_tables(H264Context *h, int free_rbsp)
 {
     int i;
@@ -2077,373 +1910,17 @@ static av_always_inline void hl_decode_mb_idct_luma(H264Context *h, int mb_type,
     }
 }
 
-static av_always_inline void hl_decode_mb_internal(H264Context *h, int simple,
-                                                   int pixel_shift)
-{
-    MpegEncContext *const s = &h->s;
-    const int mb_x    = s->mb_x;
-    const int mb_y    = s->mb_y;
-    const int mb_xy   = h->mb_xy;
-    const int mb_type = s->current_picture.f.mb_type[mb_xy];
-    uint8_t *dest_y, *dest_cb, *dest_cr;
-    int linesize, uvlinesize /*dct_offset*/;
-    int i, j;
-    int *block_offset = &h->block_offset[0];
-    const int transform_bypass = !simple && (s->qscale == 0 && h->sps.transform_bypass);
-    /* is_h264 should always be true if SVQ3 is disabled. */
-    const int is_h264 = !CONFIG_SVQ3_DECODER || simple || s->codec_id == CODEC_ID_H264;
-    void (*idct_add)(uint8_t *dst, DCTELEM *block, int stride);
-    const int block_h   = 16 >> s->chroma_y_shift;
-    const int chroma422 = CHROMA422;
-
-    dest_y  = s->current_picture.f.data[0] + ((mb_x << pixel_shift)     + mb_y * s->linesize)  * 16;
-    dest_cb = s->current_picture.f.data[1] +  (mb_x << pixel_shift) * 8 + mb_y * s->uvlinesize * block_h;
-    dest_cr = s->current_picture.f.data[2] +  (mb_x << pixel_shift) * 8 + mb_y * s->uvlinesize * block_h;
-
-    s->dsp.prefetch(dest_y  + (s->mb_x & 3) * 4 * s->linesize   + (64 << pixel_shift), s->linesize,       4);
-    s->dsp.prefetch(dest_cb + (s->mb_x & 7)     * s->uvlinesize + (64 << pixel_shift), dest_cr - dest_cb, 2);
-
-    h->list_counts[mb_xy] = h->list_count;
-
-    if (!simple && MB_FIELD) {
-        linesize     = h->mb_linesize = s->linesize * 2;
-        uvlinesize   = h->mb_uvlinesize = s->uvlinesize * 2;
-        block_offset = &h->block_offset[48];
-        if (mb_y & 1) { // FIXME move out of this function?
-            dest_y  -= s->linesize * 15;
-            dest_cb -= s->uvlinesize * (block_h - 1);
-            dest_cr -= s->uvlinesize * (block_h - 1);
-        }
-        if (FRAME_MBAFF) {
-            int list;
-            for (list = 0; list < h->list_count; list++) {
-                if (!USES_LIST(mb_type, list))
-                    continue;
-                if (IS_16X16(mb_type)) {
-                    int8_t *ref = &h->ref_cache[list][scan8[0]];
-                    fill_rectangle(ref, 4, 4, 8, (16 + *ref) ^ (s->mb_y & 1), 1);
-                } else {
-                    for (i = 0; i < 16; i += 4) {
-                        int ref = h->ref_cache[list][scan8[i]];
-                        if (ref >= 0)
-                            fill_rectangle(&h->ref_cache[list][scan8[i]], 2, 2,
-                                           8, (16 + ref) ^ (s->mb_y & 1), 1);
-                    }
-                }
-            }
-        }
-    } else {
-        linesize   = h->mb_linesize   = s->linesize;
-        uvlinesize = h->mb_uvlinesize = s->uvlinesize;
-        // dct_offset = s->linesize * 16;
-    }
-
-    if (!simple && IS_INTRA_PCM(mb_type)) {
-        if (pixel_shift) {
-            const int bit_depth = h->sps.bit_depth_luma;
-            int j;
-            GetBitContext gb;
-            init_get_bits(&gb, (uint8_t *)h->mb,
-                          ff_h264_mb_sizes[h->sps.chroma_format_idc] * bit_depth);
-
-            for (i = 0; i < 16; i++) {
-                uint16_t *tmp_y = (uint16_t *)(dest_y + i * linesize);
-                for (j = 0; j < 16; j++)
-                    tmp_y[j] = get_bits(&gb, bit_depth);
-            }
-            if (simple || !CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY)) {
-                if (!h->sps.chroma_format_idc) {
-                    for (i = 0; i < block_h; i++) {
-                        uint16_t *tmp_cb = (uint16_t *)(dest_cb + i * uvlinesize);
-                        for (j = 0; j < 8; j++)
-                            tmp_cb[j] = 1 << (bit_depth - 1);
-                    }
-                    for (i = 0; i < block_h; i++) {
-                        uint16_t *tmp_cr = (uint16_t *)(dest_cr + i * uvlinesize);
-                        for (j = 0; j < 8; j++)
-                            tmp_cr[j] = 1 << (bit_depth - 1);
-                    }
-                } else {
-                    for (i = 0; i < block_h; i++) {
-                        uint16_t *tmp_cb = (uint16_t *)(dest_cb + i * uvlinesize);
-                        for (j = 0; j < 8; j++)
-                            tmp_cb[j] = get_bits(&gb, bit_depth);
-                    }
-                    for (i = 0; i < block_h; i++) {
-                        uint16_t *tmp_cr = (uint16_t *)(dest_cr + i * uvlinesize);
-                        for (j = 0; j < 8; j++)
-                            tmp_cr[j] = get_bits(&gb, bit_depth);
-                    }
-                }
-            }
-        } else {
-            for (i = 0; i < 16; i++)
-                memcpy(dest_y + i * linesize, (uint8_t *)h->mb + i * 16, 16);
-            if (simple || !CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY)) {
-                if (!h->sps.chroma_format_idc) {
-                    for (i = 0; i < block_h; i++) {
-                        memset(dest_cb + i * uvlinesize, 128, 8);
-                        memset(dest_cr + i * uvlinesize, 128, 8);
-                    }
-                } else {
-                    uint8_t *src_cb = (uint8_t *)h->mb + 256;
-                    uint8_t *src_cr = (uint8_t *)h->mb + 256 + block_h * 8;
-                    for (i = 0; i < block_h; i++) {
-                        memcpy(dest_cb + i * uvlinesize, src_cb + i * 8, 8);
-                        memcpy(dest_cr + i * uvlinesize, src_cr + i * 8, 8);
-                    }
-                }
-            }
-        }
-    } else {
-        if (IS_INTRA(mb_type)) {
-            if (h->deblocking_filter)
-                xchg_mb_border(h, dest_y, dest_cb, dest_cr, linesize,
-                               uvlinesize, 1, 0, simple, pixel_shift);
-
-            if (simple || !CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY)) {
-                h->hpc.pred8x8[h->chroma_pred_mode](dest_cb, uvlinesize);
-                h->hpc.pred8x8[h->chroma_pred_mode](dest_cr, uvlinesize);
-            }
-
-            hl_decode_mb_predict_luma(h, mb_type, is_h264, simple,
-                                      transform_bypass, pixel_shift,
-                                      block_offset, linesize, dest_y, 0);
-
-            if (h->deblocking_filter)
-                xchg_mb_border(h, dest_y, dest_cb, dest_cr, linesize,
-                               uvlinesize, 0, 0, simple, pixel_shift);
-        } else if (is_h264) {
-            if (chroma422) {
-                hl_motion_422(h, dest_y, dest_cb, dest_cr,
-                              s->me.qpel_put, s->dsp.put_h264_chroma_pixels_tab,
-                              s->me.qpel_avg, s->dsp.avg_h264_chroma_pixels_tab,
-                              h->h264dsp.weight_h264_pixels_tab,
-                              h->h264dsp.biweight_h264_pixels_tab,
-                              pixel_shift);
-            } else {
-                hl_motion_420(h, dest_y, dest_cb, dest_cr,
-                              s->me.qpel_put, s->dsp.put_h264_chroma_pixels_tab,
-                              s->me.qpel_avg, s->dsp.avg_h264_chroma_pixels_tab,
-                              h->h264dsp.weight_h264_pixels_tab,
-                              h->h264dsp.biweight_h264_pixels_tab,
-                              pixel_shift);
-            }
-        }
-
-        hl_decode_mb_idct_luma(h, mb_type, is_h264, simple, transform_bypass,
-                               pixel_shift, block_offset, linesize, dest_y, 0);
-
-        if ((simple || !CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY)) &&
-            (h->cbp & 0x30)) {
-            uint8_t *dest[2] = { dest_cb, dest_cr };
-            if (transform_bypass) {
-                if (IS_INTRA(mb_type) && h->sps.profile_idc == 244 &&
-                    (h->chroma_pred_mode == VERT_PRED8x8 ||
-                     h->chroma_pred_mode == HOR_PRED8x8)) {
-                    h->hpc.pred8x8_add[h->chroma_pred_mode](dest[0],
-                                                            block_offset + 16,
-                                                            h->mb + (16 * 16 * 1 << pixel_shift),
-                                                            uvlinesize);
-                    h->hpc.pred8x8_add[h->chroma_pred_mode](dest[1],
-                                                            block_offset + 32,
-                                                            h->mb + (16 * 16 * 2 << pixel_shift),
-                                                            uvlinesize);
-                } else {
-                    idct_add = s->dsp.add_pixels4;
-                    for (j = 1; j < 3; j++) {
-                        for (i = j * 16; i < j * 16 + 4; i++)
-                            if (h->non_zero_count_cache[scan8[i]] ||
-                                dctcoef_get(h->mb, pixel_shift, i * 16))
-                                idct_add(dest[j - 1] + block_offset[i],
-                                         h->mb + (i * 16 << pixel_shift),
-                                         uvlinesize);
-                        if (chroma422) {
-                            for (i = j * 16 + 4; i < j * 16 + 8; i++)
-                                if (h->non_zero_count_cache[scan8[i + 4]] ||
-                                    dctcoef_get(h->mb, pixel_shift, i * 16))
-                                    idct_add(dest[j - 1] + block_offset[i + 4],
-                                             h->mb + (i * 16 << pixel_shift),
-                                             uvlinesize);
-                        }
-                    }
-                }
-            } else {
-                if (is_h264) {
-                    int qp[2];
-                    if (chroma422) {
-                        qp[0] = h->chroma_qp[0] + 3;
-                        qp[1] = h->chroma_qp[1] + 3;
-                    } else {
-                        qp[0] = h->chroma_qp[0];
-                        qp[1] = h->chroma_qp[1];
-                    }
-                    if (h->non_zero_count_cache[scan8[CHROMA_DC_BLOCK_INDEX + 0]])
-                        h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + (16 * 16 * 1 << pixel_shift),
-                                                               h->dequant4_coeff[IS_INTRA(mb_type) ? 1 : 4][qp[0]][0]);
-                    if (h->non_zero_count_cache[scan8[CHROMA_DC_BLOCK_INDEX + 1]])
-                        h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + (16 * 16 * 2 << pixel_shift),
-                                                               h->dequant4_coeff[IS_INTRA(mb_type) ? 2 : 5][qp[1]][0]);
-                    h->h264dsp.h264_idct_add8(dest, block_offset,
-                                              h->mb, uvlinesize,
-                                              h->non_zero_count_cache);
-                } else if (CONFIG_SVQ3_DECODER) {
-                    h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + 16 * 16 * 1,
-                                                           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 * 2,
-                                                           h->dequant4_coeff[IS_INTRA(mb_type) ? 2 : 5][h->chroma_qp[1]][0]);
-                    for (j = 1; j < 3; j++) {
-                        for (i = j * 16; i < j * 16 + 4; i++)
-                            if (h->non_zero_count_cache[scan8[i]] || h->mb[i * 16]) {
-                                uint8_t *const ptr = dest[j - 1] + block_offset[i];
-                                ff_svq3_add_idct_c(ptr, h->mb + i * 16,
-                                                   uvlinesize,
-                                                   ff_h264_chroma_qp[0][s->qscale + 12] - 12, 2);
-                            }
-                    }
-                }
-            }
-        }
-    }
-    if (h->cbp || IS_INTRA(mb_type)) {
-        s->dsp.clear_blocks(h->mb);
-        s->dsp.clear_blocks(h->mb + (24 * 16 << pixel_shift));
-    }
-}
-
-static av_always_inline void hl_decode_mb_444_internal(H264Context *h,
-                                                       int simple,
-                                                       int pixel_shift)
-{
-    MpegEncContext *const s = &h->s;
-    const int mb_x    = s->mb_x;
-    const int mb_y    = s->mb_y;
-    const int mb_xy   = h->mb_xy;
-    const int mb_type = s->current_picture.f.mb_type[mb_xy];
-    uint8_t *dest[3];
-    int linesize;
-    int i, j, p;
-    int *block_offset = &h->block_offset[0];
-    const int transform_bypass = !simple && (s->qscale == 0 && h->sps.transform_bypass);
-    const int plane_count      = (simple || !CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY)) ? 3 : 1;
-
-    for (p = 0; p < plane_count; p++) {
-        dest[p] = s->current_picture.f.data[p] +
-                  ((mb_x << pixel_shift) + mb_y * s->linesize) * 16;
-        s->dsp.prefetch(dest[p] + (s->mb_x & 3) * 4 * s->linesize + (64 << pixel_shift),
-                        s->linesize, 4);
-    }
-
-    h->list_counts[mb_xy] = h->list_count;
-
-    if (!simple && MB_FIELD) {
-        linesize     = h->mb_linesize = h->mb_uvlinesize = s->linesize * 2;
-        block_offset = &h->block_offset[48];
-        if (mb_y & 1) // FIXME move out of this function?
-            for (p = 0; p < 3; p++)
-                dest[p] -= s->linesize * 15;
-        if (FRAME_MBAFF) {
-            int list;
-            for (list = 0; list < h->list_count; list++) {
-                if (!USES_LIST(mb_type, list))
-                    continue;
-                if (IS_16X16(mb_type)) {
-                    int8_t *ref = &h->ref_cache[list][scan8[0]];
-                    fill_rectangle(ref, 4, 4, 8, (16 + *ref) ^ (s->mb_y & 1), 1);
-                } else {
-                    for (i = 0; i < 16; i += 4) {
-                        int ref = h->ref_cache[list][scan8[i]];
-                        if (ref >= 0)
-                            fill_rectangle(&h->ref_cache[list][scan8[i]], 2, 2,
-                                           8, (16 + ref) ^ (s->mb_y & 1), 1);
-                    }
-                }
-            }
-        }
-    } else {
-        linesize = h->mb_linesize = h->mb_uvlinesize = s->linesize;
-    }
-
-    if (!simple && IS_INTRA_PCM(mb_type)) {
-        if (pixel_shift) {
-            const int bit_depth = h->sps.bit_depth_luma;
-            GetBitContext gb;
-            init_get_bits(&gb, (uint8_t *)h->mb, 768 * bit_depth);
-
-            for (p = 0; p < plane_count; p++)
-                for (i = 0; i < 16; i++) {
-                    uint16_t *tmp = (uint16_t *)(dest[p] + i * linesize);
-                    for (j = 0; j < 16; j++)
-                        tmp[j] = get_bits(&gb, bit_depth);
-                }
-        } else {
-            for (p = 0; p < plane_count; p++)
-                for (i = 0; i < 16; i++)
-                    memcpy(dest[p] + i * linesize,
-                           (uint8_t *)h->mb + p * 256 + i * 16, 16);
-        }
-    } else {
-        if (IS_INTRA(mb_type)) {
-            if (h->deblocking_filter)
-                xchg_mb_border(h, dest[0], dest[1], dest[2], linesize,
-                               linesize, 1, 1, simple, pixel_shift);
-
-            for (p = 0; p < plane_count; p++)
-                hl_decode_mb_predict_luma(h, mb_type, 1, simple,
-                                          transform_bypass, pixel_shift,
-                                          block_offset, linesize, dest[p], p);
-
-            if (h->deblocking_filter)
-                xchg_mb_border(h, dest[0], dest[1], dest[2], linesize,
-                               linesize, 0, 1, simple, pixel_shift);
-        } else {
-            hl_motion(h, dest[0], dest[1], dest[2],
-                      s->me.qpel_put, s->dsp.put_h264_chroma_pixels_tab,
-                      s->me.qpel_avg, s->dsp.avg_h264_chroma_pixels_tab,
-                      h->h264dsp.weight_h264_pixels_tab,
-                      h->h264dsp.biweight_h264_pixels_tab, pixel_shift, 3);
-        }
-
-        for (p = 0; p < plane_count; p++)
-            hl_decode_mb_idct_luma(h, mb_type, 1, simple, transform_bypass,
-                                   pixel_shift, block_offset, linesize,
-                                   dest[p], p);
-    }
-    if (h->cbp || IS_INTRA(mb_type)) {
-        s->dsp.clear_blocks(h->mb);
-        s->dsp.clear_blocks(h->mb + (24 * 16 << pixel_shift));
-    }
-}
-
-/**
- * Process a macroblock; this case avoids checks for expensive uncommon cases.
- */
-#define hl_decode_mb_simple(sh, bits)                          \
-static void hl_decode_mb_simple_ ## bits(H264Context *h)       \
-{                                                              \
-    hl_decode_mb_internal(h, 1, sh);                           \
-}
+#define BITS   8
+#define SIMPLE 1
+#include "h264_mb_template.c"
 
-hl_decode_mb_simple(0, 8)
-hl_decode_mb_simple(1, 16)
+#undef  BITS
+#define BITS   16
+#include "h264_mb_template.c"
 
-/**
- * Process a macroblock; this handles edge cases, such as interlacing.
- */
-static av_noinline void hl_decode_mb_complex(H264Context *h)
-{
-    hl_decode_mb_internal(h, 0, h->pixel_shift);
-}
-
-static av_noinline void hl_decode_mb_444_complex(H264Context *h)
-{
-    hl_decode_mb_444_internal(h, 0, h->pixel_shift);
-}
-
-static av_noinline void hl_decode_mb_444_simple(H264Context *h)
-{
-    hl_decode_mb_444_internal(h, 1, 0);
-}
+#undef  SIMPLE
+#define SIMPLE 0
+#include "h264_mb_template.c"
 
 void ff_h264_hl_decode_mb(H264Context *h)
 {
@@ -2456,7 +1933,7 @@ void ff_h264_hl_decode_mb(H264Context *h)
         if (is_complex || h->pixel_shift)
             hl_decode_mb_444_complex(h);
         else
-            hl_decode_mb_444_simple(h);
+            hl_decode_mb_444_simple_8(h);
     } else if (is_complex) {
         hl_decode_mb_complex(h);
     } else if (h->pixel_shift) {
diff --git a/libavcodec/h264_mb_template.c b/libavcodec/h264_mb_template.c
new file mode 100644 (file)
index 0000000..b7856cb
--- /dev/null
@@ -0,0 +1,380 @@
+/*
+ * H.26L/H.264/AVC/JVT/14496-10/... decoder
+ * Copyright (c) 2003 Michael Niedermayer <michaelni@gmx.at>
+ *
+ * This file is part of Libav.
+ *
+ * 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.
+ *
+ * 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
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#undef FUNC
+#undef PIXEL_SHIFT
+
+#if SIMPLE
+#   define FUNC(n) AV_JOIN(n ## _simple_, BITS)
+#   define PIXEL_SHIFT (BITS >> 4)
+#else
+#   define FUNC(n) n ## _complex
+#   define PIXEL_SHIFT h->pixel_shift
+#endif
+
+#undef  CHROMA_IDC
+#define CHROMA_IDC 1
+#include "h264_mc_template.c"
+
+#undef  CHROMA_IDC
+#define CHROMA_IDC 2
+#include "h264_mc_template.c"
+
+static av_noinline void FUNC(hl_decode_mb)(H264Context *h)
+{
+    MpegEncContext *const s = &h->s;
+    const int mb_x    = s->mb_x;
+    const int mb_y    = s->mb_y;
+    const int mb_xy   = h->mb_xy;
+    const int mb_type = s->current_picture.f.mb_type[mb_xy];
+    uint8_t *dest_y, *dest_cb, *dest_cr;
+    int linesize, uvlinesize /*dct_offset*/;
+    int i, j;
+    int *block_offset = &h->block_offset[0];
+    const int transform_bypass = !SIMPLE && (s->qscale == 0 && h->sps.transform_bypass);
+    /* is_h264 should always be true if SVQ3 is disabled. */
+    const int is_h264 = !CONFIG_SVQ3_DECODER || SIMPLE || s->codec_id == CODEC_ID_H264;
+    void (*idct_add)(uint8_t *dst, DCTELEM *block, int stride);
+    const int block_h   = 16 >> s->chroma_y_shift;
+    const int chroma422 = CHROMA422;
+
+    dest_y  = s->current_picture.f.data[0] + ((mb_x << PIXEL_SHIFT)     + mb_y * s->linesize)  * 16;
+    dest_cb = s->current_picture.f.data[1] +  (mb_x << PIXEL_SHIFT) * 8 + mb_y * s->uvlinesize * block_h;
+    dest_cr = s->current_picture.f.data[2] +  (mb_x << PIXEL_SHIFT) * 8 + mb_y * s->uvlinesize * block_h;
+
+    s->dsp.prefetch(dest_y  + (s->mb_x & 3) * 4 * s->linesize   + (64 << PIXEL_SHIFT), s->linesize,       4);
+    s->dsp.prefetch(dest_cb + (s->mb_x & 7)     * s->uvlinesize + (64 << PIXEL_SHIFT), dest_cr - dest_cb, 2);
+
+    h->list_counts[mb_xy] = h->list_count;
+
+    if (!SIMPLE && MB_FIELD) {
+        linesize     = h->mb_linesize = s->linesize * 2;
+        uvlinesize   = h->mb_uvlinesize = s->uvlinesize * 2;
+        block_offset = &h->block_offset[48];
+        if (mb_y & 1) { // FIXME move out of this function?
+            dest_y  -= s->linesize * 15;
+            dest_cb -= s->uvlinesize * (block_h - 1);
+            dest_cr -= s->uvlinesize * (block_h - 1);
+        }
+        if (FRAME_MBAFF) {
+            int list;
+            for (list = 0; list < h->list_count; list++) {
+                if (!USES_LIST(mb_type, list))
+                    continue;
+                if (IS_16X16(mb_type)) {
+                    int8_t *ref = &h->ref_cache[list][scan8[0]];
+                    fill_rectangle(ref, 4, 4, 8, (16 + *ref) ^ (s->mb_y & 1), 1);
+                } else {
+                    for (i = 0; i < 16; i += 4) {
+                        int ref = h->ref_cache[list][scan8[i]];
+                        if (ref >= 0)
+                            fill_rectangle(&h->ref_cache[list][scan8[i]], 2, 2,
+                                           8, (16 + ref) ^ (s->mb_y & 1), 1);
+                    }
+                }
+            }
+        }
+    } else {
+        linesize   = h->mb_linesize   = s->linesize;
+        uvlinesize = h->mb_uvlinesize = s->uvlinesize;
+        // dct_offset = s->linesize * 16;
+    }
+
+    if (!SIMPLE && IS_INTRA_PCM(mb_type)) {
+        if (PIXEL_SHIFT) {
+            const int bit_depth = h->sps.bit_depth_luma;
+            int j;
+            GetBitContext gb;
+            init_get_bits(&gb, (uint8_t *)h->mb,
+                          ff_h264_mb_sizes[h->sps.chroma_format_idc] * bit_depth);
+
+            for (i = 0; i < 16; i++) {
+                uint16_t *tmp_y = (uint16_t *)(dest_y + i * linesize);
+                for (j = 0; j < 16; j++)
+                    tmp_y[j] = get_bits(&gb, bit_depth);
+            }
+            if (SIMPLE || !CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY)) {
+                if (!h->sps.chroma_format_idc) {
+                    for (i = 0; i < block_h; i++) {
+                        uint16_t *tmp_cb = (uint16_t *)(dest_cb + i * uvlinesize);
+                        for (j = 0; j < 8; j++)
+                            tmp_cb[j] = 1 << (bit_depth - 1);
+                    }
+                    for (i = 0; i < block_h; i++) {
+                        uint16_t *tmp_cr = (uint16_t *)(dest_cr + i * uvlinesize);
+                        for (j = 0; j < 8; j++)
+                            tmp_cr[j] = 1 << (bit_depth - 1);
+                    }
+                } else {
+                    for (i = 0; i < block_h; i++) {
+                        uint16_t *tmp_cb = (uint16_t *)(dest_cb + i * uvlinesize);
+                        for (j = 0; j < 8; j++)
+                            tmp_cb[j] = get_bits(&gb, bit_depth);
+                    }
+                    for (i = 0; i < block_h; i++) {
+                        uint16_t *tmp_cr = (uint16_t *)(dest_cr + i * uvlinesize);
+                        for (j = 0; j < 8; j++)
+                            tmp_cr[j] = get_bits(&gb, bit_depth);
+                    }
+                }
+            }
+        } else {
+            for (i = 0; i < 16; i++)
+                memcpy(dest_y + i * linesize, (uint8_t *)h->mb + i * 16, 16);
+            if (SIMPLE || !CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY)) {
+                if (!h->sps.chroma_format_idc) {
+                    for (i = 0; i < block_h; i++) {
+                        memset(dest_cb + i * uvlinesize, 128, 8);
+                        memset(dest_cr + i * uvlinesize, 128, 8);
+                    }
+                } else {
+                    uint8_t *src_cb = (uint8_t *)h->mb + 256;
+                    uint8_t *src_cr = (uint8_t *)h->mb + 256 + block_h * 8;
+                    for (i = 0; i < block_h; i++) {
+                        memcpy(dest_cb + i * uvlinesize, src_cb + i * 8, 8);
+                        memcpy(dest_cr + i * uvlinesize, src_cr + i * 8, 8);
+                    }
+                }
+            }
+        }
+    } else {
+        if (IS_INTRA(mb_type)) {
+            if (h->deblocking_filter)
+                xchg_mb_border(h, dest_y, dest_cb, dest_cr, linesize,
+                               uvlinesize, 1, 0, SIMPLE, PIXEL_SHIFT);
+
+            if (SIMPLE || !CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY)) {
+                h->hpc.pred8x8[h->chroma_pred_mode](dest_cb, uvlinesize);
+                h->hpc.pred8x8[h->chroma_pred_mode](dest_cr, uvlinesize);
+            }
+
+            hl_decode_mb_predict_luma(h, mb_type, is_h264, SIMPLE,
+                                      transform_bypass, PIXEL_SHIFT,
+                                      block_offset, linesize, dest_y, 0);
+
+            if (h->deblocking_filter)
+                xchg_mb_border(h, dest_y, dest_cb, dest_cr, linesize,
+                               uvlinesize, 0, 0, SIMPLE, PIXEL_SHIFT);
+        } else if (is_h264) {
+            if (chroma422) {
+                FUNC(hl_motion_422)(h, dest_y, dest_cb, dest_cr,
+                              s->me.qpel_put, s->dsp.put_h264_chroma_pixels_tab,
+                              s->me.qpel_avg, s->dsp.avg_h264_chroma_pixels_tab,
+                              h->h264dsp.weight_h264_pixels_tab,
+                              h->h264dsp.biweight_h264_pixels_tab);
+            } else {
+                FUNC(hl_motion_420)(h, dest_y, dest_cb, dest_cr,
+                              s->me.qpel_put, s->dsp.put_h264_chroma_pixels_tab,
+                              s->me.qpel_avg, s->dsp.avg_h264_chroma_pixels_tab,
+                              h->h264dsp.weight_h264_pixels_tab,
+                              h->h264dsp.biweight_h264_pixels_tab);
+            }
+        }
+
+        hl_decode_mb_idct_luma(h, mb_type, is_h264, SIMPLE, transform_bypass,
+                               PIXEL_SHIFT, block_offset, linesize, dest_y, 0);
+
+        if ((SIMPLE || !CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY)) &&
+            (h->cbp & 0x30)) {
+            uint8_t *dest[2] = { dest_cb, dest_cr };
+            if (transform_bypass) {
+                if (IS_INTRA(mb_type) && h->sps.profile_idc == 244 &&
+                    (h->chroma_pred_mode == VERT_PRED8x8 ||
+                     h->chroma_pred_mode == HOR_PRED8x8)) {
+                    h->hpc.pred8x8_add[h->chroma_pred_mode](dest[0],
+                                                            block_offset + 16,
+                                                            h->mb + (16 * 16 * 1 << PIXEL_SHIFT),
+                                                            uvlinesize);
+                    h->hpc.pred8x8_add[h->chroma_pred_mode](dest[1],
+                                                            block_offset + 32,
+                                                            h->mb + (16 * 16 * 2 << PIXEL_SHIFT),
+                                                            uvlinesize);
+                } else {
+                    idct_add = s->dsp.add_pixels4;
+                    for (j = 1; j < 3; j++) {
+                        for (i = j * 16; i < j * 16 + 4; i++)
+                            if (h->non_zero_count_cache[scan8[i]] ||
+                                dctcoef_get(h->mb, PIXEL_SHIFT, i * 16))
+                                idct_add(dest[j - 1] + block_offset[i],
+                                         h->mb + (i * 16 << PIXEL_SHIFT),
+                                         uvlinesize);
+                        if (chroma422) {
+                            for (i = j * 16 + 4; i < j * 16 + 8; i++)
+                                if (h->non_zero_count_cache[scan8[i + 4]] ||
+                                    dctcoef_get(h->mb, PIXEL_SHIFT, i * 16))
+                                    idct_add(dest[j - 1] + block_offset[i + 4],
+                                             h->mb + (i * 16 << PIXEL_SHIFT),
+                                             uvlinesize);
+                        }
+                    }
+                }
+            } else {
+                if (is_h264) {
+                    int qp[2];
+                    if (chroma422) {
+                        qp[0] = h->chroma_qp[0] + 3;
+                        qp[1] = h->chroma_qp[1] + 3;
+                    } else {
+                        qp[0] = h->chroma_qp[0];
+                        qp[1] = h->chroma_qp[1];
+                    }
+                    if (h->non_zero_count_cache[scan8[CHROMA_DC_BLOCK_INDEX + 0]])
+                        h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + (16 * 16 * 1 << PIXEL_SHIFT),
+                                                               h->dequant4_coeff[IS_INTRA(mb_type) ? 1 : 4][qp[0]][0]);
+                    if (h->non_zero_count_cache[scan8[CHROMA_DC_BLOCK_INDEX + 1]])
+                        h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + (16 * 16 * 2 << PIXEL_SHIFT),
+                                                               h->dequant4_coeff[IS_INTRA(mb_type) ? 2 : 5][qp[1]][0]);
+                    h->h264dsp.h264_idct_add8(dest, block_offset,
+                                              h->mb, uvlinesize,
+                                              h->non_zero_count_cache);
+                } else if (CONFIG_SVQ3_DECODER) {
+                    h->h264dsp.h264_chroma_dc_dequant_idct(h->mb + 16 * 16 * 1,
+                                                           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 * 2,
+                                                           h->dequant4_coeff[IS_INTRA(mb_type) ? 2 : 5][h->chroma_qp[1]][0]);
+                    for (j = 1; j < 3; j++) {
+                        for (i = j * 16; i < j * 16 + 4; i++)
+                            if (h->non_zero_count_cache[scan8[i]] || h->mb[i * 16]) {
+                                uint8_t *const ptr = dest[j - 1] + block_offset[i];
+                                ff_svq3_add_idct_c(ptr, h->mb + i * 16,
+                                                   uvlinesize,
+                                                   ff_h264_chroma_qp[0][s->qscale + 12] - 12, 2);
+                            }
+                    }
+                }
+            }
+        }
+    }
+    if (h->cbp || IS_INTRA(mb_type)) {
+        s->dsp.clear_blocks(h->mb);
+        s->dsp.clear_blocks(h->mb + (24 * 16 << PIXEL_SHIFT));
+    }
+}
+
+#if !SIMPLE || BITS == 8
+
+#undef  CHROMA_IDC
+#define CHROMA_IDC 3
+#include "h264_mc_template.c"
+
+static av_noinline void FUNC(hl_decode_mb_444)(H264Context *h)
+{
+    MpegEncContext *const s = &h->s;
+    const int mb_x    = s->mb_x;
+    const int mb_y    = s->mb_y;
+    const int mb_xy   = h->mb_xy;
+    const int mb_type = s->current_picture.f.mb_type[mb_xy];
+    uint8_t *dest[3];
+    int linesize;
+    int i, j, p;
+    int *block_offset = &h->block_offset[0];
+    const int transform_bypass = !SIMPLE && (s->qscale == 0 && h->sps.transform_bypass);
+    const int plane_count      = (SIMPLE || !CONFIG_GRAY || !(s->flags & CODEC_FLAG_GRAY)) ? 3 : 1;
+
+    for (p = 0; p < plane_count; p++) {
+        dest[p] = s->current_picture.f.data[p] +
+                  ((mb_x << PIXEL_SHIFT) + mb_y * s->linesize) * 16;
+        s->dsp.prefetch(dest[p] + (s->mb_x & 3) * 4 * s->linesize + (64 << PIXEL_SHIFT),
+                        s->linesize, 4);
+    }
+
+    h->list_counts[mb_xy] = h->list_count;
+
+    if (!SIMPLE && MB_FIELD) {
+        linesize     = h->mb_linesize = h->mb_uvlinesize = s->linesize * 2;
+        block_offset = &h->block_offset[48];
+        if (mb_y & 1) // FIXME move out of this function?
+            for (p = 0; p < 3; p++)
+                dest[p] -= s->linesize * 15;
+        if (FRAME_MBAFF) {
+            int list;
+            for (list = 0; list < h->list_count; list++) {
+                if (!USES_LIST(mb_type, list))
+                    continue;
+                if (IS_16X16(mb_type)) {
+                    int8_t *ref = &h->ref_cache[list][scan8[0]];
+                    fill_rectangle(ref, 4, 4, 8, (16 + *ref) ^ (s->mb_y & 1), 1);
+                } else {
+                    for (i = 0; i < 16; i += 4) {
+                        int ref = h->ref_cache[list][scan8[i]];
+                        if (ref >= 0)
+                            fill_rectangle(&h->ref_cache[list][scan8[i]], 2, 2,
+                                           8, (16 + ref) ^ (s->mb_y & 1), 1);
+                    }
+                }
+            }
+        }
+    } else {
+        linesize = h->mb_linesize = h->mb_uvlinesize = s->linesize;
+    }
+
+    if (!SIMPLE && IS_INTRA_PCM(mb_type)) {
+        if (PIXEL_SHIFT) {
+            const int bit_depth = h->sps.bit_depth_luma;
+            GetBitContext gb;
+            init_get_bits(&gb, (uint8_t *)h->mb, 768 * bit_depth);
+
+            for (p = 0; p < plane_count; p++)
+                for (i = 0; i < 16; i++) {
+                    uint16_t *tmp = (uint16_t *)(dest[p] + i * linesize);
+                    for (j = 0; j < 16; j++)
+                        tmp[j] = get_bits(&gb, bit_depth);
+                }
+        } else {
+            for (p = 0; p < plane_count; p++)
+                for (i = 0; i < 16; i++)
+                    memcpy(dest[p] + i * linesize,
+                           (uint8_t *)h->mb + p * 256 + i * 16, 16);
+        }
+    } else {
+        if (IS_INTRA(mb_type)) {
+            if (h->deblocking_filter)
+                xchg_mb_border(h, dest[0], dest[1], dest[2], linesize,
+                               linesize, 1, 1, SIMPLE, PIXEL_SHIFT);
+
+            for (p = 0; p < plane_count; p++)
+                hl_decode_mb_predict_luma(h, mb_type, 1, SIMPLE,
+                                          transform_bypass, PIXEL_SHIFT,
+                                          block_offset, linesize, dest[p], p);
+
+            if (h->deblocking_filter)
+                xchg_mb_border(h, dest[0], dest[1], dest[2], linesize,
+                               linesize, 0, 1, SIMPLE, PIXEL_SHIFT);
+        } else {
+            FUNC(hl_motion_444)(h, dest[0], dest[1], dest[2],
+                      s->me.qpel_put, s->dsp.put_h264_chroma_pixels_tab,
+                      s->me.qpel_avg, s->dsp.avg_h264_chroma_pixels_tab,
+                      h->h264dsp.weight_h264_pixels_tab,
+                      h->h264dsp.biweight_h264_pixels_tab);
+        }
+
+        for (p = 0; p < plane_count; p++)
+            hl_decode_mb_idct_luma(h, mb_type, 1, SIMPLE, transform_bypass,
+                                   PIXEL_SHIFT, block_offset, linesize,
+                                   dest[p], p);
+    }
+    if (h->cbp || IS_INTRA(mb_type)) {
+        s->dsp.clear_blocks(h->mb);
+        s->dsp.clear_blocks(h->mb + (24 * 16 << PIXEL_SHIFT));
+    }
+}
+
+#endif
diff --git a/libavcodec/h264_mc_template.c b/libavcodec/h264_mc_template.c
new file mode 100644 (file)
index 0000000..a3af39b
--- /dev/null
@@ -0,0 +1,160 @@
+/*
+ * H.26L/H.264/AVC/JVT/14496-10/... decoder
+ * Copyright (c) 2003 Michael Niedermayer <michaelni@gmx.at>
+ *
+ * This file is part of Libav.
+ *
+ * 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.
+ *
+ * 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
+ * License along with Libav; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#undef MCFUNC
+
+#if   CHROMA_IDC == 1
+#   define MCFUNC(n) FUNC(n ## _420)
+#elif CHROMA_IDC == 2
+#   define MCFUNC(n) FUNC(n ## _422)
+#elif CHROMA_IDC == 3
+#   define MCFUNC(n) FUNC(n ## _444)
+#endif
+
+#undef  mc_part
+#define mc_part MCFUNC(mc_part)
+
+static void mc_part(H264Context *h, int n, int square,
+                    int 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,
+                    h264_weight_func *weight_op,
+                    h264_biweight_func *weight_avg,
+                    int list0, int list1)
+{
+    if ((h->use_weight == 2 && list0 && list1 &&
+         (h->implicit_weight[h->ref_cache[0][scan8[n]]][h->ref_cache[1][scan8[n]]][h->s.mb_y & 1] != 32)) ||
+        h->use_weight == 1)
+        mc_part_weighted(h, n, square, height, delta, dest_y, dest_cb, dest_cr,
+                         x_offset, y_offset, qpix_put, chroma_put,
+                         weight_op[0], weight_op[1], weight_avg[0],
+                         weight_avg[1], list0, list1, PIXEL_SHIFT, CHROMA_IDC);
+    else
+        mc_part_std(h, n, square, height, delta, dest_y, dest_cb, dest_cr,
+                    x_offset, y_offset, qpix_put, chroma_put, qpix_avg,
+                    chroma_avg, list0, list1, PIXEL_SHIFT, CHROMA_IDC);
+}
+
+static void MCFUNC(hl_motion)(H264Context *h, uint8_t *dest_y,
+                              uint8_t *dest_cb, uint8_t *dest_cr,
+                              qpel_mc_func(*qpix_put)[16],
+                              h264_chroma_mc_func(*chroma_put),
+                              qpel_mc_func(*qpix_avg)[16],
+                              h264_chroma_mc_func(*chroma_avg),
+                              h264_weight_func *weight_op,
+                              h264_biweight_func *weight_avg)
+{
+    MpegEncContext *const s = &h->s;
+    const int mb_xy   = h->mb_xy;
+    const int mb_type = s->current_picture.f.mb_type[mb_xy];
+
+    assert(IS_INTER(mb_type));
+
+    if (HAVE_THREADS && (s->avctx->active_thread_type & FF_THREAD_FRAME))
+        await_references(h);
+    prefetch_motion(h, 0, PIXEL_SHIFT, CHROMA_IDC);
+
+    if (IS_16X16(mb_type)) {
+        mc_part(h, 0, 1, 16, 0, dest_y, dest_cb, dest_cr, 0, 0,
+                qpix_put[0], chroma_put[0], qpix_avg[0], chroma_avg[0],
+                weight_op, weight_avg,
+                IS_DIR(mb_type, 0, 0), IS_DIR(mb_type, 0, 1));
+    } else if (IS_16X8(mb_type)) {
+        mc_part(h, 0, 0, 8, 8 << PIXEL_SHIFT, dest_y, dest_cb, dest_cr, 0, 0,
+                qpix_put[1], chroma_put[0], qpix_avg[1], chroma_avg[0],
+                weight_op, weight_avg,
+                IS_DIR(mb_type, 0, 0), IS_DIR(mb_type, 0, 1));
+        mc_part(h, 8, 0, 8, 8 << PIXEL_SHIFT, dest_y, dest_cb, dest_cr, 0, 4,
+                qpix_put[1], chroma_put[0], qpix_avg[1], chroma_avg[0],
+                weight_op, weight_avg,
+                IS_DIR(mb_type, 1, 0), IS_DIR(mb_type, 1, 1));
+    } else if (IS_8X16(mb_type)) {
+        mc_part(h, 0, 0, 16, 8 * h->mb_linesize, dest_y, dest_cb, dest_cr, 0, 0,
+                qpix_put[1], chroma_put[1], qpix_avg[1], chroma_avg[1],
+                &weight_op[1], &weight_avg[1],
+                IS_DIR(mb_type, 0, 0), IS_DIR(mb_type, 0, 1));
+        mc_part(h, 4, 0, 16, 8 * h->mb_linesize, dest_y, dest_cb, dest_cr, 4, 0,
+                qpix_put[1], chroma_put[1], qpix_avg[1], chroma_avg[1],
+                &weight_op[1], &weight_avg[1],
+                IS_DIR(mb_type, 1, 0), IS_DIR(mb_type, 1, 1));
+    } else {
+        int i;
+
+        assert(IS_8X8(mb_type));
+
+        for (i = 0; i < 4; i++) {
+            const int sub_mb_type = h->sub_mb_type[i];
+            const int n  = 4 * i;
+            int x_offset = (i & 1) << 2;
+            int y_offset = (i & 2) << 1;
+
+            if (IS_SUB_8X8(sub_mb_type)) {
+                mc_part(h, n, 1, 8, 0, dest_y, dest_cb, dest_cr,
+                        x_offset, y_offset,
+                        qpix_put[1], chroma_put[1], qpix_avg[1], chroma_avg[1],
+                        &weight_op[1], &weight_avg[1],
+                        IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1));
+            } else if (IS_SUB_8X4(sub_mb_type)) {
+                mc_part(h, n, 0, 4, 4 << PIXEL_SHIFT, dest_y, dest_cb, dest_cr,
+                        x_offset, y_offset,
+                        qpix_put[2], chroma_put[1], qpix_avg[2], chroma_avg[1],
+                        &weight_op[1], &weight_avg[1],
+                        IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1));
+                mc_part(h, n + 2, 0, 4, 4 << PIXEL_SHIFT,
+                        dest_y, dest_cb, dest_cr, x_offset, y_offset + 2,
+                        qpix_put[2], chroma_put[1], qpix_avg[2], chroma_avg[1],
+                        &weight_op[1], &weight_avg[1],
+                        IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1));
+            } else if (IS_SUB_4X8(sub_mb_type)) {
+                mc_part(h, n, 0, 8, 4 * h->mb_linesize,
+                        dest_y, dest_cb, dest_cr, x_offset, y_offset,
+                        qpix_put[2], chroma_put[2], qpix_avg[2], chroma_avg[2],
+                        &weight_op[2], &weight_avg[2],
+                        IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1));
+                mc_part(h, n + 1, 0, 8, 4 * h->mb_linesize,
+                        dest_y, dest_cb, dest_cr, x_offset + 2, y_offset,
+                        qpix_put[2], chroma_put[2], qpix_avg[2], chroma_avg[2],
+                        &weight_op[2], &weight_avg[2],
+                        IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1));
+            } else {
+                int j;
+                assert(IS_SUB_4X4(sub_mb_type));
+                for (j = 0; j < 4; j++) {
+                    int sub_x_offset = x_offset + 2 * (j & 1);
+                    int sub_y_offset = y_offset + (j & 2);
+                    mc_part(h, n + j, 1, 4, 0,
+                            dest_y, dest_cb, dest_cr, sub_x_offset, sub_y_offset,
+                            qpix_put[2], chroma_put[2], qpix_avg[2], chroma_avg[2],
+                            &weight_op[2], &weight_avg[2],
+                            IS_DIR(sub_mb_type, 0, 0), IS_DIR(sub_mb_type, 0, 1));
+                }
+            }
+        }
+    }
+
+    prefetch_motion(h, 1, PIXEL_SHIFT, CHROMA_IDC);
+}
+