diff options
Diffstat (limited to 'contrib/ffmpeg/libavcodec/h264.c')
| -rw-r--r-- | contrib/ffmpeg/libavcodec/h264.c | 3174 |
1 files changed, 1275 insertions, 1899 deletions
diff --git a/contrib/ffmpeg/libavcodec/h264.c b/contrib/ffmpeg/libavcodec/h264.c index 4d72dc2ff..cd6facb9b 100644 --- a/contrib/ffmpeg/libavcodec/h264.c +++ b/contrib/ffmpeg/libavcodec/h264.c @@ -17,7 +17,6 @@ * You should have received a copy of the GNU Lesser General Public * License along with FFmpeg; if not, write to the Free Software * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA - * */ /** @@ -26,367 +25,25 @@ * @author Michael Niedermayer <michaelni@gmx.at> */ -#include "common.h" #include "dsputil.h" #include "avcodec.h" #include "mpegvideo.h" +#include "h264.h" #include "h264data.h" +#include "h264_parser.h" #include "golomb.h" +#include "rectangle.h" #include "cabac.h" //#undef NDEBUG #include <assert.h> -#define interlaced_dct interlaced_dct_is_a_bad_name -#define mb_intra mb_intra_isnt_initalized_see_mb_type - -#define LUMA_DC_BLOCK_INDEX 25 -#define CHROMA_DC_BLOCK_INDEX 26 - -#define CHROMA_DC_COEFF_TOKEN_VLC_BITS 8 -#define COEFF_TOKEN_VLC_BITS 8 -#define TOTAL_ZEROS_VLC_BITS 9 -#define CHROMA_DC_TOTAL_ZEROS_VLC_BITS 3 -#define RUN_VLC_BITS 3 -#define RUN7_VLC_BITS 6 - -#define MAX_SPS_COUNT 32 -#define MAX_PPS_COUNT 256 - -#define MAX_MMCO_COUNT 66 - -/* Compiling in interlaced support reduces the speed - * of progressive decoding by about 2%. */ -#define ALLOW_INTERLACE - -#ifdef ALLOW_INTERLACE -#define MB_MBAFF h->mb_mbaff -#define MB_FIELD h->mb_field_decoding_flag -#define FRAME_MBAFF h->mb_aff_frame -#else -#define MB_MBAFF 0 -#define MB_FIELD 0 -#define FRAME_MBAFF 0 -#undef IS_INTERLACED -#define IS_INTERLACED(mb_type) 0 -#endif - -/** - * Sequence parameter set - */ -typedef struct SPS{ - - int profile_idc; - int level_idc; - int transform_bypass; ///< qpprime_y_zero_transform_bypass_flag - int log2_max_frame_num; ///< log2_max_frame_num_minus4 + 4 - int poc_type; ///< pic_order_cnt_type - int log2_max_poc_lsb; ///< log2_max_pic_order_cnt_lsb_minus4 - int delta_pic_order_always_zero_flag; - int offset_for_non_ref_pic; - int offset_for_top_to_bottom_field; - int poc_cycle_length; ///< num_ref_frames_in_pic_order_cnt_cycle - int ref_frame_count; ///< num_ref_frames - int gaps_in_frame_num_allowed_flag; - int mb_width; ///< frame_width_in_mbs_minus1 + 1 - int mb_height; ///< frame_height_in_mbs_minus1 + 1 - int frame_mbs_only_flag; - int mb_aff; ///<mb_adaptive_frame_field_flag - int direct_8x8_inference_flag; - int crop; ///< frame_cropping_flag - int crop_left; ///< frame_cropping_rect_left_offset - int crop_right; ///< frame_cropping_rect_right_offset - int crop_top; ///< frame_cropping_rect_top_offset - int crop_bottom; ///< frame_cropping_rect_bottom_offset - int vui_parameters_present_flag; - AVRational sar; - int timing_info_present_flag; - uint32_t num_units_in_tick; - uint32_t time_scale; - int fixed_frame_rate_flag; - short offset_for_ref_frame[256]; //FIXME dyn aloc? - int bitstream_restriction_flag; - int num_reorder_frames; - int scaling_matrix_present; - uint8_t scaling_matrix4[6][16]; - uint8_t scaling_matrix8[2][64]; -}SPS; - -/** - * Picture parameter set - */ -typedef struct PPS{ - unsigned int sps_id; - int cabac; ///< entropy_coding_mode_flag - int pic_order_present; ///< pic_order_present_flag - int slice_group_count; ///< num_slice_groups_minus1 + 1 - int mb_slice_group_map_type; - unsigned int ref_count[2]; ///< num_ref_idx_l0/1_active_minus1 + 1 - int weighted_pred; ///< weighted_pred_flag - int weighted_bipred_idc; - int init_qp; ///< pic_init_qp_minus26 + 26 - int init_qs; ///< pic_init_qs_minus26 + 26 - int chroma_qp_index_offset; - int deblocking_filter_parameters_present; ///< deblocking_filter_parameters_present_flag - int constrained_intra_pred; ///< constrained_intra_pred_flag - int redundant_pic_cnt_present; ///< redundant_pic_cnt_present_flag - int transform_8x8_mode; ///< transform_8x8_mode_flag - uint8_t scaling_matrix4[6][16]; - uint8_t scaling_matrix8[2][64]; -}PPS; - -/** - * Memory management control operation opcode. - */ -typedef enum MMCOOpcode{ - MMCO_END=0, - MMCO_SHORT2UNUSED, - MMCO_LONG2UNUSED, - MMCO_SHORT2LONG, - MMCO_SET_MAX_LONG, - MMCO_RESET, - MMCO_LONG, -} MMCOOpcode; - -/** - * Memory management control operation. - */ -typedef struct MMCO{ - MMCOOpcode opcode; - int short_frame_num; - int long_index; -} MMCO; - /** - * H264Context + * Value of Picture.reference when Picture is not a reference picture, but + * is held for delayed output. */ -typedef struct H264Context{ - MpegEncContext s; - int nal_ref_idc; - int nal_unit_type; - uint8_t *rbsp_buffer; - unsigned int rbsp_buffer_size; - - /** - * Used to parse AVC variant of h264 - */ - int is_avc; ///< this flag is != 0 if codec is avc1 - int got_avcC; ///< flag used to parse avcC data only once - int nal_length_size; ///< Number of bytes used for nal length (1, 2 or 4) - - int chroma_qp; //QPc - - int prev_mb_skipped; - int next_mb_skipped; - - //prediction stuff - int chroma_pred_mode; - int intra16x16_pred_mode; - - int top_mb_xy; - int left_mb_xy[2]; - - int8_t intra4x4_pred_mode_cache[5*8]; - int8_t (*intra4x4_pred_mode)[8]; - void (*pred4x4 [9+3])(uint8_t *src, uint8_t *topright, int stride);//FIXME move to dsp? - void (*pred8x8l [9+3])(uint8_t *src, int topleft, int topright, int stride); - void (*pred8x8 [4+3])(uint8_t *src, int stride); - void (*pred16x16[4+3])(uint8_t *src, int stride); - unsigned int topleft_samples_available; - unsigned int top_samples_available; - unsigned int topright_samples_available; - unsigned int left_samples_available; - uint8_t (*top_borders[2])[16+2*8]; - uint8_t left_border[2*(17+2*9)]; - - /** - * non zero coeff count cache. - * is 64 if not available. - */ - DECLARE_ALIGNED_8(uint8_t, non_zero_count_cache[6*8]); - uint8_t (*non_zero_count)[16]; - - /** - * Motion vector cache. - */ - DECLARE_ALIGNED_8(int16_t, mv_cache[2][5*8][2]); - DECLARE_ALIGNED_8(int8_t, ref_cache[2][5*8]); -#define LIST_NOT_USED -1 //FIXME rename? -#define PART_NOT_AVAILABLE -2 - - /** - * is 1 if the specific list MV&references are set to 0,0,-2. - */ - int mv_cache_clean[2]; - - /** - * number of neighbors (top and/or left) that used 8x8 dct - */ - int neighbor_transform_size; - - /** - * block_offset[ 0..23] for frame macroblocks - * block_offset[24..47] for field macroblocks - */ - int block_offset[2*(16+8)]; - - uint32_t *mb2b_xy; //FIXME are these 4 a good idea? - uint32_t *mb2b8_xy; - int b_stride; //FIXME use s->b4_stride - int b8_stride; - - int mb_linesize; ///< may be equal to s->linesize or s->linesize*2, for mbaff - int mb_uvlinesize; - - int emu_edge_width; - int emu_edge_height; - - int halfpel_flag; - int thirdpel_flag; - - int unknown_svq3_flag; - int next_slice_index; - - SPS sps_buffer[MAX_SPS_COUNT]; - SPS sps; ///< current sps - - PPS pps_buffer[MAX_PPS_COUNT]; - /** - * current pps - */ - PPS pps; //FIXME move to Picture perhaps? (->no) do we need that? - - uint32_t dequant4_buffer[6][52][16]; - uint32_t dequant8_buffer[2][52][64]; - uint32_t (*dequant4_coeff[6])[16]; - uint32_t (*dequant8_coeff[2])[64]; - int dequant_coeff_pps; ///< reinit tables when pps changes - - int slice_num; - uint8_t *slice_table_base; - uint8_t *slice_table; ///< slice_table_base + 2*mb_stride + 1 - int slice_type; - int slice_type_fixed; - - //interlacing specific flags - int mb_aff_frame; - int mb_field_decoding_flag; - int mb_mbaff; ///< mb_aff_frame && mb_field_decoding_flag - - unsigned int sub_mb_type[4]; - - //POC stuff - int poc_lsb; - int poc_msb; - int delta_poc_bottom; - int delta_poc[2]; - int frame_num; - int prev_poc_msb; ///< poc_msb of the last reference pic for POC type 0 - int prev_poc_lsb; ///< poc_lsb of the last reference pic for POC type 0 - int frame_num_offset; ///< for POC type 2 - int prev_frame_num_offset; ///< for POC type 2 - int prev_frame_num; ///< frame_num of the last pic for POC type 1/2 - - /** - * frame_num for frames or 2*frame_num for field pics. - */ - int curr_pic_num; - - /** - * max_frame_num or 2*max_frame_num for field pics. - */ - int max_pic_num; - - //Weighted pred stuff - int use_weight; - int use_weight_chroma; - int luma_log2_weight_denom; - int chroma_log2_weight_denom; - int luma_weight[2][48]; - int luma_offset[2][48]; - int chroma_weight[2][48][2]; - int chroma_offset[2][48][2]; - int implicit_weight[48][48]; - - //deblock - int deblocking_filter; ///< disable_deblocking_filter_idc with 1<->0 - int slice_alpha_c0_offset; - int slice_beta_offset; - - int redundant_pic_count; - - int direct_spatial_mv_pred; - int dist_scale_factor[16]; - int dist_scale_factor_field[32]; - int map_col_to_list0[2][16]; - int map_col_to_list0_field[2][32]; - - /** - * num_ref_idx_l0/1_active_minus1 + 1 - */ - unsigned int ref_count[2]; ///< counts frames or fields, depending on current mb mode - unsigned int list_count; - Picture *short_ref[32]; - Picture *long_ref[32]; - Picture default_ref_list[2][32]; - Picture ref_list[2][48]; ///< 0..15: frame refs, 16..47: mbaff field refs - Picture *delayed_pic[18]; //FIXME size? - Picture *delayed_output_pic; - - /** - * memory management control operations buffer. - */ - MMCO mmco[MAX_MMCO_COUNT]; - int mmco_index; - - int long_ref_count; ///< number of actual long term references - int short_ref_count; ///< number of actual short term references - - //data partitioning - GetBitContext intra_gb; - GetBitContext inter_gb; - GetBitContext *intra_gb_ptr; - GetBitContext *inter_gb_ptr; - - DECLARE_ALIGNED_8(DCTELEM, mb[16*24]); - DCTELEM mb_padding[256]; ///< as mb is addressed by scantable[i] and scantable is uint8_t we can either check that i is not to large or ensure that there is some unused stuff after mb - - /** - * Cabac - */ - CABACContext cabac; - uint8_t cabac_state[460]; - int cabac_init_idc; - - /* 0x100 -> non null luma_dc, 0x80/0x40 -> non null chroma_dc (cb/cr), 0x?0 -> chroma_cbp(0,1,2), 0x0? luma_cbp */ - uint16_t *cbp_table; - int cbp; - int top_cbp; - int left_cbp; - /* chroma_pred_mode for i4x4 or i16x16, else 0 */ - uint8_t *chroma_pred_mode_table; - int last_qscale_diff; - int16_t (*mvd_table[2])[2]; - DECLARE_ALIGNED_8(int16_t, mvd_cache[2][5*8][2]); - uint8_t *direct_table; - uint8_t direct_cache[5*8]; - - uint8_t zigzag_scan[16]; - uint8_t zigzag_scan8x8[64]; - uint8_t zigzag_scan8x8_cavlc[64]; - uint8_t field_scan[16]; - uint8_t field_scan8x8[64]; - uint8_t field_scan8x8_cavlc[64]; - const uint8_t *zigzag_scan_q0; - const uint8_t *zigzag_scan8x8_q0; - const uint8_t *zigzag_scan8x8_cavlc_q0; - const uint8_t *field_scan_q0; - const uint8_t *field_scan8x8_q0; - const uint8_t *field_scan8x8_cavlc_q0; - - int x264_build; -}H264Context; +#define DELAYED_PIC_REF 4 static VLC coeff_token_vlc[4]; static VLC chroma_dc_coeff_token_vlc; @@ -419,109 +76,23 @@ const uint8_t ff_div6[52]={ }; -/** - * fill a rectangle. - * @param h height of the rectangle, should be a constant - * @param w width of the rectangle, should be a constant - * @param size the size of val (1 or 4), should be a constant - */ -static av_always_inline void fill_rectangle(void *vp, int w, int h, int stride, uint32_t val, int size){ - uint8_t *p= (uint8_t*)vp; - assert(size==1 || size==4); - assert(w<=4); - - w *= size; - stride *= size; - - assert((((long)vp)&(FFMIN(w, STRIDE_ALIGN)-1)) == 0); - assert((stride&(w-1))==0); - if(w==2){ - const uint16_t v= size==4 ? val : val*0x0101; - *(uint16_t*)(p + 0*stride)= v; - if(h==1) return; - *(uint16_t*)(p + 1*stride)= v; - if(h==2) return; - *(uint16_t*)(p + 2*stride)= - *(uint16_t*)(p + 3*stride)= v; - }else if(w==4){ - const uint32_t v= size==4 ? val : val*0x01010101; - *(uint32_t*)(p + 0*stride)= v; - if(h==1) return; - *(uint32_t*)(p + 1*stride)= v; - if(h==2) return; - *(uint32_t*)(p + 2*stride)= - *(uint32_t*)(p + 3*stride)= v; - }else if(w==8){ - //gcc can't optimize 64bit math on x86_32 -#if defined(ARCH_X86_64) || (defined(MP_WORDSIZE) && MP_WORDSIZE >= 64) - const uint64_t v= val*0x0100000001ULL; - *(uint64_t*)(p + 0*stride)= v; - if(h==1) return; - *(uint64_t*)(p + 1*stride)= v; - if(h==2) return; - *(uint64_t*)(p + 2*stride)= - *(uint64_t*)(p + 3*stride)= v; - }else if(w==16){ - const uint64_t v= val*0x0100000001ULL; - *(uint64_t*)(p + 0+0*stride)= - *(uint64_t*)(p + 8+0*stride)= - *(uint64_t*)(p + 0+1*stride)= - *(uint64_t*)(p + 8+1*stride)= v; - if(h==2) return; - *(uint64_t*)(p + 0+2*stride)= - *(uint64_t*)(p + 8+2*stride)= - *(uint64_t*)(p + 0+3*stride)= - *(uint64_t*)(p + 8+3*stride)= v; -#else - *(uint32_t*)(p + 0+0*stride)= - *(uint32_t*)(p + 4+0*stride)= val; - if(h==1) return; - *(uint32_t*)(p + 0+1*stride)= - *(uint32_t*)(p + 4+1*stride)= val; - if(h==2) return; - *(uint32_t*)(p + 0+2*stride)= - *(uint32_t*)(p + 4+2*stride)= - *(uint32_t*)(p + 0+3*stride)= - *(uint32_t*)(p + 4+3*stride)= val; - }else if(w==16){ - *(uint32_t*)(p + 0+0*stride)= - *(uint32_t*)(p + 4+0*stride)= - *(uint32_t*)(p + 8+0*stride)= - *(uint32_t*)(p +12+0*stride)= - *(uint32_t*)(p + 0+1*stride)= - *(uint32_t*)(p + 4+1*stride)= - *(uint32_t*)(p + 8+1*stride)= - *(uint32_t*)(p +12+1*stride)= val; - if(h==2) return; - *(uint32_t*)(p + 0+2*stride)= - *(uint32_t*)(p + 4+2*stride)= - *(uint32_t*)(p + 8+2*stride)= - *(uint32_t*)(p +12+2*stride)= - *(uint32_t*)(p + 0+3*stride)= - *(uint32_t*)(p + 4+3*stride)= - *(uint32_t*)(p + 8+3*stride)= - *(uint32_t*)(p +12+3*stride)= val; -#endif - }else - assert(0); - assert(h==4); -} - static void fill_caches(H264Context *h, int mb_type, int for_deblock){ MpegEncContext * const s = &h->s; const int mb_xy= s->mb_x + s->mb_y*s->mb_stride; int topleft_xy, top_xy, topright_xy, left_xy[2]; int topleft_type, top_type, topright_type, left_type[2]; int left_block[8]; + int topleft_partition= -1; int i; + top_xy = mb_xy - (s->mb_stride << FIELD_PICTURE); + //FIXME deblocking could skip the intra and nnz parts. - if(for_deblock && (h->slice_num == 1 || h->slice_table[mb_xy] == h->slice_table[mb_xy-s->mb_stride]) && !FRAME_MBAFF) + if(for_deblock && (h->slice_num == 1 || h->slice_table[mb_xy] == h->slice_table[top_xy]) && !FRAME_MBAFF) return; //wow what a mess, why didn't they simplify the interlacing&intra stuff, i can't imagine that these complex rules are worth it - top_xy = mb_xy - s->mb_stride; topleft_xy = top_xy - 1; topright_xy= top_xy + 1; left_xy[1] = left_xy[0] = mb_xy-1; @@ -556,6 +127,10 @@ static void fill_caches(H264Context *h, int mb_type, int for_deblock){ : (!curr_mb_frame_flag && !topleft_mb_frame_flag) // top macroblock ) { topleft_xy -= s->mb_stride; + } else if(bottom && curr_mb_frame_flag && !left_mb_frame_flag) { + topleft_xy += s->mb_stride; + // take topleft mv from the middle of the mb, as opposed to all other modes which use the bottom-right partition + topleft_partition = 0; } if (bottom ? !curr_mb_frame_flag // bottom macroblock @@ -833,8 +408,8 @@ static void fill_caches(H264Context *h, int mb_type, int for_deblock){ continue; if(USES_LIST(topleft_type, list)){ - const int b_xy = h->mb2b_xy[topleft_xy] + 3 + 3*h->b_stride; - const int b8_xy= h->mb2b8_xy[topleft_xy] + 1 + h->b8_stride; + const int b_xy = h->mb2b_xy[topleft_xy] + 3 + h->b_stride + (topleft_partition & 2*h->b_stride); + const int b8_xy= h->mb2b8_xy[topleft_xy] + 1 + (topleft_partition & h->b8_stride); *(uint32_t*)h->mv_cache[list][scan8[0] - 1 - 1*8]= *(uint32_t*)s->current_picture.motion_val[list][b_xy]; h->ref_cache[list][scan8[0] - 1 - 1*8]= s->current_picture.ref_index[list][b8_xy]; }else{ @@ -1131,7 +706,7 @@ static inline int fetch_diagonal_mv(H264Context *h, const int16_t **C, int i, in #define SET_DIAG_MV(MV_OP, REF_OP, X4, Y4)\ const int x4 = X4, y4 = Y4;\ const int mb_type = mb_types[(x4>>2)+(y4>>2)*s->mb_stride];\ - if(!USES_LIST(mb_type,list) && !IS_8X8(mb_type))\ + if(!USES_LIST(mb_type,list))\ return LIST_NOT_USED;\ mv = s->current_picture_ptr->motion_val[list][x4 + y4*h->b_stride];\ h->mv_cache[list][scan8[0]-2][0] = mv[0];\ @@ -1152,7 +727,7 @@ static inline int fetch_diagonal_mv(H264Context *h, const int16_t **C, int i, in && !IS_INTERLACED(mb_types[h->left_mb_xy[0]]) && i >= scan8[0]+8){ // leftshift will turn LIST_NOT_USED into PART_NOT_AVAILABLE, but that's ok. - SET_DIAG_MV(>>1, <<1, s->mb_x*4-1, (s->mb_y&~1)*4 - 1 + ((i-scan8[0])>>3)*2); + SET_DIAG_MV(/2, <<1, s->mb_x*4-1, (s->mb_y&~1)*4 - 1 + ((i-scan8[0])>>3)*2); } } #undef SET_DIAG_MV @@ -1447,14 +1022,76 @@ static inline void pred_direct_motion(H264Context * const h, int *mb_type){ } if(ref[1] < 0){ - *mb_type &= ~MB_TYPE_P0L1; - sub_mb_type &= ~MB_TYPE_P0L1; + if(!is_b8x8) + *mb_type &= ~MB_TYPE_L1; + sub_mb_type &= ~MB_TYPE_L1; }else if(ref[0] < 0){ - *mb_type &= ~MB_TYPE_P0L0; - sub_mb_type &= ~MB_TYPE_P0L0; + if(!is_b8x8) + *mb_type &= ~MB_TYPE_L0; + sub_mb_type &= ~MB_TYPE_L0; } - if(IS_16X16(*mb_type)){ + if(IS_INTERLACED(*mb_type) != IS_INTERLACED(mb_type_col)){ + int pair_xy = s->mb_x + (s->mb_y&~1)*s->mb_stride; + int mb_types_col[2]; + int b8_stride = h->b8_stride; + int b4_stride = h->b_stride; + + *mb_type = (*mb_type & ~MB_TYPE_16x16) | MB_TYPE_8x8; + + if(IS_INTERLACED(*mb_type)){ + mb_types_col[0] = h->ref_list[1][0].mb_type[pair_xy]; + mb_types_col[1] = h->ref_list[1][0].mb_type[pair_xy+s->mb_stride]; + if(s->mb_y&1){ + l1ref0 -= 2*b8_stride; + l1ref1 -= 2*b8_stride; + l1mv0 -= 4*b4_stride; + l1mv1 -= 4*b4_stride; + } + b8_stride *= 3; + b4_stride *= 6; + }else{ + int cur_poc = s->current_picture_ptr->poc; + int *col_poc = h->ref_list[1]->field_poc; + int col_parity = FFABS(col_poc[0] - cur_poc) >= FFABS(col_poc[1] - cur_poc); + int dy = 2*col_parity - (s->mb_y&1); + mb_types_col[0] = + mb_types_col[1] = h->ref_list[1][0].mb_type[pair_xy + col_parity*s->mb_stride]; + l1ref0 += dy*b8_stride; + l1ref1 += dy*b8_stride; + l1mv0 += 2*dy*b4_stride; + l1mv1 += 2*dy*b4_stride; + b8_stride = 0; + } + + for(i8=0; i8<4; i8++){ + int x8 = i8&1; + int y8 = i8>>1; + int xy8 = x8+y8*b8_stride; + int xy4 = 3*x8+y8*b4_stride; + int a=0, b=0; + + if(is_b8x8 && !IS_DIRECT(h->sub_mb_type[i8])) + continue; + h->sub_mb_type[i8] = sub_mb_type; + + fill_rectangle(&h->ref_cache[0][scan8[i8*4]], 2, 2, 8, (uint8_t)ref[0], 1); + fill_rectangle(&h->ref_cache[1][scan8[i8*4]], 2, 2, 8, (uint8_t)ref[1], 1); + if(!IS_INTRA(mb_types_col[y8]) + && ( (l1ref0[xy8] == 0 && FFABS(l1mv0[xy4][0]) <= 1 && FFABS(l1mv0[xy4][1]) <= 1) + || (l1ref0[xy8] < 0 && l1ref1[xy8] == 0 && FFABS(l1mv1[xy4][0]) <= 1 && FFABS(l1mv1[xy4][1]) <= 1))){ + if(ref[0] > 0) + a= pack16to32(mv[0][0],mv[0][1]); + if(ref[1] > 0) + b= pack16to32(mv[1][0],mv[1][1]); + }else{ + a= pack16to32(mv[0][0],mv[0][1]); + b= pack16to32(mv[1][0],mv[1][1]); + } + fill_rectangle(&h->mv_cache[0][scan8[i8*4]], 2, 2, 8, a, 4); + fill_rectangle(&h->mv_cache[1][scan8[i8*4]], 2, 2, 8, b, 4); + } + }else if(IS_16X16(*mb_type)){ int a=0, b=0; fill_rectangle(&h->ref_cache[0][scan8[0]], 4, 4, 8, (uint8_t)ref[0], 1); @@ -1738,9 +1375,10 @@ static inline void write_back_motion(H264Context *h, int mb_type){ * @param dst_length is the number of decoded bytes FIXME here or a decode rbsp tailing? * @returns decoded bytes, might be src+1 if no escapes */ -static uint8_t *decode_nal(H264Context *h, uint8_t *src, int *dst_length, int *consumed, int length){ +static const uint8_t *decode_nal(H264Context *h, const uint8_t *src, int *dst_length, int *consumed, int length){ int i, si, di; uint8_t *dst; + int bufidx; // src[0]&0x80; //forbidden bit h->nal_ref_idc= src[0]>>5; @@ -1769,8 +1407,9 @@ static uint8_t *decode_nal(H264Context *h, uint8_t *src, int *dst_length, int *c return src; } - h->rbsp_buffer= av_fast_realloc(h->rbsp_buffer, &h->rbsp_buffer_size, length); - dst= h->rbsp_buffer; + bufidx = h->nal_unit_type == NAL_DPC ? 1 : 0; // use second escape buffer for inter data + h->rbsp_buffer[bufidx]= av_fast_realloc(h->rbsp_buffer[bufidx], &h->rbsp_buffer_size[bufidx], length); + dst= h->rbsp_buffer[bufidx]; if (dst == NULL){ return NULL; @@ -1795,7 +1434,7 @@ static uint8_t *decode_nal(H264Context *h, uint8_t *src, int *dst_length, int *c *dst_length= di; *consumed= si + 1;//+1 for the header -//FIXME store exact number of bits in the getbitcontext (its needed for decoding) +//FIXME store exact number of bits in the getbitcontext (it is needed for decoding) return dst; } @@ -1803,7 +1442,7 @@ static uint8_t *decode_nal(H264Context *h, uint8_t *src, int *dst_length, int *c * identifies the exact end of the bitstream * @return the length of the trailing, or 0 if damaged */ -static int decode_rbsp_trailing(H264Context *h, uint8_t *src){ +static int decode_rbsp_trailing(H264Context *h, const uint8_t *src){ int v= *src; int r; @@ -1946,12 +1585,11 @@ static void chroma_dc_dct_c(DCTELEM *block){ /** * gets the chroma qp. */ -static inline int get_chroma_qp(int chroma_qp_index_offset, int qscale){ - - return chroma_qp[av_clip(qscale + chroma_qp_index_offset, 0, 51)]; +static inline int get_chroma_qp(H264Context *h, int t, int qscale){ + return h->pps.chroma_qp_table[t][qscale & 0xff]; } -//FIXME need to check that this doesnt overflow signed 32 bit for low qp, i am not sure, it's very close +//FIXME need to check that this does not overflow signed 32 bit for low qp, i am not sure, it's very close //FIXME check that gcc inlines this (and optimizes intra & separate_dc stuff away) static inline int quantize_c(DCTELEM *block, uint8_t *scantable, int qscale, int intra, int separate_dc){ int i; @@ -2030,722 +1668,6 @@ static inline int quantize_c(DCTELEM *block, uint8_t *scantable, int qscale, int return last_non_zero; } -static void pred4x4_vertical_c(uint8_t *src, uint8_t *topright, int stride){ - const uint32_t a= ((uint32_t*)(src-stride))[0]; - ((uint32_t*)(src+0*stride))[0]= a; - ((uint32_t*)(src+1*stride))[0]= a; - ((uint32_t*)(src+2*stride))[0]= a; - ((uint32_t*)(src+3*stride))[0]= a; -} - -static void pred4x4_horizontal_c(uint8_t *src, uint8_t *topright, int stride){ - ((uint32_t*)(src+0*stride))[0]= src[-1+0*stride]*0x01010101; - ((uint32_t*)(src+1*stride))[0]= src[-1+1*stride]*0x01010101; - ((uint32_t*)(src+2*stride))[0]= src[-1+2*stride]*0x01010101; - ((uint32_t*)(src+3*stride))[0]= src[-1+3*stride]*0x01010101; -} - -static void pred4x4_dc_c(uint8_t *src, uint8_t *topright, int stride){ - const int dc= ( src[-stride] + src[1-stride] + src[2-stride] + src[3-stride] - + src[-1+0*stride] + src[-1+1*stride] + src[-1+2*stride] + src[-1+3*stride] + 4) >>3; - - ((uint32_t*)(src+0*stride))[0]= - ((uint32_t*)(src+1*stride))[0]= - ((uint32_t*)(src+2*stride))[0]= - ((uint32_t*)(src+3*stride))[0]= dc* 0x01010101; -} - -static void pred4x4_left_dc_c(uint8_t *src, uint8_t *topright, int stride){ - const int dc= ( src[-1+0*stride] + src[-1+1*stride] + src[-1+2*stride] + src[-1+3*stride] + 2) >>2; - - ((uint32_t*)(src+0*stride))[0]= - ((uint32_t*)(src+1*stride))[0]= - ((uint32_t*)(src+2*stride))[0]= - ((uint32_t*)(src+3*stride))[0]= dc* 0x01010101; -} - -static void pred4x4_top_dc_c(uint8_t *src, uint8_t *topright, int stride){ - const int dc= ( src[-stride] + src[1-stride] + src[2-stride] + src[3-stride] + 2) >>2; - - ((uint32_t*)(src+0*stride))[0]= - ((uint32_t*)(src+1*stride))[0]= - ((uint32_t*)(src+2*stride))[0]= - ((uint32_t*)(src+3*stride))[0]= dc* 0x01010101; -} - -static void pred4x4_128_dc_c(uint8_t *src, uint8_t *topright, int stride){ - ((uint32_t*)(src+0*stride))[0]= - ((uint32_t*)(src+1*stride))[0]= - ((uint32_t*)(src+2*stride))[0]= - ((uint32_t*)(src+3*stride))[0]= 128U*0x01010101U; -} - - -#define LOAD_TOP_RIGHT_EDGE\ - const int t4= topright[0];\ - const int t5= topright[1];\ - const int t6= topright[2];\ - const int t7= topright[3];\ - -#define LOAD_LEFT_EDGE\ - const int l0= src[-1+0*stride];\ - const int l1= src[-1+1*stride];\ - const int l2= src[-1+2*stride];\ - const int l3= src[-1+3*stride];\ - -#define LOAD_TOP_EDGE\ - const int t0= src[ 0-1*stride];\ - const int t1= src[ 1-1*stride];\ - const int t2= src[ 2-1*stride];\ - const int t3= src[ 3-1*stride];\ - -static void pred4x4_down_right_c(uint8_t *src, uint8_t *topright, int stride){ - const int lt= src[-1-1*stride]; - LOAD_TOP_EDGE - LOAD_LEFT_EDGE - - src[0+3*stride]=(l3 + 2*l2 + l1 + 2)>>2; - src[0+2*stride]= - src[1+3*stride]=(l2 + 2*l1 + l0 + 2)>>2; - src[0+1*stride]= - src[1+2*stride]= - src[2+3*stride]=(l1 + 2*l0 + lt + 2)>>2; - src[0+0*stride]= - src[1+1*stride]= - src[2+2*stride]= - src[3+3*stride]=(l0 + 2*lt + t0 + 2)>>2; - src[1+0*stride]= - src[2+1*stride]= - src[3+2*stride]=(lt + 2*t0 + t1 + 2)>>2; - src[2+0*stride]= - src[3+1*stride]=(t0 + 2*t1 + t2 + 2)>>2; - src[3+0*stride]=(t1 + 2*t2 + t3 + 2)>>2; -} - -static void pred4x4_down_left_c(uint8_t *src, uint8_t *topright, int stride){ - LOAD_TOP_EDGE - LOAD_TOP_RIGHT_EDGE -// LOAD_LEFT_EDGE - - src[0+0*stride]=(t0 + t2 + 2*t1 + 2)>>2; - src[1+0*stride]= - src[0+1*stride]=(t1 + t3 + 2*t2 + 2)>>2; - src[2+0*stride]= - src[1+1*stride]= - src[0+2*stride]=(t2 + t4 + 2*t3 + 2)>>2; - src[3+0*stride]= - src[2+1*stride]= - src[1+2*stride]= - src[0+3*stride]=(t3 + t5 + 2*t4 + 2)>>2; - src[3+1*stride]= - src[2+2*stride]= - src[1+3*stride]=(t4 + t6 + 2*t5 + 2)>>2; - src[3+2*stride]= - src[2+3*stride]=(t5 + t7 + 2*t6 + 2)>>2; - src[3+3*stride]=(t6 + 3*t7 + 2)>>2; -} - -static void pred4x4_vertical_right_c(uint8_t *src, uint8_t *topright, int stride){ - const int lt= src[-1-1*stride]; - LOAD_TOP_EDGE - LOAD_LEFT_EDGE - const __attribute__((unused)) int unu= l3; - - src[0+0*stride]= - src[1+2*stride]=(lt + t0 + 1)>>1; - src[1+0*stride]= - src[2+2*stride]=(t0 + t1 + 1)>>1; - src[2+0*stride]= - src[3+2*stride]=(t1 + t2 + 1)>>1; - src[3+0*stride]=(t2 + t3 + 1)>>1; - src[0+1*stride]= - src[1+3*stride]=(l0 + 2*lt + t0 + 2)>>2; - src[1+1*stride]= - src[2+3*stride]=(lt + 2*t0 + t1 + 2)>>2; - src[2+1*stride]= - src[3+3*stride]=(t0 + 2*t1 + t2 + 2)>>2; - src[3+1*stride]=(t1 + 2*t2 + t3 + 2)>>2; - src[0+2*stride]=(lt + 2*l0 + l1 + 2)>>2; - src[0+3*stride]=(l0 + 2*l1 + l2 + 2)>>2; -} - -static void pred4x4_vertical_left_c(uint8_t *src, uint8_t *topright, int stride){ - LOAD_TOP_EDGE - LOAD_TOP_RIGHT_EDGE - const __attribute__((unused)) int unu= t7; - - src[0+0*stride]=(t0 + t1 + 1)>>1; - src[1+0*stride]= - src[0+2*stride]=(t1 + t2 + 1)>>1; - src[2+0*stride]= - src[1+2*stride]=(t2 + t3 + 1)>>1; - src[3+0*stride]= - src[2+2*stride]=(t3 + t4+ 1)>>1; - src[3+2*stride]=(t4 + t5+ 1)>>1; - src[0+1*stride]=(t0 + 2*t1 + t2 + 2)>>2; - src[1+1*stride]= - src[0+3*stride]=(t1 + 2*t2 + t3 + 2)>>2; - src[2+1*stride]= - src[1+3*stride]=(t2 + 2*t3 + t4 + 2)>>2; - src[3+1*stride]= - src[2+3*stride]=(t3 + 2*t4 + t5 + 2)>>2; - src[3+3*stride]=(t4 + 2*t5 + t6 + 2)>>2; -} - -static void pred4x4_horizontal_up_c(uint8_t *src, uint8_t *topright, int stride){ - LOAD_LEFT_EDGE - - src[0+0*stride]=(l0 + l1 + 1)>>1; - src[1+0*stride]=(l0 + 2*l1 + l2 + 2)>>2; - src[2+0*stride]= - src[0+1*stride]=(l1 + l2 + 1)>>1; - src[3+0*stride]= - src[1+1*stride]=(l1 + 2*l2 + l3 + 2)>>2; - src[2+1*stride]= - src[0+2*stride]=(l2 + l3 + 1)>>1; - src[3+1*stride]= - src[1+2*stride]=(l2 + 2*l3 + l3 + 2)>>2; - src[3+2*stride]= - src[1+3*stride]= - src[0+3*stride]= - src[2+2*stride]= - src[2+3*stride]= - src[3+3*stride]=l3; -} - -static void pred4x4_horizontal_down_c(uint8_t *src, uint8_t *topright, int stride){ - const int lt= src[-1-1*stride]; - LOAD_TOP_EDGE - LOAD_LEFT_EDGE - const __attribute__((unused)) int unu= t3; - - src[0+0*stride]= - src[2+1*stride]=(lt + l0 + 1)>>1; - src[1+0*stride]= - src[3+1*stride]=(l0 + 2*lt + t0 + 2)>>2; - src[2+0*stride]=(lt + 2*t0 + t1 + 2)>>2; - src[3+0*stride]=(t0 + 2*t1 + t2 + 2)>>2; - src[0+1*stride]= - src[2+2*stride]=(l0 + l1 + 1)>>1; - src[1+1*stride]= - src[3+2*stride]=(lt + 2*l0 + l1 + 2)>>2; - src[0+2*stride]= - src[2+3*stride]=(l1 + l2+ 1)>>1; - src[1+2*stride]= - src[3+3*stride]=(l0 + 2*l1 + l2 + 2)>>2; - src[0+3*stride]=(l2 + l3 + 1)>>1; - src[1+3*stride]=(l1 + 2*l2 + l3 + 2)>>2; -} - -void ff_pred16x16_vertical_c(uint8_t *src, int stride){ - int i; - const uint32_t a= ((uint32_t*)(src-stride))[0]; - const uint32_t b= ((uint32_t*)(src-stride))[1]; - const uint32_t c= ((uint32_t*)(src-stride))[2]; - const uint32_t d= ((uint32_t*)(src-stride))[3]; - - for(i=0; i<16; i++){ - ((uint32_t*)(src+i*stride))[0]= a; - ((uint32_t*)(src+i*stride))[1]= b; - ((uint32_t*)(src+i*stride))[2]= c; - ((uint32_t*)(src+i*stride))[3]= d; - } -} - -void ff_pred16x16_horizontal_c(uint8_t *src, int stride){ - int i; - - for(i=0; i<16; i++){ - ((uint32_t*)(src+i*stride))[0]= - ((uint32_t*)(src+i*stride))[1]= - ((uint32_t*)(src+i*stride))[2]= - ((uint32_t*)(src+i*stride))[3]= src[-1+i*stride]*0x01010101; - } -} - -void ff_pred16x16_dc_c(uint8_t *src, int stride){ - int i, dc=0; - - for(i=0;i<16; i++){ - dc+= src[-1+i*stride]; - } - - for(i=0;i<16; i++){ - dc+= src[i-stride]; - } - - dc= 0x01010101*((dc + 16)>>5); - - for(i=0; i<16; i++){ - ((uint32_t*)(src+i*stride))[0]= - ((uint32_t*)(src+i*stride))[1]= - ((uint32_t*)(src+i*stride))[2]= - ((uint32_t*)(src+i*stride))[3]= dc; - } -} - -static void pred16x16_left_dc_c(uint8_t *src, int stride){ - int i, dc=0; - - for(i=0;i<16; i++){ - dc+= src[-1+i*stride]; - } - - dc= 0x01010101*((dc + 8)>>4); - - for(i=0; i<16; i++){ - ((uint32_t*)(src+i*stride))[0]= - ((uint32_t*)(src+i*stride))[1]= - ((uint32_t*)(src+i*stride))[2]= - ((uint32_t*)(src+i*stride))[3]= dc; - } -} - -static void pred16x16_top_dc_c(uint8_t *src, int stride){ - int i, dc=0; - - for(i=0;i<16; i++){ - dc+= src[i-stride]; - } - dc= 0x01010101*((dc + 8)>>4); - - for(i=0; i<16; i++){ - ((uint32_t*)(src+i*stride))[0]= - ((uint32_t*)(src+i*stride))[1]= - ((uint32_t*)(src+i*stride))[2]= - ((uint32_t*)(src+i*stride))[3]= dc; - } -} - -void ff_pred16x16_128_dc_c(uint8_t *src, int stride){ - int i; - - for(i=0; i<16; i++){ - ((uint32_t*)(src+i*stride))[0]= - ((uint32_t*)(src+i*stride))[1]= - ((uint32_t*)(src+i*stride))[2]= - ((uint32_t*)(src+i*stride))[3]= 0x01010101U*128U; - } -} - -static inline void pred16x16_plane_compat_c(uint8_t *src, int stride, const int svq3){ - int i, j, k; - int a; - uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; - const uint8_t * const src0 = src+7-stride; - const uint8_t *src1 = src+8*stride-1; - const uint8_t *src2 = src1-2*stride; // == src+6*stride-1; - int H = src0[1] - src0[-1]; - int V = src1[0] - src2[ 0]; - for(k=2; k<=8; ++k) { - src1 += stride; src2 -= stride; - H += k*(src0[k] - src0[-k]); - V += k*(src1[0] - src2[ 0]); - } - if(svq3){ - H = ( 5*(H/4) ) / 16; - V = ( 5*(V/4) ) / 16; - - /* required for 100% accuracy */ - i = H; H = V; V = i; - }else{ - H = ( 5*H+32 ) >> 6; - V = ( 5*V+32 ) >> 6; - } - - a = 16*(src1[0] + src2[16] + 1) - 7*(V+H); - for(j=16; j>0; --j) { - int b = a; - a += V; - for(i=-16; i<0; i+=4) { - src[16+i] = cm[ (b ) >> 5 ]; - src[17+i] = cm[ (b+ H) >> 5 ]; - src[18+i] = cm[ (b+2*H) >> 5 ]; - src[19+i] = cm[ (b+3*H) >> 5 ]; - b += 4*H; - } - src += stride; - } -} - -void ff_pred16x16_plane_c(uint8_t *src, int stride){ - pred16x16_plane_compat_c(src, stride, 0); -} - -void ff_pred8x8_vertical_c(uint8_t *src, int stride){ - int i; - const uint32_t a= ((uint32_t*)(src-stride))[0]; - const uint32_t b= ((uint32_t*)(src-stride))[1]; - - for(i=0; i<8; i++){ - ((uint32_t*)(src+i*stride))[0]= a; - ((uint32_t*)(src+i*stride))[1]= b; - } -} - -void ff_pred8x8_horizontal_c(uint8_t *src, int stride){ - int i; - - for(i=0; i<8; i++){ - ((uint32_t*)(src+i*stride))[0]= - ((uint32_t*)(src+i*stride))[1]= src[-1+i*stride]*0x01010101; - } -} - -void ff_pred8x8_128_dc_c(uint8_t *src, int stride){ - int i; - - for(i=0; i<8; i++){ - ((uint32_t*)(src+i*stride))[0]= - ((uint32_t*)(src+i*stride))[1]= 0x01010101U*128U; - } -} - -static void pred8x8_left_dc_c(uint8_t *src, int stride){ - int i; - int dc0, dc2; - - dc0=dc2=0; - for(i=0;i<4; i++){ - dc0+= src[-1+i*stride]; - dc2+= src[-1+(i+4)*stride]; - } - dc0= 0x01010101*((dc0 + 2)>>2); - dc2= 0x01010101*((dc2 + 2)>>2); - - for(i=0; i<4; i++){ - ((uint32_t*)(src+i*stride))[0]= - ((uint32_t*)(src+i*stride))[1]= dc0; - } - for(i=4; i<8; i++){ - ((uint32_t*)(src+i*stride))[0]= - ((uint32_t*)(src+i*stride))[1]= dc2; - } -} - -static void pred8x8_top_dc_c(uint8_t *src, int stride){ - int i; - int dc0, dc1; - - dc0=dc1=0; - for(i=0;i<4; i++){ - dc0+= src[i-stride]; - dc1+= src[4+i-stride]; - } - dc0= 0x01010101*((dc0 + 2)>>2); - dc1= 0x01010101*((dc1 + 2)>>2); - - for(i=0; i<4; i++){ - ((uint32_t*)(src+i*stride))[0]= dc0; - ((uint32_t*)(src+i*stride))[1]= dc1; - } - for(i=4; i<8; i++){ - ((uint32_t*)(src+i*stride))[0]= dc0; - ((uint32_t*)(src+i*stride))[1]= dc1; - } -} - - -void ff_pred8x8_dc_c(uint8_t *src, int stride){ - int i; - int dc0, dc1, dc2, dc3; - - dc0=dc1=dc2=0; - for(i=0;i<4; i++){ - dc0+= src[-1+i*stride] + src[i-stride]; - dc1+= src[4+i-stride]; - dc2+= src[-1+(i+4)*stride]; - } - dc3= 0x01010101*((dc1 + dc2 + 4)>>3); - dc0= 0x01010101*((dc0 + 4)>>3); - dc1= 0x01010101*((dc1 + 2)>>2); - dc2= 0x01010101*((dc2 + 2)>>2); - - for(i=0; i<4; i++){ - ((uint32_t*)(src+i*stride))[0]= dc0; - ((uint32_t*)(src+i*stride))[1]= dc1; - } - for(i=4; i<8; i++){ - ((uint32_t*)(src+i*stride))[0]= dc2; - ((uint32_t*)(src+i*stride))[1]= dc3; - } -} - -void ff_pred8x8_plane_c(uint8_t *src, int stride){ - int j, k; - int a; - uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; - const uint8_t * const src0 = src+3-stride; - const uint8_t *src1 = src+4*stride-1; - const uint8_t *src2 = src1-2*stride; // == src+2*stride-1; - int H = src0[1] - src0[-1]; - int V = src1[0] - src2[ 0]; - for(k=2; k<=4; ++k) { - src1 += stride; src2 -= stride; - H += k*(src0[k] - src0[-k]); - V += k*(src1[0] - src2[ 0]); - } - H = ( 17*H+16 ) >> 5; - V = ( 17*V+16 ) >> 5; - - a = 16*(src1[0] + src2[8]+1) - 3*(V+H); - for(j=8; j>0; --j) { - int b = a; - a += V; - src[0] = cm[ (b ) >> 5 ]; - src[1] = cm[ (b+ H) >> 5 ]; - src[2] = cm[ (b+2*H) >> 5 ]; - src[3] = cm[ (b+3*H) >> 5 ]; - src[4] = cm[ (b+4*H) >> 5 ]; - src[5] = cm[ (b+5*H) >> 5 ]; - src[6] = cm[ (b+6*H) >> 5 ]; - src[7] = cm[ (b+7*H) >> 5 ]; - src += stride; - } -} - -#define SRC(x,y) src[(x)+(y)*stride] -#define PL(y) \ - const int l##y = (SRC(-1,y-1) + 2*SRC(-1,y) + SRC(-1,y+1) + 2) >> 2; -#define PREDICT_8x8_LOAD_LEFT \ - const int l0 = ((has_topleft ? SRC(-1,-1) : SRC(-1,0)) \ - + 2*SRC(-1,0) + SRC(-1,1) + 2) >> 2; \ - PL(1) PL(2) PL(3) PL(4) PL(5) PL(6) \ - const int l7 attribute_unused = (SRC(-1,6) + 3*SRC(-1,7) + 2) >> 2 - -#define PT(x) \ - const int t##x = (SRC(x-1,-1) + 2*SRC(x,-1) + SRC(x+1,-1) + 2) >> 2; -#define PREDICT_8x8_LOAD_TOP \ - const int t0 = ((has_topleft ? SRC(-1,-1) : SRC(0,-1)) \ - + 2*SRC(0,-1) + SRC(1,-1) + 2) >> 2; \ - PT(1) PT(2) PT(3) PT(4) PT(5) PT(6) \ - const int t7 attribute_unused = ((has_topright ? SRC(8,-1) : SRC(7,-1)) \ - + 2*SRC(7,-1) + SRC(6,-1) + 2) >> 2 - -#define PTR(x) \ - t##x = (SRC(x-1,-1) + 2*SRC(x,-1) + SRC(x+1,-1) + 2) >> 2; -#define PREDICT_8x8_LOAD_TOPRIGHT \ - int t8, t9, t10, t11, t12, t13, t14, t15; \ - if(has_topright) { \ - PTR(8) PTR(9) PTR(10) PTR(11) PTR(12) PTR(13) PTR(14) \ - t15 = (SRC(14,-1) + 3*SRC(15,-1) + 2) >> 2; \ - } else t8=t9=t10=t11=t12=t13=t14=t15= SRC(7,-1); - -#define PREDICT_8x8_LOAD_TOPLEFT \ - const int lt = (SRC(-1,0) + 2*SRC(-1,-1) + SRC(0,-1) + 2) >> 2 - -#define PREDICT_8x8_DC(v) \ - int y; \ - for( y = 0; y < 8; y++ ) { \ - ((uint32_t*)src)[0] = \ - ((uint32_t*)src)[1] = v; \ - src += stride; \ - } - -static void pred8x8l_128_dc_c(uint8_t *src, int has_topleft, int has_topright, int stride) -{ - PREDICT_8x8_DC(0x80808080); -} -static void pred8x8l_left_dc_c(uint8_t *src, int has_topleft, int has_topright, int stride) -{ - PREDICT_8x8_LOAD_LEFT; - const uint32_t dc = ((l0+l1+l2+l3+l4+l5+l6+l7+4) >> 3) * 0x01010101; - PREDICT_8x8_DC(dc); -} -static void pred8x8l_top_dc_c(uint8_t *src, int has_topleft, int has_topright, int stride) -{ - PREDICT_8x8_LOAD_TOP; - const uint32_t dc = ((t0+t1+t2+t3+t4+t5+t6+t7+4) >> 3) * 0x01010101; - PREDICT_8x8_DC(dc); -} -static void pred8x8l_dc_c(uint8_t *src, int has_topleft, int has_topright, int stride) -{ - PREDICT_8x8_LOAD_LEFT; - PREDICT_8x8_LOAD_TOP; - const uint32_t dc = ((l0+l1+l2+l3+l4+l5+l6+l7 - +t0+t1+t2+t3+t4+t5+t6+t7+8) >> 4) * 0x01010101; - PREDICT_8x8_DC(dc); -} -static void pred8x8l_horizontal_c(uint8_t *src, int has_topleft, int has_topright, int stride) -{ - PREDICT_8x8_LOAD_LEFT; -#define ROW(y) ((uint32_t*)(src+y*stride))[0] =\ - ((uint32_t*)(src+y*stride))[1] = 0x01010101 * l##y - ROW(0); ROW(1); ROW(2); ROW(3); ROW(4); ROW(5); ROW(6); ROW(7); -#undef ROW -} -static void pred8x8l_vertical_c(uint8_t *src, int has_topleft, int has_topright, int stride) -{ - int y; - PREDICT_8x8_LOAD_TOP; - src[0] = t0; - src[1] = t1; - src[2] = t2; - src[3] = t3; - src[4] = t4; - src[5] = t5; - src[6] = t6; - src[7] = t7; - for( y = 1; y < 8; y++ ) - *(uint64_t*)(src+y*stride) = *(uint64_t*)src; -} -static void pred8x8l_down_left_c(uint8_t *src, int has_topleft, int has_topright, int stride) -{ - PREDICT_8x8_LOAD_TOP; - PREDICT_8x8_LOAD_TOPRIGHT; - SRC(0,0)= (t0 + 2*t1 + t2 + 2) >> 2; - SRC(0,1)=SRC(1,0)= (t1 + 2*t2 + t3 + 2) >> 2; - SRC(0,2)=SRC(1,1)=SRC(2,0)= (t2 + 2*t3 + t4 + 2) >> 2; - SRC(0,3)=SRC(1,2)=SRC(2,1)=SRC(3,0)= (t3 + 2*t4 + t5 + 2) >> 2; - SRC(0,4)=SRC(1,3)=SRC(2,2)=SRC(3,1)=SRC(4,0)= (t4 + 2*t5 + t6 + 2) >> 2; - SRC(0,5)=SRC(1,4)=SRC(2,3)=SRC(3,2)=SRC(4,1)=SRC(5,0)= (t5 + 2*t6 + t7 + 2) >> 2; - SRC(0,6)=SRC(1,5)=SRC(2,4)=SRC(3,3)=SRC(4,2)=SRC(5,1)=SRC(6,0)= (t6 + 2*t7 + t8 + 2) >> 2; - SRC(0,7)=SRC(1,6)=SRC(2,5)=SRC(3,4)=SRC(4,3)=SRC(5,2)=SRC(6,1)=SRC(7,0)= (t7 + 2*t8 + t9 + 2) >> 2; - SRC(1,7)=SRC(2,6)=SRC(3,5)=SRC(4,4)=SRC(5,3)=SRC(6,2)=SRC(7,1)= (t8 + 2*t9 + t10 + 2) >> 2; - SRC(2,7)=SRC(3,6)=SRC(4,5)=SRC(5,4)=SRC(6,3)=SRC(7,2)= (t9 + 2*t10 + t11 + 2) >> 2; - SRC(3,7)=SRC(4,6)=SRC(5,5)=SRC(6,4)=SRC(7,3)= (t10 + 2*t11 + t12 + 2) >> 2; - SRC(4,7)=SRC(5,6)=SRC(6,5)=SRC(7,4)= (t11 + 2*t12 + t13 + 2) >> 2; - SRC(5,7)=SRC(6,6)=SRC(7,5)= (t12 + 2*t13 + t14 + 2) >> 2; - SRC(6,7)=SRC(7,6)= (t13 + 2*t14 + t15 + 2) >> 2; - SRC(7,7)= (t14 + 3*t15 + 2) >> 2; -} -static void pred8x8l_down_right_c(uint8_t *src, int has_topleft, int has_topright, int stride) -{ - PREDICT_8x8_LOAD_TOP; - PREDICT_8x8_LOAD_LEFT; - PREDICT_8x8_LOAD_TOPLEFT; - SRC(0,7)= (l7 + 2*l6 + l5 + 2) >> 2; - SRC(0,6)=SRC(1,7)= (l6 + 2*l5 + l4 + 2) >> 2; - SRC(0,5)=SRC(1,6)=SRC(2,7)= (l5 + 2*l4 + l3 + 2) >> 2; - SRC(0,4)=SRC(1,5)=SRC(2,6)=SRC(3,7)= (l4 + 2*l3 + l2 + 2) >> 2; - SRC(0,3)=SRC(1,4)=SRC(2,5)=SRC(3,6)=SRC(4,7)= (l3 + 2*l2 + l1 + 2) >> 2; - SRC(0,2)=SRC(1,3)=SRC(2,4)=SRC(3,5)=SRC(4,6)=SRC(5,7)= (l2 + 2*l1 + l0 + 2) >> 2; - SRC(0,1)=SRC(1,2)=SRC(2,3)=SRC(3,4)=SRC(4,5)=SRC(5,6)=SRC(6,7)= (l1 + 2*l0 + lt + 2) >> 2; - SRC(0,0)=SRC(1,1)=SRC(2,2)=SRC(3,3)=SRC(4,4)=SRC(5,5)=SRC(6,6)=SRC(7,7)= (l0 + 2*lt + t0 + 2) >> 2; - SRC(1,0)=SRC(2,1)=SRC(3,2)=SRC(4,3)=SRC(5,4)=SRC(6,5)=SRC(7,6)= (lt + 2*t0 + t1 + 2) >> 2; - SRC(2,0)=SRC(3,1)=SRC(4,2)=SRC(5,3)=SRC(6,4)=SRC(7,5)= (t0 + 2*t1 + t2 + 2) >> 2; - SRC(3,0)=SRC(4,1)=SRC(5,2)=SRC(6,3)=SRC(7,4)= (t1 + 2*t2 + t3 + 2) >> 2; - SRC(4,0)=SRC(5,1)=SRC(6,2)=SRC(7,3)= (t2 + 2*t3 + t4 + 2) >> 2; - SRC(5,0)=SRC(6,1)=SRC(7,2)= (t3 + 2*t4 + t5 + 2) >> 2; - SRC(6,0)=SRC(7,1)= (t4 + 2*t5 + t6 + 2) >> 2; - SRC(7,0)= (t5 + 2*t6 + t7 + 2) >> 2; - -} -static void pred8x8l_vertical_right_c(uint8_t *src, int has_topleft, int has_topright, int stride) -{ - PREDICT_8x8_LOAD_TOP; - PREDICT_8x8_LOAD_LEFT; - PREDICT_8x8_LOAD_TOPLEFT; - SRC(0,6)= (l5 + 2*l4 + l3 + 2) >> 2; - SRC(0,7)= (l6 + 2*l5 + l4 + 2) >> 2; - SRC(0,4)=SRC(1,6)= (l3 + 2*l2 + l1 + 2) >> 2; - SRC(0,5)=SRC(1,7)= (l4 + 2*l3 + l2 + 2) >> 2; - SRC(0,2)=SRC(1,4)=SRC(2,6)= (l1 + 2*l0 + lt + 2) >> 2; - SRC(0,3)=SRC(1,5)=SRC(2,7)= (l2 + 2*l1 + l0 + 2) >> 2; - SRC(0,1)=SRC(1,3)=SRC(2,5)=SRC(3,7)= (l0 + 2*lt + t0 + 2) >> 2; - SRC(0,0)=SRC(1,2)=SRC(2,4)=SRC(3,6)= (lt + t0 + 1) >> 1; - SRC(1,1)=SRC(2,3)=SRC(3,5)=SRC(4,7)= (lt + 2*t0 + t1 + 2) >> 2; - SRC(1,0)=SRC(2,2)=SRC(3,4)=SRC(4,6)= (t0 + t1 + 1) >> 1; - SRC(2,1)=SRC(3,3)=SRC(4,5)=SRC(5,7)= (t0 + 2*t1 + t2 + 2) >> 2; - SRC(2,0)=SRC(3,2)=SRC(4,4)=SRC(5,6)= (t1 + t2 + 1) >> 1; - SRC(3,1)=SRC(4,3)=SRC(5,5)=SRC(6,7)= (t1 + 2*t2 + t3 + 2) >> 2; - SRC(3,0)=SRC(4,2)=SRC(5,4)=SRC(6,6)= (t2 + t3 + 1) >> 1; - SRC(4,1)=SRC(5,3)=SRC(6,5)=SRC(7,7)= (t2 + 2*t3 + t4 + 2) >> 2; - SRC(4,0)=SRC(5,2)=SRC(6,4)=SRC(7,6)= (t3 + t4 + 1) >> 1; - SRC(5,1)=SRC(6,3)=SRC(7,5)= (t3 + 2*t4 + t5 + 2) >> 2; - SRC(5,0)=SRC(6,2)=SRC(7,4)= (t4 + t5 + 1) >> 1; - SRC(6,1)=SRC(7,3)= (t4 + 2*t5 + t6 + 2) >> 2; - SRC(6,0)=SRC(7,2)= (t5 + t6 + 1) >> 1; - SRC(7,1)= (t5 + 2*t6 + t7 + 2) >> 2; - SRC(7,0)= (t6 + t7 + 1) >> 1; -} -static void pred8x8l_horizontal_down_c(uint8_t *src, int has_topleft, int has_topright, int stride) -{ - PREDICT_8x8_LOAD_TOP; - PREDICT_8x8_LOAD_LEFT; - PREDICT_8x8_LOAD_TOPLEFT; - SRC(0,7)= (l6 + l7 + 1) >> 1; - SRC(1,7)= (l5 + 2*l6 + l7 + 2) >> 2; - SRC(0,6)=SRC(2,7)= (l5 + l6 + 1) >> 1; - SRC(1,6)=SRC(3,7)= (l4 + 2*l5 + l6 + 2) >> 2; - SRC(0,5)=SRC(2,6)=SRC(4,7)= (l4 + l5 + 1) >> 1; - SRC(1,5)=SRC(3,6)=SRC(5,7)= (l3 + 2*l4 + l5 + 2) >> 2; - SRC(0,4)=SRC(2,5)=SRC(4,6)=SRC(6,7)= (l3 + l4 + 1) >> 1; - SRC(1,4)=SRC(3,5)=SRC(5,6)=SRC(7,7)= (l2 + 2*l3 + l4 + 2) >> 2; - SRC(0,3)=SRC(2,4)=SRC(4,5)=SRC(6,6)= (l2 + l3 + 1) >> 1; - SRC(1,3)=SRC(3,4)=SRC(5,5)=SRC(7,6)= (l1 + 2*l2 + l3 + 2) >> 2; - SRC(0,2)=SRC(2,3)=SRC(4,4)=SRC(6,5)= (l1 + l2 + 1) >> 1; - SRC(1,2)=SRC(3,3)=SRC(5,4)=SRC(7,5)= (l0 + 2*l1 + l2 + 2) >> 2; - SRC(0,1)=SRC(2,2)=SRC(4,3)=SRC(6,4)= (l0 + l1 + 1) >> 1; - SRC(1,1)=SRC(3,2)=SRC(5,3)=SRC(7,4)= (lt + 2*l0 + l1 + 2) >> 2; - SRC(0,0)=SRC(2,1)=SRC(4,2)=SRC(6,3)= (lt + l0 + 1) >> 1; - SRC(1,0)=SRC(3,1)=SRC(5,2)=SRC(7,3)= (l0 + 2*lt + t0 + 2) >> 2; - SRC(2,0)=SRC(4,1)=SRC(6,2)= (t1 + 2*t0 + lt + 2) >> 2; - SRC(3,0)=SRC(5,1)=SRC(7,2)= (t2 + 2*t1 + t0 + 2) >> 2; - SRC(4,0)=SRC(6,1)= (t3 + 2*t2 + t1 + 2) >> 2; - SRC(5,0)=SRC(7,1)= (t4 + 2*t3 + t2 + 2) >> 2; - SRC(6,0)= (t5 + 2*t4 + t3 + 2) >> 2; - SRC(7,0)= (t6 + 2*t5 + t4 + 2) >> 2; -} -static void pred8x8l_vertical_left_c(uint8_t *src, int has_topleft, int has_topright, int stride) -{ - PREDICT_8x8_LOAD_TOP; - PREDICT_8x8_LOAD_TOPRIGHT; - SRC(0,0)= (t0 + t1 + 1) >> 1; - SRC(0,1)= (t0 + 2*t1 + t2 + 2) >> 2; - SRC(0,2)=SRC(1,0)= (t1 + t2 + 1) >> 1; - SRC(0,3)=SRC(1,1)= (t1 + 2*t2 + t3 + 2) >> 2; - SRC(0,4)=SRC(1,2)=SRC(2,0)= (t2 + t3 + 1) >> 1; - SRC(0,5)=SRC(1,3)=SRC(2,1)= (t2 + 2*t3 + t4 + 2) >> 2; - SRC(0,6)=SRC(1,4)=SRC(2,2)=SRC(3,0)= (t3 + t4 + 1) >> 1; - SRC(0,7)=SRC(1,5)=SRC(2,3)=SRC(3,1)= (t3 + 2*t4 + t5 + 2) >> 2; - SRC(1,6)=SRC(2,4)=SRC(3,2)=SRC(4,0)= (t4 + t5 + 1) >> 1; - SRC(1,7)=SRC(2,5)=SRC(3,3)=SRC(4,1)= (t4 + 2*t5 + t6 + 2) >> 2; - SRC(2,6)=SRC(3,4)=SRC(4,2)=SRC(5,0)= (t5 + t6 + 1) >> 1; - SRC(2,7)=SRC(3,5)=SRC(4,3)=SRC(5,1)= (t5 + 2*t6 + t7 + 2) >> 2; - SRC(3,6)=SRC(4,4)=SRC(5,2)=SRC(6,0)= (t6 + t7 + 1) >> 1; - SRC(3,7)=SRC(4,5)=SRC(5,3)=SRC(6,1)= (t6 + 2*t7 + t8 + 2) >> 2; - SRC(4,6)=SRC(5,4)=SRC(6,2)=SRC(7,0)= (t7 + t8 + 1) >> 1; - SRC(4,7)=SRC(5,5)=SRC(6,3)=SRC(7,1)= (t7 + 2*t8 + t9 + 2) >> 2; - SRC(5,6)=SRC(6,4)=SRC(7,2)= (t8 + t9 + 1) >> 1; - SRC(5,7)=SRC(6,5)=SRC(7,3)= (t8 + 2*t9 + t10 + 2) >> 2; - SRC(6,6)=SRC(7,4)= (t9 + t10 + 1) >> 1; - SRC(6,7)=SRC(7,5)= (t9 + 2*t10 + t11 + 2) >> 2; - SRC(7,6)= (t10 + t11 + 1) >> 1; - SRC(7,7)= (t10 + 2*t11 + t12 + 2) >> 2; -} -static void pred8x8l_horizontal_up_c(uint8_t *src, int has_topleft, int has_topright, int stride) -{ - PREDICT_8x8_LOAD_LEFT; - SRC(0,0)= (l0 + l1 + 1) >> 1; - SRC(1,0)= (l0 + 2*l1 + l2 + 2) >> 2; - SRC(0,1)=SRC(2,0)= (l1 + l2 + 1) >> 1; - SRC(1,1)=SRC(3,0)= (l1 + 2*l2 + l3 + 2) >> 2; - SRC(0,2)=SRC(2,1)=SRC(4,0)= (l2 + l3 + 1) >> 1; - SRC(1,2)=SRC(3,1)=SRC(5,0)= (l2 + 2*l3 + l4 + 2) >> 2; - SRC(0,3)=SRC(2,2)=SRC(4,1)=SRC(6,0)= (l3 + l4 + 1) >> 1; - SRC(1,3)=SRC(3,2)=SRC(5,1)=SRC(7,0)= (l3 + 2*l4 + l5 + 2) >> 2; - SRC(0,4)=SRC(2,3)=SRC(4,2)=SRC(6,1)= (l4 + l5 + 1) >> 1; - SRC(1,4)=SRC(3,3)=SRC(5,2)=SRC(7,1)= (l4 + 2*l5 + l6 + 2) >> 2; - SRC(0,5)=SRC(2,4)=SRC(4,3)=SRC(6,2)= (l5 + l6 + 1) >> 1; - SRC(1,5)=SRC(3,4)=SRC(5,3)=SRC(7,2)= (l5 + 2*l6 + l7 + 2) >> 2; - SRC(0,6)=SRC(2,5)=SRC(4,4)=SRC(6,3)= (l6 + l7 + 1) >> 1; - SRC(1,6)=SRC(3,5)=SRC(5,4)=SRC(7,3)= (l6 + 3*l7 + 2) >> 2; - SRC(0,7)=SRC(1,7)=SRC(2,6)=SRC(2,7)=SRC(3,6)= - SRC(3,7)=SRC(4,5)=SRC(4,6)=SRC(4,7)=SRC(5,5)= - SRC(5,6)=SRC(5,7)=SRC(6,4)=SRC(6,5)=SRC(6,6)= - SRC(6,7)=SRC(7,4)=SRC(7,5)=SRC(7,6)=SRC(7,7)= l7; -} -#undef PREDICT_8x8_LOAD_LEFT -#undef PREDICT_8x8_LOAD_TOP -#undef PREDICT_8x8_LOAD_TOPLEFT -#undef PREDICT_8x8_LOAD_TOPRIGHT -#undef PREDICT_8x8_DC -#undef PTR -#undef PT -#undef PL -#undef SRC - static inline void mc_dir_part(H264Context *h, Picture *pic, int n, 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, @@ -2762,7 +1684,7 @@ static inline void mc_dir_part(H264Context *h, Picture *pic, int n, int square, const int full_mx= mx>>2; const int full_my= my>>2; const int pic_width = 16*s->mb_width; - const int pic_height = 16*s->mb_height >> MB_MBAFF; + const int pic_height = 16*s->mb_height >> MB_FIELD; if(!pic->data[0]) //FIXME this is unacceptable, some senseable error concealment must be done for missing reference frames return; @@ -2784,11 +1706,11 @@ static inline void mc_dir_part(H264Context *h, Picture *pic, int n, int square, qpix_op[luma_xy](dest_y + delta, src_y + delta, h->mb_linesize); } - if(s->flags&CODEC_FLAG_GRAY) return; + if(ENABLE_GRAY && s->flags&CODEC_FLAG_GRAY) return; - if(MB_MBAFF){ + if(MB_FIELD){ // chroma offset when predicting from a field of opposite parity - my += 2 * ((s->mb_y & 1) - (h->ref_cache[list][scan8[n]] & 1)); + my += 2 * ((s->mb_y & 1) - (pic->reference - 1)); emu |= (my>>3) < 0 || (my>>3) + 8 >= (pic_height>>1); } src_cb= pic->data[1] + (mx>>3) + (my>>3)*h->mb_uvlinesize; @@ -2821,7 +1743,7 @@ static inline void mc_part_std(H264Context *h, int n, int square, int chroma_hei dest_cb += x_offset + y_offset*h->mb_uvlinesize; dest_cr += x_offset + y_offset*h->mb_uvlinesize; x_offset += 8*s->mb_x; - y_offset += 8*(s->mb_y >> MB_MBAFF); + y_offset += 8*(s->mb_y >> MB_FIELD); if(list0){ Picture *ref= &h->ref_list[0][ h->ref_cache[0][ scan8[n] ] ]; @@ -2854,7 +1776,7 @@ static inline void mc_part_weighted(H264Context *h, int n, int square, int chrom dest_cb += x_offset + y_offset*h->mb_uvlinesize; dest_cr += x_offset + y_offset*h->mb_uvlinesize; x_offset += 8*s->mb_x; - y_offset += 8*(s->mb_y >> MB_MBAFF); + y_offset += 8*(s->mb_y >> MB_FIELD); if(list0 && list1){ /* don't optimize for luma-only case, since B-frames usually @@ -3029,7 +1951,7 @@ static void hl_motion(H264Context *h, uint8_t *dest_y, uint8_t *dest_cb, uint8_t prefetch_motion(h, 1); } -static void decode_init_vlc(){ +static void decode_init_vlc(void){ static int done = 0; if (!done) { @@ -3068,56 +1990,9 @@ static void decode_init_vlc(){ } } -/** - * Sets the intra prediction function pointers. - */ -static void init_pred_ptrs(H264Context *h){ -// MpegEncContext * const s = &h->s; - - h->pred4x4[VERT_PRED ]= pred4x4_vertical_c; - h->pred4x4[HOR_PRED ]= pred4x4_horizontal_c; - h->pred4x4[DC_PRED ]= pred4x4_dc_c; - h->pred4x4[DIAG_DOWN_LEFT_PRED ]= pred4x4_down_left_c; - h->pred4x4[DIAG_DOWN_RIGHT_PRED]= pred4x4_down_right_c; - h->pred4x4[VERT_RIGHT_PRED ]= pred4x4_vertical_right_c; - h->pred4x4[HOR_DOWN_PRED ]= pred4x4_horizontal_down_c; - h->pred4x4[VERT_LEFT_PRED ]= pred4x4_vertical_left_c; - h->pred4x4[HOR_UP_PRED ]= pred4x4_horizontal_up_c; - h->pred4x4[LEFT_DC_PRED ]= pred4x4_left_dc_c; - h->pred4x4[TOP_DC_PRED ]= pred4x4_top_dc_c; - h->pred4x4[DC_128_PRED ]= pred4x4_128_dc_c; - - h->pred8x8l[VERT_PRED ]= pred8x8l_vertical_c; - h->pred8x8l[HOR_PRED ]= pred8x8l_horizontal_c; - h->pred8x8l[DC_PRED ]= pred8x8l_dc_c; - h->pred8x8l[DIAG_DOWN_LEFT_PRED ]= pred8x8l_down_left_c; - h->pred8x8l[DIAG_DOWN_RIGHT_PRED]= pred8x8l_down_right_c; - h->pred8x8l[VERT_RIGHT_PRED ]= pred8x8l_vertical_right_c; - h->pred8x8l[HOR_DOWN_PRED ]= pred8x8l_horizontal_down_c; - h->pred8x8l[VERT_LEFT_PRED ]= pred8x8l_vertical_left_c; - h->pred8x8l[HOR_UP_PRED ]= pred8x8l_horizontal_up_c; - h->pred8x8l[LEFT_DC_PRED ]= pred8x8l_left_dc_c; - h->pred8x8l[TOP_DC_PRED ]= pred8x8l_top_dc_c; - h->pred8x8l[DC_128_PRED ]= pred8x8l_128_dc_c; - - h->pred8x8[DC_PRED8x8 ]= ff_pred8x8_dc_c; - h->pred8x8[VERT_PRED8x8 ]= ff_pred8x8_vertical_c; - h->pred8x8[HOR_PRED8x8 ]= ff_pred8x8_horizontal_c; - h->pred8x8[PLANE_PRED8x8 ]= ff_pred8x8_plane_c; - h->pred8x8[LEFT_DC_PRED8x8]= pred8x8_left_dc_c; - h->pred8x8[TOP_DC_PRED8x8 ]= pred8x8_top_dc_c; - h->pred8x8[DC_128_PRED8x8 ]= ff_pred8x8_128_dc_c; - - h->pred16x16[DC_PRED8x8 ]= ff_pred16x16_dc_c; - h->pred16x16[VERT_PRED8x8 ]= ff_pred16x16_vertical_c; - h->pred16x16[HOR_PRED8x8 ]= ff_pred16x16_horizontal_c; - h->pred16x16[PLANE_PRED8x8 ]= ff_pred16x16_plane_c; - h->pred16x16[LEFT_DC_PRED8x8]= pred16x16_left_dc_c; - h->pred16x16[TOP_DC_PRED8x8 ]= pred16x16_top_dc_c; - h->pred16x16[DC_128_PRED8x8 ]= ff_pred16x16_128_dc_c; -} - static void free_tables(H264Context *h){ + int i; + H264Context *hx; av_freep(&h->intra4x4_pred_mode); av_freep(&h->chroma_pred_mode_table); av_freep(&h->cbp_table); @@ -3126,14 +2001,24 @@ static void free_tables(H264Context *h){ av_freep(&h->direct_table); av_freep(&h->non_zero_count); av_freep(&h->slice_table_base); - av_freep(&h->top_borders[1]); - av_freep(&h->top_borders[0]); h->slice_table= NULL; av_freep(&h->mb2b_xy); av_freep(&h->mb2b8_xy); - av_freep(&h->s.obmc_scratchpad); + for(i = 0; i < MAX_SPS_COUNT; i++) + av_freep(h->sps_buffers + i); + + for(i = 0; i < MAX_PPS_COUNT; i++) + av_freep(h->pps_buffers + i); + + for(i = 0; i < h->s.avctx->thread_count; i++) { + hx = h->thread_context[i]; + if(!hx) continue; + av_freep(&hx->top_borders[1]); + av_freep(&hx->top_borders[0]); + av_freep(&hx->s.obmc_scratchpad); + } } static void init_dequant8_coeff_table(H264Context *h){ @@ -3214,16 +2099,12 @@ static int alloc_tables(H264Context *h){ CHECKED_ALLOCZ(h->non_zero_count , big_mb_num * 16 * sizeof(uint8_t)) CHECKED_ALLOCZ(h->slice_table_base , (big_mb_num+s->mb_stride) * sizeof(uint8_t)) - CHECKED_ALLOCZ(h->top_borders[0] , s->mb_width * (16+8+8) * sizeof(uint8_t)) - CHECKED_ALLOCZ(h->top_borders[1] , s->mb_width * (16+8+8) * sizeof(uint8_t)) CHECKED_ALLOCZ(h->cbp_table, big_mb_num * sizeof(uint16_t)) - if( h->pps.cabac ) { - CHECKED_ALLOCZ(h->chroma_pred_mode_table, big_mb_num * sizeof(uint8_t)) - CHECKED_ALLOCZ(h->mvd_table[0], 32*big_mb_num * sizeof(uint16_t)); - CHECKED_ALLOCZ(h->mvd_table[1], 32*big_mb_num * sizeof(uint16_t)); - CHECKED_ALLOCZ(h->direct_table, 32*big_mb_num * sizeof(uint8_t)); - } + CHECKED_ALLOCZ(h->chroma_pred_mode_table, big_mb_num * sizeof(uint8_t)) + CHECKED_ALLOCZ(h->mvd_table[0], 32*big_mb_num * sizeof(uint16_t)); + CHECKED_ALLOCZ(h->mvd_table[1], 32*big_mb_num * sizeof(uint16_t)); + CHECKED_ALLOCZ(h->direct_table, 32*big_mb_num * sizeof(uint8_t)); memset(h->slice_table_base, -1, (big_mb_num+s->mb_stride) * sizeof(uint8_t)); h->slice_table= h->slice_table_base + s->mb_stride*2 + 1; @@ -3252,6 +2133,38 @@ fail: return -1; } +/** + * Mimic alloc_tables(), but for every context thread. + */ +static void clone_tables(H264Context *dst, H264Context *src){ + dst->intra4x4_pred_mode = src->intra4x4_pred_mode; + dst->non_zero_count = src->non_zero_count; + dst->slice_table = src->slice_table; + dst->cbp_table = src->cbp_table; + dst->mb2b_xy = src->mb2b_xy; + dst->mb2b8_xy = src->mb2b8_xy; + dst->chroma_pred_mode_table = src->chroma_pred_mode_table; + dst->mvd_table[0] = src->mvd_table[0]; + dst->mvd_table[1] = src->mvd_table[1]; + dst->direct_table = src->direct_table; + + dst->s.obmc_scratchpad = NULL; + ff_h264_pred_init(&dst->hpc, src->s.codec_id); +} + +/** + * Init context + * Allocate buffers which are not shared amongst multiple threads. + */ +static int context_init(H264Context *h){ + CHECKED_ALLOCZ(h->top_borders[0], h->s.mb_width * (16+8+8) * sizeof(uint8_t)) + CHECKED_ALLOCZ(h->top_borders[1], h->s.mb_width * (16+8+8) * sizeof(uint8_t)) + + return 0; +fail: + return -1; // free_tables will clean up for us +} + static void common_init(H264Context *h){ MpegEncContext * const s = &h->s; @@ -3259,7 +2172,7 @@ static void common_init(H264Context *h){ s->height = s->avctx->height; s->codec_id= s->avctx->codec->id; - init_pred_ptrs(h); + ff_h264_pred_init(&h->hpc, s->codec_id); h->dequant_coeff_pps= -1; s->unrestricted_mv=1; @@ -3283,6 +2196,7 @@ static int decode_init(AVCodecContext *avctx){ // set defaults // s->decode_mb= ff_h263_decode_mb; + s->quarter_sample = 1; s->low_delay= 1; avctx->pix_fmt= PIX_FMT_YUV420P; @@ -3296,6 +2210,7 @@ static int decode_init(AVCodecContext *avctx){ h->is_avc = 0; } + h->thread_context[0] = h; return 0; } @@ -3306,6 +2221,13 @@ static int frame_start(H264Context *h){ if(MPV_frame_start(s, s->avctx) < 0) return -1; ff_er_frame_start(s); + /* + * MPV_frame_start uses pict_type to derive key_frame. + * This is incorrect for H.264; IDR markings must be used. + * Zero here; IDR markings per slice in frame or fields are OR'd in later. + * See decode_nal_units(). + */ + s->current_picture_ptr->key_frame= 0; assert(s->linesize && s->uvlinesize); @@ -3322,18 +2244,19 @@ static int frame_start(H264Context *h){ /* can't be in alloc_tables because linesize isn't known there. * FIXME: redo bipred weight to not require extra buffer? */ - if(!s->obmc_scratchpad) - s->obmc_scratchpad = av_malloc(16*2*s->linesize + 8*2*s->uvlinesize); + for(i = 0; i < s->avctx->thread_count; i++) + if(!h->thread_context[i]->s.obmc_scratchpad) + h->thread_context[i]->s.obmc_scratchpad = av_malloc(16*2*s->linesize + 8*2*s->uvlinesize); /* some macroblocks will be accessed before they're available */ - if(FRAME_MBAFF) + if(FRAME_MBAFF || s->avctx->thread_count > 1) memset(h->slice_table, -1, (s->mb_height*s->mb_stride-1) * sizeof(uint8_t)); // s->decode= (s->flags&CODEC_FLAG_PSNR) || !s->encoding || s->current_picture.reference /*|| h->contains_intra*/ || 1; return 0; } -static inline void backup_mb_border(H264Context *h, uint8_t *src_y, uint8_t *src_cb, uint8_t *src_cr, int linesize, int uvlinesize){ +static inline void backup_mb_border(H264Context *h, uint8_t *src_y, uint8_t *src_cb, uint8_t *src_cr, int linesize, int uvlinesize, int simple){ MpegEncContext * const s = &h->s; int i; @@ -3351,7 +2274,7 @@ static inline void backup_mb_border(H264Context *h, uint8_t *src_y, uint8_t *src *(uint64_t*)(h->top_borders[0][s->mb_x]+0)= *(uint64_t*)(src_y + 16*linesize); *(uint64_t*)(h->top_borders[0][s->mb_x]+8)= *(uint64_t*)(src_y +8+16*linesize); - if(!(s->flags&CODEC_FLAG_GRAY)){ + if(simple || !ENABLE_GRAY || !(s->flags&CODEC_FLAG_GRAY)){ h->left_border[17 ]= h->top_borders[0][s->mb_x][16+7]; h->left_border[17+9]= h->top_borders[0][s->mb_x][24+7]; for(i=1; i<9; i++){ @@ -3363,12 +2286,22 @@ static inline void backup_mb_border(H264Context *h, uint8_t *src_y, uint8_t *src } } -static inline void xchg_mb_border(H264Context *h, uint8_t *src_y, uint8_t *src_cb, uint8_t *src_cr, int linesize, int uvlinesize, int xchg){ +static inline void xchg_mb_border(H264Context *h, uint8_t *src_y, uint8_t *src_cb, uint8_t *src_cr, int linesize, int uvlinesize, int xchg, int simple){ MpegEncContext * const s = &h->s; int temp8, i; uint64_t temp64; - int deblock_left = (s->mb_x > 0); - int deblock_top = (s->mb_y > 0); + int deblock_left; + int deblock_top; + int mb_xy; + + if(h->deblocking_filter == 2) { + mb_xy = s->mb_x + s->mb_y*s->mb_stride; + deblock_left = h->slice_table[mb_xy] == h->slice_table[mb_xy - 1]; + deblock_top = h->slice_table[mb_xy] == h->slice_table[h->top_mb_xy]; + } else { + deblock_left = (s->mb_x > 0); + deblock_top = (s->mb_y > 0); + } src_y -= linesize + 1; src_cb -= uvlinesize + 1; @@ -3394,7 +2327,7 @@ b= t; } } - if(!(s->flags&CODEC_FLAG_GRAY)){ + if(simple || !ENABLE_GRAY || !(s->flags&CODEC_FLAG_GRAY)){ if(deblock_left){ for(i = !deblock_top; i<9; i++){ XCHG(h->left_border[i+17 ], src_cb[i*uvlinesize], temp8, xchg); @@ -3429,7 +2362,7 @@ static inline void backup_pair_border(H264Context *h, uint8_t *src_y, uint8_t *s *(uint64_t*)(h->top_borders[1][s->mb_x]+0)= *(uint64_t*)(src_y + 33*linesize); *(uint64_t*)(h->top_borders[1][s->mb_x]+8)= *(uint64_t*)(src_y +8+33*linesize); - if(!(s->flags&CODEC_FLAG_GRAY)){ + if(!ENABLE_GRAY || !(s->flags&CODEC_FLAG_GRAY)){ h->left_border[34 ]= h->top_borders[0][s->mb_x][16+7]; h->left_border[34+ 1]= h->top_borders[1][s->mb_x][16+7]; h->left_border[34+18 ]= h->top_borders[0][s->mb_x][24+7]; @@ -3481,7 +2414,7 @@ b= t; } } - if(!(s->flags&CODEC_FLAG_GRAY)){ + if(!ENABLE_GRAY || !(s->flags&CODEC_FLAG_GRAY)){ if(deblock_left){ for(i = (!deblock_top) << 1; i<18; i++){ XCHG(h->left_border[i+34 ], src_cb[i*uvlinesize], temp8, xchg); @@ -3497,7 +2430,7 @@ b= t; } } -static void av_always_inline hl_decode_mb_internal(H264Context *h, int simple){ +static av_always_inline void hl_decode_mb_internal(H264Context *h, int simple){ MpegEncContext * const s = &h->s; const int mb_x= s->mb_x; const int mb_y= s->mb_y; @@ -3535,13 +2468,13 @@ static void av_always_inline hl_decode_mb_internal(H264Context *h, int simple){ 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); + fill_rectangle(ref, 4, 4, 8, (16+*ref)^(s->mb_y&1), 1); }else{ for(i=0; i<16; i+=4){ //FIXME can refs be smaller than 8x8 when !direct_8x8_inference ? 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); + fill_rectangle(&h->ref_cache[list][scan8[i]], 2, 2, 8, (16+ref)^(s->mb_y&1), 1); } } } @@ -3601,11 +2534,11 @@ static void av_always_inline hl_decode_mb_internal(H264Context *h, int simple){ } else { if(IS_INTRA(mb_type)){ if(h->deblocking_filter && (simple || !FRAME_MBAFF)) - xchg_mb_border(h, dest_y, dest_cb, dest_cr, linesize, uvlinesize, 1); + xchg_mb_border(h, dest_y, dest_cb, dest_cr, linesize, uvlinesize, 1, simple); - if(simple || !(s->flags&CODEC_FLAG_GRAY)){ - h->pred8x8[ h->chroma_pred_mode ](dest_cb, uvlinesize); - h->pred8x8[ h->chroma_pred_mode ](dest_cr, uvlinesize); + if(simple || !ENABLE_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); } if(IS_INTRA4x4(mb_type)){ @@ -3615,7 +2548,7 @@ static void av_always_inline hl_decode_mb_internal(H264Context *h, int simple){ uint8_t * const ptr= dest_y + block_offset[i]; const int dir= h->intra4x4_pred_mode_cache[ scan8[i] ]; const int nnz = h->non_zero_count_cache[ scan8[i] ]; - h->pred8x8l[ dir ](ptr, (h->topleft_samples_available<<i)&0x8000, + h->hpc.pred8x8l[ dir ](ptr, (h->topleft_samples_available<<i)&0x8000, (h->topright_samples_available<<i)&0x4000, linesize); if(nnz){ if(nnz == 1 && h->mb[i*16]) @@ -3642,7 +2575,7 @@ static void av_always_inline hl_decode_mb_internal(H264Context *h, int simple){ }else topright= NULL; - h->pred4x4[ dir ](ptr, topright, linesize); + h->hpc.pred4x4[ dir ](ptr, topright, linesize); nnz = h->non_zero_count_cache[ scan8[i] ]; if(nnz){ if(is_h264){ @@ -3656,15 +2589,15 @@ static void av_always_inline hl_decode_mb_internal(H264Context *h, int simple){ } } }else{ - h->pred16x16[ h->intra16x16_pred_mode ](dest_y , linesize); + h->hpc.pred16x16[ h->intra16x16_pred_mode ](dest_y , linesize); if(is_h264){ if(!transform_bypass) - h264_luma_dc_dequant_idct_c(h->mb, s->qscale, h->dequant4_coeff[IS_INTRA(mb_type) ? 0:3][s->qscale][0]); + h264_luma_dc_dequant_idct_c(h->mb, s->qscale, h->dequant4_coeff[0][s->qscale][0]); }else svq3_luma_dc_dequant_idct_c(h->mb, s->qscale); } if(h->deblocking_filter && (simple || !FRAME_MBAFF)) - xchg_mb_border(h, dest_y, dest_cb, dest_cr, linesize, uvlinesize, 0); + xchg_mb_border(h, dest_y, dest_cb, dest_cr, linesize, uvlinesize, 0, simple); }else if(is_h264){ hl_motion(h, dest_y, dest_cb, dest_cr, s->me.qpel_put, s->dsp.put_h264_chroma_pixels_tab, @@ -3704,15 +2637,15 @@ static void av_always_inline hl_decode_mb_internal(H264Context *h, int simple){ } } - if(simple || !(s->flags&CODEC_FLAG_GRAY)){ + if(simple || !ENABLE_GRAY || !(s->flags&CODEC_FLAG_GRAY)){ uint8_t *dest[2] = {dest_cb, dest_cr}; if(transform_bypass){ idct_add = idct_dc_add = s->dsp.add_pixels4; }else{ idct_add = s->dsp.h264_idct_add; idct_dc_add = s->dsp.h264_idct_dc_add; - chroma_dc_dequant_idct_c(h->mb + 16*16, h->chroma_qp, h->dequant4_coeff[IS_INTRA(mb_type) ? 1:4][h->chroma_qp][0]); - chroma_dc_dequant_idct_c(h->mb + 16*16+4*16, h->chroma_qp, h->dequant4_coeff[IS_INTRA(mb_type) ? 2:5][h->chroma_qp][0]); + chroma_dc_dequant_idct_c(h->mb + 16*16, h->chroma_qp[0], 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->chroma_qp[1], h->dequant4_coeff[IS_INTRA(mb_type) ? 2:5][h->chroma_qp[1]][0]); } if(is_h264){ for(i=16; i<16+8; i++){ @@ -3754,17 +2687,19 @@ static void av_always_inline hl_decode_mb_internal(H264Context *h, int simple){ s->mb_y--; tprintf(h->s.avctx, "call mbaff filter_mb mb_x:%d mb_y:%d pair_dest_y = %p, dest_y = %p\n", mb_x, mb_y, pair_dest_y, dest_y); fill_caches(h, mb_type_top, 1); //FIXME don't fill stuff which isn't used by filter_mb - h->chroma_qp = get_chroma_qp(h->pps.chroma_qp_index_offset, s->current_picture.qscale_table[mb_xy]); + h->chroma_qp[0] = get_chroma_qp(h, 0, s->current_picture.qscale_table[mb_xy]); + h->chroma_qp[1] = get_chroma_qp(h, 1, s->current_picture.qscale_table[mb_xy]); filter_mb(h, mb_x, mb_y, pair_dest_y, pair_dest_cb, pair_dest_cr, linesize, uvlinesize); // bottom s->mb_y++; tprintf(h->s.avctx, "call mbaff filter_mb\n"); fill_caches(h, mb_type_bottom, 1); //FIXME don't fill stuff which isn't used by filter_mb - h->chroma_qp = get_chroma_qp(h->pps.chroma_qp_index_offset, s->current_picture.qscale_table[mb_xy+s->mb_stride]); + h->chroma_qp[0] = get_chroma_qp(h, 0, s->current_picture.qscale_table[mb_xy+s->mb_stride]); + h->chroma_qp[1] = get_chroma_qp(h, 1, s->current_picture.qscale_table[mb_xy+s->mb_stride]); filter_mb(h, mb_x, mb_y+1, dest_y, dest_cb, dest_cr, linesize, uvlinesize); } else { tprintf(h->s.avctx, "call filter_mb\n"); - backup_mb_border(h, dest_y, dest_cb, dest_cr, linesize, uvlinesize); + backup_mb_border(h, dest_y, dest_cb, dest_cr, linesize, uvlinesize, simple); fill_caches(h, mb_type, 1); //FIXME don't fill stuff which isn't used by filter_mb filter_mb_fast(h, mb_x, mb_y, dest_y, dest_cb, dest_cr, linesize, uvlinesize); } @@ -3791,7 +2726,7 @@ static void hl_decode_mb(H264Context *h){ const int mb_y= s->mb_y; const int mb_xy= mb_x + mb_y*s->mb_stride; const int mb_type= s->current_picture.mb_type[mb_xy]; - int is_complex = FRAME_MBAFF || MB_FIELD || IS_INTRA_PCM(mb_type) || s->codec_id != CODEC_ID_H264 || (s->flags&CODEC_FLAG_GRAY) || s->encoding; + int is_complex = FRAME_MBAFF || MB_FIELD || IS_INTRA_PCM(mb_type) || s->codec_id != CODEC_ID_H264 || (ENABLE_GRAY && (s->flags&CODEC_FLAG_GRAY)) || s->encoding; if(!s->decode) return; @@ -3801,6 +2736,105 @@ static void hl_decode_mb(H264Context *h){ else hl_decode_mb_simple(h); } +static void pic_as_field(Picture *pic, const int parity){ + int i; + for (i = 0; i < 4; ++i) { + if (parity == PICT_BOTTOM_FIELD) + pic->data[i] += pic->linesize[i]; + pic->reference = parity; + pic->linesize[i] *= 2; + } +} + +static int split_field_copy(Picture *dest, Picture *src, + int parity, int id_add){ + int match = !!(src->reference & parity); + + if (match) { + *dest = *src; + pic_as_field(dest, parity); + dest->pic_id *= 2; + dest->pic_id += id_add; + } + + return match; +} + +/** + * Split one reference list into field parts, interleaving by parity + * as per H.264 spec section 8.2.4.2.5. Output fields have their data pointers + * set to look at the actual start of data for that field. + * + * @param dest output list + * @param dest_len maximum number of fields to put in dest + * @param src the source reference list containing fields and/or field pairs + * (aka short_ref/long_ref, or + * refFrameListXShortTerm/refFrameListLongTerm in spec-speak) + * @param src_len number of Picture's in source (pairs and unmatched fields) + * @param parity the parity of the picture being decoded/needing + * these ref pics (PICT_{TOP,BOTTOM}_FIELD) + * @return number of fields placed in dest + */ +static int split_field_half_ref_list(Picture *dest, int dest_len, + Picture *src, int src_len, int parity){ + int same_parity = 1; + int same_i = 0; + int opp_i = 0; + int out_i; + int field_output; + + for (out_i = 0; out_i < dest_len; out_i += field_output) { + if (same_parity && same_i < src_len) { + field_output = split_field_copy(dest + out_i, src + same_i, + parity, 1); + same_parity = !field_output; + same_i++; + + } else if (opp_i < src_len) { + field_output = split_field_copy(dest + out_i, src + opp_i, + PICT_FRAME - parity, 0); + same_parity = field_output; + opp_i++; + + } else { + break; + } + } + + return out_i; +} + +/** + * Split the reference frame list into a reference field list. + * This implements H.264 spec 8.2.4.2.5 for a combined input list. + * The input list contains both reference field pairs and + * unmatched reference fields; it is ordered as spec describes + * RefPicListX for frames in 8.2.4.2.1 and 8.2.4.2.3, except that + * unmatched field pairs are also present. Conceptually this is equivalent + * to concatenation of refFrameListXShortTerm with refFrameListLongTerm. + * + * @param dest output reference list where ordered fields are to be placed + * @param dest_len max number of fields to place at dest + * @param src source reference list, as described above + * @param src_len number of pictures (pairs and unmatched fields) in src + * @param parity parity of field being currently decoded + * (one of PICT_{TOP,BOTTOM}_FIELD) + * @param long_i index into src array that holds first long reference picture, + * or src_len if no long refs present. + */ +static int split_field_ref_list(Picture *dest, int dest_len, + Picture *src, int src_len, + int parity, int long_i){ + + int i = split_field_half_ref_list(dest, dest_len, src, long_i, parity); + dest += i; + dest_len -= i; + + i += split_field_half_ref_list(dest, dest_len, src + long_i, + src_len - long_i, parity); + return i; +} + /** * fills the default_ref_list. */ @@ -3808,9 +2842,25 @@ static int fill_default_ref_list(H264Context *h){ MpegEncContext * const s = &h->s; int i; int smallest_poc_greater_than_current = -1; + int structure_sel; Picture sorted_short_ref[32]; + Picture field_entry_list[2][32]; + Picture *frame_list[2]; + + if (FIELD_PICTURE) { + structure_sel = PICT_FRAME; + frame_list[0] = field_entry_list[0]; + frame_list[1] = field_entry_list[1]; + } else { + structure_sel = 0; + frame_list[0] = h->default_ref_list[0]; + frame_list[1] = h->default_ref_list[1]; + } if(h->slice_type==B_TYPE){ + int list; + int len[2]; + int short_len[2]; int out_i; int limit= INT_MIN; @@ -3838,71 +2888,92 @@ static int fill_default_ref_list(H264Context *h){ } } } - } - if(s->picture_structure == PICT_FRAME){ - if(h->slice_type==B_TYPE){ - int list; - tprintf(h->s.avctx, "current poc: %d, smallest_poc_greater_than_current: %d\n", s->current_picture_ptr->poc, smallest_poc_greater_than_current); + tprintf(h->s.avctx, "current poc: %d, smallest_poc_greater_than_current: %d\n", s->current_picture_ptr->poc, smallest_poc_greater_than_current); - // find the largest poc - for(list=0; list<2; list++){ - int index = 0; - int j= -99; - int step= list ? -1 : 1; - - for(i=0; i<h->short_ref_count && index < h->ref_count[list]; i++, j+=step) { - while(j<0 || j>= h->short_ref_count){ - if(j != -99 && step == (list ? -1 : 1)) - return -1; - step = -step; - j= smallest_poc_greater_than_current + (step>>1); - } - if(sorted_short_ref[j].reference != 3) continue; - h->default_ref_list[list][index ]= sorted_short_ref[j]; - h->default_ref_list[list][index++].pic_id= sorted_short_ref[j].frame_num; + // find the largest poc + for(list=0; list<2; list++){ + int index = 0; + int j= -99; + int step= list ? -1 : 1; + + for(i=0; i<h->short_ref_count && index < h->ref_count[list]; i++, j+=step) { + int sel; + while(j<0 || j>= h->short_ref_count){ + if(j != -99 && step == (list ? -1 : 1)) + return -1; + step = -step; + j= smallest_poc_greater_than_current + (step>>1); } + sel = sorted_short_ref[j].reference | structure_sel; + if(sel != PICT_FRAME) continue; + frame_list[list][index ]= sorted_short_ref[j]; + frame_list[list][index++].pic_id= sorted_short_ref[j].frame_num; + } + short_len[list] = index; - for(i = 0; i < 16 && index < h->ref_count[ list ]; i++){ - if(h->long_ref[i] == NULL) continue; - if(h->long_ref[i]->reference != 3) continue; + for(i = 0; i < 16 && index < h->ref_count[ list ]; i++){ + int sel; + if(h->long_ref[i] == NULL) continue; + sel = h->long_ref[i]->reference | structure_sel; + if(sel != PICT_FRAME) continue; - h->default_ref_list[ list ][index ]= *h->long_ref[i]; - h->default_ref_list[ list ][index++].pic_id= i;; - } + frame_list[ list ][index ]= *h->long_ref[i]; + frame_list[ list ][index++].pic_id= i; + } + len[list] = index; + } - if(list && (smallest_poc_greater_than_current<=0 || smallest_poc_greater_than_current>=h->short_ref_count) && (1 < index)){ - // swap the two first elements of L1 when - // L0 and L1 are identical - Picture temp= h->default_ref_list[1][0]; - h->default_ref_list[1][0] = h->default_ref_list[1][1]; - h->default_ref_list[1][1] = temp; - } + for(list=0; list<2; list++){ + if (FIELD_PICTURE) + len[list] = split_field_ref_list(h->default_ref_list[list], + h->ref_count[list], + frame_list[list], + len[list], + s->picture_structure, + short_len[list]); + + // swap the two first elements of L1 when L0 and L1 are identical + if(list && len[0] > 1 && len[0] == len[1]) + for(i=0; h->default_ref_list[0][i].data[0] == h->default_ref_list[1][i].data[0]; i++) + if(i == len[0]){ + FFSWAP(Picture, h->default_ref_list[1][0], h->default_ref_list[1][1]); + break; + } - if(index < h->ref_count[ list ]) - memset(&h->default_ref_list[list][index], 0, sizeof(Picture)*(h->ref_count[ list ] - index)); - } - }else{ - int index=0; - for(i=0; i<h->short_ref_count; i++){ - if(h->short_ref[i]->reference != 3) continue; //FIXME refernce field shit - h->default_ref_list[0][index ]= *h->short_ref[i]; - h->default_ref_list[0][index++].pic_id= h->short_ref[i]->frame_num; - } - for(i = 0; i < 16; i++){ - if(h->long_ref[i] == NULL) continue; - if(h->long_ref[i]->reference != 3) continue; - h->default_ref_list[0][index ]= *h->long_ref[i]; - h->default_ref_list[0][index++].pic_id= i;; - } - if(index < h->ref_count[0]) - memset(&h->default_ref_list[0][index], 0, sizeof(Picture)*(h->ref_count[0] - index)); + if(len[list] < h->ref_count[ list ]) + memset(&h->default_ref_list[list][len[list]], 0, sizeof(Picture)*(h->ref_count[ list ] - len[list])); } - }else{ //FIELD - if(h->slice_type==B_TYPE){ - }else{ - //FIXME second field balh + + + }else{ + int index=0; + int short_len; + for(i=0; i<h->short_ref_count; i++){ + int sel; + sel = h->short_ref[i]->reference | structure_sel; + if(sel != PICT_FRAME) continue; + frame_list[0][index ]= *h->short_ref[i]; + frame_list[0][index++].pic_id= h->short_ref[i]->frame_num; + } + short_len = index; + for(i = 0; i < 16; i++){ + int sel; + if(h->long_ref[i] == NULL) continue; + sel = h->long_ref[i]->reference | structure_sel; + if(sel != PICT_FRAME) continue; + frame_list[0][index ]= *h->long_ref[i]; + frame_list[0][index++].pic_id= i; } + + if (FIELD_PICTURE) + index = split_field_ref_list(h->default_ref_list[0], + h->ref_count[0], frame_list[0], + index, s->picture_structure, + short_len); + + if(index < h->ref_count[0]) + memset(&h->default_ref_list[0][index], 0, sizeof(Picture)*(h->ref_count[0] - index)); } #ifdef TRACE for (i=0; i<h->ref_count[0]; i++) { @@ -3910,7 +2981,7 @@ static int fill_default_ref_list(H264Context *h){ } if(h->slice_type==B_TYPE){ for (i=0; i<h->ref_count[1]; i++) { - tprintf(h->s.avctx, "List1: %s fn:%d 0x%p\n", (h->default_ref_list[1][i].long_ref ? "LT" : "ST"), h->default_ref_list[1][i].pic_id, h->default_ref_list[0][i].data[0]); + tprintf(h->s.avctx, "List1: %s fn:%d 0x%p\n", (h->default_ref_list[1][i].long_ref ? "LT" : "ST"), h->default_ref_list[1][i].pic_id, h->default_ref_list[1][i].data[0]); } } #endif @@ -3920,9 +2991,33 @@ static int fill_default_ref_list(H264Context *h){ static void print_short_term(H264Context *h); static void print_long_term(H264Context *h); +/** + * Extract structure information about the picture described by pic_num in + * the current decoding context (frame or field). Note that pic_num is + * picture number without wrapping (so, 0<=pic_num<max_pic_num). + * @param pic_num picture number for which to extract structure information + * @param structure one of PICT_XXX describing structure of picture + * with pic_num + * @return frame number (short term) or long term index of picture + * described by pic_num + */ +static int pic_num_extract(H264Context *h, int pic_num, int *structure){ + MpegEncContext * const s = &h->s; + + *structure = s->picture_structure; + if(FIELD_PICTURE){ + if (!(pic_num & 1)) + /* opposite field */ + *structure ^= PICT_FRAME; + pic_num >>= 1; + } + + return pic_num; +} + static int decode_ref_pic_list_reordering(H264Context *h){ MpegEncContext * const s = &h->s; - int list, index; + int list, index, pic_structure; print_short_term(h); print_long_term(h); @@ -3951,8 +3046,9 @@ static int decode_ref_pic_list_reordering(H264Context *h){ if(reordering_of_pic_nums_idc<3){ if(reordering_of_pic_nums_idc<2){ const unsigned int abs_diff_pic_num= get_ue_golomb(&s->gb) + 1; + int frame_num; - if(abs_diff_pic_num >= h->max_pic_num){ + if(abs_diff_pic_num > h->max_pic_num){ av_log(h->s.avctx, AV_LOG_ERROR, "abs_diff_pic_num overflow\n"); return -1; } @@ -3961,25 +3057,34 @@ static int decode_ref_pic_list_reordering(H264Context *h){ else pred+= abs_diff_pic_num; pred &= h->max_pic_num - 1; + frame_num = pic_num_extract(h, pred, &pic_structure); + for(i= h->short_ref_count-1; i>=0; i--){ ref = h->short_ref[i]; - assert(ref->reference == 3); + assert(ref->reference); assert(!ref->long_ref); - if(ref->data[0] != NULL && ref->frame_num == pred && ref->long_ref == 0) // ignore non existing pictures by testing data[0] pointer + if(ref->data[0] != NULL && + ref->frame_num == frame_num && + (ref->reference & pic_structure) && + ref->long_ref == 0) // ignore non existing pictures by testing data[0] pointer break; } if(i>=0) - ref->pic_id= ref->frame_num; + ref->pic_id= pred; }else{ + int long_idx; pic_id= get_ue_golomb(&s->gb); //long_term_pic_idx - if(pic_id>31){ + + long_idx= pic_num_extract(h, pic_id, &pic_structure); + + if(long_idx>31){ av_log(h->s.avctx, AV_LOG_ERROR, "long_term_pic_idx overflow\n"); return -1; } - ref = h->long_ref[pic_id]; - if(ref){ + ref = h->long_ref[long_idx]; + assert(!(ref && !ref->reference)); + if(ref && (ref->reference & pic_structure)){ ref->pic_id= pic_id; - assert(ref->reference == 3); assert(ref->long_ref); i=0; }else{ @@ -3999,6 +3104,9 @@ static int decode_ref_pic_list_reordering(H264Context *h){ h->ref_list[list][i]= h->ref_list[list][i-1]; } h->ref_list[list][index]= *ref; + if (FIELD_PICTURE){ + pic_as_field(&h->ref_list[list][index], pic_structure); + } } }else{ av_log(h->s.avctx, AV_LOG_ERROR, "illegal reordering_of_pic_nums_idc\n"); @@ -4029,9 +3137,11 @@ static void fill_mbaff_ref_list(H264Context *h){ field[0] = *frame; for(j=0; j<3; j++) field[0].linesize[j] <<= 1; + field[0].reference = PICT_TOP_FIELD; field[1] = field[0]; for(j=0; j<3; j++) field[1].data[j] += frame->linesize[j]; + field[1].reference = PICT_BOTTOM_FIELD; h->luma_weight[list][16+2*i] = h->luma_weight[list][16+2*i+1] = h->luma_weight[list][i]; h->luma_offset[list][16+2*i] = h->luma_offset[list][16+2*i+1] = h->luma_offset[list][i]; @@ -4137,17 +3247,32 @@ static void implicit_weight_table(H264Context *h){ } } -static inline void unreference_pic(H264Context *h, Picture *pic){ +/** + * Mark a picture as no longer needed for reference. The refmask + * argument allows unreferencing of individual fields or the whole frame. + * If the picture becomes entirely unreferenced, but is being held for + * display purposes, it is marked as such. + * @param refmask mask of fields to unreference; the mask is bitwise + * anded with the reference marking of pic + * @return non-zero if pic becomes entirely unreferenced (except possibly + * for display purposes) zero if one of the fields remains in + * reference + */ +static inline int unreference_pic(H264Context *h, Picture *pic, int refmask){ int i; - pic->reference=0; - if(pic == h->delayed_output_pic) - pic->reference=1; - else{ - for(i = 0; h->delayed_pic[i]; i++) - if(pic == h->delayed_pic[i]){ - pic->reference=1; - break; - } + if (pic->reference &= refmask) { + return 0; + } else { + if(pic == h->delayed_output_pic) + pic->reference=DELAYED_PIC_REF; + else{ + for(i = 0; h->delayed_pic[i]; i++) + if(pic == h->delayed_pic[i]){ + pic->reference=DELAYED_PIC_REF; + break; + } + } + return 1; } } @@ -4159,14 +3284,14 @@ static void idr(H264Context *h){ for(i=0; i<16; i++){ if (h->long_ref[i] != NULL) { - unreference_pic(h, h->long_ref[i]); + unreference_pic(h, h->long_ref[i], 0); h->long_ref[i]= NULL; } } h->long_ref_count=0; for(i=0; i<h->short_ref_count; i++){ - unreference_pic(h, h->short_ref[i]); + unreference_pic(h, h->short_ref[i], 0); h->short_ref[i]= NULL; } h->short_ref_count=0; @@ -4187,27 +3312,28 @@ static void flush_dpb(AVCodecContext *avctx){ idr(h); if(h->s.current_picture_ptr) h->s.current_picture_ptr->reference= 0; + h->s.first_field= 0; + ff_mpeg_flush(avctx); } /** - * - * @return the removed picture or NULL if an error occurs + * Find a Picture in the short term reference list by frame number. + * @param frame_num frame number to search for + * @param idx the index into h->short_ref where returned picture is found + * undefined if no picture found. + * @return pointer to the found picture, or NULL if no pic with the provided + * frame number is found */ -static Picture * remove_short(H264Context *h, int frame_num){ +static Picture * find_short(H264Context *h, int frame_num, int *idx){ MpegEncContext * const s = &h->s; int i; - if(s->avctx->debug&FF_DEBUG_MMCO) - av_log(h->s.avctx, AV_LOG_DEBUG, "remove short %d count %d\n", frame_num, h->short_ref_count); - for(i=0; i<h->short_ref_count; i++){ Picture *pic= h->short_ref[i]; if(s->avctx->debug&FF_DEBUG_MMCO) av_log(h->s.avctx, AV_LOG_DEBUG, "%d %d %p\n", i, pic->frame_num, pic); - if(pic->frame_num == frame_num){ - h->short_ref[i]= NULL; - memmove(&h->short_ref[i], &h->short_ref[i+1], (h->short_ref_count - i - 1)*sizeof(Picture*)); - h->short_ref_count--; + if(pic->frame_num == frame_num) { + *idx = i; return pic; } } @@ -4215,6 +3341,49 @@ static Picture * remove_short(H264Context *h, int frame_num){ } /** + * Remove a picture from the short term reference list by its index in + * that list. This does no checking on the provided index; it is assumed + * to be valid. Other list entries are shifted down. + * @param i index into h->short_ref of picture to remove. + */ +static void remove_short_at_index(H264Context *h, int i){ + assert(i > 0 && i < h->short_ref_count); + h->short_ref[i]= NULL; + if (--h->short_ref_count) + memmove(&h->short_ref[i], &h->short_ref[i+1], (h->short_ref_count - i)*sizeof(Picture*)); +} + +/** + * + * @return the removed picture or NULL if an error occurs + */ +static Picture * remove_short(H264Context *h, int frame_num){ + MpegEncContext * const s = &h->s; + Picture *pic; + int i; + + if(s->avctx->debug&FF_DEBUG_MMCO) + av_log(h->s.avctx, AV_LOG_DEBUG, "remove short %d count %d\n", frame_num, h->short_ref_count); + + pic = find_short(h, frame_num, &i); + if (pic) + remove_short_at_index(h, i); + + return pic; +} + +/** + * Remove a picture from the long term reference list by its index in + * that list. This does no checking on the provided index; it is assumed + * to be valid. The removed entry is set to NULL. Other entries are unaffected. + * @param i index into h->long_ref of picture to remove. + */ +static void remove_long_at_index(H264Context *h, int i){ + h->long_ref[i]= NULL; + h->long_ref_count--; +} + +/** * * @return the removed picture or NULL if an error occurs */ @@ -4222,8 +3391,8 @@ static Picture * remove_long(H264Context *h, int i){ Picture *pic; pic= h->long_ref[i]; - h->long_ref[i]= NULL; - if(pic) h->long_ref_count--; + if (pic) + remove_long_at_index(h, i); return pic; } @@ -4264,77 +3433,143 @@ static void print_long_term(H264Context *h) { static int execute_ref_pic_marking(H264Context *h, MMCO *mmco, int mmco_count){ MpegEncContext * const s = &h->s; int i, j; - int current_is_long=0; + int current_ref_assigned=0; Picture *pic; if((s->avctx->debug&FF_DEBUG_MMCO) && mmco_count==0) av_log(h->s.avctx, AV_LOG_DEBUG, "no mmco here\n"); for(i=0; i<mmco_count; i++){ + int structure, frame_num, unref_pic; if(s->avctx->debug&FF_DEBUG_MMCO) - av_log(h->s.avctx, AV_LOG_DEBUG, "mmco:%d %d %d\n", h->mmco[i].opcode, h->mmco[i].short_frame_num, h->mmco[i].long_index); + av_log(h->s.avctx, AV_LOG_DEBUG, "mmco:%d %d %d\n", h->mmco[i].opcode, h->mmco[i].short_pic_num, h->mmco[i].long_arg); switch(mmco[i].opcode){ case MMCO_SHORT2UNUSED: - pic= remove_short(h, mmco[i].short_frame_num); - if(pic) - unreference_pic(h, pic); - else if(s->avctx->debug&FF_DEBUG_MMCO) - av_log(h->s.avctx, AV_LOG_DEBUG, "mmco: remove_short() failure\n"); + if(s->avctx->debug&FF_DEBUG_MMCO) + av_log(h->s.avctx, AV_LOG_DEBUG, "mmco: unref short %d count %d\n", h->mmco[i].short_pic_num, h->short_ref_count); + frame_num = pic_num_extract(h, mmco[i].short_pic_num, &structure); + pic = find_short(h, frame_num, &j); + if (pic) { + if (unreference_pic(h, pic, structure ^ PICT_FRAME)) + remove_short_at_index(h, j); + } else if(s->avctx->debug&FF_DEBUG_MMCO) + av_log(h->s.avctx, AV_LOG_DEBUG, "mmco: unref short failure\n"); break; case MMCO_SHORT2LONG: - pic= remove_long(h, mmco[i].long_index); - if(pic) unreference_pic(h, pic); + if (FIELD_PICTURE && mmco[i].long_arg < h->long_ref_count && + h->long_ref[mmco[i].long_arg]->frame_num == + mmco[i].short_pic_num / 2) { + /* do nothing, we've already moved this field pair. */ + } else { + int frame_num = mmco[i].short_pic_num >> FIELD_PICTURE; - h->long_ref[ mmco[i].long_index ]= remove_short(h, mmco[i].short_frame_num); - if (h->long_ref[ mmco[i].long_index ]){ - h->long_ref[ mmco[i].long_index ]->long_ref=1; - h->long_ref_count++; + pic= remove_long(h, mmco[i].long_arg); + if(pic) unreference_pic(h, pic, 0); + + h->long_ref[ mmco[i].long_arg ]= remove_short(h, frame_num); + if (h->long_ref[ mmco[i].long_arg ]){ + h->long_ref[ mmco[i].long_arg ]->long_ref=1; + h->long_ref_count++; + } } break; case MMCO_LONG2UNUSED: - pic= remove_long(h, mmco[i].long_index); - if(pic) - unreference_pic(h, pic); - else if(s->avctx->debug&FF_DEBUG_MMCO) - av_log(h->s.avctx, AV_LOG_DEBUG, "mmco: remove_long() failure\n"); + j = pic_num_extract(h, mmco[i].long_arg, &structure); + pic = h->long_ref[j]; + if (pic) { + if (unreference_pic(h, pic, structure ^ PICT_FRAME)) + remove_long_at_index(h, j); + } else if(s->avctx->debug&FF_DEBUG_MMCO) + av_log(h->s.avctx, AV_LOG_DEBUG, "mmco: unref long failure\n"); break; case MMCO_LONG: - pic= remove_long(h, mmco[i].long_index); - if(pic) unreference_pic(h, pic); + unref_pic = 1; + if (FIELD_PICTURE && !s->first_field) { + if (h->long_ref[mmco[i].long_arg] == s->current_picture_ptr) { + /* Just mark second field as referenced */ + unref_pic = 0; + } else if (s->current_picture_ptr->reference) { + /* First field in pair is in short term list or + * at a different long term index. + * This is not allowed; see 7.4.3, notes 2 and 3. + * Report the problem and keep the pair where it is, + * and mark this field valid. + */ + av_log(h->s.avctx, AV_LOG_ERROR, + "illegal long term reference assignment for second " + "field in complementary field pair (first field is " + "short term or has non-matching long index)\n"); + unref_pic = 0; + } + } - h->long_ref[ mmco[i].long_index ]= s->current_picture_ptr; - h->long_ref[ mmco[i].long_index ]->long_ref=1; - h->long_ref_count++; + if (unref_pic) { + pic= remove_long(h, mmco[i].long_arg); + if(pic) unreference_pic(h, pic, 0); - current_is_long=1; + h->long_ref[ mmco[i].long_arg ]= s->current_picture_ptr; + h->long_ref[ mmco[i].long_arg ]->long_ref=1; + h->long_ref_count++; + } + + s->current_picture_ptr->reference |= s->picture_structure; + current_ref_assigned=1; break; case MMCO_SET_MAX_LONG: - assert(mmco[i].long_index <= 16); + assert(mmco[i].long_arg <= 16); // just remove the long term which index is greater than new max - for(j = mmco[i].long_index; j<16; j++){ + for(j = mmco[i].long_arg; j<16; j++){ pic = remove_long(h, j); - if (pic) unreference_pic(h, pic); + if (pic) unreference_pic(h, pic, 0); } break; case MMCO_RESET: while(h->short_ref_count){ pic= remove_short(h, h->short_ref[0]->frame_num); - if(pic) unreference_pic(h, pic); + if(pic) unreference_pic(h, pic, 0); } for(j = 0; j < 16; j++) { pic= remove_long(h, j); - if(pic) unreference_pic(h, pic); + if(pic) unreference_pic(h, pic, 0); } break; default: assert(0); } } - if(!current_is_long){ + if (!current_ref_assigned && FIELD_PICTURE && + !s->first_field && s->current_picture_ptr->reference) { + + /* Second field of complementary field pair; the first field of + * which is already referenced. If short referenced, it + * should be first entry in short_ref. If not, it must exist + * in long_ref; trying to put it on the short list here is an + * error in the encoded bit stream (ref: 7.4.3, NOTE 2 and 3). + */ + if (h->short_ref_count && h->short_ref[0] == s->current_picture_ptr) { + /* Just mark the second field valid */ + s->current_picture_ptr->reference = PICT_FRAME; + } else if (s->current_picture_ptr->long_ref) { + av_log(h->s.avctx, AV_LOG_ERROR, "illegal short term reference " + "assignment for second field " + "in complementary field pair " + "(first field is long term)\n"); + } else { + /* + * First field in reference, but not in any sensible place on our + * reference lists. This shouldn't happen unless reference + * handling somewhere else is wrong. + */ + assert(0); + } + current_ref_assigned = 1; + } + + if(!current_ref_assigned){ pic= remove_short(h, s->current_picture_ptr->frame_num); if(pic){ - unreference_pic(h, pic); + unreference_pic(h, pic, 0); av_log(h->s.avctx, AV_LOG_ERROR, "illegal short term buffer state detected\n"); } @@ -4344,6 +3579,32 @@ static int execute_ref_pic_marking(H264Context *h, MMCO *mmco, int mmco_count){ h->short_ref[0]= s->current_picture_ptr; h->short_ref[0]->long_ref=0; h->short_ref_count++; + s->current_picture_ptr->reference |= s->picture_structure; + } + + if (h->long_ref_count + h->short_ref_count > h->sps.ref_frame_count){ + + /* We have too many reference frames, probably due to corrupted + * stream. Need to discard one frame. Prevents overrun of the + * short_ref and long_ref buffers. + */ + av_log(h->s.avctx, AV_LOG_ERROR, + "number of reference frames exceeds max (probably " + "corrupt input), discarding one\n"); + + if (h->long_ref_count) { + for (i = 0; i < 16; ++i) + if (h->long_ref[i]) + break; + + assert(i < 16); + pic = h->long_ref[i]; + remove_long_at_index(h, i); + } else { + pic = h->short_ref[h->short_ref_count - 1]; + remove_short_at_index(h, h->short_ref_count - 1); + } + unreference_pic(h, pic, 0); } print_short_term(h); @@ -4351,39 +3612,39 @@ static int execute_ref_pic_marking(H264Context *h, MMCO *mmco, int mmco_count){ return 0; } -static int decode_ref_pic_marking(H264Context *h){ +static int decode_ref_pic_marking(H264Context *h, GetBitContext *gb){ MpegEncContext * const s = &h->s; int i; if(h->nal_unit_type == NAL_IDR_SLICE){ //FIXME fields - s->broken_link= get_bits1(&s->gb) -1; - h->mmco[0].long_index= get_bits1(&s->gb) - 1; // current_long_term_idx - if(h->mmco[0].long_index == -1) + s->broken_link= get_bits1(gb) -1; + h->mmco[0].long_arg= get_bits1(gb) - 1; // current_long_term_idx + if(h->mmco[0].long_arg == -1) h->mmco_index= 0; else{ h->mmco[0].opcode= MMCO_LONG; h->mmco_index= 1; } }else{ - if(get_bits1(&s->gb)){ // adaptive_ref_pic_marking_mode_flag + if(get_bits1(gb)){ // adaptive_ref_pic_marking_mode_flag for(i= 0; i<MAX_MMCO_COUNT; i++) { - MMCOOpcode opcode= get_ue_golomb(&s->gb);; + MMCOOpcode opcode= get_ue_golomb(gb); h->mmco[i].opcode= opcode; if(opcode==MMCO_SHORT2UNUSED || opcode==MMCO_SHORT2LONG){ - h->mmco[i].short_frame_num= (h->frame_num - get_ue_golomb(&s->gb) - 1) & ((1<<h->sps.log2_max_frame_num)-1); //FIXME fields -/* if(h->mmco[i].short_frame_num >= h->short_ref_count || h->short_ref[ h->mmco[i].short_frame_num ] == NULL){ + h->mmco[i].short_pic_num= (h->curr_pic_num - get_ue_golomb(gb) - 1) & (h->max_pic_num - 1); +/* if(h->mmco[i].short_pic_num >= h->short_ref_count || h->short_ref[ h->mmco[i].short_pic_num ] == NULL){ av_log(s->avctx, AV_LOG_ERROR, "illegal short ref in memory management control operation %d\n", mmco); return -1; }*/ } if(opcode==MMCO_SHORT2LONG || opcode==MMCO_LONG2UNUSED || opcode==MMCO_LONG || opcode==MMCO_SET_MAX_LONG){ - unsigned int long_index= get_ue_golomb(&s->gb); - if(/*h->mmco[i].long_index >= h->long_ref_count || h->long_ref[ h->mmco[i].long_index ] == NULL*/ long_index >= 16){ + unsigned int long_arg= get_ue_golomb(gb); + if(long_arg >= 32 || (long_arg >= 16 && !(opcode == MMCO_LONG2UNUSED && FIELD_PICTURE))){ av_log(h->s.avctx, AV_LOG_ERROR, "illegal long ref in memory management control operation %d\n", opcode); return -1; } - h->mmco[i].long_index= long_index; + h->mmco[i].long_arg= long_arg; } if(opcode > (unsigned)MMCO_LONG){ @@ -4397,10 +3658,17 @@ static int decode_ref_pic_marking(H264Context *h){ }else{ assert(h->long_ref_count + h->short_ref_count <= h->sps.ref_frame_count); - if(h->long_ref_count + h->short_ref_count == h->sps.ref_frame_count){ //FIXME fields + if(h->short_ref_count && h->long_ref_count + h->short_ref_count == h->sps.ref_frame_count && + !(FIELD_PICTURE && !s->first_field && s->current_picture_ptr->reference)) { h->mmco[0].opcode= MMCO_SHORT2UNUSED; - h->mmco[0].short_frame_num= h->short_ref[ h->short_ref_count - 1 ]->frame_num; + h->mmco[0].short_pic_num= h->short_ref[ h->short_ref_count - 1 ]->frame_num; h->mmco_index= 1; + if (FIELD_PICTURE) { + h->mmco[0].short_pic_num *= 2; + h->mmco[1].opcode= MMCO_SHORT2UNUSED; + h->mmco[1].short_pic_num= h->mmco[0].short_pic_num + 1; + h->mmco_index= 2; + } }else h->mmco_index= 0; } @@ -4488,37 +3756,135 @@ static int init_poc(H264Context *h){ field_poc[1]= poc; } - if(s->picture_structure != PICT_BOTTOM_FIELD) + if(s->picture_structure != PICT_BOTTOM_FIELD) { s->current_picture_ptr->field_poc[0]= field_poc[0]; - if(s->picture_structure != PICT_TOP_FIELD) + s->current_picture_ptr->poc = field_poc[0]; + } + if(s->picture_structure != PICT_TOP_FIELD) { s->current_picture_ptr->field_poc[1]= field_poc[1]; - if(s->picture_structure == PICT_FRAME) // FIXME field pix? - s->current_picture_ptr->poc= FFMIN(field_poc[0], field_poc[1]); + s->current_picture_ptr->poc = field_poc[1]; + } + if(!FIELD_PICTURE || !s->first_field) { + Picture *cur = s->current_picture_ptr; + cur->poc= FFMIN(cur->field_poc[0], cur->field_poc[1]); + } return 0; } + +/** + * initialize scan tables + */ +static void init_scan_tables(H264Context *h){ + MpegEncContext * const s = &h->s; + int i; + if(s->dsp.h264_idct_add == ff_h264_idct_add_c){ //FIXME little ugly + memcpy(h->zigzag_scan, zigzag_scan, 16*sizeof(uint8_t)); + memcpy(h-> field_scan, field_scan, 16*sizeof(uint8_t)); + }else{ + for(i=0; i<16; i++){ +#define T(x) (x>>2) | ((x<<2) & 0xF) + h->zigzag_scan[i] = T(zigzag_scan[i]); + h-> field_scan[i] = T( field_scan[i]); +#undef T + } + } + if(s->dsp.h264_idct8_add == ff_h264_idct8_add_c){ + memcpy(h->zigzag_scan8x8, zigzag_scan8x8, 64*sizeof(uint8_t)); + memcpy(h->zigzag_scan8x8_cavlc, zigzag_scan8x8_cavlc, 64*sizeof(uint8_t)); + memcpy(h->field_scan8x8, field_scan8x8, 64*sizeof(uint8_t)); + memcpy(h->field_scan8x8_cavlc, field_scan8x8_cavlc, 64*sizeof(uint8_t)); + }else{ + for(i=0; i<64; i++){ +#define T(x) (x>>3) | ((x&7)<<3) + h->zigzag_scan8x8[i] = T(zigzag_scan8x8[i]); + h->zigzag_scan8x8_cavlc[i] = T(zigzag_scan8x8_cavlc[i]); + h->field_scan8x8[i] = T(field_scan8x8[i]); + h->field_scan8x8_cavlc[i] = T(field_scan8x8_cavlc[i]); +#undef T + } + } + if(h->sps.transform_bypass){ //FIXME same ugly + h->zigzag_scan_q0 = zigzag_scan; + h->zigzag_scan8x8_q0 = zigzag_scan8x8; + h->zigzag_scan8x8_cavlc_q0 = zigzag_scan8x8_cavlc; + h->field_scan_q0 = field_scan; + h->field_scan8x8_q0 = field_scan8x8; + h->field_scan8x8_cavlc_q0 = field_scan8x8_cavlc; + }else{ + h->zigzag_scan_q0 = h->zigzag_scan; + h->zigzag_scan8x8_q0 = h->zigzag_scan8x8; + h->zigzag_scan8x8_cavlc_q0 = h->zigzag_scan8x8_cavlc; + h->field_scan_q0 = h->field_scan; + h->field_scan8x8_q0 = h->field_scan8x8; + h->field_scan8x8_cavlc_q0 = h->field_scan8x8_cavlc; + } +} + +/** + * Replicates H264 "master" context to thread contexts. + */ +static void clone_slice(H264Context *dst, H264Context *src) +{ + memcpy(dst->block_offset, src->block_offset, sizeof(dst->block_offset)); + dst->s.current_picture_ptr = src->s.current_picture_ptr; + dst->s.current_picture = src->s.current_picture; + dst->s.linesize = src->s.linesize; + dst->s.uvlinesize = src->s.uvlinesize; + dst->s.first_field = src->s.first_field; + + dst->prev_poc_msb = src->prev_poc_msb; + dst->prev_poc_lsb = src->prev_poc_lsb; + dst->prev_frame_num_offset = src->prev_frame_num_offset; + dst->prev_frame_num = src->prev_frame_num; + dst->short_ref_count = src->short_ref_count; + + memcpy(dst->short_ref, src->short_ref, sizeof(dst->short_ref)); + memcpy(dst->long_ref, src->long_ref, sizeof(dst->long_ref)); + memcpy(dst->default_ref_list, src->default_ref_list, sizeof(dst->default_ref_list)); + memcpy(dst->ref_list, src->ref_list, sizeof(dst->ref_list)); + + memcpy(dst->dequant4_coeff, src->dequant4_coeff, sizeof(src->dequant4_coeff)); + memcpy(dst->dequant8_coeff, src->dequant8_coeff, sizeof(src->dequant8_coeff)); +} + /** * decodes a slice header. * this will allso call MPV_common_init() and frame_start() as needed + * + * @param h h264context + * @param h0 h264 master context (differs from 'h' when doing sliced based parallel decoding) + * + * @return 0 if okay, <0 if an error occured, 1 if decoding must not be multithreaded */ -static int decode_slice_header(H264Context *h){ +static int decode_slice_header(H264Context *h, H264Context *h0){ MpegEncContext * const s = &h->s; + MpegEncContext * const s0 = &h0->s; unsigned int first_mb_in_slice; unsigned int pps_id; int num_ref_idx_active_override_flag; static const uint8_t slice_type_map[5]= {P_TYPE, B_TYPE, I_TYPE, SP_TYPE, SI_TYPE}; - unsigned int slice_type, tmp; + unsigned int slice_type, tmp, i; int default_ref_list_done = 0; + int last_pic_structure; - s->current_picture.reference= h->nal_ref_idc != 0; s->dropable= h->nal_ref_idc == 0; + if((s->avctx->flags2 & CODEC_FLAG2_FAST) && !h->nal_ref_idc){ + s->me.qpel_put= s->dsp.put_2tap_qpel_pixels_tab; + s->me.qpel_avg= s->dsp.avg_2tap_qpel_pixels_tab; + }else{ + s->me.qpel_put= s->dsp.put_h264_qpel_pixels_tab; + s->me.qpel_avg= s->dsp.avg_h264_qpel_pixels_tab; + } + first_mb_in_slice= get_ue_golomb(&s->gb); if((s->flags2 & CODEC_FLAG2_CHUNKS) && first_mb_in_slice == 0){ - h->slice_num = 0; - s->current_picture_ptr= NULL; + h0->current_slice = 0; + if (!s0->first_field) + s->current_picture_ptr= NULL; } slice_type= get_ue_golomb(&s->gb); @@ -4534,31 +3900,36 @@ static int decode_slice_header(H264Context *h){ slice_type= slice_type_map[ slice_type ]; if (slice_type == I_TYPE - || (h->slice_num != 0 && slice_type == h->slice_type) ) { + || (h0->current_slice != 0 && slice_type == h0->last_slice_type) ) { default_ref_list_done = 1; } h->slice_type= slice_type; s->pict_type= h->slice_type; // to make a few old func happy, it's wrong though + if (s->pict_type == B_TYPE && s0->last_picture_ptr == NULL) { + av_log(h->s.avctx, AV_LOG_ERROR, + "B picture before any references, skipping\n"); + return -1; + } pps_id= get_ue_golomb(&s->gb); if(pps_id>=MAX_PPS_COUNT){ av_log(h->s.avctx, AV_LOG_ERROR, "pps_id out of range\n"); return -1; } - h->pps= h->pps_buffer[pps_id]; - if(h->pps.slice_group_count == 0){ + if(!h0->pps_buffers[pps_id]) { av_log(h->s.avctx, AV_LOG_ERROR, "non existing PPS referenced\n"); return -1; } + h->pps= *h0->pps_buffers[pps_id]; - h->sps= h->sps_buffer[ h->pps.sps_id ]; - if(h->sps.log2_max_frame_num == 0){ + if(!h0->sps_buffers[h->pps.sps_id]) { av_log(h->s.avctx, AV_LOG_ERROR, "non existing SPS referenced\n"); return -1; } + h->sps = *h0->sps_buffers[h->pps.sps_id]; - if(h->dequant_coeff_pps != pps_id){ + if(h == h0 && h->dequant_coeff_pps != pps_id){ h->dequant_coeff_pps = pps_id; init_dequant_tables(h); } @@ -4577,58 +3948,35 @@ static int decode_slice_header(H264Context *h){ if (s->context_initialized && ( s->width != s->avctx->width || s->height != s->avctx->height)) { + if(h != h0) + return -1; // width / height changed during parallelized decoding free_tables(h); MPV_common_end(s); } if (!s->context_initialized) { + if(h != h0) + return -1; // we cant (re-)initialize context during parallel decoding if (MPV_common_init(s) < 0) return -1; + s->first_field = 0; - if(s->dsp.h264_idct_add == ff_h264_idct_add_c){ //FIXME little ugly - memcpy(h->zigzag_scan, zigzag_scan, 16*sizeof(uint8_t)); - memcpy(h-> field_scan, field_scan, 16*sizeof(uint8_t)); - }else{ - int i; - for(i=0; i<16; i++){ -#define T(x) (x>>2) | ((x<<2) & 0xF) - h->zigzag_scan[i] = T(zigzag_scan[i]); - h-> field_scan[i] = T( field_scan[i]); -#undef T - } - } - if(s->dsp.h264_idct8_add == ff_h264_idct8_add_c){ - memcpy(h->zigzag_scan8x8, zigzag_scan8x8, 64*sizeof(uint8_t)); - memcpy(h->zigzag_scan8x8_cavlc, zigzag_scan8x8_cavlc, 64*sizeof(uint8_t)); - memcpy(h->field_scan8x8, field_scan8x8, 64*sizeof(uint8_t)); - memcpy(h->field_scan8x8_cavlc, field_scan8x8_cavlc, 64*sizeof(uint8_t)); - }else{ - int i; - for(i=0; i<64; i++){ -#define T(x) (x>>3) | ((x&7)<<3) - h->zigzag_scan8x8[i] = T(zigzag_scan8x8[i]); - h->zigzag_scan8x8_cavlc[i] = T(zigzag_scan8x8_cavlc[i]); - h->field_scan8x8[i] = T(field_scan8x8[i]); - h->field_scan8x8_cavlc[i] = T(field_scan8x8_cavlc[i]); -#undef T - } - } - if(h->sps.transform_bypass){ //FIXME same ugly - h->zigzag_scan_q0 = zigzag_scan; - h->zigzag_scan8x8_q0 = zigzag_scan8x8; - h->zigzag_scan8x8_cavlc_q0 = zigzag_scan8x8_cavlc; - h->field_scan_q0 = field_scan; - h->field_scan8x8_q0 = field_scan8x8; - h->field_scan8x8_cavlc_q0 = field_scan8x8_cavlc; - }else{ - h->zigzag_scan_q0 = h->zigzag_scan; - h->zigzag_scan8x8_q0 = h->zigzag_scan8x8; - h->zigzag_scan8x8_cavlc_q0 = h->zigzag_scan8x8_cavlc; - h->field_scan_q0 = h->field_scan; - h->field_scan8x8_q0 = h->field_scan8x8; - h->field_scan8x8_cavlc_q0 = h->field_scan8x8_cavlc; + init_scan_tables(h); + alloc_tables(h); + + for(i = 1; i < s->avctx->thread_count; i++) { + H264Context *c; + c = h->thread_context[i] = av_malloc(sizeof(H264Context)); + memcpy(c, h->s.thread_context[i], sizeof(MpegEncContext)); + memset(&c->s + 1, 0, sizeof(H264Context) - sizeof(MpegEncContext)); + c->sps = h->sps; + c->pps = h->pps; + init_scan_tables(c); + clone_tables(c, h); } - alloc_tables(h); + for(i = 0; i < s->avctx->thread_count; i++) + if(context_init(h->thread_context[i]) < 0) + return -1; s->avctx->width = s->width; s->avctx->height = s->height; @@ -4645,42 +3993,90 @@ static int decode_slice_header(H264Context *h){ } } - if(h->slice_num == 0){ - if(frame_start(h) < 0) - return -1; - } - - s->current_picture_ptr->frame_num= //FIXME frame_num cleanup h->frame_num= get_bits(&s->gb, h->sps.log2_max_frame_num); h->mb_mbaff = 0; h->mb_aff_frame = 0; + last_pic_structure = s0->picture_structure; if(h->sps.frame_mbs_only_flag){ s->picture_structure= PICT_FRAME; }else{ if(get_bits1(&s->gb)) { //field_pic_flag s->picture_structure= PICT_TOP_FIELD + get_bits1(&s->gb); //bottom_field_flag - av_log(h->s.avctx, AV_LOG_ERROR, "PAFF interlacing is not implemented\n"); } else { s->picture_structure= PICT_FRAME; h->mb_aff_frame = h->sps.mb_aff; } } + + if(h0->current_slice == 0){ + /* See if we have a decoded first field looking for a pair... */ + if (s0->first_field) { + assert(s0->current_picture_ptr); + assert(s0->current_picture_ptr->data[0]); + assert(s0->current_picture_ptr->reference != DELAYED_PIC_REF); + + /* figure out if we have a complementary field pair */ + if (!FIELD_PICTURE || s->picture_structure == last_pic_structure) { + /* + * Previous field is unmatched. Don't display it, but let it + * remain for reference if marked as such. + */ + s0->current_picture_ptr = NULL; + s0->first_field = FIELD_PICTURE; + + } else { + if (h->nal_ref_idc && + s0->current_picture_ptr->reference && + s0->current_picture_ptr->frame_num != h->frame_num) { + /* + * This and previous field were reference, but had + * different frame_nums. Consider this field first in + * pair. Throw away previous field except for reference + * purposes. + */ + s0->first_field = 1; + s0->current_picture_ptr = NULL; + + } else { + /* Second field in complementary pair */ + s0->first_field = 0; + } + } + + } else { + /* Frame or first field in a potentially complementary pair */ + assert(!s0->current_picture_ptr); + s0->first_field = FIELD_PICTURE; + } + + if((!FIELD_PICTURE || s0->first_field) && frame_start(h) < 0) { + s0->first_field = 0; + return -1; + } + } + if(h != h0) + clone_slice(h, h0); + + s->current_picture_ptr->frame_num= h->frame_num; //FIXME frame_num cleanup + assert(s->mb_num == s->mb_width * s->mb_height); - if(first_mb_in_slice << h->mb_aff_frame >= s->mb_num || + if(first_mb_in_slice << FIELD_OR_MBAFF_PICTURE >= s->mb_num || first_mb_in_slice >= s->mb_num){ av_log(h->s.avctx, AV_LOG_ERROR, "first_mb_in_slice overflow\n"); return -1; } s->resync_mb_x = s->mb_x = first_mb_in_slice % s->mb_width; - s->resync_mb_y = s->mb_y = (first_mb_in_slice / s->mb_width) << h->mb_aff_frame; + s->resync_mb_y = s->mb_y = (first_mb_in_slice / s->mb_width) << FIELD_OR_MBAFF_PICTURE; + if (s->picture_structure == PICT_BOTTOM_FIELD) + s->resync_mb_y = s->mb_y = s->mb_y + 1; assert(s->mb_y < s->mb_height); if(s->picture_structure==PICT_FRAME){ h->curr_pic_num= h->frame_num; h->max_pic_num= 1<< h->sps.log2_max_frame_num; }else{ - h->curr_pic_num= 2*h->frame_num; + h->curr_pic_num= 2*h->frame_num + 1; h->max_pic_num= 1<<(h->sps.log2_max_frame_num + 1); } @@ -4716,8 +4112,6 @@ static int decode_slice_header(H264Context *h){ if(h->slice_type == P_TYPE || h->slice_type == SP_TYPE || h->slice_type == B_TYPE){ if(h->slice_type == B_TYPE){ h->direct_spatial_mv_pred= get_bits1(&s->gb); - if(h->sps.mb_aff && h->direct_spatial_mv_pred) - av_log(h->s.avctx, AV_LOG_ERROR, "MBAFF + spatial direct mode is not implemented\n"); } num_ref_idx_active_override_flag= get_bits1(&s->gb); @@ -4754,8 +4148,8 @@ static int decode_slice_header(H264Context *h){ else h->use_weight = 0; - if(s->current_picture.reference) - decode_ref_pic_marking(h); + if(h->nal_ref_idc) + decode_ref_pic_marking(h0, &s->gb); if(FRAME_MBAFF) fill_mbaff_ref_list(h); @@ -4776,7 +4170,8 @@ static int decode_slice_header(H264Context *h){ return -1; } s->qscale= tmp; - h->chroma_qp = get_chroma_qp(h->pps.chroma_qp_index_offset, s->qscale); + h->chroma_qp[0] = get_chroma_qp(h, 0, s->qscale); + h->chroma_qp[1] = get_chroma_qp(h, 1, s->qscale); //FIXME qscale / qp ... stuff if(h->slice_type == SP_TYPE){ get_bits1(&s->gb); /* sp_for_switch_flag */ @@ -4803,21 +4198,39 @@ static int decode_slice_header(H264Context *h){ h->slice_beta_offset = get_se_golomb(&s->gb) << 1; } } + if( s->avctx->skip_loop_filter >= AVDISCARD_ALL ||(s->avctx->skip_loop_filter >= AVDISCARD_NONKEY && h->slice_type != I_TYPE) ||(s->avctx->skip_loop_filter >= AVDISCARD_BIDIR && h->slice_type == B_TYPE) ||(s->avctx->skip_loop_filter >= AVDISCARD_NONREF && h->nal_ref_idc == 0)) h->deblocking_filter= 0; + if(h->deblocking_filter == 1 && h0->max_contexts > 1) { + if(s->avctx->flags2 & CODEC_FLAG2_FAST) { + /* Cheat slightly for speed: + Do not bother to deblock across slices. */ + h->deblocking_filter = 2; + } else { + h0->max_contexts = 1; + if(!h0->single_decode_warning) { + av_log(s->avctx, AV_LOG_INFO, "Cannot parallelize deblocking type 1, decoding such frames in sequential order\n"); + h0->single_decode_warning = 1; + } + if(h != h0) + return 1; // deblocking switched inside frame + } + } + #if 0 //FMO if( h->pps.num_slice_groups > 1 && h->pps.mb_slice_group_map_type >= 3 && h->pps.mb_slice_group_map_type <= 5) slice_group_change_cycle= get_bits(&s->gb, ?); #endif - h->slice_num++; + h0->last_slice_type = slice_type; + h->slice_num = ++h0->current_slice; h->emu_edge_width= (s->flags&CODEC_FLAG_EMU_EDGE) ? 0 : 16; - h->emu_edge_height= FRAME_MBAFF ? 0 : h->emu_edge_width; + h->emu_edge_height= (FRAME_MBAFF || FIELD_PICTURE) ? 0 : h->emu_edge_width; if(s->avctx->debug&FF_DEBUG_PICT_INFO){ av_log(h->s.avctx, AV_LOG_DEBUG, "slice:%d %s mb:%d %c pps:%u frame:%d poc:%d/%d ref:%d/%d qp:%d loop:%d:%d:%d weight:%d%s\n", @@ -4835,14 +4248,6 @@ static int decode_slice_header(H264Context *h){ ); } - if((s->avctx->flags2 & CODEC_FLAG2_FAST) && !s->current_picture.reference){ - s->me.qpel_put= s->dsp.put_2tap_qpel_pixels_tab; - s->me.qpel_avg= s->dsp.avg_2tap_qpel_pixels_tab; - }else{ - s->me.qpel_put= s->dsp.put_h264_qpel_pixels_tab; - s->me.qpel_avg= s->dsp.avg_h264_qpel_pixels_tab; - } - return 0; } @@ -5161,7 +4566,7 @@ decode_intra_mb: if(IS_INTRA_PCM(mb_type)){ unsigned int x, y; - // we assume these blocks are very rare so we dont optimize it + // We assume these blocks are very rare so we do not optimize it. align_get_bits(&s->gb); // The pixels are stored in the same order as levels in h->mb array. @@ -5189,7 +4594,8 @@ decode_intra_mb: // In deblocking, the quantizer is 0 s->current_picture.qscale_table[mb_xy]= 0; - h->chroma_qp = get_chroma_qp(h->pps.chroma_qp_index_offset, 0); + h->chroma_qp[0] = get_chroma_qp(h, 0, 0); + h->chroma_qp[1] = get_chroma_qp(h, 1, 0); // All coeffs are present memset(h->non_zero_count[mb_xy], 16, 16); @@ -5299,8 +4705,6 @@ decode_intra_mb: dct8x8_allowed = get_dct8x8_allowed(h); for(list=0; list<h->list_count; list++){ - const int ref_count= IS_REF0(mb_type) ? 1 : h->ref_count[list]; - for(i=0; i<4; i++){ if(IS_DIRECT(h->sub_mb_type[i])) { h->ref_cache[list][ scan8[4*i] ] = h->ref_cache[list][ scan8[4*i]+1 ]; @@ -5465,7 +4869,7 @@ decode_intra_mb: if(cbp || IS_INTRA16x16(mb_type)){ int i8x8, i4x4, chroma_idx; - int chroma_qp, dquant; + int dquant; GetBitContext *gb= IS_INTRA(mb_type) ? h->intra_gb_ptr : h->inter_gb_ptr; const uint8_t *scan, *scan8x8, *dc_scan; @@ -5494,7 +4898,8 @@ decode_intra_mb: else s->qscale-= 52; } - h->chroma_qp= chroma_qp= get_chroma_qp(h->pps.chroma_qp_index_offset, s->qscale); + h->chroma_qp[0]= get_chroma_qp(h, 0, s->qscale); + h->chroma_qp[1]= get_chroma_qp(h, 1, s->qscale); if(IS_INTRA16x16(mb_type)){ if( decode_residual(h, h->intra_gb_ptr, h->mb, LUMA_DC_BLOCK_INDEX, dc_scan, h->dequant4_coeff[0][s->qscale], 16) < 0){ return -1; //FIXME continue if partitioned and other return -1 too @@ -5552,9 +4957,10 @@ decode_intra_mb: if(cbp&0x20){ for(chroma_idx=0; chroma_idx<2; chroma_idx++){ + const uint32_t *qmul = h->dequant4_coeff[chroma_idx+1+(IS_INTRA( mb_type ) ? 0:3)][h->chroma_qp[chroma_idx]]; for(i4x4=0; i4x4<4; i4x4++){ const int index= 16 + 4*chroma_idx + i4x4; - if( decode_residual(h, gb, h->mb + 16*index, index, scan + 1, h->dequant4_coeff[chroma_idx+1+(IS_INTRA( mb_type ) ? 0:3)][chroma_qp], 15) < 0){ + if( decode_residual(h, gb, h->mb + 16*index, index, scan + 1, qmul, 15) < 0){ return -1; } } @@ -5713,7 +5119,7 @@ static int decode_cabac_mb_skip( H264Context *h, int mb_x, int mb_y ) { }else{ int mb_xy = mb_x + mb_y*s->mb_stride; mba_xy = mb_xy - 1; - mbb_xy = mb_xy - s->mb_stride; + mbb_xy = mb_xy - (s->mb_stride << FIELD_PICTURE); } if( h->slice_table[mba_xy] == h->slice_num && !IS_SKIP( s->current_picture.mb_type[mba_xy] )) @@ -5766,65 +5172,20 @@ static int decode_cabac_mb_chroma_pre_mode( H264Context *h) { return 3; } -static const uint8_t block_idx_x[16] = { - 0, 1, 0, 1, 2, 3, 2, 3, 0, 1, 0, 1, 2, 3, 2, 3 -}; -static const uint8_t block_idx_y[16] = { - 0, 0, 1, 1, 0, 0, 1, 1, 2, 2, 3, 3, 2, 2, 3, 3 -}; -static const uint8_t block_idx_xy[4][4] = { - { 0, 2, 8, 10}, - { 1, 3, 9, 11}, - { 4, 6, 12, 14}, - { 5, 7, 13, 15} -}; - static int decode_cabac_mb_cbp_luma( H264Context *h) { - int cbp = 0; - int cbp_b = -1; - int i8x8; - - if( h->slice_table[h->top_mb_xy] == h->slice_num ) { - cbp_b = h->top_cbp; - tprintf(h->s.avctx, "cbp_b = top_cbp = %x\n", cbp_b); - } - - for( i8x8 = 0; i8x8 < 4; i8x8++ ) { - int cbp_a = -1; - int x, y; - int ctx = 0; - - x = block_idx_x[4*i8x8]; - y = block_idx_y[4*i8x8]; - - if( x > 0 ) - cbp_a = cbp; - else if( h->slice_table[h->left_mb_xy[0]] == h->slice_num ) { - cbp_a = h->left_cbp; - tprintf(h->s.avctx, "cbp_a = left_cbp = %x\n", cbp_a); - } - - if( y > 0 ) - cbp_b = cbp; - - /* No need to test for skip as we put 0 for skip block */ - /* No need to test for IPCM as we put 1 for IPCM block */ - if( cbp_a >= 0 ) { - int i8x8a = block_idx_xy[(x-1)&0x03][y]/4; - if( ((cbp_a >> i8x8a)&0x01) == 0 ) - ctx++; - } - - if( cbp_b >= 0 ) { - int i8x8b = block_idx_xy[x][(y-1)&0x03]/4; - if( ((cbp_b >> i8x8b)&0x01) == 0 ) - ctx += 2; - } - - if( get_cabac( &h->cabac, &h->cabac_state[73 + ctx] ) ) { - cbp |= 1 << i8x8; - } - } + int cbp_b, cbp_a, ctx, cbp = 0; + + cbp_a = h->slice_table[h->left_mb_xy[0]] == h->slice_num ? h->left_cbp : -1; + cbp_b = h->slice_table[h->top_mb_xy] == h->slice_num ? h->top_cbp : -1; + + ctx = !(cbp_a & 0x02) + 2 * !(cbp_b & 0x04); + cbp |= get_cabac_noinline(&h->cabac, &h->cabac_state[73 + ctx]); + ctx = !(cbp & 0x01) + 2 * !(cbp_b & 0x08); + cbp |= get_cabac_noinline(&h->cabac, &h->cabac_state[73 + ctx]) << 1; + ctx = !(cbp_a & 0x08) + 2 * !(cbp & 0x01); + cbp |= get_cabac_noinline(&h->cabac, &h->cabac_state[73 + ctx]) << 2; + ctx = !(cbp & 0x04) + 2 * !(cbp & 0x02); + cbp |= get_cabac_noinline(&h->cabac, &h->cabac_state[73 + ctx]) << 3; return cbp; } static int decode_cabac_mb_cbp_chroma( H264Context *h) { @@ -5846,16 +5207,9 @@ static int decode_cabac_mb_cbp_chroma( H264Context *h) { return 1 + get_cabac_noinline( &h->cabac, &h->cabac_state[77 + ctx] ); } static int decode_cabac_mb_dqp( H264Context *h) { - MpegEncContext * const s = &h->s; - int mbn_xy; int ctx = 0; int val = 0; - if( s->mb_x > 0 ) - mbn_xy = s->mb_x + s->mb_y*s->mb_stride - 1; - else - mbn_xy = s->mb_width - 1 + (s->mb_y-1)*s->mb_stride; - if( h->last_qscale_diff != 0 ) ctx++; @@ -5978,7 +5332,7 @@ static int decode_cabac_mb_mvd( H264Context *h, int list, int n, int l ) { return get_cabac_bypass_sign( &h->cabac, -mvd ); } -static int inline get_cabac_cbf_ctx( H264Context *h, int cat, int idx ) { +static inline int get_cabac_cbf_ctx( H264Context *h, int cat, int idx ) { int nza, nzb; int ctx = 0; @@ -6006,14 +5360,14 @@ static int inline get_cabac_cbf_ctx( H264Context *h, int cat, int idx ) { return ctx + 4 * cat; } -static const __attribute((used)) uint8_t last_coeff_flag_offset_8x8[63] = { +DECLARE_ASM_CONST(1, const uint8_t, last_coeff_flag_offset_8x8[63]) = { 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 6, 6, 6, 6, 7, 7, 7, 7, 8, 8, 8 }; -static int decode_cabac_residual( H264Context *h, DCTELEM *block, int cat, int n, const uint8_t *scantable, const uint32_t *qmul, int max_coeff) { +static void decode_cabac_residual( H264Context *h, DCTELEM *block, int cat, int n, const uint8_t *scantable, const uint32_t *qmul, int max_coeff) { const int mb_xy = h->s.mb_x + h->s.mb_y*h->s.mb_stride; static const int significant_coeff_flag_offset[2][6] = { { 105+0, 105+15, 105+29, 105+44, 105+47, 402 }, @@ -6039,7 +5393,7 @@ static int decode_cabac_residual( H264Context *h, DCTELEM *block, int cat, int n int index[64]; - int last; + int av_unused last; int coeff_count = 0; int abslevel1 = 1; @@ -6083,7 +5437,7 @@ static int decode_cabac_residual( H264Context *h, DCTELEM *block, int cat, int n h->cabac.low = cc.low ; h->cabac.bytestream= cc.bytestream; #endif - return 0; + return; } } @@ -6111,7 +5465,7 @@ static int decode_cabac_residual( H264Context *h, DCTELEM *block, int cat, int n index[coeff_count++] = last;\ } const uint8_t *sig_off = significant_coeff_flag_offset_8x8[MB_FIELD]; -#if defined(ARCH_X86) && defined(CONFIG_7REGS) && defined(CONFIG_EBX_AVAILABLE) && !( defined(ARCH_X86_64) && defined(PIC) ) +#if defined(ARCH_X86) && defined(HAVE_7REGS) && defined(HAVE_EBX_AVAILABLE) && !defined(BROKEN_RELOCATIONS) coeff_count= decode_significance_8x8_x86(CC, significant_coeff_ctx_base, index, sig_off); } else { coeff_count= decode_significance_x86(CC, max_coeff, significant_coeff_ctx_base, index); @@ -6144,7 +5498,7 @@ static int decode_cabac_residual( H264Context *h, DCTELEM *block, int cat, int n if( !qmul ) { block[j] = get_cabac_bypass_sign( CC, -1); }else{ - block[j] = (get_cabac_bypass_sign( CC, -qmul[j]) + 32) >> 6;; + block[j] = (get_cabac_bypass_sign( CC, -qmul[j]) + 32) >> 6; } abslevel1++; @@ -6184,10 +5538,10 @@ static int decode_cabac_residual( H264Context *h, DCTELEM *block, int cat, int n h->cabac.low = cc.low ; h->cabac.bytestream= cc.bytestream; #endif - return 0; + } -static void inline compute_mb_neighbors(H264Context *h) +static inline void compute_mb_neighbors(H264Context *h) { MpegEncContext * const s = &h->s; const int mb_xy = s->mb_x + s->mb_y*s->mb_stride; @@ -6209,6 +5563,8 @@ static void inline compute_mb_neighbors(H264Context *h) if (left_mb_frame_flag != curr_mb_frame_flag) { h->left_mb_xy[0] = pair_xy - 1; } + } else if (FIELD_PICTURE) { + h->top_mb_xy -= s->mb_stride; } return; } @@ -6304,7 +5660,7 @@ decode_intra_mb: const uint8_t *ptr; unsigned int x, y; - // We assume these blocks are very rare so we dont optimize it. + // We assume these blocks are very rare so we do not optimize it. // FIXME The two following lines get the bitstream position in the cabac // decode, I think it should be done by a function in cabac.h (or cabac.c). ptr= h->cabac.bytestream; @@ -6343,7 +5699,8 @@ decode_intra_mb: h->chroma_pred_mode_table[mb_xy] = 0; // In deblocking, the quantizer is 0 s->current_picture.qscale_table[mb_xy]= 0; - h->chroma_qp = get_chroma_qp(h->pps.chroma_qp_index_offset, 0); + h->chroma_qp[0] = get_chroma_qp(h, 0, 0); + h->chroma_qp[1] = get_chroma_qp(h, 1, 0); // All coeffs are present memset(h->non_zero_count[mb_xy], 16, 16); s->current_picture.mb_type[mb_xy]= mb_type; @@ -6399,6 +5756,10 @@ decode_intra_mb: if( IS_DIRECT(h->sub_mb_type[0] | h->sub_mb_type[1] | h->sub_mb_type[2] | h->sub_mb_type[3]) ) { pred_direct_motion(h, &mb_type); + h->ref_cache[0][scan8[4]] = + h->ref_cache[1][scan8[4]] = + h->ref_cache[0][scan8[12]] = + h->ref_cache[1][scan8[12]] = PART_NOT_AVAILABLE; if( h->ref_count[0] > 1 || h->ref_count[1] > 1 ) { for( i = 0; i < 4; i++ ) if( IS_DIRECT(h->sub_mb_type[i]) ) @@ -6434,11 +5795,11 @@ decode_intra_mb: for(list=0; list<h->list_count; list++){ for(i=0; i<4; i++){ + h->ref_cache[list][ scan8[4*i] ]=h->ref_cache[list][ scan8[4*i]+1 ]; if(IS_DIRECT(h->sub_mb_type[i])){ fill_rectangle(h->mvd_cache[list][scan8[4*i]], 2, 2, 8, 0, 4); continue; } - h->ref_cache[list][ scan8[4*i] ]=h->ref_cache[list][ scan8[4*i]+1 ]; if(IS_DIR(h->sub_mb_type[i], 0, list) && !IS_DIRECT(h->sub_mb_type[i])){ const int sub_mb_type= h->sub_mb_type[i]; @@ -6597,6 +5958,7 @@ decode_intra_mb: if( cbp || IS_INTRA16x16( mb_type ) ) { const uint8_t *scan, *scan8x8, *dc_scan; + const uint32_t *qmul; int dqp; if(IS_INTERLACED(mb_type)){ @@ -6619,18 +5981,19 @@ decode_intra_mb: if(s->qscale<0) s->qscale+= 52; else s->qscale-= 52; } - h->chroma_qp = get_chroma_qp(h->pps.chroma_qp_index_offset, s->qscale); + h->chroma_qp[0] = get_chroma_qp(h, 0, s->qscale); + h->chroma_qp[1] = get_chroma_qp(h, 1, s->qscale); if( IS_INTRA16x16( mb_type ) ) { int i; //av_log( s->avctx, AV_LOG_ERROR, "INTRA16x16 DC\n" ); - if( decode_cabac_residual( h, h->mb, 0, 0, dc_scan, NULL, 16) < 0) - return -1; + decode_cabac_residual( h, h->mb, 0, 0, dc_scan, NULL, 16); + if( cbp&15 ) { + qmul = h->dequant4_coeff[0][s->qscale]; for( i = 0; i < 16; i++ ) { //av_log( s->avctx, AV_LOG_ERROR, "INTRA16x16 AC:%d\n", i ); - if( decode_cabac_residual(h, h->mb + 16*i, 1, i, scan + 1, h->dequant4_coeff[0][s->qscale], 15) < 0 ) - return -1; + decode_cabac_residual(h, h->mb + 16*i, 1, i, scan + 1, qmul, 15); } } else { fill_rectangle(&h->non_zero_count_cache[scan8[0]], 4, 4, 8, 0, 1); @@ -6640,17 +6003,17 @@ decode_intra_mb: for( i8x8 = 0; i8x8 < 4; i8x8++ ) { if( cbp & (1<<i8x8) ) { if( IS_8x8DCT(mb_type) ) { - if( decode_cabac_residual(h, h->mb + 64*i8x8, 5, 4*i8x8, - scan8x8, h->dequant8_coeff[IS_INTRA( mb_type ) ? 0:1][s->qscale], 64) < 0 ) - return -1; - } else - for( i4x4 = 0; i4x4 < 4; i4x4++ ) { - const int index = 4*i8x8 + i4x4; - //av_log( s->avctx, AV_LOG_ERROR, "Luma4x4: %d\n", index ); + decode_cabac_residual(h, h->mb + 64*i8x8, 5, 4*i8x8, + scan8x8, h->dequant8_coeff[IS_INTRA( mb_type ) ? 0:1][s->qscale], 64); + } else { + qmul = h->dequant4_coeff[IS_INTRA( mb_type ) ? 0:3][s->qscale]; + for( i4x4 = 0; i4x4 < 4; i4x4++ ) { + const int index = 4*i8x8 + i4x4; + //av_log( s->avctx, AV_LOG_ERROR, "Luma4x4: %d\n", index ); //START_TIMER - if( decode_cabac_residual(h, h->mb + 16*index, 2, index, scan, h->dequant4_coeff[IS_INTRA( mb_type ) ? 0:3][s->qscale], 16) < 0 ) - return -1; + decode_cabac_residual(h, h->mb + 16*index, 2, index, scan, qmul, 16); //STOP_TIMER("decode_residual") + } } } else { uint8_t * const nnz= &h->non_zero_count_cache[ scan8[4*i8x8] ]; @@ -6663,19 +6026,18 @@ decode_intra_mb: int c; for( c = 0; c < 2; c++ ) { //av_log( s->avctx, AV_LOG_ERROR, "INTRA C%d-DC\n",c ); - if( decode_cabac_residual(h, h->mb + 256 + 16*4*c, 3, c, chroma_dc_scan, NULL, 4) < 0) - return -1; + decode_cabac_residual(h, h->mb + 256 + 16*4*c, 3, c, chroma_dc_scan, NULL, 4); } } if( cbp&0x20 ) { int c, i; for( c = 0; c < 2; c++ ) { + qmul = h->dequant4_coeff[c+1+(IS_INTRA( mb_type ) ? 0:3)][h->chroma_qp[c]]; for( i = 0; i < 4; i++ ) { const int index = 16 + 4 * c + i; //av_log( s->avctx, AV_LOG_ERROR, "INTRA C%d-AC %d\n",c, index - 16 ); - if( decode_cabac_residual(h, h->mb + 16*index, 4, index - 16, scan + 1, h->dequant4_coeff[c+1+(IS_INTRA( mb_type ) ? 0:3)][h->chroma_qp], 15) < 0) - return -1; + decode_cabac_residual(h, h->mb + 16*index, 4, index - 16, scan + 1, qmul, 15); } } } else { @@ -7009,23 +6371,27 @@ static void filter_mb_edgech( H264Context *h, uint8_t *pix, int stride, int16_t static void filter_mb_fast( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8_t *img_cb, uint8_t *img_cr, unsigned int linesize, unsigned int uvlinesize) { MpegEncContext * const s = &h->s; + int mb_y_firstrow = s->picture_structure == PICT_BOTTOM_FIELD; int mb_xy, mb_type; int qp, qp0, qp1, qpc, qpc0, qpc1, qp_thresh; - if(mb_x==0 || mb_y==0 || !s->dsp.h264_loop_filter_strength) { + mb_xy = mb_x + mb_y*s->mb_stride; + + if(mb_x==0 || mb_y==mb_y_firstrow || !s->dsp.h264_loop_filter_strength || h->pps.chroma_qp_diff || + (h->deblocking_filter == 2 && (h->slice_table[mb_xy] != h->slice_table[h->top_mb_xy] || + h->slice_table[mb_xy] != h->slice_table[mb_xy - 1]))) { filter_mb(h, mb_x, mb_y, img_y, img_cb, img_cr, linesize, uvlinesize); return; } assert(!FRAME_MBAFF); - mb_xy = mb_x + mb_y*s->mb_stride; mb_type = s->current_picture.mb_type[mb_xy]; qp = s->current_picture.qscale_table[mb_xy]; qp0 = s->current_picture.qscale_table[mb_xy-1]; qp1 = s->current_picture.qscale_table[h->top_mb_xy]; - qpc = get_chroma_qp( h->pps.chroma_qp_index_offset, qp ); - qpc0 = get_chroma_qp( h->pps.chroma_qp_index_offset, qp0 ); - qpc1 = get_chroma_qp( h->pps.chroma_qp_index_offset, qp1 ); + qpc = get_chroma_qp( h, 0, qp ); + qpc0 = get_chroma_qp( h, 0, qp0 ); + qpc1 = get_chroma_qp( h, 0, qp1 ); qp0 = (qp + qp0 + 1) >> 1; qp1 = (qp + qp1 + 1) >> 1; qpc0 = (qpc + qpc0 + 1) >> 1; @@ -7038,17 +6404,18 @@ static void filter_mb_fast( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, if( IS_INTRA(mb_type) ) { int16_t bS4[4] = {4,4,4,4}; int16_t bS3[4] = {3,3,3,3}; + int16_t *bSH = FIELD_PICTURE ? bS3 : bS4; if( IS_8x8DCT(mb_type) ) { filter_mb_edgev( h, &img_y[4*0], linesize, bS4, qp0 ); filter_mb_edgev( h, &img_y[4*2], linesize, bS3, qp ); - filter_mb_edgeh( h, &img_y[4*0*linesize], linesize, bS4, qp1 ); + filter_mb_edgeh( h, &img_y[4*0*linesize], linesize, bSH, qp1 ); filter_mb_edgeh( h, &img_y[4*2*linesize], linesize, bS3, qp ); } else { filter_mb_edgev( h, &img_y[4*0], linesize, bS4, qp0 ); filter_mb_edgev( h, &img_y[4*1], linesize, bS3, qp ); filter_mb_edgev( h, &img_y[4*2], linesize, bS3, qp ); filter_mb_edgev( h, &img_y[4*3], linesize, bS3, qp ); - filter_mb_edgeh( h, &img_y[4*0*linesize], linesize, bS4, qp1 ); + filter_mb_edgeh( h, &img_y[4*0*linesize], linesize, bSH, qp1 ); filter_mb_edgeh( h, &img_y[4*1*linesize], linesize, bS3, qp ); filter_mb_edgeh( h, &img_y[4*2*linesize], linesize, bS3, qp ); filter_mb_edgeh( h, &img_y[4*3*linesize], linesize, bS3, qp ); @@ -7057,9 +6424,9 @@ static void filter_mb_fast( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, filter_mb_edgecv( h, &img_cb[2*2], uvlinesize, bS3, qpc ); filter_mb_edgecv( h, &img_cr[2*0], uvlinesize, bS4, qpc0 ); filter_mb_edgecv( h, &img_cr[2*2], uvlinesize, bS3, qpc ); - filter_mb_edgech( h, &img_cb[2*0*uvlinesize], uvlinesize, bS4, qpc1 ); + filter_mb_edgech( h, &img_cb[2*0*uvlinesize], uvlinesize, bSH, qpc1 ); filter_mb_edgech( h, &img_cb[2*2*uvlinesize], uvlinesize, bS3, qpc ); - filter_mb_edgech( h, &img_cr[2*0*uvlinesize], uvlinesize, bS4, qpc1 ); + filter_mb_edgech( h, &img_cr[2*0*uvlinesize], uvlinesize, bSH, qpc1 ); filter_mb_edgech( h, &img_cr[2*2*uvlinesize], uvlinesize, bS3, qpc ); return; } else { @@ -7083,7 +6450,7 @@ static void filter_mb_fast( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, if( IS_INTRA(s->current_picture.mb_type[mb_xy-1]) ) bSv[0][0] = 0x0004000400040004ULL; if( IS_INTRA(s->current_picture.mb_type[h->top_mb_xy]) ) - bSv[1][0] = 0x0004000400040004ULL; + bSv[1][0] = FIELD_PICTURE ? 0x0003000300030003ULL : 0x0004000400040004ULL; #define FILTER(hv,dir,edge)\ if(bSv[dir][edge]) {\ @@ -7131,7 +6498,7 @@ static void filter_mb( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8 //for sufficiently low qp, filtering wouldn't do anything //this is a conservative estimate: could also check beta_offset and more accurate chroma_qp if(!FRAME_MBAFF){ - int qp_thresh = 15 - h->slice_alpha_c0_offset - FFMAX(0, h->pps.chroma_qp_index_offset); + int qp_thresh = 15 - h->slice_alpha_c0_offset - FFMAX(0, FFMAX(h->pps.chroma_qp_index_offset[0], h->pps.chroma_qp_index_offset[1])); int qp = s->current_picture.qscale_table[mb_xy]; if(qp <= qp_thresh && (mb_x == 0 || ((qp + s->current_picture.qscale_table[mb_xy-1] + 1)>>1) <= qp_thresh) @@ -7154,7 +6521,8 @@ static void filter_mb( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8 const int left_mb_xy[2] = { pair_xy-1, pair_xy-1+s->mb_stride }; int16_t bS[8]; int qp[2]; - int chroma_qp[2]; + int bqp[2]; + int rqp[2]; int mb_qp, mbn0_qp, mbn1_qp; int i; first_vertical_edge_done = 1; @@ -7180,18 +6548,22 @@ static void filter_mb( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8 mbn0_qp = s->current_picture.qscale_table[left_mb_xy[0]]; mbn1_qp = s->current_picture.qscale_table[left_mb_xy[1]]; qp[0] = ( mb_qp + mbn0_qp + 1 ) >> 1; - chroma_qp[0] = ( get_chroma_qp( h->pps.chroma_qp_index_offset, mb_qp ) + - get_chroma_qp( h->pps.chroma_qp_index_offset, mbn0_qp ) + 1 ) >> 1; + bqp[0] = ( get_chroma_qp( h, 0, mb_qp ) + + get_chroma_qp( h, 0, mbn0_qp ) + 1 ) >> 1; + rqp[0] = ( get_chroma_qp( h, 1, mb_qp ) + + get_chroma_qp( h, 1, mbn0_qp ) + 1 ) >> 1; qp[1] = ( mb_qp + mbn1_qp + 1 ) >> 1; - chroma_qp[1] = ( get_chroma_qp( h->pps.chroma_qp_index_offset, mb_qp ) + - get_chroma_qp( h->pps.chroma_qp_index_offset, mbn1_qp ) + 1 ) >> 1; + bqp[1] = ( get_chroma_qp( h, 0, mb_qp ) + + get_chroma_qp( h, 0, mbn1_qp ) + 1 ) >> 1; + rqp[1] = ( get_chroma_qp( h, 1, mb_qp ) + + get_chroma_qp( h, 1, mbn1_qp ) + 1 ) >> 1; /* Filter edge */ - tprintf(s->avctx, "filter mb:%d/%d MBAFF, QPy:%d/%d, QPc:%d/%d ls:%d uvls:%d", mb_x, mb_y, qp[0], qp[1], chroma_qp[0], chroma_qp[1], linesize, uvlinesize); + tprintf(s->avctx, "filter mb:%d/%d MBAFF, QPy:%d/%d, QPb:%d/%d QPr:%d/%d ls:%d uvls:%d", mb_x, mb_y, qp[0], qp[1], bqp[0], bqp[1], rqp[0], rqp[1], linesize, uvlinesize); { int i; for (i = 0; i < 8; i++) tprintf(s->avctx, " bS[%d]:%d", i, bS[i]); tprintf(s->avctx, "\n"); } filter_mb_mbaff_edgev ( h, &img_y [0], linesize, bS, qp ); - filter_mb_mbaff_edgecv( h, &img_cb[0], uvlinesize, bS, chroma_qp ); - filter_mb_mbaff_edgecv( h, &img_cr[0], uvlinesize, bS, chroma_qp ); + filter_mb_mbaff_edgecv( h, &img_cb[0], uvlinesize, bS, bqp ); + filter_mb_mbaff_edgecv( h, &img_cr[0], uvlinesize, bS, rqp ); } /* dir : 0 -> vertical edge, 1 -> horizontal edge */ for( dir = 0; dir < 2; dir++ ) @@ -7229,7 +6601,7 @@ static void filter_mb( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8 unsigned int tmp_linesize = 2 * linesize; unsigned int tmp_uvlinesize = 2 * uvlinesize; int mbn_xy = mb_xy - 2 * s->mb_stride; - int qp, chroma_qp; + int qp; int i, j; int16_t bS[4]; @@ -7253,10 +6625,10 @@ static void filter_mb( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8 tprintf(s->avctx, "filter mb:%d/%d dir:%d edge:%d, QPy:%d ls:%d uvls:%d", mb_x, mb_y, dir, edge, qp, tmp_linesize, tmp_uvlinesize); { int i; for (i = 0; i < 4; i++) tprintf(s->avctx, " bS[%d]:%d", i, bS[i]); tprintf(s->avctx, "\n"); } filter_mb_edgeh( h, &img_y[j*linesize], tmp_linesize, bS, qp ); - chroma_qp = ( h->chroma_qp + - get_chroma_qp( h->pps.chroma_qp_index_offset, s->current_picture.qscale_table[mbn_xy] ) + 1 ) >> 1; - filter_mb_edgech( h, &img_cb[j*uvlinesize], tmp_uvlinesize, bS, chroma_qp ); - filter_mb_edgech( h, &img_cr[j*uvlinesize], tmp_uvlinesize, bS, chroma_qp ); + filter_mb_edgech( h, &img_cb[j*uvlinesize], tmp_uvlinesize, bS, + ( h->chroma_qp[0] + get_chroma_qp( h, 0, s->current_picture.qscale_table[mbn_xy] ) + 1 ) >> 1); + filter_mb_edgech( h, &img_cr[j*uvlinesize], tmp_uvlinesize, bS, + ( h->chroma_qp[1] + get_chroma_qp( h, 1, s->current_picture.qscale_table[mbn_xy] ) + 1 ) >> 1); } start = 1; @@ -7353,25 +6725,25 @@ static void filter_mb( H264Context *h, int mb_x, int mb_y, uint8_t *img_y, uint8 if( dir == 0 ) { filter_mb_edgev( h, &img_y[4*edge], linesize, bS, qp ); if( (edge&1) == 0 ) { - int chroma_qp = ( h->chroma_qp + - get_chroma_qp( h->pps.chroma_qp_index_offset, s->current_picture.qscale_table[mbn_xy] ) + 1 ) >> 1; - filter_mb_edgecv( h, &img_cb[2*edge], uvlinesize, bS, chroma_qp ); - filter_mb_edgecv( h, &img_cr[2*edge], uvlinesize, bS, chroma_qp ); + filter_mb_edgecv( h, &img_cb[2*edge], uvlinesize, bS, + ( h->chroma_qp[0] + get_chroma_qp( h, 0, s->current_picture.qscale_table[mbn_xy] ) + 1 ) >> 1); + filter_mb_edgecv( h, &img_cr[2*edge], uvlinesize, bS, + ( h->chroma_qp[1] + get_chroma_qp( h, 1, s->current_picture.qscale_table[mbn_xy] ) + 1 ) >> 1); } } else { filter_mb_edgeh( h, &img_y[4*edge*linesize], linesize, bS, qp ); if( (edge&1) == 0 ) { - int chroma_qp = ( h->chroma_qp + - get_chroma_qp( h->pps.chroma_qp_index_offset, s->current_picture.qscale_table[mbn_xy] ) + 1 ) >> 1; - filter_mb_edgech( h, &img_cb[2*edge*uvlinesize], uvlinesize, bS, chroma_qp ); - filter_mb_edgech( h, &img_cr[2*edge*uvlinesize], uvlinesize, bS, chroma_qp ); + filter_mb_edgech( h, &img_cb[2*edge*uvlinesize], uvlinesize, bS, + ( h->chroma_qp[0] + get_chroma_qp( h, 0, s->current_picture.qscale_table[mbn_xy] ) + 1 ) >> 1); + filter_mb_edgech( h, &img_cr[2*edge*uvlinesize], uvlinesize, bS, + ( h->chroma_qp[1] + get_chroma_qp( h, 1, s->current_picture.qscale_table[mbn_xy] ) + 1 ) >> 1); } } } } } -static int decode_slice(H264Context *h){ +static int decode_slice(struct AVCodecContext *avctx, H264Context *h){ MpegEncContext * const s = &h->s; const int part_mask= s->partitioned_frame ? (AC_END|AC_ERROR) : 0x7F; @@ -7421,7 +6793,7 @@ static int decode_slice(H264Context *h){ eos = get_cabac_terminate( &h->cabac ); if( ret < 0 || h->cabac.bytestream > h->cabac.bytestream_end + 2) { - av_log(h->s.avctx, AV_LOG_ERROR, "error while decoding MB %d %d, bytestream (%d)\n", s->mb_x, s->mb_y, h->cabac.bytestream_end - h->cabac.bytestream); + av_log(h->s.avctx, AV_LOG_ERROR, "error while decoding MB %d %d, bytestream (%td)\n", s->mb_x, s->mb_y, h->cabac.bytestream_end - h->cabac.bytestream); ff_er_add_slice(s, s->resync_mb_x, s->resync_mb_y, s->mb_x, s->mb_y, (AC_ERROR|DC_ERROR|MV_ERROR)&part_mask); return -1; } @@ -7430,7 +6802,7 @@ static int decode_slice(H264Context *h){ s->mb_x = 0; ff_draw_horiz_band(s, 16*s->mb_y, 16); ++s->mb_y; - if(FRAME_MBAFF) { + if(FIELD_OR_MBAFF_PICTURE) { ++s->mb_y; } } @@ -7467,7 +6839,7 @@ static int decode_slice(H264Context *h){ s->mb_x=0; ff_draw_horiz_band(s, 16*s->mb_y, 16); ++s->mb_y; - if(FRAME_MBAFF) { + if(FIELD_OR_MBAFF_PICTURE) { ++s->mb_y; } if(s->mb_y >= s->mb_height){ @@ -7636,7 +7008,7 @@ static inline int decode_vui_parameters(H264Context *h, SPS *sps){ if( aspect_ratio_idc == EXTENDED_SAR ) { sps->sar.num= get_bits(&s->gb, 16); sps->sar.den= get_bits(&s->gb, 16); - }else if(aspect_ratio_idc < 14){ + }else if(aspect_ratio_idc < sizeof(pixel_aspect)/sizeof(*pixel_aspect)){ sps->sar= pixel_aspect[aspect_ratio_idc]; }else{ av_log(h->s.avctx, AV_LOG_ERROR, "illegal aspect ratio\n"); @@ -7753,6 +7125,26 @@ static void decode_scaling_matrices(H264Context *h, SPS *sps, PPS *pps, int is_s } } +/** + * Returns and optionally allocates SPS / PPS structures in the supplied array 'vec' + */ +static void * +alloc_parameter_set(H264Context *h, void **vec, const unsigned int id, const unsigned int max, + const size_t size, const char *name) +{ + if(id>=max) { + av_log(h->s.avctx, AV_LOG_ERROR, "%s_id (%d) out of range\n", name, id); + return NULL; + } + + if(!vec[id]) { + vec[id] = av_mallocz(size); + if(vec[id] == NULL) + av_log(h->s.avctx, AV_LOG_ERROR, "cannot allocate memory for %s\n", name); + } + return vec[id]; +} + static inline int decode_seq_parameter_set(H264Context *h){ MpegEncContext * const s = &h->s; int profile_idc, level_idc; @@ -7769,13 +7161,10 @@ static inline int decode_seq_parameter_set(H264Context *h){ level_idc= get_bits(&s->gb, 8); sps_id= get_ue_golomb(&s->gb); - if (sps_id >= MAX_SPS_COUNT){ - // ok it has gone out of hand, someone is sending us bad stuff. - av_log(h->s.avctx, AV_LOG_ERROR, "illegal sps_id (%d)\n", sps_id); + sps = alloc_parameter_set(h, (void **)h->sps_buffers, sps_id, MAX_SPS_COUNT, sizeof(SPS), "sps"); + if(sps == NULL) return -1; - } - sps= &h->sps_buffer[ sps_id ]; sps->profile_idc= profile_idc; sps->level_idc= level_idc; @@ -7814,8 +7203,9 @@ static inline int decode_seq_parameter_set(H264Context *h){ } tmp= get_ue_golomb(&s->gb); - if(tmp > MAX_PICTURE_COUNT-2){ + if(tmp > MAX_PICTURE_COUNT-2 || tmp >= 32){ av_log(h->s.avctx, AV_LOG_ERROR, "too many reference frames\n"); + return -1; } sps->ref_frame_count= tmp; sps->gaps_in_frame_num_allowed_flag= get_bits1(&s->gb); @@ -7880,19 +7270,25 @@ static inline int decode_seq_parameter_set(H264Context *h){ return 0; } +static void +build_qp_table(PPS *pps, int t, int index) +{ + int i; + for(i = 0; i < 255; i++) + pps->chroma_qp_table[t][i & 0xff] = chroma_qp[av_clip(i + index, 0, 51)]; +} + static inline int decode_picture_parameter_set(H264Context *h, int bit_length){ MpegEncContext * const s = &h->s; unsigned int tmp, pps_id= get_ue_golomb(&s->gb); PPS *pps; - if(pps_id>=MAX_PPS_COUNT){ - av_log(h->s.avctx, AV_LOG_ERROR, "pps_id out of range\n"); + pps = alloc_parameter_set(h, (void **)h->pps_buffers, pps_id, MAX_PPS_COUNT, sizeof(PPS), "pps"); + if(pps == NULL) return -1; - } - pps = &h->pps_buffer[pps_id]; tmp= get_ue_golomb(&s->gb); - if(tmp>=MAX_SPS_COUNT){ + if(tmp>=MAX_SPS_COUNT || h->sps_buffers[tmp] == NULL){ av_log(h->s.avctx, AV_LOG_ERROR, "sps_id out of range\n"); return -1; } @@ -7950,7 +7346,7 @@ static inline int decode_picture_parameter_set(H264Context *h, int bit_length){ pps->weighted_bipred_idc= get_bits(&s->gb, 2); pps->init_qp= get_se_golomb(&s->gb) + 26; pps->init_qs= get_se_golomb(&s->gb) + 26; - pps->chroma_qp_index_offset= get_se_golomb(&s->gb); + pps->chroma_qp_index_offset[0]= get_se_golomb(&s->gb); pps->deblocking_filter_parameters_present= get_bits1(&s->gb); pps->constrained_intra_pred= get_bits1(&s->gb); pps->redundant_pic_cnt_present = get_bits1(&s->gb); @@ -7962,18 +7358,27 @@ static inline int decode_picture_parameter_set(H264Context *h, int bit_length){ if(get_bits_count(&s->gb) < bit_length){ pps->transform_8x8_mode= get_bits1(&s->gb); - decode_scaling_matrices(h, &h->sps_buffer[pps->sps_id], pps, 0, pps->scaling_matrix4, pps->scaling_matrix8); - get_se_golomb(&s->gb); //second_chroma_qp_index_offset + decode_scaling_matrices(h, h->sps_buffers[pps->sps_id], pps, 0, pps->scaling_matrix4, pps->scaling_matrix8); + pps->chroma_qp_index_offset[1]= get_se_golomb(&s->gb); //second_chroma_qp_index_offset + } else { + pps->chroma_qp_index_offset[1]= pps->chroma_qp_index_offset[0]; } + build_qp_table(pps, 0, pps->chroma_qp_index_offset[0]); + if(pps->chroma_qp_index_offset[0] != pps->chroma_qp_index_offset[1]) { + build_qp_table(pps, 1, pps->chroma_qp_index_offset[1]); + h->pps.chroma_qp_diff= 1; + } else + memcpy(pps->chroma_qp_table[1], pps->chroma_qp_table[0], 256); + if(s->avctx->debug&FF_DEBUG_PICT_INFO){ - av_log(h->s.avctx, AV_LOG_DEBUG, "pps:%u sps:%u %s slice_groups:%d ref:%d/%d %s qp:%d/%d/%d %s %s %s %s\n", + av_log(h->s.avctx, AV_LOG_DEBUG, "pps:%u sps:%u %s slice_groups:%d ref:%d/%d %s qp:%d/%d/%d/%d %s %s %s %s\n", pps_id, pps->sps_id, pps->cabac ? "CABAC" : "CAVLC", pps->slice_group_count, pps->ref_count[0], pps->ref_count[1], pps->weighted_pred ? "weighted" : "", - pps->init_qp, pps->init_qs, pps->chroma_qp_index_offset, + pps->init_qp, pps->init_qs, pps->chroma_qp_index_offset[0], pps->chroma_qp_index_offset[1], pps->deblocking_filter_parameters_present ? "LPAR" : "", pps->constrained_intra_pred ? "CONSTR" : "", pps->redundant_pic_cnt_present ? "REDU" : "", @@ -7985,119 +7390,49 @@ static inline int decode_picture_parameter_set(H264Context *h, int bit_length){ } /** - * finds the end of the current frame in the bitstream. - * @return the position of the first byte of the next frame, or -1 + * Call decode_slice() for each context. + * + * @param h h264 master context + * @param context_count number of contexts to execute */ -static int find_frame_end(H264Context *h, const uint8_t *buf, int buf_size){ +static void execute_decode_slices(H264Context *h, int context_count){ + MpegEncContext * const s = &h->s; + AVCodecContext * const avctx= s->avctx; + H264Context *hx; int i; - uint32_t state; - ParseContext *pc = &(h->s.parse_context); -//printf("first %02X%02X%02X%02X\n", buf[0], buf[1],buf[2],buf[3]); -// mb_addr= pc->mb_addr - 1; - state= pc->state; - if(state>13) - state= 7; - - for(i=0; i<buf_size; i++){ - if(state==7){ - for(; i<buf_size; i++){ - if(!buf[i]){ - state=2; - break; - } - } - }else if(state<=2){ - if(buf[i]==1) state^= 5; //2->7, 1->4, 0->5 - else if(buf[i]) state = 7; - else state>>=1; //2->1, 1->0, 0->0 - }else if(state<=5){ - int v= buf[i] & 0x1F; - if(v==7 || v==8 || v==9){ - if(pc->frame_start_found){ - i++; -found: - pc->state=7; - pc->frame_start_found= 0; - return i-(state&5); - } - }else if(v==1 || v==2 || v==5){ - if(pc->frame_start_found){ - state+=8; - continue; - }else - pc->frame_start_found = 1; - } - state= 7; - }else{ - if(buf[i] & 0x80) - goto found; - state= 7; - } - } - pc->state= state; - return END_NOT_FOUND; -} - -#ifdef CONFIG_H264_PARSER -static int h264_parse(AVCodecParserContext *s, - AVCodecContext *avctx, - uint8_t **poutbuf, int *poutbuf_size, - const uint8_t *buf, int buf_size) -{ - H264Context *h = s->priv_data; - ParseContext *pc = &h->s.parse_context; - int next; - - if(s->flags & PARSER_FLAG_COMPLETE_FRAMES){ - next= buf_size; - }else{ - next= find_frame_end(h, buf, buf_size); - - if (ff_combine_frame(pc, next, (uint8_t **)&buf, &buf_size) < 0) { - *poutbuf = NULL; - *poutbuf_size = 0; - return buf_size; - } - if(next<0){ - find_frame_end(h, &pc->buffer[pc->last_index + next], -next); //update state + if(context_count == 1) { + decode_slice(avctx, h); + } else { + for(i = 1; i < context_count; i++) { + hx = h->thread_context[i]; + hx->s.error_resilience = avctx->error_resilience; + hx->s.error_count = 0; } - } - - *poutbuf = (uint8_t *)buf; - *poutbuf_size = buf_size; - return next; -} -static int h264_split(AVCodecContext *avctx, - const uint8_t *buf, int buf_size) -{ - int i; - uint32_t state = -1; - int has_sps= 0; + avctx->execute(avctx, (void *)decode_slice, + (void **)h->thread_context, NULL, context_count); - for(i=0; i<=buf_size; i++){ - if((state&0xFFFFFF1F) == 0x107) - has_sps=1; -/* if((state&0xFFFFFF1F) == 0x101 || (state&0xFFFFFF1F) == 0x102 || (state&0xFFFFFF1F) == 0x105){ - }*/ - if((state&0xFFFFFF00) == 0x100 && (state&0xFFFFFF1F) != 0x107 && (state&0xFFFFFF1F) != 0x108 && (state&0xFFFFFF1F) != 0x109){ - if(has_sps){ - while(i>4 && buf[i-5]==0) i--; - return i-4; - } - } - if (i<buf_size) - state= (state<<8) | buf[i]; + /* pull back stuff from slices to master context */ + hx = h->thread_context[context_count - 1]; + s->mb_x = hx->s.mb_x; + s->mb_y = hx->s.mb_y; + s->dropable = hx->s.dropable; + s->picture_structure = hx->s.picture_structure; + for(i = 1; i < context_count; i++) + h->s.error_count += h->thread_context[i]->s.error_count; } - return 0; } -#endif /* CONFIG_H264_PARSER */ -static int decode_nal_units(H264Context *h, uint8_t *buf, int buf_size){ + +static int decode_nal_units(H264Context *h, const uint8_t *buf, int buf_size){ MpegEncContext * const s = &h->s; AVCodecContext * const avctx= s->avctx; int buf_index=0; + H264Context *hx; ///< thread context + int context_count = 0; + + h->max_contexts = avctx->thread_count; #if 0 int i; for(i=0; i<50; i++){ @@ -8105,54 +7440,58 @@ static int decode_nal_units(H264Context *h, uint8_t *buf, int buf_size){ } #endif if(!(s->flags2 & CODEC_FLAG2_CHUNKS)){ - h->slice_num = 0; - s->current_picture_ptr= NULL; + h->current_slice = 0; + if (!s->first_field) + s->current_picture_ptr= NULL; } for(;;){ int consumed; int dst_length; int bit_length; - uint8_t *ptr; + const uint8_t *ptr; int i, nalsize = 0; - - if(h->is_avc) { - if(buf_index >= buf_size) break; - nalsize = 0; - for(i = 0; i < h->nal_length_size; i++) - nalsize = (nalsize << 8) | buf[buf_index++]; - if(nalsize <= 1 || nalsize > buf_size){ - if(nalsize == 1){ - buf_index++; - continue; - }else{ - av_log(h->s.avctx, AV_LOG_ERROR, "AVC: nal size %d\n", nalsize); - break; + int err; + + if(h->is_avc) { + if(buf_index >= buf_size) break; + nalsize = 0; + for(i = 0; i < h->nal_length_size; i++) + nalsize = (nalsize << 8) | buf[buf_index++]; + if(nalsize <= 1 || (nalsize+buf_index > buf_size)){ + if(nalsize == 1){ + buf_index++; + continue; + }else{ + av_log(h->s.avctx, AV_LOG_ERROR, "AVC: nal size %d\n", nalsize); + break; + } + } + } else { + // start code prefix search + for(; buf_index + 3 < buf_size; buf_index++){ + // This should always succeed in the first iteration. + if(buf[buf_index] == 0 && buf[buf_index+1] == 0 && buf[buf_index+2] == 1) + break; } - } - } else { - // start code prefix search - for(; buf_index + 3 < buf_size; buf_index++){ - // this should allways succeed in the first iteration - if(buf[buf_index] == 0 && buf[buf_index+1] == 0 && buf[buf_index+2] == 1) - break; - } - if(buf_index+3 >= buf_size) break; + if(buf_index+3 >= buf_size) break; - buf_index+=3; - } + buf_index+=3; + } + + hx = h->thread_context[context_count]; - ptr= decode_nal(h, buf + buf_index, &dst_length, &consumed, h->is_avc ? nalsize : buf_size - buf_index); - if (ptr==NULL || dst_length <= 0){ + ptr= decode_nal(hx, buf + buf_index, &dst_length, &consumed, h->is_avc ? nalsize : buf_size - buf_index); + if (ptr==NULL || dst_length < 0){ return -1; } - while(ptr[dst_length - 1] == 0 && dst_length > 1) + while(ptr[dst_length - 1] == 0 && dst_length > 0) dst_length--; - bit_length= 8*dst_length - decode_rbsp_trailing(h, ptr + dst_length - 1); + bit_length= !dst_length ? 0 : (8*dst_length - decode_rbsp_trailing(h, ptr + dst_length - 1)); if(s->avctx->debug&FF_DEBUG_STARTCODE){ - av_log(h->s.avctx, AV_LOG_DEBUG, "NAL %d at %d/%d length %d\n", h->nal_unit_type, buf_index, buf_size, dst_length); + av_log(h->s.avctx, AV_LOG_DEBUG, "NAL %d at %d/%d length %d\n", hx->nal_unit_type, buf_index, buf_size, dst_length); } if (h->is_avc && (nalsize != consumed)) @@ -8160,57 +7499,60 @@ static int decode_nal_units(H264Context *h, uint8_t *buf, int buf_size){ buf_index += consumed; - if( (s->hurry_up == 1 && h->nal_ref_idc == 0) //FIXME dont discard SEI id + if( (s->hurry_up == 1 && h->nal_ref_idc == 0) //FIXME do not discard SEI id ||(avctx->skip_frame >= AVDISCARD_NONREF && h->nal_ref_idc == 0)) continue; - switch(h->nal_unit_type){ + again: + err = 0; + switch(hx->nal_unit_type){ case NAL_IDR_SLICE: + if (h->nal_unit_type != NAL_IDR_SLICE) { + av_log(h->s.avctx, AV_LOG_ERROR, "Invalid mix of idr and non-idr slices"); + return -1; + } idr(h); //FIXME ensure we don't loose some frames if there is reordering case NAL_SLICE: - init_get_bits(&s->gb, ptr, bit_length); - h->intra_gb_ptr= - h->inter_gb_ptr= &s->gb; - s->data_partitioning = 0; - - if(decode_slice_header(h) < 0){ - av_log(h->s.avctx, AV_LOG_ERROR, "decode_slice_header error\n"); - break; - } - s->current_picture_ptr->key_frame= (h->nal_unit_type == NAL_IDR_SLICE); - if(h->redundant_pic_count==0 && s->hurry_up < 5 - && (avctx->skip_frame < AVDISCARD_NONREF || h->nal_ref_idc) - && (avctx->skip_frame < AVDISCARD_BIDIR || h->slice_type!=B_TYPE) - && (avctx->skip_frame < AVDISCARD_NONKEY || h->slice_type==I_TYPE) + init_get_bits(&hx->s.gb, ptr, bit_length); + hx->intra_gb_ptr= + hx->inter_gb_ptr= &hx->s.gb; + hx->s.data_partitioning = 0; + + if((err = decode_slice_header(hx, h))) + break; + + s->current_picture_ptr->key_frame|= (hx->nal_unit_type == NAL_IDR_SLICE); + if(hx->redundant_pic_count==0 && hx->s.hurry_up < 5 + && (avctx->skip_frame < AVDISCARD_NONREF || hx->nal_ref_idc) + && (avctx->skip_frame < AVDISCARD_BIDIR || hx->slice_type!=B_TYPE) + && (avctx->skip_frame < AVDISCARD_NONKEY || hx->slice_type==I_TYPE) && avctx->skip_frame < AVDISCARD_ALL) - decode_slice(h); + context_count++; break; case NAL_DPA: - init_get_bits(&s->gb, ptr, bit_length); - h->intra_gb_ptr= - h->inter_gb_ptr= NULL; - s->data_partitioning = 1; + init_get_bits(&hx->s.gb, ptr, bit_length); + hx->intra_gb_ptr= + hx->inter_gb_ptr= NULL; + hx->s.data_partitioning = 1; - if(decode_slice_header(h) < 0){ - av_log(h->s.avctx, AV_LOG_ERROR, "decode_slice_header error\n"); - } + err = decode_slice_header(hx, h); break; case NAL_DPB: - init_get_bits(&h->intra_gb, ptr, bit_length); - h->intra_gb_ptr= &h->intra_gb; + init_get_bits(&hx->intra_gb, ptr, bit_length); + hx->intra_gb_ptr= &hx->intra_gb; break; case NAL_DPC: - init_get_bits(&h->inter_gb, ptr, bit_length); - h->inter_gb_ptr= &h->inter_gb; + init_get_bits(&hx->inter_gb, ptr, bit_length); + hx->inter_gb_ptr= &hx->inter_gb; - if(h->redundant_pic_count==0 && h->intra_gb_ptr && s->data_partitioning + if(hx->redundant_pic_count==0 && hx->intra_gb_ptr && hx->s.data_partitioning && s->context_initialized && s->hurry_up < 5 - && (avctx->skip_frame < AVDISCARD_NONREF || h->nal_ref_idc) - && (avctx->skip_frame < AVDISCARD_BIDIR || h->slice_type!=B_TYPE) - && (avctx->skip_frame < AVDISCARD_NONKEY || h->slice_type==I_TYPE) + && (avctx->skip_frame < AVDISCARD_NONREF || hx->nal_ref_idc) + && (avctx->skip_frame < AVDISCARD_BIDIR || hx->slice_type!=B_TYPE) + && (avctx->skip_frame < AVDISCARD_NONKEY || hx->slice_type==I_TYPE) && avctx->skip_frame < AVDISCARD_ALL) - decode_slice(h); + context_count++; break; case NAL_SEI: init_get_bits(&s->gb, ptr, bit_length); @@ -8240,10 +7582,29 @@ static int decode_nal_units(H264Context *h, uint8_t *buf, int buf_size){ case NAL_AUXILIARY_SLICE: break; default: - av_log(avctx, AV_LOG_ERROR, "Unknown NAL code: %d\n", h->nal_unit_type); + av_log(avctx, AV_LOG_DEBUG, "Unknown NAL code: %d (%d bits)\n", h->nal_unit_type, bit_length); + } + + if(context_count == h->max_contexts) { + execute_decode_slices(h, context_count); + context_count = 0; } - } + if (err < 0) + av_log(h->s.avctx, AV_LOG_ERROR, "decode_slice_header error\n"); + else if(err == 1) { + /* Slice could not be decoded in parallel mode, copy down + * NAL unit stuff to context 0 and restart. Note that + * rbsp_buffer is not transfered, but since we no longer + * run in parallel mode this should not be an issue. */ + h->nal_unit_type = hx->nal_unit_type; + h->nal_ref_idc = hx->nal_ref_idc; + hx = h; + goto again; + } + } + if(context_count) + execute_decode_slices(h, context_count); return buf_index; } @@ -8257,7 +7618,7 @@ static int get_consumed_bytes(MpegEncContext *s, int pos, int buf_size){ return pos; }else{ - if(pos==0) pos=1; //avoid infinite loops (i doubt thats needed but ...) + if(pos==0) pos=1; //avoid infinite loops (i doubt that is needed but ...) if(pos+10>buf_size) pos=buf_size; // oops ;) return pos; @@ -8266,7 +7627,7 @@ static int get_consumed_bytes(MpegEncContext *s, int pos, int buf_size){ static int decode_frame(AVCodecContext *avctx, void *data, int *data_size, - uint8_t *buf, int buf_size) + const uint8_t *buf, int buf_size) { H264Context *h = avctx->priv_data; MpegEncContext *s = &h->s; @@ -8302,9 +7663,9 @@ static int decode_frame(AVCodecContext *avctx, } if(s->flags&CODEC_FLAG_TRUNCATED){ - int next= find_frame_end(h, buf, buf_size); + int next= ff_h264_find_frame_end(h, buf, buf_size); - if( ff_combine_frame(&s->parse_context, next, &buf, &buf_size) < 0 ) + if( ff_combine_frame(&s->parse_context, next, (const uint8_t **)&buf, &buf_size) < 0 ) return buf_size; //printf("next:%d buf_size:%d last_index:%d\n", next, buf_size, s->parse_context.last_index); } @@ -8360,6 +7721,7 @@ static int decode_frame(AVCodecContext *avctx, return -1; if(!(s->flags2 & CODEC_FLAG2_CHUNKS) && !s->current_picture_ptr){ + if (avctx->skip_frame >= AVDISCARD_NONREF || s->hurry_up) return 0; av_log(avctx, AV_LOG_ERROR, "no frame!\n"); return -1; } @@ -8377,87 +7739,109 @@ static int decode_frame(AVCodecContext *avctx, h->prev_frame_num_offset= h->frame_num_offset; h->prev_frame_num= h->frame_num; - if(s->current_picture_ptr->reference){ + if(!s->dropable) { h->prev_poc_msb= h->poc_msb; h->prev_poc_lsb= h->poc_lsb; - } - if(s->current_picture_ptr->reference) execute_ref_pic_marking(h, h->mmco, h->mmco_index); + } - ff_er_frame_end(s); + /* + * FIXME: Error handling code does not seem to support interlaced + * when slices span multiple rows + * The ff_er_add_slice calls don't work right for bottom + * fields; they cause massive erroneous error concealing + * Error marking covers both fields (top and bottom). + * This causes a mismatched s->error_count + * and a bad error table. Further, the error count goes to + * INT_MAX when called for bottom field, because mb_y is + * past end by one (callers fault) and resync_mb_y != 0 + * causes problems for the first MB line, too. + */ + if (!FIELD_PICTURE) + ff_er_frame_end(s); MPV_frame_end(s); - //FIXME do something with unavailable reference frames + if (s->first_field) { + /* Wait for second field. */ + *data_size = 0; -#if 0 //decode order - *data_size = sizeof(AVFrame); -#else - /* Sort B-frames into display order */ + } else { + cur->interlaced_frame = FIELD_OR_MBAFF_PICTURE; + /* Derive top_field_first from field pocs. */ + cur->top_field_first = cur->field_poc[0] < cur->field_poc[1]; - if(h->sps.bitstream_restriction_flag - && s->avctx->has_b_frames < h->sps.num_reorder_frames){ - s->avctx->has_b_frames = h->sps.num_reorder_frames; - s->low_delay = 0; - } + //FIXME do something with unavailable reference frames - pics = 0; - while(h->delayed_pic[pics]) pics++; +#if 0 //decode order + *data_size = sizeof(AVFrame); +#else + /* Sort B-frames into display order */ - assert(pics+1 < sizeof(h->delayed_pic) / sizeof(h->delayed_pic[0])); + if(h->sps.bitstream_restriction_flag + && s->avctx->has_b_frames < h->sps.num_reorder_frames){ + s->avctx->has_b_frames = h->sps.num_reorder_frames; + s->low_delay = 0; + } - h->delayed_pic[pics++] = cur; - if(cur->reference == 0) - cur->reference = 1; + pics = 0; + while(h->delayed_pic[pics]) pics++; - cross_idr = 0; - for(i=0; h->delayed_pic[i]; i++) - if(h->delayed_pic[i]->key_frame || h->delayed_pic[i]->poc==0) - cross_idr = 1; + assert(pics+1 < sizeof(h->delayed_pic) / sizeof(h->delayed_pic[0])); - out = h->delayed_pic[0]; - out_idx = 0; - for(i=1; h->delayed_pic[i] && !h->delayed_pic[i]->key_frame; i++) - if(h->delayed_pic[i]->poc < out->poc){ - out = h->delayed_pic[i]; - out_idx = i; - } + h->delayed_pic[pics++] = cur; + if(cur->reference == 0) + cur->reference = DELAYED_PIC_REF; - out_of_order = !cross_idr && prev && out->poc < prev->poc; - if(h->sps.bitstream_restriction_flag && s->avctx->has_b_frames >= h->sps.num_reorder_frames) - { } - else if(prev && pics <= s->avctx->has_b_frames) - out = prev; - else if((out_of_order && pics-1 == s->avctx->has_b_frames && pics < 15) - || (s->low_delay && - ((!cross_idr && prev && out->poc > prev->poc + 2) - || cur->pict_type == B_TYPE))) - { - s->low_delay = 0; - s->avctx->has_b_frames++; - out = prev; - } - else if(out_of_order) - out = prev; + cross_idr = 0; + for(i=0; h->delayed_pic[i]; i++) + if(h->delayed_pic[i]->key_frame || h->delayed_pic[i]->poc==0) + cross_idr = 1; - if(out_of_order || pics > s->avctx->has_b_frames){ - for(i=out_idx; h->delayed_pic[i]; i++) - h->delayed_pic[i] = h->delayed_pic[i+1]; - } + out = h->delayed_pic[0]; + out_idx = 0; + for(i=1; h->delayed_pic[i] && !h->delayed_pic[i]->key_frame; i++) + if(h->delayed_pic[i]->poc < out->poc){ + out = h->delayed_pic[i]; + out_idx = i; + } - if(prev == out) - *data_size = 0; - else - *data_size = sizeof(AVFrame); - if(prev && prev != out && prev->reference == 1) - prev->reference = 0; - h->delayed_output_pic = out; + out_of_order = !cross_idr && prev && out->poc < prev->poc; + if(h->sps.bitstream_restriction_flag && s->avctx->has_b_frames >= h->sps.num_reorder_frames) + { } + else if(prev && pics <= s->avctx->has_b_frames) + out = prev; + else if((out_of_order && pics-1 == s->avctx->has_b_frames && pics < 15) + || (s->low_delay && + ((!cross_idr && prev && out->poc > prev->poc + 2) + || cur->pict_type == B_TYPE))) + { + s->low_delay = 0; + s->avctx->has_b_frames++; + out = prev; + } + else if(out_of_order) + out = prev; + + if(out_of_order || pics > s->avctx->has_b_frames){ + for(i=out_idx; h->delayed_pic[i]; i++) + h->delayed_pic[i] = h->delayed_pic[i+1]; + } + + if(prev == out) + *data_size = 0; + else + *data_size = sizeof(AVFrame); + if(prev && prev != out && prev->reference == DELAYED_PIC_REF) + prev->reference = 0; + h->delayed_output_pic = out; #endif - if(out) - *pict= *(AVFrame*)out; - else - av_log(avctx, AV_LOG_DEBUG, "no picture\n"); + if(out) + *pict= *(AVFrame*)out; + else + av_log(avctx, AV_LOG_DEBUG, "no picture\n"); + } } assert(pict->data[0] || !*data_size); @@ -8466,7 +7850,7 @@ static int decode_frame(AVCodecContext *avctx, #if 0 //? /* Return the Picture timestamp as the frame number */ - /* we substract 1 because it is added on utils.c */ + /* we subtract 1 because it is added on utils.c */ avctx->frame_number = s->picture_number - 1; #endif return get_consumed_bytes(s, buf_index, buf_size); @@ -8491,10 +7875,12 @@ static inline void fill_mb_avail(H264Context *h){ } #endif -#if 0 //selftest +#ifdef TEST +#undef printf +#undef random #define COUNT 8000 #define SIZE (COUNT*40) -int main(){ +int main(void){ int i; uint8_t temp[SIZE]; PutBitContext pb; @@ -8523,7 +7909,7 @@ int main(){ START_TIMER j= get_ue_golomb(&gb); if(j != i){ - printf("missmatch! at %d (%d should be %d) bits:%6X\n", i, j, i, s); + printf("mismatch! at %d (%d should be %d) bits:%6X\n", i, j, i, s); // return -1; } STOP_TIMER("get_ue_golomb"); @@ -8548,12 +7934,13 @@ int main(){ START_TIMER j= get_se_golomb(&gb); if(j != i - COUNT/2){ - printf("missmatch! at %d (%d should be %d) bits:%6X\n", i, j, i, s); + printf("mismatch! at %d (%d should be %d) bits:%6X\n", i, j, i, s); // return -1; } STOP_TIMER("get_se_golomb"); } +#if 0 printf("testing 4x4 (I)DCT\n"); DCTELEM block[16]; @@ -8593,14 +7980,12 @@ int main(){ } } printf("error=%f max_error=%d\n", ((float)error)/COUNT/16, (int)max_error ); -#if 0 printf("testing quantizer\n"); for(qp=0; qp<52; qp++){ for(i=0; i<16; i++) src1_block[i]= src2_block[i]= random()%255; } -#endif printf("Testing NAL layer\n"); uint8_t bitstream[COUNT]; @@ -8652,17 +8037,18 @@ int main(){ } if(memcmp(bitstream, out, COUNT)){ - printf("missmatch\n"); + printf("mismatch\n"); return -1; } } +#endif printf("Testing RBSP\n"); return 0; } -#endif +#endif /* TEST */ static int decode_end(AVCodecContext *avctx) @@ -8670,7 +8056,8 @@ static int decode_end(AVCodecContext *avctx) H264Context *h = avctx->priv_data; MpegEncContext *s = &h->s; - av_freep(&h->rbsp_buffer); + av_freep(&h->rbsp_buffer[0]); + av_freep(&h->rbsp_buffer[1]); free_tables(h); //FIXME cleanup init stuff perhaps MPV_common_end(s); @@ -8693,15 +8080,4 @@ AVCodec h264_decoder = { .flush= flush_dpb, }; -#ifdef CONFIG_H264_PARSER -AVCodecParser h264_parser = { - { CODEC_ID_H264 }, - sizeof(H264Context), - NULL, - h264_parse, - ff_parse_close, - h264_split, -}; -#endif - #include "svq3.c" |