2 * H.26L/H.264/AVC/JVT/14496-10/... reference picture handling
3 * Copyright (c) 2003 Michael Niedermayer <michaelni@gmx.at>
5 * This file is part of Libav.
7 * Libav is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
12 * Libav is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with Libav; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
24 * H.264 / AVC / MPEG4 part10 reference picture handling.
25 * @author Michael Niedermayer <michaelni@gmx.at>
38 static void pic_as_field(Picture *pic, const int parity){
40 for (i = 0; i < 4; ++i) {
41 if (parity == PICT_BOTTOM_FIELD)
42 pic->f.data[i] += pic->f.linesize[i];
43 pic->f.reference = parity;
44 pic->f.linesize[i] *= 2;
46 pic->poc= pic->field_poc[parity == PICT_BOTTOM_FIELD];
49 static int split_field_copy(Picture *dest, Picture *src,
50 int parity, int id_add){
51 int match = !!(src->f.reference & parity);
55 if(parity != PICT_FRAME){
56 pic_as_field(dest, parity);
58 dest->pic_id += id_add;
65 static int build_def_list(Picture *def, Picture **in, int len, int is_long, int sel){
69 while(i[0]<len || i[1]<len){
70 while (i[0] < len && !(in[ i[0] ] && (in[ i[0] ]->f.reference & sel)))
72 while (i[1] < len && !(in[ i[1] ] && (in[ i[1] ]->f.reference & (sel^3))))
75 in[ i[0] ]->pic_id= is_long ? i[0] : in[ i[0] ]->frame_num;
76 split_field_copy(&def[index++], in[ i[0]++ ], sel , 1);
79 in[ i[1] ]->pic_id= is_long ? i[1] : in[ i[1] ]->frame_num;
80 split_field_copy(&def[index++], in[ i[1]++ ], sel^3, 0);
87 static int add_sorted(Picture **sorted, Picture **src, int len, int limit, int dir){
92 best_poc= dir ? INT_MIN : INT_MAX;
95 const int poc= src[i]->poc;
96 if(((poc > limit) ^ dir) && ((poc < best_poc) ^ dir)){
98 sorted[out_i]= src[i];
101 if(best_poc == (dir ? INT_MIN : INT_MAX))
103 limit= sorted[out_i++]->poc - dir;
108 int ff_h264_fill_default_ref_list(H264Context *h){
109 MpegEncContext * const s = &h->s;
112 if(h->slice_type_nos==AV_PICTURE_TYPE_B){
118 cur_poc= s->current_picture_ptr->field_poc[ s->picture_structure == PICT_BOTTOM_FIELD ];
120 cur_poc= s->current_picture_ptr->poc;
122 for(list= 0; list<2; list++){
123 len= add_sorted(sorted , h->short_ref, h->short_ref_count, cur_poc, 1^list);
124 len+=add_sorted(sorted+len, h->short_ref, h->short_ref_count, cur_poc, 0^list);
126 len= build_def_list(h->default_ref_list[list] , sorted , len, 0, s->picture_structure);
127 len+=build_def_list(h->default_ref_list[list]+len, h->long_ref, 16 , 1, s->picture_structure);
130 if(len < h->ref_count[list])
131 memset(&h->default_ref_list[list][len], 0, sizeof(Picture)*(h->ref_count[list] - len));
135 if(lens[0] == lens[1] && lens[1] > 1){
136 for (i = 0; h->default_ref_list[0][i].f.data[0] == h->default_ref_list[1][i].f.data[0] && i < lens[0]; i++);
138 FFSWAP(Picture, h->default_ref_list[1][0], h->default_ref_list[1][1]);
141 len = build_def_list(h->default_ref_list[0] , h->short_ref, h->short_ref_count, 0, s->picture_structure);
142 len+= build_def_list(h->default_ref_list[0]+len, h-> long_ref, 16 , 1, s->picture_structure);
144 if(len < h->ref_count[0])
145 memset(&h->default_ref_list[0][len], 0, sizeof(Picture)*(h->ref_count[0] - len));
148 for (i=0; i<h->ref_count[0]; i++) {
149 tprintf(h->s.avctx, "List0: %s fn:%d 0x%p\n", (h->default_ref_list[0][i].long_ref ? "LT" : "ST"), h->default_ref_list[0][i].pic_id, h->default_ref_list[0][i].data[0]);
151 if(h->slice_type_nos==AV_PICTURE_TYPE_B){
152 for (i=0; i<h->ref_count[1]; i++) {
153 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]);
160 static void print_short_term(H264Context *h);
161 static void print_long_term(H264Context *h);
164 * Extract structure information about the picture described by pic_num in
165 * the current decoding context (frame or field). Note that pic_num is
166 * picture number without wrapping (so, 0<=pic_num<max_pic_num).
167 * @param pic_num picture number for which to extract structure information
168 * @param structure one of PICT_XXX describing structure of picture
170 * @return frame number (short term) or long term index of picture
171 * described by pic_num
173 static int pic_num_extract(H264Context *h, int pic_num, int *structure){
174 MpegEncContext * const s = &h->s;
176 *structure = s->picture_structure;
180 *structure ^= PICT_FRAME;
187 int ff_h264_decode_ref_pic_list_reordering(H264Context *h){
188 MpegEncContext * const s = &h->s;
189 int list, index, pic_structure;
194 for(list=0; list<h->list_count; list++){
195 memcpy(h->ref_list[list], h->default_ref_list[list], sizeof(Picture)*h->ref_count[list]);
197 if(get_bits1(&s->gb)){
198 int pred= h->curr_pic_num;
200 for(index=0; ; index++){
201 unsigned int reordering_of_pic_nums_idc= get_ue_golomb_31(&s->gb);
206 if(reordering_of_pic_nums_idc==3)
209 if(index >= h->ref_count[list]){
210 av_log(h->s.avctx, AV_LOG_ERROR, "reference count overflow\n");
214 if(reordering_of_pic_nums_idc<3){
215 if(reordering_of_pic_nums_idc<2){
216 const unsigned int abs_diff_pic_num= get_ue_golomb(&s->gb) + 1;
219 if(abs_diff_pic_num > h->max_pic_num){
220 av_log(h->s.avctx, AV_LOG_ERROR, "abs_diff_pic_num overflow\n");
224 if(reordering_of_pic_nums_idc == 0) pred-= abs_diff_pic_num;
225 else pred+= abs_diff_pic_num;
226 pred &= h->max_pic_num - 1;
228 frame_num = pic_num_extract(h, pred, &pic_structure);
230 for(i= h->short_ref_count-1; i>=0; i--){
231 ref = h->short_ref[i];
232 assert(ref->f.reference);
233 assert(!ref->long_ref);
235 ref->frame_num == frame_num &&
236 (ref->f.reference & pic_structure)
244 pic_id= get_ue_golomb(&s->gb); //long_term_pic_idx
246 long_idx= pic_num_extract(h, pic_id, &pic_structure);
249 av_log(h->s.avctx, AV_LOG_ERROR, "long_term_pic_idx overflow\n");
252 ref = h->long_ref[long_idx];
253 assert(!(ref && !ref->f.reference));
254 if (ref && (ref->f.reference & pic_structure)) {
256 assert(ref->long_ref);
264 av_log(h->s.avctx, AV_LOG_ERROR, "reference picture missing during reorder\n");
265 memset(&h->ref_list[list][index], 0, sizeof(Picture)); //FIXME
267 for(i=index; i+1<h->ref_count[list]; i++){
268 if(ref->long_ref == h->ref_list[list][i].long_ref && ref->pic_id == h->ref_list[list][i].pic_id)
271 for(; i > index; i--){
272 h->ref_list[list][i]= h->ref_list[list][i-1];
274 h->ref_list[list][index]= *ref;
276 pic_as_field(&h->ref_list[list][index], pic_structure);
280 av_log(h->s.avctx, AV_LOG_ERROR, "illegal reordering_of_pic_nums_idc\n");
286 for(list=0; list<h->list_count; list++){
287 for(index= 0; index < h->ref_count[list]; index++){
288 if (!h->ref_list[list][index].f.data[0]) {
289 av_log(h->s.avctx, AV_LOG_ERROR, "Missing reference picture\n");
290 if (h->default_ref_list[list][0].f.data[0])
291 h->ref_list[list][index]= h->default_ref_list[list][0];
301 void ff_h264_fill_mbaff_ref_list(H264Context *h){
303 for(list=0; list<2; list++){ //FIXME try list_count
304 for(i=0; i<h->ref_count[list]; i++){
305 Picture *frame = &h->ref_list[list][i];
306 Picture *field = &h->ref_list[list][16+2*i];
309 field[0].f.linesize[j] <<= 1;
310 field[0].f.reference = PICT_TOP_FIELD;
311 field[0].poc= field[0].field_poc[0];
314 field[1].f.data[j] += frame->f.linesize[j];
315 field[1].f.reference = PICT_BOTTOM_FIELD;
316 field[1].poc= field[1].field_poc[1];
318 h->luma_weight[16+2*i][list][0] = h->luma_weight[16+2*i+1][list][0] = h->luma_weight[i][list][0];
319 h->luma_weight[16+2*i][list][1] = h->luma_weight[16+2*i+1][list][1] = h->luma_weight[i][list][1];
321 h->chroma_weight[16+2*i][list][j][0] = h->chroma_weight[16+2*i+1][list][j][0] = h->chroma_weight[i][list][j][0];
322 h->chroma_weight[16+2*i][list][j][1] = h->chroma_weight[16+2*i+1][list][j][1] = h->chroma_weight[i][list][j][1];
329 * Mark a picture as no longer needed for reference. The refmask
330 * argument allows unreferencing of individual fields or the whole frame.
331 * If the picture becomes entirely unreferenced, but is being held for
332 * display purposes, it is marked as such.
333 * @param refmask mask of fields to unreference; the mask is bitwise
334 * anded with the reference marking of pic
335 * @return non-zero if pic becomes entirely unreferenced (except possibly
336 * for display purposes) zero if one of the fields remains in
339 static inline int unreference_pic(H264Context *h, Picture *pic, int refmask){
341 if (pic->f.reference &= refmask) {
344 for(i = 0; h->delayed_pic[i]; i++)
345 if(pic == h->delayed_pic[i]){
346 pic->f.reference = DELAYED_PIC_REF;
354 * Find a Picture in the short term reference list by frame number.
355 * @param frame_num frame number to search for
356 * @param idx the index into h->short_ref where returned picture is found
357 * undefined if no picture found.
358 * @return pointer to the found picture, or NULL if no pic with the provided
359 * frame number is found
361 static Picture * find_short(H264Context *h, int frame_num, int *idx){
362 MpegEncContext * const s = &h->s;
365 for(i=0; i<h->short_ref_count; i++){
366 Picture *pic= h->short_ref[i];
367 if(s->avctx->debug&FF_DEBUG_MMCO)
368 av_log(h->s.avctx, AV_LOG_DEBUG, "%d %d %p\n", i, pic->frame_num, pic);
369 if(pic->frame_num == frame_num) {
378 * Remove a picture from the short term reference list by its index in
379 * that list. This does no checking on the provided index; it is assumed
380 * to be valid. Other list entries are shifted down.
381 * @param i index into h->short_ref of picture to remove.
383 static void remove_short_at_index(H264Context *h, int i){
384 assert(i >= 0 && i < h->short_ref_count);
385 h->short_ref[i]= NULL;
386 if (--h->short_ref_count)
387 memmove(&h->short_ref[i], &h->short_ref[i+1], (h->short_ref_count - i)*sizeof(Picture*));
392 * @return the removed picture or NULL if an error occurs
394 static Picture * remove_short(H264Context *h, int frame_num, int ref_mask){
395 MpegEncContext * const s = &h->s;
399 if(s->avctx->debug&FF_DEBUG_MMCO)
400 av_log(h->s.avctx, AV_LOG_DEBUG, "remove short %d count %d\n", frame_num, h->short_ref_count);
402 pic = find_short(h, frame_num, &i);
404 if(unreference_pic(h, pic, ref_mask))
405 remove_short_at_index(h, i);
412 * Remove a picture from the long term reference list by its index in
414 * @return the removed picture or NULL if an error occurs
416 static Picture * remove_long(H264Context *h, int i, int ref_mask){
421 if(unreference_pic(h, pic, ref_mask)){
422 assert(h->long_ref[i]->long_ref == 1);
423 h->long_ref[i]->long_ref= 0;
424 h->long_ref[i]= NULL;
432 void ff_h264_remove_all_refs(H264Context *h){
436 remove_long(h, i, 0);
438 assert(h->long_ref_count==0);
440 for(i=0; i<h->short_ref_count; i++){
441 unreference_pic(h, h->short_ref[i], 0);
442 h->short_ref[i]= NULL;
444 h->short_ref_count=0;
448 * print short term list
450 static void print_short_term(H264Context *h) {
452 if(h->s.avctx->debug&FF_DEBUG_MMCO) {
453 av_log(h->s.avctx, AV_LOG_DEBUG, "short term list:\n");
454 for(i=0; i<h->short_ref_count; i++){
455 Picture *pic= h->short_ref[i];
456 av_log(h->s.avctx, AV_LOG_DEBUG, "%d fn:%d poc:%d %p\n",
457 i, pic->frame_num, pic->poc, pic->f.data[0]);
463 * print long term list
465 static void print_long_term(H264Context *h) {
467 if(h->s.avctx->debug&FF_DEBUG_MMCO) {
468 av_log(h->s.avctx, AV_LOG_DEBUG, "long term list:\n");
469 for(i = 0; i < 16; i++){
470 Picture *pic= h->long_ref[i];
472 av_log(h->s.avctx, AV_LOG_DEBUG, "%d fn:%d poc:%d %p\n",
473 i, pic->frame_num, pic->poc, pic->f.data[0]);
479 void ff_generate_sliding_window_mmcos(H264Context *h) {
480 MpegEncContext * const s = &h->s;
481 assert(h->long_ref_count + h->short_ref_count <= h->sps.ref_frame_count);
484 if(h->short_ref_count && h->long_ref_count + h->short_ref_count == h->sps.ref_frame_count &&
485 !(FIELD_PICTURE && !s->first_field && s->current_picture_ptr->f.reference)) {
486 h->mmco[0].opcode= MMCO_SHORT2UNUSED;
487 h->mmco[0].short_pic_num= h->short_ref[ h->short_ref_count - 1 ]->frame_num;
490 h->mmco[0].short_pic_num *= 2;
491 h->mmco[1].opcode= MMCO_SHORT2UNUSED;
492 h->mmco[1].short_pic_num= h->mmco[0].short_pic_num + 1;
498 int ff_h264_execute_ref_pic_marking(H264Context *h, MMCO *mmco, int mmco_count){
499 MpegEncContext * const s = &h->s;
501 int current_ref_assigned=0, err=0;
502 Picture *av_uninit(pic);
504 if((s->avctx->debug&FF_DEBUG_MMCO) && mmco_count==0)
505 av_log(h->s.avctx, AV_LOG_DEBUG, "no mmco here\n");
507 for(i=0; i<mmco_count; i++){
508 int av_uninit(structure), av_uninit(frame_num);
509 if(s->avctx->debug&FF_DEBUG_MMCO)
510 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);
512 if( mmco[i].opcode == MMCO_SHORT2UNUSED
513 || mmco[i].opcode == MMCO_SHORT2LONG){
514 frame_num = pic_num_extract(h, mmco[i].short_pic_num, &structure);
515 pic = find_short(h, frame_num, &j);
517 if(mmco[i].opcode != MMCO_SHORT2LONG || !h->long_ref[mmco[i].long_arg]
518 || h->long_ref[mmco[i].long_arg]->frame_num != frame_num) {
519 av_log(h->s.avctx, AV_LOG_ERROR, "mmco: unref short failure\n");
520 err = AVERROR_INVALIDDATA;
526 switch(mmco[i].opcode){
527 case MMCO_SHORT2UNUSED:
528 if(s->avctx->debug&FF_DEBUG_MMCO)
529 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);
530 remove_short(h, frame_num, structure ^ PICT_FRAME);
532 case MMCO_SHORT2LONG:
533 if (h->long_ref[mmco[i].long_arg] != pic)
534 remove_long(h, mmco[i].long_arg, 0);
536 remove_short_at_index(h, j);
537 h->long_ref[ mmco[i].long_arg ]= pic;
538 if (h->long_ref[ mmco[i].long_arg ]){
539 h->long_ref[ mmco[i].long_arg ]->long_ref=1;
543 case MMCO_LONG2UNUSED:
544 j = pic_num_extract(h, mmco[i].long_arg, &structure);
545 pic = h->long_ref[j];
547 remove_long(h, j, structure ^ PICT_FRAME);
548 } else if(s->avctx->debug&FF_DEBUG_MMCO)
549 av_log(h->s.avctx, AV_LOG_DEBUG, "mmco: unref long failure\n");
552 // Comment below left from previous code as it is an interresting note.
553 /* First field in pair is in short term list or
554 * at a different long term index.
555 * This is not allowed; see 7.4.3.3, notes 2 and 3.
556 * Report the problem and keep the pair where it is,
557 * and mark this field valid.
560 if (h->long_ref[mmco[i].long_arg] != s->current_picture_ptr) {
561 remove_long(h, mmco[i].long_arg, 0);
563 h->long_ref[ mmco[i].long_arg ]= s->current_picture_ptr;
564 h->long_ref[ mmco[i].long_arg ]->long_ref=1;
568 s->current_picture_ptr->f.reference |= s->picture_structure;
569 current_ref_assigned=1;
571 case MMCO_SET_MAX_LONG:
572 assert(mmco[i].long_arg <= 16);
573 // just remove the long term which index is greater than new max
574 for(j = mmco[i].long_arg; j<16; j++){
575 remove_long(h, j, 0);
579 while(h->short_ref_count){
580 remove_short(h, h->short_ref[0]->frame_num, 0);
582 for(j = 0; j < 16; j++) {
583 remove_long(h, j, 0);
586 s->current_picture_ptr->frame_num= 0;
588 s->current_picture_ptr->mmco_reset=1;
594 if (!current_ref_assigned) {
595 /* Second field of complementary field pair; the first field of
596 * which is already referenced. If short referenced, it
597 * should be first entry in short_ref. If not, it must exist
598 * in long_ref; trying to put it on the short list here is an
599 * error in the encoded bit stream (ref: 7.4.3.3, NOTE 2 and 3).
601 if (h->short_ref_count && h->short_ref[0] == s->current_picture_ptr) {
602 /* Just mark the second field valid */
603 s->current_picture_ptr->f.reference = PICT_FRAME;
604 } else if (s->current_picture_ptr->long_ref) {
605 av_log(h->s.avctx, AV_LOG_ERROR, "illegal short term reference "
606 "assignment for second field "
607 "in complementary field pair "
608 "(first field is long term)\n");
609 err = AVERROR_INVALIDDATA;
611 pic= remove_short(h, s->current_picture_ptr->frame_num, 0);
613 av_log(h->s.avctx, AV_LOG_ERROR, "illegal short term buffer state detected\n");
614 err = AVERROR_INVALIDDATA;
617 if(h->short_ref_count)
618 memmove(&h->short_ref[1], &h->short_ref[0], h->short_ref_count*sizeof(Picture*));
620 h->short_ref[0]= s->current_picture_ptr;
621 h->short_ref_count++;
622 s->current_picture_ptr->f.reference |= s->picture_structure;
626 if (h->long_ref_count + h->short_ref_count -
627 (h->short_ref[0] == s->current_picture_ptr) > h->sps.ref_frame_count){
629 /* We have too many reference frames, probably due to corrupted
630 * stream. Need to discard one frame. Prevents overrun of the
631 * short_ref and long_ref buffers.
633 av_log(h->s.avctx, AV_LOG_ERROR,
634 "number of reference frames (%d+%d) exceeds max (%d; probably "
635 "corrupt input), discarding one\n",
636 h->long_ref_count, h->short_ref_count, h->sps.ref_frame_count);
637 err = AVERROR_INVALIDDATA;
639 if (h->long_ref_count && !h->short_ref_count) {
640 for (i = 0; i < 16; ++i)
645 remove_long(h, i, 0);
647 pic = h->short_ref[h->short_ref_count - 1];
648 remove_short(h, pic->frame_num, 0);
654 return (h->s.avctx->err_recognition & AV_EF_EXPLODE) ? err : 0;
657 int ff_h264_decode_ref_pic_marking(H264Context *h, GetBitContext *gb){
658 MpegEncContext * const s = &h->s;
662 if(h->nal_unit_type == NAL_IDR_SLICE){ //FIXME fields
663 s->broken_link= get_bits1(gb) -1;
665 h->mmco[0].opcode= MMCO_LONG;
666 h->mmco[0].long_arg= 0;
670 if(get_bits1(gb)){ // adaptive_ref_pic_marking_mode_flag
671 for(i= 0; i<MAX_MMCO_COUNT; i++) {
672 MMCOOpcode opcode= get_ue_golomb_31(gb);
674 h->mmco[i].opcode= opcode;
675 if(opcode==MMCO_SHORT2UNUSED || opcode==MMCO_SHORT2LONG){
676 h->mmco[i].short_pic_num= (h->curr_pic_num - get_ue_golomb(gb) - 1) & (h->max_pic_num - 1);
677 /* if(h->mmco[i].short_pic_num >= h->short_ref_count || h->short_ref[ h->mmco[i].short_pic_num ] == NULL){
678 av_log(s->avctx, AV_LOG_ERROR, "illegal short ref in memory management control operation %d\n", mmco);
682 if(opcode==MMCO_SHORT2LONG || opcode==MMCO_LONG2UNUSED || opcode==MMCO_LONG || opcode==MMCO_SET_MAX_LONG){
683 unsigned int long_arg= get_ue_golomb_31(gb);
684 if(long_arg >= 32 || (long_arg >= 16 && !(opcode == MMCO_SET_MAX_LONG && long_arg == 16) && !(opcode == MMCO_LONG2UNUSED && FIELD_PICTURE))){
685 av_log(h->s.avctx, AV_LOG_ERROR, "illegal long ref in memory management control operation %d\n", opcode);
688 h->mmco[i].long_arg= long_arg;
691 if(opcode > (unsigned)MMCO_LONG){
692 av_log(h->s.avctx, AV_LOG_ERROR, "illegal memory management control operation %d\n", opcode);
695 if(opcode == MMCO_END)
700 ff_generate_sliding_window_mmcos(h);