You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

1114 lines
37KB

  1. /*
  2. * H.26L/H.264/AVC/JVT/14496-10/... decoder
  3. * Copyright (c) 2003 Michael Niedermayer <michaelni@gmx.at>
  4. *
  5. * This file is part of FFmpeg.
  6. *
  7. * FFmpeg 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.
  11. *
  12. * FFmpeg 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.
  16. *
  17. * You should have received a copy of the GNU Lesser General Public
  18. * License along with FFmpeg; if not, write to the Free Software
  19. * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  20. */
  21. /**
  22. * @file
  23. * H.264 / AVC / MPEG-4 part10 codec.
  24. * @author Michael Niedermayer <michaelni@gmx.at>
  25. */
  26. #define UNCHECKED_BITSTREAM_READER 1
  27. #include "libavutil/avassert.h"
  28. #include "libavutil/display.h"
  29. #include "libavutil/imgutils.h"
  30. #include "libavutil/opt.h"
  31. #include "libavutil/stereo3d.h"
  32. #include "libavutil/video_enc_params.h"
  33. #include "internal.h"
  34. #include "bytestream.h"
  35. #include "cabac.h"
  36. #include "cabac_functions.h"
  37. #include "error_resilience.h"
  38. #include "avcodec.h"
  39. #include "h264.h"
  40. #include "h264dec.h"
  41. #include "h2645_parse.h"
  42. #include "h264data.h"
  43. #include "h264chroma.h"
  44. #include "h264_mvpred.h"
  45. #include "h264_ps.h"
  46. #include "golomb.h"
  47. #include "hwconfig.h"
  48. #include "mathops.h"
  49. #include "me_cmp.h"
  50. #include "mpegutils.h"
  51. #include "profiles.h"
  52. #include "rectangle.h"
  53. #include "thread.h"
  54. const uint16_t ff_h264_mb_sizes[4] = { 256, 384, 512, 768 };
  55. int avpriv_h264_has_num_reorder_frames(AVCodecContext *avctx)
  56. {
  57. H264Context *h = avctx->priv_data;
  58. return h && h->ps.sps ? h->ps.sps->num_reorder_frames : 0;
  59. }
  60. static void h264_er_decode_mb(void *opaque, int ref, int mv_dir, int mv_type,
  61. int (*mv)[2][4][2],
  62. int mb_x, int mb_y, int mb_intra, int mb_skipped)
  63. {
  64. H264Context *h = opaque;
  65. H264SliceContext *sl = &h->slice_ctx[0];
  66. sl->mb_x = mb_x;
  67. sl->mb_y = mb_y;
  68. sl->mb_xy = mb_x + mb_y * h->mb_stride;
  69. memset(sl->non_zero_count_cache, 0, sizeof(sl->non_zero_count_cache));
  70. av_assert1(ref >= 0);
  71. /* FIXME: It is possible albeit uncommon that slice references
  72. * differ between slices. We take the easy approach and ignore
  73. * it for now. If this turns out to have any relevance in
  74. * practice then correct remapping should be added. */
  75. if (ref >= sl->ref_count[0])
  76. ref = 0;
  77. if (!sl->ref_list[0][ref].data[0]) {
  78. av_log(h->avctx, AV_LOG_DEBUG, "Reference not available for error concealing\n");
  79. ref = 0;
  80. }
  81. if ((sl->ref_list[0][ref].reference&3) != 3) {
  82. av_log(h->avctx, AV_LOG_DEBUG, "Reference invalid\n");
  83. return;
  84. }
  85. fill_rectangle(&h->cur_pic.ref_index[0][4 * sl->mb_xy],
  86. 2, 2, 2, ref, 1);
  87. fill_rectangle(&sl->ref_cache[0][scan8[0]], 4, 4, 8, ref, 1);
  88. fill_rectangle(sl->mv_cache[0][scan8[0]], 4, 4, 8,
  89. pack16to32((*mv)[0][0][0], (*mv)[0][0][1]), 4);
  90. sl->mb_mbaff =
  91. sl->mb_field_decoding_flag = 0;
  92. ff_h264_hl_decode_mb(h, &h->slice_ctx[0]);
  93. }
  94. void ff_h264_draw_horiz_band(const H264Context *h, H264SliceContext *sl,
  95. int y, int height)
  96. {
  97. AVCodecContext *avctx = h->avctx;
  98. const AVFrame *src = h->cur_pic.f;
  99. const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(avctx->pix_fmt);
  100. int vshift = desc->log2_chroma_h;
  101. const int field_pic = h->picture_structure != PICT_FRAME;
  102. if (field_pic) {
  103. height <<= 1;
  104. y <<= 1;
  105. }
  106. height = FFMIN(height, avctx->height - y);
  107. if (field_pic && h->first_field && !(avctx->slice_flags & SLICE_FLAG_ALLOW_FIELD))
  108. return;
  109. if (avctx->draw_horiz_band) {
  110. int offset[AV_NUM_DATA_POINTERS];
  111. int i;
  112. offset[0] = y * src->linesize[0];
  113. offset[1] =
  114. offset[2] = (y >> vshift) * src->linesize[1];
  115. for (i = 3; i < AV_NUM_DATA_POINTERS; i++)
  116. offset[i] = 0;
  117. emms_c();
  118. avctx->draw_horiz_band(avctx, src, offset,
  119. y, h->picture_structure, height);
  120. }
  121. }
  122. void ff_h264_free_tables(H264Context *h)
  123. {
  124. int i;
  125. av_freep(&h->intra4x4_pred_mode);
  126. av_freep(&h->chroma_pred_mode_table);
  127. av_freep(&h->cbp_table);
  128. av_freep(&h->mvd_table[0]);
  129. av_freep(&h->mvd_table[1]);
  130. av_freep(&h->direct_table);
  131. av_freep(&h->non_zero_count);
  132. av_freep(&h->slice_table_base);
  133. h->slice_table = NULL;
  134. av_freep(&h->list_counts);
  135. av_freep(&h->mb2b_xy);
  136. av_freep(&h->mb2br_xy);
  137. av_buffer_pool_uninit(&h->qscale_table_pool);
  138. av_buffer_pool_uninit(&h->mb_type_pool);
  139. av_buffer_pool_uninit(&h->motion_val_pool);
  140. av_buffer_pool_uninit(&h->ref_index_pool);
  141. for (i = 0; i < h->nb_slice_ctx; i++) {
  142. H264SliceContext *sl = &h->slice_ctx[i];
  143. av_freep(&sl->dc_val_base);
  144. av_freep(&sl->er.mb_index2xy);
  145. av_freep(&sl->er.error_status_table);
  146. av_freep(&sl->er.er_temp_buffer);
  147. av_freep(&sl->bipred_scratchpad);
  148. av_freep(&sl->edge_emu_buffer);
  149. av_freep(&sl->top_borders[0]);
  150. av_freep(&sl->top_borders[1]);
  151. sl->bipred_scratchpad_allocated = 0;
  152. sl->edge_emu_buffer_allocated = 0;
  153. sl->top_borders_allocated[0] = 0;
  154. sl->top_borders_allocated[1] = 0;
  155. }
  156. }
  157. int ff_h264_alloc_tables(H264Context *h)
  158. {
  159. const int big_mb_num = h->mb_stride * (h->mb_height + 1);
  160. const int row_mb_num = 2*h->mb_stride*FFMAX(h->nb_slice_ctx, 1);
  161. int x, y;
  162. FF_ALLOCZ_ARRAY_OR_GOTO(h->avctx, h->intra4x4_pred_mode,
  163. row_mb_num, 8 * sizeof(uint8_t), fail)
  164. h->slice_ctx[0].intra4x4_pred_mode = h->intra4x4_pred_mode;
  165. FF_ALLOCZ_OR_GOTO(h->avctx, h->non_zero_count,
  166. big_mb_num * 48 * sizeof(uint8_t), fail)
  167. FF_ALLOCZ_OR_GOTO(h->avctx, h->slice_table_base,
  168. (big_mb_num + h->mb_stride) * sizeof(*h->slice_table_base), fail)
  169. FF_ALLOCZ_OR_GOTO(h->avctx, h->cbp_table,
  170. big_mb_num * sizeof(uint16_t), fail)
  171. FF_ALLOCZ_OR_GOTO(h->avctx, h->chroma_pred_mode_table,
  172. big_mb_num * sizeof(uint8_t), fail)
  173. FF_ALLOCZ_ARRAY_OR_GOTO(h->avctx, h->mvd_table[0],
  174. row_mb_num, 16 * sizeof(uint8_t), fail);
  175. FF_ALLOCZ_ARRAY_OR_GOTO(h->avctx, h->mvd_table[1],
  176. row_mb_num, 16 * sizeof(uint8_t), fail);
  177. h->slice_ctx[0].mvd_table[0] = h->mvd_table[0];
  178. h->slice_ctx[0].mvd_table[1] = h->mvd_table[1];
  179. FF_ALLOCZ_OR_GOTO(h->avctx, h->direct_table,
  180. 4 * big_mb_num * sizeof(uint8_t), fail);
  181. FF_ALLOCZ_OR_GOTO(h->avctx, h->list_counts,
  182. big_mb_num * sizeof(uint8_t), fail)
  183. memset(h->slice_table_base, -1,
  184. (big_mb_num + h->mb_stride) * sizeof(*h->slice_table_base));
  185. h->slice_table = h->slice_table_base + h->mb_stride * 2 + 1;
  186. FF_ALLOCZ_OR_GOTO(h->avctx, h->mb2b_xy,
  187. big_mb_num * sizeof(uint32_t), fail);
  188. FF_ALLOCZ_OR_GOTO(h->avctx, h->mb2br_xy,
  189. big_mb_num * sizeof(uint32_t), fail);
  190. for (y = 0; y < h->mb_height; y++)
  191. for (x = 0; x < h->mb_width; x++) {
  192. const int mb_xy = x + y * h->mb_stride;
  193. const int b_xy = 4 * x + 4 * y * h->b_stride;
  194. h->mb2b_xy[mb_xy] = b_xy;
  195. h->mb2br_xy[mb_xy] = 8 * (FMO ? mb_xy : (mb_xy % (2 * h->mb_stride)));
  196. }
  197. return 0;
  198. fail:
  199. return AVERROR(ENOMEM);
  200. }
  201. /**
  202. * Init context
  203. * Allocate buffers which are not shared amongst multiple threads.
  204. */
  205. int ff_h264_slice_context_init(H264Context *h, H264SliceContext *sl)
  206. {
  207. ERContext *er = &sl->er;
  208. int mb_array_size = h->mb_height * h->mb_stride;
  209. int y_size = (2 * h->mb_width + 1) * (2 * h->mb_height + 1);
  210. int c_size = h->mb_stride * (h->mb_height + 1);
  211. int yc_size = y_size + 2 * c_size;
  212. int x, y, i;
  213. sl->ref_cache[0][scan8[5] + 1] =
  214. sl->ref_cache[0][scan8[7] + 1] =
  215. sl->ref_cache[0][scan8[13] + 1] =
  216. sl->ref_cache[1][scan8[5] + 1] =
  217. sl->ref_cache[1][scan8[7] + 1] =
  218. sl->ref_cache[1][scan8[13] + 1] = PART_NOT_AVAILABLE;
  219. if (sl != h->slice_ctx) {
  220. memset(er, 0, sizeof(*er));
  221. } else
  222. if (CONFIG_ERROR_RESILIENCE) {
  223. /* init ER */
  224. er->avctx = h->avctx;
  225. er->decode_mb = h264_er_decode_mb;
  226. er->opaque = h;
  227. er->quarter_sample = 1;
  228. er->mb_num = h->mb_num;
  229. er->mb_width = h->mb_width;
  230. er->mb_height = h->mb_height;
  231. er->mb_stride = h->mb_stride;
  232. er->b8_stride = h->mb_width * 2 + 1;
  233. // error resilience code looks cleaner with this
  234. FF_ALLOCZ_OR_GOTO(h->avctx, er->mb_index2xy,
  235. (h->mb_num + 1) * sizeof(int), fail);
  236. for (y = 0; y < h->mb_height; y++)
  237. for (x = 0; x < h->mb_width; x++)
  238. er->mb_index2xy[x + y * h->mb_width] = x + y * h->mb_stride;
  239. er->mb_index2xy[h->mb_height * h->mb_width] = (h->mb_height - 1) *
  240. h->mb_stride + h->mb_width;
  241. FF_ALLOCZ_OR_GOTO(h->avctx, er->error_status_table,
  242. mb_array_size * sizeof(uint8_t), fail);
  243. FF_ALLOC_OR_GOTO(h->avctx, er->er_temp_buffer,
  244. h->mb_height * h->mb_stride * (4*sizeof(int) + 1), fail);
  245. FF_ALLOCZ_OR_GOTO(h->avctx, sl->dc_val_base,
  246. yc_size * sizeof(int16_t), fail);
  247. er->dc_val[0] = sl->dc_val_base + h->mb_width * 2 + 2;
  248. er->dc_val[1] = sl->dc_val_base + y_size + h->mb_stride + 1;
  249. er->dc_val[2] = er->dc_val[1] + c_size;
  250. for (i = 0; i < yc_size; i++)
  251. sl->dc_val_base[i] = 1024;
  252. }
  253. return 0;
  254. fail:
  255. return AVERROR(ENOMEM); // ff_h264_free_tables will clean up for us
  256. }
  257. static int h264_init_context(AVCodecContext *avctx, H264Context *h)
  258. {
  259. int i;
  260. h->avctx = avctx;
  261. h->cur_chroma_format_idc = -1;
  262. h->width_from_caller = avctx->width;
  263. h->height_from_caller = avctx->height;
  264. h->workaround_bugs = avctx->workaround_bugs;
  265. h->flags = avctx->flags;
  266. h->poc.prev_poc_msb = 1 << 16;
  267. h->recovery_frame = -1;
  268. h->frame_recovered = 0;
  269. h->poc.prev_frame_num = -1;
  270. h->sei.frame_packing.arrangement_cancel_flag = -1;
  271. h->sei.unregistered.x264_build = -1;
  272. h->next_outputed_poc = INT_MIN;
  273. for (i = 0; i < MAX_DELAYED_PIC_COUNT; i++)
  274. h->last_pocs[i] = INT_MIN;
  275. ff_h264_sei_uninit(&h->sei);
  276. h->nb_slice_ctx = (avctx->active_thread_type & FF_THREAD_SLICE) ? avctx->thread_count : 1;
  277. h->slice_ctx = av_mallocz_array(h->nb_slice_ctx, sizeof(*h->slice_ctx));
  278. if (!h->slice_ctx) {
  279. h->nb_slice_ctx = 0;
  280. return AVERROR(ENOMEM);
  281. }
  282. for (i = 0; i < H264_MAX_PICTURE_COUNT; i++) {
  283. h->DPB[i].f = av_frame_alloc();
  284. if (!h->DPB[i].f)
  285. return AVERROR(ENOMEM);
  286. }
  287. h->cur_pic.f = av_frame_alloc();
  288. if (!h->cur_pic.f)
  289. return AVERROR(ENOMEM);
  290. h->last_pic_for_ec.f = av_frame_alloc();
  291. if (!h->last_pic_for_ec.f)
  292. return AVERROR(ENOMEM);
  293. for (i = 0; i < h->nb_slice_ctx; i++)
  294. h->slice_ctx[i].h264 = h;
  295. return 0;
  296. }
  297. static av_cold int h264_decode_end(AVCodecContext *avctx)
  298. {
  299. H264Context *h = avctx->priv_data;
  300. int i;
  301. ff_h264_remove_all_refs(h);
  302. ff_h264_free_tables(h);
  303. for (i = 0; i < H264_MAX_PICTURE_COUNT; i++) {
  304. ff_h264_unref_picture(h, &h->DPB[i]);
  305. av_frame_free(&h->DPB[i].f);
  306. }
  307. memset(h->delayed_pic, 0, sizeof(h->delayed_pic));
  308. h->cur_pic_ptr = NULL;
  309. av_freep(&h->slice_ctx);
  310. h->nb_slice_ctx = 0;
  311. ff_h264_sei_uninit(&h->sei);
  312. ff_h264_ps_uninit(&h->ps);
  313. ff_h2645_packet_uninit(&h->pkt);
  314. ff_h264_unref_picture(h, &h->cur_pic);
  315. av_frame_free(&h->cur_pic.f);
  316. ff_h264_unref_picture(h, &h->last_pic_for_ec);
  317. av_frame_free(&h->last_pic_for_ec.f);
  318. return 0;
  319. }
  320. static AVOnce h264_vlc_init = AV_ONCE_INIT;
  321. static av_cold int h264_decode_init(AVCodecContext *avctx)
  322. {
  323. H264Context *h = avctx->priv_data;
  324. int ret;
  325. ret = h264_init_context(avctx, h);
  326. if (ret < 0)
  327. return ret;
  328. ret = ff_thread_once(&h264_vlc_init, ff_h264_decode_init_vlc);
  329. if (ret != 0) {
  330. av_log(avctx, AV_LOG_ERROR, "pthread_once has failed.");
  331. return AVERROR_UNKNOWN;
  332. }
  333. if (avctx->ticks_per_frame == 1) {
  334. if(h->avctx->time_base.den < INT_MAX/2) {
  335. h->avctx->time_base.den *= 2;
  336. } else
  337. h->avctx->time_base.num /= 2;
  338. }
  339. avctx->ticks_per_frame = 2;
  340. if (!avctx->internal->is_copy) {
  341. if (avctx->extradata_size > 0 && avctx->extradata) {
  342. ret = ff_h264_decode_extradata(avctx->extradata, avctx->extradata_size,
  343. &h->ps, &h->is_avc, &h->nal_length_size,
  344. avctx->err_recognition, avctx);
  345. if (ret < 0) {
  346. int explode = avctx->err_recognition & AV_EF_EXPLODE;
  347. av_log(avctx, explode ? AV_LOG_ERROR: AV_LOG_WARNING,
  348. "Error decoding the extradata\n");
  349. if (explode) {
  350. return ret;
  351. }
  352. ret = 0;
  353. }
  354. }
  355. }
  356. if (h->ps.sps && h->ps.sps->bitstream_restriction_flag &&
  357. h->avctx->has_b_frames < h->ps.sps->num_reorder_frames) {
  358. h->avctx->has_b_frames = h->ps.sps->num_reorder_frames;
  359. }
  360. ff_h264_flush_change(h);
  361. if (h->enable_er < 0 && (avctx->active_thread_type & FF_THREAD_SLICE))
  362. h->enable_er = 0;
  363. if (h->enable_er && (avctx->active_thread_type & FF_THREAD_SLICE)) {
  364. av_log(avctx, AV_LOG_WARNING,
  365. "Error resilience with slice threads is enabled. It is unsafe and unsupported and may crash. "
  366. "Use it at your own risk\n");
  367. }
  368. return 0;
  369. }
  370. /**
  371. * instantaneous decoder refresh.
  372. */
  373. static void idr(H264Context *h)
  374. {
  375. int i;
  376. ff_h264_remove_all_refs(h);
  377. h->poc.prev_frame_num =
  378. h->poc.prev_frame_num_offset = 0;
  379. h->poc.prev_poc_msb = 1<<16;
  380. h->poc.prev_poc_lsb = -1;
  381. for (i = 0; i < MAX_DELAYED_PIC_COUNT; i++)
  382. h->last_pocs[i] = INT_MIN;
  383. }
  384. /* forget old pics after a seek */
  385. void ff_h264_flush_change(H264Context *h)
  386. {
  387. int i, j;
  388. h->next_outputed_poc = INT_MIN;
  389. h->prev_interlaced_frame = 1;
  390. idr(h);
  391. h->poc.prev_frame_num = -1;
  392. if (h->cur_pic_ptr) {
  393. h->cur_pic_ptr->reference = 0;
  394. for (j=i=0; h->delayed_pic[i]; i++)
  395. if (h->delayed_pic[i] != h->cur_pic_ptr)
  396. h->delayed_pic[j++] = h->delayed_pic[i];
  397. h->delayed_pic[j] = NULL;
  398. }
  399. ff_h264_unref_picture(h, &h->last_pic_for_ec);
  400. h->first_field = 0;
  401. h->recovery_frame = -1;
  402. h->frame_recovered = 0;
  403. h->current_slice = 0;
  404. h->mmco_reset = 1;
  405. }
  406. static void h264_decode_flush(AVCodecContext *avctx)
  407. {
  408. H264Context *h = avctx->priv_data;
  409. int i;
  410. memset(h->delayed_pic, 0, sizeof(h->delayed_pic));
  411. ff_h264_flush_change(h);
  412. ff_h264_sei_uninit(&h->sei);
  413. for (i = 0; i < H264_MAX_PICTURE_COUNT; i++)
  414. ff_h264_unref_picture(h, &h->DPB[i]);
  415. h->cur_pic_ptr = NULL;
  416. ff_h264_unref_picture(h, &h->cur_pic);
  417. h->mb_y = 0;
  418. ff_h264_free_tables(h);
  419. h->context_initialized = 0;
  420. }
  421. static int get_last_needed_nal(H264Context *h)
  422. {
  423. int nals_needed = 0;
  424. int first_slice = 0;
  425. int i, ret;
  426. for (i = 0; i < h->pkt.nb_nals; i++) {
  427. H2645NAL *nal = &h->pkt.nals[i];
  428. GetBitContext gb;
  429. /* packets can sometimes contain multiple PPS/SPS,
  430. * e.g. two PAFF field pictures in one packet, or a demuxer
  431. * which splits NALs strangely if so, when frame threading we
  432. * can't start the next thread until we've read all of them */
  433. switch (nal->type) {
  434. case H264_NAL_SPS:
  435. case H264_NAL_PPS:
  436. nals_needed = i;
  437. break;
  438. case H264_NAL_DPA:
  439. case H264_NAL_IDR_SLICE:
  440. case H264_NAL_SLICE:
  441. ret = init_get_bits8(&gb, nal->data + 1, nal->size - 1);
  442. if (ret < 0) {
  443. av_log(h->avctx, AV_LOG_ERROR, "Invalid zero-sized VCL NAL unit\n");
  444. if (h->avctx->err_recognition & AV_EF_EXPLODE)
  445. return ret;
  446. break;
  447. }
  448. if (!get_ue_golomb_long(&gb) || // first_mb_in_slice
  449. !first_slice ||
  450. first_slice != nal->type)
  451. nals_needed = i;
  452. if (!first_slice)
  453. first_slice = nal->type;
  454. }
  455. }
  456. return nals_needed;
  457. }
  458. static void debug_green_metadata(const H264SEIGreenMetaData *gm, void *logctx)
  459. {
  460. av_log(logctx, AV_LOG_DEBUG, "Green Metadata Info SEI message\n");
  461. av_log(logctx, AV_LOG_DEBUG, " green_metadata_type: %d\n", gm->green_metadata_type);
  462. if (gm->green_metadata_type == 0) {
  463. av_log(logctx, AV_LOG_DEBUG, " green_metadata_period_type: %d\n", gm->period_type);
  464. if (gm->period_type == 2)
  465. av_log(logctx, AV_LOG_DEBUG, " green_metadata_num_seconds: %d\n", gm->num_seconds);
  466. else if (gm->period_type == 3)
  467. av_log(logctx, AV_LOG_DEBUG, " green_metadata_num_pictures: %d\n", gm->num_pictures);
  468. av_log(logctx, AV_LOG_DEBUG, " SEI GREEN Complexity Metrics: %f %f %f %f\n",
  469. (float)gm->percent_non_zero_macroblocks/255,
  470. (float)gm->percent_intra_coded_macroblocks/255,
  471. (float)gm->percent_six_tap_filtering/255,
  472. (float)gm->percent_alpha_point_deblocking_instance/255);
  473. } else if (gm->green_metadata_type == 1) {
  474. av_log(logctx, AV_LOG_DEBUG, " xsd_metric_type: %d\n", gm->xsd_metric_type);
  475. if (gm->xsd_metric_type == 0)
  476. av_log(logctx, AV_LOG_DEBUG, " xsd_metric_value: %f\n",
  477. (float)gm->xsd_metric_value/100);
  478. }
  479. }
  480. static int decode_nal_units(H264Context *h, const uint8_t *buf, int buf_size)
  481. {
  482. AVCodecContext *const avctx = h->avctx;
  483. int nals_needed = 0; ///< number of NALs that need decoding before the next frame thread starts
  484. int idr_cleared=0;
  485. int i, ret = 0;
  486. h->has_slice = 0;
  487. h->nal_unit_type= 0;
  488. if (!(avctx->flags2 & AV_CODEC_FLAG2_CHUNKS)) {
  489. h->current_slice = 0;
  490. if (!h->first_field) {
  491. h->cur_pic_ptr = NULL;
  492. ff_h264_sei_uninit(&h->sei);
  493. }
  494. }
  495. if (h->nal_length_size == 4) {
  496. if (buf_size > 8 && AV_RB32(buf) == 1 && AV_RB32(buf+5) > (unsigned)buf_size) {
  497. h->is_avc = 0;
  498. }else if(buf_size > 3 && AV_RB32(buf) > 1 && AV_RB32(buf) <= (unsigned)buf_size)
  499. h->is_avc = 1;
  500. }
  501. ret = ff_h2645_packet_split(&h->pkt, buf, buf_size, avctx, h->is_avc, h->nal_length_size,
  502. avctx->codec_id, 0, 0);
  503. if (ret < 0) {
  504. av_log(avctx, AV_LOG_ERROR,
  505. "Error splitting the input into NAL units.\n");
  506. return ret;
  507. }
  508. if (avctx->active_thread_type & FF_THREAD_FRAME)
  509. nals_needed = get_last_needed_nal(h);
  510. if (nals_needed < 0)
  511. return nals_needed;
  512. for (i = 0; i < h->pkt.nb_nals; i++) {
  513. H2645NAL *nal = &h->pkt.nals[i];
  514. int max_slice_ctx, err;
  515. if (avctx->skip_frame >= AVDISCARD_NONREF &&
  516. nal->ref_idc == 0 && nal->type != H264_NAL_SEI)
  517. continue;
  518. // FIXME these should stop being context-global variables
  519. h->nal_ref_idc = nal->ref_idc;
  520. h->nal_unit_type = nal->type;
  521. err = 0;
  522. switch (nal->type) {
  523. case H264_NAL_IDR_SLICE:
  524. if ((nal->data[1] & 0xFC) == 0x98) {
  525. av_log(h->avctx, AV_LOG_ERROR, "Invalid inter IDR frame\n");
  526. h->next_outputed_poc = INT_MIN;
  527. ret = -1;
  528. goto end;
  529. }
  530. if(!idr_cleared) {
  531. idr(h); // FIXME ensure we don't lose some frames if there is reordering
  532. }
  533. idr_cleared = 1;
  534. h->has_recovery_point = 1;
  535. case H264_NAL_SLICE:
  536. h->has_slice = 1;
  537. if ((err = ff_h264_queue_decode_slice(h, nal))) {
  538. H264SliceContext *sl = h->slice_ctx + h->nb_slice_ctx_queued;
  539. sl->ref_count[0] = sl->ref_count[1] = 0;
  540. break;
  541. }
  542. if (h->current_slice == 1) {
  543. if (avctx->active_thread_type & FF_THREAD_FRAME &&
  544. i >= nals_needed && !h->setup_finished && h->cur_pic_ptr) {
  545. ff_thread_finish_setup(avctx);
  546. h->setup_finished = 1;
  547. }
  548. if (h->avctx->hwaccel &&
  549. (ret = h->avctx->hwaccel->start_frame(h->avctx, buf, buf_size)) < 0)
  550. goto end;
  551. }
  552. max_slice_ctx = avctx->hwaccel ? 1 : h->nb_slice_ctx;
  553. if (h->nb_slice_ctx_queued == max_slice_ctx) {
  554. if (h->avctx->hwaccel) {
  555. ret = avctx->hwaccel->decode_slice(avctx, nal->raw_data, nal->raw_size);
  556. h->nb_slice_ctx_queued = 0;
  557. } else
  558. ret = ff_h264_execute_decode_slices(h);
  559. if (ret < 0 && (h->avctx->err_recognition & AV_EF_EXPLODE))
  560. goto end;
  561. }
  562. break;
  563. case H264_NAL_DPA:
  564. case H264_NAL_DPB:
  565. case H264_NAL_DPC:
  566. avpriv_request_sample(avctx, "data partitioning");
  567. break;
  568. case H264_NAL_SEI:
  569. ret = ff_h264_sei_decode(&h->sei, &nal->gb, &h->ps, avctx);
  570. h->has_recovery_point = h->has_recovery_point || h->sei.recovery_point.recovery_frame_cnt != -1;
  571. if (avctx->debug & FF_DEBUG_GREEN_MD)
  572. debug_green_metadata(&h->sei.green_metadata, h->avctx);
  573. if (ret < 0 && (h->avctx->err_recognition & AV_EF_EXPLODE))
  574. goto end;
  575. break;
  576. case H264_NAL_SPS: {
  577. GetBitContext tmp_gb = nal->gb;
  578. if (avctx->hwaccel && avctx->hwaccel->decode_params) {
  579. ret = avctx->hwaccel->decode_params(avctx,
  580. nal->type,
  581. nal->raw_data,
  582. nal->raw_size);
  583. if (ret < 0)
  584. goto end;
  585. }
  586. if (ff_h264_decode_seq_parameter_set(&tmp_gb, avctx, &h->ps, 0) >= 0)
  587. break;
  588. av_log(h->avctx, AV_LOG_DEBUG,
  589. "SPS decoding failure, trying again with the complete NAL\n");
  590. init_get_bits8(&tmp_gb, nal->raw_data + 1, nal->raw_size - 1);
  591. if (ff_h264_decode_seq_parameter_set(&tmp_gb, avctx, &h->ps, 0) >= 0)
  592. break;
  593. ff_h264_decode_seq_parameter_set(&nal->gb, avctx, &h->ps, 1);
  594. break;
  595. }
  596. case H264_NAL_PPS:
  597. if (avctx->hwaccel && avctx->hwaccel->decode_params) {
  598. ret = avctx->hwaccel->decode_params(avctx,
  599. nal->type,
  600. nal->raw_data,
  601. nal->raw_size);
  602. if (ret < 0)
  603. goto end;
  604. }
  605. ret = ff_h264_decode_picture_parameter_set(&nal->gb, avctx, &h->ps,
  606. nal->size_bits);
  607. if (ret < 0 && (h->avctx->err_recognition & AV_EF_EXPLODE))
  608. goto end;
  609. break;
  610. case H264_NAL_AUD:
  611. case H264_NAL_END_SEQUENCE:
  612. case H264_NAL_END_STREAM:
  613. case H264_NAL_FILLER_DATA:
  614. case H264_NAL_SPS_EXT:
  615. case H264_NAL_AUXILIARY_SLICE:
  616. break;
  617. default:
  618. av_log(avctx, AV_LOG_DEBUG, "Unknown NAL code: %d (%d bits)\n",
  619. nal->type, nal->size_bits);
  620. }
  621. if (err < 0) {
  622. av_log(h->avctx, AV_LOG_ERROR, "decode_slice_header error\n");
  623. }
  624. }
  625. ret = ff_h264_execute_decode_slices(h);
  626. if (ret < 0 && (h->avctx->err_recognition & AV_EF_EXPLODE))
  627. goto end;
  628. // set decode_error_flags to allow users to detect concealed decoding errors
  629. if ((ret < 0 || h->slice_ctx->er.error_occurred) && h->cur_pic_ptr) {
  630. h->cur_pic_ptr->f->decode_error_flags |= FF_DECODE_ERROR_DECODE_SLICES;
  631. }
  632. ret = 0;
  633. end:
  634. #if CONFIG_ERROR_RESILIENCE
  635. /*
  636. * FIXME: Error handling code does not seem to support interlaced
  637. * when slices span multiple rows
  638. * The ff_er_add_slice calls don't work right for bottom
  639. * fields; they cause massive erroneous error concealing
  640. * Error marking covers both fields (top and bottom).
  641. * This causes a mismatched s->error_count
  642. * and a bad error table. Further, the error count goes to
  643. * INT_MAX when called for bottom field, because mb_y is
  644. * past end by one (callers fault) and resync_mb_y != 0
  645. * causes problems for the first MB line, too.
  646. */
  647. if (!FIELD_PICTURE(h) && h->current_slice && h->enable_er) {
  648. H264SliceContext *sl = h->slice_ctx;
  649. int use_last_pic = h->last_pic_for_ec.f->buf[0] && !sl->ref_count[0];
  650. ff_h264_set_erpic(&sl->er.cur_pic, h->cur_pic_ptr);
  651. if (use_last_pic) {
  652. ff_h264_set_erpic(&sl->er.last_pic, &h->last_pic_for_ec);
  653. sl->ref_list[0][0].parent = &h->last_pic_for_ec;
  654. memcpy(sl->ref_list[0][0].data, h->last_pic_for_ec.f->data, sizeof(sl->ref_list[0][0].data));
  655. memcpy(sl->ref_list[0][0].linesize, h->last_pic_for_ec.f->linesize, sizeof(sl->ref_list[0][0].linesize));
  656. sl->ref_list[0][0].reference = h->last_pic_for_ec.reference;
  657. } else if (sl->ref_count[0]) {
  658. ff_h264_set_erpic(&sl->er.last_pic, sl->ref_list[0][0].parent);
  659. } else
  660. ff_h264_set_erpic(&sl->er.last_pic, NULL);
  661. if (sl->ref_count[1])
  662. ff_h264_set_erpic(&sl->er.next_pic, sl->ref_list[1][0].parent);
  663. sl->er.ref_count = sl->ref_count[0];
  664. ff_er_frame_end(&sl->er);
  665. if (use_last_pic)
  666. memset(&sl->ref_list[0][0], 0, sizeof(sl->ref_list[0][0]));
  667. }
  668. #endif /* CONFIG_ERROR_RESILIENCE */
  669. /* clean up */
  670. if (h->cur_pic_ptr && !h->droppable && h->has_slice) {
  671. ff_thread_report_progress(&h->cur_pic_ptr->tf, INT_MAX,
  672. h->picture_structure == PICT_BOTTOM_FIELD);
  673. }
  674. return (ret < 0) ? ret : buf_size;
  675. }
  676. /**
  677. * Return the number of bytes consumed for building the current frame.
  678. */
  679. static int get_consumed_bytes(int pos, int buf_size)
  680. {
  681. if (pos == 0)
  682. pos = 1; // avoid infinite loops (I doubt that is needed but...)
  683. if (pos + 10 > buf_size)
  684. pos = buf_size; // oops ;)
  685. return pos;
  686. }
  687. static int h264_export_enc_params(AVFrame *f, H264Picture *p)
  688. {
  689. AVVideoEncParams *par;
  690. unsigned int nb_mb = p->mb_height * p->mb_width;
  691. unsigned int x, y;
  692. par = av_video_enc_params_create_side_data(f, AV_VIDEO_ENC_PARAMS_H264, nb_mb);
  693. if (!par)
  694. return AVERROR(ENOMEM);
  695. par->qp = p->pps->init_qp;
  696. par->delta_qp[1][0] = p->pps->chroma_qp_index_offset[0];
  697. par->delta_qp[1][1] = p->pps->chroma_qp_index_offset[0];
  698. par->delta_qp[2][0] = p->pps->chroma_qp_index_offset[1];
  699. par->delta_qp[2][1] = p->pps->chroma_qp_index_offset[1];
  700. for (y = 0; y < p->mb_height; y++)
  701. for (x = 0; x < p->mb_width; x++) {
  702. const unsigned int block_idx = y * p->mb_width + x;
  703. const unsigned int mb_xy = y * p->mb_stride + x;
  704. AVVideoBlockParams *b = av_video_enc_params_block(par, block_idx);
  705. b->src_x = x * 16;
  706. b->src_y = y * 16;
  707. b->w = 16;
  708. b->h = 16;
  709. b->delta_qp = p->qscale_table[mb_xy] - par->qp;
  710. }
  711. return 0;
  712. }
  713. static int output_frame(H264Context *h, AVFrame *dst, H264Picture *srcp)
  714. {
  715. AVFrame *src = srcp->f;
  716. int ret;
  717. ret = av_frame_ref(dst, src);
  718. if (ret < 0)
  719. return ret;
  720. av_dict_set(&dst->metadata, "stereo_mode", ff_h264_sei_stereo_mode(&h->sei.frame_packing), 0);
  721. if (srcp->sei_recovery_frame_cnt == 0)
  722. dst->key_frame = 1;
  723. if (h->avctx->export_side_data & AV_CODEC_EXPORT_DATA_VIDEO_ENC_PARAMS) {
  724. ret = h264_export_enc_params(dst, srcp);
  725. if (ret < 0)
  726. goto fail;
  727. }
  728. return 0;
  729. fail:
  730. av_frame_unref(dst);
  731. return ret;
  732. }
  733. static int is_extra(const uint8_t *buf, int buf_size)
  734. {
  735. int cnt= buf[5]&0x1f;
  736. const uint8_t *p= buf+6;
  737. if (!cnt)
  738. return 0;
  739. while(cnt--){
  740. int nalsize= AV_RB16(p) + 2;
  741. if(nalsize > buf_size - (p-buf) || (p[2] & 0x9F) != 7)
  742. return 0;
  743. p += nalsize;
  744. }
  745. cnt = *(p++);
  746. if(!cnt)
  747. return 0;
  748. while(cnt--){
  749. int nalsize= AV_RB16(p) + 2;
  750. if(nalsize > buf_size - (p-buf) || (p[2] & 0x9F) != 8)
  751. return 0;
  752. p += nalsize;
  753. }
  754. return 1;
  755. }
  756. static int finalize_frame(H264Context *h, AVFrame *dst, H264Picture *out, int *got_frame)
  757. {
  758. int ret;
  759. if (((h->avctx->flags & AV_CODEC_FLAG_OUTPUT_CORRUPT) ||
  760. (h->avctx->flags2 & AV_CODEC_FLAG2_SHOW_ALL) ||
  761. out->recovered)) {
  762. if (!h->avctx->hwaccel &&
  763. (out->field_poc[0] == INT_MAX ||
  764. out->field_poc[1] == INT_MAX)
  765. ) {
  766. int p;
  767. AVFrame *f = out->f;
  768. int field = out->field_poc[0] == INT_MAX;
  769. uint8_t *dst_data[4];
  770. int linesizes[4];
  771. const uint8_t *src_data[4];
  772. av_log(h->avctx, AV_LOG_DEBUG, "Duplicating field %d to fill missing\n", field);
  773. for (p = 0; p<4; p++) {
  774. dst_data[p] = f->data[p] + (field^1)*f->linesize[p];
  775. src_data[p] = f->data[p] + field *f->linesize[p];
  776. linesizes[p] = 2*f->linesize[p];
  777. }
  778. av_image_copy(dst_data, linesizes, src_data, linesizes,
  779. f->format, f->width, f->height>>1);
  780. }
  781. ret = output_frame(h, dst, out);
  782. if (ret < 0)
  783. return ret;
  784. *got_frame = 1;
  785. if (CONFIG_MPEGVIDEO) {
  786. ff_print_debug_info2(h->avctx, dst, NULL,
  787. out->mb_type,
  788. out->qscale_table,
  789. out->motion_val,
  790. NULL,
  791. h->mb_width, h->mb_height, h->mb_stride, 1);
  792. }
  793. }
  794. return 0;
  795. }
  796. static int send_next_delayed_frame(H264Context *h, AVFrame *dst_frame,
  797. int *got_frame, int buf_index)
  798. {
  799. int ret, i, out_idx;
  800. H264Picture *out = h->delayed_pic[0];
  801. h->cur_pic_ptr = NULL;
  802. h->first_field = 0;
  803. out_idx = 0;
  804. for (i = 1;
  805. h->delayed_pic[i] &&
  806. !h->delayed_pic[i]->f->key_frame &&
  807. !h->delayed_pic[i]->mmco_reset;
  808. i++)
  809. if (h->delayed_pic[i]->poc < out->poc) {
  810. out = h->delayed_pic[i];
  811. out_idx = i;
  812. }
  813. for (i = out_idx; h->delayed_pic[i]; i++)
  814. h->delayed_pic[i] = h->delayed_pic[i + 1];
  815. if (out) {
  816. out->reference &= ~DELAYED_PIC_REF;
  817. ret = finalize_frame(h, dst_frame, out, got_frame);
  818. if (ret < 0)
  819. return ret;
  820. }
  821. return buf_index;
  822. }
  823. static int h264_decode_frame(AVCodecContext *avctx, void *data,
  824. int *got_frame, AVPacket *avpkt)
  825. {
  826. const uint8_t *buf = avpkt->data;
  827. int buf_size = avpkt->size;
  828. H264Context *h = avctx->priv_data;
  829. AVFrame *pict = data;
  830. int buf_index;
  831. int ret;
  832. h->flags = avctx->flags;
  833. h->setup_finished = 0;
  834. h->nb_slice_ctx_queued = 0;
  835. ff_h264_unref_picture(h, &h->last_pic_for_ec);
  836. /* end of stream, output what is still in the buffers */
  837. if (buf_size == 0)
  838. return send_next_delayed_frame(h, pict, got_frame, 0);
  839. if (h->is_avc && av_packet_get_side_data(avpkt, AV_PKT_DATA_NEW_EXTRADATA, NULL)) {
  840. int side_size;
  841. uint8_t *side = av_packet_get_side_data(avpkt, AV_PKT_DATA_NEW_EXTRADATA, &side_size);
  842. if (is_extra(side, side_size))
  843. ff_h264_decode_extradata(side, side_size,
  844. &h->ps, &h->is_avc, &h->nal_length_size,
  845. avctx->err_recognition, avctx);
  846. }
  847. if (h->is_avc && buf_size >= 9 && buf[0]==1 && buf[2]==0 && (buf[4]&0xFC)==0xFC) {
  848. if (is_extra(buf, buf_size))
  849. return ff_h264_decode_extradata(buf, buf_size,
  850. &h->ps, &h->is_avc, &h->nal_length_size,
  851. avctx->err_recognition, avctx);
  852. }
  853. buf_index = decode_nal_units(h, buf, buf_size);
  854. if (buf_index < 0)
  855. return AVERROR_INVALIDDATA;
  856. if (!h->cur_pic_ptr && h->nal_unit_type == H264_NAL_END_SEQUENCE) {
  857. av_assert0(buf_index <= buf_size);
  858. return send_next_delayed_frame(h, pict, got_frame, buf_index);
  859. }
  860. if (!(avctx->flags2 & AV_CODEC_FLAG2_CHUNKS) && (!h->cur_pic_ptr || !h->has_slice)) {
  861. if (avctx->skip_frame >= AVDISCARD_NONREF ||
  862. buf_size >= 4 && !memcmp("Q264", buf, 4))
  863. return buf_size;
  864. av_log(avctx, AV_LOG_ERROR, "no frame!\n");
  865. return AVERROR_INVALIDDATA;
  866. }
  867. if (!(avctx->flags2 & AV_CODEC_FLAG2_CHUNKS) ||
  868. (h->mb_y >= h->mb_height && h->mb_height)) {
  869. if ((ret = ff_h264_field_end(h, &h->slice_ctx[0], 0)) < 0)
  870. return ret;
  871. /* Wait for second field. */
  872. if (h->next_output_pic) {
  873. ret = finalize_frame(h, pict, h->next_output_pic, got_frame);
  874. if (ret < 0)
  875. return ret;
  876. }
  877. }
  878. av_assert0(pict->buf[0] || !*got_frame);
  879. ff_h264_unref_picture(h, &h->last_pic_for_ec);
  880. return get_consumed_bytes(buf_index, buf_size);
  881. }
  882. #define OFFSET(x) offsetof(H264Context, x)
  883. #define VD AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_DECODING_PARAM
  884. static const AVOption h264_options[] = {
  885. { "is_avc", "is avc", OFFSET(is_avc), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, 0 },
  886. { "nal_length_size", "nal_length_size", OFFSET(nal_length_size), AV_OPT_TYPE_INT, {.i64 = 0}, 0, 4, 0 },
  887. { "enable_er", "Enable error resilience on damaged frames (unsafe)", OFFSET(enable_er), AV_OPT_TYPE_BOOL, { .i64 = -1 }, -1, 1, VD },
  888. { "x264_build", "Assume this x264 version if no x264 version found in any SEI", OFFSET(x264_build), AV_OPT_TYPE_INT, {.i64 = -1}, -1, INT_MAX, VD },
  889. { NULL },
  890. };
  891. static const AVClass h264_class = {
  892. .class_name = "H264 Decoder",
  893. .item_name = av_default_item_name,
  894. .option = h264_options,
  895. .version = LIBAVUTIL_VERSION_INT,
  896. };
  897. AVCodec ff_h264_decoder = {
  898. .name = "h264",
  899. .long_name = NULL_IF_CONFIG_SMALL("H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10"),
  900. .type = AVMEDIA_TYPE_VIDEO,
  901. .id = AV_CODEC_ID_H264,
  902. .priv_data_size = sizeof(H264Context),
  903. .init = h264_decode_init,
  904. .close = h264_decode_end,
  905. .decode = h264_decode_frame,
  906. .capabilities = /*AV_CODEC_CAP_DRAW_HORIZ_BAND |*/ AV_CODEC_CAP_DR1 |
  907. AV_CODEC_CAP_DELAY | AV_CODEC_CAP_SLICE_THREADS |
  908. AV_CODEC_CAP_FRAME_THREADS,
  909. .hw_configs = (const AVCodecHWConfigInternal*[]) {
  910. #if CONFIG_H264_DXVA2_HWACCEL
  911. HWACCEL_DXVA2(h264),
  912. #endif
  913. #if CONFIG_H264_D3D11VA_HWACCEL
  914. HWACCEL_D3D11VA(h264),
  915. #endif
  916. #if CONFIG_H264_D3D11VA2_HWACCEL
  917. HWACCEL_D3D11VA2(h264),
  918. #endif
  919. #if CONFIG_H264_NVDEC_HWACCEL
  920. HWACCEL_NVDEC(h264),
  921. #endif
  922. #if CONFIG_H264_VAAPI_HWACCEL
  923. HWACCEL_VAAPI(h264),
  924. #endif
  925. #if CONFIG_H264_VDPAU_HWACCEL
  926. HWACCEL_VDPAU(h264),
  927. #endif
  928. #if CONFIG_H264_VIDEOTOOLBOX_HWACCEL
  929. HWACCEL_VIDEOTOOLBOX(h264),
  930. #endif
  931. NULL
  932. },
  933. .caps_internal = FF_CODEC_CAP_INIT_THREADSAFE | FF_CODEC_CAP_EXPORTS_CROPPING |
  934. FF_CODEC_CAP_ALLOCATE_PROGRESS | FF_CODEC_CAP_INIT_CLEANUP,
  935. .flush = h264_decode_flush,
  936. .update_thread_context = ONLY_IF_THREADS_ENABLED(ff_h264_update_thread_context),
  937. .profiles = NULL_IF_CONFIG_SMALL(ff_h264_profiles),
  938. .priv_class = &h264_class,
  939. };