diff options
Diffstat (limited to 'libavcodec/h264.c')
| -rw-r--r-- | libavcodec/h264.c | 613 |
1 files changed, 391 insertions, 222 deletions
diff --git a/libavcodec/h264.c b/libavcodec/h264.c index 1a0b341..3bd840d 100644 --- a/libavcodec/h264.c +++ b/libavcodec/h264.c @@ -2,20 +2,20 @@ * H.26L/H.264/AVC/JVT/14496-10/... decoder * Copyright (c) 2003 Michael Niedermayer <michaelni@gmx.at> * - * This file is part of Libav. + * This file is part of FFmpeg. * - * Libav is free software; you can redistribute it and/or + * FFmpeg is free software; you can redistribute it and/or * modify it under the terms of the GNU Lesser General Public * License as published by the Free Software Foundation; either * version 2.1 of the License, or (at your option) any later version. * - * Libav is distributed in the hope that it will be useful, + * FFmpeg is distributed in the hope that it will be useful, * but WITHOUT ANY WARRANTY; without even the implied warranty of * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU * Lesser General Public License for more details. * * You should have received a copy of the GNU Lesser General Public - * License along with Libav; if not, write to the Free Software + * License along with FFmpeg; if not, write to the Free Software * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA */ @@ -25,7 +25,10 @@ * @author Michael Niedermayer <michaelni@gmx.at> */ +#define UNCHECKED_BITSTREAM_READER 1 + #include "libavutil/imgutils.h" +#include "libavutil/opt.h" #include "internal.h" #include "cabac.h" #include "cabac_functions.h" @@ -50,13 +53,17 @@ const uint16_t ff_h264_mb_sizes[4] = { 256, 384, 512, 768 }; static const uint8_t rem6[QP_MAX_NUM + 1] = { 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, - 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, + 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, + 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, 0, 1, 2, 3, 4, 5, + 0, 1, 2, 3, }; static const uint8_t div6[QP_MAX_NUM + 1] = { 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 4, 4, 4, 4, 4, 4, 5, 5, 5, 5, 5, 5, 6, 6, 6, 6, 6, 6, - 7, 7, 7, 7, 7, 7, 8, 8, 8, 8, 8, 8, 9, 9, 9, 9, 9, 9, 10, 10, 10, 10, + 7, 7, 7, 7, 7, 7, 8, 8, 8, 8, 8, 8, 9, 9, 9, 9, 9, 9, 10, 10, 10, + 10,10,10,11,11,11,11,11,11,12,12,12,12,12,12,13,13,13, 13, 13, 13, + 14,14,14,14, }; static const enum AVPixelFormat hwaccel_pixfmt_list_h264_jpeg_420[] = { @@ -67,6 +74,12 @@ static const enum AVPixelFormat hwaccel_pixfmt_list_h264_jpeg_420[] = { AV_PIX_FMT_NONE }; +int avpriv_h264_has_num_reorder_frames(AVCodecContext *avctx) +{ + H264Context *h = avctx->priv_data; + return h ? h->sps.num_reorder_frames : 0; +} + /** * Check if the top & left blocks are available if needed and * change the dc mode so it only uses the available blocks. @@ -220,21 +233,27 @@ const uint8_t *ff_h264_decode_nal(H264Context *h, const uint8_t *src, } #endif - if (i >= length - 1) { // no escaped 0 - *dst_length = length; - *consumed = length + 1; // +1 for the header - return src; - } - // use second escape buffer for inter data bufidx = h->nal_unit_type == NAL_DPC ? 1 : 0; - av_fast_malloc(&h->rbsp_buffer[bufidx], &h->rbsp_buffer_size[bufidx], - length + FF_INPUT_BUFFER_PADDING_SIZE); + + si = h->rbsp_buffer_size[bufidx]; + av_fast_padded_malloc(&h->rbsp_buffer[bufidx], &h->rbsp_buffer_size[bufidx], length+MAX_MBPAIR_SIZE); dst = h->rbsp_buffer[bufidx]; if (dst == NULL) return NULL; + if(i>=length-1){ //no escaped 0 + *dst_length= length; + *consumed= length+1; //+1 for the header + if(h->s.avctx->flags2 & CODEC_FLAG2_FAST){ + return src; + }else{ + memcpy(dst, src, length); + return dst; + } + } + memcpy(dst, src, i); si = di = i; while (si + 2 < length) { @@ -369,7 +388,7 @@ static void await_references(H264Context *h) } else { int i; - assert(IS_8X8(mb_type)); + av_assert2(IS_8X8(mb_type)); for (i = 0; i < 4; i++) { const int sub_mb_type = h->sub_mb_type[i]; @@ -401,7 +420,7 @@ static void await_references(H264Context *h) nrefs); } else { int j; - assert(IS_SUB_4X4(sub_mb_type)); + av_assert2(IS_SUB_4X4(sub_mb_type)); for (j = 0; j < 4; j++) { int sub_y_offset = y_offset + 2 * (j & 2); get_lowest_part_y(h, refs, n + j, 4, sub_y_offset, @@ -740,9 +759,7 @@ static av_always_inline void prefetch_motion(H264Context *h, int list, s->vdsp.prefetch(src[1] + off, s->linesize, 4); s->vdsp.prefetch(src[2] + off, s->linesize, 4); } else { - off = ((mx >> 1) << pixel_shift) + - ((my >> 1) + (s->mb_x & 7)) * s->uvlinesize + - (64 << pixel_shift); + off= (((mx>>1)+64)<<pixel_shift) + ((my>>1) + (s->mb_x&7))*s->uvlinesize; s->vdsp.prefetch(src[1] + off, src[2] - src[1], 2); } } @@ -859,7 +876,7 @@ int ff_h264_alloc_tables(H264Context *h) { MpegEncContext *const s = &h->s; const int big_mb_num = s->mb_stride * (s->mb_height + 1); - const int row_mb_num = s->mb_stride * 2 * s->avctx->thread_count; + const int row_mb_num = 2*s->mb_stride*FFMAX(s->avctx->thread_count, 1); int x, y; FF_ALLOCZ_OR_GOTO(h->s.avctx, h->intra4x4_pred_mode, @@ -965,12 +982,18 @@ static av_cold void common_init(H264Context *h) s->height = s->avctx->height; s->codec_id = s->avctx->codec->id; - ff_h264dsp_init(&h->h264dsp, 8, 1); - ff_h264_pred_init(&h->hpc, s->codec_id, 8, 1); + s->avctx->bits_per_raw_sample = 8; + h->cur_chroma_format_idc = 1; + + ff_h264dsp_init(&h->h264dsp, + s->avctx->bits_per_raw_sample, h->cur_chroma_format_idc); + ff_h264_pred_init(&h->hpc, s->codec_id, + s->avctx->bits_per_raw_sample, h->cur_chroma_format_idc); h->dequant_coeff_pps = -1; s->unrestricted_mv = 1; + s->dsp.dct_bits = 16; /* needed so that IDCT permutation is known early */ ff_dsputil_init(&s->dsp, s->avctx); ff_videodsp_init(&s->vdsp, 8); @@ -979,17 +1002,20 @@ static av_cold void common_init(H264Context *h) memset(h->pps.scaling_matrix8, 16, 2 * 64 * sizeof(uint8_t)); } -int ff_h264_decode_extradata(H264Context *h) +int ff_h264_decode_extradata(H264Context *h, const uint8_t *buf, int size) { AVCodecContext *avctx = h->s.avctx; - if (avctx->extradata[0] == 1) { + if (!buf || size <= 0) + return -1; + + if (buf[0] == 1) { int i, cnt, nalsize; - unsigned char *p = avctx->extradata; + const unsigned char *p = buf; h->is_avc = 1; - if (avctx->extradata_size < 7) { + if (size < 7) { av_log(avctx, AV_LOG_ERROR, "avcC too short\n"); return -1; } @@ -1001,7 +1027,7 @@ int ff_h264_decode_extradata(H264Context *h) p += 6; for (i = 0; i < cnt; i++) { nalsize = AV_RB16(p) + 2; - if (p - avctx->extradata + nalsize > avctx->extradata_size) + if(nalsize > size - (p-buf)) return -1; if (decode_nal_units(h, p, nalsize, 1) < 0) { av_log(avctx, AV_LOG_ERROR, @@ -1014,7 +1040,7 @@ int ff_h264_decode_extradata(H264Context *h) cnt = *(p++); // Number of pps for (i = 0; i < cnt; i++) { nalsize = AV_RB16(p) + 2; - if (p - avctx->extradata + nalsize > avctx->extradata_size) + if(nalsize > size - (p-buf)) return -1; if (decode_nal_units(h, p, nalsize, 1) < 0) { av_log(avctx, AV_LOG_ERROR, @@ -1024,13 +1050,13 @@ int ff_h264_decode_extradata(H264Context *h) p += nalsize; } // Now store right nal length size, that will be used to parse all other nals - h->nal_length_size = (avctx->extradata[4] & 0x03) + 1; + h->nal_length_size = (buf[4] & 0x03) + 1; } else { h->is_avc = 0; - if (decode_nal_units(h, avctx->extradata, avctx->extradata_size, 1) < 0) + if (decode_nal_units(h, buf, size, 1) < 0) return -1; } - return 0; + return size; } av_cold int ff_h264_decode_init(AVCodecContext *avctx) @@ -1065,17 +1091,24 @@ av_cold int ff_h264_decode_init(AVCodecContext *avctx) for (i = 0; i < MAX_DELAYED_PIC_COUNT; i++) h->last_pocs[i] = INT_MIN; h->prev_poc_msb = 1 << 16; + h->prev_frame_num = -1; h->x264_build = -1; ff_h264_reset_sei(h); if (avctx->codec_id == AV_CODEC_ID_H264) { - if (avctx->ticks_per_frame == 1) - s->avctx->time_base.den *= 2; + if (avctx->ticks_per_frame == 1) { + if(s->avctx->time_base.den < INT_MAX/2) { + s->avctx->time_base.den *= 2; + } else + s->avctx->time_base.num /= 2; + } avctx->ticks_per_frame = 2; } if (avctx->extradata_size > 0 && avctx->extradata && - ff_h264_decode_extradata(h)) + ff_h264_decode_extradata(h, avctx->extradata, avctx->extradata_size) < 0) { + ff_h264_free_context(h); return -1; + } if (h->sps.bitstream_restriction_flag && s->avctx->has_b_frames < h->sps.num_reorder_frames) { @@ -1083,6 +1116,8 @@ av_cold int ff_h264_decode_init(AVCodecContext *avctx) s->low_delay = 0; } + ff_init_cabac_states(); + return 0; } @@ -1148,7 +1183,7 @@ static int decode_update_thread_context(AVCodecContext *dst, int inited = s->context_initialized, err; int i; - if (dst == src || !s1->context_initialized) + if (dst == src) return 0; if (inited && @@ -1166,6 +1201,13 @@ static int decode_update_thread_context(AVCodecContext *dst, s->height = s1->height; s->mb_height = s1->mb_height; h->b_stride = h1->b_stride; + // SPS/PPS + copy_parameter_set((void **)h->sps_buffers, (void **)h1->sps_buffers, + MAX_SPS_COUNT, sizeof(SPS)); + h->sps = h1->sps; + copy_parameter_set((void **)h->pps_buffers, (void **)h1->pps_buffers, + MAX_PPS_COUNT, sizeof(PPS)); + h->pps = h1->pps; if ((err = h264_slice_header_init(h, 1)) < 0) { av_log(h->s.avctx, AV_LOG_ERROR, "h264_slice_header_init() failed"); @@ -1199,12 +1241,16 @@ static int decode_update_thread_context(AVCodecContext *dst, sizeof(H264Context) - sizeof(MpegEncContext)); memset(h->sps_buffers, 0, sizeof(h->sps_buffers)); memset(h->pps_buffers, 0, sizeof(h->pps_buffers)); + + if (s1->context_initialized) { if (ff_h264_alloc_tables(h) < 0) { av_log(dst, AV_LOG_ERROR, "Could not allocate memory for h264\n"); return AVERROR(ENOMEM); } context_init(h); + } + for (i = 0; i < 2; i++) { h->rbsp_buffer[i] = NULL; h->rbsp_buffer_size[i] = 0; @@ -1219,7 +1265,7 @@ static int decode_update_thread_context(AVCodecContext *dst, /* frame_start may not be called for the next thread (if it's decoding * a bottom field) so this has to be allocated here */ - if (!h->bipred_scratchpad) + if (!h->bipred_scratchpad && s->linesize) h->bipred_scratchpad = av_malloc(16 * 6 * s->linesize); // extradata/NAL handling @@ -1261,6 +1307,7 @@ static int decode_update_thread_context(AVCodecContext *dst, MAX_DELAYED_PIC_COUNT + 2, s, s1); h->last_slice_type = h1->last_slice_type; + h->sync = h1->sync; if (!s->current_picture_ptr) return 0; @@ -1293,6 +1340,7 @@ int ff_h264_frame_start(H264Context *h) * See decode_nal_units(). */ s->current_picture_ptr->f.key_frame = 0; + s->current_picture_ptr->sync = 0; s->current_picture_ptr->mmco_reset = 0; assert(s->linesize && s->uvlinesize); @@ -1355,7 +1403,6 @@ static void decode_postinit(H264Context *h, int setup_finished) Picture *out = s->current_picture_ptr; Picture *cur = s->current_picture_ptr; int i, pics, out_of_order, out_idx; - int invalid = 0, cnt = 0; s->current_picture_ptr->f.qscale_type = FF_QSCALE_TYPE_H264; s->current_picture_ptr->f.pict_type = s->pict_type; @@ -1439,6 +1486,8 @@ static void decode_postinit(H264Context *h, int setup_finished) } } + cur->mmco_reset = h->mmco_reset; + h->mmco_reset = 0; // FIXME do something with unavailable reference frames /* Sort B-frames into display order */ @@ -1455,79 +1504,51 @@ static void decode_postinit(H264Context *h, int setup_finished) s->low_delay = 0; } + for (i = 0; 1; i++) { + if(i == MAX_DELAYED_PIC_COUNT || cur->poc < h->last_pocs[i]){ + if(i) + h->last_pocs[i-1] = cur->poc; + break; + } else if(i) { + h->last_pocs[i-1]= h->last_pocs[i]; + } + } + out_of_order = MAX_DELAYED_PIC_COUNT - i; + if( cur->f.pict_type == AV_PICTURE_TYPE_B + || (h->last_pocs[MAX_DELAYED_PIC_COUNT-2] > INT_MIN && h->last_pocs[MAX_DELAYED_PIC_COUNT-1] - h->last_pocs[MAX_DELAYED_PIC_COUNT-2] > 2)) + out_of_order = FFMAX(out_of_order, 1); + if(s->avctx->has_b_frames < out_of_order && !h->sps.bitstream_restriction_flag){ + av_log(s->avctx, AV_LOG_VERBOSE, "Increasing reorder buffer to %d\n", out_of_order); + s->avctx->has_b_frames = out_of_order; + s->low_delay = 0; + } + pics = 0; while (h->delayed_pic[pics]) pics++; - assert(pics <= MAX_DELAYED_PIC_COUNT); + av_assert0(pics <= MAX_DELAYED_PIC_COUNT); h->delayed_pic[pics++] = cur; if (cur->f.reference == 0) cur->f.reference = DELAYED_PIC_REF; - /* Frame reordering. This code takes pictures from coding order and sorts - * them by their incremental POC value into display order. It supports POC - * gaps, MMCO reset codes and random resets. - * A "display group" can start either with a IDR frame (f.key_frame = 1), - * and/or can be closed down with a MMCO reset code. In sequences where - * there is no delay, we can't detect that (since the frame was already - * output to the user), so we also set h->mmco_reset to detect the MMCO - * reset code. - * FIXME: if we detect insufficient delays (as per s->avctx->has_b_frames), - * we increase the delay between input and output. All frames affected by - * the lag (e.g. those that should have been output before another frame - * that we already returned to the user) will be dropped. This is a bug - * that we will fix later. */ - for (i = 0; i < MAX_DELAYED_PIC_COUNT; i++) { - cnt += out->poc < h->last_pocs[i]; - invalid += out->poc == INT_MIN; - } - if (!h->mmco_reset && !cur->f.key_frame && - cnt + invalid == MAX_DELAYED_PIC_COUNT && cnt > 0) { - h->mmco_reset = 2; - if (pics > 1) - h->delayed_pic[pics - 2]->mmco_reset = 2; - } - if (h->mmco_reset || cur->f.key_frame) { - for (i = 0; i < MAX_DELAYED_PIC_COUNT; i++) - h->last_pocs[i] = INT_MIN; - cnt = 0; - invalid = MAX_DELAYED_PIC_COUNT; - } - out = h->delayed_pic[0]; + out = h->delayed_pic[0]; out_idx = 0; - for (i = 1; i < MAX_DELAYED_PIC_COUNT && - h->delayed_pic[i] && - !h->delayed_pic[i - 1]->mmco_reset && - !h->delayed_pic[i]->f.key_frame; + for (i = 1; h->delayed_pic[i] && + !h->delayed_pic[i]->f.key_frame && + !h->delayed_pic[i]->mmco_reset; i++) if (h->delayed_pic[i]->poc < out->poc) { out = h->delayed_pic[i]; out_idx = i; } if (s->avctx->has_b_frames == 0 && - (h->delayed_pic[0]->f.key_frame || h->mmco_reset)) + (h->delayed_pic[0]->f.key_frame || h->delayed_pic[0]->mmco_reset)) h->next_outputed_poc = INT_MIN; - out_of_order = !out->f.key_frame && !h->mmco_reset && - (out->poc < h->next_outputed_poc); + out_of_order = out->poc < h->next_outputed_poc; - if (h->sps.bitstream_restriction_flag && - s->avctx->has_b_frames >= h->sps.num_reorder_frames) { - } else if (out_of_order && pics - 1 == s->avctx->has_b_frames && - s->avctx->has_b_frames < MAX_DELAYED_PIC_COUNT) { - if (invalid + cnt < MAX_DELAYED_PIC_COUNT) { - s->avctx->has_b_frames = FFMAX(s->avctx->has_b_frames, cnt); - } - s->low_delay = 0; - } else if (s->low_delay && - ((h->next_outputed_poc != INT_MIN && - out->poc > h->next_outputed_poc + 2) || - cur->f.pict_type == AV_PICTURE_TYPE_B)) { - s->low_delay = 0; - s->avctx->has_b_frames++; - } - - if (pics > s->avctx->has_b_frames) { + if (out_of_order || pics > s->avctx->has_b_frames) { out->f.reference &= ~DELAYED_PIC_REF; // for frame threading, the owner must be the second field's thread or // else the first thread can release the picture and reuse it unsafely @@ -1535,28 +1556,18 @@ static void decode_postinit(H264Context *h, int setup_finished) for (i = out_idx; h->delayed_pic[i]; i++) h->delayed_pic[i] = h->delayed_pic[i + 1]; } - memmove(h->last_pocs, &h->last_pocs[1], - sizeof(*h->last_pocs) * (MAX_DELAYED_PIC_COUNT - 1)); - h->last_pocs[MAX_DELAYED_PIC_COUNT - 1] = cur->poc; if (!out_of_order && pics > s->avctx->has_b_frames) { h->next_output_pic = out; - if (out->mmco_reset) { - if (out_idx > 0) { - h->next_outputed_poc = out->poc; - h->delayed_pic[out_idx - 1]->mmco_reset = out->mmco_reset; - } else { - h->next_outputed_poc = INT_MIN; - } - } else { - if (out_idx == 0 && pics > 1 && h->delayed_pic[0]->f.key_frame) { - h->next_outputed_poc = INT_MIN; - } else { - h->next_outputed_poc = out->poc; - } - } - h->mmco_reset = 0; + if (out_idx == 0 && h->delayed_pic[0] && (h->delayed_pic[0]->f.key_frame || h->delayed_pic[0]->mmco_reset)) { + h->next_outputed_poc = INT_MIN; + } else + h->next_outputed_poc = out->poc; } else { - av_log(s->avctx, AV_LOG_DEBUG, "no picture\n"); + av_log(s->avctx, AV_LOG_DEBUG, "no picture %s\n", out_of_order ? "ooo" : ""); + } + + if (h->next_output_pic && h->next_output_pic->sync) { + h->sync |= 2; } if (setup_finished) @@ -1829,7 +1840,7 @@ static av_always_inline void hl_decode_mb_predict_luma(H264Context *h, uint64_t tr_high; if (dir == DIAG_DOWN_LEFT_PRED || dir == VERT_LEFT_PRED) { const int topright_avail = (h->topright_samples_available << i) & 0x8000; - assert(s->mb_y || linesize <= block_offset[i]); + av_assert2(s->mb_y || linesize <= block_offset[i]); if (!topright_avail) { if (pixel_shift) { tr_high = ((uint16_t *)ptr)[3 - linesize / 2] * 0x0001000100010001ULL; @@ -2123,22 +2134,23 @@ static void implicit_weight_table(H264Context *h, int field) */ static void idr(H264Context *h) { + int i; ff_h264_remove_all_refs(h); h->prev_frame_num = 0; h->prev_frame_num_offset = 0; - h->prev_poc_msb = + h->prev_poc_msb = 1<<16; h->prev_poc_lsb = 0; + for (i = 0; i < MAX_DELAYED_PIC_COUNT; i++) + h->last_pocs[i] = INT_MIN; } /* forget old pics after a seek */ static void flush_change(H264Context *h) { - int i; - for (i = 0; i < MAX_DELAYED_PIC_COUNT; i++) - h->last_pocs[i] = INT_MIN; h->outputed_poc = h->next_outputed_poc = INT_MIN; h->prev_interlaced_frame = 1; idr(h); + h->prev_frame_num = -1; if (h->s.current_picture_ptr) h->s.current_picture_ptr->f.reference = 0; h->s.first_field = 0; @@ -2147,6 +2159,10 @@ static void flush_change(H264Context *h) memset(h->default_ref_list[0], 0, sizeof(h->default_ref_list[0])); memset(h->default_ref_list[1], 0, sizeof(h->default_ref_list[1])); ff_h264_reset_sei(h); + h->recovery_frame= -1; + h->sync= 0; + h->list_count = 0; + h->current_slice = 0; } /* forget old pics after a seek */ @@ -2155,7 +2171,7 @@ static void flush_dpb(AVCodecContext *avctx) H264Context *h = avctx->priv_data; int i; - for (i = 0; i < MAX_DELAYED_PIC_COUNT; i++) { + for (i = 0; i <= MAX_DELAYED_PIC_COUNT; i++) { if (h->delayed_pic[i]) h->delayed_pic[i]->f.reference = 0; h->delayed_pic[i] = NULL; @@ -2264,19 +2280,19 @@ static void init_scan_tables(H264Context *h) #undef T } if (h->sps.transform_bypass) { // FIXME same ugly - h->zigzag_scan_q0 = zigzag_scan; - h->zigzag_scan8x8_q0 = ff_zigzag_direct; - 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; + memcpy(h->zigzag_scan_q0 , zigzag_scan , sizeof(h->zigzag_scan_q0 )); + memcpy(h->zigzag_scan8x8_q0 , ff_zigzag_direct , sizeof(h->zigzag_scan8x8_q0 )); + memcpy(h->zigzag_scan8x8_cavlc_q0 , zigzag_scan8x8_cavlc , sizeof(h->zigzag_scan8x8_cavlc_q0)); + memcpy(h->field_scan_q0 , field_scan , sizeof(h->field_scan_q0 )); + memcpy(h->field_scan8x8_q0 , field_scan8x8 , sizeof(h->field_scan8x8_q0 )); + memcpy(h->field_scan8x8_cavlc_q0 , field_scan8x8_cavlc , sizeof(h->field_scan8x8_cavlc_q0 )); } 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; + memcpy(h->zigzag_scan_q0 , h->zigzag_scan , sizeof(h->zigzag_scan_q0 )); + memcpy(h->zigzag_scan8x8_q0 , h->zigzag_scan8x8 , sizeof(h->zigzag_scan8x8_q0 )); + memcpy(h->zigzag_scan8x8_cavlc_q0 , h->zigzag_scan8x8_cavlc , sizeof(h->zigzag_scan8x8_cavlc_q0)); + memcpy(h->field_scan_q0 , h->field_scan , sizeof(h->field_scan_q0 )); + memcpy(h->field_scan8x8_q0 , h->field_scan8x8 , sizeof(h->field_scan8x8_q0 )); + memcpy(h->field_scan8x8_cavlc_q0 , h->field_scan8x8_cavlc , sizeof(h->field_scan8x8_cavlc_q0 )); } } @@ -2429,7 +2445,9 @@ static int h264_set_parameter_from_sps(H264Context *h) "VDPAU decoding does not support video colorspace.\n"); return AVERROR_INVALIDDATA; } - if (h->sps.bit_depth_luma >= 8 && h->sps.bit_depth_luma <= 10) { + if (h->sps.bit_depth_luma >= 8 && h->sps.bit_depth_luma <= 14 && + h->sps.bit_depth_luma != 11 && h->sps.bit_depth_luma != 13 && + (h->sps.bit_depth_luma != 9 || !CHROMA422)) { s->avctx->bits_per_raw_sample = h->sps.bit_depth_luma; h->cur_chroma_format_idc = h->sps.chroma_format_idc; h->pixel_shift = h->sps.bit_depth_luma > 8; @@ -2476,13 +2494,38 @@ static enum PixelFormat get_pixel_format(H264Context *h) else return AV_PIX_FMT_YUV420P10; break; - case 8: + case 12: + if (CHROMA444) { + if (s->avctx->colorspace == AVCOL_SPC_RGB) { + return AV_PIX_FMT_GBRP12; + } else + return AV_PIX_FMT_YUV444P12; + } else if (CHROMA422) + return AV_PIX_FMT_YUV422P12; + else + return AV_PIX_FMT_YUV420P12; + break; + case 14: if (CHROMA444) { if (s->avctx->colorspace == AVCOL_SPC_RGB) { - return AV_PIX_FMT_GBRP; + return AV_PIX_FMT_GBRP14; } else - return s->avctx->color_range == AVCOL_RANGE_JPEG ? AV_PIX_FMT_YUVJ444P - : AV_PIX_FMT_YUV444P; + return AV_PIX_FMT_YUV444P14; + } else if (CHROMA422) + return AV_PIX_FMT_YUV422P14; + else + return AV_PIX_FMT_YUV420P14; + break; + case 8: + if (CHROMA444) { + if (s->avctx->colorspace == AVCOL_SPC_RGB) { + av_log(h->s.avctx, AV_LOG_DEBUG, "Detected GBR colorspace.\n"); + return AV_PIX_FMT_GBR24P; + } else if (s->avctx->colorspace == AVCOL_SPC_YCGCO) { + av_log(h->s.avctx, AV_LOG_WARNING, "Detected unsupported YCgCo colorspace.\n"); + } + return s->avctx->color_range == AVCOL_RANGE_JPEG ? AV_PIX_FMT_YUVJ444P + : AV_PIX_FMT_YUV444P; } else if (CHROMA422) { return s->avctx->color_range == AVCOL_RANGE_JPEG ? AV_PIX_FMT_YUVJ422P : AV_PIX_FMT_YUV422P; @@ -2506,7 +2549,20 @@ static int h264_slice_header_init(H264Context *h, int reinit) MpegEncContext *const s = &h->s; int i, ret; - avcodec_set_dimensions(s->avctx, s->width, s->height); + if( FFALIGN(s->avctx->width , 16 ) == s->width + && FFALIGN(s->avctx->height, 16*(2 - h->sps.frame_mbs_only_flag)) == s->height + && !h->sps.crop_right && !h->sps.crop_bottom + && (s->avctx->width != s->width || s->avctx->height && s->height) + ) { + av_log(h->s.avctx, AV_LOG_DEBUG, "Using externally provided dimensions\n"); + s->avctx->coded_width = s->width; + s->avctx->coded_height = s->height; + } else{ + avcodec_set_dimensions(s->avctx, s->width, s->height); + s->avctx->width -= (2>>CHROMA444)*FFMIN(h->sps.crop_right, (8<<CHROMA444)-1); + s->avctx->height -= (1<<s->chroma_y_shift)*FFMIN(h->sps.crop_bottom, (16>>s->chroma_y_shift)-1) * (2 - h->sps.frame_mbs_only_flag); + } + s->avctx->sample_aspect_ratio = h->sps.sar; av_assert0(s->avctx->sample_aspect_ratio.den); @@ -2557,6 +2613,7 @@ static int h264_slice_header_init(H264Context *h, int reinit) c->sps = h->sps; c->pps = h->pps; c->pixel_shift = h->pixel_shift; + c->cur_chroma_format_idc = h->cur_chroma_format_idc; init_scan_tables(c); clone_tables(c, h, i); } @@ -2587,10 +2644,11 @@ static int decode_slice_header(H264Context *h, H264Context *h0) MpegEncContext *const s0 = &h0->s; unsigned int first_mb_in_slice; unsigned int pps_id; - int num_ref_idx_active_override_flag, max_refs, ret; + int num_ref_idx_active_override_flag, ret; unsigned int slice_type, tmp, i, j; int default_ref_list_done = 0; int last_pic_structure, last_pic_droppable; + int must_reinit; int needs_reinit = 0; enum AVPixelFormat pix_fmt; @@ -2604,7 +2662,7 @@ static int decode_slice_header(H264Context *h, H264Context *h0) s->me.qpel_avg = s->dsp.avg_h264_qpel_pixels_tab; } - first_mb_in_slice = get_ue_golomb(&s->gb); + first_mb_in_slice = get_ue_golomb_long(&s->gb); if (first_mb_in_slice == 0) { // FIXME better field boundary detection if (h0->current_slice && FIELD_PICTURE) { @@ -2626,7 +2684,7 @@ static int decode_slice_header(H264Context *h, H264Context *h0) if (slice_type > 9) { av_log(h->s.avctx, AV_LOG_ERROR, "slice type too large (%d) at %d %d\n", - h->slice_type, s->mb_x, s->mb_y); + slice_type, s->mb_x, s->mb_y); return -1; } if (slice_type > 4) { @@ -2648,7 +2706,7 @@ static int decode_slice_header(H264Context *h, H264Context *h0) 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"); + av_log(h->s.avctx, AV_LOG_ERROR, "pps_id %d out of range\n", pps_id); return -1; } if (!h0->pps_buffers[pps_id]) { @@ -2674,6 +2732,13 @@ static int decode_slice_header(H264Context *h, H264Context *h0) h->current_sps_id = h->pps.sps_id; h->sps = *h0->sps_buffers[h->pps.sps_id]; + if (s->mb_width != h->sps.mb_width || + s->mb_height != h->sps.mb_height * (2 - h->sps.frame_mbs_only_flag) || + s->avctx->bits_per_raw_sample != h->sps.bit_depth_luma || + h->cur_chroma_format_idc != h->sps.chroma_format_idc + ) + needs_reinit = 1; + if ((ret = h264_set_parameter_from_sps(h)) < 0) return ret; } @@ -2682,9 +2747,13 @@ static int decode_slice_header(H264Context *h, H264Context *h0) s->avctx->level = h->sps.level_idc; s->avctx->refs = h->sps.ref_frame_count; - if (s->mb_width != h->sps.mb_width || - s->mb_height != h->sps.mb_height * (2 - h->sps.frame_mbs_only_flag)) - needs_reinit = 1; + must_reinit = (s->context_initialized && + ( 16*h->sps.mb_width != s->avctx->coded_width + || 16*h->sps.mb_height * (2 - h->sps.frame_mbs_only_flag) != s->avctx->coded_height + || s->avctx->bits_per_raw_sample != h->sps.bit_depth_luma + || h->cur_chroma_format_idc != h->sps.chroma_format_idc + || av_cmp_q(h->sps.sar, s->avctx->sample_aspect_ratio))); + s->mb_width = h->sps.mb_width; s->mb_height = h->sps.mb_height * (2 - h->sps.frame_mbs_only_flag); @@ -2693,21 +2762,12 @@ static int decode_slice_header(H264Context *h, H264Context *h0) s->chroma_y_shift = h->sps.chroma_format_idc <= 1; // 400 uses yuv420p - s->width = 16 * s->mb_width - (2 >> CHROMA444) * FFMIN(h->sps.crop_right, (8 << CHROMA444) - 1); - if (h->sps.frame_mbs_only_flag) - s->height = 16 * s->mb_height - (1 << s->chroma_y_shift) * FFMIN(h->sps.crop_bottom, (16 >> s->chroma_y_shift) - 1); - else - s->height = 16 * s->mb_height - (2 << s->chroma_y_shift) * FFMIN(h->sps.crop_bottom, (16 >> s->chroma_y_shift) - 1); - - if (FFALIGN(s->avctx->width, 16) == s->width && - FFALIGN(s->avctx->height, 16) == s->height) { - s->width = s->avctx->width; - s->height = s->avctx->height; - } + s->width = 16 * s->mb_width; + s->height = 16 * s->mb_height; if (h->sps.video_signal_type_present_flag) { - s->avctx->color_range = h->sps.full_range ? AVCOL_RANGE_JPEG - : AVCOL_RANGE_MPEG; + s->avctx->color_range = h->sps.full_range>0 ? AVCOL_RANGE_JPEG + : AVCOL_RANGE_MPEG; if (h->sps.colour_description_present_flag) { s->avctx->color_primaries = h->sps.color_primaries; s->avctx->color_trc = h->sps.color_trc; @@ -2715,6 +2775,7 @@ static int decode_slice_header(H264Context *h, H264Context *h0) } } + ret = get_pixel_format(h); if (ret < 0) return ret; @@ -2724,11 +2785,9 @@ static int decode_slice_header(H264Context *h, H264Context *h0) s->avctx->pix_fmt = pix_fmt; if (s->context_initialized && - (s->width != s->avctx->width || - s->height != s->avctx->height || - pix_fmt != s->avctx->pix_fmt || + ( needs_reinit || - av_cmp_q(h->sps.sar, s->avctx->sample_aspect_ratio))) { + must_reinit)) { if (h != h0) { av_log(s->avctx, AV_LOG_ERROR, "changing width/height on " @@ -2778,6 +2837,10 @@ static int decode_slice_header(H264Context *h, H264Context *h0) if (h->sps.frame_mbs_only_flag) { s->picture_structure = PICT_FRAME; } else { + if (!h->sps.direct_8x8_inference_flag && slice_type == AV_PICTURE_TYPE_B) { + av_log(h->s.avctx, AV_LOG_ERROR, "This stream was generated by a broken encoder, invalid 8x8 inference\n"); + return -1; + } if (get_bits1(&s->gb)) { // field_pic_flag s->picture_structure = PICT_TOP_FIELD + get_bits1(&s->gb); // bottom_field_flag } else { @@ -2805,7 +2868,7 @@ static int decode_slice_header(H264Context *h, H264Context *h0) } else { /* Shorten frame num gaps so we don't have to allocate reference * frames just to throw them away */ - if (h->frame_num != h->prev_frame_num) { + if (h->frame_num != h->prev_frame_num && h->prev_frame_num >= 0) { int unwrap_prev_frame_num = h->prev_frame_num; int max_frame_num = 1 << h->sps.log2_max_frame_num; @@ -2887,7 +2950,7 @@ static int decode_slice_header(H264Context *h, H264Context *h0) } } - while (h->frame_num != h->prev_frame_num && + while (h->frame_num != h->prev_frame_num && h->prev_frame_num >= 0 && !s0->first_field && h->frame_num != (h->prev_frame_num + 1) % (1 << h->sps.log2_max_frame_num)) { Picture *prev = h->short_ref_count ? h->short_ref[0] : NULL; av_log(h->s.avctx, AV_LOG_DEBUG, "Frame num gap %d %d\n", @@ -2936,6 +2999,8 @@ static int decode_slice_header(H264Context *h, H264Context *h0) s0->first_field = FIELD_PICTURE; } else { if (s0->current_picture_ptr->frame_num != h->frame_num) { + ff_thread_report_progress((AVFrame*)s0->current_picture_ptr, INT_MAX, + s0->picture_structure==PICT_BOTTOM_FIELD); /* This and the previous field had different frame_nums. * Consider this field first in pair. Throw away previous * one except for reference purposes. */ @@ -2965,7 +3030,7 @@ static int decode_slice_header(H264Context *h, H264Context *h0) s->current_picture_ptr->frame_num = h->frame_num; // FIXME frame_num cleanup - assert(s->mb_num == s->mb_width * s->mb_height); + av_assert1(s->mb_num == s->mb_width * s->mb_height); 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"); @@ -2975,7 +3040,7 @@ static int decode_slice_header(H264Context *h, H264Context *h0) 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); + av_assert1(s->mb_y < s->mb_height); if (s->picture_structure == PICT_FRAME) { h->curr_pic_num = h->frame_num; @@ -3012,19 +3077,26 @@ static int decode_slice_header(H264Context *h, H264Context *h0) h->ref_count[1] = h->pps.ref_count[1]; if (h->slice_type_nos != AV_PICTURE_TYPE_I) { + unsigned max[2]; + max[0] = max[1] = s->picture_structure == PICT_FRAME ? 15 : 31; + if (h->slice_type_nos == AV_PICTURE_TYPE_B) h->direct_spatial_mv_pred = get_bits1(&s->gb); num_ref_idx_active_override_flag = get_bits1(&s->gb); if (num_ref_idx_active_override_flag) { h->ref_count[0] = get_ue_golomb(&s->gb) + 1; - if (h->ref_count[0] < 1) - return AVERROR_INVALIDDATA; if (h->slice_type_nos == AV_PICTURE_TYPE_B) { h->ref_count[1] = get_ue_golomb(&s->gb) + 1; - if (h->ref_count[1] < 1) - return AVERROR_INVALIDDATA; - } + } else + // full range is spec-ok in this case, even for frames + h->ref_count[1] = 1; + } + + if (h->ref_count[0]-1 > max[0] || h->ref_count[1]-1 > max[1]){ + av_log(h->s.avctx, AV_LOG_ERROR, "reference overflow %u > %u or %u > %u\n", h->ref_count[0]-1, max[0], h->ref_count[1]-1, max[1]); + h->ref_count[0] = h->ref_count[1] = 1; + return AVERROR_INVALIDDATA; } if (h->slice_type_nos == AV_PICTURE_TYPE_B) @@ -3032,15 +3104,7 @@ static int decode_slice_header(H264Context *h, H264Context *h0) else h->list_count = 1; } else - h->list_count = 0; - - max_refs = s->picture_structure == PICT_FRAME ? 16 : 32; - - if (h->ref_count[0] > max_refs || h->ref_count[1] > max_refs) { - av_log(h->s.avctx, AV_LOG_ERROR, "reference overflow\n"); - h->ref_count[0] = h->ref_count[1] = 1; - return AVERROR_INVALIDDATA; - } + h->ref_count[1]= h->ref_count[0]= h->list_count= 0; if (!default_ref_list_done) ff_h264_fill_default_ref_list(h); @@ -3183,9 +3247,14 @@ static int decode_slice_header(H264Context *h, H264Context *h0) h0->last_slice_type = slice_type; h->slice_num = ++h0->current_slice; - if (h->slice_num >= MAX_SLICES) { - av_log(s->avctx, AV_LOG_ERROR, - "Too many slices, increase MAX_SLICES and recompile\n"); + + if (h->slice_num) + h0->slice_row[(h->slice_num-1)&(MAX_SLICES-1)]= s->resync_mb_y; + if ( h0->slice_row[h->slice_num&(MAX_SLICES-1)] + 3 >= s->resync_mb_y + && h0->slice_row[h->slice_num&(MAX_SLICES-1)] <= s->resync_mb_y + && h->slice_num >= MAX_SLICES) { + //in case of ASO this check needs to be updated depending on how we decide to assign slice numbers in this case + av_log(s->avctx, AV_LOG_WARNING, "Possibly too many slices (%d >= %d), increase MAX_SLICES and recompile if there are artifacts\n", h->slice_num, MAX_SLICES); } for (j = 0; j < 2; j++) { @@ -3285,7 +3354,7 @@ static av_always_inline void fill_filter_caches_inter(H264Context *h, if (USES_LIST(top_type, list)) { const int b_xy = h->mb2b_xy[top_xy] + 3 * b_stride; const int b8_xy = 4 * top_xy + 2; - int (*ref2frm)[64] = h->ref2frm[h->slice_table[top_xy] & (MAX_SLICES - 1)][0] + (MB_MBAFF ? 20 : 2); + int (*ref2frm)[64] = (void*)(h->ref2frm[h->slice_table[top_xy] & (MAX_SLICES - 1)][0] + (MB_MBAFF ? 20 : 2)); AV_COPY128(mv_dst - 1 * 8, s->current_picture.f.motion_val[list][b_xy + 0]); ref_cache[0 - 1 * 8] = ref_cache[1 - 1 * 8] = ref2frm[list][s->current_picture.f.ref_index[list][b8_xy + 0]]; @@ -3300,7 +3369,7 @@ static av_always_inline void fill_filter_caches_inter(H264Context *h, if (USES_LIST(left_type[LTOP], list)) { const int b_xy = h->mb2b_xy[left_xy[LTOP]] + 3; const int b8_xy = 4 * left_xy[LTOP] + 1; - int (*ref2frm)[64] = h->ref2frm[h->slice_table[left_xy[LTOP]] & (MAX_SLICES - 1)][0] + (MB_MBAFF ? 20 : 2); + int (*ref2frm)[64] =(void*)( h->ref2frm[h->slice_table[left_xy[LTOP]] & (MAX_SLICES - 1)][0] + (MB_MBAFF ? 20 : 2)); AV_COPY32(mv_dst - 1 + 0, s->current_picture.f.motion_val[list][b_xy + b_stride * 0]); AV_COPY32(mv_dst - 1 + 8, s->current_picture.f.motion_val[list][b_xy + b_stride * 1]); AV_COPY32(mv_dst - 1 + 16, s->current_picture.f.motion_val[list][b_xy + b_stride * 2]); @@ -3333,7 +3402,7 @@ static av_always_inline void fill_filter_caches_inter(H264Context *h, { int8_t *ref = &s->current_picture.f.ref_index[list][4 * mb_xy]; - int (*ref2frm)[64] = h->ref2frm[h->slice_num & (MAX_SLICES - 1)][0] + (MB_MBAFF ? 20 : 2); + int (*ref2frm)[64] = (void*)(h->ref2frm[h->slice_num & (MAX_SLICES - 1)][0] + (MB_MBAFF ? 20 : 2)); uint32_t ref01 = (pack16to32(ref2frm[list][ref[0]], ref2frm[list][ref[1]]) & 0x00FF00FF) * 0x0101; uint32_t ref23 = (pack16to32(ref2frm[list][ref[2]], ref2frm[list][ref[3]]) & 0x00FF00FF) * 0x0101; AV_WN32A(&ref_cache[0 * 8], ref01); @@ -3636,7 +3705,6 @@ static int decode_slice(struct AVCodecContext *avctx, void *arg) align_get_bits(&s->gb); /* init cabac */ - ff_init_cabac_states(&h->cabac); ff_init_cabac_decoder(&h->cabac, s->gb.buffer + get_bits_count(&s->gb) / 8, (get_bits_left(&s->gb) + 7) / 8); @@ -3672,7 +3740,9 @@ static int decode_slice(struct AVCodecContext *avctx, void *arg) loop_filter(h, lf_x_start, s->mb_x + 1); return 0; } - if (ret < 0 || h->cabac.bytestream > h->cabac.bytestream_end + 2) { + if (h->cabac.bytestream > h->cabac.bytestream_end + 2 ) + av_log(h->s.avctx, AV_LOG_DEBUG, "bytestream overread %td\n", h->cabac.bytestream_end - h->cabac.bytestream); + if (ret < 0 || h->cabac.bytestream > h->cabac.bytestream_end + 4) { av_log(h->s.avctx, AV_LOG_ERROR, "error while decoding MB %d %d, bytestream (%td)\n", s->mb_x, s->mb_y, @@ -3743,7 +3813,8 @@ static int decode_slice(struct AVCodecContext *avctx, void *arg) tprintf(s->avctx, "slice end %d %d\n", get_bits_count(&s->gb), s->gb.size_in_bits); - if (get_bits_left(&s->gb) == 0) { + if ( get_bits_left(&s->gb) == 0 + || get_bits_left(&s->gb) > 0 && !(s->avctx->err_recognition & AV_EF_AGGRESSIVE)) { ff_er_add_slice(s, s->resync_mb_x, s->resync_mb_y, s->mb_x - 1, s->mb_y, ER_MB_END & part_mask); @@ -3751,7 +3822,7 @@ static int decode_slice(struct AVCodecContext *avctx, void *arg) return 0; } else { ff_er_add_slice(s, s->resync_mb_x, s->resync_mb_y, - s->mb_x - 1, s->mb_y, + s->mb_x, s->mb_y, ER_MB_END & part_mask); return -1; @@ -3804,6 +3875,7 @@ static int execute_decode_slices(H264Context *h, int context_count) hx = h->thread_context[i]; hx->s.err_recognition = avctx->err_recognition; hx->s.error_count = 0; + hx->x264_build = h->x264_build; } avctx->execute(avctx, decode_slice, h->thread_context, @@ -3834,7 +3906,13 @@ static int decode_nal_units(H264Context *h, const uint8_t *buf, int buf_size, int pass = !(avctx->active_thread_type & FF_THREAD_FRAME); int nals_needed = 0; ///< number of NALs that need decoding before the next frame thread starts int nal_index; + int idr_cleared=0; + int first_slice = 0; + + h->nal_unit_type= 0; + if(!s->slice_context_count) + s->slice_context_count= 1; h->max_contexts = s->slice_context_count; if (!(s->flags2 & CODEC_FLAG2_CHUNKS)) { h->current_slice = 0; @@ -3843,6 +3921,13 @@ static int decode_nal_units(H264Context *h, const uint8_t *buf, int buf_size, ff_h264_reset_sei(h); } + if (h->nal_length_size == 4) { + if (buf_size > 8 && AV_RB32(buf) == 1 && AV_RB32(buf+5) > (unsigned)buf_size) { + h->is_avc = 0; + }else if(buf_size > 3 && AV_RB32(buf) > 1 && AV_RB32(buf) <= (unsigned)buf_size) + h->is_avc = 1; + } + for (; pass <= 1; pass++) { buf_index = 0; context_count = 0; @@ -3902,16 +3987,14 @@ static int decode_nal_units(H264Context *h, const uint8_t *buf, int buf_size, s->workaround_bugs |= FF_BUG_TRUNCATED; if (!(s->workaround_bugs & FF_BUG_TRUNCATED)) - while (ptr[dst_length - 1] == 0 && dst_length > 0) + while(dst_length > 0 && ptr[dst_length - 1] == 0) dst_length--; 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", - hx->nal_unit_type, buf_index, buf_size, dst_length); + av_log(h->s.avctx, AV_LOG_DEBUG, "NAL %d/%d at %d/%d length %d pass %d\n", hx->nal_unit_type, hx->nal_ref_idc, buf_index, buf_size, dst_length, pass); if (h->is_avc && (nalsize != consumed) && nalsize) av_log(h->s.avctx, AV_LOG_DEBUG, @@ -3935,38 +4018,56 @@ static int decode_nal_units(H264Context *h, const uint8_t *buf, int buf_size, case NAL_IDR_SLICE: case NAL_SLICE: init_get_bits(&hx->s.gb, ptr, bit_length); - if (!get_ue_golomb(&hx->s.gb)) + if (!get_ue_golomb(&hx->s.gb) || !first_slice) nals_needed = nal_index; + if (!first_slice) + first_slice = hx->nal_unit_type; } continue; } + if (!first_slice) + switch (hx->nal_unit_type) { + case NAL_DPA: + case NAL_IDR_SLICE: + case NAL_SLICE: + first_slice = hx->nal_unit_type; + } + // FIXME do not discard SEI id if (avctx->skip_frame >= AVDISCARD_NONREF && h->nal_ref_idc == 0) continue; again: - /* Ignore every NAL unit type except PPS and SPS during extradata + /* Ignore per frame NAL unit type during extradata * parsing. Decoding slices is not possible in codec init * with frame-mt */ - if (parse_extradata && HAVE_THREADS && - (s->avctx->active_thread_type & FF_THREAD_FRAME) && - (hx->nal_unit_type != NAL_PPS && - hx->nal_unit_type != NAL_SPS)) { - av_log(avctx, AV_LOG_INFO, "Ignoring NAL unit %d during " - "extradata parsing\n", hx->nal_unit_type); - hx->nal_unit_type = NAL_FF_IGNORE; + if (parse_extradata) { + switch (hx->nal_unit_type) { + case NAL_IDR_SLICE: + case NAL_SLICE: + case NAL_DPA: + case NAL_DPB: + case NAL_DPC: + case NAL_AUXILIARY_SLICE: + av_log(h->s.avctx, AV_LOG_WARNING, "Ignoring NAL %d in global header/extradata\n", hx->nal_unit_type); + hx->nal_unit_type = NAL_FF_IGNORE; + } } + err = 0; + switch (hx->nal_unit_type) { case NAL_IDR_SLICE: - if (h->nal_unit_type != NAL_IDR_SLICE) { + if (first_slice != NAL_IDR_SLICE) { av_log(h->s.avctx, AV_LOG_ERROR, "Invalid mix of idr and non-idr slices\n"); buf_index = -1; goto end; } - idr(h); // FIXME ensure we don't lose some frames if there is reordering + if(!idr_cleared) + idr(h); // FIXME ensure we don't lose some frames if there is reordering + idr_cleared = 1; case NAL_SLICE: init_get_bits(&hx->s.gb, ptr, bit_length); hx->intra_gb_ptr = @@ -3976,9 +4077,30 @@ again: if ((err = decode_slice_header(hx, h))) break; + if (h->sei_recovery_frame_cnt >= 0 && (h->frame_num != h->sei_recovery_frame_cnt || hx->slice_type_nos != AV_PICTURE_TYPE_I)) + h->valid_recovery_point = 1; + + if ( h->sei_recovery_frame_cnt >= 0 + && ( h->recovery_frame<0 + || ((h->recovery_frame - h->frame_num) & ((1 << h->sps.log2_max_frame_num)-1)) > h->sei_recovery_frame_cnt)) { + h->recovery_frame = (h->frame_num + h->sei_recovery_frame_cnt) % + (1 << h->sps.log2_max_frame_num); + + if (!h->valid_recovery_point) + h->recovery_frame = h->frame_num; + } + s->current_picture_ptr->f.key_frame |= - (hx->nal_unit_type == NAL_IDR_SLICE) || - (h->sei_recovery_frame_cnt >= 0); + (hx->nal_unit_type == NAL_IDR_SLICE); + + if (h->recovery_frame == h->frame_num) { + s->current_picture_ptr->sync |= 1; + h->recovery_frame = -1; + } + + h->sync |= !!s->current_picture_ptr->f.key_frame; + h->sync |= 3*!!(s->flags2 & CODEC_FLAG2_SHOW_ALL); + s->current_picture_ptr->sync |= h->sync; if (h->current_slice == 1) { if (!(s->flags2 & CODEC_FLAG2_CHUNKS)) @@ -4035,6 +4157,9 @@ again: init_get_bits(&hx->inter_gb, ptr, bit_length); hx->inter_gb_ptr = &hx->inter_gb; + av_log(h->s.avctx, AV_LOG_ERROR, "Partitioned H.264 support is incomplete\n"); + break; + if (hx->redundant_pic_count == 0 && hx->intra_gb_ptr && hx->s.data_partitioning && @@ -4054,19 +4179,18 @@ again: break; case NAL_SPS: init_get_bits(&s->gb, ptr, bit_length); - if (ff_h264_decode_seq_parameter_set(h) < 0 && - h->is_avc && (nalsize != consumed) && nalsize) { + if (ff_h264_decode_seq_parameter_set(h) < 0 && (h->is_avc ? (nalsize != consumed) && nalsize : 1)) { av_log(h->s.avctx, AV_LOG_DEBUG, "SPS decoding failure, trying again with the complete NAL\n"); - init_get_bits(&s->gb, buf + buf_index + 1 - consumed, - 8 * (nalsize - 1)); + if (h->is_avc) + av_assert0(next_avc - buf_index + consumed == nalsize); + if ((next_avc - buf_index + consumed - 1) >= INT_MAX/8) + break; + init_get_bits(&s->gb, &buf[buf_index + 1 - consumed], + 8*(next_avc - buf_index + consumed - 1)); ff_h264_decode_seq_parameter_set(h); } - if (h264_set_parameter_from_sps(h) < 0) { - buf_index = -1; - goto end; - } break; case NAL_PPS: init_get_bits(&s->gb, ptr, bit_length); @@ -4141,15 +4265,15 @@ static int decode_frame(AVCodecContext *avctx, void *data, MpegEncContext *s = &h->s; AVFrame *pict = data; int buf_index = 0; + Picture *out; + int i, out_idx; s->flags = avctx->flags; s->flags2 = avctx->flags2; /* end of stream, output what is still in the buffers */ -out: if (buf_size == 0) { - Picture *out; - int i, out_idx; + out: s->current_picture_ptr = NULL; @@ -4176,19 +4300,42 @@ out: return buf_index; } + if(h->is_avc && buf_size >= 9 && buf[0]==1 && buf[2]==0 && (buf[4]&0xFC)==0xFC && (buf[5]&0x1F) && buf[8]==0x67){ + int cnt= buf[5]&0x1f; + const uint8_t *p= buf+6; + while(cnt--){ + int nalsize= AV_RB16(p) + 2; + if(nalsize > buf_size - (p-buf) || p[2]!=0x67) + goto not_extra; + p += nalsize; + } + cnt = *(p++); + if(!cnt) + goto not_extra; + while(cnt--){ + int nalsize= AV_RB16(p) + 2; + if(nalsize > buf_size - (p-buf) || p[2]!=0x68) + goto not_extra; + p += nalsize; + } + + return ff_h264_decode_extradata(h, buf, buf_size); + } +not_extra: buf_index = decode_nal_units(h, buf, buf_size, 0); if (buf_index < 0) return -1; if (!s->current_picture_ptr && h->nal_unit_type == NAL_END_SEQUENCE) { - buf_size = 0; + av_assert0(buf_index <= buf_size); goto out; } if (!(s->flags2 & CODEC_FLAG2_CHUNKS) && !s->current_picture_ptr) { - if (avctx->skip_frame >= AVDISCARD_NONREF) - return 0; + if (avctx->skip_frame >= AVDISCARD_NONREF || + buf_size >= 4 && !memcmp("Q264", buf, 4)) + return buf_size; av_log(avctx, AV_LOG_ERROR, "no frame!\n"); return -1; } @@ -4201,10 +4348,9 @@ out: field_end(h, 0); h->context_reinitialized = 0; - if (!h->next_output_pic) { - /* Wait for second field. */ - *got_frame = 0; - } else { + /* Wait for second field. */ + *got_frame = 0; + if (h->next_output_pic && (h->next_output_pic->sync || h->sync>1)) { *got_frame = 1; *pict = h->next_output_pic->f; } @@ -4234,6 +4380,7 @@ static av_cold int h264_decode_end(AVCodecContext *avctx) H264Context *h = avctx->priv_data; MpegEncContext *s = &h->s; + ff_h264_remove_all_refs(h); ff_h264_free_context(h); ff_MPV_common_end(s); @@ -4260,6 +4407,26 @@ static const AVProfile profiles[] = { { FF_PROFILE_UNKNOWN }, }; +static const AVOption h264_options[] = { + {"is_avc", "is avc", offsetof(H264Context, is_avc), FF_OPT_TYPE_INT, {.i64 = 0}, 0, 1, 0}, + {"nal_length_size", "nal_length_size", offsetof(H264Context, nal_length_size), FF_OPT_TYPE_INT, {.i64 = 0}, 0, 4, 0}, + {NULL} +}; + +static const AVClass h264_class = { + .class_name = "H264 Decoder", + .item_name = av_default_item_name, + .option = h264_options, + .version = LIBAVUTIL_VERSION_INT, +}; + +static const AVClass h264_vdpau_class = { + .class_name = "H264 VDPAU Decoder", + .item_name = av_default_item_name, + .option = h264_options, + .version = LIBAVUTIL_VERSION_INT, +}; + AVCodec ff_h264_decoder = { .name = "h264", .type = AVMEDIA_TYPE_VIDEO, @@ -4276,6 +4443,7 @@ AVCodec ff_h264_decoder = { .init_thread_copy = ONLY_IF_THREADS_ENABLED(decode_init_thread_copy), .update_thread_context = ONLY_IF_THREADS_ENABLED(decode_update_thread_context), .profiles = NULL_IF_CONFIG_SMALL(profiles), + .priv_class = &h264_class, }; #if CONFIG_H264_VDPAU_DECODER @@ -4293,5 +4461,6 @@ AVCodec ff_h264_vdpau_decoder = { .pix_fmts = (const enum AVPixelFormat[]) { AV_PIX_FMT_VDPAU_H264, AV_PIX_FMT_NONE}, .profiles = NULL_IF_CONFIG_SMALL(profiles), + .priv_class = &h264_vdpau_class, }; #endif |
