[FFmpeg-trac] #10961(avcodec:new): RTSP h264 hardware decode shows a very strange log: Increasing reorder buffer to 1920921104

[FFmpeg]

#10961: RTSP h264 hardware decode shows a very strange log: Increasing reorder
buffer to 1920921104
-------------------------------------+-------------------------------------
             Reporter:  hechuan      |                     Type:  task
               Status:  new          |                 Priority:  critical
            Component:  avcodec      |                  Version:  4.2.9
             Keywords:  FFmpeg       |               Blocked By:
  decode hw_decode RTSP              |
             Blocking:               |  Reproduced by developer:  0
Analyzed by developer:  0            |
-------------------------------------+-------------------------------------
 Use FFmpeg to hardware decode (h264_nvdec) two RTSP h264 IP camera.Eache
 camera has a seperate thread to read frame, decode and scale.After run
 some time, ffmpeg show some very strange log as blow:

 1710848158.543170 [243] [debug] non-existing PPS 1920921104 referenced
 1710848158.543174 [243] [debug] decode_slice_header error
 1710848158.543174 [243] [debug] no frame!
 1710848158.647788 [243] [debug] non-existing PPS 1920957856 referenced
 1710848158.647867 [243] [debug] non-existing PPS 1920921104 referenced
 1710848158.647872 [243] [debug] decode_slice_header error
 1710848158.647873 [243] [debug] no frame!
 1710848158.705852 [243] [debug] error while decoding MB 1920958160 0,
 bytestream -6
 1710848158.763328 [243] [debug] Increasing reorder buffer to 1920921104
 1710848160.782519 [265] [debug] Increasing reorder buffer to 635491232
 1710864812.410619 [260] [debug] SEI type 1101054320 size 0 truncated at
 136
 1710864812.410726 [260] [debug] SEI type 1101055152 size 0 truncated at
 134
 1710864812.453565 [265] [debug] SEI type 635527536 size 0 truncated at 136
 1710864812.453646 [265] [debug] SEI type 635528368 size 0 truncated at 132
 1710864812.470501 [260] [debug] SEI type 1101054320 size 0 truncated at
 136
 1710864812.470550 [260] [debug] SEI type 1101055152 size 0 truncated at
 134
 1710864812.500707 [260] [debug] SEI type 1101054320 size 0 truncated at
 136

 Used version: 4.2.9

 I read ffmpeg source code, the key code as blow:

     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] -
 (int64_t)h->last_pocs[MAX_DELAYED_PIC_COUNT-2] > 2))
         out_of_order = FFMAX(out_of_order, 1);
     if (out_of_order == MAX_DELAYED_PIC_COUNT) {
         av_log(h->avctx, AV_LOG_VERBOSE, "Invalid POC %d<%d\n", cur->poc,
 h->last_pocs[0]);
         for (i = 1; i < MAX_DELAYED_PIC_COUNT; i++)
             h->last_pocs[i] = INT_MIN;
         h->last_pocs[0] = cur->poc;
         cur->mmco_reset = 1;
     } else if(h->avctx->has_b_frames < out_of_order &&
 !sps->bitstream_restriction_flag){
         int loglevel = h->avctx->frame_number > 1 ? AV_LOG_WARNING :
 AV_LOG_VERBOSE;
         av_log(h->avctx, loglevel, "Increasing reorder buffer to %d\n",
 out_of_order);
         h->avctx->has_b_frames = out_of_order;
     }

 It seems that, out_of_order will never be a so large number!!!
 If so, I think hardware decoder will appear "No decoder surfaces left"
 error.
-- 
Ticket URL: <https://trac.ffmpeg.org/ticket/10961>
FFmpeg <https://ffmpeg.org>
FFmpeg issue tracker