Recherche avancée

Sur d’autres sites (9049)

• Why am I getting FPE when using swresample 1.1 ?

  24 janvier 2013, par mystafer

  I am building a project to view a video feed from an IP camera in Android using FFmpeg 1.1.

  I'm attempting to use swresample in an Android project and getting a floating point exception when calling swr_convert. I stepped through the swresample code and found one line in libswresample/swresample.c function swri_realloc_audio where the variables a->bps and a->ch_count are zero causing the FPE.

  int swri_realloc_audio(AudioData *a, int count){
  
      int i, countb;
  
      AudioData old;
  
  
  
      LOGD("in swri_realloc_audio - bps[%d], ch_count[%d]", a->bps, a->ch_count);
  
      if(count < 0 || count > INT_MAX/2/a->bps/a->ch_count)
  
      return AVERROR(EINVAL);

  01-21 17:29:09.612 : D/swresample.c(18789) : in swri_realloc_audio - bps[0], ch_count[0]

  I found bug ticket #1834 in the FFmpeg project that sounds like the exact same issue, but it was resolved by calling swr_init. However, my code does call this function and still crashes. Here is my JNI code :

  SwrContext* resampleCtx = swr_alloc_set_opts(NULL,
  
          AV_CH_LAYOUT_MONO, AV_SAMPLE_FMT_S16, pAudioCodecCtx->sample_rate,
  
          pAudioCodecCtx->channel_layout, pAudioCodecCtx->sample_fmt,
  
          pAudioCodecCtx->sample_rate, 0, 0);
  
  
  
  swr_init(resampleCtx);
  
  LOGD("Resample context initialized");
  
  
  
  int dataSize = swr_convert(resampleCtx,
  
          &pAudioOutBuffer, AVCODEC_MAX_AUDIO_FRAME_SIZE / 2,
  
          (const uint8_t**) &(pFrame->data[0]), pFrame->nb_samples);
  
  LOGD("Resample conversion complete");
  
  
  
  swr_free(&resampleCtx);
  
  LOGD("Obtained data size - dataSize[%d]", dataSize);

  I'm confused because I don't seem to have any control over the variable a in the swri_realloc_audio function. I stepped through the code and noticed that it is from the variable resampleCtx->postin. This variable is copied from resampleCtx->in in the swr_init function, but I don't see where in is ever set to anything.

  What am I doing wrong ? Is it in my code or is there a problem in swresample ?

• avutil/common : add av_rint64_clip

  1er novembre 2015, par Ganesh Ajjanagadde

  avutil/common : add av_rint64_clip
  The rationale for this function is reflected in the documentation for
  
  it, and is copied here :
  Clip a double value into the long long amin-amax range.
  
  This function is needed because conversion of floating point to integers when
  
  it does not fit in the integer’s representation does not necessarily saturate
  
  correctly (usually converted to a cvttsd2si on x86) which saturates numbers
  
  > INT64_MAX to INT64_MIN. The standard marks such conversions as undefined
  
  behavior, allowing this sort of mathematically bogus conversions. This provides
  
  a safe alternative that is slower obviously but assures safety and better
  
  mathematical behavior.
  
  API :
  
  @param a value to clip
  
  @param amin minimum value of the clip range
  
  @param amax maximum value of the clip range
  
  @return clipped value
  Note that a priori if one can guarantee from the calling side that the
  
  double is in range, it is safe to simply do an explicit/implicit cast,
  
  and that will be far faster. However, otherwise this function should be
  
  used.
  avutil minor version is bumped.
  Reviewed-by : Ronald S. Bultje <rsbultje@gmail.com>
  
  Signed-off-by : Ganesh Ajjanagadde <gajjanagadde@gmail.com>
  

  • [DH] libavutil/common.h
  • [DH] libavutil/version.h

• ffmpeg quits if one output stream fails - can this be prevented ? [duplicate]

  5 décembre 2017, par Caius Jard

  This question already has an answer here :

  • ffmpeg rtmp and local file output
    
    1 answer
    
  • FFMPEG : How to stream to multiple outputs with the same encoding independently
    
    2 answers
    

  I’m responsible for maintaining a device that streams a live webcam feed to a remote relay server, and simultaneously writes a version of the stream to the local disk. It does this by a single instance of ffmpeg that has two outputs - one to the local disk, and one over rtsp to the streaming server

  I’m encountering a problem where by if the streaming server disconencts for any reason, ffmpeg quits. I’m not really bothered if the live stream is lost, but it’s a big problem that the local recording is lost also - it’s not a huge detriment to the particular business process if it cannot be watched live, but losing the locally stored copy is a disaster

  ffmpeg is started with a command line similar to :

  ffmpeg -thread_queue_size 4096 -async 1 -f v4l2 
  
    -input_format mjpeg -framerate 30 -video_size 1280x720 
  
    -i /dev/video0 -thread_queue_size 4096 -async 1 -f alsa 
  
    -i plughw:CARD=Set,DEV=0 -r 30 -c:a aac -b:a 96k -c:v h264 -b:v 983040 
  
    -profile:v baseline -preset veryfast -pix_fmt yuv420p 
  
    -f tee -map 0:v -map 1:a 
  
      [f=matroska]'/var/recordings/yyyy-mm-dd/backup.mkv'|
  
      [f=rtsp:rtsp_transport=tcp]rtsp://streamingserver.com:1234/session.sdp`

  Is there any way (command line switch etc) that ffmpeg can be made to carry on if an output stream is lost, rather than quitting ?