Hello everyone, I hope that someone can point me into the right direction with 
this issue.

I have a decoding loop that can optionally use HW acceleration.
The code has been tested on different machines with both Intel and NVIDIA 
graphic cards and works without issues in most cases.
I'm seeing a memory leak though during a test on a laptop with the following 
specs:


  *   Intel i7
  *   NVIDIA quadro p520
  *   File being played is a H.264 1920x1080
  *   HW device is AV_HWDEVICE_TYPE_D3D11VA

The leak will eat up all the available memory very quickly and crash the 
machine.

When running on this machine avcodec_receive_frame(d->avctx, cur_frame) will 
always return -11 (AVERROR(EAGAIN)) while avcodec_send_packet executes without 
any issue and it is continuously called even if avcodec_receive_frame is not 
dequeuing any frame from AVCodecContext*  d->avctx.

av_hwdevice_ctx_create is executed without any problem and creating the context 
correctly.

cur_frame is not allocated and de-allocated at each step of the decoding loop 
as in 
hw_decode.c<https://github.com/FFmpeg/FFmpeg/blob/master/doc/examples/hw_decode.c>
 example but is allocated only once inside a init() function:
d->valid_frame = av_frame_alloc();
d->sw_valid_frame = av_frame_alloc();

and destroyed once in a destroy() function:
av_frame_free(&d->valid_frame);
av_frame_free(&d->sw_valid_frame);
 av_packet_unref(&d->pkt);
 avcodec_free_context(&d->avctx);

The leak seems is not due to any missing de-allocation.
Any help is much appreciated. Thank you!

Andrea
This email and any attachments to it may be confidential and are intended 
solely for the use of the individual to whom it is addressed.
If you are not the intended recipient of this email, you must neither take any 
action based upon its contents, nor copy or show it to anyone.
Please contact the sender if you believe you have received this email in error. 
_______________________________________________
Libav-user mailing list
[email protected]
https://ffmpeg.org/mailman/listinfo/libav-user

To unsubscribe, visit link above, or email
[email protected] with subject "unsubscribe".

Reply via email to