Zhao Zhili:
> Remove the check on dv_demux since dv_fctx will leak if allocate
> dv_demux failed.
> ---
> libavformat/mov.c | 7 +++----
> 1 file changed, 3 insertions(+), 4 deletions(-)
>
> diff --git a/libavformat/mov.c b/libavformat/mov.c
> index adc52de947..f179b6efdd 100644
> --- a/libavformat/mov.c
> +++ b/libavformat/mov.c
> @@ -7357,10 +7357,9 @@ static int mov_read_close(AVFormatContext *s)
> av_freep(&sc->coll);
> }
>
> - if (mov->dv_demux) {
> - avformat_free_context(mov->dv_fctx);
> - mov->dv_fctx = NULL;
> - }
> + av_freep(&mov->dv_demux);
> + avformat_free_context(mov->dv_fctx);
> + mov->dv_fctx = NULL;
>
> if (mov->meta_keys) {
> for (i = 1; i < mov->meta_keys_count; i++) {
>
I found samples [1] to test the dv audio codepath; it indeed showed two
kinds of leaks: The leaks of AVBufferRef and AVBuffer per packet (fixed
by my patch) and the leak of the DVDemuxContext (fixed by your patch).
The latter leak is a regression since
edb1af7c466ebb28bfdb0c076e498e527b43d24f.
I intend to apply the two patches in two days if there are no objections
from other devs (or if the maintainer of mov.c doesn't do it earlier).
- Andreas
[1]: Use the samples from
http://samples.ffmpeg.org/ffmpeg-bugs/trac/ticket4671/dir1.tar.bz2
together with the enable_drefs option set.
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
