On Mon, Apr 01, 2024 at 11:35:56PM +0200, Andreas Rheinhardt wrote:
> Michael Niedermayer:
> > Fixes: memleak
> > Fixes: 67714/clusterfuzz-testcase-minimized-ffmpeg_dem_MOV_fuzzer-5671570999476224
> > 
> > Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
> > Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
> > ---
> >  libavformat/demux_utils.c | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/libavformat/demux_utils.c b/libavformat/demux_utils.c
> > index 86f551245be..96e6e20d1ec 100644
> > --- a/libavformat/demux_utils.c
> > +++ b/libavformat/demux_utils.c
> > @@ -123,9 +123,9 @@ int ff_add_attached_pic(AVFormatContext *s, AVStream *st0, AVIOContext *pb,
> >      if (!st && !(st = avformat_new_stream(s, NULL)))
> >          return AVERROR(ENOMEM);
> >      pkt = &st->attached_pic;
> > +    av_packet_unref(pkt);
> >      if (buf) {
> >          av_assert1(*buf);
> > -        av_packet_unref(pkt);
> >          pkt->buf  = *buf;
> >          pkt->data = (*buf)->data;
> >          pkt->size = (*buf)->size - AV_INPUT_BUFFER_PADDING_SIZE;
> 
> This seems to be from the ff_add_attached_pic() call in
> mov_read_chapters() with the referenced stream having been created in
> mov_read_covr(). The latter does not set a proper id -- it just takes
> what avformat_new_stream() sets as id on every new stream (namely zero).
> So it makes no real sense to compare it to the ids contained in
> chapter_tracks (can really every track be reinterpreted as chapter
> track?). But I am no mov/mp4 expert.
> Anyway, does the following fix it?
> 
> diff --git a/libavformat/mov.c b/libavformat/mov.c
> index 7bdeeb99f9..51d97296f1 100644
> --- a/libavformat/mov.c
> +++ b/libavformat/mov.c
> @@ -8820,7 +8820,7 @@ static void mov_read_chapters(AVFormatContext *s)
> 
>          if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO) {
>              st->disposition |= AV_DISPOSITION_ATTACHED_PIC |
> AV_DISPOSITION_TIMED_THUMBNAILS;
> -            if (sti->nb_index_entries) {
> +            if (!st->attached_pic.data && sti->nb_index_entries) {
>                  // Retrieve the first frame, if possible
>                  AVIndexEntry *sample = &sti->index_entries[0];
>                  if (avio_seek(sc->pb, sample->pos, SEEK_SET) !=
> sample->pos) {
> s

Yes, this fixes it too, please apply and backport as needed

thx

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Homeopathy is like voting while filling the ballot out with transparent ink.
Sometimes the outcome one wanted occurs. Rarely its worse than filling out
a ballot properly.