[FFmpeg-devel] [PATCH] avcodec/amr_parser: passthrough if channels is unknown

[Paul B Mahol]

On Tue, Oct 5, 2021 at 11:03 PM James Almer <jamrial at gmail.com> wrote:

> On 10/5/2021 5:51 PM, Michael Niedermayer wrote:
> > Fixes: division by 0
> > Fixes:
> 39562/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_AMRWB_fuzzer-5448834960982016
> > Fixes:
> 39589/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_AMRWB_fuzzer-6119205334810624
> >
> > Found-by: continuous fuzzing process
> https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
> > Signed-off-by: Michael Niedermayer <michael at niedermayer.cc>
> > ---
> >   libavcodec/amr_parser.c | 2 +-
> >   1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/libavcodec/amr_parser.c b/libavcodec/amr_parser.c
> > index 79258d4d0cf..9fb7711fee0 100644
> > --- a/libavcodec/amr_parser.c
> > +++ b/libavcodec/amr_parser.c
> > @@ -62,7 +62,7 @@ static int amr_parse(AVCodecParserContext *s1,
> >       *poutbuf_size = 0;
> >       *poutbuf = NULL;
> >
> > -    if (s1->flags & PARSER_FLAG_COMPLETE_FRAMES) {
> > +    if (s1->flags & PARSER_FLAG_COMPLETE_FRAMES || !avctx->channels) {
>
> No line in the parser divides by avctx->channels. There's however one
> doing a modulo avctx->channels. Wouldn't this mean the compiler is being
> overzealous with optimizations?
>
> Also, instead of this please do the same as the decoder, and set
> channels to 1 and layout to mono if nothing is already set. Forcing the
> "complete frames" path when no channels are set feels like a hacky
> workaround.
> If the relevant parser flag is not set, then you must not treat packets
> as if they contained complete frames.
>

Agreed also with that solution.


>
> >           next = buf_size;
> >       } else {
> >           int ch, offset = 0;
> >
>
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>
> To unsubscribe, visit link above, or email
> ffmpeg-devel-request at ffmpeg.org with subject "unsubscribe".
>