Re: [FFmpeg-devel] [PATCH] avcodec/vp9_superframe_split_bsf: Check in size

[Andreas Rheinhardt <andreas.rheinhardt@outlook.com>]

Michael Niedermayer:
> Fixes: Out of array read
> Fixes: 45137/clusterfuzz-testcase-minimized-ffmpeg_BSF_VP9_SUPERFRAME_SPLIT_fuzzer-4984270639202304
> 
> Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
> Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
> ---
>  libavcodec/vp9_superframe_split_bsf.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/libavcodec/vp9_superframe_split_bsf.c b/libavcodec/vp9_superframe_split_bsf.c
> index ed0444561a..481484a4f0 100644
> --- a/libavcodec/vp9_superframe_split_bsf.c
> +++ b/libavcodec/vp9_superframe_split_bsf.c
> @@ -51,7 +51,7 @@ static int vp9_superframe_split_filter(AVBSFContext *ctx, AVPacket *out)
>              return ret;
>          in = s->buffer_pkt;
>  
> -        marker = in->data[in->size - 1];
> +        marker = in->size ? in->data[in->size - 1] : 0;
>          if ((marker & 0xe0) == 0xc0) {
>              int length_size = 1 + ((marker >> 3) & 0x3);
>              int   nb_frames = 1 + (marker & 0x7);

There is a second place in this BSF where data might be read in the
absence of data, namely if one of the frames in a superframe have size
of zero (its attempted to read its profile; no actual read takes place
due to the checks of the get_bits API, but it is nevertheless invalid
data). See
https://patchwork.ffmpeg.org/project/ffmpeg/patch/20200530160541.29517-7-andreas.rheinhardt@gmail.com/;
also see
https://patchwork.ffmpeg.org/project/ffmpeg/patch/20200530160541.29517-11-andreas.rheinhardt@gmail.com/
and
https://patchwork.ffmpeg.org/project/ffmpeg/patch/20200530160541.29517-1-andreas.rheinhardt@gmail.com/

- Andreas
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".