[FFmpeg-devel] [PATCH 3/3] avcodec/huffyuvdec: Check input buffer size

[Paul B Mahol]

On 2/5/18, Michael Niedermayer <michael at niedermayer.cc> wrote:
> On Mon, Feb 05, 2018 at 10:04:47AM +0100, Paul B Mahol wrote:
>> On 2/5/18, Michael Niedermayer <michael at niedermayer.cc> wrote:
>> > On Thu, Feb 01, 2018 at 02:36:16AM +0100, Michael Niedermayer wrote:
>> >> On Wed, Jan 31, 2018 at 07:56:06PM +0100, Paul B Mahol wrote:
>> >> > On 1/31/18, Michael Niedermayer <michael at niedermayer.cc> wrote:
>> >> > > Fixes: Timeout
>> >> > > Fixes: 5487/clusterfuzz-testcase-4696837035393024
>> >> > >
>> >> > > 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/huffyuvdec.c | 3 +++
>> >> > >  1 file changed, 3 insertions(+)
>> >> > >
>> >> > > diff --git a/libavcodec/huffyuvdec.c b/libavcodec/huffyuvdec.c
>> >> > > index 979c4b9d5c..66357bfb40 100644
>> >> > > --- a/libavcodec/huffyuvdec.c
>> >> > > +++ b/libavcodec/huffyuvdec.c
>> >> > > @@ -919,6 +919,9 @@ static int decode_frame(AVCodecContext *avctx,
>> >> > > void
>> >> > > *data, int *got_frame,
>> >> > >      AVFrame *const p = data;
>> >> > >      int table_size = 0, ret;
>> >> > >
>> >> > > +    if (buf_size < (width * height + 7)/8)
>> >> > > +        return AVERROR_INVALIDDATA;
>> >> > > +
>> >> >
>> >> > Are you sure this is enough?
>> >>
>> >> I dont know if thats the only way the decoder can be made to waste
>> >> large amounts of CPU with little input data
>> >>
>> >> I do belive it stops this specific class of issues though
>> >>
>> >>
>> >> >
>> >> > Something similar you had already posted long ago.
>> >>
>> >> for other decoders, yes. Did i forget a patch for huffyuv ?
>> >
>> > i will apply this in a few days unless someone has objections or
>> > sees some possible imrpovment
>>
>> Are you sure this does not break decoding of valid files?
>
> huffyuv encodes samples using huffman codes, the smallest is 1 bit so
> w*h bits should be the minimum
>
>

ok