[FFmpeg-devel] [PATCH 2/3] avcodec/iff: Check bpp for validity

[Peter Ross]

On Wed, Jun 19, 2019 at 11:44:36AM +0200, Michael Niedermayer wrote:
> On Wed, Jun 19, 2019 at 07:34:19PM +1000, Peter Ross wrote:
> > On Wed, Jun 19, 2019 at 01:53:02AM +0200, Michael Niedermayer wrote:
> > > Fixes: shift exponent -100663046 is negative
> > > Fixes: out of array access
> > > Fixes: 15270/clusterfuzz-testcase-minimized-ffmpeg_AV_CODEC_ID_IFF_ILBM_fuzzer-5727829913763840
> > > 
> > > 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/iff.c | 4 ++--
> > >  1 file changed, 2 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/libavcodec/iff.c b/libavcodec/iff.c
> > > index 33cf2e3a94..48a0340604 100644
> > > --- a/libavcodec/iff.c
> > > +++ b/libavcodec/iff.c
> > > @@ -382,7 +382,7 @@ static av_cold int decode_init(AVCodecContext *avctx)
> > >      IffContext *s = avctx->priv_data;
> > >      int err;
> > >  
> > > -    if (avctx->bits_per_coded_sample <= 8) {
> > > +    if (avctx->bits_per_coded_sample <= 8U) {
> > >          int palette_size;
> > >  
> > >          if (avctx->extradata_size >= 2)
> > > @@ -391,7 +391,7 @@ static av_cold int decode_init(AVCodecContext *avctx)
> > >              palette_size = 0;
> > >          avctx->pix_fmt = (avctx->bits_per_coded_sample < 8) ||
> > >                           (avctx->extradata_size >= 2 && palette_size) ? AV_PIX_FMT_PAL8 : AV_PIX_FMT_GRAY8;
> > > -    } else if (avctx->bits_per_coded_sample <= 32) {
> > > +    } else if (avctx->bits_per_coded_sample <= 32U) {
> > >          if (avctx->codec_tag == MKTAG('R', 'G', 'B', '8')) {
> > >              avctx->pix_fmt = AV_PIX_FMT_RGB32;
> > >          } else if (avctx->codec_tag == MKTAG('R', 'G', 'B', 'N')) {
> > > -- 
> > > 2.21.0
> > 
> > hey michael, can you post the test case file?
> >       
> > AVCodecParameters.bits_per_coded_sample is signed, but looking at *libavformat*/iff.c,
> > i can't fathom how its getting below zero.
> 
> even better i can explain it
> in target_dec_fuzzer.c there is:
>         ctx->bits_per_coded_sample              = bytestream2_get_le32(&gbc);
> 
> quite possibly this is not possble with libavformat and this codec currently.
> A API user from his own demuxer or a future demuxer in libavformat could set
> it though.

understood, this patch is ok.

i fear there are many of other places where signed test is done on bits_per_coded sample
leading to shift explosion. given enough time, fuzz will eventually catch them all.

how radical of a change would just making AVCodecParameters.bits_per_coded_sample unsigned be?

-- Peter
(A907 E02F A6E5 0CD2 34CD 20D2 6760 79C5 AC40 DD6B)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 195 bytes
Desc: not available
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20190619/565fa2bd/attachment.sig>