[FFmpeg-devel] [PATCH v2 1/2] avcodec/vvc_mp4toannexb: check bytes left for nalu_len

[Andreas Rheinhardt]

Nuo Mi:
> On Fri, Feb 9, 2024 at 7:40 PM Andreas Rheinhardt <
> andreas.rheinhardt at outlook.com> wrote:
> 
>> Nuo Mi:
>>> Fixes: fuzzer timeout
>>> Fixes:
>> 65253/clusterfuzz-testcase-minimized-ffmpeg_BSF_VVC_MP4TOANNEXB_fuzzer-4972412487467008
>>>
>>> Found-by: continuous fuzzing process
>> https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
>>> Signed-off-by: Michael Niedermayer <michael at niedermayer.cc>
>>> Signed-off-by: Andreas Rheinhardt <andreas.rheinhardt at outlook.com>
>>> ---
>>>  libavcodec/bsf/vvc_mp4toannexb.c | 7 ++++---
>>>  1 file changed, 4 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/libavcodec/bsf/vvc_mp4toannexb.c
>> b/libavcodec/bsf/vvc_mp4toannexb.c
>>> index 25c3726918..36bdae8f49 100644
>>> --- a/libavcodec/bsf/vvc_mp4toannexb.c
>>> +++ b/libavcodec/bsf/vvc_mp4toannexb.c
>>> @@ -155,10 +155,11 @@ static int vvc_extradata_to_annexb(AVBSFContext
>> *ctx)
>>>          }
>>>
>>>          for (j = 0; j < cnt; j++) {
>>> -            int nalu_len = bytestream2_get_be16(&gb);
>>> +            const int nalu_len = bytestream2_get_be16(&gb);
>>>
>>> -            if (4 + AV_INPUT_BUFFER_PADDING_SIZE + nalu_len >
>>> -                SIZE_MAX - new_extradata_size) {
>>> +            if (!nalu_len ||
>>> +                nalu_len > bytestream2_get_bytes_left(&gb) ||
>>> +                4 + AV_INPUT_BUFFER_PADDING_SIZE + nalu_len > SIZE_MAX
>> - new_extradata_size) {
>>>                  ret = AVERROR_INVALIDDATA;
>>>                  goto fail;
>>>              }
>>
>> What about growing the packet?
>>
> Hi Andreas,
> Do you mean growing the packet only once for all nalus?
> However, this would change the original behavior and result in more
> duplicate code between the HEVC and VVC implementations.
> I can do it, but I'll refactor duplications to h2656_mp4toannexb.c first.
> Do you think that's okay?
> Thank you
> 

I meant a simple check before growing the packet to ensure that there is
enough input left as the header claims there to be. Your original patch
added such a check, so I presumed that there is none. But there is one
(both here and in hevc_mp4toannexb). So my above point is moot.

- Andreas