[FFmpeg-devel] [PATCH 3/3] avformat/mov: Check for EOF in mov_read_meta()

[James Almer]

On 9/15/2019 4:34 PM, Michael Niedermayer wrote:
> On Sat, Aug 31, 2019 at 10:36:32AM -0300, James Almer wrote:
>> On 8/31/2019 5:47 AM, Michael Niedermayer wrote:
>>> On Fri, Aug 30, 2019 at 08:57:29PM -0300, James Almer wrote:
>>>> On 8/30/2019 8:25 PM, Michael Niedermayer wrote:
>>>>> Fixes: Timeout (195sec -> 2ms)
>>>>> Fixes: 16735/clusterfuzz-testcase-minimized-ffmpeg_DEMUXER_fuzzer-5090676403863552
>>>>>
>>>>> Found-by: continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
>>>>> Signed-off-by: Michael Niedermayer <michael at niedermayer.cc>
>>>>> ---
>>>>>  libavformat/mov.c | 5 ++++-
>>>>>  1 file changed, 4 insertions(+), 1 deletion(-)
>>>>>
>>>>> diff --git a/libavformat/mov.c b/libavformat/mov.c
>>>>> index 675b915906..46c544b61f 100644
>>>>> --- a/libavformat/mov.c
>>>>> +++ b/libavformat/mov.c
>>>>> @@ -4419,7 +4419,10 @@ static int mov_read_custom(MOVContext *c, AVIOContext *pb, MOVAtom atom)
>>>>>  static int mov_read_meta(MOVContext *c, AVIOContext *pb, MOVAtom atom)
>>>>>  {
>>>>>      while (atom.size > 8) {
>>>>> -        uint32_t tag = avio_rl32(pb);
>>>>> +        uint32_t tag;
>>>>> +        if (avio_feof(pb))
>>>>> +            return AVERROR_EOF;
>>>>> +        tag = avio_rl32(pb);
>>>>>          atom.size -= 4;
>>>>>          if (tag == MKTAG('h','d','l','r')) {
>>>>>              avio_seek(pb, -8, SEEK_CUR);
>>>>
>>>> Maybe do something like "while (atom.size > 8 && !avio_feof(pb))"
>>>> instead, which is similar to the loop in mov_read_default.
>>>
>>> Can do but why ?
>>> the code in the patch returns an error if the atom is truncated
>>> the change suggested does not return an error if the atom is truncated
>>> on its own this doesnt sound better
>>>
>>> Thanks
>>
>> There's Marton's "avformat/utils: return pending IO error on EOF in
>> av_read_frame()" patch to check in generic code if avio_feof() != 0 is
>> an actual EOF or an IO error, so if you make this code here simply break
>> the loop, same as it's done in mov_read_default(), then said generic
>> code would be triggered and return the proper error code once the
>> current packet is done processing.
> 
> are you against the original patch in this thread ?
> from reading this its not clear to me if you dislike the original
> patch or not ?
> 
> thanks

No, I'm not.