On Tue, Aug 18, 2020 at 10:25 PM Jan Ekström <jee...@gmail.com> wrote: > > The first two commits fix both of the fuzzing samples I have on hand. > One being from #8845, and another provided privately by Michael. Changes have > been tested both with clang 10's ASAN as well as standard valgrind. >
For the record I have a feeling that the actual reason for the issues is an underlying issue where a ChannelElement in a list gets allocated, but then also freed (yet not actually set to nullptr?), and my not strict enough validation based on valid samples just happened to bring it to the surface. Since I got publicly hurried and called out to "Please fix it or revert ASAP!", here is the thing that anyone sane enough will attempt to do to get people off their backs to get more breathing room: Here's more stringent checks so that 22.2 will only be probed if the configuration aligns exactly as it does for valid streams, and it seems to remove the symptoms with regards to all of the provided fuzzed samples. I hope y'all have much more fun time than I have. Jan _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
