On Tue, Apr 28, 2020 at 4:12 PM Lynne <d...@lynne.ee> wrote: > > Apr 28, 2020, 14:05 by mattias.wad...@gmail.com: > > > On Tue, Apr 28, 2020 at 1:59 PM Lynne <d...@lynne.ee> wrote: > > > >> > >> This makes decoding far more robust, since OggS, the ogg magic, > >> can be commonly found randomly in streams, which previously made > >> the demuxer think there's a new stream or a change in such. > >> > >> Patch attached. > >> > > > > Maybe nitpick, could seek back even longer than size on crc mismatch? > > > > Thanks a lot for reviewing the patches. > I was thinking about seeking back further to perhaps the version byte, but > unfortunately, that's > not possible. If ffio_ensure_seekback is called multiple times, the last call > overwrites the previous > calls and we lose the ability to seek before it. > Since the only place at which we know the exact size of the page is when its > signaled, that's the > only point we can ensure we can seek back to. > Before that, we can seek back from the header's end to the version byte. > Unfortunately, that > would mean verifying the header before the checksum, which as you've pointed > out, is bad > for robustness. > Still, this is definitely an improvement, since previously the demuxer didn't > seek back at all.
I see, thanks for the explaining. Also better than in my patch where it would skip the whole page. > Anyway, I've implemented checking the version byte after reading the CRC as > you suggested > and have attached the new patch. Nice, test files works fine now Would it make sense to conditionally ignore crc mismatch based on s->error_recognition & AV_EF_CRCCHECK ? > > _______________________________________________ > 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". _______________________________________________ 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".