On Fri, Aug 28, 2020 at 11:01 PM Xiaohan Wang (王消寒) <[email protected]> wrote: > > Resend with @chromium.org account. Sorry for the noise.
I think something like this could be the underlying issue that got hit when the 22.2 check was too lax and set incorrect parameters earlier in the parsing process? What this defines it as is: "TYPE_SCE, 0th of this type, AAC_CHANNEL_OFF" I guess, which maybe is OK? I do see quite a few definitions of this in various places in libavcodec/aacdec_template.c, though, with `git grep -E "layout_map\[MAX_ELEM_ID.*\*.*4\]\[3\]"`. I do wonder if those should also be similarly initialized? For the record, the sanity check needed to be fixed anyways, and that also fixed the fuzzing issues reported upstream: http://git.videolan.org/?p=ffmpeg.git;a=commit;h=d6f293353c94c7ce200f6e0975ae3de49787f91f Jan _______________________________________________ ffmpeg-devel mailing list [email protected] https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email [email protected] with subject "unsubscribe".
