Am 19.08.19 um 23:22 schrieb Michael Niedermayer: > On Mon, Aug 19, 2019 at 05:09:37PM +0200, Thilo Borgmann wrote: >> Am 19.08.19 um 14:27 schrieb Michael Niedermayer: >>> On Mon, Aug 19, 2019 at 07:41:43AM +0200, Thilo Borgmann wrote: >>>> Am 19.08.19 um 01:30 schrieb Michael Niedermayer: >>>>> There seems to be no limit in the specification and upto 64k could be >>>>> stored >>>>> 64 is chooses as limit as thats also used for AAC and is what a int64 mask >>>>> can handle >>>>> >>>>> An alternative to this patch would be a max_channels variable >>>> >>>> There's a conformance file containing 512 channels, that should be the >>>> default max value. >>> >>> will apply with that value later >> >> Decoding of that is already stopped via FF_SANE_NB_CHANNELS in >> lavc/internal.h. >> That is currently set to 256U. I guess pushing that to 512U might already be >> enough without any change to single decoders? > > the problem of out of memory is that the als decoder allocates some pretty > large things per channel. A check on channels needs to happen before > this. > The existing checks didnt achieve that. > > I can of course post a patchset that uses FF_SANE_NB_CHANNELS in als and > bump it to the next number if people prefer this ? > > Note though that increasing FF_SANE_NB_CHANNELS may also increase timeouts > in other decoders.
I see the problem. However, right now alsdec is fine with any number of channels and the CLI prevents decoding anything above FF_SANE_NB_CHANNELS (checks are in lavc/decoding.c and lavc/utils.c; this would also stop any FATE test on 512 channels file). So having alsdec check for a value higher than FF_SANE_NB_CHANNELS (like 512) would have no effect for CLI users, right? -Thilo >> Once fixed, I will add a FATE test for the 512 channel conformance file. > > thx > >> >> -Thilo >> _______________________________________________ >> 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". > _______________________________________________ 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".
