On Thu, Mar 12, 2015 at 07:41:47PM +0100, Luca Barbato wrote:
> On 12/03/15 19:35, Diego Biurrun wrote:
> >On Thu, Mar 12, 2015 at 04:54:01PM +0100, Niels Möller wrote:
> >>Diego Biurrun <[email protected]> writes:
> >>>I see no more issues, I just want to run this through Oracle, once the
> >>>TDSC decoder is through.
> >>
> >>Thanks for taking care of this!
> >
> >No issues so far, but let's wait until tomorrow for the whole run.
> >
> >>>+                if (chset->downmix_ncoeffs > DCA_XLL_DMIX_NCOEFFS_MAX) {
> >>>+                    av_log(s->avctx, AV_LOG_WARNING,
> >>>+                           "XLL: Skipping %d downmix coefficients, 
> >>>exceeding implementation limit %d\n",
> >>>+                           chset->downmix_ncoeffs, 
> >>>DCA_XLL_DMIX_NCOEFFS_MAX);
> >>>+                    skip_bits_long(&s->gb, 9 * chset->downmix_ncoeffs);
> >>>+                    chset->downmix_ncoeffs = 0;
> >>
> >>I still think it would make sense with a return AVERROR_PATCHWELCOME
> >>here. The alternatives are to either
> >>
> >>1. Make the code reading chset->downmix_coeffs (that's close to the end
> >>    of ff_dca_xll_decode_audio) check for chset->downmix_ncoeffs == 0,
> >>    and do something not too insane, or error out at that time.
> >>
> >>2. Allocate chset->downmix_coeffs dynamically.
> >>
> >>Otherwise, we risk reading garbage, as well as reading beyond the end of
> >>the buffer.
> >
> >I'll go for the return AVERROR, as in the attached patch, OK?
> 
> I'd ask for samples.

Changed locally.

Diego
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to