tag 677035 upstream stop On Mon, Jun 11, 2012 at 3:04 PM, <g.bona...@libero.it> wrote: > On Mon, Jun 11, 2012 at 12:05:05PM +0200, Reinhard Tartler wrote: >> >> In order to check if that bug has already been fixed upstream, could >> you please install the libavcodec-extra-53 package from backports.org >> and check if the segmentation fault still occurs? >> > > I've just tried libav-tools 4:0.8-2~bpo60+1 and libavcodec53 segfaults too, so > perhaps the problem is present upstream. > > After my submission, I've noticed that it crashes with horizontal sizes > 626, 632, and 852, but not with sizes 624 and 640. > > Coupled with the name of the top-of-stack function (sad16_sse2), that suggests > a workaround: use horizontal sizes that are multiples of 16. > > Although I've had no time to examine the code (and I'm no expert in assembly) > I guess that the problem might be caused by a wrongly aligned or sized buffer.
That's an interesting observation. This bug needs to be forwarded upstream. http://bugzilla.libav.org -- regards, Reinhard -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
