On Tue, Sep 02, 2014 at 01:59:50AM +0200, Michael Niedermayer wrote: > On Mon, Sep 01, 2014 at 11:25:57PM +0200, Reimar Döffinger wrote: > > On Mon, Sep 01, 2014 at 10:59:51PM +0200, Reimar Döffinger wrote: > > > On Mon, Sep 01, 2014 at 10:19:36PM +0200, Michael Niedermayer wrote: > > > > On Mon, Sep 01, 2014 at 08:59:43PM +0200, Reimar Döffinger wrote: > > > > > Signed-off-by: Reimar Döffinger <reimar.doeffin...@gmx.de> > > > > > --- > > > > > libavcodec/mpeg12.c | 21 ++++++++++----------- > > > > > libavcodec/mpegvideo.c | 12 ++++++++---- > > > > > libavcodec/rl.h | 8 ++------ > > > > > 3 files changed, 20 insertions(+), 21 deletions(-) > > > > > > > > assuming the amount is ok on the stack the patch LGTM > > > > > > I reduced the mpeg12 to the 680 required - since that's a > > > local function I think it's unnecessary to be conservative there. > > > For the other one... > > > Well, HP-UX seems to use 16 kB stack. I guess that we'd have > > > problems there (probably even without the patch). But I > > > don't think we'll have any volunteers to test and fix FFmpeg > > > on that (I doubt anyone would use HP-UX voluntarily at all..). > > > Next closest is OpenBSD with 64 kB (though maybe that is outdated > > > information?). > > > We'll be testing that, but of course that is small enough that > > > it can easily fail if the calling application already used up quite > > > a bit of it (and the ff_init_vlc_sparse uses up another 3 kB). > > > On the other hand, ffv1enc uses a lot more than that: > > > uint8_t best_state[256][256]; > > > (I suspect that should probably be fixed > 64 kB I'd surely consider > > > unreasonable, and it will definitely crash on older OpenBSD at least). > > > > Btw. should we maybe set up a fate test with ulimit -s 64 (or so - > > higher might be ok as well)? > > this sounds like a great idea > i could easily do it but my boxes already run quite a few fate clients > so i wont add a new one for that > but if noone else sets up a seperate one for that then someone ping me > and ill add it to a random existing one
I also have 30+ FATE instance on the same machine, including very slow and consuming runs (multiple valgrind memcheck/helgrind/drd, tsan, ...) so I'd prefer not to add this one either, sorry. -- Clément B.
pgpLkFCXIBguU.pgp
Description: PGP signature
_______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel