Michael Niedermayer <[email protected]> added the comment:

On Wed, Jun 03, 2009 at 07:51:26PM +0000, Reimar Döffinger wrote:
> 
> Reimar Döffinger <[email protected]> added the comment:
> 
> On Wed, Jun 03, 2009 at 07:32:41PM +0000, Baptiste Coudurier wrote:
> > 
> > Baptiste Coudurier <[email protected]> added the comment:
> > 
> > On 6/3/2009 11:24 AM, Reimar Döffinger wrote:
> > > Reimar Döffinger <[email protected]> added the comment:
> > > 
> > > On Mon, Jun 01, 2009 at 09:30:15PM +0000, Baptiste Coudurier wrote:
> > >> Baptiste Coudurier <[email protected]> added the comment:
> > >>
> > >> Attached valgrind output.
> > > 
> > > Huh? That doesn't show anything really wrong.
> > 
> > Huh? Did you see all the multiple uinitialized reads ?
> > Are they really harmless ?
> 
> Well, maybe not exactly harmless (though they might be), but they in
> themselves are almost certainly not the cause of the crash.
> So let me call them "at least in comparison harmless".
> Finally valgrind managed to reach the problematic point,
> these are some of the errors it gives:
> 
> ==4536== Invalid read of size 1
> ==4536==    at 0x59E1C0: ff_er_frame_end (error_resilience.c:979)
> ==4536==    by 0x5B7ED0: ff_h263_decode_frame (h263dec.c:705)
> ==4536==    by 0x495ABC: avcodec_decode_video2 (utils.c:577)
> ==4536==    by 0x408276: output_packet (ffmpeg.c:1255)
> ==4536==    by 0x40B4FE: av_encode (ffmpeg.c:2203)
> ==4536==    by 0x40BD1B: main (ffmpeg.c:3958)
> ==4536==  Address 0x5a47de0 is 0 bytes after a block of size 188,944 alloc'd
> ==4536==    at 0x4A065A0: memalign (vg_replace_malloc.c:431)
> ==4536==    by 0x4A0665A: posix_memalign (vg_replace_malloc.c:540)
> ==4536==    by 0x8340F1: av_malloc (mem.c:66)
> ==4536==    by 0x496160: avcodec_default_get_buffer (utils.c:312)
> ==4536==    by 0x4C5605: ff_alloc_picture (mpegvideo.c:195)
> ==4536==    by 0x4C6AB5: MPV_frame_start (mpegvideo.c:920)
> ==4536==    by 0x5B7CA7: ff_h263_decode_frame (h263dec.c:626)
> ==4536==    by 0x495ABC: avcodec_decode_video2 (utils.c:577)
> ==4536==    by 0x408276: output_packet (ffmpeg.c:1255)
> ==4536==    by 0x40B4FE: av_encode (ffmpeg.c:2203)
> ==4536==    by 0x40BD1B: main (ffmpeg.c:3958)
> 
> ==4536== Invalid write of size 1
> ==4536==    at 0x59E75A: ff_er_frame_end (error_resilience.c:55)

this one is on put_dc()
are the mb_width&mb_height variables still correct when this happens?
any width/height changes in this video

btw, wget: unable to resolve host address `fs1.fex.hit.ua'

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

It is dangerous to be right in matters on which the established authorities
are wrong. -- Voltaire

_____________________________________________________
FFmpeg issue tracker <[email protected]>
<https://roundup.ffmpeg.org/roundup/ffmpeg/issue1135>
_____________________________________________________

Reply via email to