[MPlayer-users] Re: mencoder crash

Corey Hickey bugfood-ml at fatooh.org
Thu Sep 11 07:44:21 CEST 2003


D Richard Felker III wrote:
> [Automatic answer: RTFM (read DOCS, FAQ), also read DOCS/bugreports.html]
> On Wed, Sep 10, 2003 at 10:49:07PM -0500, Jonathan Rogers wrote:
> 
>>[Automatic answer: RTFM (read DOCS, FAQ), also read DOCS/bugreports.html]
>>Michael Niedermayer wrote:
>>
>>>On Tuesday 09 September 2003 12:01, Jonathan Rogers wrote:
>>>
>>>>The crash happened right at the end of a movie or show, though not at
>>>>the end of the VOB, right? Did you concatenate a bunch of titles? I've
>>>>observed that the last chapter of a DVD title is usually empty. Maybe
>>>>it's just an end marker. I've also observed the crash you had, usually
>>>>right at the end of the movie. Coincidence? I don't think so.
>>>
>>>this may be "fixed" in latest ffmpeg developer cvs (= tomorrows ffmpeg 
>>>annonymous cvs) allthough the correct fix would be to not reinit the 
>>>encoder at all ...
>>
>>Do you know why mencoder feels it needs to reinitialize the encoder? Is 
>>it some peculiarity of DVD VOBs, or just a mencoder peculiarity?
> 
> 
> It's not mencoder per se, but either vd_ffmpeg.c or libavcodec. For
> some reason it thinks something has changed in the stream which
> requires reinit, but reinit is broken in mencoder.
> 
> BTW, I've never seen this happen, and I assume it does not happen
> unless you cat vob files together,

No, actually (in my experience) it has a tendency to happen right at the
end of encoding a file produced by -dumpstream. I don't know if that's
exactly the same problem, but it looks similar. An old mail I wrote is
here:
http://www.mplayerhq.hu/pipermail/mplayer-dev-eng/2003-August/020145.html

It's possible to dd the dumpstream down to the very end (last 150k or
so) and have the problem be reproducible.
A file which exhibits problems:
http://bugfood.casa-z.org/mpbug/end-of-dvd-segfault.vob

A description of how to reproduce:
http://bugfood.casa-z.org/mpbug/end-of-dvd-segfault.txt

A file which DOES NOT seem to have cause the problem:
http://bugfood.casa-z.org/mpbug/nosegfault.vob


Last night I sent a mail to Michael Niedermayer, and he promptly
responded with a couple patches (attached). I didn't have a chance to
test them until this afternoon, but I've been throwing various DVDs at
different versions of mencoder all evening.

The second patch Michael sent is the one soon to appear in anonymous
ffmpeg CVS. I applied it, but it didn't seem to solve the problem for 
me. (Note that not all files make the problem occur.)

The first patch applies to mplayer CVS from April 28th, and appears to
fix quite nicely the related problem of divx2pass.log getting filled
with zeroes when using vmax_b_frames. I manually applied the patch to
current CVS, built mplayer, and am currently testing. I'll know in about
an hour if it's successful.

-Corey
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: fix_double_vo_config.patch
URL: <http://lists.mplayerhq.hu/pipermail/mplayer-users/attachments/20030910/0e6ba409/attachment.asc>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: list
URL: <http://lists.mplayerhq.hu/pipermail/mplayer-users/attachments/20030910/0e6ba409/attachment.txt>


More information about the MPlayer-users mailing list