[x264-devel] [Bug Report] Segmentation Fault in 2nd Pass

[Rodolfo J. Quesada Zumbado]

Hi,

I use x264 (libx264.so.59) within MEncoder, and had always had great 
results.

Before starting a batch encoding session I always update to the latest 
head in git and I found recently that encoding will segfault in the 2nd 
pass no matter what, but a few seconds (About 10) after starting it. I 
did updated my x264 git to the latest and greatest just a few minutes 
ago, so the problem persists.

I tested it in three (old) computers, one Athlon XP, Celeron-M notebook 
and a Celeron Tualatin. All have the same errors.

I compile x264 with:

./configure --enable-shared --prefix=/usr/

And build x264 using gcc-4.3.0 under ArchLinux (with the latest updates).

And use it in MEncoder inside a script with the options:

X264ENCOPTS="subq=7:partitions=all:8x8dct:me=umh:frameref=5:bframes=3:\
b_pyramid:weight_b:cabac:deblock:trellis=2:\
bitrate=$BITRATE:threads=$THREADS"

With BITRATE=400 and THREADS="auto".

The divx2pass.log file reports the options as:

#options: 688x432 fps=2997/125 cabac=1 ref=1 deblock=1:0:0 analyse=0x1:0 
me=dia subme=1 brdo=0 mixed_ref=0 me_range=16 chroma_me=1 trellis=0 
8x8dct=0 cqm=0 deadzone=21,11 chroma_qp_offset=0 threads=1 nr=0 
decimate=1 mbaff=0 bframes=3 b_pyramid=1 b_adapt=1 b_bias=0 direct=1 
wpredb=0 bime=0 keyint=250 keyint_min=25 scenecut=40 rc=abr bitrate=400 
ratetol=1.0 rceq='blurCplx^(1-qComp)' qcomp=1.00 qpmin=10 qpmax=51 
qpstep=4 ip_ratio=1.40 pb_ratio=1.30 aq=2:1.00

The errors appears in console as:

18754 Segmentation fault      mencoder -endpos 100 -vf scale=-9:-9 -oac 
pcm -ovc x264 -x264encopts $X264ENCOPTS:$PASS "$1" -o "n$1"

With X264ENCOPTS as stated before and PASS="pass=1:turbo=2" in the first 
pass, and PASS="pass=2" in the second.

And in dmesg as:

mencoder[18754]: segfault at 38032a8d ip b7fd4d00 sp bf87b8c0 error 4 in 
libx264.so.59[b7f81000+b7000]

Before the latest updates, everything was working perfectly (Well, about 
a month ago). I also updated to the latest MPlayer version in their SVN. 
But as the dmesg reports that the segfault occurred in libx264.so.59, 
I'm reporting the bug to this mailing list instead of MPlayer's.

After testing a few old releases from the daily tarball repository, I 
found that the release "x264-snapshot-20080430-2245" doesn't segfaults, 
although it is painfully slow in the first pass, even using turbo=2.

I tested other two releases, x264-snapshot-20080520-2245 and 
x264-snapshot-20080531-2245, both segfaults.

I come from a subversion world, so I don't know much about git (yet), 
but if somebody could hint me how to revert to a previous release I can 
continue testing until I get the commit the produced this segfault.

You are doing a great job with x264, I found that the video quality is 
incredible even at very low bitrates.

-- 
Rodolfo J. Quesada Zumbado
rquesada at roqz.net