#598818 xorg: X crashes with segmentation fault when Xine window is out of the screen

Package:
xserver-xorg-video-nouveau
Source:
xserver-xorg-video-nouveau
Description:
X.Org X server -- Nouveau display driver
Submitter:
ProgVal
Date:
2010-10-04 09:33:12 UTC
Severity:
important
#598818#5
Date:
2010-10-02 11:32:59 UTC
From:
To:
Hello,

I was trying to look for a movie with Xine. Then, I accidentally moved the
windows out of my screen (the left border), and X crashed.
I launched it again (I always run X with startx as a non-root user), tryed to
reproduce the bug, and it appears again.

I made several other tests, as viewing another video, or viewing the first one
with DragonPlayer : no problem.
Then, I tryed again to reproduce it with the first video in Xine at the bottom,
the right : nothing happens.
Then, I tested to make it got out of the screen at the left, nothing happened
instanly, but, after some random moves, X crashed again.

Now, I am unable to reproduce the bug again.


The movie's name is Sintel.2010.2K.Theora-VODO.mp4 (I say that because it seems
to have some influence on the bug)

Best regards,
ProgVal

#598818#10
Date:
2010-10-02 12:12:16 UTC
From:
To:
Hi.

ProgVal <progval@gmail.com> (02/10/2010):

Can you please attach X log of the crash, as well as dmesg?

Not sure which one it is, but Debian's 2.6.32 includes many
DRM-related fixes from later kerrnel versions, which should improve
the “graphics experience”.

Mraw,
KiBi.

#598818#15
Date:
2010-10-02 12:28:55 UTC
From:
To:
Hi back
0: /usr/bin/X (xorg_backtrace+0x28) [0x4a1f88]
1: /usr/bin/X (0x400000+0x64a99) [0x464a99]
2: /lib/libpthread.so.0 (0x7f67ec62c000+0xef60) [0x7f67ec63af60]
3: /usr/lib/xorg/modules/drivers/nouveau_drv.so (0x7f67e914a000+0x9ce4)
[0x7f67e9153ce4]
4: /usr/bin/X (0x400000+0x12dbd0) [0x52dbd0]
5: /usr/lib/xorg/modules/extensions/libextmod.so (0x7f67ea220000+0x134bd)
[0x7f67ea2334bd]
6: /usr/bin/X (0x400000+0x31594) [0x431594]
7: /usr/bin/X (0x400000+0x25bca) [0x425bca]
8: /lib/libc.so.6 (__libc_start_main+0xfd) [0x7f67eb130c4d]
9: /usr/bin/X (0x400000+0x25779) [0x425779]
Segmentation fault at address 0x7f67e5b96000

Fatal server error:
Caught signal 11 (Segmentation fault). Server aborting

I am not sure I understand that, but I'll try to answer.
I got the sources using apt-get source ; I didn't edit anything out of the
menuconfig.

I don't have any non-free package on my system (except of the wifi driver),
and, I didn't edit anything about the graphics (Xorg, Nouveau, ...)

Best regards,
ProgVal

#598818#20
Date:
2010-10-02 12:39:00 UTC
From:
To:
ProgVal <progval@gmail.com> (02/10/2010):

Hi again :)

OK, good, “just” a segfault. Is there any chance to install debugging
packages (xserver-xorg-core-dbg and xserver-xorg-video-nouveau-dbg),
and to run X inside gdb, so as to get a full backtrace?

If that makes things easier for you, you can get a core dump first,
and then get the full backtrace out of it, post-mortem.

Yeah, I failed to phrase my question properly. The idea was to
determine whether you were running “just” a 2.6.32(.y) upstream
release, or whether you had some patches on top of it. Since you're
using Debian's source, that's alright, you should have DRM patches,
which is good. (It might be nice to know which exact kernel version
you fetched through apt-get source, though.)

Mraw,
KiBi.

#598818#25
Date:
2010-10-02 13:19:12 UTC
From:
To:
When I edit /etc/X11/xinit/xserverrc to replace
exec /usr/bin/X -nolisten tcp "$@"
by
exec /usr/lib/debug/usr/bin/Xorg -nolisten tcp "$@"
I get an error saying me there is a syntax error (a parenthesis is missing
at line 1)



Excuse me if I don't do what I should do, but I'm new to Debian

#598818#30
Date:
2010-10-02 13:23:50 UTC
From:
To:
ProgVal <progval@gmail.com> (02/10/2010):

No problem. :) Run X as usual, then attach it in gdb (outside X) using
“attach $pid”, then type “continue”.

Mraw,
KiBi.

#598818#35
Date:
2010-10-02 14:04:22 UTC
From:
To:
I tryed to do it, but attaching gdb to X makes all my computer freeze after
pressing Ctrl + Alt + F7 (whether I type "continue" or not). I run startx
from tty1 and gdb from tty2.

But I have a new idea about the why of the bug : it has more chances to
appear if I've underclocked my CPUs and if the Xine window is bigger.