Version: 1:7.7+3~deb7u0 xorg-server 2:1.12.4-6 (Julien Cristau <jcristau@debian.org>) My graphic card is a Riva TNT2 M64 I'm not sure which package has the problem. The x crashes inmediatly from the first reboot, returning to text console. Only lasts with the x just fresh installed. It happens, at least, from testing version March 2012 ; but x works whith testing 2012-01-16, only crashing ocassionally although more than usual. I tried in stable 7.0 compiling drm-nouveau ( 2.45 ) , nouveau ( 1.0.6 ) , and kernel ( 3.2.46 ) , all from original sources, without debian patches. It gets more stable with nouveau, no changes with the kernel; the lasting of x is variable , but remains unstable, in the scope of maybe half one hour, even leaving without activity, using a text console. I also compiled xorg-server , kernel and fvwm in a testing version, but not drm-nouveau nor nouveau, and the x continued crashing just seconds after launching them. If you launch repeatedly the x you get continuous crashes getting always the same messages from x. Always the problem is now in libexa. Some testing versions ago, the broken library was random. Particularities of my system are: - Installed just minimum from cd installer - added packages from dvds with apt-get ( included x ) - fvwm - athlon xp - no internet X was perfectly stable with nv and fvwm from potato to squeeze, and is stable with nouveau and slackware14 with kernel 3.2.46 , but has some crashes with 3.4.x ( but different, from the kernel module ) Xorg backtrace : Backtrace: [ 381.001] 0: /usr/bin/X (xorg_backtrace+0x49) [0xb7699769] [ 381.001] 1: /usr/bin/X (0xb751c000+0x181186) [0xb769d186] [ 381.001] 2: (vdso) (__kernel_rt_sigreturn+0x0) [0xb74fd40c] [ 381.001] 3: /usr/lib/xorg/modules/libexa.so (0xb6f16000+0x979f) [0xb6f1f79f] [ 381.001] 4: /usr/lib/xorg/modules/libexa.so (0xb6f16000+0x99d8) [0xb6f1f9d8] [ 381.001] 5: /usr/bin/X (miCopyRegion+0x17c) [0xb7675f6c] [ 381.002] 6: /usr/bin/X (miDoCopy+0x4f0) [0xb76765a0] [ 381.002] 7: /usr/lib/xorg/modules/libexa.so (0xb6f16000+0x80c2) [0xb6f1e0c2] [ 381.002] 8: /usr/bin/X (0xb751c000+0x10929d) [0xb762529d] [ 381.002] 9: /usr/bin/X (0xb751c000+0x37d23) [0xb7553d23] [ 381.002] 10: /usr/bin/X (0xb751c000+0x3c375) [0xb7558375] [ 381.002] 11: /usr/bin/X (0xb751c000+0x29e95) [0xb7545e95] [ 381.002] 12: /lib/i386-linux-gnu/i686/cmov/libc.so.6 (__libc_start_main+0xe6) [0xb71c1e46] [ 381.002] 13: /usr/bin/X (0xb751c000+0x2a1e9) [0xb75461e9] [ 381.002] [ 381.002] Segmentation fault at address 0xbf9135d0 [ 381.002] Fatal server error: [ 381.002] Caught signal 11 (Segmentation fault). Server aborting This backtrace is with nouveau , drm-nouveau and kernel compiled as commented before , but the same , i think , than with debian original packages.
Dear Maintainer, I *think* I see the same problem, at least I have about the same backtrace. I see the problem when displaying pictures e.g. using display. Attached is the gdb backtrace. I you need more debug symbols e.g. for glibc or xserver-xorg-video-nouveau please call back. Many thanks! Hilmar [ 10338.982] Backtrace: [ 10338.983] 0: /usr/bin/Xorg (xorg_backtrace+0x49) [0xb76f57b9] [ 10338.983] 1: /usr/bin/Xorg (0xb7578000+0x1811d6) [0xb76f91d6] [ 10338.983] 2: (vdso) (__kernel_rt_sigreturn+0x0) [0xb755940c] [ 10338.983] 3: /usr/lib/xorg/modules/drivers/nouveau_drv.so (0xb6fc1000+0x5185) [0xb6fc6185] [ 10338.983] 4: /usr/lib/xorg/modules/drivers/nouveau_drv.so (0xb6fc1000+0x6b1c) [0xb6fc7b1c] [ 10338.983] 5: /usr/lib/xorg/modules/drivers/nouveau_drv.so (0xb6fc1000+0x11faf) [0xb6fd2faf] [ 10338.983] 6: /usr/lib/xorg/modules/libexa.so (0xb6f85000+0x95e4) [0xb6f8e5e4] [ 10338.983] 7: /usr/lib/xorg/modules/libexa.so (0xb6f85000+0x99d8) [0xb6f8e9d8] [ 10338.983] 8: /usr/bin/Xorg (miCopyRegion+0x17c) [0xb76d1fbc] [ 10338.983] 9: /usr/bin/Xorg (miDoCopy+0x4f0) [0xb76d25f0] [ 10338.983] 10: /usr/lib/xorg/modules/libexa.so (0xb6f85000+0x80c2) [0xb6f8d0c2] [ 10338.983] 11: /usr/bin/Xorg (0xb7578000+0x1092ed) [0xb76812ed] [ 10338.983] 12: /usr/bin/Xorg (0xb7578000+0x37d23) [0xb75afd23] [ 10338.983] 13: /usr/bin/Xorg (0xb7578000+0x3c375) [0xb75b4375] [ 10338.983] 14: /usr/bin/Xorg (0xb7578000+0x29e95) [0xb75a1e95] [ 10338.983] 15: /lib/i386-linux-gnu/libc.so.6 (__libc_start_main+0xe6) [0xb7224e16] [ 10338.984] 16: /usr/bin/Xorg (0xb7578000+0x2a1e9) [0xb75a21e9] [ 10338.984] [ 10338.984] Segmentation fault at address 0x20 [ 10338.984] Fatal server error: [ 10338.984] Caught signal 11 (Segmentation fault). Server aborting
severity 712545 important stop On 14.02.14 Hilmar Preusse (hille42@web.de) wrote: Hi, I guess debug of the nouveau code is relevant. Here is the bt of the same core dump after installing xserver-xorg-video-nouveau-dbg. Further I think that bug is important -> raising sev. H.
Am 14.02.2014 um 19:42 schrieb Hilmar Preusse:
This might be https://bugs.freedesktop.org/show_bug.cgi?id=71116.
Could you please test whether the attached patch for
xserver-xorg-video-nouveau in wheezy helps?
TIA,
Sven
Hi, I guess I rather look at https://bugs.debian.org/688082 . Is this a patch for it? H.
Am 14.02.2014 um 23:00 schrieb Hilmar Preusse:
I don't think so, but then again I don't really understand why that bug
suddenly had been fixed for Thiemo Nagel.
Cheers,
Sven
Hi, He says, it is fixed latest in 1:1.0.10-1 (Debian testing). Unfortunately I can't upgrade my system to testing. Would it be possible to have a fix in Debian stable? H.
Am 15.02.2014 um 16:01 schrieb Hilmar Preusse:
Yes. What I mean is that I cannot find anything in the commit logs of
xserver-xorg-nouveau which would explain this to me.
You could rebuild libdrm and xserver-xorg-video-nouveau from testing on
wheezy to see if the problem is fixed in current versions.
If we can identify a fix, certainly.
Cheers,
Sven
Hi,
The only one I found was the following from upstream changelog:
commit 2fa3397e348161a3394e2b456f065921272a056a
Author: Ilia Mirkin <imirkin@alum.mit.edu>
Date: Fri May 3 09:59:39 2013 -0400
nouveau_xv: Avoid reading off the end of the source image on NV50+
The 'w' argument to NVCopyNV12ColorPlanes is used to index into the
source image. line_len is rounded up to 8 on NV50+, so if the source
image (+ left offset) is not rounded to 8, NVCopyNV12ColorPlanes could
read past the end of the array and crash X. This change can cause the
last few horizontal pixels of dst to not be initialized, but they
should be truncated by the renderer anyways.
See https://bugs.freedesktop.org/show_bug.cgi?id=63263
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
No, I don't have an NV50+ card, but rather a NV40 card. Could you
have a short look if my problem is related to this issue? If yes,
I'll try to compile a patched package for testing.
Thanks!
H.
Am 15.02.2014 um 22:53 schrieb Hilmar Preusse: Probably not, but I'm not totally sure. Here is the commit diff if you want to try:
Hi, As predicted the patch does not solve my problem. I guess I have to go the hard way of backports to fix my problem (hopefully). Couly you contact upstream if they know about the problem and if it was fixed w/o noticing? Many thanks! H.
Hi, OK, I can't follow Thiemos observation, that nouveau 1.0.10 solves the problem. I now have ii libdrm-nouveau2:i386 2.4.52-1 ii xserver-xorg-video-nouveau 1:1.0.10-1 ii xserver-xorg-video-nouveau-dbg 1:1.0.10-1 The X server still crashes, the gdb backtrace looks about the same (bug seems to be in nouveau code). Do you think that should be reported @upstream? Thanks! Hilmar
Am 19.02.2014 um 15:07 schrieb Hilmar Preusse: They will likely complain about your old kernel and mesa versions, but you can try (I don't think those matter in this case). Please see http://nouveau.freedesktop.org/wiki/Bugs/ for instructions. Cheers, Sven
On 19.02.14 Sven Joachim (svenjoac@gmx.de) wrote: Hi, The nouveau driver is not linked w/ mesa and a userspace program. Both should not matter. <quote> If you are using packages from your distribution, send the bug reports to your distribution and not directly to us. </quote> As first I'd re-open #688082 and then contact Thiemo and the submitter if they still see the problem (or vice versa [first contact, then re-open]). Do you agree? Yes, I can open the ticket @n.f.o/wiki/Bugs later on. Regards, Hilmar
Am 19.02.2014 um 21:18 schrieb Hilmar Preusse:
Many bugs reported against xserver-xorg-video-nouveau are actually
kernel bugs or bugs in the 3D driver (package libgl1-mesa-dri). But not
this one, it seems.
They obviously don't want to deal with outdated distribution packages
(like the ones in Debian stable), but we're shipping the latest versions
of libdrm and the X driver in testing and unstable, and we're not
patching them.
Yes, your latest backtrace looks very similar to the one reported
originally in #688082.
Cheers,
Sven
On 19.02.14 Sven Joachim (svenjoac@gmx.de) wrote: Hi, https://bugs.freedesktop.org/show_bug.cgi?id=70510 Seems to describe my problem. Comment added. H.
On 14.02.14 Sven Joachim (svenjoac@gmx.de) wrote: Hi, For now I've solved my personal problem by pluggin in a ATI Radeon 4650 card and disable the on-board NVidia device -> X-Server is stable now. ;-) H.
Dear Maintainer, *** Please consider answering these questions, where appropriate *** * What led up to the situation? Updated Xorg using aptitude to latest testing branch. Upon restarting lightdm the Xserver crashes after loading the initial GTK-greeter screen, then reloads repeatedly which makes it impossible to login to the machine without rebooting into recovery mode.
Please file your own bug and include the relevant logs. Cheers, Julien
--
Herzliche Glückwünsche !!!
Es wurde eine Spende in Höhe von 1.000.000,00 € geleistet
Zur Verfügung gestellt von der Azim Premji Foundation. Für weitere
Informationen und Ansprüche senden Sie mir jetzt eine E-Mail an:
Charityfoundation299@gmail.com
Danke Azim Prejim