Xorg freeze after "switch user" applet is invoked

Bug #577997 reported by BAndy
34
This bug affects 7 people
Affects Status Importance Assigned to Milestone
NVIDIA Drivers Ubuntu
Invalid
Undecided
Unassigned
xf86-video-ati
Invalid
Undecided
Unassigned
nvidia-graphics-drivers-180 (Ubuntu)
Invalid
Undecided
Unassigned

Bug Description

Binary package hint: xorg

After the 2nd invocation of the "switch user" applet in order to switch back to the original user, the screen goes blank and no login prompt appears. I can login to the machine remotely, which shows X is running at 100% CPU usage. Restarting GDM fails to restart X. The only way I've found to proceed is to forceably kill (ie kill -9) the X process at which point the GDM login screen reappears. This is ubuntu 10.04 (which was just upgraded from 9.10).

ProblemType: Bug
DistroRelease: Ubuntu 10.04
Package: xorg 1:7.5+5ubuntu1
ProcVersionSignature: Ubuntu 2.6.32-22.33-generic 2.6.32.11+drm33.2
Uname: Linux 2.6.32-22-generic i686
NonfreeKernelModules: nvidia
Architecture: i386
Date: Sun May 9 22:20:42 2010
DkmsStatus: nvidia-173, 173.14.22, 2.6.32-22-generic, i686: installed
ProcCmdLine: root=UUID=6ab876bb-15b6-478e-8d89-8da260d8f4e7 ro
ProcEnviron:
 LANGUAGE=en_GB:en
 PATH=(custom, no user)
 LANG=en_GB.utf8
 SHELL=/bin/bash
SourcePackage: xorg
Symptom: display
Title: Xorg freeze
dmi.bios.date: 05/19/2006
dmi.bios.vendor: Intel Corp.
dmi.bios.version: NT94510J.86A.3869.2006.0519.1533
dmi.board.asset.tag: Base Board Asset Tag
dmi.board.name: D945GTP
dmi.board.vendor: Intel Corporation
dmi.board.version: AAC97834-304
dmi.chassis.type: 2
dmi.modalias: dmi:bvnIntelCorp.:bvrNT94510J.86A.3869.2006.0519.1533:bd05/19/2006:svn:pn:pvr:rvnIntelCorporation:rnD945GTP:rvrAAC97834-304:cvn:ct2:cvr:
system:
 distro: Ubuntu
 codename: lucid
 architecture: i686
 kernel: 2.6.32-22-generic

Revision history for this message
BAndy (apb-deactivatedaccount) wrote :
Bryce Harrington (bryce)
affects: xorg (Ubuntu) → nvidia-graphics-drivers-180 (Ubuntu)
Revision history for this message
BAndy (apb-deactivatedaccount) wrote :

The switch user crash hang I reported here seems to be related to :

https://bugs.launchpad.net/ubuntu/+source/xserver-xorg-video-ati/+bug/529882

Revision history for this message
Brien Hunsaker (brien-hunsaker) wrote :

I have the same problem but I have an ati card

Revision history for this message
BAndy (apb-deactivatedaccount) wrote :

It happened again when I forgot about switch-user triggering this bug, but this time the end of my Xorg.0.log file says :

(II) NVIDIA(0): Initialized GPU GART.
[mi] EQ overflowing. The server is probably stuck in an infinite loop.

Backtrace:
0: /usr/bin/X (xorg_backtrace+0x3b) [0x80e937b]
1: /usr/bin/X (mieqEnqueue+0x1ab) [0x80e8b6b]
2: /usr/bin/X (xf86PostMotionEventP+0xd2) [0x80c2d02]
3: /usr/lib/xorg/modules/input/evdev_drv.so (0x326000+0x48a1) [0x32a8a1]
4: /usr/lib/xorg/modules/input/evdev_drv.so (0x326000+0x4b96) [0x32ab96]
5: /usr/bin/X (0x8048000+0x6d5bf) [0x80b55bf]
6: /usr/bin/X (0x8048000+0x122794) [0x816a794]
7: (vdso) (__kernel_sigreturn+0x0) [0xdc4400]

Revision history for this message
Adrian Popa (adrian-popa-gh) wrote :

I'm not sure if it's exactly the same bug, but when my X crashes (I have two displays), I can see the mouse pointer jumping violently between the edges of the two screens. I am not using user-switcher - I am just using desktop applications (some terminals, firefox, thunderbird and a touch of java).
I can't be 100% sure, but the problem appeared after installing (and rebooting to) the new kernel 2.6.32-22-generic.
I have the binary Nvidia drivers, but they haven't been a problem so far.

I will try to see if the bug reproduces with the previous kernel (I have only upgraded to 10.04 for 4 days, so my previous experience might not be concusive).

The end of my Xorg.log file is:
(WW) Jun 04 16:46:00 NVIDIA(1): WAIT (0, 6, 0x8000, 0x000057f0, 0x000057f0)
(WW) Jun 04 16:46:04 NVIDIA(1): WAIT (2, 6, 0x8000, 0x00004300, 0x000089b4)
(WW) Jun 04 16:46:04 NVIDIA(1): WAIT (0, 6, 0x8000, 0x000089b4, 0x000089b4)
(WW) Jun 04 16:46:07 NVIDIA(1): WAIT (2, 6, 0x8000, 0x00005980, 0x00009d5c)
[mi] EQ overflowing. The server is probably stuck in an infinite loop.

Backtrace:
0: /usr/bin/X (xorg_backtrace+0x3b) [0x80e937b]
1: /usr/bin/X (mieqEnqueue+0x1ab) [0x80e8b6b]
2: /usr/bin/X (xf86PostMotionEventP+0xd2) [0x80c2d02]
3: /usr/lib/xorg/modules/input/evdev_drv.so (0x21c000+0x48a1) [0x2208a1]
4: /usr/lib/xorg/modules/input/evdev_drv.so (0x21c000+0x4b96) [0x220b96]
5: /usr/bin/X (0x8048000+0x6d5bf) [0x80b55bf]
6: /usr/bin/X (0x8048000+0x122794) [0x816a794]
7: (vdso) (__kernel_sigreturn+0x0) [0x7da400]
(WW) Jun 04 16:46:08 NVIDIA(1): WAIT (0, 6, 0x8000, 0x00009d5c, 0x00009d5c)

By the way, I'm using Xfce, but the same nvidia driver version and kernel... Let me know if you need more logs.

Revision history for this message
Adrian Popa (adrian-popa-gh) wrote :

I've managed to reproduce the same bug on a different PC, running nvidia 96 driver and also having a dual display. The only difference is that I didn't get a stack trace in Xorg.0.log when I killed X. The trigger in all cases has been moving the mouse cursor between the screens (going from one screen to the other).
AndyB - can you try to reproduce the issue by either moving the mouse pointer quickly between your screens, or by dragging a window between the two screens? If what I'm seeing is the same bug as what you have, you should have a crash independent of the login switcher...
By the way, it seems it's not connected to the kernel update - I tested on systems before and after the update and I reproduced the bug.

Regards.

Revision history for this message
BAndy (apb-deactivatedaccount) wrote :

I don't have two screens so can't reproduce a crash that way. Apart from the user-switcher app causing the crash, the only other time I've seen it is after a normal (first user) logout - when X re-initializes it just hangs using 100% CPU before the gdm login window appears.

Revision history for this message
Adrian Popa (adrian-popa-gh) wrote :

I've been able to work around the problem by limiting the mouse and window transitions between screens. It seems the problem is not how many transitions you make between the screens, but how fast you make them. If you mostly work in in between the screens, you have a higher chance of hitting this bug. Anyway, I'm still afraid for my session stability (anyone can crash my X by wiggling the mouse while the screensaver is on) so I'm looking for a fix.

Revision history for this message
Bodinux (bodinux) wrote :

Switching user using the nvidia drivers leads to a crash of my system. My solution was to remove the nvidia driver.

I am using a single screen.

Revision history for this message
Launchpad Janitor (janitor) wrote :

Status changed to 'Confirmed' because the bug affects multiple users.

Changed in nvidia-graphics-drivers-180 (Ubuntu):
status: New → Confirmed
Revision history for this message
Hans Deragon (deragon) wrote :

Same problem on 12.04 with ATI Radeon HD 4770 using the open source video driver.

Revision history for this message
In , Hans Deragon (deragon) wrote :

Login as a user in Ubuntu 12.04, then switch to another user. Screen becomes blank and inactivated (LCD monitor does not receive any signal).

Occurs with ATI Radeon HD 4770. Problem also reported by others with NVidia driver, thus seam unrelated to driver. Not sure how to classify this bug.

Tracked on LaunchPad Bug #577997 (https://bugs.launchpad.net/nvidia-drivers-ubuntu/+bug/577997).

Changed in xf86-video-ati:
importance: Unknown → High
status: Unknown → Confirmed
Revision history for this message
dino99 (9d9) wrote :

That version is no more supported; please open a new bug report if the actual archive found version also has the same issue.

Changed in nvidia-graphics-drivers-180 (Ubuntu):
status: Confirmed → Invalid
Changed in nvidia-drivers-ubuntu:
status: New → Invalid
Changed in xf86-video-ati:
importance: High → Undecided
status: Confirmed → New
status: New → Invalid
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.