Public bug reported:
I am running Lucid with a nVidia Corporation GT216 [GeForce GT 220]. I
have had this exact configuration for > 1 year and it had been running
stable.
Starting approximately 1 month ago, I began to experience strange screen
black out and lock ups when running 3d games. I have seen this with
Trine, Steel Storm, Aquaria, etc.
These lock ups seem to be accompanies by the following messages in
/var/log/messages:
kernel: [100573.805121] NVRM: os_pci_init_handle: invalid context!
Sometimes, I can log into the system via SSH. In one case I managed to
capture dmesg output while the screen was locked. I found the
following:
[92682.852528] NVRM: Xid (0001:00): 16, Head Count 3419
[92682.852607] NVRM: Xid (0001:00): 16, Head 0001 Count 0012
[92881.022338] INFO: task compiz:1590 blocked for more than 120 seconds.
[92881.022344] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this
message.
[92881.022348] compizD 0682 0 1590 1524 0x
[92881.022355] f1177dd0 0086 f5bfccac 0682 c084c760 f592a8e4
c084c760
[92881.022366] 582b9f36 544c c084c760 c084c760 f592a8e4 c084c760 c084c760
f699c400
[92881.022376] 582aa9c2 544c f592a640 7fff 7fff f0c7c688 f1177e30
c058dc2d
[92881.022386] Call Trace:
[92881.022399] [] schedule_timeout+0x1ad/0x280
[92881.022408] [] ? update_curr+0x169/0x2a0
[92881.022413] [] ? update_curr+0x169/0x2a0
[92881.022419] [] ? getnstimeofday+0x56/0x110
[92881.022424] [] wait_for_common+0xa6/0x130
[92881.022429] [] ? default_wake_function+0x0/0x20
[92881.022434] [] wait_for_completion+0x17/0x20
[92881.022680] [] os_acquire_sema+0x77/0x80 [nvidia]
[92881.022928] [] ? _nv002152rm+0xc0/0xc7 [nvidia]
[92881.023188] [] _nv006668rm+0xa/0x21 [nvidia]
[92881.023451] [] ? _nv008443rm+0x34/0x7c [nvidia]
[92881.023674] [] ? _nv004517rm+0x190/0x544 [nvidia]
[92881.023935] [] ? rm_ioctl+0x3e/0x6d [nvidia]
[92881.024144] [] ? nv_kern_ioctl+0x151/0x470 [nvidia]
[92881.024153] [] ? __slab_free+0x7c/0xe0
[92881.024361] [] ? nv_kern_unlocked_ioctl+0x20/0x30 [nvidia]
[92881.024505] [] ? nv_kern_unlocked_ioctl+0x0/0x30 [nvidia]
[92881.024604] [] ? vfs_ioctl+0x21/0x90
[92881.024609] [] ? remove_vma+0x47/0x60
[92881.024614] [] ? do_vfs_ioctl+0x79/0x310
[92881.024618] [] ? sys_ioctl+0x67/0x80
[92881.024623] [] ? syscall_call+0x7/0xb
[92881.024628] [] ? __math_state_restore+0x59/0x90
[92881.024683] INFO: task trine-bin32:8548 blocked for more than 120 seconds.
[92881.024687] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this
message.
[92881.024690] trine-bin32 D 81c6 0 8548 1 0x
[92881.024697] f543ddd0 0086 f1176000 81c6 c084c760 c71002a4
c084c760
[92881.024707] 3fcd17b6 544c c084c760 c084c760 c71002a4 c084c760 c084c760
edb34600
[92881.024716] 544c c710 7fff 7fff f0c7c688 f543de30
c058dc2d
[92881.024726] Call Trace:
[92881.024733] [] schedule_timeout+0x1ad/0x280
[92881.024739] [] ? unix_stream_recvmsg+0x21a/0x460
[92881.024745] [] ? update_curr+0x169/0x2a0
[92881.024750] [] ? getnstimeofday+0x56/0x110
[92881.024755] [] wait_for_common+0xa6/0x130
[92881.024759] [] ? default_wake_function+0x0/0x20
[92881.024764] [] wait_for_completion+0x17/0x20
[92881.024996] [] os_acquire_sema+0x77/0x80 [nvidia]
[92881.025232] [] ? _nv002152rm+0xc0/0xc7 [nvidia]
[92881.025489] [] _nv006668rm+0xa/0x21 [nvidia]
[92881.025749] [] ? _nv008443rm+0x34/0x7c [nvidia]
[92881.025971] [] ? _nv004517rm+0x190/0x544 [nvidia]
[92881.026226] [] ? rm_ioctl+0x3e/0x6d [nvidia]
[92881.026441] [] ? nv_kern_ioctl+0x151/0x470 [nvidia]
[92881.026650] [] ? nv_kern_vma_release+0x6a/0x130 [nvidia]
[92881.026864] [] ? nv_kern_unlocked_ioctl+0x20/0x30 [nvidia]
[92881.027073] [] ? nv_kern_unlocked_ioctl+0x0/0x30 [nvidia]
[92881.027079] [] ? vfs_ioctl+0x21/0x90
[92881.027084] [] ? remove_vma+0x47/0x60
[92881.027088] [] ? do_vfs_ioctl+0x79/0x310
[92881.027092] [] ? sys_ioctl+0x67/0x80
[92881.027097] [] ? syscall_call+0x7/0xb
[92881.027102] [] ? __math_state_restore+0x59/0x90
As I said, the system had been running very stable and I had no similar
problems w/ 3D gaming until about 1 month ago.
To try to work around this I have so far:
Enabled Sync to VBlank in OpenGL Settings
Set PowerMizer to "Max Performance"
Reinstalled Nvidia drivers
Disabled PowerMizer by adding the following to xorg.conf:
Option "RegistryDwords" "PowerMizerEnable=0x1; PerfLevelSrc=0x3322;
PowerMizerDefaultAC=0x1"
Disabled Desktop Effects
Disabled Desktop Sharing
None of these has helped.
I can reproduce this reliably by playing Trine for 5- 10 minutes.
I have full dmesg output and output of nvidia-bug-report.sh if that
helps.
ProblemType: Bug
DistroRelease: Ubuntu 10.04
Package: nvidia-current 195.36.24-0ubuntu1~10.04
ProcVersionSignature: Ubuntu 2.6.32-30.59-generic 2.6.32.29+drm33.13
Uname: Linux 2.6.32-30-generic i686
NonfreeKernelModules