INFO: task * blocked for more than 120 seconds.

[stoogiebuncho]

I've been getting random freeze-ups from time to time. Sometimes I'll go for a few weeks without it happening, sometimes it will happen every day.

When it freezes, the computer becomes completely unresponsive and I have to press Alt+SysRq+R-S-E-I-U-B to get it to reboot.

Here is the kern.log for the latest freeze:

Mar 22 17:07:33 Compy kernel: [ 2520.436029] INFO: task i915/1:339 blocked for more than 120 seconds.
Mar 22 17:07:33 Compy kernel: [ 2520.436034] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
Mar 22 17:07:33 Compy kernel: [ 2520.436039] i915/1        D c08185c0     0   339      2 0x00000000
Mar 22 17:07:33 Compy kernel: [ 2520.436047]  f659df04 00000046 f696581c c08185c0 f6965a88 c08185c0 a77c63e7 00000205
Mar 22 17:07:33 Compy kernel: [ 2520.436059]  c08185c0 c08185c0 f6965a88 c08185c0 a77c5587 00000205 c08185c0 eebc0000
Mar 22 17:07:33 Compy kernel: [ 2520.436070]  f69657f0 f68ba814 f68ba818 ffffffff f659df30 c0572b76 f69025b0 f68ba81c
Mar 22 17:07:33 Compy kernel: [ 2520.436081] Call Trace:
Mar 22 17:07:33 Compy kernel: [ 2520.436093]  [<c0572b76>] __mutex_lock_slowpath+0xc6/0x130
Mar 22 17:07:33 Compy kernel: [ 2520.436100]  [<c0572a90>] mutex_lock+0x20/0x40
Mar 22 17:07:33 Compy kernel: [ 2520.436129]  [<f83af8fa>] i915_gem_retire_work_handler+0x2a/0x70 [i915]
Mar 22 17:07:33 Compy kernel: [ 2520.436137]  [<c01579de>] run_workqueue+0x6e/0x140
Mar 22 17:07:33 Compy kernel: [ 2520.436159]  [<f83af8d0>] ? i915_gem_retire_work_handler+0x0/0x70 [i915]
Mar 22 17:07:33 Compy kernel: [ 2520.436166]  [<c0157b38>] worker_thread+0x88/0xe0
Mar 22 17:07:33 Compy kernel: [ 2520.436173]  [<c015c180>] ? autoremove_wake_function+0x0/0x40
Mar 22 17:07:33 Compy kernel: [ 2520.436179]  [<c0157ab0>] ? worker_thread+0x0/0xe0
Mar 22 17:07:33 Compy kernel: [ 2520.436185]  [<c015be8c>] kthread+0x7c/0x90
Mar 22 17:07:33 Compy kernel: [ 2520.436190]  [<c015be10>] ? kthread+0x0/0x90
Mar 22 17:07:33 Compy kernel: [ 2520.436196]  [<c0104047>] kernel_thread_helper+0x7/0x10

This repeats over and over.

Most of the stuff I've found online says this problem is from the 2.6.26 kernel, but I'm running 2.6.31-20-generic.

I'm at a loss as to what to do next. If anyone has found a workaround for this problem, I would appreciate it.

[stoogiebuncho]

Doesn't seem like I'm getting much response, so I'll ask an additional question.

Has anyone else been having this problem, or is it just me?

Could this be a hardware problem, or is it more likely to be related to the kernel?

Thanks.

[danep]

I have been getting this error only for the cron task, it started after I rebooted for a kernel update a few weeks ago. It's really bad- whenever the error occurs the system becomes unresponsive to SSH and IMAP requests, but apache still seems to return webpages just fine. It usually lasts 30-60 minutes and goes away on its own. Very weird. I would very much appreciate some help tracing the root of this problem.

[computermacgyver]

You're not alone, stoogiebuncho,
https://bugs.launchpad.net/ubuntu/+s...ux/+bug/429321

I'm having the problem as well. It appears to be connected to an out of date video driver in the kernal. I'm still looking for a good solution now.

I think your issue, danep, is different. Those with the above issue report still being able to login with SSH, etc., (I have yet to confirm this in my case).

[computermacgyver]

The problem hasn't happened to me again since diabling visual effects (Compiz Fusion). I have no steps to reliabily reproduce the problem; so, I am not sure if it is solved or not.

Try to disable visual effects by
1) right click on the deskop and select "change desktop background"
2) Click the "Visual Effects" tab
3) Change it to none.
This is a per user setting and (as far as I know) has to be changed for each user using the machine.

[stoogiebuncho]

The problem hasn't happened to me again since diabling visual effects (Compiz Fusion). I have no steps to reliabily reproduce the problem; so, I am not sure if it is solved or not.

Try to disable visual effects by
1) right click on the deskop and select "change desktop background"
2) Click the "Visual Effects" tab
3) Change it to none.
This is a per user setting and (as far as I know) has to be changed for each user using the machine.

Thanks for the information - I'll keep an eye on that bug report. From what it describes, it certainly seems rational that disabling visual effect could solve the problems.

I use Gnome-Do Docky so I'm hesitant to disable compiz since I would lose that functionality, and right now my computer has only been freezing once a week or less. If it becomes more frequent I'll probably give it a try, though.

[tarvid]

Happening to me too. On a server. No X.

Home workstation instable since processor upgrade. Switching from compiz to metacity and firefox to chromium helped.

These are kernel bugs and the only thing I can find in common is load. Recent kernels would rather quit than fight.

[kbron]

Hi guys.

Just wanted to let you know this issue is not exclusive to Ubuntu. I have a CentOS 5.5 suffering from the same problem, to the point that I had to totally disable Xorg in order for the box to keep running more than 10 minutes (I'm booting in runlevel 3)

It seems to be driver for Intel's i915 video card.
(some more info: I know there's a video card that Intel totally dropped the support for Linux, but I don't remember which one was it, nor where I read it .... -- sorry, too lazy to google --)

BTW, it's not the first time I've seen this! Some years ago I had exactly the same issue with an ATI card.

[kbron]

One last thing I almost forgot to mention:

I can confirm that in Ubuntu you can SSH the box and do some forensics such as /var/log analysis and of course dmesg analysis. As a matter of fact, dmesg is the only "log" (so to speak) I've been able to find that really blames the crash to i915 -- which I'm pretty sure it's the cause of this nightmare, by the way. When done you can safely shutdown/reboot the box.

Bad news for CentOS: this issue causes a kernel crash that not even a "Raising Skinny Elephants Is Utterly Boring" sequence will be able to help you
(solution: just a good old cold boot to get the box alive again)

[stoogiebuncho]

Since moving to Ubuntu 10.04 I have not had this problem. Maybe it's just a fluke, but I'm not complaining.