X server hung, kernel said "task blocked for >120 secs" (2.6.30)

X server hung, kernel said "task blocked for >120 secs" (2.6.30)

[Sanjoy Mahajan]

I'm running vanilla 2.6.30 on an otherwise Debian 'unstable' system -- a
TP T60 with Intel graphics and wireless (no tainting).  The X server
[xorg 1.6.1.901 (1.6.2 RC 1), intel driver 2.7.1] hung in a way that it
often has with other versions: The mouse cursor moves around as normal,
but the keyboard or mouse buttons don't do anything.  I used sysrq half
competently to send TERM and KILL signals and then reboot.

The log message below was in the /var/log/messages from when the problem
happened.  Should I report this information to freedesktop.org or is it
a kernel issue belonging on the bugzilla?

[228600.676053] INFO: task events/0:9 blocked for more than 120 seconds.
[228600.676058] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[228600.676061] events/0      D 31efa204     0     9      2
[228600.676066]  c1bfc040 00000046 f7121a60 31efa204 c0439040 c0436134 c0101f70 f708f0c0
[228600.676075]  f708f274 00000000 00000000 f7121a8c c1bfc040 00000000 c011d842 c1bfc580
[228600.676083]  f708f0c0 f708f274 0366353a f626cac0 00000000 c0439040 c02f19c2 c03a8300
[228600.676090] Call Trace:
[228600.676099]  [<c0101f70>] ? __switch_to+0xbf/0x140
[228600.676105]  [<c011d842>] ? pick_next_task_fair+0x80/0x87
[228600.676111]  [<c02f19c2>] ? __schedule+0x6f2/0x74d
[228600.676115]  [<c02f1fc4>] ? __mutex_lock_common+0xe1/0x136
[228600.676120]  [<c02f2028>] ? __mutex_lock_slowpath+0xf/0x11
[228600.676123]  [<c02f1e78>] ? mutex_lock+0x10/0x1e
[228600.676127]  [<c02f1e78>] ? mutex_lock+0x10/0x1e
[228600.676133]  [<c0133e97>] ? queue_delayed_work_on+0x9c/0xa8
[228600.676150]  [<f8ff74f3>] ? i915_gem_retire_work_handler+0x1c/0x4e [i915]
[228600.676155]  [<c0133701>] ? worker_thread+0x13d/0x1bf
[228600.676168]  [<f8ff74d7>] ? i915_gem_retire_work_handler+0x0/0x4e [i915]
[228600.676174]  [<c01365a6>] ? autoremove_wake_function+0x0/0x2d
[228600.676178]  [<c01335c4>] ? worker_thread+0x0/0x1bf
[228600.676182]  [<c01362b6>] ? kthread+0x42/0x67
[228600.676186]  [<c0136274>] ? kthread+0x0/0x67
[228600.676190]  [<c0103ab7>] ? kernel_thread_helper+0x7/0x10
[228611.402991] SysRq : Keyboard mode set to system default

Re: X server hung, kernel said "task blocked for >120 secs" (2.6.30)

[Andrew Morton]

(cc dri-devel)

On Sun, 14 Jun 2009 21:46:05 -0400
Sanjoy Mahajan <sanjoy@MIT.EDU> wrote:

> I'm running vanilla 2.6.30 on an otherwise Debian 'unstable' system -- a
> TP T60 with Intel graphics and wireless (no tainting).  The X server
> [xorg 1.6.1.901 (1.6.2 RC 1), intel driver 2.7.1] hung in a way that it
> often has with other versions: The mouse cursor moves around as normal,
> but the keyboard or mouse buttons don't do anything.  I used sysrq half
> competently to send TERM and KILL signals and then reboot.
> 
> The log message below was in the /var/log/messages from when the problem
> happened.  Should I report this information to freedesktop.org or is it
> a kernel issue belonging on the bugzilla?
> 
> [228600.676053] INFO: task events/0:9 blocked for more than 120 seconds.
> [228600.676058] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
> [228600.676061] events/0      D 31efa204     0     9      2
> [228600.676066]  c1bfc040 00000046 f7121a60 31efa204 c0439040 c0436134 c0101f70 f708f0c0
> [228600.676075]  f708f274 00000000 00000000 f7121a8c c1bfc040 00000000 c011d842 c1bfc580
> [228600.676083]  f708f0c0 f708f274 0366353a f626cac0 00000000 c0439040 c02f19c2 c03a8300
> [228600.676090] Call Trace:
> [228600.676099]  [<c0101f70>] ? __switch_to+0xbf/0x140
> [228600.676105]  [<c011d842>] ? pick_next_task_fair+0x80/0x87
> [228600.676111]  [<c02f19c2>] ? __schedule+0x6f2/0x74d
> [228600.676115]  [<c02f1fc4>] ? __mutex_lock_common+0xe1/0x136
> [228600.676120]  [<c02f2028>] ? __mutex_lock_slowpath+0xf/0x11
> [228600.676123]  [<c02f1e78>] ? mutex_lock+0x10/0x1e
> [228600.676127]  [<c02f1e78>] ? mutex_lock+0x10/0x1e
> [228600.676133]  [<c0133e97>] ? queue_delayed_work_on+0x9c/0xa8
> [228600.676150]  [<f8ff74f3>] ? i915_gem_retire_work_handler+0x1c/0x4e [i915]
> [228600.676155]  [<c0133701>] ? worker_thread+0x13d/0x1bf
> [228600.676168]  [<f8ff74d7>] ? i915_gem_retire_work_handler+0x0/0x4e [i915]
> [228600.676174]  [<c01365a6>] ? autoremove_wake_function+0x0/0x2d
> [228600.676178]  [<c01335c4>] ? worker_thread+0x0/0x1bf
> [228600.676182]  [<c01362b6>] ? kthread+0x42/0x67
> [228600.676186]  [<c0136274>] ? kthread+0x0/0x67
> [228600.676190]  [<c0103ab7>] ? kernel_thread_helper+0x7/0x10
> [228611.402991] SysRq : Keyboard mode set to system default

I assume this is a regression?  Since 2.6.29?

Thanks.

Re: X server hung, kernel said "task blocked for >120 secs" (2.6.30)

[Sanjoy Mahajan]

[Have been away for 10 days and just saw the message]

> I assume this is a regression?  Since 2.6.29?

I never saw it before using 2.6.30 and haven't been able to reproduce
it even running 2.6.30.  

So it could be a regression.  Or it could be a recurring problem (I had
a few X lockups with earlier kernels) now exposed by a blocked-task
timer (the "INFO: task events/0:9 blocked for more than 120 seconds"
message) -- if that notification code is new in 2.6.30?

-Sanjoy

`Until lions have their historians, tales of the hunt shall always
 glorify the hunters.'  --African Proverb

Re: X server hung, kernel said "task blocked for >120 secs" (2.6.30)

[Eric Anholt]

[-- Attachment #1: Type: text/plain, Size: 907 bytes --]

On Wed, 2009-07-01 at 12:48 -0400, Sanjoy Mahajan wrote:
> [Have been away for 10 days and just saw the message]
> 
> > I assume this is a regression?  Since 2.6.29?
> 
> I never saw it before using 2.6.30 and haven't been able to reproduce
> it even running 2.6.30.  
> 
> So it could be a regression.  Or it could be a recurring problem (I had
> a few X lockups with earlier kernels) now exposed by a blocked-task
> timer (the "INFO: task events/0:9 blocked for more than 120 seconds"
> message) -- if that notification code is new in 2.6.30?

This message appears when the GPU is hung but somebody is still trying
to use it.  The problem is that the GPU is hung, not the presence of the
message (so you can't use the message to indicate any particular bug,
and the bug is almost always in userland).

-- 
Eric Anholt
eric@anholt.net                         eric.anholt@intel.com



[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 197 bytes --]