[arch-general] khugepaged hangs and filesystem unresponsive

[pants]

Good evening,

I just experienced a major problem with my system while listening to a
music file in mpd from an xfs filesystem over a mdadm raid6.  A kernel
error was thrown, with the following error.log entry:

output: /var/log/error.log
> Aug 26 23:34:50 localhost kernel: [283781.061258] xhci_hcd 0000:0b:00.0: ERROR Transfer event TRB DMA ptr not part of current TD
> Aug 26 23:34:50 localhost kernel: [283781.062268] xhci_hcd 0000:0b:00.0: ERROR Transfer event TRB DMA ptr not part of current TD
> Aug 26 23:34:50 localhost kernel: [283781.063273] xhci_hcd 0000:0b:00.0: ERROR Transfer event TRB DMA ptr not part of current TD
> Aug 26 23:34:50 localhost kernel: [283781.064245] xhci_hcd 0000:0b:00.0: ERROR Transfer event TRB DMA ptr not part of current TD
> Aug 26 23:34:51 localhost kernel: [283782.058901] timeout: still 1 active urbs..
> Aug 26 23:38:48 localhost kernel: [284019.080666] INFO: task mpd:707 blocked for more than 120 seconds.
> Aug 26 23:38:48 localhost kernel: [284019.080696] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
> Aug 26 23:40:48 localhost kernel: [284139.071419] INFO: task khugepaged:32 blocked for more than 120 seconds.
> Aug 26 23:40:48 localhost kernel: [284139.071451] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
> Aug 26 23:40:48 localhost kernel: [284139.071589] INFO: task mpd:525 blocked for more than 120 seconds.
> Aug 26 23:40:48 localhost kernel: [284139.071613] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
> Aug 26 23:40:48 localhost kernel: [284139.071721] INFO: task mpd:707 blocked for more than 120 seconds.
> Aug 26 23:40:48 localhost kernel: [284139.071744] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
> Aug 26 23:40:48 localhost kernel: [284139.071943] INFO: task mplayer:28316 blocked for more than 120 seconds.
> Aug 26 23:40:48 localhost kernel: [284139.071968] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
> Aug 26 23:42:48 localhost kernel: [284259.062189] INFO: task khugepaged:32 blocked for more than 120 seconds.
> Aug 26 23:42:48 localhost kernel: [284259.062220] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
> Aug 26 23:42:48 localhost kernel: [284259.062358] INFO: task mpd:525 blocked for more than 120 seconds.
> Aug 26 23:42:48 localhost kernel: [284259.062382] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
> Aug 26 23:42:48 localhost kernel: [284259.062489] INFO: task mpd:702 blocked for more than 120 seconds.
> Aug 26 23:42:48 localhost kernel: [284259.062512] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
> Aug 26 23:42:48 localhost kernel: [284259.062688] INFO: task mpd:703 blocked for more than 120 seconds.
> Aug 26 23:42:48 localhost kernel: [284259.062712] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
> Aug 26 23:42:48 localhost kernel: [284259.062829] INFO: task mpd:704 blocked for more than 120 seconds.
> Aug 26 23:42:48 localhost kernel: [284259.062852] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.

Attempts to access other files on the same filesystem after the incident
caused the applications used to also go into interruptible sleep (see
the mplayer processes that appear later in the log).  I was forced to
kill and unmount what I could, then force the system down.  Afterwards,
I could replicate the error by attempting to read the file in question
at the same point

Even if you have no solution, pointing me towards the relevant kernel
mailing list would be very helpful.

Thanks,

pants.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 490 bytes
Desc: not available
URL: <http://mailman.archlinux.org/pipermail/arch-general/attachments/20120827/1c348e7a/attachment.asc>