Re: "BUG: soft lockup - CPU#0 stuck for 61s! [kswapd0:184]"

Re: "BUG: soft lockup - CPU#0 stuck for 61s! [kswapd0:184]"

[KOSAKI Motohiro]

> Hi,
> 
> I just wanted to report a bug upstream that is affecting the latest
> versions of at least both Fedora and Ubuntu. CPUs somehow lock up
> under load, producing errors of the form "BUG: soft lockup - CPU#0
> stuck for 61s! [kswapd0:184]"
> 
> The Fedora Bug report is here:
> https://bugzilla.redhat.com/show_bug.cgi?id=649694 -- however you can
> find lots of references to the error message on other distributions
> (including Ubuntu) by googling  "bug soft lockup cpu stuck".
> 
> Lockups seem to happen on server-class hardware under heavy loads when
> the machine is swapping.  This can lead to the entire machine locking
> up in some reported cases (although so far only individual CPUs seem
> to have locked up in my case, not the entire machine).  The point at
> which the CPU hangs varies -- see the dmesg output I attached to the
> Fedora bug report above.
> 
> My machine is a 12-way Xeon X5680 system with ext3, AFS and XFS
> filesystems (XFS is running on hardware RAID).  Please let me know if
> you need other info that would be helpful to diagnosing the problem.

AFAIK, This isssue was already fixed by Mel.

http://kerneltrap.org/mailarchive/linux-kernel/2010/10/27/4637977



--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>

Re: "BUG: soft lockup - CPU#0 stuck for 61s! [kswapd0:184]"

[Luke Hutchison]

On Tue, Nov 9, 2010 at 12:33 AM, KOSAKI Motohiro
<kosaki.motohiro@jp.fujitsu.com> wrote:
> AFAIK, This isssue was already fixed by Mel.
>
> http://kerneltrap.org/mailarchive/linux-kernel/2010/10/27/4637977

Yes, based on where the CPU lockups were occurring
(zone_nr_free_pages, zone_watermark_ok), this fix does seem to address
the problem I described.  I assume the other lockup points
(_raw_spin_unlock_irqrestore, find_next_bit, sleeping_prematurely,
test_tsk_thread_flag) are also caused by the NR_FREE_PAGES problem?

Thank you for the link, I'll put it into the Fedora bug report and
hopefully a fix will be pushed out sometime soon.

Luke

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>