Bug 12574 - possible circular locking dependency detected
Summary: possible circular locking dependency detected
Status: CLOSED INSUFFICIENT_DATA
Alias: None
Product: Power Management
Classification: Unclassified
Component: Hibernation/Suspend (show other bugs)
Hardware: All Linux
: P1 normal
Assignee: power-management_other
URL:
Keywords:
Depends on:
Blocks: 7216 12398
  Show dependency tree
 
Reported: 2009-01-29 11:35 UTC by Michael S. Tsirkin
Modified: 2009-04-26 11:01 UTC (History)
1 user (show)

See Also:
Kernel Version: 2.6.29-rc3
Subsystem:
Regression: Yes
Bisected commit-id:


Attachments
full dmesg output including the warning (121.46 KB, application/octet-stream)
2009-01-29 11:36 UTC, Michael S. Tsirkin
Details
kernel 2.6.29-rc3 .config with lockdep enabled (75.08 KB, application/octet-stream)
2009-01-29 11:37 UTC, Michael S. Tsirkin
Details

Description Michael S. Tsirkin 2009-01-29 11:35:58 UTC
Latest working kernel version: 2.6.28
Earliest failing kernel version: 2.6.29-rc3
Distribution: ubuntu gutsy
Hardware Environment: thinkpad T60p
Software Environment: lockdep enabled
Problem Description: after suspend/resume,
circular lock dependency warning
appears in dmesg output:

[ 2108.325226] ACPI: Preparing to enter system sleep state S3
[ 2108.345297] Disabling non-boot CPUs ...
[ 2108.345692]
[ 2108.345694] =======================================================
[ 2108.345697] [ INFO: possible circular locking dependency detected ]
[ 2108.345702] 2.6.29-rc3-mst-debug #94
[ 2108.345705] -------------------------------------------------------
[ 2108.345708] bash/6787 is trying to acquire lock:
[ 2108.345712]  (&per_cpu(cpu_policy_rwsem, cpu)){----}, at: [<c02c02b1>] lock_policy_rwsem_write+0x31/0x70
[ 2108.345727]
[ 2108.345729] but task is already holding lock:
[ 2108.345732]  (&cpu_hotplug.lock){--..}, at: [<c012d7ea>] cpu_hotplug_begin+0x1a/0x50
[ 2108.345741]
[ 2108.345742] which lock already depends on the new lock.
[ 2108.345744]


Full message and .config will attached.

Steps to reproduce: suspend, resume.
Comment 1 Michael S. Tsirkin 2009-01-29 11:36:50 UTC
Created attachment 20040 [details]
full dmesg output including the warning
Comment 2 Michael S. Tsirkin 2009-01-29 11:37:55 UTC
Created attachment 20041 [details]
kernel 2.6.29-rc3 .config with lockdep enabled
Comment 3 Rafael J. Wysocki 2009-02-14 14:23:23 UTC
References : http://lkml.org/lkml/2009/2/9/205
Comment 4 Rafael J. Wysocki 2009-02-15 12:44:37 UTC
On Sunday 15 February 2009, Michael S. Tsirkin wrote:
> It's still there in -rc5.
> 
> On Sat, Feb 14, 2009 at 10:38 PM, Rafael J. Wysocki <rjw@sisk.pl> wrote:
> > This message has been generated automatically as a part of a report
> > of recent regressions.
> >
> > The following bug entry is on the current list of known regressions
> > from 2.6.28.  Please verify if it still should be listed and let me know
> > (either way).
> >
> >
> > Bug-Entry       : http://bugzilla.kernel.org/show_bug.cgi?id=12574
> > Subject         : possible circular locking dependency detected
> > Submitter       : Michael S. Tsirkin <m.s.tsirkin@gmail.com>
> > Date            : 2009-01-29 11:35 (17 days old)
> > References      : http://lkml.org/lkml/2009/2/9/205
Comment 5 Rafael J. Wysocki 2009-03-03 14:42:01 UTC
On Tuesday 03 March 2009, Michael S. Tsirkin wrote:
> On Tue, Mar 3, 2009 at 9:25 PM, Rafael J. Wysocki <rjw@sisk.pl> wrote:
> > This message has been generated automatically as a part of a report
> > of recent regressions.
> >
> > The following bug entry is on the current list of known regressions
> > from 2.6.28.  Please verify if it still should be listed and let me know
> > (either way).
> >
> >
> > Bug-Entry       : http://bugzilla.kernel.org/show_bug.cgi?id=12574
> > Subject         : possible circular locking dependency detected
> > Submitter       : Michael S. Tsirkin <m.s.tsirkin@gmail.com>
> > Date            : 2009-01-29 11:35 (34 days old)
> > References      : http://lkml.org/lkml/2009/2/9/205
> >
> >
> >
> 
> I don't seem to observe this warning in v2.6.29-rc6 anymore.
Comment 6 Michael S. Tsirkin 2009-03-04 15:34:26 UTC
I see this again in 2.6.29-rc7.
Reopening.
Comment 7 Zhang Rui 2009-03-18 19:55:48 UTC
as the problem exists in 2.6.29-rc3 while it doesn't in 2.6.28
could you please use git-bisect to see which commit introduces this bug?
Comment 8 Michael S. Tsirkin 2009-03-19 01:12:21 UTC
>
> as the problem exists in 2.6.29-rc3 while it doesn't in 2.6.28
> could you please use git-bisect to see which commit introduces this bug?
>

Is this really necessary? I am not sure I can get my hands on this machine.
It seems pretty clear where the circular dependency is -
the real question is IMO what the right fix is.
<div dir="ltr"><div class="gmail_quote"><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">as the problem exists in 2.6.29-rc3 while it doesn&#39;t in 2.6.28<br>

could you please use git-bisect to see which commit introduces this bug?<br>
</blockquote></div><br>Is this really necessary? I am not sure I can get my hands on this machine.<br>It seems pretty clear where the circular dependency is -<br>the real question is IMO what the right fix is.<br><br></div>
Comment 9 Rafael J. Wysocki 2009-03-19 06:03:55 UTC
On Monday 16 March 2009, Michael S. Tsirkin wrote:
> On Sat, Mar 14, 2009 at 08:05:29PM +0100, Rafael J. Wysocki wrote:
> > This message has been generated automatically as a part of a report
> > of recent regressions.
> > 
> > The following bug entry is on the current list of known regressions
> > from 2.6.28.  Please verify if it still should be listed and let me know
> > (either way).
> > 
> > 
> > Bug-Entry   : http://bugzilla.kernel.org/show_bug.cgi?id=12574
> > Subject             : possible circular locking dependency detected
> > Submitter   : Michael S. Tsirkin <m.s.tsirkin@gmail.com>
> > Date                : 2009-01-29 11:35 (45 days old)
> > References  : http://lkml.org/lkml/2009/2/9/205
> > 
> 
> It's still there in rc8.
Comment 10 Rafael J. Wysocki 2009-04-26 11:00:53 UTC
It seems we can't proceed any further here.  Closing.

Note You need to log in before you can comment on or make changes to this bug.