Bug 709544

Summary: INFO: possible circular locking dependency detected (ext4_evict_inode &sb->s_type->i_mutex_key vs &mm->mmap_sem)
Product: [Fedora] Fedora Reporter: koukou73gr
Component: kernelAssignee: Eric Sandeen <esandeen>
Status: CLOSED CURRENTRELEASE QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: medium Docs Contact:
Priority: unspecified    
Version: 15CC: gansalmon, itamar, jonathan, kernel-maint, madhu.chinakonda
Target Milestone: ---   
Target Release: ---   
Hardware: x86_64   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2012-03-17 16:31:36 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Attachments:
Description Flags
INFO trace
none
dmesg messages after thawing.
none
dmesg, 2.6.40.4-5.fc15
none
dmesg, 2.6.40.4-5.fc15, continued none

Description koukou73gr 2011-06-01 00:18:16 UTC
Created attachment 502136 [details]
INFO trace

Description of problem:
Sometimes while hibernating (mostly the first time after a fresh restart) I get and INFO trace, presumably while the system tries to shutdown the non-boot core of the CPU.

Hibernation process completes however and the system is able to resume successfully as well afterwards.

Version-Release number of selected component (if applicable):
2.6.38.6-27.fc15.x86_64.debug #1

How reproducible:
Often but not every time. However, I haven't seen it happening again in subsequent hibernation/thaw cycles. If I cold-boot the system, it is bound to happen again.

Steps to Reproduce:
1. Boot
2. Login
3. Hibernate gnome-shell. I have set my power button to hibernation and use this.
  
Actual results:
Trace generated.

Expected results:
No trace.

Additional info:
This is on a Dell Latitude E4300, 6GB RAM, Intel GM45 with an SP9600 CPU. 

This laptop seems to hang into ACPI ( "ACPI: While loop taking a really long time. loop_count=..." ) after some number of hibernation/thaw cycles and/or throw random oops traces during hibernation/thaw that I was unable to save so far. Hence I run a debug kernel and hope to catch something more in my logs just before that happens. This INFO trace is the first catch.

Comment 1 koukou73gr 2011-06-01 00:20:01 UTC
Created attachment 502137 [details]
dmesg messages after thawing.

Comment 2 Josh Boyer 2011-09-27 15:22:44 UTC
Does this still happen with the latest 2.6.40.x F15 kernel?

Comment 3 koukou73gr 2011-10-02 16:31:58 UTC
Sort of...

Just tested with 2.6.40.4-5.fc15.x86_64.debug today, booted, hibernated, resumed, hibernated and resumed again. I got an INFO trace for a possible deadlock after the first resume, see attached dmesg.

I've been testing even with the 3.x series, no matter what kernel version so far, after 3-4 hibernate/resume cycles my system will echo INFO/BUG traces and either fail to hibernate again, crash upon next resume, fail to bring up wired or wireless network cards and generally come to a state where I am forced to make a clean boot.

Pretty bad for a laptop with no exotic hardware whatsoever :(

Thanks for looking into this.

Comment 4 koukou73gr 2011-10-02 16:34:12 UTC
Created attachment 525946 [details]
dmesg, 2.6.40.4-5.fc15

Comment 5 koukou73gr 2011-10-03 11:29:24 UTC
And then again, after third hibernate/resume cycle, there was an INFO trace logged during hibernation. Attached dmesg picks up exaclty where the previous one left off, this is the same clean boot cycle.

Comment 6 koukou73gr 2011-10-03 11:30:35 UTC
Created attachment 526029 [details]
dmesg, 2.6.40.4-5.fc15, continued

Comment 7 Dave Jones 2012-02-06 17:42:15 UTC
this looks familiar.  I think this may have been fixed in latest kernels ?

Comment 8 koukou73gr 2012-02-06 21:15:03 UTC
(In reply to comment #7)
> this looks familiar.  I think this may have been fixed in latest kernels ?

I have stopped using hibernation on this laptop since my last posting here, as it was simply unreliable. I'll give it a go again and report back.

Thanks for looking this way Dave.

Comment 9 koukou73gr 2012-03-17 16:31:36 UTC
Alright, I tested hibernation again for sometime. I can admit tha what I outlined above does not happen now. I still get freezes and/or OOPSes after 3-4 cycles, just not of the same kind and reproducibility.

So I guess this should be closed.

Thanks to everybody who took my ticket into account and I hope it helped in improving the Linux kernel.