Bug 791121

Summary: [kde-runtime-4.8.0-4] Process /usr/bin/nepomukservicestub was killed by signal 11 (SIGSEGV)
Product: [Fedora] Fedora Reporter: Martin Kho <rh-bugzilla>
Component: kde-runtimeAssignee: Than Ngo <than>
Status: CLOSED RAWHIDE QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: rawhideCC: jreznik, kevinc, kevin, ltinkl, rdieter, rnovacek, ry, smparrish, than
Target Milestone: ---Keywords: Patch, Reopened, Upstream
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: kde-runtime-4.8.1-1 Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2012-03-10 13:42:26 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
Backtrace created by abrt
none
xsession-errors none

Description Martin Kho 2012-02-16 09:08:39 UTC
Created attachment 562441 [details]
Backtrace created by abrt

Description of problem:
This is fedora 17 branched! The crash happened after a new mail was arrived ;-(. In .xsession-errors the following message appears:

:[/usr/bin/nepomukservicestub] QObject::connect: Cannot connect XSyncBasedPoller::destroyed() to (null)::_k_x11FilterDestroyed()
:ProcessControl: Application '/usr/bin/nepomukservicestub' stopped unexpected (Process crashed)
:Application '/usr/bin/nepomukservicestub nepomukstorage' crashed! 4 restarts left.
:Application '/usr/bin/nepomukservicestub nepomukbackupsync' exited normally...
:Application '/usr/bin/nepomukservicestub nepomukqueryservice' exited normally...
:Application '/usr/bin/nepomukservicestub nepomukfileindexer' exited normally...
:Application '/usr/bin/nepomukservicestub nepomukfilewatch' exited normally...


Version-Release number of selected component (if applicable):
kde-runtime-4.8.0-4.fc17.x86_64

How reproducible:
Always (when new mail has arrived)

Steps to Reproduce:
1. Just wait till new mail has arrived
2.
3.
  
Actual results:
Don't know the real effect. May be mail won't be indexed?

Expected results:
Mail get indexed

Additional info:

Comment 1 Martin Kho 2012-02-17 18:20:33 UTC
Hi,

I'v good news and bad news. The good news is that I haven't had any crashes lately. :-) The bad news is that I don't know what solved it. I suspect it was an update to selinux or may be systemd?

So I think this report can be considered as a 'false positive'. Please close it.

Thanks and sorry for the noise.

Martin Kho

Comment 2 Martin Kho 2012-02-17 18:52:09 UTC
Hi,

Huh, ho stop! Right after I had written comment #1 nepomukservicestub crashed
again. Just before the crash the following messages were written to
.xsession-errors:

"akonadi_nepomuk_feeder(1192) ItemQueue::fetchJobResult: Not all items were
fetched:  1 2 
akonadi_nepomuk_feeder(1192) ItemQueue::batchJobResult: "Message did not
receive a reply (timeout by message bus)""


Please don't close this report (jet) ;-(

Martin Kho

PS. Just had a collision with Kevin :-)

Comment 3 Martin Kho 2012-02-17 19:01:02 UTC
Created attachment 563968 [details]
xsession-errors

Because I see more badness in .xsession-errors I uploaded it. May be it can give more information.

Comment 4 Martin Kho 2012-02-21 09:11:25 UTC
Hi,

It seems SELinux has something to do with this issue. In /var/log/audit/audit.log I see:

"type=ANOM_ABEND msg=audit(1329814757.733:87): auid=1000 uid=1000 gid=1000 ses=1 subj=unconfined_u:unconfined_r:unconfined_t:s0 pid=1677 comm="nepomukservices" reason="memory violation" sig=11"

No idea what this means?

Martin Kho

Comment 5 Kevin Kofler 2012-02-21 14:32:37 UTC
That doesn't look like a SELinux AVC to me, but only as an audit logging of the segfault. It just means that the program crashed with SIGSEGV (signal 11), which we already knew from ABRT's report.

Comment 6 Martin Kho 2012-02-21 14:54:35 UTC
Hi,

Yeah, I already was afraid that, that would be the anwser :-). The problem is that I now run with selinux=0 (selinux disabled) as kernel parameter and don't have seen crashes. Isn't this weird?

Martin Kho

Comment 7 Martin Kho 2012-02-21 15:19:41 UTC
Hurray, it's not related to selinux. nepomukservicestub still crashes :-) It just took a little longer. Before nepomukservicestub crashed kmail/akonadi was asking for it to do 'something'.

From .xsession-errors:
"akonadi_nepomuk_feeder(1389) ItemQueue::batchJobResult: "Message did not receive a reply (timeout by message bus)"
ProcessControl: Application '/usr/bin/nepomukservicestub' stopped unexpected (Process crashed)"

Martin Kho

Comment 8 Kevin Kofler 2012-02-21 15:25:59 UTC
FYI, SELinux does respond to some blocked operations by segfaulting the offending process, but that would trigger an AVC before the segfault (unless it's "dontaudit"ed, which wouldn't make much sense for a denial crashing the program).

Comment 9 Rex Dieter 2012-02-22 19:10:37 UTC
In my F17-alpha RC4 install, I had some serious issues, at least until I upgraded to selinux-policy-3.10.0-91 (currently in updates-testing).

Comment 10 Martin Kho 2012-02-22 21:34:10 UTC
Hi,

@Rex: I'm running version -91 since this morning (CET), but still get the crashes. Some other issues (e.g log in delay of ~30 seconds) are gone.

Martin Kho

Comment 11 Martin Kho 2012-02-23 17:20:28 UTC
Hi,

Based on a discussion on the nepomuk mailinglist [1] I got the idea that the debug enabled kernels could make a difference wrt the crashes I see. I installed kernel 3.3.0-0.rc4.git0.1.fc17.x86_64 (debugging disabled) and had no crashes. akonadi_nepomuk_feeder is doing its job as one would expect.


Martin Kho 


[1] http://mail.kde.org/pipermail/nepomuk/2011-November/002116.html

Comment 12 Martin Kho 2012-02-28 09:03:19 UTC
Hi,

A little update: Now running a debug enabled kernel again (3.3.0-0.rc4.git1.4.fc17.x86_64) and nepomukservicestub crashes again. Not saying that there is no bug in nepomukservicestub. The debug kernel just triggers the bug, IMHO.

Martin Kho

Comment 13 Martin Kho 2012-02-28 18:28:59 UTC
Hi,

Today is a very fruitful day. First was bug #796969 [1] found and solved so that I can reboot/shutdown again. Next I saw that Sebastian Trueg has committed a patch (rev. 8bcedfcc) for this issue (nepomukservicestub crash) [2]. I've rebuild kde-runtime and it looks like that everything is working fine.

Martin Kho


[1] https://bugzilla.redhat.com/show_bug.cgi?id=796969
[2] https://projects.kde.org/projects/kde/kde-runtime/repository/revisions/8bcedfccd412603bf5744a15be3cb97d48a4923e/diff/nepomuk/services/backupsync/lib/resourceidentifier.cpp

Comment 14 Rex Dieter 2012-02-28 18:46:25 UTC
win!

given 4.8.1 will be landing within a couple of days, let's wait for that.

Comment 15 Martin Kho 2012-03-10 13:14:48 UTC
Hi Rex,

Yes, the commit (comment #13) made it in 4.8.1 and everything is fine again. So please close this report.

Thanks,

Martin Kho

PS. 4.8.1 is even better, really great work!