Bug 533521

Summary:	RFE: changes to component selection when ABRT detects a segfault in /usr/bin/python
Product:	[Fedora] Fedora	Reporter:	Dave Malcolm <dmalcolm>
Component:	abrt	Assignee:	Jiri Moskovcak <jmoskovc>
Status:	CLOSED ERRATA	QA Contact:	Fedora Extras Quality Assurance <extras-qa>
Severity:	medium	Docs Contact:
Priority:	medium
Version:	rawhide	CC:	anton, berrange, dfediuck, dvlasenk, dwmw2, iprikryl, jmoskovc, kklic, mnowak, npajkovs
Target Milestone:	---
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:	1.0.6-1.fc12	Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2010-02-09 21:02:57 UTC	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Dave Malcolm 2009-11-07 00:35:40 UTC

Description of problem:
I'm getting a steady trickle of bugs assigned to me when /usr/bin/python segfaults.

The great majority of these appear to be due to assertion failures in C extension modules as called by the scripts being run.

For example, see bug 533460.  This appears to be an assertion failure within gobject python wrappings defined specifically for the elisa package.

For a segfault involving a python process, where e.g.
  cmdline: /usr/bin/python /usr/bin/elisa
would it be possible to determine the package owning the script, and if it is owned, assign the bug to that component, rather than to "python"?  (rpm -qf /usr/bin/elisa in this example)?

Thanks

Comment 1 Daniel Berrangé 2009-12-01 20:11:52 UTC

Regardless of whether the crash occurrs in a code that's part of the python runtime, or application code, this should always be reported against the RPM corresponding to the application if there is one. It is not at all helpful for all virt-manager bugs to be reported against 'python', instead of 'virt-manager'.

Agree with Dave that doing a rpm -qf on the script's filename seems like a pretty good idea to get the vast majority of reports going to the right place in BZ.

Comment 2 Jiri Moskovcak 2009-12-02 12:09:08 UTC

Yes, should be possible to parse the real app name. The current problem is that if some app creates a coredump, ABRT detects it and just read /proc/<pid>/exe to derminde the executable and in the case of bug in python extension it says /usr/bin/python, but we already have a code that reads the cmdline from coredump, which should help with this bug.

Jirka

Comment 3 Dave Malcolm 2009-12-02 14:18:37 UTC

(flipping component back from "0xFFFF" to "abrt"; issue with bugzilla's web UI?)

Comment 4 Dave Malcolm 2009-12-03 19:45:33 UTC

In https://bugzilla.redhat.com/show_bug.cgi?id=543517#c1 jmoskovc said:
> I have written a patch for #533521 and now it's in the testing phase, so you
> should be soon freed of bugs caused by extensions.

Comment 5 Denys Vlasenko 2010-01-25 14:25:35 UTC

This is already done in git and it should be in 1.0.3 too IIRC. Changing to "MODIFIED"

Comment 6 Fedora Update System 2010-02-03 15:17:20 UTC

abrt-1.0.6-1.fc12 has been submitted as an update for Fedora 12.
http://admin.fedoraproject.org/updates/abrt-1.0.6-1.fc12

Comment 7 Fedora Update System 2010-02-05 01:31:45 UTC

abrt-1.0.6-1.fc12 has been pushed to the Fedora 12 testing repository.  If problems still persist, please make note of it in this bug report.
 If you want to test the update, you can install it with 
 su -c 'yum --enablerepo=updates-testing update abrt'.  You can provide feedback for this update here: http://admin.fedoraproject.org/updates/F12/FEDORA-2010-1470

Comment 8 Fedora Update System 2010-02-09 21:01:49 UTC

abrt-1.0.6-1.fc12 has been pushed to the Fedora 12 stable repository.  If problems still persist, please make note of it in this bug report.