apport-gtk crashes when the crash report is not writable

[Till Ulen]

Binary package hint: apport

(Was: apport-gtk crashed with IOError in delete_report().)

Bug #129146 is about handling the case when the crash report is not readable. I stumbled upon another bug in apport where it crashes while trying to write to a .crash file that is not writable.

Has this problem been reported already? I couldn't find any existing bugs on it.

How to reproduce:

1. Crash a program running with root privileges. I used bug #67995 to crash Synaptic.

2. Find the fresh report in /var/crash and change its group and permissions so that it is readable but not writable by your current user:
    $ cd /var/crash
    $ sudo chgrp username _usr_sbin_synaptic.0.crash
    $ sudo chmod g+r _usr_sbin_synaptic.0.crash

3. Open the crash report under your current user.

Step one can actually be omitted, but this scenario is intended to show how you can run into this problem in real life. When I did, I didn't realize that apport needed to write to the crash report in order to send it.

ProblemType: Crash
Architecture: i386
CrashReports:
 640:0:1001:5152387:2008-03-05 02:23:22.000000000 +0600:2008-03-05 02:21:45.000000000 +0600:/var/crash/_usr_sbin_synaptic.0.crash
 600:1001:1001:18014:2008-03-05 02:48:14.000000000 +0600:2008-03-05 02:48:15.000000000 +0600:/var/crash/_usr_share_apport_apport-gtk.1001.crash
Date: Wed Mar 5 02:48:15 2008
DistroRelease: Ubuntu 7.10
ExecutablePath: /usr/share/apport/apport-gtk
InterpreterPath: /usr/bin/python2.5
Package: apport-gtk 0.98
PackageArchitecture: all
ProcCmdline: /usr/bin/python /usr/share/apport/apport-gtk -c /var/crash/_usr_sbin_synaptic.0.crash
ProcCwd: /home/users/alexkon
ProcEnviron:
 PATH=/home/users/alexkon/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games
 LANG=en_US.UTF-8
 SHELL=/bin/bash
PythonArgs: ['/usr/share/apport/apport-gtk', '-c', '/var/crash/_usr_sbin_synaptic.0.crash']
SourcePackage: apport
Title: apport-gtk crashed with IOError in delete_report()
Uname: Linux chronos 2.6.22-14-generic #1 SMP Tue Feb 12 07:42:25 UTC 2008 i686 GNU/Linux

[Daniel Hahler (blueyed)]

Thank you for your bug report.
This appears to be a duplicate of bug 195750, which was marked private until now.
I'm marking it as such.

[Till Ulen (tillulen)]

Is it necessary to mark duplicate bugs as Invalid? I thought that
status was reserved to irrelevant bug reports, and that no report of a
genuine bug should be marked as Invalid. See
<https://help.launchpad.net/BugStatuses>.

On Sat, Mar 8, 2008 at 4:31 AM, Daniel Hahler wrote:
[ ... ]
>
> ** Changed in: apport (Ubuntu)
> Importance: Undecided => Medium
> Status: New => Invalid
>
> ** This bug has been marked a duplicate of bug 195750
> apport-cli crashed with IOError in delete_report()

[Daniel Hahler (blueyed)]

Alexander, this page says:
  Invalid: the report describes the software's normal behaviour, or is unsuitable for any other reason.

I think "unsuitable for any other reason" fits here.

I've just asked about this in #ubuntu-bugs (IRC) and the desktop team handles this the same way (setting it to "Invalid"), but others don't.
I couldn't find any policy for this, e.g. on https://wiki.ubuntu.com/Bugs/MarkingDuplicate

This should probably get discussed on the ubuntu-bugsquad mailing list (https://lists.ubuntu.com/mailman/listinfo/Ubuntu-bugsquad).