gnome-power-manager crashed with SIGSEGV in free()

StacktraceTop:?? ()

Thanks for the stack trace (the complete one is here: http://launchpadlibrarian.net/28836332/Stacktrace.txt).

How repeatable is the crash? Are there steps you can do to recreate it? When you say "the screen turns black" - is it because of GPM screen blanking, screen saver, or this crash?
Thanks!

Thanks for your care.

Immediately before the crash, there is a very brief screen blanking. It is definitely not due to the screen saver because it happens when I work.

Don't know if it's of interest, but in order to keep my battery in good condition, I always use it whenever possible, down to ~ 10% then I recharge it fully. I'm going to be careful in order to gather more precise conditions to reproduce this bug.

I have just seen this issue. No steps to reproduce as far as I could tell. System was idle and then notified me that power-manager had crashed triggering apport.

Thanks for the confirmation.
Could one of you please attach the resulting log file of: /usr/share/gnome-power-manager/gnome-power-bugreport &> gpm.log to the report?

Also, the backtrace isn't complete yet. Could you follow the instructions at:
https://wiki.ubuntu.com/Backtrace
to get a backtrace. The apport trace is failing, so we have to use the manual gdb trace.

Please find the gpm.log file as per your request. I'm now backtracing using gdb. Chris, can you also make this procedure, we'll have more chance triggering the crash.

Here is the gdb backtrace. I have also added dbgsym in libglib and libdbus. There are still "??" from libdbus-glib. I'll add dbgsym in this package too in my next attempt to reproduce the bug.

I think the bug triggers when reaching full charge, with a screensaver enabled (I'm using GLCells). I was not able to reproduce the bug yesterday with screensaver disabled (though it is not a certitude).

Confirming, this bug triggers when reaching full charge (was able to reproduce it 3 times).

Here is the *very* full gdb trace with dbgsym in all libraries.

Phenomenal work! Thanks for putting this together and going through the extra steps for the manual backtrace.

This bug has been reported to the developers of the software. You can track it and make comments at: http://bugzilla.gnome.org/show_bug.cgi?id=588259

The upstream author, Richard Hughes has committed the fix upstream:

------- Comment #3 from Richard Hughes 2009-07-11 08:00 UTC -------
Excellent bug, thanks for reporting this. I've fixes this in git master, it was
a typo in the signal setup.

We'll either have to wait to get this merged into ubuntu, or someone here can make the patch.

Here is a patch.

Here is a debdiff patch fixing this bug.

Thanks a lot Arnaud.

This package uses a patching system [1,2] so you'll have to change your debdiff a bit:

1) rename LP397248.patch to something more descriptive like "crash-when-battery-charge-full.patch"
2) move it to debian/patches
3) build and test
4) get the debdiff

I can't tell when the next GPM revision will be released upstream, but this is something worth doing.

1: https://wiki.ubuntu.com/PackagingGuide/Complete
2: https://wiki.ubuntu.com/PackagingGuide/Complete#Patch%20Systems

Also, you can go onto IRC (freenode) and ask someone from #ubuntu-motu to sponsor your patch when your done.

Here is a proposed bzr branch to merge to main branch lp:~gnome-power-manager-team/gnome-power/trunk.

This bug made me learn a lot about the Ubuntu bug patching system !

Merged, thanks!

This bug was fixed in the package gnome-power-manager - 2.27.2-0ubuntu2

---------------
gnome-power-manager (2.27.2-0ubuntu2) karmic; urgency=low

  [ Martin Pitt ]
  * Add crash-when-battery-charge-full.patch: Fix crash when battery gets
    fully charged. Thanks to Arnaud Faucher! (LP: #397248)

  [ Scott Howard ]
  * debian/local/source_gnome-power-manager.py: Add gnome-power-bugreport.sh
    output. Upstream uses this script for triaging. (LP: #397993)

 -- Martin Pitt <email address hidden> Mon, 13 Jul 2009 10:10:56 +0200