trinity-users@lists.pearsoncomputing.net

Message: previous - next
Month: January 2015

Re: [trinity-users] desktop crashing

From: Michele Calgaro <michele.calgaro@...>
Date: Tue, 13 Jan 2015 14:51:37 +0900
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

On 2015/01/13 12:00 PM, dep wrote:
> (gdb) print *((int*)(0xda51f8)+2) $1 = 30334
That's good. It means the mutex is locked by thread 1 (LWP 30334).

I actually noticed just now (my fault) that you do not have the debug
symbols installed for tdelibs, tdebase and tqt3. Can you install them,
exit gdb, reattach to the kdesktop process and run again the
thread apply all bt
command? That should provide more detailed information (i.e. line
numbers) about where each thread and frame is.
- From a quick look at the current incomplete stack frame, thread 1
(which is blocking thread 2) is blocked inside a malloc() call, which
as far as I can tell is usually a sign that the heap is corrupted
(perhaps a pointer deallocated twice). Now we are kind of stretching
to the limit of my knowledge as well, so I will need to guess/look up
what's the best way to go.

By the way, when kdesktop_lock crashes do you have any crash report
window popping up in your system?

In the mean time, can you try the following things:

1) install the debug symbols as said above and provide the updated
stack frames

2) kill kdesktop and run it again (or reboot your system, even
better). From CLI (outside TDE), run:
ps aux |grep kdesktop
Take note of the kdesktop_lock pid. Run:
gdb --pid=<kdesktop_lock pid>
After gdb attaches to the process, continue execution with:
c
At this stage kdesktop_lock should be running normally. Leave
everything running until kdesktop_lock crashes again. At this point
gdb should provide some info about the crash (perhaps a segmentation
fault - SIGSEGV). When it happens you should be able to get the stack
frame for kdesktop_lock (again thread apply all bt) since gdb is still
attached to the process. Do not exit gdb after that, not sure what
else we may need. All this is somehow experimental, if something
doesn't go to plan, please let me know.

Cheers
  Michele

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2
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=Gykb
-----END PGP SIGNATURE-----