[Bug 196207] GTK/Glib applications crash when running under KDE4.3x (malloc_printerr, free(): invalid pointer)
Michael Pyne
mpyne at kde.org
Mon Jul 13 23:54:14 BST 2009
https://bugs.kde.org/show_bug.cgi?id=196207
Michael Pyne <mpyne at kde.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|RESOLVED |UNCONFIRMED
CC| |mpyne at kde.org
Resolution|DUPLICATE |
--- Comment #31 from Michael Pyne <mpyne kde org> 2009-07-14 00:54:08 ---
I've been experiencing what I believe is this bug, and here's what I (think) I
know:
- It's in relation to glib. QT_NO_GLIB=1 before running affected applications
seems to stop the crashes.
- It's in relation to glibc 2.10.1. I didn't experience this issue until I
upgraded to this version of glibc.
- It's related to the malloc check code:
* Running MALLOC_CHECK_=2 or 3 causes a SIGABRT after some songs with JuK,
using phonon-gst (and the glib Qt event loop, obviously).
* Tone MALLOC_CHECK_ to a 1 (check, give message, but don't abort) and JuK
will eventually crash with SIGSEGV instead of SIGABRT, which is an indication
of eventual heap corruption.
* unset MALLOC_CHECK_ and I can't get JuK to crash using my normal reproducing
technique (clicking "Next" as fast as I can during playback). I haven't run it
for long enough to where I think there won't be a SIGSEGV but it lasts much
longer than with checking enabled but disregarded.
- It's related to KDE 4. I've been able to reproduce with a gstreamer-only
test application I have (though much less frequently), except when running in
KDE 4. Running KDE 4 applications in fluxbox still leads to crashes as well.
The other application besides JuK I get issues with is KTorrent.
Reopening because I'm really not quite sure of what to make of it, or where the
true fault lies (although at this point I suspect something in glibc's memory
checking code).
--
Configure bugmail: https://bugs.kde.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.
More information about the Unassigned-bugs
mailing list