Bug 411652 – crash of "Software sources" in IA__gdk_keymap_get_direction()

After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.

Bug 411652 - crash of "Software sources" in IA__gdk_keymap_get_direction()


Summary:	crash of "Software sources" in IA__gdk_keymap_get_direction()


Status:	RESOLVED NOTABUG

Product:	pygtk
Classification:	Bindings
Component:	gtk
Version:	2.12.x
Hardware:	Other Linux

Importance:	Normal critical
Target Milestone:	---
Assigned To:	Nobody's working on this now (help wanted and appreciated)
QA Contact:	Python bindings maintainers

URL:
Whiteboard:

Depends on:
Blocks:

Reported:	2007-02-24 18:55 UTC by Sebastien Bacher
Modified:	2008-02-01 19:09 UTC

See Also:
GNOME target:	---
GNOME version:	2.19/2.20

Description Sebastien Bacher 2007-02-24 18:55:59 UTC

That bug has been described on https://launchpad.net/bugs/86698

"...
crash of "Software surces" while switching tabs
...
"

Debug backtrace for the crash with GTK 2.10.9:

+ Trace 113574

Thread 1 (process 5911)

#0 IA__gdk_keymap_get_direction
at gdkkeys-x11.c line 702
#1 gtk_text_view_check_keymap_direction
at gtktextview.c line 5900
#2 gtk_text_view_ensure_layout
at gtktextview.c line 5965
#3 gtk_text_view_update_layout_width
at gtktextview.c line 1814
#4 gtk_text_view_value_changed
at gtktextview.c line 6785
#5 _gtk_marshal_VOID__OBJECT_OBJECT
at gtkmarshalers.c line 2202
#6 g_type_class_meta_marshal
at gclosure.c line 567
#7 IA__g_closure_invoke
at gclosure.c line 490
#8 signal_emit_unlocked_R
at gsignal.c line 2478
#9 IA__g_signal_emit_valist
#10 IA__g_signal_emit
at gsignal.c line 2243
#11 IA__gtk_widget_set_scroll_adjustments
at gtkwidget.c line 3982
#12 gtk_scrolled_window_add
at gtkscrolledwindow.c line 1560
#13 IA__g_cclosure_marshal_VOID__OBJECT
at gmarshal.c line 636
#14 g_type_class_meta_marshal
at gclosure.c line 567
#15 IA__g_closure_invoke
at gclosure.c line 490
#16 signal_emit_unlocked_R
at gsignal.c line 2370
#17 IA__g_signal_emit_valist
at gsignal.c line 2199
#18 IA__g_signal_emit
at gsignal.c line 2243
#19 IA__gtk_container_add
at gtkcontainer.c line 959
#20 glade_standard_build_children
from /usr/lib/libglade-2.0.so.0
#21 glade_xml_set_common_params
from /usr/lib/libglade-2.0.so.0
#22 glade_xml_build_widget
from /usr/lib/libglade-2.0.so.0

Comment 1 Sebastien Bacher 2008-02-01 17:52:03 UTC

that looks like a bug in gtk_text_view_check_keymap_direction (), gdk_keymap_get_for_display () is called to get the keymap and can return null and then gdk_keymap_get_direction() is call and crash on null argument

Comment 2 Sebastien Bacher 2008-02-01 18:02:08 UTC

keymap = GET_EFFECTIVE_KEYMAP (keymap) is used so it should handle keymap correctly, not sure what is going wrong there

Comment 3 Christian Kellner 2008-02-01 19:02:08 UTC

Actually that bug is caused by the pygtk changes of bug #316877. import pygtk will not fail when the display could not be opened but only spit out a warning message causing all pygtk applications that are not prepared for that to proceed with their logic most likely hitting a gtk function that needs a valid display (like gtk_text_view_check_keymap_direction) resulting in a crash inside of gtk. Of course you shouldnt use gtk when gtk_init_check () returns FALSE. So this is not a gtk bug, but a pygtk application bug. 

I'll reassign to pygtk therefore since its likely that a few applications will hit this (update-manager and Gdebi in ubuntu already did) leaving it to the pygtk folks to decided what do to with it (probably NOTABUG or something). ;-)

Comment 4 Johan (not receiving bugmail) Dahlin 2008-02-01 19:09:48 UTC

The change of not raising an error is documented:

http://live.gnome.org/PyGTK/WhatsNew210

Closing as NOTABUG.