Bug 100697

Summary: cairo segfaults with gtk-qt theme
Product: [openSUSE] SUSE LINUX 10.0 Reporter: Christoph Thiel <cthiel>
Component: GNOMEAssignee: E-mail List <gnome-bugs>
Status: RESOLVED FIXED QA Contact: E-mail List <qa-bugs>
Severity: Critical    
Priority: P5 - None CC: federico, gnome-bugs, meissner
Version: Preview 4   
Target Milestone: ---   
Hardware: All   
OS: All   
Whiteboard:
Found By: Other Services Priority:
Business Priority: Blocker: ---
Marketing QA Status: --- IT Deployment: ---
Bug Depends on:    
Bug Blocks: 102955    
Attachments: rpm -qa

Description Christoph Thiel 2005-08-04 10:14:39 UTC 
cthiel@f33:~> realplay

(realplay.bin:23762): Gdk-WARNING **: Using Cairo rendering requires the
drawable argument to
have a specified colormap. All windows have a colormap,
however, pixmaps only have colormap by default if they
were created with a non-NULL window argument. Otherwise
a colormap must be set on them with gdk_drawable_set_colormap
/usr/bin/realplay: line 75: 23762 Segmentation fault      $REALPLAYBIN "$@"
cthiel@f33:~>
Comment 1 Stanislav Brabec 2005-08-04 10:58:54 UTC 
Can you be more specific, when it occurs. I cannot repeat it.

Which gtk theme are you using? Could you try different GTK+2 theme or run it in
GNOME?
Comment 2 Christoph Thiel 2005-08-04 11:01:08 UTC 
I'm running it on KDE. I'll attach "rpm -qa" output.

To reproduce it, just run realplayer and then try to click on the menu.
Comment 4 Stanislav Brabec 2005-08-04 12:32:30 UTC 
Please try again with different gtk+ theme (or GNOME session).
Comment 5 Stanislav Brabec 2005-08-04 12:38:39 UTC 
Confirming RealPlayer crash with gtk-qt-engine and pressing File menu.

Confirming total crash of GNOME session while switching theme.

(AMD64)
Comment 6 Christoph Thiel 2005-08-04 15:19:11 UTC 
This problem is _not_ AMD64-only. My workstation is i386 and it fails there, too.
Comment 7 Stanislav Brabec 2005-08-05 10:20:25 UTC 
*** Bug 102325 has been marked as a duplicate of this bug. ***
Comment 8 Christoph Thiel 2005-08-05 16:00:43 UTC 
Well, the same (?) problem can be triggered with gimp. Just start it and hit on
the menu.

Here's the backtrace:

#0  0x4055b965 in cairo_test_xlib_disable_render () from /usr/lib/libcairo.so.1
#1  0x4055b9e2 in cairo_test_xlib_disable_render () from /usr/lib/libcairo.so.1
#2  0xbfd7948c in ?? ()
#3  0x00000000 in ?? ()
#4  0x00000000 in ?? ()
#5  0x00000002 in ?? ()
#6  0x00000000 in ?? ()
#7  0x00000000 in ?? ()
#8  0xbfd79338 in ?? ()
#9  0x4055bc40 in cairo_test_xlib_disable_render () from /usr/lib/libcairo.so.1
#10 0x00000000 in ?? ()
#11 0x00000002 in ?? ()
#12 0x00000000 in ?? ()
#13 0x4043f658 in ?? () from /opt/gnome/lib/libgtk-x11-2.0.so.0
#14 0x4043f658 in ?? () from /opt/gnome/lib/libgtk-x11-2.0.so.0
#15 0x40290a80 in gtk_menu_bar_get_pack_direction () from
/opt/gnome/lib/libgtk-x11-2.0.so.0
Previous frame inner to this frame (corrupt stack?)
Comment 9 Stephan Kulow 2005-08-05 16:04:06 UTC 
this looks pretty much like a gtk2 regression
Comment 10 Robert O'Callahan 2005-08-08 19:06:11 UTC 
The warning message seems clear enough. New versions of GTK have cairo
integrated so that's where the regression is coming from. Maybe gtk-qt needs to
be updated.

The stacktrace is mostly garbage. Can you get a good stacktrace with gdb?
Comment 11 Chris Lahey 2005-08-10 16:56:40 UTC 
I've talked to the gtk people.  They say that drawing to pixmaps without a
colormap was always against the rules, but wasn't as enforced as it is now.
Comment 12 JP Rosevear 2005-08-11 21:55:23 UTC 
Chris has submitted a fix for this to beta 2.
Comment 13 Stephan Kulow 2005-08-15 08:54:50 UTC 
*** Bug 104634 has been marked as a duplicate of this bug. ***