Bug#535626: Crash with "BadImplementation (server does not implement operation)"

[Ulrich Eckhardt]

Package: libwebkit-1.0-2
Version: 1.1.10-2

I initially observed this bug in midori but was suggested to test webkit 
directly, which promptly showed the same error. What I do is this:

$ /usr/lib/webkit-1.0-2/libexec/GtkLauncher http://space-invasion.de
open dsp: Device or resource busy
The program 'GtkLauncher' received an X Window System error.
This probably reflects a bug in the program.
The error was 'BadImplementation (server does not implement operation)'.
  (Details: serial 33 error_code 17 request_code 146 minor_code 5)
  (Note to programmers: normally, X errors are reported asynchronously;
   that is, you will receive the error a while after causing it.
   To debug your program, run it with the --sync command line
   option to change this behavior. You can then get a meaningful
   backtrace from your debugger if you break on the gdk_x_error() function.)

When I do this (start it in gdb and use the --sync commandline option) it 
doesn't change anything, the error stays the same and the breakpoint is never 
reached (like in [1]).


Okay, what else....
- The page initially loaded a bit, enough to reset the window title and 
display some graphics.
- I actually downloaded and installed the debug webkit library, but I couldn't 
figure out how to use this. No readme. No documentation. The included 
GtkLauncher is not executable. Bummer.
- I actually tried to run midori through xtrace[2] but all I got was a messed-
up X session which I could only recover from through a remote connection from 
a different machine. Heaven knows what xtrace does with things it doesn't 
understand itself...
- Looking at /usr/include/xcb/glx.h, 146 is defined as XCB_GLX_IS_TEXTURE, 
maybe the toolkit in question tries to use GLX (which I otherwise have 
activated, as glxinfo tells me) without querying its availability first?
- I'm running on a PowerPC machine, which is big-endian as opposed to the more 
common x86 machines, so you might not be able to see the effect.


Good luck and drop me a note if I can help test further things!

Uli

[1] http://bugs.debian.org:80/cgi-bin/bugreport.cgi?bug=510697, same useless 
hint how to debug this.
[2] Run xtrace in one terminal, it will tell you where it starts an X server, 
e.g. ":9". You then invoke midori with "DISPLAY=:9 midori".