[csw-users] gtk segmentation fault

[James Lee]

On 13/01/06, 03:48:00, Denis "I." Morozov <d_frost at mail.ru> wrote regarding 
Re: [csw-users] gtk segmentation fault:

> I executed "$ gdb /opt/csw/bin/gaim ./core":

> (gdb) where
> #0  0xfe81f82c in _lwp_kill () from /usr/lib/libc.so.1
> #1  0xfe7d0a24 in raise () from /usr/lib/libc.so.1
> #2  0xfe7b6ce0 in abort () from /usr/lib/libc.so.1
> #3  0xfdf461a8 in __sighndlr () from /usr/lib/libthread.so.1
> #4  0xfdf3fed8 in call_user_handler () from /usr/lib/libthread.so.1
> #5  <signal handler called>
> #6  0x00000000 in ?? ()
> #7  0xfe2da378 in fbRasterizeTrapezoid () from /opt/csw/lib/libcairo.so.2
> #8  0xfe2daa70 in pixman_compositeGeneral () from 
/opt/csw/lib/libcairo.so.2
> #9  0xfe2caa6c in _cairo_pixman_composite () from 
/opt/csw/lib/libcairo.so.2
> #10 0xfe2a5344 in _cairo_image_surface_assume_ownership_of_data ()
>     from /opt/csw/lib/libcairo.so.2
> #11 0xfe2a5344 in _cairo_image_surface_assume_ownership_of_data ()
>     from /opt/csw/lib/libcairo.so.2
> Previous frame identical to this frame (corrupt stack?)
> (gdb)


This is very similar to an error I have seen, see bug:
http://www.blastwave.org/mantis/view_bug_page.php?f_id=0001318

What happened was cairo set a function pointer to null, then called it.
It uses null because it doesn't understand how to use the visual it has
been given (eg wrong colour depth).  Line #6: the 0x00000000 is a null
function address and ?? is because it can't have a name.  Line #7 and
below suggests the cause is in cairo and my own debugging work proved
this to be so.

Try replacing pango and gtk2 (my theory is that the fault is in cairo
but we have to replace the packages that call it).

# pkgrm CSWpango CSWgtk2

and install those from stable:

# pkg-get -s http://path/to/mirror/stable -U -i libpango gtk2





James.