gtk+-2

[Dennis Veatch]

On Friday, November 11, 2011 2:26:13 AM Zachary McGrew wrote:
> Ever since the latest glibc update my system has been a bit wonky. I
> found a patch I need to commit to fix a bunch of segfaults in SDL &
> OpenGL apps, but I still can't track down why my gtk+ update is
> segfaulting.
> 
> I've turned core dumps on, but it's not terribly useful. The following
> is the relevant part of the compile that fails (it's in the make
> install part):
> 
> /usr/bin/make  install-data-hook
> make[4]: Entering directory `/usr/src/gtk+-2.24.7/modules/input'
> /bin/sh /usr/src/gtk+-2.24.7/install-sh -d /etc/gtk-2.0
> ../../gtk/gtk-query-immodules-2.0 > /etc/gtk-2.0/gtk.immodules
> /bin/sh: line 12: 17352 Segmentation fault      (core dumped)
> ../../gtk/gtk-query-immodules-2.0 > /etc/gtk-2.0/gtk.immodules
> make[4]: *** [install-data-hook] Error 139
> 
> ----------------------------------------
> 
> And here is what gdb says of the core dump:
> 
> Reading symbols from
> /usr/src/gtk+-2.24.7/gtk/.libs/lt-gtk-query-immodules-2.0...done.
> 
> warning: core file may not match specified executable file.
> [New LWP 17352]
> 
> warning: Can't read pathname for load map: Input/output error.
> [Thread debugging using libthread_db enabled]
> Core was generated by
> `/usr/src/gtk+-2.24.7/gtk/.libs/lt-gtk-query-immodules-2.0'.
> Program terminated with signal 11, Segmentation fault.
> #0  0x00000000 in ?? ()
> 
> ----------------------------------------
> 
> If I run the command that's segfaulting outside of lin it works
> perfectly (no segfault).

What you describe about gtk+-2 has bitten me on several occasion's and afaict 
it is not glibc related since it has happened over several glibc versions. 

For me iirc it has always happened during a lunar rebuild after I have X, KDE 
and a slew of other things installed. So I thought it might have something to 
do with lunar rebuild + 1000 or more things installed and running out of 
buffers, memory, hitting some upper level ability of bash or something; I 
don't think that is the case.

Google reveals many hits but the majority were not useful, except for one; 
remove the nvidia blobs, then lin gkt+-2 and reinstall the blob. 

That has always worked for me. The strange thing is; the seg fault does not 
always happen with the blobs installed and does not seem to be dependent on a 
particular blob version.


-- 
Dennis Veatch
Lunar-Linux Developer
http://lunar-linux.org