Re: [gtk-list] Re: Random Segv's

["Gregory A. McLean" <gregm randomc com>]

> 
> 
> "Gregory A. McLean" <gregm@randomc.com> writes:
> 
> > -----BEGIN PGP SIGNED MESSAGE-----
> > 
> > 
> > Hi!
> > 
> > In gxsnmp I have been getting some 'random' seg faults on exit, here's a 
> > backtrace:
> >  #0  0x4011de36 in chunk_free () at malloc.c:2824
> >  #1  0x4011dd01 in __cfree () at malloc.c:2684
> >  #2  0x400d92dd in g_free () at gmem.c:591
> >  #3  0x400764f4 in gtk_object_finalize () at gtkobject.c:421
> 
> [....]
> 
> > The puzzling thing is this is happening at program exit, but not every time..
>  
> The symptoms indicate that the memory is getting doubly freed. I can't
> say for sure if this is your problem or GTK's problem, but it
> is some sort of reference counting problem.
> 
> Try running the program as:
> 
>   gxsnmp --gtk-debug objects
> 

That produced this:
** ERROR **: sigsegv caught
living objects (211):
0x80ccdc8: GtkLabel ref_count=1
0x80ccd80: GtkButton ref_count=1
0x80ccd10: GtkLabel ref_count=1
0x80cccc8: GtkButton ref_count=1
0x80ccc58: GtkLabel ref_count=1
0x80ccc10: GtkButton ref_count=1
0x80ccba0: GtkHButtonBox ref_count=1
0x80ccb58: GtkHSeparator ref_count=1
[trimmed in the intrest of bandwidth]
0x8070310: GtkWindow ref_count=1
living objects count = 211

> That should give some clues to what is causing the problem.

Back to gdb now that I got _all_ libs compiled with the -g option...
> 
> > Anyone have any clues? Is this some wierd glibc -> gtk interaction am I
> > doing something wrong?
> 
> It is related to glibc in that not all malloc's dump core on
> double frees, but it certainly is not _caused_ by using glibc.
> 
> Regards,
>                                         Owen
> 
> -- 
> To unsubscribe: mail -s unsubscribe gtk-list-request@redhat.com < /dev/null