fc4 on x86_64, gtk2 weird behaviour

[Roger Heflin]

> -----Original Message-----
> From: fedora-list-bounces at redhat.com 
> [mailto:fedora-list-bounces at redhat.com] On Behalf Of oleksandr korneta
> Sent: Thursday, January 05, 2006 9:00 PM
> To: mai11ist at fastmail.fm; For users of Fedora Core releases
> Subject: Re: fc4 on x86_64, gtk2 weird behaviour
> 
> so seems like no one knows what the problem might be...
> not good
> 
> 
> on 01/04/2006 11:15 PM oleksandr korneta wrote:
> > Hello All,
> > 
> > I'm fighting with FC4 trying to make it work on x86_64 
> (Turion) based 
> > laptop. Honestly saying there is a whole bunch of troubles 
> so far. But I
> >    hope (still) to handle those.
> > 
> > Anyway here is the point. Something weird has happened to 
> all gtk2-based 
> >   applications. Whenever you try to open Open/Save dialog 
> window the 
> > program just hangs and you have to kill it. If you run it 
> as regular 
> > user there is no output to console, however running 
> something (let's 
> > say
> > gedit) from the root terminal seems to yield this:
> > 
> > # gedit
> > *** glibc detected *** gedit: malloc(): memory corruption: 
> > 0x00000000009a7c80 ***
> > 
> > 
> > again application just hangs.
> > 
> > I have to say that, firefox (1.5) behaves a bit 
> differently. It also 
> > gets stuck, but not always. Sometimes it works just fine 
> for some reason.
> > 
> > The only thing I recall, one can blame on is that I've 
> updated gnome 
> > to
> > 2.12 from http://www.nrpms.net repository. But I'm not sure 
> about that 
> > being a reason.
> > 
> > any suggestions?
> > 


I have seen the firefox crash, there is a bugzilla report for it.

Compiling firefox-1.5 myself with -O1 seems to have made it stable
and decently fast.   With -O2 it crashes on a save on certain web
pages, with -O0 is it stable but slow.

The crashes are all deep in gtk2, and I had this crash on the 
default fc4 setup, and also with a number of gtk updates applied.

I think it is a gtk2 bug, possibly an memory overrun, as the -O1 seems
to prevent it from happening (less packing of memory?), all of my 
tests are only by recompiling firefox, not by messing with gtk2.

                                          Roger