*** glibc detected *** munmap_chunk(): invalid pointer: 0xb7f1a470 ***

[NTICompass]

Originally Posted by JG^

i've had the same problem with all 180.* drivers (gentoo 32bit), but the new 180.35 seems to resolve this. firefox-bin works fine now without the munmap_chunk crash.
although i can't be sure if the driver upgrade alone fixed it, because portage also updated 21 other packages that day (qt-core poppler texinfo dialog libXi randrproto libXfont pixman man-pages fontforge ruby qt-sql xf86-video-vesa raptor soprano poppler-bindings imagemagick xine-lib gst-plugins-faad vlc libquicktime).

if it is the driver, thanks for fixing it

Using 180.37 of Nvidia drivers on Gentoo x64 still has this error.

[rapsure]

If you have upgraded from gcc 4.1.2 -> 4.3.x and haven't done an emerge -e system && emerge -e world then please do it to make sure that the problem still exists.

[kernelOfTruth]

Originally Posted by rtcm

AFAIK, the glitz cairo backend is *completly unsupported* by the cairo developers...

since when ?

the time I started using bery, compiz and compiz-fusion it was still in gentoo-wiki.com and afaik a needed dependency for it to work

so it isn't needed anymore ?

yeah ! let's re-compile the system

hopefully this fixes my problems too

[rtcm]

Originally Posted by kernelOfTruth

since when ?

the time I started using bery, compiz and compiz-fusion it was still in gentoo-wiki.com and afaik a needed dependency for it to work

so it isn't needed anymore ?

yeah ! let's re-compile the system

hopefully this fixes my problems too

Check the thread starting at http://lists.cairographics.org/archi...ry/016646.html

It has some more interesting posts in the March archive.

[Mike Playle]

It's interesting to come across this thread.

I've been having a similar problem with an application I've been developing. I haven't submitted a bug report because I haven't been able to trim it down to a usefully small example, and I couldn't be sure that it wasn't my own bug!

For what it's worth the problem seemed to be triggered by a call to glBindFramebuffer().

[kernelOfTruth]

Originally Posted by rtcm

Check the thread starting at http://lists.cairographics.org/archi...ry/016646.html

It has some more interesting posts in the March archive.

thanks a lot rtcm !

unfortunately the ebuilds seem to disagree:

emerge: there are no ebuilds built with USE flags to satisfy "x11-libs/cairo[glitz]".
!!! One of the following packages is required to complete your request:
- x11-libs/cairo-1.8.6-r1 (Change USE: +glitz)
(dependency required by "x11-plugins/compiz-plugins-extra-0.8.2" [ebuild])
(dependency required by "@compiz-fusion" [argument])

anyone has compiz(-fusion) with glitz-disabled cairo running here ?

[hdas]

Late here, but just to say, I am running compiz-fusion without glitz on Gentoo.

Just to clarify, I realized from my Ubuntu system, where these crashes didn't happen, that there is no glitz installed in Ubuntu. So I added USE=-glitz and recompiled cairo. And then did a revdep-rebuild to compile everything with USE=-glitz. I edited the compiz ebuild to remove the glitz dependency (and other ebuilds which require glitz or cairo with glitz). The result - it all emerges fine and works fine. So I now have a completely glitz-free system and no gtk app segfaults any more. However, since nvidia-177 series used to work fine, I still feel Nvidia should look into it and rectify. And so do the glitz guys, since this doesn't happen with Qt or such.

[KDE]

Firefox 3.5 on 180.60 works for me.