Red Hat Bugzilla – Bug 209338
compiz segfaults xserver on radeon with exa enabled
Last modified: 2007-11-30 17:11:45 EST
Description of problem:
When i start compiz (either command-line or via desktop-effects) and have the
line Option "AccelMethod" "EXA" in my xorg.conf Xorg segfaults. When i remove
the offending option, compiz runs fine. This all was working fine around the
FC6Test2 time, however i havent followed development close, so i cant say when
When Xorg segfaults i get the following in my dmesg:
kernel: Xorg general protection rip:2aaaaaabeb9c rsp:e87a2949a36315fe error:0
The closest thing searching the web and bugzilla which sounds similiar to my
problem is in the xorg-ml, however i havent found an follow-up to this post:
Version-Release number of selected component (if applicable):
uptodate fc6-rawhide sys (as of 10/04/06)
Steps to Reproduce:
1. change AccelMethod to EXA
2. start compiz, either manually or via desktop-effects
3. watch the crash
Xorg should not crash
ATI X700 PCIe Mobility
I rebuild xorg-x11-server-Xorg-1.1.1-47, but commented out patch #107
Now compiz wont segfault Xorg when EXA is enabled, everything is fine. The patch
was added on August, 7th. This was the time when FC6test2 was new and i had no
problems. So could someone explain me what this patch does, and why it is needed
(as it breaks exa, at least on my x700).
Yes, compiz and EXA aren't compatible, the offscreen-pixmaps patch is XAA
specific. Right now the solution is "don't use EXA", and with compiz EXA vs XAA
doesn't make a difference anyway. There's should be more graceful error
handling, sure, or maybe we should just make it work with EXA.
Thank you for the info, Kristian. What is the purpose of the offscreen-pixmaps
patch ? Could it be possible to document this behaviour (it took me some time to
find out, that the EXA option is to blame for the crashes is experienced with
Thank you anyway for the good work.
Fedora Core 5 and Fedora Core 6 are, as we're sure you've noticed, no longer
test releases. We're cleaning up the bug database and making sure important bug
reports filed against these test releases don't get lost. It would be helpful if
you could test this issue with a released version of Fedora or with the latest
development / test release. Thanks for your help and for your patience.
[This is a bulk message for all open FC5/FC6 test release bugs. I'm adding
myself to the CC list for each bug, so I'll see any comments you make after this
and do my best to make sure every issue gets proper attention.]
Even after almost three months there has been no activity on these bugs, so we
consider them obsolete and they are being closed as INSUFFICIENT_DATA.
Please, if you are able to reproduce this bug on currently supported (FC6 and
F7) non-test distribution, or if there is any other reason why this bug should
be kept alive, please, reopen, change the product to something more recent and
add an additional information to the comment of such bug.
Thank you very much.