System is DragonFly BSD, xserver 1.4.0 and intel driver fresh 2.2 branch. DRI is fresh git from git://ww2.fs.ei.tum.de/~corecode/git/drm.git (DragonFly port is not yet submitted to upstream), mesa 7.0.2.
Server crashes while I move glxgears window. It might take some seconds, but is reproduceable 100%. It doesn't happen with DRI/EXA (although there is different crash, but it can't be triggered with glxgears). X is restarted after crash, but it isn't able to start any more - I have to reboot machine. Logs/conf/backtraces are from Thinkpad X61s laptop with 965GM chip, but the same happens in desktop with 965Q chip as well.
Backtrace isn't always the same, but the pattern is.
Created attachment 13608 [details]
Created attachment 13609 [details]
Xorg log from crashing session.
Created attachment 13610 [details]
Log from new session - X isn't able to start any more because of errors found in hardware state.
Created attachment 13611 [details]
Although I'm afraid that it's already consequence, there is one of backtraces I get from X core. The patter is always the same, but there might be "... ProcPolyLine(); DamageDamageRegion(); cwPolylines(); XAAPolylinesFallback(); I830Sync(); ..." sequence in it.
Just for info, upgrading Mesa to the latest 7.0 branch code didn't change anything.
I also believe I see this bug.
Ubuntu Hardy, xserver-xorg-video-intel_2.4.1 built from source.
Created attachment 18988 [details]
Log from the affected system - Does not show the issue, but shows configuration
I'm afraid XAA won't be supported and will be removed. Please use EXA (default).
XAA is dead. If it hurts, don't use it.