This is probably the same issue as bug 5376 but I have different testcase and setup so I created a new bug. Hardware and OS config: Thinkpad T43 with ATI Radeon Mobility X300 (M22) 5460 (PCIE) Debian unstable/experimental Linux 2.6.21 kernel (custom build) xserver-xorg 7.2-2 xserver-xorg-video-ati 6.6.191-1 libgl1-mesa-dri 6.5.2-4 libdrm2 2.3.0-4 The problem occurs only when using the hardware cursor with the radeon driver (the default). I am launching a 'failsafe' session from kdm, only a single terminal is running, no window manager or anything else. I have the crystal xcursor theme installed as the default cursor theme. When I launch the GL app neverball in windowed mode, I can cause the entire system to freeze by moving the mouse cursor into and out of the GL window repeatedly. Neverball changes the mouse pointer to a black and white arrow so it looks like switching the cursor is triggering the problem. I cannot reproduce the problem when running glxgears and moving the cursor into and out of the window, glxgears does not appear to be changing the cursor image. Enabling the software cursor with "SWcursor" "on" does not cause the system to freeze anymore, although the cursor flickers and is barely visible in GL windows. The "SilkenMouse" option does not have any effect, the system locks up whether it is set on or off. I'm also not using MergedFB. I will attach my xorg.conf and Xorg.0.log, let me know if you need any more information.
Created attachment 9807 [details] xorg.conf
Created attachment 9808 [details] Xorg.0.log
Does it also happen without Option "DynamicClocks"?
Yes, the freeze still occurs with DynamicClocks set on or off.
Created attachment 9994 [details] [review] HW cursor cleanup Would this patch happen to help at all?
That patch does happen to help, in fact I'd say it fixed the problem on my system. I built the driver from a fresh git checkout (ati-6.6.191-18-ga3ee422) and could quickly cause a lockup, but after applying your patch I have not been able to lock my system yet after a couple of hours of testing. Thank you very much! I have experienced this problem with every recent xserver release and with kernels back to 2.6.15 but had never been able to consistently reproduce it until now because it seemed so random.
Pushed to git, thanks for testing.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.