For a long time now, X sometimes hangs at startup, the screen remains black. I haven't really tried it out, but I'd say the success rate is about 50%. By using "git bisect" I now have found out the latest revision that seems to always work for me, it's 36799e3ea7ddde5fce528e8bc17ea942933a9111 . Ironically the next commit is supposed to "fix some remaining lockup scenarios" when "setting the memory map" - well, on my system, it seems to create one :-) Please tell me if you need more information.
Created attachment 7595 [details] Xorg log of the crash
Created attachment 7596 [details] log of working revision
Created attachment 7597 [details] xorg.conf
By taking revision ati-6_6_2 (because head didn't compile) an backing out the changes from 4d6656062129da0489eb4bc898871379ba891d8f, which is the commit that adds workarounds to fix some lockups, I now have a driver that hasn't locked up so far and even allows using aiglx. But unfortunately an other problem reappeared :-( again, in half of the cases X performance seems to be very low (although, for example, aiglx still works). I have attached the diff of Xorg.log in the good and in the bad case. I use 24bit colours at the moment, I'll try 16bit too.
Created attachment 7652 [details] Diff between good and bad Xorg.log
This seems to be a very strange problem. Now I've tried to use the fglrx driver, I even unloaded the radeon module before starting the X server again, but X is _still_ slow. Any ideas what could be wrong?
Could you try the latest GIT version of xserver-xorg-video-ati? The recent fix to bug 9284 solved hang problems for many people on Radeon 9500/9600 cards at least. But you also might have something else, of course. Also note that installing fglrx generally breaks open source 3D from working, as it installs its own libGL among else. It's best to (try to) completely remove fglrx before testing with the open source driver.
I'm sorry, I don't have that card anymore.
So, i mark it as a duplicate of another r300 lockup. Even thought your lockup did might came from somethings else, i believe that most, if not all, r300 lockup are somehow deeply related. *** This bug has been marked as a duplicate of bug 9252 ***
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.