I suppose I would first need to reinstall Git versions of all components along with a 2.6.33rc kernel and I'd rather not go through that again :P
Look through the bugs on:
https://bugs.freedesktop.org
If you see one that matches your bug, add a comment about your issue; if not, file a new bug.
I suppose I would first need to reinstall Git versions of all components along with a 2.6.33rc kernel and I'd rather not go through that again :P
Thanks to the people in #radeon on freenode! They pointed me to this bugreport in the kernel tracker: http://bugzilla.kernel.org/show_bug.cgi?id=15186
Though it is marked as regression I already had this with vanilla sources in 2.6.32 (so for me it is no regression). The issue though was fixed with the 2nd patch that was attached, as I wrote in my comments.
Yes, the issue itself is basically what can be observed in the screenshots RealNC. Looks like this (crude hack) does solve things, will report back if I still see those issues.
Oh, I might as well mention another bug I've been experiencing. On my dual screen setup (1280x1024 + 1680x1050), the framebuffer resolution for both screens is set to the lower resolution, 1280x1024.
So on my large screen, I get a white box around the framebuffer in the space it didn't fill.
It's not a bug. It's a limitation of the kernel framebuffer interface. It has no concept of multiple heads. In order to make sure no console information is lost the console is run at the largest mode supported by all attached heads. If you want full output flexibility, you'll need to use the kms interface like the X driver uses.