Comment 9 for bug 558657

Hm. Following the cryptic references in the RH bug, it's possible that there's a kernel commit that would fix this:

commit 1cfa49c2f1f3f42cf166438c17dafac577610836
Author: Ben Skeggs <email address hidden>
Date: Tue Feb 9 16:59:26 2010 +1000

    drm/nv50: EVO uses a real ring buffer, not jumps back to start like PFIFO

    Tested on NV84GL and NV96 with cursor state caching disabled to force
    lots of EVO use, no problems seen.

    There's been some reports of EVO randomly dying and reporting an error
    during cursor state changes for some people, I've been unable to
    reproduce on my hardware but hopefully this is the culprit.

    Signed-off-by: Ben Skeggs <email address hidden>

I thought we had that in our tree, but it seems that we don't.