Comment 58 for bug 556232

Allan, actually you're correct, with current -ati in Lucid we've moved to Kernel modesetting (KMS) which means the kernel drm driver sets up external screens and so on. So your findings that the issue is tied to particular kernel versions is in keeping with this, and it sounds like this is a regression introduced presumably at or before 2.6.32-10 when you started noticing the problem.

Reviewing the changelog for the -10 kernel sounds like the next step for this bug, but I'll turn it over to the kernel team at this point. Good luck.