Thanks, Anisse, but Dave's answer was wrong anyway. Using a different user-space-driver wouldn't have magically fixed the bug in the kernel KMS driver part.
Eventually something newer will pop up in Debian SID and/or Debian Backports. I'll then pick it. As the performance of the device doesn't bother me, I see no need. The more I stay at what Debian provides, the less I have to maintain by myself :-)
