Comment 12 for bug 537640

Robert, thanks for identifying that radeon.new_pll=0 works around this. That indicates this is an issue in the kernel drm, so needs either an appropriate patch or quirk to resolve, so I'm refiling this to the kernel team to look into.