On Mon, Jul 28, 2008 at 19:34:34 +0300, Arthur A wrote: > Anyway, since we don't yet have a reliable workaround and this can cause > data loss (just happened!) I'd like to ask that the severity of this bug > be bumped up. Bumping the severity won't magically fix the bug, so no. Does the ExaNoComposite option work around the bug? What about disabling DRI? Cheers, Julien