It turns out that if we have two clients, one opening /dev/dri/card0 and one opending /dev/dri/controlD64, both will attempt to become active master. That seems a little illogical to me (and makes the vmwgfx driver BUG). Shouldn't we have at most one active master per device, and hence attempt to replace file_priv->minor->master with file_priv->minor->dev->active_master /Thomas _______________________________________________ dri-devel mailing list dri-devel@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/dri-devel