Is clean-room reverse engineering even viable? That is, would anyone agree that an independent discovery of what bit does what is even possible, and this is not lifted from known drivers?
To put it another way -- the IP protections on the current library are a matter of copyright law, and the protections are on the code itself. The facts of which bits do what -- which are embodied in that code -- are not protected (or protectable!), and can be reused freely.
Though given Oracle's current legal proceedings that might change...