> Also I'd like to hear about the principle of such a module from the
> core team. Since it partially defeats the kernel security are there
> other ways to get the frambuffers mapped in memory?  Are there
> additional precautions to take in the driver?

Although I'm not on the core team, let me voice my opinion.  The way I
read your driver (quickly :-), you open up the security hole the new
security in 0.9C is supposed to close.  It would perhaps be better if the
aperture driver only permitted access to specific parts of physical
memory, i.e. those parts where you find your frame buffer.  I'm not too
well aquainted with the variations between different i386 graphics cards,
but this may lead to device dependencies creeping into the driver, or it
may be difficult to do for other reasons.

Regards,

- Havard

------------------------------------------------------------------------------