Greywolf <greywolf@starwolf.com> writes:
> Thus spake Perry E. Metzger ("PEM> ") sometime Today...
> 
> PEM> Consider, for example, that all such cards have DMA controllers on
> PEM> board -- DMA controllers which can be programmed to move bits into and
> PEM> out of arbitrary portions of memory.
> 
> Oh, icky.  I was completely unaware of this!  Thank you for the edification.
> 
> However, since the DMA controllers can be programmed, what if it was
> set up that only the kernel could get to the DMA, and thus could,
> at that point, dictate where the DMA was going to happen?

Sure. That's called "moving X into the kernel." In order for X to work
as currently built, it needs the ability to do fine grained control of
the video card.

Until we start somehow building kernel X drivers (likely never),
you're stuck.


-- 
Perry E. Metzger		perry@piermont.com