>On Wed, 03 Mar 1999 16:55:30 -0800 

>That's bus_space_mmap(), and it gives a value that you pass to
>pmap_phys_address().
>
>bus_space_mmap() isn't implemented yet, but needs to be.

okay. understood.  gotcha. That also covers buses which just cannot do
DMA, but which do have mmap()able shared-memory framebuffers.
but what should we do in the meantime?

How about i dink in a ``machine_btop()'' into the TC code, in order to
clean up Nisimura-san's wscons drivers and to make progress with the
ioctl asic DMA support (for faking mmap() for audio?)?

On the STRICT understanding that it will go away when
bus_space_mmap() exists and provides the necessary functionality?