Subject: Re: user mapping pci memory
To: Gordon W. Ross <gwr@mc.com>
From: Chris G. Demetriou <cgd@cs.cmu.edu>
List: tech-kern
Date: 03/12/1997 12:32:41
> [ ... A potential solution to the mmap annoyances ... ]

What you suggest could probably work to allow general MI exporting of
address space, but still doesn't solve the problems of:

	(1) how to communicate to user-land that what's mapped really
	    doesn't start at the beginning of the page,

	(2) transforming the physical address that the device/bus may
	    present into a system bus physical address, and

	(3) sanely informing the user-land process how it's supposed
	    to access the mapped region.



cgd