On Tue, Apr 10, 2001 at 12:49:38PM -0400, der Mouse wrote:

 > Perhaps it would be interesting to try uvm_io versus copying through a
 > kernel buffer and see how it affects performance for various I/O sizes
 > (and on various machines - I'm sure uvm_io is more expensive on some
 > architectures than others!).

I hate to burst your bubble, but uvm_io() copies through a kernel
buffer :-) It temporarily maps the user destination pages into kernel
space and does the I/O to them.

-- 
        -- Jason R. Thorpe <thorpej@zembu.com>