Jason Thorpe writes:
 > On Mon, 15 Mar 1999 13:31:43 -0800 
 >  "Neil A. Carson" <neil@causality.com> wrote:
 > 
 >  > So I had this thought last night, on systems where context switches
 >  > involve long-time cache invalidations. When doing the context switch in
 >  > the kernel, do not invalidate the user mappings [until you must].
 >  > Does this make a whole lot of sense? Comments from Jonathan et al
 >  > welcome.
 > 
 > It makes perfect sense... and can be implemented completely in
 > machdep code :-)

FYI: the Sun3 port already does this optimization.
(See pmap_activate and pmap_switch for details.)