> Certainly faster, the difference is probably even bigger for the slower 
> risc-pc's.
> 
> The fork/exit loop comes in with:
> 8.43 real         0.00 user         2.07 sys
> compared with previous to this patch
> 9.17 real         0.00 user         2.06 sys

So, I finally got around to propagating the pmap changes I had made at 
home to my shark here at work.  As you will recall, when I first posted 
timings for this test I was running a fairly standard SHARK kernel with no 
local hacks.  The timings were nearly identical to the kinetic numbers 
above, vis

Shark/NetBSD(SA110/233MHz)  	real 9.2  user 0.02 sys 1.9

With my pmap changes applied these numbers come down to

real 7.6  user 0.02  sys 1.7

And with Jason's patch, I now get

real 6.9  user 0.02  sys 1.6

which in my book is pretty darn close to 25% improvement on the original 
figures.

Time to polish off those changes and get them posted me-thinks.

R.