> Since, however, UVM has been changed so that pmap_enter() is
> allowed to fail in the cases which the PT page stealing would
> have helped.  In these cases, uvm_fault() will then simply
> wait until memory is available, and then retry the fault.

The page daemon is likely to swap out processes in the meantime,
effectively doing the same thing (stealing their page table pages);
this stealing happens in a much safer context.

It's also worth noting that I added a printf to the PTP steal path on
the i386mp branch, and I've never actually seen that printf happen on
my MP test machine, even under relatively heavy load; it's much more
likely that processes get SIGKILLed from lack of swap..

So, at least on x86, the ptp-stealing code path is large, complex,
runs extremely rarely, and is no longer necessary for correct
operation even under extreme stress.  Sounds like a perfect candidate
for elimination.

					- Bill