sommerfeld@orchard.arlington.ma.us said:
> > Finally, I've been looking at implementing pmap_copy to duplicate an 
> > entire pmap (this is potentially cheaper than taking a large number of
> > faults when a new process starts up). [...]

> There was a verdict in a related case, though: there was a pmap_copy
> implementation in the i386 pmap; it turned out to be a net lose and
> was disabled. 

I can understand why it might be: my initial experiments have shown a 
similar small loss in performance with this feature (which is a direct 
port of the i386 version but without some of the bells and whistles).  
However, the rules on the ARM may be sufficiently different that being 
able to map the pmap cacheable during the copy might be enough to more 
than recover the loss.

R.