On Tue, 2005-02-15 at 22:23, Chuck Silvers wrote:
> On Tue, Feb 15, 2005 at 05:00:09PM -0500, Stephan Uphoff wrote:
> > Mhhh .. for SMP it could be beneficial just to call pmap_remove for the
> > whole range before the pmap_enter calls to cut down on the number of IPI
> > calls. (At least for multi page reads on i386)
> 
> hmm, on the second thought, the IPI coalescing should be taken care of
> by deferring them until pmap_update() (which I was also missing).
> the i386 pmap doesn't take advantage of this currently, though.

Yes, fixing the i386 pmap to take advantage of  pmap_update() is
probably a better idea.

I am planning another attack on Manuel Bouyer's "TLB IPI rendezvous
failed..." problem sometime in the next couple of weeks and might as
well take a look at  pmap_update() while I am in the neighborhood.

Stephan