> Perhaps is it better to change the calculation of (uvmexp.active +
> uvmexp.inactive + uvmexp.free) in uvmpd_scan_inactive() to
> uvmexp.npages?

I don't think it's better since pagedaemon can do something only for
pages on the queues.

> BTW, do you have any idea what was happening at the following case?
> It seems the anonymous memory didn't exceed anonmax at that point,
> although it's not 100% sure.

There is no exact number (e.g., vmstat -s etc) of `that point' but at
least the number in the
http://mail-index.netbsd.org/current-users/2004/10/07/0014.html tells
that anon reaches to the max (600MB of anon vs 700MB of pages on the
queue).

enami.