K 10
svn:author
V 5
kmacy
K 8
svn:date
V 27
2010-04-27T05:39:13.264785Z
K 7
svn:log
V 416
As it stands now, with the exception of hold_count, all vm_page
modification  acquires both the page queue lock and the appropriate
page lock. The assumption being that over time we can incrementally
wean ourselves from the page queue mutex where possible. The next
field to rely strictly on page_lock will be wire_count.

This currently appears to work ok on normal compile workloads in
addition to stress2's swap.

END