K 10
svn:author
V 7
attilio
K 8
svn:date
V 27
2013-03-22T22:52:02.345587Z
K 7
svn:log
V 766
- Save some further atomics and introduce vm_page_wakeup_locked() that
  assumes vm_page_lock held.  It is implemented as an hard function
  right now because it is only a WIP and it avoids a namespace
  pollution.  It will be cleaned before to merge back.
  Maybe it makes sense to always acquire the vm_page_lock() in the
  users for vm_page_wakeup() and get rid of vm_page_wakeup_locked()
  finally.  This will be discussed more.
- Fix a bug where vm_page_remove() is called for UNNAMED pages and
  the page lock is not held.  Don't make assumption about the lock in
  that case and just deal with the situation, even if it is a tricky
  pattern. We use a boolean_t rather than a mtx pointer. [0]

Sponsored by:	EMC / Isilon storage division
[0] Reported by:	pho

END