On Sun, Sep 05, 2004 at 10:58:07AM -0700, Linus Torvalds wrote:
> 
> 
> On Sun, 5 Sep 2004, Arjan van de Ven wrote:
> > 
> > well... we have a reverse mapping now. What is stopping us from doing
> > physical defragmentation ?
> 
> Nothing but replacement policy, really, and the fact that not everything
> is rmappable.
> 
> I think we should _normally_ honor replacement policy, the way we do now.  

yes it absolutely is quite a heavy hammer. 
However right now the alternative (free a LOT of memory and hope it
collapses into higher order ones) is even heavier, freeing the wrong 8 pages
is less of a disturbance than freeing 8000 of the mostly wrong pages ;) 

I absolutely agree this heavy hammer only should trigger if there is a
request for a higher order page that isn't there. Doing it from a special
thread does make sense, makes it relatively easy to keep track of such
wakeups and more importantly rate limit them etc.