Re: Addendum to Daniel Phillips [RFC] use-once patch

[Steven Walter <srwalter@yahoo.com>]

On Fri, Jul 27, 2001 at 04:38:18PM +0200, Daniel Phillips wrote:
> > may be
> > +                * it is the best candidate for swapping.
> > +                */
> > +               if ((page->age > PAGE_AGE_MAX) && (page_count(page)
> > <= 1)) {
> > +                       page->age = PAGE_AGE_START;
> > +               } else {
> > +                       page->age += PAGE_AGE_ADV;
> > +                       if (page->age > PAGE_AGE_MAX) {
> > +                               page->age = PAGE_AGE_MAX;
> > +                       }
> > +                       return;
> > +               }
> 
> I noticed your good benchmark results below, but I'm having some 
> trouble understanding how this works.  How can page->age ever become 
> greater than PAGE_AGE_MAX?  Also, I don't see any reference to 
> PAGE_MAX_USE.  Comments?

What if page->age is equal to PAGE_AGE_MAX when it hits the 'else'
statement.  It will be unconditionally incremented by PAGE_AGE_ADV, and
then it will be greater than PAGE_AGE_MAX.  The inner 'if' statement
catches this, and sets the age back to PAGE_AGE_MAX
-- 
-Steven
In a time of universal deceit, telling the truth is a revolutionary act.
			-- George Orwell