At 13:47 8/23/95 -0400, Jonathan Shapiro wrote:
>The ager runs, and places some set of object (page or node) frames
>into the free list ...
>
>Subsequent to the ager running, but prior to the object's frame being
>allocated off the free list, some process runs and prepares a key to
>an object whose frame has been placed on the free list. The object is
>then promoted into the youngest generation, which removes the object
>frame from the free list as a side effect. This is what I mean by
>revivification.
If a particular revivfication would drive the free pages below the limit needed for migration (or some other proof of progress), it could be dis-allowed and the read could be made from the disk. This should be uncommon since the free-list pages are not recently used.
Bill