From: mingo/UNIX (mingo@pc5829.hil.siemens.co.at)
Date: ## 04/25/96 12:24 ##
> > On Thu, 25 Apr 1996
> > Hemment_Mark/HEMEL_IOPS_INTERNATIONAL-SALES@uniplex.co.uk wrote:
> >
> > the _minimum_ that is being used for page tables by user processes
> > is 20*2 = 40 pages (160KB). Ok, thats not v. large, but some of those
> > processes spent alot of time asleep (hours maybe). Why should they be
> > allowed to hold on to a valuable resource which they are not using?
> 160KByte ... unless there is an elegant way to do it without loosing
> cycles in the page fault handler ... no way! :) I would rather use my
> shadow memory then :))))
Ok, so if a secondary page table for a process (that has been idle
for sometime) needs to be faulted in, that's one extra fault on the
first reference to the 4MB range it covers. That's not much of an
overhead...
While the page table(s) are out, the released page(s) are available for
other uses. As more free memory is available to the system it
_should_ reduce the need to page-out (and fault back in again) other
pages which are more more frequently referenced than those table
pages.
It's a balancing act - taking a resource from where it is not being
used effectively, and giving it to another area that can use it better.
The problem is getting the balance right (espically on a Friday
night...:)
markhe