Re: Consistent page aging....

[ebiederm@xmission.com (Eric W. Biederman)]

Marcelo Tosatti <marcelo@conectiva.com.br> writes:

> On 25 Jul 2001, Eric W. Biederman wrote:
> 
> > 
> > Be very clear on this because I sense some confusion.  We don't
> > ``require'' allocation of swap space to do aging. 
> 
> Right now, we have to make anon pages become swap cache pages (which need
> swap space allocated) to be able to age them in the LRU lists.

Nothing important in the aging algorithm on the active list requires
that the page be in the page cache.  The only immediate difference is
that the test to see if a page is busy needs to be updated to not
assume the page is in the page cache.  

> Sure, we do aging before by just scanning the pte's and the process of
> adding a page to the swapcache is already some kind of aging. 
> 
> I'm talking about the aging in the LRU lists here. 

Me to.
 
> There is no confusion. Its the way 2.4 VM works. 

It is a minor implemtation detail of the 2.4 VM not a fundamental
property.  If fundamental routines like add_to_page_cache didn't make
the assumption that a page they are being called on wasn't on an lru
list it might be straight forward to change.   But as it is you have
to be very careful to make that kind of change.

> See? 

Nope.

Eric
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/