Re: Page Replacement in 2.6.10

[Rahul Iyer <rni@xxxxxxxxxxxxxx>]

Hi Rik,
Yes that would work :) I was looking at a 32 bit identifier, but seems 
like that won't happen. However, just one more question.

kswapd calls balance_pgdat() which inturns calls 2 major functions:
- shrink_zone
- shrink_slab

shrink_zone calls shrink_cache, which creates a list of pages that it 
should evict. It then calls shrink_list with this list. shrink_list 
calls try_to_unmap, which does this:

if (PageAnon(page))
                ret = try_to_unmap_anon(page);
        else
                ret = try_to_unmap_file(page);


Now here's where i'm confused:
Where are the disk buffers being handled? wasn't the page cache and 
buffer cache unified to give the "new and improved" page cache? To 
elaborate, suppose i read a file from disk via a read() for instance. 
The disk blocks that i read would be cached somewhere right? I assumed 
it would be in the "buffer cache", which in Linux's case is the page 
cache. But i cannot see these types of pages being handled :(
Thanks
Rahul



Rik van Riel wrote:

> On Sun, 27 Mar 2005, Rahul Iyer wrote:
>
>  
>
> > ok, i can store the swp_entry_t as an id for the anonymous pages... but what
> > about file backed ones? there doesn't seem to be any id as there needs to be
> > none. Am i right here?
> >    
>
> How about storing (mapping, offset) tuples ?
>
>  

--
Kernelnewbies: Help each other learn about the Linux kernel.
Archive:       http://mail.nl.linux.org/kernelnewbies/
FAQ:           http://kernelnewbies.org/faq/