[Devel] Re: [PATCH 6/6 mm] memcgroup: revert swap_state mods

[Hugh Dickins]

On Fri, 9 Nov 2007, KAMEZAWA Hiroyuki wrote:
> On Fri, 9 Nov 2007 07:14:22 +0000 (GMT)
> Hugh Dickins <hugh at veritas.com> wrote:
> 
> > If we're charging rss and we're charging cache, it seems obvious that
> > we should be charging swapcache - as has been done.  But in practice
> > that doesn't work out so well: both swapin readahead and swapoff leave
> > the majority of pages charged to the wrong cgroup (the cgroup that
> > happened to read them in, rather than the cgroup to which they belong).
> 
> Thank you. I welcome this patch :)

Thank you!  But perhaps less welcome if I don't confirm...

> Could I confirm a change in the logic  ?
> 
>  * Before this patch, wrong swapcache charge is added to one who
>    called try_to_free_page().

try_to_free_pages?  No, I don't think any wrong charge was made
there.  It was when reading in swap pages.  The usual way is by
swapin_readahead, which reads in a cluster of swap pages, which
are quite likely to belong to different memcgroups, but were all
charged to the one which is doing the fault on its target page.
Another way is in swapoff, where they all got charged to whoever
was doing the swapoff (and the charging in unuse_pte was a no-op).

>  * After this patch, anonymous page's charge will drop to 0 when
>    page_remove_rmap() is called.

Yes, when its final (usually its only) page_remove_rmap is called.

Hugh
_______________________________________________
Containers mailing list
Containers at lists.linux-foundation.org
https://lists.linux-foundation.org/mailman/listinfo/containers