lkml.org 
[lkml]   [2009]   [Oct]   [15]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 8/9] swap_info: note SWAP_MAP_SHMEM
On Thu, 15 Oct 2009 23:23:24 +0100 (BST)
Hugh Dickins <hugh.dickins@tiscali.co.uk> wrote:

> On Thu, 15 Oct 2009, KAMEZAWA Hiroyuki wrote:
> > On Thu, 15 Oct 2009 01:57:28 +0100 (BST)
> > Hugh Dickins <hugh.dickins@tiscali.co.uk> wrote:
> >
> > > While we're fiddling with the swap_map values, let's assign a particular
> > > value to shmem/tmpfs swap pages: their swap counts are never incremented,
> > > and it helps swapoff's try_to_unuse() a little if it can immediately
> > > distinguish those pages from process pages.
> > >
> > > Since we've no use for SWAP_MAP_BAD | COUNT_CONTINUED,
> > > we might as well use that 0xbf value for SWAP_MAP_SHMEM.
> > >
> > > Signed-off-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
> >
> > I welcome this!
>
> Ah, I did wonder whether you might find some memcg use for it too:
> I'm guessing your welcome means that you do have some such in mind.
>
yes, I'm thinking I can use this or not on memcg for simplifying memcg's hooks
for shmem. It's complicated ;)
I have to test memcg+shmem carefully again after this patch but I think
there will be no trouble, now.

> (By the way, there's no particular need to use that 0xbf value:
> during most of my testing I was using SWAP_MAP_SHMEM 0x3e and
> SWAP_MAP_MAX 0x3d; but then noticed that 0xbf just happened to be
> free, and also happened to sail through the tests in the right way.
> But if it ever becomes a nuisance there, no problem to move it.)
>

Hmm. I myself have no troubles whatever free vaule is used.
let me clarify..

xx00 0000
xx11 1110 - swap count max
01xx xxxx - swap has cache
1xxx xxxx - swap count has continuation
1x11 1111 - swap for shmem

seems not very bad.

Regards,
-Kame



\
 
 \ /
  Last update: 2009-10-16 02:11    [W:0.246 / U:0.056 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site