* [-mm] Make the memory controller more desktop responsive (v2)
@ 2008-04-04 13:21 Balbir Singh
2008-04-04 14:01 ` KOSAKI Motohiro
0 siblings, 1 reply; 3+ messages in thread
From: Balbir Singh @ 2008-04-04 13:21 UTC (permalink / raw)
To: Andrew Morton
Cc: YAMAMOTO Takashi, Paul Menage, linux-kernel, linux-mm,
Pavel Emelianov, Balbir Singh, KAMEZAWA Hiroyuki
Changelog v1
------------
Move back the retries to 5 (KAMEZAWA)
This patch makes the memory controller more responsive on my desktop.
Here is what the patch does
1. It sets all cached pages as inactive. We were by default marking
all pages as active, thus forcing us to go through two passes for
reclaiming pages
2. Removes congestion_wait(), since we already have that logic in
do_try_to_free_pages()
Comments? Flames?
The patch is against 2.6.25-rc8-mm1
Signed-off-by: Balbir Singh <balbir@linux.vnet.ibm.com>
---
mm/memcontrol.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff -puN mm/memcontrol.c~memory-controller-improve-responsiveness mm/memcontrol.c
--- linux-2.6.25-rc8/mm/memcontrol.c~memory-controller-improve-responsiveness 2008-04-04 18:29:57.000000000 +0530
+++ linux-2.6.25-rc8-balbir/mm/memcontrol.c 2008-04-04 18:30:08.000000000 +0530
@@ -604,7 +604,6 @@ retry:
mem_cgroup_out_of_memory(mem, gfp_mask);
goto out;
}
- congestion_wait(WRITE, HZ/10);
}
pc->ref_cnt = 1;
@@ -612,7 +611,7 @@ retry:
pc->page = page;
pc->flags = PAGE_CGROUP_FLAG_ACTIVE;
if (ctype == MEM_CGROUP_CHARGE_TYPE_CACHE)
- pc->flags |= PAGE_CGROUP_FLAG_CACHE;
+ pc->flags = PAGE_CGROUP_FLAG_CACHE;
lock_page_cgroup(page);
if (page_get_page_cgroup(page)) {
_
--
Warm Regards,
Balbir Singh
Linux Technology Center
IBM, ISTL
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [-mm] Make the memory controller more desktop responsive (v2)
2008-04-04 13:21 [-mm] Make the memory controller more desktop responsive (v2) Balbir Singh
@ 2008-04-04 14:01 ` KOSAKI Motohiro
2008-04-04 14:02 ` KOSAKI Motohiro
0 siblings, 1 reply; 3+ messages in thread
From: KOSAKI Motohiro @ 2008-04-04 14:01 UTC (permalink / raw)
To: Balbir Singh
Cc: Andrew Morton, YAMAMOTO Takashi, Paul Menage, linux-kernel,
linux-mm, Pavel Emelianov, KAMEZAWA Hiroyuki
Hi
> @@ -612,7 +611,7 @@ retry:
> pc->page = page;
> pc->flags = PAGE_CGROUP_FLAG_ACTIVE;
> if (ctype == MEM_CGROUP_CHARGE_TYPE_CACHE)
> - pc->flags |= PAGE_CGROUP_FLAG_CACHE;
> + pc->flags = PAGE_CGROUP_FLAG_CACHE;
>
> lock_page_cgroup(page);
> if (page_get_page_cgroup(page)) {
Yes.
in general, cache page create as cold page.
if not, large file copy or streaming file drop all cache easily.
Reviewd-by: KOSAKI
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [-mm] Make the memory controller more desktop responsive (v2)
2008-04-04 14:01 ` KOSAKI Motohiro
@ 2008-04-04 14:02 ` KOSAKI Motohiro
0 siblings, 0 replies; 3+ messages in thread
From: KOSAKI Motohiro @ 2008-04-04 14:02 UTC (permalink / raw)
To: Balbir Singh
Cc: Andrew Morton, YAMAMOTO Takashi, Paul Menage, linux-kernel,
linux-mm, Pavel Emelianov, KAMEZAWA Hiroyuki
On Fri, Apr 4, 2008 at 11:01 PM, KOSAKI Motohiro
<m-kosaki@ceres.dti.ne.jp> wrote:
> Hi
>
>
> > @@ -612,7 +611,7 @@ retry:
> > pc->page = page;
> > pc->flags = PAGE_CGROUP_FLAG_ACTIVE;
> > if (ctype == MEM_CGROUP_CHARGE_TYPE_CACHE)
> > - pc->flags |= PAGE_CGROUP_FLAG_CACHE;
> > + pc->flags = PAGE_CGROUP_FLAG_CACHE;
> >
> > lock_page_cgroup(page);
> > if (page_get_page_cgroup(page)) {
>
> Yes.
> in general, cache page create as cold page.
> if not, large file copy or streaming file drop all cache easily.
>
> Reviewd-by: KOSAKI
sorry, I sent premature document.
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2008-04-04 14:02 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2008-04-04 13:21 [-mm] Make the memory controller more desktop responsive (v2) Balbir Singh
2008-04-04 14:01 ` KOSAKI Motohiro
2008-04-04 14:02 ` KOSAKI Motohiro
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox