From: "Pádraig Brady" <P@draigBrady.com>
To: Mel Gorman <mgorman@suse.de>
Cc: linux-mm@kvack.org
Subject: Re: sandy bridge kswapd0 livelock with pagecache
Date: Fri, 24 Jun 2011 14:10:30 +0100 [thread overview]
Message-ID: <4E048CC6.5010400@draigBrady.com> (raw)
In-Reply-To: <20110624114444.GP9396@suse.de>
On 24/06/11 12:44, Mel Gorman wrote:
> On Thu, Jun 23, 2011 at 08:25:40PM +0100, P?draig Brady wrote:
>> On 23/06/11 17:59, Mel Gorman wrote:
>>> On Thu, Jun 23, 2011 at 04:32:27PM +0100, P?draig Brady wrote:
>>>> On 23/06/11 16:24, Mel Gorman wrote:
>>>>>
>>>>> Theory 2 it is then. This is to be applied on top of the patch for
>>>>> theory 1.
>>>>>
>>>>> ==== CUT HERE ====
>>>>> mm: vmscan: Prevent kswapd doing excessive work when classzone is unreclaimable
>>>>
>>>> No joy :(
>>>>
>>>
>>> Joy is indeed rapidly fleeing the vicinity.
>>>
>>> Check /proc/sys/vm/laptop_mode . If it's set, unset it and try again.
>>
>> It was not set
>>
>>>
>>> diff --git a/mm/vmscan.c b/mm/vmscan.c
>>> index dce95dd..c8c0f5a 100644
>>> --- a/mm/vmscan.c
>>> +++ b/mm/vmscan.c
>>> @@ -2426,19 +2426,19 @@ loop_again:
>>> * zone has way too many pages free already.
>>> */
>>> if (!zone_watermark_ok_safe(zone, order,
>>> - 8*high_wmark_pages(zone), end_zone, 0))
>>
>> Note 8 was not in my tree.
>> Manually applied patch makes no difference :(
>> Well maybe kswapd0 started spinning a little later.
>>
>
> Gack :)
>
> On further reflection "mm: vmscan: Prevent kswapd doing excessive
> work when classzone is unreclaimable" was off but it was along the
> right lines in that the balancing classzone was not being considered
> when going to sleep.
>
> The following is a patch against mainline 2.6.38.8 and is a
> roll-up of four separate patches that includes a new modification to
> sleeping_prematurely. Because the stack I am working off has changed
> significantly, it's far easier if you apply this on top of a vanilla
> fedora branch of 2.6.38 and test rather than trying to backout and
> reapply. Depending on when you checked out or if you have applied the
> BALANCE_GAP patch yourself, it might collide with 8*high_wmark_pages
> but the resolution should be straight-forward.
>
> Thanks for persisting.
Bingo!
--
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/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
next prev parent reply other threads:[~2011-06-24 13:10 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-06-21 9:53 sandy bridge kswapd0 livelock with pagecache Pádraig Brady
2011-06-21 10:39 ` Mel Gorman
2011-06-21 10:47 ` Pádraig Brady
2011-06-21 11:34 ` Mel Gorman
2011-06-21 11:59 ` Pádraig Brady
2011-06-21 13:07 ` Mel Gorman
2011-06-21 14:23 ` Pádraig Brady
2011-06-22 9:44 ` Mel Gorman
2011-06-22 10:19 ` Pádraig Brady
2011-06-23 11:46 ` Mel Gorman
2011-06-23 13:04 ` Pádraig Brady
2011-06-23 15:24 ` Mel Gorman
2011-06-23 15:32 ` Pádraig Brady
2011-06-23 16:59 ` Mel Gorman
2011-06-23 19:25 ` Pádraig Brady
2011-06-24 11:44 ` Mel Gorman
2011-06-24 13:10 ` Pádraig Brady [this message]
2011-06-24 15:04 ` Mel Gorman
2011-06-24 6:33 ` Shaohua Li
2011-06-21 14:34 ` Mel Gorman
2011-06-21 15:29 ` Pádraig Brady
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=4E048CC6.5010400@draigBrady.com \
--to=p@draigbrady.com \
--cc=linux-mm@kvack.org \
--cc=mgorman@suse.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.