mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: "Dmitry Adamushko" <dmitry.adamushko@gmail.com>
To: "Micah Dowty" <micah@vmware.com>
Cc: "Ingo Molnar" <mingo@elte.hu>,
	"Christoph Lameter" <clameter@sgi.com>,
	"Kyle Moffett" <mrmacman_g4@mac.com>,
	"Cyrus Massoumi" <cyrusm@gmx.net>,
	"LKML Kernel" <linux-kernel@vger.kernel.org>,
	"Andrew Morton" <akpm@osdl.org>, "Mike Galbraith" <efault@gmx.de>,
	"Paul Menage" <menage@google.com>,
	"Peter Williams" <pwil3058@bigpond.net.au>
Subject: Re: High priority tasks break SMP balancer?
Date: Mon, 19 Nov 2007 23:22:06 +0100	[thread overview]
Message-ID: <b647ffbd0711191422q6be59c44gfbc72f0fc3167768@mail.gmail.com> (raw)
In-Reply-To: <20071119185116.GA28173@vmware.com>

On 19/11/2007, Micah Dowty <micah@vmware.com> wrote:
> [ ... ]
>
> micah@micah-64:~$ cat /proc/schedstat
> version 14
> timestamp 4366722208
> cpu0 0 0 0 785868111 0 828020771 12621812 22703872 14096041 9504937730116 23603703739504 815398959
> domain0 03 8766051 8760123 95 14544377 6374 0 13 8760110 7369 6771 0 20835105 605 0 1 6770 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 7026998 0 24826
> cpu1 0 0 0 783505304 0 860700497 26129151 41776419 34749418 10682815341474 1753703225423 834571346
> domain0 03 8685595 8677710 224 7037051 8127 0 76 8677634 7500 7320 7 471762 179 0 8 7312 ===> 0 0 0 0 0 0 0 0 <=== 1 0 1 0 0 0 0 0 0 8607831 0 36753

> micah@micah-64:~$ cat /proc/schedstat
> version 14
> timestamp 4366723231
> cpu0 0 0 0 785868111 0 828021128 12621812 22703965 14096130 9509029340381 23603703795868 815399316
> domain0 03 8766051 8760123 95 14544377 6374 0 13 8760110 7374 6776 0 20835105 605 0 1 6775 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 7027010 0 24826
> cpu1 0 0 0 783505304 0 860708521 26132612 41780521 34753508 10683770853336 1753758309667 834575909
> domain0 03 8685718 8677833 224 7037051 8127 0 76 8677757 7500 7320 7 471762 179 0 8 7312 ===> 0 0 0 0 0 0 0 0 <=== 1 0 1 0 0 0 0 0 0 8607835 0 36753

You seem to have a configuration with domains which don't have
SD_BALANCE_NEWIDLE  on (CONFIG_NUMA?) as there are no events (all
zeros above) for CPU_NEWLY_IDLE.

this one is being triggered whenever a cpu becomes idle (schedule()
--> idle_balance() --> load_balance_newidle()).

(this flag is a bit #1 == 2)

cat /proc/sys/kernel/sched_domain/cpu0/domain0/flags

?

it can be changed with "echo new_value > ..."

does a change of it make any difference?

moreover, /proc/sys/kernel/sched_domain/cpu1/domain0/newidle_idx seems
to be responsible for a source of the load for calculating the busiest
group. e.g. with newidle_idx == 0, the current load on the queue is
used instead of cpu_load[].


>
> Thanks,
> --Micah
>

-- 
Best regards,
Dmitry Adamushko

  reply	other threads:[~2007-11-19 22:22 UTC|newest]

Thread overview: 35+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2007-11-09 22:34 Micah Dowty
2007-11-09 23:56 ` Cyrus Massoumi
2007-11-10  0:11   ` Micah Dowty
2007-11-14 18:39     ` Micah Dowty
2007-11-15 18:48     ` Kyle Moffett
2007-11-15 19:14       ` Micah Dowty
2007-11-15 20:07         ` Christoph Lameter
2007-11-15 20:24           ` Micah Dowty
2007-11-15 21:28             ` Christoph Lameter
2007-11-15 21:35               ` Micah Dowty
2007-11-16  2:31                 ` Christoph Lameter
2007-11-16  2:44                   ` Micah Dowty
2007-11-16  6:07                     ` Ingo Molnar
2007-11-16  9:19                       ` Micah Dowty
2007-11-16 10:45                         ` Ingo Molnar
2007-11-16 10:48                       ` Micah Dowty
2007-11-16 22:12                         ` Christoph Lameter
2007-11-16 10:48                       ` Dmitry Adamushko
2007-11-16 22:14                         ` Micah Dowty
2007-11-16 23:26                           ` Dmitry Adamushko
2007-11-17  1:03                             ` Micah Dowty
2007-11-17 19:10                               ` Dmitry Adamushko
2007-11-19 18:51                                 ` Micah Dowty
2007-11-19 22:22                                   ` Dmitry Adamushko [this message]
2007-11-19 23:05                                     ` Micah Dowty
2007-11-20  5:57                                       ` Ingo Molnar
2007-11-20 18:06                                         ` Micah Dowty
2007-11-20 21:47                                           ` Dmitry Adamushko
2007-11-22  7:46                                             ` Micah Dowty
2007-11-22 12:53                                               ` Dmitry Adamushko
2007-11-26 19:44                                                 ` Micah Dowty
2007-11-27  9:21                                                   ` Dmitry Adamushko
2007-11-27 17:13                                                     ` Micah Dowty
2007-11-16 19:13         ` David Newall
2007-11-16 21:38           ` Micah Dowty

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=b647ffbd0711191422q6be59c44gfbc72f0fc3167768@mail.gmail.com \
    --to=dmitry.adamushko@gmail.com \
    --cc=akpm@osdl.org \
    --cc=clameter@sgi.com \
    --cc=cyrusm@gmx.net \
    --cc=efault@gmx.de \
    --cc=linux-kernel@vger.kernel.org \
    --cc=menage@google.com \
    --cc=micah@vmware.com \
    --cc=mingo@elte.hu \
    --cc=mrmacman_g4@mac.com \
    --cc=pwil3058@bigpond.net.au \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®