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
next prev parent 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®