[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20061019191623.5279bc77.kamezawa.hiroyu@jp.fujitsu.com>
Date: Thu, 19 Oct 2006 19:16:23 +0900
From: KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>
To: "Siddha, Suresh B" <suresh.b.siddha@...el.com>
Cc: clameter@....com, nickpiggin@...oo.com.au,
suresh.b.siddha@...el.com, mingo@...e.hu, pwil3058@...pond.net.au,
linux-kernel@...r.kernel.org,
"kaneshige.kenji@...t.fujitsu.com" <kaneshige.kenji@...t.fujitsu.com>
Subject: Re: [RFC] sched_tick with interrupts enabled
On Wed, 18 Oct 2006 19:19:00 -0700
"Siddha, Suresh B" <suresh.b.siddha@...el.com> wrote:
> On Wed, Oct 18, 2006 at 02:59:07PM -0700, Christoph Lameter wrote:
> > load_balancing has the potential of running for some time if f.e.
> > sched_domains for a system with 1024 processors have to be balanced.
> > We currently do all of that with interrupts disabled. So we may be unable
> > to service interrupts for some time. Most of that time is potentially
> > spend in rebalance_tick.
>
> Did you see an issue because of this or just theoretical?
>
IIRC, Fujitsu's 64-cpu ia64/SMP system sufferred from this issue *in older kernel*.
Now, we avoid it by creating NUMA nodes and dividing scheduler domains.
Situation was...:
5 runnable processes which were pinned to a cpu in a 64cpu system.
the system was always rebalanced and seems to be hanged.
-Kame
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists