Re: [PATCH] sched: Remove WAKEUP_PREEMPT feature check in entity_tick

From: Yong Zhang <yong.zhang0@gmail.com>
To: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Lin Ming <ming.m.lin@intel.com>, "mingo@elte.hu" <mingo@elte.hu>,
	lkml <linux-kernel@vger.kernel.org>,
	Mike Galbraith <efault@gmx.de>
Subject: Re: [PATCH] sched: Remove WAKEUP_PREEMPT feature check in entity_tick
Date: Fri, 29 Jul 2011 16:46:37 +0800	[thread overview]
Message-ID: <20110729084637.GC12106@zhy> (raw)
In-Reply-To: <1311927650.5890.217.camel@twins>

On Fri, Jul 29, 2011 at 10:20:50AM +0200, Peter Zijlstra wrote:
> On Fri, 2011-07-29 at 16:18 +0800, Yong Zhang wrote:
> > > Nah, if there is 1 runnable task it will always run, preemption simply
> > > doesn't matter. There's nothing to preempt it with.
> > 
> > Hmmm, so the newly waked task could be scheduled a little later.
> > That means schedule tick judge everything. 
> 
> Oh, are you referring to the case where a task gets woken on an idle
> remote cpu?

Not really.

Let's take UP for example, we have cpu-hug task A and threadirq B.

	n tick		--->		n+1 tick
set_tsk_need_resched(A);
		B comes in and
		wake up thread-B;

So for system on which we disable WAKEUP_PREEMPT,
if we don't have that check, thread-B will wait until n+1 tick comes
to get to run.
But if we have that check, thread-B will get to run after IRQ-B returns.

Thanks,
Yong

-- 
Only stand for myself