[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20160809231955.GA3273@intel.com>
Date: Wed, 10 Aug 2016 07:19:55 +0800
From: Yuyang Du <yuyang.du@...el.com>
To: Peter Zijlstra <peterz@...radead.org>
Cc: Dietmar Eggemann <dietmar.eggemann@....com>,
Vincent Guittot <vincent.guittot@...aro.org>,
Ingo Molnar <mingo@...nel.org>,
linux-kernel <linux-kernel@...r.kernel.org>,
Mike Galbraith <umgwanakikbuti@...il.com>,
Benjamin Segall <bsegall@...gle.com>,
Paul Turner <pjt@...gle.com>,
Morten Rasmussen <morten.rasmussen@....com>,
Matt Fleming <matt@...eblueprint.co.uk>
Subject: Re: [PATCH 4/4] sched,fair: Fix PELT integrity for new tasks
On Fri, Jun 24, 2016 at 03:03:24PM +0200, Peter Zijlstra wrote:
> > Task's last_update_time means this task is detached from fair queue. This
> > (re)definition is by all means much better than migrating. No?
>
> I would maybe redefine it as an up-to-date marker for a migration across
> a clock discontinuity. Both CPU migration and group movement suffer from
> this, albeit for different reasons.
>
> In the CPU migration case we simply cannot tell time by our refusal to
> acquire the old rq lock. So we age to the last time we 'know' and then
> mark it up-to-date.
>
> For the cgroup move the timelines simply _are_ discontinuous. So we have
> to mark it up-to-date after we update it to the instant of detach, such
> that when we attach it to the new group we don't try to age it across
> the time difference.
Got it. I'm glad to see you worked out all these.
It's like I am replying to an email in the last century, sorry, but I was
distracted for many other things.
Powered by blists - more mailing lists