Re: [PATCH 0/6] Optimize the cpu hotplug locking -v2

[Posted October 15, 2013 by corbet]

From:		Ingo Molnar <mingo-AT-kernel.org>
To:		Andrew Morton <akpm-AT-linux-foundation.org>
Subject:		Re: [PATCH 0/6] Optimize the cpu hotplug locking -v2
Date:		Thu, 10 Oct 2013 18:52:29 +0200
Message-ID:		<20131010165229.GC12998@gmail.com>
Cc:		Oleg Nesterov <oleg-AT-redhat.com>, "Srivatsa S. Bhat" <srivatsa.bhat-AT-linux.vnet.ibm.com>, Peter Zijlstra <peterz-AT-infradead.org>, Paul McKenney <paulmck-AT-linux.vnet.ibm.com>, Mel Gorman <mgorman-AT-suse.de>, Rik van Riel <riel-AT-redhat.com>, Srikar Dronamraju <srikar-AT-linux.vnet.ibm.com>, Andrea Arcangeli <aarcange-AT-redhat.com>, Johannes Weiner <hannes-AT-cmpxchg.org>, Thomas Gleixner <tglx-AT-linutronix.de>, Steven Rostedt <rostedt-AT-goodmis.org>, Linus Torvalds <torvalds-AT-linux-foundation.org>, linux-kernel-AT-vger.kernel.org
Archive‑link:		Article


* Andrew Morton <akpm@linux-foundation.org> wrote:

> On Thu, 10 Oct 2013 17:26:12 +0200 Oleg Nesterov <oleg@redhat.com> wrote:
> 
> > On 10/10, Ingo Molnar wrote:
> > >
> > > * Peter Zijlstra <peterz@infradead.org> wrote:
> > >
> > > > But the thing is; our sense of NR_CPUS has shifted, where it used to be
> > > > ok to do something like:
> > > >
> > > >   for_each_cpu()
> > > >
> > > > With preemption disabled; it gets to be less and less sane to do 
> > > > so, simply because 'common' hardware has 256+ CPUs these days. If 
> > > > we cannot rely on preempt disable to exclude hotplug, we must use 
> > > > get_online_cpus(), but get_online_cpus() is global state and thus 
> > > > cannot be used at any sort of frequency.
> > >
> > > So ... why not make it _really_ cheap, i.e. the read lock costing 
> > > nothing, and tie CPU hotplug to freezing all tasks in the system?
> > >
> > > Actual CPU hot unplugging and repluggin is _ridiculously_ rare in a 
> > > system, I don't understand how we tolerate _any_ overhead from this 
> > > utter slowpath.
> > 
> > Well, iirc Srivatsa (cc'ed) pointed out that some systems do 
> > cpu_down/up quite often to save the power.
> 
> cpu hotremove already uses stop_machine, so such an approach shouldn't 
> actually worsen things (a lot) for them?

Also, using CPU hotremove to save power, instead of implementing proper 
power aware scheduling, is very broken to begin with.

Thanks,

	Ingo