lkml.org 
[lkml]   [2012]   [Sep]   [13]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: [RFC] sched: nohz_idle_balance
From
Date
On Thu, 2012-09-13 at 10:19 +0200, Peter Zijlstra wrote: 
> On Thu, 2012-09-13 at 08:49 +0200, Mike Galbraith wrote:
> > On Thu, 2012-09-13 at 06:11 +0200, Vincent Guittot wrote:
> > > On tickless system, one CPU runs load balance for all idle CPUs.
> > > The cpu_load of this CPU is updated before starting the load balance
> > > of each other idle CPUs. We should instead update the cpu_load of the balance_cpu.
> > >
> > > Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org>
> > > ---
> > > kernel/sched/fair.c | 11 ++++++-----
> > > 1 file changed, 6 insertions(+), 5 deletions(-)
> > >
> > > diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> > > index 1ca4fe4..9ae3a5b 100644
> > > --- a/kernel/sched/fair.c
> > > +++ b/kernel/sched/fair.c
> > > @@ -4794,14 +4794,15 @@ static void nohz_idle_balance(int this_cpu, enum cpu_idle_type idle)
> > > if (need_resched())
> > > break;
> > >
> > > - raw_spin_lock_irq(&this_rq->lock);
> > > - update_rq_clock(this_rq);
> > > - update_idle_cpu_load(this_rq);
> > > - raw_spin_unlock_irq(&this_rq->lock);
> > > + rq = cpu_rq(balance_cpu);
> > > +
> > > + raw_spin_lock_irq(&rq->lock);
> > > + update_rq_clock(rq);
> > > + update_idle_cpu_load(rq);
> > > + raw_spin_unlock_irq(&rq->lock);
> > >
> > > rebalance_domains(balance_cpu, CPU_IDLE);
> > >
> > > - rq = cpu_rq(balance_cpu);
> > > if (time_after(this_rq->next_balance, rq->next_balance))
> > > this_rq->next_balance = rq->next_balance;
> > > }
> >
> > Ew, banging locks and updating clocks to what good end?
>
> Well, updating the load statistics on the cpu you're going to balance
> seems like a good end to me.. ;-) No point updating the local statistics
> N times and leaving the ones you're going to balance stale for god knows
> how long.

Sure, the goal is fine, but I wonder about the price vs payoff. I was
thinking perhaps the redundant updates should go away instead, unless
stats are shown to be causing real world pain.

-Mike



\
 
 \ /
  Last update: 2012-09-13 13:01    [W:0.067 / U:0.332 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site