On Fri, Jun 13, 2014 at 08:55:48AM -0700, Paul E. McKenney wrote:
> On Fri, Jun 13, 2014 at 02:55:45PM +0200, Frederic Weisbecker wrote:
> > On Thu, Jun 12, 2014 at 07:05:54PM -0700, Paul E. McKenney wrote:
> > > On Thu, Jun 12, 2014 at 06:24:32PM -0700, Paul E. McKenney wrote:
> > > > On Fri, Jun 13, 2014 at 02:16:59AM +0200, Frederic Weisbecker wrote:
> > > > > CONFIG_NO_HZ_FULL may be enabled widely on distros nowadays but actual
> > > > > users should be a tiny minority, if actually any.
> > > > >
> > > > > Also there is a risk that affining the GP kthread to a single CPU
> > > > > could
> > > > > end up noticeably reducing RCU performances and increasing energy
> > > > > consumption.
> > > > >
> > > > > So lets affine the GP kthread only when nohz full is actually used
> > > > > (ie: when the nohz_full= parameter is filled or
> > > > > CONFIG_NO_HZ_FULL_ALL=y)
> > > > >
> > > > > Signed-off-by: Frederic Weisbecker <fweis...@gmail.com>
> > > > > Cc: Josh Triplett <j...@joshtriplett.org>
> > > > > Cc: Steven Rostedt <rost...@goodmis.org>
> > > > > Cc: Mathieu Desnoyers <mathieu.desnoy...@efficios.com>
> > > > > ---
> > > > > kernel/rcu/tree_plugin.h | 10 +++++++---
> > > > > 1 file changed, 7 insertions(+), 3 deletions(-)
> > > > >
> > > > > diff --git a/kernel/rcu/tree_plugin.h b/kernel/rcu/tree_plugin.h
> > > > > index cbc2c45..726f52c 100644
> > > > > --- a/kernel/rcu/tree_plugin.h
> > > > > +++ b/kernel/rcu/tree_plugin.h
> > > > > @@ -2843,12 +2843,16 @@ static bool rcu_nohz_full_cpu(struct
> > > > > rcu_state *rsp)
> > > > > */
> > > > > static void rcu_bind_gp_kthread(void)
> > > > > {
> > > > > -#ifdef CONFIG_NO_HZ_FULL
> > > > > - int cpu = ACCESS_ONCE(tick_do_timer_cpu);
> > > > > + int cpu;
> > > > > +
> > > > > + if (!tick_nohz_full_enabled())
> > > > > + return;
> > > > > +
> > > > > + cpu = ACCESS_ONCE(tick_do_timer_cpu);
> > > > >
> > > > > if (cpu < 0 || cpu >= nr_cpu_ids)
> > > > > return;
> > > > > +
> > > > > if (raw_smp_processor_id() != cpu)
> > > > > set_cpus_allowed_ptr(current, cpumask_of(cpu));
> > > > > -#endif /* #ifdef CONFIG_NO_HZ_FULL */
> > > > > }
> > > >
> > > > Hello, Frederic,
> > > >
> > > > I have the following queued. Shall I port yours on top of mine, or is
> > > > there an issue with mine?
> > >
> > > OK, I suppose I should show you what it looks like ported on top of mine.
> >
> > No need to keep my patch as long as yours goes in. It fixes all the issue.
> >
> > >
> > > I didn't understand the removal of "#ifdef CONFIG_NO_HZ_FULL", so I
> > > left that.
> >
> > Yeah that's because the:
> >
> > if (!tick_nohz_full_enabled())
> > return;
> >
> > returns unconditionally if CONFIG_NO_HZ_FULL=n. So the whole function
> > becomes dead code that should be detected and removed by gcc. So same
> > result as with the ifdef.
>
> Well, that part of your patch is worthwhile, then! And I do therefore
> need to restructure to invoke tick_nohz_full_enabled() at the very
> beginning of the function.
Well if no CPU is full dynticks, the sole effect of this is that the kthread
will be re-affined globally... So no big issue.
Although that can make you spare a call to set_cpus_allowed_ptr(), which can
be worthwile after all :)
>
> > > I also didn't understand how dumping the GP kthreads onto
> > > a single CPU could affect energy efficiency all that much, so I omitted
> > > that from the commit log. If I am missing something, please enlighten me.
> >
> > Because if the kthread is pinned to CPU 0 and a grace period is in progress,
> > the kthread is going to disturb the CPU 0 regardless of its possibly deep
> > sleep
> > state. OTOH if the kthread is widely affine, it can be scheduled to idle
> > CPUs
> > that are less cold and thus wake them from less deep power state.
> >
> > Well that all assuming that the scheduler takes care of these deep idle
> > state.
> > But I believe it does iirc...
>
> My guess is that enough stuff gets dumped on CPU 0 to make this a non-issue,
> but point taken nonetheless.
Ok.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
