Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1281761 > unrolled thread

Re: [PATCH 4/7] sched: Account rr and fifo tasks separately

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [PATCH 4/7] sched: Account rr and fifo tasks separately Peter Zijlstra <peterz@infradead.org> - 2015-12-02 14:00 +0100
    Re: [PATCH 4/7] sched: Account rr and fifo tasks separately Frederic Weisbecker <fweisbec@gmail.com> - 2015-12-02 15:20 +0100

#1281761 — Re: [PATCH 4/7] sched: Account rr and fifo tasks separately

On Fri, Nov 13, 2015 at 03:22:06PM +0100, Frederic Weisbecker wrote:
> In order to evaluate tick dependency, we need to account SCHED_RR and
> SCHED_FIFO tasks separately as those policies don't have the same
> preemption requirements.
> 
> We still keep rt_nr_running as a cache to avoid additions between nr_rr
> and nr_fifo all over the place.

In which case you only need one of nr_fifo/nr_rr. Less accounting is
better.

Pick the one you need for the nohz_full condition, and leave the other.
A quick look at sched_can_stop_tick() seems to suggest nr_rr is the
interesting one. nr_rr < 2 should allow stopping the tick.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [next] | [standalone]

#1281851

On Wed, Dec 02, 2015 at 01:53:39PM +0100, Peter Zijlstra wrote:
> On Fri, Nov 13, 2015 at 03:22:06PM +0100, Frederic Weisbecker wrote:
> > In order to evaluate tick dependency, we need to account SCHED_RR and
> > SCHED_FIFO tasks separately as those policies don't have the same
> > preemption requirements.
> > 
> > We still keep rt_nr_running as a cache to avoid additions between nr_rr
> > and nr_fifo all over the place.
> 
> In which case you only need one of nr_fifo/nr_rr. Less accounting is
> better.
> 
> Pick the one you need for the nohz_full condition, and leave the other.
> A quick look at sched_can_stop_tick() seems to suggest nr_rr is the
> interesting one. nr_rr < 2 should allow stopping the tick.

Sounds pretty good! I'm going to do that!

Thanks!
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web