From: Vincent Guittot <vincent.guittot@linaro.org>
To: Scott Wood <swood@redhat.com>
Cc: Ingo Molnar <mingo@redhat.com>,
Peter Zijlstra <peterz@infradead.org>,
Dietmar Eggemann <dietmar.eggemann@arm.com>,
Steven Rostedt <rostedt@goodmis.org>,
Mel Gorman <mgorman@suse.de>,
Valentin Schneider <valentin.schneider@arm.com>,
linux-kernel <linux-kernel@vger.kernel.org>,
linux-rt-users <linux-rt-users@vger.kernel.org>,
Sebastian Andrzej Siewior <bigeasy@linutronix.de>,
Thomas Gleixner <tglx@linutronix.de>
Subject: Re: [PATCH v2 1/3] sched/fair: Call newidle_balance() from balance_callback on PREEMPT_RT
Date: Wed, 5 May 2021 14:13:25 +0200 [thread overview]
Message-ID: <CAKfTPtBrGF72Zf37vPS6r4wy_jKagznBbiqKWpPyMEB_u4wo1g@mail.gmail.com> (raw)
In-Reply-To: <20210428232821.2506201-2-swood@redhat.com>
On Thu, 29 Apr 2021 at 01:28, Scott Wood <swood@redhat.com> wrote:
>
> This is required in order to be able to enable interrupts in the next
> patch. This is limited to PREEMPT_RT to avoid adding potentially
> measurable overhead to the non-RT case (requiring a double switch when
> pulling a task onto a newly idle cpu).
IIUC, only the newidle_balance is a problem and not the idle load
balance that runs softirq. In this case, why not skipping
newidle_balance entirely in case of preempt_rt and kick an idle load
balance instead as you switch to idle thread context anyway
>
> update_misfit_status() is factored out for the PREEMPT_RT case, to ensure
> that the misfit status is kept consistent before dropping the lock.
>
> Signed-off-by: Scott Wood <swood@redhat.com>
> ---
> v2: Use a balance callback, and limit to PREEMPT_RT
>
> kernel/sched/fair.c | 34 ++++++++++++++++++++++++++++++----
> 1 file changed, 30 insertions(+), 4 deletions(-)
>
> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> index 794c2cb945f8..ff369c38a5b5 100644
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -5660,6 +5660,9 @@ static void dequeue_task_fair(struct rq *rq, struct task_struct *p, int flags)
>
> #ifdef CONFIG_SMP
>
> +static const bool newidle_balance_in_callback = IS_ENABLED(CONFIG_PREEMPT_RT);
> +static DEFINE_PER_CPU(struct callback_head, rebalance_head);
> +
> /* Working cpumask for: load_balance, load_balance_newidle. */
> DEFINE_PER_CPU(cpumask_var_t, load_balance_mask);
> DEFINE_PER_CPU(cpumask_var_t, select_idle_mask);
> @@ -10549,7 +10552,7 @@ static inline void nohz_newidle_balance(struct rq *this_rq) { }
> * 0 - failed, no new tasks
> * > 0 - success, new (fair) tasks present
> */
> -static int newidle_balance(struct rq *this_rq, struct rq_flags *rf)
> +static int do_newidle_balance(struct rq *this_rq, struct rq_flags *rf)
> {
> unsigned long next_balance = jiffies + HZ;
> int this_cpu = this_rq->cpu;
> @@ -10557,7 +10560,9 @@ static int newidle_balance(struct rq *this_rq, struct rq_flags *rf)
> int pulled_task = 0;
> u64 curr_cost = 0;
>
> - update_misfit_status(NULL, this_rq);
> + if (!newidle_balance_in_callback)
> + update_misfit_status(NULL, this_rq);
> +
> /*
> * We must set idle_stamp _before_ calling idle_balance(), such that we
> * measure the duration of idle_balance() as idle time.
> @@ -10576,7 +10581,8 @@ static int newidle_balance(struct rq *this_rq, struct rq_flags *rf)
> * further scheduler activity on it and we're being very careful to
> * re-start the picking loop.
> */
> - rq_unpin_lock(this_rq, rf);
> + if (!newidle_balance_in_callback)
> + rq_unpin_lock(this_rq, rf);
>
> if (this_rq->avg_idle < sysctl_sched_migration_cost ||
> !READ_ONCE(this_rq->rd->overload)) {
> @@ -10655,11 +10661,31 @@ static int newidle_balance(struct rq *this_rq, struct rq_flags *rf)
> if (pulled_task)
> this_rq->idle_stamp = 0;
>
> - rq_repin_lock(this_rq, rf);
> + if (!newidle_balance_in_callback)
> + rq_repin_lock(this_rq, rf);
>
> return pulled_task;
> }
>
> +static void newidle_balance_cb(struct rq *this_rq)
> +{
> + update_rq_clock(this_rq);
> + do_newidle_balance(this_rq, NULL);
> +}
> +
> +static int newidle_balance(struct rq *this_rq, struct rq_flags *rf)
> +{
> + if (newidle_balance_in_callback) {
> + update_misfit_status(NULL, this_rq);
> + queue_balance_callback(this_rq,
> + &per_cpu(rebalance_head, this_rq->cpu),
> + newidle_balance_cb);
> + return 0;
> + }
> +
> + return do_newidle_balance(this_rq, rf);
> +}
> +
> /*
> * run_rebalance_domains is triggered when needed from the scheduler tick.
> * Also triggered for nohz idle balancing (with nohz_balancing_kick set).
> --
> 2.27.0
>
next prev parent reply other threads:[~2021-05-05 12:14 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-04-28 23:28 [PATCH v2 0/3] newidle_balance() PREEMPT_RT latency mitigations Scott Wood
2021-04-28 23:28 ` [PATCH v2 1/3] sched/fair: Call newidle_balance() from balance_callback on PREEMPT_RT Scott Wood
2021-05-05 12:13 ` Vincent Guittot [this message]
2021-05-07 15:19 ` Valentin Schneider
2021-04-28 23:28 ` [PATCH v2 2/3] sched/fair: Enable interrupts when dropping lock in newidle_balance() Scott Wood
2021-04-28 23:28 ` [PATCH v2 3/3] sched/fair: break out of newidle balancing if an RT task appears Scott Wood
2021-04-29 4:11 ` kernel test robot
2021-04-29 6:37 ` kernel test robot
2021-05-07 11:03 ` Peter Zijlstra
2021-05-15 7:29 ` Mike Galbraith
2021-05-15 8:36 ` Mike Galbraith
2021-04-29 7:12 ` [PATCH v2 0/3] newidle_balance() PREEMPT_RT latency mitigations Vincent Guittot
2021-05-01 22:03 ` Scott Wood
2021-05-02 3:25 ` Mike Galbraith
2021-05-03 16:33 ` Scott Wood
2021-05-03 18:52 ` Mike Galbraith
2021-05-03 21:57 ` Scott Wood
2021-05-04 4:07 ` Mike Galbraith
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CAKfTPtBrGF72Zf37vPS6r4wy_jKagznBbiqKWpPyMEB_u4wo1g@mail.gmail.com \
--to=vincent.guittot@linaro.org \
--cc=bigeasy@linutronix.de \
--cc=dietmar.eggemann@arm.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-rt-users@vger.kernel.org \
--cc=mgorman@suse.de \
--cc=mingo@redhat.com \
--cc=peterz@infradead.org \
--cc=rostedt@goodmis.org \
--cc=swood@redhat.com \
--cc=tglx@linutronix.de \
--cc=valentin.schneider@arm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).