Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   Scott Wood <swood@redhat.com>
To:     Ingo Molnar <mingo@redhat.com>,
        Peter Zijlstra <peterz@infradead.org>,
        Vincent Guittot <vincent.guittot@linaro.org>
Cc:     Dietmar Eggemann <dietmar.eggemann@arm.com>,
        Steven Rostedt <rostedt@goodmis.org>,
        Mel Gorman <mgorman@suse.de>,
        Valentin Schneider <valentin.schneider@arm.com>,
        linux-kernel@vger.kernel.org, linux-rt-users@vger.kernel.org,
        Sebastian Andrzej Siewior <bigeasy@linutronix.de>,
        Thomas Gleixner <tglx@linutronix.de>,
        Scott Wood <swood@redhat.com>
Subject: [PATCH v2 2/3] sched/fair: Enable interrupts when dropping lock in newidle_balance()
Date:   Wed, 28 Apr 2021 18:28:20 -0500
Message-Id: <20210428232821.2506201-3-swood@redhat.com>
In-Reply-To: <20210428232821.2506201-1-swood@redhat.com>
References: <20210428232821.2506201-1-swood@redhat.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

When combined with the next patch, which breaks out of rebalancing
when an RT task is runnable, significant latency reductions are seen
on systems with many CPUs.

Signed-off-by: Scott Wood <swood@redhat.com>
---
 kernel/sched/fair.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index ff369c38a5b5..aa8c87b6aff8 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -10521,6 +10521,8 @@ static void nohz_newidle_balance(struct rq *this_rq)
 		return;
 
 	raw_spin_unlock(&this_rq->lock);
+	if (newidle_balance_in_callback)
+		local_irq_enable();
 	/*
 	 * This CPU is going to be idle and blocked load of idle CPUs
 	 * need to be updated. Run the ilb locally as it is a good
@@ -10529,6 +10531,8 @@ static void nohz_newidle_balance(struct rq *this_rq)
 	 */
 	if (!_nohz_idle_balance(this_rq, NOHZ_STATS_KICK, CPU_NEWLY_IDLE))
 		kick_ilb(NOHZ_STATS_KICK);
+	if (newidle_balance_in_callback)
+		local_irq_disable();
 	raw_spin_lock(&this_rq->lock);
 }
 
@@ -10599,6 +10603,8 @@ static int do_newidle_balance(struct rq *this_rq, struct rq_flags *rf)
 	}
 
 	raw_spin_unlock(&this_rq->lock);
+	if (newidle_balance_in_callback)
+		local_irq_enable();
 
 	update_blocked_averages(this_cpu);
 	rcu_read_lock();
@@ -10636,6 +10642,8 @@ static int do_newidle_balance(struct rq *this_rq, struct rq_flags *rf)
 	}
 	rcu_read_unlock();
 
+	if (newidle_balance_in_callback)
+		local_irq_disable();
 	raw_spin_lock(&this_rq->lock);
 
 	if (curr_cost > this_rq->max_idle_balance_cost)
-- 
2.27.0