sched/fair: Trigger the update of blocked load on newly idle cpu

Instead of waking up a random and already idle CPU, we can take advantage
of this_cpu being about to enter idle to run the ILB and update the
blocked load.

Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Valentin Schneider <valentin.schneider@arm.com>
Link: https://lkml.kernel.org/r/20210224133007.28644-7-vincent.guittot@linaro.org

4 files changed, 35 insertions, 4 deletions

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index f9dfb349146e..361974efc243 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -737,7 +737,7 @@ static void nohz_csd_func(void *info)
 	/*
 	 * Release the rq::nohz_csd.
 	 */
-	flags = atomic_fetch_andnot(NOHZ_KICK_MASK, nohz_flags(cpu));
+	flags = atomic_fetch_andnot(NOHZ_KICK_MASK | NOHZ_NEWILB_KICK, nohz_flags(cpu));
 	WARN_ON(!(flags & NOHZ_KICK_MASK));
 
 	rq->idle_balance = idle_cpu(cpu);
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 356a2456c5e9..e87e1b3bcdca 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -10453,6 +10453,24 @@ static bool nohz_idle_balance(struct rq *this_rq, enum cpu_idle_type idle)
 	return true;
 }
 
+/*
+ * Check if we need to run the ILB for updating blocked load before entering
+ * idle state.
+ */
+void nohz_run_idle_balance(int cpu)
+{
+	unsigned int flags;
+
+	flags = atomic_fetch_andnot(NOHZ_NEWILB_KICK, nohz_flags(cpu));
+
+	/*
+	 * Update the blocked load only if no SCHED_SOFTIRQ is about to happen
+	 * (ie NOHZ_STATS_KICK set) and will do the same.
+	 */
+	if ((flags == NOHZ_NEWILB_KICK) && !need_resched())
+		_nohz_idle_balance(cpu_rq(cpu), NOHZ_STATS_KICK, CPU_IDLE);
+}
+
 static void nohz_newidle_balance(struct rq *this_rq)
 {
 	int this_cpu = this_rq->cpu;
@@ -10474,10 +10492,10 @@ static void nohz_newidle_balance(struct rq *this_rq)
 		return;
 
 	/*
-	 * Blocked load of idle CPUs need to be updated.
-	 * Kick an ILB to update statistics.
+	 * Set the need to trigger ILB in order to update blocked load
+	 * before entering idle state.
 	 */
-	kick_ilb(NOHZ_STATS_KICK);
+	atomic_or(NOHZ_NEWILB_KICK, nohz_flags(this_cpu));
 }
 
 #else /* !CONFIG_NO_HZ_COMMON */
diff --git a/kernel/sched/idle.c b/kernel/sched/idle.c
index 7199e6f23789..7a92d6054aba 100644
--- a/kernel/sched/idle.c
+++ b/kernel/sched/idle.c
@@ -261,6 +261,12 @@ exit_idle:
 static void do_idle(void)
 {
 	int cpu = smp_processor_id();
+
+	/*
+	 * Check if we need to update blocked load
+	 */
+	nohz_run_idle_balance(cpu);
+
 	/*
 	 * If the arch has a polling bit, we maintain an invariant:
 	 *
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index 10a1522b1e30..0ddc9a6ff03a 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -2385,9 +2385,11 @@ extern void cfs_bandwidth_usage_dec(void);
 #ifdef CONFIG_NO_HZ_COMMON
 #define NOHZ_BALANCE_KICK_BIT	0
 #define NOHZ_STATS_KICK_BIT	1
+#define NOHZ_NEWILB_KICK_BIT	2
 
 #define NOHZ_BALANCE_KICK	BIT(NOHZ_BALANCE_KICK_BIT)
 #define NOHZ_STATS_KICK		BIT(NOHZ_STATS_KICK_BIT)
+#define NOHZ_NEWILB_KICK	BIT(NOHZ_NEWILB_KICK_BIT)
 
 #define NOHZ_KICK_MASK	(NOHZ_BALANCE_KICK | NOHZ_STATS_KICK)
 
@@ -2398,6 +2400,11 @@ extern void nohz_balance_exit_idle(struct rq *rq);
 static inline void nohz_balance_exit_idle(struct rq *rq) { }
 #endif
 
+#if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
+extern void nohz_run_idle_balance(int cpu);
+#else
+static inline void nohz_run_idle_balance(int cpu) { }
+#endif
 
 #ifdef CONFIG_SMP
 static inline