[PATCH v10 12/15] sched: Add over-utilization/tipping point indicator

From: Quentin Perret <quentin.perret@arm.com>
To: peterz@infradead.org, rjw@rjwysocki.net,
	linux-kernel@vger.kernel.org, linux-pm@vger.kernel.org
Cc: gregkh@linuxfoundation.org, mingo@redhat.com,
	dietmar.eggemann@arm.com, morten.rasmussen@arm.com,
	chris.redpath@arm.com, patrick.bellasi@arm.com,
	valentin.schneider@arm.com, vincent.guittot@linaro.org,
	thara.gopinath@linaro.org, viresh.kumar@linaro.org,
	tkjos@google.com, joel@joelfernandes.org, smuckle@google.com,
	adharmap@codeaurora.org, skannan@codeaurora.org,
	pkondeti@codeaurora.org, juri.lelli@redhat.com,
	edubezval@gmail.com, srinivas.pandruvada@linux.intel.com,
	currojerez@riseup.net, javi.merino@kernel.org,
	quentin.perret@arm.com
Subject: [PATCH v10 12/15] sched: Add over-utilization/tipping point indicator
Date: Mon,  3 Dec 2018 09:56:25 +0000	[thread overview]
Message-ID: <20181203095628.11858-13-quentin.perret@arm.com> (raw)
In-Reply-To: <20181203095628.11858-1-quentin.perret@arm.com>

From: Morten Rasmussen <morten.rasmussen@arm.com>

Energy-aware scheduling is only meant to be active while the system is
_not_ over-utilized. That is, there are spare cycles available to shift
tasks around based on their actual utilization to get a more
energy-efficient task distribution without depriving any tasks. When
above the tipping point task placement is done the traditional way based
on load_avg, spreading the tasks across as many cpus as possible based
on priority scaled load to preserve smp_nice. Below the tipping point we
want to use util_avg instead. We need to define a criteria for when we
make the switch.

The util_avg for each cpu converges towards 100% regardless of how many
additional tasks we may put on it. If we define over-utilized as:

sum_{cpus}(rq.cfs.avg.util_avg) + margin > sum_{cpus}(rq.capacity)

some individual cpus may be over-utilized running multiple tasks even
when the above condition is false. That should be okay as long as we try
to spread the tasks out to avoid per-cpu over-utilization as much as
possible and if all tasks have the _same_ priority. If the latter isn't
true, we have to consider priority to preserve smp_nice.

For example, we could have n_cpus nice=-10 util_avg=55% tasks and
n_cpus/2 nice=0 util_avg=60% tasks. Balancing based on util_avg we are
likely to end up with nice=-10 tasks sharing cpus and nice=0 tasks
getting their own as we 1.5*n_cpus tasks in total and 55%+55% is less
over-utilized than 55%+60% for those cpus that have to be shared. The
system utilization is only 85% of the system capacity, but we are
breaking smp_nice.

To be sure not to break smp_nice, we have defined over-utilization
conservatively as when any cpu in the system is fully utilized at its
highest frequency instead:

cpu_rq(any).cfs.avg.util_avg + margin > cpu_rq(any).capacity

IOW, as soon as one cpu is (nearly) 100% utilized, we switch to load_avg
to factor in priority to preserve smp_nice.

With this definition, we can skip periodic load-balance as no cpu has an
always-running task when the system is not over-utilized. All tasks will
be periodic and we can balance them at wake-up. This conservative
condition does however mean that some scenarios that could benefit from
energy-aware decisions even if one cpu is fully utilized would not get
those benefits.

For systems where some cpus might have reduced capacity on some cpus
(RT-pressure and/or big.LITTLE), we want periodic load-balance checks as
soon a just a single cpu is fully utilized as it might one of those with
reduced capacity and in that case we want to migrate it.

cc: Ingo Molnar <mingo@redhat.com>
cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Morten Rasmussen <morten.rasmussen@arm.com>
[ Added a comment explaining why new tasks are not accounted during
  overutilization detection ]
Signed-off-by: Quentin Perret <quentin.perret@arm.com>
---
 kernel/sched/fair.c  | 59 ++++++++++++++++++++++++++++++++++++++++++--
 kernel/sched/sched.h |  4 +++
 2 files changed, 61 insertions(+), 2 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index e21f37129395..c3b2dad72c9c 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -5082,6 +5082,24 @@ static inline void hrtick_update(struct rq *rq)
 }
 #endif
 
+#ifdef CONFIG_SMP
+static inline unsigned long cpu_util(int cpu);
+static unsigned long capacity_of(int cpu);
+
+static inline bool cpu_overutilized(int cpu)
+{
+	return (capacity_of(cpu) * 1024) < (cpu_util(cpu) * capacity_margin);
+}
+
+static inline void update_overutilized_status(struct rq *rq)
+{
+	if (!READ_ONCE(rq->rd->overutilized) && cpu_overutilized(rq->cpu))
+		WRITE_ONCE(rq->rd->overutilized, SG_OVERUTILIZED);
+}
+#else
+static inline void update_overutilized_status(struct rq *rq) { }
+#endif
+
 /*
  * The enqueue_task method is called before nr_running is
  * increased. Here we update the fair scheduling stats and
@@ -5139,8 +5157,26 @@ enqueue_task_fair(struct rq *rq, struct task_struct *p, int flags)
 		update_cfs_group(se);
 	}
 
-	if (!se)
+	if (!se) {
 		add_nr_running(rq, 1);
+		/*
+		 * Since new tasks are assigned an initial util_avg equal to
+		 * half of the spare capacity of their CPU, tiny tasks have the
+		 * ability to cross the overutilized threshold, which will
+		 * result in the load balancer ruining all the task placement
+		 * done by EAS. As a way to mitigate that effect, do not account
+		 * for the first enqueue operation of new tasks during the
+		 * overutilized flag detection.
+		 *
+		 * A better way of solving this problem would be to wait for
+		 * the PELT signals of tasks to converge before taking them
+		 * into account, but that is not straightforward to implement,
+		 * and the following generally works well enough in practice.
+		 */
+		if (flags & ENQUEUE_WAKEUP)
+			update_overutilized_status(rq);
+
+	}
 
 	hrtick_update(rq);
 }
@@ -7940,6 +7976,9 @@ static inline void update_sg_lb_stats(struct lb_env *env,
 		if (nr_running > 1)
 			*sg_status |= SG_OVERLOAD;
 
+		if (cpu_overutilized(i))
+			*sg_status |= SG_OVERUTILIZED;
+
 #ifdef CONFIG_NUMA_BALANCING
 		sgs->nr_numa_running += rq->nr_numa_running;
 		sgs->nr_preferred_running += rq->nr_preferred_running;
@@ -8170,8 +8209,15 @@ static inline void update_sd_lb_stats(struct lb_env *env, struct sd_lb_stats *sd
 		env->fbq_type = fbq_classify_group(&sds->busiest_stat);
 
 	if (!env->sd->parent) {
+		struct root_domain *rd = env->dst_rq->rd;
+
 		/* update overload indicator if we are at root domain */
-		WRITE_ONCE(env->dst_rq->rd->overload, sg_status & SG_OVERLOAD);
+		WRITE_ONCE(rd->overload, sg_status & SG_OVERLOAD);
+
+		/* Update over-utilization (tipping point, U >= 0) indicator */
+		WRITE_ONCE(rd->overutilized, sg_status & SG_OVERUTILIZED);
+	} else if (sg_status & SG_OVERUTILIZED) {
+		WRITE_ONCE(env->dst_rq->rd->overutilized, SG_OVERUTILIZED);
 	}
 }
 
@@ -8398,6 +8444,14 @@ static struct sched_group *find_busiest_group(struct lb_env *env)
 	 * this level.
 	 */
 	update_sd_lb_stats(env, &sds);
+
+	if (static_branch_unlikely(&sched_energy_present)) {
+		struct root_domain *rd = env->dst_rq->rd;
+
+		if (rcu_dereference(rd->pd) && !READ_ONCE(rd->overutilized))
+			goto out_balanced;
+	}
+
 	local = &sds.local_stat;
 	busiest = &sds.busiest_stat;
 
@@ -9798,6 +9852,7 @@ static void task_tick_fair(struct rq *rq, struct task_struct *curr, int queued)
 		task_tick_numa(rq, curr);
 
 	update_misfit_status(curr, rq);
+	update_overutilized_status(task_rq(curr));
 }
 
 /*
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index 4c1e4b73f40d..28d9209554dc 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -717,6 +717,7 @@ struct perf_domain {
 
 /* Scheduling group status flags */
 #define SG_OVERLOAD		0x1 /* More than one runnable task on a CPU. */
+#define SG_OVERUTILIZED		0x2 /* One or more CPUs are over-utilized. */
 
 /*
  * We add the notion of a root-domain which will be used to define per-domain
@@ -740,6 +741,9 @@ struct root_domain {
 	 */
 	int			overload;
 
+	/* Indicate one or more cpus over-utilized (tipping point) */
+	int			overutilized;
+
 	/*
 	 * The bit corresponding to a CPU gets set here if such CPU has more
 	 * than one runnable -deadline task (as it is below for RT tasks).
-- 
2.19.2

