CHROMIUM: sched: update the average of nr_running

Doing a Exponential moving average per nr_running++/-- does not
guarantee a fixed sample rate which induces errors if there are lots of
threads being enqueued/dequeued from the rq (Linpack mt). Instead of
keeping track of the avg, the scheduler now keeps track of the integral
of nr_running and allows the readers to perform filtering on top.

Original-author: Sai Charan Gurrappadi <sgurrappadi@nvidia.com>

Change-Id: Id946654f32fa8be0eaf9d8fa7c9a8039b5ef9fab
Signed-off-by: Joseph Lo <josephl@nvidia.com>
Signed-off-by: Andrew Bresticker <abrestic@chromium.org>
Reviewed-on: https://chromium-review.googlesource.com/174694
Reviewed-on: https://chromium-review.googlesource.com/272853
[jstultz: fwdported to 4.4]
Signed-off-by: John Stultz <john.stultz@linaro.org>

2 files changed, 77 insertions, 2 deletions

diff --git a/kernel/sched/core.c b/kernel/sched/core.c
index 9e25e68747d7..42c45ccc1f62 100644
--- a/kernel/sched/core.c
+++ b/kernel/sched/core.c
@@ -2768,6 +2768,36 @@ unsigned long nr_iowait_cpu(int cpu)
 	return atomic_read(&this->nr_iowait);
 }
 
+#ifdef CONFIG_CPU_QUIET
+u64 nr_running_integral(unsigned int cpu)
+{
+	unsigned int seqcnt;
+	u64 integral;
+	struct rq *q;
+
+	if (cpu >= nr_cpu_ids)
+		return 0;
+
+	q = cpu_rq(cpu);
+
+	/*
+	 * Update average to avoid reading stalled value if there were
+	 * no run-queue changes for a long time. On the other hand if
+	 * the changes are happening right now, just read current value
+	 * directly.
+	 */
+
+	seqcnt = read_seqcount_begin(&q->ave_seqcnt);
+	integral = do_nr_running_integral(q);
+	if (read_seqcount_retry(&q->ave_seqcnt, seqcnt)) {
+		read_seqcount_begin(&q->ave_seqcnt);
+		integral = q->nr_running_integral;
+	}
+
+	return integral;
+}
+#endif
+
 void get_iowait_load(unsigned long *nr_waiters, unsigned long *load)
 {
 	struct rq *rq = this_rq();
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index db3f3db9849c..5713513adb9d 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -594,6 +594,14 @@ struct rq {
 #ifdef CONFIG_NO_HZ_FULL
 	unsigned long last_sched_tick;
 #endif
+
+#ifdef CONFIG_CPU_QUIET
+	/* time-based average load */
+	u64 nr_last_stamp;
+	u64 nr_running_integral;
+	seqcount_t ave_seqcnt;
+#endif
+
 	/* capture load from *all* tasks on this cpu: */
 	struct load_weight load;
 	unsigned long nr_load_updates;
@@ -1353,7 +1361,7 @@ extern void init_entity_runnable_average(struct sched_entity *se);
 
 extern void init_max_cpu_capacity(struct max_cpu_capacity *mcc);
 
-static inline void add_nr_running(struct rq *rq, unsigned count)
+static inline void __add_nr_running(struct rq *rq, unsigned count)
 {
 	unsigned prev_nr = rq->nr_running;
 
@@ -1381,11 +1389,48 @@ static inline void add_nr_running(struct rq *rq, unsigned count)
 	}
 }
 
-static inline void sub_nr_running(struct rq *rq, unsigned count)
+static inline void __sub_nr_running(struct rq *rq, unsigned count)
 {
 	rq->nr_running -= count;
 }
 
+#ifdef CONFIG_CPU_QUIET
+#define NR_AVE_SCALE(x)		((x) << FSHIFT)
+static inline u64 do_nr_running_integral(struct rq *rq)
+{
+	s64 nr, deltax;
+	u64 nr_running_integral = rq->nr_running_integral;
+
+	deltax = rq->clock_task - rq->nr_last_stamp;
+	nr = NR_AVE_SCALE(rq->nr_running);
+
+	nr_running_integral += nr * deltax;
+
+	return nr_running_integral;
+}
+
+static inline void add_nr_running(struct rq *rq, unsigned count)
+{
+	write_seqcount_begin(&rq->ave_seqcnt);
+	rq->nr_running_integral = do_nr_running_integral(rq);
+	rq->nr_last_stamp = rq->clock_task;
+	__add_nr_running(rq, count);
+	write_seqcount_end(&rq->ave_seqcnt);
+}
+
+static inline void sub_nr_running(struct rq *rq, unsigned count)
+{
+	write_seqcount_begin(&rq->ave_seqcnt);
+	rq->nr_running_integral = do_nr_running_integral(rq);
+	rq->nr_last_stamp = rq->clock_task;
+	__sub_nr_running(rq, count);
+	write_seqcount_end(&rq->ave_seqcnt);
+}
+#else
+#define add_nr_running __add_nr_running
+#define sub_nr_running __sub_nr_running
+#endif
+
 static inline void rq_last_tick_reset(struct rq *rq)
 {
 #ifdef CONFIG_NO_HZ_FULL