Subject: [DEBUG PATCH] sched/fair: profile distribute_cfs_runtime()

---
 kernel/sched/fair.c  | 17 +++++++++++++++++
 kernel/sched/sched.h |  2 ++
 2 files changed, 19 insertions(+)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index da3f728b27725..e3546274a162d 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -5009,6 +5009,8 @@ static int tg_unthrottle_up(struct task_group *tg, void *data)
 	cfs_rq->throttled_clock_pelt_time += rq_clock_pelt(rq) -
 					cfs_rq->throttled_clock_pelt;
 
+	rq->unthrottled_cfs_rq++;
+
 	/* Re-enqueue the tasks that have been throttled at this level. */
 	list_for_each_entry_safe(p, tmp, &cfs_rq->throttled_limbo_list, throttle_node) {
 		list_del_init(&p->throttle_node);
@@ -5017,6 +5019,7 @@ static int tg_unthrottle_up(struct task_group *tg, void *data)
 		 * due to affinity change while p is throttled.
 		 */
 		enqueue_task_fair(rq_of(cfs_rq), p, ENQUEUE_WAKEUP);
+		rq->unthrottled_task++;
 	}
 
 	/* Add cfs_rq with load or one or more already running entities to the list */
@@ -5193,7 +5196,9 @@ static void distribute_cfs_runtime(struct cfs_bandwidth *cfs_b)
 {
 	struct cfs_rq *cfs_rq;
 	u64 runtime, remaining = 1;
+	unsigned int unthrottled_rqs = 0, unthrottled_cfs_rq = 0, unthrottled_task = 0;
 
+	trace_printk("cpu%d: begins\n", raw_smp_processor_id());
 	rcu_read_lock();
 	list_for_each_entry_rcu(cfs_rq, &cfs_b->throttled_cfs_rq,
 				throttled_list) {
@@ -5201,6 +5206,7 @@ static void distribute_cfs_runtime(struct cfs_bandwidth *cfs_b)
 		struct rq_flags rf;
 
 		rq_lock_irqsave(rq, &rf);
+		rq->unthrottled_cfs_rq = rq->unthrottled_task = 0;
 		if (!cfs_rq_throttled(cfs_rq))
 			goto next;
 
@@ -5222,12 +5228,23 @@ static void distribute_cfs_runtime(struct cfs_bandwidth *cfs_b)
 			unthrottle_cfs_rq(cfs_rq);
 
 next:
+		trace_printk("cpu%d: cpu%d unthrottled_cfs_rq=%d/%d, unthrottled_task=%d/%d, remaining=%Lu\n",
+				raw_smp_processor_id(), cpu_of(rq),
+				rq->unthrottled_cfs_rq, unthrottled_cfs_rq,
+				rq->unthrottled_task, unthrottled_task, remaining);
+
+		unthrottled_cfs_rq += rq->unthrottled_cfs_rq;
+		unthrottled_task += rq->unthrottled_task;
+		unthrottled_rqs++;
+		rq->unthrottled_cfs_rq = rq->unthrottled_task = 0;
 		rq_unlock_irqrestore(rq, &rf);
 
 		if (!remaining)
 			break;
 	}
 	rcu_read_unlock();
+	trace_printk("cpu%d: finishes: unthrottled_rqs=%u, unthrottled_cfs_rq=%u, unthrottled_task=%u\n",
+			raw_smp_processor_id(), unthrottled_rqs, unthrottled_cfs_rq, unthrottled_task);
 }
 
 /*
diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h
index e0e05847855f0..bd3a11582d5b6 100644
--- a/kernel/sched/sched.h
+++ b/kernel/sched/sched.h
@@ -1118,6 +1118,8 @@ struct rq {
 	unsigned int		core_forceidle_occupation;
 	u64			core_forceidle_start;
 #endif
+	unsigned int		unthrottled_cfs_rq;
+	unsigned int		unthrottled_task;
 };
 
 #ifdef CONFIG_FAIR_GROUP_SCHED
-- 
2.39.5