sched: Update_cfs_shares at period edge

Now that our measurement intervals are small (~1ms) we can amortize the posting
of update_shares() to be about each period overflow.  This is a large cost
saving for frequently switching tasks.
Signed-off-by: Paul Turner <pjt@google.com>
Reviewed-by: Ben Segall <bsegall@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20120823141507.200772172@google.comSigned-off-by: Ingo Molnar <mingo@kernel.org>

kernel/sched/fair.c

@@ -1187,6 +1187,7 @@ static void update_cfs_rq_blocked_load(struct cfs_rq *cfs_rq, int force_update)
 	}
 
 	__update_cfs_rq_tg_load_contrib(cfs_rq, force_update);
+	update_cfs_shares(cfs_rq);
 }
 
 static inline void update_rq_runnable_avg(struct rq *rq, int runnable)
@@ -1396,9 +1397,8 @@ enqueue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags)
 	 * Update run-time statistics of the 'current'.
 	 */
 	update_curr(cfs_rq);
-	enqueue_entity_load_avg(cfs_rq, se, flags & ENQUEUE_WAKEUP);
 	account_entity_enqueue(cfs_rq, se);
-	update_cfs_shares(cfs_rq);
+	enqueue_entity_load_avg(cfs_rq, se, flags & ENQUEUE_WAKEUP);
 
 	if (flags & ENQUEUE_WAKEUP) {
 		place_entity(cfs_rq, se, 0);
@@ -1471,7 +1471,6 @@ dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags)
 	 * Update run-time statistics of the 'current'.
 	 */
 	update_curr(cfs_rq);
-	dequeue_entity_load_avg(cfs_rq, se, flags & DEQUEUE_SLEEP);
 
 	update_stats_dequeue(cfs_rq, se);
 	if (flags & DEQUEUE_SLEEP) {
@@ -1491,8 +1490,8 @@ dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags)
 
 	if (se != cfs_rq->curr)
 		__dequeue_entity(cfs_rq, se);
-	se->on_rq = 0;
 	account_entity_dequeue(cfs_rq, se);
+	dequeue_entity_load_avg(cfs_rq, se, flags & DEQUEUE_SLEEP);
 
 	/*
 	 * Normalize the entity after updating the min_vruntime because the
@@ -1506,7 +1505,7 @@ dequeue_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int flags)
 	return_cfs_rq_runtime(cfs_rq);
 
 	update_min_vruntime(cfs_rq);
-	update_cfs_shares(cfs_rq);
+	se->on_rq = 0;
 }
 
 /*
@@ -2518,8 +2517,8 @@ enqueue_task_fair(struct rq *rq, struct task_struct *p, int flags)
 		if (cfs_rq_throttled(cfs_rq))
 			break;
 
-		update_cfs_shares(cfs_rq);
 		update_entity_load_avg(se, 1);
+		update_cfs_rq_blocked_load(cfs_rq, 0);
 	}
 
 	if (!se) {
@@ -2579,8 +2578,8 @@ static void dequeue_task_fair(struct rq *rq, struct task_struct *p, int flags)
 		if (cfs_rq_throttled(cfs_rq))
 			break;
 
-		update_cfs_shares(cfs_rq);
 		update_entity_load_avg(se, 1);
+		update_cfs_rq_blocked_load(cfs_rq, 0);
 	}
 
 	if (!se) {
@@ -5639,8 +5638,11 @@ int sched_group_set_shares(struct task_group *tg, unsigned long shares)
 		se = tg->se[i];
 		/* Propagate contribution to hierarchy */
 		raw_spin_lock_irqsave(&rq->lock, flags);
-		for_each_sched_entity(se)
+		for_each_sched_entity(se) {
 			update_cfs_shares(group_cfs_rq(se));
+			/* update contribution to parent */
+			update_entity_load_avg(se, 1);
+		}
 		raw_spin_unlock_irqrestore(&rq->lock, flags);
 	}