sched_rt.c: Avoid unnecessary dequeue and enqueue of pushable tasks in set_cpus_allowed_rt()

Migration status depends on a difference of weight from 0 and 1. If weight > 1 (<= 1) and old weight <= 1 (> 1) then task becomes pushable (not pushable). We are not insterested in exact values of it, is it 3 or 4, for example.

Now if we are changing affinity from a set of 3 cpus to a set of 4, the task will be dequeued and enqueued sequentially without important difference in comparison with initial state. The only difference is in internal representation of plist queue of pushable tasks and the fact that the task may won't be the first in a sequence of the same priority tasks. But it seems to me it gives nothing.

sched_rt.c renamed from sched_rt.c.orig +1 -0

@@ -1572,43 +1572,37 @@ static void set_cpus_allowed_rt(struct t
 				const struct cpumask *new_mask)
 {
 	int weight = cpumask_weight(new_mask);
+	struct rq *rq;
 
 	BUG_ON(!rt_task(p));
 

sched_rt.c renamed from sched_rt.c.orig +18 -25

-		if (!task_current(rq, p)) {
-			/*
-			 * Make sure we dequeue this task from the pushable list
-			 * before going further.  It will either remain off of
-			 * the list because we are no longer pushable, or it
-			 * will be requeued.
-			 */
-			if (p->rt.nr_cpus_allowed > 1)
-				dequeue_pushable_task(rq, p);
-
-			/*
-			 * Requeue if our weight is changing and still > 1
-			 */
-			if (weight > 1)
-				enqueue_pushable_task(rq, p);
-
-		}
-
-		if ((p->rt.nr_cpus_allowed <= 1) && (weight > 1)) {
-			rq->rt.rt_nr_migratory++;
-		} else if ((p->rt.nr_cpus_allowed > 1) && (weight <= 1)) {
-			BUG_ON(!rq->rt.rt_nr_migratory);
-			rq->rt.rt_nr_migratory--;
-		}
+	if (!p->on_rq)
+		return;
 
-		update_rt_migration(&rq->rt);
+	rq = task_rq(p);
+
+	/*
+	 * Several cpus were allowed but now it's not so OR vice versa
+	 */
+	if (weight <= 1) {
+		if (!task_current(rq, p))
+			dequeue_pushable_task(rq, p);
+		BUG_ON(!rq->rt.rt_nr_migratory);
+		rq->rt.rt_nr_migratory--;
+	} else {
+		if (!task_current(rq, p))
+			enqueue_pushable_task(rq, p);
+		rq->rt.rt_nr_migratory++;
 	}
+
+	update_rt_migration(&rq->rt);
 }
 
 /* Assumes rq->lock is held */
 

Quoted 6 lines

 Migration status depends on a difference of weight from 0 and 1. If weight > 1 (<= 1) and old weight <= 1 (> 1) then task becomes pushable (not pushable). We are not insterested in exact values of it, is it 3 or 4, for example.

Quoted 6 lines

 Now if we are changing affinity from a set of 3 cpus to a set of 4, the task will be dequeued and enqueued sequentially without important difference in comparison with initial state. The only difference is in internal representation of plist queue of pushable tasks and the fact that the task may won't be the first in a sequence of the same priority tasks. But it seems to me it gives nothing.

Looks reasonable, although I can't say I really understand this code. Add Gregory.

Quoted 4 lines

Signed-off-by Tkhai Kirill <[email protected]>

 --- kernel/sched_rt.c.orig 2011-12-02 00:29:11.970243145 +0400 +++ kernel/sched_rt.c 2011-12-02 00:37:43.622846606 +0400

please use -p1

Quoted 8 lines

 @@ -1572,43 +1572,37 @@ static void set_cpus_allowed_rt(struct t const struct cpumask *new_mask) {
          int weight = cpumask_weight(new_mask);
 + struct rq *rq;

          BUG_ON(!rt_task(p));

Quoted 10 lines

          /*
 - * Update the migration status of the RQ if we have an RT task
 - * which is running AND changing its weight value.
 + * Just exit if it's not necessary to change migration status */
 - if (p->on_rq && (weight != p->rt.nr_cpus_allowed)) {
 - struct rq *rq = task_rq(p);
 + if ((p->rt.nr_cpus_allowed <= 1 && weight <= 1)
 + || (p->rt.nr_cpus_allowed > 1 && weight > 1))
 + return;

Subjective, but may be

        if ((p->rt.nr_cpus_allowed > 1) != (weight > 1)) return;

looks more understandable?

20.12.2011, 21:44, "Oleg Nesterov" <[email protected]>:

Quoted 12 lines · Kirill Tkhai

Quoted 11 lines

  Migration status depends on a difference of weight from 0 and 1. If weight > 1 (<= 1) and old weight <= 1 (> 1) then task becomes pushable (not pushable). We are not insterested in exact values of it, is it 3 or 4, for example.

  Now if we are changing affinity from a set of 3 cpus to a set of 4, the task will be dequeued and enqueued sequentially without important difference in comparison with initial state. The only difference is in internal representation of plist queue of pushable tasks and the fact that the task may won't be the first in a sequence of the same priority tasks. But it seems to me it gives nothing.

Quoted 7 lines

 Looks reasonable, although I can't say I really understand this code. Add Gregory.

Quoted 4 lines

Signed-off-by Tkhai Kirill <[email protected]>

  --- kernel/sched_rt.c.orig 2011-12-02 00:29:11.970243145 +0400 +++ kernel/sched_rt.c 2011-12-02 00:37:43.622846606 +0400

 please use -p1

Sorry, this time I'm sending "git diffed" output.

Quoted 18 lines

Quoted 8 lines

  @@ -1572,43 +1572,37 @@ static void set_cpus_allowed_rt(struct t const struct cpumask *new_mask) {
           int weight = cpumask_weight(new_mask);
  + struct rq *rq;

           BUG_ON(!rt_task(p));

Quoted 10 lines

           /*
  - * Update the migration status of the RQ if we have an RT task
  - * which is running AND changing its weight value.
  + * Just exit if it's not necessary to change migration status */
  - if (p->on_rq && (weight != p->rt.nr_cpus_allowed)) {
  - struct rq *rq = task_rq(p);
  + if ((p->rt.nr_cpus_allowed <= 1 && weight <= 1)
  + || (p->rt.nr_cpus_allowed > 1 && weight > 1))
  + return;

Quoted 6 lines

 Subjective, but may be

         if ((p->rt.nr_cpus_allowed > 1) != (weight > 1)) return;

 looks more understandable?

Yes, thanks.

---

kernel/sched/rt.c +15 -28

index 3640ebb..4467f4d 100644
@@ -1774,43 +1774,36 @@ static void set_cpus_allowed_rt(struct task_struct *p,
                                  const struct cpumask *new_mask)
  {
          int weight = cpumask_weight(new_mask);
+ struct rq *rq;
 
          BUG_ON(!rt_task(p));
 
          /*
- * Update the migration status of the RQ if we have an RT task
- * which is running AND changing its weight value.
+ * Just exit if it's not necessary to change migration status
           */
- if (p->on_rq && (weight != p->rt.nr_cpus_allowed)) {
- struct rq *rq = task_rq(p);
-
- if (!task_current(rq, p)) {
- /*
- * Make sure we dequeue this task from the pushable list
- * before going further.  It will either remain off of
- * the list because we are no longer pushable, or it
- * will be requeued.
- */
- if (p->rt.nr_cpus_allowed > 1)
- dequeue_pushable_task(rq, p);
-
- /*
- * Requeue if our weight is changing and still > 1
- */
- if (weight > 1)
- enqueue_pushable_task(rq, p);
+ if ((p->rt.nr_cpus_allowed > 1) != (weight > 1))
+ return;
 
- }
+ if (!p->on_rq)
+ return;
 
- if ((p->rt.nr_cpus_allowed <= 1) && (weight > 1)) {
- rq->rt.rt_nr_migratory++;
- } else if ((p->rt.nr_cpus_allowed > 1) && (weight <= 1)) {
- BUG_ON(!rq->rt.rt_nr_migratory);
- rq->rt.rt_nr_migratory--;
- }
+ rq = task_rq(p);
 
- update_rt_migration(&rq->rt);
+ /*
+ * Several cpus were allowed but now it's not so OR vice versa
+ */
+ if (weight <= 1) {
+ if (!task_current(rq, p))
+ dequeue_pushable_task(rq, p);
+ BUG_ON(!rq->rt.rt_nr_migratory);
+ rq->rt.rt_nr_migratory--;

[PATCH]sched_rt.c: Avoid unnecessary dequeue and enqueue of pushable tasks in set_cpus_allowed_rt()

Migration status depends on a difference of weight from 0 and 1. If weight > 1 (<= 1) and old weight <= 1 (> 1) then task becomes pushable (not pushable). We are not insterested in exact values of it, is it 3 or 4, for example.

Now if we are changing affinity from a set of 3 cpus to a set of 4, the task will be dequeued and enqueued sequentially without important difference in comparison with initial state. The only difference is in internal representation of plist queue of pushable tasks and the fact that the task may won't be the first in a sequence of the same priority tasks. But it seems to me it gives nothing.

Signed-off-by Tkhai Kirill <[email protected]>

sched_rt.c renamed from sched_rt.c.orig +23 -30

@@ -1572,43 +1572,37 @@ static void set_cpus_allowed_rt(struct t
 				const struct cpumask *new_mask)
 {
 	int weight = cpumask_weight(new_mask);
+	struct rq *rq;
 
 	BUG_ON(!rt_task(p));
 
 	/*
-	 * Update the migration status of the RQ if we have an RT task
-	 * which is running AND changing its weight value.
+	 * Just exit if it's not necessary to change migration status
 	 */
-	if (p->on_rq && (weight != p->rt.nr_cpus_allowed)) {
-		struct rq *rq = task_rq(p);
+	if ((p->rt.nr_cpus_allowed <= 1 && weight <= 1)
+		|| (p->rt.nr_cpus_allowed > 1 && weight > 1))
+		return;
 
-		if (!task_current(rq, p)) {
-			/*
-			 * Make sure we dequeue this task from the pushable list
-			 * before going further.  It will either remain off of
-			 * the list because we are no longer pushable, or it
-			 * will be requeued.
-			 */
-			if (p->rt.nr_cpus_allowed > 1)
-				dequeue_pushable_task(rq, p);
-
-			/*
-			 * Requeue if our weight is changing and still > 1
-			 */
-			if (weight > 1)
-				enqueue_pushable_task(rq, p);
-
-		}
-
-		if ((p->rt.nr_cpus_allowed <= 1) && (weight > 1)) {
-			rq->rt.rt_nr_migratory++;
-		} else if ((p->rt.nr_cpus_allowed > 1) && (weight <= 1)) {
-			BUG_ON(!rq->rt.rt_nr_migratory);
-			rq->rt.rt_nr_migratory--;
-		}
+	if (!p->on_rq)
+		return;
 
-		update_rt_migration(&rq->rt);
+	rq = task_rq(p);
+
+	/*
+	 * Several cpus were allowed but now it's not so OR vice versa
+	 */
+	if (weight <= 1) {
+		if (!task_current(rq, p))
+			dequeue_pushable_task(rq, p);
+		BUG_ON(!rq->rt.rt_nr_migratory);
+		rq->rt.rt_nr_migratory--;
+	} else {
+		if (!task_current(rq, p))
+			enqueue_pushable_task(rq, p);
+		rq->rt.rt_nr_migratory++;
 	}
+
+	update_rt_migration(&rq->rt);
 }
 
 /* Assumes rq->lock is held */
 
 
 
--

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

I wasn't on the Cc of the original message, but it was bounced to me awhile ago. I'm cleaning out my email and came across it.

Can you send me the latest version of this patch, either against latest Linus, or against tip/master.

Thanks,

-- Steve

P.S. I'll be at ELC this week so it may not get processed right away.

13.02.2012, 21:23, "Steven Rostedt" <[email protected]>:

Quoted 11 lines

 I wasn't on the Cc of the original message, but it was bounced to me awhile ago. I'm cleaning out my email and came across it.

 Can you send me the latest version of this patch, either against latest Linus, or against tip/master.

 Thanks,

 -- Steve

 P.S. I'll be at ELC this week so it may not get processed right away.

Migration status depends on a difference of weight from 0 and 1. If weight > 1 (<= 1) and old weight <= 1 (> 1) then task becomes pushable (or not pushable). We are not insterested in its exact values, is it 3 or 4, for example.

Now if we are changing affinity from a set of 3 cpus to a set of 4, the- task will be dequeued and enqueued sequentially without important difference in comparison with initial state. The only difference is in internal representation of plist queue of pushable tasks and the fact that the task may won't be the first in a sequence of the same priority tasks. But it seems to me it gives nothing.

Signed-off-by Tkhai Kirill <[email protected]>

Steven, what is about the patch from my previous message? Is everything ok?

Regards,
Kirill

19.02.2012, 18:17, "Kirill Tkhai" <[email protected]>:

Quoted 27 lines

13.02.2012, 21:23, "Steven Rostedt" <[email protected]>:

Quoted 11 lines

I wasn't on the Cc of the original message, but it was bounced to me awhile ago. I'm cleaning out my email and came across it.

Can you send me the latest version of this patch, either against latest Linus, or against tip/master.

Thanks,

-- Steve

P.S. I'll be at ELC this week so it may not get processed right away.

Migration status depends on a difference of weight from 0 and 1. If weight > 1 (<= 1) and old weight <= 1 (> 1) then task becomes pushable (or not pushable). We are not insterested in its exact values, is it 3 or 4, for example.

Now if we are changing affinity from a set of 3 cpus to a set of 4, the- task will be dequeued and enqueued sequentially without important difference in comparison with initial state. The only difference is in internal representation of plist queue of pushable tasks and the fact that the task may won't be the first in a sequence of the same priority tasks. But it seems to me it gives nothing.

Signed-off-by Tkhai Kirill <[email protected]>

kernel/sched/rt.c +0 -14

 	 */
-	if (p->on_rq && (weight != p->rt.nr_cpus_allowed)) {
-		struct rq *rq = task_rq(p);
-
-		if (!task_current(rq, p)) {
-			/*
-			 * Make sure we dequeue this task from the pushable list
-			 * before going further.  It will either remain off of
-			 * the list because we are no longer pushable, or it
-			 * will be requeued.
-			 */
-			if (p->rt.nr_cpus_allowed > 1)
-				dequeue_pushable_task(rq, p);
-
-			/*

kernel/sched/rt.c +12 -0

+	 */
+	if (weight <= 1) {
+		if (!task_current(rq, p))
+			dequeue_pushable_task(rq, p);
+		BUG_ON(!rq->rt.rt_nr_migratory);
+		rq->rt.rt_nr_migratory--;
+	} else {
+		if (!task_current(rq, p))
+			enqueue_pushable_task(rq, p);
+		rq->rt.rt_nr_migratory++;
 	}
+
+	update_rt_migration(&rq->rt);
 }
 
 /* Assumes rq->lock is held */