add requeue task

Quoted 1 line · Con Kolivas

add requeue task

ack for this bit, but please split this one out:

| Change the granularity code to requeue tasks at their best priority
| instead of changing priority while they're running. This keeps tasks | at their top interactive level during their whole timeslice.

this is a behavioral change and should go into the 'possible negative effect on interactivity' bucket. (of course it could very likely have a positive effect as well - the bucket is just to separate the risks.)

Quoted 7 lines

| Change the granularity code to requeue tasks at their best priority
| instead of changing priority while they're running. This keeps tasks 
| at their top interactive level during their whole timeslice.

this is a behavioral change and should go into the 'possible negative effect on interactivity' bucket. (of course it could very likely have a positive effect as well - the bucket is just to separate the risks.)

Actually I'd like to say I did it for positive effect to counter the change that occurred with dropping interactive credit. That was what I found in my testing, and I'd like them both to go together into -mm.

add requeue task

------------------------------------------------------------------------

We can requeue tasks for cheaper then doing a complete dequeue followed by an enqueue. Add the requeue_task function and perform it where possible.

Change the granularity code to requeue tasks at their best priority instead of changing priority while they're running. This keeps tasks at their top interactive level during their whole timeslice.

Signed-off-by Con Kolivas <[email protected]>

Index: linux-2.6.10-rc1-mm2/kernel/sched.c
===================================================================

kernel/sched.c +11 -2

@@ -579,6 +579,16 @@ static void enqueue_task(struct task_str
 }
 
 /*
+ * Put task to the end of the run list without the overhead of dequeue
+ * followed by enqueue.
+ */
+static void requeue_task(struct task_struct *p, prio_array_t *array)
+{
+	list_del(&p->run_list);
+	list_add_tail(&p->run_list, array->queue + p->prio);
+}
+
+/*
  * Used by the migration code - we pull tasks from the head of the
  * remote queue so we want these tasks to show up at the head of the
  * local queue:

@@ -2425,8 +2435,7 @@ void scheduler_tick(void)
 			set_tsk_need_resched(p);
 
 			/* put it at the end of the queue: */
-			dequeue_task(p, rq->active);
-			enqueue_task(p, rq->active);
+			requeue_task(p, rq->active);
 		}
 		goto out_unlock;
 	}

@@ -2467,10 +2476,8 @@ void scheduler_tick(void)
 			(p->time_slice >= TIMESLICE_GRANULARITY(p)) &&

kernel/sched.c +1 -2

@@ -3569,8 +3576,14 @@ asmlinkage long sys_sched_yield(void)
 	} else if (!rq->expired->nr_active)
 		schedstat_inc(rq, yld_exp_empty);
 
-	dequeue_task(current, array);
-	enqueue_task(current, target);
+	if (array != target) {

add requeue task

Quoted 4 lines · Con Kolivas

add requeue task

Quoted 9 lines

------------------------------------------------------------------------

We can requeue tasks for cheaper then doing a complete dequeue followed by an enqueue. Add the requeue_task function and perform it where possible.

Change the granularity code to requeue tasks at their best priority instead of changing priority while they're running. This keeps tasks at their top interactive level during their whole timeslice.

I wonder... these things are all in sufficiently rarely used places, that the icache miss might be more costly than the operations saved.

But....

Quoted 9 lines

Signed-off-by Con Kolivas <[email protected]>

Index: linux-2.6.10-rc1-mm2/kernel/sched.c
===================================================================

kernel/sched.c +0 -0

@@ -579,6 +579,16 @@ static void enqueue_task(struct task_str
 }
 

kernel/sched.c static void enqueue_task(struct task_str · line 589

Quoted 9 lines

kernel/sched.c +8 -0

 /*
+ * Put task to the end of the run list without the overhead of dequeue
+ * followed by enqueue.
+ */
+static void requeue_task(struct task_struct *p, prio_array_t *array)
+{
+	list_del(&p->run_list);
+	list_add_tail(&p->run_list, array->queue + p->prio);
+}

Quoted 12 lines · Nick Piggin

kernel/sched.c static void enqueue_task(struct task_str · line 589

Quoted 11 lines · Con Kolivas

add requeue task

/*
+ * Put task to the end of the run list without the overhead of dequeue
+ * followed by enqueue.
+ */

kernel/sched.c +5 -0

+static void requeue_task(struct task_struct *p, prio_array_t *array)
+{
+	list_del(&p->run_list);
+	list_add_tail(&p->run_list, array->queue + p->prio);
+}

Shouldn't this be

list_move_tail(&p->run_list, array->queue + p->prio);

?

Quoted 4 lines

kernel/sched.c static void enqueue_task(struct task_str · line 589

Quoted 3 lines · Con Kolivas

kernel/sched.c +3 -0

+	list_del(&p->run_list);
+	list_add_tail(&p->run_list, array->queue + p->prio);
+}

Shouldn't this be

list_move_tail(&p->run_list, array->queue + p->prio);