sched/deadline: Make dl_rq->pushable_dl_tasks update drive dl_rq->ove…

…rloaded

dl_rq->dl_nr_migratory is increased whenever a DL entity is enqueued and it has
nr_cpus_allowed > 1. Unlike the pushable_dl_tasks tree, dl_rq->dl_nr_migratory
includes a dl_rq's current task. This means a dl_rq can have a migratable
current, N non-migratable queued tasks, and be flagged as overloaded and have
its CPU set in the dlo_mask, despite having an empty pushable_tasks tree.

Make an dl_rq's overload logic be driven by {enqueue,dequeue}_pushable_dl_task(),
in other words make DL RQs only be flagged as overloaded if they have at
least one runnable-but-not-current migratable task.

 o push_dl_task() is unaffected, as it is a no-op if there are no pushable
   tasks.

 o pull_dl_task() now no longer scans runqueues whose sole migratable task is
   their current one, which it can't do anything about anyway.
   It may also now pull tasks to a DL RQ with dl_nr_running > 1 if only its
   current task is migratable.

Since dl_rq->dl_nr_migratory becomes unused, remove it.

RT had the exact same mechanism (rt_rq->rt_nr_migratory) which was dropped
in favour of relying on rt_rq->pushable_tasks, see:

  612f769 ("sched/rt: Make rt_rq->pushable_tasks updates drive rto_mask")

Signed-off-by: Valentin Schneider <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>
Acked-by: Juri Lelli <[email protected]>
Link: https://lore.kernel.org/r/[email protected]

kernel/sched/deadline.c

@@ -509,7 +509,6 @@ void init_dl_rq(struct dl_rq *dl_rq)
 	/* zero means no -deadline tasks */
 	dl_rq->earliest_dl.curr = dl_rq->earliest_dl.next = 0;
 
-	dl_rq->dl_nr_migratory = 0;
 	dl_rq->overloaded = 0;
 	dl_rq->pushable_dl_tasks_root = RB_ROOT_CACHED;
 #else
@@ -553,39 +552,6 @@ static inline void dl_clear_overload(struct rq *rq)
 	cpumask_clear_cpu(rq->cpu, rq->rd->dlo_mask);
 }
 
-static void update_dl_migration(struct dl_rq *dl_rq)
-{
-	if (dl_rq->dl_nr_migratory && dl_rq->dl_nr_running > 1) {
-		if (!dl_rq->overloaded) {
-			dl_set_overload(rq_of_dl_rq(dl_rq));
-			dl_rq->overloaded = 1;
-		}
-	} else if (dl_rq->overloaded) {
-		dl_clear_overload(rq_of_dl_rq(dl_rq));
-		dl_rq->overloaded = 0;
-	}
-}
-
-static void inc_dl_migration(struct sched_dl_entity *dl_se, struct dl_rq *dl_rq)
-{
-	struct task_struct *p = dl_task_of(dl_se);
-
-	if (p->nr_cpus_allowed > 1)
-		dl_rq->dl_nr_migratory++;
-
-	update_dl_migration(dl_rq);
-}
-
-static void dec_dl_migration(struct sched_dl_entity *dl_se, struct dl_rq *dl_rq)
-{
-	struct task_struct *p = dl_task_of(dl_se);
-
-	if (p->nr_cpus_allowed > 1)
-		dl_rq->dl_nr_migratory--;
-
-	update_dl_migration(dl_rq);
-}
-
 #define __node_2_pdl(node) \
 	rb_entry((node), struct task_struct, pushable_dl_tasks)
 
@@ -594,6 +560,11 @@ static inline bool __pushable_less(struct rb_node *a, const struct rb_node *b)
 	return dl_entity_preempt(&__node_2_pdl(a)->dl, &__node_2_pdl(b)->dl);
 }
 
+static inline int has_pushable_dl_tasks(struct rq *rq)
+{
+	return !RB_EMPTY_ROOT(&rq->dl.pushable_dl_tasks_root.rb_root);
+}
+
 /*
  * The list of pushable -deadline task is not a plist, like in
  * sched_rt.c, it is an rb-tree with tasks ordered by deadline.
@@ -609,6 +580,11 @@ static void enqueue_pushable_dl_task(struct rq *rq, struct task_struct *p)
 				 __pushable_less);
 	if (leftmost)
 		rq->dl.earliest_dl.next = p->dl.deadline;
+
+	if (!rq->dl.overloaded) {
+		dl_set_overload(rq);
+		rq->dl.overloaded = 1;
+	}
 }
 
 static void dequeue_pushable_dl_task(struct rq *rq, struct task_struct *p)
@@ -625,11 +601,11 @@ static void dequeue_pushable_dl_task(struct rq *rq, struct task_struct *p)
 		dl_rq->earliest_dl.next = __node_2_pdl(leftmost)->dl.deadline;
 
 	RB_CLEAR_NODE(&p->pushable_dl_tasks);
-}
 
-static inline int has_pushable_dl_tasks(struct rq *rq)
-{
-	return !RB_EMPTY_ROOT(&rq->dl.pushable_dl_tasks_root.rb_root);
+	if (!has_pushable_dl_tasks(rq) && rq->dl.overloaded) {
+		dl_clear_overload(rq);
+		rq->dl.overloaded = 0;
+	}
 }
 
 static int push_dl_task(struct rq *rq);
@@ -1504,7 +1480,6 @@ void inc_dl_tasks(struct sched_dl_entity *dl_se, struct dl_rq *dl_rq)
 	add_nr_running(rq_of_dl_rq(dl_rq), 1);
 
 	inc_dl_deadline(dl_rq, deadline);
-	inc_dl_migration(dl_se, dl_rq);
 }
 
 static inline
@@ -1518,7 +1493,6 @@ void dec_dl_tasks(struct sched_dl_entity *dl_se, struct dl_rq *dl_rq)
 	sub_nr_running(rq_of_dl_rq(dl_rq), 1);
 
 	dec_dl_deadline(dl_rq, dl_se->deadline);
-	dec_dl_migration(dl_se, dl_rq);
 }
 
 static inline bool __dl_less(struct rb_node *a, const struct rb_node *b)
@@ -2291,9 +2265,6 @@ static int push_dl_task(struct rq *rq)
 	struct rq *later_rq;
 	int ret = 0;
 
-	if (!rq->dl.overloaded)
-		return 0;
-
 	next_task = pick_next_pushable_dl_task(rq);
 	if (!next_task)
 		return 0;

kernel/sched/debug.c

@@ -745,7 +745,6 @@ void print_dl_rq(struct seq_file *m, int cpu, struct dl_rq *dl_rq)
 
 	PU(dl_nr_running);
 #ifdef CONFIG_SMP
-	PU(dl_nr_migratory);
 	dl_bw = &cpu_rq(cpu)->rd->dl_bw;
 #else
 	dl_bw = &dl_rq->dl_bw;

kernel/sched/sched.h

@@ -707,7 +707,6 @@ struct dl_rq {
 		u64		next;
 	} earliest_dl;
 
-	unsigned int		dl_nr_migratory;
 	int			overloaded;
 
 	/*