[PATCH] sched/fair: Cleanup task->numa_work initialization

[PATCH] sched/fair: Cleanup task->numa_work initialization

[Thomas Gleixner]

- Resolve the ancient TODO by setting the numa_work function in
  init_numa_balancing() which is called on fork().

- Make task_numa_work() static as it's not used outside of the fair
  scheduler and lacks a prototype as well.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
---
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index bc9cfeaac8bd..f47869c0cdad 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -1103,6 +1103,7 @@ static struct numa_group *deref_curr_numa_group(struct task_struct *p)
 
 static inline unsigned long group_faults_priv(struct numa_group *ng);
 static inline unsigned long group_faults_shared(struct numa_group *ng);
+static void task_numa_work(struct callback_head *work);
 
 static unsigned int task_nr_scan_windows(struct task_struct *p)
 {
@@ -1203,6 +1204,7 @@ void init_numa_balancing(unsigned long clone_flags, struct task_struct *p)
 	p->node_stamp			= 0;
 	p->numa_scan_seq		= mm ? mm->numa_scan_seq : 0;
 	p->numa_scan_period		= sysctl_numa_balancing_scan_delay;
+	p->numa_work.func		= task_numa_work;
 	p->numa_work.next		= &p->numa_work;
 	p->numa_faults			= NULL;
 	RCU_INIT_POINTER(p->numa_group, NULL);
@@ -2523,7 +2525,7 @@ static void reset_ptenuma_scan(struct task_struct *p)
  * The expensive part of numa migration is done from task_work context.
  * Triggered from task_tick_numa().
  */
-void task_numa_work(struct callback_head *work)
+static void task_numa_work(struct callback_head *work)
 {
 	unsigned long migrate, next_scan, now = jiffies;
 	struct task_struct *p = current;
@@ -2693,10 +2695,8 @@ static void task_tick_numa(struct rq *rq, struct task_struct *curr)
 			curr->numa_scan_period = task_scan_start(curr);
 		curr->node_stamp += period;
 
-		if (!time_before(jiffies, curr->mm->numa_next_scan)) {
-			init_task_work(work, task_numa_work); /* TODO: move this into sched_fork() */
+		if (!time_before(jiffies, curr->mm->numa_next_scan))
 			task_work_add(curr, work, true);
-		}
 	}
 }
 

Re: [PATCH] sched/fair: Cleanup task->numa_work initialization

[Valentin Schneider]

Hi,

On 01/08/2019 21:51, Thomas Gleixner wrote:
> - Resolve the ancient TODO by setting the numa_work function in
>   init_numa_balancing() which is called on fork().
> 
> - Make task_numa_work() static as it's not used outside of the fair
>   scheduler and lacks a prototype as well.
> 
> Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

Looks like I beat you to it...

https://lore.kernel.org/lkml/20190715102508.32434-1-valentin.schneider@arm.com/

Do I win anything? :D