Skip to contentSkip to conversation

little schedule() cleanup: use cached current value

PATCH In review Oleg Nesterov 2 participants 4 messages Nov 9, 2004

lore
Share & tools
lore

Oleg Nesterov

author
#0 lore raw

PATCH 1/1 little schedule() cleanup: use cached current value

Hello.

schedule() can use prev/next instead of get_current().

Oleg.

Signed-off-by Oleg Nesterov <[email protected]>
kernel/sched.c renamed from kernel/sched.c~ +4 -4
@@ -2508,7 +2508,7 @@ need_resched:
 	 * The idle thread is not allowed to schedule!
 	 * Remove this check after it has been exercised a bit.
 	 */
-	if (unlikely(current == rq->idle) && current->state != TASK_RUNNING) {
+	if (unlikely(prev == rq->idle) && prev->state != TASK_RUNNING) {
 		printk(KERN_ERR "bad: scheduling from the idle thread!\n");
 		dump_stack();
 	}

@@ -2531,8 +2531,8 @@ need_resched:
 
 	spin_lock_irq(&rq->lock);
 
-	if (unlikely(current->flags & PF_DEAD))
-		current->state = EXIT_DEAD;
+	if (unlikely(prev->flags & PF_DEAD))
+		prev->state = EXIT_DEAD;
 	/*
 	 * if entering off of a kernel preemption go straight
 	 * to picking the next task.

@@ -2636,7 +2636,7 @@ switch_tasks:
 	} else
 		spin_unlock_irq(&rq->lock);
 
-	reacquire_kernel_lock(current);
+	reacquire_kernel_lock(next);
 	preempt_enable_no_resched();
 	if (unlikely(test_thread_flag(TIF_NEED_RESCHED)))
 		goto need_resched;
#1 lore raw

kernel/sched.c switch_tasks: · line 2642

Quoted 9 lines · Oleg Nesterov
kernel/sched.c renamed from kernel/sched.c~ +1 -1
@@ -2636,7 +2636,7 @@ switch_tasks:
 	} else
 		spin_unlock_irq(&rq->lock);
 
-	reacquire_kernel_lock(current);
+	reacquire_kernel_lock(next);
 	preempt_enable_no_resched();
 	if (unlikely(test_thread_flag(TIF_NEED_RESCHED)))
 		goto need_resched;

nack. We switch the kernel stack in switch_to() so 'next' here is the old task we switched to before we went off the CPU. The value of 'next' doesnt get carried over a context-switch. (we do carry 'prev' over via assembly trickery because we need it, but not 'next'.) This change would most likely result in rare, hard-to-track-down BKL-related crashes all around the place.

the other bits are OK, please resend them.

Ingo

author
#2 lore raw
Quoted 2 lines · Ingo Molnar

nack. We switch the kernel stack in switch_to() so 'next' here is the old task we switched to before we went off the CPU.

yes, sorry. here is corrected patch.

schedule() can use prev instead of get_current().

Signed-off-by Oleg Nesterov <[email protected]>
kernel/sched.c renamed from kernel/sched.c~ +3 -3
@@ -2508,7 +2508,7 @@ need_resched:
 	 * The idle thread is not allowed to schedule!
 	 * Remove this check after it has been exercised a bit.
 	 */
-	if (unlikely(current == rq->idle) && current->state != TASK_RUNNING) {
+	if (unlikely(prev == rq->idle) && prev->state != TASK_RUNNING) {
 		printk(KERN_ERR "bad: scheduling from the idle thread!\n");
 		dump_stack();
 	}

@@ -2531,8 +2531,8 @@ need_resched:
 
 	spin_lock_irq(&rq->lock);
 
-	if (unlikely(current->flags & PF_DEAD))
-		current->state = EXIT_DEAD;
+	if (unlikely(prev->flags & PF_DEAD))
+		prev->state = EXIT_DEAD;
 	/*
 	 * if entering off of a kernel preemption go straight
 	 * to picking the next task.