sched: Don't run cpu-online with balance_push() enabled

We don't need to push away tasks when we come online, mark the push
complete right before the CPU dies.

XXX hotplug state machine has trouble with rollback here.

Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Reviewed-by: Valentin Schneider <[email protected]>
Tested-by: Valentin Schneider <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]

Author: Peter Zijlstra

kernel/sched/core.c

@@ -7320,10 +7320,12 @@ static void balance_push_set(int cpu, bool on)
 	struct rq_flags rf;
 
 	rq_lock_irqsave(rq, &rf);
-	if (on)
+	if (on) {
+		WARN_ON_ONCE(rq->balance_callback);
 		rq->balance_callback = &balance_push_callback;
-	else
+	} else if (rq->balance_callback == &balance_push_callback) {
 		rq->balance_callback = NULL;
+	}
 	rq_unlock_irqrestore(rq, &rf);
 }
 
@@ -7441,6 +7443,10 @@ int sched_cpu_activate(unsigned int cpu)
 	struct rq *rq = cpu_rq(cpu);
 	struct rq_flags rf;
 
+	/*
+	 * Make sure that when the hotplug state machine does a roll-back
+	 * we clear balance_push. Ideally that would happen earlier...
+	 */
 	balance_push_set(cpu, false);
 
 #ifdef CONFIG_SCHED_SMT
@@ -7608,6 +7614,12 @@ int sched_cpu_dying(unsigned int cpu)
 	}
 	rq_unlock_irqrestore(rq, &rf);
 
+	/*
+	 * Now that the CPU is offline, make sure we're welcome
+	 * to new tasks once we come back up.
+	 */
+	balance_push_set(cpu, false);
+
 	calc_load_migrate(rq);
 	update_max_interval();
 	nohz_balance_exit_idle(rq);