sched/hotplug: Convert cpu_[in]active notifiers to state machine

Now that we reduced everything into single notifiers, it's simple to move them
into the hotplug state machine space.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: rt@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>

include/linux/cpu.h

@@ -59,18 +59,6 @@ struct notifier_block;
  * CPU notifier priorities.
  */
 enum {
-	/*
-	 * SCHED_ACTIVE marks a cpu which is coming up active during
-	 * CPU_ONLINE and CPU_DOWN_FAILED and must be the first notifier.  Is
-	 * also cpuset according to cpu_active mask right after activating the
-	 * cpu. During CPU_DOWN_PREPARE, SCHED_INACTIVE reversed the operation.
-	 *
-	 * This ordering guarantees consistent cpu_active mask and
-	 * migration behavior to all cpu notifiers.
-	 */
-	CPU_PRI_SCHED_ACTIVE	= INT_MAX,
-	CPU_PRI_SCHED_INACTIVE	= INT_MIN,
-
 	/* migration should happen before other stuff but after perf */
 	CPU_PRI_PERF		= 20,
 	CPU_PRI_MIGRATION	= 10,

include/linux/cpuhotplug.h

@@ -13,6 +13,7 @@ enum cpuhp_state {
 	CPUHP_AP_ONLINE,
 	CPUHP_TEARDOWN_CPU,
 	CPUHP_AP_ONLINE_IDLE,
+	CPUHP_AP_ACTIVE,
 	CPUHP_AP_SMPBOOT_THREADS,
 	CPUHP_AP_NOTIFY_ONLINE,
 	CPUHP_AP_ONLINE_DYN,

include/linux/sched.h

@@ -373,6 +373,8 @@ extern void trap_init(void);
 extern void update_process_times(int user);
 extern void scheduler_tick(void);
 extern int sched_cpu_starting(unsigned int cpu);
+extern int sched_cpu_activate(unsigned int cpu);
+extern int sched_cpu_deactivate(unsigned int cpu);
 
 extern void sched_show_task(struct task_struct *p);
 

kernel/cpu.c

@@ -923,8 +923,6 @@ void cpuhp_online_idle(enum cpuhp_state state)
 
 	st->state = CPUHP_AP_ONLINE_IDLE;
 
-	/* The cpu is marked online, set it active now */
-	set_cpu_active(cpu, true);
 	/* Unpark the stopper thread and the hotplug thread of this cpu */
 	stop_machine_unpark(cpu);
 	kthread_unpark(st->thread);
@@ -1259,6 +1257,12 @@ static struct cpuhp_step cpuhp_ap_states[] = {
 	[CPUHP_AP_ONLINE] = {
 		.name			= "ap:online",
 	},
+	/* First state is scheduler control. Interrupts are enabled */
+	[CPUHP_AP_ACTIVE] = {
+		.name			= "sched:active",
+		.startup		= sched_cpu_activate,
+		.teardown		= sched_cpu_deactivate,
+	},
 	/* Handle smpboot threads park/unpark */
 	[CPUHP_AP_SMPBOOT_THREADS] = {
 		.name			= "smpboot:threads",

kernel/sched/core.c

@@ -6634,9 +6634,6 @@ static void sched_domains_numa_masks_set(unsigned int cpu)
 	int node = cpu_to_node(cpu);
 	int i, j;
 
-	if (!sched_smp_initialized)
-		return;
-
 	for (i = 0; i < sched_domains_numa_levels; i++) {
 		for (j = 0; j < nr_node_ids; j++) {
 			if (node_distance(j, node) <= sched_domains_numa_distance[i])
@@ -6649,9 +6646,6 @@ static void sched_domains_numa_masks_clear(unsigned int cpu)
 {
 	int i, j;
 
-	if (!sched_smp_initialized)
-		return;
-
 	for (i = 0; i < sched_domains_numa_levels; i++) {
 		for (j = 0; j < nr_node_ids; j++)
 			cpumask_clear_cpu(cpu, sched_domains_numa_masks[i][j]);
@@ -7051,12 +7045,9 @@ static int num_cpus_frozen;	/* used to mark begin/end of suspend/resume */
  * If we come here as part of a suspend/resume, don't touch cpusets because we
  * want to restore it back to its original state upon resume anyway.
  */
-static void cpuset_cpu_active(bool frozen)
+static void cpuset_cpu_active(void)
 {
-	if (!sched_smp_initialized)
+	if (cpuhp_tasks_frozen) {
-		return;
-
-	if (frozen) {
 		/*
 		 * num_cpus_frozen tracks how many CPUs are involved in suspend
 		 * resume sequence. As long as this is not the last online
@@ -7077,17 +7068,14 @@ static void cpuset_cpu_active(bool frozen)
 	cpuset_update_active_cpus(true);
 }
 
-static int cpuset_cpu_inactive(unsigned int cpu, bool frozen)
+static int cpuset_cpu_inactive(unsigned int cpu)
 {
 	unsigned long flags;
 	struct dl_bw *dl_b;
 	bool overflow;
 	int cpus;
 
-	if (!sched_smp_initialized)
+	if (!cpuhp_tasks_frozen) {
-		return 0;
-
-	if (!frozen) {
 		rcu_read_lock_sched();
 		dl_b = dl_bw_of(cpu);
 
@@ -7108,42 +7096,33 @@ static int cpuset_cpu_inactive(unsigned int cpu, bool frozen)
 	return 0;
 }
 
-static int sched_cpu_active(struct notifier_block *nfb, unsigned long action,
+int sched_cpu_activate(unsigned int cpu)
-			    void *hcpu)
 {
-	unsigned int cpu = (unsigned long)hcpu;
-
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_DOWN_FAILED:
-	case CPU_ONLINE:
 	set_cpu_active(cpu, true);
+
+	if (sched_smp_initialized) {
 		sched_domains_numa_masks_set(cpu);
-		cpuset_cpu_active(action & CPU_TASKS_FROZEN);
+		cpuset_cpu_active();
-		return NOTIFY_OK;
-	default:
-		return NOTIFY_DONE;
 	}
+	return 0;
 }
 
-static int sched_cpu_inactive(struct notifier_block *nfb,
+int sched_cpu_deactivate(unsigned int cpu)
-					unsigned long action, void *hcpu)
 {
-	unsigned int cpu = (unsigned long)hcpu;
 	int ret;
 
-	switch (action & ~CPU_TASKS_FROZEN) {
-	case CPU_DOWN_PREPARE:
 	set_cpu_active(cpu, false);
-		ret = cpuset_cpu_inactive(cpu, action & CPU_TASKS_FROZEN);
+
+	if (!sched_smp_initialized)
+		return 0;
+
+	ret = cpuset_cpu_inactive(cpu);
 	if (ret) {
 		set_cpu_active(cpu, true);
-			return notifier_from_errno(ret);
+		return ret;
 	}
 	sched_domains_numa_masks_clear(cpu);
-		return NOTIFY_OK;
+	return 0;
-	default:
-		return NOTIFY_DONE;
-	}
 }
 
 int sched_cpu_starting(unsigned int cpu)
@@ -7197,10 +7176,6 @@ static int __init migration_init(void)
 	migration_call(&migration_notifier, CPU_ONLINE, cpu);
 	register_cpu_notifier(&migration_notifier);
 
-	/* Register cpu active notifiers */
-	cpu_notifier(sched_cpu_active, CPU_PRI_SCHED_ACTIVE);
-	cpu_notifier(sched_cpu_inactive, CPU_PRI_SCHED_INACTIVE);
-
 	return 0;
 }
 early_initcall(migration_init);