sched/idle: Add support for tasks that inject idle

Idle injection drivers such as Intel powerclamp and ACPI PAD drivers use
realtime tasks to take control of CPU then inject idle. There are two
issues with this approach:

 1. Low efficiency: injected idle task is treated as busy so sched ticks
    do not stop during injected idle period, the result of these
    unwanted wakeups can be ~20% loss in power savings.

 2. Idle accounting: injected idle time is presented to user as busy.

This patch addresses the issues by introducing a new PF_IDLE flag which
allows any given task to be treated as idle task while the flag is set.
Therefore, idle injection tasks can run through the normal flow of NOHZ
idle enter/exit to get the correct accounting as well as tick stop when
possible.

The implication is that idle task is then no longer limited to PID == 0.

Acked-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Jacob Pan <jacob.jun.pan@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>

include/linux/cpu.h

@@ -245,6 +245,8 @@ void arch_cpu_idle_dead(void);
 int cpu_report_state(int cpu);
 int cpu_check_up_prepare(int cpu);
 void cpu_set_state_online(int cpu);
+void play_idle(unsigned long duration_ms);
+
 #ifdef CONFIG_HOTPLUG_CPU
 bool cpu_wait_death(unsigned int cpu, int seconds);
 bool cpu_report_death(void);

include/linux/sched.h

@@ -2254,6 +2254,7 @@ extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut,
 /*
  * Per process flags
  */
+#define PF_IDLE		0x00000002	/* I am an IDLE thread */
 #define PF_EXITING	0x00000004	/* getting shut down */
 #define PF_EXITPIDONE	0x00000008	/* pi exit done on shut down */
 #define PF_VCPU		0x00000010	/* I'm a virtual CPU */
@@ -2609,7 +2610,7 @@ extern struct task_struct *idle_task(int cpu);
  */
 static inline bool is_idle_task(const struct task_struct *p)
 {
-	return p->pid == 0;
+	return !!(p->flags & PF_IDLE);
 }
 extern struct task_struct *curr_task(int cpu);
 extern void ia64_set_curr_task(int cpu, struct task_struct *p);

kernel/fork.c

@@ -1537,7 +1537,7 @@ static __latent_entropy struct task_struct *copy_process(
 		goto bad_fork_cleanup_count;
 
 	delayacct_tsk_init(p);	/* Must remain after dup_task_struct() */
-	p->flags &= ~(PF_SUPERPRIV | PF_WQ_WORKER);
+	p->flags &= ~(PF_SUPERPRIV | PF_WQ_WORKER | PF_IDLE);
 	p->flags |= PF_FORKNOEXEC;
 	INIT_LIST_HEAD(&p->children);
 	INIT_LIST_HEAD(&p->sibling);

kernel/sched/core.c

@@ -5285,6 +5285,7 @@ void init_idle(struct task_struct *idle, int cpu)
 	__sched_fork(0, idle);
 	idle->state = TASK_RUNNING;
 	idle->se.exec_start = sched_clock();
+	idle->flags |= PF_IDLE;
 
 	kasan_unpoison_task_stack(idle);
 

kernel/sched/idle.c

@@ -205,76 +205,65 @@ exit_idle:
  *
  * Called with polling cleared.
  */
-static void cpu_idle_loop(void)
+static void do_idle(void)
 {
-	int cpu = smp_processor_id();
+	/*
+	 * If the arch has a polling bit, we maintain an invariant:
+	 *
+	 * Our polling bit is clear if we're not scheduled (i.e. if rq->curr !=
+	 * rq->idle). This means that, if rq->idle has the polling bit set,
+	 * then setting need_resched is guaranteed to cause the CPU to
+	 * reschedule.
+	 */
 
-	while (1) {
-		/*
-		 * If the arch has a polling bit, we maintain an invariant:
-		 *
-		 * Our polling bit is clear if we're not scheduled (i.e. if
-		 * rq->curr != rq->idle).  This means that, if rq->idle has
-		 * the polling bit set, then setting need_resched is
-		 * guaranteed to cause the cpu to reschedule.
-		 */
+	__current_set_polling();
+	tick_nohz_idle_enter();
 
-		__current_set_polling();
-		quiet_vmstat();
-		tick_nohz_idle_enter();
+	while (!need_resched()) {
+		check_pgt_cache();
+		rmb();
 
-		while (!need_resched()) {
-			check_pgt_cache();
-			rmb();
-
-			if (cpu_is_offline(cpu)) {
-				cpuhp_report_idle_dead();
-				arch_cpu_idle_dead();
-			}
-
-			local_irq_disable();
-			arch_cpu_idle_enter();
-
-			/*
-			 * In poll mode we reenable interrupts and spin.
-			 *
-			 * Also if we detected in the wakeup from idle
-			 * path that the tick broadcast device expired
-			 * for us, we don't want to go deep idle as we
-			 * know that the IPI is going to arrive right
-			 * away
-			 */
-			if (cpu_idle_force_poll || tick_check_broadcast_expired())
-				cpu_idle_poll();
-			else
-				cpuidle_idle_call();
-
-			arch_cpu_idle_exit();
+		if (cpu_is_offline(smp_processor_id())) {
+			cpuhp_report_idle_dead();
+			arch_cpu_idle_dead();
 		}
 
-		/*
-		 * Since we fell out of the loop above, we know
-		 * TIF_NEED_RESCHED must be set, propagate it into
-		 * PREEMPT_NEED_RESCHED.
-		 *
-		 * This is required because for polling idle loops we will
-		 * not have had an IPI to fold the state for us.
-		 */
-		preempt_set_need_resched();
-		tick_nohz_idle_exit();
-		__current_clr_polling();
+		local_irq_disable();
+		arch_cpu_idle_enter();
 
 		/*
-		 * We promise to call sched_ttwu_pending and reschedule
-		 * if need_resched is set while polling is set.  That
-		 * means that clearing polling needs to be visible
-		 * before doing these things.
+		 * In poll mode we reenable interrupts and spin. Also if we
+		 * detected in the wakeup from idle path that the tick
+		 * broadcast device expired for us, we don't want to go deep
+		 * idle as we know that the IPI is going to arrive right away.
 		 */
-		smp_mb__after_atomic();
-
-		sched_ttwu_pending();
-		schedule_preempt_disabled();
+		if (cpu_idle_force_poll || tick_check_broadcast_expired())
+			cpu_idle_poll();
+		else
+			cpuidle_idle_call();
+		arch_cpu_idle_exit();
 	}
+
+	/*
+	 * Since we fell out of the loop above, we know TIF_NEED_RESCHED must
+	 * be set, propagate it into PREEMPT_NEED_RESCHED.
+	 *
+	 * This is required because for polling idle loops we will not have had
+	 * an IPI to fold the state for us.
+	 */
+	preempt_set_need_resched();
+	tick_nohz_idle_exit();
+	__current_clr_polling();
+
+	/*
+	 * We promise to call sched_ttwu_pending() and reschedule if
+	 * need_resched() is set while polling is set. That means that clearing
+	 * polling needs to be visible before doing these things.
+	 */
+	smp_mb__after_atomic();
+
+	sched_ttwu_pending();
+	schedule_preempt_disabled();
 }
 
 bool cpu_in_idle(unsigned long pc)
@@ -283,6 +272,56 @@ bool cpu_in_idle(unsigned long pc)
 		pc < (unsigned long)__cpuidle_text_end;
 }
 
+struct idle_timer {
+	struct hrtimer timer;
+	int done;
+};
+
+static enum hrtimer_restart idle_inject_timer_fn(struct hrtimer *timer)
+{
+	struct idle_timer *it = container_of(timer, struct idle_timer, timer);
+
+	WRITE_ONCE(it->done, 1);
+	set_tsk_need_resched(current);
+
+	return HRTIMER_NORESTART;
+}
+
+void play_idle(unsigned long duration_ms)
+{
+	struct idle_timer it;
+
+	/*
+	 * Only FIFO tasks can disable the tick since they don't need the forced
+	 * preemption.
+	 */
+	WARN_ON_ONCE(current->policy != SCHED_FIFO);
+	WARN_ON_ONCE(current->nr_cpus_allowed != 1);
+	WARN_ON_ONCE(!(current->flags & PF_KTHREAD));
+	WARN_ON_ONCE(!(current->flags & PF_NO_SETAFFINITY));
+	WARN_ON_ONCE(!duration_ms);
+
+	rcu_sleep_check();
+	preempt_disable();
+	current->flags |= PF_IDLE;
+	cpuidle_use_deepest_state(true);
+
+	it.done = 0;
+	hrtimer_init_on_stack(&it.timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
+	it.timer.function = idle_inject_timer_fn;
+	hrtimer_start(&it.timer, ms_to_ktime(duration_ms), HRTIMER_MODE_REL_PINNED);
+
+	while (!READ_ONCE(it.done))
+		do_idle();
+
+	cpuidle_use_deepest_state(false);
+	current->flags &= ~PF_IDLE;
+
+	preempt_fold_need_resched();
+	preempt_enable();
+}
+EXPORT_SYMBOL_GPL(play_idle);
+
 void cpu_startup_entry(enum cpuhp_state state)
 {
 	/*
@@ -302,5 +341,6 @@ void cpu_startup_entry(enum cpuhp_state state)
 #endif
 	arch_cpu_idle_prepare();
 	cpuhp_online_idle(state);
-	cpu_idle_loop();
+	while (1)
+		do_idle();
 }