kernel_tick.md

Kernel code reading notes: timer tick

As Robert Love mentions in his book, a timer provides a mechanism for triggering an interrupt at a periodic rate. Such timer interrupts orchestrate OS scheduling, update resource bookkeeping, update the wall clock, etc. We are going to take a look into the Linux timer event handler.

https://github.com/torvalds/linux/blob/master/kernel/time/clockevents.c

/**
 * clockevents_register_device - register a clock event device
 * @dev:	device to register
 */
void clockevents_register_device(struct clock_event_device *dev)
{
	unsigned long flags;

	/* Initialize state to DETACHED */
	clockevent_set_state(dev, CLOCK_EVT_STATE_DETACHED);

	if (!dev->cpumask) {
		WARN_ON(num_possible_cpus() > 1);
		dev->cpumask = cpumask_of(smp_processor_id());
	}

	if (dev->cpumask == cpu_all_mask) {
		WARN(1, "%s cpumask == cpu_all_mask, using cpu_possible_mask instead\n",
		     dev->name);
		dev->cpumask = cpu_possible_mask;
	}

	raw_spin_lock_irqsave(&clockevents_lock, flags);

	list_add(&dev->list, &clockevent_devices);
	tick_check_new_device(dev);
	clockevents_notify_released();

	raw_spin_unlock_irqrestore(&clockevents_lock, flags);
}
EXPORT_SYMBOL_GPL(clockevents_register_device);

tick_check_new_device() is defined in https://github.com/torvalds/linux/blob/master/kernel/time/tick-common.c

/*
 * Check, if the new registered device should be used. Called with
 * clockevents_lock held and interrupts disabled.
 */
void tick_check_new_device(struct clock_event_device *newdev)
{
	struct clock_event_device *curdev;
	struct tick_device *td;
	int cpu;

	cpu = smp_processor_id();
	td = &per_cpu(tick_cpu_device, cpu);
	curdev = td->evtdev;

	/* cpu local device ? */
	if (!tick_check_percpu(curdev, newdev, cpu))
		goto out_bc;

	/* Preference decision */
	if (!tick_check_preferred(curdev, newdev))
		goto out_bc;

	if (!try_module_get(newdev->owner))
		return;

	/*
	 * Replace the eventually existing device by the new
	 * device. If the current device is the broadcast device, do
	 * not give it back to the clockevents layer !
	 */
	if (tick_is_broadcast_device(curdev)) {
		clockevents_shutdown(curdev);
		curdev = NULL;
	}
	clockevents_exchange_device(curdev, newdev);
	tick_setup_device(td, newdev, cpu, cpumask_of(cpu));
	if (newdev->features & CLOCK_EVT_FEAT_ONESHOT)
		tick_oneshot_notify();
	return;

out_bc:
	/*
	 * Can the new device be used as a broadcast device ?
	 */
	tick_install_broadcast_device(newdev);
}

tick_check_new_device() calls tick_setup_device() to setup per cpu tick interrupt handler. newdev is a clock_event_device struct which describes a timer event device. This chapter has more details on it. (You may also want to look at other chapters of this book. It explains kernel notions very well.). https://0xax.gitbooks.io/linux-insides/content/Timers/linux-timers-5.html

If the timer device is a broadcast device, the execution flow goes to out_bc, which calls tick_install_broadcast_device(). A broadcast timer device does not belong to a certain CPU. A non-broadcast timer will hibernate along with its CPU, and therefore, the CPU package needs an outside timer to wake it up. A broadcast timer will do it.

tick_setup_device() is defined as follow,

static void tick_setup_device(struct tick_device *td,
			      struct clock_event_device *newdev, int cpu,
			      const struct cpumask *cpumask)
{
	void (*handler)(struct clock_event_device *) = NULL;
	ktime_t next_event = 0;

	/*
	 * First device setup ?
	 */
	if (!td->evtdev) {
		/*
		 * If no cpu took the do_timer update, assign it to
		 * this cpu:
		 */
		if (tick_do_timer_cpu == TICK_DO_TIMER_BOOT) {
			tick_do_timer_cpu = cpu;

			tick_next_period = ktime_get();
			tick_period = NSEC_PER_SEC / HZ;
#ifdef CONFIG_NO_HZ_FULL
			/*
			 * The boot CPU may be nohz_full, in which case set
			 * tick_do_timer_boot_cpu so the first housekeeping
			 * secondary that comes up will take do_timer from
			 * us.
			 */
			if (tick_nohz_full_cpu(cpu))
				tick_do_timer_boot_cpu = cpu;

		} else if (tick_do_timer_boot_cpu != -1 &&
						!tick_nohz_full_cpu(cpu)) {
			tick_take_do_timer_from_boot();
			tick_do_timer_boot_cpu = -1;
			WARN_ON(tick_do_timer_cpu != cpu);
#endif
		}

		/*
		 * Startup in periodic mode first.
		 */
		td->mode = TICKDEV_MODE_PERIODIC;
	} else {
		handler = td->evtdev->event_handler;
		next_event = td->evtdev->next_event;
		td->evtdev->event_handler = clockevents_handle_noop;
	}

	td->evtdev = newdev;

	/*
	 * When the device is not per cpu, pin the interrupt to the
	 * current cpu:
	 */
	if (!cpumask_equal(newdev->cpumask, cpumask))
		irq_set_affinity(newdev->irq, cpumask);

	/*
	 * When global broadcasting is active, check if the current
	 * device is registered as a placeholder for broadcast mode.
	 * This allows us to handle this x86 misfeature in a generic
	 * way. This function also returns !=0 when we keep the
	 * current active broadcast state for this CPU.
	 */
	if (tick_device_uses_broadcast(newdev, cpu))
		return;

	if (td->mode == TICKDEV_MODE_PERIODIC)
		tick_setup_periodic(newdev, 0);
	else
		tick_setup_oneshot(newdev, handler, next_event);
}

It does a lot of checks, and eventually calls tick_setup_periodic():

void tick_setup_periodic(struct clock_event_device *dev, int broadcast)
{
	tick_set_periodic_handler(dev, broadcast);

	/* Broadcast setup ? */
	if (!tick_device_is_functional(dev))
		return;

	if ((dev->features & CLOCK_EVT_FEAT_PERIODIC) &&
	    !tick_broadcast_oneshot_active()) {
		clockevents_switch_state(dev, CLOCK_EVT_STATE_PERIODIC);
	} else {
		unsigned int seq;
		ktime_t next;

		do {
			seq = read_seqbegin(&jiffies_lock);
			next = tick_next_period;
		} while (read_seqretry(&jiffies_lock, seq));

		clockevents_switch_state(dev, CLOCK_EVT_STATE_ONESHOT);

		for (;;) {
			if (!clockevents_program_event(dev, next, false))
				return;
			next = ktime_add(next, tick_period);
		}
	}
}

tick_set_periodic_handler() simply writes the interrupt handler to dev->event_handler. The function is defined in https://github.com/torvalds/linux/blob/master/kernel/time/tick-broadcast.c

void tick_set_periodic_handler(struct clock_event_device *dev, int broadcast)
{
	if (!broadcast)
		dev->event_handler = tick_handle_periodic;
	else
		dev->event_handler = tick_handle_periodic_broadcast;
}

tick_handle_periodic() is defined in https://github.com/torvalds/linux/blob/master/kernel/time/tick-common.c

void tick_handle_periodic(struct clock_event_device *dev)
{
	int cpu = smp_processor_id();
	ktime_t next = dev->next_event;

	tick_periodic(cpu);

#if defined(CONFIG_HIGH_RES_TIMERS) || defined(CONFIG_NO_HZ_COMMON)
	/*
	 * The cpu might have transitioned to HIGHRES or NOHZ mode via
	 * update_process_times() -> run_local_timers() ->
	 * hrtimer_run_queues().
	 */
	if (dev->event_handler != tick_handle_periodic)
		return;
#endif

	if (!clockevent_state_oneshot(dev))
		return;
	for (;;) {
		/*
		 * Setup the next period for devices, which do not have
		 * periodic mode:
		 */
		next = ktime_add(next, tick_period);

		if (!clockevents_program_event(dev, next, false))
			return;
		/*
		 * Have to be careful here. If we're in oneshot mode,
		 * before we call tick_periodic() in a loop, we need
		 * to be sure we're using a real hardware clocksource.
		 * Otherwise we could get trapped in an infinite
		 * loop, as the tick_periodic() increments jiffies,
		 * which then will increment time, possibly causing
		 * the loop to trigger again and again.
		 */
		if (timekeeping_valid_for_hres())
			tick_periodic(cpu);
	}
}

Upon every timer tick interrupt, tick_handle_periodic() calls tick_periodic(). If the device is a fixed timer, it will simply return. Otherwise, if the timer is a "oneshot" device, it sets up the next timer interval (i.e. when the next tick interrupt should arrive).

In the same file, tick_periodic is defined as,

static void tick_periodic(int cpu)
{
	if (tick_do_timer_cpu == cpu) {
		write_seqlock(&jiffies_lock);

		/* Keep track of the next tick event */
		tick_next_period = ktime_add(tick_next_period, tick_period);

		do_timer(1);
		write_sequnlock(&jiffies_lock);
		update_wall_time();
	}

	update_process_times(user_mode(get_irq_regs()));
	profile_tick(CPU_PROFILING);
}

do_timer updates the jiffies counter and kernel load statistics. It is defined in https://github.com/torvalds/linux/blob/master/kernel/time/timekeeping.c

void do_timer(unsigned long ticks)
{
	jiffies_64 += ticks;
	calc_global_load(ticks);
}

update_process_times() is defined in https://github.com/torvalds/linux/blob/master/kernel/time/timer.c It eats a tick from the current process for fair scheduling purposes.

/*
 * Called from the timer interrupt handler to charge one tick to the current
 * process.  user_tick is 1 if the tick is user time, 0 for system.
 */
void update_process_times(int user_tick)
{
	struct task_struct *p = current;

	/* Note: this timer irq context must be accounted for as well. */
	account_process_tick(p, user_tick);
	run_local_timers();
	rcu_sched_clock_irq(user_tick);
#ifdef CONFIG_IRQ_WORK
	if (in_irq())
		irq_work_tick();
#endif
	scheduler_tick();
	if (IS_ENABLED(CONFIG_POSIX_TIMERS))
		run_posix_cpu_timers();
}