[mirror_ubuntu-jammy-kernel.git] / kernel / time / tick-common.c

/*
 * linux/kernel/time/tick-common.c
 *
 * This file contains the base functions to manage periodic tick
 * related events.
 *
 * Copyright(C) 2005-2006, Thomas Gleixner <tglx@linutronix.de>
 * Copyright(C) 2005-2007, Red Hat, Inc., Ingo Molnar
 * Copyright(C) 2006-2007, Timesys Corp., Thomas Gleixner
 *
 * This code is licenced under the GPL version 2. For details see
 * kernel-base/COPYING.
 */
#include <linux/cpu.h>
#include <linux/err.h>
#include <linux/hrtimer.h>
#include <linux/interrupt.h>
#include <linux/percpu.h>
#include <linux/profile.h>
#include <linux/sched.h>
#include <linux/module.h>

#include <asm/irq_regs.h>

#include "tick-internal.h"

/*
 * Tick devices
 */
DEFINE_PER_CPU(struct tick_device, tick_cpu_device);
/*
 * Tick next event: keeps track of the tick time
 */
ktime_t tick_next_period;
ktime_t tick_period;

/*
 * tick_do_timer_cpu is a timer core internal variable which holds the CPU NR
 * which is responsible for calling do_timer(), i.e. the timekeeping stuff. This
 * variable has two functions:
 *
 * 1) Prevent a thundering herd issue of a gazillion of CPUs trying to grab the
 *    timekeeping lock all at once. Only the CPU which is assigned to do the
 *    update is handling it.
 *
 * 2) Hand off the duty in the NOHZ idle case by setting the value to
 *    TICK_DO_TIMER_NONE, i.e. a non existing CPU. So the next cpu which looks
 *    at it will take over and keep the time keeping alive.  The handover
 *    procedure also covers cpu hotplug.
 */
int tick_do_timer_cpu __read_mostly = TICK_DO_TIMER_BOOT;

/*
 * Debugging: see timer_list.c
 */
struct tick_device *tick_get_device(int cpu)
{
	return &per_cpu(tick_cpu_device, cpu);
}

/**
 * tick_is_oneshot_available - check for a oneshot capable event device
 */
int tick_is_oneshot_available(void)
{
	struct clock_event_device *dev = __this_cpu_read(tick_cpu_device.evtdev);

	if (!dev || !(dev->features & CLOCK_EVT_FEAT_ONESHOT))
		return 0;
	if (!(dev->features & CLOCK_EVT_FEAT_C3STOP))
		return 1;
	return tick_broadcast_oneshot_available();
}

/*
 * Periodic tick
 */
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);
}

/*
 * Event handler for periodic ticks
 */
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);
	}
}

/*
 * Setup the device for a periodic tick
 */
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 long 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);
		}
	}
}

/*
 * Setup the tick device
 */
static void tick_setup_device(struct tick_device *td,
			      struct clock_event_device *newdev, int cpu,
			      const struct cpumask *cpumask)
{
	ktime_t next_event;
	void (*handler)(struct clock_event_device *) = NULL;

	/*
	 * 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) {
			if (!tick_nohz_full_cpu(cpu))
				tick_do_timer_cpu = cpu;
			else
				tick_do_timer_cpu = TICK_DO_TIMER_NONE;
			tick_next_period = ktime_get();
			tick_period = ktime_set(0, NSEC_PER_SEC / HZ);
		}

		/*
		 * 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);
}

void tick_install_replacement(struct clock_event_device *newdev)
{
	struct tick_device *td = this_cpu_ptr(&tick_cpu_device);
	int cpu = smp_processor_id();

	clockevents_exchange_device(td->evtdev, newdev);
	tick_setup_device(td, newdev, cpu, cpumask_of(cpu));
	if (newdev->features & CLOCK_EVT_FEAT_ONESHOT)
		tick_oneshot_notify();
}

static bool tick_check_percpu(struct clock_event_device *curdev,
			      struct clock_event_device *newdev, int cpu)
{
	if (!cpumask_test_cpu(cpu, newdev->cpumask))
		return false;
	if (cpumask_equal(newdev->cpumask, cpumask_of(cpu)))
		return true;
	/* Check if irq affinity can be set */
	if (newdev->irq >= 0 && !irq_can_set_affinity(newdev->irq))
		return false;
	/* Prefer an existing cpu local device */
	if (curdev && cpumask_equal(curdev->cpumask, cpumask_of(cpu)))
		return false;
	return true;
}

static bool tick_check_preferred(struct clock_event_device *curdev,
				 struct clock_event_device *newdev)
{
	/* Prefer oneshot capable device */
	if (!(newdev->features & CLOCK_EVT_FEAT_ONESHOT)) {
		if (curdev && (curdev->features & CLOCK_EVT_FEAT_ONESHOT))
			return false;
		if (tick_oneshot_mode_active())
			return false;
	}

	/*
	 * Use the higher rated one, but prefer a CPU local device with a lower
	 * rating than a non-CPU local device
	 */
	return !curdev ||
		newdev->rating > curdev->rating ||
	       !cpumask_equal(curdev->cpumask, newdev->cpumask);
}

/*
 * Check whether the new device is a better fit than curdev. curdev
 * can be NULL !
 */
bool tick_check_replacement(struct clock_event_device *curdev,
			    struct clock_event_device *newdev)
{
	if (!tick_check_percpu(curdev, newdev, smp_processor_id()))
		return false;

	return tick_check_preferred(curdev, newdev);
}

/*
 * 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();
	if (!cpumask_test_cpu(cpu, newdev->cpumask))
		goto out_bc;

	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);
}

#ifdef CONFIG_HOTPLUG_CPU
/*
 * Transfer the do_timer job away from a dying cpu.
 *
 * Called with interrupts disabled. Not locking required. If
 * tick_do_timer_cpu is owned by this cpu, nothing can change it.
 */
void tick_handover_do_timer(void)
{
	if (tick_do_timer_cpu == smp_processor_id()) {
		int cpu = cpumask_first(cpu_online_mask);

		tick_do_timer_cpu = (cpu < nr_cpu_ids) ? cpu :
			TICK_DO_TIMER_NONE;
	}
}

/*
 * Shutdown an event device on a given cpu:
 *
 * This is called on a life CPU, when a CPU is dead. So we cannot
 * access the hardware device itself.
 * We just set the mode and remove it from the lists.
 */
void tick_shutdown(unsigned int cpu)
{
	struct tick_device *td = &per_cpu(tick_cpu_device, cpu);
	struct clock_event_device *dev = td->evtdev;

	td->mode = TICKDEV_MODE_PERIODIC;
	if (dev) {
		/*
		 * Prevent that the clock events layer tries to call
		 * the set mode function!
		 */
		dev->state = CLOCK_EVT_STATE_DETACHED;
		dev->mode = CLOCK_EVT_MODE_UNUSED;
		clockevents_exchange_device(dev, NULL);
		dev->event_handler = clockevents_handle_noop;
		td->evtdev = NULL;
	}
}
#endif

/**
 * tick_suspend_local - Suspend the local tick device
 *
 * Called from the local cpu for freeze with interrupts disabled.
 *
 * No locks required. Nothing can change the per cpu device.
 */
void tick_suspend_local(void)
{
	struct tick_device *td = this_cpu_ptr(&tick_cpu_device);

	clockevents_shutdown(td->evtdev);
}

/**
 * tick_resume_local - Resume the local tick device
 *
 * Called from the local CPU for unfreeze or XEN resume magic.
 *
 * No locks required. Nothing can change the per cpu device.
 */
void tick_resume_local(void)
{
	struct tick_device *td = this_cpu_ptr(&tick_cpu_device);
	bool broadcast = tick_resume_check_broadcast();

	clockevents_tick_resume(td->evtdev);
	if (!broadcast) {
		if (td->mode == TICKDEV_MODE_PERIODIC)
			tick_setup_periodic(td->evtdev, 0);
		else
			tick_resume_oneshot();
	}
}

/**
 * tick_suspend - Suspend the tick and the broadcast device
 *
 * Called from syscore_suspend() via timekeeping_suspend with only one
 * CPU online and interrupts disabled or from tick_unfreeze() under
 * tick_freeze_lock.
 *
 * No locks required. Nothing can change the per cpu device.
 */
void tick_suspend(void)
{
	tick_suspend_local();
	tick_suspend_broadcast();
}

/**
 * tick_resume - Resume the tick and the broadcast device
 *
 * Called from syscore_resume() via timekeeping_resume with only one
 * CPU online and interrupts disabled.
 *
 * No locks required. Nothing can change the per cpu device.
 */
void tick_resume(void)
{
	tick_resume_broadcast();
	tick_resume_local();
}

static DEFINE_RAW_SPINLOCK(tick_freeze_lock);
static unsigned int tick_freeze_depth;

/**
 * tick_freeze - Suspend the local tick and (possibly) timekeeping.
 *
 * Check if this is the last online CPU executing the function and if so,
 * suspend timekeeping.  Otherwise suspend the local tick.
 *
 * Call with interrupts disabled.  Must be balanced with %tick_unfreeze().
 * Interrupts must not be enabled before the subsequent %tick_unfreeze().
 */
void tick_freeze(void)
{
	raw_spin_lock(&tick_freeze_lock);

	tick_freeze_depth++;
	if (tick_freeze_depth == num_online_cpus())
		timekeeping_suspend();
	else
		tick_suspend_local();

	raw_spin_unlock(&tick_freeze_lock);
}

/**
 * tick_unfreeze - Resume the local tick and (possibly) timekeeping.
 *
 * Check if this is the first CPU executing the function and if so, resume
 * timekeeping.  Otherwise resume the local tick.
 *
 * Call with interrupts disabled.  Must be balanced with %tick_freeze().
 * Interrupts must not be enabled after the preceding %tick_freeze().
 */
void tick_unfreeze(void)
{
	raw_spin_lock(&tick_freeze_lock);

	if (tick_freeze_depth == num_online_cpus())
		timekeeping_resume();
	else
		tick_resume_local();

	tick_freeze_depth--;

	raw_spin_unlock(&tick_freeze_lock);
}

/**
 * tick_init - initialize the tick control
 */
void __init tick_init(void)
{
	tick_broadcast_init();
	tick_nohz_init();
}
Commit	Line	Data
906568c9 TG	1	/*
	2	* linux/kernel/time/tick-common.c
	3	*
	4	* This file contains the base functions to manage periodic tick
	5	* related events.
	6	*
	7	* Copyright(C) 2005-2006, Thomas Gleixner <tglx@linutronix.de>
	8	* Copyright(C) 2005-2007, Red Hat, Inc., Ingo Molnar
	9	* Copyright(C) 2006-2007, Timesys Corp., Thomas Gleixner
	10	*
	11	* This code is licenced under the GPL version 2. For details see
	12	* kernel-base/COPYING.
	13	*/
	14	#include <linux/cpu.h>
	15	#include <linux/err.h>
	16	#include <linux/hrtimer.h>
d7b90689	17	#include <linux/interrupt.h>
906568c9 TG	18	#include <linux/percpu.h>
	19	#include <linux/profile.h>
	20	#include <linux/sched.h>
ccf33d68	21	#include <linux/module.h>
906568c9	22
d7b90689 RK	23	#include <asm/irq_regs.h>
d7b90689 RK	24
f8381cba TG	25	#include "tick-internal.h"
f8381cba TG	26
906568c9 TG	27	/*
	28	* Tick devices
	29	*/
f8381cba	30	DEFINE_PER_CPU(struct tick_device, tick_cpu_device);
906568c9 TG	31	/*
	32	* Tick next event: keeps track of the tick time
	33	*/
f8381cba TG	34	ktime_t tick_next_period;
f8381cba TG	35	ktime_t tick_period;
050ded1b AM	36
	37	/*
	38	* tick_do_timer_cpu is a timer core internal variable which holds the CPU NR
	39	* which is responsible for calling do_timer(), i.e. the timekeeping stuff. This
	40	* variable has two functions:
	41	*
	42	* 1) Prevent a thundering herd issue of a gazillion of CPUs trying to grab the
	43	* timekeeping lock all at once. Only the CPU which is assigned to do the
	44	* update is handling it.
	45	*
	46	* 2) Hand off the duty in the NOHZ idle case by setting the value to
	47	* TICK_DO_TIMER_NONE, i.e. a non existing CPU. So the next cpu which looks
	48	* at it will take over and keep the time keeping alive. The handover
	49	* procedure also covers cpu hotplug.
	50	*/
6441402b	51	int tick_do_timer_cpu __read_mostly = TICK_DO_TIMER_BOOT;
906568c9	52
289f480a IM	53	/*
	54	* Debugging: see timer_list.c
	55	*/
	56	struct tick_device *tick_get_device(int cpu)
	57	{
	58	return &per_cpu(tick_cpu_device, cpu);
	59	}
	60
79bf2bb3 TG	61	/**
	62	* tick_is_oneshot_available - check for a oneshot capable event device
	63	*/
	64	int tick_is_oneshot_available(void)
	65	{
909ea964	66	struct clock_event_device *dev = __this_cpu_read(tick_cpu_device.evtdev);
79bf2bb3	67
3a142a06 TG	68	if (!dev \|\| !(dev->features & CLOCK_EVT_FEAT_ONESHOT))
	69	return 0;
	70	if (!(dev->features & CLOCK_EVT_FEAT_C3STOP))
	71	return 1;
	72	return tick_broadcast_oneshot_available();
79bf2bb3 TG	73	}
79bf2bb3 TG	74
906568c9 TG	75	/*
	76	* Periodic tick
	77	*/
	78	static void tick_periodic(int cpu)
	79	{
	80	if (tick_do_timer_cpu == cpu) {
d6ad4187	81	write_seqlock(&jiffies_lock);
906568c9 TG	82
	83	/* Keep track of the next tick event */
	84	tick_next_period = ktime_add(tick_next_period, tick_period);
	85
	86	do_timer(1);
d6ad4187	87	write_sequnlock(&jiffies_lock);
47a1b796	88	update_wall_time();
906568c9 TG	89	}
	90
	91	update_process_times(user_mode(get_irq_regs()));
	92	profile_tick(CPU_PROFILING);
	93	}
	94
	95	/*
	96	* Event handler for periodic ticks
	97	*/
	98	void tick_handle_periodic(struct clock_event_device *dev)
	99	{
	100	int cpu = smp_processor_id();
b97f0291	101	ktime_t next = dev->next_event;
906568c9 TG	102
	103	tick_periodic(cpu);
	104
c6eb3f70 TG	105	#if defined(CONFIG_HIGH_RES_TIMERS) \|\| defined(CONFIG_NO_HZ_COMMON)
	106	/*
	107	* The cpu might have transitioned to HIGHRES or NOHZ mode via
	108	* update_process_times() -> run_local_timers() ->
	109	* hrtimer_run_queues().
	110	*/
	111	if (dev->event_handler != tick_handle_periodic)
	112	return;
	113	#endif
	114
472c4a94	115	if (!clockevent_state_oneshot(dev))
906568c9	116	return;
906568c9	117	for (;;) {
b97f0291 VK	118	/*
	119	* Setup the next period for devices, which do not have
	120	* periodic mode:
	121	*/
	122	next = ktime_add(next, tick_period);
	123
d1748302	124	if (!clockevents_program_event(dev, next, false))
906568c9	125	return;
74a03b69 JS	126	/*
	127	* Have to be careful here. If we're in oneshot mode,
	128	* before we call tick_periodic() in a loop, we need
	129	* to be sure we're using a real hardware clocksource.
	130	* Otherwise we could get trapped in an infinite
	131	* loop, as the tick_periodic() increments jiffies,
cacb3c76	132	* which then will increment time, possibly causing
74a03b69 JS	133	* the loop to trigger again and again.
	134	*/
	135	if (timekeeping_valid_for_hres())
	136	tick_periodic(cpu);
906568c9 TG	137	}
	138	}
	139
	140	/*
	141	* Setup the device for a periodic tick
	142	*/
f8381cba	143	void tick_setup_periodic(struct clock_event_device *dev, int broadcast)
906568c9	144	{
f8381cba TG	145	tick_set_periodic_handler(dev, broadcast);
	146
	147	/* Broadcast setup ? */
	148	if (!tick_device_is_functional(dev))
	149	return;
906568c9	150
27ce4cb4 TG	151	if ((dev->features & CLOCK_EVT_FEAT_PERIODIC) &&
27ce4cb4 TG	152	!tick_broadcast_oneshot_active()) {
d7eb231c	153	clockevents_switch_state(dev, CLOCK_EVT_STATE_PERIODIC);
906568c9 TG	154	} else {
	155	unsigned long seq;
	156	ktime_t next;
	157
	158	do {
d6ad4187	159	seq = read_seqbegin(&jiffies_lock);
906568c9	160	next = tick_next_period;
d6ad4187	161	} while (read_seqretry(&jiffies_lock, seq));
906568c9	162
d7eb231c	163	clockevents_switch_state(dev, CLOCK_EVT_STATE_ONESHOT);
906568c9 TG	164
906568c9 TG	165	for (;;) {
d1748302	166	if (!clockevents_program_event(dev, next, false))
906568c9 TG	167	return;
	168	next = ktime_add(next, tick_period);
	169	}
	170	}
	171	}
	172
	173	/*
	174	* Setup the tick device
	175	*/
	176	static void tick_setup_device(struct tick_device *td,
	177	struct clock_event_device *newdev, int cpu,
0de26520	178	const struct cpumask *cpumask)
906568c9 TG	179	{
	180	ktime_t next_event;
	181	void (handler)(struct clock_event_device ) = NULL;
	182
	183	/*
	184	* First device setup ?
	185	*/
	186	if (!td->evtdev) {
	187	/*
	188	* If no cpu took the do_timer update, assign it to
	189	* this cpu:
	190	*/
6441402b	191	if (tick_do_timer_cpu == TICK_DO_TIMER_BOOT) {
c5bfece2	192	if (!tick_nohz_full_cpu(cpu))
a382bf93 FW	193	tick_do_timer_cpu = cpu;
	194	else
	195	tick_do_timer_cpu = TICK_DO_TIMER_NONE;
906568c9 TG	196	tick_next_period = ktime_get();
	197	tick_period = ktime_set(0, NSEC_PER_SEC / HZ);
	198	}
	199
	200	/*
	201	* Startup in periodic mode first.
	202	*/
	203	td->mode = TICKDEV_MODE_PERIODIC;
	204	} else {
	205	handler = td->evtdev->event_handler;
	206	next_event = td->evtdev->next_event;
7c1e7689	207	td->evtdev->event_handler = clockevents_handle_noop;
906568c9 TG	208	}
	209
	210	td->evtdev = newdev;
	211
	212	/*
	213	* When the device is not per cpu, pin the interrupt to the
	214	* current cpu:
	215	*/
320ab2b0	216	if (!cpumask_equal(newdev->cpumask, cpumask))
0de26520	217	irq_set_affinity(newdev->irq, cpumask);
906568c9	218
f8381cba TG	219	/*
	220	* When global broadcasting is active, check if the current
	221	* device is registered as a placeholder for broadcast mode.
	222	* This allows us to handle this x86 misfeature in a generic
07bd1172 TG	223	* way. This function also returns !=0 when we keep the
07bd1172 TG	224	* current active broadcast state for this CPU.
f8381cba TG	225	*/
	226	if (tick_device_uses_broadcast(newdev, cpu))
	227	return;
	228
906568c9 TG	229	if (td->mode == TICKDEV_MODE_PERIODIC)
906568c9 TG	230	tick_setup_periodic(newdev, 0);
79bf2bb3 TG	231	else
79bf2bb3 TG	232	tick_setup_oneshot(newdev, handler, next_event);
906568c9 TG	233	}
906568c9 TG	234
03e13cf5 TG	235	void tick_install_replacement(struct clock_event_device *newdev)
03e13cf5 TG	236	{
22127e93	237	struct tick_device *td = this_cpu_ptr(&tick_cpu_device);
03e13cf5 TG	238	int cpu = smp_processor_id();
	239
	240	clockevents_exchange_device(td->evtdev, newdev);
	241	tick_setup_device(td, newdev, cpu, cpumask_of(cpu));
	242	if (newdev->features & CLOCK_EVT_FEAT_ONESHOT)
	243	tick_oneshot_notify();
	244	}
	245
45cb8e01 TG	246	static bool tick_check_percpu(struct clock_event_device *curdev,
	247	struct clock_event_device *newdev, int cpu)
	248	{
	249	if (!cpumask_test_cpu(cpu, newdev->cpumask))
	250	return false;
	251	if (cpumask_equal(newdev->cpumask, cpumask_of(cpu)))
	252	return true;
	253	/* Check if irq affinity can be set */
	254	if (newdev->irq >= 0 && !irq_can_set_affinity(newdev->irq))
	255	return false;
	256	/* Prefer an existing cpu local device */
	257	if (curdev && cpumask_equal(curdev->cpumask, cpumask_of(cpu)))
	258	return false;
	259	return true;
	260	}
	261
	262	static bool tick_check_preferred(struct clock_event_device *curdev,
	263	struct clock_event_device *newdev)
	264	{
	265	/* Prefer oneshot capable device */
	266	if (!(newdev->features & CLOCK_EVT_FEAT_ONESHOT)) {
	267	if (curdev && (curdev->features & CLOCK_EVT_FEAT_ONESHOT))
	268	return false;
	269	if (tick_oneshot_mode_active())
	270	return false;
	271	}
	272
70e5975d SB	273	/*
	274	* Use the higher rated one, but prefer a CPU local device with a lower
	275	* rating than a non-CPU local device
	276	*/
	277	return !curdev \|\|
	278	newdev->rating > curdev->rating \|\|
	279	!cpumask_equal(curdev->cpumask, newdev->cpumask);
45cb8e01 TG	280	}
45cb8e01 TG	281
03e13cf5 TG	282	/*
	283	* Check whether the new device is a better fit than curdev. curdev
	284	* can be NULL !
	285	*/
	286	bool tick_check_replacement(struct clock_event_device *curdev,
	287	struct clock_event_device *newdev)
	288	{
521c4299	289	if (!tick_check_percpu(curdev, newdev, smp_processor_id()))
03e13cf5 TG	290	return false;
	291
	292	return tick_check_preferred(curdev, newdev);
	293	}
	294
906568c9	295	/*
7126cac4 TG	296	* Check, if the new registered device should be used. Called with
7126cac4 TG	297	* clockevents_lock held and interrupts disabled.
906568c9	298	*/
7172a286	299	void tick_check_new_device(struct clock_event_device *newdev)
906568c9 TG	300	{
	301	struct clock_event_device *curdev;
	302	struct tick_device *td;
7172a286	303	int cpu;
906568c9 TG	304
906568c9 TG	305	cpu = smp_processor_id();
320ab2b0	306	if (!cpumask_test_cpu(cpu, newdev->cpumask))
4a93232d	307	goto out_bc;
906568c9 TG	308
	309	td = &per_cpu(tick_cpu_device, cpu);
	310	curdev = td->evtdev;
906568c9 TG	311
906568c9 TG	312	/* cpu local device ? */
45cb8e01 TG	313	if (!tick_check_percpu(curdev, newdev, cpu))
45cb8e01 TG	314	goto out_bc;
906568c9	315
45cb8e01 TG	316	/* Preference decision */
	317	if (!tick_check_preferred(curdev, newdev))
	318	goto out_bc;
906568c9	319
ccf33d68 TG	320	if (!try_module_get(newdev->owner))
	321	return;
	322
906568c9 TG	323	/*
906568c9 TG	324	* Replace the eventually existing device by the new
f8381cba TG	325	* device. If the current device is the broadcast device, do
f8381cba TG	326	* not give it back to the clockevents layer !
906568c9	327	*/
f8381cba	328	if (tick_is_broadcast_device(curdev)) {
2344abbc	329	clockevents_shutdown(curdev);
f8381cba TG	330	curdev = NULL;
f8381cba TG	331	}
906568c9	332	clockevents_exchange_device(curdev, newdev);
6b954823	333	tick_setup_device(td, newdev, cpu, cpumask_of(cpu));
79bf2bb3 TG	334	if (newdev->features & CLOCK_EVT_FEAT_ONESHOT)
79bf2bb3 TG	335	tick_oneshot_notify();
7172a286	336	return;
f8381cba TG	337
	338	out_bc:
	339	/*
	340	* Can the new device be used as a broadcast device ?
	341	*/
7172a286	342	tick_install_broadcast_device(newdev);
906568c9 TG	343	}
906568c9 TG	344
52c063d1	345	#ifdef CONFIG_HOTPLUG_CPU
94df7de0 SD	346	/*
	347	* Transfer the do_timer job away from a dying cpu.
	348	*
52c063d1 TG	349	* Called with interrupts disabled. Not locking required. If
52c063d1 TG	350	* tick_do_timer_cpu is owned by this cpu, nothing can change it.
94df7de0	351	*/
52c063d1	352	void tick_handover_do_timer(void)
94df7de0	353	{
52c063d1	354	if (tick_do_timer_cpu == smp_processor_id()) {
94df7de0 SD	355	int cpu = cpumask_first(cpu_online_mask);
	356
	357	tick_do_timer_cpu = (cpu < nr_cpu_ids) ? cpu :
	358	TICK_DO_TIMER_NONE;
	359	}
	360	}
	361
906568c9 TG	362	/*
	363	* Shutdown an event device on a given cpu:
	364	*
	365	* This is called on a life CPU, when a CPU is dead. So we cannot
	366	* access the hardware device itself.
	367	* We just set the mode and remove it from the lists.
	368	*/
a49b116d	369	void tick_shutdown(unsigned int cpu)
906568c9	370	{
a49b116d	371	struct tick_device *td = &per_cpu(tick_cpu_device, cpu);
906568c9	372	struct clock_event_device *dev = td->evtdev;
906568c9	373
906568c9 TG	374	td->mode = TICKDEV_MODE_PERIODIC;
	375	if (dev) {
	376	/*
	377	* Prevent that the clock events layer tries to call
	378	* the set mode function!
	379	*/
77e32c89	380	dev->state = CLOCK_EVT_STATE_DETACHED;
906568c9 TG	381	dev->mode = CLOCK_EVT_MODE_UNUSED;
906568c9 TG	382	clockevents_exchange_device(dev, NULL);
6f7a05d7	383	dev->event_handler = clockevents_handle_noop;
906568c9 TG	384	td->evtdev = NULL;
906568c9 TG	385	}
906568c9	386	}
a49b116d	387	#endif
906568c9	388
4ffee521	389	/**
f46481d0	390	* tick_suspend_local - Suspend the local tick device
4ffee521	391	*
f46481d0	392	* Called from the local cpu for freeze with interrupts disabled.
4ffee521 TG	393	*
	394	* No locks required. Nothing can change the per cpu device.
	395	*/
7270d11c	396	void tick_suspend_local(void)
6321dd60	397	{
22127e93	398	struct tick_device *td = this_cpu_ptr(&tick_cpu_device);
6321dd60	399
2344abbc	400	clockevents_shutdown(td->evtdev);
6321dd60 TG	401	}
6321dd60 TG	402
4ffee521	403	/**
f46481d0	404	* tick_resume_local - Resume the local tick device
4ffee521	405	*
f46481d0	406	* Called from the local CPU for unfreeze or XEN resume magic.
4ffee521 TG	407	*
	408	* No locks required. Nothing can change the per cpu device.
	409	*/
f46481d0	410	void tick_resume_local(void)
6321dd60	411	{
f46481d0 TG	412	struct tick_device *td = this_cpu_ptr(&tick_cpu_device);
f46481d0 TG	413	bool broadcast = tick_resume_check_broadcast();
6321dd60	414
554ef387	415	clockevents_tick_resume(td->evtdev);
18de5bc4 TG	416	if (!broadcast) {
	417	if (td->mode == TICKDEV_MODE_PERIODIC)
	418	tick_setup_periodic(td->evtdev, 0);
	419	else
	420	tick_resume_oneshot();
	421	}
6321dd60 TG	422	}
6321dd60 TG	423
f46481d0 TG	424	/**
	425	* tick_suspend - Suspend the tick and the broadcast device
	426	*
	427	* Called from syscore_suspend() via timekeeping_suspend with only one
	428	* CPU online and interrupts disabled or from tick_unfreeze() under
	429	* tick_freeze_lock.
	430	*
	431	* No locks required. Nothing can change the per cpu device.
	432	*/
	433	void tick_suspend(void)
	434	{
	435	tick_suspend_local();
	436	tick_suspend_broadcast();
	437	}
	438
	439	/**
	440	* tick_resume - Resume the tick and the broadcast device
	441	*
	442	* Called from syscore_resume() via timekeeping_resume with only one
	443	* CPU online and interrupts disabled.
	444	*
	445	* No locks required. Nothing can change the per cpu device.
	446	*/
	447	void tick_resume(void)
	448	{
	449	tick_resume_broadcast();
	450	tick_resume_local();
	451	}
	452
124cf911 RW	453	static DEFINE_RAW_SPINLOCK(tick_freeze_lock);
	454	static unsigned int tick_freeze_depth;
	455
	456	/**
	457	* tick_freeze - Suspend the local tick and (possibly) timekeeping.
	458	*
	459	* Check if this is the last online CPU executing the function and if so,
	460	* suspend timekeeping. Otherwise suspend the local tick.
	461	*
	462	* Call with interrupts disabled. Must be balanced with %tick_unfreeze().
	463	* Interrupts must not be enabled before the subsequent %tick_unfreeze().
	464	*/
	465	void tick_freeze(void)
	466	{
	467	raw_spin_lock(&tick_freeze_lock);
	468
	469	tick_freeze_depth++;
def74708	470	if (tick_freeze_depth == num_online_cpus())
124cf911	471	timekeeping_suspend();
def74708	472	else
f46481d0	473	tick_suspend_local();
124cf911 RW	474
	475	raw_spin_unlock(&tick_freeze_lock);
	476	}
	477
	478	/**
	479	* tick_unfreeze - Resume the local tick and (possibly) timekeeping.
	480	*
	481	* Check if this is the first CPU executing the function and if so, resume
	482	* timekeeping. Otherwise resume the local tick.
	483	*
	484	* Call with interrupts disabled. Must be balanced with %tick_freeze().
	485	* Interrupts must not be enabled after the preceding %tick_freeze().
	486	*/
	487	void tick_unfreeze(void)
	488	{
	489	raw_spin_lock(&tick_freeze_lock);
	490
	491	if (tick_freeze_depth == num_online_cpus())
	492	timekeeping_resume();
	493	else
422fe750	494	tick_resume_local();
124cf911 RW	495
	496	tick_freeze_depth--;
	497
	498	raw_spin_unlock(&tick_freeze_lock);
	499	}
	500
906568c9 TG	501	/**
906568c9 TG	502	* tick_init - initialize the tick control
906568c9 TG	503	*/
	504	void __init tick_init(void)
	505	{
b352bc1c	506	tick_broadcast_init();
a80e49e2	507	tick_nohz_init();
906568c9	508	}