[deliverable/linux.git] / drivers / cpuidle / cpuidle.c

/*
 * cpuidle.c - core cpuidle infrastructure
 *
 * (C) 2006-2007 Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
 *               Shaohua Li <shaohua.li@intel.com>
 *               Adam Belay <abelay@novell.com>
 *
 * This code is licenced under the GPL.
 */

#include <linux/kernel.h>
#include <linux/mutex.h>
#include <linux/sched.h>
#include <linux/notifier.h>
#include <linux/pm_qos.h>
#include <linux/cpu.h>
#include <linux/cpuidle.h>
#include <linux/ktime.h>
#include <linux/hrtimer.h>
#include <linux/module.h>
#include <trace/events/power.h>

#include "cpuidle.h"

DEFINE_PER_CPU(struct cpuidle_device *, cpuidle_devices);

DEFINE_MUTEX(cpuidle_lock);
LIST_HEAD(cpuidle_detected_devices);

static int enabled_devices;
static int off __read_mostly;
static int initialized __read_mostly;

int cpuidle_disabled(void)
{
	return off;
}
void disable_cpuidle(void)
{
	off = 1;
}

static int __cpuidle_register_device(struct cpuidle_device *dev);

static inline int cpuidle_enter(struct cpuidle_device *dev,
				struct cpuidle_driver *drv, int index)
{
	struct cpuidle_state *target_state = &drv->states[index];
	return target_state->enter(dev, drv, index);
}

static inline int cpuidle_enter_tk(struct cpuidle_device *dev,
			       struct cpuidle_driver *drv, int index)
{
	return cpuidle_wrap_enter(dev, drv, index, cpuidle_enter);
}

typedef int (*cpuidle_enter_t)(struct cpuidle_device *dev,
			       struct cpuidle_driver *drv, int index);

static cpuidle_enter_t cpuidle_enter_ops;

/**
 * cpuidle_play_dead - cpu off-lining
 *
 * Returns in case of an error or no driver
 */
int cpuidle_play_dead(void)
{
	struct cpuidle_device *dev = __this_cpu_read(cpuidle_devices);
	struct cpuidle_driver *drv = cpuidle_get_driver();
	int i, dead_state = -1;
	int power_usage = -1;

	if (!drv)
		return -ENODEV;

	/* Find lowest-power state that supports long-term idle */
	for (i = CPUIDLE_DRIVER_STATE_START; i < drv->state_count; i++) {
		struct cpuidle_state *s = &drv->states[i];

		if (s->power_usage < power_usage && s->enter_dead) {
			power_usage = s->power_usage;
			dead_state = i;
		}
	}

	if (dead_state != -1)
		return drv->states[dead_state].enter_dead(dev, dead_state);

	return -ENODEV;
}

/**
 * cpuidle_idle_call - the main idle loop
 *
 * NOTE: no locks or semaphores should be used here
 * return non-zero on failure
 */
int cpuidle_idle_call(void)
{
	struct cpuidle_device *dev = __this_cpu_read(cpuidle_devices);
	struct cpuidle_driver *drv = cpuidle_get_driver();
	int next_state, entered_state;

	if (off)
		return -ENODEV;

	if (!initialized)
		return -ENODEV;

	/* check if the device is ready */
	if (!dev || !dev->enabled)
		return -EBUSY;

#if 0
	/* shows regressions, re-enable for 2.6.29 */
	/*
	 * run any timers that can be run now, at this point
	 * before calculating the idle duration etc.
	 */
	hrtimer_peek_ahead_timers();
#endif

	/* ask the governor for the next state */
	next_state = cpuidle_curr_governor->select(drv, dev);
	if (need_resched()) {
		local_irq_enable();
		return 0;
	}

	trace_power_start_rcuidle(POWER_CSTATE, next_state, dev->cpu);
	trace_cpu_idle_rcuidle(next_state, dev->cpu);

	entered_state = cpuidle_enter_ops(dev, drv, next_state);

	trace_power_end_rcuidle(dev->cpu);
	trace_cpu_idle_rcuidle(PWR_EVENT_EXIT, dev->cpu);

	if (entered_state >= 0) {
		/* Update cpuidle counters */
		/* This can be moved to within driver enter routine
		 * but that results in multiple copies of same code.
		 */
		dev->states_usage[entered_state].time +=
				(unsigned long long)dev->last_residency;
		dev->states_usage[entered_state].usage++;
	} else {
		dev->last_residency = 0;
	}

	/* give the governor an opportunity to reflect on the outcome */
	if (cpuidle_curr_governor->reflect)
		cpuidle_curr_governor->reflect(dev, entered_state);

	return 0;
}

/**
 * cpuidle_install_idle_handler - installs the cpuidle idle loop handler
 */
void cpuidle_install_idle_handler(void)
{
	if (enabled_devices) {
		/* Make sure all changes finished before we switch to new idle */
		smp_wmb();
		initialized = 1;
	}
}

/**
 * cpuidle_uninstall_idle_handler - uninstalls the cpuidle idle loop handler
 */
void cpuidle_uninstall_idle_handler(void)
{
	if (enabled_devices) {
		initialized = 0;
		kick_all_cpus_sync();
	}
}

/**
 * cpuidle_pause_and_lock - temporarily disables CPUIDLE
 */
void cpuidle_pause_and_lock(void)
{
	mutex_lock(&cpuidle_lock);
	cpuidle_uninstall_idle_handler();
}

EXPORT_SYMBOL_GPL(cpuidle_pause_and_lock);

/**
 * cpuidle_resume_and_unlock - resumes CPUIDLE operation
 */
void cpuidle_resume_and_unlock(void)
{
	cpuidle_install_idle_handler();
	mutex_unlock(&cpuidle_lock);
}

EXPORT_SYMBOL_GPL(cpuidle_resume_and_unlock);

/**
 * cpuidle_wrap_enter - performs timekeeping and irqen around enter function
 * @dev: pointer to a valid cpuidle_device object
 * @drv: pointer to a valid cpuidle_driver object
 * @index: index of the target cpuidle state.
 */
int cpuidle_wrap_enter(struct cpuidle_device *dev,
				struct cpuidle_driver *drv, int index,
				int (*enter)(struct cpuidle_device *dev,
					struct cpuidle_driver *drv, int index))
{
	ktime_t time_start, time_end;
	s64 diff;

	time_start = ktime_get();

	index = enter(dev, drv, index);

	time_end = ktime_get();

	local_irq_enable();

	diff = ktime_to_us(ktime_sub(time_end, time_start));
	if (diff > INT_MAX)
		diff = INT_MAX;

	dev->last_residency = (int) diff;

	return index;
}

#ifdef CONFIG_ARCH_HAS_CPU_RELAX
static int poll_idle(struct cpuidle_device *dev,
		struct cpuidle_driver *drv, int index)
{
	ktime_t	t1, t2;
	s64 diff;

	t1 = ktime_get();
	local_irq_enable();
	while (!need_resched())
		cpu_relax();

	t2 = ktime_get();
	diff = ktime_to_us(ktime_sub(t2, t1));
	if (diff > INT_MAX)
		diff = INT_MAX;

	dev->last_residency = (int) diff;

	return index;
}

static void poll_idle_init(struct cpuidle_driver *drv)
{
	struct cpuidle_state *state = &drv->states[0];

	snprintf(state->name, CPUIDLE_NAME_LEN, "POLL");
	snprintf(state->desc, CPUIDLE_DESC_LEN, "CPUIDLE CORE POLL IDLE");
	state->exit_latency = 0;
	state->target_residency = 0;
	state->power_usage = -1;
	state->flags = 0;
	state->enter = poll_idle;
	state->disable = 0;
}
#else
static void poll_idle_init(struct cpuidle_driver *drv) {}
#endif /* CONFIG_ARCH_HAS_CPU_RELAX */

/**
 * cpuidle_enable_device - enables idle PM for a CPU
 * @dev: the CPU
 *
 * This function must be called between cpuidle_pause_and_lock and
 * cpuidle_resume_and_unlock when used externally.
 */
int cpuidle_enable_device(struct cpuidle_device *dev)
{
	int ret, i;
	struct cpuidle_driver *drv = cpuidle_get_driver();

	if (dev->enabled)
		return 0;
	if (!drv || !cpuidle_curr_governor)
		return -EIO;
	if (!dev->state_count)
		dev->state_count = drv->state_count;

	if (dev->registered == 0) {
		ret = __cpuidle_register_device(dev);
		if (ret)
			return ret;
	}

	cpuidle_enter_ops = drv->en_core_tk_irqen ?
		cpuidle_enter_tk : cpuidle_enter;

	poll_idle_init(drv);

	if ((ret = cpuidle_add_state_sysfs(dev)))
		return ret;

	if (cpuidle_curr_governor->enable &&
	    (ret = cpuidle_curr_governor->enable(drv, dev)))
		goto fail_sysfs;

	for (i = 0; i < dev->state_count; i++) {
		dev->states_usage[i].usage = 0;
		dev->states_usage[i].time = 0;
	}
	dev->last_residency = 0;

	smp_wmb();

	dev->enabled = 1;

	enabled_devices++;
	return 0;

fail_sysfs:
	cpuidle_remove_state_sysfs(dev);

	return ret;
}

EXPORT_SYMBOL_GPL(cpuidle_enable_device);

/**
 * cpuidle_disable_device - disables idle PM for a CPU
 * @dev: the CPU
 *
 * This function must be called between cpuidle_pause_and_lock and
 * cpuidle_resume_and_unlock when used externally.
 */
void cpuidle_disable_device(struct cpuidle_device *dev)
{
	if (!dev->enabled)
		return;
	if (!cpuidle_get_driver() || !cpuidle_curr_governor)
		return;

	dev->enabled = 0;

	if (cpuidle_curr_governor->disable)
		cpuidle_curr_governor->disable(cpuidle_get_driver(), dev);

	cpuidle_remove_state_sysfs(dev);
	enabled_devices--;
}

EXPORT_SYMBOL_GPL(cpuidle_disable_device);

/**
 * __cpuidle_register_device - internal register function called before register
 * and enable routines
 * @dev: the cpu
 *
 * cpuidle_lock mutex must be held before this is called
 */
static int __cpuidle_register_device(struct cpuidle_device *dev)
{
	int ret;
	struct device *cpu_dev = get_cpu_device((unsigned long)dev->cpu);
	struct cpuidle_driver *cpuidle_driver = cpuidle_get_driver();

	if (!dev)
		return -EINVAL;
	if (!try_module_get(cpuidle_driver->owner))
		return -EINVAL;

	init_completion(&dev->kobj_unregister);

	per_cpu(cpuidle_devices, dev->cpu) = dev;
	list_add(&dev->device_list, &cpuidle_detected_devices);
	if ((ret = cpuidle_add_sysfs(cpu_dev))) {
		module_put(cpuidle_driver->owner);
		return ret;
	}

	dev->registered = 1;
	return 0;
}

/**
 * cpuidle_register_device - registers a CPU's idle PM feature
 * @dev: the cpu
 */
int cpuidle_register_device(struct cpuidle_device *dev)
{
	int ret;

	mutex_lock(&cpuidle_lock);

	if ((ret = __cpuidle_register_device(dev))) {
		mutex_unlock(&cpuidle_lock);
		return ret;
	}

	cpuidle_enable_device(dev);
	cpuidle_install_idle_handler();

	mutex_unlock(&cpuidle_lock);

	return 0;

}

EXPORT_SYMBOL_GPL(cpuidle_register_device);

/**
 * cpuidle_unregister_device - unregisters a CPU's idle PM feature
 * @dev: the cpu
 */
void cpuidle_unregister_device(struct cpuidle_device *dev)
{
	struct device *cpu_dev = get_cpu_device((unsigned long)dev->cpu);
	struct cpuidle_driver *cpuidle_driver = cpuidle_get_driver();

	if (dev->registered == 0)
		return;

	cpuidle_pause_and_lock();

	cpuidle_disable_device(dev);

	cpuidle_remove_sysfs(cpu_dev);
	list_del(&dev->device_list);
	wait_for_completion(&dev->kobj_unregister);
	per_cpu(cpuidle_devices, dev->cpu) = NULL;

	cpuidle_resume_and_unlock();

	module_put(cpuidle_driver->owner);
}

EXPORT_SYMBOL_GPL(cpuidle_unregister_device);

#ifdef CONFIG_SMP

static void smp_callback(void *v)
{
	/* we already woke the CPU up, nothing more to do */
}

/*
 * This function gets called when a part of the kernel has a new latency
 * requirement.  This means we need to get all processors out of their C-state,
 * and then recalculate a new suitable C-state. Just do a cross-cpu IPI; that
 * wakes them all right up.
 */
static int cpuidle_latency_notify(struct notifier_block *b,
		unsigned long l, void *v)
{
	smp_call_function(smp_callback, NULL, 1);
	return NOTIFY_OK;
}

static struct notifier_block cpuidle_latency_notifier = {
	.notifier_call = cpuidle_latency_notify,
};

static inline void latency_notifier_init(struct notifier_block *n)
{
	pm_qos_add_notifier(PM_QOS_CPU_DMA_LATENCY, n);
}

#else /* CONFIG_SMP */

#define latency_notifier_init(x) do { } while (0)

#endif /* CONFIG_SMP */

/**
 * cpuidle_init - core initializer
 */
static int __init cpuidle_init(void)
{
	int ret;

	if (cpuidle_disabled())
		return -ENODEV;

	ret = cpuidle_add_interface(cpu_subsys.dev_root);
	if (ret)
		return ret;

	latency_notifier_init(&cpuidle_latency_notifier);

	return 0;
}

module_param(off, int, 0444);
core_initcall(cpuidle_init);
Commit	Line	Data
4f86d3a8 LB	1	/*
	2	* cpuidle.c - core cpuidle infrastructure
	3	*
	4	* (C) 2006-2007 Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
	5	* Shaohua Li <shaohua.li@intel.com>
	6	* Adam Belay <abelay@novell.com>
	7	*
	8	* This code is licenced under the GPL.
	9	*/
	10
	11	#include <linux/kernel.h>
	12	#include <linux/mutex.h>
	13	#include <linux/sched.h>
	14	#include <linux/notifier.h>
e8db0be1	15	#include <linux/pm_qos.h>
4f86d3a8 LB	16	#include <linux/cpu.h>
4f86d3a8 LB	17	#include <linux/cpuidle.h>
9a0b8415	18	#include <linux/ktime.h>
2e94d1f7	19	#include <linux/hrtimer.h>
884b17e1	20	#include <linux/module.h>
288f023e	21	#include <trace/events/power.h>
4f86d3a8 LB	22
	23	#include "cpuidle.h"
	24
	25	DEFINE_PER_CPU(struct cpuidle_device *, cpuidle_devices);
4f86d3a8 LB	26
	27	DEFINE_MUTEX(cpuidle_lock);
	28	LIST_HEAD(cpuidle_detected_devices);
4f86d3a8 LB	29
4f86d3a8 LB	30	static int enabled_devices;
62027aea	31	static int off __read_mostly;
a0bfa137	32	static int initialized __read_mostly;
62027aea LB	33
	34	int cpuidle_disabled(void)
	35	{
	36	return off;
	37	}
d91ee586 LB	38	void disable_cpuidle(void)
	39	{
	40	off = 1;
	41	}
4f86d3a8	42
dcb84f33 VP	43	static int __cpuidle_register_device(struct cpuidle_device *dev);
dcb84f33 VP	44
e1689795 RL	45	static inline int cpuidle_enter(struct cpuidle_device *dev,
	46	struct cpuidle_driver *drv, int index)
	47	{
	48	struct cpuidle_state *target_state = &drv->states[index];
	49	return target_state->enter(dev, drv, index);
	50	}
	51
	52	static inline int cpuidle_enter_tk(struct cpuidle_device *dev,
	53	struct cpuidle_driver *drv, int index)
	54	{
	55	return cpuidle_wrap_enter(dev, drv, index, cpuidle_enter);
	56	}
	57
	58	typedef int (cpuidle_enter_t)(struct cpuidle_device dev,
	59	struct cpuidle_driver *drv, int index);
	60
	61	static cpuidle_enter_t cpuidle_enter_ops;
	62
1a022e3f BO	63	/**
	64	* cpuidle_play_dead - cpu off-lining
	65	*
ee01e663	66	* Returns in case of an error or no driver
1a022e3f BO	67	*/
	68	int cpuidle_play_dead(void)
	69	{
	70	struct cpuidle_device *dev = __this_cpu_read(cpuidle_devices);
	71	struct cpuidle_driver *drv = cpuidle_get_driver();
	72	int i, dead_state = -1;
	73	int power_usage = -1;
	74
ee01e663 TK	75	if (!drv)
	76	return -ENODEV;
	77
1a022e3f BO	78	/* Find lowest-power state that supports long-term idle */
	79	for (i = CPUIDLE_DRIVER_STATE_START; i < drv->state_count; i++) {
	80	struct cpuidle_state *s = &drv->states[i];
	81
	82	if (s->power_usage < power_usage && s->enter_dead) {
	83	power_usage = s->power_usage;
	84	dead_state = i;
	85	}
	86	}
	87
	88	if (dead_state != -1)
	89	return drv->states[dead_state].enter_dead(dev, dead_state);
	90
	91	return -ENODEV;
	92	}
	93
4f86d3a8 LB	94	/**
	95	* cpuidle_idle_call - the main idle loop
	96	*
	97	* NOTE: no locks or semaphores should be used here
a0bfa137	98	* return non-zero on failure
4f86d3a8	99	*/
a0bfa137	100	int cpuidle_idle_call(void)
4f86d3a8	101	{
4a6f4fe8	102	struct cpuidle_device *dev = __this_cpu_read(cpuidle_devices);
46bcfad7	103	struct cpuidle_driver *drv = cpuidle_get_driver();
e978aa7d	104	int next_state, entered_state;
4f86d3a8	105
a0bfa137 LB	106	if (off)
	107	return -ENODEV;
	108
	109	if (!initialized)
	110	return -ENODEV;
	111
4f86d3a8	112	/* check if the device is ready */
a0bfa137 LB	113	if (!dev \|\| !dev->enabled)
a0bfa137 LB	114	return -EBUSY;
4f86d3a8	115
9a655837 AV	116	#if 0
9a655837 AV	117	/* shows regressions, re-enable for 2.6.29 */
2e94d1f7 AV	118	/*
	119	* run any timers that can be run now, at this point
	120	* before calculating the idle duration etc.
	121	*/
	122	hrtimer_peek_ahead_timers();
9a655837	123	#endif
71abbbf8	124
4f86d3a8	125	/* ask the governor for the next state */
46bcfad7	126	next_state = cpuidle_curr_governor->select(drv, dev);
246eb7f0 KH	127	if (need_resched()) {
246eb7f0 KH	128	local_irq_enable();
a0bfa137	129	return 0;
246eb7f0 KH	130	}
246eb7f0 KH	131
76027ea8 SR	132	trace_power_start_rcuidle(POWER_CSTATE, next_state, dev->cpu);
76027ea8 SR	133	trace_cpu_idle_rcuidle(next_state, dev->cpu);
f77cfe4e	134
e1689795	135	entered_state = cpuidle_enter_ops(dev, drv, next_state);
f77cfe4e	136
76027ea8 SR	137	trace_power_end_rcuidle(dev->cpu);
76027ea8 SR	138	trace_cpu_idle_rcuidle(PWR_EVENT_EXIT, dev->cpu);
f77cfe4e	139
e978aa7d DD	140	if (entered_state >= 0) {
	141	/* Update cpuidle counters */
	142	/* This can be moved to within driver enter routine
	143	* but that results in multiple copies of same code.
	144	*/
4202735e	145	dev->states_usage[entered_state].time +=
e978aa7d	146	(unsigned long long)dev->last_residency;
4202735e	147	dev->states_usage[entered_state].usage++;
e1689795 RL	148	} else {
e1689795 RL	149	dev->last_residency = 0;
e978aa7d	150	}
4f86d3a8 LB	151
	152	/* give the governor an opportunity to reflect on the outcome */
	153	if (cpuidle_curr_governor->reflect)
e978aa7d	154	cpuidle_curr_governor->reflect(dev, entered_state);
a0bfa137 LB	155
a0bfa137 LB	156	return 0;
4f86d3a8 LB	157	}
	158
	159	/**
	160	* cpuidle_install_idle_handler - installs the cpuidle idle loop handler
	161	*/
	162	void cpuidle_install_idle_handler(void)
	163	{
a0bfa137	164	if (enabled_devices) {
4f86d3a8 LB	165	/* Make sure all changes finished before we switch to new idle */
4f86d3a8 LB	166	smp_wmb();
a0bfa137	167	initialized = 1;
4f86d3a8 LB	168	}
	169	}
	170
	171	/**
	172	* cpuidle_uninstall_idle_handler - uninstalls the cpuidle idle loop handler
	173	*/
	174	void cpuidle_uninstall_idle_handler(void)
	175	{
a0bfa137 LB	176	if (enabled_devices) {
a0bfa137 LB	177	initialized = 0;
4a162513	178	kick_all_cpus_sync();
4f86d3a8 LB	179	}
	180	}
	181
	182	/**
	183	* cpuidle_pause_and_lock - temporarily disables CPUIDLE
	184	*/
	185	void cpuidle_pause_and_lock(void)
	186	{
	187	mutex_lock(&cpuidle_lock);
	188	cpuidle_uninstall_idle_handler();
	189	}
	190
	191	EXPORT_SYMBOL_GPL(cpuidle_pause_and_lock);
	192
	193	/**
	194	* cpuidle_resume_and_unlock - resumes CPUIDLE operation
	195	*/
	196	void cpuidle_resume_and_unlock(void)
	197	{
	198	cpuidle_install_idle_handler();
	199	mutex_unlock(&cpuidle_lock);
	200	}
	201
	202	EXPORT_SYMBOL_GPL(cpuidle_resume_and_unlock);
	203
e1689795 RL	204	/**
	205	* cpuidle_wrap_enter - performs timekeeping and irqen around enter function
	206	* @dev: pointer to a valid cpuidle_device object
	207	* @drv: pointer to a valid cpuidle_driver object
	208	* @index: index of the target cpuidle state.
	209	*/
	210	int cpuidle_wrap_enter(struct cpuidle_device *dev,
	211	struct cpuidle_driver *drv, int index,
	212	int (enter)(struct cpuidle_device dev,
	213	struct cpuidle_driver *drv, int index))
	214	{
	215	ktime_t time_start, time_end;
	216	s64 diff;
	217
	218	time_start = ktime_get();
	219
	220	index = enter(dev, drv, index);
	221
	222	time_end = ktime_get();
	223
	224	local_irq_enable();
	225
	226	diff = ktime_to_us(ktime_sub(time_end, time_start));
	227	if (diff > INT_MAX)
	228	diff = INT_MAX;
	229
	230	dev->last_residency = (int) diff;
	231
	232	return index;
	233	}
	234
d8c216cf	235	#ifdef CONFIG_ARCH_HAS_CPU_RELAX
46bcfad7 DD	236	static int poll_idle(struct cpuidle_device *dev,
46bcfad7 DD	237	struct cpuidle_driver *drv, int index)
d8c216cf RW	238	{
	239	ktime_t t1, t2;
	240	s64 diff;
d8c216cf RW	241
	242	t1 = ktime_get();
	243	local_irq_enable();
	244	while (!need_resched())
	245	cpu_relax();
	246
	247	t2 = ktime_get();
	248	diff = ktime_to_us(ktime_sub(t2, t1));
	249	if (diff > INT_MAX)
	250	diff = INT_MAX;
	251
e978aa7d DD	252	dev->last_residency = (int) diff;
	253
	254	return index;
d8c216cf RW	255	}
d8c216cf RW	256
46bcfad7	257	static void poll_idle_init(struct cpuidle_driver *drv)
d8c216cf	258	{
46bcfad7	259	struct cpuidle_state *state = &drv->states[0];
d8c216cf	260
720f1c30	261	snprintf(state->name, CPUIDLE_NAME_LEN, "POLL");
d8c216cf RW	262	snprintf(state->desc, CPUIDLE_DESC_LEN, "CPUIDLE CORE POLL IDLE");
	263	state->exit_latency = 0;
	264	state->target_residency = 0;
	265	state->power_usage = -1;
d247632c	266	state->flags = 0;
d8c216cf	267	state->enter = poll_idle;
3a53396b	268	state->disable = 0;
d8c216cf RW	269	}
d8c216cf RW	270	#else
46bcfad7	271	static void poll_idle_init(struct cpuidle_driver *drv) {}
d8c216cf RW	272	#endif /* CONFIG_ARCH_HAS_CPU_RELAX */
d8c216cf RW	273
4f86d3a8 LB	274	/**
	275	* cpuidle_enable_device - enables idle PM for a CPU
	276	* @dev: the CPU
	277	*
	278	* This function must be called between cpuidle_pause_and_lock and
	279	* cpuidle_resume_and_unlock when used externally.
	280	*/
	281	int cpuidle_enable_device(struct cpuidle_device *dev)
	282	{
	283	int ret, i;
e1689795	284	struct cpuidle_driver *drv = cpuidle_get_driver();
4f86d3a8 LB	285
	286	if (dev->enabled)
	287	return 0;
e1689795	288	if (!drv \|\| !cpuidle_curr_governor)
4f86d3a8 LB	289	return -EIO;
4f86d3a8 LB	290	if (!dev->state_count)
fc850f39	291	dev->state_count = drv->state_count;
4f86d3a8	292
dcb84f33 VP	293	if (dev->registered == 0) {
	294	ret = __cpuidle_register_device(dev);
	295	if (ret)
	296	return ret;
	297	}
	298
e1689795 RL	299	cpuidle_enter_ops = drv->en_core_tk_irqen ?
	300	cpuidle_enter_tk : cpuidle_enter;
	301
	302	poll_idle_init(drv);
d8c216cf	303
4f86d3a8 LB	304	if ((ret = cpuidle_add_state_sysfs(dev)))
	305	return ret;
	306
	307	if (cpuidle_curr_governor->enable &&
e1689795	308	(ret = cpuidle_curr_governor->enable(drv, dev)))
4f86d3a8 LB	309	goto fail_sysfs;
	310
	311	for (i = 0; i < dev->state_count; i++) {
4202735e DD	312	dev->states_usage[i].usage = 0;
4202735e DD	313	dev->states_usage[i].time = 0;
4f86d3a8 LB	314	}
4f86d3a8 LB	315	dev->last_residency = 0;
4f86d3a8 LB	316
	317	smp_wmb();
	318
	319	dev->enabled = 1;
	320
	321	enabled_devices++;
	322	return 0;
	323
	324	fail_sysfs:
	325	cpuidle_remove_state_sysfs(dev);
	326
	327	return ret;
	328	}
	329
	330	EXPORT_SYMBOL_GPL(cpuidle_enable_device);
	331
	332	/**
	333	* cpuidle_disable_device - disables idle PM for a CPU
	334	* @dev: the CPU
	335	*
	336	* This function must be called between cpuidle_pause_and_lock and
	337	* cpuidle_resume_and_unlock when used externally.
	338	*/
	339	void cpuidle_disable_device(struct cpuidle_device *dev)
	340	{
	341	if (!dev->enabled)
	342	return;
752138df	343	if (!cpuidle_get_driver() \|\| !cpuidle_curr_governor)
4f86d3a8 LB	344	return;
	345
	346	dev->enabled = 0;
	347
	348	if (cpuidle_curr_governor->disable)
46bcfad7	349	cpuidle_curr_governor->disable(cpuidle_get_driver(), dev);
4f86d3a8 LB	350
	351	cpuidle_remove_state_sysfs(dev);
	352	enabled_devices--;
	353	}
	354
	355	EXPORT_SYMBOL_GPL(cpuidle_disable_device);
	356
	357	/**
dcb84f33 VP	358	* __cpuidle_register_device - internal register function called before register
dcb84f33 VP	359	* and enable routines
4f86d3a8	360	* @dev: the cpu
dcb84f33 VP	361	*
dcb84f33 VP	362	* cpuidle_lock mutex must be held before this is called
4f86d3a8	363	*/
dcb84f33	364	static int __cpuidle_register_device(struct cpuidle_device *dev)
4f86d3a8 LB	365	{
4f86d3a8 LB	366	int ret;
8a25a2fd	367	struct device *cpu_dev = get_cpu_device((unsigned long)dev->cpu);
752138df	368	struct cpuidle_driver *cpuidle_driver = cpuidle_get_driver();
4f86d3a8	369
8a25a2fd	370	if (!dev)
4f86d3a8	371	return -EINVAL;
752138df	372	if (!try_module_get(cpuidle_driver->owner))
4f86d3a8 LB	373	return -EINVAL;
	374
	375	init_completion(&dev->kobj_unregister);
	376
4f86d3a8 LB	377	per_cpu(cpuidle_devices, dev->cpu) = dev;
4f86d3a8 LB	378	list_add(&dev->device_list, &cpuidle_detected_devices);
8a25a2fd	379	if ((ret = cpuidle_add_sysfs(cpu_dev))) {
752138df	380	module_put(cpuidle_driver->owner);
4f86d3a8 LB	381	return ret;
	382	}
	383
dcb84f33 VP	384	dev->registered = 1;
	385	return 0;
	386	}
	387
	388	/**
	389	* cpuidle_register_device - registers a CPU's idle PM feature
	390	* @dev: the cpu
	391	*/
	392	int cpuidle_register_device(struct cpuidle_device *dev)
	393	{
	394	int ret;
	395
	396	mutex_lock(&cpuidle_lock);
	397
	398	if ((ret = __cpuidle_register_device(dev))) {
	399	mutex_unlock(&cpuidle_lock);
	400	return ret;
	401	}
	402
4f86d3a8 LB	403	cpuidle_enable_device(dev);
	404	cpuidle_install_idle_handler();
	405
	406	mutex_unlock(&cpuidle_lock);
	407
	408	return 0;
	409
	410	}
	411
	412	EXPORT_SYMBOL_GPL(cpuidle_register_device);
	413
	414	/**
	415	* cpuidle_unregister_device - unregisters a CPU's idle PM feature
	416	* @dev: the cpu
	417	*/
	418	void cpuidle_unregister_device(struct cpuidle_device *dev)
	419	{
8a25a2fd	420	struct device *cpu_dev = get_cpu_device((unsigned long)dev->cpu);
752138df	421	struct cpuidle_driver *cpuidle_driver = cpuidle_get_driver();
4f86d3a8	422
dcb84f33 VP	423	if (dev->registered == 0)
	424	return;
	425
4f86d3a8 LB	426	cpuidle_pause_and_lock();
	427
	428	cpuidle_disable_device(dev);
	429
8a25a2fd	430	cpuidle_remove_sysfs(cpu_dev);
4f86d3a8 LB	431	list_del(&dev->device_list);
	432	wait_for_completion(&dev->kobj_unregister);
	433	per_cpu(cpuidle_devices, dev->cpu) = NULL;
	434
	435	cpuidle_resume_and_unlock();
	436
752138df	437	module_put(cpuidle_driver->owner);
4f86d3a8 LB	438	}
	439
	440	EXPORT_SYMBOL_GPL(cpuidle_unregister_device);
	441
	442	#ifdef CONFIG_SMP
	443
	444	static void smp_callback(void *v)
	445	{
	446	/* we already woke the CPU up, nothing more to do */
	447	}
	448
	449	/*
	450	* This function gets called when a part of the kernel has a new latency
	451	* requirement. This means we need to get all processors out of their C-state,
	452	* and then recalculate a new suitable C-state. Just do a cross-cpu IPI; that
	453	* wakes them all right up.
	454	*/
	455	static int cpuidle_latency_notify(struct notifier_block *b,
	456	unsigned long l, void *v)
	457	{
8691e5a8	458	smp_call_function(smp_callback, NULL, 1);
4f86d3a8 LB	459	return NOTIFY_OK;
	460	}
	461
	462	static struct notifier_block cpuidle_latency_notifier = {
	463	.notifier_call = cpuidle_latency_notify,
	464	};
	465
d82b3518 MG	466	static inline void latency_notifier_init(struct notifier_block *n)
	467	{
	468	pm_qos_add_notifier(PM_QOS_CPU_DMA_LATENCY, n);
	469	}
4f86d3a8 LB	470
	471	#else /* CONFIG_SMP */
	472
	473	#define latency_notifier_init(x) do { } while (0)
	474
	475	#endif /* CONFIG_SMP */
	476
	477	/**
	478	* cpuidle_init - core initializer
	479	*/
	480	static int __init cpuidle_init(void)
	481	{
	482	int ret;
	483
62027aea LB	484	if (cpuidle_disabled())
	485	return -ENODEV;
	486
8a25a2fd	487	ret = cpuidle_add_interface(cpu_subsys.dev_root);
4f86d3a8 LB	488	if (ret)
	489	return ret;
	490
	491	latency_notifier_init(&cpuidle_latency_notifier);
	492
	493	return 0;
	494	}
	495
62027aea	496	module_param(off, int, 0444);
4f86d3a8	497	core_initcall(cpuidle_init);