[mirror_ubuntu-jammy-kernel.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_params.h>
#include <linux/cpu.h>
#include <linux/cpuidle.h>
#include <linux/ktime.h>
#include <linux/hrtimer.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;
}

#if defined(CONFIG_ARCH_HAS_CPU_IDLE_WAIT)
static void cpuidle_kick_cpus(void)
{
	cpu_idle_wait();
}
#elif defined(CONFIG_SMP)
# error "Arch needs cpu_idle_wait() equivalent here"
#else /* !CONFIG_ARCH_HAS_CPU_IDLE_WAIT && !CONFIG_SMP */
static void cpuidle_kick_cpus(void) {}
#endif

static int __cpuidle_register_device(struct cpuidle_device *dev);

/**
 * 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_state *target_state;
	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(dev);
	if (need_resched()) {
		local_irq_enable();
		return 0;
	}

	target_state = &dev->states[next_state];

	trace_power_start(POWER_CSTATE, next_state, dev->cpu);
	trace_cpu_idle(next_state, dev->cpu);

	entered_state = target_state->enter(dev, next_state);

	trace_power_end(dev->cpu);
	trace_cpu_idle(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++;
	}

	/* 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;
		cpuidle_kick_cpus();
	}
}

/**
 * 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);

#ifdef CONFIG_ARCH_HAS_CPU_RELAX
static int poll_idle(struct cpuidle_device *dev, 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_device *dev)
{
	struct cpuidle_state *state = &dev->states[0];
	struct cpuidle_state_usage *state_usage = &dev->states_usage[0];

	cpuidle_set_statedata(state_usage, NULL);

	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;
}
#else
static void poll_idle_init(struct cpuidle_device *dev) {}
#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;

	if (dev->enabled)
		return 0;
	if (!cpuidle_get_driver() || !cpuidle_curr_governor)
		return -EIO;
	if (!dev->state_count)
		return -EINVAL;

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

	poll_idle_init(dev);

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

	if (cpuidle_curr_governor->enable &&
	    (ret = cpuidle_curr_governor->enable(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(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 sys_device *sys_dev = get_cpu_sysdev((unsigned long)dev->cpu);
	struct cpuidle_driver *cpuidle_driver = cpuidle_get_driver();

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

	init_completion(&dev->kobj_unregister);

	/*
	 * cpuidle driver should set the dev->power_specified bit
	 * before registering the device if the driver provides
	 * power_usage numbers.
	 *
	 * For those devices whose ->power_specified is not set,
	 * we fill in power_usage with decreasing values as the
	 * cpuidle code has an implicit assumption that state Cn
	 * uses less power than C(n-1).
	 *
	 * With CONFIG_ARCH_HAS_CPU_RELAX, C0 is already assigned
	 * an power value of -1.  So we use -2, -3, etc, for other
	 * c-states.
	 */
	if (!dev->power_specified) {
		int i;
		for (i = CPUIDLE_DRIVER_STATE_START; i < dev->state_count; i++)
			dev->states[i].power_usage = -1 - i;
	}

	per_cpu(cpuidle_devices, dev->cpu) = dev;
	list_add(&dev->device_list, &cpuidle_detected_devices);
	if ((ret = cpuidle_add_sysfs(sys_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 sys_device *sys_dev = get_cpu_sysdev((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(sys_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_class_sysfs(&cpu_sysdev_class);
	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>
d82b3518	15	#include <linux/pm_qos_params.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>
288f023e	20	#include <trace/events/power.h>
4f86d3a8 LB	21
	22	#include "cpuidle.h"
	23
	24	DEFINE_PER_CPU(struct cpuidle_device *, cpuidle_devices);
4f86d3a8 LB	25
	26	DEFINE_MUTEX(cpuidle_lock);
	27	LIST_HEAD(cpuidle_detected_devices);
4f86d3a8 LB	28
4f86d3a8 LB	29	static int enabled_devices;
62027aea	30	static int off __read_mostly;
a0bfa137	31	static int initialized __read_mostly;
62027aea LB	32
	33	int cpuidle_disabled(void)
	34	{
	35	return off;
	36	}
d91ee586 LB	37	void disable_cpuidle(void)
	38	{
	39	off = 1;
	40	}
4f86d3a8	41
a6869cc4 VP	42	#if defined(CONFIG_ARCH_HAS_CPU_IDLE_WAIT)
	43	static void cpuidle_kick_cpus(void)
	44	{
	45	cpu_idle_wait();
	46	}
	47	#elif defined(CONFIG_SMP)
	48	# error "Arch needs cpu_idle_wait() equivalent here"
	49	#else /* !CONFIG_ARCH_HAS_CPU_IDLE_WAIT && !CONFIG_SMP */
	50	static void cpuidle_kick_cpus(void) {}
	51	#endif
	52
dcb84f33 VP	53	static int __cpuidle_register_device(struct cpuidle_device *dev);
dcb84f33 VP	54
4f86d3a8 LB	55	/**
	56	* cpuidle_idle_call - the main idle loop
	57	*
	58	* NOTE: no locks or semaphores should be used here
a0bfa137	59	* return non-zero on failure
4f86d3a8	60	*/
a0bfa137	61	int cpuidle_idle_call(void)
4f86d3a8	62	{
4a6f4fe8	63	struct cpuidle_device *dev = __this_cpu_read(cpuidle_devices);
4f86d3a8	64	struct cpuidle_state *target_state;
e978aa7d	65	int next_state, entered_state;
4f86d3a8	66
a0bfa137 LB	67	if (off)
	68	return -ENODEV;
	69
	70	if (!initialized)
	71	return -ENODEV;
	72
4f86d3a8	73	/* check if the device is ready */
a0bfa137 LB	74	if (!dev \|\| !dev->enabled)
a0bfa137 LB	75	return -EBUSY;
4f86d3a8	76
9a655837 AV	77	#if 0
9a655837 AV	78	/* shows regressions, re-enable for 2.6.29 */
2e94d1f7 AV	79	/*
	80	* run any timers that can be run now, at this point
	81	* before calculating the idle duration etc.
	82	*/
	83	hrtimer_peek_ahead_timers();
9a655837	84	#endif
71abbbf8	85
4f86d3a8 LB	86	/* ask the governor for the next state */
4f86d3a8 LB	87	next_state = cpuidle_curr_governor->select(dev);
246eb7f0 KH	88	if (need_resched()) {
246eb7f0 KH	89	local_irq_enable();
a0bfa137	90	return 0;
246eb7f0 KH	91	}
246eb7f0 KH	92
4f86d3a8 LB	93	target_state = &dev->states[next_state];
4f86d3a8 LB	94
f77cfe4e TR	95	trace_power_start(POWER_CSTATE, next_state, dev->cpu);
	96	trace_cpu_idle(next_state, dev->cpu);
	97
e978aa7d	98	entered_state = target_state->enter(dev, next_state);
f77cfe4e TR	99
	100	trace_power_end(dev->cpu);
	101	trace_cpu_idle(PWR_EVENT_EXIT, dev->cpu);
	102
e978aa7d DD	103	if (entered_state >= 0) {
	104	/* Update cpuidle counters */
	105	/* This can be moved to within driver enter routine
	106	* but that results in multiple copies of same code.
	107	*/
4202735e	108	dev->states_usage[entered_state].time +=
e978aa7d	109	(unsigned long long)dev->last_residency;
4202735e	110	dev->states_usage[entered_state].usage++;
e978aa7d	111	}
4f86d3a8 LB	112
	113	/* give the governor an opportunity to reflect on the outcome */
	114	if (cpuidle_curr_governor->reflect)
e978aa7d	115	cpuidle_curr_governor->reflect(dev, entered_state);
a0bfa137 LB	116
a0bfa137 LB	117	return 0;
4f86d3a8 LB	118	}
	119
	120	/**
	121	* cpuidle_install_idle_handler - installs the cpuidle idle loop handler
	122	*/
	123	void cpuidle_install_idle_handler(void)
	124	{
a0bfa137	125	if (enabled_devices) {
4f86d3a8 LB	126	/* Make sure all changes finished before we switch to new idle */
4f86d3a8 LB	127	smp_wmb();
a0bfa137	128	initialized = 1;
4f86d3a8 LB	129	}
	130	}
	131
	132	/**
	133	* cpuidle_uninstall_idle_handler - uninstalls the cpuidle idle loop handler
	134	*/
	135	void cpuidle_uninstall_idle_handler(void)
	136	{
a0bfa137 LB	137	if (enabled_devices) {
a0bfa137 LB	138	initialized = 0;
a6869cc4	139	cpuidle_kick_cpus();
4f86d3a8 LB	140	}
	141	}
	142
	143	/**
	144	* cpuidle_pause_and_lock - temporarily disables CPUIDLE
	145	*/
	146	void cpuidle_pause_and_lock(void)
	147	{
	148	mutex_lock(&cpuidle_lock);
	149	cpuidle_uninstall_idle_handler();
	150	}
	151
	152	EXPORT_SYMBOL_GPL(cpuidle_pause_and_lock);
	153
	154	/**
	155	* cpuidle_resume_and_unlock - resumes CPUIDLE operation
	156	*/
	157	void cpuidle_resume_and_unlock(void)
	158	{
	159	cpuidle_install_idle_handler();
	160	mutex_unlock(&cpuidle_lock);
	161	}
	162
	163	EXPORT_SYMBOL_GPL(cpuidle_resume_and_unlock);
	164
d8c216cf	165	#ifdef CONFIG_ARCH_HAS_CPU_RELAX
e978aa7d	166	static int poll_idle(struct cpuidle_device *dev, int index)
d8c216cf RW	167	{
	168	ktime_t t1, t2;
	169	s64 diff;
d8c216cf RW	170
	171	t1 = ktime_get();
	172	local_irq_enable();
	173	while (!need_resched())
	174	cpu_relax();
	175
	176	t2 = ktime_get();
	177	diff = ktime_to_us(ktime_sub(t2, t1));
	178	if (diff > INT_MAX)
	179	diff = INT_MAX;
	180
e978aa7d DD	181	dev->last_residency = (int) diff;
	182
	183	return index;
d8c216cf RW	184	}
	185
	186	static void poll_idle_init(struct cpuidle_device *dev)
	187	{
	188	struct cpuidle_state *state = &dev->states[0];
4202735e	189	struct cpuidle_state_usage *state_usage = &dev->states_usage[0];
d8c216cf	190
4202735e	191	cpuidle_set_statedata(state_usage, NULL);
d8c216cf	192
720f1c30	193	snprintf(state->name, CPUIDLE_NAME_LEN, "POLL");
d8c216cf RW	194	snprintf(state->desc, CPUIDLE_DESC_LEN, "CPUIDLE CORE POLL IDLE");
	195	state->exit_latency = 0;
	196	state->target_residency = 0;
	197	state->power_usage = -1;
d247632c	198	state->flags = 0;
d8c216cf RW	199	state->enter = poll_idle;
	200	}
	201	#else
	202	static void poll_idle_init(struct cpuidle_device *dev) {}
	203	#endif /* CONFIG_ARCH_HAS_CPU_RELAX */
	204
4f86d3a8 LB	205	/**
	206	* cpuidle_enable_device - enables idle PM for a CPU
	207	* @dev: the CPU
	208	*
	209	* This function must be called between cpuidle_pause_and_lock and
	210	* cpuidle_resume_and_unlock when used externally.
	211	*/
	212	int cpuidle_enable_device(struct cpuidle_device *dev)
	213	{
	214	int ret, i;
	215
	216	if (dev->enabled)
	217	return 0;
752138df	218	if (!cpuidle_get_driver() \|\| !cpuidle_curr_governor)
4f86d3a8 LB	219	return -EIO;
	220	if (!dev->state_count)
	221	return -EINVAL;
	222
dcb84f33 VP	223	if (dev->registered == 0) {
	224	ret = __cpuidle_register_device(dev);
	225	if (ret)
	226	return ret;
	227	}
	228
d8c216cf RW	229	poll_idle_init(dev);
d8c216cf RW	230
4f86d3a8 LB	231	if ((ret = cpuidle_add_state_sysfs(dev)))
	232	return ret;
	233
	234	if (cpuidle_curr_governor->enable &&
	235	(ret = cpuidle_curr_governor->enable(dev)))
	236	goto fail_sysfs;
	237
	238	for (i = 0; i < dev->state_count; i++) {
4202735e DD	239	dev->states_usage[i].usage = 0;
4202735e DD	240	dev->states_usage[i].time = 0;
4f86d3a8 LB	241	}
4f86d3a8 LB	242	dev->last_residency = 0;
4f86d3a8 LB	243
	244	smp_wmb();
	245
	246	dev->enabled = 1;
	247
	248	enabled_devices++;
	249	return 0;
	250
	251	fail_sysfs:
	252	cpuidle_remove_state_sysfs(dev);
	253
	254	return ret;
	255	}
	256
	257	EXPORT_SYMBOL_GPL(cpuidle_enable_device);
	258
	259	/**
	260	* cpuidle_disable_device - disables idle PM for a CPU
	261	* @dev: the CPU
	262	*
	263	* This function must be called between cpuidle_pause_and_lock and
	264	* cpuidle_resume_and_unlock when used externally.
	265	*/
	266	void cpuidle_disable_device(struct cpuidle_device *dev)
	267	{
	268	if (!dev->enabled)
	269	return;
752138df	270	if (!cpuidle_get_driver() \|\| !cpuidle_curr_governor)
4f86d3a8 LB	271	return;
	272
	273	dev->enabled = 0;
	274
	275	if (cpuidle_curr_governor->disable)
	276	cpuidle_curr_governor->disable(dev);
	277
	278	cpuidle_remove_state_sysfs(dev);
	279	enabled_devices--;
	280	}
	281
	282	EXPORT_SYMBOL_GPL(cpuidle_disable_device);
	283
	284	/**
dcb84f33 VP	285	* __cpuidle_register_device - internal register function called before register
dcb84f33 VP	286	* and enable routines
4f86d3a8	287	* @dev: the cpu
dcb84f33 VP	288	*
dcb84f33 VP	289	* cpuidle_lock mutex must be held before this is called
4f86d3a8	290	*/
dcb84f33	291	static int __cpuidle_register_device(struct cpuidle_device *dev)
4f86d3a8 LB	292	{
	293	int ret;
	294	struct sys_device *sys_dev = get_cpu_sysdev((unsigned long)dev->cpu);
752138df	295	struct cpuidle_driver *cpuidle_driver = cpuidle_get_driver();
4f86d3a8 LB	296
	297	if (!sys_dev)
	298	return -EINVAL;
752138df	299	if (!try_module_get(cpuidle_driver->owner))
4f86d3a8 LB	300	return -EINVAL;
	301
	302	init_completion(&dev->kobj_unregister);
	303
71abbbf8 AL	304	/*
	305	* cpuidle driver should set the dev->power_specified bit
	306	* before registering the device if the driver provides
	307	* power_usage numbers.
	308	*
	309	* For those devices whose ->power_specified is not set,
	310	* we fill in power_usage with decreasing values as the
	311	* cpuidle code has an implicit assumption that state Cn
	312	* uses less power than C(n-1).
	313	*
	314	* With CONFIG_ARCH_HAS_CPU_RELAX, C0 is already assigned
	315	* an power value of -1. So we use -2, -3, etc, for other
	316	* c-states.
	317	*/
	318	if (!dev->power_specified) {
	319	int i;
	320	for (i = CPUIDLE_DRIVER_STATE_START; i < dev->state_count; i++)
	321	dev->states[i].power_usage = -1 - i;
	322	}
	323
4f86d3a8 LB	324	per_cpu(cpuidle_devices, dev->cpu) = dev;
	325	list_add(&dev->device_list, &cpuidle_detected_devices);
	326	if ((ret = cpuidle_add_sysfs(sys_dev))) {
752138df	327	module_put(cpuidle_driver->owner);
4f86d3a8 LB	328	return ret;
	329	}
	330
dcb84f33 VP	331	dev->registered = 1;
	332	return 0;
	333	}
	334
	335	/**
	336	* cpuidle_register_device - registers a CPU's idle PM feature
	337	* @dev: the cpu
	338	*/
	339	int cpuidle_register_device(struct cpuidle_device *dev)
	340	{
	341	int ret;
	342
	343	mutex_lock(&cpuidle_lock);
	344
	345	if ((ret = __cpuidle_register_device(dev))) {
	346	mutex_unlock(&cpuidle_lock);
	347	return ret;
	348	}
	349
4f86d3a8 LB	350	cpuidle_enable_device(dev);
	351	cpuidle_install_idle_handler();
	352
	353	mutex_unlock(&cpuidle_lock);
	354
	355	return 0;
	356
	357	}
	358
	359	EXPORT_SYMBOL_GPL(cpuidle_register_device);
	360
	361	/**
	362	* cpuidle_unregister_device - unregisters a CPU's idle PM feature
	363	* @dev: the cpu
	364	*/
	365	void cpuidle_unregister_device(struct cpuidle_device *dev)
	366	{
	367	struct sys_device *sys_dev = get_cpu_sysdev((unsigned long)dev->cpu);
752138df	368	struct cpuidle_driver *cpuidle_driver = cpuidle_get_driver();
4f86d3a8	369
dcb84f33 VP	370	if (dev->registered == 0)
	371	return;
	372
4f86d3a8 LB	373	cpuidle_pause_and_lock();
	374
	375	cpuidle_disable_device(dev);
	376
	377	cpuidle_remove_sysfs(sys_dev);
	378	list_del(&dev->device_list);
	379	wait_for_completion(&dev->kobj_unregister);
	380	per_cpu(cpuidle_devices, dev->cpu) = NULL;
	381
	382	cpuidle_resume_and_unlock();
	383
752138df	384	module_put(cpuidle_driver->owner);
4f86d3a8 LB	385	}
	386
	387	EXPORT_SYMBOL_GPL(cpuidle_unregister_device);
	388
	389	#ifdef CONFIG_SMP
	390
	391	static void smp_callback(void *v)
	392	{
	393	/* we already woke the CPU up, nothing more to do */
	394	}
	395
	396	/*
	397	* This function gets called when a part of the kernel has a new latency
	398	* requirement. This means we need to get all processors out of their C-state,
	399	* and then recalculate a new suitable C-state. Just do a cross-cpu IPI; that
	400	* wakes them all right up.
	401	*/
	402	static int cpuidle_latency_notify(struct notifier_block *b,
	403	unsigned long l, void *v)
	404	{
8691e5a8	405	smp_call_function(smp_callback, NULL, 1);
4f86d3a8 LB	406	return NOTIFY_OK;
	407	}
	408
	409	static struct notifier_block cpuidle_latency_notifier = {
	410	.notifier_call = cpuidle_latency_notify,
	411	};
	412
d82b3518 MG	413	static inline void latency_notifier_init(struct notifier_block *n)
	414	{
	415	pm_qos_add_notifier(PM_QOS_CPU_DMA_LATENCY, n);
	416	}
4f86d3a8 LB	417
	418	#else /* CONFIG_SMP */
	419
	420	#define latency_notifier_init(x) do { } while (0)
	421
	422	#endif /* CONFIG_SMP */
	423
	424	/**
	425	* cpuidle_init - core initializer
	426	*/
	427	static int __init cpuidle_init(void)
	428	{
	429	int ret;
	430
62027aea LB	431	if (cpuidle_disabled())
	432	return -ENODEV;
	433
4f86d3a8 LB	434	ret = cpuidle_add_class_sysfs(&cpu_sysdev_class);
	435	if (ret)
	436	return ret;
	437
	438	latency_notifier_init(&cpuidle_latency_notifier);
	439
	440	return 0;
	441	}
	442
62027aea	443	module_param(off, int, 0444);
4f86d3a8	444	core_initcall(cpuidle_init);