[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_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 void (*pm_idle_old)(void);

static int enabled_devices;
static int off __read_mostly;

int cpuidle_disabled(void)
{
	return off;
}

#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
 */
static void cpuidle_idle_call(void)
{
	struct cpuidle_device *dev = __this_cpu_read(cpuidle_devices);
	struct cpuidle_state *target_state;
	int next_state;

	/* check if the device is ready */
	if (!dev || !dev->enabled) {
		if (pm_idle_old)
			pm_idle_old();
		else
#if defined(CONFIG_ARCH_HAS_DEFAULT_IDLE)
			default_idle();
#else
			local_irq_enable();
#endif
		return;
	}

#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

	/*
	 * Call the device's prepare function before calling the
	 * governor's select function.  ->prepare gives the device's
	 * cpuidle driver a chance to update any dynamic information
	 * of its cpuidle states for the current idle period, e.g.
	 * state availability, latencies, residencies, etc.
	 */
	if (dev->prepare)
		dev->prepare(dev);

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

	target_state = &dev->states[next_state];

	/* enter the state and update stats */
	dev->last_state = target_state;

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

	dev->last_residency = target_state->enter(dev, target_state);

	trace_power_end(dev->cpu);
	trace_cpu_idle(PWR_EVENT_EXIT, dev->cpu);

	if (dev->last_state)
		target_state = dev->last_state;

	target_state->time += (unsigned long long)dev->last_residency;
	target_state->usage++;

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

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

/**
 * cpuidle_uninstall_idle_handler - uninstalls the cpuidle idle loop handler
 */
void cpuidle_uninstall_idle_handler(void)
{
	if (enabled_devices && pm_idle_old && (pm_idle != pm_idle_old)) {
		pm_idle = pm_idle_old;
		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, struct cpuidle_state *st)
{
	ktime_t	t1, t2;
	s64 diff;
	int ret;

	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;

	ret = (int) diff;
	return ret;
}

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

	cpuidle_set_statedata(state, 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[i].usage = 0;
		dev->states[i].time = 0;
	}
	dev->last_residency = 0;
	dev->last_state = NULL;

	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;

	pm_idle_old = pm_idle;

	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);
	28	static void (*pm_idle_old)(void);
	29
	30	static int enabled_devices;
62027aea LB	31	static int off __read_mostly;
	32
	33	int cpuidle_disabled(void)
	34	{
	35	return off;
	36	}
4f86d3a8	37
a6869cc4 VP	38	#if defined(CONFIG_ARCH_HAS_CPU_IDLE_WAIT)
	39	static void cpuidle_kick_cpus(void)
	40	{
	41	cpu_idle_wait();
	42	}
	43	#elif defined(CONFIG_SMP)
	44	# error "Arch needs cpu_idle_wait() equivalent here"
	45	#else /* !CONFIG_ARCH_HAS_CPU_IDLE_WAIT && !CONFIG_SMP */
	46	static void cpuidle_kick_cpus(void) {}
	47	#endif
	48
dcb84f33 VP	49	static int __cpuidle_register_device(struct cpuidle_device *dev);
dcb84f33 VP	50
4f86d3a8 LB	51	/**
	52	* cpuidle_idle_call - the main idle loop
	53	*
	54	* NOTE: no locks or semaphores should be used here
	55	*/
	56	static void cpuidle_idle_call(void)
	57	{
4a6f4fe8	58	struct cpuidle_device *dev = __this_cpu_read(cpuidle_devices);
4f86d3a8 LB	59	struct cpuidle_state *target_state;
	60	int next_state;
	61
	62	/* check if the device is ready */
	63	if (!dev \|\| !dev->enabled) {
	64	if (pm_idle_old)
	65	pm_idle_old();
	66	else
89cedfef VP	67	#if defined(CONFIG_ARCH_HAS_DEFAULT_IDLE)
	68	default_idle();
	69	#else
4f86d3a8	70	local_irq_enable();
89cedfef	71	#endif
4f86d3a8 LB	72	return;
	73	}
	74
9a655837 AV	75	#if 0
9a655837 AV	76	/* shows regressions, re-enable for 2.6.29 */
2e94d1f7 AV	77	/*
	78	* run any timers that can be run now, at this point
	79	* before calculating the idle duration etc.
	80	*/
	81	hrtimer_peek_ahead_timers();
9a655837	82	#endif
71abbbf8 AL	83
	84	/*
	85	* Call the device's prepare function before calling the
	86	* governor's select function. ->prepare gives the device's
	87	* cpuidle driver a chance to update any dynamic information
	88	* of its cpuidle states for the current idle period, e.g.
	89	* state availability, latencies, residencies, etc.
	90	*/
	91	if (dev->prepare)
	92	dev->prepare(dev);
	93
4f86d3a8 LB	94	/* ask the governor for the next state */
4f86d3a8 LB	95	next_state = cpuidle_curr_governor->select(dev);
246eb7f0 KH	96	if (need_resched()) {
246eb7f0 KH	97	local_irq_enable();
4f86d3a8	98	return;
246eb7f0 KH	99	}
246eb7f0 KH	100
4f86d3a8 LB	101	target_state = &dev->states[next_state];
	102
	103	/* enter the state and update stats */
4f86d3a8	104	dev->last_state = target_state;
f77cfe4e TR	105
	106	trace_power_start(POWER_CSTATE, next_state, dev->cpu);
	107	trace_cpu_idle(next_state, dev->cpu);
	108
887e301a	109	dev->last_residency = target_state->enter(dev, target_state);
f77cfe4e TR	110
	111	trace_power_end(dev->cpu);
	112	trace_cpu_idle(PWR_EVENT_EXIT, dev->cpu);
	113
887e301a VP	114	if (dev->last_state)
	115	target_state = dev->last_state;
	116
8b78cf60	117	target_state->time += (unsigned long long)dev->last_residency;
4f86d3a8 LB	118	target_state->usage++;
	119
	120	/* give the governor an opportunity to reflect on the outcome */
	121	if (cpuidle_curr_governor->reflect)
	122	cpuidle_curr_governor->reflect(dev);
	123	}
	124
	125	/**
	126	* cpuidle_install_idle_handler - installs the cpuidle idle loop handler
	127	*/
	128	void cpuidle_install_idle_handler(void)
	129	{
	130	if (enabled_devices && (pm_idle != cpuidle_idle_call)) {
	131	/* Make sure all changes finished before we switch to new idle */
	132	smp_wmb();
	133	pm_idle = cpuidle_idle_call;
	134	}
	135	}
	136
	137	/**
	138	* cpuidle_uninstall_idle_handler - uninstalls the cpuidle idle loop handler
	139	*/
	140	void cpuidle_uninstall_idle_handler(void)
	141	{
b032bf70	142	if (enabled_devices && pm_idle_old && (pm_idle != pm_idle_old)) {
4f86d3a8	143	pm_idle = pm_idle_old;
a6869cc4	144	cpuidle_kick_cpus();
4f86d3a8 LB	145	}
	146	}
	147
	148	/**
	149	* cpuidle_pause_and_lock - temporarily disables CPUIDLE
	150	*/
	151	void cpuidle_pause_and_lock(void)
	152	{
	153	mutex_lock(&cpuidle_lock);
	154	cpuidle_uninstall_idle_handler();
	155	}
	156
	157	EXPORT_SYMBOL_GPL(cpuidle_pause_and_lock);
	158
	159	/**
	160	* cpuidle_resume_and_unlock - resumes CPUIDLE operation
	161	*/
	162	void cpuidle_resume_and_unlock(void)
	163	{
	164	cpuidle_install_idle_handler();
	165	mutex_unlock(&cpuidle_lock);
	166	}
	167
	168	EXPORT_SYMBOL_GPL(cpuidle_resume_and_unlock);
	169
d8c216cf RW	170	#ifdef CONFIG_ARCH_HAS_CPU_RELAX
	171	static int poll_idle(struct cpuidle_device dev, struct cpuidle_state st)
	172	{
	173	ktime_t t1, t2;
	174	s64 diff;
	175	int ret;
	176
	177	t1 = ktime_get();
	178	local_irq_enable();
	179	while (!need_resched())
	180	cpu_relax();
	181
	182	t2 = ktime_get();
	183	diff = ktime_to_us(ktime_sub(t2, t1));
	184	if (diff > INT_MAX)
	185	diff = INT_MAX;
	186
	187	ret = (int) diff;
	188	return ret;
	189	}
	190
	191	static void poll_idle_init(struct cpuidle_device *dev)
	192	{
	193	struct cpuidle_state *state = &dev->states[0];
	194
	195	cpuidle_set_statedata(state, NULL);
	196
720f1c30	197	snprintf(state->name, CPUIDLE_NAME_LEN, "POLL");
d8c216cf RW	198	snprintf(state->desc, CPUIDLE_DESC_LEN, "CPUIDLE CORE POLL IDLE");
	199	state->exit_latency = 0;
	200	state->target_residency = 0;
	201	state->power_usage = -1;
d247632c	202	state->flags = 0;
d8c216cf RW	203	state->enter = poll_idle;
	204	}
	205	#else
	206	static void poll_idle_init(struct cpuidle_device *dev) {}
	207	#endif /* CONFIG_ARCH_HAS_CPU_RELAX */
	208
4f86d3a8 LB	209	/**
	210	* cpuidle_enable_device - enables idle PM for a CPU
	211	* @dev: the CPU
	212	*
	213	* This function must be called between cpuidle_pause_and_lock and
	214	* cpuidle_resume_and_unlock when used externally.
	215	*/
	216	int cpuidle_enable_device(struct cpuidle_device *dev)
	217	{
	218	int ret, i;
	219
	220	if (dev->enabled)
	221	return 0;
752138df	222	if (!cpuidle_get_driver() \|\| !cpuidle_curr_governor)
4f86d3a8 LB	223	return -EIO;
	224	if (!dev->state_count)
	225	return -EINVAL;
	226
dcb84f33 VP	227	if (dev->registered == 0) {
	228	ret = __cpuidle_register_device(dev);
	229	if (ret)
	230	return ret;
	231	}
	232
d8c216cf RW	233	poll_idle_init(dev);
d8c216cf RW	234
4f86d3a8 LB	235	if ((ret = cpuidle_add_state_sysfs(dev)))
	236	return ret;
	237
	238	if (cpuidle_curr_governor->enable &&
	239	(ret = cpuidle_curr_governor->enable(dev)))
	240	goto fail_sysfs;
	241
	242	for (i = 0; i < dev->state_count; i++) {
	243	dev->states[i].usage = 0;
	244	dev->states[i].time = 0;
	245	}
	246	dev->last_residency = 0;
	247	dev->last_state = NULL;
	248
	249	smp_wmb();
	250
	251	dev->enabled = 1;
	252
	253	enabled_devices++;
	254	return 0;
	255
	256	fail_sysfs:
	257	cpuidle_remove_state_sysfs(dev);
	258
	259	return ret;
	260	}
	261
	262	EXPORT_SYMBOL_GPL(cpuidle_enable_device);
	263
	264	/**
	265	* cpuidle_disable_device - disables idle PM for a CPU
	266	* @dev: the CPU
	267	*
	268	* This function must be called between cpuidle_pause_and_lock and
	269	* cpuidle_resume_and_unlock when used externally.
	270	*/
	271	void cpuidle_disable_device(struct cpuidle_device *dev)
	272	{
	273	if (!dev->enabled)
	274	return;
752138df	275	if (!cpuidle_get_driver() \|\| !cpuidle_curr_governor)
4f86d3a8 LB	276	return;
	277
	278	dev->enabled = 0;
	279
	280	if (cpuidle_curr_governor->disable)
	281	cpuidle_curr_governor->disable(dev);
	282
	283	cpuidle_remove_state_sysfs(dev);
	284	enabled_devices--;
	285	}
	286
	287	EXPORT_SYMBOL_GPL(cpuidle_disable_device);
	288
	289	/**
dcb84f33 VP	290	* __cpuidle_register_device - internal register function called before register
dcb84f33 VP	291	* and enable routines
4f86d3a8	292	* @dev: the cpu
dcb84f33 VP	293	*
dcb84f33 VP	294	* cpuidle_lock mutex must be held before this is called
4f86d3a8	295	*/
dcb84f33	296	static int __cpuidle_register_device(struct cpuidle_device *dev)
4f86d3a8 LB	297	{
	298	int ret;
	299	struct sys_device *sys_dev = get_cpu_sysdev((unsigned long)dev->cpu);
752138df	300	struct cpuidle_driver *cpuidle_driver = cpuidle_get_driver();
4f86d3a8 LB	301
	302	if (!sys_dev)
	303	return -EINVAL;
752138df	304	if (!try_module_get(cpuidle_driver->owner))
4f86d3a8 LB	305	return -EINVAL;
	306
	307	init_completion(&dev->kobj_unregister);
	308
71abbbf8 AL	309	/*
	310	* cpuidle driver should set the dev->power_specified bit
	311	* before registering the device if the driver provides
	312	* power_usage numbers.
	313	*
	314	* For those devices whose ->power_specified is not set,
	315	* we fill in power_usage with decreasing values as the
	316	* cpuidle code has an implicit assumption that state Cn
	317	* uses less power than C(n-1).
	318	*
	319	* With CONFIG_ARCH_HAS_CPU_RELAX, C0 is already assigned
	320	* an power value of -1. So we use -2, -3, etc, for other
	321	* c-states.
	322	*/
	323	if (!dev->power_specified) {
	324	int i;
	325	for (i = CPUIDLE_DRIVER_STATE_START; i < dev->state_count; i++)
	326	dev->states[i].power_usage = -1 - i;
	327	}
	328
4f86d3a8 LB	329	per_cpu(cpuidle_devices, dev->cpu) = dev;
	330	list_add(&dev->device_list, &cpuidle_detected_devices);
	331	if ((ret = cpuidle_add_sysfs(sys_dev))) {
752138df	332	module_put(cpuidle_driver->owner);
4f86d3a8 LB	333	return ret;
	334	}
	335
dcb84f33 VP	336	dev->registered = 1;
	337	return 0;
	338	}
	339
	340	/**
	341	* cpuidle_register_device - registers a CPU's idle PM feature
	342	* @dev: the cpu
	343	*/
	344	int cpuidle_register_device(struct cpuidle_device *dev)
	345	{
	346	int ret;
	347
	348	mutex_lock(&cpuidle_lock);
	349
	350	if ((ret = __cpuidle_register_device(dev))) {
	351	mutex_unlock(&cpuidle_lock);
	352	return ret;
	353	}
	354
4f86d3a8 LB	355	cpuidle_enable_device(dev);
	356	cpuidle_install_idle_handler();
	357
	358	mutex_unlock(&cpuidle_lock);
	359
	360	return 0;
	361
	362	}
	363
	364	EXPORT_SYMBOL_GPL(cpuidle_register_device);
	365
	366	/**
	367	* cpuidle_unregister_device - unregisters a CPU's idle PM feature
	368	* @dev: the cpu
	369	*/
	370	void cpuidle_unregister_device(struct cpuidle_device *dev)
	371	{
	372	struct sys_device *sys_dev = get_cpu_sysdev((unsigned long)dev->cpu);
752138df	373	struct cpuidle_driver *cpuidle_driver = cpuidle_get_driver();
4f86d3a8	374
dcb84f33 VP	375	if (dev->registered == 0)
	376	return;
	377
4f86d3a8 LB	378	cpuidle_pause_and_lock();
	379
	380	cpuidle_disable_device(dev);
	381
	382	cpuidle_remove_sysfs(sys_dev);
	383	list_del(&dev->device_list);
	384	wait_for_completion(&dev->kobj_unregister);
	385	per_cpu(cpuidle_devices, dev->cpu) = NULL;
	386
	387	cpuidle_resume_and_unlock();
	388
752138df	389	module_put(cpuidle_driver->owner);
4f86d3a8 LB	390	}
	391
	392	EXPORT_SYMBOL_GPL(cpuidle_unregister_device);
	393
	394	#ifdef CONFIG_SMP
	395
	396	static void smp_callback(void *v)
	397	{
	398	/* we already woke the CPU up, nothing more to do */
	399	}
	400
	401	/*
	402	* This function gets called when a part of the kernel has a new latency
	403	* requirement. This means we need to get all processors out of their C-state,
	404	* and then recalculate a new suitable C-state. Just do a cross-cpu IPI; that
	405	* wakes them all right up.
	406	*/
	407	static int cpuidle_latency_notify(struct notifier_block *b,
	408	unsigned long l, void *v)
	409	{
8691e5a8	410	smp_call_function(smp_callback, NULL, 1);
4f86d3a8 LB	411	return NOTIFY_OK;
	412	}
	413
	414	static struct notifier_block cpuidle_latency_notifier = {
	415	.notifier_call = cpuidle_latency_notify,
	416	};
	417
d82b3518 MG	418	static inline void latency_notifier_init(struct notifier_block *n)
	419	{
	420	pm_qos_add_notifier(PM_QOS_CPU_DMA_LATENCY, n);
	421	}
4f86d3a8 LB	422
	423	#else /* CONFIG_SMP */
	424
	425	#define latency_notifier_init(x) do { } while (0)
	426
	427	#endif /* CONFIG_SMP */
	428
	429	/**
	430	* cpuidle_init - core initializer
	431	*/
	432	static int __init cpuidle_init(void)
	433	{
	434	int ret;
	435
62027aea LB	436	if (cpuidle_disabled())
	437	return -ENODEV;
	438
4f86d3a8 LB	439	pm_idle_old = pm_idle;
	440
	441	ret = cpuidle_add_class_sysfs(&cpu_sysdev_class);
	442	if (ret)
	443	return ret;
	444
	445	latency_notifier_init(&cpuidle_latency_notifier);
	446
	447	return 0;
	448	}
	449
62027aea	450	module_param(off, int, 0444);
4f86d3a8	451	core_initcall(cpuidle_init);