cpufreq.c 62.4 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5
/*
 *  linux/drivers/cpufreq/cpufreq.c
 *
 *  Copyright (C) 2001 Russell King
 *            (C) 2002 - 2003 Dominik Brodowski <linux@brodo.de>
6
 *            (C) 2013 Viresh Kumar <viresh.kumar@linaro.org>
L
Linus Torvalds 已提交
7
 *
8
 *  Oct 2005 - Ashok Raj <ashok.raj@intel.com>
9
 *	Added handling for CPU hotplug
10 11
 *  Feb 2006 - Jacob Shin <jacob.shin@amd.com>
 *	Fix handling for CPU hotplug -- affected CPUs
12
 *
L
Linus Torvalds 已提交
13 14 15 16 17
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 */

V
Viresh Kumar 已提交
18 19
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

20
#include <linux/cpu.h>
L
Linus Torvalds 已提交
21 22 23
#include <linux/cpufreq.h>
#include <linux/delay.h>
#include <linux/device.h>
24 25 26
#include <linux/init.h>
#include <linux/kernel_stat.h>
#include <linux/module.h>
27
#include <linux/mutex.h>
28
#include <linux/slab.h>
29
#include <linux/suspend.h>
30
#include <linux/syscore_ops.h>
31
#include <linux/tick.h>
32 33
#include <trace/events/power.h>

V
Viresh Kumar 已提交
34
static LIST_HEAD(cpufreq_policy_list);
35 36 37 38 39 40 41

static inline bool policy_is_inactive(struct cpufreq_policy *policy)
{
	return cpumask_empty(policy->cpus);
}

/* Macros to iterate over CPU policies */
42 43 44
#define for_each_suitable_policy(__policy, __active)			 \
	list_for_each_entry(__policy, &cpufreq_policy_list, policy_list) \
		if ((__active) == !policy_is_inactive(__policy))
45 46 47 48 49 50 51

#define for_each_active_policy(__policy)		\
	for_each_suitable_policy(__policy, true)
#define for_each_inactive_policy(__policy)		\
	for_each_suitable_policy(__policy, false)

#define for_each_policy(__policy)			\
V
Viresh Kumar 已提交
52 53
	list_for_each_entry(__policy, &cpufreq_policy_list, policy_list)

54 55 56 57 58
/* Iterate over governors */
static LIST_HEAD(cpufreq_governor_list);
#define for_each_governor(__governor)				\
	list_for_each_entry(__governor, &cpufreq_governor_list, governor_list)

L
Linus Torvalds 已提交
59
/**
D
Dave Jones 已提交
60
 * The "cpufreq driver" - the arch- or hardware-dependent low
L
Linus Torvalds 已提交
61 62 63
 * level driver of CPUFreq support, and its spinlock. This lock
 * also protects the cpufreq_cpu_data array.
 */
64
static struct cpufreq_driver *cpufreq_driver;
65
static DEFINE_PER_CPU(struct cpufreq_policy *, cpufreq_cpu_data);
66 67
static DEFINE_RWLOCK(cpufreq_driver_lock);

68 69
/* Flag to suspend/resume CPUFreq governors */
static bool cpufreq_suspended;
L
Linus Torvalds 已提交
70

71 72 73 74 75
static inline bool has_target(void)
{
	return cpufreq_driver->target_index || cpufreq_driver->target;
}

L
Linus Torvalds 已提交
76
/* internal prototypes */
77
static int cpufreq_governor(struct cpufreq_policy *policy, unsigned int event);
78
static unsigned int __cpufreq_get(struct cpufreq_policy *policy);
L
Linus Torvalds 已提交
79 80

/**
81 82
 * Two notifier lists: the "policy" list is involved in the
 * validation process for a new CPU frequency policy; the
L
Linus Torvalds 已提交
83 84 85 86
 * "transition" list for kernel code that needs to handle
 * changes to devices when the CPU clock speed changes.
 * The mutex locks both lists.
 */
87
static BLOCKING_NOTIFIER_HEAD(cpufreq_policy_notifier_list);
88
static struct srcu_notifier_head cpufreq_transition_notifier_list;
L
Linus Torvalds 已提交
89

90
static bool init_cpufreq_transition_notifier_list_called;
91 92 93
static int __init init_cpufreq_transition_notifier_list(void)
{
	srcu_init_notifier_head(&cpufreq_transition_notifier_list);
94
	init_cpufreq_transition_notifier_list_called = true;
95 96
	return 0;
}
97
pure_initcall(init_cpufreq_transition_notifier_list);
L
Linus Torvalds 已提交
98

99
static int off __read_mostly;
100
static int cpufreq_disabled(void)
101 102 103 104 105 106 107
{
	return off;
}
void disable_cpufreq(void)
{
	off = 1;
}
108
static DEFINE_MUTEX(cpufreq_governor_mutex);
L
Linus Torvalds 已提交
109

110 111
bool have_governor_per_policy(void)
{
112
	return !!(cpufreq_driver->flags & CPUFREQ_HAVE_GOVERNOR_PER_POLICY);
113
}
114
EXPORT_SYMBOL_GPL(have_governor_per_policy);
115

116 117 118 119 120 121 122 123 124
struct kobject *get_governor_parent_kobj(struct cpufreq_policy *policy)
{
	if (have_governor_per_policy())
		return &policy->kobj;
	else
		return cpufreq_global_kobject;
}
EXPORT_SYMBOL_GPL(get_governor_parent_kobj);

125 126 127 128 129 130 131 132 133
struct cpufreq_frequency_table *cpufreq_frequency_get_table(unsigned int cpu)
{
	struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);

	return policy && !policy_is_inactive(policy) ?
		policy->freq_table : NULL;
}
EXPORT_SYMBOL_GPL(cpufreq_frequency_get_table);

134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168
static inline u64 get_cpu_idle_time_jiffy(unsigned int cpu, u64 *wall)
{
	u64 idle_time;
	u64 cur_wall_time;
	u64 busy_time;

	cur_wall_time = jiffies64_to_cputime64(get_jiffies_64());

	busy_time = kcpustat_cpu(cpu).cpustat[CPUTIME_USER];
	busy_time += kcpustat_cpu(cpu).cpustat[CPUTIME_SYSTEM];
	busy_time += kcpustat_cpu(cpu).cpustat[CPUTIME_IRQ];
	busy_time += kcpustat_cpu(cpu).cpustat[CPUTIME_SOFTIRQ];
	busy_time += kcpustat_cpu(cpu).cpustat[CPUTIME_STEAL];
	busy_time += kcpustat_cpu(cpu).cpustat[CPUTIME_NICE];

	idle_time = cur_wall_time - busy_time;
	if (wall)
		*wall = cputime_to_usecs(cur_wall_time);

	return cputime_to_usecs(idle_time);
}

u64 get_cpu_idle_time(unsigned int cpu, u64 *wall, int io_busy)
{
	u64 idle_time = get_cpu_idle_time_us(cpu, io_busy ? wall : NULL);

	if (idle_time == -1ULL)
		return get_cpu_idle_time_jiffy(cpu, wall);
	else if (!io_busy)
		idle_time += get_cpu_iowait_time_us(cpu, wall);

	return idle_time;
}
EXPORT_SYMBOL_GPL(get_cpu_idle_time);

169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190
/*
 * This is a generic cpufreq init() routine which can be used by cpufreq
 * drivers of SMP systems. It will do following:
 * - validate & show freq table passed
 * - set policies transition latency
 * - policy->cpus with all possible CPUs
 */
int cpufreq_generic_init(struct cpufreq_policy *policy,
		struct cpufreq_frequency_table *table,
		unsigned int transition_latency)
{
	int ret;

	ret = cpufreq_table_validate_and_show(policy, table);
	if (ret) {
		pr_err("%s: invalid frequency table: %d\n", __func__, ret);
		return ret;
	}

	policy->cpuinfo.transition_latency = transition_latency;

	/*
191
	 * The driver only supports the SMP configuration where all processors
192 193 194 195 196 197 198 199
	 * share the clock and voltage and clock.
	 */
	cpumask_setall(policy->cpus);

	return 0;
}
EXPORT_SYMBOL_GPL(cpufreq_generic_init);

200
struct cpufreq_policy *cpufreq_cpu_get_raw(unsigned int cpu)
201 202 203
{
	struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);

204 205
	return policy && cpumask_test_cpu(cpu, policy->cpus) ? policy : NULL;
}
206
EXPORT_SYMBOL_GPL(cpufreq_cpu_get_raw);
207 208 209 210 211

unsigned int cpufreq_generic_get(unsigned int cpu)
{
	struct cpufreq_policy *policy = cpufreq_cpu_get_raw(cpu);

212
	if (!policy || IS_ERR(policy->clk)) {
213 214
		pr_err("%s: No %s associated to cpu: %d\n",
		       __func__, policy ? "clk" : "policy", cpu);
215 216 217 218 219 220 221
		return 0;
	}

	return clk_get_rate(policy->clk) / 1000;
}
EXPORT_SYMBOL_GPL(cpufreq_generic_get);

222 223 224 225 226 227 228 229 230 231 232 233 234
/**
 * cpufreq_cpu_get: returns policy for a cpu and marks it busy.
 *
 * @cpu: cpu to find policy for.
 *
 * This returns policy for 'cpu', returns NULL if it doesn't exist.
 * It also increments the kobject reference count to mark it busy and so would
 * require a corresponding call to cpufreq_cpu_put() to decrement it back.
 * If corresponding call cpufreq_cpu_put() isn't made, the policy wouldn't be
 * freed as that depends on the kobj count.
 *
 * Return: A valid policy on success, otherwise NULL on failure.
 */
235
struct cpufreq_policy *cpufreq_cpu_get(unsigned int cpu)
L
Linus Torvalds 已提交
236
{
237
	struct cpufreq_policy *policy = NULL;
L
Linus Torvalds 已提交
238 239
	unsigned long flags;

240
	if (WARN_ON(cpu >= nr_cpu_ids))
241 242
		return NULL;

L
Linus Torvalds 已提交
243
	/* get the cpufreq driver */
244
	read_lock_irqsave(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
245

246 247
	if (cpufreq_driver) {
		/* get the CPU */
248
		policy = cpufreq_cpu_get_raw(cpu);
249 250 251
		if (policy)
			kobject_get(&policy->kobj);
	}
L
Linus Torvalds 已提交
252

253
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
254

255
	return policy;
256
}
L
Linus Torvalds 已提交
257 258
EXPORT_SYMBOL_GPL(cpufreq_cpu_get);

259 260 261 262 263 264 265 266
/**
 * cpufreq_cpu_put: Decrements the usage count of a policy
 *
 * @policy: policy earlier returned by cpufreq_cpu_get().
 *
 * This decrements the kobject reference count incremented earlier by calling
 * cpufreq_cpu_get().
 */
267
void cpufreq_cpu_put(struct cpufreq_policy *policy)
L
Linus Torvalds 已提交
268
{
269
	kobject_put(&policy->kobj);
L
Linus Torvalds 已提交
270 271 272 273 274 275 276 277 278 279 280 281
}
EXPORT_SYMBOL_GPL(cpufreq_cpu_put);

/*********************************************************************
 *            EXTERNALLY AFFECTING FREQUENCY CHANGES                 *
 *********************************************************************/

/**
 * adjust_jiffies - adjust the system "loops_per_jiffy"
 *
 * This function alters the system "loops_per_jiffy" for the clock
 * speed change. Note that loops_per_jiffy cannot be updated on SMP
282
 * systems as each CPU might be scaled differently. So, use the arch
L
Linus Torvalds 已提交
283 284
 * per-CPU loops_per_jiffy value wherever possible.
 */
285
static void adjust_jiffies(unsigned long val, struct cpufreq_freqs *ci)
L
Linus Torvalds 已提交
286
{
287 288 289 290
#ifndef CONFIG_SMP
	static unsigned long l_p_j_ref;
	static unsigned int l_p_j_ref_freq;

L
Linus Torvalds 已提交
291 292 293 294 295 296
	if (ci->flags & CPUFREQ_CONST_LOOPS)
		return;

	if (!l_p_j_ref_freq) {
		l_p_j_ref = loops_per_jiffy;
		l_p_j_ref_freq = ci->old;
297 298
		pr_debug("saving %lu as reference value for loops_per_jiffy; freq is %u kHz\n",
			 l_p_j_ref, l_p_j_ref_freq);
L
Linus Torvalds 已提交
299
	}
300
	if (val == CPUFREQ_POSTCHANGE && ci->old != ci->new) {
301 302
		loops_per_jiffy = cpufreq_scale(l_p_j_ref, l_p_j_ref_freq,
								ci->new);
303 304
		pr_debug("scaling loops_per_jiffy to %lu for frequency %u kHz\n",
			 loops_per_jiffy, ci->new);
L
Linus Torvalds 已提交
305 306
	}
#endif
307
}
L
Linus Torvalds 已提交
308

309
static void __cpufreq_notify_transition(struct cpufreq_policy *policy,
310
		struct cpufreq_freqs *freqs, unsigned int state)
L
Linus Torvalds 已提交
311 312 313
{
	BUG_ON(irqs_disabled());

314 315 316
	if (cpufreq_disabled())
		return;

317
	freqs->flags = cpufreq_driver->flags;
318
	pr_debug("notification %u of frequency transition to %u kHz\n",
319
		 state, freqs->new);
L
Linus Torvalds 已提交
320 321

	switch (state) {
322

L
Linus Torvalds 已提交
323
	case CPUFREQ_PRECHANGE:
324
		/* detect if the driver reported a value as "old frequency"
325 326
		 * which is not equal to what the cpufreq core thinks is
		 * "old frequency".
L
Linus Torvalds 已提交
327
		 */
328
		if (!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
329 330
			if ((policy) && (policy->cpu == freqs->cpu) &&
			    (policy->cur) && (policy->cur != freqs->old)) {
331 332
				pr_debug("Warning: CPU frequency is %u, cpufreq assumed %u kHz\n",
					 freqs->old, policy->cur);
333
				freqs->old = policy->cur;
L
Linus Torvalds 已提交
334 335
			}
		}
336
		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
337
				CPUFREQ_PRECHANGE, freqs);
L
Linus Torvalds 已提交
338 339
		adjust_jiffies(CPUFREQ_PRECHANGE, freqs);
		break;
340

L
Linus Torvalds 已提交
341 342
	case CPUFREQ_POSTCHANGE:
		adjust_jiffies(CPUFREQ_POSTCHANGE, freqs);
343 344
		pr_debug("FREQ: %lu - CPU: %lu\n",
			 (unsigned long)freqs->new, (unsigned long)freqs->cpu);
345
		trace_cpu_frequency(freqs->new, freqs->cpu);
346
		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
347
				CPUFREQ_POSTCHANGE, freqs);
348 349
		if (likely(policy) && likely(policy->cpu == freqs->cpu))
			policy->cur = freqs->new;
L
Linus Torvalds 已提交
350 351 352
		break;
	}
}
353

354 355 356 357 358 359 360 361
/**
 * cpufreq_notify_transition - call notifier chain and adjust_jiffies
 * on frequency transition.
 *
 * This function calls the transition notifiers and the "adjust_jiffies"
 * function. It is called twice on all CPU frequency changes that have
 * external effects.
 */
362
static void cpufreq_notify_transition(struct cpufreq_policy *policy,
363 364 365 366 367
		struct cpufreq_freqs *freqs, unsigned int state)
{
	for_each_cpu(freqs->cpu, policy->cpus)
		__cpufreq_notify_transition(policy, freqs, state);
}
L
Linus Torvalds 已提交
368

369
/* Do post notifications when there are chances that transition has failed */
370
static void cpufreq_notify_post_transition(struct cpufreq_policy *policy,
371 372 373 374 375 376 377 378 379 380 381
		struct cpufreq_freqs *freqs, int transition_failed)
{
	cpufreq_notify_transition(policy, freqs, CPUFREQ_POSTCHANGE);
	if (!transition_failed)
		return;

	swap(freqs->old, freqs->new);
	cpufreq_notify_transition(policy, freqs, CPUFREQ_PRECHANGE);
	cpufreq_notify_transition(policy, freqs, CPUFREQ_POSTCHANGE);
}

382 383 384
void cpufreq_freq_transition_begin(struct cpufreq_policy *policy,
		struct cpufreq_freqs *freqs)
{
385 386 387 388 389 390 391 392 393 394 395 396

	/*
	 * Catch double invocations of _begin() which lead to self-deadlock.
	 * ASYNC_NOTIFICATION drivers are left out because the cpufreq core
	 * doesn't invoke _begin() on their behalf, and hence the chances of
	 * double invocations are very low. Moreover, there are scenarios
	 * where these checks can emit false-positive warnings in these
	 * drivers; so we avoid that by skipping them altogether.
	 */
	WARN_ON(!(cpufreq_driver->flags & CPUFREQ_ASYNC_NOTIFICATION)
				&& current == policy->transition_task);

397 398 399 400 401 402 403 404 405 406 407
wait:
	wait_event(policy->transition_wait, !policy->transition_ongoing);

	spin_lock(&policy->transition_lock);

	if (unlikely(policy->transition_ongoing)) {
		spin_unlock(&policy->transition_lock);
		goto wait;
	}

	policy->transition_ongoing = true;
408
	policy->transition_task = current;
409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424

	spin_unlock(&policy->transition_lock);

	cpufreq_notify_transition(policy, freqs, CPUFREQ_PRECHANGE);
}
EXPORT_SYMBOL_GPL(cpufreq_freq_transition_begin);

void cpufreq_freq_transition_end(struct cpufreq_policy *policy,
		struct cpufreq_freqs *freqs, int transition_failed)
{
	if (unlikely(WARN_ON(!policy->transition_ongoing)))
		return;

	cpufreq_notify_post_transition(policy, freqs, transition_failed);

	policy->transition_ongoing = false;
425
	policy->transition_task = NULL;
426 427 428 429 430

	wake_up(&policy->transition_wait);
}
EXPORT_SYMBOL_GPL(cpufreq_freq_transition_end);

L
Linus Torvalds 已提交
431 432 433 434

/*********************************************************************
 *                          SYSFS INTERFACE                          *
 *********************************************************************/
435
static ssize_t show_boost(struct kobject *kobj,
436 437 438 439 440 441 442 443 444 445 446 447 448 449 450
				 struct attribute *attr, char *buf)
{
	return sprintf(buf, "%d\n", cpufreq_driver->boost_enabled);
}

static ssize_t store_boost(struct kobject *kobj, struct attribute *attr,
				  const char *buf, size_t count)
{
	int ret, enable;

	ret = sscanf(buf, "%d", &enable);
	if (ret != 1 || enable < 0 || enable > 1)
		return -EINVAL;

	if (cpufreq_boost_trigger_state(enable)) {
451 452
		pr_err("%s: Cannot %s BOOST!\n",
		       __func__, enable ? "enable" : "disable");
453 454 455
		return -EINVAL;
	}

456 457
	pr_debug("%s: cpufreq BOOST %s\n",
		 __func__, enable ? "enabled" : "disabled");
458 459 460 461

	return count;
}
define_one_global_rw(boost);
L
Linus Torvalds 已提交
462

463
static struct cpufreq_governor *find_governor(const char *str_governor)
464 465 466
{
	struct cpufreq_governor *t;

467
	for_each_governor(t)
468
		if (!strncasecmp(str_governor, t->name, CPUFREQ_NAME_LEN))
469 470 471 472 473
			return t;

	return NULL;
}

L
Linus Torvalds 已提交
474 475 476
/**
 * cpufreq_parse_governor - parse a governor string
 */
D
Dave Jones 已提交
477
static int cpufreq_parse_governor(char *str_governor, unsigned int *policy,
L
Linus Torvalds 已提交
478 479
				struct cpufreq_governor **governor)
{
480
	int err = -EINVAL;
481 482

	if (cpufreq_driver->setpolicy) {
483
		if (!strncasecmp(str_governor, "performance", CPUFREQ_NAME_LEN)) {
L
Linus Torvalds 已提交
484
			*policy = CPUFREQ_POLICY_PERFORMANCE;
485
			err = 0;
486
		} else if (!strncasecmp(str_governor, "powersave",
487
						CPUFREQ_NAME_LEN)) {
L
Linus Torvalds 已提交
488
			*policy = CPUFREQ_POLICY_POWERSAVE;
489
			err = 0;
L
Linus Torvalds 已提交
490
		}
491
	} else {
L
Linus Torvalds 已提交
492
		struct cpufreq_governor *t;
493

494
		mutex_lock(&cpufreq_governor_mutex);
495

496
		t = find_governor(str_governor);
497

498
		if (t == NULL) {
499
			int ret;
500

501 502 503
			mutex_unlock(&cpufreq_governor_mutex);
			ret = request_module("cpufreq_%s", str_governor);
			mutex_lock(&cpufreq_governor_mutex);
504

505
			if (ret == 0)
506
				t = find_governor(str_governor);
507 508
		}

509 510 511
		if (t != NULL) {
			*governor = t;
			err = 0;
L
Linus Torvalds 已提交
512
		}
513

514
		mutex_unlock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
515
	}
516
	return err;
L
Linus Torvalds 已提交
517 518 519
}

/**
520 521
 * cpufreq_per_cpu_attr_read() / show_##file_name() -
 * print out cpufreq information
L
Linus Torvalds 已提交
522 523 524 525 526
 *
 * Write out information from cpufreq_driver->policy[cpu]; object must be
 * "unsigned int".
 */

527 528
#define show_one(file_name, object)			\
static ssize_t show_##file_name				\
D
Dave Jones 已提交
529
(struct cpufreq_policy *policy, char *buf)		\
530
{							\
531
	return sprintf(buf, "%u\n", policy->object);	\
L
Linus Torvalds 已提交
532 533 534 535
}

show_one(cpuinfo_min_freq, cpuinfo.min_freq);
show_one(cpuinfo_max_freq, cpuinfo.max_freq);
536
show_one(cpuinfo_transition_latency, cpuinfo.transition_latency);
L
Linus Torvalds 已提交
537 538
show_one(scaling_min_freq, min);
show_one(scaling_max_freq, max);
539

540
static ssize_t show_scaling_cur_freq(struct cpufreq_policy *policy, char *buf)
541 542 543 544 545 546 547 548 549
{
	ssize_t ret;

	if (cpufreq_driver && cpufreq_driver->setpolicy && cpufreq_driver->get)
		ret = sprintf(buf, "%u\n", cpufreq_driver->get(policy->cpu));
	else
		ret = sprintf(buf, "%u\n", policy->cur);
	return ret;
}
L
Linus Torvalds 已提交
550

551
static int cpufreq_set_policy(struct cpufreq_policy *policy,
552
				struct cpufreq_policy *new_policy);
553

L
Linus Torvalds 已提交
554 555 556 557 558
/**
 * cpufreq_per_cpu_attr_write() / store_##file_name() - sysfs write access
 */
#define store_one(file_name, object)			\
static ssize_t store_##file_name					\
D
Dave Jones 已提交
559
(struct cpufreq_policy *policy, const char *buf, size_t count)		\
L
Linus Torvalds 已提交
560
{									\
561
	int ret, temp;							\
L
Linus Torvalds 已提交
562 563
	struct cpufreq_policy new_policy;				\
									\
564
	memcpy(&new_policy, policy, sizeof(*policy));			\
L
Linus Torvalds 已提交
565
									\
566
	ret = sscanf(buf, "%u", &new_policy.object);			\
L
Linus Torvalds 已提交
567 568 569
	if (ret != 1)							\
		return -EINVAL;						\
									\
570
	temp = new_policy.object;					\
571
	ret = cpufreq_set_policy(policy, &new_policy);		\
572 573
	if (!ret)							\
		policy->user_policy.object = temp;			\
L
Linus Torvalds 已提交
574 575 576 577
									\
	return ret ? ret : count;					\
}

578 579
store_one(scaling_min_freq, min);
store_one(scaling_max_freq, max);
L
Linus Torvalds 已提交
580 581 582 583

/**
 * show_cpuinfo_cur_freq - current CPU frequency as detected by hardware
 */
D
Dave Jones 已提交
584 585
static ssize_t show_cpuinfo_cur_freq(struct cpufreq_policy *policy,
					char *buf)
L
Linus Torvalds 已提交
586
{
587
	unsigned int cur_freq = __cpufreq_get(policy);
L
Linus Torvalds 已提交
588 589 590 591 592 593 594 595
	if (!cur_freq)
		return sprintf(buf, "<unknown>");
	return sprintf(buf, "%u\n", cur_freq);
}

/**
 * show_scaling_governor - show the current policy for the specified CPU
 */
D
Dave Jones 已提交
596
static ssize_t show_scaling_governor(struct cpufreq_policy *policy, char *buf)
L
Linus Torvalds 已提交
597
{
598
	if (policy->policy == CPUFREQ_POLICY_POWERSAVE)
L
Linus Torvalds 已提交
599 600 601 602
		return sprintf(buf, "powersave\n");
	else if (policy->policy == CPUFREQ_POLICY_PERFORMANCE)
		return sprintf(buf, "performance\n");
	else if (policy->governor)
603
		return scnprintf(buf, CPUFREQ_NAME_PLEN, "%s\n",
604
				policy->governor->name);
L
Linus Torvalds 已提交
605 606 607 608 609 610
	return -EINVAL;
}

/**
 * store_scaling_governor - store policy for the specified CPU
 */
D
Dave Jones 已提交
611 612
static ssize_t store_scaling_governor(struct cpufreq_policy *policy,
					const char *buf, size_t count)
L
Linus Torvalds 已提交
613
{
614
	int ret;
L
Linus Torvalds 已提交
615 616 617
	char	str_governor[16];
	struct cpufreq_policy new_policy;

618
	memcpy(&new_policy, policy, sizeof(*policy));
L
Linus Torvalds 已提交
619

620
	ret = sscanf(buf, "%15s", str_governor);
L
Linus Torvalds 已提交
621 622 623
	if (ret != 1)
		return -EINVAL;

624 625
	if (cpufreq_parse_governor(str_governor, &new_policy.policy,
						&new_policy.governor))
L
Linus Torvalds 已提交
626 627
		return -EINVAL;

628
	ret = cpufreq_set_policy(policy, &new_policy);
629
	return ret ? ret : count;
L
Linus Torvalds 已提交
630 631 632 633 634
}

/**
 * show_scaling_driver - show the cpufreq driver currently loaded
 */
D
Dave Jones 已提交
635
static ssize_t show_scaling_driver(struct cpufreq_policy *policy, char *buf)
L
Linus Torvalds 已提交
636
{
637
	return scnprintf(buf, CPUFREQ_NAME_PLEN, "%s\n", cpufreq_driver->name);
L
Linus Torvalds 已提交
638 639 640 641 642
}

/**
 * show_scaling_available_governors - show the available CPUfreq governors
 */
D
Dave Jones 已提交
643 644
static ssize_t show_scaling_available_governors(struct cpufreq_policy *policy,
						char *buf)
L
Linus Torvalds 已提交
645 646 647 648
{
	ssize_t i = 0;
	struct cpufreq_governor *t;

649
	if (!has_target()) {
L
Linus Torvalds 已提交
650 651 652 653
		i += sprintf(buf, "performance powersave");
		goto out;
	}

654
	for_each_governor(t) {
655 656
		if (i >= (ssize_t) ((PAGE_SIZE / sizeof(char))
		    - (CPUFREQ_NAME_LEN + 2)))
L
Linus Torvalds 已提交
657
			goto out;
658
		i += scnprintf(&buf[i], CPUFREQ_NAME_PLEN, "%s ", t->name);
L
Linus Torvalds 已提交
659
	}
660
out:
L
Linus Torvalds 已提交
661 662 663
	i += sprintf(&buf[i], "\n");
	return i;
}
664

665
ssize_t cpufreq_show_cpus(const struct cpumask *mask, char *buf)
L
Linus Torvalds 已提交
666 667 668 669
{
	ssize_t i = 0;
	unsigned int cpu;

670
	for_each_cpu(cpu, mask) {
L
Linus Torvalds 已提交
671 672 673 674
		if (i)
			i += scnprintf(&buf[i], (PAGE_SIZE - i - 2), " ");
		i += scnprintf(&buf[i], (PAGE_SIZE - i - 2), "%u", cpu);
		if (i >= (PAGE_SIZE - 5))
675
			break;
L
Linus Torvalds 已提交
676 677 678 679
	}
	i += sprintf(&buf[i], "\n");
	return i;
}
680
EXPORT_SYMBOL_GPL(cpufreq_show_cpus);
L
Linus Torvalds 已提交
681

682 683 684 685 686 687
/**
 * show_related_cpus - show the CPUs affected by each transition even if
 * hw coordination is in use
 */
static ssize_t show_related_cpus(struct cpufreq_policy *policy, char *buf)
{
688
	return cpufreq_show_cpus(policy->related_cpus, buf);
689 690 691 692 693 694 695
}

/**
 * show_affected_cpus - show the CPUs affected by each transition
 */
static ssize_t show_affected_cpus(struct cpufreq_policy *policy, char *buf)
{
696
	return cpufreq_show_cpus(policy->cpus, buf);
697 698
}

699
static ssize_t store_scaling_setspeed(struct cpufreq_policy *policy,
D
Dave Jones 已提交
700
					const char *buf, size_t count)
701 702 703 704
{
	unsigned int freq = 0;
	unsigned int ret;

705
	if (!policy->governor || !policy->governor->store_setspeed)
706 707 708 709 710 711 712 713 714 715 716 717 718
		return -EINVAL;

	ret = sscanf(buf, "%u", &freq);
	if (ret != 1)
		return -EINVAL;

	policy->governor->store_setspeed(policy, freq);

	return count;
}

static ssize_t show_scaling_setspeed(struct cpufreq_policy *policy, char *buf)
{
719
	if (!policy->governor || !policy->governor->show_setspeed)
720 721 722 723
		return sprintf(buf, "<unsupported>\n");

	return policy->governor->show_setspeed(policy, buf);
}
L
Linus Torvalds 已提交
724

725
/**
726
 * show_bios_limit - show the current cpufreq HW/BIOS limitation
727 728 729 730 731
 */
static ssize_t show_bios_limit(struct cpufreq_policy *policy, char *buf)
{
	unsigned int limit;
	int ret;
732 733
	if (cpufreq_driver->bios_limit) {
		ret = cpufreq_driver->bios_limit(policy->cpu, &limit);
734 735 736 737 738 739
		if (!ret)
			return sprintf(buf, "%u\n", limit);
	}
	return sprintf(buf, "%u\n", policy->cpuinfo.max_freq);
}

740 741 742 743 744 745 746 747 748 749 750 751 752 753
cpufreq_freq_attr_ro_perm(cpuinfo_cur_freq, 0400);
cpufreq_freq_attr_ro(cpuinfo_min_freq);
cpufreq_freq_attr_ro(cpuinfo_max_freq);
cpufreq_freq_attr_ro(cpuinfo_transition_latency);
cpufreq_freq_attr_ro(scaling_available_governors);
cpufreq_freq_attr_ro(scaling_driver);
cpufreq_freq_attr_ro(scaling_cur_freq);
cpufreq_freq_attr_ro(bios_limit);
cpufreq_freq_attr_ro(related_cpus);
cpufreq_freq_attr_ro(affected_cpus);
cpufreq_freq_attr_rw(scaling_min_freq);
cpufreq_freq_attr_rw(scaling_max_freq);
cpufreq_freq_attr_rw(scaling_governor);
cpufreq_freq_attr_rw(scaling_setspeed);
L
Linus Torvalds 已提交
754

D
Dave Jones 已提交
755
static struct attribute *default_attrs[] = {
L
Linus Torvalds 已提交
756 757
	&cpuinfo_min_freq.attr,
	&cpuinfo_max_freq.attr,
758
	&cpuinfo_transition_latency.attr,
L
Linus Torvalds 已提交
759 760 761
	&scaling_min_freq.attr,
	&scaling_max_freq.attr,
	&affected_cpus.attr,
762
	&related_cpus.attr,
L
Linus Torvalds 已提交
763 764 765
	&scaling_governor.attr,
	&scaling_driver.attr,
	&scaling_available_governors.attr,
766
	&scaling_setspeed.attr,
L
Linus Torvalds 已提交
767 768 769
	NULL
};

770 771
#define to_policy(k) container_of(k, struct cpufreq_policy, kobj)
#define to_attr(a) container_of(a, struct freq_attr, attr)
L
Linus Torvalds 已提交
772

773
static ssize_t show(struct kobject *kobj, struct attribute *attr, char *buf)
L
Linus Torvalds 已提交
774
{
D
Dave Jones 已提交
775 776
	struct cpufreq_policy *policy = to_policy(kobj);
	struct freq_attr *fattr = to_attr(attr);
777
	ssize_t ret;
778

779
	down_read(&policy->rwsem);
780
	ret = fattr->show(policy, buf);
781
	up_read(&policy->rwsem);
782

L
Linus Torvalds 已提交
783 784 785
	return ret;
}

D
Dave Jones 已提交
786 787
static ssize_t store(struct kobject *kobj, struct attribute *attr,
		     const char *buf, size_t count)
L
Linus Torvalds 已提交
788
{
D
Dave Jones 已提交
789 790
	struct cpufreq_policy *policy = to_policy(kobj);
	struct freq_attr *fattr = to_attr(attr);
791
	ssize_t ret = -EINVAL;
792

793 794
	get_online_cpus();

795 796
	if (cpu_online(policy->cpu)) {
		down_write(&policy->rwsem);
797
		ret = fattr->store(policy, buf, count);
798 799
		up_write(&policy->rwsem);
	}
800

801 802
	put_online_cpus();

L
Linus Torvalds 已提交
803 804 805
	return ret;
}

D
Dave Jones 已提交
806
static void cpufreq_sysfs_release(struct kobject *kobj)
L
Linus Torvalds 已提交
807
{
D
Dave Jones 已提交
808
	struct cpufreq_policy *policy = to_policy(kobj);
809
	pr_debug("last reference is dropped\n");
L
Linus Torvalds 已提交
810 811 812
	complete(&policy->kobj_unregister);
}

813
static const struct sysfs_ops sysfs_ops = {
L
Linus Torvalds 已提交
814 815 816 817 818 819 820 821 822 823
	.show	= show,
	.store	= store,
};

static struct kobj_type ktype_cpufreq = {
	.sysfs_ops	= &sysfs_ops,
	.default_attrs	= default_attrs,
	.release	= cpufreq_sysfs_release,
};

824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853
static int add_cpu_dev_symlink(struct cpufreq_policy *policy, int cpu)
{
	struct device *cpu_dev;

	pr_debug("%s: Adding symlink for CPU: %u\n", __func__, cpu);

	if (!policy)
		return 0;

	cpu_dev = get_cpu_device(cpu);
	if (WARN_ON(!cpu_dev))
		return 0;

	return sysfs_create_link(&cpu_dev->kobj, &policy->kobj, "cpufreq");
}

static void remove_cpu_dev_symlink(struct cpufreq_policy *policy, int cpu)
{
	struct device *cpu_dev;

	pr_debug("%s: Removing symlink for CPU: %u\n", __func__, cpu);

	cpu_dev = get_cpu_device(cpu);
	if (WARN_ON(!cpu_dev))
		return;

	sysfs_remove_link(&cpu_dev->kobj, "cpufreq");
}

/* Add/remove symlinks for all related CPUs */
854
static int cpufreq_add_dev_symlink(struct cpufreq_policy *policy)
855 856 857 858
{
	unsigned int j;
	int ret = 0;

859
	/* Some related CPUs might not be present (physically hotplugged) */
860
	for_each_cpu(j, policy->real_cpus) {
861
		ret = add_cpu_dev_symlink(policy, j);
862 863
		if (ret)
			break;
864
	}
865

866 867 868
	return ret;
}

869 870 871 872 873
static void cpufreq_remove_dev_symlink(struct cpufreq_policy *policy)
{
	unsigned int j;

	/* Some related CPUs might not be present (physically hotplugged) */
874
	for_each_cpu(j, policy->real_cpus)
875 876 877
		remove_cpu_dev_symlink(policy, j);
}

878
static int cpufreq_add_dev_interface(struct cpufreq_policy *policy)
879 880 881 882 883
{
	struct freq_attr **drv_attr;
	int ret = 0;

	/* set up files for this cpu device */
884
	drv_attr = cpufreq_driver->attr;
V
Viresh Kumar 已提交
885
	while (drv_attr && *drv_attr) {
886 887
		ret = sysfs_create_file(&policy->kobj, &((*drv_attr)->attr));
		if (ret)
888
			return ret;
889 890
		drv_attr++;
	}
891
	if (cpufreq_driver->get) {
892 893
		ret = sysfs_create_file(&policy->kobj, &cpuinfo_cur_freq.attr);
		if (ret)
894
			return ret;
895
	}
896 897 898

	ret = sysfs_create_file(&policy->kobj, &scaling_cur_freq.attr);
	if (ret)
899
		return ret;
900

901
	if (cpufreq_driver->bios_limit) {
902 903
		ret = sysfs_create_file(&policy->kobj, &bios_limit.attr);
		if (ret)
904
			return ret;
905
	}
906

907
	return cpufreq_add_dev_symlink(policy);
908 909
}

910 911 912 913 914
__weak struct cpufreq_governor *cpufreq_default_governor(void)
{
	return NULL;
}

915
static int cpufreq_init_policy(struct cpufreq_policy *policy)
916
{
917
	struct cpufreq_governor *gov = NULL;
918 919
	struct cpufreq_policy new_policy;

920
	memcpy(&new_policy, policy, sizeof(*policy));
921

922
	/* Update governor of new_policy to the governor used before hotplug */
923
	gov = find_governor(policy->last_governor);
924
	if (gov) {
925 926
		pr_debug("Restoring governor %s for cpu %d\n",
				policy->governor->name, policy->cpu);
927 928 929 930 931
	} else {
		gov = cpufreq_default_governor();
		if (!gov)
			return -ENODATA;
	}
932 933 934

	new_policy.governor = gov;

935 936 937 938 939 940 941 942
	/* Use the default policy if there is no last_policy. */
	if (cpufreq_driver->setpolicy) {
		if (policy->last_policy)
			new_policy.policy = policy->last_policy;
		else
			cpufreq_parse_governor(gov->name, &new_policy.policy,
					       NULL);
	}
943
	/* set default policy */
944
	return cpufreq_set_policy(policy, &new_policy);
945 946
}

947
static int cpufreq_add_policy_cpu(struct cpufreq_policy *policy, unsigned int cpu)
948
{
949
	int ret = 0;
950

951 952 953 954
	/* Has this CPU been taken care of already? */
	if (cpumask_test_cpu(cpu, policy->cpus))
		return 0;

955
	down_write(&policy->rwsem);
956
	if (has_target()) {
957
		ret = cpufreq_governor(policy, CPUFREQ_GOV_STOP);
958 959
		if (ret) {
			pr_err("%s: Failed to stop governor\n", __func__);
960
			goto unlock;
961 962
		}
	}
963 964

	cpumask_set_cpu(cpu, policy->cpus);
V
Viresh Kumar 已提交
965

966
	if (has_target()) {
967
		ret = cpufreq_governor(policy, CPUFREQ_GOV_START);
968
		if (!ret)
969
			ret = cpufreq_governor(policy, CPUFREQ_GOV_LIMITS);
970

971
		if (ret)
972
			pr_err("%s: Failed to start governor\n", __func__);
973
	}
974

975 976 977
unlock:
	up_write(&policy->rwsem);
	return ret;
978
}
L
Linus Torvalds 已提交
979

980 981 982 983 984 985 986
static void handle_update(struct work_struct *work)
{
	struct cpufreq_policy *policy =
		container_of(work, struct cpufreq_policy, update);
	unsigned int cpu = policy->cpu;
	pr_debug("handle_update for cpu %u called\n", cpu);
	cpufreq_update_policy(cpu);
987
}
L
Linus Torvalds 已提交
988

989
static struct cpufreq_policy *cpufreq_policy_alloc(unsigned int cpu)
990
{
991
	struct device *dev = get_cpu_device(cpu);
992
	struct cpufreq_policy *policy;
993
	int ret;
994

995 996 997
	if (WARN_ON(!dev))
		return NULL;

998 999 1000 1001 1002 1003 1004 1005 1006 1007
	policy = kzalloc(sizeof(*policy), GFP_KERNEL);
	if (!policy)
		return NULL;

	if (!alloc_cpumask_var(&policy->cpus, GFP_KERNEL))
		goto err_free_policy;

	if (!zalloc_cpumask_var(&policy->related_cpus, GFP_KERNEL))
		goto err_free_cpumask;

1008 1009 1010
	if (!zalloc_cpumask_var(&policy->real_cpus, GFP_KERNEL))
		goto err_free_rcpumask;

1011 1012 1013 1014 1015 1016 1017
	ret = kobject_init_and_add(&policy->kobj, &ktype_cpufreq,
				   cpufreq_global_kobject, "policy%u", cpu);
	if (ret) {
		pr_err("%s: failed to init policy->kobj: %d\n", __func__, ret);
		goto err_free_real_cpus;
	}

1018
	INIT_LIST_HEAD(&policy->policy_list);
1019
	init_rwsem(&policy->rwsem);
1020 1021
	spin_lock_init(&policy->transition_lock);
	init_waitqueue_head(&policy->transition_wait);
1022 1023
	init_completion(&policy->kobj_unregister);
	INIT_WORK(&policy->update, handle_update);
1024

1025
	policy->cpu = cpu;
1026 1027
	return policy;

1028 1029
err_free_real_cpus:
	free_cpumask_var(policy->real_cpus);
1030 1031
err_free_rcpumask:
	free_cpumask_var(policy->related_cpus);
1032 1033 1034 1035 1036 1037 1038 1039
err_free_cpumask:
	free_cpumask_var(policy->cpus);
err_free_policy:
	kfree(policy);

	return NULL;
}

1040
static void cpufreq_policy_put_kobj(struct cpufreq_policy *policy, bool notify)
1041 1042 1043 1044
{
	struct kobject *kobj;
	struct completion *cmp;

1045 1046 1047
	if (notify)
		blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
					     CPUFREQ_REMOVE_POLICY, policy);
1048

1049 1050
	down_write(&policy->rwsem);
	cpufreq_remove_dev_symlink(policy);
1051 1052
	kobj = &policy->kobj;
	cmp = &policy->kobj_unregister;
1053
	up_write(&policy->rwsem);
1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065
	kobject_put(kobj);

	/*
	 * We need to make sure that the underlying kobj is
	 * actually not referenced anymore by anybody before we
	 * proceed with unloading.
	 */
	pr_debug("waiting for dropping of refcount\n");
	wait_for_completion(cmp);
	pr_debug("wait complete\n");
}

1066
static void cpufreq_policy_free(struct cpufreq_policy *policy, bool notify)
1067
{
1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078
	unsigned long flags;
	int cpu;

	/* Remove policy from list */
	write_lock_irqsave(&cpufreq_driver_lock, flags);
	list_del(&policy->policy_list);

	for_each_cpu(cpu, policy->related_cpus)
		per_cpu(cpufreq_cpu_data, cpu) = NULL;
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);

1079
	cpufreq_policy_put_kobj(policy, notify);
1080
	free_cpumask_var(policy->real_cpus);
1081 1082 1083 1084 1085
	free_cpumask_var(policy->related_cpus);
	free_cpumask_var(policy->cpus);
	kfree(policy);
}

1086
static int cpufreq_online(unsigned int cpu)
L
Linus Torvalds 已提交
1087
{
1088
	struct cpufreq_policy *policy;
1089
	bool new_policy;
L
Linus Torvalds 已提交
1090
	unsigned long flags;
1091 1092
	unsigned int j;
	int ret;
1093

1094
	pr_debug("%s: bringing CPU%u online\n", __func__, cpu);
1095

1096
	/* Check if this CPU already has a policy to manage it */
1097
	policy = per_cpu(cpufreq_cpu_data, cpu);
1098
	if (policy) {
1099
		WARN_ON(!cpumask_test_cpu(cpu, policy->related_cpus));
1100
		if (!policy_is_inactive(policy))
1101
			return cpufreq_add_policy_cpu(policy, cpu);
L
Linus Torvalds 已提交
1102

1103
		/* This is the only online CPU for the policy.  Start over. */
1104
		new_policy = false;
1105 1106 1107 1108 1109
		down_write(&policy->rwsem);
		policy->cpu = cpu;
		policy->governor = NULL;
		up_write(&policy->rwsem);
	} else {
1110
		new_policy = true;
1111
		policy = cpufreq_policy_alloc(cpu);
1112
		if (!policy)
1113
			return -ENOMEM;
1114
	}
1115

1116
	cpumask_copy(policy->cpus, cpumask_of(cpu));
L
Linus Torvalds 已提交
1117 1118 1119 1120

	/* call driver. From then on the cpufreq must be able
	 * to accept all calls to ->verify and ->setpolicy for this CPU
	 */
1121
	ret = cpufreq_driver->init(policy);
L
Linus Torvalds 已提交
1122
	if (ret) {
1123
		pr_debug("initialization failed\n");
1124
		goto out_free_policy;
L
Linus Torvalds 已提交
1125
	}
V
Viresh Kumar 已提交
1126

1127 1128
	down_write(&policy->rwsem);

1129
	if (new_policy) {
1130
		/* related_cpus should at least include policy->cpus. */
1131
		cpumask_copy(policy->related_cpus, policy->cpus);
1132
		/* Remember CPUs present at the policy creation time. */
1133
		cpumask_and(policy->real_cpus, policy->cpus, cpu_present_mask);
1134
	}
1135

1136 1137 1138 1139 1140 1141
	/*
	 * affected cpus must always be the one, which are online. We aren't
	 * managing offline cpus here.
	 */
	cpumask_and(policy->cpus, policy->cpus, cpu_online_mask);

1142
	if (new_policy) {
1143 1144
		policy->user_policy.min = policy->min;
		policy->user_policy.max = policy->max;
1145

1146 1147 1148 1149 1150
		write_lock_irqsave(&cpufreq_driver_lock, flags);
		for_each_cpu(j, policy->related_cpus)
			per_cpu(cpufreq_cpu_data, j) = policy;
		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
	}
1151

1152
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
1153 1154 1155
		policy->cur = cpufreq_driver->get(policy->cpu);
		if (!policy->cur) {
			pr_err("%s: ->get() failed\n", __func__);
1156
			goto out_exit_policy;
1157 1158 1159
		}
	}

1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199
	/*
	 * Sometimes boot loaders set CPU frequency to a value outside of
	 * frequency table present with cpufreq core. In such cases CPU might be
	 * unstable if it has to run on that frequency for long duration of time
	 * and so its better to set it to a frequency which is specified in
	 * freq-table. This also makes cpufreq stats inconsistent as
	 * cpufreq-stats would fail to register because current frequency of CPU
	 * isn't found in freq-table.
	 *
	 * Because we don't want this change to effect boot process badly, we go
	 * for the next freq which is >= policy->cur ('cur' must be set by now,
	 * otherwise we will end up setting freq to lowest of the table as 'cur'
	 * is initialized to zero).
	 *
	 * We are passing target-freq as "policy->cur - 1" otherwise
	 * __cpufreq_driver_target() would simply fail, as policy->cur will be
	 * equal to target-freq.
	 */
	if ((cpufreq_driver->flags & CPUFREQ_NEED_INITIAL_FREQ_CHECK)
	    && has_target()) {
		/* Are we running at unknown frequency ? */
		ret = cpufreq_frequency_table_get_index(policy, policy->cur);
		if (ret == -EINVAL) {
			/* Warn user and fix it */
			pr_warn("%s: CPU%d: Running at unlisted freq: %u KHz\n",
				__func__, policy->cpu, policy->cur);
			ret = __cpufreq_driver_target(policy, policy->cur - 1,
				CPUFREQ_RELATION_L);

			/*
			 * Reaching here after boot in a few seconds may not
			 * mean that system will remain stable at "unknown"
			 * frequency for longer duration. Hence, a BUG_ON().
			 */
			BUG_ON(ret);
			pr_warn("%s: CPU%d: Unlisted initial frequency changed to: %u KHz\n",
				__func__, policy->cpu, policy->cur);
		}
	}

1200 1201 1202
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
				     CPUFREQ_START, policy);

1203
	if (new_policy) {
1204
		ret = cpufreq_add_dev_interface(policy);
1205
		if (ret)
1206
			goto out_exit_policy;
1207 1208
		blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
				CPUFREQ_CREATE_POLICY, policy);
1209

1210 1211 1212 1213
		write_lock_irqsave(&cpufreq_driver_lock, flags);
		list_add(&policy->policy_list, &cpufreq_policy_list);
		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
	}
1214

1215 1216 1217 1218
	ret = cpufreq_init_policy(policy);
	if (ret) {
		pr_err("%s: Failed to initialize policy for cpu: %d (%d)\n",
		       __func__, cpu, ret);
1219 1220 1221
		/* cpufreq_policy_free() will notify based on this */
		new_policy = false;
		goto out_exit_policy;
1222
	}
1223

1224
	up_write(&policy->rwsem);
1225

1226
	kobject_uevent(&policy->kobj, KOBJ_ADD);
1227 1228 1229 1230 1231

	/* Callback for handling stuff after policy is ready */
	if (cpufreq_driver->ready)
		cpufreq_driver->ready(policy);

1232
	pr_debug("initialization complete\n");
1233

L
Linus Torvalds 已提交
1234 1235
	return 0;

1236
out_exit_policy:
1237 1238
	up_write(&policy->rwsem);

1239 1240
	if (cpufreq_driver->exit)
		cpufreq_driver->exit(policy);
1241
out_free_policy:
1242
	cpufreq_policy_free(policy, !new_policy);
L
Linus Torvalds 已提交
1243 1244 1245
	return ret;
}

1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270
/**
 * cpufreq_add_dev - the cpufreq interface for a CPU device.
 * @dev: CPU device.
 * @sif: Subsystem interface structure pointer (not used)
 */
static int cpufreq_add_dev(struct device *dev, struct subsys_interface *sif)
{
	unsigned cpu = dev->id;
	int ret;

	dev_dbg(dev, "%s: adding CPU%u\n", __func__, cpu);

	if (cpu_online(cpu)) {
		ret = cpufreq_online(cpu);
	} else {
		/*
		 * A hotplug notifier will follow and we will handle it as CPU
		 * online then.  For now, just create the sysfs link, unless
		 * there is no policy or the link is already present.
		 */
		struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);

		ret = policy && !cpumask_test_and_set_cpu(cpu, policy->real_cpus)
			? add_cpu_dev_symlink(policy, cpu) : 0;
	}
1271

L
Linus Torvalds 已提交
1272 1273 1274
	return ret;
}

1275
static void cpufreq_offline(unsigned int cpu)
L
Linus Torvalds 已提交
1276
{
1277
	struct cpufreq_policy *policy;
1278
	int ret;
L
Linus Torvalds 已提交
1279

1280
	pr_debug("%s: unregistering CPU %u\n", __func__, cpu);
L
Linus Torvalds 已提交
1281

1282
	policy = cpufreq_cpu_get_raw(cpu);
1283
	if (!policy) {
1284
		pr_debug("%s: No cpu_data found\n", __func__);
1285
		return;
L
Linus Torvalds 已提交
1286 1287
	}

1288
	down_write(&policy->rwsem);
1289
	if (has_target()) {
1290
		ret = cpufreq_governor(policy, CPUFREQ_GOV_STOP);
1291
		if (ret)
1292
			pr_err("%s: Failed to stop governor\n", __func__);
1293
	}
L
Linus Torvalds 已提交
1294

1295
	cpumask_clear_cpu(cpu, policy->cpus);
1296

1297 1298 1299 1300
	if (policy_is_inactive(policy)) {
		if (has_target())
			strncpy(policy->last_governor, policy->governor->name,
				CPUFREQ_NAME_LEN);
1301 1302
		else
			policy->last_policy = policy->policy;
1303 1304 1305 1306
	} else if (cpu == policy->cpu) {
		/* Nominate new CPU */
		policy->cpu = cpumask_any(policy->cpus);
	}
1307

1308 1309 1310
	/* Start governor again for active policy */
	if (!policy_is_inactive(policy)) {
		if (has_target()) {
1311
			ret = cpufreq_governor(policy, CPUFREQ_GOV_START);
1312
			if (!ret)
1313
				ret = cpufreq_governor(policy, CPUFREQ_GOV_LIMITS);
1314

1315 1316 1317
			if (ret)
				pr_err("%s: Failed to start governor\n", __func__);
		}
1318

1319
		goto unlock;
1320 1321
	}

1322 1323
	if (cpufreq_driver->stop_cpu)
		cpufreq_driver->stop_cpu(policy);
1324 1325 1326

	/* If cpu is last user of policy, free policy */
	if (has_target()) {
1327
		ret = cpufreq_governor(policy, CPUFREQ_GOV_POLICY_EXIT);
1328
		if (ret)
1329
			pr_err("%s: Failed to exit governor\n", __func__);
1330
	}
L
Linus Torvalds 已提交
1331

1332 1333 1334 1335 1336
	/*
	 * Perform the ->exit() even during light-weight tear-down,
	 * since this is a core component, and is essential for the
	 * subsequent light-weight ->init() to succeed.
	 */
1337
	if (cpufreq_driver->exit) {
1338
		cpufreq_driver->exit(policy);
1339 1340
		policy->freq_table = NULL;
	}
1341 1342 1343

unlock:
	up_write(&policy->rwsem);
L
Linus Torvalds 已提交
1344 1345
}

1346
/**
1347
 * cpufreq_remove_dev - remove a CPU device
1348 1349 1350
 *
 * Removes the cpufreq interface for a CPU device.
 */
1351
static void cpufreq_remove_dev(struct device *dev, struct subsys_interface *sif)
1352
{
1353
	unsigned int cpu = dev->id;
1354
	struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);
1355

1356
	if (!policy)
1357
		return;
1358

1359 1360
	if (cpu_online(cpu))
		cpufreq_offline(cpu);
1361

1362
	cpumask_clear_cpu(cpu, policy->real_cpus);
1363
	remove_cpu_dev_symlink(policy, cpu);
1364

1365
	if (cpumask_empty(policy->real_cpus))
1366
		cpufreq_policy_free(policy, true);
1367 1368
}

L
Linus Torvalds 已提交
1369
/**
1370 1371
 *	cpufreq_out_of_sync - If actual and saved CPU frequency differs, we're
 *	in deep trouble.
1372
 *	@policy: policy managing CPUs
L
Linus Torvalds 已提交
1373 1374
 *	@new_freq: CPU frequency the CPU actually runs at
 *
1375 1376
 *	We adjust to current frequency first, and need to clean up later.
 *	So either call to cpufreq_update_policy() or schedule handle_update()).
L
Linus Torvalds 已提交
1377
 */
1378
static void cpufreq_out_of_sync(struct cpufreq_policy *policy,
1379
				unsigned int new_freq)
L
Linus Torvalds 已提交
1380 1381
{
	struct cpufreq_freqs freqs;
1382

1383
	pr_debug("Warning: CPU frequency out of sync: cpufreq and timing core thinks of %u, is %u kHz\n",
1384
		 policy->cur, new_freq);
L
Linus Torvalds 已提交
1385

1386
	freqs.old = policy->cur;
L
Linus Torvalds 已提交
1387
	freqs.new = new_freq;
1388

1389 1390
	cpufreq_freq_transition_begin(policy, &freqs);
	cpufreq_freq_transition_end(policy, &freqs, 0);
L
Linus Torvalds 已提交
1391 1392
}

1393
/**
D
Dhaval Giani 已提交
1394
 * cpufreq_quick_get - get the CPU frequency (in kHz) from policy->cur
1395 1396 1397 1398 1399 1400 1401
 * @cpu: CPU number
 *
 * This is the last known freq, without actually getting it from the driver.
 * Return value will be same as what is shown in scaling_cur_freq in sysfs.
 */
unsigned int cpufreq_quick_get(unsigned int cpu)
{
1402
	struct cpufreq_policy *policy;
1403
	unsigned int ret_freq = 0;
1404

1405 1406
	if (cpufreq_driver && cpufreq_driver->setpolicy && cpufreq_driver->get)
		return cpufreq_driver->get(cpu);
1407 1408

	policy = cpufreq_cpu_get(cpu);
1409
	if (policy) {
1410
		ret_freq = policy->cur;
1411 1412 1413
		cpufreq_cpu_put(policy);
	}

D
Dave Jones 已提交
1414
	return ret_freq;
1415 1416 1417
}
EXPORT_SYMBOL(cpufreq_quick_get);

1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437
/**
 * cpufreq_quick_get_max - get the max reported CPU frequency for this CPU
 * @cpu: CPU number
 *
 * Just return the max possible frequency for a given CPU.
 */
unsigned int cpufreq_quick_get_max(unsigned int cpu)
{
	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
	unsigned int ret_freq = 0;

	if (policy) {
		ret_freq = policy->max;
		cpufreq_cpu_put(policy);
	}

	return ret_freq;
}
EXPORT_SYMBOL(cpufreq_quick_get_max);

1438
static unsigned int __cpufreq_get(struct cpufreq_policy *policy)
L
Linus Torvalds 已提交
1439
{
1440
	unsigned int ret_freq = 0;
1441

1442
	if (!cpufreq_driver->get)
D
Dave Jones 已提交
1443
		return ret_freq;
L
Linus Torvalds 已提交
1444

1445
	ret_freq = cpufreq_driver->get(policy->cpu);
L
Linus Torvalds 已提交
1446

1447 1448 1449 1450
	/* Updating inactive policies is invalid, so avoid doing that. */
	if (unlikely(policy_is_inactive(policy)))
		return ret_freq;

1451
	if (ret_freq && policy->cur &&
1452
		!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
1453 1454 1455
		/* verify no discrepancy between actual and
					saved value exists */
		if (unlikely(ret_freq != policy->cur)) {
1456
			cpufreq_out_of_sync(policy, ret_freq);
L
Linus Torvalds 已提交
1457 1458 1459 1460
			schedule_work(&policy->update);
		}
	}

D
Dave Jones 已提交
1461
	return ret_freq;
1462
}
L
Linus Torvalds 已提交
1463

1464 1465 1466 1467 1468 1469 1470 1471
/**
 * cpufreq_get - get the current CPU frequency (in kHz)
 * @cpu: CPU number
 *
 * Get the CPU current (static) CPU frequency
 */
unsigned int cpufreq_get(unsigned int cpu)
{
1472
	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
1473 1474
	unsigned int ret_freq = 0;

1475 1476
	if (policy) {
		down_read(&policy->rwsem);
1477
		ret_freq = __cpufreq_get(policy);
1478
		up_read(&policy->rwsem);
1479

1480 1481
		cpufreq_cpu_put(policy);
	}
1482

D
Dave Jones 已提交
1483
	return ret_freq;
L
Linus Torvalds 已提交
1484 1485 1486
}
EXPORT_SYMBOL(cpufreq_get);

1487 1488 1489 1490 1491
static struct subsys_interface cpufreq_interface = {
	.name		= "cpufreq",
	.subsys		= &cpu_subsys,
	.add_dev	= cpufreq_add_dev,
	.remove_dev	= cpufreq_remove_dev,
1492 1493
};

1494 1495 1496 1497 1498 1499 1500 1501 1502
/*
 * In case platform wants some specific frequency to be configured
 * during suspend..
 */
int cpufreq_generic_suspend(struct cpufreq_policy *policy)
{
	int ret;

	if (!policy->suspend_freq) {
1503 1504
		pr_debug("%s: suspend_freq not defined\n", __func__);
		return 0;
1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519
	}

	pr_debug("%s: Setting suspend-freq: %u\n", __func__,
			policy->suspend_freq);

	ret = __cpufreq_driver_target(policy, policy->suspend_freq,
			CPUFREQ_RELATION_H);
	if (ret)
		pr_err("%s: unable to set suspend-freq: %u. err: %d\n",
				__func__, policy->suspend_freq, ret);

	return ret;
}
EXPORT_SYMBOL(cpufreq_generic_suspend);

1520
/**
1521
 * cpufreq_suspend() - Suspend CPUFreq governors
1522
 *
1523 1524 1525 1526
 * Called during system wide Suspend/Hibernate cycles for suspending governors
 * as some platforms can't change frequency after this point in suspend cycle.
 * Because some of the devices (like: i2c, regulators, etc) they use for
 * changing frequency are suspended quickly after this point.
1527
 */
1528
void cpufreq_suspend(void)
1529
{
1530
	struct cpufreq_policy *policy;
1531
	int ret;
1532

1533 1534
	if (!cpufreq_driver)
		return;
1535

1536
	if (!has_target())
1537
		goto suspend;
1538

1539 1540
	pr_debug("%s: Suspending Governors\n", __func__);

1541
	for_each_active_policy(policy) {
1542
		down_write(&policy->rwsem);
1543
		ret = cpufreq_governor(policy, CPUFREQ_GOV_STOP);
1544 1545 1546
		up_write(&policy->rwsem);

		if (ret)
1547 1548 1549 1550 1551 1552
			pr_err("%s: Failed to stop governor for policy: %p\n",
				__func__, policy);
		else if (cpufreq_driver->suspend
		    && cpufreq_driver->suspend(policy))
			pr_err("%s: Failed to suspend driver: %p\n", __func__,
				policy);
1553
	}
1554 1555 1556

suspend:
	cpufreq_suspended = true;
1557 1558
}

L
Linus Torvalds 已提交
1559
/**
1560
 * cpufreq_resume() - Resume CPUFreq governors
L
Linus Torvalds 已提交
1561
 *
1562 1563
 * Called during system wide Suspend/Hibernate cycle for resuming governors that
 * are suspended with cpufreq_suspend().
L
Linus Torvalds 已提交
1564
 */
1565
void cpufreq_resume(void)
L
Linus Torvalds 已提交
1566
{
1567
	struct cpufreq_policy *policy;
1568
	int ret;
L
Linus Torvalds 已提交
1569

1570 1571
	if (!cpufreq_driver)
		return;
L
Linus Torvalds 已提交
1572

1573 1574
	cpufreq_suspended = false;

1575
	if (!has_target())
1576
		return;
L
Linus Torvalds 已提交
1577

1578
	pr_debug("%s: Resuming Governors\n", __func__);
L
Linus Torvalds 已提交
1579

1580
	for_each_active_policy(policy) {
1581
		if (cpufreq_driver->resume && cpufreq_driver->resume(policy)) {
1582 1583
			pr_err("%s: Failed to resume driver: %p\n", __func__,
				policy);
1584 1585
		} else {
			down_write(&policy->rwsem);
1586
			ret = cpufreq_governor(policy, CPUFREQ_GOV_START);
1587
			if (!ret)
1588
				cpufreq_governor(policy, CPUFREQ_GOV_LIMITS);
1589 1590 1591 1592 1593 1594
			up_write(&policy->rwsem);

			if (ret)
				pr_err("%s: Failed to start governor for policy: %p\n",
				       __func__, policy);
		}
1595
	}
1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606

	/*
	 * schedule call cpufreq_update_policy() for first-online CPU, as that
	 * wouldn't be hotplugged-out on suspend. It will verify that the
	 * current freq is in sync with what we believe it to be.
	 */
	policy = cpufreq_cpu_get_raw(cpumask_first(cpu_online_mask));
	if (WARN_ON(!policy))
		return;

	schedule_work(&policy->update);
1607
}
L
Linus Torvalds 已提交
1608

1609 1610 1611 1612 1613 1614 1615 1616
/**
 *	cpufreq_get_current_driver - return current driver's name
 *
 *	Return the name string of the currently loaded cpufreq driver
 *	or NULL, if none.
 */
const char *cpufreq_get_current_driver(void)
{
1617 1618 1619 1620
	if (cpufreq_driver)
		return cpufreq_driver->name;

	return NULL;
1621 1622
}
EXPORT_SYMBOL_GPL(cpufreq_get_current_driver);
L
Linus Torvalds 已提交
1623

1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638
/**
 *	cpufreq_get_driver_data - return current driver data
 *
 *	Return the private data of the currently loaded cpufreq
 *	driver, or NULL if no cpufreq driver is loaded.
 */
void *cpufreq_get_driver_data(void)
{
	if (cpufreq_driver)
		return cpufreq_driver->driver_data;

	return NULL;
}
EXPORT_SYMBOL_GPL(cpufreq_get_driver_data);

L
Linus Torvalds 已提交
1639 1640 1641 1642 1643 1644 1645 1646 1647
/*********************************************************************
 *                     NOTIFIER LISTS INTERFACE                      *
 *********************************************************************/

/**
 *	cpufreq_register_notifier - register a driver with cpufreq
 *	@nb: notifier function to register
 *      @list: CPUFREQ_TRANSITION_NOTIFIER or CPUFREQ_POLICY_NOTIFIER
 *
1648
 *	Add a driver to one of two lists: either a list of drivers that
L
Linus Torvalds 已提交
1649 1650 1651 1652 1653
 *      are notified about clock rate changes (once before and once after
 *      the transition), or a list of drivers that are notified about
 *      changes in cpufreq policy.
 *
 *	This function may sleep, and has the same return conditions as
1654
 *	blocking_notifier_chain_register.
L
Linus Torvalds 已提交
1655 1656 1657 1658 1659
 */
int cpufreq_register_notifier(struct notifier_block *nb, unsigned int list)
{
	int ret;

1660 1661 1662
	if (cpufreq_disabled())
		return -EINVAL;

1663 1664
	WARN_ON(!init_cpufreq_transition_notifier_list_called);

L
Linus Torvalds 已提交
1665 1666
	switch (list) {
	case CPUFREQ_TRANSITION_NOTIFIER:
1667
		ret = srcu_notifier_chain_register(
1668
				&cpufreq_transition_notifier_list, nb);
L
Linus Torvalds 已提交
1669 1670
		break;
	case CPUFREQ_POLICY_NOTIFIER:
1671 1672
		ret = blocking_notifier_chain_register(
				&cpufreq_policy_notifier_list, nb);
L
Linus Torvalds 已提交
1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684
		break;
	default:
		ret = -EINVAL;
	}

	return ret;
}
EXPORT_SYMBOL(cpufreq_register_notifier);

/**
 *	cpufreq_unregister_notifier - unregister a driver with cpufreq
 *	@nb: notifier block to be unregistered
1685
 *	@list: CPUFREQ_TRANSITION_NOTIFIER or CPUFREQ_POLICY_NOTIFIER
L
Linus Torvalds 已提交
1686 1687 1688 1689
 *
 *	Remove a driver from the CPU frequency notifier list.
 *
 *	This function may sleep, and has the same return conditions as
1690
 *	blocking_notifier_chain_unregister.
L
Linus Torvalds 已提交
1691 1692 1693 1694 1695
 */
int cpufreq_unregister_notifier(struct notifier_block *nb, unsigned int list)
{
	int ret;

1696 1697 1698
	if (cpufreq_disabled())
		return -EINVAL;

L
Linus Torvalds 已提交
1699 1700
	switch (list) {
	case CPUFREQ_TRANSITION_NOTIFIER:
1701
		ret = srcu_notifier_chain_unregister(
1702
				&cpufreq_transition_notifier_list, nb);
L
Linus Torvalds 已提交
1703 1704
		break;
	case CPUFREQ_POLICY_NOTIFIER:
1705 1706
		ret = blocking_notifier_chain_unregister(
				&cpufreq_policy_notifier_list, nb);
L
Linus Torvalds 已提交
1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720
		break;
	default:
		ret = -EINVAL;
	}

	return ret;
}
EXPORT_SYMBOL(cpufreq_unregister_notifier);


/*********************************************************************
 *                              GOVERNORS                            *
 *********************************************************************/

1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746
/* Must set freqs->new to intermediate frequency */
static int __target_intermediate(struct cpufreq_policy *policy,
				 struct cpufreq_freqs *freqs, int index)
{
	int ret;

	freqs->new = cpufreq_driver->get_intermediate(policy, index);

	/* We don't need to switch to intermediate freq */
	if (!freqs->new)
		return 0;

	pr_debug("%s: cpu: %d, switching to intermediate freq: oldfreq: %u, intermediate freq: %u\n",
		 __func__, policy->cpu, freqs->old, freqs->new);

	cpufreq_freq_transition_begin(policy, freqs);
	ret = cpufreq_driver->target_intermediate(policy, index);
	cpufreq_freq_transition_end(policy, freqs, ret);

	if (ret)
		pr_err("%s: Failed to change to intermediate frequency: %d\n",
		       __func__, ret);

	return ret;
}

1747 1748 1749
static int __target_index(struct cpufreq_policy *policy,
			  struct cpufreq_frequency_table *freq_table, int index)
{
1750 1751
	struct cpufreq_freqs freqs = {.old = policy->cur, .flags = 0};
	unsigned int intermediate_freq = 0;
1752 1753 1754 1755 1756
	int retval = -EINVAL;
	bool notify;

	notify = !(cpufreq_driver->flags & CPUFREQ_ASYNC_NOTIFICATION);
	if (notify) {
1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767
		/* Handle switching to intermediate frequency */
		if (cpufreq_driver->get_intermediate) {
			retval = __target_intermediate(policy, &freqs, index);
			if (retval)
				return retval;

			intermediate_freq = freqs.new;
			/* Set old freq to intermediate */
			if (intermediate_freq)
				freqs.old = freqs.new;
		}
1768

1769
		freqs.new = freq_table[index].frequency;
1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780
		pr_debug("%s: cpu: %d, oldfreq: %u, new freq: %u\n",
			 __func__, policy->cpu, freqs.old, freqs.new);

		cpufreq_freq_transition_begin(policy, &freqs);
	}

	retval = cpufreq_driver->target_index(policy, index);
	if (retval)
		pr_err("%s: Failed to change cpu frequency: %d\n", __func__,
		       retval);

1781
	if (notify) {
1782 1783
		cpufreq_freq_transition_end(policy, &freqs, retval);

1784 1785 1786 1787
		/*
		 * Failed after setting to intermediate freq? Driver should have
		 * reverted back to initial frequency and so should we. Check
		 * here for intermediate_freq instead of get_intermediate, in
1788
		 * case we haven't switched to intermediate freq at all.
1789 1790 1791 1792 1793 1794 1795 1796 1797
		 */
		if (unlikely(retval && intermediate_freq)) {
			freqs.old = intermediate_freq;
			freqs.new = policy->restore_freq;
			cpufreq_freq_transition_begin(policy, &freqs);
			cpufreq_freq_transition_end(policy, &freqs, 0);
		}
	}

1798 1799 1800
	return retval;
}

L
Linus Torvalds 已提交
1801 1802 1803 1804
int __cpufreq_driver_target(struct cpufreq_policy *policy,
			    unsigned int target_freq,
			    unsigned int relation)
{
1805
	unsigned int old_target_freq = target_freq;
1806 1807
	struct cpufreq_frequency_table *freq_table;
	int index, retval;
1808

1809 1810 1811
	if (cpufreq_disabled())
		return -ENODEV;

1812 1813 1814 1815 1816 1817 1818
	/* Make sure that target_freq is within supported range */
	if (target_freq > policy->max)
		target_freq = policy->max;
	if (target_freq < policy->min)
		target_freq = policy->min;

	pr_debug("target for CPU %u: %u kHz, relation %u, requested %u kHz\n",
1819
		 policy->cpu, target_freq, relation, old_target_freq);
1820

1821 1822 1823 1824 1825 1826
	/*
	 * This might look like a redundant call as we are checking it again
	 * after finding index. But it is left intentionally for cases where
	 * exactly same freq is called again and so we can save on few function
	 * calls.
	 */
1827 1828 1829
	if (target_freq == policy->cur)
		return 0;

1830 1831 1832
	/* Save last value to restore later on errors */
	policy->restore_freq = policy->cur;

1833
	if (cpufreq_driver->target)
1834
		return cpufreq_driver->target(policy, target_freq, relation);
1835

1836 1837
	if (!cpufreq_driver->target_index)
		return -EINVAL;
1838

1839 1840 1841 1842 1843
	freq_table = cpufreq_frequency_get_table(policy->cpu);
	if (unlikely(!freq_table)) {
		pr_err("%s: Unable to find freq_table\n", __func__);
		return -EINVAL;
	}
1844

1845 1846 1847 1848 1849
	retval = cpufreq_frequency_table_target(policy, freq_table, target_freq,
						relation, &index);
	if (unlikely(retval)) {
		pr_err("%s: Unable to find matching freq\n", __func__);
		return retval;
1850 1851
	}

1852 1853 1854 1855
	if (freq_table[index].frequency == policy->cur)
		return 0;

	return __target_index(policy, freq_table, index);
L
Linus Torvalds 已提交
1856 1857 1858 1859 1860 1861 1862
}
EXPORT_SYMBOL_GPL(__cpufreq_driver_target);

int cpufreq_driver_target(struct cpufreq_policy *policy,
			  unsigned int target_freq,
			  unsigned int relation)
{
1863
	int ret = -EINVAL;
L
Linus Torvalds 已提交
1864

1865
	down_write(&policy->rwsem);
L
Linus Torvalds 已提交
1866 1867 1868

	ret = __cpufreq_driver_target(policy, target_freq, relation);

1869
	up_write(&policy->rwsem);
L
Linus Torvalds 已提交
1870 1871 1872 1873 1874

	return ret;
}
EXPORT_SYMBOL_GPL(cpufreq_driver_target);

1875 1876 1877 1878 1879
__weak struct cpufreq_governor *cpufreq_fallback_governor(void)
{
	return NULL;
}

1880
static int cpufreq_governor(struct cpufreq_policy *policy, unsigned int event)
L
Linus Torvalds 已提交
1881
{
1882
	int ret;
1883

1884 1885 1886
	/* Don't start any governor operations if we are entering suspend */
	if (cpufreq_suspended)
		return 0;
1887 1888 1889 1890 1891 1892
	/*
	 * Governor might not be initiated here if ACPI _PPC changed
	 * notification happened, so check it.
	 */
	if (!policy->governor)
		return -EINVAL;
1893

1894 1895 1896
	if (policy->governor->max_transition_latency &&
	    policy->cpuinfo.transition_latency >
	    policy->governor->max_transition_latency) {
1897 1898 1899
		struct cpufreq_governor *gov = cpufreq_fallback_governor();

		if (gov) {
1900 1901
			pr_warn("%s governor failed, too long transition latency of HW, fallback to %s governor\n",
				policy->governor->name, gov->name);
1902
			policy->governor = gov;
1903 1904
		} else {
			return -EINVAL;
1905
		}
1906
	}
L
Linus Torvalds 已提交
1907

1908 1909 1910
	if (event == CPUFREQ_GOV_POLICY_INIT)
		if (!try_module_get(policy->governor->owner))
			return -EINVAL;
L
Linus Torvalds 已提交
1911

1912
	pr_debug("%s: for CPU %u, event %u\n", __func__, policy->cpu, event);
1913

L
Linus Torvalds 已提交
1914 1915
	ret = policy->governor->governor(policy, event);

1916 1917 1918 1919 1920 1921
	if (!ret) {
		if (event == CPUFREQ_GOV_POLICY_INIT)
			policy->governor->initialized++;
		else if (event == CPUFREQ_GOV_POLICY_EXIT)
			policy->governor->initialized--;
	}
1922

1923 1924
	if (((event == CPUFREQ_GOV_POLICY_INIT) && ret) ||
			((event == CPUFREQ_GOV_POLICY_EXIT) && !ret))
L
Linus Torvalds 已提交
1925 1926 1927 1928 1929 1930 1931
		module_put(policy->governor->owner);

	return ret;
}

int cpufreq_register_governor(struct cpufreq_governor *governor)
{
1932
	int err;
L
Linus Torvalds 已提交
1933 1934 1935 1936

	if (!governor)
		return -EINVAL;

1937 1938 1939
	if (cpufreq_disabled())
		return -ENODEV;

1940
	mutex_lock(&cpufreq_governor_mutex);
1941

1942
	governor->initialized = 0;
1943
	err = -EBUSY;
1944
	if (!find_governor(governor->name)) {
1945 1946
		err = 0;
		list_add(&governor->governor_list, &cpufreq_governor_list);
L
Linus Torvalds 已提交
1947 1948
	}

1949
	mutex_unlock(&cpufreq_governor_mutex);
1950
	return err;
L
Linus Torvalds 已提交
1951 1952 1953 1954 1955
}
EXPORT_SYMBOL_GPL(cpufreq_register_governor);

void cpufreq_unregister_governor(struct cpufreq_governor *governor)
{
1956 1957
	struct cpufreq_policy *policy;
	unsigned long flags;
1958

L
Linus Torvalds 已提交
1959 1960 1961
	if (!governor)
		return;

1962 1963 1964
	if (cpufreq_disabled())
		return;

1965 1966 1967
	/* clear last_governor for all inactive policies */
	read_lock_irqsave(&cpufreq_driver_lock, flags);
	for_each_inactive_policy(policy) {
1968 1969
		if (!strcmp(policy->last_governor, governor->name)) {
			policy->governor = NULL;
1970
			strcpy(policy->last_governor, "\0");
1971
		}
1972
	}
1973
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
1974

1975
	mutex_lock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
1976
	list_del(&governor->governor_list);
1977
	mutex_unlock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988
	return;
}
EXPORT_SYMBOL_GPL(cpufreq_unregister_governor);


/*********************************************************************
 *                          POLICY INTERFACE                         *
 *********************************************************************/

/**
 * cpufreq_get_policy - get the current cpufreq_policy
1989 1990
 * @policy: struct cpufreq_policy into which the current cpufreq_policy
 *	is written
L
Linus Torvalds 已提交
1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
 *
 * Reads the current cpufreq policy.
 */
int cpufreq_get_policy(struct cpufreq_policy *policy, unsigned int cpu)
{
	struct cpufreq_policy *cpu_policy;
	if (!policy)
		return -EINVAL;

	cpu_policy = cpufreq_cpu_get(cpu);
	if (!cpu_policy)
		return -EINVAL;

2004
	memcpy(policy, cpu_policy, sizeof(*policy));
L
Linus Torvalds 已提交
2005 2006 2007 2008 2009 2010

	cpufreq_cpu_put(cpu_policy);
	return 0;
}
EXPORT_SYMBOL(cpufreq_get_policy);

2011
/*
2012 2013
 * policy : current policy.
 * new_policy: policy to be set.
2014
 */
2015
static int cpufreq_set_policy(struct cpufreq_policy *policy,
2016
				struct cpufreq_policy *new_policy)
L
Linus Torvalds 已提交
2017
{
2018 2019
	struct cpufreq_governor *old_gov;
	int ret;
L
Linus Torvalds 已提交
2020

2021 2022
	pr_debug("setting new policy for CPU %u: %u - %u kHz\n",
		 new_policy->cpu, new_policy->min, new_policy->max);
L
Linus Torvalds 已提交
2023

2024
	memcpy(&new_policy->cpuinfo, &policy->cpuinfo, sizeof(policy->cpuinfo));
L
Linus Torvalds 已提交
2025

2026 2027 2028 2029 2030
	/*
	* This check works well when we store new min/max freq attributes,
	* because new_policy is a copy of policy with one field updated.
	*/
	if (new_policy->min > new_policy->max)
2031
		return -EINVAL;
2032

L
Linus Torvalds 已提交
2033
	/* verify the cpu speed can be set within this limit */
2034
	ret = cpufreq_driver->verify(new_policy);
L
Linus Torvalds 已提交
2035
	if (ret)
2036
		return ret;
L
Linus Torvalds 已提交
2037 2038

	/* adjust if necessary - all reasons */
2039
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2040
			CPUFREQ_ADJUST, new_policy);
L
Linus Torvalds 已提交
2041

2042 2043 2044 2045
	/*
	 * verify the cpu speed can be set within this limit, which might be
	 * different to the first one
	 */
2046
	ret = cpufreq_driver->verify(new_policy);
2047
	if (ret)
2048
		return ret;
L
Linus Torvalds 已提交
2049 2050

	/* notification of the new policy */
2051
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2052
			CPUFREQ_NOTIFY, new_policy);
L
Linus Torvalds 已提交
2053

2054 2055
	policy->min = new_policy->min;
	policy->max = new_policy->max;
L
Linus Torvalds 已提交
2056

2057
	pr_debug("new min and max freqs are %u - %u kHz\n",
2058
		 policy->min, policy->max);
L
Linus Torvalds 已提交
2059

2060
	if (cpufreq_driver->setpolicy) {
2061
		policy->policy = new_policy->policy;
2062
		pr_debug("setting range\n");
2063 2064
		return cpufreq_driver->setpolicy(new_policy);
	}
L
Linus Torvalds 已提交
2065

2066 2067
	if (new_policy->governor == policy->governor)
		goto out;
2068

2069 2070 2071 2072 2073 2074
	pr_debug("governor switch\n");

	/* save old, working values */
	old_gov = policy->governor;
	/* end old governor */
	if (old_gov) {
2075
		ret = cpufreq_governor(policy, CPUFREQ_GOV_STOP);
2076 2077 2078 2079 2080 2081 2082
		if (ret) {
			/* This can happen due to race with other operations */
			pr_debug("%s: Failed to Stop Governor: %s (%d)\n",
				 __func__, old_gov->name, ret);
			return ret;
		}

2083
		ret = cpufreq_governor(policy, CPUFREQ_GOV_POLICY_EXIT);
2084 2085 2086 2087 2088
		if (ret) {
			pr_err("%s: Failed to Exit Governor: %s (%d)\n",
			       __func__, old_gov->name, ret);
			return ret;
		}
L
Linus Torvalds 已提交
2089 2090
	}

2091 2092
	/* start new governor */
	policy->governor = new_policy->governor;
2093
	ret = cpufreq_governor(policy, CPUFREQ_GOV_POLICY_INIT);
2094
	if (!ret) {
2095
		ret = cpufreq_governor(policy, CPUFREQ_GOV_START);
2096
		if (!ret)
2097 2098
			goto out;

2099
		cpufreq_governor(policy, CPUFREQ_GOV_POLICY_EXIT);
2100 2101 2102 2103 2104 2105
	}

	/* new governor failed, so re-start old one */
	pr_debug("starting governor %s failed\n", policy->governor->name);
	if (old_gov) {
		policy->governor = old_gov;
2106
		if (cpufreq_governor(policy, CPUFREQ_GOV_POLICY_INIT))
2107 2108
			policy->governor = NULL;
		else
2109
			cpufreq_governor(policy, CPUFREQ_GOV_START);
2110 2111
	}

2112
	return ret;
2113 2114 2115

 out:
	pr_debug("governor: change or update limits\n");
2116
	return cpufreq_governor(policy, CPUFREQ_GOV_LIMITS);
L
Linus Torvalds 已提交
2117 2118 2119 2120 2121 2122
}

/**
 *	cpufreq_update_policy - re-evaluate an existing cpufreq policy
 *	@cpu: CPU which shall be re-evaluated
 *
L
Lucas De Marchi 已提交
2123
 *	Useful for policy notifiers which have different necessities
L
Linus Torvalds 已提交
2124 2125 2126 2127
 *	at different times.
 */
int cpufreq_update_policy(unsigned int cpu)
{
2128 2129
	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
	struct cpufreq_policy new_policy;
2130
	int ret;
L
Linus Torvalds 已提交
2131

2132 2133
	if (!policy)
		return -ENODEV;
L
Linus Torvalds 已提交
2134

2135
	down_write(&policy->rwsem);
L
Linus Torvalds 已提交
2136

2137
	pr_debug("updating policy for CPU %u\n", cpu);
2138
	memcpy(&new_policy, policy, sizeof(*policy));
2139 2140
	new_policy.min = policy->user_policy.min;
	new_policy.max = policy->user_policy.max;
L
Linus Torvalds 已提交
2141

2142 2143 2144 2145
	/*
	 * BIOS might change freq behind our back
	 * -> ask driver for current freq and notify governors about a change
	 */
2146
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
2147
		new_policy.cur = cpufreq_driver->get(cpu);
2148 2149
		if (WARN_ON(!new_policy.cur)) {
			ret = -EIO;
2150
			goto unlock;
2151 2152
		}

2153
		if (!policy->cur) {
2154
			pr_debug("Driver did not initialize current freq\n");
2155
			policy->cur = new_policy.cur;
2156
		} else {
2157
			if (policy->cur != new_policy.cur && has_target())
2158
				cpufreq_out_of_sync(policy, new_policy.cur);
2159
		}
2160 2161
	}

2162
	ret = cpufreq_set_policy(policy, &new_policy);
L
Linus Torvalds 已提交
2163

2164
unlock:
2165
	up_write(&policy->rwsem);
2166

2167
	cpufreq_cpu_put(policy);
L
Linus Torvalds 已提交
2168 2169 2170 2171
	return ret;
}
EXPORT_SYMBOL(cpufreq_update_policy);

2172
static int cpufreq_cpu_callback(struct notifier_block *nfb,
2173 2174 2175 2176
					unsigned long action, void *hcpu)
{
	unsigned int cpu = (unsigned long)hcpu;

2177 2178 2179 2180
	switch (action & ~CPU_TASKS_FROZEN) {
	case CPU_ONLINE:
		cpufreq_online(cpu);
		break;
2181

2182
	case CPU_DOWN_PREPARE:
2183
		cpufreq_offline(cpu);
2184
		break;
2185

2186 2187 2188
	case CPU_DOWN_FAILED:
		cpufreq_online(cpu);
		break;
2189 2190 2191 2192
	}
	return NOTIFY_OK;
}

2193
static struct notifier_block __refdata cpufreq_cpu_notifier = {
2194
	.notifier_call = cpufreq_cpu_callback,
2195
};
L
Linus Torvalds 已提交
2196

2197 2198 2199 2200 2201 2202 2203 2204 2205
/*********************************************************************
 *               BOOST						     *
 *********************************************************************/
static int cpufreq_boost_set_sw(int state)
{
	struct cpufreq_frequency_table *freq_table;
	struct cpufreq_policy *policy;
	int ret = -EINVAL;

2206
	for_each_active_policy(policy) {
2207 2208 2209 2210 2211 2212 2213 2214 2215
		freq_table = cpufreq_frequency_get_table(policy->cpu);
		if (freq_table) {
			ret = cpufreq_frequency_table_cpuinfo(policy,
							freq_table);
			if (ret) {
				pr_err("%s: Policy frequency update failed\n",
				       __func__);
				break;
			}
2216 2217

			down_write(&policy->rwsem);
2218
			policy->user_policy.max = policy->max;
2219
			cpufreq_governor(policy, CPUFREQ_GOV_LIMITS);
2220
			up_write(&policy->rwsem);
2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244
		}
	}

	return ret;
}

int cpufreq_boost_trigger_state(int state)
{
	unsigned long flags;
	int ret = 0;

	if (cpufreq_driver->boost_enabled == state)
		return 0;

	write_lock_irqsave(&cpufreq_driver_lock, flags);
	cpufreq_driver->boost_enabled = state;
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);

	ret = cpufreq_driver->set_boost(state);
	if (ret) {
		write_lock_irqsave(&cpufreq_driver_lock, flags);
		cpufreq_driver->boost_enabled = !state;
		write_unlock_irqrestore(&cpufreq_driver_lock, flags);

2245 2246
		pr_err("%s: Cannot %s BOOST\n",
		       __func__, state ? "enable" : "disable");
2247 2248 2249 2250 2251
	}

	return ret;
}

2252
static bool cpufreq_boost_supported(void)
2253
{
2254
	return likely(cpufreq_driver) && cpufreq_driver->set_boost;
2255 2256
}

2257 2258 2259 2260
static int create_boost_sysfs_file(void)
{
	int ret;

2261
	ret = sysfs_create_file(cpufreq_global_kobject, &boost.attr);
2262 2263 2264 2265 2266 2267 2268 2269 2270 2271
	if (ret)
		pr_err("%s: cannot register global BOOST sysfs file\n",
		       __func__);

	return ret;
}

static void remove_boost_sysfs_file(void)
{
	if (cpufreq_boost_supported())
2272
		sysfs_remove_file(cpufreq_global_kobject, &boost.attr);
2273 2274 2275 2276 2277 2278 2279 2280 2281 2282
}

int cpufreq_enable_boost_support(void)
{
	if (!cpufreq_driver)
		return -EINVAL;

	if (cpufreq_boost_supported())
		return 0;

2283
	cpufreq_driver->set_boost = cpufreq_boost_set_sw;
2284 2285 2286 2287 2288 2289

	/* This will get removed on driver unregister */
	return create_boost_sysfs_file();
}
EXPORT_SYMBOL_GPL(cpufreq_enable_boost_support);

2290 2291 2292 2293 2294 2295
int cpufreq_boost_enabled(void)
{
	return cpufreq_driver->boost_enabled;
}
EXPORT_SYMBOL_GPL(cpufreq_boost_enabled);

L
Linus Torvalds 已提交
2296 2297 2298 2299 2300 2301 2302 2303 2304
/*********************************************************************
 *               REGISTER / UNREGISTER CPUFREQ DRIVER                *
 *********************************************************************/

/**
 * cpufreq_register_driver - register a CPU Frequency driver
 * @driver_data: A struct cpufreq_driver containing the values#
 * submitted by the CPU Frequency driver.
 *
2305
 * Registers a CPU Frequency driver to this core code. This code
2306
 * returns zero on success, -EEXIST when another driver got here first
2307
 * (and isn't unregistered in the meantime).
L
Linus Torvalds 已提交
2308 2309
 *
 */
2310
int cpufreq_register_driver(struct cpufreq_driver *driver_data)
L
Linus Torvalds 已提交
2311 2312 2313 2314
{
	unsigned long flags;
	int ret;

2315 2316 2317
	if (cpufreq_disabled())
		return -ENODEV;

L
Linus Torvalds 已提交
2318
	if (!driver_data || !driver_data->verify || !driver_data->init ||
2319
	    !(driver_data->setpolicy || driver_data->target_index ||
2320 2321
		    driver_data->target) ||
	     (driver_data->setpolicy && (driver_data->target_index ||
2322 2323
		    driver_data->target)) ||
	     (!!driver_data->get_intermediate != !!driver_data->target_intermediate))
L
Linus Torvalds 已提交
2324 2325
		return -EINVAL;

2326
	pr_debug("trying to register driver %s\n", driver_data->name);
L
Linus Torvalds 已提交
2327

2328 2329 2330
	/* Protect against concurrent CPU online/offline. */
	get_online_cpus();

2331
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2332
	if (cpufreq_driver) {
2333
		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2334 2335
		ret = -EEXIST;
		goto out;
L
Linus Torvalds 已提交
2336
	}
2337
	cpufreq_driver = driver_data;
2338
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
2339

2340 2341 2342
	if (driver_data->setpolicy)
		driver_data->flags |= CPUFREQ_CONST_LOOPS;

2343 2344 2345 2346 2347
	if (cpufreq_boost_supported()) {
		ret = create_boost_sysfs_file();
		if (ret)
			goto err_null_driver;
	}
2348

2349
	ret = subsys_interface_register(&cpufreq_interface);
2350
	if (ret)
2351
		goto err_boost_unreg;
L
Linus Torvalds 已提交
2352

2353 2354
	if (!(cpufreq_driver->flags & CPUFREQ_STICKY) &&
	    list_empty(&cpufreq_policy_list)) {
L
Linus Torvalds 已提交
2355
		/* if all ->init() calls failed, unregister */
2356 2357 2358
		pr_debug("%s: No CPU initialized for driver %s\n", __func__,
			 driver_data->name);
		goto err_if_unreg;
L
Linus Torvalds 已提交
2359 2360
	}

2361
	register_hotcpu_notifier(&cpufreq_cpu_notifier);
2362
	pr_debug("driver %s up and running\n", driver_data->name);
L
Linus Torvalds 已提交
2363

2364 2365 2366 2367
out:
	put_online_cpus();
	return ret;

2368 2369
err_if_unreg:
	subsys_interface_unregister(&cpufreq_interface);
2370
err_boost_unreg:
2371
	remove_boost_sysfs_file();
2372
err_null_driver:
2373
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2374
	cpufreq_driver = NULL;
2375
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2376
	goto out;
L
Linus Torvalds 已提交
2377 2378 2379 2380 2381 2382
}
EXPORT_SYMBOL_GPL(cpufreq_register_driver);

/**
 * cpufreq_unregister_driver - unregister the current CPUFreq driver
 *
2383
 * Unregister the current CPUFreq driver. Only call this if you have
L
Linus Torvalds 已提交
2384 2385 2386 2387
 * the right to do so, i.e. if you have succeeded in initialising before!
 * Returns zero if successful, and -EINVAL if the cpufreq_driver is
 * currently not initialised.
 */
2388
int cpufreq_unregister_driver(struct cpufreq_driver *driver)
L
Linus Torvalds 已提交
2389 2390 2391
{
	unsigned long flags;

2392
	if (!cpufreq_driver || (driver != cpufreq_driver))
L
Linus Torvalds 已提交
2393 2394
		return -EINVAL;

2395
	pr_debug("unregistering driver %s\n", driver->name);
L
Linus Torvalds 已提交
2396

2397 2398
	/* Protect against concurrent cpu hotplug */
	get_online_cpus();
2399
	subsys_interface_unregister(&cpufreq_interface);
2400
	remove_boost_sysfs_file();
2401
	unregister_hotcpu_notifier(&cpufreq_cpu_notifier);
L
Linus Torvalds 已提交
2402

2403
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2404

2405
	cpufreq_driver = NULL;
2406

2407
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2408
	put_online_cpus();
L
Linus Torvalds 已提交
2409 2410 2411 2412

	return 0;
}
EXPORT_SYMBOL_GPL(cpufreq_unregister_driver);
2413

2414 2415 2416 2417 2418 2419 2420 2421
/*
 * Stop cpufreq at shutdown to make sure it isn't holding any locks
 * or mutexes when secondary CPUs are halted.
 */
static struct syscore_ops cpufreq_syscore_ops = {
	.shutdown = cpufreq_suspend,
};

2422 2423 2424
struct kobject *cpufreq_global_kobject;
EXPORT_SYMBOL(cpufreq_global_kobject);

2425 2426
static int __init cpufreq_core_init(void)
{
2427 2428 2429
	if (cpufreq_disabled())
		return -ENODEV;

2430
	cpufreq_global_kobject = kobject_create_and_add("cpufreq", &cpu_subsys.dev_root->kobj);
2431 2432
	BUG_ON(!cpufreq_global_kobject);

2433 2434
	register_syscore_ops(&cpufreq_syscore_ops);

2435 2436 2437
	return 0;
}
core_initcall(cpufreq_core_init);