cpufreq.c 65.8 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 unsigned int __cpufreq_get(struct cpufreq_policy *policy);
78 79
static int cpufreq_init_governor(struct cpufreq_policy *policy);
static void cpufreq_exit_governor(struct cpufreq_policy *policy);
80
static int cpufreq_start_governor(struct cpufreq_policy *policy);
81 82
static void cpufreq_stop_governor(struct cpufreq_policy *policy);
static void cpufreq_governor_limits(struct cpufreq_policy *policy);
83

L
Linus Torvalds 已提交
84
/**
85 86
 * 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
87 88 89 90
 * "transition" list for kernel code that needs to handle
 * changes to devices when the CPU clock speed changes.
 * The mutex locks both lists.
 */
91
static BLOCKING_NOTIFIER_HEAD(cpufreq_policy_notifier_list);
92
static struct srcu_notifier_head cpufreq_transition_notifier_list;
L
Linus Torvalds 已提交
93

94
static bool init_cpufreq_transition_notifier_list_called;
95 96 97
static int __init init_cpufreq_transition_notifier_list(void)
{
	srcu_init_notifier_head(&cpufreq_transition_notifier_list);
98
	init_cpufreq_transition_notifier_list_called = true;
99 100
	return 0;
}
101
pure_initcall(init_cpufreq_transition_notifier_list);
L
Linus Torvalds 已提交
102

103
static int off __read_mostly;
104
static int cpufreq_disabled(void)
105 106 107 108 109 110 111
{
	return off;
}
void disable_cpufreq(void)
{
	off = 1;
}
112
static DEFINE_MUTEX(cpufreq_governor_mutex);
L
Linus Torvalds 已提交
113

114 115
bool have_governor_per_policy(void)
{
116
	return !!(cpufreq_driver->flags & CPUFREQ_HAVE_GOVERNOR_PER_POLICY);
117
}
118
EXPORT_SYMBOL_GPL(have_governor_per_policy);
119

120 121 122 123 124 125 126 127 128
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);

129 130 131 132 133 134
static inline u64 get_cpu_idle_time_jiffy(unsigned int cpu, u64 *wall)
{
	u64 idle_time;
	u64 cur_wall_time;
	u64 busy_time;

135
	cur_wall_time = jiffies64_to_nsecs(get_jiffies_64());
136 137 138 139 140 141 142 143 144 145

	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)
146
		*wall = div_u64(cur_wall_time, NSEC_PER_USEC);
147

148
	return div_u64(idle_time, NSEC_PER_USEC);
149 150 151 152 153 154 155 156 157 158 159 160 161 162 163
}

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);

164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185
/*
 * 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;

	/*
186
	 * The driver only supports the SMP configuration where all processors
187 188 189 190 191 192 193 194
	 * share the clock and voltage and clock.
	 */
	cpumask_setall(policy->cpus);

	return 0;
}
EXPORT_SYMBOL_GPL(cpufreq_generic_init);

195
struct cpufreq_policy *cpufreq_cpu_get_raw(unsigned int cpu)
196 197 198
{
	struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);

199 200
	return policy && cpumask_test_cpu(cpu, policy->cpus) ? policy : NULL;
}
201
EXPORT_SYMBOL_GPL(cpufreq_cpu_get_raw);
202 203 204 205 206

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

207
	if (!policy || IS_ERR(policy->clk)) {
208 209
		pr_err("%s: No %s associated to cpu: %d\n",
		       __func__, policy ? "clk" : "policy", cpu);
210 211 212 213 214 215 216
		return 0;
	}

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

217 218 219 220 221 222 223 224 225 226 227 228 229
/**
 * 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.
 */
230
struct cpufreq_policy *cpufreq_cpu_get(unsigned int cpu)
L
Linus Torvalds 已提交
231
{
232
	struct cpufreq_policy *policy = NULL;
L
Linus Torvalds 已提交
233 234
	unsigned long flags;

235
	if (WARN_ON(cpu >= nr_cpu_ids))
236 237
		return NULL;

L
Linus Torvalds 已提交
238
	/* get the cpufreq driver */
239
	read_lock_irqsave(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
240

241 242
	if (cpufreq_driver) {
		/* get the CPU */
243
		policy = cpufreq_cpu_get_raw(cpu);
244 245 246
		if (policy)
			kobject_get(&policy->kobj);
	}
L
Linus Torvalds 已提交
247

248
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
249

250
	return policy;
251
}
L
Linus Torvalds 已提交
252 253
EXPORT_SYMBOL_GPL(cpufreq_cpu_get);

254 255 256 257 258 259 260 261
/**
 * 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().
 */
262
void cpufreq_cpu_put(struct cpufreq_policy *policy)
L
Linus Torvalds 已提交
263
{
264
	kobject_put(&policy->kobj);
L
Linus Torvalds 已提交
265 266 267 268 269 270 271 272 273 274 275 276
}
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
277
 * systems as each CPU might be scaled differently. So, use the arch
L
Linus Torvalds 已提交
278 279
 * per-CPU loops_per_jiffy value wherever possible.
 */
280
static void adjust_jiffies(unsigned long val, struct cpufreq_freqs *ci)
L
Linus Torvalds 已提交
281
{
282 283 284 285
#ifndef CONFIG_SMP
	static unsigned long l_p_j_ref;
	static unsigned int l_p_j_ref_freq;

L
Linus Torvalds 已提交
286 287 288 289 290 291
	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;
292 293
		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
294
	}
295
	if (val == CPUFREQ_POSTCHANGE && ci->old != ci->new) {
296 297
		loops_per_jiffy = cpufreq_scale(l_p_j_ref, l_p_j_ref_freq,
								ci->new);
298 299
		pr_debug("scaling loops_per_jiffy to %lu for frequency %u kHz\n",
			 loops_per_jiffy, ci->new);
L
Linus Torvalds 已提交
300 301
	}
#endif
302
}
L
Linus Torvalds 已提交
303

304
static void __cpufreq_notify_transition(struct cpufreq_policy *policy,
305
		struct cpufreq_freqs *freqs, unsigned int state)
L
Linus Torvalds 已提交
306 307 308
{
	BUG_ON(irqs_disabled());

309 310 311
	if (cpufreq_disabled())
		return;

312
	freqs->flags = cpufreq_driver->flags;
313
	pr_debug("notification %u of frequency transition to %u kHz\n",
314
		 state, freqs->new);
L
Linus Torvalds 已提交
315 316

	switch (state) {
317

L
Linus Torvalds 已提交
318
	case CPUFREQ_PRECHANGE:
319
		/* detect if the driver reported a value as "old frequency"
320 321
		 * which is not equal to what the cpufreq core thinks is
		 * "old frequency".
L
Linus Torvalds 已提交
322
		 */
323
		if (!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
324 325
			if ((policy) && (policy->cpu == freqs->cpu) &&
			    (policy->cur) && (policy->cur != freqs->old)) {
326 327
				pr_debug("Warning: CPU frequency is %u, cpufreq assumed %u kHz\n",
					 freqs->old, policy->cur);
328
				freqs->old = policy->cur;
L
Linus Torvalds 已提交
329 330
			}
		}
331
		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
332
				CPUFREQ_PRECHANGE, freqs);
L
Linus Torvalds 已提交
333 334
		adjust_jiffies(CPUFREQ_PRECHANGE, freqs);
		break;
335

L
Linus Torvalds 已提交
336 337
	case CPUFREQ_POSTCHANGE:
		adjust_jiffies(CPUFREQ_POSTCHANGE, freqs);
338 339
		pr_debug("FREQ: %lu - CPU: %lu\n",
			 (unsigned long)freqs->new, (unsigned long)freqs->cpu);
340
		trace_cpu_frequency(freqs->new, freqs->cpu);
341
		cpufreq_stats_record_transition(policy, freqs->new);
342
		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
343
				CPUFREQ_POSTCHANGE, freqs);
344 345
		if (likely(policy) && likely(policy->cpu == freqs->cpu))
			policy->cur = freqs->new;
L
Linus Torvalds 已提交
346 347 348
		break;
	}
}
349

350 351 352 353 354 355 356 357
/**
 * 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.
 */
358
static void cpufreq_notify_transition(struct cpufreq_policy *policy,
359 360 361 362 363
		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
364

365
/* Do post notifications when there are chances that transition has failed */
366
static void cpufreq_notify_post_transition(struct cpufreq_policy *policy,
367 368 369 370 371 372 373 374 375 376 377
		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);
}

378 379 380
void cpufreq_freq_transition_begin(struct cpufreq_policy *policy,
		struct cpufreq_freqs *freqs)
{
381 382 383 384 385 386 387 388 389 390 391 392

	/*
	 * 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);

393 394 395 396 397 398 399 400 401 402 403
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;
404
	policy->transition_task = current;
405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420

	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;
421
	policy->transition_task = NULL;
422 423 424 425 426

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

427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478
/*
 * Fast frequency switching status count.  Positive means "enabled", negative
 * means "disabled" and 0 means "not decided yet".
 */
static int cpufreq_fast_switch_count;
static DEFINE_MUTEX(cpufreq_fast_switch_lock);

static void cpufreq_list_transition_notifiers(void)
{
	struct notifier_block *nb;

	pr_info("Registered transition notifiers:\n");

	mutex_lock(&cpufreq_transition_notifier_list.mutex);

	for (nb = cpufreq_transition_notifier_list.head; nb; nb = nb->next)
		pr_info("%pF\n", nb->notifier_call);

	mutex_unlock(&cpufreq_transition_notifier_list.mutex);
}

/**
 * cpufreq_enable_fast_switch - Enable fast frequency switching for policy.
 * @policy: cpufreq policy to enable fast frequency switching for.
 *
 * Try to enable fast frequency switching for @policy.
 *
 * The attempt will fail if there is at least one transition notifier registered
 * at this point, as fast frequency switching is quite fundamentally at odds
 * with transition notifiers.  Thus if successful, it will make registration of
 * transition notifiers fail going forward.
 */
void cpufreq_enable_fast_switch(struct cpufreq_policy *policy)
{
	lockdep_assert_held(&policy->rwsem);

	if (!policy->fast_switch_possible)
		return;

	mutex_lock(&cpufreq_fast_switch_lock);
	if (cpufreq_fast_switch_count >= 0) {
		cpufreq_fast_switch_count++;
		policy->fast_switch_enabled = true;
	} else {
		pr_warn("CPU%u: Fast frequency switching not enabled\n",
			policy->cpu);
		cpufreq_list_transition_notifiers();
	}
	mutex_unlock(&cpufreq_fast_switch_lock);
}
EXPORT_SYMBOL_GPL(cpufreq_enable_fast_switch);

479 480 481 482 483
/**
 * cpufreq_disable_fast_switch - Disable fast frequency switching for policy.
 * @policy: cpufreq policy to disable fast frequency switching for.
 */
void cpufreq_disable_fast_switch(struct cpufreq_policy *policy)
484 485 486 487 488 489 490 491 492
{
	mutex_lock(&cpufreq_fast_switch_lock);
	if (policy->fast_switch_enabled) {
		policy->fast_switch_enabled = false;
		if (!WARN_ON(cpufreq_fast_switch_count <= 0))
			cpufreq_fast_switch_count--;
	}
	mutex_unlock(&cpufreq_fast_switch_lock);
}
493
EXPORT_SYMBOL_GPL(cpufreq_disable_fast_switch);
L
Linus Torvalds 已提交
494

495 496 497 498 499 500 501 502 503 504 505 506 507 508 509
/**
 * cpufreq_driver_resolve_freq - Map a target frequency to a driver-supported
 * one.
 * @target_freq: target frequency to resolve.
 *
 * The target to driver frequency mapping is cached in the policy.
 *
 * Return: Lowest driver-supported frequency greater than or equal to the
 * given target_freq, subject to policy (min/max) and driver limitations.
 */
unsigned int cpufreq_driver_resolve_freq(struct cpufreq_policy *policy,
					 unsigned int target_freq)
{
	target_freq = clamp_val(target_freq, policy->min, policy->max);
	policy->cached_target_freq = target_freq;
510 511 512 513 514 515 516 517 518 519

	if (cpufreq_driver->target_index) {
		int idx;

		idx = cpufreq_frequency_table_target(policy, target_freq,
						     CPUFREQ_RELATION_L);
		policy->cached_resolved_idx = idx;
		return policy->freq_table[idx].frequency;
	}

520 521
	if (cpufreq_driver->resolve_freq)
		return cpufreq_driver->resolve_freq(policy, target_freq);
522 523

	return target_freq;
524
}
525
EXPORT_SYMBOL_GPL(cpufreq_driver_resolve_freq);
526

L
Linus Torvalds 已提交
527 528 529
/*********************************************************************
 *                          SYSFS INTERFACE                          *
 *********************************************************************/
530
static ssize_t show_boost(struct kobject *kobj,
531 532 533 534 535 536 537 538 539 540 541 542 543 544 545
				 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)) {
546 547
		pr_err("%s: Cannot %s BOOST!\n",
		       __func__, enable ? "enable" : "disable");
548 549 550
		return -EINVAL;
	}

551 552
	pr_debug("%s: cpufreq BOOST %s\n",
		 __func__, enable ? "enabled" : "disabled");
553 554 555 556

	return count;
}
define_one_global_rw(boost);
L
Linus Torvalds 已提交
557

558
static struct cpufreq_governor *find_governor(const char *str_governor)
559 560 561
{
	struct cpufreq_governor *t;

562
	for_each_governor(t)
563
		if (!strncasecmp(str_governor, t->name, CPUFREQ_NAME_LEN))
564 565 566 567 568
			return t;

	return NULL;
}

L
Linus Torvalds 已提交
569 570 571
/**
 * cpufreq_parse_governor - parse a governor string
 */
D
Dave Jones 已提交
572
static int cpufreq_parse_governor(char *str_governor, unsigned int *policy,
L
Linus Torvalds 已提交
573 574
				struct cpufreq_governor **governor)
{
575
	int err = -EINVAL;
576 577

	if (cpufreq_driver->setpolicy) {
578
		if (!strncasecmp(str_governor, "performance", CPUFREQ_NAME_LEN)) {
L
Linus Torvalds 已提交
579
			*policy = CPUFREQ_POLICY_PERFORMANCE;
580
			err = 0;
581
		} else if (!strncasecmp(str_governor, "powersave",
582
						CPUFREQ_NAME_LEN)) {
L
Linus Torvalds 已提交
583
			*policy = CPUFREQ_POLICY_POWERSAVE;
584
			err = 0;
L
Linus Torvalds 已提交
585
		}
586
	} else {
L
Linus Torvalds 已提交
587
		struct cpufreq_governor *t;
588

589
		mutex_lock(&cpufreq_governor_mutex);
590

591
		t = find_governor(str_governor);
592

593
		if (t == NULL) {
594
			int ret;
595

596 597 598
			mutex_unlock(&cpufreq_governor_mutex);
			ret = request_module("cpufreq_%s", str_governor);
			mutex_lock(&cpufreq_governor_mutex);
599

600
			if (ret == 0)
601
				t = find_governor(str_governor);
602 603
		}

604 605 606
		if (t != NULL) {
			*governor = t;
			err = 0;
L
Linus Torvalds 已提交
607
		}
608

609
		mutex_unlock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
610
	}
611
	return err;
L
Linus Torvalds 已提交
612 613 614
}

/**
615 616
 * cpufreq_per_cpu_attr_read() / show_##file_name() -
 * print out cpufreq information
L
Linus Torvalds 已提交
617 618 619 620 621
 *
 * Write out information from cpufreq_driver->policy[cpu]; object must be
 * "unsigned int".
 */

622 623
#define show_one(file_name, object)			\
static ssize_t show_##file_name				\
D
Dave Jones 已提交
624
(struct cpufreq_policy *policy, char *buf)		\
625
{							\
626
	return sprintf(buf, "%u\n", policy->object);	\
L
Linus Torvalds 已提交
627 628 629 630
}

show_one(cpuinfo_min_freq, cpuinfo.min_freq);
show_one(cpuinfo_max_freq, cpuinfo.max_freq);
631
show_one(cpuinfo_transition_latency, cpuinfo.transition_latency);
L
Linus Torvalds 已提交
632 633
show_one(scaling_min_freq, min);
show_one(scaling_max_freq, max);
634

635 636 637 638 639
__weak unsigned int arch_freq_get_on_cpu(int cpu)
{
	return 0;
}

640
static ssize_t show_scaling_cur_freq(struct cpufreq_policy *policy, char *buf)
641 642
{
	ssize_t ret;
643
	unsigned int freq;
644

645 646 647 648 649
	freq = arch_freq_get_on_cpu(policy->cpu);
	if (freq)
		ret = sprintf(buf, "%u\n", freq);
	else if (cpufreq_driver && cpufreq_driver->setpolicy &&
			cpufreq_driver->get)
650 651 652 653 654
		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
655

656
static int cpufreq_set_policy(struct cpufreq_policy *policy,
657
				struct cpufreq_policy *new_policy);
658

L
Linus Torvalds 已提交
659 660 661 662 663
/**
 * 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
664
(struct cpufreq_policy *policy, const char *buf, size_t count)		\
L
Linus Torvalds 已提交
665
{									\
666
	int ret, temp;							\
L
Linus Torvalds 已提交
667 668
	struct cpufreq_policy new_policy;				\
									\
669
	memcpy(&new_policy, policy, sizeof(*policy));			\
L
Linus Torvalds 已提交
670
									\
671
	ret = sscanf(buf, "%u", &new_policy.object);			\
L
Linus Torvalds 已提交
672 673 674
	if (ret != 1)							\
		return -EINVAL;						\
									\
675
	temp = new_policy.object;					\
676
	ret = cpufreq_set_policy(policy, &new_policy);		\
677 678
	if (!ret)							\
		policy->user_policy.object = temp;			\
L
Linus Torvalds 已提交
679 680 681 682
									\
	return ret ? ret : count;					\
}

683 684
store_one(scaling_min_freq, min);
store_one(scaling_max_freq, max);
L
Linus Torvalds 已提交
685 686 687 688

/**
 * show_cpuinfo_cur_freq - current CPU frequency as detected by hardware
 */
D
Dave Jones 已提交
689 690
static ssize_t show_cpuinfo_cur_freq(struct cpufreq_policy *policy,
					char *buf)
L
Linus Torvalds 已提交
691
{
692
	unsigned int cur_freq = __cpufreq_get(policy);
693 694 695 696 697

	if (cur_freq)
		return sprintf(buf, "%u\n", cur_freq);

	return sprintf(buf, "<unknown>\n");
L
Linus Torvalds 已提交
698 699 700 701 702
}

/**
 * show_scaling_governor - show the current policy for the specified CPU
 */
D
Dave Jones 已提交
703
static ssize_t show_scaling_governor(struct cpufreq_policy *policy, char *buf)
L
Linus Torvalds 已提交
704
{
705
	if (policy->policy == CPUFREQ_POLICY_POWERSAVE)
L
Linus Torvalds 已提交
706 707 708 709
		return sprintf(buf, "powersave\n");
	else if (policy->policy == CPUFREQ_POLICY_PERFORMANCE)
		return sprintf(buf, "performance\n");
	else if (policy->governor)
710
		return scnprintf(buf, CPUFREQ_NAME_PLEN, "%s\n",
711
				policy->governor->name);
L
Linus Torvalds 已提交
712 713 714 715 716 717
	return -EINVAL;
}

/**
 * store_scaling_governor - store policy for the specified CPU
 */
D
Dave Jones 已提交
718 719
static ssize_t store_scaling_governor(struct cpufreq_policy *policy,
					const char *buf, size_t count)
L
Linus Torvalds 已提交
720
{
721
	int ret;
L
Linus Torvalds 已提交
722 723 724
	char	str_governor[16];
	struct cpufreq_policy new_policy;

725
	memcpy(&new_policy, policy, sizeof(*policy));
L
Linus Torvalds 已提交
726

727
	ret = sscanf(buf, "%15s", str_governor);
L
Linus Torvalds 已提交
728 729 730
	if (ret != 1)
		return -EINVAL;

731 732
	if (cpufreq_parse_governor(str_governor, &new_policy.policy,
						&new_policy.governor))
L
Linus Torvalds 已提交
733 734
		return -EINVAL;

735
	ret = cpufreq_set_policy(policy, &new_policy);
736
	return ret ? ret : count;
L
Linus Torvalds 已提交
737 738 739 740 741
}

/**
 * show_scaling_driver - show the cpufreq driver currently loaded
 */
D
Dave Jones 已提交
742
static ssize_t show_scaling_driver(struct cpufreq_policy *policy, char *buf)
L
Linus Torvalds 已提交
743
{
744
	return scnprintf(buf, CPUFREQ_NAME_PLEN, "%s\n", cpufreq_driver->name);
L
Linus Torvalds 已提交
745 746 747 748 749
}

/**
 * show_scaling_available_governors - show the available CPUfreq governors
 */
D
Dave Jones 已提交
750 751
static ssize_t show_scaling_available_governors(struct cpufreq_policy *policy,
						char *buf)
L
Linus Torvalds 已提交
752 753 754 755
{
	ssize_t i = 0;
	struct cpufreq_governor *t;

756
	if (!has_target()) {
L
Linus Torvalds 已提交
757 758 759 760
		i += sprintf(buf, "performance powersave");
		goto out;
	}

761
	for_each_governor(t) {
762 763
		if (i >= (ssize_t) ((PAGE_SIZE / sizeof(char))
		    - (CPUFREQ_NAME_LEN + 2)))
L
Linus Torvalds 已提交
764
			goto out;
765
		i += scnprintf(&buf[i], CPUFREQ_NAME_PLEN, "%s ", t->name);
L
Linus Torvalds 已提交
766
	}
767
out:
L
Linus Torvalds 已提交
768 769 770
	i += sprintf(&buf[i], "\n");
	return i;
}
771

772
ssize_t cpufreq_show_cpus(const struct cpumask *mask, char *buf)
L
Linus Torvalds 已提交
773 774 775 776
{
	ssize_t i = 0;
	unsigned int cpu;

777
	for_each_cpu(cpu, mask) {
L
Linus Torvalds 已提交
778 779 780 781
		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))
782
			break;
L
Linus Torvalds 已提交
783 784 785 786
	}
	i += sprintf(&buf[i], "\n");
	return i;
}
787
EXPORT_SYMBOL_GPL(cpufreq_show_cpus);
L
Linus Torvalds 已提交
788

789 790 791 792 793 794
/**
 * 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)
{
795
	return cpufreq_show_cpus(policy->related_cpus, buf);
796 797 798 799 800 801 802
}

/**
 * show_affected_cpus - show the CPUs affected by each transition
 */
static ssize_t show_affected_cpus(struct cpufreq_policy *policy, char *buf)
{
803
	return cpufreq_show_cpus(policy->cpus, buf);
804 805
}

806
static ssize_t store_scaling_setspeed(struct cpufreq_policy *policy,
D
Dave Jones 已提交
807
					const char *buf, size_t count)
808 809 810 811
{
	unsigned int freq = 0;
	unsigned int ret;

812
	if (!policy->governor || !policy->governor->store_setspeed)
813 814 815 816 817 818 819 820 821 822 823 824 825
		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)
{
826
	if (!policy->governor || !policy->governor->show_setspeed)
827 828 829 830
		return sprintf(buf, "<unsupported>\n");

	return policy->governor->show_setspeed(policy, buf);
}
L
Linus Torvalds 已提交
831

832
/**
833
 * show_bios_limit - show the current cpufreq HW/BIOS limitation
834 835 836 837 838
 */
static ssize_t show_bios_limit(struct cpufreq_policy *policy, char *buf)
{
	unsigned int limit;
	int ret;
839 840
	if (cpufreq_driver->bios_limit) {
		ret = cpufreq_driver->bios_limit(policy->cpu, &limit);
841 842 843 844 845 846
		if (!ret)
			return sprintf(buf, "%u\n", limit);
	}
	return sprintf(buf, "%u\n", policy->cpuinfo.max_freq);
}

847 848 849 850 851 852 853 854 855 856 857 858 859 860
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
861

D
Dave Jones 已提交
862
static struct attribute *default_attrs[] = {
L
Linus Torvalds 已提交
863 864
	&cpuinfo_min_freq.attr,
	&cpuinfo_max_freq.attr,
865
	&cpuinfo_transition_latency.attr,
L
Linus Torvalds 已提交
866 867 868
	&scaling_min_freq.attr,
	&scaling_max_freq.attr,
	&affected_cpus.attr,
869
	&related_cpus.attr,
L
Linus Torvalds 已提交
870 871 872
	&scaling_governor.attr,
	&scaling_driver.attr,
	&scaling_available_governors.attr,
873
	&scaling_setspeed.attr,
L
Linus Torvalds 已提交
874 875 876
	NULL
};

877 878
#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
879

880
static ssize_t show(struct kobject *kobj, struct attribute *attr, char *buf)
L
Linus Torvalds 已提交
881
{
D
Dave Jones 已提交
882 883
	struct cpufreq_policy *policy = to_policy(kobj);
	struct freq_attr *fattr = to_attr(attr);
884
	ssize_t ret;
885

886
	down_read(&policy->rwsem);
887
	ret = fattr->show(policy, buf);
888
	up_read(&policy->rwsem);
889

L
Linus Torvalds 已提交
890 891 892
	return ret;
}

D
Dave Jones 已提交
893 894
static ssize_t store(struct kobject *kobj, struct attribute *attr,
		     const char *buf, size_t count)
L
Linus Torvalds 已提交
895
{
D
Dave Jones 已提交
896 897
	struct cpufreq_policy *policy = to_policy(kobj);
	struct freq_attr *fattr = to_attr(attr);
898
	ssize_t ret = -EINVAL;
899

900
	cpus_read_lock();
901

902 903
	if (cpu_online(policy->cpu)) {
		down_write(&policy->rwsem);
904
		ret = fattr->store(policy, buf, count);
905 906
		up_write(&policy->rwsem);
	}
907

908
	cpus_read_unlock();
909

L
Linus Torvalds 已提交
910 911 912
	return ret;
}

D
Dave Jones 已提交
913
static void cpufreq_sysfs_release(struct kobject *kobj)
L
Linus Torvalds 已提交
914
{
D
Dave Jones 已提交
915
	struct cpufreq_policy *policy = to_policy(kobj);
916
	pr_debug("last reference is dropped\n");
L
Linus Torvalds 已提交
917 918 919
	complete(&policy->kobj_unregister);
}

920
static const struct sysfs_ops sysfs_ops = {
L
Linus Torvalds 已提交
921 922 923 924 925 926 927 928 929 930
	.show	= show,
	.store	= store,
};

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

931
static void add_cpu_dev_symlink(struct cpufreq_policy *policy, unsigned int cpu)
932
{
933 934 935 936 937 938 939 940
	struct device *dev = get_cpu_device(cpu);

	if (!dev)
		return;

	if (cpumask_test_and_set_cpu(cpu, policy->real_cpus))
		return;

941
	dev_dbg(dev, "%s: Adding symlink\n", __func__);
942 943
	if (sysfs_create_link(&dev->kobj, &policy->kobj, "cpufreq"))
		dev_err(dev, "cpufreq symlink creation failed\n");
944 945
}

946 947
static void remove_cpu_dev_symlink(struct cpufreq_policy *policy,
				   struct device *dev)
948
{
949 950
	dev_dbg(dev, "%s: Removing symlink\n", __func__);
	sysfs_remove_link(&dev->kobj, "cpufreq");
951 952
}

953
static int cpufreq_add_dev_interface(struct cpufreq_policy *policy)
954 955 956 957 958
{
	struct freq_attr **drv_attr;
	int ret = 0;

	/* set up files for this cpu device */
959
	drv_attr = cpufreq_driver->attr;
V
Viresh Kumar 已提交
960
	while (drv_attr && *drv_attr) {
961 962
		ret = sysfs_create_file(&policy->kobj, &((*drv_attr)->attr));
		if (ret)
963
			return ret;
964 965
		drv_attr++;
	}
966
	if (cpufreq_driver->get) {
967 968
		ret = sysfs_create_file(&policy->kobj, &cpuinfo_cur_freq.attr);
		if (ret)
969
			return ret;
970
	}
971 972 973

	ret = sysfs_create_file(&policy->kobj, &scaling_cur_freq.attr);
	if (ret)
974
		return ret;
975

976
	if (cpufreq_driver->bios_limit) {
977 978
		ret = sysfs_create_file(&policy->kobj, &bios_limit.attr);
		if (ret)
979
			return ret;
980
	}
981

982
	return 0;
983 984
}

985 986 987 988 989
__weak struct cpufreq_governor *cpufreq_default_governor(void)
{
	return NULL;
}

990
static int cpufreq_init_policy(struct cpufreq_policy *policy)
991
{
992
	struct cpufreq_governor *gov = NULL;
993 994
	struct cpufreq_policy new_policy;

995
	memcpy(&new_policy, policy, sizeof(*policy));
996

997
	/* Update governor of new_policy to the governor used before hotplug */
998
	gov = find_governor(policy->last_governor);
999
	if (gov) {
1000 1001
		pr_debug("Restoring governor %s for cpu %d\n",
				policy->governor->name, policy->cpu);
1002 1003 1004 1005 1006
	} else {
		gov = cpufreq_default_governor();
		if (!gov)
			return -ENODATA;
	}
1007 1008 1009

	new_policy.governor = gov;

1010 1011 1012 1013 1014 1015 1016 1017
	/* 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);
	}
1018
	/* set default policy */
1019
	return cpufreq_set_policy(policy, &new_policy);
1020 1021
}

1022
static int cpufreq_add_policy_cpu(struct cpufreq_policy *policy, unsigned int cpu)
1023
{
1024
	int ret = 0;
1025

1026 1027 1028 1029
	/* Has this CPU been taken care of already? */
	if (cpumask_test_cpu(cpu, policy->cpus))
		return 0;

1030
	down_write(&policy->rwsem);
1031 1032
	if (has_target())
		cpufreq_stop_governor(policy);
1033 1034

	cpumask_set_cpu(cpu, policy->cpus);
V
Viresh Kumar 已提交
1035

1036
	if (has_target()) {
1037
		ret = cpufreq_start_governor(policy);
1038
		if (ret)
1039
			pr_err("%s: Failed to start governor\n", __func__);
1040
	}
1041 1042
	up_write(&policy->rwsem);
	return ret;
1043
}
L
Linus Torvalds 已提交
1044

1045 1046 1047 1048 1049 1050 1051
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);
1052
}
L
Linus Torvalds 已提交
1053

1054
static struct cpufreq_policy *cpufreq_policy_alloc(unsigned int cpu)
1055 1056
{
	struct cpufreq_policy *policy;
1057
	int ret;
1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068

	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;

1069 1070 1071
	if (!zalloc_cpumask_var(&policy->real_cpus, GFP_KERNEL))
		goto err_free_rcpumask;

1072 1073 1074 1075 1076 1077 1078
	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;
	}

1079
	INIT_LIST_HEAD(&policy->policy_list);
1080
	init_rwsem(&policy->rwsem);
1081 1082
	spin_lock_init(&policy->transition_lock);
	init_waitqueue_head(&policy->transition_wait);
1083 1084
	init_completion(&policy->kobj_unregister);
	INIT_WORK(&policy->update, handle_update);
1085

1086
	policy->cpu = cpu;
1087 1088
	return policy;

1089 1090
err_free_real_cpus:
	free_cpumask_var(policy->real_cpus);
1091 1092
err_free_rcpumask:
	free_cpumask_var(policy->related_cpus);
1093 1094 1095 1096 1097 1098 1099 1100
err_free_cpumask:
	free_cpumask_var(policy->cpus);
err_free_policy:
	kfree(policy);

	return NULL;
}

1101
static void cpufreq_policy_put_kobj(struct cpufreq_policy *policy)
1102 1103 1104 1105
{
	struct kobject *kobj;
	struct completion *cmp;

1106
	down_write(&policy->rwsem);
1107
	cpufreq_stats_free_table(policy);
1108 1109
	kobj = &policy->kobj;
	cmp = &policy->kobj_unregister;
1110
	up_write(&policy->rwsem);
1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122
	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");
}

1123
static void cpufreq_policy_free(struct cpufreq_policy *policy)
1124
{
1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135
	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);

1136
	cpufreq_policy_put_kobj(policy);
1137
	free_cpumask_var(policy->real_cpus);
1138 1139 1140 1141 1142
	free_cpumask_var(policy->related_cpus);
	free_cpumask_var(policy->cpus);
	kfree(policy);
}

1143
static int cpufreq_online(unsigned int cpu)
L
Linus Torvalds 已提交
1144
{
1145
	struct cpufreq_policy *policy;
1146
	bool new_policy;
L
Linus Torvalds 已提交
1147
	unsigned long flags;
1148 1149
	unsigned int j;
	int ret;
1150

1151
	pr_debug("%s: bringing CPU%u online\n", __func__, cpu);
1152

1153
	/* Check if this CPU already has a policy to manage it */
1154
	policy = per_cpu(cpufreq_cpu_data, cpu);
1155
	if (policy) {
1156
		WARN_ON(!cpumask_test_cpu(cpu, policy->related_cpus));
1157
		if (!policy_is_inactive(policy))
1158
			return cpufreq_add_policy_cpu(policy, cpu);
L
Linus Torvalds 已提交
1159

1160
		/* This is the only online CPU for the policy.  Start over. */
1161
		new_policy = false;
1162 1163 1164 1165 1166
		down_write(&policy->rwsem);
		policy->cpu = cpu;
		policy->governor = NULL;
		up_write(&policy->rwsem);
	} else {
1167
		new_policy = true;
1168
		policy = cpufreq_policy_alloc(cpu);
1169
		if (!policy)
1170
			return -ENOMEM;
1171
	}
1172

1173
	cpumask_copy(policy->cpus, cpumask_of(cpu));
L
Linus Torvalds 已提交
1174 1175 1176 1177

	/* call driver. From then on the cpufreq must be able
	 * to accept all calls to ->verify and ->setpolicy for this CPU
	 */
1178
	ret = cpufreq_driver->init(policy);
L
Linus Torvalds 已提交
1179
	if (ret) {
1180
		pr_debug("initialization failed\n");
1181
		goto out_free_policy;
L
Linus Torvalds 已提交
1182
	}
V
Viresh Kumar 已提交
1183

1184 1185
	down_write(&policy->rwsem);

1186
	if (new_policy) {
1187
		/* related_cpus should at least include policy->cpus. */
1188
		cpumask_copy(policy->related_cpus, policy->cpus);
1189
	}
1190

1191 1192 1193 1194 1195 1196
	/*
	 * 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);

1197
	if (new_policy) {
1198 1199
		policy->user_policy.min = policy->min;
		policy->user_policy.max = policy->max;
1200

1201
		for_each_cpu(j, policy->related_cpus) {
1202
			per_cpu(cpufreq_cpu_data, j) = policy;
1203 1204
			add_cpu_dev_symlink(policy, j);
		}
1205 1206 1207
	} else {
		policy->min = policy->user_policy.min;
		policy->max = policy->user_policy.max;
1208
	}
1209

1210
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
1211 1212 1213
		policy->cur = cpufreq_driver->get(policy->cpu);
		if (!policy->cur) {
			pr_err("%s: ->get() failed\n", __func__);
1214
			goto out_exit_policy;
1215 1216 1217
		}
	}

1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257
	/*
	 * 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);
		}
	}

1258
	if (new_policy) {
1259
		ret = cpufreq_add_dev_interface(policy);
1260
		if (ret)
1261
			goto out_exit_policy;
1262 1263

		cpufreq_stats_create_table(policy);
1264

1265 1266 1267 1268
		write_lock_irqsave(&cpufreq_driver_lock, flags);
		list_add(&policy->policy_list, &cpufreq_policy_list);
		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
	}
1269

1270 1271 1272 1273
	ret = cpufreq_init_policy(policy);
	if (ret) {
		pr_err("%s: Failed to initialize policy for cpu: %d (%d)\n",
		       __func__, cpu, ret);
1274 1275 1276
		/* cpufreq_policy_free() will notify based on this */
		new_policy = false;
		goto out_exit_policy;
1277
	}
1278

1279
	up_write(&policy->rwsem);
1280

1281
	kobject_uevent(&policy->kobj, KOBJ_ADD);
1282 1283 1284 1285 1286

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

1287
	pr_debug("initialization complete\n");
1288

L
Linus Torvalds 已提交
1289 1290
	return 0;

1291
out_exit_policy:
1292 1293
	up_write(&policy->rwsem);

1294 1295
	if (cpufreq_driver->exit)
		cpufreq_driver->exit(policy);
1296 1297 1298 1299

	for_each_cpu(j, policy->real_cpus)
		remove_cpu_dev_symlink(policy, get_cpu_device(j));

1300
out_free_policy:
1301
	cpufreq_policy_free(policy);
L
Linus Torvalds 已提交
1302 1303 1304
	return ret;
}

1305 1306 1307 1308 1309 1310 1311
/**
 * 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)
{
1312
	struct cpufreq_policy *policy;
1313
	unsigned cpu = dev->id;
1314
	int ret;
1315 1316 1317

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

1318 1319 1320 1321 1322
	if (cpu_online(cpu)) {
		ret = cpufreq_online(cpu);
		if (ret)
			return ret;
	}
1323

1324
	/* Create sysfs link on CPU registration */
1325
	policy = per_cpu(cpufreq_cpu_data, cpu);
1326 1327
	if (policy)
		add_cpu_dev_symlink(policy, cpu);
1328

1329
	return 0;
L
Linus Torvalds 已提交
1330 1331
}

1332
static int cpufreq_offline(unsigned int cpu)
L
Linus Torvalds 已提交
1333
{
1334
	struct cpufreq_policy *policy;
1335
	int ret;
L
Linus Torvalds 已提交
1336

1337
	pr_debug("%s: unregistering CPU %u\n", __func__, cpu);
L
Linus Torvalds 已提交
1338

1339
	policy = cpufreq_cpu_get_raw(cpu);
1340
	if (!policy) {
1341
		pr_debug("%s: No cpu_data found\n", __func__);
1342
		return 0;
L
Linus Torvalds 已提交
1343 1344
	}

1345
	down_write(&policy->rwsem);
1346 1347
	if (has_target())
		cpufreq_stop_governor(policy);
L
Linus Torvalds 已提交
1348

1349
	cpumask_clear_cpu(cpu, policy->cpus);
1350

1351 1352 1353 1354
	if (policy_is_inactive(policy)) {
		if (has_target())
			strncpy(policy->last_governor, policy->governor->name,
				CPUFREQ_NAME_LEN);
1355 1356
		else
			policy->last_policy = policy->policy;
1357 1358 1359 1360
	} else if (cpu == policy->cpu) {
		/* Nominate new CPU */
		policy->cpu = cpumask_any(policy->cpus);
	}
1361

1362 1363 1364
	/* Start governor again for active policy */
	if (!policy_is_inactive(policy)) {
		if (has_target()) {
1365
			ret = cpufreq_start_governor(policy);
1366 1367 1368
			if (ret)
				pr_err("%s: Failed to start governor\n", __func__);
		}
1369

1370
		goto unlock;
1371 1372
	}

1373 1374
	if (cpufreq_driver->stop_cpu)
		cpufreq_driver->stop_cpu(policy);
1375

1376 1377
	if (has_target())
		cpufreq_exit_governor(policy);
L
Linus Torvalds 已提交
1378

1379 1380 1381 1382 1383
	/*
	 * 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.
	 */
1384
	if (cpufreq_driver->exit) {
1385
		cpufreq_driver->exit(policy);
1386 1387
		policy->freq_table = NULL;
	}
1388 1389 1390

unlock:
	up_write(&policy->rwsem);
1391
	return 0;
L
Linus Torvalds 已提交
1392 1393
}

1394
/**
1395
 * cpufreq_remove_dev - remove a CPU device
1396 1397 1398
 *
 * Removes the cpufreq interface for a CPU device.
 */
1399
static void cpufreq_remove_dev(struct device *dev, struct subsys_interface *sif)
1400
{
1401
	unsigned int cpu = dev->id;
1402
	struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);
1403

1404
	if (!policy)
1405
		return;
1406

1407 1408
	if (cpu_online(cpu))
		cpufreq_offline(cpu);
1409

1410
	cpumask_clear_cpu(cpu, policy->real_cpus);
1411
	remove_cpu_dev_symlink(policy, dev);
1412

1413
	if (cpumask_empty(policy->real_cpus))
1414
		cpufreq_policy_free(policy);
1415 1416
}

L
Linus Torvalds 已提交
1417
/**
1418 1419
 *	cpufreq_out_of_sync - If actual and saved CPU frequency differs, we're
 *	in deep trouble.
1420
 *	@policy: policy managing CPUs
L
Linus Torvalds 已提交
1421 1422
 *	@new_freq: CPU frequency the CPU actually runs at
 *
1423 1424
 *	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
1425
 */
1426
static void cpufreq_out_of_sync(struct cpufreq_policy *policy,
1427
				unsigned int new_freq)
L
Linus Torvalds 已提交
1428 1429
{
	struct cpufreq_freqs freqs;
1430

1431
	pr_debug("Warning: CPU frequency out of sync: cpufreq and timing core thinks of %u, is %u kHz\n",
1432
		 policy->cur, new_freq);
L
Linus Torvalds 已提交
1433

1434
	freqs.old = policy->cur;
L
Linus Torvalds 已提交
1435
	freqs.new = new_freq;
1436

1437 1438
	cpufreq_freq_transition_begin(policy, &freqs);
	cpufreq_freq_transition_end(policy, &freqs, 0);
L
Linus Torvalds 已提交
1439 1440
}

1441
/**
D
Dhaval Giani 已提交
1442
 * cpufreq_quick_get - get the CPU frequency (in kHz) from policy->cur
1443 1444 1445 1446 1447 1448 1449
 * @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)
{
1450
	struct cpufreq_policy *policy;
1451
	unsigned int ret_freq = 0;
1452
	unsigned long flags;
1453

1454 1455 1456 1457 1458 1459 1460 1461 1462
	read_lock_irqsave(&cpufreq_driver_lock, flags);

	if (cpufreq_driver && cpufreq_driver->setpolicy && cpufreq_driver->get) {
		ret_freq = cpufreq_driver->get(cpu);
		read_unlock_irqrestore(&cpufreq_driver_lock, flags);
		return ret_freq;
	}

	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
1463 1464

	policy = cpufreq_cpu_get(cpu);
1465
	if (policy) {
1466
		ret_freq = policy->cur;
1467 1468 1469
		cpufreq_cpu_put(policy);
	}

D
Dave Jones 已提交
1470
	return ret_freq;
1471 1472 1473
}
EXPORT_SYMBOL(cpufreq_quick_get);

1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493
/**
 * 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);

1494
static unsigned int __cpufreq_get(struct cpufreq_policy *policy)
L
Linus Torvalds 已提交
1495
{
1496
	unsigned int ret_freq = 0;
1497

1498
	if (!cpufreq_driver->get)
D
Dave Jones 已提交
1499
		return ret_freq;
L
Linus Torvalds 已提交
1500

1501
	ret_freq = cpufreq_driver->get(policy->cpu);
L
Linus Torvalds 已提交
1502

1503 1504 1505 1506 1507 1508
	/*
	 * Updating inactive policies is invalid, so avoid doing that.  Also
	 * if fast frequency switching is used with the given policy, the check
	 * against policy->cur is pointless, so skip it in that case too.
	 */
	if (unlikely(policy_is_inactive(policy)) || policy->fast_switch_enabled)
1509 1510
		return ret_freq;

1511
	if (ret_freq && policy->cur &&
1512
		!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
1513 1514 1515
		/* verify no discrepancy between actual and
					saved value exists */
		if (unlikely(ret_freq != policy->cur)) {
1516
			cpufreq_out_of_sync(policy, ret_freq);
L
Linus Torvalds 已提交
1517 1518 1519 1520
			schedule_work(&policy->update);
		}
	}

D
Dave Jones 已提交
1521
	return ret_freq;
1522
}
L
Linus Torvalds 已提交
1523

1524 1525 1526 1527 1528 1529 1530 1531
/**
 * 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)
{
1532
	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
1533 1534
	unsigned int ret_freq = 0;

1535 1536
	if (policy) {
		down_read(&policy->rwsem);
1537 1538 1539 1540

		if (!policy_is_inactive(policy))
			ret_freq = __cpufreq_get(policy);

1541
		up_read(&policy->rwsem);
1542

1543 1544
		cpufreq_cpu_put(policy);
	}
1545

D
Dave Jones 已提交
1546
	return ret_freq;
L
Linus Torvalds 已提交
1547 1548 1549
}
EXPORT_SYMBOL(cpufreq_get);

1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567
static unsigned int cpufreq_update_current_freq(struct cpufreq_policy *policy)
{
	unsigned int new_freq;

	new_freq = cpufreq_driver->get(policy->cpu);
	if (!new_freq)
		return 0;

	if (!policy->cur) {
		pr_debug("cpufreq: Driver did not initialize current freq\n");
		policy->cur = new_freq;
	} else if (policy->cur != new_freq && has_target()) {
		cpufreq_out_of_sync(policy, new_freq);
	}

	return new_freq;
}

1568 1569 1570 1571 1572
static struct subsys_interface cpufreq_interface = {
	.name		= "cpufreq",
	.subsys		= &cpu_subsys,
	.add_dev	= cpufreq_add_dev,
	.remove_dev	= cpufreq_remove_dev,
1573 1574
};

1575 1576 1577 1578 1579 1580 1581 1582 1583
/*
 * 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) {
1584 1585
		pr_debug("%s: suspend_freq not defined\n", __func__);
		return 0;
1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600
	}

	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);

1601
/**
1602
 * cpufreq_suspend() - Suspend CPUFreq governors
1603
 *
1604 1605 1606 1607
 * 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.
1608
 */
1609
void cpufreq_suspend(void)
1610
{
1611
	struct cpufreq_policy *policy;
1612

1613 1614
	if (!cpufreq_driver)
		return;
1615

1616
	if (!has_target() && !cpufreq_driver->suspend)
1617
		goto suspend;
1618

1619 1620
	pr_debug("%s: Suspending Governors\n", __func__);

1621
	for_each_active_policy(policy) {
1622 1623
		if (has_target()) {
			down_write(&policy->rwsem);
1624
			cpufreq_stop_governor(policy);
1625 1626 1627 1628
			up_write(&policy->rwsem);
		}

		if (cpufreq_driver->suspend && cpufreq_driver->suspend(policy))
1629 1630
			pr_err("%s: Failed to suspend driver: %p\n", __func__,
				policy);
1631
	}
1632 1633 1634

suspend:
	cpufreq_suspended = true;
1635 1636
}

L
Linus Torvalds 已提交
1637
/**
1638
 * cpufreq_resume() - Resume CPUFreq governors
L
Linus Torvalds 已提交
1639
 *
1640 1641
 * Called during system wide Suspend/Hibernate cycle for resuming governors that
 * are suspended with cpufreq_suspend().
L
Linus Torvalds 已提交
1642
 */
1643
void cpufreq_resume(void)
L
Linus Torvalds 已提交
1644
{
1645
	struct cpufreq_policy *policy;
1646
	int ret;
L
Linus Torvalds 已提交
1647

1648 1649
	if (!cpufreq_driver)
		return;
L
Linus Torvalds 已提交
1650

1651 1652
	cpufreq_suspended = false;

1653
	if (!has_target() && !cpufreq_driver->resume)
1654
		return;
L
Linus Torvalds 已提交
1655

1656
	pr_debug("%s: Resuming Governors\n", __func__);
L
Linus Torvalds 已提交
1657

1658
	for_each_active_policy(policy) {
1659
		if (cpufreq_driver->resume && cpufreq_driver->resume(policy)) {
1660 1661
			pr_err("%s: Failed to resume driver: %p\n", __func__,
				policy);
1662
		} else if (has_target()) {
1663
			down_write(&policy->rwsem);
1664
			ret = cpufreq_start_governor(policy);
1665 1666 1667 1668 1669 1670
			up_write(&policy->rwsem);

			if (ret)
				pr_err("%s: Failed to start governor for policy: %p\n",
				       __func__, policy);
		}
1671 1672
	}
}
L
Linus Torvalds 已提交
1673

1674 1675 1676 1677 1678 1679 1680 1681
/**
 *	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)
{
1682 1683 1684 1685
	if (cpufreq_driver)
		return cpufreq_driver->name;

	return NULL;
1686 1687
}
EXPORT_SYMBOL_GPL(cpufreq_get_current_driver);
L
Linus Torvalds 已提交
1688

1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703
/**
 *	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
1704 1705 1706 1707 1708 1709 1710 1711 1712
/*********************************************************************
 *                     NOTIFIER LISTS INTERFACE                      *
 *********************************************************************/

/**
 *	cpufreq_register_notifier - register a driver with cpufreq
 *	@nb: notifier function to register
 *      @list: CPUFREQ_TRANSITION_NOTIFIER or CPUFREQ_POLICY_NOTIFIER
 *
1713
 *	Add a driver to one of two lists: either a list of drivers that
L
Linus Torvalds 已提交
1714 1715 1716 1717 1718
 *      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
1719
 *	blocking_notifier_chain_register.
L
Linus Torvalds 已提交
1720 1721 1722 1723 1724
 */
int cpufreq_register_notifier(struct notifier_block *nb, unsigned int list)
{
	int ret;

1725 1726 1727
	if (cpufreq_disabled())
		return -EINVAL;

1728 1729
	WARN_ON(!init_cpufreq_transition_notifier_list_called);

L
Linus Torvalds 已提交
1730 1731
	switch (list) {
	case CPUFREQ_TRANSITION_NOTIFIER:
1732 1733 1734 1735 1736 1737
		mutex_lock(&cpufreq_fast_switch_lock);

		if (cpufreq_fast_switch_count > 0) {
			mutex_unlock(&cpufreq_fast_switch_lock);
			return -EBUSY;
		}
1738
		ret = srcu_notifier_chain_register(
1739
				&cpufreq_transition_notifier_list, nb);
1740 1741 1742 1743
		if (!ret)
			cpufreq_fast_switch_count--;

		mutex_unlock(&cpufreq_fast_switch_lock);
L
Linus Torvalds 已提交
1744 1745
		break;
	case CPUFREQ_POLICY_NOTIFIER:
1746 1747
		ret = blocking_notifier_chain_register(
				&cpufreq_policy_notifier_list, nb);
L
Linus Torvalds 已提交
1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759
		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
1760
 *	@list: CPUFREQ_TRANSITION_NOTIFIER or CPUFREQ_POLICY_NOTIFIER
L
Linus Torvalds 已提交
1761 1762 1763 1764
 *
 *	Remove a driver from the CPU frequency notifier list.
 *
 *	This function may sleep, and has the same return conditions as
1765
 *	blocking_notifier_chain_unregister.
L
Linus Torvalds 已提交
1766 1767 1768 1769 1770
 */
int cpufreq_unregister_notifier(struct notifier_block *nb, unsigned int list)
{
	int ret;

1771 1772 1773
	if (cpufreq_disabled())
		return -EINVAL;

L
Linus Torvalds 已提交
1774 1775
	switch (list) {
	case CPUFREQ_TRANSITION_NOTIFIER:
1776 1777
		mutex_lock(&cpufreq_fast_switch_lock);

1778
		ret = srcu_notifier_chain_unregister(
1779
				&cpufreq_transition_notifier_list, nb);
1780 1781 1782 1783
		if (!ret && !WARN_ON(cpufreq_fast_switch_count >= 0))
			cpufreq_fast_switch_count++;

		mutex_unlock(&cpufreq_fast_switch_lock);
L
Linus Torvalds 已提交
1784 1785
		break;
	case CPUFREQ_POLICY_NOTIFIER:
1786 1787
		ret = blocking_notifier_chain_unregister(
				&cpufreq_policy_notifier_list, nb);
L
Linus Torvalds 已提交
1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801
		break;
	default:
		ret = -EINVAL;
	}

	return ret;
}
EXPORT_SYMBOL(cpufreq_unregister_notifier);


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

1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826
/**
 * cpufreq_driver_fast_switch - Carry out a fast CPU frequency switch.
 * @policy: cpufreq policy to switch the frequency for.
 * @target_freq: New frequency to set (may be approximate).
 *
 * Carry out a fast frequency switch without sleeping.
 *
 * The driver's ->fast_switch() callback invoked by this function must be
 * suitable for being called from within RCU-sched read-side critical sections
 * and it is expected to select the minimum available frequency greater than or
 * equal to @target_freq (CPUFREQ_RELATION_L).
 *
 * This function must not be called if policy->fast_switch_enabled is unset.
 *
 * Governors calling this function must guarantee that it will never be invoked
 * twice in parallel for the same policy and that it will never be called in
 * parallel with either ->target() or ->target_index() for the same policy.
 *
 * If CPUFREQ_ENTRY_INVALID is returned by the driver's ->fast_switch()
 * callback to indicate an error condition, the hardware configuration must be
 * preserved.
 */
unsigned int cpufreq_driver_fast_switch(struct cpufreq_policy *policy,
					unsigned int target_freq)
{
1827
	target_freq = clamp_val(target_freq, policy->min, policy->max);
1828 1829 1830 1831 1832

	return cpufreq_driver->fast_switch(policy, target_freq);
}
EXPORT_SYMBOL_GPL(cpufreq_driver_fast_switch);

1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858
/* 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;
}

1859
static int __target_index(struct cpufreq_policy *policy, int index)
1860
{
1861 1862
	struct cpufreq_freqs freqs = {.old = policy->cur, .flags = 0};
	unsigned int intermediate_freq = 0;
1863
	unsigned int newfreq = policy->freq_table[index].frequency;
1864 1865 1866
	int retval = -EINVAL;
	bool notify;

1867 1868 1869
	if (newfreq == policy->cur)
		return 0;

1870 1871
	notify = !(cpufreq_driver->flags & CPUFREQ_ASYNC_NOTIFICATION);
	if (notify) {
1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882
		/* 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;
		}
1883

1884
		freqs.new = newfreq;
1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895
		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);

1896
	if (notify) {
1897 1898
		cpufreq_freq_transition_end(policy, &freqs, retval);

1899 1900 1901 1902
		/*
		 * 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
1903
		 * case we haven't switched to intermediate freq at all.
1904 1905 1906 1907 1908 1909 1910 1911 1912
		 */
		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);
		}
	}

1913 1914 1915
	return retval;
}

L
Linus Torvalds 已提交
1916 1917 1918 1919
int __cpufreq_driver_target(struct cpufreq_policy *policy,
			    unsigned int target_freq,
			    unsigned int relation)
{
1920
	unsigned int old_target_freq = target_freq;
1921
	int index;
1922

1923 1924 1925
	if (cpufreq_disabled())
		return -ENODEV;

1926
	/* Make sure that target_freq is within supported range */
1927
	target_freq = clamp_val(target_freq, policy->min, policy->max);
1928 1929

	pr_debug("target for CPU %u: %u kHz, relation %u, requested %u kHz\n",
1930
		 policy->cpu, target_freq, relation, old_target_freq);
1931

1932 1933 1934 1935 1936 1937
	/*
	 * 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.
	 */
1938 1939 1940
	if (target_freq == policy->cur)
		return 0;

1941 1942 1943
	/* Save last value to restore later on errors */
	policy->restore_freq = policy->cur;

1944
	if (cpufreq_driver->target)
1945
		return cpufreq_driver->target(policy, target_freq, relation);
1946

1947 1948
	if (!cpufreq_driver->target_index)
		return -EINVAL;
1949

1950
	index = cpufreq_frequency_table_target(policy, target_freq, relation);
1951

1952
	return __target_index(policy, index);
L
Linus Torvalds 已提交
1953 1954 1955 1956 1957 1958 1959
}
EXPORT_SYMBOL_GPL(__cpufreq_driver_target);

int cpufreq_driver_target(struct cpufreq_policy *policy,
			  unsigned int target_freq,
			  unsigned int relation)
{
1960
	int ret = -EINVAL;
L
Linus Torvalds 已提交
1961

1962
	down_write(&policy->rwsem);
L
Linus Torvalds 已提交
1963 1964 1965

	ret = __cpufreq_driver_target(policy, target_freq, relation);

1966
	up_write(&policy->rwsem);
L
Linus Torvalds 已提交
1967 1968 1969 1970 1971

	return ret;
}
EXPORT_SYMBOL_GPL(cpufreq_driver_target);

1972 1973 1974 1975 1976
__weak struct cpufreq_governor *cpufreq_fallback_governor(void)
{
	return NULL;
}

1977
static int cpufreq_init_governor(struct cpufreq_policy *policy)
L
Linus Torvalds 已提交
1978
{
1979
	int ret;
1980

1981 1982 1983
	/* Don't start any governor operations if we are entering suspend */
	if (cpufreq_suspended)
		return 0;
1984 1985 1986 1987 1988 1989
	/*
	 * Governor might not be initiated here if ACPI _PPC changed
	 * notification happened, so check it.
	 */
	if (!policy->governor)
		return -EINVAL;
1990

1991 1992 1993
	if (policy->governor->max_transition_latency &&
	    policy->cpuinfo.transition_latency >
	    policy->governor->max_transition_latency) {
1994 1995 1996
		struct cpufreq_governor *gov = cpufreq_fallback_governor();

		if (gov) {
1997 1998
			pr_warn("%s governor failed, too long transition latency of HW, fallback to %s governor\n",
				policy->governor->name, gov->name);
1999
			policy->governor = gov;
2000 2001
		} else {
			return -EINVAL;
2002
		}
2003
	}
L
Linus Torvalds 已提交
2004

2005 2006
	if (!try_module_get(policy->governor->owner))
		return -EINVAL;
2007

2008
	pr_debug("%s: for CPU %u\n", __func__, policy->cpu);
L
Linus Torvalds 已提交
2009

2010 2011 2012
	if (policy->governor->init) {
		ret = policy->governor->init(policy);
		if (ret) {
2013
			module_put(policy->governor->owner);
2014 2015
			return ret;
		}
2016
	}
L
Linus Torvalds 已提交
2017

2018 2019 2020 2021 2022 2023 2024 2025 2026 2027
	return 0;
}

static void cpufreq_exit_governor(struct cpufreq_policy *policy)
{
	if (cpufreq_suspended || !policy->governor)
		return;

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

2028 2029
	if (policy->governor->exit)
		policy->governor->exit(policy);
2030 2031

	module_put(policy->governor->owner);
L
Linus Torvalds 已提交
2032 2033
}

2034 2035 2036 2037
static int cpufreq_start_governor(struct cpufreq_policy *policy)
{
	int ret;

2038 2039 2040 2041 2042 2043 2044 2045
	if (cpufreq_suspended)
		return 0;

	if (!policy->governor)
		return -EINVAL;

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

2046 2047 2048
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy)
		cpufreq_update_current_freq(policy);

2049 2050 2051 2052 2053 2054 2055 2056
	if (policy->governor->start) {
		ret = policy->governor->start(policy);
		if (ret)
			return ret;
	}

	if (policy->governor->limits)
		policy->governor->limits(policy);
2057 2058

	return 0;
2059 2060
}

2061 2062 2063 2064 2065 2066 2067
static void cpufreq_stop_governor(struct cpufreq_policy *policy)
{
	if (cpufreq_suspended || !policy->governor)
		return;

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

2068 2069
	if (policy->governor->stop)
		policy->governor->stop(policy);
2070 2071 2072 2073 2074 2075 2076 2077 2078
}

static void cpufreq_governor_limits(struct cpufreq_policy *policy)
{
	if (cpufreq_suspended || !policy->governor)
		return;

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

2079 2080
	if (policy->governor->limits)
		policy->governor->limits(policy);
2081 2082
}

L
Linus Torvalds 已提交
2083 2084
int cpufreq_register_governor(struct cpufreq_governor *governor)
{
2085
	int err;
L
Linus Torvalds 已提交
2086 2087 2088 2089

	if (!governor)
		return -EINVAL;

2090 2091 2092
	if (cpufreq_disabled())
		return -ENODEV;

2093
	mutex_lock(&cpufreq_governor_mutex);
2094

2095
	err = -EBUSY;
2096
	if (!find_governor(governor->name)) {
2097 2098
		err = 0;
		list_add(&governor->governor_list, &cpufreq_governor_list);
L
Linus Torvalds 已提交
2099 2100
	}

2101
	mutex_unlock(&cpufreq_governor_mutex);
2102
	return err;
L
Linus Torvalds 已提交
2103 2104 2105 2106 2107
}
EXPORT_SYMBOL_GPL(cpufreq_register_governor);

void cpufreq_unregister_governor(struct cpufreq_governor *governor)
{
2108 2109
	struct cpufreq_policy *policy;
	unsigned long flags;
2110

L
Linus Torvalds 已提交
2111 2112 2113
	if (!governor)
		return;

2114 2115 2116
	if (cpufreq_disabled())
		return;

2117 2118 2119
	/* clear last_governor for all inactive policies */
	read_lock_irqsave(&cpufreq_driver_lock, flags);
	for_each_inactive_policy(policy) {
2120 2121
		if (!strcmp(policy->last_governor, governor->name)) {
			policy->governor = NULL;
2122
			strcpy(policy->last_governor, "\0");
2123
		}
2124
	}
2125
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
2126

2127
	mutex_lock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
2128
	list_del(&governor->governor_list);
2129
	mutex_unlock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140
	return;
}
EXPORT_SYMBOL_GPL(cpufreq_unregister_governor);


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

/**
 * cpufreq_get_policy - get the current cpufreq_policy
2141 2142
 * @policy: struct cpufreq_policy into which the current cpufreq_policy
 *	is written
L
Linus Torvalds 已提交
2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155
 *
 * 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;

2156
	memcpy(policy, cpu_policy, sizeof(*policy));
L
Linus Torvalds 已提交
2157 2158 2159 2160 2161 2162

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

2163
/*
2164 2165
 * policy : current policy.
 * new_policy: policy to be set.
2166
 */
2167
static int cpufreq_set_policy(struct cpufreq_policy *policy,
2168
				struct cpufreq_policy *new_policy)
L
Linus Torvalds 已提交
2169
{
2170 2171
	struct cpufreq_governor *old_gov;
	int ret;
L
Linus Torvalds 已提交
2172

2173 2174
	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
2175

2176
	memcpy(&new_policy->cpuinfo, &policy->cpuinfo, sizeof(policy->cpuinfo));
L
Linus Torvalds 已提交
2177

2178 2179 2180 2181 2182
	/*
	* 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)
2183
		return -EINVAL;
2184

L
Linus Torvalds 已提交
2185
	/* verify the cpu speed can be set within this limit */
2186
	ret = cpufreq_driver->verify(new_policy);
L
Linus Torvalds 已提交
2187
	if (ret)
2188
		return ret;
L
Linus Torvalds 已提交
2189 2190

	/* adjust if necessary - all reasons */
2191
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2192
			CPUFREQ_ADJUST, new_policy);
L
Linus Torvalds 已提交
2193

2194 2195 2196 2197
	/*
	 * verify the cpu speed can be set within this limit, which might be
	 * different to the first one
	 */
2198
	ret = cpufreq_driver->verify(new_policy);
2199
	if (ret)
2200
		return ret;
L
Linus Torvalds 已提交
2201 2202

	/* notification of the new policy */
2203
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2204
			CPUFREQ_NOTIFY, new_policy);
L
Linus Torvalds 已提交
2205

2206 2207
	policy->min = new_policy->min;
	policy->max = new_policy->max;
L
Linus Torvalds 已提交
2208

2209 2210
	policy->cached_target_freq = UINT_MAX;

2211
	pr_debug("new min and max freqs are %u - %u kHz\n",
2212
		 policy->min, policy->max);
L
Linus Torvalds 已提交
2213

2214
	if (cpufreq_driver->setpolicy) {
2215
		policy->policy = new_policy->policy;
2216
		pr_debug("setting range\n");
2217 2218
		return cpufreq_driver->setpolicy(new_policy);
	}
L
Linus Torvalds 已提交
2219

2220 2221
	if (new_policy->governor == policy->governor) {
		pr_debug("cpufreq: governor limits update\n");
2222
		cpufreq_governor_limits(policy);
2223
		return 0;
2224
	}
2225

2226 2227 2228 2229 2230 2231
	pr_debug("governor switch\n");

	/* save old, working values */
	old_gov = policy->governor;
	/* end old governor */
	if (old_gov) {
2232
		cpufreq_stop_governor(policy);
2233
		cpufreq_exit_governor(policy);
L
Linus Torvalds 已提交
2234 2235
	}

2236 2237
	/* start new governor */
	policy->governor = new_policy->governor;
2238
	ret = cpufreq_init_governor(policy);
2239
	if (!ret) {
2240 2241 2242 2243 2244
		ret = cpufreq_start_governor(policy);
		if (!ret) {
			pr_debug("cpufreq: governor change\n");
			return 0;
		}
2245
		cpufreq_exit_governor(policy);
2246 2247 2248 2249 2250 2251
	}

	/* 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;
2252
		if (cpufreq_init_governor(policy))
2253 2254
			policy->governor = NULL;
		else
2255
			cpufreq_start_governor(policy);
2256 2257
	}

2258
	return ret;
L
Linus Torvalds 已提交
2259 2260 2261 2262 2263 2264
}

/**
 *	cpufreq_update_policy - re-evaluate an existing cpufreq policy
 *	@cpu: CPU which shall be re-evaluated
 *
L
Lucas De Marchi 已提交
2265
 *	Useful for policy notifiers which have different necessities
L
Linus Torvalds 已提交
2266 2267
 *	at different times.
 */
2268
void cpufreq_update_policy(unsigned int cpu)
L
Linus Torvalds 已提交
2269
{
2270 2271
	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
	struct cpufreq_policy new_policy;
L
Linus Torvalds 已提交
2272

2273
	if (!policy)
2274
		return;
L
Linus Torvalds 已提交
2275

2276
	down_write(&policy->rwsem);
L
Linus Torvalds 已提交
2277

2278
	if (policy_is_inactive(policy))
2279 2280
		goto unlock;

2281
	pr_debug("updating policy for CPU %u\n", cpu);
2282
	memcpy(&new_policy, policy, sizeof(*policy));
2283 2284
	new_policy.min = policy->user_policy.min;
	new_policy.max = policy->user_policy.max;
L
Linus Torvalds 已提交
2285

2286 2287 2288 2289
	/*
	 * BIOS might change freq behind our back
	 * -> ask driver for current freq and notify governors about a change
	 */
2290
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
2291
		if (cpufreq_suspended)
2292
			goto unlock;
2293

2294
		new_policy.cur = cpufreq_update_current_freq(policy);
2295
		if (WARN_ON(!new_policy.cur))
2296
			goto unlock;
2297 2298
	}

2299
	cpufreq_set_policy(policy, &new_policy);
L
Linus Torvalds 已提交
2300

2301
unlock:
2302
	up_write(&policy->rwsem);
2303

2304
	cpufreq_cpu_put(policy);
L
Linus Torvalds 已提交
2305 2306 2307
}
EXPORT_SYMBOL(cpufreq_update_policy);

2308 2309 2310 2311 2312 2313 2314 2315
/*********************************************************************
 *               BOOST						     *
 *********************************************************************/
static int cpufreq_boost_set_sw(int state)
{
	struct cpufreq_policy *policy;
	int ret = -EINVAL;

2316
	for_each_active_policy(policy) {
2317 2318
		if (!policy->freq_table)
			continue;
2319

2320 2321 2322 2323 2324 2325
		ret = cpufreq_frequency_table_cpuinfo(policy,
						      policy->freq_table);
		if (ret) {
			pr_err("%s: Policy frequency update failed\n",
			       __func__);
			break;
2326
		}
2327 2328 2329 2330 2331

		down_write(&policy->rwsem);
		policy->user_policy.max = policy->max;
		cpufreq_governor_limits(policy);
		up_write(&policy->rwsem);
2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354
	}

	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);

2355 2356
		pr_err("%s: Cannot %s BOOST\n",
		       __func__, state ? "enable" : "disable");
2357 2358 2359 2360 2361
	}

	return ret;
}

2362
static bool cpufreq_boost_supported(void)
2363
{
2364
	return likely(cpufreq_driver) && cpufreq_driver->set_boost;
2365 2366
}

2367 2368 2369 2370
static int create_boost_sysfs_file(void)
{
	int ret;

2371
	ret = sysfs_create_file(cpufreq_global_kobject, &boost.attr);
2372 2373 2374 2375 2376 2377 2378 2379 2380 2381
	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())
2382
		sysfs_remove_file(cpufreq_global_kobject, &boost.attr);
2383 2384 2385 2386 2387 2388 2389 2390 2391 2392
}

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

	if (cpufreq_boost_supported())
		return 0;

2393
	cpufreq_driver->set_boost = cpufreq_boost_set_sw;
2394 2395 2396 2397 2398 2399

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

2400 2401 2402 2403 2404 2405
int cpufreq_boost_enabled(void)
{
	return cpufreq_driver->boost_enabled;
}
EXPORT_SYMBOL_GPL(cpufreq_boost_enabled);

L
Linus Torvalds 已提交
2406 2407 2408
/*********************************************************************
 *               REGISTER / UNREGISTER CPUFREQ DRIVER                *
 *********************************************************************/
2409
static enum cpuhp_state hp_online;
L
Linus Torvalds 已提交
2410

2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424
static int cpuhp_cpufreq_online(unsigned int cpu)
{
	cpufreq_online(cpu);

	return 0;
}

static int cpuhp_cpufreq_offline(unsigned int cpu)
{
	cpufreq_offline(cpu);

	return 0;
}

L
Linus Torvalds 已提交
2425 2426 2427 2428 2429
/**
 * cpufreq_register_driver - register a CPU Frequency driver
 * @driver_data: A struct cpufreq_driver containing the values#
 * submitted by the CPU Frequency driver.
 *
2430
 * Registers a CPU Frequency driver to this core code. This code
2431
 * returns zero on success, -EEXIST when another driver got here first
2432
 * (and isn't unregistered in the meantime).
L
Linus Torvalds 已提交
2433 2434
 *
 */
2435
int cpufreq_register_driver(struct cpufreq_driver *driver_data)
L
Linus Torvalds 已提交
2436 2437 2438 2439
{
	unsigned long flags;
	int ret;

2440 2441 2442
	if (cpufreq_disabled())
		return -ENODEV;

L
Linus Torvalds 已提交
2443
	if (!driver_data || !driver_data->verify || !driver_data->init ||
2444
	    !(driver_data->setpolicy || driver_data->target_index ||
2445 2446
		    driver_data->target) ||
	     (driver_data->setpolicy && (driver_data->target_index ||
2447 2448
		    driver_data->target)) ||
	     (!!driver_data->get_intermediate != !!driver_data->target_intermediate))
L
Linus Torvalds 已提交
2449 2450
		return -EINVAL;

2451
	pr_debug("trying to register driver %s\n", driver_data->name);
L
Linus Torvalds 已提交
2452

2453
	/* Protect against concurrent CPU online/offline. */
2454
	cpus_read_lock();
2455

2456
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2457
	if (cpufreq_driver) {
2458
		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2459 2460
		ret = -EEXIST;
		goto out;
L
Linus Torvalds 已提交
2461
	}
2462
	cpufreq_driver = driver_data;
2463
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
2464

2465 2466 2467
	if (driver_data->setpolicy)
		driver_data->flags |= CPUFREQ_CONST_LOOPS;

2468 2469 2470 2471 2472
	if (cpufreq_boost_supported()) {
		ret = create_boost_sysfs_file();
		if (ret)
			goto err_null_driver;
	}
2473

2474
	ret = subsys_interface_register(&cpufreq_interface);
2475
	if (ret)
2476
		goto err_boost_unreg;
L
Linus Torvalds 已提交
2477

2478 2479
	if (!(cpufreq_driver->flags & CPUFREQ_STICKY) &&
	    list_empty(&cpufreq_policy_list)) {
L
Linus Torvalds 已提交
2480
		/* if all ->init() calls failed, unregister */
2481
		ret = -ENODEV;
2482 2483 2484
		pr_debug("%s: No CPU initialized for driver %s\n", __func__,
			 driver_data->name);
		goto err_if_unreg;
L
Linus Torvalds 已提交
2485 2486
	}

2487 2488 2489 2490
	ret = cpuhp_setup_state_nocalls_cpuslocked(CPUHP_AP_ONLINE_DYN,
						   "cpufreq:online",
						   cpuhp_cpufreq_online,
						   cpuhp_cpufreq_offline);
2491 2492 2493
	if (ret < 0)
		goto err_if_unreg;
	hp_online = ret;
2494
	ret = 0;
2495

2496
	pr_debug("driver %s up and running\n", driver_data->name);
2497
	goto out;
2498

2499 2500
err_if_unreg:
	subsys_interface_unregister(&cpufreq_interface);
2501
err_boost_unreg:
2502
	remove_boost_sysfs_file();
2503
err_null_driver:
2504
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2505
	cpufreq_driver = NULL;
2506
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2507
out:
2508
	cpus_read_unlock();
2509
	return ret;
L
Linus Torvalds 已提交
2510 2511 2512 2513 2514 2515
}
EXPORT_SYMBOL_GPL(cpufreq_register_driver);

/**
 * cpufreq_unregister_driver - unregister the current CPUFreq driver
 *
2516
 * Unregister the current CPUFreq driver. Only call this if you have
L
Linus Torvalds 已提交
2517 2518 2519 2520
 * 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.
 */
2521
int cpufreq_unregister_driver(struct cpufreq_driver *driver)
L
Linus Torvalds 已提交
2522 2523 2524
{
	unsigned long flags;

2525
	if (!cpufreq_driver || (driver != cpufreq_driver))
L
Linus Torvalds 已提交
2526 2527
		return -EINVAL;

2528
	pr_debug("unregistering driver %s\n", driver->name);
L
Linus Torvalds 已提交
2529

2530
	/* Protect against concurrent cpu hotplug */
2531
	cpus_read_lock();
2532
	subsys_interface_unregister(&cpufreq_interface);
2533
	remove_boost_sysfs_file();
2534
	cpuhp_remove_state_nocalls_cpuslocked(hp_online);
L
Linus Torvalds 已提交
2535

2536
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2537

2538
	cpufreq_driver = NULL;
2539

2540
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2541
	cpus_read_unlock();
L
Linus Torvalds 已提交
2542 2543 2544 2545

	return 0;
}
EXPORT_SYMBOL_GPL(cpufreq_unregister_driver);
2546

2547 2548 2549 2550 2551 2552 2553 2554
/*
 * 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,
};

2555 2556 2557
struct kobject *cpufreq_global_kobject;
EXPORT_SYMBOL(cpufreq_global_kobject);

2558 2559
static int __init cpufreq_core_init(void)
{
2560 2561 2562
	if (cpufreq_disabled())
		return -ENODEV;

2563
	cpufreq_global_kobject = kobject_create_and_add("cpufreq", &cpu_subsys.dev_root->kobj);
2564 2565
	BUG_ON(!cpufreq_global_kobject);

2566 2567
	register_syscore_ops(&cpufreq_syscore_ops);

2568 2569
	return 0;
}
2570
module_param(off, int, 0444);
2571
core_initcall(cpufreq_core_init);