cpufreq.c 66.3 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 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89

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

static bool suitable_policy(struct cpufreq_policy *policy, bool active)
{
	return active == !policy_is_inactive(policy);
}

/* Finds Next Acive/Inactive policy */
static struct cpufreq_policy *next_policy(struct cpufreq_policy *policy,
					  bool active)
{
	do {
		policy = list_next_entry(policy, policy_list);

		/* No more policies in the list */
		if (&policy->policy_list == &cpufreq_policy_list)
			return NULL;
	} while (!suitable_policy(policy, active));

	return policy;
}

static struct cpufreq_policy *first_policy(bool active)
{
	struct cpufreq_policy *policy;

	/* No policies in the list */
	if (list_empty(&cpufreq_policy_list))
		return NULL;

	policy = list_first_entry(&cpufreq_policy_list, typeof(*policy),
				  policy_list);

	if (!suitable_policy(policy, active))
		policy = next_policy(policy, active);

	return policy;
}

/* Macros to iterate over CPU policies */
#define for_each_suitable_policy(__policy, __active)	\
	for (__policy = first_policy(__active);		\
	     __policy;					\
	     __policy = next_policy(__policy, __active))

#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
90 91
	list_for_each_entry(__policy, &cpufreq_policy_list, policy_list)

92 93 94 95 96
/* 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
97
/**
D
Dave Jones 已提交
98
 * The "cpufreq driver" - the arch- or hardware-dependent low
L
Linus Torvalds 已提交
99 100 101
 * level driver of CPUFreq support, and its spinlock. This lock
 * also protects the cpufreq_cpu_data array.
 */
102
static struct cpufreq_driver *cpufreq_driver;
103
static DEFINE_PER_CPU(struct cpufreq_policy *, cpufreq_cpu_data);
104
static DEFINE_RWLOCK(cpufreq_driver_lock);
105
DEFINE_MUTEX(cpufreq_governor_lock);
106

107 108
/* Flag to suspend/resume CPUFreq governors */
static bool cpufreq_suspended;
L
Linus Torvalds 已提交
109

110 111 112 113 114
static inline bool has_target(void)
{
	return cpufreq_driver->target_index || cpufreq_driver->target;
}

L
Linus Torvalds 已提交
115
/* internal prototypes */
116 117
static int __cpufreq_governor(struct cpufreq_policy *policy,
		unsigned int event);
118
static unsigned int __cpufreq_get(struct cpufreq_policy *policy);
119
static void handle_update(struct work_struct *work);
L
Linus Torvalds 已提交
120 121

/**
122 123
 * 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
124 125 126 127
 * "transition" list for kernel code that needs to handle
 * changes to devices when the CPU clock speed changes.
 * The mutex locks both lists.
 */
128
static BLOCKING_NOTIFIER_HEAD(cpufreq_policy_notifier_list);
129
static struct srcu_notifier_head cpufreq_transition_notifier_list;
L
Linus Torvalds 已提交
130

131
static bool init_cpufreq_transition_notifier_list_called;
132 133 134
static int __init init_cpufreq_transition_notifier_list(void)
{
	srcu_init_notifier_head(&cpufreq_transition_notifier_list);
135
	init_cpufreq_transition_notifier_list_called = true;
136 137
	return 0;
}
138
pure_initcall(init_cpufreq_transition_notifier_list);
L
Linus Torvalds 已提交
139

140
static int off __read_mostly;
141
static int cpufreq_disabled(void)
142 143 144 145 146 147 148
{
	return off;
}
void disable_cpufreq(void)
{
	off = 1;
}
149
static DEFINE_MUTEX(cpufreq_governor_mutex);
L
Linus Torvalds 已提交
150

151 152
bool have_governor_per_policy(void)
{
153
	return !!(cpufreq_driver->flags & CPUFREQ_HAVE_GOVERNOR_PER_POLICY);
154
}
155
EXPORT_SYMBOL_GPL(have_governor_per_policy);
156

157 158 159 160 161 162 163 164 165
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);

166 167 168 169 170 171 172 173 174
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);

175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209
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);

210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231
/*
 * 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;

	/*
232
	 * The driver only supports the SMP configuration where all processors
233 234 235 236 237 238 239 240
	 * share the clock and voltage and clock.
	 */
	cpumask_setall(policy->cpus);

	return 0;
}
EXPORT_SYMBOL_GPL(cpufreq_generic_init);

241 242
/* Only for cpufreq core internal use */
struct cpufreq_policy *cpufreq_cpu_get_raw(unsigned int cpu)
243 244 245
{
	struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);

246 247 248 249 250 251 252
	return policy && cpumask_test_cpu(cpu, policy->cpus) ? policy : NULL;
}

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

253
	if (!policy || IS_ERR(policy->clk)) {
254 255
		pr_err("%s: No %s associated to cpu: %d\n",
		       __func__, policy ? "clk" : "policy", cpu);
256 257 258 259 260 261 262
		return 0;
	}

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

263 264 265 266 267 268 269 270 271 272 273 274 275
/**
 * 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.
 */
276
struct cpufreq_policy *cpufreq_cpu_get(unsigned int cpu)
L
Linus Torvalds 已提交
277
{
278
	struct cpufreq_policy *policy = NULL;
L
Linus Torvalds 已提交
279 280
	unsigned long flags;

281
	if (WARN_ON(cpu >= nr_cpu_ids))
282 283
		return NULL;

L
Linus Torvalds 已提交
284
	/* get the cpufreq driver */
285
	read_lock_irqsave(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
286

287 288
	if (cpufreq_driver) {
		/* get the CPU */
289
		policy = cpufreq_cpu_get_raw(cpu);
290 291 292
		if (policy)
			kobject_get(&policy->kobj);
	}
L
Linus Torvalds 已提交
293

294
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
295

296
	return policy;
297
}
L
Linus Torvalds 已提交
298 299
EXPORT_SYMBOL_GPL(cpufreq_cpu_get);

300 301 302 303 304 305 306 307
/**
 * 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().
 */
308
void cpufreq_cpu_put(struct cpufreq_policy *policy)
L
Linus Torvalds 已提交
309
{
310
	kobject_put(&policy->kobj);
L
Linus Torvalds 已提交
311 312 313 314 315 316 317 318 319 320 321 322
}
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
323
 * systems as each CPU might be scaled differently. So, use the arch
L
Linus Torvalds 已提交
324 325
 * per-CPU loops_per_jiffy value wherever possible.
 */
326
static void adjust_jiffies(unsigned long val, struct cpufreq_freqs *ci)
L
Linus Torvalds 已提交
327
{
328 329 330 331
#ifndef CONFIG_SMP
	static unsigned long l_p_j_ref;
	static unsigned int l_p_j_ref_freq;

L
Linus Torvalds 已提交
332 333 334 335 336 337
	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;
338 339
		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
340
	}
341
	if (val == CPUFREQ_POSTCHANGE && ci->old != ci->new) {
342 343
		loops_per_jiffy = cpufreq_scale(l_p_j_ref, l_p_j_ref_freq,
								ci->new);
344 345
		pr_debug("scaling loops_per_jiffy to %lu for frequency %u kHz\n",
			 loops_per_jiffy, ci->new);
L
Linus Torvalds 已提交
346 347
	}
#endif
348
}
L
Linus Torvalds 已提交
349

350
static void __cpufreq_notify_transition(struct cpufreq_policy *policy,
351
		struct cpufreq_freqs *freqs, unsigned int state)
L
Linus Torvalds 已提交
352 353 354
{
	BUG_ON(irqs_disabled());

355 356 357
	if (cpufreq_disabled())
		return;

358
	freqs->flags = cpufreq_driver->flags;
359
	pr_debug("notification %u of frequency transition to %u kHz\n",
360
		 state, freqs->new);
L
Linus Torvalds 已提交
361 362

	switch (state) {
363

L
Linus Torvalds 已提交
364
	case CPUFREQ_PRECHANGE:
365
		/* detect if the driver reported a value as "old frequency"
366 367
		 * which is not equal to what the cpufreq core thinks is
		 * "old frequency".
L
Linus Torvalds 已提交
368
		 */
369
		if (!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
370 371
			if ((policy) && (policy->cpu == freqs->cpu) &&
			    (policy->cur) && (policy->cur != freqs->old)) {
372 373
				pr_debug("Warning: CPU frequency is %u, cpufreq assumed %u kHz\n",
					 freqs->old, policy->cur);
374
				freqs->old = policy->cur;
L
Linus Torvalds 已提交
375 376
			}
		}
377
		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
378
				CPUFREQ_PRECHANGE, freqs);
L
Linus Torvalds 已提交
379 380
		adjust_jiffies(CPUFREQ_PRECHANGE, freqs);
		break;
381

L
Linus Torvalds 已提交
382 383
	case CPUFREQ_POSTCHANGE:
		adjust_jiffies(CPUFREQ_POSTCHANGE, freqs);
384 385
		pr_debug("FREQ: %lu - CPU: %lu\n",
			 (unsigned long)freqs->new, (unsigned long)freqs->cpu);
386
		trace_cpu_frequency(freqs->new, freqs->cpu);
387
		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
388
				CPUFREQ_POSTCHANGE, freqs);
389 390
		if (likely(policy) && likely(policy->cpu == freqs->cpu))
			policy->cur = freqs->new;
L
Linus Torvalds 已提交
391 392 393
		break;
	}
}
394

395 396 397 398 399 400 401 402
/**
 * 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.
 */
403
static void cpufreq_notify_transition(struct cpufreq_policy *policy,
404 405 406 407 408
		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
409

410
/* Do post notifications when there are chances that transition has failed */
411
static void cpufreq_notify_post_transition(struct cpufreq_policy *policy,
412 413 414 415 416 417 418 419 420 421 422
		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);
}

423 424 425
void cpufreq_freq_transition_begin(struct cpufreq_policy *policy,
		struct cpufreq_freqs *freqs)
{
426 427 428 429 430 431 432 433 434 435 436 437

	/*
	 * 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);

438 439 440 441 442 443 444 445 446 447 448
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;
449
	policy->transition_task = current;
450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465

	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;
466
	policy->transition_task = NULL;
467 468 469 470 471

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

L
Linus Torvalds 已提交
472 473 474 475

/*********************************************************************
 *                          SYSFS INTERFACE                          *
 *********************************************************************/
476
static ssize_t show_boost(struct kobject *kobj,
477 478 479 480 481 482 483 484 485 486 487 488 489 490 491
				 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)) {
492 493
		pr_err("%s: Cannot %s BOOST!\n",
		       __func__, enable ? "enable" : "disable");
494 495 496
		return -EINVAL;
	}

497 498
	pr_debug("%s: cpufreq BOOST %s\n",
		 __func__, enable ? "enabled" : "disabled");
499 500 501 502

	return count;
}
define_one_global_rw(boost);
L
Linus Torvalds 已提交
503

504
static struct cpufreq_governor *find_governor(const char *str_governor)
505 506 507
{
	struct cpufreq_governor *t;

508
	for_each_governor(t)
509
		if (!strncasecmp(str_governor, t->name, CPUFREQ_NAME_LEN))
510 511 512 513 514
			return t;

	return NULL;
}

L
Linus Torvalds 已提交
515 516 517
/**
 * cpufreq_parse_governor - parse a governor string
 */
D
Dave Jones 已提交
518
static int cpufreq_parse_governor(char *str_governor, unsigned int *policy,
L
Linus Torvalds 已提交
519 520
				struct cpufreq_governor **governor)
{
521
	int err = -EINVAL;
522 523

	if (!cpufreq_driver)
524 525
		goto out;

526
	if (cpufreq_driver->setpolicy) {
527
		if (!strncasecmp(str_governor, "performance", CPUFREQ_NAME_LEN)) {
L
Linus Torvalds 已提交
528
			*policy = CPUFREQ_POLICY_PERFORMANCE;
529
			err = 0;
530
		} else if (!strncasecmp(str_governor, "powersave",
531
						CPUFREQ_NAME_LEN)) {
L
Linus Torvalds 已提交
532
			*policy = CPUFREQ_POLICY_POWERSAVE;
533
			err = 0;
L
Linus Torvalds 已提交
534
		}
535
	} else {
L
Linus Torvalds 已提交
536
		struct cpufreq_governor *t;
537

538
		mutex_lock(&cpufreq_governor_mutex);
539

540
		t = find_governor(str_governor);
541

542
		if (t == NULL) {
543
			int ret;
544

545 546 547
			mutex_unlock(&cpufreq_governor_mutex);
			ret = request_module("cpufreq_%s", str_governor);
			mutex_lock(&cpufreq_governor_mutex);
548

549
			if (ret == 0)
550
				t = find_governor(str_governor);
551 552
		}

553 554 555
		if (t != NULL) {
			*governor = t;
			err = 0;
L
Linus Torvalds 已提交
556
		}
557

558
		mutex_unlock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
559
	}
560
out:
561
	return err;
L
Linus Torvalds 已提交
562 563 564
}

/**
565 566
 * cpufreq_per_cpu_attr_read() / show_##file_name() -
 * print out cpufreq information
L
Linus Torvalds 已提交
567 568 569 570 571
 *
 * Write out information from cpufreq_driver->policy[cpu]; object must be
 * "unsigned int".
 */

572 573
#define show_one(file_name, object)			\
static ssize_t show_##file_name				\
D
Dave Jones 已提交
574
(struct cpufreq_policy *policy, char *buf)		\
575
{							\
576
	return sprintf(buf, "%u\n", policy->object);	\
L
Linus Torvalds 已提交
577 578 579 580
}

show_one(cpuinfo_min_freq, cpuinfo.min_freq);
show_one(cpuinfo_max_freq, cpuinfo.max_freq);
581
show_one(cpuinfo_transition_latency, cpuinfo.transition_latency);
L
Linus Torvalds 已提交
582 583
show_one(scaling_min_freq, min);
show_one(scaling_max_freq, max);
584

585
static ssize_t show_scaling_cur_freq(struct cpufreq_policy *policy, char *buf)
586 587 588 589 590 591 592 593 594
{
	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
595

596
static int cpufreq_set_policy(struct cpufreq_policy *policy,
597
				struct cpufreq_policy *new_policy);
598

L
Linus Torvalds 已提交
599 600 601 602 603
/**
 * 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
604
(struct cpufreq_policy *policy, const char *buf, size_t count)		\
L
Linus Torvalds 已提交
605
{									\
606
	int ret, temp;							\
L
Linus Torvalds 已提交
607 608 609 610 611 612
	struct cpufreq_policy new_policy;				\
									\
	ret = cpufreq_get_policy(&new_policy, policy->cpu);		\
	if (ret)							\
		return -EINVAL;						\
									\
613
	ret = sscanf(buf, "%u", &new_policy.object);			\
L
Linus Torvalds 已提交
614 615 616
	if (ret != 1)							\
		return -EINVAL;						\
									\
617
	temp = new_policy.object;					\
618
	ret = cpufreq_set_policy(policy, &new_policy);		\
619 620
	if (!ret)							\
		policy->user_policy.object = temp;			\
L
Linus Torvalds 已提交
621 622 623 624
									\
	return ret ? ret : count;					\
}

625 626
store_one(scaling_min_freq, min);
store_one(scaling_max_freq, max);
L
Linus Torvalds 已提交
627 628 629 630

/**
 * show_cpuinfo_cur_freq - current CPU frequency as detected by hardware
 */
D
Dave Jones 已提交
631 632
static ssize_t show_cpuinfo_cur_freq(struct cpufreq_policy *policy,
					char *buf)
L
Linus Torvalds 已提交
633
{
634
	unsigned int cur_freq = __cpufreq_get(policy);
L
Linus Torvalds 已提交
635 636 637 638 639 640 641 642
	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
643
static ssize_t show_scaling_governor(struct cpufreq_policy *policy, char *buf)
L
Linus Torvalds 已提交
644
{
645
	if (policy->policy == CPUFREQ_POLICY_POWERSAVE)
L
Linus Torvalds 已提交
646 647 648 649
		return sprintf(buf, "powersave\n");
	else if (policy->policy == CPUFREQ_POLICY_PERFORMANCE)
		return sprintf(buf, "performance\n");
	else if (policy->governor)
650
		return scnprintf(buf, CPUFREQ_NAME_PLEN, "%s\n",
651
				policy->governor->name);
L
Linus Torvalds 已提交
652 653 654 655 656 657
	return -EINVAL;
}

/**
 * store_scaling_governor - store policy for the specified CPU
 */
D
Dave Jones 已提交
658 659
static ssize_t store_scaling_governor(struct cpufreq_policy *policy,
					const char *buf, size_t count)
L
Linus Torvalds 已提交
660
{
661
	int ret;
L
Linus Torvalds 已提交
662 663 664 665 666 667 668
	char	str_governor[16];
	struct cpufreq_policy new_policy;

	ret = cpufreq_get_policy(&new_policy, policy->cpu);
	if (ret)
		return ret;

669
	ret = sscanf(buf, "%15s", str_governor);
L
Linus Torvalds 已提交
670 671 672
	if (ret != 1)
		return -EINVAL;

673 674
	if (cpufreq_parse_governor(str_governor, &new_policy.policy,
						&new_policy.governor))
L
Linus Torvalds 已提交
675 676
		return -EINVAL;

677
	ret = cpufreq_set_policy(policy, &new_policy);
678 679 680 681

	policy->user_policy.policy = policy->policy;
	policy->user_policy.governor = policy->governor;

682 683 684 685
	if (ret)
		return ret;
	else
		return count;
L
Linus Torvalds 已提交
686 687 688 689 690
}

/**
 * show_scaling_driver - show the cpufreq driver currently loaded
 */
D
Dave Jones 已提交
691
static ssize_t show_scaling_driver(struct cpufreq_policy *policy, char *buf)
L
Linus Torvalds 已提交
692
{
693
	return scnprintf(buf, CPUFREQ_NAME_PLEN, "%s\n", cpufreq_driver->name);
L
Linus Torvalds 已提交
694 695 696 697 698
}

/**
 * show_scaling_available_governors - show the available CPUfreq governors
 */
D
Dave Jones 已提交
699 700
static ssize_t show_scaling_available_governors(struct cpufreq_policy *policy,
						char *buf)
L
Linus Torvalds 已提交
701 702 703 704
{
	ssize_t i = 0;
	struct cpufreq_governor *t;

705
	if (!has_target()) {
L
Linus Torvalds 已提交
706 707 708 709
		i += sprintf(buf, "performance powersave");
		goto out;
	}

710
	for_each_governor(t) {
711 712
		if (i >= (ssize_t) ((PAGE_SIZE / sizeof(char))
		    - (CPUFREQ_NAME_LEN + 2)))
L
Linus Torvalds 已提交
713
			goto out;
714
		i += scnprintf(&buf[i], CPUFREQ_NAME_PLEN, "%s ", t->name);
L
Linus Torvalds 已提交
715
	}
716
out:
L
Linus Torvalds 已提交
717 718 719
	i += sprintf(&buf[i], "\n");
	return i;
}
720

721
ssize_t cpufreq_show_cpus(const struct cpumask *mask, char *buf)
L
Linus Torvalds 已提交
722 723 724 725
{
	ssize_t i = 0;
	unsigned int cpu;

726
	for_each_cpu(cpu, mask) {
L
Linus Torvalds 已提交
727 728 729 730
		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))
731
			break;
L
Linus Torvalds 已提交
732 733 734 735
	}
	i += sprintf(&buf[i], "\n");
	return i;
}
736
EXPORT_SYMBOL_GPL(cpufreq_show_cpus);
L
Linus Torvalds 已提交
737

738 739 740 741 742 743
/**
 * 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)
{
744
	return cpufreq_show_cpus(policy->related_cpus, buf);
745 746 747 748 749 750 751
}

/**
 * show_affected_cpus - show the CPUs affected by each transition
 */
static ssize_t show_affected_cpus(struct cpufreq_policy *policy, char *buf)
{
752
	return cpufreq_show_cpus(policy->cpus, buf);
753 754
}

755
static ssize_t store_scaling_setspeed(struct cpufreq_policy *policy,
D
Dave Jones 已提交
756
					const char *buf, size_t count)
757 758 759 760
{
	unsigned int freq = 0;
	unsigned int ret;

761
	if (!policy->governor || !policy->governor->store_setspeed)
762 763 764 765 766 767 768 769 770 771 772 773 774
		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)
{
775
	if (!policy->governor || !policy->governor->show_setspeed)
776 777 778 779
		return sprintf(buf, "<unsupported>\n");

	return policy->governor->show_setspeed(policy, buf);
}
L
Linus Torvalds 已提交
780

781
/**
782
 * show_bios_limit - show the current cpufreq HW/BIOS limitation
783 784 785 786 787
 */
static ssize_t show_bios_limit(struct cpufreq_policy *policy, char *buf)
{
	unsigned int limit;
	int ret;
788 789
	if (cpufreq_driver->bios_limit) {
		ret = cpufreq_driver->bios_limit(policy->cpu, &limit);
790 791 792 793 794 795
		if (!ret)
			return sprintf(buf, "%u\n", limit);
	}
	return sprintf(buf, "%u\n", policy->cpuinfo.max_freq);
}

796 797 798 799 800 801 802 803 804 805 806 807 808 809
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
810

D
Dave Jones 已提交
811
static struct attribute *default_attrs[] = {
L
Linus Torvalds 已提交
812 813
	&cpuinfo_min_freq.attr,
	&cpuinfo_max_freq.attr,
814
	&cpuinfo_transition_latency.attr,
L
Linus Torvalds 已提交
815 816 817
	&scaling_min_freq.attr,
	&scaling_max_freq.attr,
	&affected_cpus.attr,
818
	&related_cpus.attr,
L
Linus Torvalds 已提交
819 820 821
	&scaling_governor.attr,
	&scaling_driver.attr,
	&scaling_available_governors.attr,
822
	&scaling_setspeed.attr,
L
Linus Torvalds 已提交
823 824 825
	NULL
};

826 827
#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
828

829
static ssize_t show(struct kobject *kobj, struct attribute *attr, char *buf)
L
Linus Torvalds 已提交
830
{
D
Dave Jones 已提交
831 832
	struct cpufreq_policy *policy = to_policy(kobj);
	struct freq_attr *fattr = to_attr(attr);
833
	ssize_t ret;
834

835
	down_read(&policy->rwsem);
836

837 838 839 840 841
	if (fattr->show)
		ret = fattr->show(policy, buf);
	else
		ret = -EIO;

842
	up_read(&policy->rwsem);
843

L
Linus Torvalds 已提交
844 845 846
	return ret;
}

D
Dave Jones 已提交
847 848
static ssize_t store(struct kobject *kobj, struct attribute *attr,
		     const char *buf, size_t count)
L
Linus Torvalds 已提交
849
{
D
Dave Jones 已提交
850 851
	struct cpufreq_policy *policy = to_policy(kobj);
	struct freq_attr *fattr = to_attr(attr);
852
	ssize_t ret = -EINVAL;
853

854 855 856 857 858
	get_online_cpus();

	if (!cpu_online(policy->cpu))
		goto unlock;

859
	down_write(&policy->rwsem);
860

861 862 863 864 865 866
	/* Updating inactive policies is invalid, so avoid doing that. */
	if (unlikely(policy_is_inactive(policy))) {
		ret = -EBUSY;
		goto unlock_policy_rwsem;
	}

867 868 869 870 871
	if (fattr->store)
		ret = fattr->store(policy, buf, count);
	else
		ret = -EIO;

872
unlock_policy_rwsem:
873
	up_write(&policy->rwsem);
874 875 876
unlock:
	put_online_cpus();

L
Linus Torvalds 已提交
877 878 879
	return ret;
}

D
Dave Jones 已提交
880
static void cpufreq_sysfs_release(struct kobject *kobj)
L
Linus Torvalds 已提交
881
{
D
Dave Jones 已提交
882
	struct cpufreq_policy *policy = to_policy(kobj);
883
	pr_debug("last reference is dropped\n");
L
Linus Torvalds 已提交
884 885 886
	complete(&policy->kobj_unregister);
}

887
static const struct sysfs_ops sysfs_ops = {
L
Linus Torvalds 已提交
888 889 890 891 892 893 894 895 896 897
	.show	= show,
	.store	= store,
};

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

898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940
struct kobject *cpufreq_global_kobject;
EXPORT_SYMBOL(cpufreq_global_kobject);

static int cpufreq_global_kobject_usage;

int cpufreq_get_global_kobject(void)
{
	if (!cpufreq_global_kobject_usage++)
		return kobject_add(cpufreq_global_kobject,
				&cpu_subsys.dev_root->kobj, "%s", "cpufreq");

	return 0;
}
EXPORT_SYMBOL(cpufreq_get_global_kobject);

void cpufreq_put_global_kobject(void)
{
	if (!--cpufreq_global_kobject_usage)
		kobject_del(cpufreq_global_kobject);
}
EXPORT_SYMBOL(cpufreq_put_global_kobject);

int cpufreq_sysfs_create_file(const struct attribute *attr)
{
	int ret = cpufreq_get_global_kobject();

	if (!ret) {
		ret = sysfs_create_file(cpufreq_global_kobject, attr);
		if (ret)
			cpufreq_put_global_kobject();
	}

	return ret;
}
EXPORT_SYMBOL(cpufreq_sysfs_create_file);

void cpufreq_sysfs_remove_file(const struct attribute *attr)
{
	sysfs_remove_file(cpufreq_global_kobject, attr);
	cpufreq_put_global_kobject();
}
EXPORT_SYMBOL(cpufreq_sysfs_remove_file);

941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970
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 */
971
static int cpufreq_add_dev_symlink(struct cpufreq_policy *policy)
972 973 974 975
{
	unsigned int j;
	int ret = 0;

976
	/* Some related CPUs might not be present (physically hotplugged) */
977
	for_each_cpu(j, policy->real_cpus) {
978
		if (j == policy->kobj_cpu)
979 980
			continue;

981
		ret = add_cpu_dev_symlink(policy, j);
982 983
		if (ret)
			break;
984
	}
985

986 987 988
	return ret;
}

989 990 991 992 993
static void cpufreq_remove_dev_symlink(struct cpufreq_policy *policy)
{
	unsigned int j;

	/* Some related CPUs might not be present (physically hotplugged) */
994
	for_each_cpu(j, policy->real_cpus) {
995 996 997 998 999 1000 1001
		if (j == policy->kobj_cpu)
			continue;

		remove_cpu_dev_symlink(policy, j);
	}
}

1002
static int cpufreq_add_dev_interface(struct cpufreq_policy *policy)
1003 1004 1005 1006 1007
{
	struct freq_attr **drv_attr;
	int ret = 0;

	/* set up files for this cpu device */
1008
	drv_attr = cpufreq_driver->attr;
V
Viresh Kumar 已提交
1009
	while (drv_attr && *drv_attr) {
1010 1011
		ret = sysfs_create_file(&policy->kobj, &((*drv_attr)->attr));
		if (ret)
1012
			return ret;
1013 1014
		drv_attr++;
	}
1015
	if (cpufreq_driver->get) {
1016 1017
		ret = sysfs_create_file(&policy->kobj, &cpuinfo_cur_freq.attr);
		if (ret)
1018
			return ret;
1019
	}
1020 1021 1022

	ret = sysfs_create_file(&policy->kobj, &scaling_cur_freq.attr);
	if (ret)
1023
		return ret;
1024

1025
	if (cpufreq_driver->bios_limit) {
1026 1027
		ret = sysfs_create_file(&policy->kobj, &bios_limit.attr);
		if (ret)
1028
			return ret;
1029
	}
1030

1031
	return cpufreq_add_dev_symlink(policy);
1032 1033
}

1034
static int cpufreq_init_policy(struct cpufreq_policy *policy)
1035
{
1036
	struct cpufreq_governor *gov = NULL;
1037 1038
	struct cpufreq_policy new_policy;

1039
	memcpy(&new_policy, policy, sizeof(*policy));
1040

1041
	/* Update governor of new_policy to the governor used before hotplug */
1042
	gov = find_governor(policy->last_governor);
1043 1044 1045 1046 1047 1048 1049 1050
	if (gov)
		pr_debug("Restoring governor %s for cpu %d\n",
				policy->governor->name, policy->cpu);
	else
		gov = CPUFREQ_DEFAULT_GOVERNOR;

	new_policy.governor = gov;

1051 1052
	/* Use the default policy if its valid. */
	if (cpufreq_driver->setpolicy)
1053
		cpufreq_parse_governor(gov->name, &new_policy.policy, NULL);
1054 1055

	/* set default policy */
1056
	return cpufreq_set_policy(policy, &new_policy);
1057 1058
}

1059
static int cpufreq_add_policy_cpu(struct cpufreq_policy *policy, unsigned int cpu)
1060
{
1061
	int ret = 0;
1062

1063 1064 1065 1066
	/* Has this CPU been taken care of already? */
	if (cpumask_test_cpu(cpu, policy->cpus))
		return 0;

1067
	if (has_target()) {
1068 1069 1070 1071 1072 1073
		ret = __cpufreq_governor(policy, CPUFREQ_GOV_STOP);
		if (ret) {
			pr_err("%s: Failed to stop governor\n", __func__);
			return ret;
		}
	}
1074

1075
	down_write(&policy->rwsem);
1076
	cpumask_set_cpu(cpu, policy->cpus);
1077
	up_write(&policy->rwsem);
V
Viresh Kumar 已提交
1078

1079
	if (has_target()) {
1080 1081 1082 1083 1084
		ret = __cpufreq_governor(policy, CPUFREQ_GOV_START);
		if (!ret)
			ret = __cpufreq_governor(policy, CPUFREQ_GOV_LIMITS);

		if (ret) {
1085 1086 1087
			pr_err("%s: Failed to start governor\n", __func__);
			return ret;
		}
1088
	}
1089

1090
	return 0;
1091
}
L
Linus Torvalds 已提交
1092

1093
static struct cpufreq_policy *cpufreq_policy_alloc(unsigned int cpu)
1094
{
1095
	struct device *dev = get_cpu_device(cpu);
1096
	struct cpufreq_policy *policy;
1097
	int ret;
1098

1099 1100 1101
	if (WARN_ON(!dev))
		return NULL;

1102 1103 1104 1105 1106 1107 1108 1109 1110 1111
	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;

1112 1113 1114
	if (!zalloc_cpumask_var(&policy->real_cpus, GFP_KERNEL))
		goto err_free_rcpumask;

1115 1116 1117 1118
	ret = kobject_init_and_add(&policy->kobj, &ktype_cpufreq, &dev->kobj,
				   "cpufreq");
	if (ret) {
		pr_err("%s: failed to init policy->kobj: %d\n", __func__, ret);
1119
		goto err_free_real_cpus;
1120 1121
	}

1122
	INIT_LIST_HEAD(&policy->policy_list);
1123
	init_rwsem(&policy->rwsem);
1124 1125
	spin_lock_init(&policy->transition_lock);
	init_waitqueue_head(&policy->transition_wait);
1126 1127
	init_completion(&policy->kobj_unregister);
	INIT_WORK(&policy->update, handle_update);
1128

1129
	policy->cpu = cpu;
1130 1131

	/* Set this once on allocation */
1132
	policy->kobj_cpu = cpu;
1133

1134 1135
	return policy;

1136 1137
err_free_real_cpus:
	free_cpumask_var(policy->real_cpus);
1138 1139
err_free_rcpumask:
	free_cpumask_var(policy->related_cpus);
1140 1141 1142 1143 1144 1145 1146 1147
err_free_cpumask:
	free_cpumask_var(policy->cpus);
err_free_policy:
	kfree(policy);

	return NULL;
}

1148
static void cpufreq_policy_put_kobj(struct cpufreq_policy *policy, bool notify)
1149 1150 1151 1152
{
	struct kobject *kobj;
	struct completion *cmp;

1153 1154 1155
	if (notify)
		blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
					     CPUFREQ_REMOVE_POLICY, policy);
1156

1157 1158
	down_write(&policy->rwsem);
	cpufreq_remove_dev_symlink(policy);
1159 1160
	kobj = &policy->kobj;
	cmp = &policy->kobj_unregister;
1161
	up_write(&policy->rwsem);
1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173
	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");
}

1174
static void cpufreq_policy_free(struct cpufreq_policy *policy, bool notify)
1175
{
1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186
	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);

1187
	cpufreq_policy_put_kobj(policy, notify);
1188
	free_cpumask_var(policy->real_cpus);
1189 1190 1191 1192 1193
	free_cpumask_var(policy->related_cpus);
	free_cpumask_var(policy->cpus);
	kfree(policy);
}

1194 1195 1196 1197 1198 1199 1200 1201 1202 1203
/**
 * cpufreq_add_dev - add a CPU device
 *
 * Adds the cpufreq interface for a CPU device.
 *
 * The Oracle says: try running cpufreq registration/unregistration concurrently
 * with with cpu hotplugging and all hell will break loose. Tried to clean this
 * mess up, but more thorough testing is needed. - Mathieu
 */
static int cpufreq_add_dev(struct device *dev, struct subsys_interface *sif)
L
Linus Torvalds 已提交
1204
{
1205
	unsigned int j, cpu = dev->id;
1206
	int ret;
1207
	struct cpufreq_policy *policy;
L
Linus Torvalds 已提交
1208
	unsigned long flags;
1209
	bool recover_policy;
1210

1211
	pr_debug("adding CPU %u\n", cpu);
L
Linus Torvalds 已提交
1212

1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223
	if (cpu_is_offline(cpu)) {
		/*
		 * Only possible if we are here from the subsys_interface add
		 * callback.  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.
		 */
		policy = per_cpu(cpufreq_cpu_data, cpu);
		return policy && !cpumask_test_and_set_cpu(cpu, policy->real_cpus)
			? add_cpu_dev_symlink(policy, cpu) : 0;
	}
1224

1225
	/* Check if this CPU already has a policy to manage it */
1226
	policy = per_cpu(cpufreq_cpu_data, cpu);
1227
	if (policy) {
1228
		WARN_ON(!cpumask_test_cpu(cpu, policy->related_cpus));
1229
		if (!policy_is_inactive(policy))
1230
			return cpufreq_add_policy_cpu(policy, cpu);
L
Linus Torvalds 已提交
1231

1232 1233 1234 1235 1236 1237 1238
		/* This is the only online CPU for the policy.  Start over. */
		recover_policy = true;
		down_write(&policy->rwsem);
		policy->cpu = cpu;
		policy->governor = NULL;
		up_write(&policy->rwsem);
	} else {
1239
		recover_policy = false;
1240
		policy = cpufreq_policy_alloc(cpu);
1241
		if (!policy)
1242
			return -ENOMEM;
1243
	}
1244

1245
	cpumask_copy(policy->cpus, cpumask_of(cpu));
L
Linus Torvalds 已提交
1246 1247 1248 1249

	/* call driver. From then on the cpufreq must be able
	 * to accept all calls to ->verify and ->setpolicy for this CPU
	 */
1250
	ret = cpufreq_driver->init(policy);
L
Linus Torvalds 已提交
1251
	if (ret) {
1252
		pr_debug("initialization failed\n");
1253
		goto out_free_policy;
L
Linus Torvalds 已提交
1254
	}
V
Viresh Kumar 已提交
1255

1256 1257
	down_write(&policy->rwsem);

1258 1259 1260 1261
	if (!recover_policy) {
		/* related_cpus should at least include policy->cpus. */
		cpumask_or(policy->related_cpus, policy->related_cpus, policy->cpus);
		/* Remember CPUs present at the policy creation time. */
1262
		cpumask_and(policy->real_cpus, policy->cpus, cpu_present_mask);
1263
	}
1264

1265 1266 1267 1268 1269 1270
	/*
	 * 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);

1271
	if (!recover_policy) {
1272 1273
		policy->user_policy.min = policy->min;
		policy->user_policy.max = policy->max;
1274

1275 1276 1277 1278 1279
		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);
	}
1280

1281
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
1282 1283 1284
		policy->cur = cpufreq_driver->get(policy->cpu);
		if (!policy->cur) {
			pr_err("%s: ->get() failed\n", __func__);
1285
			goto out_exit_policy;
1286 1287 1288
		}
	}

1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328
	/*
	 * 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);
		}
	}

1329 1330 1331
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
				     CPUFREQ_START, policy);

1332
	if (!recover_policy) {
1333
		ret = cpufreq_add_dev_interface(policy);
1334
		if (ret)
1335
			goto out_exit_policy;
1336 1337
		blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
				CPUFREQ_CREATE_POLICY, policy);
1338

1339 1340 1341 1342
		write_lock_irqsave(&cpufreq_driver_lock, flags);
		list_add(&policy->policy_list, &cpufreq_policy_list);
		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
	}
1343

1344 1345 1346 1347 1348 1349
	ret = cpufreq_init_policy(policy);
	if (ret) {
		pr_err("%s: Failed to initialize policy for cpu: %d (%d)\n",
		       __func__, cpu, ret);
		goto out_remove_policy_notify;
	}
1350

1351
	if (!recover_policy) {
1352 1353 1354
		policy->user_policy.policy = policy->policy;
		policy->user_policy.governor = policy->governor;
	}
1355
	up_write(&policy->rwsem);
1356

1357
	kobject_uevent(&policy->kobj, KOBJ_ADD);
1358 1359 1360 1361 1362

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

1363
	pr_debug("initialization complete\n");
1364

L
Linus Torvalds 已提交
1365 1366
	return 0;

1367 1368 1369
out_remove_policy_notify:
	/* cpufreq_policy_free() will notify based on this */
	recover_policy = true;
1370
out_exit_policy:
1371 1372
	up_write(&policy->rwsem);

1373 1374
	if (cpufreq_driver->exit)
		cpufreq_driver->exit(policy);
1375
out_free_policy:
1376
	cpufreq_policy_free(policy, recover_policy);
L
Linus Torvalds 已提交
1377 1378 1379
	return ret;
}

1380
static void cpufreq_offline_prepare(unsigned int cpu)
L
Linus Torvalds 已提交
1381
{
1382
	struct cpufreq_policy *policy;
L
Linus Torvalds 已提交
1383

1384
	pr_debug("%s: unregistering CPU %u\n", __func__, cpu);
L
Linus Torvalds 已提交
1385

1386
	policy = cpufreq_cpu_get_raw(cpu);
1387
	if (!policy) {
1388
		pr_debug("%s: No cpu_data found\n", __func__);
1389
		return;
L
Linus Torvalds 已提交
1390 1391
	}

1392
	if (has_target()) {
1393
		int ret = __cpufreq_governor(policy, CPUFREQ_GOV_STOP);
1394
		if (ret)
1395
			pr_err("%s: Failed to stop governor\n", __func__);
1396
	}
L
Linus Torvalds 已提交
1397

1398
	down_write(&policy->rwsem);
1399
	cpumask_clear_cpu(cpu, policy->cpus);
1400

1401 1402 1403 1404 1405 1406 1407 1408
	if (policy_is_inactive(policy)) {
		if (has_target())
			strncpy(policy->last_governor, policy->governor->name,
				CPUFREQ_NAME_LEN);
	} else if (cpu == policy->cpu) {
		/* Nominate new CPU */
		policy->cpu = cpumask_any(policy->cpus);
	}
1409
	up_write(&policy->rwsem);
1410

1411 1412 1413
	/* Start governor again for active policy */
	if (!policy_is_inactive(policy)) {
		if (has_target()) {
1414
			int ret = __cpufreq_governor(policy, CPUFREQ_GOV_START);
1415 1416
			if (!ret)
				ret = __cpufreq_governor(policy, CPUFREQ_GOV_LIMITS);
1417

1418 1419 1420 1421
			if (ret)
				pr_err("%s: Failed to start governor\n", __func__);
		}
	} else if (cpufreq_driver->stop_cpu) {
1422
		cpufreq_driver->stop_cpu(policy);
1423
	}
1424 1425
}

1426
static void cpufreq_offline_finish(unsigned int cpu)
1427
{
1428
	struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);
1429 1430 1431

	if (!policy) {
		pr_debug("%s: No cpu_data found\n", __func__);
1432
		return;
1433 1434
	}

1435 1436
	/* Only proceed for inactive policies */
	if (!policy_is_inactive(policy))
1437
		return;
1438 1439 1440

	/* If cpu is last user of policy, free policy */
	if (has_target()) {
1441
		int ret = __cpufreq_governor(policy, CPUFREQ_GOV_POLICY_EXIT);
1442
		if (ret)
1443
			pr_err("%s: Failed to exit governor\n", __func__);
1444
	}
L
Linus Torvalds 已提交
1445

1446 1447 1448 1449 1450 1451 1452
	/*
	 * 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.
	 */
	if (cpufreq_driver->exit)
		cpufreq_driver->exit(policy);
L
Linus Torvalds 已提交
1453 1454
}

1455
/**
1456
 * cpufreq_remove_dev - remove a CPU device
1457 1458 1459
 *
 * Removes the cpufreq interface for a CPU device.
 */
1460
static int cpufreq_remove_dev(struct device *dev, struct subsys_interface *sif)
1461
{
1462
	unsigned int cpu = dev->id;
1463
	struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);
1464

1465 1466
	if (!policy)
		return 0;
1467

1468
	if (cpu_online(cpu)) {
1469 1470
		cpufreq_offline_prepare(cpu);
		cpufreq_offline_finish(cpu);
1471
	}
1472

1473
	cpumask_clear_cpu(cpu, policy->real_cpus);
1474

1475
	if (cpumask_empty(policy->real_cpus)) {
1476
		cpufreq_policy_free(policy, true);
1477
		return 0;
1478
	}
1479

1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490
	if (cpu != policy->kobj_cpu) {
		remove_cpu_dev_symlink(policy, cpu);
	} else {
		/*
		 * The CPU owning the policy object is going away.  Move it to
		 * another suitable CPU.
		 */
		unsigned int new_cpu = cpumask_first(policy->real_cpus);
		struct device *new_dev = get_cpu_device(new_cpu);

		dev_dbg(dev, "%s: Moving policy object to CPU%u\n", __func__, new_cpu);
1491

1492 1493 1494 1495
		sysfs_remove_link(&new_dev->kobj, "cpufreq");
		policy->kobj_cpu = new_cpu;
		WARN_ON(kobject_move(&policy->kobj, &new_dev->kobj));
	}
1496

1497
	return 0;
1498 1499
}

1500
static void handle_update(struct work_struct *work)
L
Linus Torvalds 已提交
1501
{
1502 1503 1504
	struct cpufreq_policy *policy =
		container_of(work, struct cpufreq_policy, update);
	unsigned int cpu = policy->cpu;
1505
	pr_debug("handle_update for cpu %u called\n", cpu);
L
Linus Torvalds 已提交
1506 1507 1508 1509
	cpufreq_update_policy(cpu);
}

/**
1510 1511
 *	cpufreq_out_of_sync - If actual and saved CPU frequency differs, we're
 *	in deep trouble.
1512
 *	@policy: policy managing CPUs
L
Linus Torvalds 已提交
1513 1514
 *	@new_freq: CPU frequency the CPU actually runs at
 *
1515 1516
 *	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
1517
 */
1518
static void cpufreq_out_of_sync(struct cpufreq_policy *policy,
1519
				unsigned int new_freq)
L
Linus Torvalds 已提交
1520 1521
{
	struct cpufreq_freqs freqs;
1522

1523
	pr_debug("Warning: CPU frequency out of sync: cpufreq and timing core thinks of %u, is %u kHz\n",
1524
		 policy->cur, new_freq);
L
Linus Torvalds 已提交
1525

1526
	freqs.old = policy->cur;
L
Linus Torvalds 已提交
1527
	freqs.new = new_freq;
1528

1529 1530
	cpufreq_freq_transition_begin(policy, &freqs);
	cpufreq_freq_transition_end(policy, &freqs, 0);
L
Linus Torvalds 已提交
1531 1532
}

1533
/**
D
Dhaval Giani 已提交
1534
 * cpufreq_quick_get - get the CPU frequency (in kHz) from policy->cur
1535 1536 1537 1538 1539 1540 1541
 * @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)
{
1542
	struct cpufreq_policy *policy;
1543
	unsigned int ret_freq = 0;
1544

1545 1546
	if (cpufreq_driver && cpufreq_driver->setpolicy && cpufreq_driver->get)
		return cpufreq_driver->get(cpu);
1547 1548

	policy = cpufreq_cpu_get(cpu);
1549
	if (policy) {
1550
		ret_freq = policy->cur;
1551 1552 1553
		cpufreq_cpu_put(policy);
	}

D
Dave Jones 已提交
1554
	return ret_freq;
1555 1556 1557
}
EXPORT_SYMBOL(cpufreq_quick_get);

1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577
/**
 * 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);

1578
static unsigned int __cpufreq_get(struct cpufreq_policy *policy)
L
Linus Torvalds 已提交
1579
{
1580
	unsigned int ret_freq = 0;
1581

1582
	if (!cpufreq_driver->get)
D
Dave Jones 已提交
1583
		return ret_freq;
L
Linus Torvalds 已提交
1584

1585
	ret_freq = cpufreq_driver->get(policy->cpu);
L
Linus Torvalds 已提交
1586

1587 1588 1589 1590
	/* Updating inactive policies is invalid, so avoid doing that. */
	if (unlikely(policy_is_inactive(policy)))
		return ret_freq;

1591
	if (ret_freq && policy->cur &&
1592
		!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
1593 1594 1595
		/* verify no discrepancy between actual and
					saved value exists */
		if (unlikely(ret_freq != policy->cur)) {
1596
			cpufreq_out_of_sync(policy, ret_freq);
L
Linus Torvalds 已提交
1597 1598 1599 1600
			schedule_work(&policy->update);
		}
	}

D
Dave Jones 已提交
1601
	return ret_freq;
1602
}
L
Linus Torvalds 已提交
1603

1604 1605 1606 1607 1608 1609 1610 1611
/**
 * 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)
{
1612
	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
1613 1614
	unsigned int ret_freq = 0;

1615 1616
	if (policy) {
		down_read(&policy->rwsem);
1617
		ret_freq = __cpufreq_get(policy);
1618
		up_read(&policy->rwsem);
1619

1620 1621
		cpufreq_cpu_put(policy);
	}
1622

D
Dave Jones 已提交
1623
	return ret_freq;
L
Linus Torvalds 已提交
1624 1625 1626
}
EXPORT_SYMBOL(cpufreq_get);

1627 1628 1629 1630 1631
static struct subsys_interface cpufreq_interface = {
	.name		= "cpufreq",
	.subsys		= &cpu_subsys,
	.add_dev	= cpufreq_add_dev,
	.remove_dev	= cpufreq_remove_dev,
1632 1633
};

1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659
/*
 * 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) {
		pr_err("%s: suspend_freq can't be zero\n", __func__);
		return -EINVAL;
	}

	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);

1660
/**
1661
 * cpufreq_suspend() - Suspend CPUFreq governors
1662
 *
1663 1664 1665 1666
 * 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.
1667
 */
1668
void cpufreq_suspend(void)
1669
{
1670
	struct cpufreq_policy *policy;
1671

1672 1673
	if (!cpufreq_driver)
		return;
1674

1675
	if (!has_target())
1676
		goto suspend;
1677

1678 1679
	pr_debug("%s: Suspending Governors\n", __func__);

1680
	for_each_active_policy(policy) {
1681 1682 1683 1684 1685 1686 1687
		if (__cpufreq_governor(policy, CPUFREQ_GOV_STOP))
			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);
1688
	}
1689 1690 1691

suspend:
	cpufreq_suspended = true;
1692 1693
}

L
Linus Torvalds 已提交
1694
/**
1695
 * cpufreq_resume() - Resume CPUFreq governors
L
Linus Torvalds 已提交
1696
 *
1697 1698
 * Called during system wide Suspend/Hibernate cycle for resuming governors that
 * are suspended with cpufreq_suspend().
L
Linus Torvalds 已提交
1699
 */
1700
void cpufreq_resume(void)
L
Linus Torvalds 已提交
1701
{
1702
	struct cpufreq_policy *policy;
L
Linus Torvalds 已提交
1703

1704 1705
	if (!cpufreq_driver)
		return;
L
Linus Torvalds 已提交
1706

1707 1708
	cpufreq_suspended = false;

1709
	if (!has_target())
1710
		return;
L
Linus Torvalds 已提交
1711

1712
	pr_debug("%s: Resuming Governors\n", __func__);
L
Linus Torvalds 已提交
1713

1714
	for_each_active_policy(policy) {
1715 1716 1717 1718
		if (cpufreq_driver->resume && cpufreq_driver->resume(policy))
			pr_err("%s: Failed to resume driver: %p\n", __func__,
				policy);
		else if (__cpufreq_governor(policy, CPUFREQ_GOV_START)
1719 1720 1721 1722
		    || __cpufreq_governor(policy, CPUFREQ_GOV_LIMITS))
			pr_err("%s: Failed to start governor for policy: %p\n",
				__func__, policy);
	}
1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733

	/*
	 * 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);
1734
}
L
Linus Torvalds 已提交
1735

1736 1737 1738 1739 1740 1741 1742 1743
/**
 *	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)
{
1744 1745 1746 1747
	if (cpufreq_driver)
		return cpufreq_driver->name;

	return NULL;
1748 1749
}
EXPORT_SYMBOL_GPL(cpufreq_get_current_driver);
L
Linus Torvalds 已提交
1750

1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765
/**
 *	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
1766 1767 1768 1769 1770 1771 1772 1773 1774
/*********************************************************************
 *                     NOTIFIER LISTS INTERFACE                      *
 *********************************************************************/

/**
 *	cpufreq_register_notifier - register a driver with cpufreq
 *	@nb: notifier function to register
 *      @list: CPUFREQ_TRANSITION_NOTIFIER or CPUFREQ_POLICY_NOTIFIER
 *
1775
 *	Add a driver to one of two lists: either a list of drivers that
L
Linus Torvalds 已提交
1776 1777 1778 1779 1780
 *      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
1781
 *	blocking_notifier_chain_register.
L
Linus Torvalds 已提交
1782 1783 1784 1785 1786
 */
int cpufreq_register_notifier(struct notifier_block *nb, unsigned int list)
{
	int ret;

1787 1788 1789
	if (cpufreq_disabled())
		return -EINVAL;

1790 1791
	WARN_ON(!init_cpufreq_transition_notifier_list_called);

L
Linus Torvalds 已提交
1792 1793
	switch (list) {
	case CPUFREQ_TRANSITION_NOTIFIER:
1794
		ret = srcu_notifier_chain_register(
1795
				&cpufreq_transition_notifier_list, nb);
L
Linus Torvalds 已提交
1796 1797
		break;
	case CPUFREQ_POLICY_NOTIFIER:
1798 1799
		ret = blocking_notifier_chain_register(
				&cpufreq_policy_notifier_list, nb);
L
Linus Torvalds 已提交
1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811
		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
1812
 *	@list: CPUFREQ_TRANSITION_NOTIFIER or CPUFREQ_POLICY_NOTIFIER
L
Linus Torvalds 已提交
1813 1814 1815 1816
 *
 *	Remove a driver from the CPU frequency notifier list.
 *
 *	This function may sleep, and has the same return conditions as
1817
 *	blocking_notifier_chain_unregister.
L
Linus Torvalds 已提交
1818 1819 1820 1821 1822
 */
int cpufreq_unregister_notifier(struct notifier_block *nb, unsigned int list)
{
	int ret;

1823 1824 1825
	if (cpufreq_disabled())
		return -EINVAL;

L
Linus Torvalds 已提交
1826 1827
	switch (list) {
	case CPUFREQ_TRANSITION_NOTIFIER:
1828
		ret = srcu_notifier_chain_unregister(
1829
				&cpufreq_transition_notifier_list, nb);
L
Linus Torvalds 已提交
1830 1831
		break;
	case CPUFREQ_POLICY_NOTIFIER:
1832 1833
		ret = blocking_notifier_chain_unregister(
				&cpufreq_policy_notifier_list, nb);
L
Linus Torvalds 已提交
1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847
		break;
	default:
		ret = -EINVAL;
	}

	return ret;
}
EXPORT_SYMBOL(cpufreq_unregister_notifier);


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

1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873
/* 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;
}

1874 1875 1876
static int __target_index(struct cpufreq_policy *policy,
			  struct cpufreq_frequency_table *freq_table, int index)
{
1877 1878
	struct cpufreq_freqs freqs = {.old = policy->cur, .flags = 0};
	unsigned int intermediate_freq = 0;
1879 1880 1881 1882 1883
	int retval = -EINVAL;
	bool notify;

	notify = !(cpufreq_driver->flags & CPUFREQ_ASYNC_NOTIFICATION);
	if (notify) {
1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894
		/* 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;
		}
1895

1896
		freqs.new = freq_table[index].frequency;
1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907
		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);

1908
	if (notify) {
1909 1910
		cpufreq_freq_transition_end(policy, &freqs, retval);

1911 1912 1913 1914
		/*
		 * 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
1915
		 * case we haven't switched to intermediate freq at all.
1916 1917 1918 1919 1920 1921 1922 1923 1924
		 */
		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);
		}
	}

1925 1926 1927
	return retval;
}

L
Linus Torvalds 已提交
1928 1929 1930 1931
int __cpufreq_driver_target(struct cpufreq_policy *policy,
			    unsigned int target_freq,
			    unsigned int relation)
{
1932
	unsigned int old_target_freq = target_freq;
1933
	int retval = -EINVAL;
1934

1935 1936 1937
	if (cpufreq_disabled())
		return -ENODEV;

1938 1939 1940 1941 1942 1943 1944
	/* 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",
1945
		 policy->cpu, target_freq, relation, old_target_freq);
1946

1947 1948 1949 1950 1951 1952
	/*
	 * 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.
	 */
1953 1954 1955
	if (target_freq == policy->cur)
		return 0;

1956 1957 1958
	/* Save last value to restore later on errors */
	policy->restore_freq = policy->cur;

1959 1960
	if (cpufreq_driver->target)
		retval = cpufreq_driver->target(policy, target_freq, relation);
1961 1962 1963
	else if (cpufreq_driver->target_index) {
		struct cpufreq_frequency_table *freq_table;
		int index;
1964

1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977
		freq_table = cpufreq_frequency_get_table(policy->cpu);
		if (unlikely(!freq_table)) {
			pr_err("%s: Unable to find freq_table\n", __func__);
			goto out;
		}

		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__);
			goto out;
		}

1978
		if (freq_table[index].frequency == policy->cur) {
1979
			retval = 0;
1980 1981 1982
			goto out;
		}

1983
		retval = __target_index(policy, freq_table, index);
1984 1985 1986
	}

out:
L
Linus Torvalds 已提交
1987 1988 1989 1990 1991 1992 1993 1994
	return retval;
}
EXPORT_SYMBOL_GPL(__cpufreq_driver_target);

int cpufreq_driver_target(struct cpufreq_policy *policy,
			  unsigned int target_freq,
			  unsigned int relation)
{
1995
	int ret = -EINVAL;
L
Linus Torvalds 已提交
1996

1997
	down_write(&policy->rwsem);
L
Linus Torvalds 已提交
1998 1999 2000

	ret = __cpufreq_driver_target(policy, target_freq, relation);

2001
	up_write(&policy->rwsem);
L
Linus Torvalds 已提交
2002 2003 2004 2005 2006

	return ret;
}
EXPORT_SYMBOL_GPL(cpufreq_driver_target);

2007 2008
static int __cpufreq_governor(struct cpufreq_policy *policy,
					unsigned int event)
L
Linus Torvalds 已提交
2009
{
2010
	int ret;
2011 2012 2013 2014 2015 2016 2017 2018 2019 2020

	/* Only must be defined when default governor is known to have latency
	   restrictions, like e.g. conservative or ondemand.
	   That this is the case is already ensured in Kconfig
	*/
#ifdef CONFIG_CPU_FREQ_GOV_PERFORMANCE
	struct cpufreq_governor *gov = &cpufreq_gov_performance;
#else
	struct cpufreq_governor *gov = NULL;
#endif
2021

2022 2023 2024
	/* Don't start any governor operations if we are entering suspend */
	if (cpufreq_suspended)
		return 0;
2025 2026 2027 2028 2029 2030
	/*
	 * Governor might not be initiated here if ACPI _PPC changed
	 * notification happened, so check it.
	 */
	if (!policy->governor)
		return -EINVAL;
2031

2032 2033 2034
	if (policy->governor->max_transition_latency &&
	    policy->cpuinfo.transition_latency >
	    policy->governor->max_transition_latency) {
2035 2036 2037
		if (!gov)
			return -EINVAL;
		else {
2038 2039
			pr_warn("%s governor failed, too long transition latency of HW, fallback to %s governor\n",
				policy->governor->name, gov->name);
2040 2041
			policy->governor = gov;
		}
2042
	}
L
Linus Torvalds 已提交
2043

2044 2045 2046
	if (event == CPUFREQ_GOV_POLICY_INIT)
		if (!try_module_get(policy->governor->owner))
			return -EINVAL;
L
Linus Torvalds 已提交
2047

2048
	pr_debug("__cpufreq_governor for CPU %u, event %u\n",
2049
		 policy->cpu, event);
2050 2051

	mutex_lock(&cpufreq_governor_lock);
2052
	if ((policy->governor_enabled && event == CPUFREQ_GOV_START)
2053 2054
	    || (!policy->governor_enabled
	    && (event == CPUFREQ_GOV_LIMITS || event == CPUFREQ_GOV_STOP))) {
2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065
		mutex_unlock(&cpufreq_governor_lock);
		return -EBUSY;
	}

	if (event == CPUFREQ_GOV_STOP)
		policy->governor_enabled = false;
	else if (event == CPUFREQ_GOV_START)
		policy->governor_enabled = true;

	mutex_unlock(&cpufreq_governor_lock);

L
Linus Torvalds 已提交
2066 2067
	ret = policy->governor->governor(policy, event);

2068 2069 2070 2071 2072
	if (!ret) {
		if (event == CPUFREQ_GOV_POLICY_INIT)
			policy->governor->initialized++;
		else if (event == CPUFREQ_GOV_POLICY_EXIT)
			policy->governor->initialized--;
2073 2074 2075 2076 2077 2078 2079 2080
	} else {
		/* Restore original values */
		mutex_lock(&cpufreq_governor_lock);
		if (event == CPUFREQ_GOV_STOP)
			policy->governor_enabled = true;
		else if (event == CPUFREQ_GOV_START)
			policy->governor_enabled = false;
		mutex_unlock(&cpufreq_governor_lock);
2081
	}
2082

2083 2084
	if (((event == CPUFREQ_GOV_POLICY_INIT) && ret) ||
			((event == CPUFREQ_GOV_POLICY_EXIT) && !ret))
L
Linus Torvalds 已提交
2085 2086 2087 2088 2089 2090 2091
		module_put(policy->governor->owner);

	return ret;
}

int cpufreq_register_governor(struct cpufreq_governor *governor)
{
2092
	int err;
L
Linus Torvalds 已提交
2093 2094 2095 2096

	if (!governor)
		return -EINVAL;

2097 2098 2099
	if (cpufreq_disabled())
		return -ENODEV;

2100
	mutex_lock(&cpufreq_governor_mutex);
2101

2102
	governor->initialized = 0;
2103
	err = -EBUSY;
2104
	if (!find_governor(governor->name)) {
2105 2106
		err = 0;
		list_add(&governor->governor_list, &cpufreq_governor_list);
L
Linus Torvalds 已提交
2107 2108
	}

2109
	mutex_unlock(&cpufreq_governor_mutex);
2110
	return err;
L
Linus Torvalds 已提交
2111 2112 2113 2114 2115
}
EXPORT_SYMBOL_GPL(cpufreq_register_governor);

void cpufreq_unregister_governor(struct cpufreq_governor *governor)
{
2116 2117
	struct cpufreq_policy *policy;
	unsigned long flags;
2118

L
Linus Torvalds 已提交
2119 2120 2121
	if (!governor)
		return;

2122 2123 2124
	if (cpufreq_disabled())
		return;

2125 2126 2127
	/* clear last_governor for all inactive policies */
	read_lock_irqsave(&cpufreq_driver_lock, flags);
	for_each_inactive_policy(policy) {
2128 2129
		if (!strcmp(policy->last_governor, governor->name)) {
			policy->governor = NULL;
2130
			strcpy(policy->last_governor, "\0");
2131
		}
2132
	}
2133
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
2134

2135
	mutex_lock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
2136
	list_del(&governor->governor_list);
2137
	mutex_unlock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148
	return;
}
EXPORT_SYMBOL_GPL(cpufreq_unregister_governor);


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

/**
 * cpufreq_get_policy - get the current cpufreq_policy
2149 2150
 * @policy: struct cpufreq_policy into which the current cpufreq_policy
 *	is written
L
Linus Torvalds 已提交
2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163
 *
 * 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;

2164
	memcpy(policy, cpu_policy, sizeof(*policy));
L
Linus Torvalds 已提交
2165 2166 2167 2168 2169 2170

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

2171
/*
2172 2173
 * policy : current policy.
 * new_policy: policy to be set.
2174
 */
2175
static int cpufreq_set_policy(struct cpufreq_policy *policy,
2176
				struct cpufreq_policy *new_policy)
L
Linus Torvalds 已提交
2177
{
2178 2179
	struct cpufreq_governor *old_gov;
	int ret;
L
Linus Torvalds 已提交
2180

2181 2182
	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
2183

2184
	memcpy(&new_policy->cpuinfo, &policy->cpuinfo, sizeof(policy->cpuinfo));
L
Linus Torvalds 已提交
2185

2186 2187
	if (new_policy->min > policy->max || new_policy->max < policy->min)
		return -EINVAL;
2188

L
Linus Torvalds 已提交
2189
	/* verify the cpu speed can be set within this limit */
2190
	ret = cpufreq_driver->verify(new_policy);
L
Linus Torvalds 已提交
2191
	if (ret)
2192
		return ret;
L
Linus Torvalds 已提交
2193 2194

	/* adjust if necessary - all reasons */
2195
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2196
			CPUFREQ_ADJUST, new_policy);
L
Linus Torvalds 已提交
2197 2198

	/* adjust if necessary - hardware incompatibility*/
2199
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2200
			CPUFREQ_INCOMPATIBLE, new_policy);
L
Linus Torvalds 已提交
2201

2202 2203 2204 2205
	/*
	 * verify the cpu speed can be set within this limit, which might be
	 * different to the first one
	 */
2206
	ret = cpufreq_driver->verify(new_policy);
2207
	if (ret)
2208
		return ret;
L
Linus Torvalds 已提交
2209 2210

	/* notification of the new policy */
2211
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2212
			CPUFREQ_NOTIFY, new_policy);
L
Linus Torvalds 已提交
2213

2214 2215
	policy->min = new_policy->min;
	policy->max = new_policy->max;
L
Linus Torvalds 已提交
2216

2217
	pr_debug("new min and max freqs are %u - %u kHz\n",
2218
		 policy->min, policy->max);
L
Linus Torvalds 已提交
2219

2220
	if (cpufreq_driver->setpolicy) {
2221
		policy->policy = new_policy->policy;
2222
		pr_debug("setting range\n");
2223 2224
		return cpufreq_driver->setpolicy(new_policy);
	}
L
Linus Torvalds 已提交
2225

2226 2227
	if (new_policy->governor == policy->governor)
		goto out;
2228

2229 2230 2231 2232 2233 2234
	pr_debug("governor switch\n");

	/* save old, working values */
	old_gov = policy->governor;
	/* end old governor */
	if (old_gov) {
2235 2236 2237 2238 2239 2240 2241 2242
		ret = __cpufreq_governor(policy, CPUFREQ_GOV_STOP);
		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;
		}

2243
		up_write(&policy->rwsem);
2244
		ret = __cpufreq_governor(policy, CPUFREQ_GOV_POLICY_EXIT);
2245
		down_write(&policy->rwsem);
2246 2247 2248 2249 2250 2251

		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
2252 2253
	}

2254 2255
	/* start new governor */
	policy->governor = new_policy->governor;
2256 2257 2258 2259
	ret = __cpufreq_governor(policy, CPUFREQ_GOV_POLICY_INIT);
	if (!ret) {
		ret = __cpufreq_governor(policy, CPUFREQ_GOV_START);
		if (!ret)
2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270
			goto out;

		up_write(&policy->rwsem);
		__cpufreq_governor(policy, CPUFREQ_GOV_POLICY_EXIT);
		down_write(&policy->rwsem);
	}

	/* 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;
2271 2272 2273 2274
		if (__cpufreq_governor(policy, CPUFREQ_GOV_POLICY_INIT))
			policy->governor = NULL;
		else
			__cpufreq_governor(policy, CPUFREQ_GOV_START);
2275 2276
	}

2277
	return ret;
2278 2279 2280 2281

 out:
	pr_debug("governor: change or update limits\n");
	return __cpufreq_governor(policy, CPUFREQ_GOV_LIMITS);
L
Linus Torvalds 已提交
2282 2283 2284 2285 2286 2287
}

/**
 *	cpufreq_update_policy - re-evaluate an existing cpufreq policy
 *	@cpu: CPU which shall be re-evaluated
 *
L
Lucas De Marchi 已提交
2288
 *	Useful for policy notifiers which have different necessities
L
Linus Torvalds 已提交
2289 2290 2291 2292
 *	at different times.
 */
int cpufreq_update_policy(unsigned int cpu)
{
2293 2294
	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
	struct cpufreq_policy new_policy;
2295
	int ret;
L
Linus Torvalds 已提交
2296

2297 2298
	if (!policy)
		return -ENODEV;
L
Linus Torvalds 已提交
2299

2300
	down_write(&policy->rwsem);
L
Linus Torvalds 已提交
2301

2302
	pr_debug("updating policy for CPU %u\n", cpu);
2303
	memcpy(&new_policy, policy, sizeof(*policy));
2304 2305 2306 2307
	new_policy.min = policy->user_policy.min;
	new_policy.max = policy->user_policy.max;
	new_policy.policy = policy->user_policy.policy;
	new_policy.governor = policy->user_policy.governor;
L
Linus Torvalds 已提交
2308

2309 2310 2311 2312
	/*
	 * BIOS might change freq behind our back
	 * -> ask driver for current freq and notify governors about a change
	 */
2313
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
2314
		new_policy.cur = cpufreq_driver->get(cpu);
2315 2316
		if (WARN_ON(!new_policy.cur)) {
			ret = -EIO;
2317
			goto unlock;
2318 2319
		}

2320
		if (!policy->cur) {
2321
			pr_debug("Driver did not initialize current freq\n");
2322
			policy->cur = new_policy.cur;
2323
		} else {
2324
			if (policy->cur != new_policy.cur && has_target())
2325
				cpufreq_out_of_sync(policy, new_policy.cur);
2326
		}
2327 2328
	}

2329
	ret = cpufreq_set_policy(policy, &new_policy);
L
Linus Torvalds 已提交
2330

2331
unlock:
2332
	up_write(&policy->rwsem);
2333

2334
	cpufreq_cpu_put(policy);
L
Linus Torvalds 已提交
2335 2336 2337 2338
	return ret;
}
EXPORT_SYMBOL(cpufreq_update_policy);

2339
static int cpufreq_cpu_callback(struct notifier_block *nfb,
2340 2341 2342
					unsigned long action, void *hcpu)
{
	unsigned int cpu = (unsigned long)hcpu;
2343
	struct device *dev;
2344

2345 2346
	dev = get_cpu_device(cpu);
	if (dev) {
2347
		switch (action & ~CPU_TASKS_FROZEN) {
2348
		case CPU_ONLINE:
2349
			cpufreq_add_dev(dev, NULL);
2350
			break;
2351

2352
		case CPU_DOWN_PREPARE:
2353
			cpufreq_offline_prepare(cpu);
2354 2355 2356
			break;

		case CPU_POST_DEAD:
2357
			cpufreq_offline_finish(cpu);
2358
			break;
2359

2360
		case CPU_DOWN_FAILED:
2361
			cpufreq_add_dev(dev, NULL);
2362 2363 2364 2365 2366 2367
			break;
		}
	}
	return NOTIFY_OK;
}

2368
static struct notifier_block __refdata cpufreq_cpu_notifier = {
2369
	.notifier_call = cpufreq_cpu_callback,
2370
};
L
Linus Torvalds 已提交
2371

2372 2373 2374 2375 2376 2377 2378 2379 2380
/*********************************************************************
 *               BOOST						     *
 *********************************************************************/
static int cpufreq_boost_set_sw(int state)
{
	struct cpufreq_frequency_table *freq_table;
	struct cpufreq_policy *policy;
	int ret = -EINVAL;

2381
	for_each_active_policy(policy) {
2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416
		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;
			}
			policy->user_policy.max = policy->max;
			__cpufreq_governor(policy, CPUFREQ_GOV_LIMITS);
		}
	}

	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);

2417 2418
		pr_err("%s: Cannot %s BOOST\n",
		       __func__, state ? "enable" : "disable");
2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438
	}

	return ret;
}

int cpufreq_boost_supported(void)
{
	if (likely(cpufreq_driver))
		return cpufreq_driver->boost_supported;

	return 0;
}
EXPORT_SYMBOL_GPL(cpufreq_boost_supported);

int cpufreq_boost_enabled(void)
{
	return cpufreq_driver->boost_enabled;
}
EXPORT_SYMBOL_GPL(cpufreq_boost_enabled);

L
Linus Torvalds 已提交
2439 2440 2441 2442 2443 2444 2445 2446 2447
/*********************************************************************
 *               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.
 *
2448
 * Registers a CPU Frequency driver to this core code. This code
L
Linus Torvalds 已提交
2449
 * returns zero on success, -EBUSY when another driver got here first
2450
 * (and isn't unregistered in the meantime).
L
Linus Torvalds 已提交
2451 2452
 *
 */
2453
int cpufreq_register_driver(struct cpufreq_driver *driver_data)
L
Linus Torvalds 已提交
2454 2455 2456 2457
{
	unsigned long flags;
	int ret;

2458 2459 2460
	if (cpufreq_disabled())
		return -ENODEV;

L
Linus Torvalds 已提交
2461
	if (!driver_data || !driver_data->verify || !driver_data->init ||
2462
	    !(driver_data->setpolicy || driver_data->target_index ||
2463 2464
		    driver_data->target) ||
	     (driver_data->setpolicy && (driver_data->target_index ||
2465 2466
		    driver_data->target)) ||
	     (!!driver_data->get_intermediate != !!driver_data->target_intermediate))
L
Linus Torvalds 已提交
2467 2468
		return -EINVAL;

2469
	pr_debug("trying to register driver %s\n", driver_data->name);
L
Linus Torvalds 已提交
2470

2471
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2472
	if (cpufreq_driver) {
2473
		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2474
		return -EEXIST;
L
Linus Torvalds 已提交
2475
	}
2476
	cpufreq_driver = driver_data;
2477
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
2478

2479 2480 2481
	if (driver_data->setpolicy)
		driver_data->flags |= CPUFREQ_CONST_LOOPS;

2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492
	if (cpufreq_boost_supported()) {
		/*
		 * Check if driver provides function to enable boost -
		 * if not, use cpufreq_boost_set_sw as default
		 */
		if (!cpufreq_driver->set_boost)
			cpufreq_driver->set_boost = cpufreq_boost_set_sw;

		ret = cpufreq_sysfs_create_file(&boost.attr);
		if (ret) {
			pr_err("%s: cannot register global BOOST sysfs file\n",
2493
			       __func__);
2494 2495 2496 2497
			goto err_null_driver;
		}
	}

2498
	ret = subsys_interface_register(&cpufreq_interface);
2499
	if (ret)
2500
		goto err_boost_unreg;
L
Linus Torvalds 已提交
2501

2502 2503
	if (!(cpufreq_driver->flags & CPUFREQ_STICKY) &&
	    list_empty(&cpufreq_policy_list)) {
L
Linus Torvalds 已提交
2504
		/* if all ->init() calls failed, unregister */
2505 2506 2507
		pr_debug("%s: No CPU initialized for driver %s\n", __func__,
			 driver_data->name);
		goto err_if_unreg;
L
Linus Torvalds 已提交
2508 2509
	}

2510
	register_hotcpu_notifier(&cpufreq_cpu_notifier);
2511
	pr_debug("driver %s up and running\n", driver_data->name);
L
Linus Torvalds 已提交
2512

2513
	return 0;
2514 2515
err_if_unreg:
	subsys_interface_unregister(&cpufreq_interface);
2516 2517 2518
err_boost_unreg:
	if (cpufreq_boost_supported())
		cpufreq_sysfs_remove_file(&boost.attr);
2519
err_null_driver:
2520
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2521
	cpufreq_driver = NULL;
2522
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
D
Dave Jones 已提交
2523
	return ret;
L
Linus Torvalds 已提交
2524 2525 2526 2527 2528 2529
}
EXPORT_SYMBOL_GPL(cpufreq_register_driver);

/**
 * cpufreq_unregister_driver - unregister the current CPUFreq driver
 *
2530
 * Unregister the current CPUFreq driver. Only call this if you have
L
Linus Torvalds 已提交
2531 2532 2533 2534
 * 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.
 */
2535
int cpufreq_unregister_driver(struct cpufreq_driver *driver)
L
Linus Torvalds 已提交
2536 2537 2538
{
	unsigned long flags;

2539
	if (!cpufreq_driver || (driver != cpufreq_driver))
L
Linus Torvalds 已提交
2540 2541
		return -EINVAL;

2542
	pr_debug("unregistering driver %s\n", driver->name);
L
Linus Torvalds 已提交
2543

2544 2545
	/* Protect against concurrent cpu hotplug */
	get_online_cpus();
2546
	subsys_interface_unregister(&cpufreq_interface);
2547 2548 2549
	if (cpufreq_boost_supported())
		cpufreq_sysfs_remove_file(&boost.attr);

2550
	unregister_hotcpu_notifier(&cpufreq_cpu_notifier);
L
Linus Torvalds 已提交
2551

2552
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2553

2554
	cpufreq_driver = NULL;
2555

2556
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2557
	put_online_cpus();
L
Linus Torvalds 已提交
2558 2559 2560 2561

	return 0;
}
EXPORT_SYMBOL_GPL(cpufreq_unregister_driver);
2562

2563 2564 2565 2566 2567 2568 2569 2570
/*
 * 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,
};

2571 2572
static int __init cpufreq_core_init(void)
{
2573 2574 2575
	if (cpufreq_disabled())
		return -ENODEV;

2576
	cpufreq_global_kobject = kobject_create();
2577 2578
	BUG_ON(!cpufreq_global_kobject);

2579 2580
	register_syscore_ops(&cpufreq_syscore_ops);

2581 2582 2583
	return 0;
}
core_initcall(cpufreq_core_init);