cpufreq.c 65.5 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 35 36 37 38 39
/* Macros to iterate over lists */
/* Iterate over online CPUs policies */
static LIST_HEAD(cpufreq_policy_list);
#define for_each_policy(__policy)				\
	list_for_each_entry(__policy, &cpufreq_policy_list, policy_list)

40 41 42 43 44
/* 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
45
/**
D
Dave Jones 已提交
46
 * The "cpufreq driver" - the arch- or hardware-dependent low
L
Linus Torvalds 已提交
47 48 49
 * level driver of CPUFreq support, and its spinlock. This lock
 * also protects the cpufreq_cpu_data array.
 */
50
static struct cpufreq_driver *cpufreq_driver;
51
static DEFINE_PER_CPU(struct cpufreq_policy *, cpufreq_cpu_data);
52
static DEFINE_PER_CPU(struct cpufreq_policy *, cpufreq_cpu_data_fallback);
53
static DEFINE_RWLOCK(cpufreq_driver_lock);
54
DEFINE_MUTEX(cpufreq_governor_lock);
55

56
/* This one keeps track of the previously set governor of a removed CPU */
57
static DEFINE_PER_CPU(char[CPUFREQ_NAME_LEN], cpufreq_cpu_governor);
L
Linus Torvalds 已提交
58

59 60
/* Flag to suspend/resume CPUFreq governors */
static bool cpufreq_suspended;
L
Linus Torvalds 已提交
61

62 63 64 65 66
static inline bool has_target(void)
{
	return cpufreq_driver->target_index || cpufreq_driver->target;
}

67 68 69 70 71 72
/*
 * rwsem to guarantee that cpufreq driver module doesn't unload during critical
 * sections
 */
static DECLARE_RWSEM(cpufreq_rwsem);

L
Linus Torvalds 已提交
73
/* internal prototypes */
74 75
static int __cpufreq_governor(struct cpufreq_policy *policy,
		unsigned int event);
76
static unsigned int __cpufreq_get(struct cpufreq_policy *policy);
77
static void handle_update(struct work_struct *work);
L
Linus Torvalds 已提交
78 79

/**
80 81
 * 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
82 83 84 85
 * "transition" list for kernel code that needs to handle
 * changes to devices when the CPU clock speed changes.
 * The mutex locks both lists.
 */
86
static BLOCKING_NOTIFIER_HEAD(cpufreq_policy_notifier_list);
87
static struct srcu_notifier_head cpufreq_transition_notifier_list;
L
Linus Torvalds 已提交
88

89
static bool init_cpufreq_transition_notifier_list_called;
90 91 92
static int __init init_cpufreq_transition_notifier_list(void)
{
	srcu_init_notifier_head(&cpufreq_transition_notifier_list);
93
	init_cpufreq_transition_notifier_list_called = true;
94 95
	return 0;
}
96
pure_initcall(init_cpufreq_transition_notifier_list);
L
Linus Torvalds 已提交
97

98
static int off __read_mostly;
99
static int cpufreq_disabled(void)
100 101 102 103 104 105 106
{
	return off;
}
void disable_cpufreq(void)
{
	off = 1;
}
107
static DEFINE_MUTEX(cpufreq_governor_mutex);
L
Linus Torvalds 已提交
108

109 110
bool have_governor_per_policy(void)
{
111
	return !!(cpufreq_driver->flags & CPUFREQ_HAVE_GOVERNOR_PER_POLICY);
112
}
113
EXPORT_SYMBOL_GPL(have_governor_per_policy);
114

115 116 117 118 119 120 121 122 123
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);

124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158
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);

159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189
/*
 * 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;

	/*
	 * The driver only supports the SMP configuartion where all processors
	 * share the clock and voltage and clock.
	 */
	cpumask_setall(policy->cpus);

	return 0;
}
EXPORT_SYMBOL_GPL(cpufreq_generic_init);

190 191 192 193 194
unsigned int cpufreq_generic_get(unsigned int cpu)
{
	struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);

	if (!policy || IS_ERR(policy->clk)) {
195 196
		pr_err("%s: No %s associated to cpu: %d\n",
		       __func__, policy ? "clk" : "policy", cpu);
197 198 199 200 201 202 203
		return 0;
	}

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

204 205 206 207 208 209
/* Only for cpufreq core internal use */
struct cpufreq_policy *cpufreq_cpu_get_raw(unsigned int cpu)
{
	return per_cpu(cpufreq_cpu_data, cpu);
}

210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226
/**
 * 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.
 *
 * It also takes a read-lock of 'cpufreq_rwsem' and doesn't put it back if a
 * valid policy is found. This is done to make sure the driver doesn't get
 * unregistered while the policy is being used.
 *
 * Return: A valid policy on success, otherwise NULL on failure.
 */
227
struct cpufreq_policy *cpufreq_cpu_get(unsigned int cpu)
L
Linus Torvalds 已提交
228
{
229
	struct cpufreq_policy *policy = NULL;
L
Linus Torvalds 已提交
230 231
	unsigned long flags;

232
	if (WARN_ON(cpu >= nr_cpu_ids))
233 234 235 236
		return NULL;

	if (!down_read_trylock(&cpufreq_rwsem))
		return NULL;
L
Linus Torvalds 已提交
237 238

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

241 242 243 244 245 246
	if (cpufreq_driver) {
		/* get the CPU */
		policy = per_cpu(cpufreq_cpu_data, cpu);
		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
	if (!policy)
251
		up_read(&cpufreq_rwsem);
L
Linus Torvalds 已提交
252

253
	return policy;
254
}
L
Linus Torvalds 已提交
255 256
EXPORT_SYMBOL_GPL(cpufreq_cpu_get);

257 258 259 260 261 262 263 264 265 266
/**
 * cpufreq_cpu_put: Decrements the usage count of a policy
 *
 * @policy: policy earlier returned by cpufreq_cpu_get().
 *
 * This decrements the kobject reference count incremented earlier by calling
 * cpufreq_cpu_get().
 *
 * It also drops the read-lock of 'cpufreq_rwsem' taken at cpufreq_cpu_get().
 */
267
void cpufreq_cpu_put(struct cpufreq_policy *policy)
L
Linus Torvalds 已提交
268
{
269 270
	kobject_put(&policy->kobj);
	up_read(&cpufreq_rwsem);
L
Linus Torvalds 已提交
271 272 273 274 275 276 277 278 279 280 281 282
}
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
283
 * systems as each CPU might be scaled differently. So, use the arch
L
Linus Torvalds 已提交
284 285
 * per-CPU loops_per_jiffy value wherever possible.
 */
286
static void adjust_jiffies(unsigned long val, struct cpufreq_freqs *ci)
L
Linus Torvalds 已提交
287
{
288 289 290 291
#ifndef CONFIG_SMP
	static unsigned long l_p_j_ref;
	static unsigned int l_p_j_ref_freq;

L
Linus Torvalds 已提交
292 293 294 295 296 297
	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;
298 299
		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
300
	}
301
	if (val == CPUFREQ_POSTCHANGE && ci->old != ci->new) {
302 303
		loops_per_jiffy = cpufreq_scale(l_p_j_ref, l_p_j_ref_freq,
								ci->new);
304 305
		pr_debug("scaling loops_per_jiffy to %lu for frequency %u kHz\n",
			 loops_per_jiffy, ci->new);
L
Linus Torvalds 已提交
306 307
	}
#endif
308
}
L
Linus Torvalds 已提交
309

310
static void __cpufreq_notify_transition(struct cpufreq_policy *policy,
311
		struct cpufreq_freqs *freqs, unsigned int state)
L
Linus Torvalds 已提交
312 313 314
{
	BUG_ON(irqs_disabled());

315 316 317
	if (cpufreq_disabled())
		return;

318
	freqs->flags = cpufreq_driver->flags;
319
	pr_debug("notification %u of frequency transition to %u kHz\n",
320
		 state, freqs->new);
L
Linus Torvalds 已提交
321 322

	switch (state) {
323

L
Linus Torvalds 已提交
324
	case CPUFREQ_PRECHANGE:
325
		/* detect if the driver reported a value as "old frequency"
326 327
		 * which is not equal to what the cpufreq core thinks is
		 * "old frequency".
L
Linus Torvalds 已提交
328
		 */
329
		if (!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
330 331
			if ((policy) && (policy->cpu == freqs->cpu) &&
			    (policy->cur) && (policy->cur != freqs->old)) {
332 333
				pr_debug("Warning: CPU frequency is %u, cpufreq assumed %u kHz\n",
					 freqs->old, policy->cur);
334
				freqs->old = policy->cur;
L
Linus Torvalds 已提交
335 336
			}
		}
337
		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
338
				CPUFREQ_PRECHANGE, freqs);
L
Linus Torvalds 已提交
339 340
		adjust_jiffies(CPUFREQ_PRECHANGE, freqs);
		break;
341

L
Linus Torvalds 已提交
342 343
	case CPUFREQ_POSTCHANGE:
		adjust_jiffies(CPUFREQ_POSTCHANGE, freqs);
344 345
		pr_debug("FREQ: %lu - CPU: %lu\n",
			 (unsigned long)freqs->new, (unsigned long)freqs->cpu);
346
		trace_cpu_frequency(freqs->new, freqs->cpu);
347
		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
348
				CPUFREQ_POSTCHANGE, freqs);
349 350
		if (likely(policy) && likely(policy->cpu == freqs->cpu))
			policy->cur = freqs->new;
L
Linus Torvalds 已提交
351 352 353
		break;
	}
}
354

355 356 357 358 359 360 361 362
/**
 * 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.
 */
363
static void cpufreq_notify_transition(struct cpufreq_policy *policy,
364 365 366 367 368
		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
369

370
/* Do post notifications when there are chances that transition has failed */
371
static void cpufreq_notify_post_transition(struct cpufreq_policy *policy,
372 373 374 375 376 377 378 379 380 381 382
		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);
}

383 384 385
void cpufreq_freq_transition_begin(struct cpufreq_policy *policy,
		struct cpufreq_freqs *freqs)
{
386 387 388 389 390 391 392 393 394 395 396 397

	/*
	 * 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);

398 399 400 401 402 403 404 405 406 407 408
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;
409
	policy->transition_task = current;
410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425

	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;
426
	policy->transition_task = NULL;
427 428 429 430 431

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

L
Linus Torvalds 已提交
432 433 434 435

/*********************************************************************
 *                          SYSFS INTERFACE                          *
 *********************************************************************/
436
static ssize_t show_boost(struct kobject *kobj,
437 438 439 440 441 442 443 444 445 446 447 448 449 450 451
				 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)) {
452 453
		pr_err("%s: Cannot %s BOOST!\n",
		       __func__, enable ? "enable" : "disable");
454 455 456
		return -EINVAL;
	}

457 458
	pr_debug("%s: cpufreq BOOST %s\n",
		 __func__, enable ? "enabled" : "disabled");
459 460 461 462

	return count;
}
define_one_global_rw(boost);
L
Linus Torvalds 已提交
463

464
static struct cpufreq_governor *find_governor(const char *str_governor)
465 466 467
{
	struct cpufreq_governor *t;

468
	for_each_governor(t)
469
		if (!strncasecmp(str_governor, t->name, CPUFREQ_NAME_LEN))
470 471 472 473 474
			return t;

	return NULL;
}

L
Linus Torvalds 已提交
475 476 477
/**
 * cpufreq_parse_governor - parse a governor string
 */
D
Dave Jones 已提交
478
static int cpufreq_parse_governor(char *str_governor, unsigned int *policy,
L
Linus Torvalds 已提交
479 480
				struct cpufreq_governor **governor)
{
481
	int err = -EINVAL;
482 483

	if (!cpufreq_driver)
484 485
		goto out;

486
	if (cpufreq_driver->setpolicy) {
487
		if (!strncasecmp(str_governor, "performance", CPUFREQ_NAME_LEN)) {
L
Linus Torvalds 已提交
488
			*policy = CPUFREQ_POLICY_PERFORMANCE;
489
			err = 0;
490
		} else if (!strncasecmp(str_governor, "powersave",
491
						CPUFREQ_NAME_LEN)) {
L
Linus Torvalds 已提交
492
			*policy = CPUFREQ_POLICY_POWERSAVE;
493
			err = 0;
L
Linus Torvalds 已提交
494
		}
495
	} else {
L
Linus Torvalds 已提交
496
		struct cpufreq_governor *t;
497

498
		mutex_lock(&cpufreq_governor_mutex);
499

500
		t = find_governor(str_governor);
501

502
		if (t == NULL) {
503
			int ret;
504

505 506 507
			mutex_unlock(&cpufreq_governor_mutex);
			ret = request_module("cpufreq_%s", str_governor);
			mutex_lock(&cpufreq_governor_mutex);
508

509
			if (ret == 0)
510
				t = find_governor(str_governor);
511 512
		}

513 514 515
		if (t != NULL) {
			*governor = t;
			err = 0;
L
Linus Torvalds 已提交
516
		}
517

518
		mutex_unlock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
519
	}
520
out:
521
	return err;
L
Linus Torvalds 已提交
522 523 524
}

/**
525 526
 * cpufreq_per_cpu_attr_read() / show_##file_name() -
 * print out cpufreq information
L
Linus Torvalds 已提交
527 528 529 530 531
 *
 * Write out information from cpufreq_driver->policy[cpu]; object must be
 * "unsigned int".
 */

532 533
#define show_one(file_name, object)			\
static ssize_t show_##file_name				\
D
Dave Jones 已提交
534
(struct cpufreq_policy *policy, char *buf)		\
535
{							\
536
	return sprintf(buf, "%u\n", policy->object);	\
L
Linus Torvalds 已提交
537 538 539 540
}

show_one(cpuinfo_min_freq, cpuinfo.min_freq);
show_one(cpuinfo_max_freq, cpuinfo.max_freq);
541
show_one(cpuinfo_transition_latency, cpuinfo.transition_latency);
L
Linus Torvalds 已提交
542 543
show_one(scaling_min_freq, min);
show_one(scaling_max_freq, max);
544

545
static ssize_t show_scaling_cur_freq(struct cpufreq_policy *policy, char *buf)
546 547 548 549 550 551 552 553 554
{
	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
555

556
static int cpufreq_set_policy(struct cpufreq_policy *policy,
557
				struct cpufreq_policy *new_policy);
558

L
Linus Torvalds 已提交
559 560 561 562 563
/**
 * 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
564
(struct cpufreq_policy *policy, const char *buf, size_t count)		\
L
Linus Torvalds 已提交
565
{									\
566
	int ret, temp;							\
L
Linus Torvalds 已提交
567 568 569 570 571 572
	struct cpufreq_policy new_policy;				\
									\
	ret = cpufreq_get_policy(&new_policy, policy->cpu);		\
	if (ret)							\
		return -EINVAL;						\
									\
573
	ret = sscanf(buf, "%u", &new_policy.object);			\
L
Linus Torvalds 已提交
574 575 576
	if (ret != 1)							\
		return -EINVAL;						\
									\
577
	temp = new_policy.object;					\
578
	ret = cpufreq_set_policy(policy, &new_policy);		\
579 580
	if (!ret)							\
		policy->user_policy.object = temp;			\
L
Linus Torvalds 已提交
581 582 583 584
									\
	return ret ? ret : count;					\
}

585 586
store_one(scaling_min_freq, min);
store_one(scaling_max_freq, max);
L
Linus Torvalds 已提交
587 588 589 590

/**
 * show_cpuinfo_cur_freq - current CPU frequency as detected by hardware
 */
D
Dave Jones 已提交
591 592
static ssize_t show_cpuinfo_cur_freq(struct cpufreq_policy *policy,
					char *buf)
L
Linus Torvalds 已提交
593
{
594
	unsigned int cur_freq = __cpufreq_get(policy);
L
Linus Torvalds 已提交
595 596 597 598 599 600 601 602
	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
603
static ssize_t show_scaling_governor(struct cpufreq_policy *policy, char *buf)
L
Linus Torvalds 已提交
604
{
605
	if (policy->policy == CPUFREQ_POLICY_POWERSAVE)
L
Linus Torvalds 已提交
606 607 608 609
		return sprintf(buf, "powersave\n");
	else if (policy->policy == CPUFREQ_POLICY_PERFORMANCE)
		return sprintf(buf, "performance\n");
	else if (policy->governor)
610
		return scnprintf(buf, CPUFREQ_NAME_PLEN, "%s\n",
611
				policy->governor->name);
L
Linus Torvalds 已提交
612 613 614 615 616 617
	return -EINVAL;
}

/**
 * store_scaling_governor - store policy for the specified CPU
 */
D
Dave Jones 已提交
618 619
static ssize_t store_scaling_governor(struct cpufreq_policy *policy,
					const char *buf, size_t count)
L
Linus Torvalds 已提交
620
{
621
	int ret;
L
Linus Torvalds 已提交
622 623 624 625 626 627 628
	char	str_governor[16];
	struct cpufreq_policy new_policy;

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

629
	ret = sscanf(buf, "%15s", str_governor);
L
Linus Torvalds 已提交
630 631 632
	if (ret != 1)
		return -EINVAL;

633 634
	if (cpufreq_parse_governor(str_governor, &new_policy.policy,
						&new_policy.governor))
L
Linus Torvalds 已提交
635 636
		return -EINVAL;

637
	ret = cpufreq_set_policy(policy, &new_policy);
638 639 640 641

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

642 643 644 645
	if (ret)
		return ret;
	else
		return count;
L
Linus Torvalds 已提交
646 647 648 649 650
}

/**
 * show_scaling_driver - show the cpufreq driver currently loaded
 */
D
Dave Jones 已提交
651
static ssize_t show_scaling_driver(struct cpufreq_policy *policy, char *buf)
L
Linus Torvalds 已提交
652
{
653
	return scnprintf(buf, CPUFREQ_NAME_PLEN, "%s\n", cpufreq_driver->name);
L
Linus Torvalds 已提交
654 655 656 657 658
}

/**
 * show_scaling_available_governors - show the available CPUfreq governors
 */
D
Dave Jones 已提交
659 660
static ssize_t show_scaling_available_governors(struct cpufreq_policy *policy,
						char *buf)
L
Linus Torvalds 已提交
661 662 663 664
{
	ssize_t i = 0;
	struct cpufreq_governor *t;

665
	if (!has_target()) {
L
Linus Torvalds 已提交
666 667 668 669
		i += sprintf(buf, "performance powersave");
		goto out;
	}

670
	for_each_governor(t) {
671 672
		if (i >= (ssize_t) ((PAGE_SIZE / sizeof(char))
		    - (CPUFREQ_NAME_LEN + 2)))
L
Linus Torvalds 已提交
673
			goto out;
674
		i += scnprintf(&buf[i], CPUFREQ_NAME_PLEN, "%s ", t->name);
L
Linus Torvalds 已提交
675
	}
676
out:
L
Linus Torvalds 已提交
677 678 679
	i += sprintf(&buf[i], "\n");
	return i;
}
680

681
ssize_t cpufreq_show_cpus(const struct cpumask *mask, char *buf)
L
Linus Torvalds 已提交
682 683 684 685
{
	ssize_t i = 0;
	unsigned int cpu;

686
	for_each_cpu(cpu, mask) {
L
Linus Torvalds 已提交
687 688 689 690
		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))
691
			break;
L
Linus Torvalds 已提交
692 693 694 695
	}
	i += sprintf(&buf[i], "\n");
	return i;
}
696
EXPORT_SYMBOL_GPL(cpufreq_show_cpus);
L
Linus Torvalds 已提交
697

698 699 700 701 702 703
/**
 * 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)
{
704
	return cpufreq_show_cpus(policy->related_cpus, buf);
705 706 707 708 709 710 711
}

/**
 * show_affected_cpus - show the CPUs affected by each transition
 */
static ssize_t show_affected_cpus(struct cpufreq_policy *policy, char *buf)
{
712
	return cpufreq_show_cpus(policy->cpus, buf);
713 714
}

715
static ssize_t store_scaling_setspeed(struct cpufreq_policy *policy,
D
Dave Jones 已提交
716
					const char *buf, size_t count)
717 718 719 720
{
	unsigned int freq = 0;
	unsigned int ret;

721
	if (!policy->governor || !policy->governor->store_setspeed)
722 723 724 725 726 727 728 729 730 731 732 733 734
		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)
{
735
	if (!policy->governor || !policy->governor->show_setspeed)
736 737 738 739
		return sprintf(buf, "<unsupported>\n");

	return policy->governor->show_setspeed(policy, buf);
}
L
Linus Torvalds 已提交
740

741
/**
742
 * show_bios_limit - show the current cpufreq HW/BIOS limitation
743 744 745 746 747
 */
static ssize_t show_bios_limit(struct cpufreq_policy *policy, char *buf)
{
	unsigned int limit;
	int ret;
748 749
	if (cpufreq_driver->bios_limit) {
		ret = cpufreq_driver->bios_limit(policy->cpu, &limit);
750 751 752 753 754 755
		if (!ret)
			return sprintf(buf, "%u\n", limit);
	}
	return sprintf(buf, "%u\n", policy->cpuinfo.max_freq);
}

756 757 758 759 760 761 762 763 764 765 766 767 768 769
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
770

D
Dave Jones 已提交
771
static struct attribute *default_attrs[] = {
L
Linus Torvalds 已提交
772 773
	&cpuinfo_min_freq.attr,
	&cpuinfo_max_freq.attr,
774
	&cpuinfo_transition_latency.attr,
L
Linus Torvalds 已提交
775 776 777
	&scaling_min_freq.attr,
	&scaling_max_freq.attr,
	&affected_cpus.attr,
778
	&related_cpus.attr,
L
Linus Torvalds 已提交
779 780 781
	&scaling_governor.attr,
	&scaling_driver.attr,
	&scaling_available_governors.attr,
782
	&scaling_setspeed.attr,
L
Linus Torvalds 已提交
783 784 785
	NULL
};

786 787
#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
788

789
static ssize_t show(struct kobject *kobj, struct attribute *attr, char *buf)
L
Linus Torvalds 已提交
790
{
D
Dave Jones 已提交
791 792
	struct cpufreq_policy *policy = to_policy(kobj);
	struct freq_attr *fattr = to_attr(attr);
793
	ssize_t ret;
794 795

	if (!down_read_trylock(&cpufreq_rwsem))
796
		return -EINVAL;
797

798
	down_read(&policy->rwsem);
799

800 801 802 803 804
	if (fattr->show)
		ret = fattr->show(policy, buf);
	else
		ret = -EIO;

805
	up_read(&policy->rwsem);
806
	up_read(&cpufreq_rwsem);
807

L
Linus Torvalds 已提交
808 809 810
	return ret;
}

D
Dave Jones 已提交
811 812
static ssize_t store(struct kobject *kobj, struct attribute *attr,
		     const char *buf, size_t count)
L
Linus Torvalds 已提交
813
{
D
Dave Jones 已提交
814 815
	struct cpufreq_policy *policy = to_policy(kobj);
	struct freq_attr *fattr = to_attr(attr);
816
	ssize_t ret = -EINVAL;
817

818 819 820 821 822
	get_online_cpus();

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

823
	if (!down_read_trylock(&cpufreq_rwsem))
824
		goto unlock;
825

826
	down_write(&policy->rwsem);
827

828 829 830 831 832
	if (fattr->store)
		ret = fattr->store(policy, buf, count);
	else
		ret = -EIO;

833
	up_write(&policy->rwsem);
834 835

	up_read(&cpufreq_rwsem);
836 837 838
unlock:
	put_online_cpus();

L
Linus Torvalds 已提交
839 840 841
	return ret;
}

D
Dave Jones 已提交
842
static void cpufreq_sysfs_release(struct kobject *kobj)
L
Linus Torvalds 已提交
843
{
D
Dave Jones 已提交
844
	struct cpufreq_policy *policy = to_policy(kobj);
845
	pr_debug("last reference is dropped\n");
L
Linus Torvalds 已提交
846 847 848
	complete(&policy->kobj_unregister);
}

849
static const struct sysfs_ops sysfs_ops = {
L
Linus Torvalds 已提交
850 851 852 853 854 855 856 857 858 859
	.show	= show,
	.store	= store,
};

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

860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902
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);

903
/* symlink affected CPUs */
904
static int cpufreq_add_dev_symlink(struct cpufreq_policy *policy)
905 906 907 908 909
{
	unsigned int j;
	int ret = 0;

	for_each_cpu(j, policy->cpus) {
910
		struct device *cpu_dev;
911

912
		if (j == policy->cpu)
913 914
			continue;

915
		pr_debug("Adding link for CPU: %u\n", j);
916 917
		cpu_dev = get_cpu_device(j);
		ret = sysfs_create_link(&cpu_dev->kobj, &policy->kobj,
918
					"cpufreq");
919 920
		if (ret)
			break;
921 922 923 924
	}
	return ret;
}

925
static int cpufreq_add_dev_interface(struct cpufreq_policy *policy,
926
				     struct device *dev)
927 928 929 930 931
{
	struct freq_attr **drv_attr;
	int ret = 0;

	/* set up files for this cpu device */
932
	drv_attr = cpufreq_driver->attr;
V
Viresh Kumar 已提交
933
	while (drv_attr && *drv_attr) {
934 935
		ret = sysfs_create_file(&policy->kobj, &((*drv_attr)->attr));
		if (ret)
936
			return ret;
937 938
		drv_attr++;
	}
939
	if (cpufreq_driver->get) {
940 941
		ret = sysfs_create_file(&policy->kobj, &cpuinfo_cur_freq.attr);
		if (ret)
942
			return ret;
943
	}
944 945 946

	ret = sysfs_create_file(&policy->kobj, &scaling_cur_freq.attr);
	if (ret)
947
		return ret;
948

949
	if (cpufreq_driver->bios_limit) {
950 951
		ret = sysfs_create_file(&policy->kobj, &bios_limit.attr);
		if (ret)
952
			return ret;
953
	}
954

955
	return cpufreq_add_dev_symlink(policy);
956 957 958 959
}

static void cpufreq_init_policy(struct cpufreq_policy *policy)
{
960
	struct cpufreq_governor *gov = NULL;
961 962 963
	struct cpufreq_policy new_policy;
	int ret = 0;

964
	memcpy(&new_policy, policy, sizeof(*policy));
965

966
	/* Update governor of new_policy to the governor used before hotplug */
967
	gov = find_governor(per_cpu(cpufreq_cpu_governor, policy->cpu));
968 969 970 971 972 973 974 975
	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;

976 977
	/* Use the default policy if its valid. */
	if (cpufreq_driver->setpolicy)
978
		cpufreq_parse_governor(gov->name, &new_policy.policy, NULL);
979 980

	/* set default policy */
981
	ret = cpufreq_set_policy(policy, &new_policy);
982
	if (ret) {
983
		pr_debug("setting policy failed\n");
984 985
		if (cpufreq_driver->exit)
			cpufreq_driver->exit(policy);
986
	}
987 988
}

989
static int cpufreq_add_policy_cpu(struct cpufreq_policy *policy,
990
				  unsigned int cpu, struct device *dev)
991
{
992
	int ret = 0;
993 994
	unsigned long flags;

995 996 997 998
	/* Has this CPU been taken care of already? */
	if (cpumask_test_cpu(cpu, policy->cpus))
		return 0;

999
	if (has_target()) {
1000 1001 1002 1003 1004 1005
		ret = __cpufreq_governor(policy, CPUFREQ_GOV_STOP);
		if (ret) {
			pr_err("%s: Failed to stop governor\n", __func__);
			return ret;
		}
	}
1006

1007
	down_write(&policy->rwsem);
V
Viresh Kumar 已提交
1008

1009
	write_lock_irqsave(&cpufreq_driver_lock, flags);
V
Viresh Kumar 已提交
1010

1011 1012
	cpumask_set_cpu(cpu, policy->cpus);
	per_cpu(cpufreq_cpu_data, cpu) = policy;
1013
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
1014

1015
	up_write(&policy->rwsem);
V
Viresh Kumar 已提交
1016

1017
	if (has_target()) {
1018 1019 1020 1021 1022
		ret = __cpufreq_governor(policy, CPUFREQ_GOV_START);
		if (!ret)
			ret = __cpufreq_governor(policy, CPUFREQ_GOV_LIMITS);

		if (ret) {
1023 1024 1025
			pr_err("%s: Failed to start governor\n", __func__);
			return ret;
		}
1026
	}
1027

1028
	return sysfs_create_link(&dev->kobj, &policy->kobj, "cpufreq");
1029
}
L
Linus Torvalds 已提交
1030

1031 1032 1033 1034 1035
static struct cpufreq_policy *cpufreq_policy_restore(unsigned int cpu)
{
	struct cpufreq_policy *policy;
	unsigned long flags;

1036
	read_lock_irqsave(&cpufreq_driver_lock, flags);
1037 1038 1039

	policy = per_cpu(cpufreq_cpu_data_fallback, cpu);

1040
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
1041

1042 1043
	if (policy)
		policy->governor = NULL;
1044

1045 1046 1047
	return policy;
}

1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061
static struct cpufreq_policy *cpufreq_policy_alloc(void)
{
	struct cpufreq_policy *policy;

	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;

1062
	INIT_LIST_HEAD(&policy->policy_list);
1063
	init_rwsem(&policy->rwsem);
1064 1065
	spin_lock_init(&policy->transition_lock);
	init_waitqueue_head(&policy->transition_wait);
1066 1067
	init_completion(&policy->kobj_unregister);
	INIT_WORK(&policy->update, handle_update);
1068

1069 1070 1071 1072 1073 1074 1075 1076 1077 1078
	return policy;

err_free_cpumask:
	free_cpumask_var(policy->cpus);
err_free_policy:
	kfree(policy);

	return NULL;
}

1079 1080 1081 1082 1083
static void cpufreq_policy_put_kobj(struct cpufreq_policy *policy)
{
	struct kobject *kobj;
	struct completion *cmp;

1084 1085 1086
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
			CPUFREQ_REMOVE_POLICY, policy);

1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102
	down_read(&policy->rwsem);
	kobj = &policy->kobj;
	cmp = &policy->kobj_unregister;
	up_read(&policy->rwsem);
	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");
}

1103 1104 1105 1106 1107 1108 1109
static void cpufreq_policy_free(struct cpufreq_policy *policy)
{
	free_cpumask_var(policy->related_cpus);
	free_cpumask_var(policy->cpus);
	kfree(policy);
}

1110 1111
static int update_policy_cpu(struct cpufreq_policy *policy, unsigned int cpu,
			     struct device *cpu_dev)
1112
{
1113 1114
	int ret;

1115
	if (WARN_ON(cpu == policy->cpu))
1116 1117 1118 1119 1120 1121 1122 1123
		return 0;

	/* Move kobject to the new policy->cpu */
	ret = kobject_move(&policy->kobj, &cpu_dev->kobj);
	if (ret) {
		pr_err("%s: Failed to move kobj: %d\n", __func__, ret);
		return ret;
	}
1124

1125
	down_write(&policy->rwsem);
1126
	policy->cpu = cpu;
1127
	up_write(&policy->rwsem);
1128

1129
	return 0;
1130 1131
}

1132 1133 1134 1135 1136 1137 1138 1139 1140 1141
/**
 * 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
1142
{
1143
	unsigned int j, cpu = dev->id;
1144
	int ret = -ENOMEM;
1145
	struct cpufreq_policy *policy;
L
Linus Torvalds 已提交
1146
	unsigned long flags;
1147
	bool recover_policy = cpufreq_suspended;
L
Linus Torvalds 已提交
1148

1149 1150 1151
	if (cpu_is_offline(cpu))
		return 0;

1152
	pr_debug("adding CPU %u\n", cpu);
L
Linus Torvalds 已提交
1153

1154 1155 1156
	if (!down_read_trylock(&cpufreq_rwsem))
		return 0;

1157
	/* Check if this CPU already has a policy to manage it */
1158
	read_lock_irqsave(&cpufreq_driver_lock, flags);
V
Viresh Kumar 已提交
1159
	for_each_policy(policy) {
1160
		if (cpumask_test_cpu(cpu, policy->related_cpus)) {
1161
			read_unlock_irqrestore(&cpufreq_driver_lock, flags);
1162
			ret = cpufreq_add_policy_cpu(policy, cpu, dev);
1163 1164
			up_read(&cpufreq_rwsem);
			return ret;
V
Viresh Kumar 已提交
1165
		}
1166
	}
1167
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
1168

1169 1170 1171 1172
	/*
	 * Restore the saved policy when doing light-weight init and fall back
	 * to the full init if that fails.
	 */
1173
	policy = recover_policy ? cpufreq_policy_restore(cpu) : NULL;
1174
	if (!policy) {
1175
		recover_policy = false;
1176
		policy = cpufreq_policy_alloc();
1177 1178 1179
		if (!policy)
			goto nomem_out;
	}
1180 1181 1182 1183 1184 1185 1186

	/*
	 * In the resume path, since we restore a saved policy, the assignment
	 * to policy->cpu is like an update of the existing policy, rather than
	 * the creation of a brand new one. So we need to perform this update
	 * by invoking update_policy_cpu().
	 */
1187 1188 1189
	if (recover_policy && cpu != policy->cpu)
		WARN_ON(update_policy_cpu(policy, cpu, dev));
	else
1190 1191
		policy->cpu = cpu;

1192
	cpumask_copy(policy->cpus, cpumask_of(cpu));
L
Linus Torvalds 已提交
1193 1194 1195 1196

	/* call driver. From then on the cpufreq must be able
	 * to accept all calls to ->verify and ->setpolicy for this CPU
	 */
1197
	ret = cpufreq_driver->init(policy);
L
Linus Torvalds 已提交
1198
	if (ret) {
1199
		pr_debug("initialization failed\n");
V
Viresh Kumar 已提交
1200
		goto err_set_policy_cpu;
L
Linus Torvalds 已提交
1201
	}
V
Viresh Kumar 已提交
1202

1203 1204
	down_write(&policy->rwsem);

1205 1206 1207 1208 1209 1210 1211 1212 1213
	/* related cpus should atleast have policy->cpus */
	cpumask_or(policy->related_cpus, policy->related_cpus, policy->cpus);

	/*
	 * 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);

1214
	if (!recover_policy) {
1215 1216
		policy->user_policy.min = policy->min;
		policy->user_policy.max = policy->max;
1217 1218 1219 1220 1221 1222 1223 1224 1225

		/* prepare interface data */
		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);
			goto err_init_policy_kobj;
		}
1226 1227
	}

1228 1229 1230 1231 1232
	write_lock_irqsave(&cpufreq_driver_lock, flags);
	for_each_cpu(j, policy->cpus)
		per_cpu(cpufreq_cpu_data, j) = policy;
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);

1233
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
1234 1235 1236 1237 1238 1239 1240
		policy->cur = cpufreq_driver->get(policy->cpu);
		if (!policy->cur) {
			pr_err("%s: ->get() failed\n", __func__);
			goto err_get_freq;
		}
	}

1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280
	/*
	 * 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);
		}
	}

1281 1282 1283
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
				     CPUFREQ_START, policy);

1284
	if (!recover_policy) {
1285
		ret = cpufreq_add_dev_interface(policy, dev);
1286 1287
		if (ret)
			goto err_out_unregister;
1288 1289
		blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
				CPUFREQ_CREATE_POLICY, policy);
1290
	}
1291

1292 1293 1294 1295
	write_lock_irqsave(&cpufreq_driver_lock, flags);
	list_add(&policy->policy_list, &cpufreq_policy_list);
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);

1296 1297
	cpufreq_init_policy(policy);

1298
	if (!recover_policy) {
1299 1300 1301
		policy->user_policy.policy = policy->policy;
		policy->user_policy.governor = policy->governor;
	}
1302
	up_write(&policy->rwsem);
1303

1304
	kobject_uevent(&policy->kobj, KOBJ_ADD);
1305

1306 1307
	up_read(&cpufreq_rwsem);

1308 1309 1310 1311
	/* Callback for handling stuff after policy is ready */
	if (cpufreq_driver->ready)
		cpufreq_driver->ready(policy);

1312
	pr_debug("initialization complete\n");
1313

L
Linus Torvalds 已提交
1314 1315 1316
	return 0;

err_out_unregister:
1317
err_get_freq:
1318
	write_lock_irqsave(&cpufreq_driver_lock, flags);
1319
	for_each_cpu(j, policy->cpus)
1320
		per_cpu(cpufreq_cpu_data, j) = NULL;
1321
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
1322

1323 1324 1325 1326 1327
	if (!recover_policy) {
		kobject_put(&policy->kobj);
		wait_for_completion(&policy->kobj_unregister);
	}
err_init_policy_kobj:
1328 1329
	up_write(&policy->rwsem);

1330 1331
	if (cpufreq_driver->exit)
		cpufreq_driver->exit(policy);
V
Viresh Kumar 已提交
1332
err_set_policy_cpu:
1333
	if (recover_policy) {
1334 1335
		/* Do not leave stale fallback data behind. */
		per_cpu(cpufreq_cpu_data_fallback, cpu) = NULL;
1336
		cpufreq_policy_put_kobj(policy);
1337
	}
1338
	cpufreq_policy_free(policy);
1339

L
Linus Torvalds 已提交
1340
nomem_out:
1341 1342
	up_read(&cpufreq_rwsem);

L
Linus Torvalds 已提交
1343 1344 1345
	return ret;
}

1346
static int __cpufreq_remove_dev_prepare(struct device *dev,
1347
					struct subsys_interface *sif)
L
Linus Torvalds 已提交
1348
{
1349
	unsigned int cpu = dev->id, cpus;
1350
	int ret;
L
Linus Torvalds 已提交
1351
	unsigned long flags;
1352
	struct cpufreq_policy *policy;
L
Linus Torvalds 已提交
1353

1354
	pr_debug("%s: unregistering CPU %u\n", __func__, cpu);
L
Linus Torvalds 已提交
1355

1356
	write_lock_irqsave(&cpufreq_driver_lock, flags);
V
Viresh Kumar 已提交
1357

1358
	policy = per_cpu(cpufreq_cpu_data, cpu);
V
Viresh Kumar 已提交
1359

1360
	/* Save the policy somewhere when doing a light-weight tear-down */
1361
	if (cpufreq_suspended)
1362
		per_cpu(cpufreq_cpu_data_fallback, cpu) = policy;
1363

1364
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
1365

1366
	if (!policy) {
1367
		pr_debug("%s: No cpu_data found\n", __func__);
L
Linus Torvalds 已提交
1368 1369 1370
		return -EINVAL;
	}

1371
	if (has_target()) {
1372 1373 1374 1375 1376
		ret = __cpufreq_governor(policy, CPUFREQ_GOV_STOP);
		if (ret) {
			pr_err("%s: Failed to stop governor\n", __func__);
			return ret;
		}
L
Linus Torvalds 已提交
1377

1378
		strncpy(per_cpu(cpufreq_cpu_governor, cpu),
1379
			policy->governor->name, CPUFREQ_NAME_LEN);
1380
	}
L
Linus Torvalds 已提交
1381

1382
	down_read(&policy->rwsem);
1383
	cpus = cpumask_weight(policy->cpus);
1384
	up_read(&policy->rwsem);
1385

1386
	if (cpu != policy->cpu) {
1387
		sysfs_remove_link(&dev->kobj, "cpufreq");
1388
	} else if (cpus > 1) {
1389 1390 1391
		/* Nominate new CPU */
		int new_cpu = cpumask_any_but(policy->cpus, cpu);
		struct device *cpu_dev = get_cpu_device(new_cpu);
1392

1393 1394 1395 1396 1397 1398 1399 1400
		sysfs_remove_link(&cpu_dev->kobj, "cpufreq");
		ret = update_policy_cpu(policy, new_cpu, cpu_dev);
		if (ret) {
			if (sysfs_create_link(&cpu_dev->kobj, &policy->kobj,
					      "cpufreq"))
				pr_err("%s: Failed to restore kobj link to cpu:%d\n",
				       __func__, cpu_dev->id);
			return ret;
L
Linus Torvalds 已提交
1401
		}
1402 1403 1404 1405

		if (!cpufreq_suspended)
			pr_debug("%s: policy Kobject moved to cpu: %d from: %d\n",
				 __func__, new_cpu, cpu);
1406
	} else if (cpufreq_driver->stop_cpu) {
1407
		cpufreq_driver->stop_cpu(policy);
L
Linus Torvalds 已提交
1408 1409
	}

1410 1411 1412 1413
	return 0;
}

static int __cpufreq_remove_dev_finish(struct device *dev,
1414
				       struct subsys_interface *sif)
1415 1416 1417 1418 1419 1420
{
	unsigned int cpu = dev->id, cpus;
	int ret;
	unsigned long flags;
	struct cpufreq_policy *policy;

1421
	write_lock_irqsave(&cpufreq_driver_lock, flags);
1422
	policy = per_cpu(cpufreq_cpu_data, cpu);
1423 1424
	per_cpu(cpufreq_cpu_data, cpu) = NULL;
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
1425 1426 1427 1428 1429 1430

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

1431
	down_write(&policy->rwsem);
1432
	cpus = cpumask_weight(policy->cpus);
1433
	cpumask_clear_cpu(cpu, policy->cpus);
1434
	up_write(&policy->rwsem);
1435

1436 1437
	/* If cpu is last user of policy, free policy */
	if (cpus == 1) {
1438
		if (has_target()) {
1439 1440 1441 1442
			ret = __cpufreq_governor(policy,
					CPUFREQ_GOV_POLICY_EXIT);
			if (ret) {
				pr_err("%s: Failed to exit governor\n",
1443
				       __func__);
1444 1445
				return ret;
			}
1446
		}
1447

1448
		if (!cpufreq_suspended)
1449
			cpufreq_policy_put_kobj(policy);
1450

1451 1452 1453 1454
		/*
		 * 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.
1455
		 */
1456
		if (cpufreq_driver->exit)
1457
			cpufreq_driver->exit(policy);
1458

1459 1460 1461 1462 1463
		/* Remove policy from list of active policies */
		write_lock_irqsave(&cpufreq_driver_lock, flags);
		list_del(&policy->policy_list);
		write_unlock_irqrestore(&cpufreq_driver_lock, flags);

1464
		if (!cpufreq_suspended)
1465
			cpufreq_policy_free(policy);
1466 1467 1468 1469 1470 1471 1472 1473
	} else if (has_target()) {
		ret = __cpufreq_governor(policy, CPUFREQ_GOV_START);
		if (!ret)
			ret = __cpufreq_governor(policy, CPUFREQ_GOV_LIMITS);

		if (ret) {
			pr_err("%s: Failed to start governor\n", __func__);
			return ret;
1474
		}
1475
	}
L
Linus Torvalds 已提交
1476 1477 1478 1479

	return 0;
}

1480
/**
1481
 * cpufreq_remove_dev - remove a CPU device
1482 1483 1484
 *
 * Removes the cpufreq interface for a CPU device.
 */
1485
static int cpufreq_remove_dev(struct device *dev, struct subsys_interface *sif)
1486
{
1487
	unsigned int cpu = dev->id;
1488
	int ret;
1489 1490 1491 1492

	if (cpu_is_offline(cpu))
		return 0;

1493
	ret = __cpufreq_remove_dev_prepare(dev, sif);
1494 1495

	if (!ret)
1496
		ret = __cpufreq_remove_dev_finish(dev, sif);
1497 1498

	return ret;
1499 1500
}

1501
static void handle_update(struct work_struct *work)
L
Linus Torvalds 已提交
1502
{
1503 1504 1505
	struct cpufreq_policy *policy =
		container_of(work, struct cpufreq_policy, update);
	unsigned int cpu = policy->cpu;
1506
	pr_debug("handle_update for cpu %u called\n", cpu);
L
Linus Torvalds 已提交
1507 1508 1509 1510
	cpufreq_update_policy(cpu);
}

/**
1511 1512
 *	cpufreq_out_of_sync - If actual and saved CPU frequency differs, we're
 *	in deep trouble.
1513
 *	@policy: policy managing CPUs
L
Linus Torvalds 已提交
1514 1515
 *	@new_freq: CPU frequency the CPU actually runs at
 *
1516 1517
 *	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
1518
 */
1519
static void cpufreq_out_of_sync(struct cpufreq_policy *policy,
1520
				unsigned int new_freq)
L
Linus Torvalds 已提交
1521 1522
{
	struct cpufreq_freqs freqs;
1523

1524
	pr_debug("Warning: CPU frequency out of sync: cpufreq and timing core thinks of %u, is %u kHz\n",
1525
		 policy->cur, new_freq);
L
Linus Torvalds 已提交
1526

1527
	freqs.old = policy->cur;
L
Linus Torvalds 已提交
1528
	freqs.new = new_freq;
1529

1530 1531
	cpufreq_freq_transition_begin(policy, &freqs);
	cpufreq_freq_transition_end(policy, &freqs, 0);
L
Linus Torvalds 已提交
1532 1533
}

1534
/**
D
Dhaval Giani 已提交
1535
 * cpufreq_quick_get - get the CPU frequency (in kHz) from policy->cur
1536 1537 1538 1539 1540 1541 1542
 * @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)
{
1543
	struct cpufreq_policy *policy;
1544
	unsigned int ret_freq = 0;
1545

1546 1547
	if (cpufreq_driver && cpufreq_driver->setpolicy && cpufreq_driver->get)
		return cpufreq_driver->get(cpu);
1548 1549

	policy = cpufreq_cpu_get(cpu);
1550
	if (policy) {
1551
		ret_freq = policy->cur;
1552 1553 1554
		cpufreq_cpu_put(policy);
	}

D
Dave Jones 已提交
1555
	return ret_freq;
1556 1557 1558
}
EXPORT_SYMBOL(cpufreq_quick_get);

1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578
/**
 * 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);

1579
static unsigned int __cpufreq_get(struct cpufreq_policy *policy)
L
Linus Torvalds 已提交
1580
{
1581
	unsigned int ret_freq = 0;
1582

1583
	if (!cpufreq_driver->get)
D
Dave Jones 已提交
1584
		return ret_freq;
L
Linus Torvalds 已提交
1585

1586
	ret_freq = cpufreq_driver->get(policy->cpu);
L
Linus Torvalds 已提交
1587

1588
	if (ret_freq && policy->cur &&
1589
		!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
1590 1591 1592
		/* verify no discrepancy between actual and
					saved value exists */
		if (unlikely(ret_freq != policy->cur)) {
1593
			cpufreq_out_of_sync(policy, ret_freq);
L
Linus Torvalds 已提交
1594 1595 1596 1597
			schedule_work(&policy->update);
		}
	}

D
Dave Jones 已提交
1598
	return ret_freq;
1599
}
L
Linus Torvalds 已提交
1600

1601 1602 1603 1604 1605 1606 1607 1608
/**
 * 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)
{
1609
	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
1610 1611
	unsigned int ret_freq = 0;

1612 1613
	if (policy) {
		down_read(&policy->rwsem);
1614
		ret_freq = __cpufreq_get(policy);
1615
		up_read(&policy->rwsem);
1616

1617 1618
		cpufreq_cpu_put(policy);
	}
1619

D
Dave Jones 已提交
1620
	return ret_freq;
L
Linus Torvalds 已提交
1621 1622 1623
}
EXPORT_SYMBOL(cpufreq_get);

1624 1625 1626 1627 1628
static struct subsys_interface cpufreq_interface = {
	.name		= "cpufreq",
	.subsys		= &cpu_subsys,
	.add_dev	= cpufreq_add_dev,
	.remove_dev	= cpufreq_remove_dev,
1629 1630
};

1631 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
/*
 * 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);

1657
/**
1658
 * cpufreq_suspend() - Suspend CPUFreq governors
1659
 *
1660 1661 1662 1663
 * 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.
1664
 */
1665
void cpufreq_suspend(void)
1666
{
1667
	struct cpufreq_policy *policy;
1668

1669 1670
	if (!cpufreq_driver)
		return;
1671

1672
	if (!has_target())
1673
		goto suspend;
1674

1675 1676
	pr_debug("%s: Suspending Governors\n", __func__);

V
Viresh Kumar 已提交
1677
	for_each_policy(policy) {
1678 1679 1680 1681 1682 1683 1684
		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);
1685
	}
1686 1687 1688

suspend:
	cpufreq_suspended = true;
1689 1690
}

L
Linus Torvalds 已提交
1691
/**
1692
 * cpufreq_resume() - Resume CPUFreq governors
L
Linus Torvalds 已提交
1693
 *
1694 1695
 * Called during system wide Suspend/Hibernate cycle for resuming governors that
 * are suspended with cpufreq_suspend().
L
Linus Torvalds 已提交
1696
 */
1697
void cpufreq_resume(void)
L
Linus Torvalds 已提交
1698
{
1699
	struct cpufreq_policy *policy;
L
Linus Torvalds 已提交
1700

1701 1702
	if (!cpufreq_driver)
		return;
L
Linus Torvalds 已提交
1703

1704 1705
	cpufreq_suspended = false;

1706
	if (!has_target())
1707
		return;
L
Linus Torvalds 已提交
1708

1709
	pr_debug("%s: Resuming Governors\n", __func__);
L
Linus Torvalds 已提交
1710

V
Viresh Kumar 已提交
1711
	for_each_policy(policy) {
1712 1713 1714 1715
		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)
1716 1717 1718 1719
		    || __cpufreq_governor(policy, CPUFREQ_GOV_LIMITS))
			pr_err("%s: Failed to start governor for policy: %p\n",
				__func__, policy);
	}
1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730

	/*
	 * 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);
1731
}
L
Linus Torvalds 已提交
1732

1733 1734 1735 1736 1737 1738 1739 1740
/**
 *	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)
{
1741 1742 1743 1744
	if (cpufreq_driver)
		return cpufreq_driver->name;

	return NULL;
1745 1746
}
EXPORT_SYMBOL_GPL(cpufreq_get_current_driver);
L
Linus Torvalds 已提交
1747

1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762
/**
 *	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
1763 1764 1765 1766 1767 1768 1769 1770 1771
/*********************************************************************
 *                     NOTIFIER LISTS INTERFACE                      *
 *********************************************************************/

/**
 *	cpufreq_register_notifier - register a driver with cpufreq
 *	@nb: notifier function to register
 *      @list: CPUFREQ_TRANSITION_NOTIFIER or CPUFREQ_POLICY_NOTIFIER
 *
1772
 *	Add a driver to one of two lists: either a list of drivers that
L
Linus Torvalds 已提交
1773 1774 1775 1776 1777
 *      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
1778
 *	blocking_notifier_chain_register.
L
Linus Torvalds 已提交
1779 1780 1781 1782 1783
 */
int cpufreq_register_notifier(struct notifier_block *nb, unsigned int list)
{
	int ret;

1784 1785 1786
	if (cpufreq_disabled())
		return -EINVAL;

1787 1788
	WARN_ON(!init_cpufreq_transition_notifier_list_called);

L
Linus Torvalds 已提交
1789 1790
	switch (list) {
	case CPUFREQ_TRANSITION_NOTIFIER:
1791
		ret = srcu_notifier_chain_register(
1792
				&cpufreq_transition_notifier_list, nb);
L
Linus Torvalds 已提交
1793 1794
		break;
	case CPUFREQ_POLICY_NOTIFIER:
1795 1796
		ret = blocking_notifier_chain_register(
				&cpufreq_policy_notifier_list, nb);
L
Linus Torvalds 已提交
1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808
		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
1809
 *	@list: CPUFREQ_TRANSITION_NOTIFIER or CPUFREQ_POLICY_NOTIFIER
L
Linus Torvalds 已提交
1810 1811 1812 1813
 *
 *	Remove a driver from the CPU frequency notifier list.
 *
 *	This function may sleep, and has the same return conditions as
1814
 *	blocking_notifier_chain_unregister.
L
Linus Torvalds 已提交
1815 1816 1817 1818 1819
 */
int cpufreq_unregister_notifier(struct notifier_block *nb, unsigned int list)
{
	int ret;

1820 1821 1822
	if (cpufreq_disabled())
		return -EINVAL;

L
Linus Torvalds 已提交
1823 1824
	switch (list) {
	case CPUFREQ_TRANSITION_NOTIFIER:
1825
		ret = srcu_notifier_chain_unregister(
1826
				&cpufreq_transition_notifier_list, nb);
L
Linus Torvalds 已提交
1827 1828
		break;
	case CPUFREQ_POLICY_NOTIFIER:
1829 1830
		ret = blocking_notifier_chain_unregister(
				&cpufreq_policy_notifier_list, nb);
L
Linus Torvalds 已提交
1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844
		break;
	default:
		ret = -EINVAL;
	}

	return ret;
}
EXPORT_SYMBOL(cpufreq_unregister_notifier);


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

1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870
/* 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;
}

1871 1872 1873
static int __target_index(struct cpufreq_policy *policy,
			  struct cpufreq_frequency_table *freq_table, int index)
{
1874 1875
	struct cpufreq_freqs freqs = {.old = policy->cur, .flags = 0};
	unsigned int intermediate_freq = 0;
1876 1877 1878 1879 1880
	int retval = -EINVAL;
	bool notify;

	notify = !(cpufreq_driver->flags & CPUFREQ_ASYNC_NOTIFICATION);
	if (notify) {
1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891
		/* 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;
		}
1892

1893
		freqs.new = freq_table[index].frequency;
1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904
		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);

1905
	if (notify) {
1906 1907
		cpufreq_freq_transition_end(policy, &freqs, retval);

1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921
		/*
		 * 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
		 * case we have't switched to intermediate freq at all.
		 */
		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);
		}
	}

1922 1923 1924
	return retval;
}

L
Linus Torvalds 已提交
1925 1926 1927 1928
int __cpufreq_driver_target(struct cpufreq_policy *policy,
			    unsigned int target_freq,
			    unsigned int relation)
{
1929
	unsigned int old_target_freq = target_freq;
1930
	int retval = -EINVAL;
1931

1932 1933 1934
	if (cpufreq_disabled())
		return -ENODEV;

1935 1936 1937 1938 1939 1940 1941
	/* 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",
1942
		 policy->cpu, target_freq, relation, old_target_freq);
1943

1944 1945 1946 1947 1948 1949
	/*
	 * 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.
	 */
1950 1951 1952
	if (target_freq == policy->cur)
		return 0;

1953 1954 1955
	/* Save last value to restore later on errors */
	policy->restore_freq = policy->cur;

1956 1957
	if (cpufreq_driver->target)
		retval = cpufreq_driver->target(policy, target_freq, relation);
1958 1959 1960
	else if (cpufreq_driver->target_index) {
		struct cpufreq_frequency_table *freq_table;
		int index;
1961

1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974
		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;
		}

1975
		if (freq_table[index].frequency == policy->cur) {
1976
			retval = 0;
1977 1978 1979
			goto out;
		}

1980
		retval = __target_index(policy, freq_table, index);
1981 1982 1983
	}

out:
L
Linus Torvalds 已提交
1984 1985 1986 1987 1988 1989 1990 1991
	return retval;
}
EXPORT_SYMBOL_GPL(__cpufreq_driver_target);

int cpufreq_driver_target(struct cpufreq_policy *policy,
			  unsigned int target_freq,
			  unsigned int relation)
{
1992
	int ret = -EINVAL;
L
Linus Torvalds 已提交
1993

1994
	down_write(&policy->rwsem);
L
Linus Torvalds 已提交
1995 1996 1997

	ret = __cpufreq_driver_target(policy, target_freq, relation);

1998
	up_write(&policy->rwsem);
L
Linus Torvalds 已提交
1999 2000 2001 2002 2003

	return ret;
}
EXPORT_SYMBOL_GPL(cpufreq_driver_target);

2004 2005
static int __cpufreq_governor(struct cpufreq_policy *policy,
					unsigned int event)
L
Linus Torvalds 已提交
2006
{
2007
	int ret;
2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

	/* 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
2018

2019 2020 2021
	/* Don't start any governor operations if we are entering suspend */
	if (cpufreq_suspended)
		return 0;
2022 2023 2024 2025 2026 2027
	/*
	 * Governor might not be initiated here if ACPI _PPC changed
	 * notification happened, so check it.
	 */
	if (!policy->governor)
		return -EINVAL;
2028

2029 2030 2031
	if (policy->governor->max_transition_latency &&
	    policy->cpuinfo.transition_latency >
	    policy->governor->max_transition_latency) {
2032 2033 2034
		if (!gov)
			return -EINVAL;
		else {
2035 2036
			pr_warn("%s governor failed, too long transition latency of HW, fallback to %s governor\n",
				policy->governor->name, gov->name);
2037 2038
			policy->governor = gov;
		}
2039
	}
L
Linus Torvalds 已提交
2040

2041 2042 2043
	if (event == CPUFREQ_GOV_POLICY_INIT)
		if (!try_module_get(policy->governor->owner))
			return -EINVAL;
L
Linus Torvalds 已提交
2044

2045
	pr_debug("__cpufreq_governor for CPU %u, event %u\n",
2046
		 policy->cpu, event);
2047 2048

	mutex_lock(&cpufreq_governor_lock);
2049
	if ((policy->governor_enabled && event == CPUFREQ_GOV_START)
2050 2051
	    || (!policy->governor_enabled
	    && (event == CPUFREQ_GOV_LIMITS || event == CPUFREQ_GOV_STOP))) {
2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062
		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
2063 2064
	ret = policy->governor->governor(policy, event);

2065 2066 2067 2068 2069
	if (!ret) {
		if (event == CPUFREQ_GOV_POLICY_INIT)
			policy->governor->initialized++;
		else if (event == CPUFREQ_GOV_POLICY_EXIT)
			policy->governor->initialized--;
2070 2071 2072 2073 2074 2075 2076 2077
	} 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);
2078
	}
2079

2080 2081
	if (((event == CPUFREQ_GOV_POLICY_INIT) && ret) ||
			((event == CPUFREQ_GOV_POLICY_EXIT) && !ret))
L
Linus Torvalds 已提交
2082 2083 2084 2085 2086 2087 2088
		module_put(policy->governor->owner);

	return ret;
}

int cpufreq_register_governor(struct cpufreq_governor *governor)
{
2089
	int err;
L
Linus Torvalds 已提交
2090 2091 2092 2093

	if (!governor)
		return -EINVAL;

2094 2095 2096
	if (cpufreq_disabled())
		return -ENODEV;

2097
	mutex_lock(&cpufreq_governor_mutex);
2098

2099
	governor->initialized = 0;
2100
	err = -EBUSY;
2101
	if (!find_governor(governor->name)) {
2102 2103
		err = 0;
		list_add(&governor->governor_list, &cpufreq_governor_list);
L
Linus Torvalds 已提交
2104 2105
	}

2106
	mutex_unlock(&cpufreq_governor_mutex);
2107
	return err;
L
Linus Torvalds 已提交
2108 2109 2110 2111 2112
}
EXPORT_SYMBOL_GPL(cpufreq_register_governor);

void cpufreq_unregister_governor(struct cpufreq_governor *governor)
{
2113 2114
	int cpu;

L
Linus Torvalds 已提交
2115 2116 2117
	if (!governor)
		return;

2118 2119 2120
	if (cpufreq_disabled())
		return;

2121 2122 2123 2124 2125 2126 2127
	for_each_present_cpu(cpu) {
		if (cpu_online(cpu))
			continue;
		if (!strcmp(per_cpu(cpufreq_cpu_governor, cpu), governor->name))
			strcpy(per_cpu(cpufreq_cpu_governor, cpu), "\0");
	}

2128
	mutex_lock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
2129
	list_del(&governor->governor_list);
2130
	mutex_unlock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141
	return;
}
EXPORT_SYMBOL_GPL(cpufreq_unregister_governor);


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

/**
 * cpufreq_get_policy - get the current cpufreq_policy
2142 2143
 * @policy: struct cpufreq_policy into which the current cpufreq_policy
 *	is written
L
Linus Torvalds 已提交
2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156
 *
 * 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;

2157
	memcpy(policy, cpu_policy, sizeof(*policy));
L
Linus Torvalds 已提交
2158 2159 2160 2161 2162 2163

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

2164
/*
2165 2166
 * policy : current policy.
 * new_policy: policy to be set.
2167
 */
2168
static int cpufreq_set_policy(struct cpufreq_policy *policy,
2169
				struct cpufreq_policy *new_policy)
L
Linus Torvalds 已提交
2170
{
2171 2172
	struct cpufreq_governor *old_gov;
	int ret;
L
Linus Torvalds 已提交
2173

2174 2175
	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
2176

2177
	memcpy(&new_policy->cpuinfo, &policy->cpuinfo, sizeof(policy->cpuinfo));
L
Linus Torvalds 已提交
2178

2179 2180
	if (new_policy->min > policy->max || new_policy->max < policy->min)
		return -EINVAL;
2181

L
Linus Torvalds 已提交
2182
	/* verify the cpu speed can be set within this limit */
2183
	ret = cpufreq_driver->verify(new_policy);
L
Linus Torvalds 已提交
2184
	if (ret)
2185
		return ret;
L
Linus Torvalds 已提交
2186 2187

	/* adjust if necessary - all reasons */
2188
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2189
			CPUFREQ_ADJUST, new_policy);
L
Linus Torvalds 已提交
2190 2191

	/* adjust if necessary - hardware incompatibility*/
2192
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2193
			CPUFREQ_INCOMPATIBLE, new_policy);
L
Linus Torvalds 已提交
2194

2195 2196 2197 2198
	/*
	 * verify the cpu speed can be set within this limit, which might be
	 * different to the first one
	 */
2199
	ret = cpufreq_driver->verify(new_policy);
2200
	if (ret)
2201
		return ret;
L
Linus Torvalds 已提交
2202 2203

	/* notification of the new policy */
2204
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2205
			CPUFREQ_NOTIFY, new_policy);
L
Linus Torvalds 已提交
2206

2207 2208
	policy->min = new_policy->min;
	policy->max = new_policy->max;
L
Linus Torvalds 已提交
2209

2210
	pr_debug("new min and max freqs are %u - %u kHz\n",
2211
		 policy->min, policy->max);
L
Linus Torvalds 已提交
2212

2213
	if (cpufreq_driver->setpolicy) {
2214
		policy->policy = new_policy->policy;
2215
		pr_debug("setting range\n");
2216 2217
		return cpufreq_driver->setpolicy(new_policy);
	}
L
Linus Torvalds 已提交
2218

2219 2220
	if (new_policy->governor == policy->governor)
		goto out;
2221

2222 2223 2224 2225 2226 2227 2228 2229
	pr_debug("governor switch\n");

	/* save old, working values */
	old_gov = policy->governor;
	/* end old governor */
	if (old_gov) {
		__cpufreq_governor(policy, CPUFREQ_GOV_STOP);
		up_write(&policy->rwsem);
2230
		__cpufreq_governor(policy, CPUFREQ_GOV_POLICY_EXIT);
2231
		down_write(&policy->rwsem);
L
Linus Torvalds 已提交
2232 2233
	}

2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257
	/* start new governor */
	policy->governor = new_policy->governor;
	if (!__cpufreq_governor(policy, CPUFREQ_GOV_POLICY_INIT)) {
		if (!__cpufreq_governor(policy, CPUFREQ_GOV_START))
			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;
		__cpufreq_governor(policy, CPUFREQ_GOV_POLICY_INIT);
		__cpufreq_governor(policy, CPUFREQ_GOV_START);
	}

	return -EINVAL;

 out:
	pr_debug("governor: change or update limits\n");
	return __cpufreq_governor(policy, CPUFREQ_GOV_LIMITS);
L
Linus Torvalds 已提交
2258 2259 2260 2261 2262 2263
}

/**
 *	cpufreq_update_policy - re-evaluate an existing cpufreq policy
 *	@cpu: CPU which shall be re-evaluated
 *
L
Lucas De Marchi 已提交
2264
 *	Useful for policy notifiers which have different necessities
L
Linus Torvalds 已提交
2265 2266 2267 2268
 *	at different times.
 */
int cpufreq_update_policy(unsigned int cpu)
{
2269 2270
	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
	struct cpufreq_policy new_policy;
2271
	int ret;
L
Linus Torvalds 已提交
2272

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

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

2278
	pr_debug("updating policy for CPU %u\n", cpu);
2279
	memcpy(&new_policy, policy, sizeof(*policy));
2280 2281 2282 2283
	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
2284

2285 2286 2287 2288
	/*
	 * BIOS might change freq behind our back
	 * -> ask driver for current freq and notify governors about a change
	 */
2289
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
2290
		new_policy.cur = cpufreq_driver->get(cpu);
2291 2292
		if (WARN_ON(!new_policy.cur)) {
			ret = -EIO;
2293
			goto unlock;
2294 2295
		}

2296
		if (!policy->cur) {
2297
			pr_debug("Driver did not initialize current freq\n");
2298
			policy->cur = new_policy.cur;
2299
		} else {
2300
			if (policy->cur != new_policy.cur && has_target())
2301
				cpufreq_out_of_sync(policy, new_policy.cur);
2302
		}
2303 2304
	}

2305
	ret = cpufreq_set_policy(policy, &new_policy);
L
Linus Torvalds 已提交
2306

2307
unlock:
2308
	up_write(&policy->rwsem);
2309

2310
	cpufreq_cpu_put(policy);
L
Linus Torvalds 已提交
2311 2312 2313 2314
	return ret;
}
EXPORT_SYMBOL(cpufreq_update_policy);

2315
static int cpufreq_cpu_callback(struct notifier_block *nfb,
2316 2317 2318
					unsigned long action, void *hcpu)
{
	unsigned int cpu = (unsigned long)hcpu;
2319
	struct device *dev;
2320

2321 2322
	dev = get_cpu_device(cpu);
	if (dev) {
2323
		switch (action & ~CPU_TASKS_FROZEN) {
2324
		case CPU_ONLINE:
2325
			cpufreq_add_dev(dev, NULL);
2326
			break;
2327

2328
		case CPU_DOWN_PREPARE:
2329
			__cpufreq_remove_dev_prepare(dev, NULL);
2330 2331 2332
			break;

		case CPU_POST_DEAD:
2333
			__cpufreq_remove_dev_finish(dev, NULL);
2334
			break;
2335

2336
		case CPU_DOWN_FAILED:
2337
			cpufreq_add_dev(dev, NULL);
2338 2339 2340 2341 2342 2343
			break;
		}
	}
	return NOTIFY_OK;
}

2344
static struct notifier_block __refdata cpufreq_cpu_notifier = {
2345
	.notifier_call = cpufreq_cpu_callback,
2346
};
L
Linus Torvalds 已提交
2347

2348 2349 2350 2351 2352 2353 2354 2355 2356
/*********************************************************************
 *               BOOST						     *
 *********************************************************************/
static int cpufreq_boost_set_sw(int state)
{
	struct cpufreq_frequency_table *freq_table;
	struct cpufreq_policy *policy;
	int ret = -EINVAL;

V
Viresh Kumar 已提交
2357
	for_each_policy(policy) {
2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392
		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);

2393 2394
		pr_err("%s: Cannot %s BOOST\n",
		       __func__, state ? "enable" : "disable");
2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414
	}

	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
2415 2416 2417 2418 2419 2420 2421 2422 2423
/*********************************************************************
 *               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.
 *
2424
 * Registers a CPU Frequency driver to this core code. This code
L
Linus Torvalds 已提交
2425
 * returns zero on success, -EBUSY when another driver got here first
2426
 * (and isn't unregistered in the meantime).
L
Linus Torvalds 已提交
2427 2428
 *
 */
2429
int cpufreq_register_driver(struct cpufreq_driver *driver_data)
L
Linus Torvalds 已提交
2430 2431 2432 2433
{
	unsigned long flags;
	int ret;

2434 2435 2436
	if (cpufreq_disabled())
		return -ENODEV;

L
Linus Torvalds 已提交
2437
	if (!driver_data || !driver_data->verify || !driver_data->init ||
2438
	    !(driver_data->setpolicy || driver_data->target_index ||
2439 2440
		    driver_data->target) ||
	     (driver_data->setpolicy && (driver_data->target_index ||
2441 2442
		    driver_data->target)) ||
	     (!!driver_data->get_intermediate != !!driver_data->target_intermediate))
L
Linus Torvalds 已提交
2443 2444
		return -EINVAL;

2445
	pr_debug("trying to register driver %s\n", driver_data->name);
L
Linus Torvalds 已提交
2446

2447
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2448
	if (cpufreq_driver) {
2449
		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2450
		return -EEXIST;
L
Linus Torvalds 已提交
2451
	}
2452
	cpufreq_driver = driver_data;
2453
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
2454

2455 2456 2457
	if (driver_data->setpolicy)
		driver_data->flags |= CPUFREQ_CONST_LOOPS;

2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468
	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",
2469
			       __func__);
2470 2471 2472 2473
			goto err_null_driver;
		}
	}

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 2482 2483
		pr_debug("%s: No CPU initialized for driver %s\n", __func__,
			 driver_data->name);
		goto err_if_unreg;
L
Linus Torvalds 已提交
2484 2485
	}

2486
	register_hotcpu_notifier(&cpufreq_cpu_notifier);
2487
	pr_debug("driver %s up and running\n", driver_data->name);
L
Linus Torvalds 已提交
2488

2489
	return 0;
2490 2491
err_if_unreg:
	subsys_interface_unregister(&cpufreq_interface);
2492 2493 2494
err_boost_unreg:
	if (cpufreq_boost_supported())
		cpufreq_sysfs_remove_file(&boost.attr);
2495
err_null_driver:
2496
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2497
	cpufreq_driver = NULL;
2498
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
D
Dave Jones 已提交
2499
	return ret;
L
Linus Torvalds 已提交
2500 2501 2502 2503 2504 2505
}
EXPORT_SYMBOL_GPL(cpufreq_register_driver);

/**
 * cpufreq_unregister_driver - unregister the current CPUFreq driver
 *
2506
 * Unregister the current CPUFreq driver. Only call this if you have
L
Linus Torvalds 已提交
2507 2508 2509 2510
 * 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.
 */
2511
int cpufreq_unregister_driver(struct cpufreq_driver *driver)
L
Linus Torvalds 已提交
2512 2513 2514
{
	unsigned long flags;

2515
	if (!cpufreq_driver || (driver != cpufreq_driver))
L
Linus Torvalds 已提交
2516 2517
		return -EINVAL;

2518
	pr_debug("unregistering driver %s\n", driver->name);
L
Linus Torvalds 已提交
2519

2520
	subsys_interface_unregister(&cpufreq_interface);
2521 2522 2523
	if (cpufreq_boost_supported())
		cpufreq_sysfs_remove_file(&boost.attr);

2524
	unregister_hotcpu_notifier(&cpufreq_cpu_notifier);
L
Linus Torvalds 已提交
2525

2526
	down_write(&cpufreq_rwsem);
2527
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2528

2529
	cpufreq_driver = NULL;
2530

2531
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2532
	up_write(&cpufreq_rwsem);
L
Linus Torvalds 已提交
2533 2534 2535 2536

	return 0;
}
EXPORT_SYMBOL_GPL(cpufreq_unregister_driver);
2537

2538 2539 2540 2541 2542 2543 2544 2545
/*
 * 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,
};

2546 2547
static int __init cpufreq_core_init(void)
{
2548 2549 2550
	if (cpufreq_disabled())
		return -ENODEV;

2551
	cpufreq_global_kobject = kobject_create();
2552 2553
	BUG_ON(!cpufreq_global_kobject);

2554 2555
	register_syscore_ops(&cpufreq_syscore_ops);

2556 2557 2558
	return 0;
}
core_initcall(cpufreq_core_init);