cpufreq.c 65.0 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/tick.h>
31 32
#include <trace/events/power.h>

L
Linus Torvalds 已提交
33
/**
D
Dave Jones 已提交
34
 * The "cpufreq driver" - the arch- or hardware-dependent low
L
Linus Torvalds 已提交
35 36 37
 * level driver of CPUFreq support, and its spinlock. This lock
 * also protects the cpufreq_cpu_data array.
 */
38
static struct cpufreq_driver *cpufreq_driver;
39
static DEFINE_PER_CPU(struct cpufreq_policy *, cpufreq_cpu_data);
40
static DEFINE_PER_CPU(struct cpufreq_policy *, cpufreq_cpu_data_fallback);
41
static DEFINE_RWLOCK(cpufreq_driver_lock);
42
DEFINE_MUTEX(cpufreq_governor_lock);
43
static LIST_HEAD(cpufreq_policy_list);
44

45
/* This one keeps track of the previously set governor of a removed CPU */
46
static DEFINE_PER_CPU(char[CPUFREQ_NAME_LEN], cpufreq_cpu_governor);
L
Linus Torvalds 已提交
47

48 49
/* Flag to suspend/resume CPUFreq governors */
static bool cpufreq_suspended;
L
Linus Torvalds 已提交
50

51 52 53 54 55
static inline bool has_target(void)
{
	return cpufreq_driver->target_index || cpufreq_driver->target;
}

56 57 58 59 60 61
/*
 * rwsem to guarantee that cpufreq driver module doesn't unload during critical
 * sections
 */
static DECLARE_RWSEM(cpufreq_rwsem);

L
Linus Torvalds 已提交
62
/* internal prototypes */
63 64
static int __cpufreq_governor(struct cpufreq_policy *policy,
		unsigned int event);
65
static unsigned int __cpufreq_get(unsigned int cpu);
66
static void handle_update(struct work_struct *work);
L
Linus Torvalds 已提交
67 68

/**
69 70
 * 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
71 72 73 74
 * "transition" list for kernel code that needs to handle
 * changes to devices when the CPU clock speed changes.
 * The mutex locks both lists.
 */
75
static BLOCKING_NOTIFIER_HEAD(cpufreq_policy_notifier_list);
76
static struct srcu_notifier_head cpufreq_transition_notifier_list;
L
Linus Torvalds 已提交
77

78
static bool init_cpufreq_transition_notifier_list_called;
79 80 81
static int __init init_cpufreq_transition_notifier_list(void)
{
	srcu_init_notifier_head(&cpufreq_transition_notifier_list);
82
	init_cpufreq_transition_notifier_list_called = true;
83 84
	return 0;
}
85
pure_initcall(init_cpufreq_transition_notifier_list);
L
Linus Torvalds 已提交
86

87
static int off __read_mostly;
88
static int cpufreq_disabled(void)
89 90 91 92 93 94 95
{
	return off;
}
void disable_cpufreq(void)
{
	off = 1;
}
L
Linus Torvalds 已提交
96
static LIST_HEAD(cpufreq_governor_list);
97
static DEFINE_MUTEX(cpufreq_governor_mutex);
L
Linus Torvalds 已提交
98

99 100
bool have_governor_per_policy(void)
{
101
	return !!(cpufreq_driver->flags & CPUFREQ_HAVE_GOVERNOR_PER_POLICY);
102
}
103
EXPORT_SYMBOL_GPL(have_governor_per_policy);
104

105 106 107 108 109 110 111 112 113
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);

114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148
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);

149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179
/*
 * 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);

180 181 182 183 184
unsigned int cpufreq_generic_get(unsigned int cpu)
{
	struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);

	if (!policy || IS_ERR(policy->clk)) {
185 186
		pr_err("%s: No %s associated to cpu: %d\n",
		       __func__, policy ? "clk" : "policy", cpu);
187 188 189 190 191 192 193
		return 0;
	}

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

194 195 196 197 198 199
/* Only for cpufreq core internal use */
struct cpufreq_policy *cpufreq_cpu_get_raw(unsigned int cpu)
{
	return per_cpu(cpufreq_cpu_data, cpu);
}

200
struct cpufreq_policy *cpufreq_cpu_get(unsigned int cpu)
L
Linus Torvalds 已提交
201
{
202
	struct cpufreq_policy *policy = NULL;
L
Linus Torvalds 已提交
203 204
	unsigned long flags;

205 206 207 208 209
	if (cpufreq_disabled() || (cpu >= nr_cpu_ids))
		return NULL;

	if (!down_read_trylock(&cpufreq_rwsem))
		return NULL;
L
Linus Torvalds 已提交
210 211

	/* get the cpufreq driver */
212
	read_lock_irqsave(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
213

214 215 216 217 218 219
	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
220

221
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
222

223
	if (!policy)
224
		up_read(&cpufreq_rwsem);
L
Linus Torvalds 已提交
225

226
	return policy;
227
}
L
Linus Torvalds 已提交
228 229
EXPORT_SYMBOL_GPL(cpufreq_cpu_get);

230
void cpufreq_cpu_put(struct cpufreq_policy *policy)
L
Linus Torvalds 已提交
231
{
232 233 234
	if (cpufreq_disabled())
		return;

235 236
	kobject_put(&policy->kobj);
	up_read(&cpufreq_rwsem);
L
Linus Torvalds 已提交
237 238 239 240 241 242 243 244 245 246 247 248
}
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
249
 * systems as each CPU might be scaled differently. So, use the arch
L
Linus Torvalds 已提交
250 251 252 253
 * per-CPU loops_per_jiffy value wherever possible.
 */
#ifndef CONFIG_SMP
static unsigned long l_p_j_ref;
254
static unsigned int l_p_j_ref_freq;
L
Linus Torvalds 已提交
255

256
static void adjust_jiffies(unsigned long val, struct cpufreq_freqs *ci)
L
Linus Torvalds 已提交
257 258 259 260 261 262 263
{
	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;
264 265
		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
266
	}
267
	if (val == CPUFREQ_POSTCHANGE && ci->old != ci->new) {
268 269
		loops_per_jiffy = cpufreq_scale(l_p_j_ref, l_p_j_ref_freq,
								ci->new);
270 271
		pr_debug("scaling loops_per_jiffy to %lu for frequency %u kHz\n",
			 loops_per_jiffy, ci->new);
L
Linus Torvalds 已提交
272 273 274
	}
}
#else
275 276 277 278
static inline void adjust_jiffies(unsigned long val, struct cpufreq_freqs *ci)
{
	return;
}
L
Linus Torvalds 已提交
279 280
#endif

281
static void __cpufreq_notify_transition(struct cpufreq_policy *policy,
282
		struct cpufreq_freqs *freqs, unsigned int state)
L
Linus Torvalds 已提交
283 284 285
{
	BUG_ON(irqs_disabled());

286 287 288
	if (cpufreq_disabled())
		return;

289
	freqs->flags = cpufreq_driver->flags;
290
	pr_debug("notification %u of frequency transition to %u kHz\n",
291
		 state, freqs->new);
L
Linus Torvalds 已提交
292 293

	switch (state) {
294

L
Linus Torvalds 已提交
295
	case CPUFREQ_PRECHANGE:
296
		/* detect if the driver reported a value as "old frequency"
297 298
		 * which is not equal to what the cpufreq core thinks is
		 * "old frequency".
L
Linus Torvalds 已提交
299
		 */
300
		if (!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
301 302
			if ((policy) && (policy->cpu == freqs->cpu) &&
			    (policy->cur) && (policy->cur != freqs->old)) {
303 304
				pr_debug("Warning: CPU frequency is %u, cpufreq assumed %u kHz\n",
					 freqs->old, policy->cur);
305
				freqs->old = policy->cur;
L
Linus Torvalds 已提交
306 307
			}
		}
308
		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
309
				CPUFREQ_PRECHANGE, freqs);
L
Linus Torvalds 已提交
310 311
		adjust_jiffies(CPUFREQ_PRECHANGE, freqs);
		break;
312

L
Linus Torvalds 已提交
313 314
	case CPUFREQ_POSTCHANGE:
		adjust_jiffies(CPUFREQ_POSTCHANGE, freqs);
315 316
		pr_debug("FREQ: %lu - CPU: %lu\n",
			 (unsigned long)freqs->new, (unsigned long)freqs->cpu);
317
		trace_cpu_frequency(freqs->new, freqs->cpu);
318
		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
319
				CPUFREQ_POSTCHANGE, freqs);
320 321
		if (likely(policy) && likely(policy->cpu == freqs->cpu))
			policy->cur = freqs->new;
L
Linus Torvalds 已提交
322 323 324
		break;
	}
}
325

326 327 328 329 330 331 332 333
/**
 * 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.
 */
334
static void cpufreq_notify_transition(struct cpufreq_policy *policy,
335 336 337 338 339
		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
340

341
/* Do post notifications when there are chances that transition has failed */
342
static void cpufreq_notify_post_transition(struct cpufreq_policy *policy,
343 344 345 346 347 348 349 350 351 352 353
		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);
}

354 355 356
void cpufreq_freq_transition_begin(struct cpufreq_policy *policy,
		struct cpufreq_freqs *freqs)
{
357 358 359 360 361 362 363 364 365 366 367 368

	/*
	 * 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);

369 370 371 372 373 374 375 376 377 378 379
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;
380
	policy->transition_task = current;
381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396

	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;
397
	policy->transition_task = NULL;
398 399 400 401 402

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

L
Linus Torvalds 已提交
403 404 405 406

/*********************************************************************
 *                          SYSFS INTERFACE                          *
 *********************************************************************/
407
static ssize_t show_boost(struct kobject *kobj,
408 409 410 411 412 413 414 415 416 417 418 419 420 421 422
				 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)) {
423 424
		pr_err("%s: Cannot %s BOOST!\n",
		       __func__, enable ? "enable" : "disable");
425 426 427
		return -EINVAL;
	}

428 429
	pr_debug("%s: cpufreq BOOST %s\n",
		 __func__, enable ? "enabled" : "disabled");
430 431 432 433

	return count;
}
define_one_global_rw(boost);
L
Linus Torvalds 已提交
434

435 436 437 438 439
static struct cpufreq_governor *__find_governor(const char *str_governor)
{
	struct cpufreq_governor *t;

	list_for_each_entry(t, &cpufreq_governor_list, governor_list)
440
		if (!strncasecmp(str_governor, t->name, CPUFREQ_NAME_LEN))
441 442 443 444 445
			return t;

	return NULL;
}

L
Linus Torvalds 已提交
446 447 448
/**
 * cpufreq_parse_governor - parse a governor string
 */
D
Dave Jones 已提交
449
static int cpufreq_parse_governor(char *str_governor, unsigned int *policy,
L
Linus Torvalds 已提交
450 451
				struct cpufreq_governor **governor)
{
452
	int err = -EINVAL;
453 454

	if (!cpufreq_driver)
455 456
		goto out;

457
	if (cpufreq_driver->setpolicy) {
458
		if (!strncasecmp(str_governor, "performance", CPUFREQ_NAME_LEN)) {
L
Linus Torvalds 已提交
459
			*policy = CPUFREQ_POLICY_PERFORMANCE;
460
			err = 0;
461
		} else if (!strncasecmp(str_governor, "powersave",
462
						CPUFREQ_NAME_LEN)) {
L
Linus Torvalds 已提交
463
			*policy = CPUFREQ_POLICY_POWERSAVE;
464
			err = 0;
L
Linus Torvalds 已提交
465
		}
466
	} else if (has_target()) {
L
Linus Torvalds 已提交
467
		struct cpufreq_governor *t;
468

469
		mutex_lock(&cpufreq_governor_mutex);
470 471 472

		t = __find_governor(str_governor);

473
		if (t == NULL) {
474
			int ret;
475

476 477 478
			mutex_unlock(&cpufreq_governor_mutex);
			ret = request_module("cpufreq_%s", str_governor);
			mutex_lock(&cpufreq_governor_mutex);
479

480 481
			if (ret == 0)
				t = __find_governor(str_governor);
482 483
		}

484 485 486
		if (t != NULL) {
			*governor = t;
			err = 0;
L
Linus Torvalds 已提交
487
		}
488

489
		mutex_unlock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
490
	}
491
out:
492
	return err;
L
Linus Torvalds 已提交
493 494 495
}

/**
496 497
 * cpufreq_per_cpu_attr_read() / show_##file_name() -
 * print out cpufreq information
L
Linus Torvalds 已提交
498 499 500 501 502
 *
 * Write out information from cpufreq_driver->policy[cpu]; object must be
 * "unsigned int".
 */

503 504
#define show_one(file_name, object)			\
static ssize_t show_##file_name				\
D
Dave Jones 已提交
505
(struct cpufreq_policy *policy, char *buf)		\
506
{							\
507
	return sprintf(buf, "%u\n", policy->object);	\
L
Linus Torvalds 已提交
508 509 510 511
}

show_one(cpuinfo_min_freq, cpuinfo.min_freq);
show_one(cpuinfo_max_freq, cpuinfo.max_freq);
512
show_one(cpuinfo_transition_latency, cpuinfo.transition_latency);
L
Linus Torvalds 已提交
513 514
show_one(scaling_min_freq, min);
show_one(scaling_max_freq, max);
515 516 517 518 519 520 521 522 523 524 525 526

static ssize_t show_scaling_cur_freq(
	struct cpufreq_policy *policy, char *buf)
{
	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
527

528
static int cpufreq_set_policy(struct cpufreq_policy *policy,
529
				struct cpufreq_policy *new_policy);
530

L
Linus Torvalds 已提交
531 532 533 534 535
/**
 * 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
536
(struct cpufreq_policy *policy, const char *buf, size_t count)		\
L
Linus Torvalds 已提交
537
{									\
538
	int ret, temp;							\
L
Linus Torvalds 已提交
539 540 541 542 543 544
	struct cpufreq_policy new_policy;				\
									\
	ret = cpufreq_get_policy(&new_policy, policy->cpu);		\
	if (ret)							\
		return -EINVAL;						\
									\
545
	ret = sscanf(buf, "%u", &new_policy.object);			\
L
Linus Torvalds 已提交
546 547 548
	if (ret != 1)							\
		return -EINVAL;						\
									\
549
	temp = new_policy.object;					\
550
	ret = cpufreq_set_policy(policy, &new_policy);		\
551 552
	if (!ret)							\
		policy->user_policy.object = temp;			\
L
Linus Torvalds 已提交
553 554 555 556
									\
	return ret ? ret : count;					\
}

557 558
store_one(scaling_min_freq, min);
store_one(scaling_max_freq, max);
L
Linus Torvalds 已提交
559 560 561 562

/**
 * show_cpuinfo_cur_freq - current CPU frequency as detected by hardware
 */
D
Dave Jones 已提交
563 564
static ssize_t show_cpuinfo_cur_freq(struct cpufreq_policy *policy,
					char *buf)
L
Linus Torvalds 已提交
565
{
566
	unsigned int cur_freq = __cpufreq_get(policy->cpu);
L
Linus Torvalds 已提交
567 568 569 570 571 572 573 574
	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
575
static ssize_t show_scaling_governor(struct cpufreq_policy *policy, char *buf)
L
Linus Torvalds 已提交
576
{
577
	if (policy->policy == CPUFREQ_POLICY_POWERSAVE)
L
Linus Torvalds 已提交
578 579 580 581
		return sprintf(buf, "powersave\n");
	else if (policy->policy == CPUFREQ_POLICY_PERFORMANCE)
		return sprintf(buf, "performance\n");
	else if (policy->governor)
582
		return scnprintf(buf, CPUFREQ_NAME_PLEN, "%s\n",
583
				policy->governor->name);
L
Linus Torvalds 已提交
584 585 586 587 588 589
	return -EINVAL;
}

/**
 * store_scaling_governor - store policy for the specified CPU
 */
D
Dave Jones 已提交
590 591
static ssize_t store_scaling_governor(struct cpufreq_policy *policy,
					const char *buf, size_t count)
L
Linus Torvalds 已提交
592
{
593
	int ret;
L
Linus Torvalds 已提交
594 595 596 597 598 599 600
	char	str_governor[16];
	struct cpufreq_policy new_policy;

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

601
	ret = sscanf(buf, "%15s", str_governor);
L
Linus Torvalds 已提交
602 603 604
	if (ret != 1)
		return -EINVAL;

605 606
	if (cpufreq_parse_governor(str_governor, &new_policy.policy,
						&new_policy.governor))
L
Linus Torvalds 已提交
607 608
		return -EINVAL;

609
	ret = cpufreq_set_policy(policy, &new_policy);
610 611 612 613

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

614 615 616 617
	if (ret)
		return ret;
	else
		return count;
L
Linus Torvalds 已提交
618 619 620 621 622
}

/**
 * show_scaling_driver - show the cpufreq driver currently loaded
 */
D
Dave Jones 已提交
623
static ssize_t show_scaling_driver(struct cpufreq_policy *policy, char *buf)
L
Linus Torvalds 已提交
624
{
625
	return scnprintf(buf, CPUFREQ_NAME_PLEN, "%s\n", cpufreq_driver->name);
L
Linus Torvalds 已提交
626 627 628 629 630
}

/**
 * show_scaling_available_governors - show the available CPUfreq governors
 */
D
Dave Jones 已提交
631 632
static ssize_t show_scaling_available_governors(struct cpufreq_policy *policy,
						char *buf)
L
Linus Torvalds 已提交
633 634 635 636
{
	ssize_t i = 0;
	struct cpufreq_governor *t;

637
	if (!has_target()) {
L
Linus Torvalds 已提交
638 639 640 641 642
		i += sprintf(buf, "performance powersave");
		goto out;
	}

	list_for_each_entry(t, &cpufreq_governor_list, governor_list) {
643 644
		if (i >= (ssize_t) ((PAGE_SIZE / sizeof(char))
		    - (CPUFREQ_NAME_LEN + 2)))
L
Linus Torvalds 已提交
645
			goto out;
646
		i += scnprintf(&buf[i], CPUFREQ_NAME_PLEN, "%s ", t->name);
L
Linus Torvalds 已提交
647
	}
648
out:
L
Linus Torvalds 已提交
649 650 651
	i += sprintf(&buf[i], "\n");
	return i;
}
652

653
ssize_t cpufreq_show_cpus(const struct cpumask *mask, char *buf)
L
Linus Torvalds 已提交
654 655 656 657
{
	ssize_t i = 0;
	unsigned int cpu;

658
	for_each_cpu(cpu, mask) {
L
Linus Torvalds 已提交
659 660 661 662
		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))
663
			break;
L
Linus Torvalds 已提交
664 665 666 667
	}
	i += sprintf(&buf[i], "\n");
	return i;
}
668
EXPORT_SYMBOL_GPL(cpufreq_show_cpus);
L
Linus Torvalds 已提交
669

670 671 672 673 674 675
/**
 * 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)
{
676
	return cpufreq_show_cpus(policy->related_cpus, buf);
677 678 679 680 681 682 683
}

/**
 * show_affected_cpus - show the CPUs affected by each transition
 */
static ssize_t show_affected_cpus(struct cpufreq_policy *policy, char *buf)
{
684
	return cpufreq_show_cpus(policy->cpus, buf);
685 686
}

687
static ssize_t store_scaling_setspeed(struct cpufreq_policy *policy,
D
Dave Jones 已提交
688
					const char *buf, size_t count)
689 690 691 692
{
	unsigned int freq = 0;
	unsigned int ret;

693
	if (!policy->governor || !policy->governor->store_setspeed)
694 695 696 697 698 699 700 701 702 703 704 705 706
		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)
{
707
	if (!policy->governor || !policy->governor->show_setspeed)
708 709 710 711
		return sprintf(buf, "<unsupported>\n");

	return policy->governor->show_setspeed(policy, buf);
}
L
Linus Torvalds 已提交
712

713
/**
714
 * show_bios_limit - show the current cpufreq HW/BIOS limitation
715 716 717 718 719
 */
static ssize_t show_bios_limit(struct cpufreq_policy *policy, char *buf)
{
	unsigned int limit;
	int ret;
720 721
	if (cpufreq_driver->bios_limit) {
		ret = cpufreq_driver->bios_limit(policy->cpu, &limit);
722 723 724 725 726 727
		if (!ret)
			return sprintf(buf, "%u\n", limit);
	}
	return sprintf(buf, "%u\n", policy->cpuinfo.max_freq);
}

728 729 730 731 732 733 734 735 736 737 738 739 740 741
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
742

D
Dave Jones 已提交
743
static struct attribute *default_attrs[] = {
L
Linus Torvalds 已提交
744 745
	&cpuinfo_min_freq.attr,
	&cpuinfo_max_freq.attr,
746
	&cpuinfo_transition_latency.attr,
L
Linus Torvalds 已提交
747 748 749
	&scaling_min_freq.attr,
	&scaling_max_freq.attr,
	&affected_cpus.attr,
750
	&related_cpus.attr,
L
Linus Torvalds 已提交
751 752 753
	&scaling_governor.attr,
	&scaling_driver.attr,
	&scaling_available_governors.attr,
754
	&scaling_setspeed.attr,
L
Linus Torvalds 已提交
755 756 757
	NULL
};

758 759
#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
760

761
static ssize_t show(struct kobject *kobj, struct attribute *attr, char *buf)
L
Linus Torvalds 已提交
762
{
D
Dave Jones 已提交
763 764
	struct cpufreq_policy *policy = to_policy(kobj);
	struct freq_attr *fattr = to_attr(attr);
765
	ssize_t ret;
766 767

	if (!down_read_trylock(&cpufreq_rwsem))
768
		return -EINVAL;
769

770
	down_read(&policy->rwsem);
771

772 773 774 775 776
	if (fattr->show)
		ret = fattr->show(policy, buf);
	else
		ret = -EIO;

777
	up_read(&policy->rwsem);
778
	up_read(&cpufreq_rwsem);
779

L
Linus Torvalds 已提交
780 781 782
	return ret;
}

D
Dave Jones 已提交
783 784
static ssize_t store(struct kobject *kobj, struct attribute *attr,
		     const char *buf, size_t count)
L
Linus Torvalds 已提交
785
{
D
Dave Jones 已提交
786 787
	struct cpufreq_policy *policy = to_policy(kobj);
	struct freq_attr *fattr = to_attr(attr);
788
	ssize_t ret = -EINVAL;
789

790 791 792 793 794
	get_online_cpus();

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

795
	if (!down_read_trylock(&cpufreq_rwsem))
796
		goto unlock;
797

798
	down_write(&policy->rwsem);
799

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

805
	up_write(&policy->rwsem);
806 807

	up_read(&cpufreq_rwsem);
808 809 810
unlock:
	put_online_cpus();

L
Linus Torvalds 已提交
811 812 813
	return ret;
}

D
Dave Jones 已提交
814
static void cpufreq_sysfs_release(struct kobject *kobj)
L
Linus Torvalds 已提交
815
{
D
Dave Jones 已提交
816
	struct cpufreq_policy *policy = to_policy(kobj);
817
	pr_debug("last reference is dropped\n");
L
Linus Torvalds 已提交
818 819 820
	complete(&policy->kobj_unregister);
}

821
static const struct sysfs_ops sysfs_ops = {
L
Linus Torvalds 已提交
822 823 824 825 826 827 828 829 830 831
	.show	= show,
	.store	= store,
};

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

832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874
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);

875
/* symlink affected CPUs */
876
static int cpufreq_add_dev_symlink(struct cpufreq_policy *policy)
877 878 879 880 881
{
	unsigned int j;
	int ret = 0;

	for_each_cpu(j, policy->cpus) {
882
		struct device *cpu_dev;
883

884
		if (j == policy->cpu)
885 886
			continue;

887
		pr_debug("Adding link for CPU: %u\n", j);
888 889
		cpu_dev = get_cpu_device(j);
		ret = sysfs_create_link(&cpu_dev->kobj, &policy->kobj,
890
					"cpufreq");
891 892
		if (ret)
			break;
893 894 895 896
	}
	return ret;
}

897
static int cpufreq_add_dev_interface(struct cpufreq_policy *policy,
898
				     struct device *dev)
899 900 901 902 903 904
{
	struct freq_attr **drv_attr;
	int ret = 0;

	/* prepare interface data */
	ret = kobject_init_and_add(&policy->kobj, &ktype_cpufreq,
905
				   &dev->kobj, "cpufreq");
906 907 908 909
	if (ret)
		return ret;

	/* set up files for this cpu device */
910
	drv_attr = cpufreq_driver->attr;
911 912 913
	while ((drv_attr) && (*drv_attr)) {
		ret = sysfs_create_file(&policy->kobj, &((*drv_attr)->attr));
		if (ret)
914
			goto err_out_kobj_put;
915 916
		drv_attr++;
	}
917
	if (cpufreq_driver->get) {
918 919
		ret = sysfs_create_file(&policy->kobj, &cpuinfo_cur_freq.attr);
		if (ret)
920
			goto err_out_kobj_put;
921
	}
922 923 924 925 926

	ret = sysfs_create_file(&policy->kobj, &scaling_cur_freq.attr);
	if (ret)
		goto err_out_kobj_put;

927
	if (cpufreq_driver->bios_limit) {
928 929
		ret = sysfs_create_file(&policy->kobj, &bios_limit.attr);
		if (ret)
930
			goto err_out_kobj_put;
931
	}
932

933
	ret = cpufreq_add_dev_symlink(policy);
934 935 936
	if (ret)
		goto err_out_kobj_put;

937 938 939 940 941 942 943 944 945 946
	return ret;

err_out_kobj_put:
	kobject_put(&policy->kobj);
	wait_for_completion(&policy->kobj_unregister);
	return ret;
}

static void cpufreq_init_policy(struct cpufreq_policy *policy)
{
947
	struct cpufreq_governor *gov = NULL;
948 949 950
	struct cpufreq_policy new_policy;
	int ret = 0;

951
	memcpy(&new_policy, policy, sizeof(*policy));
952

953 954 955 956 957 958 959 960 961 962
	/* Update governor of new_policy to the governor used before hotplug */
	gov = __find_governor(per_cpu(cpufreq_cpu_governor, policy->cpu));
	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;

963 964
	/* Use the default policy if its valid. */
	if (cpufreq_driver->setpolicy)
965
		cpufreq_parse_governor(gov->name, &new_policy.policy, NULL);
966 967

	/* set default policy */
968
	ret = cpufreq_set_policy(policy, &new_policy);
969
	if (ret) {
970
		pr_debug("setting policy failed\n");
971 972
		if (cpufreq_driver->exit)
			cpufreq_driver->exit(policy);
973
	}
974 975
}

976
#ifdef CONFIG_HOTPLUG_CPU
977
static int cpufreq_add_policy_cpu(struct cpufreq_policy *policy,
978
				  unsigned int cpu, struct device *dev)
979
{
980
	int ret = 0;
981 982
	unsigned long flags;

983
	if (has_target()) {
984 985 986 987 988 989
		ret = __cpufreq_governor(policy, CPUFREQ_GOV_STOP);
		if (ret) {
			pr_err("%s: Failed to stop governor\n", __func__);
			return ret;
		}
	}
990

991
	down_write(&policy->rwsem);
V
Viresh Kumar 已提交
992

993
	write_lock_irqsave(&cpufreq_driver_lock, flags);
V
Viresh Kumar 已提交
994

995 996
	cpumask_set_cpu(cpu, policy->cpus);
	per_cpu(cpufreq_cpu_data, cpu) = policy;
997
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
998

999
	up_write(&policy->rwsem);
V
Viresh Kumar 已提交
1000

1001
	if (has_target()) {
1002 1003 1004 1005 1006
		ret = __cpufreq_governor(policy, CPUFREQ_GOV_START);
		if (!ret)
			ret = __cpufreq_governor(policy, CPUFREQ_GOV_LIMITS);

		if (ret) {
1007 1008 1009
			pr_err("%s: Failed to start governor\n", __func__);
			return ret;
		}
1010
	}
1011

1012
	return sysfs_create_link(&dev->kobj, &policy->kobj, "cpufreq");
1013 1014
}
#endif
L
Linus Torvalds 已提交
1015

1016 1017 1018 1019 1020
static struct cpufreq_policy *cpufreq_policy_restore(unsigned int cpu)
{
	struct cpufreq_policy *policy;
	unsigned long flags;

1021
	read_lock_irqsave(&cpufreq_driver_lock, flags);
1022 1023 1024

	policy = per_cpu(cpufreq_cpu_data_fallback, cpu);

1025
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
1026

1027 1028
	if (policy)
		policy->governor = NULL;
1029

1030 1031 1032
	return policy;
}

1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046
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;

1047
	INIT_LIST_HEAD(&policy->policy_list);
1048
	init_rwsem(&policy->rwsem);
1049 1050
	spin_lock_init(&policy->transition_lock);
	init_waitqueue_head(&policy->transition_wait);
1051

1052 1053 1054 1055 1056 1057 1058 1059 1060 1061
	return policy;

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

	return NULL;
}

1062 1063 1064 1065 1066
static void cpufreq_policy_put_kobj(struct cpufreq_policy *policy)
{
	struct kobject *kobj;
	struct completion *cmp;

1067 1068 1069
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
			CPUFREQ_REMOVE_POLICY, policy);

1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085
	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");
}

1086 1087 1088 1089 1090 1091 1092
static void cpufreq_policy_free(struct cpufreq_policy *policy)
{
	free_cpumask_var(policy->related_cpus);
	free_cpumask_var(policy->cpus);
	kfree(policy);
}

1093 1094
static int update_policy_cpu(struct cpufreq_policy *policy, unsigned int cpu,
			     struct device *cpu_dev)
1095
{
1096 1097
	int ret;

1098
	if (WARN_ON(cpu == policy->cpu))
1099 1100 1101 1102 1103 1104 1105 1106
		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;
	}
1107

1108
	down_write(&policy->rwsem);
1109

1110 1111 1112
	policy->last_cpu = policy->cpu;
	policy->cpu = cpu;

1113
	up_write(&policy->rwsem);
1114

1115 1116
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
			CPUFREQ_UPDATE_POLICY_CPU, policy);
1117 1118

	return 0;
1119 1120
}

1121
static int __cpufreq_add_dev(struct device *dev, struct subsys_interface *sif)
L
Linus Torvalds 已提交
1122
{
1123
	unsigned int j, cpu = dev->id;
1124
	int ret = -ENOMEM;
L
Linus Torvalds 已提交
1125 1126
	struct cpufreq_policy *policy;
	unsigned long flags;
1127
	bool recover_policy = cpufreq_suspended;
1128
#ifdef CONFIG_HOTPLUG_CPU
1129
	struct cpufreq_policy *tpolicy;
1130
#endif
L
Linus Torvalds 已提交
1131

1132 1133 1134
	if (cpu_is_offline(cpu))
		return 0;

1135
	pr_debug("adding CPU %u\n", cpu);
L
Linus Torvalds 已提交
1136 1137 1138 1139 1140 1141

#ifdef CONFIG_SMP
	/* check whether a different CPU already registered this
	 * CPU because it is in the same boat. */
	policy = cpufreq_cpu_get(cpu);
	if (unlikely(policy)) {
1142
		cpufreq_cpu_put(policy);
L
Linus Torvalds 已提交
1143 1144
		return 0;
	}
1145
#endif
1146

1147 1148 1149
	if (!down_read_trylock(&cpufreq_rwsem))
		return 0;

1150 1151
#ifdef CONFIG_HOTPLUG_CPU
	/* Check if this cpu was hot-unplugged earlier and has siblings */
1152
	read_lock_irqsave(&cpufreq_driver_lock, flags);
1153 1154
	list_for_each_entry(tpolicy, &cpufreq_policy_list, policy_list) {
		if (cpumask_test_cpu(cpu, tpolicy->related_cpus)) {
1155
			read_unlock_irqrestore(&cpufreq_driver_lock, flags);
1156
			ret = cpufreq_add_policy_cpu(tpolicy, cpu, dev);
1157 1158
			up_read(&cpufreq_rwsem);
			return ret;
V
Viresh Kumar 已提交
1159
		}
1160
	}
1161
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
1162 1163
#endif

1164 1165 1166 1167
	/*
	 * Restore the saved policy when doing light-weight init and fall back
	 * to the full init if that fails.
	 */
1168
	policy = recover_policy ? cpufreq_policy_restore(cpu) : NULL;
1169
	if (!policy) {
1170
		recover_policy = false;
1171
		policy = cpufreq_policy_alloc();
1172 1173 1174
		if (!policy)
			goto nomem_out;
	}
1175 1176 1177 1178 1179 1180 1181

	/*
	 * 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().
	 */
1182 1183 1184
	if (recover_policy && cpu != policy->cpu)
		WARN_ON(update_policy_cpu(policy, cpu, dev));
	else
1185 1186
		policy->cpu = cpu;

1187
	cpumask_copy(policy->cpus, cpumask_of(cpu));
L
Linus Torvalds 已提交
1188 1189

	init_completion(&policy->kobj_unregister);
1190
	INIT_WORK(&policy->update, handle_update);
L
Linus Torvalds 已提交
1191 1192 1193 1194

	/* call driver. From then on the cpufreq must be able
	 * to accept all calls to ->verify and ->setpolicy for this CPU
	 */
1195
	ret = cpufreq_driver->init(policy);
L
Linus Torvalds 已提交
1196
	if (ret) {
1197
		pr_debug("initialization failed\n");
V
Viresh Kumar 已提交
1198
		goto err_set_policy_cpu;
L
Linus Torvalds 已提交
1199
	}
V
Viresh Kumar 已提交
1200

1201 1202 1203 1204 1205 1206 1207 1208 1209
	/* 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);

1210
	if (!recover_policy) {
1211 1212 1213 1214
		policy->user_policy.min = policy->min;
		policy->user_policy.max = policy->max;
	}

1215
	down_write(&policy->rwsem);
1216 1217 1218 1219 1220
	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);

1221
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
1222 1223 1224 1225 1226 1227 1228
		policy->cur = cpufreq_driver->get(policy->cpu);
		if (!policy->cur) {
			pr_err("%s: ->get() failed\n", __func__);
			goto err_get_freq;
		}
	}

1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268
	/*
	 * 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);
		}
	}

1269 1270 1271
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
				     CPUFREQ_START, policy);

1272
	if (!recover_policy) {
1273
		ret = cpufreq_add_dev_interface(policy, dev);
1274 1275
		if (ret)
			goto err_out_unregister;
1276 1277
		blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
				CPUFREQ_CREATE_POLICY, policy);
1278
	}
1279

1280 1281 1282 1283
	write_lock_irqsave(&cpufreq_driver_lock, flags);
	list_add(&policy->policy_list, &cpufreq_policy_list);
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);

1284 1285
	cpufreq_init_policy(policy);

1286
	if (!recover_policy) {
1287 1288 1289
		policy->user_policy.policy = policy->policy;
		policy->user_policy.governor = policy->governor;
	}
1290
	up_write(&policy->rwsem);
1291

1292
	kobject_uevent(&policy->kobj, KOBJ_ADD);
1293 1294
	up_read(&cpufreq_rwsem);

1295
	pr_debug("initialization complete\n");
1296

L
Linus Torvalds 已提交
1297 1298 1299
	return 0;

err_out_unregister:
1300
err_get_freq:
1301
	write_lock_irqsave(&cpufreq_driver_lock, flags);
1302
	for_each_cpu(j, policy->cpus)
1303
		per_cpu(cpufreq_cpu_data, j) = NULL;
1304
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
1305

1306 1307
	up_write(&policy->rwsem);

1308 1309
	if (cpufreq_driver->exit)
		cpufreq_driver->exit(policy);
V
Viresh Kumar 已提交
1310
err_set_policy_cpu:
1311
	if (recover_policy) {
1312 1313
		/* Do not leave stale fallback data behind. */
		per_cpu(cpufreq_cpu_data_fallback, cpu) = NULL;
1314
		cpufreq_policy_put_kobj(policy);
1315
	}
1316
	cpufreq_policy_free(policy);
1317

L
Linus Torvalds 已提交
1318
nomem_out:
1319 1320
	up_read(&cpufreq_rwsem);

L
Linus Torvalds 已提交
1321 1322 1323
	return ret;
}

1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334
/**
 * 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)
{
1335
	return __cpufreq_add_dev(dev, sif);
1336 1337
}

1338
static int __cpufreq_remove_dev_prepare(struct device *dev,
1339
					struct subsys_interface *sif)
L
Linus Torvalds 已提交
1340
{
1341
	unsigned int cpu = dev->id, cpus;
1342
	int ret;
L
Linus Torvalds 已提交
1343
	unsigned long flags;
1344
	struct cpufreq_policy *policy;
L
Linus Torvalds 已提交
1345

1346
	pr_debug("%s: unregistering CPU %u\n", __func__, cpu);
L
Linus Torvalds 已提交
1347

1348
	write_lock_irqsave(&cpufreq_driver_lock, flags);
V
Viresh Kumar 已提交
1349

1350
	policy = per_cpu(cpufreq_cpu_data, cpu);
V
Viresh Kumar 已提交
1351

1352
	/* Save the policy somewhere when doing a light-weight tear-down */
1353
	if (cpufreq_suspended)
1354
		per_cpu(cpufreq_cpu_data_fallback, cpu) = policy;
1355

1356
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
1357

1358
	if (!policy) {
1359
		pr_debug("%s: No cpu_data found\n", __func__);
L
Linus Torvalds 已提交
1360 1361 1362
		return -EINVAL;
	}

1363
	if (has_target()) {
1364 1365 1366 1367 1368 1369
		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
1370

1371
	if (!cpufreq_driver->setpolicy)
1372
		strncpy(per_cpu(cpufreq_cpu_governor, cpu),
1373
			policy->governor->name, CPUFREQ_NAME_LEN);
L
Linus Torvalds 已提交
1374

1375
	down_read(&policy->rwsem);
1376
	cpus = cpumask_weight(policy->cpus);
1377
	up_read(&policy->rwsem);
1378

1379
	if (cpu != policy->cpu) {
1380
		sysfs_remove_link(&dev->kobj, "cpufreq");
1381
	} else if (cpus > 1) {
1382 1383 1384
		/* Nominate new CPU */
		int new_cpu = cpumask_any_but(policy->cpus, cpu);
		struct device *cpu_dev = get_cpu_device(new_cpu);
1385

1386 1387 1388 1389 1390 1391 1392 1393
		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
1394
		}
1395 1396 1397 1398

		if (!cpufreq_suspended)
			pr_debug("%s: policy Kobject moved to cpu: %d from: %d\n",
				 __func__, new_cpu, cpu);
1399
	} else if (cpufreq_driver->stop_cpu) {
1400
		cpufreq_driver->stop_cpu(policy);
L
Linus Torvalds 已提交
1401 1402
	}

1403 1404 1405 1406
	return 0;
}

static int __cpufreq_remove_dev_finish(struct device *dev,
1407
				       struct subsys_interface *sif)
1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422
{
	unsigned int cpu = dev->id, cpus;
	int ret;
	unsigned long flags;
	struct cpufreq_policy *policy;

	read_lock_irqsave(&cpufreq_driver_lock, flags);
	policy = per_cpu(cpufreq_cpu_data, cpu);
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);

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

1423
	down_write(&policy->rwsem);
1424
	cpus = cpumask_weight(policy->cpus);
1425 1426 1427

	if (cpus > 1)
		cpumask_clear_cpu(cpu, policy->cpus);
1428
	up_write(&policy->rwsem);
1429

1430 1431
	/* If cpu is last user of policy, free policy */
	if (cpus == 1) {
1432
		if (has_target()) {
1433 1434 1435 1436
			ret = __cpufreq_governor(policy,
					CPUFREQ_GOV_POLICY_EXIT);
			if (ret) {
				pr_err("%s: Failed to exit governor\n",
1437
				       __func__);
1438 1439
				return ret;
			}
1440
		}
1441

1442
		if (!cpufreq_suspended)
1443
			cpufreq_policy_put_kobj(policy);
1444

1445 1446 1447 1448
		/*
		 * 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.
1449
		 */
1450
		if (cpufreq_driver->exit)
1451
			cpufreq_driver->exit(policy);
1452

1453 1454 1455 1456 1457
		/* 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);

1458
		if (!cpufreq_suspended)
1459
			cpufreq_policy_free(policy);
1460 1461 1462 1463 1464 1465 1466 1467
	} 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;
1468
		}
1469
	}
L
Linus Torvalds 已提交
1470

1471
	per_cpu(cpufreq_cpu_data, cpu) = NULL;
L
Linus Torvalds 已提交
1472 1473 1474
	return 0;
}

1475
/**
1476
 * cpufreq_remove_dev - remove a CPU device
1477 1478 1479
 *
 * Removes the cpufreq interface for a CPU device.
 */
1480
static int cpufreq_remove_dev(struct device *dev, struct subsys_interface *sif)
1481
{
1482
	unsigned int cpu = dev->id;
1483
	int ret;
1484 1485 1486 1487

	if (cpu_is_offline(cpu))
		return 0;

1488
	ret = __cpufreq_remove_dev_prepare(dev, sif);
1489 1490

	if (!ret)
1491
		ret = __cpufreq_remove_dev_finish(dev, sif);
1492 1493

	return ret;
1494 1495
}

1496
static void handle_update(struct work_struct *work)
L
Linus Torvalds 已提交
1497
{
1498 1499 1500
	struct cpufreq_policy *policy =
		container_of(work, struct cpufreq_policy, update);
	unsigned int cpu = policy->cpu;
1501
	pr_debug("handle_update for cpu %u called\n", cpu);
L
Linus Torvalds 已提交
1502 1503 1504 1505
	cpufreq_update_policy(cpu);
}

/**
1506 1507
 *	cpufreq_out_of_sync - If actual and saved CPU frequency differs, we're
 *	in deep trouble.
L
Linus Torvalds 已提交
1508 1509 1510 1511
 *	@cpu: cpu number
 *	@old_freq: CPU frequency the kernel thinks the CPU runs at
 *	@new_freq: CPU frequency the CPU actually runs at
 *
1512 1513
 *	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
1514
 */
1515 1516
static void cpufreq_out_of_sync(unsigned int cpu, unsigned int old_freq,
				unsigned int new_freq)
L
Linus Torvalds 已提交
1517
{
1518
	struct cpufreq_policy *policy;
L
Linus Torvalds 已提交
1519
	struct cpufreq_freqs freqs;
1520 1521
	unsigned long flags;

1522 1523
	pr_debug("Warning: CPU frequency out of sync: cpufreq and timing core thinks of %u, is %u kHz\n",
		 old_freq, new_freq);
L
Linus Torvalds 已提交
1524 1525 1526

	freqs.old = old_freq;
	freqs.new = new_freq;
1527 1528 1529 1530 1531

	read_lock_irqsave(&cpufreq_driver_lock, flags);
	policy = per_cpu(cpufreq_cpu_data, cpu);
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);

1532 1533
	cpufreq_freq_transition_begin(policy, &freqs);
	cpufreq_freq_transition_end(policy, &freqs, 0);
L
Linus Torvalds 已提交
1534 1535
}

1536
/**
D
Dhaval Giani 已提交
1537
 * cpufreq_quick_get - get the CPU frequency (in kHz) from policy->cur
1538 1539 1540 1541 1542 1543 1544
 * @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)
{
1545
	struct cpufreq_policy *policy;
1546
	unsigned int ret_freq = 0;
1547

1548 1549
	if (cpufreq_driver && cpufreq_driver->setpolicy && cpufreq_driver->get)
		return cpufreq_driver->get(cpu);
1550 1551

	policy = cpufreq_cpu_get(cpu);
1552
	if (policy) {
1553
		ret_freq = policy->cur;
1554 1555 1556
		cpufreq_cpu_put(policy);
	}

D
Dave Jones 已提交
1557
	return ret_freq;
1558 1559 1560
}
EXPORT_SYMBOL(cpufreq_quick_get);

1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580
/**
 * 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);

1581
static unsigned int __cpufreq_get(unsigned int cpu)
L
Linus Torvalds 已提交
1582
{
1583
	struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);
1584
	unsigned int ret_freq = 0;
1585

1586
	if (!cpufreq_driver->get)
D
Dave Jones 已提交
1587
		return ret_freq;
L
Linus Torvalds 已提交
1588

1589
	ret_freq = cpufreq_driver->get(cpu);
L
Linus Torvalds 已提交
1590

1591
	if (ret_freq && policy->cur &&
1592
		!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
1593 1594 1595 1596
		/* verify no discrepancy between actual and
					saved value exists */
		if (unlikely(ret_freq != policy->cur)) {
			cpufreq_out_of_sync(cpu, policy->cur, 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 1617 1618
	if (policy) {
		down_read(&policy->rwsem);
		ret_freq = __cpufreq_get(cpu);
		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 1680 1681 1682 1683 1684 1685 1686 1687
	pr_debug("%s: Suspending Governors\n", __func__);

	list_for_each_entry(policy, &cpufreq_policy_list, policy_list) {
		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
	list_for_each_entry(policy, &cpufreq_policy_list, policy_list) {
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
		    || __cpufreq_governor(policy, CPUFREQ_GOV_LIMITS))
			pr_err("%s: Failed to start governor for policy: %p\n",
				__func__, policy);
L
Linus Torvalds 已提交
1722

1723 1724 1725 1726 1727 1728 1729 1730 1731
		/*
		 * schedule call cpufreq_update_policy() for boot CPU, i.e. last
		 * policy in list. It will verify that the current freq is in
		 * sync with what we believe it to be.
		 */
		if (list_is_last(&policy->policy_list, &cpufreq_policy_list))
			schedule_work(&policy->update);
	}
}
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 2006
/*
 * when "event" is CPUFREQ_GOV_LIMITS
 */
L
Linus Torvalds 已提交
2007

2008 2009
static int __cpufreq_governor(struct cpufreq_policy *policy,
					unsigned int event)
L
Linus Torvalds 已提交
2010
{
2011
	int ret;
2012 2013 2014 2015 2016 2017 2018 2019 2020 2021

	/* 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
2022

2023 2024 2025 2026
	/* Don't start any governor operations if we are entering suspend */
	if (cpufreq_suspended)
		return 0;

2027 2028 2029
	if (policy->governor->max_transition_latency &&
	    policy->cpuinfo.transition_latency >
	    policy->governor->max_transition_latency) {
2030 2031 2032
		if (!gov)
			return -EINVAL;
		else {
2033 2034
			pr_warn("%s governor failed, too long transition latency of HW, fallback to %s governor\n",
				policy->governor->name, gov->name);
2035 2036
			policy->governor = gov;
		}
2037
	}
L
Linus Torvalds 已提交
2038

2039 2040 2041
	if (event == CPUFREQ_GOV_POLICY_INIT)
		if (!try_module_get(policy->governor->owner))
			return -EINVAL;
L
Linus Torvalds 已提交
2042

2043
	pr_debug("__cpufreq_governor for CPU %u, event %u\n",
2044
		 policy->cpu, event);
2045 2046

	mutex_lock(&cpufreq_governor_lock);
2047
	if ((policy->governor_enabled && event == CPUFREQ_GOV_START)
2048 2049
	    || (!policy->governor_enabled
	    && (event == CPUFREQ_GOV_LIMITS || event == CPUFREQ_GOV_STOP))) {
2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060
		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
2061 2062
	ret = policy->governor->governor(policy, event);

2063 2064 2065 2066 2067
	if (!ret) {
		if (event == CPUFREQ_GOV_POLICY_INIT)
			policy->governor->initialized++;
		else if (event == CPUFREQ_GOV_POLICY_EXIT)
			policy->governor->initialized--;
2068 2069 2070 2071 2072 2073 2074 2075
	} 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);
2076
	}
2077

2078 2079
	if (((event == CPUFREQ_GOV_POLICY_INIT) && ret) ||
			((event == CPUFREQ_GOV_POLICY_EXIT) && !ret))
L
Linus Torvalds 已提交
2080 2081 2082 2083 2084 2085 2086
		module_put(policy->governor->owner);

	return ret;
}

int cpufreq_register_governor(struct cpufreq_governor *governor)
{
2087
	int err;
L
Linus Torvalds 已提交
2088 2089 2090 2091

	if (!governor)
		return -EINVAL;

2092 2093 2094
	if (cpufreq_disabled())
		return -ENODEV;

2095
	mutex_lock(&cpufreq_governor_mutex);
2096

2097
	governor->initialized = 0;
2098 2099 2100 2101
	err = -EBUSY;
	if (__find_governor(governor->name) == NULL) {
		err = 0;
		list_add(&governor->governor_list, &cpufreq_governor_list);
L
Linus Torvalds 已提交
2102 2103
	}

2104
	mutex_unlock(&cpufreq_governor_mutex);
2105
	return err;
L
Linus Torvalds 已提交
2106 2107 2108 2109 2110
}
EXPORT_SYMBOL_GPL(cpufreq_register_governor);

void cpufreq_unregister_governor(struct cpufreq_governor *governor)
{
2111 2112
	int cpu;

L
Linus Torvalds 已提交
2113 2114 2115
	if (!governor)
		return;

2116 2117 2118
	if (cpufreq_disabled())
		return;

2119 2120 2121 2122 2123 2124 2125
	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");
	}

2126
	mutex_lock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
2127
	list_del(&governor->governor_list);
2128
	mutex_unlock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139
	return;
}
EXPORT_SYMBOL_GPL(cpufreq_unregister_governor);


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

/**
 * cpufreq_get_policy - get the current cpufreq_policy
2140 2141
 * @policy: struct cpufreq_policy into which the current cpufreq_policy
 *	is written
L
Linus Torvalds 已提交
2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154
 *
 * 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;

2155
	memcpy(policy, cpu_policy, sizeof(*policy));
L
Linus Torvalds 已提交
2156 2157 2158 2159 2160 2161

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

2162
/*
2163 2164
 * policy : current policy.
 * new_policy: policy to be set.
2165
 */
2166
static int cpufreq_set_policy(struct cpufreq_policy *policy,
2167
				struct cpufreq_policy *new_policy)
L
Linus Torvalds 已提交
2168
{
2169 2170
	struct cpufreq_governor *old_gov;
	int ret;
L
Linus Torvalds 已提交
2171

2172 2173
	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
2174

2175
	memcpy(&new_policy->cpuinfo, &policy->cpuinfo, sizeof(policy->cpuinfo));
L
Linus Torvalds 已提交
2176

2177 2178
	if (new_policy->min > policy->max || new_policy->max < policy->min)
		return -EINVAL;
2179

L
Linus Torvalds 已提交
2180
	/* verify the cpu speed can be set within this limit */
2181
	ret = cpufreq_driver->verify(new_policy);
L
Linus Torvalds 已提交
2182
	if (ret)
2183
		return ret;
L
Linus Torvalds 已提交
2184 2185

	/* adjust if necessary - all reasons */
2186
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2187
			CPUFREQ_ADJUST, new_policy);
L
Linus Torvalds 已提交
2188 2189

	/* adjust if necessary - hardware incompatibility*/
2190
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2191
			CPUFREQ_INCOMPATIBLE, new_policy);
L
Linus Torvalds 已提交
2192

2193 2194 2195 2196
	/*
	 * verify the cpu speed can be set within this limit, which might be
	 * different to the first one
	 */
2197
	ret = cpufreq_driver->verify(new_policy);
2198
	if (ret)
2199
		return ret;
L
Linus Torvalds 已提交
2200 2201

	/* notification of the new policy */
2202
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2203
			CPUFREQ_NOTIFY, new_policy);
L
Linus Torvalds 已提交
2204

2205 2206
	policy->min = new_policy->min;
	policy->max = new_policy->max;
L
Linus Torvalds 已提交
2207

2208
	pr_debug("new min and max freqs are %u - %u kHz\n",
2209
		 policy->min, policy->max);
L
Linus Torvalds 已提交
2210

2211
	if (cpufreq_driver->setpolicy) {
2212
		policy->policy = new_policy->policy;
2213
		pr_debug("setting range\n");
2214 2215
		return cpufreq_driver->setpolicy(new_policy);
	}
L
Linus Torvalds 已提交
2216

2217 2218
	if (new_policy->governor == policy->governor)
		goto out;
2219

2220 2221 2222 2223 2224 2225 2226 2227
	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);
2228
		__cpufreq_governor(policy, CPUFREQ_GOV_POLICY_EXIT);
2229
		down_write(&policy->rwsem);
L
Linus Torvalds 已提交
2230 2231
	}

2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255
	/* 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
2256 2257 2258 2259 2260 2261
}

/**
 *	cpufreq_update_policy - re-evaluate an existing cpufreq policy
 *	@cpu: CPU which shall be re-evaluated
 *
L
Lucas De Marchi 已提交
2262
 *	Useful for policy notifiers which have different necessities
L
Linus Torvalds 已提交
2263 2264 2265 2266
 *	at different times.
 */
int cpufreq_update_policy(unsigned int cpu)
{
2267 2268
	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
	struct cpufreq_policy new_policy;
2269
	int ret;
L
Linus Torvalds 已提交
2270

2271 2272
	if (!policy)
		return -ENODEV;
L
Linus Torvalds 已提交
2273

2274
	down_write(&policy->rwsem);
L
Linus Torvalds 已提交
2275

2276
	pr_debug("updating policy for CPU %u\n", cpu);
2277
	memcpy(&new_policy, policy, sizeof(*policy));
2278 2279 2280 2281
	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
2282

2283 2284 2285 2286
	/*
	 * BIOS might change freq behind our back
	 * -> ask driver for current freq and notify governors about a change
	 */
2287
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
2288
		new_policy.cur = cpufreq_driver->get(cpu);
2289 2290
		if (WARN_ON(!new_policy.cur)) {
			ret = -EIO;
2291
			goto unlock;
2292 2293
		}

2294
		if (!policy->cur) {
2295
			pr_debug("Driver did not initialize current freq\n");
2296
			policy->cur = new_policy.cur;
2297
		} else {
2298
			if (policy->cur != new_policy.cur && has_target())
2299 2300
				cpufreq_out_of_sync(cpu, policy->cur,
								new_policy.cur);
2301
		}
2302 2303
	}

2304
	ret = cpufreq_set_policy(policy, &new_policy);
L
Linus Torvalds 已提交
2305

2306
unlock:
2307
	up_write(&policy->rwsem);
2308

2309
	cpufreq_cpu_put(policy);
L
Linus Torvalds 已提交
2310 2311 2312 2313
	return ret;
}
EXPORT_SYMBOL(cpufreq_update_policy);

2314
static int cpufreq_cpu_callback(struct notifier_block *nfb,
2315 2316 2317
					unsigned long action, void *hcpu)
{
	unsigned int cpu = (unsigned long)hcpu;
2318
	struct device *dev;
2319

2320 2321
	dev = get_cpu_device(cpu);
	if (dev) {
2322
		switch (action & ~CPU_TASKS_FROZEN) {
2323
		case CPU_ONLINE:
2324
			__cpufreq_add_dev(dev, NULL);
2325
			break;
2326

2327
		case CPU_DOWN_PREPARE:
2328
			__cpufreq_remove_dev_prepare(dev, NULL);
2329 2330 2331
			break;

		case CPU_POST_DEAD:
2332
			__cpufreq_remove_dev_finish(dev, NULL);
2333
			break;
2334

2335
		case CPU_DOWN_FAILED:
2336
			__cpufreq_add_dev(dev, NULL);
2337 2338 2339 2340 2341 2342
			break;
		}
	}
	return NOTIFY_OK;
}

2343
static struct notifier_block __refdata cpufreq_cpu_notifier = {
2344
	.notifier_call = cpufreq_cpu_callback,
2345
};
L
Linus Torvalds 已提交
2346

2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 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
/*********************************************************************
 *               BOOST						     *
 *********************************************************************/
static int cpufreq_boost_set_sw(int state)
{
	struct cpufreq_frequency_table *freq_table;
	struct cpufreq_policy *policy;
	int ret = -EINVAL;

	list_for_each_entry(policy, &cpufreq_policy_list, policy_list) {
		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);

2392 2393
		pr_err("%s: Cannot %s BOOST\n",
		       __func__, state ? "enable" : "disable");
2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413
	}

	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
2414 2415 2416 2417 2418 2419 2420 2421 2422
/*********************************************************************
 *               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.
 *
2423
 * Registers a CPU Frequency driver to this core code. This code
L
Linus Torvalds 已提交
2424
 * returns zero on success, -EBUSY when another driver got here first
2425
 * (and isn't unregistered in the meantime).
L
Linus Torvalds 已提交
2426 2427
 *
 */
2428
int cpufreq_register_driver(struct cpufreq_driver *driver_data)
L
Linus Torvalds 已提交
2429 2430 2431 2432
{
	unsigned long flags;
	int ret;

2433 2434 2435
	if (cpufreq_disabled())
		return -ENODEV;

L
Linus Torvalds 已提交
2436
	if (!driver_data || !driver_data->verify || !driver_data->init ||
2437
	    !(driver_data->setpolicy || driver_data->target_index ||
2438 2439
		    driver_data->target) ||
	     (driver_data->setpolicy && (driver_data->target_index ||
2440 2441
		    driver_data->target)) ||
	     (!!driver_data->get_intermediate != !!driver_data->target_intermediate))
L
Linus Torvalds 已提交
2442 2443
		return -EINVAL;

2444
	pr_debug("trying to register driver %s\n", driver_data->name);
L
Linus Torvalds 已提交
2445 2446 2447 2448

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

2449
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2450
	if (cpufreq_driver) {
2451
		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2452
		return -EEXIST;
L
Linus Torvalds 已提交
2453
	}
2454
	cpufreq_driver = driver_data;
2455
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
2456

2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467
	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",
2468
			       __func__);
2469 2470 2471 2472
			goto err_null_driver;
		}
	}

2473
	ret = subsys_interface_register(&cpufreq_interface);
2474
	if (ret)
2475
		goto err_boost_unreg;
L
Linus Torvalds 已提交
2476

2477
	if (!(cpufreq_driver->flags & CPUFREQ_STICKY)) {
L
Linus Torvalds 已提交
2478 2479 2480 2481
		int i;
		ret = -ENODEV;

		/* check for at least one working CPU */
2482 2483
		for (i = 0; i < nr_cpu_ids; i++)
			if (cpu_possible(i) && per_cpu(cpufreq_cpu_data, i)) {
L
Linus Torvalds 已提交
2484
				ret = 0;
2485 2486
				break;
			}
L
Linus Torvalds 已提交
2487 2488 2489

		/* if all ->init() calls failed, unregister */
		if (ret) {
2490
			pr_debug("no CPU initialized for driver %s\n",
2491
				 driver_data->name);
2492
			goto err_if_unreg;
L
Linus Torvalds 已提交
2493 2494 2495
		}
	}

2496
	register_hotcpu_notifier(&cpufreq_cpu_notifier);
2497
	pr_debug("driver %s up and running\n", driver_data->name);
L
Linus Torvalds 已提交
2498

2499
	return 0;
2500 2501
err_if_unreg:
	subsys_interface_unregister(&cpufreq_interface);
2502 2503 2504
err_boost_unreg:
	if (cpufreq_boost_supported())
		cpufreq_sysfs_remove_file(&boost.attr);
2505
err_null_driver:
2506
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2507
	cpufreq_driver = NULL;
2508
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
D
Dave Jones 已提交
2509
	return ret;
L
Linus Torvalds 已提交
2510 2511 2512 2513 2514 2515
}
EXPORT_SYMBOL_GPL(cpufreq_register_driver);

/**
 * cpufreq_unregister_driver - unregister the current CPUFreq driver
 *
2516
 * Unregister the current CPUFreq driver. Only call this if you have
L
Linus Torvalds 已提交
2517 2518 2519 2520
 * the right to do so, i.e. if you have succeeded in initialising before!
 * Returns zero if successful, and -EINVAL if the cpufreq_driver is
 * currently not initialised.
 */
2521
int cpufreq_unregister_driver(struct cpufreq_driver *driver)
L
Linus Torvalds 已提交
2522 2523 2524
{
	unsigned long flags;

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

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

2530
	subsys_interface_unregister(&cpufreq_interface);
2531 2532 2533
	if (cpufreq_boost_supported())
		cpufreq_sysfs_remove_file(&boost.attr);

2534
	unregister_hotcpu_notifier(&cpufreq_cpu_notifier);
L
Linus Torvalds 已提交
2535

2536
	down_write(&cpufreq_rwsem);
2537
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2538

2539
	cpufreq_driver = NULL;
2540

2541
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2542
	up_write(&cpufreq_rwsem);
L
Linus Torvalds 已提交
2543 2544 2545 2546

	return 0;
}
EXPORT_SYMBOL_GPL(cpufreq_unregister_driver);
2547 2548 2549

static int __init cpufreq_core_init(void)
{
2550 2551 2552
	if (cpufreq_disabled())
		return -ENODEV;

2553
	cpufreq_global_kobject = kobject_create();
2554 2555
	BUG_ON(!cpufreq_global_kobject);

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