cpufreq.c 64.3 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5
/*
 *  linux/drivers/cpufreq/cpufreq.c
 *
 *  Copyright (C) 2001 Russell King
 *            (C) 2002 - 2003 Dominik Brodowski <linux@brodo.de>
6
 *            (C) 2013 Viresh Kumar <viresh.kumar@linaro.org>
L
Linus Torvalds 已提交
7
 *
8
 *  Oct 2005 - Ashok Raj <ashok.raj@intel.com>
9
 *	Added handling for CPU hotplug
10 11
 *  Feb 2006 - Jacob Shin <jacob.shin@amd.com>
 *	Fix handling for CPU hotplug -- affected CPUs
12
 *
L
Linus Torvalds 已提交
13 14 15 16 17
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 */

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

20
#include <linux/cpu.h>
L
Linus Torvalds 已提交
21 22 23
#include <linux/cpufreq.h>
#include <linux/delay.h>
#include <linux/device.h>
24 25 26
#include <linux/init.h>
#include <linux/kernel_stat.h>
#include <linux/module.h>
27
#include <linux/mutex.h>
28
#include <linux/slab.h>
29
#include <linux/suspend.h>
30
#include <linux/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 (!strnicmp(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) {
L
Linus Torvalds 已提交
458 459
		if (!strnicmp(str_governor, "performance", CPUFREQ_NAME_LEN)) {
			*policy = CPUFREQ_POLICY_PERFORMANCE;
460
			err = 0;
461 462
		} else if (!strnicmp(str_governor, "powersave",
						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
513 514 515 516
show_one(scaling_min_freq, min);
show_one(scaling_max_freq, max);
show_one(scaling_cur_freq, cur);

517
static int cpufreq_set_policy(struct cpufreq_policy *policy,
518
				struct cpufreq_policy *new_policy);
519

L
Linus Torvalds 已提交
520 521 522 523 524
/**
 * 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
525
(struct cpufreq_policy *policy, const char *buf, size_t count)		\
L
Linus Torvalds 已提交
526
{									\
527
	int ret;							\
L
Linus Torvalds 已提交
528 529 530 531 532 533
	struct cpufreq_policy new_policy;				\
									\
	ret = cpufreq_get_policy(&new_policy, policy->cpu);		\
	if (ret)							\
		return -EINVAL;						\
									\
534
	ret = sscanf(buf, "%u", &new_policy.object);			\
L
Linus Torvalds 已提交
535 536 537
	if (ret != 1)							\
		return -EINVAL;						\
									\
538
	ret = cpufreq_set_policy(policy, &new_policy);		\
539
	policy->user_policy.object = policy->object;			\
L
Linus Torvalds 已提交
540 541 542 543
									\
	return ret ? ret : count;					\
}

544 545
store_one(scaling_min_freq, min);
store_one(scaling_max_freq, max);
L
Linus Torvalds 已提交
546 547 548 549

/**
 * show_cpuinfo_cur_freq - current CPU frequency as detected by hardware
 */
D
Dave Jones 已提交
550 551
static ssize_t show_cpuinfo_cur_freq(struct cpufreq_policy *policy,
					char *buf)
L
Linus Torvalds 已提交
552
{
553
	unsigned int cur_freq = __cpufreq_get(policy->cpu);
L
Linus Torvalds 已提交
554 555 556 557 558 559 560 561
	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
562
static ssize_t show_scaling_governor(struct cpufreq_policy *policy, char *buf)
L
Linus Torvalds 已提交
563
{
564
	if (policy->policy == CPUFREQ_POLICY_POWERSAVE)
L
Linus Torvalds 已提交
565 566 567 568
		return sprintf(buf, "powersave\n");
	else if (policy->policy == CPUFREQ_POLICY_PERFORMANCE)
		return sprintf(buf, "performance\n");
	else if (policy->governor)
569
		return scnprintf(buf, CPUFREQ_NAME_PLEN, "%s\n",
570
				policy->governor->name);
L
Linus Torvalds 已提交
571 572 573 574 575 576
	return -EINVAL;
}

/**
 * store_scaling_governor - store policy for the specified CPU
 */
D
Dave Jones 已提交
577 578
static ssize_t store_scaling_governor(struct cpufreq_policy *policy,
					const char *buf, size_t count)
L
Linus Torvalds 已提交
579
{
580
	int ret;
L
Linus Torvalds 已提交
581 582 583 584 585 586 587
	char	str_governor[16];
	struct cpufreq_policy new_policy;

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

588
	ret = sscanf(buf, "%15s", str_governor);
L
Linus Torvalds 已提交
589 590 591
	if (ret != 1)
		return -EINVAL;

592 593
	if (cpufreq_parse_governor(str_governor, &new_policy.policy,
						&new_policy.governor))
L
Linus Torvalds 已提交
594 595
		return -EINVAL;

596
	ret = cpufreq_set_policy(policy, &new_policy);
597 598 599 600

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

601 602 603 604
	if (ret)
		return ret;
	else
		return count;
L
Linus Torvalds 已提交
605 606 607 608 609
}

/**
 * show_scaling_driver - show the cpufreq driver currently loaded
 */
D
Dave Jones 已提交
610
static ssize_t show_scaling_driver(struct cpufreq_policy *policy, char *buf)
L
Linus Torvalds 已提交
611
{
612
	return scnprintf(buf, CPUFREQ_NAME_PLEN, "%s\n", cpufreq_driver->name);
L
Linus Torvalds 已提交
613 614 615 616 617
}

/**
 * show_scaling_available_governors - show the available CPUfreq governors
 */
D
Dave Jones 已提交
618 619
static ssize_t show_scaling_available_governors(struct cpufreq_policy *policy,
						char *buf)
L
Linus Torvalds 已提交
620 621 622 623
{
	ssize_t i = 0;
	struct cpufreq_governor *t;

624
	if (!has_target()) {
L
Linus Torvalds 已提交
625 626 627 628 629
		i += sprintf(buf, "performance powersave");
		goto out;
	}

	list_for_each_entry(t, &cpufreq_governor_list, governor_list) {
630 631
		if (i >= (ssize_t) ((PAGE_SIZE / sizeof(char))
		    - (CPUFREQ_NAME_LEN + 2)))
L
Linus Torvalds 已提交
632
			goto out;
633
		i += scnprintf(&buf[i], CPUFREQ_NAME_PLEN, "%s ", t->name);
L
Linus Torvalds 已提交
634
	}
635
out:
L
Linus Torvalds 已提交
636 637 638
	i += sprintf(&buf[i], "\n");
	return i;
}
639

640
ssize_t cpufreq_show_cpus(const struct cpumask *mask, char *buf)
L
Linus Torvalds 已提交
641 642 643 644
{
	ssize_t i = 0;
	unsigned int cpu;

645
	for_each_cpu(cpu, mask) {
L
Linus Torvalds 已提交
646 647 648 649
		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))
650
			break;
L
Linus Torvalds 已提交
651 652 653 654
	}
	i += sprintf(&buf[i], "\n");
	return i;
}
655
EXPORT_SYMBOL_GPL(cpufreq_show_cpus);
L
Linus Torvalds 已提交
656

657 658 659 660 661 662
/**
 * 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)
{
663
	return cpufreq_show_cpus(policy->related_cpus, buf);
664 665 666 667 668 669 670
}

/**
 * show_affected_cpus - show the CPUs affected by each transition
 */
static ssize_t show_affected_cpus(struct cpufreq_policy *policy, char *buf)
{
671
	return cpufreq_show_cpus(policy->cpus, buf);
672 673
}

674
static ssize_t store_scaling_setspeed(struct cpufreq_policy *policy,
D
Dave Jones 已提交
675
					const char *buf, size_t count)
676 677 678 679
{
	unsigned int freq = 0;
	unsigned int ret;

680
	if (!policy->governor || !policy->governor->store_setspeed)
681 682 683 684 685 686 687 688 689 690 691 692 693
		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)
{
694
	if (!policy->governor || !policy->governor->show_setspeed)
695 696 697 698
		return sprintf(buf, "<unsupported>\n");

	return policy->governor->show_setspeed(policy, buf);
}
L
Linus Torvalds 已提交
699

700
/**
701
 * show_bios_limit - show the current cpufreq HW/BIOS limitation
702 703 704 705 706
 */
static ssize_t show_bios_limit(struct cpufreq_policy *policy, char *buf)
{
	unsigned int limit;
	int ret;
707 708
	if (cpufreq_driver->bios_limit) {
		ret = cpufreq_driver->bios_limit(policy->cpu, &limit);
709 710 711 712 713 714
		if (!ret)
			return sprintf(buf, "%u\n", limit);
	}
	return sprintf(buf, "%u\n", policy->cpuinfo.max_freq);
}

715 716 717 718 719 720 721 722 723 724 725 726 727 728
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
729

D
Dave Jones 已提交
730
static struct attribute *default_attrs[] = {
L
Linus Torvalds 已提交
731 732
	&cpuinfo_min_freq.attr,
	&cpuinfo_max_freq.attr,
733
	&cpuinfo_transition_latency.attr,
L
Linus Torvalds 已提交
734 735 736
	&scaling_min_freq.attr,
	&scaling_max_freq.attr,
	&affected_cpus.attr,
737
	&related_cpus.attr,
L
Linus Torvalds 已提交
738 739 740
	&scaling_governor.attr,
	&scaling_driver.attr,
	&scaling_available_governors.attr,
741
	&scaling_setspeed.attr,
L
Linus Torvalds 已提交
742 743 744
	NULL
};

745 746
#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
747

748
static ssize_t show(struct kobject *kobj, struct attribute *attr, char *buf)
L
Linus Torvalds 已提交
749
{
D
Dave Jones 已提交
750 751
	struct cpufreq_policy *policy = to_policy(kobj);
	struct freq_attr *fattr = to_attr(attr);
752
	ssize_t ret;
753 754

	if (!down_read_trylock(&cpufreq_rwsem))
755
		return -EINVAL;
756

757
	down_read(&policy->rwsem);
758

759 760 761 762 763
	if (fattr->show)
		ret = fattr->show(policy, buf);
	else
		ret = -EIO;

764
	up_read(&policy->rwsem);
765
	up_read(&cpufreq_rwsem);
766

L
Linus Torvalds 已提交
767 768 769
	return ret;
}

D
Dave Jones 已提交
770 771
static ssize_t store(struct kobject *kobj, struct attribute *attr,
		     const char *buf, size_t count)
L
Linus Torvalds 已提交
772
{
D
Dave Jones 已提交
773 774
	struct cpufreq_policy *policy = to_policy(kobj);
	struct freq_attr *fattr = to_attr(attr);
775
	ssize_t ret = -EINVAL;
776

777 778 779 780 781
	get_online_cpus();

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

782
	if (!down_read_trylock(&cpufreq_rwsem))
783
		goto unlock;
784

785
	down_write(&policy->rwsem);
786

787 788 789 790 791
	if (fattr->store)
		ret = fattr->store(policy, buf, count);
	else
		ret = -EIO;

792
	up_write(&policy->rwsem);
793 794

	up_read(&cpufreq_rwsem);
795 796 797
unlock:
	put_online_cpus();

L
Linus Torvalds 已提交
798 799 800
	return ret;
}

D
Dave Jones 已提交
801
static void cpufreq_sysfs_release(struct kobject *kobj)
L
Linus Torvalds 已提交
802
{
D
Dave Jones 已提交
803
	struct cpufreq_policy *policy = to_policy(kobj);
804
	pr_debug("last reference is dropped\n");
L
Linus Torvalds 已提交
805 806 807
	complete(&policy->kobj_unregister);
}

808
static const struct sysfs_ops sysfs_ops = {
L
Linus Torvalds 已提交
809 810 811 812 813 814 815 816 817 818
	.show	= show,
	.store	= store,
};

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

819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861
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);

862
/* symlink affected CPUs */
863
static int cpufreq_add_dev_symlink(struct cpufreq_policy *policy)
864 865 866 867 868
{
	unsigned int j;
	int ret = 0;

	for_each_cpu(j, policy->cpus) {
869
		struct device *cpu_dev;
870

871
		if (j == policy->cpu)
872 873
			continue;

874
		pr_debug("Adding link for CPU: %u\n", j);
875 876
		cpu_dev = get_cpu_device(j);
		ret = sysfs_create_link(&cpu_dev->kobj, &policy->kobj,
877
					"cpufreq");
878 879
		if (ret)
			break;
880 881 882 883
	}
	return ret;
}

884
static int cpufreq_add_dev_interface(struct cpufreq_policy *policy,
885
				     struct device *dev)
886 887 888 889 890 891
{
	struct freq_attr **drv_attr;
	int ret = 0;

	/* prepare interface data */
	ret = kobject_init_and_add(&policy->kobj, &ktype_cpufreq,
892
				   &dev->kobj, "cpufreq");
893 894 895 896
	if (ret)
		return ret;

	/* set up files for this cpu device */
897
	drv_attr = cpufreq_driver->attr;
898 899 900
	while ((drv_attr) && (*drv_attr)) {
		ret = sysfs_create_file(&policy->kobj, &((*drv_attr)->attr));
		if (ret)
901
			goto err_out_kobj_put;
902 903
		drv_attr++;
	}
904
	if (cpufreq_driver->get) {
905 906
		ret = sysfs_create_file(&policy->kobj, &cpuinfo_cur_freq.attr);
		if (ret)
907
			goto err_out_kobj_put;
908
	}
909
	if (has_target()) {
910 911
		ret = sysfs_create_file(&policy->kobj, &scaling_cur_freq.attr);
		if (ret)
912
			goto err_out_kobj_put;
913
	}
914
	if (cpufreq_driver->bios_limit) {
915 916
		ret = sysfs_create_file(&policy->kobj, &bios_limit.attr);
		if (ret)
917
			goto err_out_kobj_put;
918
	}
919

920
	ret = cpufreq_add_dev_symlink(policy);
921 922 923
	if (ret)
		goto err_out_kobj_put;

924 925 926 927 928 929 930 931 932 933
	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)
{
934
	struct cpufreq_governor *gov = NULL;
935 936 937
	struct cpufreq_policy new_policy;
	int ret = 0;

938
	memcpy(&new_policy, policy, sizeof(*policy));
939

940 941 942 943 944 945 946 947 948 949
	/* 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;

950 951
	/* Use the default policy if its valid. */
	if (cpufreq_driver->setpolicy)
952
		cpufreq_parse_governor(gov->name, &new_policy.policy, NULL);
953 954

	/* set default policy */
955
	ret = cpufreq_set_policy(policy, &new_policy);
956
	if (ret) {
957
		pr_debug("setting policy failed\n");
958 959
		if (cpufreq_driver->exit)
			cpufreq_driver->exit(policy);
960
	}
961 962
}

963
#ifdef CONFIG_HOTPLUG_CPU
964
static int cpufreq_add_policy_cpu(struct cpufreq_policy *policy,
965
				  unsigned int cpu, struct device *dev)
966
{
967
	int ret = 0;
968 969
	unsigned long flags;

970
	if (has_target()) {
971 972 973 974 975 976
		ret = __cpufreq_governor(policy, CPUFREQ_GOV_STOP);
		if (ret) {
			pr_err("%s: Failed to stop governor\n", __func__);
			return ret;
		}
	}
977

978
	down_write(&policy->rwsem);
V
Viresh Kumar 已提交
979

980
	write_lock_irqsave(&cpufreq_driver_lock, flags);
V
Viresh Kumar 已提交
981

982 983
	cpumask_set_cpu(cpu, policy->cpus);
	per_cpu(cpufreq_cpu_data, cpu) = policy;
984
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
985

986
	up_write(&policy->rwsem);
V
Viresh Kumar 已提交
987

988
	if (has_target()) {
989 990 991 992 993
		ret = __cpufreq_governor(policy, CPUFREQ_GOV_START);
		if (!ret)
			ret = __cpufreq_governor(policy, CPUFREQ_GOV_LIMITS);

		if (ret) {
994 995 996
			pr_err("%s: Failed to start governor\n", __func__);
			return ret;
		}
997
	}
998

999
	return sysfs_create_link(&dev->kobj, &policy->kobj, "cpufreq");
1000 1001
}
#endif
L
Linus Torvalds 已提交
1002

1003 1004 1005 1006 1007
static struct cpufreq_policy *cpufreq_policy_restore(unsigned int cpu)
{
	struct cpufreq_policy *policy;
	unsigned long flags;

1008
	read_lock_irqsave(&cpufreq_driver_lock, flags);
1009 1010 1011

	policy = per_cpu(cpufreq_cpu_data_fallback, cpu);

1012
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
1013

1014 1015
	policy->governor = NULL;

1016 1017 1018
	return policy;
}

1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032
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;

1033
	INIT_LIST_HEAD(&policy->policy_list);
1034
	init_rwsem(&policy->rwsem);
1035 1036
	spin_lock_init(&policy->transition_lock);
	init_waitqueue_head(&policy->transition_wait);
1037

1038 1039 1040 1041 1042 1043 1044 1045 1046 1047
	return policy;

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

	return NULL;
}

1048 1049 1050 1051 1052
static void cpufreq_policy_put_kobj(struct cpufreq_policy *policy)
{
	struct kobject *kobj;
	struct completion *cmp;

1053 1054 1055
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
			CPUFREQ_REMOVE_POLICY, policy);

1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071
	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");
}

1072 1073 1074 1075 1076 1077 1078
static void cpufreq_policy_free(struct cpufreq_policy *policy)
{
	free_cpumask_var(policy->related_cpus);
	free_cpumask_var(policy->cpus);
	kfree(policy);
}

1079 1080
static int update_policy_cpu(struct cpufreq_policy *policy, unsigned int cpu,
			     struct device *cpu_dev)
1081
{
1082 1083
	int ret;

1084
	if (WARN_ON(cpu == policy->cpu))
1085 1086 1087 1088 1089 1090 1091 1092
		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;
	}
1093

1094
	down_write(&policy->rwsem);
1095

1096 1097 1098
	policy->last_cpu = policy->cpu;
	policy->cpu = cpu;

1099
	up_write(&policy->rwsem);
1100

1101 1102
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
			CPUFREQ_UPDATE_POLICY_CPU, policy);
1103 1104

	return 0;
1105 1106
}

1107
static int __cpufreq_add_dev(struct device *dev, struct subsys_interface *sif)
L
Linus Torvalds 已提交
1108
{
1109
	unsigned int j, cpu = dev->id;
1110
	int ret = -ENOMEM;
L
Linus Torvalds 已提交
1111 1112
	struct cpufreq_policy *policy;
	unsigned long flags;
1113
	bool recover_policy = cpufreq_suspended;
1114
#ifdef CONFIG_HOTPLUG_CPU
1115
	struct cpufreq_policy *tpolicy;
1116
#endif
L
Linus Torvalds 已提交
1117

1118 1119 1120
	if (cpu_is_offline(cpu))
		return 0;

1121
	pr_debug("adding CPU %u\n", cpu);
L
Linus Torvalds 已提交
1122 1123 1124 1125 1126 1127

#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)) {
1128
		cpufreq_cpu_put(policy);
L
Linus Torvalds 已提交
1129 1130
		return 0;
	}
1131
#endif
1132

1133 1134 1135
	if (!down_read_trylock(&cpufreq_rwsem))
		return 0;

1136 1137
#ifdef CONFIG_HOTPLUG_CPU
	/* Check if this cpu was hot-unplugged earlier and has siblings */
1138
	read_lock_irqsave(&cpufreq_driver_lock, flags);
1139 1140
	list_for_each_entry(tpolicy, &cpufreq_policy_list, policy_list) {
		if (cpumask_test_cpu(cpu, tpolicy->related_cpus)) {
1141
			read_unlock_irqrestore(&cpufreq_driver_lock, flags);
1142
			ret = cpufreq_add_policy_cpu(tpolicy, cpu, dev);
1143 1144
			up_read(&cpufreq_rwsem);
			return ret;
V
Viresh Kumar 已提交
1145
		}
1146
	}
1147
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
1148 1149
#endif

1150 1151 1152 1153
	/*
	 * Restore the saved policy when doing light-weight init and fall back
	 * to the full init if that fails.
	 */
1154
	policy = recover_policy ? cpufreq_policy_restore(cpu) : NULL;
1155
	if (!policy) {
1156
		recover_policy = false;
1157
		policy = cpufreq_policy_alloc();
1158 1159 1160
		if (!policy)
			goto nomem_out;
	}
1161 1162 1163 1164 1165 1166 1167

	/*
	 * 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().
	 */
1168 1169 1170
	if (recover_policy && cpu != policy->cpu)
		WARN_ON(update_policy_cpu(policy, cpu, dev));
	else
1171 1172
		policy->cpu = cpu;

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

	init_completion(&policy->kobj_unregister);
1176
	INIT_WORK(&policy->update, handle_update);
L
Linus Torvalds 已提交
1177 1178 1179 1180

	/* call driver. From then on the cpufreq must be able
	 * to accept all calls to ->verify and ->setpolicy for this CPU
	 */
1181
	ret = cpufreq_driver->init(policy);
L
Linus Torvalds 已提交
1182
	if (ret) {
1183
		pr_debug("initialization failed\n");
V
Viresh Kumar 已提交
1184
		goto err_set_policy_cpu;
L
Linus Torvalds 已提交
1185
	}
V
Viresh Kumar 已提交
1186

1187 1188 1189 1190 1191 1192 1193 1194 1195
	/* 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);

1196
	if (!recover_policy) {
1197 1198 1199 1200
		policy->user_policy.min = policy->min;
		policy->user_policy.max = policy->max;
	}

1201
	down_write(&policy->rwsem);
1202 1203 1204 1205 1206
	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);

1207
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
1208 1209 1210 1211 1212 1213 1214
		policy->cur = cpufreq_driver->get(policy->cpu);
		if (!policy->cur) {
			pr_err("%s: ->get() failed\n", __func__);
			goto err_get_freq;
		}
	}

1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254
	/*
	 * 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);
		}
	}

1255 1256 1257
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
				     CPUFREQ_START, policy);

1258
	if (!recover_policy) {
1259
		ret = cpufreq_add_dev_interface(policy, dev);
1260 1261
		if (ret)
			goto err_out_unregister;
1262 1263
		blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
				CPUFREQ_CREATE_POLICY, policy);
1264
	}
1265

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

1270 1271
	cpufreq_init_policy(policy);

1272
	if (!recover_policy) {
1273 1274 1275
		policy->user_policy.policy = policy->policy;
		policy->user_policy.governor = policy->governor;
	}
1276
	up_write(&policy->rwsem);
1277

1278
	kobject_uevent(&policy->kobj, KOBJ_ADD);
1279 1280
	up_read(&cpufreq_rwsem);

1281
	pr_debug("initialization complete\n");
1282

L
Linus Torvalds 已提交
1283 1284 1285
	return 0;

err_out_unregister:
1286
err_get_freq:
1287
	write_lock_irqsave(&cpufreq_driver_lock, flags);
1288
	for_each_cpu(j, policy->cpus)
1289
		per_cpu(cpufreq_cpu_data, j) = NULL;
1290
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
1291

1292 1293
	if (cpufreq_driver->exit)
		cpufreq_driver->exit(policy);
V
Viresh Kumar 已提交
1294
err_set_policy_cpu:
1295
	if (recover_policy) {
1296 1297
		/* Do not leave stale fallback data behind. */
		per_cpu(cpufreq_cpu_data_fallback, cpu) = NULL;
1298
		cpufreq_policy_put_kobj(policy);
1299
	}
1300
	cpufreq_policy_free(policy);
1301

L
Linus Torvalds 已提交
1302
nomem_out:
1303 1304
	up_read(&cpufreq_rwsem);

L
Linus Torvalds 已提交
1305 1306 1307
	return ret;
}

1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318
/**
 * 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)
{
1319
	return __cpufreq_add_dev(dev, sif);
1320 1321
}

1322
static int __cpufreq_remove_dev_prepare(struct device *dev,
1323
					struct subsys_interface *sif)
L
Linus Torvalds 已提交
1324
{
1325
	unsigned int cpu = dev->id, cpus;
1326
	int ret;
L
Linus Torvalds 已提交
1327
	unsigned long flags;
1328
	struct cpufreq_policy *policy;
L
Linus Torvalds 已提交
1329

1330
	pr_debug("%s: unregistering CPU %u\n", __func__, cpu);
L
Linus Torvalds 已提交
1331

1332
	write_lock_irqsave(&cpufreq_driver_lock, flags);
V
Viresh Kumar 已提交
1333

1334
	policy = per_cpu(cpufreq_cpu_data, cpu);
V
Viresh Kumar 已提交
1335

1336
	/* Save the policy somewhere when doing a light-weight tear-down */
1337
	if (cpufreq_suspended)
1338
		per_cpu(cpufreq_cpu_data_fallback, cpu) = policy;
1339

1340
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
1341

1342
	if (!policy) {
1343
		pr_debug("%s: No cpu_data found\n", __func__);
L
Linus Torvalds 已提交
1344 1345 1346
		return -EINVAL;
	}

1347
	if (has_target()) {
1348 1349 1350 1351 1352 1353
		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
1354

1355
	if (!cpufreq_driver->setpolicy)
1356
		strncpy(per_cpu(cpufreq_cpu_governor, cpu),
1357
			policy->governor->name, CPUFREQ_NAME_LEN);
L
Linus Torvalds 已提交
1358

1359
	down_read(&policy->rwsem);
1360
	cpus = cpumask_weight(policy->cpus);
1361
	up_read(&policy->rwsem);
1362

1363
	if (cpu != policy->cpu) {
1364
		sysfs_remove_link(&dev->kobj, "cpufreq");
1365
	} else if (cpus > 1) {
1366 1367 1368
		/* Nominate new CPU */
		int new_cpu = cpumask_any_but(policy->cpus, cpu);
		struct device *cpu_dev = get_cpu_device(new_cpu);
1369

1370 1371 1372 1373 1374 1375 1376 1377
		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
1378
		}
1379 1380 1381 1382

		if (!cpufreq_suspended)
			pr_debug("%s: policy Kobject moved to cpu: %d from: %d\n",
				 __func__, new_cpu, cpu);
1383
	} else if (cpufreq_driver->stop_cpu) {
1384
		cpufreq_driver->stop_cpu(policy);
L
Linus Torvalds 已提交
1385 1386
	}

1387 1388 1389 1390
	return 0;
}

static int __cpufreq_remove_dev_finish(struct device *dev,
1391
				       struct subsys_interface *sif)
1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406
{
	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;
	}

1407
	down_write(&policy->rwsem);
1408
	cpus = cpumask_weight(policy->cpus);
1409 1410 1411

	if (cpus > 1)
		cpumask_clear_cpu(cpu, policy->cpus);
1412
	up_write(&policy->rwsem);
1413

1414 1415
	/* If cpu is last user of policy, free policy */
	if (cpus == 1) {
1416
		if (has_target()) {
1417 1418 1419 1420
			ret = __cpufreq_governor(policy,
					CPUFREQ_GOV_POLICY_EXIT);
			if (ret) {
				pr_err("%s: Failed to exit governor\n",
1421
				       __func__);
1422 1423
				return ret;
			}
1424
		}
1425

1426
		if (!cpufreq_suspended)
1427
			cpufreq_policy_put_kobj(policy);
1428

1429 1430 1431 1432
		/*
		 * 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.
1433
		 */
1434
		if (cpufreq_driver->exit)
1435
			cpufreq_driver->exit(policy);
1436

1437 1438 1439 1440 1441
		/* 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);

1442
		if (!cpufreq_suspended)
1443
			cpufreq_policy_free(policy);
1444 1445 1446 1447 1448 1449 1450 1451
	} 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;
1452
		}
1453
	}
L
Linus Torvalds 已提交
1454

1455
	per_cpu(cpufreq_cpu_data, cpu) = NULL;
L
Linus Torvalds 已提交
1456 1457 1458
	return 0;
}

1459
/**
1460
 * cpufreq_remove_dev - remove a CPU device
1461 1462 1463
 *
 * Removes the cpufreq interface for a CPU device.
 */
1464
static int cpufreq_remove_dev(struct device *dev, struct subsys_interface *sif)
1465
{
1466
	unsigned int cpu = dev->id;
1467
	int ret;
1468 1469 1470 1471

	if (cpu_is_offline(cpu))
		return 0;

1472
	ret = __cpufreq_remove_dev_prepare(dev, sif);
1473 1474

	if (!ret)
1475
		ret = __cpufreq_remove_dev_finish(dev, sif);
1476 1477

	return ret;
1478 1479
}

1480
static void handle_update(struct work_struct *work)
L
Linus Torvalds 已提交
1481
{
1482 1483 1484
	struct cpufreq_policy *policy =
		container_of(work, struct cpufreq_policy, update);
	unsigned int cpu = policy->cpu;
1485
	pr_debug("handle_update for cpu %u called\n", cpu);
L
Linus Torvalds 已提交
1486 1487 1488 1489
	cpufreq_update_policy(cpu);
}

/**
1490 1491
 *	cpufreq_out_of_sync - If actual and saved CPU frequency differs, we're
 *	in deep trouble.
L
Linus Torvalds 已提交
1492 1493 1494 1495
 *	@cpu: cpu number
 *	@old_freq: CPU frequency the kernel thinks the CPU runs at
 *	@new_freq: CPU frequency the CPU actually runs at
 *
1496 1497
 *	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
1498
 */
1499 1500
static void cpufreq_out_of_sync(unsigned int cpu, unsigned int old_freq,
				unsigned int new_freq)
L
Linus Torvalds 已提交
1501
{
1502
	struct cpufreq_policy *policy;
L
Linus Torvalds 已提交
1503
	struct cpufreq_freqs freqs;
1504 1505
	unsigned long flags;

1506 1507
	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
1508 1509 1510

	freqs.old = old_freq;
	freqs.new = new_freq;
1511 1512 1513 1514 1515

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

1516 1517
	cpufreq_freq_transition_begin(policy, &freqs);
	cpufreq_freq_transition_end(policy, &freqs, 0);
L
Linus Torvalds 已提交
1518 1519
}

1520
/**
D
Dhaval Giani 已提交
1521
 * cpufreq_quick_get - get the CPU frequency (in kHz) from policy->cur
1522 1523 1524 1525 1526 1527 1528
 * @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)
{
1529
	struct cpufreq_policy *policy;
1530
	unsigned int ret_freq = 0;
1531

1532 1533
	if (cpufreq_driver && cpufreq_driver->setpolicy && cpufreq_driver->get)
		return cpufreq_driver->get(cpu);
1534 1535

	policy = cpufreq_cpu_get(cpu);
1536
	if (policy) {
1537
		ret_freq = policy->cur;
1538 1539 1540
		cpufreq_cpu_put(policy);
	}

D
Dave Jones 已提交
1541
	return ret_freq;
1542 1543 1544
}
EXPORT_SYMBOL(cpufreq_quick_get);

1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564
/**
 * 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);

1565
static unsigned int __cpufreq_get(unsigned int cpu)
L
Linus Torvalds 已提交
1566
{
1567
	struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);
1568
	unsigned int ret_freq = 0;
1569

1570
	if (!cpufreq_driver->get)
D
Dave Jones 已提交
1571
		return ret_freq;
L
Linus Torvalds 已提交
1572

1573
	ret_freq = cpufreq_driver->get(cpu);
L
Linus Torvalds 已提交
1574

1575
	if (ret_freq && policy->cur &&
1576
		!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
1577 1578 1579 1580
		/* 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
1581 1582 1583 1584
			schedule_work(&policy->update);
		}
	}

D
Dave Jones 已提交
1585
	return ret_freq;
1586
}
L
Linus Torvalds 已提交
1587

1588 1589 1590 1591 1592 1593 1594 1595
/**
 * 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)
{
1596
	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
1597 1598
	unsigned int ret_freq = 0;

1599 1600 1601 1602
	if (policy) {
		down_read(&policy->rwsem);
		ret_freq = __cpufreq_get(cpu);
		up_read(&policy->rwsem);
1603

1604 1605
		cpufreq_cpu_put(policy);
	}
1606

D
Dave Jones 已提交
1607
	return ret_freq;
L
Linus Torvalds 已提交
1608 1609 1610
}
EXPORT_SYMBOL(cpufreq_get);

1611 1612 1613 1614 1615
static struct subsys_interface cpufreq_interface = {
	.name		= "cpufreq",
	.subsys		= &cpu_subsys,
	.add_dev	= cpufreq_add_dev,
	.remove_dev	= cpufreq_remove_dev,
1616 1617
};

1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643
/*
 * 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);

1644
/**
1645
 * cpufreq_suspend() - Suspend CPUFreq governors
1646
 *
1647 1648 1649 1650
 * 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.
1651
 */
1652
void cpufreq_suspend(void)
1653
{
1654
	struct cpufreq_policy *policy;
1655

1656 1657
	if (!cpufreq_driver)
		return;
1658

1659 1660
	if (!has_target())
		return;
1661

1662 1663 1664 1665 1666 1667 1668 1669 1670 1671
	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);
1672 1673
	}

1674
	cpufreq_suspended = true;
1675 1676
}

L
Linus Torvalds 已提交
1677
/**
1678
 * cpufreq_resume() - Resume CPUFreq governors
L
Linus Torvalds 已提交
1679
 *
1680 1681
 * Called during system wide Suspend/Hibernate cycle for resuming governors that
 * are suspended with cpufreq_suspend().
L
Linus Torvalds 已提交
1682
 */
1683
void cpufreq_resume(void)
L
Linus Torvalds 已提交
1684
{
1685
	struct cpufreq_policy *policy;
L
Linus Torvalds 已提交
1686

1687 1688
	if (!cpufreq_driver)
		return;
L
Linus Torvalds 已提交
1689

1690
	if (!has_target())
1691
		return;
L
Linus Torvalds 已提交
1692

1693
	pr_debug("%s: Resuming Governors\n", __func__);
L
Linus Torvalds 已提交
1694

1695
	cpufreq_suspended = false;
1696

1697
	list_for_each_entry(policy, &cpufreq_policy_list, policy_list) {
1698 1699 1700 1701
		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)
1702 1703 1704
		    || __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
1705

1706 1707 1708 1709 1710 1711 1712 1713 1714
		/*
		 * 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
1715

1716 1717 1718 1719 1720 1721 1722 1723
/**
 *	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)
{
1724 1725 1726 1727
	if (cpufreq_driver)
		return cpufreq_driver->name;

	return NULL;
1728 1729
}
EXPORT_SYMBOL_GPL(cpufreq_get_current_driver);
L
Linus Torvalds 已提交
1730 1731 1732 1733 1734 1735 1736 1737 1738 1739

/*********************************************************************
 *                     NOTIFIER LISTS INTERFACE                      *
 *********************************************************************/

/**
 *	cpufreq_register_notifier - register a driver with cpufreq
 *	@nb: notifier function to register
 *      @list: CPUFREQ_TRANSITION_NOTIFIER or CPUFREQ_POLICY_NOTIFIER
 *
1740
 *	Add a driver to one of two lists: either a list of drivers that
L
Linus Torvalds 已提交
1741 1742 1743 1744 1745
 *      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
1746
 *	blocking_notifier_chain_register.
L
Linus Torvalds 已提交
1747 1748 1749 1750 1751
 */
int cpufreq_register_notifier(struct notifier_block *nb, unsigned int list)
{
	int ret;

1752 1753 1754
	if (cpufreq_disabled())
		return -EINVAL;

1755 1756
	WARN_ON(!init_cpufreq_transition_notifier_list_called);

L
Linus Torvalds 已提交
1757 1758
	switch (list) {
	case CPUFREQ_TRANSITION_NOTIFIER:
1759
		ret = srcu_notifier_chain_register(
1760
				&cpufreq_transition_notifier_list, nb);
L
Linus Torvalds 已提交
1761 1762
		break;
	case CPUFREQ_POLICY_NOTIFIER:
1763 1764
		ret = blocking_notifier_chain_register(
				&cpufreq_policy_notifier_list, nb);
L
Linus Torvalds 已提交
1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776
		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
1777
 *	@list: CPUFREQ_TRANSITION_NOTIFIER or CPUFREQ_POLICY_NOTIFIER
L
Linus Torvalds 已提交
1778 1779 1780 1781
 *
 *	Remove a driver from the CPU frequency notifier list.
 *
 *	This function may sleep, and has the same return conditions as
1782
 *	blocking_notifier_chain_unregister.
L
Linus Torvalds 已提交
1783 1784 1785 1786 1787
 */
int cpufreq_unregister_notifier(struct notifier_block *nb, unsigned int list)
{
	int ret;

1788 1789 1790
	if (cpufreq_disabled())
		return -EINVAL;

L
Linus Torvalds 已提交
1791 1792
	switch (list) {
	case CPUFREQ_TRANSITION_NOTIFIER:
1793
		ret = srcu_notifier_chain_unregister(
1794
				&cpufreq_transition_notifier_list, nb);
L
Linus Torvalds 已提交
1795 1796
		break;
	case CPUFREQ_POLICY_NOTIFIER:
1797 1798
		ret = blocking_notifier_chain_unregister(
				&cpufreq_policy_notifier_list, nb);
L
Linus Torvalds 已提交
1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812
		break;
	default:
		ret = -EINVAL;
	}

	return ret;
}
EXPORT_SYMBOL(cpufreq_unregister_notifier);


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

1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838
/* 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;
}

1839 1840 1841
static int __target_index(struct cpufreq_policy *policy,
			  struct cpufreq_frequency_table *freq_table, int index)
{
1842 1843
	struct cpufreq_freqs freqs = {.old = policy->cur, .flags = 0};
	unsigned int intermediate_freq = 0;
1844 1845 1846 1847 1848
	int retval = -EINVAL;
	bool notify;

	notify = !(cpufreq_driver->flags & CPUFREQ_ASYNC_NOTIFICATION);
	if (notify) {
1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859
		/* 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;
		}
1860

1861
		freqs.new = freq_table[index].frequency;
1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872
		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);

1873
	if (notify) {
1874 1875
		cpufreq_freq_transition_end(policy, &freqs, retval);

1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889
		/*
		 * 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);
		}
	}

1890 1891 1892
	return retval;
}

L
Linus Torvalds 已提交
1893 1894 1895 1896
int __cpufreq_driver_target(struct cpufreq_policy *policy,
			    unsigned int target_freq,
			    unsigned int relation)
{
1897
	unsigned int old_target_freq = target_freq;
1898
	int retval = -EINVAL;
1899

1900 1901 1902
	if (cpufreq_disabled())
		return -ENODEV;

1903 1904 1905 1906 1907 1908 1909
	/* 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",
1910
		 policy->cpu, target_freq, relation, old_target_freq);
1911

1912 1913 1914 1915 1916 1917
	/*
	 * 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.
	 */
1918 1919 1920
	if (target_freq == policy->cur)
		return 0;

1921 1922 1923
	/* Save last value to restore later on errors */
	policy->restore_freq = policy->cur;

1924 1925
	if (cpufreq_driver->target)
		retval = cpufreq_driver->target(policy, target_freq, relation);
1926 1927 1928
	else if (cpufreq_driver->target_index) {
		struct cpufreq_frequency_table *freq_table;
		int index;
1929

1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942
		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;
		}

1943
		if (freq_table[index].frequency == policy->cur) {
1944
			retval = 0;
1945 1946 1947
			goto out;
		}

1948
		retval = __target_index(policy, freq_table, index);
1949 1950 1951
	}

out:
L
Linus Torvalds 已提交
1952 1953 1954 1955 1956 1957 1958 1959
	return retval;
}
EXPORT_SYMBOL_GPL(__cpufreq_driver_target);

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

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

	ret = __cpufreq_driver_target(policy, target_freq, relation);

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

	return ret;
}
EXPORT_SYMBOL_GPL(cpufreq_driver_target);

1972 1973 1974
/*
 * when "event" is CPUFREQ_GOV_LIMITS
 */
L
Linus Torvalds 已提交
1975

1976 1977
static int __cpufreq_governor(struct cpufreq_policy *policy,
					unsigned int event)
L
Linus Torvalds 已提交
1978
{
1979
	int ret;
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989

	/* 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
1990

1991 1992 1993 1994
	/* Don't start any governor operations if we are entering suspend */
	if (cpufreq_suspended)
		return 0;

1995 1996 1997
	if (policy->governor->max_transition_latency &&
	    policy->cpuinfo.transition_latency >
	    policy->governor->max_transition_latency) {
1998 1999 2000
		if (!gov)
			return -EINVAL;
		else {
2001 2002
			pr_warn("%s governor failed, too long transition latency of HW, fallback to %s governor\n",
				policy->governor->name, gov->name);
2003 2004
			policy->governor = gov;
		}
2005
	}
L
Linus Torvalds 已提交
2006

2007 2008 2009
	if (event == CPUFREQ_GOV_POLICY_INIT)
		if (!try_module_get(policy->governor->owner))
			return -EINVAL;
L
Linus Torvalds 已提交
2010

2011
	pr_debug("__cpufreq_governor for CPU %u, event %u\n",
2012
		 policy->cpu, event);
2013 2014

	mutex_lock(&cpufreq_governor_lock);
2015
	if ((policy->governor_enabled && event == CPUFREQ_GOV_START)
2016 2017
	    || (!policy->governor_enabled
	    && (event == CPUFREQ_GOV_LIMITS || event == CPUFREQ_GOV_STOP))) {
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
		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
2029 2030
	ret = policy->governor->governor(policy, event);

2031 2032 2033 2034 2035
	if (!ret) {
		if (event == CPUFREQ_GOV_POLICY_INIT)
			policy->governor->initialized++;
		else if (event == CPUFREQ_GOV_POLICY_EXIT)
			policy->governor->initialized--;
2036 2037 2038 2039 2040 2041 2042 2043
	} 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);
2044
	}
2045

2046 2047
	if (((event == CPUFREQ_GOV_POLICY_INIT) && ret) ||
			((event == CPUFREQ_GOV_POLICY_EXIT) && !ret))
L
Linus Torvalds 已提交
2048 2049 2050 2051 2052 2053 2054
		module_put(policy->governor->owner);

	return ret;
}

int cpufreq_register_governor(struct cpufreq_governor *governor)
{
2055
	int err;
L
Linus Torvalds 已提交
2056 2057 2058 2059

	if (!governor)
		return -EINVAL;

2060 2061 2062
	if (cpufreq_disabled())
		return -ENODEV;

2063
	mutex_lock(&cpufreq_governor_mutex);
2064

2065
	governor->initialized = 0;
2066 2067 2068 2069
	err = -EBUSY;
	if (__find_governor(governor->name) == NULL) {
		err = 0;
		list_add(&governor->governor_list, &cpufreq_governor_list);
L
Linus Torvalds 已提交
2070 2071
	}

2072
	mutex_unlock(&cpufreq_governor_mutex);
2073
	return err;
L
Linus Torvalds 已提交
2074 2075 2076 2077 2078
}
EXPORT_SYMBOL_GPL(cpufreq_register_governor);

void cpufreq_unregister_governor(struct cpufreq_governor *governor)
{
2079 2080
	int cpu;

L
Linus Torvalds 已提交
2081 2082 2083
	if (!governor)
		return;

2084 2085 2086
	if (cpufreq_disabled())
		return;

2087 2088 2089 2090 2091 2092 2093
	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");
	}

2094
	mutex_lock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
2095
	list_del(&governor->governor_list);
2096
	mutex_unlock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107
	return;
}
EXPORT_SYMBOL_GPL(cpufreq_unregister_governor);


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

/**
 * cpufreq_get_policy - get the current cpufreq_policy
2108 2109
 * @policy: struct cpufreq_policy into which the current cpufreq_policy
 *	is written
L
Linus Torvalds 已提交
2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122
 *
 * 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;

2123
	memcpy(policy, cpu_policy, sizeof(*policy));
L
Linus Torvalds 已提交
2124 2125 2126 2127 2128 2129

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

2130
/*
2131 2132
 * policy : current policy.
 * new_policy: policy to be set.
2133
 */
2134
static int cpufreq_set_policy(struct cpufreq_policy *policy,
2135
				struct cpufreq_policy *new_policy)
L
Linus Torvalds 已提交
2136
{
2137 2138
	struct cpufreq_governor *old_gov;
	int ret;
L
Linus Torvalds 已提交
2139

2140 2141
	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
2142

2143
	memcpy(&new_policy->cpuinfo, &policy->cpuinfo, sizeof(policy->cpuinfo));
L
Linus Torvalds 已提交
2144

2145 2146
	if (new_policy->min > policy->max || new_policy->max < policy->min)
		return -EINVAL;
2147

L
Linus Torvalds 已提交
2148
	/* verify the cpu speed can be set within this limit */
2149
	ret = cpufreq_driver->verify(new_policy);
L
Linus Torvalds 已提交
2150
	if (ret)
2151
		return ret;
L
Linus Torvalds 已提交
2152 2153

	/* adjust if necessary - all reasons */
2154
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2155
			CPUFREQ_ADJUST, new_policy);
L
Linus Torvalds 已提交
2156 2157

	/* adjust if necessary - hardware incompatibility*/
2158
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2159
			CPUFREQ_INCOMPATIBLE, new_policy);
L
Linus Torvalds 已提交
2160

2161 2162 2163 2164
	/*
	 * verify the cpu speed can be set within this limit, which might be
	 * different to the first one
	 */
2165
	ret = cpufreq_driver->verify(new_policy);
2166
	if (ret)
2167
		return ret;
L
Linus Torvalds 已提交
2168 2169

	/* notification of the new policy */
2170
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2171
			CPUFREQ_NOTIFY, new_policy);
L
Linus Torvalds 已提交
2172

2173 2174
	policy->min = new_policy->min;
	policy->max = new_policy->max;
L
Linus Torvalds 已提交
2175

2176
	pr_debug("new min and max freqs are %u - %u kHz\n",
2177
		 policy->min, policy->max);
L
Linus Torvalds 已提交
2178

2179
	if (cpufreq_driver->setpolicy) {
2180
		policy->policy = new_policy->policy;
2181
		pr_debug("setting range\n");
2182 2183
		return cpufreq_driver->setpolicy(new_policy);
	}
L
Linus Torvalds 已提交
2184

2185 2186
	if (new_policy->governor == policy->governor)
		goto out;
2187

2188 2189 2190 2191 2192 2193 2194 2195
	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);
2196
		__cpufreq_governor(policy, CPUFREQ_GOV_POLICY_EXIT);
2197
		down_write(&policy->rwsem);
L
Linus Torvalds 已提交
2198 2199
	}

2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223
	/* 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
2224 2225 2226 2227 2228 2229
}

/**
 *	cpufreq_update_policy - re-evaluate an existing cpufreq policy
 *	@cpu: CPU which shall be re-evaluated
 *
L
Lucas De Marchi 已提交
2230
 *	Useful for policy notifiers which have different necessities
L
Linus Torvalds 已提交
2231 2232 2233 2234
 *	at different times.
 */
int cpufreq_update_policy(unsigned int cpu)
{
2235 2236
	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
	struct cpufreq_policy new_policy;
2237
	int ret;
L
Linus Torvalds 已提交
2238

2239 2240
	if (!policy)
		return -ENODEV;
L
Linus Torvalds 已提交
2241

2242
	down_write(&policy->rwsem);
L
Linus Torvalds 已提交
2243

2244
	pr_debug("updating policy for CPU %u\n", cpu);
2245
	memcpy(&new_policy, policy, sizeof(*policy));
2246 2247 2248 2249
	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
2250

2251 2252 2253 2254
	/*
	 * BIOS might change freq behind our back
	 * -> ask driver for current freq and notify governors about a change
	 */
2255
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
2256
		new_policy.cur = cpufreq_driver->get(cpu);
2257 2258
		if (WARN_ON(!new_policy.cur)) {
			ret = -EIO;
2259
			goto unlock;
2260 2261
		}

2262
		if (!policy->cur) {
2263
			pr_debug("Driver did not initialize current freq\n");
2264
			policy->cur = new_policy.cur;
2265
		} else {
2266
			if (policy->cur != new_policy.cur && has_target())
2267 2268
				cpufreq_out_of_sync(cpu, policy->cur,
								new_policy.cur);
2269
		}
2270 2271
	}

2272
	ret = cpufreq_set_policy(policy, &new_policy);
L
Linus Torvalds 已提交
2273

2274
unlock:
2275
	up_write(&policy->rwsem);
2276

2277
	cpufreq_cpu_put(policy);
L
Linus Torvalds 已提交
2278 2279 2280 2281
	return ret;
}
EXPORT_SYMBOL(cpufreq_update_policy);

2282
static int cpufreq_cpu_callback(struct notifier_block *nfb,
2283 2284 2285
					unsigned long action, void *hcpu)
{
	unsigned int cpu = (unsigned long)hcpu;
2286
	struct device *dev;
2287

2288 2289
	dev = get_cpu_device(cpu);
	if (dev) {
2290
		switch (action & ~CPU_TASKS_FROZEN) {
2291
		case CPU_ONLINE:
2292
			__cpufreq_add_dev(dev, NULL);
2293
			break;
2294

2295
		case CPU_DOWN_PREPARE:
2296
			__cpufreq_remove_dev_prepare(dev, NULL);
2297 2298 2299
			break;

		case CPU_POST_DEAD:
2300
			__cpufreq_remove_dev_finish(dev, NULL);
2301
			break;
2302

2303
		case CPU_DOWN_FAILED:
2304
			__cpufreq_add_dev(dev, NULL);
2305 2306 2307 2308 2309 2310
			break;
		}
	}
	return NOTIFY_OK;
}

2311
static struct notifier_block __refdata cpufreq_cpu_notifier = {
2312
	.notifier_call = cpufreq_cpu_callback,
2313
};
L
Linus Torvalds 已提交
2314

2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359
/*********************************************************************
 *               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);

2360 2361
		pr_err("%s: Cannot %s BOOST\n",
		       __func__, state ? "enable" : "disable");
2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381
	}

	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
2382 2383 2384 2385 2386 2387 2388 2389 2390
/*********************************************************************
 *               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.
 *
2391
 * Registers a CPU Frequency driver to this core code. This code
L
Linus Torvalds 已提交
2392
 * returns zero on success, -EBUSY when another driver got here first
2393
 * (and isn't unregistered in the meantime).
L
Linus Torvalds 已提交
2394 2395
 *
 */
2396
int cpufreq_register_driver(struct cpufreq_driver *driver_data)
L
Linus Torvalds 已提交
2397 2398 2399 2400
{
	unsigned long flags;
	int ret;

2401 2402 2403
	if (cpufreq_disabled())
		return -ENODEV;

L
Linus Torvalds 已提交
2404
	if (!driver_data || !driver_data->verify || !driver_data->init ||
2405
	    !(driver_data->setpolicy || driver_data->target_index ||
2406 2407
		    driver_data->target) ||
	     (driver_data->setpolicy && (driver_data->target_index ||
2408 2409
		    driver_data->target)) ||
	     (!!driver_data->get_intermediate != !!driver_data->target_intermediate))
L
Linus Torvalds 已提交
2410 2411
		return -EINVAL;

2412
	pr_debug("trying to register driver %s\n", driver_data->name);
L
Linus Torvalds 已提交
2413 2414 2415 2416

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

2417
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2418
	if (cpufreq_driver) {
2419
		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2420
		return -EEXIST;
L
Linus Torvalds 已提交
2421
	}
2422
	cpufreq_driver = driver_data;
2423
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
2424

2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435
	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",
2436
			       __func__);
2437 2438 2439 2440
			goto err_null_driver;
		}
	}

2441
	ret = subsys_interface_register(&cpufreq_interface);
2442
	if (ret)
2443
		goto err_boost_unreg;
L
Linus Torvalds 已提交
2444

2445
	if (!(cpufreq_driver->flags & CPUFREQ_STICKY)) {
L
Linus Torvalds 已提交
2446 2447 2448 2449
		int i;
		ret = -ENODEV;

		/* check for at least one working CPU */
2450 2451
		for (i = 0; i < nr_cpu_ids; i++)
			if (cpu_possible(i) && per_cpu(cpufreq_cpu_data, i)) {
L
Linus Torvalds 已提交
2452
				ret = 0;
2453 2454
				break;
			}
L
Linus Torvalds 已提交
2455 2456 2457

		/* if all ->init() calls failed, unregister */
		if (ret) {
2458
			pr_debug("no CPU initialized for driver %s\n",
2459
				 driver_data->name);
2460
			goto err_if_unreg;
L
Linus Torvalds 已提交
2461 2462 2463
		}
	}

2464
	register_hotcpu_notifier(&cpufreq_cpu_notifier);
2465
	pr_debug("driver %s up and running\n", driver_data->name);
L
Linus Torvalds 已提交
2466

2467
	return 0;
2468 2469
err_if_unreg:
	subsys_interface_unregister(&cpufreq_interface);
2470 2471 2472
err_boost_unreg:
	if (cpufreq_boost_supported())
		cpufreq_sysfs_remove_file(&boost.attr);
2473
err_null_driver:
2474
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2475
	cpufreq_driver = NULL;
2476
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
D
Dave Jones 已提交
2477
	return ret;
L
Linus Torvalds 已提交
2478 2479 2480 2481 2482 2483
}
EXPORT_SYMBOL_GPL(cpufreq_register_driver);

/**
 * cpufreq_unregister_driver - unregister the current CPUFreq driver
 *
2484
 * Unregister the current CPUFreq driver. Only call this if you have
L
Linus Torvalds 已提交
2485 2486 2487 2488
 * 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.
 */
2489
int cpufreq_unregister_driver(struct cpufreq_driver *driver)
L
Linus Torvalds 已提交
2490 2491 2492
{
	unsigned long flags;

2493
	if (!cpufreq_driver || (driver != cpufreq_driver))
L
Linus Torvalds 已提交
2494 2495
		return -EINVAL;

2496
	pr_debug("unregistering driver %s\n", driver->name);
L
Linus Torvalds 已提交
2497

2498
	subsys_interface_unregister(&cpufreq_interface);
2499 2500 2501
	if (cpufreq_boost_supported())
		cpufreq_sysfs_remove_file(&boost.attr);

2502
	unregister_hotcpu_notifier(&cpufreq_cpu_notifier);
L
Linus Torvalds 已提交
2503

2504
	down_write(&cpufreq_rwsem);
2505
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2506

2507
	cpufreq_driver = NULL;
2508

2509
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2510
	up_write(&cpufreq_rwsem);
L
Linus Torvalds 已提交
2511 2512 2513 2514

	return 0;
}
EXPORT_SYMBOL_GPL(cpufreq_unregister_driver);
2515 2516 2517

static int __init cpufreq_core_init(void)
{
2518 2519 2520
	if (cpufreq_disabled())
		return -ENODEV;

2521
	cpufreq_global_kobject = kobject_create();
2522 2523
	BUG_ON(!cpufreq_global_kobject);

2524 2525 2526
	return 0;
}
core_initcall(cpufreq_core_init);