cpufreq.c 64.7 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5
/*
 *  linux/drivers/cpufreq/cpufreq.c
 *
 *  Copyright (C) 2001 Russell King
 *            (C) 2002 - 2003 Dominik Brodowski <linux@brodo.de>
6
 *            (C) 2013 Viresh Kumar <viresh.kumar@linaro.org>
L
Linus Torvalds 已提交
7
 *
8
 *  Oct 2005 - Ashok Raj <ashok.raj@intel.com>
9
 *	Added handling for CPU hotplug
10 11
 *  Feb 2006 - Jacob Shin <jacob.shin@amd.com>
 *	Fix handling for CPU hotplug -- affected CPUs
12
 *
L
Linus Torvalds 已提交
13 14 15 16 17
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 */

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

20
#include <linux/cpu.h>
L
Linus Torvalds 已提交
21 22 23
#include <linux/cpufreq.h>
#include <linux/delay.h>
#include <linux/device.h>
24 25 26
#include <linux/init.h>
#include <linux/kernel_stat.h>
#include <linux/module.h>
27
#include <linux/mutex.h>
28
#include <linux/slab.h>
29
#include <linux/suspend.h>
30
#include <linux/syscore_ops.h>
31
#include <linux/tick.h>
32 33
#include <trace/events/power.h>

L
Linus Torvalds 已提交
34
/**
D
Dave Jones 已提交
35
 * The "cpufreq driver" - the arch- or hardware-dependent low
L
Linus Torvalds 已提交
36 37 38
 * level driver of CPUFreq support, and its spinlock. This lock
 * also protects the cpufreq_cpu_data array.
 */
39
static struct cpufreq_driver *cpufreq_driver;
40
static DEFINE_PER_CPU(struct cpufreq_policy *, cpufreq_cpu_data);
41
static DEFINE_PER_CPU(struct cpufreq_policy *, cpufreq_cpu_data_fallback);
42
static DEFINE_RWLOCK(cpufreq_driver_lock);
43
DEFINE_MUTEX(cpufreq_governor_lock);
44
static LIST_HEAD(cpufreq_policy_list);
45

46
/* This one keeps track of the previously set governor of a removed CPU */
47
static DEFINE_PER_CPU(char[CPUFREQ_NAME_LEN], cpufreq_cpu_governor);
L
Linus Torvalds 已提交
48

49 50
/* Flag to suspend/resume CPUFreq governors */
static bool cpufreq_suspended;
L
Linus Torvalds 已提交
51

52 53 54 55 56
static inline bool has_target(void)
{
	return cpufreq_driver->target_index || cpufreq_driver->target;
}

57 58 59 60 61 62
/*
 * rwsem to guarantee that cpufreq driver module doesn't unload during critical
 * sections
 */
static DECLARE_RWSEM(cpufreq_rwsem);

L
Linus Torvalds 已提交
63
/* internal prototypes */
64 65
static int __cpufreq_governor(struct cpufreq_policy *policy,
		unsigned int event);
66
static unsigned int __cpufreq_get(struct cpufreq_policy *policy);
67
static void handle_update(struct work_struct *work);
L
Linus Torvalds 已提交
68 69

/**
70 71
 * 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
72 73 74 75
 * "transition" list for kernel code that needs to handle
 * changes to devices when the CPU clock speed changes.
 * The mutex locks both lists.
 */
76
static BLOCKING_NOTIFIER_HEAD(cpufreq_policy_notifier_list);
77
static struct srcu_notifier_head cpufreq_transition_notifier_list;
L
Linus Torvalds 已提交
78

79
static bool init_cpufreq_transition_notifier_list_called;
80 81 82
static int __init init_cpufreq_transition_notifier_list(void)
{
	srcu_init_notifier_head(&cpufreq_transition_notifier_list);
83
	init_cpufreq_transition_notifier_list_called = true;
84 85
	return 0;
}
86
pure_initcall(init_cpufreq_transition_notifier_list);
L
Linus Torvalds 已提交
87

88
static int off __read_mostly;
89
static int cpufreq_disabled(void)
90 91 92 93 94 95 96
{
	return off;
}
void disable_cpufreq(void)
{
	off = 1;
}
L
Linus Torvalds 已提交
97
static LIST_HEAD(cpufreq_governor_list);
98
static DEFINE_MUTEX(cpufreq_governor_mutex);
L
Linus Torvalds 已提交
99

100 101
bool have_governor_per_policy(void)
{
102
	return !!(cpufreq_driver->flags & CPUFREQ_HAVE_GOVERNOR_PER_POLICY);
103
}
104
EXPORT_SYMBOL_GPL(have_governor_per_policy);
105

106 107 108 109 110 111 112 113 114
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);

115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149
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);

150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180
/*
 * 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);

181 182 183 184 185
unsigned int cpufreq_generic_get(unsigned int cpu)
{
	struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);

	if (!policy || IS_ERR(policy->clk)) {
186 187
		pr_err("%s: No %s associated to cpu: %d\n",
		       __func__, policy ? "clk" : "policy", cpu);
188 189 190 191 192 193 194
		return 0;
	}

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

195 196 197 198 199 200
/* Only for cpufreq core internal use */
struct cpufreq_policy *cpufreq_cpu_get_raw(unsigned int cpu)
{
	return per_cpu(cpufreq_cpu_data, cpu);
}

201
struct cpufreq_policy *cpufreq_cpu_get(unsigned int cpu)
L
Linus Torvalds 已提交
202
{
203
	struct cpufreq_policy *policy = NULL;
L
Linus Torvalds 已提交
204 205
	unsigned long flags;

206 207 208 209 210
	if (cpufreq_disabled() || (cpu >= nr_cpu_ids))
		return NULL;

	if (!down_read_trylock(&cpufreq_rwsem))
		return NULL;
L
Linus Torvalds 已提交
211 212

	/* get the cpufreq driver */
213
	read_lock_irqsave(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
214

215 216 217 218 219 220
	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
221

222
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
223

224
	if (!policy)
225
		up_read(&cpufreq_rwsem);
L
Linus Torvalds 已提交
226

227
	return policy;
228
}
L
Linus Torvalds 已提交
229 230
EXPORT_SYMBOL_GPL(cpufreq_cpu_get);

231
void cpufreq_cpu_put(struct cpufreq_policy *policy)
L
Linus Torvalds 已提交
232
{
233 234 235
	if (cpufreq_disabled())
		return;

236 237
	kobject_put(&policy->kobj);
	up_read(&cpufreq_rwsem);
L
Linus Torvalds 已提交
238 239 240 241 242 243 244 245 246 247 248 249
}
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
250
 * systems as each CPU might be scaled differently. So, use the arch
L
Linus Torvalds 已提交
251 252
 * per-CPU loops_per_jiffy value wherever possible.
 */
253
static void adjust_jiffies(unsigned long val, struct cpufreq_freqs *ci)
L
Linus Torvalds 已提交
254
{
255 256 257 258
#ifndef CONFIG_SMP
	static unsigned long l_p_j_ref;
	static unsigned int l_p_j_ref_freq;

L
Linus Torvalds 已提交
259 260 261 262 263 264
	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;
265 266
		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
267
	}
268
	if (val == CPUFREQ_POSTCHANGE && ci->old != ci->new) {
269 270
		loops_per_jiffy = cpufreq_scale(l_p_j_ref, l_p_j_ref_freq,
								ci->new);
271 272
		pr_debug("scaling loops_per_jiffy to %lu for frequency %u kHz\n",
			 loops_per_jiffy, ci->new);
L
Linus Torvalds 已提交
273 274
	}
#endif
275
}
L
Linus Torvalds 已提交
276

277
static void __cpufreq_notify_transition(struct cpufreq_policy *policy,
278
		struct cpufreq_freqs *freqs, unsigned int state)
L
Linus Torvalds 已提交
279 280 281
{
	BUG_ON(irqs_disabled());

282 283 284
	if (cpufreq_disabled())
		return;

285
	freqs->flags = cpufreq_driver->flags;
286
	pr_debug("notification %u of frequency transition to %u kHz\n",
287
		 state, freqs->new);
L
Linus Torvalds 已提交
288 289

	switch (state) {
290

L
Linus Torvalds 已提交
291
	case CPUFREQ_PRECHANGE:
292
		/* detect if the driver reported a value as "old frequency"
293 294
		 * which is not equal to what the cpufreq core thinks is
		 * "old frequency".
L
Linus Torvalds 已提交
295
		 */
296
		if (!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
297 298
			if ((policy) && (policy->cpu == freqs->cpu) &&
			    (policy->cur) && (policy->cur != freqs->old)) {
299 300
				pr_debug("Warning: CPU frequency is %u, cpufreq assumed %u kHz\n",
					 freqs->old, policy->cur);
301
				freqs->old = policy->cur;
L
Linus Torvalds 已提交
302 303
			}
		}
304
		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
305
				CPUFREQ_PRECHANGE, freqs);
L
Linus Torvalds 已提交
306 307
		adjust_jiffies(CPUFREQ_PRECHANGE, freqs);
		break;
308

L
Linus Torvalds 已提交
309 310
	case CPUFREQ_POSTCHANGE:
		adjust_jiffies(CPUFREQ_POSTCHANGE, freqs);
311 312
		pr_debug("FREQ: %lu - CPU: %lu\n",
			 (unsigned long)freqs->new, (unsigned long)freqs->cpu);
313
		trace_cpu_frequency(freqs->new, freqs->cpu);
314
		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
315
				CPUFREQ_POSTCHANGE, freqs);
316 317
		if (likely(policy) && likely(policy->cpu == freqs->cpu))
			policy->cur = freqs->new;
L
Linus Torvalds 已提交
318 319 320
		break;
	}
}
321

322 323 324 325 326 327 328 329
/**
 * 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.
 */
330
static void cpufreq_notify_transition(struct cpufreq_policy *policy,
331 332 333 334 335
		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
336

337
/* Do post notifications when there are chances that transition has failed */
338
static void cpufreq_notify_post_transition(struct cpufreq_policy *policy,
339 340 341 342 343 344 345 346 347 348 349
		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);
}

350 351 352
void cpufreq_freq_transition_begin(struct cpufreq_policy *policy,
		struct cpufreq_freqs *freqs)
{
353 354 355 356 357 358 359 360 361 362 363 364

	/*
	 * 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);

365 366 367 368 369 370 371 372 373 374 375
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;
376
	policy->transition_task = current;
377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392

	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;
393
	policy->transition_task = NULL;
394 395 396 397 398

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

L
Linus Torvalds 已提交
399 400 401 402

/*********************************************************************
 *                          SYSFS INTERFACE                          *
 *********************************************************************/
403
static ssize_t show_boost(struct kobject *kobj,
404 405 406 407 408 409 410 411 412 413 414 415 416 417 418
				 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)) {
419 420
		pr_err("%s: Cannot %s BOOST!\n",
		       __func__, enable ? "enable" : "disable");
421 422 423
		return -EINVAL;
	}

424 425
	pr_debug("%s: cpufreq BOOST %s\n",
		 __func__, enable ? "enabled" : "disabled");
426 427 428 429

	return count;
}
define_one_global_rw(boost);
L
Linus Torvalds 已提交
430

431
static struct cpufreq_governor *find_governor(const char *str_governor)
432 433 434 435
{
	struct cpufreq_governor *t;

	list_for_each_entry(t, &cpufreq_governor_list, governor_list)
436
		if (!strncasecmp(str_governor, t->name, CPUFREQ_NAME_LEN))
437 438 439 440 441
			return t;

	return NULL;
}

L
Linus Torvalds 已提交
442 443 444
/**
 * cpufreq_parse_governor - parse a governor string
 */
D
Dave Jones 已提交
445
static int cpufreq_parse_governor(char *str_governor, unsigned int *policy,
L
Linus Torvalds 已提交
446 447
				struct cpufreq_governor **governor)
{
448
	int err = -EINVAL;
449 450

	if (!cpufreq_driver)
451 452
		goto out;

453
	if (cpufreq_driver->setpolicy) {
454
		if (!strncasecmp(str_governor, "performance", CPUFREQ_NAME_LEN)) {
L
Linus Torvalds 已提交
455
			*policy = CPUFREQ_POLICY_PERFORMANCE;
456
			err = 0;
457
		} else if (!strncasecmp(str_governor, "powersave",
458
						CPUFREQ_NAME_LEN)) {
L
Linus Torvalds 已提交
459
			*policy = CPUFREQ_POLICY_POWERSAVE;
460
			err = 0;
L
Linus Torvalds 已提交
461
		}
462
	} else {
L
Linus Torvalds 已提交
463
		struct cpufreq_governor *t;
464

465
		mutex_lock(&cpufreq_governor_mutex);
466

467
		t = find_governor(str_governor);
468

469
		if (t == NULL) {
470
			int ret;
471

472 473 474
			mutex_unlock(&cpufreq_governor_mutex);
			ret = request_module("cpufreq_%s", str_governor);
			mutex_lock(&cpufreq_governor_mutex);
475

476
			if (ret == 0)
477
				t = find_governor(str_governor);
478 479
		}

480 481 482
		if (t != NULL) {
			*governor = t;
			err = 0;
L
Linus Torvalds 已提交
483
		}
484

485
		mutex_unlock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
486
	}
487
out:
488
	return err;
L
Linus Torvalds 已提交
489 490 491
}

/**
492 493
 * cpufreq_per_cpu_attr_read() / show_##file_name() -
 * print out cpufreq information
L
Linus Torvalds 已提交
494 495 496 497 498
 *
 * Write out information from cpufreq_driver->policy[cpu]; object must be
 * "unsigned int".
 */

499 500
#define show_one(file_name, object)			\
static ssize_t show_##file_name				\
D
Dave Jones 已提交
501
(struct cpufreq_policy *policy, char *buf)		\
502
{							\
503
	return sprintf(buf, "%u\n", policy->object);	\
L
Linus Torvalds 已提交
504 505 506 507
}

show_one(cpuinfo_min_freq, cpuinfo.min_freq);
show_one(cpuinfo_max_freq, cpuinfo.max_freq);
508
show_one(cpuinfo_transition_latency, cpuinfo.transition_latency);
L
Linus Torvalds 已提交
509 510
show_one(scaling_min_freq, min);
show_one(scaling_max_freq, max);
511

512
static ssize_t show_scaling_cur_freq(struct cpufreq_policy *policy, char *buf)
513 514 515 516 517 518 519 520 521
{
	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
522

523
static int cpufreq_set_policy(struct cpufreq_policy *policy,
524
				struct cpufreq_policy *new_policy);
525

L
Linus Torvalds 已提交
526 527 528 529 530
/**
 * 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
531
(struct cpufreq_policy *policy, const char *buf, size_t count)		\
L
Linus Torvalds 已提交
532
{									\
533
	int ret, temp;							\
L
Linus Torvalds 已提交
534 535 536 537 538 539
	struct cpufreq_policy new_policy;				\
									\
	ret = cpufreq_get_policy(&new_policy, policy->cpu);		\
	if (ret)							\
		return -EINVAL;						\
									\
540
	ret = sscanf(buf, "%u", &new_policy.object);			\
L
Linus Torvalds 已提交
541 542 543
	if (ret != 1)							\
		return -EINVAL;						\
									\
544
	temp = new_policy.object;					\
545
	ret = cpufreq_set_policy(policy, &new_policy);		\
546 547
	if (!ret)							\
		policy->user_policy.object = temp;			\
L
Linus Torvalds 已提交
548 549 550 551
									\
	return ret ? ret : count;					\
}

552 553
store_one(scaling_min_freq, min);
store_one(scaling_max_freq, max);
L
Linus Torvalds 已提交
554 555 556 557

/**
 * show_cpuinfo_cur_freq - current CPU frequency as detected by hardware
 */
D
Dave Jones 已提交
558 559
static ssize_t show_cpuinfo_cur_freq(struct cpufreq_policy *policy,
					char *buf)
L
Linus Torvalds 已提交
560
{
561
	unsigned int cur_freq = __cpufreq_get(policy);
L
Linus Torvalds 已提交
562 563 564 565 566 567 568 569
	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
570
static ssize_t show_scaling_governor(struct cpufreq_policy *policy, char *buf)
L
Linus Torvalds 已提交
571
{
572
	if (policy->policy == CPUFREQ_POLICY_POWERSAVE)
L
Linus Torvalds 已提交
573 574 575 576
		return sprintf(buf, "powersave\n");
	else if (policy->policy == CPUFREQ_POLICY_PERFORMANCE)
		return sprintf(buf, "performance\n");
	else if (policy->governor)
577
		return scnprintf(buf, CPUFREQ_NAME_PLEN, "%s\n",
578
				policy->governor->name);
L
Linus Torvalds 已提交
579 580 581 582 583 584
	return -EINVAL;
}

/**
 * store_scaling_governor - store policy for the specified CPU
 */
D
Dave Jones 已提交
585 586
static ssize_t store_scaling_governor(struct cpufreq_policy *policy,
					const char *buf, size_t count)
L
Linus Torvalds 已提交
587
{
588
	int ret;
L
Linus Torvalds 已提交
589 590 591 592 593 594 595
	char	str_governor[16];
	struct cpufreq_policy new_policy;

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

596
	ret = sscanf(buf, "%15s", str_governor);
L
Linus Torvalds 已提交
597 598 599
	if (ret != 1)
		return -EINVAL;

600 601
	if (cpufreq_parse_governor(str_governor, &new_policy.policy,
						&new_policy.governor))
L
Linus Torvalds 已提交
602 603
		return -EINVAL;

604
	ret = cpufreq_set_policy(policy, &new_policy);
605 606 607 608

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

609 610 611 612
	if (ret)
		return ret;
	else
		return count;
L
Linus Torvalds 已提交
613 614 615 616 617
}

/**
 * show_scaling_driver - show the cpufreq driver currently loaded
 */
D
Dave Jones 已提交
618
static ssize_t show_scaling_driver(struct cpufreq_policy *policy, char *buf)
L
Linus Torvalds 已提交
619
{
620
	return scnprintf(buf, CPUFREQ_NAME_PLEN, "%s\n", cpufreq_driver->name);
L
Linus Torvalds 已提交
621 622 623 624 625
}

/**
 * show_scaling_available_governors - show the available CPUfreq governors
 */
D
Dave Jones 已提交
626 627
static ssize_t show_scaling_available_governors(struct cpufreq_policy *policy,
						char *buf)
L
Linus Torvalds 已提交
628 629 630 631
{
	ssize_t i = 0;
	struct cpufreq_governor *t;

632
	if (!has_target()) {
L
Linus Torvalds 已提交
633 634 635 636 637
		i += sprintf(buf, "performance powersave");
		goto out;
	}

	list_for_each_entry(t, &cpufreq_governor_list, governor_list) {
638 639
		if (i >= (ssize_t) ((PAGE_SIZE / sizeof(char))
		    - (CPUFREQ_NAME_LEN + 2)))
L
Linus Torvalds 已提交
640
			goto out;
641
		i += scnprintf(&buf[i], CPUFREQ_NAME_PLEN, "%s ", t->name);
L
Linus Torvalds 已提交
642
	}
643
out:
L
Linus Torvalds 已提交
644 645 646
	i += sprintf(&buf[i], "\n");
	return i;
}
647

648
ssize_t cpufreq_show_cpus(const struct cpumask *mask, char *buf)
L
Linus Torvalds 已提交
649 650 651 652
{
	ssize_t i = 0;
	unsigned int cpu;

653
	for_each_cpu(cpu, mask) {
L
Linus Torvalds 已提交
654 655 656 657
		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))
658
			break;
L
Linus Torvalds 已提交
659 660 661 662
	}
	i += sprintf(&buf[i], "\n");
	return i;
}
663
EXPORT_SYMBOL_GPL(cpufreq_show_cpus);
L
Linus Torvalds 已提交
664

665 666 667 668 669 670
/**
 * 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)
{
671
	return cpufreq_show_cpus(policy->related_cpus, buf);
672 673 674 675 676 677 678
}

/**
 * show_affected_cpus - show the CPUs affected by each transition
 */
static ssize_t show_affected_cpus(struct cpufreq_policy *policy, char *buf)
{
679
	return cpufreq_show_cpus(policy->cpus, buf);
680 681
}

682
static ssize_t store_scaling_setspeed(struct cpufreq_policy *policy,
D
Dave Jones 已提交
683
					const char *buf, size_t count)
684 685 686 687
{
	unsigned int freq = 0;
	unsigned int ret;

688
	if (!policy->governor || !policy->governor->store_setspeed)
689 690 691 692 693 694 695 696 697 698 699 700 701
		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)
{
702
	if (!policy->governor || !policy->governor->show_setspeed)
703 704 705 706
		return sprintf(buf, "<unsupported>\n");

	return policy->governor->show_setspeed(policy, buf);
}
L
Linus Torvalds 已提交
707

708
/**
709
 * show_bios_limit - show the current cpufreq HW/BIOS limitation
710 711 712 713 714
 */
static ssize_t show_bios_limit(struct cpufreq_policy *policy, char *buf)
{
	unsigned int limit;
	int ret;
715 716
	if (cpufreq_driver->bios_limit) {
		ret = cpufreq_driver->bios_limit(policy->cpu, &limit);
717 718 719 720 721 722
		if (!ret)
			return sprintf(buf, "%u\n", limit);
	}
	return sprintf(buf, "%u\n", policy->cpuinfo.max_freq);
}

723 724 725 726 727 728 729 730 731 732 733 734 735 736
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
737

D
Dave Jones 已提交
738
static struct attribute *default_attrs[] = {
L
Linus Torvalds 已提交
739 740
	&cpuinfo_min_freq.attr,
	&cpuinfo_max_freq.attr,
741
	&cpuinfo_transition_latency.attr,
L
Linus Torvalds 已提交
742 743 744
	&scaling_min_freq.attr,
	&scaling_max_freq.attr,
	&affected_cpus.attr,
745
	&related_cpus.attr,
L
Linus Torvalds 已提交
746 747 748
	&scaling_governor.attr,
	&scaling_driver.attr,
	&scaling_available_governors.attr,
749
	&scaling_setspeed.attr,
L
Linus Torvalds 已提交
750 751 752
	NULL
};

753 754
#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
755

756
static ssize_t show(struct kobject *kobj, struct attribute *attr, char *buf)
L
Linus Torvalds 已提交
757
{
D
Dave Jones 已提交
758 759
	struct cpufreq_policy *policy = to_policy(kobj);
	struct freq_attr *fattr = to_attr(attr);
760
	ssize_t ret;
761 762

	if (!down_read_trylock(&cpufreq_rwsem))
763
		return -EINVAL;
764

765
	down_read(&policy->rwsem);
766

767 768 769 770 771
	if (fattr->show)
		ret = fattr->show(policy, buf);
	else
		ret = -EIO;

772
	up_read(&policy->rwsem);
773
	up_read(&cpufreq_rwsem);
774

L
Linus Torvalds 已提交
775 776 777
	return ret;
}

D
Dave Jones 已提交
778 779
static ssize_t store(struct kobject *kobj, struct attribute *attr,
		     const char *buf, size_t count)
L
Linus Torvalds 已提交
780
{
D
Dave Jones 已提交
781 782
	struct cpufreq_policy *policy = to_policy(kobj);
	struct freq_attr *fattr = to_attr(attr);
783
	ssize_t ret = -EINVAL;
784

785 786 787 788 789
	get_online_cpus();

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

790
	if (!down_read_trylock(&cpufreq_rwsem))
791
		goto unlock;
792

793
	down_write(&policy->rwsem);
794

795 796 797 798 799
	if (fattr->store)
		ret = fattr->store(policy, buf, count);
	else
		ret = -EIO;

800
	up_write(&policy->rwsem);
801 802

	up_read(&cpufreq_rwsem);
803 804 805
unlock:
	put_online_cpus();

L
Linus Torvalds 已提交
806 807 808
	return ret;
}

D
Dave Jones 已提交
809
static void cpufreq_sysfs_release(struct kobject *kobj)
L
Linus Torvalds 已提交
810
{
D
Dave Jones 已提交
811
	struct cpufreq_policy *policy = to_policy(kobj);
812
	pr_debug("last reference is dropped\n");
L
Linus Torvalds 已提交
813 814 815
	complete(&policy->kobj_unregister);
}

816
static const struct sysfs_ops sysfs_ops = {
L
Linus Torvalds 已提交
817 818 819 820 821 822 823 824 825 826
	.show	= show,
	.store	= store,
};

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

827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869
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);

870
/* symlink affected CPUs */
871
static int cpufreq_add_dev_symlink(struct cpufreq_policy *policy)
872 873 874 875 876
{
	unsigned int j;
	int ret = 0;

	for_each_cpu(j, policy->cpus) {
877
		struct device *cpu_dev;
878

879
		if (j == policy->cpu)
880 881
			continue;

882
		pr_debug("Adding link for CPU: %u\n", j);
883 884
		cpu_dev = get_cpu_device(j);
		ret = sysfs_create_link(&cpu_dev->kobj, &policy->kobj,
885
					"cpufreq");
886 887
		if (ret)
			break;
888 889 890 891
	}
	return ret;
}

892
static int cpufreq_add_dev_interface(struct cpufreq_policy *policy,
893
				     struct device *dev)
894 895 896 897 898
{
	struct freq_attr **drv_attr;
	int ret = 0;

	/* set up files for this cpu device */
899
	drv_attr = cpufreq_driver->attr;
V
Viresh Kumar 已提交
900
	while (drv_attr && *drv_attr) {
901 902
		ret = sysfs_create_file(&policy->kobj, &((*drv_attr)->attr));
		if (ret)
903
			return ret;
904 905
		drv_attr++;
	}
906
	if (cpufreq_driver->get) {
907 908
		ret = sysfs_create_file(&policy->kobj, &cpuinfo_cur_freq.attr);
		if (ret)
909
			return ret;
910
	}
911 912 913

	ret = sysfs_create_file(&policy->kobj, &scaling_cur_freq.attr);
	if (ret)
914
		return ret;
915

916
	if (cpufreq_driver->bios_limit) {
917 918
		ret = sysfs_create_file(&policy->kobj, &bios_limit.attr);
		if (ret)
919
			return ret;
920
	}
921

922
	return cpufreq_add_dev_symlink(policy);
923 924 925 926
}

static void cpufreq_init_policy(struct cpufreq_policy *policy)
{
927
	struct cpufreq_governor *gov = NULL;
928 929 930
	struct cpufreq_policy new_policy;
	int ret = 0;

931
	memcpy(&new_policy, policy, sizeof(*policy));
932

933
	/* Update governor of new_policy to the governor used before hotplug */
934
	gov = find_governor(per_cpu(cpufreq_cpu_governor, policy->cpu));
935 936 937 938 939 940 941 942
	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;

943 944
	/* Use the default policy if its valid. */
	if (cpufreq_driver->setpolicy)
945
		cpufreq_parse_governor(gov->name, &new_policy.policy, NULL);
946 947

	/* set default policy */
948
	ret = cpufreq_set_policy(policy, &new_policy);
949
	if (ret) {
950
		pr_debug("setting policy failed\n");
951 952
		if (cpufreq_driver->exit)
			cpufreq_driver->exit(policy);
953
	}
954 955
}

956
static int cpufreq_add_policy_cpu(struct cpufreq_policy *policy,
957
				  unsigned int cpu, struct device *dev)
958
{
959
	int ret = 0;
960 961
	unsigned long flags;

962
	if (has_target()) {
963 964 965 966 967 968
		ret = __cpufreq_governor(policy, CPUFREQ_GOV_STOP);
		if (ret) {
			pr_err("%s: Failed to stop governor\n", __func__);
			return ret;
		}
	}
969

970
	down_write(&policy->rwsem);
V
Viresh Kumar 已提交
971

972
	write_lock_irqsave(&cpufreq_driver_lock, flags);
V
Viresh Kumar 已提交
973

974 975
	cpumask_set_cpu(cpu, policy->cpus);
	per_cpu(cpufreq_cpu_data, cpu) = policy;
976
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
977

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

980
	if (has_target()) {
981 982 983 984 985
		ret = __cpufreq_governor(policy, CPUFREQ_GOV_START);
		if (!ret)
			ret = __cpufreq_governor(policy, CPUFREQ_GOV_LIMITS);

		if (ret) {
986 987 988
			pr_err("%s: Failed to start governor\n", __func__);
			return ret;
		}
989
	}
990

991
	return sysfs_create_link(&dev->kobj, &policy->kobj, "cpufreq");
992
}
L
Linus Torvalds 已提交
993

994 995 996 997 998
static struct cpufreq_policy *cpufreq_policy_restore(unsigned int cpu)
{
	struct cpufreq_policy *policy;
	unsigned long flags;

999
	read_lock_irqsave(&cpufreq_driver_lock, flags);
1000 1001 1002

	policy = per_cpu(cpufreq_cpu_data_fallback, cpu);

1003
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
1004

1005 1006
	if (policy)
		policy->governor = NULL;
1007

1008 1009 1010
	return policy;
}

1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024
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;

1025
	INIT_LIST_HEAD(&policy->policy_list);
1026
	init_rwsem(&policy->rwsem);
1027 1028
	spin_lock_init(&policy->transition_lock);
	init_waitqueue_head(&policy->transition_wait);
1029

1030 1031 1032 1033 1034 1035 1036 1037 1038 1039
	return policy;

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

	return NULL;
}

1040 1041 1042 1043 1044
static void cpufreq_policy_put_kobj(struct cpufreq_policy *policy)
{
	struct kobject *kobj;
	struct completion *cmp;

1045 1046 1047
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
			CPUFREQ_REMOVE_POLICY, policy);

1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063
	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");
}

1064 1065 1066 1067 1068 1069 1070
static void cpufreq_policy_free(struct cpufreq_policy *policy)
{
	free_cpumask_var(policy->related_cpus);
	free_cpumask_var(policy->cpus);
	kfree(policy);
}

1071 1072
static int update_policy_cpu(struct cpufreq_policy *policy, unsigned int cpu,
			     struct device *cpu_dev)
1073
{
1074 1075
	int ret;

1076
	if (WARN_ON(cpu == policy->cpu))
1077 1078 1079 1080 1081 1082 1083 1084
		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;
	}
1085

1086
	down_write(&policy->rwsem);
1087

1088 1089 1090
	policy->last_cpu = policy->cpu;
	policy->cpu = cpu;

1091
	up_write(&policy->rwsem);
1092

1093 1094
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
			CPUFREQ_UPDATE_POLICY_CPU, policy);
1095 1096

	return 0;
1097 1098
}

1099
static int __cpufreq_add_dev(struct device *dev, struct subsys_interface *sif)
L
Linus Torvalds 已提交
1100
{
1101
	unsigned int j, cpu = dev->id;
1102
	int ret = -ENOMEM;
1103
	struct cpufreq_policy *policy;
L
Linus Torvalds 已提交
1104
	unsigned long flags;
1105
	bool recover_policy = cpufreq_suspended;
L
Linus Torvalds 已提交
1106

1107 1108 1109
	if (cpu_is_offline(cpu))
		return 0;

1110
	pr_debug("adding CPU %u\n", cpu);
L
Linus Torvalds 已提交
1111 1112 1113

	/* check whether a different CPU already registered this
	 * CPU because it is in the same boat. */
1114 1115
	policy = cpufreq_cpu_get_raw(cpu);
	if (unlikely(policy))
L
Linus Torvalds 已提交
1116
		return 0;
1117

1118 1119 1120
	if (!down_read_trylock(&cpufreq_rwsem))
		return 0;

1121
	/* Check if this cpu was hot-unplugged earlier and has siblings */
1122
	read_lock_irqsave(&cpufreq_driver_lock, flags);
1123 1124
	list_for_each_entry(policy, &cpufreq_policy_list, policy_list) {
		if (cpumask_test_cpu(cpu, policy->related_cpus)) {
1125
			read_unlock_irqrestore(&cpufreq_driver_lock, flags);
1126
			ret = cpufreq_add_policy_cpu(policy, cpu, dev);
1127 1128
			up_read(&cpufreq_rwsem);
			return ret;
V
Viresh Kumar 已提交
1129
		}
1130
	}
1131
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
1132

1133 1134 1135 1136
	/*
	 * Restore the saved policy when doing light-weight init and fall back
	 * to the full init if that fails.
	 */
1137
	policy = recover_policy ? cpufreq_policy_restore(cpu) : NULL;
1138
	if (!policy) {
1139
		recover_policy = false;
1140
		policy = cpufreq_policy_alloc();
1141 1142 1143
		if (!policy)
			goto nomem_out;
	}
1144 1145 1146 1147 1148 1149 1150

	/*
	 * 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().
	 */
1151 1152 1153
	if (recover_policy && cpu != policy->cpu)
		WARN_ON(update_policy_cpu(policy, cpu, dev));
	else
1154 1155
		policy->cpu = cpu;

1156
	cpumask_copy(policy->cpus, cpumask_of(cpu));
L
Linus Torvalds 已提交
1157 1158

	init_completion(&policy->kobj_unregister);
1159
	INIT_WORK(&policy->update, handle_update);
L
Linus Torvalds 已提交
1160 1161 1162 1163

	/* call driver. From then on the cpufreq must be able
	 * to accept all calls to ->verify and ->setpolicy for this CPU
	 */
1164
	ret = cpufreq_driver->init(policy);
L
Linus Torvalds 已提交
1165
	if (ret) {
1166
		pr_debug("initialization failed\n");
V
Viresh Kumar 已提交
1167
		goto err_set_policy_cpu;
L
Linus Torvalds 已提交
1168
	}
V
Viresh Kumar 已提交
1169

1170 1171
	down_write(&policy->rwsem);

1172 1173 1174 1175 1176 1177 1178 1179 1180
	/* 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);

1181
	if (!recover_policy) {
1182 1183
		policy->user_policy.min = policy->min;
		policy->user_policy.max = policy->max;
1184 1185 1186 1187 1188 1189 1190 1191 1192

		/* prepare interface data */
		ret = kobject_init_and_add(&policy->kobj, &ktype_cpufreq,
					   &dev->kobj, "cpufreq");
		if (ret) {
			pr_err("%s: failed to init policy->kobj: %d\n",
			       __func__, ret);
			goto err_init_policy_kobj;
		}
1193 1194
	}

1195 1196 1197 1198 1199
	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);

1200
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
1201 1202 1203 1204 1205 1206 1207
		policy->cur = cpufreq_driver->get(policy->cpu);
		if (!policy->cur) {
			pr_err("%s: ->get() failed\n", __func__);
			goto err_get_freq;
		}
	}

1208 1209 1210 1211 1212 1213 1214 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
	/*
	 * 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);
		}
	}

1248 1249 1250
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
				     CPUFREQ_START, policy);

1251
	if (!recover_policy) {
1252
		ret = cpufreq_add_dev_interface(policy, dev);
1253 1254
		if (ret)
			goto err_out_unregister;
1255 1256
		blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
				CPUFREQ_CREATE_POLICY, policy);
1257
	}
1258

1259 1260 1261 1262
	write_lock_irqsave(&cpufreq_driver_lock, flags);
	list_add(&policy->policy_list, &cpufreq_policy_list);
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);

1263 1264
	cpufreq_init_policy(policy);

1265
	if (!recover_policy) {
1266 1267 1268
		policy->user_policy.policy = policy->policy;
		policy->user_policy.governor = policy->governor;
	}
1269
	up_write(&policy->rwsem);
1270

1271
	kobject_uevent(&policy->kobj, KOBJ_ADD);
1272

1273 1274
	up_read(&cpufreq_rwsem);

1275 1276 1277 1278
	/* Callback for handling stuff after policy is ready */
	if (cpufreq_driver->ready)
		cpufreq_driver->ready(policy);

1279
	pr_debug("initialization complete\n");
1280

L
Linus Torvalds 已提交
1281 1282 1283
	return 0;

err_out_unregister:
1284
err_get_freq:
1285
	write_lock_irqsave(&cpufreq_driver_lock, flags);
1286
	for_each_cpu(j, policy->cpus)
1287
		per_cpu(cpufreq_cpu_data, j) = NULL;
1288
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
1289

1290 1291 1292 1293 1294
	if (!recover_policy) {
		kobject_put(&policy->kobj);
		wait_for_completion(&policy->kobj_unregister);
	}
err_init_policy_kobj:
1295 1296
	up_write(&policy->rwsem);

1297 1298
	if (cpufreq_driver->exit)
		cpufreq_driver->exit(policy);
V
Viresh Kumar 已提交
1299
err_set_policy_cpu:
1300
	if (recover_policy) {
1301 1302
		/* Do not leave stale fallback data behind. */
		per_cpu(cpufreq_cpu_data_fallback, cpu) = NULL;
1303
		cpufreq_policy_put_kobj(policy);
1304
	}
1305
	cpufreq_policy_free(policy);
1306

L
Linus Torvalds 已提交
1307
nomem_out:
1308 1309
	up_read(&cpufreq_rwsem);

L
Linus Torvalds 已提交
1310 1311 1312
	return ret;
}

1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323
/**
 * 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)
{
1324
	return __cpufreq_add_dev(dev, sif);
1325 1326
}

1327
static int __cpufreq_remove_dev_prepare(struct device *dev,
1328
					struct subsys_interface *sif)
L
Linus Torvalds 已提交
1329
{
1330
	unsigned int cpu = dev->id, cpus;
1331
	int ret;
L
Linus Torvalds 已提交
1332
	unsigned long flags;
1333
	struct cpufreq_policy *policy;
L
Linus Torvalds 已提交
1334

1335
	pr_debug("%s: unregistering CPU %u\n", __func__, cpu);
L
Linus Torvalds 已提交
1336

1337
	write_lock_irqsave(&cpufreq_driver_lock, flags);
V
Viresh Kumar 已提交
1338

1339
	policy = per_cpu(cpufreq_cpu_data, cpu);
V
Viresh Kumar 已提交
1340

1341
	/* Save the policy somewhere when doing a light-weight tear-down */
1342
	if (cpufreq_suspended)
1343
		per_cpu(cpufreq_cpu_data_fallback, cpu) = policy;
1344

1345
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
1346

1347
	if (!policy) {
1348
		pr_debug("%s: No cpu_data found\n", __func__);
L
Linus Torvalds 已提交
1349 1350 1351
		return -EINVAL;
	}

1352
	if (has_target()) {
1353 1354 1355 1356 1357
		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
1358

1359
		strncpy(per_cpu(cpufreq_cpu_governor, cpu),
1360
			policy->governor->name, CPUFREQ_NAME_LEN);
1361
	}
L
Linus Torvalds 已提交
1362

1363
	down_read(&policy->rwsem);
1364
	cpus = cpumask_weight(policy->cpus);
1365
	up_read(&policy->rwsem);
1366

1367
	if (cpu != policy->cpu) {
1368
		sysfs_remove_link(&dev->kobj, "cpufreq");
1369
	} else if (cpus > 1) {
1370 1371 1372
		/* Nominate new CPU */
		int new_cpu = cpumask_any_but(policy->cpus, cpu);
		struct device *cpu_dev = get_cpu_device(new_cpu);
1373

1374 1375 1376 1377 1378 1379 1380 1381
		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
1382
		}
1383 1384 1385 1386

		if (!cpufreq_suspended)
			pr_debug("%s: policy Kobject moved to cpu: %d from: %d\n",
				 __func__, new_cpu, cpu);
1387
	} else if (cpufreq_driver->stop_cpu) {
1388
		cpufreq_driver->stop_cpu(policy);
L
Linus Torvalds 已提交
1389 1390
	}

1391 1392 1393 1394
	return 0;
}

static int __cpufreq_remove_dev_finish(struct device *dev,
1395
				       struct subsys_interface *sif)
1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410
{
	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;
	}

1411
	down_write(&policy->rwsem);
1412
	cpus = cpumask_weight(policy->cpus);
1413 1414 1415

	if (cpus > 1)
		cpumask_clear_cpu(cpu, policy->cpus);
1416
	up_write(&policy->rwsem);
1417

1418 1419
	/* If cpu is last user of policy, free policy */
	if (cpus == 1) {
1420
		if (has_target()) {
1421 1422 1423 1424
			ret = __cpufreq_governor(policy,
					CPUFREQ_GOV_POLICY_EXIT);
			if (ret) {
				pr_err("%s: Failed to exit governor\n",
1425
				       __func__);
1426 1427
				return ret;
			}
1428
		}
1429

1430
		if (!cpufreq_suspended)
1431
			cpufreq_policy_put_kobj(policy);
1432

1433 1434 1435 1436
		/*
		 * 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.
1437
		 */
1438
		if (cpufreq_driver->exit)
1439
			cpufreq_driver->exit(policy);
1440

1441 1442 1443 1444 1445
		/* 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);

1446
		if (!cpufreq_suspended)
1447
			cpufreq_policy_free(policy);
1448 1449 1450 1451 1452 1453 1454 1455
	} 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;
1456
		}
1457
	}
L
Linus Torvalds 已提交
1458

1459
	per_cpu(cpufreq_cpu_data, cpu) = NULL;
L
Linus Torvalds 已提交
1460 1461 1462
	return 0;
}

1463
/**
1464
 * cpufreq_remove_dev - remove a CPU device
1465 1466 1467
 *
 * Removes the cpufreq interface for a CPU device.
 */
1468
static int cpufreq_remove_dev(struct device *dev, struct subsys_interface *sif)
1469
{
1470
	unsigned int cpu = dev->id;
1471
	int ret;
1472 1473 1474 1475

	if (cpu_is_offline(cpu))
		return 0;

1476
	ret = __cpufreq_remove_dev_prepare(dev, sif);
1477 1478

	if (!ret)
1479
		ret = __cpufreq_remove_dev_finish(dev, sif);
1480 1481

	return ret;
1482 1483
}

1484
static void handle_update(struct work_struct *work)
L
Linus Torvalds 已提交
1485
{
1486 1487 1488
	struct cpufreq_policy *policy =
		container_of(work, struct cpufreq_policy, update);
	unsigned int cpu = policy->cpu;
1489
	pr_debug("handle_update for cpu %u called\n", cpu);
L
Linus Torvalds 已提交
1490 1491 1492 1493
	cpufreq_update_policy(cpu);
}

/**
1494 1495
 *	cpufreq_out_of_sync - If actual and saved CPU frequency differs, we're
 *	in deep trouble.
1496
 *	@policy: policy managing CPUs
L
Linus Torvalds 已提交
1497 1498
 *	@new_freq: CPU frequency the CPU actually runs at
 *
1499 1500
 *	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
1501
 */
1502
static void cpufreq_out_of_sync(struct cpufreq_policy *policy,
1503
				unsigned int new_freq)
L
Linus Torvalds 已提交
1504 1505
{
	struct cpufreq_freqs freqs;
1506

1507
	pr_debug("Warning: CPU frequency out of sync: cpufreq and timing core thinks of %u, is %u kHz\n",
1508
		 policy->cur, new_freq);
L
Linus Torvalds 已提交
1509

1510
	freqs.old = policy->cur;
L
Linus Torvalds 已提交
1511
	freqs.new = new_freq;
1512

1513 1514
	cpufreq_freq_transition_begin(policy, &freqs);
	cpufreq_freq_transition_end(policy, &freqs, 0);
L
Linus Torvalds 已提交
1515 1516
}

1517
/**
D
Dhaval Giani 已提交
1518
 * cpufreq_quick_get - get the CPU frequency (in kHz) from policy->cur
1519 1520 1521 1522 1523 1524 1525
 * @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)
{
1526
	struct cpufreq_policy *policy;
1527
	unsigned int ret_freq = 0;
1528

1529 1530
	if (cpufreq_driver && cpufreq_driver->setpolicy && cpufreq_driver->get)
		return cpufreq_driver->get(cpu);
1531 1532

	policy = cpufreq_cpu_get(cpu);
1533
	if (policy) {
1534
		ret_freq = policy->cur;
1535 1536 1537
		cpufreq_cpu_put(policy);
	}

D
Dave Jones 已提交
1538
	return ret_freq;
1539 1540 1541
}
EXPORT_SYMBOL(cpufreq_quick_get);

1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561
/**
 * 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);

1562
static unsigned int __cpufreq_get(struct cpufreq_policy *policy)
L
Linus Torvalds 已提交
1563
{
1564
	unsigned int ret_freq = 0;
1565

1566
	if (!cpufreq_driver->get)
D
Dave Jones 已提交
1567
		return ret_freq;
L
Linus Torvalds 已提交
1568

1569
	ret_freq = cpufreq_driver->get(policy->cpu);
L
Linus Torvalds 已提交
1570

1571
	if (ret_freq && policy->cur &&
1572
		!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
1573 1574 1575
		/* verify no discrepancy between actual and
					saved value exists */
		if (unlikely(ret_freq != policy->cur)) {
1576
			cpufreq_out_of_sync(policy, ret_freq);
L
Linus Torvalds 已提交
1577 1578 1579 1580
			schedule_work(&policy->update);
		}
	}

D
Dave Jones 已提交
1581
	return ret_freq;
1582
}
L
Linus Torvalds 已提交
1583

1584 1585 1586 1587 1588 1589 1590 1591
/**
 * 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)
{
1592
	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
1593 1594
	unsigned int ret_freq = 0;

1595 1596
	if (policy) {
		down_read(&policy->rwsem);
1597
		ret_freq = __cpufreq_get(policy);
1598
		up_read(&policy->rwsem);
1599

1600 1601
		cpufreq_cpu_put(policy);
	}
1602

D
Dave Jones 已提交
1603
	return ret_freq;
L
Linus Torvalds 已提交
1604 1605 1606
}
EXPORT_SYMBOL(cpufreq_get);

1607 1608 1609 1610 1611
static struct subsys_interface cpufreq_interface = {
	.name		= "cpufreq",
	.subsys		= &cpu_subsys,
	.add_dev	= cpufreq_add_dev,
	.remove_dev	= cpufreq_remove_dev,
1612 1613
};

1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639
/*
 * 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);

1640
/**
1641
 * cpufreq_suspend() - Suspend CPUFreq governors
1642
 *
1643 1644 1645 1646
 * 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.
1647
 */
1648
void cpufreq_suspend(void)
1649
{
1650
	struct cpufreq_policy *policy;
1651

1652 1653
	if (!cpufreq_driver)
		return;
1654

1655
	if (!has_target())
1656
		goto suspend;
1657

1658 1659 1660 1661 1662 1663 1664 1665 1666 1667
	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);
1668
	}
1669 1670 1671

suspend:
	cpufreq_suspended = true;
1672 1673
}

L
Linus Torvalds 已提交
1674
/**
1675
 * cpufreq_resume() - Resume CPUFreq governors
L
Linus Torvalds 已提交
1676
 *
1677 1678
 * Called during system wide Suspend/Hibernate cycle for resuming governors that
 * are suspended with cpufreq_suspend().
L
Linus Torvalds 已提交
1679
 */
1680
void cpufreq_resume(void)
L
Linus Torvalds 已提交
1681
{
1682
	struct cpufreq_policy *policy;
L
Linus Torvalds 已提交
1683

1684 1685
	if (!cpufreq_driver)
		return;
L
Linus Torvalds 已提交
1686

1687 1688
	cpufreq_suspended = false;

1689
	if (!has_target())
1690
		return;
L
Linus Torvalds 已提交
1691

1692
	pr_debug("%s: Resuming Governors\n", __func__);
L
Linus Torvalds 已提交
1693

1694
	list_for_each_entry(policy, &cpufreq_policy_list, policy_list) {
1695 1696 1697 1698
		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)
1699 1700 1701
		    || __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
1702

1703 1704 1705 1706 1707 1708 1709 1710 1711
		/*
		 * 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
1712

1713 1714 1715 1716 1717 1718 1719 1720
/**
 *	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)
{
1721 1722 1723 1724
	if (cpufreq_driver)
		return cpufreq_driver->name;

	return NULL;
1725 1726
}
EXPORT_SYMBOL_GPL(cpufreq_get_current_driver);
L
Linus Torvalds 已提交
1727

1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742
/**
 *	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
1743 1744 1745 1746 1747 1748 1749 1750 1751
/*********************************************************************
 *                     NOTIFIER LISTS INTERFACE                      *
 *********************************************************************/

/**
 *	cpufreq_register_notifier - register a driver with cpufreq
 *	@nb: notifier function to register
 *      @list: CPUFREQ_TRANSITION_NOTIFIER or CPUFREQ_POLICY_NOTIFIER
 *
1752
 *	Add a driver to one of two lists: either a list of drivers that
L
Linus Torvalds 已提交
1753 1754 1755 1756 1757
 *      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
1758
 *	blocking_notifier_chain_register.
L
Linus Torvalds 已提交
1759 1760 1761 1762 1763
 */
int cpufreq_register_notifier(struct notifier_block *nb, unsigned int list)
{
	int ret;

1764 1765 1766
	if (cpufreq_disabled())
		return -EINVAL;

1767 1768
	WARN_ON(!init_cpufreq_transition_notifier_list_called);

L
Linus Torvalds 已提交
1769 1770
	switch (list) {
	case CPUFREQ_TRANSITION_NOTIFIER:
1771
		ret = srcu_notifier_chain_register(
1772
				&cpufreq_transition_notifier_list, nb);
L
Linus Torvalds 已提交
1773 1774
		break;
	case CPUFREQ_POLICY_NOTIFIER:
1775 1776
		ret = blocking_notifier_chain_register(
				&cpufreq_policy_notifier_list, nb);
L
Linus Torvalds 已提交
1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788
		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
1789
 *	@list: CPUFREQ_TRANSITION_NOTIFIER or CPUFREQ_POLICY_NOTIFIER
L
Linus Torvalds 已提交
1790 1791 1792 1793
 *
 *	Remove a driver from the CPU frequency notifier list.
 *
 *	This function may sleep, and has the same return conditions as
1794
 *	blocking_notifier_chain_unregister.
L
Linus Torvalds 已提交
1795 1796 1797 1798 1799
 */
int cpufreq_unregister_notifier(struct notifier_block *nb, unsigned int list)
{
	int ret;

1800 1801 1802
	if (cpufreq_disabled())
		return -EINVAL;

L
Linus Torvalds 已提交
1803 1804
	switch (list) {
	case CPUFREQ_TRANSITION_NOTIFIER:
1805
		ret = srcu_notifier_chain_unregister(
1806
				&cpufreq_transition_notifier_list, nb);
L
Linus Torvalds 已提交
1807 1808
		break;
	case CPUFREQ_POLICY_NOTIFIER:
1809 1810
		ret = blocking_notifier_chain_unregister(
				&cpufreq_policy_notifier_list, nb);
L
Linus Torvalds 已提交
1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824
		break;
	default:
		ret = -EINVAL;
	}

	return ret;
}
EXPORT_SYMBOL(cpufreq_unregister_notifier);


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

1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850
/* 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;
}

1851 1852 1853
static int __target_index(struct cpufreq_policy *policy,
			  struct cpufreq_frequency_table *freq_table, int index)
{
1854 1855
	struct cpufreq_freqs freqs = {.old = policy->cur, .flags = 0};
	unsigned int intermediate_freq = 0;
1856 1857 1858 1859 1860
	int retval = -EINVAL;
	bool notify;

	notify = !(cpufreq_driver->flags & CPUFREQ_ASYNC_NOTIFICATION);
	if (notify) {
1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871
		/* 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;
		}
1872

1873
		freqs.new = freq_table[index].frequency;
1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884
		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);

1885
	if (notify) {
1886 1887
		cpufreq_freq_transition_end(policy, &freqs, retval);

1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901
		/*
		 * 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);
		}
	}

1902 1903 1904
	return retval;
}

L
Linus Torvalds 已提交
1905 1906 1907 1908
int __cpufreq_driver_target(struct cpufreq_policy *policy,
			    unsigned int target_freq,
			    unsigned int relation)
{
1909
	unsigned int old_target_freq = target_freq;
1910
	int retval = -EINVAL;
1911

1912 1913 1914
	if (cpufreq_disabled())
		return -ENODEV;

1915 1916 1917 1918 1919 1920 1921
	/* 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",
1922
		 policy->cpu, target_freq, relation, old_target_freq);
1923

1924 1925 1926 1927 1928 1929
	/*
	 * 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.
	 */
1930 1931 1932
	if (target_freq == policy->cur)
		return 0;

1933 1934 1935
	/* Save last value to restore later on errors */
	policy->restore_freq = policy->cur;

1936 1937
	if (cpufreq_driver->target)
		retval = cpufreq_driver->target(policy, target_freq, relation);
1938 1939 1940
	else if (cpufreq_driver->target_index) {
		struct cpufreq_frequency_table *freq_table;
		int index;
1941

1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954
		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;
		}

1955
		if (freq_table[index].frequency == policy->cur) {
1956
			retval = 0;
1957 1958 1959
			goto out;
		}

1960
		retval = __target_index(policy, freq_table, index);
1961 1962 1963
	}

out:
L
Linus Torvalds 已提交
1964 1965 1966 1967 1968 1969 1970 1971
	return retval;
}
EXPORT_SYMBOL_GPL(__cpufreq_driver_target);

int cpufreq_driver_target(struct cpufreq_policy *policy,
			  unsigned int target_freq,
			  unsigned int relation)
{
1972
	int ret = -EINVAL;
L
Linus Torvalds 已提交
1973

1974
	down_write(&policy->rwsem);
L
Linus Torvalds 已提交
1975 1976 1977

	ret = __cpufreq_driver_target(policy, target_freq, relation);

1978
	up_write(&policy->rwsem);
L
Linus Torvalds 已提交
1979 1980 1981 1982 1983

	return ret;
}
EXPORT_SYMBOL_GPL(cpufreq_driver_target);

1984 1985
static int __cpufreq_governor(struct cpufreq_policy *policy,
					unsigned int event)
L
Linus Torvalds 已提交
1986
{
1987
	int ret;
1988 1989 1990 1991 1992 1993 1994 1995 1996 1997

	/* 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
1998

1999 2000 2001
	/* Don't start any governor operations if we are entering suspend */
	if (cpufreq_suspended)
		return 0;
2002 2003 2004 2005 2006 2007
	/*
	 * Governor might not be initiated here if ACPI _PPC changed
	 * notification happened, so check it.
	 */
	if (!policy->governor)
		return -EINVAL;
2008

2009 2010 2011
	if (policy->governor->max_transition_latency &&
	    policy->cpuinfo.transition_latency >
	    policy->governor->max_transition_latency) {
2012 2013 2014
		if (!gov)
			return -EINVAL;
		else {
2015 2016
			pr_warn("%s governor failed, too long transition latency of HW, fallback to %s governor\n",
				policy->governor->name, gov->name);
2017 2018
			policy->governor = gov;
		}
2019
	}
L
Linus Torvalds 已提交
2020

2021 2022 2023
	if (event == CPUFREQ_GOV_POLICY_INIT)
		if (!try_module_get(policy->governor->owner))
			return -EINVAL;
L
Linus Torvalds 已提交
2024

2025
	pr_debug("__cpufreq_governor for CPU %u, event %u\n",
2026
		 policy->cpu, event);
2027 2028

	mutex_lock(&cpufreq_governor_lock);
2029
	if ((policy->governor_enabled && event == CPUFREQ_GOV_START)
2030 2031
	    || (!policy->governor_enabled
	    && (event == CPUFREQ_GOV_LIMITS || event == CPUFREQ_GOV_STOP))) {
2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042
		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
2043 2044
	ret = policy->governor->governor(policy, event);

2045 2046 2047 2048 2049
	if (!ret) {
		if (event == CPUFREQ_GOV_POLICY_INIT)
			policy->governor->initialized++;
		else if (event == CPUFREQ_GOV_POLICY_EXIT)
			policy->governor->initialized--;
2050 2051 2052 2053 2054 2055 2056 2057
	} 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);
2058
	}
2059

2060 2061
	if (((event == CPUFREQ_GOV_POLICY_INIT) && ret) ||
			((event == CPUFREQ_GOV_POLICY_EXIT) && !ret))
L
Linus Torvalds 已提交
2062 2063 2064 2065 2066 2067 2068
		module_put(policy->governor->owner);

	return ret;
}

int cpufreq_register_governor(struct cpufreq_governor *governor)
{
2069
	int err;
L
Linus Torvalds 已提交
2070 2071 2072 2073

	if (!governor)
		return -EINVAL;

2074 2075 2076
	if (cpufreq_disabled())
		return -ENODEV;

2077
	mutex_lock(&cpufreq_governor_mutex);
2078

2079
	governor->initialized = 0;
2080
	err = -EBUSY;
2081
	if (!find_governor(governor->name)) {
2082 2083
		err = 0;
		list_add(&governor->governor_list, &cpufreq_governor_list);
L
Linus Torvalds 已提交
2084 2085
	}

2086
	mutex_unlock(&cpufreq_governor_mutex);
2087
	return err;
L
Linus Torvalds 已提交
2088 2089 2090 2091 2092
}
EXPORT_SYMBOL_GPL(cpufreq_register_governor);

void cpufreq_unregister_governor(struct cpufreq_governor *governor)
{
2093 2094
	int cpu;

L
Linus Torvalds 已提交
2095 2096 2097
	if (!governor)
		return;

2098 2099 2100
	if (cpufreq_disabled())
		return;

2101 2102 2103 2104 2105 2106 2107
	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");
	}

2108
	mutex_lock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
2109
	list_del(&governor->governor_list);
2110
	mutex_unlock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121
	return;
}
EXPORT_SYMBOL_GPL(cpufreq_unregister_governor);


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

/**
 * cpufreq_get_policy - get the current cpufreq_policy
2122 2123
 * @policy: struct cpufreq_policy into which the current cpufreq_policy
 *	is written
L
Linus Torvalds 已提交
2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136
 *
 * 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;

2137
	memcpy(policy, cpu_policy, sizeof(*policy));
L
Linus Torvalds 已提交
2138 2139 2140 2141 2142 2143

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

2144
/*
2145 2146
 * policy : current policy.
 * new_policy: policy to be set.
2147
 */
2148
static int cpufreq_set_policy(struct cpufreq_policy *policy,
2149
				struct cpufreq_policy *new_policy)
L
Linus Torvalds 已提交
2150
{
2151 2152
	struct cpufreq_governor *old_gov;
	int ret;
L
Linus Torvalds 已提交
2153

2154 2155
	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
2156

2157
	memcpy(&new_policy->cpuinfo, &policy->cpuinfo, sizeof(policy->cpuinfo));
L
Linus Torvalds 已提交
2158

2159 2160
	if (new_policy->min > policy->max || new_policy->max < policy->min)
		return -EINVAL;
2161

L
Linus Torvalds 已提交
2162
	/* verify the cpu speed can be set within this limit */
2163
	ret = cpufreq_driver->verify(new_policy);
L
Linus Torvalds 已提交
2164
	if (ret)
2165
		return ret;
L
Linus Torvalds 已提交
2166 2167

	/* adjust if necessary - all reasons */
2168
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2169
			CPUFREQ_ADJUST, new_policy);
L
Linus Torvalds 已提交
2170 2171

	/* adjust if necessary - hardware incompatibility*/
2172
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2173
			CPUFREQ_INCOMPATIBLE, new_policy);
L
Linus Torvalds 已提交
2174

2175 2176 2177 2178
	/*
	 * verify the cpu speed can be set within this limit, which might be
	 * different to the first one
	 */
2179
	ret = cpufreq_driver->verify(new_policy);
2180
	if (ret)
2181
		return ret;
L
Linus Torvalds 已提交
2182 2183

	/* notification of the new policy */
2184
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2185
			CPUFREQ_NOTIFY, new_policy);
L
Linus Torvalds 已提交
2186

2187 2188
	policy->min = new_policy->min;
	policy->max = new_policy->max;
L
Linus Torvalds 已提交
2189

2190
	pr_debug("new min and max freqs are %u - %u kHz\n",
2191
		 policy->min, policy->max);
L
Linus Torvalds 已提交
2192

2193
	if (cpufreq_driver->setpolicy) {
2194
		policy->policy = new_policy->policy;
2195
		pr_debug("setting range\n");
2196 2197
		return cpufreq_driver->setpolicy(new_policy);
	}
L
Linus Torvalds 已提交
2198

2199 2200
	if (new_policy->governor == policy->governor)
		goto out;
2201

2202 2203 2204 2205 2206 2207 2208 2209
	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);
2210
		__cpufreq_governor(policy, CPUFREQ_GOV_POLICY_EXIT);
2211
		down_write(&policy->rwsem);
L
Linus Torvalds 已提交
2212 2213
	}

2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237
	/* 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
2238 2239 2240 2241 2242 2243
}

/**
 *	cpufreq_update_policy - re-evaluate an existing cpufreq policy
 *	@cpu: CPU which shall be re-evaluated
 *
L
Lucas De Marchi 已提交
2244
 *	Useful for policy notifiers which have different necessities
L
Linus Torvalds 已提交
2245 2246 2247 2248
 *	at different times.
 */
int cpufreq_update_policy(unsigned int cpu)
{
2249 2250
	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
	struct cpufreq_policy new_policy;
2251
	int ret;
L
Linus Torvalds 已提交
2252

2253 2254
	if (!policy)
		return -ENODEV;
L
Linus Torvalds 已提交
2255

2256
	down_write(&policy->rwsem);
L
Linus Torvalds 已提交
2257

2258
	pr_debug("updating policy for CPU %u\n", cpu);
2259
	memcpy(&new_policy, policy, sizeof(*policy));
2260 2261 2262 2263
	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
2264

2265 2266 2267 2268
	/*
	 * BIOS might change freq behind our back
	 * -> ask driver for current freq and notify governors about a change
	 */
2269
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
2270
		new_policy.cur = cpufreq_driver->get(cpu);
2271 2272
		if (WARN_ON(!new_policy.cur)) {
			ret = -EIO;
2273
			goto unlock;
2274 2275
		}

2276
		if (!policy->cur) {
2277
			pr_debug("Driver did not initialize current freq\n");
2278
			policy->cur = new_policy.cur;
2279
		} else {
2280
			if (policy->cur != new_policy.cur && has_target())
2281
				cpufreq_out_of_sync(policy, new_policy.cur);
2282
		}
2283 2284
	}

2285
	ret = cpufreq_set_policy(policy, &new_policy);
L
Linus Torvalds 已提交
2286

2287
unlock:
2288
	up_write(&policy->rwsem);
2289

2290
	cpufreq_cpu_put(policy);
L
Linus Torvalds 已提交
2291 2292 2293 2294
	return ret;
}
EXPORT_SYMBOL(cpufreq_update_policy);

2295
static int cpufreq_cpu_callback(struct notifier_block *nfb,
2296 2297 2298
					unsigned long action, void *hcpu)
{
	unsigned int cpu = (unsigned long)hcpu;
2299
	struct device *dev;
2300

2301 2302
	dev = get_cpu_device(cpu);
	if (dev) {
2303
		switch (action & ~CPU_TASKS_FROZEN) {
2304
		case CPU_ONLINE:
2305
			__cpufreq_add_dev(dev, NULL);
2306
			break;
2307

2308
		case CPU_DOWN_PREPARE:
2309
			__cpufreq_remove_dev_prepare(dev, NULL);
2310 2311 2312
			break;

		case CPU_POST_DEAD:
2313
			__cpufreq_remove_dev_finish(dev, NULL);
2314
			break;
2315

2316
		case CPU_DOWN_FAILED:
2317
			__cpufreq_add_dev(dev, NULL);
2318 2319 2320 2321 2322 2323
			break;
		}
	}
	return NOTIFY_OK;
}

2324
static struct notifier_block __refdata cpufreq_cpu_notifier = {
2325
	.notifier_call = cpufreq_cpu_callback,
2326
};
L
Linus Torvalds 已提交
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 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372
/*********************************************************************
 *               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);

2373 2374
		pr_err("%s: Cannot %s BOOST\n",
		       __func__, state ? "enable" : "disable");
2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394
	}

	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
2395 2396 2397 2398 2399 2400 2401 2402 2403
/*********************************************************************
 *               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.
 *
2404
 * Registers a CPU Frequency driver to this core code. This code
L
Linus Torvalds 已提交
2405
 * returns zero on success, -EBUSY when another driver got here first
2406
 * (and isn't unregistered in the meantime).
L
Linus Torvalds 已提交
2407 2408
 *
 */
2409
int cpufreq_register_driver(struct cpufreq_driver *driver_data)
L
Linus Torvalds 已提交
2410 2411 2412 2413
{
	unsigned long flags;
	int ret;

2414 2415 2416
	if (cpufreq_disabled())
		return -ENODEV;

L
Linus Torvalds 已提交
2417
	if (!driver_data || !driver_data->verify || !driver_data->init ||
2418
	    !(driver_data->setpolicy || driver_data->target_index ||
2419 2420
		    driver_data->target) ||
	     (driver_data->setpolicy && (driver_data->target_index ||
2421 2422
		    driver_data->target)) ||
	     (!!driver_data->get_intermediate != !!driver_data->target_intermediate))
L
Linus Torvalds 已提交
2423 2424
		return -EINVAL;

2425
	pr_debug("trying to register driver %s\n", driver_data->name);
L
Linus Torvalds 已提交
2426

2427
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2428
	if (cpufreq_driver) {
2429
		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2430
		return -EEXIST;
L
Linus Torvalds 已提交
2431
	}
2432
	cpufreq_driver = driver_data;
2433
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
2434

2435 2436 2437
	if (driver_data->setpolicy)
		driver_data->flags |= CPUFREQ_CONST_LOOPS;

2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448
	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",
2449
			       __func__);
2450 2451 2452 2453
			goto err_null_driver;
		}
	}

2454
	ret = subsys_interface_register(&cpufreq_interface);
2455
	if (ret)
2456
		goto err_boost_unreg;
L
Linus Torvalds 已提交
2457

2458 2459
	if (!(cpufreq_driver->flags & CPUFREQ_STICKY) &&
	    list_empty(&cpufreq_policy_list)) {
L
Linus Torvalds 已提交
2460
		/* if all ->init() calls failed, unregister */
2461 2462 2463
		pr_debug("%s: No CPU initialized for driver %s\n", __func__,
			 driver_data->name);
		goto err_if_unreg;
L
Linus Torvalds 已提交
2464 2465
	}

2466
	register_hotcpu_notifier(&cpufreq_cpu_notifier);
2467
	pr_debug("driver %s up and running\n", driver_data->name);
L
Linus Torvalds 已提交
2468

2469
	return 0;
2470 2471
err_if_unreg:
	subsys_interface_unregister(&cpufreq_interface);
2472 2473 2474
err_boost_unreg:
	if (cpufreq_boost_supported())
		cpufreq_sysfs_remove_file(&boost.attr);
2475
err_null_driver:
2476
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2477
	cpufreq_driver = NULL;
2478
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
D
Dave Jones 已提交
2479
	return ret;
L
Linus Torvalds 已提交
2480 2481 2482 2483 2484 2485
}
EXPORT_SYMBOL_GPL(cpufreq_register_driver);

/**
 * cpufreq_unregister_driver - unregister the current CPUFreq driver
 *
2486
 * Unregister the current CPUFreq driver. Only call this if you have
L
Linus Torvalds 已提交
2487 2488 2489 2490
 * 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.
 */
2491
int cpufreq_unregister_driver(struct cpufreq_driver *driver)
L
Linus Torvalds 已提交
2492 2493 2494
{
	unsigned long flags;

2495
	if (!cpufreq_driver || (driver != cpufreq_driver))
L
Linus Torvalds 已提交
2496 2497
		return -EINVAL;

2498
	pr_debug("unregistering driver %s\n", driver->name);
L
Linus Torvalds 已提交
2499

2500
	subsys_interface_unregister(&cpufreq_interface);
2501 2502 2503
	if (cpufreq_boost_supported())
		cpufreq_sysfs_remove_file(&boost.attr);

2504
	unregister_hotcpu_notifier(&cpufreq_cpu_notifier);
L
Linus Torvalds 已提交
2505

2506
	down_write(&cpufreq_rwsem);
2507
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2508

2509
	cpufreq_driver = NULL;
2510

2511
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2512
	up_write(&cpufreq_rwsem);
L
Linus Torvalds 已提交
2513 2514 2515 2516

	return 0;
}
EXPORT_SYMBOL_GPL(cpufreq_unregister_driver);
2517

2518 2519 2520 2521 2522 2523 2524 2525
/*
 * Stop cpufreq at shutdown to make sure it isn't holding any locks
 * or mutexes when secondary CPUs are halted.
 */
static struct syscore_ops cpufreq_syscore_ops = {
	.shutdown = cpufreq_suspend,
};

2526 2527
static int __init cpufreq_core_init(void)
{
2528 2529 2530
	if (cpufreq_disabled())
		return -ENODEV;

2531
	cpufreq_global_kobject = kobject_create();
2532 2533
	BUG_ON(!cpufreq_global_kobject);

2534 2535
	register_syscore_ops(&cpufreq_syscore_ops);

2536 2537 2538
	return 0;
}
core_initcall(cpufreq_core_init);