cpufreq.c 64.6 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
	if (cpu >= nr_cpu_ids)
207 208 209 210
		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
	kobject_put(&policy->kobj);
	up_read(&cpufreq_rwsem);
L
Linus Torvalds 已提交
235 236 237 238 239 240 241 242 243 244 245 246
}
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
247
 * systems as each CPU might be scaled differently. So, use the arch
L
Linus Torvalds 已提交
248 249
 * per-CPU loops_per_jiffy value wherever possible.
 */
250
static void adjust_jiffies(unsigned long val, struct cpufreq_freqs *ci)
L
Linus Torvalds 已提交
251
{
252 253 254 255
#ifndef CONFIG_SMP
	static unsigned long l_p_j_ref;
	static unsigned int l_p_j_ref_freq;

L
Linus Torvalds 已提交
256 257 258 259 260 261
	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;
262 263
		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
264
	}
265
	if (val == CPUFREQ_POSTCHANGE && ci->old != ci->new) {
266 267
		loops_per_jiffy = cpufreq_scale(l_p_j_ref, l_p_j_ref_freq,
								ci->new);
268 269
		pr_debug("scaling loops_per_jiffy to %lu for frequency %u kHz\n",
			 loops_per_jiffy, ci->new);
L
Linus Torvalds 已提交
270 271
	}
#endif
272
}
L
Linus Torvalds 已提交
273

274
static void __cpufreq_notify_transition(struct cpufreq_policy *policy,
275
		struct cpufreq_freqs *freqs, unsigned int state)
L
Linus Torvalds 已提交
276 277 278
{
	BUG_ON(irqs_disabled());

279 280 281
	if (cpufreq_disabled())
		return;

282
	freqs->flags = cpufreq_driver->flags;
283
	pr_debug("notification %u of frequency transition to %u kHz\n",
284
		 state, freqs->new);
L
Linus Torvalds 已提交
285 286

	switch (state) {
287

L
Linus Torvalds 已提交
288
	case CPUFREQ_PRECHANGE:
289
		/* detect if the driver reported a value as "old frequency"
290 291
		 * which is not equal to what the cpufreq core thinks is
		 * "old frequency".
L
Linus Torvalds 已提交
292
		 */
293
		if (!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
294 295
			if ((policy) && (policy->cpu == freqs->cpu) &&
			    (policy->cur) && (policy->cur != freqs->old)) {
296 297
				pr_debug("Warning: CPU frequency is %u, cpufreq assumed %u kHz\n",
					 freqs->old, policy->cur);
298
				freqs->old = policy->cur;
L
Linus Torvalds 已提交
299 300
			}
		}
301
		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
302
				CPUFREQ_PRECHANGE, freqs);
L
Linus Torvalds 已提交
303 304
		adjust_jiffies(CPUFREQ_PRECHANGE, freqs);
		break;
305

L
Linus Torvalds 已提交
306 307
	case CPUFREQ_POSTCHANGE:
		adjust_jiffies(CPUFREQ_POSTCHANGE, freqs);
308 309
		pr_debug("FREQ: %lu - CPU: %lu\n",
			 (unsigned long)freqs->new, (unsigned long)freqs->cpu);
310
		trace_cpu_frequency(freqs->new, freqs->cpu);
311
		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
312
				CPUFREQ_POSTCHANGE, freqs);
313 314
		if (likely(policy) && likely(policy->cpu == freqs->cpu))
			policy->cur = freqs->new;
L
Linus Torvalds 已提交
315 316 317
		break;
	}
}
318

319 320 321 322 323 324 325 326
/**
 * 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.
 */
327
static void cpufreq_notify_transition(struct cpufreq_policy *policy,
328 329 330 331 332
		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
333

334
/* Do post notifications when there are chances that transition has failed */
335
static void cpufreq_notify_post_transition(struct cpufreq_policy *policy,
336 337 338 339 340 341 342 343 344 345 346
		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);
}

347 348 349
void cpufreq_freq_transition_begin(struct cpufreq_policy *policy,
		struct cpufreq_freqs *freqs)
{
350 351 352 353 354 355 356 357 358 359 360 361

	/*
	 * 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);

362 363 364 365 366 367 368 369 370 371 372
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;
373
	policy->transition_task = current;
374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389

	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;
390
	policy->transition_task = NULL;
391 392 393 394 395

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

L
Linus Torvalds 已提交
396 397 398 399

/*********************************************************************
 *                          SYSFS INTERFACE                          *
 *********************************************************************/
400
static ssize_t show_boost(struct kobject *kobj,
401 402 403 404 405 406 407 408 409 410 411 412 413 414 415
				 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)) {
416 417
		pr_err("%s: Cannot %s BOOST!\n",
		       __func__, enable ? "enable" : "disable");
418 419 420
		return -EINVAL;
	}

421 422
	pr_debug("%s: cpufreq BOOST %s\n",
		 __func__, enable ? "enabled" : "disabled");
423 424 425 426

	return count;
}
define_one_global_rw(boost);
L
Linus Torvalds 已提交
427

428
static struct cpufreq_governor *find_governor(const char *str_governor)
429 430 431 432
{
	struct cpufreq_governor *t;

	list_for_each_entry(t, &cpufreq_governor_list, governor_list)
433
		if (!strncasecmp(str_governor, t->name, CPUFREQ_NAME_LEN))
434 435 436 437 438
			return t;

	return NULL;
}

L
Linus Torvalds 已提交
439 440 441
/**
 * cpufreq_parse_governor - parse a governor string
 */
D
Dave Jones 已提交
442
static int cpufreq_parse_governor(char *str_governor, unsigned int *policy,
L
Linus Torvalds 已提交
443 444
				struct cpufreq_governor **governor)
{
445
	int err = -EINVAL;
446 447

	if (!cpufreq_driver)
448 449
		goto out;

450
	if (cpufreq_driver->setpolicy) {
451
		if (!strncasecmp(str_governor, "performance", CPUFREQ_NAME_LEN)) {
L
Linus Torvalds 已提交
452
			*policy = CPUFREQ_POLICY_PERFORMANCE;
453
			err = 0;
454
		} else if (!strncasecmp(str_governor, "powersave",
455
						CPUFREQ_NAME_LEN)) {
L
Linus Torvalds 已提交
456
			*policy = CPUFREQ_POLICY_POWERSAVE;
457
			err = 0;
L
Linus Torvalds 已提交
458
		}
459
	} else {
L
Linus Torvalds 已提交
460
		struct cpufreq_governor *t;
461

462
		mutex_lock(&cpufreq_governor_mutex);
463

464
		t = find_governor(str_governor);
465

466
		if (t == NULL) {
467
			int ret;
468

469 470 471
			mutex_unlock(&cpufreq_governor_mutex);
			ret = request_module("cpufreq_%s", str_governor);
			mutex_lock(&cpufreq_governor_mutex);
472

473
			if (ret == 0)
474
				t = find_governor(str_governor);
475 476
		}

477 478 479
		if (t != NULL) {
			*governor = t;
			err = 0;
L
Linus Torvalds 已提交
480
		}
481

482
		mutex_unlock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
483
	}
484
out:
485
	return err;
L
Linus Torvalds 已提交
486 487 488
}

/**
489 490
 * cpufreq_per_cpu_attr_read() / show_##file_name() -
 * print out cpufreq information
L
Linus Torvalds 已提交
491 492 493 494 495
 *
 * Write out information from cpufreq_driver->policy[cpu]; object must be
 * "unsigned int".
 */

496 497
#define show_one(file_name, object)			\
static ssize_t show_##file_name				\
D
Dave Jones 已提交
498
(struct cpufreq_policy *policy, char *buf)		\
499
{							\
500
	return sprintf(buf, "%u\n", policy->object);	\
L
Linus Torvalds 已提交
501 502 503 504
}

show_one(cpuinfo_min_freq, cpuinfo.min_freq);
show_one(cpuinfo_max_freq, cpuinfo.max_freq);
505
show_one(cpuinfo_transition_latency, cpuinfo.transition_latency);
L
Linus Torvalds 已提交
506 507
show_one(scaling_min_freq, min);
show_one(scaling_max_freq, max);
508

509
static ssize_t show_scaling_cur_freq(struct cpufreq_policy *policy, char *buf)
510 511 512 513 514 515 516 517 518
{
	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
519

520
static int cpufreq_set_policy(struct cpufreq_policy *policy,
521
				struct cpufreq_policy *new_policy);
522

L
Linus Torvalds 已提交
523 524 525 526 527
/**
 * 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
528
(struct cpufreq_policy *policy, const char *buf, size_t count)		\
L
Linus Torvalds 已提交
529
{									\
530
	int ret, temp;							\
L
Linus Torvalds 已提交
531 532 533 534 535 536
	struct cpufreq_policy new_policy;				\
									\
	ret = cpufreq_get_policy(&new_policy, policy->cpu);		\
	if (ret)							\
		return -EINVAL;						\
									\
537
	ret = sscanf(buf, "%u", &new_policy.object);			\
L
Linus Torvalds 已提交
538 539 540
	if (ret != 1)							\
		return -EINVAL;						\
									\
541
	temp = new_policy.object;					\
542
	ret = cpufreq_set_policy(policy, &new_policy);		\
543 544
	if (!ret)							\
		policy->user_policy.object = temp;			\
L
Linus Torvalds 已提交
545 546 547 548
									\
	return ret ? ret : count;					\
}

549 550
store_one(scaling_min_freq, min);
store_one(scaling_max_freq, max);
L
Linus Torvalds 已提交
551 552 553 554

/**
 * show_cpuinfo_cur_freq - current CPU frequency as detected by hardware
 */
D
Dave Jones 已提交
555 556
static ssize_t show_cpuinfo_cur_freq(struct cpufreq_policy *policy,
					char *buf)
L
Linus Torvalds 已提交
557
{
558
	unsigned int cur_freq = __cpufreq_get(policy);
L
Linus Torvalds 已提交
559 560 561 562 563 564 565 566
	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
567
static ssize_t show_scaling_governor(struct cpufreq_policy *policy, char *buf)
L
Linus Torvalds 已提交
568
{
569
	if (policy->policy == CPUFREQ_POLICY_POWERSAVE)
L
Linus Torvalds 已提交
570 571 572 573
		return sprintf(buf, "powersave\n");
	else if (policy->policy == CPUFREQ_POLICY_PERFORMANCE)
		return sprintf(buf, "performance\n");
	else if (policy->governor)
574
		return scnprintf(buf, CPUFREQ_NAME_PLEN, "%s\n",
575
				policy->governor->name);
L
Linus Torvalds 已提交
576 577 578 579 580 581
	return -EINVAL;
}

/**
 * store_scaling_governor - store policy for the specified CPU
 */
D
Dave Jones 已提交
582 583
static ssize_t store_scaling_governor(struct cpufreq_policy *policy,
					const char *buf, size_t count)
L
Linus Torvalds 已提交
584
{
585
	int ret;
L
Linus Torvalds 已提交
586 587 588 589 590 591 592
	char	str_governor[16];
	struct cpufreq_policy new_policy;

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

593
	ret = sscanf(buf, "%15s", str_governor);
L
Linus Torvalds 已提交
594 595 596
	if (ret != 1)
		return -EINVAL;

597 598
	if (cpufreq_parse_governor(str_governor, &new_policy.policy,
						&new_policy.governor))
L
Linus Torvalds 已提交
599 600
		return -EINVAL;

601
	ret = cpufreq_set_policy(policy, &new_policy);
602 603 604 605

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

606 607 608 609
	if (ret)
		return ret;
	else
		return count;
L
Linus Torvalds 已提交
610 611 612 613 614
}

/**
 * show_scaling_driver - show the cpufreq driver currently loaded
 */
D
Dave Jones 已提交
615
static ssize_t show_scaling_driver(struct cpufreq_policy *policy, char *buf)
L
Linus Torvalds 已提交
616
{
617
	return scnprintf(buf, CPUFREQ_NAME_PLEN, "%s\n", cpufreq_driver->name);
L
Linus Torvalds 已提交
618 619 620 621 622
}

/**
 * show_scaling_available_governors - show the available CPUfreq governors
 */
D
Dave Jones 已提交
623 624
static ssize_t show_scaling_available_governors(struct cpufreq_policy *policy,
						char *buf)
L
Linus Torvalds 已提交
625 626 627 628
{
	ssize_t i = 0;
	struct cpufreq_governor *t;

629
	if (!has_target()) {
L
Linus Torvalds 已提交
630 631 632 633 634
		i += sprintf(buf, "performance powersave");
		goto out;
	}

	list_for_each_entry(t, &cpufreq_governor_list, governor_list) {
635 636
		if (i >= (ssize_t) ((PAGE_SIZE / sizeof(char))
		    - (CPUFREQ_NAME_LEN + 2)))
L
Linus Torvalds 已提交
637
			goto out;
638
		i += scnprintf(&buf[i], CPUFREQ_NAME_PLEN, "%s ", t->name);
L
Linus Torvalds 已提交
639
	}
640
out:
L
Linus Torvalds 已提交
641 642 643
	i += sprintf(&buf[i], "\n");
	return i;
}
644

645
ssize_t cpufreq_show_cpus(const struct cpumask *mask, char *buf)
L
Linus Torvalds 已提交
646 647 648 649
{
	ssize_t i = 0;
	unsigned int cpu;

650
	for_each_cpu(cpu, mask) {
L
Linus Torvalds 已提交
651 652 653 654
		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))
655
			break;
L
Linus Torvalds 已提交
656 657 658 659
	}
	i += sprintf(&buf[i], "\n");
	return i;
}
660
EXPORT_SYMBOL_GPL(cpufreq_show_cpus);
L
Linus Torvalds 已提交
661

662 663 664 665 666 667
/**
 * 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)
{
668
	return cpufreq_show_cpus(policy->related_cpus, buf);
669 670 671 672 673 674 675
}

/**
 * show_affected_cpus - show the CPUs affected by each transition
 */
static ssize_t show_affected_cpus(struct cpufreq_policy *policy, char *buf)
{
676
	return cpufreq_show_cpus(policy->cpus, buf);
677 678
}

679
static ssize_t store_scaling_setspeed(struct cpufreq_policy *policy,
D
Dave Jones 已提交
680
					const char *buf, size_t count)
681 682 683 684
{
	unsigned int freq = 0;
	unsigned int ret;

685
	if (!policy->governor || !policy->governor->store_setspeed)
686 687 688 689 690 691 692 693 694 695 696 697 698
		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)
{
699
	if (!policy->governor || !policy->governor->show_setspeed)
700 701 702 703
		return sprintf(buf, "<unsupported>\n");

	return policy->governor->show_setspeed(policy, buf);
}
L
Linus Torvalds 已提交
704

705
/**
706
 * show_bios_limit - show the current cpufreq HW/BIOS limitation
707 708 709 710 711
 */
static ssize_t show_bios_limit(struct cpufreq_policy *policy, char *buf)
{
	unsigned int limit;
	int ret;
712 713
	if (cpufreq_driver->bios_limit) {
		ret = cpufreq_driver->bios_limit(policy->cpu, &limit);
714 715 716 717 718 719
		if (!ret)
			return sprintf(buf, "%u\n", limit);
	}
	return sprintf(buf, "%u\n", policy->cpuinfo.max_freq);
}

720 721 722 723 724 725 726 727 728 729 730 731 732 733
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
734

D
Dave Jones 已提交
735
static struct attribute *default_attrs[] = {
L
Linus Torvalds 已提交
736 737
	&cpuinfo_min_freq.attr,
	&cpuinfo_max_freq.attr,
738
	&cpuinfo_transition_latency.attr,
L
Linus Torvalds 已提交
739 740 741
	&scaling_min_freq.attr,
	&scaling_max_freq.attr,
	&affected_cpus.attr,
742
	&related_cpus.attr,
L
Linus Torvalds 已提交
743 744 745
	&scaling_governor.attr,
	&scaling_driver.attr,
	&scaling_available_governors.attr,
746
	&scaling_setspeed.attr,
L
Linus Torvalds 已提交
747 748 749
	NULL
};

750 751
#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
752

753
static ssize_t show(struct kobject *kobj, struct attribute *attr, char *buf)
L
Linus Torvalds 已提交
754
{
D
Dave Jones 已提交
755 756
	struct cpufreq_policy *policy = to_policy(kobj);
	struct freq_attr *fattr = to_attr(attr);
757
	ssize_t ret;
758 759

	if (!down_read_trylock(&cpufreq_rwsem))
760
		return -EINVAL;
761

762
	down_read(&policy->rwsem);
763

764 765 766 767 768
	if (fattr->show)
		ret = fattr->show(policy, buf);
	else
		ret = -EIO;

769
	up_read(&policy->rwsem);
770
	up_read(&cpufreq_rwsem);
771

L
Linus Torvalds 已提交
772 773 774
	return ret;
}

D
Dave Jones 已提交
775 776
static ssize_t store(struct kobject *kobj, struct attribute *attr,
		     const char *buf, size_t count)
L
Linus Torvalds 已提交
777
{
D
Dave Jones 已提交
778 779
	struct cpufreq_policy *policy = to_policy(kobj);
	struct freq_attr *fattr = to_attr(attr);
780
	ssize_t ret = -EINVAL;
781

782 783 784 785 786
	get_online_cpus();

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

787
	if (!down_read_trylock(&cpufreq_rwsem))
788
		goto unlock;
789

790
	down_write(&policy->rwsem);
791

792 793 794 795 796
	if (fattr->store)
		ret = fattr->store(policy, buf, count);
	else
		ret = -EIO;

797
	up_write(&policy->rwsem);
798 799

	up_read(&cpufreq_rwsem);
800 801 802
unlock:
	put_online_cpus();

L
Linus Torvalds 已提交
803 804 805
	return ret;
}

D
Dave Jones 已提交
806
static void cpufreq_sysfs_release(struct kobject *kobj)
L
Linus Torvalds 已提交
807
{
D
Dave Jones 已提交
808
	struct cpufreq_policy *policy = to_policy(kobj);
809
	pr_debug("last reference is dropped\n");
L
Linus Torvalds 已提交
810 811 812
	complete(&policy->kobj_unregister);
}

813
static const struct sysfs_ops sysfs_ops = {
L
Linus Torvalds 已提交
814 815 816 817 818 819 820 821 822 823
	.show	= show,
	.store	= store,
};

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

824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866
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);

867
/* symlink affected CPUs */
868
static int cpufreq_add_dev_symlink(struct cpufreq_policy *policy)
869 870 871 872 873
{
	unsigned int j;
	int ret = 0;

	for_each_cpu(j, policy->cpus) {
874
		struct device *cpu_dev;
875

876
		if (j == policy->cpu)
877 878
			continue;

879
		pr_debug("Adding link for CPU: %u\n", j);
880 881
		cpu_dev = get_cpu_device(j);
		ret = sysfs_create_link(&cpu_dev->kobj, &policy->kobj,
882
					"cpufreq");
883 884
		if (ret)
			break;
885 886 887 888
	}
	return ret;
}

889
static int cpufreq_add_dev_interface(struct cpufreq_policy *policy,
890
				     struct device *dev)
891 892 893 894 895
{
	struct freq_attr **drv_attr;
	int ret = 0;

	/* set up files for this cpu device */
896
	drv_attr = cpufreq_driver->attr;
V
Viresh Kumar 已提交
897
	while (drv_attr && *drv_attr) {
898 899
		ret = sysfs_create_file(&policy->kobj, &((*drv_attr)->attr));
		if (ret)
900
			return ret;
901 902
		drv_attr++;
	}
903
	if (cpufreq_driver->get) {
904 905
		ret = sysfs_create_file(&policy->kobj, &cpuinfo_cur_freq.attr);
		if (ret)
906
			return ret;
907
	}
908 909 910

	ret = sysfs_create_file(&policy->kobj, &scaling_cur_freq.attr);
	if (ret)
911
		return ret;
912

913
	if (cpufreq_driver->bios_limit) {
914 915
		ret = sysfs_create_file(&policy->kobj, &bios_limit.attr);
		if (ret)
916
			return ret;
917
	}
918

919
	return cpufreq_add_dev_symlink(policy);
920 921 922 923
}

static void cpufreq_init_policy(struct cpufreq_policy *policy)
{
924
	struct cpufreq_governor *gov = NULL;
925 926 927
	struct cpufreq_policy new_policy;
	int ret = 0;

928
	memcpy(&new_policy, policy, sizeof(*policy));
929

930
	/* Update governor of new_policy to the governor used before hotplug */
931
	gov = find_governor(per_cpu(cpufreq_cpu_governor, policy->cpu));
932 933 934 935 936 937 938 939
	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;

940 941
	/* Use the default policy if its valid. */
	if (cpufreq_driver->setpolicy)
942
		cpufreq_parse_governor(gov->name, &new_policy.policy, NULL);
943 944

	/* set default policy */
945
	ret = cpufreq_set_policy(policy, &new_policy);
946
	if (ret) {
947
		pr_debug("setting policy failed\n");
948 949
		if (cpufreq_driver->exit)
			cpufreq_driver->exit(policy);
950
	}
951 952
}

953
static int cpufreq_add_policy_cpu(struct cpufreq_policy *policy,
954
				  unsigned int cpu, struct device *dev)
955
{
956
	int ret = 0;
957 958
	unsigned long flags;

959
	if (has_target()) {
960 961 962 963 964 965
		ret = __cpufreq_governor(policy, CPUFREQ_GOV_STOP);
		if (ret) {
			pr_err("%s: Failed to stop governor\n", __func__);
			return ret;
		}
	}
966

967
	down_write(&policy->rwsem);
V
Viresh Kumar 已提交
968

969
	write_lock_irqsave(&cpufreq_driver_lock, flags);
V
Viresh Kumar 已提交
970

971 972
	cpumask_set_cpu(cpu, policy->cpus);
	per_cpu(cpufreq_cpu_data, cpu) = policy;
973
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
974

975
	up_write(&policy->rwsem);
V
Viresh Kumar 已提交
976

977
	if (has_target()) {
978 979 980 981 982
		ret = __cpufreq_governor(policy, CPUFREQ_GOV_START);
		if (!ret)
			ret = __cpufreq_governor(policy, CPUFREQ_GOV_LIMITS);

		if (ret) {
983 984 985
			pr_err("%s: Failed to start governor\n", __func__);
			return ret;
		}
986
	}
987

988
	return sysfs_create_link(&dev->kobj, &policy->kobj, "cpufreq");
989
}
L
Linus Torvalds 已提交
990

991 992 993 994 995
static struct cpufreq_policy *cpufreq_policy_restore(unsigned int cpu)
{
	struct cpufreq_policy *policy;
	unsigned long flags;

996
	read_lock_irqsave(&cpufreq_driver_lock, flags);
997 998 999

	policy = per_cpu(cpufreq_cpu_data_fallback, cpu);

1000
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
1001

1002 1003
	if (policy)
		policy->governor = NULL;
1004

1005 1006 1007
	return policy;
}

1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021
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;

1022
	INIT_LIST_HEAD(&policy->policy_list);
1023
	init_rwsem(&policy->rwsem);
1024 1025
	spin_lock_init(&policy->transition_lock);
	init_waitqueue_head(&policy->transition_wait);
1026 1027
	init_completion(&policy->kobj_unregister);
	INIT_WORK(&policy->update, handle_update);
1028

1029 1030 1031 1032 1033 1034 1035 1036 1037 1038
	return policy;

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

	return NULL;
}

1039 1040 1041 1042 1043
static void cpufreq_policy_put_kobj(struct cpufreq_policy *policy)
{
	struct kobject *kobj;
	struct completion *cmp;

1044 1045 1046
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
			CPUFREQ_REMOVE_POLICY, policy);

1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062
	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");
}

1063 1064 1065 1066 1067 1068 1069
static void cpufreq_policy_free(struct cpufreq_policy *policy)
{
	free_cpumask_var(policy->related_cpus);
	free_cpumask_var(policy->cpus);
	kfree(policy);
}

1070 1071
static int update_policy_cpu(struct cpufreq_policy *policy, unsigned int cpu,
			     struct device *cpu_dev)
1072
{
1073 1074
	int ret;

1075
	if (WARN_ON(cpu == policy->cpu))
1076 1077 1078 1079 1080 1081 1082 1083
		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;
	}
1084

1085
	down_write(&policy->rwsem);
1086
	policy->cpu = cpu;
1087
	up_write(&policy->rwsem);
1088

1089
	return 0;
1090 1091
}

1092
static int __cpufreq_add_dev(struct device *dev, struct subsys_interface *sif)
L
Linus Torvalds 已提交
1093
{
1094
	unsigned int j, cpu = dev->id;
1095
	int ret = -ENOMEM;
1096
	struct cpufreq_policy *policy;
L
Linus Torvalds 已提交
1097
	unsigned long flags;
1098
	bool recover_policy = cpufreq_suspended;
L
Linus Torvalds 已提交
1099

1100 1101 1102
	if (cpu_is_offline(cpu))
		return 0;

1103
	pr_debug("adding CPU %u\n", cpu);
L
Linus Torvalds 已提交
1104 1105 1106

	/* check whether a different CPU already registered this
	 * CPU because it is in the same boat. */
1107 1108
	policy = cpufreq_cpu_get_raw(cpu);
	if (unlikely(policy))
L
Linus Torvalds 已提交
1109
		return 0;
1110

1111 1112 1113
	if (!down_read_trylock(&cpufreq_rwsem))
		return 0;

1114
	/* Check if this cpu was hot-unplugged earlier and has siblings */
1115
	read_lock_irqsave(&cpufreq_driver_lock, flags);
1116 1117
	list_for_each_entry(policy, &cpufreq_policy_list, policy_list) {
		if (cpumask_test_cpu(cpu, policy->related_cpus)) {
1118
			read_unlock_irqrestore(&cpufreq_driver_lock, flags);
1119
			ret = cpufreq_add_policy_cpu(policy, cpu, dev);
1120 1121
			up_read(&cpufreq_rwsem);
			return ret;
V
Viresh Kumar 已提交
1122
		}
1123
	}
1124
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
1125

1126 1127 1128 1129
	/*
	 * Restore the saved policy when doing light-weight init and fall back
	 * to the full init if that fails.
	 */
1130
	policy = recover_policy ? cpufreq_policy_restore(cpu) : NULL;
1131
	if (!policy) {
1132
		recover_policy = false;
1133
		policy = cpufreq_policy_alloc();
1134 1135 1136
		if (!policy)
			goto nomem_out;
	}
1137 1138 1139 1140 1141 1142 1143

	/*
	 * 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().
	 */
1144 1145 1146
	if (recover_policy && cpu != policy->cpu)
		WARN_ON(update_policy_cpu(policy, cpu, dev));
	else
1147 1148
		policy->cpu = cpu;

1149
	cpumask_copy(policy->cpus, cpumask_of(cpu));
L
Linus Torvalds 已提交
1150 1151 1152 1153

	/* call driver. From then on the cpufreq must be able
	 * to accept all calls to ->verify and ->setpolicy for this CPU
	 */
1154
	ret = cpufreq_driver->init(policy);
L
Linus Torvalds 已提交
1155
	if (ret) {
1156
		pr_debug("initialization failed\n");
V
Viresh Kumar 已提交
1157
		goto err_set_policy_cpu;
L
Linus Torvalds 已提交
1158
	}
V
Viresh Kumar 已提交
1159

1160 1161
	down_write(&policy->rwsem);

1162 1163 1164 1165 1166 1167 1168 1169 1170
	/* 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);

1171
	if (!recover_policy) {
1172 1173
		policy->user_policy.min = policy->min;
		policy->user_policy.max = policy->max;
1174 1175 1176 1177 1178 1179 1180 1181 1182

		/* 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;
		}
1183 1184
	}

1185 1186 1187 1188 1189
	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);

1190
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
1191 1192 1193 1194 1195 1196 1197
		policy->cur = cpufreq_driver->get(policy->cpu);
		if (!policy->cur) {
			pr_err("%s: ->get() failed\n", __func__);
			goto err_get_freq;
		}
	}

1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 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
	/*
	 * 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);
		}
	}

1238 1239 1240
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
				     CPUFREQ_START, policy);

1241
	if (!recover_policy) {
1242
		ret = cpufreq_add_dev_interface(policy, dev);
1243 1244
		if (ret)
			goto err_out_unregister;
1245 1246
		blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
				CPUFREQ_CREATE_POLICY, policy);
1247
	}
1248

1249 1250 1251 1252
	write_lock_irqsave(&cpufreq_driver_lock, flags);
	list_add(&policy->policy_list, &cpufreq_policy_list);
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);

1253 1254
	cpufreq_init_policy(policy);

1255
	if (!recover_policy) {
1256 1257 1258
		policy->user_policy.policy = policy->policy;
		policy->user_policy.governor = policy->governor;
	}
1259
	up_write(&policy->rwsem);
1260

1261
	kobject_uevent(&policy->kobj, KOBJ_ADD);
1262

1263 1264
	up_read(&cpufreq_rwsem);

1265 1266 1267 1268
	/* Callback for handling stuff after policy is ready */
	if (cpufreq_driver->ready)
		cpufreq_driver->ready(policy);

1269
	pr_debug("initialization complete\n");
1270

L
Linus Torvalds 已提交
1271 1272 1273
	return 0;

err_out_unregister:
1274
err_get_freq:
1275
	write_lock_irqsave(&cpufreq_driver_lock, flags);
1276
	for_each_cpu(j, policy->cpus)
1277
		per_cpu(cpufreq_cpu_data, j) = NULL;
1278
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
1279

1280 1281 1282 1283 1284
	if (!recover_policy) {
		kobject_put(&policy->kobj);
		wait_for_completion(&policy->kobj_unregister);
	}
err_init_policy_kobj:
1285 1286
	up_write(&policy->rwsem);

1287 1288
	if (cpufreq_driver->exit)
		cpufreq_driver->exit(policy);
V
Viresh Kumar 已提交
1289
err_set_policy_cpu:
1290
	if (recover_policy) {
1291 1292
		/* Do not leave stale fallback data behind. */
		per_cpu(cpufreq_cpu_data_fallback, cpu) = NULL;
1293
		cpufreq_policy_put_kobj(policy);
1294
	}
1295
	cpufreq_policy_free(policy);
1296

L
Linus Torvalds 已提交
1297
nomem_out:
1298 1299
	up_read(&cpufreq_rwsem);

L
Linus Torvalds 已提交
1300 1301 1302
	return ret;
}

1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313
/**
 * 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)
{
1314
	return __cpufreq_add_dev(dev, sif);
1315 1316
}

1317
static int __cpufreq_remove_dev_prepare(struct device *dev,
1318
					struct subsys_interface *sif)
L
Linus Torvalds 已提交
1319
{
1320
	unsigned int cpu = dev->id, cpus;
1321
	int ret;
L
Linus Torvalds 已提交
1322
	unsigned long flags;
1323
	struct cpufreq_policy *policy;
L
Linus Torvalds 已提交
1324

1325
	pr_debug("%s: unregistering CPU %u\n", __func__, cpu);
L
Linus Torvalds 已提交
1326

1327
	write_lock_irqsave(&cpufreq_driver_lock, flags);
V
Viresh Kumar 已提交
1328

1329
	policy = per_cpu(cpufreq_cpu_data, cpu);
V
Viresh Kumar 已提交
1330

1331
	/* Save the policy somewhere when doing a light-weight tear-down */
1332
	if (cpufreq_suspended)
1333
		per_cpu(cpufreq_cpu_data_fallback, cpu) = policy;
1334

1335
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
1336

1337
	if (!policy) {
1338
		pr_debug("%s: No cpu_data found\n", __func__);
L
Linus Torvalds 已提交
1339 1340 1341
		return -EINVAL;
	}

1342
	if (has_target()) {
1343 1344 1345 1346 1347
		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
1348

1349
		strncpy(per_cpu(cpufreq_cpu_governor, cpu),
1350
			policy->governor->name, CPUFREQ_NAME_LEN);
1351
	}
L
Linus Torvalds 已提交
1352

1353
	down_read(&policy->rwsem);
1354
	cpus = cpumask_weight(policy->cpus);
1355
	up_read(&policy->rwsem);
1356

1357
	if (cpu != policy->cpu) {
1358
		sysfs_remove_link(&dev->kobj, "cpufreq");
1359
	} else if (cpus > 1) {
1360 1361 1362
		/* Nominate new CPU */
		int new_cpu = cpumask_any_but(policy->cpus, cpu);
		struct device *cpu_dev = get_cpu_device(new_cpu);
1363

1364 1365 1366 1367 1368 1369 1370 1371
		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
1372
		}
1373 1374 1375 1376

		if (!cpufreq_suspended)
			pr_debug("%s: policy Kobject moved to cpu: %d from: %d\n",
				 __func__, new_cpu, cpu);
1377
	} else if (cpufreq_driver->stop_cpu) {
1378
		cpufreq_driver->stop_cpu(policy);
L
Linus Torvalds 已提交
1379 1380
	}

1381 1382 1383 1384
	return 0;
}

static int __cpufreq_remove_dev_finish(struct device *dev,
1385
				       struct subsys_interface *sif)
1386 1387 1388 1389 1390 1391
{
	unsigned int cpu = dev->id, cpus;
	int ret;
	unsigned long flags;
	struct cpufreq_policy *policy;

1392
	write_lock_irqsave(&cpufreq_driver_lock, flags);
1393
	policy = per_cpu(cpufreq_cpu_data, cpu);
1394 1395
	per_cpu(cpufreq_cpu_data, cpu) = NULL;
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
1396 1397 1398 1399 1400 1401

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

1402
	down_write(&policy->rwsem);
1403
	cpus = cpumask_weight(policy->cpus);
1404 1405 1406

	if (cpus > 1)
		cpumask_clear_cpu(cpu, policy->cpus);
1407
	up_write(&policy->rwsem);
1408

1409 1410
	/* If cpu is last user of policy, free policy */
	if (cpus == 1) {
1411
		if (has_target()) {
1412 1413 1414 1415
			ret = __cpufreq_governor(policy,
					CPUFREQ_GOV_POLICY_EXIT);
			if (ret) {
				pr_err("%s: Failed to exit governor\n",
1416
				       __func__);
1417 1418
				return ret;
			}
1419
		}
1420

1421
		if (!cpufreq_suspended)
1422
			cpufreq_policy_put_kobj(policy);
1423

1424 1425 1426 1427
		/*
		 * 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.
1428
		 */
1429
		if (cpufreq_driver->exit)
1430
			cpufreq_driver->exit(policy);
1431

1432 1433 1434 1435 1436
		/* 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);

1437
		if (!cpufreq_suspended)
1438
			cpufreq_policy_free(policy);
1439 1440 1441 1442 1443 1444 1445 1446
	} 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;
1447
		}
1448
	}
L
Linus Torvalds 已提交
1449 1450 1451 1452

	return 0;
}

1453
/**
1454
 * cpufreq_remove_dev - remove a CPU device
1455 1456 1457
 *
 * Removes the cpufreq interface for a CPU device.
 */
1458
static int cpufreq_remove_dev(struct device *dev, struct subsys_interface *sif)
1459
{
1460
	unsigned int cpu = dev->id;
1461
	int ret;
1462 1463 1464 1465

	if (cpu_is_offline(cpu))
		return 0;

1466
	ret = __cpufreq_remove_dev_prepare(dev, sif);
1467 1468

	if (!ret)
1469
		ret = __cpufreq_remove_dev_finish(dev, sif);
1470 1471

	return ret;
1472 1473
}

1474
static void handle_update(struct work_struct *work)
L
Linus Torvalds 已提交
1475
{
1476 1477 1478
	struct cpufreq_policy *policy =
		container_of(work, struct cpufreq_policy, update);
	unsigned int cpu = policy->cpu;
1479
	pr_debug("handle_update for cpu %u called\n", cpu);
L
Linus Torvalds 已提交
1480 1481 1482 1483
	cpufreq_update_policy(cpu);
}

/**
1484 1485
 *	cpufreq_out_of_sync - If actual and saved CPU frequency differs, we're
 *	in deep trouble.
1486
 *	@policy: policy managing CPUs
L
Linus Torvalds 已提交
1487 1488
 *	@new_freq: CPU frequency the CPU actually runs at
 *
1489 1490
 *	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
1491
 */
1492
static void cpufreq_out_of_sync(struct cpufreq_policy *policy,
1493
				unsigned int new_freq)
L
Linus Torvalds 已提交
1494 1495
{
	struct cpufreq_freqs freqs;
1496

1497
	pr_debug("Warning: CPU frequency out of sync: cpufreq and timing core thinks of %u, is %u kHz\n",
1498
		 policy->cur, new_freq);
L
Linus Torvalds 已提交
1499

1500
	freqs.old = policy->cur;
L
Linus Torvalds 已提交
1501
	freqs.new = new_freq;
1502

1503 1504
	cpufreq_freq_transition_begin(policy, &freqs);
	cpufreq_freq_transition_end(policy, &freqs, 0);
L
Linus Torvalds 已提交
1505 1506
}

1507
/**
D
Dhaval Giani 已提交
1508
 * cpufreq_quick_get - get the CPU frequency (in kHz) from policy->cur
1509 1510 1511 1512 1513 1514 1515
 * @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)
{
1516
	struct cpufreq_policy *policy;
1517
	unsigned int ret_freq = 0;
1518

1519 1520
	if (cpufreq_driver && cpufreq_driver->setpolicy && cpufreq_driver->get)
		return cpufreq_driver->get(cpu);
1521 1522

	policy = cpufreq_cpu_get(cpu);
1523
	if (policy) {
1524
		ret_freq = policy->cur;
1525 1526 1527
		cpufreq_cpu_put(policy);
	}

D
Dave Jones 已提交
1528
	return ret_freq;
1529 1530 1531
}
EXPORT_SYMBOL(cpufreq_quick_get);

1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551
/**
 * 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);

1552
static unsigned int __cpufreq_get(struct cpufreq_policy *policy)
L
Linus Torvalds 已提交
1553
{
1554
	unsigned int ret_freq = 0;
1555

1556
	if (!cpufreq_driver->get)
D
Dave Jones 已提交
1557
		return ret_freq;
L
Linus Torvalds 已提交
1558

1559
	ret_freq = cpufreq_driver->get(policy->cpu);
L
Linus Torvalds 已提交
1560

1561
	if (ret_freq && policy->cur &&
1562
		!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
1563 1564 1565
		/* verify no discrepancy between actual and
					saved value exists */
		if (unlikely(ret_freq != policy->cur)) {
1566
			cpufreq_out_of_sync(policy, ret_freq);
L
Linus Torvalds 已提交
1567 1568 1569 1570
			schedule_work(&policy->update);
		}
	}

D
Dave Jones 已提交
1571
	return ret_freq;
1572
}
L
Linus Torvalds 已提交
1573

1574 1575 1576 1577 1578 1579 1580 1581
/**
 * 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)
{
1582
	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
1583 1584
	unsigned int ret_freq = 0;

1585 1586
	if (policy) {
		down_read(&policy->rwsem);
1587
		ret_freq = __cpufreq_get(policy);
1588
		up_read(&policy->rwsem);
1589

1590 1591
		cpufreq_cpu_put(policy);
	}
1592

D
Dave Jones 已提交
1593
	return ret_freq;
L
Linus Torvalds 已提交
1594 1595 1596
}
EXPORT_SYMBOL(cpufreq_get);

1597 1598 1599 1600 1601
static struct subsys_interface cpufreq_interface = {
	.name		= "cpufreq",
	.subsys		= &cpu_subsys,
	.add_dev	= cpufreq_add_dev,
	.remove_dev	= cpufreq_remove_dev,
1602 1603
};

1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629
/*
 * 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);

1630
/**
1631
 * cpufreq_suspend() - Suspend CPUFreq governors
1632
 *
1633 1634 1635 1636
 * 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.
1637
 */
1638
void cpufreq_suspend(void)
1639
{
1640
	struct cpufreq_policy *policy;
1641

1642 1643
	if (!cpufreq_driver)
		return;
1644

1645
	if (!has_target())
1646
		goto suspend;
1647

1648 1649 1650 1651 1652 1653 1654 1655 1656 1657
	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);
1658
	}
1659 1660 1661

suspend:
	cpufreq_suspended = true;
1662 1663
}

L
Linus Torvalds 已提交
1664
/**
1665
 * cpufreq_resume() - Resume CPUFreq governors
L
Linus Torvalds 已提交
1666
 *
1667 1668
 * Called during system wide Suspend/Hibernate cycle for resuming governors that
 * are suspended with cpufreq_suspend().
L
Linus Torvalds 已提交
1669
 */
1670
void cpufreq_resume(void)
L
Linus Torvalds 已提交
1671
{
1672
	struct cpufreq_policy *policy;
L
Linus Torvalds 已提交
1673

1674 1675
	if (!cpufreq_driver)
		return;
L
Linus Torvalds 已提交
1676

1677 1678
	cpufreq_suspended = false;

1679
	if (!has_target())
1680
		return;
L
Linus Torvalds 已提交
1681

1682
	pr_debug("%s: Resuming Governors\n", __func__);
L
Linus Torvalds 已提交
1683

1684
	list_for_each_entry(policy, &cpufreq_policy_list, policy_list) {
1685 1686 1687 1688
		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)
1689 1690 1691
		    || __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
1692

1693 1694 1695 1696 1697 1698 1699 1700 1701
		/*
		 * 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
1702

1703 1704 1705 1706 1707 1708 1709 1710
/**
 *	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)
{
1711 1712 1713 1714
	if (cpufreq_driver)
		return cpufreq_driver->name;

	return NULL;
1715 1716
}
EXPORT_SYMBOL_GPL(cpufreq_get_current_driver);
L
Linus Torvalds 已提交
1717

1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732
/**
 *	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
1733 1734 1735 1736 1737 1738 1739 1740 1741
/*********************************************************************
 *                     NOTIFIER LISTS INTERFACE                      *
 *********************************************************************/

/**
 *	cpufreq_register_notifier - register a driver with cpufreq
 *	@nb: notifier function to register
 *      @list: CPUFREQ_TRANSITION_NOTIFIER or CPUFREQ_POLICY_NOTIFIER
 *
1742
 *	Add a driver to one of two lists: either a list of drivers that
L
Linus Torvalds 已提交
1743 1744 1745 1746 1747
 *      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
1748
 *	blocking_notifier_chain_register.
L
Linus Torvalds 已提交
1749 1750 1751 1752 1753
 */
int cpufreq_register_notifier(struct notifier_block *nb, unsigned int list)
{
	int ret;

1754 1755 1756
	if (cpufreq_disabled())
		return -EINVAL;

1757 1758
	WARN_ON(!init_cpufreq_transition_notifier_list_called);

L
Linus Torvalds 已提交
1759 1760
	switch (list) {
	case CPUFREQ_TRANSITION_NOTIFIER:
1761
		ret = srcu_notifier_chain_register(
1762
				&cpufreq_transition_notifier_list, nb);
L
Linus Torvalds 已提交
1763 1764
		break;
	case CPUFREQ_POLICY_NOTIFIER:
1765 1766
		ret = blocking_notifier_chain_register(
				&cpufreq_policy_notifier_list, nb);
L
Linus Torvalds 已提交
1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778
		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
1779
 *	@list: CPUFREQ_TRANSITION_NOTIFIER or CPUFREQ_POLICY_NOTIFIER
L
Linus Torvalds 已提交
1780 1781 1782 1783
 *
 *	Remove a driver from the CPU frequency notifier list.
 *
 *	This function may sleep, and has the same return conditions as
1784
 *	blocking_notifier_chain_unregister.
L
Linus Torvalds 已提交
1785 1786 1787 1788 1789
 */
int cpufreq_unregister_notifier(struct notifier_block *nb, unsigned int list)
{
	int ret;

1790 1791 1792
	if (cpufreq_disabled())
		return -EINVAL;

L
Linus Torvalds 已提交
1793 1794
	switch (list) {
	case CPUFREQ_TRANSITION_NOTIFIER:
1795
		ret = srcu_notifier_chain_unregister(
1796
				&cpufreq_transition_notifier_list, nb);
L
Linus Torvalds 已提交
1797 1798
		break;
	case CPUFREQ_POLICY_NOTIFIER:
1799 1800
		ret = blocking_notifier_chain_unregister(
				&cpufreq_policy_notifier_list, nb);
L
Linus Torvalds 已提交
1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814
		break;
	default:
		ret = -EINVAL;
	}

	return ret;
}
EXPORT_SYMBOL(cpufreq_unregister_notifier);


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

1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840
/* 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;
}

1841 1842 1843
static int __target_index(struct cpufreq_policy *policy,
			  struct cpufreq_frequency_table *freq_table, int index)
{
1844 1845
	struct cpufreq_freqs freqs = {.old = policy->cur, .flags = 0};
	unsigned int intermediate_freq = 0;
1846 1847 1848 1849 1850
	int retval = -EINVAL;
	bool notify;

	notify = !(cpufreq_driver->flags & CPUFREQ_ASYNC_NOTIFICATION);
	if (notify) {
1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861
		/* 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;
		}
1862

1863
		freqs.new = freq_table[index].frequency;
1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874
		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);

1875
	if (notify) {
1876 1877
		cpufreq_freq_transition_end(policy, &freqs, retval);

1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891
		/*
		 * 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);
		}
	}

1892 1893 1894
	return retval;
}

L
Linus Torvalds 已提交
1895 1896 1897 1898
int __cpufreq_driver_target(struct cpufreq_policy *policy,
			    unsigned int target_freq,
			    unsigned int relation)
{
1899
	unsigned int old_target_freq = target_freq;
1900
	int retval = -EINVAL;
1901

1902 1903 1904
	if (cpufreq_disabled())
		return -ENODEV;

1905 1906 1907 1908 1909 1910 1911
	/* 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",
1912
		 policy->cpu, target_freq, relation, old_target_freq);
1913

1914 1915 1916 1917 1918 1919
	/*
	 * 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.
	 */
1920 1921 1922
	if (target_freq == policy->cur)
		return 0;

1923 1924 1925
	/* Save last value to restore later on errors */
	policy->restore_freq = policy->cur;

1926 1927
	if (cpufreq_driver->target)
		retval = cpufreq_driver->target(policy, target_freq, relation);
1928 1929 1930
	else if (cpufreq_driver->target_index) {
		struct cpufreq_frequency_table *freq_table;
		int index;
1931

1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944
		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;
		}

1945
		if (freq_table[index].frequency == policy->cur) {
1946
			retval = 0;
1947 1948 1949
			goto out;
		}

1950
		retval = __target_index(policy, freq_table, index);
1951 1952 1953
	}

out:
L
Linus Torvalds 已提交
1954 1955 1956 1957 1958 1959 1960 1961
	return retval;
}
EXPORT_SYMBOL_GPL(__cpufreq_driver_target);

int cpufreq_driver_target(struct cpufreq_policy *policy,
			  unsigned int target_freq,
			  unsigned int relation)
{
1962
	int ret = -EINVAL;
L
Linus Torvalds 已提交
1963

1964
	down_write(&policy->rwsem);
L
Linus Torvalds 已提交
1965 1966 1967

	ret = __cpufreq_driver_target(policy, target_freq, relation);

1968
	up_write(&policy->rwsem);
L
Linus Torvalds 已提交
1969 1970 1971 1972 1973

	return ret;
}
EXPORT_SYMBOL_GPL(cpufreq_driver_target);

1974 1975
static int __cpufreq_governor(struct cpufreq_policy *policy,
					unsigned int event)
L
Linus Torvalds 已提交
1976
{
1977
	int ret;
1978 1979 1980 1981 1982 1983 1984 1985 1986 1987

	/* 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
1988

1989 1990 1991
	/* Don't start any governor operations if we are entering suspend */
	if (cpufreq_suspended)
		return 0;
1992 1993 1994 1995 1996 1997
	/*
	 * Governor might not be initiated here if ACPI _PPC changed
	 * notification happened, so check it.
	 */
	if (!policy->governor)
		return -EINVAL;
1998

1999 2000 2001
	if (policy->governor->max_transition_latency &&
	    policy->cpuinfo.transition_latency >
	    policy->governor->max_transition_latency) {
2002 2003 2004
		if (!gov)
			return -EINVAL;
		else {
2005 2006
			pr_warn("%s governor failed, too long transition latency of HW, fallback to %s governor\n",
				policy->governor->name, gov->name);
2007 2008
			policy->governor = gov;
		}
2009
	}
L
Linus Torvalds 已提交
2010

2011 2012 2013
	if (event == CPUFREQ_GOV_POLICY_INIT)
		if (!try_module_get(policy->governor->owner))
			return -EINVAL;
L
Linus Torvalds 已提交
2014

2015
	pr_debug("__cpufreq_governor for CPU %u, event %u\n",
2016
		 policy->cpu, event);
2017 2018

	mutex_lock(&cpufreq_governor_lock);
2019
	if ((policy->governor_enabled && event == CPUFREQ_GOV_START)
2020 2021
	    || (!policy->governor_enabled
	    && (event == CPUFREQ_GOV_LIMITS || event == CPUFREQ_GOV_STOP))) {
2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032
		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
2033 2034
	ret = policy->governor->governor(policy, event);

2035 2036 2037 2038 2039
	if (!ret) {
		if (event == CPUFREQ_GOV_POLICY_INIT)
			policy->governor->initialized++;
		else if (event == CPUFREQ_GOV_POLICY_EXIT)
			policy->governor->initialized--;
2040 2041 2042 2043 2044 2045 2046 2047
	} 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);
2048
	}
2049

2050 2051
	if (((event == CPUFREQ_GOV_POLICY_INIT) && ret) ||
			((event == CPUFREQ_GOV_POLICY_EXIT) && !ret))
L
Linus Torvalds 已提交
2052 2053 2054 2055 2056 2057 2058
		module_put(policy->governor->owner);

	return ret;
}

int cpufreq_register_governor(struct cpufreq_governor *governor)
{
2059
	int err;
L
Linus Torvalds 已提交
2060 2061 2062 2063

	if (!governor)
		return -EINVAL;

2064 2065 2066
	if (cpufreq_disabled())
		return -ENODEV;

2067
	mutex_lock(&cpufreq_governor_mutex);
2068

2069
	governor->initialized = 0;
2070
	err = -EBUSY;
2071
	if (!find_governor(governor->name)) {
2072 2073
		err = 0;
		list_add(&governor->governor_list, &cpufreq_governor_list);
L
Linus Torvalds 已提交
2074 2075
	}

2076
	mutex_unlock(&cpufreq_governor_mutex);
2077
	return err;
L
Linus Torvalds 已提交
2078 2079 2080 2081 2082
}
EXPORT_SYMBOL_GPL(cpufreq_register_governor);

void cpufreq_unregister_governor(struct cpufreq_governor *governor)
{
2083 2084
	int cpu;

L
Linus Torvalds 已提交
2085 2086 2087
	if (!governor)
		return;

2088 2089 2090
	if (cpufreq_disabled())
		return;

2091 2092 2093 2094 2095 2096 2097
	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");
	}

2098
	mutex_lock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
2099
	list_del(&governor->governor_list);
2100
	mutex_unlock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111
	return;
}
EXPORT_SYMBOL_GPL(cpufreq_unregister_governor);


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

/**
 * cpufreq_get_policy - get the current cpufreq_policy
2112 2113
 * @policy: struct cpufreq_policy into which the current cpufreq_policy
 *	is written
L
Linus Torvalds 已提交
2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126
 *
 * 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;

2127
	memcpy(policy, cpu_policy, sizeof(*policy));
L
Linus Torvalds 已提交
2128 2129 2130 2131 2132 2133

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

2134
/*
2135 2136
 * policy : current policy.
 * new_policy: policy to be set.
2137
 */
2138
static int cpufreq_set_policy(struct cpufreq_policy *policy,
2139
				struct cpufreq_policy *new_policy)
L
Linus Torvalds 已提交
2140
{
2141 2142
	struct cpufreq_governor *old_gov;
	int ret;
L
Linus Torvalds 已提交
2143

2144 2145
	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
2146

2147
	memcpy(&new_policy->cpuinfo, &policy->cpuinfo, sizeof(policy->cpuinfo));
L
Linus Torvalds 已提交
2148

2149 2150
	if (new_policy->min > policy->max || new_policy->max < policy->min)
		return -EINVAL;
2151

L
Linus Torvalds 已提交
2152
	/* verify the cpu speed can be set within this limit */
2153
	ret = cpufreq_driver->verify(new_policy);
L
Linus Torvalds 已提交
2154
	if (ret)
2155
		return ret;
L
Linus Torvalds 已提交
2156 2157

	/* adjust if necessary - all reasons */
2158
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2159
			CPUFREQ_ADJUST, new_policy);
L
Linus Torvalds 已提交
2160 2161

	/* adjust if necessary - hardware incompatibility*/
2162
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2163
			CPUFREQ_INCOMPATIBLE, new_policy);
L
Linus Torvalds 已提交
2164

2165 2166 2167 2168
	/*
	 * verify the cpu speed can be set within this limit, which might be
	 * different to the first one
	 */
2169
	ret = cpufreq_driver->verify(new_policy);
2170
	if (ret)
2171
		return ret;
L
Linus Torvalds 已提交
2172 2173

	/* notification of the new policy */
2174
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2175
			CPUFREQ_NOTIFY, new_policy);
L
Linus Torvalds 已提交
2176

2177 2178
	policy->min = new_policy->min;
	policy->max = new_policy->max;
L
Linus Torvalds 已提交
2179

2180
	pr_debug("new min and max freqs are %u - %u kHz\n",
2181
		 policy->min, policy->max);
L
Linus Torvalds 已提交
2182

2183
	if (cpufreq_driver->setpolicy) {
2184
		policy->policy = new_policy->policy;
2185
		pr_debug("setting range\n");
2186 2187
		return cpufreq_driver->setpolicy(new_policy);
	}
L
Linus Torvalds 已提交
2188

2189 2190
	if (new_policy->governor == policy->governor)
		goto out;
2191

2192 2193 2194 2195 2196 2197 2198 2199
	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);
2200
		__cpufreq_governor(policy, CPUFREQ_GOV_POLICY_EXIT);
2201
		down_write(&policy->rwsem);
L
Linus Torvalds 已提交
2202 2203
	}

2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227
	/* 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
2228 2229 2230 2231 2232 2233
}

/**
 *	cpufreq_update_policy - re-evaluate an existing cpufreq policy
 *	@cpu: CPU which shall be re-evaluated
 *
L
Lucas De Marchi 已提交
2234
 *	Useful for policy notifiers which have different necessities
L
Linus Torvalds 已提交
2235 2236 2237 2238
 *	at different times.
 */
int cpufreq_update_policy(unsigned int cpu)
{
2239 2240
	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
	struct cpufreq_policy new_policy;
2241
	int ret;
L
Linus Torvalds 已提交
2242

2243 2244
	if (!policy)
		return -ENODEV;
L
Linus Torvalds 已提交
2245

2246
	down_write(&policy->rwsem);
L
Linus Torvalds 已提交
2247

2248
	pr_debug("updating policy for CPU %u\n", cpu);
2249
	memcpy(&new_policy, policy, sizeof(*policy));
2250 2251 2252 2253
	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
2254

2255 2256 2257 2258
	/*
	 * BIOS might change freq behind our back
	 * -> ask driver for current freq and notify governors about a change
	 */
2259
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
2260
		new_policy.cur = cpufreq_driver->get(cpu);
2261 2262
		if (WARN_ON(!new_policy.cur)) {
			ret = -EIO;
2263
			goto unlock;
2264 2265
		}

2266
		if (!policy->cur) {
2267
			pr_debug("Driver did not initialize current freq\n");
2268
			policy->cur = new_policy.cur;
2269
		} else {
2270
			if (policy->cur != new_policy.cur && has_target())
2271
				cpufreq_out_of_sync(policy, new_policy.cur);
2272
		}
2273 2274
	}

2275
	ret = cpufreq_set_policy(policy, &new_policy);
L
Linus Torvalds 已提交
2276

2277
unlock:
2278
	up_write(&policy->rwsem);
2279

2280
	cpufreq_cpu_put(policy);
L
Linus Torvalds 已提交
2281 2282 2283 2284
	return ret;
}
EXPORT_SYMBOL(cpufreq_update_policy);

2285
static int cpufreq_cpu_callback(struct notifier_block *nfb,
2286 2287 2288
					unsigned long action, void *hcpu)
{
	unsigned int cpu = (unsigned long)hcpu;
2289
	struct device *dev;
2290

2291 2292
	dev = get_cpu_device(cpu);
	if (dev) {
2293
		switch (action & ~CPU_TASKS_FROZEN) {
2294
		case CPU_ONLINE:
2295
			__cpufreq_add_dev(dev, NULL);
2296
			break;
2297

2298
		case CPU_DOWN_PREPARE:
2299
			__cpufreq_remove_dev_prepare(dev, NULL);
2300 2301 2302
			break;

		case CPU_POST_DEAD:
2303
			__cpufreq_remove_dev_finish(dev, NULL);
2304
			break;
2305

2306
		case CPU_DOWN_FAILED:
2307
			__cpufreq_add_dev(dev, NULL);
2308 2309 2310 2311 2312 2313
			break;
		}
	}
	return NOTIFY_OK;
}

2314
static struct notifier_block __refdata cpufreq_cpu_notifier = {
2315
	.notifier_call = cpufreq_cpu_callback,
2316
};
L
Linus Torvalds 已提交
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 2360 2361 2362
/*********************************************************************
 *               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);

2363 2364
		pr_err("%s: Cannot %s BOOST\n",
		       __func__, state ? "enable" : "disable");
2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384
	}

	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
2385 2386 2387 2388 2389 2390 2391 2392 2393
/*********************************************************************
 *               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.
 *
2394
 * Registers a CPU Frequency driver to this core code. This code
L
Linus Torvalds 已提交
2395
 * returns zero on success, -EBUSY when another driver got here first
2396
 * (and isn't unregistered in the meantime).
L
Linus Torvalds 已提交
2397 2398
 *
 */
2399
int cpufreq_register_driver(struct cpufreq_driver *driver_data)
L
Linus Torvalds 已提交
2400 2401 2402 2403
{
	unsigned long flags;
	int ret;

2404 2405 2406
	if (cpufreq_disabled())
		return -ENODEV;

L
Linus Torvalds 已提交
2407
	if (!driver_data || !driver_data->verify || !driver_data->init ||
2408
	    !(driver_data->setpolicy || driver_data->target_index ||
2409 2410
		    driver_data->target) ||
	     (driver_data->setpolicy && (driver_data->target_index ||
2411 2412
		    driver_data->target)) ||
	     (!!driver_data->get_intermediate != !!driver_data->target_intermediate))
L
Linus Torvalds 已提交
2413 2414
		return -EINVAL;

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

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

2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438
	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",
2439
			       __func__);
2440 2441 2442 2443
			goto err_null_driver;
		}
	}

2444
	ret = subsys_interface_register(&cpufreq_interface);
2445
	if (ret)
2446
		goto err_boost_unreg;
L
Linus Torvalds 已提交
2447

2448 2449
	if (!(cpufreq_driver->flags & CPUFREQ_STICKY) &&
	    list_empty(&cpufreq_policy_list)) {
L
Linus Torvalds 已提交
2450
		/* if all ->init() calls failed, unregister */
2451 2452 2453
		pr_debug("%s: No CPU initialized for driver %s\n", __func__,
			 driver_data->name);
		goto err_if_unreg;
L
Linus Torvalds 已提交
2454 2455
	}

2456
	register_hotcpu_notifier(&cpufreq_cpu_notifier);
2457
	pr_debug("driver %s up and running\n", driver_data->name);
L
Linus Torvalds 已提交
2458

2459
	return 0;
2460 2461
err_if_unreg:
	subsys_interface_unregister(&cpufreq_interface);
2462 2463 2464
err_boost_unreg:
	if (cpufreq_boost_supported())
		cpufreq_sysfs_remove_file(&boost.attr);
2465
err_null_driver:
2466
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2467
	cpufreq_driver = NULL;
2468
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
D
Dave Jones 已提交
2469
	return ret;
L
Linus Torvalds 已提交
2470 2471 2472 2473 2474 2475
}
EXPORT_SYMBOL_GPL(cpufreq_register_driver);

/**
 * cpufreq_unregister_driver - unregister the current CPUFreq driver
 *
2476
 * Unregister the current CPUFreq driver. Only call this if you have
L
Linus Torvalds 已提交
2477 2478 2479 2480
 * 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.
 */
2481
int cpufreq_unregister_driver(struct cpufreq_driver *driver)
L
Linus Torvalds 已提交
2482 2483 2484
{
	unsigned long flags;

2485
	if (!cpufreq_driver || (driver != cpufreq_driver))
L
Linus Torvalds 已提交
2486 2487
		return -EINVAL;

2488
	pr_debug("unregistering driver %s\n", driver->name);
L
Linus Torvalds 已提交
2489

2490
	subsys_interface_unregister(&cpufreq_interface);
2491 2492 2493
	if (cpufreq_boost_supported())
		cpufreq_sysfs_remove_file(&boost.attr);

2494
	unregister_hotcpu_notifier(&cpufreq_cpu_notifier);
L
Linus Torvalds 已提交
2495

2496
	down_write(&cpufreq_rwsem);
2497
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2498

2499
	cpufreq_driver = NULL;
2500

2501
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2502
	up_write(&cpufreq_rwsem);
L
Linus Torvalds 已提交
2503 2504 2505 2506

	return 0;
}
EXPORT_SYMBOL_GPL(cpufreq_unregister_driver);
2507

2508 2509 2510 2511 2512 2513 2514 2515
/*
 * 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,
};

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
	register_syscore_ops(&cpufreq_syscore_ops);

2526 2527 2528
	return 0;
}
core_initcall(cpufreq_core_init);