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>

V
Viresh Kumar 已提交
34 35 36 37 38 39
/* Macros to iterate over lists */
/* Iterate over online CPUs policies */
static LIST_HEAD(cpufreq_policy_list);
#define for_each_policy(__policy)				\
	list_for_each_entry(__policy, &cpufreq_policy_list, policy_list)

40 41 42 43 44
/* Iterate over governors */
static LIST_HEAD(cpufreq_governor_list);
#define for_each_governor(__governor)				\
	list_for_each_entry(__governor, &cpufreq_governor_list, governor_list)

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

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

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

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

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

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

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

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

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

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

115 116 117 118 119 120 121 122 123
struct kobject *get_governor_parent_kobj(struct cpufreq_policy *policy)
{
	if (have_governor_per_policy())
		return &policy->kobj;
	else
		return cpufreq_global_kobject;
}
EXPORT_SYMBOL_GPL(get_governor_parent_kobj);

124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158
static inline u64 get_cpu_idle_time_jiffy(unsigned int cpu, u64 *wall)
{
	u64 idle_time;
	u64 cur_wall_time;
	u64 busy_time;

	cur_wall_time = jiffies64_to_cputime64(get_jiffies_64());

	busy_time = kcpustat_cpu(cpu).cpustat[CPUTIME_USER];
	busy_time += kcpustat_cpu(cpu).cpustat[CPUTIME_SYSTEM];
	busy_time += kcpustat_cpu(cpu).cpustat[CPUTIME_IRQ];
	busy_time += kcpustat_cpu(cpu).cpustat[CPUTIME_SOFTIRQ];
	busy_time += kcpustat_cpu(cpu).cpustat[CPUTIME_STEAL];
	busy_time += kcpustat_cpu(cpu).cpustat[CPUTIME_NICE];

	idle_time = cur_wall_time - busy_time;
	if (wall)
		*wall = cputime_to_usecs(cur_wall_time);

	return cputime_to_usecs(idle_time);
}

u64 get_cpu_idle_time(unsigned int cpu, u64 *wall, int io_busy)
{
	u64 idle_time = get_cpu_idle_time_us(cpu, io_busy ? wall : NULL);

	if (idle_time == -1ULL)
		return get_cpu_idle_time_jiffy(cpu, wall);
	else if (!io_busy)
		idle_time += get_cpu_iowait_time_us(cpu, wall);

	return idle_time;
}
EXPORT_SYMBOL_GPL(get_cpu_idle_time);

159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189
/*
 * This is a generic cpufreq init() routine which can be used by cpufreq
 * drivers of SMP systems. It will do following:
 * - validate & show freq table passed
 * - set policies transition latency
 * - policy->cpus with all possible CPUs
 */
int cpufreq_generic_init(struct cpufreq_policy *policy,
		struct cpufreq_frequency_table *table,
		unsigned int transition_latency)
{
	int ret;

	ret = cpufreq_table_validate_and_show(policy, table);
	if (ret) {
		pr_err("%s: invalid frequency table: %d\n", __func__, ret);
		return ret;
	}

	policy->cpuinfo.transition_latency = transition_latency;

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

	return 0;
}
EXPORT_SYMBOL_GPL(cpufreq_generic_init);

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

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

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

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

210
struct cpufreq_policy *cpufreq_cpu_get(unsigned int cpu)
L
Linus Torvalds 已提交
211
{
212
	struct cpufreq_policy *policy = NULL;
L
Linus Torvalds 已提交
213 214
	unsigned long flags;

215
	if (cpu >= nr_cpu_ids)
216 217 218 219
		return NULL;

	if (!down_read_trylock(&cpufreq_rwsem))
		return NULL;
L
Linus Torvalds 已提交
220 221

	/* get the cpufreq driver */
222
	read_lock_irqsave(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
223

224 225 226 227 228 229
	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
230

231
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
232

233
	if (!policy)
234
		up_read(&cpufreq_rwsem);
L
Linus Torvalds 已提交
235

236
	return policy;
237
}
L
Linus Torvalds 已提交
238 239
EXPORT_SYMBOL_GPL(cpufreq_cpu_get);

240
void cpufreq_cpu_put(struct cpufreq_policy *policy)
L
Linus Torvalds 已提交
241
{
242 243
	kobject_put(&policy->kobj);
	up_read(&cpufreq_rwsem);
L
Linus Torvalds 已提交
244 245 246 247 248 249 250 251 252 253 254 255
}
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
256
 * systems as each CPU might be scaled differently. So, use the arch
L
Linus Torvalds 已提交
257 258
 * per-CPU loops_per_jiffy value wherever possible.
 */
259
static void adjust_jiffies(unsigned long val, struct cpufreq_freqs *ci)
L
Linus Torvalds 已提交
260
{
261 262 263 264
#ifndef CONFIG_SMP
	static unsigned long l_p_j_ref;
	static unsigned int l_p_j_ref_freq;

L
Linus Torvalds 已提交
265 266 267 268 269 270
	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;
271 272
		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
273
	}
274
	if (val == CPUFREQ_POSTCHANGE && ci->old != ci->new) {
275 276
		loops_per_jiffy = cpufreq_scale(l_p_j_ref, l_p_j_ref_freq,
								ci->new);
277 278
		pr_debug("scaling loops_per_jiffy to %lu for frequency %u kHz\n",
			 loops_per_jiffy, ci->new);
L
Linus Torvalds 已提交
279 280
	}
#endif
281
}
L
Linus Torvalds 已提交
282

283
static void __cpufreq_notify_transition(struct cpufreq_policy *policy,
284
		struct cpufreq_freqs *freqs, unsigned int state)
L
Linus Torvalds 已提交
285 286 287
{
	BUG_ON(irqs_disabled());

288 289 290
	if (cpufreq_disabled())
		return;

291
	freqs->flags = cpufreq_driver->flags;
292
	pr_debug("notification %u of frequency transition to %u kHz\n",
293
		 state, freqs->new);
L
Linus Torvalds 已提交
294 295

	switch (state) {
296

L
Linus Torvalds 已提交
297
	case CPUFREQ_PRECHANGE:
298
		/* detect if the driver reported a value as "old frequency"
299 300
		 * which is not equal to what the cpufreq core thinks is
		 * "old frequency".
L
Linus Torvalds 已提交
301
		 */
302
		if (!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
303 304
			if ((policy) && (policy->cpu == freqs->cpu) &&
			    (policy->cur) && (policy->cur != freqs->old)) {
305 306
				pr_debug("Warning: CPU frequency is %u, cpufreq assumed %u kHz\n",
					 freqs->old, policy->cur);
307
				freqs->old = policy->cur;
L
Linus Torvalds 已提交
308 309
			}
		}
310
		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
311
				CPUFREQ_PRECHANGE, freqs);
L
Linus Torvalds 已提交
312 313
		adjust_jiffies(CPUFREQ_PRECHANGE, freqs);
		break;
314

L
Linus Torvalds 已提交
315 316
	case CPUFREQ_POSTCHANGE:
		adjust_jiffies(CPUFREQ_POSTCHANGE, freqs);
317 318
		pr_debug("FREQ: %lu - CPU: %lu\n",
			 (unsigned long)freqs->new, (unsigned long)freqs->cpu);
319
		trace_cpu_frequency(freqs->new, freqs->cpu);
320
		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
321
				CPUFREQ_POSTCHANGE, freqs);
322 323
		if (likely(policy) && likely(policy->cpu == freqs->cpu))
			policy->cur = freqs->new;
L
Linus Torvalds 已提交
324 325 326
		break;
	}
}
327

328 329 330 331 332 333 334 335
/**
 * 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.
 */
336
static void cpufreq_notify_transition(struct cpufreq_policy *policy,
337 338 339 340 341
		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
342

343
/* Do post notifications when there are chances that transition has failed */
344
static void cpufreq_notify_post_transition(struct cpufreq_policy *policy,
345 346 347 348 349 350 351 352 353 354 355
		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);
}

356 357 358
void cpufreq_freq_transition_begin(struct cpufreq_policy *policy,
		struct cpufreq_freqs *freqs)
{
359 360 361 362 363 364 365 366 367 368 369 370

	/*
	 * 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);

371 372 373 374 375 376 377 378 379 380 381
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;
382
	policy->transition_task = current;
383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398

	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;
399
	policy->transition_task = NULL;
400 401 402 403 404

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

L
Linus Torvalds 已提交
405 406 407 408

/*********************************************************************
 *                          SYSFS INTERFACE                          *
 *********************************************************************/
409
static ssize_t show_boost(struct kobject *kobj,
410 411 412 413 414 415 416 417 418 419 420 421 422 423 424
				 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)) {
425 426
		pr_err("%s: Cannot %s BOOST!\n",
		       __func__, enable ? "enable" : "disable");
427 428 429
		return -EINVAL;
	}

430 431
	pr_debug("%s: cpufreq BOOST %s\n",
		 __func__, enable ? "enabled" : "disabled");
432 433 434 435

	return count;
}
define_one_global_rw(boost);
L
Linus Torvalds 已提交
436

437
static struct cpufreq_governor *find_governor(const char *str_governor)
438 439 440
{
	struct cpufreq_governor *t;

441
	for_each_governor(t)
442
		if (!strncasecmp(str_governor, t->name, CPUFREQ_NAME_LEN))
443 444 445 446 447
			return t;

	return NULL;
}

L
Linus Torvalds 已提交
448 449 450
/**
 * cpufreq_parse_governor - parse a governor string
 */
D
Dave Jones 已提交
451
static int cpufreq_parse_governor(char *str_governor, unsigned int *policy,
L
Linus Torvalds 已提交
452 453
				struct cpufreq_governor **governor)
{
454
	int err = -EINVAL;
455 456

	if (!cpufreq_driver)
457 458
		goto out;

459
	if (cpufreq_driver->setpolicy) {
460
		if (!strncasecmp(str_governor, "performance", CPUFREQ_NAME_LEN)) {
L
Linus Torvalds 已提交
461
			*policy = CPUFREQ_POLICY_PERFORMANCE;
462
			err = 0;
463
		} else if (!strncasecmp(str_governor, "powersave",
464
						CPUFREQ_NAME_LEN)) {
L
Linus Torvalds 已提交
465
			*policy = CPUFREQ_POLICY_POWERSAVE;
466
			err = 0;
L
Linus Torvalds 已提交
467
		}
468
	} else {
L
Linus Torvalds 已提交
469
		struct cpufreq_governor *t;
470

471
		mutex_lock(&cpufreq_governor_mutex);
472

473
		t = find_governor(str_governor);
474

475
		if (t == NULL) {
476
			int ret;
477

478 479 480
			mutex_unlock(&cpufreq_governor_mutex);
			ret = request_module("cpufreq_%s", str_governor);
			mutex_lock(&cpufreq_governor_mutex);
481

482
			if (ret == 0)
483
				t = find_governor(str_governor);
484 485
		}

486 487 488
		if (t != NULL) {
			*governor = t;
			err = 0;
L
Linus Torvalds 已提交
489
		}
490

491
		mutex_unlock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
492
	}
493
out:
494
	return err;
L
Linus Torvalds 已提交
495 496 497
}

/**
498 499
 * cpufreq_per_cpu_attr_read() / show_##file_name() -
 * print out cpufreq information
L
Linus Torvalds 已提交
500 501 502 503 504
 *
 * Write out information from cpufreq_driver->policy[cpu]; object must be
 * "unsigned int".
 */

505 506
#define show_one(file_name, object)			\
static ssize_t show_##file_name				\
D
Dave Jones 已提交
507
(struct cpufreq_policy *policy, char *buf)		\
508
{							\
509
	return sprintf(buf, "%u\n", policy->object);	\
L
Linus Torvalds 已提交
510 511 512 513
}

show_one(cpuinfo_min_freq, cpuinfo.min_freq);
show_one(cpuinfo_max_freq, cpuinfo.max_freq);
514
show_one(cpuinfo_transition_latency, cpuinfo.transition_latency);
L
Linus Torvalds 已提交
515 516
show_one(scaling_min_freq, min);
show_one(scaling_max_freq, max);
517

518
static ssize_t show_scaling_cur_freq(struct cpufreq_policy *policy, char *buf)
519 520 521 522 523 524 525 526 527
{
	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
528

529
static int cpufreq_set_policy(struct cpufreq_policy *policy,
530
				struct cpufreq_policy *new_policy);
531

L
Linus Torvalds 已提交
532 533 534 535 536
/**
 * 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
537
(struct cpufreq_policy *policy, const char *buf, size_t count)		\
L
Linus Torvalds 已提交
538
{									\
539
	int ret, temp;							\
L
Linus Torvalds 已提交
540 541 542 543 544 545
	struct cpufreq_policy new_policy;				\
									\
	ret = cpufreq_get_policy(&new_policy, policy->cpu);		\
	if (ret)							\
		return -EINVAL;						\
									\
546
	ret = sscanf(buf, "%u", &new_policy.object);			\
L
Linus Torvalds 已提交
547 548 549
	if (ret != 1)							\
		return -EINVAL;						\
									\
550
	temp = new_policy.object;					\
551
	ret = cpufreq_set_policy(policy, &new_policy);		\
552 553
	if (!ret)							\
		policy->user_policy.object = temp;			\
L
Linus Torvalds 已提交
554 555 556 557
									\
	return ret ? ret : count;					\
}

558 559
store_one(scaling_min_freq, min);
store_one(scaling_max_freq, max);
L
Linus Torvalds 已提交
560 561 562 563

/**
 * show_cpuinfo_cur_freq - current CPU frequency as detected by hardware
 */
D
Dave Jones 已提交
564 565
static ssize_t show_cpuinfo_cur_freq(struct cpufreq_policy *policy,
					char *buf)
L
Linus Torvalds 已提交
566
{
567
	unsigned int cur_freq = __cpufreq_get(policy);
L
Linus Torvalds 已提交
568 569 570 571 572 573 574 575
	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
576
static ssize_t show_scaling_governor(struct cpufreq_policy *policy, char *buf)
L
Linus Torvalds 已提交
577
{
578
	if (policy->policy == CPUFREQ_POLICY_POWERSAVE)
L
Linus Torvalds 已提交
579 580 581 582
		return sprintf(buf, "powersave\n");
	else if (policy->policy == CPUFREQ_POLICY_PERFORMANCE)
		return sprintf(buf, "performance\n");
	else if (policy->governor)
583
		return scnprintf(buf, CPUFREQ_NAME_PLEN, "%s\n",
584
				policy->governor->name);
L
Linus Torvalds 已提交
585 586 587 588 589 590
	return -EINVAL;
}

/**
 * store_scaling_governor - store policy for the specified CPU
 */
D
Dave Jones 已提交
591 592
static ssize_t store_scaling_governor(struct cpufreq_policy *policy,
					const char *buf, size_t count)
L
Linus Torvalds 已提交
593
{
594
	int ret;
L
Linus Torvalds 已提交
595 596 597 598 599 600 601
	char	str_governor[16];
	struct cpufreq_policy new_policy;

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

602
	ret = sscanf(buf, "%15s", str_governor);
L
Linus Torvalds 已提交
603 604 605
	if (ret != 1)
		return -EINVAL;

606 607
	if (cpufreq_parse_governor(str_governor, &new_policy.policy,
						&new_policy.governor))
L
Linus Torvalds 已提交
608 609
		return -EINVAL;

610
	ret = cpufreq_set_policy(policy, &new_policy);
611 612 613 614

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

615 616 617 618
	if (ret)
		return ret;
	else
		return count;
L
Linus Torvalds 已提交
619 620 621 622 623
}

/**
 * show_scaling_driver - show the cpufreq driver currently loaded
 */
D
Dave Jones 已提交
624
static ssize_t show_scaling_driver(struct cpufreq_policy *policy, char *buf)
L
Linus Torvalds 已提交
625
{
626
	return scnprintf(buf, CPUFREQ_NAME_PLEN, "%s\n", cpufreq_driver->name);
L
Linus Torvalds 已提交
627 628 629 630 631
}

/**
 * show_scaling_available_governors - show the available CPUfreq governors
 */
D
Dave Jones 已提交
632 633
static ssize_t show_scaling_available_governors(struct cpufreq_policy *policy,
						char *buf)
L
Linus Torvalds 已提交
634 635 636 637
{
	ssize_t i = 0;
	struct cpufreq_governor *t;

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

643
	for_each_governor(t) {
644 645
		if (i >= (ssize_t) ((PAGE_SIZE / sizeof(char))
		    - (CPUFREQ_NAME_LEN + 2)))
L
Linus Torvalds 已提交
646
			goto out;
647
		i += scnprintf(&buf[i], CPUFREQ_NAME_PLEN, "%s ", t->name);
L
Linus Torvalds 已提交
648
	}
649
out:
L
Linus Torvalds 已提交
650 651 652
	i += sprintf(&buf[i], "\n");
	return i;
}
653

654
ssize_t cpufreq_show_cpus(const struct cpumask *mask, char *buf)
L
Linus Torvalds 已提交
655 656 657 658
{
	ssize_t i = 0;
	unsigned int cpu;

659
	for_each_cpu(cpu, mask) {
L
Linus Torvalds 已提交
660 661 662 663
		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))
664
			break;
L
Linus Torvalds 已提交
665 666 667 668
	}
	i += sprintf(&buf[i], "\n");
	return i;
}
669
EXPORT_SYMBOL_GPL(cpufreq_show_cpus);
L
Linus Torvalds 已提交
670

671 672 673 674 675 676
/**
 * 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)
{
677
	return cpufreq_show_cpus(policy->related_cpus, buf);
678 679 680 681 682 683 684
}

/**
 * show_affected_cpus - show the CPUs affected by each transition
 */
static ssize_t show_affected_cpus(struct cpufreq_policy *policy, char *buf)
{
685
	return cpufreq_show_cpus(policy->cpus, buf);
686 687
}

688
static ssize_t store_scaling_setspeed(struct cpufreq_policy *policy,
D
Dave Jones 已提交
689
					const char *buf, size_t count)
690 691 692 693
{
	unsigned int freq = 0;
	unsigned int ret;

694
	if (!policy->governor || !policy->governor->store_setspeed)
695 696 697 698 699 700 701 702 703 704 705 706 707
		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)
{
708
	if (!policy->governor || !policy->governor->show_setspeed)
709 710 711 712
		return sprintf(buf, "<unsupported>\n");

	return policy->governor->show_setspeed(policy, buf);
}
L
Linus Torvalds 已提交
713

714
/**
715
 * show_bios_limit - show the current cpufreq HW/BIOS limitation
716 717 718 719 720
 */
static ssize_t show_bios_limit(struct cpufreq_policy *policy, char *buf)
{
	unsigned int limit;
	int ret;
721 722
	if (cpufreq_driver->bios_limit) {
		ret = cpufreq_driver->bios_limit(policy->cpu, &limit);
723 724 725 726 727 728
		if (!ret)
			return sprintf(buf, "%u\n", limit);
	}
	return sprintf(buf, "%u\n", policy->cpuinfo.max_freq);
}

729 730 731 732 733 734 735 736 737 738 739 740 741 742
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
743

D
Dave Jones 已提交
744
static struct attribute *default_attrs[] = {
L
Linus Torvalds 已提交
745 746
	&cpuinfo_min_freq.attr,
	&cpuinfo_max_freq.attr,
747
	&cpuinfo_transition_latency.attr,
L
Linus Torvalds 已提交
748 749 750
	&scaling_min_freq.attr,
	&scaling_max_freq.attr,
	&affected_cpus.attr,
751
	&related_cpus.attr,
L
Linus Torvalds 已提交
752 753 754
	&scaling_governor.attr,
	&scaling_driver.attr,
	&scaling_available_governors.attr,
755
	&scaling_setspeed.attr,
L
Linus Torvalds 已提交
756 757 758
	NULL
};

759 760
#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
761

762
static ssize_t show(struct kobject *kobj, struct attribute *attr, char *buf)
L
Linus Torvalds 已提交
763
{
D
Dave Jones 已提交
764 765
	struct cpufreq_policy *policy = to_policy(kobj);
	struct freq_attr *fattr = to_attr(attr);
766
	ssize_t ret;
767 768

	if (!down_read_trylock(&cpufreq_rwsem))
769
		return -EINVAL;
770

771
	down_read(&policy->rwsem);
772

773 774 775 776 777
	if (fattr->show)
		ret = fattr->show(policy, buf);
	else
		ret = -EIO;

778
	up_read(&policy->rwsem);
779
	up_read(&cpufreq_rwsem);
780

L
Linus Torvalds 已提交
781 782 783
	return ret;
}

D
Dave Jones 已提交
784 785
static ssize_t store(struct kobject *kobj, struct attribute *attr,
		     const char *buf, size_t count)
L
Linus Torvalds 已提交
786
{
D
Dave Jones 已提交
787 788
	struct cpufreq_policy *policy = to_policy(kobj);
	struct freq_attr *fattr = to_attr(attr);
789
	ssize_t ret = -EINVAL;
790

791 792 793 794 795
	get_online_cpus();

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

796
	if (!down_read_trylock(&cpufreq_rwsem))
797
		goto unlock;
798

799
	down_write(&policy->rwsem);
800

801 802 803 804 805
	if (fattr->store)
		ret = fattr->store(policy, buf, count);
	else
		ret = -EIO;

806
	up_write(&policy->rwsem);
807 808

	up_read(&cpufreq_rwsem);
809 810 811
unlock:
	put_online_cpus();

L
Linus Torvalds 已提交
812 813 814
	return ret;
}

D
Dave Jones 已提交
815
static void cpufreq_sysfs_release(struct kobject *kobj)
L
Linus Torvalds 已提交
816
{
D
Dave Jones 已提交
817
	struct cpufreq_policy *policy = to_policy(kobj);
818
	pr_debug("last reference is dropped\n");
L
Linus Torvalds 已提交
819 820 821
	complete(&policy->kobj_unregister);
}

822
static const struct sysfs_ops sysfs_ops = {
L
Linus Torvalds 已提交
823 824 825 826 827 828 829 830 831 832
	.show	= show,
	.store	= store,
};

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

833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875
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);

876
/* symlink affected CPUs */
877
static int cpufreq_add_dev_symlink(struct cpufreq_policy *policy)
878 879 880 881 882
{
	unsigned int j;
	int ret = 0;

	for_each_cpu(j, policy->cpus) {
883
		struct device *cpu_dev;
884

885
		if (j == policy->cpu)
886 887
			continue;

888
		pr_debug("Adding link for CPU: %u\n", j);
889 890
		cpu_dev = get_cpu_device(j);
		ret = sysfs_create_link(&cpu_dev->kobj, &policy->kobj,
891
					"cpufreq");
892 893
		if (ret)
			break;
894 895 896 897
	}
	return ret;
}

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

	/* set up files for this cpu device */
905
	drv_attr = cpufreq_driver->attr;
V
Viresh Kumar 已提交
906
	while (drv_attr && *drv_attr) {
907 908
		ret = sysfs_create_file(&policy->kobj, &((*drv_attr)->attr));
		if (ret)
909
			return ret;
910 911
		drv_attr++;
	}
912
	if (cpufreq_driver->get) {
913 914
		ret = sysfs_create_file(&policy->kobj, &cpuinfo_cur_freq.attr);
		if (ret)
915
			return ret;
916
	}
917 918 919

	ret = sysfs_create_file(&policy->kobj, &scaling_cur_freq.attr);
	if (ret)
920
		return ret;
921

922
	if (cpufreq_driver->bios_limit) {
923 924
		ret = sysfs_create_file(&policy->kobj, &bios_limit.attr);
		if (ret)
925
			return ret;
926
	}
927

928
	return cpufreq_add_dev_symlink(policy);
929 930 931 932
}

static void cpufreq_init_policy(struct cpufreq_policy *policy)
{
933
	struct cpufreq_governor *gov = NULL;
934 935 936
	struct cpufreq_policy new_policy;
	int ret = 0;

937
	memcpy(&new_policy, policy, sizeof(*policy));
938

939
	/* Update governor of new_policy to the governor used before hotplug */
940
	gov = find_governor(per_cpu(cpufreq_cpu_governor, policy->cpu));
941 942 943 944 945 946 947 948
	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;

949 950
	/* Use the default policy if its valid. */
	if (cpufreq_driver->setpolicy)
951
		cpufreq_parse_governor(gov->name, &new_policy.policy, NULL);
952 953

	/* set default policy */
954
	ret = cpufreq_set_policy(policy, &new_policy);
955
	if (ret) {
956
		pr_debug("setting policy failed\n");
957 958
		if (cpufreq_driver->exit)
			cpufreq_driver->exit(policy);
959
	}
960 961
}

962
static int cpufreq_add_policy_cpu(struct cpufreq_policy *policy,
963
				  unsigned int cpu, struct device *dev)
964
{
965
	int ret = 0;
966 967
	unsigned long flags;

968
	if (has_target()) {
969 970 971 972 973 974
		ret = __cpufreq_governor(policy, CPUFREQ_GOV_STOP);
		if (ret) {
			pr_err("%s: Failed to stop governor\n", __func__);
			return ret;
		}
	}
975

976
	down_write(&policy->rwsem);
V
Viresh Kumar 已提交
977

978
	write_lock_irqsave(&cpufreq_driver_lock, flags);
V
Viresh Kumar 已提交
979

980 981
	cpumask_set_cpu(cpu, policy->cpus);
	per_cpu(cpufreq_cpu_data, cpu) = policy;
982
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
983

984
	up_write(&policy->rwsem);
V
Viresh Kumar 已提交
985

986
	if (has_target()) {
987 988 989 990 991
		ret = __cpufreq_governor(policy, CPUFREQ_GOV_START);
		if (!ret)
			ret = __cpufreq_governor(policy, CPUFREQ_GOV_LIMITS);

		if (ret) {
992 993 994
			pr_err("%s: Failed to start governor\n", __func__);
			return ret;
		}
995
	}
996

997
	return sysfs_create_link(&dev->kobj, &policy->kobj, "cpufreq");
998
}
L
Linus Torvalds 已提交
999

1000 1001 1002 1003 1004
static struct cpufreq_policy *cpufreq_policy_restore(unsigned int cpu)
{
	struct cpufreq_policy *policy;
	unsigned long flags;

1005
	read_lock_irqsave(&cpufreq_driver_lock, flags);
1006 1007 1008

	policy = per_cpu(cpufreq_cpu_data_fallback, cpu);

1009
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
1010

1011 1012
	if (policy)
		policy->governor = NULL;
1013

1014 1015 1016
	return policy;
}

1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030
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;

1031
	INIT_LIST_HEAD(&policy->policy_list);
1032
	init_rwsem(&policy->rwsem);
1033 1034
	spin_lock_init(&policy->transition_lock);
	init_waitqueue_head(&policy->transition_wait);
1035 1036
	init_completion(&policy->kobj_unregister);
	INIT_WORK(&policy->update, handle_update);
1037

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

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

	return NULL;
}

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

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

1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071
	down_read(&policy->rwsem);
	kobj = &policy->kobj;
	cmp = &policy->kobj_unregister;
	up_read(&policy->rwsem);
	kobject_put(kobj);

	/*
	 * We need to make sure that the underlying kobj is
	 * actually not referenced anymore by anybody before we
	 * proceed with unloading.
	 */
	pr_debug("waiting for dropping of refcount\n");
	wait_for_completion(cmp);
	pr_debug("wait complete\n");
}

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

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

1084
	if (WARN_ON(cpu == policy->cpu))
1085 1086 1087 1088 1089 1090 1091 1092
		return 0;

	/* Move kobject to the new policy->cpu */
	ret = kobject_move(&policy->kobj, &cpu_dev->kobj);
	if (ret) {
		pr_err("%s: Failed to move kobj: %d\n", __func__, ret);
		return ret;
	}
1093

1094
	down_write(&policy->rwsem);
1095
	policy->cpu = cpu;
1096
	up_write(&policy->rwsem);
1097

1098
	return 0;
1099 1100
}

1101
static int __cpufreq_add_dev(struct device *dev, struct subsys_interface *sif)
L
Linus Torvalds 已提交
1102
{
1103
	unsigned int j, cpu = dev->id;
1104
	int ret = -ENOMEM;
1105
	struct cpufreq_policy *policy;
L
Linus Torvalds 已提交
1106
	unsigned long flags;
1107
	bool recover_policy = cpufreq_suspended;
L
Linus Torvalds 已提交
1108

1109 1110 1111
	if (cpu_is_offline(cpu))
		return 0;

1112
	pr_debug("adding CPU %u\n", cpu);
L
Linus Torvalds 已提交
1113 1114 1115

	/* check whether a different CPU already registered this
	 * CPU because it is in the same boat. */
1116 1117
	policy = cpufreq_cpu_get_raw(cpu);
	if (unlikely(policy))
L
Linus Torvalds 已提交
1118
		return 0;
1119

1120 1121 1122
	if (!down_read_trylock(&cpufreq_rwsem))
		return 0;

1123
	/* Check if this cpu was hot-unplugged earlier and has siblings */
1124
	read_lock_irqsave(&cpufreq_driver_lock, flags);
V
Viresh Kumar 已提交
1125
	for_each_policy(policy) {
1126
		if (cpumask_test_cpu(cpu, policy->related_cpus)) {
1127
			read_unlock_irqrestore(&cpufreq_driver_lock, flags);
1128
			ret = cpufreq_add_policy_cpu(policy, cpu, dev);
1129 1130
			up_read(&cpufreq_rwsem);
			return ret;
V
Viresh Kumar 已提交
1131
		}
1132
	}
1133
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
1134

1135 1136 1137 1138
	/*
	 * Restore the saved policy when doing light-weight init and fall back
	 * to the full init if that fails.
	 */
1139
	policy = recover_policy ? cpufreq_policy_restore(cpu) : NULL;
1140
	if (!policy) {
1141
		recover_policy = false;
1142
		policy = cpufreq_policy_alloc();
1143 1144 1145
		if (!policy)
			goto nomem_out;
	}
1146 1147 1148 1149 1150 1151 1152

	/*
	 * 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().
	 */
1153 1154 1155
	if (recover_policy && cpu != policy->cpu)
		WARN_ON(update_policy_cpu(policy, cpu, dev));
	else
1156 1157
		policy->cpu = cpu;

1158
	cpumask_copy(policy->cpus, cpumask_of(cpu));
L
Linus Torvalds 已提交
1159 1160 1161 1162

	/* call driver. From then on the cpufreq must be able
	 * to accept all calls to ->verify and ->setpolicy for this CPU
	 */
1163
	ret = cpufreq_driver->init(policy);
L
Linus Torvalds 已提交
1164
	if (ret) {
1165
		pr_debug("initialization failed\n");
V
Viresh Kumar 已提交
1166
		goto err_set_policy_cpu;
L
Linus Torvalds 已提交
1167
	}
V
Viresh Kumar 已提交
1168

1169 1170
	down_write(&policy->rwsem);

1171 1172 1173 1174 1175 1176 1177 1178 1179
	/* 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);

1180
	if (!recover_policy) {
1181 1182
		policy->user_policy.min = policy->min;
		policy->user_policy.max = policy->max;
1183 1184 1185 1186 1187 1188 1189 1190 1191

		/* 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;
		}
1192 1193
	}

1194 1195 1196 1197 1198
	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);

1199
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
1200 1201 1202 1203 1204 1205 1206
		policy->cur = cpufreq_driver->get(policy->cpu);
		if (!policy->cur) {
			pr_err("%s: ->get() failed\n", __func__);
			goto err_get_freq;
		}
	}

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 1238 1239 1240 1241 1242 1243 1244 1245 1246
	/*
	 * 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);
		}
	}

1247 1248 1249
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
				     CPUFREQ_START, policy);

1250
	if (!recover_policy) {
1251
		ret = cpufreq_add_dev_interface(policy, dev);
1252 1253
		if (ret)
			goto err_out_unregister;
1254 1255
		blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
				CPUFREQ_CREATE_POLICY, policy);
1256
	}
1257

1258 1259 1260 1261
	write_lock_irqsave(&cpufreq_driver_lock, flags);
	list_add(&policy->policy_list, &cpufreq_policy_list);
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);

1262 1263
	cpufreq_init_policy(policy);

1264
	if (!recover_policy) {
1265 1266 1267
		policy->user_policy.policy = policy->policy;
		policy->user_policy.governor = policy->governor;
	}
1268
	up_write(&policy->rwsem);
1269

1270
	kobject_uevent(&policy->kobj, KOBJ_ADD);
1271

1272 1273
	up_read(&cpufreq_rwsem);

1274 1275 1276 1277
	/* Callback for handling stuff after policy is ready */
	if (cpufreq_driver->ready)
		cpufreq_driver->ready(policy);

1278
	pr_debug("initialization complete\n");
1279

L
Linus Torvalds 已提交
1280 1281 1282
	return 0;

err_out_unregister:
1283
err_get_freq:
1284
	write_lock_irqsave(&cpufreq_driver_lock, flags);
1285
	for_each_cpu(j, policy->cpus)
1286
		per_cpu(cpufreq_cpu_data, j) = NULL;
1287
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
1288

1289 1290 1291 1292 1293
	if (!recover_policy) {
		kobject_put(&policy->kobj);
		wait_for_completion(&policy->kobj_unregister);
	}
err_init_policy_kobj:
1294 1295
	up_write(&policy->rwsem);

1296 1297
	if (cpufreq_driver->exit)
		cpufreq_driver->exit(policy);
V
Viresh Kumar 已提交
1298
err_set_policy_cpu:
1299
	if (recover_policy) {
1300 1301
		/* Do not leave stale fallback data behind. */
		per_cpu(cpufreq_cpu_data_fallback, cpu) = NULL;
1302
		cpufreq_policy_put_kobj(policy);
1303
	}
1304
	cpufreq_policy_free(policy);
1305

L
Linus Torvalds 已提交
1306
nomem_out:
1307 1308
	up_read(&cpufreq_rwsem);

L
Linus Torvalds 已提交
1309 1310 1311
	return ret;
}

1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322
/**
 * 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)
{
1323
	return __cpufreq_add_dev(dev, sif);
1324 1325
}

1326
static int __cpufreq_remove_dev_prepare(struct device *dev,
1327
					struct subsys_interface *sif)
L
Linus Torvalds 已提交
1328
{
1329
	unsigned int cpu = dev->id, cpus;
1330
	int ret;
L
Linus Torvalds 已提交
1331
	unsigned long flags;
1332
	struct cpufreq_policy *policy;
L
Linus Torvalds 已提交
1333

1334
	pr_debug("%s: unregistering CPU %u\n", __func__, cpu);
L
Linus Torvalds 已提交
1335

1336
	write_lock_irqsave(&cpufreq_driver_lock, flags);
V
Viresh Kumar 已提交
1337

1338
	policy = per_cpu(cpufreq_cpu_data, cpu);
V
Viresh Kumar 已提交
1339

1340
	/* Save the policy somewhere when doing a light-weight tear-down */
1341
	if (cpufreq_suspended)
1342
		per_cpu(cpufreq_cpu_data_fallback, cpu) = policy;
1343

1344
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
1345

1346
	if (!policy) {
1347
		pr_debug("%s: No cpu_data found\n", __func__);
L
Linus Torvalds 已提交
1348 1349 1350
		return -EINVAL;
	}

1351
	if (has_target()) {
1352 1353 1354 1355 1356
		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
1357

1358
		strncpy(per_cpu(cpufreq_cpu_governor, cpu),
1359
			policy->governor->name, CPUFREQ_NAME_LEN);
1360
	}
L
Linus Torvalds 已提交
1361

1362
	down_read(&policy->rwsem);
1363
	cpus = cpumask_weight(policy->cpus);
1364
	up_read(&policy->rwsem);
1365

1366
	if (cpu != policy->cpu) {
1367
		sysfs_remove_link(&dev->kobj, "cpufreq");
1368
	} else if (cpus > 1) {
1369 1370 1371
		/* Nominate new CPU */
		int new_cpu = cpumask_any_but(policy->cpus, cpu);
		struct device *cpu_dev = get_cpu_device(new_cpu);
1372

1373 1374 1375 1376 1377 1378 1379 1380
		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
1381
		}
1382 1383 1384 1385

		if (!cpufreq_suspended)
			pr_debug("%s: policy Kobject moved to cpu: %d from: %d\n",
				 __func__, new_cpu, cpu);
1386
	} else if (cpufreq_driver->stop_cpu) {
1387
		cpufreq_driver->stop_cpu(policy);
L
Linus Torvalds 已提交
1388 1389
	}

1390 1391 1392 1393
	return 0;
}

static int __cpufreq_remove_dev_finish(struct device *dev,
1394
				       struct subsys_interface *sif)
1395 1396 1397 1398 1399 1400
{
	unsigned int cpu = dev->id, cpus;
	int ret;
	unsigned long flags;
	struct cpufreq_policy *policy;

1401
	write_lock_irqsave(&cpufreq_driver_lock, flags);
1402
	policy = per_cpu(cpufreq_cpu_data, cpu);
1403 1404
	per_cpu(cpufreq_cpu_data, cpu) = NULL;
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
1405 1406 1407 1408 1409 1410

	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 1460 1461

	return 0;
}

1462
/**
1463
 * cpufreq_remove_dev - remove a CPU device
1464 1465 1466
 *
 * Removes the cpufreq interface for a CPU device.
 */
1467
static int cpufreq_remove_dev(struct device *dev, struct subsys_interface *sif)
1468
{
1469
	unsigned int cpu = dev->id;
1470
	int ret;
1471 1472 1473 1474

	if (cpu_is_offline(cpu))
		return 0;

1475
	ret = __cpufreq_remove_dev_prepare(dev, sif);
1476 1477

	if (!ret)
1478
		ret = __cpufreq_remove_dev_finish(dev, sif);
1479 1480

	return ret;
1481 1482
}

1483
static void handle_update(struct work_struct *work)
L
Linus Torvalds 已提交
1484
{
1485 1486 1487
	struct cpufreq_policy *policy =
		container_of(work, struct cpufreq_policy, update);
	unsigned int cpu = policy->cpu;
1488
	pr_debug("handle_update for cpu %u called\n", cpu);
L
Linus Torvalds 已提交
1489 1490 1491 1492
	cpufreq_update_policy(cpu);
}

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

1506
	pr_debug("Warning: CPU frequency out of sync: cpufreq and timing core thinks of %u, is %u kHz\n",
1507
		 policy->cur, new_freq);
L
Linus Torvalds 已提交
1508

1509
	freqs.old = policy->cur;
L
Linus Torvalds 已提交
1510
	freqs.new = new_freq;
1511

1512 1513
	cpufreq_freq_transition_begin(policy, &freqs);
	cpufreq_freq_transition_end(policy, &freqs, 0);
L
Linus Torvalds 已提交
1514 1515
}

1516
/**
D
Dhaval Giani 已提交
1517
 * cpufreq_quick_get - get the CPU frequency (in kHz) from policy->cur
1518 1519 1520 1521 1522 1523 1524
 * @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)
{
1525
	struct cpufreq_policy *policy;
1526
	unsigned int ret_freq = 0;
1527

1528 1529
	if (cpufreq_driver && cpufreq_driver->setpolicy && cpufreq_driver->get)
		return cpufreq_driver->get(cpu);
1530 1531

	policy = cpufreq_cpu_get(cpu);
1532
	if (policy) {
1533
		ret_freq = policy->cur;
1534 1535 1536
		cpufreq_cpu_put(policy);
	}

D
Dave Jones 已提交
1537
	return ret_freq;
1538 1539 1540
}
EXPORT_SYMBOL(cpufreq_quick_get);

1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560
/**
 * 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);

1561
static unsigned int __cpufreq_get(struct cpufreq_policy *policy)
L
Linus Torvalds 已提交
1562
{
1563
	unsigned int ret_freq = 0;
1564

1565
	if (!cpufreq_driver->get)
D
Dave Jones 已提交
1566
		return ret_freq;
L
Linus Torvalds 已提交
1567

1568
	ret_freq = cpufreq_driver->get(policy->cpu);
L
Linus Torvalds 已提交
1569

1570
	if (ret_freq && policy->cur &&
1571
		!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
1572 1573 1574
		/* verify no discrepancy between actual and
					saved value exists */
		if (unlikely(ret_freq != policy->cur)) {
1575
			cpufreq_out_of_sync(policy, ret_freq);
L
Linus Torvalds 已提交
1576 1577 1578 1579
			schedule_work(&policy->update);
		}
	}

D
Dave Jones 已提交
1580
	return ret_freq;
1581
}
L
Linus Torvalds 已提交
1582

1583 1584 1585 1586 1587 1588 1589 1590
/**
 * 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)
{
1591
	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
1592 1593
	unsigned int ret_freq = 0;

1594 1595
	if (policy) {
		down_read(&policy->rwsem);
1596
		ret_freq = __cpufreq_get(policy);
1597
		up_read(&policy->rwsem);
1598

1599 1600
		cpufreq_cpu_put(policy);
	}
1601

D
Dave Jones 已提交
1602
	return ret_freq;
L
Linus Torvalds 已提交
1603 1604 1605
}
EXPORT_SYMBOL(cpufreq_get);

1606 1607 1608 1609 1610
static struct subsys_interface cpufreq_interface = {
	.name		= "cpufreq",
	.subsys		= &cpu_subsys,
	.add_dev	= cpufreq_add_dev,
	.remove_dev	= cpufreq_remove_dev,
1611 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
/*
 * 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);

1639
/**
1640
 * cpufreq_suspend() - Suspend CPUFreq governors
1641
 *
1642 1643 1644 1645
 * 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.
1646
 */
1647
void cpufreq_suspend(void)
1648
{
1649
	struct cpufreq_policy *policy;
1650

1651 1652
	if (!cpufreq_driver)
		return;
1653

1654
	if (!has_target())
1655
		goto suspend;
1656

1657 1658
	pr_debug("%s: Suspending Governors\n", __func__);

V
Viresh Kumar 已提交
1659
	for_each_policy(policy) {
1660 1661 1662 1663 1664 1665 1666
		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);
1667
	}
1668 1669 1670

suspend:
	cpufreq_suspended = true;
1671 1672
}

L
Linus Torvalds 已提交
1673
/**
1674
 * cpufreq_resume() - Resume CPUFreq governors
L
Linus Torvalds 已提交
1675
 *
1676 1677
 * Called during system wide Suspend/Hibernate cycle for resuming governors that
 * are suspended with cpufreq_suspend().
L
Linus Torvalds 已提交
1678
 */
1679
void cpufreq_resume(void)
L
Linus Torvalds 已提交
1680
{
1681
	struct cpufreq_policy *policy;
L
Linus Torvalds 已提交
1682

1683 1684
	if (!cpufreq_driver)
		return;
L
Linus Torvalds 已提交
1685

1686 1687
	cpufreq_suspended = false;

1688
	if (!has_target())
1689
		return;
L
Linus Torvalds 已提交
1690

1691
	pr_debug("%s: Resuming Governors\n", __func__);
L
Linus Torvalds 已提交
1692

V
Viresh Kumar 已提交
1693
	for_each_policy(policy) {
1694 1695 1696 1697
		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)
1698 1699 1700 1701
		    || __cpufreq_governor(policy, CPUFREQ_GOV_LIMITS))
			pr_err("%s: Failed to start governor for policy: %p\n",
				__func__, policy);
	}
1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712

	/*
	 * schedule call cpufreq_update_policy() for first-online CPU, as that
	 * wouldn't be hotplugged-out on suspend. It will verify that the
	 * current freq is in sync with what we believe it to be.
	 */
	policy = cpufreq_cpu_get_raw(cpumask_first(cpu_online_mask));
	if (WARN_ON(!policy))
		return;

	schedule_work(&policy->update);
1713
}
L
Linus Torvalds 已提交
1714

1715 1716 1717 1718 1719 1720 1721 1722
/**
 *	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)
{
1723 1724 1725 1726
	if (cpufreq_driver)
		return cpufreq_driver->name;

	return NULL;
1727 1728
}
EXPORT_SYMBOL_GPL(cpufreq_get_current_driver);
L
Linus Torvalds 已提交
1729

1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744
/**
 *	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
1745 1746 1747 1748 1749 1750 1751 1752 1753
/*********************************************************************
 *                     NOTIFIER LISTS INTERFACE                      *
 *********************************************************************/

/**
 *	cpufreq_register_notifier - register a driver with cpufreq
 *	@nb: notifier function to register
 *      @list: CPUFREQ_TRANSITION_NOTIFIER or CPUFREQ_POLICY_NOTIFIER
 *
1754
 *	Add a driver to one of two lists: either a list of drivers that
L
Linus Torvalds 已提交
1755 1756 1757 1758 1759
 *      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
1760
 *	blocking_notifier_chain_register.
L
Linus Torvalds 已提交
1761 1762 1763 1764 1765
 */
int cpufreq_register_notifier(struct notifier_block *nb, unsigned int list)
{
	int ret;

1766 1767 1768
	if (cpufreq_disabled())
		return -EINVAL;

1769 1770
	WARN_ON(!init_cpufreq_transition_notifier_list_called);

L
Linus Torvalds 已提交
1771 1772
	switch (list) {
	case CPUFREQ_TRANSITION_NOTIFIER:
1773
		ret = srcu_notifier_chain_register(
1774
				&cpufreq_transition_notifier_list, nb);
L
Linus Torvalds 已提交
1775 1776
		break;
	case CPUFREQ_POLICY_NOTIFIER:
1777 1778
		ret = blocking_notifier_chain_register(
				&cpufreq_policy_notifier_list, nb);
L
Linus Torvalds 已提交
1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790
		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
1791
 *	@list: CPUFREQ_TRANSITION_NOTIFIER or CPUFREQ_POLICY_NOTIFIER
L
Linus Torvalds 已提交
1792 1793 1794 1795
 *
 *	Remove a driver from the CPU frequency notifier list.
 *
 *	This function may sleep, and has the same return conditions as
1796
 *	blocking_notifier_chain_unregister.
L
Linus Torvalds 已提交
1797 1798 1799 1800 1801
 */
int cpufreq_unregister_notifier(struct notifier_block *nb, unsigned int list)
{
	int ret;

1802 1803 1804
	if (cpufreq_disabled())
		return -EINVAL;

L
Linus Torvalds 已提交
1805 1806
	switch (list) {
	case CPUFREQ_TRANSITION_NOTIFIER:
1807
		ret = srcu_notifier_chain_unregister(
1808
				&cpufreq_transition_notifier_list, nb);
L
Linus Torvalds 已提交
1809 1810
		break;
	case CPUFREQ_POLICY_NOTIFIER:
1811 1812
		ret = blocking_notifier_chain_unregister(
				&cpufreq_policy_notifier_list, nb);
L
Linus Torvalds 已提交
1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826
		break;
	default:
		ret = -EINVAL;
	}

	return ret;
}
EXPORT_SYMBOL(cpufreq_unregister_notifier);


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

1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852
/* 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;
}

1853 1854 1855
static int __target_index(struct cpufreq_policy *policy,
			  struct cpufreq_frequency_table *freq_table, int index)
{
1856 1857
	struct cpufreq_freqs freqs = {.old = policy->cur, .flags = 0};
	unsigned int intermediate_freq = 0;
1858 1859 1860 1861 1862
	int retval = -EINVAL;
	bool notify;

	notify = !(cpufreq_driver->flags & CPUFREQ_ASYNC_NOTIFICATION);
	if (notify) {
1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873
		/* 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;
		}
1874

1875
		freqs.new = freq_table[index].frequency;
1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886
		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);

1887
	if (notify) {
1888 1889
		cpufreq_freq_transition_end(policy, &freqs, retval);

1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903
		/*
		 * 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);
		}
	}

1904 1905 1906
	return retval;
}

L
Linus Torvalds 已提交
1907 1908 1909 1910
int __cpufreq_driver_target(struct cpufreq_policy *policy,
			    unsigned int target_freq,
			    unsigned int relation)
{
1911
	unsigned int old_target_freq = target_freq;
1912
	int retval = -EINVAL;
1913

1914 1915 1916
	if (cpufreq_disabled())
		return -ENODEV;

1917 1918 1919 1920 1921 1922 1923
	/* 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",
1924
		 policy->cpu, target_freq, relation, old_target_freq);
1925

1926 1927 1928 1929 1930 1931
	/*
	 * 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.
	 */
1932 1933 1934
	if (target_freq == policy->cur)
		return 0;

1935 1936 1937
	/* Save last value to restore later on errors */
	policy->restore_freq = policy->cur;

1938 1939
	if (cpufreq_driver->target)
		retval = cpufreq_driver->target(policy, target_freq, relation);
1940 1941 1942
	else if (cpufreq_driver->target_index) {
		struct cpufreq_frequency_table *freq_table;
		int index;
1943

1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956
		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;
		}

1957
		if (freq_table[index].frequency == policy->cur) {
1958
			retval = 0;
1959 1960 1961
			goto out;
		}

1962
		retval = __target_index(policy, freq_table, index);
1963 1964 1965
	}

out:
L
Linus Torvalds 已提交
1966 1967 1968 1969 1970 1971 1972 1973
	return retval;
}
EXPORT_SYMBOL_GPL(__cpufreq_driver_target);

int cpufreq_driver_target(struct cpufreq_policy *policy,
			  unsigned int target_freq,
			  unsigned int relation)
{
1974
	int ret = -EINVAL;
L
Linus Torvalds 已提交
1975

1976
	down_write(&policy->rwsem);
L
Linus Torvalds 已提交
1977 1978 1979

	ret = __cpufreq_driver_target(policy, target_freq, relation);

1980
	up_write(&policy->rwsem);
L
Linus Torvalds 已提交
1981 1982 1983 1984 1985

	return ret;
}
EXPORT_SYMBOL_GPL(cpufreq_driver_target);

1986 1987
static int __cpufreq_governor(struct cpufreq_policy *policy,
					unsigned int event)
L
Linus Torvalds 已提交
1988
{
1989
	int ret;
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999

	/* 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
2000

2001 2002 2003
	/* Don't start any governor operations if we are entering suspend */
	if (cpufreq_suspended)
		return 0;
2004 2005 2006 2007 2008 2009
	/*
	 * Governor might not be initiated here if ACPI _PPC changed
	 * notification happened, so check it.
	 */
	if (!policy->governor)
		return -EINVAL;
2010

2011 2012 2013
	if (policy->governor->max_transition_latency &&
	    policy->cpuinfo.transition_latency >
	    policy->governor->max_transition_latency) {
2014 2015 2016
		if (!gov)
			return -EINVAL;
		else {
2017 2018
			pr_warn("%s governor failed, too long transition latency of HW, fallback to %s governor\n",
				policy->governor->name, gov->name);
2019 2020
			policy->governor = gov;
		}
2021
	}
L
Linus Torvalds 已提交
2022

2023 2024 2025
	if (event == CPUFREQ_GOV_POLICY_INIT)
		if (!try_module_get(policy->governor->owner))
			return -EINVAL;
L
Linus Torvalds 已提交
2026

2027
	pr_debug("__cpufreq_governor for CPU %u, event %u\n",
2028
		 policy->cpu, event);
2029 2030

	mutex_lock(&cpufreq_governor_lock);
2031
	if ((policy->governor_enabled && event == CPUFREQ_GOV_START)
2032 2033
	    || (!policy->governor_enabled
	    && (event == CPUFREQ_GOV_LIMITS || event == CPUFREQ_GOV_STOP))) {
2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044
		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
2045 2046
	ret = policy->governor->governor(policy, event);

2047 2048 2049 2050 2051
	if (!ret) {
		if (event == CPUFREQ_GOV_POLICY_INIT)
			policy->governor->initialized++;
		else if (event == CPUFREQ_GOV_POLICY_EXIT)
			policy->governor->initialized--;
2052 2053 2054 2055 2056 2057 2058 2059
	} 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);
2060
	}
2061

2062 2063
	if (((event == CPUFREQ_GOV_POLICY_INIT) && ret) ||
			((event == CPUFREQ_GOV_POLICY_EXIT) && !ret))
L
Linus Torvalds 已提交
2064 2065 2066 2067 2068 2069 2070
		module_put(policy->governor->owner);

	return ret;
}

int cpufreq_register_governor(struct cpufreq_governor *governor)
{
2071
	int err;
L
Linus Torvalds 已提交
2072 2073 2074 2075

	if (!governor)
		return -EINVAL;

2076 2077 2078
	if (cpufreq_disabled())
		return -ENODEV;

2079
	mutex_lock(&cpufreq_governor_mutex);
2080

2081
	governor->initialized = 0;
2082
	err = -EBUSY;
2083
	if (!find_governor(governor->name)) {
2084 2085
		err = 0;
		list_add(&governor->governor_list, &cpufreq_governor_list);
L
Linus Torvalds 已提交
2086 2087
	}

2088
	mutex_unlock(&cpufreq_governor_mutex);
2089
	return err;
L
Linus Torvalds 已提交
2090 2091 2092 2093 2094
}
EXPORT_SYMBOL_GPL(cpufreq_register_governor);

void cpufreq_unregister_governor(struct cpufreq_governor *governor)
{
2095 2096
	int cpu;

L
Linus Torvalds 已提交
2097 2098 2099
	if (!governor)
		return;

2100 2101 2102
	if (cpufreq_disabled())
		return;

2103 2104 2105 2106 2107 2108 2109
	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");
	}

2110
	mutex_lock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
2111
	list_del(&governor->governor_list);
2112
	mutex_unlock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123
	return;
}
EXPORT_SYMBOL_GPL(cpufreq_unregister_governor);


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

/**
 * cpufreq_get_policy - get the current cpufreq_policy
2124 2125
 * @policy: struct cpufreq_policy into which the current cpufreq_policy
 *	is written
L
Linus Torvalds 已提交
2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138
 *
 * 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;

2139
	memcpy(policy, cpu_policy, sizeof(*policy));
L
Linus Torvalds 已提交
2140 2141 2142 2143 2144 2145

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

2146
/*
2147 2148
 * policy : current policy.
 * new_policy: policy to be set.
2149
 */
2150
static int cpufreq_set_policy(struct cpufreq_policy *policy,
2151
				struct cpufreq_policy *new_policy)
L
Linus Torvalds 已提交
2152
{
2153 2154
	struct cpufreq_governor *old_gov;
	int ret;
L
Linus Torvalds 已提交
2155

2156 2157
	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
2158

2159
	memcpy(&new_policy->cpuinfo, &policy->cpuinfo, sizeof(policy->cpuinfo));
L
Linus Torvalds 已提交
2160

2161 2162
	if (new_policy->min > policy->max || new_policy->max < policy->min)
		return -EINVAL;
2163

L
Linus Torvalds 已提交
2164
	/* verify the cpu speed can be set within this limit */
2165
	ret = cpufreq_driver->verify(new_policy);
L
Linus Torvalds 已提交
2166
	if (ret)
2167
		return ret;
L
Linus Torvalds 已提交
2168 2169

	/* adjust if necessary - all reasons */
2170
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2171
			CPUFREQ_ADJUST, new_policy);
L
Linus Torvalds 已提交
2172 2173

	/* adjust if necessary - hardware incompatibility*/
2174
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2175
			CPUFREQ_INCOMPATIBLE, new_policy);
L
Linus Torvalds 已提交
2176

2177 2178 2179 2180
	/*
	 * verify the cpu speed can be set within this limit, which might be
	 * different to the first one
	 */
2181
	ret = cpufreq_driver->verify(new_policy);
2182
	if (ret)
2183
		return ret;
L
Linus Torvalds 已提交
2184 2185

	/* notification of the new policy */
2186
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2187
			CPUFREQ_NOTIFY, new_policy);
L
Linus Torvalds 已提交
2188

2189 2190
	policy->min = new_policy->min;
	policy->max = new_policy->max;
L
Linus Torvalds 已提交
2191

2192
	pr_debug("new min and max freqs are %u - %u kHz\n",
2193
		 policy->min, policy->max);
L
Linus Torvalds 已提交
2194

2195
	if (cpufreq_driver->setpolicy) {
2196
		policy->policy = new_policy->policy;
2197
		pr_debug("setting range\n");
2198 2199
		return cpufreq_driver->setpolicy(new_policy);
	}
L
Linus Torvalds 已提交
2200

2201 2202
	if (new_policy->governor == policy->governor)
		goto out;
2203

2204 2205 2206 2207 2208 2209 2210 2211
	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);
2212
		__cpufreq_governor(policy, CPUFREQ_GOV_POLICY_EXIT);
2213
		down_write(&policy->rwsem);
L
Linus Torvalds 已提交
2214 2215
	}

2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239
	/* 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
2240 2241 2242 2243 2244 2245
}

/**
 *	cpufreq_update_policy - re-evaluate an existing cpufreq policy
 *	@cpu: CPU which shall be re-evaluated
 *
L
Lucas De Marchi 已提交
2246
 *	Useful for policy notifiers which have different necessities
L
Linus Torvalds 已提交
2247 2248 2249 2250
 *	at different times.
 */
int cpufreq_update_policy(unsigned int cpu)
{
2251 2252
	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
	struct cpufreq_policy new_policy;
2253
	int ret;
L
Linus Torvalds 已提交
2254

2255 2256
	if (!policy)
		return -ENODEV;
L
Linus Torvalds 已提交
2257

2258
	down_write(&policy->rwsem);
L
Linus Torvalds 已提交
2259

2260
	pr_debug("updating policy for CPU %u\n", cpu);
2261
	memcpy(&new_policy, policy, sizeof(*policy));
2262 2263 2264 2265
	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
2266

2267 2268 2269 2270
	/*
	 * BIOS might change freq behind our back
	 * -> ask driver for current freq and notify governors about a change
	 */
2271
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
2272
		new_policy.cur = cpufreq_driver->get(cpu);
2273 2274
		if (WARN_ON(!new_policy.cur)) {
			ret = -EIO;
2275
			goto unlock;
2276 2277
		}

2278
		if (!policy->cur) {
2279
			pr_debug("Driver did not initialize current freq\n");
2280
			policy->cur = new_policy.cur;
2281
		} else {
2282
			if (policy->cur != new_policy.cur && has_target())
2283
				cpufreq_out_of_sync(policy, new_policy.cur);
2284
		}
2285 2286
	}

2287
	ret = cpufreq_set_policy(policy, &new_policy);
L
Linus Torvalds 已提交
2288

2289
unlock:
2290
	up_write(&policy->rwsem);
2291

2292
	cpufreq_cpu_put(policy);
L
Linus Torvalds 已提交
2293 2294 2295 2296
	return ret;
}
EXPORT_SYMBOL(cpufreq_update_policy);

2297
static int cpufreq_cpu_callback(struct notifier_block *nfb,
2298 2299 2300
					unsigned long action, void *hcpu)
{
	unsigned int cpu = (unsigned long)hcpu;
2301
	struct device *dev;
2302

2303 2304
	dev = get_cpu_device(cpu);
	if (dev) {
2305
		switch (action & ~CPU_TASKS_FROZEN) {
2306
		case CPU_ONLINE:
2307
			__cpufreq_add_dev(dev, NULL);
2308
			break;
2309

2310
		case CPU_DOWN_PREPARE:
2311
			__cpufreq_remove_dev_prepare(dev, NULL);
2312 2313 2314
			break;

		case CPU_POST_DEAD:
2315
			__cpufreq_remove_dev_finish(dev, NULL);
2316
			break;
2317

2318
		case CPU_DOWN_FAILED:
2319
			__cpufreq_add_dev(dev, NULL);
2320 2321 2322 2323 2324 2325
			break;
		}
	}
	return NOTIFY_OK;
}

2326
static struct notifier_block __refdata cpufreq_cpu_notifier = {
2327
	.notifier_call = cpufreq_cpu_callback,
2328
};
L
Linus Torvalds 已提交
2329

2330 2331 2332 2333 2334 2335 2336 2337 2338
/*********************************************************************
 *               BOOST						     *
 *********************************************************************/
static int cpufreq_boost_set_sw(int state)
{
	struct cpufreq_frequency_table *freq_table;
	struct cpufreq_policy *policy;
	int ret = -EINVAL;

V
Viresh Kumar 已提交
2339
	for_each_policy(policy) {
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 2373 2374
		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);

2375 2376
		pr_err("%s: Cannot %s BOOST\n",
		       __func__, state ? "enable" : "disable");
2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396
	}

	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
2397 2398 2399 2400 2401 2402 2403 2404 2405
/*********************************************************************
 *               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.
 *
2406
 * Registers a CPU Frequency driver to this core code. This code
L
Linus Torvalds 已提交
2407
 * returns zero on success, -EBUSY when another driver got here first
2408
 * (and isn't unregistered in the meantime).
L
Linus Torvalds 已提交
2409 2410
 *
 */
2411
int cpufreq_register_driver(struct cpufreq_driver *driver_data)
L
Linus Torvalds 已提交
2412 2413 2414 2415
{
	unsigned long flags;
	int ret;

2416 2417 2418
	if (cpufreq_disabled())
		return -ENODEV;

L
Linus Torvalds 已提交
2419
	if (!driver_data || !driver_data->verify || !driver_data->init ||
2420
	    !(driver_data->setpolicy || driver_data->target_index ||
2421 2422
		    driver_data->target) ||
	     (driver_data->setpolicy && (driver_data->target_index ||
2423 2424
		    driver_data->target)) ||
	     (!!driver_data->get_intermediate != !!driver_data->target_intermediate))
L
Linus Torvalds 已提交
2425 2426
		return -EINVAL;

2427
	pr_debug("trying to register driver %s\n", driver_data->name);
L
Linus Torvalds 已提交
2428

2429
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2430
	if (cpufreq_driver) {
2431
		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2432
		return -EEXIST;
L
Linus Torvalds 已提交
2433
	}
2434
	cpufreq_driver = driver_data;
2435
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
2436

2437 2438 2439
	if (driver_data->setpolicy)
		driver_data->flags |= CPUFREQ_CONST_LOOPS;

2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450
	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",
2451
			       __func__);
2452 2453 2454 2455
			goto err_null_driver;
		}
	}

2456
	ret = subsys_interface_register(&cpufreq_interface);
2457
	if (ret)
2458
		goto err_boost_unreg;
L
Linus Torvalds 已提交
2459

2460 2461
	if (!(cpufreq_driver->flags & CPUFREQ_STICKY) &&
	    list_empty(&cpufreq_policy_list)) {
L
Linus Torvalds 已提交
2462
		/* if all ->init() calls failed, unregister */
2463 2464 2465
		pr_debug("%s: No CPU initialized for driver %s\n", __func__,
			 driver_data->name);
		goto err_if_unreg;
L
Linus Torvalds 已提交
2466 2467
	}

2468
	register_hotcpu_notifier(&cpufreq_cpu_notifier);
2469
	pr_debug("driver %s up and running\n", driver_data->name);
L
Linus Torvalds 已提交
2470

2471
	return 0;
2472 2473
err_if_unreg:
	subsys_interface_unregister(&cpufreq_interface);
2474 2475 2476
err_boost_unreg:
	if (cpufreq_boost_supported())
		cpufreq_sysfs_remove_file(&boost.attr);
2477
err_null_driver:
2478
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2479
	cpufreq_driver = NULL;
2480
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
D
Dave Jones 已提交
2481
	return ret;
L
Linus Torvalds 已提交
2482 2483 2484 2485 2486 2487
}
EXPORT_SYMBOL_GPL(cpufreq_register_driver);

/**
 * cpufreq_unregister_driver - unregister the current CPUFreq driver
 *
2488
 * Unregister the current CPUFreq driver. Only call this if you have
L
Linus Torvalds 已提交
2489 2490 2491 2492
 * 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.
 */
2493
int cpufreq_unregister_driver(struct cpufreq_driver *driver)
L
Linus Torvalds 已提交
2494 2495 2496
{
	unsigned long flags;

2497
	if (!cpufreq_driver || (driver != cpufreq_driver))
L
Linus Torvalds 已提交
2498 2499
		return -EINVAL;

2500
	pr_debug("unregistering driver %s\n", driver->name);
L
Linus Torvalds 已提交
2501

2502
	subsys_interface_unregister(&cpufreq_interface);
2503 2504 2505
	if (cpufreq_boost_supported())
		cpufreq_sysfs_remove_file(&boost.attr);

2506
	unregister_hotcpu_notifier(&cpufreq_cpu_notifier);
L
Linus Torvalds 已提交
2507

2508
	down_write(&cpufreq_rwsem);
2509
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2510

2511
	cpufreq_driver = NULL;
2512

2513
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2514
	up_write(&cpufreq_rwsem);
L
Linus Torvalds 已提交
2515 2516 2517 2518

	return 0;
}
EXPORT_SYMBOL_GPL(cpufreq_unregister_driver);
2519

2520 2521 2522 2523 2524 2525 2526 2527
/*
 * 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,
};

2528 2529
static int __init cpufreq_core_init(void)
{
2530 2531 2532
	if (cpufreq_disabled())
		return -ENODEV;

2533
	cpufreq_global_kobject = kobject_create();
2534 2535
	BUG_ON(!cpufreq_global_kobject);

2536 2537
	register_syscore_ops(&cpufreq_syscore_ops);

2538 2539 2540
	return 0;
}
core_initcall(cpufreq_core_init);