cpufreq.c 65.9 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
static LIST_HEAD(cpufreq_policy_list);
35 36 37 38 39 40 41

static inline bool policy_is_inactive(struct cpufreq_policy *policy)
{
	return cpumask_empty(policy->cpus);
}

/* Macros to iterate over CPU policies */
42 43 44
#define for_each_suitable_policy(__policy, __active)			 \
	list_for_each_entry(__policy, &cpufreq_policy_list, policy_list) \
		if ((__active) == !policy_is_inactive(__policy))
45 46 47 48 49 50 51

#define for_each_active_policy(__policy)		\
	for_each_suitable_policy(__policy, true)
#define for_each_inactive_policy(__policy)		\
	for_each_suitable_policy(__policy, false)

#define for_each_policy(__policy)			\
V
Viresh Kumar 已提交
52 53
	list_for_each_entry(__policy, &cpufreq_policy_list, policy_list)

54 55 56 57 58
/* 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
59
/**
D
Dave Jones 已提交
60
 * The "cpufreq driver" - the arch- or hardware-dependent low
L
Linus Torvalds 已提交
61 62 63
 * level driver of CPUFreq support, and its spinlock. This lock
 * also protects the cpufreq_cpu_data array.
 */
64
static struct cpufreq_driver *cpufreq_driver;
65
static DEFINE_PER_CPU(struct cpufreq_policy *, cpufreq_cpu_data);
66 67
static DEFINE_RWLOCK(cpufreq_driver_lock);

68 69
/* Flag to suspend/resume CPUFreq governors */
static bool cpufreq_suspended;
L
Linus Torvalds 已提交
70

71 72 73 74 75
static inline bool has_target(void)
{
	return cpufreq_driver->target_index || cpufreq_driver->target;
}

L
Linus Torvalds 已提交
76
/* internal prototypes */
77
static unsigned int __cpufreq_get(struct cpufreq_policy *policy);
78 79
static int cpufreq_init_governor(struct cpufreq_policy *policy);
static void cpufreq_exit_governor(struct cpufreq_policy *policy);
80
static int cpufreq_start_governor(struct cpufreq_policy *policy);
81 82
static void cpufreq_stop_governor(struct cpufreq_policy *policy);
static void cpufreq_governor_limits(struct cpufreq_policy *policy);
83

L
Linus Torvalds 已提交
84
/**
85 86
 * 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
87 88 89 90
 * "transition" list for kernel code that needs to handle
 * changes to devices when the CPU clock speed changes.
 * The mutex locks both lists.
 */
91
static BLOCKING_NOTIFIER_HEAD(cpufreq_policy_notifier_list);
92
static struct srcu_notifier_head cpufreq_transition_notifier_list;
L
Linus Torvalds 已提交
93

94
static bool init_cpufreq_transition_notifier_list_called;
95 96 97
static int __init init_cpufreq_transition_notifier_list(void)
{
	srcu_init_notifier_head(&cpufreq_transition_notifier_list);
98
	init_cpufreq_transition_notifier_list_called = true;
99 100
	return 0;
}
101
pure_initcall(init_cpufreq_transition_notifier_list);
L
Linus Torvalds 已提交
102

103
static int off __read_mostly;
104
static int cpufreq_disabled(void)
105 106 107 108 109 110 111
{
	return off;
}
void disable_cpufreq(void)
{
	off = 1;
}
112
static DEFINE_MUTEX(cpufreq_governor_mutex);
L
Linus Torvalds 已提交
113

114 115
bool have_governor_per_policy(void)
{
116
	return !!(cpufreq_driver->flags & CPUFREQ_HAVE_GOVERNOR_PER_POLICY);
117
}
118
EXPORT_SYMBOL_GPL(have_governor_per_policy);
119

120 121 122 123 124 125 126 127 128
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);

129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163
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);

164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185
/*
 * 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;

	/*
186
	 * The driver only supports the SMP configuration where all processors
187 188 189 190 191 192 193 194
	 * share the clock and voltage and clock.
	 */
	cpumask_setall(policy->cpus);

	return 0;
}
EXPORT_SYMBOL_GPL(cpufreq_generic_init);

195
struct cpufreq_policy *cpufreq_cpu_get_raw(unsigned int cpu)
196 197 198
{
	struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);

199 200
	return policy && cpumask_test_cpu(cpu, policy->cpus) ? policy : NULL;
}
201
EXPORT_SYMBOL_GPL(cpufreq_cpu_get_raw);
202 203 204 205 206

unsigned int cpufreq_generic_get(unsigned int cpu)
{
	struct cpufreq_policy *policy = cpufreq_cpu_get_raw(cpu);

207
	if (!policy || IS_ERR(policy->clk)) {
208 209
		pr_err("%s: No %s associated to cpu: %d\n",
		       __func__, policy ? "clk" : "policy", cpu);
210 211 212 213 214 215 216
		return 0;
	}

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

217 218 219 220 221 222 223 224 225 226 227 228 229
/**
 * cpufreq_cpu_get: returns policy for a cpu and marks it busy.
 *
 * @cpu: cpu to find policy for.
 *
 * This returns policy for 'cpu', returns NULL if it doesn't exist.
 * It also increments the kobject reference count to mark it busy and so would
 * require a corresponding call to cpufreq_cpu_put() to decrement it back.
 * If corresponding call cpufreq_cpu_put() isn't made, the policy wouldn't be
 * freed as that depends on the kobj count.
 *
 * Return: A valid policy on success, otherwise NULL on failure.
 */
230
struct cpufreq_policy *cpufreq_cpu_get(unsigned int cpu)
L
Linus Torvalds 已提交
231
{
232
	struct cpufreq_policy *policy = NULL;
L
Linus Torvalds 已提交
233 234
	unsigned long flags;

235
	if (WARN_ON(cpu >= nr_cpu_ids))
236 237
		return NULL;

L
Linus Torvalds 已提交
238
	/* get the cpufreq driver */
239
	read_lock_irqsave(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
240

241 242
	if (cpufreq_driver) {
		/* get the CPU */
243
		policy = cpufreq_cpu_get_raw(cpu);
244 245 246
		if (policy)
			kobject_get(&policy->kobj);
	}
L
Linus Torvalds 已提交
247

248
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
249

250
	return policy;
251
}
L
Linus Torvalds 已提交
252 253
EXPORT_SYMBOL_GPL(cpufreq_cpu_get);

254 255 256 257 258 259 260 261
/**
 * cpufreq_cpu_put: Decrements the usage count of a policy
 *
 * @policy: policy earlier returned by cpufreq_cpu_get().
 *
 * This decrements the kobject reference count incremented earlier by calling
 * cpufreq_cpu_get().
 */
262
void cpufreq_cpu_put(struct cpufreq_policy *policy)
L
Linus Torvalds 已提交
263
{
264
	kobject_put(&policy->kobj);
L
Linus Torvalds 已提交
265 266 267 268 269 270 271 272 273 274 275 276
}
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
277
 * systems as each CPU might be scaled differently. So, use the arch
L
Linus Torvalds 已提交
278 279
 * per-CPU loops_per_jiffy value wherever possible.
 */
280
static void adjust_jiffies(unsigned long val, struct cpufreq_freqs *ci)
L
Linus Torvalds 已提交
281
{
282 283 284 285
#ifndef CONFIG_SMP
	static unsigned long l_p_j_ref;
	static unsigned int l_p_j_ref_freq;

L
Linus Torvalds 已提交
286 287 288 289 290 291
	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;
292 293
		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
294
	}
295
	if (val == CPUFREQ_POSTCHANGE && ci->old != ci->new) {
296 297
		loops_per_jiffy = cpufreq_scale(l_p_j_ref, l_p_j_ref_freq,
								ci->new);
298 299
		pr_debug("scaling loops_per_jiffy to %lu for frequency %u kHz\n",
			 loops_per_jiffy, ci->new);
L
Linus Torvalds 已提交
300 301
	}
#endif
302
}
L
Linus Torvalds 已提交
303

304
static void __cpufreq_notify_transition(struct cpufreq_policy *policy,
305
		struct cpufreq_freqs *freqs, unsigned int state)
L
Linus Torvalds 已提交
306 307 308
{
	BUG_ON(irqs_disabled());

309 310 311
	if (cpufreq_disabled())
		return;

312
	freqs->flags = cpufreq_driver->flags;
313
	pr_debug("notification %u of frequency transition to %u kHz\n",
314
		 state, freqs->new);
L
Linus Torvalds 已提交
315 316

	switch (state) {
317

L
Linus Torvalds 已提交
318
	case CPUFREQ_PRECHANGE:
319
		/* detect if the driver reported a value as "old frequency"
320 321
		 * which is not equal to what the cpufreq core thinks is
		 * "old frequency".
L
Linus Torvalds 已提交
322
		 */
323
		if (!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
324 325
			if ((policy) && (policy->cpu == freqs->cpu) &&
			    (policy->cur) && (policy->cur != freqs->old)) {
326 327
				pr_debug("Warning: CPU frequency is %u, cpufreq assumed %u kHz\n",
					 freqs->old, policy->cur);
328
				freqs->old = policy->cur;
L
Linus Torvalds 已提交
329 330
			}
		}
331
		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
332
				CPUFREQ_PRECHANGE, freqs);
L
Linus Torvalds 已提交
333 334
		adjust_jiffies(CPUFREQ_PRECHANGE, freqs);
		break;
335

L
Linus Torvalds 已提交
336 337
	case CPUFREQ_POSTCHANGE:
		adjust_jiffies(CPUFREQ_POSTCHANGE, freqs);
338 339
		pr_debug("FREQ: %lu - CPU: %lu\n",
			 (unsigned long)freqs->new, (unsigned long)freqs->cpu);
340
		trace_cpu_frequency(freqs->new, freqs->cpu);
341
		cpufreq_stats_record_transition(policy, freqs->new);
342
		srcu_notifier_call_chain(&cpufreq_transition_notifier_list,
343
				CPUFREQ_POSTCHANGE, freqs);
344 345
		if (likely(policy) && likely(policy->cpu == freqs->cpu))
			policy->cur = freqs->new;
L
Linus Torvalds 已提交
346 347 348
		break;
	}
}
349

350 351 352 353 354 355 356 357
/**
 * 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.
 */
358
static void cpufreq_notify_transition(struct cpufreq_policy *policy,
359 360 361 362 363
		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
364

365
/* Do post notifications when there are chances that transition has failed */
366
static void cpufreq_notify_post_transition(struct cpufreq_policy *policy,
367 368 369 370 371 372 373 374 375 376 377
		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);
}

378 379 380
void cpufreq_freq_transition_begin(struct cpufreq_policy *policy,
		struct cpufreq_freqs *freqs)
{
381 382 383 384 385 386 387 388 389 390 391 392

	/*
	 * 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);

393 394 395 396 397 398 399 400 401 402 403
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;
404
	policy->transition_task = current;
405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420

	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;
421
	policy->transition_task = NULL;
422 423 424 425 426

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

427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478
/*
 * Fast frequency switching status count.  Positive means "enabled", negative
 * means "disabled" and 0 means "not decided yet".
 */
static int cpufreq_fast_switch_count;
static DEFINE_MUTEX(cpufreq_fast_switch_lock);

static void cpufreq_list_transition_notifiers(void)
{
	struct notifier_block *nb;

	pr_info("Registered transition notifiers:\n");

	mutex_lock(&cpufreq_transition_notifier_list.mutex);

	for (nb = cpufreq_transition_notifier_list.head; nb; nb = nb->next)
		pr_info("%pF\n", nb->notifier_call);

	mutex_unlock(&cpufreq_transition_notifier_list.mutex);
}

/**
 * cpufreq_enable_fast_switch - Enable fast frequency switching for policy.
 * @policy: cpufreq policy to enable fast frequency switching for.
 *
 * Try to enable fast frequency switching for @policy.
 *
 * The attempt will fail if there is at least one transition notifier registered
 * at this point, as fast frequency switching is quite fundamentally at odds
 * with transition notifiers.  Thus if successful, it will make registration of
 * transition notifiers fail going forward.
 */
void cpufreq_enable_fast_switch(struct cpufreq_policy *policy)
{
	lockdep_assert_held(&policy->rwsem);

	if (!policy->fast_switch_possible)
		return;

	mutex_lock(&cpufreq_fast_switch_lock);
	if (cpufreq_fast_switch_count >= 0) {
		cpufreq_fast_switch_count++;
		policy->fast_switch_enabled = true;
	} else {
		pr_warn("CPU%u: Fast frequency switching not enabled\n",
			policy->cpu);
		cpufreq_list_transition_notifiers();
	}
	mutex_unlock(&cpufreq_fast_switch_lock);
}
EXPORT_SYMBOL_GPL(cpufreq_enable_fast_switch);

479 480 481 482 483
/**
 * cpufreq_disable_fast_switch - Disable fast frequency switching for policy.
 * @policy: cpufreq policy to disable fast frequency switching for.
 */
void cpufreq_disable_fast_switch(struct cpufreq_policy *policy)
484 485 486 487 488 489 490 491 492
{
	mutex_lock(&cpufreq_fast_switch_lock);
	if (policy->fast_switch_enabled) {
		policy->fast_switch_enabled = false;
		if (!WARN_ON(cpufreq_fast_switch_count <= 0))
			cpufreq_fast_switch_count--;
	}
	mutex_unlock(&cpufreq_fast_switch_lock);
}
493
EXPORT_SYMBOL_GPL(cpufreq_disable_fast_switch);
L
Linus Torvalds 已提交
494 495 496 497

/*********************************************************************
 *                          SYSFS INTERFACE                          *
 *********************************************************************/
498
static ssize_t show_boost(struct kobject *kobj,
499 500 501 502 503 504 505 506 507 508 509 510 511 512 513
				 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)) {
514 515
		pr_err("%s: Cannot %s BOOST!\n",
		       __func__, enable ? "enable" : "disable");
516 517 518
		return -EINVAL;
	}

519 520
	pr_debug("%s: cpufreq BOOST %s\n",
		 __func__, enable ? "enabled" : "disabled");
521 522 523 524

	return count;
}
define_one_global_rw(boost);
L
Linus Torvalds 已提交
525

526
static struct cpufreq_governor *find_governor(const char *str_governor)
527 528 529
{
	struct cpufreq_governor *t;

530
	for_each_governor(t)
531
		if (!strncasecmp(str_governor, t->name, CPUFREQ_NAME_LEN))
532 533 534 535 536
			return t;

	return NULL;
}

L
Linus Torvalds 已提交
537 538 539
/**
 * cpufreq_parse_governor - parse a governor string
 */
D
Dave Jones 已提交
540
static int cpufreq_parse_governor(char *str_governor, unsigned int *policy,
L
Linus Torvalds 已提交
541 542
				struct cpufreq_governor **governor)
{
543
	int err = -EINVAL;
544 545

	if (cpufreq_driver->setpolicy) {
546
		if (!strncasecmp(str_governor, "performance", CPUFREQ_NAME_LEN)) {
L
Linus Torvalds 已提交
547
			*policy = CPUFREQ_POLICY_PERFORMANCE;
548
			err = 0;
549
		} else if (!strncasecmp(str_governor, "powersave",
550
						CPUFREQ_NAME_LEN)) {
L
Linus Torvalds 已提交
551
			*policy = CPUFREQ_POLICY_POWERSAVE;
552
			err = 0;
L
Linus Torvalds 已提交
553
		}
554
	} else {
L
Linus Torvalds 已提交
555
		struct cpufreq_governor *t;
556

557
		mutex_lock(&cpufreq_governor_mutex);
558

559
		t = find_governor(str_governor);
560

561
		if (t == NULL) {
562
			int ret;
563

564 565 566
			mutex_unlock(&cpufreq_governor_mutex);
			ret = request_module("cpufreq_%s", str_governor);
			mutex_lock(&cpufreq_governor_mutex);
567

568
			if (ret == 0)
569
				t = find_governor(str_governor);
570 571
		}

572 573 574
		if (t != NULL) {
			*governor = t;
			err = 0;
L
Linus Torvalds 已提交
575
		}
576

577
		mutex_unlock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
578
	}
579
	return err;
L
Linus Torvalds 已提交
580 581 582
}

/**
583 584
 * cpufreq_per_cpu_attr_read() / show_##file_name() -
 * print out cpufreq information
L
Linus Torvalds 已提交
585 586 587 588 589
 *
 * Write out information from cpufreq_driver->policy[cpu]; object must be
 * "unsigned int".
 */

590 591
#define show_one(file_name, object)			\
static ssize_t show_##file_name				\
D
Dave Jones 已提交
592
(struct cpufreq_policy *policy, char *buf)		\
593
{							\
594
	return sprintf(buf, "%u\n", policy->object);	\
L
Linus Torvalds 已提交
595 596 597 598
}

show_one(cpuinfo_min_freq, cpuinfo.min_freq);
show_one(cpuinfo_max_freq, cpuinfo.max_freq);
599
show_one(cpuinfo_transition_latency, cpuinfo.transition_latency);
L
Linus Torvalds 已提交
600 601
show_one(scaling_min_freq, min);
show_one(scaling_max_freq, max);
602

603
static ssize_t show_scaling_cur_freq(struct cpufreq_policy *policy, char *buf)
604 605 606 607 608 609 610 611 612
{
	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
613

614
static int cpufreq_set_policy(struct cpufreq_policy *policy,
615
				struct cpufreq_policy *new_policy);
616

L
Linus Torvalds 已提交
617 618 619 620 621
/**
 * 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
622
(struct cpufreq_policy *policy, const char *buf, size_t count)		\
L
Linus Torvalds 已提交
623
{									\
624
	int ret, temp;							\
L
Linus Torvalds 已提交
625 626
	struct cpufreq_policy new_policy;				\
									\
627
	memcpy(&new_policy, policy, sizeof(*policy));			\
L
Linus Torvalds 已提交
628
									\
629
	ret = sscanf(buf, "%u", &new_policy.object);			\
L
Linus Torvalds 已提交
630 631 632
	if (ret != 1)							\
		return -EINVAL;						\
									\
633
	temp = new_policy.object;					\
634
	ret = cpufreq_set_policy(policy, &new_policy);		\
635 636
	if (!ret)							\
		policy->user_policy.object = temp;			\
L
Linus Torvalds 已提交
637 638 639 640
									\
	return ret ? ret : count;					\
}

641 642
store_one(scaling_min_freq, min);
store_one(scaling_max_freq, max);
L
Linus Torvalds 已提交
643 644 645 646

/**
 * show_cpuinfo_cur_freq - current CPU frequency as detected by hardware
 */
D
Dave Jones 已提交
647 648
static ssize_t show_cpuinfo_cur_freq(struct cpufreq_policy *policy,
					char *buf)
L
Linus Torvalds 已提交
649
{
650
	unsigned int cur_freq = __cpufreq_get(policy);
L
Linus Torvalds 已提交
651 652 653 654 655 656 657 658
	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
659
static ssize_t show_scaling_governor(struct cpufreq_policy *policy, char *buf)
L
Linus Torvalds 已提交
660
{
661
	if (policy->policy == CPUFREQ_POLICY_POWERSAVE)
L
Linus Torvalds 已提交
662 663 664 665
		return sprintf(buf, "powersave\n");
	else if (policy->policy == CPUFREQ_POLICY_PERFORMANCE)
		return sprintf(buf, "performance\n");
	else if (policy->governor)
666
		return scnprintf(buf, CPUFREQ_NAME_PLEN, "%s\n",
667
				policy->governor->name);
L
Linus Torvalds 已提交
668 669 670 671 672 673
	return -EINVAL;
}

/**
 * store_scaling_governor - store policy for the specified CPU
 */
D
Dave Jones 已提交
674 675
static ssize_t store_scaling_governor(struct cpufreq_policy *policy,
					const char *buf, size_t count)
L
Linus Torvalds 已提交
676
{
677
	int ret;
L
Linus Torvalds 已提交
678 679 680
	char	str_governor[16];
	struct cpufreq_policy new_policy;

681
	memcpy(&new_policy, policy, sizeof(*policy));
L
Linus Torvalds 已提交
682

683
	ret = sscanf(buf, "%15s", str_governor);
L
Linus Torvalds 已提交
684 685 686
	if (ret != 1)
		return -EINVAL;

687 688
	if (cpufreq_parse_governor(str_governor, &new_policy.policy,
						&new_policy.governor))
L
Linus Torvalds 已提交
689 690
		return -EINVAL;

691
	ret = cpufreq_set_policy(policy, &new_policy);
692
	return ret ? ret : count;
L
Linus Torvalds 已提交
693 694 695 696 697
}

/**
 * show_scaling_driver - show the cpufreq driver currently loaded
 */
D
Dave Jones 已提交
698
static ssize_t show_scaling_driver(struct cpufreq_policy *policy, char *buf)
L
Linus Torvalds 已提交
699
{
700
	return scnprintf(buf, CPUFREQ_NAME_PLEN, "%s\n", cpufreq_driver->name);
L
Linus Torvalds 已提交
701 702 703 704 705
}

/**
 * show_scaling_available_governors - show the available CPUfreq governors
 */
D
Dave Jones 已提交
706 707
static ssize_t show_scaling_available_governors(struct cpufreq_policy *policy,
						char *buf)
L
Linus Torvalds 已提交
708 709 710 711
{
	ssize_t i = 0;
	struct cpufreq_governor *t;

712
	if (!has_target()) {
L
Linus Torvalds 已提交
713 714 715 716
		i += sprintf(buf, "performance powersave");
		goto out;
	}

717
	for_each_governor(t) {
718 719
		if (i >= (ssize_t) ((PAGE_SIZE / sizeof(char))
		    - (CPUFREQ_NAME_LEN + 2)))
L
Linus Torvalds 已提交
720
			goto out;
721
		i += scnprintf(&buf[i], CPUFREQ_NAME_PLEN, "%s ", t->name);
L
Linus Torvalds 已提交
722
	}
723
out:
L
Linus Torvalds 已提交
724 725 726
	i += sprintf(&buf[i], "\n");
	return i;
}
727

728
ssize_t cpufreq_show_cpus(const struct cpumask *mask, char *buf)
L
Linus Torvalds 已提交
729 730 731 732
{
	ssize_t i = 0;
	unsigned int cpu;

733
	for_each_cpu(cpu, mask) {
L
Linus Torvalds 已提交
734 735 736 737
		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))
738
			break;
L
Linus Torvalds 已提交
739 740 741 742
	}
	i += sprintf(&buf[i], "\n");
	return i;
}
743
EXPORT_SYMBOL_GPL(cpufreq_show_cpus);
L
Linus Torvalds 已提交
744

745 746 747 748 749 750
/**
 * 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)
{
751
	return cpufreq_show_cpus(policy->related_cpus, buf);
752 753 754 755 756 757 758
}

/**
 * show_affected_cpus - show the CPUs affected by each transition
 */
static ssize_t show_affected_cpus(struct cpufreq_policy *policy, char *buf)
{
759
	return cpufreq_show_cpus(policy->cpus, buf);
760 761
}

762
static ssize_t store_scaling_setspeed(struct cpufreq_policy *policy,
D
Dave Jones 已提交
763
					const char *buf, size_t count)
764 765 766 767
{
	unsigned int freq = 0;
	unsigned int ret;

768
	if (!policy->governor || !policy->governor->store_setspeed)
769 770 771 772 773 774 775 776 777 778 779 780 781
		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)
{
782
	if (!policy->governor || !policy->governor->show_setspeed)
783 784 785 786
		return sprintf(buf, "<unsupported>\n");

	return policy->governor->show_setspeed(policy, buf);
}
L
Linus Torvalds 已提交
787

788
/**
789
 * show_bios_limit - show the current cpufreq HW/BIOS limitation
790 791 792 793 794
 */
static ssize_t show_bios_limit(struct cpufreq_policy *policy, char *buf)
{
	unsigned int limit;
	int ret;
795 796
	if (cpufreq_driver->bios_limit) {
		ret = cpufreq_driver->bios_limit(policy->cpu, &limit);
797 798 799 800 801 802
		if (!ret)
			return sprintf(buf, "%u\n", limit);
	}
	return sprintf(buf, "%u\n", policy->cpuinfo.max_freq);
}

803 804 805 806 807 808 809 810 811 812 813 814 815 816
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
817

D
Dave Jones 已提交
818
static struct attribute *default_attrs[] = {
L
Linus Torvalds 已提交
819 820
	&cpuinfo_min_freq.attr,
	&cpuinfo_max_freq.attr,
821
	&cpuinfo_transition_latency.attr,
L
Linus Torvalds 已提交
822 823 824
	&scaling_min_freq.attr,
	&scaling_max_freq.attr,
	&affected_cpus.attr,
825
	&related_cpus.attr,
L
Linus Torvalds 已提交
826 827 828
	&scaling_governor.attr,
	&scaling_driver.attr,
	&scaling_available_governors.attr,
829
	&scaling_setspeed.attr,
L
Linus Torvalds 已提交
830 831 832
	NULL
};

833 834
#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
835

836
static ssize_t show(struct kobject *kobj, struct attribute *attr, char *buf)
L
Linus Torvalds 已提交
837
{
D
Dave Jones 已提交
838 839
	struct cpufreq_policy *policy = to_policy(kobj);
	struct freq_attr *fattr = to_attr(attr);
840
	ssize_t ret;
841

842
	down_read(&policy->rwsem);
843
	ret = fattr->show(policy, buf);
844
	up_read(&policy->rwsem);
845

L
Linus Torvalds 已提交
846 847 848
	return ret;
}

D
Dave Jones 已提交
849 850
static ssize_t store(struct kobject *kobj, struct attribute *attr,
		     const char *buf, size_t count)
L
Linus Torvalds 已提交
851
{
D
Dave Jones 已提交
852 853
	struct cpufreq_policy *policy = to_policy(kobj);
	struct freq_attr *fattr = to_attr(attr);
854
	ssize_t ret = -EINVAL;
855

856 857
	get_online_cpus();

858 859
	if (cpu_online(policy->cpu)) {
		down_write(&policy->rwsem);
860
		ret = fattr->store(policy, buf, count);
861 862
		up_write(&policy->rwsem);
	}
863

864 865
	put_online_cpus();

L
Linus Torvalds 已提交
866 867 868
	return ret;
}

D
Dave Jones 已提交
869
static void cpufreq_sysfs_release(struct kobject *kobj)
L
Linus Torvalds 已提交
870
{
D
Dave Jones 已提交
871
	struct cpufreq_policy *policy = to_policy(kobj);
872
	pr_debug("last reference is dropped\n");
L
Linus Torvalds 已提交
873 874 875
	complete(&policy->kobj_unregister);
}

876
static const struct sysfs_ops sysfs_ops = {
L
Linus Torvalds 已提交
877 878 879 880 881 882 883 884 885 886
	.show	= show,
	.store	= store,
};

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

887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916
static int add_cpu_dev_symlink(struct cpufreq_policy *policy, int cpu)
{
	struct device *cpu_dev;

	pr_debug("%s: Adding symlink for CPU: %u\n", __func__, cpu);

	if (!policy)
		return 0;

	cpu_dev = get_cpu_device(cpu);
	if (WARN_ON(!cpu_dev))
		return 0;

	return sysfs_create_link(&cpu_dev->kobj, &policy->kobj, "cpufreq");
}

static void remove_cpu_dev_symlink(struct cpufreq_policy *policy, int cpu)
{
	struct device *cpu_dev;

	pr_debug("%s: Removing symlink for CPU: %u\n", __func__, cpu);

	cpu_dev = get_cpu_device(cpu);
	if (WARN_ON(!cpu_dev))
		return;

	sysfs_remove_link(&cpu_dev->kobj, "cpufreq");
}

/* Add/remove symlinks for all related CPUs */
917
static int cpufreq_add_dev_symlink(struct cpufreq_policy *policy)
918 919 920 921
{
	unsigned int j;
	int ret = 0;

922
	/* Some related CPUs might not be present (physically hotplugged) */
923
	for_each_cpu(j, policy->real_cpus) {
924
		ret = add_cpu_dev_symlink(policy, j);
925 926
		if (ret)
			break;
927
	}
928

929 930 931
	return ret;
}

932 933 934 935 936
static void cpufreq_remove_dev_symlink(struct cpufreq_policy *policy)
{
	unsigned int j;

	/* Some related CPUs might not be present (physically hotplugged) */
937
	for_each_cpu(j, policy->real_cpus)
938 939 940
		remove_cpu_dev_symlink(policy, j);
}

941
static int cpufreq_add_dev_interface(struct cpufreq_policy *policy)
942 943 944 945 946
{
	struct freq_attr **drv_attr;
	int ret = 0;

	/* set up files for this cpu device */
947
	drv_attr = cpufreq_driver->attr;
V
Viresh Kumar 已提交
948
	while (drv_attr && *drv_attr) {
949 950
		ret = sysfs_create_file(&policy->kobj, &((*drv_attr)->attr));
		if (ret)
951
			return ret;
952 953
		drv_attr++;
	}
954
	if (cpufreq_driver->get) {
955 956
		ret = sysfs_create_file(&policy->kobj, &cpuinfo_cur_freq.attr);
		if (ret)
957
			return ret;
958
	}
959 960 961

	ret = sysfs_create_file(&policy->kobj, &scaling_cur_freq.attr);
	if (ret)
962
		return ret;
963

964
	if (cpufreq_driver->bios_limit) {
965 966
		ret = sysfs_create_file(&policy->kobj, &bios_limit.attr);
		if (ret)
967
			return ret;
968
	}
969

970
	return cpufreq_add_dev_symlink(policy);
971 972
}

973 974 975 976 977
__weak struct cpufreq_governor *cpufreq_default_governor(void)
{
	return NULL;
}

978
static int cpufreq_init_policy(struct cpufreq_policy *policy)
979
{
980
	struct cpufreq_governor *gov = NULL;
981 982
	struct cpufreq_policy new_policy;

983
	memcpy(&new_policy, policy, sizeof(*policy));
984

985
	/* Update governor of new_policy to the governor used before hotplug */
986
	gov = find_governor(policy->last_governor);
987
	if (gov) {
988 989
		pr_debug("Restoring governor %s for cpu %d\n",
				policy->governor->name, policy->cpu);
990 991 992 993 994
	} else {
		gov = cpufreq_default_governor();
		if (!gov)
			return -ENODATA;
	}
995 996 997

	new_policy.governor = gov;

998 999 1000 1001 1002 1003 1004 1005
	/* Use the default policy if there is no last_policy. */
	if (cpufreq_driver->setpolicy) {
		if (policy->last_policy)
			new_policy.policy = policy->last_policy;
		else
			cpufreq_parse_governor(gov->name, &new_policy.policy,
					       NULL);
	}
1006
	/* set default policy */
1007
	return cpufreq_set_policy(policy, &new_policy);
1008 1009
}

1010
static int cpufreq_add_policy_cpu(struct cpufreq_policy *policy, unsigned int cpu)
1011
{
1012
	int ret = 0;
1013

1014 1015 1016 1017
	/* Has this CPU been taken care of already? */
	if (cpumask_test_cpu(cpu, policy->cpus))
		return 0;

1018
	down_write(&policy->rwsem);
1019 1020
	if (has_target())
		cpufreq_stop_governor(policy);
1021 1022

	cpumask_set_cpu(cpu, policy->cpus);
V
Viresh Kumar 已提交
1023

1024
	if (has_target()) {
1025
		ret = cpufreq_start_governor(policy);
1026
		if (ret)
1027
			pr_err("%s: Failed to start governor\n", __func__);
1028
	}
1029 1030
	up_write(&policy->rwsem);
	return ret;
1031
}
L
Linus Torvalds 已提交
1032

1033 1034 1035 1036 1037 1038 1039
static void handle_update(struct work_struct *work)
{
	struct cpufreq_policy *policy =
		container_of(work, struct cpufreq_policy, update);
	unsigned int cpu = policy->cpu;
	pr_debug("handle_update for cpu %u called\n", cpu);
	cpufreq_update_policy(cpu);
1040
}
L
Linus Torvalds 已提交
1041

1042
static struct cpufreq_policy *cpufreq_policy_alloc(unsigned int cpu)
1043
{
1044
	struct device *dev = get_cpu_device(cpu);
1045
	struct cpufreq_policy *policy;
1046
	int ret;
1047

1048 1049 1050
	if (WARN_ON(!dev))
		return NULL;

1051 1052 1053 1054 1055 1056 1057 1058 1059 1060
	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;

1061 1062 1063
	if (!zalloc_cpumask_var(&policy->real_cpus, GFP_KERNEL))
		goto err_free_rcpumask;

1064 1065 1066 1067 1068 1069 1070
	ret = kobject_init_and_add(&policy->kobj, &ktype_cpufreq,
				   cpufreq_global_kobject, "policy%u", cpu);
	if (ret) {
		pr_err("%s: failed to init policy->kobj: %d\n", __func__, ret);
		goto err_free_real_cpus;
	}

1071
	INIT_LIST_HEAD(&policy->policy_list);
1072
	init_rwsem(&policy->rwsem);
1073 1074
	spin_lock_init(&policy->transition_lock);
	init_waitqueue_head(&policy->transition_wait);
1075 1076
	init_completion(&policy->kobj_unregister);
	INIT_WORK(&policy->update, handle_update);
1077

1078
	policy->cpu = cpu;
1079 1080
	return policy;

1081 1082
err_free_real_cpus:
	free_cpumask_var(policy->real_cpus);
1083 1084
err_free_rcpumask:
	free_cpumask_var(policy->related_cpus);
1085 1086 1087 1088 1089 1090 1091 1092
err_free_cpumask:
	free_cpumask_var(policy->cpus);
err_free_policy:
	kfree(policy);

	return NULL;
}

1093
static void cpufreq_policy_put_kobj(struct cpufreq_policy *policy, bool notify)
1094 1095 1096 1097
{
	struct kobject *kobj;
	struct completion *cmp;

1098 1099 1100
	if (notify)
		blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
					     CPUFREQ_REMOVE_POLICY, policy);
1101

1102
	down_write(&policy->rwsem);
1103
	cpufreq_stats_free_table(policy);
1104
	cpufreq_remove_dev_symlink(policy);
1105 1106
	kobj = &policy->kobj;
	cmp = &policy->kobj_unregister;
1107
	up_write(&policy->rwsem);
1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119
	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");
}

1120
static void cpufreq_policy_free(struct cpufreq_policy *policy, bool notify)
1121
{
1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132
	unsigned long flags;
	int cpu;

	/* Remove policy from list */
	write_lock_irqsave(&cpufreq_driver_lock, flags);
	list_del(&policy->policy_list);

	for_each_cpu(cpu, policy->related_cpus)
		per_cpu(cpufreq_cpu_data, cpu) = NULL;
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);

1133
	cpufreq_policy_put_kobj(policy, notify);
1134
	free_cpumask_var(policy->real_cpus);
1135 1136 1137 1138 1139
	free_cpumask_var(policy->related_cpus);
	free_cpumask_var(policy->cpus);
	kfree(policy);
}

1140
static int cpufreq_online(unsigned int cpu)
L
Linus Torvalds 已提交
1141
{
1142
	struct cpufreq_policy *policy;
1143
	bool new_policy;
L
Linus Torvalds 已提交
1144
	unsigned long flags;
1145 1146
	unsigned int j;
	int ret;
1147

1148
	pr_debug("%s: bringing CPU%u online\n", __func__, cpu);
1149

1150
	/* Check if this CPU already has a policy to manage it */
1151
	policy = per_cpu(cpufreq_cpu_data, cpu);
1152
	if (policy) {
1153
		WARN_ON(!cpumask_test_cpu(cpu, policy->related_cpus));
1154
		if (!policy_is_inactive(policy))
1155
			return cpufreq_add_policy_cpu(policy, cpu);
L
Linus Torvalds 已提交
1156

1157
		/* This is the only online CPU for the policy.  Start over. */
1158
		new_policy = false;
1159 1160 1161 1162 1163
		down_write(&policy->rwsem);
		policy->cpu = cpu;
		policy->governor = NULL;
		up_write(&policy->rwsem);
	} else {
1164
		new_policy = true;
1165
		policy = cpufreq_policy_alloc(cpu);
1166
		if (!policy)
1167
			return -ENOMEM;
1168
	}
1169

1170
	cpumask_copy(policy->cpus, cpumask_of(cpu));
L
Linus Torvalds 已提交
1171 1172 1173 1174

	/* call driver. From then on the cpufreq must be able
	 * to accept all calls to ->verify and ->setpolicy for this CPU
	 */
1175
	ret = cpufreq_driver->init(policy);
L
Linus Torvalds 已提交
1176
	if (ret) {
1177
		pr_debug("initialization failed\n");
1178
		goto out_free_policy;
L
Linus Torvalds 已提交
1179
	}
V
Viresh Kumar 已提交
1180

1181 1182
	down_write(&policy->rwsem);

1183
	if (new_policy) {
1184
		/* related_cpus should at least include policy->cpus. */
1185
		cpumask_copy(policy->related_cpus, policy->cpus);
1186
		/* Remember CPUs present at the policy creation time. */
1187
		cpumask_and(policy->real_cpus, policy->cpus, cpu_present_mask);
1188
	}
1189

1190 1191 1192 1193 1194 1195
	/*
	 * affected cpus must always be the one, which are online. We aren't
	 * managing offline cpus here.
	 */
	cpumask_and(policy->cpus, policy->cpus, cpu_online_mask);

1196
	if (new_policy) {
1197 1198
		policy->user_policy.min = policy->min;
		policy->user_policy.max = policy->max;
1199

1200 1201 1202 1203 1204
		write_lock_irqsave(&cpufreq_driver_lock, flags);
		for_each_cpu(j, policy->related_cpus)
			per_cpu(cpufreq_cpu_data, j) = policy;
		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
	}
1205

1206
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
1207 1208 1209
		policy->cur = cpufreq_driver->get(policy->cpu);
		if (!policy->cur) {
			pr_err("%s: ->get() failed\n", __func__);
1210
			goto out_exit_policy;
1211 1212 1213
		}
	}

1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253
	/*
	 * 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);
		}
	}

1254
	if (new_policy) {
1255
		ret = cpufreq_add_dev_interface(policy);
1256
		if (ret)
1257
			goto out_exit_policy;
1258 1259

		cpufreq_stats_create_table(policy);
1260 1261
		blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
				CPUFREQ_CREATE_POLICY, policy);
1262

1263 1264 1265 1266
		write_lock_irqsave(&cpufreq_driver_lock, flags);
		list_add(&policy->policy_list, &cpufreq_policy_list);
		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
	}
1267

1268 1269 1270
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
				     CPUFREQ_START, policy);

1271 1272 1273 1274
	ret = cpufreq_init_policy(policy);
	if (ret) {
		pr_err("%s: Failed to initialize policy for cpu: %d (%d)\n",
		       __func__, cpu, ret);
1275 1276 1277
		/* cpufreq_policy_free() will notify based on this */
		new_policy = false;
		goto out_exit_policy;
1278
	}
1279

1280
	up_write(&policy->rwsem);
1281

1282
	kobject_uevent(&policy->kobj, KOBJ_ADD);
1283 1284 1285 1286 1287

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

1288
	pr_debug("initialization complete\n");
1289

L
Linus Torvalds 已提交
1290 1291
	return 0;

1292
out_exit_policy:
1293 1294
	up_write(&policy->rwsem);

1295 1296
	if (cpufreq_driver->exit)
		cpufreq_driver->exit(policy);
1297
out_free_policy:
1298
	cpufreq_policy_free(policy, !new_policy);
L
Linus Torvalds 已提交
1299 1300 1301
	return ret;
}

1302 1303 1304 1305 1306 1307 1308
/**
 * cpufreq_add_dev - the cpufreq interface for a CPU device.
 * @dev: CPU device.
 * @sif: Subsystem interface structure pointer (not used)
 */
static int cpufreq_add_dev(struct device *dev, struct subsys_interface *sif)
{
1309
	struct cpufreq_policy *policy;
1310 1311 1312 1313
	unsigned cpu = dev->id;

	dev_dbg(dev, "%s: adding CPU%u\n", __func__, cpu);

1314 1315
	if (cpu_online(cpu))
		return cpufreq_online(cpu);
1316

1317 1318 1319 1320 1321 1322 1323 1324
	/*
	 * A hotplug notifier will follow and we will handle it as CPU online
	 * then.  For now, just create the sysfs link, unless there is no policy
	 * or the link is already present.
	 */
	policy = per_cpu(cpufreq_cpu_data, cpu);
	if (!policy || cpumask_test_and_set_cpu(cpu, policy->real_cpus))
		return 0;
1325

1326
	return add_cpu_dev_symlink(policy, cpu);
L
Linus Torvalds 已提交
1327 1328
}

1329
static void cpufreq_offline(unsigned int cpu)
L
Linus Torvalds 已提交
1330
{
1331
	struct cpufreq_policy *policy;
1332
	int ret;
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
	policy = cpufreq_cpu_get_raw(cpu);
1337
	if (!policy) {
1338
		pr_debug("%s: No cpu_data found\n", __func__);
1339
		return;
L
Linus Torvalds 已提交
1340 1341
	}

1342
	down_write(&policy->rwsem);
1343 1344
	if (has_target())
		cpufreq_stop_governor(policy);
L
Linus Torvalds 已提交
1345

1346
	cpumask_clear_cpu(cpu, policy->cpus);
1347

1348 1349 1350 1351
	if (policy_is_inactive(policy)) {
		if (has_target())
			strncpy(policy->last_governor, policy->governor->name,
				CPUFREQ_NAME_LEN);
1352 1353
		else
			policy->last_policy = policy->policy;
1354 1355 1356 1357
	} else if (cpu == policy->cpu) {
		/* Nominate new CPU */
		policy->cpu = cpumask_any(policy->cpus);
	}
1358

1359 1360 1361
	/* Start governor again for active policy */
	if (!policy_is_inactive(policy)) {
		if (has_target()) {
1362
			ret = cpufreq_start_governor(policy);
1363 1364 1365
			if (ret)
				pr_err("%s: Failed to start governor\n", __func__);
		}
1366

1367
		goto unlock;
1368 1369
	}

1370 1371
	if (cpufreq_driver->stop_cpu)
		cpufreq_driver->stop_cpu(policy);
1372

1373 1374
	if (has_target())
		cpufreq_exit_governor(policy);
L
Linus Torvalds 已提交
1375

1376 1377 1378 1379 1380
	/*
	 * 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.
	 */
1381
	if (cpufreq_driver->exit) {
1382
		cpufreq_driver->exit(policy);
1383 1384
		policy->freq_table = NULL;
	}
1385 1386 1387

unlock:
	up_write(&policy->rwsem);
L
Linus Torvalds 已提交
1388 1389
}

1390
/**
1391
 * cpufreq_remove_dev - remove a CPU device
1392 1393 1394
 *
 * Removes the cpufreq interface for a CPU device.
 */
1395
static void cpufreq_remove_dev(struct device *dev, struct subsys_interface *sif)
1396
{
1397
	unsigned int cpu = dev->id;
1398
	struct cpufreq_policy *policy = per_cpu(cpufreq_cpu_data, cpu);
1399

1400
	if (!policy)
1401
		return;
1402

1403 1404
	if (cpu_online(cpu))
		cpufreq_offline(cpu);
1405

1406
	cpumask_clear_cpu(cpu, policy->real_cpus);
1407
	remove_cpu_dev_symlink(policy, cpu);
1408

1409
	if (cpumask_empty(policy->real_cpus))
1410
		cpufreq_policy_free(policy, true);
1411 1412
}

L
Linus Torvalds 已提交
1413
/**
1414 1415
 *	cpufreq_out_of_sync - If actual and saved CPU frequency differs, we're
 *	in deep trouble.
1416
 *	@policy: policy managing CPUs
L
Linus Torvalds 已提交
1417 1418
 *	@new_freq: CPU frequency the CPU actually runs at
 *
1419 1420
 *	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
1421
 */
1422
static void cpufreq_out_of_sync(struct cpufreq_policy *policy,
1423
				unsigned int new_freq)
L
Linus Torvalds 已提交
1424 1425
{
	struct cpufreq_freqs freqs;
1426

1427
	pr_debug("Warning: CPU frequency out of sync: cpufreq and timing core thinks of %u, is %u kHz\n",
1428
		 policy->cur, new_freq);
L
Linus Torvalds 已提交
1429

1430
	freqs.old = policy->cur;
L
Linus Torvalds 已提交
1431
	freqs.new = new_freq;
1432

1433 1434
	cpufreq_freq_transition_begin(policy, &freqs);
	cpufreq_freq_transition_end(policy, &freqs, 0);
L
Linus Torvalds 已提交
1435 1436
}

1437
/**
D
Dhaval Giani 已提交
1438
 * cpufreq_quick_get - get the CPU frequency (in kHz) from policy->cur
1439 1440 1441 1442 1443 1444 1445
 * @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)
{
1446
	struct cpufreq_policy *policy;
1447
	unsigned int ret_freq = 0;
1448
	unsigned long flags;
1449

1450 1451 1452 1453 1454 1455 1456 1457 1458
	read_lock_irqsave(&cpufreq_driver_lock, flags);

	if (cpufreq_driver && cpufreq_driver->setpolicy && cpufreq_driver->get) {
		ret_freq = cpufreq_driver->get(cpu);
		read_unlock_irqrestore(&cpufreq_driver_lock, flags);
		return ret_freq;
	}

	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
1459 1460

	policy = cpufreq_cpu_get(cpu);
1461
	if (policy) {
1462
		ret_freq = policy->cur;
1463 1464 1465
		cpufreq_cpu_put(policy);
	}

D
Dave Jones 已提交
1466
	return ret_freq;
1467 1468 1469
}
EXPORT_SYMBOL(cpufreq_quick_get);

1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489
/**
 * 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);

1490
static unsigned int __cpufreq_get(struct cpufreq_policy *policy)
L
Linus Torvalds 已提交
1491
{
1492
	unsigned int ret_freq = 0;
1493

1494
	if (!cpufreq_driver->get)
D
Dave Jones 已提交
1495
		return ret_freq;
L
Linus Torvalds 已提交
1496

1497
	ret_freq = cpufreq_driver->get(policy->cpu);
L
Linus Torvalds 已提交
1498

1499 1500 1501 1502 1503 1504
	/*
	 * Updating inactive policies is invalid, so avoid doing that.  Also
	 * if fast frequency switching is used with the given policy, the check
	 * against policy->cur is pointless, so skip it in that case too.
	 */
	if (unlikely(policy_is_inactive(policy)) || policy->fast_switch_enabled)
1505 1506
		return ret_freq;

1507
	if (ret_freq && policy->cur &&
1508
		!(cpufreq_driver->flags & CPUFREQ_CONST_LOOPS)) {
1509 1510 1511
		/* verify no discrepancy between actual and
					saved value exists */
		if (unlikely(ret_freq != policy->cur)) {
1512
			cpufreq_out_of_sync(policy, ret_freq);
L
Linus Torvalds 已提交
1513 1514 1515 1516
			schedule_work(&policy->update);
		}
	}

D
Dave Jones 已提交
1517
	return ret_freq;
1518
}
L
Linus Torvalds 已提交
1519

1520 1521 1522 1523 1524 1525 1526 1527
/**
 * 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)
{
1528
	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
1529 1530
	unsigned int ret_freq = 0;

1531 1532
	if (policy) {
		down_read(&policy->rwsem);
1533
		ret_freq = __cpufreq_get(policy);
1534
		up_read(&policy->rwsem);
1535

1536 1537
		cpufreq_cpu_put(policy);
	}
1538

D
Dave Jones 已提交
1539
	return ret_freq;
L
Linus Torvalds 已提交
1540 1541 1542
}
EXPORT_SYMBOL(cpufreq_get);

1543 1544 1545 1546
static unsigned int cpufreq_update_current_freq(struct cpufreq_policy *policy)
{
	unsigned int new_freq;

1547 1548 1549
	if (cpufreq_suspended)
		return 0;

1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563
	new_freq = cpufreq_driver->get(policy->cpu);
	if (!new_freq)
		return 0;

	if (!policy->cur) {
		pr_debug("cpufreq: Driver did not initialize current freq\n");
		policy->cur = new_freq;
	} else if (policy->cur != new_freq && has_target()) {
		cpufreq_out_of_sync(policy, new_freq);
	}

	return new_freq;
}

1564 1565 1566 1567 1568
static struct subsys_interface cpufreq_interface = {
	.name		= "cpufreq",
	.subsys		= &cpu_subsys,
	.add_dev	= cpufreq_add_dev,
	.remove_dev	= cpufreq_remove_dev,
1569 1570
};

1571 1572 1573 1574 1575 1576 1577 1578 1579
/*
 * 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) {
1580 1581
		pr_debug("%s: suspend_freq not defined\n", __func__);
		return 0;
1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596
	}

	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);

1597
/**
1598
 * cpufreq_suspend() - Suspend CPUFreq governors
1599
 *
1600 1601 1602 1603
 * 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.
1604
 */
1605
void cpufreq_suspend(void)
1606
{
1607
	struct cpufreq_policy *policy;
1608

1609 1610
	if (!cpufreq_driver)
		return;
1611

1612
	if (!has_target() && !cpufreq_driver->suspend)
1613
		goto suspend;
1614

1615 1616
	pr_debug("%s: Suspending Governors\n", __func__);

1617
	for_each_active_policy(policy) {
1618 1619
		if (has_target()) {
			down_write(&policy->rwsem);
1620
			cpufreq_stop_governor(policy);
1621 1622 1623 1624
			up_write(&policy->rwsem);
		}

		if (cpufreq_driver->suspend && cpufreq_driver->suspend(policy))
1625 1626
			pr_err("%s: Failed to suspend driver: %p\n", __func__,
				policy);
1627
	}
1628 1629 1630

suspend:
	cpufreq_suspended = true;
1631 1632
}

L
Linus Torvalds 已提交
1633
/**
1634
 * cpufreq_resume() - Resume CPUFreq governors
L
Linus Torvalds 已提交
1635
 *
1636 1637
 * Called during system wide Suspend/Hibernate cycle for resuming governors that
 * are suspended with cpufreq_suspend().
L
Linus Torvalds 已提交
1638
 */
1639
void cpufreq_resume(void)
L
Linus Torvalds 已提交
1640
{
1641
	struct cpufreq_policy *policy;
1642
	int ret;
L
Linus Torvalds 已提交
1643

1644 1645
	if (!cpufreq_driver)
		return;
L
Linus Torvalds 已提交
1646

1647 1648
	cpufreq_suspended = false;

1649
	if (!has_target() && !cpufreq_driver->resume)
1650
		return;
L
Linus Torvalds 已提交
1651

1652
	pr_debug("%s: Resuming Governors\n", __func__);
L
Linus Torvalds 已提交
1653

1654
	for_each_active_policy(policy) {
1655
		if (cpufreq_driver->resume && cpufreq_driver->resume(policy)) {
1656 1657
			pr_err("%s: Failed to resume driver: %p\n", __func__,
				policy);
1658
		} else if (has_target()) {
1659
			down_write(&policy->rwsem);
1660
			ret = cpufreq_start_governor(policy);
1661 1662 1663 1664 1665 1666
			up_write(&policy->rwsem);

			if (ret)
				pr_err("%s: Failed to start governor for policy: %p\n",
				       __func__, policy);
		}
1667 1668
	}
}
L
Linus Torvalds 已提交
1669

1670 1671 1672 1673 1674 1675 1676 1677
/**
 *	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)
{
1678 1679 1680 1681
	if (cpufreq_driver)
		return cpufreq_driver->name;

	return NULL;
1682 1683
}
EXPORT_SYMBOL_GPL(cpufreq_get_current_driver);
L
Linus Torvalds 已提交
1684

1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699
/**
 *	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
1700 1701 1702 1703 1704 1705 1706 1707 1708
/*********************************************************************
 *                     NOTIFIER LISTS INTERFACE                      *
 *********************************************************************/

/**
 *	cpufreq_register_notifier - register a driver with cpufreq
 *	@nb: notifier function to register
 *      @list: CPUFREQ_TRANSITION_NOTIFIER or CPUFREQ_POLICY_NOTIFIER
 *
1709
 *	Add a driver to one of two lists: either a list of drivers that
L
Linus Torvalds 已提交
1710 1711 1712 1713 1714
 *      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
1715
 *	blocking_notifier_chain_register.
L
Linus Torvalds 已提交
1716 1717 1718 1719 1720
 */
int cpufreq_register_notifier(struct notifier_block *nb, unsigned int list)
{
	int ret;

1721 1722 1723
	if (cpufreq_disabled())
		return -EINVAL;

1724 1725
	WARN_ON(!init_cpufreq_transition_notifier_list_called);

L
Linus Torvalds 已提交
1726 1727
	switch (list) {
	case CPUFREQ_TRANSITION_NOTIFIER:
1728 1729 1730 1731 1732 1733
		mutex_lock(&cpufreq_fast_switch_lock);

		if (cpufreq_fast_switch_count > 0) {
			mutex_unlock(&cpufreq_fast_switch_lock);
			return -EBUSY;
		}
1734
		ret = srcu_notifier_chain_register(
1735
				&cpufreq_transition_notifier_list, nb);
1736 1737 1738 1739
		if (!ret)
			cpufreq_fast_switch_count--;

		mutex_unlock(&cpufreq_fast_switch_lock);
L
Linus Torvalds 已提交
1740 1741
		break;
	case CPUFREQ_POLICY_NOTIFIER:
1742 1743
		ret = blocking_notifier_chain_register(
				&cpufreq_policy_notifier_list, nb);
L
Linus Torvalds 已提交
1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755
		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
1756
 *	@list: CPUFREQ_TRANSITION_NOTIFIER or CPUFREQ_POLICY_NOTIFIER
L
Linus Torvalds 已提交
1757 1758 1759 1760
 *
 *	Remove a driver from the CPU frequency notifier list.
 *
 *	This function may sleep, and has the same return conditions as
1761
 *	blocking_notifier_chain_unregister.
L
Linus Torvalds 已提交
1762 1763 1764 1765 1766
 */
int cpufreq_unregister_notifier(struct notifier_block *nb, unsigned int list)
{
	int ret;

1767 1768 1769
	if (cpufreq_disabled())
		return -EINVAL;

L
Linus Torvalds 已提交
1770 1771
	switch (list) {
	case CPUFREQ_TRANSITION_NOTIFIER:
1772 1773
		mutex_lock(&cpufreq_fast_switch_lock);

1774
		ret = srcu_notifier_chain_unregister(
1775
				&cpufreq_transition_notifier_list, nb);
1776 1777 1778 1779
		if (!ret && !WARN_ON(cpufreq_fast_switch_count >= 0))
			cpufreq_fast_switch_count++;

		mutex_unlock(&cpufreq_fast_switch_lock);
L
Linus Torvalds 已提交
1780 1781
		break;
	case CPUFREQ_POLICY_NOTIFIER:
1782 1783
		ret = blocking_notifier_chain_unregister(
				&cpufreq_policy_notifier_list, nb);
L
Linus Torvalds 已提交
1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797
		break;
	default:
		ret = -EINVAL;
	}

	return ret;
}
EXPORT_SYMBOL(cpufreq_unregister_notifier);


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

1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828
/**
 * cpufreq_driver_fast_switch - Carry out a fast CPU frequency switch.
 * @policy: cpufreq policy to switch the frequency for.
 * @target_freq: New frequency to set (may be approximate).
 *
 * Carry out a fast frequency switch without sleeping.
 *
 * The driver's ->fast_switch() callback invoked by this function must be
 * suitable for being called from within RCU-sched read-side critical sections
 * and it is expected to select the minimum available frequency greater than or
 * equal to @target_freq (CPUFREQ_RELATION_L).
 *
 * This function must not be called if policy->fast_switch_enabled is unset.
 *
 * Governors calling this function must guarantee that it will never be invoked
 * twice in parallel for the same policy and that it will never be called in
 * parallel with either ->target() or ->target_index() for the same policy.
 *
 * If CPUFREQ_ENTRY_INVALID is returned by the driver's ->fast_switch()
 * callback to indicate an error condition, the hardware configuration must be
 * preserved.
 */
unsigned int cpufreq_driver_fast_switch(struct cpufreq_policy *policy,
					unsigned int target_freq)
{
	clamp_val(target_freq, policy->min, policy->max);

	return cpufreq_driver->fast_switch(policy, target_freq);
}
EXPORT_SYMBOL_GPL(cpufreq_driver_fast_switch);

1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854
/* 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;
}

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

1863 1864 1865
	if (newfreq == policy->cur)
		return 0;

1866 1867
	notify = !(cpufreq_driver->flags & CPUFREQ_ASYNC_NOTIFICATION);
	if (notify) {
1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878
		/* 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;
		}
1879

1880
		freqs.new = newfreq;
1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891
		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);

1892
	if (notify) {
1893 1894
		cpufreq_freq_transition_end(policy, &freqs, retval);

1895 1896 1897 1898
		/*
		 * 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
1899
		 * case we haven't switched to intermediate freq at all.
1900 1901 1902 1903 1904 1905 1906 1907 1908
		 */
		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);
		}
	}

1909 1910 1911
	return retval;
}

L
Linus Torvalds 已提交
1912 1913 1914 1915
int __cpufreq_driver_target(struct cpufreq_policy *policy,
			    unsigned int target_freq,
			    unsigned int relation)
{
1916
	unsigned int old_target_freq = target_freq;
1917
	int index, retval;
1918

1919 1920 1921
	if (cpufreq_disabled())
		return -ENODEV;

1922
	/* Make sure that target_freq is within supported range */
1923
	target_freq = clamp_val(target_freq, policy->min, policy->max);
1924 1925

	pr_debug("target for CPU %u: %u kHz, relation %u, requested %u kHz\n",
1926
		 policy->cpu, target_freq, relation, old_target_freq);
1927

1928 1929 1930 1931 1932 1933
	/*
	 * 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.
	 */
1934 1935 1936
	if (target_freq == policy->cur)
		return 0;

1937 1938 1939
	/* Save last value to restore later on errors */
	policy->restore_freq = policy->cur;

1940
	if (cpufreq_driver->target)
1941
		return cpufreq_driver->target(policy, target_freq, relation);
1942

1943 1944
	if (!cpufreq_driver->target_index)
		return -EINVAL;
1945

1946 1947
	retval = cpufreq_frequency_table_target(policy, target_freq, relation,
						&index);
1948 1949 1950
	if (unlikely(retval)) {
		pr_err("%s: Unable to find matching freq\n", __func__);
		return retval;
1951 1952
	}

1953
	return __target_index(policy, index);
L
Linus Torvalds 已提交
1954 1955 1956 1957 1958 1959 1960
}
EXPORT_SYMBOL_GPL(__cpufreq_driver_target);

int cpufreq_driver_target(struct cpufreq_policy *policy,
			  unsigned int target_freq,
			  unsigned int relation)
{
1961
	int ret = -EINVAL;
L
Linus Torvalds 已提交
1962

1963
	down_write(&policy->rwsem);
L
Linus Torvalds 已提交
1964 1965 1966

	ret = __cpufreq_driver_target(policy, target_freq, relation);

1967
	up_write(&policy->rwsem);
L
Linus Torvalds 已提交
1968 1969 1970 1971 1972

	return ret;
}
EXPORT_SYMBOL_GPL(cpufreq_driver_target);

1973 1974 1975 1976 1977
__weak struct cpufreq_governor *cpufreq_fallback_governor(void)
{
	return NULL;
}

1978
static int cpufreq_init_governor(struct cpufreq_policy *policy)
L
Linus Torvalds 已提交
1979
{
1980
	int ret;
1981

1982 1983 1984
	/* Don't start any governor operations if we are entering suspend */
	if (cpufreq_suspended)
		return 0;
1985 1986 1987 1988 1989 1990
	/*
	 * Governor might not be initiated here if ACPI _PPC changed
	 * notification happened, so check it.
	 */
	if (!policy->governor)
		return -EINVAL;
1991

1992 1993 1994 1995
	if (policy->governor->max_transition_latency &&
	    policy->cpuinfo.transition_latency >
	    policy->governor->max_transition_latency) {
		struct cpufreq_governor *gov = cpufreq_fallback_governor();
L
Linus Torvalds 已提交
1996

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

2006 2007
	if (!try_module_get(policy->governor->owner))
		return -EINVAL;
2008

2009
	pr_debug("%s: for CPU %u\n", __func__, policy->cpu);
L
Linus Torvalds 已提交
2010

2011 2012 2013 2014 2015 2016
	if (policy->governor->init) {
		ret = policy->governor->init(policy);
		if (ret) {
			module_put(policy->governor->owner);
			return ret;
		}
2017
	}
L
Linus Torvalds 已提交
2018

2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
	return 0;
}

static void cpufreq_exit_governor(struct cpufreq_policy *policy)
{
	if (cpufreq_suspended || !policy->governor)
		return;

	pr_debug("%s: for CPU %u\n", __func__, policy->cpu);

2029 2030
	if (policy->governor->exit)
		policy->governor->exit(policy);
2031 2032

	module_put(policy->governor->owner);
L
Linus Torvalds 已提交
2033 2034
}

2035 2036 2037 2038
static int cpufreq_start_governor(struct cpufreq_policy *policy)
{
	int ret;

2039 2040 2041 2042 2043 2044 2045 2046
	if (cpufreq_suspended)
		return 0;

	if (!policy->governor)
		return -EINVAL;

	pr_debug("%s: for CPU %u\n", __func__, policy->cpu);

2047 2048 2049
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy)
		cpufreq_update_current_freq(policy);

2050 2051 2052 2053 2054 2055 2056 2057
	if (policy->governor->start) {
		ret = policy->governor->start(policy);
		if (ret)
			return ret;
	}

	if (policy->governor->limits)
		policy->governor->limits(policy);
2058 2059

	return 0;
2060 2061
}

2062 2063 2064 2065 2066 2067 2068
static void cpufreq_stop_governor(struct cpufreq_policy *policy)
{
	if (cpufreq_suspended || !policy->governor)
		return;

	pr_debug("%s: for CPU %u\n", __func__, policy->cpu);

2069 2070
	if (policy->governor->stop)
		policy->governor->stop(policy);
2071 2072 2073 2074 2075 2076 2077 2078 2079
}

static void cpufreq_governor_limits(struct cpufreq_policy *policy)
{
	if (cpufreq_suspended || !policy->governor)
		return;

	pr_debug("%s: for CPU %u\n", __func__, policy->cpu);

2080 2081
	if (policy->governor->limits)
		policy->governor->limits(policy);
2082 2083
}

L
Linus Torvalds 已提交
2084 2085
int cpufreq_register_governor(struct cpufreq_governor *governor)
{
2086
	int err;
L
Linus Torvalds 已提交
2087 2088 2089 2090

	if (!governor)
		return -EINVAL;

2091 2092 2093
	if (cpufreq_disabled())
		return -ENODEV;

2094
	mutex_lock(&cpufreq_governor_mutex);
2095

2096
	err = -EBUSY;
2097
	if (!find_governor(governor->name)) {
2098 2099
		err = 0;
		list_add(&governor->governor_list, &cpufreq_governor_list);
L
Linus Torvalds 已提交
2100 2101
	}

2102
	mutex_unlock(&cpufreq_governor_mutex);
2103
	return err;
L
Linus Torvalds 已提交
2104 2105 2106 2107 2108
}
EXPORT_SYMBOL_GPL(cpufreq_register_governor);

void cpufreq_unregister_governor(struct cpufreq_governor *governor)
{
2109 2110
	struct cpufreq_policy *policy;
	unsigned long flags;
2111

L
Linus Torvalds 已提交
2112 2113 2114
	if (!governor)
		return;

2115 2116 2117
	if (cpufreq_disabled())
		return;

2118 2119 2120
	/* clear last_governor for all inactive policies */
	read_lock_irqsave(&cpufreq_driver_lock, flags);
	for_each_inactive_policy(policy) {
2121 2122
		if (!strcmp(policy->last_governor, governor->name)) {
			policy->governor = NULL;
2123
			strcpy(policy->last_governor, "\0");
2124
		}
2125
	}
2126
	read_unlock_irqrestore(&cpufreq_driver_lock, flags);
2127

2128
	mutex_lock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
2129
	list_del(&governor->governor_list);
2130
	mutex_unlock(&cpufreq_governor_mutex);
L
Linus Torvalds 已提交
2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141
	return;
}
EXPORT_SYMBOL_GPL(cpufreq_unregister_governor);


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

/**
 * cpufreq_get_policy - get the current cpufreq_policy
2142 2143
 * @policy: struct cpufreq_policy into which the current cpufreq_policy
 *	is written
L
Linus Torvalds 已提交
2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156
 *
 * 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;

2157
	memcpy(policy, cpu_policy, sizeof(*policy));
L
Linus Torvalds 已提交
2158 2159 2160 2161 2162 2163

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

2164
/*
2165 2166
 * policy : current policy.
 * new_policy: policy to be set.
2167
 */
2168
static int cpufreq_set_policy(struct cpufreq_policy *policy,
2169
				struct cpufreq_policy *new_policy)
L
Linus Torvalds 已提交
2170
{
2171 2172
	struct cpufreq_governor *old_gov;
	int ret;
L
Linus Torvalds 已提交
2173

2174 2175
	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
2176

2177
	memcpy(&new_policy->cpuinfo, &policy->cpuinfo, sizeof(policy->cpuinfo));
L
Linus Torvalds 已提交
2178

2179 2180 2181 2182 2183
	/*
	* This check works well when we store new min/max freq attributes,
	* because new_policy is a copy of policy with one field updated.
	*/
	if (new_policy->min > new_policy->max)
2184
		return -EINVAL;
2185

L
Linus Torvalds 已提交
2186
	/* verify the cpu speed can be set within this limit */
2187
	ret = cpufreq_driver->verify(new_policy);
L
Linus Torvalds 已提交
2188
	if (ret)
2189
		return ret;
L
Linus Torvalds 已提交
2190 2191

	/* adjust if necessary - all reasons */
2192
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2193
			CPUFREQ_ADJUST, new_policy);
L
Linus Torvalds 已提交
2194

2195 2196 2197 2198
	/*
	 * verify the cpu speed can be set within this limit, which might be
	 * different to the first one
	 */
2199
	ret = cpufreq_driver->verify(new_policy);
2200
	if (ret)
2201
		return ret;
L
Linus Torvalds 已提交
2202 2203

	/* notification of the new policy */
2204
	blocking_notifier_call_chain(&cpufreq_policy_notifier_list,
2205
			CPUFREQ_NOTIFY, new_policy);
L
Linus Torvalds 已提交
2206

2207 2208
	policy->min = new_policy->min;
	policy->max = new_policy->max;
L
Linus Torvalds 已提交
2209

2210
	pr_debug("new min and max freqs are %u - %u kHz\n",
2211
		 policy->min, policy->max);
L
Linus Torvalds 已提交
2212

2213
	if (cpufreq_driver->setpolicy) {
2214
		policy->policy = new_policy->policy;
2215
		pr_debug("setting range\n");
2216 2217
		return cpufreq_driver->setpolicy(new_policy);
	}
L
Linus Torvalds 已提交
2218

2219 2220
	if (new_policy->governor == policy->governor) {
		pr_debug("cpufreq: governor limits update\n");
2221
		cpufreq_governor_limits(policy);
2222
		return 0;
2223
	}
2224

2225 2226 2227 2228 2229 2230
	pr_debug("governor switch\n");

	/* save old, working values */
	old_gov = policy->governor;
	/* end old governor */
	if (old_gov) {
2231
		cpufreq_stop_governor(policy);
2232
		cpufreq_exit_governor(policy);
L
Linus Torvalds 已提交
2233 2234
	}

2235 2236
	/* start new governor */
	policy->governor = new_policy->governor;
2237
	ret = cpufreq_init_governor(policy);
2238
	if (!ret) {
2239 2240 2241 2242 2243
		ret = cpufreq_start_governor(policy);
		if (!ret) {
			pr_debug("cpufreq: governor change\n");
			return 0;
		}
2244
		cpufreq_exit_governor(policy);
2245 2246 2247 2248 2249 2250
	}

	/* 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;
2251
		if (cpufreq_init_governor(policy))
2252 2253
			policy->governor = NULL;
		else
2254
			cpufreq_start_governor(policy);
2255 2256
	}

2257
	return ret;
L
Linus Torvalds 已提交
2258 2259 2260 2261 2262 2263
}

/**
 *	cpufreq_update_policy - re-evaluate an existing cpufreq policy
 *	@cpu: CPU which shall be re-evaluated
 *
L
Lucas De Marchi 已提交
2264
 *	Useful for policy notifiers which have different necessities
L
Linus Torvalds 已提交
2265 2266 2267 2268
 *	at different times.
 */
int cpufreq_update_policy(unsigned int cpu)
{
2269 2270
	struct cpufreq_policy *policy = cpufreq_cpu_get(cpu);
	struct cpufreq_policy new_policy;
2271
	int ret;
L
Linus Torvalds 已提交
2272

2273 2274
	if (!policy)
		return -ENODEV;
L
Linus Torvalds 已提交
2275

2276
	down_write(&policy->rwsem);
L
Linus Torvalds 已提交
2277

2278
	pr_debug("updating policy for CPU %u\n", cpu);
2279
	memcpy(&new_policy, policy, sizeof(*policy));
2280 2281
	new_policy.min = policy->user_policy.min;
	new_policy.max = policy->user_policy.max;
L
Linus Torvalds 已提交
2282

2283 2284 2285 2286
	/*
	 * BIOS might change freq behind our back
	 * -> ask driver for current freq and notify governors about a change
	 */
2287
	if (cpufreq_driver->get && !cpufreq_driver->setpolicy) {
2288
		new_policy.cur = cpufreq_update_current_freq(policy);
2289 2290
		if (WARN_ON(!new_policy.cur)) {
			ret = -EIO;
2291
			goto unlock;
2292
		}
2293 2294
	}

2295
	ret = cpufreq_set_policy(policy, &new_policy);
L
Linus Torvalds 已提交
2296

2297
unlock:
2298
	up_write(&policy->rwsem);
2299

2300
	cpufreq_cpu_put(policy);
L
Linus Torvalds 已提交
2301 2302 2303 2304
	return ret;
}
EXPORT_SYMBOL(cpufreq_update_policy);

2305
static int cpufreq_cpu_callback(struct notifier_block *nfb,
2306 2307 2308 2309
					unsigned long action, void *hcpu)
{
	unsigned int cpu = (unsigned long)hcpu;

2310 2311
	switch (action & ~CPU_TASKS_FROZEN) {
	case CPU_ONLINE:
2312
	case CPU_DOWN_FAILED:
2313 2314
		cpufreq_online(cpu);
		break;
2315

2316
	case CPU_DOWN_PREPARE:
2317
		cpufreq_offline(cpu);
2318
		break;
2319 2320 2321 2322
	}
	return NOTIFY_OK;
}

2323
static struct notifier_block __refdata cpufreq_cpu_notifier = {
2324
	.notifier_call = cpufreq_cpu_callback,
2325
};
L
Linus Torvalds 已提交
2326

2327 2328 2329 2330 2331 2332 2333 2334
/*********************************************************************
 *               BOOST						     *
 *********************************************************************/
static int cpufreq_boost_set_sw(int state)
{
	struct cpufreq_policy *policy;
	int ret = -EINVAL;

2335
	for_each_active_policy(policy) {
2336 2337
		if (!policy->freq_table)
			continue;
2338

2339 2340 2341 2342 2343 2344
		ret = cpufreq_frequency_table_cpuinfo(policy,
						      policy->freq_table);
		if (ret) {
			pr_err("%s: Policy frequency update failed\n",
			       __func__);
			break;
2345
		}
2346 2347 2348 2349 2350

		down_write(&policy->rwsem);
		policy->user_policy.max = policy->max;
		cpufreq_governor_limits(policy);
		up_write(&policy->rwsem);
2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373
	}

	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);

2374 2375
		pr_err("%s: Cannot %s BOOST\n",
		       __func__, state ? "enable" : "disable");
2376 2377 2378 2379 2380
	}

	return ret;
}

2381
static bool cpufreq_boost_supported(void)
2382
{
2383
	return likely(cpufreq_driver) && cpufreq_driver->set_boost;
2384 2385
}

2386 2387 2388 2389
static int create_boost_sysfs_file(void)
{
	int ret;

2390
	ret = sysfs_create_file(cpufreq_global_kobject, &boost.attr);
2391 2392 2393 2394 2395 2396 2397 2398 2399 2400
	if (ret)
		pr_err("%s: cannot register global BOOST sysfs file\n",
		       __func__);

	return ret;
}

static void remove_boost_sysfs_file(void)
{
	if (cpufreq_boost_supported())
2401
		sysfs_remove_file(cpufreq_global_kobject, &boost.attr);
2402 2403 2404 2405 2406 2407 2408 2409 2410 2411
}

int cpufreq_enable_boost_support(void)
{
	if (!cpufreq_driver)
		return -EINVAL;

	if (cpufreq_boost_supported())
		return 0;

2412
	cpufreq_driver->set_boost = cpufreq_boost_set_sw;
2413 2414 2415 2416 2417 2418

	/* This will get removed on driver unregister */
	return create_boost_sysfs_file();
}
EXPORT_SYMBOL_GPL(cpufreq_enable_boost_support);

2419 2420 2421 2422 2423 2424
int cpufreq_boost_enabled(void)
{
	return cpufreq_driver->boost_enabled;
}
EXPORT_SYMBOL_GPL(cpufreq_boost_enabled);

L
Linus Torvalds 已提交
2425 2426 2427 2428 2429 2430 2431 2432 2433
/*********************************************************************
 *               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.
 *
2434
 * Registers a CPU Frequency driver to this core code. This code
2435
 * returns zero on success, -EEXIST when another driver got here first
2436
 * (and isn't unregistered in the meantime).
L
Linus Torvalds 已提交
2437 2438
 *
 */
2439
int cpufreq_register_driver(struct cpufreq_driver *driver_data)
L
Linus Torvalds 已提交
2440 2441 2442 2443
{
	unsigned long flags;
	int ret;

2444 2445 2446
	if (cpufreq_disabled())
		return -ENODEV;

L
Linus Torvalds 已提交
2447
	if (!driver_data || !driver_data->verify || !driver_data->init ||
2448
	    !(driver_data->setpolicy || driver_data->target_index ||
2449 2450
		    driver_data->target) ||
	     (driver_data->setpolicy && (driver_data->target_index ||
2451 2452
		    driver_data->target)) ||
	     (!!driver_data->get_intermediate != !!driver_data->target_intermediate))
L
Linus Torvalds 已提交
2453 2454
		return -EINVAL;

2455
	pr_debug("trying to register driver %s\n", driver_data->name);
L
Linus Torvalds 已提交
2456

2457 2458 2459
	/* Protect against concurrent CPU online/offline. */
	get_online_cpus();

2460
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2461
	if (cpufreq_driver) {
2462
		write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2463 2464
		ret = -EEXIST;
		goto out;
L
Linus Torvalds 已提交
2465
	}
2466
	cpufreq_driver = driver_data;
2467
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
L
Linus Torvalds 已提交
2468

2469 2470 2471
	if (driver_data->setpolicy)
		driver_data->flags |= CPUFREQ_CONST_LOOPS;

2472 2473 2474 2475 2476
	if (cpufreq_boost_supported()) {
		ret = create_boost_sysfs_file();
		if (ret)
			goto err_null_driver;
	}
2477

2478
	ret = subsys_interface_register(&cpufreq_interface);
2479
	if (ret)
2480
		goto err_boost_unreg;
L
Linus Torvalds 已提交
2481

2482 2483
	if (!(cpufreq_driver->flags & CPUFREQ_STICKY) &&
	    list_empty(&cpufreq_policy_list)) {
L
Linus Torvalds 已提交
2484
		/* if all ->init() calls failed, unregister */
2485 2486 2487
		pr_debug("%s: No CPU initialized for driver %s\n", __func__,
			 driver_data->name);
		goto err_if_unreg;
L
Linus Torvalds 已提交
2488 2489
	}

2490
	register_hotcpu_notifier(&cpufreq_cpu_notifier);
2491
	pr_debug("driver %s up and running\n", driver_data->name);
2492
	goto out;
2493

2494 2495
err_if_unreg:
	subsys_interface_unregister(&cpufreq_interface);
2496
err_boost_unreg:
2497
	remove_boost_sysfs_file();
2498
err_null_driver:
2499
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2500
	cpufreq_driver = NULL;
2501
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2502 2503 2504
out:
	put_online_cpus();
	return ret;
L
Linus Torvalds 已提交
2505 2506 2507 2508 2509 2510
}
EXPORT_SYMBOL_GPL(cpufreq_register_driver);

/**
 * cpufreq_unregister_driver - unregister the current CPUFreq driver
 *
2511
 * Unregister the current CPUFreq driver. Only call this if you have
L
Linus Torvalds 已提交
2512 2513 2514 2515
 * 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.
 */
2516
int cpufreq_unregister_driver(struct cpufreq_driver *driver)
L
Linus Torvalds 已提交
2517 2518 2519
{
	unsigned long flags;

2520
	if (!cpufreq_driver || (driver != cpufreq_driver))
L
Linus Torvalds 已提交
2521 2522
		return -EINVAL;

2523
	pr_debug("unregistering driver %s\n", driver->name);
L
Linus Torvalds 已提交
2524

2525 2526
	/* Protect against concurrent cpu hotplug */
	get_online_cpus();
2527
	subsys_interface_unregister(&cpufreq_interface);
2528
	remove_boost_sysfs_file();
2529
	unregister_hotcpu_notifier(&cpufreq_cpu_notifier);
L
Linus Torvalds 已提交
2530

2531
	write_lock_irqsave(&cpufreq_driver_lock, flags);
2532

2533
	cpufreq_driver = NULL;
2534

2535
	write_unlock_irqrestore(&cpufreq_driver_lock, flags);
2536
	put_online_cpus();
L
Linus Torvalds 已提交
2537 2538 2539 2540

	return 0;
}
EXPORT_SYMBOL_GPL(cpufreq_unregister_driver);
2541

2542 2543 2544 2545 2546 2547 2548 2549
/*
 * 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,
};

2550 2551 2552
struct kobject *cpufreq_global_kobject;
EXPORT_SYMBOL(cpufreq_global_kobject);

2553 2554
static int __init cpufreq_core_init(void)
{
2555 2556 2557
	if (cpufreq_disabled())
		return -ENODEV;

2558
	cpufreq_global_kobject = kobject_create_and_add("cpufreq", &cpu_subsys.dev_root->kobj);
2559 2560
	BUG_ON(!cpufreq_global_kobject);

2561 2562
	register_syscore_ops(&cpufreq_syscore_ops);

2563 2564 2565
	return 0;
}
core_initcall(cpufreq_core_init);