cpuidle.c 11.8 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14
/*
 * cpuidle.c - core cpuidle infrastructure
 *
 * (C) 2006-2007 Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
 *               Shaohua Li <shaohua.li@intel.com>
 *               Adam Belay <abelay@novell.com>
 *
 * This code is licenced under the GPL.
 */

#include <linux/kernel.h>
#include <linux/mutex.h>
#include <linux/sched.h>
#include <linux/notifier.h>
15
#include <linux/pm_qos.h>
16 17
#include <linux/cpu.h>
#include <linux/cpuidle.h>
18
#include <linux/ktime.h>
19
#include <linux/hrtimer.h>
20
#include <linux/module.h>
21
#include <trace/events/power.h>
22 23 24 25 26 27 28 29 30

#include "cpuidle.h"

DEFINE_PER_CPU(struct cpuidle_device *, cpuidle_devices);

DEFINE_MUTEX(cpuidle_lock);
LIST_HEAD(cpuidle_detected_devices);

static int enabled_devices;
31
static int off __read_mostly;
L
Len Brown 已提交
32
static int initialized __read_mostly;
33 34 35 36 37

int cpuidle_disabled(void)
{
	return off;
}
38 39 40 41
void disable_cpuidle(void)
{
	off = 1;
}
42

43 44
static int __cpuidle_register_device(struct cpuidle_device *dev);

45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62
static inline int cpuidle_enter(struct cpuidle_device *dev,
				struct cpuidle_driver *drv, int index)
{
	struct cpuidle_state *target_state = &drv->states[index];
	return target_state->enter(dev, drv, index);
}

static inline int cpuidle_enter_tk(struct cpuidle_device *dev,
			       struct cpuidle_driver *drv, int index)
{
	return cpuidle_wrap_enter(dev, drv, index, cpuidle_enter);
}

typedef int (*cpuidle_enter_t)(struct cpuidle_device *dev,
			       struct cpuidle_driver *drv, int index);

static cpuidle_enter_t cpuidle_enter_ops;

63 64 65
/**
 * cpuidle_play_dead - cpu off-lining
 *
66
 * Returns in case of an error or no driver
67 68 69 70
 */
int cpuidle_play_dead(void)
{
	struct cpuidle_device *dev = __this_cpu_read(cpuidle_devices);
D
Daniel Lezcano 已提交
71
	struct cpuidle_driver *drv = cpuidle_get_cpu_driver(dev);
72
	int i, dead_state = -1;
73
	int power_usage = INT_MAX;
74

75 76 77
	if (!drv)
		return -ENODEV;

78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93
	/* Find lowest-power state that supports long-term idle */
	for (i = CPUIDLE_DRIVER_STATE_START; i < drv->state_count; i++) {
		struct cpuidle_state *s = &drv->states[i];

		if (s->power_usage < power_usage && s->enter_dead) {
			power_usage = s->power_usage;
			dead_state = i;
		}
	}

	if (dead_state != -1)
		return drv->states[dead_state].enter_dead(dev, dead_state);

	return -ENODEV;
}

94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111
/**
 * cpuidle_enter_state - enter the state and update stats
 * @dev: cpuidle device for this cpu
 * @drv: cpuidle driver for this cpu
 * @next_state: index into drv->states of the state to enter
 */
int cpuidle_enter_state(struct cpuidle_device *dev, struct cpuidle_driver *drv,
		int next_state)
{
	int entered_state;

	entered_state = cpuidle_enter_ops(dev, drv, next_state);

	if (entered_state >= 0) {
		/* Update cpuidle counters */
		/* This can be moved to within driver enter routine
		 * but that results in multiple copies of same code.
		 */
112
		dev->states_usage[entered_state].time += dev->last_residency;
113 114 115 116 117 118 119 120
		dev->states_usage[entered_state].usage++;
	} else {
		dev->last_residency = 0;
	}

	return entered_state;
}

121 122 123 124
/**
 * cpuidle_idle_call - the main idle loop
 *
 * NOTE: no locks or semaphores should be used here
L
Len Brown 已提交
125
 * return non-zero on failure
126
 */
L
Len Brown 已提交
127
int cpuidle_idle_call(void)
128
{
129
	struct cpuidle_device *dev = __this_cpu_read(cpuidle_devices);
D
Daniel Lezcano 已提交
130
	struct cpuidle_driver *drv;
131
	int next_state, entered_state;
132

L
Len Brown 已提交
133 134 135 136 137 138
	if (off)
		return -ENODEV;

	if (!initialized)
		return -ENODEV;

139
	/* check if the device is ready */
L
Len Brown 已提交
140 141
	if (!dev || !dev->enabled)
		return -EBUSY;
142

D
Daniel Lezcano 已提交
143 144
	drv = cpuidle_get_cpu_driver(dev);

145
	/* ask the governor for the next state */
146
	next_state = cpuidle_curr_governor->select(drv, dev);
147
	if (need_resched()) {
148 149 150 151
		dev->last_residency = 0;
		/* give the governor an opportunity to reflect on the outcome */
		if (cpuidle_curr_governor->reflect)
			cpuidle_curr_governor->reflect(dev, next_state);
152
		local_irq_enable();
L
Len Brown 已提交
153
		return 0;
154 155
	}

156 157
	trace_power_start_rcuidle(POWER_CSTATE, next_state, dev->cpu);
	trace_cpu_idle_rcuidle(next_state, dev->cpu);
158

159 160 161 162 163
	if (cpuidle_state_is_coupled(dev, drv, next_state))
		entered_state = cpuidle_enter_state_coupled(dev, drv,
							    next_state);
	else
		entered_state = cpuidle_enter_state(dev, drv, next_state);
164

165 166
	trace_power_end_rcuidle(dev->cpu);
	trace_cpu_idle_rcuidle(PWR_EVENT_EXIT, dev->cpu);
167

168 169
	/* give the governor an opportunity to reflect on the outcome */
	if (cpuidle_curr_governor->reflect)
170
		cpuidle_curr_governor->reflect(dev, entered_state);
L
Len Brown 已提交
171 172

	return 0;
173 174 175 176 177 178 179
}

/**
 * cpuidle_install_idle_handler - installs the cpuidle idle loop handler
 */
void cpuidle_install_idle_handler(void)
{
L
Len Brown 已提交
180
	if (enabled_devices) {
181 182
		/* Make sure all changes finished before we switch to new idle */
		smp_wmb();
L
Len Brown 已提交
183
		initialized = 1;
184 185 186 187 188 189 190 191
	}
}

/**
 * cpuidle_uninstall_idle_handler - uninstalls the cpuidle idle loop handler
 */
void cpuidle_uninstall_idle_handler(void)
{
L
Len Brown 已提交
192 193
	if (enabled_devices) {
		initialized = 0;
194
		kick_all_cpus_sync();
195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219
	}
}

/**
 * cpuidle_pause_and_lock - temporarily disables CPUIDLE
 */
void cpuidle_pause_and_lock(void)
{
	mutex_lock(&cpuidle_lock);
	cpuidle_uninstall_idle_handler();
}

EXPORT_SYMBOL_GPL(cpuidle_pause_and_lock);

/**
 * cpuidle_resume_and_unlock - resumes CPUIDLE operation
 */
void cpuidle_resume_and_unlock(void)
{
	cpuidle_install_idle_handler();
	mutex_unlock(&cpuidle_lock);
}

EXPORT_SYMBOL_GPL(cpuidle_resume_and_unlock);

220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235
/* Currently used in suspend/resume path to suspend cpuidle */
void cpuidle_pause(void)
{
	mutex_lock(&cpuidle_lock);
	cpuidle_uninstall_idle_handler();
	mutex_unlock(&cpuidle_lock);
}

/* Currently used in suspend/resume path to resume cpuidle */
void cpuidle_resume(void)
{
	mutex_lock(&cpuidle_lock);
	cpuidle_install_idle_handler();
	mutex_unlock(&cpuidle_lock);
}

236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266
/**
 * cpuidle_wrap_enter - performs timekeeping and irqen around enter function
 * @dev: pointer to a valid cpuidle_device object
 * @drv: pointer to a valid cpuidle_driver object
 * @index: index of the target cpuidle state.
 */
int cpuidle_wrap_enter(struct cpuidle_device *dev,
				struct cpuidle_driver *drv, int index,
				int (*enter)(struct cpuidle_device *dev,
					struct cpuidle_driver *drv, int index))
{
	ktime_t time_start, time_end;
	s64 diff;

	time_start = ktime_get();

	index = enter(dev, drv, index);

	time_end = ktime_get();

	local_irq_enable();

	diff = ktime_to_us(ktime_sub(time_end, time_start));
	if (diff > INT_MAX)
		diff = INT_MAX;

	dev->last_residency = (int) diff;

	return index;
}

267
#ifdef CONFIG_ARCH_HAS_CPU_RELAX
268 269
static int poll_idle(struct cpuidle_device *dev,
		struct cpuidle_driver *drv, int index)
270 271 272 273 274 275 276 277 278 279 280 281 282 283
{
	ktime_t	t1, t2;
	s64 diff;

	t1 = ktime_get();
	local_irq_enable();
	while (!need_resched())
		cpu_relax();

	t2 = ktime_get();
	diff = ktime_to_us(ktime_sub(t2, t1));
	if (diff > INT_MAX)
		diff = INT_MAX;

284 285 286
	dev->last_residency = (int) diff;

	return index;
287 288
}

289
static void poll_idle_init(struct cpuidle_driver *drv)
290
{
291
	struct cpuidle_state *state = &drv->states[0];
292

293
	snprintf(state->name, CPUIDLE_NAME_LEN, "POLL");
294 295 296 297
	snprintf(state->desc, CPUIDLE_DESC_LEN, "CPUIDLE CORE POLL IDLE");
	state->exit_latency = 0;
	state->target_residency = 0;
	state->power_usage = -1;
298
	state->flags = 0;
299
	state->enter = poll_idle;
300
	state->disabled = false;
301 302
}
#else
303
static void poll_idle_init(struct cpuidle_driver *drv) {}
304 305
#endif /* CONFIG_ARCH_HAS_CPU_RELAX */

306 307 308 309 310 311 312 313 314 315
/**
 * cpuidle_enable_device - enables idle PM for a CPU
 * @dev: the CPU
 *
 * This function must be called between cpuidle_pause_and_lock and
 * cpuidle_resume_and_unlock when used externally.
 */
int cpuidle_enable_device(struct cpuidle_device *dev)
{
	int ret, i;
D
Daniel Lezcano 已提交
316
	struct cpuidle_driver *drv;
317

318 319 320
	if (!dev)
		return -EINVAL;

321 322
	if (dev->enabled)
		return 0;
D
Daniel Lezcano 已提交
323 324 325

	drv = cpuidle_get_cpu_driver(dev);

326
	if (!drv || !cpuidle_curr_governor)
327
		return -EIO;
D
Daniel Lezcano 已提交
328

329
	if (!dev->state_count)
330
		dev->state_count = drv->state_count;
331

332 333 334 335 336 337
	if (dev->registered == 0) {
		ret = __cpuidle_register_device(dev);
		if (ret)
			return ret;
	}

338 339 340 341
	cpuidle_enter_ops = drv->en_core_tk_irqen ?
		cpuidle_enter_tk : cpuidle_enter;

	poll_idle_init(drv);
342

D
Daniel Lezcano 已提交
343 344
	ret = cpuidle_add_device_sysfs(dev);
	if (ret)
345 346 347
		return ret;

	if (cpuidle_curr_governor->enable &&
348
	    (ret = cpuidle_curr_governor->enable(drv, dev)))
349 350 351
		goto fail_sysfs;

	for (i = 0; i < dev->state_count; i++) {
352 353
		dev->states_usage[i].usage = 0;
		dev->states_usage[i].time = 0;
354 355 356 357 358 359 360 361 362 363 364
	}
	dev->last_residency = 0;

	smp_wmb();

	dev->enabled = 1;

	enabled_devices++;
	return 0;

fail_sysfs:
D
Daniel Lezcano 已提交
365
	cpuidle_remove_device_sysfs(dev);
366 367 368 369 370 371 372 373 374 375 376 377 378 379 380

	return ret;
}

EXPORT_SYMBOL_GPL(cpuidle_enable_device);

/**
 * cpuidle_disable_device - disables idle PM for a CPU
 * @dev: the CPU
 *
 * This function must be called between cpuidle_pause_and_lock and
 * cpuidle_resume_and_unlock when used externally.
 */
void cpuidle_disable_device(struct cpuidle_device *dev)
{
D
Daniel Lezcano 已提交
381 382
	struct cpuidle_driver *drv = cpuidle_get_cpu_driver(dev);

383
	if (!dev || !dev->enabled)
384
		return;
D
Daniel Lezcano 已提交
385 386

	if (!drv || !cpuidle_curr_governor)
387 388 389 390 391
		return;

	dev->enabled = 0;

	if (cpuidle_curr_governor->disable)
D
Daniel Lezcano 已提交
392
		cpuidle_curr_governor->disable(drv, dev);
393

D
Daniel Lezcano 已提交
394
	cpuidle_remove_device_sysfs(dev);
395 396 397 398 399 400
	enabled_devices--;
}

EXPORT_SYMBOL_GPL(cpuidle_disable_device);

/**
401 402
 * __cpuidle_register_device - internal register function called before register
 * and enable routines
403
 * @dev: the cpu
404 405
 *
 * cpuidle_lock mutex must be held before this is called
406
 */
407
static int __cpuidle_register_device(struct cpuidle_device *dev)
408 409
{
	int ret;
D
Daniel Lezcano 已提交
410
	struct cpuidle_driver *drv = cpuidle_get_cpu_driver(dev);
411

D
Daniel Lezcano 已提交
412
	if (!try_module_get(drv->owner))
413 414 415 416
		return -EINVAL;

	per_cpu(cpuidle_devices, dev->cpu) = dev;
	list_add(&dev->device_list, &cpuidle_detected_devices);
417
	ret = cpuidle_add_sysfs(dev);
418 419
	if (ret)
		goto err_sysfs;
420

421 422 423
	ret = cpuidle_coupled_register_device(dev);
	if (ret)
		goto err_coupled;
424

425 426
	dev->registered = 1;
	return 0;
427

428
err_coupled:
429
	cpuidle_remove_sysfs(dev);
430 431 432
err_sysfs:
	list_del(&dev->device_list);
	per_cpu(cpuidle_devices, dev->cpu) = NULL;
D
Daniel Lezcano 已提交
433
	module_put(drv->owner);
434
	return ret;
435 436 437 438 439 440 441 442 443 444
}

/**
 * cpuidle_register_device - registers a CPU's idle PM feature
 * @dev: the cpu
 */
int cpuidle_register_device(struct cpuidle_device *dev)
{
	int ret;

445 446 447
	if (!dev)
		return -EINVAL;

448 449 450 451 452 453 454
	mutex_lock(&cpuidle_lock);

	if ((ret = __cpuidle_register_device(dev))) {
		mutex_unlock(&cpuidle_lock);
		return ret;
	}

455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471
	cpuidle_enable_device(dev);
	cpuidle_install_idle_handler();

	mutex_unlock(&cpuidle_lock);

	return 0;

}

EXPORT_SYMBOL_GPL(cpuidle_register_device);

/**
 * cpuidle_unregister_device - unregisters a CPU's idle PM feature
 * @dev: the cpu
 */
void cpuidle_unregister_device(struct cpuidle_device *dev)
{
D
Daniel Lezcano 已提交
472
	struct cpuidle_driver *drv = cpuidle_get_cpu_driver(dev);
473

474 475 476
	if (dev->registered == 0)
		return;

477 478 479 480
	cpuidle_pause_and_lock();

	cpuidle_disable_device(dev);

481
	cpuidle_remove_sysfs(dev);
482 483 484
	list_del(&dev->device_list);
	per_cpu(cpuidle_devices, dev->cpu) = NULL;

485 486
	cpuidle_coupled_unregister_device(dev);

487 488
	cpuidle_resume_and_unlock();

D
Daniel Lezcano 已提交
489
	module_put(drv->owner);
490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509
}

EXPORT_SYMBOL_GPL(cpuidle_unregister_device);

#ifdef CONFIG_SMP

static void smp_callback(void *v)
{
	/* we already woke the CPU up, nothing more to do */
}

/*
 * This function gets called when a part of the kernel has a new latency
 * requirement.  This means we need to get all processors out of their C-state,
 * and then recalculate a new suitable C-state. Just do a cross-cpu IPI; that
 * wakes them all right up.
 */
static int cpuidle_latency_notify(struct notifier_block *b,
		unsigned long l, void *v)
{
510
	smp_call_function(smp_callback, NULL, 1);
511 512 513 514 515 516 517
	return NOTIFY_OK;
}

static struct notifier_block cpuidle_latency_notifier = {
	.notifier_call = cpuidle_latency_notify,
};

M
Mark Gross 已提交
518 519 520 521
static inline void latency_notifier_init(struct notifier_block *n)
{
	pm_qos_add_notifier(PM_QOS_CPU_DMA_LATENCY, n);
}
522 523 524 525 526 527 528 529 530 531 532 533 534 535

#else /* CONFIG_SMP */

#define latency_notifier_init(x) do { } while (0)

#endif /* CONFIG_SMP */

/**
 * cpuidle_init - core initializer
 */
static int __init cpuidle_init(void)
{
	int ret;

536 537 538
	if (cpuidle_disabled())
		return -ENODEV;

539
	ret = cpuidle_add_interface(cpu_subsys.dev_root);
540 541 542 543 544 545 546 547
	if (ret)
		return ret;

	latency_notifier_init(&cpuidle_latency_notifier);

	return 0;
}

548
module_param(off, int, 0444);
549
core_initcall(cpuidle_init);