cpuidle.c 11.6 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

V
Venki Pallipadi 已提交
43 44 45 46 47 48 49 50 51 52 53
#if defined(CONFIG_ARCH_HAS_CPU_IDLE_WAIT)
static void cpuidle_kick_cpus(void)
{
	cpu_idle_wait();
}
#elif defined(CONFIG_SMP)
# error "Arch needs cpu_idle_wait() equivalent here"
#else /* !CONFIG_ARCH_HAS_CPU_IDLE_WAIT && !CONFIG_SMP */
static void cpuidle_kick_cpus(void) {}
#endif

54 55
static int __cpuidle_register_device(struct cpuidle_device *dev);

56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73
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;

74 75 76
/**
 * cpuidle_play_dead - cpu off-lining
 *
77
 * Returns in case of an error or no driver
78 79 80 81 82 83 84 85
 */
int cpuidle_play_dead(void)
{
	struct cpuidle_device *dev = __this_cpu_read(cpuidle_devices);
	struct cpuidle_driver *drv = cpuidle_get_driver();
	int i, dead_state = -1;
	int power_usage = -1;

86 87 88
	if (!drv)
		return -ENODEV;

89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104
	/* 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;
}

105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132
/**
 * 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.
		 */
		dev->states_usage[entered_state].time +=
				(unsigned long long)dev->last_residency;
		dev->states_usage[entered_state].usage++;
	} else {
		dev->last_residency = 0;
	}

	return entered_state;
}

133 134 135 136
/**
 * cpuidle_idle_call - the main idle loop
 *
 * NOTE: no locks or semaphores should be used here
L
Len Brown 已提交
137
 * return non-zero on failure
138
 */
L
Len Brown 已提交
139
int cpuidle_idle_call(void)
140
{
141
	struct cpuidle_device *dev = __this_cpu_read(cpuidle_devices);
142
	struct cpuidle_driver *drv = cpuidle_get_driver();
143
	int next_state, entered_state;
144

L
Len Brown 已提交
145 146 147 148 149 150
	if (off)
		return -ENODEV;

	if (!initialized)
		return -ENODEV;

151
	/* check if the device is ready */
L
Len Brown 已提交
152 153
	if (!dev || !dev->enabled)
		return -EBUSY;
154

155 156
#if 0
	/* shows regressions, re-enable for 2.6.29 */
157 158 159 160 161
	/*
	 * run any timers that can be run now, at this point
	 * before calculating the idle duration etc.
	 */
	hrtimer_peek_ahead_timers();
162
#endif
163

164
	/* ask the governor for the next state */
165
	next_state = cpuidle_curr_governor->select(drv, dev);
166 167
	if (need_resched()) {
		local_irq_enable();
L
Len Brown 已提交
168
		return 0;
169 170
	}

171 172
	trace_power_start_rcuidle(POWER_CSTATE, next_state, dev->cpu);
	trace_cpu_idle_rcuidle(next_state, dev->cpu);
173

174
	entered_state = cpuidle_enter_state(dev, drv, next_state);
175

176 177
	trace_power_end_rcuidle(dev->cpu);
	trace_cpu_idle_rcuidle(PWR_EVENT_EXIT, dev->cpu);
178

179 180
	/* give the governor an opportunity to reflect on the outcome */
	if (cpuidle_curr_governor->reflect)
181
		cpuidle_curr_governor->reflect(dev, entered_state);
L
Len Brown 已提交
182 183

	return 0;
184 185 186 187 188 189 190
}

/**
 * cpuidle_install_idle_handler - installs the cpuidle idle loop handler
 */
void cpuidle_install_idle_handler(void)
{
L
Len Brown 已提交
191
	if (enabled_devices) {
192 193
		/* Make sure all changes finished before we switch to new idle */
		smp_wmb();
L
Len Brown 已提交
194
		initialized = 1;
195 196 197 198 199 200 201 202
	}
}

/**
 * cpuidle_uninstall_idle_handler - uninstalls the cpuidle idle loop handler
 */
void cpuidle_uninstall_idle_handler(void)
{
L
Len Brown 已提交
203 204
	if (enabled_devices) {
		initialized = 0;
V
Venki Pallipadi 已提交
205
		cpuidle_kick_cpus();
206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230
	}
}

/**
 * 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);

231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261
/**
 * 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;
}

262
#ifdef CONFIG_ARCH_HAS_CPU_RELAX
263 264
static int poll_idle(struct cpuidle_device *dev,
		struct cpuidle_driver *drv, int index)
265 266 267 268 269 270 271 272 273 274 275 276 277 278
{
	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;

279 280 281
	dev->last_residency = (int) diff;

	return index;
282 283
}

284
static void poll_idle_init(struct cpuidle_driver *drv)
285
{
286
	struct cpuidle_state *state = &drv->states[0];
287

288
	snprintf(state->name, CPUIDLE_NAME_LEN, "POLL");
289 290 291 292
	snprintf(state->desc, CPUIDLE_DESC_LEN, "CPUIDLE CORE POLL IDLE");
	state->exit_latency = 0;
	state->target_residency = 0;
	state->power_usage = -1;
293
	state->flags = 0;
294
	state->enter = poll_idle;
295
	state->disable = 0;
296 297
}
#else
298
static void poll_idle_init(struct cpuidle_driver *drv) {}
299 300
#endif /* CONFIG_ARCH_HAS_CPU_RELAX */

301 302 303 304 305 306 307 308 309 310
/**
 * 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;
311
	struct cpuidle_driver *drv = cpuidle_get_driver();
312 313 314

	if (dev->enabled)
		return 0;
315
	if (!drv || !cpuidle_curr_governor)
316 317
		return -EIO;
	if (!dev->state_count)
318
		dev->state_count = drv->state_count;
319

320 321 322 323 324 325
	if (dev->registered == 0) {
		ret = __cpuidle_register_device(dev);
		if (ret)
			return ret;
	}

326 327 328 329
	cpuidle_enter_ops = drv->en_core_tk_irqen ?
		cpuidle_enter_tk : cpuidle_enter;

	poll_idle_init(drv);
330

331 332 333 334
	if ((ret = cpuidle_add_state_sysfs(dev)))
		return ret;

	if (cpuidle_curr_governor->enable &&
335
	    (ret = cpuidle_curr_governor->enable(drv, dev)))
336 337 338
		goto fail_sysfs;

	for (i = 0; i < dev->state_count; i++) {
339 340
		dev->states_usage[i].usage = 0;
		dev->states_usage[i].time = 0;
341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369
	}
	dev->last_residency = 0;

	smp_wmb();

	dev->enabled = 1;

	enabled_devices++;
	return 0;

fail_sysfs:
	cpuidle_remove_state_sysfs(dev);

	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)
{
	if (!dev->enabled)
		return;
370
	if (!cpuidle_get_driver() || !cpuidle_curr_governor)
371 372 373 374 375
		return;

	dev->enabled = 0;

	if (cpuidle_curr_governor->disable)
376
		cpuidle_curr_governor->disable(cpuidle_get_driver(), dev);
377 378 379 380 381 382 383 384

	cpuidle_remove_state_sysfs(dev);
	enabled_devices--;
}

EXPORT_SYMBOL_GPL(cpuidle_disable_device);

/**
385 386
 * __cpuidle_register_device - internal register function called before register
 * and enable routines
387
 * @dev: the cpu
388 389
 *
 * cpuidle_lock mutex must be held before this is called
390
 */
391
static int __cpuidle_register_device(struct cpuidle_device *dev)
392 393
{
	int ret;
394
	struct device *cpu_dev = get_cpu_device((unsigned long)dev->cpu);
395
	struct cpuidle_driver *cpuidle_driver = cpuidle_get_driver();
396

397
	if (!dev)
398
		return -EINVAL;
399
	if (!try_module_get(cpuidle_driver->owner))
400 401 402 403 404 405
		return -EINVAL;

	init_completion(&dev->kobj_unregister);

	per_cpu(cpuidle_devices, dev->cpu) = dev;
	list_add(&dev->device_list, &cpuidle_detected_devices);
406 407 408
	ret = cpuidle_add_sysfs(cpu_dev);
	if (ret)
		goto err_sysfs;
409

410 411
	dev->registered = 1;
	return 0;
412 413 414 415 416 417

err_sysfs:
	list_del(&dev->device_list);
	per_cpu(cpuidle_devices, dev->cpu) = NULL;
	module_put(cpuidle_driver->owner);
	return ret;
418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434
}

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

	mutex_lock(&cpuidle_lock);

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

435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451
	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)
{
452
	struct device *cpu_dev = get_cpu_device((unsigned long)dev->cpu);
453
	struct cpuidle_driver *cpuidle_driver = cpuidle_get_driver();
454

455 456 457
	if (dev->registered == 0)
		return;

458 459 460 461
	cpuidle_pause_and_lock();

	cpuidle_disable_device(dev);

462
	cpuidle_remove_sysfs(cpu_dev);
463 464 465 466 467 468
	list_del(&dev->device_list);
	wait_for_completion(&dev->kobj_unregister);
	per_cpu(cpuidle_devices, dev->cpu) = NULL;

	cpuidle_resume_and_unlock();

469
	module_put(cpuidle_driver->owner);
470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489
}

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)
{
490
	smp_call_function(smp_callback, NULL, 1);
491 492 493 494 495 496 497
	return NOTIFY_OK;
}

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

M
Mark Gross 已提交
498 499 500 501
static inline void latency_notifier_init(struct notifier_block *n)
{
	pm_qos_add_notifier(PM_QOS_CPU_DMA_LATENCY, n);
}
502 503 504 505 506 507 508 509 510 511 512 513 514 515

#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;

516 517 518
	if (cpuidle_disabled())
		return -ENODEV;

519
	ret = cpuidle_add_interface(cpu_subsys.dev_root);
520 521 522 523 524 525 526 527
	if (ret)
		return ret;

	latency_notifier_init(&cpuidle_latency_notifier);

	return 0;
}

528
module_param(off, int, 0444);
529
core_initcall(cpuidle_init);