cpuidle.c 12.8 KB
Newer Older
1 2 3 4 5 6 7 8 9 10
/*
 * 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.
 */

11
#include <linux/clockchips.h>
12 13 14 15
#include <linux/kernel.h>
#include <linux/mutex.h>
#include <linux/sched.h>
#include <linux/notifier.h>
16
#include <linux/pm_qos.h>
17 18
#include <linux/cpu.h>
#include <linux/cpuidle.h>
19
#include <linux/ktime.h>
20
#include <linux/hrtimer.h>
21
#include <linux/module.h>
22
#include <trace/events/power.h>
23 24 25 26

#include "cpuidle.h"

DEFINE_PER_CPU(struct cpuidle_device *, cpuidle_devices);
27
DEFINE_PER_CPU(struct cpuidle_device, cpuidle_dev);
28 29 30 31 32

DEFINE_MUTEX(cpuidle_lock);
LIST_HEAD(cpuidle_detected_devices);

static int enabled_devices;
33
static int off __read_mostly;
L
Len Brown 已提交
34
static int initialized __read_mostly;
35
static bool use_deepest_state __read_mostly;
36 37 38 39 40

int cpuidle_disabled(void)
{
	return off;
}
41 42 43 44
void disable_cpuidle(void)
{
	off = 1;
}
45

46 47 48
/**
 * cpuidle_play_dead - cpu off-lining
 *
49
 * Returns in case of an error or no driver
50 51 52 53
 */
int cpuidle_play_dead(void)
{
	struct cpuidle_device *dev = __this_cpu_read(cpuidle_devices);
D
Daniel Lezcano 已提交
54
	struct cpuidle_driver *drv = cpuidle_get_cpu_driver(dev);
55
	int i;
56

57 58 59
	if (!drv)
		return -ENODEV;

60
	/* Find lowest-power state that supports long-term idle */
61 62 63
	for (i = drv->state_count - 1; i >= CPUIDLE_DRIVER_STATE_START; i--)
		if (drv->states[i].enter_dead)
			return drv->states[i].enter_dead(dev, i);
64 65 66 67

	return -ENODEV;
}

68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106
/**
 * cpuidle_use_deepest_state - Enable/disable the "deepest idle" mode.
 * @enable: Whether enable or disable the feature.
 *
 * If the "deepest idle" mode is enabled, cpuidle will ignore the governor and
 * always use the state with the greatest exit latency (out of the states that
 * are not disabled).
 *
 * This function can only be called after cpuidle_pause() to avoid races.
 */
void cpuidle_use_deepest_state(bool enable)
{
	use_deepest_state = enable;
}

/**
 * cpuidle_find_deepest_state - Find the state of the greatest exit latency.
 * @drv: cpuidle driver for a given CPU.
 * @dev: cpuidle device for a given CPU.
 */
static int cpuidle_find_deepest_state(struct cpuidle_driver *drv,
				      struct cpuidle_device *dev)
{
	unsigned int latency_req = 0;
	int i, ret = CPUIDLE_DRIVER_STATE_START - 1;

	for (i = CPUIDLE_DRIVER_STATE_START; i < drv->state_count; i++) {
		struct cpuidle_state *s = &drv->states[i];
		struct cpuidle_state_usage *su = &dev->states_usage[i];

		if (s->disabled || su->disable || s->exit_latency <= latency_req)
			continue;

		latency_req = s->exit_latency;
		ret = i;
	}
	return ret;
}

107 108 109 110 111 112 113
/**
 * 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,
114
			int index)
115 116 117
{
	int entered_state;

118 119 120 121 122 123 124 125 126 127
	struct cpuidle_state *target_state = &drv->states[index];
	ktime_t time_start, time_end;
	s64 diff;

	time_start = ktime_get();

	entered_state = target_state->enter(dev, drv, index);

	time_end = ktime_get();

128 129
	if (!cpuidle_state_is_coupled(dev, drv, entered_state))
		local_irq_enable();
130 131 132 133 134 135

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

	dev->last_residency = (int) diff;
136 137 138 139 140 141

	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.
		 */
142
		dev->states_usage[entered_state].time += dev->last_residency;
143 144 145 146 147 148 149 150
		dev->states_usage[entered_state].usage++;
	} else {
		dev->last_residency = 0;
	}

	return entered_state;
}

151
/**
152 153 154 155
 * cpuidle_select - ask the cpuidle framework to choose an idle state
 *
 * @drv: the cpuidle driver
 * @dev: the cpuidle device
156
 *
157
 * Returns the index of the idle state.
158
 */
159
int cpuidle_select(struct cpuidle_driver *drv, struct cpuidle_device *dev)
160
{
161 162 163 164 165 166
	if (off || !initialized)
		return -ENODEV;

	if (!drv || !dev || !dev->enabled)
		return -EBUSY;

167 168 169
	if (unlikely(use_deepest_state))
		return cpuidle_find_deepest_state(drv, dev);

170 171
	return cpuidle_curr_governor->select(drv, dev);
}
172

173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189
/**
 * cpuidle_enter - enter into the specified idle state
 *
 * @drv:   the cpuidle driver tied with the cpu
 * @dev:   the cpuidle device
 * @index: the index in the idle state table
 *
 * Returns the index in the idle state, < 0 in case of error.
 * The error code depends on the backend driver
 */
int cpuidle_enter(struct cpuidle_driver *drv, struct cpuidle_device *dev,
		  int index)
{
	if (cpuidle_state_is_coupled(dev, drv, index))
		return cpuidle_enter_state_coupled(dev, drv, index);
	return cpuidle_enter_state(dev, drv, index);
}
190

191 192 193 194 195 196 197 198 199 200
/**
 * cpuidle_reflect - tell the underlying governor what was the state
 * we were in
 *
 * @dev  : the cpuidle device
 * @index: the index in the idle state table
 *
 */
void cpuidle_reflect(struct cpuidle_device *dev, int index)
{
201
	if (cpuidle_curr_governor->reflect && !unlikely(use_deepest_state))
202
		cpuidle_curr_governor->reflect(dev, index);
203 204 205 206 207 208 209
}

/**
 * cpuidle_install_idle_handler - installs the cpuidle idle loop handler
 */
void cpuidle_install_idle_handler(void)
{
L
Len Brown 已提交
210
	if (enabled_devices) {
211 212
		/* Make sure all changes finished before we switch to new idle */
		smp_wmb();
L
Len Brown 已提交
213
		initialized = 1;
214 215 216 217 218 219 220 221
	}
}

/**
 * cpuidle_uninstall_idle_handler - uninstalls the cpuidle idle loop handler
 */
void cpuidle_uninstall_idle_handler(void)
{
L
Len Brown 已提交
222 223
	if (enabled_devices) {
		initialized = 0;
224
		kick_all_cpus_sync();
225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249
	}
}

/**
 * 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);

250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265
/* 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);
}

266 267 268 269 270 271 272 273 274
/**
 * 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)
{
275
	int ret;
D
Daniel Lezcano 已提交
276
	struct cpuidle_driver *drv;
277

278 279 280
	if (!dev)
		return -EINVAL;

281 282
	if (dev->enabled)
		return 0;
D
Daniel Lezcano 已提交
283 284 285

	drv = cpuidle_get_cpu_driver(dev);

286
	if (!drv || !cpuidle_curr_governor)
287
		return -EIO;
D
Daniel Lezcano 已提交
288

289 290 291
	if (!dev->registered)
		return -EINVAL;

292
	if (!dev->state_count)
293
		dev->state_count = drv->state_count;
294

D
Daniel Lezcano 已提交
295 296
	ret = cpuidle_add_device_sysfs(dev);
	if (ret)
297 298 299
		return ret;

	if (cpuidle_curr_governor->enable &&
300
	    (ret = cpuidle_curr_governor->enable(drv, dev)))
301 302 303 304 305 306 307 308 309 310
		goto fail_sysfs;

	smp_wmb();

	dev->enabled = 1;

	enabled_devices++;
	return 0;

fail_sysfs:
D
Daniel Lezcano 已提交
311
	cpuidle_remove_device_sysfs(dev);
312 313 314 315 316 317 318 319 320 321 322 323 324 325 326

	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
327 328
	struct cpuidle_driver *drv = cpuidle_get_cpu_driver(dev);

329
	if (!dev || !dev->enabled)
330
		return;
D
Daniel Lezcano 已提交
331 332

	if (!drv || !cpuidle_curr_governor)
333 334 335 336 337
		return;

	dev->enabled = 0;

	if (cpuidle_curr_governor->disable)
D
Daniel Lezcano 已提交
338
		cpuidle_curr_governor->disable(drv, dev);
339

D
Daniel Lezcano 已提交
340
	cpuidle_remove_device_sysfs(dev);
341 342 343 344 345
	enabled_devices--;
}

EXPORT_SYMBOL_GPL(cpuidle_disable_device);

346 347 348 349 350 351 352 353 354
static void __cpuidle_unregister_device(struct cpuidle_device *dev)
{
	struct cpuidle_driver *drv = cpuidle_get_cpu_driver(dev);

	list_del(&dev->device_list);
	per_cpu(cpuidle_devices, dev->cpu) = NULL;
	module_put(drv->owner);
}

355
static void __cpuidle_device_init(struct cpuidle_device *dev)
356 357 358 359 360
{
	memset(dev->states_usage, 0, sizeof(dev->states_usage));
	dev->last_residency = 0;
}

361
/**
362 363
 * __cpuidle_register_device - internal register function called before register
 * and enable routines
364
 * @dev: the cpu
365 366
 *
 * cpuidle_lock mutex must be held before this is called
367
 */
368
static int __cpuidle_register_device(struct cpuidle_device *dev)
369 370
{
	int ret;
D
Daniel Lezcano 已提交
371
	struct cpuidle_driver *drv = cpuidle_get_cpu_driver(dev);
372

D
Daniel Lezcano 已提交
373
	if (!try_module_get(drv->owner))
374 375 376 377 378
		return -EINVAL;

	per_cpu(cpuidle_devices, dev->cpu) = dev;
	list_add(&dev->device_list, &cpuidle_detected_devices);

379
	ret = cpuidle_coupled_register_device(dev);
380
	if (ret)
381
		__cpuidle_unregister_device(dev);
382 383
	else
		dev->registered = 1;
384

385
	return ret;
386 387 388 389 390 391 392 393
}

/**
 * cpuidle_register_device - registers a CPU's idle PM feature
 * @dev: the cpu
 */
int cpuidle_register_device(struct cpuidle_device *dev)
{
394
	int ret = -EBUSY;
395

396 397 398
	if (!dev)
		return -EINVAL;

399 400
	mutex_lock(&cpuidle_lock);

401 402 403
	if (dev->registered)
		goto out_unlock;

404
	__cpuidle_device_init(dev);
405

406 407 408 409 410 411 412
	ret = __cpuidle_register_device(dev);
	if (ret)
		goto out_unlock;

	ret = cpuidle_add_sysfs(dev);
	if (ret)
		goto out_unregister;
413

414
	ret = cpuidle_enable_device(dev);
415 416
	if (ret)
		goto out_sysfs;
417

418 419
	cpuidle_install_idle_handler();

420
out_unlock:
421 422
	mutex_unlock(&cpuidle_lock);

423 424 425 426 427 428 429
	return ret;

out_sysfs:
	cpuidle_remove_sysfs(dev);
out_unregister:
	__cpuidle_unregister_device(dev);
	goto out_unlock;
430 431 432 433 434 435 436 437 438 439
}

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)
{
440
	if (!dev || dev->registered == 0)
441 442
		return;

443 444 445 446
	cpuidle_pause_and_lock();

	cpuidle_disable_device(dev);

447
	cpuidle_remove_sysfs(dev);
448 449

	__cpuidle_unregister_device(dev);
450

451 452
	cpuidle_coupled_unregister_device(dev);

453 454 455 456 457
	cpuidle_resume_and_unlock();
}

EXPORT_SYMBOL_GPL(cpuidle_unregister_device);

D
Daniel Lezcano 已提交
458
/**
459 460 461 462 463 464 465 466 467 468 469
 * cpuidle_unregister: unregister a driver and the devices. This function
 * can be used only if the driver has been previously registered through
 * the cpuidle_register function.
 *
 * @drv: a valid pointer to a struct cpuidle_driver
 */
void cpuidle_unregister(struct cpuidle_driver *drv)
{
	int cpu;
	struct cpuidle_device *device;

470
	for_each_cpu(cpu, drv->cpumask) {
471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501
		device = &per_cpu(cpuidle_dev, cpu);
		cpuidle_unregister_device(device);
	}

	cpuidle_unregister_driver(drv);
}
EXPORT_SYMBOL_GPL(cpuidle_unregister);

/**
 * cpuidle_register: registers the driver and the cpu devices with the
 * coupled_cpus passed as parameter. This function is used for all common
 * initialization pattern there are in the arch specific drivers. The
 * devices is globally defined in this file.
 *
 * @drv         : a valid pointer to a struct cpuidle_driver
 * @coupled_cpus: a cpumask for the coupled states
 *
 * Returns 0 on success, < 0 otherwise
 */
int cpuidle_register(struct cpuidle_driver *drv,
		     const struct cpumask *const coupled_cpus)
{
	int ret, cpu;
	struct cpuidle_device *device;

	ret = cpuidle_register_driver(drv);
	if (ret) {
		pr_err("failed to register cpuidle driver\n");
		return ret;
	}

502
	for_each_cpu(cpu, drv->cpumask) {
503 504 505 506 507
		device = &per_cpu(cpuidle_dev, cpu);
		device->cpu = cpu;

#ifdef CONFIG_ARCH_NEEDS_CPU_IDLE_COUPLED
		/*
508
		 * On multiplatform for ARM, the coupled idle states could be
509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528
		 * enabled in the kernel even if the cpuidle driver does not
		 * use it. Note, coupled_cpus is a struct copy.
		 */
		if (coupled_cpus)
			device->coupled_cpus = *coupled_cpus;
#endif
		ret = cpuidle_register_device(device);
		if (!ret)
			continue;

		pr_err("Failed to register cpuidle device for cpu%d\n", cpu);

		cpuidle_unregister(drv);
		break;
	}

	return ret;
}
EXPORT_SYMBOL_GPL(cpuidle_register);

529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544
#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)
{
545
	smp_call_function(smp_callback, NULL, 1);
546 547 548 549 550 551 552
	return NOTIFY_OK;
}

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

M
Mark Gross 已提交
553 554 555 556
static inline void latency_notifier_init(struct notifier_block *n)
{
	pm_qos_add_notifier(PM_QOS_CPU_DMA_LATENCY, n);
}
557 558 559 560 561 562 563 564 565 566 567 568 569 570

#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;

571 572 573
	if (cpuidle_disabled())
		return -ENODEV;

574
	ret = cpuidle_add_interface(cpu_subsys.dev_root);
575 576 577 578 579 580 581 582
	if (ret)
		return ret;

	latency_notifier_init(&cpuidle_latency_notifier);

	return 0;
}

583
module_param(off, int, 0444);
584
core_initcall(cpuidle_init);