1 // SPDX-License-Identifier: GPL-2.0
2 /*
3  *  thermal.c - Generic Thermal Management Sysfs support.
4  *
5  *  Copyright (C) 2008 Intel Corp
6  *  Copyright (C) 2008 Zhang Rui <rui.zhang@intel.com>
7  *  Copyright (C) 2008 Sujith Thomas <sujith.thomas@intel.com>
8  */
9 
10 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
11 
12 #include <linux/module.h>
13 #include <linux/device.h>
14 #include <linux/err.h>
15 #include <linux/slab.h>
16 #include <linux/kdev_t.h>
17 #include <linux/idr.h>
18 #include <linux/thermal.h>
19 #include <linux/reboot.h>
20 #include <linux/string.h>
21 #include <linux/of.h>
22 #include <net/netlink.h>
23 #include <net/genetlink.h>
24 #include <linux/suspend.h>
25 
26 #define CREATE_TRACE_POINTS
27 #include <trace/events/thermal.h>
28 
29 #include "thermal_core.h"
30 #include "thermal_hwmon.h"
31 
32 MODULE_AUTHOR("Zhang Rui");
33 MODULE_DESCRIPTION("Generic thermal management sysfs support");
34 MODULE_LICENSE("GPL v2");
35 
36 static DEFINE_IDA(thermal_tz_ida);
37 static DEFINE_IDA(thermal_cdev_ida);
38 
39 static LIST_HEAD(thermal_tz_list);
40 static LIST_HEAD(thermal_cdev_list);
41 static LIST_HEAD(thermal_governor_list);
42 
43 static DEFINE_MUTEX(thermal_list_lock);
44 static DEFINE_MUTEX(thermal_governor_lock);
45 static DEFINE_MUTEX(poweroff_lock);
46 
47 static atomic_t in_suspend;
48 static bool power_off_triggered;
49 
50 static struct thermal_governor *def_governor;
51 
52 /*
53  * Governor section: set of functions to handle thermal governors
54  *
55  * Functions to help in the life cycle of thermal governors within
56  * the thermal core and by the thermal governor code.
57  */
58 
59 static struct thermal_governor *__find_governor(const char *name)
60 {
61 	struct thermal_governor *pos;
62 
63 	if (!name || !name[0])
64 		return def_governor;
65 
66 	list_for_each_entry(pos, &thermal_governor_list, governor_list)
67 		if (!strncasecmp(name, pos->name, THERMAL_NAME_LENGTH))
68 			return pos;
69 
70 	return NULL;
71 }
72 
73 /**
74  * bind_previous_governor() - bind the previous governor of the thermal zone
75  * @tz:		a valid pointer to a struct thermal_zone_device
76  * @failed_gov_name:	the name of the governor that failed to register
77  *
78  * Register the previous governor of the thermal zone after a new
79  * governor has failed to be bound.
80  */
81 static void bind_previous_governor(struct thermal_zone_device *tz,
82 				   const char *failed_gov_name)
83 {
84 	if (tz->governor && tz->governor->bind_to_tz) {
85 		if (tz->governor->bind_to_tz(tz)) {
86 			dev_err(&tz->device,
87 				"governor %s failed to bind and the previous one (%s) failed to bind again, thermal zone %s has no governor\n",
88 				failed_gov_name, tz->governor->name, tz->type);
89 			tz->governor = NULL;
90 		}
91 	}
92 }
93 
94 /**
95  * thermal_set_governor() - Switch to another governor
96  * @tz:		a valid pointer to a struct thermal_zone_device
97  * @new_gov:	pointer to the new governor
98  *
99  * Change the governor of thermal zone @tz.
100  *
101  * Return: 0 on success, an error if the new governor's bind_to_tz() failed.
102  */
103 static int thermal_set_governor(struct thermal_zone_device *tz,
104 				struct thermal_governor *new_gov)
105 {
106 	int ret = 0;
107 
108 	if (tz->governor && tz->governor->unbind_from_tz)
109 		tz->governor->unbind_from_tz(tz);
110 
111 	if (new_gov && new_gov->bind_to_tz) {
112 		ret = new_gov->bind_to_tz(tz);
113 		if (ret) {
114 			bind_previous_governor(tz, new_gov->name);
115 
116 			return ret;
117 		}
118 	}
119 
120 	tz->governor = new_gov;
121 
122 	return ret;
123 }
124 
125 int thermal_register_governor(struct thermal_governor *governor)
126 {
127 	int err;
128 	const char *name;
129 	struct thermal_zone_device *pos;
130 
131 	if (!governor)
132 		return -EINVAL;
133 
134 	mutex_lock(&thermal_governor_lock);
135 
136 	err = -EBUSY;
137 	if (!__find_governor(governor->name)) {
138 		bool match_default;
139 
140 		err = 0;
141 		list_add(&governor->governor_list, &thermal_governor_list);
142 		match_default = !strncmp(governor->name,
143 					 DEFAULT_THERMAL_GOVERNOR,
144 					 THERMAL_NAME_LENGTH);
145 
146 		if (!def_governor && match_default)
147 			def_governor = governor;
148 	}
149 
150 	mutex_lock(&thermal_list_lock);
151 
152 	list_for_each_entry(pos, &thermal_tz_list, node) {
153 		/*
154 		 * only thermal zones with specified tz->tzp->governor_name
155 		 * may run with tz->govenor unset
156 		 */
157 		if (pos->governor)
158 			continue;
159 
160 		name = pos->tzp->governor_name;
161 
162 		if (!strncasecmp(name, governor->name, THERMAL_NAME_LENGTH)) {
163 			int ret;
164 
165 			ret = thermal_set_governor(pos, governor);
166 			if (ret)
167 				dev_err(&pos->device,
168 					"Failed to set governor %s for thermal zone %s: %d\n",
169 					governor->name, pos->type, ret);
170 		}
171 	}
172 
173 	mutex_unlock(&thermal_list_lock);
174 	mutex_unlock(&thermal_governor_lock);
175 
176 	return err;
177 }
178 
179 void thermal_unregister_governor(struct thermal_governor *governor)
180 {
181 	struct thermal_zone_device *pos;
182 
183 	if (!governor)
184 		return;
185 
186 	mutex_lock(&thermal_governor_lock);
187 
188 	if (!__find_governor(governor->name))
189 		goto exit;
190 
191 	mutex_lock(&thermal_list_lock);
192 
193 	list_for_each_entry(pos, &thermal_tz_list, node) {
194 		if (!strncasecmp(pos->governor->name, governor->name,
195 				 THERMAL_NAME_LENGTH))
196 			thermal_set_governor(pos, NULL);
197 	}
198 
199 	mutex_unlock(&thermal_list_lock);
200 	list_del(&governor->governor_list);
201 exit:
202 	mutex_unlock(&thermal_governor_lock);
203 }
204 
205 int thermal_zone_device_set_policy(struct thermal_zone_device *tz,
206 				   char *policy)
207 {
208 	struct thermal_governor *gov;
209 	int ret = -EINVAL;
210 
211 	mutex_lock(&thermal_governor_lock);
212 	mutex_lock(&tz->lock);
213 
214 	gov = __find_governor(strim(policy));
215 	if (!gov)
216 		goto exit;
217 
218 	ret = thermal_set_governor(tz, gov);
219 
220 exit:
221 	mutex_unlock(&tz->lock);
222 	mutex_unlock(&thermal_governor_lock);
223 
224 	return ret;
225 }
226 
227 int thermal_build_list_of_policies(char *buf)
228 {
229 	struct thermal_governor *pos;
230 	ssize_t count = 0;
231 	ssize_t size = PAGE_SIZE;
232 
233 	mutex_lock(&thermal_governor_lock);
234 
235 	list_for_each_entry(pos, &thermal_governor_list, governor_list) {
236 		size = PAGE_SIZE - count;
237 		count += scnprintf(buf + count, size, "%s ", pos->name);
238 	}
239 	count += scnprintf(buf + count, size, "\n");
240 
241 	mutex_unlock(&thermal_governor_lock);
242 
243 	return count;
244 }
245 
246 static int __init thermal_register_governors(void)
247 {
248 	int result;
249 
250 	result = thermal_gov_step_wise_register();
251 	if (result)
252 		return result;
253 
254 	result = thermal_gov_fair_share_register();
255 	if (result)
256 		return result;
257 
258 	result = thermal_gov_bang_bang_register();
259 	if (result)
260 		return result;
261 
262 	result = thermal_gov_user_space_register();
263 	if (result)
264 		return result;
265 
266 	return thermal_gov_power_allocator_register();
267 }
268 
269 static void __init thermal_unregister_governors(void)
270 {
271 	thermal_gov_step_wise_unregister();
272 	thermal_gov_fair_share_unregister();
273 	thermal_gov_bang_bang_unregister();
274 	thermal_gov_user_space_unregister();
275 	thermal_gov_power_allocator_unregister();
276 }
277 
278 /*
279  * Zone update section: main control loop applied to each zone while monitoring
280  *
281  * in polling mode. The monitoring is done using a workqueue.
282  * Same update may be done on a zone by calling thermal_zone_device_update().
283  *
284  * An update means:
285  * - Non-critical trips will invoke the governor responsible for that zone;
286  * - Hot trips will produce a notification to userspace;
287  * - Critical trip point will cause a system shutdown.
288  */
289 static void thermal_zone_device_set_polling(struct thermal_zone_device *tz,
290 					    int delay)
291 {
292 	if (delay > 1000)
293 		mod_delayed_work(system_freezable_power_efficient_wq,
294 				 &tz->poll_queue,
295 				 round_jiffies(msecs_to_jiffies(delay)));
296 	else if (delay)
297 		mod_delayed_work(system_freezable_power_efficient_wq,
298 				 &tz->poll_queue,
299 				 msecs_to_jiffies(delay));
300 	else
301 		cancel_delayed_work(&tz->poll_queue);
302 }
303 
304 static void monitor_thermal_zone(struct thermal_zone_device *tz)
305 {
306 	mutex_lock(&tz->lock);
307 
308 	if (tz->passive)
309 		thermal_zone_device_set_polling(tz, tz->passive_delay);
310 	else if (tz->polling_delay)
311 		thermal_zone_device_set_polling(tz, tz->polling_delay);
312 	else
313 		thermal_zone_device_set_polling(tz, 0);
314 
315 	mutex_unlock(&tz->lock);
316 }
317 
318 static void handle_non_critical_trips(struct thermal_zone_device *tz, int trip)
319 {
320 	tz->governor ? tz->governor->throttle(tz, trip) :
321 		       def_governor->throttle(tz, trip);
322 }
323 
324 /**
325  * thermal_emergency_poweroff_func - emergency poweroff work after a known delay
326  * @work: work_struct associated with the emergency poweroff function
327  *
328  * This function is called in very critical situations to force
329  * a kernel poweroff after a configurable timeout value.
330  */
331 static void thermal_emergency_poweroff_func(struct work_struct *work)
332 {
333 	/*
334 	 * We have reached here after the emergency thermal shutdown
335 	 * Waiting period has expired. This means orderly_poweroff has
336 	 * not been able to shut off the system for some reason.
337 	 * Try to shut down the system immediately using kernel_power_off
338 	 * if populated
339 	 */
340 	WARN(1, "Attempting kernel_power_off: Temperature too high\n");
341 	kernel_power_off();
342 
343 	/*
344 	 * Worst of the worst case trigger emergency restart
345 	 */
346 	WARN(1, "Attempting emergency_restart: Temperature too high\n");
347 	emergency_restart();
348 }
349 
350 static DECLARE_DELAYED_WORK(thermal_emergency_poweroff_work,
351 			    thermal_emergency_poweroff_func);
352 
353 /**
354  * thermal_emergency_poweroff - Trigger an emergency system poweroff
355  *
356  * This may be called from any critical situation to trigger a system shutdown
357  * after a known period of time. By default this is not scheduled.
358  */
359 static void thermal_emergency_poweroff(void)
360 {
361 	int poweroff_delay_ms = CONFIG_THERMAL_EMERGENCY_POWEROFF_DELAY_MS;
362 	/*
363 	 * poweroff_delay_ms must be a carefully profiled positive value.
364 	 * Its a must for thermal_emergency_poweroff_work to be scheduled
365 	 */
366 	if (poweroff_delay_ms <= 0)
367 		return;
368 	schedule_delayed_work(&thermal_emergency_poweroff_work,
369 			      msecs_to_jiffies(poweroff_delay_ms));
370 }
371 
372 static void handle_critical_trips(struct thermal_zone_device *tz,
373 				  int trip, enum thermal_trip_type trip_type)
374 {
375 	int trip_temp;
376 
377 	tz->ops->get_trip_temp(tz, trip, &trip_temp);
378 
379 	/* If we have not crossed the trip_temp, we do not care. */
380 	if (trip_temp <= 0 || tz->temperature < trip_temp)
381 		return;
382 
383 	trace_thermal_zone_trip(tz, trip, trip_type);
384 
385 	if (tz->ops->notify)
386 		tz->ops->notify(tz, trip, trip_type);
387 
388 	if (trip_type == THERMAL_TRIP_CRITICAL) {
389 		dev_emerg(&tz->device,
390 			  "critical temperature reached (%d C), shutting down\n",
391 			  tz->temperature / 1000);
392 		mutex_lock(&poweroff_lock);
393 		if (!power_off_triggered) {
394 			/*
395 			 * Queue a backup emergency shutdown in the event of
396 			 * orderly_poweroff failure
397 			 */
398 			thermal_emergency_poweroff();
399 			orderly_poweroff(true);
400 			power_off_triggered = true;
401 		}
402 		mutex_unlock(&poweroff_lock);
403 	}
404 }
405 
406 static void handle_thermal_trip(struct thermal_zone_device *tz, int trip)
407 {
408 	enum thermal_trip_type type;
409 
410 	/* Ignore disabled trip points */
411 	if (test_bit(trip, &tz->trips_disabled))
412 		return;
413 
414 	tz->ops->get_trip_type(tz, trip, &type);
415 
416 	if (type == THERMAL_TRIP_CRITICAL || type == THERMAL_TRIP_HOT)
417 		handle_critical_trips(tz, trip, type);
418 	else
419 		handle_non_critical_trips(tz, trip);
420 	/*
421 	 * Alright, we handled this trip successfully.
422 	 * So, start monitoring again.
423 	 */
424 	monitor_thermal_zone(tz);
425 }
426 
427 static void update_temperature(struct thermal_zone_device *tz)
428 {
429 	int temp, ret;
430 
431 	ret = thermal_zone_get_temp(tz, &temp);
432 	if (ret) {
433 		if (ret != -EAGAIN)
434 			dev_warn(&tz->device,
435 				 "failed to read out thermal zone (%d)\n",
436 				 ret);
437 		return;
438 	}
439 
440 	mutex_lock(&tz->lock);
441 	tz->last_temperature = tz->temperature;
442 	tz->temperature = temp;
443 	mutex_unlock(&tz->lock);
444 
445 	trace_thermal_temperature(tz);
446 	if (tz->last_temperature == THERMAL_TEMP_INVALID)
447 		dev_dbg(&tz->device, "last_temperature N/A, current_temperature=%d\n",
448 			tz->temperature);
449 	else
450 		dev_dbg(&tz->device, "last_temperature=%d, current_temperature=%d\n",
451 			tz->last_temperature, tz->temperature);
452 }
453 
454 static void thermal_zone_device_init(struct thermal_zone_device *tz)
455 {
456 	struct thermal_instance *pos;
457 	tz->temperature = THERMAL_TEMP_INVALID;
458 	list_for_each_entry(pos, &tz->thermal_instances, tz_node)
459 		pos->initialized = false;
460 }
461 
462 static void thermal_zone_device_reset(struct thermal_zone_device *tz)
463 {
464 	tz->passive = 0;
465 	thermal_zone_device_init(tz);
466 }
467 
468 void thermal_zone_device_update(struct thermal_zone_device *tz,
469 				enum thermal_notify_event event)
470 {
471 	int count;
472 
473 	if (atomic_read(&in_suspend))
474 		return;
475 
476 	if (!tz->ops->get_temp)
477 		return;
478 
479 	update_temperature(tz);
480 
481 	thermal_zone_set_trips(tz);
482 
483 	tz->notify_event = event;
484 
485 	for (count = 0; count < tz->trips; count++)
486 		handle_thermal_trip(tz, count);
487 }
488 EXPORT_SYMBOL_GPL(thermal_zone_device_update);
489 
490 /**
491  * thermal_notify_framework - Sensor drivers use this API to notify framework
492  * @tz:		thermal zone device
493  * @trip:	indicates which trip point has been crossed
494  *
495  * This function handles the trip events from sensor drivers. It starts
496  * throttling the cooling devices according to the policy configured.
497  * For CRITICAL and HOT trip points, this notifies the respective drivers,
498  * and does actual throttling for other trip points i.e ACTIVE and PASSIVE.
499  * The throttling policy is based on the configured platform data; if no
500  * platform data is provided, this uses the step_wise throttling policy.
501  */
502 void thermal_notify_framework(struct thermal_zone_device *tz, int trip)
503 {
504 	handle_thermal_trip(tz, trip);
505 }
506 EXPORT_SYMBOL_GPL(thermal_notify_framework);
507 
508 static void thermal_zone_device_check(struct work_struct *work)
509 {
510 	struct thermal_zone_device *tz = container_of(work, struct
511 						      thermal_zone_device,
512 						      poll_queue.work);
513 	thermal_zone_device_update(tz, THERMAL_EVENT_UNSPECIFIED);
514 }
515 
516 /*
517  * Power actor section: interface to power actors to estimate power
518  *
519  * Set of functions used to interact to cooling devices that know
520  * how to estimate their devices power consumption.
521  */
522 
523 /**
524  * power_actor_get_max_power() - get the maximum power that a cdev can consume
525  * @cdev:	pointer to &thermal_cooling_device
526  * @tz:		a valid thermal zone device pointer
527  * @max_power:	pointer in which to store the maximum power
528  *
529  * Calculate the maximum power consumption in milliwats that the
530  * cooling device can currently consume and store it in @max_power.
531  *
532  * Return: 0 on success, -EINVAL if @cdev doesn't support the
533  * power_actor API or -E* on other error.
534  */
535 int power_actor_get_max_power(struct thermal_cooling_device *cdev,
536 			      struct thermal_zone_device *tz, u32 *max_power)
537 {
538 	if (!cdev_is_power_actor(cdev))
539 		return -EINVAL;
540 
541 	return cdev->ops->state2power(cdev, tz, 0, max_power);
542 }
543 
544 /**
545  * power_actor_get_min_power() - get the mainimum power that a cdev can consume
546  * @cdev:	pointer to &thermal_cooling_device
547  * @tz:		a valid thermal zone device pointer
548  * @min_power:	pointer in which to store the minimum power
549  *
550  * Calculate the minimum power consumption in milliwatts that the
551  * cooling device can currently consume and store it in @min_power.
552  *
553  * Return: 0 on success, -EINVAL if @cdev doesn't support the
554  * power_actor API or -E* on other error.
555  */
556 int power_actor_get_min_power(struct thermal_cooling_device *cdev,
557 			      struct thermal_zone_device *tz, u32 *min_power)
558 {
559 	unsigned long max_state;
560 	int ret;
561 
562 	if (!cdev_is_power_actor(cdev))
563 		return -EINVAL;
564 
565 	ret = cdev->ops->get_max_state(cdev, &max_state);
566 	if (ret)
567 		return ret;
568 
569 	return cdev->ops->state2power(cdev, tz, max_state, min_power);
570 }
571 
572 /**
573  * power_actor_set_power() - limit the maximum power a cooling device consumes
574  * @cdev:	pointer to &thermal_cooling_device
575  * @instance:	thermal instance to update
576  * @power:	the power in milliwatts
577  *
578  * Set the cooling device to consume at most @power milliwatts. The limit is
579  * expected to be a cap at the maximum power consumption.
580  *
581  * Return: 0 on success, -EINVAL if the cooling device does not
582  * implement the power actor API or -E* for other failures.
583  */
584 int power_actor_set_power(struct thermal_cooling_device *cdev,
585 			  struct thermal_instance *instance, u32 power)
586 {
587 	unsigned long state;
588 	int ret;
589 
590 	if (!cdev_is_power_actor(cdev))
591 		return -EINVAL;
592 
593 	ret = cdev->ops->power2state(cdev, instance->tz, power, &state);
594 	if (ret)
595 		return ret;
596 
597 	instance->target = state;
598 	mutex_lock(&cdev->lock);
599 	cdev->updated = false;
600 	mutex_unlock(&cdev->lock);
601 	thermal_cdev_update(cdev);
602 
603 	return 0;
604 }
605 
606 void thermal_zone_device_rebind_exception(struct thermal_zone_device *tz,
607 					  const char *cdev_type, size_t size)
608 {
609 	struct thermal_cooling_device *cdev = NULL;
610 
611 	mutex_lock(&thermal_list_lock);
612 	list_for_each_entry(cdev, &thermal_cdev_list, node) {
613 		/* skip non matching cdevs */
614 		if (strncmp(cdev_type, cdev->type, size))
615 			continue;
616 
617 		/* re binding the exception matching the type pattern */
618 		thermal_zone_bind_cooling_device(tz, THERMAL_TRIPS_NONE, cdev,
619 						 THERMAL_NO_LIMIT,
620 						 THERMAL_NO_LIMIT,
621 						 THERMAL_WEIGHT_DEFAULT);
622 	}
623 	mutex_unlock(&thermal_list_lock);
624 }
625 
626 void thermal_zone_device_unbind_exception(struct thermal_zone_device *tz,
627 					  const char *cdev_type, size_t size)
628 {
629 	struct thermal_cooling_device *cdev = NULL;
630 
631 	mutex_lock(&thermal_list_lock);
632 	list_for_each_entry(cdev, &thermal_cdev_list, node) {
633 		/* skip non matching cdevs */
634 		if (strncmp(cdev_type, cdev->type, size))
635 			continue;
636 		/* unbinding the exception matching the type pattern */
637 		thermal_zone_unbind_cooling_device(tz, THERMAL_TRIPS_NONE,
638 						   cdev);
639 	}
640 	mutex_unlock(&thermal_list_lock);
641 }
642 
643 /*
644  * Device management section: cooling devices, zones devices, and binding
645  *
646  * Set of functions provided by the thermal core for:
647  * - cooling devices lifecycle: registration, unregistration,
648  *				binding, and unbinding.
649  * - thermal zone devices lifecycle: registration, unregistration,
650  *				     binding, and unbinding.
651  */
652 
653 /**
654  * thermal_zone_bind_cooling_device() - bind a cooling device to a thermal zone
655  * @tz:		pointer to struct thermal_zone_device
656  * @trip:	indicates which trip point the cooling devices is
657  *		associated with in this thermal zone.
658  * @cdev:	pointer to struct thermal_cooling_device
659  * @upper:	the Maximum cooling state for this trip point.
660  *		THERMAL_NO_LIMIT means no upper limit,
661  *		and the cooling device can be in max_state.
662  * @lower:	the Minimum cooling state can be used for this trip point.
663  *		THERMAL_NO_LIMIT means no lower limit,
664  *		and the cooling device can be in cooling state 0.
665  * @weight:	The weight of the cooling device to be bound to the
666  *		thermal zone. Use THERMAL_WEIGHT_DEFAULT for the
667  *		default value
668  *
669  * This interface function bind a thermal cooling device to the certain trip
670  * point of a thermal zone device.
671  * This function is usually called in the thermal zone device .bind callback.
672  *
673  * Return: 0 on success, the proper error value otherwise.
674  */
675 int thermal_zone_bind_cooling_device(struct thermal_zone_device *tz,
676 				     int trip,
677 				     struct thermal_cooling_device *cdev,
678 				     unsigned long upper, unsigned long lower,
679 				     unsigned int weight)
680 {
681 	struct thermal_instance *dev;
682 	struct thermal_instance *pos;
683 	struct thermal_zone_device *pos1;
684 	struct thermal_cooling_device *pos2;
685 	unsigned long max_state;
686 	int result, ret;
687 
688 	if (trip >= tz->trips || (trip < 0 && trip != THERMAL_TRIPS_NONE))
689 		return -EINVAL;
690 
691 	list_for_each_entry(pos1, &thermal_tz_list, node) {
692 		if (pos1 == tz)
693 			break;
694 	}
695 	list_for_each_entry(pos2, &thermal_cdev_list, node) {
696 		if (pos2 == cdev)
697 			break;
698 	}
699 
700 	if (tz != pos1 || cdev != pos2)
701 		return -EINVAL;
702 
703 	ret = cdev->ops->get_max_state(cdev, &max_state);
704 	if (ret)
705 		return ret;
706 
707 	/* lower default 0, upper default max_state */
708 	lower = lower == THERMAL_NO_LIMIT ? 0 : lower;
709 	upper = upper == THERMAL_NO_LIMIT ? max_state : upper;
710 
711 	if (lower > upper || upper > max_state)
712 		return -EINVAL;
713 
714 	dev = kzalloc(sizeof(*dev), GFP_KERNEL);
715 	if (!dev)
716 		return -ENOMEM;
717 	dev->tz = tz;
718 	dev->cdev = cdev;
719 	dev->trip = trip;
720 	dev->upper = upper;
721 	dev->lower = lower;
722 	dev->target = THERMAL_NO_TARGET;
723 	dev->weight = weight;
724 
725 	result = ida_simple_get(&tz->ida, 0, 0, GFP_KERNEL);
726 	if (result < 0)
727 		goto free_mem;
728 
729 	dev->id = result;
730 	sprintf(dev->name, "cdev%d", dev->id);
731 	result =
732 	    sysfs_create_link(&tz->device.kobj, &cdev->device.kobj, dev->name);
733 	if (result)
734 		goto release_ida;
735 
736 	sprintf(dev->attr_name, "cdev%d_trip_point", dev->id);
737 	sysfs_attr_init(&dev->attr.attr);
738 	dev->attr.attr.name = dev->attr_name;
739 	dev->attr.attr.mode = 0444;
740 	dev->attr.show = trip_point_show;
741 	result = device_create_file(&tz->device, &dev->attr);
742 	if (result)
743 		goto remove_symbol_link;
744 
745 	sprintf(dev->weight_attr_name, "cdev%d_weight", dev->id);
746 	sysfs_attr_init(&dev->weight_attr.attr);
747 	dev->weight_attr.attr.name = dev->weight_attr_name;
748 	dev->weight_attr.attr.mode = S_IWUSR | S_IRUGO;
749 	dev->weight_attr.show = weight_show;
750 	dev->weight_attr.store = weight_store;
751 	result = device_create_file(&tz->device, &dev->weight_attr);
752 	if (result)
753 		goto remove_trip_file;
754 
755 	mutex_lock(&tz->lock);
756 	mutex_lock(&cdev->lock);
757 	list_for_each_entry(pos, &tz->thermal_instances, tz_node)
758 		if (pos->tz == tz && pos->trip == trip && pos->cdev == cdev) {
759 			result = -EEXIST;
760 			break;
761 		}
762 	if (!result) {
763 		list_add_tail(&dev->tz_node, &tz->thermal_instances);
764 		list_add_tail(&dev->cdev_node, &cdev->thermal_instances);
765 		atomic_set(&tz->need_update, 1);
766 	}
767 	mutex_unlock(&cdev->lock);
768 	mutex_unlock(&tz->lock);
769 
770 	if (!result)
771 		return 0;
772 
773 	device_remove_file(&tz->device, &dev->weight_attr);
774 remove_trip_file:
775 	device_remove_file(&tz->device, &dev->attr);
776 remove_symbol_link:
777 	sysfs_remove_link(&tz->device.kobj, dev->name);
778 release_ida:
779 	ida_simple_remove(&tz->ida, dev->id);
780 free_mem:
781 	kfree(dev);
782 	return result;
783 }
784 EXPORT_SYMBOL_GPL(thermal_zone_bind_cooling_device);
785 
786 /**
787  * thermal_zone_unbind_cooling_device() - unbind a cooling device from a
788  *					  thermal zone.
789  * @tz:		pointer to a struct thermal_zone_device.
790  * @trip:	indicates which trip point the cooling devices is
791  *		associated with in this thermal zone.
792  * @cdev:	pointer to a struct thermal_cooling_device.
793  *
794  * This interface function unbind a thermal cooling device from the certain
795  * trip point of a thermal zone device.
796  * This function is usually called in the thermal zone device .unbind callback.
797  *
798  * Return: 0 on success, the proper error value otherwise.
799  */
800 int thermal_zone_unbind_cooling_device(struct thermal_zone_device *tz,
801 				       int trip,
802 				       struct thermal_cooling_device *cdev)
803 {
804 	struct thermal_instance *pos, *next;
805 
806 	mutex_lock(&tz->lock);
807 	mutex_lock(&cdev->lock);
808 	list_for_each_entry_safe(pos, next, &tz->thermal_instances, tz_node) {
809 		if (pos->tz == tz && pos->trip == trip && pos->cdev == cdev) {
810 			list_del(&pos->tz_node);
811 			list_del(&pos->cdev_node);
812 			mutex_unlock(&cdev->lock);
813 			mutex_unlock(&tz->lock);
814 			goto unbind;
815 		}
816 	}
817 	mutex_unlock(&cdev->lock);
818 	mutex_unlock(&tz->lock);
819 
820 	return -ENODEV;
821 
822 unbind:
823 	device_remove_file(&tz->device, &pos->weight_attr);
824 	device_remove_file(&tz->device, &pos->attr);
825 	sysfs_remove_link(&tz->device.kobj, pos->name);
826 	ida_simple_remove(&tz->ida, pos->id);
827 	kfree(pos);
828 	return 0;
829 }
830 EXPORT_SYMBOL_GPL(thermal_zone_unbind_cooling_device);
831 
832 static void thermal_release(struct device *dev)
833 {
834 	struct thermal_zone_device *tz;
835 	struct thermal_cooling_device *cdev;
836 
837 	if (!strncmp(dev_name(dev), "thermal_zone",
838 		     sizeof("thermal_zone") - 1)) {
839 		tz = to_thermal_zone(dev);
840 		thermal_zone_destroy_device_groups(tz);
841 		kfree(tz);
842 	} else if (!strncmp(dev_name(dev), "cooling_device",
843 			    sizeof("cooling_device") - 1)) {
844 		cdev = to_cooling_device(dev);
845 		kfree(cdev);
846 	}
847 }
848 
849 static struct class thermal_class = {
850 	.name = "thermal",
851 	.dev_release = thermal_release,
852 };
853 
854 static inline
855 void print_bind_err_msg(struct thermal_zone_device *tz,
856 			struct thermal_cooling_device *cdev, int ret)
857 {
858 	dev_err(&tz->device, "binding zone %s with cdev %s failed:%d\n",
859 		tz->type, cdev->type, ret);
860 }
861 
862 static void __bind(struct thermal_zone_device *tz, int mask,
863 		   struct thermal_cooling_device *cdev,
864 		   unsigned long *limits,
865 		   unsigned int weight)
866 {
867 	int i, ret;
868 
869 	for (i = 0; i < tz->trips; i++) {
870 		if (mask & (1 << i)) {
871 			unsigned long upper, lower;
872 
873 			upper = THERMAL_NO_LIMIT;
874 			lower = THERMAL_NO_LIMIT;
875 			if (limits) {
876 				lower = limits[i * 2];
877 				upper = limits[i * 2 + 1];
878 			}
879 			ret = thermal_zone_bind_cooling_device(tz, i, cdev,
880 							       upper, lower,
881 							       weight);
882 			if (ret)
883 				print_bind_err_msg(tz, cdev, ret);
884 		}
885 	}
886 }
887 
888 static void bind_cdev(struct thermal_cooling_device *cdev)
889 {
890 	int i, ret;
891 	const struct thermal_zone_params *tzp;
892 	struct thermal_zone_device *pos = NULL;
893 
894 	mutex_lock(&thermal_list_lock);
895 
896 	list_for_each_entry(pos, &thermal_tz_list, node) {
897 		if (!pos->tzp && !pos->ops->bind)
898 			continue;
899 
900 		if (pos->ops->bind) {
901 			ret = pos->ops->bind(pos, cdev);
902 			if (ret)
903 				print_bind_err_msg(pos, cdev, ret);
904 			continue;
905 		}
906 
907 		tzp = pos->tzp;
908 		if (!tzp || !tzp->tbp)
909 			continue;
910 
911 		for (i = 0; i < tzp->num_tbps; i++) {
912 			if (tzp->tbp[i].cdev || !tzp->tbp[i].match)
913 				continue;
914 			if (tzp->tbp[i].match(pos, cdev))
915 				continue;
916 			tzp->tbp[i].cdev = cdev;
917 			__bind(pos, tzp->tbp[i].trip_mask, cdev,
918 			       tzp->tbp[i].binding_limits,
919 			       tzp->tbp[i].weight);
920 		}
921 	}
922 
923 	mutex_unlock(&thermal_list_lock);
924 }
925 
926 /**
927  * __thermal_cooling_device_register() - register a new thermal cooling device
928  * @np:		a pointer to a device tree node.
929  * @type:	the thermal cooling device type.
930  * @devdata:	device private data.
931  * @ops:		standard thermal cooling devices callbacks.
932  *
933  * This interface function adds a new thermal cooling device (fan/processor/...)
934  * to /sys/class/thermal/ folder as cooling_device[0-*]. It tries to bind itself
935  * to all the thermal zone devices registered at the same time.
936  * It also gives the opportunity to link the cooling device to a device tree
937  * node, so that it can be bound to a thermal zone created out of device tree.
938  *
939  * Return: a pointer to the created struct thermal_cooling_device or an
940  * ERR_PTR. Caller must check return value with IS_ERR*() helpers.
941  */
942 static struct thermal_cooling_device *
943 __thermal_cooling_device_register(struct device_node *np,
944 				  const char *type, void *devdata,
945 				  const struct thermal_cooling_device_ops *ops)
946 {
947 	struct thermal_cooling_device *cdev;
948 	struct thermal_zone_device *pos = NULL;
949 	int result;
950 
951 	if (type && strlen(type) >= THERMAL_NAME_LENGTH)
952 		return ERR_PTR(-EINVAL);
953 
954 	if (!ops || !ops->get_max_state || !ops->get_cur_state ||
955 	    !ops->set_cur_state)
956 		return ERR_PTR(-EINVAL);
957 
958 	cdev = kzalloc(sizeof(*cdev), GFP_KERNEL);
959 	if (!cdev)
960 		return ERR_PTR(-ENOMEM);
961 
962 	result = ida_simple_get(&thermal_cdev_ida, 0, 0, GFP_KERNEL);
963 	if (result < 0) {
964 		kfree(cdev);
965 		return ERR_PTR(result);
966 	}
967 
968 	cdev->id = result;
969 	strlcpy(cdev->type, type ? : "", sizeof(cdev->type));
970 	mutex_init(&cdev->lock);
971 	INIT_LIST_HEAD(&cdev->thermal_instances);
972 	cdev->np = np;
973 	cdev->ops = ops;
974 	cdev->updated = false;
975 	cdev->device.class = &thermal_class;
976 	cdev->devdata = devdata;
977 	thermal_cooling_device_setup_sysfs(cdev);
978 	dev_set_name(&cdev->device, "cooling_device%d", cdev->id);
979 	result = device_register(&cdev->device);
980 	if (result) {
981 		ida_simple_remove(&thermal_cdev_ida, cdev->id);
982 		kfree(cdev);
983 		return ERR_PTR(result);
984 	}
985 
986 	/* Add 'this' new cdev to the global cdev list */
987 	mutex_lock(&thermal_list_lock);
988 	list_add(&cdev->node, &thermal_cdev_list);
989 	mutex_unlock(&thermal_list_lock);
990 
991 	/* Update binding information for 'this' new cdev */
992 	bind_cdev(cdev);
993 
994 	mutex_lock(&thermal_list_lock);
995 	list_for_each_entry(pos, &thermal_tz_list, node)
996 		if (atomic_cmpxchg(&pos->need_update, 1, 0))
997 			thermal_zone_device_update(pos,
998 						   THERMAL_EVENT_UNSPECIFIED);
999 	mutex_unlock(&thermal_list_lock);
1000 
1001 	return cdev;
1002 }
1003 
1004 /**
1005  * thermal_cooling_device_register() - register a new thermal cooling device
1006  * @type:	the thermal cooling device type.
1007  * @devdata:	device private data.
1008  * @ops:		standard thermal cooling devices callbacks.
1009  *
1010  * This interface function adds a new thermal cooling device (fan/processor/...)
1011  * to /sys/class/thermal/ folder as cooling_device[0-*]. It tries to bind itself
1012  * to all the thermal zone devices registered at the same time.
1013  *
1014  * Return: a pointer to the created struct thermal_cooling_device or an
1015  * ERR_PTR. Caller must check return value with IS_ERR*() helpers.
1016  */
1017 struct thermal_cooling_device *
1018 thermal_cooling_device_register(const char *type, void *devdata,
1019 				const struct thermal_cooling_device_ops *ops)
1020 {
1021 	return __thermal_cooling_device_register(NULL, type, devdata, ops);
1022 }
1023 EXPORT_SYMBOL_GPL(thermal_cooling_device_register);
1024 
1025 /**
1026  * thermal_of_cooling_device_register() - register an OF thermal cooling device
1027  * @np:		a pointer to a device tree node.
1028  * @type:	the thermal cooling device type.
1029  * @devdata:	device private data.
1030  * @ops:		standard thermal cooling devices callbacks.
1031  *
1032  * This function will register a cooling device with device tree node reference.
1033  * This interface function adds a new thermal cooling device (fan/processor/...)
1034  * to /sys/class/thermal/ folder as cooling_device[0-*]. It tries to bind itself
1035  * to all the thermal zone devices registered at the same time.
1036  *
1037  * Return: a pointer to the created struct thermal_cooling_device or an
1038  * ERR_PTR. Caller must check return value with IS_ERR*() helpers.
1039  */
1040 struct thermal_cooling_device *
1041 thermal_of_cooling_device_register(struct device_node *np,
1042 				   const char *type, void *devdata,
1043 				   const struct thermal_cooling_device_ops *ops)
1044 {
1045 	return __thermal_cooling_device_register(np, type, devdata, ops);
1046 }
1047 EXPORT_SYMBOL_GPL(thermal_of_cooling_device_register);
1048 
1049 static void thermal_cooling_device_release(struct device *dev, void *res)
1050 {
1051 	thermal_cooling_device_unregister(
1052 				*(struct thermal_cooling_device **)res);
1053 }
1054 
1055 /**
1056  * devm_thermal_of_cooling_device_register() - register an OF thermal cooling
1057  *					       device
1058  * @dev:	a valid struct device pointer of a sensor device.
1059  * @np:		a pointer to a device tree node.
1060  * @type:	the thermal cooling device type.
1061  * @devdata:	device private data.
1062  * @ops:	standard thermal cooling devices callbacks.
1063  *
1064  * This function will register a cooling device with device tree node reference.
1065  * This interface function adds a new thermal cooling device (fan/processor/...)
1066  * to /sys/class/thermal/ folder as cooling_device[0-*]. It tries to bind itself
1067  * to all the thermal zone devices registered at the same time.
1068  *
1069  * Return: a pointer to the created struct thermal_cooling_device or an
1070  * ERR_PTR. Caller must check return value with IS_ERR*() helpers.
1071  */
1072 struct thermal_cooling_device *
1073 devm_thermal_of_cooling_device_register(struct device *dev,
1074 				struct device_node *np,
1075 				char *type, void *devdata,
1076 				const struct thermal_cooling_device_ops *ops)
1077 {
1078 	struct thermal_cooling_device **ptr, *tcd;
1079 
1080 	ptr = devres_alloc(thermal_cooling_device_release, sizeof(*ptr),
1081 			   GFP_KERNEL);
1082 	if (!ptr)
1083 		return ERR_PTR(-ENOMEM);
1084 
1085 	tcd = __thermal_cooling_device_register(np, type, devdata, ops);
1086 	if (IS_ERR(tcd)) {
1087 		devres_free(ptr);
1088 		return tcd;
1089 	}
1090 
1091 	*ptr = tcd;
1092 	devres_add(dev, ptr);
1093 
1094 	return tcd;
1095 }
1096 EXPORT_SYMBOL_GPL(devm_thermal_of_cooling_device_register);
1097 
1098 static void __unbind(struct thermal_zone_device *tz, int mask,
1099 		     struct thermal_cooling_device *cdev)
1100 {
1101 	int i;
1102 
1103 	for (i = 0; i < tz->trips; i++)
1104 		if (mask & (1 << i))
1105 			thermal_zone_unbind_cooling_device(tz, i, cdev);
1106 }
1107 
1108 /**
1109  * thermal_cooling_device_unregister - removes a thermal cooling device
1110  * @cdev:	the thermal cooling device to remove.
1111  *
1112  * thermal_cooling_device_unregister() must be called when a registered
1113  * thermal cooling device is no longer needed.
1114  */
1115 void thermal_cooling_device_unregister(struct thermal_cooling_device *cdev)
1116 {
1117 	int i;
1118 	const struct thermal_zone_params *tzp;
1119 	struct thermal_zone_device *tz;
1120 	struct thermal_cooling_device *pos = NULL;
1121 
1122 	if (!cdev)
1123 		return;
1124 
1125 	mutex_lock(&thermal_list_lock);
1126 	list_for_each_entry(pos, &thermal_cdev_list, node)
1127 		if (pos == cdev)
1128 			break;
1129 	if (pos != cdev) {
1130 		/* thermal cooling device not found */
1131 		mutex_unlock(&thermal_list_lock);
1132 		return;
1133 	}
1134 	list_del(&cdev->node);
1135 
1136 	/* Unbind all thermal zones associated with 'this' cdev */
1137 	list_for_each_entry(tz, &thermal_tz_list, node) {
1138 		if (tz->ops->unbind) {
1139 			tz->ops->unbind(tz, cdev);
1140 			continue;
1141 		}
1142 
1143 		if (!tz->tzp || !tz->tzp->tbp)
1144 			continue;
1145 
1146 		tzp = tz->tzp;
1147 		for (i = 0; i < tzp->num_tbps; i++) {
1148 			if (tzp->tbp[i].cdev == cdev) {
1149 				__unbind(tz, tzp->tbp[i].trip_mask, cdev);
1150 				tzp->tbp[i].cdev = NULL;
1151 			}
1152 		}
1153 	}
1154 
1155 	mutex_unlock(&thermal_list_lock);
1156 
1157 	ida_simple_remove(&thermal_cdev_ida, cdev->id);
1158 	device_del(&cdev->device);
1159 	thermal_cooling_device_destroy_sysfs(cdev);
1160 	put_device(&cdev->device);
1161 }
1162 EXPORT_SYMBOL_GPL(thermal_cooling_device_unregister);
1163 
1164 static void bind_tz(struct thermal_zone_device *tz)
1165 {
1166 	int i, ret;
1167 	struct thermal_cooling_device *pos = NULL;
1168 	const struct thermal_zone_params *tzp = tz->tzp;
1169 
1170 	if (!tzp && !tz->ops->bind)
1171 		return;
1172 
1173 	mutex_lock(&thermal_list_lock);
1174 
1175 	/* If there is ops->bind, try to use ops->bind */
1176 	if (tz->ops->bind) {
1177 		list_for_each_entry(pos, &thermal_cdev_list, node) {
1178 			ret = tz->ops->bind(tz, pos);
1179 			if (ret)
1180 				print_bind_err_msg(tz, pos, ret);
1181 		}
1182 		goto exit;
1183 	}
1184 
1185 	if (!tzp || !tzp->tbp)
1186 		goto exit;
1187 
1188 	list_for_each_entry(pos, &thermal_cdev_list, node) {
1189 		for (i = 0; i < tzp->num_tbps; i++) {
1190 			if (tzp->tbp[i].cdev || !tzp->tbp[i].match)
1191 				continue;
1192 			if (tzp->tbp[i].match(tz, pos))
1193 				continue;
1194 			tzp->tbp[i].cdev = pos;
1195 			__bind(tz, tzp->tbp[i].trip_mask, pos,
1196 			       tzp->tbp[i].binding_limits,
1197 			       tzp->tbp[i].weight);
1198 		}
1199 	}
1200 exit:
1201 	mutex_unlock(&thermal_list_lock);
1202 }
1203 
1204 /**
1205  * thermal_zone_device_register() - register a new thermal zone device
1206  * @type:	the thermal zone device type
1207  * @trips:	the number of trip points the thermal zone support
1208  * @mask:	a bit string indicating the writeablility of trip points
1209  * @devdata:	private device data
1210  * @ops:	standard thermal zone device callbacks
1211  * @tzp:	thermal zone platform parameters
1212  * @passive_delay: number of milliseconds to wait between polls when
1213  *		   performing passive cooling
1214  * @polling_delay: number of milliseconds to wait between polls when checking
1215  *		   whether trip points have been crossed (0 for interrupt
1216  *		   driven systems)
1217  *
1218  * This interface function adds a new thermal zone device (sensor) to
1219  * /sys/class/thermal folder as thermal_zone[0-*]. It tries to bind all the
1220  * thermal cooling devices registered at the same time.
1221  * thermal_zone_device_unregister() must be called when the device is no
1222  * longer needed. The passive cooling depends on the .get_trend() return value.
1223  *
1224  * Return: a pointer to the created struct thermal_zone_device or an
1225  * in case of error, an ERR_PTR. Caller must check return value with
1226  * IS_ERR*() helpers.
1227  */
1228 struct thermal_zone_device *
1229 thermal_zone_device_register(const char *type, int trips, int mask,
1230 			     void *devdata, struct thermal_zone_device_ops *ops,
1231 			     struct thermal_zone_params *tzp, int passive_delay,
1232 			     int polling_delay)
1233 {
1234 	struct thermal_zone_device *tz;
1235 	enum thermal_trip_type trip_type;
1236 	int trip_temp;
1237 	int result;
1238 	int count;
1239 	struct thermal_governor *governor;
1240 
1241 	if (!type || strlen(type) == 0)
1242 		return ERR_PTR(-EINVAL);
1243 
1244 	if (type && strlen(type) >= THERMAL_NAME_LENGTH)
1245 		return ERR_PTR(-EINVAL);
1246 
1247 	if (trips > THERMAL_MAX_TRIPS || trips < 0 || mask >> trips)
1248 		return ERR_PTR(-EINVAL);
1249 
1250 	if (!ops)
1251 		return ERR_PTR(-EINVAL);
1252 
1253 	if (trips > 0 && (!ops->get_trip_type || !ops->get_trip_temp))
1254 		return ERR_PTR(-EINVAL);
1255 
1256 	tz = kzalloc(sizeof(*tz), GFP_KERNEL);
1257 	if (!tz)
1258 		return ERR_PTR(-ENOMEM);
1259 
1260 	INIT_LIST_HEAD(&tz->thermal_instances);
1261 	ida_init(&tz->ida);
1262 	mutex_init(&tz->lock);
1263 	result = ida_simple_get(&thermal_tz_ida, 0, 0, GFP_KERNEL);
1264 	if (result < 0)
1265 		goto free_tz;
1266 
1267 	tz->id = result;
1268 	strlcpy(tz->type, type, sizeof(tz->type));
1269 	tz->ops = ops;
1270 	tz->tzp = tzp;
1271 	tz->device.class = &thermal_class;
1272 	tz->devdata = devdata;
1273 	tz->trips = trips;
1274 	tz->passive_delay = passive_delay;
1275 	tz->polling_delay = polling_delay;
1276 
1277 	/* sys I/F */
1278 	/* Add nodes that are always present via .groups */
1279 	result = thermal_zone_create_device_groups(tz, mask);
1280 	if (result)
1281 		goto remove_id;
1282 
1283 	/* A new thermal zone needs to be updated anyway. */
1284 	atomic_set(&tz->need_update, 1);
1285 
1286 	dev_set_name(&tz->device, "thermal_zone%d", tz->id);
1287 	result = device_register(&tz->device);
1288 	if (result)
1289 		goto remove_device_groups;
1290 
1291 	for (count = 0; count < trips; count++) {
1292 		if (tz->ops->get_trip_type(tz, count, &trip_type))
1293 			set_bit(count, &tz->trips_disabled);
1294 		if (tz->ops->get_trip_temp(tz, count, &trip_temp))
1295 			set_bit(count, &tz->trips_disabled);
1296 		/* Check for bogus trip points */
1297 		if (trip_temp == 0)
1298 			set_bit(count, &tz->trips_disabled);
1299 	}
1300 
1301 	/* Update 'this' zone's governor information */
1302 	mutex_lock(&thermal_governor_lock);
1303 
1304 	if (tz->tzp)
1305 		governor = __find_governor(tz->tzp->governor_name);
1306 	else
1307 		governor = def_governor;
1308 
1309 	result = thermal_set_governor(tz, governor);
1310 	if (result) {
1311 		mutex_unlock(&thermal_governor_lock);
1312 		goto unregister;
1313 	}
1314 
1315 	mutex_unlock(&thermal_governor_lock);
1316 
1317 	if (!tz->tzp || !tz->tzp->no_hwmon) {
1318 		result = thermal_add_hwmon_sysfs(tz);
1319 		if (result)
1320 			goto unregister;
1321 	}
1322 
1323 	mutex_lock(&thermal_list_lock);
1324 	list_add_tail(&tz->node, &thermal_tz_list);
1325 	mutex_unlock(&thermal_list_lock);
1326 
1327 	/* Bind cooling devices for this zone */
1328 	bind_tz(tz);
1329 
1330 	INIT_DELAYED_WORK(&tz->poll_queue, thermal_zone_device_check);
1331 
1332 	thermal_zone_device_reset(tz);
1333 	/* Update the new thermal zone and mark it as already updated. */
1334 	if (atomic_cmpxchg(&tz->need_update, 1, 0))
1335 		thermal_zone_device_update(tz, THERMAL_EVENT_UNSPECIFIED);
1336 
1337 	return tz;
1338 
1339 unregister:
1340 	ida_simple_remove(&thermal_tz_ida, tz->id);
1341 	device_unregister(&tz->device);
1342 	return ERR_PTR(result);
1343 
1344 remove_device_groups:
1345 	thermal_zone_destroy_device_groups(tz);
1346 remove_id:
1347 	ida_simple_remove(&thermal_tz_ida, tz->id);
1348 free_tz:
1349 	kfree(tz);
1350 	return ERR_PTR(result);
1351 }
1352 EXPORT_SYMBOL_GPL(thermal_zone_device_register);
1353 
1354 /**
1355  * thermal_device_unregister - removes the registered thermal zone device
1356  * @tz: the thermal zone device to remove
1357  */
1358 void thermal_zone_device_unregister(struct thermal_zone_device *tz)
1359 {
1360 	int i;
1361 	const struct thermal_zone_params *tzp;
1362 	struct thermal_cooling_device *cdev;
1363 	struct thermal_zone_device *pos = NULL;
1364 
1365 	if (!tz)
1366 		return;
1367 
1368 	tzp = tz->tzp;
1369 
1370 	mutex_lock(&thermal_list_lock);
1371 	list_for_each_entry(pos, &thermal_tz_list, node)
1372 		if (pos == tz)
1373 			break;
1374 	if (pos != tz) {
1375 		/* thermal zone device not found */
1376 		mutex_unlock(&thermal_list_lock);
1377 		return;
1378 	}
1379 	list_del(&tz->node);
1380 
1381 	/* Unbind all cdevs associated with 'this' thermal zone */
1382 	list_for_each_entry(cdev, &thermal_cdev_list, node) {
1383 		if (tz->ops->unbind) {
1384 			tz->ops->unbind(tz, cdev);
1385 			continue;
1386 		}
1387 
1388 		if (!tzp || !tzp->tbp)
1389 			break;
1390 
1391 		for (i = 0; i < tzp->num_tbps; i++) {
1392 			if (tzp->tbp[i].cdev == cdev) {
1393 				__unbind(tz, tzp->tbp[i].trip_mask, cdev);
1394 				tzp->tbp[i].cdev = NULL;
1395 			}
1396 		}
1397 	}
1398 
1399 	mutex_unlock(&thermal_list_lock);
1400 
1401 	thermal_zone_device_set_polling(tz, 0);
1402 
1403 	thermal_set_governor(tz, NULL);
1404 
1405 	thermal_remove_hwmon_sysfs(tz);
1406 	ida_simple_remove(&thermal_tz_ida, tz->id);
1407 	ida_destroy(&tz->ida);
1408 	mutex_destroy(&tz->lock);
1409 	device_unregister(&tz->device);
1410 }
1411 EXPORT_SYMBOL_GPL(thermal_zone_device_unregister);
1412 
1413 /**
1414  * thermal_zone_get_zone_by_name() - search for a zone and returns its ref
1415  * @name: thermal zone name to fetch the temperature
1416  *
1417  * When only one zone is found with the passed name, returns a reference to it.
1418  *
1419  * Return: On success returns a reference to an unique thermal zone with
1420  * matching name equals to @name, an ERR_PTR otherwise (-EINVAL for invalid
1421  * paramenters, -ENODEV for not found and -EEXIST for multiple matches).
1422  */
1423 struct thermal_zone_device *thermal_zone_get_zone_by_name(const char *name)
1424 {
1425 	struct thermal_zone_device *pos = NULL, *ref = ERR_PTR(-EINVAL);
1426 	unsigned int found = 0;
1427 
1428 	if (!name)
1429 		goto exit;
1430 
1431 	mutex_lock(&thermal_list_lock);
1432 	list_for_each_entry(pos, &thermal_tz_list, node)
1433 		if (!strncasecmp(name, pos->type, THERMAL_NAME_LENGTH)) {
1434 			found++;
1435 			ref = pos;
1436 		}
1437 	mutex_unlock(&thermal_list_lock);
1438 
1439 	/* nothing has been found, thus an error code for it */
1440 	if (found == 0)
1441 		ref = ERR_PTR(-ENODEV);
1442 	else if (found > 1)
1443 	/* Success only when an unique zone is found */
1444 		ref = ERR_PTR(-EEXIST);
1445 
1446 exit:
1447 	return ref;
1448 }
1449 EXPORT_SYMBOL_GPL(thermal_zone_get_zone_by_name);
1450 
1451 #ifdef CONFIG_NET
1452 static const struct genl_multicast_group thermal_event_mcgrps[] = {
1453 	{ .name = THERMAL_GENL_MCAST_GROUP_NAME, },
1454 };
1455 
1456 static struct genl_family thermal_event_genl_family __ro_after_init = {
1457 	.module = THIS_MODULE,
1458 	.name = THERMAL_GENL_FAMILY_NAME,
1459 	.version = THERMAL_GENL_VERSION,
1460 	.maxattr = THERMAL_GENL_ATTR_MAX,
1461 	.mcgrps = thermal_event_mcgrps,
1462 	.n_mcgrps = ARRAY_SIZE(thermal_event_mcgrps),
1463 };
1464 
1465 int thermal_generate_netlink_event(struct thermal_zone_device *tz,
1466 				   enum events event)
1467 {
1468 	struct sk_buff *skb;
1469 	struct nlattr *attr;
1470 	struct thermal_genl_event *thermal_event;
1471 	void *msg_header;
1472 	int size;
1473 	int result;
1474 	static unsigned int thermal_event_seqnum;
1475 
1476 	if (!tz)
1477 		return -EINVAL;
1478 
1479 	/* allocate memory */
1480 	size = nla_total_size(sizeof(struct thermal_genl_event)) +
1481 	       nla_total_size(0);
1482 
1483 	skb = genlmsg_new(size, GFP_ATOMIC);
1484 	if (!skb)
1485 		return -ENOMEM;
1486 
1487 	/* add the genetlink message header */
1488 	msg_header = genlmsg_put(skb, 0, thermal_event_seqnum++,
1489 				 &thermal_event_genl_family, 0,
1490 				 THERMAL_GENL_CMD_EVENT);
1491 	if (!msg_header) {
1492 		nlmsg_free(skb);
1493 		return -ENOMEM;
1494 	}
1495 
1496 	/* fill the data */
1497 	attr = nla_reserve(skb, THERMAL_GENL_ATTR_EVENT,
1498 			   sizeof(struct thermal_genl_event));
1499 
1500 	if (!attr) {
1501 		nlmsg_free(skb);
1502 		return -EINVAL;
1503 	}
1504 
1505 	thermal_event = nla_data(attr);
1506 	if (!thermal_event) {
1507 		nlmsg_free(skb);
1508 		return -EINVAL;
1509 	}
1510 
1511 	memset(thermal_event, 0, sizeof(struct thermal_genl_event));
1512 
1513 	thermal_event->orig = tz->id;
1514 	thermal_event->event = event;
1515 
1516 	/* send multicast genetlink message */
1517 	genlmsg_end(skb, msg_header);
1518 
1519 	result = genlmsg_multicast(&thermal_event_genl_family, skb, 0,
1520 				   0, GFP_ATOMIC);
1521 	if (result)
1522 		dev_err(&tz->device, "Failed to send netlink event:%d", result);
1523 
1524 	return result;
1525 }
1526 EXPORT_SYMBOL_GPL(thermal_generate_netlink_event);
1527 
1528 static int __init genetlink_init(void)
1529 {
1530 	return genl_register_family(&thermal_event_genl_family);
1531 }
1532 
1533 static void genetlink_exit(void)
1534 {
1535 	genl_unregister_family(&thermal_event_genl_family);
1536 }
1537 #else /* !CONFIG_NET */
1538 static inline int genetlink_init(void) { return 0; }
1539 static inline void genetlink_exit(void) {}
1540 #endif /* !CONFIG_NET */
1541 
1542 static int thermal_pm_notify(struct notifier_block *nb,
1543 			     unsigned long mode, void *_unused)
1544 {
1545 	struct thermal_zone_device *tz;
1546 	enum thermal_device_mode tz_mode;
1547 
1548 	switch (mode) {
1549 	case PM_HIBERNATION_PREPARE:
1550 	case PM_RESTORE_PREPARE:
1551 	case PM_SUSPEND_PREPARE:
1552 		atomic_set(&in_suspend, 1);
1553 		break;
1554 	case PM_POST_HIBERNATION:
1555 	case PM_POST_RESTORE:
1556 	case PM_POST_SUSPEND:
1557 		atomic_set(&in_suspend, 0);
1558 		list_for_each_entry(tz, &thermal_tz_list, node) {
1559 			tz_mode = THERMAL_DEVICE_ENABLED;
1560 			if (tz->ops->get_mode)
1561 				tz->ops->get_mode(tz, &tz_mode);
1562 
1563 			if (tz_mode == THERMAL_DEVICE_DISABLED)
1564 				continue;
1565 
1566 			thermal_zone_device_init(tz);
1567 			thermal_zone_device_update(tz,
1568 						   THERMAL_EVENT_UNSPECIFIED);
1569 		}
1570 		break;
1571 	default:
1572 		break;
1573 	}
1574 	return 0;
1575 }
1576 
1577 static struct notifier_block thermal_pm_nb = {
1578 	.notifier_call = thermal_pm_notify,
1579 };
1580 
1581 static int __init thermal_init(void)
1582 {
1583 	int result;
1584 
1585 	mutex_init(&poweroff_lock);
1586 	result = thermal_register_governors();
1587 	if (result)
1588 		goto error;
1589 
1590 	result = class_register(&thermal_class);
1591 	if (result)
1592 		goto unregister_governors;
1593 
1594 	result = genetlink_init();
1595 	if (result)
1596 		goto unregister_class;
1597 
1598 	result = of_parse_thermal_zones();
1599 	if (result)
1600 		goto exit_netlink;
1601 
1602 	result = register_pm_notifier(&thermal_pm_nb);
1603 	if (result)
1604 		pr_warn("Thermal: Can not register suspend notifier, return %d\n",
1605 			result);
1606 
1607 	return 0;
1608 
1609 exit_netlink:
1610 	genetlink_exit();
1611 unregister_class:
1612 	class_unregister(&thermal_class);
1613 unregister_governors:
1614 	thermal_unregister_governors();
1615 error:
1616 	ida_destroy(&thermal_tz_ida);
1617 	ida_destroy(&thermal_cdev_ida);
1618 	mutex_destroy(&thermal_list_lock);
1619 	mutex_destroy(&thermal_governor_lock);
1620 	mutex_destroy(&poweroff_lock);
1621 	return result;
1622 }
1623 fs_initcall(thermal_init);
1624