1==================================================
2Runtime Power Management Framework for I/O Devices
3==================================================
4
5(C) 2009-2011 Rafael J. Wysocki <rjw@sisk.pl>, Novell Inc.
6
7(C) 2010 Alan Stern <stern@rowland.harvard.edu>
8
9(C) 2014 Intel Corp., Rafael J. Wysocki <rafael.j.wysocki@intel.com>
10
111. Introduction
12===============
13
14Support for runtime power management (runtime PM) of I/O devices is provided
15at the power management core (PM core) level by means of:
16
17* The power management workqueue pm_wq in which bus types and device drivers can
18  put their PM-related work items.  It is strongly recommended that pm_wq be
19  used for queuing all work items related to runtime PM, because this allows
20  them to be synchronized with system-wide power transitions (suspend to RAM,
21  hibernation and resume from system sleep states).  pm_wq is declared in
22  include/linux/pm_runtime.h and defined in kernel/power/main.c.
23
24* A number of runtime PM fields in the 'power' member of 'struct device' (which
25  is of the type 'struct dev_pm_info', defined in include/linux/pm.h) that can
26  be used for synchronizing runtime PM operations with one another.
27
28* Three device runtime PM callbacks in 'struct dev_pm_ops' (defined in
29  include/linux/pm.h).
30
31* A set of helper functions defined in drivers/base/power/runtime.c that can be
32  used for carrying out runtime PM operations in such a way that the
33  synchronization between them is taken care of by the PM core.  Bus types and
34  device drivers are encouraged to use these functions.
35
36The runtime PM callbacks present in 'struct dev_pm_ops', the device runtime PM
37fields of 'struct dev_pm_info' and the core helper functions provided for
38runtime PM are described below.
39
402. Device Runtime PM Callbacks
41==============================
42
43There are three device runtime PM callbacks defined in 'struct dev_pm_ops'::
44
45  struct dev_pm_ops {
46	...
47	int (*runtime_suspend)(struct device *dev);
48	int (*runtime_resume)(struct device *dev);
49	int (*runtime_idle)(struct device *dev);
50	...
51  };
52
53The ->runtime_suspend(), ->runtime_resume() and ->runtime_idle() callbacks
54are executed by the PM core for the device's subsystem that may be either of
55the following:
56
57  1. PM domain of the device, if the device's PM domain object, dev->pm_domain,
58     is present.
59
60  2. Device type of the device, if both dev->type and dev->type->pm are present.
61
62  3. Device class of the device, if both dev->class and dev->class->pm are
63     present.
64
65  4. Bus type of the device, if both dev->bus and dev->bus->pm are present.
66
67If the subsystem chosen by applying the above rules doesn't provide the relevant
68callback, the PM core will invoke the corresponding driver callback stored in
69dev->driver->pm directly (if present).
70
71The PM core always checks which callback to use in the order given above, so the
72priority order of callbacks from high to low is: PM domain, device type, class
73and bus type.  Moreover, the high-priority one will always take precedence over
74a low-priority one.  The PM domain, bus type, device type and class callbacks
75are referred to as subsystem-level callbacks in what follows.
76
77By default, the callbacks are always invoked in process context with interrupts
78enabled.  However, the pm_runtime_irq_safe() helper function can be used to tell
79the PM core that it is safe to run the ->runtime_suspend(), ->runtime_resume()
80and ->runtime_idle() callbacks for the given device in atomic context with
81interrupts disabled.  This implies that the callback routines in question must
82not block or sleep, but it also means that the synchronous helper functions
83listed at the end of Section 4 may be used for that device within an interrupt
84handler or generally in an atomic context.
85
86The subsystem-level suspend callback, if present, is _entirely_ _responsible_
87for handling the suspend of the device as appropriate, which may, but need not
88include executing the device driver's own ->runtime_suspend() callback (from the
89PM core's point of view it is not necessary to implement a ->runtime_suspend()
90callback in a device driver as long as the subsystem-level suspend callback
91knows what to do to handle the device).
92
93  * Once the subsystem-level suspend callback (or the driver suspend callback,
94    if invoked directly) has completed successfully for the given device, the PM
95    core regards the device as suspended, which need not mean that it has been
96    put into a low power state.  It is supposed to mean, however, that the
97    device will not process data and will not communicate with the CPU(s) and
98    RAM until the appropriate resume callback is executed for it.  The runtime
99    PM status of a device after successful execution of the suspend callback is
100    'suspended'.
101
102  * If the suspend callback returns -EBUSY or -EAGAIN, the device's runtime PM
103    status remains 'active', which means that the device _must_ be fully
104    operational afterwards.
105
106  * If the suspend callback returns an error code different from -EBUSY and
107    -EAGAIN, the PM core regards this as a fatal error and will refuse to run
108    the helper functions described in Section 4 for the device until its status
109    is directly set to  either 'active', or 'suspended' (the PM core provides
110    special helper functions for this purpose).
111
112In particular, if the driver requires remote wakeup capability (i.e. hardware
113mechanism allowing the device to request a change of its power state, such as
114PCI PME) for proper functioning and device_can_wakeup() returns 'false' for the
115device, then ->runtime_suspend() should return -EBUSY.  On the other hand, if
116device_can_wakeup() returns 'true' for the device and the device is put into a
117low-power state during the execution of the suspend callback, it is expected
118that remote wakeup will be enabled for the device.  Generally, remote wakeup
119should be enabled for all input devices put into low-power states at run time.
120
121The subsystem-level resume callback, if present, is **entirely responsible** for
122handling the resume of the device as appropriate, which may, but need not
123include executing the device driver's own ->runtime_resume() callback (from the
124PM core's point of view it is not necessary to implement a ->runtime_resume()
125callback in a device driver as long as the subsystem-level resume callback knows
126what to do to handle the device).
127
128  * Once the subsystem-level resume callback (or the driver resume callback, if
129    invoked directly) has completed successfully, the PM core regards the device
130    as fully operational, which means that the device _must_ be able to complete
131    I/O operations as needed.  The runtime PM status of the device is then
132    'active'.
133
134  * If the resume callback returns an error code, the PM core regards this as a
135    fatal error and will refuse to run the helper functions described in Section
136    4 for the device, until its status is directly set to either 'active', or
137    'suspended' (by means of special helper functions provided by the PM core
138    for this purpose).
139
140The idle callback (a subsystem-level one, if present, or the driver one) is
141executed by the PM core whenever the device appears to be idle, which is
142indicated to the PM core by two counters, the device's usage counter and the
143counter of 'active' children of the device.
144
145  * If any of these counters is decreased using a helper function provided by
146    the PM core and it turns out to be equal to zero, the other counter is
147    checked.  If that counter also is equal to zero, the PM core executes the
148    idle callback with the device as its argument.
149
150The action performed by the idle callback is totally dependent on the subsystem
151(or driver) in question, but the expected and recommended action is to check
152if the device can be suspended (i.e. if all of the conditions necessary for
153suspending the device are satisfied) and to queue up a suspend request for the
154device in that case.  If there is no idle callback, or if the callback returns
1550, then the PM core will attempt to carry out a runtime suspend of the device,
156also respecting devices configured for autosuspend.  In essence this means a
157call to pm_runtime_autosuspend() (do note that drivers needs to update the
158device last busy mark, pm_runtime_mark_last_busy(), to control the delay under
159this circumstance).  To prevent this (for example, if the callback routine has
160started a delayed suspend), the routine must return a non-zero value.  Negative
161error return codes are ignored by the PM core.
162
163The helper functions provided by the PM core, described in Section 4, guarantee
164that the following constraints are met with respect to runtime PM callbacks for
165one device:
166
167(1) The callbacks are mutually exclusive (e.g. it is forbidden to execute
168    ->runtime_suspend() in parallel with ->runtime_resume() or with another
169    instance of ->runtime_suspend() for the same device) with the exception that
170    ->runtime_suspend() or ->runtime_resume() can be executed in parallel with
171    ->runtime_idle() (although ->runtime_idle() will not be started while any
172    of the other callbacks is being executed for the same device).
173
174(2) ->runtime_idle() and ->runtime_suspend() can only be executed for 'active'
175    devices (i.e. the PM core will only execute ->runtime_idle() or
176    ->runtime_suspend() for the devices the runtime PM status of which is
177    'active').
178
179(3) ->runtime_idle() and ->runtime_suspend() can only be executed for a device
180    the usage counter of which is equal to zero _and_ either the counter of
181    'active' children of which is equal to zero, or the 'power.ignore_children'
182    flag of which is set.
183
184(4) ->runtime_resume() can only be executed for 'suspended' devices  (i.e. the
185    PM core will only execute ->runtime_resume() for the devices the runtime
186    PM status of which is 'suspended').
187
188Additionally, the helper functions provided by the PM core obey the following
189rules:
190
191  * If ->runtime_suspend() is about to be executed or there's a pending request
192    to execute it, ->runtime_idle() will not be executed for the same device.
193
194  * A request to execute or to schedule the execution of ->runtime_suspend()
195    will cancel any pending requests to execute ->runtime_idle() for the same
196    device.
197
198  * If ->runtime_resume() is about to be executed or there's a pending request
199    to execute it, the other callbacks will not be executed for the same device.
200
201  * A request to execute ->runtime_resume() will cancel any pending or
202    scheduled requests to execute the other callbacks for the same device,
203    except for scheduled autosuspends.
204
2053. Runtime PM Device Fields
206===========================
207
208The following device runtime PM fields are present in 'struct dev_pm_info', as
209defined in include/linux/pm.h:
210
211  `struct timer_list suspend_timer;`
212    - timer used for scheduling (delayed) suspend and autosuspend requests
213
214  `unsigned long timer_expires;`
215    - timer expiration time, in jiffies (if this is different from zero, the
216      timer is running and will expire at that time, otherwise the timer is not
217      running)
218
219  `struct work_struct work;`
220    - work structure used for queuing up requests (i.e. work items in pm_wq)
221
222  `wait_queue_head_t wait_queue;`
223    - wait queue used if any of the helper functions needs to wait for another
224      one to complete
225
226  `spinlock_t lock;`
227    - lock used for synchronization
228
229  `atomic_t usage_count;`
230    - the usage counter of the device
231
232  `atomic_t child_count;`
233    - the count of 'active' children of the device
234
235  `unsigned int ignore_children;`
236    - if set, the value of child_count is ignored (but still updated)
237
238  `unsigned int disable_depth;`
239    - used for disabling the helper functions (they work normally if this is
240      equal to zero); the initial value of it is 1 (i.e. runtime PM is
241      initially disabled for all devices)
242
243  `int runtime_error;`
244    - if set, there was a fatal error (one of the callbacks returned error code
245      as described in Section 2), so the helper functions will not work until
246      this flag is cleared; this is the error code returned by the failing
247      callback
248
249  `unsigned int idle_notification;`
250    - if set, ->runtime_idle() is being executed
251
252  `unsigned int request_pending;`
253    - if set, there's a pending request (i.e. a work item queued up into pm_wq)
254
255  `enum rpm_request request;`
256    - type of request that's pending (valid if request_pending is set)
257
258  `unsigned int deferred_resume;`
259    - set if ->runtime_resume() is about to be run while ->runtime_suspend() is
260      being executed for that device and it is not practical to wait for the
261      suspend to complete; means "start a resume as soon as you've suspended"
262
263  `enum rpm_status runtime_status;`
264    - the runtime PM status of the device; this field's initial value is
265      RPM_SUSPENDED, which means that each device is initially regarded by the
266      PM core as 'suspended', regardless of its real hardware status
267
268  `unsigned int runtime_auto;`
269    - if set, indicates that the user space has allowed the device driver to
270      power manage the device at run time via the /sys/devices/.../power/control
271      `interface;` it may only be modified with the help of the
272      pm_runtime_allow() and pm_runtime_forbid() helper functions
273
274  `unsigned int no_callbacks;`
275    - indicates that the device does not use the runtime PM callbacks (see
276      Section 8); it may be modified only by the pm_runtime_no_callbacks()
277      helper function
278
279  `unsigned int irq_safe;`
280    - indicates that the ->runtime_suspend() and ->runtime_resume() callbacks
281      will be invoked with the spinlock held and interrupts disabled
282
283  `unsigned int use_autosuspend;`
284    - indicates that the device's driver supports delayed autosuspend (see
285      Section 9); it may be modified only by the
286      pm_runtime{_dont}_use_autosuspend() helper functions
287
288  `unsigned int timer_autosuspends;`
289    - indicates that the PM core should attempt to carry out an autosuspend
290      when the timer expires rather than a normal suspend
291
292  `int autosuspend_delay;`
293    - the delay time (in milliseconds) to be used for autosuspend
294
295  `unsigned long last_busy;`
296    - the time (in jiffies) when the pm_runtime_mark_last_busy() helper
297      function was last called for this device; used in calculating inactivity
298      periods for autosuspend
299
300All of the above fields are members of the 'power' member of 'struct device'.
301
3024. Runtime PM Device Helper Functions
303=====================================
304
305The following runtime PM helper functions are defined in
306drivers/base/power/runtime.c and include/linux/pm_runtime.h:
307
308  `void pm_runtime_init(struct device *dev);`
309    - initialize the device runtime PM fields in 'struct dev_pm_info'
310
311  `void pm_runtime_remove(struct device *dev);`
312    - make sure that the runtime PM of the device will be disabled after
313      removing the device from device hierarchy
314
315  `int pm_runtime_idle(struct device *dev);`
316    - execute the subsystem-level idle callback for the device; returns an
317      error code on failure, where -EINPROGRESS means that ->runtime_idle() is
318      already being executed; if there is no callback or the callback returns 0
319      then run pm_runtime_autosuspend(dev) and return its result
320
321  `int pm_runtime_suspend(struct device *dev);`
322    - execute the subsystem-level suspend callback for the device; returns 0 on
323      success, 1 if the device's runtime PM status was already 'suspended', or
324      error code on failure, where -EAGAIN or -EBUSY means it is safe to attempt
325      to suspend the device again in future and -EACCES means that
326      'power.disable_depth' is different from 0
327
328  `int pm_runtime_autosuspend(struct device *dev);`
329    - same as pm_runtime_suspend() except that the autosuspend delay is taken
330      `into account;` if pm_runtime_autosuspend_expiration() says the delay has
331      not yet expired then an autosuspend is scheduled for the appropriate time
332      and 0 is returned
333
334  `int pm_runtime_resume(struct device *dev);`
335    - execute the subsystem-level resume callback for the device; returns 0 on
336      success, 1 if the device's runtime PM status was already 'active' or
337      error code on failure, where -EAGAIN means it may be safe to attempt to
338      resume the device again in future, but 'power.runtime_error' should be
339      checked additionally, and -EACCES means that 'power.disable_depth' is
340      different from 0
341
342  `int pm_request_idle(struct device *dev);`
343    - submit a request to execute the subsystem-level idle callback for the
344      device (the request is represented by a work item in pm_wq); returns 0 on
345      success or error code if the request has not been queued up
346
347  `int pm_request_autosuspend(struct device *dev);`
348    - schedule the execution of the subsystem-level suspend callback for the
349      device when the autosuspend delay has expired; if the delay has already
350      expired then the work item is queued up immediately
351
352  `int pm_schedule_suspend(struct device *dev, unsigned int delay);`
353    - schedule the execution of the subsystem-level suspend callback for the
354      device in future, where 'delay' is the time to wait before queuing up a
355      suspend work item in pm_wq, in milliseconds (if 'delay' is zero, the work
356      item is queued up immediately); returns 0 on success, 1 if the device's PM
357      runtime status was already 'suspended', or error code if the request
358      hasn't been scheduled (or queued up if 'delay' is 0); if the execution of
359      ->runtime_suspend() is already scheduled and not yet expired, the new
360      value of 'delay' will be used as the time to wait
361
362  `int pm_request_resume(struct device *dev);`
363    - submit a request to execute the subsystem-level resume callback for the
364      device (the request is represented by a work item in pm_wq); returns 0 on
365      success, 1 if the device's runtime PM status was already 'active', or
366      error code if the request hasn't been queued up
367
368  `void pm_runtime_get_noresume(struct device *dev);`
369    - increment the device's usage counter
370
371  `int pm_runtime_get(struct device *dev);`
372    - increment the device's usage counter, run pm_request_resume(dev) and
373      return its result
374
375  `int pm_runtime_get_sync(struct device *dev);`
376    - increment the device's usage counter, run pm_runtime_resume(dev) and
377      return its result
378
379  `int pm_runtime_get_if_in_use(struct device *dev);`
380    - return -EINVAL if 'power.disable_depth' is nonzero; otherwise, if the
381      runtime PM status is RPM_ACTIVE and the runtime PM usage counter is
382      nonzero, increment the counter and return 1; otherwise return 0 without
383      changing the counter
384
385  `int pm_runtime_get_if_active(struct device *dev, bool ign_usage_count);`
386    - return -EINVAL if 'power.disable_depth' is nonzero; otherwise, if the
387      runtime PM status is RPM_ACTIVE, and either ign_usage_count is true
388      or the device's usage_count is non-zero, increment the counter and
389      return 1; otherwise return 0 without changing the counter
390
391  `void pm_runtime_put_noidle(struct device *dev);`
392    - decrement the device's usage counter
393
394  `int pm_runtime_put(struct device *dev);`
395    - decrement the device's usage counter; if the result is 0 then run
396      pm_request_idle(dev) and return its result
397
398  `int pm_runtime_put_autosuspend(struct device *dev);`
399    - decrement the device's usage counter; if the result is 0 then run
400      pm_request_autosuspend(dev) and return its result
401
402  `int pm_runtime_put_sync(struct device *dev);`
403    - decrement the device's usage counter; if the result is 0 then run
404      pm_runtime_idle(dev) and return its result
405
406  `int pm_runtime_put_sync_suspend(struct device *dev);`
407    - decrement the device's usage counter; if the result is 0 then run
408      pm_runtime_suspend(dev) and return its result
409
410  `int pm_runtime_put_sync_autosuspend(struct device *dev);`
411    - decrement the device's usage counter; if the result is 0 then run
412      pm_runtime_autosuspend(dev) and return its result
413
414  `void pm_runtime_enable(struct device *dev);`
415    - decrement the device's 'power.disable_depth' field; if that field is equal
416      to zero, the runtime PM helper functions can execute subsystem-level
417      callbacks described in Section 2 for the device
418
419  `int pm_runtime_disable(struct device *dev);`
420    - increment the device's 'power.disable_depth' field (if the value of that
421      field was previously zero, this prevents subsystem-level runtime PM
422      callbacks from being run for the device), make sure that all of the
423      pending runtime PM operations on the device are either completed or
424      canceled; returns 1 if there was a resume request pending and it was
425      necessary to execute the subsystem-level resume callback for the device
426      to satisfy that request, otherwise 0 is returned
427
428  `int pm_runtime_barrier(struct device *dev);`
429    - check if there's a resume request pending for the device and resume it
430      (synchronously) in that case, cancel any other pending runtime PM requests
431      regarding it and wait for all runtime PM operations on it in progress to
432      complete; returns 1 if there was a resume request pending and it was
433      necessary to execute the subsystem-level resume callback for the device to
434      satisfy that request, otherwise 0 is returned
435
436  `void pm_suspend_ignore_children(struct device *dev, bool enable);`
437    - set/unset the power.ignore_children flag of the device
438
439  `int pm_runtime_set_active(struct device *dev);`
440    - clear the device's 'power.runtime_error' flag, set the device's runtime
441      PM status to 'active' and update its parent's counter of 'active'
442      children as appropriate (it is only valid to use this function if
443      'power.runtime_error' is set or 'power.disable_depth' is greater than
444      zero); it will fail and return error code if the device has a parent
445      which is not active and the 'power.ignore_children' flag of which is unset
446
447  `void pm_runtime_set_suspended(struct device *dev);`
448    - clear the device's 'power.runtime_error' flag, set the device's runtime
449      PM status to 'suspended' and update its parent's counter of 'active'
450      children as appropriate (it is only valid to use this function if
451      'power.runtime_error' is set or 'power.disable_depth' is greater than
452      zero)
453
454  `bool pm_runtime_active(struct device *dev);`
455    - return true if the device's runtime PM status is 'active' or its
456      'power.disable_depth' field is not equal to zero, or false otherwise
457
458  `bool pm_runtime_suspended(struct device *dev);`
459    - return true if the device's runtime PM status is 'suspended' and its
460      'power.disable_depth' field is equal to zero, or false otherwise
461
462  `bool pm_runtime_status_suspended(struct device *dev);`
463    - return true if the device's runtime PM status is 'suspended'
464
465  `void pm_runtime_allow(struct device *dev);`
466    - set the power.runtime_auto flag for the device and decrease its usage
467      counter (used by the /sys/devices/.../power/control interface to
468      effectively allow the device to be power managed at run time)
469
470  `void pm_runtime_forbid(struct device *dev);`
471    - unset the power.runtime_auto flag for the device and increase its usage
472      counter (used by the /sys/devices/.../power/control interface to
473      effectively prevent the device from being power managed at run time)
474
475  `void pm_runtime_no_callbacks(struct device *dev);`
476    - set the power.no_callbacks flag for the device and remove the runtime
477      PM attributes from /sys/devices/.../power (or prevent them from being
478      added when the device is registered)
479
480  `void pm_runtime_irq_safe(struct device *dev);`
481    - set the power.irq_safe flag for the device, causing the runtime-PM
482      callbacks to be invoked with interrupts off
483
484  `bool pm_runtime_is_irq_safe(struct device *dev);`
485    - return true if power.irq_safe flag was set for the device, causing
486      the runtime-PM callbacks to be invoked with interrupts off
487
488  `void pm_runtime_mark_last_busy(struct device *dev);`
489    - set the power.last_busy field to the current time
490
491  `void pm_runtime_use_autosuspend(struct device *dev);`
492    - set the power.use_autosuspend flag, enabling autosuspend delays; call
493      pm_runtime_get_sync if the flag was previously cleared and
494      power.autosuspend_delay is negative
495
496  `void pm_runtime_dont_use_autosuspend(struct device *dev);`
497    - clear the power.use_autosuspend flag, disabling autosuspend delays;
498      decrement the device's usage counter if the flag was previously set and
499      power.autosuspend_delay is negative; call pm_runtime_idle
500
501  `void pm_runtime_set_autosuspend_delay(struct device *dev, int delay);`
502    - set the power.autosuspend_delay value to 'delay' (expressed in
503      milliseconds); if 'delay' is negative then runtime suspends are
504      prevented; if power.use_autosuspend is set, pm_runtime_get_sync may be
505      called or the device's usage counter may be decremented and
506      pm_runtime_idle called depending on if power.autosuspend_delay is
507      changed to or from a negative value; if power.use_autosuspend is clear,
508      pm_runtime_idle is called
509
510  `unsigned long pm_runtime_autosuspend_expiration(struct device *dev);`
511    - calculate the time when the current autosuspend delay period will expire,
512      based on power.last_busy and power.autosuspend_delay; if the delay time
513      is 1000 ms or larger then the expiration time is rounded up to the
514      nearest second; returns 0 if the delay period has already expired or
515      power.use_autosuspend isn't set, otherwise returns the expiration time
516      in jiffies
517
518It is safe to execute the following helper functions from interrupt context:
519
520- pm_request_idle()
521- pm_request_autosuspend()
522- pm_schedule_suspend()
523- pm_request_resume()
524- pm_runtime_get_noresume()
525- pm_runtime_get()
526- pm_runtime_put_noidle()
527- pm_runtime_put()
528- pm_runtime_put_autosuspend()
529- pm_runtime_enable()
530- pm_suspend_ignore_children()
531- pm_runtime_set_active()
532- pm_runtime_set_suspended()
533- pm_runtime_suspended()
534- pm_runtime_mark_last_busy()
535- pm_runtime_autosuspend_expiration()
536
537If pm_runtime_irq_safe() has been called for a device then the following helper
538functions may also be used in interrupt context:
539
540- pm_runtime_idle()
541- pm_runtime_suspend()
542- pm_runtime_autosuspend()
543- pm_runtime_resume()
544- pm_runtime_get_sync()
545- pm_runtime_put_sync()
546- pm_runtime_put_sync_suspend()
547- pm_runtime_put_sync_autosuspend()
548
5495. Runtime PM Initialization, Device Probing and Removal
550========================================================
551
552Initially, the runtime PM is disabled for all devices, which means that the
553majority of the runtime PM helper functions described in Section 4 will return
554-EAGAIN until pm_runtime_enable() is called for the device.
555
556In addition to that, the initial runtime PM status of all devices is
557'suspended', but it need not reflect the actual physical state of the device.
558Thus, if the device is initially active (i.e. it is able to process I/O), its
559runtime PM status must be changed to 'active', with the help of
560pm_runtime_set_active(), before pm_runtime_enable() is called for the device.
561
562However, if the device has a parent and the parent's runtime PM is enabled,
563calling pm_runtime_set_active() for the device will affect the parent, unless
564the parent's 'power.ignore_children' flag is set.  Namely, in that case the
565parent won't be able to suspend at run time, using the PM core's helper
566functions, as long as the child's status is 'active', even if the child's
567runtime PM is still disabled (i.e. pm_runtime_enable() hasn't been called for
568the child yet or pm_runtime_disable() has been called for it).  For this reason,
569once pm_runtime_set_active() has been called for the device, pm_runtime_enable()
570should be called for it too as soon as reasonably possible or its runtime PM
571status should be changed back to 'suspended' with the help of
572pm_runtime_set_suspended().
573
574If the default initial runtime PM status of the device (i.e. 'suspended')
575reflects the actual state of the device, its bus type's or its driver's
576->probe() callback will likely need to wake it up using one of the PM core's
577helper functions described in Section 4.  In that case, pm_runtime_resume()
578should be used.  Of course, for this purpose the device's runtime PM has to be
579enabled earlier by calling pm_runtime_enable().
580
581Note, if the device may execute pm_runtime calls during the probe (such as
582if it is registers with a subsystem that may call back in) then the
583pm_runtime_get_sync() call paired with a pm_runtime_put() call will be
584appropriate to ensure that the device is not put back to sleep during the
585probe. This can happen with systems such as the network device layer.
586
587It may be desirable to suspend the device once ->probe() has finished.
588Therefore the driver core uses the asynchronous pm_request_idle() to submit a
589request to execute the subsystem-level idle callback for the device at that
590time.  A driver that makes use of the runtime autosuspend feature, may want to
591update the last busy mark before returning from ->probe().
592
593Moreover, the driver core prevents runtime PM callbacks from racing with the bus
594notifier callback in __device_release_driver(), which is necessary, because the
595notifier is used by some subsystems to carry out operations affecting the
596runtime PM functionality.  It does so by calling pm_runtime_get_sync() before
597driver_sysfs_remove() and the BUS_NOTIFY_UNBIND_DRIVER notifications.  This
598resumes the device if it's in the suspended state and prevents it from
599being suspended again while those routines are being executed.
600
601To allow bus types and drivers to put devices into the suspended state by
602calling pm_runtime_suspend() from their ->remove() routines, the driver core
603executes pm_runtime_put_sync() after running the BUS_NOTIFY_UNBIND_DRIVER
604notifications in __device_release_driver().  This requires bus types and
605drivers to make their ->remove() callbacks avoid races with runtime PM directly,
606but also it allows of more flexibility in the handling of devices during the
607removal of their drivers.
608
609Drivers in ->remove() callback should undo the runtime PM changes done
610in ->probe(). Usually this means calling pm_runtime_disable(),
611pm_runtime_dont_use_autosuspend() etc.
612
613The user space can effectively disallow the driver of the device to power manage
614it at run time by changing the value of its /sys/devices/.../power/control
615attribute to "on", which causes pm_runtime_forbid() to be called.  In principle,
616this mechanism may also be used by the driver to effectively turn off the
617runtime power management of the device until the user space turns it on.
618Namely, during the initialization the driver can make sure that the runtime PM
619status of the device is 'active' and call pm_runtime_forbid().  It should be
620noted, however, that if the user space has already intentionally changed the
621value of /sys/devices/.../power/control to "auto" to allow the driver to power
622manage the device at run time, the driver may confuse it by using
623pm_runtime_forbid() this way.
624
6256. Runtime PM and System Sleep
626==============================
627
628Runtime PM and system sleep (i.e., system suspend and hibernation, also known
629as suspend-to-RAM and suspend-to-disk) interact with each other in a couple of
630ways.  If a device is active when a system sleep starts, everything is
631straightforward.  But what should happen if the device is already suspended?
632
633The device may have different wake-up settings for runtime PM and system sleep.
634For example, remote wake-up may be enabled for runtime suspend but disallowed
635for system sleep (device_may_wakeup(dev) returns 'false').  When this happens,
636the subsystem-level system suspend callback is responsible for changing the
637device's wake-up setting (it may leave that to the device driver's system
638suspend routine).  It may be necessary to resume the device and suspend it again
639in order to do so.  The same is true if the driver uses different power levels
640or other settings for runtime suspend and system sleep.
641
642During system resume, the simplest approach is to bring all devices back to full
643power, even if they had been suspended before the system suspend began.  There
644are several reasons for this, including:
645
646  * The device might need to switch power levels, wake-up settings, etc.
647
648  * Remote wake-up events might have been lost by the firmware.
649
650  * The device's children may need the device to be at full power in order
651    to resume themselves.
652
653  * The driver's idea of the device state may not agree with the device's
654    physical state.  This can happen during resume from hibernation.
655
656  * The device might need to be reset.
657
658  * Even though the device was suspended, if its usage counter was > 0 then most
659    likely it would need a runtime resume in the near future anyway.
660
661If the device had been suspended before the system suspend began and it's
662brought back to full power during resume, then its runtime PM status will have
663to be updated to reflect the actual post-system sleep status.  The way to do
664this is:
665
666	 - pm_runtime_disable(dev);
667	 - pm_runtime_set_active(dev);
668	 - pm_runtime_enable(dev);
669
670The PM core always increments the runtime usage counter before calling the
671->suspend() callback and decrements it after calling the ->resume() callback.
672Hence disabling runtime PM temporarily like this will not cause any runtime
673suspend attempts to be permanently lost.  If the usage count goes to zero
674following the return of the ->resume() callback, the ->runtime_idle() callback
675will be invoked as usual.
676
677On some systems, however, system sleep is not entered through a global firmware
678or hardware operation.  Instead, all hardware components are put into low-power
679states directly by the kernel in a coordinated way.  Then, the system sleep
680state effectively follows from the states the hardware components end up in
681and the system is woken up from that state by a hardware interrupt or a similar
682mechanism entirely under the kernel's control.  As a result, the kernel never
683gives control away and the states of all devices during resume are precisely
684known to it.  If that is the case and none of the situations listed above takes
685place (in particular, if the system is not waking up from hibernation), it may
686be more efficient to leave the devices that had been suspended before the system
687suspend began in the suspended state.
688
689To this end, the PM core provides a mechanism allowing some coordination between
690different levels of device hierarchy.  Namely, if a system suspend .prepare()
691callback returns a positive number for a device, that indicates to the PM core
692that the device appears to be runtime-suspended and its state is fine, so it
693may be left in runtime suspend provided that all of its descendants are also
694left in runtime suspend.  If that happens, the PM core will not execute any
695system suspend and resume callbacks for all of those devices, except for the
696complete callback, which is then entirely responsible for handling the device
697as appropriate.  This only applies to system suspend transitions that are not
698related to hibernation (see Documentation/driver-api/pm/devices.rst for more
699information).
700
701The PM core does its best to reduce the probability of race conditions between
702the runtime PM and system suspend/resume (and hibernation) callbacks by carrying
703out the following operations:
704
705  * During system suspend pm_runtime_get_noresume() is called for every device
706    right before executing the subsystem-level .prepare() callback for it and
707    pm_runtime_barrier() is called for every device right before executing the
708    subsystem-level .suspend() callback for it.  In addition to that the PM core
709    calls  __pm_runtime_disable() with 'false' as the second argument for every
710    device right before executing the subsystem-level .suspend_late() callback
711    for it.
712
713  * During system resume pm_runtime_enable() and pm_runtime_put() are called for
714    every device right after executing the subsystem-level .resume_early()
715    callback and right after executing the subsystem-level .complete() callback
716    for it, respectively.
717
7187. Generic subsystem callbacks
719
720Subsystems may wish to conserve code space by using the set of generic power
721management callbacks provided by the PM core, defined in
722driver/base/power/generic_ops.c:
723
724  `int pm_generic_runtime_suspend(struct device *dev);`
725    - invoke the ->runtime_suspend() callback provided by the driver of this
726      device and return its result, or return 0 if not defined
727
728  `int pm_generic_runtime_resume(struct device *dev);`
729    - invoke the ->runtime_resume() callback provided by the driver of this
730      device and return its result, or return 0 if not defined
731
732  `int pm_generic_suspend(struct device *dev);`
733    - if the device has not been suspended at run time, invoke the ->suspend()
734      callback provided by its driver and return its result, or return 0 if not
735      defined
736
737  `int pm_generic_suspend_noirq(struct device *dev);`
738    - if pm_runtime_suspended(dev) returns "false", invoke the ->suspend_noirq()
739      callback provided by the device's driver and return its result, or return
740      0 if not defined
741
742  `int pm_generic_resume(struct device *dev);`
743    - invoke the ->resume() callback provided by the driver of this device and,
744      if successful, change the device's runtime PM status to 'active'
745
746  `int pm_generic_resume_noirq(struct device *dev);`
747    - invoke the ->resume_noirq() callback provided by the driver of this device
748
749  `int pm_generic_freeze(struct device *dev);`
750    - if the device has not been suspended at run time, invoke the ->freeze()
751      callback provided by its driver and return its result, or return 0 if not
752      defined
753
754  `int pm_generic_freeze_noirq(struct device *dev);`
755    - if pm_runtime_suspended(dev) returns "false", invoke the ->freeze_noirq()
756      callback provided by the device's driver and return its result, or return
757      0 if not defined
758
759  `int pm_generic_thaw(struct device *dev);`
760    - if the device has not been suspended at run time, invoke the ->thaw()
761      callback provided by its driver and return its result, or return 0 if not
762      defined
763
764  `int pm_generic_thaw_noirq(struct device *dev);`
765    - if pm_runtime_suspended(dev) returns "false", invoke the ->thaw_noirq()
766      callback provided by the device's driver and return its result, or return
767      0 if not defined
768
769  `int pm_generic_poweroff(struct device *dev);`
770    - if the device has not been suspended at run time, invoke the ->poweroff()
771      callback provided by its driver and return its result, or return 0 if not
772      defined
773
774  `int pm_generic_poweroff_noirq(struct device *dev);`
775    - if pm_runtime_suspended(dev) returns "false", run the ->poweroff_noirq()
776      callback provided by the device's driver and return its result, or return
777      0 if not defined
778
779  `int pm_generic_restore(struct device *dev);`
780    - invoke the ->restore() callback provided by the driver of this device and,
781      if successful, change the device's runtime PM status to 'active'
782
783  `int pm_generic_restore_noirq(struct device *dev);`
784    - invoke the ->restore_noirq() callback provided by the device's driver
785
786These functions are the defaults used by the PM core, if a subsystem doesn't
787provide its own callbacks for ->runtime_idle(), ->runtime_suspend(),
788->runtime_resume(), ->suspend(), ->suspend_noirq(), ->resume(),
789->resume_noirq(), ->freeze(), ->freeze_noirq(), ->thaw(), ->thaw_noirq(),
790->poweroff(), ->poweroff_noirq(), ->restore(), ->restore_noirq() in the
791subsystem-level dev_pm_ops structure.
792
793Device drivers that wish to use the same function as a system suspend, freeze,
794poweroff and runtime suspend callback, and similarly for system resume, thaw,
795restore, and runtime resume, can achieve this with the help of the
796UNIVERSAL_DEV_PM_OPS macro defined in include/linux/pm.h (possibly setting its
797last argument to NULL).
798
7998. "No-Callback" Devices
800========================
801
802Some "devices" are only logical sub-devices of their parent and cannot be
803power-managed on their own.  (The prototype example is a USB interface.  Entire
804USB devices can go into low-power mode or send wake-up requests, but neither is
805possible for individual interfaces.)  The drivers for these devices have no
806need of runtime PM callbacks; if the callbacks did exist, ->runtime_suspend()
807and ->runtime_resume() would always return 0 without doing anything else and
808->runtime_idle() would always call pm_runtime_suspend().
809
810Subsystems can tell the PM core about these devices by calling
811pm_runtime_no_callbacks().  This should be done after the device structure is
812initialized and before it is registered (although after device registration is
813also okay).  The routine will set the device's power.no_callbacks flag and
814prevent the non-debugging runtime PM sysfs attributes from being created.
815
816When power.no_callbacks is set, the PM core will not invoke the
817->runtime_idle(), ->runtime_suspend(), or ->runtime_resume() callbacks.
818Instead it will assume that suspends and resumes always succeed and that idle
819devices should be suspended.
820
821As a consequence, the PM core will never directly inform the device's subsystem
822or driver about runtime power changes.  Instead, the driver for the device's
823parent must take responsibility for telling the device's driver when the
824parent's power state changes.
825
8269. Autosuspend, or automatically-delayed suspends
827=================================================
828
829Changing a device's power state isn't free; it requires both time and energy.
830A device should be put in a low-power state only when there's some reason to
831think it will remain in that state for a substantial time.  A common heuristic
832says that a device which hasn't been used for a while is liable to remain
833unused; following this advice, drivers should not allow devices to be suspended
834at runtime until they have been inactive for some minimum period.  Even when
835the heuristic ends up being non-optimal, it will still prevent devices from
836"bouncing" too rapidly between low-power and full-power states.
837
838The term "autosuspend" is an historical remnant.  It doesn't mean that the
839device is automatically suspended (the subsystem or driver still has to call
840the appropriate PM routines); rather it means that runtime suspends will
841automatically be delayed until the desired period of inactivity has elapsed.
842
843Inactivity is determined based on the power.last_busy field.  Drivers should
844call pm_runtime_mark_last_busy() to update this field after carrying out I/O,
845typically just before calling pm_runtime_put_autosuspend().  The desired length
846of the inactivity period is a matter of policy.  Subsystems can set this length
847initially by calling pm_runtime_set_autosuspend_delay(), but after device
848registration the length should be controlled by user space, using the
849/sys/devices/.../power/autosuspend_delay_ms attribute.
850
851In order to use autosuspend, subsystems or drivers must call
852pm_runtime_use_autosuspend() (preferably before registering the device), and
853thereafter they should use the various `*_autosuspend()` helper functions
854instead of the non-autosuspend counterparts::
855
856	Instead of: pm_runtime_suspend    use: pm_runtime_autosuspend;
857	Instead of: pm_schedule_suspend   use: pm_request_autosuspend;
858	Instead of: pm_runtime_put        use: pm_runtime_put_autosuspend;
859	Instead of: pm_runtime_put_sync   use: pm_runtime_put_sync_autosuspend.
860
861Drivers may also continue to use the non-autosuspend helper functions; they
862will behave normally, which means sometimes taking the autosuspend delay into
863account (see pm_runtime_idle).
864
865Under some circumstances a driver or subsystem may want to prevent a device
866from autosuspending immediately, even though the usage counter is zero and the
867autosuspend delay time has expired.  If the ->runtime_suspend() callback
868returns -EAGAIN or -EBUSY, and if the next autosuspend delay expiration time is
869in the future (as it normally would be if the callback invoked
870pm_runtime_mark_last_busy()), the PM core will automatically reschedule the
871autosuspend.  The ->runtime_suspend() callback can't do this rescheduling
872itself because no suspend requests of any kind are accepted while the device is
873suspending (i.e., while the callback is running).
874
875The implementation is well suited for asynchronous use in interrupt contexts.
876However such use inevitably involves races, because the PM core can't
877synchronize ->runtime_suspend() callbacks with the arrival of I/O requests.
878This synchronization must be handled by the driver, using its private lock.
879Here is a schematic pseudo-code example::
880
881	foo_read_or_write(struct foo_priv *foo, void *data)
882	{
883		lock(&foo->private_lock);
884		add_request_to_io_queue(foo, data);
885		if (foo->num_pending_requests++ == 0)
886			pm_runtime_get(&foo->dev);
887		if (!foo->is_suspended)
888			foo_process_next_request(foo);
889		unlock(&foo->private_lock);
890	}
891
892	foo_io_completion(struct foo_priv *foo, void *req)
893	{
894		lock(&foo->private_lock);
895		if (--foo->num_pending_requests == 0) {
896			pm_runtime_mark_last_busy(&foo->dev);
897			pm_runtime_put_autosuspend(&foo->dev);
898		} else {
899			foo_process_next_request(foo);
900		}
901		unlock(&foo->private_lock);
902		/* Send req result back to the user ... */
903	}
904
905	int foo_runtime_suspend(struct device *dev)
906	{
907		struct foo_priv foo = container_of(dev, ...);
908		int ret = 0;
909
910		lock(&foo->private_lock);
911		if (foo->num_pending_requests > 0) {
912			ret = -EBUSY;
913		} else {
914			/* ... suspend the device ... */
915			foo->is_suspended = 1;
916		}
917		unlock(&foo->private_lock);
918		return ret;
919	}
920
921	int foo_runtime_resume(struct device *dev)
922	{
923		struct foo_priv foo = container_of(dev, ...);
924
925		lock(&foo->private_lock);
926		/* ... resume the device ... */
927		foo->is_suspended = 0;
928		pm_runtime_mark_last_busy(&foo->dev);
929		if (foo->num_pending_requests > 0)
930			foo_process_next_request(foo);
931		unlock(&foo->private_lock);
932		return 0;
933	}
934
935The important point is that after foo_io_completion() asks for an autosuspend,
936the foo_runtime_suspend() callback may race with foo_read_or_write().
937Therefore foo_runtime_suspend() has to check whether there are any pending I/O
938requests (while holding the private lock) before allowing the suspend to
939proceed.
940
941In addition, the power.autosuspend_delay field can be changed by user space at
942any time.  If a driver cares about this, it can call
943pm_runtime_autosuspend_expiration() from within the ->runtime_suspend()
944callback while holding its private lock.  If the function returns a nonzero
945value then the delay has not yet expired and the callback should return
946-EAGAIN.
947