1Device Power Management
2
3Copyright (c) 2010-2011 Rafael J. Wysocki <rjw@sisk.pl>, Novell Inc.
4Copyright (c) 2010 Alan Stern <stern@rowland.harvard.edu>
5
6
7Most of the code in Linux is device drivers, so most of the Linux power
8management (PM) code is also driver-specific.  Most drivers will do very
9little; others, especially for platforms with small batteries (like cell
10phones), will do a lot.
11
12This writeup gives an overview of how drivers interact with system-wide
13power management goals, emphasizing the models and interfaces that are
14shared by everything that hooks up to the driver model core.  Read it as
15background for the domain-specific work you'd do with any specific driver.
16
17
18Two Models for Device Power Management
19======================================
20Drivers will use one or both of these models to put devices into low-power
21states:
22
23    System Sleep model:
24	Drivers can enter low-power states as part of entering system-wide
25	low-power states like "suspend" (also known as "suspend-to-RAM"), or
26	(mostly for systems with disks) "hibernation" (also known as
27	"suspend-to-disk").
28
29	This is something that device, bus, and class drivers collaborate on
30	by implementing various role-specific suspend and resume methods to
31	cleanly power down hardware and software subsystems, then reactivate
32	them without loss of data.
33
34	Some drivers can manage hardware wakeup events, which make the system
35	leave the low-power state.  This feature may be enabled or disabled
36	using the relevant /sys/devices/.../power/wakeup file (for Ethernet
37	drivers the ioctl interface used by ethtool may also be used for this
38	purpose); enabling it may cost some power usage, but let the whole
39	system enter low-power states more often.
40
41    Runtime Power Management model:
42	Devices may also be put into low-power states while the system is
43	running, independently of other power management activity in principle.
44	However, devices are not generally independent of each other (for
45	example, a parent device cannot be suspended unless all of its child
46	devices have been suspended).  Moreover, depending on the bus type the
47	device is on, it may be necessary to carry out some bus-specific
48	operations on the device for this purpose.  Devices put into low power
49	states at run time may require special handling during system-wide power
50	transitions (suspend or hibernation).
51
52	For these reasons not only the device driver itself, but also the
53	appropriate subsystem (bus type, device type or device class) driver and
54	the PM core are involved in runtime power management.  As in the system
55	sleep power management case, they need to collaborate by implementing
56	various role-specific suspend and resume methods, so that the hardware
57	is cleanly powered down and reactivated without data or service loss.
58
59There's not a lot to be said about those low-power states except that they are
60very system-specific, and often device-specific.  Also, that if enough devices
61have been put into low-power states (at runtime), the effect may be very similar
62to entering some system-wide low-power state (system sleep) ... and that
63synergies exist, so that several drivers using runtime PM might put the system
64into a state where even deeper power saving options are available.
65
66Most suspended devices will have quiesced all I/O: no more DMA or IRQs (except
67for wakeup events), no more data read or written, and requests from upstream
68drivers are no longer accepted.  A given bus or platform may have different
69requirements though.
70
71Examples of hardware wakeup events include an alarm from a real time clock,
72network wake-on-LAN packets, keyboard or mouse activity, and media insertion
73or removal (for PCMCIA, MMC/SD, USB, and so on).
74
75
76Interfaces for Entering System Sleep States
77===========================================
78There are programming interfaces provided for subsystems (bus type, device type,
79device class) and device drivers to allow them to participate in the power
80management of devices they are concerned with.  These interfaces cover both
81system sleep and runtime power management.
82
83
84Device Power Management Operations
85----------------------------------
86Device power management operations, at the subsystem level as well as at the
87device driver level, are implemented by defining and populating objects of type
88struct dev_pm_ops:
89
90struct dev_pm_ops {
91	int (*prepare)(struct device *dev);
92	void (*complete)(struct device *dev);
93	int (*suspend)(struct device *dev);
94	int (*resume)(struct device *dev);
95	int (*freeze)(struct device *dev);
96	int (*thaw)(struct device *dev);
97	int (*poweroff)(struct device *dev);
98	int (*restore)(struct device *dev);
99	int (*suspend_noirq)(struct device *dev);
100	int (*resume_noirq)(struct device *dev);
101	int (*freeze_noirq)(struct device *dev);
102	int (*thaw_noirq)(struct device *dev);
103	int (*poweroff_noirq)(struct device *dev);
104	int (*restore_noirq)(struct device *dev);
105	int (*runtime_suspend)(struct device *dev);
106	int (*runtime_resume)(struct device *dev);
107	int (*runtime_idle)(struct device *dev);
108};
109
110This structure is defined in include/linux/pm.h and the methods included in it
111are also described in that file.  Their roles will be explained in what follows.
112For now, it should be sufficient to remember that the last three methods are
113specific to runtime power management while the remaining ones are used during
114system-wide power transitions.
115
116There also is a deprecated "old" or "legacy" interface for power management
117operations available at least for some subsystems.  This approach does not use
118struct dev_pm_ops objects and it is suitable only for implementing system sleep
119power management methods.  Therefore it is not described in this document, so
120please refer directly to the source code for more information about it.
121
122
123Subsystem-Level Methods
124-----------------------
125The core methods to suspend and resume devices reside in struct dev_pm_ops
126pointed to by the ops member of struct dev_pm_domain, or by the pm member of
127struct bus_type, struct device_type and struct class.  They are mostly of
128interest to the people writing infrastructure for platforms and buses, like PCI
129or USB, or device type and device class drivers.  They also are relevant to the
130writers of device drivers whose subsystems (PM domains, device types, device
131classes and bus types) don't provide all power management methods.
132
133Bus drivers implement these methods as appropriate for the hardware and the
134drivers using it; PCI works differently from USB, and so on.  Not many people
135write subsystem-level drivers; most driver code is a "device driver" that builds
136on top of bus-specific framework code.
137
138For more information on these driver calls, see the description later;
139they are called in phases for every device, respecting the parent-child
140sequencing in the driver model tree.
141
142
143/sys/devices/.../power/wakeup files
144-----------------------------------
145All device objects in the driver model contain fields that control the handling
146of system wakeup events (hardware signals that can force the system out of a
147sleep state).  These fields are initialized by bus or device driver code using
148device_set_wakeup_capable() and device_set_wakeup_enable(), defined in
149include/linux/pm_wakeup.h.
150
151The "power.can_wakeup" flag just records whether the device (and its driver) can
152physically support wakeup events.  The device_set_wakeup_capable() routine
153affects this flag.  The "power.wakeup" field is a pointer to an object of type
154struct wakeup_source used for controlling whether or not the device should use
155its system wakeup mechanism and for notifying the PM core of system wakeup
156events signaled by the device.  This object is only present for wakeup-capable
157devices (i.e. devices whose "can_wakeup" flags are set) and is created (or
158removed) by device_set_wakeup_capable().
159
160Whether or not a device is capable of issuing wakeup events is a hardware
161matter, and the kernel is responsible for keeping track of it.  By contrast,
162whether or not a wakeup-capable device should issue wakeup events is a policy
163decision, and it is managed by user space through a sysfs attribute: the
164"power/wakeup" file.  User space can write the strings "enabled" or "disabled"
165to it to indicate whether or not, respectively, the device is supposed to signal
166system wakeup.  This file is only present if the "power.wakeup" object exists
167for the given device and is created (or removed) along with that object, by
168device_set_wakeup_capable().  Reads from the file will return the corresponding
169string.
170
171The "power/wakeup" file is supposed to contain the "disabled" string initially
172for the majority of devices; the major exceptions are power buttons, keyboards,
173and Ethernet adapters whose WoL (wake-on-LAN) feature has been set up with
174ethtool.  It should also default to "enabled" for devices that don't generate
175wakeup requests on their own but merely forward wakeup requests from one bus to
176another (like PCI Express ports).
177
178The device_may_wakeup() routine returns true only if the "power.wakeup" object
179exists and the corresponding "power/wakeup" file contains the string "enabled".
180This information is used by subsystems, like the PCI bus type code, to see
181whether or not to enable the devices' wakeup mechanisms.  If device wakeup
182mechanisms are enabled or disabled directly by drivers, they also should use
183device_may_wakeup() to decide what to do during a system sleep transition.
184Device drivers, however, are not supposed to call device_set_wakeup_enable()
185directly in any case.
186
187It ought to be noted that system wakeup is conceptually different from "remote
188wakeup" used by runtime power management, although it may be supported by the
189same physical mechanism.  Remote wakeup is a feature allowing devices in
190low-power states to trigger specific interrupts to signal conditions in which
191they should be put into the full-power state.  Those interrupts may or may not
192be used to signal system wakeup events, depending on the hardware design.  On
193some systems it is impossible to trigger them from system sleep states.  In any
194case, remote wakeup should always be enabled for runtime power management for
195all devices and drivers that support it.
196
197/sys/devices/.../power/control files
198------------------------------------
199Each device in the driver model has a flag to control whether it is subject to
200runtime power management.  This flag, called runtime_auto, is initialized by the
201bus type (or generally subsystem) code using pm_runtime_allow() or
202pm_runtime_forbid(); the default is to allow runtime power management.
203
204The setting can be adjusted by user space by writing either "on" or "auto" to
205the device's power/control sysfs file.  Writing "auto" calls pm_runtime_allow(),
206setting the flag and allowing the device to be runtime power-managed by its
207driver.  Writing "on" calls pm_runtime_forbid(), clearing the flag, returning
208the device to full power if it was in a low-power state, and preventing the
209device from being runtime power-managed.  User space can check the current value
210of the runtime_auto flag by reading the file.
211
212The device's runtime_auto flag has no effect on the handling of system-wide
213power transitions.  In particular, the device can (and in the majority of cases
214should and will) be put into a low-power state during a system-wide transition
215to a sleep state even though its runtime_auto flag is clear.
216
217For more information about the runtime power management framework, refer to
218Documentation/power/runtime_pm.txt.
219
220
221Calling Drivers to Enter and Leave System Sleep States
222======================================================
223When the system goes into a sleep state, each device's driver is asked to
224suspend the device by putting it into a state compatible with the target
225system state.  That's usually some version of "off", but the details are
226system-specific.  Also, wakeup-enabled devices will usually stay partly
227functional in order to wake the system.
228
229When the system leaves that low-power state, the device's driver is asked to
230resume it by returning it to full power.  The suspend and resume operations
231always go together, and both are multi-phase operations.
232
233For simple drivers, suspend might quiesce the device using class code
234and then turn its hardware as "off" as possible during suspend_noirq.  The
235matching resume calls would then completely reinitialize the hardware
236before reactivating its class I/O queues.
237
238More power-aware drivers might prepare the devices for triggering system wakeup
239events.
240
241
242Call Sequence Guarantees
243------------------------
244To ensure that bridges and similar links needing to talk to a device are
245available when the device is suspended or resumed, the device tree is
246walked in a bottom-up order to suspend devices.  A top-down order is
247used to resume those devices.
248
249The ordering of the device tree is defined by the order in which devices
250get registered:  a child can never be registered, probed or resumed before
251its parent; and can't be removed or suspended after that parent.
252
253The policy is that the device tree should match hardware bus topology.
254(Or at least the control bus, for devices which use multiple busses.)
255In particular, this means that a device registration may fail if the parent of
256the device is suspending (i.e. has been chosen by the PM core as the next
257device to suspend) or has already suspended, as well as after all of the other
258devices have been suspended.  Device drivers must be prepared to cope with such
259situations.
260
261
262System Power Management Phases
263------------------------------
264Suspending or resuming the system is done in several phases.  Different phases
265are used for standby or memory sleep states ("suspend-to-RAM") and the
266hibernation state ("suspend-to-disk").  Each phase involves executing callbacks
267for every device before the next phase begins.  Not all busses or classes
268support all these callbacks and not all drivers use all the callbacks.  The
269various phases always run after tasks have been frozen and before they are
270unfrozen.  Furthermore, the *_noirq phases run at a time when IRQ handlers have
271been disabled (except for those marked with the IRQF_NO_SUSPEND flag).
272
273All phases use PM domain, bus, type, class or driver callbacks (that is, methods
274defined in dev->pm_domain->ops, dev->bus->pm, dev->type->pm, dev->class->pm or
275dev->driver->pm).  These callbacks are regarded by the PM core as mutually
276exclusive.  Moreover, PM domain callbacks always take precedence over all of the
277other callbacks and, for example, type callbacks take precedence over bus, class
278and driver callbacks.  To be precise, the following rules are used to determine
279which callback to execute in the given phase:
280
281    1.	If dev->pm_domain is present, the PM core will choose the callback
282	included in dev->pm_domain->ops for execution
283
284    2.	Otherwise, if both dev->type and dev->type->pm are present, the callback
285	included in dev->type->pm will be chosen for execution.
286
287    3.	Otherwise, if both dev->class and dev->class->pm are present, the
288	callback included in dev->class->pm will be chosen for execution.
289
290    4.	Otherwise, if both dev->bus and dev->bus->pm are present, the callback
291	included in dev->bus->pm will be chosen for execution.
292
293This allows PM domains and device types to override callbacks provided by bus
294types or device classes if necessary.
295
296The PM domain, type, class and bus callbacks may in turn invoke device- or
297driver-specific methods stored in dev->driver->pm, but they don't have to do
298that.
299
300If the subsystem callback chosen for execution is not present, the PM core will
301execute the corresponding method from dev->driver->pm instead if there is one.
302
303
304Entering System Suspend
305-----------------------
306When the system goes into the standby or memory sleep state, the phases are:
307
308		prepare, suspend, suspend_noirq.
309
310    1.	The prepare phase is meant to prevent races by preventing new devices
311	from being registered; the PM core would never know that all the
312	children of a device had been suspended if new children could be
313	registered at will.  (By contrast, devices may be unregistered at any
314	time.)  Unlike the other suspend-related phases, during the prepare
315	phase the device tree is traversed top-down.
316
317	After the prepare callback method returns, no new children may be
318	registered below the device.  The method may also prepare the device or
319	driver in some way for the upcoming system power transition, but it
320	should not put the device into a low-power state.
321
322    2.	The suspend methods should quiesce the device to stop it from performing
323	I/O.  They also may save the device registers and put it into the
324	appropriate low-power state, depending on the bus type the device is on,
325	and they may enable wakeup events.
326
327    3.	The suspend_noirq phase occurs after IRQ handlers have been disabled,
328	which means that the driver's interrupt handler will not be called while
329	the callback method is running.  The methods should save the values of
330	the device's registers that weren't saved previously and finally put the
331	device into the appropriate low-power state.
332
333	The majority of subsystems and device drivers need not implement this
334	callback.  However, bus types allowing devices to share interrupt
335	vectors, like PCI, generally need it; otherwise a driver might encounter
336	an error during the suspend phase by fielding a shared interrupt
337	generated by some other device after its own device had been set to low
338	power.
339
340At the end of these phases, drivers should have stopped all I/O transactions
341(DMA, IRQs), saved enough state that they can re-initialize or restore previous
342state (as needed by the hardware), and placed the device into a low-power state.
343On many platforms they will gate off one or more clock sources; sometimes they
344will also switch off power supplies or reduce voltages.  (Drivers supporting
345runtime PM may already have performed some or all of these steps.)
346
347If device_may_wakeup(dev) returns true, the device should be prepared for
348generating hardware wakeup signals to trigger a system wakeup event when the
349system is in the sleep state.  For example, enable_irq_wake() might identify
350GPIO signals hooked up to a switch or other external hardware, and
351pci_enable_wake() does something similar for the PCI PME signal.
352
353If any of these callbacks returns an error, the system won't enter the desired
354low-power state.  Instead the PM core will unwind its actions by resuming all
355the devices that were suspended.
356
357
358Leaving System Suspend
359----------------------
360When resuming from standby or memory sleep, the phases are:
361
362		resume_noirq, resume, complete.
363
364    1.	The resume_noirq callback methods should perform any actions needed
365	before the driver's interrupt handlers are invoked.  This generally
366	means undoing the actions of the suspend_noirq phase.  If the bus type
367	permits devices to share interrupt vectors, like PCI, the method should
368	bring the device and its driver into a state in which the driver can
369	recognize if the device is the source of incoming interrupts, if any,
370	and handle them correctly.
371
372	For example, the PCI bus type's ->pm.resume_noirq() puts the device into
373	the full-power state (D0 in the PCI terminology) and restores the
374	standard configuration registers of the device.  Then it calls the
375	device driver's ->pm.resume_noirq() method to perform device-specific
376	actions.
377
378    2.	The resume methods should bring the the device back to its operating
379	state, so that it can perform normal I/O.  This generally involves
380	undoing the actions of the suspend phase.
381
382    3.	The complete phase uses only a bus callback.  The method should undo the
383	actions of the prepare phase.  Note, however, that new children may be
384	registered below the device as soon as the resume callbacks occur; it's
385	not necessary to wait until the complete phase.
386
387At the end of these phases, drivers should be as functional as they were before
388suspending: I/O can be performed using DMA and IRQs, and the relevant clocks are
389gated on.  Even if the device was in a low-power state before the system sleep
390because of runtime power management, afterwards it should be back in its
391full-power state.  There are multiple reasons why it's best to do this; they are
392discussed in more detail in Documentation/power/runtime_pm.txt.
393
394However, the details here may again be platform-specific.  For example,
395some systems support multiple "run" states, and the mode in effect at
396the end of resume might not be the one which preceded suspension.
397That means availability of certain clocks or power supplies changed,
398which could easily affect how a driver works.
399
400Drivers need to be able to handle hardware which has been reset since the
401suspend methods were called, for example by complete reinitialization.
402This may be the hardest part, and the one most protected by NDA'd documents
403and chip errata.  It's simplest if the hardware state hasn't changed since
404the suspend was carried out, but that can't be guaranteed (in fact, it usually
405is not the case).
406
407Drivers must also be prepared to notice that the device has been removed
408while the system was powered down, whenever that's physically possible.
409PCMCIA, MMC, USB, Firewire, SCSI, and even IDE are common examples of busses
410where common Linux platforms will see such removal.  Details of how drivers
411will notice and handle such removals are currently bus-specific, and often
412involve a separate thread.
413
414These callbacks may return an error value, but the PM core will ignore such
415errors since there's nothing it can do about them other than printing them in
416the system log.
417
418
419Entering Hibernation
420--------------------
421Hibernating the system is more complicated than putting it into the standby or
422memory sleep state, because it involves creating and saving a system image.
423Therefore there are more phases for hibernation, with a different set of
424callbacks.  These phases always run after tasks have been frozen and memory has
425been freed.
426
427The general procedure for hibernation is to quiesce all devices (freeze), create
428an image of the system memory while everything is stable, reactivate all
429devices (thaw), write the image to permanent storage, and finally shut down the
430system (poweroff).  The phases used to accomplish this are:
431
432	prepare, freeze, freeze_noirq, thaw_noirq, thaw, complete,
433	prepare, poweroff, poweroff_noirq
434
435    1.	The prepare phase is discussed in the "Entering System Suspend" section
436	above.
437
438    2.	The freeze methods should quiesce the device so that it doesn't generate
439	IRQs or DMA, and they may need to save the values of device registers.
440	However the device does not have to be put in a low-power state, and to
441	save time it's best not to do so.  Also, the device should not be
442	prepared to generate wakeup events.
443
444    3.	The freeze_noirq phase is analogous to the suspend_noirq phase discussed
445	above, except again that the device should not be put in a low-power
446	state and should not be allowed to generate wakeup events.
447
448At this point the system image is created.  All devices should be inactive and
449the contents of memory should remain undisturbed while this happens, so that the
450image forms an atomic snapshot of the system state.
451
452    4.	The thaw_noirq phase is analogous to the resume_noirq phase discussed
453	above.  The main difference is that its methods can assume the device is
454	in the same state as at the end of the freeze_noirq phase.
455
456    5.	The thaw phase is analogous to the resume phase discussed above.  Its
457	methods should bring the device back to an operating state, so that it
458	can be used for saving the image if necessary.
459
460    6.	The complete phase is discussed in the "Leaving System Suspend" section
461	above.
462
463At this point the system image is saved, and the devices then need to be
464prepared for the upcoming system shutdown.  This is much like suspending them
465before putting the system into the standby or memory sleep state, and the phases
466are similar.
467
468    7.	The prepare phase is discussed above.
469
470    8.	The poweroff phase is analogous to the suspend phase.
471
472    9.	The poweroff_noirq phase is analogous to the suspend_noirq phase.
473
474The poweroff and poweroff_noirq callbacks should do essentially the same things
475as the suspend and suspend_noirq callbacks.  The only notable difference is that
476they need not store the device register values, because the registers should
477already have been stored during the freeze or freeze_noirq phases.
478
479
480Leaving Hibernation
481-------------------
482Resuming from hibernation is, again, more complicated than resuming from a sleep
483state in which the contents of main memory are preserved, because it requires
484a system image to be loaded into memory and the pre-hibernation memory contents
485to be restored before control can be passed back to the image kernel.
486
487Although in principle, the image might be loaded into memory and the
488pre-hibernation memory contents restored by the boot loader, in practice this
489can't be done because boot loaders aren't smart enough and there is no
490established protocol for passing the necessary information.  So instead, the
491boot loader loads a fresh instance of the kernel, called the boot kernel, into
492memory and passes control to it in the usual way.  Then the boot kernel reads
493the system image, restores the pre-hibernation memory contents, and passes
494control to the image kernel.  Thus two different kernels are involved in
495resuming from hibernation.  In fact, the boot kernel may be completely different
496from the image kernel: a different configuration and even a different version.
497This has important consequences for device drivers and their subsystems.
498
499To be able to load the system image into memory, the boot kernel needs to
500include at least a subset of device drivers allowing it to access the storage
501medium containing the image, although it doesn't need to include all of the
502drivers present in the image kernel.  After the image has been loaded, the
503devices managed by the boot kernel need to be prepared for passing control back
504to the image kernel.  This is very similar to the initial steps involved in
505creating a system image, and it is accomplished in the same way, using prepare,
506freeze, and freeze_noirq phases.  However the devices affected by these phases
507are only those having drivers in the boot kernel; other devices will still be in
508whatever state the boot loader left them.
509
510Should the restoration of the pre-hibernation memory contents fail, the boot
511kernel would go through the "thawing" procedure described above, using the
512thaw_noirq, thaw, and complete phases, and then continue running normally.  This
513happens only rarely.  Most often the pre-hibernation memory contents are
514restored successfully and control is passed to the image kernel, which then
515becomes responsible for bringing the system back to the working state.
516
517To achieve this, the image kernel must restore the devices' pre-hibernation
518functionality.  The operation is much like waking up from the memory sleep
519state, although it involves different phases:
520
521	restore_noirq, restore, complete
522
523    1.	The restore_noirq phase is analogous to the resume_noirq phase.
524
525    2.	The restore phase is analogous to the resume phase.
526
527    3.	The complete phase is discussed above.
528
529The main difference from resume[_noirq] is that restore[_noirq] must assume the
530device has been accessed and reconfigured by the boot loader or the boot kernel.
531Consequently the state of the device may be different from the state remembered
532from the freeze and freeze_noirq phases.  The device may even need to be reset
533and completely re-initialized.  In many cases this difference doesn't matter, so
534the resume[_noirq] and restore[_norq] method pointers can be set to the same
535routines.  Nevertheless, different callback pointers are used in case there is a
536situation where it actually matters.
537
538
539Device Power Management Domains
540-------------------------------
541Sometimes devices share reference clocks or other power resources.  In those
542cases it generally is not possible to put devices into low-power states
543individually.  Instead, a set of devices sharing a power resource can be put
544into a low-power state together at the same time by turning off the shared
545power resource.  Of course, they also need to be put into the full-power state
546together, by turning the shared power resource on.  A set of devices with this
547property is often referred to as a power domain.
548
549Support for power domains is provided through the pm_domain field of struct
550device.  This field is a pointer to an object of type struct dev_pm_domain,
551defined in include/linux/pm.h, providing a set of power management callbacks
552analogous to the subsystem-level and device driver callbacks that are executed
553for the given device during all power transitions, instead of the respective
554subsystem-level callbacks.  Specifically, if a device's pm_domain pointer is
555not NULL, the ->suspend() callback from the object pointed to by it will be
556executed instead of its subsystem's (e.g. bus type's) ->suspend() callback and
557anlogously for all of the remaining callbacks.  In other words, power management
558domain callbacks, if defined for the given device, always take precedence over
559the callbacks provided by the device's subsystem (e.g. bus type).
560
561The support for device power management domains is only relevant to platforms
562needing to use the same device driver power management callbacks in many
563different power domain configurations and wanting to avoid incorporating the
564support for power domains into subsystem-level callbacks, for example by
565modifying the platform bus type.  Other platforms need not implement it or take
566it into account in any way.
567
568
569Device Low Power (suspend) States
570---------------------------------
571Device low-power states aren't standard.  One device might only handle
572"on" and "off, while another might support a dozen different versions of
573"on" (how many engines are active?), plus a state that gets back to "on"
574faster than from a full "off".
575
576Some busses define rules about what different suspend states mean.  PCI
577gives one example:  after the suspend sequence completes, a non-legacy
578PCI device may not perform DMA or issue IRQs, and any wakeup events it
579issues would be issued through the PME# bus signal.  Plus, there are
580several PCI-standard device states, some of which are optional.
581
582In contrast, integrated system-on-chip processors often use IRQs as the
583wakeup event sources (so drivers would call enable_irq_wake) and might
584be able to treat DMA completion as a wakeup event (sometimes DMA can stay
585active too, it'd only be the CPU and some peripherals that sleep).
586
587Some details here may be platform-specific.  Systems may have devices that
588can be fully active in certain sleep states, such as an LCD display that's
589refreshed using DMA while most of the system is sleeping lightly ... and
590its frame buffer might even be updated by a DSP or other non-Linux CPU while
591the Linux control processor stays idle.
592
593Moreover, the specific actions taken may depend on the target system state.
594One target system state might allow a given device to be very operational;
595another might require a hard shut down with re-initialization on resume.
596And two different target systems might use the same device in different
597ways; the aforementioned LCD might be active in one product's "standby",
598but a different product using the same SOC might work differently.
599
600
601Power Management Notifiers
602--------------------------
603There are some operations that cannot be carried out by the power management
604callbacks discussed above, because the callbacks occur too late or too early.
605To handle these cases, subsystems and device drivers may register power
606management notifiers that are called before tasks are frozen and after they have
607been thawed.  Generally speaking, the PM notifiers are suitable for performing
608actions that either require user space to be available, or at least won't
609interfere with user space.
610
611For details refer to Documentation/power/notifiers.txt.
612
613
614Runtime Power Management
615========================
616Many devices are able to dynamically power down while the system is still
617running. This feature is useful for devices that are not being used, and
618can offer significant power savings on a running system.  These devices
619often support a range of runtime power states, which might use names such
620as "off", "sleep", "idle", "active", and so on.  Those states will in some
621cases (like PCI) be partially constrained by the bus the device uses, and will
622usually include hardware states that are also used in system sleep states.
623
624A system-wide power transition can be started while some devices are in low
625power states due to runtime power management.  The system sleep PM callbacks
626should recognize such situations and react to them appropriately, but the
627necessary actions are subsystem-specific.
628
629In some cases the decision may be made at the subsystem level while in other
630cases the device driver may be left to decide.  In some cases it may be
631desirable to leave a suspended device in that state during a system-wide power
632transition, but in other cases the device must be put back into the full-power
633state temporarily, for example so that its system wakeup capability can be
634disabled.  This all depends on the hardware and the design of the subsystem and
635device driver in question.
636
637During system-wide resume from a sleep state it's easiest to put devices into
638the full-power state, as explained in Documentation/power/runtime_pm.txt.  Refer
639to that document for more information regarding this particular issue as well as
640for information on the device runtime power management framework in general.
641