driver-api/pm/devices.rst

1 .. SPDX-License-Identifier: GPL-2.0
10 :Copyright: |copy| 2010-2011 Rafael J. Wysocki <rjw@sisk.pl>, Novell Inc.
18 management (PM) code is also driver-specific.  Most drivers will do very
22 This writeup gives an overview of how drivers interact with system-wide
25 background for the domain-specific work you'd do with any specific driver.
31 Drivers will use one or both of these models to put devices into low-power
34     System Sleep model:
36 	Drivers can enter low-power states as part of entering system-wide
37 	low-power states like "suspend" (also known as "suspend-to-RAM"), or
39 	"suspend-to-disk").
42 	by implementing various role-specific suspend and resume methods to
46 	Some drivers can manage hardware wakeup events, which make the system
47 	leave the low-power state.  This feature may be enabled or disabled
51 	whole system enter low-power states more often.
55 	Devices may also be put into low-power states while the system is
60 	device is on, it may be necessary to carry out some bus-specific
62 	states at run time may require special handling during system-wide power
67 	the PM core are involved in runtime power management.  As in the system
69 	various role-specific suspend and resume methods, so that the hardware
72 There's not a lot to be said about those low-power states except that they are
73 very system-specific, and often device-specific.  Also, that if enough devices
74 have been put into low-power states (at runtime), the effect may be very similar
75 to entering some system-wide low-power state (system sleep) ... and that
76 synergies exist, so that several drivers using runtime PM might put the system
85 network wake-on-LAN packets, keyboard or mouse activity, and media insertion
88 Interfaces for Entering System Sleep States
94 system sleep and runtime power management.
98 ----------------------------------
105 management while the remaining ones are used during system-wide power
110 struct dev_pm_ops objects and it is suitable only for implementing system
116 Subsystem-Level Methods
117 -----------------------
130 write subsystem-level drivers; most driver code is a "device driver" that builds
131 on top of bus-specific framework code.
134 they are called in phases for every device, respecting the parent-child
139 -------------------------------------------
142 of system wakeup events (hardware signals that can force the system out of a
152 its system wakeup mechanism and for notifying the PM core of system wakeup
153 events signaled by the device.  This object is only present for wakeup-capable
159 whether or not a wakeup-capable device should issue wakeup events is a policy
163 to signal system wakeup.  This file is only present if the
170 Ethernet adapters whose WoL (wake-on-LAN) feature has been set up with ethtool.
181 during a system sleep transition.  Device drivers, however, are not expected to
184 It ought to be noted that system wakeup is conceptually different from "remote
187 low-power states to trigger specific interrupts to signal conditions in which
188 they should be put into the full-power state.  Those interrupts may or may not
189 be used to signal system wakeup events, depending on the hardware design.  On
190 some systems it is impossible to trigger them from system sleep states.  In any
196 --------------------------------------------
207 runtime power-managed by its driver.  Writing "on" calls
209 power if it was in a low-power state, and preventing the
210 device from being runtime power-managed.  User space can check the current value
214 system-wide power transitions.  In particular, the device can (and in the
215 majority of cases should and will) be put into a low-power state during a
216 system-wide transition to a sleep state even though its :c:member:`runtime_auto`
223 Calling Drivers to Enter and Leave System Sleep States
226 When the system goes into a sleep state, each device's driver is asked to
228 system state.  That's usually some version of "off", but the details are
229 system-specific.  Also, wakeup-enabled devices will usually stay partly
230 functional in order to wake the system.
232 When the system leaves that low-power state, the device's driver is asked to
234 always go together, and both are multi-phase operations.
241 More power-aware drivers might prepare the devices for triggering system wakeup
246 ------------------------
250 walked in a bottom-up order to suspend devices.  A top-down order is
266 System Power Management Phases
267 ------------------------------
269 Suspending or resuming the system is done in several phases.  Different phases
270 are used for suspend-to-idle, shallow (standby), and deep ("suspend-to-RAM")
271 sleep states and the hibernation state ("suspend-to-disk").  Each phase involves
280 defined in ``dev->pm_domain->ops``, ``dev->bus->pm``, ``dev->type->pm``,
281 ``dev->class->pm`` or ``dev->driver->pm``).  These callbacks are regarded by the
287     1.	If ``dev->pm_domain`` is present, the PM core will choose the callback
288 	provided by ``dev->pm_domain->ops`` for execution.
290     2.	Otherwise, if both ``dev->type`` and ``dev->type->pm`` are present, the
291 	callback provided by ``dev->type->pm`` will be chosen for execution.
293     3.	Otherwise, if both ``dev->class`` and ``dev->class->pm`` are present,
294 	the callback provided by ``dev->class->pm`` will be chosen for
297     4.	Otherwise, if both ``dev->bus`` and ``dev->bus->pm`` are present, the
298 	callback provided by ``dev->bus->pm`` will be chosen for execution.
303 The PM domain, type, class and bus callbacks may in turn invoke device- or
304 driver-specific methods stored in ``dev->driver->pm``, but they don't have to do
308 execute the corresponding method from the ``dev->driver->pm`` set instead if
312 Entering System Suspend
313 -----------------------
315 When the system goes into the freeze, standby or memory sleep state,
323 	suspend-related phases, during the ``prepare`` phase the device
324 	hierarchy is traversed top-down.
326 	After the ``->prepare`` callback method returns, no new children may be
328 	driver in some way for the upcoming system power transition, but it
329 	should not put the device into a low-power state.  Moreover, if the
330 	device supports runtime power management, the ``->prepare`` callback
336 	safely leave the device in runtime suspend (if runtime-suspended
340 	and all of them (including the device itself) are runtime-suspended, the
344 	the ``->complete`` callback will be the next one invoked after the
345 	``->prepare`` callback and is entirely responsible for putting the
348 	Note that this direct-complete procedure applies even if the device is
349 	disabled for runtime PM; only the runtime-PM status matters.  It follows
350 	that if a device has system-sleep callbacks but does not support runtime
352 	is because all such devices are initially set to runtime-suspended with
360 	these flags is set, the PM core will not apply the direct-complete
364 	the return value of the ``->prepare`` callback provided by the driver
366 	``->prepare`` callback if the driver's one also has returned a positive
369     2.	The ``->suspend`` methods should quiesce the device to stop it from
371 	the appropriate low-power state, depending on the bus type the device is
375 	``->suspend`` methods provided by subsystems (bus types and PM domains
377 	to the devices before their drivers' ``->suspend`` methods are called.
382 	suspend in their ``->suspend`` methods).  In fact, the PM core prevents
385 	the ``->prepare`` callback (and calling :c:func:`pm_runtime_put` after
386 	issuing the ``->complete`` callback).
395 	the callback method is running.  The ``->suspend_noirq`` methods should
397 	and finally put the device into the appropriate low-power state.
407 (DMA, IRQs), saved enough state that they can re-initialize or restore previous
408 state (as needed by the hardware), and placed the device into a low-power state.
414 prepared for generating hardware wakeup signals to trigger a system wakeup event
415 when the system is in the sleep state.  For example, :c:func:`enable_irq_wake()`
419 If any of these callbacks returns an error, the system won't enter the desired
420 low-power state.  Instead, the PM core will unwind its actions by resuming all
424 Leaving System Suspend
425 ----------------------
430     1.	The ``->resume_noirq`` callback methods should perform any actions
438 	For example, the PCI bus type's ``->pm.resume_noirq()`` puts the device
439 	into the full-power state (D0 in the PCI terminology) and restores the
441 	device driver's ``->pm.resume_noirq()`` method to perform device-specific
444     2.	The ``->resume_early`` methods should prepare devices for the execution
448     3.	The ``->resume`` methods should bring the device back to its operating
453         For this reason, unlike the other resume-related phases, during the
454         ``complete`` phase the device hierarchy is traversed bottom-up.
457 	soon as the ``->resume`` callbacks occur; it's not necessary to wait
460 	Moreover, if the preceding ``->prepare`` callback returned a positive
462 	whole system suspend and resume (its ``->suspend``, ``->suspend_late``,
463 	``->suspend_noirq``, ``->resume_noirq``,
464 	``->resume_early``, and ``->resume`` callbacks may have been
465 	skipped).  In that case, the ``->complete`` callback is entirely
466 	responsible for putting the device into a consistent state after system
469 	the case, the ``->complete`` callback can consult the device's
471 	``->complete`` callback is being run then the direct-complete mechanism
479 However, the details here may again be platform-specific.  For example,
490 system sleep entered was suspend-to-idle.  For the other system sleep states
491 that may not be the case (and usually isn't for ACPI-defined system sleep
495 while the system was powered down, whenever that's physically possible.
498 will notice and handle such removals are currently bus-specific, and often
503 the system log.
507 --------------------
509 Hibernating the system is more complicated than putting it into sleep states,
510 because it involves creating and saving a system image.  Therefore there are
515 create an image of the system memory while everything is stable, reactivate all
517 the system ("power off").  The phases used to accomplish this are: ``prepare``,
522     1.	The ``prepare`` phase is discussed in the "Entering System Suspend"
525     2.	The ``->freeze`` methods should quiesce the device so that it doesn't
527 	registers.  However the device does not have to be put in a low-power
533 	low-power state and should not be allowed to generate wakeup events.
537 	a low-power state and should not be allowed to generate wakeup events.
539 At this point the system image is created.  All devices should be inactive and
541 image forms an atomic snapshot of the system state.
556     8.	The ``complete`` phase is discussed in the "Leaving System Suspend"
559 At this point the system image is saved, and the devices then need to be
560 prepared for the upcoming system shutdown.  This is much like suspending them
561 before putting the system into the suspend-to-idle, shallow or deep sleep state,
572 The ``->poweroff``, ``->poweroff_late`` and ``->poweroff_noirq`` callbacks
573 should do essentially the same things as the ``->suspend``, ``->suspend_late``
574 and ``->suspend_noirq`` callbacks, respectively.  A notable difference is
577 ``freeze_noirq`` phases.  Also, on many machines the firmware will power-down
578 the entire system, so it is not necessary for the callback to put the device in
579 a low-power state.
583 -------------------
587 a system image to be loaded into memory and the pre-hibernation memory contents
591 pre-hibernation memory contents restored by the boot loader, in practice this
596 reads the system image, restores the pre-hibernation memory contents, and passes
603 To be able to load the system image into memory, the restore kernel needs to
609 creating a system image, and it is accomplished in the same way, using
614 Should the restoration of the pre-hibernation memory contents fail, the restore
618 pre-hibernation memory contents are restored successfully and control is passed
619 to the image kernel, which then becomes responsible for bringing the system back
622 To achieve this, the image kernel must restore the devices' pre-hibernation
640 reset and completely re-initialized.  In many cases this difference doesn't
641 matter, so the ``->resume[_early|_noirq]`` and ``->restore[_early|_norq]``
658 For details refer to Documentation/driver-api/pm/notifiers.rst.
661 Device Low-Power (suspend) States
664 Device low-power states aren't standard.  One device might only handle
670 gives one example: after the suspend sequence completes, a non-legacy
673 several PCI-standard device states, some of which are optional.
675 In contrast, integrated system-on-chip processors often use IRQs as the
680 Some details here may be platform-specific.  Systems may have devices that
682 refreshed using DMA while most of the system is sleeping lightly ... and
683 its frame buffer might even be updated by a DSP or other non-Linux CPU while
686 Moreover, the specific actions taken may depend on the target system state.
687 One target system state might allow a given device to be very operational;
688 another might require a hard shut down with re-initialization on resume.
698 cases it generally is not possible to put devices into low-power states
700 into a low-power state together at the same time by turning off the shared
701 power resource.  Of course, they also need to be put into the full-power state
705 sub-domain of the parent domain.
710 of power management callbacks analogous to the subsystem-level and device driver
712 instead of the respective subsystem-level callbacks.  Specifically, if a
713 device's :c:member:`pm_domain` pointer is not NULL, the ``->suspend()`` callback
715 (e.g. bus type's) ``->suspend()`` callback and analogously for all of the
723 support for power domains into subsystem-level callbacks, for example by
727 Devices may be defined as IRQ-safe which indicates to the PM core that their
730 IRQ-safe device belongs to a PM domain, the runtime PM of the domain will be
731 disallowed, unless the domain itself is defined as IRQ-safe. However, it
732 makes sense to define a PM domain as IRQ-safe only if all the devices in it
733 are IRQ-safe. Moreover, if an IRQ-safe domain has a parent domain, the runtime
734 PM of the parent is only allowed if the parent itself is IRQ-safe too with the
735 additional restriction that all child domains of an IRQ-safe parent must also
736 be IRQ-safe.
742 Many devices are able to dynamically power down while the system is still
744 can offer significant power savings on a running system.  These devices
748 usually include hardware states that are also used in system sleep states.
750 A system-wide power transition can be started while some devices are in low
751 power states due to runtime power management.  The system sleep PM callbacks
753 necessary actions are subsystem-specific.
757 desirable to leave a suspended device in that state during a system-wide power
758 transition, but in other cases the device must be put back into the full-power
759 state temporarily, for example so that its system wakeup capability can be
763 If it is necessary to resume a device from runtime suspend during a system-wide
765 :c:func:`pm_runtime_resume` from the ``->suspend`` callback (or the ``->freeze``
766 or ``->poweroff`` callback for transitions related to hibernation) of either the
769 from their ``->prepare`` and ``->suspend`` callbacks (or equivalent) *before*
770 invoking device drivers' ``->suspend`` callbacks (or equivalent).
775 ------------------------------------------
778 suspend upfront in their ``->suspend`` callbacks, but that may not be really
779 necessary if the device's driver can cope with runtime-suspended devices.
784 Setting that flag causes the PM core and middle-layer code
785 (bus types, PM domains etc.) to skip the ``->suspend_late`` and
786 ``->suspend_noirq`` callbacks provided by the driver if the device remains in
787 runtime suspend throughout those phases of the system-wide suspend (and
788 similarly for the "freeze" and "poweroff" parts of system hibernation).
791 be valid in general.]  If the middle-layer system-wide PM callbacks are present
797 In addition, with ``DPM_FLAG_SMART_SUSPEND`` set, the driver's ``->thaw_noirq``
798 and ``->thaw_early`` callbacks are skipped in hibernation if the device remained
800 middle-layer callbacks are present for the device, they are responsible for
805 --------------------------------------------
807 During system-wide resume from a sleep state it's easiest to put devices into
808 the full-power state, as explained in Documentation/power/runtime_pm.rst.
812 system transitions to the working state, especially if those devices had been in
813 runtime suspend before the preceding system-wide suspend (or analogous)
817 indicate to the PM core and middle-layer code that they allow their "noirq" and
819 after system-wide PM transitions to the working state.  Whether or not that is
820 the case generally depends on the state of the device before the given system
821 suspend-resume cycle and on the type of the system transition under way.
833 "suspend" phase of suspend-type transitions.  If the driver or the middle layer
835 being skipped during the subsequent system resume transition, it should
836 clear :c:member:`power.may_skip_resume` in its ``->suspend``, ``->suspend_late``
837 or ``->suspend_noirq`` callback.  [Note that the drivers setting
839 their ``->suspend`` callback in case the other two are skipped.]
849 "suspended" by the PM core.  Otherwise, if the device was runtime-suspended
850 during the preceding system-wide suspend transition and its
855 system-wide resume-type transitions.]
859 callbacks are skipped, its system-wide "noirq" and "early" resume callbacks, if
862 driver must be prepared to cope with the invocation of its system-wide resume
863 callbacks back-to-back with its ``->runtime_suspend`` one (without the
864 intervening ``->runtime_resume`` and system-wide suspend callbacks) and the
867 ``->suspend_late`` callback pointer points to the same function as its
868 ``->runtime_suspend`` one and its ``->resume_early`` callback pointer points to
869 the same function as the ``->runtime_resume`` one, while none of the other
870 system-wide suspend-resume callbacks of the driver are present, for example.]
873 system-wide "noirq" and "early" resume callbacks may be skipped while its "late"
876 needs to be able to cope with the invocation of its ``->runtime_resume``
877 callback back-to-back with its "late" and "noirq" suspend ones.  [For instance,
880 functions for runtime PM and system-wide suspend/resume.]