]>
Commit | Line | Data |
---|---|---|
624f6ec8 RW |
1 | Device Power Management |
2 | ||
7538e3db | 3 | Copyright (c) 2010-2011 Rafael J. Wysocki <rjw@sisk.pl>, Novell Inc. |
d6f9cda1 AS |
4 | Copyright (c) 2010 Alan Stern <stern@rowland.harvard.edu> |
5 | ||
624f6ec8 | 6 | |
4fc08400 | 7 | Most of the code in Linux is device drivers, so most of the Linux power |
d6f9cda1 AS |
8 | management (PM) code is also driver-specific. Most drivers will do very |
9 | little; others, especially for platforms with small batteries (like cell | |
10 | phones), will do a lot. | |
4fc08400 DB |
11 | |
12 | This writeup gives an overview of how drivers interact with system-wide | |
13 | power management goals, emphasizing the models and interfaces that are | |
14 | shared by everything that hooks up to the driver model core. Read it as | |
15 | background for the domain-specific work you'd do with any specific driver. | |
16 | ||
17 | ||
18 | Two Models for Device Power Management | |
19 | ====================================== | |
20 | Drivers will use one or both of these models to put devices into low-power | |
21 | states: | |
22 | ||
23 | System Sleep model: | |
d6f9cda1 AS |
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"). | |
4fc08400 DB |
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 | |
d6f9cda1 | 35 | leave the low-power state. This feature may be enabled or disabled |
624f6ec8 RW |
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 | |
d6f9cda1 | 39 | system enter low-power states more often. |
4fc08400 DB |
40 | |
41 | Runtime Power Management model: | |
d6f9cda1 | 42 | Devices may also be put into low-power states while the system is |
624f6ec8 RW |
43 | running, independently of other power management activity in principle. |
44 | However, devices are not generally independent of each other (for | |
d6f9cda1 AS |
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 | |
624f6ec8 | 47 | device is on, it may be necessary to carry out some bus-specific |
d6f9cda1 AS |
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). | |
624f6ec8 RW |
51 | |
52 | For these reasons not only the device driver itself, but also the | |
d6f9cda1 AS |
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 | ||
59 | There's not a lot to be said about those low-power states except that they are | |
60 | very system-specific, and often device-specific. Also, that if enough devices | |
61 | have been put into low-power states (at runtime), the effect may be very similar | |
62 | to entering some system-wide low-power state (system sleep) ... and that | |
63 | synergies exist, so that several drivers using runtime PM might put the system | |
64 | into a state where even deeper power saving options are available. | |
65 | ||
66 | Most suspended devices will have quiesced all I/O: no more DMA or IRQs (except | |
67 | for wakeup events), no more data read or written, and requests from upstream | |
68 | drivers are no longer accepted. A given bus or platform may have different | |
69 | requirements though. | |
4fc08400 DB |
70 | |
71 | Examples of hardware wakeup events include an alarm from a real time clock, | |
72 | network wake-on-LAN packets, keyboard or mouse activity, and media insertion | |
73 | or removal (for PCMCIA, MMC/SD, USB, and so on). | |
74 | ||
75 | ||
76 | Interfaces for Entering System Sleep States | |
77 | =========================================== | |
d6f9cda1 AS |
78 | There are programming interfaces provided for subsystems (bus type, device type, |
79 | device class) and device drivers to allow them to participate in the power | |
80 | management of devices they are concerned with. These interfaces cover both | |
81 | system sleep and runtime power management. | |
624f6ec8 RW |
82 | |
83 | ||
84 | Device Power Management Operations | |
85 | ---------------------------------- | |
86 | Device power management operations, at the subsystem level as well as at the | |
87 | device driver level, are implemented by defining and populating objects of type | |
88 | struct dev_pm_ops: | |
89 | ||
90 | struct 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 | }; | |
4fc08400 | 109 | |
624f6ec8 RW |
110 | This structure is defined in include/linux/pm.h and the methods included in it |
111 | are also described in that file. Their roles will be explained in what follows. | |
d6f9cda1 AS |
112 | For now, it should be sufficient to remember that the last three methods are |
113 | specific to runtime power management while the remaining ones are used during | |
624f6ec8 | 114 | system-wide power transitions. |
4fc08400 | 115 | |
d6f9cda1 AS |
116 | There also is a deprecated "old" or "legacy" interface for power management |
117 | operations available at least for some subsystems. This approach does not use | |
118 | struct dev_pm_ops objects and it is suitable only for implementing system sleep | |
119 | power management methods. Therefore it is not described in this document, so | |
120 | please refer directly to the source code for more information about it. | |
624f6ec8 RW |
121 | |
122 | ||
123 | Subsystem-Level Methods | |
124 | ----------------------- | |
125 | The core methods to suspend and resume devices reside in struct dev_pm_ops | |
5841eb64 RW |
126 | pointed to by the ops member of struct dev_pm_domain, or by the pm member of |
127 | struct bus_type, struct device_type and struct class. They are mostly of | |
128 | interest to the people writing infrastructure for platforms and buses, like PCI | |
35cd133c RW |
129 | or USB, or device type and device class drivers. They also are relevant to the |
130 | writers of device drivers whose subsystems (PM domains, device types, device | |
131 | classes and bus types) don't provide all power management methods. | |
1da177e4 | 132 | |
d6f9cda1 AS |
133 | Bus drivers implement these methods as appropriate for the hardware and the |
134 | drivers using it; PCI works differently from USB, and so on. Not many people | |
135 | write subsystem-level drivers; most driver code is a "device driver" that builds | |
136 | on top of bus-specific framework code. | |
4fc08400 DB |
137 | |
138 | For more information on these driver calls, see the description later; | |
139 | they are called in phases for every device, respecting the parent-child | |
624f6ec8 | 140 | sequencing in the driver model tree. |
4fc08400 DB |
141 | |
142 | ||
143 | /sys/devices/.../power/wakeup files | |
144 | ----------------------------------- | |
fafba48d RW |
145 | All device objects in the driver model contain fields that control the handling |
146 | of system wakeup events (hardware signals that can force the system out of a | |
147 | sleep state). These fields are initialized by bus or device driver code using | |
d6f9cda1 AS |
148 | device_set_wakeup_capable() and device_set_wakeup_enable(), defined in |
149 | include/linux/pm_wakeup.h. | |
4fc08400 | 150 | |
fafba48d | 151 | The "power.can_wakeup" flag just records whether the device (and its driver) can |
d6f9cda1 | 152 | physically support wakeup events. The device_set_wakeup_capable() routine |
fafba48d RW |
153 | affects this flag. The "power.wakeup" field is a pointer to an object of type |
154 | struct wakeup_source used for controlling whether or not the device should use | |
155 | its system wakeup mechanism and for notifying the PM core of system wakeup | |
156 | events signaled by the device. This object is only present for wakeup-capable | |
157 | devices (i.e. devices whose "can_wakeup" flags are set) and is created (or | |
158 | removed) by device_set_wakeup_capable(). | |
d6f9cda1 AS |
159 | |
160 | Whether or not a device is capable of issuing wakeup events is a hardware | |
161 | matter, and the kernel is responsible for keeping track of it. By contrast, | |
162 | whether or not a wakeup-capable device should issue wakeup events is a policy | |
163 | decision, and it is managed by user space through a sysfs attribute: the | |
fafba48d RW |
164 | "power/wakeup" file. User space can write the strings "enabled" or "disabled" |
165 | to it to indicate whether or not, respectively, the device is supposed to signal | |
166 | system wakeup. This file is only present if the "power.wakeup" object exists | |
167 | for the given device and is created (or removed) along with that object, by | |
168 | device_set_wakeup_capable(). Reads from the file will return the corresponding | |
169 | string. | |
170 | ||
171 | The "power/wakeup" file is supposed to contain the "disabled" string initially | |
172 | for the majority of devices; the major exceptions are power buttons, keyboards, | |
173 | and Ethernet adapters whose WoL (wake-on-LAN) feature has been set up with | |
174 | ethtool. It should also default to "enabled" for devices that don't generate | |
175 | wakeup requests on their own but merely forward wakeup requests from one bus to | |
176 | another (like PCI Express ports). | |
177 | ||
178 | The device_may_wakeup() routine returns true only if the "power.wakeup" object | |
179 | exists and the corresponding "power/wakeup" file contains the string "enabled". | |
cb8f51bd RW |
180 | This information is used by subsystems, like the PCI bus type code, to see |
181 | whether or not to enable the devices' wakeup mechanisms. If device wakeup | |
182 | mechanisms are enabled or disabled directly by drivers, they also should use | |
183 | device_may_wakeup() to decide what to do during a system sleep transition. | |
fafba48d RW |
184 | Device drivers, however, are not supposed to call device_set_wakeup_enable() |
185 | directly in any case. | |
186 | ||
187 | It ought to be noted that system wakeup is conceptually different from "remote | |
188 | wakeup" used by runtime power management, although it may be supported by the | |
189 | same physical mechanism. Remote wakeup is a feature allowing devices in | |
190 | low-power states to trigger specific interrupts to signal conditions in which | |
191 | they should be put into the full-power state. Those interrupts may or may not | |
192 | be used to signal system wakeup events, depending on the hardware design. On | |
193 | some systems it is impossible to trigger them from system sleep states. In any | |
194 | case, remote wakeup should always be enabled for runtime power management for | |
195 | all devices and drivers that support it. | |
624f6ec8 RW |
196 | |
197 | /sys/devices/.../power/control files | |
198 | ------------------------------------ | |
d6f9cda1 AS |
199 | Each device in the driver model has a flag to control whether it is subject to |
200 | runtime power management. This flag, called runtime_auto, is initialized by the | |
201 | bus type (or generally subsystem) code using pm_runtime_allow() or | |
202 | pm_runtime_forbid(); the default is to allow runtime power management. | |
203 | ||
204 | The setting can be adjusted by user space by writing either "on" or "auto" to | |
205 | the device's power/control sysfs file. Writing "auto" calls pm_runtime_allow(), | |
206 | setting the flag and allowing the device to be runtime power-managed by its | |
207 | driver. Writing "on" calls pm_runtime_forbid(), clearing the flag, returning | |
208 | the device to full power if it was in a low-power state, and preventing the | |
209 | device from being runtime power-managed. User space can check the current value | |
210 | of the runtime_auto flag by reading the file. | |
624f6ec8 RW |
211 | |
212 | The device's runtime_auto flag has no effect on the handling of system-wide | |
d6f9cda1 AS |
213 | power transitions. In particular, the device can (and in the majority of cases |
214 | should and will) be put into a low-power state during a system-wide transition | |
215 | to a sleep state even though its runtime_auto flag is clear. | |
624f6ec8 | 216 | |
d6f9cda1 AS |
217 | For more information about the runtime power management framework, refer to |
218 | Documentation/power/runtime_pm.txt. | |
4fc08400 DB |
219 | |
220 | ||
d6f9cda1 AS |
221 | Calling Drivers to Enter and Leave System Sleep States |
222 | ====================================================== | |
223 | When the system goes into a sleep state, each device's driver is asked to | |
224 | suspend the device by putting it into a state compatible with the target | |
4fc08400 DB |
225 | system state. That's usually some version of "off", but the details are |
226 | system-specific. Also, wakeup-enabled devices will usually stay partly | |
227 | functional in order to wake the system. | |
228 | ||
d6f9cda1 AS |
229 | When the system leaves that low-power state, the device's driver is asked to |
230 | resume it by returning it to full power. The suspend and resume operations | |
231 | always go together, and both are multi-phase operations. | |
4fc08400 | 232 | |
d6f9cda1 AS |
233 | For simple drivers, suspend might quiesce the device using class code |
234 | and then turn its hardware as "off" as possible during suspend_noirq. The | |
4fc08400 DB |
235 | matching resume calls would then completely reinitialize the hardware |
236 | before reactivating its class I/O queues. | |
237 | ||
624f6ec8 RW |
238 | More power-aware drivers might prepare the devices for triggering system wakeup |
239 | events. | |
4fc08400 DB |
240 | |
241 | ||
242 | Call Sequence Guarantees | |
243 | ------------------------ | |
624f6ec8 | 244 | To ensure that bridges and similar links needing to talk to a device are |
4fc08400 DB |
245 | available when the device is suspended or resumed, the device tree is |
246 | walked in a bottom-up order to suspend devices. A top-down order is | |
247 | used to resume those devices. | |
248 | ||
249 | The ordering of the device tree is defined by the order in which devices | |
250 | get registered: a child can never be registered, probed or resumed before | |
251 | its parent; and can't be removed or suspended after that parent. | |
252 | ||
253 | The policy is that the device tree should match hardware bus topology. | |
254 | (Or at least the control bus, for devices which use multiple busses.) | |
58aca232 | 255 | In particular, this means that a device registration may fail if the parent of |
624f6ec8 | 256 | the device is suspending (i.e. has been chosen by the PM core as the next |
58aca232 RW |
257 | device to suspend) or has already suspended, as well as after all of the other |
258 | devices have been suspended. Device drivers must be prepared to cope with such | |
259 | situations. | |
4fc08400 DB |
260 | |
261 | ||
d6f9cda1 AS |
262 | System Power Management Phases |
263 | ------------------------------ | |
264 | Suspending or resuming the system is done in several phases. Different phases | |
265 | are used for standby or memory sleep states ("suspend-to-RAM") and the | |
266 | hibernation state ("suspend-to-disk"). Each phase involves executing callbacks | |
267 | for every device before the next phase begins. Not all busses or classes | |
268 | support all these callbacks and not all drivers use all the callbacks. The | |
269 | various phases always run after tasks have been frozen and before they are | |
270 | unfrozen. Furthermore, the *_noirq phases run at a time when IRQ handlers have | |
fa8ce723 | 271 | been disabled (except for those marked with the IRQF_NO_SUSPEND flag). |
624f6ec8 | 272 | |
35cd133c RW |
273 | All phases use PM domain, bus, type, class or driver callbacks (that is, methods |
274 | defined in dev->pm_domain->ops, dev->bus->pm, dev->type->pm, dev->class->pm or | |
275 | dev->driver->pm). These callbacks are regarded by the PM core as mutually | |
276 | exclusive. Moreover, PM domain callbacks always take precedence over all of the | |
277 | other callbacks and, for example, type callbacks take precedence over bus, class | |
278 | and driver callbacks. To be precise, the following rules are used to determine | |
279 | which callback to execute in the given phase: | |
5841eb64 | 280 | |
35cd133c RW |
281 | 1. If dev->pm_domain is present, the PM core will choose the callback |
282 | included in dev->pm_domain->ops for execution | |
5841eb64 RW |
283 | |
284 | 2. Otherwise, if both dev->type and dev->type->pm are present, the callback | |
35cd133c | 285 | included in dev->type->pm will be chosen for execution. |
5841eb64 RW |
286 | |
287 | 3. Otherwise, if both dev->class and dev->class->pm are present, the | |
35cd133c | 288 | callback included in dev->class->pm will be chosen for execution. |
5841eb64 RW |
289 | |
290 | 4. Otherwise, if both dev->bus and dev->bus->pm are present, the callback | |
35cd133c | 291 | included in dev->bus->pm will be chosen for execution. |
5841eb64 RW |
292 | |
293 | This allows PM domains and device types to override callbacks provided by bus | |
294 | types or device classes if necessary. | |
4fc08400 | 295 | |
35cd133c RW |
296 | The PM domain, type, class and bus callbacks may in turn invoke device- or |
297 | driver-specific methods stored in dev->driver->pm, but they don't have to do | |
298 | that. | |
299 | ||
300 | If the subsystem callback chosen for execution is not present, the PM core will | |
301 | execute the corresponding method from dev->driver->pm instead if there is one. | |
4fc08400 | 302 | |
4fc08400 | 303 | |
d6f9cda1 AS |
304 | Entering System Suspend |
305 | ----------------------- | |
306 | When 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 | ||
91e7c75b RW |
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 | |
fa8ce723 RW |
319 | driver in some way for the upcoming system power transition, but it |
320 | should not put the device into a low-power state. | |
d6f9cda1 AS |
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. | |
624f6ec8 RW |
332 | |
333 | The majority of subsystems and device drivers need not implement this | |
d6f9cda1 AS |
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 | ||
340 | At 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 | |
342 | state (as needed by the hardware), and placed the device into a low-power state. | |
343 | On many platforms they will gate off one or more clock sources; sometimes they | |
344 | will also switch off power supplies or reduce voltages. (Drivers supporting | |
345 | runtime PM may already have performed some or all of these steps.) | |
624f6ec8 RW |
346 | |
347 | If device_may_wakeup(dev) returns true, the device should be prepared for | |
d6f9cda1 AS |
348 | generating hardware wakeup signals to trigger a system wakeup event when the |
349 | system is in the sleep state. For example, enable_irq_wake() might identify | |
624f6ec8 RW |
350 | GPIO signals hooked up to a switch or other external hardware, and |
351 | pci_enable_wake() does something similar for the PCI PME signal. | |
352 | ||
d6f9cda1 AS |
353 | If any of these callbacks returns an error, the system won't enter the desired |
354 | low-power state. Instead the PM core will unwind its actions by resuming all | |
355 | the devices that were suspended. | |
4fc08400 DB |
356 | |
357 | ||
d6f9cda1 AS |
358 | Leaving System Suspend |
359 | ---------------------- | |
360 | When resuming from standby or memory sleep, the phases are: | |
4fc08400 | 361 | |
d6f9cda1 | 362 | resume_noirq, resume, complete. |
4fc08400 | 363 | |
d6f9cda1 AS |
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. | |
4fc08400 | 371 | |
624f6ec8 | 372 | For example, the PCI bus type's ->pm.resume_noirq() puts the device into |
d6f9cda1 AS |
373 | the full-power state (D0 in the PCI terminology) and restores the |
374 | standard configuration registers of the device. Then it calls the | |
624f6ec8 | 375 | device driver's ->pm.resume_noirq() method to perform device-specific |
d6f9cda1 | 376 | actions. |
4fc08400 | 377 | |
d6f9cda1 AS |
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. | |
4fc08400 | 381 | |
d6f9cda1 AS |
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. | |
4fc08400 | 386 | |
d6f9cda1 AS |
387 | At the end of these phases, drivers should be as functional as they were before |
388 | suspending: I/O can be performed using DMA and IRQs, and the relevant clocks are | |
389 | gated on. Even if the device was in a low-power state before the system sleep | |
390 | because of runtime power management, afterwards it should be back in its | |
391 | full-power state. There are multiple reasons why it's best to do this; they are | |
392 | discussed in more detail in Documentation/power/runtime_pm.txt. | |
4fc08400 DB |
393 | |
394 | However, the details here may again be platform-specific. For example, | |
395 | some systems support multiple "run" states, and the mode in effect at | |
624f6ec8 | 396 | the end of resume might not be the one which preceded suspension. |
4fc08400 DB |
397 | That means availability of certain clocks or power supplies changed, |
398 | which could easily affect how a driver works. | |
399 | ||
4fc08400 DB |
400 | Drivers need to be able to handle hardware which has been reset since the |
401 | suspend methods were called, for example by complete reinitialization. | |
402 | This may be the hardest part, and the one most protected by NDA'd documents | |
403 | and chip errata. It's simplest if the hardware state hasn't changed since | |
25985edc | 404 | the suspend was carried out, but that can't be guaranteed (in fact, it usually |
624f6ec8 | 405 | is not the case). |
4fc08400 DB |
406 | |
407 | Drivers must also be prepared to notice that the device has been removed | |
d6f9cda1 | 408 | while the system was powered down, whenever that's physically possible. |
4fc08400 DB |
409 | PCMCIA, MMC, USB, Firewire, SCSI, and even IDE are common examples of busses |
410 | where common Linux platforms will see such removal. Details of how drivers | |
411 | will notice and handle such removals are currently bus-specific, and often | |
412 | involve a separate thread. | |
1da177e4 | 413 | |
d6f9cda1 AS |
414 | These callbacks may return an error value, but the PM core will ignore such |
415 | errors since there's nothing it can do about them other than printing them in | |
416 | the system log. | |
1da177e4 | 417 | |
d6f9cda1 AS |
418 | |
419 | Entering Hibernation | |
420 | -------------------- | |
421 | Hibernating the system is more complicated than putting it into the standby or | |
422 | memory sleep state, because it involves creating and saving a system image. | |
423 | Therefore there are more phases for hibernation, with a different set of | |
424 | callbacks. These phases always run after tasks have been frozen and memory has | |
425 | been freed. | |
426 | ||
427 | The general procedure for hibernation is to quiesce all devices (freeze), create | |
428 | an image of the system memory while everything is stable, reactivate all | |
429 | devices (thaw), write the image to permanent storage, and finally shut down the | |
430 | system (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 | ||
448 | At this point the system image is created. All devices should be inactive and | |
449 | the contents of memory should remain undisturbed while this happens, so that the | |
450 | image 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 | ||
463 | At this point the system image is saved, and the devices then need to be | |
464 | prepared for the upcoming system shutdown. This is much like suspending them | |
465 | before putting the system into the standby or memory sleep state, and the phases | |
466 | are 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 | ||
474 | The poweroff and poweroff_noirq callbacks should do essentially the same things | |
475 | as the suspend and suspend_noirq callbacks. The only notable difference is that | |
476 | they need not store the device register values, because the registers should | |
477 | already have been stored during the freeze or freeze_noirq phases. | |
478 | ||
479 | ||
480 | Leaving Hibernation | |
481 | ------------------- | |
624f6ec8 RW |
482 | Resuming from hibernation is, again, more complicated than resuming from a sleep |
483 | state in which the contents of main memory are preserved, because it requires | |
484 | a system image to be loaded into memory and the pre-hibernation memory contents | |
485 | to be restored before control can be passed back to the image kernel. | |
486 | ||
d6f9cda1 AS |
487 | Although in principle, the image might be loaded into memory and the |
488 | pre-hibernation memory contents restored by the boot loader, in practice this | |
489 | can't be done because boot loaders aren't smart enough and there is no | |
490 | established protocol for passing the necessary information. So instead, the | |
491 | boot loader loads a fresh instance of the kernel, called the boot kernel, into | |
492 | memory and passes control to it in the usual way. Then the boot kernel reads | |
493 | the system image, restores the pre-hibernation memory contents, and passes | |
494 | control to the image kernel. Thus two different kernels are involved in | |
495 | resuming from hibernation. In fact, the boot kernel may be completely different | |
496 | from the image kernel: a different configuration and even a different version. | |
497 | This has important consequences for device drivers and their subsystems. | |
498 | ||
499 | To be able to load the system image into memory, the boot kernel needs to | |
500 | include at least a subset of device drivers allowing it to access the storage | |
501 | medium containing the image, although it doesn't need to include all of the | |
502 | drivers present in the image kernel. After the image has been loaded, the | |
503 | devices managed by the boot kernel need to be prepared for passing control back | |
504 | to the image kernel. This is very similar to the initial steps involved in | |
505 | creating a system image, and it is accomplished in the same way, using prepare, | |
506 | freeze, and freeze_noirq phases. However the devices affected by these phases | |
507 | are only those having drivers in the boot kernel; other devices will still be in | |
508 | whatever state the boot loader left them. | |
624f6ec8 RW |
509 | |
510 | Should the restoration of the pre-hibernation memory contents fail, the boot | |
d6f9cda1 AS |
511 | kernel would go through the "thawing" procedure described above, using the |
512 | thaw_noirq, thaw, and complete phases, and then continue running normally. This | |
513 | happens only rarely. Most often the pre-hibernation memory contents are | |
514 | restored successfully and control is passed to the image kernel, which then | |
515 | becomes responsible for bringing the system back to the working state. | |
624f6ec8 | 516 | |
d6f9cda1 AS |
517 | To achieve this, the image kernel must restore the devices' pre-hibernation |
518 | functionality. The operation is much like waking up from the memory sleep | |
519 | state, although it involves different phases: | |
624f6ec8 | 520 | |
d6f9cda1 | 521 | restore_noirq, restore, complete |
624f6ec8 | 522 | |
d6f9cda1 | 523 | 1. The restore_noirq phase is analogous to the resume_noirq phase. |
624f6ec8 | 524 | |
d6f9cda1 | 525 | 2. The restore phase is analogous to the resume phase. |
624f6ec8 | 526 | |
d6f9cda1 | 527 | 3. The complete phase is discussed above. |
624f6ec8 | 528 | |
d6f9cda1 AS |
529 | The main difference from resume[_noirq] is that restore[_noirq] must assume the |
530 | device has been accessed and reconfigured by the boot loader or the boot kernel. | |
531 | Consequently the state of the device may be different from the state remembered | |
532 | from the freeze and freeze_noirq phases. The device may even need to be reset | |
533 | and completely re-initialized. In many cases this difference doesn't matter, so | |
534 | the resume[_noirq] and restore[_norq] method pointers can be set to the same | |
535 | routines. Nevertheless, different callback pointers are used in case there is a | |
536 | situation where it actually matters. | |
1da177e4 | 537 | |
1da177e4 | 538 | |
564b905a RW |
539 | Device Power Management Domains |
540 | ------------------------------- | |
7538e3db RW |
541 | Sometimes devices share reference clocks or other power resources. In those |
542 | cases it generally is not possible to put devices into low-power states | |
543 | individually. Instead, a set of devices sharing a power resource can be put | |
544 | into a low-power state together at the same time by turning off the shared | |
545 | power resource. Of course, they also need to be put into the full-power state | |
546 | together, by turning the shared power resource on. A set of devices with this | |
547 | property is often referred to as a power domain. | |
548 | ||
564b905a RW |
549 | Support for power domains is provided through the pm_domain field of struct |
550 | device. This field is a pointer to an object of type struct dev_pm_domain, | |
7538e3db RW |
551 | defined in include/linux/pm.h, providing a set of power management callbacks |
552 | analogous to the subsystem-level and device driver callbacks that are executed | |
ca9c6890 RW |
553 | for the given device during all power transitions, instead of the respective |
554 | subsystem-level callbacks. Specifically, if a device's pm_domain pointer is | |
555 | not NULL, the ->suspend() callback from the object pointed to by it will be | |
556 | executed instead of its subsystem's (e.g. bus type's) ->suspend() callback and | |
557 | anlogously for all of the remaining callbacks. In other words, power management | |
558 | domain callbacks, if defined for the given device, always take precedence over | |
559 | the callbacks provided by the device's subsystem (e.g. bus type). | |
560 | ||
561 | The support for device power management domains is only relevant to platforms | |
562 | needing to use the same device driver power management callbacks in many | |
563 | different power domain configurations and wanting to avoid incorporating the | |
564 | support for power domains into subsystem-level callbacks, for example by | |
565 | modifying the platform bus type. Other platforms need not implement it or take | |
566 | it into account in any way. | |
7538e3db RW |
567 | |
568 | ||
d6f9cda1 AS |
569 | Device Low Power (suspend) States |
570 | --------------------------------- | |
571 | Device 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" | |
574 | faster than from a full "off". | |
575 | ||
576 | Some busses define rules about what different suspend states mean. PCI | |
577 | gives one example: after the suspend sequence completes, a non-legacy | |
578 | PCI device may not perform DMA or issue IRQs, and any wakeup events it | |
579 | issues would be issued through the PME# bus signal. Plus, there are | |
580 | several PCI-standard device states, some of which are optional. | |
581 | ||
582 | In contrast, integrated system-on-chip processors often use IRQs as the | |
583 | wakeup event sources (so drivers would call enable_irq_wake) and might | |
584 | be able to treat DMA completion as a wakeup event (sometimes DMA can stay | |
585 | active too, it'd only be the CPU and some peripherals that sleep). | |
586 | ||
587 | Some details here may be platform-specific. Systems may have devices that | |
588 | can be fully active in certain sleep states, such as an LCD display that's | |
589 | refreshed using DMA while most of the system is sleeping lightly ... and | |
590 | its frame buffer might even be updated by a DSP or other non-Linux CPU while | |
591 | the Linux control processor stays idle. | |
592 | ||
593 | Moreover, the specific actions taken may depend on the target system state. | |
594 | One target system state might allow a given device to be very operational; | |
595 | another might require a hard shut down with re-initialization on resume. | |
596 | And two different target systems might use the same device in different | |
597 | ways; the aforementioned LCD might be active in one product's "standby", | |
598 | but a different product using the same SOC might work differently. | |
599 | ||
600 | ||
624f6ec8 RW |
601 | Power Management Notifiers |
602 | -------------------------- | |
d6f9cda1 AS |
603 | There are some operations that cannot be carried out by the power management |
604 | callbacks discussed above, because the callbacks occur too late or too early. | |
605 | To handle these cases, subsystems and device drivers may register power | |
606 | management notifiers that are called before tasks are frozen and after they have | |
607 | been thawed. Generally speaking, the PM notifiers are suitable for performing | |
608 | actions that either require user space to be available, or at least won't | |
609 | interfere with user space. | |
624f6ec8 RW |
610 | |
611 | For details refer to Documentation/power/notifiers.txt. | |
612 | ||
613 | ||
4fc08400 DB |
614 | Runtime Power Management |
615 | ======================== | |
616 | Many devices are able to dynamically power down while the system is still | |
617 | running. This feature is useful for devices that are not being used, and | |
618 | can offer significant power savings on a running system. These devices | |
619 | often support a range of runtime power states, which might use names such | |
620 | as "off", "sleep", "idle", "active", and so on. Those states will in some | |
d6f9cda1 | 621 | cases (like PCI) be partially constrained by the bus the device uses, and will |
4fc08400 DB |
622 | usually include hardware states that are also used in system sleep states. |
623 | ||
d6f9cda1 AS |
624 | A system-wide power transition can be started while some devices are in low |
625 | power states due to runtime power management. The system sleep PM callbacks | |
626 | should recognize such situations and react to them appropriately, but the | |
627 | necessary actions are subsystem-specific. | |
628 | ||
629 | In some cases the decision may be made at the subsystem level while in other | |
630 | cases the device driver may be left to decide. In some cases it may be | |
631 | desirable to leave a suspended device in that state during a system-wide power | |
632 | transition, but in other cases the device must be put back into the full-power | |
633 | state temporarily, for example so that its system wakeup capability can be | |
634 | disabled. This all depends on the hardware and the design of the subsystem and | |
635 | device driver in question. | |
636 | ||
455716e9 RW |
637 | During system-wide resume from a sleep state it's easiest to put devices into |
638 | the full-power state, as explained in Documentation/power/runtime_pm.txt. Refer | |
639 | to that document for more information regarding this particular issue as well as | |
624f6ec8 | 640 | for information on the device runtime power management framework in general. |