1.. SPDX-License-Identifier: GPL-2.0
2.. include:: <isonum.txt>
3
4===================
5System Sleep States
6===================
7
8:Copyright: |copy| 2017 Intel Corporation
9
10:Author: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
11
12
13Sleep states are global low-power states of the entire system in which user
14space code cannot be executed and the overall system activity is significantly
15reduced.
16
17
18Sleep States That Can Be Supported
19==================================
20
21Depending on its configuration and the capabilities of the platform it runs on,
22the Linux kernel can support up to four system sleep states, including
23hibernation and up to three variants of system suspend.  The sleep states that
24can be supported by the kernel are listed below.
25
26.. _s2idle:
27
28Suspend-to-Idle
29---------------
30
31This is a generic, pure software, light-weight variant of system suspend (also
32referred to as S2I or S2Idle).  It allows more energy to be saved relative to
33runtime idle by freezing user space, suspending the timekeeping and putting all
34I/O devices into low-power states (possibly lower-power than available in the
35working state), such that the processors can spend time in their deepest idle
36states while the system is suspended.
37
38The system is woken up from this state by in-band interrupts, so theoretically
39any devices that can cause interrupts to be generated in the working state can
40also be set up as wakeup devices for S2Idle.
41
42This state can be used on platforms without support for :ref:`standby <standby>`
43or :ref:`suspend-to-RAM <s2ram>`, or it can be used in addition to any of the
44deeper system suspend variants to provide reduced resume latency.  It is always
45supported if the :c:macro:`CONFIG_SUSPEND` kernel configuration option is set.
46
47.. _standby:
48
49Standby
50-------
51
52This state, if supported, offers moderate, but real, energy savings, while
53providing a relatively straightforward transition back to the working state.  No
54operating state is lost (the system core logic retains power), so the system can
55go back to where it left off easily enough.
56
57In addition to freezing user space, suspending the timekeeping and putting all
58I/O devices into low-power states, which is done for :ref:`suspend-to-idle
59<s2idle>` too, nonboot CPUs are taken offline and all low-level system functions
60are suspended during transitions into this state.  For this reason, it should
61allow more energy to be saved relative to :ref:`suspend-to-idle <s2idle>`, but
62the resume latency will generally be greater than for that state.
63
64The set of devices that can wake up the system from this state usually is
65reduced relative to :ref:`suspend-to-idle <s2idle>` and it may be necessary to
66rely on the platform for setting up the wakeup functionality as appropriate.
67
68This state is supported if the :c:macro:`CONFIG_SUSPEND` kernel configuration
69option is set and the support for it is registered by the platform with the
70core system suspend subsystem.  On ACPI-based systems this state is mapped to
71the S1 system state defined by ACPI.
72
73.. _s2ram:
74
75Suspend-to-RAM
76--------------
77
78This state (also referred to as STR or S2RAM), if supported, offers significant
79energy savings as everything in the system is put into a low-power state, except
80for memory, which should be placed into the self-refresh mode to retain its
81contents.  All of the steps carried out when entering :ref:`standby <standby>`
82are also carried out during transitions to S2RAM.  Additional operations may
83take place depending on the platform capabilities.  In particular, on ACPI-based
84systems the kernel passes control to the platform firmware (BIOS) as the last
85step during S2RAM transitions and that usually results in powering down some
86more low-level components that are not directly controlled by the kernel.
87
88The state of devices and CPUs is saved and held in memory.  All devices are
89suspended and put into low-power states.  In many cases, all peripheral buses
90lose power when entering S2RAM, so devices must be able to handle the transition
91back to the "on" state.
92
93On ACPI-based systems S2RAM requires some minimal boot-strapping code in the
94platform firmware to resume the system from it.  This may be the case on other
95platforms too.
96
97The set of devices that can wake up the system from S2RAM usually is reduced
98relative to :ref:`suspend-to-idle <s2idle>` and :ref:`standby <standby>` and it
99may be necessary to rely on the platform for setting up the wakeup functionality
100as appropriate.
101
102S2RAM is supported if the :c:macro:`CONFIG_SUSPEND` kernel configuration option
103is set and the support for it is registered by the platform with the core system
104suspend subsystem.  On ACPI-based systems it is mapped to the S3 system state
105defined by ACPI.
106
107.. _hibernation:
108
109Hibernation
110-----------
111
112This state (also referred to as Suspend-to-Disk or STD) offers the greatest
113energy savings and can be used even in the absence of low-level platform support
114for system suspend.  However, it requires some low-level code for resuming the
115system to be present for the underlying CPU architecture.
116
117Hibernation is significantly different from any of the system suspend variants.
118It takes three system state changes to put it into hibernation and two system
119state changes to resume it.
120
121First, when hibernation is triggered, the kernel stops all system activity and
122creates a snapshot image of memory to be written into persistent storage.  Next,
123the system goes into a state in which the snapshot image can be saved, the image
124is written out and finally the system goes into the target low-power state in
125which power is cut from almost all of its hardware components, including memory,
126except for a limited set of wakeup devices.
127
128Once the snapshot image has been written out, the system may either enter a
129special low-power state (like ACPI S4), or it may simply power down itself.
130Powering down means minimum power draw and it allows this mechanism to work on
131any system.  However, entering a special low-power state may allow additional
132means of system wakeup to be used  (e.g. pressing a key on the keyboard or
133opening a laptop lid).
134
135After wakeup, control goes to the platform firmware that runs a boot loader
136which boots a fresh instance of the kernel (control may also go directly to
137the boot loader, depending on the system configuration, but anyway it causes
138a fresh instance of the kernel to be booted).  That new instance of the kernel
139(referred to as the ``restore kernel``) looks for a hibernation image in
140persistent storage and if one is found, it is loaded into memory.  Next, all
141activity in the system is stopped and the restore kernel overwrites itself with
142the image contents and jumps into a special trampoline area in the original
143kernel stored in the image (referred to as the ``image kernel``), which is where
144the special architecture-specific low-level code is needed.  Finally, the
145image kernel restores the system to the pre-hibernation state and allows user
146space to run again.
147
148Hibernation is supported if the :c:macro:`CONFIG_HIBERNATION` kernel
149configuration option is set.  However, this option can only be set if support
150for the given CPU architecture includes the low-level code for system resume.
151
152
153Basic ``sysfs`` Interfaces for System Suspend and Hibernation
154=============================================================
155
156The power management subsystem provides userspace with a unified ``sysfs``
157interface for system sleep regardless of the underlying system architecture or
158platform.  That interface is located in the :file:`/sys/power/` directory
159(assuming that ``sysfs`` is mounted at :file:`/sys`) and it consists of the
160following attributes (files):
161
162``state``
163	This file contains a list of strings representing sleep states supported
164	by the kernel.  Writing one of these strings into it causes the kernel
165	to start a transition of the system into the sleep state represented by
166	that string.
167
168	In particular, the "disk", "freeze" and "standby" strings represent the
169	:ref:`hibernation <hibernation>`, :ref:`suspend-to-idle <s2idle>` and
170	:ref:`standby <standby>` sleep states, respectively.  The "mem" string
171	is interpreted in accordance with the contents of the ``mem_sleep`` file
172	described below.
173
174	If the kernel does not support any system sleep states, this file is
175	not present.
176
177``mem_sleep``
178	This file contains a list of strings representing supported system
179	suspend	variants and allows user space to select the variant to be
180	associated with the "mem" string in the ``state`` file described above.
181
182	The strings that may be present in this file are "s2idle", "shallow"
183	and "deep".  The "s2idle" string always represents :ref:`suspend-to-idle
184	<s2idle>` and, by convention, "shallow" and "deep" represent
185	:ref:`standby <standby>` and :ref:`suspend-to-RAM <s2ram>`,
186	respectively.
187
188	Writing one of the listed strings into this file causes the system
189	suspend variant represented by it to be associated with the "mem" string
190	in the ``state`` file.  The string representing the suspend variant
191	currently associated with the "mem" string in the ``state`` file is
192	shown in square brackets.
193
194	If the kernel does not support system suspend, this file is not present.
195
196``disk``
197	This file controls the operating mode of hibernation (Suspend-to-Disk).
198	Specifically, it tells the kernel what to do after creating a
199	hibernation image.
200
201	Reading from it returns a list of supported options encoded as:
202
203	``platform``
204		Put the system into a special low-power state (e.g. ACPI S4) to
205		make additional wakeup options available and possibly allow the
206		platform firmware to take a simplified initialization path after
207		wakeup.
208
209		It is only available if the platform provides a special
210		mechanism to put the system to sleep after creating a
211		hibernation image (platforms with ACPI do that as a rule, for
212		example).
213
214	``shutdown``
215		Power off the system.
216
217	``reboot``
218		Reboot the system (useful for diagnostics mostly).
219
220	``suspend``
221		Hybrid system suspend.  Put the system into the suspend sleep
222		state selected through the ``mem_sleep`` file described above.
223		If the system is successfully woken up from that state, discard
224		the hibernation image and continue.  Otherwise, use the image
225		to restore the previous state of the system.
226
227		It is available if system suspend is supported.
228
229	``test_resume``
230		Diagnostic operation.  Load the image as though the system had
231		just woken up from hibernation and the currently running kernel
232		instance was a restore kernel and follow up with full system
233		resume.
234
235	Writing one of the strings listed above into this file causes the option
236	represented by it to be selected.
237
238	The currently selected option is shown in square brackets, which means
239	that the operation represented by it will be carried out after creating
240	and saving the image when hibernation is triggered by writing ``disk``
241	to :file:`/sys/power/state`.
242
243	If the kernel does not support hibernation, this file is not present.
244
245``image_size``
246	This file controls the size of hibernation images.
247
248	It can be written a string representing a non-negative integer that will
249	be used as a best-effort upper limit of the image size, in bytes.  The
250	hibernation core will do its best to ensure that the image size will not
251	exceed that number, but if that turns out to be impossible to achieve, a
252	hibernation image will still be created and its size will be as small as
253	possible.  In particular, writing '0' to this file causes the size of
254	hibernation images to be minimum.
255
256	Reading from it returns the current image size limit, which is set to
257	around 2/5 of the available RAM size by default.
258
259``pm_trace``
260	This file controls the "PM trace" mechanism saving the last suspend
261	or resume event point in the RTC memory across reboots.  It helps to
262	debug hard lockups or reboots due to device driver failures that occur
263	during system suspend or resume (which is more common) more effectively.
264
265	If it contains "1", the fingerprint of each suspend/resume event point
266	in turn will be stored in the RTC memory (overwriting the actual RTC
267	information), so it will survive a system crash if one occurs right
268	after storing it and it can be used later to identify the driver that
269	caused the crash to happen.
270
271	It contains "0" by default, which may be changed to "1" by writing a
272	string representing a nonzero integer into it.
273
274According to the above, there are two ways to make the system go into the
275:ref:`suspend-to-idle <s2idle>` state.  The first one is to write "freeze"
276directly to :file:`/sys/power/state`.  The second one is to write "s2idle" to
277:file:`/sys/power/mem_sleep` and then to write "mem" to
278:file:`/sys/power/state`.  Likewise, there are two ways to make the system go
279into the :ref:`standby <standby>` state (the strings to write to the control
280files in that case are "standby" or "shallow" and "mem", respectively) if that
281state is supported by the platform.  However, there is only one way to make the
282system go into the :ref:`suspend-to-RAM <s2ram>` state (write "deep" into
283:file:`/sys/power/mem_sleep` and "mem" into :file:`/sys/power/state`).
284
285The default suspend variant (ie. the one to be used without writing anything
286into :file:`/sys/power/mem_sleep`) is either "deep" (on the majority of systems
287supporting :ref:`suspend-to-RAM <s2ram>`) or "s2idle", but it can be overridden
288by the value of the ``mem_sleep_default`` parameter in the kernel command line.
289On some systems with ACPI, depending on the information in the ACPI tables, the
290default may be "s2idle" even if :ref:`suspend-to-RAM <s2ram>` is supported in
291principle.
292