Merge tag 'hardening-v6.7-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/kees...
[linux-block.git] / Documentation / power / swsusp.rst
CommitLineData
151f4e2b
MCC
1============
2Swap suspend
3============
4
5Some warnings, first.
6
7.. warning::
8
9 **BIG FAT WARNING**
10
11 If you touch anything on disk between suspend and resume...
12 ...kiss your data goodbye.
13
14 If you do resume from initrd after your filesystems are mounted...
15 ...bye bye root partition.
16
17 [this is actually same case as above]
18
19 If you have unsupported ( ) devices using DMA, you may have some
20 problems. If your disk driver does not support suspend... (IDE does),
21 it may cause some problems, too. If you change kernel command line
22 between suspend and resume, it may do something wrong. If you change
23 your hardware while system is suspended... well, it was not good idea;
24 but it will probably only crash.
25
26 ( ) suspend/resume support is needed to make it safe.
27
28 If you have any filesystems on USB devices mounted before software suspend,
29 they won't be accessible after resume and you may lose data, as though
30 you have unplugged the USB devices with mounted filesystems on them;
31 see the FAQ below for details. (This is not true for more traditional
32 power states like "standby", which normally don't turn USB off.)
33
34Swap partition:
35 You need to append resume=/dev/your_swap_partition to kernel command
36 line or specify it using /sys/power/resume.
37
38Swap file:
39 If using a swapfile you can also specify a resume offset using
40 resume_offset=<number> on the kernel command line or specify it
41 in /sys/power/resume_offset.
42
43After preparing then you suspend by::
44
45 echo shutdown > /sys/power/disk; echo disk > /sys/power/state
46
47- If you feel ACPI works pretty well on your system, you might try::
48
49 echo platform > /sys/power/disk; echo disk > /sys/power/state
50
51- If you would like to write hibernation image to swap and then suspend
52 to RAM (provided your platform supports it), you can try::
53
54 echo suspend > /sys/power/disk; echo disk > /sys/power/state
55
56- If you have SATA disks, you'll need recent kernels with SATA suspend
57 support. For suspend and resume to work, make sure your disk drivers
58 are built into kernel -- not modules. [There's way to make
59 suspend/resume with modular disk drivers, see FAQ, but you probably
60 should not do that.]
61
62If you want to limit the suspend image size to N bytes, do::
63
64 echo N > /sys/power/image_size
65
66before suspend (it is limited to around 2/5 of available RAM by default).
67
68- The resume process checks for the presence of the resume device,
69 if found, it then checks the contents for the hibernation image signature.
70 If both are found, it resumes the hibernation image.
71
72- The resume process may be triggered in two ways:
73
74 1) During lateinit: If resume=/dev/your_swap_partition is specified on
75 the kernel command line, lateinit runs the resume process. If the
76 resume device has not been probed yet, the resume process fails and
77 bootup continues.
78 2) Manually from an initrd or initramfs: May be run from
79 the init script by using the /sys/power/resume file. It is vital
80 that this be done prior to remounting any filesystems (even as
81 read-only) otherwise data may be corrupted.
82
83Article about goals and implementation of Software Suspend for Linux
84====================================================================
85
86Author: Gábor Kuti
87Last revised: 2003-10-20 by Pavel Machek
88
89Idea and goals to achieve
90-------------------------
91
92Nowadays it is common in several laptops that they have a suspend button. It
93saves the state of the machine to a filesystem or to a partition and switches
94to standby mode. Later resuming the machine the saved state is loaded back to
95ram and the machine can continue its work. It has two real benefits. First we
96save ourselves the time machine goes down and later boots up, energy costs
97are real high when running from batteries. The other gain is that we don't have
98to interrupt our programs so processes that are calculating something for a long
99time shouldn't need to be written interruptible.
100
101swsusp saves the state of the machine into active swaps and then reboots or
102powerdowns. You must explicitly specify the swap partition to resume from with
103`resume=` kernel option. If signature is found it loads and restores saved
104state. If the option `noresume` is specified as a boot parameter, it skips
105the resuming. If the option `hibernate=nocompress` is specified as a boot
106parameter, it saves hibernation image without compression.
107
108In the meantime while the system is suspended you should not add/remove any
109of the hardware, write to the filesystems, etc.
110
111Sleep states summary
112====================
113
114There are three different interfaces you can use, /proc/acpi should
115work like this:
116
117In a really perfect world::
118
119 echo 1 > /proc/acpi/sleep # for standby
120 echo 2 > /proc/acpi/sleep # for suspend to ram
1992b66d
BH
121 echo 3 > /proc/acpi/sleep # for suspend to ram, but with more power
122 # conservative
151f4e2b
MCC
123 echo 4 > /proc/acpi/sleep # for suspend to disk
124 echo 5 > /proc/acpi/sleep # for shutdown unfriendly the system
125
126and perhaps::
127
128 echo 4b > /proc/acpi/sleep # for suspend to disk via s4bios
129
130Frequently Asked Questions
131==========================
132
133Q:
134 well, suspending a server is IMHO a really stupid thing,
135 but... (Diego Zuccato):
136
137A:
138 You bought new UPS for your server. How do you install it without
139 bringing machine down? Suspend to disk, rearrange power cables,
140 resume.
141
142 You have your server on UPS. Power died, and UPS is indicating 30
143 seconds to failure. What do you do? Suspend to disk.
144
145
146Q:
147 Maybe I'm missing something, but why don't the regular I/O paths work?
148
149A:
150 We do use the regular I/O paths. However we cannot restore the data
151 to its original location as we load it. That would create an
152 inconsistent kernel state which would certainly result in an oops.
153 Instead, we load the image into unused memory and then atomically copy
154 it back to it original location. This implies, of course, a maximum
155 image size of half the amount of memory.
156
157 There are two solutions to this:
158
159 * require half of memory to be free during suspend. That way you can
160 read "new" data onto free spots, then cli and copy
161
162 * assume we had special "polling" ide driver that only uses memory
163 between 0-640KB. That way, I'd have to make sure that 0-640KB is free
164 during suspending, but otherwise it would work...
165
166 suspend2 shares this fundamental limitation, but does not include user
167 data and disk caches into "used memory" by saving them in
168 advance. That means that the limitation goes away in practice.
169
170Q:
171 Does linux support ACPI S4?
172
173A:
174 Yes. That's what echo platform > /sys/power/disk does.
175
176Q:
177 What is 'suspend2'?
178
179A:
180 suspend2 is 'Software Suspend 2', a forked implementation of
181 suspend-to-disk which is available as separate patches for 2.4 and 2.6
182 kernels from swsusp.sourceforge.net. It includes support for SMP, 4GB
183 highmem and preemption. It also has a extensible architecture that
184 allows for arbitrary transformations on the image (compression,
185 encryption) and arbitrary backends for writing the image (eg to swap
186 or an NFS share[Work In Progress]). Questions regarding suspend2
187 should be sent to the mailing list available through the suspend2
188 website, and not to the Linux Kernel Mailing List. We are working
189 toward merging suspend2 into the mainline kernel.
190
191Q:
192 What is the freezing of tasks and why are we using it?
193
194A:
195 The freezing of tasks is a mechanism by which user space processes and some
1992b66d
BH
196 kernel threads are controlled during hibernation or system-wide suspend (on
197 some architectures). See freezing-of-tasks.txt for details.
151f4e2b
MCC
198
199Q:
200 What is the difference between "platform" and "shutdown"?
201
202A:
203 shutdown:
204 save state in linux, then tell bios to powerdown
205
206 platform:
207 save state in linux, then tell bios to powerdown and blink
208 "suspended led"
209
210 "platform" is actually right thing to do where supported, but
211 "shutdown" is most reliable (except on ACPI systems).
212
213Q:
214 I do not understand why you have such strong objections to idea of
215 selective suspend.
216
217A:
218 Do selective suspend during runtime power management, that's okay. But
219 it's useless for suspend-to-disk. (And I do not see how you could use
220 it for suspend-to-ram, I hope you do not want that).
221
222 Lets see, so you suggest to
223
224 * SUSPEND all but swap device and parents
225 * Snapshot
226 * Write image to disk
227 * SUSPEND swap device and parents
228 * Powerdown
229
230 Oh no, that does not work, if swap device or its parents uses DMA,
231 you've corrupted data. You'd have to do
232
233 * SUSPEND all but swap device and parents
234 * FREEZE swap device and parents
235 * Snapshot
236 * UNFREEZE swap device and parents
237 * Write
238 * SUSPEND swap device and parents
239
240 Which means that you still need that FREEZE state, and you get more
241 complicated code. (And I have not yet introduce details like system
242 devices).
243
244Q:
245 There don't seem to be any generally useful behavioral
246 distinctions between SUSPEND and FREEZE.
247
248A:
249 Doing SUSPEND when you are asked to do FREEZE is always correct,
250 but it may be unnecessarily slow. If you want your driver to stay simple,
251 slowness may not matter to you. It can always be fixed later.
252
253 For devices like disk it does matter, you do not want to spindown for
254 FREEZE.
255
256Q:
257 After resuming, system is paging heavily, leading to very bad interactivity.
258
259A:
260 Try running::
261
262 cat /proc/[0-9]*/maps | grep / | sed 's:.* /:/:' | sort -u | while read file
263 do
264 test -f "$file" && cat "$file" > /dev/null
265 done
266
267 after resume. swapoff -a; swapon -a may also be useful.
268
269Q:
270 What happens to devices during swsusp? They seem to be resumed
271 during system suspend?
272
273A:
274 That's correct. We need to resume them if we want to write image to
275 disk. Whole sequence goes like
276
277 **Suspend part**
278
279 running system, user asks for suspend-to-disk
280
281 user processes are stopped
282
283 suspend(PMSG_FREEZE): devices are frozen so that they don't interfere
284 with state snapshot
285
1992b66d
BH
286 state snapshot: copy of whole used memory is taken with interrupts
287 disabled
151f4e2b
MCC
288
289 resume(): devices are woken up so that we can write image to swap
290
291 write image to swap
292
293 suspend(PMSG_SUSPEND): suspend devices so that we can power off
294
295 turn the power off
296
297 **Resume part**
298
299 (is actually pretty similar)
300
301 running system, user asks for suspend-to-disk
302
303 user processes are stopped (in common case there are none,
304 but with resume-from-initrd, no one knows)
305
306 read image from disk
307
308 suspend(PMSG_FREEZE): devices are frozen so that they don't interfere
309 with image restoration
310
311 image restoration: rewrite memory with image
312
313 resume(): devices are woken up so that system can continue
314
315 thaw all user processes
316
317Q:
318 What is this 'Encrypt suspend image' for?
319
320A:
321 First of all: it is not a replacement for dm-crypt encrypted swap.
322 It cannot protect your computer while it is suspended. Instead it does
323 protect from leaking sensitive data after resume from suspend.
324
325 Think of the following: you suspend while an application is running
326 that keeps sensitive data in memory. The application itself prevents
327 the data from being swapped out. Suspend, however, must write these
328 data to swap to be able to resume later on. Without suspend encryption
329 your sensitive data are then stored in plaintext on disk. This means
330 that after resume your sensitive data are accessible to all
331 applications having direct access to the swap device which was used
332 for suspend. If you don't need swap after resume these data can remain
333 on disk virtually forever. Thus it can happen that your system gets
334 broken in weeks later and sensitive data which you thought were
335 encrypted and protected are retrieved and stolen from the swap device.
336 To prevent this situation you should use 'Encrypt suspend image'.
337
338 During suspend a temporary key is created and this key is used to
339 encrypt the data written to disk. When, during resume, the data was
340 read back into memory the temporary key is destroyed which simply
341 means that all data written to disk during suspend are then
342 inaccessible so they can't be stolen later on. The only thing that
343 you must then take care of is that you call 'mkswap' for the swap
344 partition used for suspend as early as possible during regular
345 boot. This asserts that any temporary key from an oopsed suspend or
346 from a failed or aborted resume is erased from the swap device.
347
348 As a rule of thumb use encrypted swap to protect your data while your
349 system is shut down or suspended. Additionally use the encrypted
350 suspend image to prevent sensitive data from being stolen after
351 resume.
352
353Q:
354 Can I suspend to a swap file?
355
356A:
357 Generally, yes, you can. However, it requires you to use the "resume=" and
1992b66d
BH
358 "resume_offset=" kernel command line parameters, so the resume from a swap
359 file cannot be initiated from an initrd or initramfs image. See
151f4e2b
MCC
360 swsusp-and-swap-files.txt for details.
361
362Q:
363 Is there a maximum system RAM size that is supported by swsusp?
364
365A:
366 It should work okay with highmem.
367
368Q:
369 Does swsusp (to disk) use only one swap partition or can it use
370 multiple swap partitions (aggregate them into one logical space)?
371
372A:
373 Only one swap partition, sorry.
374
375Q:
376 If my application(s) causes lots of memory & swap space to be used
377 (over half of the total system RAM), is it correct that it is likely
378 to be useless to try to suspend to disk while that app is running?
379
380A:
381 No, it should work okay, as long as your app does not mlock()
382 it. Just prepare big enough swap partition.
383
384Q:
385 What information is useful for debugging suspend-to-disk problems?
386
387A:
388 Well, last messages on the screen are always useful. If something
389 is broken, it is usually some kernel driver, therefore trying with as
390 little as possible modules loaded helps a lot. I also prefer people to
391 suspend from console, preferably without X running. Booting with
392 init=/bin/bash, then swapon and starting suspend sequence manually
393 usually does the trick. Then it is good idea to try with latest
394 vanilla kernel.
395
396Q:
397 How can distributions ship a swsusp-supporting kernel with modular
398 disk drivers (especially SATA)?
399
400A:
401 Well, it can be done, load the drivers, then do echo into
402 /sys/power/resume file from initrd. Be sure not to mount
403 anything, not even read-only mount, or you are going to lose your
404 data.
405
406Q:
407 How do I make suspend more verbose?
408
409A:
410 If you want to see any non-error kernel messages on the virtual
411 terminal the kernel switches to during suspend, you have to set the
412 kernel console loglevel to at least 4 (KERN_WARNING), for example by
413 doing::
414
415 # save the old loglevel
416 read LOGLEVEL DUMMY < /proc/sys/kernel/printk
417 # set the loglevel so we see the progress bar.
418 # if the level is higher than needed, we leave it alone.
419 if [ $LOGLEVEL -lt 5 ]; then
420 echo 5 > /proc/sys/kernel/printk
421 fi
422
423 IMG_SZ=0
424 read IMG_SZ < /sys/power/image_size
425 echo -n disk > /sys/power/state
426 RET=$?
427 #
428 # the logic here is:
429 # if image_size > 0 (without kernel support, IMG_SZ will be zero),
430 # then try again with image_size set to zero.
431 if [ $RET -ne 0 -a $IMG_SZ -ne 0 ]; then # try again with minimal image size
432 echo 0 > /sys/power/image_size
433 echo -n disk > /sys/power/state
434 RET=$?
435 fi
436
437 # restore previous loglevel
438 echo $LOGLEVEL > /proc/sys/kernel/printk
439 exit $RET
440
441Q:
442 Is this true that if I have a mounted filesystem on a USB device and
443 I suspend to disk, I can lose data unless the filesystem has been mounted
444 with "sync"?
445
446A:
447 That's right ... if you disconnect that device, you may lose data.
448 In fact, even with "-o sync" you can lose data if your programs have
449 information in buffers they haven't written out to a disk you disconnect,
450 or if you disconnect before the device finished saving data you wrote.
451
452 Software suspend normally powers down USB controllers, which is equivalent
453 to disconnecting all USB devices attached to your system.
454
455 Your system might well support low-power modes for its USB controllers
456 while the system is asleep, maintaining the connection, using true sleep
457 modes like "suspend-to-RAM" or "standby". (Don't write "disk" to the
458 /sys/power/state file; write "standby" or "mem".) We've not seen any
459 hardware that can use these modes through software suspend, although in
460 theory some systems might support "platform" modes that won't break the
461 USB connections.
462
463 Remember that it's always a bad idea to unplug a disk drive containing a
464 mounted filesystem. That's true even when your system is asleep! The
465 safest thing is to unmount all filesystems on removable media (such USB,
466 Firewire, CompactFlash, MMC, external SATA, or even IDE hotplug bays)
467 before suspending; then remount them after resuming.
468
469 There is a work-around for this problem. For more information, see
470 Documentation/driver-api/usb/persist.rst.
471
472Q:
473 Can I suspend-to-disk using a swap partition under LVM?
474
475A:
476 Yes and No. You can suspend successfully, but the kernel will not be able
477 to resume on its own. You need an initramfs that can recognize the resume
478 situation, activate the logical volume containing the swap volume (but not
479 touch any filesystems!), and eventually call::
480
481 echo -n "$major:$minor" > /sys/power/resume
482
483 where $major and $minor are the respective major and minor device numbers of
484 the swap volume.
485
486 uswsusp works with LVM, too. See http://suspend.sourceforge.net/
487
488Q:
489 I upgraded the kernel from 2.6.15 to 2.6.16. Both kernels were
490 compiled with the similar configuration files. Anyway I found that
491 suspend to disk (and resume) is much slower on 2.6.16 compared to
492 2.6.15. Any idea for why that might happen or how can I speed it up?
493
494A:
495 This is because the size of the suspend image is now greater than
496 for 2.6.15 (by saving more data we can get more responsive system
497 after resume).
498
499 There's the /sys/power/image_size knob that controls the size of the
500 image. If you set it to 0 (eg. by echo 0 > /sys/power/image_size as
501 root), the 2.6.15 behavior should be restored. If it is still too
502 slow, take a look at suspend.sf.net -- userland suspend is faster and
503 supports LZF compression to speed it up further.