1Some warnings, first.
3 * BIG FAT WARNING *********************************************************
4 *
5 * If you touch anything on disk between suspend and resume...
6 * ...kiss your data goodbye.
7 *
8 * If you do resume from initrd after your filesystems are mounted...
9 * ...bye bye root partition.
10 * [this is actually same case as above]
11 *
12 * If you have unsupported (*) devices using DMA, you may have some
13 * problems. If your disk driver does not support suspend... (IDE does),
14 * it may cause some problems, too. If you change kernel command line
15 * between suspend and resume, it may do something wrong. If you change
16 * your hardware while system is suspended... well, it was not good idea;
17 * but it will probably only crash.
18 *
19 * (*) suspend/resume support is needed to make it safe.
20 *
21 * If you have any filesystems on USB devices mounted before software suspend,
22 * they won't be accessible after resume and you may lose data, as though
23 * you have unplugged the USB devices with mounted filesystems on them;
24 * see the FAQ below for details. (This is not true for more traditional
25 * power states like "standby", which normally don't turn USB off.)
27You need to append resume=/dev/your_swap_partition to kernel command
28line. Then you suspend by
30echo shutdown > /sys/power/disk; echo disk > /sys/power/state
32. If you feel ACPI works pretty well on your system, you might try
34echo platform > /sys/power/disk; echo disk > /sys/power/state
36. If you have SATA disks, you'll need recent kernels with SATA suspend
37support. For suspend and resume to work, make sure your disk drivers
38are built into kernel -- not modules. [There's way to make
39suspend/resume with modular disk drivers, see FAQ, but you probably
40should not do that.]
42If you want to limit the suspend image size to N bytes, do
44echo N > /sys/power/image_size
46before suspend (it is limited to 500 MB by default).
49Article about goals and implementation of Software Suspend for Linux
51Author: G‚ábor Kuti
52Last revised: 2003-10-20 by Pavel Machek
54Idea and goals to achieve
56Nowadays it is common in several laptops that they have a suspend button. It
57saves the state of the machine to a filesystem or to a partition and switches
58to standby mode. Later resuming the machine the saved state is loaded back to
59ram and the machine can continue its work. It has two real benefits. First we
60save ourselves the time machine goes down and later boots up, energy costs
61are real high when running from batteries. The other gain is that we don't have to
62interrupt our programs so processes that are calculating something for a long
63time shouldn't need to be written interruptible.
65swsusp saves the state of the machine into active swaps and then reboots or
66powerdowns. You must explicitly specify the swap partition to resume from with
67``resume='' kernel option. If signature is found it loads and restores saved
68state. If the option ``noresume'' is specified as a boot parameter, it skips
69the resuming. If the option ``hibernate=nocompress'' is specified as a boot
70parameter, it saves hibernation image without compression.
72In the meantime while the system is suspended you should not add/remove any
73of the hardware, write to the filesystems, etc.
75Sleep states summary
78There are three different interfaces you can use, /proc/acpi should
79work like this:
81In a really perfect world:
82echo 1 > /proc/acpi/sleep # for standby
83echo 2 > /proc/acpi/sleep # for suspend to ram
84echo 3 > /proc/acpi/sleep # for suspend to ram, but with more power conservative
85echo 4 > /proc/acpi/sleep # for suspend to disk
86echo 5 > /proc/acpi/sleep # for shutdown unfriendly the system
88and perhaps
89echo 4b > /proc/acpi/sleep # for suspend to disk via s4bios
91Frequently Asked Questions
94Q: well, suspending a server is IMHO a really stupid thing,
95but... (Diego Zuccato):
97A: You bought new UPS for your server. How do you install it without
98bringing machine down? Suspend to disk, rearrange power cables,
101You have your server on UPS. Power died, and UPS is indicating 30
102seconds to failure. What do you do? Suspend to disk.
105Q: Maybe I'm missing something, but why don't the regular I/O paths work?
107A: We do use the regular I/O paths. However we cannot restore the data
108to its original location as we load it. That would create an
109inconsistent kernel state which would certainly result in an oops.
110Instead, we load the image into unused memory and then atomically copy
111it back to it original location. This implies, of course, a maximum
112image size of half the amount of memory.
114There are two solutions to this:
116* require half of memory to be free during suspend. That way you can
117read "new" data onto free spots, then cli and copy
119* assume we had special "polling" ide driver that only uses memory
120between 0-640KB. That way, I'd have to make sure that 0-640KB is free
121during suspending, but otherwise it would work...
123suspend2 shares this fundamental limitation, but does not include user
124data and disk caches into "used memory" by saving them in
125advance. That means that the limitation goes away in practice.
127Q: Does linux support ACPI S4?
129A: Yes. That's what echo platform > /sys/power/disk does.
131Q: What is 'suspend2'?
133A: suspend2 is 'Software Suspend 2', a forked implementation of
134suspend-to-disk which is available as separate patches for 2.4 and 2.6
135kernels from It includes support for SMP, 4GB
136highmem and preemption. It also has a extensible architecture that
137allows for arbitrary transformations on the image (compression,
138encryption) and arbitrary backends for writing the image (eg to swap
139or an NFS share[Work In Progress]). Questions regarding suspend2
140should be sent to the mailing list available through the suspend2
141website, and not to the Linux Kernel Mailing List. We are working
142toward merging suspend2 into the mainline kernel.
144Q: What is the freezing of tasks and why are we using it?
146A: The freezing of tasks is a mechanism by which user space processes and some
147kernel threads are controlled during hibernation or system-wide suspend (on some
148architectures). See freezing-of-tasks.txt for details.
150Q: What is the difference between "platform" and "shutdown"?
154shutdown: save state in linux, then tell bios to powerdown
156platform: save state in linux, then tell bios to powerdown and blink
157          "suspended led"
159"platform" is actually right thing to do where supported, but
160"shutdown" is most reliable (except on ACPI systems).
162Q: I do not understand why you have such strong objections to idea of
163selective suspend.
165A: Do selective suspend during runtime power management, that's okay. But
166it's useless for suspend-to-disk. (And I do not see how you could use
167it for suspend-to-ram, I hope you do not want that).
169Lets see, so you suggest to
171* SUSPEND all but swap device and parents
172* Snapshot
173* Write image to disk
174* SUSPEND swap device and parents
175* Powerdown
177Oh no, that does not work, if swap device or its parents uses DMA,
178you've corrupted data. You'd have to do
180* SUSPEND all but swap device and parents
181* FREEZE swap device and parents
182* Snapshot
183* UNFREEZE swap device and parents
184* Write
185* SUSPEND swap device and parents
187Which means that you still need that FREEZE state, and you get more
188complicated code. (And I have not yet introduce details like system
191Q: There don't seem to be any generally useful behavioral
192distinctions between SUSPEND and FREEZE.
194A: Doing SUSPEND when you are asked to do FREEZE is always correct,
195but it may be unnecessarily slow. If you want your driver to stay simple,
196slowness may not matter to you. It can always be fixed later.
198For devices like disk it does matter, you do not want to spindown for
201Q: After resuming, system is paging heavily, leading to very bad interactivity.
203A: Try running
205cat `cat /proc/[0-9]*/maps | grep / | sed 's:.* /:/:' | sort -u` > /dev/null
207after resume. swapoff -a; swapon -a may also be useful.
209Q: What happens to devices during swsusp? They seem to be resumed
210during system suspend?
212A: That's correct. We need to resume them if we want to write image to
213disk. Whole sequence goes like
215      Suspend part
216      ~~~~~~~~~~~~
217      running system, user asks for suspend-to-disk
219      user processes are stopped
221      suspend(PMSG_FREEZE): devices are frozen so that they don't interfere
222                    with state snapshot
224      state snapshot: copy of whole used memory is taken with interrupts disabled
226      resume(): devices are woken up so that we can write image to swap
228      write image to swap
230      suspend(PMSG_SUSPEND): suspend devices so that we can power off
232      turn the power off
234      Resume part
235      ~~~~~~~~~~~
236      (is actually pretty similar)
238      running system, user asks for suspend-to-disk
240      user processes are stopped (in common case there are none, but with resume-from-initrd, no one knows)
242      read image from disk
244      suspend(PMSG_FREEZE): devices are frozen so that they don't interfere
245                    with image restoration
247      image restoration: rewrite memory with image
249      resume(): devices are woken up so that system can continue
251      thaw all user processes
253Q: What is this 'Encrypt suspend image' for?
255A: First of all: it is not a replacement for dm-crypt encrypted swap.
256It cannot protect your computer while it is suspended. Instead it does
257protect from leaking sensitive data after resume from suspend.
259Think of the following: you suspend while an application is running
260that keeps sensitive data in memory. The application itself prevents
261the data from being swapped out. Suspend, however, must write these
262data to swap to be able to resume later on. Without suspend encryption
263your sensitive data are then stored in plaintext on disk. This means
264that after resume your sensitive data are accessible to all
265applications having direct access to the swap device which was used
266for suspend. If you don't need swap after resume these data can remain
267on disk virtually forever. Thus it can happen that your system gets
268broken in weeks later and sensitive data which you thought were
269encrypted and protected are retrieved and stolen from the swap device.
270To prevent this situation you should use 'Encrypt suspend image'.
272During suspend a temporary key is created and this key is used to
273encrypt the data written to disk. When, during resume, the data was
274read back into memory the temporary key is destroyed which simply
275means that all data written to disk during suspend are then
276inaccessible so they can't be stolen later on. The only thing that
277you must then take care of is that you call 'mkswap' for the swap
278partition used for suspend as early as possible during regular
279boot. This asserts that any temporary key from an oopsed suspend or
280from a failed or aborted resume is erased from the swap device.
282As a rule of thumb use encrypted swap to protect your data while your
283system is shut down or suspended. Additionally use the encrypted
284suspend image to prevent sensitive data from being stolen after
287Q: Can I suspend to a swap file?
289A: Generally, yes, you can. However, it requires you to use the "resume=" and
290"resume_offset=" kernel command line parameters, so the resume from a swap file
291cannot be initiated from an initrd or initramfs image. See
292swsusp-and-swap-files.txt for details.
294Q: Is there a maximum system RAM size that is supported by swsusp?
296A: It should work okay with highmem.
298Q: Does swsusp (to disk) use only one swap partition or can it use
299multiple swap partitions (aggregate them into one logical space)?
301A: Only one swap partition, sorry.
303Q: If my application(s) causes lots of memory & swap space to be used
304(over half of the total system RAM), is it correct that it is likely
305to be useless to try to suspend to disk while that app is running?
307A: No, it should work okay, as long as your app does not mlock()
308it. Just prepare big enough swap partition.
310Q: What information is useful for debugging suspend-to-disk problems?
312A: Well, last messages on the screen are always useful. If something
313is broken, it is usually some kernel driver, therefore trying with as
314little as possible modules loaded helps a lot. I also prefer people to
315suspend from console, preferably without X running. Booting with
316init=/bin/bash, then swapon and starting suspend sequence manually
317usually does the trick. Then it is good idea to try with latest
318vanilla kernel.
320Q: How can distributions ship a swsusp-supporting kernel with modular
321disk drivers (especially SATA)?
323A: Well, it can be done, load the drivers, then do echo into
324/sys/power/disk/resume file from initrd. Be sure not to mount
325anything, not even read-only mount, or you are going to lose your
328Q: How do I make suspend more verbose?
330A: If you want to see any non-error kernel messages on the virtual
331terminal the kernel switches to during suspend, you have to set the
332kernel console loglevel to at least 4 (KERN_WARNING), for example by
335    # save the old loglevel
336    read LOGLEVEL DUMMY < /proc/sys/kernel/printk
337    # set the loglevel so we see the progress bar.
338    # if the level is higher than needed, we leave it alone.
339    if [ $LOGLEVEL -lt 5 ]; then
340            echo 5 > /proc/sys/kernel/printk
341        fi
343        IMG_SZ=0
344        read IMG_SZ < /sys/power/image_size
345        echo -n disk > /sys/power/state
346        RET=$?
347        #
348        # the logic here is:
349        # if image_size > 0 (without kernel support, IMG_SZ will be zero),
350        # then try again with image_size set to zero.
351    if [ $RET -ne 0 -a $IMG_SZ -ne 0 ]; then # try again with minimal image size
352                echo 0 > /sys/power/image_size
353                echo -n disk > /sys/power/state
354                RET=$?
355        fi
357    # restore previous loglevel
358    echo $LOGLEVEL > /proc/sys/kernel/printk
359    exit $RET
361Q: Is this true that if I have a mounted filesystem on a USB device and
362I suspend to disk, I can lose data unless the filesystem has been mounted
363with "sync"?
365A: That's right ... if you disconnect that device, you may lose data.
366In fact, even with "-o sync" you can lose data if your programs have
367information in buffers they haven't written out to a disk you disconnect,
368or if you disconnect before the device finished saving data you wrote.
370Software suspend normally powers down USB controllers, which is equivalent
371to disconnecting all USB devices attached to your system.
373Your system might well support low-power modes for its USB controllers
374while the system is asleep, maintaining the connection, using true sleep
375modes like "suspend-to-RAM" or "standby". (Don't write "disk" to the
376/sys/power/state file; write "standby" or "mem".) We've not seen any
377hardware that can use these modes through software suspend, although in
378theory some systems might support "platform" modes that won't break the
379USB connections.
381Remember that it's always a bad idea to unplug a disk drive containing a
382mounted filesystem. That's true even when your system is asleep! The
383safest thing is to unmount all filesystems on removable media (such USB,
384Firewire, CompactFlash, MMC, external SATA, or even IDE hotplug bays)
385before suspending; then remount them after resuming.
387There is a work-around for this problem. For more information, see
390Q: Can I suspend-to-disk using a swap partition under LVM?
392A: No. You can suspend successfully, but you'll not be able to
393resume. uswsusp should be able to work with LVM. See
395Q: I upgraded the kernel from 2.6.15 to 2.6.16. Both kernels were
396compiled with the similar configuration files. Anyway I found that
397suspend to disk (and resume) is much slower on 2.6.16 compared to
3982.6.15. Any idea for why that might happen or how can I speed it up?
400A: This is because the size of the suspend image is now greater than
401for 2.6.15 (by saving more data we can get more responsive system
402after resume).
404There's the /sys/power/image_size knob that controls the size of the
405image. If you set it to 0 (eg. by echo 0 > /sys/power/image_size as
406root), the 2.6.15 behavior should be restored. If it is still too
407slow, take a look at -- userland suspend is faster and
408supports LZF compression to speed it up further.

Archive Download this file