swsusp.txt 16 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407
  1. Some warnings, first.
  2. * BIG FAT WARNING *********************************************************
  3. *
  4. * If you touch anything on disk between suspend and resume...
  5. * ...kiss your data goodbye.
  6. *
  7. * If you do resume from initrd after your filesystems are mounted...
  8. * ...bye bye root partition.
  9. * [this is actually same case as above]
  10. *
  11. * If you have unsupported (*) devices using DMA, you may have some
  12. * problems. If your disk driver does not support suspend... (IDE does),
  13. * it may cause some problems, too. If you change kernel command line
  14. * between suspend and resume, it may do something wrong. If you change
  15. * your hardware while system is suspended... well, it was not good idea;
  16. * but it will probably only crash.
  17. *
  18. * (*) suspend/resume support is needed to make it safe.
  19. *
  20. * If you have any filesystems on USB devices mounted before software suspend,
  21. * they won't be accessible after resume and you may lose data, as though
  22. * you have unplugged the USB devices with mounted filesystems on them;
  23. * see the FAQ below for details. (This is not true for more traditional
  24. * power states like "standby", which normally don't turn USB off.)
  25. You need to append resume=/dev/your_swap_partition to kernel command
  26. line. Then you suspend by
  27. echo shutdown > /sys/power/disk; echo disk > /sys/power/state
  28. . If you feel ACPI works pretty well on your system, you might try
  29. echo platform > /sys/power/disk; echo disk > /sys/power/state
  30. . If you have SATA disks, you'll need recent kernels with SATA suspend
  31. support. For suspend and resume to work, make sure your disk drivers
  32. are built into kernel -- not modules. [There's way to make
  33. suspend/resume with modular disk drivers, see FAQ, but you probably
  34. should not do that.]
  35. If you want to limit the suspend image size to N bytes, do
  36. echo N > /sys/power/image_size
  37. before suspend (it is limited to 500 MB by default).
  38. Article about goals and implementation of Software Suspend for Linux
  39. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  40. Author: G‚ábor Kuti
  41. Last revised: 2003-10-20 by Pavel Machek
  42. Idea and goals to achieve
  43. Nowadays it is common in several laptops that they have a suspend button. It
  44. saves the state of the machine to a filesystem or to a partition and switches
  45. to standby mode. Later resuming the machine the saved state is loaded back to
  46. ram and the machine can continue its work. It has two real benefits. First we
  47. save ourselves the time machine goes down and later boots up, energy costs
  48. are real high when running from batteries. The other gain is that we don't have to
  49. interrupt our programs so processes that are calculating something for a long
  50. time shouldn't need to be written interruptible.
  51. swsusp saves the state of the machine into active swaps and then reboots or
  52. powerdowns. You must explicitly specify the swap partition to resume from with
  53. ``resume='' kernel option. If signature is found it loads and restores saved
  54. state. If the option ``noresume'' is specified as a boot parameter, it skips
  55. the resuming.
  56. In the meantime while the system is suspended you should not add/remove any
  57. of the hardware, write to the filesystems, etc.
  58. Sleep states summary
  59. ====================
  60. There are three different interfaces you can use, /proc/acpi should
  61. work like this:
  62. In a really perfect world:
  63. echo 1 > /proc/acpi/sleep # for standby
  64. echo 2 > /proc/acpi/sleep # for suspend to ram
  65. echo 3 > /proc/acpi/sleep # for suspend to ram, but with more power conservative
  66. echo 4 > /proc/acpi/sleep # for suspend to disk
  67. echo 5 > /proc/acpi/sleep # for shutdown unfriendly the system
  68. and perhaps
  69. echo 4b > /proc/acpi/sleep # for suspend to disk via s4bios
  70. Frequently Asked Questions
  71. ==========================
  72. Q: well, suspending a server is IMHO a really stupid thing,
  73. but... (Diego Zuccato):
  74. A: You bought new UPS for your server. How do you install it without
  75. bringing machine down? Suspend to disk, rearrange power cables,
  76. resume.
  77. You have your server on UPS. Power died, and UPS is indicating 30
  78. seconds to failure. What do you do? Suspend to disk.
  79. Q: Maybe I'm missing something, but why don't the regular I/O paths work?
  80. A: We do use the regular I/O paths. However we cannot restore the data
  81. to its original location as we load it. That would create an
  82. inconsistent kernel state which would certainly result in an oops.
  83. Instead, we load the image into unused memory and then atomically copy
  84. it back to it original location. This implies, of course, a maximum
  85. image size of half the amount of memory.
  86. There are two solutions to this:
  87. * require half of memory to be free during suspend. That way you can
  88. read "new" data onto free spots, then cli and copy
  89. * assume we had special "polling" ide driver that only uses memory
  90. between 0-640KB. That way, I'd have to make sure that 0-640KB is free
  91. during suspending, but otherwise it would work...
  92. suspend2 shares this fundamental limitation, but does not include user
  93. data and disk caches into "used memory" by saving them in
  94. advance. That means that the limitation goes away in practice.
  95. Q: Does linux support ACPI S4?
  96. A: Yes. That's what echo platform > /sys/power/disk does.
  97. Q: What is 'suspend2'?
  98. A: suspend2 is 'Software Suspend 2', a forked implementation of
  99. suspend-to-disk which is available as separate patches for 2.4 and 2.6
  100. kernels from swsusp.sourceforge.net. It includes support for SMP, 4GB
  101. highmem and preemption. It also has a extensible architecture that
  102. allows for arbitrary transformations on the image (compression,
  103. encryption) and arbitrary backends for writing the image (eg to swap
  104. or an NFS share[Work In Progress]). Questions regarding suspend2
  105. should be sent to the mailing list available through the suspend2
  106. website, and not to the Linux Kernel Mailing List. We are working
  107. toward merging suspend2 into the mainline kernel.
  108. Q: What is the freezing of tasks and why are we using it?
  109. A: The freezing of tasks is a mechanism by which user space processes and some
  110. kernel threads are controlled during hibernation or system-wide suspend (on some
  111. architectures). See freezing-of-tasks.txt for details.
  112. Q: What is the difference between "platform" and "shutdown"?
  113. A:
  114. shutdown: save state in linux, then tell bios to powerdown
  115. platform: save state in linux, then tell bios to powerdown and blink
  116. "suspended led"
  117. "platform" is actually right thing to do where supported, but
  118. "shutdown" is most reliable (except on ACPI systems).
  119. Q: I do not understand why you have such strong objections to idea of
  120. selective suspend.
  121. A: Do selective suspend during runtime power management, that's okay. But
  122. it's useless for suspend-to-disk. (And I do not see how you could use
  123. it for suspend-to-ram, I hope you do not want that).
  124. Lets see, so you suggest to
  125. * SUSPEND all but swap device and parents
  126. * Snapshot
  127. * Write image to disk
  128. * SUSPEND swap device and parents
  129. * Powerdown
  130. Oh no, that does not work, if swap device or its parents uses DMA,
  131. you've corrupted data. You'd have to do
  132. * SUSPEND all but swap device and parents
  133. * FREEZE swap device and parents
  134. * Snapshot
  135. * UNFREEZE swap device and parents
  136. * Write
  137. * SUSPEND swap device and parents
  138. Which means that you still need that FREEZE state, and you get more
  139. complicated code. (And I have not yet introduce details like system
  140. devices).
  141. Q: There don't seem to be any generally useful behavioral
  142. distinctions between SUSPEND and FREEZE.
  143. A: Doing SUSPEND when you are asked to do FREEZE is always correct,
  144. but it may be unneccessarily slow. If you want your driver to stay simple,
  145. slowness may not matter to you. It can always be fixed later.
  146. For devices like disk it does matter, you do not want to spindown for
  147. FREEZE.
  148. Q: After resuming, system is paging heavily, leading to very bad interactivity.
  149. A: Try running
  150. cat `cat /proc/[0-9]*/maps | grep / | sed 's:.* /:/:' | sort -u` > /dev/null
  151. after resume. swapoff -a; swapon -a may also be useful.
  152. Q: What happens to devices during swsusp? They seem to be resumed
  153. during system suspend?
  154. A: That's correct. We need to resume them if we want to write image to
  155. disk. Whole sequence goes like
  156. Suspend part
  157. ~~~~~~~~~~~~
  158. running system, user asks for suspend-to-disk
  159. user processes are stopped
  160. suspend(PMSG_FREEZE): devices are frozen so that they don't interfere
  161. with state snapshot
  162. state snapshot: copy of whole used memory is taken with interrupts disabled
  163. resume(): devices are woken up so that we can write image to swap
  164. write image to swap
  165. suspend(PMSG_SUSPEND): suspend devices so that we can power off
  166. turn the power off
  167. Resume part
  168. ~~~~~~~~~~~
  169. (is actually pretty similar)
  170. running system, user asks for suspend-to-disk
  171. user processes are stopped (in common case there are none, but with resume-from-initrd, noone knows)
  172. read image from disk
  173. suspend(PMSG_FREEZE): devices are frozen so that they don't interfere
  174. with image restoration
  175. image restoration: rewrite memory with image
  176. resume(): devices are woken up so that system can continue
  177. thaw all user processes
  178. Q: What is this 'Encrypt suspend image' for?
  179. A: First of all: it is not a replacement for dm-crypt encrypted swap.
  180. It cannot protect your computer while it is suspended. Instead it does
  181. protect from leaking sensitive data after resume from suspend.
  182. Think of the following: you suspend while an application is running
  183. that keeps sensitive data in memory. The application itself prevents
  184. the data from being swapped out. Suspend, however, must write these
  185. data to swap to be able to resume later on. Without suspend encryption
  186. your sensitive data are then stored in plaintext on disk. This means
  187. that after resume your sensitive data are accessible to all
  188. applications having direct access to the swap device which was used
  189. for suspend. If you don't need swap after resume these data can remain
  190. on disk virtually forever. Thus it can happen that your system gets
  191. broken in weeks later and sensitive data which you thought were
  192. encrypted and protected are retrieved and stolen from the swap device.
  193. To prevent this situation you should use 'Encrypt suspend image'.
  194. During suspend a temporary key is created and this key is used to
  195. encrypt the data written to disk. When, during resume, the data was
  196. read back into memory the temporary key is destroyed which simply
  197. means that all data written to disk during suspend are then
  198. inaccessible so they can't be stolen later on. The only thing that
  199. you must then take care of is that you call 'mkswap' for the swap
  200. partition used for suspend as early as possible during regular
  201. boot. This asserts that any temporary key from an oopsed suspend or
  202. from a failed or aborted resume is erased from the swap device.
  203. As a rule of thumb use encrypted swap to protect your data while your
  204. system is shut down or suspended. Additionally use the encrypted
  205. suspend image to prevent sensitive data from being stolen after
  206. resume.
  207. Q: Can I suspend to a swap file?
  208. A: Generally, yes, you can. However, it requires you to use the "resume=" and
  209. "resume_offset=" kernel command line parameters, so the resume from a swap file
  210. cannot be initiated from an initrd or initramfs image. See
  211. swsusp-and-swap-files.txt for details.
  212. Q: Is there a maximum system RAM size that is supported by swsusp?
  213. A: It should work okay with highmem.
  214. Q: Does swsusp (to disk) use only one swap partition or can it use
  215. multiple swap partitions (aggregate them into one logical space)?
  216. A: Only one swap partition, sorry.
  217. Q: If my application(s) causes lots of memory & swap space to be used
  218. (over half of the total system RAM), is it correct that it is likely
  219. to be useless to try to suspend to disk while that app is running?
  220. A: No, it should work okay, as long as your app does not mlock()
  221. it. Just prepare big enough swap partition.
  222. Q: What information is useful for debugging suspend-to-disk problems?
  223. A: Well, last messages on the screen are always useful. If something
  224. is broken, it is usually some kernel driver, therefore trying with as
  225. little as possible modules loaded helps a lot. I also prefer people to
  226. suspend from console, preferably without X running. Booting with
  227. init=/bin/bash, then swapon and starting suspend sequence manually
  228. usually does the trick. Then it is good idea to try with latest
  229. vanilla kernel.
  230. Q: How can distributions ship a swsusp-supporting kernel with modular
  231. disk drivers (especially SATA)?
  232. A: Well, it can be done, load the drivers, then do echo into
  233. /sys/power/disk/resume file from initrd. Be sure not to mount
  234. anything, not even read-only mount, or you are going to lose your
  235. data.
  236. Q: How do I make suspend more verbose?
  237. A: If you want to see any non-error kernel messages on the virtual
  238. terminal the kernel switches to during suspend, you have to set the
  239. kernel console loglevel to at least 4 (KERN_WARNING), for example by
  240. doing
  241. # save the old loglevel
  242. read LOGLEVEL DUMMY < /proc/sys/kernel/printk
  243. # set the loglevel so we see the progress bar.
  244. # if the level is higher than needed, we leave it alone.
  245. if [ $LOGLEVEL -lt 5 ]; then
  246. echo 5 > /proc/sys/kernel/printk
  247. fi
  248. IMG_SZ=0
  249. read IMG_SZ < /sys/power/image_size
  250. echo -n disk > /sys/power/state
  251. RET=$?
  252. #
  253. # the logic here is:
  254. # if image_size > 0 (without kernel support, IMG_SZ will be zero),
  255. # then try again with image_size set to zero.
  256. if [ $RET -ne 0 -a $IMG_SZ -ne 0 ]; then # try again with minimal image size
  257. echo 0 > /sys/power/image_size
  258. echo -n disk > /sys/power/state
  259. RET=$?
  260. fi
  261. # restore previous loglevel
  262. echo $LOGLEVEL > /proc/sys/kernel/printk
  263. exit $RET
  264. Q: Is this true that if I have a mounted filesystem on a USB device and
  265. I suspend to disk, I can lose data unless the filesystem has been mounted
  266. with "sync"?
  267. A: That's right ... if you disconnect that device, you may lose data.
  268. In fact, even with "-o sync" you can lose data if your programs have
  269. information in buffers they haven't written out to a disk you disconnect,
  270. or if you disconnect before the device finished saving data you wrote.
  271. Software suspend normally powers down USB controllers, which is equivalent
  272. to disconnecting all USB devices attached to your system.
  273. Your system might well support low-power modes for its USB controllers
  274. while the system is asleep, maintaining the connection, using true sleep
  275. modes like "suspend-to-RAM" or "standby". (Don't write "disk" to the
  276. /sys/power/state file; write "standby" or "mem".) We've not seen any
  277. hardware that can use these modes through software suspend, although in
  278. theory some systems might support "platform" modes that won't break the
  279. USB connections.
  280. Remember that it's always a bad idea to unplug a disk drive containing a
  281. mounted filesystem. That's true even when your system is asleep! The
  282. safest thing is to unmount all filesystems on removable media (such USB,
  283. Firewire, CompactFlash, MMC, external SATA, or even IDE hotplug bays)
  284. before suspending; then remount them after resuming.
  285. There is a work-around for this problem. For more information, see
  286. Documentation/usb/persist.txt.
  287. Q: Can I suspend-to-disk using a swap partition under LVM?
  288. A: No. You can suspend successfully, but you'll not be able to
  289. resume. uswsusp should be able to work with LVM. See suspend.sf.net.
  290. Q: I upgraded the kernel from 2.6.15 to 2.6.16. Both kernels were
  291. compiled with the similar configuration files. Anyway I found that
  292. suspend to disk (and resume) is much slower on 2.6.16 compared to
  293. 2.6.15. Any idea for why that might happen or how can I speed it up?
  294. A: This is because the size of the suspend image is now greater than
  295. for 2.6.15 (by saving more data we can get more responsive system
  296. after resume).
  297. There's the /sys/power/image_size knob that controls the size of the
  298. image. If you set it to 0 (eg. by echo 0 > /sys/power/image_size as
  299. root), the 2.6.15 behavior should be restored. If it is still too
  300. slow, take a look at suspend.sf.net -- userland suspend is faster and
  301. supports LZF compression to speed it up further.