linux/Documentation/power/swsusp.txt
<<
>>
Prefs
   1Some warnings, first.
   2
   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.)
  26
  27You need to append resume=/dev/your_swap_partition to kernel command
  28line. Then you suspend by
  29
  30echo shutdown > /sys/power/disk; echo disk > /sys/power/state
  31
  32. If you feel ACPI works pretty well on your system, you might try
  33
  34echo platform > /sys/power/disk; echo disk > /sys/power/state
  35
  36. If you would like to write hibernation image to swap and then suspend
  37to RAM (provided your platform supports it), you can try
  38
  39echo suspend > /sys/power/disk; echo disk > /sys/power/state
  40
  41. If you have SATA disks, you'll need recent kernels with SATA suspend
  42support. For suspend and resume to work, make sure your disk drivers
  43are built into kernel -- not modules. [There's way to make
  44suspend/resume with modular disk drivers, see FAQ, but you probably
  45should not do that.]
  46
  47If you want to limit the suspend image size to N bytes, do
  48
  49echo N > /sys/power/image_size
  50
  51before suspend (it is limited to 500 MB by default).
  52
  53
  54Article about goals and implementation of Software Suspend for Linux
  55~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  56Author: G‚ábor Kuti
  57Last revised: 2003-10-20 by Pavel Machek
  58
  59Idea and goals to achieve
  60
  61Nowadays it is common in several laptops that they have a suspend button. It
  62saves the state of the machine to a filesystem or to a partition and switches
  63to standby mode. Later resuming the machine the saved state is loaded back to
  64ram and the machine can continue its work. It has two real benefits. First we
  65save ourselves the time machine goes down and later boots up, energy costs
  66are real high when running from batteries. The other gain is that we don't have to
  67interrupt our programs so processes that are calculating something for a long
  68time shouldn't need to be written interruptible.
  69
  70swsusp saves the state of the machine into active swaps and then reboots or
  71powerdowns.  You must explicitly specify the swap partition to resume from with
  72``resume='' kernel option. If signature is found it loads and restores saved
  73state. If the option ``noresume'' is specified as a boot parameter, it skips
  74the resuming.  If the option ``hibernate=nocompress'' is specified as a boot
  75parameter, it saves hibernation image without compression.
  76
  77In the meantime while the system is suspended you should not add/remove any
  78of the hardware, write to the filesystems, etc.
  79
  80Sleep states summary
  81====================
  82
  83There are three different interfaces you can use, /proc/acpi should
  84work like this:
  85
  86In a really perfect world:
  87echo 1 > /proc/acpi/sleep       # for standby
  88echo 2 > /proc/acpi/sleep       # for suspend to ram
  89echo 3 > /proc/acpi/sleep       # for suspend to ram, but with more power conservative
  90echo 4 > /proc/acpi/sleep       # for suspend to disk
  91echo 5 > /proc/acpi/sleep       # for shutdown unfriendly the system
  92
  93and perhaps
  94echo 4b > /proc/acpi/sleep      # for suspend to disk via s4bios
  95
  96Frequently Asked Questions
  97==========================
  98
  99Q: well, suspending a server is IMHO a really stupid thing,
 100but... (Diego Zuccato):
 101
 102A: You bought new UPS for your server. How do you install it without
 103bringing machine down? Suspend to disk, rearrange power cables,
 104resume.
 105
 106You have your server on UPS. Power died, and UPS is indicating 30
 107seconds to failure. What do you do? Suspend to disk.
 108
 109
 110Q: Maybe I'm missing something, but why don't the regular I/O paths work?
 111
 112A: We do use the regular I/O paths. However we cannot restore the data
 113to its original location as we load it. That would create an
 114inconsistent kernel state which would certainly result in an oops.
 115Instead, we load the image into unused memory and then atomically copy
 116it back to it original location. This implies, of course, a maximum
 117image size of half the amount of memory.
 118
 119There are two solutions to this:
 120
 121* require half of memory to be free during suspend. That way you can
 122read "new" data onto free spots, then cli and copy
 123
 124* assume we had special "polling" ide driver that only uses memory
 125between 0-640KB. That way, I'd have to make sure that 0-640KB is free
 126during suspending, but otherwise it would work...
 127
 128suspend2 shares this fundamental limitation, but does not include user
 129data and disk caches into "used memory" by saving them in
 130advance. That means that the limitation goes away in practice.
 131
 132Q: Does linux support ACPI S4?
 133
 134A: Yes. That's what echo platform > /sys/power/disk does.
 135
 136Q: What is 'suspend2'?
 137
 138A: suspend2 is 'Software Suspend 2', a forked implementation of
 139suspend-to-disk which is available as separate patches for 2.4 and 2.6
 140kernels from swsusp.sourceforge.net. It includes support for SMP, 4GB
 141highmem and preemption. It also has a extensible architecture that
 142allows for arbitrary transformations on the image (compression,
 143encryption) and arbitrary backends for writing the image (eg to swap
 144or an NFS share[Work In Progress]). Questions regarding suspend2
 145should be sent to the mailing list available through the suspend2
 146website, and not to the Linux Kernel Mailing List. We are working
 147toward merging suspend2 into the mainline kernel.
 148
 149Q: What is the freezing of tasks and why are we using it?
 150
 151A: The freezing of tasks is a mechanism by which user space processes and some
 152kernel threads are controlled during hibernation or system-wide suspend (on some
 153architectures).  See freezing-of-tasks.txt for details.
 154
 155Q: What is the difference between "platform" and "shutdown"?
 156
 157A:
 158
 159shutdown: save state in linux, then tell bios to powerdown
 160
 161platform: save state in linux, then tell bios to powerdown and blink
 162          "suspended led"
 163
 164"platform" is actually right thing to do where supported, but
 165"shutdown" is most reliable (except on ACPI systems).
 166
 167Q: I do not understand why you have such strong objections to idea of
 168selective suspend.
 169
 170A: Do selective suspend during runtime power management, that's okay. But
 171it's useless for suspend-to-disk. (And I do not see how you could use
 172it for suspend-to-ram, I hope you do not want that).
 173
 174Lets see, so you suggest to
 175
 176* SUSPEND all but swap device and parents
 177* Snapshot
 178* Write image to disk
 179* SUSPEND swap device and parents
 180* Powerdown
 181
 182Oh no, that does not work, if swap device or its parents uses DMA,
 183you've corrupted data. You'd have to do
 184
 185* SUSPEND all but swap device and parents
 186* FREEZE swap device and parents
 187* Snapshot
 188* UNFREEZE swap device and parents
 189* Write
 190* SUSPEND swap device and parents
 191
 192Which means that you still need that FREEZE state, and you get more
 193complicated code. (And I have not yet introduce details like system
 194devices).
 195
 196Q: There don't seem to be any generally useful behavioral
 197distinctions between SUSPEND and FREEZE.
 198
 199A: Doing SUSPEND when you are asked to do FREEZE is always correct,
 200but it may be unnecessarily slow. If you want your driver to stay simple,
 201slowness may not matter to you. It can always be fixed later.
 202
 203For devices like disk it does matter, you do not want to spindown for
 204FREEZE.
 205
 206Q: After resuming, system is paging heavily, leading to very bad interactivity.
 207
 208A: Try running
 209
 210cat `cat /proc/[0-9]*/maps | grep / | sed 's:.* /:/:' | sort -u` > /dev/null
 211
 212after resume. swapoff -a; swapon -a may also be useful.
 213
 214Q: What happens to devices during swsusp? They seem to be resumed
 215during system suspend?
 216
 217A: That's correct. We need to resume them if we want to write image to
 218disk. Whole sequence goes like
 219
 220      Suspend part
 221      ~~~~~~~~~~~~
 222      running system, user asks for suspend-to-disk
 223
 224      user processes are stopped
 225
 226      suspend(PMSG_FREEZE): devices are frozen so that they don't interfere
 227                      with state snapshot
 228
 229      state snapshot: copy of whole used memory is taken with interrupts disabled
 230
 231      resume(): devices are woken up so that we can write image to swap
 232
 233      write image to swap
 234
 235      suspend(PMSG_SUSPEND): suspend devices so that we can power off
 236
 237      turn the power off
 238
 239      Resume part
 240      ~~~~~~~~~~~
 241      (is actually pretty similar)
 242
 243      running system, user asks for suspend-to-disk
 244
 245      user processes are stopped (in common case there are none, but with resume-from-initrd, no one knows)
 246
 247      read image from disk
 248
 249      suspend(PMSG_FREEZE): devices are frozen so that they don't interfere
 250                      with image restoration
 251
 252      image restoration: rewrite memory with image
 253
 254      resume(): devices are woken up so that system can continue
 255
 256      thaw all user processes
 257
 258Q: What is this 'Encrypt suspend image' for?
 259
 260A: First of all: it is not a replacement for dm-crypt encrypted swap.
 261It cannot protect your computer while it is suspended. Instead it does
 262protect from leaking sensitive data after resume from suspend.
 263
 264Think of the following: you suspend while an application is running
 265that keeps sensitive data in memory. The application itself prevents
 266the data from being swapped out. Suspend, however, must write these
 267data to swap to be able to resume later on. Without suspend encryption
 268your sensitive data are then stored in plaintext on disk.  This means
 269that after resume your sensitive data are accessible to all
 270applications having direct access to the swap device which was used
 271for suspend. If you don't need swap after resume these data can remain
 272on disk virtually forever. Thus it can happen that your system gets
 273broken in weeks later and sensitive data which you thought were
 274encrypted and protected are retrieved and stolen from the swap device.
 275To prevent this situation you should use 'Encrypt suspend image'.
 276
 277During suspend a temporary key is created and this key is used to
 278encrypt the data written to disk. When, during resume, the data was
 279read back into memory the temporary key is destroyed which simply
 280means that all data written to disk during suspend are then
 281inaccessible so they can't be stolen later on.  The only thing that
 282you must then take care of is that you call 'mkswap' for the swap
 283partition used for suspend as early as possible during regular
 284boot. This asserts that any temporary key from an oopsed suspend or
 285from a failed or aborted resume is erased from the swap device.
 286
 287As a rule of thumb use encrypted swap to protect your data while your
 288system is shut down or suspended. Additionally use the encrypted
 289suspend image to prevent sensitive data from being stolen after
 290resume.
 291
 292Q: Can I suspend to a swap file?
 293
 294A: Generally, yes, you can.  However, it requires you to use the "resume=" and
 295"resume_offset=" kernel command line parameters, so the resume from a swap file
 296cannot be initiated from an initrd or initramfs image.  See
 297swsusp-and-swap-files.txt for details.
 298
 299Q: Is there a maximum system RAM size that is supported by swsusp?
 300
 301A: It should work okay with highmem.
 302
 303Q: Does swsusp (to disk) use only one swap partition or can it use
 304multiple swap partitions (aggregate them into one logical space)?
 305
 306A: Only one swap partition, sorry.
 307
 308Q: If my application(s) causes lots of memory & swap space to be used
 309(over half of the total system RAM), is it correct that it is likely
 310to be useless to try to suspend to disk while that app is running?
 311
 312A: No, it should work okay, as long as your app does not mlock()
 313it. Just prepare big enough swap partition.
 314
 315Q: What information is useful for debugging suspend-to-disk problems?
 316
 317A: Well, last messages on the screen are always useful. If something
 318is broken, it is usually some kernel driver, therefore trying with as
 319little as possible modules loaded helps a lot. I also prefer people to
 320suspend from console, preferably without X running. Booting with
 321init=/bin/bash, then swapon and starting suspend sequence manually
 322usually does the trick. Then it is good idea to try with latest
 323vanilla kernel.
 324
 325Q: How can distributions ship a swsusp-supporting kernel with modular
 326disk drivers (especially SATA)?
 327
 328A: Well, it can be done, load the drivers, then do echo into
 329/sys/power/disk/resume file from initrd. Be sure not to mount
 330anything, not even read-only mount, or you are going to lose your
 331data.
 332
 333Q: How do I make suspend more verbose?
 334
 335A: If you want to see any non-error kernel messages on the virtual
 336terminal the kernel switches to during suspend, you have to set the
 337kernel console loglevel to at least 4 (KERN_WARNING), for example by
 338doing
 339
 340        # save the old loglevel
 341        read LOGLEVEL DUMMY < /proc/sys/kernel/printk
 342        # set the loglevel so we see the progress bar.
 343        # if the level is higher than needed, we leave it alone.
 344        if [ $LOGLEVEL -lt 5 ]; then
 345                echo 5 > /proc/sys/kernel/printk
 346                fi
 347
 348        IMG_SZ=0
 349        read IMG_SZ < /sys/power/image_size
 350        echo -n disk > /sys/power/state
 351        RET=$?
 352        #
 353        # the logic here is:
 354        # if image_size > 0 (without kernel support, IMG_SZ will be zero),
 355        # then try again with image_size set to zero.
 356        if [ $RET -ne 0 -a $IMG_SZ -ne 0 ]; then # try again with minimal image size
 357                echo 0 > /sys/power/image_size
 358                echo -n disk > /sys/power/state
 359                RET=$?
 360        fi
 361
 362        # restore previous loglevel
 363        echo $LOGLEVEL > /proc/sys/kernel/printk
 364        exit $RET
 365
 366Q: Is this true that if I have a mounted filesystem on a USB device and
 367I suspend to disk, I can lose data unless the filesystem has been mounted
 368with "sync"?
 369
 370A: That's right ... if you disconnect that device, you may lose data.
 371In fact, even with "-o sync" you can lose data if your programs have
 372information in buffers they haven't written out to a disk you disconnect,
 373or if you disconnect before the device finished saving data you wrote.
 374
 375Software suspend normally powers down USB controllers, which is equivalent
 376to disconnecting all USB devices attached to your system.
 377
 378Your system might well support low-power modes for its USB controllers
 379while the system is asleep, maintaining the connection, using true sleep
 380modes like "suspend-to-RAM" or "standby".  (Don't write "disk" to the
 381/sys/power/state file; write "standby" or "mem".)  We've not seen any
 382hardware that can use these modes through software suspend, although in
 383theory some systems might support "platform" modes that won't break the
 384USB connections.
 385
 386Remember that it's always a bad idea to unplug a disk drive containing a
 387mounted filesystem.  That's true even when your system is asleep!  The
 388safest thing is to unmount all filesystems on removable media (such USB,
 389Firewire, CompactFlash, MMC, external SATA, or even IDE hotplug bays)
 390before suspending; then remount them after resuming.
 391
 392There is a work-around for this problem.  For more information, see
 393Documentation/usb/persist.txt.
 394
 395Q: Can I suspend-to-disk using a swap partition under LVM?
 396
 397A: No. You can suspend successfully, but you'll not be able to
 398resume. uswsusp should be able to work with LVM. See suspend.sf.net.
 399
 400Q: I upgraded the kernel from 2.6.15 to 2.6.16. Both kernels were
 401compiled with the similar configuration files. Anyway I found that
 402suspend to disk (and resume) is much slower on 2.6.16 compared to
 4032.6.15. Any idea for why that might happen or how can I speed it up?
 404
 405A: This is because the size of the suspend image is now greater than
 406for 2.6.15 (by saving more data we can get more responsive system
 407after resume).
 408
 409There's the /sys/power/image_size knob that controls the size of the
 410image.  If you set it to 0 (eg. by echo 0 > /sys/power/image_size as
 411root), the 2.6.15 behavior should be restored.  If it is still too
 412slow, take a look at suspend.sf.net -- userland suspend is faster and
 413supports LZF compression to speed it up further.
 414
lxr.linux.no kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.