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 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.] 41 42If you want to limit the suspend image size to N bytes, do 43 44echo N > /sys/power/image_size 45 46before suspend (it is limited to 500 MB by default). 47 48 49Article about goals and implementation of Software Suspend for Linux 50~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 51Author: G����bor Kuti 52Last revised: 2003-10-20 by Pavel Machek 53 54Idea and goals to achieve 55 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. 64 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. 70 71In the meantime while the system is suspended you should not add/remove any 72of the hardware, write to the filesystems, etc. 73 74Sleep states summary 75==================== 76 77There are three different interfaces you can use, /proc/acpi should 78work like this: 79 80In a really perfect world: 81echo 1 > /proc/acpi/sleep # for standby 82echo 2 > /proc/acpi/sleep # for suspend to ram 83echo 3 > /proc/acpi/sleep # for suspend to ram, but with more power conservative 84echo 4 > /proc/acpi/sleep # for suspend to disk 85echo 5 > /proc/acpi/sleep # for shutdown unfriendly the system 86 87and perhaps 88echo 4b > /proc/acpi/sleep # for suspend to disk via s4bios 89 90Frequently Asked Questions 91========================== 92 93Q: well, suspending a server is IMHO a really stupid thing, 94but... (Diego Zuccato): 95 96A: You bought new UPS for your server. How do you install it without 97bringing machine down? Suspend to disk, rearrange power cables, 98resume. 99 100You have your server on UPS. Power died, and UPS is indicating 30 101seconds to failure. What do you do? Suspend to disk. 102 103 104Q: Maybe I'm missing something, but why don't the regular I/O paths work? 105 106A: We do use the regular I/O paths. However we cannot restore the data 107to its original location as we load it. That would create an 108inconsistent kernel state which would certainly result in an oops. 109Instead, we load the image into unused memory and then atomically copy 110it back to it original location. This implies, of course, a maximum 111image size of half the amount of memory. 112 113There are two solutions to this: 114 115* require half of memory to be free during suspend. That way you can 116read "new" data onto free spots, then cli and copy 117 118* assume we had special "polling" ide driver that only uses memory 119between 0-640KB. That way, I'd have to make sure that 0-640KB is free 120during suspending, but otherwise it would work... 121 122suspend2 shares this fundamental limitation, but does not include user 123data and disk caches into "used memory" by saving them in 124advance. That means that the limitation goes away in practice. 125 126Q: Does linux support ACPI S4? 127 128A: Yes. That's what echo platform > /sys/power/disk does. 129 130Q: What is 'suspend2'? 131 132A: suspend2 is 'Software Suspend 2', a forked implementation of 133suspend-to-disk which is available as separate patches for 2.4 and 2.6 134kernels from swsusp.sourceforge.net. It includes support for SMP, 4GB 135highmem and preemption. It also has a extensible architecture that 136allows for arbitrary transformations on the image (compression, 137encryption) and arbitrary backends for writing the image (eg to swap 138or an NFS share[Work In Progress]). Questions regarding suspend2 139should be sent to the mailing list available through the suspend2 140website, and not to the Linux Kernel Mailing List. We are working 141toward merging suspend2 into the mainline kernel. 142 143Q: What is the freezing of tasks and why are we using it? 144 145A: The freezing of tasks is a mechanism by which user space processes and some 146kernel threads are controlled during hibernation or system-wide suspend (on some 147architectures). See freezing-of-tasks.txt for details. 148 149Q: What is the difference between "platform" and "shutdown"? 150 151A: 152 153shutdown: save state in linux, then tell bios to powerdown 154 155platform: save state in linux, then tell bios to powerdown and blink 156 "suspended led" 157 158"platform" is actually right thing to do where supported, but 159"shutdown" is most reliable (except on ACPI systems). 160 161Q: I do not understand why you have such strong objections to idea of 162selective suspend. 163 164A: Do selective suspend during runtime power management, that's okay. But 165it's useless for suspend-to-disk. (And I do not see how you could use 166it for suspend-to-ram, I hope you do not want that). 167 168Lets see, so you suggest to 169 170* SUSPEND all but swap device and parents 171* Snapshot 172* Write image to disk 173* SUSPEND swap device and parents 174* Powerdown 175 176Oh no, that does not work, if swap device or its parents uses DMA, 177you've corrupted data. You'd have to do 178 179* SUSPEND all but swap device and parents 180* FREEZE swap device and parents 181* Snapshot 182* UNFREEZE swap device and parents 183* Write 184* SUSPEND swap device and parents 185 186Which means that you still need that FREEZE state, and you get more 187complicated code. (And I have not yet introduce details like system 188devices). 189 190Q: There don't seem to be any generally useful behavioral 191distinctions between SUSPEND and FREEZE. 192 193A: Doing SUSPEND when you are asked to do FREEZE is always correct, 194but it may be unneccessarily slow. If you want your driver to stay simple, 195slowness may not matter to you. It can always be fixed later. 196 197For devices like disk it does matter, you do not want to spindown for 198FREEZE. 199 200Q: After resuming, system is paging heavily, leading to very bad interactivity. 201 202A: Try running 203 204cat `cat /proc/[0-9]*/maps | grep / | sed 's:.* /:/:' | sort -u` > /dev/null 205 206after resume. swapoff -a; swapon -a may also be useful. 207 208Q: What happens to devices during swsusp? They seem to be resumed 209during system suspend? 210 211A: That's correct. We need to resume them if we want to write image to 212disk. Whole sequence goes like 213 214 Suspend part 215 ~~~~~~~~~~~~ 216 running system, user asks for suspend-to-disk 217 218 user processes are stopped 219 220 suspend(PMSG_FREEZE): devices are frozen so that they don't interfere 221 with state snapshot 222 223 state snapshot: copy of whole used memory is taken with interrupts disabled 224 225 resume(): devices are woken up so that we can write image to swap 226 227 write image to swap 228 229 suspend(PMSG_SUSPEND): suspend devices so that we can power off 230 231 turn the power off 232 233 Resume part 234 ~~~~~~~~~~~ 235 (is actually pretty similar) 236 237 running system, user asks for suspend-to-disk 238 239 user processes are stopped (in common case there are none, but with resume-from-initrd, noone knows) 240 241 read image from disk 242 243 suspend(PMSG_FREEZE): devices are frozen so that they don't interfere 244 with image restoration 245 246 image restoration: rewrite memory with image 247 248 resume(): devices are woken up so that system can continue 249 250 thaw all user processes 251 252Q: What is this 'Encrypt suspend image' for? 253 254A: First of all: it is not a replacement for dm-crypt encrypted swap. 255It cannot protect your computer while it is suspended. Instead it does 256protect from leaking sensitive data after resume from suspend. 257 258Think of the following: you suspend while an application is running 259that keeps sensitive data in memory. The application itself prevents 260the data from being swapped out. Suspend, however, must write these 261data to swap to be able to resume later on. Without suspend encryption 262your sensitive data are then stored in plaintext on disk. This means 263that after resume your sensitive data are accessible to all 264applications having direct access to the swap device which was used 265for suspend. If you don't need swap after resume these data can remain 266on disk virtually forever. Thus it can happen that your system gets 267broken in weeks later and sensitive data which you thought were 268encrypted and protected are retrieved and stolen from the swap device. 269To prevent this situation you should use 'Encrypt suspend image'. 270 271During suspend a temporary key is created and this key is used to 272encrypt the data written to disk. When, during resume, the data was 273read back into memory the temporary key is destroyed which simply 274means that all data written to disk during suspend are then 275inaccessible so they can't be stolen later on. The only thing that 276you must then take care of is that you call 'mkswap' for the swap 277partition used for suspend as early as possible during regular 278boot. This asserts that any temporary key from an oopsed suspend or 279from a failed or aborted resume is erased from the swap device. 280 281As a rule of thumb use encrypted swap to protect your data while your 282system is shut down or suspended. Additionally use the encrypted 283suspend image to prevent sensitive data from being stolen after 284resume. 285 286Q: Can I suspend to a swap file? 287 288A: Generally, yes, you can. However, it requires you to use the "resume=" and 289"resume_offset=" kernel command line parameters, so the resume from a swap file 290cannot be initiated from an initrd or initramfs image. See 291swsusp-and-swap-files.txt for details. 292 293Q: Is there a maximum system RAM size that is supported by swsusp? 294 295A: It should work okay with highmem. 296 297Q: Does swsusp (to disk) use only one swap partition or can it use 298multiple swap partitions (aggregate them into one logical space)? 299 300A: Only one swap partition, sorry. 301 302Q: If my application(s) causes lots of memory & swap space to be used 303(over half of the total system RAM), is it correct that it is likely 304to be useless to try to suspend to disk while that app is running? 305 306A: No, it should work okay, as long as your app does not mlock() 307it. Just prepare big enough swap partition. 308 309Q: What information is useful for debugging suspend-to-disk problems? 310 311A: Well, last messages on the screen are always useful. If something 312is broken, it is usually some kernel driver, therefore trying with as 313little as possible modules loaded helps a lot. I also prefer people to 314suspend from console, preferably without X running. Booting with 315init=/bin/bash, then swapon and starting suspend sequence manually 316usually does the trick. Then it is good idea to try with latest 317vanilla kernel. 318 319Q: How can distributions ship a swsusp-supporting kernel with modular 320disk drivers (especially SATA)? 321 322A: Well, it can be done, load the drivers, then do echo into 323/sys/power/disk/resume file from initrd. Be sure not to mount 324anything, not even read-only mount, or you are going to lose your 325data. 326 327Q: How do I make suspend more verbose? 328 329A: If you want to see any non-error kernel messages on the virtual 330terminal the kernel switches to during suspend, you have to set the 331kernel console loglevel to at least 4 (KERN_WARNING), for example by 332doing 333 334 # save the old loglevel 335 read LOGLEVEL DUMMY < /proc/sys/kernel/printk 336 # set the loglevel so we see the progress bar. 337 # if the level is higher than needed, we leave it alone. 338 if [ $LOGLEVEL -lt 5 ]; then 339 echo 5 > /proc/sys/kernel/printk 340 fi 341 342 IMG_SZ=0 343 read IMG_SZ < /sys/power/image_size 344 echo -n disk > /sys/power/state 345 RET=$? 346 # 347 # the logic here is: 348 # if image_size > 0 (without kernel support, IMG_SZ will be zero), 349 # then try again with image_size set to zero. 350 if [ $RET -ne 0 -a $IMG_SZ -ne 0 ]; then # try again with minimal image size 351 echo 0 > /sys/power/image_size 352 echo -n disk > /sys/power/state 353 RET=$? 354 fi 355 356 # restore previous loglevel 357 echo $LOGLEVEL > /proc/sys/kernel/printk 358 exit $RET 359 360Q: Is this true that if I have a mounted filesystem on a USB device and 361I suspend to disk, I can lose data unless the filesystem has been mounted 362with "sync"? 363 364A: That's right ... if you disconnect that device, you may lose data. 365In fact, even with "-o sync" you can lose data if your programs have 366information in buffers they haven't written out to a disk you disconnect, 367or if you disconnect before the device finished saving data you wrote. 368 369Software suspend normally powers down USB controllers, which is equivalent 370to disconnecting all USB devices attached to your system. 371 372Your system might well support low-power modes for its USB controllers 373while the system is asleep, maintaining the connection, using true sleep 374modes like "suspend-to-RAM" or "standby". (Don't write "disk" to the 375/sys/power/state file; write "standby" or "mem".) We've not seen any 376hardware that can use these modes through software suspend, although in 377theory some systems might support "platform" modes that won't break the 378USB connections. 379 380Remember that it's always a bad idea to unplug a disk drive containing a 381mounted filesystem. That's true even when your system is asleep! The 382safest thing is to unmount all filesystems on removable media (such USB, 383Firewire, CompactFlash, MMC, external SATA, or even IDE hotplug bays) 384before suspending; then remount them after resuming. 385 386There is a work-around for this problem. For more information, see 387Documentation/usb/persist.txt. 388 389Q: Can I suspend-to-disk using a swap partition under LVM? 390 391A: No. You can suspend successfully, but you'll not be able to 392resume. uswsusp should be able to work with LVM. See suspend.sf.net. 393 394Q: I upgraded the kernel from 2.6.15 to 2.6.16. Both kernels were 395compiled with the similar configuration files. Anyway I found that 396suspend to disk (and resume) is much slower on 2.6.16 compared to 3972.6.15. Any idea for why that might happen or how can I speed it up? 398 399A: This is because the size of the suspend image is now greater than 400for 2.6.15 (by saving more data we can get more responsive system 401after resume). 402 403There's the /sys/power/image_size knob that controls the size of the 404image. If you set it to 0 (eg. by echo 0 > /sys/power/image_size as 405root), the 2.6.15 behavior should be restored. If it is still too 406slow, take a look at suspend.sf.net -- userland suspend is faster and 407supports LZF compression to speed it up further. 408