[lkml]   [2009]   [May]   [6]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
Patch in this message
Subject[PATCH 1/19] TuxOnIce: Documentation
This patch creates new Documentation/power files for describing
TuxOnIce, and a MAINTAINERS entry for the code.

Signed-off-by: Nigel Cunningham <>
Documentation/power/tuxonice-internals.txt | 477 ++++++++++++++
Documentation/power/tuxonice.txt | 924 ++++++++++++++++++++++++++++
3 files changed, 1408 insertions(+), 0 deletions(-)
create mode 100644 Documentation/power/tuxonice-internals.txt
create mode 100644 Documentation/power/tuxonice.txt

diff --git a/Documentation/power/tuxonice-internals.txt b/Documentation/power/tuxonice-internals.txt
new file mode 100644
index 0000000..7a96186
--- /dev/null
+++ b/Documentation/power/tuxonice-internals.txt
@@ -0,0 +1,477 @@
+ TuxOnIce 3.0 Internal Documentation.
+ Updated to 26 March 2009
+1. Introduction.
+ TuxOnIce 3.0 is an addition to the Linux Kernel, designed to
+ allow the user to quickly shutdown and quickly boot a computer, without
+ needing to close documents or programs. It is equivalent to the
+ hibernate facility in some laptops. This implementation, however,
+ requires no special BIOS or hardware support.
+ The code in these files is based upon the original implementation
+ prepared by Gabor Kuti and additional work by Pavel Machek and a
+ host of others. This code has been substantially reworked by Nigel
+ Cunningham, again with the help and testing of many others, not the
+ least of whom is Michael Frank. At its heart, however, the operation is
+ essentially the same as Gabor's version.
+2. Overview of operation.
+ The basic sequence of operations is as follows:
+ a. Quiesce all other activity.
+ b. Ensure enough memory and storage space are available, and attempt
+ to free memory/storage if necessary.
+ c. Allocate the required memory and storage space.
+ d. Write the image.
+ e. Power down.
+ There are a number of complicating factors which mean that things are
+ not as simple as the above would imply, however...
+ o The activity of each process must be stopped at a point where it will
+ not be holding locks necessary for saving the image, or unexpectedly
+ restart operations due to something like a timeout and thereby make
+ our image inconsistent.
+ o It is desirous that we sync outstanding I/O to disk before calculating
+ image statistics. This reduces corruption if one should suspend but
+ then not resume, and also makes later parts of the operation safer (see
+ below).
+ o We need to get as close as we can to an atomic copy of the data.
+ Inconsistencies in the image will result in inconsistent memory contents at
+ resume time, and thus in instability of the system and/or file system
+ corruption. This would appear to imply a maximum image size of one half of
+ the amount of RAM, but we have a solution... (again, below).
+ o In 2.6, we choose to play nicely with the other suspend-to-disk
+ implementations.
+3. Detailed description of internals.
+ a. Quiescing activity.
+ Safely quiescing the system is achieved using three separate but related
+ aspects.
+ First, we note that the vast majority of processes don't need to run during
+ suspend. They can be 'frozen'. We therefore implement a refrigerator
+ routine, which processes enter and in which they remain until the cycle is
+ complete. Processes enter the refrigerator via try_to_freeze() invocations
+ at appropriate places. A process cannot be frozen in any old place. It
+ must not be holding locks that will be needed for writing the image or
+ freezing other processes. For this reason, userspace processes generally
+ enter the refrigerator via the signal handling code, and kernel threads at
+ the place in their event loops where they drop locks and yield to other
+ processes or sleep.
+ The task of freezing processes is complicated by the fact that there can be
+ interdependencies between processes. Freezing process A before process B may
+ mean that process B cannot be frozen, because it stops at waiting for
+ process A rather than in the refrigerator. This issue is seen where
+ userspace waits on freezeable kernel threads or fuse filesystem threads. To
+ address this issue, we implement the following algorithm for quiescing
+ activity:
+ - Freeze filesystems (including fuse - userspace programs starting
+ new requests are immediately frozen; programs already running
+ requests complete their work before being frozen in the next
+ step)
+ - Freeze userspace
+ - Thaw filesystems (this is safe now that userspace is frozen and no
+ fuse requests are outstanding).
+ - Invoke sys_sync (noop on fuse).
+ - Freeze filesystems
+ - Freeze kernel threads
+ If we need to free memory, we thaw kernel threads and filesystems, but not
+ userspace. We can then free caches without worrying about deadlocks due to
+ swap files being on frozen filesystems or such like.
+ b. Ensure enough memory & storage are available.
+ We have a number of constraints to meet in order to be able to successfully
+ suspend and resume.
+ First, the image will be written in two parts, described below. One of these
+ parts needs to have an atomic copy made, which of course implies a maximum
+ size of one half of the amount of system memory. The other part ('pageset')
+ is not atomically copied, and can therefore be as large or small as desired.
+ Second, we have constraints on the amount of storage available. In these
+ calculations, we may also consider any compression that will be done. The
+ cryptoapi module allows the user to configure an expected compression ratio.
+ Third, the user can specify an arbitrary limit on the image size, in
+ megabytes. This limit is treated as a soft limit, so that we don't fail the
+ attempt to suspend if we cannot meet this constraint.
+ c. Allocate the required memory and storage space.
+ Having done the initial freeze, we determine whether the above constraints
+ are met, and seek to allocate the metadata for the image. If the constraints
+ are not met, or we fail to allocate the required space for the metadata, we
+ seek to free the amount of memory that we calculate is needed and try again.
+ We allow up to four iterations of this loop before aborting the cycle. If we
+ do fail, it should only be because of a bug in TuxOnIce's calculations.
+ These steps are merged together in the prepare_image function, found in
+ prepare_image.c. The functions are merged because of the cyclical nature
+ of the problem of calculating how much memory and storage is needed. Since
+ the data structures containing the information about the image must
+ themselves take memory and use storage, the amount of memory and storage
+ required changes as we prepare the image. Since the changes are not large,
+ only one or two iterations will be required to achieve a solution.
+ The recursive nature of the algorithm is miminised by keeping user space
+ frozen while preparing the image, and by the fact that our records of which
+ pages are to be saved and which pageset they are saved in use bitmaps (so
+ that changes in number or fragmentation of the pages to be saved don't
+ feedback via changes in the amount of memory needed for metadata). The
+ recursiveness is thus limited to any extra slab pages allocated to store the
+ extents that record storage used, and the effects of seeking to free memory.
+ d. Write the image.
+ We previously mentioned the need to create an atomic copy of the data, and
+ the half-of-memory limitation that is implied in this. This limitation is
+ circumvented by dividing the memory to be saved into two parts, called
+ pagesets.
+ Pageset2 contains most of the page cache - the pages on the active and
+ inactive LRU lists that aren't needed or modified while TuxOnIce is
+ running, so they can be safely written without an atomic copy. They are
+ therefore saved first and reloaded last. While saving these pages,
+ TuxOnIce carefully ensures that the work of writing the pages doesn't make
+ the image inconsistent. With the support for Kernel (Video) Mode Setting
+ going into the kernel at the time of writing, we need to check for pages
+ on the LRU that are used by KMS, and exclude them from pageset2. They are
+ atomically copied as part of pageset 1.
+ Once pageset2 has been saved, we prepare to do the atomic copy of remaining
+ memory. As part of the preparation, we power down drivers, thereby providing
+ them with the opportunity to have their state recorded in the image. The
+ amount of memory allocated by drivers for this is usually negligible, but if
+ DRI is in use, video drivers may require significants amounts. Ideally we
+ would be able to query drivers while preparing the image as to the amount of
+ memory they will need. Unfortunately no such mechanism exists at the time of
+ writing. For this reason, TuxOnIce allows the user to set an
+ 'extra_pages_allowance', which is used to seek to ensure sufficient memory
+ is available for drivers at this point. TuxOnIce also lets the user set this
+ value to 0. In this case, a test driver suspend is done while preparing the
+ image, and the difference (plus a margin) used instead. TuxOnIce will also
+ automatically restart the hibernation process (twice at most) if it finds
+ that the extra pages allowance is not sufficient. It will then use what was
+ actually needed (plus a margin, again). Failure to hibernate should thus
+ be an extremely rare occurence.
+ Having suspended the drivers, we save the CPU context before making an
+ atomic copy of pageset1, resuming the drivers and saving the atomic copy.
+ After saving the two pagesets, we just need to save our metadata before
+ powering down.
+ As we mentioned earlier, the contents of pageset2 pages aren't needed once
+ they've been saved. We therefore use them as the destination of our atomic
+ copy. In the unlikely event that pageset1 is larger, extra pages are
+ allocated while the image is being prepared. This is normally only a real
+ possibility when the system has just been booted and the page cache is
+ small.
+ This is where we need to be careful about syncing, however. Pageset2 will
+ probably contain filesystem meta data. If this is overwritten with pageset1
+ and then a sync occurs, the filesystem will be corrupted - at least until
+ resume time and another sync of the restored data. Since there is a
+ possibility that the user might not resume or (may it never be!) that
+ TuxOnIce might oops, we do our utmost to avoid syncing filesystems after
+ copying pageset1.
+ e. Power down.
+ Powering down uses standard kernel routines. TuxOnIce supports powering down
+ using the ACPI S3, S4 and S5 methods or the kernel's non-ACPI power-off.
+ Supporting suspend to ram (S3) as a power off option might sound strange,
+ but it allows the user to quickly get their system up and running again if
+ the battery doesn't run out (we just need to re-read the overwritten pages)
+ and if the battery does run out (or the user removes power), they can still
+ resume.
+4. Data Structures.
+ TuxOnIce uses three main structures to store its metadata and configuration
+ information:
+ a) Pageflags bitmaps.
+ TuxOnIce records which pages will be in pageset1, pageset2, the destination
+ of the atomic copy and the source of the atomically restored image using
+ bitmaps. The code used is that written for swsusp, with small improvements
+ to match TuxOnIce's requirements.
+ The pageset1 bitmap is thus easily stored in the image header for use at
+ resume time.
+ As mentioned above, using bitmaps also means that the amount of memory and
+ storage required for recording the above information is constant. This
+ greatly simplifies the work of preparing the image. In earlier versions of
+ TuxOnIce, extents were used to record which pages would be stored. In that
+ case, however, eating memory could result in greater fragmentation of the
+ lists of pages, which in turn required more memory to store the extents and
+ more storage in the image header. These could in turn require further
+ freeing of memory, and another iteration. All of this complexity is removed
+ by having bitmaps.
+ Bitmaps also make a lot of sense because TuxOnIce only ever iterates
+ through the lists. There is therefore no cost to not being able to find the
+ nth page in order 0 time. We only need to worry about the cost of finding
+ the n+1th page, given the location of the nth page. Bitwise optimisations
+ help here.
+ b) Extents for block data.
+ TuxOnIce supports writing the image to multiple block devices. In the case
+ of swap, multiple partitions and/or files may be in use, and we happily use
+ them all (with the exception of compcache pages, which we allocate but do
+ not use). This use of multiple block devices is accomplished as follows:
+ Whatever the actual source of the allocated storage, the destination of the
+ image can be viewed in terms of one or more block devices, and on each
+ device, a list of sectors. To simplify matters, we only use contiguous,
+ PAGE_SIZE aligned sectors, like the swap code does.
+ Since sector numbers on each bdev may well not start at 0, it makes much
+ more sense to use extents here. Contiguous ranges of pages can thus be
+ represented in the extents by contiguous values.
+ Variations in block size are taken account of in transforming this data
+ into the parameters for bio submission.
+ We can thus implement a layer of abstraction wherein the core of TuxOnIce
+ doesn't have to worry about which device we're currently writing to or
+ where in the device we are. It simply requests that the next page in the
+ pageset or header be written, leaving the details to this lower layer.
+ The lower layer remembers where in the sequence of devices and blocks each
+ pageset starts. The header always starts at the beginning of the allocated
+ storage.
+ So extents are:
+ struct extent {
+ unsigned long minimum, maximum;
+ struct extent *next;
+ }
+ These are combined into chains of extents for a device:
+ struct extent_chain {
+ int size; /* size of the extent ie sum (max-min+1) */
+ int allocs, frees;
+ char *name;
+ struct extent *first, *last_touched;
+ };
+ For each bdev, we need to store a little more info:
+ struct suspend_bdev_info {
+ struct block_device *bdev;
+ dev_t dev_t;
+ int bmap_shift;
+ int blocks_per_page;
+ };
+ The dev_t is used to identify the device in the stored image. As a result,
+ we expect devices at resume time to have the same major and minor numbers
+ as they had while suspending. This is primarily a concern where the user
+ utilises LVM for storage, as they will need to dmsetup their partitions in
+ such a way as to maintain this consistency at resume time.
+ bmap_shift and blocks_per_page apply the effects of variations in blocks
+ per page settings for the filesystem and underlying bdev. For most
+ filesystems, these are the same, but for xfs, they can have independant
+ values.
+ Combining these two structures together, we have everything we need to
+ record what devices and what blocks on each device are being used to
+ store the image, and to submit i/o using bio_submit.
+ The last elements in the picture are a means of recording how the storage
+ is being used.
+ We do this first and foremost by implementing a layer of abstraction on
+ top of the devices and extent chains which allows us to view however many
+ devices there might be as one long storage tape, with a single 'head' that
+ tracks a 'current position' on the tape:
+ struct extent_iterate_state {
+ struct extent_chain *chains;
+ int num_chains;
+ int current_chain;
+ struct extent *current_extent;
+ unsigned long current_offset;
+ };
+ That is, *chains points to an array of size num_chains of extent chains.
+ For the filewriter, this is always a single chain. For the swapwriter, the
+ array is of size MAX_SWAPFILES.
+ current_chain, current_extent and current_offset thus point to the current
+ index in the chains array (and into a matching array of struct
+ suspend_bdev_info), the current extent in that chain (to optimise access),
+ and the current value in the offset.
+ The image is divided into three parts:
+ - The header
+ - Pageset 1
+ - Pageset 2
+ The header always starts at the first device and first block. We know its
+ size before we begin to save the image because we carefully account for
+ everything that will be stored in it.
+ The second pageset (LRU) is stored first. It begins on the next page after
+ the end of the header.
+ The first pageset is stored second. It's start location is only known once
+ pageset2 has been saved, since pageset2 may be compressed as it is written.
+ This location is thus recorded at the end of saving pageset2. It is page
+ aligned also.
+ Since this information is needed at resume time, and the location of extents
+ in memory will differ at resume time, this needs to be stored in a portable
+ way:
+ struct extent_iterate_saved_state {
+ int chain_num;
+ int extent_num;
+ unsigned long offset;
+ };
+ We can thus implement a layer of abstraction wherein the core of TuxOnIce
+ doesn't have to worry about which device we're currently writing to or
+ where in the device we are. It simply requests that the next page in the
+ pageset or header be written, leaving the details to this layer, and
+ invokes the routines to remember and restore the position, without having
+ to worry about the details of how the data is arranged on disk or such like.
+ c) Modules
+ One aim in designing TuxOnIce was to make it flexible. We wanted to allow
+ for the implementation of different methods of transforming a page to be
+ written to disk and different methods of getting the pages stored.
+ In early versions (the betas and perhaps Suspend1), compression support was
+ inlined in the image writing code, and the data structures and code for
+ managing swap were intertwined with the rest of the code. A number of people
+ had expressed interest in implementing image encryption, and alternative
+ methods of storing the image.
+ In order to achieve this, TuxOnIce was given a modular design.
+ A module is a single file which encapsulates the functionality needed
+ to transform a pageset of data (encryption or compression, for example),
+ or to write the pageset to a device. The former type of module is called
+ a 'page-transformer', the later a 'writer'.
+ Modules are linked together in pipeline fashion. There may be zero or more
+ page transformers in a pipeline, and there is always exactly one writer.
+ The pipeline follows this pattern:
+ ---------------------------------
+ | TuxOnIce Core |
+ ---------------------------------
+ |
+ |
+ ---------------------------------
+ | Page transformer 1 |
+ ---------------------------------
+ |
+ |
+ ---------------------------------
+ | Page transformer 2 |
+ ---------------------------------
+ |
+ |
+ ---------------------------------
+ | Writer |
+ ---------------------------------
+ During the writing of an image, the core code feeds pages one at a time
+ to the first module. This module performs whatever transformations it
+ implements on the incoming data, completely consuming the incoming data and
+ feeding output in a similar manner to the next module.
+ All routines are SMP safe, and the final result of the transformations is
+ written with an index (provided by the core) and size of the output by the
+ writer. As a result, we can have multithreaded I/O without needing to
+ worry about the sequence in which pages are written (or read).
+ During reading, the pipeline works in the reverse direction. The core code
+ calls the first module with the address of a buffer which should be filled.
+ (Note that the buffer size is always PAGE_SIZE at this time). This module
+ will in turn request data from the next module and so on down until the
+ writer is made to read from the stored image.
+ Part of definition of the structure of a module thus looks like this:
+ int (*rw_init) (int rw, int stream_number);
+ int (*rw_cleanup) (int rw);
+ int (*write_chunk) (struct page *buffer_page);
+ int (*read_chunk) (struct page *buffer_page, int sync);
+ It should be noted that the _cleanup routine may be called before the
+ full stream of data has been read or written. While writing the image,
+ the user may (depending upon settings) choose to abort suspending, and
+ if we are in the midst of writing the last portion of the image, a portion
+ of the second pageset may be reread. This may also happen if an error
+ occurs and we seek to abort the process of writing the image.
+ The modular design is also useful in a number of other ways. It provides
+ a means where by we can add support for:
+ - providing overall initialisation and cleanup routines;
+ - serialising configuration information in the image header;
+ - providing debugging information to the user;
+ - determining memory and image storage requirements;
+ - dis/enabling components at run-time;
+ - configuring the module (see below);
+ ...and routines for writers specific to their work:
+ - Parsing a resume= location;
+ - Determining whether an image exists;
+ - Marking a resume as having been attempted;
+ - Invalidating an image;
+ Since some parts of the core - the user interface and storage manager
+ support - have use for some of these functions, they are registered as
+ 'miscellaneous' modules as well.
+ d) Sysfs data structures.
+ This brings us naturally to support for configuring TuxOnIce. We desired to
+ provide a way to make TuxOnIce as flexible and configurable as possible.
+ The user shouldn't have to reboot just because they want to now hibernate to
+ a file instead of a partition, for example.
+ To accomplish this, TuxOnIce implements a very generic means whereby the
+ core and modules can register new sysfs entries. All TuxOnIce entries use
+ a single _store and _show routine, both of which are found in
+ tuxonice_sysfs.c in the kernel/power directory. These routines handle the
+ most common operations - getting and setting the values of bits, integers,
+ longs, unsigned longs and strings in one place, and allow overrides for
+ customised get and set options as well as side-effect routines for all
+ reads and writes.
+ When combined with some simple macros, a new sysfs entry can then be defined
+ in just a couple of lines:
+ SYSFS_INT("progress_granularity", SYSFS_RW, &progress_granularity, 1,
+ 2048, 0, NULL),
+ This defines a sysfs entry named "progress_granularity" which is rw and
+ allows the user to access an integer stored at &progress_granularity, giving
+ it a value between 1 and 2048 inclusive.
+ Sysfs entries are registered under /sys/power/tuxonice, and entries for
+ modules are located in a subdirectory named after the module.
diff --git a/Documentation/power/tuxonice.txt b/Documentation/power/tuxonice.txt
new file mode 100644
index 0000000..bfc0df5
--- /dev/null
+++ b/Documentation/power/tuxonice.txt
@@ -0,0 +1,924 @@
+ --- TuxOnIce, version 3.0 ---
+1. What is it?
+2. Why would you want it?
+3. What do you need to use it?
+4. Why not just use the version already in the kernel?
+5. How do you use it?
+6. What do all those entries in /sys/power/tuxonice do?
+7. How do you get support?
+8. I think I've found a bug. What should I do?
+9. When will XXX be supported?
+10 How does it work?
+11. Who wrote TuxOnIce?
+1. What is it?
+ Imagine you're sitting at your computer, working away. For some reason, you
+ need to turn off your computer for a while - perhaps it's time to go home
+ for the day. When you come back to your computer next, you're going to want
+ to carry on where you left off. Now imagine that you could push a button and
+ have your computer store the contents of its memory to disk and power down.
+ Then, when you next start up your computer, it loads that image back into
+ memory and you can carry on from where you were, just as if you'd never
+ turned the computer off. You have far less time to start up, no reopening of
+ applications or finding what directory you put that file in yesterday.
+ That's what TuxOnIce does.
+ TuxOnIce has a long heritage. It began life as work by Gabor Kuti, who,
+ with some help from Pavel Machek, got an early version going in 1999. The
+ project was then taken over by Florent Chabaud while still in alpha version
+ numbers. Nigel Cunningham came on the scene when Florent was unable to
+ continue, moving the project into betas, then 1.0, 2.0 and so on up to
+ the present series. During the 2.0 series, the name was contracted to
+ Suspend2 and the website created. Beginning around July 2007,
+ a transition to calling the software TuxOnIce was made, to seek to help
+ make it clear that TuxOnIce is more concerned with hibernation than suspend
+ to ram.
+ Pavel Machek's swsusp code, which was merged around 2.5.17 retains the
+ original name, and was essentially a fork of the beta code until Rafael
+ Wysocki came on the scene in 2005 and began to improve it further.
+2. Why would you want it?
+ Why wouldn't you want it?
+ Being able to save the state of your system and quickly restore it improves
+ your productivity - you get a useful system in far less time than through
+ the normal boot process. You also get to be completely 'green', using zero
+ power, or as close to that as possible (the computer may still provide
+ minimal power to some devices, so they can initiate a power on, but that
+ will be the same amount of power as would be used if you told the computer
+ to shutdown.
+3. What do you need to use it?
+ a. Kernel Support.
+ i) The TuxOnIce patch.
+ TuxOnIce is part of the Linux Kernel. This version is not part of Linus's
+ 2.6 tree at the moment, so you will need to download the kernel source and
+ apply the latest patch. Having done that, enable the appropriate options in
+ make [menu|x]config (under Power Management Options - look for "Enhanced
+ Hibernation"), compile and install your kernel. TuxOnIce works with SMP,
+ Highmem, preemption, fuse filesystems, x86-32, PPC and x86_64.
+ TuxOnIce patches are available from
+ ii) Compression support.
+ Compression support is implemented via the cryptoapi. You will therefore want
+ to select any Cryptoapi transforms that you want to use on your image from
+ the Cryptoapi menu while configuring your kernel. We recommend the use of the
+ LZO compression method - it is very fast and still achieves good compression.
+ You can also tell TuxOnIce to write its image to an encrypted and/or
+ compressed filesystem/swap partition. In that case, you don't need to do
+ anything special for TuxOnIce when it comes to kernel configuration.
+ iii) Configuring other options.
+ While you're configuring your kernel, try to configure as much as possible
+ to build as modules. We recommend this because there are a number of drivers
+ that are still in the process of implementing proper power management
+ support. In those cases, the best way to work around their current lack is
+ to build them as modules and remove the modules while hibernating. You might
+ also bug the driver authors to get their support up to speed, or even help!
+ b. Storage.
+ i) Swap.
+ TuxOnIce can store the hibernation image in your swap partition, a swap file or
+ a combination thereof. Whichever combination you choose, you will probably
+ want to create enough swap space to store the largest image you could have,
+ plus the space you'd normally use for swap. A good rule of thumb would be
+ to calculate the amount of swap you'd want without using TuxOnIce, and then
+ add the amount of memory you have. This swapspace can be arranged in any way
+ you'd like. It can be in one partition or file, or spread over a number. The
+ only requirement is that they be active when you start a hibernation cycle.
+ There is one exception to this requirement. TuxOnIce has the ability to turn
+ on one swap file or partition at the start of hibernating and turn it back off
+ at the end. If you want to ensure you have enough memory to store a image
+ when your memory is fully used, you might want to make one swap partition or
+ file for 'normal' use, and another for TuxOnIce to activate & deactivate
+ automatically. (Further details below).
+ ii) Normal files.
+ TuxOnIce includes a 'file allocator'. The file allocator can store your
+ image in a simple file. Since Linux has the concept of everything being a
+ file, this is more powerful than it initially sounds. If, for example, you
+ were to set up a network block device file, you could hibernate to a network
+ server. This has been tested and works to a point, but nbd itself isn't
+ stateless enough for our purposes.
+ Take extra care when setting up the file allocator. If you just type
+ commands without thinking and then try to hibernate, you could cause
+ irreversible corruption on your filesystems! Make sure you have backups.
+ Most people will only want to hibernate to a local file. To achieve that, do
+ something along the lines of:
+ echo "TuxOnIce" > /hibernation-file
+ dd if=/dev/zero bs=1M count=512 >> hibernation-file
+ This will create a 512MB file called /hibernation-file. To get TuxOnIce to use
+ it:
+ echo /hibernation-file > /sys/power/tuxonice/file/target
+ Then
+ cat /sys/power/tuxonice/resume
+ Put the results of this into your bootloader's configuration (see also step
+ C, below):
+ # cat /sys/power/tuxonice/resume
+ file:/dev/hda2:0x1e001
+ In this example, we would edit the append= line of our lilo.conf|menu.lst
+ so that it included:
+ resume=file:/dev/hda2:0x1e001
+ For those who are thinking 'Could I make the file sparse?', the answer is
+ 'No!'. At the moment, there is no way for TuxOnIce to fill in the holes in
+ a sparse file while hibernating. In the longer term (post merge!), I'd like
+ to change things so that the file could be dynamically resized and have
+ holes filled as needed. Right now, however, that's not possible and not a
+ priority.
+ c. Bootloader configuration.
+ Using TuxOnIce also requires that you add an extra parameter to
+ your lilo.conf or equivalent. Here's an example for a swap partition:
+ append="resume=swap:/dev/hda1"
+ This would tell TuxOnIce that /dev/hda1 is a swap partition you
+ have. TuxOnIce will use the swap signature of this partition as a
+ pointer to your data when you hibernate. This means that (in this example)
+ /dev/hda1 doesn't need to be _the_ swap partition where all of your data
+ is actually stored. It just needs to be a swap partition that has a
+ valid signature.
+ You don't need to have a swap partition for this purpose. TuxOnIce
+ can also use a swap file, but usage is a little more complex. Having made
+ your swap file, turn it on and do
+ cat /sys/power/tuxonice/swap/headerlocations
+ (this assumes you've already compiled your kernel with TuxOnIce
+ support and booted it). The results of the cat command will tell you
+ what you need to put in lilo.conf:
+ For swap partitions like /dev/hda1, simply use resume=/dev/hda1.
+ For swapfile `swapfile`, use resume=swap:/dev/hda2:0x242d.
+ If the swapfile changes for any reason (it is moved to a different
+ location, it is deleted and recreated, or the filesystem is
+ defragmented) then you will have to check
+ /sys/power/tuxonice/swap/headerlocations for a new resume_block value.
+ Once you've compiled and installed the kernel and adjusted your bootloader
+ configuration, you should only need to reboot for the most basic part
+ of TuxOnIce to be ready.
+ If you only compile in the swap allocator, or only compile in the file
+ allocator, you don't need to add the "swap:" part of the resume=
+ parameters above. resume=/dev/hda2:0x242d will work just as well. If you
+ have compiled both and your storage is on swap, you can also use this
+ format (the swap allocator is the default allocator).
+ When compiling your kernel, one of the options in the 'Power Management
+ Support' menu, just above the 'Enhanced Hibernation (TuxOnIce)' entry is
+ called 'Default resume partition'. This can be used to set a default value
+ for the resume= parameter.
+ d. The hibernate script.
+ Since the driver model in 2.6 kernels is still being developed, you may need
+ to do more than just configure TuxOnIce. Users of TuxOnIce usually start the
+ process via a script which prepares for the hibernation cycle, tells the
+ kernel to do its stuff and then restore things afterwards. This script might
+ involve:
+ - Switching to a text console and back if X doesn't like the video card
+ status on resume.
+ - Un/reloading drivers that don't play well with hibernation.
+ Note that you might not be able to unload some drivers if there are
+ processes using them. You might have to kill off processes that hold
+ devices open. Hint: if your X server accesses an USB mouse, doing a
+ 'chvt' to a text console releases the device and you can unload the
+ module.
+ Check out the latest script (available on
+ e. The userspace user interface.
+ TuxOnIce has very limited support for displaying status if you only apply
+ the kernel patch - it can printk messages, but that is all. In addition,
+ some of the functions mentioned in this document (such as cancelling a cycle
+ or performing interactive debugging) are unavailable. To utilise these
+ functions, or simply get a nice display, you need the 'userui' component.
+ Userui comes in three flavours, usplash, fbsplash and text. Text should
+ work on any console. Usplash and fbsplash require the appropriate
+ (distro specific?) support.
+ To utilise a userui, TuxOnIce just needs to be told where to find the
+ userspace binary:
+ echo "/usr/local/sbin/tuxoniceui_fbsplash" > /sys/power/tuxonice/user_interface/program
+ The hibernate script can do this for you, and a default value for this
+ setting can be configured when compiling the kernel. This path is also
+ stored in the image header, so if you have an initrd or initramfs, you can
+ use the userui during the first part of resuming (prior to the atomic
+ restore) by putting the binary in the same path in your initrd/ramfs.
+ Alternatively, you can put it in a different location and do an echo
+ similar to the above prior to the echo > do_resume. The value saved in the
+ image header will then be ignored.
+4. Why not just use the version already in the kernel?
+ The version in the vanilla kernel has a number of drawbacks. The most
+ serious of these are:
+ - it has a maximum image size of 1/2 total memory;
+ - it doesn't allocate storage until after it has snapshotted memory.
+ This means that you can't be sure hibernating will work until you
+ see it start to write the image;
+ - it does not allow you to press escape to cancel a cycle;
+ - it does not allow you to press escape to cancel resuming;
+ - it does not allow you to automatically swapon a file when
+ starting a cycle;
+ - it does not allow you to use multiple swap partitions or files;
+ - it does not allow you to use ordinary files;
+ - it just invalidates an image and continues to boot if you
+ accidentally boot the wrong kernel after hibernating;
+ - it doesn't support any sort of nice display while hibernating;
+ - it is moving toward requiring that you have an initrd/initramfs
+ to ever have a hope of resuming (uswsusp). While uswsusp will
+ address some of the concerns above, it won't address all of them,
+ and will be more complicated to get set up;
+ - it doesn't have support for suspend-to-both (write a hibernation
+ image, then suspend to ram; I think this is known as ReadySafe
+ under M$).
+5. How do you use it?
+ A hibernation cycle can be started directly by doing:
+ echo > /sys/power/tuxonice/do_hibernate
+ In practice, though, you'll probably want to use the hibernate script
+ to unload modules, configure the kernel the way you like it and so on.
+ In that case, you'd do (as root):
+ hibernate
+ See the hibernate script's man page for more details on the options it
+ takes.
+ If you're using the text or splash user interface modules, one feature of
+ TuxOnIce that you might find useful is that you can press Escape at any time
+ during hibernating, and the process will be aborted.
+ Due to the way hibernation works, this means you'll have your system back and
+ perfectly usable almost instantly. The only exception is when it's at the
+ very end of writing the image. Then it will need to reload a small (usually
+ 4-50MBs, depending upon the image characteristics) portion first.
+ Likewise, when resuming, you can press escape and resuming will be aborted.
+ The computer will then powerdown again according to settings at that time for
+ the powerdown method or rebooting.
+ You can change the settings for powering down while the image is being
+ written by pressing 'R' to toggle rebooting and 'O' to toggle between
+ suspending to ram and powering down completely).
+ If you run into problems with resuming, adding the "noresume" option to
+ the kernel command line will let you skip the resume step and recover your
+ system. This option shouldn't normally be needed, because TuxOnIce modifies
+ the image header prior to the atomic restore, and will thus prompt you
+ if it detects that you've tried to resume an image before (this flag is
+ removed if you press Escape to cancel a resume, so you won't be prompted
+ then).
+ Recent kernels (2.6.24 onwards) add support for resuming from a different
+ kernel to the one that was hibernated (thanks to Rafael for his work on
+ this - I've just embraced and enhanced the support for TuxOnIce). This
+ should further reduce the need for you to use the noresume option.
+6. What do all those entries in /sys/power/tuxonice do?
+ /sys/power/tuxonice is the directory which contains files you can use to
+ tune and configure TuxOnIce to your liking. The exact contents of
+ the directory will depend upon the version of TuxOnIce you're
+ running and the options you selected at compile time. In the following
+ descriptions, names in brackets refer to compile time options.
+ (Note that they're all dependant upon you having selected CONFIG_TUXONICE
+ in the first place!).
+ Since the values of these settings can open potential security risks, the
+ writeable ones are accessible only to the root user. You may want to
+ configure sudo to allow you to invoke your hibernate script as an ordinary
+ user.
+ - alt_resume_param
+ Instead of powering down after writing a hibernation image, TuxOnIce
+ supports resuming from a different image. This entry lets you set the
+ location of the signature for that image (the resume= value you'd use
+ for it). Using an alternate image and keep_image mode, you can do things
+ like using an alternate image to power down an uninterruptible power
+ supply.
+ - block_io/target_outstanding_io
+ This value controls the amount of memory that the block I/O code says it
+ needs when the core code is calculating how much memory is needed for
+ hibernating and for resuming. It doesn't directly control the amount of
+ I/O that is submitted at any one time - that depends on the amount of
+ available memory (we may have more available than we asked for), the
+ throughput that is being achieved and the ability of the CPU to keep up
+ with disk throughput (particularly where we're compressing pages).
+ - compression/algorithm
+ Set the cryptoapi algorithm used for compressing the image.
+ - compression/expected_compression
+ These values allow you to set an expected compression ratio, which TuxOnice
+ will use in calculating whether it meets constraints on the image size. If
+ this expected compression ratio is not attained, the hibernation cycle will
+ abort, so it is wise to allow some spare. You can see what compression
+ ratio is achieved in the logs after hibernating.
+ - debug_info:
+ This file returns information about your configuration that may be helpful
+ in diagnosing problems with hibernating.
+ - did_suspend_to_both:
+ This file can be used when you hibernate with powerdown method 3 (ie suspend
+ to ram after writing the image). There can be two outcomes in this case. We
+ can resume from the suspend-to-ram before the battery runs out, or we can run
+ out of juice and and up resuming like normal. This entry lets you find out,
+ post resume, which way we went. If the value is 1, we resumed from suspend
+ to ram. This can be useful when actions need to be run post suspend-to-ram
+ that don't need to be run if we did the normal resume from power off.
+ - do_hibernate:
+ When anything is written to this file, the kernel side of TuxOnIce will
+ begin to attempt to write an image to disk and power down. You'll normally
+ want to run the hibernate script instead, to get modules unloaded first.
+ - do_resume:
+ When anything is written to this file TuxOnIce will attempt to read and
+ restore an image. If there is no image, it will return almost immediately.
+ If an image exists, the echo > will never return. Instead, the original
+ kernel context will be restored and the original echo > do_hibernate will
+ return.
+ - */enabled
+ These option can be used to temporarily disable various parts of TuxOnIce.
+ - extra_pages_allowance
+ When TuxOnIce does its atomic copy, it calls the driver model suspend
+ and resume methods. If you have DRI enabled with a driver such as fglrx,
+ this can result in the driver allocating a substantial amount of memory
+ for storing its state. Extra_pages_allowance tells TuxOnIce how much
+ extra memory it should ensure is available for those allocations. If
+ your attempts at hibernating end with a message in dmesg indicating that
+ insufficient extra pages were allowed, you need to increase this value.
+ - file/target:
+ Read this value to get the current setting. Write to it to point TuxOnice
+ at a new storage location for the file allocator. See section 3.b.ii above
+ for details of how to set up the file allocator.
+ - freezer_test
+ This entry can be used to get TuxOnIce to just test the freezer and prepare
+ an image without actually doing a hibernation cycle. It is useful for
+ diagnosing freezing and image preparation issues.
+ - full_pageset2
+ TuxOnIce divides the pages that are stored in an image into two sets. The
+ difference between the two sets is that pages in pageset 1 are atomically
+ copied, and pages in pageset 2 are written to disk without being copied
+ first. A page CAN be written to disk without being copied first if and only
+ if its contents will not be modified or used at any time after userspace
+ processes are frozen. A page MUST be in pageset 1 if its contents are
+ modified or used at any time after userspace processes have been frozen.
+ Normally (ie if this option is enabled), TuxOnIce will put all pages on the
+ per-zone LRUs in pageset2, then remove those pages used by any userspace
+ user interface helper and TuxOnIce storage manager that are running,
+ together with pages used by the GEM memory manager introduced around 2.6.28
+ kernels.
+ If this option is disabled, a much more conservative approach will be taken.
+ The only pages in pageset2 will be those belonging to userspace processes,
+ with the exclusion of those belonging to the TuxOnIce userspace helpers
+ mentioned above. This will result in a much smaller pageset2, and will
+ therefore result in smaller images than are possible with this option
+ enabled.
+ - ignore_rootfs
+ TuxOnIce records which device is mounted as the root filesystem when
+ writing the hibernation image. It will normally check at resume time that
+ this device isn't already mounted - that would be a cause of filesystem
+ corruption. In some particular cases (RAM based root filesystems), you
+ might want to disable this check. This option allows you to do that.
+ - image_exists:
+ Can be used in a script to determine whether a valid image exists at the
+ location currently pointed to by resume=. Returns up to three lines.
+ The first is whether an image exists (-1 for unsure, otherwise 0 or 1).
+ If an image eixsts, additional lines will return the machine and version.
+ Echoing anything to this entry removes any current image.
+ - image_size_limit:
+ The maximum size of hibernation image written to disk, measured in megabytes
+ (1024*1024).
+ - last_result:
+ The result of the last hibernation cycle, as defined in
+ include/linux/suspend-debug.h with the values SUSPEND_ABORTED to
+ SUSPEND_KEPT_IMAGE. This is a bitmask.
+ - late_cpu_hotplug:
+ This sysfs entry controls whether cpu hotplugging is done - as normal - just
+ before (unplug) and after (replug) the atomic copy/restore (so that all
+ CPUs/cores are available for multithreaded I/O). The alternative is to
+ unplug all secondary CPUs/cores at the start of hibernating/resuming, and
+ replug them at the end of resuming. No multithreaded I/O will be possible in
+ this configuration, but the odd machine has been reported to require it.
+ - lid_file:
+ This determines which ACPI button file we look in to determine whether the
+ lid is open or closed after resuming from suspend to disk or power off.
+ If the entry is set to "lid/LID", we'll open /proc/acpi/button/lid/LID/state
+ and check its contents at the appropriate moment. See post_wake_state below
+ for more details on how this entry is used.
+ - log_everything (CONFIG_PM_DEBUG):
+ Setting this option results in all messages printed being logged. Normally,
+ only a subset are logged, so as to not slow the process and not clutter the
+ logs. Useful for debugging. It can be toggled during a cycle by pressing
+ 'L'.
+ - no_load_direct:
+ This is a debugging option. If, when loading the atomically copied pages of
+ an image, TuxOnIce finds that the destination address for a page is free,
+ it will normally allocate the image, load the data directly into that
+ address and skip it in the atomic restore. If this option is disabled, the
+ page will be loaded somewhere else and atomically restored like other pages.
+ - no_flusher_thread:
+ When doing multithreaded I/O (see below), the first online CPU can be used
+ to _just_ submit compressed pages when writing the image, rather than
+ compressing and submitting data. This option is normally disabled, but has
+ been included because Nigel would like to see whether it will be more useful
+ as the number of cores/cpus in computers increases.
+ - no_multithreaded_io:
+ TuxOnIce will normally create one thread per cpu/core on your computer,
+ each of which will then perform I/O. This will generally result in
+ throughput that's the maximum the storage medium can handle. There
+ shouldn't be any reason to disable multithreaded I/O now, but this option
+ has been retained for debugging purposes.
+ - no_pageset2
+ See the entry for full_pageset2 above for an explanation of pagesets.
+ Enabling this option causes TuxOnIce to do an atomic copy of all pages,
+ thereby limiting the maximum image size to 1/2 of memory, as swsusp does.
+ - no_pageset2_if_unneeded
+ See the entry for full_pageset2 above for an explanation of pagesets.
+ Enabling this option causes TuxOnIce to act like no_pageset2 was enabled
+ if and only it isn't needed anyway. This option may still make TuxOnIce
+ less reliable because pageset2 pages are normally used to store the
+ atomic copy - drivers that want to do allocations of larger amounts of
+ memory in one shot will be more likely to find that those amounts aren't
+ available if this option is enabled.
+ - pause_between_steps (CONFIG_PM_DEBUG):
+ This option is used during debugging, to make TuxOnIce pause between
+ each step of the process. It is ignored when the nice display is on.
+ - post_wake_state:
+ TuxOnIce provides support for automatically waking after a user-selected
+ delay, and using a different powerdown method if the lid is still closed.
+ (Yes, we're assuming a laptop). This entry lets you choose what state
+ should be entered next. The values are those described under
+ powerdown_method, below. It can be used to suspend to RAM after hibernating,
+ then powerdown properly (say) 20 minutes. It can also be used to power down
+ properly, then wake at (say) 6.30am and suspend to RAM until you're ready
+ to use the machine.
+ - powerdown_method:
+ Used to select a method by which TuxOnIce should powerdown after writing the
+ image. Currently:
+ 0: Don't use ACPI to power off.
+ 3: Attempt to enter Suspend-to-ram.
+ 4: Attempt to enter ACPI S4 mode.
+ 5: Attempt to power down via ACPI S5 mode.
+ Note that these options are highly dependant upon your hardware & software:
+ 3: When succesful, your machine suspends to ram instead of powering off.
+ The advantage of using this mode is that it doesn't matter whether your
+ battery has enough charge to make it through to your next resume. If it
+ lasts, you will simply resume from suspend to ram (and the image on disk
+ will be discarded). If the battery runs out, you will resume from disk
+ instead. The disadvantage is that it takes longer than a normal
+ suspend-to-ram to enter the state, since the suspend-to-disk image needs
+ to be written first.
+ 4/5: When successful, your machine will be off and comsume (almost) no power.
+ But it might still react to some external events like opening the lid or
+ trafic on a network or usb device. For the bios, resume is then the same
+ as warm boot, similar to a situation where you used the command `reboot'
+ to reboot your machine. If your machine has problems on warm boot or if
+ you want to protect your machine with the bios password, this is probably
+ not the right choice. Mode 4 may be necessary on some machines where ACPI
+ wake up methods need to be run to properly reinitialise hardware after a
+ hibernation cycle.
+ 0: Switch the machine completely off. The only possible wakeup is the power
+ button. For the bios, resume is then the same as a cold boot, in
+ particular you would have to provide your bios boot password if your
+ machine uses that feature for booting.
+ - progressbar_granularity_limit:
+ This option can be used to limit the granularity of the progress bar
+ displayed with a bootsplash screen. The value is the maximum number of
+ steps. That is, 10 will make the progress bar jump in 10% increments.
+ - reboot:
+ This option causes TuxOnIce to reboot rather than powering down
+ at the end of saving an image. It can be toggled during a cycle by pressing
+ 'R'.
+ - resume:
+ This sysfs entry can be used to read and set the location in which TuxOnIce
+ will look for the signature of an image - the value set using resume= at
+ boot time or CONFIG_PM_STD_PARTITION ("Default resume partition"). By
+ writing to this file as well as modifying your bootloader's configuration
+ file (eg menu.lst), you can set or reset the location of your image or the
+ method of storing the image without rebooting.
+ - replace_swsusp (CONFIG_TOI_REPLACE_SWSUSP):
+ This option makes
+ echo disk > /sys/power/state
+ activate TuxOnIce instead of swsusp. Regardless of whether this option is
+ enabled, any invocation of swsusp's resume time trigger will cause TuxOnIce
+ to check for an image too. This is due to the fact that at resume time, we
+ can't know whether this option was enabled until we see if an image is there
+ for us to resume from. (And when an image exists, we don't care whether we
+ did replace swsusp anyway - we just want to resume).
+ - resume_commandline:
+ This entry can be read after resuming to see the commandline that was used
+ when resuming began. You might use this to set up two bootloader entries
+ that are the same apart from the fact that one includes a extra append=
+ argument "at_work=1". You could then grep resume_commandline in your
+ post-resume scripts and configure networking (for example) differently
+ depending upon whether you're at home or work. resume_commandline can be
+ set to arbitrary text if you wish to remove sensitive contents.
+ - swap/swapfilename:
+ This entry is used to specify the swapfile or partition that
+ TuxOnIce will attempt to swapon/swapoff automatically. Thus, if
+ I normally use /dev/hda1 for swap, and want to use /dev/hda2 for specifically
+ for my hibernation image, I would
+ echo /dev/hda2 > /sys/power/tuxonice/swap/swapfile
+ /dev/hda2 would then be automatically swapon'd and swapoff'd. Note that the
+ swapon and swapoff occur while other processes are frozen (including kswapd)
+ so this swap file will not be used up when attempting to free memory. The
+ parition/file is also given the highest priority, so other swapfiles/partitions
+ will only be used to save the image when this one is filled.
+ The value of this file is used by headerlocations along with any currently
+ activated swapfiles/partitions.
+ - swap/headerlocations:
+ This option tells you the resume= options to use for swap devices you
+ currently have activated. It is particularly useful when you only want to
+ use a swap file to store your image. See above for further details.
+ - test_bio
+ This is a debugging option. When enabled, TuxOnIce will not hibernate.
+ Instead, when asked to write an image, it will skip the atomic copy,
+ just doing the writing of the image and then returning control to the
+ user at the point where it would have powered off. This is useful for
+ testing throughput in different configurations.
+ - test_filter_speed
+ This is a debugging option. When enabled, TuxOnIce will not hibernate.
+ Instead, when asked to write an image, it will not write anything or do
+ an atomic copy, but will only run any enabled compression algorithm on the
+ data that would have been written (the source pages of the atomic copy in
+ the case of pageset 1). This is useful for comparing the performance of
+ compression algorithms and for determining the extent to which an upgrade
+ to your storage method would improve hibernation speed.
+ - user_interface/debug_sections (CONFIG_PM_DEBUG):
+ This value, together with the console log level, controls what debugging
+ information is displayed. The console log level determines the level of
+ detail, and this value determines what detail is displayed. This value is
+ a bit vector, and the meaning of the bits can be found in the kernel tree
+ in include/linux/tuxonice.h. It can be overridden using the kernel's
+ command line option suspend_dbg.
+ - user_interface/default_console_level (CONFIG_PM_DEBUG):
+ This determines the value of the console log level at the start of a
+ hibernation cycle. If debugging is compiled in, the console log level can be
+ changed during a cycle by pressing the digit keys. Meanings are:
+ 0: Nice display.
+ 1: Nice display plus numerical progress.
+ 2: Errors only.
+ 3: Low level debugging info.
+ 4: Medium level debugging info.
+ 5: High level debugging info.
+ 6: Verbose debugging info.
+ - user_interface/enable_escape:
+ Setting this to "1" will enable you abort a hibernation cycle or resuming by
+ pressing escape, "0" (default) disables this feature. Note that enabling
+ this option means that you cannot initiate a hibernation cycle and then walk
+ away from your computer, expecting it to be secure. With feature disabled,
+ you can validly have this expectation once TuxOnice begins to write the
+ image to disk. (Prior to this point, it is possible that TuxOnice might
+ about because of failure to freeze all processes or because constraints
+ on its ability to save the image are not met).
+ - user_interface/program
+ This entry is used to tell TuxOnice what userspace program to use for
+ providing a user interface while hibernating. The program uses a netlink
+ socket to pass messages back and forward to the kernel, allowing all of the
+ functions formerly implemented in the kernel user interface components.
+ - version:
+ The version of TuxOnIce you have compiled into the currently running kernel.
+ - wake_alarm_dir:
+ As mentioned above (post_wake_state), TuxOnIce supports automatically waking
+ after some delay. This entry allows you to select which wake alarm to use.
+ It should contain the value "rtc0" if you're wanting to use
+ /sys/class/rtc/rtc0.
+ - wake_delay:
+ This value determines the delay from the end of writing the image until the
+ wake alarm is triggered. You can set an absolute time by writing the desired
+ time into /sys/class/rtc/<wake_alarm_dir>/wakealarm and leaving these values
+ empty.
+ Note that for the wakeup to actually occur, you may need to modify entries
+ in /proc/acpi/wakeup. This is done by echoing the name of the button in the
+ first column (eg PBTN) into the file.
+7. How do you get support?
+ Glad you asked. TuxOnIce is being actively maintained and supported
+ by Nigel (the guy doing most of the kernel coding at the moment), Bernard
+ (who maintains the hibernate script and userspace user interface components)
+ and its users.
+ Resources availble include HowTos, FAQs and a Wiki, all available via
+ You can find the mailing lists there.
+8. I think I've found a bug. What should I do?
+ By far and a way, the most common problems people have with TuxOnIce
+ related to drivers not having adequate power management support. In this
+ case, it is not a bug with TuxOnIce, but we can still help you. As we
+ mentioned above, such issues can usually be worked around by building the
+ functionality as modules and unloading them while hibernating. Please visit
+ the Wiki for up-to-date lists of known issues and work arounds.
+ If this information doesn't help, try running:
+ hibernate --bug-report
+ ..and sending the output to the users mailing list.
+ Good information on how to provide us with useful information from an
+ oops is found in the file REPORTING-BUGS, in the top level directory
+ of the kernel tree. If you get an oops, please especially note the
+ information about running what is printed on the screen through ksymoops.
+ The raw information is useless.
+9. When will XXX be supported?
+ If there's a feature missing from TuxOnIce that you'd like, feel free to
+ ask. We try to be obliging, within reason.
+ Patches are welcome. Please send to the list.
+10. How does it work?
+ TuxOnIce does its work in a number of steps.
+ a. Freezing system activity.
+ The first main stage in hibernating is to stop all other activity. This is
+ achieved in stages. Processes are considered in fours groups, which we will
+ describe in reverse order for clarity's sake: Threads with the PF_NOFREEZE
+ flag, kernel threads without this flag, userspace processes with the
+ PF_SYNCTHREAD flag and all other processes. The first set (PF_NOFREEZE) are
+ untouched by the refrigerator code. They are allowed to run during hibernating
+ and resuming, and are used to support user interaction, storage access or the
+ like. Other kernel threads (those unneeded while hibernating) are frozen last.
+ This leaves us with userspace processes that need to be frozen. When a
+ process enters one of the *_sync system calls, we set a PF_SYNCTHREAD flag on
+ that process for the duration of that call. Processes that have this flag are
+ frozen after processes without it, so that we can seek to ensure that dirty
+ data is synced to disk as quickly as possible in a situation where other
+ processes may be submitting writes at the same time. Freezing the processes
+ that are submitting data stops new I/O from being submitted. Syncthreads can
+ then cleanly finish their work. So the order is:
+ - Userspace processes without PF_SYNCTHREAD or PF_NOFREEZE;
+ - Userspace processes with PF_SYNCTHREAD (they won't have NOFREEZE);
+ - Kernel processes without PF_NOFREEZE.
+ b. Eating memory.
+ For a successful hibernation cycle, you need to have enough disk space to store the
+ image and enough memory for the various limitations of TuxOnIce's
+ algorithm. You can also specify a maximum image size. In order to attain
+ to those constraints, TuxOnIce may 'eat' memory. If, after freezing
+ processes, the constraints aren't met, TuxOnIce will thaw all the
+ other processes and begin to eat memory until its calculations indicate
+ the constraints are met. It will then freeze processes again and recheck
+ its calculations.
+ c. Allocation of storage.
+ Next, TuxOnIce allocates the storage that will be used to save
+ the image.
+ The core of TuxOnIce knows nothing about how or where pages are stored. We
+ therefore request the active allocator (remember you might have compiled in
+ more than one!) to allocate enough storage for our expect image size. If
+ this request cannot be fulfilled, we eat more memory and try again. If it
+ is fulfiled, we seek to allocate additional storage, just in case our
+ expected compression ratio (if any) isn't achieved. This time, however, we
+ just continue if we can't allocate enough storage.
+ If these calls to our allocator change the characteristics of the image
+ such that we haven't allocated enough memory, we also loop. (The allocator
+ may well need to allocate space for its storage information).
+ d. Write the first part of the image.
+ TuxOnIce stores the image in two sets of pages called 'pagesets'.
+ Pageset 2 contains pages on the active and inactive lists; essentially
+ the page cache. Pageset 1 contains all other pages, including the kernel.
+ We use two pagesets for one important reason: We need to make an atomic copy
+ of the kernel to ensure consistency of the image. Without a second pageset,
+ that would limit us to an image that was at most half the amount of memory
+ available. Using two pagesets allows us to store a full image. Since pageset
+ 2 pages won't be needed in saving pageset 1, we first save pageset 2 pages.
+ We can then make our atomic copy of the remaining pages using both pageset 2
+ pages and any other pages that are free. While saving both pagesets, we are
+ careful not to corrupt the image. Among other things, we use lowlevel block
+ I/O routines that don't change the pagecache contents.
+ The next step, then, is writing pageset 2.
+ e. Suspending drivers and storing processor context.
+ Having written pageset2, TuxOnIce calls the power management functions to
+ notify drivers of the hibernation, and saves the processor state in preparation
+ for the atomic copy of memory we are about to make.
+ f. Atomic copy.
+ At this stage, everything else but the TuxOnIce code is halted. Processes
+ are frozen or idling, drivers are quiesced and have stored (ideally and where
+ necessary) their configuration in memory we are about to atomically copy.
+ In our lowlevel architecture specific code, we have saved the CPU state.
+ We can therefore now do our atomic copy before resuming drivers etc.
+ g. Save the atomic copy (pageset 1).
+ TuxOnice can then write the atomic copy of the remaining pages. Since we
+ have copied the pages into other locations, we can continue to use the
+ normal block I/O routines without fear of corruption our image.
+ f. Save the image header.
+ Nearly there! We save our settings and other parameters needed for
+ reloading pageset 1 in an 'image header'. We also tell our allocator to
+ serialise its data at this stage, so that it can reread the image at resume
+ time.
+ g. Set the image header.
+ Finally, we edit the header at our resume= location. The signature is
+ changed by the allocator to reflect the fact that an image exists, and to
+ point to the start of that data if necessary (swap allocator).
+ h. Power down.
+ Or reboot if we're debugging and the appropriate option is selected.
+ Whew!
+ Reloading the image.
+ --------------------
+ Reloading the image is essentially the reverse of all the above. We load
+ our copy of pageset 1, being careful to choose locations that aren't going
+ to be overwritten as we copy it back (We start very early in the boot
+ process, so there are no other processes to quiesce here). We then copy
+ pageset 1 back to its original location in memory and restore the process
+ context. We are now running with the original kernel. Next, we reload the
+ pageset 2 pages, free the memory and swap used by TuxOnIce, restore
+ the pageset header and restart processes. Sounds easy in comparison to
+ hibernating, doesn't it!
+ There is of course more to TuxOnIce than this, but this explanation
+ should be a good start. If there's interest, I'll write further
+ documentation on range pages and the low level I/O.
+11. Who wrote TuxOnIce?
+ (Answer based on the writings of Florent Chabaud, credits in files and
+ Nigel's limited knowledge; apologies to anyone missed out!)
+ The main developers of TuxOnIce have been...
+ Gabor Kuti
+ Pavel Machek
+ Florent Chabaud
+ Bernard Blackham
+ Nigel Cunningham
+ Significant portions of swsusp, the code in the vanilla kernel which
+ TuxOnIce enhances, have been worked on by Rafael Wysocki. Thanks should
+ also be expressed to him.
+ The above mentioned developers have been aided in their efforts by a host
+ of hundreds, if not thousands of testers and people who have submitted bug
+ fixes & suggestions. Of special note are the efforts of Michael Frank, who
+ had his computers repetitively hibernate and resume for literally tens of
+ thousands of cycles and developed scripts to stress the system and test
+ TuxOnIce far beyond the point most of us (Nigel included!) would consider
+ testing. His efforts have contributed as much to TuxOnIce as any of the
+ names above.
index 2b349ba..d43cd3a 100644
@@ -5630,6 +5630,13 @@ S: Maintained
F: drivers/tc/
F: include/linux/tc.h

+P: Nigel Cunningham
+S: Maintained
P: Dario Ballabio

 \ /
  Last update: 2009-05-06 17:07    [W:0.257 / U:0.240 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site