xref: /linux/Documentation/admin-guide/cgroup-v2.rst (revision 51a8f9d7f587290944d6fc733d1f897091c63159)
1.. _cgroup-v2:
2
3================
4Control Group v2
5================
6
7:Date: October, 2015
8:Author: Tejun Heo <tj@kernel.org>
9
10This is the authoritative documentation on the design, interface and
11conventions of cgroup v2.  It describes all userland-visible aspects
12of cgroup including core and specific controller behaviors.  All
13future changes must be reflected in this document.  Documentation for
14v1 is available under :ref:`Documentation/admin-guide/cgroup-v1/index.rst <cgroup-v1>`.
15
16.. CONTENTS
17
18   1. Introduction
19     1-1. Terminology
20     1-2. What is cgroup?
21   2. Basic Operations
22     2-1. Mounting
23     2-2. Organizing Processes and Threads
24       2-2-1. Processes
25       2-2-2. Threads
26     2-3. [Un]populated Notification
27     2-4. Controlling Controllers
28       2-4-1. Enabling and Disabling
29       2-4-2. Top-down Constraint
30       2-4-3. No Internal Process Constraint
31     2-5. Delegation
32       2-5-1. Model of Delegation
33       2-5-2. Delegation Containment
34     2-6. Guidelines
35       2-6-1. Organize Once and Control
36       2-6-2. Avoid Name Collisions
37   3. Resource Distribution Models
38     3-1. Weights
39     3-2. Limits
40     3-3. Protections
41     3-4. Allocations
42   4. Interface Files
43     4-1. Format
44     4-2. Conventions
45     4-3. Core Interface Files
46   5. Controllers
47     5-1. CPU
48       5-1-1. CPU Interface Files
49     5-2. Memory
50       5-2-1. Memory Interface Files
51       5-2-2. Usage Guidelines
52       5-2-3. Memory Ownership
53     5-3. IO
54       5-3-1. IO Interface Files
55       5-3-2. Writeback
56       5-3-3. IO Latency
57         5-3-3-1. How IO Latency Throttling Works
58         5-3-3-2. IO Latency Interface Files
59       5-3-4. IO Priority
60     5-4. PID
61       5-4-1. PID Interface Files
62     5-5. Cpuset
63       5.5-1. Cpuset Interface Files
64     5-6. Device
65     5-7. RDMA
66       5-7-1. RDMA Interface Files
67     5-8. HugeTLB
68       5.8-1. HugeTLB Interface Files
69     5-9. Misc
70       5.9-1 Miscellaneous cgroup Interface Files
71       5.9-2 Migration and Ownership
72     5-10. Others
73       5-10-1. perf_event
74     5-N. Non-normative information
75       5-N-1. CPU controller root cgroup process behaviour
76       5-N-2. IO controller root cgroup process behaviour
77   6. Namespace
78     6-1. Basics
79     6-2. The Root and Views
80     6-3. Migration and setns(2)
81     6-4. Interaction with Other Namespaces
82   P. Information on Kernel Programming
83     P-1. Filesystem Support for Writeback
84   D. Deprecated v1 Core Features
85   R. Issues with v1 and Rationales for v2
86     R-1. Multiple Hierarchies
87     R-2. Thread Granularity
88     R-3. Competition Between Inner Nodes and Threads
89     R-4. Other Interface Issues
90     R-5. Controller Issues and Remedies
91       R-5-1. Memory
92
93
94Introduction
95============
96
97Terminology
98-----------
99
100"cgroup" stands for "control group" and is never capitalized.  The
101singular form is used to designate the whole feature and also as a
102qualifier as in "cgroup controllers".  When explicitly referring to
103multiple individual control groups, the plural form "cgroups" is used.
104
105
106What is cgroup?
107---------------
108
109cgroup is a mechanism to organize processes hierarchically and
110distribute system resources along the hierarchy in a controlled and
111configurable manner.
112
113cgroup is largely composed of two parts - the core and controllers.
114cgroup core is primarily responsible for hierarchically organizing
115processes.  A cgroup controller is usually responsible for
116distributing a specific type of system resource along the hierarchy
117although there are utility controllers which serve purposes other than
118resource distribution.
119
120cgroups form a tree structure and every process in the system belongs
121to one and only one cgroup.  All threads of a process belong to the
122same cgroup.  On creation, all processes are put in the cgroup that
123the parent process belongs to at the time.  A process can be migrated
124to another cgroup.  Migration of a process doesn't affect already
125existing descendant processes.
126
127Following certain structural constraints, controllers may be enabled or
128disabled selectively on a cgroup.  All controller behaviors are
129hierarchical - if a controller is enabled on a cgroup, it affects all
130processes which belong to the cgroups consisting the inclusive
131sub-hierarchy of the cgroup.  When a controller is enabled on a nested
132cgroup, it always restricts the resource distribution further.  The
133restrictions set closer to the root in the hierarchy can not be
134overridden from further away.
135
136
137Basic Operations
138================
139
140Mounting
141--------
142
143Unlike v1, cgroup v2 has only single hierarchy.  The cgroup v2
144hierarchy can be mounted with the following mount command::
145
146  # mount -t cgroup2 none $MOUNT_POINT
147
148cgroup2 filesystem has the magic number 0x63677270 ("cgrp").  All
149controllers which support v2 and are not bound to a v1 hierarchy are
150automatically bound to the v2 hierarchy and show up at the root.
151Controllers which are not in active use in the v2 hierarchy can be
152bound to other hierarchies.  This allows mixing v2 hierarchy with the
153legacy v1 multiple hierarchies in a fully backward compatible way.
154
155A controller can be moved across hierarchies only after the controller
156is no longer referenced in its current hierarchy.  Because per-cgroup
157controller states are destroyed asynchronously and controllers may
158have lingering references, a controller may not show up immediately on
159the v2 hierarchy after the final umount of the previous hierarchy.
160Similarly, a controller should be fully disabled to be moved out of
161the unified hierarchy and it may take some time for the disabled
162controller to become available for other hierarchies; furthermore, due
163to inter-controller dependencies, other controllers may need to be
164disabled too.
165
166While useful for development and manual configurations, moving
167controllers dynamically between the v2 and other hierarchies is
168strongly discouraged for production use.  It is recommended to decide
169the hierarchies and controller associations before starting using the
170controllers after system boot.
171
172During transition to v2, system management software might still
173automount the v1 cgroup filesystem and so hijack all controllers
174during boot, before manual intervention is possible. To make testing
175and experimenting easier, the kernel parameter cgroup_no_v1= allows
176disabling controllers in v1 and make them always available in v2.
177
178cgroup v2 currently supports the following mount options.
179
180  nsdelegate
181	Consider cgroup namespaces as delegation boundaries.  This
182	option is system wide and can only be set on mount or modified
183	through remount from the init namespace.  The mount option is
184	ignored on non-init namespace mounts.  Please refer to the
185	Delegation section for details.
186
187  favordynmods
188        Reduce the latencies of dynamic cgroup modifications such as
189        task migrations and controller on/offs at the cost of making
190        hot path operations such as forks and exits more expensive.
191        The static usage pattern of creating a cgroup, enabling
192        controllers, and then seeding it with CLONE_INTO_CGROUP is
193        not affected by this option.
194
195  memory_localevents
196        Only populate memory.events with data for the current cgroup,
197        and not any subtrees. This is legacy behaviour, the default
198        behaviour without this option is to include subtree counts.
199        This option is system wide and can only be set on mount or
200        modified through remount from the init namespace. The mount
201        option is ignored on non-init namespace mounts.
202
203  memory_recursiveprot
204        Recursively apply memory.min and memory.low protection to
205        entire subtrees, without requiring explicit downward
206        propagation into leaf cgroups.  This allows protecting entire
207        subtrees from one another, while retaining free competition
208        within those subtrees.  This should have been the default
209        behavior but is a mount-option to avoid regressing setups
210        relying on the original semantics (e.g. specifying bogusly
211        high 'bypass' protection values at higher tree levels).
212
213
214Organizing Processes and Threads
215--------------------------------
216
217Processes
218~~~~~~~~~
219
220Initially, only the root cgroup exists to which all processes belong.
221A child cgroup can be created by creating a sub-directory::
222
223  # mkdir $CGROUP_NAME
224
225A given cgroup may have multiple child cgroups forming a tree
226structure.  Each cgroup has a read-writable interface file
227"cgroup.procs".  When read, it lists the PIDs of all processes which
228belong to the cgroup one-per-line.  The PIDs are not ordered and the
229same PID may show up more than once if the process got moved to
230another cgroup and then back or the PID got recycled while reading.
231
232A process can be migrated into a cgroup by writing its PID to the
233target cgroup's "cgroup.procs" file.  Only one process can be migrated
234on a single write(2) call.  If a process is composed of multiple
235threads, writing the PID of any thread migrates all threads of the
236process.
237
238When a process forks a child process, the new process is born into the
239cgroup that the forking process belongs to at the time of the
240operation.  After exit, a process stays associated with the cgroup
241that it belonged to at the time of exit until it's reaped; however, a
242zombie process does not appear in "cgroup.procs" and thus can't be
243moved to another cgroup.
244
245A cgroup which doesn't have any children or live processes can be
246destroyed by removing the directory.  Note that a cgroup which doesn't
247have any children and is associated only with zombie processes is
248considered empty and can be removed::
249
250  # rmdir $CGROUP_NAME
251
252"/proc/$PID/cgroup" lists a process's cgroup membership.  If legacy
253cgroup is in use in the system, this file may contain multiple lines,
254one for each hierarchy.  The entry for cgroup v2 is always in the
255format "0::$PATH"::
256
257  # cat /proc/842/cgroup
258  ...
259  0::/test-cgroup/test-cgroup-nested
260
261If the process becomes a zombie and the cgroup it was associated with
262is removed subsequently, " (deleted)" is appended to the path::
263
264  # cat /proc/842/cgroup
265  ...
266  0::/test-cgroup/test-cgroup-nested (deleted)
267
268
269Threads
270~~~~~~~
271
272cgroup v2 supports thread granularity for a subset of controllers to
273support use cases requiring hierarchical resource distribution across
274the threads of a group of processes.  By default, all threads of a
275process belong to the same cgroup, which also serves as the resource
276domain to host resource consumptions which are not specific to a
277process or thread.  The thread mode allows threads to be spread across
278a subtree while still maintaining the common resource domain for them.
279
280Controllers which support thread mode are called threaded controllers.
281The ones which don't are called domain controllers.
282
283Marking a cgroup threaded makes it join the resource domain of its
284parent as a threaded cgroup.  The parent may be another threaded
285cgroup whose resource domain is further up in the hierarchy.  The root
286of a threaded subtree, that is, the nearest ancestor which is not
287threaded, is called threaded domain or thread root interchangeably and
288serves as the resource domain for the entire subtree.
289
290Inside a threaded subtree, threads of a process can be put in
291different cgroups and are not subject to the no internal process
292constraint - threaded controllers can be enabled on non-leaf cgroups
293whether they have threads in them or not.
294
295As the threaded domain cgroup hosts all the domain resource
296consumptions of the subtree, it is considered to have internal
297resource consumptions whether there are processes in it or not and
298can't have populated child cgroups which aren't threaded.  Because the
299root cgroup is not subject to no internal process constraint, it can
300serve both as a threaded domain and a parent to domain cgroups.
301
302The current operation mode or type of the cgroup is shown in the
303"cgroup.type" file which indicates whether the cgroup is a normal
304domain, a domain which is serving as the domain of a threaded subtree,
305or a threaded cgroup.
306
307On creation, a cgroup is always a domain cgroup and can be made
308threaded by writing "threaded" to the "cgroup.type" file.  The
309operation is single direction::
310
311  # echo threaded > cgroup.type
312
313Once threaded, the cgroup can't be made a domain again.  To enable the
314thread mode, the following conditions must be met.
315
316- As the cgroup will join the parent's resource domain.  The parent
317  must either be a valid (threaded) domain or a threaded cgroup.
318
319- When the parent is an unthreaded domain, it must not have any domain
320  controllers enabled or populated domain children.  The root is
321  exempt from this requirement.
322
323Topology-wise, a cgroup can be in an invalid state.  Please consider
324the following topology::
325
326  A (threaded domain) - B (threaded) - C (domain, just created)
327
328C is created as a domain but isn't connected to a parent which can
329host child domains.  C can't be used until it is turned into a
330threaded cgroup.  "cgroup.type" file will report "domain (invalid)" in
331these cases.  Operations which fail due to invalid topology use
332EOPNOTSUPP as the errno.
333
334A domain cgroup is turned into a threaded domain when one of its child
335cgroup becomes threaded or threaded controllers are enabled in the
336"cgroup.subtree_control" file while there are processes in the cgroup.
337A threaded domain reverts to a normal domain when the conditions
338clear.
339
340When read, "cgroup.threads" contains the list of the thread IDs of all
341threads in the cgroup.  Except that the operations are per-thread
342instead of per-process, "cgroup.threads" has the same format and
343behaves the same way as "cgroup.procs".  While "cgroup.threads" can be
344written to in any cgroup, as it can only move threads inside the same
345threaded domain, its operations are confined inside each threaded
346subtree.
347
348The threaded domain cgroup serves as the resource domain for the whole
349subtree, and, while the threads can be scattered across the subtree,
350all the processes are considered to be in the threaded domain cgroup.
351"cgroup.procs" in a threaded domain cgroup contains the PIDs of all
352processes in the subtree and is not readable in the subtree proper.
353However, "cgroup.procs" can be written to from anywhere in the subtree
354to migrate all threads of the matching process to the cgroup.
355
356Only threaded controllers can be enabled in a threaded subtree.  When
357a threaded controller is enabled inside a threaded subtree, it only
358accounts for and controls resource consumptions associated with the
359threads in the cgroup and its descendants.  All consumptions which
360aren't tied to a specific thread belong to the threaded domain cgroup.
361
362Because a threaded subtree is exempt from no internal process
363constraint, a threaded controller must be able to handle competition
364between threads in a non-leaf cgroup and its child cgroups.  Each
365threaded controller defines how such competitions are handled.
366
367
368[Un]populated Notification
369--------------------------
370
371Each non-root cgroup has a "cgroup.events" file which contains
372"populated" field indicating whether the cgroup's sub-hierarchy has
373live processes in it.  Its value is 0 if there is no live process in
374the cgroup and its descendants; otherwise, 1.  poll and [id]notify
375events are triggered when the value changes.  This can be used, for
376example, to start a clean-up operation after all processes of a given
377sub-hierarchy have exited.  The populated state updates and
378notifications are recursive.  Consider the following sub-hierarchy
379where the numbers in the parentheses represent the numbers of processes
380in each cgroup::
381
382  A(4) - B(0) - C(1)
383              \ D(0)
384
385A, B and C's "populated" fields would be 1 while D's 0.  After the one
386process in C exits, B and C's "populated" fields would flip to "0" and
387file modified events will be generated on the "cgroup.events" files of
388both cgroups.
389
390
391Controlling Controllers
392-----------------------
393
394Enabling and Disabling
395~~~~~~~~~~~~~~~~~~~~~~
396
397Each cgroup has a "cgroup.controllers" file which lists all
398controllers available for the cgroup to enable::
399
400  # cat cgroup.controllers
401  cpu io memory
402
403No controller is enabled by default.  Controllers can be enabled and
404disabled by writing to the "cgroup.subtree_control" file::
405
406  # echo "+cpu +memory -io" > cgroup.subtree_control
407
408Only controllers which are listed in "cgroup.controllers" can be
409enabled.  When multiple operations are specified as above, either they
410all succeed or fail.  If multiple operations on the same controller
411are specified, the last one is effective.
412
413Enabling a controller in a cgroup indicates that the distribution of
414the target resource across its immediate children will be controlled.
415Consider the following sub-hierarchy.  The enabled controllers are
416listed in parentheses::
417
418  A(cpu,memory) - B(memory) - C()
419                            \ D()
420
421As A has "cpu" and "memory" enabled, A will control the distribution
422of CPU cycles and memory to its children, in this case, B.  As B has
423"memory" enabled but not "CPU", C and D will compete freely on CPU
424cycles but their division of memory available to B will be controlled.
425
426As a controller regulates the distribution of the target resource to
427the cgroup's children, enabling it creates the controller's interface
428files in the child cgroups.  In the above example, enabling "cpu" on B
429would create the "cpu." prefixed controller interface files in C and
430D.  Likewise, disabling "memory" from B would remove the "memory."
431prefixed controller interface files from C and D.  This means that the
432controller interface files - anything which doesn't start with
433"cgroup." are owned by the parent rather than the cgroup itself.
434
435
436Top-down Constraint
437~~~~~~~~~~~~~~~~~~~
438
439Resources are distributed top-down and a cgroup can further distribute
440a resource only if the resource has been distributed to it from the
441parent.  This means that all non-root "cgroup.subtree_control" files
442can only contain controllers which are enabled in the parent's
443"cgroup.subtree_control" file.  A controller can be enabled only if
444the parent has the controller enabled and a controller can't be
445disabled if one or more children have it enabled.
446
447
448No Internal Process Constraint
449~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
450
451Non-root cgroups can distribute domain resources to their children
452only when they don't have any processes of their own.  In other words,
453only domain cgroups which don't contain any processes can have domain
454controllers enabled in their "cgroup.subtree_control" files.
455
456This guarantees that, when a domain controller is looking at the part
457of the hierarchy which has it enabled, processes are always only on
458the leaves.  This rules out situations where child cgroups compete
459against internal processes of the parent.
460
461The root cgroup is exempt from this restriction.  Root contains
462processes and anonymous resource consumption which can't be associated
463with any other cgroups and requires special treatment from most
464controllers.  How resource consumption in the root cgroup is governed
465is up to each controller (for more information on this topic please
466refer to the Non-normative information section in the Controllers
467chapter).
468
469Note that the restriction doesn't get in the way if there is no
470enabled controller in the cgroup's "cgroup.subtree_control".  This is
471important as otherwise it wouldn't be possible to create children of a
472populated cgroup.  To control resource distribution of a cgroup, the
473cgroup must create children and transfer all its processes to the
474children before enabling controllers in its "cgroup.subtree_control"
475file.
476
477
478Delegation
479----------
480
481Model of Delegation
482~~~~~~~~~~~~~~~~~~~
483
484A cgroup can be delegated in two ways.  First, to a less privileged
485user by granting write access of the directory and its "cgroup.procs",
486"cgroup.threads" and "cgroup.subtree_control" files to the user.
487Second, if the "nsdelegate" mount option is set, automatically to a
488cgroup namespace on namespace creation.
489
490Because the resource control interface files in a given directory
491control the distribution of the parent's resources, the delegatee
492shouldn't be allowed to write to them.  For the first method, this is
493achieved by not granting access to these files.  For the second, the
494kernel rejects writes to all files other than "cgroup.procs" and
495"cgroup.subtree_control" on a namespace root from inside the
496namespace.
497
498The end results are equivalent for both delegation types.  Once
499delegated, the user can build sub-hierarchy under the directory,
500organize processes inside it as it sees fit and further distribute the
501resources it received from the parent.  The limits and other settings
502of all resource controllers are hierarchical and regardless of what
503happens in the delegated sub-hierarchy, nothing can escape the
504resource restrictions imposed by the parent.
505
506Currently, cgroup doesn't impose any restrictions on the number of
507cgroups in or nesting depth of a delegated sub-hierarchy; however,
508this may be limited explicitly in the future.
509
510
511Delegation Containment
512~~~~~~~~~~~~~~~~~~~~~~
513
514A delegated sub-hierarchy is contained in the sense that processes
515can't be moved into or out of the sub-hierarchy by the delegatee.
516
517For delegations to a less privileged user, this is achieved by
518requiring the following conditions for a process with a non-root euid
519to migrate a target process into a cgroup by writing its PID to the
520"cgroup.procs" file.
521
522- The writer must have write access to the "cgroup.procs" file.
523
524- The writer must have write access to the "cgroup.procs" file of the
525  common ancestor of the source and destination cgroups.
526
527The above two constraints ensure that while a delegatee may migrate
528processes around freely in the delegated sub-hierarchy it can't pull
529in from or push out to outside the sub-hierarchy.
530
531For an example, let's assume cgroups C0 and C1 have been delegated to
532user U0 who created C00, C01 under C0 and C10 under C1 as follows and
533all processes under C0 and C1 belong to U0::
534
535  ~~~~~~~~~~~~~ - C0 - C00
536  ~ cgroup    ~      \ C01
537  ~ hierarchy ~
538  ~~~~~~~~~~~~~ - C1 - C10
539
540Let's also say U0 wants to write the PID of a process which is
541currently in C10 into "C00/cgroup.procs".  U0 has write access to the
542file; however, the common ancestor of the source cgroup C10 and the
543destination cgroup C00 is above the points of delegation and U0 would
544not have write access to its "cgroup.procs" files and thus the write
545will be denied with -EACCES.
546
547For delegations to namespaces, containment is achieved by requiring
548that both the source and destination cgroups are reachable from the
549namespace of the process which is attempting the migration.  If either
550is not reachable, the migration is rejected with -ENOENT.
551
552
553Guidelines
554----------
555
556Organize Once and Control
557~~~~~~~~~~~~~~~~~~~~~~~~~
558
559Migrating a process across cgroups is a relatively expensive operation
560and stateful resources such as memory are not moved together with the
561process.  This is an explicit design decision as there often exist
562inherent trade-offs between migration and various hot paths in terms
563of synchronization cost.
564
565As such, migrating processes across cgroups frequently as a means to
566apply different resource restrictions is discouraged.  A workload
567should be assigned to a cgroup according to the system's logical and
568resource structure once on start-up.  Dynamic adjustments to resource
569distribution can be made by changing controller configuration through
570the interface files.
571
572
573Avoid Name Collisions
574~~~~~~~~~~~~~~~~~~~~~
575
576Interface files for a cgroup and its children cgroups occupy the same
577directory and it is possible to create children cgroups which collide
578with interface files.
579
580All cgroup core interface files are prefixed with "cgroup." and each
581controller's interface files are prefixed with the controller name and
582a dot.  A controller's name is composed of lower case alphabets and
583'_'s but never begins with an '_' so it can be used as the prefix
584character for collision avoidance.  Also, interface file names won't
585start or end with terms which are often used in categorizing workloads
586such as job, service, slice, unit or workload.
587
588cgroup doesn't do anything to prevent name collisions and it's the
589user's responsibility to avoid them.
590
591
592Resource Distribution Models
593============================
594
595cgroup controllers implement several resource distribution schemes
596depending on the resource type and expected use cases.  This section
597describes major schemes in use along with their expected behaviors.
598
599
600Weights
601-------
602
603A parent's resource is distributed by adding up the weights of all
604active children and giving each the fraction matching the ratio of its
605weight against the sum.  As only children which can make use of the
606resource at the moment participate in the distribution, this is
607work-conserving.  Due to the dynamic nature, this model is usually
608used for stateless resources.
609
610All weights are in the range [1, 10000] with the default at 100.  This
611allows symmetric multiplicative biases in both directions at fine
612enough granularity while staying in the intuitive range.
613
614As long as the weight is in range, all configuration combinations are
615valid and there is no reason to reject configuration changes or
616process migrations.
617
618"cpu.weight" proportionally distributes CPU cycles to active children
619and is an example of this type.
620
621
622Limits
623------
624
625A child can only consume upto the configured amount of the resource.
626Limits can be over-committed - the sum of the limits of children can
627exceed the amount of resource available to the parent.
628
629Limits are in the range [0, max] and defaults to "max", which is noop.
630
631As limits can be over-committed, all configuration combinations are
632valid and there is no reason to reject configuration changes or
633process migrations.
634
635"io.max" limits the maximum BPS and/or IOPS that a cgroup can consume
636on an IO device and is an example of this type.
637
638
639Protections
640-----------
641
642A cgroup is protected upto the configured amount of the resource
643as long as the usages of all its ancestors are under their
644protected levels.  Protections can be hard guarantees or best effort
645soft boundaries.  Protections can also be over-committed in which case
646only upto the amount available to the parent is protected among
647children.
648
649Protections are in the range [0, max] and defaults to 0, which is
650noop.
651
652As protections can be over-committed, all configuration combinations
653are valid and there is no reason to reject configuration changes or
654process migrations.
655
656"memory.low" implements best-effort memory protection and is an
657example of this type.
658
659
660Allocations
661-----------
662
663A cgroup is exclusively allocated a certain amount of a finite
664resource.  Allocations can't be over-committed - the sum of the
665allocations of children can not exceed the amount of resource
666available to the parent.
667
668Allocations are in the range [0, max] and defaults to 0, which is no
669resource.
670
671As allocations can't be over-committed, some configuration
672combinations are invalid and should be rejected.  Also, if the
673resource is mandatory for execution of processes, process migrations
674may be rejected.
675
676"cpu.rt.max" hard-allocates realtime slices and is an example of this
677type.
678
679
680Interface Files
681===============
682
683Format
684------
685
686All interface files should be in one of the following formats whenever
687possible::
688
689  New-line separated values
690  (when only one value can be written at once)
691
692	VAL0\n
693	VAL1\n
694	...
695
696  Space separated values
697  (when read-only or multiple values can be written at once)
698
699	VAL0 VAL1 ...\n
700
701  Flat keyed
702
703	KEY0 VAL0\n
704	KEY1 VAL1\n
705	...
706
707  Nested keyed
708
709	KEY0 SUB_KEY0=VAL00 SUB_KEY1=VAL01...
710	KEY1 SUB_KEY0=VAL10 SUB_KEY1=VAL11...
711	...
712
713For a writable file, the format for writing should generally match
714reading; however, controllers may allow omitting later fields or
715implement restricted shortcuts for most common use cases.
716
717For both flat and nested keyed files, only the values for a single key
718can be written at a time.  For nested keyed files, the sub key pairs
719may be specified in any order and not all pairs have to be specified.
720
721
722Conventions
723-----------
724
725- Settings for a single feature should be contained in a single file.
726
727- The root cgroup should be exempt from resource control and thus
728  shouldn't have resource control interface files.
729
730- The default time unit is microseconds.  If a different unit is ever
731  used, an explicit unit suffix must be present.
732
733- A parts-per quantity should use a percentage decimal with at least
734  two digit fractional part - e.g. 13.40.
735
736- If a controller implements weight based resource distribution, its
737  interface file should be named "weight" and have the range [1,
738  10000] with 100 as the default.  The values are chosen to allow
739  enough and symmetric bias in both directions while keeping it
740  intuitive (the default is 100%).
741
742- If a controller implements an absolute resource guarantee and/or
743  limit, the interface files should be named "min" and "max"
744  respectively.  If a controller implements best effort resource
745  guarantee and/or limit, the interface files should be named "low"
746  and "high" respectively.
747
748  In the above four control files, the special token "max" should be
749  used to represent upward infinity for both reading and writing.
750
751- If a setting has a configurable default value and keyed specific
752  overrides, the default entry should be keyed with "default" and
753  appear as the first entry in the file.
754
755  The default value can be updated by writing either "default $VAL" or
756  "$VAL".
757
758  When writing to update a specific override, "default" can be used as
759  the value to indicate removal of the override.  Override entries
760  with "default" as the value must not appear when read.
761
762  For example, a setting which is keyed by major:minor device numbers
763  with integer values may look like the following::
764
765    # cat cgroup-example-interface-file
766    default 150
767    8:0 300
768
769  The default value can be updated by::
770
771    # echo 125 > cgroup-example-interface-file
772
773  or::
774
775    # echo "default 125" > cgroup-example-interface-file
776
777  An override can be set by::
778
779    # echo "8:16 170" > cgroup-example-interface-file
780
781  and cleared by::
782
783    # echo "8:0 default" > cgroup-example-interface-file
784    # cat cgroup-example-interface-file
785    default 125
786    8:16 170
787
788- For events which are not very high frequency, an interface file
789  "events" should be created which lists event key value pairs.
790  Whenever a notifiable event happens, file modified event should be
791  generated on the file.
792
793
794Core Interface Files
795--------------------
796
797All cgroup core files are prefixed with "cgroup."
798
799  cgroup.type
800	A read-write single value file which exists on non-root
801	cgroups.
802
803	When read, it indicates the current type of the cgroup, which
804	can be one of the following values.
805
806	- "domain" : A normal valid domain cgroup.
807
808	- "domain threaded" : A threaded domain cgroup which is
809          serving as the root of a threaded subtree.
810
811	- "domain invalid" : A cgroup which is in an invalid state.
812	  It can't be populated or have controllers enabled.  It may
813	  be allowed to become a threaded cgroup.
814
815	- "threaded" : A threaded cgroup which is a member of a
816          threaded subtree.
817
818	A cgroup can be turned into a threaded cgroup by writing
819	"threaded" to this file.
820
821  cgroup.procs
822	A read-write new-line separated values file which exists on
823	all cgroups.
824
825	When read, it lists the PIDs of all processes which belong to
826	the cgroup one-per-line.  The PIDs are not ordered and the
827	same PID may show up more than once if the process got moved
828	to another cgroup and then back or the PID got recycled while
829	reading.
830
831	A PID can be written to migrate the process associated with
832	the PID to the cgroup.  The writer should match all of the
833	following conditions.
834
835	- It must have write access to the "cgroup.procs" file.
836
837	- It must have write access to the "cgroup.procs" file of the
838	  common ancestor of the source and destination cgroups.
839
840	When delegating a sub-hierarchy, write access to this file
841	should be granted along with the containing directory.
842
843	In a threaded cgroup, reading this file fails with EOPNOTSUPP
844	as all the processes belong to the thread root.  Writing is
845	supported and moves every thread of the process to the cgroup.
846
847  cgroup.threads
848	A read-write new-line separated values file which exists on
849	all cgroups.
850
851	When read, it lists the TIDs of all threads which belong to
852	the cgroup one-per-line.  The TIDs are not ordered and the
853	same TID may show up more than once if the thread got moved to
854	another cgroup and then back or the TID got recycled while
855	reading.
856
857	A TID can be written to migrate the thread associated with the
858	TID to the cgroup.  The writer should match all of the
859	following conditions.
860
861	- It must have write access to the "cgroup.threads" file.
862
863	- The cgroup that the thread is currently in must be in the
864          same resource domain as the destination cgroup.
865
866	- It must have write access to the "cgroup.procs" file of the
867	  common ancestor of the source and destination cgroups.
868
869	When delegating a sub-hierarchy, write access to this file
870	should be granted along with the containing directory.
871
872  cgroup.controllers
873	A read-only space separated values file which exists on all
874	cgroups.
875
876	It shows space separated list of all controllers available to
877	the cgroup.  The controllers are not ordered.
878
879  cgroup.subtree_control
880	A read-write space separated values file which exists on all
881	cgroups.  Starts out empty.
882
883	When read, it shows space separated list of the controllers
884	which are enabled to control resource distribution from the
885	cgroup to its children.
886
887	Space separated list of controllers prefixed with '+' or '-'
888	can be written to enable or disable controllers.  A controller
889	name prefixed with '+' enables the controller and '-'
890	disables.  If a controller appears more than once on the list,
891	the last one is effective.  When multiple enable and disable
892	operations are specified, either all succeed or all fail.
893
894  cgroup.events
895	A read-only flat-keyed file which exists on non-root cgroups.
896	The following entries are defined.  Unless specified
897	otherwise, a value change in this file generates a file
898	modified event.
899
900	  populated
901		1 if the cgroup or its descendants contains any live
902		processes; otherwise, 0.
903	  frozen
904		1 if the cgroup is frozen; otherwise, 0.
905
906  cgroup.max.descendants
907	A read-write single value files.  The default is "max".
908
909	Maximum allowed number of descent cgroups.
910	If the actual number of descendants is equal or larger,
911	an attempt to create a new cgroup in the hierarchy will fail.
912
913  cgroup.max.depth
914	A read-write single value files.  The default is "max".
915
916	Maximum allowed descent depth below the current cgroup.
917	If the actual descent depth is equal or larger,
918	an attempt to create a new child cgroup will fail.
919
920  cgroup.stat
921	A read-only flat-keyed file with the following entries:
922
923	  nr_descendants
924		Total number of visible descendant cgroups.
925
926	  nr_dying_descendants
927		Total number of dying descendant cgroups. A cgroup becomes
928		dying after being deleted by a user. The cgroup will remain
929		in dying state for some time undefined time (which can depend
930		on system load) before being completely destroyed.
931
932		A process can't enter a dying cgroup under any circumstances,
933		a dying cgroup can't revive.
934
935		A dying cgroup can consume system resources not exceeding
936		limits, which were active at the moment of cgroup deletion.
937
938  cgroup.freeze
939	A read-write single value file which exists on non-root cgroups.
940	Allowed values are "0" and "1". The default is "0".
941
942	Writing "1" to the file causes freezing of the cgroup and all
943	descendant cgroups. This means that all belonging processes will
944	be stopped and will not run until the cgroup will be explicitly
945	unfrozen. Freezing of the cgroup may take some time; when this action
946	is completed, the "frozen" value in the cgroup.events control file
947	will be updated to "1" and the corresponding notification will be
948	issued.
949
950	A cgroup can be frozen either by its own settings, or by settings
951	of any ancestor cgroups. If any of ancestor cgroups is frozen, the
952	cgroup will remain frozen.
953
954	Processes in the frozen cgroup can be killed by a fatal signal.
955	They also can enter and leave a frozen cgroup: either by an explicit
956	move by a user, or if freezing of the cgroup races with fork().
957	If a process is moved to a frozen cgroup, it stops. If a process is
958	moved out of a frozen cgroup, it becomes running.
959
960	Frozen status of a cgroup doesn't affect any cgroup tree operations:
961	it's possible to delete a frozen (and empty) cgroup, as well as
962	create new sub-cgroups.
963
964  cgroup.kill
965	A write-only single value file which exists in non-root cgroups.
966	The only allowed value is "1".
967
968	Writing "1" to the file causes the cgroup and all descendant cgroups to
969	be killed. This means that all processes located in the affected cgroup
970	tree will be killed via SIGKILL.
971
972	Killing a cgroup tree will deal with concurrent forks appropriately and
973	is protected against migrations.
974
975	In a threaded cgroup, writing this file fails with EOPNOTSUPP as
976	killing cgroups is a process directed operation, i.e. it affects
977	the whole thread-group.
978
979  cgroup.pressure
980	A read-write single value file that allowed values are "0" and "1".
981	The default is "1".
982
983	Writing "0" to the file will disable the cgroup PSI accounting.
984	Writing "1" to the file will re-enable the cgroup PSI accounting.
985
986	This control attribute is not hierarchical, so disable or enable PSI
987	accounting in a cgroup does not affect PSI accounting in descendants
988	and doesn't need pass enablement via ancestors from root.
989
990	The reason this control attribute exists is that PSI accounts stalls for
991	each cgroup separately and aggregates it at each level of the hierarchy.
992	This may cause non-negligible overhead for some workloads when under
993	deep level of the hierarchy, in which case this control attribute can
994	be used to disable PSI accounting in the non-leaf cgroups.
995
996  irq.pressure
997	A read-write nested-keyed file.
998
999	Shows pressure stall information for IRQ/SOFTIRQ. See
1000	:ref:`Documentation/accounting/psi.rst <psi>` for details.
1001
1002Controllers
1003===========
1004
1005.. _cgroup-v2-cpu:
1006
1007CPU
1008---
1009
1010The "cpu" controllers regulates distribution of CPU cycles.  This
1011controller implements weight and absolute bandwidth limit models for
1012normal scheduling policy and absolute bandwidth allocation model for
1013realtime scheduling policy.
1014
1015In all the above models, cycles distribution is defined only on a temporal
1016base and it does not account for the frequency at which tasks are executed.
1017The (optional) utilization clamping support allows to hint the schedutil
1018cpufreq governor about the minimum desired frequency which should always be
1019provided by a CPU, as well as the maximum desired frequency, which should not
1020be exceeded by a CPU.
1021
1022WARNING: cgroup2 doesn't yet support control of realtime processes and
1023the cpu controller can only be enabled when all RT processes are in
1024the root cgroup.  Be aware that system management software may already
1025have placed RT processes into nonroot cgroups during the system boot
1026process, and these processes may need to be moved to the root cgroup
1027before the cpu controller can be enabled.
1028
1029
1030CPU Interface Files
1031~~~~~~~~~~~~~~~~~~~
1032
1033All time durations are in microseconds.
1034
1035  cpu.stat
1036	A read-only flat-keyed file.
1037	This file exists whether the controller is enabled or not.
1038
1039	It always reports the following three stats:
1040
1041	- usage_usec
1042	- user_usec
1043	- system_usec
1044
1045	and the following three when the controller is enabled:
1046
1047	- nr_periods
1048	- nr_throttled
1049	- throttled_usec
1050	- nr_bursts
1051	- burst_usec
1052
1053  cpu.weight
1054	A read-write single value file which exists on non-root
1055	cgroups.  The default is "100".
1056
1057	The weight in the range [1, 10000].
1058
1059  cpu.weight.nice
1060	A read-write single value file which exists on non-root
1061	cgroups.  The default is "0".
1062
1063	The nice value is in the range [-20, 19].
1064
1065	This interface file is an alternative interface for
1066	"cpu.weight" and allows reading and setting weight using the
1067	same values used by nice(2).  Because the range is smaller and
1068	granularity is coarser for the nice values, the read value is
1069	the closest approximation of the current weight.
1070
1071  cpu.max
1072	A read-write two value file which exists on non-root cgroups.
1073	The default is "max 100000".
1074
1075	The maximum bandwidth limit.  It's in the following format::
1076
1077	  $MAX $PERIOD
1078
1079	which indicates that the group may consume upto $MAX in each
1080	$PERIOD duration.  "max" for $MAX indicates no limit.  If only
1081	one number is written, $MAX is updated.
1082
1083  cpu.max.burst
1084	A read-write single value file which exists on non-root
1085	cgroups.  The default is "0".
1086
1087	The burst in the range [0, $MAX].
1088
1089  cpu.pressure
1090	A read-write nested-keyed file.
1091
1092	Shows pressure stall information for CPU. See
1093	:ref:`Documentation/accounting/psi.rst <psi>` for details.
1094
1095  cpu.uclamp.min
1096        A read-write single value file which exists on non-root cgroups.
1097        The default is "0", i.e. no utilization boosting.
1098
1099        The requested minimum utilization (protection) as a percentage
1100        rational number, e.g. 12.34 for 12.34%.
1101
1102        This interface allows reading and setting minimum utilization clamp
1103        values similar to the sched_setattr(2). This minimum utilization
1104        value is used to clamp the task specific minimum utilization clamp.
1105
1106        The requested minimum utilization (protection) is always capped by
1107        the current value for the maximum utilization (limit), i.e.
1108        `cpu.uclamp.max`.
1109
1110  cpu.uclamp.max
1111        A read-write single value file which exists on non-root cgroups.
1112        The default is "max". i.e. no utilization capping
1113
1114        The requested maximum utilization (limit) as a percentage rational
1115        number, e.g. 98.76 for 98.76%.
1116
1117        This interface allows reading and setting maximum utilization clamp
1118        values similar to the sched_setattr(2). This maximum utilization
1119        value is used to clamp the task specific maximum utilization clamp.
1120
1121
1122
1123Memory
1124------
1125
1126The "memory" controller regulates distribution of memory.  Memory is
1127stateful and implements both limit and protection models.  Due to the
1128intertwining between memory usage and reclaim pressure and the
1129stateful nature of memory, the distribution model is relatively
1130complex.
1131
1132While not completely water-tight, all major memory usages by a given
1133cgroup are tracked so that the total memory consumption can be
1134accounted and controlled to a reasonable extent.  Currently, the
1135following types of memory usages are tracked.
1136
1137- Userland memory - page cache and anonymous memory.
1138
1139- Kernel data structures such as dentries and inodes.
1140
1141- TCP socket buffers.
1142
1143The above list may expand in the future for better coverage.
1144
1145
1146Memory Interface Files
1147~~~~~~~~~~~~~~~~~~~~~~
1148
1149All memory amounts are in bytes.  If a value which is not aligned to
1150PAGE_SIZE is written, the value may be rounded up to the closest
1151PAGE_SIZE multiple when read back.
1152
1153  memory.current
1154	A read-only single value file which exists on non-root
1155	cgroups.
1156
1157	The total amount of memory currently being used by the cgroup
1158	and its descendants.
1159
1160  memory.min
1161	A read-write single value file which exists on non-root
1162	cgroups.  The default is "0".
1163
1164	Hard memory protection.  If the memory usage of a cgroup
1165	is within its effective min boundary, the cgroup's memory
1166	won't be reclaimed under any conditions. If there is no
1167	unprotected reclaimable memory available, OOM killer
1168	is invoked. Above the effective min boundary (or
1169	effective low boundary if it is higher), pages are reclaimed
1170	proportionally to the overage, reducing reclaim pressure for
1171	smaller overages.
1172
1173	Effective min boundary is limited by memory.min values of
1174	all ancestor cgroups. If there is memory.min overcommitment
1175	(child cgroup or cgroups are requiring more protected memory
1176	than parent will allow), then each child cgroup will get
1177	the part of parent's protection proportional to its
1178	actual memory usage below memory.min.
1179
1180	Putting more memory than generally available under this
1181	protection is discouraged and may lead to constant OOMs.
1182
1183	If a memory cgroup is not populated with processes,
1184	its memory.min is ignored.
1185
1186  memory.low
1187	A read-write single value file which exists on non-root
1188	cgroups.  The default is "0".
1189
1190	Best-effort memory protection.  If the memory usage of a
1191	cgroup is within its effective low boundary, the cgroup's
1192	memory won't be reclaimed unless there is no reclaimable
1193	memory available in unprotected cgroups.
1194	Above the effective low	boundary (or
1195	effective min boundary if it is higher), pages are reclaimed
1196	proportionally to the overage, reducing reclaim pressure for
1197	smaller overages.
1198
1199	Effective low boundary is limited by memory.low values of
1200	all ancestor cgroups. If there is memory.low overcommitment
1201	(child cgroup or cgroups are requiring more protected memory
1202	than parent will allow), then each child cgroup will get
1203	the part of parent's protection proportional to its
1204	actual memory usage below memory.low.
1205
1206	Putting more memory than generally available under this
1207	protection is discouraged.
1208
1209  memory.high
1210	A read-write single value file which exists on non-root
1211	cgroups.  The default is "max".
1212
1213	Memory usage throttle limit.  This is the main mechanism to
1214	control memory usage of a cgroup.  If a cgroup's usage goes
1215	over the high boundary, the processes of the cgroup are
1216	throttled and put under heavy reclaim pressure.
1217
1218	Going over the high limit never invokes the OOM killer and
1219	under extreme conditions the limit may be breached.
1220
1221  memory.max
1222	A read-write single value file which exists on non-root
1223	cgroups.  The default is "max".
1224
1225	Memory usage hard limit.  This is the final protection
1226	mechanism.  If a cgroup's memory usage reaches this limit and
1227	can't be reduced, the OOM killer is invoked in the cgroup.
1228	Under certain circumstances, the usage may go over the limit
1229	temporarily.
1230
1231	In default configuration regular 0-order allocations always
1232	succeed unless OOM killer chooses current task as a victim.
1233
1234	Some kinds of allocations don't invoke the OOM killer.
1235	Caller could retry them differently, return into userspace
1236	as -ENOMEM or silently ignore in cases like disk readahead.
1237
1238	This is the ultimate protection mechanism.  As long as the
1239	high limit is used and monitored properly, this limit's
1240	utility is limited to providing the final safety net.
1241
1242  memory.reclaim
1243	A write-only nested-keyed file which exists for all cgroups.
1244
1245	This is a simple interface to trigger memory reclaim in the
1246	target cgroup.
1247
1248	This file accepts a single key, the number of bytes to reclaim.
1249	No nested keys are currently supported.
1250
1251	Example::
1252
1253	  echo "1G" > memory.reclaim
1254
1255	The interface can be later extended with nested keys to
1256	configure the reclaim behavior. For example, specify the
1257	type of memory to reclaim from (anon, file, ..).
1258
1259	Please note that the kernel can over or under reclaim from
1260	the target cgroup. If less bytes are reclaimed than the
1261	specified amount, -EAGAIN is returned.
1262
1263	Please note that the proactive reclaim (triggered by this
1264	interface) is not meant to indicate memory pressure on the
1265	memory cgroup. Therefore socket memory balancing triggered by
1266	the memory reclaim normally is not exercised in this case.
1267	This means that the networking layer will not adapt based on
1268	reclaim induced by memory.reclaim.
1269
1270  memory.peak
1271	A read-only single value file which exists on non-root
1272	cgroups.
1273
1274	The max memory usage recorded for the cgroup and its
1275	descendants since the creation of the cgroup.
1276
1277  memory.oom.group
1278	A read-write single value file which exists on non-root
1279	cgroups.  The default value is "0".
1280
1281	Determines whether the cgroup should be treated as
1282	an indivisible workload by the OOM killer. If set,
1283	all tasks belonging to the cgroup or to its descendants
1284	(if the memory cgroup is not a leaf cgroup) are killed
1285	together or not at all. This can be used to avoid
1286	partial kills to guarantee workload integrity.
1287
1288	Tasks with the OOM protection (oom_score_adj set to -1000)
1289	are treated as an exception and are never killed.
1290
1291	If the OOM killer is invoked in a cgroup, it's not going
1292	to kill any tasks outside of this cgroup, regardless
1293	memory.oom.group values of ancestor cgroups.
1294
1295  memory.events
1296	A read-only flat-keyed file which exists on non-root cgroups.
1297	The following entries are defined.  Unless specified
1298	otherwise, a value change in this file generates a file
1299	modified event.
1300
1301	Note that all fields in this file are hierarchical and the
1302	file modified event can be generated due to an event down the
1303	hierarchy. For the local events at the cgroup level see
1304	memory.events.local.
1305
1306	  low
1307		The number of times the cgroup is reclaimed due to
1308		high memory pressure even though its usage is under
1309		the low boundary.  This usually indicates that the low
1310		boundary is over-committed.
1311
1312	  high
1313		The number of times processes of the cgroup are
1314		throttled and routed to perform direct memory reclaim
1315		because the high memory boundary was exceeded.  For a
1316		cgroup whose memory usage is capped by the high limit
1317		rather than global memory pressure, this event's
1318		occurrences are expected.
1319
1320	  max
1321		The number of times the cgroup's memory usage was
1322		about to go over the max boundary.  If direct reclaim
1323		fails to bring it down, the cgroup goes to OOM state.
1324
1325	  oom
1326		The number of time the cgroup's memory usage was
1327		reached the limit and allocation was about to fail.
1328
1329		This event is not raised if the OOM killer is not
1330		considered as an option, e.g. for failed high-order
1331		allocations or if caller asked to not retry attempts.
1332
1333	  oom_kill
1334		The number of processes belonging to this cgroup
1335		killed by any kind of OOM killer.
1336
1337          oom_group_kill
1338                The number of times a group OOM has occurred.
1339
1340  memory.events.local
1341	Similar to memory.events but the fields in the file are local
1342	to the cgroup i.e. not hierarchical. The file modified event
1343	generated on this file reflects only the local events.
1344
1345  memory.stat
1346	A read-only flat-keyed file which exists on non-root cgroups.
1347
1348	This breaks down the cgroup's memory footprint into different
1349	types of memory, type-specific details, and other information
1350	on the state and past events of the memory management system.
1351
1352	All memory amounts are in bytes.
1353
1354	The entries are ordered to be human readable, and new entries
1355	can show up in the middle. Don't rely on items remaining in a
1356	fixed position; use the keys to look up specific values!
1357
1358	If the entry has no per-node counter (or not show in the
1359	memory.numa_stat). We use 'npn' (non-per-node) as the tag
1360	to indicate that it will not show in the memory.numa_stat.
1361
1362	  anon
1363		Amount of memory used in anonymous mappings such as
1364		brk(), sbrk(), and mmap(MAP_ANONYMOUS)
1365
1366	  file
1367		Amount of memory used to cache filesystem data,
1368		including tmpfs and shared memory.
1369
1370	  kernel (npn)
1371		Amount of total kernel memory, including
1372		(kernel_stack, pagetables, percpu, vmalloc, slab) in
1373		addition to other kernel memory use cases.
1374
1375	  kernel_stack
1376		Amount of memory allocated to kernel stacks.
1377
1378	  pagetables
1379                Amount of memory allocated for page tables.
1380
1381	  sec_pagetables
1382		Amount of memory allocated for secondary page tables,
1383		this currently includes KVM mmu allocations on x86
1384		and arm64.
1385
1386	  percpu (npn)
1387		Amount of memory used for storing per-cpu kernel
1388		data structures.
1389
1390	  sock (npn)
1391		Amount of memory used in network transmission buffers
1392
1393	  vmalloc (npn)
1394		Amount of memory used for vmap backed memory.
1395
1396	  shmem
1397		Amount of cached filesystem data that is swap-backed,
1398		such as tmpfs, shm segments, shared anonymous mmap()s
1399
1400	  zswap
1401		Amount of memory consumed by the zswap compression backend.
1402
1403	  zswapped
1404		Amount of application memory swapped out to zswap.
1405
1406	  file_mapped
1407		Amount of cached filesystem data mapped with mmap()
1408
1409	  file_dirty
1410		Amount of cached filesystem data that was modified but
1411		not yet written back to disk
1412
1413	  file_writeback
1414		Amount of cached filesystem data that was modified and
1415		is currently being written back to disk
1416
1417	  swapcached
1418		Amount of swap cached in memory. The swapcache is accounted
1419		against both memory and swap usage.
1420
1421	  anon_thp
1422		Amount of memory used in anonymous mappings backed by
1423		transparent hugepages
1424
1425	  file_thp
1426		Amount of cached filesystem data backed by transparent
1427		hugepages
1428
1429	  shmem_thp
1430		Amount of shm, tmpfs, shared anonymous mmap()s backed by
1431		transparent hugepages
1432
1433	  inactive_anon, active_anon, inactive_file, active_file, unevictable
1434		Amount of memory, swap-backed and filesystem-backed,
1435		on the internal memory management lists used by the
1436		page reclaim algorithm.
1437
1438		As these represent internal list state (eg. shmem pages are on anon
1439		memory management lists), inactive_foo + active_foo may not be equal to
1440		the value for the foo counter, since the foo counter is type-based, not
1441		list-based.
1442
1443	  slab_reclaimable
1444		Part of "slab" that might be reclaimed, such as
1445		dentries and inodes.
1446
1447	  slab_unreclaimable
1448		Part of "slab" that cannot be reclaimed on memory
1449		pressure.
1450
1451	  slab (npn)
1452		Amount of memory used for storing in-kernel data
1453		structures.
1454
1455	  workingset_refault_anon
1456		Number of refaults of previously evicted anonymous pages.
1457
1458	  workingset_refault_file
1459		Number of refaults of previously evicted file pages.
1460
1461	  workingset_activate_anon
1462		Number of refaulted anonymous pages that were immediately
1463		activated.
1464
1465	  workingset_activate_file
1466		Number of refaulted file pages that were immediately activated.
1467
1468	  workingset_restore_anon
1469		Number of restored anonymous pages which have been detected as
1470		an active workingset before they got reclaimed.
1471
1472	  workingset_restore_file
1473		Number of restored file pages which have been detected as an
1474		active workingset before they got reclaimed.
1475
1476	  workingset_nodereclaim
1477		Number of times a shadow node has been reclaimed
1478
1479	  pgscan (npn)
1480		Amount of scanned pages (in an inactive LRU list)
1481
1482	  pgsteal (npn)
1483		Amount of reclaimed pages
1484
1485	  pgscan_kswapd (npn)
1486		Amount of scanned pages by kswapd (in an inactive LRU list)
1487
1488	  pgscan_direct (npn)
1489		Amount of scanned pages directly  (in an inactive LRU list)
1490
1491	  pgscan_khugepaged (npn)
1492		Amount of scanned pages by khugepaged  (in an inactive LRU list)
1493
1494	  pgsteal_kswapd (npn)
1495		Amount of reclaimed pages by kswapd
1496
1497	  pgsteal_direct (npn)
1498		Amount of reclaimed pages directly
1499
1500	  pgsteal_khugepaged (npn)
1501		Amount of reclaimed pages by khugepaged
1502
1503	  pgfault (npn)
1504		Total number of page faults incurred
1505
1506	  pgmajfault (npn)
1507		Number of major page faults incurred
1508
1509	  pgrefill (npn)
1510		Amount of scanned pages (in an active LRU list)
1511
1512	  pgactivate (npn)
1513		Amount of pages moved to the active LRU list
1514
1515	  pgdeactivate (npn)
1516		Amount of pages moved to the inactive LRU list
1517
1518	  pglazyfree (npn)
1519		Amount of pages postponed to be freed under memory pressure
1520
1521	  pglazyfreed (npn)
1522		Amount of reclaimed lazyfree pages
1523
1524	  thp_fault_alloc (npn)
1525		Number of transparent hugepages which were allocated to satisfy
1526		a page fault. This counter is not present when CONFIG_TRANSPARENT_HUGEPAGE
1527                is not set.
1528
1529	  thp_collapse_alloc (npn)
1530		Number of transparent hugepages which were allocated to allow
1531		collapsing an existing range of pages. This counter is not
1532		present when CONFIG_TRANSPARENT_HUGEPAGE is not set.
1533
1534  memory.numa_stat
1535	A read-only nested-keyed file which exists on non-root cgroups.
1536
1537	This breaks down the cgroup's memory footprint into different
1538	types of memory, type-specific details, and other information
1539	per node on the state of the memory management system.
1540
1541	This is useful for providing visibility into the NUMA locality
1542	information within an memcg since the pages are allowed to be
1543	allocated from any physical node. One of the use case is evaluating
1544	application performance by combining this information with the
1545	application's CPU allocation.
1546
1547	All memory amounts are in bytes.
1548
1549	The output format of memory.numa_stat is::
1550
1551	  type N0=<bytes in node 0> N1=<bytes in node 1> ...
1552
1553	The entries are ordered to be human readable, and new entries
1554	can show up in the middle. Don't rely on items remaining in a
1555	fixed position; use the keys to look up specific values!
1556
1557	The entries can refer to the memory.stat.
1558
1559  memory.swap.current
1560	A read-only single value file which exists on non-root
1561	cgroups.
1562
1563	The total amount of swap currently being used by the cgroup
1564	and its descendants.
1565
1566  memory.swap.high
1567	A read-write single value file which exists on non-root
1568	cgroups.  The default is "max".
1569
1570	Swap usage throttle limit.  If a cgroup's swap usage exceeds
1571	this limit, all its further allocations will be throttled to
1572	allow userspace to implement custom out-of-memory procedures.
1573
1574	This limit marks a point of no return for the cgroup. It is NOT
1575	designed to manage the amount of swapping a workload does
1576	during regular operation. Compare to memory.swap.max, which
1577	prohibits swapping past a set amount, but lets the cgroup
1578	continue unimpeded as long as other memory can be reclaimed.
1579
1580	Healthy workloads are not expected to reach this limit.
1581
1582  memory.swap.max
1583	A read-write single value file which exists on non-root
1584	cgroups.  The default is "max".
1585
1586	Swap usage hard limit.  If a cgroup's swap usage reaches this
1587	limit, anonymous memory of the cgroup will not be swapped out.
1588
1589  memory.swap.events
1590	A read-only flat-keyed file which exists on non-root cgroups.
1591	The following entries are defined.  Unless specified
1592	otherwise, a value change in this file generates a file
1593	modified event.
1594
1595	  high
1596		The number of times the cgroup's swap usage was over
1597		the high threshold.
1598
1599	  max
1600		The number of times the cgroup's swap usage was about
1601		to go over the max boundary and swap allocation
1602		failed.
1603
1604	  fail
1605		The number of times swap allocation failed either
1606		because of running out of swap system-wide or max
1607		limit.
1608
1609	When reduced under the current usage, the existing swap
1610	entries are reclaimed gradually and the swap usage may stay
1611	higher than the limit for an extended period of time.  This
1612	reduces the impact on the workload and memory management.
1613
1614  memory.zswap.current
1615	A read-only single value file which exists on non-root
1616	cgroups.
1617
1618	The total amount of memory consumed by the zswap compression
1619	backend.
1620
1621  memory.zswap.max
1622	A read-write single value file which exists on non-root
1623	cgroups.  The default is "max".
1624
1625	Zswap usage hard limit. If a cgroup's zswap pool reaches this
1626	limit, it will refuse to take any more stores before existing
1627	entries fault back in or are written out to disk.
1628
1629  memory.pressure
1630	A read-only nested-keyed file.
1631
1632	Shows pressure stall information for memory. See
1633	:ref:`Documentation/accounting/psi.rst <psi>` for details.
1634
1635
1636Usage Guidelines
1637~~~~~~~~~~~~~~~~
1638
1639"memory.high" is the main mechanism to control memory usage.
1640Over-committing on high limit (sum of high limits > available memory)
1641and letting global memory pressure to distribute memory according to
1642usage is a viable strategy.
1643
1644Because breach of the high limit doesn't trigger the OOM killer but
1645throttles the offending cgroup, a management agent has ample
1646opportunities to monitor and take appropriate actions such as granting
1647more memory or terminating the workload.
1648
1649Determining whether a cgroup has enough memory is not trivial as
1650memory usage doesn't indicate whether the workload can benefit from
1651more memory.  For example, a workload which writes data received from
1652network to a file can use all available memory but can also operate as
1653performant with a small amount of memory.  A measure of memory
1654pressure - how much the workload is being impacted due to lack of
1655memory - is necessary to determine whether a workload needs more
1656memory; unfortunately, memory pressure monitoring mechanism isn't
1657implemented yet.
1658
1659
1660Memory Ownership
1661~~~~~~~~~~~~~~~~
1662
1663A memory area is charged to the cgroup which instantiated it and stays
1664charged to the cgroup until the area is released.  Migrating a process
1665to a different cgroup doesn't move the memory usages that it
1666instantiated while in the previous cgroup to the new cgroup.
1667
1668A memory area may be used by processes belonging to different cgroups.
1669To which cgroup the area will be charged is in-deterministic; however,
1670over time, the memory area is likely to end up in a cgroup which has
1671enough memory allowance to avoid high reclaim pressure.
1672
1673If a cgroup sweeps a considerable amount of memory which is expected
1674to be accessed repeatedly by other cgroups, it may make sense to use
1675POSIX_FADV_DONTNEED to relinquish the ownership of memory areas
1676belonging to the affected files to ensure correct memory ownership.
1677
1678
1679IO
1680--
1681
1682The "io" controller regulates the distribution of IO resources.  This
1683controller implements both weight based and absolute bandwidth or IOPS
1684limit distribution; however, weight based distribution is available
1685only if cfq-iosched is in use and neither scheme is available for
1686blk-mq devices.
1687
1688
1689IO Interface Files
1690~~~~~~~~~~~~~~~~~~
1691
1692  io.stat
1693	A read-only nested-keyed file.
1694
1695	Lines are keyed by $MAJ:$MIN device numbers and not ordered.
1696	The following nested keys are defined.
1697
1698	  ======	=====================
1699	  rbytes	Bytes read
1700	  wbytes	Bytes written
1701	  rios		Number of read IOs
1702	  wios		Number of write IOs
1703	  dbytes	Bytes discarded
1704	  dios		Number of discard IOs
1705	  ======	=====================
1706
1707	An example read output follows::
1708
1709	  8:16 rbytes=1459200 wbytes=314773504 rios=192 wios=353 dbytes=0 dios=0
1710	  8:0 rbytes=90430464 wbytes=299008000 rios=8950 wios=1252 dbytes=50331648 dios=3021
1711
1712  io.cost.qos
1713	A read-write nested-keyed file which exists only on the root
1714	cgroup.
1715
1716	This file configures the Quality of Service of the IO cost
1717	model based controller (CONFIG_BLK_CGROUP_IOCOST) which
1718	currently implements "io.weight" proportional control.  Lines
1719	are keyed by $MAJ:$MIN device numbers and not ordered.  The
1720	line for a given device is populated on the first write for
1721	the device on "io.cost.qos" or "io.cost.model".  The following
1722	nested keys are defined.
1723
1724	  ======	=====================================
1725	  enable	Weight-based control enable
1726	  ctrl		"auto" or "user"
1727	  rpct		Read latency percentile    [0, 100]
1728	  rlat		Read latency threshold
1729	  wpct		Write latency percentile   [0, 100]
1730	  wlat		Write latency threshold
1731	  min		Minimum scaling percentage [1, 10000]
1732	  max		Maximum scaling percentage [1, 10000]
1733	  ======	=====================================
1734
1735	The controller is disabled by default and can be enabled by
1736	setting "enable" to 1.  "rpct" and "wpct" parameters default
1737	to zero and the controller uses internal device saturation
1738	state to adjust the overall IO rate between "min" and "max".
1739
1740	When a better control quality is needed, latency QoS
1741	parameters can be configured.  For example::
1742
1743	  8:16 enable=1 ctrl=auto rpct=95.00 rlat=75000 wpct=95.00 wlat=150000 min=50.00 max=150.0
1744
1745	shows that on sdb, the controller is enabled, will consider
1746	the device saturated if the 95th percentile of read completion
1747	latencies is above 75ms or write 150ms, and adjust the overall
1748	IO issue rate between 50% and 150% accordingly.
1749
1750	The lower the saturation point, the better the latency QoS at
1751	the cost of aggregate bandwidth.  The narrower the allowed
1752	adjustment range between "min" and "max", the more conformant
1753	to the cost model the IO behavior.  Note that the IO issue
1754	base rate may be far off from 100% and setting "min" and "max"
1755	blindly can lead to a significant loss of device capacity or
1756	control quality.  "min" and "max" are useful for regulating
1757	devices which show wide temporary behavior changes - e.g. a
1758	ssd which accepts writes at the line speed for a while and
1759	then completely stalls for multiple seconds.
1760
1761	When "ctrl" is "auto", the parameters are controlled by the
1762	kernel and may change automatically.  Setting "ctrl" to "user"
1763	or setting any of the percentile and latency parameters puts
1764	it into "user" mode and disables the automatic changes.  The
1765	automatic mode can be restored by setting "ctrl" to "auto".
1766
1767  io.cost.model
1768	A read-write nested-keyed file which exists only on the root
1769	cgroup.
1770
1771	This file configures the cost model of the IO cost model based
1772	controller (CONFIG_BLK_CGROUP_IOCOST) which currently
1773	implements "io.weight" proportional control.  Lines are keyed
1774	by $MAJ:$MIN device numbers and not ordered.  The line for a
1775	given device is populated on the first write for the device on
1776	"io.cost.qos" or "io.cost.model".  The following nested keys
1777	are defined.
1778
1779	  =====		================================
1780	  ctrl		"auto" or "user"
1781	  model		The cost model in use - "linear"
1782	  =====		================================
1783
1784	When "ctrl" is "auto", the kernel may change all parameters
1785	dynamically.  When "ctrl" is set to "user" or any other
1786	parameters are written to, "ctrl" become "user" and the
1787	automatic changes are disabled.
1788
1789	When "model" is "linear", the following model parameters are
1790	defined.
1791
1792	  =============	========================================
1793	  [r|w]bps	The maximum sequential IO throughput
1794	  [r|w]seqiops	The maximum 4k sequential IOs per second
1795	  [r|w]randiops	The maximum 4k random IOs per second
1796	  =============	========================================
1797
1798	From the above, the builtin linear model determines the base
1799	costs of a sequential and random IO and the cost coefficient
1800	for the IO size.  While simple, this model can cover most
1801	common device classes acceptably.
1802
1803	The IO cost model isn't expected to be accurate in absolute
1804	sense and is scaled to the device behavior dynamically.
1805
1806	If needed, tools/cgroup/iocost_coef_gen.py can be used to
1807	generate device-specific coefficients.
1808
1809  io.weight
1810	A read-write flat-keyed file which exists on non-root cgroups.
1811	The default is "default 100".
1812
1813	The first line is the default weight applied to devices
1814	without specific override.  The rest are overrides keyed by
1815	$MAJ:$MIN device numbers and not ordered.  The weights are in
1816	the range [1, 10000] and specifies the relative amount IO time
1817	the cgroup can use in relation to its siblings.
1818
1819	The default weight can be updated by writing either "default
1820	$WEIGHT" or simply "$WEIGHT".  Overrides can be set by writing
1821	"$MAJ:$MIN $WEIGHT" and unset by writing "$MAJ:$MIN default".
1822
1823	An example read output follows::
1824
1825	  default 100
1826	  8:16 200
1827	  8:0 50
1828
1829  io.max
1830	A read-write nested-keyed file which exists on non-root
1831	cgroups.
1832
1833	BPS and IOPS based IO limit.  Lines are keyed by $MAJ:$MIN
1834	device numbers and not ordered.  The following nested keys are
1835	defined.
1836
1837	  =====		==================================
1838	  rbps		Max read bytes per second
1839	  wbps		Max write bytes per second
1840	  riops		Max read IO operations per second
1841	  wiops		Max write IO operations per second
1842	  =====		==================================
1843
1844	When writing, any number of nested key-value pairs can be
1845	specified in any order.  "max" can be specified as the value
1846	to remove a specific limit.  If the same key is specified
1847	multiple times, the outcome is undefined.
1848
1849	BPS and IOPS are measured in each IO direction and IOs are
1850	delayed if limit is reached.  Temporary bursts are allowed.
1851
1852	Setting read limit at 2M BPS and write at 120 IOPS for 8:16::
1853
1854	  echo "8:16 rbps=2097152 wiops=120" > io.max
1855
1856	Reading returns the following::
1857
1858	  8:16 rbps=2097152 wbps=max riops=max wiops=120
1859
1860	Write IOPS limit can be removed by writing the following::
1861
1862	  echo "8:16 wiops=max" > io.max
1863
1864	Reading now returns the following::
1865
1866	  8:16 rbps=2097152 wbps=max riops=max wiops=max
1867
1868  io.pressure
1869	A read-only nested-keyed file.
1870
1871	Shows pressure stall information for IO. See
1872	:ref:`Documentation/accounting/psi.rst <psi>` for details.
1873
1874
1875Writeback
1876~~~~~~~~~
1877
1878Page cache is dirtied through buffered writes and shared mmaps and
1879written asynchronously to the backing filesystem by the writeback
1880mechanism.  Writeback sits between the memory and IO domains and
1881regulates the proportion of dirty memory by balancing dirtying and
1882write IOs.
1883
1884The io controller, in conjunction with the memory controller,
1885implements control of page cache writeback IOs.  The memory controller
1886defines the memory domain that dirty memory ratio is calculated and
1887maintained for and the io controller defines the io domain which
1888writes out dirty pages for the memory domain.  Both system-wide and
1889per-cgroup dirty memory states are examined and the more restrictive
1890of the two is enforced.
1891
1892cgroup writeback requires explicit support from the underlying
1893filesystem.  Currently, cgroup writeback is implemented on ext2, ext4,
1894btrfs, f2fs, and xfs.  On other filesystems, all writeback IOs are
1895attributed to the root cgroup.
1896
1897There are inherent differences in memory and writeback management
1898which affects how cgroup ownership is tracked.  Memory is tracked per
1899page while writeback per inode.  For the purpose of writeback, an
1900inode is assigned to a cgroup and all IO requests to write dirty pages
1901from the inode are attributed to that cgroup.
1902
1903As cgroup ownership for memory is tracked per page, there can be pages
1904which are associated with different cgroups than the one the inode is
1905associated with.  These are called foreign pages.  The writeback
1906constantly keeps track of foreign pages and, if a particular foreign
1907cgroup becomes the majority over a certain period of time, switches
1908the ownership of the inode to that cgroup.
1909
1910While this model is enough for most use cases where a given inode is
1911mostly dirtied by a single cgroup even when the main writing cgroup
1912changes over time, use cases where multiple cgroups write to a single
1913inode simultaneously are not supported well.  In such circumstances, a
1914significant portion of IOs are likely to be attributed incorrectly.
1915As memory controller assigns page ownership on the first use and
1916doesn't update it until the page is released, even if writeback
1917strictly follows page ownership, multiple cgroups dirtying overlapping
1918areas wouldn't work as expected.  It's recommended to avoid such usage
1919patterns.
1920
1921The sysctl knobs which affect writeback behavior are applied to cgroup
1922writeback as follows.
1923
1924  vm.dirty_background_ratio, vm.dirty_ratio
1925	These ratios apply the same to cgroup writeback with the
1926	amount of available memory capped by limits imposed by the
1927	memory controller and system-wide clean memory.
1928
1929  vm.dirty_background_bytes, vm.dirty_bytes
1930	For cgroup writeback, this is calculated into ratio against
1931	total available memory and applied the same way as
1932	vm.dirty[_background]_ratio.
1933
1934
1935IO Latency
1936~~~~~~~~~~
1937
1938This is a cgroup v2 controller for IO workload protection.  You provide a group
1939with a latency target, and if the average latency exceeds that target the
1940controller will throttle any peers that have a lower latency target than the
1941protected workload.
1942
1943The limits are only applied at the peer level in the hierarchy.  This means that
1944in the diagram below, only groups A, B, and C will influence each other, and
1945groups D and F will influence each other.  Group G will influence nobody::
1946
1947			[root]
1948		/	   |		\
1949		A	   B		C
1950	       /  \        |
1951	      D    F	   G
1952
1953
1954So the ideal way to configure this is to set io.latency in groups A, B, and C.
1955Generally you do not want to set a value lower than the latency your device
1956supports.  Experiment to find the value that works best for your workload.
1957Start at higher than the expected latency for your device and watch the
1958avg_lat value in io.stat for your workload group to get an idea of the
1959latency you see during normal operation.  Use the avg_lat value as a basis for
1960your real setting, setting at 10-15% higher than the value in io.stat.
1961
1962How IO Latency Throttling Works
1963~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
1964
1965io.latency is work conserving; so as long as everybody is meeting their latency
1966target the controller doesn't do anything.  Once a group starts missing its
1967target it begins throttling any peer group that has a higher target than itself.
1968This throttling takes 2 forms:
1969
1970- Queue depth throttling.  This is the number of outstanding IO's a group is
1971  allowed to have.  We will clamp down relatively quickly, starting at no limit
1972  and going all the way down to 1 IO at a time.
1973
1974- Artificial delay induction.  There are certain types of IO that cannot be
1975  throttled without possibly adversely affecting higher priority groups.  This
1976  includes swapping and metadata IO.  These types of IO are allowed to occur
1977  normally, however they are "charged" to the originating group.  If the
1978  originating group is being throttled you will see the use_delay and delay
1979  fields in io.stat increase.  The delay value is how many microseconds that are
1980  being added to any process that runs in this group.  Because this number can
1981  grow quite large if there is a lot of swapping or metadata IO occurring we
1982  limit the individual delay events to 1 second at a time.
1983
1984Once the victimized group starts meeting its latency target again it will start
1985unthrottling any peer groups that were throttled previously.  If the victimized
1986group simply stops doing IO the global counter will unthrottle appropriately.
1987
1988IO Latency Interface Files
1989~~~~~~~~~~~~~~~~~~~~~~~~~~
1990
1991  io.latency
1992	This takes a similar format as the other controllers.
1993
1994		"MAJOR:MINOR target=<target time in microseconds>"
1995
1996  io.stat
1997	If the controller is enabled you will see extra stats in io.stat in
1998	addition to the normal ones.
1999
2000	  depth
2001		This is the current queue depth for the group.
2002
2003	  avg_lat
2004		This is an exponential moving average with a decay rate of 1/exp
2005		bound by the sampling interval.  The decay rate interval can be
2006		calculated by multiplying the win value in io.stat by the
2007		corresponding number of samples based on the win value.
2008
2009	  win
2010		The sampling window size in milliseconds.  This is the minimum
2011		duration of time between evaluation events.  Windows only elapse
2012		with IO activity.  Idle periods extend the most recent window.
2013
2014IO Priority
2015~~~~~~~~~~~
2016
2017A single attribute controls the behavior of the I/O priority cgroup policy,
2018namely the blkio.prio.class attribute. The following values are accepted for
2019that attribute:
2020
2021  no-change
2022	Do not modify the I/O priority class.
2023
2024  none-to-rt
2025	For requests that do not have an I/O priority class (NONE),
2026	change the I/O priority class into RT. Do not modify
2027	the I/O priority class of other requests.
2028
2029  restrict-to-be
2030	For requests that do not have an I/O priority class or that have I/O
2031	priority class RT, change it into BE. Do not modify the I/O priority
2032	class of requests that have priority class IDLE.
2033
2034  idle
2035	Change the I/O priority class of all requests into IDLE, the lowest
2036	I/O priority class.
2037
2038The following numerical values are associated with the I/O priority policies:
2039
2040+-------------+---+
2041| no-change   | 0 |
2042+-------------+---+
2043| none-to-rt  | 1 |
2044+-------------+---+
2045| rt-to-be    | 2 |
2046+-------------+---+
2047| all-to-idle | 3 |
2048+-------------+---+
2049
2050The numerical value that corresponds to each I/O priority class is as follows:
2051
2052+-------------------------------+---+
2053| IOPRIO_CLASS_NONE             | 0 |
2054+-------------------------------+---+
2055| IOPRIO_CLASS_RT (real-time)   | 1 |
2056+-------------------------------+---+
2057| IOPRIO_CLASS_BE (best effort) | 2 |
2058+-------------------------------+---+
2059| IOPRIO_CLASS_IDLE             | 3 |
2060+-------------------------------+---+
2061
2062The algorithm to set the I/O priority class for a request is as follows:
2063
2064- Translate the I/O priority class policy into a number.
2065- Change the request I/O priority class into the maximum of the I/O priority
2066  class policy number and the numerical I/O priority class.
2067
2068PID
2069---
2070
2071The process number controller is used to allow a cgroup to stop any
2072new tasks from being fork()'d or clone()'d after a specified limit is
2073reached.
2074
2075The number of tasks in a cgroup can be exhausted in ways which other
2076controllers cannot prevent, thus warranting its own controller.  For
2077example, a fork bomb is likely to exhaust the number of tasks before
2078hitting memory restrictions.
2079
2080Note that PIDs used in this controller refer to TIDs, process IDs as
2081used by the kernel.
2082
2083
2084PID Interface Files
2085~~~~~~~~~~~~~~~~~~~
2086
2087  pids.max
2088	A read-write single value file which exists on non-root
2089	cgroups.  The default is "max".
2090
2091	Hard limit of number of processes.
2092
2093  pids.current
2094	A read-only single value file which exists on all cgroups.
2095
2096	The number of processes currently in the cgroup and its
2097	descendants.
2098
2099Organisational operations are not blocked by cgroup policies, so it is
2100possible to have pids.current > pids.max.  This can be done by either
2101setting the limit to be smaller than pids.current, or attaching enough
2102processes to the cgroup such that pids.current is larger than
2103pids.max.  However, it is not possible to violate a cgroup PID policy
2104through fork() or clone(). These will return -EAGAIN if the creation
2105of a new process would cause a cgroup policy to be violated.
2106
2107
2108Cpuset
2109------
2110
2111The "cpuset" controller provides a mechanism for constraining
2112the CPU and memory node placement of tasks to only the resources
2113specified in the cpuset interface files in a task's current cgroup.
2114This is especially valuable on large NUMA systems where placing jobs
2115on properly sized subsets of the systems with careful processor and
2116memory placement to reduce cross-node memory access and contention
2117can improve overall system performance.
2118
2119The "cpuset" controller is hierarchical.  That means the controller
2120cannot use CPUs or memory nodes not allowed in its parent.
2121
2122
2123Cpuset Interface Files
2124~~~~~~~~~~~~~~~~~~~~~~
2125
2126  cpuset.cpus
2127	A read-write multiple values file which exists on non-root
2128	cpuset-enabled cgroups.
2129
2130	It lists the requested CPUs to be used by tasks within this
2131	cgroup.  The actual list of CPUs to be granted, however, is
2132	subjected to constraints imposed by its parent and can differ
2133	from the requested CPUs.
2134
2135	The CPU numbers are comma-separated numbers or ranges.
2136	For example::
2137
2138	  # cat cpuset.cpus
2139	  0-4,6,8-10
2140
2141	An empty value indicates that the cgroup is using the same
2142	setting as the nearest cgroup ancestor with a non-empty
2143	"cpuset.cpus" or all the available CPUs if none is found.
2144
2145	The value of "cpuset.cpus" stays constant until the next update
2146	and won't be affected by any CPU hotplug events.
2147
2148  cpuset.cpus.effective
2149	A read-only multiple values file which exists on all
2150	cpuset-enabled cgroups.
2151
2152	It lists the onlined CPUs that are actually granted to this
2153	cgroup by its parent.  These CPUs are allowed to be used by
2154	tasks within the current cgroup.
2155
2156	If "cpuset.cpus" is empty, the "cpuset.cpus.effective" file shows
2157	all the CPUs from the parent cgroup that can be available to
2158	be used by this cgroup.  Otherwise, it should be a subset of
2159	"cpuset.cpus" unless none of the CPUs listed in "cpuset.cpus"
2160	can be granted.  In this case, it will be treated just like an
2161	empty "cpuset.cpus".
2162
2163	Its value will be affected by CPU hotplug events.
2164
2165  cpuset.mems
2166	A read-write multiple values file which exists on non-root
2167	cpuset-enabled cgroups.
2168
2169	It lists the requested memory nodes to be used by tasks within
2170	this cgroup.  The actual list of memory nodes granted, however,
2171	is subjected to constraints imposed by its parent and can differ
2172	from the requested memory nodes.
2173
2174	The memory node numbers are comma-separated numbers or ranges.
2175	For example::
2176
2177	  # cat cpuset.mems
2178	  0-1,3
2179
2180	An empty value indicates that the cgroup is using the same
2181	setting as the nearest cgroup ancestor with a non-empty
2182	"cpuset.mems" or all the available memory nodes if none
2183	is found.
2184
2185	The value of "cpuset.mems" stays constant until the next update
2186	and won't be affected by any memory nodes hotplug events.
2187
2188	Setting a non-empty value to "cpuset.mems" causes memory of
2189	tasks within the cgroup to be migrated to the designated nodes if
2190	they are currently using memory outside of the designated nodes.
2191
2192	There is a cost for this memory migration.  The migration
2193	may not be complete and some memory pages may be left behind.
2194	So it is recommended that "cpuset.mems" should be set properly
2195	before spawning new tasks into the cpuset.  Even if there is
2196	a need to change "cpuset.mems" with active tasks, it shouldn't
2197	be done frequently.
2198
2199  cpuset.mems.effective
2200	A read-only multiple values file which exists on all
2201	cpuset-enabled cgroups.
2202
2203	It lists the onlined memory nodes that are actually granted to
2204	this cgroup by its parent. These memory nodes are allowed to
2205	be used by tasks within the current cgroup.
2206
2207	If "cpuset.mems" is empty, it shows all the memory nodes from the
2208	parent cgroup that will be available to be used by this cgroup.
2209	Otherwise, it should be a subset of "cpuset.mems" unless none of
2210	the memory nodes listed in "cpuset.mems" can be granted.  In this
2211	case, it will be treated just like an empty "cpuset.mems".
2212
2213	Its value will be affected by memory nodes hotplug events.
2214
2215  cpuset.cpus.partition
2216	A read-write single value file which exists on non-root
2217	cpuset-enabled cgroups.  This flag is owned by the parent cgroup
2218	and is not delegatable.
2219
2220	It accepts only the following input values when written to.
2221
2222	  ==========	=====================================
2223	  "member"	Non-root member of a partition
2224	  "root"	Partition root
2225	  "isolated"	Partition root without load balancing
2226	  ==========	=====================================
2227
2228	The root cgroup is always a partition root and its state
2229	cannot be changed.  All other non-root cgroups start out as
2230	"member".
2231
2232	When set to "root", the current cgroup is the root of a new
2233	partition or scheduling domain that comprises itself and all
2234	its descendants except those that are separate partition roots
2235	themselves and their descendants.
2236
2237	When set to "isolated", the CPUs in that partition root will
2238	be in an isolated state without any load balancing from the
2239	scheduler.  Tasks placed in such a partition with multiple
2240	CPUs should be carefully distributed and bound to each of the
2241	individual CPUs for optimal performance.
2242
2243	The value shown in "cpuset.cpus.effective" of a partition root
2244	is the CPUs that the partition root can dedicate to a potential
2245	new child partition root. The new child subtracts available
2246	CPUs from its parent "cpuset.cpus.effective".
2247
2248	A partition root ("root" or "isolated") can be in one of the
2249	two possible states - valid or invalid.  An invalid partition
2250	root is in a degraded state where some state information may
2251	be retained, but behaves more like a "member".
2252
2253	All possible state transitions among "member", "root" and
2254	"isolated" are allowed.
2255
2256	On read, the "cpuset.cpus.partition" file can show the following
2257	values.
2258
2259	  =============================	=====================================
2260	  "member"			Non-root member of a partition
2261	  "root"			Partition root
2262	  "isolated"			Partition root without load balancing
2263	  "root invalid (<reason>)"	Invalid partition root
2264	  "isolated invalid (<reason>)"	Invalid isolated partition root
2265	  =============================	=====================================
2266
2267	In the case of an invalid partition root, a descriptive string on
2268	why the partition is invalid is included within parentheses.
2269
2270	For a partition root to become valid, the following conditions
2271	must be met.
2272
2273	1) The "cpuset.cpus" is exclusive with its siblings , i.e. they
2274	   are not shared by any of its siblings (exclusivity rule).
2275	2) The parent cgroup is a valid partition root.
2276	3) The "cpuset.cpus" is not empty and must contain at least
2277	   one of the CPUs from parent's "cpuset.cpus", i.e. they overlap.
2278	4) The "cpuset.cpus.effective" cannot be empty unless there is
2279	   no task associated with this partition.
2280
2281	External events like hotplug or changes to "cpuset.cpus" can
2282	cause a valid partition root to become invalid and vice versa.
2283	Note that a task cannot be moved to a cgroup with empty
2284	"cpuset.cpus.effective".
2285
2286	For a valid partition root with the sibling cpu exclusivity
2287	rule enabled, changes made to "cpuset.cpus" that violate the
2288	exclusivity rule will invalidate the partition as well as its
2289	sibiling partitions with conflicting cpuset.cpus values. So
2290	care must be taking in changing "cpuset.cpus".
2291
2292	A valid non-root parent partition may distribute out all its CPUs
2293	to its child partitions when there is no task associated with it.
2294
2295	Care must be taken to change a valid partition root to
2296	"member" as all its child partitions, if present, will become
2297	invalid causing disruption to tasks running in those child
2298	partitions. These inactivated partitions could be recovered if
2299	their parent is switched back to a partition root with a proper
2300	set of "cpuset.cpus".
2301
2302	Poll and inotify events are triggered whenever the state of
2303	"cpuset.cpus.partition" changes.  That includes changes caused
2304	by write to "cpuset.cpus.partition", cpu hotplug or other
2305	changes that modify the validity status of the partition.
2306	This will allow user space agents to monitor unexpected changes
2307	to "cpuset.cpus.partition" without the need to do continuous
2308	polling.
2309
2310
2311Device controller
2312-----------------
2313
2314Device controller manages access to device files. It includes both
2315creation of new device files (using mknod), and access to the
2316existing device files.
2317
2318Cgroup v2 device controller has no interface files and is implemented
2319on top of cgroup BPF. To control access to device files, a user may
2320create bpf programs of type BPF_PROG_TYPE_CGROUP_DEVICE and attach
2321them to cgroups with BPF_CGROUP_DEVICE flag. On an attempt to access a
2322device file, corresponding BPF programs will be executed, and depending
2323on the return value the attempt will succeed or fail with -EPERM.
2324
2325A BPF_PROG_TYPE_CGROUP_DEVICE program takes a pointer to the
2326bpf_cgroup_dev_ctx structure, which describes the device access attempt:
2327access type (mknod/read/write) and device (type, major and minor numbers).
2328If the program returns 0, the attempt fails with -EPERM, otherwise it
2329succeeds.
2330
2331An example of BPF_PROG_TYPE_CGROUP_DEVICE program may be found in
2332tools/testing/selftests/bpf/progs/dev_cgroup.c in the kernel source tree.
2333
2334
2335RDMA
2336----
2337
2338The "rdma" controller regulates the distribution and accounting of
2339RDMA resources.
2340
2341RDMA Interface Files
2342~~~~~~~~~~~~~~~~~~~~
2343
2344  rdma.max
2345	A readwrite nested-keyed file that exists for all the cgroups
2346	except root that describes current configured resource limit
2347	for a RDMA/IB device.
2348
2349	Lines are keyed by device name and are not ordered.
2350	Each line contains space separated resource name and its configured
2351	limit that can be distributed.
2352
2353	The following nested keys are defined.
2354
2355	  ==========	=============================
2356	  hca_handle	Maximum number of HCA Handles
2357	  hca_object 	Maximum number of HCA Objects
2358	  ==========	=============================
2359
2360	An example for mlx4 and ocrdma device follows::
2361
2362	  mlx4_0 hca_handle=2 hca_object=2000
2363	  ocrdma1 hca_handle=3 hca_object=max
2364
2365  rdma.current
2366	A read-only file that describes current resource usage.
2367	It exists for all the cgroup except root.
2368
2369	An example for mlx4 and ocrdma device follows::
2370
2371	  mlx4_0 hca_handle=1 hca_object=20
2372	  ocrdma1 hca_handle=1 hca_object=23
2373
2374HugeTLB
2375-------
2376
2377The HugeTLB controller allows to limit the HugeTLB usage per control group and
2378enforces the controller limit during page fault.
2379
2380HugeTLB Interface Files
2381~~~~~~~~~~~~~~~~~~~~~~~
2382
2383  hugetlb.<hugepagesize>.current
2384	Show current usage for "hugepagesize" hugetlb.  It exists for all
2385	the cgroup except root.
2386
2387  hugetlb.<hugepagesize>.max
2388	Set/show the hard limit of "hugepagesize" hugetlb usage.
2389	The default value is "max".  It exists for all the cgroup except root.
2390
2391  hugetlb.<hugepagesize>.events
2392	A read-only flat-keyed file which exists on non-root cgroups.
2393
2394	  max
2395		The number of allocation failure due to HugeTLB limit
2396
2397  hugetlb.<hugepagesize>.events.local
2398	Similar to hugetlb.<hugepagesize>.events but the fields in the file
2399	are local to the cgroup i.e. not hierarchical. The file modified event
2400	generated on this file reflects only the local events.
2401
2402  hugetlb.<hugepagesize>.numa_stat
2403	Similar to memory.numa_stat, it shows the numa information of the
2404        hugetlb pages of <hugepagesize> in this cgroup.  Only active in
2405        use hugetlb pages are included.  The per-node values are in bytes.
2406
2407Misc
2408----
2409
2410The Miscellaneous cgroup provides the resource limiting and tracking
2411mechanism for the scalar resources which cannot be abstracted like the other
2412cgroup resources. Controller is enabled by the CONFIG_CGROUP_MISC config
2413option.
2414
2415A resource can be added to the controller via enum misc_res_type{} in the
2416include/linux/misc_cgroup.h file and the corresponding name via misc_res_name[]
2417in the kernel/cgroup/misc.c file. Provider of the resource must set its
2418capacity prior to using the resource by calling misc_cg_set_capacity().
2419
2420Once a capacity is set then the resource usage can be updated using charge and
2421uncharge APIs. All of the APIs to interact with misc controller are in
2422include/linux/misc_cgroup.h.
2423
2424Misc Interface Files
2425~~~~~~~~~~~~~~~~~~~~
2426
2427Miscellaneous controller provides 3 interface files. If two misc resources (res_a and res_b) are registered then:
2428
2429  misc.capacity
2430        A read-only flat-keyed file shown only in the root cgroup.  It shows
2431        miscellaneous scalar resources available on the platform along with
2432        their quantities::
2433
2434	  $ cat misc.capacity
2435	  res_a 50
2436	  res_b 10
2437
2438  misc.current
2439        A read-only flat-keyed file shown in the non-root cgroups.  It shows
2440        the current usage of the resources in the cgroup and its children.::
2441
2442	  $ cat misc.current
2443	  res_a 3
2444	  res_b 0
2445
2446  misc.max
2447        A read-write flat-keyed file shown in the non root cgroups. Allowed
2448        maximum usage of the resources in the cgroup and its children.::
2449
2450	  $ cat misc.max
2451	  res_a max
2452	  res_b 4
2453
2454	Limit can be set by::
2455
2456	  # echo res_a 1 > misc.max
2457
2458	Limit can be set to max by::
2459
2460	  # echo res_a max > misc.max
2461
2462        Limits can be set higher than the capacity value in the misc.capacity
2463        file.
2464
2465  misc.events
2466	A read-only flat-keyed file which exists on non-root cgroups. The
2467	following entries are defined. Unless specified otherwise, a value
2468	change in this file generates a file modified event. All fields in
2469	this file are hierarchical.
2470
2471	  max
2472		The number of times the cgroup's resource usage was
2473		about to go over the max boundary.
2474
2475Migration and Ownership
2476~~~~~~~~~~~~~~~~~~~~~~~
2477
2478A miscellaneous scalar resource is charged to the cgroup in which it is used
2479first, and stays charged to that cgroup until that resource is freed. Migrating
2480a process to a different cgroup does not move the charge to the destination
2481cgroup where the process has moved.
2482
2483Others
2484------
2485
2486perf_event
2487~~~~~~~~~~
2488
2489perf_event controller, if not mounted on a legacy hierarchy, is
2490automatically enabled on the v2 hierarchy so that perf events can
2491always be filtered by cgroup v2 path.  The controller can still be
2492moved to a legacy hierarchy after v2 hierarchy is populated.
2493
2494
2495Non-normative information
2496-------------------------
2497
2498This section contains information that isn't considered to be a part of
2499the stable kernel API and so is subject to change.
2500
2501
2502CPU controller root cgroup process behaviour
2503~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2504
2505When distributing CPU cycles in the root cgroup each thread in this
2506cgroup is treated as if it was hosted in a separate child cgroup of the
2507root cgroup. This child cgroup weight is dependent on its thread nice
2508level.
2509
2510For details of this mapping see sched_prio_to_weight array in
2511kernel/sched/core.c file (values from this array should be scaled
2512appropriately so the neutral - nice 0 - value is 100 instead of 1024).
2513
2514
2515IO controller root cgroup process behaviour
2516~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2517
2518Root cgroup processes are hosted in an implicit leaf child node.
2519When distributing IO resources this implicit child node is taken into
2520account as if it was a normal child cgroup of the root cgroup with a
2521weight value of 200.
2522
2523
2524Namespace
2525=========
2526
2527Basics
2528------
2529
2530cgroup namespace provides a mechanism to virtualize the view of the
2531"/proc/$PID/cgroup" file and cgroup mounts.  The CLONE_NEWCGROUP clone
2532flag can be used with clone(2) and unshare(2) to create a new cgroup
2533namespace.  The process running inside the cgroup namespace will have
2534its "/proc/$PID/cgroup" output restricted to cgroupns root.  The
2535cgroupns root is the cgroup of the process at the time of creation of
2536the cgroup namespace.
2537
2538Without cgroup namespace, the "/proc/$PID/cgroup" file shows the
2539complete path of the cgroup of a process.  In a container setup where
2540a set of cgroups and namespaces are intended to isolate processes the
2541"/proc/$PID/cgroup" file may leak potential system level information
2542to the isolated processes.  For example::
2543
2544  # cat /proc/self/cgroup
2545  0::/batchjobs/container_id1
2546
2547The path '/batchjobs/container_id1' can be considered as system-data
2548and undesirable to expose to the isolated processes.  cgroup namespace
2549can be used to restrict visibility of this path.  For example, before
2550creating a cgroup namespace, one would see::
2551
2552  # ls -l /proc/self/ns/cgroup
2553  lrwxrwxrwx 1 root root 0 2014-07-15 10:37 /proc/self/ns/cgroup -> cgroup:[4026531835]
2554  # cat /proc/self/cgroup
2555  0::/batchjobs/container_id1
2556
2557After unsharing a new namespace, the view changes::
2558
2559  # ls -l /proc/self/ns/cgroup
2560  lrwxrwxrwx 1 root root 0 2014-07-15 10:35 /proc/self/ns/cgroup -> cgroup:[4026532183]
2561  # cat /proc/self/cgroup
2562  0::/
2563
2564When some thread from a multi-threaded process unshares its cgroup
2565namespace, the new cgroupns gets applied to the entire process (all
2566the threads).  This is natural for the v2 hierarchy; however, for the
2567legacy hierarchies, this may be unexpected.
2568
2569A cgroup namespace is alive as long as there are processes inside or
2570mounts pinning it.  When the last usage goes away, the cgroup
2571namespace is destroyed.  The cgroupns root and the actual cgroups
2572remain.
2573
2574
2575The Root and Views
2576------------------
2577
2578The 'cgroupns root' for a cgroup namespace is the cgroup in which the
2579process calling unshare(2) is running.  For example, if a process in
2580/batchjobs/container_id1 cgroup calls unshare, cgroup
2581/batchjobs/container_id1 becomes the cgroupns root.  For the
2582init_cgroup_ns, this is the real root ('/') cgroup.
2583
2584The cgroupns root cgroup does not change even if the namespace creator
2585process later moves to a different cgroup::
2586
2587  # ~/unshare -c # unshare cgroupns in some cgroup
2588  # cat /proc/self/cgroup
2589  0::/
2590  # mkdir sub_cgrp_1
2591  # echo 0 > sub_cgrp_1/cgroup.procs
2592  # cat /proc/self/cgroup
2593  0::/sub_cgrp_1
2594
2595Each process gets its namespace-specific view of "/proc/$PID/cgroup"
2596
2597Processes running inside the cgroup namespace will be able to see
2598cgroup paths (in /proc/self/cgroup) only inside their root cgroup.
2599From within an unshared cgroupns::
2600
2601  # sleep 100000 &
2602  [1] 7353
2603  # echo 7353 > sub_cgrp_1/cgroup.procs
2604  # cat /proc/7353/cgroup
2605  0::/sub_cgrp_1
2606
2607From the initial cgroup namespace, the real cgroup path will be
2608visible::
2609
2610  $ cat /proc/7353/cgroup
2611  0::/batchjobs/container_id1/sub_cgrp_1
2612
2613From a sibling cgroup namespace (that is, a namespace rooted at a
2614different cgroup), the cgroup path relative to its own cgroup
2615namespace root will be shown.  For instance, if PID 7353's cgroup
2616namespace root is at '/batchjobs/container_id2', then it will see::
2617
2618  # cat /proc/7353/cgroup
2619  0::/../container_id2/sub_cgrp_1
2620
2621Note that the relative path always starts with '/' to indicate that
2622its relative to the cgroup namespace root of the caller.
2623
2624
2625Migration and setns(2)
2626----------------------
2627
2628Processes inside a cgroup namespace can move into and out of the
2629namespace root if they have proper access to external cgroups.  For
2630example, from inside a namespace with cgroupns root at
2631/batchjobs/container_id1, and assuming that the global hierarchy is
2632still accessible inside cgroupns::
2633
2634  # cat /proc/7353/cgroup
2635  0::/sub_cgrp_1
2636  # echo 7353 > batchjobs/container_id2/cgroup.procs
2637  # cat /proc/7353/cgroup
2638  0::/../container_id2
2639
2640Note that this kind of setup is not encouraged.  A task inside cgroup
2641namespace should only be exposed to its own cgroupns hierarchy.
2642
2643setns(2) to another cgroup namespace is allowed when:
2644
2645(a) the process has CAP_SYS_ADMIN against its current user namespace
2646(b) the process has CAP_SYS_ADMIN against the target cgroup
2647    namespace's userns
2648
2649No implicit cgroup changes happen with attaching to another cgroup
2650namespace.  It is expected that the someone moves the attaching
2651process under the target cgroup namespace root.
2652
2653
2654Interaction with Other Namespaces
2655---------------------------------
2656
2657Namespace specific cgroup hierarchy can be mounted by a process
2658running inside a non-init cgroup namespace::
2659
2660  # mount -t cgroup2 none $MOUNT_POINT
2661
2662This will mount the unified cgroup hierarchy with cgroupns root as the
2663filesystem root.  The process needs CAP_SYS_ADMIN against its user and
2664mount namespaces.
2665
2666The virtualization of /proc/self/cgroup file combined with restricting
2667the view of cgroup hierarchy by namespace-private cgroupfs mount
2668provides a properly isolated cgroup view inside the container.
2669
2670
2671Information on Kernel Programming
2672=================================
2673
2674This section contains kernel programming information in the areas
2675where interacting with cgroup is necessary.  cgroup core and
2676controllers are not covered.
2677
2678
2679Filesystem Support for Writeback
2680--------------------------------
2681
2682A filesystem can support cgroup writeback by updating
2683address_space_operations->writepage[s]() to annotate bio's using the
2684following two functions.
2685
2686  wbc_init_bio(@wbc, @bio)
2687	Should be called for each bio carrying writeback data and
2688	associates the bio with the inode's owner cgroup and the
2689	corresponding request queue.  This must be called after
2690	a queue (device) has been associated with the bio and
2691	before submission.
2692
2693  wbc_account_cgroup_owner(@wbc, @page, @bytes)
2694	Should be called for each data segment being written out.
2695	While this function doesn't care exactly when it's called
2696	during the writeback session, it's the easiest and most
2697	natural to call it as data segments are added to a bio.
2698
2699With writeback bio's annotated, cgroup support can be enabled per
2700super_block by setting SB_I_CGROUPWB in ->s_iflags.  This allows for
2701selective disabling of cgroup writeback support which is helpful when
2702certain filesystem features, e.g. journaled data mode, are
2703incompatible.
2704
2705wbc_init_bio() binds the specified bio to its cgroup.  Depending on
2706the configuration, the bio may be executed at a lower priority and if
2707the writeback session is holding shared resources, e.g. a journal
2708entry, may lead to priority inversion.  There is no one easy solution
2709for the problem.  Filesystems can try to work around specific problem
2710cases by skipping wbc_init_bio() and using bio_associate_blkg()
2711directly.
2712
2713
2714Deprecated v1 Core Features
2715===========================
2716
2717- Multiple hierarchies including named ones are not supported.
2718
2719- All v1 mount options are not supported.
2720
2721- The "tasks" file is removed and "cgroup.procs" is not sorted.
2722
2723- "cgroup.clone_children" is removed.
2724
2725- /proc/cgroups is meaningless for v2.  Use "cgroup.controllers" file
2726  at the root instead.
2727
2728
2729Issues with v1 and Rationales for v2
2730====================================
2731
2732Multiple Hierarchies
2733--------------------
2734
2735cgroup v1 allowed an arbitrary number of hierarchies and each
2736hierarchy could host any number of controllers.  While this seemed to
2737provide a high level of flexibility, it wasn't useful in practice.
2738
2739For example, as there is only one instance of each controller, utility
2740type controllers such as freezer which can be useful in all
2741hierarchies could only be used in one.  The issue is exacerbated by
2742the fact that controllers couldn't be moved to another hierarchy once
2743hierarchies were populated.  Another issue was that all controllers
2744bound to a hierarchy were forced to have exactly the same view of the
2745hierarchy.  It wasn't possible to vary the granularity depending on
2746the specific controller.
2747
2748In practice, these issues heavily limited which controllers could be
2749put on the same hierarchy and most configurations resorted to putting
2750each controller on its own hierarchy.  Only closely related ones, such
2751as the cpu and cpuacct controllers, made sense to be put on the same
2752hierarchy.  This often meant that userland ended up managing multiple
2753similar hierarchies repeating the same steps on each hierarchy
2754whenever a hierarchy management operation was necessary.
2755
2756Furthermore, support for multiple hierarchies came at a steep cost.
2757It greatly complicated cgroup core implementation but more importantly
2758the support for multiple hierarchies restricted how cgroup could be
2759used in general and what controllers was able to do.
2760
2761There was no limit on how many hierarchies there might be, which meant
2762that a thread's cgroup membership couldn't be described in finite
2763length.  The key might contain any number of entries and was unlimited
2764in length, which made it highly awkward to manipulate and led to
2765addition of controllers which existed only to identify membership,
2766which in turn exacerbated the original problem of proliferating number
2767of hierarchies.
2768
2769Also, as a controller couldn't have any expectation regarding the
2770topologies of hierarchies other controllers might be on, each
2771controller had to assume that all other controllers were attached to
2772completely orthogonal hierarchies.  This made it impossible, or at
2773least very cumbersome, for controllers to cooperate with each other.
2774
2775In most use cases, putting controllers on hierarchies which are
2776completely orthogonal to each other isn't necessary.  What usually is
2777called for is the ability to have differing levels of granularity
2778depending on the specific controller.  In other words, hierarchy may
2779be collapsed from leaf towards root when viewed from specific
2780controllers.  For example, a given configuration might not care about
2781how memory is distributed beyond a certain level while still wanting
2782to control how CPU cycles are distributed.
2783
2784
2785Thread Granularity
2786------------------
2787
2788cgroup v1 allowed threads of a process to belong to different cgroups.
2789This didn't make sense for some controllers and those controllers
2790ended up implementing different ways to ignore such situations but
2791much more importantly it blurred the line between API exposed to
2792individual applications and system management interface.
2793
2794Generally, in-process knowledge is available only to the process
2795itself; thus, unlike service-level organization of processes,
2796categorizing threads of a process requires active participation from
2797the application which owns the target process.
2798
2799cgroup v1 had an ambiguously defined delegation model which got abused
2800in combination with thread granularity.  cgroups were delegated to
2801individual applications so that they can create and manage their own
2802sub-hierarchies and control resource distributions along them.  This
2803effectively raised cgroup to the status of a syscall-like API exposed
2804to lay programs.
2805
2806First of all, cgroup has a fundamentally inadequate interface to be
2807exposed this way.  For a process to access its own knobs, it has to
2808extract the path on the target hierarchy from /proc/self/cgroup,
2809construct the path by appending the name of the knob to the path, open
2810and then read and/or write to it.  This is not only extremely clunky
2811and unusual but also inherently racy.  There is no conventional way to
2812define transaction across the required steps and nothing can guarantee
2813that the process would actually be operating on its own sub-hierarchy.
2814
2815cgroup controllers implemented a number of knobs which would never be
2816accepted as public APIs because they were just adding control knobs to
2817system-management pseudo filesystem.  cgroup ended up with interface
2818knobs which were not properly abstracted or refined and directly
2819revealed kernel internal details.  These knobs got exposed to
2820individual applications through the ill-defined delegation mechanism
2821effectively abusing cgroup as a shortcut to implementing public APIs
2822without going through the required scrutiny.
2823
2824This was painful for both userland and kernel.  Userland ended up with
2825misbehaving and poorly abstracted interfaces and kernel exposing and
2826locked into constructs inadvertently.
2827
2828
2829Competition Between Inner Nodes and Threads
2830-------------------------------------------
2831
2832cgroup v1 allowed threads to be in any cgroups which created an
2833interesting problem where threads belonging to a parent cgroup and its
2834children cgroups competed for resources.  This was nasty as two
2835different types of entities competed and there was no obvious way to
2836settle it.  Different controllers did different things.
2837
2838The cpu controller considered threads and cgroups as equivalents and
2839mapped nice levels to cgroup weights.  This worked for some cases but
2840fell flat when children wanted to be allocated specific ratios of CPU
2841cycles and the number of internal threads fluctuated - the ratios
2842constantly changed as the number of competing entities fluctuated.
2843There also were other issues.  The mapping from nice level to weight
2844wasn't obvious or universal, and there were various other knobs which
2845simply weren't available for threads.
2846
2847The io controller implicitly created a hidden leaf node for each
2848cgroup to host the threads.  The hidden leaf had its own copies of all
2849the knobs with ``leaf_`` prefixed.  While this allowed equivalent
2850control over internal threads, it was with serious drawbacks.  It
2851always added an extra layer of nesting which wouldn't be necessary
2852otherwise, made the interface messy and significantly complicated the
2853implementation.
2854
2855The memory controller didn't have a way to control what happened
2856between internal tasks and child cgroups and the behavior was not
2857clearly defined.  There were attempts to add ad-hoc behaviors and
2858knobs to tailor the behavior to specific workloads which would have
2859led to problems extremely difficult to resolve in the long term.
2860
2861Multiple controllers struggled with internal tasks and came up with
2862different ways to deal with it; unfortunately, all the approaches were
2863severely flawed and, furthermore, the widely different behaviors
2864made cgroup as a whole highly inconsistent.
2865
2866This clearly is a problem which needs to be addressed from cgroup core
2867in a uniform way.
2868
2869
2870Other Interface Issues
2871----------------------
2872
2873cgroup v1 grew without oversight and developed a large number of
2874idiosyncrasies and inconsistencies.  One issue on the cgroup core side
2875was how an empty cgroup was notified - a userland helper binary was
2876forked and executed for each event.  The event delivery wasn't
2877recursive or delegatable.  The limitations of the mechanism also led
2878to in-kernel event delivery filtering mechanism further complicating
2879the interface.
2880
2881Controller interfaces were problematic too.  An extreme example is
2882controllers completely ignoring hierarchical organization and treating
2883all cgroups as if they were all located directly under the root
2884cgroup.  Some controllers exposed a large amount of inconsistent
2885implementation details to userland.
2886
2887There also was no consistency across controllers.  When a new cgroup
2888was created, some controllers defaulted to not imposing extra
2889restrictions while others disallowed any resource usage until
2890explicitly configured.  Configuration knobs for the same type of
2891control used widely differing naming schemes and formats.  Statistics
2892and information knobs were named arbitrarily and used different
2893formats and units even in the same controller.
2894
2895cgroup v2 establishes common conventions where appropriate and updates
2896controllers so that they expose minimal and consistent interfaces.
2897
2898
2899Controller Issues and Remedies
2900------------------------------
2901
2902Memory
2903~~~~~~
2904
2905The original lower boundary, the soft limit, is defined as a limit
2906that is per default unset.  As a result, the set of cgroups that
2907global reclaim prefers is opt-in, rather than opt-out.  The costs for
2908optimizing these mostly negative lookups are so high that the
2909implementation, despite its enormous size, does not even provide the
2910basic desirable behavior.  First off, the soft limit has no
2911hierarchical meaning.  All configured groups are organized in a global
2912rbtree and treated like equal peers, regardless where they are located
2913in the hierarchy.  This makes subtree delegation impossible.  Second,
2914the soft limit reclaim pass is so aggressive that it not just
2915introduces high allocation latencies into the system, but also impacts
2916system performance due to overreclaim, to the point where the feature
2917becomes self-defeating.
2918
2919The memory.low boundary on the other hand is a top-down allocated
2920reserve.  A cgroup enjoys reclaim protection when it's within its
2921effective low, which makes delegation of subtrees possible. It also
2922enjoys having reclaim pressure proportional to its overage when
2923above its effective low.
2924
2925The original high boundary, the hard limit, is defined as a strict
2926limit that can not budge, even if the OOM killer has to be called.
2927But this generally goes against the goal of making the most out of the
2928available memory.  The memory consumption of workloads varies during
2929runtime, and that requires users to overcommit.  But doing that with a
2930strict upper limit requires either a fairly accurate prediction of the
2931working set size or adding slack to the limit.  Since working set size
2932estimation is hard and error prone, and getting it wrong results in
2933OOM kills, most users tend to err on the side of a looser limit and
2934end up wasting precious resources.
2935
2936The memory.high boundary on the other hand can be set much more
2937conservatively.  When hit, it throttles allocations by forcing them
2938into direct reclaim to work off the excess, but it never invokes the
2939OOM killer.  As a result, a high boundary that is chosen too
2940aggressively will not terminate the processes, but instead it will
2941lead to gradual performance degradation.  The user can monitor this
2942and make corrections until the minimal memory footprint that still
2943gives acceptable performance is found.
2944
2945In extreme cases, with many concurrent allocations and a complete
2946breakdown of reclaim progress within the group, the high boundary can
2947be exceeded.  But even then it's mostly better to satisfy the
2948allocation from the slack available in other groups or the rest of the
2949system than killing the group.  Otherwise, memory.max is there to
2950limit this type of spillover and ultimately contain buggy or even
2951malicious applications.
2952
2953Setting the original memory.limit_in_bytes below the current usage was
2954subject to a race condition, where concurrent charges could cause the
2955limit setting to fail. memory.max on the other hand will first set the
2956limit to prevent new charges, and then reclaim and OOM kill until the
2957new limit is met - or the task writing to memory.max is killed.
2958
2959The combined memory+swap accounting and limiting is replaced by real
2960control over swap space.
2961
2962The main argument for a combined memory+swap facility in the original
2963cgroup design was that global or parental pressure would always be
2964able to swap all anonymous memory of a child group, regardless of the
2965child's own (possibly untrusted) configuration.  However, untrusted
2966groups can sabotage swapping by other means - such as referencing its
2967anonymous memory in a tight loop - and an admin can not assume full
2968swappability when overcommitting untrusted jobs.
2969
2970For trusted jobs, on the other hand, a combined counter is not an
2971intuitive userspace interface, and it flies in the face of the idea
2972that cgroup controllers should account and limit specific physical
2973resources.  Swap space is a resource like all others in the system,
2974and that's why unified hierarchy allows distributing it separately.
2975