xref: /linux/Documentation/admin-guide/cgroup-v2.rst (revision 4fc4187984e5dbd7ad63c3acf0b228fa9bf298d3)
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  memory_hugetlb_accounting
214        Count HugeTLB memory usage towards the cgroup's overall
215        memory usage for the memory controller (for the purpose of
216        statistics reporting and memory protetion). This is a new
217        behavior that could regress existing setups, so it must be
218        explicitly opted in with this mount option.
219
220        A few caveats to keep in mind:
221
222        * There is no HugeTLB pool management involved in the memory
223          controller. The pre-allocated pool does not belong to anyone.
224          Specifically, when a new HugeTLB folio is allocated to
225          the pool, it is not accounted for from the perspective of the
226          memory controller. It is only charged to a cgroup when it is
227          actually used (for e.g at page fault time). Host memory
228          overcommit management has to consider this when configuring
229          hard limits. In general, HugeTLB pool management should be
230          done via other mechanisms (such as the HugeTLB controller).
231        * Failure to charge a HugeTLB folio to the memory controller
232          results in SIGBUS. This could happen even if the HugeTLB pool
233          still has pages available (but the cgroup limit is hit and
234          reclaim attempt fails).
235        * Charging HugeTLB memory towards the memory controller affects
236          memory protection and reclaim dynamics. Any userspace tuning
237          (of low, min limits for e.g) needs to take this into account.
238        * HugeTLB pages utilized while this option is not selected
239          will not be tracked by the memory controller (even if cgroup
240          v2 is remounted later on).
241
242  pids_localevents
243        The option restores v1-like behavior of pids.events:max, that is only
244        local (inside cgroup proper) fork failures are counted. Without this
245        option pids.events.max represents any pids.max enforcemnt across
246        cgroup's subtree.
247
248
249
250Organizing Processes and Threads
251--------------------------------
252
253Processes
254~~~~~~~~~
255
256Initially, only the root cgroup exists to which all processes belong.
257A child cgroup can be created by creating a sub-directory::
258
259  # mkdir $CGROUP_NAME
260
261A given cgroup may have multiple child cgroups forming a tree
262structure.  Each cgroup has a read-writable interface file
263"cgroup.procs".  When read, it lists the PIDs of all processes which
264belong to the cgroup one-per-line.  The PIDs are not ordered and the
265same PID may show up more than once if the process got moved to
266another cgroup and then back or the PID got recycled while reading.
267
268A process can be migrated into a cgroup by writing its PID to the
269target cgroup's "cgroup.procs" file.  Only one process can be migrated
270on a single write(2) call.  If a process is composed of multiple
271threads, writing the PID of any thread migrates all threads of the
272process.
273
274When a process forks a child process, the new process is born into the
275cgroup that the forking process belongs to at the time of the
276operation.  After exit, a process stays associated with the cgroup
277that it belonged to at the time of exit until it's reaped; however, a
278zombie process does not appear in "cgroup.procs" and thus can't be
279moved to another cgroup.
280
281A cgroup which doesn't have any children or live processes can be
282destroyed by removing the directory.  Note that a cgroup which doesn't
283have any children and is associated only with zombie processes is
284considered empty and can be removed::
285
286  # rmdir $CGROUP_NAME
287
288"/proc/$PID/cgroup" lists a process's cgroup membership.  If legacy
289cgroup is in use in the system, this file may contain multiple lines,
290one for each hierarchy.  The entry for cgroup v2 is always in the
291format "0::$PATH"::
292
293  # cat /proc/842/cgroup
294  ...
295  0::/test-cgroup/test-cgroup-nested
296
297If the process becomes a zombie and the cgroup it was associated with
298is removed subsequently, " (deleted)" is appended to the path::
299
300  # cat /proc/842/cgroup
301  ...
302  0::/test-cgroup/test-cgroup-nested (deleted)
303
304
305Threads
306~~~~~~~
307
308cgroup v2 supports thread granularity for a subset of controllers to
309support use cases requiring hierarchical resource distribution across
310the threads of a group of processes.  By default, all threads of a
311process belong to the same cgroup, which also serves as the resource
312domain to host resource consumptions which are not specific to a
313process or thread.  The thread mode allows threads to be spread across
314a subtree while still maintaining the common resource domain for them.
315
316Controllers which support thread mode are called threaded controllers.
317The ones which don't are called domain controllers.
318
319Marking a cgroup threaded makes it join the resource domain of its
320parent as a threaded cgroup.  The parent may be another threaded
321cgroup whose resource domain is further up in the hierarchy.  The root
322of a threaded subtree, that is, the nearest ancestor which is not
323threaded, is called threaded domain or thread root interchangeably and
324serves as the resource domain for the entire subtree.
325
326Inside a threaded subtree, threads of a process can be put in
327different cgroups and are not subject to the no internal process
328constraint - threaded controllers can be enabled on non-leaf cgroups
329whether they have threads in them or not.
330
331As the threaded domain cgroup hosts all the domain resource
332consumptions of the subtree, it is considered to have internal
333resource consumptions whether there are processes in it or not and
334can't have populated child cgroups which aren't threaded.  Because the
335root cgroup is not subject to no internal process constraint, it can
336serve both as a threaded domain and a parent to domain cgroups.
337
338The current operation mode or type of the cgroup is shown in the
339"cgroup.type" file which indicates whether the cgroup is a normal
340domain, a domain which is serving as the domain of a threaded subtree,
341or a threaded cgroup.
342
343On creation, a cgroup is always a domain cgroup and can be made
344threaded by writing "threaded" to the "cgroup.type" file.  The
345operation is single direction::
346
347  # echo threaded > cgroup.type
348
349Once threaded, the cgroup can't be made a domain again.  To enable the
350thread mode, the following conditions must be met.
351
352- As the cgroup will join the parent's resource domain.  The parent
353  must either be a valid (threaded) domain or a threaded cgroup.
354
355- When the parent is an unthreaded domain, it must not have any domain
356  controllers enabled or populated domain children.  The root is
357  exempt from this requirement.
358
359Topology-wise, a cgroup can be in an invalid state.  Please consider
360the following topology::
361
362  A (threaded domain) - B (threaded) - C (domain, just created)
363
364C is created as a domain but isn't connected to a parent which can
365host child domains.  C can't be used until it is turned into a
366threaded cgroup.  "cgroup.type" file will report "domain (invalid)" in
367these cases.  Operations which fail due to invalid topology use
368EOPNOTSUPP as the errno.
369
370A domain cgroup is turned into a threaded domain when one of its child
371cgroup becomes threaded or threaded controllers are enabled in the
372"cgroup.subtree_control" file while there are processes in the cgroup.
373A threaded domain reverts to a normal domain when the conditions
374clear.
375
376When read, "cgroup.threads" contains the list of the thread IDs of all
377threads in the cgroup.  Except that the operations are per-thread
378instead of per-process, "cgroup.threads" has the same format and
379behaves the same way as "cgroup.procs".  While "cgroup.threads" can be
380written to in any cgroup, as it can only move threads inside the same
381threaded domain, its operations are confined inside each threaded
382subtree.
383
384The threaded domain cgroup serves as the resource domain for the whole
385subtree, and, while the threads can be scattered across the subtree,
386all the processes are considered to be in the threaded domain cgroup.
387"cgroup.procs" in a threaded domain cgroup contains the PIDs of all
388processes in the subtree and is not readable in the subtree proper.
389However, "cgroup.procs" can be written to from anywhere in the subtree
390to migrate all threads of the matching process to the cgroup.
391
392Only threaded controllers can be enabled in a threaded subtree.  When
393a threaded controller is enabled inside a threaded subtree, it only
394accounts for and controls resource consumptions associated with the
395threads in the cgroup and its descendants.  All consumptions which
396aren't tied to a specific thread belong to the threaded domain cgroup.
397
398Because a threaded subtree is exempt from no internal process
399constraint, a threaded controller must be able to handle competition
400between threads in a non-leaf cgroup and its child cgroups.  Each
401threaded controller defines how such competitions are handled.
402
403Currently, the following controllers are threaded and can be enabled
404in a threaded cgroup::
405
406- cpu
407- cpuset
408- perf_event
409- pids
410
411[Un]populated Notification
412--------------------------
413
414Each non-root cgroup has a "cgroup.events" file which contains
415"populated" field indicating whether the cgroup's sub-hierarchy has
416live processes in it.  Its value is 0 if there is no live process in
417the cgroup and its descendants; otherwise, 1.  poll and [id]notify
418events are triggered when the value changes.  This can be used, for
419example, to start a clean-up operation after all processes of a given
420sub-hierarchy have exited.  The populated state updates and
421notifications are recursive.  Consider the following sub-hierarchy
422where the numbers in the parentheses represent the numbers of processes
423in each cgroup::
424
425  A(4) - B(0) - C(1)
426              \ D(0)
427
428A, B and C's "populated" fields would be 1 while D's 0.  After the one
429process in C exits, B and C's "populated" fields would flip to "0" and
430file modified events will be generated on the "cgroup.events" files of
431both cgroups.
432
433
434Controlling Controllers
435-----------------------
436
437Enabling and Disabling
438~~~~~~~~~~~~~~~~~~~~~~
439
440Each cgroup has a "cgroup.controllers" file which lists all
441controllers available for the cgroup to enable::
442
443  # cat cgroup.controllers
444  cpu io memory
445
446No controller is enabled by default.  Controllers can be enabled and
447disabled by writing to the "cgroup.subtree_control" file::
448
449  # echo "+cpu +memory -io" > cgroup.subtree_control
450
451Only controllers which are listed in "cgroup.controllers" can be
452enabled.  When multiple operations are specified as above, either they
453all succeed or fail.  If multiple operations on the same controller
454are specified, the last one is effective.
455
456Enabling a controller in a cgroup indicates that the distribution of
457the target resource across its immediate children will be controlled.
458Consider the following sub-hierarchy.  The enabled controllers are
459listed in parentheses::
460
461  A(cpu,memory) - B(memory) - C()
462                            \ D()
463
464As A has "cpu" and "memory" enabled, A will control the distribution
465of CPU cycles and memory to its children, in this case, B.  As B has
466"memory" enabled but not "CPU", C and D will compete freely on CPU
467cycles but their division of memory available to B will be controlled.
468
469As a controller regulates the distribution of the target resource to
470the cgroup's children, enabling it creates the controller's interface
471files in the child cgroups.  In the above example, enabling "cpu" on B
472would create the "cpu." prefixed controller interface files in C and
473D.  Likewise, disabling "memory" from B would remove the "memory."
474prefixed controller interface files from C and D.  This means that the
475controller interface files - anything which doesn't start with
476"cgroup." are owned by the parent rather than the cgroup itself.
477
478
479Top-down Constraint
480~~~~~~~~~~~~~~~~~~~
481
482Resources are distributed top-down and a cgroup can further distribute
483a resource only if the resource has been distributed to it from the
484parent.  This means that all non-root "cgroup.subtree_control" files
485can only contain controllers which are enabled in the parent's
486"cgroup.subtree_control" file.  A controller can be enabled only if
487the parent has the controller enabled and a controller can't be
488disabled if one or more children have it enabled.
489
490
491No Internal Process Constraint
492~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
493
494Non-root cgroups can distribute domain resources to their children
495only when they don't have any processes of their own.  In other words,
496only domain cgroups which don't contain any processes can have domain
497controllers enabled in their "cgroup.subtree_control" files.
498
499This guarantees that, when a domain controller is looking at the part
500of the hierarchy which has it enabled, processes are always only on
501the leaves.  This rules out situations where child cgroups compete
502against internal processes of the parent.
503
504The root cgroup is exempt from this restriction.  Root contains
505processes and anonymous resource consumption which can't be associated
506with any other cgroups and requires special treatment from most
507controllers.  How resource consumption in the root cgroup is governed
508is up to each controller (for more information on this topic please
509refer to the Non-normative information section in the Controllers
510chapter).
511
512Note that the restriction doesn't get in the way if there is no
513enabled controller in the cgroup's "cgroup.subtree_control".  This is
514important as otherwise it wouldn't be possible to create children of a
515populated cgroup.  To control resource distribution of a cgroup, the
516cgroup must create children and transfer all its processes to the
517children before enabling controllers in its "cgroup.subtree_control"
518file.
519
520
521Delegation
522----------
523
524Model of Delegation
525~~~~~~~~~~~~~~~~~~~
526
527A cgroup can be delegated in two ways.  First, to a less privileged
528user by granting write access of the directory and its "cgroup.procs",
529"cgroup.threads" and "cgroup.subtree_control" files to the user.
530Second, if the "nsdelegate" mount option is set, automatically to a
531cgroup namespace on namespace creation.
532
533Because the resource control interface files in a given directory
534control the distribution of the parent's resources, the delegatee
535shouldn't be allowed to write to them.  For the first method, this is
536achieved by not granting access to these files.  For the second, the
537kernel rejects writes to all files other than "cgroup.procs" and
538"cgroup.subtree_control" on a namespace root from inside the
539namespace.
540
541The end results are equivalent for both delegation types.  Once
542delegated, the user can build sub-hierarchy under the directory,
543organize processes inside it as it sees fit and further distribute the
544resources it received from the parent.  The limits and other settings
545of all resource controllers are hierarchical and regardless of what
546happens in the delegated sub-hierarchy, nothing can escape the
547resource restrictions imposed by the parent.
548
549Currently, cgroup doesn't impose any restrictions on the number of
550cgroups in or nesting depth of a delegated sub-hierarchy; however,
551this may be limited explicitly in the future.
552
553
554Delegation Containment
555~~~~~~~~~~~~~~~~~~~~~~
556
557A delegated sub-hierarchy is contained in the sense that processes
558can't be moved into or out of the sub-hierarchy by the delegatee.
559
560For delegations to a less privileged user, this is achieved by
561requiring the following conditions for a process with a non-root euid
562to migrate a target process into a cgroup by writing its PID to the
563"cgroup.procs" file.
564
565- The writer must have write access to the "cgroup.procs" file.
566
567- The writer must have write access to the "cgroup.procs" file of the
568  common ancestor of the source and destination cgroups.
569
570The above two constraints ensure that while a delegatee may migrate
571processes around freely in the delegated sub-hierarchy it can't pull
572in from or push out to outside the sub-hierarchy.
573
574For an example, let's assume cgroups C0 and C1 have been delegated to
575user U0 who created C00, C01 under C0 and C10 under C1 as follows and
576all processes under C0 and C1 belong to U0::
577
578  ~~~~~~~~~~~~~ - C0 - C00
579  ~ cgroup    ~      \ C01
580  ~ hierarchy ~
581  ~~~~~~~~~~~~~ - C1 - C10
582
583Let's also say U0 wants to write the PID of a process which is
584currently in C10 into "C00/cgroup.procs".  U0 has write access to the
585file; however, the common ancestor of the source cgroup C10 and the
586destination cgroup C00 is above the points of delegation and U0 would
587not have write access to its "cgroup.procs" files and thus the write
588will be denied with -EACCES.
589
590For delegations to namespaces, containment is achieved by requiring
591that both the source and destination cgroups are reachable from the
592namespace of the process which is attempting the migration.  If either
593is not reachable, the migration is rejected with -ENOENT.
594
595
596Guidelines
597----------
598
599Organize Once and Control
600~~~~~~~~~~~~~~~~~~~~~~~~~
601
602Migrating a process across cgroups is a relatively expensive operation
603and stateful resources such as memory are not moved together with the
604process.  This is an explicit design decision as there often exist
605inherent trade-offs between migration and various hot paths in terms
606of synchronization cost.
607
608As such, migrating processes across cgroups frequently as a means to
609apply different resource restrictions is discouraged.  A workload
610should be assigned to a cgroup according to the system's logical and
611resource structure once on start-up.  Dynamic adjustments to resource
612distribution can be made by changing controller configuration through
613the interface files.
614
615
616Avoid Name Collisions
617~~~~~~~~~~~~~~~~~~~~~
618
619Interface files for a cgroup and its children cgroups occupy the same
620directory and it is possible to create children cgroups which collide
621with interface files.
622
623All cgroup core interface files are prefixed with "cgroup." and each
624controller's interface files are prefixed with the controller name and
625a dot.  A controller's name is composed of lower case alphabets and
626'_'s but never begins with an '_' so it can be used as the prefix
627character for collision avoidance.  Also, interface file names won't
628start or end with terms which are often used in categorizing workloads
629such as job, service, slice, unit or workload.
630
631cgroup doesn't do anything to prevent name collisions and it's the
632user's responsibility to avoid them.
633
634
635Resource Distribution Models
636============================
637
638cgroup controllers implement several resource distribution schemes
639depending on the resource type and expected use cases.  This section
640describes major schemes in use along with their expected behaviors.
641
642
643Weights
644-------
645
646A parent's resource is distributed by adding up the weights of all
647active children and giving each the fraction matching the ratio of its
648weight against the sum.  As only children which can make use of the
649resource at the moment participate in the distribution, this is
650work-conserving.  Due to the dynamic nature, this model is usually
651used for stateless resources.
652
653All weights are in the range [1, 10000] with the default at 100.  This
654allows symmetric multiplicative biases in both directions at fine
655enough granularity while staying in the intuitive range.
656
657As long as the weight is in range, all configuration combinations are
658valid and there is no reason to reject configuration changes or
659process migrations.
660
661"cpu.weight" proportionally distributes CPU cycles to active children
662and is an example of this type.
663
664
665.. _cgroupv2-limits-distributor:
666
667Limits
668------
669
670A child can only consume up to the configured amount of the resource.
671Limits can be over-committed - the sum of the limits of children can
672exceed the amount of resource available to the parent.
673
674Limits are in the range [0, max] and defaults to "max", which is noop.
675
676As limits can be over-committed, all configuration combinations are
677valid and there is no reason to reject configuration changes or
678process migrations.
679
680"io.max" limits the maximum BPS and/or IOPS that a cgroup can consume
681on an IO device and is an example of this type.
682
683.. _cgroupv2-protections-distributor:
684
685Protections
686-----------
687
688A cgroup is protected up to the configured amount of the resource
689as long as the usages of all its ancestors are under their
690protected levels.  Protections can be hard guarantees or best effort
691soft boundaries.  Protections can also be over-committed in which case
692only up to the amount available to the parent is protected among
693children.
694
695Protections are in the range [0, max] and defaults to 0, which is
696noop.
697
698As protections can be over-committed, all configuration combinations
699are valid and there is no reason to reject configuration changes or
700process migrations.
701
702"memory.low" implements best-effort memory protection and is an
703example of this type.
704
705
706Allocations
707-----------
708
709A cgroup is exclusively allocated a certain amount of a finite
710resource.  Allocations can't be over-committed - the sum of the
711allocations of children can not exceed the amount of resource
712available to the parent.
713
714Allocations are in the range [0, max] and defaults to 0, which is no
715resource.
716
717As allocations can't be over-committed, some configuration
718combinations are invalid and should be rejected.  Also, if the
719resource is mandatory for execution of processes, process migrations
720may be rejected.
721
722"cpu.rt.max" hard-allocates realtime slices and is an example of this
723type.
724
725
726Interface Files
727===============
728
729Format
730------
731
732All interface files should be in one of the following formats whenever
733possible::
734
735  New-line separated values
736  (when only one value can be written at once)
737
738	VAL0\n
739	VAL1\n
740	...
741
742  Space separated values
743  (when read-only or multiple values can be written at once)
744
745	VAL0 VAL1 ...\n
746
747  Flat keyed
748
749	KEY0 VAL0\n
750	KEY1 VAL1\n
751	...
752
753  Nested keyed
754
755	KEY0 SUB_KEY0=VAL00 SUB_KEY1=VAL01...
756	KEY1 SUB_KEY0=VAL10 SUB_KEY1=VAL11...
757	...
758
759For a writable file, the format for writing should generally match
760reading; however, controllers may allow omitting later fields or
761implement restricted shortcuts for most common use cases.
762
763For both flat and nested keyed files, only the values for a single key
764can be written at a time.  For nested keyed files, the sub key pairs
765may be specified in any order and not all pairs have to be specified.
766
767
768Conventions
769-----------
770
771- Settings for a single feature should be contained in a single file.
772
773- The root cgroup should be exempt from resource control and thus
774  shouldn't have resource control interface files.
775
776- The default time unit is microseconds.  If a different unit is ever
777  used, an explicit unit suffix must be present.
778
779- A parts-per quantity should use a percentage decimal with at least
780  two digit fractional part - e.g. 13.40.
781
782- If a controller implements weight based resource distribution, its
783  interface file should be named "weight" and have the range [1,
784  10000] with 100 as the default.  The values are chosen to allow
785  enough and symmetric bias in both directions while keeping it
786  intuitive (the default is 100%).
787
788- If a controller implements an absolute resource guarantee and/or
789  limit, the interface files should be named "min" and "max"
790  respectively.  If a controller implements best effort resource
791  guarantee and/or limit, the interface files should be named "low"
792  and "high" respectively.
793
794  In the above four control files, the special token "max" should be
795  used to represent upward infinity for both reading and writing.
796
797- If a setting has a configurable default value and keyed specific
798  overrides, the default entry should be keyed with "default" and
799  appear as the first entry in the file.
800
801  The default value can be updated by writing either "default $VAL" or
802  "$VAL".
803
804  When writing to update a specific override, "default" can be used as
805  the value to indicate removal of the override.  Override entries
806  with "default" as the value must not appear when read.
807
808  For example, a setting which is keyed by major:minor device numbers
809  with integer values may look like the following::
810
811    # cat cgroup-example-interface-file
812    default 150
813    8:0 300
814
815  The default value can be updated by::
816
817    # echo 125 > cgroup-example-interface-file
818
819  or::
820
821    # echo "default 125" > cgroup-example-interface-file
822
823  An override can be set by::
824
825    # echo "8:16 170" > cgroup-example-interface-file
826
827  and cleared by::
828
829    # echo "8:0 default" > cgroup-example-interface-file
830    # cat cgroup-example-interface-file
831    default 125
832    8:16 170
833
834- For events which are not very high frequency, an interface file
835  "events" should be created which lists event key value pairs.
836  Whenever a notifiable event happens, file modified event should be
837  generated on the file.
838
839
840Core Interface Files
841--------------------
842
843All cgroup core files are prefixed with "cgroup."
844
845  cgroup.type
846	A read-write single value file which exists on non-root
847	cgroups.
848
849	When read, it indicates the current type of the cgroup, which
850	can be one of the following values.
851
852	- "domain" : A normal valid domain cgroup.
853
854	- "domain threaded" : A threaded domain cgroup which is
855          serving as the root of a threaded subtree.
856
857	- "domain invalid" : A cgroup which is in an invalid state.
858	  It can't be populated or have controllers enabled.  It may
859	  be allowed to become a threaded cgroup.
860
861	- "threaded" : A threaded cgroup which is a member of a
862          threaded subtree.
863
864	A cgroup can be turned into a threaded cgroup by writing
865	"threaded" to this file.
866
867  cgroup.procs
868	A read-write new-line separated values file which exists on
869	all cgroups.
870
871	When read, it lists the PIDs of all processes which belong to
872	the cgroup one-per-line.  The PIDs are not ordered and the
873	same PID may show up more than once if the process got moved
874	to another cgroup and then back or the PID got recycled while
875	reading.
876
877	A PID can be written to migrate the process associated with
878	the PID to the cgroup.  The writer should match all of the
879	following conditions.
880
881	- It must have write access to the "cgroup.procs" file.
882
883	- It must have write access to the "cgroup.procs" file of the
884	  common ancestor of the source and destination cgroups.
885
886	When delegating a sub-hierarchy, write access to this file
887	should be granted along with the containing directory.
888
889	In a threaded cgroup, reading this file fails with EOPNOTSUPP
890	as all the processes belong to the thread root.  Writing is
891	supported and moves every thread of the process to the cgroup.
892
893  cgroup.threads
894	A read-write new-line separated values file which exists on
895	all cgroups.
896
897	When read, it lists the TIDs of all threads which belong to
898	the cgroup one-per-line.  The TIDs are not ordered and the
899	same TID may show up more than once if the thread got moved to
900	another cgroup and then back or the TID got recycled while
901	reading.
902
903	A TID can be written to migrate the thread associated with the
904	TID to the cgroup.  The writer should match all of the
905	following conditions.
906
907	- It must have write access to the "cgroup.threads" file.
908
909	- The cgroup that the thread is currently in must be in the
910          same resource domain as the destination cgroup.
911
912	- It must have write access to the "cgroup.procs" file of the
913	  common ancestor of the source and destination cgroups.
914
915	When delegating a sub-hierarchy, write access to this file
916	should be granted along with the containing directory.
917
918  cgroup.controllers
919	A read-only space separated values file which exists on all
920	cgroups.
921
922	It shows space separated list of all controllers available to
923	the cgroup.  The controllers are not ordered.
924
925  cgroup.subtree_control
926	A read-write space separated values file which exists on all
927	cgroups.  Starts out empty.
928
929	When read, it shows space separated list of the controllers
930	which are enabled to control resource distribution from the
931	cgroup to its children.
932
933	Space separated list of controllers prefixed with '+' or '-'
934	can be written to enable or disable controllers.  A controller
935	name prefixed with '+' enables the controller and '-'
936	disables.  If a controller appears more than once on the list,
937	the last one is effective.  When multiple enable and disable
938	operations are specified, either all succeed or all fail.
939
940  cgroup.events
941	A read-only flat-keyed file which exists on non-root cgroups.
942	The following entries are defined.  Unless specified
943	otherwise, a value change in this file generates a file
944	modified event.
945
946	  populated
947		1 if the cgroup or its descendants contains any live
948		processes; otherwise, 0.
949	  frozen
950		1 if the cgroup is frozen; otherwise, 0.
951
952  cgroup.max.descendants
953	A read-write single value files.  The default is "max".
954
955	Maximum allowed number of descent cgroups.
956	If the actual number of descendants is equal or larger,
957	an attempt to create a new cgroup in the hierarchy will fail.
958
959  cgroup.max.depth
960	A read-write single value files.  The default is "max".
961
962	Maximum allowed descent depth below the current cgroup.
963	If the actual descent depth is equal or larger,
964	an attempt to create a new child cgroup will fail.
965
966  cgroup.stat
967	A read-only flat-keyed file with the following entries:
968
969	  nr_descendants
970		Total number of visible descendant cgroups.
971
972	  nr_dying_descendants
973		Total number of dying descendant cgroups. A cgroup becomes
974		dying after being deleted by a user. The cgroup will remain
975		in dying state for some time undefined time (which can depend
976		on system load) before being completely destroyed.
977
978		A process can't enter a dying cgroup under any circumstances,
979		a dying cgroup can't revive.
980
981		A dying cgroup can consume system resources not exceeding
982		limits, which were active at the moment of cgroup deletion.
983
984  cgroup.freeze
985	A read-write single value file which exists on non-root cgroups.
986	Allowed values are "0" and "1". The default is "0".
987
988	Writing "1" to the file causes freezing of the cgroup and all
989	descendant cgroups. This means that all belonging processes will
990	be stopped and will not run until the cgroup will be explicitly
991	unfrozen. Freezing of the cgroup may take some time; when this action
992	is completed, the "frozen" value in the cgroup.events control file
993	will be updated to "1" and the corresponding notification will be
994	issued.
995
996	A cgroup can be frozen either by its own settings, or by settings
997	of any ancestor cgroups. If any of ancestor cgroups is frozen, the
998	cgroup will remain frozen.
999
1000	Processes in the frozen cgroup can be killed by a fatal signal.
1001	They also can enter and leave a frozen cgroup: either by an explicit
1002	move by a user, or if freezing of the cgroup races with fork().
1003	If a process is moved to a frozen cgroup, it stops. If a process is
1004	moved out of a frozen cgroup, it becomes running.
1005
1006	Frozen status of a cgroup doesn't affect any cgroup tree operations:
1007	it's possible to delete a frozen (and empty) cgroup, as well as
1008	create new sub-cgroups.
1009
1010  cgroup.kill
1011	A write-only single value file which exists in non-root cgroups.
1012	The only allowed value is "1".
1013
1014	Writing "1" to the file causes the cgroup and all descendant cgroups to
1015	be killed. This means that all processes located in the affected cgroup
1016	tree will be killed via SIGKILL.
1017
1018	Killing a cgroup tree will deal with concurrent forks appropriately and
1019	is protected against migrations.
1020
1021	In a threaded cgroup, writing this file fails with EOPNOTSUPP as
1022	killing cgroups is a process directed operation, i.e. it affects
1023	the whole thread-group.
1024
1025  cgroup.pressure
1026	A read-write single value file that allowed values are "0" and "1".
1027	The default is "1".
1028
1029	Writing "0" to the file will disable the cgroup PSI accounting.
1030	Writing "1" to the file will re-enable the cgroup PSI accounting.
1031
1032	This control attribute is not hierarchical, so disable or enable PSI
1033	accounting in a cgroup does not affect PSI accounting in descendants
1034	and doesn't need pass enablement via ancestors from root.
1035
1036	The reason this control attribute exists is that PSI accounts stalls for
1037	each cgroup separately and aggregates it at each level of the hierarchy.
1038	This may cause non-negligible overhead for some workloads when under
1039	deep level of the hierarchy, in which case this control attribute can
1040	be used to disable PSI accounting in the non-leaf cgroups.
1041
1042  irq.pressure
1043	A read-write nested-keyed file.
1044
1045	Shows pressure stall information for IRQ/SOFTIRQ. See
1046	:ref:`Documentation/accounting/psi.rst <psi>` for details.
1047
1048Controllers
1049===========
1050
1051.. _cgroup-v2-cpu:
1052
1053CPU
1054---
1055
1056The "cpu" controllers regulates distribution of CPU cycles.  This
1057controller implements weight and absolute bandwidth limit models for
1058normal scheduling policy and absolute bandwidth allocation model for
1059realtime scheduling policy.
1060
1061In all the above models, cycles distribution is defined only on a temporal
1062base and it does not account for the frequency at which tasks are executed.
1063The (optional) utilization clamping support allows to hint the schedutil
1064cpufreq governor about the minimum desired frequency which should always be
1065provided by a CPU, as well as the maximum desired frequency, which should not
1066be exceeded by a CPU.
1067
1068WARNING: cgroup2 doesn't yet support control of realtime processes. For
1069a kernel built with the CONFIG_RT_GROUP_SCHED option enabled for group
1070scheduling of realtime processes, the cpu controller can only be enabled
1071when all RT processes are in the root cgroup.  This limitation does
1072not apply if CONFIG_RT_GROUP_SCHED is disabled.  Be aware that system
1073management software may already have placed RT processes into nonroot
1074cgroups during the system boot process, and these processes may need
1075to be moved to the root cgroup before the cpu controller can be enabled
1076with a CONFIG_RT_GROUP_SCHED enabled kernel.
1077
1078
1079CPU Interface Files
1080~~~~~~~~~~~~~~~~~~~
1081
1082All time durations are in microseconds.
1083
1084  cpu.stat
1085	A read-only flat-keyed file.
1086	This file exists whether the controller is enabled or not.
1087
1088	It always reports the following three stats:
1089
1090	- usage_usec
1091	- user_usec
1092	- system_usec
1093
1094	and the following five when the controller is enabled:
1095
1096	- nr_periods
1097	- nr_throttled
1098	- throttled_usec
1099	- nr_bursts
1100	- burst_usec
1101
1102  cpu.weight
1103	A read-write single value file which exists on non-root
1104	cgroups.  The default is "100".
1105
1106	For non idle groups (cpu.idle = 0), the weight is in the
1107	range [1, 10000].
1108
1109	If the cgroup has been configured to be SCHED_IDLE (cpu.idle = 1),
1110	then the weight will show as a 0.
1111
1112  cpu.weight.nice
1113	A read-write single value file which exists on non-root
1114	cgroups.  The default is "0".
1115
1116	The nice value is in the range [-20, 19].
1117
1118	This interface file is an alternative interface for
1119	"cpu.weight" and allows reading and setting weight using the
1120	same values used by nice(2).  Because the range is smaller and
1121	granularity is coarser for the nice values, the read value is
1122	the closest approximation of the current weight.
1123
1124  cpu.max
1125	A read-write two value file which exists on non-root cgroups.
1126	The default is "max 100000".
1127
1128	The maximum bandwidth limit.  It's in the following format::
1129
1130	  $MAX $PERIOD
1131
1132	which indicates that the group may consume up to $MAX in each
1133	$PERIOD duration.  "max" for $MAX indicates no limit.  If only
1134	one number is written, $MAX is updated.
1135
1136  cpu.max.burst
1137	A read-write single value file which exists on non-root
1138	cgroups.  The default is "0".
1139
1140	The burst in the range [0, $MAX].
1141
1142  cpu.pressure
1143	A read-write nested-keyed file.
1144
1145	Shows pressure stall information for CPU. See
1146	:ref:`Documentation/accounting/psi.rst <psi>` for details.
1147
1148  cpu.uclamp.min
1149        A read-write single value file which exists on non-root cgroups.
1150        The default is "0", i.e. no utilization boosting.
1151
1152        The requested minimum utilization (protection) as a percentage
1153        rational number, e.g. 12.34 for 12.34%.
1154
1155        This interface allows reading and setting minimum utilization clamp
1156        values similar to the sched_setattr(2). This minimum utilization
1157        value is used to clamp the task specific minimum utilization clamp.
1158
1159        The requested minimum utilization (protection) is always capped by
1160        the current value for the maximum utilization (limit), i.e.
1161        `cpu.uclamp.max`.
1162
1163  cpu.uclamp.max
1164        A read-write single value file which exists on non-root cgroups.
1165        The default is "max". i.e. no utilization capping
1166
1167        The requested maximum utilization (limit) as a percentage rational
1168        number, e.g. 98.76 for 98.76%.
1169
1170        This interface allows reading and setting maximum utilization clamp
1171        values similar to the sched_setattr(2). This maximum utilization
1172        value is used to clamp the task specific maximum utilization clamp.
1173
1174  cpu.idle
1175	A read-write single value file which exists on non-root cgroups.
1176	The default is 0.
1177
1178	This is the cgroup analog of the per-task SCHED_IDLE sched policy.
1179	Setting this value to a 1 will make the scheduling policy of the
1180	cgroup SCHED_IDLE. The threads inside the cgroup will retain their
1181	own relative priorities, but the cgroup itself will be treated as
1182	very low priority relative to its peers.
1183
1184
1185
1186Memory
1187------
1188
1189The "memory" controller regulates distribution of memory.  Memory is
1190stateful and implements both limit and protection models.  Due to the
1191intertwining between memory usage and reclaim pressure and the
1192stateful nature of memory, the distribution model is relatively
1193complex.
1194
1195While not completely water-tight, all major memory usages by a given
1196cgroup are tracked so that the total memory consumption can be
1197accounted and controlled to a reasonable extent.  Currently, the
1198following types of memory usages are tracked.
1199
1200- Userland memory - page cache and anonymous memory.
1201
1202- Kernel data structures such as dentries and inodes.
1203
1204- TCP socket buffers.
1205
1206The above list may expand in the future for better coverage.
1207
1208
1209Memory Interface Files
1210~~~~~~~~~~~~~~~~~~~~~~
1211
1212All memory amounts are in bytes.  If a value which is not aligned to
1213PAGE_SIZE is written, the value may be rounded up to the closest
1214PAGE_SIZE multiple when read back.
1215
1216  memory.current
1217	A read-only single value file which exists on non-root
1218	cgroups.
1219
1220	The total amount of memory currently being used by the cgroup
1221	and its descendants.
1222
1223  memory.min
1224	A read-write single value file which exists on non-root
1225	cgroups.  The default is "0".
1226
1227	Hard memory protection.  If the memory usage of a cgroup
1228	is within its effective min boundary, the cgroup's memory
1229	won't be reclaimed under any conditions. If there is no
1230	unprotected reclaimable memory available, OOM killer
1231	is invoked. Above the effective min boundary (or
1232	effective low boundary if it is higher), pages are reclaimed
1233	proportionally to the overage, reducing reclaim pressure for
1234	smaller overages.
1235
1236	Effective min boundary is limited by memory.min values of
1237	all ancestor cgroups. If there is memory.min overcommitment
1238	(child cgroup or cgroups are requiring more protected memory
1239	than parent will allow), then each child cgroup will get
1240	the part of parent's protection proportional to its
1241	actual memory usage below memory.min.
1242
1243	Putting more memory than generally available under this
1244	protection is discouraged and may lead to constant OOMs.
1245
1246	If a memory cgroup is not populated with processes,
1247	its memory.min is ignored.
1248
1249  memory.low
1250	A read-write single value file which exists on non-root
1251	cgroups.  The default is "0".
1252
1253	Best-effort memory protection.  If the memory usage of a
1254	cgroup is within its effective low boundary, the cgroup's
1255	memory won't be reclaimed unless there is no reclaimable
1256	memory available in unprotected cgroups.
1257	Above the effective low	boundary (or
1258	effective min boundary if it is higher), pages are reclaimed
1259	proportionally to the overage, reducing reclaim pressure for
1260	smaller overages.
1261
1262	Effective low boundary is limited by memory.low values of
1263	all ancestor cgroups. If there is memory.low overcommitment
1264	(child cgroup or cgroups are requiring more protected memory
1265	than parent will allow), then each child cgroup will get
1266	the part of parent's protection proportional to its
1267	actual memory usage below memory.low.
1268
1269	Putting more memory than generally available under this
1270	protection is discouraged.
1271
1272  memory.high
1273	A read-write single value file which exists on non-root
1274	cgroups.  The default is "max".
1275
1276	Memory usage throttle limit.  If a cgroup's usage goes
1277	over the high boundary, the processes of the cgroup are
1278	throttled and put under heavy reclaim pressure.
1279
1280	Going over the high limit never invokes the OOM killer and
1281	under extreme conditions the limit may be breached. The high
1282	limit should be used in scenarios where an external process
1283	monitors the limited cgroup to alleviate heavy reclaim
1284	pressure.
1285
1286  memory.max
1287	A read-write single value file which exists on non-root
1288	cgroups.  The default is "max".
1289
1290	Memory usage hard limit.  This is the main mechanism to limit
1291	memory usage of a cgroup.  If a cgroup's memory usage reaches
1292	this limit and can't be reduced, the OOM killer is invoked in
1293	the cgroup. Under certain circumstances, the usage may go
1294	over the limit temporarily.
1295
1296	In default configuration regular 0-order allocations always
1297	succeed unless OOM killer chooses current task as a victim.
1298
1299	Some kinds of allocations don't invoke the OOM killer.
1300	Caller could retry them differently, return into userspace
1301	as -ENOMEM or silently ignore in cases like disk readahead.
1302
1303  memory.reclaim
1304	A write-only nested-keyed file which exists for all cgroups.
1305
1306	This is a simple interface to trigger memory reclaim in the
1307	target cgroup.
1308
1309	Example::
1310
1311	  echo "1G" > memory.reclaim
1312
1313	Please note that the kernel can over or under reclaim from
1314	the target cgroup. If less bytes are reclaimed than the
1315	specified amount, -EAGAIN is returned.
1316
1317	Please note that the proactive reclaim (triggered by this
1318	interface) is not meant to indicate memory pressure on the
1319	memory cgroup. Therefore socket memory balancing triggered by
1320	the memory reclaim normally is not exercised in this case.
1321	This means that the networking layer will not adapt based on
1322	reclaim induced by memory.reclaim.
1323
1324The following nested keys are defined.
1325
1326	  ==========            ================================
1327	  swappiness            Swappiness value to reclaim with
1328	  ==========            ================================
1329
1330	Specifying a swappiness value instructs the kernel to perform
1331	the reclaim with that swappiness value. Note that this has the
1332	same semantics as vm.swappiness applied to memcg reclaim with
1333	all the existing limitations and potential future extensions.
1334
1335  memory.peak
1336	A read-write single value file which exists on non-root cgroups.
1337
1338	The max memory usage recorded for the cgroup and its descendants since
1339	either the creation of the cgroup or the most recent reset for that FD.
1340
1341	A write of any non-empty string to this file resets it to the
1342	current memory usage for subsequent reads through the same
1343	file descriptor.
1344
1345  memory.oom.group
1346	A read-write single value file which exists on non-root
1347	cgroups.  The default value is "0".
1348
1349	Determines whether the cgroup should be treated as
1350	an indivisible workload by the OOM killer. If set,
1351	all tasks belonging to the cgroup or to its descendants
1352	(if the memory cgroup is not a leaf cgroup) are killed
1353	together or not at all. This can be used to avoid
1354	partial kills to guarantee workload integrity.
1355
1356	Tasks with the OOM protection (oom_score_adj set to -1000)
1357	are treated as an exception and are never killed.
1358
1359	If the OOM killer is invoked in a cgroup, it's not going
1360	to kill any tasks outside of this cgroup, regardless
1361	memory.oom.group values of ancestor cgroups.
1362
1363  memory.events
1364	A read-only flat-keyed file which exists on non-root cgroups.
1365	The following entries are defined.  Unless specified
1366	otherwise, a value change in this file generates a file
1367	modified event.
1368
1369	Note that all fields in this file are hierarchical and the
1370	file modified event can be generated due to an event down the
1371	hierarchy. For the local events at the cgroup level see
1372	memory.events.local.
1373
1374	  low
1375		The number of times the cgroup is reclaimed due to
1376		high memory pressure even though its usage is under
1377		the low boundary.  This usually indicates that the low
1378		boundary is over-committed.
1379
1380	  high
1381		The number of times processes of the cgroup are
1382		throttled and routed to perform direct memory reclaim
1383		because the high memory boundary was exceeded.  For a
1384		cgroup whose memory usage is capped by the high limit
1385		rather than global memory pressure, this event's
1386		occurrences are expected.
1387
1388	  max
1389		The number of times the cgroup's memory usage was
1390		about to go over the max boundary.  If direct reclaim
1391		fails to bring it down, the cgroup goes to OOM state.
1392
1393	  oom
1394		The number of time the cgroup's memory usage was
1395		reached the limit and allocation was about to fail.
1396
1397		This event is not raised if the OOM killer is not
1398		considered as an option, e.g. for failed high-order
1399		allocations or if caller asked to not retry attempts.
1400
1401	  oom_kill
1402		The number of processes belonging to this cgroup
1403		killed by any kind of OOM killer.
1404
1405          oom_group_kill
1406                The number of times a group OOM has occurred.
1407
1408  memory.events.local
1409	Similar to memory.events but the fields in the file are local
1410	to the cgroup i.e. not hierarchical. The file modified event
1411	generated on this file reflects only the local events.
1412
1413  memory.stat
1414	A read-only flat-keyed file which exists on non-root cgroups.
1415
1416	This breaks down the cgroup's memory footprint into different
1417	types of memory, type-specific details, and other information
1418	on the state and past events of the memory management system.
1419
1420	All memory amounts are in bytes.
1421
1422	The entries are ordered to be human readable, and new entries
1423	can show up in the middle. Don't rely on items remaining in a
1424	fixed position; use the keys to look up specific values!
1425
1426	If the entry has no per-node counter (or not show in the
1427	memory.numa_stat). We use 'npn' (non-per-node) as the tag
1428	to indicate that it will not show in the memory.numa_stat.
1429
1430	  anon
1431		Amount of memory used in anonymous mappings such as
1432		brk(), sbrk(), and mmap(MAP_ANONYMOUS)
1433
1434	  file
1435		Amount of memory used to cache filesystem data,
1436		including tmpfs and shared memory.
1437
1438	  kernel (npn)
1439		Amount of total kernel memory, including
1440		(kernel_stack, pagetables, percpu, vmalloc, slab) in
1441		addition to other kernel memory use cases.
1442
1443	  kernel_stack
1444		Amount of memory allocated to kernel stacks.
1445
1446	  pagetables
1447                Amount of memory allocated for page tables.
1448
1449	  sec_pagetables
1450		Amount of memory allocated for secondary page tables,
1451		this currently includes KVM mmu allocations on x86
1452		and arm64 and IOMMU page tables.
1453
1454	  percpu (npn)
1455		Amount of memory used for storing per-cpu kernel
1456		data structures.
1457
1458	  sock (npn)
1459		Amount of memory used in network transmission buffers
1460
1461	  vmalloc (npn)
1462		Amount of memory used for vmap backed memory.
1463
1464	  shmem
1465		Amount of cached filesystem data that is swap-backed,
1466		such as tmpfs, shm segments, shared anonymous mmap()s
1467
1468	  zswap
1469		Amount of memory consumed by the zswap compression backend.
1470
1471	  zswapped
1472		Amount of application memory swapped out to zswap.
1473
1474	  file_mapped
1475		Amount of cached filesystem data mapped with mmap()
1476
1477	  file_dirty
1478		Amount of cached filesystem data that was modified but
1479		not yet written back to disk
1480
1481	  file_writeback
1482		Amount of cached filesystem data that was modified and
1483		is currently being written back to disk
1484
1485	  swapcached
1486		Amount of swap cached in memory. The swapcache is accounted
1487		against both memory and swap usage.
1488
1489	  anon_thp
1490		Amount of memory used in anonymous mappings backed by
1491		transparent hugepages
1492
1493	  file_thp
1494		Amount of cached filesystem data backed by transparent
1495		hugepages
1496
1497	  shmem_thp
1498		Amount of shm, tmpfs, shared anonymous mmap()s backed by
1499		transparent hugepages
1500
1501	  inactive_anon, active_anon, inactive_file, active_file, unevictable
1502		Amount of memory, swap-backed and filesystem-backed,
1503		on the internal memory management lists used by the
1504		page reclaim algorithm.
1505
1506		As these represent internal list state (eg. shmem pages are on anon
1507		memory management lists), inactive_foo + active_foo may not be equal to
1508		the value for the foo counter, since the foo counter is type-based, not
1509		list-based.
1510
1511	  slab_reclaimable
1512		Part of "slab" that might be reclaimed, such as
1513		dentries and inodes.
1514
1515	  slab_unreclaimable
1516		Part of "slab" that cannot be reclaimed on memory
1517		pressure.
1518
1519	  slab (npn)
1520		Amount of memory used for storing in-kernel data
1521		structures.
1522
1523	  workingset_refault_anon
1524		Number of refaults of previously evicted anonymous pages.
1525
1526	  workingset_refault_file
1527		Number of refaults of previously evicted file pages.
1528
1529	  workingset_activate_anon
1530		Number of refaulted anonymous pages that were immediately
1531		activated.
1532
1533	  workingset_activate_file
1534		Number of refaulted file pages that were immediately activated.
1535
1536	  workingset_restore_anon
1537		Number of restored anonymous pages which have been detected as
1538		an active workingset before they got reclaimed.
1539
1540	  workingset_restore_file
1541		Number of restored file pages which have been detected as an
1542		active workingset before they got reclaimed.
1543
1544	  workingset_nodereclaim
1545		Number of times a shadow node has been reclaimed
1546
1547	  pgscan (npn)
1548		Amount of scanned pages (in an inactive LRU list)
1549
1550	  pgsteal (npn)
1551		Amount of reclaimed pages
1552
1553	  pgscan_kswapd (npn)
1554		Amount of scanned pages by kswapd (in an inactive LRU list)
1555
1556	  pgscan_direct (npn)
1557		Amount of scanned pages directly  (in an inactive LRU list)
1558
1559	  pgscan_khugepaged (npn)
1560		Amount of scanned pages by khugepaged  (in an inactive LRU list)
1561
1562	  pgsteal_kswapd (npn)
1563		Amount of reclaimed pages by kswapd
1564
1565	  pgsteal_direct (npn)
1566		Amount of reclaimed pages directly
1567
1568	  pgsteal_khugepaged (npn)
1569		Amount of reclaimed pages by khugepaged
1570
1571	  pgfault (npn)
1572		Total number of page faults incurred
1573
1574	  pgmajfault (npn)
1575		Number of major page faults incurred
1576
1577	  pgrefill (npn)
1578		Amount of scanned pages (in an active LRU list)
1579
1580	  pgactivate (npn)
1581		Amount of pages moved to the active LRU list
1582
1583	  pgdeactivate (npn)
1584		Amount of pages moved to the inactive LRU list
1585
1586	  pglazyfree (npn)
1587		Amount of pages postponed to be freed under memory pressure
1588
1589	  pglazyfreed (npn)
1590		Amount of reclaimed lazyfree pages
1591
1592	  zswpin
1593		Number of pages moved in to memory from zswap.
1594
1595	  zswpout
1596		Number of pages moved out of memory to zswap.
1597
1598	  zswpwb
1599		Number of pages written from zswap to swap.
1600
1601	  thp_fault_alloc (npn)
1602		Number of transparent hugepages which were allocated to satisfy
1603		a page fault. This counter is not present when CONFIG_TRANSPARENT_HUGEPAGE
1604                is not set.
1605
1606	  thp_collapse_alloc (npn)
1607		Number of transparent hugepages which were allocated to allow
1608		collapsing an existing range of pages. This counter is not
1609		present when CONFIG_TRANSPARENT_HUGEPAGE is not set.
1610
1611	  thp_swpout (npn)
1612		Number of transparent hugepages which are swapout in one piece
1613		without splitting.
1614
1615	  thp_swpout_fallback (npn)
1616		Number of transparent hugepages which were split before swapout.
1617		Usually because failed to allocate some continuous swap space
1618		for the huge page.
1619
1620	  numa_pages_migrated (npn)
1621		Number of pages migrated by NUMA balancing.
1622
1623	  numa_pte_updates (npn)
1624		Number of pages whose page table entries are modified by
1625		NUMA balancing to produce NUMA hinting faults on access.
1626
1627	  numa_hint_faults (npn)
1628		Number of NUMA hinting faults.
1629
1630	  pgdemote_kswapd
1631		Number of pages demoted by kswapd.
1632
1633	  pgdemote_direct
1634		Number of pages demoted directly.
1635
1636	  pgdemote_khugepaged
1637		Number of pages demoted by khugepaged.
1638
1639  memory.numa_stat
1640	A read-only nested-keyed file which exists on non-root cgroups.
1641
1642	This breaks down the cgroup's memory footprint into different
1643	types of memory, type-specific details, and other information
1644	per node on the state of the memory management system.
1645
1646	This is useful for providing visibility into the NUMA locality
1647	information within an memcg since the pages are allowed to be
1648	allocated from any physical node. One of the use case is evaluating
1649	application performance by combining this information with the
1650	application's CPU allocation.
1651
1652	All memory amounts are in bytes.
1653
1654	The output format of memory.numa_stat is::
1655
1656	  type N0=<bytes in node 0> N1=<bytes in node 1> ...
1657
1658	The entries are ordered to be human readable, and new entries
1659	can show up in the middle. Don't rely on items remaining in a
1660	fixed position; use the keys to look up specific values!
1661
1662	The entries can refer to the memory.stat.
1663
1664  memory.swap.current
1665	A read-only single value file which exists on non-root
1666	cgroups.
1667
1668	The total amount of swap currently being used by the cgroup
1669	and its descendants.
1670
1671  memory.swap.high
1672	A read-write single value file which exists on non-root
1673	cgroups.  The default is "max".
1674
1675	Swap usage throttle limit.  If a cgroup's swap usage exceeds
1676	this limit, all its further allocations will be throttled to
1677	allow userspace to implement custom out-of-memory procedures.
1678
1679	This limit marks a point of no return for the cgroup. It is NOT
1680	designed to manage the amount of swapping a workload does
1681	during regular operation. Compare to memory.swap.max, which
1682	prohibits swapping past a set amount, but lets the cgroup
1683	continue unimpeded as long as other memory can be reclaimed.
1684
1685	Healthy workloads are not expected to reach this limit.
1686
1687  memory.swap.peak
1688	A read-write single value file which exists on non-root cgroups.
1689
1690	The max swap usage recorded for the cgroup and its descendants since
1691	the creation of the cgroup or the most recent reset for that FD.
1692
1693	A write of any non-empty string to this file resets it to the
1694	current memory usage for subsequent reads through the same
1695	file descriptor.
1696
1697  memory.swap.max
1698	A read-write single value file which exists on non-root
1699	cgroups.  The default is "max".
1700
1701	Swap usage hard limit.  If a cgroup's swap usage reaches this
1702	limit, anonymous memory of the cgroup will not be swapped out.
1703
1704  memory.swap.events
1705	A read-only flat-keyed file which exists on non-root cgroups.
1706	The following entries are defined.  Unless specified
1707	otherwise, a value change in this file generates a file
1708	modified event.
1709
1710	  high
1711		The number of times the cgroup's swap usage was over
1712		the high threshold.
1713
1714	  max
1715		The number of times the cgroup's swap usage was about
1716		to go over the max boundary and swap allocation
1717		failed.
1718
1719	  fail
1720		The number of times swap allocation failed either
1721		because of running out of swap system-wide or max
1722		limit.
1723
1724	When reduced under the current usage, the existing swap
1725	entries are reclaimed gradually and the swap usage may stay
1726	higher than the limit for an extended period of time.  This
1727	reduces the impact on the workload and memory management.
1728
1729  memory.zswap.current
1730	A read-only single value file which exists on non-root
1731	cgroups.
1732
1733	The total amount of memory consumed by the zswap compression
1734	backend.
1735
1736  memory.zswap.max
1737	A read-write single value file which exists on non-root
1738	cgroups.  The default is "max".
1739
1740	Zswap usage hard limit. If a cgroup's zswap pool reaches this
1741	limit, it will refuse to take any more stores before existing
1742	entries fault back in or are written out to disk.
1743
1744  memory.zswap.writeback
1745	A read-write single value file. The default value is "1". The
1746	initial value of the root cgroup is 1, and when a new cgroup is
1747	created, it inherits the current value of its parent.
1748
1749	When this is set to 0, all swapping attempts to swapping devices
1750	are disabled. This included both zswap writebacks, and swapping due
1751	to zswap store failures. If the zswap store failures are recurring
1752	(for e.g if the pages are incompressible), users can observe
1753	reclaim inefficiency after disabling writeback (because the same
1754	pages might be rejected again and again).
1755
1756	Note that this is subtly different from setting memory.swap.max to
1757	0, as it still allows for pages to be written to the zswap pool.
1758	This setting has no effect if zswap is disabled, and swapping
1759	is allowed unless memory.swap.max is set to 0.
1760
1761  memory.pressure
1762	A read-only nested-keyed file.
1763
1764	Shows pressure stall information for memory. See
1765	:ref:`Documentation/accounting/psi.rst <psi>` for details.
1766
1767
1768Usage Guidelines
1769~~~~~~~~~~~~~~~~
1770
1771"memory.high" is the main mechanism to control memory usage.
1772Over-committing on high limit (sum of high limits > available memory)
1773and letting global memory pressure to distribute memory according to
1774usage is a viable strategy.
1775
1776Because breach of the high limit doesn't trigger the OOM killer but
1777throttles the offending cgroup, a management agent has ample
1778opportunities to monitor and take appropriate actions such as granting
1779more memory or terminating the workload.
1780
1781Determining whether a cgroup has enough memory is not trivial as
1782memory usage doesn't indicate whether the workload can benefit from
1783more memory.  For example, a workload which writes data received from
1784network to a file can use all available memory but can also operate as
1785performant with a small amount of memory.  A measure of memory
1786pressure - how much the workload is being impacted due to lack of
1787memory - is necessary to determine whether a workload needs more
1788memory; unfortunately, memory pressure monitoring mechanism isn't
1789implemented yet.
1790
1791
1792Memory Ownership
1793~~~~~~~~~~~~~~~~
1794
1795A memory area is charged to the cgroup which instantiated it and stays
1796charged to the cgroup until the area is released.  Migrating a process
1797to a different cgroup doesn't move the memory usages that it
1798instantiated while in the previous cgroup to the new cgroup.
1799
1800A memory area may be used by processes belonging to different cgroups.
1801To which cgroup the area will be charged is in-deterministic; however,
1802over time, the memory area is likely to end up in a cgroup which has
1803enough memory allowance to avoid high reclaim pressure.
1804
1805If a cgroup sweeps a considerable amount of memory which is expected
1806to be accessed repeatedly by other cgroups, it may make sense to use
1807POSIX_FADV_DONTNEED to relinquish the ownership of memory areas
1808belonging to the affected files to ensure correct memory ownership.
1809
1810
1811IO
1812--
1813
1814The "io" controller regulates the distribution of IO resources.  This
1815controller implements both weight based and absolute bandwidth or IOPS
1816limit distribution; however, weight based distribution is available
1817only if cfq-iosched is in use and neither scheme is available for
1818blk-mq devices.
1819
1820
1821IO Interface Files
1822~~~~~~~~~~~~~~~~~~
1823
1824  io.stat
1825	A read-only nested-keyed file.
1826
1827	Lines are keyed by $MAJ:$MIN device numbers and not ordered.
1828	The following nested keys are defined.
1829
1830	  ======	=====================
1831	  rbytes	Bytes read
1832	  wbytes	Bytes written
1833	  rios		Number of read IOs
1834	  wios		Number of write IOs
1835	  dbytes	Bytes discarded
1836	  dios		Number of discard IOs
1837	  ======	=====================
1838
1839	An example read output follows::
1840
1841	  8:16 rbytes=1459200 wbytes=314773504 rios=192 wios=353 dbytes=0 dios=0
1842	  8:0 rbytes=90430464 wbytes=299008000 rios=8950 wios=1252 dbytes=50331648 dios=3021
1843
1844  io.cost.qos
1845	A read-write nested-keyed file which exists only on the root
1846	cgroup.
1847
1848	This file configures the Quality of Service of the IO cost
1849	model based controller (CONFIG_BLK_CGROUP_IOCOST) which
1850	currently implements "io.weight" proportional control.  Lines
1851	are keyed by $MAJ:$MIN device numbers and not ordered.  The
1852	line for a given device is populated on the first write for
1853	the device on "io.cost.qos" or "io.cost.model".  The following
1854	nested keys are defined.
1855
1856	  ======	=====================================
1857	  enable	Weight-based control enable
1858	  ctrl		"auto" or "user"
1859	  rpct		Read latency percentile    [0, 100]
1860	  rlat		Read latency threshold
1861	  wpct		Write latency percentile   [0, 100]
1862	  wlat		Write latency threshold
1863	  min		Minimum scaling percentage [1, 10000]
1864	  max		Maximum scaling percentage [1, 10000]
1865	  ======	=====================================
1866
1867	The controller is disabled by default and can be enabled by
1868	setting "enable" to 1.  "rpct" and "wpct" parameters default
1869	to zero and the controller uses internal device saturation
1870	state to adjust the overall IO rate between "min" and "max".
1871
1872	When a better control quality is needed, latency QoS
1873	parameters can be configured.  For example::
1874
1875	  8:16 enable=1 ctrl=auto rpct=95.00 rlat=75000 wpct=95.00 wlat=150000 min=50.00 max=150.0
1876
1877	shows that on sdb, the controller is enabled, will consider
1878	the device saturated if the 95th percentile of read completion
1879	latencies is above 75ms or write 150ms, and adjust the overall
1880	IO issue rate between 50% and 150% accordingly.
1881
1882	The lower the saturation point, the better the latency QoS at
1883	the cost of aggregate bandwidth.  The narrower the allowed
1884	adjustment range between "min" and "max", the more conformant
1885	to the cost model the IO behavior.  Note that the IO issue
1886	base rate may be far off from 100% and setting "min" and "max"
1887	blindly can lead to a significant loss of device capacity or
1888	control quality.  "min" and "max" are useful for regulating
1889	devices which show wide temporary behavior changes - e.g. a
1890	ssd which accepts writes at the line speed for a while and
1891	then completely stalls for multiple seconds.
1892
1893	When "ctrl" is "auto", the parameters are controlled by the
1894	kernel and may change automatically.  Setting "ctrl" to "user"
1895	or setting any of the percentile and latency parameters puts
1896	it into "user" mode and disables the automatic changes.  The
1897	automatic mode can be restored by setting "ctrl" to "auto".
1898
1899  io.cost.model
1900	A read-write nested-keyed file which exists only on the root
1901	cgroup.
1902
1903	This file configures the cost model of the IO cost model based
1904	controller (CONFIG_BLK_CGROUP_IOCOST) which currently
1905	implements "io.weight" proportional control.  Lines are keyed
1906	by $MAJ:$MIN device numbers and not ordered.  The line for a
1907	given device is populated on the first write for the device on
1908	"io.cost.qos" or "io.cost.model".  The following nested keys
1909	are defined.
1910
1911	  =====		================================
1912	  ctrl		"auto" or "user"
1913	  model		The cost model in use - "linear"
1914	  =====		================================
1915
1916	When "ctrl" is "auto", the kernel may change all parameters
1917	dynamically.  When "ctrl" is set to "user" or any other
1918	parameters are written to, "ctrl" become "user" and the
1919	automatic changes are disabled.
1920
1921	When "model" is "linear", the following model parameters are
1922	defined.
1923
1924	  =============	========================================
1925	  [r|w]bps	The maximum sequential IO throughput
1926	  [r|w]seqiops	The maximum 4k sequential IOs per second
1927	  [r|w]randiops	The maximum 4k random IOs per second
1928	  =============	========================================
1929
1930	From the above, the builtin linear model determines the base
1931	costs of a sequential and random IO and the cost coefficient
1932	for the IO size.  While simple, this model can cover most
1933	common device classes acceptably.
1934
1935	The IO cost model isn't expected to be accurate in absolute
1936	sense and is scaled to the device behavior dynamically.
1937
1938	If needed, tools/cgroup/iocost_coef_gen.py can be used to
1939	generate device-specific coefficients.
1940
1941  io.weight
1942	A read-write flat-keyed file which exists on non-root cgroups.
1943	The default is "default 100".
1944
1945	The first line is the default weight applied to devices
1946	without specific override.  The rest are overrides keyed by
1947	$MAJ:$MIN device numbers and not ordered.  The weights are in
1948	the range [1, 10000] and specifies the relative amount IO time
1949	the cgroup can use in relation to its siblings.
1950
1951	The default weight can be updated by writing either "default
1952	$WEIGHT" or simply "$WEIGHT".  Overrides can be set by writing
1953	"$MAJ:$MIN $WEIGHT" and unset by writing "$MAJ:$MIN default".
1954
1955	An example read output follows::
1956
1957	  default 100
1958	  8:16 200
1959	  8:0 50
1960
1961  io.max
1962	A read-write nested-keyed file which exists on non-root
1963	cgroups.
1964
1965	BPS and IOPS based IO limit.  Lines are keyed by $MAJ:$MIN
1966	device numbers and not ordered.  The following nested keys are
1967	defined.
1968
1969	  =====		==================================
1970	  rbps		Max read bytes per second
1971	  wbps		Max write bytes per second
1972	  riops		Max read IO operations per second
1973	  wiops		Max write IO operations per second
1974	  =====		==================================
1975
1976	When writing, any number of nested key-value pairs can be
1977	specified in any order.  "max" can be specified as the value
1978	to remove a specific limit.  If the same key is specified
1979	multiple times, the outcome is undefined.
1980
1981	BPS and IOPS are measured in each IO direction and IOs are
1982	delayed if limit is reached.  Temporary bursts are allowed.
1983
1984	Setting read limit at 2M BPS and write at 120 IOPS for 8:16::
1985
1986	  echo "8:16 rbps=2097152 wiops=120" > io.max
1987
1988	Reading returns the following::
1989
1990	  8:16 rbps=2097152 wbps=max riops=max wiops=120
1991
1992	Write IOPS limit can be removed by writing the following::
1993
1994	  echo "8:16 wiops=max" > io.max
1995
1996	Reading now returns the following::
1997
1998	  8:16 rbps=2097152 wbps=max riops=max wiops=max
1999
2000  io.pressure
2001	A read-only nested-keyed file.
2002
2003	Shows pressure stall information for IO. See
2004	:ref:`Documentation/accounting/psi.rst <psi>` for details.
2005
2006
2007Writeback
2008~~~~~~~~~
2009
2010Page cache is dirtied through buffered writes and shared mmaps and
2011written asynchronously to the backing filesystem by the writeback
2012mechanism.  Writeback sits between the memory and IO domains and
2013regulates the proportion of dirty memory by balancing dirtying and
2014write IOs.
2015
2016The io controller, in conjunction with the memory controller,
2017implements control of page cache writeback IOs.  The memory controller
2018defines the memory domain that dirty memory ratio is calculated and
2019maintained for and the io controller defines the io domain which
2020writes out dirty pages for the memory domain.  Both system-wide and
2021per-cgroup dirty memory states are examined and the more restrictive
2022of the two is enforced.
2023
2024cgroup writeback requires explicit support from the underlying
2025filesystem.  Currently, cgroup writeback is implemented on ext2, ext4,
2026btrfs, f2fs, and xfs.  On other filesystems, all writeback IOs are
2027attributed to the root cgroup.
2028
2029There are inherent differences in memory and writeback management
2030which affects how cgroup ownership is tracked.  Memory is tracked per
2031page while writeback per inode.  For the purpose of writeback, an
2032inode is assigned to a cgroup and all IO requests to write dirty pages
2033from the inode are attributed to that cgroup.
2034
2035As cgroup ownership for memory is tracked per page, there can be pages
2036which are associated with different cgroups than the one the inode is
2037associated with.  These are called foreign pages.  The writeback
2038constantly keeps track of foreign pages and, if a particular foreign
2039cgroup becomes the majority over a certain period of time, switches
2040the ownership of the inode to that cgroup.
2041
2042While this model is enough for most use cases where a given inode is
2043mostly dirtied by a single cgroup even when the main writing cgroup
2044changes over time, use cases where multiple cgroups write to a single
2045inode simultaneously are not supported well.  In such circumstances, a
2046significant portion of IOs are likely to be attributed incorrectly.
2047As memory controller assigns page ownership on the first use and
2048doesn't update it until the page is released, even if writeback
2049strictly follows page ownership, multiple cgroups dirtying overlapping
2050areas wouldn't work as expected.  It's recommended to avoid such usage
2051patterns.
2052
2053The sysctl knobs which affect writeback behavior are applied to cgroup
2054writeback as follows.
2055
2056  vm.dirty_background_ratio, vm.dirty_ratio
2057	These ratios apply the same to cgroup writeback with the
2058	amount of available memory capped by limits imposed by the
2059	memory controller and system-wide clean memory.
2060
2061  vm.dirty_background_bytes, vm.dirty_bytes
2062	For cgroup writeback, this is calculated into ratio against
2063	total available memory and applied the same way as
2064	vm.dirty[_background]_ratio.
2065
2066
2067IO Latency
2068~~~~~~~~~~
2069
2070This is a cgroup v2 controller for IO workload protection.  You provide a group
2071with a latency target, and if the average latency exceeds that target the
2072controller will throttle any peers that have a lower latency target than the
2073protected workload.
2074
2075The limits are only applied at the peer level in the hierarchy.  This means that
2076in the diagram below, only groups A, B, and C will influence each other, and
2077groups D and F will influence each other.  Group G will influence nobody::
2078
2079			[root]
2080		/	   |		\
2081		A	   B		C
2082	       /  \        |
2083	      D    F	   G
2084
2085
2086So the ideal way to configure this is to set io.latency in groups A, B, and C.
2087Generally you do not want to set a value lower than the latency your device
2088supports.  Experiment to find the value that works best for your workload.
2089Start at higher than the expected latency for your device and watch the
2090avg_lat value in io.stat for your workload group to get an idea of the
2091latency you see during normal operation.  Use the avg_lat value as a basis for
2092your real setting, setting at 10-15% higher than the value in io.stat.
2093
2094How IO Latency Throttling Works
2095~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2096
2097io.latency is work conserving; so as long as everybody is meeting their latency
2098target the controller doesn't do anything.  Once a group starts missing its
2099target it begins throttling any peer group that has a higher target than itself.
2100This throttling takes 2 forms:
2101
2102- Queue depth throttling.  This is the number of outstanding IO's a group is
2103  allowed to have.  We will clamp down relatively quickly, starting at no limit
2104  and going all the way down to 1 IO at a time.
2105
2106- Artificial delay induction.  There are certain types of IO that cannot be
2107  throttled without possibly adversely affecting higher priority groups.  This
2108  includes swapping and metadata IO.  These types of IO are allowed to occur
2109  normally, however they are "charged" to the originating group.  If the
2110  originating group is being throttled you will see the use_delay and delay
2111  fields in io.stat increase.  The delay value is how many microseconds that are
2112  being added to any process that runs in this group.  Because this number can
2113  grow quite large if there is a lot of swapping or metadata IO occurring we
2114  limit the individual delay events to 1 second at a time.
2115
2116Once the victimized group starts meeting its latency target again it will start
2117unthrottling any peer groups that were throttled previously.  If the victimized
2118group simply stops doing IO the global counter will unthrottle appropriately.
2119
2120IO Latency Interface Files
2121~~~~~~~~~~~~~~~~~~~~~~~~~~
2122
2123  io.latency
2124	This takes a similar format as the other controllers.
2125
2126		"MAJOR:MINOR target=<target time in microseconds>"
2127
2128  io.stat
2129	If the controller is enabled you will see extra stats in io.stat in
2130	addition to the normal ones.
2131
2132	  depth
2133		This is the current queue depth for the group.
2134
2135	  avg_lat
2136		This is an exponential moving average with a decay rate of 1/exp
2137		bound by the sampling interval.  The decay rate interval can be
2138		calculated by multiplying the win value in io.stat by the
2139		corresponding number of samples based on the win value.
2140
2141	  win
2142		The sampling window size in milliseconds.  This is the minimum
2143		duration of time between evaluation events.  Windows only elapse
2144		with IO activity.  Idle periods extend the most recent window.
2145
2146IO Priority
2147~~~~~~~~~~~
2148
2149A single attribute controls the behavior of the I/O priority cgroup policy,
2150namely the io.prio.class attribute. The following values are accepted for
2151that attribute:
2152
2153  no-change
2154	Do not modify the I/O priority class.
2155
2156  promote-to-rt
2157	For requests that have a non-RT I/O priority class, change it into RT.
2158	Also change the priority level of these requests to 4. Do not modify
2159	the I/O priority of requests that have priority class RT.
2160
2161  restrict-to-be
2162	For requests that do not have an I/O priority class or that have I/O
2163	priority class RT, change it into BE. Also change the priority level
2164	of these requests to 0. Do not modify the I/O priority class of
2165	requests that have priority class IDLE.
2166
2167  idle
2168	Change the I/O priority class of all requests into IDLE, the lowest
2169	I/O priority class.
2170
2171  none-to-rt
2172	Deprecated. Just an alias for promote-to-rt.
2173
2174The following numerical values are associated with the I/O priority policies:
2175
2176+----------------+---+
2177| no-change      | 0 |
2178+----------------+---+
2179| promote-to-rt  | 1 |
2180+----------------+---+
2181| restrict-to-be | 2 |
2182+----------------+---+
2183| idle           | 3 |
2184+----------------+---+
2185
2186The numerical value that corresponds to each I/O priority class is as follows:
2187
2188+-------------------------------+---+
2189| IOPRIO_CLASS_NONE             | 0 |
2190+-------------------------------+---+
2191| IOPRIO_CLASS_RT (real-time)   | 1 |
2192+-------------------------------+---+
2193| IOPRIO_CLASS_BE (best effort) | 2 |
2194+-------------------------------+---+
2195| IOPRIO_CLASS_IDLE             | 3 |
2196+-------------------------------+---+
2197
2198The algorithm to set the I/O priority class for a request is as follows:
2199
2200- If I/O priority class policy is promote-to-rt, change the request I/O
2201  priority class to IOPRIO_CLASS_RT and change the request I/O priority
2202  level to 4.
2203- If I/O priority class policy is not promote-to-rt, translate the I/O priority
2204  class policy into a number, then change the request I/O priority class
2205  into the maximum of the I/O priority class policy number and the numerical
2206  I/O priority class.
2207
2208PID
2209---
2210
2211The process number controller is used to allow a cgroup to stop any
2212new tasks from being fork()'d or clone()'d after a specified limit is
2213reached.
2214
2215The number of tasks in a cgroup can be exhausted in ways which other
2216controllers cannot prevent, thus warranting its own controller.  For
2217example, a fork bomb is likely to exhaust the number of tasks before
2218hitting memory restrictions.
2219
2220Note that PIDs used in this controller refer to TIDs, process IDs as
2221used by the kernel.
2222
2223
2224PID Interface Files
2225~~~~~~~~~~~~~~~~~~~
2226
2227  pids.max
2228	A read-write single value file which exists on non-root
2229	cgroups.  The default is "max".
2230
2231	Hard limit of number of processes.
2232
2233  pids.current
2234	A read-only single value file which exists on non-root cgroups.
2235
2236	The number of processes currently in the cgroup and its
2237	descendants.
2238
2239  pids.peak
2240	A read-only single value file which exists on non-root cgroups.
2241
2242	The maximum value that the number of processes in the cgroup and its
2243	descendants has ever reached.
2244
2245  pids.events
2246	A read-only flat-keyed file which exists on non-root cgroups. Unless
2247	specified otherwise, a value change in this file generates a file
2248	modified event. The following entries are defined.
2249
2250	  max
2251		The number of times the cgroup's total number of processes hit the pids.max
2252		limit (see also pids_localevents).
2253
2254  pids.events.local
2255	Similar to pids.events but the fields in the file are local
2256	to the cgroup i.e. not hierarchical. The file modified event
2257	generated on this file reflects only the local events.
2258
2259Organisational operations are not blocked by cgroup policies, so it is
2260possible to have pids.current > pids.max.  This can be done by either
2261setting the limit to be smaller than pids.current, or attaching enough
2262processes to the cgroup such that pids.current is larger than
2263pids.max.  However, it is not possible to violate a cgroup PID policy
2264through fork() or clone(). These will return -EAGAIN if the creation
2265of a new process would cause a cgroup policy to be violated.
2266
2267
2268Cpuset
2269------
2270
2271The "cpuset" controller provides a mechanism for constraining
2272the CPU and memory node placement of tasks to only the resources
2273specified in the cpuset interface files in a task's current cgroup.
2274This is especially valuable on large NUMA systems where placing jobs
2275on properly sized subsets of the systems with careful processor and
2276memory placement to reduce cross-node memory access and contention
2277can improve overall system performance.
2278
2279The "cpuset" controller is hierarchical.  That means the controller
2280cannot use CPUs or memory nodes not allowed in its parent.
2281
2282
2283Cpuset Interface Files
2284~~~~~~~~~~~~~~~~~~~~~~
2285
2286  cpuset.cpus
2287	A read-write multiple values file which exists on non-root
2288	cpuset-enabled cgroups.
2289
2290	It lists the requested CPUs to be used by tasks within this
2291	cgroup.  The actual list of CPUs to be granted, however, is
2292	subjected to constraints imposed by its parent and can differ
2293	from the requested CPUs.
2294
2295	The CPU numbers are comma-separated numbers or ranges.
2296	For example::
2297
2298	  # cat cpuset.cpus
2299	  0-4,6,8-10
2300
2301	An empty value indicates that the cgroup is using the same
2302	setting as the nearest cgroup ancestor with a non-empty
2303	"cpuset.cpus" or all the available CPUs if none is found.
2304
2305	The value of "cpuset.cpus" stays constant until the next update
2306	and won't be affected by any CPU hotplug events.
2307
2308  cpuset.cpus.effective
2309	A read-only multiple values file which exists on all
2310	cpuset-enabled cgroups.
2311
2312	It lists the onlined CPUs that are actually granted to this
2313	cgroup by its parent.  These CPUs are allowed to be used by
2314	tasks within the current cgroup.
2315
2316	If "cpuset.cpus" is empty, the "cpuset.cpus.effective" file shows
2317	all the CPUs from the parent cgroup that can be available to
2318	be used by this cgroup.  Otherwise, it should be a subset of
2319	"cpuset.cpus" unless none of the CPUs listed in "cpuset.cpus"
2320	can be granted.  In this case, it will be treated just like an
2321	empty "cpuset.cpus".
2322
2323	Its value will be affected by CPU hotplug events.
2324
2325  cpuset.mems
2326	A read-write multiple values file which exists on non-root
2327	cpuset-enabled cgroups.
2328
2329	It lists the requested memory nodes to be used by tasks within
2330	this cgroup.  The actual list of memory nodes granted, however,
2331	is subjected to constraints imposed by its parent and can differ
2332	from the requested memory nodes.
2333
2334	The memory node numbers are comma-separated numbers or ranges.
2335	For example::
2336
2337	  # cat cpuset.mems
2338	  0-1,3
2339
2340	An empty value indicates that the cgroup is using the same
2341	setting as the nearest cgroup ancestor with a non-empty
2342	"cpuset.mems" or all the available memory nodes if none
2343	is found.
2344
2345	The value of "cpuset.mems" stays constant until the next update
2346	and won't be affected by any memory nodes hotplug events.
2347
2348	Setting a non-empty value to "cpuset.mems" causes memory of
2349	tasks within the cgroup to be migrated to the designated nodes if
2350	they are currently using memory outside of the designated nodes.
2351
2352	There is a cost for this memory migration.  The migration
2353	may not be complete and some memory pages may be left behind.
2354	So it is recommended that "cpuset.mems" should be set properly
2355	before spawning new tasks into the cpuset.  Even if there is
2356	a need to change "cpuset.mems" with active tasks, it shouldn't
2357	be done frequently.
2358
2359  cpuset.mems.effective
2360	A read-only multiple values file which exists on all
2361	cpuset-enabled cgroups.
2362
2363	It lists the onlined memory nodes that are actually granted to
2364	this cgroup by its parent. These memory nodes are allowed to
2365	be used by tasks within the current cgroup.
2366
2367	If "cpuset.mems" is empty, it shows all the memory nodes from the
2368	parent cgroup that will be available to be used by this cgroup.
2369	Otherwise, it should be a subset of "cpuset.mems" unless none of
2370	the memory nodes listed in "cpuset.mems" can be granted.  In this
2371	case, it will be treated just like an empty "cpuset.mems".
2372
2373	Its value will be affected by memory nodes hotplug events.
2374
2375  cpuset.cpus.exclusive
2376	A read-write multiple values file which exists on non-root
2377	cpuset-enabled cgroups.
2378
2379	It lists all the exclusive CPUs that are allowed to be used
2380	to create a new cpuset partition.  Its value is not used
2381	unless the cgroup becomes a valid partition root.  See the
2382	"cpuset.cpus.partition" section below for a description of what
2383	a cpuset partition is.
2384
2385	When the cgroup becomes a partition root, the actual exclusive
2386	CPUs that are allocated to that partition are listed in
2387	"cpuset.cpus.exclusive.effective" which may be different
2388	from "cpuset.cpus.exclusive".  If "cpuset.cpus.exclusive"
2389	has previously been set, "cpuset.cpus.exclusive.effective"
2390	is always a subset of it.
2391
2392	Users can manually set it to a value that is different from
2393	"cpuset.cpus".	One constraint in setting it is that the list of
2394	CPUs must be exclusive with respect to "cpuset.cpus.exclusive"
2395	of its sibling.  If "cpuset.cpus.exclusive" of a sibling cgroup
2396	isn't set, its "cpuset.cpus" value, if set, cannot be a subset
2397	of it to leave at least one CPU available when the exclusive
2398	CPUs are taken away.
2399
2400	For a parent cgroup, any one of its exclusive CPUs can only
2401	be distributed to at most one of its child cgroups.  Having an
2402	exclusive CPU appearing in two or more of its child cgroups is
2403	not allowed (the exclusivity rule).  A value that violates the
2404	exclusivity rule will be rejected with a write error.
2405
2406	The root cgroup is a partition root and all its available CPUs
2407	are in its exclusive CPU set.
2408
2409  cpuset.cpus.exclusive.effective
2410	A read-only multiple values file which exists on all non-root
2411	cpuset-enabled cgroups.
2412
2413	This file shows the effective set of exclusive CPUs that
2414	can be used to create a partition root.  The content
2415	of this file will always be a subset of its parent's
2416	"cpuset.cpus.exclusive.effective" if its parent is not the root
2417	cgroup.  It will also be a subset of "cpuset.cpus.exclusive"
2418	if it is set.  If "cpuset.cpus.exclusive" is not set, it is
2419	treated to have an implicit value of "cpuset.cpus" in the
2420	formation of local partition.
2421
2422  cpuset.cpus.isolated
2423	A read-only and root cgroup only multiple values file.
2424
2425	This file shows the set of all isolated CPUs used in existing
2426	isolated partitions. It will be empty if no isolated partition
2427	is created.
2428
2429  cpuset.cpus.partition
2430	A read-write single value file which exists on non-root
2431	cpuset-enabled cgroups.  This flag is owned by the parent cgroup
2432	and is not delegatable.
2433
2434	It accepts only the following input values when written to.
2435
2436	  ==========	=====================================
2437	  "member"	Non-root member of a partition
2438	  "root"	Partition root
2439	  "isolated"	Partition root without load balancing
2440	  ==========	=====================================
2441
2442	A cpuset partition is a collection of cpuset-enabled cgroups with
2443	a partition root at the top of the hierarchy and its descendants
2444	except those that are separate partition roots themselves and
2445	their descendants.  A partition has exclusive access to the
2446	set of exclusive CPUs allocated to it.	Other cgroups outside
2447	of that partition cannot use any CPUs in that set.
2448
2449	There are two types of partitions - local and remote.  A local
2450	partition is one whose parent cgroup is also a valid partition
2451	root.  A remote partition is one whose parent cgroup is not a
2452	valid partition root itself.  Writing to "cpuset.cpus.exclusive"
2453	is optional for the creation of a local partition as its
2454	"cpuset.cpus.exclusive" file will assume an implicit value that
2455	is the same as "cpuset.cpus" if it is not set.	Writing the
2456	proper "cpuset.cpus.exclusive" values down the cgroup hierarchy
2457	before the target partition root is mandatory for the creation
2458	of a remote partition.
2459
2460	Currently, a remote partition cannot be created under a local
2461	partition.  All the ancestors of a remote partition root except
2462	the root cgroup cannot be a partition root.
2463
2464	The root cgroup is always a partition root and its state cannot
2465	be changed.  All other non-root cgroups start out as "member".
2466
2467	When set to "root", the current cgroup is the root of a new
2468	partition or scheduling domain.  The set of exclusive CPUs is
2469	determined by the value of its "cpuset.cpus.exclusive.effective".
2470
2471	When set to "isolated", the CPUs in that partition will be in
2472	an isolated state without any load balancing from the scheduler
2473	and excluded from the unbound workqueues.  Tasks placed in such
2474	a partition with multiple CPUs should be carefully distributed
2475	and bound to each of the individual CPUs for optimal performance.
2476
2477	A partition root ("root" or "isolated") can be in one of the
2478	two possible states - valid or invalid.  An invalid partition
2479	root is in a degraded state where some state information may
2480	be retained, but behaves more like a "member".
2481
2482	All possible state transitions among "member", "root" and
2483	"isolated" are allowed.
2484
2485	On read, the "cpuset.cpus.partition" file can show the following
2486	values.
2487
2488	  =============================	=====================================
2489	  "member"			Non-root member of a partition
2490	  "root"			Partition root
2491	  "isolated"			Partition root without load balancing
2492	  "root invalid (<reason>)"	Invalid partition root
2493	  "isolated invalid (<reason>)"	Invalid isolated partition root
2494	  =============================	=====================================
2495
2496	In the case of an invalid partition root, a descriptive string on
2497	why the partition is invalid is included within parentheses.
2498
2499	For a local partition root to be valid, the following conditions
2500	must be met.
2501
2502	1) The parent cgroup is a valid partition root.
2503	2) The "cpuset.cpus.exclusive.effective" file cannot be empty,
2504	   though it may contain offline CPUs.
2505	3) The "cpuset.cpus.effective" cannot be empty unless there is
2506	   no task associated with this partition.
2507
2508	For a remote partition root to be valid, all the above conditions
2509	except the first one must be met.
2510
2511	External events like hotplug or changes to "cpuset.cpus" or
2512	"cpuset.cpus.exclusive" can cause a valid partition root to
2513	become invalid and vice versa.	Note that a task cannot be
2514	moved to a cgroup with empty "cpuset.cpus.effective".
2515
2516	A valid non-root parent partition may distribute out all its CPUs
2517	to its child local partitions when there is no task associated
2518	with it.
2519
2520	Care must be taken to change a valid partition root to "member"
2521	as all its child local partitions, if present, will become
2522	invalid causing disruption to tasks running in those child
2523	partitions. These inactivated partitions could be recovered if
2524	their parent is switched back to a partition root with a proper
2525	value in "cpuset.cpus" or "cpuset.cpus.exclusive".
2526
2527	Poll and inotify events are triggered whenever the state of
2528	"cpuset.cpus.partition" changes.  That includes changes caused
2529	by write to "cpuset.cpus.partition", cpu hotplug or other
2530	changes that modify the validity status of the partition.
2531	This will allow user space agents to monitor unexpected changes
2532	to "cpuset.cpus.partition" without the need to do continuous
2533	polling.
2534
2535	A user can pre-configure certain CPUs to an isolated state
2536	with load balancing disabled at boot time with the "isolcpus"
2537	kernel boot command line option.  If those CPUs are to be put
2538	into a partition, they have to be used in an isolated partition.
2539
2540
2541Device controller
2542-----------------
2543
2544Device controller manages access to device files. It includes both
2545creation of new device files (using mknod), and access to the
2546existing device files.
2547
2548Cgroup v2 device controller has no interface files and is implemented
2549on top of cgroup BPF. To control access to device files, a user may
2550create bpf programs of type BPF_PROG_TYPE_CGROUP_DEVICE and attach
2551them to cgroups with BPF_CGROUP_DEVICE flag. On an attempt to access a
2552device file, corresponding BPF programs will be executed, and depending
2553on the return value the attempt will succeed or fail with -EPERM.
2554
2555A BPF_PROG_TYPE_CGROUP_DEVICE program takes a pointer to the
2556bpf_cgroup_dev_ctx structure, which describes the device access attempt:
2557access type (mknod/read/write) and device (type, major and minor numbers).
2558If the program returns 0, the attempt fails with -EPERM, otherwise it
2559succeeds.
2560
2561An example of BPF_PROG_TYPE_CGROUP_DEVICE program may be found in
2562tools/testing/selftests/bpf/progs/dev_cgroup.c in the kernel source tree.
2563
2564
2565RDMA
2566----
2567
2568The "rdma" controller regulates the distribution and accounting of
2569RDMA resources.
2570
2571RDMA Interface Files
2572~~~~~~~~~~~~~~~~~~~~
2573
2574  rdma.max
2575	A readwrite nested-keyed file that exists for all the cgroups
2576	except root that describes current configured resource limit
2577	for a RDMA/IB device.
2578
2579	Lines are keyed by device name and are not ordered.
2580	Each line contains space separated resource name and its configured
2581	limit that can be distributed.
2582
2583	The following nested keys are defined.
2584
2585	  ==========	=============================
2586	  hca_handle	Maximum number of HCA Handles
2587	  hca_object 	Maximum number of HCA Objects
2588	  ==========	=============================
2589
2590	An example for mlx4 and ocrdma device follows::
2591
2592	  mlx4_0 hca_handle=2 hca_object=2000
2593	  ocrdma1 hca_handle=3 hca_object=max
2594
2595  rdma.current
2596	A read-only file that describes current resource usage.
2597	It exists for all the cgroup except root.
2598
2599	An example for mlx4 and ocrdma device follows::
2600
2601	  mlx4_0 hca_handle=1 hca_object=20
2602	  ocrdma1 hca_handle=1 hca_object=23
2603
2604HugeTLB
2605-------
2606
2607The HugeTLB controller allows to limit the HugeTLB usage per control group and
2608enforces the controller limit during page fault.
2609
2610HugeTLB Interface Files
2611~~~~~~~~~~~~~~~~~~~~~~~
2612
2613  hugetlb.<hugepagesize>.current
2614	Show current usage for "hugepagesize" hugetlb.  It exists for all
2615	the cgroup except root.
2616
2617  hugetlb.<hugepagesize>.max
2618	Set/show the hard limit of "hugepagesize" hugetlb usage.
2619	The default value is "max".  It exists for all the cgroup except root.
2620
2621  hugetlb.<hugepagesize>.events
2622	A read-only flat-keyed file which exists on non-root cgroups.
2623
2624	  max
2625		The number of allocation failure due to HugeTLB limit
2626
2627  hugetlb.<hugepagesize>.events.local
2628	Similar to hugetlb.<hugepagesize>.events but the fields in the file
2629	are local to the cgroup i.e. not hierarchical. The file modified event
2630	generated on this file reflects only the local events.
2631
2632  hugetlb.<hugepagesize>.numa_stat
2633	Similar to memory.numa_stat, it shows the numa information of the
2634        hugetlb pages of <hugepagesize> in this cgroup.  Only active in
2635        use hugetlb pages are included.  The per-node values are in bytes.
2636
2637Misc
2638----
2639
2640The Miscellaneous cgroup provides the resource limiting and tracking
2641mechanism for the scalar resources which cannot be abstracted like the other
2642cgroup resources. Controller is enabled by the CONFIG_CGROUP_MISC config
2643option.
2644
2645A resource can be added to the controller via enum misc_res_type{} in the
2646include/linux/misc_cgroup.h file and the corresponding name via misc_res_name[]
2647in the kernel/cgroup/misc.c file. Provider of the resource must set its
2648capacity prior to using the resource by calling misc_cg_set_capacity().
2649
2650Once a capacity is set then the resource usage can be updated using charge and
2651uncharge APIs. All of the APIs to interact with misc controller are in
2652include/linux/misc_cgroup.h.
2653
2654Misc Interface Files
2655~~~~~~~~~~~~~~~~~~~~
2656
2657Miscellaneous controller provides 3 interface files. If two misc resources (res_a and res_b) are registered then:
2658
2659  misc.capacity
2660        A read-only flat-keyed file shown only in the root cgroup.  It shows
2661        miscellaneous scalar resources available on the platform along with
2662        their quantities::
2663
2664	  $ cat misc.capacity
2665	  res_a 50
2666	  res_b 10
2667
2668  misc.current
2669        A read-only flat-keyed file shown in the all cgroups.  It shows
2670        the current usage of the resources in the cgroup and its children.::
2671
2672	  $ cat misc.current
2673	  res_a 3
2674	  res_b 0
2675
2676  misc.peak
2677        A read-only flat-keyed file shown in all cgroups.  It shows the
2678        historical maximum usage of the resources in the cgroup and its
2679        children.::
2680
2681	  $ cat misc.peak
2682	  res_a 10
2683	  res_b 8
2684
2685  misc.max
2686        A read-write flat-keyed file shown in the non root cgroups. Allowed
2687        maximum usage of the resources in the cgroup and its children.::
2688
2689	  $ cat misc.max
2690	  res_a max
2691	  res_b 4
2692
2693	Limit can be set by::
2694
2695	  # echo res_a 1 > misc.max
2696
2697	Limit can be set to max by::
2698
2699	  # echo res_a max > misc.max
2700
2701        Limits can be set higher than the capacity value in the misc.capacity
2702        file.
2703
2704  misc.events
2705	A read-only flat-keyed file which exists on non-root cgroups. The
2706	following entries are defined. Unless specified otherwise, a value
2707	change in this file generates a file modified event. All fields in
2708	this file are hierarchical.
2709
2710	  max
2711		The number of times the cgroup's resource usage was
2712		about to go over the max boundary.
2713
2714  misc.events.local
2715        Similar to misc.events but the fields in the file are local to the
2716        cgroup i.e. not hierarchical. The file modified event generated on
2717        this file reflects only the local events.
2718
2719Migration and Ownership
2720~~~~~~~~~~~~~~~~~~~~~~~
2721
2722A miscellaneous scalar resource is charged to the cgroup in which it is used
2723first, and stays charged to that cgroup until that resource is freed. Migrating
2724a process to a different cgroup does not move the charge to the destination
2725cgroup where the process has moved.
2726
2727Others
2728------
2729
2730perf_event
2731~~~~~~~~~~
2732
2733perf_event controller, if not mounted on a legacy hierarchy, is
2734automatically enabled on the v2 hierarchy so that perf events can
2735always be filtered by cgroup v2 path.  The controller can still be
2736moved to a legacy hierarchy after v2 hierarchy is populated.
2737
2738
2739Non-normative information
2740-------------------------
2741
2742This section contains information that isn't considered to be a part of
2743the stable kernel API and so is subject to change.
2744
2745
2746CPU controller root cgroup process behaviour
2747~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2748
2749When distributing CPU cycles in the root cgroup each thread in this
2750cgroup is treated as if it was hosted in a separate child cgroup of the
2751root cgroup. This child cgroup weight is dependent on its thread nice
2752level.
2753
2754For details of this mapping see sched_prio_to_weight array in
2755kernel/sched/core.c file (values from this array should be scaled
2756appropriately so the neutral - nice 0 - value is 100 instead of 1024).
2757
2758
2759IO controller root cgroup process behaviour
2760~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
2761
2762Root cgroup processes are hosted in an implicit leaf child node.
2763When distributing IO resources this implicit child node is taken into
2764account as if it was a normal child cgroup of the root cgroup with a
2765weight value of 200.
2766
2767
2768Namespace
2769=========
2770
2771Basics
2772------
2773
2774cgroup namespace provides a mechanism to virtualize the view of the
2775"/proc/$PID/cgroup" file and cgroup mounts.  The CLONE_NEWCGROUP clone
2776flag can be used with clone(2) and unshare(2) to create a new cgroup
2777namespace.  The process running inside the cgroup namespace will have
2778its "/proc/$PID/cgroup" output restricted to cgroupns root.  The
2779cgroupns root is the cgroup of the process at the time of creation of
2780the cgroup namespace.
2781
2782Without cgroup namespace, the "/proc/$PID/cgroup" file shows the
2783complete path of the cgroup of a process.  In a container setup where
2784a set of cgroups and namespaces are intended to isolate processes the
2785"/proc/$PID/cgroup" file may leak potential system level information
2786to the isolated processes.  For example::
2787
2788  # cat /proc/self/cgroup
2789  0::/batchjobs/container_id1
2790
2791The path '/batchjobs/container_id1' can be considered as system-data
2792and undesirable to expose to the isolated processes.  cgroup namespace
2793can be used to restrict visibility of this path.  For example, before
2794creating a cgroup namespace, one would see::
2795
2796  # ls -l /proc/self/ns/cgroup
2797  lrwxrwxrwx 1 root root 0 2014-07-15 10:37 /proc/self/ns/cgroup -> cgroup:[4026531835]
2798  # cat /proc/self/cgroup
2799  0::/batchjobs/container_id1
2800
2801After unsharing a new namespace, the view changes::
2802
2803  # ls -l /proc/self/ns/cgroup
2804  lrwxrwxrwx 1 root root 0 2014-07-15 10:35 /proc/self/ns/cgroup -> cgroup:[4026532183]
2805  # cat /proc/self/cgroup
2806  0::/
2807
2808When some thread from a multi-threaded process unshares its cgroup
2809namespace, the new cgroupns gets applied to the entire process (all
2810the threads).  This is natural for the v2 hierarchy; however, for the
2811legacy hierarchies, this may be unexpected.
2812
2813A cgroup namespace is alive as long as there are processes inside or
2814mounts pinning it.  When the last usage goes away, the cgroup
2815namespace is destroyed.  The cgroupns root and the actual cgroups
2816remain.
2817
2818
2819The Root and Views
2820------------------
2821
2822The 'cgroupns root' for a cgroup namespace is the cgroup in which the
2823process calling unshare(2) is running.  For example, if a process in
2824/batchjobs/container_id1 cgroup calls unshare, cgroup
2825/batchjobs/container_id1 becomes the cgroupns root.  For the
2826init_cgroup_ns, this is the real root ('/') cgroup.
2827
2828The cgroupns root cgroup does not change even if the namespace creator
2829process later moves to a different cgroup::
2830
2831  # ~/unshare -c # unshare cgroupns in some cgroup
2832  # cat /proc/self/cgroup
2833  0::/
2834  # mkdir sub_cgrp_1
2835  # echo 0 > sub_cgrp_1/cgroup.procs
2836  # cat /proc/self/cgroup
2837  0::/sub_cgrp_1
2838
2839Each process gets its namespace-specific view of "/proc/$PID/cgroup"
2840
2841Processes running inside the cgroup namespace will be able to see
2842cgroup paths (in /proc/self/cgroup) only inside their root cgroup.
2843From within an unshared cgroupns::
2844
2845  # sleep 100000 &
2846  [1] 7353
2847  # echo 7353 > sub_cgrp_1/cgroup.procs
2848  # cat /proc/7353/cgroup
2849  0::/sub_cgrp_1
2850
2851From the initial cgroup namespace, the real cgroup path will be
2852visible::
2853
2854  $ cat /proc/7353/cgroup
2855  0::/batchjobs/container_id1/sub_cgrp_1
2856
2857From a sibling cgroup namespace (that is, a namespace rooted at a
2858different cgroup), the cgroup path relative to its own cgroup
2859namespace root will be shown.  For instance, if PID 7353's cgroup
2860namespace root is at '/batchjobs/container_id2', then it will see::
2861
2862  # cat /proc/7353/cgroup
2863  0::/../container_id2/sub_cgrp_1
2864
2865Note that the relative path always starts with '/' to indicate that
2866its relative to the cgroup namespace root of the caller.
2867
2868
2869Migration and setns(2)
2870----------------------
2871
2872Processes inside a cgroup namespace can move into and out of the
2873namespace root if they have proper access to external cgroups.  For
2874example, from inside a namespace with cgroupns root at
2875/batchjobs/container_id1, and assuming that the global hierarchy is
2876still accessible inside cgroupns::
2877
2878  # cat /proc/7353/cgroup
2879  0::/sub_cgrp_1
2880  # echo 7353 > batchjobs/container_id2/cgroup.procs
2881  # cat /proc/7353/cgroup
2882  0::/../container_id2
2883
2884Note that this kind of setup is not encouraged.  A task inside cgroup
2885namespace should only be exposed to its own cgroupns hierarchy.
2886
2887setns(2) to another cgroup namespace is allowed when:
2888
2889(a) the process has CAP_SYS_ADMIN against its current user namespace
2890(b) the process has CAP_SYS_ADMIN against the target cgroup
2891    namespace's userns
2892
2893No implicit cgroup changes happen with attaching to another cgroup
2894namespace.  It is expected that the someone moves the attaching
2895process under the target cgroup namespace root.
2896
2897
2898Interaction with Other Namespaces
2899---------------------------------
2900
2901Namespace specific cgroup hierarchy can be mounted by a process
2902running inside a non-init cgroup namespace::
2903
2904  # mount -t cgroup2 none $MOUNT_POINT
2905
2906This will mount the unified cgroup hierarchy with cgroupns root as the
2907filesystem root.  The process needs CAP_SYS_ADMIN against its user and
2908mount namespaces.
2909
2910The virtualization of /proc/self/cgroup file combined with restricting
2911the view of cgroup hierarchy by namespace-private cgroupfs mount
2912provides a properly isolated cgroup view inside the container.
2913
2914
2915Information on Kernel Programming
2916=================================
2917
2918This section contains kernel programming information in the areas
2919where interacting with cgroup is necessary.  cgroup core and
2920controllers are not covered.
2921
2922
2923Filesystem Support for Writeback
2924--------------------------------
2925
2926A filesystem can support cgroup writeback by updating
2927address_space_operations->writepage[s]() to annotate bio's using the
2928following two functions.
2929
2930  wbc_init_bio(@wbc, @bio)
2931	Should be called for each bio carrying writeback data and
2932	associates the bio with the inode's owner cgroup and the
2933	corresponding request queue.  This must be called after
2934	a queue (device) has been associated with the bio and
2935	before submission.
2936
2937  wbc_account_cgroup_owner(@wbc, @page, @bytes)
2938	Should be called for each data segment being written out.
2939	While this function doesn't care exactly when it's called
2940	during the writeback session, it's the easiest and most
2941	natural to call it as data segments are added to a bio.
2942
2943With writeback bio's annotated, cgroup support can be enabled per
2944super_block by setting SB_I_CGROUPWB in ->s_iflags.  This allows for
2945selective disabling of cgroup writeback support which is helpful when
2946certain filesystem features, e.g. journaled data mode, are
2947incompatible.
2948
2949wbc_init_bio() binds the specified bio to its cgroup.  Depending on
2950the configuration, the bio may be executed at a lower priority and if
2951the writeback session is holding shared resources, e.g. a journal
2952entry, may lead to priority inversion.  There is no one easy solution
2953for the problem.  Filesystems can try to work around specific problem
2954cases by skipping wbc_init_bio() and using bio_associate_blkg()
2955directly.
2956
2957
2958Deprecated v1 Core Features
2959===========================
2960
2961- Multiple hierarchies including named ones are not supported.
2962
2963- All v1 mount options are not supported.
2964
2965- The "tasks" file is removed and "cgroup.procs" is not sorted.
2966
2967- "cgroup.clone_children" is removed.
2968
2969- /proc/cgroups is meaningless for v2.  Use "cgroup.controllers" file
2970  at the root instead.
2971
2972
2973Issues with v1 and Rationales for v2
2974====================================
2975
2976Multiple Hierarchies
2977--------------------
2978
2979cgroup v1 allowed an arbitrary number of hierarchies and each
2980hierarchy could host any number of controllers.  While this seemed to
2981provide a high level of flexibility, it wasn't useful in practice.
2982
2983For example, as there is only one instance of each controller, utility
2984type controllers such as freezer which can be useful in all
2985hierarchies could only be used in one.  The issue is exacerbated by
2986the fact that controllers couldn't be moved to another hierarchy once
2987hierarchies were populated.  Another issue was that all controllers
2988bound to a hierarchy were forced to have exactly the same view of the
2989hierarchy.  It wasn't possible to vary the granularity depending on
2990the specific controller.
2991
2992In practice, these issues heavily limited which controllers could be
2993put on the same hierarchy and most configurations resorted to putting
2994each controller on its own hierarchy.  Only closely related ones, such
2995as the cpu and cpuacct controllers, made sense to be put on the same
2996hierarchy.  This often meant that userland ended up managing multiple
2997similar hierarchies repeating the same steps on each hierarchy
2998whenever a hierarchy management operation was necessary.
2999
3000Furthermore, support for multiple hierarchies came at a steep cost.
3001It greatly complicated cgroup core implementation but more importantly
3002the support for multiple hierarchies restricted how cgroup could be
3003used in general and what controllers was able to do.
3004
3005There was no limit on how many hierarchies there might be, which meant
3006that a thread's cgroup membership couldn't be described in finite
3007length.  The key might contain any number of entries and was unlimited
3008in length, which made it highly awkward to manipulate and led to
3009addition of controllers which existed only to identify membership,
3010which in turn exacerbated the original problem of proliferating number
3011of hierarchies.
3012
3013Also, as a controller couldn't have any expectation regarding the
3014topologies of hierarchies other controllers might be on, each
3015controller had to assume that all other controllers were attached to
3016completely orthogonal hierarchies.  This made it impossible, or at
3017least very cumbersome, for controllers to cooperate with each other.
3018
3019In most use cases, putting controllers on hierarchies which are
3020completely orthogonal to each other isn't necessary.  What usually is
3021called for is the ability to have differing levels of granularity
3022depending on the specific controller.  In other words, hierarchy may
3023be collapsed from leaf towards root when viewed from specific
3024controllers.  For example, a given configuration might not care about
3025how memory is distributed beyond a certain level while still wanting
3026to control how CPU cycles are distributed.
3027
3028
3029Thread Granularity
3030------------------
3031
3032cgroup v1 allowed threads of a process to belong to different cgroups.
3033This didn't make sense for some controllers and those controllers
3034ended up implementing different ways to ignore such situations but
3035much more importantly it blurred the line between API exposed to
3036individual applications and system management interface.
3037
3038Generally, in-process knowledge is available only to the process
3039itself; thus, unlike service-level organization of processes,
3040categorizing threads of a process requires active participation from
3041the application which owns the target process.
3042
3043cgroup v1 had an ambiguously defined delegation model which got abused
3044in combination with thread granularity.  cgroups were delegated to
3045individual applications so that they can create and manage their own
3046sub-hierarchies and control resource distributions along them.  This
3047effectively raised cgroup to the status of a syscall-like API exposed
3048to lay programs.
3049
3050First of all, cgroup has a fundamentally inadequate interface to be
3051exposed this way.  For a process to access its own knobs, it has to
3052extract the path on the target hierarchy from /proc/self/cgroup,
3053construct the path by appending the name of the knob to the path, open
3054and then read and/or write to it.  This is not only extremely clunky
3055and unusual but also inherently racy.  There is no conventional way to
3056define transaction across the required steps and nothing can guarantee
3057that the process would actually be operating on its own sub-hierarchy.
3058
3059cgroup controllers implemented a number of knobs which would never be
3060accepted as public APIs because they were just adding control knobs to
3061system-management pseudo filesystem.  cgroup ended up with interface
3062knobs which were not properly abstracted or refined and directly
3063revealed kernel internal details.  These knobs got exposed to
3064individual applications through the ill-defined delegation mechanism
3065effectively abusing cgroup as a shortcut to implementing public APIs
3066without going through the required scrutiny.
3067
3068This was painful for both userland and kernel.  Userland ended up with
3069misbehaving and poorly abstracted interfaces and kernel exposing and
3070locked into constructs inadvertently.
3071
3072
3073Competition Between Inner Nodes and Threads
3074-------------------------------------------
3075
3076cgroup v1 allowed threads to be in any cgroups which created an
3077interesting problem where threads belonging to a parent cgroup and its
3078children cgroups competed for resources.  This was nasty as two
3079different types of entities competed and there was no obvious way to
3080settle it.  Different controllers did different things.
3081
3082The cpu controller considered threads and cgroups as equivalents and
3083mapped nice levels to cgroup weights.  This worked for some cases but
3084fell flat when children wanted to be allocated specific ratios of CPU
3085cycles and the number of internal threads fluctuated - the ratios
3086constantly changed as the number of competing entities fluctuated.
3087There also were other issues.  The mapping from nice level to weight
3088wasn't obvious or universal, and there were various other knobs which
3089simply weren't available for threads.
3090
3091The io controller implicitly created a hidden leaf node for each
3092cgroup to host the threads.  The hidden leaf had its own copies of all
3093the knobs with ``leaf_`` prefixed.  While this allowed equivalent
3094control over internal threads, it was with serious drawbacks.  It
3095always added an extra layer of nesting which wouldn't be necessary
3096otherwise, made the interface messy and significantly complicated the
3097implementation.
3098
3099The memory controller didn't have a way to control what happened
3100between internal tasks and child cgroups and the behavior was not
3101clearly defined.  There were attempts to add ad-hoc behaviors and
3102knobs to tailor the behavior to specific workloads which would have
3103led to problems extremely difficult to resolve in the long term.
3104
3105Multiple controllers struggled with internal tasks and came up with
3106different ways to deal with it; unfortunately, all the approaches were
3107severely flawed and, furthermore, the widely different behaviors
3108made cgroup as a whole highly inconsistent.
3109
3110This clearly is a problem which needs to be addressed from cgroup core
3111in a uniform way.
3112
3113
3114Other Interface Issues
3115----------------------
3116
3117cgroup v1 grew without oversight and developed a large number of
3118idiosyncrasies and inconsistencies.  One issue on the cgroup core side
3119was how an empty cgroup was notified - a userland helper binary was
3120forked and executed for each event.  The event delivery wasn't
3121recursive or delegatable.  The limitations of the mechanism also led
3122to in-kernel event delivery filtering mechanism further complicating
3123the interface.
3124
3125Controller interfaces were problematic too.  An extreme example is
3126controllers completely ignoring hierarchical organization and treating
3127all cgroups as if they were all located directly under the root
3128cgroup.  Some controllers exposed a large amount of inconsistent
3129implementation details to userland.
3130
3131There also was no consistency across controllers.  When a new cgroup
3132was created, some controllers defaulted to not imposing extra
3133restrictions while others disallowed any resource usage until
3134explicitly configured.  Configuration knobs for the same type of
3135control used widely differing naming schemes and formats.  Statistics
3136and information knobs were named arbitrarily and used different
3137formats and units even in the same controller.
3138
3139cgroup v2 establishes common conventions where appropriate and updates
3140controllers so that they expose minimal and consistent interfaces.
3141
3142
3143Controller Issues and Remedies
3144------------------------------
3145
3146Memory
3147~~~~~~
3148
3149The original lower boundary, the soft limit, is defined as a limit
3150that is per default unset.  As a result, the set of cgroups that
3151global reclaim prefers is opt-in, rather than opt-out.  The costs for
3152optimizing these mostly negative lookups are so high that the
3153implementation, despite its enormous size, does not even provide the
3154basic desirable behavior.  First off, the soft limit has no
3155hierarchical meaning.  All configured groups are organized in a global
3156rbtree and treated like equal peers, regardless where they are located
3157in the hierarchy.  This makes subtree delegation impossible.  Second,
3158the soft limit reclaim pass is so aggressive that it not just
3159introduces high allocation latencies into the system, but also impacts
3160system performance due to overreclaim, to the point where the feature
3161becomes self-defeating.
3162
3163The memory.low boundary on the other hand is a top-down allocated
3164reserve.  A cgroup enjoys reclaim protection when it's within its
3165effective low, which makes delegation of subtrees possible. It also
3166enjoys having reclaim pressure proportional to its overage when
3167above its effective low.
3168
3169The original high boundary, the hard limit, is defined as a strict
3170limit that can not budge, even if the OOM killer has to be called.
3171But this generally goes against the goal of making the most out of the
3172available memory.  The memory consumption of workloads varies during
3173runtime, and that requires users to overcommit.  But doing that with a
3174strict upper limit requires either a fairly accurate prediction of the
3175working set size or adding slack to the limit.  Since working set size
3176estimation is hard and error prone, and getting it wrong results in
3177OOM kills, most users tend to err on the side of a looser limit and
3178end up wasting precious resources.
3179
3180The memory.high boundary on the other hand can be set much more
3181conservatively.  When hit, it throttles allocations by forcing them
3182into direct reclaim to work off the excess, but it never invokes the
3183OOM killer.  As a result, a high boundary that is chosen too
3184aggressively will not terminate the processes, but instead it will
3185lead to gradual performance degradation.  The user can monitor this
3186and make corrections until the minimal memory footprint that still
3187gives acceptable performance is found.
3188
3189In extreme cases, with many concurrent allocations and a complete
3190breakdown of reclaim progress within the group, the high boundary can
3191be exceeded.  But even then it's mostly better to satisfy the
3192allocation from the slack available in other groups or the rest of the
3193system than killing the group.  Otherwise, memory.max is there to
3194limit this type of spillover and ultimately contain buggy or even
3195malicious applications.
3196
3197Setting the original memory.limit_in_bytes below the current usage was
3198subject to a race condition, where concurrent charges could cause the
3199limit setting to fail. memory.max on the other hand will first set the
3200limit to prevent new charges, and then reclaim and OOM kill until the
3201new limit is met - or the task writing to memory.max is killed.
3202
3203The combined memory+swap accounting and limiting is replaced by real
3204control over swap space.
3205
3206The main argument for a combined memory+swap facility in the original
3207cgroup design was that global or parental pressure would always be
3208able to swap all anonymous memory of a child group, regardless of the
3209child's own (possibly untrusted) configuration.  However, untrusted
3210groups can sabotage swapping by other means - such as referencing its
3211anonymous memory in a tight loop - and an admin can not assume full
3212swappability when overcommitting untrusted jobs.
3213
3214For trusted jobs, on the other hand, a combined counter is not an
3215intuitive userspace interface, and it flies in the face of the idea
3216that cgroup controllers should account and limit specific physical
3217resources.  Swap space is a resource like all others in the system,
3218and that's why unified hierarchy allows distributing it separately.
3219