cgroup.h 29.9 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14
#ifndef _LINUX_CGROUP_H
#define _LINUX_CGROUP_H
/*
 *  cgroup interface
 *
 *  Copyright (C) 2003 BULL SA
 *  Copyright (C) 2004-2006 Silicon Graphics, Inc.
 *
 */

#include <linux/sched.h>
#include <linux/cpumask.h>
#include <linux/nodemask.h>
#include <linux/rcupdate.h>
15
#include <linux/rculist.h>
B
Balbir Singh 已提交
16
#include <linux/cgroupstats.h>
17
#include <linux/rwsem.h>
K
KAMEZAWA Hiroyuki 已提交
18
#include <linux/idr.h>
19
#include <linux/workqueue.h>
20
#include <linux/fs.h>
21
#include <linux/percpu-refcount.h>
22
#include <linux/seq_file.h>
T
Tejun Heo 已提交
23
#include <linux/kernfs.h>
24
#include <linux/wait.h>
25 26 27

#ifdef CONFIG_CGROUPS

28
struct cgroup_root;
29
struct cgroup_subsys;
30
struct cgroup;
31 32 33

extern int cgroup_init_early(void);
extern int cgroup_init(void);
34
extern void cgroup_fork(struct task_struct *p);
35
extern void cgroup_post_fork(struct task_struct *p);
36
extern void cgroup_exit(struct task_struct *p);
B
Balbir Singh 已提交
37 38
extern int cgroupstats_build(struct cgroupstats *stats,
				struct dentry *dentry);
39

Z
Zefan Li 已提交
40 41
extern int proc_cgroup_show(struct seq_file *m, struct pid_namespace *ns,
			    struct pid *pid, struct task_struct *tsk);
42

T
Tejun Heo 已提交
43
/* define the enumeration of all cgroup subsystems */
44
#define SUBSYS(_x) _x ## _cgrp_id,
45 46
enum cgroup_subsys_id {
#include <linux/cgroup_subsys.h>
47
	CGROUP_SUBSYS_COUNT,
48 49 50
};
#undef SUBSYS

T
Tejun Heo 已提交
51 52 53 54 55 56 57
/*
 * Per-subsystem/per-cgroup state maintained by the system.  This is the
 * fundamental structural building block that controllers deal with.
 *
 * Fields marked with "PI:" are public and immutable and may be accessed
 * directly without synchronization.
 */
58
struct cgroup_subsys_state {
T
Tejun Heo 已提交
59
	/* PI: the cgroup that this css is attached to */
60 61
	struct cgroup *cgroup;

T
Tejun Heo 已提交
62
	/* PI: the cgroup subsystem that this css is attached to */
63 64
	struct cgroup_subsys *ss;

65 66
	/* reference count - access via css_[try]get() and css_put() */
	struct percpu_ref refcnt;
67

T
Tejun Heo 已提交
68
	/* PI: the parent css */
69 70
	struct cgroup_subsys_state *parent;

71 72 73 74
	/* siblings list anchored at the parent's ->children */
	struct list_head sibling;
	struct list_head children;

75
	/*
T
Tejun Heo 已提交
76
	 * PI: Subsys-unique ID.  0 is unused and root is always 1.  The
77 78 79 80
	 * matching css can be looked up using css_from_id().
	 */
	int id;

81
	unsigned int flags;
82

83 84 85 86 87 88 89 90
	/*
	 * Monotonically increasing unique serial number which defines a
	 * uniform order among all csses.  It's guaranteed that all
	 * ->children lists are in the ascending order of ->serial_nr and
	 * used to allow interrupting and resuming iterations.
	 */
	u64 serial_nr;

91 92
	/* percpu_ref killing and RCU release */
	struct rcu_head rcu_head;
93
	struct work_struct destroy_work;
94 95 96 97
};

/* bits in struct cgroup_subsys_state flags field */
enum {
98
	CSS_NO_REF	= (1 << 0), /* no reference counting for this css */
99
	CSS_ONLINE	= (1 << 1), /* between ->css_online() and ->css_offline() */
100
	CSS_RELEASED	= (1 << 2), /* refcnt reached zero, released */
101 102
};

103 104 105 106 107
/**
 * css_get - obtain a reference on the specified css
 * @css: target css
 *
 * The caller must already have a reference.
108 109 110
 */
static inline void css_get(struct cgroup_subsys_state *css)
{
111 112
	if (!(css->flags & CSS_NO_REF))
		percpu_ref_get(&css->refcnt);
113
}
P
Paul Menage 已提交
114

115 116 117 118 119 120 121 122 123 124 125 126 127
/**
 * css_get_many - obtain references on the specified css
 * @css: target css
 * @n: number of references to get
 *
 * The caller must already have a reference.
 */
static inline void css_get_many(struct cgroup_subsys_state *css, unsigned int n)
{
	if (!(css->flags & CSS_NO_REF))
		percpu_ref_get_many(&css->refcnt, n);
}

T
Tejun Heo 已提交
128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145
/**
 * css_tryget - try to obtain a reference on the specified css
 * @css: target css
 *
 * Obtain a reference on @css unless it already has reached zero and is
 * being released.  This function doesn't care whether @css is on or
 * offline.  The caller naturally needs to ensure that @css is accessible
 * but doesn't have to be holding a reference on it - IOW, RCU protected
 * access is good enough for this function.  Returns %true if a reference
 * count was successfully obtained; %false otherwise.
 */
static inline bool css_tryget(struct cgroup_subsys_state *css)
{
	if (!(css->flags & CSS_NO_REF))
		return percpu_ref_tryget(&css->refcnt);
	return true;
}

146
/**
147
 * css_tryget_online - try to obtain a reference on the specified css if online
148 149
 * @css: target css
 *
150 151
 * Obtain a reference on @css if it's online.  The caller naturally needs
 * to ensure that @css is accessible but doesn't have to be holding a
152 153 154 155
 * reference on it - IOW, RCU protected access is good enough for this
 * function.  Returns %true if a reference count was successfully obtained;
 * %false otherwise.
 */
156
static inline bool css_tryget_online(struct cgroup_subsys_state *css)
P
Paul Menage 已提交
157
{
158 159 160
	if (!(css->flags & CSS_NO_REF))
		return percpu_ref_tryget_live(&css->refcnt);
	return true;
P
Paul Menage 已提交
161 162
}

163 164 165 166
/**
 * css_put - put a css reference
 * @css: target css
 *
167
 * Put a reference obtained via css_get() and css_tryget_online().
168
 */
169 170
static inline void css_put(struct cgroup_subsys_state *css)
{
171 172
	if (!(css->flags & CSS_NO_REF))
		percpu_ref_put(&css->refcnt);
173 174
}

175 176 177 178 179 180 181 182 183 184 185 186 187
/**
 * css_put_many - put css references
 * @css: target css
 * @n: number of references to put
 *
 * Put references obtained via css_get() and css_tryget_online().
 */
static inline void css_put_many(struct cgroup_subsys_state *css, unsigned int n)
{
	if (!(css->flags & CSS_NO_REF))
		percpu_ref_put_many(&css->refcnt, n);
}

188 189 190 191
/* bits in struct cgroup flags field */
enum {
	/* Control Group requires release notifications to userspace */
	CGRP_NOTIFY_ON_RELEASE,
192
	/*
193 194 195
	 * Clone the parent's configuration when creating a new child
	 * cpuset cgroup.  For historical reasons, this option can be
	 * specified at mount time and thus is implemented here.
196
	 */
197
	CGRP_CPUSET_CLONE_CHILDREN,
198 199
};

200
struct cgroup {
201 202 203
	/* self css with NULL ->ss, points back to this cgroup */
	struct cgroup_subsys_state self;

204 205
	unsigned long flags;		/* "unsigned long" so bitops work */

206 207 208
	/*
	 * idr allocated in-hierarchy ID.
	 *
209 210
	 * ID 0 is not used, the ID of the root cgroup is always 1, and a
	 * new cgroup will be assigned with a smallest available ID.
211 212
	 *
	 * Allocating/Removing ID must be protected by cgroup_mutex.
213 214
	 */
	int id;
T
Tejun Heo 已提交
215

216 217 218 219 220 221 222 223
	/*
	 * If this cgroup contains any tasks, it contributes one to
	 * populated_cnt.  All children with non-zero popuplated_cnt of
	 * their own contribute one.  The count is zero iff there's no task
	 * in this cgroup or its subtree.
	 */
	int populated_cnt;

T
Tejun Heo 已提交
224
	struct kernfs_node *kn;		/* cgroup kernfs entry */
225
	struct kernfs_node *populated_kn; /* kn for "cgroup.subtree_populated" */
226

227 228 229 230 231
	/*
	 * The bitmask of subsystems enabled on the child cgroups.
	 * ->subtree_control is the one configured through
	 * "cgroup.subtree_control" while ->child_subsys_mask is the
	 * effective one which may have more subsystems enabled.
232 233
	 * Controller knobs are made available iff it's enabled in
	 * ->subtree_control.
234 235
	 */
	unsigned int subtree_control;
236
	unsigned int child_subsys_mask;
237

238
	/* Private pointers for each registered subsystem */
239
	struct cgroup_subsys_state __rcu *subsys[CGROUP_SUBSYS_COUNT];
240

241
	struct cgroup_root *root;
242 243

	/*
244 245
	 * List of cgrp_cset_links pointing at css_sets with tasks in this
	 * cgroup.  Protected by css_set_lock.
246
	 */
247
	struct list_head cset_links;
248

T
Tejun Heo 已提交
249 250 251 252 253 254 255 256 257
	/*
	 * On the default hierarchy, a css_set for a cgroup with some
	 * susbsys disabled will point to css's which are associated with
	 * the closest ancestor which has the subsys enabled.  The
	 * following lists all css_sets which point to this cgroup's css
	 * for the given subsystem.
	 */
	struct list_head e_csets[CGROUP_SUBSYS_COUNT];

258 259 260 261 262 263
	/*
	 * list of pidlists, up to two for each namespace (one for procs, one
	 * for tasks); created on demand.
	 */
	struct list_head pidlists;
	struct mutex pidlist_mutex;
264

265 266
	/* used to wait for offlining of csses */
	wait_queue_head_t offline_waitq;
267 268 269

	/* used to schedule release agent */
	struct work_struct release_agent_work;
270 271
};

272 273
#define MAX_CGROUP_ROOT_NAMELEN 64

274
/* cgroup_root->flags */
275
enum {
276
	CGRP_ROOT_SANE_BEHAVIOR	= (1 << 0), /* __DEVEL__sane_behavior specified */
277 278 279 280 281
	CGRP_ROOT_NOPREFIX	= (1 << 1), /* mounted subsystems have no named prefix */
	CGRP_ROOT_XATTR		= (1 << 2), /* supports extended attributes */
};

/*
282
 * A cgroup_root represents the root of a cgroup hierarchy, and may be
T
Tejun Heo 已提交
283
 * associated with a kernfs_root to form an active hierarchy.  This is
284 285
 * internal to cgroup core.  Don't access directly from controllers.
 */
286
struct cgroup_root {
T
Tejun Heo 已提交
287
	struct kernfs_root *kf_root;
288

289
	/* The bitmask of subsystems attached to this hierarchy */
290
	unsigned int subsys_mask;
291

292 293 294
	/* Unique id for this hierarchy. */
	int hierarchy_id;

T
Tejun Heo 已提交
295
	/* The root cgroup.  Root is destroyed on its release. */
296
	struct cgroup cgrp;
297

298 299
	/* Number of cgroups in the hierarchy, used only for /proc/cgroups */
	atomic_t nr_cgrps;
300 301 302 303 304

	/* A list running through the active hierarchies */
	struct list_head root_list;

	/* Hierarchy-specific flags */
305
	unsigned int flags;
306 307

	/* IDs for cgroups in this hierarchy */
308
	struct idr cgroup_idr;
309 310 311 312 313 314 315 316

	/* The path to use for release notifications. */
	char release_agent_path[PATH_MAX];

	/* The name for this hierarchy - may be empty */
	char name[MAX_CGROUP_ROOT_NAMELEN];
};

P
Paul Menage 已提交
317 318
/*
 * A css_set is a structure holding pointers to a set of
319 320
 * cgroup_subsys_state objects. This saves space in the task struct
 * object and speeds up fork()/exit(), since a single inc/dec and a
P
Paul Menage 已提交
321 322
 * list_add()/del() can bump the reference count on the entire cgroup
 * set for a task.
323 324 325 326 327
 */

struct css_set {

	/* Reference count */
328
	atomic_t refcount;
329

330 331 332 333 334 335
	/*
	 * List running through all cgroup groups in the same hash
	 * slot. Protected by css_set_lock
	 */
	struct hlist_node hlist;

336
	/*
T
Tejun Heo 已提交
337 338 339 340 341
	 * Lists running through all tasks using this cgroup group.
	 * mg_tasks lists tasks which belong to this cset but are in the
	 * process of being migrated out or in.  Protected by
	 * css_set_rwsem, but, during migration, once tasks are moved to
	 * mg_tasks, it can be read safely while holding cgroup_mutex.
342 343
	 */
	struct list_head tasks;
T
Tejun Heo 已提交
344
	struct list_head mg_tasks;
345 346

	/*
347 348
	 * List of cgrp_cset_links pointing at cgroups referenced from this
	 * css_set.  Protected by css_set_lock.
349
	 */
350
	struct list_head cgrp_links;
351

T
Tejun Heo 已提交
352 353 354
	/* the default cgroup associated with this css_set */
	struct cgroup *dfl_cgrp;

355
	/*
T
Tejun Heo 已提交
356 357 358
	 * Set of subsystem states, one for each subsystem. This array is
	 * immutable after creation apart from the init_css_set during
	 * subsystem registration (at boot time).
359 360
	 */
	struct cgroup_subsys_state *subsys[CGROUP_SUBSYS_COUNT];
361

362 363 364 365
	/*
	 * List of csets participating in the on-going migration either as
	 * source or destination.  Protected by cgroup_mutex.
	 */
366
	struct list_head mg_preload_node;
367 368 369 370 371 372 373 374 375 376 377 378
	struct list_head mg_node;

	/*
	 * If this cset is acting as the source of migration the following
	 * two fields are set.  mg_src_cgrp is the source cgroup of the
	 * on-going migration and mg_dst_cset is the destination cset the
	 * target tasks on this cset should be migrated to.  Protected by
	 * cgroup_mutex.
	 */
	struct cgroup *mg_src_cgrp;
	struct css_set *mg_dst_cset;

T
Tejun Heo 已提交
379 380 381 382 383 384 385 386 387
	/*
	 * On the default hierarhcy, ->subsys[ssid] may point to a css
	 * attached to an ancestor instead of the cgroup this css_set is
	 * associated with.  The following node is anchored at
	 * ->subsys[ssid]->cgroup->e_csets[ssid] and provides a way to
	 * iterate through all css's attached to a given cgroup.
	 */
	struct list_head e_cset_node[CGROUP_SUBSYS_COUNT];

388 389
	/* For RCU-protected deletion */
	struct rcu_head rcu_head;
390 391
};

P
Paul Menage 已提交
392 393
/*
 * struct cftype: handler definitions for cgroup control files
394 395
 *
 * When reading/writing to a file:
A
Al Viro 已提交
396 397
 *	- the cgroup to use is file->f_path.dentry->d_parent->d_fsdata
 *	- the 'cftype' of the file is file->f_path.dentry->d_fsdata
398 399
 */

400
/* cftype->flags */
401
enum {
L
Li Zefan 已提交
402 403
	CFTYPE_ONLY_ON_ROOT	= (1 << 0),	/* only create on root cgrp */
	CFTYPE_NOT_ON_ROOT	= (1 << 1),	/* don't create on root cgrp */
T
Tejun Heo 已提交
404
	CFTYPE_NO_PREFIX	= (1 << 3),	/* (DON'T USE FOR NEW FILES) no subsys prefix */
405 406 407 408

	/* internal flags, do not use outside cgroup core proper */
	__CFTYPE_ONLY_ON_DFL	= (1 << 16),	/* only on default hierarchy */
	__CFTYPE_NOT_ON_DFL	= (1 << 17),	/* not on default hierarchy */
409
};
410 411 412

#define MAX_CFTYPE_NAME		64

413
struct cftype {
P
Paul Menage 已提交
414 415
	/*
	 * By convention, the name should begin with the name of the
416 417
	 * subsystem, followed by a period.  Zero length string indicates
	 * end of cftype array.
P
Paul Menage 已提交
418
	 */
419 420
	char name[MAX_CFTYPE_NAME];
	int private;
L
Li Zefan 已提交
421 422 423 424
	/*
	 * If not 0, file mode is set to this value, otherwise it will
	 * be figured out automatically
	 */
A
Al Viro 已提交
425
	umode_t mode;
426 427

	/*
428
	 * The maximum length of string, excluding trailing nul, that can
429
	 * be passed to write.  If < PAGE_SIZE-1, PAGE_SIZE-1 is assumed.
430 431 432
	 */
	size_t max_write_len;

433 434 435
	/* CFTYPE_* flags */
	unsigned int flags;

436
	/*
T
Tejun Heo 已提交
437 438
	 * Fields used for internal bookkeeping.  Initialized automatically
	 * during registration.
439
	 */
T
Tejun Heo 已提交
440 441
	struct cgroup_subsys *ss;	/* NULL for cgroup core files */
	struct list_head node;		/* anchored at ss->cfts */
T
Tejun Heo 已提交
442 443
	struct kernfs_ops *kf_ops;

444
	/*
445
	 * read_u64() is a shortcut for the common case of returning a
446 447
	 * single integer. Use it in place of read()
	 */
448
	u64 (*read_u64)(struct cgroup_subsys_state *css, struct cftype *cft);
449 450 451
	/*
	 * read_s64() is a signed version of read_u64()
	 */
452
	s64 (*read_s64)(struct cgroup_subsys_state *css, struct cftype *cft);
453 454 455

	/* generic seq_file read interface */
	int (*seq_show)(struct seq_file *sf, void *v);
456

457 458 459 460 461
	/* optional ops, implement all or none */
	void *(*seq_start)(struct seq_file *sf, loff_t *ppos);
	void *(*seq_next)(struct seq_file *sf, void *v, loff_t *ppos);
	void (*seq_stop)(struct seq_file *sf, void *v);

462
	/*
463
	 * write_u64() is a shortcut for the common case of accepting
464 465 466
	 * a single integer (as parsed by simple_strtoull) from
	 * userspace. Use in place of write(); return 0 or error.
	 */
467 468
	int (*write_u64)(struct cgroup_subsys_state *css, struct cftype *cft,
			 u64 val);
469 470 471
	/*
	 * write_s64() is a signed version of write_u64()
	 */
472 473
	int (*write_s64)(struct cgroup_subsys_state *css, struct cftype *cft,
			 s64 val);
474

T
Tejun Heo 已提交
475 476 477 478 479 480 481 482 483
	/*
	 * write() is the generic write callback which maps directly to
	 * kernfs write operation and overrides all other operations.
	 * Maximum write size is determined by ->max_write_len.  Use
	 * of_css/cft() to access the associated css and cft.
	 */
	ssize_t (*write)(struct kernfs_open_file *of,
			 char *buf, size_t nbytes, loff_t off);

T
Tejun Heo 已提交
484 485 486
#ifdef CONFIG_DEBUG_LOCK_ALLOC
	struct lock_class_key	lockdep_key;
#endif
487 488
};

T
Tejun Heo 已提交
489
extern struct cgroup_root cgrp_dfl_root;
T
Tejun Heo 已提交
490
extern struct css_set init_css_set;
T
Tejun Heo 已提交
491

492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544
/**
 * cgroup_on_dfl - test whether a cgroup is on the default hierarchy
 * @cgrp: the cgroup of interest
 *
 * The default hierarchy is the v2 interface of cgroup and this function
 * can be used to test whether a cgroup is on the default hierarchy for
 * cases where a subsystem should behave differnetly depending on the
 * interface version.
 *
 * The set of behaviors which change on the default hierarchy are still
 * being determined and the mount option is prefixed with __DEVEL__.
 *
 * List of changed behaviors:
 *
 * - Mount options "noprefix", "xattr", "clone_children", "release_agent"
 *   and "name" are disallowed.
 *
 * - When mounting an existing superblock, mount options should match.
 *
 * - Remount is disallowed.
 *
 * - rename(2) is disallowed.
 *
 * - "tasks" is removed.  Everything should be at process granularity.  Use
 *   "cgroup.procs" instead.
 *
 * - "cgroup.procs" is not sorted.  pids will be unique unless they got
 *   recycled inbetween reads.
 *
 * - "release_agent" and "notify_on_release" are removed.  Replacement
 *   notification mechanism will be implemented.
 *
 * - "cgroup.clone_children" is removed.
 *
 * - "cgroup.subtree_populated" is available.  Its value is 0 if the cgroup
 *   and its descendants contain no task; otherwise, 1.  The file also
 *   generates kernfs notification which can be monitored through poll and
 *   [di]notify when the value of the file changes.
 *
 * - cpuset: tasks will be kept in empty cpusets when hotplug happens and
 *   take masks of ancestors with non-empty cpus/mems, instead of being
 *   moved to an ancestor.
 *
 * - cpuset: a task can be moved into an empty cpuset, and again it takes
 *   masks of ancestors.
 *
 * - memcg: use_hierarchy is on by default and the cgroup file for the flag
 *   is not created.
 *
 * - blkcg: blk-throttle becomes properly hierarchical.
 *
 * - debug: disallowed on the default hierarchy.
 */
T
Tejun Heo 已提交
545 546 547 548 549
static inline bool cgroup_on_dfl(const struct cgroup *cgrp)
{
	return cgrp->root == &cgrp_dfl_root;
}

550 551 552 553 554 555
/* no synchronization, the result can only be used as a hint */
static inline bool cgroup_has_tasks(struct cgroup *cgrp)
{
	return !list_empty(&cgrp->cset_links);
}

556
/* returns ino associated with a cgroup */
T
Tejun Heo 已提交
557 558
static inline ino_t cgroup_ino(struct cgroup *cgrp)
{
559
	return cgrp->kn->ino;
T
Tejun Heo 已提交
560 561
}

T
Tejun Heo 已提交
562 563
/* cft/css accessors for cftype->write() operation */
static inline struct cftype *of_cft(struct kernfs_open_file *of)
564
{
T
Tejun Heo 已提交
565
	return of->kn->priv;
566 567
}

T
Tejun Heo 已提交
568 569 570 571 572 573 574 575 576 577 578 579
struct cgroup_subsys_state *of_css(struct kernfs_open_file *of);

/* cft/css accessors for cftype->seq_*() operations */
static inline struct cftype *seq_cft(struct seq_file *seq)
{
	return of_cft(seq->private);
}

static inline struct cgroup_subsys_state *seq_css(struct seq_file *seq)
{
	return of_css(seq->private);
}
580

T
Tejun Heo 已提交
581 582 583 584 585 586 587
/*
 * Name / path handling functions.  All are thin wrappers around the kernfs
 * counterparts and can be called under any context.
 */

static inline int cgroup_name(struct cgroup *cgrp, char *buf, size_t buflen)
{
588
	return kernfs_name(cgrp->kn, buf, buflen);
T
Tejun Heo 已提交
589 590 591 592 593
}

static inline char * __must_check cgroup_path(struct cgroup *cgrp, char *buf,
					      size_t buflen)
{
594
	return kernfs_path(cgrp->kn, buf, buflen);
T
Tejun Heo 已提交
595 596 597 598
}

static inline void pr_cont_cgroup_name(struct cgroup *cgrp)
{
599
	pr_cont_kernfs_name(cgrp->kn);
T
Tejun Heo 已提交
600 601 602 603
}

static inline void pr_cont_cgroup_path(struct cgroup *cgrp)
{
604
	pr_cont_kernfs_path(cgrp->kn);
T
Tejun Heo 已提交
605 606 607 608
}

char *task_cgroup_path(struct task_struct *task, char *buf, size_t buflen);

609
int cgroup_add_dfl_cftypes(struct cgroup_subsys *ss, struct cftype *cfts);
610
int cgroup_add_legacy_cftypes(struct cgroup_subsys *ss, struct cftype *cfts);
611
int cgroup_rm_cftypes(struct cftype *cfts);
612

613
bool cgroup_is_descendant(struct cgroup *cgrp, struct cgroup *ancestor);
614

615 616 617 618 619 620 621 622 623 624 625 626 627
/*
 * Control Group taskset, used to pass around set of tasks to cgroup_subsys
 * methods.
 */
struct cgroup_taskset;
struct task_struct *cgroup_taskset_first(struct cgroup_taskset *tset);
struct task_struct *cgroup_taskset_next(struct cgroup_taskset *tset);

/**
 * cgroup_taskset_for_each - iterate cgroup_taskset
 * @task: the loop cursor
 * @tset: taskset to iterate
 */
628
#define cgroup_taskset_for_each(task, tset)				\
629
	for ((task) = cgroup_taskset_first((tset)); (task);		\
630
	     (task) = cgroup_taskset_next((tset)))
631

632 633 634 635
/*
 * Control Group subsystem type.
 * See Documentation/cgroups/cgroups.txt for details
 */
636 637

struct cgroup_subsys {
638 639 640
	struct cgroup_subsys_state *(*css_alloc)(struct cgroup_subsys_state *parent_css);
	int (*css_online)(struct cgroup_subsys_state *css);
	void (*css_offline)(struct cgroup_subsys_state *css);
641
	void (*css_released)(struct cgroup_subsys_state *css);
642
	void (*css_free)(struct cgroup_subsys_state *css);
643
	void (*css_reset)(struct cgroup_subsys_state *css);
644
	void (*css_e_css_changed)(struct cgroup_subsys_state *css);
645 646 647 648 649 650 651

	int (*can_attach)(struct cgroup_subsys_state *css,
			  struct cgroup_taskset *tset);
	void (*cancel_attach)(struct cgroup_subsys_state *css,
			      struct cgroup_taskset *tset);
	void (*attach)(struct cgroup_subsys_state *css,
		       struct cgroup_taskset *tset);
652
	void (*fork)(struct task_struct *task);
653 654
	void (*exit)(struct cgroup_subsys_state *css,
		     struct cgroup_subsys_state *old_css,
655
		     struct task_struct *task);
656
	void (*bind)(struct cgroup_subsys_state *root_css);
657

658
	int disabled;
659
	int early_init;
660

661 662 663 664 665 666 667 668 669 670 671 672 673 674 675
	/*
	 * If %false, this subsystem is properly hierarchical -
	 * configuration, resource accounting and restriction on a parent
	 * cgroup cover those of its children.  If %true, hierarchy support
	 * is broken in some ways - some subsystems ignore hierarchy
	 * completely while others are only implemented half-way.
	 *
	 * It's now disallowed to create nested cgroups if the subsystem is
	 * broken and cgroup core will emit a warning message on such
	 * cases.  Eventually, all subsystems will be made properly
	 * hierarchical and this will go away.
	 */
	bool broken_hierarchy;
	bool warned_broken_hierarchy;

676
	/* the following two fields are initialized automtically during boot */
677
	int id;
678 679 680
#define MAX_CGROUP_TYPE_NAMELEN 32
	const char *name;

T
Tejun Heo 已提交
681
	/* link to parent, protected by cgroup_lock() */
682
	struct cgroup_root *root;
683

684 685 686
	/* idr for css->id */
	struct idr css_idr;

T
Tejun Heo 已提交
687 688 689 690 691
	/*
	 * List of cftypes.  Each entry is the first entry of an array
	 * terminated by zero length name.
	 */
	struct list_head cfts;
692

693 694 695 696 697 698
	/*
	 * Base cftypes which are automatically registered.  The two can
	 * point to the same array.
	 */
	struct cftype *dfl_cftypes;	/* for the default hierarchy */
	struct cftype *legacy_cftypes;	/* for the legacy hierarchies */
699 700 701 702 703 704 705 706 707

	/*
	 * A subsystem may depend on other subsystems.  When such subsystem
	 * is enabled on a cgroup, the depended-upon subsystems are enabled
	 * together if available.  Subsystems enabled due to dependency are
	 * not visible to userland until explicitly enabled.  The following
	 * specifies the mask of subsystems that this one depends on.
	 */
	unsigned int depends_on;
708 709
};

710
#define SUBSYS(_x) extern struct cgroup_subsys _x ## _cgrp_subsys;
711 712 713
#include <linux/cgroup_subsys.h>
#undef SUBSYS

714 715 716 717 718 719 720 721 722 723 724 725
/**
 * task_css_set_check - obtain a task's css_set with extra access conditions
 * @task: the task to obtain css_set for
 * @__c: extra condition expression to be passed to rcu_dereference_check()
 *
 * A task's css_set is RCU protected, initialized and exited while holding
 * task_lock(), and can only be modified while holding both cgroup_mutex
 * and task_lock() while the task is alive.  This macro verifies that the
 * caller is inside proper critical section and returns @task's css_set.
 *
 * The caller can also specify additional allowed conditions via @__c, such
 * as locks used during the cgroup_subsys::attach() methods.
726
 */
T
Tejun Heo 已提交
727 728
#ifdef CONFIG_PROVE_RCU
extern struct mutex cgroup_mutex;
729
extern struct rw_semaphore css_set_rwsem;
730 731
#define task_css_set_check(task, __c)					\
	rcu_dereference_check((task)->cgroups,				\
732 733 734
		lockdep_is_held(&cgroup_mutex) ||			\
		lockdep_is_held(&css_set_rwsem) ||			\
		((task)->flags & PF_EXITING) || (__c))
T
Tejun Heo 已提交
735
#else
736 737
#define task_css_set_check(task, __c)					\
	rcu_dereference((task)->cgroups)
T
Tejun Heo 已提交
738
#endif
739

740
/**
741
 * task_css_check - obtain css for (task, subsys) w/ extra access conds
742 743 744 745 746 747 748
 * @task: the target task
 * @subsys_id: the target subsystem ID
 * @__c: extra condition expression to be passed to rcu_dereference_check()
 *
 * Return the cgroup_subsys_state for the (@task, @subsys_id) pair.  The
 * synchronization rules are the same as task_css_set_check().
 */
749
#define task_css_check(task, subsys_id, __c)				\
750 751 752 753 754 755 756 757 758 759 760 761 762 763
	task_css_set_check((task), (__c))->subsys[(subsys_id)]

/**
 * task_css_set - obtain a task's css_set
 * @task: the task to obtain css_set for
 *
 * See task_css_set_check().
 */
static inline struct css_set *task_css_set(struct task_struct *task)
{
	return task_css_set_check(task, false);
}

/**
764
 * task_css - obtain css for (task, subsys)
765 766 767
 * @task: the target task
 * @subsys_id: the target subsystem ID
 *
768
 * See task_css_check().
769
 */
770 771
static inline struct cgroup_subsys_state *task_css(struct task_struct *task,
						   int subsys_id)
772
{
773
	return task_css_check(task, subsys_id, false);
774 775
}

T
Tejun Heo 已提交
776 777 778 779 780 781 782 783 784 785 786 787 788 789
/**
 * task_css_is_root - test whether a task belongs to the root css
 * @task: the target task
 * @subsys_id: the target subsystem ID
 *
 * Test whether @task belongs to the root css on the specified subsystem.
 * May be invoked in any context.
 */
static inline bool task_css_is_root(struct task_struct *task, int subsys_id)
{
	return task_css_check(task, subsys_id, true) ==
		init_css_set.subsys[subsys_id];
}

790 791
static inline struct cgroup *task_cgroup(struct task_struct *task,
					 int subsys_id)
792
{
793
	return task_css(task, subsys_id)->cgroup;
794 795
}

796 797
struct cgroup_subsys_state *css_next_child(struct cgroup_subsys_state *pos,
					   struct cgroup_subsys_state *parent);
798

799 800
struct cgroup_subsys_state *css_from_id(int id, struct cgroup_subsys *ss);

801
/**
802 803 804
 * css_for_each_child - iterate through children of a css
 * @pos: the css * to use as the loop cursor
 * @parent: css whose children to walk
805
 *
806
 * Walk @parent's children.  Must be called under rcu_read_lock().
807
 *
808 809 810 811 812 813
 * If a subsystem synchronizes ->css_online() and the start of iteration, a
 * css which finished ->css_online() is guaranteed to be visible in the
 * future iterations and will stay visible until the last reference is put.
 * A css which hasn't finished ->css_online() or already finished
 * ->css_offline() may show up during traversal.  It's each subsystem's
 * responsibility to synchronize against on/offlining.
814 815 816 817
 *
 * It is allowed to temporarily drop RCU read lock during iteration.  The
 * caller is responsible for ensuring that @pos remains accessible until
 * the start of the next iteration by, for example, bumping the css refcnt.
818
 */
819 820 821
#define css_for_each_child(pos, parent)					\
	for ((pos) = css_next_child(NULL, (parent)); (pos);		\
	     (pos) = css_next_child((pos), (parent)))
822

823 824 825 826 827 828
struct cgroup_subsys_state *
css_next_descendant_pre(struct cgroup_subsys_state *pos,
			struct cgroup_subsys_state *css);

struct cgroup_subsys_state *
css_rightmost_descendant(struct cgroup_subsys_state *pos);
829 830

/**
831 832 833
 * css_for_each_descendant_pre - pre-order walk of a css's descendants
 * @pos: the css * to use as the loop cursor
 * @root: css whose descendants to walk
834
 *
835
 * Walk @root's descendants.  @root is included in the iteration and the
836
 * first node to be visited.  Must be called under rcu_read_lock().
837
 *
838 839 840 841 842 843
 * If a subsystem synchronizes ->css_online() and the start of iteration, a
 * css which finished ->css_online() is guaranteed to be visible in the
 * future iterations and will stay visible until the last reference is put.
 * A css which hasn't finished ->css_online() or already finished
 * ->css_offline() may show up during traversal.  It's each subsystem's
 * responsibility to synchronize against on/offlining.
844
 *
845
 * For example, the following guarantees that a descendant can't escape
846 847
 * state updates of its ancestors.
 *
848
 * my_online(@css)
849
 * {
850 851
 *	Lock @css's parent and @css;
 *	Inherit state from the parent;
852 853 854
 *	Unlock both.
 * }
 *
855
 * my_update_state(@css)
856
 * {
857
 *	css_for_each_descendant_pre(@pos, @css) {
858
 *		Lock @pos;
859 860 861 862
 *		if (@pos == @css)
 *			Update @css's state;
 *		else
 *			Verify @pos is alive and inherit state from its parent;
863 864 865 866 867 868 869 870 871 872
 *		Unlock @pos;
 *	}
 * }
 *
 * As long as the inheriting step, including checking the parent state, is
 * enclosed inside @pos locking, double-locking the parent isn't necessary
 * while inheriting.  The state update to the parent is guaranteed to be
 * visible by walking order and, as long as inheriting operations to the
 * same @pos are atomic to each other, multiple updates racing each other
 * still result in the correct state.  It's guaranateed that at least one
873
 * inheritance happens for any css after the latest update to its parent.
874 875 876 877 878
 *
 * If checking parent's state requires locking the parent, each inheriting
 * iteration should lock and unlock both @pos->parent and @pos.
 *
 * Alternatively, a subsystem may choose to use a single global lock to
879
 * synchronize ->css_online() and ->css_offline() against tree-walking
880
 * operations.
881 882 883 884
 *
 * It is allowed to temporarily drop RCU read lock during iteration.  The
 * caller is responsible for ensuring that @pos remains accessible until
 * the start of the next iteration by, for example, bumping the css refcnt.
885
 */
886 887 888
#define css_for_each_descendant_pre(pos, css)				\
	for ((pos) = css_next_descendant_pre(NULL, (css)); (pos);	\
	     (pos) = css_next_descendant_pre((pos), (css)))
889

890 891 892
struct cgroup_subsys_state *
css_next_descendant_post(struct cgroup_subsys_state *pos,
			 struct cgroup_subsys_state *css);
893 894

/**
895 896 897
 * css_for_each_descendant_post - post-order walk of a css's descendants
 * @pos: the css * to use as the loop cursor
 * @css: css whose descendants to walk
898
 *
899
 * Similar to css_for_each_descendant_pre() but performs post-order
900
 * traversal instead.  @root is included in the iteration and the last
901 902 903 904 905 906 907 908 909 910 911
 * node to be visited.
 *
 * If a subsystem synchronizes ->css_online() and the start of iteration, a
 * css which finished ->css_online() is guaranteed to be visible in the
 * future iterations and will stay visible until the last reference is put.
 * A css which hasn't finished ->css_online() or already finished
 * ->css_offline() may show up during traversal.  It's each subsystem's
 * responsibility to synchronize against on/offlining.
 *
 * Note that the walk visibility guarantee example described in pre-order
 * walk doesn't apply the same to post-order walks.
912
 */
913 914 915
#define css_for_each_descendant_post(pos, css)				\
	for ((pos) = css_next_descendant_post(NULL, (css)); (pos);	\
	     (pos) = css_next_descendant_post((pos), (css)))
916

917 918
bool css_has_online_children(struct cgroup_subsys_state *css);

919 920
/* A css_task_iter should be treated as an opaque object */
struct css_task_iter {
921 922
	struct cgroup_subsys		*ss;

T
Tejun Heo 已提交
923 924 925 926 927 928
	struct list_head		*cset_pos;
	struct list_head		*cset_head;

	struct list_head		*task_pos;
	struct list_head		*tasks_head;
	struct list_head		*mg_tasks_head;
929 930
};

931 932 933 934
void css_task_iter_start(struct cgroup_subsys_state *css,
			 struct css_task_iter *it);
struct task_struct *css_task_iter_next(struct css_task_iter *it);
void css_task_iter_end(struct css_task_iter *it);
T
Tejun Heo 已提交
935

M
Michael S. Tsirkin 已提交
936
int cgroup_attach_task_all(struct task_struct *from, struct task_struct *);
937
int cgroup_transfer_tasks(struct cgroup *to, struct cgroup *from);
M
Michael S. Tsirkin 已提交
938

T
Tejun Heo 已提交
939 940
struct cgroup_subsys_state *cgroup_get_e_css(struct cgroup *cgroup,
					     struct cgroup_subsys *ss);
941 942
struct cgroup_subsys_state *css_tryget_online_from_dir(struct dentry *dentry,
						       struct cgroup_subsys *ss);
K
KAMEZAWA Hiroyuki 已提交
943

944 945
#else /* !CONFIG_CGROUPS */

946 947
struct cgroup_subsys_state;

948 949
static inline int cgroup_init_early(void) { return 0; }
static inline int cgroup_init(void) { return 0; }
950
static inline void cgroup_fork(struct task_struct *p) {}
951
static inline void cgroup_post_fork(struct task_struct *p) {}
952
static inline void cgroup_exit(struct task_struct *p) {}
953

B
Balbir Singh 已提交
954 955 956 957 958
static inline int cgroupstats_build(struct cgroupstats *stats,
					struct dentry *dentry)
{
	return -EINVAL;
}
959

960 961
static inline void css_put(struct cgroup_subsys_state *css) {}

962
/* No cgroups - nothing to do */
M
Michael S. Tsirkin 已提交
963 964 965 966 967
static inline int cgroup_attach_task_all(struct task_struct *from,
					 struct task_struct *t)
{
	return 0;
}
968

969 970 971
#endif /* !CONFIG_CGROUPS */

#endif /* _LINUX_CGROUP_H */