cgroup.h 27.8 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 30
struct cgroup_subsys;
struct inode;
31
struct cgroup;
32 33 34

extern int cgroup_init_early(void);
extern int cgroup_init(void);
35
extern void cgroup_fork(struct task_struct *p);
36
extern void cgroup_post_fork(struct task_struct *p);
37
extern void cgroup_exit(struct task_struct *p);
B
Balbir Singh 已提交
38 39
extern int cgroupstats_build(struct cgroupstats *stats,
				struct dentry *dentry);
40

41
extern int proc_cgroup_show(struct seq_file *, void *);
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

51 52
/* Per-subsystem/per-cgroup state maintained by the system. */
struct cgroup_subsys_state {
53
	/* the cgroup that this css is attached to */
54 55
	struct cgroup *cgroup;

56 57 58
	/* the cgroup subsystem that this css is attached to */
	struct cgroup_subsys *ss;

59 60
	/* reference count - access via css_[try]get() and css_put() */
	struct percpu_ref refcnt;
61

62 63 64
	/* the parent css */
	struct cgroup_subsys_state *parent;

65 66 67 68 69 70
	/*
	 * Subsys-unique ID.  0 is unused and root is always 1.  The
	 * matching css can be looked up using css_from_id().
	 */
	int id;

71
	unsigned int flags;
72

73 74
	/* percpu_ref killing and RCU release */
	struct rcu_head rcu_head;
75
	struct work_struct destroy_work;
76 77 78 79
};

/* bits in struct cgroup_subsys_state flags field */
enum {
80
	CSS_ONLINE	= (1 << 1), /* between ->css_online() and ->css_offline() */
81 82
};

83 84 85 86 87
/**
 * css_get - obtain a reference on the specified css
 * @css: target css
 *
 * The caller must already have a reference.
88 89 90
 */
static inline void css_get(struct cgroup_subsys_state *css)
{
91
	percpu_ref_get(&css->refcnt);
92
}
P
Paul Menage 已提交
93

94
/**
95
 * css_tryget_online - try to obtain a reference on the specified css if online
96 97
 * @css: target css
 *
98 99
 * 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
100 101 102 103
 * reference on it - IOW, RCU protected access is good enough for this
 * function.  Returns %true if a reference count was successfully obtained;
 * %false otherwise.
 */
104
static inline bool css_tryget_online(struct cgroup_subsys_state *css)
P
Paul Menage 已提交
105
{
106
	return percpu_ref_tryget_live(&css->refcnt);
P
Paul Menage 已提交
107 108
}

109 110 111 112
/**
 * css_put - put a css reference
 * @css: target css
 *
113
 * Put a reference obtained via css_get() and css_tryget_online().
114
 */
115 116
static inline void css_put(struct cgroup_subsys_state *css)
{
117
	percpu_ref_put(&css->refcnt);
118 119
}

120 121 122
/* bits in struct cgroup flags field */
enum {
	/* Control Group is dead */
123
	CGRP_DEAD,
P
Paul Menage 已提交
124 125 126 127
	/*
	 * Control Group has previously had a child cgroup or a task,
	 * but no longer (only if CGRP_NOTIFY_ON_RELEASE is set)
	 */
128 129 130
	CGRP_RELEASABLE,
	/* Control Group requires release notifications to userspace */
	CGRP_NOTIFY_ON_RELEASE,
131
	/*
132 133 134
	 * 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.
135
	 */
136
	CGRP_CPUSET_CLONE_CHILDREN,
137 138
};

139
struct cgroup {
140 141 142
	/* self css with NULL ->ss, points back to this cgroup */
	struct cgroup_subsys_state self;

143 144
	unsigned long flags;		/* "unsigned long" so bitops work */

145 146 147
	/*
	 * idr allocated in-hierarchy ID.
	 *
148 149
	 * 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.
150 151
	 *
	 * Allocating/Removing ID must be protected by cgroup_mutex.
152 153
	 */
	int id;
T
Tejun Heo 已提交
154

155 156 157 158 159 160 161 162
	/*
	 * 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;

163 164 165 166 167 168 169
	/*
	 * We link our 'sibling' struct into our parent's 'children'.
	 * Our children link their 'sibling' into our 'children'.
	 */
	struct list_head sibling;	/* my parent's children */
	struct list_head children;	/* my children */

P
Paul Menage 已提交
170
	struct cgroup *parent;		/* my parent */
T
Tejun Heo 已提交
171
	struct kernfs_node *kn;		/* cgroup kernfs entry */
172
	struct kernfs_node *populated_kn; /* kn for "cgroup.subtree_populated" */
173

174 175 176 177 178 179 180 181
	/*
	 * Monotonically increasing unique serial number which defines a
	 * uniform order among all cgroups.  It's guaranteed that all
	 * ->children lists are in the ascending order of ->serial_nr.
	 * It's used to allow interrupting and resuming iterations.
	 */
	u64 serial_nr;

182
	/* the bitmask of subsystems enabled on the child cgroups */
183
	unsigned int child_subsys_mask;
184

185
	/* Private pointers for each registered subsystem */
186
	struct cgroup_subsys_state __rcu *subsys[CGROUP_SUBSYS_COUNT];
187

188
	struct cgroup_root *root;
189 190

	/*
191 192
	 * List of cgrp_cset_links pointing at css_sets with tasks in this
	 * cgroup.  Protected by css_set_lock.
193
	 */
194
	struct list_head cset_links;
195

T
Tejun Heo 已提交
196 197 198 199 200 201 202 203 204
	/*
	 * 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];

205 206 207 208 209 210
	/*
	 * Linked list running through all cgroups that can
	 * potentially be reaped by the release agent. Protected by
	 * release_list_lock
	 */
	struct list_head release_list;
211

212 213 214 215 216 217
	/*
	 * 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;
218

219 220
	/* used to wait for offlining of csses */
	wait_queue_head_t offline_waitq;
221 222
};

223 224
#define MAX_CGROUP_ROOT_NAMELEN 64

225
/* cgroup_root->flags */
226
enum {
227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242
	/*
	 * Unfortunately, cgroup core and various controllers are riddled
	 * with idiosyncrasies and pointless options.  The following flag,
	 * when set, will force sane behavior - some options are forced on,
	 * others are disallowed, and some controllers will change their
	 * hierarchical or other behaviors.
	 *
	 * The set of behaviors affected by this flag are still being
	 * determined and developed and the mount option for this flag is
	 * prefixed with __DEVEL__.  The prefix will be dropped once we
	 * reach the point where all behaviors are compatible with the
	 * planned unified hierarchy, which will automatically turn on this
	 * flag.
	 *
	 * The followings are the behaviors currently affected this flag.
	 *
243 244
	 * - Mount options "noprefix", "xattr", "clone_children",
	 *   "release_agent" and "name" are disallowed.
245 246 247 248 249 250
	 *
	 * - When mounting an existing superblock, mount options should
	 *   match.
	 *
	 * - Remount is disallowed.
	 *
251 252
	 * - rename(2) is disallowed.
	 *
253 254
	 * - "tasks" is removed.  Everything should be at process
	 *   granularity.  Use "cgroup.procs" instead.
255
	 *
256 257 258
	 * - "cgroup.procs" is not sorted.  pids will be unique unless they
	 *   got recycled inbetween reads.
	 *
259 260
	 * - "release_agent" and "notify_on_release" are removed.
	 *   Replacement notification mechanism will be implemented.
261
	 *
262
	 * - "cgroup.clone_children" is removed.
263
	 *
264 265 266 267 268 269
	 * - "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.
	 *
T
Tejun Heo 已提交
270 271 272
	 * - If mount is requested with sane_behavior but without any
	 *   subsystem, the default unified hierarchy is mounted.
	 *
273 274 275 276
	 * - 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.
	 *
277 278
	 * - cpuset: a task can be moved into an empty cpuset, and again it
	 *   takes masks of ancestors.
279
	 *
280 281
	 * - memcg: use_hierarchy is on by default and the cgroup file for
	 *   the flag is not created.
282
	 *
283
	 * - blkcg: blk-throttle becomes properly hierarchical.
284 285 286
	 */
	CGRP_ROOT_SANE_BEHAVIOR	= (1 << 0),

287 288
	CGRP_ROOT_NOPREFIX	= (1 << 1), /* mounted subsystems have no named prefix */
	CGRP_ROOT_XATTR		= (1 << 2), /* supports extended attributes */
289 290 291

	/* mount options live below bit 16 */
	CGRP_ROOT_OPTION_MASK	= (1 << 16) - 1,
292 293 294
};

/*
295
 * A cgroup_root represents the root of a cgroup hierarchy, and may be
T
Tejun Heo 已提交
296
 * associated with a kernfs_root to form an active hierarchy.  This is
297 298
 * internal to cgroup core.  Don't access directly from controllers.
 */
299
struct cgroup_root {
T
Tejun Heo 已提交
300
	struct kernfs_root *kf_root;
301

302
	/* The bitmask of subsystems attached to this hierarchy */
303
	unsigned int subsys_mask;
304

305 306 307
	/* Unique id for this hierarchy. */
	int hierarchy_id;

T
Tejun Heo 已提交
308
	/* The root cgroup.  Root is destroyed on its release. */
309
	struct cgroup cgrp;
310

311 312
	/* Number of cgroups in the hierarchy, used only for /proc/cgroups */
	atomic_t nr_cgrps;
313 314 315 316 317

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

	/* Hierarchy-specific flags */
318
	unsigned int flags;
319 320

	/* IDs for cgroups in this hierarchy */
321
	struct idr cgroup_idr;
322 323 324 325 326 327 328 329

	/* 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
330 331
/*
 * A css_set is a structure holding pointers to a set of
332 333
 * 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
334 335
 * list_add()/del() can bump the reference count on the entire cgroup
 * set for a task.
336 337 338 339 340
 */

struct css_set {

	/* Reference count */
341
	atomic_t refcount;
342

343 344 345 346 347 348
	/*
	 * List running through all cgroup groups in the same hash
	 * slot. Protected by css_set_lock
	 */
	struct hlist_node hlist;

349
	/*
T
Tejun Heo 已提交
350 351 352 353 354
	 * 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.
355 356
	 */
	struct list_head tasks;
T
Tejun Heo 已提交
357
	struct list_head mg_tasks;
358 359

	/*
360 361
	 * List of cgrp_cset_links pointing at cgroups referenced from this
	 * css_set.  Protected by css_set_lock.
362
	 */
363
	struct list_head cgrp_links;
364

T
Tejun Heo 已提交
365 366 367
	/* the default cgroup associated with this css_set */
	struct cgroup *dfl_cgrp;

368
	/*
T
Tejun Heo 已提交
369 370 371
	 * 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).
372 373
	 */
	struct cgroup_subsys_state *subsys[CGROUP_SUBSYS_COUNT];
374

375 376 377 378
	/*
	 * List of csets participating in the on-going migration either as
	 * source or destination.  Protected by cgroup_mutex.
	 */
379
	struct list_head mg_preload_node;
380 381 382 383 384 385 386 387 388 389 390 391
	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
392 393 394 395 396 397 398 399 400
	/*
	 * 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];

401 402
	/* For RCU-protected deletion */
	struct rcu_head rcu_head;
403 404
};

P
Paul Menage 已提交
405 406
/*
 * struct cftype: handler definitions for cgroup control files
407 408
 *
 * When reading/writing to a file:
L
Li Zefan 已提交
409
 *	- the cgroup to use is file->f_dentry->d_parent->d_fsdata
410 411 412
 *	- the 'cftype' of the file is file->f_dentry->d_fsdata
 */

413
/* cftype->flags */
414
enum {
L
Li Zefan 已提交
415 416
	CFTYPE_ONLY_ON_ROOT	= (1 << 0),	/* only create on root cgrp */
	CFTYPE_NOT_ON_ROOT	= (1 << 1),	/* don't create on root cgrp */
417
	CFTYPE_INSANE		= (1 << 2),	/* don't create if sane_behavior */
T
Tejun Heo 已提交
418
	CFTYPE_NO_PREFIX	= (1 << 3),	/* (DON'T USE FOR NEW FILES) no subsys prefix */
T
Tejun Heo 已提交
419
	CFTYPE_ONLY_ON_DFL	= (1 << 4),	/* only on default hierarchy */
420
};
421 422 423

#define MAX_CFTYPE_NAME		64

424
struct cftype {
P
Paul Menage 已提交
425 426
	/*
	 * By convention, the name should begin with the name of the
427 428
	 * subsystem, followed by a period.  Zero length string indicates
	 * end of cftype array.
P
Paul Menage 已提交
429
	 */
430 431
	char name[MAX_CFTYPE_NAME];
	int private;
L
Li Zefan 已提交
432 433 434 435
	/*
	 * If not 0, file mode is set to this value, otherwise it will
	 * be figured out automatically
	 */
A
Al Viro 已提交
436
	umode_t mode;
437 438

	/*
439
	 * The maximum length of string, excluding trailing nul, that can
440
	 * be passed to write.  If < PAGE_SIZE-1, PAGE_SIZE-1 is assumed.
441 442 443
	 */
	size_t max_write_len;

444 445 446
	/* CFTYPE_* flags */
	unsigned int flags;

447
	/*
T
Tejun Heo 已提交
448 449
	 * Fields used for internal bookkeeping.  Initialized automatically
	 * during registration.
450
	 */
T
Tejun Heo 已提交
451 452
	struct cgroup_subsys *ss;	/* NULL for cgroup core files */
	struct list_head node;		/* anchored at ss->cfts */
T
Tejun Heo 已提交
453 454
	struct kernfs_ops *kf_ops;

455
	/*
456
	 * read_u64() is a shortcut for the common case of returning a
457 458
	 * single integer. Use it in place of read()
	 */
459
	u64 (*read_u64)(struct cgroup_subsys_state *css, struct cftype *cft);
460 461 462
	/*
	 * read_s64() is a signed version of read_u64()
	 */
463
	s64 (*read_s64)(struct cgroup_subsys_state *css, struct cftype *cft);
464 465 466

	/* generic seq_file read interface */
	int (*seq_show)(struct seq_file *sf, void *v);
467

468 469 470 471 472
	/* 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);

473
	/*
474
	 * write_u64() is a shortcut for the common case of accepting
475 476 477
	 * a single integer (as parsed by simple_strtoull) from
	 * userspace. Use in place of write(); return 0 or error.
	 */
478 479
	int (*write_u64)(struct cgroup_subsys_state *css, struct cftype *cft,
			 u64 val);
480 481 482
	/*
	 * write_s64() is a signed version of write_u64()
	 */
483 484
	int (*write_s64)(struct cgroup_subsys_state *css, struct cftype *cft,
			 s64 val);
485

T
Tejun Heo 已提交
486 487 488 489 490 491 492 493 494
	/*
	 * 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
495 496 497
#ifdef CONFIG_DEBUG_LOCK_ALLOC
	struct lock_class_key	lockdep_key;
#endif
498 499
};

T
Tejun Heo 已提交
500
extern struct cgroup_root cgrp_dfl_root;
T
Tejun Heo 已提交
501
extern struct css_set init_css_set;
T
Tejun Heo 已提交
502 503 504 505 506 507

static inline bool cgroup_on_dfl(const struct cgroup *cgrp)
{
	return cgrp->root == &cgrp_dfl_root;
}

508 509 510 511 512 513 514 515 516
/*
 * See the comment above CGRP_ROOT_SANE_BEHAVIOR for details.  This
 * function can be called as long as @cgrp is accessible.
 */
static inline bool cgroup_sane_behavior(const struct cgroup *cgrp)
{
	return cgrp->root->flags & CGRP_ROOT_SANE_BEHAVIOR;
}

517 518 519 520 521 522
/* 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);
}

T
Tejun Heo 已提交
523 524 525
/* returns ino associated with a cgroup, 0 indicates unmounted root */
static inline ino_t cgroup_ino(struct cgroup *cgrp)
{
T
Tejun Heo 已提交
526 527
	if (cgrp->kn)
		return cgrp->kn->ino;
T
Tejun Heo 已提交
528 529 530 531
	else
		return 0;
}

T
Tejun Heo 已提交
532 533
/* cft/css accessors for cftype->write() operation */
static inline struct cftype *of_cft(struct kernfs_open_file *of)
534
{
T
Tejun Heo 已提交
535
	return of->kn->priv;
536 537
}

T
Tejun Heo 已提交
538 539 540 541 542 543 544 545 546 547 548 549
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);
}
550

T
Tejun Heo 已提交
551 552 553 554 555 556 557
/*
 * 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)
{
558
	return kernfs_name(cgrp->kn, buf, buflen);
T
Tejun Heo 已提交
559 560 561 562 563
}

static inline char * __must_check cgroup_path(struct cgroup *cgrp, char *buf,
					      size_t buflen)
{
564
	return kernfs_path(cgrp->kn, buf, buflen);
T
Tejun Heo 已提交
565 566 567 568
}

static inline void pr_cont_cgroup_name(struct cgroup *cgrp)
{
569
	pr_cont_kernfs_name(cgrp->kn);
T
Tejun Heo 已提交
570 571 572 573
}

static inline void pr_cont_cgroup_path(struct cgroup *cgrp)
{
574
	pr_cont_kernfs_path(cgrp->kn);
T
Tejun Heo 已提交
575 576 577 578
}

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

A
Aristeu Rozanski 已提交
579
int cgroup_add_cftypes(struct cgroup_subsys *ss, struct cftype *cfts);
580
int cgroup_rm_cftypes(struct cftype *cfts);
581

582
bool cgroup_is_descendant(struct cgroup *cgrp, struct cgroup *ancestor);
583

584 585 586 587 588 589 590 591 592 593 594 595 596
/*
 * 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
 */
597
#define cgroup_taskset_for_each(task, tset)				\
598
	for ((task) = cgroup_taskset_first((tset)); (task);		\
599
	     (task) = cgroup_taskset_next((tset)))
600

601 602 603 604
/*
 * Control Group subsystem type.
 * See Documentation/cgroups/cgroups.txt for details
 */
605 606

struct cgroup_subsys {
607 608 609 610 611 612 613 614 615 616 617
	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);
	void (*css_free)(struct cgroup_subsys_state *css);

	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);
618
	void (*fork)(struct task_struct *task);
619 620
	void (*exit)(struct cgroup_subsys_state *css,
		     struct cgroup_subsys_state *old_css,
621
		     struct task_struct *task);
622
	void (*bind)(struct cgroup_subsys_state *root_css);
623

624
	int disabled;
625
	int early_init;
626

627 628 629 630 631 632 633 634 635 636 637 638 639 640 641
	/*
	 * 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;

642
	/* the following two fields are initialized automtically during boot */
643
	int id;
644 645 646
#define MAX_CGROUP_TYPE_NAMELEN 32
	const char *name;

T
Tejun Heo 已提交
647
	/* link to parent, protected by cgroup_lock() */
648
	struct cgroup_root *root;
649

650 651 652
	/* idr for css->id */
	struct idr css_idr;

T
Tejun Heo 已提交
653 654 655 656 657
	/*
	 * List of cftypes.  Each entry is the first entry of an array
	 * terminated by zero length name.
	 */
	struct list_head cfts;
658

659
	/* base cftypes, automatically registered with subsys itself */
660
	struct cftype *base_cftypes;
661 662
};

663
#define SUBSYS(_x) extern struct cgroup_subsys _x ## _cgrp_subsys;
664 665 666
#include <linux/cgroup_subsys.h>
#undef SUBSYS

T
Tejun Heo 已提交
667 668 669 670 671 672 673 674 675 676
/**
 * css_parent - find the parent css
 * @css: the target cgroup_subsys_state
 *
 * Return the parent css of @css.  This function is guaranteed to return
 * non-NULL parent as long as @css isn't the root.
 */
static inline
struct cgroup_subsys_state *css_parent(struct cgroup_subsys_state *css)
{
677
	return css->parent;
T
Tejun Heo 已提交
678 679
}

680 681 682 683 684 685 686 687 688 689 690 691
/**
 * 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.
692
 */
T
Tejun Heo 已提交
693 694
#ifdef CONFIG_PROVE_RCU
extern struct mutex cgroup_mutex;
695
extern struct rw_semaphore css_set_rwsem;
696 697
#define task_css_set_check(task, __c)					\
	rcu_dereference_check((task)->cgroups,				\
698 699 700
		lockdep_is_held(&cgroup_mutex) ||			\
		lockdep_is_held(&css_set_rwsem) ||			\
		((task)->flags & PF_EXITING) || (__c))
T
Tejun Heo 已提交
701
#else
702 703
#define task_css_set_check(task, __c)					\
	rcu_dereference((task)->cgroups)
T
Tejun Heo 已提交
704
#endif
705

706
/**
707
 * task_css_check - obtain css for (task, subsys) w/ extra access conds
708 709 710 711 712 713 714
 * @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().
 */
715
#define task_css_check(task, subsys_id, __c)				\
716 717 718 719 720 721 722 723 724 725 726 727 728 729
	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);
}

/**
730
 * task_css - obtain css for (task, subsys)
731 732 733
 * @task: the target task
 * @subsys_id: the target subsystem ID
 *
734
 * See task_css_check().
735
 */
736 737
static inline struct cgroup_subsys_state *task_css(struct task_struct *task,
						   int subsys_id)
738
{
739
	return task_css_check(task, subsys_id, false);
740 741
}

T
Tejun Heo 已提交
742 743 744 745 746 747 748 749 750 751 752 753 754 755
/**
 * 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];
}

756 757
static inline struct cgroup *task_cgroup(struct task_struct *task,
					 int subsys_id)
758
{
759
	return task_css(task, subsys_id)->cgroup;
760 761
}

762 763
struct cgroup_subsys_state *css_next_child(struct cgroup_subsys_state *pos,
					   struct cgroup_subsys_state *parent);
764

765 766
struct cgroup_subsys_state *css_from_id(int id, struct cgroup_subsys *ss);

767
/**
768 769 770
 * 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
771
 *
772 773
 * Walk @parent's children.  Must be called under rcu_read_lock().  A child
 * css which hasn't finished ->css_online() or already has finished
774
 * ->css_offline() may show up during traversal and it's each subsystem's
775 776
 * responsibility to verify that each @pos is alive.
 *
777
 * If a subsystem synchronizes against the parent in its ->css_online() and
778
 * before starting iterating, a css which finished ->css_online() is
779
 * guaranteed to be visible in the future iterations.
780 781 782 783
 *
 * 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.
784
 */
785 786 787
#define css_for_each_child(pos, parent)					\
	for ((pos) = css_next_child(NULL, (parent)); (pos);		\
	     (pos) = css_next_child((pos), (parent)))
788

789 790 791 792 793 794
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);
795 796

/**
797 798 799
 * 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
800
 *
801 802
 * Walk @root's descendants.  @root is included in the iteration and the
 * first node to be visited.  Must be called under rcu_read_lock().  A
803
 * descendant css which hasn't finished ->css_online() or already has
804
 * finished ->css_offline() may show up during traversal and it's each
805 806
 * subsystem's responsibility to verify that each @pos is alive.
 *
807 808
 * If a subsystem synchronizes against the parent in its ->css_online() and
 * before starting iterating, and synchronizes against @pos on each
809
 * iteration, any descendant css which finished ->css_online() is
810 811 812 813 814
 * guaranteed to be visible in the future iterations.
 *
 * In other words, the following guarantees that a descendant can't escape
 * state updates of its ancestors.
 *
815
 * my_online(@css)
816
 * {
817 818
 *	Lock @css's parent and @css;
 *	Inherit state from the parent;
819 820 821
 *	Unlock both.
 * }
 *
822
 * my_update_state(@css)
823
 * {
824
 *	css_for_each_descendant_pre(@pos, @css) {
825
 *		Lock @pos;
826 827 828 829
 *		if (@pos == @css)
 *			Update @css's state;
 *		else
 *			Verify @pos is alive and inherit state from its parent;
830 831 832 833 834 835 836 837 838 839
 *		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
840
 * inheritance happens for any css after the latest update to its parent.
841 842 843 844 845
 *
 * 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
846
 * synchronize ->css_online() and ->css_offline() against tree-walking
847
 * operations.
848 849 850 851
 *
 * 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.
852
 */
853 854 855
#define css_for_each_descendant_pre(pos, css)				\
	for ((pos) = css_next_descendant_pre(NULL, (css)); (pos);	\
	     (pos) = css_next_descendant_pre((pos), (css)))
856

857 858 859
struct cgroup_subsys_state *
css_next_descendant_post(struct cgroup_subsys_state *pos,
			 struct cgroup_subsys_state *css);
860 861

/**
862 863 864
 * 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
865
 *
866
 * Similar to css_for_each_descendant_pre() but performs post-order
867 868 869
 * traversal instead.  @root is included in the iteration and the last
 * node to be visited.  Note that the walk visibility guarantee described
 * in pre-order walk doesn't apply the same to post-order walks.
870
 */
871 872 873
#define css_for_each_descendant_post(pos, css)				\
	for ((pos) = css_next_descendant_post(NULL, (css)); (pos);	\
	     (pos) = css_next_descendant_post((pos), (css)))
874

875 876
/* A css_task_iter should be treated as an opaque object */
struct css_task_iter {
877 878
	struct cgroup_subsys		*ss;

T
Tejun Heo 已提交
879 880 881 882 883 884
	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;
885 886
};

887 888 889 890
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
891

M
Michael S. Tsirkin 已提交
892
int cgroup_attach_task_all(struct task_struct *from, struct task_struct *);
893
int cgroup_transfer_tasks(struct cgroup *to, struct cgroup *from);
M
Michael S. Tsirkin 已提交
894

895 896
struct cgroup_subsys_state *css_tryget_online_from_dir(struct dentry *dentry,
						       struct cgroup_subsys *ss);
K
KAMEZAWA Hiroyuki 已提交
897

898 899 900 901
#else /* !CONFIG_CGROUPS */

static inline int cgroup_init_early(void) { return 0; }
static inline int cgroup_init(void) { return 0; }
902
static inline void cgroup_fork(struct task_struct *p) {}
903
static inline void cgroup_post_fork(struct task_struct *p) {}
904
static inline void cgroup_exit(struct task_struct *p) {}
905

B
Balbir Singh 已提交
906 907 908 909 910
static inline int cgroupstats_build(struct cgroupstats *stats,
					struct dentry *dentry)
{
	return -EINVAL;
}
911

912
/* No cgroups - nothing to do */
M
Michael S. Tsirkin 已提交
913 914 915 916 917
static inline int cgroup_attach_task_all(struct task_struct *from,
					 struct task_struct *t)
{
	return 0;
}
918

919 920 921
#endif /* !CONFIG_CGROUPS */

#endif /* _LINUX_CGROUP_H */