cgroup.h 26.9 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 25 26

#ifdef CONFIG_CGROUPS

27
struct cgroup_root;
28 29
struct cgroup_subsys;
struct inode;
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

40
extern int proc_cgroup_show(struct seq_file *, void *);
41

T
Tejun Heo 已提交
42
/* define the enumeration of all cgroup subsystems */
43
#define SUBSYS(_x) _x ## _cgrp_id,
44 45
enum cgroup_subsys_id {
#include <linux/cgroup_subsys.h>
46
	CGROUP_SUBSYS_COUNT,
47 48 49
};
#undef SUBSYS

50 51
/* Per-subsystem/per-cgroup state maintained by the system. */
struct cgroup_subsys_state {
52
	/* the cgroup that this css is attached to */
53 54
	struct cgroup *cgroup;

55 56 57
	/* the cgroup subsystem that this css is attached to */
	struct cgroup_subsys *ss;

58 59
	/* reference count - access via css_[try]get() and css_put() */
	struct percpu_ref refcnt;
60

61 62 63
	/* the parent css */
	struct cgroup_subsys_state *parent;

64
	unsigned long flags;
65

66 67
	/* percpu_ref killing and RCU release */
	struct rcu_head rcu_head;
68
	struct work_struct destroy_work;
69 70 71 72
};

/* bits in struct cgroup_subsys_state flags field */
enum {
73
	CSS_ROOT	= (1 << 0), /* this CSS is the root of the subsystem */
74
	CSS_ONLINE	= (1 << 1), /* between ->css_online() and ->css_offline() */
75 76
};

77 78 79 80 81
/**
 * css_get - obtain a reference on the specified css
 * @css: target css
 *
 * The caller must already have a reference.
82 83 84 85
 */
static inline void css_get(struct cgroup_subsys_state *css)
{
	/* We don't need to reference count the root state */
86
	if (!(css->flags & CSS_ROOT))
87
		percpu_ref_get(&css->refcnt);
88
}
P
Paul Menage 已提交
89

90 91 92 93 94 95 96 97 98 99
/**
 * css_tryget - try to obtain a reference on the specified css
 * @css: target css
 *
 * Obtain a reference on @css if it's alive.  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.
 */
P
Paul Menage 已提交
100 101
static inline bool css_tryget(struct cgroup_subsys_state *css)
{
102
	if (css->flags & CSS_ROOT)
P
Paul Menage 已提交
103
		return true;
104
	return percpu_ref_tryget(&css->refcnt);
P
Paul Menage 已提交
105 106
}

107 108 109 110 111 112
/**
 * css_put - put a css reference
 * @css: target css
 *
 * Put a reference obtained via css_get() and css_tryget().
 */
113 114
static inline void css_put(struct cgroup_subsys_state *css)
{
115
	if (!(css->flags & CSS_ROOT))
116
		percpu_ref_put(&css->refcnt);
117 118
}

119 120 121
/* bits in struct cgroup flags field */
enum {
	/* Control Group is dead */
122
	CGRP_DEAD,
P
Paul Menage 已提交
123 124 125 126
	/*
	 * Control Group has previously had a child cgroup or a task,
	 * but no longer (only if CGRP_NOTIFY_ON_RELEASE is set)
	 */
127 128 129
	CGRP_RELEASABLE,
	/* Control Group requires release notifications to userspace */
	CGRP_NOTIFY_ON_RELEASE,
130
	/*
131 132 133
	 * 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.
134
	 */
135
	CGRP_CPUSET_CLONE_CHILDREN,
136 137
	/* see the comment above CGRP_ROOT_SANE_BEHAVIOR for details */
	CGRP_SANE_BEHAVIOR,
138 139
};

140 141 142
struct cgroup {
	unsigned long flags;		/* "unsigned long" so bitops work */

143 144 145 146 147
	/*
	 * idr allocated in-hierarchy ID.
	 *
	 * The ID of the root cgroup is always 0, and a new cgroup
	 * will be assigned with a smallest available ID.
148 149
	 *
	 * Allocating/Removing ID must be protected by cgroup_mutex.
150 151
	 */
	int id;
T
Tejun Heo 已提交
152

153 154 155
	/* the number of attached css's */
	int nr_css;

T
Tejun Heo 已提交
156 157
	atomic_t refcnt;

158 159 160 161 162 163 164
	/*
	 * 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
165
	struct cgroup *parent;		/* my parent */
T
Tejun Heo 已提交
166
	struct kernfs_node *kn;		/* cgroup kernfs entry */
167

168 169 170 171 172 173 174 175
	/*
	 * 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;

176 177
	/* the bitmask of subsystems enabled on the child cgroups */
	unsigned long child_subsys_mask;
178

179
	/* Private pointers for each registered subsystem */
180
	struct cgroup_subsys_state __rcu *subsys[CGROUP_SUBSYS_COUNT];
181

182
	struct cgroup_root *root;
183 184

	/*
185 186
	 * List of cgrp_cset_links pointing at css_sets with tasks in this
	 * cgroup.  Protected by css_set_lock.
187
	 */
188
	struct list_head cset_links;
189

T
Tejun Heo 已提交
190 191 192 193 194 195 196 197 198
	/*
	 * 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];

199 200 201 202 203 204
	/*
	 * 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;
205

206 207 208 209 210 211
	/*
	 * 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;
212

T
Tejun Heo 已提交
213 214 215
	/* dummy css with NULL ->ss, points back to this cgroup */
	struct cgroup_subsys_state dummy_css;

216
	/* For css percpu_ref killing and RCU-protected deletion */
217
	struct rcu_head rcu_head;
218
	struct work_struct destroy_work;
219 220
};

221 222
#define MAX_CGROUP_ROOT_NAMELEN 64

223
/* cgroup_root->flags */
224
enum {
225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240
	/*
	 * 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.
	 *
241 242
	 * - Mount options "noprefix", "xattr", "clone_children",
	 *   "release_agent" and "name" are disallowed.
243 244 245 246 247 248
	 *
	 * - When mounting an existing superblock, mount options should
	 *   match.
	 *
	 * - Remount is disallowed.
	 *
249 250
	 * - rename(2) is disallowed.
	 *
251 252
	 * - "tasks" is removed.  Everything should be at process
	 *   granularity.  Use "cgroup.procs" instead.
253
	 *
254 255 256
	 * - "cgroup.procs" is not sorted.  pids will be unique unless they
	 *   got recycled inbetween reads.
	 *
257 258
	 * - "release_agent" and "notify_on_release" are removed.
	 *   Replacement notification mechanism will be implemented.
259
	 *
260
	 * - "cgroup.clone_children" is removed.
261
	 *
T
Tejun Heo 已提交
262 263 264
	 * - If mount is requested with sane_behavior but without any
	 *   subsystem, the default unified hierarchy is mounted.
	 *
265 266 267 268
	 * - 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.
	 *
269 270
	 * - cpuset: a task can be moved into an empty cpuset, and again it
	 *   takes masks of ancestors.
271
	 *
272 273
	 * - memcg: use_hierarchy is on by default and the cgroup file for
	 *   the flag is not created.
274
	 *
275
	 * - blkcg: blk-throttle becomes properly hierarchical.
276 277 278
	 */
	CGRP_ROOT_SANE_BEHAVIOR	= (1 << 0),

279 280
	CGRP_ROOT_NOPREFIX	= (1 << 1), /* mounted subsystems have no named prefix */
	CGRP_ROOT_XATTR		= (1 << 2), /* supports extended attributes */
281 282 283

	/* mount options live below bit 16 */
	CGRP_ROOT_OPTION_MASK	= (1 << 16) - 1,
284 285 286
};

/*
287
 * A cgroup_root represents the root of a cgroup hierarchy, and may be
T
Tejun Heo 已提交
288
 * associated with a kernfs_root to form an active hierarchy.  This is
289 290
 * internal to cgroup core.  Don't access directly from controllers.
 */
291
struct cgroup_root {
T
Tejun Heo 已提交
292
	struct kernfs_root *kf_root;
293

294 295 296
	/* The bitmask of subsystems attached to this hierarchy */
	unsigned long subsys_mask;

297 298 299
	/* Unique id for this hierarchy. */
	int hierarchy_id;

T
Tejun Heo 已提交
300
	/* The root cgroup.  Root is destroyed on its release. */
301
	struct cgroup cgrp;
302

303 304
	/* Number of cgroups in the hierarchy, used only for /proc/cgroups */
	atomic_t nr_cgrps;
305 306 307 308 309 310 311 312

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

	/* Hierarchy-specific flags */
	unsigned long flags;

	/* IDs for cgroups in this hierarchy */
313
	struct idr cgroup_idr;
314 315 316 317 318 319 320 321

	/* 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
322 323
/*
 * A css_set is a structure holding pointers to a set of
324 325
 * 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
326 327
 * list_add()/del() can bump the reference count on the entire cgroup
 * set for a task.
328 329 330 331 332
 */

struct css_set {

	/* Reference count */
333
	atomic_t refcount;
334

335 336 337 338 339 340
	/*
	 * List running through all cgroup groups in the same hash
	 * slot. Protected by css_set_lock
	 */
	struct hlist_node hlist;

341
	/*
T
Tejun Heo 已提交
342 343 344 345 346
	 * 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.
347 348
	 */
	struct list_head tasks;
T
Tejun Heo 已提交
349
	struct list_head mg_tasks;
350 351

	/*
352 353
	 * List of cgrp_cset_links pointing at cgroups referenced from this
	 * css_set.  Protected by css_set_lock.
354
	 */
355
	struct list_head cgrp_links;
356

T
Tejun Heo 已提交
357 358 359
	/* the default cgroup associated with this css_set */
	struct cgroup *dfl_cgrp;

360
	/*
T
Tejun Heo 已提交
361 362 363
	 * 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).
364 365
	 */
	struct cgroup_subsys_state *subsys[CGROUP_SUBSYS_COUNT];
366

367 368 369 370
	/*
	 * List of csets participating in the on-going migration either as
	 * source or destination.  Protected by cgroup_mutex.
	 */
371
	struct list_head mg_preload_node;
372 373 374 375 376 377 378 379 380 381 382 383
	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
384 385 386 387 388 389 390 391 392
	/*
	 * 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];

393 394
	/* For RCU-protected deletion */
	struct rcu_head rcu_head;
395 396
};

P
Paul Menage 已提交
397 398
/*
 * struct cftype: handler definitions for cgroup control files
399 400
 *
 * When reading/writing to a file:
L
Li Zefan 已提交
401
 *	- the cgroup to use is file->f_dentry->d_parent->d_fsdata
402 403 404
 *	- the 'cftype' of the file is file->f_dentry->d_fsdata
 */

405
/* cftype->flags */
406
enum {
L
Li Zefan 已提交
407 408
	CFTYPE_ONLY_ON_ROOT	= (1 << 0),	/* only create on root cgrp */
	CFTYPE_NOT_ON_ROOT	= (1 << 1),	/* don't create on root cgrp */
409
	CFTYPE_INSANE		= (1 << 2),	/* don't create if sane_behavior */
T
Tejun Heo 已提交
410
	CFTYPE_NO_PREFIX	= (1 << 3),	/* (DON'T USE FOR NEW FILES) no subsys prefix */
T
Tejun Heo 已提交
411
	CFTYPE_ONLY_ON_DFL	= (1 << 4),	/* only on default hierarchy */
412
};
413 414 415

#define MAX_CFTYPE_NAME		64

416
struct cftype {
P
Paul Menage 已提交
417 418
	/*
	 * By convention, the name should begin with the name of the
419 420
	 * subsystem, followed by a period.  Zero length string indicates
	 * end of cftype array.
P
Paul Menage 已提交
421
	 */
422 423
	char name[MAX_CFTYPE_NAME];
	int private;
L
Li Zefan 已提交
424 425 426 427
	/*
	 * If not 0, file mode is set to this value, otherwise it will
	 * be figured out automatically
	 */
A
Al Viro 已提交
428
	umode_t mode;
429 430

	/*
431 432 433
	 * The maximum length of string, excluding trailing nul, that can
	 * be passed to write_string.  If < PAGE_SIZE-1, PAGE_SIZE-1 is
	 * assumed.
434 435 436
	 */
	size_t max_write_len;

437 438 439
	/* CFTYPE_* flags */
	unsigned int flags;

440
	/*
T
Tejun Heo 已提交
441 442
	 * Fields used for internal bookkeeping.  Initialized automatically
	 * during registration.
443
	 */
T
Tejun Heo 已提交
444 445
	struct cgroup_subsys *ss;	/* NULL for cgroup core files */
	struct list_head node;		/* anchored at ss->cfts */
T
Tejun Heo 已提交
446 447
	struct kernfs_ops *kf_ops;

448
	/*
449
	 * read_u64() is a shortcut for the common case of returning a
450 451
	 * single integer. Use it in place of read()
	 */
452
	u64 (*read_u64)(struct cgroup_subsys_state *css, struct cftype *cft);
453 454 455
	/*
	 * read_s64() is a signed version of read_u64()
	 */
456
	s64 (*read_s64)(struct cgroup_subsys_state *css, struct cftype *cft);
457 458 459

	/* generic seq_file read interface */
	int (*seq_show)(struct seq_file *sf, void *v);
460

461 462 463 464 465
	/* 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);

466
	/*
467
	 * write_u64() is a shortcut for the common case of accepting
468 469 470
	 * a single integer (as parsed by simple_strtoull) from
	 * userspace. Use in place of write(); return 0 or error.
	 */
471 472
	int (*write_u64)(struct cgroup_subsys_state *css, struct cftype *cft,
			 u64 val);
473 474 475
	/*
	 * write_s64() is a signed version of write_u64()
	 */
476 477
	int (*write_s64)(struct cgroup_subsys_state *css, struct cftype *cft,
			 s64 val);
478

479 480 481 482 483
	/*
	 * write_string() is passed a nul-terminated kernelspace
	 * buffer of maximum length determined by max_write_len.
	 * Returns 0 or -ve error code.
	 */
484
	int (*write_string)(struct cgroup_subsys_state *css, struct cftype *cft,
485
			    char *buffer);
486 487 488 489 490 491
	/*
	 * trigger() callback can be used to get some kick from the
	 * userspace, when the actual string written is not important
	 * at all. The private field can be used to determine the
	 * kick type for multiplexing.
	 */
492
	int (*trigger)(struct cgroup_subsys_state *css, unsigned int event);
T
Tejun Heo 已提交
493 494 495 496

#ifdef CONFIG_DEBUG_LOCK_ALLOC
	struct lock_class_key	lockdep_key;
#endif
497 498
};

T
Tejun Heo 已提交
499 500 501 502 503 504 505
extern struct cgroup_root cgrp_dfl_root;

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

506 507 508 509 510 511 512 513 514
/*
 * 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;
}

515 516 517 518 519 520
/* 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
521 522 523
/* 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
524 525
	if (cgrp->kn)
		return cgrp->kn->ino;
T
Tejun Heo 已提交
526 527 528 529
	else
		return 0;
}

530 531
static inline struct cftype *seq_cft(struct seq_file *seq)
{
T
Tejun Heo 已提交
532 533 534
	struct kernfs_open_file *of = seq->private;

	return of->kn->priv;
535 536
}

537 538
struct cgroup_subsys_state *seq_css(struct seq_file *seq);

T
Tejun Heo 已提交
539 540 541 542 543 544 545
/*
 * 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)
{
546
	return kernfs_name(cgrp->kn, buf, buflen);
T
Tejun Heo 已提交
547 548 549 550 551
}

static inline char * __must_check cgroup_path(struct cgroup *cgrp, char *buf,
					      size_t buflen)
{
552
	return kernfs_path(cgrp->kn, buf, buflen);
T
Tejun Heo 已提交
553 554 555 556
}

static inline void pr_cont_cgroup_name(struct cgroup *cgrp)
{
557
	pr_cont_kernfs_name(cgrp->kn);
T
Tejun Heo 已提交
558 559 560 561
}

static inline void pr_cont_cgroup_path(struct cgroup *cgrp)
{
562
	pr_cont_kernfs_path(cgrp->kn);
T
Tejun Heo 已提交
563 564 565 566
}

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

A
Aristeu Rozanski 已提交
567
int cgroup_add_cftypes(struct cgroup_subsys *ss, struct cftype *cfts);
568
int cgroup_rm_cftypes(struct cftype *cfts);
569

570
bool cgroup_is_descendant(struct cgroup *cgrp, struct cgroup *ancestor);
571

572 573 574 575 576 577 578 579 580 581 582 583 584
/*
 * 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
 */
585
#define cgroup_taskset_for_each(task, tset)				\
586
	for ((task) = cgroup_taskset_first((tset)); (task);		\
587
	     (task) = cgroup_taskset_next((tset)))
588

589 590 591 592
/*
 * Control Group subsystem type.
 * See Documentation/cgroups/cgroups.txt for details
 */
593 594

struct cgroup_subsys {
595 596 597 598 599 600 601 602 603 604 605
	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);
606
	void (*fork)(struct task_struct *task);
607 608
	void (*exit)(struct cgroup_subsys_state *css,
		     struct cgroup_subsys_state *old_css,
609
		     struct task_struct *task);
610
	void (*bind)(struct cgroup_subsys_state *root_css);
611

612
	int disabled;
613
	int early_init;
614

615 616 617 618 619 620 621 622 623 624 625 626 627 628 629
	/*
	 * 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;

630
	/* the following two fields are initialized automtically during boot */
631
	int id;
632 633 634
#define MAX_CGROUP_TYPE_NAMELEN 32
	const char *name;

T
Tejun Heo 已提交
635
	/* link to parent, protected by cgroup_lock() */
636
	struct cgroup_root *root;
637

T
Tejun Heo 已提交
638 639 640 641 642
	/*
	 * List of cftypes.  Each entry is the first entry of an array
	 * terminated by zero length name.
	 */
	struct list_head cfts;
643

644
	/* base cftypes, automatically registered with subsys itself */
645
	struct cftype *base_cftypes;
646 647
};

648
#define SUBSYS(_x) extern struct cgroup_subsys _x ## _cgrp_subsys;
649 650 651
#include <linux/cgroup_subsys.h>
#undef SUBSYS

T
Tejun Heo 已提交
652 653 654 655 656 657 658 659 660 661
/**
 * 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)
{
662
	return css->parent;
T
Tejun Heo 已提交
663 664
}

665 666 667 668 669 670 671 672 673 674 675 676
/**
 * 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.
677
 */
T
Tejun Heo 已提交
678 679
#ifdef CONFIG_PROVE_RCU
extern struct mutex cgroup_mutex;
680
extern struct rw_semaphore css_set_rwsem;
681 682
#define task_css_set_check(task, __c)					\
	rcu_dereference_check((task)->cgroups,				\
683 684 685
		lockdep_is_held(&cgroup_mutex) ||			\
		lockdep_is_held(&css_set_rwsem) ||			\
		((task)->flags & PF_EXITING) || (__c))
T
Tejun Heo 已提交
686
#else
687 688
#define task_css_set_check(task, __c)					\
	rcu_dereference((task)->cgroups)
T
Tejun Heo 已提交
689
#endif
690

691
/**
692
 * task_css_check - obtain css for (task, subsys) w/ extra access conds
693 694 695 696 697 698 699
 * @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().
 */
700
#define task_css_check(task, subsys_id, __c)				\
701 702 703 704 705 706 707 708 709 710 711 712 713 714
	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);
}

/**
715
 * task_css - obtain css for (task, subsys)
716 717 718
 * @task: the target task
 * @subsys_id: the target subsystem ID
 *
719
 * See task_css_check().
720
 */
721 722
static inline struct cgroup_subsys_state *task_css(struct task_struct *task,
						   int subsys_id)
723
{
724
	return task_css_check(task, subsys_id, false);
725 726
}

727 728
static inline struct cgroup *task_cgroup(struct task_struct *task,
					 int subsys_id)
729
{
730
	return task_css(task, subsys_id)->cgroup;
731 732
}

733 734
struct cgroup_subsys_state *css_next_child(struct cgroup_subsys_state *pos,
					   struct cgroup_subsys_state *parent);
735

736 737
struct cgroup_subsys_state *css_from_id(int id, struct cgroup_subsys *ss);

738
/**
739 740 741
 * 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
742
 *
743 744
 * Walk @parent's children.  Must be called under rcu_read_lock().  A child
 * css which hasn't finished ->css_online() or already has finished
745
 * ->css_offline() may show up during traversal and it's each subsystem's
746 747
 * responsibility to verify that each @pos is alive.
 *
748
 * If a subsystem synchronizes against the parent in its ->css_online() and
749
 * before starting iterating, a css which finished ->css_online() is
750
 * guaranteed to be visible in the future iterations.
751 752 753 754
 *
 * 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.
755
 */
756 757 758
#define css_for_each_child(pos, parent)					\
	for ((pos) = css_next_child(NULL, (parent)); (pos);		\
	     (pos) = css_next_child((pos), (parent)))
759

760 761 762 763 764 765
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);
766 767

/**
768 769 770
 * 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
771
 *
772 773
 * 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
774
 * descendant css which hasn't finished ->css_online() or already has
775
 * finished ->css_offline() may show up during traversal and it's each
776 777
 * subsystem's responsibility to verify that each @pos is alive.
 *
778 779
 * If a subsystem synchronizes against the parent in its ->css_online() and
 * before starting iterating, and synchronizes against @pos on each
780
 * iteration, any descendant css which finished ->css_online() is
781 782 783 784 785
 * 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.
 *
786
 * my_online(@css)
787
 * {
788 789
 *	Lock @css's parent and @css;
 *	Inherit state from the parent;
790 791 792
 *	Unlock both.
 * }
 *
793
 * my_update_state(@css)
794
 * {
795
 *	css_for_each_descendant_pre(@pos, @css) {
796
 *		Lock @pos;
797 798 799 800
 *		if (@pos == @css)
 *			Update @css's state;
 *		else
 *			Verify @pos is alive and inherit state from its parent;
801 802 803 804 805 806 807 808 809 810
 *		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
811
 * inheritance happens for any css after the latest update to its parent.
812 813 814 815 816
 *
 * 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
817
 * synchronize ->css_online() and ->css_offline() against tree-walking
818
 * operations.
819 820 821 822
 *
 * 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.
823
 */
824 825 826
#define css_for_each_descendant_pre(pos, css)				\
	for ((pos) = css_next_descendant_pre(NULL, (css)); (pos);	\
	     (pos) = css_next_descendant_pre((pos), (css)))
827

828 829 830
struct cgroup_subsys_state *
css_next_descendant_post(struct cgroup_subsys_state *pos,
			 struct cgroup_subsys_state *css);
831 832

/**
833 834 835
 * 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
836
 *
837
 * Similar to css_for_each_descendant_pre() but performs post-order
838 839 840
 * 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.
841
 */
842 843 844
#define css_for_each_descendant_post(pos, css)				\
	for ((pos) = css_next_descendant_post(NULL, (css)); (pos);	\
	     (pos) = css_next_descendant_post((pos), (css)))
845

846 847
/* A css_task_iter should be treated as an opaque object */
struct css_task_iter {
848 849
	struct cgroup_subsys		*ss;

T
Tejun Heo 已提交
850 851 852 853 854 855
	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;
856 857
};

858 859 860 861
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
862

M
Michael S. Tsirkin 已提交
863
int cgroup_attach_task_all(struct task_struct *from, struct task_struct *);
864
int cgroup_transfer_tasks(struct cgroup *to, struct cgroup *from);
M
Michael S. Tsirkin 已提交
865

866 867
struct cgroup_subsys_state *css_tryget_from_dir(struct dentry *dentry,
						struct cgroup_subsys *ss);
K
KAMEZAWA Hiroyuki 已提交
868

869 870 871 872
#else /* !CONFIG_CGROUPS */

static inline int cgroup_init_early(void) { return 0; }
static inline int cgroup_init(void) { return 0; }
873
static inline void cgroup_fork(struct task_struct *p) {}
874
static inline void cgroup_post_fork(struct task_struct *p) {}
875
static inline void cgroup_exit(struct task_struct *p) {}
876

B
Balbir Singh 已提交
877 878 879 880 881
static inline int cgroupstats_build(struct cgroupstats *stats,
					struct dentry *dentry)
{
	return -EINVAL;
}
882

883
/* No cgroups - nothing to do */
M
Michael S. Tsirkin 已提交
884 885 886 887 888
static inline int cgroup_attach_task_all(struct task_struct *from,
					 struct task_struct *t)
{
	return 0;
}
889

890 891 892
#endif /* !CONFIG_CGROUPS */

#endif /* _LINUX_CGROUP_H */