cgroup.h 23.0 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/prio_heap.h>
18
#include <linux/rwsem.h>
K
KAMEZAWA Hiroyuki 已提交
19
#include <linux/idr.h>
20
#include <linux/workqueue.h>
A
Aristeu Rozanski 已提交
21
#include <linux/xattr.h>
22 23 24 25 26 27

#ifdef CONFIG_CGROUPS

struct cgroupfs_root;
struct cgroup_subsys;
struct inode;
28
struct cgroup;
K
KAMEZAWA Hiroyuki 已提交
29
struct css_id;
30 31 32

extern int cgroup_init_early(void);
extern int cgroup_init(void);
33
extern int cgroup_lock_is_held(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, int run_callbacks);
B
Balbir Singh 已提交
37 38
extern int cgroupstats_build(struct cgroupstats *stats,
				struct dentry *dentry);
39
extern int cgroup_load_subsys(struct cgroup_subsys *ss);
B
Ben Blum 已提交
40
extern void cgroup_unload_subsys(struct cgroup_subsys *ss);
41

42
extern const struct file_operations proc_cgroup_operations;
43

44 45 46 47 48
/*
 * Define the enumeration of all cgroup subsystems.
 *
 * We define ids for builtin subsystems and then modular ones.
 */
49 50
#define SUBSYS(_x) _x ## _subsys_id,
enum cgroup_subsys_id {
51
#define IS_SUBSYS_ENABLED(option) IS_BUILTIN(option)
52
#include <linux/cgroup_subsys.h>
53 54 55 56 57 58 59 60
#undef IS_SUBSYS_ENABLED
	CGROUP_BUILTIN_SUBSYS_COUNT,

	__CGROUP_SUBSYS_TEMP_PLACEHOLDER = CGROUP_BUILTIN_SUBSYS_COUNT - 1,

#define IS_SUBSYS_ENABLED(option) IS_MODULE(option)
#include <linux/cgroup_subsys.h>
#undef IS_SUBSYS_ENABLED
61
	CGROUP_SUBSYS_COUNT,
62 63 64
};
#undef SUBSYS

65 66
/* Per-subsystem/per-cgroup state maintained by the system. */
struct cgroup_subsys_state {
P
Paul Menage 已提交
67 68
	/*
	 * The cgroup that this subsystem is attached to. Useful
69
	 * for subsystems that want to know about the cgroup
P
Paul Menage 已提交
70 71
	 * hierarchy structure
	 */
72 73
	struct cgroup *cgroup;

P
Paul Menage 已提交
74 75
	/*
	 * State maintained by the cgroup system to allow subsystems
P
Paul Menage 已提交
76
	 * to be "busy". Should be accessed via css_get(),
77
	 * css_tryget() and css_put().
P
Paul Menage 已提交
78
	 */
79 80 81 82

	atomic_t refcnt;

	unsigned long flags;
K
KAMEZAWA Hiroyuki 已提交
83
	/* ID for this css, if possible */
A
Arnd Bergmann 已提交
84
	struct css_id __rcu *id;
85 86 87

	/* Used to put @cgroup->dentry on the last css_put() */
	struct work_struct dput_work;
88 89 90 91
};

/* bits in struct cgroup_subsys_state flags field */
enum {
92
	CSS_ROOT	= (1 << 0), /* this CSS is the root of the subsystem */
93
	CSS_ONLINE	= (1 << 1), /* between ->css_online() and ->css_offline() */
94 95
};

96 97 98 99 100 101
/* Caller must verify that the css is not for root cgroup */
static inline void __css_get(struct cgroup_subsys_state *css, int count)
{
	atomic_add(count, &css->refcnt);
}

102
/*
P
Paul Menage 已提交
103 104 105 106
 * Call css_get() to hold a reference on the css; it can be used
 * for a reference obtained via:
 * - an existing ref-counted reference to the css
 * - task->cgroups for a locked task
107 108 109 110 111
 */

static inline void css_get(struct cgroup_subsys_state *css)
{
	/* We don't need to reference count the root state */
112
	if (!(css->flags & CSS_ROOT))
113
		__css_get(css, 1);
114
}
P
Paul Menage 已提交
115 116 117 118 119 120 121

/*
 * Call css_tryget() to take a reference on a css if your existing
 * (known-valid) reference isn't already ref-counted. Returns false if
 * the css has been destroyed.
 */

122
extern bool __css_tryget(struct cgroup_subsys_state *css);
P
Paul Menage 已提交
123 124
static inline bool css_tryget(struct cgroup_subsys_state *css)
{
125
	if (css->flags & CSS_ROOT)
P
Paul Menage 已提交
126
		return true;
127
	return __css_tryget(css);
P
Paul Menage 已提交
128 129
}

130 131
/*
 * css_put() should be called to release a reference taken by
P
Paul Menage 已提交
132
 * css_get() or css_tryget()
133 134
 */

135
extern void __css_put(struct cgroup_subsys_state *css);
136 137
static inline void css_put(struct cgroup_subsys_state *css)
{
138
	if (!(css->flags & CSS_ROOT))
139
		__css_put(css);
140 141
}

142 143 144 145
/* bits in struct cgroup flags field */
enum {
	/* Control Group is dead */
	CGRP_REMOVED,
P
Paul Menage 已提交
146 147 148 149
	/*
	 * Control Group has previously had a child cgroup or a task,
	 * but no longer (only if CGRP_NOTIFY_ON_RELEASE is set)
	 */
150 151 152
	CGRP_RELEASABLE,
	/* Control Group requires release notifications to userspace */
	CGRP_NOTIFY_ON_RELEASE,
153
	/*
154 155 156
	 * 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.
157
	 */
158
	CGRP_CPUSET_CLONE_CHILDREN,
159 160
};

161 162 163 164 165
struct cgroup_name {
	struct rcu_head rcu_head;
	char name[];
};

166 167 168
struct cgroup {
	unsigned long flags;		/* "unsigned long" so bitops work */

P
Paul Menage 已提交
169 170 171 172
	/*
	 * count users of this cgroup. >0 means busy, but doesn't
	 * necessarily indicate the number of tasks in the cgroup
	 */
173 174
	atomic_t count;

T
Tejun Heo 已提交
175 176
	int id;				/* ida allocated in-hierarchy ID */

177 178 179 180 181 182
	/*
	 * 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 */
T
Tejun Heo 已提交
183
	struct list_head files;		/* my files */
184

P
Paul Menage 已提交
185
	struct cgroup *parent;		/* my parent */
186
	struct dentry *dentry;		/* cgroup fs entry, RCU protected */
187

188 189 190 191 192 193 194 195 196 197 198 199 200
	/*
	 * This is a copy of dentry->d_name, and it's needed because
	 * we can't use dentry->d_name in cgroup_path().
	 *
	 * You must acquire rcu_read_lock() to access cgrp->name, and
	 * the only place that can change it is rename(), which is
	 * protected by parent dir's i_mutex.
	 *
	 * Normally you should use cgroup_name() wrapper rather than
	 * access it directly.
	 */
	struct cgroup_name __rcu *name;

201 202 203 204 205
	/* Private pointers for each registered subsystem */
	struct cgroup_subsys_state *subsys[CGROUP_SUBSYS_COUNT];

	struct cgroupfs_root *root;
	struct cgroup *top_cgroup;
206 207 208 209 210 211

	/*
	 * List of cg_cgroup_links pointing at css_sets with
	 * tasks in this cgroup. Protected by css_set_lock
	 */
	struct list_head css_sets;
212

213
	struct list_head allcg_node;	/* cgroupfs_root->allcg_list */
214
	struct list_head cft_q_node;	/* used during cftype add/rm */
215

216 217 218 219 220 221
	/*
	 * 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;
222

223 224 225 226 227 228
	/*
	 * 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;
229 230 231

	/* For RCU-protected deletion */
	struct rcu_head rcu_head;
232
	struct work_struct free_work;
233

L
Lucas De Marchi 已提交
234
	/* List of events which userspace want to receive */
235 236
	struct list_head event_list;
	spinlock_t event_list_lock;
A
Aristeu Rozanski 已提交
237 238 239

	/* directory xattrs */
	struct simple_xattrs xattrs;
240 241
};

P
Paul Menage 已提交
242 243
/*
 * A css_set is a structure holding pointers to a set of
244 245
 * 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
246 247
 * list_add()/del() can bump the reference count on the entire cgroup
 * set for a task.
248 249 250 251 252
 */

struct css_set {

	/* Reference count */
253
	atomic_t refcount;
254

255 256 257 258 259 260
	/*
	 * List running through all cgroup groups in the same hash
	 * slot. Protected by css_set_lock
	 */
	struct hlist_node hlist;

261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276
	/*
	 * List running through all tasks using this cgroup
	 * group. Protected by css_set_lock
	 */
	struct list_head tasks;

	/*
	 * List of cg_cgroup_link objects on link chains from
	 * cgroups referenced from this css_set. Protected by
	 * css_set_lock
	 */
	struct list_head cg_links;

	/*
	 * Set of subsystem states, one for each subsystem. This array
	 * is immutable after creation apart from the init_css_set
B
Ben Blum 已提交
277 278
	 * during subsystem registration (at boot time) and modular subsystem
	 * loading/unloading.
279 280
	 */
	struct cgroup_subsys_state *subsys[CGROUP_SUBSYS_COUNT];
281 282 283

	/* For RCU-protected deletion */
	struct rcu_head rcu_head;
284 285
};

286 287 288 289 290 291 292 293 294 295
/*
 * cgroup_map_cb is an abstract callback API for reporting map-valued
 * control files
 */

struct cgroup_map_cb {
	int (*fill)(struct cgroup_map_cb *cb, const char *key, u64 value);
	void *state;
};

P
Paul Menage 已提交
296 297
/*
 * struct cftype: handler definitions for cgroup control files
298 299
 *
 * When reading/writing to a file:
L
Li Zefan 已提交
300
 *	- the cgroup to use is file->f_dentry->d_parent->d_fsdata
301 302 303
 *	- the 'cftype' of the file is file->f_dentry->d_fsdata
 */

304 305
/* cftype->flags */
#define CFTYPE_ONLY_ON_ROOT	(1U << 0)	/* only create on root cg */
306
#define CFTYPE_NOT_ON_ROOT	(1U << 1)	/* don't create on root cg */
307 308 309

#define MAX_CFTYPE_NAME		64

310
struct cftype {
P
Paul Menage 已提交
311 312
	/*
	 * By convention, the name should begin with the name of the
313 314
	 * subsystem, followed by a period.  Zero length string indicates
	 * end of cftype array.
P
Paul Menage 已提交
315
	 */
316 317
	char name[MAX_CFTYPE_NAME];
	int private;
L
Li Zefan 已提交
318 319 320 321
	/*
	 * If not 0, file mode is set to this value, otherwise it will
	 * be figured out automatically
	 */
A
Al Viro 已提交
322
	umode_t mode;
323 324 325 326 327 328 329

	/*
	 * If non-zero, defines the maximum length of string that can
	 * be passed to write_string; defaults to 64
	 */
	size_t max_write_len;

330 331 332
	/* CFTYPE_* flags */
	unsigned int flags;

A
Aristeu Rozanski 已提交
333 334 335
	/* file xattrs */
	struct simple_xattrs xattrs;

336 337 338 339
	int (*open)(struct inode *inode, struct file *file);
	ssize_t (*read)(struct cgroup *cgrp, struct cftype *cft,
			struct file *file,
			char __user *buf, size_t nbytes, loff_t *ppos);
340
	/*
341
	 * read_u64() is a shortcut for the common case of returning a
342 343
	 * single integer. Use it in place of read()
	 */
344
	u64 (*read_u64)(struct cgroup *cgrp, struct cftype *cft);
345 346 347
	/*
	 * read_s64() is a signed version of read_u64()
	 */
348
	s64 (*read_s64)(struct cgroup *cgrp, struct cftype *cft);
349 350 351 352 353 354
	/*
	 * read_map() is used for defining a map of key/value
	 * pairs. It should call cb->fill(cb, key, value) for each
	 * entry. The key/value pairs (and their ordering) should not
	 * change between reboots.
	 */
355 356
	int (*read_map)(struct cgroup *cont, struct cftype *cft,
			struct cgroup_map_cb *cb);
357 358 359 360
	/*
	 * read_seq_string() is used for outputting a simple sequence
	 * using seqfile.
	 */
361 362
	int (*read_seq_string)(struct cgroup *cont, struct cftype *cft,
			       struct seq_file *m);
363

364 365 366
	ssize_t (*write)(struct cgroup *cgrp, struct cftype *cft,
			 struct file *file,
			 const char __user *buf, size_t nbytes, loff_t *ppos);
367 368

	/*
369
	 * write_u64() is a shortcut for the common case of accepting
370 371 372
	 * a single integer (as parsed by simple_strtoull) from
	 * userspace. Use in place of write(); return 0 or error.
	 */
373
	int (*write_u64)(struct cgroup *cgrp, struct cftype *cft, u64 val);
374 375 376
	/*
	 * write_s64() is a signed version of write_u64()
	 */
377
	int (*write_s64)(struct cgroup *cgrp, struct cftype *cft, s64 val);
378

379 380 381 382 383 384 385
	/*
	 * write_string() is passed a nul-terminated kernelspace
	 * buffer of maximum length determined by max_write_len.
	 * Returns 0 or -ve error code.
	 */
	int (*write_string)(struct cgroup *cgrp, struct cftype *cft,
			    const char *buffer);
386 387 388 389 390 391 392 393
	/*
	 * 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.
	 */
	int (*trigger)(struct cgroup *cgrp, unsigned int event);

394
	int (*release)(struct inode *inode, struct file *file);
395 396 397 398 399 400 401 402 403 404 405 406 407 408 409

	/*
	 * register_event() callback will be used to add new userspace
	 * waiter for changes related to the cftype. Implement it if
	 * you want to provide this functionality. Use eventfd_signal()
	 * on eventfd to send notification to userspace.
	 */
	int (*register_event)(struct cgroup *cgrp, struct cftype *cft,
			struct eventfd_ctx *eventfd, const char *args);
	/*
	 * unregister_event() callback will be called when userspace
	 * closes the eventfd or on cgroup removing.
	 * This callback must be implemented, if you want provide
	 * notification functionality.
	 */
410
	void (*unregister_event)(struct cgroup *cgrp, struct cftype *cft,
411
			struct eventfd_ctx *eventfd);
412 413
};

414 415 416 417 418 419 420
/*
 * cftype_sets describe cftypes belonging to a subsystem and are chained at
 * cgroup_subsys->cftsets.  Each cftset points to an array of cftypes
 * terminated by zero length name.
 */
struct cftype_set {
	struct list_head		node;	/* chained at subsys->cftsets */
A
Aristeu Rozanski 已提交
421
	struct cftype			*cfts;
422 423
};

424 425 426 427 428 429
struct cgroup_scanner {
	struct cgroup *cg;
	int (*test_task)(struct task_struct *p, struct cgroup_scanner *scan);
	void (*process_task)(struct task_struct *p,
			struct cgroup_scanner *scan);
	struct ptr_heap *heap;
430
	void *data;
431 432
};

433 434 435 436 437 438
/* Caller should hold rcu_read_lock() */
static inline const char *cgroup_name(const struct cgroup *cgrp)
{
	return rcu_dereference(cgrp->name)->name;
}

A
Aristeu Rozanski 已提交
439 440
int cgroup_add_cftypes(struct cgroup_subsys *ss, struct cftype *cfts);
int cgroup_rm_cftypes(struct cgroup_subsys *ss, struct cftype *cfts);
441

L
Li Zefan 已提交
442
int cgroup_is_removed(const struct cgroup *cgrp);
443

L
Li Zefan 已提交
444
int cgroup_path(const struct cgroup *cgrp, char *buf, int buflen);
445

L
Li Zefan 已提交
446
int cgroup_task_count(const struct cgroup *cgrp);
447

448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469
/*
 * 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);
struct cgroup *cgroup_taskset_cur_cgroup(struct cgroup_taskset *tset);
int cgroup_taskset_size(struct cgroup_taskset *tset);

/**
 * cgroup_taskset_for_each - iterate cgroup_taskset
 * @task: the loop cursor
 * @skip_cgrp: skip if task's cgroup matches this, %NULL to iterate through all
 * @tset: taskset to iterate
 */
#define cgroup_taskset_for_each(task, skip_cgrp, tset)			\
	for ((task) = cgroup_taskset_first((tset)); (task);		\
	     (task) = cgroup_taskset_next((tset)))			\
		if (!(skip_cgrp) ||					\
		    cgroup_taskset_cur_cgroup((tset)) != (skip_cgrp))

470 471 472 473
/*
 * Control Group subsystem type.
 * See Documentation/cgroups/cgroups.txt for details
 */
474 475

struct cgroup_subsys {
476 477 478 479 480
	struct cgroup_subsys_state *(*css_alloc)(struct cgroup *cgrp);
	int (*css_online)(struct cgroup *cgrp);
	void (*css_offline)(struct cgroup *cgrp);
	void (*css_free)(struct cgroup *cgrp);

481 482 483 484 485 486 487
	int (*can_attach)(struct cgroup *cgrp, struct cgroup_taskset *tset);
	void (*cancel_attach)(struct cgroup *cgrp, struct cgroup_taskset *tset);
	void (*attach)(struct cgroup *cgrp, struct cgroup_taskset *tset);
	void (*fork)(struct task_struct *task);
	void (*exit)(struct cgroup *cgrp, struct cgroup *old_cgrp,
		     struct task_struct *task);
	void (*bind)(struct cgroup *root);
488

489 490
	int subsys_id;
	int active;
491
	int disabled;
492
	int early_init;
K
KAMEZAWA Hiroyuki 已提交
493 494 495 496 497
	/*
	 * True if this subsys uses ID. ID is not available before cgroup_init()
	 * (not available in early_init time.)
	 */
	bool use_id;
498

499 500 501 502 503 504 505 506 507 508 509 510 511 512 513
	/*
	 * 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;

514 515 516
#define MAX_CGROUP_TYPE_NAMELEN 32
	const char *name;

517 518
	/*
	 * Link to parent, and list entry in parent's children.
L
Li Zefan 已提交
519
	 * Protected by cgroup_lock()
520 521
	 */
	struct cgroupfs_root *root;
522
	struct list_head sibling;
K
KAMEZAWA Hiroyuki 已提交
523 524
	/* used when use_id == true */
	struct idr idr;
525
	spinlock_t id_lock;
526

527 528 529 530 531 532 533
	/* list of cftype_sets */
	struct list_head cftsets;

	/* base cftypes, automatically [de]registered with subsys itself */
	struct cftype *base_cftypes;
	struct cftype_set base_cftset;

534 535
	/* should be defined only by modular subsystems */
	struct module *module;
536 537 538
};

#define SUBSYS(_x) extern struct cgroup_subsys _x ## _subsys;
539
#define IS_SUBSYS_ENABLED(option) IS_BUILTIN(option)
540
#include <linux/cgroup_subsys.h>
541
#undef IS_SUBSYS_ENABLED
542 543 544
#undef SUBSYS

static inline struct cgroup_subsys_state *cgroup_subsys_state(
L
Li Zefan 已提交
545
	struct cgroup *cgrp, int subsys_id)
546
{
L
Li Zefan 已提交
547
	return cgrp->subsys[subsys_id];
548 549
}

550 551 552 553 554 555 556 557 558 559 560 561
/*
 * function to get the cgroup_subsys_state which allows for extra
 * rcu_dereference_check() conditions, such as locks used during the
 * cgroup_subsys::attach() methods.
 */
#define task_subsys_state_check(task, subsys_id, __c)			\
	rcu_dereference_check(task->cgroups->subsys[subsys_id],		\
			      lockdep_is_held(&task->alloc_lock) ||	\
			      cgroup_lock_is_held() || (__c))

static inline struct cgroup_subsys_state *
task_subsys_state(struct task_struct *task, int subsys_id)
562
{
563
	return task_subsys_state_check(task, subsys_id, false);
564 565 566 567 568 569 570 571
}

static inline struct cgroup* task_cgroup(struct task_struct *task,
					       int subsys_id)
{
	return task_subsys_state(task, subsys_id)->cgroup;
}

572 573 574 575 576 577
/**
 * cgroup_for_each_child - iterate through children of a cgroup
 * @pos: the cgroup * to use as the loop cursor
 * @cgroup: cgroup whose children to walk
 *
 * Walk @cgroup's children.  Must be called under rcu_read_lock().  A child
578 579
 * cgroup which hasn't finished ->css_online() or already has finished
 * ->css_offline() may show up during traversal and it's each subsystem's
580 581
 * responsibility to verify that each @pos is alive.
 *
582 583 584
 * If a subsystem synchronizes against the parent in its ->css_online() and
 * before starting iterating, a cgroup which finished ->css_online() is
 * guaranteed to be visible in the future iterations.
585 586 587 588 589 590
 */
#define cgroup_for_each_child(pos, cgroup)				\
	list_for_each_entry_rcu(pos, &(cgroup)->children, sibling)

struct cgroup *cgroup_next_descendant_pre(struct cgroup *pos,
					  struct cgroup *cgroup);
591
struct cgroup *cgroup_rightmost_descendant(struct cgroup *pos);
592 593 594 595 596 597 598

/**
 * cgroup_for_each_descendant_pre - pre-order walk of a cgroup's descendants
 * @pos: the cgroup * to use as the loop cursor
 * @cgroup: cgroup whose descendants to walk
 *
 * Walk @cgroup's descendants.  Must be called under rcu_read_lock().  A
599 600
 * descendant cgroup which hasn't finished ->css_online() or already has
 * finished ->css_offline() may show up during traversal and it's each
601 602
 * subsystem's responsibility to verify that each @pos is alive.
 *
603 604 605
 * If a subsystem synchronizes against the parent in its ->css_online() and
 * before starting iterating, and synchronizes against @pos on each
 * iteration, any descendant cgroup which finished ->css_offline() is
606 607 608 609 610
 * 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.
 *
611
 * my_online(@cgrp)
612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643
 * {
 *	Lock @cgrp->parent and @cgrp;
 *	Inherit state from @cgrp->parent;
 *	Unlock both.
 * }
 *
 * my_update_state(@cgrp)
 * {
 *	Lock @cgrp;
 *	Update @cgrp's state;
 *	Unlock @cgrp;
 *
 *	cgroup_for_each_descendant_pre(@pos, @cgrp) {
 *		Lock @pos;
 *		Verify @pos is alive and inherit state from @pos->parent;
 *		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
 * inheritance happens for any cgroup after the latest update to its
 * parent.
 *
 * 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
644
 * synchronize ->css_online() and ->css_offline() against tree-walking
645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666
 * operations.
 */
#define cgroup_for_each_descendant_pre(pos, cgroup)			\
	for (pos = cgroup_next_descendant_pre(NULL, (cgroup)); (pos);	\
	     pos = cgroup_next_descendant_pre((pos), (cgroup)))

struct cgroup *cgroup_next_descendant_post(struct cgroup *pos,
					   struct cgroup *cgroup);

/**
 * cgroup_for_each_descendant_post - post-order walk of a cgroup's descendants
 * @pos: the cgroup * to use as the loop cursor
 * @cgroup: cgroup whose descendants to walk
 *
 * Similar to cgroup_for_each_descendant_pre() but performs post-order
 * traversal instead.  Note that the walk visibility guarantee described in
 * pre-order walk doesn't apply the same to post-order walks.
 */
#define cgroup_for_each_descendant_post(pos, cgroup)			\
	for (pos = cgroup_next_descendant_post(NULL, (cgroup)); (pos);	\
	     pos = cgroup_next_descendant_post((pos), (cgroup)))

667 668 669 670 671 672
/* A cgroup_iter should be treated as an opaque object */
struct cgroup_iter {
	struct list_head *cg_link;
	struct list_head *task;
};

P
Paul Menage 已提交
673 674
/*
 * To iterate across the tasks in a cgroup:
675
 *
676
 * 1) call cgroup_iter_start to initialize an iterator
677 678 679 680 681
 *
 * 2) call cgroup_iter_next() to retrieve member tasks until it
 *    returns NULL or until you want to end the iteration
 *
 * 3) call cgroup_iter_end() to destroy the iterator.
682
 *
P
Paul Menage 已提交
683 684 685 686
 * Or, call cgroup_scan_tasks() to iterate through every task in a
 * cgroup - cgroup_scan_tasks() holds the css_set_lock when calling
 * the test_task() callback, but not while calling the process_task()
 * callback.
687
 */
L
Li Zefan 已提交
688 689
void cgroup_iter_start(struct cgroup *cgrp, struct cgroup_iter *it);
struct task_struct *cgroup_iter_next(struct cgroup *cgrp,
690
					struct cgroup_iter *it);
L
Li Zefan 已提交
691
void cgroup_iter_end(struct cgroup *cgrp, struct cgroup_iter *it);
692
int cgroup_scan_tasks(struct cgroup_scanner *scan);
M
Michael S. Tsirkin 已提交
693
int cgroup_attach_task_all(struct task_struct *from, struct task_struct *);
694
int cgroup_transfer_tasks(struct cgroup *to, struct cgroup *from);
M
Michael S. Tsirkin 已提交
695

K
KAMEZAWA Hiroyuki 已提交
696 697 698 699 700 701 702 703
/*
 * CSS ID is ID for cgroup_subsys_state structs under subsys. This only works
 * if cgroup_subsys.use_id == true. It can be used for looking up and scanning.
 * CSS ID is assigned at cgroup allocation (create) automatically
 * and removed when subsys calls free_css_id() function. This is because
 * the lifetime of cgroup_subsys_state is subsys's matter.
 *
 * Looking up and scanning function should be called under rcu_read_lock().
L
Li Zefan 已提交
704
 * Taking cgroup_mutex is not necessary for following calls.
K
KAMEZAWA Hiroyuki 已提交
705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727
 * But the css returned by this routine can be "not populated yet" or "being
 * destroyed". The caller should check css and cgroup's status.
 */

/*
 * Typically Called at ->destroy(), or somewhere the subsys frees
 * cgroup_subsys_state.
 */
void free_css_id(struct cgroup_subsys *ss, struct cgroup_subsys_state *css);

/* Find a cgroup_subsys_state which has given ID */

struct cgroup_subsys_state *css_lookup(struct cgroup_subsys *ss, int id);

/*
 * Get a cgroup whose id is greater than or equal to id under tree of root.
 * Returning a cgroup_subsys_state or NULL.
 */
struct cgroup_subsys_state *css_get_next(struct cgroup_subsys *ss, int id,
		struct cgroup_subsys_state *root, int *foundid);

/* Returns true if root is ancestor of cg */
bool css_is_ancestor(struct cgroup_subsys_state *cg,
728
		     const struct cgroup_subsys_state *root);
K
KAMEZAWA Hiroyuki 已提交
729 730 731 732

/* Get id and depth of css */
unsigned short css_id(struct cgroup_subsys_state *css);
unsigned short css_depth(struct cgroup_subsys_state *css);
S
Stephane Eranian 已提交
733
struct cgroup_subsys_state *cgroup_css_from_dir(struct file *f, int id);
K
KAMEZAWA Hiroyuki 已提交
734

735 736 737 738
#else /* !CONFIG_CGROUPS */

static inline int cgroup_init_early(void) { return 0; }
static inline int cgroup_init(void) { return 0; }
739
static inline void cgroup_fork(struct task_struct *p) {}
740
static inline void cgroup_post_fork(struct task_struct *p) {}
741
static inline void cgroup_exit(struct task_struct *p, int callbacks) {}
742 743 744

static inline void cgroup_lock(void) {}
static inline void cgroup_unlock(void) {}
B
Balbir Singh 已提交
745 746 747 748 749
static inline int cgroupstats_build(struct cgroupstats *stats,
					struct dentry *dentry)
{
	return -EINVAL;
}
750

751
/* No cgroups - nothing to do */
M
Michael S. Tsirkin 已提交
752 753 754 755 756
static inline int cgroup_attach_task_all(struct task_struct *from,
					 struct task_struct *t)
{
	return 0;
}
757

758 759 760
#endif /* !CONFIG_CGROUPS */

#endif /* _LINUX_CGROUP_H */