cgroup.h 25.8 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/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
#include <linux/fs.h>
23
#include <linux/percpu-refcount.h>
24 25 26 27 28 29

#ifdef CONFIG_CGROUPS

struct cgroupfs_root;
struct cgroup_subsys;
struct inode;
30
struct cgroup;
K
KAMEZAWA Hiroyuki 已提交
31
struct css_id;
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, int run_callbacks);
B
Balbir Singh 已提交
38 39
extern int cgroupstats_build(struct cgroupstats *stats,
				struct dentry *dentry);
40
extern int cgroup_load_subsys(struct cgroup_subsys *ss);
B
Ben Blum 已提交
41
extern void cgroup_unload_subsys(struct cgroup_subsys *ss);
42

43
extern int proc_cgroup_show(struct seq_file *, void *);
44

45 46 47 48 49
/*
 * Define the enumeration of all cgroup subsystems.
 *
 * We define ids for builtin subsystems and then modular ones.
 */
50 51
#define SUBSYS(_x) _x ## _subsys_id,
enum cgroup_subsys_id {
52
#define IS_SUBSYS_ENABLED(option) IS_BUILTIN(option)
53
#include <linux/cgroup_subsys.h>
54 55 56 57 58 59
#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)
60
#include <linux/cgroup_subsys.h>
61
#undef IS_SUBSYS_ENABLED
62
	CGROUP_SUBSYS_COUNT,
63 64 65
};
#undef SUBSYS

66 67
/* Per-subsystem/per-cgroup state maintained by the system. */
struct cgroup_subsys_state {
68
	/* the cgroup that this css is attached to */
69 70
	struct cgroup *cgroup;

71 72 73
	/* the cgroup subsystem that this css is attached to */
	struct cgroup_subsys *ss;

74 75
	/* reference count - access via css_[try]get() and css_put() */
	struct percpu_ref refcnt;
76

77 78 79
	/* the parent css */
	struct cgroup_subsys_state *parent;

80
	unsigned long flags;
K
KAMEZAWA Hiroyuki 已提交
81
	/* ID for this css, if possible */
A
Arnd Bergmann 已提交
82
	struct css_id __rcu *id;
83

84 85
	/* percpu_ref killing and RCU release */
	struct rcu_head rcu_head;
86
	struct work_struct destroy_work;
87 88 89 90
};

/* bits in struct cgroup_subsys_state flags field */
enum {
91
	CSS_ROOT	= (1 << 0), /* this CSS is the root of the subsystem */
92
	CSS_ONLINE	= (1 << 1), /* between ->css_online() and ->css_offline() */
93 94
};

95 96 97 98 99
/**
 * css_get - obtain a reference on the specified css
 * @css: target css
 *
 * The caller must already have a reference.
100 101 102 103
 */
static inline void css_get(struct cgroup_subsys_state *css)
{
	/* We don't need to reference count the root state */
104
	if (!(css->flags & CSS_ROOT))
105
		percpu_ref_get(&css->refcnt);
106
}
P
Paul Menage 已提交
107

108 109 110 111 112 113 114 115 116 117
/**
 * 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
118 119
static inline bool css_tryget(struct cgroup_subsys_state *css)
{
120
	if (css->flags & CSS_ROOT)
P
Paul Menage 已提交
121
		return true;
122
	return percpu_ref_tryget(&css->refcnt);
P
Paul Menage 已提交
123 124
}

125 126 127 128 129 130
/**
 * css_put - put a css reference
 * @css: target css
 *
 * Put a reference obtained via css_get() and css_tryget().
 */
131 132
static inline void css_put(struct cgroup_subsys_state *css)
{
133
	if (!(css->flags & CSS_ROOT))
134
		percpu_ref_put(&css->refcnt);
135 136
}

137 138 139
/* bits in struct cgroup flags field */
enum {
	/* Control Group is dead */
140
	CGRP_DEAD,
P
Paul Menage 已提交
141 142 143 144
	/*
	 * Control Group has previously had a child cgroup or a task,
	 * but no longer (only if CGRP_NOTIFY_ON_RELEASE is set)
	 */
145 146 147
	CGRP_RELEASABLE,
	/* Control Group requires release notifications to userspace */
	CGRP_NOTIFY_ON_RELEASE,
148
	/*
149 150 151
	 * 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.
152
	 */
153
	CGRP_CPUSET_CLONE_CHILDREN,
154 155
	/* see the comment above CGRP_ROOT_SANE_BEHAVIOR for details */
	CGRP_SANE_BEHAVIOR,
156 157
};

158 159 160
struct cgroup_name {
	struct rcu_head rcu_head;
	char name[];
161 162
};

163 164 165
struct cgroup {
	unsigned long flags;		/* "unsigned long" so bitops work */

166 167 168 169 170 171 172
	/*
	 * 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.
	 */
	int id;
T
Tejun Heo 已提交
173

174 175 176
	/* the number of attached css's */
	int nr_css;

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
	/*
	 * 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;

196 197 198 199 200 201 202 203 204 205 206 207 208
	/*
	 * 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;

209
	/* Private pointers for each registered subsystem */
210
	struct cgroup_subsys_state __rcu *subsys[CGROUP_SUBSYS_COUNT];
211 212

	struct cgroupfs_root *root;
213 214

	/*
215 216
	 * List of cgrp_cset_links pointing at css_sets with tasks in this
	 * cgroup.  Protected by css_set_lock.
217
	 */
218
	struct list_head cset_links;
219 220 221 222 223 224 225

	/*
	 * 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;
226

227 228 229 230 231 232
	/*
	 * 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;
233

T
Tejun Heo 已提交
234 235 236
	/* dummy css with NULL ->ss, points back to this cgroup */
	struct cgroup_subsys_state dummy_css;

237
	/* For css percpu_ref killing and RCU-protected deletion */
238
	struct rcu_head rcu_head;
239
	struct work_struct destroy_work;
240

A
Aristeu Rozanski 已提交
241 242
	/* directory xattrs */
	struct simple_xattrs xattrs;
243 244
};

245 246 247 248
#define MAX_CGROUP_ROOT_NAMELEN 64

/* cgroupfs_root->flags */
enum {
249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272
	/*
	 * 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.
	 *
	 * - Mount options "noprefix" and "clone_children" are disallowed.
	 *   Also, cgroupfs file cgroup.clone_children is not created.
	 *
	 * - When mounting an existing superblock, mount options should
	 *   match.
	 *
	 * - Remount is disallowed.
	 *
273 274
	 * - rename(2) is disallowed.
	 *
275 276
	 * - "tasks" is removed.  Everything should be at process
	 *   granularity.  Use "cgroup.procs" instead.
277
	 *
278 279 280
	 * - "cgroup.procs" is not sorted.  pids will be unique unless they
	 *   got recycled inbetween reads.
	 *
281 282
	 * - "release_agent" and "notify_on_release" are removed.
	 *   Replacement notification mechanism will be implemented.
283
	 *
284 285 286 287
	 * - 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.
	 *
288 289
	 * - cpuset: a task can be moved into an empty cpuset, and again it
	 *   takes masks of ancestors.
290
	 *
291 292
	 * - memcg: use_hierarchy is on by default and the cgroup file for
	 *   the flag is not created.
293
	 *
294
	 * - blkcg: blk-throttle becomes properly hierarchical.
295 296 297
	 */
	CGRP_ROOT_SANE_BEHAVIOR	= (1 << 0),

298 299
	CGRP_ROOT_NOPREFIX	= (1 << 1), /* mounted subsystems have no named prefix */
	CGRP_ROOT_XATTR		= (1 << 2), /* supports extended attributes */
300 301 302 303 304

	/* mount options live below bit 16 */
	CGRP_ROOT_OPTION_MASK	= (1 << 16) - 1,

	CGRP_ROOT_SUBSYS_BOUND	= (1 << 16), /* subsystems finished binding */
305 306 307 308 309 310 311 312 313 314
};

/*
 * A cgroupfs_root represents the root of a cgroup hierarchy, and may be
 * associated with a superblock to form an active hierarchy.  This is
 * internal to cgroup core.  Don't access directly from controllers.
 */
struct cgroupfs_root {
	struct super_block *sb;

315
	/* The bitmask of subsystems attached to this hierarchy */
316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336
	unsigned long subsys_mask;

	/* Unique id for this hierarchy. */
	int hierarchy_id;

	/* A list running through the attached subsystems */
	struct list_head subsys_list;

	/* The root cgroup for this hierarchy */
	struct cgroup top_cgroup;

	/* Tracks how many cgroups are currently defined in hierarchy.*/
	int number_of_cgroups;

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

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

	/* IDs for cgroups in this hierarchy */
337
	struct idr cgroup_idr;
338 339 340 341 342 343 344 345

	/* 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
346 347
/*
 * A css_set is a structure holding pointers to a set of
348 349
 * 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
350 351
 * list_add()/del() can bump the reference count on the entire cgroup
 * set for a task.
352 353 354 355 356
 */

struct css_set {

	/* Reference count */
357
	atomic_t refcount;
358

359 360 361 362 363 364
	/*
	 * List running through all cgroup groups in the same hash
	 * slot. Protected by css_set_lock
	 */
	struct hlist_node hlist;

365 366 367 368 369 370 371
	/*
	 * List running through all tasks using this cgroup
	 * group. Protected by css_set_lock
	 */
	struct list_head tasks;

	/*
372 373
	 * List of cgrp_cset_links pointing at cgroups referenced from this
	 * css_set.  Protected by css_set_lock.
374
	 */
375
	struct list_head cgrp_links;
376 377 378 379

	/*
	 * 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
380 381
	 * during subsystem registration (at boot time) and modular subsystem
	 * loading/unloading.
382 383
	 */
	struct cgroup_subsys_state *subsys[CGROUP_SUBSYS_COUNT];
384 385 386

	/* For RCU-protected deletion */
	struct rcu_head rcu_head;
387 388
};

P
Paul Menage 已提交
389 390
/*
 * struct cftype: handler definitions for cgroup control files
391 392
 *
 * When reading/writing to a file:
L
Li Zefan 已提交
393
 *	- the cgroup to use is file->f_dentry->d_parent->d_fsdata
394 395 396
 *	- the 'cftype' of the file is file->f_dentry->d_fsdata
 */

397
/* cftype->flags */
398
enum {
L
Li Zefan 已提交
399 400
	CFTYPE_ONLY_ON_ROOT	= (1 << 0),	/* only create on root cgrp */
	CFTYPE_NOT_ON_ROOT	= (1 << 1),	/* don't create on root cgrp */
401
	CFTYPE_INSANE		= (1 << 2),	/* don't create if sane_behavior */
T
Tejun Heo 已提交
402
	CFTYPE_NO_PREFIX	= (1 << 3),	/* (DON'T USE FOR NEW FILES) no subsys prefix */
403
};
404 405 406

#define MAX_CFTYPE_NAME		64

407
struct cftype {
P
Paul Menage 已提交
408 409
	/*
	 * By convention, the name should begin with the name of the
410 411
	 * subsystem, followed by a period.  Zero length string indicates
	 * end of cftype array.
P
Paul Menage 已提交
412
	 */
413 414
	char name[MAX_CFTYPE_NAME];
	int private;
L
Li Zefan 已提交
415 416 417 418
	/*
	 * If not 0, file mode is set to this value, otherwise it will
	 * be figured out automatically
	 */
A
Al Viro 已提交
419
	umode_t mode;
420 421 422 423 424 425 426

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

427 428 429
	/* CFTYPE_* flags */
	unsigned int flags;

430 431 432 433 434 435
	/*
	 * The subsys this file belongs to.  Initialized automatically
	 * during registration.  NULL for cgroup core files.
	 */
	struct cgroup_subsys *ss;

436
	int (*open)(struct inode *inode, struct file *file);
437
	/*
438
	 * read_u64() is a shortcut for the common case of returning a
439 440
	 * single integer. Use it in place of read()
	 */
441
	u64 (*read_u64)(struct cgroup_subsys_state *css, struct cftype *cft);
442 443 444
	/*
	 * read_s64() is a signed version of read_u64()
	 */
445
	s64 (*read_s64)(struct cgroup_subsys_state *css, struct cftype *cft);
446 447 448 449
	/*
	 * read_seq_string() is used for outputting a simple sequence
	 * using seqfile.
	 */
450 451
	int (*read_seq_string)(struct cgroup_subsys_state *css,
			       struct cftype *cft, struct seq_file *m);
452

453
	/*
454
	 * write_u64() is a shortcut for the common case of accepting
455 456 457
	 * a single integer (as parsed by simple_strtoull) from
	 * userspace. Use in place of write(); return 0 or error.
	 */
458 459
	int (*write_u64)(struct cgroup_subsys_state *css, struct cftype *cft,
			 u64 val);
460 461 462
	/*
	 * write_s64() is a signed version of write_u64()
	 */
463 464
	int (*write_s64)(struct cgroup_subsys_state *css, struct cftype *cft,
			 s64 val);
465

466 467 468 469 470
	/*
	 * write_string() is passed a nul-terminated kernelspace
	 * buffer of maximum length determined by max_write_len.
	 * Returns 0 or -ve error code.
	 */
471
	int (*write_string)(struct cgroup_subsys_state *css, struct cftype *cft,
472
			    const char *buffer);
473 474 475 476 477 478
	/*
	 * 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.
	 */
479
	int (*trigger)(struct cgroup_subsys_state *css, unsigned int event);
480 481
};

482 483 484 485 486 487 488
/*
 * 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
489
	struct cftype			*cfts;
490 491
};

492 493 494 495 496 497 498 499 500
/*
 * 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;
}

501 502 503 504 505 506
/* 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
507
int cgroup_add_cftypes(struct cgroup_subsys *ss, struct cftype *cfts);
508
int cgroup_rm_cftypes(struct cftype *cfts);
509

510
bool cgroup_is_descendant(struct cgroup *cgrp, struct cgroup *ancestor);
511

L
Li Zefan 已提交
512
int cgroup_path(const struct cgroup *cgrp, char *buf, int buflen);
513
int task_cgroup_path(struct task_struct *task, char *buf, size_t buflen);
514

L
Li Zefan 已提交
515
int cgroup_task_count(const struct cgroup *cgrp);
516

517 518 519 520 521 522 523
/*
 * 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);
524 525
struct cgroup_subsys_state *cgroup_taskset_cur_css(struct cgroup_taskset *tset,
						   int subsys_id);
526 527 528 529 530
int cgroup_taskset_size(struct cgroup_taskset *tset);

/**
 * cgroup_taskset_for_each - iterate cgroup_taskset
 * @task: the loop cursor
531
 * @skip_css: skip if task's css matches this, %NULL to iterate through all
532 533
 * @tset: taskset to iterate
 */
534
#define cgroup_taskset_for_each(task, skip_css, tset)			\
535 536
	for ((task) = cgroup_taskset_first((tset)); (task);		\
	     (task) = cgroup_taskset_next((tset)))			\
537 538 539
		if (!(skip_css) ||					\
		    cgroup_taskset_cur_css((tset),			\
			(skip_css)->ss->subsys_id) != (skip_css))
540

541 542 543 544
/*
 * Control Group subsystem type.
 * See Documentation/cgroups/cgroups.txt for details
 */
545 546

struct cgroup_subsys {
547 548 549 550 551 552 553 554 555 556 557
	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);
558
	void (*fork)(struct task_struct *task);
559 560
	void (*exit)(struct cgroup_subsys_state *css,
		     struct cgroup_subsys_state *old_css,
561
		     struct task_struct *task);
562
	void (*bind)(struct cgroup_subsys_state *root_css);
563

564
	int subsys_id;
565
	int disabled;
566
	int early_init;
567

568 569 570 571 572 573 574 575 576 577 578 579 580 581 582
	/*
	 * 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;

583 584 585
#define MAX_CGROUP_TYPE_NAMELEN 32
	const char *name;

586 587
	/*
	 * Link to parent, and list entry in parent's children.
L
Li Zefan 已提交
588
	 * Protected by cgroup_lock()
589 590
	 */
	struct cgroupfs_root *root;
591
	struct list_head sibling;
592

593 594 595 596 597 598 599
	/* 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;

600 601
	/* should be defined only by modular subsystems */
	struct module *module;
602 603 604
};

#define SUBSYS(_x) extern struct cgroup_subsys _x ## _subsys;
605
#define IS_SUBSYS_ENABLED(option) IS_BUILTIN(option)
606
#include <linux/cgroup_subsys.h>
607
#undef IS_SUBSYS_ENABLED
608 609
#undef SUBSYS

T
Tejun Heo 已提交
610 611 612 613 614 615 616 617 618 619
/**
 * 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)
{
620
	return css->parent;
T
Tejun Heo 已提交
621 622
}

623 624 625 626 627 628 629 630 631 632 633 634
/**
 * 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.
635
 */
T
Tejun Heo 已提交
636 637
#ifdef CONFIG_PROVE_RCU
extern struct mutex cgroup_mutex;
638 639 640 641
#define task_css_set_check(task, __c)					\
	rcu_dereference_check((task)->cgroups,				\
		lockdep_is_held(&(task)->alloc_lock) ||			\
		lockdep_is_held(&cgroup_mutex) || (__c))
T
Tejun Heo 已提交
642
#else
643 644
#define task_css_set_check(task, __c)					\
	rcu_dereference((task)->cgroups)
T
Tejun Heo 已提交
645
#endif
646

647
/**
648
 * task_css_check - obtain css for (task, subsys) w/ extra access conds
649 650 651 652 653 654 655
 * @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().
 */
656
#define task_css_check(task, subsys_id, __c)				\
657 658 659 660 661 662 663 664 665 666 667 668 669 670
	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);
}

/**
671
 * task_css - obtain css for (task, subsys)
672 673 674
 * @task: the target task
 * @subsys_id: the target subsystem ID
 *
675
 * See task_css_check().
676
 */
677 678
static inline struct cgroup_subsys_state *task_css(struct task_struct *task,
						   int subsys_id)
679
{
680
	return task_css_check(task, subsys_id, false);
681 682
}

683 684
static inline struct cgroup *task_cgroup(struct task_struct *task,
					 int subsys_id)
685
{
686
	return task_css(task, subsys_id)->cgroup;
687 688
}

689 690
struct cgroup_subsys_state *css_next_child(struct cgroup_subsys_state *pos,
					   struct cgroup_subsys_state *parent);
691

692 693
struct cgroup_subsys_state *css_from_id(int id, struct cgroup_subsys *ss);

694
/**
695 696 697
 * 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
698
 *
699 700
 * Walk @parent's children.  Must be called under rcu_read_lock().  A child
 * css which hasn't finished ->css_online() or already has finished
701
 * ->css_offline() may show up during traversal and it's each subsystem's
702 703
 * responsibility to verify that each @pos is alive.
 *
704
 * If a subsystem synchronizes against the parent in its ->css_online() and
705
 * before starting iterating, a css which finished ->css_online() is
706
 * guaranteed to be visible in the future iterations.
707 708 709 710
 *
 * 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.
711
 */
712 713 714
#define css_for_each_child(pos, parent)					\
	for ((pos) = css_next_child(NULL, (parent)); (pos);		\
	     (pos) = css_next_child((pos), (parent)))
715

716 717 718 719 720 721
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);
722 723

/**
724 725 726
 * 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
727
 *
728 729
 * 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
730
 * descendant css which hasn't finished ->css_online() or already has
731
 * finished ->css_offline() may show up during traversal and it's each
732 733
 * subsystem's responsibility to verify that each @pos is alive.
 *
734 735
 * If a subsystem synchronizes against the parent in its ->css_online() and
 * before starting iterating, and synchronizes against @pos on each
736
 * iteration, any descendant css which finished ->css_online() is
737 738 739 740 741
 * 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.
 *
742
 * my_online(@css)
743
 * {
744 745
 *	Lock @css's parent and @css;
 *	Inherit state from the parent;
746 747 748
 *	Unlock both.
 * }
 *
749
 * my_update_state(@css)
750
 * {
751
 *	css_for_each_descendant_pre(@pos, @css) {
752
 *		Lock @pos;
753 754 755 756
 *		if (@pos == @css)
 *			Update @css's state;
 *		else
 *			Verify @pos is alive and inherit state from its parent;
757 758 759 760 761 762 763 764 765 766
 *		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
767
 * inheritance happens for any css after the latest update to its parent.
768 769 770 771 772
 *
 * 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
773
 * synchronize ->css_online() and ->css_offline() against tree-walking
774
 * operations.
775 776 777 778
 *
 * 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.
779
 */
780 781 782
#define css_for_each_descendant_pre(pos, css)				\
	for ((pos) = css_next_descendant_pre(NULL, (css)); (pos);	\
	     (pos) = css_next_descendant_pre((pos), (css)))
783

784 785 786
struct cgroup_subsys_state *
css_next_descendant_post(struct cgroup_subsys_state *pos,
			 struct cgroup_subsys_state *css);
787 788

/**
789 790 791
 * 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
792
 *
793
 * Similar to css_for_each_descendant_pre() but performs post-order
794 795 796
 * 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.
797
 */
798 799 800
#define css_for_each_descendant_post(pos, css)				\
	for ((pos) = css_next_descendant_post(NULL, (css)); (pos);	\
	     (pos) = css_next_descendant_post((pos), (css)))
801

802 803 804
/* A css_task_iter should be treated as an opaque object */
struct css_task_iter {
	struct cgroup_subsys_state	*origin_css;
805 806
	struct list_head		*cset_link;
	struct list_head		*task;
807 808
};

809 810 811 812
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
813

814 815 816 817
int css_scan_tasks(struct cgroup_subsys_state *css,
		   bool (*test)(struct task_struct *, void *),
		   void (*process)(struct task_struct *, void *),
		   void *data, struct ptr_heap *heap);
T
Tejun Heo 已提交
818

M
Michael S. Tsirkin 已提交
819
int cgroup_attach_task_all(struct task_struct *from, struct task_struct *);
820
int cgroup_transfer_tasks(struct cgroup *to, struct cgroup *from);
M
Michael S. Tsirkin 已提交
821

822 823
struct cgroup_subsys_state *css_from_dir(struct dentry *dentry,
					 struct cgroup_subsys *ss);
K
KAMEZAWA Hiroyuki 已提交
824

825 826 827 828
#else /* !CONFIG_CGROUPS */

static inline int cgroup_init_early(void) { return 0; }
static inline int cgroup_init(void) { return 0; }
829
static inline void cgroup_fork(struct task_struct *p) {}
830
static inline void cgroup_post_fork(struct task_struct *p) {}
831
static inline void cgroup_exit(struct task_struct *p, int callbacks) {}
832

B
Balbir Singh 已提交
833 834 835 836 837
static inline int cgroupstats_build(struct cgroupstats *stats,
					struct dentry *dentry)
{
	return -EINVAL;
}
838

839
/* No cgroups - nothing to do */
M
Michael S. Tsirkin 已提交
840 841 842 843 844
static inline int cgroup_attach_task_all(struct task_struct *from,
					 struct task_struct *t)
{
	return 0;
}
845

846 847 848
#endif /* !CONFIG_CGROUPS */

#endif /* _LINUX_CGROUP_H */