memcontrol.h 30.4 KB
Newer Older
B
Balbir Singh 已提交
1 2 3 4 5
/* memcontrol.h - Memory Controller
 *
 * Copyright IBM Corporation, 2007
 * Author Balbir Singh <balbir@linux.vnet.ibm.com>
 *
6 7 8
 * Copyright 2007 OpenVZ SWsoft Inc
 * Author: Pavel Emelianov <xemul@openvz.org>
 *
B
Balbir Singh 已提交
9 10 11 12 13 14 15 16 17 18 19 20 21
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 */

#ifndef _LINUX_MEMCONTROL_H
#define _LINUX_MEMCONTROL_H
22
#include <linux/cgroup.h>
23
#include <linux/vm_event_item.h>
24
#include <linux/hardirq.h>
25
#include <linux/jump_label.h>
M
Michal Hocko 已提交
26 27 28
#include <linux/page_counter.h>
#include <linux/vmpressure.h>
#include <linux/eventfd.h>
29 30
#include <linux/mm.h>
#include <linux/vmstat.h>
M
Michal Hocko 已提交
31
#include <linux/writeback.h>
32
#include <linux/page-flags.h>
33

34
struct mem_cgroup;
35 36
struct page;
struct mm_struct;
37
struct kmem_cache;
38

39 40 41 42 43 44 45 46 47
/* Cgroup-specific page state, on top of universal node page state */
enum memcg_stat_item {
	MEMCG_CACHE = NR_VM_NODE_STAT_ITEMS,
	MEMCG_RSS,
	MEMCG_RSS_HUGE,
	MEMCG_SWAP,
	MEMCG_SOCK,
	/* XXX: why are these zone and not node counters? */
	MEMCG_KERNEL_STACK_KB,
48
	MEMCG_NR_STAT,
49 50
};

51 52
enum memcg_memory_event {
	MEMCG_LOW,
53 54 55
	MEMCG_HIGH,
	MEMCG_MAX,
	MEMCG_OOM,
R
Roman Gushchin 已提交
56
	MEMCG_OOM_KILL,
57 58
	MEMCG_SWAP_MAX,
	MEMCG_SWAP_FAIL,
59
	MEMCG_NR_MEMORY_EVENTS,
60 61
};

R
Roman Gushchin 已提交
62 63 64 65 66 67
enum mem_cgroup_protection {
	MEMCG_PROT_NONE,
	MEMCG_PROT_LOW,
	MEMCG_PROT_MIN,
};

68
struct mem_cgroup_reclaim_cookie {
69
	pg_data_t *pgdat;
70 71 72 73
	int priority;
	unsigned int generation;
};

74 75 76 77 78 79 80 81 82 83
#ifdef CONFIG_MEMCG

#define MEM_CGROUP_ID_SHIFT	16
#define MEM_CGROUP_ID_MAX	USHRT_MAX

struct mem_cgroup_id {
	int id;
	atomic_t ref;
};

M
Michal Hocko 已提交
84 85 86 87 88 89 90 91 92 93 94 95 96 97
/*
 * Per memcg event counter is incremented at every pagein/pageout. With THP,
 * it will be incremated by the number of pages. This counter is used for
 * for trigger some periodic events. This is straightforward and better
 * than using jiffies etc. to handle periodic memcg event.
 */
enum mem_cgroup_events_target {
	MEM_CGROUP_TARGET_THRESH,
	MEM_CGROUP_TARGET_SOFTLIMIT,
	MEM_CGROUP_TARGET_NUMAINFO,
	MEM_CGROUP_NTARGETS,
};

struct mem_cgroup_stat_cpu {
98
	long count[MEMCG_NR_STAT];
99
	unsigned long events[NR_VM_EVENT_ITEMS];
M
Michal Hocko 已提交
100 101 102 103 104 105 106 107 108 109
	unsigned long nr_page_events;
	unsigned long targets[MEM_CGROUP_NTARGETS];
};

struct mem_cgroup_reclaim_iter {
	struct mem_cgroup *position;
	/* scan generation, increased every round-trip */
	unsigned int generation;
};

110 111 112 113
struct lruvec_stat {
	long count[NR_VM_NODE_STAT_ITEMS];
};

M
Michal Hocko 已提交
114 115 116
/*
 * per-zone information in memory controller.
 */
117
struct mem_cgroup_per_node {
M
Michal Hocko 已提交
118
	struct lruvec		lruvec;
119 120 121 122

	struct lruvec_stat __percpu *lruvec_stat_cpu;
	atomic_long_t		lruvec_stat[NR_VM_NODE_STAT_ITEMS];

123
	unsigned long		lru_zone_size[MAX_NR_ZONES][NR_LRU_LISTS];
M
Michal Hocko 已提交
124 125 126 127 128 129 130

	struct mem_cgroup_reclaim_iter	iter[DEF_PRIORITY + 1];

	struct rb_node		tree_node;	/* RB tree node */
	unsigned long		usage_in_excess;/* Set to the value by which */
						/* the soft limit is exceeded*/
	bool			on_tree;
131 132 133
	bool			congested;	/* memcg has many dirty pages */
						/* backed by a congested BDI */

M
Michal Hocko 已提交
134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163
	struct mem_cgroup	*memcg;		/* Back pointer, we cannot */
						/* use container_of	   */
};

struct mem_cgroup_threshold {
	struct eventfd_ctx *eventfd;
	unsigned long threshold;
};

/* For threshold */
struct mem_cgroup_threshold_ary {
	/* An array index points to threshold just below or equal to usage. */
	int current_threshold;
	/* Size of entries[] */
	unsigned int size;
	/* Array of thresholds */
	struct mem_cgroup_threshold entries[0];
};

struct mem_cgroup_thresholds {
	/* Primary thresholds array */
	struct mem_cgroup_threshold_ary *primary;
	/*
	 * Spare threshold array.
	 * This is needed to make mem_cgroup_unregister_event() "never fail".
	 * It must be able to store at least primary->size - 1 entries.
	 */
	struct mem_cgroup_threshold_ary *spare;
};

164 165 166 167 168 169
enum memcg_kmem_state {
	KMEM_NONE,
	KMEM_ALLOCATED,
	KMEM_ONLINE,
};

170 171 172 173 174 175 176 177 178
#if defined(CONFIG_SMP)
struct memcg_padding {
	char x[0];
} ____cacheline_internodealigned_in_smp;
#define MEMCG_PADDING(name)      struct memcg_padding name;
#else
#define MEMCG_PADDING(name)
#endif

M
Michal Hocko 已提交
179 180 181 182 183 184 185 186 187
/*
 * The memory controller data structure. The memory controller controls both
 * page cache and RSS per cgroup. We would eventually like to provide
 * statistics based on the statistics developed by Rik Van Riel for clock-pro,
 * to help the administrator determine what knobs to tune.
 */
struct mem_cgroup {
	struct cgroup_subsys_state css;

188 189 190
	/* Private memcg ID. Used to ID objects that outlive the cgroup */
	struct mem_cgroup_id id;

M
Michal Hocko 已提交
191 192
	/* Accounted resources */
	struct page_counter memory;
193
	struct page_counter swap;
194 195

	/* Legacy consumer-oriented counters */
M
Michal Hocko 已提交
196 197
	struct page_counter memsw;
	struct page_counter kmem;
198
	struct page_counter tcpmem;
M
Michal Hocko 已提交
199

200
	/* Upper bound of normal memory consumption range */
M
Michal Hocko 已提交
201 202
	unsigned long high;

203 204 205
	/* Range enforcement for interrupt charges */
	struct work_struct high_work;

M
Michal Hocko 已提交
206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223
	unsigned long soft_limit;

	/* vmpressure notifications */
	struct vmpressure vmpressure;

	/*
	 * Should the accounting and control be hierarchical, per subtree?
	 */
	bool use_hierarchy;

	/* protected by memcg_oom_lock */
	bool		oom_lock;
	int		under_oom;

	int	swappiness;
	/* OOM-Killer disable */
	int		oom_kill_disable;

224
	/* memory.events */
225 226
	struct cgroup_file events_file;

227 228 229
	/* handle for "memory.swap.events" */
	struct cgroup_file swap_events_file;

M
Michal Hocko 已提交
230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246
	/* protect arrays of thresholds */
	struct mutex thresholds_lock;

	/* thresholds for memory usage. RCU-protected */
	struct mem_cgroup_thresholds thresholds;

	/* thresholds for mem+swap usage. RCU-protected */
	struct mem_cgroup_thresholds memsw_thresholds;

	/* For oom notifier event fd */
	struct list_head oom_notify;

	/*
	 * Should we move charges of a task when a task is moved into this
	 * mem_cgroup ? And what type of charges should we move ?
	 */
	unsigned long move_charge_at_immigrate;
247 248 249 250 251 252
	/* taken only while moving_account > 0 */
	spinlock_t		move_lock;
	unsigned long		move_lock_flags;

	MEMCG_PADDING(_pad1_);

M
Michal Hocko 已提交
253 254 255 256 257
	/*
	 * set > 0 if pages under this cgroup are moving to other cgroup.
	 */
	atomic_t		moving_account;
	struct task_struct	*move_lock_task;
258

259
	/* memory.stat */
260
	struct mem_cgroup_stat_cpu __percpu *stat_cpu;
261 262 263

	MEMCG_PADDING(_pad2_);

264
	atomic_long_t		stat[MEMCG_NR_STAT];
265
	atomic_long_t		events[NR_VM_EVENT_ITEMS];
266
	atomic_long_t memory_events[MEMCG_NR_MEMORY_EVENTS];
M
Michal Hocko 已提交
267

268 269 270
	unsigned long		socket_pressure;

	/* Legacy tcp memory accounting */
271 272
	bool			tcpmem_active;
	int			tcpmem_pressure;
273

274
#ifndef CONFIG_SLOB
M
Michal Hocko 已提交
275 276
        /* Index in the kmem_cache->memcg_params.memcg_caches array */
	int kmemcg_id;
277
	enum memcg_kmem_state kmem_state;
278
	struct list_head kmem_caches;
M
Michal Hocko 已提交
279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299
#endif

	int last_scanned_node;
#if MAX_NUMNODES > 1
	nodemask_t	scan_nodes;
	atomic_t	numainfo_events;
	atomic_t	numainfo_updating;
#endif

#ifdef CONFIG_CGROUP_WRITEBACK
	struct list_head cgwb_list;
	struct wb_domain cgwb_domain;
#endif

	/* List of events which userspace want to receive */
	struct list_head event_list;
	spinlock_t event_list_lock;

	struct mem_cgroup_per_node *nodeinfo[0];
	/* WARNING: nodeinfo must be the last member here */
};
300

301 302 303 304 305 306
/*
 * size of first charge trial. "32" comes from vmscan.c's magic value.
 * TODO: maybe necessary to use big numbers in big irons.
 */
#define MEMCG_CHARGE_BATCH 32U

307
extern struct mem_cgroup *root_mem_cgroup;
T
Tejun Heo 已提交
308

309 310 311 312 313
static inline bool mem_cgroup_disabled(void)
{
	return !cgroup_subsys_enabled(memory_cgrp_subsys);
}

R
Roman Gushchin 已提交
314 315
enum mem_cgroup_protection mem_cgroup_protected(struct mem_cgroup *root,
						struct mem_cgroup *memcg);
316

317
int mem_cgroup_try_charge(struct page *page, struct mm_struct *mm,
318 319
			  gfp_t gfp_mask, struct mem_cgroup **memcgp,
			  bool compound);
320 321 322
int mem_cgroup_try_charge_delay(struct page *page, struct mm_struct *mm,
			  gfp_t gfp_mask, struct mem_cgroup **memcgp,
			  bool compound);
323
void mem_cgroup_commit_charge(struct page *page, struct mem_cgroup *memcg,
324 325 326
			      bool lrucare, bool compound);
void mem_cgroup_cancel_charge(struct page *page, struct mem_cgroup *memcg,
		bool compound);
327
void mem_cgroup_uncharge(struct page *page);
328
void mem_cgroup_uncharge_list(struct list_head *page_list);
329

330
void mem_cgroup_migrate(struct page *oldpage, struct page *newpage);
331

332 333
static struct mem_cgroup_per_node *
mem_cgroup_nodeinfo(struct mem_cgroup *memcg, int nid)
334
{
335
	return memcg->nodeinfo[nid];
336 337 338
}

/**
339 340
 * mem_cgroup_lruvec - get the lru list vector for a node or a memcg zone
 * @node: node of the wanted lruvec
341 342
 * @memcg: memcg of the wanted lruvec
 *
343 344
 * Returns the lru list vector holding pages for a given @node or a given
 * @memcg and @zone. This can be the node lruvec, if the memory controller
345 346
 * is disabled.
 */
347
static inline struct lruvec *mem_cgroup_lruvec(struct pglist_data *pgdat,
348
				struct mem_cgroup *memcg)
349
{
350
	struct mem_cgroup_per_node *mz;
351 352 353
	struct lruvec *lruvec;

	if (mem_cgroup_disabled()) {
354
		lruvec = node_lruvec(pgdat);
355 356 357
		goto out;
	}

358
	mz = mem_cgroup_nodeinfo(memcg, pgdat->node_id);
359 360 361 362
	lruvec = &mz->lruvec;
out:
	/*
	 * Since a node can be onlined after the mem_cgroup was created,
M
Mel Gorman 已提交
363
	 * we have to be prepared to initialize lruvec->pgdat here;
364 365
	 * and if offlined then reonlined, we need to reinitialize it.
	 */
366 367
	if (unlikely(lruvec->pgdat != pgdat))
		lruvec->pgdat = pgdat;
368 369 370
	return lruvec;
}

M
Mel Gorman 已提交
371
struct lruvec *mem_cgroup_page_lruvec(struct page *, struct pglist_data *);
372

373
bool task_in_mem_cgroup(struct task_struct *task, struct mem_cgroup *memcg);
374
struct mem_cgroup *mem_cgroup_from_task(struct task_struct *p);
375

M
Michal Hocko 已提交
376 377 378 379 380
static inline
struct mem_cgroup *mem_cgroup_from_css(struct cgroup_subsys_state *css){
	return css ? container_of(css, struct mem_cgroup, css) : NULL;
}

381 382 383 384 385
static inline void mem_cgroup_put(struct mem_cgroup *memcg)
{
	css_put(&memcg->css);
}

386 387 388
#define mem_cgroup_from_counter(counter, member)	\
	container_of(counter, struct mem_cgroup, member)

M
Michal Hocko 已提交
389 390 391 392
struct mem_cgroup *mem_cgroup_iter(struct mem_cgroup *,
				   struct mem_cgroup *,
				   struct mem_cgroup_reclaim_cookie *);
void mem_cgroup_iter_break(struct mem_cgroup *, struct mem_cgroup *);
393 394
int mem_cgroup_scan_tasks(struct mem_cgroup *,
			  int (*)(struct task_struct *, void *), void *);
M
Michal Hocko 已提交
395

396 397 398 399 400
static inline unsigned short mem_cgroup_id(struct mem_cgroup *memcg)
{
	if (mem_cgroup_disabled())
		return 0;

401
	return memcg->id.id;
402
}
403
struct mem_cgroup *mem_cgroup_from_id(unsigned short id);
404

405 406 407 408 409 410 411 412 413 414 415
static inline struct mem_cgroup *lruvec_memcg(struct lruvec *lruvec)
{
	struct mem_cgroup_per_node *mz;

	if (mem_cgroup_disabled())
		return NULL;

	mz = container_of(lruvec, struct mem_cgroup_per_node, lruvec);
	return mz->memcg;
}

416 417 418 419 420 421 422 423 424 425 426 427 428 429
/**
 * parent_mem_cgroup - find the accounting parent of a memcg
 * @memcg: memcg whose parent to find
 *
 * Returns the parent memcg, or NULL if this is the root or the memory
 * controller is in legacy no-hierarchy mode.
 */
static inline struct mem_cgroup *parent_mem_cgroup(struct mem_cgroup *memcg)
{
	if (!memcg->memory.parent)
		return NULL;
	return mem_cgroup_from_counter(memcg->memory.parent, memory);
}

M
Michal Hocko 已提交
430 431 432 433 434 435 436 437 438
static inline bool mem_cgroup_is_descendant(struct mem_cgroup *memcg,
			      struct mem_cgroup *root)
{
	if (root == memcg)
		return true;
	if (!root->use_hierarchy)
		return false;
	return cgroup_is_descendant(memcg->css.cgroup, root->css.cgroup);
}
G
Glauber Costa 已提交
439

440 441
static inline bool mm_match_cgroup(struct mm_struct *mm,
				   struct mem_cgroup *memcg)
442
{
443
	struct mem_cgroup *task_memcg;
444
	bool match = false;
445

446
	rcu_read_lock();
447
	task_memcg = mem_cgroup_from_task(rcu_dereference(mm->owner));
448
	if (task_memcg)
449
		match = mem_cgroup_is_descendant(task_memcg, memcg);
450
	rcu_read_unlock();
451
	return match;
452
}
453

454
struct cgroup_subsys_state *mem_cgroup_css_from_page(struct page *page);
455
ino_t page_cgroup_ino(struct page *page);
456

457 458 459 460 461 462 463
static inline bool mem_cgroup_online(struct mem_cgroup *memcg)
{
	if (mem_cgroup_disabled())
		return true;
	return !!(memcg->css.flags & CSS_ONLINE);
}

464 465 466
/*
 * For memory reclaim.
 */
467
int mem_cgroup_select_victim_node(struct mem_cgroup *memcg);
M
Michal Hocko 已提交
468 469

void mem_cgroup_update_lru_size(struct lruvec *lruvec, enum lru_list lru,
470
		int zid, int nr_pages);
M
Michal Hocko 已提交
471

472 473 474
unsigned long mem_cgroup_node_nr_lru_pages(struct mem_cgroup *memcg,
					   int nid, unsigned int lru_mask);

M
Michal Hocko 已提交
475 476 477
static inline
unsigned long mem_cgroup_get_lru_size(struct lruvec *lruvec, enum lru_list lru)
{
478
	struct mem_cgroup_per_node *mz;
479 480
	unsigned long nr_pages = 0;
	int zid;
M
Michal Hocko 已提交
481

482
	mz = container_of(lruvec, struct mem_cgroup_per_node, lruvec);
483 484 485 486 487 488 489 490 491 492 493 494 495
	for (zid = 0; zid < MAX_NR_ZONES; zid++)
		nr_pages += mz->lru_zone_size[zid][lru];
	return nr_pages;
}

static inline
unsigned long mem_cgroup_get_zone_lru_size(struct lruvec *lruvec,
		enum lru_list lru, int zone_idx)
{
	struct mem_cgroup_per_node *mz;

	mz = container_of(lruvec, struct mem_cgroup_per_node, lruvec);
	return mz->lru_zone_size[zone_idx][lru];
M
Michal Hocko 已提交
496 497
}

498 499
void mem_cgroup_handle_over_high(void);

500
unsigned long mem_cgroup_get_max(struct mem_cgroup *memcg);
501

502 503
void mem_cgroup_print_oom_info(struct mem_cgroup *memcg,
				struct task_struct *p);
504

505
static inline void mem_cgroup_oom_enable(void)
506
{
T
Tejun Heo 已提交
507 508
	WARN_ON(current->memcg_may_oom);
	current->memcg_may_oom = 1;
509 510
}

511
static inline void mem_cgroup_oom_disable(void)
512
{
T
Tejun Heo 已提交
513 514
	WARN_ON(!current->memcg_may_oom);
	current->memcg_may_oom = 0;
515 516
}

517 518
static inline bool task_in_memcg_oom(struct task_struct *p)
{
T
Tejun Heo 已提交
519
	return p->memcg_in_oom;
520 521
}

522
bool mem_cgroup_oom_synchronize(bool wait);
523

A
Andrew Morton 已提交
524
#ifdef CONFIG_MEMCG_SWAP
525 526
extern int do_swap_account;
#endif
527

528 529
struct mem_cgroup *lock_page_memcg(struct page *page);
void __unlock_page_memcg(struct mem_cgroup *memcg);
J
Johannes Weiner 已提交
530
void unlock_page_memcg(struct page *page);
531

532
/* idx can be of type enum memcg_stat_item or node_stat_item */
533
static inline unsigned long memcg_page_state(struct mem_cgroup *memcg,
534
					     int idx)
535
{
536 537 538 539 540 541
	long x = atomic_long_read(&memcg->stat[idx]);
#ifdef CONFIG_SMP
	if (x < 0)
		x = 0;
#endif
	return x;
542 543
}

544
/* idx can be of type enum memcg_stat_item or node_stat_item */
545
static inline void __mod_memcg_state(struct mem_cgroup *memcg,
546
				     int idx, int val)
547
{
548 549 550 551 552 553 554 555 556 557 558
	long x;

	if (mem_cgroup_disabled())
		return;

	x = val + __this_cpu_read(memcg->stat_cpu->count[idx]);
	if (unlikely(abs(x) > MEMCG_CHARGE_BATCH)) {
		atomic_long_add(x, &memcg->stat[idx]);
		x = 0;
	}
	__this_cpu_write(memcg->stat_cpu->count[idx], x);
559 560
}

561
/* idx can be of type enum memcg_stat_item or node_stat_item */
562
static inline void mod_memcg_state(struct mem_cgroup *memcg,
563
				   int idx, int val)
564
{
565 566 567
	unsigned long flags;

	local_irq_save(flags);
568
	__mod_memcg_state(memcg, idx, val);
569
	local_irq_restore(flags);
570 571
}

M
Michal Hocko 已提交
572
/**
573
 * mod_memcg_page_state - update page state statistics
J
Johannes Weiner 已提交
574
 * @page: the page
M
Michal Hocko 已提交
575 576 577
 * @idx: page state item to account
 * @val: number of pages (positive or negative)
 *
578 579 580
 * The @page must be locked or the caller must use lock_page_memcg()
 * to prevent double accounting when the page is concurrently being
 * moved to another memcg:
581
 *
582
 *   lock_page(page) or lock_page_memcg(page)
583
 *   if (TestClearPageState(page))
584
 *     mod_memcg_page_state(page, state, -1);
585
 *   unlock_page(page) or unlock_page_memcg(page)
586 587
 *
 * Kernel pages are an exception to this, since they'll never move.
M
Michal Hocko 已提交
588
 */
589
static inline void __mod_memcg_page_state(struct page *page,
590
					  int idx, int val)
591 592 593 594 595
{
	if (page->mem_cgroup)
		__mod_memcg_state(page->mem_cgroup, idx, val);
}

596
static inline void mod_memcg_page_state(struct page *page,
597
					int idx, int val)
M
Michal Hocko 已提交
598
{
J
Johannes Weiner 已提交
599
	if (page->mem_cgroup)
600
		mod_memcg_state(page->mem_cgroup, idx, val);
M
Michal Hocko 已提交
601 602
}

603 604
static inline unsigned long lruvec_page_state(struct lruvec *lruvec,
					      enum node_stat_item idx)
605
{
606
	struct mem_cgroup_per_node *pn;
607
	long x;
608 609 610 611 612

	if (mem_cgroup_disabled())
		return node_page_state(lruvec_pgdat(lruvec), idx);

	pn = container_of(lruvec, struct mem_cgroup_per_node, lruvec);
613 614 615 616 617 618
	x = atomic_long_read(&pn->lruvec_stat[idx]);
#ifdef CONFIG_SMP
	if (x < 0)
		x = 0;
#endif
	return x;
619 620
}

621 622
static inline void __mod_lruvec_state(struct lruvec *lruvec,
				      enum node_stat_item idx, int val)
623
{
624
	struct mem_cgroup_per_node *pn;
625
	long x;
626

627
	/* Update node */
628
	__mod_node_page_state(lruvec_pgdat(lruvec), idx, val);
629

630 631
	if (mem_cgroup_disabled())
		return;
632

633
	pn = container_of(lruvec, struct mem_cgroup_per_node, lruvec);
634 635

	/* Update memcg */
636
	__mod_memcg_state(pn->memcg, idx, val);
637 638

	/* Update lruvec */
639 640 641 642 643 644
	x = val + __this_cpu_read(pn->lruvec_stat_cpu->count[idx]);
	if (unlikely(abs(x) > MEMCG_CHARGE_BATCH)) {
		atomic_long_add(x, &pn->lruvec_stat[idx]);
		x = 0;
	}
	__this_cpu_write(pn->lruvec_stat_cpu->count[idx], x);
645 646 647 648 649
}

static inline void mod_lruvec_state(struct lruvec *lruvec,
				    enum node_stat_item idx, int val)
{
650 651 652
	unsigned long flags;

	local_irq_save(flags);
653
	__mod_lruvec_state(lruvec, idx, val);
654
	local_irq_restore(flags);
655 656 657 658 659
}

static inline void __mod_lruvec_page_state(struct page *page,
					   enum node_stat_item idx, int val)
{
660 661
	pg_data_t *pgdat = page_pgdat(page);
	struct lruvec *lruvec;
662

663 664 665
	/* Untracked pages have no memcg, no lruvec. Update only the node */
	if (!page->mem_cgroup) {
		__mod_node_page_state(pgdat, idx, val);
666
		return;
667 668 669 670
	}

	lruvec = mem_cgroup_lruvec(pgdat, page->mem_cgroup);
	__mod_lruvec_state(lruvec, idx, val);
671 672 673 674 675
}

static inline void mod_lruvec_page_state(struct page *page,
					 enum node_stat_item idx, int val)
{
676 677 678
	unsigned long flags;

	local_irq_save(flags);
679
	__mod_lruvec_page_state(page, idx, val);
680
	local_irq_restore(flags);
681 682
}

683
unsigned long mem_cgroup_soft_limit_reclaim(pg_data_t *pgdat, int order,
684 685
						gfp_t gfp_mask,
						unsigned long *total_scanned);
D
David Rientjes 已提交
686

687
static inline void __count_memcg_events(struct mem_cgroup *memcg,
688 689
					enum vm_event_item idx,
					unsigned long count)
690
{
691 692 693 694 695 696 697 698 699 700 701
	unsigned long x;

	if (mem_cgroup_disabled())
		return;

	x = count + __this_cpu_read(memcg->stat_cpu->events[idx]);
	if (unlikely(x > MEMCG_CHARGE_BATCH)) {
		atomic_long_add(x, &memcg->events[idx]);
		x = 0;
	}
	__this_cpu_write(memcg->stat_cpu->events[idx], x);
702 703
}

704
static inline void count_memcg_events(struct mem_cgroup *memcg,
705 706
				      enum vm_event_item idx,
				      unsigned long count)
707
{
708 709 710
	unsigned long flags;

	local_irq_save(flags);
711
	__count_memcg_events(memcg, idx, count);
712
	local_irq_restore(flags);
713 714 715
}

static inline void count_memcg_page_event(struct page *page,
716
					  enum vm_event_item idx)
717 718 719 720 721 722 723
{
	if (page->mem_cgroup)
		count_memcg_events(page->mem_cgroup, idx, 1);
}

static inline void count_memcg_event_mm(struct mm_struct *mm,
					enum vm_event_item idx)
724
{
M
Michal Hocko 已提交
725 726
	struct mem_cgroup *memcg;

727 728
	if (mem_cgroup_disabled())
		return;
M
Michal Hocko 已提交
729 730 731

	rcu_read_lock();
	memcg = mem_cgroup_from_task(rcu_dereference(mm->owner));
R
Roman Gushchin 已提交
732
	if (likely(memcg))
733
		count_memcg_events(memcg, idx, 1);
M
Michal Hocko 已提交
734
	rcu_read_unlock();
735
}
736

737 738
static inline void memcg_memory_event(struct mem_cgroup *memcg,
				      enum memcg_memory_event event)
739
{
740
	atomic_long_inc(&memcg->memory_events[event]);
741 742 743
	cgroup_file_notify(&memcg->events_file);
}

R
Roman Gushchin 已提交
744 745 746 747 748 749 750 751 752 753 754 755 756 757 758
static inline void memcg_memory_event_mm(struct mm_struct *mm,
					 enum memcg_memory_event event)
{
	struct mem_cgroup *memcg;

	if (mem_cgroup_disabled())
		return;

	rcu_read_lock();
	memcg = mem_cgroup_from_task(rcu_dereference(mm->owner));
	if (likely(memcg))
		memcg_memory_event(memcg, event);
	rcu_read_unlock();
}

759
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
760
void mem_cgroup_split_huge_fixup(struct page *head);
761 762
#endif

A
Andrew Morton 已提交
763
#else /* CONFIG_MEMCG */
764 765 766 767

#define MEM_CGROUP_ID_SHIFT	0
#define MEM_CGROUP_ID_MAX	0

768 769
struct mem_cgroup;

770 771 772 773 774
static inline bool mem_cgroup_disabled(void)
{
	return true;
}

775 776
static inline void memcg_memory_event(struct mem_cgroup *memcg,
				      enum memcg_memory_event event)
777 778 779
{
}

R
Roman Gushchin 已提交
780 781 782 783 784
static inline void memcg_memory_event_mm(struct mm_struct *mm,
					 enum memcg_memory_event event)
{
}

R
Roman Gushchin 已提交
785 786
static inline enum mem_cgroup_protection mem_cgroup_protected(
	struct mem_cgroup *root, struct mem_cgroup *memcg)
787
{
R
Roman Gushchin 已提交
788
	return MEMCG_PROT_NONE;
789 790
}

791 792
static inline int mem_cgroup_try_charge(struct page *page, struct mm_struct *mm,
					gfp_t gfp_mask,
793 794
					struct mem_cgroup **memcgp,
					bool compound)
795
{
796
	*memcgp = NULL;
797 798 799
	return 0;
}

800 801 802 803 804 805 806 807 808 809
static inline int mem_cgroup_try_charge_delay(struct page *page,
					      struct mm_struct *mm,
					      gfp_t gfp_mask,
					      struct mem_cgroup **memcgp,
					      bool compound)
{
	*memcgp = NULL;
	return 0;
}

810 811
static inline void mem_cgroup_commit_charge(struct page *page,
					    struct mem_cgroup *memcg,
812
					    bool lrucare, bool compound)
813 814 815
{
}

816
static inline void mem_cgroup_cancel_charge(struct page *page,
817 818
					    struct mem_cgroup *memcg,
					    bool compound)
819 820 821
{
}

822
static inline void mem_cgroup_uncharge(struct page *page)
823 824 825
{
}

826
static inline void mem_cgroup_uncharge_list(struct list_head *page_list)
827 828 829
{
}

830
static inline void mem_cgroup_migrate(struct page *old, struct page *new)
831 832 833
{
}

834
static inline struct lruvec *mem_cgroup_lruvec(struct pglist_data *pgdat,
835
				struct mem_cgroup *memcg)
K
KAMEZAWA Hiroyuki 已提交
836
{
837
	return node_lruvec(pgdat);
K
KAMEZAWA Hiroyuki 已提交
838 839
}

840
static inline struct lruvec *mem_cgroup_page_lruvec(struct page *page,
M
Mel Gorman 已提交
841
						    struct pglist_data *pgdat)
842
{
M
Mel Gorman 已提交
843
	return &pgdat->lruvec;
844 845
}

846
static inline bool mm_match_cgroup(struct mm_struct *mm,
847
		struct mem_cgroup *memcg)
848
{
849
	return true;
850 851
}

852 853
static inline bool task_in_mem_cgroup(struct task_struct *task,
				      const struct mem_cgroup *memcg)
854
{
855
	return true;
856 857
}

858 859 860 861
static inline void mem_cgroup_put(struct mem_cgroup *memcg)
{
}

862 863 864 865 866 867 868 869 870 871 872 873 874
static inline struct mem_cgroup *
mem_cgroup_iter(struct mem_cgroup *root,
		struct mem_cgroup *prev,
		struct mem_cgroup_reclaim_cookie *reclaim)
{
	return NULL;
}

static inline void mem_cgroup_iter_break(struct mem_cgroup *root,
					 struct mem_cgroup *prev)
{
}

875 876 877 878 879 880
static inline int mem_cgroup_scan_tasks(struct mem_cgroup *memcg,
		int (*fn)(struct task_struct *, void *), void *arg)
{
	return 0;
}

881
static inline unsigned short mem_cgroup_id(struct mem_cgroup *memcg)
882
{
883 884 885 886 887 888 889 890
	return 0;
}

static inline struct mem_cgroup *mem_cgroup_from_id(unsigned short id)
{
	WARN_ON_ONCE(id);
	/* XXX: This should always return root_mem_cgroup */
	return NULL;
891
}
892

893 894 895 896 897
static inline struct mem_cgroup *lruvec_memcg(struct lruvec *lruvec)
{
	return NULL;
}

898
static inline bool mem_cgroup_online(struct mem_cgroup *memcg)
899
{
900
	return true;
901 902
}

903
static inline unsigned long
904
mem_cgroup_get_lru_size(struct lruvec *lruvec, enum lru_list lru)
905 906 907
{
	return 0;
}
908 909 910 911 912 913
static inline
unsigned long mem_cgroup_get_zone_lru_size(struct lruvec *lruvec,
		enum lru_list lru, int zone_idx)
{
	return 0;
}
914

915 916 917 918 919 920 921
static inline unsigned long
mem_cgroup_node_nr_lru_pages(struct mem_cgroup *memcg,
			     int nid, unsigned int lru_mask)
{
	return 0;
}

922
static inline unsigned long mem_cgroup_get_max(struct mem_cgroup *memcg)
923 924 925 926
{
	return 0;
}

927 928 929 930 931
static inline void
mem_cgroup_print_oom_info(struct mem_cgroup *memcg, struct task_struct *p)
{
}

932 933 934 935 936 937
static inline struct mem_cgroup *lock_page_memcg(struct page *page)
{
	return NULL;
}

static inline void __unlock_page_memcg(struct mem_cgroup *memcg)
938 939 940
{
}

J
Johannes Weiner 已提交
941
static inline void unlock_page_memcg(struct page *page)
942 943 944
{
}

945 946 947 948
static inline void mem_cgroup_handle_over_high(void)
{
}

949
static inline void mem_cgroup_oom_enable(void)
950 951 952
{
}

953
static inline void mem_cgroup_oom_disable(void)
954 955 956
{
}

957 958 959 960 961
static inline bool task_in_memcg_oom(struct task_struct *p)
{
	return false;
}

962
static inline bool mem_cgroup_oom_synchronize(bool wait)
963 964 965 966
{
	return false;
}

967
static inline unsigned long memcg_page_state(struct mem_cgroup *memcg,
968
					     int idx)
969 970 971 972
{
	return 0;
}

973
static inline void __mod_memcg_state(struct mem_cgroup *memcg,
974
				     int idx,
975
				     int nr)
976 977 978
{
}

979
static inline void mod_memcg_state(struct mem_cgroup *memcg,
980
				   int idx,
981
				   int nr)
982 983 984
{
}

985
static inline void __mod_memcg_page_state(struct page *page,
986
					  int idx,
987
					  int nr)
988 989 990
{
}

991
static inline void mod_memcg_page_state(struct page *page,
992
					int idx,
993
					int nr)
994 995 996
{
}

997 998
static inline unsigned long lruvec_page_state(struct lruvec *lruvec,
					      enum node_stat_item idx)
999
{
1000
	return node_page_state(lruvec_pgdat(lruvec), idx);
1001 1002
}

1003 1004
static inline void __mod_lruvec_state(struct lruvec *lruvec,
				      enum node_stat_item idx, int val)
1005
{
1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024
	__mod_node_page_state(lruvec_pgdat(lruvec), idx, val);
}

static inline void mod_lruvec_state(struct lruvec *lruvec,
				    enum node_stat_item idx, int val)
{
	mod_node_page_state(lruvec_pgdat(lruvec), idx, val);
}

static inline void __mod_lruvec_page_state(struct page *page,
					   enum node_stat_item idx, int val)
{
	__mod_node_page_state(page_pgdat(page), idx, val);
}

static inline void mod_lruvec_page_state(struct page *page,
					 enum node_stat_item idx, int val)
{
	mod_node_page_state(page_pgdat(page), idx, val);
1025 1026
}

1027
static inline
1028
unsigned long mem_cgroup_soft_limit_reclaim(pg_data_t *pgdat, int order,
1029 1030
					    gfp_t gfp_mask,
					    unsigned long *total_scanned)
1031
{
1032
	return 0;
1033 1034
}

1035
static inline void mem_cgroup_split_huge_fixup(struct page *head)
1036 1037 1038
{
}

1039 1040 1041 1042 1043 1044 1045
static inline void count_memcg_events(struct mem_cgroup *memcg,
				      enum vm_event_item idx,
				      unsigned long count)
{
}

static inline void count_memcg_page_event(struct page *page,
1046
					  int idx)
1047 1048 1049
{
}

1050
static inline
1051
void count_memcg_event_mm(struct mm_struct *mm, enum vm_event_item idx)
1052 1053
{
}
A
Andrew Morton 已提交
1054
#endif /* CONFIG_MEMCG */
1055

1056
/* idx can be of type enum memcg_stat_item or node_stat_item */
1057
static inline void __inc_memcg_state(struct mem_cgroup *memcg,
1058
				     int idx)
1059 1060 1061 1062
{
	__mod_memcg_state(memcg, idx, 1);
}

1063
/* idx can be of type enum memcg_stat_item or node_stat_item */
1064
static inline void __dec_memcg_state(struct mem_cgroup *memcg,
1065
				     int idx)
1066 1067 1068 1069
{
	__mod_memcg_state(memcg, idx, -1);
}

1070
/* idx can be of type enum memcg_stat_item or node_stat_item */
1071
static inline void __inc_memcg_page_state(struct page *page,
1072
					  int idx)
1073 1074 1075 1076
{
	__mod_memcg_page_state(page, idx, 1);
}

1077
/* idx can be of type enum memcg_stat_item or node_stat_item */
1078
static inline void __dec_memcg_page_state(struct page *page,
1079
					  int idx)
1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107
{
	__mod_memcg_page_state(page, idx, -1);
}

static inline void __inc_lruvec_state(struct lruvec *lruvec,
				      enum node_stat_item idx)
{
	__mod_lruvec_state(lruvec, idx, 1);
}

static inline void __dec_lruvec_state(struct lruvec *lruvec,
				      enum node_stat_item idx)
{
	__mod_lruvec_state(lruvec, idx, -1);
}

static inline void __inc_lruvec_page_state(struct page *page,
					   enum node_stat_item idx)
{
	__mod_lruvec_page_state(page, idx, 1);
}

static inline void __dec_lruvec_page_state(struct page *page,
					   enum node_stat_item idx)
{
	__mod_lruvec_page_state(page, idx, -1);
}

1108
/* idx can be of type enum memcg_stat_item or node_stat_item */
1109
static inline void inc_memcg_state(struct mem_cgroup *memcg,
1110
				   int idx)
1111 1112 1113 1114
{
	mod_memcg_state(memcg, idx, 1);
}

1115
/* idx can be of type enum memcg_stat_item or node_stat_item */
1116
static inline void dec_memcg_state(struct mem_cgroup *memcg,
1117
				   int idx)
1118 1119 1120 1121
{
	mod_memcg_state(memcg, idx, -1);
}

1122
/* idx can be of type enum memcg_stat_item or node_stat_item */
1123
static inline void inc_memcg_page_state(struct page *page,
1124
					int idx)
1125 1126 1127 1128
{
	mod_memcg_page_state(page, idx, 1);
}

1129
/* idx can be of type enum memcg_stat_item or node_stat_item */
1130
static inline void dec_memcg_page_state(struct page *page,
1131
					int idx)
1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159
{
	mod_memcg_page_state(page, idx, -1);
}

static inline void inc_lruvec_state(struct lruvec *lruvec,
				    enum node_stat_item idx)
{
	mod_lruvec_state(lruvec, idx, 1);
}

static inline void dec_lruvec_state(struct lruvec *lruvec,
				    enum node_stat_item idx)
{
	mod_lruvec_state(lruvec, idx, -1);
}

static inline void inc_lruvec_page_state(struct page *page,
					 enum node_stat_item idx)
{
	mod_lruvec_page_state(page, idx, 1);
}

static inline void dec_lruvec_page_state(struct page *page,
					 enum node_stat_item idx)
{
	mod_lruvec_page_state(page, idx, -1);
}

1160
#ifdef CONFIG_CGROUP_WRITEBACK
T
Tejun Heo 已提交
1161 1162

struct wb_domain *mem_cgroup_wb_domain(struct bdi_writeback *wb);
1163 1164 1165
void mem_cgroup_wb_stats(struct bdi_writeback *wb, unsigned long *pfilepages,
			 unsigned long *pheadroom, unsigned long *pdirty,
			 unsigned long *pwriteback);
T
Tejun Heo 已提交
1166 1167 1168 1169 1170 1171 1172 1173

#else	/* CONFIG_CGROUP_WRITEBACK */

static inline struct wb_domain *mem_cgroup_wb_domain(struct bdi_writeback *wb)
{
	return NULL;
}

1174
static inline void mem_cgroup_wb_stats(struct bdi_writeback *wb,
1175 1176
				       unsigned long *pfilepages,
				       unsigned long *pheadroom,
1177 1178 1179 1180 1181
				       unsigned long *pdirty,
				       unsigned long *pwriteback)
{
}

T
Tejun Heo 已提交
1182
#endif	/* CONFIG_CGROUP_WRITEBACK */
1183

G
Glauber Costa 已提交
1184
struct sock;
1185 1186
bool mem_cgroup_charge_skmem(struct mem_cgroup *memcg, unsigned int nr_pages);
void mem_cgroup_uncharge_skmem(struct mem_cgroup *memcg, unsigned int nr_pages);
1187
#ifdef CONFIG_MEMCG
1188 1189
extern struct static_key_false memcg_sockets_enabled_key;
#define mem_cgroup_sockets_enabled static_branch_unlikely(&memcg_sockets_enabled_key)
1190 1191
void mem_cgroup_sk_alloc(struct sock *sk);
void mem_cgroup_sk_free(struct sock *sk);
1192
static inline bool mem_cgroup_under_socket_pressure(struct mem_cgroup *memcg)
1193
{
1194
	if (!cgroup_subsys_on_dfl(memory_cgrp_subsys) && memcg->tcpmem_pressure)
1195 1196 1197 1198 1199 1200
		return true;
	do {
		if (time_before(jiffies, memcg->socket_pressure))
			return true;
	} while ((memcg = parent_mem_cgroup(memcg)));
	return false;
1201 1202
}
#else
1203
#define mem_cgroup_sockets_enabled 0
1204 1205
static inline void mem_cgroup_sk_alloc(struct sock *sk) { };
static inline void mem_cgroup_sk_free(struct sock *sk) { };
1206
static inline bool mem_cgroup_under_socket_pressure(struct mem_cgroup *memcg)
1207 1208 1209 1210
{
	return false;
}
#endif
1211

1212 1213 1214 1215 1216 1217 1218
struct kmem_cache *memcg_kmem_get_cache(struct kmem_cache *cachep);
void memcg_kmem_put_cache(struct kmem_cache *cachep);
int memcg_kmem_charge_memcg(struct page *page, gfp_t gfp, int order,
			    struct mem_cgroup *memcg);
int memcg_kmem_charge(struct page *page, gfp_t gfp, int order);
void memcg_kmem_uncharge(struct page *page, int order);

1219
#if defined(CONFIG_MEMCG) && !defined(CONFIG_SLOB)
1220
extern struct static_key_false memcg_kmem_enabled_key;
1221
extern struct workqueue_struct *memcg_kmem_cache_wq;
1222

1223
extern int memcg_nr_cache_ids;
1224 1225
void memcg_get_cache_ids(void);
void memcg_put_cache_ids(void);
1226 1227 1228 1229 1230 1231

/*
 * Helper macro to loop through all memcg-specific caches. Callers must still
 * check if the cache is valid (it is either valid or NULL).
 * the slab_mutex must be held when looping through those caches
 */
1232
#define for_each_memcg_cache_index(_idx)	\
1233
	for ((_idx) = 0; (_idx) < memcg_nr_cache_ids; (_idx)++)
1234

1235 1236
static inline bool memcg_kmem_enabled(void)
{
1237
	return static_branch_unlikely(&memcg_kmem_enabled_key);
1238 1239
}

M
Michal Hocko 已提交
1240
/*
J
Jesper Dangaard Brouer 已提交
1241
 * helper for accessing a memcg's index. It will be used as an index in the
M
Michal Hocko 已提交
1242 1243 1244 1245 1246 1247 1248
 * child cache array in kmem_cache, and also to derive its name. This function
 * will return -1 when this is not a kmem-limited memcg.
 */
static inline int memcg_cache_id(struct mem_cgroup *memcg)
{
	return memcg ? memcg->kmemcg_id : -1;
}
1249

1250
#else
1251 1252 1253
#define for_each_memcg_cache_index(_idx)	\
	for (; NULL; )

1254 1255 1256 1257 1258
static inline bool memcg_kmem_enabled(void)
{
	return false;
}

1259 1260 1261 1262 1263
static inline int memcg_cache_id(struct mem_cgroup *memcg)
{
	return -1;
}

1264 1265 1266 1267 1268 1269 1270 1271
static inline void memcg_get_cache_ids(void)
{
}

static inline void memcg_put_cache_ids(void)
{
}

1272 1273
#endif /* CONFIG_MEMCG && !CONFIG_SLOB */

B
Balbir Singh 已提交
1274
#endif /* _LINUX_MEMCONTROL_H */