memcontrol.c 186 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0-or-later
2 3 4 5 6
/* memcontrol.c - Memory Controller
 *
 * Copyright IBM Corporation, 2007
 * Author Balbir Singh <balbir@linux.vnet.ibm.com>
 *
7 8 9
 * Copyright 2007 OpenVZ SWsoft Inc
 * Author: Pavel Emelianov <xemul@openvz.org>
 *
10 11 12 13
 * Memory thresholds
 * Copyright (C) 2009 Nokia Corporation
 * Author: Kirill A. Shutemov
 *
14 15 16 17
 * Kernel Memory Controller
 * Copyright (C) 2012 Parallels Inc. and Google Inc.
 * Authors: Glauber Costa and Suleiman Souhlal
 *
18 19 20 21 22
 * Native page reclaim
 * Charge lifetime sanitation
 * Lockless page tracking & accounting
 * Unified hierarchy configuration model
 * Copyright (C) 2015 Red Hat, Inc., Johannes Weiner
23 24
 */

25
#include <linux/page_counter.h>
26 27
#include <linux/memcontrol.h>
#include <linux/cgroup.h>
28
#include <linux/pagewalk.h>
29
#include <linux/sched/mm.h>
30
#include <linux/shmem_fs.h>
31
#include <linux/hugetlb.h>
KAMEZAWA Hiroyuki's avatar
KAMEZAWA Hiroyuki committed
32
#include <linux/pagemap.h>
33
#include <linux/vm_event_item.h>
34
#include <linux/smp.h>
35
#include <linux/page-flags.h>
36
#include <linux/backing-dev.h>
37 38
#include <linux/bit_spinlock.h>
#include <linux/rcupdate.h>
39
#include <linux/limits.h>
40
#include <linux/export.h>
41
#include <linux/mutex.h>
42
#include <linux/rbtree.h>
43
#include <linux/slab.h>
44
#include <linux/swap.h>
45
#include <linux/swapops.h>
46
#include <linux/spinlock.h>
47
#include <linux/eventfd.h>
48
#include <linux/poll.h>
49
#include <linux/sort.h>
50
#include <linux/fs.h>
51
#include <linux/seq_file.h>
52
#include <linux/vmpressure.h>
53
#include <linux/mm_inline.h>
54
#include <linux/swap_cgroup.h>
55
#include <linux/cpu.h>
56
#include <linux/oom.h>
57
#include <linux/lockdep.h>
58
#include <linux/file.h>
59
#include <linux/tracehook.h>
60
#include <linux/psi.h>
61
#include <linux/seq_buf.h>
KAMEZAWA Hiroyuki's avatar
KAMEZAWA Hiroyuki committed
62
#include "internal.h"
Glauber Costa's avatar
Glauber Costa committed
63
#include <net/sock.h>
Michal Hocko's avatar
Michal Hocko committed
64
#include <net/ip.h>
65
#include "slab.h"
66

67
#include <linux/uaccess.h>
68

69 70
#include <trace/events/vmscan.h>

71 72
struct cgroup_subsys memory_cgrp_subsys __read_mostly;
EXPORT_SYMBOL(memory_cgrp_subsys);
73

74 75
struct mem_cgroup *root_mem_cgroup __read_mostly;

76
#define MEM_CGROUP_RECLAIM_RETRIES	5
77

78 79 80
/* Socket memory accounting disabled? */
static bool cgroup_memory_nosocket;

81 82 83
/* Kernel memory accounting disabled? */
static bool cgroup_memory_nokmem;

84
/* Whether the swap controller is active */
Andrew Morton's avatar
Andrew Morton committed
85
#ifdef CONFIG_MEMCG_SWAP
86 87
int do_swap_account __read_mostly;
#else
88
#define do_swap_account		0
89 90
#endif

91 92 93 94
#ifdef CONFIG_CGROUP_WRITEBACK
static DECLARE_WAIT_QUEUE_HEAD(memcg_cgwb_frn_waitq);
#endif

95 96 97 98 99 100
/* Whether legacy memory+swap accounting is active */
static bool do_memsw_account(void)
{
	return !cgroup_subsys_on_dfl(memory_cgrp_subsys) && do_swap_account;
}

101 102
#define THRESHOLDS_EVENTS_TARGET 128
#define SOFTLIMIT_EVENTS_TARGET 1024
103

104 105 106 107 108
/*
 * Cgroups above their limits are maintained in a RB-Tree, independent of
 * their hierarchy representation
 */

109
struct mem_cgroup_tree_per_node {
110
	struct rb_root rb_root;
111
	struct rb_node *rb_rightmost;
112 113 114 115 116 117 118 119 120
	spinlock_t lock;
};

struct mem_cgroup_tree {
	struct mem_cgroup_tree_per_node *rb_tree_per_node[MAX_NUMNODES];
};

static struct mem_cgroup_tree soft_limit_tree __read_mostly;

KAMEZAWA Hiroyuki's avatar
KAMEZAWA Hiroyuki committed
121 122 123 124 125
/* for OOM */
struct mem_cgroup_eventfd_list {
	struct list_head list;
	struct eventfd_ctx *eventfd;
};
126

127 128 129
/*
 * cgroup_event represents events which userspace want to receive.
 */
130
struct mem_cgroup_event {
131
	/*
132
	 * memcg which the event belongs to.
133
	 */
134
	struct mem_cgroup *memcg;
135 136 137 138 139 140 141 142
	/*
	 * eventfd to signal userspace about the event.
	 */
	struct eventfd_ctx *eventfd;
	/*
	 * Each of these stored in a list by the cgroup.
	 */
	struct list_head list;
143 144 145 146 147
	/*
	 * register_event() callback will be used to add new userspace
	 * waiter for changes related to this event.  Use eventfd_signal()
	 * on eventfd to send notification to userspace.
	 */
148
	int (*register_event)(struct mem_cgroup *memcg,
Tejun Heo's avatar
Tejun Heo committed
149
			      struct eventfd_ctx *eventfd, const char *args);
150 151 152 153 154
	/*
	 * unregister_event() callback will be called when userspace closes
	 * the eventfd or on cgroup removing.  This callback must be set,
	 * if you want provide notification functionality.
	 */
155
	void (*unregister_event)(struct mem_cgroup *memcg,
156
				 struct eventfd_ctx *eventfd);
157 158 159 160 161 162
	/*
	 * All fields below needed to unregister event when
	 * userspace closes eventfd.
	 */
	poll_table pt;
	wait_queue_head_t *wqh;
163
	wait_queue_entry_t wait;
164 165 166
	struct work_struct remove;
};

167 168
static void mem_cgroup_threshold(struct mem_cgroup *memcg);
static void mem_cgroup_oom_notify(struct mem_cgroup *memcg);
169

170 171
/* Stuffs for move charges at task migration. */
/*
172
 * Types of charges to be moved.
173
 */
174 175 176
#define MOVE_ANON	0x1U
#define MOVE_FILE	0x2U
#define MOVE_MASK	(MOVE_ANON | MOVE_FILE)
177

178 179
/* "mc" and its members are protected by cgroup_mutex */
static struct move_charge_struct {
180
	spinlock_t	  lock; /* for from, to */
181
	struct mm_struct  *mm;
182 183
	struct mem_cgroup *from;
	struct mem_cgroup *to;
184
	unsigned long flags;
185
	unsigned long precharge;
186
	unsigned long moved_charge;
187
	unsigned long moved_swap;
188 189 190
	struct task_struct *moving_task;	/* a task moving charges */
	wait_queue_head_t waitq;		/* a waitq for other context */
} mc = {
191
	.lock = __SPIN_LOCK_UNLOCKED(mc.lock),
192 193
	.waitq = __WAIT_QUEUE_HEAD_INITIALIZER(mc.waitq),
};
194

195 196 197 198
/*
 * Maximum loops in mem_cgroup_hierarchical_reclaim(), used for soft
 * limit reclaim to prevent infinite loops, if they ever occur.
 */
199
#define	MEM_CGROUP_MAX_RECLAIM_LOOPS		100
200
#define	MEM_CGROUP_MAX_SOFT_LIMIT_RECLAIM_LOOPS	2
201

202 203
enum charge_type {
	MEM_CGROUP_CHARGE_TYPE_CACHE = 0,
204
	MEM_CGROUP_CHARGE_TYPE_ANON,
KAMEZAWA Hiroyuki's avatar
KAMEZAWA Hiroyuki committed
205
	MEM_CGROUP_CHARGE_TYPE_SWAPOUT,	/* for accounting swapcache */
KAMEZAWA Hiroyuki's avatar
KAMEZAWA Hiroyuki committed
206
	MEM_CGROUP_CHARGE_TYPE_DROP,	/* a page was unused swap cache */
207 208 209
	NR_CHARGE_TYPE,
};

210
/* for encoding cft->private value on file */
211 212 213 214
enum res_type {
	_MEM,
	_MEMSWAP,
	_OOM_TYPE,
215
	_KMEM,
Vladimir Davydov's avatar
Vladimir Davydov committed
216
	_TCP,
217 218
};

219 220
#define MEMFILE_PRIVATE(x, val)	((x) << 16 | (val))
#define MEMFILE_TYPE(val)	((val) >> 16 & 0xffff)
221
#define MEMFILE_ATTR(val)	((val) & 0xffff)
KAMEZAWA Hiroyuki's avatar
KAMEZAWA Hiroyuki committed
222 223
/* Used for OOM nofiier */
#define OOM_CONTROL		(0)
224

225 226 227 228 229 230 231 232 233 234 235 236 237 238 239
/*
 * Iteration constructs for visiting all cgroups (under a tree).  If
 * loops are exited prematurely (break), mem_cgroup_iter_break() must
 * be used for reference counting.
 */
#define for_each_mem_cgroup_tree(iter, root)		\
	for (iter = mem_cgroup_iter(root, NULL, NULL);	\
	     iter != NULL;				\
	     iter = mem_cgroup_iter(root, iter, NULL))

#define for_each_mem_cgroup(iter)			\
	for (iter = mem_cgroup_iter(NULL, NULL, NULL);	\
	     iter != NULL;				\
	     iter = mem_cgroup_iter(NULL, iter, NULL))

240 241 242 243 244 245
static inline bool should_force_charge(void)
{
	return tsk_is_oom_victim(current) || fatal_signal_pending(current) ||
		(current->flags & PF_EXITING);
}

246 247 248 249 250 251 252 253 254 255 256 257 258
/* Some nice accessors for the vmpressure. */
struct vmpressure *memcg_to_vmpressure(struct mem_cgroup *memcg)
{
	if (!memcg)
		memcg = root_mem_cgroup;
	return &memcg->vmpressure;
}

struct cgroup_subsys_state *vmpressure_to_css(struct vmpressure *vmpr)
{
	return &container_of(vmpr, struct mem_cgroup, vmpressure)->css;
}

259
#ifdef CONFIG_MEMCG_KMEM
260
/*
261
 * This will be the memcg's index in each cache's ->memcg_params.memcg_caches.
Li Zefan's avatar
Li Zefan committed
262 263 264 265 266
 * The main reason for not using cgroup id for this:
 *  this works better in sparse environments, where we have a lot of memcgs,
 *  but only a few kmem-limited. Or also, if we have, for instance, 200
 *  memcgs, and none but the 200th is kmem-limited, we'd have to have a
 *  200 entry array for that.
267
 *
268 269
 * The current size of the caches array is stored in memcg_nr_cache_ids. It
 * will double each time we have to increase it.
270
 */
271 272
static DEFINE_IDA(memcg_cache_ida);
int memcg_nr_cache_ids;
273

274 275 276 277 278 279 280 281 282 283 284 285 286
/* Protects memcg_nr_cache_ids */
static DECLARE_RWSEM(memcg_cache_ids_sem);

void memcg_get_cache_ids(void)
{
	down_read(&memcg_cache_ids_sem);
}

void memcg_put_cache_ids(void)
{
	up_read(&memcg_cache_ids_sem);
}

287 288 289 290 291 292
/*
 * MIN_SIZE is different than 1, because we would like to avoid going through
 * the alloc/free process all the time. In a small machine, 4 kmem-limited
 * cgroups is a reasonable guess. In the future, it could be a parameter or
 * tunable, but that is strictly not necessary.
 *
Li Zefan's avatar
Li Zefan committed
293
 * MAX_SIZE should be as large as the number of cgrp_ids. Ideally, we could get
294 295
 * this constant directly from cgroup, but it is understandable that this is
 * better kept as an internal representation in cgroup.c. In any case, the
Li Zefan's avatar
Li Zefan committed
296
 * cgrp_id space is not getting any smaller, and we don't have to necessarily
297 298 299
 * increase ours as well if it increases.
 */
#define MEMCG_CACHES_MIN_SIZE 4
Li Zefan's avatar
Li Zefan committed
300
#define MEMCG_CACHES_MAX_SIZE MEM_CGROUP_ID_MAX
301

302 303 304 305 306 307
/*
 * A lot of the calls to the cache allocation functions are expected to be
 * inlined by the compiler. Since the calls to memcg_kmem_get_cache are
 * conditional to this static branch, we'll have to allow modules that does
 * kmem_cache_alloc and the such to see this symbol as well
 */
308
DEFINE_STATIC_KEY_FALSE(memcg_kmem_enabled_key);
309
EXPORT_SYMBOL(memcg_kmem_enabled_key);
310

311
struct workqueue_struct *memcg_kmem_cache_wq;
312
#endif
313

314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411
static int memcg_shrinker_map_size;
static DEFINE_MUTEX(memcg_shrinker_map_mutex);

static void memcg_free_shrinker_map_rcu(struct rcu_head *head)
{
	kvfree(container_of(head, struct memcg_shrinker_map, rcu));
}

static int memcg_expand_one_shrinker_map(struct mem_cgroup *memcg,
					 int size, int old_size)
{
	struct memcg_shrinker_map *new, *old;
	int nid;

	lockdep_assert_held(&memcg_shrinker_map_mutex);

	for_each_node(nid) {
		old = rcu_dereference_protected(
			mem_cgroup_nodeinfo(memcg, nid)->shrinker_map, true);
		/* Not yet online memcg */
		if (!old)
			return 0;

		new = kvmalloc(sizeof(*new) + size, GFP_KERNEL);
		if (!new)
			return -ENOMEM;

		/* Set all old bits, clear all new bits */
		memset(new->map, (int)0xff, old_size);
		memset((void *)new->map + old_size, 0, size - old_size);

		rcu_assign_pointer(memcg->nodeinfo[nid]->shrinker_map, new);
		call_rcu(&old->rcu, memcg_free_shrinker_map_rcu);
	}

	return 0;
}

static void memcg_free_shrinker_maps(struct mem_cgroup *memcg)
{
	struct mem_cgroup_per_node *pn;
	struct memcg_shrinker_map *map;
	int nid;

	if (mem_cgroup_is_root(memcg))
		return;

	for_each_node(nid) {
		pn = mem_cgroup_nodeinfo(memcg, nid);
		map = rcu_dereference_protected(pn->shrinker_map, true);
		if (map)
			kvfree(map);
		rcu_assign_pointer(pn->shrinker_map, NULL);
	}
}

static int memcg_alloc_shrinker_maps(struct mem_cgroup *memcg)
{
	struct memcg_shrinker_map *map;
	int nid, size, ret = 0;

	if (mem_cgroup_is_root(memcg))
		return 0;

	mutex_lock(&memcg_shrinker_map_mutex);
	size = memcg_shrinker_map_size;
	for_each_node(nid) {
		map = kvzalloc(sizeof(*map) + size, GFP_KERNEL);
		if (!map) {
			memcg_free_shrinker_maps(memcg);
			ret = -ENOMEM;
			break;
		}
		rcu_assign_pointer(memcg->nodeinfo[nid]->shrinker_map, map);
	}
	mutex_unlock(&memcg_shrinker_map_mutex);

	return ret;
}

int memcg_expand_shrinker_maps(int new_id)
{
	int size, old_size, ret = 0;
	struct mem_cgroup *memcg;

	size = DIV_ROUND_UP(new_id + 1, BITS_PER_LONG) * sizeof(unsigned long);
	old_size = memcg_shrinker_map_size;
	if (size <= old_size)
		return 0;

	mutex_lock(&memcg_shrinker_map_mutex);
	if (!root_mem_cgroup)
		goto unlock;

	for_each_mem_cgroup(memcg) {
		if (mem_cgroup_is_root(memcg))
			continue;
		ret = memcg_expand_one_shrinker_map(memcg, size, old_size);
412 413
		if (ret) {
			mem_cgroup_iter_break(NULL, memcg);
414
			goto unlock;
415
		}
416 417 418 419 420 421 422
	}
unlock:
	if (!ret)
		memcg_shrinker_map_size = size;
	mutex_unlock(&memcg_shrinker_map_mutex);
	return ret;
}
423 424 425 426 427 428 429 430

void memcg_set_shrinker_bit(struct mem_cgroup *memcg, int nid, int shrinker_id)
{
	if (shrinker_id >= 0 && memcg && !mem_cgroup_is_root(memcg)) {
		struct memcg_shrinker_map *map;

		rcu_read_lock();
		map = rcu_dereference(memcg->nodeinfo[nid]->shrinker_map);
431 432
		/* Pairs with smp mb in shrink_slab() */
		smp_mb__before_atomic();
433 434 435 436 437
		set_bit(shrinker_id, map->map);
		rcu_read_unlock();
	}
}

438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454
/**
 * mem_cgroup_css_from_page - css of the memcg associated with a page
 * @page: page of interest
 *
 * If memcg is bound to the default hierarchy, css of the memcg associated
 * with @page is returned.  The returned css remains associated with @page
 * until it is released.
 *
 * If memcg is bound to a traditional hierarchy, the css of root_mem_cgroup
 * is returned.
 */
struct cgroup_subsys_state *mem_cgroup_css_from_page(struct page *page)
{
	struct mem_cgroup *memcg;

	memcg = page->mem_cgroup;

455
	if (!memcg || !cgroup_subsys_on_dfl(memory_cgrp_subsys))
456 457 458 459 460
		memcg = root_mem_cgroup;

	return &memcg->css;
}

461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479
/**
 * page_cgroup_ino - return inode number of the memcg a page is charged to
 * @page: the page
 *
 * Look up the closest online ancestor of the memory cgroup @page is charged to
 * and return its inode number or 0 if @page is not charged to any cgroup. It
 * is safe to call this function without holding a reference to @page.
 *
 * Note, this function is inherently racy, because there is nothing to prevent
 * the cgroup inode from getting torn down and potentially reallocated a moment
 * after page_cgroup_ino() returns, so it only should be used by callers that
 * do not care (such as procfs interfaces).
 */
ino_t page_cgroup_ino(struct page *page)
{
	struct mem_cgroup *memcg;
	unsigned long ino = 0;

	rcu_read_lock();
480
	if (PageSlab(page) && !PageTail(page))
481 482 483
		memcg = memcg_from_slab_page(page);
	else
		memcg = READ_ONCE(page->mem_cgroup);
484 485 486 487 488 489 490 491
	while (memcg && !(memcg->css.flags & CSS_ONLINE))
		memcg = parent_mem_cgroup(memcg);
	if (memcg)
		ino = cgroup_ino(memcg->css.cgroup);
	rcu_read_unlock();
	return ino;
}

492 493
static struct mem_cgroup_per_node *
mem_cgroup_page_nodeinfo(struct mem_cgroup *memcg, struct page *page)
494
{
495
	int nid = page_to_nid(page);
496

497
	return memcg->nodeinfo[nid];
498 499
}

500 501
static struct mem_cgroup_tree_per_node *
soft_limit_tree_node(int nid)
502
{
503
	return soft_limit_tree.rb_tree_per_node[nid];
504 505
}

506
static struct mem_cgroup_tree_per_node *
507 508 509 510
soft_limit_tree_from_page(struct page *page)
{
	int nid = page_to_nid(page);

511
	return soft_limit_tree.rb_tree_per_node[nid];
512 513
}

514 515
static void __mem_cgroup_insert_exceeded(struct mem_cgroup_per_node *mz,
					 struct mem_cgroup_tree_per_node *mctz,
516
					 unsigned long new_usage_in_excess)
517 518 519
{
	struct rb_node **p = &mctz->rb_root.rb_node;
	struct rb_node *parent = NULL;
520
	struct mem_cgroup_per_node *mz_node;
521
	bool rightmost = true;
522 523 524 525 526 527 528 529 530

	if (mz->on_tree)
		return;

	mz->usage_in_excess = new_usage_in_excess;
	if (!mz->usage_in_excess)
		return;
	while (*p) {
		parent = *p;
531
		mz_node = rb_entry(parent, struct mem_cgroup_per_node,
532
					tree_node);
533
		if (mz->usage_in_excess < mz_node->usage_in_excess) {
534
			p = &(*p)->rb_left;
535 536 537
			rightmost = false;
		}

538 539 540 541 542 543 544
		/*
		 * We can't avoid mem cgroups that are over their soft
		 * limit by the same amount
		 */
		else if (mz->usage_in_excess >= mz_node->usage_in_excess)
			p = &(*p)->rb_right;
	}
545 546 547 548

	if (rightmost)
		mctz->rb_rightmost = &mz->tree_node;

549 550 551 552 553
	rb_link_node(&mz->tree_node, parent, p);
	rb_insert_color(&mz->tree_node, &mctz->rb_root);
	mz->on_tree = true;
}

554 555
static void __mem_cgroup_remove_exceeded(struct mem_cgroup_per_node *mz,
					 struct mem_cgroup_tree_per_node *mctz)
556 557 558
{
	if (!mz->on_tree)
		return;
559 560 561 562

	if (&mz->tree_node == mctz->rb_rightmost)
		mctz->rb_rightmost = rb_prev(&mz->tree_node);

563 564 565 566
	rb_erase(&mz->tree_node, &mctz->rb_root);
	mz->on_tree = false;
}

567 568
static void mem_cgroup_remove_exceeded(struct mem_cgroup_per_node *mz,
				       struct mem_cgroup_tree_per_node *mctz)
569
{
570 571 572
	unsigned long flags;

	spin_lock_irqsave(&mctz->lock, flags);
573
	__mem_cgroup_remove_exceeded(mz, mctz);
574
	spin_unlock_irqrestore(&mctz->lock, flags);
575 576
}

577 578 579
static unsigned long soft_limit_excess(struct mem_cgroup *memcg)
{
	unsigned long nr_pages = page_counter_read(&memcg->memory);
580
	unsigned long soft_limit = READ_ONCE(memcg->soft_limit);
581 582 583 584 585 586 587
	unsigned long excess = 0;

	if (nr_pages > soft_limit)
		excess = nr_pages - soft_limit;

	return excess;
}
588 589 590

static void mem_cgroup_update_tree(struct mem_cgroup *memcg, struct page *page)
{
591
	unsigned long excess;
592 593
	struct mem_cgroup_per_node *mz;
	struct mem_cgroup_tree_per_node *mctz;
594

595
	mctz = soft_limit_tree_from_page(page);
596 597
	if (!mctz)
		return;
598 599 600 601 602
	/*
	 * Necessary to update all ancestors when hierarchy is used.
	 * because their event counter is not touched.
	 */
	for (; memcg; memcg = parent_mem_cgroup(memcg)) {
603
		mz = mem_cgroup_page_nodeinfo(memcg, page);
604
		excess = soft_limit_excess(memcg);
605 606 607 608 609
		/*
		 * We have to update the tree if mz is on RB-tree or
		 * mem is over its softlimit.
		 */
		if (excess || mz->on_tree) {
610 611 612
			unsigned long flags;

			spin_lock_irqsave(&mctz->lock, flags);
613 614
			/* if on-tree, remove it */
			if (mz->on_tree)
615
				__mem_cgroup_remove_exceeded(mz, mctz);
616 617 618 619
			/*
			 * Insert again. mz->usage_in_excess will be updated.
			 * If excess is 0, no tree ops.
			 */
620
			__mem_cgroup_insert_exceeded(mz, mctz, excess);
621
			spin_unlock_irqrestore(&mctz->lock, flags);
622 623 624 625 626 627
		}
	}
}

static void mem_cgroup_remove_from_trees(struct mem_cgroup *memcg)
{
628 629 630
	struct mem_cgroup_tree_per_node *mctz;
	struct mem_cgroup_per_node *mz;
	int nid;
631

632
	for_each_node(nid) {
633 634
		mz = mem_cgroup_nodeinfo(memcg, nid);
		mctz = soft_limit_tree_node(nid);
635 636
		if (mctz)
			mem_cgroup_remove_exceeded(mz, mctz);
637 638 639
	}
}

640 641
static struct mem_cgroup_per_node *
__mem_cgroup_largest_soft_limit_node(struct mem_cgroup_tree_per_node *mctz)
642
{
643
	struct mem_cgroup_per_node *mz;
644 645 646

retry:
	mz = NULL;
647
	if (!mctz->rb_rightmost)
648 649
		goto done;		/* Nothing to reclaim from */

650 651
	mz = rb_entry(mctz->rb_rightmost,
		      struct mem_cgroup_per_node, tree_node);
652 653 654 655 656
	/*
	 * Remove the node now but someone else can add it back,
	 * we will to add it back at the end of reclaim to its correct
	 * position in the tree.
	 */
657
	__mem_cgroup_remove_exceeded(mz, mctz);
658
	if (!soft_limit_excess(mz->memcg) ||
659
	    !css_tryget_online(&mz->memcg->css))
660 661 662 663 664
		goto retry;
done:
	return mz;
}

665 666
static struct mem_cgroup_per_node *
mem_cgroup_largest_soft_limit_node(struct mem_cgroup_tree_per_node *mctz)
667
{
668
	struct mem_cgroup_per_node *mz;
669

670
	spin_lock_irq(&mctz->lock);
671
	mz = __mem_cgroup_largest_soft_limit_node(mctz);
672
	spin_unlock_irq(&mctz->lock);
673 674 675
	return mz;
}

676 677 678 679 680 681 682 683 684 685 686 687 688 689 690
/**
 * __mod_memcg_state - update cgroup memory statistics
 * @memcg: the memory cgroup
 * @idx: the stat item - can be enum memcg_stat_item or enum node_stat_item
 * @val: delta to add to the counter, can be negative
 */
void __mod_memcg_state(struct mem_cgroup *memcg, int idx, int val)
{
	long x;

	if (mem_cgroup_disabled())
		return;

	x = val + __this_cpu_read(memcg->vmstats_percpu->stat[idx]);
	if (unlikely(abs(x) > MEMCG_CHARGE_BATCH)) {
691 692
		struct mem_cgroup *mi;

693 694 695 696 697
		/*
		 * Batch local counters to keep them in sync with
		 * the hierarchical ones.
		 */
		__this_cpu_add(memcg->vmstats_local->stat[idx], x);
698 699
		for (mi = memcg; mi; mi = parent_mem_cgroup(mi))
			atomic_long_add(x, &mi->vmstats[idx]);
700 701 702 703 704
		x = 0;
	}
	__this_cpu_write(memcg->vmstats_percpu->stat[idx], x);
}

705 706 707 708 709 710 711 712 713 714 715
static struct mem_cgroup_per_node *
parent_nodeinfo(struct mem_cgroup_per_node *pn, int nid)
{
	struct mem_cgroup *parent;

	parent = parent_mem_cgroup(pn->memcg);
	if (!parent)
		return NULL;
	return mem_cgroup_nodeinfo(parent, nid);
}

716 717 718 719 720 721 722 723 724 725 726 727 728
/**
 * __mod_lruvec_state - update lruvec memory statistics
 * @lruvec: the lruvec
 * @idx: the stat item
 * @val: delta to add to the counter, can be negative
 *
 * The lruvec is the intersection of the NUMA node and a cgroup. This
 * function updates the all three counters that are affected by a
 * change of state at this level: per-node, per-cgroup, per-lruvec.
 */
void __mod_lruvec_state(struct lruvec *lruvec, enum node_stat_item idx,
			int val)
{
729
	pg_data_t *pgdat = lruvec_pgdat(lruvec);
730
	struct mem_cgroup_per_node *pn;
731
	struct mem_cgroup *memcg;
732 733 734
	long x;

	/* Update node */
735
	__mod_node_page_state(pgdat, idx, val);
736 737 738 739 740

	if (mem_cgroup_disabled())
		return;

	pn = container_of(lruvec, struct mem_cgroup_per_node, lruvec);
741
	memcg = pn->memcg;
742 743

	/* Update memcg */
744
	__mod_memcg_state(memcg, idx, val);
745

746 747 748
	/* Update lruvec */
	__this_cpu_add(pn->lruvec_stat_local->count[idx], val);

749 750
	x = val + __this_cpu_read(pn->lruvec_stat_cpu->count[idx]);
	if (unlikely(abs(x) > MEMCG_CHARGE_BATCH)) {
751 752 753 754
		struct mem_cgroup_per_node *pi;

		for (pi = pn; pi; pi = parent_nodeinfo(pi, pgdat->node_id))
			atomic_long_add(x, &pi->lruvec_stat[idx]);
755 756 757 758 759
		x = 0;
	}
	__this_cpu_write(pn->lruvec_stat_cpu->count[idx], x);
}

760 761 762 763 764 765 766 767 768 769 770 771 772 773
void __mod_lruvec_slab_state(void *p, enum node_stat_item idx, int val)
{
	struct page *page = virt_to_head_page(p);
	pg_data_t *pgdat = page_pgdat(page);
	struct mem_cgroup *memcg;
	struct lruvec *lruvec;

	rcu_read_lock();
	memcg = memcg_from_slab_page(page);

	/* Untracked pages have no memcg, no lruvec. Update only the node */
	if (!memcg || memcg == root_mem_cgroup) {
		__mod_node_page_state(pgdat, idx, val);
	} else {
774
		lruvec = mem_cgroup_lruvec(memcg, pgdat);
775 776 777 778 779
		__mod_lruvec_state(lruvec, idx, val);
	}
	rcu_read_unlock();
}

780 781 782 783 784 785 786 787 788 789 790
void mod_memcg_obj_state(void *p, int idx, int val)
{
	struct mem_cgroup *memcg;

	rcu_read_lock();
	memcg = mem_cgroup_from_obj(p);
	if (memcg)
		mod_memcg_state(memcg, idx, val);
	rcu_read_unlock();
}

791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806
/**
 * __count_memcg_events - account VM events in a cgroup
 * @memcg: the memory cgroup
 * @idx: the event item
 * @count: the number of events that occured
 */
void __count_memcg_events(struct mem_cgroup *memcg, enum vm_event_item idx,
			  unsigned long count)
{
	unsigned long x;

	if (mem_cgroup_disabled())
		return;

	x = count + __this_cpu_read(memcg->vmstats_percpu->events[idx]);
	if (unlikely(x > MEMCG_CHARGE_BATCH)) {
807 808
		struct mem_cgroup *mi;

809 810 811 812 813
		/*
		 * Batch local counters to keep them in sync with
		 * the hierarchical ones.
		 */
		__this_cpu_add(memcg->vmstats_local->events[idx], x);
814 815
		for (mi = memcg; mi; mi = parent_mem_cgroup(mi))
			atomic_long_add(x, &mi->vmevents[idx]);
816 817 818 819 820
		x = 0;
	}
	__this_cpu_write(memcg->vmstats_percpu->events[idx], x);
}

821
static unsigned long memcg_events(struct mem_cgroup *memcg, int event)
822
{
823
	return atomic_long_read(&memcg->vmevents[event]);
824 825
}

826 827
static unsigned long memcg_events_local(struct mem_cgroup *memcg, int event)
{
828 829 830 831 832 833
	long x = 0;
	int cpu;

	for_each_possible_cpu(cpu)
		x += per_cpu(memcg->vmstats_local->events[event], cpu);
	return x;
834 835
}

836
static void mem_cgroup_charge_statistics(struct mem_cgroup *memcg,
837
					 struct page *page,
838
					 bool compound, int nr_pages)
839
{
840 841 842 843
	/*
	 * Here, RSS means 'mapped anon' and anon's SwapCache. Shmem/tmpfs is
	 * counted as CACHE even if it's on ANON LRU.
	 */
844
	if (PageAnon(page))
845
		__mod_memcg_state(memcg, MEMCG_RSS, nr_pages);
846
	else {
847
		__mod_memcg_state(memcg, MEMCG_CACHE, nr_pages);
848
		if (PageSwapBacked(page))
849
			__mod_memcg_state(memcg, NR_SHMEM, nr_pages);
850
	}
851

852 853
	if (compound) {
		VM_BUG_ON_PAGE(!PageTransHuge(page), page);
854
		__mod_memcg_state(memcg, MEMCG_RSS_HUGE, nr_pages);
855
	}
856

857 858
	/* pagein of a big page is an event. So, ignore page size */
	if (nr_pages > 0)
859
		__count_memcg_events(memcg, PGPGIN, 1);
860
	else {
861
		__count_memcg_events(memcg, PGPGOUT, 1);
862 863
		nr_pages = -nr_pages; /* for event */
	}
864

865
	__this_cpu_add(memcg->vmstats_percpu->nr_page_events, nr_pages);
866 867
}

868 869
static bool mem_cgroup_event_ratelimit(struct mem_cgroup *memcg,
				       enum mem_cgroup_events_target target)
870 871 872
{
	unsigned long val, next;

873 874
	val = __this_cpu_read(memcg->vmstats_percpu->nr_page_events);
	next = __this_cpu_read(memcg->vmstats_percpu->targets[target]);
875
	/* from time_after() in jiffies.h */
876
	if ((long)(next - val) < 0) {
877 878 879 880
		switch (target) {
		case MEM_CGROUP_TARGET_THRESH:
			next = val + THRESHOLDS_EVENTS_TARGET;
			break;
881 882 883
		case MEM_CGROUP_TARGET_SOFTLIMIT:
			next = val + SOFTLIMIT_EVENTS_TARGET;
			break;
884 885 886
		default:
			break;
		}
887
		__this_cpu_write(memcg->vmstats_percpu->targets[target], next);
888
		return true;
889
	}
890
	return false;
891 892 893 894 895 896
}

/*
 * Check events in order.
 *
 */
897
static void memcg_check_events(struct mem_cgroup *memcg, struct page *page)
898 899
{
	/* threshold event is triggered in finer grain than soft limit */
900 901
	if (unlikely(mem_cgroup_event_ratelimit(memcg,
						MEM_CGROUP_TARGET_THRESH))) {
902
		bool do_softlimit;
903

904 905
		do_softlimit = mem_cgroup_event_ratelimit(memcg,
						MEM_CGROUP_TARGET_SOFTLIMIT);
906
		mem_cgroup_threshold(memcg);
907 908
		if (unlikely(do_softlimit))
			mem_cgroup_update_tree(memcg, page);
909
	}
910 911
}

912
struct mem_cgroup *mem_cgroup_from_task(struct task_struct *p)
913
{
914 915 916 917 918 919 920 921
	/*
	 * mm_update_next_owner() may clear mm->owner to NULL
	 * if it races with swapoff, page migration, etc.
	 * So this can be called with p == NULL.
	 */
	if (unlikely(!p))
		return NULL;

922
	return mem_cgroup_from_css(task_css(p, memory_cgrp_id));
923
}
Michal Hocko's avatar
Michal Hocko committed
924
EXPORT_SYMBOL(mem_cgroup_from_task);
925

926 927 928 929 930 931 932 933 934
/**
 * get_mem_cgroup_from_mm: Obtain a reference on given mm_struct's memcg.
 * @mm: mm from which memcg should be extracted. It can be NULL.
 *
 * Obtain a reference on mm->memcg and returns it if successful. Otherwise
 * root_mem_cgroup is returned. However if mem_cgroup is disabled, NULL is
 * returned.
 */
struct mem_cgroup *get_mem_cgroup_from_mm(struct mm_struct *mm)
935
{
936 937 938 939
	struct mem_cgroup *memcg;

	if (mem_cgroup_disabled())
		return NULL;
940

941 942
	rcu_read_lock();
	do {
943 944 945 946 947 948
		/*
		 * Page cache insertions can happen withou an
		 * actual mm context, e.g. during disk probing
		 * on boot, loopback IO, acct() writes etc.
		 */
		if (unlikely(!mm))
949
			memcg = root_mem_cgroup;
950 951 952 953 954
		else {
			memcg = mem_cgroup_from_task(rcu_dereference(mm->owner));
			if (unlikely(!memcg))
				memcg = root_mem_cgroup;
		}
955
	} while (!css_tryget(&memcg->css));
956
	rcu_read_unlock();
957
	return memcg;
958
}
959 960
EXPORT_SYMBOL(get_mem_cgroup_from_mm);

961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982
/**
 * get_mem_cgroup_from_page: Obtain a reference on given page's memcg.
 * @page: page from which memcg should be extracted.
 *
 * Obtain a reference on page->memcg and returns it if successful. Otherwise
 * root_mem_cgroup is returned.
 */
struct mem_cgroup *get_mem_cgroup_from_page(struct page *page)
{
	struct mem_cgroup *memcg = page->mem_cgroup;

	if (mem_cgroup_disabled())
		return NULL;

	rcu_read_lock();
	if (!memcg || !css_tryget_online(&memcg->css))
		memcg = root_mem_cgroup;
	rcu_read_unlock();
	return memcg;
}
EXPORT_SYMBOL(get_mem_cgroup_from_page);

983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998
/**
 * If current->active_memcg is non-NULL, do not fallback to current->mm->memcg.
 */
static __always_inline struct mem_cgroup *get_mem_cgroup_from_current(void)
{
	if (unlikely(current->active_memcg)) {
		struct mem_cgroup *memcg = root_mem_cgroup;

		rcu_read_lock();
		if (css_tryget_online(&current->active_memcg->css))
			memcg = current->active_memcg;
		rcu_read_unlock();
		return memcg;
	}
	return get_mem_cgroup_from_mm(current->mm);
}
999

1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012
/**
 * mem_cgroup_iter - iterate over memory cgroup hierarchy
 * @root: hierarchy root
 * @prev: previously returned memcg, NULL on first invocation
 * @reclaim: cookie for shared reclaim walks, NULL for full walks
 *
 * Returns references to children of the hierarchy below @root, or
 * @root itself, or %NULL after a full round-trip.
 *
 * Caller must pass the return value in @prev on subsequent
 * invocations for reference counting, or use mem_cgroup_iter_break()
 * to cancel a hierarchy walk before the round-trip is complete.
 *
1013
 * Reclaimers can specify a node and a priority level in @reclaim to
1014
 * divide up the memcgs in the hierarchy among all concurrent