memcontrol.c 131.3 KB
Newer Older
B
Balbir Singh 已提交
1 2 3 4 5
/* memcontrol.c - 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>
 *
9 10 11 12
 * Memory thresholds
 * Copyright (C) 2009 Nokia Corporation
 * Author: Kirill A. Shutemov
 *
B
Balbir Singh 已提交
13 14 15 16 17 18 19 20 21 22 23 24 25 26
 * 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.
 */

#include <linux/res_counter.h>
#include <linux/memcontrol.h>
#include <linux/cgroup.h>
27
#include <linux/mm.h>
28
#include <linux/hugetlb.h>
K
KAMEZAWA Hiroyuki 已提交
29
#include <linux/pagemap.h>
30
#include <linux/smp.h>
31
#include <linux/page-flags.h>
32
#include <linux/backing-dev.h>
33 34
#include <linux/bit_spinlock.h>
#include <linux/rcupdate.h>
35
#include <linux/limits.h>
36
#include <linux/mutex.h>
37
#include <linux/rbtree.h>
38
#include <linux/slab.h>
39
#include <linux/swap.h>
40
#include <linux/swapops.h>
41
#include <linux/spinlock.h>
42 43
#include <linux/eventfd.h>
#include <linux/sort.h>
44
#include <linux/fs.h>
45
#include <linux/seq_file.h>
46
#include <linux/vmalloc.h>
47
#include <linux/mm_inline.h>
48
#include <linux/page_cgroup.h>
49
#include <linux/cpu.h>
50
#include <linux/oom.h>
K
KAMEZAWA Hiroyuki 已提交
51
#include "internal.h"
B
Balbir Singh 已提交
52

53 54
#include <asm/uaccess.h>

55 56
#include <trace/events/vmscan.h>

57 58
struct cgroup_subsys mem_cgroup_subsys __read_mostly;
#define MEM_CGROUP_RECLAIM_RETRIES	5
59
struct mem_cgroup *root_mem_cgroup __read_mostly;
B
Balbir Singh 已提交
60

61
#ifdef CONFIG_CGROUP_MEM_RES_CTLR_SWAP
L
Li Zefan 已提交
62
/* Turned on only when memory cgroup is enabled && really_do_swap_account = 1 */
63
int do_swap_account __read_mostly;
64 65 66 67 68 69 70 71

/* for remember boot option*/
#ifdef CONFIG_CGROUP_MEM_RES_CTLR_SWAP_ENABLED
static int really_do_swap_account __initdata = 1;
#else
static int really_do_swap_account __initdata = 0;
#endif

72 73 74 75
#else
#define do_swap_account		(0)
#endif

76 77 78 79 80 81 82 83 84
/*
 * Per memcg event counter is incremented at every pagein/pageout. This counter
 * is used for trigger some periodic events. This is straightforward and better
 * than using jiffies etc. to handle periodic memcg event.
 *
 * These values will be used as !((event) & ((1 <<(thresh)) - 1))
 */
#define THRESHOLDS_EVENTS_THRESH (7) /* once in 128 */
#define SOFTLIMIT_EVENTS_THRESH (10) /* once in 1024 */
85

86 87 88 89 90 91 92 93
/*
 * Statistics for memory cgroup.
 */
enum mem_cgroup_stat_index {
	/*
	 * For MEM_CONTAINER_TYPE_ALL, usage = pagecache + rss.
	 */
	MEM_CGROUP_STAT_CACHE, 	   /* # of pages charged as cache */
94
	MEM_CGROUP_STAT_RSS,	   /* # of pages charged as anon rss */
95
	MEM_CGROUP_STAT_FILE_MAPPED,  /* # of pages charged as file rss */
96 97
	MEM_CGROUP_STAT_PGPGIN_COUNT,	/* # of pages paged in */
	MEM_CGROUP_STAT_PGPGOUT_COUNT,	/* # of pages paged out */
98
	MEM_CGROUP_STAT_SWAPOUT, /* # of pages, swapped out */
99 100 101
	MEM_CGROUP_STAT_DATA, /* end of data requires synchronization */
	/* incremented at every  pagein/pageout */
	MEM_CGROUP_EVENTS = MEM_CGROUP_STAT_DATA,
102
	MEM_CGROUP_ON_MOVE,	/* someone is moving account between groups */
103 104 105 106 107 108 109 110

	MEM_CGROUP_STAT_NSTATS,
};

struct mem_cgroup_stat_cpu {
	s64 count[MEM_CGROUP_STAT_NSTATS];
};

111 112 113 114
/*
 * per-zone information in memory controller.
 */
struct mem_cgroup_per_zone {
115 116 117
	/*
	 * spin_lock to protect the per cgroup LRU
	 */
118 119
	struct list_head	lists[NR_LRU_LISTS];
	unsigned long		count[NR_LRU_LISTS];
K
KOSAKI Motohiro 已提交
120 121

	struct zone_reclaim_stat reclaim_stat;
122 123 124 125
	struct rb_node		tree_node;	/* RB tree node */
	unsigned long long	usage_in_excess;/* Set to the value by which */
						/* the soft limit is exceeded*/
	bool			on_tree;
126 127
	struct mem_cgroup	*mem;		/* Back pointer, we cannot */
						/* use container_of	   */
128 129 130 131 132 133 134 135 136 137 138 139
};
/* Macro for accessing counter */
#define MEM_CGROUP_ZSTAT(mz, idx)	((mz)->count[(idx)])

struct mem_cgroup_per_node {
	struct mem_cgroup_per_zone zoneinfo[MAX_NR_ZONES];
};

struct mem_cgroup_lru_info {
	struct mem_cgroup_per_node *nodeinfo[MAX_NUMNODES];
};

140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159
/*
 * Cgroups above their limits are maintained in a RB-Tree, independent of
 * their hierarchy representation
 */

struct mem_cgroup_tree_per_zone {
	struct rb_root rb_root;
	spinlock_t lock;
};

struct mem_cgroup_tree_per_node {
	struct mem_cgroup_tree_per_zone rb_tree_per_zone[MAX_NR_ZONES];
};

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;

160 161 162 163 164
struct mem_cgroup_threshold {
	struct eventfd_ctx *eventfd;
	u64 threshold;
};

K
KAMEZAWA Hiroyuki 已提交
165
/* For threshold */
166 167
struct mem_cgroup_threshold_ary {
	/* An array index points to threshold just below usage. */
168
	int current_threshold;
169 170 171 172 173
	/* Size of entries[] */
	unsigned int size;
	/* Array of thresholds */
	struct mem_cgroup_threshold entries[0];
};
174 175 176 177 178 179 180 181 182 183 184 185

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;
};

K
KAMEZAWA Hiroyuki 已提交
186 187 188 189 190
/* for OOM */
struct mem_cgroup_eventfd_list {
	struct list_head list;
	struct eventfd_ctx *eventfd;
};
191 192

static void mem_cgroup_threshold(struct mem_cgroup *mem);
K
KAMEZAWA Hiroyuki 已提交
193
static void mem_cgroup_oom_notify(struct mem_cgroup *mem);
194

B
Balbir Singh 已提交
195 196 197 198 199 200 201
/*
 * 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.
 *
 * TODO: Add a water mark for the memory controller. Reclaim will begin when
202 203 204
 * we hit the water mark. May be even add a low water mark, such that
 * no reclaim occurs from a cgroup at it's low water mark, this is
 * a feature that will be implemented much later in the future.
B
Balbir Singh 已提交
205 206 207 208 209 210 211
 */
struct mem_cgroup {
	struct cgroup_subsys_state css;
	/*
	 * the counter to account for memory usage
	 */
	struct res_counter res;
212 213 214 215
	/*
	 * the counter to account for mem+swap usage.
	 */
	struct res_counter memsw;
216 217 218 219
	/*
	 * Per cgroup active and inactive list, similar to the
	 * per zone LRU lists.
	 */
220
	struct mem_cgroup_lru_info info;
221
	/*
222
	 * While reclaiming in a hierarchy, we cache the last child we
K
KAMEZAWA Hiroyuki 已提交
223
	 * reclaimed from.
224
	 */
K
KAMEZAWA Hiroyuki 已提交
225
	int last_scanned_child;
226 227 228 229
	/*
	 * Should the accounting and control be hierarchical, per subtree?
	 */
	bool use_hierarchy;
K
KAMEZAWA Hiroyuki 已提交
230
	atomic_t	oom_lock;
231
	atomic_t	refcnt;
232

K
KOSAKI Motohiro 已提交
233
	unsigned int	swappiness;
234 235
	/* OOM-Killer disable */
	int		oom_kill_disable;
K
KOSAKI Motohiro 已提交
236

237 238 239
	/* set when res.limit == memsw.limit */
	bool		memsw_is_minimum;

240 241 242 243
	/* protect arrays of thresholds */
	struct mutex thresholds_lock;

	/* thresholds for memory usage. RCU-protected */
244
	struct mem_cgroup_thresholds thresholds;
245

246
	/* thresholds for mem+swap usage. RCU-protected */
247
	struct mem_cgroup_thresholds memsw_thresholds;
248

K
KAMEZAWA Hiroyuki 已提交
249 250 251
	/* For oom notifier event fd */
	struct list_head oom_notify;

252 253 254 255 256
	/*
	 * 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;
257
	/*
258
	 * percpu counter.
259
	 */
260
	struct mem_cgroup_stat_cpu *stat;
261 262 263 264 265 266
	/*
	 * used when a cpu is offlined or other synchronizations
	 * See mem_cgroup_read_stat().
	 */
	struct mem_cgroup_stat_cpu nocpu_base;
	spinlock_t pcp_counter_lock;
B
Balbir Singh 已提交
267 268
};

269 270 271 272 273 274
/* Stuffs for move charges at task migration. */
/*
 * Types of charges to be moved. "move_charge_at_immitgrate" is treated as a
 * left-shifted bitmap of these types.
 */
enum move_type {
275
	MOVE_CHARGE_TYPE_ANON,	/* private anonymous page and swap of it */
276
	MOVE_CHARGE_TYPE_FILE,	/* file page(including tmpfs) and swap of it */
277 278 279
	NR_MOVE_TYPE,
};

280 281
/* "mc" and its members are protected by cgroup_mutex */
static struct move_charge_struct {
282
	spinlock_t	  lock; /* for from, to */
283 284 285
	struct mem_cgroup *from;
	struct mem_cgroup *to;
	unsigned long precharge;
286
	unsigned long moved_charge;
287
	unsigned long moved_swap;
288 289 290
	struct task_struct *moving_task;	/* a task moving charges */
	wait_queue_head_t waitq;		/* a waitq for other context */
} mc = {
291
	.lock = __SPIN_LOCK_UNLOCKED(mc.lock),
292 293
	.waitq = __WAIT_QUEUE_HEAD_INITIALIZER(mc.waitq),
};
294

D
Daisuke Nishimura 已提交
295 296 297 298 299 300
static bool move_anon(void)
{
	return test_bit(MOVE_CHARGE_TYPE_ANON,
					&mc.to->move_charge_at_immigrate);
}

301 302 303 304 305 306
static bool move_file(void)
{
	return test_bit(MOVE_CHARGE_TYPE_FILE,
					&mc.to->move_charge_at_immigrate);
}

307 308 309 310 311 312 313
/*
 * Maximum loops in mem_cgroup_hierarchical_reclaim(), used for soft
 * limit reclaim to prevent infinite loops, if they ever occur.
 */
#define	MEM_CGROUP_MAX_RECLAIM_LOOPS		(100)
#define	MEM_CGROUP_MAX_SOFT_LIMIT_RECLAIM_LOOPS	(2)

314 315 316
enum charge_type {
	MEM_CGROUP_CHARGE_TYPE_CACHE = 0,
	MEM_CGROUP_CHARGE_TYPE_MAPPED,
317
	MEM_CGROUP_CHARGE_TYPE_SHMEM,	/* used by page migration of shmem */
318
	MEM_CGROUP_CHARGE_TYPE_FORCE,	/* used by force_empty */
K
KAMEZAWA Hiroyuki 已提交
319
	MEM_CGROUP_CHARGE_TYPE_SWAPOUT,	/* for accounting swapcache */
K
KAMEZAWA Hiroyuki 已提交
320
	MEM_CGROUP_CHARGE_TYPE_DROP,	/* a page was unused swap cache */
321 322 323
	NR_CHARGE_TYPE,
};

324 325 326
/* for encoding cft->private value on file */
#define _MEM			(0)
#define _MEMSWAP		(1)
K
KAMEZAWA Hiroyuki 已提交
327
#define _OOM_TYPE		(2)
328 329 330
#define MEMFILE_PRIVATE(x, val)	(((x) << 16) | (val))
#define MEMFILE_TYPE(val)	(((val) >> 16) & 0xffff)
#define MEMFILE_ATTR(val)	((val) & 0xffff)
K
KAMEZAWA Hiroyuki 已提交
331 332
/* Used for OOM nofiier */
#define OOM_CONTROL		(0)
333

334 335 336 337 338 339 340
/*
 * Reclaim flags for mem_cgroup_hierarchical_reclaim
 */
#define MEM_CGROUP_RECLAIM_NOSWAP_BIT	0x0
#define MEM_CGROUP_RECLAIM_NOSWAP	(1 << MEM_CGROUP_RECLAIM_NOSWAP_BIT)
#define MEM_CGROUP_RECLAIM_SHRINK_BIT	0x1
#define MEM_CGROUP_RECLAIM_SHRINK	(1 << MEM_CGROUP_RECLAIM_SHRINK_BIT)
341 342
#define MEM_CGROUP_RECLAIM_SOFT_BIT	0x2
#define MEM_CGROUP_RECLAIM_SOFT		(1 << MEM_CGROUP_RECLAIM_SOFT_BIT)
343

344 345
static void mem_cgroup_get(struct mem_cgroup *mem);
static void mem_cgroup_put(struct mem_cgroup *mem);
346
static struct mem_cgroup *parent_mem_cgroup(struct mem_cgroup *mem);
347
static void drain_all_stock_async(void);
348

349 350 351 352 353 354
static struct mem_cgroup_per_zone *
mem_cgroup_zoneinfo(struct mem_cgroup *mem, int nid, int zid)
{
	return &mem->info.nodeinfo[nid]->zoneinfo[zid];
}

355 356 357 358 359
struct cgroup_subsys_state *mem_cgroup_css(struct mem_cgroup *mem)
{
	return &mem->css;
}

360
static struct mem_cgroup_per_zone *
361
page_cgroup_zoneinfo(struct mem_cgroup *mem, struct page *page)
362
{
363 364
	int nid = page_to_nid(page);
	int zid = page_zonenum(page);
365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384

	return mem_cgroup_zoneinfo(mem, nid, zid);
}

static struct mem_cgroup_tree_per_zone *
soft_limit_tree_node_zone(int nid, int zid)
{
	return &soft_limit_tree.rb_tree_per_node[nid]->rb_tree_per_zone[zid];
}

static struct mem_cgroup_tree_per_zone *
soft_limit_tree_from_page(struct page *page)
{
	int nid = page_to_nid(page);
	int zid = page_zonenum(page);

	return &soft_limit_tree.rb_tree_per_node[nid]->rb_tree_per_zone[zid];
}

static void
385
__mem_cgroup_insert_exceeded(struct mem_cgroup *mem,
386
				struct mem_cgroup_per_zone *mz,
387 388
				struct mem_cgroup_tree_per_zone *mctz,
				unsigned long long new_usage_in_excess)
389 390 391 392 393 394 395 396
{
	struct rb_node **p = &mctz->rb_root.rb_node;
	struct rb_node *parent = NULL;
	struct mem_cgroup_per_zone *mz_node;

	if (mz->on_tree)
		return;

397 398 399
	mz->usage_in_excess = new_usage_in_excess;
	if (!mz->usage_in_excess)
		return;
400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415
	while (*p) {
		parent = *p;
		mz_node = rb_entry(parent, struct mem_cgroup_per_zone,
					tree_node);
		if (mz->usage_in_excess < mz_node->usage_in_excess)
			p = &(*p)->rb_left;
		/*
		 * 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;
	}
	rb_link_node(&mz->tree_node, parent, p);
	rb_insert_color(&mz->tree_node, &mctz->rb_root);
	mz->on_tree = true;
416 417 418 419 420 421 422 423 424 425 426 427 428
}

static void
__mem_cgroup_remove_exceeded(struct mem_cgroup *mem,
				struct mem_cgroup_per_zone *mz,
				struct mem_cgroup_tree_per_zone *mctz)
{
	if (!mz->on_tree)
		return;
	rb_erase(&mz->tree_node, &mctz->rb_root);
	mz->on_tree = false;
}

429 430 431 432 433 434
static void
mem_cgroup_remove_exceeded(struct mem_cgroup *mem,
				struct mem_cgroup_per_zone *mz,
				struct mem_cgroup_tree_per_zone *mctz)
{
	spin_lock(&mctz->lock);
435
	__mem_cgroup_remove_exceeded(mem, mz, mctz);
436 437 438 439 440 441
	spin_unlock(&mctz->lock);
}


static void mem_cgroup_update_tree(struct mem_cgroup *mem, struct page *page)
{
442
	unsigned long long excess;
443 444
	struct mem_cgroup_per_zone *mz;
	struct mem_cgroup_tree_per_zone *mctz;
445 446
	int nid = page_to_nid(page);
	int zid = page_zonenum(page);
447 448 449
	mctz = soft_limit_tree_from_page(page);

	/*
450 451
	 * Necessary to update all ancestors when hierarchy is used.
	 * because their event counter is not touched.
452
	 */
453 454
	for (; mem; mem = parent_mem_cgroup(mem)) {
		mz = mem_cgroup_zoneinfo(mem, nid, zid);
455
		excess = res_counter_soft_limit_excess(&mem->res);
456 457 458 459
		/*
		 * We have to update the tree if mz is on RB-tree or
		 * mem is over its softlimit.
		 */
460
		if (excess || mz->on_tree) {
461 462 463 464 465
			spin_lock(&mctz->lock);
			/* if on-tree, remove it */
			if (mz->on_tree)
				__mem_cgroup_remove_exceeded(mem, mz, mctz);
			/*
466 467
			 * Insert again. mz->usage_in_excess will be updated.
			 * If excess is 0, no tree ops.
468
			 */
469
			__mem_cgroup_insert_exceeded(mem, mz, mctz, excess);
470 471
			spin_unlock(&mctz->lock);
		}
472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489
	}
}

static void mem_cgroup_remove_from_trees(struct mem_cgroup *mem)
{
	int node, zone;
	struct mem_cgroup_per_zone *mz;
	struct mem_cgroup_tree_per_zone *mctz;

	for_each_node_state(node, N_POSSIBLE) {
		for (zone = 0; zone < MAX_NR_ZONES; zone++) {
			mz = mem_cgroup_zoneinfo(mem, node, zone);
			mctz = soft_limit_tree_node_zone(node, zone);
			mem_cgroup_remove_exceeded(mem, mz, mctz);
		}
	}
}

490 491 492 493
static struct mem_cgroup_per_zone *
__mem_cgroup_largest_soft_limit_node(struct mem_cgroup_tree_per_zone *mctz)
{
	struct rb_node *rightmost = NULL;
494
	struct mem_cgroup_per_zone *mz;
495 496

retry:
497
	mz = NULL;
498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526
	rightmost = rb_last(&mctz->rb_root);
	if (!rightmost)
		goto done;		/* Nothing to reclaim from */

	mz = rb_entry(rightmost, struct mem_cgroup_per_zone, tree_node);
	/*
	 * 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.
	 */
	__mem_cgroup_remove_exceeded(mz->mem, mz, mctz);
	if (!res_counter_soft_limit_excess(&mz->mem->res) ||
		!css_tryget(&mz->mem->css))
		goto retry;
done:
	return mz;
}

static struct mem_cgroup_per_zone *
mem_cgroup_largest_soft_limit_node(struct mem_cgroup_tree_per_zone *mctz)
{
	struct mem_cgroup_per_zone *mz;

	spin_lock(&mctz->lock);
	mz = __mem_cgroup_largest_soft_limit_node(mctz);
	spin_unlock(&mctz->lock);
	return mz;
}

527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545
/*
 * Implementation Note: reading percpu statistics for memcg.
 *
 * Both of vmstat[] and percpu_counter has threshold and do periodic
 * synchronization to implement "quick" read. There are trade-off between
 * reading cost and precision of value. Then, we may have a chance to implement
 * a periodic synchronizion of counter in memcg's counter.
 *
 * But this _read() function is used for user interface now. The user accounts
 * memory usage by memory cgroup and he _always_ requires exact value because
 * he accounts memory. Even if we provide quick-and-fuzzy read, we always
 * have to visit all online cpus and make sum. So, for now, unnecessary
 * synchronization is not implemented. (just implemented for cpu hotplug)
 *
 * If there are kernel internal actions which can make use of some not-exact
 * value, and reading all cpu value can be performance bottleneck in some
 * common workload, threashold and synchonization as vmstat[] should be
 * implemented.
 */
546 547 548 549 550 551
static s64 mem_cgroup_read_stat(struct mem_cgroup *mem,
		enum mem_cgroup_stat_index idx)
{
	int cpu;
	s64 val = 0;

552 553
	get_online_cpus();
	for_each_online_cpu(cpu)
554
		val += per_cpu(mem->stat->count[idx], cpu);
555 556 557 558 559 560
#ifdef CONFIG_HOTPLUG_CPU
	spin_lock(&mem->pcp_counter_lock);
	val += mem->nocpu_base.count[idx];
	spin_unlock(&mem->pcp_counter_lock);
#endif
	put_online_cpus();
561 562 563 564 565 566 567 568 569 570 571 572
	return val;
}

static s64 mem_cgroup_local_usage(struct mem_cgroup *mem)
{
	s64 ret;

	ret = mem_cgroup_read_stat(mem, MEM_CGROUP_STAT_RSS);
	ret += mem_cgroup_read_stat(mem, MEM_CGROUP_STAT_CACHE);
	return ret;
}

573 574 575 576
static void mem_cgroup_swap_statistics(struct mem_cgroup *mem,
					 bool charge)
{
	int val = (charge) ? 1 : -1;
577
	this_cpu_add(mem->stat->count[MEM_CGROUP_STAT_SWAPOUT], val);
578 579
}

580
static void mem_cgroup_charge_statistics(struct mem_cgroup *mem,
581
					 bool file, int nr_pages)
582
{
583 584
	preempt_disable();

585 586
	if (file)
		__this_cpu_add(mem->stat->count[MEM_CGROUP_STAT_CACHE], nr_pages);
587
	else
588
		__this_cpu_add(mem->stat->count[MEM_CGROUP_STAT_RSS], nr_pages);
589

590 591
	/* pagein of a big page is an event. So, ignore page size */
	if (nr_pages > 0)
592
		__this_cpu_inc(mem->stat->count[MEM_CGROUP_STAT_PGPGIN_COUNT]);
593
	else {
594
		__this_cpu_inc(mem->stat->count[MEM_CGROUP_STAT_PGPGOUT_COUNT]);
595 596
		nr_pages = -nr_pages; /* for event */
	}
597 598

	__this_cpu_add(mem->stat->count[MEM_CGROUP_EVENTS], nr_pages);
599

600
	preempt_enable();
601 602
}

K
KAMEZAWA Hiroyuki 已提交
603
static unsigned long mem_cgroup_get_local_zonestat(struct mem_cgroup *mem,
604
					enum lru_list idx)
605 606 607 608 609 610 611 612 613 614 615
{
	int nid, zid;
	struct mem_cgroup_per_zone *mz;
	u64 total = 0;

	for_each_online_node(nid)
		for (zid = 0; zid < MAX_NR_ZONES; zid++) {
			mz = mem_cgroup_zoneinfo(mem, nid, zid);
			total += MEM_CGROUP_ZSTAT(mz, idx);
		}
	return total;
616 617
}

618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640
static bool __memcg_event_check(struct mem_cgroup *mem, int event_mask_shift)
{
	s64 val;

	val = this_cpu_read(mem->stat->count[MEM_CGROUP_EVENTS]);

	return !(val & ((1 << event_mask_shift) - 1));
}

/*
 * Check events in order.
 *
 */
static void memcg_check_events(struct mem_cgroup *mem, struct page *page)
{
	/* threshold event is triggered in finer grain than soft limit */
	if (unlikely(__memcg_event_check(mem, THRESHOLDS_EVENTS_THRESH))) {
		mem_cgroup_threshold(mem);
		if (unlikely(__memcg_event_check(mem, SOFTLIMIT_EVENTS_THRESH)))
			mem_cgroup_update_tree(mem, page);
	}
}

641
static struct mem_cgroup *mem_cgroup_from_cont(struct cgroup *cont)
B
Balbir Singh 已提交
642 643 644 645 646 647
{
	return container_of(cgroup_subsys_state(cont,
				mem_cgroup_subsys_id), struct mem_cgroup,
				css);
}

648
struct mem_cgroup *mem_cgroup_from_task(struct task_struct *p)
649
{
650 651 652 653 654 655 656 657
	/*
	 * 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;

658 659 660 661
	return container_of(task_subsys_state(p, mem_cgroup_subsys_id),
				struct mem_cgroup, css);
}

662 663 664
static struct mem_cgroup *try_get_mem_cgroup_from_mm(struct mm_struct *mm)
{
	struct mem_cgroup *mem = NULL;
665 666 667

	if (!mm)
		return NULL;
668 669 670 671 672 673 674 675 676 677 678 679 680 681 682
	/*
	 * Because we have no locks, mm->owner's may be being moved to other
	 * cgroup. We use css_tryget() here even if this looks
	 * pessimistic (rather than adding locks here).
	 */
	rcu_read_lock();
	do {
		mem = mem_cgroup_from_task(rcu_dereference(mm->owner));
		if (unlikely(!mem))
			break;
	} while (!css_tryget(&mem->css));
	rcu_read_unlock();
	return mem;
}

K
KAMEZAWA Hiroyuki 已提交
683 684
/* The caller has to guarantee "mem" exists before calling this */
static struct mem_cgroup *mem_cgroup_start_loop(struct mem_cgroup *mem)
K
KAMEZAWA Hiroyuki 已提交
685
{
686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707
	struct cgroup_subsys_state *css;
	int found;

	if (!mem) /* ROOT cgroup has the smallest ID */
		return root_mem_cgroup; /*css_put/get against root is ignored*/
	if (!mem->use_hierarchy) {
		if (css_tryget(&mem->css))
			return mem;
		return NULL;
	}
	rcu_read_lock();
	/*
	 * searching a memory cgroup which has the smallest ID under given
	 * ROOT cgroup. (ID >= 1)
	 */
	css = css_get_next(&mem_cgroup_subsys, 1, &mem->css, &found);
	if (css && css_tryget(css))
		mem = container_of(css, struct mem_cgroup, css);
	else
		mem = NULL;
	rcu_read_unlock();
	return mem;
K
KAMEZAWA Hiroyuki 已提交
708 709 710 711 712 713 714 715 716
}

static struct mem_cgroup *mem_cgroup_get_next(struct mem_cgroup *iter,
					struct mem_cgroup *root,
					bool cond)
{
	int nextid = css_id(&iter->css) + 1;
	int found;
	int hierarchy_used;
K
KAMEZAWA Hiroyuki 已提交
717 718
	struct cgroup_subsys_state *css;

K
KAMEZAWA Hiroyuki 已提交
719
	hierarchy_used = iter->use_hierarchy;
K
KAMEZAWA Hiroyuki 已提交
720

K
KAMEZAWA Hiroyuki 已提交
721
	css_put(&iter->css);
722 723
	/* If no ROOT, walk all, ignore hierarchy */
	if (!cond || (root && !hierarchy_used))
K
KAMEZAWA Hiroyuki 已提交
724
		return NULL;
K
KAMEZAWA Hiroyuki 已提交
725

726 727 728
	if (!root)
		root = root_mem_cgroup;

K
KAMEZAWA Hiroyuki 已提交
729 730
	do {
		iter = NULL;
K
KAMEZAWA Hiroyuki 已提交
731
		rcu_read_lock();
K
KAMEZAWA Hiroyuki 已提交
732 733 734

		css = css_get_next(&mem_cgroup_subsys, nextid,
				&root->css, &found);
K
KAMEZAWA Hiroyuki 已提交
735
		if (css && css_tryget(css))
K
KAMEZAWA Hiroyuki 已提交
736
			iter = container_of(css, struct mem_cgroup, css);
K
KAMEZAWA Hiroyuki 已提交
737
		rcu_read_unlock();
K
KAMEZAWA Hiroyuki 已提交
738
		/* If css is NULL, no more cgroups will be found */
K
KAMEZAWA Hiroyuki 已提交
739
		nextid = found + 1;
K
KAMEZAWA Hiroyuki 已提交
740
	} while (css && !iter);
K
KAMEZAWA Hiroyuki 已提交
741

K
KAMEZAWA Hiroyuki 已提交
742
	return iter;
K
KAMEZAWA Hiroyuki 已提交
743
}
K
KAMEZAWA Hiroyuki 已提交
744 745 746 747 748 749 750 751 752 753 754 755 756
/*
 * for_eacn_mem_cgroup_tree() for visiting all cgroup under tree. Please
 * be careful that "break" loop is not allowed. We have reference count.
 * Instead of that modify "cond" to be false and "continue" to exit the loop.
 */
#define for_each_mem_cgroup_tree_cond(iter, root, cond)	\
	for (iter = mem_cgroup_start_loop(root);\
	     iter != NULL;\
	     iter = mem_cgroup_get_next(iter, root, cond))

#define for_each_mem_cgroup_tree(iter, root) \
	for_each_mem_cgroup_tree_cond(iter, root, true)

757 758 759
#define for_each_mem_cgroup_all(iter) \
	for_each_mem_cgroup_tree_cond(iter, NULL, true)

K
KAMEZAWA Hiroyuki 已提交
760

761 762 763 764 765
static inline bool mem_cgroup_is_root(struct mem_cgroup *mem)
{
	return (mem == root_mem_cgroup);
}

K
KAMEZAWA Hiroyuki 已提交
766 767 768 769 770 771 772 773 774 775 776 777 778
/*
 * Following LRU functions are allowed to be used without PCG_LOCK.
 * Operations are called by routine of global LRU independently from memcg.
 * What we have to take care of here is validness of pc->mem_cgroup.
 *
 * Changes to pc->mem_cgroup happens when
 * 1. charge
 * 2. moving account
 * In typical case, "charge" is done before add-to-lru. Exception is SwapCache.
 * It is added to LRU before charge.
 * If PCG_USED bit is not set, page_cgroup is not added to this private LRU.
 * When moving account, the page is not on LRU. It's isolated.
 */
779

K
KAMEZAWA Hiroyuki 已提交
780 781 782 783
void mem_cgroup_del_lru_list(struct page *page, enum lru_list lru)
{
	struct page_cgroup *pc;
	struct mem_cgroup_per_zone *mz;
784

785
	if (mem_cgroup_disabled())
K
KAMEZAWA Hiroyuki 已提交
786 787 788
		return;
	pc = lookup_page_cgroup(page);
	/* can happen while we handle swapcache. */
789
	if (!TestClearPageCgroupAcctLRU(pc))
K
KAMEZAWA Hiroyuki 已提交
790
		return;
791
	VM_BUG_ON(!pc->mem_cgroup);
792 793 794 795
	/*
	 * We don't check PCG_USED bit. It's cleared when the "page" is finally
	 * removed from global LRU.
	 */
796
	mz = page_cgroup_zoneinfo(pc->mem_cgroup, page);
797 798
	/* huge page split is done under lru_lock. so, we have no races. */
	MEM_CGROUP_ZSTAT(mz, lru) -= 1 << compound_order(page);
799 800 801
	if (mem_cgroup_is_root(pc->mem_cgroup))
		return;
	VM_BUG_ON(list_empty(&pc->lru));
K
KAMEZAWA Hiroyuki 已提交
802
	list_del_init(&pc->lru);
803 804
}

K
KAMEZAWA Hiroyuki 已提交
805
void mem_cgroup_del_lru(struct page *page)
806
{
K
KAMEZAWA Hiroyuki 已提交
807 808
	mem_cgroup_del_lru_list(page, page_lru(page));
}
809

810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831
/*
 * Writeback is about to end against a page which has been marked for immediate
 * reclaim.  If it still appears to be reclaimable, move it to the tail of the
 * inactive list.
 */
void mem_cgroup_rotate_reclaimable_page(struct page *page)
{
	struct mem_cgroup_per_zone *mz;
	struct page_cgroup *pc;
	enum lru_list lru = page_lru(page);

	if (mem_cgroup_disabled())
		return;

	pc = lookup_page_cgroup(page);
	/* unused or root page is not rotated. */
	if (!PageCgroupUsed(pc))
		return;
	/* Ensure pc->mem_cgroup is visible after reading PCG_USED. */
	smp_rmb();
	if (mem_cgroup_is_root(pc->mem_cgroup))
		return;
832
	mz = page_cgroup_zoneinfo(pc->mem_cgroup, page);
833 834 835
	list_move_tail(&pc->lru, &mz->lists[lru]);
}

K
KAMEZAWA Hiroyuki 已提交
836 837 838 839
void mem_cgroup_rotate_lru_list(struct page *page, enum lru_list lru)
{
	struct mem_cgroup_per_zone *mz;
	struct page_cgroup *pc;
840

841
	if (mem_cgroup_disabled())
K
KAMEZAWA Hiroyuki 已提交
842
		return;
843

K
KAMEZAWA Hiroyuki 已提交
844
	pc = lookup_page_cgroup(page);
845
	/* unused or root page is not rotated. */
846 847 848 849 850
	if (!PageCgroupUsed(pc))
		return;
	/* Ensure pc->mem_cgroup is visible after reading PCG_USED. */
	smp_rmb();
	if (mem_cgroup_is_root(pc->mem_cgroup))
K
KAMEZAWA Hiroyuki 已提交
851
		return;
852
	mz = page_cgroup_zoneinfo(pc->mem_cgroup, page);
K
KAMEZAWA Hiroyuki 已提交
853
	list_move(&pc->lru, &mz->lists[lru]);
854 855
}

K
KAMEZAWA Hiroyuki 已提交
856
void mem_cgroup_add_lru_list(struct page *page, enum lru_list lru)
857
{
K
KAMEZAWA Hiroyuki 已提交
858 859
	struct page_cgroup *pc;
	struct mem_cgroup_per_zone *mz;
860

861
	if (mem_cgroup_disabled())
K
KAMEZAWA Hiroyuki 已提交
862 863
		return;
	pc = lookup_page_cgroup(page);
864
	VM_BUG_ON(PageCgroupAcctLRU(pc));
K
KAMEZAWA Hiroyuki 已提交
865
	if (!PageCgroupUsed(pc))
L
Lee Schermerhorn 已提交
866
		return;
867 868
	/* Ensure pc->mem_cgroup is visible after reading PCG_USED. */
	smp_rmb();
869
	mz = page_cgroup_zoneinfo(pc->mem_cgroup, page);
870 871
	/* huge page split is done under lru_lock. so, we have no races. */
	MEM_CGROUP_ZSTAT(mz, lru) += 1 << compound_order(page);
872 873 874
	SetPageCgroupAcctLRU(pc);
	if (mem_cgroup_is_root(pc->mem_cgroup))
		return;
K
KAMEZAWA Hiroyuki 已提交
875 876
	list_add(&pc->lru, &mz->lists[lru]);
}
877

K
KAMEZAWA Hiroyuki 已提交
878
/*
879 880 881 882 883
 * At handling SwapCache, pc->mem_cgroup may be changed while it's linked to
 * lru because the page may.be reused after it's fully uncharged (because of
 * SwapCache behavior).To handle that, unlink page_cgroup from LRU when charge
 * it again. This function is only used to charge SwapCache. It's done under
 * lock_page and expected that zone->lru_lock is never held.
K
KAMEZAWA Hiroyuki 已提交
884
 */
885
static void mem_cgroup_lru_del_before_commit_swapcache(struct page *page)
K
KAMEZAWA Hiroyuki 已提交
886
{
887 888 889 890 891 892 893 894 895 896 897 898
	unsigned long flags;
	struct zone *zone = page_zone(page);
	struct page_cgroup *pc = lookup_page_cgroup(page);

	spin_lock_irqsave(&zone->lru_lock, flags);
	/*
	 * Forget old LRU when this page_cgroup is *not* used. This Used bit
	 * is guarded by lock_page() because the page is SwapCache.
	 */
	if (!PageCgroupUsed(pc))
		mem_cgroup_del_lru_list(page, page_lru(page));
	spin_unlock_irqrestore(&zone->lru_lock, flags);
K
KAMEZAWA Hiroyuki 已提交
899 900
}

901 902 903 904 905 906 907 908
static void mem_cgroup_lru_add_after_commit_swapcache(struct page *page)
{
	unsigned long flags;
	struct zone *zone = page_zone(page);
	struct page_cgroup *pc = lookup_page_cgroup(page);

	spin_lock_irqsave(&zone->lru_lock, flags);
	/* link when the page is linked to LRU but page_cgroup isn't */
909
	if (PageLRU(page) && !PageCgroupAcctLRU(pc))
910 911 912 913 914
		mem_cgroup_add_lru_list(page, page_lru(page));
	spin_unlock_irqrestore(&zone->lru_lock, flags);
}


K
KAMEZAWA Hiroyuki 已提交
915 916 917
void mem_cgroup_move_lists(struct page *page,
			   enum lru_list from, enum lru_list to)
{
918
	if (mem_cgroup_disabled())
K
KAMEZAWA Hiroyuki 已提交
919 920 921
		return;
	mem_cgroup_del_lru_list(page, from);
	mem_cgroup_add_lru_list(page, to);
922 923
}

924 925 926
int task_in_mem_cgroup(struct task_struct *task, const struct mem_cgroup *mem)
{
	int ret;
927
	struct mem_cgroup *curr = NULL;
928
	struct task_struct *p;
929

930 931 932 933 934
	p = find_lock_task_mm(task);
	if (!p)
		return 0;
	curr = try_get_mem_cgroup_from_mm(p->mm);
	task_unlock(p);
935 936
	if (!curr)
		return 0;
937 938 939 940 941 942 943
	/*
	 * We should check use_hierarchy of "mem" not "curr". Because checking
	 * use_hierarchy of "curr" here make this function true if hierarchy is
	 * enabled in "curr" and "curr" is a child of "mem" in *cgroup*
	 * hierarchy(even if use_hierarchy is disabled in "mem").
	 */
	if (mem->use_hierarchy)
944 945 946 947
		ret = css_is_ancestor(&curr->css, &mem->css);
	else
		ret = (curr == mem);
	css_put(&curr->css);
948 949 950
	return ret;
}

951
static int calc_inactive_ratio(struct mem_cgroup *memcg, unsigned long *present_pages)
952 953 954
{
	unsigned long active;
	unsigned long inactive;
955 956
	unsigned long gb;
	unsigned long inactive_ratio;
957

K
KAMEZAWA Hiroyuki 已提交
958 959
	inactive = mem_cgroup_get_local_zonestat(memcg, LRU_INACTIVE_ANON);
	active = mem_cgroup_get_local_zonestat(memcg, LRU_ACTIVE_ANON);
960

961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987
	gb = (inactive + active) >> (30 - PAGE_SHIFT);
	if (gb)
		inactive_ratio = int_sqrt(10 * gb);
	else
		inactive_ratio = 1;

	if (present_pages) {
		present_pages[0] = inactive;
		present_pages[1] = active;
	}

	return inactive_ratio;
}

int mem_cgroup_inactive_anon_is_low(struct mem_cgroup *memcg)
{
	unsigned long active;
	unsigned long inactive;
	unsigned long present_pages[2];
	unsigned long inactive_ratio;

	inactive_ratio = calc_inactive_ratio(memcg, present_pages);

	inactive = present_pages[0];
	active = present_pages[1];

	if (inactive * inactive_ratio < active)
988 989 990 991 992
		return 1;

	return 0;
}

993 994 995 996 997 998 999 1000 1001 1002 1003
int mem_cgroup_inactive_file_is_low(struct mem_cgroup *memcg)
{
	unsigned long active;
	unsigned long inactive;

	inactive = mem_cgroup_get_local_zonestat(memcg, LRU_INACTIVE_FILE);
	active = mem_cgroup_get_local_zonestat(memcg, LRU_ACTIVE_FILE);

	return (active > inactive);
}

1004 1005 1006 1007
unsigned long mem_cgroup_zone_nr_pages(struct mem_cgroup *memcg,
				       struct zone *zone,
				       enum lru_list lru)
{
1008
	int nid = zone_to_nid(zone);
1009 1010 1011 1012 1013 1014
	int zid = zone_idx(zone);
	struct mem_cgroup_per_zone *mz = mem_cgroup_zoneinfo(memcg, nid, zid);

	return MEM_CGROUP_ZSTAT(mz, lru);
}

K
KOSAKI Motohiro 已提交
1015 1016 1017
struct zone_reclaim_stat *mem_cgroup_get_reclaim_stat(struct mem_cgroup *memcg,
						      struct zone *zone)
{
1018
	int nid = zone_to_nid(zone);
K
KOSAKI Motohiro 已提交
1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034
	int zid = zone_idx(zone);
	struct mem_cgroup_per_zone *mz = mem_cgroup_zoneinfo(memcg, nid, zid);

	return &mz->reclaim_stat;
}

struct zone_reclaim_stat *
mem_cgroup_get_reclaim_stat_from_page(struct page *page)
{
	struct page_cgroup *pc;
	struct mem_cgroup_per_zone *mz;

	if (mem_cgroup_disabled())
		return NULL;

	pc = lookup_page_cgroup(page);
1035 1036
	if (!PageCgroupUsed(pc))
		return NULL;
1037 1038
	/* Ensure pc->mem_cgroup is visible after reading PCG_USED. */
	smp_rmb();
1039
	mz = page_cgroup_zoneinfo(pc->mem_cgroup, page);
K
KOSAKI Motohiro 已提交
1040 1041 1042
	return &mz->reclaim_stat;
}

1043 1044 1045 1046 1047
unsigned long mem_cgroup_isolate_pages(unsigned long nr_to_scan,
					struct list_head *dst,
					unsigned long *scanned, int order,
					int mode, struct zone *z,
					struct mem_cgroup *mem_cont,
1048
					int active, int file)
1049 1050 1051 1052 1053 1054
{
	unsigned long nr_taken = 0;
	struct page *page;
	unsigned long scan;
	LIST_HEAD(pc_list);
	struct list_head *src;
1055
	struct page_cgroup *pc, *tmp;
1056
	int nid = zone_to_nid(z);
1057 1058
	int zid = zone_idx(z);
	struct mem_cgroup_per_zone *mz;
1059
	int lru = LRU_FILE * file + active;
1060
	int ret;
1061

1062
	BUG_ON(!mem_cont);
1063
	mz = mem_cgroup_zoneinfo(mem_cont, nid, zid);
1064
	src = &mz->lists[lru];
1065

1066 1067
	scan = 0;
	list_for_each_entry_safe_reverse(pc, tmp, src, lru) {
H
Hugh Dickins 已提交
1068
		if (scan >= nr_to_scan)
1069
			break;
K
KAMEZAWA Hiroyuki 已提交
1070

1071 1072
		if (unlikely(!PageCgroupUsed(pc)))
			continue;
1073

1074
		page = lookup_cgroup_page(pc);
1075

H
Hugh Dickins 已提交
1076
		if (unlikely(!PageLRU(page)))
1077 1078
			continue;

H
Hugh Dickins 已提交
1079
		scan++;
1080 1081 1082
		ret = __isolate_lru_page(page, mode, file);
		switch (ret) {
		case 0:
1083
			list_move(&page->lru, dst);
1084
			mem_cgroup_del_lru(page);
1085
			nr_taken += hpage_nr_pages(page);
1086 1087 1088 1089 1090 1091 1092
			break;
		case -EBUSY:
			/* we don't affect global LRU but rotate in our LRU */
			mem_cgroup_rotate_lru_list(page, page_lru(page));
			break;
		default:
			break;
1093 1094 1095 1096
		}
	}

	*scanned = scan;
1097 1098 1099 1100

	trace_mm_vmscan_memcg_isolate(0, nr_to_scan, scan, nr_taken,
				      0, 0, 0, mode);

1101 1102 1103
	return nr_taken;
}

1104 1105 1106
#define mem_cgroup_from_res_counter(counter, member)	\
	container_of(counter, struct mem_cgroup, member)

1107
/**
1108 1109
 * mem_cgroup_margin - calculate chargeable space of a memory cgroup
 * @mem: the memory cgroup
1110
 *
1111 1112
 * Returns the maximum amount of memory @mem can be charged with, in
 * bytes.
1113
 */
1114
static unsigned long long mem_cgroup_margin(struct mem_cgroup *mem)
1115
{
1116 1117 1118 1119 1120 1121
	unsigned long long margin;

	margin = res_counter_margin(&mem->res);
	if (do_swap_account)
		margin = min(margin, res_counter_margin(&mem->memsw));
	return margin;
1122 1123
}

K
KOSAKI Motohiro 已提交
1124 1125 1126 1127 1128 1129 1130 1131
static unsigned int get_swappiness(struct mem_cgroup *memcg)
{
	struct cgroup *cgrp = memcg->css.cgroup;

	/* root ? */
	if (cgrp->parent == NULL)
		return vm_swappiness;

1132
	return memcg->swappiness;
K
KOSAKI Motohiro 已提交
1133 1134
}

1135 1136 1137
static void mem_cgroup_start_move(struct mem_cgroup *mem)
{
	int cpu;
1138 1139 1140 1141

	get_online_cpus();
	spin_lock(&mem->pcp_counter_lock);
	for_each_online_cpu(cpu)
1142
		per_cpu(mem->stat->count[MEM_CGROUP_ON_MOVE], cpu) += 1;
1143 1144 1145
	mem->nocpu_base.count[MEM_CGROUP_ON_MOVE] += 1;
	spin_unlock(&mem->pcp_counter_lock);
	put_online_cpus();
1146 1147 1148 1149 1150 1151 1152 1153 1154 1155

	synchronize_rcu();
}

static void mem_cgroup_end_move(struct mem_cgroup *mem)
{
	int cpu;

	if (!mem)
		return;
1156 1157 1158
	get_online_cpus();
	spin_lock(&mem->pcp_counter_lock);
	for_each_online_cpu(cpu)
1159
		per_cpu(mem->stat->count[MEM_CGROUP_ON_MOVE], cpu) -= 1;
1160 1161 1162
	mem->nocpu_base.count[MEM_CGROUP_ON_MOVE] -= 1;
	spin_unlock(&mem->pcp_counter_lock);
	put_online_cpus();
1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180
}
/*
 * 2 routines for checking "mem" is under move_account() or not.
 *
 * mem_cgroup_stealed() - checking a cgroup is mc.from or not. This is used
 *			  for avoiding race in accounting. If true,
 *			  pc->mem_cgroup may be overwritten.
 *
 * mem_cgroup_under_move() - checking a cgroup is mc.from or mc.to or
 *			  under hierarchy of moving cgroups. This is for
 *			  waiting at hith-memory prressure caused by "move".
 */

static bool mem_cgroup_stealed(struct mem_cgroup *mem)
{
	VM_BUG_ON(!rcu_read_lock_held());
	return this_cpu_read(mem->stat->count[MEM_CGROUP_ON_MOVE]) > 0;
}
1181 1182 1183

static bool mem_cgroup_under_move(struct mem_cgroup *mem)
{
1184 1185
	struct mem_cgroup *from;
	struct mem_cgroup *to;
1186
	bool ret = false;
1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201
	/*
	 * Unlike task_move routines, we access mc.to, mc.from not under
	 * mutual exclusion by cgroup_mutex. Here, we take spinlock instead.
	 */
	spin_lock(&mc.lock);
	from = mc.from;
	to = mc.to;
	if (!from)
		goto unlock;
	if (from == mem || to == mem
	    || (mem->use_hierarchy && css_is_ancestor(&from->css, &mem->css))
	    || (mem->use_hierarchy && css_is_ancestor(&to->css,	&mem->css)))
		ret = true;
unlock:
	spin_unlock(&mc.lock);
1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220
	return ret;
}

static bool mem_cgroup_wait_acct_move(struct mem_cgroup *mem)
{
	if (mc.moving_task && current != mc.moving_task) {
		if (mem_cgroup_under_move(mem)) {
			DEFINE_WAIT(wait);
			prepare_to_wait(&mc.waitq, &wait, TASK_INTERRUPTIBLE);
			/* moving charge context might have finished. */
			if (mc.moving_task)
				schedule();
			finish_wait(&mc.waitq, &wait);
			return true;
		}
	}
	return false;
}

1221
/**
1222
 * mem_cgroup_print_oom_info: Called from OOM with tasklist_lock held in read mode.
1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240
 * @memcg: The memory cgroup that went over limit
 * @p: Task that is going to be killed
 *
 * NOTE: @memcg and @p's mem_cgroup can be different when hierarchy is
 * enabled
 */
void mem_cgroup_print_oom_info(struct mem_cgroup *memcg, struct task_struct *p)
{
	struct cgroup *task_cgrp;
	struct cgroup *mem_cgrp;
	/*
	 * Need a buffer in BSS, can't rely on allocations. The code relies
	 * on the assumption that OOM is serialized for memory controller.
	 * If this assumption is broken, revisit this code.
	 */
	static char memcg_name[PATH_MAX];
	int ret;

1241
	if (!memcg || !p)
1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287
		return;


	rcu_read_lock();

	mem_cgrp = memcg->css.cgroup;
	task_cgrp = task_cgroup(p, mem_cgroup_subsys_id);

	ret = cgroup_path(task_cgrp, memcg_name, PATH_MAX);
	if (ret < 0) {
		/*
		 * Unfortunately, we are unable to convert to a useful name
		 * But we'll still print out the usage information
		 */
		rcu_read_unlock();
		goto done;
	}
	rcu_read_unlock();

	printk(KERN_INFO "Task in %s killed", memcg_name);

	rcu_read_lock();
	ret = cgroup_path(mem_cgrp, memcg_name, PATH_MAX);
	if (ret < 0) {
		rcu_read_unlock();
		goto done;
	}
	rcu_read_unlock();

	/*
	 * Continues from above, so we don't need an KERN_ level
	 */
	printk(KERN_CONT " as a result of limit of %s\n", memcg_name);
done:

	printk(KERN_INFO "memory: usage %llukB, limit %llukB, failcnt %llu\n",
		res_counter_read_u64(&memcg->res, RES_USAGE) >> 10,
		res_counter_read_u64(&memcg->res, RES_LIMIT) >> 10,
		res_counter_read_u64(&memcg->res, RES_FAILCNT));
	printk(KERN_INFO "memory+swap: usage %llukB, limit %llukB, "
		"failcnt %llu\n",
		res_counter_read_u64(&memcg->memsw, RES_USAGE) >> 10,
		res_counter_read_u64(&memcg->memsw, RES_LIMIT) >> 10,
		res_counter_read_u64(&memcg->memsw, RES_FAILCNT));
}

1288 1289 1290 1291 1292 1293 1294
/*
 * This function returns the number of memcg under hierarchy tree. Returns
 * 1(self count) if no children.
 */
static int mem_cgroup_count_children(struct mem_cgroup *mem)
{
	int num = 0;
K
KAMEZAWA Hiroyuki 已提交
1295 1296 1297 1298
	struct mem_cgroup *iter;

	for_each_mem_cgroup_tree(iter, mem)
		num++;
1299 1300 1301
	return num;
}

D
David Rientjes 已提交
1302 1303 1304 1305 1306 1307 1308 1309
/*
 * Return the memory (and swap, if configured) limit for a memcg.
 */
u64 mem_cgroup_get_limit(struct mem_cgroup *memcg)
{
	u64 limit;
	u64 memsw;

1310 1311 1312
	limit = res_counter_read_u64(&memcg->res, RES_LIMIT);
	limit += total_swap_pages << PAGE_SHIFT;

D
David Rientjes 已提交
1313 1314 1315 1316 1317 1318 1319 1320
	memsw = res_counter_read_u64(&memcg->memsw, RES_LIMIT);
	/*
	 * If memsw is finite and limits the amount of swap space available
	 * to this memcg, return that limit.
	 */
	return min(limit, memsw);
}

1321
/*
K
KAMEZAWA Hiroyuki 已提交
1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361
 * Visit the first child (need not be the first child as per the ordering
 * of the cgroup list, since we track last_scanned_child) of @mem and use
 * that to reclaim free pages from.
 */
static struct mem_cgroup *
mem_cgroup_select_victim(struct mem_cgroup *root_mem)
{
	struct mem_cgroup *ret = NULL;
	struct cgroup_subsys_state *css;
	int nextid, found;

	if (!root_mem->use_hierarchy) {
		css_get(&root_mem->css);
		ret = root_mem;
	}

	while (!ret) {
		rcu_read_lock();
		nextid = root_mem->last_scanned_child + 1;
		css = css_get_next(&mem_cgroup_subsys, nextid, &root_mem->css,
				   &found);
		if (css && css_tryget(css))
			ret = container_of(css, struct mem_cgroup, css);

		rcu_read_unlock();
		/* Updates scanning parameter */
		if (!css) {
			/* this means start scan from ID:1 */
			root_mem->last_scanned_child = 0;
		} else
			root_mem->last_scanned_child = found;
	}

	return ret;
}

/*
 * Scan the hierarchy if needed to reclaim memory. We remember the last child
 * we reclaimed from, so that we don't end up penalizing one child extensively
 * based on its position in the children list.
1362 1363
 *
 * root_mem is the original ancestor that we've been reclaim from.
K
KAMEZAWA Hiroyuki 已提交
1364 1365 1366
 *
 * We give up and return to the caller when we visit root_mem twice.
 * (other groups can be removed while we're walking....)
1367 1368
 *
 * If shrink==true, for avoiding to free too much, this returns immedieately.
1369 1370
 */
static int mem_cgroup_hierarchical_reclaim(struct mem_cgroup *root_mem,
1371
						struct zone *zone,
1372 1373
						gfp_t gfp_mask,
						unsigned long reclaim_options)
1374
{
K
KAMEZAWA Hiroyuki 已提交
1375 1376 1377
	struct mem_cgroup *victim;
	int ret, total = 0;
	int loop = 0;
1378 1379
	bool noswap = reclaim_options & MEM_CGROUP_RECLAIM_NOSWAP;
	bool shrink = reclaim_options & MEM_CGROUP_RECLAIM_SHRINK;
1380
	bool check_soft = reclaim_options & MEM_CGROUP_RECLAIM_SOFT;
1381 1382 1383
	unsigned long excess;

	excess = res_counter_soft_limit_excess(&root_mem->res) >> PAGE_SHIFT;
K
KAMEZAWA Hiroyuki 已提交
1384

1385 1386 1387 1388
	/* If memsw_is_minimum==1, swap-out is of-no-use. */
	if (root_mem->memsw_is_minimum)
		noswap = true;

1389
	while (1) {
K
KAMEZAWA Hiroyuki 已提交
1390
		victim = mem_cgroup_select_victim(root_mem);
1391
		if (victim == root_mem) {
K
KAMEZAWA Hiroyuki 已提交
1392
			loop++;
1393 1394
			if (loop >= 1)
				drain_all_stock_async();
1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417
			if (loop >= 2) {
				/*
				 * If we have not been able to reclaim
				 * anything, it might because there are
				 * no reclaimable pages under this hierarchy
				 */
				if (!check_soft || !total) {
					css_put(&victim->css);
					break;
				}
				/*
				 * We want to do more targetted reclaim.
				 * excess >> 2 is not to excessive so as to
				 * reclaim too much, nor too less that we keep
				 * coming back to reclaim from this cgroup
				 */
				if (total >= (excess >> 2) ||
					(loop > MEM_CGROUP_MAX_RECLAIM_LOOPS)) {
					css_put(&victim->css);
					break;
				}
			}
		}
1418
		if (!mem_cgroup_local_usage(victim)) {
K
KAMEZAWA Hiroyuki 已提交
1419 1420
			/* this cgroup's local usage == 0 */
			css_put(&victim->css);
1421 1422
			continue;
		}
K
KAMEZAWA Hiroyuki 已提交
1423
		/* we use swappiness of local cgroup */
1424 1425
		if (check_soft)
			ret = mem_cgroup_shrink_node_zone(victim, gfp_mask,
1426
				noswap, get_swappiness(victim), zone);
1427 1428 1429
		else
			ret = try_to_free_mem_cgroup_pages(victim, gfp_mask,
						noswap, get_swappiness(victim));
K
KAMEZAWA Hiroyuki 已提交
1430
		css_put(&victim->css);
1431 1432 1433 1434 1435 1436 1437
		/*
		 * At shrinking usage, we can't check we should stop here or
		 * reclaim more. It's depends on callers. last_scanned_child
		 * will work enough for keeping fairness under tree.
		 */
		if (shrink)
			return ret;
K
KAMEZAWA Hiroyuki 已提交
1438
		total += ret;
1439
		if (check_soft) {
1440
			if (!res_counter_soft_limit_excess(&root_mem->res))
1441
				return total;
1442
		} else if (mem_cgroup_margin(root_mem))
K
KAMEZAWA Hiroyuki 已提交
1443
			return 1 + total;
1444
	}
K
KAMEZAWA Hiroyuki 已提交
1445
	return total;
1446 1447
}

K
KAMEZAWA Hiroyuki 已提交
1448 1449 1450 1451 1452 1453
/*
 * Check OOM-Killer is already running under our hierarchy.
 * If someone is running, return false.
 */
static bool mem_cgroup_oom_lock(struct mem_cgroup *mem)
{
K
KAMEZAWA Hiroyuki 已提交
1454 1455
	int x, lock_count = 0;
	struct mem_cgroup *iter;
1456

K
KAMEZAWA Hiroyuki 已提交
1457 1458 1459 1460
	for_each_mem_cgroup_tree(iter, mem) {
		x = atomic_inc_return(&iter->oom_lock);
		lock_count = max(x, lock_count);
	}
K
KAMEZAWA Hiroyuki 已提交
1461 1462 1463 1464

	if (lock_count == 1)
		return true;
	return false;
1465
}
1466

K
KAMEZAWA Hiroyuki 已提交
1467
static int mem_cgroup_oom_unlock(struct mem_cgroup *mem)
1468
{
K
KAMEZAWA Hiroyuki 已提交
1469 1470
	struct mem_cgroup *iter;

K
KAMEZAWA Hiroyuki 已提交
1471 1472 1473 1474 1475
	/*
	 * When a new child is created while the hierarchy is under oom,
	 * mem_cgroup_oom_lock() may not be called. We have to use
	 * atomic_add_unless() here.
	 */
K
KAMEZAWA Hiroyuki 已提交
1476 1477
	for_each_mem_cgroup_tree(iter, mem)
		atomic_add_unless(&iter->oom_lock, -1, 0);
1478 1479 1480
	return 0;
}

K
KAMEZAWA Hiroyuki 已提交
1481 1482 1483 1484

static DEFINE_MUTEX(memcg_oom_mutex);
static DECLARE_WAIT_QUEUE_HEAD(memcg_oom_waitq);

K
KAMEZAWA Hiroyuki 已提交
1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520
struct oom_wait_info {
	struct mem_cgroup *mem;
	wait_queue_t	wait;
};

static int memcg_oom_wake_function(wait_queue_t *wait,
	unsigned mode, int sync, void *arg)
{
	struct mem_cgroup *wake_mem = (struct mem_cgroup *)arg;
	struct oom_wait_info *oom_wait_info;

	oom_wait_info = container_of(wait, struct oom_wait_info, wait);

	if (oom_wait_info->mem == wake_mem)
		goto wakeup;
	/* if no hierarchy, no match */
	if (!oom_wait_info->mem->use_hierarchy || !wake_mem->use_hierarchy)
		return 0;
	/*
	 * Both of oom_wait_info->mem and wake_mem are stable under us.
	 * Then we can use css_is_ancestor without taking care of RCU.
	 */
	if (!css_is_ancestor(&oom_wait_info->mem->css, &wake_mem->css) &&
	    !css_is_ancestor(&wake_mem->css, &oom_wait_info->mem->css))
		return 0;

wakeup:
	return autoremove_wake_function(wait, mode, sync, arg);
}

static void memcg_wakeup_oom(struct mem_cgroup *mem)
{
	/* for filtering, pass "mem" as argument. */
	__wake_up(&memcg_oom_waitq, TASK_NORMAL, 0, mem);
}

1521 1522
static void memcg_oom_recover(struct mem_cgroup *mem)
{
1523
	if (mem && atomic_read(&mem->oom_lock))
1524 1525 1526
		memcg_wakeup_oom(mem);
}

K
KAMEZAWA Hiroyuki 已提交
1527 1528 1529 1530
/*
 * try to call OOM killer. returns false if we should exit memory-reclaim loop.
 */
bool mem_cgroup_handle_oom(struct mem_cgroup *mem, gfp_t mask)
1531
{
K
KAMEZAWA Hiroyuki 已提交
1532
	struct oom_wait_info owait;
1533
	bool locked, need_to_kill;
K
KAMEZAWA Hiroyuki 已提交
1534

K
KAMEZAWA Hiroyuki 已提交
1535 1536 1537 1538 1539
	owait.mem = mem;
	owait.wait.flags = 0;
	owait.wait.func = memcg_oom_wake_function;
	owait.wait.private = current;
	INIT_LIST_HEAD(&owait.wait.task_list);
1540
	need_to_kill = true;
K
KAMEZAWA Hiroyuki 已提交
1541 1542 1543 1544 1545 1546 1547 1548
	/* At first, try to OOM lock hierarchy under mem.*/
	mutex_lock(&memcg_oom_mutex);
	locked = mem_cgroup_oom_lock(mem);
	/*
	 * Even if signal_pending(), we can't quit charge() loop without
	 * accounting. So, UNINTERRUPTIBLE is appropriate. But SIGKILL
	 * under OOM is always welcomed, use TASK_KILLABLE here.
	 */
1549 1550 1551 1552
	prepare_to_wait(&memcg_oom_waitq, &owait.wait, TASK_KILLABLE);
	if (!locked || mem->oom_kill_disable)
		need_to_kill = false;
	if (locked)
K
KAMEZAWA Hiroyuki 已提交
1553
		mem_cgroup_oom_notify(mem);
K
KAMEZAWA Hiroyuki 已提交
1554 1555
	mutex_unlock(&memcg_oom_mutex);

1556 1557
	if (need_to_kill) {
		finish_wait(&memcg_oom_waitq, &owait.wait);
K
KAMEZAWA Hiroyuki 已提交
1558
		mem_cgroup_out_of_memory(mem, mask);
1559
	} else {
K
KAMEZAWA Hiroyuki 已提交
1560
		schedule();
K
KAMEZAWA Hiroyuki 已提交
1561
		finish_wait(&memcg_oom_waitq, &owait.wait);
K
KAMEZAWA Hiroyuki 已提交
1562 1563 1564
	}
	mutex_lock(&memcg_oom_mutex);
	mem_cgroup_oom_unlock(mem);
K
KAMEZAWA Hiroyuki 已提交
1565
	memcg_wakeup_oom(mem);
K
KAMEZAWA Hiroyuki 已提交
1566 1567 1568 1569 1570 1571 1572
	mutex_unlock(&memcg_oom_mutex);

	if (test_thread_flag(TIF_MEMDIE) || fatal_signal_pending(current))
		return false;
	/* Give chance to dying process */
	schedule_timeout(1);
	return true;
1573 1574
}

1575 1576 1577
/*
 * Currently used to update mapped file statistics, but the routine can be
 * generalized to update other statistics as well.
1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596
 *
 * Notes: Race condition
 *
 * We usually use page_cgroup_lock() for accessing page_cgroup member but
 * it tends to be costly. But considering some conditions, we doesn't need
 * to do so _always_.
 *
 * Considering "charge", lock_page_cgroup() is not required because all
 * file-stat operations happen after a page is attached to radix-tree. There
 * are no race with "charge".
 *
 * Considering "uncharge", we know that memcg doesn't clear pc->mem_cgroup
 * at "uncharge" intentionally. So, we always see valid pc->mem_cgroup even
 * if there are race with "uncharge". Statistics itself is properly handled
 * by flags.
 *
 * Considering "move", this is an only case we see a race. To make the race
 * small, we check MEM_CGROUP_ON_MOVE percpu value and detect there are
 * possibility of race condition. If there is, we take a lock.
1597
 */
1598

1599 1600
void mem_cgroup_update_page_stat(struct page *page,
				 enum mem_cgroup_page_stat_item idx, int val)
1601 1602
{
	struct mem_cgroup *mem;
1603 1604
	struct page_cgroup *pc = lookup_page_cgroup(page);
	bool need_unlock = false;
1605
	unsigned long uninitialized_var(flags);
1606 1607 1608 1609

	if (unlikely(!pc))
		return;

1610
	rcu_read_lock();
1611
	mem = pc->mem_cgroup;
1612 1613 1614
	if (unlikely(!mem || !PageCgroupUsed(pc)))
		goto out;
	/* pc->mem_cgroup is unstable ? */
1615
	if (unlikely(mem_cgroup_stealed(mem)) || PageTransHuge(page)) {
1616
		/* take a lock against to access pc->mem_cgroup */
1617
		move_lock_page_cgroup(pc, &flags);
1618 1619 1620 1621 1622
		need_unlock = true;
		mem = pc->mem_cgroup;
		if (!mem || !PageCgroupUsed(pc))
			goto out;
	}
1623 1624

	switch (idx) {
1625
	case MEMCG_NR_FILE_MAPPED:
1626 1627 1628
		if (val > 0)
			SetPageCgroupFileMapped(pc);
		else if (!page_mapped(page))
1629
			ClearPageCgroupFileMapped(pc);
1630
		idx = MEM_CGROUP_STAT_FILE_MAPPED;
1631 1632 1633
		break;
	default:
		BUG();
1634
	}
1635

1636 1637
	this_cpu_add(mem->stat->count[idx], val);

1638 1639
out:
	if (unlikely(need_unlock))
1640
		move_unlock_page_cgroup(pc, &flags);
1641 1642
	rcu_read_unlock();
	return;
1643
}
1644
EXPORT_SYMBOL(mem_cgroup_update_page_stat);
1645

1646 1647 1648 1649 1650 1651 1652
/*
 * size of first charge trial. "32" comes from vmscan.c's magic value.
 * TODO: maybe necessary to use big numbers in big irons.
 */
#define CHARGE_SIZE	(32 * PAGE_SIZE)
struct memcg_stock_pcp {
	struct mem_cgroup *cached; /* this never be root cgroup */
1653
	unsigned int nr_pages;
1654 1655 1656 1657 1658 1659
	struct work_struct work;
};
static DEFINE_PER_CPU(struct memcg_stock_pcp, memcg_stock);
static atomic_t memcg_drain_count;

/*
1660
 * Try to consume stocked charge on this cpu. If success, one page is consumed
1661 1662 1663 1664 1665 1666 1667 1668 1669 1670
 * from local stock and true is returned. If the stock is 0 or charges from a
 * cgroup which is not current target, returns false. This stock will be
 * refilled.
 */
static bool consume_stock(struct mem_cgroup *mem)
{
	struct memcg_stock_pcp *stock;
	bool ret = true;

	stock = &get_cpu_var(memcg_stock);
1671 1672
	if (mem == stock->cached && stock->nr_pages)
		stock->nr_pages--;
1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685
	else /* need to call res_counter_charge */
		ret = false;
	put_cpu_var(memcg_stock);
	return ret;
}

/*
 * Returns stocks cached in percpu to res_counter and reset cached information.
 */
static void drain_stock(struct memcg_stock_pcp *stock)
{
	struct mem_cgroup *old = stock->cached;

1686 1687 1688 1689
	if (stock->nr_pages) {
		unsigned long bytes = stock->nr_pages * PAGE_SIZE;

		res_counter_uncharge(&old->res, bytes);
1690
		if (do_swap_account)
1691 1692
			res_counter_uncharge(&old->memsw, bytes);
		stock->nr_pages = 0;
1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708
	}
	stock->cached = NULL;
}

/*
 * This must be called under preempt disabled or must be called by
 * a thread which is pinned to local cpu.
 */
static void drain_local_stock(struct work_struct *dummy)
{
	struct memcg_stock_pcp *stock = &__get_cpu_var(memcg_stock);
	drain_stock(stock);
}

/*
 * Cache charges(val) which is from res_counter, to local per_cpu area.
1709
 * This will be consumed by consume_stock() function, later.
1710
 */
1711
static void refill_stock(struct mem_cgroup *mem, unsigned int nr_pages)
1712 1713 1714 1715 1716 1717 1718
{
	struct memcg_stock_pcp *stock = &get_cpu_var(memcg_stock);

	if (stock->cached != mem) { /* reset if necessary */
		drain_stock(stock);
		stock->cached = mem;
	}
1719
	stock->nr_pages += nr_pages;
1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760
	put_cpu_var(memcg_stock);
}

/*
 * Tries to drain stocked charges in other cpus. This function is asynchronous
 * and just put a work per cpu for draining localy on each cpu. Caller can
 * expects some charges will be back to res_counter later but cannot wait for
 * it.
 */
static void drain_all_stock_async(void)
{
	int cpu;
	/* This function is for scheduling "drain" in asynchronous way.
	 * The result of "drain" is not directly handled by callers. Then,
	 * if someone is calling drain, we don't have to call drain more.
	 * Anyway, WORK_STRUCT_PENDING check in queue_work_on() will catch if
	 * there is a race. We just do loose check here.
	 */
	if (atomic_read(&memcg_drain_count))
		return;
	/* Notify other cpus that system-wide "drain" is running */
	atomic_inc(&memcg_drain_count);
	get_online_cpus();
	for_each_online_cpu(cpu) {
		struct memcg_stock_pcp *stock = &per_cpu(memcg_stock, cpu);
		schedule_work_on(cpu, &stock->work);
	}
 	put_online_cpus();
	atomic_dec(&memcg_drain_count);
	/* We don't wait for flush_work */
}

/* This is a synchronous drain interface. */
static void drain_all_stock_sync(void)
{
	/* called when force_empty is called */
	atomic_inc(&memcg_drain_count);
	schedule_on_each_cpu(drain_local_stock);
	atomic_dec(&memcg_drain_count);
}

1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775
/*
 * This function drains percpu counter value from DEAD cpu and
 * move it to local cpu. Note that this function can be preempted.
 */
static void mem_cgroup_drain_pcp_counter(struct mem_cgroup *mem, int cpu)
{
	int i;

	spin_lock(&mem->pcp_counter_lock);
	for (i = 0; i < MEM_CGROUP_STAT_DATA; i++) {
		s64 x = per_cpu(mem->stat->count[i], cpu);

		per_cpu(mem->stat->count[i], cpu) = 0;
		mem->nocpu_base.count[i] += x;
	}
1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786
	/* need to clear ON_MOVE value, works as a kind of lock. */
	per_cpu(mem->stat->count[MEM_CGROUP_ON_MOVE], cpu) = 0;
	spin_unlock(&mem->pcp_counter_lock);
}

static void synchronize_mem_cgroup_on_move(struct mem_cgroup *mem, int cpu)
{
	int idx = MEM_CGROUP_ON_MOVE;

	spin_lock(&mem->pcp_counter_lock);
	per_cpu(mem->stat->count[idx], cpu) = mem->nocpu_base.count[idx];
1787 1788 1789 1790
	spin_unlock(&mem->pcp_counter_lock);
}

static int __cpuinit memcg_cpu_hotplug_callback(struct notifier_block *nb,
1791 1792 1793 1794 1795
					unsigned long action,
					void *hcpu)
{
	int cpu = (unsigned long)hcpu;
	struct memcg_stock_pcp *stock;
1796
	struct mem_cgroup *iter;
1797

1798 1799 1800 1801 1802 1803
	if ((action == CPU_ONLINE)) {
		for_each_mem_cgroup_all(iter)
			synchronize_mem_cgroup_on_move(iter, cpu);
		return NOTIFY_OK;
	}

1804
	if ((action != CPU_DEAD) || action != CPU_DEAD_FROZEN)
1805
		return NOTIFY_OK;
1806 1807 1808 1809

	for_each_mem_cgroup_all(iter)
		mem_cgroup_drain_pcp_counter(iter, cpu);

1810 1811 1812 1813 1814
	stock = &per_cpu(memcg_stock, cpu);
	drain_stock(stock);
	return NOTIFY_OK;
}

1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841

/* See __mem_cgroup_try_charge() for details */
enum {
	CHARGE_OK,		/* success */
	CHARGE_RETRY,		/* need to retry but retry is not bad */
	CHARGE_NOMEM,		/* we can't do more. return -ENOMEM */
	CHARGE_WOULDBLOCK,	/* GFP_WAIT wasn't set and no enough res. */
	CHARGE_OOM_DIE,		/* the current is killed because of OOM */
};

static int __mem_cgroup_do_charge(struct mem_cgroup *mem, gfp_t gfp_mask,
				int csize, bool oom_check)
{
	struct mem_cgroup *mem_over_limit;
	struct res_counter *fail_res;
	unsigned long flags = 0;
	int ret;

	ret = res_counter_charge(&mem->res, csize, &fail_res);

	if (likely(!ret)) {
		if (!do_swap_account)
			return CHARGE_OK;
		ret = res_counter_charge(&mem->memsw, csize, &fail_res);
		if (likely(!ret))
			return CHARGE_OK;

1842
		res_counter_uncharge(&mem->res, csize);
1843 1844 1845 1846
		mem_over_limit = mem_cgroup_from_res_counter(fail_res, memsw);
		flags |= MEM_CGROUP_RECLAIM_NOSWAP;
	} else
		mem_over_limit = mem_cgroup_from_res_counter(fail_res, res);
1847 1848 1849 1850 1851 1852 1853 1854 1855
	/*
	 * csize can be either a huge page (HPAGE_SIZE), a batch of
	 * regular pages (CHARGE_SIZE), or a single regular page
	 * (PAGE_SIZE).
	 *
	 * Never reclaim on behalf of optional batching, retry with a
	 * single page instead.
	 */
	if (csize == CHARGE_SIZE)
1856 1857 1858 1859 1860 1861
		return CHARGE_RETRY;

	if (!(gfp_mask & __GFP_WAIT))
		return CHARGE_WOULDBLOCK;

	ret = mem_cgroup_hierarchical_reclaim(mem_over_limit, NULL,
1862
					      gfp_mask, flags);
1863
	if (mem_cgroup_margin(mem_over_limit) >= csize)
1864
		return CHARGE_RETRY;
1865
	/*
1866 1867 1868 1869 1870 1871 1872
	 * Even though the limit is exceeded at this point, reclaim
	 * may have been able to free some pages.  Retry the charge
	 * before killing the task.
	 *
	 * Only for regular pages, though: huge pages are rather
	 * unlikely to succeed so close to the limit, and we fall back
	 * to regular pages anyway in case of failure.
1873
	 */
1874
	if (csize == PAGE_SIZE && ret)
1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893
		return CHARGE_RETRY;

	/*
	 * At task move, charge accounts can be doubly counted. So, it's
	 * better to wait until the end of task_move if something is going on.
	 */
	if (mem_cgroup_wait_acct_move(mem_over_limit))
		return CHARGE_RETRY;

	/* If we don't need to call oom-killer at el, return immediately */
	if (!oom_check)
		return CHARGE_NOMEM;
	/* check OOM */
	if (!mem_cgroup_handle_oom(mem_over_limit, gfp_mask))
		return CHARGE_OOM_DIE;

	return CHARGE_RETRY;
}

1894 1895 1896
/*
 * Unlike exported interface, "oom" parameter is added. if oom==true,
 * oom-killer can be invoked.
1897
 */
1898
static int __mem_cgroup_try_charge(struct mm_struct *mm,
A
Andrea Arcangeli 已提交
1899 1900 1901
				   gfp_t gfp_mask,
				   struct mem_cgroup **memcg, bool oom,
				   int page_size)
1902
{
1903 1904 1905
	int nr_oom_retries = MEM_CGROUP_RECLAIM_RETRIES;
	struct mem_cgroup *mem = NULL;
	int ret;
A
Andrea Arcangeli 已提交
1906
	int csize = max(CHARGE_SIZE, (unsigned long) page_size);
1907

K
KAMEZAWA Hiroyuki 已提交
1908 1909 1910 1911 1912 1913 1914 1915
	/*
	 * Unlike gloval-vm's OOM-kill, we're not in memory shortage
	 * in system level. So, allow to go ahead dying process in addition to
	 * MEMDIE process.
	 */
	if (unlikely(test_thread_flag(TIF_MEMDIE)
		     || fatal_signal_pending(current)))
		goto bypass;
1916

1917
	/*
1918 1919
	 * We always charge the cgroup the mm_struct belongs to.
	 * The mm_struct's mem_cgroup changes on task migration if the
1920 1921 1922
	 * thread group leader migrates. It's possible that mm is not
	 * set, if so charge the init_mm (happens for pagecache usage).
	 */
K
KAMEZAWA Hiroyuki 已提交
1923 1924 1925 1926
	if (!*memcg && !mm)
		goto bypass;
again:
	if (*memcg) { /* css should be a valid one */
1927
		mem = *memcg;
K
KAMEZAWA Hiroyuki 已提交
1928 1929 1930
		VM_BUG_ON(css_is_removed(&mem->css));
		if (mem_cgroup_is_root(mem))
			goto done;
A
Andrea Arcangeli 已提交
1931
		if (page_size == PAGE_SIZE && consume_stock(mem))
K
KAMEZAWA Hiroyuki 已提交
1932
			goto done;
1933 1934
		css_get(&mem->css);
	} else {
K
KAMEZAWA Hiroyuki 已提交
1935
		struct task_struct *p;
1936

K
KAMEZAWA Hiroyuki 已提交
1937 1938 1939
		rcu_read_lock();
		p = rcu_dereference(mm->owner);
		/*
1940 1941 1942 1943 1944 1945 1946 1947
		 * Because we don't have task_lock(), "p" can exit.
		 * In that case, "mem" can point to root or p can be NULL with
		 * race with swapoff. Then, we have small risk of mis-accouning.
		 * But such kind of mis-account by race always happens because
		 * we don't have cgroup_mutex(). It's overkill and we allo that
		 * small race, here.
		 * (*) swapoff at el will charge against mm-struct not against
		 * task-struct. So, mm->owner can be NULL.
K
KAMEZAWA Hiroyuki 已提交
1948 1949
		 */
		mem = mem_cgroup_from_task(p);
1950
		if (!mem || mem_cgroup_is_root(mem)) {
K
KAMEZAWA Hiroyuki 已提交
1951 1952 1953
			rcu_read_unlock();
			goto done;
		}
A
Andrea Arcangeli 已提交
1954
		if (page_size == PAGE_SIZE && consume_stock(mem)) {
K
KAMEZAWA Hiroyuki 已提交
1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972
			/*
			 * It seems dagerous to access memcg without css_get().
			 * But considering how consume_stok works, it's not
			 * necessary. If consume_stock success, some charges
			 * from this memcg are cached on this cpu. So, we
			 * don't need to call css_get()/css_tryget() before
			 * calling consume_stock().
			 */
			rcu_read_unlock();
			goto done;
		}
		/* after here, we may be blocked. we need to get refcnt */
		if (!css_tryget(&mem->css)) {
			rcu_read_unlock();
			goto again;
		}
		rcu_read_unlock();
	}
1973

1974 1975
	do {
		bool oom_check;
1976

1977
		/* If killed, bypass charge */
K
KAMEZAWA Hiroyuki 已提交
1978 1979
		if (fatal_signal_pending(current)) {
			css_put(&mem->css);
1980
			goto bypass;
K
KAMEZAWA Hiroyuki 已提交
1981
		}
1982

1983 1984 1985 1986
		oom_check = false;
		if (oom && !nr_oom_retries) {
			oom_check = true;
			nr_oom_retries = MEM_CGROUP_RECLAIM_RETRIES;
1987
		}
1988

1989
		ret = __mem_cgroup_do_charge(mem, gfp_mask, csize, oom_check);
1990

1991 1992 1993 1994
		switch (ret) {
		case CHARGE_OK:
			break;
		case CHARGE_RETRY: /* not in OOM situation but retry */
A
Andrea Arcangeli 已提交
1995
			csize = page_size;
K
KAMEZAWA Hiroyuki 已提交
1996 1997 1998
			css_put(&mem->css);
			mem = NULL;
			goto again;
1999
		case CHARGE_WOULDBLOCK: /* !__GFP_WAIT */
K
KAMEZAWA Hiroyuki 已提交
2000
			css_put(&mem->css);
2001 2002
			goto nomem;
		case CHARGE_NOMEM: /* OOM routine works */
K
KAMEZAWA Hiroyuki 已提交
2003 2004
			if (!oom) {
				css_put(&mem->css);
K
KAMEZAWA Hiroyuki 已提交
2005
				goto nomem;
K
KAMEZAWA Hiroyuki 已提交
2006
			}
2007 2008 2009 2010
			/* If oom, we never return -ENOMEM */
			nr_oom_retries--;
			break;
		case CHARGE_OOM_DIE: /* Killed by OOM Killer */
K
KAMEZAWA Hiroyuki 已提交
2011
			css_put(&mem->css);
K
KAMEZAWA Hiroyuki 已提交
2012
			goto bypass;
2013
		}
2014 2015
	} while (ret != CHARGE_OK);

A
Andrea Arcangeli 已提交
2016
	if (csize > page_size)
2017
		refill_stock(mem, (csize - page_size) >> PAGE_SHIFT);
K
KAMEZAWA Hiroyuki 已提交
2018
	css_put(&mem->css);
2019
done:
K
KAMEZAWA Hiroyuki 已提交
2020
	*memcg = mem;
2021 2022
	return 0;
nomem:
K
KAMEZAWA Hiroyuki 已提交
2023
	*memcg = NULL;
2024
	return -ENOMEM;
K
KAMEZAWA Hiroyuki 已提交
2025 2026 2027
bypass:
	*memcg = NULL;
	return 0;
2028
}
2029

2030 2031 2032 2033 2034
/*
 * Somemtimes we have to undo a charge we got by try_charge().
 * This function is for that and do uncharge, put css's refcnt.
 * gotten by try_charge().
 */
2035
static void __mem_cgroup_cancel_charge(struct mem_cgroup *mem,
2036
				       unsigned int nr_pages)
2037 2038
{
	if (!mem_cgroup_is_root(mem)) {
2039 2040 2041
		unsigned long bytes = nr_pages * PAGE_SIZE;

		res_counter_uncharge(&mem->res, bytes);
2042
		if (do_swap_account)
2043
			res_counter_uncharge(&mem->memsw, bytes);
2044
	}
2045 2046
}

2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065
/*
 * A helper function to get mem_cgroup from ID. must be called under
 * rcu_read_lock(). The caller must check css_is_removed() or some if
 * it's concern. (dropping refcnt from swap can be called against removed
 * memcg.)
 */
static struct mem_cgroup *mem_cgroup_lookup(unsigned short id)
{
	struct cgroup_subsys_state *css;

	/* ID 0 is unused ID */
	if (!id)
		return NULL;
	css = css_lookup(&mem_cgroup_subsys, id);
	if (!css)
		return NULL;
	return container_of(css, struct mem_cgroup, css);
}

2066
struct mem_cgroup *try_get_mem_cgroup_from_page(struct page *page)
2067
{
2068
	struct mem_cgroup *mem = NULL;
2069
	struct page_cgroup *pc;
2070
	unsigned short id;
2071 2072
	swp_entry_t ent;

2073 2074 2075
	VM_BUG_ON(!PageLocked(page));

	pc = lookup_page_cgroup(page);
2076
	lock_page_cgroup(pc);
2077
	if (PageCgroupUsed(pc)) {
2078
		mem = pc->mem_cgroup;
2079 2080
		if (mem && !css_tryget(&mem->css))
			mem = NULL;
2081
	} else if (PageSwapCache(page)) {
2082
		ent.val = page_private(page);
2083 2084 2085 2086 2087 2088
		id = lookup_swap_cgroup(ent);
		rcu_read_lock();
		mem = mem_cgroup_lookup(id);
		if (mem && !css_tryget(&mem->css))
			mem = NULL;
		rcu_read_unlock();
2089
	}
2090
	unlock_page_cgroup(pc);
2091 2092 2093
	return mem;
}

2094
static void __mem_cgroup_commit_charge(struct mem_cgroup *mem,
2095
				       struct page *page,
2096 2097 2098
				       struct page_cgroup *pc,
				       enum charge_type ctype,
				       int page_size)
2099
{
2100 2101 2102 2103 2104
	int nr_pages = page_size >> PAGE_SHIFT;

	lock_page_cgroup(pc);
	if (unlikely(PageCgroupUsed(pc))) {
		unlock_page_cgroup(pc);
2105
		__mem_cgroup_cancel_charge(mem, nr_pages);
2106 2107 2108 2109 2110 2111
		return;
	}
	/*
	 * we don't need page_cgroup_lock about tail pages, becase they are not
	 * accessed by any other context at this point.
	 */
2112
	pc->mem_cgroup = mem;
2113 2114 2115 2116 2117 2118 2119
	/*
	 * We access a page_cgroup asynchronously without lock_page_cgroup().
	 * Especially when a page_cgroup is taken from a page, pc->mem_cgroup
	 * is accessed after testing USED bit. To make pc->mem_cgroup visible
	 * before USED bit, we need memory barrier here.
	 * See mem_cgroup_add_lru_list(), etc.
 	 */
K
KAMEZAWA Hiroyuki 已提交
2120
	smp_wmb();
2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133
	switch (ctype) {
	case MEM_CGROUP_CHARGE_TYPE_CACHE:
	case MEM_CGROUP_CHARGE_TYPE_SHMEM:
		SetPageCgroupCache(pc);
		SetPageCgroupUsed(pc);
		break;
	case MEM_CGROUP_CHARGE_TYPE_MAPPED:
		ClearPageCgroupCache(pc);
		SetPageCgroupUsed(pc);
		break;
	default:
		break;
	}
2134

2135
	mem_cgroup_charge_statistics(mem, PageCgroupCache(pc), nr_pages);
2136
	unlock_page_cgroup(pc);
2137 2138 2139 2140 2141
	/*
	 * "charge_statistics" updated event counter. Then, check it.
	 * Insert ancestor (and ancestor's ancestors), to softlimit RB-tree.
	 * if they exceeds softlimit.
	 */
2142
	memcg_check_events(mem, page);
2143
}
2144

2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158
#ifdef CONFIG_TRANSPARENT_HUGEPAGE

#define PCGF_NOCOPY_AT_SPLIT ((1 << PCG_LOCK) | (1 << PCG_MOVE_LOCK) |\
			(1 << PCG_ACCT_LRU) | (1 << PCG_MIGRATION))
/*
 * Because tail pages are not marked as "used", set it. We're under
 * zone->lru_lock, 'splitting on pmd' and compund_lock.
 */
void mem_cgroup_split_huge_fixup(struct page *head, struct page *tail)
{
	struct page_cgroup *head_pc = lookup_page_cgroup(head);
	struct page_cgroup *tail_pc = lookup_page_cgroup(tail);
	unsigned long flags;

2159 2160
	if (mem_cgroup_disabled())
		return;
2161
	/*
2162
	 * We have no races with charge/uncharge but will have races with
2163 2164 2165 2166 2167 2168
	 * page state accounting.
	 */
	move_lock_page_cgroup(head_pc, &flags);

	tail_pc->mem_cgroup = head_pc->mem_cgroup;
	smp_wmb(); /* see __commit_charge() */
2169 2170 2171 2172 2173 2174 2175 2176 2177 2178
	if (PageCgroupAcctLRU(head_pc)) {
		enum lru_list lru;
		struct mem_cgroup_per_zone *mz;

		/*
		 * LRU flags cannot be copied because we need to add tail
		 *.page to LRU by generic call and our hook will be called.
		 * We hold lru_lock, then, reduce counter directly.
		 */
		lru = page_lru(head);
2179
		mz = page_cgroup_zoneinfo(head_pc->mem_cgroup, head);
2180 2181
		MEM_CGROUP_ZSTAT(mz, lru) -= 1;
	}
2182 2183 2184 2185 2186
	tail_pc->flags = head_pc->flags & ~PCGF_NOCOPY_AT_SPLIT;
	move_unlock_page_cgroup(head_pc, &flags);
}
#endif

2187
/**
2188
 * mem_cgroup_move_account - move account of the page
2189
 * @page: the page
2190 2191 2192
 * @pc:	page_cgroup of the page.
 * @from: mem_cgroup which the page is moved from.
 * @to:	mem_cgroup which the page is moved to. @from != @to.
2193
 * @uncharge: whether we should call uncharge and css_put against @from.
2194
 * @charge_size: number of bytes to charge (regular or huge page)
2195 2196
 *
 * The caller must confirm following.
K
KAMEZAWA Hiroyuki 已提交
2197
 * - page is not on LRU (isolate_page() is useful.)
2198
 * - compound_lock is held when charge_size > PAGE_SIZE
2199
 *
2200 2201 2202 2203
 * This function doesn't do "charge" nor css_get to new cgroup. It should be
 * done by a caller(__mem_cgroup_try_charge would be usefull). If @uncharge is
 * true, this function does "uncharge" from old cgroup, but it doesn't if
 * @uncharge is false, so a caller should do "uncharge".
2204
 */
2205
static int mem_cgroup_move_account(struct page *page, struct page_cgroup *pc,
2206 2207
				   struct mem_cgroup *from, struct mem_cgroup *to,
				   bool uncharge, int charge_size)
2208
{
2209
	int nr_pages = charge_size >> PAGE_SHIFT;
2210 2211
	unsigned long flags;
	int ret;
2212

2213
	VM_BUG_ON(from == to);
2214
	VM_BUG_ON(PageLRU(page));
2215 2216 2217 2218 2219 2220 2221
	/*
	 * The page is isolated from LRU. So, collapse function
	 * will not handle this page. But page splitting can happen.
	 * Do this check under compound_page_lock(). The caller should
	 * hold it.
	 */
	ret = -EBUSY;
2222
	if (charge_size > PAGE_SIZE && !PageTransHuge(page))
2223 2224 2225 2226 2227 2228 2229 2230 2231
		goto out;

	lock_page_cgroup(pc);

	ret = -EINVAL;
	if (!PageCgroupUsed(pc) || pc->mem_cgroup != from)
		goto unlock;

	move_lock_page_cgroup(pc, &flags);
2232

2233
	if (PageCgroupFileMapped(pc)) {
2234 2235 2236 2237 2238
		/* Update mapped_file data for mem_cgroup */
		preempt_disable();
		__this_cpu_dec(from->stat->count[MEM_CGROUP_STAT_FILE_MAPPED]);
		__this_cpu_inc(to->stat->count[MEM_CGROUP_STAT_FILE_MAPPED]);
		preempt_enable();
2239
	}
2240
	mem_cgroup_charge_statistics(from, PageCgroupCache(pc), -nr_pages);
2241 2242
	if (uncharge)
		/* This is not "cancel", but cancel_charge does all we need. */
2243
		__mem_cgroup_cancel_charge(from, nr_pages);
2244

2245
	/* caller should have done css_get */
K
KAMEZAWA Hiroyuki 已提交
2246
	pc->mem_cgroup = to;
2247
	mem_cgroup_charge_statistics(to, PageCgroupCache(pc), nr_pages);
2248 2249 2250
	/*
	 * We charges against "to" which may not have any tasks. Then, "to"
	 * can be under rmdir(). But in current implementation, caller of
2251 2252 2253
	 * this function is just force_empty() and move charge, so it's
	 * garanteed that "to" is never removed. So, we don't check rmdir
	 * status here.
2254
	 */
2255 2256 2257
	move_unlock_page_cgroup(pc, &flags);
	ret = 0;
unlock:
2258
	unlock_page_cgroup(pc);
2259 2260 2261
	/*
	 * check events
	 */
2262 2263
	memcg_check_events(to, page);
	memcg_check_events(from, page);
2264
out:
2265 2266 2267 2268 2269 2270 2271
	return ret;
}

/*
 * move charges to its parent.
 */

2272 2273
static int mem_cgroup_move_parent(struct page *page,
				  struct page_cgroup *pc,
2274 2275 2276 2277 2278 2279
				  struct mem_cgroup *child,
				  gfp_t gfp_mask)
{
	struct cgroup *cg = child->css.cgroup;
	struct cgroup *pcg = cg->parent;
	struct mem_cgroup *parent;
2280
	int page_size = PAGE_SIZE;
2281
	unsigned long flags;
2282 2283 2284 2285 2286 2287
	int ret;

	/* Is ROOT ? */
	if (!pcg)
		return -EINVAL;

2288 2289 2290 2291 2292
	ret = -EBUSY;
	if (!get_page_unless_zero(page))
		goto out;
	if (isolate_lru_page(page))
		goto put;
2293 2294 2295

	if (PageTransHuge(page))
		page_size = HPAGE_SIZE;
K
KAMEZAWA Hiroyuki 已提交
2296

2297
	parent = mem_cgroup_from_cont(pcg);
2298 2299
	ret = __mem_cgroup_try_charge(NULL, gfp_mask,
				&parent, false, page_size);
2300
	if (ret || !parent)
2301
		goto put_back;
2302

2303
	if (page_size > PAGE_SIZE)
2304 2305
		flags = compound_lock_irqsave(page);

2306
	ret = mem_cgroup_move_account(page, pc, child, parent, true, page_size);
2307
	if (ret)
2308
		__mem_cgroup_cancel_charge(parent, page_size >> PAGE_SHIFT);
2309

2310
	if (page_size > PAGE_SIZE)
2311
		compound_unlock_irqrestore(page, flags);
2312
put_back:
K
KAMEZAWA Hiroyuki 已提交
2313
	putback_lru_page(page);
2314
put:
2315
	put_page(page);
2316
out:
2317 2318 2319
	return ret;
}

2320 2321 2322 2323 2324 2325 2326
/*
 * Charge the memory controller for page usage.
 * Return
 * 0 if the charge was successful
 * < 0 if the cgroup is over its limit
 */
static int mem_cgroup_charge_common(struct page *page, struct mm_struct *mm,
2327
				gfp_t gfp_mask, enum charge_type ctype)
2328
{
2329
	struct mem_cgroup *mem = NULL;
2330
	int page_size = PAGE_SIZE;
2331
	struct page_cgroup *pc;
2332
	bool oom = true;
2333
	int ret;
A
Andrea Arcangeli 已提交
2334

A
Andrea Arcangeli 已提交
2335
	if (PageTransHuge(page)) {
A
Andrea Arcangeli 已提交
2336
		page_size <<= compound_order(page);
A
Andrea Arcangeli 已提交
2337
		VM_BUG_ON(!PageTransHuge(page));
2338 2339 2340 2341 2342
		/*
		 * Never OOM-kill a process for a huge page.  The
		 * fault handler will fall back to regular pages.
		 */
		oom = false;
A
Andrea Arcangeli 已提交
2343
	}
2344 2345

	pc = lookup_page_cgroup(page);
2346
	BUG_ON(!pc); /* XXX: remove this and move pc lookup into commit */
2347

2348
	ret = __mem_cgroup_try_charge(mm, gfp_mask, &mem, oom, page_size);
2349
	if (ret || !mem)
2350 2351
		return ret;

2352
	__mem_cgroup_commit_charge(mem, page, pc, ctype, page_size);
2353 2354 2355
	return 0;
}

2356 2357
int mem_cgroup_newpage_charge(struct page *page,
			      struct mm_struct *mm, gfp_t gfp_mask)
2358
{
2359
	if (mem_cgroup_disabled())
2360
		return 0;
2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371
	/*
	 * If already mapped, we don't have to account.
	 * If page cache, page->mapping has address_space.
	 * But page->mapping may have out-of-use anon_vma pointer,
	 * detecit it by PageAnon() check. newly-mapped-anon's page->mapping
	 * is NULL.
  	 */
	if (page_mapped(page) || (page->mapping && !PageAnon(page)))
		return 0;
	if (unlikely(!mm))
		mm = &init_mm;
2372
	return mem_cgroup_charge_common(page, mm, gfp_mask,
2373
				MEM_CGROUP_CHARGE_TYPE_MAPPED);
2374 2375
}

D
Daisuke Nishimura 已提交
2376 2377 2378 2379
static void
__mem_cgroup_commit_charge_swapin(struct page *page, struct mem_cgroup *ptr,
					enum charge_type ctype);

2380 2381
int mem_cgroup_cache_charge(struct page *page, struct mm_struct *mm,
				gfp_t gfp_mask)
2382
{
2383 2384
	int ret;

2385
	if (mem_cgroup_disabled())
2386
		return 0;
2387 2388
	if (PageCompound(page))
		return 0;
2389 2390 2391 2392 2393 2394 2395 2396
	/*
	 * Corner case handling. This is called from add_to_page_cache()
	 * in usual. But some FS (shmem) precharges this page before calling it
	 * and call add_to_page_cache() with GFP_NOWAIT.
	 *
	 * For GFP_NOWAIT case, the page may be pre-charged before calling
	 * add_to_page_cache(). (See shmem.c) check it here and avoid to call
	 * charge twice. (It works but has to pay a bit larger cost.)
2397 2398
	 * And when the page is SwapCache, it should take swap information
	 * into account. This is under lock_page() now.
2399 2400 2401 2402
	 */
	if (!(gfp_mask & __GFP_WAIT)) {
		struct page_cgroup *pc;

2403 2404 2405 2406 2407 2408
		pc = lookup_page_cgroup(page);
		if (!pc)
			return 0;
		lock_page_cgroup(pc);
		if (PageCgroupUsed(pc)) {
			unlock_page_cgroup(pc);
2409 2410
			return 0;
		}
2411
		unlock_page_cgroup(pc);
2412 2413
	}

2414
	if (unlikely(!mm))
2415
		mm = &init_mm;
2416

2417 2418
	if (page_is_file_cache(page))
		return mem_cgroup_charge_common(page, mm, gfp_mask,
2419
				MEM_CGROUP_CHARGE_TYPE_CACHE);
2420

D
Daisuke Nishimura 已提交
2421 2422
	/* shmem */
	if (PageSwapCache(page)) {
2423
		struct mem_cgroup *mem;
2424

D
Daisuke Nishimura 已提交
2425 2426 2427 2428 2429 2430
		ret = mem_cgroup_try_charge_swapin(mm, page, gfp_mask, &mem);
		if (!ret)
			__mem_cgroup_commit_charge_swapin(page, mem,
					MEM_CGROUP_CHARGE_TYPE_SHMEM);
	} else
		ret = mem_cgroup_charge_common(page, mm, gfp_mask,
2431
					MEM_CGROUP_CHARGE_TYPE_SHMEM);
2432 2433

	return ret;
2434 2435
}

2436 2437 2438
/*
 * While swap-in, try_charge -> commit or cancel, the page is locked.
 * And when try_charge() successfully returns, one refcnt to memcg without
2439
 * struct page_cgroup is acquired. This refcnt will be consumed by
2440 2441
 * "commit()" or removed by "cancel()"
 */
2442 2443 2444 2445 2446
int mem_cgroup_try_charge_swapin(struct mm_struct *mm,
				 struct page *page,
				 gfp_t mask, struct mem_cgroup **ptr)
{
	struct mem_cgroup *mem;
2447
	int ret;
2448

2449 2450
	*ptr = NULL;

2451
	if (mem_cgroup_disabled())
2452 2453 2454 2455 2456 2457
		return 0;

	if (!do_swap_account)
		goto charge_cur_mm;
	/*
	 * A racing thread's fault, or swapoff, may have already updated
H
Hugh Dickins 已提交
2458 2459 2460
	 * the pte, and even removed page from swap cache: in those cases
	 * do_swap_page()'s pte_same() test will fail; but there's also a
	 * KSM case which does need to charge the page.
2461 2462
	 */
	if (!PageSwapCache(page))
H
Hugh Dickins 已提交
2463
		goto charge_cur_mm;
2464
	mem = try_get_mem_cgroup_from_page(page);
2465 2466
	if (!mem)
		goto charge_cur_mm;
2467
	*ptr = mem;
A
Andrea Arcangeli 已提交
2468
	ret = __mem_cgroup_try_charge(NULL, mask, ptr, true, PAGE_SIZE);
2469 2470
	css_put(&mem->css);
	return ret;
2471 2472 2473
charge_cur_mm:
	if (unlikely(!mm))
		mm = &init_mm;
A
Andrea Arcangeli 已提交
2474
	return __mem_cgroup_try_charge(mm, mask, ptr, true, PAGE_SIZE);
2475 2476
}

D
Daisuke Nishimura 已提交
2477 2478 2479
static void
__mem_cgroup_commit_charge_swapin(struct page *page, struct mem_cgroup *ptr,
					enum charge_type ctype)
2480 2481 2482
{
	struct page_cgroup *pc;

2483
	if (mem_cgroup_disabled())
2484 2485 2486
		return;
	if (!ptr)
		return;
2487
	cgroup_exclude_rmdir(&ptr->css);
2488
	pc = lookup_page_cgroup(page);
2489
	mem_cgroup_lru_del_before_commit_swapcache(page);
2490
	__mem_cgroup_commit_charge(ptr, page, pc, ctype, PAGE_SIZE);
2491
	mem_cgroup_lru_add_after_commit_swapcache(page);
2492 2493 2494
	/*
	 * Now swap is on-memory. This means this page may be
	 * counted both as mem and swap....double count.
2495 2496 2497
	 * Fix it by uncharging from memsw. Basically, this SwapCache is stable
	 * under lock_page(). But in do_swap_page()::memory.c, reuse_swap_page()
	 * may call delete_from_swap_cache() before reach here.
2498
	 */
2499
	if (do_swap_account && PageSwapCache(page)) {
2500
		swp_entry_t ent = {.val = page_private(page)};
2501
		unsigned short id;
2502
		struct mem_cgroup *memcg;
2503 2504 2505 2506

		id = swap_cgroup_record(ent, 0);
		rcu_read_lock();
		memcg = mem_cgroup_lookup(id);
2507
		if (memcg) {
2508 2509 2510 2511
			/*
			 * This recorded memcg can be obsolete one. So, avoid
			 * calling css_tryget
			 */
2512
			if (!mem_cgroup_is_root(memcg))
2513
				res_counter_uncharge(&memcg->memsw, PAGE_SIZE);
2514
			mem_cgroup_swap_statistics(memcg, false);
2515 2516
			mem_cgroup_put(memcg);
		}
2517
		rcu_read_unlock();
2518
	}
2519 2520 2521 2522 2523 2524
	/*
	 * At swapin, we may charge account against cgroup which has no tasks.
	 * So, rmdir()->pre_destroy() can be called while we do this charge.
	 * In that case, we need to call pre_destroy() again. check it here.
	 */
	cgroup_release_and_wakeup_rmdir(&ptr->css);
2525 2526
}

D
Daisuke Nishimura 已提交
2527 2528 2529 2530 2531 2532
void mem_cgroup_commit_charge_swapin(struct page *page, struct mem_cgroup *ptr)
{
	__mem_cgroup_commit_charge_swapin(page, ptr,
					MEM_CGROUP_CHARGE_TYPE_MAPPED);
}

2533 2534
void mem_cgroup_cancel_charge_swapin(struct mem_cgroup *mem)
{
2535
	if (mem_cgroup_disabled())
2536 2537 2538
		return;
	if (!mem)
		return;
2539
	__mem_cgroup_cancel_charge(mem, 1);
2540 2541
}

2542
static void
A
Andrea Arcangeli 已提交
2543 2544
__do_uncharge(struct mem_cgroup *mem, const enum charge_type ctype,
	      int page_size)
2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559
{
	struct memcg_batch_info *batch = NULL;
	bool uncharge_memsw = true;
	/* If swapout, usage of swap doesn't decrease */
	if (!do_swap_account || ctype == MEM_CGROUP_CHARGE_TYPE_SWAPOUT)
		uncharge_memsw = false;

	batch = &current->memcg_batch;
	/*
	 * In usual, we do css_get() when we remember memcg pointer.
	 * But in this case, we keep res->usage until end of a series of
	 * uncharges. Then, it's ok to ignore memcg's refcnt.
	 */
	if (!batch->memcg)
		batch->memcg = mem;
2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570
	/*
	 * do_batch > 0 when unmapping pages or inode invalidate/truncate.
	 * In those cases, all pages freed continously can be expected to be in
	 * the same cgroup and we have chance to coalesce uncharges.
	 * But we do uncharge one by one if this is killed by OOM(TIF_MEMDIE)
	 * because we want to do uncharge as soon as possible.
	 */

	if (!batch->do_batch || test_thread_flag(TIF_MEMDIE))
		goto direct_uncharge;

A
Andrea Arcangeli 已提交
2571 2572 2573
	if (page_size != PAGE_SIZE)
		goto direct_uncharge;

2574 2575 2576 2577 2578 2579 2580 2581
	/*
	 * In typical case, batch->memcg == mem. This means we can
	 * merge a series of uncharges to an uncharge of res_counter.
	 * If not, we uncharge res_counter ony by one.
	 */
	if (batch->memcg != mem)
		goto direct_uncharge;
	/* remember freed charge and uncharge it later */
2582
	batch->nr_pages++;
2583
	if (uncharge_memsw)
2584
		batch->memsw_nr_pages++;
2585 2586
	return;
direct_uncharge:
A
Andrea Arcangeli 已提交
2587
	res_counter_uncharge(&mem->res, page_size);
2588
	if (uncharge_memsw)
A
Andrea Arcangeli 已提交
2589
		res_counter_uncharge(&mem->memsw, page_size);
2590 2591
	if (unlikely(batch->memcg != mem))
		memcg_oom_recover(mem);
2592 2593
	return;
}
2594

2595
/*
2596
 * uncharge if !page_mapped(page)
2597
 */
2598
static struct mem_cgroup *
2599
__mem_cgroup_uncharge_common(struct page *page, enum charge_type ctype)
2600
{
2601
	int count;
H
Hugh Dickins 已提交
2602
	struct page_cgroup *pc;
2603
	struct mem_cgroup *mem = NULL;
A
Andrea Arcangeli 已提交
2604
	int page_size = PAGE_SIZE;
2605

2606
	if (mem_cgroup_disabled())
2607
		return NULL;
2608

K
KAMEZAWA Hiroyuki 已提交
2609
	if (PageSwapCache(page))
2610
		return NULL;
K
KAMEZAWA Hiroyuki 已提交
2611

A
Andrea Arcangeli 已提交
2612
	if (PageTransHuge(page)) {
A
Andrea Arcangeli 已提交
2613
		page_size <<= compound_order(page);
A
Andrea Arcangeli 已提交
2614 2615
		VM_BUG_ON(!PageTransHuge(page));
	}
A
Andrea Arcangeli 已提交
2616

2617
	count = page_size >> PAGE_SHIFT;
2618
	/*
2619
	 * Check if our page_cgroup is valid
2620
	 */
2621 2622
	pc = lookup_page_cgroup(page);
	if (unlikely(!pc || !PageCgroupUsed(pc)))
2623
		return NULL;
2624

2625
	lock_page_cgroup(pc);
K
KAMEZAWA Hiroyuki 已提交
2626

2627 2628
	mem = pc->mem_cgroup;

K
KAMEZAWA Hiroyuki 已提交
2629 2630 2631 2632 2633
	if (!PageCgroupUsed(pc))
		goto unlock_out;

	switch (ctype) {
	case MEM_CGROUP_CHARGE_TYPE_MAPPED:
K
KAMEZAWA Hiroyuki 已提交
2634
	case MEM_CGROUP_CHARGE_TYPE_DROP:
2635 2636
		/* See mem_cgroup_prepare_migration() */
		if (page_mapped(page) || PageCgroupMigration(pc))
K
KAMEZAWA Hiroyuki 已提交
2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647
			goto unlock_out;
		break;
	case MEM_CGROUP_CHARGE_TYPE_SWAPOUT:
		if (!PageAnon(page)) {	/* Shared memory */
			if (page->mapping && !page_is_file_cache(page))
				goto unlock_out;
		} else if (page_mapped(page)) /* Anon */
				goto unlock_out;
		break;
	default:
		break;
2648
	}
K
KAMEZAWA Hiroyuki 已提交
2649

2650
	mem_cgroup_charge_statistics(mem, PageCgroupCache(pc), -count);
K
KAMEZAWA Hiroyuki 已提交
2651

2652
	ClearPageCgroupUsed(pc);
2653 2654 2655 2656 2657 2658
	/*
	 * pc->mem_cgroup is not cleared here. It will be accessed when it's
	 * freed from LRU. This is safe because uncharged page is expected not
	 * to be reused (freed soon). Exception is SwapCache, it's handled by
	 * special functions.
	 */
2659

2660
	unlock_page_cgroup(pc);
K
KAMEZAWA Hiroyuki 已提交
2661 2662 2663 2664
	/*
	 * even after unlock, we have mem->res.usage here and this memcg
	 * will never be freed.
	 */
2665
	memcg_check_events(mem, page);
K
KAMEZAWA Hiroyuki 已提交
2666 2667 2668 2669 2670
	if (do_swap_account && ctype == MEM_CGROUP_CHARGE_TYPE_SWAPOUT) {
		mem_cgroup_swap_statistics(mem, true);
		mem_cgroup_get(mem);
	}
	if (!mem_cgroup_is_root(mem))
A
Andrea Arcangeli 已提交
2671
		__do_uncharge(mem, ctype, page_size);
2672

2673
	return mem;
K
KAMEZAWA Hiroyuki 已提交
2674 2675 2676

unlock_out:
	unlock_page_cgroup(pc);
2677
	return NULL;
2678 2679
}

2680 2681
void mem_cgroup_uncharge_page(struct page *page)
{
2682 2683 2684 2685 2686
	/* early check. */
	if (page_mapped(page))
		return;
	if (page->mapping && !PageAnon(page))
		return;
2687 2688 2689 2690 2691 2692
	__mem_cgroup_uncharge_common(page, MEM_CGROUP_CHARGE_TYPE_MAPPED);
}

void mem_cgroup_uncharge_cache_page(struct page *page)
{
	VM_BUG_ON(page_mapped(page));
2693
	VM_BUG_ON(page->mapping);
2694 2695 2696
	__mem_cgroup_uncharge_common(page, MEM_CGROUP_CHARGE_TYPE_CACHE);
}

2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710
/*
 * Batch_start/batch_end is called in unmap_page_range/invlidate/trucate.
 * In that cases, pages are freed continuously and we can expect pages
 * are in the same memcg. All these calls itself limits the number of
 * pages freed at once, then uncharge_start/end() is called properly.
 * This may be called prural(2) times in a context,
 */

void mem_cgroup_uncharge_start(void)
{
	current->memcg_batch.do_batch++;
	/* We can do nest. */
	if (current->memcg_batch.do_batch == 1) {
		current->memcg_batch.memcg = NULL;
2711 2712
		current->memcg_batch.nr_pages = 0;
		current->memcg_batch.memsw_nr_pages = 0;
2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732
	}
}

void mem_cgroup_uncharge_end(void)
{
	struct memcg_batch_info *batch = &current->memcg_batch;

	if (!batch->do_batch)
		return;

	batch->do_batch--;
	if (batch->do_batch) /* If stacked, do nothing. */
		return;

	if (!batch->memcg)
		return;
	/*
	 * This "batch->memcg" is valid without any css_get/put etc...
	 * bacause we hide charges behind us.
	 */
2733 2734 2735 2736 2737 2738
	if (batch->nr_pages)
		res_counter_uncharge(&batch->memcg->res,
				     batch->nr_pages * PAGE_SIZE);
	if (batch->memsw_nr_pages)
		res_counter_uncharge(&batch->memcg->memsw,
				     batch->memsw_nr_pages * PAGE_SIZE);
2739
	memcg_oom_recover(batch->memcg);
2740 2741 2742 2743
	/* forget this pointer (for sanity check) */
	batch->memcg = NULL;
}

2744
#ifdef CONFIG_SWAP
2745
/*
2746
 * called after __delete_from_swap_cache() and drop "page" account.
2747 2748
 * memcg information is recorded to swap_cgroup of "ent"
 */
K
KAMEZAWA Hiroyuki 已提交
2749 2750
void
mem_cgroup_uncharge_swapcache(struct page *page, swp_entry_t ent, bool swapout)
2751 2752
{
	struct mem_cgroup *memcg;
K
KAMEZAWA Hiroyuki 已提交
2753 2754 2755 2756 2757 2758
	int ctype = MEM_CGROUP_CHARGE_TYPE_SWAPOUT;

	if (!swapout) /* this was a swap cache but the swap is unused ! */
		ctype = MEM_CGROUP_CHARGE_TYPE_DROP;

	memcg = __mem_cgroup_uncharge_common(page, ctype);
2759

K
KAMEZAWA Hiroyuki 已提交
2760 2761 2762 2763 2764
	/*
	 * record memcg information,  if swapout && memcg != NULL,
	 * mem_cgroup_get() was called in uncharge().
	 */
	if (do_swap_account && swapout && memcg)
2765
		swap_cgroup_record(ent, css_id(&memcg->css));
2766
}
2767
#endif
2768 2769 2770 2771 2772 2773 2774

#ifdef CONFIG_CGROUP_MEM_RES_CTLR_SWAP
/*
 * called from swap_entry_free(). remove record in swap_cgroup and
 * uncharge "memsw" account.
 */
void mem_cgroup_uncharge_swap(swp_entry_t ent)
K
KAMEZAWA Hiroyuki 已提交
2775
{
2776
	struct mem_cgroup *memcg;
2777
	unsigned short id;
2778 2779 2780 2781

	if (!do_swap_account)
		return;

2782 2783 2784
	id = swap_cgroup_record(ent, 0);
	rcu_read_lock();
	memcg = mem_cgroup_lookup(id);
2785
	if (memcg) {
2786 2787 2788 2789
		/*
		 * We uncharge this because swap is freed.
		 * This memcg can be obsolete one. We avoid calling css_tryget
		 */
2790
		if (!mem_cgroup_is_root(memcg))
2791
			res_counter_uncharge(&memcg->memsw, PAGE_SIZE);
2792
		mem_cgroup_swap_statistics(memcg, false);
2793 2794
		mem_cgroup_put(memcg);
	}
2795
	rcu_read_unlock();
K
KAMEZAWA Hiroyuki 已提交
2796
}
2797 2798 2799 2800 2801 2802

/**
 * mem_cgroup_move_swap_account - move swap charge and swap_cgroup's record.
 * @entry: swap entry to be moved
 * @from:  mem_cgroup which the entry is moved from
 * @to:  mem_cgroup which the entry is moved to
2803
 * @need_fixup: whether we should fixup res_counters and refcounts.
2804 2805 2806 2807 2808 2809 2810 2811 2812 2813
 *
 * It succeeds only when the swap_cgroup's record for this entry is the same
 * as the mem_cgroup's id of @from.
 *
 * Returns 0 on success, -EINVAL on failure.
 *
 * The caller must have charged to @to, IOW, called res_counter_charge() about
 * both res and memsw, and called css_get().
 */
static int mem_cgroup_move_swap_account(swp_entry_t entry,
2814
		struct mem_cgroup *from, struct mem_cgroup *to, bool need_fixup)
2815 2816 2817 2818 2819 2820 2821 2822
{
	unsigned short old_id, new_id;

	old_id = css_id(&from->css);
	new_id = css_id(&to->css);

	if (swap_cgroup_cmpxchg(entry, old_id, new_id) == old_id) {
		mem_cgroup_swap_statistics(from, false);
2823
		mem_cgroup_swap_statistics(to, true);
2824
		/*
2825 2826 2827 2828 2829 2830
		 * This function is only called from task migration context now.
		 * It postpones res_counter and refcount handling till the end
		 * of task migration(mem_cgroup_clear_mc()) for performance
		 * improvement. But we cannot postpone mem_cgroup_get(to)
		 * because if the process that has been moved to @to does
		 * swap-in, the refcount of @to might be decreased to 0.
2831 2832
		 */
		mem_cgroup_get(to);
2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843
		if (need_fixup) {
			if (!mem_cgroup_is_root(from))
				res_counter_uncharge(&from->memsw, PAGE_SIZE);
			mem_cgroup_put(from);
			/*
			 * we charged both to->res and to->memsw, so we should
			 * uncharge to->res.
			 */
			if (!mem_cgroup_is_root(to))
				res_counter_uncharge(&to->res, PAGE_SIZE);
		}
2844 2845 2846 2847 2848 2849
		return 0;
	}
	return -EINVAL;
}
#else
static inline int mem_cgroup_move_swap_account(swp_entry_t entry,
2850
		struct mem_cgroup *from, struct mem_cgroup *to, bool need_fixup)
2851 2852 2853
{
	return -EINVAL;
}
2854
#endif
K
KAMEZAWA Hiroyuki 已提交
2855

2856
/*
2857 2858
 * Before starting migration, account PAGE_SIZE to mem_cgroup that the old
 * page belongs to.
2859
 */
2860
int mem_cgroup_prepare_migration(struct page *page,
2861
	struct page *newpage, struct mem_cgroup **ptr, gfp_t gfp_mask)
2862 2863
{
	struct page_cgroup *pc;
2864
	struct mem_cgroup *mem = NULL;
2865
	enum charge_type ctype;
2866
	int ret = 0;
2867

2868 2869
	*ptr = NULL;

A
Andrea Arcangeli 已提交
2870
	VM_BUG_ON(PageTransHuge(page));
2871
	if (mem_cgroup_disabled())
2872 2873
		return 0;

2874 2875 2876
	pc = lookup_page_cgroup(page);
	lock_page_cgroup(pc);
	if (PageCgroupUsed(pc)) {
2877 2878
		mem = pc->mem_cgroup;
		css_get(&mem->css);
2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909
		/*
		 * At migrating an anonymous page, its mapcount goes down
		 * to 0 and uncharge() will be called. But, even if it's fully
		 * unmapped, migration may fail and this page has to be
		 * charged again. We set MIGRATION flag here and delay uncharge
		 * until end_migration() is called
		 *
		 * Corner Case Thinking
		 * A)
		 * When the old page was mapped as Anon and it's unmap-and-freed
		 * while migration was ongoing.
		 * If unmap finds the old page, uncharge() of it will be delayed
		 * until end_migration(). If unmap finds a new page, it's
		 * uncharged when it make mapcount to be 1->0. If unmap code
		 * finds swap_migration_entry, the new page will not be mapped
		 * and end_migration() will find it(mapcount==0).
		 *
		 * B)
		 * When the old page was mapped but migraion fails, the kernel
		 * remaps it. A charge for it is kept by MIGRATION flag even
		 * if mapcount goes down to 0. We can do remap successfully
		 * without charging it again.
		 *
		 * C)
		 * The "old" page is under lock_page() until the end of
		 * migration, so, the old page itself will not be swapped-out.
		 * If the new page is swapped out before end_migraton, our
		 * hook to usual swap-out path will catch the event.
		 */
		if (PageAnon(page))
			SetPageCgroupMigration(pc);
2910
	}
2911
	unlock_page_cgroup(pc);
2912 2913 2914 2915 2916 2917
	/*
	 * If the page is not charged at this point,
	 * we return here.
	 */
	if (!mem)
		return 0;
2918

A
Andrea Arcangeli 已提交
2919
	*ptr = mem;
2920
	ret = __mem_cgroup_try_charge(NULL, gfp_mask, ptr, false, PAGE_SIZE);
2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932
	css_put(&mem->css);/* drop extra refcnt */
	if (ret || *ptr == NULL) {
		if (PageAnon(page)) {
			lock_page_cgroup(pc);
			ClearPageCgroupMigration(pc);
			unlock_page_cgroup(pc);
			/*
			 * The old page may be fully unmapped while we kept it.
			 */
			mem_cgroup_uncharge_page(page);
		}
		return -ENOMEM;
2933
	}
2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946
	/*
	 * We charge new page before it's used/mapped. So, even if unlock_page()
	 * is called before end_migration, we can catch all events on this new
	 * page. In the case new page is migrated but not remapped, new page's
	 * mapcount will be finally 0 and we call uncharge in end_migration().
	 */
	pc = lookup_page_cgroup(newpage);
	if (PageAnon(page))
		ctype = MEM_CGROUP_CHARGE_TYPE_MAPPED;
	else if (page_is_file_cache(page))
		ctype = MEM_CGROUP_CHARGE_TYPE_CACHE;
	else
		ctype = MEM_CGROUP_CHARGE_TYPE_SHMEM;
2947
	__mem_cgroup_commit_charge(mem, page, pc, ctype, PAGE_SIZE);
2948
	return ret;
2949
}
2950

2951
/* remove redundant charge if migration failed*/
2952
void mem_cgroup_end_migration(struct mem_cgroup *mem,
2953
	struct page *oldpage, struct page *newpage, bool migration_ok)
2954
{
2955
	struct page *used, *unused;
2956 2957 2958 2959
	struct page_cgroup *pc;

	if (!mem)
		return;
2960
	/* blocks rmdir() */
2961
	cgroup_exclude_rmdir(&mem->css);
2962
	if (!migration_ok) {
2963 2964
		used = oldpage;
		unused = newpage;
2965
	} else {
2966
		used = newpage;
2967 2968
		unused = oldpage;
	}
2969
	/*
2970 2971 2972
	 * We disallowed uncharge of pages under migration because mapcount
	 * of the page goes down to zero, temporarly.
	 * Clear the flag and check the page should be charged.
2973
	 */
2974 2975 2976 2977
	pc = lookup_page_cgroup(oldpage);
	lock_page_cgroup(pc);
	ClearPageCgroupMigration(pc);
	unlock_page_cgroup(pc);
2978

2979 2980
	__mem_cgroup_uncharge_common(unused, MEM_CGROUP_CHARGE_TYPE_FORCE);

2981
	/*
2982 2983 2984 2985 2986 2987
	 * If a page is a file cache, radix-tree replacement is very atomic
	 * and we can skip this check. When it was an Anon page, its mapcount
	 * goes down to 0. But because we added MIGRATION flage, it's not
	 * uncharged yet. There are several case but page->mapcount check
	 * and USED bit check in mem_cgroup_uncharge_page() will do enough
	 * check. (see prepare_charge() also)
2988
	 */
2989 2990
	if (PageAnon(used))
		mem_cgroup_uncharge_page(used);
2991
	/*
2992 2993
	 * At migration, we may charge account against cgroup which has no
	 * tasks.
2994 2995 2996 2997
	 * So, rmdir()->pre_destroy() can be called while we do this charge.
	 * In that case, we need to call pre_destroy() again. check it here.
	 */
	cgroup_release_and_wakeup_rmdir(&mem->css);
2998
}
2999

3000
/*
3001 3002 3003 3004 3005 3006
 * A call to try to shrink memory usage on charge failure at shmem's swapin.
 * Calling hierarchical_reclaim is not enough because we should update
 * last_oom_jiffies to prevent pagefault_out_of_memory from invoking global OOM.
 * Moreover considering hierarchy, we should reclaim from the mem_over_limit,
 * not from the memcg which this page would be charged to.
 * try_charge_swapin does all of these works properly.
3007
 */
3008
int mem_cgroup_shmem_charge_fallback(struct page *page,
3009 3010
			    struct mm_struct *mm,
			    gfp_t gfp_mask)
3011
{
3012
	struct mem_cgroup *mem;
3013
	int ret;
3014

3015
	if (mem_cgroup_disabled())
3016
		return 0;
3017

3018 3019 3020
	ret = mem_cgroup_try_charge_swapin(mm, page, gfp_mask, &mem);
	if (!ret)
		mem_cgroup_cancel_charge_swapin(mem); /* it does !mem check */
3021

3022
	return ret;
3023 3024
}

3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070
#ifdef CONFIG_DEBUG_VM
static struct page_cgroup *lookup_page_cgroup_used(struct page *page)
{
	struct page_cgroup *pc;

	pc = lookup_page_cgroup(page);
	if (likely(pc) && PageCgroupUsed(pc))
		return pc;
	return NULL;
}

bool mem_cgroup_bad_page_check(struct page *page)
{
	if (mem_cgroup_disabled())
		return false;

	return lookup_page_cgroup_used(page) != NULL;
}

void mem_cgroup_print_bad_page(struct page *page)
{
	struct page_cgroup *pc;

	pc = lookup_page_cgroup_used(page);
	if (pc) {
		int ret = -1;
		char *path;

		printk(KERN_ALERT "pc:%p pc->flags:%lx pc->mem_cgroup:%p",
		       pc, pc->flags, pc->mem_cgroup);

		path = kmalloc(PATH_MAX, GFP_KERNEL);
		if (path) {
			rcu_read_lock();
			ret = cgroup_path(pc->mem_cgroup->css.cgroup,
							path, PATH_MAX);
			rcu_read_unlock();
		}

		printk(KERN_CONT "(%s)\n",
				(ret < 0) ? "cannot get the path" : path);
		kfree(path);
	}
}
#endif

3071 3072
static DEFINE_MUTEX(set_limit_mutex);

3073
static int mem_cgroup_resize_limit(struct mem_cgroup *memcg,
3074
				unsigned long long val)
3075
{
3076
	int retry_count;
3077
	u64 memswlimit, memlimit;
3078
	int ret = 0;
3079 3080
	int children = mem_cgroup_count_children(memcg);
	u64 curusage, oldusage;
3081
	int enlarge;
3082 3083 3084 3085 3086 3087 3088 3089 3090

	/*
	 * For keeping hierarchical_reclaim simple, how long we should retry
	 * is depends on callers. We set our retry-count to be function
	 * of # of children which we should visit in this loop.
	 */
	retry_count = MEM_CGROUP_RECLAIM_RETRIES * children;

	oldusage = res_counter_read_u64(&memcg->res, RES_USAGE);
3091

3092
	enlarge = 0;
3093
	while (retry_count) {
3094 3095 3096 3097
		if (signal_pending(current)) {
			ret = -EINTR;
			break;
		}
3098 3099 3100 3101 3102 3103 3104 3105 3106 3107
		/*
		 * Rather than hide all in some function, I do this in
		 * open coded manner. You see what this really does.
		 * We have to guarantee mem->res.limit < mem->memsw.limit.
		 */
		mutex_lock(&set_limit_mutex);
		memswlimit = res_counter_read_u64(&memcg->memsw, RES_LIMIT);
		if (memswlimit < val) {
			ret = -EINVAL;
			mutex_unlock(&set_limit_mutex);
3108 3109
			break;
		}
3110 3111 3112 3113 3114

		memlimit = res_counter_read_u64(&memcg->res, RES_LIMIT);
		if (memlimit < val)
			enlarge = 1;

3115
		ret = res_counter_set_limit(&memcg->res, val);
3116 3117 3118 3119 3120 3121
		if (!ret) {
			if (memswlimit == val)
				memcg->memsw_is_minimum = true;
			else
				memcg->memsw_is_minimum = false;
		}
3122 3123 3124 3125 3126
		mutex_unlock(&set_limit_mutex);

		if (!ret)
			break;

3127
		mem_cgroup_hierarchical_reclaim(memcg, NULL, GFP_KERNEL,
3128
						MEM_CGROUP_RECLAIM_SHRINK);
3129 3130 3131 3132 3133 3134
		curusage = res_counter_read_u64(&memcg->res, RES_USAGE);
		/* Usage is reduced ? */
  		if (curusage >= oldusage)
			retry_count--;
		else
			oldusage = curusage;
3135
	}
3136 3137
	if (!ret && enlarge)
		memcg_oom_recover(memcg);
3138

3139 3140 3141
	return ret;
}

L
Li Zefan 已提交
3142 3143
static int mem_cgroup_resize_memsw_limit(struct mem_cgroup *memcg,
					unsigned long long val)
3144
{
3145
	int retry_count;
3146
	u64 memlimit, memswlimit, oldusage, curusage;
3147 3148
	int children = mem_cgroup_count_children(memcg);
	int ret = -EBUSY;
3149
	int enlarge = 0;
3150

3151 3152 3153
	/* see mem_cgroup_resize_res_limit */
 	retry_count = children * MEM_CGROUP_RECLAIM_RETRIES;
	oldusage = res_counter_read_u64(&memcg->memsw, RES_USAGE);
3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170
	while (retry_count) {
		if (signal_pending(current)) {
			ret = -EINTR;
			break;
		}
		/*
		 * Rather than hide all in some function, I do this in
		 * open coded manner. You see what this really does.
		 * We have to guarantee mem->res.limit < mem->memsw.limit.
		 */
		mutex_lock(&set_limit_mutex);
		memlimit = res_counter_read_u64(&memcg->res, RES_LIMIT);
		if (memlimit > val) {
			ret = -EINVAL;
			mutex_unlock(&set_limit_mutex);
			break;
		}
3171 3172 3173
		memswlimit = res_counter_read_u64(&memcg->memsw, RES_LIMIT);
		if (memswlimit < val)
			enlarge = 1;
3174
		ret = res_counter_set_limit(&memcg->memsw, val);
3175 3176 3177 3178 3179 3180
		if (!ret) {
			if (memlimit == val)
				memcg->memsw_is_minimum = true;
			else
				memcg->memsw_is_minimum = false;
		}
3181 3182 3183 3184 3185
		mutex_unlock(&set_limit_mutex);

		if (!ret)
			break;

3186
		mem_cgroup_hierarchical_reclaim(memcg, NULL, GFP_KERNEL,
3187 3188
						MEM_CGROUP_RECLAIM_NOSWAP |
						MEM_CGROUP_RECLAIM_SHRINK);
3189
		curusage = res_counter_read_u64(&memcg->memsw, RES_USAGE);
3190
		/* Usage is reduced ? */
3191
		if (curusage >= oldusage)
3192
			retry_count--;
3193 3194
		else
			oldusage = curusage;
3195
	}
3196 3197
	if (!ret && enlarge)
		memcg_oom_recover(memcg);
3198 3199 3200
	return ret;
}

3201
unsigned long mem_cgroup_soft_limit_reclaim(struct zone *zone, int order,
3202
					    gfp_t gfp_mask)
3203 3204 3205 3206 3207 3208
{
	unsigned long nr_reclaimed = 0;
	struct mem_cgroup_per_zone *mz, *next_mz = NULL;
	unsigned long reclaimed;
	int loop = 0;
	struct mem_cgroup_tree_per_zone *mctz;
3209
	unsigned long long excess;
3210 3211 3212 3213

	if (order > 0)
		return 0;

3214
	mctz = soft_limit_tree_node_zone(zone_to_nid(zone), zone_idx(zone));
3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261
	/*
	 * This loop can run a while, specially if mem_cgroup's continuously
	 * keep exceeding their soft limit and putting the system under
	 * pressure
	 */
	do {
		if (next_mz)
			mz = next_mz;
		else
			mz = mem_cgroup_largest_soft_limit_node(mctz);
		if (!mz)
			break;

		reclaimed = mem_cgroup_hierarchical_reclaim(mz->mem, zone,
						gfp_mask,
						MEM_CGROUP_RECLAIM_SOFT);
		nr_reclaimed += reclaimed;
		spin_lock(&mctz->lock);

		/*
		 * If we failed to reclaim anything from this memory cgroup
		 * it is time to move on to the next cgroup
		 */
		next_mz = NULL;
		if (!reclaimed) {
			do {
				/*
				 * Loop until we find yet another one.
				 *
				 * By the time we get the soft_limit lock
				 * again, someone might have aded the
				 * group back on the RB tree. Iterate to
				 * make sure we get a different mem.
				 * mem_cgroup_largest_soft_limit_node returns
				 * NULL if no other cgroup is present on
				 * the tree
				 */
				next_mz =
				__mem_cgroup_largest_soft_limit_node(mctz);
				if (next_mz == mz) {
					css_put(&next_mz->mem->css);
					next_mz = NULL;
				} else /* next_mz == NULL or other memcg */
					break;
			} while (1);
		}
		__mem_cgroup_remove_exceeded(mz->mem, mz, mctz);
3262
		excess = res_counter_soft_limit_excess(&mz->mem->res);
3263 3264 3265 3266 3267 3268 3269 3270
		/*
		 * One school of thought says that we should not add
		 * back the node to the tree if reclaim returns 0.
		 * But our reclaim could return 0, simply because due
		 * to priority we are exposing a smaller subset of
		 * memory to reclaim from. Consider this as a longer
		 * term TODO.
		 */
3271 3272
		/* If excess == 0, no tree ops */
		__mem_cgroup_insert_exceeded(mz->mem, mz, mctz, excess);
3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290
		spin_unlock(&mctz->lock);
		css_put(&mz->mem->css);
		loop++;
		/*
		 * Could not reclaim anything and there are no more
		 * mem cgroups to try or we seem to be looping without
		 * reclaiming anything.
		 */
		if (!nr_reclaimed &&
			(next_mz == NULL ||
			loop > MEM_CGROUP_MAX_SOFT_LIMIT_RECLAIM_LOOPS))
			break;
	} while (!nr_reclaimed);
	if (next_mz)
		css_put(&next_mz->mem->css);
	return nr_reclaimed;
}

3291 3292 3293 3294
/*
 * This routine traverse page_cgroup in given list and drop them all.
 * *And* this routine doesn't reclaim page itself, just removes page_cgroup.
 */
3295
static int mem_cgroup_force_empty_list(struct mem_cgroup *mem,
K
KAMEZAWA Hiroyuki 已提交
3296
				int node, int zid, enum lru_list lru)
3297
{
K
KAMEZAWA Hiroyuki 已提交
3298 3299
	struct zone *zone;
	struct mem_cgroup_per_zone *mz;
3300
	struct page_cgroup *pc, *busy;
K
KAMEZAWA Hiroyuki 已提交
3301
	unsigned long flags, loop;
3302
	struct list_head *list;
3303
	int ret = 0;
3304

K
KAMEZAWA Hiroyuki 已提交
3305 3306
	zone = &NODE_DATA(node)->node_zones[zid];
	mz = mem_cgroup_zoneinfo(mem, node, zid);
3307
	list = &mz->lists[lru];
3308

3309 3310 3311 3312 3313
	loop = MEM_CGROUP_ZSTAT(mz, lru);
	/* give some margin against EBUSY etc...*/
	loop += 256;
	busy = NULL;
	while (loop--) {
3314 3315
		struct page *page;

3316
		ret = 0;
K
KAMEZAWA Hiroyuki 已提交
3317
		spin_lock_irqsave(&zone->lru_lock, flags);
3318
		if (list_empty(list)) {
K
KAMEZAWA Hiroyuki 已提交
3319
			spin_unlock_irqrestore(&zone->lru_lock, flags);
3320
			break;
3321 3322 3323 3324
		}
		pc = list_entry(list->prev, struct page_cgroup, lru);
		if (busy == pc) {
			list_move(&pc->lru, list);
3325
			busy = NULL;
K
KAMEZAWA Hiroyuki 已提交
3326
			spin_unlock_irqrestore(&zone->lru_lock, flags);
3327 3328
			continue;
		}
K
KAMEZAWA Hiroyuki 已提交
3329
		spin_unlock_irqrestore(&zone->lru_lock, flags);
3330

3331
		page = lookup_cgroup_page(pc);
3332 3333

		ret = mem_cgroup_move_parent(page, pc, mem, GFP_KERNEL);
3334
		if (ret == -ENOMEM)
3335
			break;
3336 3337 3338 3339 3340 3341 3342

		if (ret == -EBUSY || ret == -EINVAL) {
			/* found lock contention or "pc" is obsolete. */
			busy = pc;
			cond_resched();
		} else
			busy = NULL;
3343
	}
K
KAMEZAWA Hiroyuki 已提交
3344

3345 3346 3347
	if (!ret && !list_empty(list))
		return -EBUSY;
	return ret;
3348 3349 3350 3351 3352 3353
}

/*
 * make mem_cgroup's charge to be 0 if there is no task.
 * This enables deleting this mem_cgroup.
 */
3354
static int mem_cgroup_force_empty(struct mem_cgroup *mem, bool free_all)
3355
{
3356 3357 3358
	int ret;
	int node, zid, shrink;
	int nr_retries = MEM_CGROUP_RECLAIM_RETRIES;
3359
	struct cgroup *cgrp = mem->css.cgroup;
3360

3361
	css_get(&mem->css);
3362 3363

	shrink = 0;
3364 3365 3366
	/* should free all ? */
	if (free_all)
		goto try_to_free;
3367
move_account:
3368
	do {
3369
		ret = -EBUSY;
3370 3371 3372 3373
		if (cgroup_task_count(cgrp) || !list_empty(&cgrp->children))
			goto out;
		ret = -EINTR;
		if (signal_pending(current))
3374
			goto out;
3375 3376
		/* This is for making all *used* pages to be on LRU. */
		lru_add_drain_all();
3377
		drain_all_stock_sync();
3378
		ret = 0;
3379
		mem_cgroup_start_move(mem);
3380
		for_each_node_state(node, N_HIGH_MEMORY) {
3381
			for (zid = 0; !ret && zid < MAX_NR_ZONES; zid++) {
3382
				enum lru_list l;
3383 3384
				for_each_lru(l) {
					ret = mem_cgroup_force_empty_list(mem,
K
KAMEZAWA Hiroyuki 已提交
3385
							node, zid, l);
3386 3387 3388
					if (ret)
						break;
				}
3389
			}
3390 3391 3392
			if (ret)
				break;
		}
3393
		mem_cgroup_end_move(mem);
3394
		memcg_oom_recover(mem);
3395 3396 3397
		/* it seems parent cgroup doesn't have enough mem */
		if (ret == -ENOMEM)
			goto try_to_free;
3398
		cond_resched();
3399 3400
	/* "ret" should also be checked to ensure all lists are empty. */
	} while (mem->res.usage > 0 || ret);
3401 3402 3403
out:
	css_put(&mem->css);
	return ret;
3404 3405

try_to_free:
3406 3407
	/* returns EBUSY if there is a task or if we come here twice. */
	if (cgroup_task_count(cgrp) || !list_empty(&cgrp->children) || shrink) {
3408 3409 3410
		ret = -EBUSY;
		goto out;
	}
3411 3412
	/* we call try-to-free pages for make this cgroup empty */
	lru_add_drain_all();
3413 3414 3415 3416
	/* try to free all pages in this cgroup */
	shrink = 1;
	while (nr_retries && mem->res.usage > 0) {
		int progress;
3417 3418 3419 3420 3421

		if (signal_pending(current)) {
			ret = -EINTR;
			goto out;
		}
K
KOSAKI Motohiro 已提交
3422 3423
		progress = try_to_free_mem_cgroup_pages(mem, GFP_KERNEL,
						false, get_swappiness(mem));
3424
		if (!progress) {
3425
			nr_retries--;
3426
			/* maybe some writeback is necessary */
3427
			congestion_wait(BLK_RW_ASYNC, HZ/10);
3428
		}
3429 3430

	}
K
KAMEZAWA Hiroyuki 已提交
3431
	lru_add_drain();
3432
	/* try move_account...there may be some *locked* pages. */
3433
	goto move_account;
3434 3435
}

3436 3437 3438 3439 3440 3441
int mem_cgroup_force_empty_write(struct cgroup *cont, unsigned int event)
{
	return mem_cgroup_force_empty(mem_cgroup_from_cont(cont), true);
}


3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459
static u64 mem_cgroup_hierarchy_read(struct cgroup *cont, struct cftype *cft)
{
	return mem_cgroup_from_cont(cont)->use_hierarchy;
}

static int mem_cgroup_hierarchy_write(struct cgroup *cont, struct cftype *cft,
					u64 val)
{
	int retval = 0;
	struct mem_cgroup *mem = mem_cgroup_from_cont(cont);
	struct cgroup *parent = cont->parent;
	struct mem_cgroup *parent_mem = NULL;

	if (parent)
		parent_mem = mem_cgroup_from_cont(parent);

	cgroup_lock();
	/*
3460
	 * If parent's use_hierarchy is set, we can't make any modifications
3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479
	 * in the child subtrees. If it is unset, then the change can
	 * occur, provided the current cgroup has no children.
	 *
	 * For the root cgroup, parent_mem is NULL, we allow value to be
	 * set if there are no children.
	 */
	if ((!parent_mem || !parent_mem->use_hierarchy) &&
				(val == 1 || val == 0)) {
		if (list_empty(&cont->children))
			mem->use_hierarchy = val;
		else
			retval = -EBUSY;
	} else
		retval = -EINVAL;
	cgroup_unlock();

	return retval;
}

3480

K
KAMEZAWA Hiroyuki 已提交
3481 3482
static u64 mem_cgroup_get_recursive_idx_stat(struct mem_cgroup *mem,
				enum mem_cgroup_stat_index idx)
3483
{
K
KAMEZAWA Hiroyuki 已提交
3484 3485
	struct mem_cgroup *iter;
	s64 val = 0;
3486

K
KAMEZAWA Hiroyuki 已提交
3487 3488 3489 3490 3491 3492 3493
	/* each per cpu's value can be minus.Then, use s64 */
	for_each_mem_cgroup_tree(iter, mem)
		val += mem_cgroup_read_stat(iter, idx);

	if (val < 0) /* race ? */
		val = 0;
	return val;
3494 3495
}

3496 3497
static inline u64 mem_cgroup_usage(struct mem_cgroup *mem, bool swap)
{
K
KAMEZAWA Hiroyuki 已提交
3498
	u64 val;
3499 3500 3501 3502 3503 3504 3505 3506

	if (!mem_cgroup_is_root(mem)) {
		if (!swap)
			return res_counter_read_u64(&mem->res, RES_USAGE);
		else
			return res_counter_read_u64(&mem->memsw, RES_USAGE);
	}

K
KAMEZAWA Hiroyuki 已提交
3507 3508
	val = mem_cgroup_get_recursive_idx_stat(mem, MEM_CGROUP_STAT_CACHE);
	val += mem_cgroup_get_recursive_idx_stat(mem, MEM_CGROUP_STAT_RSS);
3509

K
KAMEZAWA Hiroyuki 已提交
3510 3511 3512
	if (swap)
		val += mem_cgroup_get_recursive_idx_stat(mem,
				MEM_CGROUP_STAT_SWAPOUT);
3513 3514 3515 3516

	return val << PAGE_SHIFT;
}

3517
static u64 mem_cgroup_read(struct cgroup *cont, struct cftype *cft)
B
Balbir Singh 已提交
3518
{
3519
	struct mem_cgroup *mem = mem_cgroup_from_cont(cont);
3520
	u64 val;
3521 3522 3523 3524 3525 3526
	int type, name;

	type = MEMFILE_TYPE(cft->private);
	name = MEMFILE_ATTR(cft->private);
	switch (type) {
	case _MEM:
3527 3528 3529
		if (name == RES_USAGE)
			val = mem_cgroup_usage(mem, false);
		else
3530
			val = res_counter_read_u64(&mem->res, name);
3531 3532
		break;
	case _MEMSWAP:
3533 3534 3535
		if (name == RES_USAGE)
			val = mem_cgroup_usage(mem, true);
		else
3536
			val = res_counter_read_u64(&mem->memsw, name);
3537 3538 3539 3540 3541 3542
		break;
	default:
		BUG();
		break;
	}
	return val;
B
Balbir Singh 已提交
3543
}
3544 3545 3546 3547
/*
 * The user of this function is...
 * RES_LIMIT.
 */
3548 3549
static int mem_cgroup_write(struct cgroup *cont, struct cftype *cft,
			    const char *buffer)
B
Balbir Singh 已提交
3550
{
3551
	struct mem_cgroup *memcg = mem_cgroup_from_cont(cont);
3552
	int type, name;
3553 3554 3555
	unsigned long long val;
	int ret;

3556 3557 3558
	type = MEMFILE_TYPE(cft->private);
	name = MEMFILE_ATTR(cft->private);
	switch (name) {
3559
	case RES_LIMIT:
3560 3561 3562 3563
		if (mem_cgroup_is_root(memcg)) { /* Can't set limit on root */
			ret = -EINVAL;
			break;
		}
3564 3565
		/* This function does all necessary parse...reuse it */
		ret = res_counter_memparse_write_strategy(buffer, &val);
3566 3567 3568
		if (ret)
			break;
		if (type == _MEM)
3569
			ret = mem_cgroup_resize_limit(memcg, val);
3570 3571
		else
			ret = mem_cgroup_resize_memsw_limit(memcg, val);
3572
		break;
3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586
	case RES_SOFT_LIMIT:
		ret = res_counter_memparse_write_strategy(buffer, &val);
		if (ret)
			break;
		/*
		 * For memsw, soft limits are hard to implement in terms
		 * of semantics, for now, we support soft limits for
		 * control without swap
		 */
		if (type == _MEM)
			ret = res_counter_set_soft_limit(&memcg->res, val);
		else
			ret = -EINVAL;
		break;
3587 3588 3589 3590 3591
	default:
		ret = -EINVAL; /* should be BUG() ? */
		break;
	}
	return ret;
B
Balbir Singh 已提交
3592 3593
}

3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621
static void memcg_get_hierarchical_limit(struct mem_cgroup *memcg,
		unsigned long long *mem_limit, unsigned long long *memsw_limit)
{
	struct cgroup *cgroup;
	unsigned long long min_limit, min_memsw_limit, tmp;

	min_limit = res_counter_read_u64(&memcg->res, RES_LIMIT);
	min_memsw_limit = res_counter_read_u64(&memcg->memsw, RES_LIMIT);
	cgroup = memcg->css.cgroup;
	if (!memcg->use_hierarchy)
		goto out;

	while (cgroup->parent) {
		cgroup = cgroup->parent;
		memcg = mem_cgroup_from_cont(cgroup);
		if (!memcg->use_hierarchy)
			break;
		tmp = res_counter_read_u64(&memcg->res, RES_LIMIT);
		min_limit = min(min_limit, tmp);
		tmp = res_counter_read_u64(&memcg->memsw, RES_LIMIT);
		min_memsw_limit = min(min_memsw_limit, tmp);
	}
out:
	*mem_limit = min_limit;
	*memsw_limit = min_memsw_limit;
	return;
}

3622
static int mem_cgroup_reset(struct cgroup *cont, unsigned int event)
3623 3624
{
	struct mem_cgroup *mem;
3625
	int type, name;
3626 3627

	mem = mem_cgroup_from_cont(cont);
3628 3629 3630
	type = MEMFILE_TYPE(event);
	name = MEMFILE_ATTR(event);
	switch (name) {
3631
	case RES_MAX_USAGE:
3632 3633 3634 3635
		if (type == _MEM)
			res_counter_reset_max(&mem->res);
		else
			res_counter_reset_max(&mem->memsw);
3636 3637
		break;
	case RES_FAILCNT:
3638 3639 3640 3641
		if (type == _MEM)
			res_counter_reset_failcnt(&mem->res);
		else
			res_counter_reset_failcnt(&mem->memsw);
3642 3643
		break;
	}
3644

3645
	return 0;
3646 3647
}

3648 3649 3650 3651 3652 3653
static u64 mem_cgroup_move_charge_read(struct cgroup *cgrp,
					struct cftype *cft)
{
	return mem_cgroup_from_cont(cgrp)->move_charge_at_immigrate;
}

3654
#ifdef CONFIG_MMU
3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672
static int mem_cgroup_move_charge_write(struct cgroup *cgrp,
					struct cftype *cft, u64 val)
{
	struct mem_cgroup *mem = mem_cgroup_from_cont(cgrp);

	if (val >= (1 << NR_MOVE_TYPE))
		return -EINVAL;
	/*
	 * We check this value several times in both in can_attach() and
	 * attach(), so we need cgroup lock to prevent this value from being
	 * inconsistent.
	 */
	cgroup_lock();
	mem->move_charge_at_immigrate = val;
	cgroup_unlock();

	return 0;
}
3673 3674 3675 3676 3677 3678 3679
#else
static int mem_cgroup_move_charge_write(struct cgroup *cgrp,
					struct cftype *cft, u64 val)
{
	return -ENOSYS;
}
#endif
3680

K
KAMEZAWA Hiroyuki 已提交
3681 3682 3683 3684 3685

/* For read statistics */
enum {
	MCS_CACHE,
	MCS_RSS,
3686
	MCS_FILE_MAPPED,
K
KAMEZAWA Hiroyuki 已提交
3687 3688
	MCS_PGPGIN,
	MCS_PGPGOUT,
3689
	MCS_SWAP,
K
KAMEZAWA Hiroyuki 已提交
3690 3691 3692 3693 3694 3695 3696 3697 3698 3699
	MCS_INACTIVE_ANON,
	MCS_ACTIVE_ANON,
	MCS_INACTIVE_FILE,
	MCS_ACTIVE_FILE,
	MCS_UNEVICTABLE,
	NR_MCS_STAT,
};

struct mcs_total_stat {
	s64 stat[NR_MCS_STAT];
3700 3701
};

K
KAMEZAWA Hiroyuki 已提交
3702 3703 3704 3705 3706 3707
struct {
	char *local_name;
	char *total_name;
} memcg_stat_strings[NR_MCS_STAT] = {
	{"cache", "total_cache"},
	{"rss", "total_rss"},
3708
	{"mapped_file", "total_mapped_file"},
K
KAMEZAWA Hiroyuki 已提交
3709 3710
	{"pgpgin", "total_pgpgin"},
	{"pgpgout", "total_pgpgout"},
3711
	{"swap", "total_swap"},
K
KAMEZAWA Hiroyuki 已提交
3712 3713 3714 3715 3716 3717 3718 3719
	{"inactive_anon", "total_inactive_anon"},
	{"active_anon", "total_active_anon"},
	{"inactive_file", "total_inactive_file"},
	{"active_file", "total_active_file"},
	{"unevictable", "total_unevictable"}
};


K
KAMEZAWA Hiroyuki 已提交
3720 3721
static void
mem_cgroup_get_local_stat(struct mem_cgroup *mem, struct mcs_total_stat *s)
K
KAMEZAWA Hiroyuki 已提交
3722 3723 3724 3725
{
	s64 val;

	/* per cpu stat */
3726
	val = mem_cgroup_read_stat(mem, MEM_CGROUP_STAT_CACHE);
K
KAMEZAWA Hiroyuki 已提交
3727
	s->stat[MCS_CACHE] += val * PAGE_SIZE;
3728
	val = mem_cgroup_read_stat(mem, MEM_CGROUP_STAT_RSS);
K
KAMEZAWA Hiroyuki 已提交
3729
	s->stat[MCS_RSS] += val * PAGE_SIZE;
3730
	val = mem_cgroup_read_stat(mem, MEM_CGROUP_STAT_FILE_MAPPED);
3731
	s->stat[MCS_FILE_MAPPED] += val * PAGE_SIZE;
3732
	val = mem_cgroup_read_stat(mem, MEM_CGROUP_STAT_PGPGIN_COUNT);
K
KAMEZAWA Hiroyuki 已提交
3733
	s->stat[MCS_PGPGIN] += val;
3734
	val = mem_cgroup_read_stat(mem, MEM_CGROUP_STAT_PGPGOUT_COUNT);
K
KAMEZAWA Hiroyuki 已提交
3735
	s->stat[MCS_PGPGOUT] += val;
3736
	if (do_swap_account) {
3737
		val = mem_cgroup_read_stat(mem, MEM_CGROUP_STAT_SWAPOUT);
3738 3739
		s->stat[MCS_SWAP] += val * PAGE_SIZE;
	}
K
KAMEZAWA Hiroyuki 已提交
3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756

	/* per zone stat */
	val = mem_cgroup_get_local_zonestat(mem, LRU_INACTIVE_ANON);
	s->stat[MCS_INACTIVE_ANON] += val * PAGE_SIZE;
	val = mem_cgroup_get_local_zonestat(mem, LRU_ACTIVE_ANON);
	s->stat[MCS_ACTIVE_ANON] += val * PAGE_SIZE;
	val = mem_cgroup_get_local_zonestat(mem, LRU_INACTIVE_FILE);
	s->stat[MCS_INACTIVE_FILE] += val * PAGE_SIZE;
	val = mem_cgroup_get_local_zonestat(mem, LRU_ACTIVE_FILE);
	s->stat[MCS_ACTIVE_FILE] += val * PAGE_SIZE;
	val = mem_cgroup_get_local_zonestat(mem, LRU_UNEVICTABLE);
	s->stat[MCS_UNEVICTABLE] += val * PAGE_SIZE;
}

static void
mem_cgroup_get_total_stat(struct mem_cgroup *mem, struct mcs_total_stat *s)
{
K
KAMEZAWA Hiroyuki 已提交
3757 3758 3759 3760
	struct mem_cgroup *iter;

	for_each_mem_cgroup_tree(iter, mem)
		mem_cgroup_get_local_stat(iter, s);
K
KAMEZAWA Hiroyuki 已提交
3761 3762
}

3763 3764
static int mem_control_stat_show(struct cgroup *cont, struct cftype *cft,
				 struct cgroup_map_cb *cb)
3765 3766
{
	struct mem_cgroup *mem_cont = mem_cgroup_from_cont(cont);
K
KAMEZAWA Hiroyuki 已提交
3767
	struct mcs_total_stat mystat;
3768 3769
	int i;

K
KAMEZAWA Hiroyuki 已提交
3770 3771
	memset(&mystat, 0, sizeof(mystat));
	mem_cgroup_get_local_stat(mem_cont, &mystat);
3772

3773 3774 3775
	for (i = 0; i < NR_MCS_STAT; i++) {
		if (i == MCS_SWAP && !do_swap_account)
			continue;
K
KAMEZAWA Hiroyuki 已提交
3776
		cb->fill(cb, memcg_stat_strings[i].local_name, mystat.stat[i]);
3777
	}
L
Lee Schermerhorn 已提交
3778

K
KAMEZAWA Hiroyuki 已提交
3779
	/* Hierarchical information */
3780 3781 3782 3783 3784 3785 3786
	{
		unsigned long long limit, memsw_limit;
		memcg_get_hierarchical_limit(mem_cont, &limit, &memsw_limit);
		cb->fill(cb, "hierarchical_memory_limit", limit);
		if (do_swap_account)
			cb->fill(cb, "hierarchical_memsw_limit", memsw_limit);
	}
K
KOSAKI Motohiro 已提交
3787

K
KAMEZAWA Hiroyuki 已提交
3788 3789
	memset(&mystat, 0, sizeof(mystat));
	mem_cgroup_get_total_stat(mem_cont, &mystat);
3790 3791 3792
	for (i = 0; i < NR_MCS_STAT; i++) {
		if (i == MCS_SWAP && !do_swap_account)
			continue;
K
KAMEZAWA Hiroyuki 已提交
3793
		cb->fill(cb, memcg_stat_strings[i].total_name, mystat.stat[i]);
3794
	}
K
KAMEZAWA Hiroyuki 已提交
3795

K
KOSAKI Motohiro 已提交
3796
#ifdef CONFIG_DEBUG_VM
3797
	cb->fill(cb, "inactive_ratio", calc_inactive_ratio(mem_cont, NULL));
K
KOSAKI Motohiro 已提交
3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824

	{
		int nid, zid;
		struct mem_cgroup_per_zone *mz;
		unsigned long recent_rotated[2] = {0, 0};
		unsigned long recent_scanned[2] = {0, 0};

		for_each_online_node(nid)
			for (zid = 0; zid < MAX_NR_ZONES; zid++) {
				mz = mem_cgroup_zoneinfo(mem_cont, nid, zid);

				recent_rotated[0] +=
					mz->reclaim_stat.recent_rotated[0];
				recent_rotated[1] +=
					mz->reclaim_stat.recent_rotated[1];
				recent_scanned[0] +=
					mz->reclaim_stat.recent_scanned[0];
				recent_scanned[1] +=
					mz->reclaim_stat.recent_scanned[1];
			}
		cb->fill(cb, "recent_rotated_anon", recent_rotated[0]);
		cb->fill(cb, "recent_rotated_file", recent_rotated[1]);
		cb->fill(cb, "recent_scanned_anon", recent_scanned[0]);
		cb->fill(cb, "recent_scanned_file", recent_scanned[1]);
	}
#endif

3825 3826 3827
	return 0;
}

K
KOSAKI Motohiro 已提交
3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839
static u64 mem_cgroup_swappiness_read(struct cgroup *cgrp, struct cftype *cft)
{
	struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp);

	return get_swappiness(memcg);
}

static int mem_cgroup_swappiness_write(struct cgroup *cgrp, struct cftype *cft,
				       u64 val)
{
	struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp);
	struct mem_cgroup *parent;
3840

K
KOSAKI Motohiro 已提交
3841 3842 3843 3844 3845 3846 3847
	if (val > 100)
		return -EINVAL;

	if (cgrp->parent == NULL)
		return -EINVAL;

	parent = mem_cgroup_from_cont(cgrp->parent);
3848 3849 3850

	cgroup_lock();

K
KOSAKI Motohiro 已提交
3851 3852
	/* If under hierarchy, only empty-root can set this value */
	if ((parent->use_hierarchy) ||
3853 3854
	    (memcg->use_hierarchy && !list_empty(&cgrp->children))) {
		cgroup_unlock();
K
KOSAKI Motohiro 已提交
3855
		return -EINVAL;
3856
	}
K
KOSAKI Motohiro 已提交
3857 3858 3859

	memcg->swappiness = val;

3860 3861
	cgroup_unlock();

K
KOSAKI Motohiro 已提交
3862 3863 3864
	return 0;
}

3865 3866 3867 3868 3869 3870 3871 3872
static void __mem_cgroup_threshold(struct mem_cgroup *memcg, bool swap)
{
	struct mem_cgroup_threshold_ary *t;
	u64 usage;
	int i;

	rcu_read_lock();
	if (!swap)
3873
		t = rcu_dereference(memcg->thresholds.primary);
3874
	else
3875
		t = rcu_dereference(memcg->memsw_thresholds.primary);
3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886

	if (!t)
		goto unlock;

	usage = mem_cgroup_usage(memcg, swap);

	/*
	 * current_threshold points to threshold just below usage.
	 * If it's not true, a threshold was crossed after last
	 * call of __mem_cgroup_threshold().
	 */
3887
	i = t->current_threshold;
3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910

	/*
	 * Iterate backward over array of thresholds starting from
	 * current_threshold and check if a threshold is crossed.
	 * If none of thresholds below usage is crossed, we read
	 * only one element of the array here.
	 */
	for (; i >= 0 && unlikely(t->entries[i].threshold > usage); i--)
		eventfd_signal(t->entries[i].eventfd, 1);

	/* i = current_threshold + 1 */
	i++;

	/*
	 * Iterate forward over array of thresholds starting from
	 * current_threshold+1 and check if a threshold is crossed.
	 * If none of thresholds above usage is crossed, we read
	 * only one element of the array here.
	 */
	for (; i < t->size && unlikely(t->entries[i].threshold <= usage); i++)
		eventfd_signal(t->entries[i].eventfd, 1);

	/* Update current_threshold */
3911
	t->current_threshold = i - 1;
3912 3913 3914 3915 3916 3917
unlock:
	rcu_read_unlock();
}

static void mem_cgroup_threshold(struct mem_cgroup *memcg)
{
3918 3919 3920 3921 3922 3923 3924
	while (memcg) {
		__mem_cgroup_threshold(memcg, false);
		if (do_swap_account)
			__mem_cgroup_threshold(memcg, true);

		memcg = parent_mem_cgroup(memcg);
	}
3925 3926 3927 3928 3929 3930 3931 3932 3933 3934
}

static int compare_thresholds(const void *a, const void *b)
{
	const struct mem_cgroup_threshold *_a = a;
	const struct mem_cgroup_threshold *_b = b;

	return _a->threshold - _b->threshold;
}

K
KAMEZAWA Hiroyuki 已提交
3935
static int mem_cgroup_oom_notify_cb(struct mem_cgroup *mem)
K
KAMEZAWA Hiroyuki 已提交
3936 3937 3938 3939 3940 3941 3942 3943 3944 3945
{
	struct mem_cgroup_eventfd_list *ev;

	list_for_each_entry(ev, &mem->oom_notify, list)
		eventfd_signal(ev->eventfd, 1);
	return 0;
}

static void mem_cgroup_oom_notify(struct mem_cgroup *mem)
{
K
KAMEZAWA Hiroyuki 已提交
3946 3947 3948 3949
	struct mem_cgroup *iter;

	for_each_mem_cgroup_tree(iter, mem)
		mem_cgroup_oom_notify_cb(iter);
K
KAMEZAWA Hiroyuki 已提交
3950 3951 3952 3953
}

static int mem_cgroup_usage_register_event(struct cgroup *cgrp,
	struct cftype *cft, struct eventfd_ctx *eventfd, const char *args)
3954 3955
{
	struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp);
3956 3957
	struct mem_cgroup_thresholds *thresholds;
	struct mem_cgroup_threshold_ary *new;
3958 3959
	int type = MEMFILE_TYPE(cft->private);
	u64 threshold, usage;
3960
	int i, size, ret;
3961 3962 3963 3964 3965 3966

	ret = res_counter_memparse_write_strategy(args, &threshold);
	if (ret)
		return ret;

	mutex_lock(&memcg->thresholds_lock);
3967

3968
	if (type == _MEM)
3969
		thresholds = &memcg->thresholds;
3970
	else if (type == _MEMSWAP)
3971
		thresholds = &memcg->memsw_thresholds;
3972 3973 3974 3975 3976 3977
	else
		BUG();

	usage = mem_cgroup_usage(memcg, type == _MEMSWAP);

	/* Check if a threshold crossed before adding a new one */
3978
	if (thresholds->primary)
3979 3980
		__mem_cgroup_threshold(memcg, type == _MEMSWAP);

3981
	size = thresholds->primary ? thresholds->primary->size + 1 : 1;
3982 3983

	/* Allocate memory for new array of thresholds */
3984
	new = kmalloc(sizeof(*new) + size * sizeof(struct mem_cgroup_threshold),
3985
			GFP_KERNEL);
3986
	if (!new) {
3987 3988 3989
		ret = -ENOMEM;
		goto unlock;
	}
3990
	new->size = size;
3991 3992

	/* Copy thresholds (if any) to new array */
3993 3994
	if (thresholds->primary) {
		memcpy(new->entries, thresholds->primary->entries, (size - 1) *
3995
				sizeof(struct mem_cgroup_threshold));
3996 3997
	}

3998
	/* Add new threshold */
3999 4000
	new->entries[size - 1].eventfd = eventfd;
	new->entries[size - 1].threshold = threshold;
4001 4002

	/* Sort thresholds. Registering of new threshold isn't time-critical */
4003
	sort(new->entries, size, sizeof(struct mem_cgroup_threshold),
4004 4005 4006
			compare_thresholds, NULL);

	/* Find current threshold */
4007
	new->current_threshold = -1;
4008
	for (i = 0; i < size; i++) {
4009
		if (new->entries[i].threshold < usage) {
4010
			/*
4011 4012
			 * new->current_threshold will not be used until
			 * rcu_assign_pointer(), so it's safe to increment
4013 4014
			 * it here.
			 */
4015
			++new->current_threshold;
4016 4017 4018
		}
	}

4019 4020 4021 4022 4023
	/* Free old spare buffer and save old primary buffer as spare */
	kfree(thresholds->spare);
	thresholds->spare = thresholds->primary;

	rcu_assign_pointer(thresholds->primary, new);
4024

4025
	/* To be sure that nobody uses thresholds */
4026 4027 4028 4029 4030 4031 4032 4033
	synchronize_rcu();

unlock:
	mutex_unlock(&memcg->thresholds_lock);

	return ret;
}

4034
static void mem_cgroup_usage_unregister_event(struct cgroup *cgrp,
K
KAMEZAWA Hiroyuki 已提交
4035
	struct cftype *cft, struct eventfd_ctx *eventfd)
4036 4037
{
	struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp);
4038 4039
	struct mem_cgroup_thresholds *thresholds;
	struct mem_cgroup_threshold_ary *new;
4040 4041
	int type = MEMFILE_TYPE(cft->private);
	u64 usage;
4042
	int i, j, size;
4043 4044 4045

	mutex_lock(&memcg->thresholds_lock);
	if (type == _MEM)
4046
		thresholds = &memcg->thresholds;
4047
	else if (type == _MEMSWAP)
4048
		thresholds = &memcg->memsw_thresholds;
4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063
	else
		BUG();

	/*
	 * Something went wrong if we trying to unregister a threshold
	 * if we don't have thresholds
	 */
	BUG_ON(!thresholds);

	usage = mem_cgroup_usage(memcg, type == _MEMSWAP);

	/* Check if a threshold crossed before removing */
	__mem_cgroup_threshold(memcg, type == _MEMSWAP);

	/* Calculate new number of threshold */
4064 4065 4066
	size = 0;
	for (i = 0; i < thresholds->primary->size; i++) {
		if (thresholds->primary->entries[i].eventfd != eventfd)
4067 4068 4069
			size++;
	}

4070
	new = thresholds->spare;
4071

4072 4073
	/* Set thresholds array to NULL if we don't have thresholds */
	if (!size) {
4074 4075
		kfree(new);
		new = NULL;
4076
		goto swap_buffers;
4077 4078
	}

4079
	new->size = size;
4080 4081

	/* Copy thresholds and find current threshold */
4082 4083 4084
	new->current_threshold = -1;
	for (i = 0, j = 0; i < thresholds->primary->size; i++) {
		if (thresholds->primary->entries[i].eventfd == eventfd)
4085 4086
			continue;

4087 4088
		new->entries[j] = thresholds->primary->entries[i];
		if (new->entries[j].threshold < usage) {
4089
			/*
4090
			 * new->current_threshold will not be used
4091 4092 4093
			 * until rcu_assign_pointer(), so it's safe to increment
			 * it here.
			 */
4094
			++new->current_threshold;
4095 4096 4097 4098
		}
		j++;
	}

4099
swap_buffers:
4100 4101 4102
	/* Swap primary and spare array */
	thresholds->spare = thresholds->primary;
	rcu_assign_pointer(thresholds->primary, new);
4103

4104
	/* To be sure that nobody uses thresholds */
4105 4106 4107 4108
	synchronize_rcu();

	mutex_unlock(&memcg->thresholds_lock);
}
4109

K
KAMEZAWA Hiroyuki 已提交
4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134
static int mem_cgroup_oom_register_event(struct cgroup *cgrp,
	struct cftype *cft, struct eventfd_ctx *eventfd, const char *args)
{
	struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp);
	struct mem_cgroup_eventfd_list *event;
	int type = MEMFILE_TYPE(cft->private);

	BUG_ON(type != _OOM_TYPE);
	event = kmalloc(sizeof(*event),	GFP_KERNEL);
	if (!event)
		return -ENOMEM;

	mutex_lock(&memcg_oom_mutex);

	event->eventfd = eventfd;
	list_add(&event->list, &memcg->oom_notify);

	/* already in OOM ? */
	if (atomic_read(&memcg->oom_lock))
		eventfd_signal(eventfd, 1);
	mutex_unlock(&memcg_oom_mutex);

	return 0;
}

4135
static void mem_cgroup_oom_unregister_event(struct cgroup *cgrp,
K
KAMEZAWA Hiroyuki 已提交
4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155
	struct cftype *cft, struct eventfd_ctx *eventfd)
{
	struct mem_cgroup *mem = mem_cgroup_from_cont(cgrp);
	struct mem_cgroup_eventfd_list *ev, *tmp;
	int type = MEMFILE_TYPE(cft->private);

	BUG_ON(type != _OOM_TYPE);

	mutex_lock(&memcg_oom_mutex);

	list_for_each_entry_safe(ev, tmp, &mem->oom_notify, list) {
		if (ev->eventfd == eventfd) {
			list_del(&ev->list);
			kfree(ev);
		}
	}

	mutex_unlock(&memcg_oom_mutex);
}

4156 4157 4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189
static int mem_cgroup_oom_control_read(struct cgroup *cgrp,
	struct cftype *cft,  struct cgroup_map_cb *cb)
{
	struct mem_cgroup *mem = mem_cgroup_from_cont(cgrp);

	cb->fill(cb, "oom_kill_disable", mem->oom_kill_disable);

	if (atomic_read(&mem->oom_lock))
		cb->fill(cb, "under_oom", 1);
	else
		cb->fill(cb, "under_oom", 0);
	return 0;
}

static int mem_cgroup_oom_control_write(struct cgroup *cgrp,
	struct cftype *cft, u64 val)
{
	struct mem_cgroup *mem = mem_cgroup_from_cont(cgrp);
	struct mem_cgroup *parent;

	/* cannot set to root cgroup and only 0 and 1 are allowed */
	if (!cgrp->parent || !((val == 0) || (val == 1)))
		return -EINVAL;

	parent = mem_cgroup_from_cont(cgrp->parent);

	cgroup_lock();
	/* oom-kill-disable is a flag for subhierarchy. */
	if ((parent->use_hierarchy) ||
	    (mem->use_hierarchy && !list_empty(&cgrp->children))) {
		cgroup_unlock();
		return -EINVAL;
	}
	mem->oom_kill_disable = val;
4190 4191
	if (!val)
		memcg_oom_recover(mem);
4192 4193 4194 4195
	cgroup_unlock();
	return 0;
}

B
Balbir Singh 已提交
4196 4197
static struct cftype mem_cgroup_files[] = {
	{
4198
		.name = "usage_in_bytes",
4199
		.private = MEMFILE_PRIVATE(_MEM, RES_USAGE),
4200
		.read_u64 = mem_cgroup_read,
K
KAMEZAWA Hiroyuki 已提交
4201 4202
		.register_event = mem_cgroup_usage_register_event,
		.unregister_event = mem_cgroup_usage_unregister_event,
B
Balbir Singh 已提交
4203
	},
4204 4205
	{
		.name = "max_usage_in_bytes",
4206
		.private = MEMFILE_PRIVATE(_MEM, RES_MAX_USAGE),
4207
		.trigger = mem_cgroup_reset,
4208 4209
		.read_u64 = mem_cgroup_read,
	},
B
Balbir Singh 已提交
4210
	{
4211
		.name = "limit_in_bytes",
4212
		.private = MEMFILE_PRIVATE(_MEM, RES_LIMIT),
4213
		.write_string = mem_cgroup_write,
4214
		.read_u64 = mem_cgroup_read,
B
Balbir Singh 已提交
4215
	},
4216 4217 4218 4219 4220 4221
	{
		.name = "soft_limit_in_bytes",
		.private = MEMFILE_PRIVATE(_MEM, RES_SOFT_LIMIT),
		.write_string = mem_cgroup_write,
		.read_u64 = mem_cgroup_read,
	},
B
Balbir Singh 已提交
4222 4223
	{
		.name = "failcnt",
4224
		.private = MEMFILE_PRIVATE(_MEM, RES_FAILCNT),
4225
		.trigger = mem_cgroup_reset,
4226
		.read_u64 = mem_cgroup_read,
B
Balbir Singh 已提交
4227
	},
4228 4229
	{
		.name = "stat",
4230
		.read_map = mem_control_stat_show,
4231
	},
4232 4233 4234 4235
	{
		.name = "force_empty",
		.trigger = mem_cgroup_force_empty_write,
	},
4236 4237 4238 4239 4240
	{
		.name = "use_hierarchy",
		.write_u64 = mem_cgroup_hierarchy_write,
		.read_u64 = mem_cgroup_hierarchy_read,
	},
K
KOSAKI Motohiro 已提交
4241 4242 4243 4244 4245
	{
		.name = "swappiness",
		.read_u64 = mem_cgroup_swappiness_read,
		.write_u64 = mem_cgroup_swappiness_write,
	},
4246 4247 4248 4249 4250
	{
		.name = "move_charge_at_immigrate",
		.read_u64 = mem_cgroup_move_charge_read,
		.write_u64 = mem_cgroup_move_charge_write,
	},
K
KAMEZAWA Hiroyuki 已提交
4251 4252
	{
		.name = "oom_control",
4253 4254
		.read_map = mem_cgroup_oom_control_read,
		.write_u64 = mem_cgroup_oom_control_write,
K
KAMEZAWA Hiroyuki 已提交
4255 4256 4257 4258
		.register_event = mem_cgroup_oom_register_event,
		.unregister_event = mem_cgroup_oom_unregister_event,
		.private = MEMFILE_PRIVATE(_OOM_TYPE, OOM_CONTROL),
	},
B
Balbir Singh 已提交
4259 4260
};

4261 4262 4263 4264 4265 4266
#ifdef CONFIG_CGROUP_MEM_RES_CTLR_SWAP
static struct cftype memsw_cgroup_files[] = {
	{
		.name = "memsw.usage_in_bytes",
		.private = MEMFILE_PRIVATE(_MEMSWAP, RES_USAGE),
		.read_u64 = mem_cgroup_read,
K
KAMEZAWA Hiroyuki 已提交
4267 4268
		.register_event = mem_cgroup_usage_register_event,
		.unregister_event = mem_cgroup_usage_unregister_event,
4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303
	},
	{
		.name = "memsw.max_usage_in_bytes",
		.private = MEMFILE_PRIVATE(_MEMSWAP, RES_MAX_USAGE),
		.trigger = mem_cgroup_reset,
		.read_u64 = mem_cgroup_read,
	},
	{
		.name = "memsw.limit_in_bytes",
		.private = MEMFILE_PRIVATE(_MEMSWAP, RES_LIMIT),
		.write_string = mem_cgroup_write,
		.read_u64 = mem_cgroup_read,
	},
	{
		.name = "memsw.failcnt",
		.private = MEMFILE_PRIVATE(_MEMSWAP, RES_FAILCNT),
		.trigger = mem_cgroup_reset,
		.read_u64 = mem_cgroup_read,
	},
};

static int register_memsw_files(struct cgroup *cont, struct cgroup_subsys *ss)
{
	if (!do_swap_account)
		return 0;
	return cgroup_add_files(cont, ss, memsw_cgroup_files,
				ARRAY_SIZE(memsw_cgroup_files));
};
#else
static int register_memsw_files(struct cgroup *cont, struct cgroup_subsys *ss)
{
	return 0;
}
#endif

4304 4305 4306
static int alloc_mem_cgroup_per_zone_info(struct mem_cgroup *mem, int node)
{
	struct mem_cgroup_per_node *pn;
4307
	struct mem_cgroup_per_zone *mz;
4308
	enum lru_list l;
4309
	int zone, tmp = node;
4310 4311 4312 4313 4314 4315 4316 4317
	/*
	 * This routine is called against possible nodes.
	 * But it's BUG to call kmalloc() against offline node.
	 *
	 * TODO: this routine can waste much memory for nodes which will
	 *       never be onlined. It's better to use memory hotplug callback
	 *       function.
	 */
4318 4319
	if (!node_state(node, N_NORMAL_MEMORY))
		tmp = -1;
4320
	pn = kzalloc_node(sizeof(*pn), GFP_KERNEL, tmp);
4321 4322
	if (!pn)
		return 1;
4323

4324
	mem->info.nodeinfo[node] = pn;
4325 4326
	for (zone = 0; zone < MAX_NR_ZONES; zone++) {
		mz = &pn->zoneinfo[zone];
4327 4328
		for_each_lru(l)
			INIT_LIST_HEAD(&mz->lists[l]);
4329
		mz->usage_in_excess = 0;
4330 4331
		mz->on_tree = false;
		mz->mem = mem;
4332
	}
4333 4334 4335
	return 0;
}

4336 4337 4338 4339 4340
static void free_mem_cgroup_per_zone_info(struct mem_cgroup *mem, int node)
{
	kfree(mem->info.nodeinfo[node]);
}

4341 4342 4343
static struct mem_cgroup *mem_cgroup_alloc(void)
{
	struct mem_cgroup *mem;
4344
	int size = sizeof(struct mem_cgroup);
4345

4346
	/* Can be very big if MAX_NUMNODES is very big */
4347
	if (size < PAGE_SIZE)
4348
		mem = kzalloc(size, GFP_KERNEL);
4349
	else
4350
		mem = vzalloc(size);
4351

4352 4353 4354
	if (!mem)
		return NULL;

4355
	mem->stat = alloc_percpu(struct mem_cgroup_stat_cpu);
4356 4357
	if (!mem->stat)
		goto out_free;
4358
	spin_lock_init(&mem->pcp_counter_lock);
4359
	return mem;
4360 4361 4362 4363 4364 4365 4366

out_free:
	if (size < PAGE_SIZE)
		kfree(mem);
	else
		vfree(mem);
	return NULL;
4367 4368
}

4369 4370 4371 4372 4373 4374 4375 4376 4377 4378 4379
/*
 * At destroying mem_cgroup, references from swap_cgroup can remain.
 * (scanning all at force_empty is too costly...)
 *
 * Instead of clearing all references at force_empty, we remember
 * the number of reference from swap_cgroup and free mem_cgroup when
 * it goes down to 0.
 *
 * Removal of cgroup itself succeeds regardless of refs from swap.
 */

4380
static void __mem_cgroup_free(struct mem_cgroup *mem)
4381
{
K
KAMEZAWA Hiroyuki 已提交
4382 4383
	int node;

4384
	mem_cgroup_remove_from_trees(mem);
K
KAMEZAWA Hiroyuki 已提交
4385 4386
	free_css_id(&mem_cgroup_subsys, &mem->css);

K
KAMEZAWA Hiroyuki 已提交
4387 4388 4389
	for_each_node_state(node, N_POSSIBLE)
		free_mem_cgroup_per_zone_info(mem, node);

4390 4391
	free_percpu(mem->stat);
	if (sizeof(struct mem_cgroup) < PAGE_SIZE)
4392 4393 4394 4395 4396
		kfree(mem);
	else
		vfree(mem);
}

4397 4398 4399 4400 4401
static void mem_cgroup_get(struct mem_cgroup *mem)
{
	atomic_inc(&mem->refcnt);
}

4402
static void __mem_cgroup_put(struct mem_cgroup *mem, int count)
4403
{
4404
	if (atomic_sub_and_test(count, &mem->refcnt)) {
4405
		struct mem_cgroup *parent = parent_mem_cgroup(mem);
4406
		__mem_cgroup_free(mem);
4407 4408 4409
		if (parent)
			mem_cgroup_put(parent);
	}
4410 4411
}

4412 4413 4414 4415 4416
static void mem_cgroup_put(struct mem_cgroup *mem)
{
	__mem_cgroup_put(mem, 1);
}

4417 4418 4419 4420 4421 4422 4423 4424 4425
/*
 * Returns the parent mem_cgroup in memcgroup hierarchy with hierarchy enabled.
 */
static struct mem_cgroup *parent_mem_cgroup(struct mem_cgroup *mem)
{
	if (!mem->res.parent)
		return NULL;
	return mem_cgroup_from_res_counter(mem->res.parent, res);
}
4426

4427 4428 4429
#ifdef CONFIG_CGROUP_MEM_RES_CTLR_SWAP
static void __init enable_swap_cgroup(void)
{
4430
	if (!mem_cgroup_disabled() && really_do_swap_account)
4431 4432 4433 4434 4435 4436 4437 4438
		do_swap_account = 1;
}
#else
static void __init enable_swap_cgroup(void)
{
}
#endif

4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463
static int mem_cgroup_soft_limit_tree_init(void)
{
	struct mem_cgroup_tree_per_node *rtpn;
	struct mem_cgroup_tree_per_zone *rtpz;
	int tmp, node, zone;

	for_each_node_state(node, N_POSSIBLE) {
		tmp = node;
		if (!node_state(node, N_NORMAL_MEMORY))
			tmp = -1;
		rtpn = kzalloc_node(sizeof(*rtpn), GFP_KERNEL, tmp);
		if (!rtpn)
			return 1;

		soft_limit_tree.rb_tree_per_node[node] = rtpn;

		for (zone = 0; zone < MAX_NR_ZONES; zone++) {
			rtpz = &rtpn->rb_tree_per_zone[zone];
			rtpz->rb_root = RB_ROOT;
			spin_lock_init(&rtpz->lock);
		}
	}
	return 0;
}

L
Li Zefan 已提交
4464
static struct cgroup_subsys_state * __ref
B
Balbir Singh 已提交
4465 4466
mem_cgroup_create(struct cgroup_subsys *ss, struct cgroup *cont)
{
4467
	struct mem_cgroup *mem, *parent;
K
KAMEZAWA Hiroyuki 已提交
4468
	long error = -ENOMEM;
4469
	int node;
B
Balbir Singh 已提交
4470

4471 4472
	mem = mem_cgroup_alloc();
	if (!mem)
K
KAMEZAWA Hiroyuki 已提交
4473
		return ERR_PTR(error);
4474

4475 4476 4477
	for_each_node_state(node, N_POSSIBLE)
		if (alloc_mem_cgroup_per_zone_info(mem, node))
			goto free_out;
4478

4479
	/* root ? */
4480
	if (cont->parent == NULL) {
4481
		int cpu;
4482
		enable_swap_cgroup();
4483
		parent = NULL;
4484
		root_mem_cgroup = mem;
4485 4486
		if (mem_cgroup_soft_limit_tree_init())
			goto free_out;
4487 4488 4489 4490 4491
		for_each_possible_cpu(cpu) {
			struct memcg_stock_pcp *stock =
						&per_cpu(memcg_stock, cpu);
			INIT_WORK(&stock->work, drain_local_stock);
		}
4492
		hotcpu_notifier(memcg_cpu_hotplug_callback, 0);
4493
	} else {
4494
		parent = mem_cgroup_from_cont(cont->parent);
4495
		mem->use_hierarchy = parent->use_hierarchy;
4496
		mem->oom_kill_disable = parent->oom_kill_disable;
4497
	}
4498

4499 4500 4501
	if (parent && parent->use_hierarchy) {
		res_counter_init(&mem->res, &parent->res);
		res_counter_init(&mem->memsw, &parent->memsw);
4502 4503 4504 4505 4506 4507 4508
		/*
		 * We increment refcnt of the parent to ensure that we can
		 * safely access it on res_counter_charge/uncharge.
		 * This refcnt will be decremented when freeing this
		 * mem_cgroup(see mem_cgroup_put).
		 */
		mem_cgroup_get(parent);
4509 4510 4511 4512
	} else {
		res_counter_init(&mem->res, NULL);
		res_counter_init(&mem->memsw, NULL);
	}
K
KAMEZAWA Hiroyuki 已提交
4513
	mem->last_scanned_child = 0;
K
KAMEZAWA Hiroyuki 已提交
4514
	INIT_LIST_HEAD(&mem->oom_notify);
4515

K
KOSAKI Motohiro 已提交
4516 4517
	if (parent)
		mem->swappiness = get_swappiness(parent);
4518
	atomic_set(&mem->refcnt, 1);
4519
	mem->move_charge_at_immigrate = 0;
4520
	mutex_init(&mem->thresholds_lock);
B
Balbir Singh 已提交
4521
	return &mem->css;
4522
free_out:
4523
	__mem_cgroup_free(mem);
4524
	root_mem_cgroup = NULL;
K
KAMEZAWA Hiroyuki 已提交
4525
	return ERR_PTR(error);
B
Balbir Singh 已提交
4526 4527
}

4528
static int mem_cgroup_pre_destroy(struct cgroup_subsys *ss,
4529 4530 4531
					struct cgroup *cont)
{
	struct mem_cgroup *mem = mem_cgroup_from_cont(cont);
4532 4533

	return mem_cgroup_force_empty(mem, false);
4534 4535
}

B
Balbir Singh 已提交
4536 4537 4538
static void mem_cgroup_destroy(struct cgroup_subsys *ss,
				struct cgroup *cont)
{
4539 4540 4541
	struct mem_cgroup *mem = mem_cgroup_from_cont(cont);

	mem_cgroup_put(mem);
B
Balbir Singh 已提交
4542 4543 4544 4545 4546
}

static int mem_cgroup_populate(struct cgroup_subsys *ss,
				struct cgroup *cont)
{
4547 4548 4549 4550 4551 4552 4553 4554
	int ret;

	ret = cgroup_add_files(cont, ss, mem_cgroup_files,
				ARRAY_SIZE(mem_cgroup_files));

	if (!ret)
		ret = register_memsw_files(cont, ss);
	return ret;
B
Balbir Singh 已提交
4555 4556
}

4557
#ifdef CONFIG_MMU
4558
/* Handlers for move charge at task migration. */
4559 4560
#define PRECHARGE_COUNT_AT_ONCE	256
static int mem_cgroup_do_precharge(unsigned long count)
4561
{
4562 4563
	int ret = 0;
	int batch_count = PRECHARGE_COUNT_AT_ONCE;
4564 4565
	struct mem_cgroup *mem = mc.to;

4566 4567 4568 4569 4570 4571 4572 4573 4574 4575 4576 4577 4578 4579 4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600
	if (mem_cgroup_is_root(mem)) {
		mc.precharge += count;
		/* we don't need css_get for root */
		return ret;
	}
	/* try to charge at once */
	if (count > 1) {
		struct res_counter *dummy;
		/*
		 * "mem" cannot be under rmdir() because we've already checked
		 * by cgroup_lock_live_cgroup() that it is not removed and we
		 * are still under the same cgroup_mutex. So we can postpone
		 * css_get().
		 */
		if (res_counter_charge(&mem->res, PAGE_SIZE * count, &dummy))
			goto one_by_one;
		if (do_swap_account && res_counter_charge(&mem->memsw,
						PAGE_SIZE * count, &dummy)) {
			res_counter_uncharge(&mem->res, PAGE_SIZE * count);
			goto one_by_one;
		}
		mc.precharge += count;
		return ret;
	}
one_by_one:
	/* fall back to one by one charge */
	while (count--) {
		if (signal_pending(current)) {
			ret = -EINTR;
			break;
		}
		if (!batch_count--) {
			batch_count = PRECHARGE_COUNT_AT_ONCE;
			cond_resched();
		}
A
Andrea Arcangeli 已提交
4601 4602
		ret = __mem_cgroup_try_charge(NULL, GFP_KERNEL, &mem, false,
					      PAGE_SIZE);
4603 4604 4605 4606 4607
		if (ret || !mem)
			/* mem_cgroup_clear_mc() will do uncharge later */
			return -ENOMEM;
		mc.precharge++;
	}
4608 4609 4610 4611 4612 4613 4614 4615
	return ret;
}

/**
 * is_target_pte_for_mc - check a pte whether it is valid for move charge
 * @vma: the vma the pte to be checked belongs
 * @addr: the address corresponding to the pte to be checked
 * @ptent: the pte to be checked
4616
 * @target: the pointer the target page or swap ent will be stored(can be NULL)
4617 4618 4619 4620 4621 4622
 *
 * Returns
 *   0(MC_TARGET_NONE): if the pte is not a target for move charge.
 *   1(MC_TARGET_PAGE): if the page corresponding to this pte is a target for
 *     move charge. if @target is not NULL, the page is stored in target->page
 *     with extra refcnt got(Callers should handle it).
4623 4624 4625
 *   2(MC_TARGET_SWAP): if the swap entry corresponding to this pte is a
 *     target for charge migration. if @target is not NULL, the entry is stored
 *     in target->ent.
4626 4627 4628 4629 4630
 *
 * Called with pte lock held.
 */
union mc_target {
	struct page	*page;
4631
	swp_entry_t	ent;
4632 4633 4634 4635 4636
};

enum mc_target_type {
	MC_TARGET_NONE,	/* not used */
	MC_TARGET_PAGE,
4637
	MC_TARGET_SWAP,
4638 4639
};

D
Daisuke Nishimura 已提交
4640 4641
static struct page *mc_handle_present_pte(struct vm_area_struct *vma,
						unsigned long addr, pte_t ptent)
4642
{
D
Daisuke Nishimura 已提交
4643
	struct page *page = vm_normal_page(vma, addr, ptent);
4644

D
Daisuke Nishimura 已提交
4645 4646 4647 4648 4649 4650
	if (!page || !page_mapped(page))
		return NULL;
	if (PageAnon(page)) {
		/* we don't move shared anon */
		if (!move_anon() || page_mapcount(page) > 2)
			return NULL;
4651 4652
	} else if (!move_file())
		/* we ignore mapcount for file pages */
D
Daisuke Nishimura 已提交
4653 4654 4655 4656 4657 4658 4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670
		return NULL;
	if (!get_page_unless_zero(page))
		return NULL;

	return page;
}

static struct page *mc_handle_swap_pte(struct vm_area_struct *vma,
			unsigned long addr, pte_t ptent, swp_entry_t *entry)
{
	int usage_count;
	struct page *page = NULL;
	swp_entry_t ent = pte_to_swp_entry(ptent);

	if (!move_anon() || non_swap_entry(ent))
		return NULL;
	usage_count = mem_cgroup_count_swap_user(ent, &page);
	if (usage_count > 1) { /* we don't move shared anon */
4671 4672
		if (page)
			put_page(page);
D
Daisuke Nishimura 已提交
4673
		return NULL;
4674
	}
D
Daisuke Nishimura 已提交
4675 4676 4677 4678 4679 4680
	if (do_swap_account)
		entry->val = ent.val;

	return page;
}

4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713
static struct page *mc_handle_file_pte(struct vm_area_struct *vma,
			unsigned long addr, pte_t ptent, swp_entry_t *entry)
{
	struct page *page = NULL;
	struct inode *inode;
	struct address_space *mapping;
	pgoff_t pgoff;

	if (!vma->vm_file) /* anonymous vma */
		return NULL;
	if (!move_file())
		return NULL;

	inode = vma->vm_file->f_path.dentry->d_inode;
	mapping = vma->vm_file->f_mapping;
	if (pte_none(ptent))
		pgoff = linear_page_index(vma, addr);
	else /* pte_file(ptent) is true */
		pgoff = pte_to_pgoff(ptent);

	/* page is moved even if it's not RSS of this task(page-faulted). */
	if (!mapping_cap_swap_backed(mapping)) { /* normal file */
		page = find_get_page(mapping, pgoff);
	} else { /* shmem/tmpfs file. we should take account of swap too. */
		swp_entry_t ent;
		mem_cgroup_get_shmem_target(inode, pgoff, &page, &ent);
		if (do_swap_account)
			entry->val = ent.val;
	}

	return page;
}

D
Daisuke Nishimura 已提交
4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725
static int is_target_pte_for_mc(struct vm_area_struct *vma,
		unsigned long addr, pte_t ptent, union mc_target *target)
{
	struct page *page = NULL;
	struct page_cgroup *pc;
	int ret = 0;
	swp_entry_t ent = { .val = 0 };

	if (pte_present(ptent))
		page = mc_handle_present_pte(vma, addr, ptent);
	else if (is_swap_pte(ptent))
		page = mc_handle_swap_pte(vma, addr, ptent, &ent);
4726 4727
	else if (pte_none(ptent) || pte_file(ptent))
		page = mc_handle_file_pte(vma, addr, ptent, &ent);
D
Daisuke Nishimura 已提交
4728 4729 4730

	if (!page && !ent.val)
		return 0;
4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744 4745
	if (page) {
		pc = lookup_page_cgroup(page);
		/*
		 * Do only loose check w/o page_cgroup lock.
		 * mem_cgroup_move_account() checks the pc is valid or not under
		 * the lock.
		 */
		if (PageCgroupUsed(pc) && pc->mem_cgroup == mc.from) {
			ret = MC_TARGET_PAGE;
			if (target)
				target->page = page;
		}
		if (!ret || !target)
			put_page(page);
	}
D
Daisuke Nishimura 已提交
4746 4747
	/* There is a swap entry and a page doesn't exist or isn't charged */
	if (ent.val && !ret &&
4748 4749 4750 4751
			css_id(&mc.from->css) == lookup_swap_cgroup(ent)) {
		ret = MC_TARGET_SWAP;
		if (target)
			target->ent = ent;
4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763
	}
	return ret;
}

static int mem_cgroup_count_precharge_pte_range(pmd_t *pmd,
					unsigned long addr, unsigned long end,
					struct mm_walk *walk)
{
	struct vm_area_struct *vma = walk->private;
	pte_t *pte;
	spinlock_t *ptl;

4764 4765
	split_huge_page_pmd(walk->mm, pmd);

4766 4767 4768 4769 4770 4771 4772
	pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
	for (; addr != end; pte++, addr += PAGE_SIZE)
		if (is_target_pte_for_mc(vma, addr, *pte, NULL))
			mc.precharge++;	/* increment precharge temporarily */
	pte_unmap_unlock(pte - 1, ptl);
	cond_resched();

4773 4774 4775
	return 0;
}

4776 4777 4778 4779 4780
static unsigned long mem_cgroup_count_precharge(struct mm_struct *mm)
{
	unsigned long precharge;
	struct vm_area_struct *vma;

4781
	down_read(&mm->mmap_sem);
4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792
	for (vma = mm->mmap; vma; vma = vma->vm_next) {
		struct mm_walk mem_cgroup_count_precharge_walk = {
			.pmd_entry = mem_cgroup_count_precharge_pte_range,
			.mm = mm,
			.private = vma,
		};
		if (is_vm_hugetlb_page(vma))
			continue;
		walk_page_range(vma->vm_start, vma->vm_end,
					&mem_cgroup_count_precharge_walk);
	}
4793
	up_read(&mm->mmap_sem);
4794 4795 4796 4797 4798 4799 4800 4801 4802

	precharge = mc.precharge;
	mc.precharge = 0;

	return precharge;
}

static int mem_cgroup_precharge_mc(struct mm_struct *mm)
{
4803 4804 4805 4806 4807
	unsigned long precharge = mem_cgroup_count_precharge(mm);

	VM_BUG_ON(mc.moving_task);
	mc.moving_task = current;
	return mem_cgroup_do_precharge(precharge);
4808 4809
}

4810 4811
/* cancels all extra charges on mc.from and mc.to, and wakes up all waiters. */
static void __mem_cgroup_clear_mc(void)
4812
{
4813 4814 4815
	struct mem_cgroup *from = mc.from;
	struct mem_cgroup *to = mc.to;

4816
	/* we must uncharge all the leftover precharges from mc.to */
4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827
	if (mc.precharge) {
		__mem_cgroup_cancel_charge(mc.to, mc.precharge);
		mc.precharge = 0;
	}
	/*
	 * we didn't uncharge from mc.from at mem_cgroup_move_account(), so
	 * we must uncharge here.
	 */
	if (mc.moved_charge) {
		__mem_cgroup_cancel_charge(mc.from, mc.moved_charge);
		mc.moved_charge = 0;
4828
	}
4829 4830 4831 4832 4833 4834 4835 4836 4837 4838 4839 4840 4841 4842 4843 4844 4845 4846 4847
	/* we must fixup refcnts and charges */
	if (mc.moved_swap) {
		/* uncharge swap account from the old cgroup */
		if (!mem_cgroup_is_root(mc.from))
			res_counter_uncharge(&mc.from->memsw,
						PAGE_SIZE * mc.moved_swap);
		__mem_cgroup_put(mc.from, mc.moved_swap);

		if (!mem_cgroup_is_root(mc.to)) {
			/*
			 * we charged both to->res and to->memsw, so we should
			 * uncharge to->res.
			 */
			res_counter_uncharge(&mc.to->res,
						PAGE_SIZE * mc.moved_swap);
		}
		/* we've already done mem_cgroup_get(mc.to) */
		mc.moved_swap = 0;
	}
4848 4849 4850 4851 4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862
	memcg_oom_recover(from);
	memcg_oom_recover(to);
	wake_up_all(&mc.waitq);
}

static void mem_cgroup_clear_mc(void)
{
	struct mem_cgroup *from = mc.from;

	/*
	 * we must clear moving_task before waking up waiters at the end of
	 * task migration.
	 */
	mc.moving_task = NULL;
	__mem_cgroup_clear_mc();
4863
	spin_lock(&mc.lock);
4864 4865
	mc.from = NULL;
	mc.to = NULL;
4866
	spin_unlock(&mc.lock);
4867
	mem_cgroup_end_move(from);
4868 4869
}

4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880 4881 4882 4883 4884 4885 4886 4887
static int mem_cgroup_can_attach(struct cgroup_subsys *ss,
				struct cgroup *cgroup,
				struct task_struct *p,
				bool threadgroup)
{
	int ret = 0;
	struct mem_cgroup *mem = mem_cgroup_from_cont(cgroup);

	if (mem->move_charge_at_immigrate) {
		struct mm_struct *mm;
		struct mem_cgroup *from = mem_cgroup_from_task(p);

		VM_BUG_ON(from == mem);

		mm = get_task_mm(p);
		if (!mm)
			return 0;
		/* We move charges only when we move a owner of the mm */
4888 4889 4890 4891
		if (mm->owner == p) {
			VM_BUG_ON(mc.from);
			VM_BUG_ON(mc.to);
			VM_BUG_ON(mc.precharge);
4892
			VM_BUG_ON(mc.moved_charge);
4893
			VM_BUG_ON(mc.moved_swap);
4894
			mem_cgroup_start_move(from);
4895
			spin_lock(&mc.lock);
4896 4897
			mc.from = from;
			mc.to = mem;
4898
			spin_unlock(&mc.lock);
4899
			/* We set mc.moving_task later */
4900 4901 4902 4903

			ret = mem_cgroup_precharge_mc(mm);
			if (ret)
				mem_cgroup_clear_mc();
4904 4905
		}
		mmput(mm);
4906 4907 4908 4909 4910 4911 4912 4913 4914
	}
	return ret;
}

static void mem_cgroup_cancel_attach(struct cgroup_subsys *ss,
				struct cgroup *cgroup,
				struct task_struct *p,
				bool threadgroup)
{
4915
	mem_cgroup_clear_mc();
4916 4917
}

4918 4919 4920
static int mem_cgroup_move_charge_pte_range(pmd_t *pmd,
				unsigned long addr, unsigned long end,
				struct mm_walk *walk)
4921
{
4922 4923 4924 4925 4926
	int ret = 0;
	struct vm_area_struct *vma = walk->private;
	pte_t *pte;
	spinlock_t *ptl;

4927
	split_huge_page_pmd(walk->mm, pmd);
4928 4929 4930 4931 4932 4933 4934 4935
retry:
	pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
	for (; addr != end; addr += PAGE_SIZE) {
		pte_t ptent = *(pte++);
		union mc_target target;
		int type;
		struct page *page;
		struct page_cgroup *pc;
4936
		swp_entry_t ent;
4937 4938 4939 4940 4941 4942 4943 4944 4945 4946 4947

		if (!mc.precharge)
			break;

		type = is_target_pte_for_mc(vma, addr, ptent, &target);
		switch (type) {
		case MC_TARGET_PAGE:
			page = target.page;
			if (isolate_lru_page(page))
				goto put;
			pc = lookup_page_cgroup(page);
4948
			if (!mem_cgroup_move_account(page, pc,
4949
					mc.from, mc.to, false, PAGE_SIZE)) {
4950
				mc.precharge--;
4951 4952
				/* we uncharge from mc.from later. */
				mc.moved_charge++;
4953 4954 4955 4956 4957
			}
			putback_lru_page(page);
put:			/* is_target_pte_for_mc() gets the page */
			put_page(page);
			break;
4958 4959
		case MC_TARGET_SWAP:
			ent = target.ent;
4960 4961
			if (!mem_cgroup_move_swap_account(ent,
						mc.from, mc.to, false)) {
4962
				mc.precharge--;
4963 4964 4965
				/* we fixup refcnts and charges later. */
				mc.moved_swap++;
			}
4966
			break;
4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980
		default:
			break;
		}
	}
	pte_unmap_unlock(pte - 1, ptl);
	cond_resched();

	if (addr != end) {
		/*
		 * We have consumed all precharges we got in can_attach().
		 * We try charge one by one, but don't do any additional
		 * charges to mc.to if we have failed in charge once in attach()
		 * phase.
		 */
4981
		ret = mem_cgroup_do_precharge(1);
4982 4983 4984 4985 4986 4987 4988 4989 4990 4991 4992 4993
		if (!ret)
			goto retry;
	}

	return ret;
}

static void mem_cgroup_move_charge(struct mm_struct *mm)
{
	struct vm_area_struct *vma;

	lru_add_drain_all();
4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006
retry:
	if (unlikely(!down_read_trylock(&mm->mmap_sem))) {
		/*
		 * Someone who are holding the mmap_sem might be waiting in
		 * waitq. So we cancel all extra charges, wake up all waiters,
		 * and retry. Because we cancel precharges, we might not be able
		 * to move enough charges, but moving charge is a best-effort
		 * feature anyway, so it wouldn't be a big problem.
		 */
		__mem_cgroup_clear_mc();
		cond_resched();
		goto retry;
	}
5007 5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024
	for (vma = mm->mmap; vma; vma = vma->vm_next) {
		int ret;
		struct mm_walk mem_cgroup_move_charge_walk = {
			.pmd_entry = mem_cgroup_move_charge_pte_range,
			.mm = mm,
			.private = vma,
		};
		if (is_vm_hugetlb_page(vma))
			continue;
		ret = walk_page_range(vma->vm_start, vma->vm_end,
						&mem_cgroup_move_charge_walk);
		if (ret)
			/*
			 * means we have consumed all precharges and failed in
			 * doing additional charge. Just abandon here.
			 */
			break;
	}
5025
	up_read(&mm->mmap_sem);
5026 5027
}

B
Balbir Singh 已提交
5028 5029 5030
static void mem_cgroup_move_task(struct cgroup_subsys *ss,
				struct cgroup *cont,
				struct cgroup *old_cont,
5031 5032
				struct task_struct *p,
				bool threadgroup)
B
Balbir Singh 已提交
5033
{
5034 5035 5036
	struct mm_struct *mm;

	if (!mc.to)
5037 5038 5039
		/* no need to move charge */
		return;

5040 5041 5042 5043 5044
	mm = get_task_mm(p);
	if (mm) {
		mem_cgroup_move_charge(mm);
		mmput(mm);
	}
5045
	mem_cgroup_clear_mc();
B
Balbir Singh 已提交
5046
}
5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067 5068
#else	/* !CONFIG_MMU */
static int mem_cgroup_can_attach(struct cgroup_subsys *ss,
				struct cgroup *cgroup,
				struct task_struct *p,
				bool threadgroup)
{
	return 0;
}
static void mem_cgroup_cancel_attach(struct cgroup_subsys *ss,
				struct cgroup *cgroup,
				struct task_struct *p,
				bool threadgroup)
{
}
static void mem_cgroup_move_task(struct cgroup_subsys *ss,
				struct cgroup *cont,
				struct cgroup *old_cont,
				struct task_struct *p,
				bool threadgroup)
{
}
#endif
B
Balbir Singh 已提交
5069

B
Balbir Singh 已提交
5070 5071 5072 5073
struct cgroup_subsys mem_cgroup_subsys = {
	.name = "memory",
	.subsys_id = mem_cgroup_subsys_id,
	.create = mem_cgroup_create,
5074
	.pre_destroy = mem_cgroup_pre_destroy,
B
Balbir Singh 已提交
5075 5076
	.destroy = mem_cgroup_destroy,
	.populate = mem_cgroup_populate,
5077 5078
	.can_attach = mem_cgroup_can_attach,
	.cancel_attach = mem_cgroup_cancel_attach,
B
Balbir Singh 已提交
5079
	.attach = mem_cgroup_move_task,
5080
	.early_init = 0,
K
KAMEZAWA Hiroyuki 已提交
5081
	.use_id = 1,
B
Balbir Singh 已提交
5082
};
5083 5084

#ifdef CONFIG_CGROUP_MEM_RES_CTLR_SWAP
5085 5086 5087
static int __init enable_swap_account(char *s)
{
	/* consider enabled if no parameter or 1 is given */
5088
	if (!(*s) || !strcmp(s, "=1"))
5089
		really_do_swap_account = 1;
5090
	else if (!strcmp(s, "=0"))
5091 5092 5093 5094
		really_do_swap_account = 0;
	return 1;
}
__setup("swapaccount", enable_swap_account);
5095 5096 5097

static int __init disable_swap_account(char *s)
{
5098
	printk_once("noswapaccount is deprecated and will be removed in 2.6.40. Use swapaccount=0 instead\n");
5099
	enable_swap_account("=0");
5100 5101 5102 5103
	return 1;
}
__setup("noswapaccount", disable_swap_account);
#endif