memcontrol.c 145.4 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 76
#else
#define do_swap_account		(0)
#endif


77 78 79 80 81 82 83 84
/*
 * 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 */
85
	MEM_CGROUP_STAT_RSS,	   /* # of pages charged as anon rss */
86
	MEM_CGROUP_STAT_FILE_MAPPED,  /* # of pages charged as file rss */
87
	MEM_CGROUP_STAT_SWAPOUT, /* # of pages, swapped out */
88
	MEM_CGROUP_STAT_DATA, /* end of data requires synchronization */
89
	MEM_CGROUP_ON_MOVE,	/* someone is moving account between groups */
90 91 92
	MEM_CGROUP_STAT_NSTATS,
};

93 94 95 96
enum mem_cgroup_events_index {
	MEM_CGROUP_EVENTS_PGPGIN,	/* # of pages paged in */
	MEM_CGROUP_EVENTS_PGPGOUT,	/* # of pages paged out */
	MEM_CGROUP_EVENTS_COUNT,	/* # of pages paged in/out */
97 98
	MEM_CGROUP_EVENTS_PGFAULT,	/* # of page-faults */
	MEM_CGROUP_EVENTS_PGMAJFAULT,	/* # of major page-faults */
99 100
	MEM_CGROUP_EVENTS_NSTATS,
};
101 102 103 104 105 106 107 108 109
/*
 * Per memcg event counter is incremented at every pagein/pageout. With THP,
 * it will be incremated by the number of pages. This counter is used for
 * for trigger some periodic events. This is straightforward and better
 * than using jiffies etc. to handle periodic memcg event.
 */
enum mem_cgroup_events_target {
	MEM_CGROUP_TARGET_THRESH,
	MEM_CGROUP_TARGET_SOFTLIMIT,
110
	MEM_CGROUP_TARGET_NUMAINFO,
111 112 113 114
	MEM_CGROUP_NTARGETS,
};
#define THRESHOLDS_EVENTS_TARGET (128)
#define SOFTLIMIT_EVENTS_TARGET (1024)
115
#define NUMAINFO_EVENTS_TARGET	(1024)
116

117
struct mem_cgroup_stat_cpu {
118
	long count[MEM_CGROUP_STAT_NSTATS];
119
	unsigned long events[MEM_CGROUP_EVENTS_NSTATS];
120
	unsigned long targets[MEM_CGROUP_NTARGETS];
121 122
};

123 124 125 126
/*
 * per-zone information in memory controller.
 */
struct mem_cgroup_per_zone {
127 128 129
	/*
	 * spin_lock to protect the per cgroup LRU
	 */
130 131
	struct list_head	lists[NR_LRU_LISTS];
	unsigned long		count[NR_LRU_LISTS];
K
KOSAKI Motohiro 已提交
132 133

	struct zone_reclaim_stat reclaim_stat;
134 135 136 137
	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;
138 139
	struct mem_cgroup	*mem;		/* Back pointer, we cannot */
						/* use container_of	   */
140 141 142 143 144 145 146 147 148 149 150 151
};
/* 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];
};

152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171
/*
 * 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;

172 173 174 175 176
struct mem_cgroup_threshold {
	struct eventfd_ctx *eventfd;
	u64 threshold;
};

K
KAMEZAWA Hiroyuki 已提交
177
/* For threshold */
178 179
struct mem_cgroup_threshold_ary {
	/* An array index points to threshold just below usage. */
180
	int current_threshold;
181 182 183 184 185
	/* Size of entries[] */
	unsigned int size;
	/* Array of thresholds */
	struct mem_cgroup_threshold entries[0];
};
186 187 188 189 190 191 192 193 194 195 196 197

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
198 199 200 201 202
/* for OOM */
struct mem_cgroup_eventfd_list {
	struct list_head list;
	struct eventfd_ctx *eventfd;
};
203 204

static void mem_cgroup_threshold(struct mem_cgroup *mem);
K
KAMEZAWA Hiroyuki 已提交
205
static void mem_cgroup_oom_notify(struct mem_cgroup *mem);
206

207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250
enum {
	SCAN_BY_LIMIT,
	SCAN_BY_SYSTEM,
	NR_SCAN_CONTEXT,
	SCAN_BY_SHRINK,	/* not recorded now */
};

enum {
	SCAN,
	SCAN_ANON,
	SCAN_FILE,
	ROTATE,
	ROTATE_ANON,
	ROTATE_FILE,
	FREED,
	FREED_ANON,
	FREED_FILE,
	ELAPSED,
	NR_SCANSTATS,
};

struct scanstat {
	spinlock_t	lock;
	unsigned long	stats[NR_SCAN_CONTEXT][NR_SCANSTATS];
	unsigned long	rootstats[NR_SCAN_CONTEXT][NR_SCANSTATS];
};

const char *scanstat_string[NR_SCANSTATS] = {
	"scanned_pages",
	"scanned_anon_pages",
	"scanned_file_pages",
	"rotated_pages",
	"rotated_anon_pages",
	"rotated_file_pages",
	"freed_pages",
	"freed_anon_pages",
	"freed_file_pages",
	"elapsed_ns",
};
#define SCANSTAT_WORD_LIMIT	"_by_limit"
#define SCANSTAT_WORD_SYSTEM	"_by_system"
#define SCANSTAT_WORD_HIERARCHY	"_under_hierarchy"


B
Balbir Singh 已提交
251 252 253 254 255 256 257
/*
 * 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
258 259 260
 * 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
261 262 263 264 265 266 267
 */
struct mem_cgroup {
	struct cgroup_subsys_state css;
	/*
	 * the counter to account for memory usage
	 */
	struct res_counter res;
268 269 270 271
	/*
	 * the counter to account for mem+swap usage.
	 */
	struct res_counter memsw;
272 273 274 275
	/*
	 * Per cgroup active and inactive list, similar to the
	 * per zone LRU lists.
	 */
276
	struct mem_cgroup_lru_info info;
277
	/*
278
	 * While reclaiming in a hierarchy, we cache the last child we
K
KAMEZAWA Hiroyuki 已提交
279
	 * reclaimed from.
280
	 */
K
KAMEZAWA Hiroyuki 已提交
281
	int last_scanned_child;
282 283 284
	int last_scanned_node;
#if MAX_NUMNODES > 1
	nodemask_t	scan_nodes;
285 286
	atomic_t	numainfo_events;
	atomic_t	numainfo_updating;
287
#endif
288 289 290 291
	/*
	 * Should the accounting and control be hierarchical, per subtree?
	 */
	bool use_hierarchy;
292 293 294 295

	bool		oom_lock;
	atomic_t	under_oom;

296
	atomic_t	refcnt;
297

298
	int	swappiness;
299 300
	/* OOM-Killer disable */
	int		oom_kill_disable;
K
KOSAKI Motohiro 已提交
301

302 303 304
	/* set when res.limit == memsw.limit */
	bool		memsw_is_minimum;

305 306 307 308
	/* protect arrays of thresholds */
	struct mutex thresholds_lock;

	/* thresholds for memory usage. RCU-protected */
309
	struct mem_cgroup_thresholds thresholds;
310

311
	/* thresholds for mem+swap usage. RCU-protected */
312
	struct mem_cgroup_thresholds memsw_thresholds;
313

K
KAMEZAWA Hiroyuki 已提交
314 315
	/* For oom notifier event fd */
	struct list_head oom_notify;
316 317
	/* For recording LRU-scan statistics */
	struct scanstat scanstat;
318 319 320 321 322
	/*
	 * 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;
323
	/*
324
	 * percpu counter.
325
	 */
326
	struct mem_cgroup_stat_cpu *stat;
327 328 329 330 331 332
	/*
	 * 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
333 334
};

335 336 337 338 339 340
/* 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 {
341
	MOVE_CHARGE_TYPE_ANON,	/* private anonymous page and swap of it */
342
	MOVE_CHARGE_TYPE_FILE,	/* file page(including tmpfs) and swap of it */
343 344 345
	NR_MOVE_TYPE,
};

346 347
/* "mc" and its members are protected by cgroup_mutex */
static struct move_charge_struct {
348
	spinlock_t	  lock; /* for from, to */
349 350 351
	struct mem_cgroup *from;
	struct mem_cgroup *to;
	unsigned long precharge;
352
	unsigned long moved_charge;
353
	unsigned long moved_swap;
354 355 356
	struct task_struct *moving_task;	/* a task moving charges */
	wait_queue_head_t waitq;		/* a waitq for other context */
} mc = {
357
	.lock = __SPIN_LOCK_UNLOCKED(mc.lock),
358 359
	.waitq = __WAIT_QUEUE_HEAD_INITIALIZER(mc.waitq),
};
360

D
Daisuke Nishimura 已提交
361 362 363 364 365 366
static bool move_anon(void)
{
	return test_bit(MOVE_CHARGE_TYPE_ANON,
					&mc.to->move_charge_at_immigrate);
}

367 368 369 370 371 372
static bool move_file(void)
{
	return test_bit(MOVE_CHARGE_TYPE_FILE,
					&mc.to->move_charge_at_immigrate);
}

373 374 375 376 377 378 379
/*
 * 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)

380 381 382
enum charge_type {
	MEM_CGROUP_CHARGE_TYPE_CACHE = 0,
	MEM_CGROUP_CHARGE_TYPE_MAPPED,
383
	MEM_CGROUP_CHARGE_TYPE_SHMEM,	/* used by page migration of shmem */
384
	MEM_CGROUP_CHARGE_TYPE_FORCE,	/* used by force_empty */
K
KAMEZAWA Hiroyuki 已提交
385
	MEM_CGROUP_CHARGE_TYPE_SWAPOUT,	/* for accounting swapcache */
K
KAMEZAWA Hiroyuki 已提交
386
	MEM_CGROUP_CHARGE_TYPE_DROP,	/* a page was unused swap cache */
387 388 389
	NR_CHARGE_TYPE,
};

390 391 392
/* for encoding cft->private value on file */
#define _MEM			(0)
#define _MEMSWAP		(1)
K
KAMEZAWA Hiroyuki 已提交
393
#define _OOM_TYPE		(2)
394 395 396
#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
397 398
/* Used for OOM nofiier */
#define OOM_CONTROL		(0)
399

400 401 402 403 404 405 406
/*
 * 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)
407 408
#define MEM_CGROUP_RECLAIM_SOFT_BIT	0x2
#define MEM_CGROUP_RECLAIM_SOFT		(1 << MEM_CGROUP_RECLAIM_SOFT_BIT)
409

410 411
static void mem_cgroup_get(struct mem_cgroup *mem);
static void mem_cgroup_put(struct mem_cgroup *mem);
412
static struct mem_cgroup *parent_mem_cgroup(struct mem_cgroup *mem);
413
static void drain_all_stock_async(struct mem_cgroup *mem);
414

415 416 417 418 419 420
static struct mem_cgroup_per_zone *
mem_cgroup_zoneinfo(struct mem_cgroup *mem, int nid, int zid)
{
	return &mem->info.nodeinfo[nid]->zoneinfo[zid];
}

421 422 423 424 425
struct cgroup_subsys_state *mem_cgroup_css(struct mem_cgroup *mem)
{
	return &mem->css;
}

426
static struct mem_cgroup_per_zone *
427
page_cgroup_zoneinfo(struct mem_cgroup *mem, struct page *page)
428
{
429 430
	int nid = page_to_nid(page);
	int zid = page_zonenum(page);
431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450

	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
451
__mem_cgroup_insert_exceeded(struct mem_cgroup *mem,
452
				struct mem_cgroup_per_zone *mz,
453 454
				struct mem_cgroup_tree_per_zone *mctz,
				unsigned long long new_usage_in_excess)
455 456 457 458 459 460 461 462
{
	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;

463 464 465
	mz->usage_in_excess = new_usage_in_excess;
	if (!mz->usage_in_excess)
		return;
466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481
	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;
482 483 484 485 486 487 488 489 490 491 492 493 494
}

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;
}

495 496 497 498 499 500
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);
501
	__mem_cgroup_remove_exceeded(mem, mz, mctz);
502 503 504 505 506 507
	spin_unlock(&mctz->lock);
}


static void mem_cgroup_update_tree(struct mem_cgroup *mem, struct page *page)
{
508
	unsigned long long excess;
509 510
	struct mem_cgroup_per_zone *mz;
	struct mem_cgroup_tree_per_zone *mctz;
511 512
	int nid = page_to_nid(page);
	int zid = page_zonenum(page);
513 514 515
	mctz = soft_limit_tree_from_page(page);

	/*
516 517
	 * Necessary to update all ancestors when hierarchy is used.
	 * because their event counter is not touched.
518
	 */
519 520
	for (; mem; mem = parent_mem_cgroup(mem)) {
		mz = mem_cgroup_zoneinfo(mem, nid, zid);
521
		excess = res_counter_soft_limit_excess(&mem->res);
522 523 524 525
		/*
		 * We have to update the tree if mz is on RB-tree or
		 * mem is over its softlimit.
		 */
526
		if (excess || mz->on_tree) {
527 528 529 530 531
			spin_lock(&mctz->lock);
			/* if on-tree, remove it */
			if (mz->on_tree)
				__mem_cgroup_remove_exceeded(mem, mz, mctz);
			/*
532 533
			 * Insert again. mz->usage_in_excess will be updated.
			 * If excess is 0, no tree ops.
534
			 */
535
			__mem_cgroup_insert_exceeded(mem, mz, mctz, excess);
536 537
			spin_unlock(&mctz->lock);
		}
538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555
	}
}

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);
		}
	}
}

556 557 558 559
static struct mem_cgroup_per_zone *
__mem_cgroup_largest_soft_limit_node(struct mem_cgroup_tree_per_zone *mctz)
{
	struct rb_node *rightmost = NULL;
560
	struct mem_cgroup_per_zone *mz;
561 562

retry:
563
	mz = NULL;
564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592
	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;
}

593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611
/*
 * 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.
 */
612 613
static long mem_cgroup_read_stat(struct mem_cgroup *mem,
				 enum mem_cgroup_stat_index idx)
614
{
615
	long val = 0;
616 617
	int cpu;

618 619
	get_online_cpus();
	for_each_online_cpu(cpu)
620
		val += per_cpu(mem->stat->count[idx], cpu);
621 622 623 624 625 626
#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();
627 628 629
	return val;
}

630 631 632 633
static void mem_cgroup_swap_statistics(struct mem_cgroup *mem,
					 bool charge)
{
	int val = (charge) ? 1 : -1;
634
	this_cpu_add(mem->stat->count[MEM_CGROUP_STAT_SWAPOUT], val);
635 636
}

637 638 639 640 641 642 643 644 645 646
void mem_cgroup_pgfault(struct mem_cgroup *mem, int val)
{
	this_cpu_add(mem->stat->events[MEM_CGROUP_EVENTS_PGFAULT], val);
}

void mem_cgroup_pgmajfault(struct mem_cgroup *mem, int val)
{
	this_cpu_add(mem->stat->events[MEM_CGROUP_EVENTS_PGMAJFAULT], val);
}

647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662
static unsigned long mem_cgroup_read_events(struct mem_cgroup *mem,
					    enum mem_cgroup_events_index idx)
{
	unsigned long val = 0;
	int cpu;

	for_each_online_cpu(cpu)
		val += per_cpu(mem->stat->events[idx], cpu);
#ifdef CONFIG_HOTPLUG_CPU
	spin_lock(&mem->pcp_counter_lock);
	val += mem->nocpu_base.events[idx];
	spin_unlock(&mem->pcp_counter_lock);
#endif
	return val;
}

663
static void mem_cgroup_charge_statistics(struct mem_cgroup *mem,
664
					 bool file, int nr_pages)
665
{
666 667
	preempt_disable();

668 669
	if (file)
		__this_cpu_add(mem->stat->count[MEM_CGROUP_STAT_CACHE], nr_pages);
670
	else
671
		__this_cpu_add(mem->stat->count[MEM_CGROUP_STAT_RSS], nr_pages);
672

673 674
	/* pagein of a big page is an event. So, ignore page size */
	if (nr_pages > 0)
675
		__this_cpu_inc(mem->stat->events[MEM_CGROUP_EVENTS_PGPGIN]);
676
	else {
677
		__this_cpu_inc(mem->stat->events[MEM_CGROUP_EVENTS_PGPGOUT]);
678 679
		nr_pages = -nr_pages; /* for event */
	}
680

681
	__this_cpu_add(mem->stat->events[MEM_CGROUP_EVENTS_COUNT], nr_pages);
682

683
	preempt_enable();
684 685
}

686 687 688
unsigned long
mem_cgroup_zone_nr_lru_pages(struct mem_cgroup *mem, int nid, int zid,
			unsigned int lru_mask)
689 690
{
	struct mem_cgroup_per_zone *mz;
691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706
	enum lru_list l;
	unsigned long ret = 0;

	mz = mem_cgroup_zoneinfo(mem, nid, zid);

	for_each_lru(l) {
		if (BIT(l) & lru_mask)
			ret += MEM_CGROUP_ZSTAT(mz, l);
	}
	return ret;
}

static unsigned long
mem_cgroup_node_nr_lru_pages(struct mem_cgroup *mem,
			int nid, unsigned int lru_mask)
{
707 708 709
	u64 total = 0;
	int zid;

710 711 712
	for (zid = 0; zid < MAX_NR_ZONES; zid++)
		total += mem_cgroup_zone_nr_lru_pages(mem, nid, zid, lru_mask);

713 714
	return total;
}
715 716 717

static unsigned long mem_cgroup_nr_lru_pages(struct mem_cgroup *mem,
			unsigned int lru_mask)
718
{
719
	int nid;
720 721
	u64 total = 0;

722 723
	for_each_node_state(nid, N_HIGH_MEMORY)
		total += mem_cgroup_node_nr_lru_pages(mem, nid, lru_mask);
724
	return total;
725 726
}

727 728 729 730 731 732 733 734 735 736 737
static bool __memcg_event_check(struct mem_cgroup *mem, int target)
{
	unsigned long val, next;

	val = this_cpu_read(mem->stat->events[MEM_CGROUP_EVENTS_COUNT]);
	next = this_cpu_read(mem->stat->targets[target]);
	/* from time_after() in jiffies.h */
	return ((long)next - (long)val < 0);
}

static void __mem_cgroup_target_update(struct mem_cgroup *mem, int target)
738
{
739
	unsigned long val, next;
740

741
	val = this_cpu_read(mem->stat->events[MEM_CGROUP_EVENTS_COUNT]);
742

743 744 745 746 747 748 749
	switch (target) {
	case MEM_CGROUP_TARGET_THRESH:
		next = val + THRESHOLDS_EVENTS_TARGET;
		break;
	case MEM_CGROUP_TARGET_SOFTLIMIT:
		next = val + SOFTLIMIT_EVENTS_TARGET;
		break;
750 751 752
	case MEM_CGROUP_TARGET_NUMAINFO:
		next = val + NUMAINFO_EVENTS_TARGET;
		break;
753 754 755 756 757
	default:
		return;
	}

	this_cpu_write(mem->stat->targets[target], next);
758 759 760 761 762 763 764 765 766
}

/*
 * 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 */
767
	if (unlikely(__memcg_event_check(mem, MEM_CGROUP_TARGET_THRESH))) {
768
		mem_cgroup_threshold(mem);
769 770
		__mem_cgroup_target_update(mem, MEM_CGROUP_TARGET_THRESH);
		if (unlikely(__memcg_event_check(mem,
771
			     MEM_CGROUP_TARGET_SOFTLIMIT))) {
772
			mem_cgroup_update_tree(mem, page);
773
			__mem_cgroup_target_update(mem,
774 775 776 777 778 779 780 781
						   MEM_CGROUP_TARGET_SOFTLIMIT);
		}
#if MAX_NUMNODES > 1
		if (unlikely(__memcg_event_check(mem,
			MEM_CGROUP_TARGET_NUMAINFO))) {
			atomic_inc(&mem->numainfo_events);
			__mem_cgroup_target_update(mem,
				MEM_CGROUP_TARGET_NUMAINFO);
782
		}
783
#endif
784 785 786
	}
}

787
static struct mem_cgroup *mem_cgroup_from_cont(struct cgroup *cont)
B
Balbir Singh 已提交
788 789 790 791 792 793
{
	return container_of(cgroup_subsys_state(cont,
				mem_cgroup_subsys_id), struct mem_cgroup,
				css);
}

794
struct mem_cgroup *mem_cgroup_from_task(struct task_struct *p)
795
{
796 797 798 799 800 801 802 803
	/*
	 * 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;

804 805 806 807
	return container_of(task_subsys_state(p, mem_cgroup_subsys_id),
				struct mem_cgroup, css);
}

808
struct mem_cgroup *try_get_mem_cgroup_from_mm(struct mm_struct *mm)
809 810
{
	struct mem_cgroup *mem = NULL;
811 812 813

	if (!mm)
		return NULL;
814 815 816 817 818 819 820 821 822 823 824 825 826 827 828
	/*
	 * 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
829 830
/* 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
831
{
832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853
	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
854 855 856 857 858 859 860 861 862
}

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
863 864
	struct cgroup_subsys_state *css;

K
KAMEZAWA Hiroyuki 已提交
865
	hierarchy_used = iter->use_hierarchy;
K
KAMEZAWA Hiroyuki 已提交
866

K
KAMEZAWA Hiroyuki 已提交
867
	css_put(&iter->css);
868 869
	/* If no ROOT, walk all, ignore hierarchy */
	if (!cond || (root && !hierarchy_used))
K
KAMEZAWA Hiroyuki 已提交
870
		return NULL;
K
KAMEZAWA Hiroyuki 已提交
871

872 873 874
	if (!root)
		root = root_mem_cgroup;

K
KAMEZAWA Hiroyuki 已提交
875 876
	do {
		iter = NULL;
K
KAMEZAWA Hiroyuki 已提交
877
		rcu_read_lock();
K
KAMEZAWA Hiroyuki 已提交
878 879 880

		css = css_get_next(&mem_cgroup_subsys, nextid,
				&root->css, &found);
K
KAMEZAWA Hiroyuki 已提交
881
		if (css && css_tryget(css))
K
KAMEZAWA Hiroyuki 已提交
882
			iter = container_of(css, struct mem_cgroup, css);
K
KAMEZAWA Hiroyuki 已提交
883
		rcu_read_unlock();
K
KAMEZAWA Hiroyuki 已提交
884
		/* If css is NULL, no more cgroups will be found */
K
KAMEZAWA Hiroyuki 已提交
885
		nextid = found + 1;
K
KAMEZAWA Hiroyuki 已提交
886
	} while (css && !iter);
K
KAMEZAWA Hiroyuki 已提交
887

K
KAMEZAWA Hiroyuki 已提交
888
	return iter;
K
KAMEZAWA Hiroyuki 已提交
889
}
K
KAMEZAWA Hiroyuki 已提交
890 891 892 893 894 895 896 897 898 899 900 901 902
/*
 * 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)

903 904 905
#define for_each_mem_cgroup_all(iter) \
	for_each_mem_cgroup_tree_cond(iter, NULL, true)

K
KAMEZAWA Hiroyuki 已提交
906

907 908 909 910 911
static inline bool mem_cgroup_is_root(struct mem_cgroup *mem)
{
	return (mem == root_mem_cgroup);
}

912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938
void mem_cgroup_count_vm_event(struct mm_struct *mm, enum vm_event_item idx)
{
	struct mem_cgroup *mem;

	if (!mm)
		return;

	rcu_read_lock();
	mem = mem_cgroup_from_task(rcu_dereference(mm->owner));
	if (unlikely(!mem))
		goto out;

	switch (idx) {
	case PGMAJFAULT:
		mem_cgroup_pgmajfault(mem, 1);
		break;
	case PGFAULT:
		mem_cgroup_pgfault(mem, 1);
		break;
	default:
		BUG();
	}
out:
	rcu_read_unlock();
}
EXPORT_SYMBOL(mem_cgroup_count_vm_event);

K
KAMEZAWA Hiroyuki 已提交
939 940 941 942 943 944 945 946 947 948 949 950 951
/*
 * 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.
 */
952

K
KAMEZAWA Hiroyuki 已提交
953 954 955 956
void mem_cgroup_del_lru_list(struct page *page, enum lru_list lru)
{
	struct page_cgroup *pc;
	struct mem_cgroup_per_zone *mz;
957

958
	if (mem_cgroup_disabled())
K
KAMEZAWA Hiroyuki 已提交
959 960 961
		return;
	pc = lookup_page_cgroup(page);
	/* can happen while we handle swapcache. */
962
	if (!TestClearPageCgroupAcctLRU(pc))
K
KAMEZAWA Hiroyuki 已提交
963
		return;
964
	VM_BUG_ON(!pc->mem_cgroup);
965 966 967 968
	/*
	 * We don't check PCG_USED bit. It's cleared when the "page" is finally
	 * removed from global LRU.
	 */
969
	mz = page_cgroup_zoneinfo(pc->mem_cgroup, page);
970 971
	/* huge page split is done under lru_lock. so, we have no races. */
	MEM_CGROUP_ZSTAT(mz, lru) -= 1 << compound_order(page);
972 973 974
	if (mem_cgroup_is_root(pc->mem_cgroup))
		return;
	VM_BUG_ON(list_empty(&pc->lru));
K
KAMEZAWA Hiroyuki 已提交
975
	list_del_init(&pc->lru);
976 977
}

K
KAMEZAWA Hiroyuki 已提交
978
void mem_cgroup_del_lru(struct page *page)
979
{
K
KAMEZAWA Hiroyuki 已提交
980 981
	mem_cgroup_del_lru_list(page, page_lru(page));
}
982

983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004
/*
 * 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;
1005
	mz = page_cgroup_zoneinfo(pc->mem_cgroup, page);
1006 1007 1008
	list_move_tail(&pc->lru, &mz->lists[lru]);
}

K
KAMEZAWA Hiroyuki 已提交
1009 1010 1011 1012
void mem_cgroup_rotate_lru_list(struct page *page, enum lru_list lru)
{
	struct mem_cgroup_per_zone *mz;
	struct page_cgroup *pc;
1013

1014
	if (mem_cgroup_disabled())
K
KAMEZAWA Hiroyuki 已提交
1015
		return;
1016

K
KAMEZAWA Hiroyuki 已提交
1017
	pc = lookup_page_cgroup(page);
1018
	/* unused or root page is not rotated. */
1019 1020 1021 1022 1023
	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
1024
		return;
1025
	mz = page_cgroup_zoneinfo(pc->mem_cgroup, page);
K
KAMEZAWA Hiroyuki 已提交
1026
	list_move(&pc->lru, &mz->lists[lru]);
1027 1028
}

K
KAMEZAWA Hiroyuki 已提交
1029
void mem_cgroup_add_lru_list(struct page *page, enum lru_list lru)
1030
{
K
KAMEZAWA Hiroyuki 已提交
1031 1032
	struct page_cgroup *pc;
	struct mem_cgroup_per_zone *mz;
1033

1034
	if (mem_cgroup_disabled())
K
KAMEZAWA Hiroyuki 已提交
1035 1036
		return;
	pc = lookup_page_cgroup(page);
1037
	VM_BUG_ON(PageCgroupAcctLRU(pc));
K
KAMEZAWA Hiroyuki 已提交
1038
	if (!PageCgroupUsed(pc))
L
Lee Schermerhorn 已提交
1039
		return;
1040 1041
	/* Ensure pc->mem_cgroup is visible after reading PCG_USED. */
	smp_rmb();
1042
	mz = page_cgroup_zoneinfo(pc->mem_cgroup, page);
1043 1044
	/* huge page split is done under lru_lock. so, we have no races. */
	MEM_CGROUP_ZSTAT(mz, lru) += 1 << compound_order(page);
1045 1046 1047
	SetPageCgroupAcctLRU(pc);
	if (mem_cgroup_is_root(pc->mem_cgroup))
		return;
K
KAMEZAWA Hiroyuki 已提交
1048 1049
	list_add(&pc->lru, &mz->lists[lru]);
}
1050

K
KAMEZAWA Hiroyuki 已提交
1051
/*
1052 1053 1054 1055
 * At handling SwapCache and other FUSE stuff, pc->mem_cgroup may be changed
 * while it's linked to lru because the page may be reused after it's fully
 * uncharged. To handle that, unlink page_cgroup from LRU when charge it again.
 * It's done under lock_page and expected that zone->lru_lock isnever held.
K
KAMEZAWA Hiroyuki 已提交
1056
 */
1057
static void mem_cgroup_lru_del_before_commit(struct page *page)
K
KAMEZAWA Hiroyuki 已提交
1058
{
1059 1060 1061 1062
	unsigned long flags;
	struct zone *zone = page_zone(page);
	struct page_cgroup *pc = lookup_page_cgroup(page);

1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073
	/*
	 * Doing this check without taking ->lru_lock seems wrong but this
	 * is safe. Because if page_cgroup's USED bit is unset, the page
	 * will not be added to any memcg's LRU. If page_cgroup's USED bit is
	 * set, the commit after this will fail, anyway.
	 * This all charge/uncharge is done under some mutual execustion.
	 * So, we don't need to taking care of changes in USED bit.
	 */
	if (likely(!PageLRU(page)))
		return;

1074 1075 1076 1077 1078 1079 1080 1081
	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
1082 1083
}

1084
static void mem_cgroup_lru_add_after_commit(struct page *page)
1085 1086 1087 1088 1089
{
	unsigned long flags;
	struct zone *zone = page_zone(page);
	struct page_cgroup *pc = lookup_page_cgroup(page);

1090 1091 1092
	/* taking care of that the page is added to LRU while we commit it */
	if (likely(!PageLRU(page)))
		return;
1093 1094
	spin_lock_irqsave(&zone->lru_lock, flags);
	/* link when the page is linked to LRU but page_cgroup isn't */
1095
	if (PageLRU(page) && !PageCgroupAcctLRU(pc))
1096 1097 1098 1099 1100
		mem_cgroup_add_lru_list(page, page_lru(page));
	spin_unlock_irqrestore(&zone->lru_lock, flags);
}


K
KAMEZAWA Hiroyuki 已提交
1101 1102 1103
void mem_cgroup_move_lists(struct page *page,
			   enum lru_list from, enum lru_list to)
{
1104
	if (mem_cgroup_disabled())
K
KAMEZAWA Hiroyuki 已提交
1105 1106 1107
		return;
	mem_cgroup_del_lru_list(page, from);
	mem_cgroup_add_lru_list(page, to);
1108 1109
}

1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124
/*
 * Checks whether given mem is same or in the root_mem's
 * hierarchy subtree
 */
static bool mem_cgroup_same_or_subtree(const struct mem_cgroup *root_mem,
		struct mem_cgroup *mem)
{
	if (root_mem != mem) {
		return (root_mem->use_hierarchy &&
			css_is_ancestor(&mem->css, &root_mem->css));
	}

	return true;
}

1125 1126 1127
int task_in_mem_cgroup(struct task_struct *task, const struct mem_cgroup *mem)
{
	int ret;
1128
	struct mem_cgroup *curr = NULL;
1129
	struct task_struct *p;
1130

1131 1132 1133 1134 1135
	p = find_lock_task_mm(task);
	if (!p)
		return 0;
	curr = try_get_mem_cgroup_from_mm(p->mm);
	task_unlock(p);
1136 1137
	if (!curr)
		return 0;
1138 1139 1140 1141 1142 1143
	/*
	 * 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").
	 */
1144
	ret = mem_cgroup_same_or_subtree(mem, curr);
1145
	css_put(&curr->css);
1146 1147 1148
	return ret;
}

1149
static int calc_inactive_ratio(struct mem_cgroup *memcg, unsigned long *present_pages)
1150 1151 1152
{
	unsigned long active;
	unsigned long inactive;
1153 1154
	unsigned long gb;
	unsigned long inactive_ratio;
1155

1156 1157
	inactive = mem_cgroup_nr_lru_pages(memcg, BIT(LRU_INACTIVE_ANON));
	active = mem_cgroup_nr_lru_pages(memcg, BIT(LRU_ACTIVE_ANON));
1158

1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185
	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)
1186 1187 1188 1189 1190
		return 1;

	return 0;
}

1191 1192 1193 1194 1195
int mem_cgroup_inactive_file_is_low(struct mem_cgroup *memcg)
{
	unsigned long active;
	unsigned long inactive;

1196 1197
	inactive = mem_cgroup_nr_lru_pages(memcg, BIT(LRU_INACTIVE_FILE));
	active = mem_cgroup_nr_lru_pages(memcg, BIT(LRU_ACTIVE_FILE));
1198 1199 1200 1201

	return (active > inactive);
}

K
KOSAKI Motohiro 已提交
1202 1203 1204
struct zone_reclaim_stat *mem_cgroup_get_reclaim_stat(struct mem_cgroup *memcg,
						      struct zone *zone)
{
1205
	int nid = zone_to_nid(zone);
K
KOSAKI Motohiro 已提交
1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221
	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);
1222 1223
	if (!PageCgroupUsed(pc))
		return NULL;
1224 1225
	/* Ensure pc->mem_cgroup is visible after reading PCG_USED. */
	smp_rmb();
1226
	mz = page_cgroup_zoneinfo(pc->mem_cgroup, page);
K
KOSAKI Motohiro 已提交
1227 1228 1229
	return &mz->reclaim_stat;
}

1230 1231 1232 1233 1234
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,
1235
					int active, int file)
1236 1237 1238 1239 1240 1241
{
	unsigned long nr_taken = 0;
	struct page *page;
	unsigned long scan;
	LIST_HEAD(pc_list);
	struct list_head *src;
1242
	struct page_cgroup *pc, *tmp;
1243
	int nid = zone_to_nid(z);
1244 1245
	int zid = zone_idx(z);
	struct mem_cgroup_per_zone *mz;
1246
	int lru = LRU_FILE * file + active;
1247
	int ret;
1248

1249
	BUG_ON(!mem_cont);
1250
	mz = mem_cgroup_zoneinfo(mem_cont, nid, zid);
1251
	src = &mz->lists[lru];
1252

1253 1254
	scan = 0;
	list_for_each_entry_safe_reverse(pc, tmp, src, lru) {
H
Hugh Dickins 已提交
1255
		if (scan >= nr_to_scan)
1256
			break;
K
KAMEZAWA Hiroyuki 已提交
1257

1258 1259
		if (unlikely(!PageCgroupUsed(pc)))
			continue;
1260

1261
		page = lookup_cgroup_page(pc);
1262

H
Hugh Dickins 已提交
1263
		if (unlikely(!PageLRU(page)))
1264 1265
			continue;

H
Hugh Dickins 已提交
1266
		scan++;
1267 1268 1269
		ret = __isolate_lru_page(page, mode, file);
		switch (ret) {
		case 0:
1270
			list_move(&page->lru, dst);
1271
			mem_cgroup_del_lru(page);
1272
			nr_taken += hpage_nr_pages(page);
1273 1274 1275 1276 1277 1278 1279
			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;
1280 1281 1282 1283
		}
	}

	*scanned = scan;
1284 1285 1286 1287

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

1288 1289 1290
	return nr_taken;
}

1291 1292 1293
#define mem_cgroup_from_res_counter(counter, member)	\
	container_of(counter, struct mem_cgroup, member)

1294
/**
1295 1296
 * mem_cgroup_margin - calculate chargeable space of a memory cgroup
 * @mem: the memory cgroup
1297
 *
1298
 * Returns the maximum amount of memory @mem can be charged with, in
1299
 * pages.
1300
 */
1301
static unsigned long mem_cgroup_margin(struct mem_cgroup *mem)
1302
{
1303 1304 1305 1306 1307
	unsigned long long margin;

	margin = res_counter_margin(&mem->res);
	if (do_swap_account)
		margin = min(margin, res_counter_margin(&mem->memsw));
1308
	return margin >> PAGE_SHIFT;
1309 1310
}

1311
int mem_cgroup_swappiness(struct mem_cgroup *memcg)
K
KOSAKI Motohiro 已提交
1312 1313 1314 1315 1316 1317 1318
{
	struct cgroup *cgrp = memcg->css.cgroup;

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

1319
	return memcg->swappiness;
K
KOSAKI Motohiro 已提交
1320 1321
}

1322 1323 1324
static void mem_cgroup_start_move(struct mem_cgroup *mem)
{
	int cpu;
1325 1326 1327 1328

	get_online_cpus();
	spin_lock(&mem->pcp_counter_lock);
	for_each_online_cpu(cpu)
1329
		per_cpu(mem->stat->count[MEM_CGROUP_ON_MOVE], cpu) += 1;
1330 1331 1332
	mem->nocpu_base.count[MEM_CGROUP_ON_MOVE] += 1;
	spin_unlock(&mem->pcp_counter_lock);
	put_online_cpus();
1333 1334 1335 1336 1337 1338 1339 1340 1341 1342

	synchronize_rcu();
}

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

	if (!mem)
		return;
1343 1344 1345
	get_online_cpus();
	spin_lock(&mem->pcp_counter_lock);
	for_each_online_cpu(cpu)
1346
		per_cpu(mem->stat->count[MEM_CGROUP_ON_MOVE], cpu) -= 1;
1347 1348 1349
	mem->nocpu_base.count[MEM_CGROUP_ON_MOVE] -= 1;
	spin_unlock(&mem->pcp_counter_lock);
	put_online_cpus();
1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367
}
/*
 * 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;
}
1368 1369 1370

static bool mem_cgroup_under_move(struct mem_cgroup *mem)
{
1371 1372
	struct mem_cgroup *from;
	struct mem_cgroup *to;
1373
	bool ret = false;
1374 1375 1376 1377 1378 1379 1380 1381 1382
	/*
	 * 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;
1383 1384 1385

	ret = mem_cgroup_same_or_subtree(mem, from)
		|| mem_cgroup_same_or_subtree(mem, to);
1386 1387
unlock:
	spin_unlock(&mc.lock);
1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406
	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;
}

1407
/**
1408
 * mem_cgroup_print_oom_info: Called from OOM with tasklist_lock held in read mode.
1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426
 * @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;

1427
	if (!memcg || !p)
1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473
		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));
}

1474 1475 1476 1477 1478 1479 1480
/*
 * 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
1481 1482 1483 1484
	struct mem_cgroup *iter;

	for_each_mem_cgroup_tree(iter, mem)
		num++;
1485 1486 1487
	return num;
}

D
David Rientjes 已提交
1488 1489 1490 1491 1492 1493 1494 1495
/*
 * Return the memory (and swap, if configured) limit for a memcg.
 */
u64 mem_cgroup_get_limit(struct mem_cgroup *memcg)
{
	u64 limit;
	u64 memsw;

1496 1497 1498
	limit = res_counter_read_u64(&memcg->res, RES_LIMIT);
	limit += total_swap_pages << PAGE_SHIFT;

D
David Rientjes 已提交
1499 1500 1501 1502 1503 1504 1505 1506
	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);
}

1507
/*
K
KAMEZAWA Hiroyuki 已提交
1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543
 * 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;
}

1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556
/**
 * test_mem_cgroup_node_reclaimable
 * @mem: the target memcg
 * @nid: the node ID to be checked.
 * @noswap : specify true here if the user wants flle only information.
 *
 * This function returns whether the specified memcg contains any
 * reclaimable pages on a node. Returns true if there are any reclaimable
 * pages in the node.
 */
static bool test_mem_cgroup_node_reclaimable(struct mem_cgroup *mem,
		int nid, bool noswap)
{
1557
	if (mem_cgroup_node_nr_lru_pages(mem, nid, LRU_ALL_FILE))
1558 1559 1560
		return true;
	if (noswap || !total_swap_pages)
		return false;
1561
	if (mem_cgroup_node_nr_lru_pages(mem, nid, LRU_ALL_ANON))
1562 1563 1564 1565
		return true;
	return false;

}
1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576
#if MAX_NUMNODES > 1

/*
 * Always updating the nodemask is not very good - even if we have an empty
 * list or the wrong list here, we can start from some node and traverse all
 * nodes based on the zonelist. So update the list loosely once per 10 secs.
 *
 */
static void mem_cgroup_may_update_nodemask(struct mem_cgroup *mem)
{
	int nid;
1577 1578 1579 1580 1581 1582 1583
	/*
	 * numainfo_events > 0 means there was at least NUMAINFO_EVENTS_TARGET
	 * pagein/pageout changes since the last update.
	 */
	if (!atomic_read(&mem->numainfo_events))
		return;
	if (atomic_inc_return(&mem->numainfo_updating) > 1)
1584 1585 1586 1587 1588 1589 1590
		return;

	/* make a nodemask where this memcg uses memory from */
	mem->scan_nodes = node_states[N_HIGH_MEMORY];

	for_each_node_mask(nid, node_states[N_HIGH_MEMORY]) {

1591 1592
		if (!test_mem_cgroup_node_reclaimable(mem, nid, false))
			node_clear(nid, mem->scan_nodes);
1593
	}
1594 1595 1596

	atomic_set(&mem->numainfo_events, 0);
	atomic_set(&mem->numainfo_updating, 0);
1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633
}

/*
 * Selecting a node where we start reclaim from. Because what we need is just
 * reducing usage counter, start from anywhere is O,K. Considering
 * memory reclaim from current node, there are pros. and cons.
 *
 * Freeing memory from current node means freeing memory from a node which
 * we'll use or we've used. So, it may make LRU bad. And if several threads
 * hit limits, it will see a contention on a node. But freeing from remote
 * node means more costs for memory reclaim because of memory latency.
 *
 * Now, we use round-robin. Better algorithm is welcomed.
 */
int mem_cgroup_select_victim_node(struct mem_cgroup *mem)
{
	int node;

	mem_cgroup_may_update_nodemask(mem);
	node = mem->last_scanned_node;

	node = next_node(node, mem->scan_nodes);
	if (node == MAX_NUMNODES)
		node = first_node(mem->scan_nodes);
	/*
	 * We call this when we hit limit, not when pages are added to LRU.
	 * No LRU may hold pages because all pages are UNEVICTABLE or
	 * memcg is too small and all pages are not on LRU. In that case,
	 * we use curret node.
	 */
	if (unlikely(node == MAX_NUMNODES))
		node = numa_node_id();

	mem->last_scanned_node = node;
	return node;
}

1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668
/*
 * Check all nodes whether it contains reclaimable pages or not.
 * For quick scan, we make use of scan_nodes. This will allow us to skip
 * unused nodes. But scan_nodes is lazily updated and may not cotain
 * enough new information. We need to do double check.
 */
bool mem_cgroup_reclaimable(struct mem_cgroup *mem, bool noswap)
{
	int nid;

	/*
	 * quick check...making use of scan_node.
	 * We can skip unused nodes.
	 */
	if (!nodes_empty(mem->scan_nodes)) {
		for (nid = first_node(mem->scan_nodes);
		     nid < MAX_NUMNODES;
		     nid = next_node(nid, mem->scan_nodes)) {

			if (test_mem_cgroup_node_reclaimable(mem, nid, noswap))
				return true;
		}
	}
	/*
	 * Check rest of nodes.
	 */
	for_each_node_state(nid, N_HIGH_MEMORY) {
		if (node_isset(nid, mem->scan_nodes))
			continue;
		if (test_mem_cgroup_node_reclaimable(mem, nid, noswap))
			return true;
	}
	return false;
}

1669 1670 1671 1672 1673
#else
int mem_cgroup_select_victim_node(struct mem_cgroup *mem)
{
	return 0;
}
1674 1675 1676 1677 1678

bool mem_cgroup_reclaimable(struct mem_cgroup *mem, bool noswap)
{
	return test_mem_cgroup_node_reclaimable(mem, 0, noswap);
}
1679 1680
#endif

1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718
static void __mem_cgroup_record_scanstat(unsigned long *stats,
			   struct memcg_scanrecord *rec)
{

	stats[SCAN] += rec->nr_scanned[0] + rec->nr_scanned[1];
	stats[SCAN_ANON] += rec->nr_scanned[0];
	stats[SCAN_FILE] += rec->nr_scanned[1];

	stats[ROTATE] += rec->nr_rotated[0] + rec->nr_rotated[1];
	stats[ROTATE_ANON] += rec->nr_rotated[0];
	stats[ROTATE_FILE] += rec->nr_rotated[1];

	stats[FREED] += rec->nr_freed[0] + rec->nr_freed[1];
	stats[FREED_ANON] += rec->nr_freed[0];
	stats[FREED_FILE] += rec->nr_freed[1];

	stats[ELAPSED] += rec->elapsed;
}

static void mem_cgroup_record_scanstat(struct memcg_scanrecord *rec)
{
	struct mem_cgroup *mem;
	int context = rec->context;

	if (context >= NR_SCAN_CONTEXT)
		return;

	mem = rec->mem;
	spin_lock(&mem->scanstat.lock);
	__mem_cgroup_record_scanstat(mem->scanstat.stats[context], rec);
	spin_unlock(&mem->scanstat.lock);

	mem = rec->root;
	spin_lock(&mem->scanstat.lock);
	__mem_cgroup_record_scanstat(mem->scanstat.rootstats[context], rec);
	spin_unlock(&mem->scanstat.lock);
}

K
KAMEZAWA Hiroyuki 已提交
1719 1720 1721 1722
/*
 * 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.
1723 1724
 *
 * root_mem is the original ancestor that we've been reclaim from.
K
KAMEZAWA Hiroyuki 已提交
1725 1726 1727
 *
 * We give up and return to the caller when we visit root_mem twice.
 * (other groups can be removed while we're walking....)
1728 1729
 *
 * If shrink==true, for avoiding to free too much, this returns immedieately.
1730 1731
 */
static int mem_cgroup_hierarchical_reclaim(struct mem_cgroup *root_mem,
1732
						struct zone *zone,
1733
						gfp_t gfp_mask,
1734 1735
						unsigned long reclaim_options,
						unsigned long *total_scanned)
1736
{
K
KAMEZAWA Hiroyuki 已提交
1737 1738 1739
	struct mem_cgroup *victim;
	int ret, total = 0;
	int loop = 0;
1740 1741
	bool noswap = reclaim_options & MEM_CGROUP_RECLAIM_NOSWAP;
	bool shrink = reclaim_options & MEM_CGROUP_RECLAIM_SHRINK;
1742
	bool check_soft = reclaim_options & MEM_CGROUP_RECLAIM_SOFT;
1743
	struct memcg_scanrecord rec;
1744
	unsigned long excess;
1745
	unsigned long scanned;
1746 1747

	excess = res_counter_soft_limit_excess(&root_mem->res) >> PAGE_SHIFT;
K
KAMEZAWA Hiroyuki 已提交
1748

1749
	/* If memsw_is_minimum==1, swap-out is of-no-use. */
1750
	if (!check_soft && !shrink && root_mem->memsw_is_minimum)
1751 1752
		noswap = true;

1753 1754 1755 1756 1757 1758 1759 1760 1761
	if (shrink)
		rec.context = SCAN_BY_SHRINK;
	else if (check_soft)
		rec.context = SCAN_BY_SYSTEM;
	else
		rec.context = SCAN_BY_LIMIT;

	rec.root = root_mem;

1762
	while (1) {
K
KAMEZAWA Hiroyuki 已提交
1763
		victim = mem_cgroup_select_victim(root_mem);
1764
		if (victim == root_mem) {
K
KAMEZAWA Hiroyuki 已提交
1765
			loop++;
1766 1767 1768 1769 1770 1771 1772
			/*
			 * We are not draining per cpu cached charges during
			 * soft limit reclaim  because global reclaim doesn't
			 * care about charges. It tries to free some memory and
			 * charges will not give any.
			 */
			if (!check_soft && loop >= 1)
1773
				drain_all_stock_async(root_mem);
1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784
			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;
				}
				/*
L
Lucas De Marchi 已提交
1785
				 * We want to do more targeted reclaim.
1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796
				 * 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;
				}
			}
		}
1797
		if (!mem_cgroup_reclaimable(victim, noswap)) {
K
KAMEZAWA Hiroyuki 已提交
1798 1799
			/* this cgroup's local usage == 0 */
			css_put(&victim->css);
1800 1801
			continue;
		}
1802 1803 1804 1805 1806 1807 1808 1809
		rec.mem = victim;
		rec.nr_scanned[0] = 0;
		rec.nr_scanned[1] = 0;
		rec.nr_rotated[0] = 0;
		rec.nr_rotated[1] = 0;
		rec.nr_freed[0] = 0;
		rec.nr_freed[1] = 0;
		rec.elapsed = 0;
K
KAMEZAWA Hiroyuki 已提交
1810
		/* we use swappiness of local cgroup */
1811
		if (check_soft) {
1812
			ret = mem_cgroup_shrink_node_zone(victim, gfp_mask,
1813 1814
				noswap, zone, &rec, &scanned);
			*total_scanned += scanned;
1815
		} else
1816
			ret = try_to_free_mem_cgroup_pages(victim, gfp_mask,
1817 1818
						noswap, &rec);
		mem_cgroup_record_scanstat(&rec);
K
KAMEZAWA Hiroyuki 已提交
1819
		css_put(&victim->css);
1820 1821 1822 1823 1824 1825 1826
		/*
		 * 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
1827
		total += ret;
1828
		if (check_soft) {
1829
			if (!res_counter_soft_limit_excess(&root_mem->res))
1830
				return total;
1831
		} else if (mem_cgroup_margin(root_mem))
1832
			return total;
1833
	}
K
KAMEZAWA Hiroyuki 已提交
1834
	return total;
1835 1836
}

K
KAMEZAWA Hiroyuki 已提交
1837 1838 1839
/*
 * Check OOM-Killer is already running under our hierarchy.
 * If someone is running, return false.
1840
 * Has to be called with memcg_oom_lock
K
KAMEZAWA Hiroyuki 已提交
1841 1842 1843
 */
static bool mem_cgroup_oom_lock(struct mem_cgroup *mem)
{
1844 1845 1846
	int lock_count = -1;
	struct mem_cgroup *iter, *failed = NULL;
	bool cond = true;
1847

1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862
	for_each_mem_cgroup_tree_cond(iter, mem, cond) {
		bool locked = iter->oom_lock;

		iter->oom_lock = true;
		if (lock_count == -1)
			lock_count = iter->oom_lock;
		else if (lock_count != locked) {
			/*
			 * this subtree of our hierarchy is already locked
			 * so we cannot give a lock.
			 */
			lock_count = 0;
			failed = iter;
			cond = false;
		}
K
KAMEZAWA Hiroyuki 已提交
1863
	}
K
KAMEZAWA Hiroyuki 已提交
1864

1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881
	if (!failed)
		goto done;

	/*
	 * OK, we failed to lock the whole subtree so we have to clean up
	 * what we set up to the failing subtree
	 */
	cond = true;
	for_each_mem_cgroup_tree_cond(iter, mem, cond) {
		if (iter == failed) {
			cond = false;
			continue;
		}
		iter->oom_lock = false;
	}
done:
	return lock_count;
1882
}
1883

1884
/*
1885
 * Has to be called with memcg_oom_lock
1886
 */
K
KAMEZAWA Hiroyuki 已提交
1887
static int mem_cgroup_oom_unlock(struct mem_cgroup *mem)
1888
{
K
KAMEZAWA Hiroyuki 已提交
1889 1890
	struct mem_cgroup *iter;

1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907
	for_each_mem_cgroup_tree(iter, mem)
		iter->oom_lock = false;
	return 0;
}

static void mem_cgroup_mark_under_oom(struct mem_cgroup *mem)
{
	struct mem_cgroup *iter;

	for_each_mem_cgroup_tree(iter, mem)
		atomic_inc(&iter->under_oom);
}

static void mem_cgroup_unmark_under_oom(struct mem_cgroup *mem)
{
	struct mem_cgroup *iter;

K
KAMEZAWA Hiroyuki 已提交
1908 1909 1910 1911 1912
	/*
	 * 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
1913
	for_each_mem_cgroup_tree(iter, mem)
1914
		atomic_add_unless(&iter->under_oom, -1, 0);
1915 1916
}

1917
static DEFINE_SPINLOCK(memcg_oom_lock);
K
KAMEZAWA Hiroyuki 已提交
1918 1919
static DECLARE_WAIT_QUEUE_HEAD(memcg_oom_waitq);

K
KAMEZAWA Hiroyuki 已提交
1920 1921 1922 1923 1924 1925 1926 1927
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)
{
1928 1929
	struct mem_cgroup *wake_mem = (struct mem_cgroup *)arg,
			  *oom_wait_mem;
K
KAMEZAWA Hiroyuki 已提交
1930 1931 1932
	struct oom_wait_info *oom_wait_info;

	oom_wait_info = container_of(wait, struct oom_wait_info, wait);
1933
	oom_wait_mem = oom_wait_info->mem;
K
KAMEZAWA Hiroyuki 已提交
1934 1935 1936 1937 1938

	/*
	 * 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.
	 */
1939 1940
	if (!mem_cgroup_same_or_subtree(oom_wait_mem, wake_mem)
			&& !mem_cgroup_same_or_subtree(wake_mem, oom_wait_mem))
K
KAMEZAWA Hiroyuki 已提交
1941 1942 1943 1944 1945 1946 1947 1948 1949 1950
		return 0;
	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);
}

1951 1952
static void memcg_oom_recover(struct mem_cgroup *mem)
{
1953
	if (mem && atomic_read(&mem->under_oom))
1954 1955 1956
		memcg_wakeup_oom(mem);
}

K
KAMEZAWA Hiroyuki 已提交
1957 1958 1959 1960
/*
 * 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)
1961
{
K
KAMEZAWA Hiroyuki 已提交
1962
	struct oom_wait_info owait;
1963
	bool locked, need_to_kill;
K
KAMEZAWA Hiroyuki 已提交
1964

K
KAMEZAWA Hiroyuki 已提交
1965 1966 1967 1968 1969
	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);
1970
	need_to_kill = true;
1971 1972
	mem_cgroup_mark_under_oom(mem);

K
KAMEZAWA Hiroyuki 已提交
1973
	/* At first, try to OOM lock hierarchy under mem.*/
1974
	spin_lock(&memcg_oom_lock);
K
KAMEZAWA Hiroyuki 已提交
1975 1976 1977 1978 1979 1980
	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.
	 */
1981 1982 1983 1984
	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
1985
		mem_cgroup_oom_notify(mem);
1986
	spin_unlock(&memcg_oom_lock);
K
KAMEZAWA Hiroyuki 已提交
1987

1988 1989
	if (need_to_kill) {
		finish_wait(&memcg_oom_waitq, &owait.wait);
K
KAMEZAWA Hiroyuki 已提交
1990
		mem_cgroup_out_of_memory(mem, mask);
1991
	} else {
K
KAMEZAWA Hiroyuki 已提交
1992
		schedule();
K
KAMEZAWA Hiroyuki 已提交
1993
		finish_wait(&memcg_oom_waitq, &owait.wait);
K
KAMEZAWA Hiroyuki 已提交
1994
	}
1995
	spin_lock(&memcg_oom_lock);
1996 1997
	if (locked)
		mem_cgroup_oom_unlock(mem);
K
KAMEZAWA Hiroyuki 已提交
1998
	memcg_wakeup_oom(mem);
1999
	spin_unlock(&memcg_oom_lock);
K
KAMEZAWA Hiroyuki 已提交
2000

2001 2002
	mem_cgroup_unmark_under_oom(mem);

K
KAMEZAWA Hiroyuki 已提交
2003 2004 2005 2006 2007
	if (test_thread_flag(TIF_MEMDIE) || fatal_signal_pending(current))
		return false;
	/* Give chance to dying process */
	schedule_timeout(1);
	return true;
2008 2009
}

2010 2011 2012
/*
 * Currently used to update mapped file statistics, but the routine can be
 * generalized to update other statistics as well.
2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031
 *
 * 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.
2032
 */
2033

2034 2035
void mem_cgroup_update_page_stat(struct page *page,
				 enum mem_cgroup_page_stat_item idx, int val)
2036 2037
{
	struct mem_cgroup *mem;
2038 2039
	struct page_cgroup *pc = lookup_page_cgroup(page);
	bool need_unlock = false;
2040
	unsigned long uninitialized_var(flags);
2041 2042 2043 2044

	if (unlikely(!pc))
		return;

2045
	rcu_read_lock();
2046
	mem = pc->mem_cgroup;
2047 2048 2049
	if (unlikely(!mem || !PageCgroupUsed(pc)))
		goto out;
	/* pc->mem_cgroup is unstable ? */
2050
	if (unlikely(mem_cgroup_stealed(mem)) || PageTransHuge(page)) {
2051
		/* take a lock against to access pc->mem_cgroup */
2052
		move_lock_page_cgroup(pc, &flags);
2053 2054 2055 2056 2057
		need_unlock = true;
		mem = pc->mem_cgroup;
		if (!mem || !PageCgroupUsed(pc))
			goto out;
	}
2058 2059

	switch (idx) {
2060
	case MEMCG_NR_FILE_MAPPED:
2061 2062 2063
		if (val > 0)
			SetPageCgroupFileMapped(pc);
		else if (!page_mapped(page))
2064
			ClearPageCgroupFileMapped(pc);
2065
		idx = MEM_CGROUP_STAT_FILE_MAPPED;
2066 2067 2068
		break;
	default:
		BUG();
2069
	}
2070

2071 2072
	this_cpu_add(mem->stat->count[idx], val);

2073 2074
out:
	if (unlikely(need_unlock))
2075
		move_unlock_page_cgroup(pc, &flags);
2076 2077
	rcu_read_unlock();
	return;
2078
}
2079
EXPORT_SYMBOL(mem_cgroup_update_page_stat);
2080

2081 2082 2083 2084
/*
 * size of first charge trial. "32" comes from vmscan.c's magic value.
 * TODO: maybe necessary to use big numbers in big irons.
 */
2085
#define CHARGE_BATCH	32U
2086 2087
struct memcg_stock_pcp {
	struct mem_cgroup *cached; /* this never be root cgroup */
2088
	unsigned int nr_pages;
2089
	struct work_struct work;
2090 2091
	unsigned long flags;
#define FLUSHING_CACHED_CHARGE	(0)
2092 2093
};
static DEFINE_PER_CPU(struct memcg_stock_pcp, memcg_stock);
2094
static DEFINE_MUTEX(percpu_charge_mutex);
2095 2096

/*
2097
 * Try to consume stocked charge on this cpu. If success, one page is consumed
2098 2099 2100 2101 2102 2103 2104 2105 2106 2107
 * 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);
2108 2109
	if (mem == stock->cached && stock->nr_pages)
		stock->nr_pages--;
2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122
	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;

2123 2124 2125 2126
	if (stock->nr_pages) {
		unsigned long bytes = stock->nr_pages * PAGE_SIZE;

		res_counter_uncharge(&old->res, bytes);
2127
		if (do_swap_account)
2128 2129
			res_counter_uncharge(&old->memsw, bytes);
		stock->nr_pages = 0;
2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141
	}
	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);
2142
	clear_bit(FLUSHING_CACHED_CHARGE, &stock->flags);
2143 2144 2145 2146
}

/*
 * Cache charges(val) which is from res_counter, to local per_cpu area.
2147
 * This will be consumed by consume_stock() function, later.
2148
 */
2149
static void refill_stock(struct mem_cgroup *mem, unsigned int nr_pages)
2150 2151 2152 2153 2154 2155 2156
{
	struct memcg_stock_pcp *stock = &get_cpu_var(memcg_stock);

	if (stock->cached != mem) { /* reset if necessary */
		drain_stock(stock);
		stock->cached = mem;
	}
2157
	stock->nr_pages += nr_pages;
2158 2159 2160 2161
	put_cpu_var(memcg_stock);
}

/*
2162 2163 2164
 * Drains all per-CPU charge caches for given root_mem resp. subtree
 * of the hierarchy under it. sync flag says whether we should block
 * until the work is done.
2165
 */
2166
static void drain_all_stock(struct mem_cgroup *root_mem, bool sync)
2167
{
2168
	int cpu, curcpu;
2169

2170 2171
	/* Notify other cpus that system-wide "drain" is running */
	get_online_cpus();
2172 2173 2174 2175 2176 2177 2178
	/*
	 * Get a hint for avoiding draining charges on the current cpu,
	 * which must be exhausted by our charging.  It is not required that
	 * this be a precise check, so we use raw_smp_processor_id() instead of
	 * getcpu()/putcpu().
	 */
	curcpu = raw_smp_processor_id();
2179 2180
	for_each_online_cpu(cpu) {
		struct memcg_stock_pcp *stock = &per_cpu(memcg_stock, cpu);
2181 2182 2183
		struct mem_cgroup *mem;

		mem = stock->cached;
2184
		if (!mem || !stock->nr_pages)
2185
			continue;
2186 2187
		if (!mem_cgroup_same_or_subtree(root_mem, mem))
			continue;
2188 2189 2190 2191 2192 2193
		if (!test_and_set_bit(FLUSHING_CACHED_CHARGE, &stock->flags)) {
			if (cpu == curcpu)
				drain_local_stock(&stock->work);
			else
				schedule_work_on(cpu, &stock->work);
		}
2194
	}
2195 2196 2197 2198 2199 2200

	if (!sync)
		goto out;

	for_each_online_cpu(cpu) {
		struct memcg_stock_pcp *stock = &per_cpu(memcg_stock, cpu);
2201
		if (test_bit(FLUSHING_CACHED_CHARGE, &stock->flags))
2202 2203 2204
			flush_work(&stock->work);
	}
out:
2205
 	put_online_cpus();
2206 2207 2208 2209 2210 2211 2212 2213 2214 2215
}

/*
 * 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(struct mem_cgroup *root_mem)
{
2216 2217 2218 2219 2220
	/*
	 * If someone calls draining, avoid adding more kworker runs.
	 */
	if (!mutex_trylock(&percpu_charge_mutex))
		return;
2221
	drain_all_stock(root_mem, false);
2222
	mutex_unlock(&percpu_charge_mutex);
2223 2224 2225
}

/* This is a synchronous drain interface. */
2226
static void drain_all_stock_sync(struct mem_cgroup *root_mem)
2227 2228
{
	/* called when force_empty is called */
2229
	mutex_lock(&percpu_charge_mutex);
2230
	drain_all_stock(root_mem, true);
2231
	mutex_unlock(&percpu_charge_mutex);
2232 2233
}

2234 2235 2236 2237 2238 2239 2240 2241 2242 2243
/*
 * 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++) {
2244
		long x = per_cpu(mem->stat->count[i], cpu);
2245 2246 2247 2248

		per_cpu(mem->stat->count[i], cpu) = 0;
		mem->nocpu_base.count[i] += x;
	}
2249 2250 2251 2252 2253 2254
	for (i = 0; i < MEM_CGROUP_EVENTS_NSTATS; i++) {
		unsigned long x = per_cpu(mem->stat->events[i], cpu);

		per_cpu(mem->stat->events[i], cpu) = 0;
		mem->nocpu_base.events[i] += x;
	}
2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265
	/* 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];
2266 2267 2268 2269
	spin_unlock(&mem->pcp_counter_lock);
}

static int __cpuinit memcg_cpu_hotplug_callback(struct notifier_block *nb,
2270 2271 2272 2273 2274
					unsigned long action,
					void *hcpu)
{
	int cpu = (unsigned long)hcpu;
	struct memcg_stock_pcp *stock;
2275
	struct mem_cgroup *iter;
2276

2277 2278 2279 2280 2281 2282
	if ((action == CPU_ONLINE)) {
		for_each_mem_cgroup_all(iter)
			synchronize_mem_cgroup_on_move(iter, cpu);
		return NOTIFY_OK;
	}

2283
	if ((action != CPU_DEAD) || action != CPU_DEAD_FROZEN)
2284
		return NOTIFY_OK;
2285 2286 2287 2288

	for_each_mem_cgroup_all(iter)
		mem_cgroup_drain_pcp_counter(iter, cpu);

2289 2290 2291 2292 2293
	stock = &per_cpu(memcg_stock, cpu);
	drain_stock(stock);
	return NOTIFY_OK;
}

2294 2295 2296 2297 2298 2299 2300 2301 2302 2303

/* 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 */
};

2304 2305
static int mem_cgroup_do_charge(struct mem_cgroup *mem, gfp_t gfp_mask,
				unsigned int nr_pages, bool oom_check)
2306
{
2307
	unsigned long csize = nr_pages * PAGE_SIZE;
2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321
	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;

2322
		res_counter_uncharge(&mem->res, csize);
2323 2324 2325 2326
		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);
2327
	/*
2328 2329
	 * nr_pages can be either a huge page (HPAGE_PMD_NR), a batch
	 * of regular pages (CHARGE_BATCH), or a single regular page (1).
2330 2331 2332 2333
	 *
	 * Never reclaim on behalf of optional batching, retry with a
	 * single page instead.
	 */
2334
	if (nr_pages == CHARGE_BATCH)
2335 2336 2337 2338 2339 2340
		return CHARGE_RETRY;

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

	ret = mem_cgroup_hierarchical_reclaim(mem_over_limit, NULL,
2341
					      gfp_mask, flags, NULL);
2342
	if (mem_cgroup_margin(mem_over_limit) >= nr_pages)
2343
		return CHARGE_RETRY;
2344
	/*
2345 2346 2347 2348 2349 2350 2351
	 * 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.
2352
	 */
2353
	if (nr_pages == 1 && ret)
2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372
		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;
}

2373 2374 2375
/*
 * Unlike exported interface, "oom" parameter is added. if oom==true,
 * oom-killer can be invoked.
2376
 */
2377
static int __mem_cgroup_try_charge(struct mm_struct *mm,
A
Andrea Arcangeli 已提交
2378
				   gfp_t gfp_mask,
2379 2380 2381
				   unsigned int nr_pages,
				   struct mem_cgroup **memcg,
				   bool oom)
2382
{
2383
	unsigned int batch = max(CHARGE_BATCH, nr_pages);
2384 2385 2386
	int nr_oom_retries = MEM_CGROUP_RECLAIM_RETRIES;
	struct mem_cgroup *mem = NULL;
	int ret;
2387

K
KAMEZAWA Hiroyuki 已提交
2388 2389 2390 2391 2392 2393 2394 2395
	/*
	 * 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;
2396

2397
	/*
2398 2399
	 * We always charge the cgroup the mm_struct belongs to.
	 * The mm_struct's mem_cgroup changes on task migration if the
2400 2401 2402
	 * 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
2403 2404 2405 2406
	if (!*memcg && !mm)
		goto bypass;
again:
	if (*memcg) { /* css should be a valid one */
2407
		mem = *memcg;
K
KAMEZAWA Hiroyuki 已提交
2408 2409 2410
		VM_BUG_ON(css_is_removed(&mem->css));
		if (mem_cgroup_is_root(mem))
			goto done;
2411
		if (nr_pages == 1 && consume_stock(mem))
K
KAMEZAWA Hiroyuki 已提交
2412
			goto done;
2413 2414
		css_get(&mem->css);
	} else {
K
KAMEZAWA Hiroyuki 已提交
2415
		struct task_struct *p;
2416

K
KAMEZAWA Hiroyuki 已提交
2417 2418 2419
		rcu_read_lock();
		p = rcu_dereference(mm->owner);
		/*
2420 2421 2422 2423 2424 2425 2426 2427
		 * 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
2428 2429
		 */
		mem = mem_cgroup_from_task(p);
2430
		if (!mem || mem_cgroup_is_root(mem)) {
K
KAMEZAWA Hiroyuki 已提交
2431 2432 2433
			rcu_read_unlock();
			goto done;
		}
2434
		if (nr_pages == 1 && consume_stock(mem)) {
K
KAMEZAWA Hiroyuki 已提交
2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452
			/*
			 * 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();
	}
2453

2454 2455
	do {
		bool oom_check;
2456

2457
		/* If killed, bypass charge */
K
KAMEZAWA Hiroyuki 已提交
2458 2459
		if (fatal_signal_pending(current)) {
			css_put(&mem->css);
2460
			goto bypass;
K
KAMEZAWA Hiroyuki 已提交
2461
		}
2462

2463 2464 2465 2466
		oom_check = false;
		if (oom && !nr_oom_retries) {
			oom_check = true;
			nr_oom_retries = MEM_CGROUP_RECLAIM_RETRIES;
2467
		}
2468

2469
		ret = mem_cgroup_do_charge(mem, gfp_mask, batch, oom_check);
2470 2471 2472 2473
		switch (ret) {
		case CHARGE_OK:
			break;
		case CHARGE_RETRY: /* not in OOM situation but retry */
2474
			batch = nr_pages;
K
KAMEZAWA Hiroyuki 已提交
2475 2476 2477
			css_put(&mem->css);
			mem = NULL;
			goto again;
2478
		case CHARGE_WOULDBLOCK: /* !__GFP_WAIT */
K
KAMEZAWA Hiroyuki 已提交
2479
			css_put(&mem->css);
2480 2481
			goto nomem;
		case CHARGE_NOMEM: /* OOM routine works */
K
KAMEZAWA Hiroyuki 已提交
2482 2483
			if (!oom) {
				css_put(&mem->css);
K
KAMEZAWA Hiroyuki 已提交
2484
				goto nomem;
K
KAMEZAWA Hiroyuki 已提交
2485
			}
2486 2487 2488 2489
			/* If oom, we never return -ENOMEM */
			nr_oom_retries--;
			break;
		case CHARGE_OOM_DIE: /* Killed by OOM Killer */
K
KAMEZAWA Hiroyuki 已提交
2490
			css_put(&mem->css);
K
KAMEZAWA Hiroyuki 已提交
2491
			goto bypass;
2492
		}
2493 2494
	} while (ret != CHARGE_OK);

2495 2496
	if (batch > nr_pages)
		refill_stock(mem, batch - nr_pages);
K
KAMEZAWA Hiroyuki 已提交
2497
	css_put(&mem->css);
2498
done:
K
KAMEZAWA Hiroyuki 已提交
2499
	*memcg = mem;
2500 2501
	return 0;
nomem:
K
KAMEZAWA Hiroyuki 已提交
2502
	*memcg = NULL;
2503
	return -ENOMEM;
K
KAMEZAWA Hiroyuki 已提交
2504 2505 2506
bypass:
	*memcg = NULL;
	return 0;
2507
}
2508

2509 2510 2511 2512 2513
/*
 * 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().
 */
2514
static void __mem_cgroup_cancel_charge(struct mem_cgroup *mem,
2515
				       unsigned int nr_pages)
2516 2517
{
	if (!mem_cgroup_is_root(mem)) {
2518 2519 2520
		unsigned long bytes = nr_pages * PAGE_SIZE;

		res_counter_uncharge(&mem->res, bytes);
2521
		if (do_swap_account)
2522
			res_counter_uncharge(&mem->memsw, bytes);
2523
	}
2524 2525
}

2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544
/*
 * 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);
}

2545
struct mem_cgroup *try_get_mem_cgroup_from_page(struct page *page)
2546
{
2547
	struct mem_cgroup *mem = NULL;
2548
	struct page_cgroup *pc;
2549
	unsigned short id;
2550 2551
	swp_entry_t ent;

2552 2553 2554
	VM_BUG_ON(!PageLocked(page));

	pc = lookup_page_cgroup(page);
2555
	lock_page_cgroup(pc);
2556
	if (PageCgroupUsed(pc)) {
2557
		mem = pc->mem_cgroup;
2558 2559
		if (mem && !css_tryget(&mem->css))
			mem = NULL;
2560
	} else if (PageSwapCache(page)) {
2561
		ent.val = page_private(page);
2562 2563 2564 2565 2566 2567
		id = lookup_swap_cgroup(ent);
		rcu_read_lock();
		mem = mem_cgroup_lookup(id);
		if (mem && !css_tryget(&mem->css))
			mem = NULL;
		rcu_read_unlock();
2568
	}
2569
	unlock_page_cgroup(pc);
2570 2571 2572
	return mem;
}

2573
static void __mem_cgroup_commit_charge(struct mem_cgroup *mem,
2574
				       struct page *page,
2575
				       unsigned int nr_pages,
2576
				       struct page_cgroup *pc,
2577
				       enum charge_type ctype)
2578
{
2579 2580 2581
	lock_page_cgroup(pc);
	if (unlikely(PageCgroupUsed(pc))) {
		unlock_page_cgroup(pc);
2582
		__mem_cgroup_cancel_charge(mem, nr_pages);
2583 2584 2585 2586 2587 2588
		return;
	}
	/*
	 * we don't need page_cgroup_lock about tail pages, becase they are not
	 * accessed by any other context at this point.
	 */
2589
	pc->mem_cgroup = mem;
2590 2591 2592 2593 2594 2595 2596
	/*
	 * 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
2597
	smp_wmb();
2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610
	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;
	}
2611

2612
	mem_cgroup_charge_statistics(mem, PageCgroupCache(pc), nr_pages);
2613
	unlock_page_cgroup(pc);
2614 2615 2616 2617 2618
	/*
	 * "charge_statistics" updated event counter. Then, check it.
	 * Insert ancestor (and ancestor's ancestors), to softlimit RB-tree.
	 * if they exceeds softlimit.
	 */
2619
	memcg_check_events(mem, page);
2620
}
2621

2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635
#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;

2636 2637
	if (mem_cgroup_disabled())
		return;
2638
	/*
2639
	 * We have no races with charge/uncharge but will have races with
2640 2641 2642 2643 2644 2645
	 * page state accounting.
	 */
	move_lock_page_cgroup(head_pc, &flags);

	tail_pc->mem_cgroup = head_pc->mem_cgroup;
	smp_wmb(); /* see __commit_charge() */
2646 2647 2648 2649 2650 2651 2652 2653 2654 2655
	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);
2656
		mz = page_cgroup_zoneinfo(head_pc->mem_cgroup, head);
2657 2658
		MEM_CGROUP_ZSTAT(mz, lru) -= 1;
	}
2659 2660 2661 2662 2663
	tail_pc->flags = head_pc->flags & ~PCGF_NOCOPY_AT_SPLIT;
	move_unlock_page_cgroup(head_pc, &flags);
}
#endif

2664
/**
2665
 * mem_cgroup_move_account - move account of the page
2666
 * @page: the page
2667
 * @nr_pages: number of regular pages (>1 for huge pages)
2668 2669 2670
 * @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.
2671
 * @uncharge: whether we should call uncharge and css_put against @from.
2672 2673
 *
 * The caller must confirm following.
K
KAMEZAWA Hiroyuki 已提交
2674
 * - page is not on LRU (isolate_page() is useful.)
2675
 * - compound_lock is held when nr_pages > 1
2676
 *
2677
 * This function doesn't do "charge" nor css_get to new cgroup. It should be
L
Lucas De Marchi 已提交
2678
 * done by a caller(__mem_cgroup_try_charge would be useful). If @uncharge is
2679 2680
 * true, this function does "uncharge" from old cgroup, but it doesn't if
 * @uncharge is false, so a caller should do "uncharge".
2681
 */
2682 2683 2684 2685 2686 2687
static int mem_cgroup_move_account(struct page *page,
				   unsigned int nr_pages,
				   struct page_cgroup *pc,
				   struct mem_cgroup *from,
				   struct mem_cgroup *to,
				   bool uncharge)
2688
{
2689 2690
	unsigned long flags;
	int ret;
2691

2692
	VM_BUG_ON(from == to);
2693
	VM_BUG_ON(PageLRU(page));
2694 2695 2696 2697 2698 2699 2700
	/*
	 * 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;
2701
	if (nr_pages > 1 && !PageTransHuge(page))
2702 2703 2704 2705 2706 2707 2708 2709 2710
		goto out;

	lock_page_cgroup(pc);

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

	move_lock_page_cgroup(pc, &flags);
2711

2712
	if (PageCgroupFileMapped(pc)) {
2713 2714 2715 2716 2717
		/* 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();
2718
	}
2719
	mem_cgroup_charge_statistics(from, PageCgroupCache(pc), -nr_pages);
2720 2721
	if (uncharge)
		/* This is not "cancel", but cancel_charge does all we need. */
2722
		__mem_cgroup_cancel_charge(from, nr_pages);
2723

2724
	/* caller should have done css_get */
K
KAMEZAWA Hiroyuki 已提交
2725
	pc->mem_cgroup = to;
2726
	mem_cgroup_charge_statistics(to, PageCgroupCache(pc), nr_pages);
2727 2728 2729
	/*
	 * We charges against "to" which may not have any tasks. Then, "to"
	 * can be under rmdir(). But in current implementation, caller of
2730
	 * this function is just force_empty() and move charge, so it's
L
Lucas De Marchi 已提交
2731
	 * guaranteed that "to" is never removed. So, we don't check rmdir
2732
	 * status here.
2733
	 */
2734 2735 2736
	move_unlock_page_cgroup(pc, &flags);
	ret = 0;
unlock:
2737
	unlock_page_cgroup(pc);
2738 2739 2740
	/*
	 * check events
	 */
2741 2742
	memcg_check_events(to, page);
	memcg_check_events(from, page);
2743
out:
2744 2745 2746 2747 2748 2749 2750
	return ret;
}

/*
 * move charges to its parent.
 */

2751 2752
static int mem_cgroup_move_parent(struct page *page,
				  struct page_cgroup *pc,
2753 2754 2755 2756 2757 2758
				  struct mem_cgroup *child,
				  gfp_t gfp_mask)
{
	struct cgroup *cg = child->css.cgroup;
	struct cgroup *pcg = cg->parent;
	struct mem_cgroup *parent;
2759
	unsigned int nr_pages;
2760
	unsigned long uninitialized_var(flags);
2761 2762 2763 2764 2765 2766
	int ret;

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

2767 2768 2769 2770 2771
	ret = -EBUSY;
	if (!get_page_unless_zero(page))
		goto out;
	if (isolate_lru_page(page))
		goto put;
2772

2773
	nr_pages = hpage_nr_pages(page);
K
KAMEZAWA Hiroyuki 已提交
2774

2775
	parent = mem_cgroup_from_cont(pcg);
2776
	ret = __mem_cgroup_try_charge(NULL, gfp_mask, nr_pages, &parent, false);
2777
	if (ret || !parent)
2778
		goto put_back;
2779

2780
	if (nr_pages > 1)
2781 2782
		flags = compound_lock_irqsave(page);

2783
	ret = mem_cgroup_move_account(page, nr_pages, pc, child, parent, true);
2784
	if (ret)
2785
		__mem_cgroup_cancel_charge(parent, nr_pages);
2786

2787
	if (nr_pages > 1)
2788
		compound_unlock_irqrestore(page, flags);
2789
put_back:
K
KAMEZAWA Hiroyuki 已提交
2790
	putback_lru_page(page);
2791
put:
2792
	put_page(page);
2793
out:
2794 2795 2796
	return ret;
}

2797 2798 2799 2800 2801 2802 2803
/*
 * 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,
2804
				gfp_t gfp_mask, enum charge_type ctype)
2805
{
2806
	struct mem_cgroup *mem = NULL;
2807
	unsigned int nr_pages = 1;
2808
	struct page_cgroup *pc;
2809
	bool oom = true;
2810
	int ret;
A
Andrea Arcangeli 已提交
2811

A
Andrea Arcangeli 已提交
2812
	if (PageTransHuge(page)) {
2813
		nr_pages <<= compound_order(page);
A
Andrea Arcangeli 已提交
2814
		VM_BUG_ON(!PageTransHuge(page));
2815 2816 2817 2818 2819
		/*
		 * 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
2820
	}
2821 2822

	pc = lookup_page_cgroup(page);
2823
	BUG_ON(!pc); /* XXX: remove this and move pc lookup into commit */
2824

2825
	ret = __mem_cgroup_try_charge(mm, gfp_mask, nr_pages, &mem, oom);
2826
	if (ret || !mem)
2827 2828
		return ret;

2829
	__mem_cgroup_commit_charge(mem, page, nr_pages, pc, ctype);
2830 2831 2832
	return 0;
}

2833 2834
int mem_cgroup_newpage_charge(struct page *page,
			      struct mm_struct *mm, gfp_t gfp_mask)
2835
{
2836
	if (mem_cgroup_disabled())
2837
		return 0;
2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848
	/*
	 * 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;
2849
	return mem_cgroup_charge_common(page, mm, gfp_mask,
2850
				MEM_CGROUP_CHARGE_TYPE_MAPPED);
2851 2852
}

D
Daisuke Nishimura 已提交
2853 2854 2855 2856
static void
__mem_cgroup_commit_charge_swapin(struct page *page, struct mem_cgroup *ptr,
					enum charge_type ctype);

2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872
static void
__mem_cgroup_commit_charge_lrucare(struct page *page, struct mem_cgroup *mem,
					enum charge_type ctype)
{
	struct page_cgroup *pc = lookup_page_cgroup(page);
	/*
	 * In some case, SwapCache, FUSE(splice_buf->radixtree), the page
	 * is already on LRU. It means the page may on some other page_cgroup's
	 * LRU. Take care of it.
	 */
	mem_cgroup_lru_del_before_commit(page);
	__mem_cgroup_commit_charge(mem, page, 1, pc, ctype);
	mem_cgroup_lru_add_after_commit(page);
	return;
}

2873 2874
int mem_cgroup_cache_charge(struct page *page, struct mm_struct *mm,
				gfp_t gfp_mask)
2875
{
2876
	struct mem_cgroup *mem = NULL;
2877 2878
	int ret;

2879
	if (mem_cgroup_disabled())
2880
		return 0;
2881 2882
	if (PageCompound(page))
		return 0;
2883

2884
	if (unlikely(!mm))
2885
		mm = &init_mm;
2886

2887 2888 2889 2890
	if (page_is_file_cache(page)) {
		ret = __mem_cgroup_try_charge(mm, gfp_mask, 1, &mem, true);
		if (ret || !mem)
			return ret;
2891

2892 2893 2894 2895 2896 2897 2898 2899 2900
		/*
		 * FUSE reuses pages without going through the final
		 * put that would remove them from the LRU list, make
		 * sure that they get relinked properly.
		 */
		__mem_cgroup_commit_charge_lrucare(page, mem,
					MEM_CGROUP_CHARGE_TYPE_CACHE);
		return ret;
	}
D
Daisuke Nishimura 已提交
2901 2902 2903 2904 2905 2906 2907 2908
	/* shmem */
	if (PageSwapCache(page)) {
		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,
2909
					MEM_CGROUP_CHARGE_TYPE_SHMEM);
2910 2911

	return ret;
2912 2913
}

2914 2915 2916
/*
 * While swap-in, try_charge -> commit or cancel, the page is locked.
 * And when try_charge() successfully returns, one refcnt to memcg without
2917
 * struct page_cgroup is acquired. This refcnt will be consumed by
2918 2919
 * "commit()" or removed by "cancel()"
 */
2920 2921 2922 2923 2924
int mem_cgroup_try_charge_swapin(struct mm_struct *mm,
				 struct page *page,
				 gfp_t mask, struct mem_cgroup **ptr)
{
	struct mem_cgroup *mem;
2925
	int ret;
2926

2927 2928
	*ptr = NULL;

2929
	if (mem_cgroup_disabled())
2930 2931 2932 2933 2934 2935
		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
2936 2937 2938
	 * 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.
2939 2940
	 */
	if (!PageSwapCache(page))
H
Hugh Dickins 已提交
2941
		goto charge_cur_mm;
2942
	mem = try_get_mem_cgroup_from_page(page);
2943 2944
	if (!mem)
		goto charge_cur_mm;
2945
	*ptr = mem;
2946
	ret = __mem_cgroup_try_charge(NULL, mask, 1, ptr, true);
2947 2948
	css_put(&mem->css);
	return ret;
2949 2950 2951
charge_cur_mm:
	if (unlikely(!mm))
		mm = &init_mm;
2952
	return __mem_cgroup_try_charge(mm, mask, 1, ptr, true);
2953 2954
}

D
Daisuke Nishimura 已提交
2955 2956 2957
static void
__mem_cgroup_commit_charge_swapin(struct page *page, struct mem_cgroup *ptr,
					enum charge_type ctype)
2958
{
2959
	if (mem_cgroup_disabled())
2960 2961 2962
		return;
	if (!ptr)
		return;
2963
	cgroup_exclude_rmdir(&ptr->css);
2964 2965

	__mem_cgroup_commit_charge_lrucare(page, ptr, ctype);
2966 2967 2968
	/*
	 * Now swap is on-memory. This means this page may be
	 * counted both as mem and swap....double count.
2969 2970 2971
	 * 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.
2972
	 */
2973
	if (do_swap_account && PageSwapCache(page)) {
2974
		swp_entry_t ent = {.val = page_private(page)};
2975
		unsigned short id;
2976
		struct mem_cgroup *memcg;
2977 2978 2979 2980

		id = swap_cgroup_record(ent, 0);
		rcu_read_lock();
		memcg = mem_cgroup_lookup(id);
2981
		if (memcg) {
2982 2983 2984 2985
			/*
			 * This recorded memcg can be obsolete one. So, avoid
			 * calling css_tryget
			 */
2986
			if (!mem_cgroup_is_root(memcg))
2987
				res_counter_uncharge(&memcg->memsw, PAGE_SIZE);
2988
			mem_cgroup_swap_statistics(memcg, false);
2989 2990
			mem_cgroup_put(memcg);
		}
2991
		rcu_read_unlock();
2992
	}
2993 2994 2995 2996 2997 2998
	/*
	 * 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);
2999 3000
}

D
Daisuke Nishimura 已提交
3001 3002 3003 3004 3005 3006
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);
}

3007 3008
void mem_cgroup_cancel_charge_swapin(struct mem_cgroup *mem)
{
3009
	if (mem_cgroup_disabled())
3010 3011 3012
		return;
	if (!mem)
		return;
3013
	__mem_cgroup_cancel_charge(mem, 1);
3014 3015
}

3016 3017 3018
static void mem_cgroup_do_uncharge(struct mem_cgroup *mem,
				   unsigned int nr_pages,
				   const enum charge_type ctype)
3019 3020 3021
{
	struct memcg_batch_info *batch = NULL;
	bool uncharge_memsw = true;
3022

3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034
	/* 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;
3035 3036
	/*
	 * do_batch > 0 when unmapping pages or inode invalidate/truncate.
L
Lucas De Marchi 已提交
3037
	 * In those cases, all pages freed continuously can be expected to be in
3038 3039 3040 3041 3042 3043 3044 3045
	 * 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;

3046
	if (nr_pages > 1)
A
Andrea Arcangeli 已提交
3047 3048
		goto direct_uncharge;

3049 3050 3051 3052 3053 3054 3055 3056
	/*
	 * 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 */
3057
	batch->nr_pages++;
3058
	if (uncharge_memsw)
3059
		batch->memsw_nr_pages++;
3060 3061
	return;
direct_uncharge:
3062
	res_counter_uncharge(&mem->res, nr_pages * PAGE_SIZE);
3063
	if (uncharge_memsw)
3064
		res_counter_uncharge(&mem->memsw, nr_pages * PAGE_SIZE);
3065 3066
	if (unlikely(batch->memcg != mem))
		memcg_oom_recover(mem);
3067 3068
	return;
}
3069

3070
/*
3071
 * uncharge if !page_mapped(page)
3072
 */
3073
static struct mem_cgroup *
3074
__mem_cgroup_uncharge_common(struct page *page, enum charge_type ctype)
3075
{
3076
	struct mem_cgroup *mem = NULL;
3077 3078
	unsigned int nr_pages = 1;
	struct page_cgroup *pc;
3079

3080
	if (mem_cgroup_disabled())
3081
		return NULL;
3082

K
KAMEZAWA Hiroyuki 已提交
3083
	if (PageSwapCache(page))
3084
		return NULL;
K
KAMEZAWA Hiroyuki 已提交
3085

A
Andrea Arcangeli 已提交
3086
	if (PageTransHuge(page)) {
3087
		nr_pages <<= compound_order(page);
A
Andrea Arcangeli 已提交
3088 3089
		VM_BUG_ON(!PageTransHuge(page));
	}
3090
	/*
3091
	 * Check if our page_cgroup is valid
3092
	 */
3093 3094
	pc = lookup_page_cgroup(page);
	if (unlikely(!pc || !PageCgroupUsed(pc)))
3095
		return NULL;
3096

3097
	lock_page_cgroup(pc);
K
KAMEZAWA Hiroyuki 已提交
3098

3099 3100
	mem = pc->mem_cgroup;

K
KAMEZAWA Hiroyuki 已提交
3101 3102 3103 3104 3105
	if (!PageCgroupUsed(pc))
		goto unlock_out;

	switch (ctype) {
	case MEM_CGROUP_CHARGE_TYPE_MAPPED:
K
KAMEZAWA Hiroyuki 已提交
3106
	case MEM_CGROUP_CHARGE_TYPE_DROP:
3107 3108
		/* See mem_cgroup_prepare_migration() */
		if (page_mapped(page) || PageCgroupMigration(pc))
K
KAMEZAWA Hiroyuki 已提交
3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119
			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;
3120
	}
K
KAMEZAWA Hiroyuki 已提交
3121

3122
	mem_cgroup_charge_statistics(mem, PageCgroupCache(pc), -nr_pages);
K
KAMEZAWA Hiroyuki 已提交
3123

3124
	ClearPageCgroupUsed(pc);
3125 3126 3127 3128 3129 3130
	/*
	 * 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.
	 */
3131

3132
	unlock_page_cgroup(pc);
K
KAMEZAWA Hiroyuki 已提交
3133 3134 3135 3136
	/*
	 * even after unlock, we have mem->res.usage here and this memcg
	 * will never be freed.
	 */
3137
	memcg_check_events(mem, page);
K
KAMEZAWA Hiroyuki 已提交
3138 3139 3140 3141 3142
	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))
3143
		mem_cgroup_do_uncharge(mem, nr_pages, ctype);
3144

3145
	return mem;
K
KAMEZAWA Hiroyuki 已提交
3146 3147 3148

unlock_out:
	unlock_page_cgroup(pc);
3149
	return NULL;
3150 3151
}

3152 3153
void mem_cgroup_uncharge_page(struct page *page)
{
3154 3155 3156 3157 3158
	/* early check. */
	if (page_mapped(page))
		return;
	if (page->mapping && !PageAnon(page))
		return;
3159 3160 3161 3162 3163 3164
	__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));
3165
	VM_BUG_ON(page->mapping);
3166 3167 3168
	__mem_cgroup_uncharge_common(page, MEM_CGROUP_CHARGE_TYPE_CACHE);
}

3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182
/*
 * 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;
3183 3184
		current->memcg_batch.nr_pages = 0;
		current->memcg_batch.memsw_nr_pages = 0;
3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204
	}
}

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.
	 */
3205 3206 3207 3208 3209 3210
	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);
3211
	memcg_oom_recover(batch->memcg);
3212 3213 3214 3215
	/* forget this pointer (for sanity check) */
	batch->memcg = NULL;
}

3216
#ifdef CONFIG_SWAP
3217
/*
3218
 * called after __delete_from_swap_cache() and drop "page" account.
3219 3220
 * memcg information is recorded to swap_cgroup of "ent"
 */
K
KAMEZAWA Hiroyuki 已提交
3221 3222
void
mem_cgroup_uncharge_swapcache(struct page *page, swp_entry_t ent, bool swapout)
3223 3224
{
	struct mem_cgroup *memcg;
K
KAMEZAWA Hiroyuki 已提交
3225 3226 3227 3228 3229 3230
	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);
3231

K
KAMEZAWA Hiroyuki 已提交
3232 3233 3234 3235 3236
	/*
	 * record memcg information,  if swapout && memcg != NULL,
	 * mem_cgroup_get() was called in uncharge().
	 */
	if (do_swap_account && swapout && memcg)
3237
		swap_cgroup_record(ent, css_id(&memcg->css));
3238
}
3239
#endif
3240 3241 3242 3243 3244 3245 3246

#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
3247
{
3248
	struct mem_cgroup *memcg;
3249
	unsigned short id;
3250 3251 3252 3253

	if (!do_swap_account)
		return;

3254 3255 3256
	id = swap_cgroup_record(ent, 0);
	rcu_read_lock();
	memcg = mem_cgroup_lookup(id);
3257
	if (memcg) {
3258 3259 3260 3261
		/*
		 * We uncharge this because swap is freed.
		 * This memcg can be obsolete one. We avoid calling css_tryget
		 */
3262
		if (!mem_cgroup_is_root(memcg))
3263
			res_counter_uncharge(&memcg->memsw, PAGE_SIZE);
3264
		mem_cgroup_swap_statistics(memcg, false);
3265 3266
		mem_cgroup_put(memcg);
	}
3267
	rcu_read_unlock();
K
KAMEZAWA Hiroyuki 已提交
3268
}
3269 3270 3271 3272 3273 3274

/**
 * 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
3275
 * @need_fixup: whether we should fixup res_counters and refcounts.
3276 3277 3278 3279 3280 3281 3282 3283 3284 3285
 *
 * 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,
3286
		struct mem_cgroup *from, struct mem_cgroup *to, bool need_fixup)
3287 3288 3289 3290 3291 3292 3293 3294
{
	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);
3295
		mem_cgroup_swap_statistics(to, true);
3296
		/*
3297 3298 3299 3300 3301 3302
		 * 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.
3303 3304
		 */
		mem_cgroup_get(to);
3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315
		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);
		}
3316 3317 3318 3319 3320 3321
		return 0;
	}
	return -EINVAL;
}
#else
static inline int mem_cgroup_move_swap_account(swp_entry_t entry,
3322
		struct mem_cgroup *from, struct mem_cgroup *to, bool need_fixup)
3323 3324 3325
{
	return -EINVAL;
}
3326
#endif
K
KAMEZAWA Hiroyuki 已提交
3327

3328
/*
3329 3330
 * Before starting migration, account PAGE_SIZE to mem_cgroup that the old
 * page belongs to.
3331
 */
3332
int mem_cgroup_prepare_migration(struct page *page,
3333
	struct page *newpage, struct mem_cgroup **ptr, gfp_t gfp_mask)
3334
{
3335
	struct mem_cgroup *mem = NULL;
3336
	struct page_cgroup *pc;
3337
	enum charge_type ctype;
3338
	int ret = 0;
3339

3340 3341
	*ptr = NULL;

A
Andrea Arcangeli 已提交
3342
	VM_BUG_ON(PageTransHuge(page));
3343
	if (mem_cgroup_disabled())
3344 3345
		return 0;

3346 3347 3348
	pc = lookup_page_cgroup(page);
	lock_page_cgroup(pc);
	if (PageCgroupUsed(pc)) {
3349 3350
		mem = pc->mem_cgroup;
		css_get(&mem->css);
3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381
		/*
		 * 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);
3382
	}
3383
	unlock_page_cgroup(pc);
3384 3385 3386 3387 3388 3389
	/*
	 * If the page is not charged at this point,
	 * we return here.
	 */
	if (!mem)
		return 0;
3390

A
Andrea Arcangeli 已提交
3391
	*ptr = mem;
3392
	ret = __mem_cgroup_try_charge(NULL, gfp_mask, 1, ptr, false);
3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404
	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;
3405
	}
3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418
	/*
	 * 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;
3419
	__mem_cgroup_commit_charge(mem, page, 1, pc, ctype);
3420
	return ret;
3421
}
3422

3423
/* remove redundant charge if migration failed*/
3424
void mem_cgroup_end_migration(struct mem_cgroup *mem,
3425
	struct page *oldpage, struct page *newpage, bool migration_ok)
3426
{
3427
	struct page *used, *unused;
3428 3429 3430 3431
	struct page_cgroup *pc;

	if (!mem)
		return;
3432
	/* blocks rmdir() */
3433
	cgroup_exclude_rmdir(&mem->css);
3434
	if (!migration_ok) {
3435 3436
		used = oldpage;
		unused = newpage;
3437
	} else {
3438
		used = newpage;
3439 3440
		unused = oldpage;
	}
3441
	/*
3442 3443 3444
	 * 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.
3445
	 */
3446 3447 3448 3449
	pc = lookup_page_cgroup(oldpage);
	lock_page_cgroup(pc);
	ClearPageCgroupMigration(pc);
	unlock_page_cgroup(pc);
3450

3451 3452
	__mem_cgroup_uncharge_common(unused, MEM_CGROUP_CHARGE_TYPE_FORCE);

3453
	/*
3454 3455 3456 3457 3458 3459
	 * 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)
3460
	 */
3461 3462
	if (PageAnon(used))
		mem_cgroup_uncharge_page(used);
3463
	/*
3464 3465
	 * At migration, we may charge account against cgroup which has no
	 * tasks.
3466 3467 3468 3469
	 * 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);
3470
}
3471

3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517
#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

3518 3519
static DEFINE_MUTEX(set_limit_mutex);

3520
static int mem_cgroup_resize_limit(struct mem_cgroup *memcg,
3521
				unsigned long long val)
3522
{
3523
	int retry_count;
3524
	u64 memswlimit, memlimit;
3525
	int ret = 0;
3526 3527
	int children = mem_cgroup_count_children(memcg);
	u64 curusage, oldusage;
3528
	int enlarge;
3529 3530 3531 3532 3533 3534 3535 3536 3537

	/*
	 * 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);
3538

3539
	enlarge = 0;
3540
	while (retry_count) {
3541 3542 3543 3544
		if (signal_pending(current)) {
			ret = -EINTR;
			break;
		}
3545 3546 3547 3548 3549 3550 3551 3552 3553 3554
		/*
		 * 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);
3555 3556
			break;
		}
3557 3558 3559 3560 3561

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

3562
		ret = res_counter_set_limit(&memcg->res, val);
3563 3564 3565 3566 3567 3568
		if (!ret) {
			if (memswlimit == val)
				memcg->memsw_is_minimum = true;
			else
				memcg->memsw_is_minimum = false;
		}
3569 3570 3571 3572 3573
		mutex_unlock(&set_limit_mutex);

		if (!ret)
			break;

3574
		mem_cgroup_hierarchical_reclaim(memcg, NULL, GFP_KERNEL,
3575 3576
						MEM_CGROUP_RECLAIM_SHRINK,
						NULL);
3577 3578 3579 3580 3581 3582
		curusage = res_counter_read_u64(&memcg->res, RES_USAGE);
		/* Usage is reduced ? */
  		if (curusage >= oldusage)
			retry_count--;
		else
			oldusage = curusage;
3583
	}
3584 3585
	if (!ret && enlarge)
		memcg_oom_recover(memcg);
3586

3587 3588 3589
	return ret;
}

L
Li Zefan 已提交
3590 3591
static int mem_cgroup_resize_memsw_limit(struct mem_cgroup *memcg,
					unsigned long long val)
3592
{
3593
	int retry_count;
3594
	u64 memlimit, memswlimit, oldusage, curusage;
3595 3596
	int children = mem_cgroup_count_children(memcg);
	int ret = -EBUSY;
3597
	int enlarge = 0;
3598

3599 3600 3601
	/* see mem_cgroup_resize_res_limit */
 	retry_count = children * MEM_CGROUP_RECLAIM_RETRIES;
	oldusage = res_counter_read_u64(&memcg->memsw, RES_USAGE);
3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618
	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;
		}
3619 3620 3621
		memswlimit = res_counter_read_u64(&memcg->memsw, RES_LIMIT);
		if (memswlimit < val)
			enlarge = 1;
3622
		ret = res_counter_set_limit(&memcg->memsw, val);
3623 3624 3625 3626 3627 3628
		if (!ret) {
			if (memlimit == val)
				memcg->memsw_is_minimum = true;
			else
				memcg->memsw_is_minimum = false;
		}
3629 3630 3631 3632 3633
		mutex_unlock(&set_limit_mutex);

		if (!ret)
			break;

3634
		mem_cgroup_hierarchical_reclaim(memcg, NULL, GFP_KERNEL,
3635
						MEM_CGROUP_RECLAIM_NOSWAP |
3636 3637
						MEM_CGROUP_RECLAIM_SHRINK,
						NULL);
3638
		curusage = res_counter_read_u64(&memcg->memsw, RES_USAGE);
3639
		/* Usage is reduced ? */
3640
		if (curusage >= oldusage)
3641
			retry_count--;
3642 3643
		else
			oldusage = curusage;
3644
	}
3645 3646
	if (!ret && enlarge)
		memcg_oom_recover(memcg);
3647 3648 3649
	return ret;
}

3650
unsigned long mem_cgroup_soft_limit_reclaim(struct zone *zone, int order,
3651 3652
					    gfp_t gfp_mask,
					    unsigned long *total_scanned)
3653 3654 3655 3656 3657 3658
{
	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;
3659
	unsigned long long excess;
3660
	unsigned long nr_scanned;
3661 3662 3663 3664

	if (order > 0)
		return 0;

3665
	mctz = soft_limit_tree_node_zone(zone_to_nid(zone), zone_idx(zone));
3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678
	/*
	 * 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;

3679
		nr_scanned = 0;
3680 3681
		reclaimed = mem_cgroup_hierarchical_reclaim(mz->mem, zone,
						gfp_mask,
3682 3683
						MEM_CGROUP_RECLAIM_SOFT,
						&nr_scanned);
3684
		nr_reclaimed += reclaimed;
3685
		*total_scanned += nr_scanned;
3686 3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707
		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);
3708
				if (next_mz == mz)
3709
					css_put(&next_mz->mem->css);
3710
				else /* next_mz == NULL or other memcg */
3711 3712 3713 3714
					break;
			} while (1);
		}
		__mem_cgroup_remove_exceeded(mz->mem, mz, mctz);
3715
		excess = res_counter_soft_limit_excess(&mz->mem->res);
3716 3717 3718 3719 3720 3721 3722 3723
		/*
		 * 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.
		 */
3724 3725
		/* If excess == 0, no tree ops */
		__mem_cgroup_insert_exceeded(mz->mem, mz, mctz, excess);
3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743
		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;
}

3744 3745 3746 3747
/*
 * This routine traverse page_cgroup in given list and drop them all.
 * *And* this routine doesn't reclaim page itself, just removes page_cgroup.
 */
3748
static int mem_cgroup_force_empty_list(struct mem_cgroup *mem,
K
KAMEZAWA Hiroyuki 已提交
3749
				int node, int zid, enum lru_list lru)
3750
{
K
KAMEZAWA Hiroyuki 已提交
3751 3752
	struct zone *zone;
	struct mem_cgroup_per_zone *mz;
3753
	struct page_cgroup *pc, *busy;
K
KAMEZAWA Hiroyuki 已提交
3754
	unsigned long flags, loop;
3755
	struct list_head *list;
3756
	int ret = 0;
3757

K
KAMEZAWA Hiroyuki 已提交
3758 3759
	zone = &NODE_DATA(node)->node_zones[zid];
	mz = mem_cgroup_zoneinfo(mem, node, zid);
3760
	list = &mz->lists[lru];
3761

3762 3763 3764 3765 3766
	loop = MEM_CGROUP_ZSTAT(mz, lru);
	/* give some margin against EBUSY etc...*/
	loop += 256;
	busy = NULL;
	while (loop--) {
3767 3768
		struct page *page;

3769
		ret = 0;
K
KAMEZAWA Hiroyuki 已提交
3770
		spin_lock_irqsave(&zone->lru_lock, flags);
3771
		if (list_empty(list)) {
K
KAMEZAWA Hiroyuki 已提交
3772
			spin_unlock_irqrestore(&zone->lru_lock, flags);
3773
			break;
3774 3775 3776 3777
		}
		pc = list_entry(list->prev, struct page_cgroup, lru);
		if (busy == pc) {
			list_move(&pc->lru, list);
3778
			busy = NULL;
K
KAMEZAWA Hiroyuki 已提交
3779
			spin_unlock_irqrestore(&zone->lru_lock, flags);
3780 3781
			continue;
		}
K
KAMEZAWA Hiroyuki 已提交
3782
		spin_unlock_irqrestore(&zone->lru_lock, flags);
3783

3784
		page = lookup_cgroup_page(pc);
3785 3786

		ret = mem_cgroup_move_parent(page, pc, mem, GFP_KERNEL);
3787
		if (ret == -ENOMEM)
3788
			break;
3789 3790 3791 3792 3793 3794 3795

		if (ret == -EBUSY || ret == -EINVAL) {
			/* found lock contention or "pc" is obsolete. */
			busy = pc;
			cond_resched();
		} else
			busy = NULL;
3796
	}
K
KAMEZAWA Hiroyuki 已提交
3797

3798 3799 3800
	if (!ret && !list_empty(list))
		return -EBUSY;
	return ret;
3801 3802 3803 3804 3805 3806
}

/*
 * make mem_cgroup's charge to be 0 if there is no task.
 * This enables deleting this mem_cgroup.
 */
3807
static int mem_cgroup_force_empty(struct mem_cgroup *mem, bool free_all)
3808
{
3809 3810 3811
	int ret;
	int node, zid, shrink;
	int nr_retries = MEM_CGROUP_RECLAIM_RETRIES;
3812
	struct cgroup *cgrp = mem->css.cgroup;
3813

3814
	css_get(&mem->css);
3815 3816

	shrink = 0;
3817 3818 3819
	/* should free all ? */
	if (free_all)
		goto try_to_free;
3820
move_account:
3821
	do {
3822
		ret = -EBUSY;
3823 3824 3825 3826
		if (cgroup_task_count(cgrp) || !list_empty(&cgrp->children))
			goto out;
		ret = -EINTR;
		if (signal_pending(current))
3827
			goto out;
3828 3829
		/* This is for making all *used* pages to be on LRU. */
		lru_add_drain_all();
3830
		drain_all_stock_sync(mem);
3831
		ret = 0;
3832
		mem_cgroup_start_move(mem);
3833
		for_each_node_state(node, N_HIGH_MEMORY) {
3834
			for (zid = 0; !ret && zid < MAX_NR_ZONES; zid++) {
3835
				enum lru_list l;
3836 3837
				for_each_lru(l) {
					ret = mem_cgroup_force_empty_list(mem,
K
KAMEZAWA Hiroyuki 已提交
3838
							node, zid, l);
3839 3840 3841
					if (ret)
						break;
				}
3842
			}
3843 3844 3845
			if (ret)
				break;
		}
3846
		mem_cgroup_end_move(mem);
3847
		memcg_oom_recover(mem);
3848 3849 3850
		/* it seems parent cgroup doesn't have enough mem */
		if (ret == -ENOMEM)
			goto try_to_free;
3851
		cond_resched();
3852 3853
	/* "ret" should also be checked to ensure all lists are empty. */
	} while (mem->res.usage > 0 || ret);
3854 3855 3856
out:
	css_put(&mem->css);
	return ret;
3857 3858

try_to_free:
3859 3860
	/* returns EBUSY if there is a task or if we come here twice. */
	if (cgroup_task_count(cgrp) || !list_empty(&cgrp->children) || shrink) {
3861 3862 3863
		ret = -EBUSY;
		goto out;
	}
3864 3865
	/* we call try-to-free pages for make this cgroup empty */
	lru_add_drain_all();
3866 3867 3868
	/* try to free all pages in this cgroup */
	shrink = 1;
	while (nr_retries && mem->res.usage > 0) {
3869
		struct memcg_scanrecord rec;
3870
		int progress;
3871 3872 3873 3874 3875

		if (signal_pending(current)) {
			ret = -EINTR;
			goto out;
		}
3876 3877 3878
		rec.context = SCAN_BY_SHRINK;
		rec.mem = mem;
		rec.root = mem;
K
KOSAKI Motohiro 已提交
3879
		progress = try_to_free_mem_cgroup_pages(mem, GFP_KERNEL,
3880
						false, &rec);
3881
		if (!progress) {
3882
			nr_retries--;
3883
			/* maybe some writeback is necessary */
3884
			congestion_wait(BLK_RW_ASYNC, HZ/10);
3885
		}
3886 3887

	}
K
KAMEZAWA Hiroyuki 已提交
3888
	lru_add_drain();
3889
	/* try move_account...there may be some *locked* pages. */
3890
	goto move_account;
3891 3892
}

3893 3894 3895 3896 3897 3898
int mem_cgroup_force_empty_write(struct cgroup *cont, unsigned int event)
{
	return mem_cgroup_force_empty(mem_cgroup_from_cont(cont), true);
}


3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916
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();
	/*
3917
	 * If parent's use_hierarchy is set, we can't make any modifications
3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936
	 * 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;
}

3937

3938 3939
static unsigned long mem_cgroup_recursive_stat(struct mem_cgroup *mem,
					       enum mem_cgroup_stat_index idx)
3940
{
K
KAMEZAWA Hiroyuki 已提交
3941
	struct mem_cgroup *iter;
3942
	long val = 0;
3943

3944
	/* Per-cpu values can be negative, use a signed accumulator */
K
KAMEZAWA Hiroyuki 已提交
3945 3946 3947 3948 3949 3950
	for_each_mem_cgroup_tree(iter, mem)
		val += mem_cgroup_read_stat(iter, idx);

	if (val < 0) /* race ? */
		val = 0;
	return val;
3951 3952
}

3953 3954
static inline u64 mem_cgroup_usage(struct mem_cgroup *mem, bool swap)
{
K
KAMEZAWA Hiroyuki 已提交
3955
	u64 val;
3956 3957 3958 3959 3960 3961 3962 3963

	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);
	}

3964 3965
	val = mem_cgroup_recursive_stat(mem, MEM_CGROUP_STAT_CACHE);
	val += mem_cgroup_recursive_stat(mem, MEM_CGROUP_STAT_RSS);
3966

K
KAMEZAWA Hiroyuki 已提交
3967
	if (swap)
3968
		val += mem_cgroup_recursive_stat(mem, MEM_CGROUP_STAT_SWAPOUT);
3969 3970 3971 3972

	return val << PAGE_SHIFT;
}

3973
static u64 mem_cgroup_read(struct cgroup *cont, struct cftype *cft)
B
Balbir Singh 已提交
3974
{
3975
	struct mem_cgroup *mem = mem_cgroup_from_cont(cont);
3976
	u64 val;
3977 3978 3979 3980 3981 3982
	int type, name;

	type = MEMFILE_TYPE(cft->private);
	name = MEMFILE_ATTR(cft->private);
	switch (type) {
	case _MEM:
3983 3984 3985
		if (name == RES_USAGE)
			val = mem_cgroup_usage(mem, false);
		else
3986
			val = res_counter_read_u64(&mem->res, name);
3987 3988
		break;
	case _MEMSWAP:
3989 3990 3991
		if (name == RES_USAGE)
			val = mem_cgroup_usage(mem, true);
		else
3992
			val = res_counter_read_u64(&mem->memsw, name);
3993 3994 3995 3996 3997 3998
		break;
	default:
		BUG();
		break;
	}
	return val;
B
Balbir Singh 已提交
3999
}
4000 4001 4002 4003
/*
 * The user of this function is...
 * RES_LIMIT.
 */
4004 4005
static int mem_cgroup_write(struct cgroup *cont, struct cftype *cft,
			    const char *buffer)
B
Balbir Singh 已提交
4006
{
4007
	struct mem_cgroup *memcg = mem_cgroup_from_cont(cont);
4008
	int type, name;
4009 4010 4011
	unsigned long long val;
	int ret;

4012 4013 4014
	type = MEMFILE_TYPE(cft->private);
	name = MEMFILE_ATTR(cft->private);
	switch (name) {
4015
	case RES_LIMIT:
4016 4017 4018 4019
		if (mem_cgroup_is_root(memcg)) { /* Can't set limit on root */
			ret = -EINVAL;
			break;
		}
4020 4021
		/* This function does all necessary parse...reuse it */
		ret = res_counter_memparse_write_strategy(buffer, &val);
4022 4023 4024
		if (ret)
			break;
		if (type == _MEM)
4025
			ret = mem_cgroup_resize_limit(memcg, val);
4026 4027
		else
			ret = mem_cgroup_resize_memsw_limit(memcg, val);
4028
		break;
4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042
	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;
4043 4044 4045 4046 4047
	default:
		ret = -EINVAL; /* should be BUG() ? */
		break;
	}
	return ret;
B
Balbir Singh 已提交
4048 4049
}

4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 4076 4077
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;
}

4078
static int mem_cgroup_reset(struct cgroup *cont, unsigned int event)
4079 4080
{
	struct mem_cgroup *mem;
4081
	int type, name;
4082 4083

	mem = mem_cgroup_from_cont(cont);
4084 4085 4086
	type = MEMFILE_TYPE(event);
	name = MEMFILE_ATTR(event);
	switch (name) {
4087
	case RES_MAX_USAGE:
4088 4089 4090 4091
		if (type == _MEM)
			res_counter_reset_max(&mem->res);
		else
			res_counter_reset_max(&mem->memsw);
4092 4093
		break;
	case RES_FAILCNT:
4094 4095 4096 4097
		if (type == _MEM)
			res_counter_reset_failcnt(&mem->res);
		else
			res_counter_reset_failcnt(&mem->memsw);
4098 4099
		break;
	}
4100

4101
	return 0;
4102 4103
}

4104 4105 4106 4107 4108 4109
static u64 mem_cgroup_move_charge_read(struct cgroup *cgrp,
					struct cftype *cft)
{
	return mem_cgroup_from_cont(cgrp)->move_charge_at_immigrate;
}

4110
#ifdef CONFIG_MMU
4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128
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;
}
4129 4130 4131 4132 4133 4134 4135
#else
static int mem_cgroup_move_charge_write(struct cgroup *cgrp,
					struct cftype *cft, u64 val)
{
	return -ENOSYS;
}
#endif
4136

K
KAMEZAWA Hiroyuki 已提交
4137 4138 4139 4140 4141

/* For read statistics */
enum {
	MCS_CACHE,
	MCS_RSS,
4142
	MCS_FILE_MAPPED,
K
KAMEZAWA Hiroyuki 已提交
4143 4144
	MCS_PGPGIN,
	MCS_PGPGOUT,
4145
	MCS_SWAP,
4146 4147
	MCS_PGFAULT,
	MCS_PGMAJFAULT,
K
KAMEZAWA Hiroyuki 已提交
4148 4149 4150 4151 4152 4153 4154 4155 4156 4157
	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];
4158 4159
};

K
KAMEZAWA Hiroyuki 已提交
4160 4161 4162 4163 4164 4165
struct {
	char *local_name;
	char *total_name;
} memcg_stat_strings[NR_MCS_STAT] = {
	{"cache", "total_cache"},
	{"rss", "total_rss"},
4166
	{"mapped_file", "total_mapped_file"},
K
KAMEZAWA Hiroyuki 已提交
4167 4168
	{"pgpgin", "total_pgpgin"},
	{"pgpgout", "total_pgpgout"},
4169
	{"swap", "total_swap"},
4170 4171
	{"pgfault", "total_pgfault"},
	{"pgmajfault", "total_pgmajfault"},
K
KAMEZAWA Hiroyuki 已提交
4172 4173 4174 4175 4176 4177 4178 4179
	{"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
4180 4181
static void
mem_cgroup_get_local_stat(struct mem_cgroup *mem, struct mcs_total_stat *s)
K
KAMEZAWA Hiroyuki 已提交
4182 4183 4184 4185
{
	s64 val;

	/* per cpu stat */
4186
	val = mem_cgroup_read_stat(mem, MEM_CGROUP_STAT_CACHE);
K
KAMEZAWA Hiroyuki 已提交
4187
	s->stat[MCS_CACHE] += val * PAGE_SIZE;
4188
	val = mem_cgroup_read_stat(mem, MEM_CGROUP_STAT_RSS);
K
KAMEZAWA Hiroyuki 已提交
4189
	s->stat[MCS_RSS] += val * PAGE_SIZE;
4190
	val = mem_cgroup_read_stat(mem, MEM_CGROUP_STAT_FILE_MAPPED);
4191
	s->stat[MCS_FILE_MAPPED] += val * PAGE_SIZE;
4192
	val = mem_cgroup_read_events(mem, MEM_CGROUP_EVENTS_PGPGIN);
K
KAMEZAWA Hiroyuki 已提交
4193
	s->stat[MCS_PGPGIN] += val;
4194
	val = mem_cgroup_read_events(mem, MEM_CGROUP_EVENTS_PGPGOUT);
K
KAMEZAWA Hiroyuki 已提交
4195
	s->stat[MCS_PGPGOUT] += val;
4196
	if (do_swap_account) {
4197
		val = mem_cgroup_read_stat(mem, MEM_CGROUP_STAT_SWAPOUT);
4198 4199
		s->stat[MCS_SWAP] += val * PAGE_SIZE;
	}
4200 4201 4202 4203
	val = mem_cgroup_read_events(mem, MEM_CGROUP_EVENTS_PGFAULT);
	s->stat[MCS_PGFAULT] += val;
	val = mem_cgroup_read_events(mem, MEM_CGROUP_EVENTS_PGMAJFAULT);
	s->stat[MCS_PGMAJFAULT] += val;
K
KAMEZAWA Hiroyuki 已提交
4204 4205

	/* per zone stat */
4206
	val = mem_cgroup_nr_lru_pages(mem, BIT(LRU_INACTIVE_ANON));
K
KAMEZAWA Hiroyuki 已提交
4207
	s->stat[MCS_INACTIVE_ANON] += val * PAGE_SIZE;
4208
	val = mem_cgroup_nr_lru_pages(mem, BIT(LRU_ACTIVE_ANON));
K
KAMEZAWA Hiroyuki 已提交
4209
	s->stat[MCS_ACTIVE_ANON] += val * PAGE_SIZE;
4210
	val = mem_cgroup_nr_lru_pages(mem, BIT(LRU_INACTIVE_FILE));
K
KAMEZAWA Hiroyuki 已提交
4211
	s->stat[MCS_INACTIVE_FILE] += val * PAGE_SIZE;
4212
	val = mem_cgroup_nr_lru_pages(mem, BIT(LRU_ACTIVE_FILE));
K
KAMEZAWA Hiroyuki 已提交
4213
	s->stat[MCS_ACTIVE_FILE] += val * PAGE_SIZE;
4214
	val = mem_cgroup_nr_lru_pages(mem, BIT(LRU_UNEVICTABLE));
K
KAMEZAWA Hiroyuki 已提交
4215 4216 4217 4218 4219 4220
	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
4221 4222 4223 4224
	struct mem_cgroup *iter;

	for_each_mem_cgroup_tree(iter, mem)
		mem_cgroup_get_local_stat(iter, s);
K
KAMEZAWA Hiroyuki 已提交
4225 4226
}

4227 4228 4229 4230 4231 4232 4233 4234 4235
#ifdef CONFIG_NUMA
static int mem_control_numa_stat_show(struct seq_file *m, void *arg)
{
	int nid;
	unsigned long total_nr, file_nr, anon_nr, unevictable_nr;
	unsigned long node_nr;
	struct cgroup *cont = m->private;
	struct mem_cgroup *mem_cont = mem_cgroup_from_cont(cont);

4236
	total_nr = mem_cgroup_nr_lru_pages(mem_cont, LRU_ALL);
4237 4238
	seq_printf(m, "total=%lu", total_nr);
	for_each_node_state(nid, N_HIGH_MEMORY) {
4239
		node_nr = mem_cgroup_node_nr_lru_pages(mem_cont, nid, LRU_ALL);
4240 4241 4242 4243
		seq_printf(m, " N%d=%lu", nid, node_nr);
	}
	seq_putc(m, '\n');

4244
	file_nr = mem_cgroup_nr_lru_pages(mem_cont, LRU_ALL_FILE);
4245 4246
	seq_printf(m, "file=%lu", file_nr);
	for_each_node_state(nid, N_HIGH_MEMORY) {
4247 4248
		node_nr = mem_cgroup_node_nr_lru_pages(mem_cont, nid,
				LRU_ALL_FILE);
4249 4250 4251 4252
		seq_printf(m, " N%d=%lu", nid, node_nr);
	}
	seq_putc(m, '\n');

4253
	anon_nr = mem_cgroup_nr_lru_pages(mem_cont, LRU_ALL_ANON);
4254 4255
	seq_printf(m, "anon=%lu", anon_nr);
	for_each_node_state(nid, N_HIGH_MEMORY) {
4256 4257
		node_nr = mem_cgroup_node_nr_lru_pages(mem_cont, nid,
				LRU_ALL_ANON);
4258 4259 4260 4261
		seq_printf(m, " N%d=%lu", nid, node_nr);
	}
	seq_putc(m, '\n');

4262
	unevictable_nr = mem_cgroup_nr_lru_pages(mem_cont, BIT(LRU_UNEVICTABLE));
4263 4264
	seq_printf(m, "unevictable=%lu", unevictable_nr);
	for_each_node_state(nid, N_HIGH_MEMORY) {
4265 4266
		node_nr = mem_cgroup_node_nr_lru_pages(mem_cont, nid,
				BIT(LRU_UNEVICTABLE));
4267 4268 4269 4270 4271 4272 4273
		seq_printf(m, " N%d=%lu", nid, node_nr);
	}
	seq_putc(m, '\n');
	return 0;
}
#endif /* CONFIG_NUMA */

4274 4275
static int mem_control_stat_show(struct cgroup *cont, struct cftype *cft,
				 struct cgroup_map_cb *cb)
4276 4277
{
	struct mem_cgroup *mem_cont = mem_cgroup_from_cont(cont);
K
KAMEZAWA Hiroyuki 已提交
4278
	struct mcs_total_stat mystat;
4279 4280
	int i;

K
KAMEZAWA Hiroyuki 已提交
4281 4282
	memset(&mystat, 0, sizeof(mystat));
	mem_cgroup_get_local_stat(mem_cont, &mystat);
4283

4284

4285 4286 4287
	for (i = 0; i < NR_MCS_STAT; i++) {
		if (i == MCS_SWAP && !do_swap_account)
			continue;
K
KAMEZAWA Hiroyuki 已提交
4288
		cb->fill(cb, memcg_stat_strings[i].local_name, mystat.stat[i]);
4289
	}
L
Lee Schermerhorn 已提交
4290

K
KAMEZAWA Hiroyuki 已提交
4291
	/* Hierarchical information */
4292 4293 4294 4295 4296 4297 4298
	{
		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
4299

K
KAMEZAWA Hiroyuki 已提交
4300 4301
	memset(&mystat, 0, sizeof(mystat));
	mem_cgroup_get_total_stat(mem_cont, &mystat);
4302 4303 4304
	for (i = 0; i < NR_MCS_STAT; i++) {
		if (i == MCS_SWAP && !do_swap_account)
			continue;
K
KAMEZAWA Hiroyuki 已提交
4305
		cb->fill(cb, memcg_stat_strings[i].total_name, mystat.stat[i]);
4306
	}
K
KAMEZAWA Hiroyuki 已提交
4307

K
KOSAKI Motohiro 已提交
4308
#ifdef CONFIG_DEBUG_VM
4309
	cb->fill(cb, "inactive_ratio", calc_inactive_ratio(mem_cont, NULL));
K
KOSAKI Motohiro 已提交
4310 4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336

	{
		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

4337 4338 4339
	return 0;
}

K
KOSAKI Motohiro 已提交
4340 4341 4342 4343
static u64 mem_cgroup_swappiness_read(struct cgroup *cgrp, struct cftype *cft)
{
	struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp);

4344
	return mem_cgroup_swappiness(memcg);
K
KOSAKI Motohiro 已提交
4345 4346 4347 4348 4349 4350 4351
}

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;
4352

K
KOSAKI Motohiro 已提交
4353 4354 4355 4356 4357 4358 4359
	if (val > 100)
		return -EINVAL;

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

	parent = mem_cgroup_from_cont(cgrp->parent);
4360 4361 4362

	cgroup_lock();

K
KOSAKI Motohiro 已提交
4363 4364
	/* If under hierarchy, only empty-root can set this value */
	if ((parent->use_hierarchy) ||
4365 4366
	    (memcg->use_hierarchy && !list_empty(&cgrp->children))) {
		cgroup_unlock();
K
KOSAKI Motohiro 已提交
4367
		return -EINVAL;
4368
	}
K
KOSAKI Motohiro 已提交
4369 4370 4371

	memcg->swappiness = val;

4372 4373
	cgroup_unlock();

K
KOSAKI Motohiro 已提交
4374 4375 4376
	return 0;
}

4377 4378 4379 4380 4381 4382 4383 4384
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)
4385
		t = rcu_dereference(memcg->thresholds.primary);
4386
	else
4387
		t = rcu_dereference(memcg->memsw_thresholds.primary);
4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398

	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().
	 */
4399
	i = t->current_threshold;
4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422

	/*
	 * 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 */
4423
	t->current_threshold = i - 1;
4424 4425 4426 4427 4428 4429
unlock:
	rcu_read_unlock();
}

static void mem_cgroup_threshold(struct mem_cgroup *memcg)
{
4430 4431 4432 4433 4434 4435 4436
	while (memcg) {
		__mem_cgroup_threshold(memcg, false);
		if (do_swap_account)
			__mem_cgroup_threshold(memcg, true);

		memcg = parent_mem_cgroup(memcg);
	}
4437 4438 4439 4440 4441 4442 4443 4444 4445 4446
}

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
4447
static int mem_cgroup_oom_notify_cb(struct mem_cgroup *mem)
K
KAMEZAWA Hiroyuki 已提交
4448 4449 4450 4451 4452 4453 4454 4455 4456 4457
{
	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
4458 4459 4460 4461
	struct mem_cgroup *iter;

	for_each_mem_cgroup_tree(iter, mem)
		mem_cgroup_oom_notify_cb(iter);
K
KAMEZAWA Hiroyuki 已提交
4462 4463 4464 4465
}

static int mem_cgroup_usage_register_event(struct cgroup *cgrp,
	struct cftype *cft, struct eventfd_ctx *eventfd, const char *args)
4466 4467
{
	struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp);
4468 4469
	struct mem_cgroup_thresholds *thresholds;
	struct mem_cgroup_threshold_ary *new;
4470 4471
	int type = MEMFILE_TYPE(cft->private);
	u64 threshold, usage;
4472
	int i, size, ret;
4473 4474 4475 4476 4477 4478

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

	mutex_lock(&memcg->thresholds_lock);
4479

4480
	if (type == _MEM)
4481
		thresholds = &memcg->thresholds;
4482
	else if (type == _MEMSWAP)
4483
		thresholds = &memcg->memsw_thresholds;
4484 4485 4486 4487 4488 4489
	else
		BUG();

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

	/* Check if a threshold crossed before adding a new one */
4490
	if (thresholds->primary)
4491 4492
		__mem_cgroup_threshold(memcg, type == _MEMSWAP);

4493
	size = thresholds->primary ? thresholds->primary->size + 1 : 1;
4494 4495

	/* Allocate memory for new array of thresholds */
4496
	new = kmalloc(sizeof(*new) + size * sizeof(struct mem_cgroup_threshold),
4497
			GFP_KERNEL);
4498
	if (!new) {
4499 4500 4501
		ret = -ENOMEM;
		goto unlock;
	}
4502
	new->size = size;
4503 4504

	/* Copy thresholds (if any) to new array */
4505 4506
	if (thresholds->primary) {
		memcpy(new->entries, thresholds->primary->entries, (size - 1) *
4507
				sizeof(struct mem_cgroup_threshold));
4508 4509
	}

4510
	/* Add new threshold */
4511 4512
	new->entries[size - 1].eventfd = eventfd;
	new->entries[size - 1].threshold = threshold;
4513 4514

	/* Sort thresholds. Registering of new threshold isn't time-critical */
4515
	sort(new->entries, size, sizeof(struct mem_cgroup_threshold),
4516 4517 4518
			compare_thresholds, NULL);

	/* Find current threshold */
4519
	new->current_threshold = -1;
4520
	for (i = 0; i < size; i++) {
4521
		if (new->entries[i].threshold < usage) {
4522
			/*
4523 4524
			 * new->current_threshold will not be used until
			 * rcu_assign_pointer(), so it's safe to increment
4525 4526
			 * it here.
			 */
4527
			++new->current_threshold;
4528 4529 4530
		}
	}

4531 4532 4533 4534 4535
	/* Free old spare buffer and save old primary buffer as spare */
	kfree(thresholds->spare);
	thresholds->spare = thresholds->primary;

	rcu_assign_pointer(thresholds->primary, new);
4536

4537
	/* To be sure that nobody uses thresholds */
4538 4539 4540 4541 4542 4543 4544 4545
	synchronize_rcu();

unlock:
	mutex_unlock(&memcg->thresholds_lock);

	return ret;
}

4546
static void mem_cgroup_usage_unregister_event(struct cgroup *cgrp,
K
KAMEZAWA Hiroyuki 已提交
4547
	struct cftype *cft, struct eventfd_ctx *eventfd)
4548 4549
{
	struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp);
4550 4551
	struct mem_cgroup_thresholds *thresholds;
	struct mem_cgroup_threshold_ary *new;
4552 4553
	int type = MEMFILE_TYPE(cft->private);
	u64 usage;
4554
	int i, j, size;
4555 4556 4557

	mutex_lock(&memcg->thresholds_lock);
	if (type == _MEM)
4558
		thresholds = &memcg->thresholds;
4559
	else if (type == _MEMSWAP)
4560
		thresholds = &memcg->memsw_thresholds;
4561 4562 4563 4564 4565 4566 4567 4568 4569 4570 4571 4572 4573 4574 4575
	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 */
4576 4577 4578
	size = 0;
	for (i = 0; i < thresholds->primary->size; i++) {
		if (thresholds->primary->entries[i].eventfd != eventfd)
4579 4580 4581
			size++;
	}

4582
	new = thresholds->spare;
4583

4584 4585
	/* Set thresholds array to NULL if we don't have thresholds */
	if (!size) {
4586 4587
		kfree(new);
		new = NULL;
4588
		goto swap_buffers;
4589 4590
	}

4591
	new->size = size;
4592 4593

	/* Copy thresholds and find current threshold */
4594 4595 4596
	new->current_threshold = -1;
	for (i = 0, j = 0; i < thresholds->primary->size; i++) {
		if (thresholds->primary->entries[i].eventfd == eventfd)
4597 4598
			continue;

4599 4600
		new->entries[j] = thresholds->primary->entries[i];
		if (new->entries[j].threshold < usage) {
4601
			/*
4602
			 * new->current_threshold will not be used
4603 4604 4605
			 * until rcu_assign_pointer(), so it's safe to increment
			 * it here.
			 */
4606
			++new->current_threshold;
4607 4608 4609 4610
		}
		j++;
	}

4611
swap_buffers:
4612 4613 4614
	/* Swap primary and spare array */
	thresholds->spare = thresholds->primary;
	rcu_assign_pointer(thresholds->primary, new);
4615

4616
	/* To be sure that nobody uses thresholds */
4617 4618 4619 4620
	synchronize_rcu();

	mutex_unlock(&memcg->thresholds_lock);
}
4621

K
KAMEZAWA Hiroyuki 已提交
4622 4623 4624 4625 4626 4627 4628 4629 4630 4631 4632 4633
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;

4634
	spin_lock(&memcg_oom_lock);
K
KAMEZAWA Hiroyuki 已提交
4635 4636 4637 4638 4639

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

	/* already in OOM ? */
4640
	if (atomic_read(&memcg->under_oom))
K
KAMEZAWA Hiroyuki 已提交
4641
		eventfd_signal(eventfd, 1);
4642
	spin_unlock(&memcg_oom_lock);
K
KAMEZAWA Hiroyuki 已提交
4643 4644 4645 4646

	return 0;
}

4647
static void mem_cgroup_oom_unregister_event(struct cgroup *cgrp,
K
KAMEZAWA Hiroyuki 已提交
4648 4649 4650 4651 4652 4653 4654 4655
	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);

4656
	spin_lock(&memcg_oom_lock);
K
KAMEZAWA Hiroyuki 已提交
4657 4658 4659 4660 4661 4662 4663 4664

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

4665
	spin_unlock(&memcg_oom_lock);
K
KAMEZAWA Hiroyuki 已提交
4666 4667
}

4668 4669 4670 4671 4672 4673 4674
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);

4675
	if (atomic_read(&mem->under_oom))
4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701
		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;
4702 4703
	if (!val)
		memcg_oom_recover(mem);
4704 4705 4706 4707
	cgroup_unlock();
	return 0;
}

4708 4709 4710 4711 4712 4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723
#ifdef CONFIG_NUMA
static const struct file_operations mem_control_numa_stat_file_operations = {
	.read = seq_read,
	.llseek = seq_lseek,
	.release = single_release,
};

static int mem_control_numa_stat_open(struct inode *unused, struct file *file)
{
	struct cgroup *cont = file->f_dentry->d_parent->d_fsdata;

	file->f_op = &mem_control_numa_stat_file_operations;
	return single_open(file, mem_control_numa_stat_show, cont);
}
#endif /* CONFIG_NUMA */

4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771
static int mem_cgroup_vmscan_stat_read(struct cgroup *cgrp,
				struct cftype *cft,
				struct cgroup_map_cb *cb)
{
	struct mem_cgroup *mem = mem_cgroup_from_cont(cgrp);
	char string[64];
	int i;

	for (i = 0; i < NR_SCANSTATS; i++) {
		strcpy(string, scanstat_string[i]);
		strcat(string, SCANSTAT_WORD_LIMIT);
		cb->fill(cb, string,  mem->scanstat.stats[SCAN_BY_LIMIT][i]);
	}

	for (i = 0; i < NR_SCANSTATS; i++) {
		strcpy(string, scanstat_string[i]);
		strcat(string, SCANSTAT_WORD_SYSTEM);
		cb->fill(cb, string,  mem->scanstat.stats[SCAN_BY_SYSTEM][i]);
	}

	for (i = 0; i < NR_SCANSTATS; i++) {
		strcpy(string, scanstat_string[i]);
		strcat(string, SCANSTAT_WORD_LIMIT);
		strcat(string, SCANSTAT_WORD_HIERARCHY);
		cb->fill(cb, string,  mem->scanstat.rootstats[SCAN_BY_LIMIT][i]);
	}
	for (i = 0; i < NR_SCANSTATS; i++) {
		strcpy(string, scanstat_string[i]);
		strcat(string, SCANSTAT_WORD_SYSTEM);
		strcat(string, SCANSTAT_WORD_HIERARCHY);
		cb->fill(cb, string,  mem->scanstat.rootstats[SCAN_BY_SYSTEM][i]);
	}
	return 0;
}

static int mem_cgroup_reset_vmscan_stat(struct cgroup *cgrp,
				unsigned int event)
{
	struct mem_cgroup *mem = mem_cgroup_from_cont(cgrp);

	spin_lock(&mem->scanstat.lock);
	memset(&mem->scanstat.stats, 0, sizeof(mem->scanstat.stats));
	memset(&mem->scanstat.rootstats, 0, sizeof(mem->scanstat.rootstats));
	spin_unlock(&mem->scanstat.lock);
	return 0;
}


B
Balbir Singh 已提交
4772 4773
static struct cftype mem_cgroup_files[] = {
	{
4774
		.name = "usage_in_bytes",
4775
		.private = MEMFILE_PRIVATE(_MEM, RES_USAGE),
4776
		.read_u64 = mem_cgroup_read,
K
KAMEZAWA Hiroyuki 已提交
4777 4778
		.register_event = mem_cgroup_usage_register_event,
		.unregister_event = mem_cgroup_usage_unregister_event,
B
Balbir Singh 已提交
4779
	},
4780 4781
	{
		.name = "max_usage_in_bytes",
4782
		.private = MEMFILE_PRIVATE(_MEM, RES_MAX_USAGE),
4783
		.trigger = mem_cgroup_reset,
4784 4785
		.read_u64 = mem_cgroup_read,
	},
B
Balbir Singh 已提交
4786
	{
4787
		.name = "limit_in_bytes",
4788
		.private = MEMFILE_PRIVATE(_MEM, RES_LIMIT),
4789
		.write_string = mem_cgroup_write,
4790
		.read_u64 = mem_cgroup_read,
B
Balbir Singh 已提交
4791
	},
4792 4793 4794 4795 4796 4797
	{
		.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
4798 4799
	{
		.name = "failcnt",
4800
		.private = MEMFILE_PRIVATE(_MEM, RES_FAILCNT),
4801
		.trigger = mem_cgroup_reset,
4802
		.read_u64 = mem_cgroup_read,
B
Balbir Singh 已提交
4803
	},
4804 4805
	{
		.name = "stat",
4806
		.read_map = mem_control_stat_show,
4807
	},
4808 4809 4810 4811
	{
		.name = "force_empty",
		.trigger = mem_cgroup_force_empty_write,
	},
4812 4813 4814 4815 4816
	{
		.name = "use_hierarchy",
		.write_u64 = mem_cgroup_hierarchy_write,
		.read_u64 = mem_cgroup_hierarchy_read,
	},
K
KOSAKI Motohiro 已提交
4817 4818 4819 4820 4821
	{
		.name = "swappiness",
		.read_u64 = mem_cgroup_swappiness_read,
		.write_u64 = mem_cgroup_swappiness_write,
	},
4822 4823 4824 4825 4826
	{
		.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
4827 4828
	{
		.name = "oom_control",
4829 4830
		.read_map = mem_cgroup_oom_control_read,
		.write_u64 = mem_cgroup_oom_control_write,
K
KAMEZAWA Hiroyuki 已提交
4831 4832 4833 4834
		.register_event = mem_cgroup_oom_register_event,
		.unregister_event = mem_cgroup_oom_unregister_event,
		.private = MEMFILE_PRIVATE(_OOM_TYPE, OOM_CONTROL),
	},
4835 4836 4837 4838
#ifdef CONFIG_NUMA
	{
		.name = "numa_stat",
		.open = mem_control_numa_stat_open,
4839
		.mode = S_IRUGO,
4840 4841
	},
#endif
4842 4843 4844 4845 4846
	{
		.name = "vmscan_stat",
		.read_map = mem_cgroup_vmscan_stat_read,
		.trigger = mem_cgroup_reset_vmscan_stat,
	},
B
Balbir Singh 已提交
4847 4848
};

4849 4850 4851 4852 4853 4854
#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
4855 4856
		.register_event = mem_cgroup_usage_register_event,
		.unregister_event = mem_cgroup_usage_unregister_event,
4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880 4881 4882 4883 4884 4885 4886 4887 4888 4889 4890 4891
	},
	{
		.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

4892 4893 4894
static int alloc_mem_cgroup_per_zone_info(struct mem_cgroup *mem, int node)
{
	struct mem_cgroup_per_node *pn;
4895
	struct mem_cgroup_per_zone *mz;
4896
	enum lru_list l;
4897
	int zone, tmp = node;
4898 4899 4900 4901 4902 4903 4904 4905
	/*
	 * 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.
	 */
4906 4907
	if (!node_state(node, N_NORMAL_MEMORY))
		tmp = -1;
4908
	pn = kzalloc_node(sizeof(*pn), GFP_KERNEL, tmp);
4909 4910
	if (!pn)
		return 1;
4911

4912
	mem->info.nodeinfo[node] = pn;
4913 4914
	for (zone = 0; zone < MAX_NR_ZONES; zone++) {
		mz = &pn->zoneinfo[zone];
4915 4916
		for_each_lru(l)
			INIT_LIST_HEAD(&mz->lists[l]);
4917
		mz->usage_in_excess = 0;
4918 4919
		mz->on_tree = false;
		mz->mem = mem;
4920
	}
4921 4922 4923
	return 0;
}

4924 4925 4926 4927 4928
static void free_mem_cgroup_per_zone_info(struct mem_cgroup *mem, int node)
{
	kfree(mem->info.nodeinfo[node]);
}

4929 4930 4931
static struct mem_cgroup *mem_cgroup_alloc(void)
{
	struct mem_cgroup *mem;
4932
	int size = sizeof(struct mem_cgroup);
4933

4934
	/* Can be very big if MAX_NUMNODES is very big */
4935
	if (size < PAGE_SIZE)
4936
		mem = kzalloc(size, GFP_KERNEL);
4937
	else
4938
		mem = vzalloc(size);
4939

4940 4941 4942
	if (!mem)
		return NULL;

4943
	mem->stat = alloc_percpu(struct mem_cgroup_stat_cpu);
4944 4945
	if (!mem->stat)
		goto out_free;
4946
	spin_lock_init(&mem->pcp_counter_lock);
4947
	return mem;
4948 4949 4950 4951 4952 4953 4954

out_free:
	if (size < PAGE_SIZE)
		kfree(mem);
	else
		vfree(mem);
	return NULL;
4955 4956
}

4957 4958 4959 4960 4961 4962 4963 4964 4965 4966 4967
/*
 * 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.
 */

4968
static void __mem_cgroup_free(struct mem_cgroup *mem)
4969
{
K
KAMEZAWA Hiroyuki 已提交
4970 4971
	int node;

4972
	mem_cgroup_remove_from_trees(mem);
K
KAMEZAWA Hiroyuki 已提交
4973 4974
	free_css_id(&mem_cgroup_subsys, &mem->css);

K
KAMEZAWA Hiroyuki 已提交
4975 4976 4977
	for_each_node_state(node, N_POSSIBLE)
		free_mem_cgroup_per_zone_info(mem, node);

4978 4979
	free_percpu(mem->stat);
	if (sizeof(struct mem_cgroup) < PAGE_SIZE)
4980 4981 4982 4983 4984
		kfree(mem);
	else
		vfree(mem);
}

4985 4986 4987 4988 4989
static void mem_cgroup_get(struct mem_cgroup *mem)
{
	atomic_inc(&mem->refcnt);
}

4990
static void __mem_cgroup_put(struct mem_cgroup *mem, int count)
4991
{
4992
	if (atomic_sub_and_test(count, &mem->refcnt)) {
4993
		struct mem_cgroup *parent = parent_mem_cgroup(mem);
4994
		__mem_cgroup_free(mem);
4995 4996 4997
		if (parent)
			mem_cgroup_put(parent);
	}
4998 4999
}

5000 5001 5002 5003 5004
static void mem_cgroup_put(struct mem_cgroup *mem)
{
	__mem_cgroup_put(mem, 1);
}

5005 5006 5007 5008 5009 5010 5011 5012 5013
/*
 * 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);
}
5014

5015 5016 5017
#ifdef CONFIG_CGROUP_MEM_RES_CTLR_SWAP
static void __init enable_swap_cgroup(void)
{
5018
	if (!mem_cgroup_disabled() && really_do_swap_account)
5019 5020 5021 5022 5023 5024 5025 5026
		do_swap_account = 1;
}
#else
static void __init enable_swap_cgroup(void)
{
}
#endif

5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051
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
5052
static struct cgroup_subsys_state * __ref
B
Balbir Singh 已提交
5053 5054
mem_cgroup_create(struct cgroup_subsys *ss, struct cgroup *cont)
{
5055
	struct mem_cgroup *mem, *parent;
K
KAMEZAWA Hiroyuki 已提交
5056
	long error = -ENOMEM;
5057
	int node;
B
Balbir Singh 已提交
5058

5059 5060
	mem = mem_cgroup_alloc();
	if (!mem)
K
KAMEZAWA Hiroyuki 已提交
5061
		return ERR_PTR(error);
5062

5063 5064 5065
	for_each_node_state(node, N_POSSIBLE)
		if (alloc_mem_cgroup_per_zone_info(mem, node))
			goto free_out;
5066

5067
	/* root ? */
5068
	if (cont->parent == NULL) {
5069
		int cpu;
5070
		enable_swap_cgroup();
5071
		parent = NULL;
5072
		root_mem_cgroup = mem;
5073 5074
		if (mem_cgroup_soft_limit_tree_init())
			goto free_out;
5075 5076 5077 5078 5079
		for_each_possible_cpu(cpu) {
			struct memcg_stock_pcp *stock =
						&per_cpu(memcg_stock, cpu);
			INIT_WORK(&stock->work, drain_local_stock);
		}
5080
		hotcpu_notifier(memcg_cpu_hotplug_callback, 0);
5081
	} else {
5082
		parent = mem_cgroup_from_cont(cont->parent);
5083
		mem->use_hierarchy = parent->use_hierarchy;
5084
		mem->oom_kill_disable = parent->oom_kill_disable;
5085
	}
5086

5087 5088 5089
	if (parent && parent->use_hierarchy) {
		res_counter_init(&mem->res, &parent->res);
		res_counter_init(&mem->memsw, &parent->memsw);
5090 5091 5092 5093 5094 5095 5096
		/*
		 * 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);
5097 5098 5099 5100
	} else {
		res_counter_init(&mem->res, NULL);
		res_counter_init(&mem->memsw, NULL);
	}
K
KAMEZAWA Hiroyuki 已提交
5101
	mem->last_scanned_child = 0;
5102
	mem->last_scanned_node = MAX_NUMNODES;
K
KAMEZAWA Hiroyuki 已提交
5103
	INIT_LIST_HEAD(&mem->oom_notify);
5104

K
KOSAKI Motohiro 已提交
5105
	if (parent)
5106
		mem->swappiness = mem_cgroup_swappiness(parent);
5107
	atomic_set(&mem->refcnt, 1);
5108
	mem->move_charge_at_immigrate = 0;
5109
	mutex_init(&mem->thresholds_lock);
5110
	spin_lock_init(&mem->scanstat.lock);
B
Balbir Singh 已提交
5111
	return &mem->css;
5112
free_out:
5113
	__mem_cgroup_free(mem);
5114
	root_mem_cgroup = NULL;
K
KAMEZAWA Hiroyuki 已提交
5115
	return ERR_PTR(error);
B
Balbir Singh 已提交
5116 5117
}

5118
static int mem_cgroup_pre_destroy(struct cgroup_subsys *ss,
5119 5120 5121
					struct cgroup *cont)
{
	struct mem_cgroup *mem = mem_cgroup_from_cont(cont);
5122 5123

	return mem_cgroup_force_empty(mem, false);
5124 5125
}

B
Balbir Singh 已提交
5126 5127 5128
static void mem_cgroup_destroy(struct cgroup_subsys *ss,
				struct cgroup *cont)
{
5129 5130 5131
	struct mem_cgroup *mem = mem_cgroup_from_cont(cont);

	mem_cgroup_put(mem);
B
Balbir Singh 已提交
5132 5133 5134 5135 5136
}

static int mem_cgroup_populate(struct cgroup_subsys *ss,
				struct cgroup *cont)
{
5137 5138 5139 5140 5141 5142 5143 5144
	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
5145 5146
}

5147
#ifdef CONFIG_MMU
5148
/* Handlers for move charge at task migration. */
5149 5150
#define PRECHARGE_COUNT_AT_ONCE	256
static int mem_cgroup_do_precharge(unsigned long count)
5151
{
5152 5153
	int ret = 0;
	int batch_count = PRECHARGE_COUNT_AT_ONCE;
5154 5155
	struct mem_cgroup *mem = mc.to;

5156 5157 5158 5159 5160 5161 5162 5163 5164 5165 5166 5167 5168 5169 5170 5171 5172 5173 5174 5175 5176 5177 5178 5179 5180 5181 5182 5183 5184 5185 5186 5187 5188 5189 5190
	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();
		}
5191
		ret = __mem_cgroup_try_charge(NULL, GFP_KERNEL, 1, &mem, false);
5192 5193 5194 5195 5196
		if (ret || !mem)
			/* mem_cgroup_clear_mc() will do uncharge later */
			return -ENOMEM;
		mc.precharge++;
	}
5197 5198 5199 5200 5201 5202 5203 5204
	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
5205
 * @target: the pointer the target page or swap ent will be stored(can be NULL)
5206 5207 5208 5209 5210 5211
 *
 * 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).
5212 5213 5214
 *   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.
5215 5216 5217 5218 5219
 *
 * Called with pte lock held.
 */
union mc_target {
	struct page	*page;
5220
	swp_entry_t	ent;
5221 5222 5223 5224 5225
};

enum mc_target_type {
	MC_TARGET_NONE,	/* not used */
	MC_TARGET_PAGE,
5226
	MC_TARGET_SWAP,
5227 5228
};

D
Daisuke Nishimura 已提交
5229 5230
static struct page *mc_handle_present_pte(struct vm_area_struct *vma,
						unsigned long addr, pte_t ptent)
5231
{
D
Daisuke Nishimura 已提交
5232
	struct page *page = vm_normal_page(vma, addr, ptent);
5233

D
Daisuke Nishimura 已提交
5234 5235 5236 5237 5238 5239
	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;
5240 5241
	} else if (!move_file())
		/* we ignore mapcount for file pages */
D
Daisuke Nishimura 已提交
5242 5243 5244 5245 5246 5247 5248 5249 5250 5251 5252 5253 5254 5255 5256 5257 5258 5259
		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 */
5260 5261
		if (page)
			put_page(page);
D
Daisuke Nishimura 已提交
5262
		return NULL;
5263
	}
D
Daisuke Nishimura 已提交
5264 5265 5266 5267 5268 5269
	if (do_swap_account)
		entry->val = ent.val;

	return page;
}

5270 5271 5272 5273 5274 5275 5276 5277 5278 5279 5280 5281 5282 5283 5284 5285 5286 5287 5288 5289 5290
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). */
5291 5292 5293 5294 5295 5296
	page = find_get_page(mapping, pgoff);

#ifdef CONFIG_SWAP
	/* shmem/tmpfs may report page out on swap: account for that too. */
	if (radix_tree_exceptional_entry(page)) {
		swp_entry_t swap = radix_to_swp_entry(page);
5297
		if (do_swap_account)
5298 5299
			*entry = swap;
		page = find_get_page(&swapper_space, swap.val);
5300
	}
5301
#endif
5302 5303 5304
	return page;
}

D
Daisuke Nishimura 已提交
5305 5306 5307 5308 5309 5310 5311 5312 5313 5314 5315 5316
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);
5317 5318
	else if (pte_none(ptent) || pte_file(ptent))
		page = mc_handle_file_pte(vma, addr, ptent, &ent);
D
Daisuke Nishimura 已提交
5319 5320 5321

	if (!page && !ent.val)
		return 0;
5322 5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333 5334 5335 5336
	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
5337 5338
	/* There is a swap entry and a page doesn't exist or isn't charged */
	if (ent.val && !ret &&
5339 5340 5341 5342
			css_id(&mc.from->css) == lookup_swap_cgroup(ent)) {
		ret = MC_TARGET_SWAP;
		if (target)
			target->ent = ent;
5343 5344 5345 5346 5347 5348 5349 5350 5351 5352 5353 5354
	}
	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;

5355 5356
	split_huge_page_pmd(walk->mm, pmd);

5357 5358 5359 5360 5361 5362 5363
	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();

5364 5365 5366
	return 0;
}

5367 5368 5369 5370 5371
static unsigned long mem_cgroup_count_precharge(struct mm_struct *mm)
{
	unsigned long precharge;
	struct vm_area_struct *vma;

5372
	down_read(&mm->mmap_sem);
5373 5374 5375 5376 5377 5378 5379 5380 5381 5382 5383
	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);
	}
5384
	up_read(&mm->mmap_sem);
5385 5386 5387 5388 5389 5390 5391 5392 5393

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

	return precharge;
}

static int mem_cgroup_precharge_mc(struct mm_struct *mm)
{
5394 5395 5396 5397 5398
	unsigned long precharge = mem_cgroup_count_precharge(mm);

	VM_BUG_ON(mc.moving_task);
	mc.moving_task = current;
	return mem_cgroup_do_precharge(precharge);
5399 5400
}

5401 5402
/* cancels all extra charges on mc.from and mc.to, and wakes up all waiters. */
static void __mem_cgroup_clear_mc(void)
5403
{
5404 5405 5406
	struct mem_cgroup *from = mc.from;
	struct mem_cgroup *to = mc.to;

5407
	/* we must uncharge all the leftover precharges from mc.to */
5408 5409 5410 5411 5412 5413 5414 5415 5416 5417 5418
	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;
5419
	}
5420 5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432 5433 5434 5435 5436 5437 5438
	/* 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;
	}
5439 5440 5441 5442 5443 5444 5445 5446 5447 5448 5449 5450 5451 5452 5453
	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();
5454
	spin_lock(&mc.lock);
5455 5456
	mc.from = NULL;
	mc.to = NULL;
5457
	spin_unlock(&mc.lock);
5458
	mem_cgroup_end_move(from);
5459 5460
}

5461 5462
static int mem_cgroup_can_attach(struct cgroup_subsys *ss,
				struct cgroup *cgroup,
5463
				struct task_struct *p)
5464 5465 5466 5467 5468 5469 5470 5471 5472 5473 5474 5475 5476 5477
{
	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 */
5478 5479 5480 5481
		if (mm->owner == p) {
			VM_BUG_ON(mc.from);
			VM_BUG_ON(mc.to);
			VM_BUG_ON(mc.precharge);
5482
			VM_BUG_ON(mc.moved_charge);
5483
			VM_BUG_ON(mc.moved_swap);
5484
			mem_cgroup_start_move(from);
5485
			spin_lock(&mc.lock);
5486 5487
			mc.from = from;
			mc.to = mem;
5488
			spin_unlock(&mc.lock);
5489
			/* We set mc.moving_task later */
5490 5491 5492 5493

			ret = mem_cgroup_precharge_mc(mm);
			if (ret)
				mem_cgroup_clear_mc();
5494 5495
		}
		mmput(mm);
5496 5497 5498 5499 5500 5501
	}
	return ret;
}

static void mem_cgroup_cancel_attach(struct cgroup_subsys *ss,
				struct cgroup *cgroup,
5502
				struct task_struct *p)
5503
{
5504
	mem_cgroup_clear_mc();
5505 5506
}

5507 5508 5509
static int mem_cgroup_move_charge_pte_range(pmd_t *pmd,
				unsigned long addr, unsigned long end,
				struct mm_walk *walk)
5510
{
5511 5512 5513 5514 5515
	int ret = 0;
	struct vm_area_struct *vma = walk->private;
	pte_t *pte;
	spinlock_t *ptl;

5516
	split_huge_page_pmd(walk->mm, pmd);
5517 5518 5519 5520 5521 5522 5523 5524
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;
5525
		swp_entry_t ent;
5526 5527 5528 5529 5530 5531 5532 5533 5534 5535 5536

		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);
5537 5538
			if (!mem_cgroup_move_account(page, 1, pc,
						     mc.from, mc.to, false)) {
5539
				mc.precharge--;
5540 5541
				/* we uncharge from mc.from later. */
				mc.moved_charge++;
5542 5543 5544 5545 5546
			}
			putback_lru_page(page);
put:			/* is_target_pte_for_mc() gets the page */
			put_page(page);
			break;
5547 5548
		case MC_TARGET_SWAP:
			ent = target.ent;
5549 5550
			if (!mem_cgroup_move_swap_account(ent,
						mc.from, mc.to, false)) {
5551
				mc.precharge--;
5552 5553 5554
				/* we fixup refcnts and charges later. */
				mc.moved_swap++;
			}
5555
			break;
5556 5557 5558 5559 5560 5561 5562 5563 5564 5565 5566 5567 5568 5569
		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.
		 */
5570
		ret = mem_cgroup_do_precharge(1);
5571 5572 5573 5574 5575 5576 5577 5578 5579 5580 5581 5582
		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();
5583 5584 5585 5586 5587 5588 5589 5590 5591 5592 5593 5594 5595
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;
	}
5596 5597 5598 5599 5600 5601 5602 5603 5604 5605 5606 5607 5608 5609 5610 5611 5612 5613
	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;
	}
5614
	up_read(&mm->mmap_sem);
5615 5616
}

B
Balbir Singh 已提交
5617 5618 5619
static void mem_cgroup_move_task(struct cgroup_subsys *ss,
				struct cgroup *cont,
				struct cgroup *old_cont,
5620
				struct task_struct *p)
B
Balbir Singh 已提交
5621
{
5622
	struct mm_struct *mm = get_task_mm(p);
5623 5624

	if (mm) {
5625 5626 5627
		if (mc.to)
			mem_cgroup_move_charge(mm);
		put_swap_token(mm);
5628 5629
		mmput(mm);
	}
5630 5631
	if (mc.to)
		mem_cgroup_clear_mc();
B
Balbir Singh 已提交
5632
}
5633 5634 5635
#else	/* !CONFIG_MMU */
static int mem_cgroup_can_attach(struct cgroup_subsys *ss,
				struct cgroup *cgroup,
5636
				struct task_struct *p)
5637 5638 5639 5640 5641
{
	return 0;
}
static void mem_cgroup_cancel_attach(struct cgroup_subsys *ss,
				struct cgroup *cgroup,
5642
				struct task_struct *p)
5643 5644 5645 5646 5647
{
}
static void mem_cgroup_move_task(struct cgroup_subsys *ss,
				struct cgroup *cont,
				struct cgroup *old_cont,
5648
				struct task_struct *p)
5649 5650 5651
{
}
#endif
B
Balbir Singh 已提交
5652

B
Balbir Singh 已提交
5653 5654 5655 5656
struct cgroup_subsys mem_cgroup_subsys = {
	.name = "memory",
	.subsys_id = mem_cgroup_subsys_id,
	.create = mem_cgroup_create,
5657
	.pre_destroy = mem_cgroup_pre_destroy,
B
Balbir Singh 已提交
5658 5659
	.destroy = mem_cgroup_destroy,
	.populate = mem_cgroup_populate,
5660 5661
	.can_attach = mem_cgroup_can_attach,
	.cancel_attach = mem_cgroup_cancel_attach,
B
Balbir Singh 已提交
5662
	.attach = mem_cgroup_move_task,
5663
	.early_init = 0,
K
KAMEZAWA Hiroyuki 已提交
5664
	.use_id = 1,
B
Balbir Singh 已提交
5665
};
5666 5667

#ifdef CONFIG_CGROUP_MEM_RES_CTLR_SWAP
5668 5669 5670
static int __init enable_swap_account(char *s)
{
	/* consider enabled if no parameter or 1 is given */
5671
	if (!strcmp(s, "1"))
5672
		really_do_swap_account = 1;
5673
	else if (!strcmp(s, "0"))
5674 5675 5676
		really_do_swap_account = 0;
	return 1;
}
5677
__setup("swapaccount=", enable_swap_account);
5678 5679

#endif