memcontrol.c 140.8 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,
};
101 102 103 104 105 106 107 108 109 110 111 112 113
/*
 * Per memcg event counter is incremented at every pagein/pageout. With THP,
 * it will be incremated by the number of pages. This counter is used for
 * for trigger some periodic events. This is straightforward and better
 * than using jiffies etc. to handle periodic memcg event.
 */
enum mem_cgroup_events_target {
	MEM_CGROUP_TARGET_THRESH,
	MEM_CGROUP_TARGET_SOFTLIMIT,
	MEM_CGROUP_NTARGETS,
};
#define THRESHOLDS_EVENTS_TARGET (128)
#define SOFTLIMIT_EVENTS_TARGET (1024)
114

115
struct mem_cgroup_stat_cpu {
116
	long count[MEM_CGROUP_STAT_NSTATS];
117
	unsigned long events[MEM_CGROUP_EVENTS_NSTATS];
118
	unsigned long targets[MEM_CGROUP_NTARGETS];
119 120
};

121 122 123 124
/*
 * per-zone information in memory controller.
 */
struct mem_cgroup_per_zone {
125 126 127
	/*
	 * spin_lock to protect the per cgroup LRU
	 */
128 129
	struct list_head	lists[NR_LRU_LISTS];
	unsigned long		count[NR_LRU_LISTS];
K
KOSAKI Motohiro 已提交
130 131

	struct zone_reclaim_stat reclaim_stat;
132 133 134 135
	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;
136 137
	struct mem_cgroup	*mem;		/* Back pointer, we cannot */
						/* use container_of	   */
138 139 140 141 142 143 144 145 146 147 148 149
};
/* 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];
};

150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169
/*
 * 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;

170 171 172 173 174
struct mem_cgroup_threshold {
	struct eventfd_ctx *eventfd;
	u64 threshold;
};

K
KAMEZAWA Hiroyuki 已提交
175
/* For threshold */
176 177
struct mem_cgroup_threshold_ary {
	/* An array index points to threshold just below usage. */
178
	int current_threshold;
179 180 181 182 183
	/* Size of entries[] */
	unsigned int size;
	/* Array of thresholds */
	struct mem_cgroup_threshold entries[0];
};
184 185 186 187 188 189 190 191 192 193 194 195

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
196 197 198 199 200
/* for OOM */
struct mem_cgroup_eventfd_list {
	struct list_head list;
	struct eventfd_ctx *eventfd;
};
201 202

static void mem_cgroup_threshold(struct mem_cgroup *mem);
K
KAMEZAWA Hiroyuki 已提交
203
static void mem_cgroup_oom_notify(struct mem_cgroup *mem);
204

B
Balbir Singh 已提交
205 206 207 208 209 210 211
/*
 * 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
212 213 214
 * 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
215 216 217 218 219 220 221
 */
struct mem_cgroup {
	struct cgroup_subsys_state css;
	/*
	 * the counter to account for memory usage
	 */
	struct res_counter res;
222 223 224 225
	/*
	 * the counter to account for mem+swap usage.
	 */
	struct res_counter memsw;
226 227 228 229
	/*
	 * Per cgroup active and inactive list, similar to the
	 * per zone LRU lists.
	 */
230
	struct mem_cgroup_lru_info info;
231
	/*
232
	 * While reclaiming in a hierarchy, we cache the last child we
K
KAMEZAWA Hiroyuki 已提交
233
	 * reclaimed from.
234
	 */
K
KAMEZAWA Hiroyuki 已提交
235
	int last_scanned_child;
236 237 238 239 240
	int last_scanned_node;
#if MAX_NUMNODES > 1
	nodemask_t	scan_nodes;
	unsigned long   next_scan_node_update;
#endif
241 242 243 244
	/*
	 * Should the accounting and control be hierarchical, per subtree?
	 */
	bool use_hierarchy;
K
KAMEZAWA Hiroyuki 已提交
245
	atomic_t	oom_lock;
246
	atomic_t	refcnt;
247

K
KOSAKI Motohiro 已提交
248
	unsigned int	swappiness;
249 250
	/* OOM-Killer disable */
	int		oom_kill_disable;
K
KOSAKI Motohiro 已提交
251

252 253 254
	/* set when res.limit == memsw.limit */
	bool		memsw_is_minimum;

255 256 257 258
	/* protect arrays of thresholds */
	struct mutex thresholds_lock;

	/* thresholds for memory usage. RCU-protected */
259
	struct mem_cgroup_thresholds thresholds;
260

261
	/* thresholds for mem+swap usage. RCU-protected */
262
	struct mem_cgroup_thresholds memsw_thresholds;
263

K
KAMEZAWA Hiroyuki 已提交
264 265 266
	/* For oom notifier event fd */
	struct list_head oom_notify;

267 268 269 270 271
	/*
	 * 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;
272
	/*
273
	 * percpu counter.
274
	 */
275
	struct mem_cgroup_stat_cpu *stat;
276 277 278 279 280 281
	/*
	 * 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
282 283
};

284 285 286 287 288 289
/* 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 {
290
	MOVE_CHARGE_TYPE_ANON,	/* private anonymous page and swap of it */
291
	MOVE_CHARGE_TYPE_FILE,	/* file page(including tmpfs) and swap of it */
292 293 294
	NR_MOVE_TYPE,
};

295 296
/* "mc" and its members are protected by cgroup_mutex */
static struct move_charge_struct {
297
	spinlock_t	  lock; /* for from, to */
298 299 300
	struct mem_cgroup *from;
	struct mem_cgroup *to;
	unsigned long precharge;
301
	unsigned long moved_charge;
302
	unsigned long moved_swap;
303 304 305
	struct task_struct *moving_task;	/* a task moving charges */
	wait_queue_head_t waitq;		/* a waitq for other context */
} mc = {
306
	.lock = __SPIN_LOCK_UNLOCKED(mc.lock),
307 308
	.waitq = __WAIT_QUEUE_HEAD_INITIALIZER(mc.waitq),
};
309

D
Daisuke Nishimura 已提交
310 311 312 313 314 315
static bool move_anon(void)
{
	return test_bit(MOVE_CHARGE_TYPE_ANON,
					&mc.to->move_charge_at_immigrate);
}

316 317 318 319 320 321
static bool move_file(void)
{
	return test_bit(MOVE_CHARGE_TYPE_FILE,
					&mc.to->move_charge_at_immigrate);
}

322 323 324 325 326 327 328
/*
 * 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)

329 330 331
enum charge_type {
	MEM_CGROUP_CHARGE_TYPE_CACHE = 0,
	MEM_CGROUP_CHARGE_TYPE_MAPPED,
332
	MEM_CGROUP_CHARGE_TYPE_SHMEM,	/* used by page migration of shmem */
333
	MEM_CGROUP_CHARGE_TYPE_FORCE,	/* used by force_empty */
K
KAMEZAWA Hiroyuki 已提交
334
	MEM_CGROUP_CHARGE_TYPE_SWAPOUT,	/* for accounting swapcache */
K
KAMEZAWA Hiroyuki 已提交
335
	MEM_CGROUP_CHARGE_TYPE_DROP,	/* a page was unused swap cache */
336 337 338
	NR_CHARGE_TYPE,
};

339 340 341
/* for encoding cft->private value on file */
#define _MEM			(0)
#define _MEMSWAP		(1)
K
KAMEZAWA Hiroyuki 已提交
342
#define _OOM_TYPE		(2)
343 344 345
#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
346 347
/* Used for OOM nofiier */
#define OOM_CONTROL		(0)
348

349 350 351 352 353 354 355
/*
 * 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)
356 357
#define MEM_CGROUP_RECLAIM_SOFT_BIT	0x2
#define MEM_CGROUP_RECLAIM_SOFT		(1 << MEM_CGROUP_RECLAIM_SOFT_BIT)
358

359 360
static void mem_cgroup_get(struct mem_cgroup *mem);
static void mem_cgroup_put(struct mem_cgroup *mem);
361
static struct mem_cgroup *parent_mem_cgroup(struct mem_cgroup *mem);
362
static void drain_all_stock_async(void);
363

364 365 366 367 368 369
static struct mem_cgroup_per_zone *
mem_cgroup_zoneinfo(struct mem_cgroup *mem, int nid, int zid)
{
	return &mem->info.nodeinfo[nid]->zoneinfo[zid];
}

370 371 372 373 374
struct cgroup_subsys_state *mem_cgroup_css(struct mem_cgroup *mem)
{
	return &mem->css;
}

375
static struct mem_cgroup_per_zone *
376
page_cgroup_zoneinfo(struct mem_cgroup *mem, struct page *page)
377
{
378 379
	int nid = page_to_nid(page);
	int zid = page_zonenum(page);
380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399

	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
400
__mem_cgroup_insert_exceeded(struct mem_cgroup *mem,
401
				struct mem_cgroup_per_zone *mz,
402 403
				struct mem_cgroup_tree_per_zone *mctz,
				unsigned long long new_usage_in_excess)
404 405 406 407 408 409 410 411
{
	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;

412 413 414
	mz->usage_in_excess = new_usage_in_excess;
	if (!mz->usage_in_excess)
		return;
415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430
	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;
431 432 433 434 435 436 437 438 439 440 441 442 443
}

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;
}

444 445 446 447 448 449
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);
450
	__mem_cgroup_remove_exceeded(mem, mz, mctz);
451 452 453 454 455 456
	spin_unlock(&mctz->lock);
}


static void mem_cgroup_update_tree(struct mem_cgroup *mem, struct page *page)
{
457
	unsigned long long excess;
458 459
	struct mem_cgroup_per_zone *mz;
	struct mem_cgroup_tree_per_zone *mctz;
460 461
	int nid = page_to_nid(page);
	int zid = page_zonenum(page);
462 463 464
	mctz = soft_limit_tree_from_page(page);

	/*
465 466
	 * Necessary to update all ancestors when hierarchy is used.
	 * because their event counter is not touched.
467
	 */
468 469
	for (; mem; mem = parent_mem_cgroup(mem)) {
		mz = mem_cgroup_zoneinfo(mem, nid, zid);
470
		excess = res_counter_soft_limit_excess(&mem->res);
471 472 473 474
		/*
		 * We have to update the tree if mz is on RB-tree or
		 * mem is over its softlimit.
		 */
475
		if (excess || mz->on_tree) {
476 477 478 479 480
			spin_lock(&mctz->lock);
			/* if on-tree, remove it */
			if (mz->on_tree)
				__mem_cgroup_remove_exceeded(mem, mz, mctz);
			/*
481 482
			 * Insert again. mz->usage_in_excess will be updated.
			 * If excess is 0, no tree ops.
483
			 */
484
			__mem_cgroup_insert_exceeded(mem, mz, mctz, excess);
485 486
			spin_unlock(&mctz->lock);
		}
487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504
	}
}

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);
		}
	}
}

505 506 507 508
static struct mem_cgroup_per_zone *
__mem_cgroup_largest_soft_limit_node(struct mem_cgroup_tree_per_zone *mctz)
{
	struct rb_node *rightmost = NULL;
509
	struct mem_cgroup_per_zone *mz;
510 511

retry:
512
	mz = NULL;
513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541
	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;
}

542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560
/*
 * 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.
 */
561 562
static long mem_cgroup_read_stat(struct mem_cgroup *mem,
				 enum mem_cgroup_stat_index idx)
563
{
564
	long val = 0;
565 566
	int cpu;

567 568
	get_online_cpus();
	for_each_online_cpu(cpu)
569
		val += per_cpu(mem->stat->count[idx], cpu);
570 571 572 573 574 575
#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();
576 577 578
	return val;
}

579
static long mem_cgroup_local_usage(struct mem_cgroup *mem)
580
{
581
	long ret;
582 583 584 585 586 587

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

588 589 590 591
static void mem_cgroup_swap_statistics(struct mem_cgroup *mem,
					 bool charge)
{
	int val = (charge) ? 1 : -1;
592
	this_cpu_add(mem->stat->count[MEM_CGROUP_STAT_SWAPOUT], val);
593 594
}

595 596 597 598 599 600 601 602 603 604
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);
}

605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620
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;
}

621
static void mem_cgroup_charge_statistics(struct mem_cgroup *mem,
622
					 bool file, int nr_pages)
623
{
624 625
	preempt_disable();

626 627
	if (file)
		__this_cpu_add(mem->stat->count[MEM_CGROUP_STAT_CACHE], nr_pages);
628
	else
629
		__this_cpu_add(mem->stat->count[MEM_CGROUP_STAT_RSS], nr_pages);
630

631 632
	/* pagein of a big page is an event. So, ignore page size */
	if (nr_pages > 0)
633
		__this_cpu_inc(mem->stat->events[MEM_CGROUP_EVENTS_PGPGIN]);
634
	else {
635
		__this_cpu_inc(mem->stat->events[MEM_CGROUP_EVENTS_PGPGOUT]);
636 637
		nr_pages = -nr_pages; /* for event */
	}
638

639
	__this_cpu_add(mem->stat->events[MEM_CGROUP_EVENTS_COUNT], nr_pages);
640

641
	preempt_enable();
642 643
}

644 645 646 647 648 649 650 651 652 653 654 655 656
static unsigned long
mem_cgroup_get_zonestat_node(struct mem_cgroup *mem, int nid, enum lru_list idx)
{
	struct mem_cgroup_per_zone *mz;
	u64 total = 0;
	int zid;

	for (zid = 0; zid < MAX_NR_ZONES; zid++) {
		mz = mem_cgroup_zoneinfo(mem, nid, zid);
		total += MEM_CGROUP_ZSTAT(mz, idx);
	}
	return total;
}
K
KAMEZAWA Hiroyuki 已提交
657
static unsigned long mem_cgroup_get_local_zonestat(struct mem_cgroup *mem,
658
					enum lru_list idx)
659
{
660
	int nid;
661 662 663
	u64 total = 0;

	for_each_online_node(nid)
664
		total += mem_cgroup_get_zonestat_node(mem, nid, idx);
665
	return total;
666 667
}

668 669 670 671 672 673 674 675 676 677 678
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)
679
{
680
	unsigned long val, next;
681

682
	val = this_cpu_read(mem->stat->events[MEM_CGROUP_EVENTS_COUNT]);
683

684 685 686 687 688 689 690 691 692 693 694 695
	switch (target) {
	case MEM_CGROUP_TARGET_THRESH:
		next = val + THRESHOLDS_EVENTS_TARGET;
		break;
	case MEM_CGROUP_TARGET_SOFTLIMIT:
		next = val + SOFTLIMIT_EVENTS_TARGET;
		break;
	default:
		return;
	}

	this_cpu_write(mem->stat->targets[target], next);
696 697 698 699 700 701 702 703 704
}

/*
 * 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 */
705
	if (unlikely(__memcg_event_check(mem, MEM_CGROUP_TARGET_THRESH))) {
706
		mem_cgroup_threshold(mem);
707 708 709
		__mem_cgroup_target_update(mem, MEM_CGROUP_TARGET_THRESH);
		if (unlikely(__memcg_event_check(mem,
			MEM_CGROUP_TARGET_SOFTLIMIT))){
710
			mem_cgroup_update_tree(mem, page);
711 712 713
			__mem_cgroup_target_update(mem,
				MEM_CGROUP_TARGET_SOFTLIMIT);
		}
714 715 716
	}
}

717
static struct mem_cgroup *mem_cgroup_from_cont(struct cgroup *cont)
B
Balbir Singh 已提交
718 719 720 721 722 723
{
	return container_of(cgroup_subsys_state(cont,
				mem_cgroup_subsys_id), struct mem_cgroup,
				css);
}

724
struct mem_cgroup *mem_cgroup_from_task(struct task_struct *p)
725
{
726 727 728 729 730 731 732 733
	/*
	 * 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;

734 735 736 737
	return container_of(task_subsys_state(p, mem_cgroup_subsys_id),
				struct mem_cgroup, css);
}

738
struct mem_cgroup *try_get_mem_cgroup_from_mm(struct mm_struct *mm)
739 740
{
	struct mem_cgroup *mem = NULL;
741 742 743

	if (!mm)
		return NULL;
744 745 746 747 748 749 750 751 752 753 754 755 756 757 758
	/*
	 * 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
759 760
/* 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
761
{
762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783
	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
784 785 786 787 788 789 790 791 792
}

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
793 794
	struct cgroup_subsys_state *css;

K
KAMEZAWA Hiroyuki 已提交
795
	hierarchy_used = iter->use_hierarchy;
K
KAMEZAWA Hiroyuki 已提交
796

K
KAMEZAWA Hiroyuki 已提交
797
	css_put(&iter->css);
798 799
	/* If no ROOT, walk all, ignore hierarchy */
	if (!cond || (root && !hierarchy_used))
K
KAMEZAWA Hiroyuki 已提交
800
		return NULL;
K
KAMEZAWA Hiroyuki 已提交
801

802 803 804
	if (!root)
		root = root_mem_cgroup;

K
KAMEZAWA Hiroyuki 已提交
805 806
	do {
		iter = NULL;
K
KAMEZAWA Hiroyuki 已提交
807
		rcu_read_lock();
K
KAMEZAWA Hiroyuki 已提交
808 809 810

		css = css_get_next(&mem_cgroup_subsys, nextid,
				&root->css, &found);
K
KAMEZAWA Hiroyuki 已提交
811
		if (css && css_tryget(css))
K
KAMEZAWA Hiroyuki 已提交
812
			iter = container_of(css, struct mem_cgroup, css);
K
KAMEZAWA Hiroyuki 已提交
813
		rcu_read_unlock();
K
KAMEZAWA Hiroyuki 已提交
814
		/* If css is NULL, no more cgroups will be found */
K
KAMEZAWA Hiroyuki 已提交
815
		nextid = found + 1;
K
KAMEZAWA Hiroyuki 已提交
816
	} while (css && !iter);
K
KAMEZAWA Hiroyuki 已提交
817

K
KAMEZAWA Hiroyuki 已提交
818
	return iter;
K
KAMEZAWA Hiroyuki 已提交
819
}
K
KAMEZAWA Hiroyuki 已提交
820 821 822 823 824 825 826 827 828 829 830 831 832
/*
 * 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)

833 834 835
#define for_each_mem_cgroup_all(iter) \
	for_each_mem_cgroup_tree_cond(iter, NULL, true)

K
KAMEZAWA Hiroyuki 已提交
836

837 838 839 840 841
static inline bool mem_cgroup_is_root(struct mem_cgroup *mem)
{
	return (mem == root_mem_cgroup);
}

842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868
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
869 870 871 872 873 874 875 876 877 878 879 880 881
/*
 * 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.
 */
882

K
KAMEZAWA Hiroyuki 已提交
883 884 885 886
void mem_cgroup_del_lru_list(struct page *page, enum lru_list lru)
{
	struct page_cgroup *pc;
	struct mem_cgroup_per_zone *mz;
887

888
	if (mem_cgroup_disabled())
K
KAMEZAWA Hiroyuki 已提交
889 890 891
		return;
	pc = lookup_page_cgroup(page);
	/* can happen while we handle swapcache. */
892
	if (!TestClearPageCgroupAcctLRU(pc))
K
KAMEZAWA Hiroyuki 已提交
893
		return;
894
	VM_BUG_ON(!pc->mem_cgroup);
895 896 897 898
	/*
	 * We don't check PCG_USED bit. It's cleared when the "page" is finally
	 * removed from global LRU.
	 */
899
	mz = page_cgroup_zoneinfo(pc->mem_cgroup, page);
900 901
	/* huge page split is done under lru_lock. so, we have no races. */
	MEM_CGROUP_ZSTAT(mz, lru) -= 1 << compound_order(page);
902 903 904
	if (mem_cgroup_is_root(pc->mem_cgroup))
		return;
	VM_BUG_ON(list_empty(&pc->lru));
K
KAMEZAWA Hiroyuki 已提交
905
	list_del_init(&pc->lru);
906 907
}

K
KAMEZAWA Hiroyuki 已提交
908
void mem_cgroup_del_lru(struct page *page)
909
{
K
KAMEZAWA Hiroyuki 已提交
910 911
	mem_cgroup_del_lru_list(page, page_lru(page));
}
912

913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934
/*
 * 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;
935
	mz = page_cgroup_zoneinfo(pc->mem_cgroup, page);
936 937 938
	list_move_tail(&pc->lru, &mz->lists[lru]);
}

K
KAMEZAWA Hiroyuki 已提交
939 940 941 942
void mem_cgroup_rotate_lru_list(struct page *page, enum lru_list lru)
{
	struct mem_cgroup_per_zone *mz;
	struct page_cgroup *pc;
943

944
	if (mem_cgroup_disabled())
K
KAMEZAWA Hiroyuki 已提交
945
		return;
946

K
KAMEZAWA Hiroyuki 已提交
947
	pc = lookup_page_cgroup(page);
948
	/* unused or root page is not rotated. */
949 950 951 952 953
	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
954
		return;
955
	mz = page_cgroup_zoneinfo(pc->mem_cgroup, page);
K
KAMEZAWA Hiroyuki 已提交
956
	list_move(&pc->lru, &mz->lists[lru]);
957 958
}

K
KAMEZAWA Hiroyuki 已提交
959
void mem_cgroup_add_lru_list(struct page *page, enum lru_list lru)
960
{
K
KAMEZAWA Hiroyuki 已提交
961 962
	struct page_cgroup *pc;
	struct mem_cgroup_per_zone *mz;
963

964
	if (mem_cgroup_disabled())
K
KAMEZAWA Hiroyuki 已提交
965 966
		return;
	pc = lookup_page_cgroup(page);
967
	VM_BUG_ON(PageCgroupAcctLRU(pc));
K
KAMEZAWA Hiroyuki 已提交
968
	if (!PageCgroupUsed(pc))
L
Lee Schermerhorn 已提交
969
		return;
970 971
	/* Ensure pc->mem_cgroup is visible after reading PCG_USED. */
	smp_rmb();
972
	mz = page_cgroup_zoneinfo(pc->mem_cgroup, page);
973 974
	/* huge page split is done under lru_lock. so, we have no races. */
	MEM_CGROUP_ZSTAT(mz, lru) += 1 << compound_order(page);
975 976 977
	SetPageCgroupAcctLRU(pc);
	if (mem_cgroup_is_root(pc->mem_cgroup))
		return;
K
KAMEZAWA Hiroyuki 已提交
978 979
	list_add(&pc->lru, &mz->lists[lru]);
}
980

K
KAMEZAWA Hiroyuki 已提交
981
/*
982 983 984 985
 * 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
986
 */
987
static void mem_cgroup_lru_del_before_commit(struct page *page)
K
KAMEZAWA Hiroyuki 已提交
988
{
989 990 991 992
	unsigned long flags;
	struct zone *zone = page_zone(page);
	struct page_cgroup *pc = lookup_page_cgroup(page);

993 994 995 996 997 998 999 1000 1001 1002 1003
	/*
	 * 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;

1004 1005 1006 1007 1008 1009 1010 1011
	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
1012 1013
}

1014
static void mem_cgroup_lru_add_after_commit(struct page *page)
1015 1016 1017 1018 1019
{
	unsigned long flags;
	struct zone *zone = page_zone(page);
	struct page_cgroup *pc = lookup_page_cgroup(page);

1020 1021 1022
	/* taking care of that the page is added to LRU while we commit it */
	if (likely(!PageLRU(page)))
		return;
1023 1024
	spin_lock_irqsave(&zone->lru_lock, flags);
	/* link when the page is linked to LRU but page_cgroup isn't */
1025
	if (PageLRU(page) && !PageCgroupAcctLRU(pc))
1026 1027 1028 1029 1030
		mem_cgroup_add_lru_list(page, page_lru(page));
	spin_unlock_irqrestore(&zone->lru_lock, flags);
}


K
KAMEZAWA Hiroyuki 已提交
1031 1032 1033
void mem_cgroup_move_lists(struct page *page,
			   enum lru_list from, enum lru_list to)
{
1034
	if (mem_cgroup_disabled())
K
KAMEZAWA Hiroyuki 已提交
1035 1036 1037
		return;
	mem_cgroup_del_lru_list(page, from);
	mem_cgroup_add_lru_list(page, to);
1038 1039
}

1040 1041 1042
int task_in_mem_cgroup(struct task_struct *task, const struct mem_cgroup *mem)
{
	int ret;
1043
	struct mem_cgroup *curr = NULL;
1044
	struct task_struct *p;
1045

1046 1047 1048 1049 1050
	p = find_lock_task_mm(task);
	if (!p)
		return 0;
	curr = try_get_mem_cgroup_from_mm(p->mm);
	task_unlock(p);
1051 1052
	if (!curr)
		return 0;
1053 1054 1055 1056 1057 1058 1059
	/*
	 * We should check use_hierarchy of "mem" not "curr". Because checking
	 * use_hierarchy of "curr" here make this function true if hierarchy is
	 * enabled in "curr" and "curr" is a child of "mem" in *cgroup*
	 * hierarchy(even if use_hierarchy is disabled in "mem").
	 */
	if (mem->use_hierarchy)
1060 1061 1062 1063
		ret = css_is_ancestor(&curr->css, &mem->css);
	else
		ret = (curr == mem);
	css_put(&curr->css);
1064 1065 1066
	return ret;
}

1067
static int calc_inactive_ratio(struct mem_cgroup *memcg, unsigned long *present_pages)
1068 1069 1070
{
	unsigned long active;
	unsigned long inactive;
1071 1072
	unsigned long gb;
	unsigned long inactive_ratio;
1073

K
KAMEZAWA Hiroyuki 已提交
1074 1075
	inactive = mem_cgroup_get_local_zonestat(memcg, LRU_INACTIVE_ANON);
	active = mem_cgroup_get_local_zonestat(memcg, LRU_ACTIVE_ANON);
1076

1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103
	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)
1104 1105 1106 1107 1108
		return 1;

	return 0;
}

1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119
int mem_cgroup_inactive_file_is_low(struct mem_cgroup *memcg)
{
	unsigned long active;
	unsigned long inactive;

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

	return (active > inactive);
}

1120 1121 1122
unsigned long mem_cgroup_zone_nr_lru_pages(struct mem_cgroup *memcg,
						struct zone *zone,
						enum lru_list lru)
1123
{
1124
	int nid = zone_to_nid(zone);
1125 1126 1127 1128 1129 1130
	int zid = zone_idx(zone);
	struct mem_cgroup_per_zone *mz = mem_cgroup_zoneinfo(memcg, nid, zid);

	return MEM_CGROUP_ZSTAT(mz, lru);
}

1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 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 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217
#ifdef CONFIG_NUMA
static unsigned long mem_cgroup_node_nr_file_lru_pages(struct mem_cgroup *memcg,
							int nid)
{
	unsigned long ret;

	ret = mem_cgroup_get_zonestat_node(memcg, nid, LRU_INACTIVE_FILE) +
		mem_cgroup_get_zonestat_node(memcg, nid, LRU_ACTIVE_FILE);

	return ret;
}

static unsigned long mem_cgroup_nr_file_lru_pages(struct mem_cgroup *memcg)
{
	u64 total = 0;
	int nid;

	for_each_node_state(nid, N_HIGH_MEMORY)
		total += mem_cgroup_node_nr_file_lru_pages(memcg, nid);

	return total;
}

static unsigned long mem_cgroup_node_nr_anon_lru_pages(struct mem_cgroup *memcg,
							int nid)
{
	unsigned long ret;

	ret = mem_cgroup_get_zonestat_node(memcg, nid, LRU_INACTIVE_ANON) +
		mem_cgroup_get_zonestat_node(memcg, nid, LRU_ACTIVE_ANON);

	return ret;
}

static unsigned long mem_cgroup_nr_anon_lru_pages(struct mem_cgroup *memcg)
{
	u64 total = 0;
	int nid;

	for_each_node_state(nid, N_HIGH_MEMORY)
		total += mem_cgroup_node_nr_anon_lru_pages(memcg, nid);

	return total;
}

static unsigned long
mem_cgroup_node_nr_unevictable_lru_pages(struct mem_cgroup *memcg, int nid)
{
	return mem_cgroup_get_zonestat_node(memcg, nid, LRU_UNEVICTABLE);
}

static unsigned long
mem_cgroup_nr_unevictable_lru_pages(struct mem_cgroup *memcg)
{
	u64 total = 0;
	int nid;

	for_each_node_state(nid, N_HIGH_MEMORY)
		total += mem_cgroup_node_nr_unevictable_lru_pages(memcg, nid);

	return total;
}

static unsigned long mem_cgroup_node_nr_lru_pages(struct mem_cgroup *memcg,
							int nid)
{
	enum lru_list l;
	u64 total = 0;

	for_each_lru(l)
		total += mem_cgroup_get_zonestat_node(memcg, nid, l);

	return total;
}

static unsigned long mem_cgroup_nr_lru_pages(struct mem_cgroup *memcg)
{
	u64 total = 0;
	int nid;

	for_each_node_state(nid, N_HIGH_MEMORY)
		total += mem_cgroup_node_nr_lru_pages(memcg, nid);

	return total;
}
#endif /* CONFIG_NUMA */

K
KOSAKI Motohiro 已提交
1218 1219 1220
struct zone_reclaim_stat *mem_cgroup_get_reclaim_stat(struct mem_cgroup *memcg,
						      struct zone *zone)
{
1221
	int nid = zone_to_nid(zone);
K
KOSAKI Motohiro 已提交
1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237
	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);
1238 1239
	if (!PageCgroupUsed(pc))
		return NULL;
1240 1241
	/* Ensure pc->mem_cgroup is visible after reading PCG_USED. */
	smp_rmb();
1242
	mz = page_cgroup_zoneinfo(pc->mem_cgroup, page);
K
KOSAKI Motohiro 已提交
1243 1244 1245
	return &mz->reclaim_stat;
}

1246 1247 1248 1249 1250
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,
1251
					int active, int file)
1252 1253 1254 1255 1256 1257
{
	unsigned long nr_taken = 0;
	struct page *page;
	unsigned long scan;
	LIST_HEAD(pc_list);
	struct list_head *src;
1258
	struct page_cgroup *pc, *tmp;
1259
	int nid = zone_to_nid(z);
1260 1261
	int zid = zone_idx(z);
	struct mem_cgroup_per_zone *mz;
1262
	int lru = LRU_FILE * file + active;
1263
	int ret;
1264

1265
	BUG_ON(!mem_cont);
1266
	mz = mem_cgroup_zoneinfo(mem_cont, nid, zid);
1267
	src = &mz->lists[lru];
1268

1269 1270
	scan = 0;
	list_for_each_entry_safe_reverse(pc, tmp, src, lru) {
H
Hugh Dickins 已提交
1271
		if (scan >= nr_to_scan)
1272
			break;
K
KAMEZAWA Hiroyuki 已提交
1273

1274 1275
		if (unlikely(!PageCgroupUsed(pc)))
			continue;
1276

1277
		page = lookup_cgroup_page(pc);
1278

H
Hugh Dickins 已提交
1279
		if (unlikely(!PageLRU(page)))
1280 1281
			continue;

H
Hugh Dickins 已提交
1282
		scan++;
1283 1284 1285
		ret = __isolate_lru_page(page, mode, file);
		switch (ret) {
		case 0:
1286
			list_move(&page->lru, dst);
1287
			mem_cgroup_del_lru(page);
1288
			nr_taken += hpage_nr_pages(page);
1289 1290 1291 1292 1293 1294 1295
			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;
1296 1297 1298 1299
		}
	}

	*scanned = scan;
1300 1301 1302 1303

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

1304 1305 1306
	return nr_taken;
}

1307 1308 1309
#define mem_cgroup_from_res_counter(counter, member)	\
	container_of(counter, struct mem_cgroup, member)

1310
/**
1311 1312
 * mem_cgroup_margin - calculate chargeable space of a memory cgroup
 * @mem: the memory cgroup
1313
 *
1314
 * Returns the maximum amount of memory @mem can be charged with, in
1315
 * pages.
1316
 */
1317
static unsigned long mem_cgroup_margin(struct mem_cgroup *mem)
1318
{
1319 1320 1321 1322 1323
	unsigned long long margin;

	margin = res_counter_margin(&mem->res);
	if (do_swap_account)
		margin = min(margin, res_counter_margin(&mem->memsw));
1324
	return margin >> PAGE_SHIFT;
1325 1326
}

K
KOSAKI Motohiro 已提交
1327 1328 1329 1330 1331 1332 1333 1334
static unsigned int get_swappiness(struct mem_cgroup *memcg)
{
	struct cgroup *cgrp = memcg->css.cgroup;

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

1335
	return memcg->swappiness;
K
KOSAKI Motohiro 已提交
1336 1337
}

1338 1339 1340
static void mem_cgroup_start_move(struct mem_cgroup *mem)
{
	int cpu;
1341 1342 1343 1344

	get_online_cpus();
	spin_lock(&mem->pcp_counter_lock);
	for_each_online_cpu(cpu)
1345
		per_cpu(mem->stat->count[MEM_CGROUP_ON_MOVE], cpu) += 1;
1346 1347 1348
	mem->nocpu_base.count[MEM_CGROUP_ON_MOVE] += 1;
	spin_unlock(&mem->pcp_counter_lock);
	put_online_cpus();
1349 1350 1351 1352 1353 1354 1355 1356 1357 1358

	synchronize_rcu();
}

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

	if (!mem)
		return;
1359 1360 1361
	get_online_cpus();
	spin_lock(&mem->pcp_counter_lock);
	for_each_online_cpu(cpu)
1362
		per_cpu(mem->stat->count[MEM_CGROUP_ON_MOVE], cpu) -= 1;
1363 1364 1365
	mem->nocpu_base.count[MEM_CGROUP_ON_MOVE] -= 1;
	spin_unlock(&mem->pcp_counter_lock);
	put_online_cpus();
1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383
}
/*
 * 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;
}
1384 1385 1386

static bool mem_cgroup_under_move(struct mem_cgroup *mem)
{
1387 1388
	struct mem_cgroup *from;
	struct mem_cgroup *to;
1389
	bool ret = false;
1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404
	/*
	 * Unlike task_move routines, we access mc.to, mc.from not under
	 * mutual exclusion by cgroup_mutex. Here, we take spinlock instead.
	 */
	spin_lock(&mc.lock);
	from = mc.from;
	to = mc.to;
	if (!from)
		goto unlock;
	if (from == mem || to == mem
	    || (mem->use_hierarchy && css_is_ancestor(&from->css, &mem->css))
	    || (mem->use_hierarchy && css_is_ancestor(&to->css,	&mem->css)))
		ret = true;
unlock:
	spin_unlock(&mc.lock);
1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423
	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;
}

1424
/**
1425
 * mem_cgroup_print_oom_info: Called from OOM with tasklist_lock held in read mode.
1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443
 * @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;

1444
	if (!memcg || !p)
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 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490
		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));
}

1491 1492 1493 1494 1495 1496 1497
/*
 * 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
1498 1499 1500 1501
	struct mem_cgroup *iter;

	for_each_mem_cgroup_tree(iter, mem)
		num++;
1502 1503 1504
	return num;
}

D
David Rientjes 已提交
1505 1506 1507 1508 1509 1510 1511 1512
/*
 * Return the memory (and swap, if configured) limit for a memcg.
 */
u64 mem_cgroup_get_limit(struct mem_cgroup *memcg)
{
	u64 limit;
	u64 memsw;

1513 1514 1515
	limit = res_counter_read_u64(&memcg->res, RES_LIMIT);
	limit += total_swap_pages << PAGE_SHIFT;

D
David Rientjes 已提交
1516 1517 1518 1519 1520 1521 1522 1523
	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);
}

1524
/*
K
KAMEZAWA Hiroyuki 已提交
1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560
 * 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;
}

1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 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 1634 1635
#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;

	if (time_after(mem->next_scan_node_update, jiffies))
		return;

	mem->next_scan_node_update = jiffies + 10*HZ;
	/* 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]) {

		if (mem_cgroup_get_zonestat_node(mem, nid, LRU_INACTIVE_FILE) ||
		    mem_cgroup_get_zonestat_node(mem, nid, LRU_ACTIVE_FILE))
			continue;

		if (total_swap_pages &&
		    (mem_cgroup_get_zonestat_node(mem, nid, LRU_INACTIVE_ANON) ||
		     mem_cgroup_get_zonestat_node(mem, nid, LRU_ACTIVE_ANON)))
			continue;
		node_clear(nid, mem->scan_nodes);
	}
}

/*
 * 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;
}

#else
int mem_cgroup_select_victim_node(struct mem_cgroup *mem)
{
	return 0;
}
#endif

K
KAMEZAWA Hiroyuki 已提交
1636 1637 1638 1639
/*
 * 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.
1640 1641
 *
 * root_mem is the original ancestor that we've been reclaim from.
K
KAMEZAWA Hiroyuki 已提交
1642 1643 1644
 *
 * We give up and return to the caller when we visit root_mem twice.
 * (other groups can be removed while we're walking....)
1645 1646
 *
 * If shrink==true, for avoiding to free too much, this returns immedieately.
1647 1648
 */
static int mem_cgroup_hierarchical_reclaim(struct mem_cgroup *root_mem,
1649
						struct zone *zone,
1650
						gfp_t gfp_mask,
1651 1652
						unsigned long reclaim_options,
						unsigned long *total_scanned)
1653
{
K
KAMEZAWA Hiroyuki 已提交
1654 1655 1656
	struct mem_cgroup *victim;
	int ret, total = 0;
	int loop = 0;
1657 1658
	bool noswap = reclaim_options & MEM_CGROUP_RECLAIM_NOSWAP;
	bool shrink = reclaim_options & MEM_CGROUP_RECLAIM_SHRINK;
1659
	bool check_soft = reclaim_options & MEM_CGROUP_RECLAIM_SOFT;
1660
	unsigned long excess;
1661
	unsigned long nr_scanned;
1662 1663

	excess = res_counter_soft_limit_excess(&root_mem->res) >> PAGE_SHIFT;
K
KAMEZAWA Hiroyuki 已提交
1664

1665
	/* If memsw_is_minimum==1, swap-out is of-no-use. */
1666
	if (!check_soft && root_mem->memsw_is_minimum)
1667 1668
		noswap = true;

1669
	while (1) {
K
KAMEZAWA Hiroyuki 已提交
1670
		victim = mem_cgroup_select_victim(root_mem);
1671
		if (victim == root_mem) {
K
KAMEZAWA Hiroyuki 已提交
1672
			loop++;
1673 1674
			if (loop >= 1)
				drain_all_stock_async();
1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685
			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
1686
				 * We want to do more targeted reclaim.
1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697
				 * 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;
				}
			}
		}
1698
		if (!mem_cgroup_local_usage(victim)) {
K
KAMEZAWA Hiroyuki 已提交
1699 1700
			/* this cgroup's local usage == 0 */
			css_put(&victim->css);
1701 1702
			continue;
		}
K
KAMEZAWA Hiroyuki 已提交
1703
		/* we use swappiness of local cgroup */
1704
		if (check_soft) {
1705
			ret = mem_cgroup_shrink_node_zone(victim, gfp_mask,
1706 1707 1708 1709
				noswap, get_swappiness(victim), zone,
				&nr_scanned);
			*total_scanned += nr_scanned;
		} else
1710 1711
			ret = try_to_free_mem_cgroup_pages(victim, gfp_mask,
						noswap, get_swappiness(victim));
K
KAMEZAWA Hiroyuki 已提交
1712
		css_put(&victim->css);
1713 1714 1715 1716 1717 1718 1719
		/*
		 * 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
1720
		total += ret;
1721
		if (check_soft) {
1722
			if (!res_counter_soft_limit_excess(&root_mem->res))
1723
				return total;
1724
		} else if (mem_cgroup_margin(root_mem))
1725
			return total;
1726
	}
K
KAMEZAWA Hiroyuki 已提交
1727
	return total;
1728 1729
}

K
KAMEZAWA Hiroyuki 已提交
1730 1731 1732 1733 1734 1735
/*
 * Check OOM-Killer is already running under our hierarchy.
 * If someone is running, return false.
 */
static bool mem_cgroup_oom_lock(struct mem_cgroup *mem)
{
K
KAMEZAWA Hiroyuki 已提交
1736 1737
	int x, lock_count = 0;
	struct mem_cgroup *iter;
1738

K
KAMEZAWA Hiroyuki 已提交
1739 1740 1741 1742
	for_each_mem_cgroup_tree(iter, mem) {
		x = atomic_inc_return(&iter->oom_lock);
		lock_count = max(x, lock_count);
	}
K
KAMEZAWA Hiroyuki 已提交
1743 1744 1745 1746

	if (lock_count == 1)
		return true;
	return false;
1747
}
1748

K
KAMEZAWA Hiroyuki 已提交
1749
static int mem_cgroup_oom_unlock(struct mem_cgroup *mem)
1750
{
K
KAMEZAWA Hiroyuki 已提交
1751 1752
	struct mem_cgroup *iter;

K
KAMEZAWA Hiroyuki 已提交
1753 1754 1755 1756 1757
	/*
	 * 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
1758 1759
	for_each_mem_cgroup_tree(iter, mem)
		atomic_add_unless(&iter->oom_lock, -1, 0);
1760 1761 1762
	return 0;
}

K
KAMEZAWA Hiroyuki 已提交
1763 1764 1765 1766

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

K
KAMEZAWA Hiroyuki 已提交
1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802
struct oom_wait_info {
	struct mem_cgroup *mem;
	wait_queue_t	wait;
};

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

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

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

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

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

1803 1804
static void memcg_oom_recover(struct mem_cgroup *mem)
{
1805
	if (mem && atomic_read(&mem->oom_lock))
1806 1807 1808
		memcg_wakeup_oom(mem);
}

K
KAMEZAWA Hiroyuki 已提交
1809 1810 1811 1812
/*
 * 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)
1813
{
K
KAMEZAWA Hiroyuki 已提交
1814
	struct oom_wait_info owait;
1815
	bool locked, need_to_kill;
K
KAMEZAWA Hiroyuki 已提交
1816

K
KAMEZAWA Hiroyuki 已提交
1817 1818 1819 1820 1821
	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);
1822
	need_to_kill = true;
K
KAMEZAWA Hiroyuki 已提交
1823 1824 1825 1826 1827 1828 1829 1830
	/* At first, try to OOM lock hierarchy under mem.*/
	mutex_lock(&memcg_oom_mutex);
	locked = mem_cgroup_oom_lock(mem);
	/*
	 * Even if signal_pending(), we can't quit charge() loop without
	 * accounting. So, UNINTERRUPTIBLE is appropriate. But SIGKILL
	 * under OOM is always welcomed, use TASK_KILLABLE here.
	 */
1831 1832 1833 1834
	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
1835
		mem_cgroup_oom_notify(mem);
K
KAMEZAWA Hiroyuki 已提交
1836 1837
	mutex_unlock(&memcg_oom_mutex);

1838 1839
	if (need_to_kill) {
		finish_wait(&memcg_oom_waitq, &owait.wait);
K
KAMEZAWA Hiroyuki 已提交
1840
		mem_cgroup_out_of_memory(mem, mask);
1841
	} else {
K
KAMEZAWA Hiroyuki 已提交
1842
		schedule();
K
KAMEZAWA Hiroyuki 已提交
1843
		finish_wait(&memcg_oom_waitq, &owait.wait);
K
KAMEZAWA Hiroyuki 已提交
1844 1845 1846
	}
	mutex_lock(&memcg_oom_mutex);
	mem_cgroup_oom_unlock(mem);
K
KAMEZAWA Hiroyuki 已提交
1847
	memcg_wakeup_oom(mem);
K
KAMEZAWA Hiroyuki 已提交
1848 1849 1850 1851 1852 1853 1854
	mutex_unlock(&memcg_oom_mutex);

	if (test_thread_flag(TIF_MEMDIE) || fatal_signal_pending(current))
		return false;
	/* Give chance to dying process */
	schedule_timeout(1);
	return true;
1855 1856
}

1857 1858 1859
/*
 * Currently used to update mapped file statistics, but the routine can be
 * generalized to update other statistics as well.
1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878
 *
 * 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.
1879
 */
1880

1881 1882
void mem_cgroup_update_page_stat(struct page *page,
				 enum mem_cgroup_page_stat_item idx, int val)
1883 1884
{
	struct mem_cgroup *mem;
1885 1886
	struct page_cgroup *pc = lookup_page_cgroup(page);
	bool need_unlock = false;
1887
	unsigned long uninitialized_var(flags);
1888 1889 1890 1891

	if (unlikely(!pc))
		return;

1892
	rcu_read_lock();
1893
	mem = pc->mem_cgroup;
1894 1895 1896
	if (unlikely(!mem || !PageCgroupUsed(pc)))
		goto out;
	/* pc->mem_cgroup is unstable ? */
1897
	if (unlikely(mem_cgroup_stealed(mem)) || PageTransHuge(page)) {
1898
		/* take a lock against to access pc->mem_cgroup */
1899
		move_lock_page_cgroup(pc, &flags);
1900 1901 1902 1903 1904
		need_unlock = true;
		mem = pc->mem_cgroup;
		if (!mem || !PageCgroupUsed(pc))
			goto out;
	}
1905 1906

	switch (idx) {
1907
	case MEMCG_NR_FILE_MAPPED:
1908 1909 1910
		if (val > 0)
			SetPageCgroupFileMapped(pc);
		else if (!page_mapped(page))
1911
			ClearPageCgroupFileMapped(pc);
1912
		idx = MEM_CGROUP_STAT_FILE_MAPPED;
1913 1914 1915
		break;
	default:
		BUG();
1916
	}
1917

1918 1919
	this_cpu_add(mem->stat->count[idx], val);

1920 1921
out:
	if (unlikely(need_unlock))
1922
		move_unlock_page_cgroup(pc, &flags);
1923 1924
	rcu_read_unlock();
	return;
1925
}
1926
EXPORT_SYMBOL(mem_cgroup_update_page_stat);
1927

1928 1929 1930 1931
/*
 * size of first charge trial. "32" comes from vmscan.c's magic value.
 * TODO: maybe necessary to use big numbers in big irons.
 */
1932
#define CHARGE_BATCH	32U
1933 1934
struct memcg_stock_pcp {
	struct mem_cgroup *cached; /* this never be root cgroup */
1935
	unsigned int nr_pages;
1936 1937 1938 1939 1940 1941
	struct work_struct work;
};
static DEFINE_PER_CPU(struct memcg_stock_pcp, memcg_stock);
static atomic_t memcg_drain_count;

/*
1942
 * Try to consume stocked charge on this cpu. If success, one page is consumed
1943 1944 1945 1946 1947 1948 1949 1950 1951 1952
 * 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);
1953 1954
	if (mem == stock->cached && stock->nr_pages)
		stock->nr_pages--;
1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967
	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;

1968 1969 1970 1971
	if (stock->nr_pages) {
		unsigned long bytes = stock->nr_pages * PAGE_SIZE;

		res_counter_uncharge(&old->res, bytes);
1972
		if (do_swap_account)
1973 1974
			res_counter_uncharge(&old->memsw, bytes);
		stock->nr_pages = 0;
1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990
	}
	stock->cached = NULL;
}

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

/*
 * Cache charges(val) which is from res_counter, to local per_cpu area.
1991
 * This will be consumed by consume_stock() function, later.
1992
 */
1993
static void refill_stock(struct mem_cgroup *mem, unsigned int nr_pages)
1994 1995 1996 1997 1998 1999 2000
{
	struct memcg_stock_pcp *stock = &get_cpu_var(memcg_stock);

	if (stock->cached != mem) { /* reset if necessary */
		drain_stock(stock);
		stock->cached = mem;
	}
2001
	stock->nr_pages += nr_pages;
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042
	put_cpu_var(memcg_stock);
}

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

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

2043 2044 2045 2046 2047 2048 2049 2050 2051 2052
/*
 * 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++) {
2053
		long x = per_cpu(mem->stat->count[i], cpu);
2054 2055 2056 2057

		per_cpu(mem->stat->count[i], cpu) = 0;
		mem->nocpu_base.count[i] += x;
	}
2058 2059 2060 2061 2062 2063
	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;
	}
2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074
	/* 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];
2075 2076 2077 2078
	spin_unlock(&mem->pcp_counter_lock);
}

static int __cpuinit memcg_cpu_hotplug_callback(struct notifier_block *nb,
2079 2080 2081 2082 2083
					unsigned long action,
					void *hcpu)
{
	int cpu = (unsigned long)hcpu;
	struct memcg_stock_pcp *stock;
2084
	struct mem_cgroup *iter;
2085

2086 2087 2088 2089 2090 2091
	if ((action == CPU_ONLINE)) {
		for_each_mem_cgroup_all(iter)
			synchronize_mem_cgroup_on_move(iter, cpu);
		return NOTIFY_OK;
	}

2092
	if ((action != CPU_DEAD) || action != CPU_DEAD_FROZEN)
2093
		return NOTIFY_OK;
2094 2095 2096 2097

	for_each_mem_cgroup_all(iter)
		mem_cgroup_drain_pcp_counter(iter, cpu);

2098 2099 2100 2101 2102
	stock = &per_cpu(memcg_stock, cpu);
	drain_stock(stock);
	return NOTIFY_OK;
}

2103 2104 2105 2106 2107 2108 2109 2110 2111 2112

/* 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 */
};

2113 2114
static int mem_cgroup_do_charge(struct mem_cgroup *mem, gfp_t gfp_mask,
				unsigned int nr_pages, bool oom_check)
2115
{
2116
	unsigned long csize = nr_pages * PAGE_SIZE;
2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130
	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;

2131
		res_counter_uncharge(&mem->res, csize);
2132 2133 2134 2135
		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);
2136
	/*
2137 2138
	 * nr_pages can be either a huge page (HPAGE_PMD_NR), a batch
	 * of regular pages (CHARGE_BATCH), or a single regular page (1).
2139 2140 2141 2142
	 *
	 * Never reclaim on behalf of optional batching, retry with a
	 * single page instead.
	 */
2143
	if (nr_pages == CHARGE_BATCH)
2144 2145 2146 2147 2148 2149
		return CHARGE_RETRY;

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

	ret = mem_cgroup_hierarchical_reclaim(mem_over_limit, NULL,
2150
					      gfp_mask, flags, NULL);
2151
	if (mem_cgroup_margin(mem_over_limit) >= nr_pages)
2152
		return CHARGE_RETRY;
2153
	/*
2154 2155 2156 2157 2158 2159 2160
	 * 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.
2161
	 */
2162
	if (nr_pages == 1 && ret)
2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181
		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;
}

2182 2183 2184
/*
 * Unlike exported interface, "oom" parameter is added. if oom==true,
 * oom-killer can be invoked.
2185
 */
2186
static int __mem_cgroup_try_charge(struct mm_struct *mm,
A
Andrea Arcangeli 已提交
2187
				   gfp_t gfp_mask,
2188 2189 2190
				   unsigned int nr_pages,
				   struct mem_cgroup **memcg,
				   bool oom)
2191
{
2192
	unsigned int batch = max(CHARGE_BATCH, nr_pages);
2193 2194 2195
	int nr_oom_retries = MEM_CGROUP_RECLAIM_RETRIES;
	struct mem_cgroup *mem = NULL;
	int ret;
2196

K
KAMEZAWA Hiroyuki 已提交
2197 2198 2199 2200 2201 2202 2203 2204
	/*
	 * 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;
2205

2206
	/*
2207 2208
	 * We always charge the cgroup the mm_struct belongs to.
	 * The mm_struct's mem_cgroup changes on task migration if the
2209 2210 2211
	 * 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
2212 2213 2214 2215
	if (!*memcg && !mm)
		goto bypass;
again:
	if (*memcg) { /* css should be a valid one */
2216
		mem = *memcg;
K
KAMEZAWA Hiroyuki 已提交
2217 2218 2219
		VM_BUG_ON(css_is_removed(&mem->css));
		if (mem_cgroup_is_root(mem))
			goto done;
2220
		if (nr_pages == 1 && consume_stock(mem))
K
KAMEZAWA Hiroyuki 已提交
2221
			goto done;
2222 2223
		css_get(&mem->css);
	} else {
K
KAMEZAWA Hiroyuki 已提交
2224
		struct task_struct *p;
2225

K
KAMEZAWA Hiroyuki 已提交
2226 2227 2228
		rcu_read_lock();
		p = rcu_dereference(mm->owner);
		/*
2229 2230 2231 2232 2233 2234 2235 2236
		 * 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
2237 2238
		 */
		mem = mem_cgroup_from_task(p);
2239
		if (!mem || mem_cgroup_is_root(mem)) {
K
KAMEZAWA Hiroyuki 已提交
2240 2241 2242
			rcu_read_unlock();
			goto done;
		}
2243
		if (nr_pages == 1 && consume_stock(mem)) {
K
KAMEZAWA Hiroyuki 已提交
2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261
			/*
			 * 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();
	}
2262

2263 2264
	do {
		bool oom_check;
2265

2266
		/* If killed, bypass charge */
K
KAMEZAWA Hiroyuki 已提交
2267 2268
		if (fatal_signal_pending(current)) {
			css_put(&mem->css);
2269
			goto bypass;
K
KAMEZAWA Hiroyuki 已提交
2270
		}
2271

2272 2273 2274 2275
		oom_check = false;
		if (oom && !nr_oom_retries) {
			oom_check = true;
			nr_oom_retries = MEM_CGROUP_RECLAIM_RETRIES;
2276
		}
2277

2278
		ret = mem_cgroup_do_charge(mem, gfp_mask, batch, oom_check);
2279 2280 2281 2282
		switch (ret) {
		case CHARGE_OK:
			break;
		case CHARGE_RETRY: /* not in OOM situation but retry */
2283
			batch = nr_pages;
K
KAMEZAWA Hiroyuki 已提交
2284 2285 2286
			css_put(&mem->css);
			mem = NULL;
			goto again;
2287
		case CHARGE_WOULDBLOCK: /* !__GFP_WAIT */
K
KAMEZAWA Hiroyuki 已提交
2288
			css_put(&mem->css);
2289 2290
			goto nomem;
		case CHARGE_NOMEM: /* OOM routine works */
K
KAMEZAWA Hiroyuki 已提交
2291 2292
			if (!oom) {
				css_put(&mem->css);
K
KAMEZAWA Hiroyuki 已提交
2293
				goto nomem;
K
KAMEZAWA Hiroyuki 已提交
2294
			}
2295 2296 2297 2298
			/* If oom, we never return -ENOMEM */
			nr_oom_retries--;
			break;
		case CHARGE_OOM_DIE: /* Killed by OOM Killer */
K
KAMEZAWA Hiroyuki 已提交
2299
			css_put(&mem->css);
K
KAMEZAWA Hiroyuki 已提交
2300
			goto bypass;
2301
		}
2302 2303
	} while (ret != CHARGE_OK);

2304 2305
	if (batch > nr_pages)
		refill_stock(mem, batch - nr_pages);
K
KAMEZAWA Hiroyuki 已提交
2306
	css_put(&mem->css);
2307
done:
K
KAMEZAWA Hiroyuki 已提交
2308
	*memcg = mem;
2309 2310
	return 0;
nomem:
K
KAMEZAWA Hiroyuki 已提交
2311
	*memcg = NULL;
2312
	return -ENOMEM;
K
KAMEZAWA Hiroyuki 已提交
2313 2314 2315
bypass:
	*memcg = NULL;
	return 0;
2316
}
2317

2318 2319 2320 2321 2322
/*
 * 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().
 */
2323
static void __mem_cgroup_cancel_charge(struct mem_cgroup *mem,
2324
				       unsigned int nr_pages)
2325 2326
{
	if (!mem_cgroup_is_root(mem)) {
2327 2328 2329
		unsigned long bytes = nr_pages * PAGE_SIZE;

		res_counter_uncharge(&mem->res, bytes);
2330
		if (do_swap_account)
2331
			res_counter_uncharge(&mem->memsw, bytes);
2332
	}
2333 2334
}

2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353
/*
 * 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);
}

2354
struct mem_cgroup *try_get_mem_cgroup_from_page(struct page *page)
2355
{
2356
	struct mem_cgroup *mem = NULL;
2357
	struct page_cgroup *pc;
2358
	unsigned short id;
2359 2360
	swp_entry_t ent;

2361 2362 2363
	VM_BUG_ON(!PageLocked(page));

	pc = lookup_page_cgroup(page);
2364
	lock_page_cgroup(pc);
2365
	if (PageCgroupUsed(pc)) {
2366
		mem = pc->mem_cgroup;
2367 2368
		if (mem && !css_tryget(&mem->css))
			mem = NULL;
2369
	} else if (PageSwapCache(page)) {
2370
		ent.val = page_private(page);
2371 2372 2373 2374 2375 2376
		id = lookup_swap_cgroup(ent);
		rcu_read_lock();
		mem = mem_cgroup_lookup(id);
		if (mem && !css_tryget(&mem->css))
			mem = NULL;
		rcu_read_unlock();
2377
	}
2378
	unlock_page_cgroup(pc);
2379 2380 2381
	return mem;
}

2382
static void __mem_cgroup_commit_charge(struct mem_cgroup *mem,
2383
				       struct page *page,
2384
				       unsigned int nr_pages,
2385
				       struct page_cgroup *pc,
2386
				       enum charge_type ctype)
2387
{
2388 2389 2390
	lock_page_cgroup(pc);
	if (unlikely(PageCgroupUsed(pc))) {
		unlock_page_cgroup(pc);
2391
		__mem_cgroup_cancel_charge(mem, nr_pages);
2392 2393 2394 2395 2396 2397
		return;
	}
	/*
	 * we don't need page_cgroup_lock about tail pages, becase they are not
	 * accessed by any other context at this point.
	 */
2398
	pc->mem_cgroup = mem;
2399 2400 2401 2402 2403 2404 2405
	/*
	 * 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
2406
	smp_wmb();
2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419
	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;
	}
2420

2421
	mem_cgroup_charge_statistics(mem, PageCgroupCache(pc), nr_pages);
2422
	unlock_page_cgroup(pc);
2423 2424 2425 2426 2427
	/*
	 * "charge_statistics" updated event counter. Then, check it.
	 * Insert ancestor (and ancestor's ancestors), to softlimit RB-tree.
	 * if they exceeds softlimit.
	 */
2428
	memcg_check_events(mem, page);
2429
}
2430

2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444
#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;

2445 2446
	if (mem_cgroup_disabled())
		return;
2447
	/*
2448
	 * We have no races with charge/uncharge but will have races with
2449 2450 2451 2452 2453 2454
	 * page state accounting.
	 */
	move_lock_page_cgroup(head_pc, &flags);

	tail_pc->mem_cgroup = head_pc->mem_cgroup;
	smp_wmb(); /* see __commit_charge() */
2455 2456 2457 2458 2459 2460 2461 2462 2463 2464
	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);
2465
		mz = page_cgroup_zoneinfo(head_pc->mem_cgroup, head);
2466 2467
		MEM_CGROUP_ZSTAT(mz, lru) -= 1;
	}
2468 2469 2470 2471 2472
	tail_pc->flags = head_pc->flags & ~PCGF_NOCOPY_AT_SPLIT;
	move_unlock_page_cgroup(head_pc, &flags);
}
#endif

2473
/**
2474
 * mem_cgroup_move_account - move account of the page
2475
 * @page: the page
2476
 * @nr_pages: number of regular pages (>1 for huge pages)
2477 2478 2479
 * @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.
2480
 * @uncharge: whether we should call uncharge and css_put against @from.
2481 2482
 *
 * The caller must confirm following.
K
KAMEZAWA Hiroyuki 已提交
2483
 * - page is not on LRU (isolate_page() is useful.)
2484
 * - compound_lock is held when nr_pages > 1
2485
 *
2486
 * This function doesn't do "charge" nor css_get to new cgroup. It should be
L
Lucas De Marchi 已提交
2487
 * done by a caller(__mem_cgroup_try_charge would be useful). If @uncharge is
2488 2489
 * true, this function does "uncharge" from old cgroup, but it doesn't if
 * @uncharge is false, so a caller should do "uncharge".
2490
 */
2491 2492 2493 2494 2495 2496
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)
2497
{
2498 2499
	unsigned long flags;
	int ret;
2500

2501
	VM_BUG_ON(from == to);
2502
	VM_BUG_ON(PageLRU(page));
2503 2504 2505 2506 2507 2508 2509
	/*
	 * 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;
2510
	if (nr_pages > 1 && !PageTransHuge(page))
2511 2512 2513 2514 2515 2516 2517 2518 2519
		goto out;

	lock_page_cgroup(pc);

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

	move_lock_page_cgroup(pc, &flags);
2520

2521
	if (PageCgroupFileMapped(pc)) {
2522 2523 2524 2525 2526
		/* 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();
2527
	}
2528
	mem_cgroup_charge_statistics(from, PageCgroupCache(pc), -nr_pages);
2529 2530
	if (uncharge)
		/* This is not "cancel", but cancel_charge does all we need. */
2531
		__mem_cgroup_cancel_charge(from, nr_pages);
2532

2533
	/* caller should have done css_get */
K
KAMEZAWA Hiroyuki 已提交
2534
	pc->mem_cgroup = to;
2535
	mem_cgroup_charge_statistics(to, PageCgroupCache(pc), nr_pages);
2536 2537 2538
	/*
	 * We charges against "to" which may not have any tasks. Then, "to"
	 * can be under rmdir(). But in current implementation, caller of
2539
	 * this function is just force_empty() and move charge, so it's
L
Lucas De Marchi 已提交
2540
	 * guaranteed that "to" is never removed. So, we don't check rmdir
2541
	 * status here.
2542
	 */
2543 2544 2545
	move_unlock_page_cgroup(pc, &flags);
	ret = 0;
unlock:
2546
	unlock_page_cgroup(pc);
2547 2548 2549
	/*
	 * check events
	 */
2550 2551
	memcg_check_events(to, page);
	memcg_check_events(from, page);
2552
out:
2553 2554 2555 2556 2557 2558 2559
	return ret;
}

/*
 * move charges to its parent.
 */

2560 2561
static int mem_cgroup_move_parent(struct page *page,
				  struct page_cgroup *pc,
2562 2563 2564 2565 2566 2567
				  struct mem_cgroup *child,
				  gfp_t gfp_mask)
{
	struct cgroup *cg = child->css.cgroup;
	struct cgroup *pcg = cg->parent;
	struct mem_cgroup *parent;
2568
	unsigned int nr_pages;
2569
	unsigned long uninitialized_var(flags);
2570 2571 2572 2573 2574 2575
	int ret;

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

2576 2577 2578 2579 2580
	ret = -EBUSY;
	if (!get_page_unless_zero(page))
		goto out;
	if (isolate_lru_page(page))
		goto put;
2581

2582
	nr_pages = hpage_nr_pages(page);
K
KAMEZAWA Hiroyuki 已提交
2583

2584
	parent = mem_cgroup_from_cont(pcg);
2585
	ret = __mem_cgroup_try_charge(NULL, gfp_mask, nr_pages, &parent, false);
2586
	if (ret || !parent)
2587
		goto put_back;
2588

2589
	if (nr_pages > 1)
2590 2591
		flags = compound_lock_irqsave(page);

2592
	ret = mem_cgroup_move_account(page, nr_pages, pc, child, parent, true);
2593
	if (ret)
2594
		__mem_cgroup_cancel_charge(parent, nr_pages);
2595

2596
	if (nr_pages > 1)
2597
		compound_unlock_irqrestore(page, flags);
2598
put_back:
K
KAMEZAWA Hiroyuki 已提交
2599
	putback_lru_page(page);
2600
put:
2601
	put_page(page);
2602
out:
2603 2604 2605
	return ret;
}

2606 2607 2608 2609 2610 2611 2612
/*
 * 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,
2613
				gfp_t gfp_mask, enum charge_type ctype)
2614
{
2615
	struct mem_cgroup *mem = NULL;
2616
	unsigned int nr_pages = 1;
2617
	struct page_cgroup *pc;
2618
	bool oom = true;
2619
	int ret;
A
Andrea Arcangeli 已提交
2620

A
Andrea Arcangeli 已提交
2621
	if (PageTransHuge(page)) {
2622
		nr_pages <<= compound_order(page);
A
Andrea Arcangeli 已提交
2623
		VM_BUG_ON(!PageTransHuge(page));
2624 2625 2626 2627 2628
		/*
		 * 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
2629
	}
2630 2631

	pc = lookup_page_cgroup(page);
2632
	BUG_ON(!pc); /* XXX: remove this and move pc lookup into commit */
2633

2634
	ret = __mem_cgroup_try_charge(mm, gfp_mask, nr_pages, &mem, oom);
2635
	if (ret || !mem)
2636 2637
		return ret;

2638
	__mem_cgroup_commit_charge(mem, page, nr_pages, pc, ctype);
2639 2640 2641
	return 0;
}

2642 2643
int mem_cgroup_newpage_charge(struct page *page,
			      struct mm_struct *mm, gfp_t gfp_mask)
2644
{
2645
	if (mem_cgroup_disabled())
2646
		return 0;
2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657
	/*
	 * 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;
2658
	return mem_cgroup_charge_common(page, mm, gfp_mask,
2659
				MEM_CGROUP_CHARGE_TYPE_MAPPED);
2660 2661
}

D
Daisuke Nishimura 已提交
2662 2663 2664 2665
static void
__mem_cgroup_commit_charge_swapin(struct page *page, struct mem_cgroup *ptr,
					enum charge_type ctype);

2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681
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;
}

2682 2683
int mem_cgroup_cache_charge(struct page *page, struct mm_struct *mm,
				gfp_t gfp_mask)
2684
{
2685
	struct mem_cgroup *mem = NULL;
2686 2687
	int ret;

2688
	if (mem_cgroup_disabled())
2689
		return 0;
2690 2691
	if (PageCompound(page))
		return 0;
2692 2693 2694 2695 2696 2697 2698 2699
	/*
	 * Corner case handling. This is called from add_to_page_cache()
	 * in usual. But some FS (shmem) precharges this page before calling it
	 * and call add_to_page_cache() with GFP_NOWAIT.
	 *
	 * For GFP_NOWAIT case, the page may be pre-charged before calling
	 * add_to_page_cache(). (See shmem.c) check it here and avoid to call
	 * charge twice. (It works but has to pay a bit larger cost.)
2700 2701
	 * And when the page is SwapCache, it should take swap information
	 * into account. This is under lock_page() now.
2702 2703 2704 2705
	 */
	if (!(gfp_mask & __GFP_WAIT)) {
		struct page_cgroup *pc;

2706 2707 2708 2709 2710 2711
		pc = lookup_page_cgroup(page);
		if (!pc)
			return 0;
		lock_page_cgroup(pc);
		if (PageCgroupUsed(pc)) {
			unlock_page_cgroup(pc);
2712 2713
			return 0;
		}
2714
		unlock_page_cgroup(pc);
2715 2716
	}

2717
	if (unlikely(!mm))
2718
		mm = &init_mm;
2719

2720 2721 2722 2723
	if (page_is_file_cache(page)) {
		ret = __mem_cgroup_try_charge(mm, gfp_mask, 1, &mem, true);
		if (ret || !mem)
			return ret;
2724

2725 2726 2727 2728 2729 2730 2731 2732 2733
		/*
		 * 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
2734 2735 2736 2737 2738 2739 2740 2741
	/* 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,
2742
					MEM_CGROUP_CHARGE_TYPE_SHMEM);
2743 2744

	return ret;
2745 2746
}

2747 2748 2749
/*
 * While swap-in, try_charge -> commit or cancel, the page is locked.
 * And when try_charge() successfully returns, one refcnt to memcg without
2750
 * struct page_cgroup is acquired. This refcnt will be consumed by
2751 2752
 * "commit()" or removed by "cancel()"
 */
2753 2754 2755 2756 2757
int mem_cgroup_try_charge_swapin(struct mm_struct *mm,
				 struct page *page,
				 gfp_t mask, struct mem_cgroup **ptr)
{
	struct mem_cgroup *mem;
2758
	int ret;
2759

2760 2761
	*ptr = NULL;

2762
	if (mem_cgroup_disabled())
2763 2764 2765 2766 2767 2768
		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
2769 2770 2771
	 * 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.
2772 2773
	 */
	if (!PageSwapCache(page))
H
Hugh Dickins 已提交
2774
		goto charge_cur_mm;
2775
	mem = try_get_mem_cgroup_from_page(page);
2776 2777
	if (!mem)
		goto charge_cur_mm;
2778
	*ptr = mem;
2779
	ret = __mem_cgroup_try_charge(NULL, mask, 1, ptr, true);
2780 2781
	css_put(&mem->css);
	return ret;
2782 2783 2784
charge_cur_mm:
	if (unlikely(!mm))
		mm = &init_mm;
2785
	return __mem_cgroup_try_charge(mm, mask, 1, ptr, true);
2786 2787
}

D
Daisuke Nishimura 已提交
2788 2789 2790
static void
__mem_cgroup_commit_charge_swapin(struct page *page, struct mem_cgroup *ptr,
					enum charge_type ctype)
2791
{
2792
	if (mem_cgroup_disabled())
2793 2794 2795
		return;
	if (!ptr)
		return;
2796
	cgroup_exclude_rmdir(&ptr->css);
2797 2798

	__mem_cgroup_commit_charge_lrucare(page, ptr, ctype);
2799 2800 2801
	/*
	 * Now swap is on-memory. This means this page may be
	 * counted both as mem and swap....double count.
2802 2803 2804
	 * 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.
2805
	 */
2806
	if (do_swap_account && PageSwapCache(page)) {
2807
		swp_entry_t ent = {.val = page_private(page)};
2808
		unsigned short id;
2809
		struct mem_cgroup *memcg;
2810 2811 2812 2813

		id = swap_cgroup_record(ent, 0);
		rcu_read_lock();
		memcg = mem_cgroup_lookup(id);
2814
		if (memcg) {
2815 2816 2817 2818
			/*
			 * This recorded memcg can be obsolete one. So, avoid
			 * calling css_tryget
			 */
2819
			if (!mem_cgroup_is_root(memcg))
2820
				res_counter_uncharge(&memcg->memsw, PAGE_SIZE);
2821
			mem_cgroup_swap_statistics(memcg, false);
2822 2823
			mem_cgroup_put(memcg);
		}
2824
		rcu_read_unlock();
2825
	}
2826 2827 2828 2829 2830 2831
	/*
	 * 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);
2832 2833
}

D
Daisuke Nishimura 已提交
2834 2835 2836 2837 2838 2839
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);
}

2840 2841
void mem_cgroup_cancel_charge_swapin(struct mem_cgroup *mem)
{
2842
	if (mem_cgroup_disabled())
2843 2844 2845
		return;
	if (!mem)
		return;
2846
	__mem_cgroup_cancel_charge(mem, 1);
2847 2848
}

2849 2850 2851
static void mem_cgroup_do_uncharge(struct mem_cgroup *mem,
				   unsigned int nr_pages,
				   const enum charge_type ctype)
2852 2853 2854
{
	struct memcg_batch_info *batch = NULL;
	bool uncharge_memsw = true;
2855

2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867
	/* 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;
2868 2869
	/*
	 * do_batch > 0 when unmapping pages or inode invalidate/truncate.
L
Lucas De Marchi 已提交
2870
	 * In those cases, all pages freed continuously can be expected to be in
2871 2872 2873 2874 2875 2876 2877 2878
	 * 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;

2879
	if (nr_pages > 1)
A
Andrea Arcangeli 已提交
2880 2881
		goto direct_uncharge;

2882 2883 2884 2885 2886 2887 2888 2889
	/*
	 * 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 */
2890
	batch->nr_pages++;
2891
	if (uncharge_memsw)
2892
		batch->memsw_nr_pages++;
2893 2894
	return;
direct_uncharge:
2895
	res_counter_uncharge(&mem->res, nr_pages * PAGE_SIZE);
2896
	if (uncharge_memsw)
2897
		res_counter_uncharge(&mem->memsw, nr_pages * PAGE_SIZE);
2898 2899
	if (unlikely(batch->memcg != mem))
		memcg_oom_recover(mem);
2900 2901
	return;
}
2902

2903
/*
2904
 * uncharge if !page_mapped(page)
2905
 */
2906
static struct mem_cgroup *
2907
__mem_cgroup_uncharge_common(struct page *page, enum charge_type ctype)
2908
{
2909
	struct mem_cgroup *mem = NULL;
2910 2911
	unsigned int nr_pages = 1;
	struct page_cgroup *pc;
2912

2913
	if (mem_cgroup_disabled())
2914
		return NULL;
2915

K
KAMEZAWA Hiroyuki 已提交
2916
	if (PageSwapCache(page))
2917
		return NULL;
K
KAMEZAWA Hiroyuki 已提交
2918

A
Andrea Arcangeli 已提交
2919
	if (PageTransHuge(page)) {
2920
		nr_pages <<= compound_order(page);
A
Andrea Arcangeli 已提交
2921 2922
		VM_BUG_ON(!PageTransHuge(page));
	}
2923
	/*
2924
	 * Check if our page_cgroup is valid
2925
	 */
2926 2927
	pc = lookup_page_cgroup(page);
	if (unlikely(!pc || !PageCgroupUsed(pc)))
2928
		return NULL;
2929

2930
	lock_page_cgroup(pc);
K
KAMEZAWA Hiroyuki 已提交
2931

2932 2933
	mem = pc->mem_cgroup;

K
KAMEZAWA Hiroyuki 已提交
2934 2935 2936 2937 2938
	if (!PageCgroupUsed(pc))
		goto unlock_out;

	switch (ctype) {
	case MEM_CGROUP_CHARGE_TYPE_MAPPED:
K
KAMEZAWA Hiroyuki 已提交
2939
	case MEM_CGROUP_CHARGE_TYPE_DROP:
2940 2941
		/* See mem_cgroup_prepare_migration() */
		if (page_mapped(page) || PageCgroupMigration(pc))
K
KAMEZAWA Hiroyuki 已提交
2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952
			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;
2953
	}
K
KAMEZAWA Hiroyuki 已提交
2954

2955
	mem_cgroup_charge_statistics(mem, PageCgroupCache(pc), -nr_pages);
K
KAMEZAWA Hiroyuki 已提交
2956

2957
	ClearPageCgroupUsed(pc);
2958 2959 2960 2961 2962 2963
	/*
	 * 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.
	 */
2964

2965
	unlock_page_cgroup(pc);
K
KAMEZAWA Hiroyuki 已提交
2966 2967 2968 2969
	/*
	 * even after unlock, we have mem->res.usage here and this memcg
	 * will never be freed.
	 */
2970
	memcg_check_events(mem, page);
K
KAMEZAWA Hiroyuki 已提交
2971 2972 2973 2974 2975
	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))
2976
		mem_cgroup_do_uncharge(mem, nr_pages, ctype);
2977

2978
	return mem;
K
KAMEZAWA Hiroyuki 已提交
2979 2980 2981

unlock_out:
	unlock_page_cgroup(pc);
2982
	return NULL;
2983 2984
}

2985 2986
void mem_cgroup_uncharge_page(struct page *page)
{
2987 2988 2989 2990 2991
	/* early check. */
	if (page_mapped(page))
		return;
	if (page->mapping && !PageAnon(page))
		return;
2992 2993 2994 2995 2996 2997
	__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));
2998
	VM_BUG_ON(page->mapping);
2999 3000 3001
	__mem_cgroup_uncharge_common(page, MEM_CGROUP_CHARGE_TYPE_CACHE);
}

3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015
/*
 * 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;
3016 3017
		current->memcg_batch.nr_pages = 0;
		current->memcg_batch.memsw_nr_pages = 0;
3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037
	}
}

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.
	 */
3038 3039 3040 3041 3042 3043
	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);
3044
	memcg_oom_recover(batch->memcg);
3045 3046 3047 3048
	/* forget this pointer (for sanity check) */
	batch->memcg = NULL;
}

3049
#ifdef CONFIG_SWAP
3050
/*
3051
 * called after __delete_from_swap_cache() and drop "page" account.
3052 3053
 * memcg information is recorded to swap_cgroup of "ent"
 */
K
KAMEZAWA Hiroyuki 已提交
3054 3055
void
mem_cgroup_uncharge_swapcache(struct page *page, swp_entry_t ent, bool swapout)
3056 3057
{
	struct mem_cgroup *memcg;
K
KAMEZAWA Hiroyuki 已提交
3058 3059 3060 3061 3062 3063
	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);
3064

K
KAMEZAWA Hiroyuki 已提交
3065 3066 3067 3068 3069
	/*
	 * record memcg information,  if swapout && memcg != NULL,
	 * mem_cgroup_get() was called in uncharge().
	 */
	if (do_swap_account && swapout && memcg)
3070
		swap_cgroup_record(ent, css_id(&memcg->css));
3071
}
3072
#endif
3073 3074 3075 3076 3077 3078 3079

#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
3080
{
3081
	struct mem_cgroup *memcg;
3082
	unsigned short id;
3083 3084 3085 3086

	if (!do_swap_account)
		return;

3087 3088 3089
	id = swap_cgroup_record(ent, 0);
	rcu_read_lock();
	memcg = mem_cgroup_lookup(id);
3090
	if (memcg) {
3091 3092 3093 3094
		/*
		 * We uncharge this because swap is freed.
		 * This memcg can be obsolete one. We avoid calling css_tryget
		 */
3095
		if (!mem_cgroup_is_root(memcg))
3096
			res_counter_uncharge(&memcg->memsw, PAGE_SIZE);
3097
		mem_cgroup_swap_statistics(memcg, false);
3098 3099
		mem_cgroup_put(memcg);
	}
3100
	rcu_read_unlock();
K
KAMEZAWA Hiroyuki 已提交
3101
}
3102 3103 3104 3105 3106 3107

/**
 * 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
3108
 * @need_fixup: whether we should fixup res_counters and refcounts.
3109 3110 3111 3112 3113 3114 3115 3116 3117 3118
 *
 * 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,
3119
		struct mem_cgroup *from, struct mem_cgroup *to, bool need_fixup)
3120 3121 3122 3123 3124 3125 3126 3127
{
	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);
3128
		mem_cgroup_swap_statistics(to, true);
3129
		/*
3130 3131 3132 3133 3134 3135
		 * 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.
3136 3137
		 */
		mem_cgroup_get(to);
3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148
		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);
		}
3149 3150 3151 3152 3153 3154
		return 0;
	}
	return -EINVAL;
}
#else
static inline int mem_cgroup_move_swap_account(swp_entry_t entry,
3155
		struct mem_cgroup *from, struct mem_cgroup *to, bool need_fixup)
3156 3157 3158
{
	return -EINVAL;
}
3159
#endif
K
KAMEZAWA Hiroyuki 已提交
3160

3161
/*
3162 3163
 * Before starting migration, account PAGE_SIZE to mem_cgroup that the old
 * page belongs to.
3164
 */
3165
int mem_cgroup_prepare_migration(struct page *page,
3166
	struct page *newpage, struct mem_cgroup **ptr, gfp_t gfp_mask)
3167
{
3168
	struct mem_cgroup *mem = NULL;
3169
	struct page_cgroup *pc;
3170
	enum charge_type ctype;
3171
	int ret = 0;
3172

3173 3174
	*ptr = NULL;

A
Andrea Arcangeli 已提交
3175
	VM_BUG_ON(PageTransHuge(page));
3176
	if (mem_cgroup_disabled())
3177 3178
		return 0;

3179 3180 3181
	pc = lookup_page_cgroup(page);
	lock_page_cgroup(pc);
	if (PageCgroupUsed(pc)) {
3182 3183
		mem = pc->mem_cgroup;
		css_get(&mem->css);
3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214
		/*
		 * 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);
3215
	}
3216
	unlock_page_cgroup(pc);
3217 3218 3219 3220 3221 3222
	/*
	 * If the page is not charged at this point,
	 * we return here.
	 */
	if (!mem)
		return 0;
3223

A
Andrea Arcangeli 已提交
3224
	*ptr = mem;
3225
	ret = __mem_cgroup_try_charge(NULL, gfp_mask, 1, ptr, false);
3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237
	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;
3238
	}
3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251
	/*
	 * 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;
3252
	__mem_cgroup_commit_charge(mem, page, 1, pc, ctype);
3253
	return ret;
3254
}
3255

3256
/* remove redundant charge if migration failed*/
3257
void mem_cgroup_end_migration(struct mem_cgroup *mem,
3258
	struct page *oldpage, struct page *newpage, bool migration_ok)
3259
{
3260
	struct page *used, *unused;
3261 3262 3263 3264
	struct page_cgroup *pc;

	if (!mem)
		return;
3265
	/* blocks rmdir() */
3266
	cgroup_exclude_rmdir(&mem->css);
3267
	if (!migration_ok) {
3268 3269
		used = oldpage;
		unused = newpage;
3270
	} else {
3271
		used = newpage;
3272 3273
		unused = oldpage;
	}
3274
	/*
3275 3276 3277
	 * 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.
3278
	 */
3279 3280 3281 3282
	pc = lookup_page_cgroup(oldpage);
	lock_page_cgroup(pc);
	ClearPageCgroupMigration(pc);
	unlock_page_cgroup(pc);
3283

3284 3285
	__mem_cgroup_uncharge_common(unused, MEM_CGROUP_CHARGE_TYPE_FORCE);

3286
	/*
3287 3288 3289 3290 3291 3292
	 * 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)
3293
	 */
3294 3295
	if (PageAnon(used))
		mem_cgroup_uncharge_page(used);
3296
	/*
3297 3298
	 * At migration, we may charge account against cgroup which has no
	 * tasks.
3299 3300 3301 3302
	 * 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);
3303
}
3304

3305
/*
3306 3307 3308 3309 3310 3311
 * A call to try to shrink memory usage on charge failure at shmem's swapin.
 * Calling hierarchical_reclaim is not enough because we should update
 * last_oom_jiffies to prevent pagefault_out_of_memory from invoking global OOM.
 * Moreover considering hierarchy, we should reclaim from the mem_over_limit,
 * not from the memcg which this page would be charged to.
 * try_charge_swapin does all of these works properly.
3312
 */
3313
int mem_cgroup_shmem_charge_fallback(struct page *page,
3314 3315
			    struct mm_struct *mm,
			    gfp_t gfp_mask)
3316
{
3317
	struct mem_cgroup *mem;
3318
	int ret;
3319

3320
	if (mem_cgroup_disabled())
3321
		return 0;
3322

3323 3324 3325
	ret = mem_cgroup_try_charge_swapin(mm, page, gfp_mask, &mem);
	if (!ret)
		mem_cgroup_cancel_charge_swapin(mem); /* it does !mem check */
3326

3327
	return ret;
3328 3329
}

3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 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
#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

3376 3377
static DEFINE_MUTEX(set_limit_mutex);

3378
static int mem_cgroup_resize_limit(struct mem_cgroup *memcg,
3379
				unsigned long long val)
3380
{
3381
	int retry_count;
3382
	u64 memswlimit, memlimit;
3383
	int ret = 0;
3384 3385
	int children = mem_cgroup_count_children(memcg);
	u64 curusage, oldusage;
3386
	int enlarge;
3387 3388 3389 3390 3391 3392 3393 3394 3395

	/*
	 * 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);
3396

3397
	enlarge = 0;
3398
	while (retry_count) {
3399 3400 3401 3402
		if (signal_pending(current)) {
			ret = -EINTR;
			break;
		}
3403 3404 3405 3406 3407 3408 3409 3410 3411 3412
		/*
		 * 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);
3413 3414
			break;
		}
3415 3416 3417 3418 3419

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

3420
		ret = res_counter_set_limit(&memcg->res, val);
3421 3422 3423 3424 3425 3426
		if (!ret) {
			if (memswlimit == val)
				memcg->memsw_is_minimum = true;
			else
				memcg->memsw_is_minimum = false;
		}
3427 3428 3429 3430 3431
		mutex_unlock(&set_limit_mutex);

		if (!ret)
			break;

3432
		mem_cgroup_hierarchical_reclaim(memcg, NULL, GFP_KERNEL,
3433 3434
						MEM_CGROUP_RECLAIM_SHRINK,
						NULL);
3435 3436 3437 3438 3439 3440
		curusage = res_counter_read_u64(&memcg->res, RES_USAGE);
		/* Usage is reduced ? */
  		if (curusage >= oldusage)
			retry_count--;
		else
			oldusage = curusage;
3441
	}
3442 3443
	if (!ret && enlarge)
		memcg_oom_recover(memcg);
3444

3445 3446 3447
	return ret;
}

L
Li Zefan 已提交
3448 3449
static int mem_cgroup_resize_memsw_limit(struct mem_cgroup *memcg,
					unsigned long long val)
3450
{
3451
	int retry_count;
3452
	u64 memlimit, memswlimit, oldusage, curusage;
3453 3454
	int children = mem_cgroup_count_children(memcg);
	int ret = -EBUSY;
3455
	int enlarge = 0;
3456

3457 3458 3459
	/* see mem_cgroup_resize_res_limit */
 	retry_count = children * MEM_CGROUP_RECLAIM_RETRIES;
	oldusage = res_counter_read_u64(&memcg->memsw, RES_USAGE);
3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476
	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;
		}
3477 3478 3479
		memswlimit = res_counter_read_u64(&memcg->memsw, RES_LIMIT);
		if (memswlimit < val)
			enlarge = 1;
3480
		ret = res_counter_set_limit(&memcg->memsw, val);
3481 3482 3483 3484 3485 3486
		if (!ret) {
			if (memlimit == val)
				memcg->memsw_is_minimum = true;
			else
				memcg->memsw_is_minimum = false;
		}
3487 3488 3489 3490 3491
		mutex_unlock(&set_limit_mutex);

		if (!ret)
			break;

3492
		mem_cgroup_hierarchical_reclaim(memcg, NULL, GFP_KERNEL,
3493
						MEM_CGROUP_RECLAIM_NOSWAP |
3494 3495
						MEM_CGROUP_RECLAIM_SHRINK,
						NULL);
3496
		curusage = res_counter_read_u64(&memcg->memsw, RES_USAGE);
3497
		/* Usage is reduced ? */
3498
		if (curusage >= oldusage)
3499
			retry_count--;
3500 3501
		else
			oldusage = curusage;
3502
	}
3503 3504
	if (!ret && enlarge)
		memcg_oom_recover(memcg);
3505 3506 3507
	return ret;
}

3508
unsigned long mem_cgroup_soft_limit_reclaim(struct zone *zone, int order,
3509 3510
					    gfp_t gfp_mask,
					    unsigned long *total_scanned)
3511 3512 3513 3514 3515 3516
{
	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;
3517
	unsigned long long excess;
3518
	unsigned long nr_scanned;
3519 3520 3521 3522

	if (order > 0)
		return 0;

3523
	mctz = soft_limit_tree_node_zone(zone_to_nid(zone), zone_idx(zone));
3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536
	/*
	 * 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;

3537
		nr_scanned = 0;
3538 3539
		reclaimed = mem_cgroup_hierarchical_reclaim(mz->mem, zone,
						gfp_mask,
3540 3541
						MEM_CGROUP_RECLAIM_SOFT,
						&nr_scanned);
3542
		nr_reclaimed += reclaimed;
3543
		*total_scanned += nr_scanned;
3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565
		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);
3566
				if (next_mz == mz)
3567
					css_put(&next_mz->mem->css);
3568
				else /* next_mz == NULL or other memcg */
3569 3570 3571 3572
					break;
			} while (1);
		}
		__mem_cgroup_remove_exceeded(mz->mem, mz, mctz);
3573
		excess = res_counter_soft_limit_excess(&mz->mem->res);
3574 3575 3576 3577 3578 3579 3580 3581
		/*
		 * 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.
		 */
3582 3583
		/* If excess == 0, no tree ops */
		__mem_cgroup_insert_exceeded(mz->mem, mz, mctz, excess);
3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601
		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;
}

3602 3603 3604 3605
/*
 * This routine traverse page_cgroup in given list and drop them all.
 * *And* this routine doesn't reclaim page itself, just removes page_cgroup.
 */
3606
static int mem_cgroup_force_empty_list(struct mem_cgroup *mem,
K
KAMEZAWA Hiroyuki 已提交
3607
				int node, int zid, enum lru_list lru)
3608
{
K
KAMEZAWA Hiroyuki 已提交
3609 3610
	struct zone *zone;
	struct mem_cgroup_per_zone *mz;
3611
	struct page_cgroup *pc, *busy;
K
KAMEZAWA Hiroyuki 已提交
3612
	unsigned long flags, loop;
3613
	struct list_head *list;
3614
	int ret = 0;
3615

K
KAMEZAWA Hiroyuki 已提交
3616 3617
	zone = &NODE_DATA(node)->node_zones[zid];
	mz = mem_cgroup_zoneinfo(mem, node, zid);
3618
	list = &mz->lists[lru];
3619

3620 3621 3622 3623 3624
	loop = MEM_CGROUP_ZSTAT(mz, lru);
	/* give some margin against EBUSY etc...*/
	loop += 256;
	busy = NULL;
	while (loop--) {
3625 3626
		struct page *page;

3627
		ret = 0;
K
KAMEZAWA Hiroyuki 已提交
3628
		spin_lock_irqsave(&zone->lru_lock, flags);
3629
		if (list_empty(list)) {
K
KAMEZAWA Hiroyuki 已提交
3630
			spin_unlock_irqrestore(&zone->lru_lock, flags);
3631
			break;
3632 3633 3634 3635
		}
		pc = list_entry(list->prev, struct page_cgroup, lru);
		if (busy == pc) {
			list_move(&pc->lru, list);
3636
			busy = NULL;
K
KAMEZAWA Hiroyuki 已提交
3637
			spin_unlock_irqrestore(&zone->lru_lock, flags);
3638 3639
			continue;
		}
K
KAMEZAWA Hiroyuki 已提交
3640
		spin_unlock_irqrestore(&zone->lru_lock, flags);
3641

3642
		page = lookup_cgroup_page(pc);
3643 3644

		ret = mem_cgroup_move_parent(page, pc, mem, GFP_KERNEL);
3645
		if (ret == -ENOMEM)
3646
			break;
3647 3648 3649 3650 3651 3652 3653

		if (ret == -EBUSY || ret == -EINVAL) {
			/* found lock contention or "pc" is obsolete. */
			busy = pc;
			cond_resched();
		} else
			busy = NULL;
3654
	}
K
KAMEZAWA Hiroyuki 已提交
3655

3656 3657 3658
	if (!ret && !list_empty(list))
		return -EBUSY;
	return ret;
3659 3660 3661 3662 3663 3664
}

/*
 * make mem_cgroup's charge to be 0 if there is no task.
 * This enables deleting this mem_cgroup.
 */
3665
static int mem_cgroup_force_empty(struct mem_cgroup *mem, bool free_all)
3666
{
3667 3668 3669
	int ret;
	int node, zid, shrink;
	int nr_retries = MEM_CGROUP_RECLAIM_RETRIES;
3670
	struct cgroup *cgrp = mem->css.cgroup;
3671

3672
	css_get(&mem->css);
3673 3674

	shrink = 0;
3675 3676 3677
	/* should free all ? */
	if (free_all)
		goto try_to_free;
3678
move_account:
3679
	do {
3680
		ret = -EBUSY;
3681 3682 3683 3684
		if (cgroup_task_count(cgrp) || !list_empty(&cgrp->children))
			goto out;
		ret = -EINTR;
		if (signal_pending(current))
3685
			goto out;
3686 3687
		/* This is for making all *used* pages to be on LRU. */
		lru_add_drain_all();
3688
		drain_all_stock_sync();
3689
		ret = 0;
3690
		mem_cgroup_start_move(mem);
3691
		for_each_node_state(node, N_HIGH_MEMORY) {
3692
			for (zid = 0; !ret && zid < MAX_NR_ZONES; zid++) {
3693
				enum lru_list l;
3694 3695
				for_each_lru(l) {
					ret = mem_cgroup_force_empty_list(mem,
K
KAMEZAWA Hiroyuki 已提交
3696
							node, zid, l);
3697 3698 3699
					if (ret)
						break;
				}
3700
			}
3701 3702 3703
			if (ret)
				break;
		}
3704
		mem_cgroup_end_move(mem);
3705
		memcg_oom_recover(mem);
3706 3707 3708
		/* it seems parent cgroup doesn't have enough mem */
		if (ret == -ENOMEM)
			goto try_to_free;
3709
		cond_resched();
3710 3711
	/* "ret" should also be checked to ensure all lists are empty. */
	} while (mem->res.usage > 0 || ret);
3712 3713 3714
out:
	css_put(&mem->css);
	return ret;
3715 3716

try_to_free:
3717 3718
	/* returns EBUSY if there is a task or if we come here twice. */
	if (cgroup_task_count(cgrp) || !list_empty(&cgrp->children) || shrink) {
3719 3720 3721
		ret = -EBUSY;
		goto out;
	}
3722 3723
	/* we call try-to-free pages for make this cgroup empty */
	lru_add_drain_all();
3724 3725 3726 3727
	/* try to free all pages in this cgroup */
	shrink = 1;
	while (nr_retries && mem->res.usage > 0) {
		int progress;
3728 3729 3730 3731 3732

		if (signal_pending(current)) {
			ret = -EINTR;
			goto out;
		}
K
KOSAKI Motohiro 已提交
3733 3734
		progress = try_to_free_mem_cgroup_pages(mem, GFP_KERNEL,
						false, get_swappiness(mem));
3735
		if (!progress) {
3736
			nr_retries--;
3737
			/* maybe some writeback is necessary */
3738
			congestion_wait(BLK_RW_ASYNC, HZ/10);
3739
		}
3740 3741

	}
K
KAMEZAWA Hiroyuki 已提交
3742
	lru_add_drain();
3743
	/* try move_account...there may be some *locked* pages. */
3744
	goto move_account;
3745 3746
}

3747 3748 3749 3750 3751 3752
int mem_cgroup_force_empty_write(struct cgroup *cont, unsigned int event)
{
	return mem_cgroup_force_empty(mem_cgroup_from_cont(cont), true);
}


3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770
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();
	/*
3771
	 * If parent's use_hierarchy is set, we can't make any modifications
3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790
	 * 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;
}

3791

3792 3793
static unsigned long mem_cgroup_recursive_stat(struct mem_cgroup *mem,
					       enum mem_cgroup_stat_index idx)
3794
{
K
KAMEZAWA Hiroyuki 已提交
3795
	struct mem_cgroup *iter;
3796
	long val = 0;
3797

3798
	/* Per-cpu values can be negative, use a signed accumulator */
K
KAMEZAWA Hiroyuki 已提交
3799 3800 3801 3802 3803 3804
	for_each_mem_cgroup_tree(iter, mem)
		val += mem_cgroup_read_stat(iter, idx);

	if (val < 0) /* race ? */
		val = 0;
	return val;
3805 3806
}

3807 3808
static inline u64 mem_cgroup_usage(struct mem_cgroup *mem, bool swap)
{
K
KAMEZAWA Hiroyuki 已提交
3809
	u64 val;
3810 3811 3812 3813 3814 3815 3816 3817

	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);
	}

3818 3819
	val = mem_cgroup_recursive_stat(mem, MEM_CGROUP_STAT_CACHE);
	val += mem_cgroup_recursive_stat(mem, MEM_CGROUP_STAT_RSS);
3820

K
KAMEZAWA Hiroyuki 已提交
3821
	if (swap)
3822
		val += mem_cgroup_recursive_stat(mem, MEM_CGROUP_STAT_SWAPOUT);
3823 3824 3825 3826

	return val << PAGE_SHIFT;
}

3827
static u64 mem_cgroup_read(struct cgroup *cont, struct cftype *cft)
B
Balbir Singh 已提交
3828
{
3829
	struct mem_cgroup *mem = mem_cgroup_from_cont(cont);
3830
	u64 val;
3831 3832 3833 3834 3835 3836
	int type, name;

	type = MEMFILE_TYPE(cft->private);
	name = MEMFILE_ATTR(cft->private);
	switch (type) {
	case _MEM:
3837 3838 3839
		if (name == RES_USAGE)
			val = mem_cgroup_usage(mem, false);
		else
3840
			val = res_counter_read_u64(&mem->res, name);
3841 3842
		break;
	case _MEMSWAP:
3843 3844 3845
		if (name == RES_USAGE)
			val = mem_cgroup_usage(mem, true);
		else
3846
			val = res_counter_read_u64(&mem->memsw, name);
3847 3848 3849 3850 3851 3852
		break;
	default:
		BUG();
		break;
	}
	return val;
B
Balbir Singh 已提交
3853
}
3854 3855 3856 3857
/*
 * The user of this function is...
 * RES_LIMIT.
 */
3858 3859
static int mem_cgroup_write(struct cgroup *cont, struct cftype *cft,
			    const char *buffer)
B
Balbir Singh 已提交
3860
{
3861
	struct mem_cgroup *memcg = mem_cgroup_from_cont(cont);
3862
	int type, name;
3863 3864 3865
	unsigned long long val;
	int ret;

3866 3867 3868
	type = MEMFILE_TYPE(cft->private);
	name = MEMFILE_ATTR(cft->private);
	switch (name) {
3869
	case RES_LIMIT:
3870 3871 3872 3873
		if (mem_cgroup_is_root(memcg)) { /* Can't set limit on root */
			ret = -EINVAL;
			break;
		}
3874 3875
		/* This function does all necessary parse...reuse it */
		ret = res_counter_memparse_write_strategy(buffer, &val);
3876 3877 3878
		if (ret)
			break;
		if (type == _MEM)
3879
			ret = mem_cgroup_resize_limit(memcg, val);
3880 3881
		else
			ret = mem_cgroup_resize_memsw_limit(memcg, val);
3882
		break;
3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896
	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;
3897 3898 3899 3900 3901
	default:
		ret = -EINVAL; /* should be BUG() ? */
		break;
	}
	return ret;
B
Balbir Singh 已提交
3902 3903
}

3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931
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;
}

3932
static int mem_cgroup_reset(struct cgroup *cont, unsigned int event)
3933 3934
{
	struct mem_cgroup *mem;
3935
	int type, name;
3936 3937

	mem = mem_cgroup_from_cont(cont);
3938 3939 3940
	type = MEMFILE_TYPE(event);
	name = MEMFILE_ATTR(event);
	switch (name) {
3941
	case RES_MAX_USAGE:
3942 3943 3944 3945
		if (type == _MEM)
			res_counter_reset_max(&mem->res);
		else
			res_counter_reset_max(&mem->memsw);
3946 3947
		break;
	case RES_FAILCNT:
3948 3949 3950 3951
		if (type == _MEM)
			res_counter_reset_failcnt(&mem->res);
		else
			res_counter_reset_failcnt(&mem->memsw);
3952 3953
		break;
	}
3954

3955
	return 0;
3956 3957
}

3958 3959 3960 3961 3962 3963
static u64 mem_cgroup_move_charge_read(struct cgroup *cgrp,
					struct cftype *cft)
{
	return mem_cgroup_from_cont(cgrp)->move_charge_at_immigrate;
}

3964
#ifdef CONFIG_MMU
3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982
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;
}
3983 3984 3985 3986 3987 3988 3989
#else
static int mem_cgroup_move_charge_write(struct cgroup *cgrp,
					struct cftype *cft, u64 val)
{
	return -ENOSYS;
}
#endif
3990

K
KAMEZAWA Hiroyuki 已提交
3991 3992 3993 3994 3995

/* For read statistics */
enum {
	MCS_CACHE,
	MCS_RSS,
3996
	MCS_FILE_MAPPED,
K
KAMEZAWA Hiroyuki 已提交
3997 3998
	MCS_PGPGIN,
	MCS_PGPGOUT,
3999
	MCS_SWAP,
4000 4001
	MCS_PGFAULT,
	MCS_PGMAJFAULT,
K
KAMEZAWA Hiroyuki 已提交
4002 4003 4004 4005 4006 4007 4008 4009 4010 4011
	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];
4012 4013
};

K
KAMEZAWA Hiroyuki 已提交
4014 4015 4016 4017 4018 4019
struct {
	char *local_name;
	char *total_name;
} memcg_stat_strings[NR_MCS_STAT] = {
	{"cache", "total_cache"},
	{"rss", "total_rss"},
4020
	{"mapped_file", "total_mapped_file"},
K
KAMEZAWA Hiroyuki 已提交
4021 4022
	{"pgpgin", "total_pgpgin"},
	{"pgpgout", "total_pgpgout"},
4023
	{"swap", "total_swap"},
4024 4025
	{"pgfault", "total_pgfault"},
	{"pgmajfault", "total_pgmajfault"},
K
KAMEZAWA Hiroyuki 已提交
4026 4027 4028 4029 4030 4031 4032 4033
	{"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
4034 4035
static void
mem_cgroup_get_local_stat(struct mem_cgroup *mem, struct mcs_total_stat *s)
K
KAMEZAWA Hiroyuki 已提交
4036 4037 4038 4039
{
	s64 val;

	/* per cpu stat */
4040
	val = mem_cgroup_read_stat(mem, MEM_CGROUP_STAT_CACHE);
K
KAMEZAWA Hiroyuki 已提交
4041
	s->stat[MCS_CACHE] += val * PAGE_SIZE;
4042
	val = mem_cgroup_read_stat(mem, MEM_CGROUP_STAT_RSS);
K
KAMEZAWA Hiroyuki 已提交
4043
	s->stat[MCS_RSS] += val * PAGE_SIZE;
4044
	val = mem_cgroup_read_stat(mem, MEM_CGROUP_STAT_FILE_MAPPED);
4045
	s->stat[MCS_FILE_MAPPED] += val * PAGE_SIZE;
4046
	val = mem_cgroup_read_events(mem, MEM_CGROUP_EVENTS_PGPGIN);
K
KAMEZAWA Hiroyuki 已提交
4047
	s->stat[MCS_PGPGIN] += val;
4048
	val = mem_cgroup_read_events(mem, MEM_CGROUP_EVENTS_PGPGOUT);
K
KAMEZAWA Hiroyuki 已提交
4049
	s->stat[MCS_PGPGOUT] += val;
4050
	if (do_swap_account) {
4051
		val = mem_cgroup_read_stat(mem, MEM_CGROUP_STAT_SWAPOUT);
4052 4053
		s->stat[MCS_SWAP] += val * PAGE_SIZE;
	}
4054 4055 4056 4057
	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
4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074

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

static void
mem_cgroup_get_total_stat(struct mem_cgroup *mem, struct mcs_total_stat *s)
{
K
KAMEZAWA Hiroyuki 已提交
4075 4076 4077 4078
	struct mem_cgroup *iter;

	for_each_mem_cgroup_tree(iter, mem)
		mem_cgroup_get_local_stat(iter, s);
K
KAMEZAWA Hiroyuki 已提交
4079 4080
}

4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125
#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);

	total_nr = mem_cgroup_nr_lru_pages(mem_cont);
	seq_printf(m, "total=%lu", total_nr);
	for_each_node_state(nid, N_HIGH_MEMORY) {
		node_nr = mem_cgroup_node_nr_lru_pages(mem_cont, nid);
		seq_printf(m, " N%d=%lu", nid, node_nr);
	}
	seq_putc(m, '\n');

	file_nr = mem_cgroup_nr_file_lru_pages(mem_cont);
	seq_printf(m, "file=%lu", file_nr);
	for_each_node_state(nid, N_HIGH_MEMORY) {
		node_nr = mem_cgroup_node_nr_file_lru_pages(mem_cont, nid);
		seq_printf(m, " N%d=%lu", nid, node_nr);
	}
	seq_putc(m, '\n');

	anon_nr = mem_cgroup_nr_anon_lru_pages(mem_cont);
	seq_printf(m, "anon=%lu", anon_nr);
	for_each_node_state(nid, N_HIGH_MEMORY) {
		node_nr = mem_cgroup_node_nr_anon_lru_pages(mem_cont, nid);
		seq_printf(m, " N%d=%lu", nid, node_nr);
	}
	seq_putc(m, '\n');

	unevictable_nr = mem_cgroup_nr_unevictable_lru_pages(mem_cont);
	seq_printf(m, "unevictable=%lu", unevictable_nr);
	for_each_node_state(nid, N_HIGH_MEMORY) {
		node_nr = mem_cgroup_node_nr_unevictable_lru_pages(mem_cont,
									nid);
		seq_printf(m, " N%d=%lu", nid, node_nr);
	}
	seq_putc(m, '\n');
	return 0;
}
#endif /* CONFIG_NUMA */

4126 4127
static int mem_control_stat_show(struct cgroup *cont, struct cftype *cft,
				 struct cgroup_map_cb *cb)
4128 4129
{
	struct mem_cgroup *mem_cont = mem_cgroup_from_cont(cont);
K
KAMEZAWA Hiroyuki 已提交
4130
	struct mcs_total_stat mystat;
4131 4132
	int i;

K
KAMEZAWA Hiroyuki 已提交
4133 4134
	memset(&mystat, 0, sizeof(mystat));
	mem_cgroup_get_local_stat(mem_cont, &mystat);
4135

4136

4137 4138 4139
	for (i = 0; i < NR_MCS_STAT; i++) {
		if (i == MCS_SWAP && !do_swap_account)
			continue;
K
KAMEZAWA Hiroyuki 已提交
4140
		cb->fill(cb, memcg_stat_strings[i].local_name, mystat.stat[i]);
4141
	}
L
Lee Schermerhorn 已提交
4142

K
KAMEZAWA Hiroyuki 已提交
4143
	/* Hierarchical information */
4144 4145 4146 4147 4148 4149 4150
	{
		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
4151

K
KAMEZAWA Hiroyuki 已提交
4152 4153
	memset(&mystat, 0, sizeof(mystat));
	mem_cgroup_get_total_stat(mem_cont, &mystat);
4154 4155 4156
	for (i = 0; i < NR_MCS_STAT; i++) {
		if (i == MCS_SWAP && !do_swap_account)
			continue;
K
KAMEZAWA Hiroyuki 已提交
4157
		cb->fill(cb, memcg_stat_strings[i].total_name, mystat.stat[i]);
4158
	}
K
KAMEZAWA Hiroyuki 已提交
4159

K
KOSAKI Motohiro 已提交
4160
#ifdef CONFIG_DEBUG_VM
4161
	cb->fill(cb, "inactive_ratio", calc_inactive_ratio(mem_cont, NULL));
K
KOSAKI Motohiro 已提交
4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186 4187 4188

	{
		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

4189 4190 4191
	return 0;
}

K
KOSAKI Motohiro 已提交
4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203
static u64 mem_cgroup_swappiness_read(struct cgroup *cgrp, struct cftype *cft)
{
	struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp);

	return get_swappiness(memcg);
}

static int mem_cgroup_swappiness_write(struct cgroup *cgrp, struct cftype *cft,
				       u64 val)
{
	struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp);
	struct mem_cgroup *parent;
4204

K
KOSAKI Motohiro 已提交
4205 4206 4207 4208 4209 4210 4211
	if (val > 100)
		return -EINVAL;

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

	parent = mem_cgroup_from_cont(cgrp->parent);
4212 4213 4214

	cgroup_lock();

K
KOSAKI Motohiro 已提交
4215 4216
	/* If under hierarchy, only empty-root can set this value */
	if ((parent->use_hierarchy) ||
4217 4218
	    (memcg->use_hierarchy && !list_empty(&cgrp->children))) {
		cgroup_unlock();
K
KOSAKI Motohiro 已提交
4219
		return -EINVAL;
4220
	}
K
KOSAKI Motohiro 已提交
4221 4222 4223

	memcg->swappiness = val;

4224 4225
	cgroup_unlock();

K
KOSAKI Motohiro 已提交
4226 4227 4228
	return 0;
}

4229 4230 4231 4232 4233 4234 4235 4236
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)
4237
		t = rcu_dereference(memcg->thresholds.primary);
4238
	else
4239
		t = rcu_dereference(memcg->memsw_thresholds.primary);
4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250

	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().
	 */
4251
	i = t->current_threshold;
4252 4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270 4271 4272 4273 4274

	/*
	 * 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 */
4275
	t->current_threshold = i - 1;
4276 4277 4278 4279 4280 4281
unlock:
	rcu_read_unlock();
}

static void mem_cgroup_threshold(struct mem_cgroup *memcg)
{
4282 4283 4284 4285 4286 4287 4288
	while (memcg) {
		__mem_cgroup_threshold(memcg, false);
		if (do_swap_account)
			__mem_cgroup_threshold(memcg, true);

		memcg = parent_mem_cgroup(memcg);
	}
4289 4290 4291 4292 4293 4294 4295 4296 4297 4298
}

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
4299
static int mem_cgroup_oom_notify_cb(struct mem_cgroup *mem)
K
KAMEZAWA Hiroyuki 已提交
4300 4301 4302 4303 4304 4305 4306 4307 4308 4309
{
	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
4310 4311 4312 4313
	struct mem_cgroup *iter;

	for_each_mem_cgroup_tree(iter, mem)
		mem_cgroup_oom_notify_cb(iter);
K
KAMEZAWA Hiroyuki 已提交
4314 4315 4316 4317
}

static int mem_cgroup_usage_register_event(struct cgroup *cgrp,
	struct cftype *cft, struct eventfd_ctx *eventfd, const char *args)
4318 4319
{
	struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp);
4320 4321
	struct mem_cgroup_thresholds *thresholds;
	struct mem_cgroup_threshold_ary *new;
4322 4323
	int type = MEMFILE_TYPE(cft->private);
	u64 threshold, usage;
4324
	int i, size, ret;
4325 4326 4327 4328 4329 4330

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

	mutex_lock(&memcg->thresholds_lock);
4331

4332
	if (type == _MEM)
4333
		thresholds = &memcg->thresholds;
4334
	else if (type == _MEMSWAP)
4335
		thresholds = &memcg->memsw_thresholds;
4336 4337 4338 4339 4340 4341
	else
		BUG();

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

	/* Check if a threshold crossed before adding a new one */
4342
	if (thresholds->primary)
4343 4344
		__mem_cgroup_threshold(memcg, type == _MEMSWAP);

4345
	size = thresholds->primary ? thresholds->primary->size + 1 : 1;
4346 4347

	/* Allocate memory for new array of thresholds */
4348
	new = kmalloc(sizeof(*new) + size * sizeof(struct mem_cgroup_threshold),
4349
			GFP_KERNEL);
4350
	if (!new) {
4351 4352 4353
		ret = -ENOMEM;
		goto unlock;
	}
4354
	new->size = size;
4355 4356

	/* Copy thresholds (if any) to new array */
4357 4358
	if (thresholds->primary) {
		memcpy(new->entries, thresholds->primary->entries, (size - 1) *
4359
				sizeof(struct mem_cgroup_threshold));
4360 4361
	}

4362
	/* Add new threshold */
4363 4364
	new->entries[size - 1].eventfd = eventfd;
	new->entries[size - 1].threshold = threshold;
4365 4366

	/* Sort thresholds. Registering of new threshold isn't time-critical */
4367
	sort(new->entries, size, sizeof(struct mem_cgroup_threshold),
4368 4369 4370
			compare_thresholds, NULL);

	/* Find current threshold */
4371
	new->current_threshold = -1;
4372
	for (i = 0; i < size; i++) {
4373
		if (new->entries[i].threshold < usage) {
4374
			/*
4375 4376
			 * new->current_threshold will not be used until
			 * rcu_assign_pointer(), so it's safe to increment
4377 4378
			 * it here.
			 */
4379
			++new->current_threshold;
4380 4381 4382
		}
	}

4383 4384 4385 4386 4387
	/* Free old spare buffer and save old primary buffer as spare */
	kfree(thresholds->spare);
	thresholds->spare = thresholds->primary;

	rcu_assign_pointer(thresholds->primary, new);
4388

4389
	/* To be sure that nobody uses thresholds */
4390 4391 4392 4393 4394 4395 4396 4397
	synchronize_rcu();

unlock:
	mutex_unlock(&memcg->thresholds_lock);

	return ret;
}

4398
static void mem_cgroup_usage_unregister_event(struct cgroup *cgrp,
K
KAMEZAWA Hiroyuki 已提交
4399
	struct cftype *cft, struct eventfd_ctx *eventfd)
4400 4401
{
	struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp);
4402 4403
	struct mem_cgroup_thresholds *thresholds;
	struct mem_cgroup_threshold_ary *new;
4404 4405
	int type = MEMFILE_TYPE(cft->private);
	u64 usage;
4406
	int i, j, size;
4407 4408 4409

	mutex_lock(&memcg->thresholds_lock);
	if (type == _MEM)
4410
		thresholds = &memcg->thresholds;
4411
	else if (type == _MEMSWAP)
4412
		thresholds = &memcg->memsw_thresholds;
4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427
	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 */
4428 4429 4430
	size = 0;
	for (i = 0; i < thresholds->primary->size; i++) {
		if (thresholds->primary->entries[i].eventfd != eventfd)
4431 4432 4433
			size++;
	}

4434
	new = thresholds->spare;
4435

4436 4437
	/* Set thresholds array to NULL if we don't have thresholds */
	if (!size) {
4438 4439
		kfree(new);
		new = NULL;
4440
		goto swap_buffers;
4441 4442
	}

4443
	new->size = size;
4444 4445

	/* Copy thresholds and find current threshold */
4446 4447 4448
	new->current_threshold = -1;
	for (i = 0, j = 0; i < thresholds->primary->size; i++) {
		if (thresholds->primary->entries[i].eventfd == eventfd)
4449 4450
			continue;

4451 4452
		new->entries[j] = thresholds->primary->entries[i];
		if (new->entries[j].threshold < usage) {
4453
			/*
4454
			 * new->current_threshold will not be used
4455 4456 4457
			 * until rcu_assign_pointer(), so it's safe to increment
			 * it here.
			 */
4458
			++new->current_threshold;
4459 4460 4461 4462
		}
		j++;
	}

4463
swap_buffers:
4464 4465 4466
	/* Swap primary and spare array */
	thresholds->spare = thresholds->primary;
	rcu_assign_pointer(thresholds->primary, new);
4467

4468
	/* To be sure that nobody uses thresholds */
4469 4470 4471 4472
	synchronize_rcu();

	mutex_unlock(&memcg->thresholds_lock);
}
4473

K
KAMEZAWA Hiroyuki 已提交
4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498
static int mem_cgroup_oom_register_event(struct cgroup *cgrp,
	struct cftype *cft, struct eventfd_ctx *eventfd, const char *args)
{
	struct mem_cgroup *memcg = mem_cgroup_from_cont(cgrp);
	struct mem_cgroup_eventfd_list *event;
	int type = MEMFILE_TYPE(cft->private);

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

	mutex_lock(&memcg_oom_mutex);

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

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

	return 0;
}

4499
static void mem_cgroup_oom_unregister_event(struct cgroup *cgrp,
K
KAMEZAWA Hiroyuki 已提交
4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512 4513 4514 4515 4516 4517 4518 4519
	struct cftype *cft, struct eventfd_ctx *eventfd)
{
	struct mem_cgroup *mem = mem_cgroup_from_cont(cgrp);
	struct mem_cgroup_eventfd_list *ev, *tmp;
	int type = MEMFILE_TYPE(cft->private);

	BUG_ON(type != _OOM_TYPE);

	mutex_lock(&memcg_oom_mutex);

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

	mutex_unlock(&memcg_oom_mutex);
}

4520 4521 4522 4523 4524 4525 4526 4527 4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553
static int mem_cgroup_oom_control_read(struct cgroup *cgrp,
	struct cftype *cft,  struct cgroup_map_cb *cb)
{
	struct mem_cgroup *mem = mem_cgroup_from_cont(cgrp);

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

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

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

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

	parent = mem_cgroup_from_cont(cgrp->parent);

	cgroup_lock();
	/* oom-kill-disable is a flag for subhierarchy. */
	if ((parent->use_hierarchy) ||
	    (mem->use_hierarchy && !list_empty(&cgrp->children))) {
		cgroup_unlock();
		return -EINVAL;
	}
	mem->oom_kill_disable = val;
4554 4555
	if (!val)
		memcg_oom_recover(mem);
4556 4557 4558 4559
	cgroup_unlock();
	return 0;
}

4560 4561 4562 4563 4564 4565 4566 4567 4568 4569 4570 4571 4572 4573 4574 4575
#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 */

B
Balbir Singh 已提交
4576 4577
static struct cftype mem_cgroup_files[] = {
	{
4578
		.name = "usage_in_bytes",
4579
		.private = MEMFILE_PRIVATE(_MEM, RES_USAGE),
4580
		.read_u64 = mem_cgroup_read,
K
KAMEZAWA Hiroyuki 已提交
4581 4582
		.register_event = mem_cgroup_usage_register_event,
		.unregister_event = mem_cgroup_usage_unregister_event,
B
Balbir Singh 已提交
4583
	},
4584 4585
	{
		.name = "max_usage_in_bytes",
4586
		.private = MEMFILE_PRIVATE(_MEM, RES_MAX_USAGE),
4587
		.trigger = mem_cgroup_reset,
4588 4589
		.read_u64 = mem_cgroup_read,
	},
B
Balbir Singh 已提交
4590
	{
4591
		.name = "limit_in_bytes",
4592
		.private = MEMFILE_PRIVATE(_MEM, RES_LIMIT),
4593
		.write_string = mem_cgroup_write,
4594
		.read_u64 = mem_cgroup_read,
B
Balbir Singh 已提交
4595
	},
4596 4597 4598 4599 4600 4601
	{
		.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
4602 4603
	{
		.name = "failcnt",
4604
		.private = MEMFILE_PRIVATE(_MEM, RES_FAILCNT),
4605
		.trigger = mem_cgroup_reset,
4606
		.read_u64 = mem_cgroup_read,
B
Balbir Singh 已提交
4607
	},
4608 4609
	{
		.name = "stat",
4610
		.read_map = mem_control_stat_show,
4611
	},
4612 4613 4614 4615
	{
		.name = "force_empty",
		.trigger = mem_cgroup_force_empty_write,
	},
4616 4617 4618 4619 4620
	{
		.name = "use_hierarchy",
		.write_u64 = mem_cgroup_hierarchy_write,
		.read_u64 = mem_cgroup_hierarchy_read,
	},
K
KOSAKI Motohiro 已提交
4621 4622 4623 4624 4625
	{
		.name = "swappiness",
		.read_u64 = mem_cgroup_swappiness_read,
		.write_u64 = mem_cgroup_swappiness_write,
	},
4626 4627 4628 4629 4630
	{
		.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
4631 4632
	{
		.name = "oom_control",
4633 4634
		.read_map = mem_cgroup_oom_control_read,
		.write_u64 = mem_cgroup_oom_control_write,
K
KAMEZAWA Hiroyuki 已提交
4635 4636 4637 4638
		.register_event = mem_cgroup_oom_register_event,
		.unregister_event = mem_cgroup_oom_unregister_event,
		.private = MEMFILE_PRIVATE(_OOM_TYPE, OOM_CONTROL),
	},
4639 4640 4641 4642
#ifdef CONFIG_NUMA
	{
		.name = "numa_stat",
		.open = mem_control_numa_stat_open,
4643
		.mode = S_IRUGO,
4644 4645
	},
#endif
B
Balbir Singh 已提交
4646 4647
};

4648 4649 4650 4651 4652 4653
#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
4654 4655
		.register_event = mem_cgroup_usage_register_event,
		.unregister_event = mem_cgroup_usage_unregister_event,
4656 4657 4658 4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690
	},
	{
		.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

4691 4692 4693
static int alloc_mem_cgroup_per_zone_info(struct mem_cgroup *mem, int node)
{
	struct mem_cgroup_per_node *pn;
4694
	struct mem_cgroup_per_zone *mz;
4695
	enum lru_list l;
4696
	int zone, tmp = node;
4697 4698 4699 4700 4701 4702 4703 4704
	/*
	 * 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.
	 */
4705 4706
	if (!node_state(node, N_NORMAL_MEMORY))
		tmp = -1;
4707
	pn = kzalloc_node(sizeof(*pn), GFP_KERNEL, tmp);
4708 4709
	if (!pn)
		return 1;
4710

4711
	mem->info.nodeinfo[node] = pn;
4712 4713
	for (zone = 0; zone < MAX_NR_ZONES; zone++) {
		mz = &pn->zoneinfo[zone];
4714 4715
		for_each_lru(l)
			INIT_LIST_HEAD(&mz->lists[l]);
4716
		mz->usage_in_excess = 0;
4717 4718
		mz->on_tree = false;
		mz->mem = mem;
4719
	}
4720 4721 4722
	return 0;
}

4723 4724 4725 4726 4727
static void free_mem_cgroup_per_zone_info(struct mem_cgroup *mem, int node)
{
	kfree(mem->info.nodeinfo[node]);
}

4728 4729 4730
static struct mem_cgroup *mem_cgroup_alloc(void)
{
	struct mem_cgroup *mem;
4731
	int size = sizeof(struct mem_cgroup);
4732

4733
	/* Can be very big if MAX_NUMNODES is very big */
4734
	if (size < PAGE_SIZE)
4735
		mem = kzalloc(size, GFP_KERNEL);
4736
	else
4737
		mem = vzalloc(size);
4738

4739 4740 4741
	if (!mem)
		return NULL;

4742
	mem->stat = alloc_percpu(struct mem_cgroup_stat_cpu);
4743 4744
	if (!mem->stat)
		goto out_free;
4745
	spin_lock_init(&mem->pcp_counter_lock);
4746
	return mem;
4747 4748 4749 4750 4751 4752 4753

out_free:
	if (size < PAGE_SIZE)
		kfree(mem);
	else
		vfree(mem);
	return NULL;
4754 4755
}

4756 4757 4758 4759 4760 4761 4762 4763 4764 4765 4766
/*
 * 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.
 */

4767
static void __mem_cgroup_free(struct mem_cgroup *mem)
4768
{
K
KAMEZAWA Hiroyuki 已提交
4769 4770
	int node;

4771
	mem_cgroup_remove_from_trees(mem);
K
KAMEZAWA Hiroyuki 已提交
4772 4773
	free_css_id(&mem_cgroup_subsys, &mem->css);

K
KAMEZAWA Hiroyuki 已提交
4774 4775 4776
	for_each_node_state(node, N_POSSIBLE)
		free_mem_cgroup_per_zone_info(mem, node);

4777 4778
	free_percpu(mem->stat);
	if (sizeof(struct mem_cgroup) < PAGE_SIZE)
4779 4780 4781 4782 4783
		kfree(mem);
	else
		vfree(mem);
}

4784 4785 4786 4787 4788
static void mem_cgroup_get(struct mem_cgroup *mem)
{
	atomic_inc(&mem->refcnt);
}

4789
static void __mem_cgroup_put(struct mem_cgroup *mem, int count)
4790
{
4791
	if (atomic_sub_and_test(count, &mem->refcnt)) {
4792
		struct mem_cgroup *parent = parent_mem_cgroup(mem);
4793
		__mem_cgroup_free(mem);
4794 4795 4796
		if (parent)
			mem_cgroup_put(parent);
	}
4797 4798
}

4799 4800 4801 4802 4803
static void mem_cgroup_put(struct mem_cgroup *mem)
{
	__mem_cgroup_put(mem, 1);
}

4804 4805 4806 4807 4808 4809 4810 4811 4812
/*
 * 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);
}
4813

4814 4815 4816
#ifdef CONFIG_CGROUP_MEM_RES_CTLR_SWAP
static void __init enable_swap_cgroup(void)
{
4817
	if (!mem_cgroup_disabled() && really_do_swap_account)
4818 4819 4820 4821 4822 4823 4824 4825
		do_swap_account = 1;
}
#else
static void __init enable_swap_cgroup(void)
{
}
#endif

4826 4827 4828 4829 4830 4831 4832 4833 4834 4835 4836 4837 4838 4839 4840 4841 4842 4843 4844 4845 4846 4847 4848 4849 4850
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
4851
static struct cgroup_subsys_state * __ref
B
Balbir Singh 已提交
4852 4853
mem_cgroup_create(struct cgroup_subsys *ss, struct cgroup *cont)
{
4854
	struct mem_cgroup *mem, *parent;
K
KAMEZAWA Hiroyuki 已提交
4855
	long error = -ENOMEM;
4856
	int node;
B
Balbir Singh 已提交
4857

4858 4859
	mem = mem_cgroup_alloc();
	if (!mem)
K
KAMEZAWA Hiroyuki 已提交
4860
		return ERR_PTR(error);
4861

4862 4863 4864
	for_each_node_state(node, N_POSSIBLE)
		if (alloc_mem_cgroup_per_zone_info(mem, node))
			goto free_out;
4865

4866
	/* root ? */
4867
	if (cont->parent == NULL) {
4868
		int cpu;
4869
		enable_swap_cgroup();
4870
		parent = NULL;
4871
		root_mem_cgroup = mem;
4872 4873
		if (mem_cgroup_soft_limit_tree_init())
			goto free_out;
4874 4875 4876 4877 4878
		for_each_possible_cpu(cpu) {
			struct memcg_stock_pcp *stock =
						&per_cpu(memcg_stock, cpu);
			INIT_WORK(&stock->work, drain_local_stock);
		}
4879
		hotcpu_notifier(memcg_cpu_hotplug_callback, 0);
4880
	} else {
4881
		parent = mem_cgroup_from_cont(cont->parent);
4882
		mem->use_hierarchy = parent->use_hierarchy;
4883
		mem->oom_kill_disable = parent->oom_kill_disable;
4884
	}
4885

4886 4887 4888
	if (parent && parent->use_hierarchy) {
		res_counter_init(&mem->res, &parent->res);
		res_counter_init(&mem->memsw, &parent->memsw);
4889 4890 4891 4892 4893 4894 4895
		/*
		 * 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);
4896 4897 4898 4899
	} else {
		res_counter_init(&mem->res, NULL);
		res_counter_init(&mem->memsw, NULL);
	}
K
KAMEZAWA Hiroyuki 已提交
4900
	mem->last_scanned_child = 0;
4901
	mem->last_scanned_node = MAX_NUMNODES;
K
KAMEZAWA Hiroyuki 已提交
4902
	INIT_LIST_HEAD(&mem->oom_notify);
4903

K
KOSAKI Motohiro 已提交
4904 4905
	if (parent)
		mem->swappiness = get_swappiness(parent);
4906
	atomic_set(&mem->refcnt, 1);
4907
	mem->move_charge_at_immigrate = 0;
4908
	mutex_init(&mem->thresholds_lock);
B
Balbir Singh 已提交
4909
	return &mem->css;
4910
free_out:
4911
	__mem_cgroup_free(mem);
4912
	root_mem_cgroup = NULL;
K
KAMEZAWA Hiroyuki 已提交
4913
	return ERR_PTR(error);
B
Balbir Singh 已提交
4914 4915
}

4916
static int mem_cgroup_pre_destroy(struct cgroup_subsys *ss,
4917 4918 4919
					struct cgroup *cont)
{
	struct mem_cgroup *mem = mem_cgroup_from_cont(cont);
4920 4921

	return mem_cgroup_force_empty(mem, false);
4922 4923
}

B
Balbir Singh 已提交
4924 4925 4926
static void mem_cgroup_destroy(struct cgroup_subsys *ss,
				struct cgroup *cont)
{
4927 4928 4929
	struct mem_cgroup *mem = mem_cgroup_from_cont(cont);

	mem_cgroup_put(mem);
B
Balbir Singh 已提交
4930 4931 4932 4933 4934
}

static int mem_cgroup_populate(struct cgroup_subsys *ss,
				struct cgroup *cont)
{
4935 4936 4937 4938 4939 4940 4941 4942
	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
4943 4944
}

4945
#ifdef CONFIG_MMU
4946
/* Handlers for move charge at task migration. */
4947 4948
#define PRECHARGE_COUNT_AT_ONCE	256
static int mem_cgroup_do_precharge(unsigned long count)
4949
{
4950 4951
	int ret = 0;
	int batch_count = PRECHARGE_COUNT_AT_ONCE;
4952 4953
	struct mem_cgroup *mem = mc.to;

4954 4955 4956 4957 4958 4959 4960 4961 4962 4963 4964 4965 4966 4967 4968 4969 4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984 4985 4986 4987 4988
	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();
		}
4989
		ret = __mem_cgroup_try_charge(NULL, GFP_KERNEL, 1, &mem, false);
4990 4991 4992 4993 4994
		if (ret || !mem)
			/* mem_cgroup_clear_mc() will do uncharge later */
			return -ENOMEM;
		mc.precharge++;
	}
4995 4996 4997 4998 4999 5000 5001 5002
	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
5003
 * @target: the pointer the target page or swap ent will be stored(can be NULL)
5004 5005 5006 5007 5008 5009
 *
 * 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).
5010 5011 5012
 *   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.
5013 5014 5015 5016 5017
 *
 * Called with pte lock held.
 */
union mc_target {
	struct page	*page;
5018
	swp_entry_t	ent;
5019 5020 5021 5022 5023
};

enum mc_target_type {
	MC_TARGET_NONE,	/* not used */
	MC_TARGET_PAGE,
5024
	MC_TARGET_SWAP,
5025 5026
};

D
Daisuke Nishimura 已提交
5027 5028
static struct page *mc_handle_present_pte(struct vm_area_struct *vma,
						unsigned long addr, pte_t ptent)
5029
{
D
Daisuke Nishimura 已提交
5030
	struct page *page = vm_normal_page(vma, addr, ptent);
5031

D
Daisuke Nishimura 已提交
5032 5033 5034 5035 5036 5037
	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;
5038 5039
	} else if (!move_file())
		/* we ignore mapcount for file pages */
D
Daisuke Nishimura 已提交
5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057
		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 */
5058 5059
		if (page)
			put_page(page);
D
Daisuke Nishimura 已提交
5060
		return NULL;
5061
	}
D
Daisuke Nishimura 已提交
5062 5063 5064 5065 5066 5067
	if (do_swap_account)
		entry->val = ent.val;

	return page;
}

5068 5069 5070 5071 5072 5073 5074 5075 5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087 5088 5089 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100
static struct page *mc_handle_file_pte(struct vm_area_struct *vma,
			unsigned long addr, pte_t ptent, swp_entry_t *entry)
{
	struct page *page = NULL;
	struct inode *inode;
	struct address_space *mapping;
	pgoff_t pgoff;

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

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

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

	return page;
}

D
Daisuke Nishimura 已提交
5101 5102 5103 5104 5105 5106 5107 5108 5109 5110 5111 5112
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);
5113 5114
	else if (pte_none(ptent) || pte_file(ptent))
		page = mc_handle_file_pte(vma, addr, ptent, &ent);
D
Daisuke Nishimura 已提交
5115 5116 5117

	if (!page && !ent.val)
		return 0;
5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128 5129 5130 5131 5132
	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
5133 5134
	/* There is a swap entry and a page doesn't exist or isn't charged */
	if (ent.val && !ret &&
5135 5136 5137 5138
			css_id(&mc.from->css) == lookup_swap_cgroup(ent)) {
		ret = MC_TARGET_SWAP;
		if (target)
			target->ent = ent;
5139 5140 5141 5142 5143 5144 5145 5146 5147 5148 5149 5150
	}
	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;

5151 5152
	split_huge_page_pmd(walk->mm, pmd);

5153 5154 5155 5156 5157 5158 5159
	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();

5160 5161 5162
	return 0;
}

5163 5164 5165 5166 5167
static unsigned long mem_cgroup_count_precharge(struct mm_struct *mm)
{
	unsigned long precharge;
	struct vm_area_struct *vma;

5168
	down_read(&mm->mmap_sem);
5169 5170 5171 5172 5173 5174 5175 5176 5177 5178 5179
	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);
	}
5180
	up_read(&mm->mmap_sem);
5181 5182 5183 5184 5185 5186 5187 5188 5189

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

	return precharge;
}

static int mem_cgroup_precharge_mc(struct mm_struct *mm)
{
5190 5191 5192 5193 5194
	unsigned long precharge = mem_cgroup_count_precharge(mm);

	VM_BUG_ON(mc.moving_task);
	mc.moving_task = current;
	return mem_cgroup_do_precharge(precharge);
5195 5196
}

5197 5198
/* cancels all extra charges on mc.from and mc.to, and wakes up all waiters. */
static void __mem_cgroup_clear_mc(void)
5199
{
5200 5201 5202
	struct mem_cgroup *from = mc.from;
	struct mem_cgroup *to = mc.to;

5203
	/* we must uncharge all the leftover precharges from mc.to */
5204 5205 5206 5207 5208 5209 5210 5211 5212 5213 5214
	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;
5215
	}
5216 5217 5218 5219 5220 5221 5222 5223 5224 5225 5226 5227 5228 5229 5230 5231 5232 5233 5234
	/* 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;
	}
5235 5236 5237 5238 5239 5240 5241 5242 5243 5244 5245 5246 5247 5248 5249
	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();
5250
	spin_lock(&mc.lock);
5251 5252
	mc.from = NULL;
	mc.to = NULL;
5253
	spin_unlock(&mc.lock);
5254
	mem_cgroup_end_move(from);
5255 5256
}

5257 5258
static int mem_cgroup_can_attach(struct cgroup_subsys *ss,
				struct cgroup *cgroup,
5259
				struct task_struct *p)
5260 5261 5262 5263 5264 5265 5266 5267 5268 5269 5270 5271 5272 5273
{
	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 */
5274 5275 5276 5277
		if (mm->owner == p) {
			VM_BUG_ON(mc.from);
			VM_BUG_ON(mc.to);
			VM_BUG_ON(mc.precharge);
5278
			VM_BUG_ON(mc.moved_charge);
5279
			VM_BUG_ON(mc.moved_swap);
5280
			mem_cgroup_start_move(from);
5281
			spin_lock(&mc.lock);
5282 5283
			mc.from = from;
			mc.to = mem;
5284
			spin_unlock(&mc.lock);
5285
			/* We set mc.moving_task later */
5286 5287 5288 5289

			ret = mem_cgroup_precharge_mc(mm);
			if (ret)
				mem_cgroup_clear_mc();
5290 5291
		}
		mmput(mm);
5292 5293 5294 5295 5296 5297
	}
	return ret;
}

static void mem_cgroup_cancel_attach(struct cgroup_subsys *ss,
				struct cgroup *cgroup,
5298
				struct task_struct *p)
5299
{
5300
	mem_cgroup_clear_mc();
5301 5302
}

5303 5304 5305
static int mem_cgroup_move_charge_pte_range(pmd_t *pmd,
				unsigned long addr, unsigned long end,
				struct mm_walk *walk)
5306
{
5307 5308 5309 5310 5311
	int ret = 0;
	struct vm_area_struct *vma = walk->private;
	pte_t *pte;
	spinlock_t *ptl;

5312
	split_huge_page_pmd(walk->mm, pmd);
5313 5314 5315 5316 5317 5318 5319 5320
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;
5321
		swp_entry_t ent;
5322 5323 5324 5325 5326 5327 5328 5329 5330 5331 5332

		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);
5333 5334
			if (!mem_cgroup_move_account(page, 1, pc,
						     mc.from, mc.to, false)) {
5335
				mc.precharge--;
5336 5337
				/* we uncharge from mc.from later. */
				mc.moved_charge++;
5338 5339 5340 5341 5342
			}
			putback_lru_page(page);
put:			/* is_target_pte_for_mc() gets the page */
			put_page(page);
			break;
5343 5344
		case MC_TARGET_SWAP:
			ent = target.ent;
5345 5346
			if (!mem_cgroup_move_swap_account(ent,
						mc.from, mc.to, false)) {
5347
				mc.precharge--;
5348 5349 5350
				/* we fixup refcnts and charges later. */
				mc.moved_swap++;
			}
5351
			break;
5352 5353 5354 5355 5356 5357 5358 5359 5360 5361 5362 5363 5364 5365
		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.
		 */
5366
		ret = mem_cgroup_do_precharge(1);
5367 5368 5369 5370 5371 5372 5373 5374 5375 5376 5377 5378
		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();
5379 5380 5381 5382 5383 5384 5385 5386 5387 5388 5389 5390 5391
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;
	}
5392 5393 5394 5395 5396 5397 5398 5399 5400 5401 5402 5403 5404 5405 5406 5407 5408 5409
	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;
	}
5410
	up_read(&mm->mmap_sem);
5411 5412
}

B
Balbir Singh 已提交
5413 5414 5415
static void mem_cgroup_move_task(struct cgroup_subsys *ss,
				struct cgroup *cont,
				struct cgroup *old_cont,
5416
				struct task_struct *p)
B
Balbir Singh 已提交
5417
{
5418
	struct mm_struct *mm = get_task_mm(p);
5419 5420

	if (mm) {
5421 5422 5423
		if (mc.to)
			mem_cgroup_move_charge(mm);
		put_swap_token(mm);
5424 5425
		mmput(mm);
	}
5426 5427
	if (mc.to)
		mem_cgroup_clear_mc();
B
Balbir Singh 已提交
5428
}
5429 5430 5431
#else	/* !CONFIG_MMU */
static int mem_cgroup_can_attach(struct cgroup_subsys *ss,
				struct cgroup *cgroup,
5432
				struct task_struct *p)
5433 5434 5435 5436 5437
{
	return 0;
}
static void mem_cgroup_cancel_attach(struct cgroup_subsys *ss,
				struct cgroup *cgroup,
5438
				struct task_struct *p)
5439 5440 5441 5442 5443
{
}
static void mem_cgroup_move_task(struct cgroup_subsys *ss,
				struct cgroup *cont,
				struct cgroup *old_cont,
5444
				struct task_struct *p)
5445 5446 5447
{
}
#endif
B
Balbir Singh 已提交
5448

B
Balbir Singh 已提交
5449 5450 5451 5452
struct cgroup_subsys mem_cgroup_subsys = {
	.name = "memory",
	.subsys_id = mem_cgroup_subsys_id,
	.create = mem_cgroup_create,
5453
	.pre_destroy = mem_cgroup_pre_destroy,
B
Balbir Singh 已提交
5454 5455
	.destroy = mem_cgroup_destroy,
	.populate = mem_cgroup_populate,
5456 5457
	.can_attach = mem_cgroup_can_attach,
	.cancel_attach = mem_cgroup_cancel_attach,
B
Balbir Singh 已提交
5458
	.attach = mem_cgroup_move_task,
5459
	.early_init = 0,
K
KAMEZAWA Hiroyuki 已提交
5460
	.use_id = 1,
B
Balbir Singh 已提交
5461
};
5462 5463

#ifdef CONFIG_CGROUP_MEM_RES_CTLR_SWAP
5464 5465 5466
static int __init enable_swap_account(char *s)
{
	/* consider enabled if no parameter or 1 is given */
5467
	if (!strcmp(s, "1"))
5468
		really_do_swap_account = 1;
5469
	else if (!strcmp(s, "0"))
5470 5471 5472
		really_do_swap_account = 0;
	return 1;
}
5473
__setup("swapaccount=", enable_swap_account);
5474 5475

#endif