vmscan.c 96.7 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
/*
 *  linux/mm/vmscan.c
 *
 *  Copyright (C) 1991, 1992, 1993, 1994  Linus Torvalds
 *
 *  Swap reorganised 29.12.95, Stephen Tweedie.
 *  kswapd added: 7.1.96  sct
 *  Removed kswapd_ctl limits, and swap out as many pages as needed
 *  to bring the system back to freepages.high: 2.4.97, Rik van Riel.
 *  Zone aware kswapd started 02/00, Kanoj Sarcar (kanoj@sgi.com).
 *  Multiqueue VM started 5.8.00, Rik van Riel.
 */

#include <linux/mm.h>
#include <linux/module.h>
16
#include <linux/gfp.h>
L
Linus Torvalds 已提交
17 18 19 20 21
#include <linux/kernel_stat.h>
#include <linux/swap.h>
#include <linux/pagemap.h>
#include <linux/init.h>
#include <linux/highmem.h>
22
#include <linux/vmstat.h>
L
Linus Torvalds 已提交
23 24 25 26 27 28 29 30 31 32 33 34
#include <linux/file.h>
#include <linux/writeback.h>
#include <linux/blkdev.h>
#include <linux/buffer_head.h>	/* for try_to_release_page(),
					buffer_heads_over_limit */
#include <linux/mm_inline.h>
#include <linux/pagevec.h>
#include <linux/backing-dev.h>
#include <linux/rmap.h>
#include <linux/topology.h>
#include <linux/cpu.h>
#include <linux/cpuset.h>
35
#include <linux/compaction.h>
L
Linus Torvalds 已提交
36 37
#include <linux/notifier.h>
#include <linux/rwsem.h>
38
#include <linux/delay.h>
39
#include <linux/kthread.h>
40
#include <linux/freezer.h>
41
#include <linux/memcontrol.h>
42
#include <linux/delayacct.h>
43
#include <linux/sysctl.h>
44
#include <linux/oom.h>
45
#include <linux/prefetch.h>
L
Linus Torvalds 已提交
46 47 48 49 50 51

#include <asm/tlbflush.h>
#include <asm/div64.h>

#include <linux/swapops.h>

52 53
#include "internal.h"

54 55 56
#define CREATE_TRACE_POINTS
#include <trace/events/vmscan.h>

57
/*
58 59 60 61 62
 * reclaim_mode determines how the inactive list is shrunk
 * RECLAIM_MODE_SINGLE: Reclaim only order-0 pages
 * RECLAIM_MODE_ASYNC:  Do not block
 * RECLAIM_MODE_SYNC:   Allow blocking e.g. call wait_on_page_writeback
 * RECLAIM_MODE_LUMPYRECLAIM: For high-order allocations, take a reference
63 64
 *			page from the LRU and reclaim all pages within a
 *			naturally aligned range
65
 * RECLAIM_MODE_COMPACTION: For high-order allocations, reclaim a number of
66
 *			order-0 pages and then compact the zone
67
 */
68 69 70 71 72 73
typedef unsigned __bitwise__ reclaim_mode_t;
#define RECLAIM_MODE_SINGLE		((__force reclaim_mode_t)0x01u)
#define RECLAIM_MODE_ASYNC		((__force reclaim_mode_t)0x02u)
#define RECLAIM_MODE_SYNC		((__force reclaim_mode_t)0x04u)
#define RECLAIM_MODE_LUMPYRECLAIM	((__force reclaim_mode_t)0x08u)
#define RECLAIM_MODE_COMPACTION		((__force reclaim_mode_t)0x10u)
74

L
Linus Torvalds 已提交
75 76 77 78
struct scan_control {
	/* Incremented by the number of inactive pages that were scanned */
	unsigned long nr_scanned;

79 80 81
	/* Number of pages freed so far during a call to shrink_zones() */
	unsigned long nr_reclaimed;

82 83 84
	/* How many pages shrink_list() should reclaim */
	unsigned long nr_to_reclaim;

85 86
	unsigned long hibernation_mode;

L
Linus Torvalds 已提交
87
	/* This context's GFP mask */
A
Al Viro 已提交
88
	gfp_t gfp_mask;
L
Linus Torvalds 已提交
89 90 91

	int may_writepage;

92 93
	/* Can mapped pages be reclaimed? */
	int may_unmap;
94

95 96 97
	/* Can pages be swapped as part of reclaim? */
	int may_swap;

A
Andy Whitcroft 已提交
98
	int order;
99

100
	/*
101 102
	 * Intend to reclaim enough continuous memory rather than reclaim
	 * enough amount of memory. i.e, mode for high order allocation.
103
	 */
104
	reclaim_mode_t reclaim_mode;
105

106 107 108
	/* Which cgroup do we reclaim from */
	struct mem_cgroup *mem_cgroup;

109 110 111 112 113
	/*
	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
	 * are scanned.
	 */
	nodemask_t	*nodemask;
L
Linus Torvalds 已提交
114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149
};

#define lru_to_page(_head) (list_entry((_head)->prev, struct page, lru))

#ifdef ARCH_HAS_PREFETCH
#define prefetch_prev_lru_page(_page, _base, _field)			\
	do {								\
		if ((_page)->lru.prev != _base) {			\
			struct page *prev;				\
									\
			prev = lru_to_page(&(_page->lru));		\
			prefetch(&prev->_field);			\
		}							\
	} while (0)
#else
#define prefetch_prev_lru_page(_page, _base, _field) do { } while (0)
#endif

#ifdef ARCH_HAS_PREFETCHW
#define prefetchw_prev_lru_page(_page, _base, _field)			\
	do {								\
		if ((_page)->lru.prev != _base) {			\
			struct page *prev;				\
									\
			prev = lru_to_page(&(_page->lru));		\
			prefetchw(&prev->_field);			\
		}							\
	} while (0)
#else
#define prefetchw_prev_lru_page(_page, _base, _field) do { } while (0)
#endif

/*
 * From 0 .. 100.  Higher means more swappy.
 */
int vm_swappiness = 60;
150
long vm_total_pages;	/* The total number of pages which the VM controls */
L
Linus Torvalds 已提交
151 152 153 154

static LIST_HEAD(shrinker_list);
static DECLARE_RWSEM(shrinker_rwsem);

155
#ifdef CONFIG_CGROUP_MEM_RES_CTLR
156
#define scanning_global_lru(sc)	(!(sc)->mem_cgroup)
157
#else
158
#define scanning_global_lru(sc)	(1)
159 160
#endif

161 162 163
static struct zone_reclaim_stat *get_reclaim_stat(struct zone *zone,
						  struct scan_control *sc)
{
164
	if (!scanning_global_lru(sc))
K
KOSAKI Motohiro 已提交
165 166
		return mem_cgroup_get_reclaim_stat(sc->mem_cgroup, zone);

167 168 169
	return &zone->reclaim_stat;
}

170 171
static unsigned long zone_nr_lru_pages(struct zone *zone,
				struct scan_control *sc, enum lru_list lru)
172
{
173
	if (!scanning_global_lru(sc))
174
		return mem_cgroup_zone_nr_lru_pages(sc->mem_cgroup, zone, lru);
175

176 177 178 179
	return zone_page_state(zone, NR_LRU_BASE + lru);
}


L
Linus Torvalds 已提交
180 181 182
/*
 * Add a shrinker callback to be called from the vm
 */
183
void register_shrinker(struct shrinker *shrinker)
L
Linus Torvalds 已提交
184
{
185 186 187 188
	shrinker->nr = 0;
	down_write(&shrinker_rwsem);
	list_add_tail(&shrinker->list, &shrinker_list);
	up_write(&shrinker_rwsem);
L
Linus Torvalds 已提交
189
}
190
EXPORT_SYMBOL(register_shrinker);
L
Linus Torvalds 已提交
191 192 193 194

/*
 * Remove one
 */
195
void unregister_shrinker(struct shrinker *shrinker)
L
Linus Torvalds 已提交
196 197 198 199 200
{
	down_write(&shrinker_rwsem);
	list_del(&shrinker->list);
	up_write(&shrinker_rwsem);
}
201
EXPORT_SYMBOL(unregister_shrinker);
L
Linus Torvalds 已提交
202

203 204 205 206 207 208 209 210
static inline int do_shrinker_shrink(struct shrinker *shrinker,
				     struct shrink_control *sc,
				     unsigned long nr_to_scan)
{
	sc->nr_to_scan = nr_to_scan;
	return (*shrinker->shrink)(shrinker, sc);
}

L
Linus Torvalds 已提交
211 212 213 214 215 216 217 218 219
#define SHRINK_BATCH 128
/*
 * Call the shrink functions to age shrinkable caches
 *
 * Here we assume it costs one seek to replace a lru page and that it also
 * takes a seek to recreate a cache object.  With this in mind we age equal
 * percentages of the lru and ageable caches.  This should balance the seeks
 * generated by these structures.
 *
S
Simon Arlott 已提交
220
 * If the vm encountered mapped pages on the LRU it increase the pressure on
L
Linus Torvalds 已提交
221 222 223 224 225 226 227
 * slab to avoid swapping.
 *
 * We do weird things to avoid (scanned*seeks*entries) overflowing 32 bits.
 *
 * `lru_pages' represents the number of on-LRU pages in all the zones which
 * are eligible for the caller's allocation attempt.  It is used for balancing
 * slab reclaim versus page reclaim.
228 229
 *
 * Returns the number of slab objects which we shrunk.
L
Linus Torvalds 已提交
230
 */
231
unsigned long shrink_slab(struct shrink_control *shrink,
232
			  unsigned long nr_pages_scanned,
233
			  unsigned long lru_pages)
L
Linus Torvalds 已提交
234 235
{
	struct shrinker *shrinker;
236
	unsigned long ret = 0;
L
Linus Torvalds 已提交
237

238 239
	if (nr_pages_scanned == 0)
		nr_pages_scanned = SWAP_CLUSTER_MAX;
L
Linus Torvalds 已提交
240

241 242 243 244 245
	if (!down_read_trylock(&shrinker_rwsem)) {
		/* Assume we'll be able to shrink next time */
		ret = 1;
		goto out;
	}
L
Linus Torvalds 已提交
246 247 248 249

	list_for_each_entry(shrinker, &shrinker_list, list) {
		unsigned long long delta;
		unsigned long total_scan;
250
		unsigned long max_pass;
251
		int shrink_ret = 0;
252 253
		long nr;
		long new_nr;
254 255
		long batch_size = shrinker->batch ? shrinker->batch
						  : SHRINK_BATCH;
L
Linus Torvalds 已提交
256

257 258 259 260 261 262 263 264 265 266
		/*
		 * copy the current shrinker scan count into a local variable
		 * and zero it so that other concurrent shrinker invocations
		 * don't also do this scanning work.
		 */
		do {
			nr = shrinker->nr;
		} while (cmpxchg(&shrinker->nr, nr, 0) != nr);

		total_scan = nr;
267 268
		max_pass = do_shrinker_shrink(shrinker, shrink, 0);
		delta = (4 * nr_pages_scanned) / shrinker->seeks;
269
		delta *= max_pass;
L
Linus Torvalds 已提交
270
		do_div(delta, lru_pages + 1);
271 272
		total_scan += delta;
		if (total_scan < 0) {
273 274
			printk(KERN_ERR "shrink_slab: %pF negative objects to "
			       "delete nr=%ld\n",
275 276
			       shrinker->shrink, total_scan);
			total_scan = max_pass;
277 278
		}

279 280 281 282 283 284 285 286 287 288 289 290 291 292 293
		/*
		 * We need to avoid excessive windup on filesystem shrinkers
		 * due to large numbers of GFP_NOFS allocations causing the
		 * shrinkers to return -1 all the time. This results in a large
		 * nr being built up so when a shrink that can do some work
		 * comes along it empties the entire cache due to nr >>>
		 * max_pass.  This is bad for sustaining a working set in
		 * memory.
		 *
		 * Hence only allow the shrinker to scan the entire cache when
		 * a large delta change is calculated directly.
		 */
		if (delta < max_pass / 4)
			total_scan = min(total_scan, max_pass / 2);

294 295 296 297 298
		/*
		 * Avoid risking looping forever due to too large nr value:
		 * never try to free more than twice the estimate number of
		 * freeable entries.
		 */
299 300
		if (total_scan > max_pass * 2)
			total_scan = max_pass * 2;
L
Linus Torvalds 已提交
301

302
		trace_mm_shrink_slab_start(shrinker, shrink, nr,
303 304 305
					nr_pages_scanned, lru_pages,
					max_pass, delta, total_scan);

306
		while (total_scan >= batch_size) {
307
			int nr_before;
L
Linus Torvalds 已提交
308

309 310
			nr_before = do_shrinker_shrink(shrinker, shrink, 0);
			shrink_ret = do_shrinker_shrink(shrinker, shrink,
311
							batch_size);
L
Linus Torvalds 已提交
312 313
			if (shrink_ret == -1)
				break;
314 315
			if (shrink_ret < nr_before)
				ret += nr_before - shrink_ret;
316 317
			count_vm_events(SLABS_SCANNED, batch_size);
			total_scan -= batch_size;
L
Linus Torvalds 已提交
318 319 320 321

			cond_resched();
		}

322 323 324 325 326 327 328 329 330 331 332 333 334
		/*
		 * move the unused scan count back into the shrinker in a
		 * manner that handles concurrent updates. If we exhausted the
		 * scan, there is no need to do an update.
		 */
		do {
			nr = shrinker->nr;
			new_nr = total_scan + nr;
			if (total_scan <= 0)
				break;
		} while (cmpxchg(&shrinker->nr, nr, new_nr) != nr);

		trace_mm_shrink_slab_end(shrinker, shrink_ret, nr, new_nr);
L
Linus Torvalds 已提交
335 336
	}
	up_read(&shrinker_rwsem);
337 338
out:
	cond_resched();
339
	return ret;
L
Linus Torvalds 已提交
340 341
}

342
static void set_reclaim_mode(int priority, struct scan_control *sc,
343 344
				   bool sync)
{
345
	reclaim_mode_t syncmode = sync ? RECLAIM_MODE_SYNC : RECLAIM_MODE_ASYNC;
346 347

	/*
348 349 350
	 * Initially assume we are entering either lumpy reclaim or
	 * reclaim/compaction.Depending on the order, we will either set the
	 * sync mode or just reclaim order-0 pages later.
351
	 */
352
	if (COMPACTION_BUILD)
353
		sc->reclaim_mode = RECLAIM_MODE_COMPACTION;
354
	else
355
		sc->reclaim_mode = RECLAIM_MODE_LUMPYRECLAIM;
356 357

	/*
358 359 360
	 * Avoid using lumpy reclaim or reclaim/compaction if possible by
	 * restricting when its set to either costly allocations or when
	 * under memory pressure
361 362
	 */
	if (sc->order > PAGE_ALLOC_COSTLY_ORDER)
363
		sc->reclaim_mode |= syncmode;
364
	else if (sc->order && priority < DEF_PRIORITY - 2)
365
		sc->reclaim_mode |= syncmode;
366
	else
367
		sc->reclaim_mode = RECLAIM_MODE_SINGLE | RECLAIM_MODE_ASYNC;
368 369
}

370
static void reset_reclaim_mode(struct scan_control *sc)
371
{
372
	sc->reclaim_mode = RECLAIM_MODE_SINGLE | RECLAIM_MODE_ASYNC;
373 374
}

L
Linus Torvalds 已提交
375 376
static inline int is_page_cache_freeable(struct page *page)
{
377 378 379 380 381
	/*
	 * A freeable page cache page is referenced only by the caller
	 * that isolated the page, the page cache radix tree and
	 * optional buffer heads at page->private.
	 */
382
	return page_count(page) - page_has_private(page) == 2;
L
Linus Torvalds 已提交
383 384
}

385 386
static int may_write_to_queue(struct backing_dev_info *bdi,
			      struct scan_control *sc)
L
Linus Torvalds 已提交
387
{
388
	if (current->flags & PF_SWAPWRITE)
L
Linus Torvalds 已提交
389 390 391 392 393
		return 1;
	if (!bdi_write_congested(bdi))
		return 1;
	if (bdi == current->backing_dev_info)
		return 1;
394 395 396 397

	/* lumpy reclaim for hugepage often need a lot of write */
	if (sc->order > PAGE_ALLOC_COSTLY_ORDER)
		return 1;
L
Linus Torvalds 已提交
398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415
	return 0;
}

/*
 * We detected a synchronous write error writing a page out.  Probably
 * -ENOSPC.  We need to propagate that into the address_space for a subsequent
 * fsync(), msync() or close().
 *
 * The tricky part is that after writepage we cannot touch the mapping: nothing
 * prevents it from being freed up.  But we have a ref on the page and once
 * that page is locked, the mapping is pinned.
 *
 * We're allowed to run sleeping lock_page() here because we know the caller has
 * __GFP_FS.
 */
static void handle_write_error(struct address_space *mapping,
				struct page *page, int error)
{
J
Jens Axboe 已提交
416
	lock_page(page);
417 418
	if (page_mapping(page) == mapping)
		mapping_set_error(mapping, error);
L
Linus Torvalds 已提交
419 420 421
	unlock_page(page);
}

422 423 424 425 426 427 428 429 430 431 432 433
/* possible outcome of pageout() */
typedef enum {
	/* failed to write page out, page is locked */
	PAGE_KEEP,
	/* move page to the active list, page is locked */
	PAGE_ACTIVATE,
	/* page has been sent to the disk successfully, page is unlocked */
	PAGE_SUCCESS,
	/* page is clean and locked */
	PAGE_CLEAN,
} pageout_t;

L
Linus Torvalds 已提交
434
/*
A
Andrew Morton 已提交
435 436
 * pageout is called by shrink_page_list() for each dirty page.
 * Calls ->writepage().
L
Linus Torvalds 已提交
437
 */
438
static pageout_t pageout(struct page *page, struct address_space *mapping,
439
			 struct scan_control *sc)
L
Linus Torvalds 已提交
440 441 442 443 444 445 446 447
{
	/*
	 * If the page is dirty, only perform writeback if that write
	 * will be non-blocking.  To prevent this allocation from being
	 * stalled by pagecache activity.  But note that there may be
	 * stalls if we need to run get_block().  We could test
	 * PagePrivate for that.
	 *
448
	 * If this process is currently in __generic_file_aio_write() against
L
Linus Torvalds 已提交
449 450 451 452 453 454 455 456 457 458 459 460 461 462 463
	 * this page's queue, we can perform writeback even if that
	 * will block.
	 *
	 * If the page is swapcache, write it back even if that would
	 * block, for some throttling. This happens by accident, because
	 * swap_backing_dev_info is bust: it doesn't reflect the
	 * congestion state of the swapdevs.  Easy to fix, if needed.
	 */
	if (!is_page_cache_freeable(page))
		return PAGE_KEEP;
	if (!mapping) {
		/*
		 * Some data journaling orphaned pages can have
		 * page->mapping == NULL while being dirty with clean buffers.
		 */
464
		if (page_has_private(page)) {
L
Linus Torvalds 已提交
465 466
			if (try_to_free_buffers(page)) {
				ClearPageDirty(page);
467
				printk("%s: orphaned page\n", __func__);
L
Linus Torvalds 已提交
468 469 470 471 472 473 474
				return PAGE_CLEAN;
			}
		}
		return PAGE_KEEP;
	}
	if (mapping->a_ops->writepage == NULL)
		return PAGE_ACTIVATE;
475
	if (!may_write_to_queue(mapping->backing_dev_info, sc))
L
Linus Torvalds 已提交
476 477 478 479 480 481 482
		return PAGE_KEEP;

	if (clear_page_dirty_for_io(page)) {
		int res;
		struct writeback_control wbc = {
			.sync_mode = WB_SYNC_NONE,
			.nr_to_write = SWAP_CLUSTER_MAX,
483 484
			.range_start = 0,
			.range_end = LLONG_MAX,
L
Linus Torvalds 已提交
485 486 487 488 489 490 491
			.for_reclaim = 1,
		};

		SetPageReclaim(page);
		res = mapping->a_ops->writepage(page, &wbc);
		if (res < 0)
			handle_write_error(mapping, page, res);
492
		if (res == AOP_WRITEPAGE_ACTIVATE) {
L
Linus Torvalds 已提交
493 494 495
			ClearPageReclaim(page);
			return PAGE_ACTIVATE;
		}
496 497 498 499 500 501

		/*
		 * Wait on writeback if requested to. This happens when
		 * direct reclaiming a large contiguous area and the
		 * first attempt to free a range of pages fails.
		 */
502
		if (PageWriteback(page) &&
503
		    (sc->reclaim_mode & RECLAIM_MODE_SYNC))
504 505
			wait_on_page_writeback(page);

L
Linus Torvalds 已提交
506 507 508 509
		if (!PageWriteback(page)) {
			/* synchronous write or broken a_ops? */
			ClearPageReclaim(page);
		}
510
		trace_mm_vmscan_writepage(page,
511
			trace_reclaim_flags(page, sc->reclaim_mode));
512
		inc_zone_page_state(page, NR_VMSCAN_WRITE);
L
Linus Torvalds 已提交
513 514 515 516 517 518
		return PAGE_SUCCESS;
	}

	return PAGE_CLEAN;
}

519
/*
N
Nick Piggin 已提交
520 521
 * Same as remove_mapping, but if the page is removed from the mapping, it
 * gets returned with a refcount of 0.
522
 */
N
Nick Piggin 已提交
523
static int __remove_mapping(struct address_space *mapping, struct page *page)
524
{
525 526
	BUG_ON(!PageLocked(page));
	BUG_ON(mapping != page_mapping(page));
527

N
Nick Piggin 已提交
528
	spin_lock_irq(&mapping->tree_lock);
529
	/*
N
Nick Piggin 已提交
530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552
	 * The non racy check for a busy page.
	 *
	 * Must be careful with the order of the tests. When someone has
	 * a ref to the page, it may be possible that they dirty it then
	 * drop the reference. So if PageDirty is tested before page_count
	 * here, then the following race may occur:
	 *
	 * get_user_pages(&page);
	 * [user mapping goes away]
	 * write_to(page);
	 *				!PageDirty(page)    [good]
	 * SetPageDirty(page);
	 * put_page(page);
	 *				!page_count(page)   [good, discard it]
	 *
	 * [oops, our write_to data is lost]
	 *
	 * Reversing the order of the tests ensures such a situation cannot
	 * escape unnoticed. The smp_rmb is needed to ensure the page->flags
	 * load is not satisfied before that of page->_count.
	 *
	 * Note that if SetPageDirty is always performed via set_page_dirty,
	 * and thus under tree_lock, then this ordering is not required.
553
	 */
N
Nick Piggin 已提交
554
	if (!page_freeze_refs(page, 2))
555
		goto cannot_free;
N
Nick Piggin 已提交
556 557 558
	/* note: atomic_cmpxchg in page_freeze_refs provides the smp_rmb */
	if (unlikely(PageDirty(page))) {
		page_unfreeze_refs(page, 2);
559
		goto cannot_free;
N
Nick Piggin 已提交
560
	}
561 562 563 564

	if (PageSwapCache(page)) {
		swp_entry_t swap = { .val = page_private(page) };
		__delete_from_swap_cache(page);
N
Nick Piggin 已提交
565
		spin_unlock_irq(&mapping->tree_lock);
566
		swapcache_free(swap, page);
N
Nick Piggin 已提交
567
	} else {
568 569 570 571
		void (*freepage)(struct page *);

		freepage = mapping->a_ops->freepage;

572
		__delete_from_page_cache(page);
N
Nick Piggin 已提交
573
		spin_unlock_irq(&mapping->tree_lock);
574
		mem_cgroup_uncharge_cache_page(page);
575 576 577

		if (freepage != NULL)
			freepage(page);
578 579 580 581 582
	}

	return 1;

cannot_free:
N
Nick Piggin 已提交
583
	spin_unlock_irq(&mapping->tree_lock);
584 585 586
	return 0;
}

N
Nick Piggin 已提交
587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606
/*
 * Attempt to detach a locked page from its ->mapping.  If it is dirty or if
 * someone else has a ref on the page, abort and return 0.  If it was
 * successfully detached, return 1.  Assumes the caller has a single ref on
 * this page.
 */
int remove_mapping(struct address_space *mapping, struct page *page)
{
	if (__remove_mapping(mapping, page)) {
		/*
		 * Unfreezing the refcount with 1 rather than 2 effectively
		 * drops the pagecache ref for us without requiring another
		 * atomic operation.
		 */
		page_unfreeze_refs(page, 1);
		return 1;
	}
	return 0;
}

L
Lee Schermerhorn 已提交
607 608 609 610 611 612 613 614 615 616 617 618 619
/**
 * putback_lru_page - put previously isolated page onto appropriate LRU list
 * @page: page to be put back to appropriate lru list
 *
 * Add previously isolated @page to appropriate LRU list.
 * Page may still be unevictable for other reasons.
 *
 * lru_lock must not be held, interrupts must be enabled.
 */
void putback_lru_page(struct page *page)
{
	int lru;
	int active = !!TestClearPageActive(page);
620
	int was_unevictable = PageUnevictable(page);
L
Lee Schermerhorn 已提交
621 622 623 624 625 626 627 628 629 630 631 632 633

	VM_BUG_ON(PageLRU(page));

redo:
	ClearPageUnevictable(page);

	if (page_evictable(page, NULL)) {
		/*
		 * For evictable pages, we can use the cache.
		 * In event of a race, worst case is we end up with an
		 * unevictable page on [in]active list.
		 * We know how to handle that.
		 */
634
		lru = active + page_lru_base_type(page);
L
Lee Schermerhorn 已提交
635 636 637 638 639 640 641 642
		lru_cache_add_lru(page, lru);
	} else {
		/*
		 * Put unevictable pages directly on zone's unevictable
		 * list.
		 */
		lru = LRU_UNEVICTABLE;
		add_page_to_unevictable_list(page);
643 644 645 646 647 648 649 650 651 652
		/*
		 * When racing with an mlock clearing (page is
		 * unlocked), make sure that if the other thread does
		 * not observe our setting of PG_lru and fails
		 * isolation, we see PG_mlocked cleared below and move
		 * the page back to the evictable list.
		 *
		 * The other side is TestClearPageMlocked().
		 */
		smp_mb();
L
Lee Schermerhorn 已提交
653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670
	}

	/*
	 * page's status can change while we move it among lru. If an evictable
	 * page is on unevictable list, it never be freed. To avoid that,
	 * check after we added it to the list, again.
	 */
	if (lru == LRU_UNEVICTABLE && page_evictable(page, NULL)) {
		if (!isolate_lru_page(page)) {
			put_page(page);
			goto redo;
		}
		/* This means someone else dropped this page from LRU
		 * So, it will be freed or putback to LRU again. There is
		 * nothing to do here.
		 */
	}

671 672 673 674 675
	if (was_unevictable && lru != LRU_UNEVICTABLE)
		count_vm_event(UNEVICTABLE_PGRESCUED);
	else if (!was_unevictable && lru == LRU_UNEVICTABLE)
		count_vm_event(UNEVICTABLE_PGCULLED);

L
Lee Schermerhorn 已提交
676 677 678
	put_page(page);		/* drop ref from isolate */
}

679 680 681
enum page_references {
	PAGEREF_RECLAIM,
	PAGEREF_RECLAIM_CLEAN,
682
	PAGEREF_KEEP,
683 684 685 686 687 688
	PAGEREF_ACTIVATE,
};

static enum page_references page_check_references(struct page *page,
						  struct scan_control *sc)
{
689
	int referenced_ptes, referenced_page;
690 691
	unsigned long vm_flags;

692 693
	referenced_ptes = page_referenced(page, 1, sc->mem_cgroup, &vm_flags);
	referenced_page = TestClearPageReferenced(page);
694 695

	/* Lumpy reclaim - ignore references */
696
	if (sc->reclaim_mode & RECLAIM_MODE_LUMPYRECLAIM)
697 698 699 700 701 702 703 704 705
		return PAGEREF_RECLAIM;

	/*
	 * Mlock lost the isolation race with us.  Let try_to_unmap()
	 * move the page to the unevictable list.
	 */
	if (vm_flags & VM_LOCKED)
		return PAGEREF_RECLAIM;

706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729
	if (referenced_ptes) {
		if (PageAnon(page))
			return PAGEREF_ACTIVATE;
		/*
		 * All mapped pages start out with page table
		 * references from the instantiating fault, so we need
		 * to look twice if a mapped file page is used more
		 * than once.
		 *
		 * Mark it and spare it for another trip around the
		 * inactive list.  Another page table reference will
		 * lead to its activation.
		 *
		 * Note: the mark is set for activated pages as well
		 * so that recently deactivated but used pages are
		 * quickly recovered.
		 */
		SetPageReferenced(page);

		if (referenced_page)
			return PAGEREF_ACTIVATE;

		return PAGEREF_KEEP;
	}
730 731

	/* Reclaim if clean, defer dirty pages to writeback */
732
	if (referenced_page && !PageSwapBacked(page))
733 734 735
		return PAGEREF_RECLAIM_CLEAN;

	return PAGEREF_RECLAIM;
736 737
}

738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755
static noinline_for_stack void free_page_list(struct list_head *free_pages)
{
	struct pagevec freed_pvec;
	struct page *page, *tmp;

	pagevec_init(&freed_pvec, 1);

	list_for_each_entry_safe(page, tmp, free_pages, lru) {
		list_del(&page->lru);
		if (!pagevec_add(&freed_pvec, page)) {
			__pagevec_free(&freed_pvec);
			pagevec_reinit(&freed_pvec);
		}
	}

	pagevec_free(&freed_pvec);
}

L
Linus Torvalds 已提交
756
/*
A
Andrew Morton 已提交
757
 * shrink_page_list() returns the number of reclaimed pages
L
Linus Torvalds 已提交
758
 */
A
Andrew Morton 已提交
759
static unsigned long shrink_page_list(struct list_head *page_list,
760
				      struct zone *zone,
761
				      struct scan_control *sc)
L
Linus Torvalds 已提交
762 763
{
	LIST_HEAD(ret_pages);
764
	LIST_HEAD(free_pages);
L
Linus Torvalds 已提交
765
	int pgactivate = 0;
766 767
	unsigned long nr_dirty = 0;
	unsigned long nr_congested = 0;
768
	unsigned long nr_reclaimed = 0;
L
Linus Torvalds 已提交
769 770 771 772

	cond_resched();

	while (!list_empty(page_list)) {
773
		enum page_references references;
L
Linus Torvalds 已提交
774 775 776 777 778 779 780 781 782
		struct address_space *mapping;
		struct page *page;
		int may_enter_fs;

		cond_resched();

		page = lru_to_page(page_list);
		list_del(&page->lru);

N
Nick Piggin 已提交
783
		if (!trylock_page(page))
L
Linus Torvalds 已提交
784 785
			goto keep;

N
Nick Piggin 已提交
786
		VM_BUG_ON(PageActive(page));
787
		VM_BUG_ON(page_zone(page) != zone);
L
Linus Torvalds 已提交
788 789

		sc->nr_scanned++;
790

N
Nick Piggin 已提交
791 792
		if (unlikely(!page_evictable(page, NULL)))
			goto cull_mlocked;
L
Lee Schermerhorn 已提交
793

794
		if (!sc->may_unmap && page_mapped(page))
795 796
			goto keep_locked;

L
Linus Torvalds 已提交
797 798 799 800
		/* Double the slab pressure for mapped and swapcache pages */
		if (page_mapped(page) || PageSwapCache(page))
			sc->nr_scanned++;

801 802 803 804 805 806 807 808 809 810 811 812
		may_enter_fs = (sc->gfp_mask & __GFP_FS) ||
			(PageSwapCache(page) && (sc->gfp_mask & __GFP_IO));

		if (PageWriteback(page)) {
			/*
			 * Synchronous reclaim is performed in two passes,
			 * first an asynchronous pass over the list to
			 * start parallel writeback, and a second synchronous
			 * pass to wait for the IO to complete.  Wait here
			 * for any page for which writeback has already
			 * started.
			 */
813
			if ((sc->reclaim_mode & RECLAIM_MODE_SYNC) &&
814
			    may_enter_fs)
815
				wait_on_page_writeback(page);
816 817 818 819
			else {
				unlock_page(page);
				goto keep_lumpy;
			}
820
		}
L
Linus Torvalds 已提交
821

822 823 824
		references = page_check_references(page, sc);
		switch (references) {
		case PAGEREF_ACTIVATE:
L
Linus Torvalds 已提交
825
			goto activate_locked;
826 827
		case PAGEREF_KEEP:
			goto keep_locked;
828 829 830 831
		case PAGEREF_RECLAIM:
		case PAGEREF_RECLAIM_CLEAN:
			; /* try to reclaim the page below */
		}
L
Linus Torvalds 已提交
832 833 834 835 836

		/*
		 * Anonymous process memory has backing store?
		 * Try to allocate it some swap space here.
		 */
N
Nick Piggin 已提交
837
		if (PageAnon(page) && !PageSwapCache(page)) {
838 839
			if (!(sc->gfp_mask & __GFP_IO))
				goto keep_locked;
840
			if (!add_to_swap(page))
L
Linus Torvalds 已提交
841
				goto activate_locked;
842
			may_enter_fs = 1;
N
Nick Piggin 已提交
843
		}
L
Linus Torvalds 已提交
844 845 846 847 848 849 850 851

		mapping = page_mapping(page);

		/*
		 * The page is mapped into the page tables of one or more
		 * processes. Try to unmap it here.
		 */
		if (page_mapped(page) && mapping) {
852
			switch (try_to_unmap(page, TTU_UNMAP)) {
L
Linus Torvalds 已提交
853 854 855 856
			case SWAP_FAIL:
				goto activate_locked;
			case SWAP_AGAIN:
				goto keep_locked;
N
Nick Piggin 已提交
857 858
			case SWAP_MLOCK:
				goto cull_mlocked;
L
Linus Torvalds 已提交
859 860 861 862 863 864
			case SWAP_SUCCESS:
				; /* try to free the page below */
			}
		}

		if (PageDirty(page)) {
865 866
			nr_dirty++;

867
			if (references == PAGEREF_RECLAIM_CLEAN)
L
Linus Torvalds 已提交
868
				goto keep_locked;
869
			if (!may_enter_fs)
L
Linus Torvalds 已提交
870
				goto keep_locked;
871
			if (!sc->may_writepage)
L
Linus Torvalds 已提交
872 873 874
				goto keep_locked;

			/* Page is dirty, try to write it out here */
875
			switch (pageout(page, mapping, sc)) {
L
Linus Torvalds 已提交
876
			case PAGE_KEEP:
877
				nr_congested++;
L
Linus Torvalds 已提交
878 879 880 881
				goto keep_locked;
			case PAGE_ACTIVATE:
				goto activate_locked;
			case PAGE_SUCCESS:
882 883 884
				if (PageWriteback(page))
					goto keep_lumpy;
				if (PageDirty(page))
L
Linus Torvalds 已提交
885
					goto keep;
886

L
Linus Torvalds 已提交
887 888 889 890
				/*
				 * A synchronous write - probably a ramdisk.  Go
				 * ahead and try to reclaim the page.
				 */
N
Nick Piggin 已提交
891
				if (!trylock_page(page))
L
Linus Torvalds 已提交
892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910
					goto keep;
				if (PageDirty(page) || PageWriteback(page))
					goto keep_locked;
				mapping = page_mapping(page);
			case PAGE_CLEAN:
				; /* try to free the page below */
			}
		}

		/*
		 * If the page has buffers, try to free the buffer mappings
		 * associated with this page. If we succeed we try to free
		 * the page as well.
		 *
		 * We do this even if the page is PageDirty().
		 * try_to_release_page() does not perform I/O, but it is
		 * possible for a page to have PageDirty set, but it is actually
		 * clean (all its buffers are clean).  This happens if the
		 * buffers were written out directly, with submit_bh(). ext3
L
Lee Schermerhorn 已提交
911
		 * will do this, as well as the blockdev mapping.
L
Linus Torvalds 已提交
912 913 914 915 916 917 918 919 920 921
		 * try_to_release_page() will discover that cleanness and will
		 * drop the buffers and mark the page clean - it can be freed.
		 *
		 * Rarely, pages can have buffers and no ->mapping.  These are
		 * the pages which were not successfully invalidated in
		 * truncate_complete_page().  We try to drop those buffers here
		 * and if that worked, and the page is no longer mapped into
		 * process address space (page_count == 1) it can be freed.
		 * Otherwise, leave the page on the LRU so it is swappable.
		 */
922
		if (page_has_private(page)) {
L
Linus Torvalds 已提交
923 924
			if (!try_to_release_page(page, sc->gfp_mask))
				goto activate_locked;
N
Nick Piggin 已提交
925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940
			if (!mapping && page_count(page) == 1) {
				unlock_page(page);
				if (put_page_testzero(page))
					goto free_it;
				else {
					/*
					 * rare race with speculative reference.
					 * the speculative reference will free
					 * this page shortly, so we may
					 * increment nr_reclaimed here (and
					 * leave it off the LRU).
					 */
					nr_reclaimed++;
					continue;
				}
			}
L
Linus Torvalds 已提交
941 942
		}

N
Nick Piggin 已提交
943
		if (!mapping || !__remove_mapping(mapping, page))
944
			goto keep_locked;
L
Linus Torvalds 已提交
945

N
Nick Piggin 已提交
946 947 948 949 950 951 952 953
		/*
		 * At this point, we have no other references and there is
		 * no way to pick any more up (removed from LRU, removed
		 * from pagecache). Can use non-atomic bitops now (and
		 * we obviously don't have to worry about waking up a process
		 * waiting on the page lock, because there are no references.
		 */
		__clear_page_locked(page);
N
Nick Piggin 已提交
954
free_it:
955
		nr_reclaimed++;
956 957 958 959 960 961

		/*
		 * Is there need to periodically free_page_list? It would
		 * appear not as the counts should be low
		 */
		list_add(&page->lru, &free_pages);
L
Linus Torvalds 已提交
962 963
		continue;

N
Nick Piggin 已提交
964
cull_mlocked:
965 966
		if (PageSwapCache(page))
			try_to_free_swap(page);
N
Nick Piggin 已提交
967 968
		unlock_page(page);
		putback_lru_page(page);
969
		reset_reclaim_mode(sc);
N
Nick Piggin 已提交
970 971
		continue;

L
Linus Torvalds 已提交
972
activate_locked:
973 974
		/* Not a candidate for swapping, so reclaim swap space. */
		if (PageSwapCache(page) && vm_swap_full())
975
			try_to_free_swap(page);
L
Lee Schermerhorn 已提交
976
		VM_BUG_ON(PageActive(page));
L
Linus Torvalds 已提交
977 978 979 980 981
		SetPageActive(page);
		pgactivate++;
keep_locked:
		unlock_page(page);
keep:
982
		reset_reclaim_mode(sc);
983
keep_lumpy:
L
Linus Torvalds 已提交
984
		list_add(&page->lru, &ret_pages);
N
Nick Piggin 已提交
985
		VM_BUG_ON(PageLRU(page) || PageUnevictable(page));
L
Linus Torvalds 已提交
986
	}
987

988 989 990 991 992 993
	/*
	 * Tag a zone as congested if all the dirty pages encountered were
	 * backed by a congested BDI. In this case, reclaimers should just
	 * back off and wait for congestion to clear because further reclaim
	 * will encounter the same problem
	 */
K
KAMEZAWA Hiroyuki 已提交
994
	if (nr_dirty && nr_dirty == nr_congested && scanning_global_lru(sc))
995 996
		zone_set_flag(zone, ZONE_CONGESTED);

997 998
	free_page_list(&free_pages);

L
Linus Torvalds 已提交
999
	list_splice(&ret_pages, page_list);
1000
	count_vm_events(PGACTIVATE, pgactivate);
1001
	return nr_reclaimed;
L
Linus Torvalds 已提交
1002 1003
}

A
Andy Whitcroft 已提交
1004 1005 1006 1007 1008 1009 1010 1011 1012 1013
/*
 * Attempt to remove the specified page from its LRU.  Only take this page
 * if it is of the appropriate PageActive status.  Pages which are being
 * freed elsewhere are also ignored.
 *
 * page:	page to consider
 * mode:	one of the LRU isolation modes defined above
 *
 * returns 0 on success, -ve errno on failure.
 */
1014
int __isolate_lru_page(struct page *page, int mode, int file)
A
Andy Whitcroft 已提交
1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029
{
	int ret = -EINVAL;

	/* Only take pages on the LRU. */
	if (!PageLRU(page))
		return ret;

	/*
	 * When checking the active state, we need to be sure we are
	 * dealing with comparible boolean values.  Take the logical not
	 * of each.
	 */
	if (mode != ISOLATE_BOTH && (!PageActive(page) != !mode))
		return ret;

1030
	if (mode != ISOLATE_BOTH && page_is_file_cache(page) != file)
1031 1032
		return ret;

L
Lee Schermerhorn 已提交
1033 1034 1035 1036 1037 1038 1039 1040
	/*
	 * When this function is being called for lumpy reclaim, we
	 * initially look into all LRU pages, active, inactive and
	 * unevictable; only give shrink_page_list evictable pages.
	 */
	if (PageUnevictable(page))
		return ret;

A
Andy Whitcroft 已提交
1041
	ret = -EBUSY;
K
KAMEZAWA Hiroyuki 已提交
1042

A
Andy Whitcroft 已提交
1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055
	if (likely(get_page_unless_zero(page))) {
		/*
		 * Be careful not to clear PageLRU until after we're
		 * sure the page is not being freed elsewhere -- the
		 * page release code relies on it.
		 */
		ClearPageLRU(page);
		ret = 0;
	}

	return ret;
}

L
Linus Torvalds 已提交
1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069
/*
 * zone->lru_lock is heavily contended.  Some of the functions that
 * shrink the lists perform better by taking out a batch of pages
 * and working on them outside the LRU lock.
 *
 * For pagecache intensive workloads, this function is the hottest
 * spot in the kernel (apart from copy_*_user functions).
 *
 * Appropriate locks must be held before calling this function.
 *
 * @nr_to_scan:	The number of pages to look through on the list.
 * @src:	The LRU list to pull pages off.
 * @dst:	The temp list to put pages on to.
 * @scanned:	The number of pages that were scanned.
A
Andy Whitcroft 已提交
1070 1071
 * @order:	The caller's attempted allocation order
 * @mode:	One of the LRU isolation modes
1072
 * @file:	True [1] if isolating file [!anon] pages
L
Linus Torvalds 已提交
1073 1074 1075
 *
 * returns how many pages were moved onto *@dst.
 */
1076 1077
static unsigned long isolate_lru_pages(unsigned long nr_to_scan,
		struct list_head *src, struct list_head *dst,
1078
		unsigned long *scanned, int order, int mode, int file)
L
Linus Torvalds 已提交
1079
{
1080
	unsigned long nr_taken = 0;
1081 1082 1083
	unsigned long nr_lumpy_taken = 0;
	unsigned long nr_lumpy_dirty = 0;
	unsigned long nr_lumpy_failed = 0;
1084
	unsigned long scan;
L
Linus Torvalds 已提交
1085

1086
	for (scan = 0; scan < nr_to_scan && !list_empty(src); scan++) {
A
Andy Whitcroft 已提交
1087 1088 1089 1090 1091 1092
		struct page *page;
		unsigned long pfn;
		unsigned long end_pfn;
		unsigned long page_pfn;
		int zone_id;

L
Linus Torvalds 已提交
1093 1094 1095
		page = lru_to_page(src);
		prefetchw_prev_lru_page(page, src, flags);

N
Nick Piggin 已提交
1096
		VM_BUG_ON(!PageLRU(page));
N
Nick Piggin 已提交
1097

1098
		switch (__isolate_lru_page(page, mode, file)) {
A
Andy Whitcroft 已提交
1099 1100
		case 0:
			list_move(&page->lru, dst);
1101
			mem_cgroup_del_lru(page);
1102
			nr_taken += hpage_nr_pages(page);
A
Andy Whitcroft 已提交
1103 1104 1105 1106 1107
			break;

		case -EBUSY:
			/* else it is being freed elsewhere */
			list_move(&page->lru, src);
1108
			mem_cgroup_rotate_lru_list(page, page_lru(page));
A
Andy Whitcroft 已提交
1109
			continue;
1110

A
Andy Whitcroft 已提交
1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122
		default:
			BUG();
		}

		if (!order)
			continue;

		/*
		 * Attempt to take all pages in the order aligned region
		 * surrounding the tag page.  Only take those pages of
		 * the same active state as that tag page.  We may safely
		 * round the target page pfn down to the requested order
L
Lucas De Marchi 已提交
1123
		 * as the mem_map is guaranteed valid out to MAX_ORDER,
A
Andy Whitcroft 已提交
1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142
		 * where that page is in a different zone we will detect
		 * it from its zone id and abort this block scan.
		 */
		zone_id = page_zone_id(page);
		page_pfn = page_to_pfn(page);
		pfn = page_pfn & ~((1 << order) - 1);
		end_pfn = pfn + (1 << order);
		for (; pfn < end_pfn; pfn++) {
			struct page *cursor_page;

			/* The target page is in the block, ignore it. */
			if (unlikely(pfn == page_pfn))
				continue;

			/* Avoid holes within the zone. */
			if (unlikely(!pfn_valid_within(pfn)))
				break;

			cursor_page = pfn_to_page(pfn);
1143

A
Andy Whitcroft 已提交
1144 1145
			/* Check that we have not crossed a zone boundary. */
			if (unlikely(page_zone_id(cursor_page) != zone_id))
1146
				break;
1147 1148 1149 1150 1151 1152 1153

			/*
			 * If we don't have enough swap space, reclaiming of
			 * anon page which don't already have a swap slot is
			 * pointless.
			 */
			if (nr_swap_pages <= 0 && PageAnon(cursor_page) &&
1154 1155
			    !PageSwapCache(cursor_page))
				break;
1156

1157
			if (__isolate_lru_page(cursor_page, mode, file) == 0) {
A
Andy Whitcroft 已提交
1158
				list_move(&cursor_page->lru, dst);
1159
				mem_cgroup_del_lru(cursor_page);
1160
				nr_taken += hpage_nr_pages(page);
1161 1162 1163
				nr_lumpy_taken++;
				if (PageDirty(cursor_page))
					nr_lumpy_dirty++;
A
Andy Whitcroft 已提交
1164
				scan++;
1165
			} else {
1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179
				/*
				 * Check if the page is freed already.
				 *
				 * We can't use page_count() as that
				 * requires compound_head and we don't
				 * have a pin on the page here. If a
				 * page is tail, we may or may not
				 * have isolated the head, so assume
				 * it's not free, it'd be tricky to
				 * track the head status without a
				 * page pin.
				 */
				if (!PageTail(cursor_page) &&
				    !atomic_read(&cursor_page->_count))
1180 1181
					continue;
				break;
A
Andy Whitcroft 已提交
1182 1183
			}
		}
1184 1185 1186 1187

		/* If we break out of the loop above, lumpy reclaim failed */
		if (pfn < end_pfn)
			nr_lumpy_failed++;
L
Linus Torvalds 已提交
1188 1189 1190
	}

	*scanned = scan;
1191 1192 1193 1194 1195 1196

	trace_mm_vmscan_lru_isolate(order,
			nr_to_scan, scan,
			nr_taken,
			nr_lumpy_taken, nr_lumpy_dirty, nr_lumpy_failed,
			mode);
L
Linus Torvalds 已提交
1197 1198 1199
	return nr_taken;
}

1200 1201 1202 1203
static unsigned long isolate_pages_global(unsigned long nr,
					struct list_head *dst,
					unsigned long *scanned, int order,
					int mode, struct zone *z,
1204
					int active, int file)
1205
{
1206
	int lru = LRU_BASE;
1207
	if (active)
1208 1209 1210 1211
		lru += LRU_ACTIVE;
	if (file)
		lru += LRU_FILE;
	return isolate_lru_pages(nr, &z->lru[lru].list, dst, scanned, order,
1212
								mode, file);
1213 1214
}

A
Andy Whitcroft 已提交
1215 1216 1217 1218
/*
 * clear_active_flags() is a helper for shrink_active_list(), clearing
 * any active bits from the pages in the list.
 */
1219 1220
static unsigned long clear_active_flags(struct list_head *page_list,
					unsigned int *count)
A
Andy Whitcroft 已提交
1221 1222
{
	int nr_active = 0;
1223
	int lru;
A
Andy Whitcroft 已提交
1224 1225
	struct page *page;

1226
	list_for_each_entry(page, page_list, lru) {
1227
		int numpages = hpage_nr_pages(page);
1228
		lru = page_lru_base_type(page);
A
Andy Whitcroft 已提交
1229
		if (PageActive(page)) {
1230
			lru += LRU_ACTIVE;
A
Andy Whitcroft 已提交
1231
			ClearPageActive(page);
1232
			nr_active += numpages;
A
Andy Whitcroft 已提交
1233
		}
1234
		if (count)
1235
			count[lru] += numpages;
1236
	}
A
Andy Whitcroft 已提交
1237 1238 1239 1240

	return nr_active;
}

1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251
/**
 * isolate_lru_page - tries to isolate a page from its LRU list
 * @page: page to isolate from its LRU list
 *
 * Isolates a @page from an LRU list, clears PageLRU and adjusts the
 * vmstat statistic corresponding to whatever LRU list the page was on.
 *
 * Returns 0 if the page was removed from an LRU list.
 * Returns -EBUSY if the page was not on an LRU list.
 *
 * The returned page will have PageLRU() cleared.  If it was found on
L
Lee Schermerhorn 已提交
1252 1253 1254
 * the active list, it will have PageActive set.  If it was found on
 * the unevictable list, it will have the PageUnevictable bit set. That flag
 * may need to be cleared by the caller before letting the page go.
1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269
 *
 * The vmstat statistic corresponding to the list on which the page was
 * found will be decremented.
 *
 * Restrictions:
 * (1) Must be called with an elevated refcount on the page. This is a
 *     fundamentnal difference from isolate_lru_pages (which is called
 *     without a stable reference).
 * (2) the lru_lock must not be held.
 * (3) interrupts must be enabled.
 */
int isolate_lru_page(struct page *page)
{
	int ret = -EBUSY;

1270 1271
	VM_BUG_ON(!page_count(page));

1272 1273 1274 1275
	if (PageLRU(page)) {
		struct zone *zone = page_zone(page);

		spin_lock_irq(&zone->lru_lock);
1276
		if (PageLRU(page)) {
L
Lee Schermerhorn 已提交
1277
			int lru = page_lru(page);
1278
			ret = 0;
1279
			get_page(page);
1280
			ClearPageLRU(page);
1281 1282

			del_page_from_lru_list(zone, page, lru);
1283 1284 1285 1286 1287 1288
		}
		spin_unlock_irq(&zone->lru_lock);
	}
	return ret;
}

1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313
/*
 * Are there way too many processes in the direct reclaim path already?
 */
static int too_many_isolated(struct zone *zone, int file,
		struct scan_control *sc)
{
	unsigned long inactive, isolated;

	if (current_is_kswapd())
		return 0;

	if (!scanning_global_lru(sc))
		return 0;

	if (file) {
		inactive = zone_page_state(zone, NR_INACTIVE_FILE);
		isolated = zone_page_state(zone, NR_ISOLATED_FILE);
	} else {
		inactive = zone_page_state(zone, NR_INACTIVE_ANON);
		isolated = zone_page_state(zone, NR_ISOLATED_ANON);
	}

	return isolated > inactive;
}

1314 1315 1316 1317
/*
 * TODO: Try merging with migrations version of putback_lru_pages
 */
static noinline_for_stack void
1318
putback_lru_pages(struct zone *zone, struct scan_control *sc,
1319 1320 1321 1322 1323
				unsigned long nr_anon, unsigned long nr_file,
				struct list_head *page_list)
{
	struct page *page;
	struct pagevec pvec;
1324
	struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(zone, sc);
1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342

	pagevec_init(&pvec, 1);

	/*
	 * Put back any unfreeable pages.
	 */
	spin_lock(&zone->lru_lock);
	while (!list_empty(page_list)) {
		int lru;
		page = lru_to_page(page_list);
		VM_BUG_ON(PageLRU(page));
		list_del(&page->lru);
		if (unlikely(!page_evictable(page, NULL))) {
			spin_unlock_irq(&zone->lru_lock);
			putback_lru_page(page);
			spin_lock_irq(&zone->lru_lock);
			continue;
		}
1343
		SetPageLRU(page);
1344
		lru = page_lru(page);
1345
		add_page_to_lru_list(zone, page, lru);
1346 1347
		if (is_active_lru(lru)) {
			int file = is_file_lru(lru);
1348 1349
			int numpages = hpage_nr_pages(page);
			reclaim_stat->recent_rotated[file] += numpages;
1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363
		}
		if (!pagevec_add(&pvec, page)) {
			spin_unlock_irq(&zone->lru_lock);
			__pagevec_release(&pvec);
			spin_lock_irq(&zone->lru_lock);
		}
	}
	__mod_zone_page_state(zone, NR_ISOLATED_ANON, -nr_anon);
	__mod_zone_page_state(zone, NR_ISOLATED_FILE, -nr_file);

	spin_unlock_irq(&zone->lru_lock);
	pagevec_release(&pvec);
}

1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394
static noinline_for_stack void update_isolated_counts(struct zone *zone,
					struct scan_control *sc,
					unsigned long *nr_anon,
					unsigned long *nr_file,
					struct list_head *isolated_list)
{
	unsigned long nr_active;
	unsigned int count[NR_LRU_LISTS] = { 0, };
	struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(zone, sc);

	nr_active = clear_active_flags(isolated_list, count);
	__count_vm_events(PGDEACTIVATE, nr_active);

	__mod_zone_page_state(zone, NR_ACTIVE_FILE,
			      -count[LRU_ACTIVE_FILE]);
	__mod_zone_page_state(zone, NR_INACTIVE_FILE,
			      -count[LRU_INACTIVE_FILE]);
	__mod_zone_page_state(zone, NR_ACTIVE_ANON,
			      -count[LRU_ACTIVE_ANON]);
	__mod_zone_page_state(zone, NR_INACTIVE_ANON,
			      -count[LRU_INACTIVE_ANON]);

	*nr_anon = count[LRU_ACTIVE_ANON] + count[LRU_INACTIVE_ANON];
	*nr_file = count[LRU_ACTIVE_FILE] + count[LRU_INACTIVE_FILE];
	__mod_zone_page_state(zone, NR_ISOLATED_ANON, *nr_anon);
	__mod_zone_page_state(zone, NR_ISOLATED_FILE, *nr_file);

	reclaim_stat->recent_scanned[0] += *nr_anon;
	reclaim_stat->recent_scanned[1] += *nr_file;
}

1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414
/*
 * Returns true if the caller should wait to clean dirty/writeback pages.
 *
 * If we are direct reclaiming for contiguous pages and we do not reclaim
 * everything in the list, try again and wait for writeback IO to complete.
 * This will stall high-order allocations noticeably. Only do that when really
 * need to free the pages under high memory pressure.
 */
static inline bool should_reclaim_stall(unsigned long nr_taken,
					unsigned long nr_freed,
					int priority,
					struct scan_control *sc)
{
	int lumpy_stall_priority;

	/* kswapd should not stall on sync IO */
	if (current_is_kswapd())
		return false;

	/* Only stall on lumpy reclaim */
1415
	if (sc->reclaim_mode & RECLAIM_MODE_SINGLE)
1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435
		return false;

	/* If we have relaimed everything on the isolated list, no stall */
	if (nr_freed == nr_taken)
		return false;

	/*
	 * For high-order allocations, there are two stall thresholds.
	 * High-cost allocations stall immediately where as lower
	 * order allocations such as stacks require the scanning
	 * priority to be much higher before stalling.
	 */
	if (sc->order > PAGE_ALLOC_COSTLY_ORDER)
		lumpy_stall_priority = DEF_PRIORITY;
	else
		lumpy_stall_priority = DEF_PRIORITY / 3;

	return priority <= lumpy_stall_priority;
}

L
Linus Torvalds 已提交
1436
/*
A
Andrew Morton 已提交
1437 1438
 * shrink_inactive_list() is a helper for shrink_zone().  It returns the number
 * of reclaimed pages
L
Linus Torvalds 已提交
1439
 */
1440 1441 1442
static noinline_for_stack unsigned long
shrink_inactive_list(unsigned long nr_to_scan, struct zone *zone,
			struct scan_control *sc, int priority, int file)
L
Linus Torvalds 已提交
1443 1444
{
	LIST_HEAD(page_list);
1445
	unsigned long nr_scanned;
1446
	unsigned long nr_reclaimed = 0;
1447 1448 1449
	unsigned long nr_taken;
	unsigned long nr_anon;
	unsigned long nr_file;
1450

1451
	while (unlikely(too_many_isolated(zone, file, sc))) {
1452
		congestion_wait(BLK_RW_ASYNC, HZ/10);
1453 1454 1455 1456 1457 1458

		/* We are about to die and free our memory. Return now. */
		if (fatal_signal_pending(current))
			return SWAP_CLUSTER_MAX;
	}

1459
	set_reclaim_mode(priority, sc, false);
L
Linus Torvalds 已提交
1460 1461
	lru_add_drain();
	spin_lock_irq(&zone->lru_lock);
1462

1463 1464 1465
	if (scanning_global_lru(sc)) {
		nr_taken = isolate_pages_global(nr_to_scan,
			&page_list, &nr_scanned, sc->order,
1466
			sc->reclaim_mode & RECLAIM_MODE_LUMPYRECLAIM ?
1467
					ISOLATE_BOTH : ISOLATE_INACTIVE,
1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478
			zone, 0, file);
		zone->pages_scanned += nr_scanned;
		if (current_is_kswapd())
			__count_zone_vm_events(PGSCAN_KSWAPD, zone,
					       nr_scanned);
		else
			__count_zone_vm_events(PGSCAN_DIRECT, zone,
					       nr_scanned);
	} else {
		nr_taken = mem_cgroup_isolate_pages(nr_to_scan,
			&page_list, &nr_scanned, sc->order,
1479
			sc->reclaim_mode & RECLAIM_MODE_LUMPYRECLAIM ?
1480
					ISOLATE_BOTH : ISOLATE_INACTIVE,
1481 1482 1483 1484 1485 1486 1487
			zone, sc->mem_cgroup,
			0, file);
		/*
		 * mem_cgroup_isolate_pages() keeps track of
		 * scanned pages on its own.
		 */
	}
1488

1489 1490 1491 1492
	if (nr_taken == 0) {
		spin_unlock_irq(&zone->lru_lock);
		return 0;
	}
A
Andy Whitcroft 已提交
1493

1494
	update_isolated_counts(zone, sc, &nr_anon, &nr_file, &page_list);
L
Linus Torvalds 已提交
1495

1496
	spin_unlock_irq(&zone->lru_lock);
1497

1498
	nr_reclaimed = shrink_page_list(&page_list, zone, sc);
1499

1500 1501
	/* Check if we should syncronously wait for writeback */
	if (should_reclaim_stall(nr_taken, nr_reclaimed, priority, sc)) {
1502
		set_reclaim_mode(priority, sc, true);
1503
		nr_reclaimed += shrink_page_list(&page_list, zone, sc);
1504
	}
1505

1506 1507 1508 1509
	local_irq_disable();
	if (current_is_kswapd())
		__count_vm_events(KSWAPD_STEAL, nr_reclaimed);
	__count_zone_vm_events(PGSTEAL, zone, nr_reclaimed);
N
Nick Piggin 已提交
1510

1511
	putback_lru_pages(zone, sc, nr_anon, nr_file, &page_list);
1512 1513 1514 1515 1516

	trace_mm_vmscan_lru_shrink_inactive(zone->zone_pgdat->node_id,
		zone_idx(zone),
		nr_scanned, nr_reclaimed,
		priority,
1517
		trace_shrink_flags(file, sc->reclaim_mode));
1518
	return nr_reclaimed;
L
Linus Torvalds 已提交
1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537
}

/*
 * This moves pages from the active list to the inactive list.
 *
 * We move them the other way if the page is referenced by one or more
 * processes, from rmap.
 *
 * If the pages are mostly unmapped, the processing is fast and it is
 * appropriate to hold zone->lru_lock across the whole operation.  But if
 * the pages are mapped, the processing is slow (page_referenced()) so we
 * should drop zone->lru_lock around each page.  It's impossible to balance
 * this, so instead we remove the pages from the LRU while processing them.
 * It is safe to rely on PG_active against the non-LRU pages in here because
 * nobody will play with that bit on a non-LRU page.
 *
 * The downside is that we have to touch page->_count against each page.
 * But we had to alter page->flags anyway.
 */
1538

1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556
static void move_active_pages_to_lru(struct zone *zone,
				     struct list_head *list,
				     enum lru_list lru)
{
	unsigned long pgmoved = 0;
	struct pagevec pvec;
	struct page *page;

	pagevec_init(&pvec, 1);

	while (!list_empty(list)) {
		page = lru_to_page(list);

		VM_BUG_ON(PageLRU(page));
		SetPageLRU(page);

		list_move(&page->lru, &zone->lru[lru].list);
		mem_cgroup_add_lru_list(page, lru);
1557
		pgmoved += hpage_nr_pages(page);
1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570

		if (!pagevec_add(&pvec, page) || list_empty(list)) {
			spin_unlock_irq(&zone->lru_lock);
			if (buffer_heads_over_limit)
				pagevec_strip(&pvec);
			__pagevec_release(&pvec);
			spin_lock_irq(&zone->lru_lock);
		}
	}
	__mod_zone_page_state(zone, NR_LRU_BASE + lru, pgmoved);
	if (!is_active_lru(lru))
		__count_vm_events(PGDEACTIVATE, pgmoved);
}
1571

A
Andrew Morton 已提交
1572
static void shrink_active_list(unsigned long nr_pages, struct zone *zone,
1573
			struct scan_control *sc, int priority, int file)
L
Linus Torvalds 已提交
1574
{
1575
	unsigned long nr_taken;
1576
	unsigned long pgscanned;
1577
	unsigned long vm_flags;
L
Linus Torvalds 已提交
1578
	LIST_HEAD(l_hold);	/* The pages which were snipped off */
1579
	LIST_HEAD(l_active);
1580
	LIST_HEAD(l_inactive);
L
Linus Torvalds 已提交
1581
	struct page *page;
1582
	struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(zone, sc);
1583
	unsigned long nr_rotated = 0;
L
Linus Torvalds 已提交
1584 1585 1586

	lru_add_drain();
	spin_lock_irq(&zone->lru_lock);
1587
	if (scanning_global_lru(sc)) {
1588 1589 1590 1591
		nr_taken = isolate_pages_global(nr_pages, &l_hold,
						&pgscanned, sc->order,
						ISOLATE_ACTIVE, zone,
						1, file);
1592
		zone->pages_scanned += pgscanned;
1593 1594 1595 1596 1597 1598 1599 1600 1601
	} else {
		nr_taken = mem_cgroup_isolate_pages(nr_pages, &l_hold,
						&pgscanned, sc->order,
						ISOLATE_ACTIVE, zone,
						sc->mem_cgroup, 1, file);
		/*
		 * mem_cgroup_isolate_pages() keeps track of
		 * scanned pages on its own.
		 */
1602
	}
1603

1604
	reclaim_stat->recent_scanned[file] += nr_taken;
1605

1606
	__count_zone_vm_events(PGREFILL, zone, pgscanned);
1607
	if (file)
1608
		__mod_zone_page_state(zone, NR_ACTIVE_FILE, -nr_taken);
1609
	else
1610
		__mod_zone_page_state(zone, NR_ACTIVE_ANON, -nr_taken);
K
KOSAKI Motohiro 已提交
1611
	__mod_zone_page_state(zone, NR_ISOLATED_ANON + file, nr_taken);
L
Linus Torvalds 已提交
1612 1613 1614 1615 1616 1617
	spin_unlock_irq(&zone->lru_lock);

	while (!list_empty(&l_hold)) {
		cond_resched();
		page = lru_to_page(&l_hold);
		list_del(&page->lru);
1618

L
Lee Schermerhorn 已提交
1619 1620 1621 1622 1623
		if (unlikely(!page_evictable(page, NULL))) {
			putback_lru_page(page);
			continue;
		}

1624
		if (page_referenced(page, 0, sc->mem_cgroup, &vm_flags)) {
1625
			nr_rotated += hpage_nr_pages(page);
1626 1627 1628 1629 1630 1631 1632 1633 1634
			/*
			 * Identify referenced, file-backed active pages and
			 * give them one more trip around the active list. So
			 * that executable code get better chances to stay in
			 * memory under moderate memory pressure.  Anon pages
			 * are not likely to be evicted by use-once streaming
			 * IO, plus JVM can create lots of anon VM_EXEC pages,
			 * so we ignore them here.
			 */
1635
			if ((vm_flags & VM_EXEC) && page_is_file_cache(page)) {
1636 1637 1638 1639
				list_add(&page->lru, &l_active);
				continue;
			}
		}
1640

1641
		ClearPageActive(page);	/* we are de-activating */
L
Linus Torvalds 已提交
1642 1643 1644
		list_add(&page->lru, &l_inactive);
	}

1645
	/*
1646
	 * Move pages back to the lru list.
1647
	 */
1648
	spin_lock_irq(&zone->lru_lock);
1649
	/*
1650 1651 1652 1653
	 * Count referenced pages from currently used mappings as rotated,
	 * even though only some of them are actually re-activated.  This
	 * helps balance scan pressure between file and anonymous pages in
	 * get_scan_ratio.
1654
	 */
1655
	reclaim_stat->recent_rotated[file] += nr_rotated;
1656

1657 1658 1659 1660
	move_active_pages_to_lru(zone, &l_active,
						LRU_ACTIVE + file * LRU_FILE);
	move_active_pages_to_lru(zone, &l_inactive,
						LRU_BASE   + file * LRU_FILE);
K
KOSAKI Motohiro 已提交
1661
	__mod_zone_page_state(zone, NR_ISOLATED_ANON + file, -nr_taken);
1662
	spin_unlock_irq(&zone->lru_lock);
L
Linus Torvalds 已提交
1663 1664
}

1665
#ifdef CONFIG_SWAP
1666
static int inactive_anon_is_low_global(struct zone *zone)
1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678
{
	unsigned long active, inactive;

	active = zone_page_state(zone, NR_ACTIVE_ANON);
	inactive = zone_page_state(zone, NR_INACTIVE_ANON);

	if (inactive * zone->inactive_ratio < active)
		return 1;

	return 0;
}

1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690
/**
 * inactive_anon_is_low - check if anonymous pages need to be deactivated
 * @zone: zone to check
 * @sc:   scan control of this context
 *
 * Returns true if the zone does not have enough inactive anon pages,
 * meaning some active anon pages need to be deactivated.
 */
static int inactive_anon_is_low(struct zone *zone, struct scan_control *sc)
{
	int low;

1691 1692 1693 1694 1695 1696 1697
	/*
	 * If we don't have swap space, anonymous page deactivation
	 * is pointless.
	 */
	if (!total_swap_pages)
		return 0;

1698
	if (scanning_global_lru(sc))
1699 1700
		low = inactive_anon_is_low_global(zone);
	else
1701
		low = mem_cgroup_inactive_anon_is_low(sc->mem_cgroup);
1702 1703
	return low;
}
1704 1705 1706 1707 1708 1709 1710
#else
static inline int inactive_anon_is_low(struct zone *zone,
					struct scan_control *sc)
{
	return 0;
}
#endif
1711

1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747
static int inactive_file_is_low_global(struct zone *zone)
{
	unsigned long active, inactive;

	active = zone_page_state(zone, NR_ACTIVE_FILE);
	inactive = zone_page_state(zone, NR_INACTIVE_FILE);

	return (active > inactive);
}

/**
 * inactive_file_is_low - check if file pages need to be deactivated
 * @zone: zone to check
 * @sc:   scan control of this context
 *
 * When the system is doing streaming IO, memory pressure here
 * ensures that active file pages get deactivated, until more
 * than half of the file pages are on the inactive list.
 *
 * Once we get to that situation, protect the system's working
 * set from being evicted by disabling active file page aging.
 *
 * This uses a different ratio than the anonymous pages, because
 * the page cache uses a use-once replacement algorithm.
 */
static int inactive_file_is_low(struct zone *zone, struct scan_control *sc)
{
	int low;

	if (scanning_global_lru(sc))
		low = inactive_file_is_low_global(zone);
	else
		low = mem_cgroup_inactive_file_is_low(sc->mem_cgroup);
	return low;
}

1748 1749 1750 1751 1752 1753 1754 1755 1756
static int inactive_list_is_low(struct zone *zone, struct scan_control *sc,
				int file)
{
	if (file)
		return inactive_file_is_low(zone, sc);
	else
		return inactive_anon_is_low(zone, sc);
}

1757
static unsigned long shrink_list(enum lru_list lru, unsigned long nr_to_scan,
1758 1759
	struct zone *zone, struct scan_control *sc, int priority)
{
1760 1761
	int file = is_file_lru(lru);

1762 1763 1764
	if (is_active_lru(lru)) {
		if (inactive_list_is_low(zone, sc, file))
		    shrink_active_list(nr_to_scan, zone, sc, priority, file);
1765 1766 1767
		return 0;
	}

R
Rik van Riel 已提交
1768
	return shrink_inactive_list(nr_to_scan, zone, sc, priority, file);
1769 1770
}

1771 1772 1773 1774 1775 1776 1777
static int vmscan_swappiness(struct scan_control *sc)
{
	if (scanning_global_lru(sc))
		return vm_swappiness;
	return mem_cgroup_swappiness(sc->mem_cgroup);
}

1778 1779 1780 1781 1782 1783
/*
 * Determine how aggressively the anon and file LRU lists should be
 * scanned.  The relative value of each set of LRU lists is determined
 * by looking at the fraction of the pages scanned we did rotate back
 * onto the active list instead of evict.
 *
1784
 * nr[0] = anon pages to scan; nr[1] = file pages to scan
1785
 */
1786 1787
static void get_scan_count(struct zone *zone, struct scan_control *sc,
					unsigned long *nr, int priority)
1788 1789 1790 1791
{
	unsigned long anon, file, free;
	unsigned long anon_prio, file_prio;
	unsigned long ap, fp;
1792
	struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(zone, sc);
1793 1794 1795
	u64 fraction[2], denominator;
	enum lru_list l;
	int noswap = 0;
1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811
	int force_scan = 0;


	anon  = zone_nr_lru_pages(zone, sc, LRU_ACTIVE_ANON) +
		zone_nr_lru_pages(zone, sc, LRU_INACTIVE_ANON);
	file  = zone_nr_lru_pages(zone, sc, LRU_ACTIVE_FILE) +
		zone_nr_lru_pages(zone, sc, LRU_INACTIVE_FILE);

	if (((anon + file) >> priority) < SWAP_CLUSTER_MAX) {
		/* kswapd does zone balancing and need to scan this zone */
		if (scanning_global_lru(sc) && current_is_kswapd())
			force_scan = 1;
		/* memcg may have small limit and need to avoid priority drop */
		if (!scanning_global_lru(sc))
			force_scan = 1;
	}
1812 1813 1814 1815 1816 1817 1818 1819 1820

	/* If we have no swap space, do not bother scanning anon pages. */
	if (!sc->may_swap || (nr_swap_pages <= 0)) {
		noswap = 1;
		fraction[0] = 0;
		fraction[1] = 1;
		denominator = 1;
		goto out;
	}
1821

1822
	if (scanning_global_lru(sc)) {
1823 1824 1825
		free  = zone_page_state(zone, NR_FREE_PAGES);
		/* If we have very few page cache pages,
		   force-scan anon pages. */
1826
		if (unlikely(file + free <= high_wmark_pages(zone))) {
1827 1828 1829 1830
			fraction[0] = 1;
			fraction[1] = 0;
			denominator = 1;
			goto out;
1831
		}
1832 1833
	}

1834 1835 1836 1837
	/*
	 * With swappiness at 100, anonymous and file have the same priority.
	 * This scanning priority is essentially the inverse of IO cost.
	 */
1838 1839
	anon_prio = vmscan_swappiness(sc);
	file_prio = 200 - vmscan_swappiness(sc);
1840

1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851
	/*
	 * OK, so we have swap space and a fair amount of page cache
	 * pages.  We use the recently rotated / recently scanned
	 * ratios to determine how valuable each cache is.
	 *
	 * Because workloads change over time (and to avoid overflow)
	 * we keep these statistics as a floating average, which ends
	 * up weighing recent references more than old ones.
	 *
	 * anon in [0], file in [1]
	 */
1852
	spin_lock_irq(&zone->lru_lock);
1853 1854 1855
	if (unlikely(reclaim_stat->recent_scanned[0] > anon / 4)) {
		reclaim_stat->recent_scanned[0] /= 2;
		reclaim_stat->recent_rotated[0] /= 2;
1856 1857
	}

1858 1859 1860
	if (unlikely(reclaim_stat->recent_scanned[1] > file / 4)) {
		reclaim_stat->recent_scanned[1] /= 2;
		reclaim_stat->recent_rotated[1] /= 2;
1861 1862 1863
	}

	/*
1864 1865 1866
	 * The amount of pressure on anon vs file pages is inversely
	 * proportional to the fraction of recently scanned pages on
	 * each list that were recently referenced and in active use.
1867
	 */
1868 1869
	ap = (anon_prio + 1) * (reclaim_stat->recent_scanned[0] + 1);
	ap /= reclaim_stat->recent_rotated[0] + 1;
1870

1871 1872
	fp = (file_prio + 1) * (reclaim_stat->recent_scanned[1] + 1);
	fp /= reclaim_stat->recent_rotated[1] + 1;
1873
	spin_unlock_irq(&zone->lru_lock);
1874

1875 1876 1877 1878 1879 1880 1881
	fraction[0] = ap;
	fraction[1] = fp;
	denominator = ap + fp + 1;
out:
	for_each_evictable_lru(l) {
		int file = is_file_lru(l);
		unsigned long scan;
1882

1883 1884 1885 1886 1887
		scan = zone_nr_lru_pages(zone, sc, l);
		if (priority || noswap) {
			scan >>= priority;
			scan = div64_u64(scan * fraction[file], denominator);
		}
1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904

		/*
		 * If zone is small or memcg is small, nr[l] can be 0.
		 * This results no-scan on this priority and priority drop down.
		 * For global direct reclaim, it can visit next zone and tend
		 * not to have problems. For global kswapd, it's for zone
		 * balancing and it need to scan a small amounts. When using
		 * memcg, priority drop can cause big latency. So, it's better
		 * to scan small amount. See may_noscan above.
		 */
		if (!scan && force_scan) {
			if (file)
				scan = SWAP_CLUSTER_MAX;
			else if (!noswap)
				scan = SWAP_CLUSTER_MAX;
		}
		nr[l] = scan;
1905
	}
1906
}
1907

1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923
/*
 * Reclaim/compaction depends on a number of pages being freed. To avoid
 * disruption to the system, a small number of order-0 pages continue to be
 * rotated and reclaimed in the normal fashion. However, by the time we get
 * back to the allocator and call try_to_compact_zone(), we ensure that
 * there are enough free pages for it to be likely successful
 */
static inline bool should_continue_reclaim(struct zone *zone,
					unsigned long nr_reclaimed,
					unsigned long nr_scanned,
					struct scan_control *sc)
{
	unsigned long pages_for_compaction;
	unsigned long inactive_lru_pages;

	/* If not in reclaim/compaction mode, stop */
1924
	if (!(sc->reclaim_mode & RECLAIM_MODE_COMPACTION))
1925 1926
		return false;

1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948
	/* Consider stopping depending on scan and reclaim activity */
	if (sc->gfp_mask & __GFP_REPEAT) {
		/*
		 * For __GFP_REPEAT allocations, stop reclaiming if the
		 * full LRU list has been scanned and we are still failing
		 * to reclaim pages. This full LRU scan is potentially
		 * expensive but a __GFP_REPEAT caller really wants to succeed
		 */
		if (!nr_reclaimed && !nr_scanned)
			return false;
	} else {
		/*
		 * For non-__GFP_REPEAT allocations which can presumably
		 * fail without consequence, stop if we failed to reclaim
		 * any pages from the last SWAP_CLUSTER_MAX number of
		 * pages that were scanned. This will return to the
		 * caller faster at the risk reclaim/compaction and
		 * the resulting allocation attempt fails
		 */
		if (!nr_reclaimed)
			return false;
	}
1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970

	/*
	 * If we have not reclaimed enough pages for compaction and the
	 * inactive lists are large enough, continue reclaiming
	 */
	pages_for_compaction = (2UL << sc->order);
	inactive_lru_pages = zone_nr_lru_pages(zone, sc, LRU_INACTIVE_ANON) +
				zone_nr_lru_pages(zone, sc, LRU_INACTIVE_FILE);
	if (sc->nr_reclaimed < pages_for_compaction &&
			inactive_lru_pages > pages_for_compaction)
		return true;

	/* If compaction would go ahead or the allocation would succeed, stop */
	switch (compaction_suitable(zone, sc->order)) {
	case COMPACT_PARTIAL:
	case COMPACT_CONTINUE:
		return false;
	default:
		return true;
	}
}

L
Linus Torvalds 已提交
1971 1972 1973
/*
 * This is a basic per-zone page freer.  Used by both kswapd and direct reclaim.
 */
1974
static void shrink_zone(int priority, struct zone *zone,
1975
				struct scan_control *sc)
L
Linus Torvalds 已提交
1976
{
1977
	unsigned long nr[NR_LRU_LISTS];
1978
	unsigned long nr_to_scan;
1979
	enum lru_list l;
1980
	unsigned long nr_reclaimed, nr_scanned;
1981
	unsigned long nr_to_reclaim = sc->nr_to_reclaim;
1982

1983 1984
restart:
	nr_reclaimed = 0;
1985
	nr_scanned = sc->nr_scanned;
1986
	get_scan_count(zone, sc, nr, priority);
L
Linus Torvalds 已提交
1987

1988 1989
	while (nr[LRU_INACTIVE_ANON] || nr[LRU_ACTIVE_FILE] ||
					nr[LRU_INACTIVE_FILE]) {
L
Lee Schermerhorn 已提交
1990
		for_each_evictable_lru(l) {
1991
			if (nr[l]) {
K
KOSAKI Motohiro 已提交
1992 1993
				nr_to_scan = min_t(unsigned long,
						   nr[l], SWAP_CLUSTER_MAX);
1994
				nr[l] -= nr_to_scan;
L
Linus Torvalds 已提交
1995

1996 1997
				nr_reclaimed += shrink_list(l, nr_to_scan,
							    zone, sc, priority);
1998
			}
L
Linus Torvalds 已提交
1999
		}
2000 2001 2002 2003 2004 2005 2006 2007
		/*
		 * On large memory systems, scan >> priority can become
		 * really large. This is fine for the starting priority;
		 * we want to put equal scanning pressure on each zone.
		 * However, if the VM has a harder time of freeing pages,
		 * with multiple processes reclaiming pages, the total
		 * freeing target can get unreasonably large.
		 */
2008
		if (nr_reclaimed >= nr_to_reclaim && priority < DEF_PRIORITY)
2009
			break;
L
Linus Torvalds 已提交
2010
	}
2011
	sc->nr_reclaimed += nr_reclaimed;
2012

2013 2014 2015 2016
	/*
	 * Even if we did not try to evict anon pages at all, we want to
	 * rebalance the anon lru active/inactive ratio.
	 */
2017
	if (inactive_anon_is_low(zone, sc))
2018 2019
		shrink_active_list(SWAP_CLUSTER_MAX, zone, sc, priority, 0);

2020 2021 2022 2023 2024
	/* reclaim/compaction might need reclaim to continue */
	if (should_continue_reclaim(zone, nr_reclaimed,
					sc->nr_scanned - nr_scanned, sc))
		goto restart;

2025
	throttle_vm_writeout(sc->gfp_mask);
L
Linus Torvalds 已提交
2026 2027 2028 2029 2030 2031 2032
}

/*
 * This is the direct reclaim path, for page-allocating processes.  We only
 * try to reclaim pages from zones which will satisfy the caller's allocation
 * request.
 *
2033 2034
 * We reclaim from a zone even if that zone is over high_wmark_pages(zone).
 * Because:
L
Linus Torvalds 已提交
2035 2036
 * a) The caller may be trying to free *extra* pages to satisfy a higher-order
 *    allocation or
2037 2038 2039
 * b) The target zone may be at high_wmark_pages(zone) but the lower zones
 *    must go *over* high_wmark_pages(zone) to satisfy the `incremental min'
 *    zone defense algorithm.
L
Linus Torvalds 已提交
2040 2041 2042 2043
 *
 * If a zone is deemed to be full of pinned pages then just give it a light
 * scan then give up on it.
 */
2044
static void shrink_zones(int priority, struct zonelist *zonelist,
2045
					struct scan_control *sc)
L
Linus Torvalds 已提交
2046
{
2047
	struct zoneref *z;
2048
	struct zone *zone;
2049 2050
	unsigned long nr_soft_reclaimed;
	unsigned long nr_soft_scanned;
2051

2052 2053
	for_each_zone_zonelist_nodemask(zone, z, zonelist,
					gfp_zone(sc->gfp_mask), sc->nodemask) {
2054
		if (!populated_zone(zone))
L
Linus Torvalds 已提交
2055
			continue;
2056 2057 2058 2059
		/*
		 * Take care memory controller reclaiming has small influence
		 * to global LRU.
		 */
2060
		if (scanning_global_lru(sc)) {
2061 2062
			if (!cpuset_zone_allowed_hardwall(zone, GFP_KERNEL))
				continue;
2063
			if (zone->all_unreclaimable && priority != DEF_PRIORITY)
2064
				continue;	/* Let kswapd poll it */
2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077
			/*
			 * This steals pages from memory cgroups over softlimit
			 * and returns the number of reclaimed pages and
			 * scanned pages. This works for global memory pressure
			 * and balancing, not for a memcg's limit.
			 */
			nr_soft_scanned = 0;
			nr_soft_reclaimed = mem_cgroup_soft_limit_reclaim(zone,
						sc->order, sc->gfp_mask,
						&nr_soft_scanned);
			sc->nr_reclaimed += nr_soft_reclaimed;
			sc->nr_scanned += nr_soft_scanned;
			/* need some check for avoid more shrink_zone() */
2078
		}
2079

2080
		shrink_zone(priority, zone, sc);
L
Linus Torvalds 已提交
2081
	}
2082 2083 2084 2085 2086 2087 2088
}

static bool zone_reclaimable(struct zone *zone)
{
	return zone->pages_scanned < zone_reclaimable_pages(zone) * 6;
}

2089
/* All zones in zonelist are unreclaimable? */
2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101
static bool all_unreclaimable(struct zonelist *zonelist,
		struct scan_control *sc)
{
	struct zoneref *z;
	struct zone *zone;

	for_each_zone_zonelist_nodemask(zone, z, zonelist,
			gfp_zone(sc->gfp_mask), sc->nodemask) {
		if (!populated_zone(zone))
			continue;
		if (!cpuset_zone_allowed_hardwall(zone, GFP_KERNEL))
			continue;
2102 2103
		if (!zone->all_unreclaimable)
			return false;
2104 2105
	}

2106
	return true;
L
Linus Torvalds 已提交
2107
}
2108

L
Linus Torvalds 已提交
2109 2110 2111 2112 2113 2114 2115 2116
/*
 * This is the main entry point to direct page reclaim.
 *
 * If a full scan of the inactive list fails to free enough memory then we
 * are "out of memory" and something needs to be killed.
 *
 * If the caller is !__GFP_FS then the probability of a failure is reasonably
 * high - the zone may be full of dirty or under-writeback pages, which this
2117 2118 2119 2120
 * caller can't do much about.  We kick the writeback threads and take explicit
 * naps in the hope that some of these pages can be written.  But if the
 * allocating task holds filesystem locks which prevent writeout this might not
 * work, and the allocation attempt will fail.
2121 2122 2123
 *
 * returns:	0, if no pages reclaimed
 * 		else, the number of pages reclaimed
L
Linus Torvalds 已提交
2124
 */
2125
static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
2126 2127
					struct scan_control *sc,
					struct shrink_control *shrink)
L
Linus Torvalds 已提交
2128 2129
{
	int priority;
2130
	unsigned long total_scanned = 0;
L
Linus Torvalds 已提交
2131
	struct reclaim_state *reclaim_state = current->reclaim_state;
2132
	struct zoneref *z;
2133
	struct zone *zone;
2134
	unsigned long writeback_threshold;
L
Linus Torvalds 已提交
2135

2136
	get_mems_allowed();
2137 2138
	delayacct_freepages_start();

2139
	if (scanning_global_lru(sc))
2140
		count_vm_event(ALLOCSTALL);
L
Linus Torvalds 已提交
2141 2142

	for (priority = DEF_PRIORITY; priority >= 0; priority--) {
2143
		sc->nr_scanned = 0;
2144
		if (!priority)
2145
			disable_swap_token(sc->mem_cgroup);
2146
		shrink_zones(priority, zonelist, sc);
2147 2148 2149 2150
		/*
		 * Don't shrink slabs when reclaiming memory from
		 * over limit cgroups
		 */
2151
		if (scanning_global_lru(sc)) {
2152
			unsigned long lru_pages = 0;
2153 2154
			for_each_zone_zonelist(zone, z, zonelist,
					gfp_zone(sc->gfp_mask)) {
2155 2156 2157 2158 2159 2160
				if (!cpuset_zone_allowed_hardwall(zone, GFP_KERNEL))
					continue;

				lru_pages += zone_reclaimable_pages(zone);
			}

2161
			shrink_slab(shrink, sc->nr_scanned, lru_pages);
2162
			if (reclaim_state) {
2163
				sc->nr_reclaimed += reclaim_state->reclaimed_slab;
2164 2165
				reclaim_state->reclaimed_slab = 0;
			}
L
Linus Torvalds 已提交
2166
		}
2167
		total_scanned += sc->nr_scanned;
2168
		if (sc->nr_reclaimed >= sc->nr_to_reclaim)
L
Linus Torvalds 已提交
2169 2170 2171 2172 2173 2174 2175 2176 2177
			goto out;

		/*
		 * Try to write back as many pages as we just scanned.  This
		 * tends to cause slow streaming writers to write data to the
		 * disk smoothly, at the dirtying rate, which is nice.   But
		 * that's undesirable in laptop mode, where we *want* lumpy
		 * writeout.  So in laptop mode, write out the whole world.
		 */
2178 2179
		writeback_threshold = sc->nr_to_reclaim + sc->nr_to_reclaim / 2;
		if (total_scanned > writeback_threshold) {
2180
			wakeup_flusher_threads(laptop_mode ? 0 : total_scanned);
2181
			sc->may_writepage = 1;
L
Linus Torvalds 已提交
2182 2183 2184
		}

		/* Take a nap, wait for some writeback to complete */
2185
		if (!sc->hibernation_mode && sc->nr_scanned &&
2186 2187 2188 2189
		    priority < DEF_PRIORITY - 2) {
			struct zone *preferred_zone;

			first_zones_zonelist(zonelist, gfp_zone(sc->gfp_mask),
2190 2191
						&cpuset_current_mems_allowed,
						&preferred_zone);
2192 2193
			wait_iff_congested(preferred_zone, BLK_RW_ASYNC, HZ/10);
		}
L
Linus Torvalds 已提交
2194
	}
2195

L
Linus Torvalds 已提交
2196
out:
2197
	delayacct_freepages_end();
2198
	put_mems_allowed();
2199

2200 2201 2202
	if (sc->nr_reclaimed)
		return sc->nr_reclaimed;

2203 2204 2205 2206 2207 2208 2209 2210
	/*
	 * As hibernation is going on, kswapd is freezed so that it can't mark
	 * the zone into all_unreclaimable. Thus bypassing all_unreclaimable
	 * check.
	 */
	if (oom_killer_disabled)
		return 0;

2211
	/* top priority shrink_zones still had more to do? don't OOM, then */
2212
	if (scanning_global_lru(sc) && !all_unreclaimable(zonelist, sc))
2213 2214 2215
		return 1;

	return 0;
L
Linus Torvalds 已提交
2216 2217
}

2218
unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
2219
				gfp_t gfp_mask, nodemask_t *nodemask)
2220
{
2221
	unsigned long nr_reclaimed;
2222 2223 2224
	struct scan_control sc = {
		.gfp_mask = gfp_mask,
		.may_writepage = !laptop_mode,
2225
		.nr_to_reclaim = SWAP_CLUSTER_MAX,
2226
		.may_unmap = 1,
2227
		.may_swap = 1,
2228 2229
		.order = order,
		.mem_cgroup = NULL,
2230
		.nodemask = nodemask,
2231
	};
2232 2233 2234
	struct shrink_control shrink = {
		.gfp_mask = sc.gfp_mask,
	};
2235

2236 2237 2238 2239
	trace_mm_vmscan_direct_reclaim_begin(order,
				sc.may_writepage,
				gfp_mask);

2240
	nr_reclaimed = do_try_to_free_pages(zonelist, &sc, &shrink);
2241 2242 2243 2244

	trace_mm_vmscan_direct_reclaim_end(nr_reclaimed);

	return nr_reclaimed;
2245 2246
}

2247
#ifdef CONFIG_CGROUP_MEM_RES_CTLR
2248

2249 2250
unsigned long mem_cgroup_shrink_node_zone(struct mem_cgroup *mem,
						gfp_t gfp_mask, bool noswap,
2251 2252
						struct zone *zone,
						unsigned long *nr_scanned)
2253 2254
{
	struct scan_control sc = {
2255
		.nr_scanned = 0,
2256
		.nr_to_reclaim = SWAP_CLUSTER_MAX,
2257 2258 2259 2260 2261 2262
		.may_writepage = !laptop_mode,
		.may_unmap = 1,
		.may_swap = !noswap,
		.order = 0,
		.mem_cgroup = mem,
	};
2263

2264 2265
	sc.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
			(GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK);
2266 2267 2268 2269 2270

	trace_mm_vmscan_memcg_softlimit_reclaim_begin(0,
						      sc.may_writepage,
						      sc.gfp_mask);

2271 2272 2273 2274 2275 2276 2277 2278
	/*
	 * NOTE: Although we can get the priority field, using it
	 * here is not a good idea, since it limits the pages we can scan.
	 * if we don't reclaim here, the shrink_zone from balance_pgdat
	 * will pick up pages from other mem cgroup's as well. We hack
	 * the priority and make it zero.
	 */
	shrink_zone(0, zone, &sc);
2279 2280 2281

	trace_mm_vmscan_memcg_softlimit_reclaim_end(sc.nr_reclaimed);

2282
	*nr_scanned = sc.nr_scanned;
2283 2284 2285
	return sc.nr_reclaimed;
}

2286
unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *mem_cont,
K
KOSAKI Motohiro 已提交
2287
					   gfp_t gfp_mask,
2288
					   bool noswap)
2289
{
2290
	struct zonelist *zonelist;
2291
	unsigned long nr_reclaimed;
2292
	int nid;
2293 2294
	struct scan_control sc = {
		.may_writepage = !laptop_mode,
2295
		.may_unmap = 1,
2296
		.may_swap = !noswap,
2297
		.nr_to_reclaim = SWAP_CLUSTER_MAX,
2298 2299
		.order = 0,
		.mem_cgroup = mem_cont,
2300
		.nodemask = NULL, /* we don't care the placement */
2301 2302 2303 2304 2305
		.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
				(GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK),
	};
	struct shrink_control shrink = {
		.gfp_mask = sc.gfp_mask,
2306 2307
	};

2308 2309 2310 2311 2312 2313 2314 2315
	/*
	 * Unlike direct reclaim via alloc_pages(), memcg's reclaim doesn't
	 * take care of from where we get pages. So the node where we start the
	 * scan does not need to be the current node.
	 */
	nid = mem_cgroup_select_victim_node(mem_cont);

	zonelist = NODE_DATA(nid)->node_zonelists;
2316 2317 2318 2319 2320

	trace_mm_vmscan_memcg_reclaim_begin(0,
					    sc.may_writepage,
					    sc.gfp_mask);

2321
	nr_reclaimed = do_try_to_free_pages(zonelist, &sc, &shrink);
2322 2323 2324 2325

	trace_mm_vmscan_memcg_reclaim_end(nr_reclaimed);

	return nr_reclaimed;
2326 2327 2328
}
#endif

2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339
/*
 * pgdat_balanced is used when checking if a node is balanced for high-order
 * allocations. Only zones that meet watermarks and are in a zone allowed
 * by the callers classzone_idx are added to balanced_pages. The total of
 * balanced pages must be at least 25% of the zones allowed by classzone_idx
 * for the node to be considered balanced. Forcing all zones to be balanced
 * for high orders can cause excessive reclaim when there are imbalanced zones.
 * The choice of 25% is due to
 *   o a 16M DMA zone that is balanced will not balance a zone on any
 *     reasonable sized machine
 *   o On all other machines, the top zone must be at least a reasonable
L
Lucas De Marchi 已提交
2340
 *     percentage of the middle zones. For example, on 32-bit x86, highmem
2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353
 *     would need to be at least 256M for it to be balance a whole node.
 *     Similarly, on x86-64 the Normal zone would need to be at least 1G
 *     to balance a node on its own. These seemed like reasonable ratios.
 */
static bool pgdat_balanced(pg_data_t *pgdat, unsigned long balanced_pages,
						int classzone_idx)
{
	unsigned long present_pages = 0;
	int i;

	for (i = 0; i <= classzone_idx; i++)
		present_pages += pgdat->node_zones[i].present_pages;

S
Shaohua Li 已提交
2354 2355
	/* A special case here: if zone has no page, we think it's balanced */
	return balanced_pages >= (present_pages >> 2);
2356 2357
}

2358
/* is kswapd sleeping prematurely? */
2359 2360
static bool sleeping_prematurely(pg_data_t *pgdat, int order, long remaining,
					int classzone_idx)
2361
{
2362
	int i;
2363 2364
	unsigned long balanced = 0;
	bool all_zones_ok = true;
2365 2366 2367

	/* If a direct reclaimer woke kswapd within HZ/10, it's premature */
	if (remaining)
2368
		return true;
2369

2370
	/* Check the watermark levels */
2371
	for (i = 0; i <= classzone_idx; i++) {
2372 2373 2374 2375 2376
		struct zone *zone = pgdat->node_zones + i;

		if (!populated_zone(zone))
			continue;

2377 2378 2379 2380 2381 2382 2383 2384
		/*
		 * balance_pgdat() skips over all_unreclaimable after
		 * DEF_PRIORITY. Effectively, it considers them balanced so
		 * they must be considered balanced here as well if kswapd
		 * is to sleep
		 */
		if (zone->all_unreclaimable) {
			balanced += zone->present_pages;
2385
			continue;
2386
		}
2387

2388
		if (!zone_watermark_ok_safe(zone, order, high_wmark_pages(zone),
2389
							i, 0))
2390 2391 2392
			all_zones_ok = false;
		else
			balanced += zone->present_pages;
2393
	}
2394

2395 2396 2397 2398 2399 2400
	/*
	 * For high-order requests, the balanced zones must contain at least
	 * 25% of the nodes pages for kswapd to sleep. For order-0, all zones
	 * must be balanced
	 */
	if (order)
2401
		return !pgdat_balanced(pgdat, balanced, classzone_idx);
2402 2403
	else
		return !all_zones_ok;
2404 2405
}

L
Linus Torvalds 已提交
2406 2407
/*
 * For kswapd, balance_pgdat() will work across all this node's zones until
2408
 * they are all at high_wmark_pages(zone).
L
Linus Torvalds 已提交
2409
 *
2410
 * Returns the final order kswapd was reclaiming at
L
Linus Torvalds 已提交
2411 2412 2413 2414 2415 2416 2417 2418 2419 2420
 *
 * There is special handling here for zones which are full of pinned pages.
 * This can happen if the pages are all mlocked, or if they are all used by
 * device drivers (say, ZONE_DMA).  Or if they are all in use by hugetlb.
 * What we do is to detect the case where all pages in the zone have been
 * scanned twice and there has been zero successful reclaim.  Mark the zone as
 * dead and from now on, only perform a short scan.  Basically we're polling
 * the zone for when the problem goes away.
 *
 * kswapd scans the zones in the highmem->normal->dma direction.  It skips
2421 2422 2423 2424 2425
 * zones which have free_pages > high_wmark_pages(zone), but once a zone is
 * found to have free_pages <= high_wmark_pages(zone), we scan that zone and the
 * lower zones regardless of the number of free pages in the lower zones. This
 * interoperates with the page allocator fallback scheme to ensure that aging
 * of pages is balanced across the zones.
L
Linus Torvalds 已提交
2426
 */
2427
static unsigned long balance_pgdat(pg_data_t *pgdat, int order,
2428
							int *classzone_idx)
L
Linus Torvalds 已提交
2429 2430
{
	int all_zones_ok;
2431
	unsigned long balanced;
L
Linus Torvalds 已提交
2432 2433
	int priority;
	int i;
2434
	int end_zone = 0;	/* Inclusive.  0 = ZONE_DMA */
2435
	unsigned long total_scanned;
L
Linus Torvalds 已提交
2436
	struct reclaim_state *reclaim_state = current->reclaim_state;
2437 2438
	unsigned long nr_soft_reclaimed;
	unsigned long nr_soft_scanned;
2439 2440
	struct scan_control sc = {
		.gfp_mask = GFP_KERNEL,
2441
		.may_unmap = 1,
2442
		.may_swap = 1,
2443 2444 2445 2446 2447
		/*
		 * kswapd doesn't want to be bailed out while reclaim. because
		 * we want to put equal scanning pressure on each zone.
		 */
		.nr_to_reclaim = ULONG_MAX,
A
Andy Whitcroft 已提交
2448
		.order = order,
2449
		.mem_cgroup = NULL,
2450
	};
2451 2452 2453
	struct shrink_control shrink = {
		.gfp_mask = sc.gfp_mask,
	};
L
Linus Torvalds 已提交
2454 2455
loop_again:
	total_scanned = 0;
2456
	sc.nr_reclaimed = 0;
C
Christoph Lameter 已提交
2457
	sc.may_writepage = !laptop_mode;
2458
	count_vm_event(PAGEOUTRUN);
L
Linus Torvalds 已提交
2459 2460 2461

	for (priority = DEF_PRIORITY; priority >= 0; priority--) {
		unsigned long lru_pages = 0;
2462
		int has_under_min_watermark_zone = 0;
L
Linus Torvalds 已提交
2463

2464 2465
		/* The swap token gets in the way of swapout... */
		if (!priority)
2466
			disable_swap_token(NULL);
2467

L
Linus Torvalds 已提交
2468
		all_zones_ok = 1;
2469
		balanced = 0;
L
Linus Torvalds 已提交
2470

2471 2472 2473 2474 2475 2476
		/*
		 * Scan in the highmem->dma direction for the highest
		 * zone which needs scanning
		 */
		for (i = pgdat->nr_zones - 1; i >= 0; i--) {
			struct zone *zone = pgdat->node_zones + i;
L
Linus Torvalds 已提交
2477

2478 2479
			if (!populated_zone(zone))
				continue;
L
Linus Torvalds 已提交
2480

2481
			if (zone->all_unreclaimable && priority != DEF_PRIORITY)
2482
				continue;
L
Linus Torvalds 已提交
2483

2484 2485 2486 2487
			/*
			 * Do some background aging of the anon list, to give
			 * pages a chance to be referenced before reclaiming.
			 */
2488
			if (inactive_anon_is_low(zone, &sc))
2489 2490 2491
				shrink_active_list(SWAP_CLUSTER_MAX, zone,
							&sc, priority, 0);

2492
			if (!zone_watermark_ok_safe(zone, order,
2493
					high_wmark_pages(zone), 0, 0)) {
2494
				end_zone = i;
A
Andrew Morton 已提交
2495
				break;
L
Linus Torvalds 已提交
2496 2497
			}
		}
A
Andrew Morton 已提交
2498 2499 2500
		if (i < 0)
			goto out;

L
Linus Torvalds 已提交
2501 2502 2503
		for (i = 0; i <= end_zone; i++) {
			struct zone *zone = pgdat->node_zones + i;

2504
			lru_pages += zone_reclaimable_pages(zone);
L
Linus Torvalds 已提交
2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517
		}

		/*
		 * Now scan the zone in the dma->highmem direction, stopping
		 * at the last zone which needs scanning.
		 *
		 * We do this because the page allocator works in the opposite
		 * direction.  This prevents the page allocator from allocating
		 * pages behind kswapd's direction of progress, which would
		 * cause too much scanning of the lower zones.
		 */
		for (i = 0; i <= end_zone; i++) {
			struct zone *zone = pgdat->node_zones + i;
2518
			int nr_slab;
2519
			unsigned long balance_gap;
L
Linus Torvalds 已提交
2520

2521
			if (!populated_zone(zone))
L
Linus Torvalds 已提交
2522 2523
				continue;

2524
			if (zone->all_unreclaimable && priority != DEF_PRIORITY)
L
Linus Torvalds 已提交
2525 2526 2527
				continue;

			sc.nr_scanned = 0;
2528

2529
			nr_soft_scanned = 0;
2530 2531 2532
			/*
			 * Call soft limit reclaim before calling shrink_zone.
			 */
2533 2534 2535 2536 2537
			nr_soft_reclaimed = mem_cgroup_soft_limit_reclaim(zone,
							order, sc.gfp_mask,
							&nr_soft_scanned);
			sc.nr_reclaimed += nr_soft_reclaimed;
			total_scanned += nr_soft_scanned;
2538

2539
			/*
2540 2541 2542 2543 2544 2545
			 * We put equal pressure on every zone, unless
			 * one zone has way too many pages free
			 * already. The "too many pages" is defined
			 * as the high wmark plus a "gap" where the
			 * gap is either the low watermark or 1%
			 * of the zone, whichever is smaller.
2546
			 */
2547 2548 2549 2550
			balance_gap = min(low_wmark_pages(zone),
				(zone->present_pages +
					KSWAPD_ZONE_BALANCE_GAP_RATIO-1) /
				KSWAPD_ZONE_BALANCE_GAP_RATIO);
2551
			if (!zone_watermark_ok_safe(zone, order,
2552
					high_wmark_pages(zone) + balance_gap,
2553
					end_zone, 0)) {
2554
				shrink_zone(priority, zone, &sc);
2555

2556 2557 2558 2559 2560 2561 2562 2563 2564
				reclaim_state->reclaimed_slab = 0;
				nr_slab = shrink_slab(&shrink, sc.nr_scanned, lru_pages);
				sc.nr_reclaimed += reclaim_state->reclaimed_slab;
				total_scanned += sc.nr_scanned;

				if (nr_slab == 0 && !zone_reclaimable(zone))
					zone->all_unreclaimable = 1;
			}

L
Linus Torvalds 已提交
2565 2566 2567 2568 2569 2570
			/*
			 * If we've done a decent amount of scanning and
			 * the reclaim ratio is low, start doing writepage
			 * even in laptop mode
			 */
			if (total_scanned > SWAP_CLUSTER_MAX * 2 &&
2571
			    total_scanned > sc.nr_reclaimed + sc.nr_reclaimed / 2)
L
Linus Torvalds 已提交
2572
				sc.may_writepage = 1;
2573

2574 2575 2576
			if (zone->all_unreclaimable) {
				if (end_zone && end_zone == i)
					end_zone--;
2577
				continue;
2578
			}
2579

2580
			if (!zone_watermark_ok_safe(zone, order,
2581 2582 2583 2584 2585 2586 2587
					high_wmark_pages(zone), end_zone, 0)) {
				all_zones_ok = 0;
				/*
				 * We are still under min water mark.  This
				 * means that we have a GFP_ATOMIC allocation
				 * failure risk. Hurry up!
				 */
2588
				if (!zone_watermark_ok_safe(zone, order,
2589 2590
					    min_wmark_pages(zone), end_zone, 0))
					has_under_min_watermark_zone = 1;
2591 2592 2593 2594 2595 2596 2597 2598 2599
			} else {
				/*
				 * If a zone reaches its high watermark,
				 * consider it to be no longer congested. It's
				 * possible there are dirty pages backed by
				 * congested BDIs but as pressure is relieved,
				 * spectulatively avoid congestion waits
				 */
				zone_clear_flag(zone, ZONE_CONGESTED);
2600
				if (i <= *classzone_idx)
2601
					balanced += zone->present_pages;
2602
			}
2603

L
Linus Torvalds 已提交
2604
		}
2605
		if (all_zones_ok || (order && pgdat_balanced(pgdat, balanced, *classzone_idx)))
L
Linus Torvalds 已提交
2606 2607 2608 2609 2610
			break;		/* kswapd: all done */
		/*
		 * OK, kswapd is getting into trouble.  Take a nap, then take
		 * another pass across the zones.
		 */
2611 2612 2613 2614 2615 2616
		if (total_scanned && (priority < DEF_PRIORITY - 2)) {
			if (has_under_min_watermark_zone)
				count_vm_event(KSWAPD_SKIP_CONGESTION_WAIT);
			else
				congestion_wait(BLK_RW_ASYNC, HZ/10);
		}
L
Linus Torvalds 已提交
2617 2618 2619 2620 2621 2622 2623

		/*
		 * We do this so kswapd doesn't build up large priorities for
		 * example when it is freeing in parallel with allocators. It
		 * matches the direct reclaim path behaviour in terms of impact
		 * on zone->*_priority.
		 */
2624
		if (sc.nr_reclaimed >= SWAP_CLUSTER_MAX)
L
Linus Torvalds 已提交
2625 2626 2627
			break;
	}
out:
2628 2629 2630

	/*
	 * order-0: All zones must meet high watermark for a balanced node
2631 2632
	 * high-order: Balanced zones must make up at least 25% of the node
	 *             for the node to be balanced
2633
	 */
2634
	if (!(all_zones_ok || (order && pgdat_balanced(pgdat, balanced, *classzone_idx)))) {
L
Linus Torvalds 已提交
2635
		cond_resched();
2636 2637 2638

		try_to_freeze();

2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655
		/*
		 * Fragmentation may mean that the system cannot be
		 * rebalanced for high-order allocations in all zones.
		 * At this point, if nr_reclaimed < SWAP_CLUSTER_MAX,
		 * it means the zones have been fully scanned and are still
		 * not balanced. For high-order allocations, there is
		 * little point trying all over again as kswapd may
		 * infinite loop.
		 *
		 * Instead, recheck all watermarks at order-0 as they
		 * are the most important. If watermarks are ok, kswapd will go
		 * back to sleep. High-order users can still perform direct
		 * reclaim if they wish.
		 */
		if (sc.nr_reclaimed < SWAP_CLUSTER_MAX)
			order = sc.order = 0;

L
Linus Torvalds 已提交
2656 2657 2658
		goto loop_again;
	}

2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688
	/*
	 * If kswapd was reclaiming at a higher order, it has the option of
	 * sleeping without all zones being balanced. Before it does, it must
	 * ensure that the watermarks for order-0 on *all* zones are met and
	 * that the congestion flags are cleared. The congestion flag must
	 * be cleared as kswapd is the only mechanism that clears the flag
	 * and it is potentially going to sleep here.
	 */
	if (order) {
		for (i = 0; i <= end_zone; i++) {
			struct zone *zone = pgdat->node_zones + i;

			if (!populated_zone(zone))
				continue;

			if (zone->all_unreclaimable && priority != DEF_PRIORITY)
				continue;

			/* Confirm the zone is balanced for order-0 */
			if (!zone_watermark_ok(zone, 0,
					high_wmark_pages(zone), 0, 0)) {
				order = sc.order = 0;
				goto loop_again;
			}

			/* If balanced, clear the congested flag */
			zone_clear_flag(zone, ZONE_CONGESTED);
		}
	}

2689 2690 2691 2692 2693 2694
	/*
	 * Return the order we were reclaiming at so sleeping_prematurely()
	 * makes a decision on the order we were last reclaiming at. However,
	 * if another caller entered the allocator slow path while kswapd
	 * was awake, order will remain at the higher level
	 */
2695
	*classzone_idx = end_zone;
2696
	return order;
L
Linus Torvalds 已提交
2697 2698
}

2699
static void kswapd_try_to_sleep(pg_data_t *pgdat, int order, int classzone_idx)
2700 2701 2702 2703 2704 2705 2706 2707 2708 2709
{
	long remaining = 0;
	DEFINE_WAIT(wait);

	if (freezing(current) || kthread_should_stop())
		return;

	prepare_to_wait(&pgdat->kswapd_wait, &wait, TASK_INTERRUPTIBLE);

	/* Try to sleep for a short interval */
2710
	if (!sleeping_prematurely(pgdat, order, remaining, classzone_idx)) {
2711 2712 2713 2714 2715 2716 2717 2718 2719
		remaining = schedule_timeout(HZ/10);
		finish_wait(&pgdat->kswapd_wait, &wait);
		prepare_to_wait(&pgdat->kswapd_wait, &wait, TASK_INTERRUPTIBLE);
	}

	/*
	 * After a short sleep, check if it was a premature sleep. If not, then
	 * go fully to sleep until explicitly woken up.
	 */
2720
	if (!sleeping_prematurely(pgdat, order, remaining, classzone_idx)) {
2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742
		trace_mm_vmscan_kswapd_sleep(pgdat->node_id);

		/*
		 * vmstat counters are not perfectly accurate and the estimated
		 * value for counters such as NR_FREE_PAGES can deviate from the
		 * true value by nr_online_cpus * threshold. To avoid the zone
		 * watermarks being breached while under pressure, we reduce the
		 * per-cpu vmstat threshold while kswapd is awake and restore
		 * them before going back to sleep.
		 */
		set_pgdat_percpu_threshold(pgdat, calculate_normal_threshold);
		schedule();
		set_pgdat_percpu_threshold(pgdat, calculate_pressure_threshold);
	} else {
		if (remaining)
			count_vm_event(KSWAPD_LOW_WMARK_HIT_QUICKLY);
		else
			count_vm_event(KSWAPD_HIGH_WMARK_HIT_QUICKLY);
	}
	finish_wait(&pgdat->kswapd_wait, &wait);
}

L
Linus Torvalds 已提交
2743 2744
/*
 * The background pageout daemon, started as a kernel thread
2745
 * from the init process.
L
Linus Torvalds 已提交
2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757
 *
 * This basically trickles out pages so that we have _some_
 * free memory available even if there is no other activity
 * that frees anything up. This is needed for things like routing
 * etc, where we otherwise might have all activity going on in
 * asynchronous contexts that cannot page things out.
 *
 * If there are applications that are active memory-allocators
 * (most normal use), this basically shouldn't matter.
 */
static int kswapd(void *p)
{
2758 2759
	unsigned long order, new_order;
	int classzone_idx, new_classzone_idx;
L
Linus Torvalds 已提交
2760 2761
	pg_data_t *pgdat = (pg_data_t*)p;
	struct task_struct *tsk = current;
2762

L
Linus Torvalds 已提交
2763 2764 2765
	struct reclaim_state reclaim_state = {
		.reclaimed_slab = 0,
	};
2766
	const struct cpumask *cpumask = cpumask_of_node(pgdat->node_id);
L
Linus Torvalds 已提交
2767

2768 2769
	lockdep_set_current_reclaim_state(GFP_KERNEL);

R
Rusty Russell 已提交
2770
	if (!cpumask_empty(cpumask))
2771
		set_cpus_allowed_ptr(tsk, cpumask);
L
Linus Torvalds 已提交
2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785
	current->reclaim_state = &reclaim_state;

	/*
	 * Tell the memory management that we're a "memory allocator",
	 * and that if we need more memory we should get access to it
	 * regardless (see "__alloc_pages()"). "kswapd" should
	 * never get caught in the normal page freeing logic.
	 *
	 * (Kswapd normally doesn't need memory anyway, but sometimes
	 * you need a small amount of memory in order to be able to
	 * page out something else, and this flag essentially protects
	 * us from recursively trying to free more memory as we're
	 * trying to free the first piece of memory in the first place).
	 */
2786
	tsk->flags |= PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD;
2787
	set_freezable();
L
Linus Torvalds 已提交
2788

2789 2790
	order = new_order = 0;
	classzone_idx = new_classzone_idx = pgdat->nr_zones - 1;
L
Linus Torvalds 已提交
2791
	for ( ; ; ) {
2792
		int ret;
2793

2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805
		/*
		 * If the last balance_pgdat was unsuccessful it's unlikely a
		 * new request of a similar or harder type will succeed soon
		 * so consider going to sleep on the basis we reclaimed at
		 */
		if (classzone_idx >= new_classzone_idx && order == new_order) {
			new_order = pgdat->kswapd_max_order;
			new_classzone_idx = pgdat->classzone_idx;
			pgdat->kswapd_max_order =  0;
			pgdat->classzone_idx = pgdat->nr_zones - 1;
		}

2806
		if (order < new_order || classzone_idx > new_classzone_idx) {
L
Linus Torvalds 已提交
2807 2808
			/*
			 * Don't sleep if someone wants a larger 'order'
2809
			 * allocation or has tigher zone constraints
L
Linus Torvalds 已提交
2810 2811
			 */
			order = new_order;
2812
			classzone_idx = new_classzone_idx;
L
Linus Torvalds 已提交
2813
		} else {
2814
			kswapd_try_to_sleep(pgdat, order, classzone_idx);
L
Linus Torvalds 已提交
2815
			order = pgdat->kswapd_max_order;
2816
			classzone_idx = pgdat->classzone_idx;
2817
			pgdat->kswapd_max_order = 0;
2818
			pgdat->classzone_idx = pgdat->nr_zones - 1;
L
Linus Torvalds 已提交
2819 2820
		}

2821 2822 2823 2824 2825 2826 2827 2828
		ret = try_to_freeze();
		if (kthread_should_stop())
			break;

		/*
		 * We can speed up thawing tasks if we don't call balance_pgdat
		 * after returning from the refrigerator
		 */
2829 2830
		if (!ret) {
			trace_mm_vmscan_kswapd_wake(pgdat->node_id, order);
2831
			order = balance_pgdat(pgdat, order, &classzone_idx);
2832
		}
L
Linus Torvalds 已提交
2833 2834 2835 2836 2837 2838 2839
	}
	return 0;
}

/*
 * A zone is low on free memory, so wake its kswapd task to service it.
 */
2840
void wakeup_kswapd(struct zone *zone, int order, enum zone_type classzone_idx)
L
Linus Torvalds 已提交
2841 2842 2843
{
	pg_data_t *pgdat;

2844
	if (!populated_zone(zone))
L
Linus Torvalds 已提交
2845 2846
		return;

2847
	if (!cpuset_zone_allowed_hardwall(zone, GFP_KERNEL))
L
Linus Torvalds 已提交
2848
		return;
2849
	pgdat = zone->zone_pgdat;
2850
	if (pgdat->kswapd_max_order < order) {
L
Linus Torvalds 已提交
2851
		pgdat->kswapd_max_order = order;
2852 2853
		pgdat->classzone_idx = min(pgdat->classzone_idx, classzone_idx);
	}
2854
	if (!waitqueue_active(&pgdat->kswapd_wait))
L
Linus Torvalds 已提交
2855
		return;
2856 2857 2858 2859
	if (zone_watermark_ok_safe(zone, order, low_wmark_pages(zone), 0, 0))
		return;

	trace_mm_vmscan_wakeup_kswapd(pgdat->node_id, zone_idx(zone), order);
2860
	wake_up_interruptible(&pgdat->kswapd_wait);
L
Linus Torvalds 已提交
2861 2862
}

2863 2864 2865 2866 2867 2868 2869 2870
/*
 * The reclaimable count would be mostly accurate.
 * The less reclaimable pages may be
 * - mlocked pages, which will be moved to unevictable list when encountered
 * - mapped pages, which may require several travels to be reclaimed
 * - dirty pages, which is not "instantly" reclaimable
 */
unsigned long global_reclaimable_pages(void)
2871
{
2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895
	int nr;

	nr = global_page_state(NR_ACTIVE_FILE) +
	     global_page_state(NR_INACTIVE_FILE);

	if (nr_swap_pages > 0)
		nr += global_page_state(NR_ACTIVE_ANON) +
		      global_page_state(NR_INACTIVE_ANON);

	return nr;
}

unsigned long zone_reclaimable_pages(struct zone *zone)
{
	int nr;

	nr = zone_page_state(zone, NR_ACTIVE_FILE) +
	     zone_page_state(zone, NR_INACTIVE_FILE);

	if (nr_swap_pages > 0)
		nr += zone_page_state(zone, NR_ACTIVE_ANON) +
		      zone_page_state(zone, NR_INACTIVE_ANON);

	return nr;
2896 2897
}

2898
#ifdef CONFIG_HIBERNATION
L
Linus Torvalds 已提交
2899
/*
2900
 * Try to free `nr_to_reclaim' of memory, system-wide, and return the number of
2901 2902 2903 2904 2905
 * freed pages.
 *
 * Rather than trying to age LRUs the aim is to preserve the overall
 * LRU order by reclaiming preferentially
 * inactive > active > active referenced > active mapped
L
Linus Torvalds 已提交
2906
 */
2907
unsigned long shrink_all_memory(unsigned long nr_to_reclaim)
L
Linus Torvalds 已提交
2908
{
2909 2910
	struct reclaim_state reclaim_state;
	struct scan_control sc = {
2911 2912 2913
		.gfp_mask = GFP_HIGHUSER_MOVABLE,
		.may_swap = 1,
		.may_unmap = 1,
2914
		.may_writepage = 1,
2915 2916 2917
		.nr_to_reclaim = nr_to_reclaim,
		.hibernation_mode = 1,
		.order = 0,
L
Linus Torvalds 已提交
2918
	};
2919 2920 2921 2922
	struct shrink_control shrink = {
		.gfp_mask = sc.gfp_mask,
	};
	struct zonelist *zonelist = node_zonelist(numa_node_id(), sc.gfp_mask);
2923 2924
	struct task_struct *p = current;
	unsigned long nr_reclaimed;
L
Linus Torvalds 已提交
2925

2926 2927 2928 2929
	p->flags |= PF_MEMALLOC;
	lockdep_set_current_reclaim_state(sc.gfp_mask);
	reclaim_state.reclaimed_slab = 0;
	p->reclaim_state = &reclaim_state;
2930

2931
	nr_reclaimed = do_try_to_free_pages(zonelist, &sc, &shrink);
2932

2933 2934 2935
	p->reclaim_state = NULL;
	lockdep_clear_current_reclaim_state();
	p->flags &= ~PF_MEMALLOC;
2936

2937
	return nr_reclaimed;
L
Linus Torvalds 已提交
2938
}
2939
#endif /* CONFIG_HIBERNATION */
L
Linus Torvalds 已提交
2940 2941 2942 2943 2944

/* It's optimal to keep kswapds on the same CPUs as their memory, but
   not required for correctness.  So if the last cpu in a node goes
   away, we get changed to run anywhere: as the first one comes back,
   restore their cpu bindings. */
2945
static int __devinit cpu_callback(struct notifier_block *nfb,
2946
				  unsigned long action, void *hcpu)
L
Linus Torvalds 已提交
2947
{
2948
	int nid;
L
Linus Torvalds 已提交
2949

2950
	if (action == CPU_ONLINE || action == CPU_ONLINE_FROZEN) {
2951
		for_each_node_state(nid, N_HIGH_MEMORY) {
2952
			pg_data_t *pgdat = NODE_DATA(nid);
2953 2954 2955
			const struct cpumask *mask;

			mask = cpumask_of_node(pgdat->node_id);
2956

2957
			if (cpumask_any_and(cpu_online_mask, mask) < nr_cpu_ids)
L
Linus Torvalds 已提交
2958
				/* One of our CPUs online: restore mask */
2959
				set_cpus_allowed_ptr(pgdat->kswapd, mask);
L
Linus Torvalds 已提交
2960 2961 2962 2963 2964
		}
	}
	return NOTIFY_OK;
}

2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986
/*
 * This kswapd start function will be called by init and node-hot-add.
 * On node-hot-add, kswapd will moved to proper cpus if cpus are hot-added.
 */
int kswapd_run(int nid)
{
	pg_data_t *pgdat = NODE_DATA(nid);
	int ret = 0;

	if (pgdat->kswapd)
		return 0;

	pgdat->kswapd = kthread_run(kswapd, pgdat, "kswapd%d", nid);
	if (IS_ERR(pgdat->kswapd)) {
		/* failure at boot is fatal */
		BUG_ON(system_state == SYSTEM_BOOTING);
		printk("Failed to start kswapd on node %d\n",nid);
		ret = -1;
	}
	return ret;
}

2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997
/*
 * Called by memory hotplug when all memory in a node is offlined.
 */
void kswapd_stop(int nid)
{
	struct task_struct *kswapd = NODE_DATA(nid)->kswapd;

	if (kswapd)
		kthread_stop(kswapd);
}

L
Linus Torvalds 已提交
2998 2999
static int __init kswapd_init(void)
{
3000
	int nid;
3001

L
Linus Torvalds 已提交
3002
	swap_setup();
3003
	for_each_node_state(nid, N_HIGH_MEMORY)
3004
 		kswapd_run(nid);
L
Linus Torvalds 已提交
3005 3006 3007 3008 3009
	hotcpu_notifier(cpu_callback, 0);
	return 0;
}

module_init(kswapd_init)
3010 3011 3012 3013 3014 3015 3016 3017 3018 3019

#ifdef CONFIG_NUMA
/*
 * Zone reclaim mode
 *
 * If non-zero call zone_reclaim when the number of free pages falls below
 * the watermarks.
 */
int zone_reclaim_mode __read_mostly;

3020
#define RECLAIM_OFF 0
3021
#define RECLAIM_ZONE (1<<0)	/* Run shrink_inactive_list on the zone */
3022 3023 3024
#define RECLAIM_WRITE (1<<1)	/* Writeout pages during reclaim */
#define RECLAIM_SWAP (1<<2)	/* Swap pages out during reclaim */

3025 3026 3027 3028 3029 3030 3031
/*
 * Priority for ZONE_RECLAIM. This determines the fraction of pages
 * of a node considered for each zone_reclaim. 4 scans 1/16th of
 * a zone.
 */
#define ZONE_RECLAIM_PRIORITY 4

3032 3033 3034 3035 3036 3037
/*
 * Percentage of pages in a zone that must be unmapped for zone_reclaim to
 * occur.
 */
int sysctl_min_unmapped_ratio = 1;

3038 3039 3040 3041 3042 3043
/*
 * If the number of slab pages in a zone grows beyond this percentage then
 * slab reclaim needs to occur.
 */
int sysctl_min_slab_ratio = 5;

3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085
static inline unsigned long zone_unmapped_file_pages(struct zone *zone)
{
	unsigned long file_mapped = zone_page_state(zone, NR_FILE_MAPPED);
	unsigned long file_lru = zone_page_state(zone, NR_INACTIVE_FILE) +
		zone_page_state(zone, NR_ACTIVE_FILE);

	/*
	 * It's possible for there to be more file mapped pages than
	 * accounted for by the pages on the file LRU lists because
	 * tmpfs pages accounted for as ANON can also be FILE_MAPPED
	 */
	return (file_lru > file_mapped) ? (file_lru - file_mapped) : 0;
}

/* Work out how many page cache pages we can reclaim in this reclaim_mode */
static long zone_pagecache_reclaimable(struct zone *zone)
{
	long nr_pagecache_reclaimable;
	long delta = 0;

	/*
	 * If RECLAIM_SWAP is set, then all file pages are considered
	 * potentially reclaimable. Otherwise, we have to worry about
	 * pages like swapcache and zone_unmapped_file_pages() provides
	 * a better estimate
	 */
	if (zone_reclaim_mode & RECLAIM_SWAP)
		nr_pagecache_reclaimable = zone_page_state(zone, NR_FILE_PAGES);
	else
		nr_pagecache_reclaimable = zone_unmapped_file_pages(zone);

	/* If we can't clean pages, remove dirty pages from consideration */
	if (!(zone_reclaim_mode & RECLAIM_WRITE))
		delta += zone_page_state(zone, NR_FILE_DIRTY);

	/* Watch for any possible underflows due to delta */
	if (unlikely(delta > nr_pagecache_reclaimable))
		delta = nr_pagecache_reclaimable;

	return nr_pagecache_reclaimable - delta;
}

3086 3087 3088
/*
 * Try to free up some pages from this zone through reclaim.
 */
3089
static int __zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
3090
{
3091
	/* Minimum pages needed in order to stay on node */
3092
	const unsigned long nr_pages = 1 << order;
3093 3094
	struct task_struct *p = current;
	struct reclaim_state reclaim_state;
3095
	int priority;
3096 3097
	struct scan_control sc = {
		.may_writepage = !!(zone_reclaim_mode & RECLAIM_WRITE),
3098
		.may_unmap = !!(zone_reclaim_mode & RECLAIM_SWAP),
3099
		.may_swap = 1,
3100 3101
		.nr_to_reclaim = max_t(unsigned long, nr_pages,
				       SWAP_CLUSTER_MAX),
3102
		.gfp_mask = gfp_mask,
3103
		.order = order,
3104
	};
3105 3106 3107
	struct shrink_control shrink = {
		.gfp_mask = sc.gfp_mask,
	};
3108
	unsigned long nr_slab_pages0, nr_slab_pages1;
3109 3110

	cond_resched();
3111 3112 3113 3114 3115 3116
	/*
	 * We need to be able to allocate from the reserves for RECLAIM_SWAP
	 * and we also need to be able to write out pages for RECLAIM_WRITE
	 * and RECLAIM_SWAP.
	 */
	p->flags |= PF_MEMALLOC | PF_SWAPWRITE;
3117
	lockdep_set_current_reclaim_state(gfp_mask);
3118 3119
	reclaim_state.reclaimed_slab = 0;
	p->reclaim_state = &reclaim_state;
3120

3121
	if (zone_pagecache_reclaimable(zone) > zone->min_unmapped_pages) {
3122 3123 3124 3125 3126 3127
		/*
		 * Free memory by calling shrink zone with increasing
		 * priorities until we have enough memory freed.
		 */
		priority = ZONE_RECLAIM_PRIORITY;
		do {
3128
			shrink_zone(priority, zone, &sc);
3129
			priority--;
3130
		} while (priority >= 0 && sc.nr_reclaimed < nr_pages);
3131
	}
3132

3133 3134
	nr_slab_pages0 = zone_page_state(zone, NR_SLAB_RECLAIMABLE);
	if (nr_slab_pages0 > zone->min_slab_pages) {
3135
		/*
3136
		 * shrink_slab() does not currently allow us to determine how
3137 3138 3139 3140
		 * many pages were freed in this zone. So we take the current
		 * number of slab pages and shake the slab until it is reduced
		 * by the same nr_pages that we used for reclaiming unmapped
		 * pages.
3141
		 *
3142 3143
		 * Note that shrink_slab will free memory on all zones and may
		 * take a long time.
3144
		 */
3145 3146 3147 3148
		for (;;) {
			unsigned long lru_pages = zone_reclaimable_pages(zone);

			/* No reclaimable slab or very low memory pressure */
3149
			if (!shrink_slab(&shrink, sc.nr_scanned, lru_pages))
3150 3151 3152 3153 3154 3155 3156 3157
				break;

			/* Freed enough memory */
			nr_slab_pages1 = zone_page_state(zone,
							NR_SLAB_RECLAIMABLE);
			if (nr_slab_pages1 + nr_pages <= nr_slab_pages0)
				break;
		}
3158 3159 3160 3161 3162

		/*
		 * Update nr_reclaimed by the number of slab pages we
		 * reclaimed from this zone.
		 */
3163 3164 3165
		nr_slab_pages1 = zone_page_state(zone, NR_SLAB_RECLAIMABLE);
		if (nr_slab_pages1 < nr_slab_pages0)
			sc.nr_reclaimed += nr_slab_pages0 - nr_slab_pages1;
3166 3167
	}

3168
	p->reclaim_state = NULL;
3169
	current->flags &= ~(PF_MEMALLOC | PF_SWAPWRITE);
3170
	lockdep_clear_current_reclaim_state();
3171
	return sc.nr_reclaimed >= nr_pages;
3172
}
3173 3174 3175 3176

int zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
{
	int node_id;
3177
	int ret;
3178 3179

	/*
3180 3181
	 * Zone reclaim reclaims unmapped file backed pages and
	 * slab pages if we are over the defined limits.
3182
	 *
3183 3184 3185 3186 3187
	 * A small portion of unmapped file backed pages is needed for
	 * file I/O otherwise pages read by file I/O will be immediately
	 * thrown out if the zone is overallocated. So we do not reclaim
	 * if less than a specified percentage of the zone is used by
	 * unmapped file backed pages.
3188
	 */
3189 3190
	if (zone_pagecache_reclaimable(zone) <= zone->min_unmapped_pages &&
	    zone_page_state(zone, NR_SLAB_RECLAIMABLE) <= zone->min_slab_pages)
3191
		return ZONE_RECLAIM_FULL;
3192

3193
	if (zone->all_unreclaimable)
3194
		return ZONE_RECLAIM_FULL;
3195

3196
	/*
3197
	 * Do not scan if the allocation should not be delayed.
3198
	 */
3199
	if (!(gfp_mask & __GFP_WAIT) || (current->flags & PF_MEMALLOC))
3200
		return ZONE_RECLAIM_NOSCAN;
3201 3202 3203 3204 3205 3206 3207

	/*
	 * Only run zone reclaim on the local zone or on zones that do not
	 * have associated processors. This will favor the local processor
	 * over remote processors and spread off node memory allocations
	 * as wide as possible.
	 */
3208
	node_id = zone_to_nid(zone);
3209
	if (node_state(node_id, N_CPU) && node_id != numa_node_id())
3210
		return ZONE_RECLAIM_NOSCAN;
3211 3212

	if (zone_test_and_set_flag(zone, ZONE_RECLAIM_LOCKED))
3213 3214
		return ZONE_RECLAIM_NOSCAN;

3215 3216 3217
	ret = __zone_reclaim(zone, gfp_mask, order);
	zone_clear_flag(zone, ZONE_RECLAIM_LOCKED);

3218 3219 3220
	if (!ret)
		count_vm_event(PGSCAN_ZONE_RECLAIM_FAILED);

3221
	return ret;
3222
}
3223
#endif
L
Lee Schermerhorn 已提交
3224 3225 3226 3227 3228 3229 3230

/*
 * page_evictable - test whether a page is evictable
 * @page: the page to test
 * @vma: the VMA in which the page is or will be mapped, may be NULL
 *
 * Test whether page is evictable--i.e., should be placed on active/inactive
N
Nick Piggin 已提交
3231 3232
 * lists vs unevictable list.  The vma argument is !NULL when called from the
 * fault path to determine how to instantate a new page.
L
Lee Schermerhorn 已提交
3233 3234
 *
 * Reasons page might not be evictable:
3235
 * (1) page's mapping marked unevictable
N
Nick Piggin 已提交
3236
 * (2) page is part of an mlocked VMA
3237
 *
L
Lee Schermerhorn 已提交
3238 3239 3240 3241
 */
int page_evictable(struct page *page, struct vm_area_struct *vma)
{

3242 3243 3244
	if (mapping_unevictable(page_mapping(page)))
		return 0;

N
Nick Piggin 已提交
3245 3246
	if (PageMlocked(page) || (vma && is_mlocked_vma(vma, page)))
		return 0;
L
Lee Schermerhorn 已提交
3247 3248 3249

	return 1;
}
3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268

/**
 * check_move_unevictable_page - check page for evictability and move to appropriate zone lru list
 * @page: page to check evictability and move to appropriate lru list
 * @zone: zone page is in
 *
 * Checks a page for evictability and moves the page to the appropriate
 * zone lru list.
 *
 * Restrictions: zone->lru_lock must be held, page must be on LRU and must
 * have PageUnevictable set.
 */
static void check_move_unevictable_page(struct page *page, struct zone *zone)
{
	VM_BUG_ON(PageActive(page));

retry:
	ClearPageUnevictable(page);
	if (page_evictable(page, NULL)) {
3269
		enum lru_list l = page_lru_base_type(page);
3270

3271 3272
		__dec_zone_state(zone, NR_UNEVICTABLE);
		list_move(&page->lru, &zone->lru[l].list);
K
KAMEZAWA Hiroyuki 已提交
3273
		mem_cgroup_move_lists(page, LRU_UNEVICTABLE, l);
3274 3275 3276 3277 3278 3279 3280 3281
		__inc_zone_state(zone, NR_INACTIVE_ANON + l);
		__count_vm_event(UNEVICTABLE_PGRESCUED);
	} else {
		/*
		 * rotate unevictable list
		 */
		SetPageUnevictable(page);
		list_move(&page->lru, &zone->lru[LRU_UNEVICTABLE].list);
K
KAMEZAWA Hiroyuki 已提交
3282
		mem_cgroup_rotate_lru_list(page, LRU_UNEVICTABLE);
3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341
		if (page_evictable(page, NULL))
			goto retry;
	}
}

/**
 * scan_mapping_unevictable_pages - scan an address space for evictable pages
 * @mapping: struct address_space to scan for evictable pages
 *
 * Scan all pages in mapping.  Check unevictable pages for
 * evictability and move them to the appropriate zone lru list.
 */
void scan_mapping_unevictable_pages(struct address_space *mapping)
{
	pgoff_t next = 0;
	pgoff_t end   = (i_size_read(mapping->host) + PAGE_CACHE_SIZE - 1) >>
			 PAGE_CACHE_SHIFT;
	struct zone *zone;
	struct pagevec pvec;

	if (mapping->nrpages == 0)
		return;

	pagevec_init(&pvec, 0);
	while (next < end &&
		pagevec_lookup(&pvec, mapping, next, PAGEVEC_SIZE)) {
		int i;
		int pg_scanned = 0;

		zone = NULL;

		for (i = 0; i < pagevec_count(&pvec); i++) {
			struct page *page = pvec.pages[i];
			pgoff_t page_index = page->index;
			struct zone *pagezone = page_zone(page);

			pg_scanned++;
			if (page_index > next)
				next = page_index;
			next++;

			if (pagezone != zone) {
				if (zone)
					spin_unlock_irq(&zone->lru_lock);
				zone = pagezone;
				spin_lock_irq(&zone->lru_lock);
			}

			if (PageLRU(page) && PageUnevictable(page))
				check_move_unevictable_page(page, zone);
		}
		if (zone)
			spin_unlock_irq(&zone->lru_lock);
		pagevec_release(&pvec);

		count_vm_events(UNEVICTABLE_PGSCANNED, pg_scanned);
	}

}
3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353

/**
 * scan_zone_unevictable_pages - check unevictable list for evictable pages
 * @zone - zone of which to scan the unevictable list
 *
 * Scan @zone's unevictable LRU lists to check for pages that have become
 * evictable.  Move those that have to @zone's inactive list where they
 * become candidates for reclaim, unless shrink_inactive_zone() decides
 * to reactivate them.  Pages that are still unevictable are rotated
 * back onto @zone's unevictable list.
 */
#define SCAN_UNEVICTABLE_BATCH_SIZE 16UL /* arbitrary lock hold batch size */
3354
static void scan_zone_unevictable_pages(struct zone *zone)
3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395
{
	struct list_head *l_unevictable = &zone->lru[LRU_UNEVICTABLE].list;
	unsigned long scan;
	unsigned long nr_to_scan = zone_page_state(zone, NR_UNEVICTABLE);

	while (nr_to_scan > 0) {
		unsigned long batch_size = min(nr_to_scan,
						SCAN_UNEVICTABLE_BATCH_SIZE);

		spin_lock_irq(&zone->lru_lock);
		for (scan = 0;  scan < batch_size; scan++) {
			struct page *page = lru_to_page(l_unevictable);

			if (!trylock_page(page))
				continue;

			prefetchw_prev_lru_page(page, l_unevictable, flags);

			if (likely(PageLRU(page) && PageUnevictable(page)))
				check_move_unevictable_page(page, zone);

			unlock_page(page);
		}
		spin_unlock_irq(&zone->lru_lock);

		nr_to_scan -= batch_size;
	}
}


/**
 * scan_all_zones_unevictable_pages - scan all unevictable lists for evictable pages
 *
 * A really big hammer:  scan all zones' unevictable LRU lists to check for
 * pages that have become evictable.  Move those back to the zones'
 * inactive list where they become candidates for reclaim.
 * This occurs when, e.g., we have unswappable pages on the unevictable lists,
 * and we add swap to the system.  As such, it runs in the context of a task
 * that has possibly/probably made some previously unevictable pages
 * evictable.
 */
3396
static void scan_all_zones_unevictable_pages(void)
3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411
{
	struct zone *zone;

	for_each_zone(zone) {
		scan_zone_unevictable_pages(zone);
	}
}

/*
 * scan_unevictable_pages [vm] sysctl handler.  On demand re-scan of
 * all nodes' unevictable lists for evictable pages
 */
unsigned long scan_unevictable_pages;

int scan_unevictable_handler(struct ctl_table *table, int write,
3412
			   void __user *buffer,
3413 3414
			   size_t *length, loff_t *ppos)
{
3415
	proc_doulongvec_minmax(table, write, buffer, length, ppos);
3416 3417 3418 3419 3420 3421 3422 3423

	if (write && *(unsigned long *)table->data)
		scan_all_zones_unevictable_pages();

	scan_unevictable_pages = 0;
	return 0;
}

3424
#ifdef CONFIG_NUMA
3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470
/*
 * per node 'scan_unevictable_pages' attribute.  On demand re-scan of
 * a specified node's per zone unevictable lists for evictable pages.
 */

static ssize_t read_scan_unevictable_node(struct sys_device *dev,
					  struct sysdev_attribute *attr,
					  char *buf)
{
	return sprintf(buf, "0\n");	/* always zero; should fit... */
}

static ssize_t write_scan_unevictable_node(struct sys_device *dev,
					   struct sysdev_attribute *attr,
					const char *buf, size_t count)
{
	struct zone *node_zones = NODE_DATA(dev->id)->node_zones;
	struct zone *zone;
	unsigned long res;
	unsigned long req = strict_strtoul(buf, 10, &res);

	if (!req)
		return 1;	/* zero is no-op */

	for (zone = node_zones; zone - node_zones < MAX_NR_ZONES; ++zone) {
		if (!populated_zone(zone))
			continue;
		scan_zone_unevictable_pages(zone);
	}
	return 1;
}


static SYSDEV_ATTR(scan_unevictable_pages, S_IRUGO | S_IWUSR,
			read_scan_unevictable_node,
			write_scan_unevictable_node);

int scan_unevictable_register_node(struct node *node)
{
	return sysdev_create_file(&node->sysdev, &attr_scan_unevictable_pages);
}

void scan_unevictable_unregister_node(struct node *node)
{
	sysdev_remove_file(&node->sysdev, &attr_scan_unevictable_pages);
}
3471
#endif