vmscan.c 79.8 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
/*
 *  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>
#include <linux/slab.h>
#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 35 36
#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>
#include <linux/notifier.h>
#include <linux/rwsem.h>
37
#include <linux/delay.h>
38
#include <linux/kthread.h>
39
#include <linux/freezer.h>
40
#include <linux/memcontrol.h>
41
#include <linux/delayacct.h>
42
#include <linux/sysctl.h>
L
Linus Torvalds 已提交
43 44 45 46 47 48

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

#include <linux/swapops.h>

49 50
#include "internal.h"

L
Linus Torvalds 已提交
51 52 53 54
struct scan_control {
	/* Incremented by the number of inactive pages that were scanned */
	unsigned long nr_scanned;

55 56 57
	/* Number of pages freed so far during a call to shrink_zones() */
	unsigned long nr_reclaimed;

58 59 60
	/* How many pages shrink_list() should reclaim */
	unsigned long nr_to_reclaim;

61 62
	unsigned long hibernation_mode;

L
Linus Torvalds 已提交
63
	/* This context's GFP mask */
A
Al Viro 已提交
64
	gfp_t gfp_mask;
L
Linus Torvalds 已提交
65 66 67

	int may_writepage;

68 69
	/* Can mapped pages be reclaimed? */
	int may_unmap;
70

71 72 73
	/* Can pages be swapped as part of reclaim? */
	int may_swap;

74
	int swappiness;
75 76

	int all_unreclaimable;
A
Andy Whitcroft 已提交
77 78

	int order;
79 80 81 82

	/* Which cgroup do we reclaim from */
	struct mem_cgroup *mem_cgroup;

83 84 85 86 87 88
	/*
	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
	 * are scanned.
	 */
	nodemask_t	*nodemask;

89 90 91 92
	/* Pluggable isolate pages callback */
	unsigned long (*isolate_pages)(unsigned long nr, struct list_head *dst,
			unsigned long *scanned, int order, int mode,
			struct zone *z, struct mem_cgroup *mem_cont,
93
			int active, int file);
L
Linus Torvalds 已提交
94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129
};

#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;
130
long vm_total_pages;	/* The total number of pages which the VM controls */
L
Linus Torvalds 已提交
131 132 133 134

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

135
#ifdef CONFIG_CGROUP_MEM_RES_CTLR
136
#define scanning_global_lru(sc)	(!(sc)->mem_cgroup)
137
#else
138
#define scanning_global_lru(sc)	(1)
139 140
#endif

141 142 143
static struct zone_reclaim_stat *get_reclaim_stat(struct zone *zone,
						  struct scan_control *sc)
{
144
	if (!scanning_global_lru(sc))
K
KOSAKI Motohiro 已提交
145 146
		return mem_cgroup_get_reclaim_stat(sc->mem_cgroup, zone);

147 148 149
	return &zone->reclaim_stat;
}

150 151
static unsigned long zone_nr_lru_pages(struct zone *zone,
				struct scan_control *sc, enum lru_list lru)
152
{
153
	if (!scanning_global_lru(sc))
154 155
		return mem_cgroup_zone_nr_pages(sc->mem_cgroup, zone, lru);

156 157 158 159
	return zone_page_state(zone, NR_LRU_BASE + lru);
}


L
Linus Torvalds 已提交
160 161 162
/*
 * Add a shrinker callback to be called from the vm
 */
163
void register_shrinker(struct shrinker *shrinker)
L
Linus Torvalds 已提交
164
{
165 166 167 168
	shrinker->nr = 0;
	down_write(&shrinker_rwsem);
	list_add_tail(&shrinker->list, &shrinker_list);
	up_write(&shrinker_rwsem);
L
Linus Torvalds 已提交
169
}
170
EXPORT_SYMBOL(register_shrinker);
L
Linus Torvalds 已提交
171 172 173 174

/*
 * Remove one
 */
175
void unregister_shrinker(struct shrinker *shrinker)
L
Linus Torvalds 已提交
176 177 178 179 180
{
	down_write(&shrinker_rwsem);
	list_del(&shrinker->list);
	up_write(&shrinker_rwsem);
}
181
EXPORT_SYMBOL(unregister_shrinker);
L
Linus Torvalds 已提交
182 183 184 185 186 187 188 189 190 191

#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
192
 * If the vm encountered mapped pages on the LRU it increase the pressure on
L
Linus Torvalds 已提交
193 194 195 196 197 198 199
 * 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.
200 201
 *
 * Returns the number of slab objects which we shrunk.
L
Linus Torvalds 已提交
202
 */
203 204
unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
			unsigned long lru_pages)
L
Linus Torvalds 已提交
205 206
{
	struct shrinker *shrinker;
207
	unsigned long ret = 0;
L
Linus Torvalds 已提交
208 209 210 211 212

	if (scanned == 0)
		scanned = SWAP_CLUSTER_MAX;

	if (!down_read_trylock(&shrinker_rwsem))
213
		return 1;	/* Assume we'll be able to shrink next time */
L
Linus Torvalds 已提交
214 215 216 217

	list_for_each_entry(shrinker, &shrinker_list, list) {
		unsigned long long delta;
		unsigned long total_scan;
218
		unsigned long max_pass = (*shrinker->shrink)(0, gfp_mask);
L
Linus Torvalds 已提交
219 220

		delta = (4 * scanned) / shrinker->seeks;
221
		delta *= max_pass;
L
Linus Torvalds 已提交
222 223
		do_div(delta, lru_pages + 1);
		shrinker->nr += delta;
224
		if (shrinker->nr < 0) {
225 226 227
			printk(KERN_ERR "shrink_slab: %pF negative objects to "
			       "delete nr=%ld\n",
			       shrinker->shrink, shrinker->nr);
228 229 230 231 232 233 234 235 236 237
			shrinker->nr = max_pass;
		}

		/*
		 * Avoid risking looping forever due to too large nr value:
		 * never try to free more than twice the estimate number of
		 * freeable entries.
		 */
		if (shrinker->nr > max_pass * 2)
			shrinker->nr = max_pass * 2;
L
Linus Torvalds 已提交
238 239 240 241 242 243 244

		total_scan = shrinker->nr;
		shrinker->nr = 0;

		while (total_scan >= SHRINK_BATCH) {
			long this_scan = SHRINK_BATCH;
			int shrink_ret;
245
			int nr_before;
L
Linus Torvalds 已提交
246

247 248
			nr_before = (*shrinker->shrink)(0, gfp_mask);
			shrink_ret = (*shrinker->shrink)(this_scan, gfp_mask);
L
Linus Torvalds 已提交
249 250
			if (shrink_ret == -1)
				break;
251 252
			if (shrink_ret < nr_before)
				ret += nr_before - shrink_ret;
253
			count_vm_events(SLABS_SCANNED, this_scan);
L
Linus Torvalds 已提交
254 255 256 257 258 259 260 261
			total_scan -= this_scan;

			cond_resched();
		}

		shrinker->nr += total_scan;
	}
	up_read(&shrinker_rwsem);
262
	return ret;
L
Linus Torvalds 已提交
263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287
}

/* Called without lock on whether page is mapped, so answer is unstable */
static inline int page_mapping_inuse(struct page *page)
{
	struct address_space *mapping;

	/* Page is in somebody's page tables. */
	if (page_mapped(page))
		return 1;

	/* Be more reluctant to reclaim swapcache than pagecache */
	if (PageSwapCache(page))
		return 1;

	mapping = page_mapping(page);
	if (!mapping)
		return 0;

	/* File is mmap'd by somebody? */
	return mapping_mapped(mapping);
}

static inline int is_page_cache_freeable(struct page *page)
{
288 289 290 291 292
	/*
	 * 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.
	 */
293
	return page_count(page) - page_has_private(page) == 2;
L
Linus Torvalds 已提交
294 295 296 297
}

static int may_write_to_queue(struct backing_dev_info *bdi)
{
298
	if (current->flags & PF_SWAPWRITE)
L
Linus Torvalds 已提交
299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322
		return 1;
	if (!bdi_write_congested(bdi))
		return 1;
	if (bdi == current->backing_dev_info)
		return 1;
	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)
{
	lock_page(page);
323 324
	if (page_mapping(page) == mapping)
		mapping_set_error(mapping, error);
L
Linus Torvalds 已提交
325 326 327
	unlock_page(page);
}

328 329 330 331 332 333
/* Request for sync pageout. */
enum pageout_io {
	PAGEOUT_IO_ASYNC,
	PAGEOUT_IO_SYNC,
};

334 335 336 337 338 339 340 341 342 343 344 345
/* 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
346
/*
A
Andrew Morton 已提交
347 348
 * pageout is called by shrink_page_list() for each dirty page.
 * Calls ->writepage().
L
Linus Torvalds 已提交
349
 */
350 351
static pageout_t pageout(struct page *page, struct address_space *mapping,
						enum pageout_io sync_writeback)
L
Linus Torvalds 已提交
352 353 354 355 356 357 358 359
{
	/*
	 * 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.
	 *
360
	 * If this process is currently in __generic_file_aio_write() against
L
Linus Torvalds 已提交
361 362 363 364 365 366 367 368 369 370 371 372 373 374 375
	 * 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.
		 */
376
		if (page_has_private(page)) {
L
Linus Torvalds 已提交
377 378
			if (try_to_free_buffers(page)) {
				ClearPageDirty(page);
379
				printk("%s: orphaned page\n", __func__);
L
Linus Torvalds 已提交
380 381 382 383 384 385 386 387 388 389 390 391 392 393 394
				return PAGE_CLEAN;
			}
		}
		return PAGE_KEEP;
	}
	if (mapping->a_ops->writepage == NULL)
		return PAGE_ACTIVATE;
	if (!may_write_to_queue(mapping->backing_dev_info))
		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,
395 396
			.range_start = 0,
			.range_end = LLONG_MAX,
L
Linus Torvalds 已提交
397 398 399 400 401 402 403 404
			.nonblocking = 1,
			.for_reclaim = 1,
		};

		SetPageReclaim(page);
		res = mapping->a_ops->writepage(page, &wbc);
		if (res < 0)
			handle_write_error(mapping, page, res);
405
		if (res == AOP_WRITEPAGE_ACTIVATE) {
L
Linus Torvalds 已提交
406 407 408
			ClearPageReclaim(page);
			return PAGE_ACTIVATE;
		}
409 410 411 412 413 414 415 416 417

		/*
		 * 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.
		 */
		if (PageWriteback(page) && sync_writeback == PAGEOUT_IO_SYNC)
			wait_on_page_writeback(page);

L
Linus Torvalds 已提交
418 419 420 421
		if (!PageWriteback(page)) {
			/* synchronous write or broken a_ops? */
			ClearPageReclaim(page);
		}
422
		inc_zone_page_state(page, NR_VMSCAN_WRITE);
L
Linus Torvalds 已提交
423 424 425 426 427 428
		return PAGE_SUCCESS;
	}

	return PAGE_CLEAN;
}

429
/*
N
Nick Piggin 已提交
430 431
 * Same as remove_mapping, but if the page is removed from the mapping, it
 * gets returned with a refcount of 0.
432
 */
N
Nick Piggin 已提交
433
static int __remove_mapping(struct address_space *mapping, struct page *page)
434
{
435 436
	BUG_ON(!PageLocked(page));
	BUG_ON(mapping != page_mapping(page));
437

N
Nick Piggin 已提交
438
	spin_lock_irq(&mapping->tree_lock);
439
	/*
N
Nick Piggin 已提交
440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462
	 * 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.
463
	 */
N
Nick Piggin 已提交
464
	if (!page_freeze_refs(page, 2))
465
		goto cannot_free;
N
Nick Piggin 已提交
466 467 468
	/* note: atomic_cmpxchg in page_freeze_refs provides the smp_rmb */
	if (unlikely(PageDirty(page))) {
		page_unfreeze_refs(page, 2);
469
		goto cannot_free;
N
Nick Piggin 已提交
470
	}
471 472 473 474

	if (PageSwapCache(page)) {
		swp_entry_t swap = { .val = page_private(page) };
		__delete_from_swap_cache(page);
N
Nick Piggin 已提交
475
		spin_unlock_irq(&mapping->tree_lock);
476
		swapcache_free(swap, page);
N
Nick Piggin 已提交
477 478
	} else {
		__remove_from_page_cache(page);
N
Nick Piggin 已提交
479
		spin_unlock_irq(&mapping->tree_lock);
480
		mem_cgroup_uncharge_cache_page(page);
481 482 483 484 485
	}

	return 1;

cannot_free:
N
Nick Piggin 已提交
486
	spin_unlock_irq(&mapping->tree_lock);
487 488 489
	return 0;
}

N
Nick Piggin 已提交
490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509
/*
 * 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
510 511 512 513 514 515 516 517 518 519 520 521 522
/**
 * 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);
523
	int was_unevictable = PageUnevictable(page);
L
Lee Schermerhorn 已提交
524 525 526 527 528 529 530 531 532 533 534 535 536

	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.
		 */
537
		lru = active + page_lru_base_type(page);
L
Lee Schermerhorn 已提交
538 539 540 541 542 543 544 545
		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);
546 547 548 549 550 551 552 553 554 555
		/*
		 * 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
556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573
	}

	/*
	 * 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.
		 */
	}

574 575 576 577 578
	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
579 580 581
	put_page(page);		/* drop ref from isolate */
}

L
Linus Torvalds 已提交
582
/*
A
Andrew Morton 已提交
583
 * shrink_page_list() returns the number of reclaimed pages
L
Linus Torvalds 已提交
584
 */
A
Andrew Morton 已提交
585
static unsigned long shrink_page_list(struct list_head *page_list,
586 587
					struct scan_control *sc,
					enum pageout_io sync_writeback)
L
Linus Torvalds 已提交
588 589 590 591
{
	LIST_HEAD(ret_pages);
	struct pagevec freed_pvec;
	int pgactivate = 0;
592
	unsigned long nr_reclaimed = 0;
593
	unsigned long vm_flags;
L
Linus Torvalds 已提交
594 595 596 597 598 599 600 601 602 603 604 605 606 607 608

	cond_resched();

	pagevec_init(&freed_pvec, 1);
	while (!list_empty(page_list)) {
		struct address_space *mapping;
		struct page *page;
		int may_enter_fs;
		int referenced;

		cond_resched();

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

N
Nick Piggin 已提交
609
		if (!trylock_page(page))
L
Linus Torvalds 已提交
610 611
			goto keep;

N
Nick Piggin 已提交
612
		VM_BUG_ON(PageActive(page));
L
Linus Torvalds 已提交
613 614

		sc->nr_scanned++;
615

N
Nick Piggin 已提交
616 617
		if (unlikely(!page_evictable(page, NULL)))
			goto cull_mlocked;
L
Lee Schermerhorn 已提交
618

619
		if (!sc->may_unmap && page_mapped(page))
620 621
			goto keep_locked;

L
Linus Torvalds 已提交
622 623 624 625
		/* Double the slab pressure for mapped and swapcache pages */
		if (page_mapped(page) || PageSwapCache(page))
			sc->nr_scanned++;

626 627 628 629 630 631 632 633 634 635 636 637 638 639
		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.
			 */
			if (sync_writeback == PAGEOUT_IO_SYNC && may_enter_fs)
				wait_on_page_writeback(page);
640
			else
641 642
				goto keep_locked;
		}
L
Linus Torvalds 已提交
643

644 645
		referenced = page_referenced(page, 1,
						sc->mem_cgroup, &vm_flags);
646 647 648 649 650
		/*
		 * In active use or really unfreeable?  Activate it.
		 * If page which have PG_mlocked lost isoltation race,
		 * try_to_unmap moves it to unevictable list
		 */
A
Andy Whitcroft 已提交
651
		if (sc->order <= PAGE_ALLOC_COSTLY_ORDER &&
652 653
					referenced && page_mapping_inuse(page)
					&& !(vm_flags & VM_LOCKED))
L
Linus Torvalds 已提交
654 655 656 657 658 659
			goto activate_locked;

		/*
		 * Anonymous process memory has backing store?
		 * Try to allocate it some swap space here.
		 */
N
Nick Piggin 已提交
660
		if (PageAnon(page) && !PageSwapCache(page)) {
661 662
			if (!(sc->gfp_mask & __GFP_IO))
				goto keep_locked;
663
			if (!add_to_swap(page))
L
Linus Torvalds 已提交
664
				goto activate_locked;
665
			may_enter_fs = 1;
N
Nick Piggin 已提交
666
		}
L
Linus Torvalds 已提交
667 668 669 670 671 672 673 674

		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) {
675
			switch (try_to_unmap(page, TTU_UNMAP)) {
L
Linus Torvalds 已提交
676 677 678 679
			case SWAP_FAIL:
				goto activate_locked;
			case SWAP_AGAIN:
				goto keep_locked;
N
Nick Piggin 已提交
680 681
			case SWAP_MLOCK:
				goto cull_mlocked;
L
Linus Torvalds 已提交
682 683 684 685 686 687
			case SWAP_SUCCESS:
				; /* try to free the page below */
			}
		}

		if (PageDirty(page)) {
A
Andy Whitcroft 已提交
688
			if (sc->order <= PAGE_ALLOC_COSTLY_ORDER && referenced)
L
Linus Torvalds 已提交
689
				goto keep_locked;
690
			if (!may_enter_fs)
L
Linus Torvalds 已提交
691
				goto keep_locked;
692
			if (!sc->may_writepage)
L
Linus Torvalds 已提交
693 694 695
				goto keep_locked;

			/* Page is dirty, try to write it out here */
696
			switch (pageout(page, mapping, sync_writeback)) {
L
Linus Torvalds 已提交
697 698 699 700 701
			case PAGE_KEEP:
				goto keep_locked;
			case PAGE_ACTIVATE:
				goto activate_locked;
			case PAGE_SUCCESS:
702
				if (PageWriteback(page) || PageDirty(page))
L
Linus Torvalds 已提交
703 704 705 706 707
					goto keep;
				/*
				 * A synchronous write - probably a ramdisk.  Go
				 * ahead and try to reclaim the page.
				 */
N
Nick Piggin 已提交
708
				if (!trylock_page(page))
L
Linus Torvalds 已提交
709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727
					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
728
		 * will do this, as well as the blockdev mapping.
L
Linus Torvalds 已提交
729 730 731 732 733 734 735 736 737 738
		 * 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.
		 */
739
		if (page_has_private(page)) {
L
Linus Torvalds 已提交
740 741
			if (!try_to_release_page(page, sc->gfp_mask))
				goto activate_locked;
N
Nick Piggin 已提交
742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757
			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
758 759
		}

N
Nick Piggin 已提交
760
		if (!mapping || !__remove_mapping(mapping, page))
761
			goto keep_locked;
L
Linus Torvalds 已提交
762

N
Nick Piggin 已提交
763 764 765 766 767 768 769 770
		/*
		 * 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
771
free_it:
772
		nr_reclaimed++;
N
Nick Piggin 已提交
773 774 775 776
		if (!pagevec_add(&freed_pvec, page)) {
			__pagevec_free(&freed_pvec);
			pagevec_reinit(&freed_pvec);
		}
L
Linus Torvalds 已提交
777 778
		continue;

N
Nick Piggin 已提交
779
cull_mlocked:
780 781
		if (PageSwapCache(page))
			try_to_free_swap(page);
N
Nick Piggin 已提交
782 783 784 785
		unlock_page(page);
		putback_lru_page(page);
		continue;

L
Linus Torvalds 已提交
786
activate_locked:
787 788
		/* Not a candidate for swapping, so reclaim swap space. */
		if (PageSwapCache(page) && vm_swap_full())
789
			try_to_free_swap(page);
L
Lee Schermerhorn 已提交
790
		VM_BUG_ON(PageActive(page));
L
Linus Torvalds 已提交
791 792 793 794 795 796
		SetPageActive(page);
		pgactivate++;
keep_locked:
		unlock_page(page);
keep:
		list_add(&page->lru, &ret_pages);
N
Nick Piggin 已提交
797
		VM_BUG_ON(PageLRU(page) || PageUnevictable(page));
L
Linus Torvalds 已提交
798 799 800
	}
	list_splice(&ret_pages, page_list);
	if (pagevec_count(&freed_pvec))
N
Nick Piggin 已提交
801
		__pagevec_free(&freed_pvec);
802
	count_vm_events(PGACTIVATE, pgactivate);
803
	return nr_reclaimed;
L
Linus Torvalds 已提交
804 805
}

A
Andy Whitcroft 已提交
806 807 808 809 810 811 812 813 814 815 816 817 818 819 820
/* LRU Isolation modes. */
#define ISOLATE_INACTIVE 0	/* Isolate inactive pages. */
#define ISOLATE_ACTIVE 1	/* Isolate active pages. */
#define ISOLATE_BOTH 2		/* Isolate both active and inactive pages. */

/*
 * 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.
 */
821
int __isolate_lru_page(struct page *page, int mode, int file)
A
Andy Whitcroft 已提交
822 823 824 825 826 827 828 829 830 831 832 833 834 835 836
{
	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;

837
	if (mode != ISOLATE_BOTH && page_is_file_cache(page) != file)
838 839
		return ret;

L
Lee Schermerhorn 已提交
840 841 842 843 844 845 846 847
	/*
	 * 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
848
	ret = -EBUSY;
K
KAMEZAWA Hiroyuki 已提交
849

A
Andy Whitcroft 已提交
850 851 852 853 854 855 856 857 858 859 860 861 862
	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
863 864 865 866 867 868 869 870 871 872 873 874 875 876
/*
 * 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
877 878
 * @order:	The caller's attempted allocation order
 * @mode:	One of the LRU isolation modes
879
 * @file:	True [1] if isolating file [!anon] pages
L
Linus Torvalds 已提交
880 881 882
 *
 * returns how many pages were moved onto *@dst.
 */
883 884
static unsigned long isolate_lru_pages(unsigned long nr_to_scan,
		struct list_head *src, struct list_head *dst,
885
		unsigned long *scanned, int order, int mode, int file)
L
Linus Torvalds 已提交
886
{
887
	unsigned long nr_taken = 0;
888
	unsigned long scan;
L
Linus Torvalds 已提交
889

890
	for (scan = 0; scan < nr_to_scan && !list_empty(src); scan++) {
A
Andy Whitcroft 已提交
891 892 893 894 895 896
		struct page *page;
		unsigned long pfn;
		unsigned long end_pfn;
		unsigned long page_pfn;
		int zone_id;

L
Linus Torvalds 已提交
897 898 899
		page = lru_to_page(src);
		prefetchw_prev_lru_page(page, src, flags);

N
Nick Piggin 已提交
900
		VM_BUG_ON(!PageLRU(page));
N
Nick Piggin 已提交
901

902
		switch (__isolate_lru_page(page, mode, file)) {
A
Andy Whitcroft 已提交
903 904
		case 0:
			list_move(&page->lru, dst);
905
			mem_cgroup_del_lru(page);
906
			nr_taken++;
A
Andy Whitcroft 已提交
907 908 909 910 911
			break;

		case -EBUSY:
			/* else it is being freed elsewhere */
			list_move(&page->lru, src);
912
			mem_cgroup_rotate_lru_list(page, page_lru(page));
A
Andy Whitcroft 已提交
913
			continue;
914

A
Andy Whitcroft 已提交
915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946
		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
		 * as the mem_map is guarenteed valid out to MAX_ORDER,
		 * 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);
947

A
Andy Whitcroft 已提交
948 949 950
			/* Check that we have not crossed a zone boundary. */
			if (unlikely(page_zone_id(cursor_page) != zone_id))
				continue;
951 952 953 954 955 956 957 958 959 960

			/*
			 * 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) &&
					!PageSwapCache(cursor_page))
				continue;

961
			if (__isolate_lru_page(cursor_page, mode, file) == 0) {
A
Andy Whitcroft 已提交
962
				list_move(&cursor_page->lru, dst);
963
				mem_cgroup_del_lru(cursor_page);
A
Andy Whitcroft 已提交
964 965 966 967
				nr_taken++;
				scan++;
			}
		}
L
Linus Torvalds 已提交
968 969 970 971 972 973
	}

	*scanned = scan;
	return nr_taken;
}

974 975 976 977 978
static unsigned long isolate_pages_global(unsigned long nr,
					struct list_head *dst,
					unsigned long *scanned, int order,
					int mode, struct zone *z,
					struct mem_cgroup *mem_cont,
979
					int active, int file)
980
{
981
	int lru = LRU_BASE;
982
	if (active)
983 984 985 986
		lru += LRU_ACTIVE;
	if (file)
		lru += LRU_FILE;
	return isolate_lru_pages(nr, &z->lru[lru].list, dst, scanned, order,
987
								mode, file);
988 989
}

A
Andy Whitcroft 已提交
990 991 992 993
/*
 * clear_active_flags() is a helper for shrink_active_list(), clearing
 * any active bits from the pages in the list.
 */
994 995
static unsigned long clear_active_flags(struct list_head *page_list,
					unsigned int *count)
A
Andy Whitcroft 已提交
996 997
{
	int nr_active = 0;
998
	int lru;
A
Andy Whitcroft 已提交
999 1000
	struct page *page;

1001
	list_for_each_entry(page, page_list, lru) {
1002
		lru = page_lru_base_type(page);
A
Andy Whitcroft 已提交
1003
		if (PageActive(page)) {
1004
			lru += LRU_ACTIVE;
A
Andy Whitcroft 已提交
1005 1006 1007
			ClearPageActive(page);
			nr_active++;
		}
1008 1009
		count[lru]++;
	}
A
Andy Whitcroft 已提交
1010 1011 1012 1013

	return nr_active;
}

1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024
/**
 * 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
1025 1026 1027
 * 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.
1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047
 *
 * 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;

	if (PageLRU(page)) {
		struct zone *zone = page_zone(page);

		spin_lock_irq(&zone->lru_lock);
		if (PageLRU(page) && get_page_unless_zero(page)) {
L
Lee Schermerhorn 已提交
1048
			int lru = page_lru(page);
1049 1050
			ret = 0;
			ClearPageLRU(page);
1051 1052

			del_page_from_lru_list(zone, page, lru);
1053 1054 1055 1056 1057 1058
		}
		spin_unlock_irq(&zone->lru_lock);
	}
	return ret;
}

1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083
/*
 * 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;
}

L
Linus Torvalds 已提交
1084
/*
A
Andrew Morton 已提交
1085 1086
 * shrink_inactive_list() is a helper for shrink_zone().  It returns the number
 * of reclaimed pages
L
Linus Torvalds 已提交
1087
 */
A
Andrew Morton 已提交
1088
static unsigned long shrink_inactive_list(unsigned long max_scan,
R
Rik van Riel 已提交
1089 1090
			struct zone *zone, struct scan_control *sc,
			int priority, int file)
L
Linus Torvalds 已提交
1091 1092 1093
{
	LIST_HEAD(page_list);
	struct pagevec pvec;
1094
	unsigned long nr_scanned = 0;
1095
	unsigned long nr_reclaimed = 0;
1096
	struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(zone, sc);
1097 1098
	int lumpy_reclaim = 0;

1099
	while (unlikely(too_many_isolated(zone, file, sc))) {
1100
		congestion_wait(BLK_RW_ASYNC, HZ/10);
1101 1102 1103 1104 1105 1106

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

1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117
	/*
	 * If we need a large contiguous chunk of memory, or have
	 * trouble getting a small set of contiguous pages, we
	 * will reclaim both active and inactive pages.
	 *
	 * We use the same threshold as pageout congestion_wait below.
	 */
	if (sc->order > PAGE_ALLOC_COSTLY_ORDER)
		lumpy_reclaim = 1;
	else if (sc->order && priority < DEF_PRIORITY - 2)
		lumpy_reclaim = 1;
L
Linus Torvalds 已提交
1118 1119 1120 1121 1122

	pagevec_init(&pvec, 1);

	lru_add_drain();
	spin_lock_irq(&zone->lru_lock);
1123
	do {
L
Linus Torvalds 已提交
1124
		struct page *page;
1125 1126 1127
		unsigned long nr_taken;
		unsigned long nr_scan;
		unsigned long nr_freed;
A
Andy Whitcroft 已提交
1128
		unsigned long nr_active;
1129
		unsigned int count[NR_LRU_LISTS] = { 0, };
1130
		int mode = lumpy_reclaim ? ISOLATE_BOTH : ISOLATE_INACTIVE;
K
KOSAKI Motohiro 已提交
1131 1132
		unsigned long nr_anon;
		unsigned long nr_file;
L
Linus Torvalds 已提交
1133

K
KOSAKI Motohiro 已提交
1134
		nr_taken = sc->isolate_pages(SWAP_CLUSTER_MAX,
1135 1136
			     &page_list, &nr_scan, sc->order, mode,
				zone, sc->mem_cgroup, 0, file);
1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150

		if (scanning_global_lru(sc)) {
			zone->pages_scanned += nr_scan;
			if (current_is_kswapd())
				__count_zone_vm_events(PGSCAN_KSWAPD, zone,
						       nr_scan);
			else
				__count_zone_vm_events(PGSCAN_DIRECT, zone,
						       nr_scan);
		}

		if (nr_taken == 0)
			goto done;

1151
		nr_active = clear_active_flags(&page_list, count);
1152
		__count_vm_events(PGDEACTIVATE, nr_active);
A
Andy Whitcroft 已提交
1153

1154 1155 1156 1157 1158 1159 1160 1161 1162
		__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]);

K
KOSAKI Motohiro 已提交
1163 1164 1165 1166
		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);
K
KOSAKI Motohiro 已提交
1167

H
Huang Shijie 已提交
1168 1169
		reclaim_stat->recent_scanned[0] += nr_anon;
		reclaim_stat->recent_scanned[1] += nr_file;
K
KOSAKI Motohiro 已提交
1170

L
Linus Torvalds 已提交
1171 1172
		spin_unlock_irq(&zone->lru_lock);

1173
		nr_scanned += nr_scan;
1174 1175 1176 1177 1178 1179 1180 1181 1182
		nr_freed = shrink_page_list(&page_list, sc, PAGEOUT_IO_ASYNC);

		/*
		 * If we are direct reclaiming for contiguous pages and we do
		 * not reclaim everything in the list, try again and wait
		 * for IO to complete. This will stall high-order allocations
		 * but that should be acceptable to the caller
		 */
		if (nr_freed < nr_taken && !current_is_kswapd() &&
1183
		    lumpy_reclaim) {
1184
			congestion_wait(BLK_RW_ASYNC, HZ/10);
1185 1186 1187 1188 1189

			/*
			 * The attempt at page out may have made some
			 * of the pages active, mark them inactive again.
			 */
1190
			nr_active = clear_active_flags(&page_list, count);
1191 1192 1193 1194 1195 1196
			count_vm_events(PGDEACTIVATE, nr_active);

			nr_freed += shrink_page_list(&page_list, sc,
							PAGEOUT_IO_SYNC);
		}

1197
		nr_reclaimed += nr_freed;
1198

N
Nick Piggin 已提交
1199
		local_irq_disable();
1200
		if (current_is_kswapd())
1201
			__count_vm_events(KSWAPD_STEAL, nr_freed);
S
Shantanu Goel 已提交
1202
		__count_zone_vm_events(PGSTEAL, zone, nr_freed);
N
Nick Piggin 已提交
1203 1204

		spin_lock(&zone->lru_lock);
L
Linus Torvalds 已提交
1205 1206 1207 1208
		/*
		 * Put back any unfreeable pages.
		 */
		while (!list_empty(&page_list)) {
L
Lee Schermerhorn 已提交
1209
			int lru;
L
Linus Torvalds 已提交
1210
			page = lru_to_page(&page_list);
N
Nick Piggin 已提交
1211
			VM_BUG_ON(PageLRU(page));
L
Linus Torvalds 已提交
1212
			list_del(&page->lru);
L
Lee Schermerhorn 已提交
1213 1214 1215 1216 1217 1218 1219 1220 1221
			if (unlikely(!page_evictable(page, NULL))) {
				spin_unlock_irq(&zone->lru_lock);
				putback_lru_page(page);
				spin_lock_irq(&zone->lru_lock);
				continue;
			}
			SetPageLRU(page);
			lru = page_lru(page);
			add_page_to_lru_list(zone, page, lru);
1222
			if (is_active_lru(lru)) {
1223
				int file = is_file_lru(lru);
1224
				reclaim_stat->recent_rotated[file]++;
1225
			}
L
Linus Torvalds 已提交
1226 1227 1228 1229 1230 1231
			if (!pagevec_add(&pvec, page)) {
				spin_unlock_irq(&zone->lru_lock);
				__pagevec_release(&pvec);
				spin_lock_irq(&zone->lru_lock);
			}
		}
K
KOSAKI Motohiro 已提交
1232 1233 1234
		__mod_zone_page_state(zone, NR_ISOLATED_ANON, -nr_anon);
		__mod_zone_page_state(zone, NR_ISOLATED_FILE, -nr_file);

1235
  	} while (nr_scanned < max_scan);
1236

L
Linus Torvalds 已提交
1237
done:
1238
	spin_unlock_irq(&zone->lru_lock);
L
Linus Torvalds 已提交
1239
	pagevec_release(&pvec);
1240
	return nr_reclaimed;
L
Linus Torvalds 已提交
1241 1242
}

1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256
/*
 * We are about to scan this zone at a certain priority level.  If that priority
 * level is smaller (ie: more urgent) than the previous priority, then note
 * that priority level within the zone.  This is done so that when the next
 * process comes in to scan this zone, it will immediately start out at this
 * priority level rather than having to build up its own scanning priority.
 * Here, this priority affects only the reclaim-mapped threshold.
 */
static inline void note_zone_scanning_priority(struct zone *zone, int priority)
{
	if (priority < zone->prev_priority)
		zone->prev_priority = priority;
}

L
Linus Torvalds 已提交
1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273
/*
 * 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.
 */
1274

1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306
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);
		pgmoved++;

		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);
}
1307

A
Andrew Morton 已提交
1308
static void shrink_active_list(unsigned long nr_pages, struct zone *zone,
1309
			struct scan_control *sc, int priority, int file)
L
Linus Torvalds 已提交
1310
{
1311
	unsigned long nr_taken;
1312
	unsigned long pgscanned;
1313
	unsigned long vm_flags;
L
Linus Torvalds 已提交
1314
	LIST_HEAD(l_hold);	/* The pages which were snipped off */
1315
	LIST_HEAD(l_active);
1316
	LIST_HEAD(l_inactive);
L
Linus Torvalds 已提交
1317
	struct page *page;
1318
	struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(zone, sc);
1319
	unsigned long nr_rotated = 0;
L
Linus Torvalds 已提交
1320 1321 1322

	lru_add_drain();
	spin_lock_irq(&zone->lru_lock);
1323
	nr_taken = sc->isolate_pages(nr_pages, &l_hold, &pgscanned, sc->order,
1324
					ISOLATE_ACTIVE, zone,
1325
					sc->mem_cgroup, 1, file);
1326 1327 1328 1329
	/*
	 * zone->pages_scanned is used for detect zone's oom
	 * mem_cgroup remembers nr_scan by itself.
	 */
1330
	if (scanning_global_lru(sc)) {
1331
		zone->pages_scanned += pgscanned;
1332
	}
1333
	reclaim_stat->recent_scanned[file] += nr_taken;
1334

1335
	__count_zone_vm_events(PGREFILL, zone, pgscanned);
1336
	if (file)
1337
		__mod_zone_page_state(zone, NR_ACTIVE_FILE, -nr_taken);
1338
	else
1339
		__mod_zone_page_state(zone, NR_ACTIVE_ANON, -nr_taken);
K
KOSAKI Motohiro 已提交
1340
	__mod_zone_page_state(zone, NR_ISOLATED_ANON + file, nr_taken);
L
Linus Torvalds 已提交
1341 1342 1343 1344 1345 1346
	spin_unlock_irq(&zone->lru_lock);

	while (!list_empty(&l_hold)) {
		cond_resched();
		page = lru_to_page(&l_hold);
		list_del(&page->lru);
1347

L
Lee Schermerhorn 已提交
1348 1349 1350 1351 1352
		if (unlikely(!page_evictable(page, NULL))) {
			putback_lru_page(page);
			continue;
		}

1353 1354
		/* page_referenced clears PageReferenced */
		if (page_mapping_inuse(page) &&
1355
		    page_referenced(page, 0, sc->mem_cgroup, &vm_flags)) {
1356
			nr_rotated++;
1357 1358 1359 1360 1361 1362 1363 1364 1365
			/*
			 * 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.
			 */
1366
			if ((vm_flags & VM_EXEC) && page_is_file_cache(page)) {
1367 1368 1369 1370
				list_add(&page->lru, &l_active);
				continue;
			}
		}
1371

1372
		ClearPageActive(page);	/* we are de-activating */
L
Linus Torvalds 已提交
1373 1374 1375
		list_add(&page->lru, &l_inactive);
	}

1376
	/*
1377
	 * Move pages back to the lru list.
1378
	 */
1379
	spin_lock_irq(&zone->lru_lock);
1380
	/*
1381 1382 1383 1384
	 * 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.
1385
	 */
1386
	reclaim_stat->recent_rotated[file] += nr_rotated;
1387

1388 1389 1390 1391
	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
1392
	__mod_zone_page_state(zone, NR_ISOLATED_ANON + file, -nr_taken);
1393
	spin_unlock_irq(&zone->lru_lock);
L
Linus Torvalds 已提交
1394 1395
}

1396
static int inactive_anon_is_low_global(struct zone *zone)
1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408
{
	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;
}

1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420
/**
 * 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;

1421
	if (scanning_global_lru(sc))
1422 1423
		low = inactive_anon_is_low_global(zone);
	else
1424
		low = mem_cgroup_inactive_anon_is_low(sc->mem_cgroup);
1425 1426 1427
	return low;
}

1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463
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;
}

1464 1465 1466 1467 1468 1469 1470 1471 1472
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);
}

1473
static unsigned long shrink_list(enum lru_list lru, unsigned long nr_to_scan,
1474 1475
	struct zone *zone, struct scan_control *sc, int priority)
{
1476 1477
	int file = is_file_lru(lru);

1478 1479 1480
	if (is_active_lru(lru)) {
		if (inactive_list_is_low(zone, sc, file))
		    shrink_active_list(nr_to_scan, zone, sc, priority, file);
1481 1482 1483
		return 0;
	}

R
Rik van Riel 已提交
1484
	return shrink_inactive_list(nr_to_scan, zone, sc, priority, file);
1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501
}

/*
 * 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.
 *
 * percent[0] specifies how much pressure to put on ram/swap backed
 * memory, while percent[1] determines pressure on the file LRUs.
 */
static void get_scan_ratio(struct zone *zone, struct scan_control *sc,
					unsigned long *percent)
{
	unsigned long anon, file, free;
	unsigned long anon_prio, file_prio;
	unsigned long ap, fp;
1502
	struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(zone, sc);
1503

1504 1505 1506 1507
	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);
1508

1509
	if (scanning_global_lru(sc)) {
1510 1511 1512
		free  = zone_page_state(zone, NR_FREE_PAGES);
		/* If we have very few page cache pages,
		   force-scan anon pages. */
1513
		if (unlikely(file + free <= high_wmark_pages(zone))) {
1514 1515 1516 1517
			percent[0] = 100;
			percent[1] = 0;
			return;
		}
1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530
	}

	/*
	 * 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]
	 */
1531
	if (unlikely(reclaim_stat->recent_scanned[0] > anon / 4)) {
1532
		spin_lock_irq(&zone->lru_lock);
1533 1534
		reclaim_stat->recent_scanned[0] /= 2;
		reclaim_stat->recent_rotated[0] /= 2;
1535 1536 1537
		spin_unlock_irq(&zone->lru_lock);
	}

1538
	if (unlikely(reclaim_stat->recent_scanned[1] > file / 4)) {
1539
		spin_lock_irq(&zone->lru_lock);
1540 1541
		reclaim_stat->recent_scanned[1] /= 2;
		reclaim_stat->recent_rotated[1] /= 2;
1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552
		spin_unlock_irq(&zone->lru_lock);
	}

	/*
	 * With swappiness at 100, anonymous and file have the same priority.
	 * This scanning priority is essentially the inverse of IO cost.
	 */
	anon_prio = sc->swappiness;
	file_prio = 200 - sc->swappiness;

	/*
1553 1554 1555
	 * 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.
1556
	 */
1557 1558
	ap = (anon_prio + 1) * (reclaim_stat->recent_scanned[0] + 1);
	ap /= reclaim_stat->recent_rotated[0] + 1;
1559

1560 1561
	fp = (file_prio + 1) * (reclaim_stat->recent_scanned[1] + 1);
	fp /= reclaim_stat->recent_rotated[1] + 1;
1562 1563 1564 1565

	/* Normalize to percentages */
	percent[0] = 100 * ap / (ap + fp + 1);
	percent[1] = 100 - percent[0];
1566 1567
}

1568 1569 1570 1571 1572
/*
 * Smallish @nr_to_scan's are deposited in @nr_saved_scan,
 * until we collected @swap_cluster_max pages to scan.
 */
static unsigned long nr_scan_try_batch(unsigned long nr_to_scan,
K
KOSAKI Motohiro 已提交
1573
				       unsigned long *nr_saved_scan)
1574 1575 1576 1577 1578 1579
{
	unsigned long nr;

	*nr_saved_scan += nr_to_scan;
	nr = *nr_saved_scan;

K
KOSAKI Motohiro 已提交
1580
	if (nr >= SWAP_CLUSTER_MAX)
1581 1582 1583 1584 1585 1586
		*nr_saved_scan = 0;
	else
		nr = 0;

	return nr;
}
1587

L
Linus Torvalds 已提交
1588 1589 1590
/*
 * This is a basic per-zone page freer.  Used by both kswapd and direct reclaim.
 */
1591
static void shrink_zone(int priority, struct zone *zone,
1592
				struct scan_control *sc)
L
Linus Torvalds 已提交
1593
{
1594
	unsigned long nr[NR_LRU_LISTS];
1595
	unsigned long nr_to_scan;
1596
	unsigned long percent[2];	/* anon @ 0; file @ 1 */
1597
	enum lru_list l;
1598
	unsigned long nr_reclaimed = sc->nr_reclaimed;
1599
	unsigned long nr_to_reclaim = sc->nr_to_reclaim;
1600
	struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(zone, sc);
1601
	int noswap = 0;
L
Linus Torvalds 已提交
1602

1603 1604 1605 1606 1607 1608 1609
	/* If we have no swap space, do not bother scanning anon pages. */
	if (!sc->may_swap || (nr_swap_pages <= 0)) {
		noswap = 1;
		percent[0] = 0;
		percent[1] = 100;
	} else
		get_scan_ratio(zone, sc, percent);
1610

L
Lee Schermerhorn 已提交
1611
	for_each_evictable_lru(l) {
1612
		int file = is_file_lru(l);
1613
		unsigned long scan;
1614

1615
		scan = zone_nr_lru_pages(zone, sc, l);
1616
		if (priority || noswap) {
1617 1618 1619
			scan >>= priority;
			scan = (scan * percent[file]) / 100;
		}
1620
		nr[l] = nr_scan_try_batch(scan,
K
KOSAKI Motohiro 已提交
1621
					  &reclaim_stat->nr_saved_scan[l]);
1622
	}
L
Linus Torvalds 已提交
1623

1624 1625
	while (nr[LRU_INACTIVE_ANON] || nr[LRU_ACTIVE_FILE] ||
					nr[LRU_INACTIVE_FILE]) {
L
Lee Schermerhorn 已提交
1626
		for_each_evictable_lru(l) {
1627
			if (nr[l]) {
K
KOSAKI Motohiro 已提交
1628 1629
				nr_to_scan = min_t(unsigned long,
						   nr[l], SWAP_CLUSTER_MAX);
1630
				nr[l] -= nr_to_scan;
L
Linus Torvalds 已提交
1631

1632 1633
				nr_reclaimed += shrink_list(l, nr_to_scan,
							    zone, sc, priority);
1634
			}
L
Linus Torvalds 已提交
1635
		}
1636 1637 1638 1639 1640 1641 1642 1643
		/*
		 * 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.
		 */
1644
		if (nr_reclaimed >= nr_to_reclaim && priority < DEF_PRIORITY)
1645
			break;
L
Linus Torvalds 已提交
1646 1647
	}

1648 1649
	sc->nr_reclaimed = nr_reclaimed;

1650 1651 1652 1653
	/*
	 * Even if we did not try to evict anon pages at all, we want to
	 * rebalance the anon lru active/inactive ratio.
	 */
1654
	if (inactive_anon_is_low(zone, sc) && nr_swap_pages > 0)
1655 1656
		shrink_active_list(SWAP_CLUSTER_MAX, zone, sc, priority, 0);

1657
	throttle_vm_writeout(sc->gfp_mask);
L
Linus Torvalds 已提交
1658 1659 1660 1661 1662 1663 1664
}

/*
 * 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.
 *
1665 1666
 * We reclaim from a zone even if that zone is over high_wmark_pages(zone).
 * Because:
L
Linus Torvalds 已提交
1667 1668
 * a) The caller may be trying to free *extra* pages to satisfy a higher-order
 *    allocation or
1669 1670 1671
 * 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
1672 1673 1674 1675
 *
 * If a zone is deemed to be full of pinned pages then just give it a light
 * scan then give up on it.
 */
1676
static void shrink_zones(int priority, struct zonelist *zonelist,
1677
					struct scan_control *sc)
L
Linus Torvalds 已提交
1678
{
1679
	enum zone_type high_zoneidx = gfp_zone(sc->gfp_mask);
1680
	struct zoneref *z;
1681
	struct zone *zone;
1682

1683
	sc->all_unreclaimable = 1;
1684 1685
	for_each_zone_zonelist_nodemask(zone, z, zonelist, high_zoneidx,
					sc->nodemask) {
1686
		if (!populated_zone(zone))
L
Linus Torvalds 已提交
1687
			continue;
1688 1689 1690 1691
		/*
		 * Take care memory controller reclaiming has small influence
		 * to global LRU.
		 */
1692
		if (scanning_global_lru(sc)) {
1693 1694 1695
			if (!cpuset_zone_allowed_hardwall(zone, GFP_KERNEL))
				continue;
			note_zone_scanning_priority(zone, priority);
L
Linus Torvalds 已提交
1696

1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709
			if (zone_is_all_unreclaimable(zone) &&
						priority != DEF_PRIORITY)
				continue;	/* Let kswapd poll it */
			sc->all_unreclaimable = 0;
		} else {
			/*
			 * Ignore cpuset limitation here. We just want to reduce
			 * # of used pages by us regardless of memory shortage.
			 */
			sc->all_unreclaimable = 0;
			mem_cgroup_note_reclaim_priority(sc->mem_cgroup,
							priority);
		}
1710

1711
		shrink_zone(priority, zone, sc);
L
Linus Torvalds 已提交
1712 1713
	}
}
1714

L
Linus Torvalds 已提交
1715 1716 1717 1718 1719 1720 1721 1722
/*
 * 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
1723 1724 1725 1726
 * 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.
1727 1728 1729
 *
 * returns:	0, if no pages reclaimed
 * 		else, the number of pages reclaimed
L
Linus Torvalds 已提交
1730
 */
1731
static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
1732
					struct scan_control *sc)
L
Linus Torvalds 已提交
1733 1734
{
	int priority;
1735
	unsigned long ret = 0;
1736
	unsigned long total_scanned = 0;
L
Linus Torvalds 已提交
1737 1738
	struct reclaim_state *reclaim_state = current->reclaim_state;
	unsigned long lru_pages = 0;
1739
	struct zoneref *z;
1740
	struct zone *zone;
1741
	enum zone_type high_zoneidx = gfp_zone(sc->gfp_mask);
1742
	unsigned long writeback_threshold;
L
Linus Torvalds 已提交
1743

1744 1745
	delayacct_freepages_start();

1746
	if (scanning_global_lru(sc))
1747 1748 1749 1750
		count_vm_event(ALLOCSTALL);
	/*
	 * mem_cgroup will not do shrink_slab.
	 */
1751
	if (scanning_global_lru(sc)) {
1752
		for_each_zone_zonelist(zone, z, zonelist, high_zoneidx) {
L
Linus Torvalds 已提交
1753

1754 1755
			if (!cpuset_zone_allowed_hardwall(zone, GFP_KERNEL))
				continue;
L
Linus Torvalds 已提交
1756

1757
			lru_pages += zone_reclaimable_pages(zone);
1758
		}
L
Linus Torvalds 已提交
1759 1760 1761
	}

	for (priority = DEF_PRIORITY; priority >= 0; priority--) {
1762
		sc->nr_scanned = 0;
1763 1764
		if (!priority)
			disable_swap_token();
1765
		shrink_zones(priority, zonelist, sc);
1766 1767 1768 1769
		/*
		 * Don't shrink slabs when reclaiming memory from
		 * over limit cgroups
		 */
1770
		if (scanning_global_lru(sc)) {
1771
			shrink_slab(sc->nr_scanned, sc->gfp_mask, lru_pages);
1772
			if (reclaim_state) {
1773
				sc->nr_reclaimed += reclaim_state->reclaimed_slab;
1774 1775
				reclaim_state->reclaimed_slab = 0;
			}
L
Linus Torvalds 已提交
1776
		}
1777
		total_scanned += sc->nr_scanned;
1778
		if (sc->nr_reclaimed >= sc->nr_to_reclaim) {
1779
			ret = sc->nr_reclaimed;
L
Linus Torvalds 已提交
1780 1781 1782 1783 1784 1785 1786 1787 1788 1789
			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.
		 */
1790 1791
		writeback_threshold = sc->nr_to_reclaim + sc->nr_to_reclaim / 2;
		if (total_scanned > writeback_threshold) {
1792
			wakeup_flusher_threads(laptop_mode ? 0 : total_scanned);
1793
			sc->may_writepage = 1;
L
Linus Torvalds 已提交
1794 1795 1796
		}

		/* Take a nap, wait for some writeback to complete */
1797 1798
		if (!sc->hibernation_mode && sc->nr_scanned &&
		    priority < DEF_PRIORITY - 2)
1799
			congestion_wait(BLK_RW_ASYNC, HZ/10);
L
Linus Torvalds 已提交
1800
	}
1801
	/* top priority shrink_zones still had more to do? don't OOM, then */
1802
	if (!sc->all_unreclaimable && scanning_global_lru(sc))
1803
		ret = sc->nr_reclaimed;
L
Linus Torvalds 已提交
1804
out:
1805 1806 1807 1808 1809 1810 1811 1812 1813
	/*
	 * Now that we've scanned all the zones at this priority level, note
	 * that level within the zone so that the next thread which performs
	 * scanning of this zone will immediately start out at this priority
	 * level.  This affects only the decision whether or not to bring
	 * mapped pages onto the inactive list.
	 */
	if (priority < 0)
		priority = 0;
L
Linus Torvalds 已提交
1814

1815
	if (scanning_global_lru(sc)) {
1816
		for_each_zone_zonelist(zone, z, zonelist, high_zoneidx) {
1817 1818 1819 1820 1821 1822 1823 1824

			if (!cpuset_zone_allowed_hardwall(zone, GFP_KERNEL))
				continue;

			zone->prev_priority = priority;
		}
	} else
		mem_cgroup_record_reclaim_priority(sc->mem_cgroup, priority);
L
Linus Torvalds 已提交
1825

1826 1827
	delayacct_freepages_end();

L
Linus Torvalds 已提交
1828 1829 1830
	return ret;
}

1831
unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
1832
				gfp_t gfp_mask, nodemask_t *nodemask)
1833 1834 1835 1836
{
	struct scan_control sc = {
		.gfp_mask = gfp_mask,
		.may_writepage = !laptop_mode,
1837
		.nr_to_reclaim = SWAP_CLUSTER_MAX,
1838
		.may_unmap = 1,
1839
		.may_swap = 1,
1840 1841 1842 1843
		.swappiness = vm_swappiness,
		.order = order,
		.mem_cgroup = NULL,
		.isolate_pages = isolate_pages_global,
1844
		.nodemask = nodemask,
1845 1846
	};

1847
	return do_try_to_free_pages(zonelist, &sc);
1848 1849
}

1850
#ifdef CONFIG_CGROUP_MEM_RES_CTLR
1851

1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883
unsigned long mem_cgroup_shrink_node_zone(struct mem_cgroup *mem,
						gfp_t gfp_mask, bool noswap,
						unsigned int swappiness,
						struct zone *zone, int nid)
{
	struct scan_control sc = {
		.may_writepage = !laptop_mode,
		.may_unmap = 1,
		.may_swap = !noswap,
		.swappiness = swappiness,
		.order = 0,
		.mem_cgroup = mem,
		.isolate_pages = mem_cgroup_isolate_pages,
	};
	nodemask_t nm  = nodemask_of_node(nid);

	sc.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
			(GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK);
	sc.nodemask = &nm;
	sc.nr_reclaimed = 0;
	sc.nr_scanned = 0;
	/*
	 * 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);
	return sc.nr_reclaimed;
}

1884
unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *mem_cont,
K
KOSAKI Motohiro 已提交
1885 1886 1887
					   gfp_t gfp_mask,
					   bool noswap,
					   unsigned int swappiness)
1888
{
1889
	struct zonelist *zonelist;
1890 1891
	struct scan_control sc = {
		.may_writepage = !laptop_mode,
1892
		.may_unmap = 1,
1893
		.may_swap = !noswap,
1894
		.nr_to_reclaim = SWAP_CLUSTER_MAX,
K
KOSAKI Motohiro 已提交
1895
		.swappiness = swappiness,
1896 1897 1898
		.order = 0,
		.mem_cgroup = mem_cont,
		.isolate_pages = mem_cgroup_isolate_pages,
1899
		.nodemask = NULL, /* we don't care the placement */
1900 1901
	};

1902 1903 1904 1905
	sc.gfp_mask = (gfp_mask & GFP_RECLAIM_MASK) |
			(GFP_HIGHUSER_MOVABLE & ~GFP_RECLAIM_MASK);
	zonelist = NODE_DATA(numa_node_id())->node_zonelists;
	return do_try_to_free_pages(zonelist, &sc);
1906 1907 1908
}
#endif

1909
/* is kswapd sleeping prematurely? */
1910
static int sleeping_prematurely(pg_data_t *pgdat, int order, long remaining)
1911
{
1912
	int i;
1913 1914 1915 1916 1917 1918

	/* If a direct reclaimer woke kswapd within HZ/10, it's premature */
	if (remaining)
		return 1;

	/* If after HZ/10, a zone is below the high mark, it's premature */
1919 1920 1921 1922 1923 1924
	for (i = 0; i < pgdat->nr_zones; i++) {
		struct zone *zone = pgdat->node_zones + i;

		if (!populated_zone(zone))
			continue;

1925 1926 1927
		if (zone_is_all_unreclaimable(zone))
			continue;

1928 1929 1930
		if (!zone_watermark_ok(zone, order, high_wmark_pages(zone),
								0, 0))
			return 1;
1931
	}
1932 1933 1934 1935

	return 0;
}

L
Linus Torvalds 已提交
1936 1937
/*
 * For kswapd, balance_pgdat() will work across all this node's zones until
1938
 * they are all at high_wmark_pages(zone).
L
Linus Torvalds 已提交
1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950
 *
 * Returns the number of pages which were actually freed.
 *
 * 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
1951 1952 1953 1954 1955
 * 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
1956
 */
1957
static unsigned long balance_pgdat(pg_data_t *pgdat, int order)
L
Linus Torvalds 已提交
1958 1959 1960 1961
{
	int all_zones_ok;
	int priority;
	int i;
1962
	unsigned long total_scanned;
L
Linus Torvalds 已提交
1963
	struct reclaim_state *reclaim_state = current->reclaim_state;
1964 1965
	struct scan_control sc = {
		.gfp_mask = GFP_KERNEL,
1966
		.may_unmap = 1,
1967
		.may_swap = 1,
1968 1969 1970 1971 1972
		/*
		 * 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,
1973
		.swappiness = vm_swappiness,
A
Andy Whitcroft 已提交
1974
		.order = order,
1975 1976
		.mem_cgroup = NULL,
		.isolate_pages = isolate_pages_global,
1977
	};
1978 1979
	/*
	 * temp_priority is used to remember the scanning priority at which
1980 1981
	 * this zone was successfully refilled to
	 * free_pages == high_wmark_pages(zone).
1982 1983
	 */
	int temp_priority[MAX_NR_ZONES];
L
Linus Torvalds 已提交
1984 1985 1986

loop_again:
	total_scanned = 0;
1987
	sc.nr_reclaimed = 0;
C
Christoph Lameter 已提交
1988
	sc.may_writepage = !laptop_mode;
1989
	count_vm_event(PAGEOUTRUN);
L
Linus Torvalds 已提交
1990

1991 1992
	for (i = 0; i < pgdat->nr_zones; i++)
		temp_priority[i] = DEF_PRIORITY;
L
Linus Torvalds 已提交
1993 1994 1995 1996

	for (priority = DEF_PRIORITY; priority >= 0; priority--) {
		int end_zone = 0;	/* Inclusive.  0 = ZONE_DMA */
		unsigned long lru_pages = 0;
1997
		int has_under_min_watermark_zone = 0;
L
Linus Torvalds 已提交
1998

1999 2000 2001 2002
		/* The swap token gets in the way of swapout... */
		if (!priority)
			disable_swap_token();

L
Linus Torvalds 已提交
2003 2004
		all_zones_ok = 1;

2005 2006 2007 2008 2009 2010
		/*
		 * 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
2011

2012 2013
			if (!populated_zone(zone))
				continue;
L
Linus Torvalds 已提交
2014

2015 2016
			if (zone_is_all_unreclaimable(zone) &&
			    priority != DEF_PRIORITY)
2017
				continue;
L
Linus Torvalds 已提交
2018

2019 2020 2021 2022
			/*
			 * Do some background aging of the anon list, to give
			 * pages a chance to be referenced before reclaiming.
			 */
2023
			if (inactive_anon_is_low(zone, &sc))
2024 2025 2026
				shrink_active_list(SWAP_CLUSTER_MAX, zone,
							&sc, priority, 0);

2027 2028
			if (!zone_watermark_ok(zone, order,
					high_wmark_pages(zone), 0, 0)) {
2029
				end_zone = i;
A
Andrew Morton 已提交
2030
				break;
L
Linus Torvalds 已提交
2031 2032
			}
		}
A
Andrew Morton 已提交
2033 2034 2035
		if (i < 0)
			goto out;

L
Linus Torvalds 已提交
2036 2037 2038
		for (i = 0; i <= end_zone; i++) {
			struct zone *zone = pgdat->node_zones + i;

2039
			lru_pages += zone_reclaimable_pages(zone);
L
Linus Torvalds 已提交
2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052
		}

		/*
		 * 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;
2053
			int nr_slab;
2054
			int nid, zid;
L
Linus Torvalds 已提交
2055

2056
			if (!populated_zone(zone))
L
Linus Torvalds 已提交
2057 2058
				continue;

2059 2060
			if (zone_is_all_unreclaimable(zone) &&
					priority != DEF_PRIORITY)
L
Linus Torvalds 已提交
2061 2062
				continue;

2063 2064
			if (!zone_watermark_ok(zone, order,
					high_wmark_pages(zone), end_zone, 0))
2065
				all_zones_ok = 0;
2066
			temp_priority[i] = priority;
L
Linus Torvalds 已提交
2067
			sc.nr_scanned = 0;
2068
			note_zone_scanning_priority(zone, priority);
2069 2070 2071 2072 2073 2074 2075 2076 2077

			nid = pgdat->node_id;
			zid = zone_idx(zone);
			/*
			 * Call soft limit reclaim before calling shrink_zone.
			 * For now we ignore the return value
			 */
			mem_cgroup_soft_limit_reclaim(zone, order, sc.gfp_mask,
							nid, zid);
2078 2079 2080 2081
			/*
			 * We put equal pressure on every zone, unless one
			 * zone has way too many pages free already.
			 */
2082 2083
			if (!zone_watermark_ok(zone, order,
					8*high_wmark_pages(zone), end_zone, 0))
2084
				shrink_zone(priority, zone, &sc);
L
Linus Torvalds 已提交
2085
			reclaim_state->reclaimed_slab = 0;
2086 2087
			nr_slab = shrink_slab(sc.nr_scanned, GFP_KERNEL,
						lru_pages);
2088
			sc.nr_reclaimed += reclaim_state->reclaimed_slab;
L
Linus Torvalds 已提交
2089
			total_scanned += sc.nr_scanned;
2090
			if (zone_is_all_unreclaimable(zone))
L
Linus Torvalds 已提交
2091
				continue;
2092
			if (nr_slab == 0 && zone->pages_scanned >=
2093
					(zone_reclaimable_pages(zone) * 6))
2094 2095
					zone_set_flag(zone,
						      ZONE_ALL_UNRECLAIMABLE);
L
Linus Torvalds 已提交
2096 2097 2098 2099 2100 2101
			/*
			 * 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 &&
2102
			    total_scanned > sc.nr_reclaimed + sc.nr_reclaimed / 2)
L
Linus Torvalds 已提交
2103
				sc.may_writepage = 1;
2104 2105 2106 2107 2108 2109 2110 2111 2112

			/*
			 * We are still under min water mark. it mean we have
			 * GFP_ATOMIC allocation failure risk. Hurry up!
			 */
			if (!zone_watermark_ok(zone, order, min_wmark_pages(zone),
					      end_zone, 0))
				has_under_min_watermark_zone = 1;

L
Linus Torvalds 已提交
2113 2114 2115 2116 2117 2118 2119
		}
		if (all_zones_ok)
			break;		/* kswapd: all done */
		/*
		 * OK, kswapd is getting into trouble.  Take a nap, then take
		 * another pass across the zones.
		 */
2120 2121 2122 2123 2124 2125
		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
2126 2127 2128 2129 2130 2131 2132

		/*
		 * 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.
		 */
2133
		if (sc.nr_reclaimed >= SWAP_CLUSTER_MAX)
L
Linus Torvalds 已提交
2134 2135 2136
			break;
	}
out:
2137 2138 2139 2140 2141
	/*
	 * Note within each zone the priority level at which this zone was
	 * brought into a happy state.  So that the next thread which scans this
	 * zone will start out at that priority level.
	 */
L
Linus Torvalds 已提交
2142 2143 2144
	for (i = 0; i < pgdat->nr_zones; i++) {
		struct zone *zone = pgdat->node_zones + i;

2145
		zone->prev_priority = temp_priority[i];
L
Linus Torvalds 已提交
2146 2147 2148
	}
	if (!all_zones_ok) {
		cond_resched();
2149 2150 2151

		try_to_freeze();

2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168
		/*
		 * 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
2169 2170 2171
		goto loop_again;
	}

2172
	return sc.nr_reclaimed;
L
Linus Torvalds 已提交
2173 2174 2175 2176
}

/*
 * The background pageout daemon, started as a kernel thread
2177
 * from the init process.
L
Linus Torvalds 已提交
2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196
 *
 * 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)
{
	unsigned long order;
	pg_data_t *pgdat = (pg_data_t*)p;
	struct task_struct *tsk = current;
	DEFINE_WAIT(wait);
	struct reclaim_state reclaim_state = {
		.reclaimed_slab = 0,
	};
2197
	const struct cpumask *cpumask = cpumask_of_node(pgdat->node_id);
L
Linus Torvalds 已提交
2198

2199 2200
	lockdep_set_current_reclaim_state(GFP_KERNEL);

R
Rusty Russell 已提交
2201
	if (!cpumask_empty(cpumask))
2202
		set_cpus_allowed_ptr(tsk, cpumask);
L
Linus Torvalds 已提交
2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216
	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).
	 */
2217
	tsk->flags |= PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD;
2218
	set_freezable();
L
Linus Torvalds 已提交
2219 2220 2221 2222

	order = 0;
	for ( ; ; ) {
		unsigned long new_order;
2223
		int ret;
2224

L
Linus Torvalds 已提交
2225 2226 2227 2228 2229 2230 2231 2232 2233 2234
		prepare_to_wait(&pgdat->kswapd_wait, &wait, TASK_INTERRUPTIBLE);
		new_order = pgdat->kswapd_max_order;
		pgdat->kswapd_max_order = 0;
		if (order < new_order) {
			/*
			 * Don't sleep if someone wants a larger 'order'
			 * allocation
			 */
			order = new_order;
		} else {
2235 2236 2237 2238
			if (!freezing(current) && !kthread_should_stop()) {
				long remaining = 0;

				/* Try to sleep for a short interval */
2239
				if (!sleeping_prematurely(pgdat, order, remaining)) {
2240 2241 2242 2243 2244 2245 2246 2247 2248 2249
					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
				 */
2250
				if (!sleeping_prematurely(pgdat, order, remaining))
2251 2252 2253
					schedule();
				else {
					if (remaining)
2254
						count_vm_event(KSWAPD_LOW_WMARK_HIT_QUICKLY);
2255
					else
2256
						count_vm_event(KSWAPD_HIGH_WMARK_HIT_QUICKLY);
2257 2258
				}
			}
2259

L
Linus Torvalds 已提交
2260 2261 2262 2263
			order = pgdat->kswapd_max_order;
		}
		finish_wait(&pgdat->kswapd_wait, &wait);

2264 2265 2266 2267 2268 2269 2270 2271 2272
		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
		 */
		if (!ret)
2273
			balance_pgdat(pgdat, order);
L
Linus Torvalds 已提交
2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284
	}
	return 0;
}

/*
 * A zone is low on free memory, so wake its kswapd task to service it.
 */
void wakeup_kswapd(struct zone *zone, int order)
{
	pg_data_t *pgdat;

2285
	if (!populated_zone(zone))
L
Linus Torvalds 已提交
2286 2287 2288
		return;

	pgdat = zone->zone_pgdat;
2289
	if (zone_watermark_ok(zone, order, low_wmark_pages(zone), 0, 0))
L
Linus Torvalds 已提交
2290 2291 2292
		return;
	if (pgdat->kswapd_max_order < order)
		pgdat->kswapd_max_order = order;
2293
	if (!cpuset_zone_allowed_hardwall(zone, GFP_KERNEL))
L
Linus Torvalds 已提交
2294
		return;
2295
	if (!waitqueue_active(&pgdat->kswapd_wait))
L
Linus Torvalds 已提交
2296
		return;
2297
	wake_up_interruptible(&pgdat->kswapd_wait);
L
Linus Torvalds 已提交
2298 2299
}

2300 2301 2302 2303 2304 2305 2306 2307
/*
 * 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)
2308
{
2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332
	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;
2333 2334
}

2335
#ifdef CONFIG_HIBERNATION
L
Linus Torvalds 已提交
2336
/*
2337
 * Try to free `nr_to_reclaim' of memory, system-wide, and return the number of
2338 2339 2340 2341 2342
 * 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
2343
 */
2344
unsigned long shrink_all_memory(unsigned long nr_to_reclaim)
L
Linus Torvalds 已提交
2345
{
2346 2347
	struct reclaim_state reclaim_state;
	struct scan_control sc = {
2348 2349 2350
		.gfp_mask = GFP_HIGHUSER_MOVABLE,
		.may_swap = 1,
		.may_unmap = 1,
2351
		.may_writepage = 1,
2352 2353 2354 2355
		.nr_to_reclaim = nr_to_reclaim,
		.hibernation_mode = 1,
		.swappiness = vm_swappiness,
		.order = 0,
2356
		.isolate_pages = isolate_pages_global,
L
Linus Torvalds 已提交
2357
	};
2358 2359 2360
	struct zonelist * zonelist = node_zonelist(numa_node_id(), sc.gfp_mask);
	struct task_struct *p = current;
	unsigned long nr_reclaimed;
L
Linus Torvalds 已提交
2361

2362 2363 2364 2365
	p->flags |= PF_MEMALLOC;
	lockdep_set_current_reclaim_state(sc.gfp_mask);
	reclaim_state.reclaimed_slab = 0;
	p->reclaim_state = &reclaim_state;
2366

2367
	nr_reclaimed = do_try_to_free_pages(zonelist, &sc);
2368

2369 2370 2371
	p->reclaim_state = NULL;
	lockdep_clear_current_reclaim_state();
	p->flags &= ~PF_MEMALLOC;
2372

2373
	return nr_reclaimed;
L
Linus Torvalds 已提交
2374
}
2375
#endif /* CONFIG_HIBERNATION */
L
Linus Torvalds 已提交
2376 2377 2378 2379 2380

/* 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. */
2381
static int __devinit cpu_callback(struct notifier_block *nfb,
2382
				  unsigned long action, void *hcpu)
L
Linus Torvalds 已提交
2383
{
2384
	int nid;
L
Linus Torvalds 已提交
2385

2386
	if (action == CPU_ONLINE || action == CPU_ONLINE_FROZEN) {
2387
		for_each_node_state(nid, N_HIGH_MEMORY) {
2388
			pg_data_t *pgdat = NODE_DATA(nid);
2389 2390 2391
			const struct cpumask *mask;

			mask = cpumask_of_node(pgdat->node_id);
2392

2393
			if (cpumask_any_and(cpu_online_mask, mask) < nr_cpu_ids)
L
Linus Torvalds 已提交
2394
				/* One of our CPUs online: restore mask */
2395
				set_cpus_allowed_ptr(pgdat->kswapd, mask);
L
Linus Torvalds 已提交
2396 2397 2398 2399 2400
		}
	}
	return NOTIFY_OK;
}

2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422
/*
 * 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;
}

2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433
/*
 * 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
2434 2435
static int __init kswapd_init(void)
{
2436
	int nid;
2437

L
Linus Torvalds 已提交
2438
	swap_setup();
2439
	for_each_node_state(nid, N_HIGH_MEMORY)
2440
 		kswapd_run(nid);
L
Linus Torvalds 已提交
2441 2442 2443 2444 2445
	hotcpu_notifier(cpu_callback, 0);
	return 0;
}

module_init(kswapd_init)
2446 2447 2448 2449 2450 2451 2452 2453 2454 2455

#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;

2456
#define RECLAIM_OFF 0
2457
#define RECLAIM_ZONE (1<<0)	/* Run shrink_inactive_list on the zone */
2458 2459 2460
#define RECLAIM_WRITE (1<<1)	/* Writeout pages during reclaim */
#define RECLAIM_SWAP (1<<2)	/* Swap pages out during reclaim */

2461 2462 2463 2464 2465 2466 2467
/*
 * 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

2468 2469 2470 2471 2472 2473
/*
 * Percentage of pages in a zone that must be unmapped for zone_reclaim to
 * occur.
 */
int sysctl_min_unmapped_ratio = 1;

2474 2475 2476 2477 2478 2479
/*
 * 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;

2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521
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;
}

2522 2523 2524
/*
 * Try to free up some pages from this zone through reclaim.
 */
2525
static int __zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
2526
{
2527
	/* Minimum pages needed in order to stay on node */
2528
	const unsigned long nr_pages = 1 << order;
2529 2530
	struct task_struct *p = current;
	struct reclaim_state reclaim_state;
2531
	int priority;
2532 2533
	struct scan_control sc = {
		.may_writepage = !!(zone_reclaim_mode & RECLAIM_WRITE),
2534
		.may_unmap = !!(zone_reclaim_mode & RECLAIM_SWAP),
2535
		.may_swap = 1,
2536 2537
		.nr_to_reclaim = max_t(unsigned long, nr_pages,
				       SWAP_CLUSTER_MAX),
2538
		.gfp_mask = gfp_mask,
2539
		.swappiness = vm_swappiness,
2540
		.order = order,
2541
		.isolate_pages = isolate_pages_global,
2542
	};
2543
	unsigned long slab_reclaimable;
2544 2545 2546

	disable_swap_token();
	cond_resched();
2547 2548 2549 2550 2551 2552
	/*
	 * 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;
2553 2554
	reclaim_state.reclaimed_slab = 0;
	p->reclaim_state = &reclaim_state;
2555

2556
	if (zone_pagecache_reclaimable(zone) > zone->min_unmapped_pages) {
2557 2558 2559 2560 2561 2562
		/*
		 * Free memory by calling shrink zone with increasing
		 * priorities until we have enough memory freed.
		 */
		priority = ZONE_RECLAIM_PRIORITY;
		do {
2563
			note_zone_scanning_priority(zone, priority);
2564
			shrink_zone(priority, zone, &sc);
2565
			priority--;
2566
		} while (priority >= 0 && sc.nr_reclaimed < nr_pages);
2567
	}
2568

2569 2570
	slab_reclaimable = zone_page_state(zone, NR_SLAB_RECLAIMABLE);
	if (slab_reclaimable > zone->min_slab_pages) {
2571
		/*
2572
		 * shrink_slab() does not currently allow us to determine how
2573 2574 2575 2576
		 * 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.
2577
		 *
2578 2579
		 * Note that shrink_slab will free memory on all zones and may
		 * take a long time.
2580
		 */
2581
		while (shrink_slab(sc.nr_scanned, gfp_mask, order) &&
2582 2583
			zone_page_state(zone, NR_SLAB_RECLAIMABLE) >
				slab_reclaimable - nr_pages)
2584
			;
2585 2586 2587 2588 2589

		/*
		 * Update nr_reclaimed by the number of slab pages we
		 * reclaimed from this zone.
		 */
2590
		sc.nr_reclaimed += slab_reclaimable -
2591
			zone_page_state(zone, NR_SLAB_RECLAIMABLE);
2592 2593
	}

2594
	p->reclaim_state = NULL;
2595
	current->flags &= ~(PF_MEMALLOC | PF_SWAPWRITE);
2596
	return sc.nr_reclaimed >= nr_pages;
2597
}
2598 2599 2600 2601

int zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
{
	int node_id;
2602
	int ret;
2603 2604

	/*
2605 2606
	 * Zone reclaim reclaims unmapped file backed pages and
	 * slab pages if we are over the defined limits.
2607
	 *
2608 2609 2610 2611 2612
	 * 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.
2613
	 */
2614 2615
	if (zone_pagecache_reclaimable(zone) <= zone->min_unmapped_pages &&
	    zone_page_state(zone, NR_SLAB_RECLAIMABLE) <= zone->min_slab_pages)
2616
		return ZONE_RECLAIM_FULL;
2617

2618
	if (zone_is_all_unreclaimable(zone))
2619
		return ZONE_RECLAIM_FULL;
2620

2621
	/*
2622
	 * Do not scan if the allocation should not be delayed.
2623
	 */
2624
	if (!(gfp_mask & __GFP_WAIT) || (current->flags & PF_MEMALLOC))
2625
		return ZONE_RECLAIM_NOSCAN;
2626 2627 2628 2629 2630 2631 2632

	/*
	 * 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.
	 */
2633
	node_id = zone_to_nid(zone);
2634
	if (node_state(node_id, N_CPU) && node_id != numa_node_id())
2635
		return ZONE_RECLAIM_NOSCAN;
2636 2637

	if (zone_test_and_set_flag(zone, ZONE_RECLAIM_LOCKED))
2638 2639
		return ZONE_RECLAIM_NOSCAN;

2640 2641 2642
	ret = __zone_reclaim(zone, gfp_mask, order);
	zone_clear_flag(zone, ZONE_RECLAIM_LOCKED);

2643 2644 2645
	if (!ret)
		count_vm_event(PGSCAN_ZONE_RECLAIM_FAILED);

2646
	return ret;
2647
}
2648
#endif
L
Lee Schermerhorn 已提交
2649 2650 2651 2652 2653 2654 2655

/*
 * 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
2656 2657
 * 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
2658 2659
 *
 * Reasons page might not be evictable:
2660
 * (1) page's mapping marked unevictable
N
Nick Piggin 已提交
2661
 * (2) page is part of an mlocked VMA
2662
 *
L
Lee Schermerhorn 已提交
2663 2664 2665 2666
 */
int page_evictable(struct page *page, struct vm_area_struct *vma)
{

2667 2668 2669
	if (mapping_unevictable(page_mapping(page)))
		return 0;

N
Nick Piggin 已提交
2670 2671
	if (PageMlocked(page) || (vma && is_mlocked_vma(vma, page)))
		return 0;
L
Lee Schermerhorn 已提交
2672 2673 2674

	return 1;
}
2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693

/**
 * 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)) {
2694
		enum lru_list l = page_lru_base_type(page);
2695

2696 2697
		__dec_zone_state(zone, NR_UNEVICTABLE);
		list_move(&page->lru, &zone->lru[l].list);
K
KAMEZAWA Hiroyuki 已提交
2698
		mem_cgroup_move_lists(page, LRU_UNEVICTABLE, l);
2699 2700 2701 2702 2703 2704 2705 2706
		__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
2707
		mem_cgroup_rotate_lru_list(page, LRU_UNEVICTABLE);
2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766
		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);
	}

}
2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778

/**
 * 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 */
2779
static void scan_zone_unevictable_pages(struct zone *zone)
2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820
{
	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.
 */
2821
static void scan_all_zones_unevictable_pages(void)
2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836
{
	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,
2837
			   void __user *buffer,
2838 2839
			   size_t *length, loff_t *ppos)
{
2840
	proc_doulongvec_minmax(table, write, buffer, length, ppos);
2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 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

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

	scan_unevictable_pages = 0;
	return 0;
}

/*
 * 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);
}