vmscan.c 79.0 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;

L
Linus Torvalds 已提交
58
	/* This context's GFP mask */
A
Al Viro 已提交
59
	gfp_t gfp_mask;
L
Linus Torvalds 已提交
60 61 62

	int may_writepage;

63 64
	/* Can mapped pages be reclaimed? */
	int may_unmap;
65

66 67 68
	/* Can pages be swapped as part of reclaim? */
	int may_swap;

L
Linus Torvalds 已提交
69 70 71 72 73
	/* This context's SWAP_CLUSTER_MAX. If freeing memory for
	 * suspend, we effectively ignore SWAP_CLUSTER_MAX.
	 * In this context, it doesn't matter that we scan the
	 * whole list at once. */
	int swap_cluster_max;
74 75

	int swappiness;
76 77

	int all_unreclaimable;
A
Andy Whitcroft 已提交
78 79

	int order;
80 81 82 83

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

84 85 86 87 88 89
	/*
	 * Nodemask of nodes allowed by the caller. If NULL, all nodes
	 * are scanned.
	 */
	nodemask_t	*nodemask;

90 91 92 93
	/* 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,
94
			int active, int file);
L
Linus Torvalds 已提交
95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130
};

#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;
131
long vm_total_pages;	/* The total number of pages which the VM controls */
L
Linus Torvalds 已提交
132 133 134 135

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

136
#ifdef CONFIG_CGROUP_MEM_RES_CTLR
137
#define scanning_global_lru(sc)	(!(sc)->mem_cgroup)
138
#else
139
#define scanning_global_lru(sc)	(1)
140 141
#endif

142 143 144
static struct zone_reclaim_stat *get_reclaim_stat(struct zone *zone,
						  struct scan_control *sc)
{
145
	if (!scanning_global_lru(sc))
K
KOSAKI Motohiro 已提交
146 147
		return mem_cgroup_get_reclaim_stat(sc->mem_cgroup, zone);

148 149 150
	return &zone->reclaim_stat;
}

151 152 153
static unsigned long zone_nr_pages(struct zone *zone, struct scan_control *sc,
				   enum lru_list lru)
{
154
	if (!scanning_global_lru(sc))
155 156
		return mem_cgroup_zone_nr_pages(sc->mem_cgroup, zone, lru);

157 158 159 160
	return zone_page_state(zone, NR_LRU_BASE + lru);
}


L
Linus Torvalds 已提交
161 162 163
/*
 * Add a shrinker callback to be called from the vm
 */
164
void register_shrinker(struct shrinker *shrinker)
L
Linus Torvalds 已提交
165
{
166 167 168 169
	shrinker->nr = 0;
	down_write(&shrinker_rwsem);
	list_add_tail(&shrinker->list, &shrinker_list);
	up_write(&shrinker_rwsem);
L
Linus Torvalds 已提交
170
}
171
EXPORT_SYMBOL(register_shrinker);
L
Linus Torvalds 已提交
172 173 174 175

/*
 * Remove one
 */
176
void unregister_shrinker(struct shrinker *shrinker)
L
Linus Torvalds 已提交
177 178 179 180 181
{
	down_write(&shrinker_rwsem);
	list_del(&shrinker->list);
	up_write(&shrinker_rwsem);
}
182
EXPORT_SYMBOL(unregister_shrinker);
L
Linus Torvalds 已提交
183 184 185 186 187 188 189 190 191 192

#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
193
 * If the vm encountered mapped pages on the LRU it increase the pressure on
L
Linus Torvalds 已提交
194 195 196 197 198 199 200
 * 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.
201 202
 *
 * Returns the number of slab objects which we shrunk.
L
Linus Torvalds 已提交
203
 */
204 205
unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
			unsigned long lru_pages)
L
Linus Torvalds 已提交
206 207
{
	struct shrinker *shrinker;
208
	unsigned long ret = 0;
L
Linus Torvalds 已提交
209 210 211 212 213

	if (scanned == 0)
		scanned = SWAP_CLUSTER_MAX;

	if (!down_read_trylock(&shrinker_rwsem))
214
		return 1;	/* Assume we'll be able to shrink next time */
L
Linus Torvalds 已提交
215 216 217 218

	list_for_each_entry(shrinker, &shrinker_list, list) {
		unsigned long long delta;
		unsigned long total_scan;
219
		unsigned long max_pass = (*shrinker->shrink)(0, gfp_mask);
L
Linus Torvalds 已提交
220 221

		delta = (4 * scanned) / shrinker->seeks;
222
		delta *= max_pass;
L
Linus Torvalds 已提交
223 224
		do_div(delta, lru_pages + 1);
		shrinker->nr += delta;
225
		if (shrinker->nr < 0) {
226 227 228
			printk(KERN_ERR "shrink_slab: %pF negative objects to "
			       "delete nr=%ld\n",
			       shrinker->shrink, shrinker->nr);
229 230 231 232 233 234 235 236 237 238
			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
239 240 241 242 243 244 245

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

		while (total_scan >= SHRINK_BATCH) {
			long this_scan = SHRINK_BATCH;
			int shrink_ret;
246
			int nr_before;
L
Linus Torvalds 已提交
247

248 249
			nr_before = (*shrinker->shrink)(0, gfp_mask);
			shrink_ret = (*shrinker->shrink)(this_scan, gfp_mask);
L
Linus Torvalds 已提交
250 251
			if (shrink_ret == -1)
				break;
252 253
			if (shrink_ret < nr_before)
				ret += nr_before - shrink_ret;
254
			count_vm_events(SLABS_SCANNED, this_scan);
L
Linus Torvalds 已提交
255 256 257 258 259 260 261 262
			total_scan -= this_scan;

			cond_resched();
		}

		shrinker->nr += total_scan;
	}
	up_read(&shrinker_rwsem);
263
	return ret;
L
Linus Torvalds 已提交
264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288
}

/* 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)
{
289
	return page_count(page) - !!page_has_private(page) == 2;
L
Linus Torvalds 已提交
290 291 292 293
}

static int may_write_to_queue(struct backing_dev_info *bdi)
{
294
	if (current->flags & PF_SWAPWRITE)
L
Linus Torvalds 已提交
295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318
		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);
319 320
	if (page_mapping(page) == mapping)
		mapping_set_error(mapping, error);
L
Linus Torvalds 已提交
321 322 323
	unlock_page(page);
}

324 325 326 327 328 329
/* Request for sync pageout. */
enum pageout_io {
	PAGEOUT_IO_ASYNC,
	PAGEOUT_IO_SYNC,
};

330 331 332 333 334 335 336 337 338 339 340 341
/* 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
342
/*
A
Andrew Morton 已提交
343 344
 * pageout is called by shrink_page_list() for each dirty page.
 * Calls ->writepage().
L
Linus Torvalds 已提交
345
 */
346 347
static pageout_t pageout(struct page *page, struct address_space *mapping,
						enum pageout_io sync_writeback)
L
Linus Torvalds 已提交
348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372
{
	/*
	 * 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.
	 *
	 * If this process is currently in generic_file_write() against
	 * 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.
	 * See swapfile.c:page_queue_congested().
	 */
	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.
		 */
373
		if (page_has_private(page)) {
L
Linus Torvalds 已提交
374 375
			if (try_to_free_buffers(page)) {
				ClearPageDirty(page);
376
				printk("%s: orphaned page\n", __func__);
L
Linus Torvalds 已提交
377 378 379 380 381 382 383 384 385 386 387 388 389 390 391
				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,
392 393
			.range_start = 0,
			.range_end = LLONG_MAX,
L
Linus Torvalds 已提交
394 395 396 397 398 399 400 401
			.nonblocking = 1,
			.for_reclaim = 1,
		};

		SetPageReclaim(page);
		res = mapping->a_ops->writepage(page, &wbc);
		if (res < 0)
			handle_write_error(mapping, page, res);
402
		if (res == AOP_WRITEPAGE_ACTIVATE) {
L
Linus Torvalds 已提交
403 404 405
			ClearPageReclaim(page);
			return PAGE_ACTIVATE;
		}
406 407 408 409 410 411 412 413 414

		/*
		 * 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
415 416 417 418
		if (!PageWriteback(page)) {
			/* synchronous write or broken a_ops? */
			ClearPageReclaim(page);
		}
419
		inc_zone_page_state(page, NR_VMSCAN_WRITE);
L
Linus Torvalds 已提交
420 421 422 423 424 425
		return PAGE_SUCCESS;
	}

	return PAGE_CLEAN;
}

426
/*
N
Nick Piggin 已提交
427 428
 * Same as remove_mapping, but if the page is removed from the mapping, it
 * gets returned with a refcount of 0.
429
 */
N
Nick Piggin 已提交
430
static int __remove_mapping(struct address_space *mapping, struct page *page)
431
{
432 433
	BUG_ON(!PageLocked(page));
	BUG_ON(mapping != page_mapping(page));
434

N
Nick Piggin 已提交
435
	spin_lock_irq(&mapping->tree_lock);
436
	/*
N
Nick Piggin 已提交
437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459
	 * 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.
460
	 */
N
Nick Piggin 已提交
461
	if (!page_freeze_refs(page, 2))
462
		goto cannot_free;
N
Nick Piggin 已提交
463 464 465
	/* note: atomic_cmpxchg in page_freeze_refs provides the smp_rmb */
	if (unlikely(PageDirty(page))) {
		page_unfreeze_refs(page, 2);
466
		goto cannot_free;
N
Nick Piggin 已提交
467
	}
468 469 470 471

	if (PageSwapCache(page)) {
		swp_entry_t swap = { .val = page_private(page) };
		__delete_from_swap_cache(page);
N
Nick Piggin 已提交
472
		spin_unlock_irq(&mapping->tree_lock);
473
		swapcache_free(swap, page);
N
Nick Piggin 已提交
474 475
	} else {
		__remove_from_page_cache(page);
N
Nick Piggin 已提交
476
		spin_unlock_irq(&mapping->tree_lock);
477
		mem_cgroup_uncharge_cache_page(page);
478 479 480 481 482
	}

	return 1;

cannot_free:
N
Nick Piggin 已提交
483
	spin_unlock_irq(&mapping->tree_lock);
484 485 486
	return 0;
}

N
Nick Piggin 已提交
487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506
/*
 * 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
507 508 509 510 511 512 513 514 515 516 517 518 519
/**
 * 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);
520
	int was_unevictable = PageUnevictable(page);
L
Lee Schermerhorn 已提交
521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560

	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.
		 */
		lru = active + page_is_file_cache(page);
		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);
	}

	/*
	 * 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.
		 */
	}

561 562 563 564 565
	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
566 567 568
	put_page(page);		/* drop ref from isolate */
}

L
Linus Torvalds 已提交
569
/*
A
Andrew Morton 已提交
570
 * shrink_page_list() returns the number of reclaimed pages
L
Linus Torvalds 已提交
571
 */
A
Andrew Morton 已提交
572
static unsigned long shrink_page_list(struct list_head *page_list,
573 574
					struct scan_control *sc,
					enum pageout_io sync_writeback)
L
Linus Torvalds 已提交
575 576 577 578
{
	LIST_HEAD(ret_pages);
	struct pagevec freed_pvec;
	int pgactivate = 0;
579
	unsigned long nr_reclaimed = 0;
580
	unsigned long vm_flags;
L
Linus Torvalds 已提交
581 582 583 584 585 586 587 588 589 590 591 592 593 594 595

	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
596
		if (!trylock_page(page))
L
Linus Torvalds 已提交
597 598
			goto keep;

N
Nick Piggin 已提交
599
		VM_BUG_ON(PageActive(page));
L
Linus Torvalds 已提交
600 601

		sc->nr_scanned++;
602

N
Nick Piggin 已提交
603 604
		if (unlikely(!page_evictable(page, NULL)))
			goto cull_mlocked;
L
Lee Schermerhorn 已提交
605

606
		if (!sc->may_unmap && page_mapped(page))
607 608
			goto keep_locked;

L
Linus Torvalds 已提交
609 610 611 612
		/* Double the slab pressure for mapped and swapcache pages */
		if (page_mapped(page) || PageSwapCache(page))
			sc->nr_scanned++;

613 614 615 616 617 618 619 620 621 622 623 624 625 626
		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);
627
			else
628 629
				goto keep_locked;
		}
L
Linus Torvalds 已提交
630

631 632
		referenced = page_referenced(page, 1,
						sc->mem_cgroup, &vm_flags);
633 634 635 636 637
		/*
		 * 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
638
		if (sc->order <= PAGE_ALLOC_COSTLY_ORDER &&
639 640
					referenced && page_mapping_inuse(page)
					&& !(vm_flags & VM_LOCKED))
L
Linus Torvalds 已提交
641 642 643 644 645 646
			goto activate_locked;

		/*
		 * Anonymous process memory has backing store?
		 * Try to allocate it some swap space here.
		 */
N
Nick Piggin 已提交
647
		if (PageAnon(page) && !PageSwapCache(page)) {
648 649
			if (!(sc->gfp_mask & __GFP_IO))
				goto keep_locked;
650
			if (!add_to_swap(page))
L
Linus Torvalds 已提交
651
				goto activate_locked;
652
			may_enter_fs = 1;
N
Nick Piggin 已提交
653
		}
L
Linus Torvalds 已提交
654 655 656 657 658 659 660 661

		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) {
662
			switch (try_to_unmap(page, 0)) {
L
Linus Torvalds 已提交
663 664 665 666
			case SWAP_FAIL:
				goto activate_locked;
			case SWAP_AGAIN:
				goto keep_locked;
N
Nick Piggin 已提交
667 668
			case SWAP_MLOCK:
				goto cull_mlocked;
L
Linus Torvalds 已提交
669 670 671 672 673 674
			case SWAP_SUCCESS:
				; /* try to free the page below */
			}
		}

		if (PageDirty(page)) {
A
Andy Whitcroft 已提交
675
			if (sc->order <= PAGE_ALLOC_COSTLY_ORDER && referenced)
L
Linus Torvalds 已提交
676
				goto keep_locked;
677
			if (!may_enter_fs)
L
Linus Torvalds 已提交
678
				goto keep_locked;
679
			if (!sc->may_writepage)
L
Linus Torvalds 已提交
680 681 682
				goto keep_locked;

			/* Page is dirty, try to write it out here */
683
			switch (pageout(page, mapping, sync_writeback)) {
L
Linus Torvalds 已提交
684 685 686 687 688
			case PAGE_KEEP:
				goto keep_locked;
			case PAGE_ACTIVATE:
				goto activate_locked;
			case PAGE_SUCCESS:
689
				if (PageWriteback(page) || PageDirty(page))
L
Linus Torvalds 已提交
690 691 692 693 694
					goto keep;
				/*
				 * A synchronous write - probably a ramdisk.  Go
				 * ahead and try to reclaim the page.
				 */
N
Nick Piggin 已提交
695
				if (!trylock_page(page))
L
Linus Torvalds 已提交
696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714
					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
715
		 * will do this, as well as the blockdev mapping.
L
Linus Torvalds 已提交
716 717 718 719 720 721 722 723 724 725
		 * 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.
		 */
726
		if (page_has_private(page)) {
L
Linus Torvalds 已提交
727 728
			if (!try_to_release_page(page, sc->gfp_mask))
				goto activate_locked;
N
Nick Piggin 已提交
729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744
			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
745 746
		}

N
Nick Piggin 已提交
747
		if (!mapping || !__remove_mapping(mapping, page))
748
			goto keep_locked;
L
Linus Torvalds 已提交
749

N
Nick Piggin 已提交
750 751 752 753 754 755 756 757
		/*
		 * 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
758
free_it:
759
		nr_reclaimed++;
N
Nick Piggin 已提交
760 761 762 763
		if (!pagevec_add(&freed_pvec, page)) {
			__pagevec_free(&freed_pvec);
			pagevec_reinit(&freed_pvec);
		}
L
Linus Torvalds 已提交
764 765
		continue;

N
Nick Piggin 已提交
766
cull_mlocked:
767 768
		if (PageSwapCache(page))
			try_to_free_swap(page);
N
Nick Piggin 已提交
769 770 771 772
		unlock_page(page);
		putback_lru_page(page);
		continue;

L
Linus Torvalds 已提交
773
activate_locked:
774 775
		/* Not a candidate for swapping, so reclaim swap space. */
		if (PageSwapCache(page) && vm_swap_full())
776
			try_to_free_swap(page);
L
Lee Schermerhorn 已提交
777
		VM_BUG_ON(PageActive(page));
L
Linus Torvalds 已提交
778 779 780 781 782 783
		SetPageActive(page);
		pgactivate++;
keep_locked:
		unlock_page(page);
keep:
		list_add(&page->lru, &ret_pages);
N
Nick Piggin 已提交
784
		VM_BUG_ON(PageLRU(page) || PageUnevictable(page));
L
Linus Torvalds 已提交
785 786 787
	}
	list_splice(&ret_pages, page_list);
	if (pagevec_count(&freed_pvec))
N
Nick Piggin 已提交
788
		__pagevec_free(&freed_pvec);
789
	count_vm_events(PGACTIVATE, pgactivate);
790
	return nr_reclaimed;
L
Linus Torvalds 已提交
791 792
}

A
Andy Whitcroft 已提交
793 794 795 796 797 798 799 800 801 802 803 804 805 806 807
/* 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.
 */
808
int __isolate_lru_page(struct page *page, int mode, int file)
A
Andy Whitcroft 已提交
809 810 811 812 813 814 815 816 817 818 819 820 821 822 823
{
	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;

824 825 826
	if (mode != ISOLATE_BOTH && (!page_is_file_cache(page) != !file))
		return ret;

L
Lee Schermerhorn 已提交
827 828 829 830 831 832 833 834
	/*
	 * 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
835
	ret = -EBUSY;
K
KAMEZAWA Hiroyuki 已提交
836

A
Andy Whitcroft 已提交
837 838 839 840 841 842 843 844 845 846 847 848 849
	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
850 851 852 853 854 855 856 857 858 859 860 861 862 863
/*
 * 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
864 865
 * @order:	The caller's attempted allocation order
 * @mode:	One of the LRU isolation modes
866
 * @file:	True [1] if isolating file [!anon] pages
L
Linus Torvalds 已提交
867 868 869
 *
 * returns how many pages were moved onto *@dst.
 */
870 871
static unsigned long isolate_lru_pages(unsigned long nr_to_scan,
		struct list_head *src, struct list_head *dst,
872
		unsigned long *scanned, int order, int mode, int file)
L
Linus Torvalds 已提交
873
{
874
	unsigned long nr_taken = 0;
875
	unsigned long scan;
L
Linus Torvalds 已提交
876

877
	for (scan = 0; scan < nr_to_scan && !list_empty(src); scan++) {
A
Andy Whitcroft 已提交
878 879 880 881 882 883
		struct page *page;
		unsigned long pfn;
		unsigned long end_pfn;
		unsigned long page_pfn;
		int zone_id;

L
Linus Torvalds 已提交
884 885 886
		page = lru_to_page(src);
		prefetchw_prev_lru_page(page, src, flags);

N
Nick Piggin 已提交
887
		VM_BUG_ON(!PageLRU(page));
N
Nick Piggin 已提交
888

889
		switch (__isolate_lru_page(page, mode, file)) {
A
Andy Whitcroft 已提交
890 891
		case 0:
			list_move(&page->lru, dst);
892
			mem_cgroup_del_lru(page);
893
			nr_taken++;
A
Andy Whitcroft 已提交
894 895 896 897 898
			break;

		case -EBUSY:
			/* else it is being freed elsewhere */
			list_move(&page->lru, src);
899
			mem_cgroup_rotate_lru_list(page, page_lru(page));
A
Andy Whitcroft 已提交
900
			continue;
901

A
Andy Whitcroft 已提交
902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933
		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);
934

A
Andy Whitcroft 已提交
935 936 937
			/* Check that we have not crossed a zone boundary. */
			if (unlikely(page_zone_id(cursor_page) != zone_id))
				continue;
938 939 940 941 942 943 944 945 946 947

			/*
			 * 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;

948
			if (__isolate_lru_page(cursor_page, mode, file) == 0) {
A
Andy Whitcroft 已提交
949
				list_move(&cursor_page->lru, dst);
950
				mem_cgroup_del_lru(cursor_page);
A
Andy Whitcroft 已提交
951 952 953 954
				nr_taken++;
				scan++;
			}
		}
L
Linus Torvalds 已提交
955 956 957 958 959 960
	}

	*scanned = scan;
	return nr_taken;
}

961 962 963 964 965
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,
966
					int active, int file)
967
{
968
	int lru = LRU_BASE;
969
	if (active)
970 971 972 973 974
		lru += LRU_ACTIVE;
	if (file)
		lru += LRU_FILE;
	return isolate_lru_pages(nr, &z->lru[lru].list, dst, scanned, order,
								mode, !!file);
975 976
}

A
Andy Whitcroft 已提交
977 978 979 980
/*
 * clear_active_flags() is a helper for shrink_active_list(), clearing
 * any active bits from the pages in the list.
 */
981 982
static unsigned long clear_active_flags(struct list_head *page_list,
					unsigned int *count)
A
Andy Whitcroft 已提交
983 984
{
	int nr_active = 0;
985
	int lru;
A
Andy Whitcroft 已提交
986 987
	struct page *page;

988 989
	list_for_each_entry(page, page_list, lru) {
		lru = page_is_file_cache(page);
A
Andy Whitcroft 已提交
990
		if (PageActive(page)) {
991
			lru += LRU_ACTIVE;
A
Andy Whitcroft 已提交
992 993 994
			ClearPageActive(page);
			nr_active++;
		}
995 996
		count[lru]++;
	}
A
Andy Whitcroft 已提交
997 998 999 1000

	return nr_active;
}

1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011
/**
 * 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
1012 1013 1014
 * 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.
1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034
 *
 * 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
1035
			int lru = page_lru(page);
1036 1037
			ret = 0;
			ClearPageLRU(page);
1038 1039

			del_page_from_lru_list(zone, page, lru);
1040 1041 1042 1043 1044 1045
		}
		spin_unlock_irq(&zone->lru_lock);
	}
	return ret;
}

1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070
/*
 * 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
1071
/*
A
Andrew Morton 已提交
1072 1073
 * shrink_inactive_list() is a helper for shrink_zone().  It returns the number
 * of reclaimed pages
L
Linus Torvalds 已提交
1074
 */
A
Andrew Morton 已提交
1075
static unsigned long shrink_inactive_list(unsigned long max_scan,
R
Rik van Riel 已提交
1076 1077
			struct zone *zone, struct scan_control *sc,
			int priority, int file)
L
Linus Torvalds 已提交
1078 1079 1080
{
	LIST_HEAD(page_list);
	struct pagevec pvec;
1081
	unsigned long nr_scanned = 0;
1082
	unsigned long nr_reclaimed = 0;
1083
	struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(zone, sc);
1084 1085
	int lumpy_reclaim = 0;

1086 1087 1088 1089 1090 1091 1092 1093
	while (unlikely(too_many_isolated(zone, file, sc))) {
		congestion_wait(WRITE, HZ/10);

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

1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104
	/*
	 * 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
1105 1106 1107 1108 1109

	pagevec_init(&pvec, 1);

	lru_add_drain();
	spin_lock_irq(&zone->lru_lock);
1110
	do {
L
Linus Torvalds 已提交
1111
		struct page *page;
1112 1113 1114
		unsigned long nr_taken;
		unsigned long nr_scan;
		unsigned long nr_freed;
A
Andy Whitcroft 已提交
1115
		unsigned long nr_active;
1116
		unsigned int count[NR_LRU_LISTS] = { 0, };
1117
		int mode = lumpy_reclaim ? ISOLATE_BOTH : ISOLATE_INACTIVE;
K
KOSAKI Motohiro 已提交
1118 1119
		unsigned long nr_anon;
		unsigned long nr_file;
L
Linus Torvalds 已提交
1120

1121
		nr_taken = sc->isolate_pages(sc->swap_cluster_max,
1122 1123
			     &page_list, &nr_scan, sc->order, mode,
				zone, sc->mem_cgroup, 0, file);
1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137

		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;

1138
		nr_active = clear_active_flags(&page_list, count);
1139
		__count_vm_events(PGDEACTIVATE, nr_active);
A
Andy Whitcroft 已提交
1140

1141 1142 1143 1144 1145 1146 1147 1148 1149
		__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
1150 1151 1152 1153
		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
1154 1155 1156 1157 1158 1159

		reclaim_stat->recent_scanned[0] += count[LRU_INACTIVE_ANON];
		reclaim_stat->recent_scanned[0] += count[LRU_ACTIVE_ANON];
		reclaim_stat->recent_scanned[1] += count[LRU_INACTIVE_FILE];
		reclaim_stat->recent_scanned[1] += count[LRU_ACTIVE_FILE];

L
Linus Torvalds 已提交
1160 1161
		spin_unlock_irq(&zone->lru_lock);

1162
		nr_scanned += nr_scan;
1163 1164 1165 1166 1167 1168 1169 1170 1171
		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() &&
1172
		    lumpy_reclaim) {
1173
			congestion_wait(BLK_RW_ASYNC, HZ/10);
1174 1175 1176 1177 1178

			/*
			 * The attempt at page out may have made some
			 * of the pages active, mark them inactive again.
			 */
1179
			nr_active = clear_active_flags(&page_list, count);
1180 1181 1182 1183 1184 1185
			count_vm_events(PGDEACTIVATE, nr_active);

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

1186
		nr_reclaimed += nr_freed;
1187

N
Nick Piggin 已提交
1188
		local_irq_disable();
1189
		if (current_is_kswapd())
1190
			__count_vm_events(KSWAPD_STEAL, nr_freed);
S
Shantanu Goel 已提交
1191
		__count_zone_vm_events(PGSTEAL, zone, nr_freed);
N
Nick Piggin 已提交
1192 1193

		spin_lock(&zone->lru_lock);
L
Linus Torvalds 已提交
1194 1195 1196 1197
		/*
		 * Put back any unfreeable pages.
		 */
		while (!list_empty(&page_list)) {
L
Lee Schermerhorn 已提交
1198
			int lru;
L
Linus Torvalds 已提交
1199
			page = lru_to_page(&page_list);
N
Nick Piggin 已提交
1200
			VM_BUG_ON(PageLRU(page));
L
Linus Torvalds 已提交
1201
			list_del(&page->lru);
L
Lee Schermerhorn 已提交
1202 1203 1204 1205 1206 1207 1208 1209 1210
			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);
1211 1212
			if (is_active_lru(lru)) {
				int file = !!is_file_lru(lru);
1213
				reclaim_stat->recent_rotated[file]++;
1214
			}
L
Linus Torvalds 已提交
1215 1216 1217 1218 1219 1220
			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
1221 1222 1223
		__mod_zone_page_state(zone, NR_ISOLATED_ANON, -nr_anon);
		__mod_zone_page_state(zone, NR_ISOLATED_FILE, -nr_file);

1224
  	} while (nr_scanned < max_scan);
1225

L
Linus Torvalds 已提交
1226
done:
1227
	spin_unlock_irq(&zone->lru_lock);
L
Linus Torvalds 已提交
1228
	pagevec_release(&pvec);
1229
	return nr_reclaimed;
L
Linus Torvalds 已提交
1230 1231
}

1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245
/*
 * 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
1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262
/*
 * 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.
 */
1263

1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295
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);
}
1296

A
Andrew Morton 已提交
1297
static void shrink_active_list(unsigned long nr_pages, struct zone *zone,
1298
			struct scan_control *sc, int priority, int file)
L
Linus Torvalds 已提交
1299
{
1300
	unsigned long nr_taken;
1301
	unsigned long pgscanned;
1302
	unsigned long vm_flags;
L
Linus Torvalds 已提交
1303
	LIST_HEAD(l_hold);	/* The pages which were snipped off */
1304
	LIST_HEAD(l_active);
1305
	LIST_HEAD(l_inactive);
L
Linus Torvalds 已提交
1306
	struct page *page;
1307
	struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(zone, sc);
1308
	unsigned long nr_rotated = 0;
L
Linus Torvalds 已提交
1309 1310 1311

	lru_add_drain();
	spin_lock_irq(&zone->lru_lock);
1312
	nr_taken = sc->isolate_pages(nr_pages, &l_hold, &pgscanned, sc->order,
1313
					ISOLATE_ACTIVE, zone,
1314
					sc->mem_cgroup, 1, file);
1315 1316 1317 1318
	/*
	 * zone->pages_scanned is used for detect zone's oom
	 * mem_cgroup remembers nr_scan by itself.
	 */
1319
	if (scanning_global_lru(sc)) {
1320
		zone->pages_scanned += pgscanned;
1321
	}
1322
	reclaim_stat->recent_scanned[!!file] += nr_taken;
1323

1324
	__count_zone_vm_events(PGREFILL, zone, pgscanned);
1325
	if (file)
1326
		__mod_zone_page_state(zone, NR_ACTIVE_FILE, -nr_taken);
1327
	else
1328
		__mod_zone_page_state(zone, NR_ACTIVE_ANON, -nr_taken);
K
KOSAKI Motohiro 已提交
1329
	__mod_zone_page_state(zone, NR_ISOLATED_ANON + file, nr_taken);
L
Linus Torvalds 已提交
1330 1331 1332 1333 1334 1335
	spin_unlock_irq(&zone->lru_lock);

	while (!list_empty(&l_hold)) {
		cond_resched();
		page = lru_to_page(&l_hold);
		list_del(&page->lru);
1336

L
Lee Schermerhorn 已提交
1337 1338 1339 1340 1341
		if (unlikely(!page_evictable(page, NULL))) {
			putback_lru_page(page);
			continue;
		}

1342 1343
		/* page_referenced clears PageReferenced */
		if (page_mapping_inuse(page) &&
1344
		    page_referenced(page, 0, sc->mem_cgroup, &vm_flags)) {
1345
			nr_rotated++;
1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359
			/*
			 * 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.
			 */
			if ((vm_flags & VM_EXEC) && !PageAnon(page)) {
				list_add(&page->lru, &l_active);
				continue;
			}
		}
1360

1361
		ClearPageActive(page);	/* we are de-activating */
L
Linus Torvalds 已提交
1362 1363 1364
		list_add(&page->lru, &l_inactive);
	}

1365
	/*
1366
	 * Move pages back to the lru list.
1367
	 */
1368
	spin_lock_irq(&zone->lru_lock);
1369
	/*
1370 1371 1372 1373
	 * 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.
1374
	 */
1375
	reclaim_stat->recent_rotated[!!file] += nr_rotated;
1376

1377 1378 1379 1380
	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
1381
	__mod_zone_page_state(zone, NR_ISOLATED_ANON + file, -nr_taken);
1382
	spin_unlock_irq(&zone->lru_lock);
L
Linus Torvalds 已提交
1383 1384
}

1385
static int inactive_anon_is_low_global(struct zone *zone)
1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397
{
	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;
}

1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409
/**
 * 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;

1410
	if (scanning_global_lru(sc))
1411 1412
		low = inactive_anon_is_low_global(zone);
	else
1413
		low = mem_cgroup_inactive_anon_is_low(sc->mem_cgroup);
1414 1415 1416
	return low;
}

1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 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
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;
}

1453
static unsigned long shrink_list(enum lru_list lru, unsigned long nr_to_scan,
1454 1455
	struct zone *zone, struct scan_control *sc, int priority)
{
1456 1457
	int file = is_file_lru(lru);

1458
	if (lru == LRU_ACTIVE_FILE && inactive_file_is_low(zone, sc)) {
1459 1460 1461 1462
		shrink_active_list(nr_to_scan, zone, sc, priority, file);
		return 0;
	}

1463
	if (lru == LRU_ACTIVE_ANON && inactive_anon_is_low(zone, sc)) {
1464
		shrink_active_list(nr_to_scan, zone, sc, priority, file);
1465 1466
		return 0;
	}
R
Rik van Riel 已提交
1467
	return shrink_inactive_list(nr_to_scan, zone, sc, priority, file);
1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484
}

/*
 * 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;
1485
	struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(zone, sc);
1486

1487 1488 1489 1490
	anon  = zone_nr_pages(zone, sc, LRU_ACTIVE_ANON) +
		zone_nr_pages(zone, sc, LRU_INACTIVE_ANON);
	file  = zone_nr_pages(zone, sc, LRU_ACTIVE_FILE) +
		zone_nr_pages(zone, sc, LRU_INACTIVE_FILE);
1491

1492
	if (scanning_global_lru(sc)) {
1493 1494 1495
		free  = zone_page_state(zone, NR_FREE_PAGES);
		/* If we have very few page cache pages,
		   force-scan anon pages. */
1496
		if (unlikely(file + free <= high_wmark_pages(zone))) {
1497 1498 1499 1500
			percent[0] = 100;
			percent[1] = 0;
			return;
		}
1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513
	}

	/*
	 * 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]
	 */
1514
	if (unlikely(reclaim_stat->recent_scanned[0] > anon / 4)) {
1515
		spin_lock_irq(&zone->lru_lock);
1516 1517
		reclaim_stat->recent_scanned[0] /= 2;
		reclaim_stat->recent_rotated[0] /= 2;
1518 1519 1520
		spin_unlock_irq(&zone->lru_lock);
	}

1521
	if (unlikely(reclaim_stat->recent_scanned[1] > file / 4)) {
1522
		spin_lock_irq(&zone->lru_lock);
1523 1524
		reclaim_stat->recent_scanned[1] /= 2;
		reclaim_stat->recent_rotated[1] /= 2;
1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535
		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;

	/*
1536 1537 1538
	 * 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.
1539
	 */
1540 1541
	ap = (anon_prio + 1) * (reclaim_stat->recent_scanned[0] + 1);
	ap /= reclaim_stat->recent_rotated[0] + 1;
1542

1543 1544
	fp = (file_prio + 1) * (reclaim_stat->recent_scanned[1] + 1);
	fp /= reclaim_stat->recent_rotated[1] + 1;
1545 1546 1547 1548

	/* Normalize to percentages */
	percent[0] = 100 * ap / (ap + fp + 1);
	percent[1] = 100 - percent[0];
1549 1550
}

1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570
/*
 * 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,
				       unsigned long *nr_saved_scan,
				       unsigned long swap_cluster_max)
{
	unsigned long nr;

	*nr_saved_scan += nr_to_scan;
	nr = *nr_saved_scan;

	if (nr >= swap_cluster_max)
		*nr_saved_scan = 0;
	else
		nr = 0;

	return nr;
}
1571

L
Linus Torvalds 已提交
1572 1573 1574
/*
 * This is a basic per-zone page freer.  Used by both kswapd and direct reclaim.
 */
1575
static void shrink_zone(int priority, struct zone *zone,
1576
				struct scan_control *sc)
L
Linus Torvalds 已提交
1577
{
1578
	unsigned long nr[NR_LRU_LISTS];
1579
	unsigned long nr_to_scan;
1580
	unsigned long percent[2];	/* anon @ 0; file @ 1 */
1581
	enum lru_list l;
1582 1583
	unsigned long nr_reclaimed = sc->nr_reclaimed;
	unsigned long swap_cluster_max = sc->swap_cluster_max;
1584
	int noswap = 0;
L
Linus Torvalds 已提交
1585

1586 1587 1588 1589 1590 1591 1592
	/* 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);
1593

L
Lee Schermerhorn 已提交
1594
	for_each_evictable_lru(l) {
1595
		int file = is_file_lru(l);
1596
		unsigned long scan;
1597

1598
		scan = zone_nr_pages(zone, sc, l);
1599
		if (priority || noswap) {
1600 1601 1602
			scan >>= priority;
			scan = (scan * percent[file]) / 100;
		}
1603 1604 1605 1606 1607
		if (scanning_global_lru(sc))
			nr[l] = nr_scan_try_batch(scan,
						  &zone->lru[l].nr_saved_scan,
						  swap_cluster_max);
		else
1608
			nr[l] = scan;
1609
	}
L
Linus Torvalds 已提交
1610

1611 1612
	while (nr[LRU_INACTIVE_ANON] || nr[LRU_ACTIVE_FILE] ||
					nr[LRU_INACTIVE_FILE]) {
L
Lee Schermerhorn 已提交
1613
		for_each_evictable_lru(l) {
1614
			if (nr[l]) {
1615
				nr_to_scan = min(nr[l], swap_cluster_max);
1616
				nr[l] -= nr_to_scan;
L
Linus Torvalds 已提交
1617

1618 1619
				nr_reclaimed += shrink_list(l, nr_to_scan,
							    zone, sc, priority);
1620
			}
L
Linus Torvalds 已提交
1621
		}
1622 1623 1624 1625 1626 1627 1628 1629
		/*
		 * 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.
		 */
1630
		if (nr_reclaimed > swap_cluster_max &&
1631 1632
			priority < DEF_PRIORITY && !current_is_kswapd())
			break;
L
Linus Torvalds 已提交
1633 1634
	}

1635 1636
	sc->nr_reclaimed = nr_reclaimed;

1637 1638 1639 1640
	/*
	 * Even if we did not try to evict anon pages at all, we want to
	 * rebalance the anon lru active/inactive ratio.
	 */
1641
	if (inactive_anon_is_low(zone, sc) && nr_swap_pages > 0)
1642 1643
		shrink_active_list(SWAP_CLUSTER_MAX, zone, sc, priority, 0);

1644
	throttle_vm_writeout(sc->gfp_mask);
L
Linus Torvalds 已提交
1645 1646 1647 1648 1649 1650 1651
}

/*
 * 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.
 *
1652 1653
 * We reclaim from a zone even if that zone is over high_wmark_pages(zone).
 * Because:
L
Linus Torvalds 已提交
1654 1655
 * a) The caller may be trying to free *extra* pages to satisfy a higher-order
 *    allocation or
1656 1657 1658
 * 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
1659 1660 1661 1662
 *
 * If a zone is deemed to be full of pinned pages then just give it a light
 * scan then give up on it.
 */
1663
static void shrink_zones(int priority, struct zonelist *zonelist,
1664
					struct scan_control *sc)
L
Linus Torvalds 已提交
1665
{
1666
	enum zone_type high_zoneidx = gfp_zone(sc->gfp_mask);
1667
	struct zoneref *z;
1668
	struct zone *zone;
1669

1670
	sc->all_unreclaimable = 1;
1671 1672
	for_each_zone_zonelist_nodemask(zone, z, zonelist, high_zoneidx,
					sc->nodemask) {
1673
		if (!populated_zone(zone))
L
Linus Torvalds 已提交
1674
			continue;
1675 1676 1677 1678
		/*
		 * Take care memory controller reclaiming has small influence
		 * to global LRU.
		 */
1679
		if (scanning_global_lru(sc)) {
1680 1681 1682
			if (!cpuset_zone_allowed_hardwall(zone, GFP_KERNEL))
				continue;
			note_zone_scanning_priority(zone, priority);
L
Linus Torvalds 已提交
1683

1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696
			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);
		}
1697

1698
		shrink_zone(priority, zone, sc);
L
Linus Torvalds 已提交
1699 1700
	}
}
1701

L
Linus Torvalds 已提交
1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713
/*
 * 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
 * caller can't do much about.  We kick pdflush 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.
1714 1715 1716
 *
 * returns:	0, if no pages reclaimed
 * 		else, the number of pages reclaimed
L
Linus Torvalds 已提交
1717
 */
1718
static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
1719
					struct scan_control *sc)
L
Linus Torvalds 已提交
1720 1721
{
	int priority;
1722
	unsigned long ret = 0;
1723
	unsigned long total_scanned = 0;
L
Linus Torvalds 已提交
1724 1725
	struct reclaim_state *reclaim_state = current->reclaim_state;
	unsigned long lru_pages = 0;
1726
	struct zoneref *z;
1727
	struct zone *zone;
1728
	enum zone_type high_zoneidx = gfp_zone(sc->gfp_mask);
L
Linus Torvalds 已提交
1729

1730 1731
	delayacct_freepages_start();

1732
	if (scanning_global_lru(sc))
1733 1734 1735 1736
		count_vm_event(ALLOCSTALL);
	/*
	 * mem_cgroup will not do shrink_slab.
	 */
1737
	if (scanning_global_lru(sc)) {
1738
		for_each_zone_zonelist(zone, z, zonelist, high_zoneidx) {
L
Linus Torvalds 已提交
1739

1740 1741
			if (!cpuset_zone_allowed_hardwall(zone, GFP_KERNEL))
				continue;
L
Linus Torvalds 已提交
1742

1743
			lru_pages += zone_reclaimable_pages(zone);
1744
		}
L
Linus Torvalds 已提交
1745 1746 1747
	}

	for (priority = DEF_PRIORITY; priority >= 0; priority--) {
1748
		sc->nr_scanned = 0;
1749 1750
		if (!priority)
			disable_swap_token();
1751
		shrink_zones(priority, zonelist, sc);
1752 1753 1754 1755
		/*
		 * Don't shrink slabs when reclaiming memory from
		 * over limit cgroups
		 */
1756
		if (scanning_global_lru(sc)) {
1757
			shrink_slab(sc->nr_scanned, sc->gfp_mask, lru_pages);
1758
			if (reclaim_state) {
1759
				sc->nr_reclaimed += reclaim_state->reclaimed_slab;
1760 1761
				reclaim_state->reclaimed_slab = 0;
			}
L
Linus Torvalds 已提交
1762
		}
1763
		total_scanned += sc->nr_scanned;
1764 1765
		if (sc->nr_reclaimed >= sc->swap_cluster_max) {
			ret = sc->nr_reclaimed;
L
Linus Torvalds 已提交
1766 1767 1768 1769 1770 1771 1772 1773 1774 1775
			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.
		 */
1776 1777
		if (total_scanned > sc->swap_cluster_max +
					sc->swap_cluster_max / 2) {
1778
			wakeup_flusher_threads(laptop_mode ? 0 : total_scanned);
1779
			sc->may_writepage = 1;
L
Linus Torvalds 已提交
1780 1781 1782
		}

		/* Take a nap, wait for some writeback to complete */
1783
		if (sc->nr_scanned && priority < DEF_PRIORITY - 2)
1784
			congestion_wait(BLK_RW_ASYNC, HZ/10);
L
Linus Torvalds 已提交
1785
	}
1786
	/* top priority shrink_zones still had more to do? don't OOM, then */
1787
	if (!sc->all_unreclaimable && scanning_global_lru(sc))
1788
		ret = sc->nr_reclaimed;
L
Linus Torvalds 已提交
1789
out:
1790 1791 1792 1793 1794 1795 1796 1797 1798
	/*
	 * 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
1799

1800
	if (scanning_global_lru(sc)) {
1801
		for_each_zone_zonelist(zone, z, zonelist, high_zoneidx) {
1802 1803 1804 1805 1806 1807 1808 1809

			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
1810

1811 1812
	delayacct_freepages_end();

L
Linus Torvalds 已提交
1813 1814 1815
	return ret;
}

1816
unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
1817
				gfp_t gfp_mask, nodemask_t *nodemask)
1818 1819 1820 1821 1822
{
	struct scan_control sc = {
		.gfp_mask = gfp_mask,
		.may_writepage = !laptop_mode,
		.swap_cluster_max = SWAP_CLUSTER_MAX,
1823
		.may_unmap = 1,
1824
		.may_swap = 1,
1825 1826 1827 1828
		.swappiness = vm_swappiness,
		.order = order,
		.mem_cgroup = NULL,
		.isolate_pages = isolate_pages_global,
1829
		.nodemask = nodemask,
1830 1831
	};

1832
	return do_try_to_free_pages(zonelist, &sc);
1833 1834
}

1835
#ifdef CONFIG_CGROUP_MEM_RES_CTLR
1836

1837
unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *mem_cont,
K
KOSAKI Motohiro 已提交
1838 1839 1840
					   gfp_t gfp_mask,
					   bool noswap,
					   unsigned int swappiness)
1841 1842 1843
{
	struct scan_control sc = {
		.may_writepage = !laptop_mode,
1844
		.may_unmap = 1,
1845
		.may_swap = !noswap,
1846
		.swap_cluster_max = SWAP_CLUSTER_MAX,
K
KOSAKI Motohiro 已提交
1847
		.swappiness = swappiness,
1848 1849 1850
		.order = 0,
		.mem_cgroup = mem_cont,
		.isolate_pages = mem_cgroup_isolate_pages,
1851
		.nodemask = NULL, /* we don't care the placement */
1852
	};
1853
	struct zonelist *zonelist;
1854

1855 1856 1857 1858
	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);
1859 1860 1861
}
#endif

L
Linus Torvalds 已提交
1862 1863
/*
 * For kswapd, balance_pgdat() will work across all this node's zones until
1864
 * they are all at high_wmark_pages(zone).
L
Linus Torvalds 已提交
1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876
 *
 * 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
1877 1878 1879 1880 1881
 * 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
1882
 */
1883
static unsigned long balance_pgdat(pg_data_t *pgdat, int order)
L
Linus Torvalds 已提交
1884 1885 1886 1887
{
	int all_zones_ok;
	int priority;
	int i;
1888
	unsigned long total_scanned;
L
Linus Torvalds 已提交
1889
	struct reclaim_state *reclaim_state = current->reclaim_state;
1890 1891
	struct scan_control sc = {
		.gfp_mask = GFP_KERNEL,
1892
		.may_unmap = 1,
1893
		.may_swap = 1,
1894 1895
		.swap_cluster_max = SWAP_CLUSTER_MAX,
		.swappiness = vm_swappiness,
A
Andy Whitcroft 已提交
1896
		.order = order,
1897 1898
		.mem_cgroup = NULL,
		.isolate_pages = isolate_pages_global,
1899
	};
1900 1901
	/*
	 * temp_priority is used to remember the scanning priority at which
1902 1903
	 * this zone was successfully refilled to
	 * free_pages == high_wmark_pages(zone).
1904 1905
	 */
	int temp_priority[MAX_NR_ZONES];
L
Linus Torvalds 已提交
1906 1907 1908

loop_again:
	total_scanned = 0;
1909
	sc.nr_reclaimed = 0;
C
Christoph Lameter 已提交
1910
	sc.may_writepage = !laptop_mode;
1911
	count_vm_event(PAGEOUTRUN);
L
Linus Torvalds 已提交
1912

1913 1914
	for (i = 0; i < pgdat->nr_zones; i++)
		temp_priority[i] = DEF_PRIORITY;
L
Linus Torvalds 已提交
1915 1916 1917 1918 1919

	for (priority = DEF_PRIORITY; priority >= 0; priority--) {
		int end_zone = 0;	/* Inclusive.  0 = ZONE_DMA */
		unsigned long lru_pages = 0;

1920 1921 1922 1923
		/* The swap token gets in the way of swapout... */
		if (!priority)
			disable_swap_token();

L
Linus Torvalds 已提交
1924 1925
		all_zones_ok = 1;

1926 1927 1928 1929 1930 1931
		/*
		 * 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
1932

1933 1934
			if (!populated_zone(zone))
				continue;
L
Linus Torvalds 已提交
1935

1936 1937
			if (zone_is_all_unreclaimable(zone) &&
			    priority != DEF_PRIORITY)
1938
				continue;
L
Linus Torvalds 已提交
1939

1940 1941 1942 1943
			/*
			 * Do some background aging of the anon list, to give
			 * pages a chance to be referenced before reclaiming.
			 */
1944
			if (inactive_anon_is_low(zone, &sc))
1945 1946 1947
				shrink_active_list(SWAP_CLUSTER_MAX, zone,
							&sc, priority, 0);

1948 1949
			if (!zone_watermark_ok(zone, order,
					high_wmark_pages(zone), 0, 0)) {
1950
				end_zone = i;
A
Andrew Morton 已提交
1951
				break;
L
Linus Torvalds 已提交
1952 1953
			}
		}
A
Andrew Morton 已提交
1954 1955 1956
		if (i < 0)
			goto out;

L
Linus Torvalds 已提交
1957 1958 1959
		for (i = 0; i <= end_zone; i++) {
			struct zone *zone = pgdat->node_zones + i;

1960
			lru_pages += zone_reclaimable_pages(zone);
L
Linus Torvalds 已提交
1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973
		}

		/*
		 * 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;
1974
			int nr_slab;
L
Linus Torvalds 已提交
1975

1976
			if (!populated_zone(zone))
L
Linus Torvalds 已提交
1977 1978
				continue;

1979 1980
			if (zone_is_all_unreclaimable(zone) &&
					priority != DEF_PRIORITY)
L
Linus Torvalds 已提交
1981 1982
				continue;

1983 1984
			if (!zone_watermark_ok(zone, order,
					high_wmark_pages(zone), end_zone, 0))
1985
				all_zones_ok = 0;
1986
			temp_priority[i] = priority;
L
Linus Torvalds 已提交
1987
			sc.nr_scanned = 0;
1988
			note_zone_scanning_priority(zone, priority);
1989 1990 1991 1992
			/*
			 * We put equal pressure on every zone, unless one
			 * zone has way too many pages free already.
			 */
1993 1994
			if (!zone_watermark_ok(zone, order,
					8*high_wmark_pages(zone), end_zone, 0))
1995
				shrink_zone(priority, zone, &sc);
L
Linus Torvalds 已提交
1996
			reclaim_state->reclaimed_slab = 0;
1997 1998
			nr_slab = shrink_slab(sc.nr_scanned, GFP_KERNEL,
						lru_pages);
1999
			sc.nr_reclaimed += reclaim_state->reclaimed_slab;
L
Linus Torvalds 已提交
2000
			total_scanned += sc.nr_scanned;
2001
			if (zone_is_all_unreclaimable(zone))
L
Linus Torvalds 已提交
2002
				continue;
2003
			if (nr_slab == 0 && zone->pages_scanned >=
2004
					(zone_reclaimable_pages(zone) * 6))
2005 2006
					zone_set_flag(zone,
						      ZONE_ALL_UNRECLAIMABLE);
L
Linus Torvalds 已提交
2007 2008 2009 2010 2011 2012
			/*
			 * 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 &&
2013
			    total_scanned > sc.nr_reclaimed + sc.nr_reclaimed / 2)
L
Linus Torvalds 已提交
2014 2015 2016 2017 2018 2019 2020 2021
				sc.may_writepage = 1;
		}
		if (all_zones_ok)
			break;		/* kswapd: all done */
		/*
		 * OK, kswapd is getting into trouble.  Take a nap, then take
		 * another pass across the zones.
		 */
2022
		if (total_scanned && priority < DEF_PRIORITY - 2)
2023
			congestion_wait(BLK_RW_ASYNC, HZ/10);
L
Linus Torvalds 已提交
2024 2025 2026 2027 2028 2029 2030

		/*
		 * 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.
		 */
2031
		if (sc.nr_reclaimed >= SWAP_CLUSTER_MAX)
L
Linus Torvalds 已提交
2032 2033 2034
			break;
	}
out:
2035 2036 2037 2038 2039
	/*
	 * 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
2040 2041 2042
	for (i = 0; i < pgdat->nr_zones; i++) {
		struct zone *zone = pgdat->node_zones + i;

2043
		zone->prev_priority = temp_priority[i];
L
Linus Torvalds 已提交
2044 2045 2046
	}
	if (!all_zones_ok) {
		cond_resched();
2047 2048 2049

		try_to_freeze();

2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066
		/*
		 * 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
2067 2068 2069
		goto loop_again;
	}

2070
	return sc.nr_reclaimed;
L
Linus Torvalds 已提交
2071 2072 2073 2074
}

/*
 * The background pageout daemon, started as a kernel thread
2075
 * from the init process.
L
Linus Torvalds 已提交
2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094
 *
 * 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,
	};
2095
	const struct cpumask *cpumask = cpumask_of_node(pgdat->node_id);
L
Linus Torvalds 已提交
2096

2097 2098
	lockdep_set_current_reclaim_state(GFP_KERNEL);

R
Rusty Russell 已提交
2099
	if (!cpumask_empty(cpumask))
2100
		set_cpus_allowed_ptr(tsk, cpumask);
L
Linus Torvalds 已提交
2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114
	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).
	 */
2115
	tsk->flags |= PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD;
2116
	set_freezable();
L
Linus Torvalds 已提交
2117 2118 2119 2120

	order = 0;
	for ( ; ; ) {
		unsigned long new_order;
2121

L
Linus Torvalds 已提交
2122 2123 2124 2125 2126 2127 2128 2129 2130 2131
		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 {
2132 2133 2134
			if (!freezing(current))
				schedule();

L
Linus Torvalds 已提交
2135 2136 2137 2138
			order = pgdat->kswapd_max_order;
		}
		finish_wait(&pgdat->kswapd_wait, &wait);

2139 2140 2141 2142 2143 2144
		if (!try_to_freeze()) {
			/* We can speed up thawing tasks if we don't call
			 * balance_pgdat after returning from the refrigerator
			 */
			balance_pgdat(pgdat, order);
		}
L
Linus Torvalds 已提交
2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155
	}
	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;

2156
	if (!populated_zone(zone))
L
Linus Torvalds 已提交
2157 2158 2159
		return;

	pgdat = zone->zone_pgdat;
2160
	if (zone_watermark_ok(zone, order, low_wmark_pages(zone), 0, 0))
L
Linus Torvalds 已提交
2161 2162 2163
		return;
	if (pgdat->kswapd_max_order < order)
		pgdat->kswapd_max_order = order;
2164
	if (!cpuset_zone_allowed_hardwall(zone, GFP_KERNEL))
L
Linus Torvalds 已提交
2165
		return;
2166
	if (!waitqueue_active(&pgdat->kswapd_wait))
L
Linus Torvalds 已提交
2167
		return;
2168
	wake_up_interruptible(&pgdat->kswapd_wait);
L
Linus Torvalds 已提交
2169 2170
}

2171 2172 2173 2174 2175 2176 2177 2178
/*
 * 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)
2179
{
2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203
	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;
2204 2205
}

2206
#ifdef CONFIG_HIBERNATION
L
Linus Torvalds 已提交
2207
/*
2208
 * Helper function for shrink_all_memory().  Tries to reclaim 'nr_pages' pages
2209
 * from LRU lists system-wide, for given pass and priority.
2210 2211 2212
 *
 * For pass > 3 we also try to shrink the LRU lists that contain a few pages
 */
2213
static void shrink_all_zones(unsigned long nr_pages, int prio,
2214
				      int pass, struct scan_control *sc)
2215 2216
{
	struct zone *zone;
2217
	unsigned long nr_reclaimed = 0;
2218

2219
	for_each_populated_zone(zone) {
2220
		enum lru_list l;
2221

2222
		if (zone_is_all_unreclaimable(zone) && prio != DEF_PRIORITY)
2223 2224
			continue;

L
Lee Schermerhorn 已提交
2225
		for_each_evictable_lru(l) {
2226 2227 2228
			enum zone_stat_item ls = NR_LRU_BASE + l;
			unsigned long lru_pages = zone_page_state(zone, ls);

L
Lee Schermerhorn 已提交
2229
			/* For pass = 0, we don't shrink the active list */
2230 2231
			if (pass == 0 && (l == LRU_ACTIVE_ANON ||
						l == LRU_ACTIVE_FILE))
2232 2233
				continue;

2234 2235
			zone->lru[l].nr_saved_scan += (lru_pages >> prio) + 1;
			if (zone->lru[l].nr_saved_scan >= nr_pages || pass > 3) {
2236 2237
				unsigned long nr_to_scan;

2238
				zone->lru[l].nr_saved_scan = 0;
2239
				nr_to_scan = min(nr_pages, lru_pages);
2240
				nr_reclaimed += shrink_list(l, nr_to_scan, zone,
2241
								sc, prio);
2242
				if (nr_reclaimed >= nr_pages) {
2243
					sc->nr_reclaimed += nr_reclaimed;
2244 2245
					return;
				}
2246 2247 2248
			}
		}
	}
2249
	sc->nr_reclaimed += nr_reclaimed;
2250 2251 2252 2253 2254 2255 2256 2257 2258
}

/*
 * Try to free `nr_pages' of memory, system-wide, and return the number of
 * 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
2259
 */
2260
unsigned long shrink_all_memory(unsigned long nr_pages)
L
Linus Torvalds 已提交
2261
{
2262 2263 2264 2265 2266
	unsigned long lru_pages, nr_slab;
	int pass;
	struct reclaim_state reclaim_state;
	struct scan_control sc = {
		.gfp_mask = GFP_KERNEL,
2267
		.may_unmap = 0,
2268
		.may_writepage = 1,
2269
		.isolate_pages = isolate_pages_global,
2270
		.nr_reclaimed = 0,
L
Linus Torvalds 已提交
2271 2272 2273
	};

	current->reclaim_state = &reclaim_state;
2274

2275
	lru_pages = global_reclaimable_pages();
2276
	nr_slab = global_page_state(NR_SLAB_RECLAIMABLE);
2277 2278 2279 2280 2281
	/* If slab caches are huge, it's better to hit them first */
	while (nr_slab >= lru_pages) {
		reclaim_state.reclaimed_slab = 0;
		shrink_slab(nr_pages, sc.gfp_mask, lru_pages);
		if (!reclaim_state.reclaimed_slab)
L
Linus Torvalds 已提交
2282
			break;
2283

2284 2285
		sc.nr_reclaimed += reclaim_state.reclaimed_slab;
		if (sc.nr_reclaimed >= nr_pages)
2286 2287 2288
			goto out;

		nr_slab -= reclaim_state.reclaimed_slab;
L
Linus Torvalds 已提交
2289
	}
2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302

	/*
	 * We try to shrink LRUs in 5 passes:
	 * 0 = Reclaim from inactive_list only
	 * 1 = Reclaim from active list but don't reclaim mapped
	 * 2 = 2nd pass of type 1
	 * 3 = Reclaim mapped (normal reclaim)
	 * 4 = 2nd pass of type 3
	 */
	for (pass = 0; pass < 5; pass++) {
		int prio;

		/* Force reclaiming mapped pages in the passes #3 and #4 */
2303
		if (pass > 2)
2304
			sc.may_unmap = 1;
2305 2306

		for (prio = DEF_PRIORITY; prio >= 0; prio--) {
2307
			unsigned long nr_to_scan = nr_pages - sc.nr_reclaimed;
2308 2309

			sc.nr_scanned = 0;
2310
			sc.swap_cluster_max = nr_to_scan;
2311 2312
			shrink_all_zones(nr_to_scan, prio, pass, &sc);
			if (sc.nr_reclaimed >= nr_pages)
2313 2314 2315
				goto out;

			reclaim_state.reclaimed_slab = 0;
2316
			shrink_slab(sc.nr_scanned, sc.gfp_mask,
2317
				    global_reclaimable_pages());
2318 2319
			sc.nr_reclaimed += reclaim_state.reclaimed_slab;
			if (sc.nr_reclaimed >= nr_pages)
2320 2321 2322
				goto out;

			if (sc.nr_scanned && prio < DEF_PRIORITY - 2)
2323
				congestion_wait(BLK_RW_ASYNC, HZ / 10);
2324
		}
2325
	}
2326 2327

	/*
2328 2329
	 * If sc.nr_reclaimed = 0, we could not shrink LRUs, but there may be
	 * something in slab caches
2330
	 */
2331
	if (!sc.nr_reclaimed) {
2332 2333
		do {
			reclaim_state.reclaimed_slab = 0;
2334 2335
			shrink_slab(nr_pages, sc.gfp_mask,
				    global_reclaimable_pages());
2336 2337 2338
			sc.nr_reclaimed += reclaim_state.reclaimed_slab;
		} while (sc.nr_reclaimed < nr_pages &&
				reclaim_state.reclaimed_slab > 0);
2339
	}
2340

2341

2342
out:
L
Linus Torvalds 已提交
2343
	current->reclaim_state = NULL;
2344

2345
	return sc.nr_reclaimed;
L
Linus Torvalds 已提交
2346
}
2347
#endif /* CONFIG_HIBERNATION */
L
Linus Torvalds 已提交
2348 2349 2350 2351 2352

/* 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. */
2353
static int __devinit cpu_callback(struct notifier_block *nfb,
2354
				  unsigned long action, void *hcpu)
L
Linus Torvalds 已提交
2355
{
2356
	int nid;
L
Linus Torvalds 已提交
2357

2358
	if (action == CPU_ONLINE || action == CPU_ONLINE_FROZEN) {
2359
		for_each_node_state(nid, N_HIGH_MEMORY) {
2360
			pg_data_t *pgdat = NODE_DATA(nid);
2361 2362 2363
			const struct cpumask *mask;

			mask = cpumask_of_node(pgdat->node_id);
2364

2365
			if (cpumask_any_and(cpu_online_mask, mask) < nr_cpu_ids)
L
Linus Torvalds 已提交
2366
				/* One of our CPUs online: restore mask */
2367
				set_cpus_allowed_ptr(pgdat->kswapd, mask);
L
Linus Torvalds 已提交
2368 2369 2370 2371 2372
		}
	}
	return NOTIFY_OK;
}

2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394
/*
 * 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;
}

L
Linus Torvalds 已提交
2395 2396
static int __init kswapd_init(void)
{
2397
	int nid;
2398

L
Linus Torvalds 已提交
2399
	swap_setup();
2400
	for_each_node_state(nid, N_HIGH_MEMORY)
2401
 		kswapd_run(nid);
L
Linus Torvalds 已提交
2402 2403 2404 2405 2406
	hotcpu_notifier(cpu_callback, 0);
	return 0;
}

module_init(kswapd_init)
2407 2408 2409 2410 2411 2412 2413 2414 2415 2416

#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;

2417
#define RECLAIM_OFF 0
2418
#define RECLAIM_ZONE (1<<0)	/* Run shrink_inactive_list on the zone */
2419 2420 2421
#define RECLAIM_WRITE (1<<1)	/* Writeout pages during reclaim */
#define RECLAIM_SWAP (1<<2)	/* Swap pages out during reclaim */

2422 2423 2424 2425 2426 2427 2428
/*
 * 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

2429 2430 2431 2432 2433 2434
/*
 * Percentage of pages in a zone that must be unmapped for zone_reclaim to
 * occur.
 */
int sysctl_min_unmapped_ratio = 1;

2435 2436 2437 2438 2439 2440
/*
 * 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;

2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482
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;
}

2483 2484 2485
/*
 * Try to free up some pages from this zone through reclaim.
 */
2486
static int __zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
2487
{
2488
	/* Minimum pages needed in order to stay on node */
2489
	const unsigned long nr_pages = 1 << order;
2490 2491
	struct task_struct *p = current;
	struct reclaim_state reclaim_state;
2492
	int priority;
2493 2494
	struct scan_control sc = {
		.may_writepage = !!(zone_reclaim_mode & RECLAIM_WRITE),
2495
		.may_unmap = !!(zone_reclaim_mode & RECLAIM_SWAP),
2496
		.may_swap = 1,
2497 2498
		.swap_cluster_max = max_t(unsigned long, nr_pages,
					SWAP_CLUSTER_MAX),
2499
		.gfp_mask = gfp_mask,
2500
		.swappiness = vm_swappiness,
2501
		.order = order,
2502
		.isolate_pages = isolate_pages_global,
2503
	};
2504
	unsigned long slab_reclaimable;
2505 2506 2507

	disable_swap_token();
	cond_resched();
2508 2509 2510 2511 2512 2513
	/*
	 * 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;
2514 2515
	reclaim_state.reclaimed_slab = 0;
	p->reclaim_state = &reclaim_state;
2516

2517
	if (zone_pagecache_reclaimable(zone) > zone->min_unmapped_pages) {
2518 2519 2520 2521 2522 2523
		/*
		 * Free memory by calling shrink zone with increasing
		 * priorities until we have enough memory freed.
		 */
		priority = ZONE_RECLAIM_PRIORITY;
		do {
2524
			note_zone_scanning_priority(zone, priority);
2525
			shrink_zone(priority, zone, &sc);
2526
			priority--;
2527
		} while (priority >= 0 && sc.nr_reclaimed < nr_pages);
2528
	}
2529

2530 2531
	slab_reclaimable = zone_page_state(zone, NR_SLAB_RECLAIMABLE);
	if (slab_reclaimable > zone->min_slab_pages) {
2532
		/*
2533
		 * shrink_slab() does not currently allow us to determine how
2534 2535 2536 2537
		 * 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.
2538
		 *
2539 2540
		 * Note that shrink_slab will free memory on all zones and may
		 * take a long time.
2541
		 */
2542
		while (shrink_slab(sc.nr_scanned, gfp_mask, order) &&
2543 2544
			zone_page_state(zone, NR_SLAB_RECLAIMABLE) >
				slab_reclaimable - nr_pages)
2545
			;
2546 2547 2548 2549 2550

		/*
		 * Update nr_reclaimed by the number of slab pages we
		 * reclaimed from this zone.
		 */
2551
		sc.nr_reclaimed += slab_reclaimable -
2552
			zone_page_state(zone, NR_SLAB_RECLAIMABLE);
2553 2554
	}

2555
	p->reclaim_state = NULL;
2556
	current->flags &= ~(PF_MEMALLOC | PF_SWAPWRITE);
2557
	return sc.nr_reclaimed >= nr_pages;
2558
}
2559 2560 2561 2562

int zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
{
	int node_id;
2563
	int ret;
2564 2565

	/*
2566 2567
	 * Zone reclaim reclaims unmapped file backed pages and
	 * slab pages if we are over the defined limits.
2568
	 *
2569 2570 2571 2572 2573
	 * 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.
2574
	 */
2575 2576
	if (zone_pagecache_reclaimable(zone) <= zone->min_unmapped_pages &&
	    zone_page_state(zone, NR_SLAB_RECLAIMABLE) <= zone->min_slab_pages)
2577
		return ZONE_RECLAIM_FULL;
2578

2579
	if (zone_is_all_unreclaimable(zone))
2580
		return ZONE_RECLAIM_FULL;
2581

2582
	/*
2583
	 * Do not scan if the allocation should not be delayed.
2584
	 */
2585
	if (!(gfp_mask & __GFP_WAIT) || (current->flags & PF_MEMALLOC))
2586
		return ZONE_RECLAIM_NOSCAN;
2587 2588 2589 2590 2591 2592 2593

	/*
	 * 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.
	 */
2594
	node_id = zone_to_nid(zone);
2595
	if (node_state(node_id, N_CPU) && node_id != numa_node_id())
2596
		return ZONE_RECLAIM_NOSCAN;
2597 2598

	if (zone_test_and_set_flag(zone, ZONE_RECLAIM_LOCKED))
2599 2600
		return ZONE_RECLAIM_NOSCAN;

2601 2602 2603
	ret = __zone_reclaim(zone, gfp_mask, order);
	zone_clear_flag(zone, ZONE_RECLAIM_LOCKED);

2604 2605 2606
	if (!ret)
		count_vm_event(PGSCAN_ZONE_RECLAIM_FAILED);

2607
	return ret;
2608
}
2609
#endif
L
Lee Schermerhorn 已提交
2610 2611 2612 2613 2614 2615 2616

/*
 * 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
2617 2618
 * 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
2619 2620
 *
 * Reasons page might not be evictable:
2621
 * (1) page's mapping marked unevictable
N
Nick Piggin 已提交
2622
 * (2) page is part of an mlocked VMA
2623
 *
L
Lee Schermerhorn 已提交
2624 2625 2626 2627
 */
int page_evictable(struct page *page, struct vm_area_struct *vma)
{

2628 2629 2630
	if (mapping_unevictable(page_mapping(page)))
		return 0;

N
Nick Piggin 已提交
2631 2632
	if (PageMlocked(page) || (vma && is_mlocked_vma(vma, page)))
		return 0;
L
Lee Schermerhorn 已提交
2633 2634 2635

	return 1;
}
2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655

/**
 * 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)) {
		enum lru_list l = LRU_INACTIVE_ANON + page_is_file_cache(page);
2656

2657 2658
		__dec_zone_state(zone, NR_UNEVICTABLE);
		list_move(&page->lru, &zone->lru[l].list);
K
KAMEZAWA Hiroyuki 已提交
2659
		mem_cgroup_move_lists(page, LRU_UNEVICTABLE, l);
2660 2661 2662 2663 2664 2665 2666 2667
		__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
2668
		mem_cgroup_rotate_lru_list(page, LRU_UNEVICTABLE);
2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727
		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);
	}

}
2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739

/**
 * 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 */
2740
static void scan_zone_unevictable_pages(struct zone *zone)
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 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781
{
	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.
 */
2782
static void scan_all_zones_unevictable_pages(void)
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 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856
{
	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,
			   struct file *file, void __user *buffer,
			   size_t *length, loff_t *ppos)
{
	proc_doulongvec_minmax(table, write, file, buffer, length, ppos);

	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);
}