swap.c 15.2 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7
/*
 *  linux/mm/swap.c
 *
 *  Copyright (C) 1991, 1992, 1993, 1994  Linus Torvalds
 */

/*
S
Simon Arlott 已提交
8
 * This file contains the default values for the operation of the
L
Linus Torvalds 已提交
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
 * Linux VM subsystem. Fine-tuning documentation can be found in
 * Documentation/sysctl/vm.txt.
 * Started 18.12.91
 * Swap aging added 23.2.95, Stephen Tweedie.
 * Buffermem limits added 12.3.98, Rik van Riel.
 */

#include <linux/mm.h>
#include <linux/sched.h>
#include <linux/kernel_stat.h>
#include <linux/swap.h>
#include <linux/mman.h>
#include <linux/pagemap.h>
#include <linux/pagevec.h>
#include <linux/init.h>
#include <linux/module.h>
#include <linux/mm_inline.h>
#include <linux/buffer_head.h>	/* for try_to_release_page() */
#include <linux/percpu_counter.h>
#include <linux/percpu.h>
#include <linux/cpu.h>
#include <linux/notifier.h>
P
Peter Zijlstra 已提交
31
#include <linux/backing-dev.h>
32
#include <linux/memcontrol.h>
33
#include <linux/gfp.h>
L
Linus Torvalds 已提交
34

35 36
#include "internal.h"

L
Linus Torvalds 已提交
37 38 39
/* How many pages do we try to swap or page in/out together? */
int page_cluster;

40
static DEFINE_PER_CPU(struct pagevec[NR_LRU_LISTS], lru_add_pvecs);
41
static DEFINE_PER_CPU(struct pagevec, lru_rotate_pvecs);
42

A
Adrian Bunk 已提交
43 44 45 46
/*
 * This path almost never happens for VM activity - pages are normally
 * freed via pagevecs.  But it gets used by networking.
 */
H
Harvey Harrison 已提交
47
static void __page_cache_release(struct page *page)
A
Adrian Bunk 已提交
48 49 50 51 52 53 54 55 56 57 58
{
	if (PageLRU(page)) {
		unsigned long flags;
		struct zone *zone = page_zone(page);

		spin_lock_irqsave(&zone->lru_lock, flags);
		VM_BUG_ON(!PageLRU(page));
		__ClearPageLRU(page);
		del_page_from_lru(zone, page);
		spin_unlock_irqrestore(&zone->lru_lock, flags);
	}
59 60 61 62 63
}

static void __put_single_page(struct page *page)
{
	__page_cache_release(page);
L
Li Hong 已提交
64
	free_hot_cold_page(page, 0);
A
Adrian Bunk 已提交
65 66
}

67
static void __put_compound_page(struct page *page)
L
Linus Torvalds 已提交
68
{
69
	compound_page_dtor *dtor;
L
Linus Torvalds 已提交
70

71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133
	__page_cache_release(page);
	dtor = get_compound_page_dtor(page);
	(*dtor)(page);
}

static void put_compound_page(struct page *page)
{
	if (unlikely(PageTail(page))) {
		/* __split_huge_page_refcount can run under us */
		struct page *page_head = page->first_page;
		smp_rmb();
		/*
		 * If PageTail is still set after smp_rmb() we can be sure
		 * that the page->first_page we read wasn't a dangling pointer.
		 * See __split_huge_page_refcount() smp_wmb().
		 */
		if (likely(PageTail(page) && get_page_unless_zero(page_head))) {
			unsigned long flags;
			/*
			 * Verify that our page_head wasn't converted
			 * to a a regular page before we got a
			 * reference on it.
			 */
			if (unlikely(!PageHead(page_head))) {
				/* PageHead is cleared after PageTail */
				smp_rmb();
				VM_BUG_ON(PageTail(page));
				goto out_put_head;
			}
			/*
			 * Only run compound_lock on a valid PageHead,
			 * after having it pinned with
			 * get_page_unless_zero() above.
			 */
			smp_mb();
			/* page_head wasn't a dangling pointer */
			flags = compound_lock_irqsave(page_head);
			if (unlikely(!PageTail(page))) {
				/* __split_huge_page_refcount run before us */
				compound_unlock_irqrestore(page_head, flags);
				VM_BUG_ON(PageHead(page_head));
			out_put_head:
				if (put_page_testzero(page_head))
					__put_single_page(page_head);
			out_put_single:
				if (put_page_testzero(page))
					__put_single_page(page);
				return;
			}
			VM_BUG_ON(page_head != page->first_page);
			/*
			 * We can release the refcount taken by
			 * get_page_unless_zero now that
			 * split_huge_page_refcount is blocked on the
			 * compound_lock.
			 */
			if (put_page_testzero(page_head))
				VM_BUG_ON(1);
			/* __split_huge_page_refcount will wait now */
			VM_BUG_ON(atomic_read(&page->_count) <= 0);
			atomic_dec(&page->_count);
			VM_BUG_ON(atomic_read(&page_head->_count) <= 0);
			compound_unlock_irqrestore(page_head, flags);
134 135 136 137 138 139
			if (put_page_testzero(page_head)) {
				if (PageHead(page_head))
					__put_compound_page(page_head);
				else
					__put_single_page(page_head);
			}
140 141 142 143 144 145 146 147 148 149
		} else {
			/* page_head is a dangling pointer */
			VM_BUG_ON(PageTail(page));
			goto out_put_single;
		}
	} else if (put_page_testzero(page)) {
		if (PageHead(page))
			__put_compound_page(page);
		else
			__put_single_page(page);
L
Linus Torvalds 已提交
150
	}
N
Nick Piggin 已提交
151 152 153 154 155 156 157
}

void put_page(struct page *page)
{
	if (unlikely(PageCompound(page)))
		put_compound_page(page);
	else if (put_page_testzero(page))
158
		__put_single_page(page);
L
Linus Torvalds 已提交
159 160 161
}
EXPORT_SYMBOL(put_page);

162
/**
163 164
 * put_pages_list() - release a list of pages
 * @pages: list of pages threaded on page->lru
165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180
 *
 * Release a list of pages which are strung together on page.lru.  Currently
 * used by read_cache_pages() and related error recovery code.
 */
void put_pages_list(struct list_head *pages)
{
	while (!list_empty(pages)) {
		struct page *victim;

		victim = list_entry(pages->prev, struct page, lru);
		list_del(&victim->lru);
		page_cache_release(victim);
	}
}
EXPORT_SYMBOL(put_pages_list);

181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200
/*
 * pagevec_move_tail() must be called with IRQ disabled.
 * Otherwise this may cause nasty races.
 */
static void pagevec_move_tail(struct pagevec *pvec)
{
	int i;
	int pgmoved = 0;
	struct zone *zone = NULL;

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

		if (pagezone != zone) {
			if (zone)
				spin_unlock(&zone->lru_lock);
			zone = pagezone;
			spin_lock(&zone->lru_lock);
		}
L
Lee Schermerhorn 已提交
201
		if (PageLRU(page) && !PageActive(page) && !PageUnevictable(page)) {
202
			int lru = page_lru_base_type(page);
203
			list_move_tail(&page->lru, &zone->lru[lru].list);
204 205 206 207 208 209 210 211 212 213
			pgmoved++;
		}
	}
	if (zone)
		spin_unlock(&zone->lru_lock);
	__count_vm_events(PGROTATED, pgmoved);
	release_pages(pvec->pages, pvec->nr, pvec->cold);
	pagevec_reinit(pvec);
}

L
Linus Torvalds 已提交
214 215 216
/*
 * Writeback is about to end against a page which has been marked for immediate
 * reclaim.  If it still appears to be reclaimable, move it to the tail of the
217
 * inactive list.
L
Linus Torvalds 已提交
218
 */
219
void  rotate_reclaimable_page(struct page *page)
L
Linus Torvalds 已提交
220
{
221
	if (!PageLocked(page) && !PageDirty(page) && !PageActive(page) &&
L
Lee Schermerhorn 已提交
222
	    !PageUnevictable(page) && PageLRU(page)) {
223 224 225 226 227 228 229 230 231 232
		struct pagevec *pvec;
		unsigned long flags;

		page_cache_get(page);
		local_irq_save(flags);
		pvec = &__get_cpu_var(lru_rotate_pvecs);
		if (!pagevec_add(pvec, page))
			pagevec_move_tail(pvec);
		local_irq_restore(flags);
	}
L
Linus Torvalds 已提交
233 234
}

K
KOSAKI Motohiro 已提交
235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254
static void update_page_reclaim_stat(struct zone *zone, struct page *page,
				     int file, int rotated)
{
	struct zone_reclaim_stat *reclaim_stat = &zone->reclaim_stat;
	struct zone_reclaim_stat *memcg_reclaim_stat;

	memcg_reclaim_stat = mem_cgroup_get_reclaim_stat_from_page(page);

	reclaim_stat->recent_scanned[file]++;
	if (rotated)
		reclaim_stat->recent_rotated[file]++;

	if (!memcg_reclaim_stat)
		return;

	memcg_reclaim_stat->recent_scanned[file]++;
	if (rotated)
		memcg_reclaim_stat->recent_rotated[file]++;
}

L
Linus Torvalds 已提交
255 256 257
/*
 * FIXME: speed this up?
 */
H
Harvey Harrison 已提交
258
void activate_page(struct page *page)
L
Linus Torvalds 已提交
259 260 261 262
{
	struct zone *zone = page_zone(page);

	spin_lock_irq(&zone->lru_lock);
L
Lee Schermerhorn 已提交
263
	if (PageLRU(page) && !PageActive(page) && !PageUnevictable(page)) {
264
		int file = page_is_file_cache(page);
265
		int lru = page_lru_base_type(page);
266 267
		del_page_from_lru_list(zone, page, lru);

L
Linus Torvalds 已提交
268
		SetPageActive(page);
269 270
		lru += LRU_ACTIVE;
		add_page_to_lru_list(zone, page, lru);
271
		__count_vm_event(PGACTIVATE);
272

273
		update_page_reclaim_stat(zone, page, file, 1);
L
Linus Torvalds 已提交
274 275 276 277 278 279 280 281 282 283 284
	}
	spin_unlock_irq(&zone->lru_lock);
}

/*
 * Mark a page as having seen activity.
 *
 * inactive,unreferenced	->	inactive,referenced
 * inactive,referenced		->	active,unreferenced
 * active,unreferenced		->	active,referenced
 */
H
Harvey Harrison 已提交
285
void mark_page_accessed(struct page *page)
L
Linus Torvalds 已提交
286
{
L
Lee Schermerhorn 已提交
287 288
	if (!PageActive(page) && !PageUnevictable(page) &&
			PageReferenced(page) && PageLRU(page)) {
L
Linus Torvalds 已提交
289 290 291 292 293 294 295 296 297
		activate_page(page);
		ClearPageReferenced(page);
	} else if (!PageReferenced(page)) {
		SetPageReferenced(page);
	}
}

EXPORT_SYMBOL(mark_page_accessed);

298
void __lru_cache_add(struct page *page, enum lru_list lru)
L
Linus Torvalds 已提交
299
{
300
	struct pagevec *pvec = &get_cpu_var(lru_add_pvecs)[lru];
L
Linus Torvalds 已提交
301 302 303

	page_cache_get(page);
	if (!pagevec_add(pvec, page))
304
		____pagevec_lru_add(pvec, lru);
L
Linus Torvalds 已提交
305 306
	put_cpu_var(lru_add_pvecs);
}
307
EXPORT_SYMBOL(__lru_cache_add);
L
Linus Torvalds 已提交
308

309 310 311 312 313 314
/**
 * lru_cache_add_lru - add a page to a page list
 * @page: the page to be added to the LRU.
 * @lru: the LRU list to which the page is added.
 */
void lru_cache_add_lru(struct page *page, enum lru_list lru)
L
Linus Torvalds 已提交
315
{
316
	if (PageActive(page)) {
L
Lee Schermerhorn 已提交
317
		VM_BUG_ON(PageUnevictable(page));
318
		ClearPageActive(page);
L
Lee Schermerhorn 已提交
319 320 321
	} else if (PageUnevictable(page)) {
		VM_BUG_ON(PageActive(page));
		ClearPageUnevictable(page);
322
	}
L
Linus Torvalds 已提交
323

L
Lee Schermerhorn 已提交
324
	VM_BUG_ON(PageLRU(page) || PageActive(page) || PageUnevictable(page));
325
	__lru_cache_add(page, lru);
L
Linus Torvalds 已提交
326 327
}

L
Lee Schermerhorn 已提交
328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348
/**
 * add_page_to_unevictable_list - add a page to the unevictable list
 * @page:  the page to be added to the unevictable list
 *
 * Add page directly to its zone's unevictable list.  To avoid races with
 * tasks that might be making the page evictable, through eg. munlock,
 * munmap or exit, while it's not on the lru, we want to add the page
 * while it's locked or otherwise "invisible" to other tasks.  This is
 * difficult to do when using the pagevec cache, so bypass that.
 */
void add_page_to_unevictable_list(struct page *page)
{
	struct zone *zone = page_zone(page);

	spin_lock_irq(&zone->lru_lock);
	SetPageUnevictable(page);
	SetPageLRU(page);
	add_page_to_lru_list(zone, page, LRU_UNEVICTABLE);
	spin_unlock_irq(&zone->lru_lock);
}

349 350 351 352 353 354
/*
 * Drain pages out of the cpu's pagevecs.
 * Either "cpu" is the current CPU, and preemption has already been
 * disabled; or "cpu" is being hot-unplugged, and is already dead.
 */
static void drain_cpu_pagevecs(int cpu)
L
Linus Torvalds 已提交
355
{
356
	struct pagevec *pvecs = per_cpu(lru_add_pvecs, cpu);
357
	struct pagevec *pvec;
358
	int lru;
L
Linus Torvalds 已提交
359

360 361 362 363 364
	for_each_lru(lru) {
		pvec = &pvecs[lru - LRU_BASE];
		if (pagevec_count(pvec))
			____pagevec_lru_add(pvec, lru);
	}
365 366 367 368 369 370 371 372 373 374

	pvec = &per_cpu(lru_rotate_pvecs, cpu);
	if (pagevec_count(pvec)) {
		unsigned long flags;

		/* No harm done if a racing interrupt already did this */
		local_irq_save(flags);
		pagevec_move_tail(pvec);
		local_irq_restore(flags);
	}
375 376 377 378
}

void lru_add_drain(void)
{
379
	drain_cpu_pagevecs(get_cpu());
380
	put_cpu();
L
Linus Torvalds 已提交
381 382
}

D
David Howells 已提交
383
static void lru_add_drain_per_cpu(struct work_struct *dummy)
384 385 386 387 388 389 390 391 392
{
	lru_add_drain();
}

/*
 * Returns 0 for success
 */
int lru_add_drain_all(void)
{
D
David Howells 已提交
393
	return schedule_on_each_cpu(lru_add_drain_per_cpu);
394 395
}

L
Linus Torvalds 已提交
396 397 398 399 400 401 402 403
/*
 * Batched page_cache_release().  Decrement the reference count on all the
 * passed pages.  If it fell to zero then remove the page from the LRU and
 * free it.
 *
 * Avoid taking zone->lru_lock if possible, but if it is taken, retain it
 * for the remainder of the operation.
 *
404 405 406 407
 * The locking in this function is against shrink_inactive_list(): we recheck
 * the page count inside the lock to see whether shrink_inactive_list()
 * grabbed the page via the LRU.  If it did, give up: shrink_inactive_list()
 * will free it.
L
Linus Torvalds 已提交
408 409 410 411 412 413
 */
void release_pages(struct page **pages, int nr, int cold)
{
	int i;
	struct pagevec pages_to_free;
	struct zone *zone = NULL;
414
	unsigned long uninitialized_var(flags);
L
Linus Torvalds 已提交
415 416 417 418 419

	pagevec_init(&pages_to_free, cold);
	for (i = 0; i < nr; i++) {
		struct page *page = pages[i];

N
Nick Piggin 已提交
420 421
		if (unlikely(PageCompound(page))) {
			if (zone) {
422
				spin_unlock_irqrestore(&zone->lru_lock, flags);
N
Nick Piggin 已提交
423 424 425 426 427 428
				zone = NULL;
			}
			put_compound_page(page);
			continue;
		}

N
Nick Piggin 已提交
429
		if (!put_page_testzero(page))
L
Linus Torvalds 已提交
430 431
			continue;

432 433
		if (PageLRU(page)) {
			struct zone *pagezone = page_zone(page);
L
Lee Schermerhorn 已提交
434

435 436
			if (pagezone != zone) {
				if (zone)
437 438
					spin_unlock_irqrestore(&zone->lru_lock,
									flags);
439
				zone = pagezone;
440
				spin_lock_irqsave(&zone->lru_lock, flags);
441
			}
N
Nick Piggin 已提交
442
			VM_BUG_ON(!PageLRU(page));
N
Nick Piggin 已提交
443
			__ClearPageLRU(page);
L
Linus Torvalds 已提交
444
			del_page_from_lru(zone, page);
445 446 447 448
		}

		if (!pagevec_add(&pages_to_free, page)) {
			if (zone) {
449
				spin_unlock_irqrestore(&zone->lru_lock, flags);
450
				zone = NULL;
L
Linus Torvalds 已提交
451
			}
452 453 454
			__pagevec_free(&pages_to_free);
			pagevec_reinit(&pages_to_free);
  		}
L
Linus Torvalds 已提交
455 456
	}
	if (zone)
457
		spin_unlock_irqrestore(&zone->lru_lock, flags);
L
Linus Torvalds 已提交
458 459 460

	pagevec_free(&pages_to_free);
}
M
Miklos Szeredi 已提交
461
EXPORT_SYMBOL(release_pages);
L
Linus Torvalds 已提交
462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479

/*
 * The pages which we're about to release may be in the deferred lru-addition
 * queues.  That would prevent them from really being freed right now.  That's
 * OK from a correctness point of view but is inefficient - those pages may be
 * cache-warm and we want to give them back to the page allocator ASAP.
 *
 * So __pagevec_release() will drain those queues here.  __pagevec_lru_add()
 * and __pagevec_lru_add_active() call release_pages() directly to avoid
 * mutual recursion.
 */
void __pagevec_release(struct pagevec *pvec)
{
	lru_add_drain();
	release_pages(pvec->pages, pagevec_count(pvec), pvec->cold);
	pagevec_reinit(pvec);
}

480 481
EXPORT_SYMBOL(__pagevec_release);

L
Linus Torvalds 已提交
482 483 484 485
/*
 * Add the passed pages to the LRU, then drop the caller's refcount
 * on them.  Reinitialises the caller's pagevec.
 */
486
void ____pagevec_lru_add(struct pagevec *pvec, enum lru_list lru)
L
Linus Torvalds 已提交
487 488 489
{
	int i;
	struct zone *zone = NULL;
490

L
Lee Schermerhorn 已提交
491
	VM_BUG_ON(is_unevictable_lru(lru));
L
Linus Torvalds 已提交
492 493 494 495

	for (i = 0; i < pagevec_count(pvec); i++) {
		struct page *page = pvec->pages[i];
		struct zone *pagezone = page_zone(page);
R
Rik van Riel 已提交
496
		int file;
K
KOSAKI Motohiro 已提交
497
		int active;
L
Linus Torvalds 已提交
498 499 500 501 502 503 504

		if (pagezone != zone) {
			if (zone)
				spin_unlock_irq(&zone->lru_lock);
			zone = pagezone;
			spin_lock_irq(&zone->lru_lock);
		}
L
Lee Schermerhorn 已提交
505 506
		VM_BUG_ON(PageActive(page));
		VM_BUG_ON(PageUnevictable(page));
N
Nick Piggin 已提交
507
		VM_BUG_ON(PageLRU(page));
N
Nick Piggin 已提交
508
		SetPageLRU(page);
K
KOSAKI Motohiro 已提交
509
		active = is_active_lru(lru);
R
Rik van Riel 已提交
510
		file = is_file_lru(lru);
K
KOSAKI Motohiro 已提交
511
		if (active)
512
			SetPageActive(page);
K
KOSAKI Motohiro 已提交
513
		update_page_reclaim_stat(zone, page, file, active);
514
		add_page_to_lru_list(zone, page, lru);
L
Linus Torvalds 已提交
515 516 517 518 519 520 521
	}
	if (zone)
		spin_unlock_irq(&zone->lru_lock);
	release_pages(pvec->pages, pvec->nr, pvec->cold);
	pagevec_reinit(pvec);
}

522
EXPORT_SYMBOL(____pagevec_lru_add);
L
Linus Torvalds 已提交
523 524 525 526 527 528 529 530 531 532 533

/*
 * Try to drop buffers from the pages in a pagevec
 */
void pagevec_strip(struct pagevec *pvec)
{
	int i;

	for (i = 0; i < pagevec_count(pvec); i++) {
		struct page *page = pvec->pages[i];

534 535
		if (page_has_private(page) && trylock_page(page)) {
			if (page_has_private(page))
536
				try_to_release_page(page, 0);
L
Linus Torvalds 已提交
537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564
			unlock_page(page);
		}
	}
}

/**
 * pagevec_lookup - gang pagecache lookup
 * @pvec:	Where the resulting pages are placed
 * @mapping:	The address_space to search
 * @start:	The starting page index
 * @nr_pages:	The maximum number of pages
 *
 * pagevec_lookup() will search for and return a group of up to @nr_pages pages
 * in the mapping.  The pages are placed in @pvec.  pagevec_lookup() takes a
 * reference against the pages in @pvec.
 *
 * The search returns a group of mapping-contiguous pages with ascending
 * indexes.  There may be holes in the indices due to not-present pages.
 *
 * pagevec_lookup() returns the number of pages which were found.
 */
unsigned pagevec_lookup(struct pagevec *pvec, struct address_space *mapping,
		pgoff_t start, unsigned nr_pages)
{
	pvec->nr = find_get_pages(mapping, start, nr_pages, pvec->pages);
	return pagevec_count(pvec);
}

565 566
EXPORT_SYMBOL(pagevec_lookup);

L
Linus Torvalds 已提交
567 568 569 570 571 572 573 574
unsigned pagevec_lookup_tag(struct pagevec *pvec, struct address_space *mapping,
		pgoff_t *index, int tag, unsigned nr_pages)
{
	pvec->nr = find_get_pages_tag(mapping, index, tag,
					nr_pages, pvec->pages);
	return pagevec_count(pvec);
}

575
EXPORT_SYMBOL(pagevec_lookup_tag);
L
Linus Torvalds 已提交
576 577 578 579 580 581

/*
 * Perform any setup for the swap system
 */
void __init swap_setup(void)
{
582
	unsigned long megs = totalram_pages >> (20 - PAGE_SHIFT);
L
Linus Torvalds 已提交
583

P
Peter Zijlstra 已提交
584 585 586 587
#ifdef CONFIG_SWAP
	bdi_init(swapper_space.backing_dev_info);
#endif

L
Linus Torvalds 已提交
588 589 590 591 592 593 594 595 596 597
	/* Use a smaller cluster for small-memory machines */
	if (megs < 16)
		page_cluster = 2;
	else
		page_cluster = 3;
	/*
	 * Right now other parts of the system means that we
	 * _really_ don't want to cluster much more
	 */
}