swap.c 27.0 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
 * 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>
24
#include <linux/export.h>
L
Linus Torvalds 已提交
25 26
#include <linux/mm_inline.h>
#include <linux/percpu_counter.h>
27
#include <linux/memremap.h>
L
Linus Torvalds 已提交
28 29 30
#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>
34
#include <linux/uio.h>
35
#include <linux/hugetlb.h>
36
#include <linux/page_idle.h>
L
Linus Torvalds 已提交
37

38 39
#include "internal.h"

40 41 42
#define CREATE_TRACE_POINTS
#include <trace/events/pagemap.h>

L
Linus Torvalds 已提交
43 44 45
/* How many pages do we try to swap or page in/out together? */
int page_cluster;

46
static DEFINE_PER_CPU(struct pagevec, lru_add_pvec);
47
static DEFINE_PER_CPU(struct pagevec, lru_rotate_pvecs);
48
static DEFINE_PER_CPU(struct pagevec, lru_deactivate_file_pvecs);
49
static DEFINE_PER_CPU(struct pagevec, lru_deactivate_pvecs);
50 51 52
#ifdef CONFIG_SMP
static DEFINE_PER_CPU(struct pagevec, activate_page_pvecs);
#endif
53

A
Adrian Bunk 已提交
54 55 56 57
/*
 * 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
58
static void __page_cache_release(struct page *page)
A
Adrian Bunk 已提交
59 60 61
{
	if (PageLRU(page)) {
		struct zone *zone = page_zone(page);
62 63
		struct lruvec *lruvec;
		unsigned long flags;
A
Adrian Bunk 已提交
64

65
		spin_lock_irqsave(zone_lru_lock(zone), flags);
M
Mel Gorman 已提交
66
		lruvec = mem_cgroup_page_lruvec(page, zone->zone_pgdat);
67
		VM_BUG_ON_PAGE(!PageLRU(page), page);
A
Adrian Bunk 已提交
68
		__ClearPageLRU(page);
69
		del_page_from_lru_list(page, lruvec, page_off_lru(page));
70
		spin_unlock_irqrestore(zone_lru_lock(zone), flags);
A
Adrian Bunk 已提交
71
	}
72
	__ClearPageWaiters(page);
73
	mem_cgroup_uncharge(page);
74 75 76 77 78
}

static void __put_single_page(struct page *page)
{
	__page_cache_release(page);
79
	free_hot_cold_page(page, false);
A
Adrian Bunk 已提交
80 81
}

82
static void __put_compound_page(struct page *page)
L
Linus Torvalds 已提交
83
{
84
	compound_page_dtor *dtor;
L
Linus Torvalds 已提交
85

86 87 88 89 90 91 92 93
	/*
	 * __page_cache_release() is supposed to be called for thp, not for
	 * hugetlb. This is because hugetlb page does never have PageLRU set
	 * (it's never listed to any LRU lists) and no memcg routines should
	 * be called for hugetlb (it has a separate hugetlb_cgroup.)
	 */
	if (!PageHuge(page))
		__page_cache_release(page);
94 95 96 97
	dtor = get_compound_page_dtor(page);
	(*dtor)(page);
}

98
void __put_page(struct page *page)
N
Nick Piggin 已提交
99 100
{
	if (unlikely(PageCompound(page)))
101 102
		__put_compound_page(page);
	else
103
		__put_single_page(page);
L
Linus Torvalds 已提交
104
}
105
EXPORT_SYMBOL(__put_page);
106

107
/**
108 109
 * put_pages_list() - release a list of pages
 * @pages: list of pages threaded on page->lru
110 111 112 113 114 115 116 117 118 119 120
 *
 * 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);
121
		put_page(victim);
122 123 124 125
	}
}
EXPORT_SYMBOL(put_pages_list);

126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147
/*
 * get_kernel_pages() - pin kernel pages in memory
 * @kiov:	An array of struct kvec structures
 * @nr_segs:	number of segments to pin
 * @write:	pinning for read/write, currently ignored
 * @pages:	array that receives pointers to the pages pinned.
 *		Should be at least nr_segs long.
 *
 * Returns number of pages pinned. This may be fewer than the number
 * requested. If nr_pages is 0 or negative, returns 0. If no pages
 * were pinned, returns -errno. Each page returned must be released
 * with a put_page() call when it is finished with.
 */
int get_kernel_pages(const struct kvec *kiov, int nr_segs, int write,
		struct page **pages)
{
	int seg;

	for (seg = 0; seg < nr_segs; seg++) {
		if (WARN_ON(kiov[seg].iov_len != PAGE_SIZE))
			return seg;

148
		pages[seg] = kmap_to_page(kiov[seg].iov_base);
149
		get_page(pages[seg]);
150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177
	}

	return seg;
}
EXPORT_SYMBOL_GPL(get_kernel_pages);

/*
 * get_kernel_page() - pin a kernel page in memory
 * @start:	starting kernel address
 * @write:	pinning for read/write, currently ignored
 * @pages:	array that receives pointer to the page pinned.
 *		Must be at least nr_segs long.
 *
 * Returns 1 if page is pinned. If the page was not pinned, returns
 * -errno. The page returned must be released with a put_page() call
 * when it is finished with.
 */
int get_kernel_page(unsigned long start, int write, struct page **pages)
{
	const struct kvec kiov = {
		.iov_base = (void *)start,
		.iov_len = PAGE_SIZE
	};

	return get_kernel_pages(&kiov, 1, write, pages);
}
EXPORT_SYMBOL_GPL(get_kernel_page);

S
Shaohua Li 已提交
178
static void pagevec_lru_move_fn(struct pagevec *pvec,
179 180
	void (*move_fn)(struct page *page, struct lruvec *lruvec, void *arg),
	void *arg)
181 182
{
	int i;
183
	struct pglist_data *pgdat = NULL;
184
	struct lruvec *lruvec;
S
Shaohua Li 已提交
185
	unsigned long flags = 0;
186 187 188

	for (i = 0; i < pagevec_count(pvec); i++) {
		struct page *page = pvec->pages[i];
189
		struct pglist_data *pagepgdat = page_pgdat(page);
190

191 192 193 194 195
		if (pagepgdat != pgdat) {
			if (pgdat)
				spin_unlock_irqrestore(&pgdat->lru_lock, flags);
			pgdat = pagepgdat;
			spin_lock_irqsave(&pgdat->lru_lock, flags);
196
		}
S
Shaohua Li 已提交
197

198
		lruvec = mem_cgroup_page_lruvec(page, pgdat);
199
		(*move_fn)(page, lruvec, arg);
200
	}
201 202
	if (pgdat)
		spin_unlock_irqrestore(&pgdat->lru_lock, flags);
203 204
	release_pages(pvec->pages, pvec->nr, pvec->cold);
	pagevec_reinit(pvec);
S
Shaohua Li 已提交
205 206
}

207 208
static void pagevec_move_tail_fn(struct page *page, struct lruvec *lruvec,
				 void *arg)
S
Shaohua Li 已提交
209 210 211 212 213
{
	int *pgmoved = arg;

	if (PageLRU(page) && !PageActive(page) && !PageUnevictable(page)) {
		enum lru_list lru = page_lru_base_type(page);
214
		list_move_tail(&page->lru, &lruvec->lists[lru]);
S
Shaohua Li 已提交
215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230
		(*pgmoved)++;
	}
}

/*
 * pagevec_move_tail() must be called with IRQ disabled.
 * Otherwise this may cause nasty races.
 */
static void pagevec_move_tail(struct pagevec *pvec)
{
	int pgmoved = 0;

	pagevec_lru_move_fn(pvec, pagevec_move_tail_fn, &pgmoved);
	__count_vm_events(PGROTATED, pgmoved);
}

L
Linus Torvalds 已提交
231 232 233
/*
 * 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
234
 * inactive list.
L
Linus Torvalds 已提交
235
 */
S
Shaohua Li 已提交
236
void rotate_reclaimable_page(struct page *page)
L
Linus Torvalds 已提交
237
{
238
	if (!PageLocked(page) && !PageDirty(page) && !PageActive(page) &&
L
Lee Schermerhorn 已提交
239
	    !PageUnevictable(page) && PageLRU(page)) {
240 241 242
		struct pagevec *pvec;
		unsigned long flags;

243
		get_page(page);
244
		local_irq_save(flags);
245
		pvec = this_cpu_ptr(&lru_rotate_pvecs);
246
		if (!pagevec_add(pvec, page) || PageCompound(page))
247 248 249
			pagevec_move_tail(pvec);
		local_irq_restore(flags);
	}
L
Linus Torvalds 已提交
250 251
}

252
static void update_page_reclaim_stat(struct lruvec *lruvec,
K
KOSAKI Motohiro 已提交
253 254
				     int file, int rotated)
{
255
	struct zone_reclaim_stat *reclaim_stat = &lruvec->reclaim_stat;
K
KOSAKI Motohiro 已提交
256 257 258 259 260 261

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

262 263
static void __activate_page(struct page *page, struct lruvec *lruvec,
			    void *arg)
L
Linus Torvalds 已提交
264
{
265
	if (PageLRU(page) && !PageActive(page) && !PageUnevictable(page)) {
266 267
		int file = page_is_file_cache(page);
		int lru = page_lru_base_type(page);
268

269
		del_page_from_lru_list(page, lruvec, lru);
270 271
		SetPageActive(page);
		lru += LRU_ACTIVE;
272
		add_page_to_lru_list(page, lruvec, lru);
273
		trace_mm_lru_activate(page);
274

275 276
		__count_vm_event(PGACTIVATE);
		update_page_reclaim_stat(lruvec, file, 1);
L
Linus Torvalds 已提交
277
	}
278 279 280 281 282 283 284 285 286 287 288
}

#ifdef CONFIG_SMP
static void activate_page_drain(int cpu)
{
	struct pagevec *pvec = &per_cpu(activate_page_pvecs, cpu);

	if (pagevec_count(pvec))
		pagevec_lru_move_fn(pvec, __activate_page, NULL);
}

289 290 291 292 293
static bool need_activate_page_drain(int cpu)
{
	return pagevec_count(&per_cpu(activate_page_pvecs, cpu)) != 0;
}

294 295
void activate_page(struct page *page)
{
296
	page = compound_head(page);
297 298 299
	if (PageLRU(page) && !PageActive(page) && !PageUnevictable(page)) {
		struct pagevec *pvec = &get_cpu_var(activate_page_pvecs);

300
		get_page(page);
301
		if (!pagevec_add(pvec, page) || PageCompound(page))
302 303 304 305 306 307 308 309 310 311
			pagevec_lru_move_fn(pvec, __activate_page, NULL);
		put_cpu_var(activate_page_pvecs);
	}
}

#else
static inline void activate_page_drain(int cpu)
{
}

312 313 314 315 316
static bool need_activate_page_drain(int cpu)
{
	return false;
}

317 318 319 320
void activate_page(struct page *page)
{
	struct zone *zone = page_zone(page);

321
	page = compound_head(page);
322
	spin_lock_irq(zone_lru_lock(zone));
M
Mel Gorman 已提交
323
	__activate_page(page, mem_cgroup_page_lruvec(page, zone->zone_pgdat), NULL);
324
	spin_unlock_irq(zone_lru_lock(zone));
L
Linus Torvalds 已提交
325
}
326
#endif
L
Linus Torvalds 已提交
327

328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354
static void __lru_cache_activate_page(struct page *page)
{
	struct pagevec *pvec = &get_cpu_var(lru_add_pvec);
	int i;

	/*
	 * Search backwards on the optimistic assumption that the page being
	 * activated has just been added to this pagevec. Note that only
	 * the local pagevec is examined as a !PageLRU page could be in the
	 * process of being released, reclaimed, migrated or on a remote
	 * pagevec that is currently being drained. Furthermore, marking
	 * a remote pagevec's page PageActive potentially hits a race where
	 * a page is marked PageActive just after it is added to the inactive
	 * list causing accounting errors and BUG_ON checks to trigger.
	 */
	for (i = pagevec_count(pvec) - 1; i >= 0; i--) {
		struct page *pagevec_page = pvec->pages[i];

		if (pagevec_page == page) {
			SetPageActive(page);
			break;
		}
	}

	put_cpu_var(lru_add_pvec);
}

L
Linus Torvalds 已提交
355 356 357 358 359 360
/*
 * Mark a page as having seen activity.
 *
 * inactive,unreferenced	->	inactive,referenced
 * inactive,referenced		->	active,unreferenced
 * active,unreferenced		->	active,referenced
361 362 363
 *
 * When a newly allocated page is not yet visible, so safe for non-atomic ops,
 * __SetPageReferenced(page) may be substituted for mark_page_accessed(page).
L
Linus Torvalds 已提交
364
 */
H
Harvey Harrison 已提交
365
void mark_page_accessed(struct page *page)
L
Linus Torvalds 已提交
366
{
367
	page = compound_head(page);
L
Lee Schermerhorn 已提交
368
	if (!PageActive(page) && !PageUnevictable(page) &&
369 370 371 372 373 374 375 376 377 378 379 380
			PageReferenced(page)) {

		/*
		 * If the page is on the LRU, queue it for activation via
		 * activate_page_pvecs. Otherwise, assume the page is on a
		 * pagevec, mark it active and it'll be moved to the active
		 * LRU on the next drain.
		 */
		if (PageLRU(page))
			activate_page(page);
		else
			__lru_cache_activate_page(page);
L
Linus Torvalds 已提交
381
		ClearPageReferenced(page);
382 383
		if (page_is_file_cache(page))
			workingset_activation(page);
L
Linus Torvalds 已提交
384 385 386
	} else if (!PageReferenced(page)) {
		SetPageReferenced(page);
	}
387 388
	if (page_is_idle(page))
		clear_page_idle(page);
L
Linus Torvalds 已提交
389 390 391
}
EXPORT_SYMBOL(mark_page_accessed);

392
static void __lru_cache_add(struct page *page)
L
Linus Torvalds 已提交
393
{
394 395
	struct pagevec *pvec = &get_cpu_var(lru_add_pvec);

396
	get_page(page);
397
	if (!pagevec_add(pvec, page) || PageCompound(page))
398
		__pagevec_lru_add(pvec);
399
	put_cpu_var(lru_add_pvec);
L
Linus Torvalds 已提交
400
}
401 402 403 404 405 406 407

/**
 * lru_cache_add: add a page to the page lists
 * @page: the page to add
 */
void lru_cache_add_anon(struct page *page)
{
408 409
	if (PageActive(page))
		ClearPageActive(page);
410 411 412 413 414
	__lru_cache_add(page);
}

void lru_cache_add_file(struct page *page)
{
415 416
	if (PageActive(page))
		ClearPageActive(page);
417 418 419
	__lru_cache_add(page);
}
EXPORT_SYMBOL(lru_cache_add_file);
L
Linus Torvalds 已提交
420

421
/**
422
 * lru_cache_add - add a page to a page list
423
 * @page: the page to be added to the LRU.
424 425 426 427 428
 *
 * Queue the page for addition to the LRU via pagevec. The decision on whether
 * to add the page to the [in]active [file|anon] list is deferred until the
 * pagevec is drained. This gives a chance for the caller of lru_cache_add()
 * have the page added to the active list using mark_page_accessed().
429
 */
430
void lru_cache_add(struct page *page)
L
Linus Torvalds 已提交
431
{
432 433
	VM_BUG_ON_PAGE(PageActive(page) && PageUnevictable(page), page);
	VM_BUG_ON_PAGE(PageLRU(page), page);
434
	__lru_cache_add(page);
L
Linus Torvalds 已提交
435 436
}

L
Lee Schermerhorn 已提交
437 438 439 440 441 442 443 444 445 446 447 448
/**
 * 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)
{
M
Mel Gorman 已提交
449
	struct pglist_data *pgdat = page_pgdat(page);
450
	struct lruvec *lruvec;
L
Lee Schermerhorn 已提交
451

M
Mel Gorman 已提交
452 453
	spin_lock_irq(&pgdat->lru_lock);
	lruvec = mem_cgroup_page_lruvec(page, pgdat);
454
	ClearPageActive(page);
L
Lee Schermerhorn 已提交
455 456
	SetPageUnevictable(page);
	SetPageLRU(page);
457
	add_page_to_lru_list(page, lruvec, LRU_UNEVICTABLE);
M
Mel Gorman 已提交
458
	spin_unlock_irq(&pgdat->lru_lock);
L
Lee Schermerhorn 已提交
459 460
}

461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494
/**
 * lru_cache_add_active_or_unevictable
 * @page:  the page to be added to LRU
 * @vma:   vma in which page is mapped for determining reclaimability
 *
 * Place @page on the active or unevictable LRU list, depending on its
 * evictability.  Note that if the page is not evictable, it goes
 * directly back onto it's zone's unevictable list, it does NOT use a
 * per cpu pagevec.
 */
void lru_cache_add_active_or_unevictable(struct page *page,
					 struct vm_area_struct *vma)
{
	VM_BUG_ON_PAGE(PageLRU(page), page);

	if (likely((vma->vm_flags & (VM_LOCKED | VM_SPECIAL)) != VM_LOCKED)) {
		SetPageActive(page);
		lru_cache_add(page);
		return;
	}

	if (!TestSetPageMlocked(page)) {
		/*
		 * We use the irq-unsafe __mod_zone_page_stat because this
		 * counter is not modified from interrupt context, and the pte
		 * lock is held(spinlock), which implies preemption disabled.
		 */
		__mod_zone_page_state(page_zone(page), NR_MLOCK,
				    hpage_nr_pages(page));
		count_vm_event(UNEVICTABLE_PGMLOCKED);
	}
	add_page_to_unevictable_list(page);
}

M
Minchan Kim 已提交
495 496 497 498 499 500
/*
 * If the page can not be invalidated, it is moved to the
 * inactive list to speed up its reclaim.  It is moved to the
 * head of the list, rather than the tail, to give the flusher
 * threads some time to write it out, as this is much more
 * effective than the single-page writeout from reclaim.
M
Minchan Kim 已提交
501 502 503 504 505 506 507 508 509 510 511 512 513 514
 *
 * If the page isn't page_mapped and dirty/writeback, the page
 * could reclaim asap using PG_reclaim.
 *
 * 1. active, mapped page -> none
 * 2. active, dirty/writeback page -> inactive, head, PG_reclaim
 * 3. inactive, mapped page -> none
 * 4. inactive, dirty/writeback page -> inactive, head, PG_reclaim
 * 5. inactive, clean -> inactive, tail
 * 6. Others -> none
 *
 * In 4, why it moves inactive's head, the VM expects the page would
 * be write it out by flusher threads as this is much more effective
 * than the single-page writeout from reclaim.
M
Minchan Kim 已提交
515
 */
516
static void lru_deactivate_file_fn(struct page *page, struct lruvec *lruvec,
517
			      void *arg)
M
Minchan Kim 已提交
518 519
{
	int lru, file;
M
Minchan Kim 已提交
520
	bool active;
M
Minchan Kim 已提交
521

M
Minchan Kim 已提交
522
	if (!PageLRU(page))
M
Minchan Kim 已提交
523 524
		return;

525 526 527
	if (PageUnevictable(page))
		return;

M
Minchan Kim 已提交
528 529 530 531
	/* Some processes are using the page */
	if (page_mapped(page))
		return;

M
Minchan Kim 已提交
532
	active = PageActive(page);
M
Minchan Kim 已提交
533 534
	file = page_is_file_cache(page);
	lru = page_lru_base_type(page);
535 536

	del_page_from_lru_list(page, lruvec, lru + active);
M
Minchan Kim 已提交
537 538
	ClearPageActive(page);
	ClearPageReferenced(page);
539
	add_page_to_lru_list(page, lruvec, lru);
M
Minchan Kim 已提交
540

M
Minchan Kim 已提交
541 542 543 544 545 546 547 548 549 550 551 552
	if (PageWriteback(page) || PageDirty(page)) {
		/*
		 * PG_reclaim could be raced with end_page_writeback
		 * It can make readahead confusing.  But race window
		 * is _really_ small and  it's non-critical problem.
		 */
		SetPageReclaim(page);
	} else {
		/*
		 * The page's writeback ends up during pagevec
		 * We moves tha page into tail of inactive.
		 */
553
		list_move_tail(&page->lru, &lruvec->lists[lru]);
M
Minchan Kim 已提交
554 555 556 557 558
		__count_vm_event(PGROTATED);
	}

	if (active)
		__count_vm_event(PGDEACTIVATE);
559
	update_page_reclaim_stat(lruvec, file, 0);
M
Minchan Kim 已提交
560 561
}

562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579

static void lru_deactivate_fn(struct page *page, struct lruvec *lruvec,
			    void *arg)
{
	if (PageLRU(page) && PageActive(page) && !PageUnevictable(page)) {
		int file = page_is_file_cache(page);
		int lru = page_lru_base_type(page);

		del_page_from_lru_list(page, lruvec, lru + LRU_ACTIVE);
		ClearPageActive(page);
		ClearPageReferenced(page);
		add_page_to_lru_list(page, lruvec, lru);

		__count_vm_event(PGDEACTIVATE);
		update_page_reclaim_stat(lruvec, file, 0);
	}
}

580 581 582 583 584
/*
 * 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.
 */
585
void lru_add_drain_cpu(int cpu)
L
Linus Torvalds 已提交
586
{
587
	struct pagevec *pvec = &per_cpu(lru_add_pvec, cpu);
L
Linus Torvalds 已提交
588

589
	if (pagevec_count(pvec))
590
		__pagevec_lru_add(pvec);
591 592 593 594 595 596 597 598 599 600

	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);
	}
M
Minchan Kim 已提交
601

602
	pvec = &per_cpu(lru_deactivate_file_pvecs, cpu);
M
Minchan Kim 已提交
603
	if (pagevec_count(pvec))
604
		pagevec_lru_move_fn(pvec, lru_deactivate_file_fn, NULL);
605

606 607 608 609
	pvec = &per_cpu(lru_deactivate_pvecs, cpu);
	if (pagevec_count(pvec))
		pagevec_lru_move_fn(pvec, lru_deactivate_fn, NULL);

610
	activate_page_drain(cpu);
M
Minchan Kim 已提交
611 612 613
}

/**
614
 * deactivate_file_page - forcefully deactivate a file page
M
Minchan Kim 已提交
615 616 617 618 619 620
 * @page: page to deactivate
 *
 * This function hints the VM that @page is a good reclaim candidate,
 * for example if its invalidation fails due to the page being dirty
 * or under writeback.
 */
621
void deactivate_file_page(struct page *page)
M
Minchan Kim 已提交
622
{
623
	/*
624 625
	 * In a workload with many unevictable page such as mprotect,
	 * unevictable page deactivation for accelerating reclaim is pointless.
626 627 628 629
	 */
	if (PageUnevictable(page))
		return;

M
Minchan Kim 已提交
630
	if (likely(get_page_unless_zero(page))) {
631
		struct pagevec *pvec = &get_cpu_var(lru_deactivate_file_pvecs);
M
Minchan Kim 已提交
632

633
		if (!pagevec_add(pvec, page) || PageCompound(page))
634 635
			pagevec_lru_move_fn(pvec, lru_deactivate_file_fn, NULL);
		put_cpu_var(lru_deactivate_file_pvecs);
M
Minchan Kim 已提交
636
	}
637 638
}

639 640 641 642 643 644 645 646 647 648 649 650 651
/**
 * deactivate_page - deactivate a page
 * @page: page to deactivate
 *
 * deactivate_page() moves @page to the inactive list if @page was on the active
 * list and was not an unevictable page.  This is done to accelerate the reclaim
 * of @page.
 */
void deactivate_page(struct page *page)
{
	if (PageLRU(page) && PageActive(page) && !PageUnevictable(page)) {
		struct pagevec *pvec = &get_cpu_var(lru_deactivate_pvecs);

652
		get_page(page);
653
		if (!pagevec_add(pvec, page) || PageCompound(page))
654 655 656 657 658
			pagevec_lru_move_fn(pvec, lru_deactivate_fn, NULL);
		put_cpu_var(lru_deactivate_pvecs);
	}
}

659 660
void lru_add_drain(void)
{
661
	lru_add_drain_cpu(get_cpu());
662
	put_cpu();
L
Linus Torvalds 已提交
663 664
}

D
David Howells 已提交
665
static void lru_add_drain_per_cpu(struct work_struct *dummy)
666 667 668 669
{
	lru_add_drain();
}

670 671
static DEFINE_PER_CPU(struct work_struct, lru_add_drain_work);

672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689
/*
 * lru_add_drain_wq is used to do lru_add_drain_all() from a WQ_MEM_RECLAIM
 * workqueue, aiding in getting memory freed.
 */
static struct workqueue_struct *lru_add_drain_wq;

static int __init lru_init(void)
{
	lru_add_drain_wq = alloc_workqueue("lru-add-drain", WQ_MEM_RECLAIM, 0);

	if (WARN(!lru_add_drain_wq,
		"Failed to create workqueue lru_add_drain_wq"))
		return -ENOMEM;

	return 0;
}
early_initcall(lru_init);

690
void lru_add_drain_all(void)
691
{
692 693 694 695 696 697 698 699 700 701 702 703 704
	static DEFINE_MUTEX(lock);
	static struct cpumask has_work;
	int cpu;

	mutex_lock(&lock);
	get_online_cpus();
	cpumask_clear(&has_work);

	for_each_online_cpu(cpu) {
		struct work_struct *work = &per_cpu(lru_add_drain_work, cpu);

		if (pagevec_count(&per_cpu(lru_add_pvec, cpu)) ||
		    pagevec_count(&per_cpu(lru_rotate_pvecs, cpu)) ||
705
		    pagevec_count(&per_cpu(lru_deactivate_file_pvecs, cpu)) ||
706
		    pagevec_count(&per_cpu(lru_deactivate_pvecs, cpu)) ||
707 708
		    need_activate_page_drain(cpu)) {
			INIT_WORK(work, lru_add_drain_per_cpu);
709
			queue_work_on(cpu, lru_add_drain_wq, work);
710 711 712 713 714 715 716 717 718
			cpumask_set_cpu(cpu, &has_work);
		}
	}

	for_each_cpu(cpu, &has_work)
		flush_work(&per_cpu(lru_add_drain_work, cpu));

	put_online_cpus();
	mutex_unlock(&lock);
719 720
}

721
/**
722
 * release_pages - batched put_page()
723 724 725
 * @pages: array of pages to release
 * @nr: number of pages
 * @cold: whether the pages are cache cold
L
Linus Torvalds 已提交
726
 *
727 728
 * Decrement the reference count on all the pages in @pages.  If it
 * fell to zero, remove the page from the LRU and free it.
L
Linus Torvalds 已提交
729
 */
730
void release_pages(struct page **pages, int nr, bool cold)
L
Linus Torvalds 已提交
731 732
{
	int i;
733
	LIST_HEAD(pages_to_free);
M
Mel Gorman 已提交
734
	struct pglist_data *locked_pgdat = NULL;
735
	struct lruvec *lruvec;
736
	unsigned long uninitialized_var(flags);
737
	unsigned int uninitialized_var(lock_batch);
L
Linus Torvalds 已提交
738 739 740 741

	for (i = 0; i < nr; i++) {
		struct page *page = pages[i];

742 743 744
		/*
		 * Make sure the IRQ-safe lock-holding time does not get
		 * excessive with a continuous string of pages from the
M
Mel Gorman 已提交
745
		 * same pgdat. The lock is held only if pgdat != NULL.
746
		 */
M
Mel Gorman 已提交
747 748 749
		if (locked_pgdat && ++lock_batch == SWAP_CLUSTER_MAX) {
			spin_unlock_irqrestore(&locked_pgdat->lru_lock, flags);
			locked_pgdat = NULL;
750 751
		}

752
		if (is_huge_zero_page(page))
753 754
			continue;

755
		page = compound_head(page);
N
Nick Piggin 已提交
756
		if (!put_page_testzero(page))
L
Linus Torvalds 已提交
757 758
			continue;

759
		if (PageCompound(page)) {
M
Mel Gorman 已提交
760 761 762
			if (locked_pgdat) {
				spin_unlock_irqrestore(&locked_pgdat->lru_lock, flags);
				locked_pgdat = NULL;
763 764 765 766 767
			}
			__put_compound_page(page);
			continue;
		}

768
		if (PageLRU(page)) {
M
Mel Gorman 已提交
769
			struct pglist_data *pgdat = page_pgdat(page);
L
Lee Schermerhorn 已提交
770

M
Mel Gorman 已提交
771 772 773
			if (pgdat != locked_pgdat) {
				if (locked_pgdat)
					spin_unlock_irqrestore(&locked_pgdat->lru_lock,
774
									flags);
775
				lock_batch = 0;
M
Mel Gorman 已提交
776 777
				locked_pgdat = pgdat;
				spin_lock_irqsave(&locked_pgdat->lru_lock, flags);
778
			}
779

M
Mel Gorman 已提交
780
			lruvec = mem_cgroup_page_lruvec(page, locked_pgdat);
781
			VM_BUG_ON_PAGE(!PageLRU(page), page);
N
Nick Piggin 已提交
782
			__ClearPageLRU(page);
783
			del_page_from_lru_list(page, lruvec, page_off_lru(page));
784 785
		}

786
		/* Clear Active bit in case of parallel mark_page_accessed */
787
		__ClearPageActive(page);
788
		__ClearPageWaiters(page);
789

790
		list_add(&page->lru, &pages_to_free);
L
Linus Torvalds 已提交
791
	}
M
Mel Gorman 已提交
792 793
	if (locked_pgdat)
		spin_unlock_irqrestore(&locked_pgdat->lru_lock, flags);
L
Linus Torvalds 已提交
794

795
	mem_cgroup_uncharge_list(&pages_to_free);
796
	free_hot_cold_page_list(&pages_to_free, cold);
L
Linus Torvalds 已提交
797
}
M
Miklos Szeredi 已提交
798
EXPORT_SYMBOL(release_pages);
L
Linus Torvalds 已提交
799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815

/*
 * 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);
}
816 817
EXPORT_SYMBOL(__pagevec_release);

818
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
819
/* used by __split_huge_page_refcount() */
820
void lru_add_page_tail(struct page *page, struct page *page_tail,
821
		       struct lruvec *lruvec, struct list_head *list)
822 823 824
{
	const int file = 0;

825 826 827
	VM_BUG_ON_PAGE(!PageHead(page), page);
	VM_BUG_ON_PAGE(PageCompound(page_tail), page);
	VM_BUG_ON_PAGE(PageLRU(page_tail), page);
828
	VM_BUG_ON(NR_CPUS != 1 &&
M
Mel Gorman 已提交
829
		  !spin_is_locked(&lruvec_pgdat(lruvec)->lru_lock));
830

831 832
	if (!list)
		SetPageLRU(page_tail);
833

834 835
	if (likely(PageLRU(page)))
		list_add_tail(&page_tail->lru, &page->lru);
836 837 838 839 840
	else if (list) {
		/* page reclaim is reclaiming a huge page */
		get_page(page_tail);
		list_add_tail(&page_tail->lru, list);
	} else {
841 842 843 844 845 846 847 848
		struct list_head *list_head;
		/*
		 * Head page has not yet been counted, as an hpage,
		 * so we must account for each subpage individually.
		 *
		 * Use the standard add function to put page_tail on the list,
		 * but then correct its position so they all end up in order.
		 */
849
		add_page_to_lru_list(page_tail, lruvec, page_lru(page_tail));
850 851
		list_head = page_tail->lru.prev;
		list_move_tail(&page_tail->lru, list_head);
852
	}
853 854

	if (!PageUnevictable(page))
855
		update_page_reclaim_stat(lruvec, file, PageActive(page_tail));
856
}
857
#endif /* CONFIG_TRANSPARENT_HUGEPAGE */
858

859 860
static void __pagevec_lru_add_fn(struct page *page, struct lruvec *lruvec,
				 void *arg)
S
Shaohua Li 已提交
861
{
862 863 864
	int file = page_is_file_cache(page);
	int active = PageActive(page);
	enum lru_list lru = page_lru(page);
S
Shaohua Li 已提交
865

866
	VM_BUG_ON_PAGE(PageLRU(page), page);
S
Shaohua Li 已提交
867 868

	SetPageLRU(page);
869 870
	add_page_to_lru_list(page, lruvec, lru);
	update_page_reclaim_stat(lruvec, file, active);
871
	trace_mm_lru_insertion(page, lru);
S
Shaohua Li 已提交
872 873
}

L
Linus Torvalds 已提交
874 875 876 877
/*
 * Add the passed pages to the LRU, then drop the caller's refcount
 * on them.  Reinitialises the caller's pagevec.
 */
878
void __pagevec_lru_add(struct pagevec *pvec)
L
Linus Torvalds 已提交
879
{
880
	pagevec_lru_move_fn(pvec, __pagevec_lru_add_fn, NULL);
L
Linus Torvalds 已提交
881
}
882
EXPORT_SYMBOL(__pagevec_lru_add);
L
Linus Torvalds 已提交
883

884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934
/**
 * pagevec_lookup_entries - gang pagecache lookup
 * @pvec:	Where the resulting entries are placed
 * @mapping:	The address_space to search
 * @start:	The starting entry index
 * @nr_entries:	The maximum number of entries
 * @indices:	The cache indices corresponding to the entries in @pvec
 *
 * pagevec_lookup_entries() will search for and return a group of up
 * to @nr_entries pages and shadow entries in the mapping.  All
 * entries are placed in @pvec.  pagevec_lookup_entries() takes a
 * reference against actual pages in @pvec.
 *
 * The search returns a group of mapping-contiguous entries with
 * ascending indexes.  There may be holes in the indices due to
 * not-present entries.
 *
 * pagevec_lookup_entries() returns the number of entries which were
 * found.
 */
unsigned pagevec_lookup_entries(struct pagevec *pvec,
				struct address_space *mapping,
				pgoff_t start, unsigned nr_pages,
				pgoff_t *indices)
{
	pvec->nr = find_get_entries(mapping, start, nr_pages,
				    pvec->pages, indices);
	return pagevec_count(pvec);
}

/**
 * pagevec_remove_exceptionals - pagevec exceptionals pruning
 * @pvec:	The pagevec to prune
 *
 * pagevec_lookup_entries() fills both pages and exceptional radix
 * tree entries into the pagevec.  This function prunes all
 * exceptionals from @pvec without leaving holes, so that it can be
 * passed on to page-only pagevec operations.
 */
void pagevec_remove_exceptionals(struct pagevec *pvec)
{
	int i, j;

	for (i = 0, j = 0; i < pagevec_count(pvec); i++) {
		struct page *page = pvec->pages[i];
		if (!radix_tree_exceptional_entry(page))
			pvec->pages[j++] = page;
	}
	pvec->nr = j;
}

L
Linus Torvalds 已提交
935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956
/**
 * 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);
}
957 958
EXPORT_SYMBOL(pagevec_lookup);

L
Linus Torvalds 已提交
959 960 961 962 963 964 965
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);
}
966
EXPORT_SYMBOL(pagevec_lookup_tag);
L
Linus Torvalds 已提交
967 968 969 970 971 972

/*
 * Perform any setup for the swap system
 */
void __init swap_setup(void)
{
973
	unsigned long megs = totalram_pages >> (20 - PAGE_SHIFT);
P
Peter Zijlstra 已提交
974
#ifdef CONFIG_SWAP
975 976
	int i;

977
	for (i = 0; i < MAX_SWAPFILES; i++)
978
		spin_lock_init(&swapper_spaces[i].tree_lock);
P
Peter Zijlstra 已提交
979 980
#endif

L
Linus Torvalds 已提交
981 982 983 984 985 986 987 988 989 990
	/* 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
	 */
}