vmscan.c 76.7 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 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);
L
Linus Torvalds 已提交
633
		/* In active use or really unfreeable?  Activate it. */
A
Andy Whitcroft 已提交
634 635
		if (sc->order <= PAGE_ALLOC_COSTLY_ORDER &&
					referenced && page_mapping_inuse(page))
L
Linus Torvalds 已提交
636 637 638 639 640 641
			goto activate_locked;

		/*
		 * Anonymous process memory has backing store?
		 * Try to allocate it some swap space here.
		 */
N
Nick Piggin 已提交
642
		if (PageAnon(page) && !PageSwapCache(page)) {
643 644
			if (!(sc->gfp_mask & __GFP_IO))
				goto keep_locked;
645
			if (!add_to_swap(page))
L
Linus Torvalds 已提交
646
				goto activate_locked;
647
			may_enter_fs = 1;
N
Nick Piggin 已提交
648
		}
L
Linus Torvalds 已提交
649 650 651 652 653 654 655 656

		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) {
657
			switch (try_to_unmap(page, 0)) {
L
Linus Torvalds 已提交
658 659 660 661
			case SWAP_FAIL:
				goto activate_locked;
			case SWAP_AGAIN:
				goto keep_locked;
N
Nick Piggin 已提交
662 663
			case SWAP_MLOCK:
				goto cull_mlocked;
L
Linus Torvalds 已提交
664 665 666 667 668 669
			case SWAP_SUCCESS:
				; /* try to free the page below */
			}
		}

		if (PageDirty(page)) {
A
Andy Whitcroft 已提交
670
			if (sc->order <= PAGE_ALLOC_COSTLY_ORDER && referenced)
L
Linus Torvalds 已提交
671
				goto keep_locked;
672
			if (!may_enter_fs)
L
Linus Torvalds 已提交
673
				goto keep_locked;
674
			if (!sc->may_writepage)
L
Linus Torvalds 已提交
675 676 677
				goto keep_locked;

			/* Page is dirty, try to write it out here */
678
			switch (pageout(page, mapping, sync_writeback)) {
L
Linus Torvalds 已提交
679 680 681 682 683
			case PAGE_KEEP:
				goto keep_locked;
			case PAGE_ACTIVATE:
				goto activate_locked;
			case PAGE_SUCCESS:
684
				if (PageWriteback(page) || PageDirty(page))
L
Linus Torvalds 已提交
685 686 687 688 689
					goto keep;
				/*
				 * A synchronous write - probably a ramdisk.  Go
				 * ahead and try to reclaim the page.
				 */
N
Nick Piggin 已提交
690
				if (!trylock_page(page))
L
Linus Torvalds 已提交
691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709
					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
710
		 * will do this, as well as the blockdev mapping.
L
Linus Torvalds 已提交
711 712 713 714 715 716 717 718 719 720
		 * 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.
		 */
721
		if (page_has_private(page)) {
L
Linus Torvalds 已提交
722 723
			if (!try_to_release_page(page, sc->gfp_mask))
				goto activate_locked;
N
Nick Piggin 已提交
724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739
			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
740 741
		}

N
Nick Piggin 已提交
742
		if (!mapping || !__remove_mapping(mapping, page))
743
			goto keep_locked;
L
Linus Torvalds 已提交
744

N
Nick Piggin 已提交
745 746 747 748 749 750 751 752
		/*
		 * 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
753
free_it:
754
		nr_reclaimed++;
N
Nick Piggin 已提交
755 756 757 758
		if (!pagevec_add(&freed_pvec, page)) {
			__pagevec_free(&freed_pvec);
			pagevec_reinit(&freed_pvec);
		}
L
Linus Torvalds 已提交
759 760
		continue;

N
Nick Piggin 已提交
761
cull_mlocked:
762 763
		if (PageSwapCache(page))
			try_to_free_swap(page);
N
Nick Piggin 已提交
764 765 766 767
		unlock_page(page);
		putback_lru_page(page);
		continue;

L
Linus Torvalds 已提交
768
activate_locked:
769 770
		/* Not a candidate for swapping, so reclaim swap space. */
		if (PageSwapCache(page) && vm_swap_full())
771
			try_to_free_swap(page);
L
Lee Schermerhorn 已提交
772
		VM_BUG_ON(PageActive(page));
L
Linus Torvalds 已提交
773 774 775 776 777 778
		SetPageActive(page);
		pgactivate++;
keep_locked:
		unlock_page(page);
keep:
		list_add(&page->lru, &ret_pages);
N
Nick Piggin 已提交
779
		VM_BUG_ON(PageLRU(page) || PageUnevictable(page));
L
Linus Torvalds 已提交
780 781 782
	}
	list_splice(&ret_pages, page_list);
	if (pagevec_count(&freed_pvec))
N
Nick Piggin 已提交
783
		__pagevec_free(&freed_pvec);
784
	count_vm_events(PGACTIVATE, pgactivate);
785
	return nr_reclaimed;
L
Linus Torvalds 已提交
786 787
}

A
Andy Whitcroft 已提交
788 789 790 791 792 793 794 795 796 797 798 799 800 801 802
/* 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.
 */
803
int __isolate_lru_page(struct page *page, int mode, int file)
A
Andy Whitcroft 已提交
804 805 806 807 808 809 810 811 812 813 814 815 816 817 818
{
	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;

819 820 821
	if (mode != ISOLATE_BOTH && (!page_is_file_cache(page) != !file))
		return ret;

L
Lee Schermerhorn 已提交
822 823 824 825 826 827 828 829
	/*
	 * 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
830
	ret = -EBUSY;
K
KAMEZAWA Hiroyuki 已提交
831

A
Andy Whitcroft 已提交
832 833 834 835 836 837 838 839
	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;
K
KAMEZAWA Hiroyuki 已提交
840
		mem_cgroup_del_lru(page);
A
Andy Whitcroft 已提交
841 842 843 844 845
	}

	return ret;
}

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

873
	for (scan = 0; scan < nr_to_scan && !list_empty(src); scan++) {
A
Andy Whitcroft 已提交
874 875 876 877 878 879
		struct page *page;
		unsigned long pfn;
		unsigned long end_pfn;
		unsigned long page_pfn;
		int zone_id;

L
Linus Torvalds 已提交
880 881 882
		page = lru_to_page(src);
		prefetchw_prev_lru_page(page, src, flags);

N
Nick Piggin 已提交
883
		VM_BUG_ON(!PageLRU(page));
N
Nick Piggin 已提交
884

885
		switch (__isolate_lru_page(page, mode, file)) {
A
Andy Whitcroft 已提交
886 887
		case 0:
			list_move(&page->lru, dst);
888
			nr_taken++;
A
Andy Whitcroft 已提交
889 890 891 892 893 894
			break;

		case -EBUSY:
			/* else it is being freed elsewhere */
			list_move(&page->lru, src);
			continue;
895

A
Andy Whitcroft 已提交
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
		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);
928

A
Andy Whitcroft 已提交
929 930 931
			/* Check that we have not crossed a zone boundary. */
			if (unlikely(page_zone_id(cursor_page) != zone_id))
				continue;
932
			switch (__isolate_lru_page(cursor_page, mode, file)) {
A
Andy Whitcroft 已提交
933 934 935 936 937 938 939 940 941 942
			case 0:
				list_move(&cursor_page->lru, dst);
				nr_taken++;
				scan++;
				break;

			case -EBUSY:
				/* else it is being freed elsewhere */
				list_move(&cursor_page->lru, src);
			default:
L
Lee Schermerhorn 已提交
943
				break;	/* ! on LRU or wrong list */
A
Andy Whitcroft 已提交
944 945
			}
		}
L
Linus Torvalds 已提交
946 947 948 949 950 951
	}

	*scanned = scan;
	return nr_taken;
}

952 953 954 955 956
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,
957
					int active, int file)
958
{
959
	int lru = LRU_BASE;
960
	if (active)
961 962 963 964 965
		lru += LRU_ACTIVE;
	if (file)
		lru += LRU_FILE;
	return isolate_lru_pages(nr, &z->lru[lru].list, dst, scanned, order,
								mode, !!file);
966 967
}

A
Andy Whitcroft 已提交
968 969 970 971
/*
 * clear_active_flags() is a helper for shrink_active_list(), clearing
 * any active bits from the pages in the list.
 */
972 973
static unsigned long clear_active_flags(struct list_head *page_list,
					unsigned int *count)
A
Andy Whitcroft 已提交
974 975
{
	int nr_active = 0;
976
	int lru;
A
Andy Whitcroft 已提交
977 978
	struct page *page;

979 980
	list_for_each_entry(page, page_list, lru) {
		lru = page_is_file_cache(page);
A
Andy Whitcroft 已提交
981
		if (PageActive(page)) {
982
			lru += LRU_ACTIVE;
A
Andy Whitcroft 已提交
983 984 985
			ClearPageActive(page);
			nr_active++;
		}
986 987
		count[lru]++;
	}
A
Andy Whitcroft 已提交
988 989 990 991

	return nr_active;
}

992 993 994 995 996 997 998 999 1000 1001 1002
/**
 * 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
1003 1004 1005
 * 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.
1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025
 *
 * 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
1026
			int lru = page_lru(page);
1027 1028
			ret = 0;
			ClearPageLRU(page);
1029 1030

			del_page_from_lru_list(zone, page, lru);
1031 1032 1033 1034 1035 1036
		}
		spin_unlock_irq(&zone->lru_lock);
	}
	return ret;
}

L
Linus Torvalds 已提交
1037
/*
A
Andrew Morton 已提交
1038 1039
 * shrink_inactive_list() is a helper for shrink_zone().  It returns the number
 * of reclaimed pages
L
Linus Torvalds 已提交
1040
 */
A
Andrew Morton 已提交
1041
static unsigned long shrink_inactive_list(unsigned long max_scan,
R
Rik van Riel 已提交
1042 1043
			struct zone *zone, struct scan_control *sc,
			int priority, int file)
L
Linus Torvalds 已提交
1044 1045 1046
{
	LIST_HEAD(page_list);
	struct pagevec pvec;
1047
	unsigned long nr_scanned = 0;
1048
	unsigned long nr_reclaimed = 0;
1049
	struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(zone, sc);
1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062
	int lumpy_reclaim = 0;

	/*
	 * 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
1063 1064 1065 1066 1067

	pagevec_init(&pvec, 1);

	lru_add_drain();
	spin_lock_irq(&zone->lru_lock);
1068
	do {
L
Linus Torvalds 已提交
1069
		struct page *page;
1070 1071 1072
		unsigned long nr_taken;
		unsigned long nr_scan;
		unsigned long nr_freed;
A
Andy Whitcroft 已提交
1073
		unsigned long nr_active;
1074
		unsigned int count[NR_LRU_LISTS] = { 0, };
1075
		int mode = lumpy_reclaim ? ISOLATE_BOTH : ISOLATE_INACTIVE;
L
Linus Torvalds 已提交
1076

1077
		nr_taken = sc->isolate_pages(sc->swap_cluster_max,
1078 1079 1080
			     &page_list, &nr_scan, sc->order, mode,
				zone, sc->mem_cgroup, 0, file);
		nr_active = clear_active_flags(&page_list, count);
1081
		__count_vm_events(PGDEACTIVATE, nr_active);
A
Andy Whitcroft 已提交
1082

1083 1084 1085 1086 1087 1088 1089 1090 1091
		__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]);

1092
		if (scanning_global_lru(sc))
1093
			zone->pages_scanned += nr_scan;
K
KOSAKI Motohiro 已提交
1094 1095 1096 1097 1098 1099

		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
1100 1101
		spin_unlock_irq(&zone->lru_lock);

1102
		nr_scanned += nr_scan;
1103 1104 1105 1106 1107 1108 1109 1110 1111
		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() &&
1112
		    lumpy_reclaim) {
1113 1114 1115 1116 1117 1118
			congestion_wait(WRITE, HZ/10);

			/*
			 * The attempt at page out may have made some
			 * of the pages active, mark them inactive again.
			 */
1119
			nr_active = clear_active_flags(&page_list, count);
1120 1121 1122 1123 1124 1125
			count_vm_events(PGDEACTIVATE, nr_active);

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

1126
		nr_reclaimed += nr_freed;
N
Nick Piggin 已提交
1127 1128
		local_irq_disable();
		if (current_is_kswapd()) {
1129 1130
			__count_zone_vm_events(PGSCAN_KSWAPD, zone, nr_scan);
			__count_vm_events(KSWAPD_STEAL, nr_freed);
1131
		} else if (scanning_global_lru(sc))
1132
			__count_zone_vm_events(PGSCAN_DIRECT, zone, nr_scan);
1133

S
Shantanu Goel 已提交
1134
		__count_zone_vm_events(PGSTEAL, zone, nr_freed);
N
Nick Piggin 已提交
1135

1136 1137 1138
		if (nr_taken == 0)
			goto done;

N
Nick Piggin 已提交
1139
		spin_lock(&zone->lru_lock);
L
Linus Torvalds 已提交
1140 1141 1142 1143
		/*
		 * Put back any unfreeable pages.
		 */
		while (!list_empty(&page_list)) {
L
Lee Schermerhorn 已提交
1144
			int lru;
L
Linus Torvalds 已提交
1145
			page = lru_to_page(&page_list);
N
Nick Piggin 已提交
1146
			VM_BUG_ON(PageLRU(page));
L
Linus Torvalds 已提交
1147
			list_del(&page->lru);
L
Lee Schermerhorn 已提交
1148 1149 1150 1151 1152 1153 1154 1155 1156
			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);
K
KOSAKI Motohiro 已提交
1157
			if (PageActive(page)) {
1158
				int file = !!page_is_file_cache(page);
1159
				reclaim_stat->recent_rotated[file]++;
1160
			}
L
Linus Torvalds 已提交
1161 1162 1163 1164 1165 1166
			if (!pagevec_add(&pvec, page)) {
				spin_unlock_irq(&zone->lru_lock);
				__pagevec_release(&pvec);
				spin_lock_irq(&zone->lru_lock);
			}
		}
1167
  	} while (nr_scanned < max_scan);
1168
	spin_unlock(&zone->lru_lock);
L
Linus Torvalds 已提交
1169
done:
1170
	local_irq_enable();
L
Linus Torvalds 已提交
1171
	pagevec_release(&pvec);
1172
	return nr_reclaimed;
L
Linus Torvalds 已提交
1173 1174
}

1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188
/*
 * 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
1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205
/*
 * 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.
 */
1206

1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243
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);
		prefetchw_prev_lru_page(page, list, flags);

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

		VM_BUG_ON(!PageActive(page));
		if (!is_active_lru(lru))
			ClearPageActive(page);	/* we are de-activating */

		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);
}
1244

A
Andrew Morton 已提交
1245
static void shrink_active_list(unsigned long nr_pages, struct zone *zone,
1246
			struct scan_control *sc, int priority, int file)
L
Linus Torvalds 已提交
1247
{
1248 1249
	unsigned long pgmoved;
	unsigned long pgscanned;
1250
	unsigned long vm_flags;
L
Linus Torvalds 已提交
1251
	LIST_HEAD(l_hold);	/* The pages which were snipped off */
1252
	LIST_HEAD(l_active);
1253
	LIST_HEAD(l_inactive);
L
Linus Torvalds 已提交
1254
	struct page *page;
1255
	struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(zone, sc);
L
Linus Torvalds 已提交
1256 1257 1258

	lru_add_drain();
	spin_lock_irq(&zone->lru_lock);
1259 1260
	pgmoved = sc->isolate_pages(nr_pages, &l_hold, &pgscanned, sc->order,
					ISOLATE_ACTIVE, zone,
1261
					sc->mem_cgroup, 1, file);
1262 1263 1264 1265
	/*
	 * zone->pages_scanned is used for detect zone's oom
	 * mem_cgroup remembers nr_scan by itself.
	 */
1266
	if (scanning_global_lru(sc)) {
1267
		zone->pages_scanned += pgscanned;
1268
	}
K
KOSAKI Motohiro 已提交
1269
	reclaim_stat->recent_scanned[!!file] += pgmoved;
1270

1271
	__count_zone_vm_events(PGREFILL, zone, pgscanned);
1272 1273 1274 1275
	if (file)
		__mod_zone_page_state(zone, NR_ACTIVE_FILE, -pgmoved);
	else
		__mod_zone_page_state(zone, NR_ACTIVE_ANON, -pgmoved);
L
Linus Torvalds 已提交
1276 1277
	spin_unlock_irq(&zone->lru_lock);

1278
	pgmoved = 0;  /* count referenced (mapping) mapped pages */
L
Linus Torvalds 已提交
1279 1280 1281 1282
	while (!list_empty(&l_hold)) {
		cond_resched();
		page = lru_to_page(&l_hold);
		list_del(&page->lru);
1283

L
Lee Schermerhorn 已提交
1284 1285 1286 1287 1288
		if (unlikely(!page_evictable(page, NULL))) {
			putback_lru_page(page);
			continue;
		}

1289 1290
		/* page_referenced clears PageReferenced */
		if (page_mapping_inuse(page) &&
1291
		    page_referenced(page, 0, sc->mem_cgroup, &vm_flags)) {
1292
			pgmoved++;
1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306
			/*
			 * 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;
			}
		}
1307

L
Linus Torvalds 已提交
1308 1309 1310
		list_add(&page->lru, &l_inactive);
	}

1311
	/*
1312
	 * Move pages back to the lru list.
1313
	 */
1314
	spin_lock_irq(&zone->lru_lock);
1315
	/*
1316 1317 1318 1319
	 * 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.
1320
	 */
K
KOSAKI Motohiro 已提交
1321
	reclaim_stat->recent_rotated[!!file] += pgmoved;
1322

1323 1324 1325 1326
	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);
1327

1328
	spin_unlock_irq(&zone->lru_lock);
L
Linus Torvalds 已提交
1329 1330
}

1331
static int inactive_anon_is_low_global(struct zone *zone)
1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343
{
	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;
}

1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355
/**
 * 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;

1356
	if (scanning_global_lru(sc))
1357 1358
		low = inactive_anon_is_low_global(zone);
	else
1359
		low = mem_cgroup_inactive_anon_is_low(sc->mem_cgroup);
1360 1361 1362
	return low;
}

1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398
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;
}

1399
static unsigned long shrink_list(enum lru_list lru, unsigned long nr_to_scan,
1400 1401
	struct zone *zone, struct scan_control *sc, int priority)
{
1402 1403
	int file = is_file_lru(lru);

1404
	if (lru == LRU_ACTIVE_FILE && inactive_file_is_low(zone, sc)) {
1405 1406 1407 1408
		shrink_active_list(nr_to_scan, zone, sc, priority, file);
		return 0;
	}

1409
	if (lru == LRU_ACTIVE_ANON && inactive_anon_is_low(zone, sc)) {
1410
		shrink_active_list(nr_to_scan, zone, sc, priority, file);
1411 1412
		return 0;
	}
R
Rik van Riel 已提交
1413
	return shrink_inactive_list(nr_to_scan, zone, sc, priority, file);
1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430
}

/*
 * 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;
1431
	struct zone_reclaim_stat *reclaim_stat = get_reclaim_stat(zone, sc);
1432

1433 1434 1435 1436
	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);
1437

1438
	if (scanning_global_lru(sc)) {
1439 1440 1441
		free  = zone_page_state(zone, NR_FREE_PAGES);
		/* If we have very few page cache pages,
		   force-scan anon pages. */
1442
		if (unlikely(file + free <= high_wmark_pages(zone))) {
1443 1444 1445 1446
			percent[0] = 100;
			percent[1] = 0;
			return;
		}
1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459
	}

	/*
	 * 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]
	 */
1460
	if (unlikely(reclaim_stat->recent_scanned[0] > anon / 4)) {
1461
		spin_lock_irq(&zone->lru_lock);
1462 1463
		reclaim_stat->recent_scanned[0] /= 2;
		reclaim_stat->recent_rotated[0] /= 2;
1464 1465 1466
		spin_unlock_irq(&zone->lru_lock);
	}

1467
	if (unlikely(reclaim_stat->recent_scanned[1] > file / 4)) {
1468
		spin_lock_irq(&zone->lru_lock);
1469 1470
		reclaim_stat->recent_scanned[1] /= 2;
		reclaim_stat->recent_rotated[1] /= 2;
1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481
		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;

	/*
1482 1483 1484
	 * 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.
1485
	 */
1486 1487
	ap = (anon_prio + 1) * (reclaim_stat->recent_scanned[0] + 1);
	ap /= reclaim_stat->recent_rotated[0] + 1;
1488

1489 1490
	fp = (file_prio + 1) * (reclaim_stat->recent_scanned[1] + 1);
	fp /= reclaim_stat->recent_rotated[1] + 1;
1491 1492 1493 1494

	/* Normalize to percentages */
	percent[0] = 100 * ap / (ap + fp + 1);
	percent[1] = 100 - percent[0];
1495 1496
}

1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516
/*
 * 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;
}
1517

L
Linus Torvalds 已提交
1518 1519 1520
/*
 * This is a basic per-zone page freer.  Used by both kswapd and direct reclaim.
 */
1521
static void shrink_zone(int priority, struct zone *zone,
1522
				struct scan_control *sc)
L
Linus Torvalds 已提交
1523
{
1524
	unsigned long nr[NR_LRU_LISTS];
1525
	unsigned long nr_to_scan;
1526
	unsigned long percent[2];	/* anon @ 0; file @ 1 */
1527
	enum lru_list l;
1528 1529
	unsigned long nr_reclaimed = sc->nr_reclaimed;
	unsigned long swap_cluster_max = sc->swap_cluster_max;
1530
	int noswap = 0;
L
Linus Torvalds 已提交
1531

1532 1533 1534 1535 1536 1537 1538
	/* 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);
1539

L
Lee Schermerhorn 已提交
1540
	for_each_evictable_lru(l) {
1541
		int file = is_file_lru(l);
1542
		unsigned long scan;
1543

1544
		scan = zone_nr_pages(zone, sc, l);
1545
		if (priority || noswap) {
1546 1547 1548
			scan >>= priority;
			scan = (scan * percent[file]) / 100;
		}
1549 1550 1551 1552 1553
		if (scanning_global_lru(sc))
			nr[l] = nr_scan_try_batch(scan,
						  &zone->lru[l].nr_saved_scan,
						  swap_cluster_max);
		else
1554
			nr[l] = scan;
1555
	}
L
Linus Torvalds 已提交
1556

1557 1558
	while (nr[LRU_INACTIVE_ANON] || nr[LRU_ACTIVE_FILE] ||
					nr[LRU_INACTIVE_FILE]) {
L
Lee Schermerhorn 已提交
1559
		for_each_evictable_lru(l) {
1560
			if (nr[l]) {
1561
				nr_to_scan = min(nr[l], swap_cluster_max);
1562
				nr[l] -= nr_to_scan;
L
Linus Torvalds 已提交
1563

1564 1565
				nr_reclaimed += shrink_list(l, nr_to_scan,
							    zone, sc, priority);
1566
			}
L
Linus Torvalds 已提交
1567
		}
1568 1569 1570 1571 1572 1573 1574 1575
		/*
		 * 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.
		 */
1576
		if (nr_reclaimed > swap_cluster_max &&
1577 1578
			priority < DEF_PRIORITY && !current_is_kswapd())
			break;
L
Linus Torvalds 已提交
1579 1580
	}

1581 1582
	sc->nr_reclaimed = nr_reclaimed;

1583 1584 1585 1586
	/*
	 * Even if we did not try to evict anon pages at all, we want to
	 * rebalance the anon lru active/inactive ratio.
	 */
1587
	if (inactive_anon_is_low(zone, sc) && nr_swap_pages > 0)
1588 1589
		shrink_active_list(SWAP_CLUSTER_MAX, zone, sc, priority, 0);

1590
	throttle_vm_writeout(sc->gfp_mask);
L
Linus Torvalds 已提交
1591 1592 1593 1594 1595 1596 1597
}

/*
 * 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.
 *
1598 1599
 * We reclaim from a zone even if that zone is over high_wmark_pages(zone).
 * Because:
L
Linus Torvalds 已提交
1600 1601
 * a) The caller may be trying to free *extra* pages to satisfy a higher-order
 *    allocation or
1602 1603 1604
 * 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
1605 1606 1607 1608
 *
 * If a zone is deemed to be full of pinned pages then just give it a light
 * scan then give up on it.
 */
1609
static void shrink_zones(int priority, struct zonelist *zonelist,
1610
					struct scan_control *sc)
L
Linus Torvalds 已提交
1611
{
1612
	enum zone_type high_zoneidx = gfp_zone(sc->gfp_mask);
1613
	struct zoneref *z;
1614
	struct zone *zone;
1615

1616
	sc->all_unreclaimable = 1;
1617 1618
	for_each_zone_zonelist_nodemask(zone, z, zonelist, high_zoneidx,
					sc->nodemask) {
1619
		if (!populated_zone(zone))
L
Linus Torvalds 已提交
1620
			continue;
1621 1622 1623 1624
		/*
		 * Take care memory controller reclaiming has small influence
		 * to global LRU.
		 */
1625
		if (scanning_global_lru(sc)) {
1626 1627 1628
			if (!cpuset_zone_allowed_hardwall(zone, GFP_KERNEL))
				continue;
			note_zone_scanning_priority(zone, priority);
L
Linus Torvalds 已提交
1629

1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642
			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);
		}
1643

1644
		shrink_zone(priority, zone, sc);
L
Linus Torvalds 已提交
1645 1646
	}
}
1647

L
Linus Torvalds 已提交
1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659
/*
 * 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.
1660 1661 1662
 *
 * returns:	0, if no pages reclaimed
 * 		else, the number of pages reclaimed
L
Linus Torvalds 已提交
1663
 */
1664
static unsigned long do_try_to_free_pages(struct zonelist *zonelist,
1665
					struct scan_control *sc)
L
Linus Torvalds 已提交
1666 1667
{
	int priority;
1668
	unsigned long ret = 0;
1669
	unsigned long total_scanned = 0;
L
Linus Torvalds 已提交
1670 1671
	struct reclaim_state *reclaim_state = current->reclaim_state;
	unsigned long lru_pages = 0;
1672
	struct zoneref *z;
1673
	struct zone *zone;
1674
	enum zone_type high_zoneidx = gfp_zone(sc->gfp_mask);
L
Linus Torvalds 已提交
1675

1676 1677
	delayacct_freepages_start();

1678
	if (scanning_global_lru(sc))
1679 1680 1681 1682
		count_vm_event(ALLOCSTALL);
	/*
	 * mem_cgroup will not do shrink_slab.
	 */
1683
	if (scanning_global_lru(sc)) {
1684
		for_each_zone_zonelist(zone, z, zonelist, high_zoneidx) {
L
Linus Torvalds 已提交
1685

1686 1687
			if (!cpuset_zone_allowed_hardwall(zone, GFP_KERNEL))
				continue;
L
Linus Torvalds 已提交
1688

1689
			lru_pages += zone_lru_pages(zone);
1690
		}
L
Linus Torvalds 已提交
1691 1692 1693
	}

	for (priority = DEF_PRIORITY; priority >= 0; priority--) {
1694
		sc->nr_scanned = 0;
1695 1696
		if (!priority)
			disable_swap_token();
1697
		shrink_zones(priority, zonelist, sc);
1698 1699 1700 1701
		/*
		 * Don't shrink slabs when reclaiming memory from
		 * over limit cgroups
		 */
1702
		if (scanning_global_lru(sc)) {
1703
			shrink_slab(sc->nr_scanned, sc->gfp_mask, lru_pages);
1704
			if (reclaim_state) {
1705
				sc->nr_reclaimed += reclaim_state->reclaimed_slab;
1706 1707
				reclaim_state->reclaimed_slab = 0;
			}
L
Linus Torvalds 已提交
1708
		}
1709
		total_scanned += sc->nr_scanned;
1710 1711
		if (sc->nr_reclaimed >= sc->swap_cluster_max) {
			ret = sc->nr_reclaimed;
L
Linus Torvalds 已提交
1712 1713 1714 1715 1716 1717 1718 1719 1720 1721
			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.
		 */
1722 1723
		if (total_scanned > sc->swap_cluster_max +
					sc->swap_cluster_max / 2) {
1724
			wakeup_pdflush(laptop_mode ? 0 : total_scanned);
1725
			sc->may_writepage = 1;
L
Linus Torvalds 已提交
1726 1727 1728
		}

		/* Take a nap, wait for some writeback to complete */
1729
		if (sc->nr_scanned && priority < DEF_PRIORITY - 2)
1730
			congestion_wait(WRITE, HZ/10);
L
Linus Torvalds 已提交
1731
	}
1732
	/* top priority shrink_zones still had more to do? don't OOM, then */
1733
	if (!sc->all_unreclaimable && scanning_global_lru(sc))
1734
		ret = sc->nr_reclaimed;
L
Linus Torvalds 已提交
1735
out:
1736 1737 1738 1739 1740 1741 1742 1743 1744
	/*
	 * 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
1745

1746
	if (scanning_global_lru(sc)) {
1747
		for_each_zone_zonelist(zone, z, zonelist, high_zoneidx) {
1748 1749 1750 1751 1752 1753 1754 1755

			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
1756

1757 1758
	delayacct_freepages_end();

L
Linus Torvalds 已提交
1759 1760 1761
	return ret;
}

1762
unsigned long try_to_free_pages(struct zonelist *zonelist, int order,
1763
				gfp_t gfp_mask, nodemask_t *nodemask)
1764 1765 1766 1767 1768
{
	struct scan_control sc = {
		.gfp_mask = gfp_mask,
		.may_writepage = !laptop_mode,
		.swap_cluster_max = SWAP_CLUSTER_MAX,
1769
		.may_unmap = 1,
1770
		.may_swap = 1,
1771 1772 1773 1774
		.swappiness = vm_swappiness,
		.order = order,
		.mem_cgroup = NULL,
		.isolate_pages = isolate_pages_global,
1775
		.nodemask = nodemask,
1776 1777
	};

1778
	return do_try_to_free_pages(zonelist, &sc);
1779 1780
}

1781
#ifdef CONFIG_CGROUP_MEM_RES_CTLR
1782

1783
unsigned long try_to_free_mem_cgroup_pages(struct mem_cgroup *mem_cont,
K
KOSAKI Motohiro 已提交
1784 1785 1786
					   gfp_t gfp_mask,
					   bool noswap,
					   unsigned int swappiness)
1787 1788 1789
{
	struct scan_control sc = {
		.may_writepage = !laptop_mode,
1790
		.may_unmap = 1,
1791
		.may_swap = !noswap,
1792
		.swap_cluster_max = SWAP_CLUSTER_MAX,
K
KOSAKI Motohiro 已提交
1793
		.swappiness = swappiness,
1794 1795 1796
		.order = 0,
		.mem_cgroup = mem_cont,
		.isolate_pages = mem_cgroup_isolate_pages,
1797
		.nodemask = NULL, /* we don't care the placement */
1798
	};
1799
	struct zonelist *zonelist;
1800

1801 1802 1803 1804
	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);
1805 1806 1807
}
#endif

L
Linus Torvalds 已提交
1808 1809
/*
 * For kswapd, balance_pgdat() will work across all this node's zones until
1810
 * they are all at high_wmark_pages(zone).
L
Linus Torvalds 已提交
1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822
 *
 * 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
1823 1824 1825 1826 1827
 * 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
1828
 */
1829
static unsigned long balance_pgdat(pg_data_t *pgdat, int order)
L
Linus Torvalds 已提交
1830 1831 1832 1833
{
	int all_zones_ok;
	int priority;
	int i;
1834
	unsigned long total_scanned;
L
Linus Torvalds 已提交
1835
	struct reclaim_state *reclaim_state = current->reclaim_state;
1836 1837
	struct scan_control sc = {
		.gfp_mask = GFP_KERNEL,
1838
		.may_unmap = 1,
1839
		.may_swap = 1,
1840 1841
		.swap_cluster_max = SWAP_CLUSTER_MAX,
		.swappiness = vm_swappiness,
A
Andy Whitcroft 已提交
1842
		.order = order,
1843 1844
		.mem_cgroup = NULL,
		.isolate_pages = isolate_pages_global,
1845
	};
1846 1847
	/*
	 * temp_priority is used to remember the scanning priority at which
1848 1849
	 * this zone was successfully refilled to
	 * free_pages == high_wmark_pages(zone).
1850 1851
	 */
	int temp_priority[MAX_NR_ZONES];
L
Linus Torvalds 已提交
1852 1853 1854

loop_again:
	total_scanned = 0;
1855
	sc.nr_reclaimed = 0;
C
Christoph Lameter 已提交
1856
	sc.may_writepage = !laptop_mode;
1857
	count_vm_event(PAGEOUTRUN);
L
Linus Torvalds 已提交
1858

1859 1860
	for (i = 0; i < pgdat->nr_zones; i++)
		temp_priority[i] = DEF_PRIORITY;
L
Linus Torvalds 已提交
1861 1862 1863 1864 1865

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

1866 1867 1868 1869
		/* The swap token gets in the way of swapout... */
		if (!priority)
			disable_swap_token();

L
Linus Torvalds 已提交
1870 1871
		all_zones_ok = 1;

1872 1873 1874 1875 1876 1877
		/*
		 * 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
1878

1879 1880
			if (!populated_zone(zone))
				continue;
L
Linus Torvalds 已提交
1881

1882 1883
			if (zone_is_all_unreclaimable(zone) &&
			    priority != DEF_PRIORITY)
1884
				continue;
L
Linus Torvalds 已提交
1885

1886 1887 1888 1889
			/*
			 * Do some background aging of the anon list, to give
			 * pages a chance to be referenced before reclaiming.
			 */
1890
			if (inactive_anon_is_low(zone, &sc))
1891 1892 1893
				shrink_active_list(SWAP_CLUSTER_MAX, zone,
							&sc, priority, 0);

1894 1895
			if (!zone_watermark_ok(zone, order,
					high_wmark_pages(zone), 0, 0)) {
1896
				end_zone = i;
A
Andrew Morton 已提交
1897
				break;
L
Linus Torvalds 已提交
1898 1899
			}
		}
A
Andrew Morton 已提交
1900 1901 1902
		if (i < 0)
			goto out;

L
Linus Torvalds 已提交
1903 1904 1905
		for (i = 0; i <= end_zone; i++) {
			struct zone *zone = pgdat->node_zones + i;

1906
			lru_pages += zone_lru_pages(zone);
L
Linus Torvalds 已提交
1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919
		}

		/*
		 * 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;
1920
			int nr_slab;
L
Linus Torvalds 已提交
1921

1922
			if (!populated_zone(zone))
L
Linus Torvalds 已提交
1923 1924
				continue;

1925 1926
			if (zone_is_all_unreclaimable(zone) &&
					priority != DEF_PRIORITY)
L
Linus Torvalds 已提交
1927 1928
				continue;

1929 1930
			if (!zone_watermark_ok(zone, order,
					high_wmark_pages(zone), end_zone, 0))
1931
				all_zones_ok = 0;
1932
			temp_priority[i] = priority;
L
Linus Torvalds 已提交
1933
			sc.nr_scanned = 0;
1934
			note_zone_scanning_priority(zone, priority);
1935 1936 1937 1938
			/*
			 * We put equal pressure on every zone, unless one
			 * zone has way too many pages free already.
			 */
1939 1940
			if (!zone_watermark_ok(zone, order,
					8*high_wmark_pages(zone), end_zone, 0))
1941
				shrink_zone(priority, zone, &sc);
L
Linus Torvalds 已提交
1942
			reclaim_state->reclaimed_slab = 0;
1943 1944
			nr_slab = shrink_slab(sc.nr_scanned, GFP_KERNEL,
						lru_pages);
1945
			sc.nr_reclaimed += reclaim_state->reclaimed_slab;
L
Linus Torvalds 已提交
1946
			total_scanned += sc.nr_scanned;
1947
			if (zone_is_all_unreclaimable(zone))
L
Linus Torvalds 已提交
1948
				continue;
1949
			if (nr_slab == 0 && zone->pages_scanned >=
1950
						(zone_lru_pages(zone) * 6))
1951 1952
					zone_set_flag(zone,
						      ZONE_ALL_UNRECLAIMABLE);
L
Linus Torvalds 已提交
1953 1954 1955 1956 1957 1958
			/*
			 * 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 &&
1959
			    total_scanned > sc.nr_reclaimed + sc.nr_reclaimed / 2)
L
Linus Torvalds 已提交
1960 1961 1962 1963 1964 1965 1966 1967
				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.
		 */
1968
		if (total_scanned && priority < DEF_PRIORITY - 2)
1969
			congestion_wait(WRITE, HZ/10);
L
Linus Torvalds 已提交
1970 1971 1972 1973 1974 1975 1976

		/*
		 * 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.
		 */
1977
		if (sc.nr_reclaimed >= SWAP_CLUSTER_MAX)
L
Linus Torvalds 已提交
1978 1979 1980
			break;
	}
out:
1981 1982 1983 1984 1985
	/*
	 * 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
1986 1987 1988
	for (i = 0; i < pgdat->nr_zones; i++) {
		struct zone *zone = pgdat->node_zones + i;

1989
		zone->prev_priority = temp_priority[i];
L
Linus Torvalds 已提交
1990 1991 1992
	}
	if (!all_zones_ok) {
		cond_resched();
1993 1994 1995

		try_to_freeze();

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
		/*
		 * 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
2013 2014 2015
		goto loop_again;
	}

2016
	return sc.nr_reclaimed;
L
Linus Torvalds 已提交
2017 2018 2019 2020
}

/*
 * The background pageout daemon, started as a kernel thread
2021
 * from the init process.
L
Linus Torvalds 已提交
2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040
 *
 * 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,
	};
2041
	const struct cpumask *cpumask = cpumask_of_node(pgdat->node_id);
L
Linus Torvalds 已提交
2042

2043 2044
	lockdep_set_current_reclaim_state(GFP_KERNEL);

R
Rusty Russell 已提交
2045
	if (!cpumask_empty(cpumask))
2046
		set_cpus_allowed_ptr(tsk, cpumask);
L
Linus Torvalds 已提交
2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060
	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).
	 */
2061
	tsk->flags |= PF_MEMALLOC | PF_SWAPWRITE | PF_KSWAPD;
2062
	set_freezable();
L
Linus Torvalds 已提交
2063 2064 2065 2066

	order = 0;
	for ( ; ; ) {
		unsigned long new_order;
2067

L
Linus Torvalds 已提交
2068 2069 2070 2071 2072 2073 2074 2075 2076 2077
		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 {
2078 2079 2080
			if (!freezing(current))
				schedule();

L
Linus Torvalds 已提交
2081 2082 2083 2084
			order = pgdat->kswapd_max_order;
		}
		finish_wait(&pgdat->kswapd_wait, &wait);

2085 2086 2087 2088 2089 2090
		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
2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101
	}
	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;

2102
	if (!populated_zone(zone))
L
Linus Torvalds 已提交
2103 2104 2105
		return;

	pgdat = zone->zone_pgdat;
2106
	if (zone_watermark_ok(zone, order, low_wmark_pages(zone), 0, 0))
L
Linus Torvalds 已提交
2107 2108 2109
		return;
	if (pgdat->kswapd_max_order < order)
		pgdat->kswapd_max_order = order;
2110
	if (!cpuset_zone_allowed_hardwall(zone, GFP_KERNEL))
L
Linus Torvalds 已提交
2111
		return;
2112
	if (!waitqueue_active(&pgdat->kswapd_wait))
L
Linus Torvalds 已提交
2113
		return;
2114
	wake_up_interruptible(&pgdat->kswapd_wait);
L
Linus Torvalds 已提交
2115 2116
}

2117 2118 2119 2120 2121 2122 2123 2124
unsigned long global_lru_pages(void)
{
	return global_page_state(NR_ACTIVE_ANON)
		+ global_page_state(NR_ACTIVE_FILE)
		+ global_page_state(NR_INACTIVE_ANON)
		+ global_page_state(NR_INACTIVE_FILE);
}

2125
#ifdef CONFIG_HIBERNATION
L
Linus Torvalds 已提交
2126
/*
2127
 * Helper function for shrink_all_memory().  Tries to reclaim 'nr_pages' pages
2128
 * from LRU lists system-wide, for given pass and priority.
2129 2130 2131
 *
 * For pass > 3 we also try to shrink the LRU lists that contain a few pages
 */
2132
static void shrink_all_zones(unsigned long nr_pages, int prio,
2133
				      int pass, struct scan_control *sc)
2134 2135
{
	struct zone *zone;
2136
	unsigned long nr_reclaimed = 0;
2137

2138
	for_each_populated_zone(zone) {
2139
		enum lru_list l;
2140

2141
		if (zone_is_all_unreclaimable(zone) && prio != DEF_PRIORITY)
2142 2143
			continue;

L
Lee Schermerhorn 已提交
2144
		for_each_evictable_lru(l) {
2145 2146 2147
			enum zone_stat_item ls = NR_LRU_BASE + l;
			unsigned long lru_pages = zone_page_state(zone, ls);

L
Lee Schermerhorn 已提交
2148
			/* For pass = 0, we don't shrink the active list */
2149 2150
			if (pass == 0 && (l == LRU_ACTIVE_ANON ||
						l == LRU_ACTIVE_FILE))
2151 2152
				continue;

2153 2154
			zone->lru[l].nr_saved_scan += (lru_pages >> prio) + 1;
			if (zone->lru[l].nr_saved_scan >= nr_pages || pass > 3) {
2155 2156
				unsigned long nr_to_scan;

2157
				zone->lru[l].nr_saved_scan = 0;
2158
				nr_to_scan = min(nr_pages, lru_pages);
2159
				nr_reclaimed += shrink_list(l, nr_to_scan, zone,
2160
								sc, prio);
2161
				if (nr_reclaimed >= nr_pages) {
2162
					sc->nr_reclaimed += nr_reclaimed;
2163 2164
					return;
				}
2165 2166 2167
			}
		}
	}
2168
	sc->nr_reclaimed += nr_reclaimed;
2169 2170 2171 2172 2173 2174 2175 2176 2177
}

/*
 * 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
2178
 */
2179
unsigned long shrink_all_memory(unsigned long nr_pages)
L
Linus Torvalds 已提交
2180
{
2181 2182 2183 2184 2185
	unsigned long lru_pages, nr_slab;
	int pass;
	struct reclaim_state reclaim_state;
	struct scan_control sc = {
		.gfp_mask = GFP_KERNEL,
2186
		.may_unmap = 0,
2187
		.may_writepage = 1,
2188
		.isolate_pages = isolate_pages_global,
2189
		.nr_reclaimed = 0,
L
Linus Torvalds 已提交
2190 2191 2192
	};

	current->reclaim_state = &reclaim_state;
2193

2194
	lru_pages = global_lru_pages();
2195
	nr_slab = global_page_state(NR_SLAB_RECLAIMABLE);
2196 2197 2198 2199 2200
	/* 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
2201
			break;
2202

2203 2204
		sc.nr_reclaimed += reclaim_state.reclaimed_slab;
		if (sc.nr_reclaimed >= nr_pages)
2205 2206 2207
			goto out;

		nr_slab -= reclaim_state.reclaimed_slab;
L
Linus Torvalds 已提交
2208
	}
2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221

	/*
	 * 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 */
2222
		if (pass > 2)
2223
			sc.may_unmap = 1;
2224 2225

		for (prio = DEF_PRIORITY; prio >= 0; prio--) {
2226
			unsigned long nr_to_scan = nr_pages - sc.nr_reclaimed;
2227 2228

			sc.nr_scanned = 0;
2229
			sc.swap_cluster_max = nr_to_scan;
2230 2231
			shrink_all_zones(nr_to_scan, prio, pass, &sc);
			if (sc.nr_reclaimed >= nr_pages)
2232 2233 2234
				goto out;

			reclaim_state.reclaimed_slab = 0;
2235
			shrink_slab(sc.nr_scanned, sc.gfp_mask,
2236
					global_lru_pages());
2237 2238
			sc.nr_reclaimed += reclaim_state.reclaimed_slab;
			if (sc.nr_reclaimed >= nr_pages)
2239 2240 2241
				goto out;

			if (sc.nr_scanned && prio < DEF_PRIORITY - 2)
2242
				congestion_wait(WRITE, HZ / 10);
2243
		}
2244
	}
2245 2246

	/*
2247 2248
	 * If sc.nr_reclaimed = 0, we could not shrink LRUs, but there may be
	 * something in slab caches
2249
	 */
2250
	if (!sc.nr_reclaimed) {
2251 2252
		do {
			reclaim_state.reclaimed_slab = 0;
2253
			shrink_slab(nr_pages, sc.gfp_mask, global_lru_pages());
2254 2255 2256
			sc.nr_reclaimed += reclaim_state.reclaimed_slab;
		} while (sc.nr_reclaimed < nr_pages &&
				reclaim_state.reclaimed_slab > 0);
2257
	}
2258

2259

2260
out:
L
Linus Torvalds 已提交
2261
	current->reclaim_state = NULL;
2262

2263
	return sc.nr_reclaimed;
L
Linus Torvalds 已提交
2264
}
2265
#endif /* CONFIG_HIBERNATION */
L
Linus Torvalds 已提交
2266 2267 2268 2269 2270

/* 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. */
2271
static int __devinit cpu_callback(struct notifier_block *nfb,
2272
				  unsigned long action, void *hcpu)
L
Linus Torvalds 已提交
2273
{
2274
	int nid;
L
Linus Torvalds 已提交
2275

2276
	if (action == CPU_ONLINE || action == CPU_ONLINE_FROZEN) {
2277
		for_each_node_state(nid, N_HIGH_MEMORY) {
2278
			pg_data_t *pgdat = NODE_DATA(nid);
2279 2280 2281
			const struct cpumask *mask;

			mask = cpumask_of_node(pgdat->node_id);
2282

2283
			if (cpumask_any_and(cpu_online_mask, mask) < nr_cpu_ids)
L
Linus Torvalds 已提交
2284
				/* One of our CPUs online: restore mask */
2285
				set_cpus_allowed_ptr(pgdat->kswapd, mask);
L
Linus Torvalds 已提交
2286 2287 2288 2289 2290
		}
	}
	return NOTIFY_OK;
}

2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312
/*
 * 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
2313 2314
static int __init kswapd_init(void)
{
2315
	int nid;
2316

L
Linus Torvalds 已提交
2317
	swap_setup();
2318
	for_each_node_state(nid, N_HIGH_MEMORY)
2319
 		kswapd_run(nid);
L
Linus Torvalds 已提交
2320 2321 2322 2323 2324
	hotcpu_notifier(cpu_callback, 0);
	return 0;
}

module_init(kswapd_init)
2325 2326 2327 2328 2329 2330 2331 2332 2333 2334

#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;

2335
#define RECLAIM_OFF 0
2336
#define RECLAIM_ZONE (1<<0)	/* Run shrink_inactive_list on the zone */
2337 2338 2339
#define RECLAIM_WRITE (1<<1)	/* Writeout pages during reclaim */
#define RECLAIM_SWAP (1<<2)	/* Swap pages out during reclaim */

2340 2341 2342 2343 2344 2345 2346
/*
 * 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

2347 2348 2349 2350 2351 2352
/*
 * Percentage of pages in a zone that must be unmapped for zone_reclaim to
 * occur.
 */
int sysctl_min_unmapped_ratio = 1;

2353 2354 2355 2356 2357 2358
/*
 * 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;

2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400
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;
}

2401 2402 2403
/*
 * Try to free up some pages from this zone through reclaim.
 */
2404
static int __zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
2405
{
2406
	/* Minimum pages needed in order to stay on node */
2407
	const unsigned long nr_pages = 1 << order;
2408 2409
	struct task_struct *p = current;
	struct reclaim_state reclaim_state;
2410
	int priority;
2411 2412
	struct scan_control sc = {
		.may_writepage = !!(zone_reclaim_mode & RECLAIM_WRITE),
2413
		.may_unmap = !!(zone_reclaim_mode & RECLAIM_SWAP),
2414
		.may_swap = 1,
2415 2416
		.swap_cluster_max = max_t(unsigned long, nr_pages,
					SWAP_CLUSTER_MAX),
2417
		.gfp_mask = gfp_mask,
2418
		.swappiness = vm_swappiness,
2419
		.order = order,
2420
		.isolate_pages = isolate_pages_global,
2421
	};
2422
	unsigned long slab_reclaimable;
2423 2424 2425

	disable_swap_token();
	cond_resched();
2426 2427 2428 2429 2430 2431
	/*
	 * 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;
2432 2433
	reclaim_state.reclaimed_slab = 0;
	p->reclaim_state = &reclaim_state;
2434

2435
	if (zone_pagecache_reclaimable(zone) > zone->min_unmapped_pages) {
2436 2437 2438 2439 2440 2441
		/*
		 * Free memory by calling shrink zone with increasing
		 * priorities until we have enough memory freed.
		 */
		priority = ZONE_RECLAIM_PRIORITY;
		do {
2442
			note_zone_scanning_priority(zone, priority);
2443
			shrink_zone(priority, zone, &sc);
2444
			priority--;
2445
		} while (priority >= 0 && sc.nr_reclaimed < nr_pages);
2446
	}
2447

2448 2449
	slab_reclaimable = zone_page_state(zone, NR_SLAB_RECLAIMABLE);
	if (slab_reclaimable > zone->min_slab_pages) {
2450
		/*
2451
		 * shrink_slab() does not currently allow us to determine how
2452 2453 2454 2455
		 * 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.
2456
		 *
2457 2458
		 * Note that shrink_slab will free memory on all zones and may
		 * take a long time.
2459
		 */
2460
		while (shrink_slab(sc.nr_scanned, gfp_mask, order) &&
2461 2462
			zone_page_state(zone, NR_SLAB_RECLAIMABLE) >
				slab_reclaimable - nr_pages)
2463
			;
2464 2465 2466 2467 2468

		/*
		 * Update nr_reclaimed by the number of slab pages we
		 * reclaimed from this zone.
		 */
2469
		sc.nr_reclaimed += slab_reclaimable -
2470
			zone_page_state(zone, NR_SLAB_RECLAIMABLE);
2471 2472
	}

2473
	p->reclaim_state = NULL;
2474
	current->flags &= ~(PF_MEMALLOC | PF_SWAPWRITE);
2475
	return sc.nr_reclaimed >= nr_pages;
2476
}
2477 2478 2479 2480

int zone_reclaim(struct zone *zone, gfp_t gfp_mask, unsigned int order)
{
	int node_id;
2481
	int ret;
2482 2483

	/*
2484 2485
	 * Zone reclaim reclaims unmapped file backed pages and
	 * slab pages if we are over the defined limits.
2486
	 *
2487 2488 2489 2490 2491
	 * 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.
2492
	 */
2493 2494
	if (zone_pagecache_reclaimable(zone) <= zone->min_unmapped_pages &&
	    zone_page_state(zone, NR_SLAB_RECLAIMABLE) <= zone->min_slab_pages)
2495
		return ZONE_RECLAIM_FULL;
2496

2497
	if (zone_is_all_unreclaimable(zone))
2498
		return ZONE_RECLAIM_FULL;
2499

2500
	/*
2501
	 * Do not scan if the allocation should not be delayed.
2502
	 */
2503
	if (!(gfp_mask & __GFP_WAIT) || (current->flags & PF_MEMALLOC))
2504
		return ZONE_RECLAIM_NOSCAN;
2505 2506 2507 2508 2509 2510 2511

	/*
	 * 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.
	 */
2512
	node_id = zone_to_nid(zone);
2513
	if (node_state(node_id, N_CPU) && node_id != numa_node_id())
2514
		return ZONE_RECLAIM_NOSCAN;
2515 2516

	if (zone_test_and_set_flag(zone, ZONE_RECLAIM_LOCKED))
2517 2518
		return ZONE_RECLAIM_NOSCAN;

2519 2520 2521
	ret = __zone_reclaim(zone, gfp_mask, order);
	zone_clear_flag(zone, ZONE_RECLAIM_LOCKED);

2522 2523 2524
	if (!ret)
		count_vm_event(PGSCAN_ZONE_RECLAIM_FAILED);

2525
	return ret;
2526
}
2527
#endif
L
Lee Schermerhorn 已提交
2528 2529 2530 2531 2532 2533 2534

/*
 * 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
2535 2536
 * 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
2537 2538
 *
 * Reasons page might not be evictable:
2539
 * (1) page's mapping marked unevictable
N
Nick Piggin 已提交
2540
 * (2) page is part of an mlocked VMA
2541
 *
L
Lee Schermerhorn 已提交
2542 2543 2544 2545
 */
int page_evictable(struct page *page, struct vm_area_struct *vma)
{

2546 2547 2548
	if (mapping_unevictable(page_mapping(page)))
		return 0;

N
Nick Piggin 已提交
2549 2550
	if (PageMlocked(page) || (vma && is_mlocked_vma(vma, page)))
		return 0;
L
Lee Schermerhorn 已提交
2551 2552 2553

	return 1;
}
2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573

/**
 * 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);
2574

2575 2576
		__dec_zone_state(zone, NR_UNEVICTABLE);
		list_move(&page->lru, &zone->lru[l].list);
K
KAMEZAWA Hiroyuki 已提交
2577
		mem_cgroup_move_lists(page, LRU_UNEVICTABLE, l);
2578 2579 2580 2581 2582 2583 2584 2585
		__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
2586
		mem_cgroup_rotate_lru_list(page, LRU_UNEVICTABLE);
2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645
		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);
	}

}
2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657

/**
 * 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 */
2658
static void scan_zone_unevictable_pages(struct zone *zone)
2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699
{
	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.
 */
2700
static void scan_all_zones_unevictable_pages(void)
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 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774
{
	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);
}