swapfile.c 55.4 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
/*
 *  linux/mm/swapfile.c
 *
 *  Copyright (C) 1991, 1992, 1993, 1994  Linus Torvalds
 *  Swap reorganised 29.12.95, Stephen Tweedie
 */

#include <linux/mm.h>
#include <linux/hugetlb.h>
#include <linux/mman.h>
#include <linux/slab.h>
#include <linux/kernel_stat.h>
#include <linux/swap.h>
#include <linux/vmalloc.h>
#include <linux/pagemap.h>
#include <linux/namei.h>
#include <linux/shm.h>
#include <linux/blkdev.h>
19
#include <linux/random.h>
L
Linus Torvalds 已提交
20 21 22 23 24 25 26 27
#include <linux/writeback.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <linux/init.h>
#include <linux/module.h>
#include <linux/rmap.h>
#include <linux/security.h>
#include <linux/backing-dev.h>
I
Ingo Molnar 已提交
28
#include <linux/mutex.h>
29
#include <linux/capability.h>
L
Linus Torvalds 已提交
30
#include <linux/syscalls.h>
31
#include <linux/memcontrol.h>
L
Linus Torvalds 已提交
32 33 34 35

#include <asm/pgtable.h>
#include <asm/tlbflush.h>
#include <linux/swapops.h>
36
#include <linux/page_cgroup.h>
L
Linus Torvalds 已提交
37

A
Adrian Bunk 已提交
38 39
static DEFINE_SPINLOCK(swap_lock);
static unsigned int nr_swapfiles;
40
long nr_swap_pages;
L
Linus Torvalds 已提交
41 42
long total_swap_pages;
static int swap_overflow;
43
static int least_priority;
L
Linus Torvalds 已提交
44 45 46 47 48 49

static const char Bad_file[] = "Bad swap file entry ";
static const char Unused_file[] = "Unused swap file entry ";
static const char Bad_offset[] = "Bad swap offset entry ";
static const char Unused_offset[] = "Unused swap offset entry ";

A
Adrian Bunk 已提交
50
static struct swap_list_t swap_list = {-1, -1};
L
Linus Torvalds 已提交
51

52
static struct swap_info_struct swap_info[MAX_SWAPFILES];
L
Linus Torvalds 已提交
53

I
Ingo Molnar 已提交
54
static DEFINE_MUTEX(swapon_mutex);
L
Linus Torvalds 已提交
55

56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81
/* For reference count accounting in swap_map */
/* enum for swap_map[] handling. internal use only */
enum {
	SWAP_MAP = 0,	/* ops for reference from swap users */
	SWAP_CACHE,	/* ops for reference from swap cache */
};

static inline int swap_count(unsigned short ent)
{
	return ent & SWAP_COUNT_MASK;
}

static inline bool swap_has_cache(unsigned short ent)
{
	return !!(ent & SWAP_HAS_CACHE);
}

static inline unsigned short encode_swapmap(int count, bool has_cache)
{
	unsigned short ret = count;

	if (has_cache)
		return SWAP_HAS_CACHE | ret;
	return ret;
}

82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107
/* returnes 1 if swap entry is freed */
static int
__try_to_reclaim_swap(struct swap_info_struct *si, unsigned long offset)
{
	int type = si - swap_info;
	swp_entry_t entry = swp_entry(type, offset);
	struct page *page;
	int ret = 0;

	page = find_get_page(&swapper_space, entry.val);
	if (!page)
		return 0;
	/*
	 * This function is called from scan_swap_map() and it's called
	 * by vmscan.c at reclaiming pages. So, we hold a lock on a page, here.
	 * We have to use trylock for avoiding deadlock. This is a special
	 * case and you should use try_to_free_swap() with explicit lock_page()
	 * in usual operations.
	 */
	if (trylock_page(page)) {
		ret = try_to_free_swap(page);
		unlock_page(page);
	}
	page_cache_release(page);
	return ret;
}
108

L
Linus Torvalds 已提交
109 110
/*
 * We need this because the bdev->unplug_fn can sleep and we cannot
111
 * hold swap_lock while calling the unplug_fn. And swap_lock
I
Ingo Molnar 已提交
112
 * cannot be turned into a mutex.
L
Linus Torvalds 已提交
113 114 115 116 117 118 119 120
 */
static DECLARE_RWSEM(swap_unplug_sem);

void swap_unplug_io_fn(struct backing_dev_info *unused_bdi, struct page *page)
{
	swp_entry_t entry;

	down_read(&swap_unplug_sem);
H
Hugh Dickins 已提交
121
	entry.val = page_private(page);
L
Linus Torvalds 已提交
122 123 124 125 126 127 128
	if (PageSwapCache(page)) {
		struct block_device *bdev = swap_info[swp_type(entry)].bdev;
		struct backing_dev_info *bdi;

		/*
		 * If the page is removed from swapcache from under us (with a
		 * racy try_to_unuse/swapoff) we need an additional reference
H
Hugh Dickins 已提交
129 130
		 * count to avoid reading garbage from page_private(page) above.
		 * If the WARN_ON triggers during a swapoff it maybe the race
L
Linus Torvalds 已提交
131 132 133 134 135 136
		 * condition and it's harmless. However if it triggers without
		 * swapoff it signals a problem.
		 */
		WARN_ON(page_count(page) <= 1);

		bdi = bdev->bd_inode->i_mapping->backing_dev_info;
M
McMullan, Jason 已提交
137
		blk_run_backing_dev(bdi, page);
L
Linus Torvalds 已提交
138 139 140 141
	}
	up_read(&swap_unplug_sem);
}

142 143 144 145 146 147 148 149 150 151 152
/*
 * swapon tell device that all the old swap contents can be discarded,
 * to allow the swap device to optimize its wear-levelling.
 */
static int discard_swap(struct swap_info_struct *si)
{
	struct swap_extent *se;
	int err = 0;

	list_for_each_entry(se, &si->extent_list, list) {
		sector_t start_block = se->start_block << (PAGE_SHIFT - 9);
153
		sector_t nr_blocks = (sector_t)se->nr_pages << (PAGE_SHIFT - 9);
154 155 156 157 158 159 160 161 162 163

		if (se->start_page == 0) {
			/* Do not discard the swap header page! */
			start_block += 1 << (PAGE_SHIFT - 9);
			nr_blocks -= 1 << (PAGE_SHIFT - 9);
			if (!nr_blocks)
				continue;
		}

		err = blkdev_issue_discard(si->bdev, start_block,
164 165
						nr_blocks, GFP_KERNEL,
						DISCARD_FL_BARRIER);
166 167 168 169 170 171 172 173
		if (err)
			break;

		cond_resched();
	}
	return err;		/* That will often be -EOPNOTSUPP */
}

174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190
/*
 * swap allocation tell device that a cluster of swap can now be discarded,
 * to allow the swap device to optimize its wear-levelling.
 */
static void discard_swap_cluster(struct swap_info_struct *si,
				 pgoff_t start_page, pgoff_t nr_pages)
{
	struct swap_extent *se = si->curr_swap_extent;
	int found_extent = 0;

	while (nr_pages) {
		struct list_head *lh;

		if (se->start_page <= start_page &&
		    start_page < se->start_page + se->nr_pages) {
			pgoff_t offset = start_page - se->start_page;
			sector_t start_block = se->start_block + offset;
191
			sector_t nr_blocks = se->nr_pages - offset;
192 193 194 195 196 197 198 199 200 201 202 203

			if (nr_blocks > nr_pages)
				nr_blocks = nr_pages;
			start_page += nr_blocks;
			nr_pages -= nr_blocks;

			if (!found_extent++)
				si->curr_swap_extent = se;

			start_block <<= PAGE_SHIFT - 9;
			nr_blocks <<= PAGE_SHIFT - 9;
			if (blkdev_issue_discard(si->bdev, start_block,
204 205
							nr_blocks, GFP_NOIO,
							DISCARD_FL_BARRIER))
206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221
				break;
		}

		lh = se->list.next;
		if (lh == &si->extent_list)
			lh = lh->next;
		se = list_entry(lh, struct swap_extent, list);
	}
}

static int wait_for_discard(void *word)
{
	schedule();
	return 0;
}

222 223 224
#define SWAPFILE_CLUSTER	256
#define LATENCY_LIMIT		256

225 226
static inline unsigned long scan_swap_map(struct swap_info_struct *si,
					  int cache)
L
Linus Torvalds 已提交
227
{
228
	unsigned long offset;
229
	unsigned long scan_base;
230
	unsigned long last_in_cluster = 0;
231
	int latency_ration = LATENCY_LIMIT;
232
	int found_free_cluster = 0;
233

234
	/*
235 236 237 238 239 240 241
	 * We try to cluster swap pages by allocating them sequentially
	 * in swap.  Once we've allocated SWAPFILE_CLUSTER pages this
	 * way, however, we resort to first-free allocation, starting
	 * a new cluster.  This prevents us from scattering swap pages
	 * all over the entire swap partition, so that we reduce
	 * overall disk seek times between swap pages.  -- sct
	 * But we do now try to find an empty cluster.  -Andrea
242
	 * And we let swap pages go all over an SSD partition.  Hugh
243 244
	 */

245
	si->flags += SWP_SCANNING;
246
	scan_base = offset = si->cluster_next;
247 248 249 250 251 252

	if (unlikely(!si->cluster_nr--)) {
		if (si->pages - si->inuse_pages < SWAPFILE_CLUSTER) {
			si->cluster_nr = SWAPFILE_CLUSTER - 1;
			goto checks;
		}
253 254 255 256 257 258 259 260 261 262 263 264 265
		if (si->flags & SWP_DISCARDABLE) {
			/*
			 * Start range check on racing allocations, in case
			 * they overlap the cluster we eventually decide on
			 * (we scan without swap_lock to allow preemption).
			 * It's hardly conceivable that cluster_nr could be
			 * wrapped during our scan, but don't depend on it.
			 */
			if (si->lowest_alloc)
				goto checks;
			si->lowest_alloc = si->max;
			si->highest_alloc = 0;
		}
266
		spin_unlock(&swap_lock);
267

268 269 270 271 272 273 274 275 276 277
		/*
		 * If seek is expensive, start searching for new cluster from
		 * start of partition, to minimize the span of allocated swap.
		 * But if seek is cheap, search from our current position, so
		 * that swap is allocated from all over the partition: if the
		 * Flash Translation Layer only remaps within limited zones,
		 * we don't want to wear out the first zone too quickly.
		 */
		if (!(si->flags & SWP_SOLIDSTATE))
			scan_base = offset = si->lowest_bit;
278 279 280 281
		last_in_cluster = offset + SWAPFILE_CLUSTER - 1;

		/* Locate the first empty (unaligned) cluster */
		for (; last_in_cluster <= si->highest_bit; offset++) {
L
Linus Torvalds 已提交
282
			if (si->swap_map[offset])
283 284
				last_in_cluster = offset + SWAPFILE_CLUSTER;
			else if (offset == last_in_cluster) {
285
				spin_lock(&swap_lock);
286 287 288
				offset -= SWAPFILE_CLUSTER - 1;
				si->cluster_next = offset;
				si->cluster_nr = SWAPFILE_CLUSTER - 1;
289
				found_free_cluster = 1;
290
				goto checks;
L
Linus Torvalds 已提交
291
			}
292 293 294 295
			if (unlikely(--latency_ration < 0)) {
				cond_resched();
				latency_ration = LATENCY_LIMIT;
			}
296
		}
297 298

		offset = si->lowest_bit;
299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319
		last_in_cluster = offset + SWAPFILE_CLUSTER - 1;

		/* Locate the first empty (unaligned) cluster */
		for (; last_in_cluster < scan_base; offset++) {
			if (si->swap_map[offset])
				last_in_cluster = offset + SWAPFILE_CLUSTER;
			else if (offset == last_in_cluster) {
				spin_lock(&swap_lock);
				offset -= SWAPFILE_CLUSTER - 1;
				si->cluster_next = offset;
				si->cluster_nr = SWAPFILE_CLUSTER - 1;
				found_free_cluster = 1;
				goto checks;
			}
			if (unlikely(--latency_ration < 0)) {
				cond_resched();
				latency_ration = LATENCY_LIMIT;
			}
		}

		offset = scan_base;
320
		spin_lock(&swap_lock);
321
		si->cluster_nr = SWAPFILE_CLUSTER - 1;
322
		si->lowest_alloc = 0;
L
Linus Torvalds 已提交
323
	}
324

325 326
checks:
	if (!(si->flags & SWP_WRITEOK))
327
		goto no_page;
328 329
	if (!si->highest_bit)
		goto no_page;
330
	if (offset > si->highest_bit)
331
		scan_base = offset = si->lowest_bit;
332 333 334 335 336 337 338 339 340 341 342 343 344

	/* reuse swap entry of cache-only swap if not busy. */
	if (vm_swap_full() && si->swap_map[offset] == SWAP_HAS_CACHE) {
		int swap_was_freed;
		spin_unlock(&swap_lock);
		swap_was_freed = __try_to_reclaim_swap(si, offset);
		spin_lock(&swap_lock);
		/* entry was freed successfully, try to use this again */
		if (swap_was_freed)
			goto checks;
		goto scan; /* check next one */
	}

345 346 347 348 349 350 351 352 353 354 355
	if (si->swap_map[offset])
		goto scan;

	if (offset == si->lowest_bit)
		si->lowest_bit++;
	if (offset == si->highest_bit)
		si->highest_bit--;
	si->inuse_pages++;
	if (si->inuse_pages == si->pages) {
		si->lowest_bit = si->max;
		si->highest_bit = 0;
L
Linus Torvalds 已提交
356
	}
357 358 359 360
	if (cache == SWAP_CACHE) /* at usual swap-out via vmscan.c */
		si->swap_map[offset] = encode_swapmap(0, true);
	else /* at suspend */
		si->swap_map[offset] = encode_swapmap(1, false);
361 362
	si->cluster_next = offset + 1;
	si->flags -= SWP_SCANNING;
363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416

	if (si->lowest_alloc) {
		/*
		 * Only set when SWP_DISCARDABLE, and there's a scan
		 * for a free cluster in progress or just completed.
		 */
		if (found_free_cluster) {
			/*
			 * To optimize wear-levelling, discard the
			 * old data of the cluster, taking care not to
			 * discard any of its pages that have already
			 * been allocated by racing tasks (offset has
			 * already stepped over any at the beginning).
			 */
			if (offset < si->highest_alloc &&
			    si->lowest_alloc <= last_in_cluster)
				last_in_cluster = si->lowest_alloc - 1;
			si->flags |= SWP_DISCARDING;
			spin_unlock(&swap_lock);

			if (offset < last_in_cluster)
				discard_swap_cluster(si, offset,
					last_in_cluster - offset + 1);

			spin_lock(&swap_lock);
			si->lowest_alloc = 0;
			si->flags &= ~SWP_DISCARDING;

			smp_mb();	/* wake_up_bit advises this */
			wake_up_bit(&si->flags, ilog2(SWP_DISCARDING));

		} else if (si->flags & SWP_DISCARDING) {
			/*
			 * Delay using pages allocated by racing tasks
			 * until the whole discard has been issued. We
			 * could defer that delay until swap_writepage,
			 * but it's easier to keep this self-contained.
			 */
			spin_unlock(&swap_lock);
			wait_on_bit(&si->flags, ilog2(SWP_DISCARDING),
				wait_for_discard, TASK_UNINTERRUPTIBLE);
			spin_lock(&swap_lock);
		} else {
			/*
			 * Note pages allocated by racing tasks while
			 * scan for a free cluster is in progress, so
			 * that its final discard can exclude them.
			 */
			if (offset < si->lowest_alloc)
				si->lowest_alloc = offset;
			if (offset > si->highest_alloc)
				si->highest_alloc = offset;
		}
	}
417
	return offset;
418

419
scan:
420
	spin_unlock(&swap_lock);
421
	while (++offset <= si->highest_bit) {
422
		if (!si->swap_map[offset]) {
423
			spin_lock(&swap_lock);
424 425
			goto checks;
		}
426 427 428 429
		if (vm_swap_full() && si->swap_map[offset] == SWAP_HAS_CACHE) {
			spin_lock(&swap_lock);
			goto checks;
		}
430 431 432 433
		if (unlikely(--latency_ration < 0)) {
			cond_resched();
			latency_ration = LATENCY_LIMIT;
		}
434
	}
435 436 437 438 439 440
	offset = si->lowest_bit;
	while (++offset < scan_base) {
		if (!si->swap_map[offset]) {
			spin_lock(&swap_lock);
			goto checks;
		}
441 442 443 444
		if (vm_swap_full() && si->swap_map[offset] == SWAP_HAS_CACHE) {
			spin_lock(&swap_lock);
			goto checks;
		}
445 446 447 448 449
		if (unlikely(--latency_ration < 0)) {
			cond_resched();
			latency_ration = LATENCY_LIMIT;
		}
	}
450
	spin_lock(&swap_lock);
451 452

no_page:
453
	si->flags -= SWP_SCANNING;
L
Linus Torvalds 已提交
454 455 456 457 458
	return 0;
}

swp_entry_t get_swap_page(void)
{
459 460 461 462
	struct swap_info_struct *si;
	pgoff_t offset;
	int type, next;
	int wrapped = 0;
L
Linus Torvalds 已提交
463

464
	spin_lock(&swap_lock);
L
Linus Torvalds 已提交
465
	if (nr_swap_pages <= 0)
466 467 468 469 470 471 472 473 474 475
		goto noswap;
	nr_swap_pages--;

	for (type = swap_list.next; type >= 0 && wrapped < 2; type = next) {
		si = swap_info + type;
		next = si->next;
		if (next < 0 ||
		    (!wrapped && si->prio != swap_info[next].prio)) {
			next = swap_list.head;
			wrapped++;
L
Linus Torvalds 已提交
476
		}
477 478 479 480 481 482 483

		if (!si->highest_bit)
			continue;
		if (!(si->flags & SWP_WRITEOK))
			continue;

		swap_list.next = next;
484 485
		/* This is called for allocating swap entry for cache */
		offset = scan_swap_map(si, SWAP_CACHE);
486 487
		if (offset) {
			spin_unlock(&swap_lock);
488
			return swp_entry(type, offset);
489
		}
490
		next = swap_list.next;
L
Linus Torvalds 已提交
491
	}
492 493 494

	nr_swap_pages++;
noswap:
495
	spin_unlock(&swap_lock);
496
	return (swp_entry_t) {0};
L
Linus Torvalds 已提交
497 498
}

499
/* The only caller of this function is now susupend routine */
500 501 502 503 504 505 506 507 508
swp_entry_t get_swap_page_of_type(int type)
{
	struct swap_info_struct *si;
	pgoff_t offset;

	spin_lock(&swap_lock);
	si = swap_info + type;
	if (si->flags & SWP_WRITEOK) {
		nr_swap_pages--;
509 510
		/* This is called for allocating swap entry, not cache */
		offset = scan_swap_map(si, SWAP_MAP);
511 512 513 514 515 516 517 518 519 520
		if (offset) {
			spin_unlock(&swap_lock);
			return swp_entry(type, offset);
		}
		nr_swap_pages++;
	}
	spin_unlock(&swap_lock);
	return (swp_entry_t) {0};
}

L
Linus Torvalds 已提交
521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538
static struct swap_info_struct * swap_info_get(swp_entry_t entry)
{
	struct swap_info_struct * p;
	unsigned long offset, type;

	if (!entry.val)
		goto out;
	type = swp_type(entry);
	if (type >= nr_swapfiles)
		goto bad_nofile;
	p = & swap_info[type];
	if (!(p->flags & SWP_USED))
		goto bad_device;
	offset = swp_offset(entry);
	if (offset >= p->max)
		goto bad_offset;
	if (!p->swap_map[offset])
		goto bad_free;
539
	spin_lock(&swap_lock);
L
Linus Torvalds 已提交
540 541 542 543 544 545 546 547 548 549 550 551 552 553 554
	return p;

bad_free:
	printk(KERN_ERR "swap_free: %s%08lx\n", Unused_offset, entry.val);
	goto out;
bad_offset:
	printk(KERN_ERR "swap_free: %s%08lx\n", Bad_offset, entry.val);
	goto out;
bad_device:
	printk(KERN_ERR "swap_free: %s%08lx\n", Unused_file, entry.val);
	goto out;
bad_nofile:
	printk(KERN_ERR "swap_free: %s%08lx\n", Bad_file, entry.val);
out:
	return NULL;
555
}
L
Linus Torvalds 已提交
556

557 558
static int swap_entry_free(struct swap_info_struct *p,
			   swp_entry_t ent, int cache)
L
Linus Torvalds 已提交
559
{
560
	unsigned long offset = swp_offset(ent);
561 562 563 564 565 566 567 568 569
	int count = swap_count(p->swap_map[offset]);
	bool has_cache;

	has_cache = swap_has_cache(p->swap_map[offset]);

	if (cache == SWAP_MAP) { /* dropping usage count of swap */
		if (count < SWAP_MAP_MAX) {
			count--;
			p->swap_map[offset] = encode_swapmap(count, has_cache);
L
Linus Torvalds 已提交
570
		}
571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587
	} else { /* dropping swap cache flag */
		VM_BUG_ON(!has_cache);
		p->swap_map[offset] = encode_swapmap(count, false);

	}
	/* return code. */
	count = p->swap_map[offset];
	/* free if no reference */
	if (!count) {
		if (offset < p->lowest_bit)
			p->lowest_bit = offset;
		if (offset > p->highest_bit)
			p->highest_bit = offset;
		if (p->prio > swap_info[swap_list.next].prio)
			swap_list.next = p - swap_info;
		nr_swap_pages++;
		p->inuse_pages--;
L
Linus Torvalds 已提交
588
	}
K
KAMEZAWA Hiroyuki 已提交
589 590
	if (!swap_count(count))
		mem_cgroup_uncharge_swap(ent);
L
Linus Torvalds 已提交
591 592 593 594 595 596 597 598 599 600 601 602 603
	return count;
}

/*
 * Caller has made sure that the swapdevice corresponding to entry
 * is still around or has not been recycled.
 */
void swap_free(swp_entry_t entry)
{
	struct swap_info_struct * p;

	p = swap_info_get(entry);
	if (p) {
604
		swap_entry_free(p, entry, SWAP_MAP);
605
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
606 607 608
	}
}

609 610 611 612 613
/*
 * Called after dropping swapcache to decrease refcnt to swap entries.
 */
void swapcache_free(swp_entry_t entry, struct page *page)
{
614
	struct swap_info_struct *p;
K
KAMEZAWA Hiroyuki 已提交
615
	int ret;
616 617 618

	p = swap_info_get(entry);
	if (p) {
K
KAMEZAWA Hiroyuki 已提交
619 620 621 622 623 624 625 626 627
		ret = swap_entry_free(p, entry, SWAP_CACHE);
		if (page) {
			bool swapout;
			if (ret)
				swapout = true; /* the end of swap out */
			else
				swapout = false; /* no more swap users! */
			mem_cgroup_uncharge_swapcache(page, entry, swapout);
		}
628 629 630
		spin_unlock(&swap_lock);
	}
	return;
631 632
}

L
Linus Torvalds 已提交
633
/*
634
 * How many references to page are currently swapped out?
L
Linus Torvalds 已提交
635
 */
636
static inline int page_swapcount(struct page *page)
L
Linus Torvalds 已提交
637
{
638 639
	int count = 0;
	struct swap_info_struct *p;
L
Linus Torvalds 已提交
640 641
	swp_entry_t entry;

H
Hugh Dickins 已提交
642
	entry.val = page_private(page);
L
Linus Torvalds 已提交
643 644
	p = swap_info_get(entry);
	if (p) {
645
		count = swap_count(p->swap_map[swp_offset(entry)]);
646
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
647
	}
648
	return count;
L
Linus Torvalds 已提交
649 650 651
}

/*
652 653 654 655
 * We can write to an anon page without COW if there are no other references
 * to it.  And as a side-effect, free up its swap: because the old content
 * on disk will never be read, and seeking back there to write new content
 * later would only waste time away from clustering.
L
Linus Torvalds 已提交
656
 */
657
int reuse_swap_page(struct page *page)
L
Linus Torvalds 已提交
658
{
659 660
	int count;

661
	VM_BUG_ON(!PageLocked(page));
662
	count = page_mapcount(page);
663
	if (count <= 1 && PageSwapCache(page)) {
664
		count += page_swapcount(page);
665 666 667 668 669
		if (count == 1 && !PageWriteback(page)) {
			delete_from_swap_cache(page);
			SetPageDirty(page);
		}
	}
670
	return count == 1;
L
Linus Torvalds 已提交
671 672 673
}

/*
674 675
 * If swap is getting full, or if there are no more mappings of this page,
 * then try_to_free_swap is called to free its swap space.
L
Linus Torvalds 已提交
676
 */
677
int try_to_free_swap(struct page *page)
L
Linus Torvalds 已提交
678
{
679
	VM_BUG_ON(!PageLocked(page));
L
Linus Torvalds 已提交
680 681 682 683 684

	if (!PageSwapCache(page))
		return 0;
	if (PageWriteback(page))
		return 0;
685
	if (page_swapcount(page))
L
Linus Torvalds 已提交
686 687
		return 0;

688 689 690
	delete_from_swap_cache(page);
	SetPageDirty(page);
	return 1;
691 692
}

L
Linus Torvalds 已提交
693 694 695 696
/*
 * Free the swap entry like above, but also try to
 * free the page cache entry if it is the last user.
 */
697
int free_swap_and_cache(swp_entry_t entry)
L
Linus Torvalds 已提交
698
{
699
	struct swap_info_struct *p;
L
Linus Torvalds 已提交
700 701
	struct page *page = NULL;

702
	if (is_migration_entry(entry))
703
		return 1;
704

L
Linus Torvalds 已提交
705 706
	p = swap_info_get(entry);
	if (p) {
707
		if (swap_entry_free(p, entry, SWAP_MAP) == SWAP_HAS_CACHE) {
708
			page = find_get_page(&swapper_space, entry.val);
N
Nick Piggin 已提交
709
			if (page && !trylock_page(page)) {
710 711 712 713
				page_cache_release(page);
				page = NULL;
			}
		}
714
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
715 716
	}
	if (page) {
717 718 719 720
		/*
		 * Not mapped elsewhere, or swap space full? Free it!
		 * Also recheck PageSwapCache now page is locked (above).
		 */
721
		if (PageSwapCache(page) && !PageWriteback(page) &&
722
				(!page_mapped(page) || vm_swap_full())) {
L
Linus Torvalds 已提交
723 724 725 726 727 728
			delete_from_swap_cache(page);
			SetPageDirty(page);
		}
		unlock_page(page);
		page_cache_release(page);
	}
729
	return p != NULL;
L
Linus Torvalds 已提交
730 731
}

732
#ifdef CONFIG_HIBERNATION
733
/*
734
 * Find the swap type that corresponds to given device (if any).
735
 *
736 737 738 739
 * @offset - number of the PAGE_SIZE-sized block of the device, starting
 * from 0, in which the swap header is expected to be located.
 *
 * This is needed for the suspend to disk (aka swsusp).
740
 */
741
int swap_type_of(dev_t device, sector_t offset, struct block_device **bdev_p)
742
{
743
	struct block_device *bdev = NULL;
744 745
	int i;

746 747 748
	if (device)
		bdev = bdget(device);

749 750
	spin_lock(&swap_lock);
	for (i = 0; i < nr_swapfiles; i++) {
751
		struct swap_info_struct *sis = swap_info + i;
752

753
		if (!(sis->flags & SWP_WRITEOK))
754
			continue;
755

756
		if (!bdev) {
757
			if (bdev_p)
758
				*bdev_p = bdgrab(sis->bdev);
759

760 761 762
			spin_unlock(&swap_lock);
			return i;
		}
763 764 765 766 767 768
		if (bdev == sis->bdev) {
			struct swap_extent *se;

			se = list_entry(sis->extent_list.next,
					struct swap_extent, list);
			if (se->start_block == offset) {
769
				if (bdev_p)
770
					*bdev_p = bdgrab(sis->bdev);
771

772 773 774 775
				spin_unlock(&swap_lock);
				bdput(bdev);
				return i;
			}
776 777 778
		}
	}
	spin_unlock(&swap_lock);
779 780 781
	if (bdev)
		bdput(bdev);

782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807
	return -ENODEV;
}

/*
 * Return either the total number of swap pages of given type, or the number
 * of free pages of that type (depending on @free)
 *
 * This is needed for software suspend
 */
unsigned int count_swap_pages(int type, int free)
{
	unsigned int n = 0;

	if (type < nr_swapfiles) {
		spin_lock(&swap_lock);
		if (swap_info[type].flags & SWP_WRITEOK) {
			n = swap_info[type].pages;
			if (free)
				n -= swap_info[type].inuse_pages;
		}
		spin_unlock(&swap_lock);
	}
	return n;
}
#endif

L
Linus Torvalds 已提交
808
/*
809 810 811
 * No need to decide whether this PTE shares the swap entry with others,
 * just let do_wp_page work it out if a write is requested later - to
 * force COW, vm_page_prot omits write permission from any private vma.
L
Linus Torvalds 已提交
812
 */
H
Hugh Dickins 已提交
813
static int unuse_pte(struct vm_area_struct *vma, pmd_t *pmd,
L
Linus Torvalds 已提交
814 815
		unsigned long addr, swp_entry_t entry, struct page *page)
{
816
	struct mem_cgroup *ptr = NULL;
H
Hugh Dickins 已提交
817 818 819 820
	spinlock_t *ptl;
	pte_t *pte;
	int ret = 1;

821
	if (mem_cgroup_try_charge_swapin(vma->vm_mm, page, GFP_KERNEL, &ptr)) {
H
Hugh Dickins 已提交
822
		ret = -ENOMEM;
823 824
		goto out_nolock;
	}
H
Hugh Dickins 已提交
825 826 827 828

	pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
	if (unlikely(!pte_same(*pte, swp_entry_to_pte(entry)))) {
		if (ret > 0)
829
			mem_cgroup_cancel_charge_swapin(ptr);
H
Hugh Dickins 已提交
830 831 832
		ret = 0;
		goto out;
	}
833

834
	inc_mm_counter(vma->vm_mm, anon_rss);
L
Linus Torvalds 已提交
835 836 837 838
	get_page(page);
	set_pte_at(vma->vm_mm, addr, pte,
		   pte_mkold(mk_pte(page, vma->vm_page_prot)));
	page_add_anon_rmap(page, vma, addr);
839
	mem_cgroup_commit_charge_swapin(page, ptr);
L
Linus Torvalds 已提交
840 841 842 843 844 845
	swap_free(entry);
	/*
	 * Move the page to the active list so it is not
	 * immediately swapped out again after swapon.
	 */
	activate_page(page);
H
Hugh Dickins 已提交
846 847
out:
	pte_unmap_unlock(pte, ptl);
848
out_nolock:
H
Hugh Dickins 已提交
849
	return ret;
L
Linus Torvalds 已提交
850 851 852 853 854 855 856
}

static int unuse_pte_range(struct vm_area_struct *vma, pmd_t *pmd,
				unsigned long addr, unsigned long end,
				swp_entry_t entry, struct page *page)
{
	pte_t swp_pte = swp_entry_to_pte(entry);
857
	pte_t *pte;
858
	int ret = 0;
L
Linus Torvalds 已提交
859

H
Hugh Dickins 已提交
860 861 862 863 864 865 866 867 868 869
	/*
	 * We don't actually need pte lock while scanning for swp_pte: since
	 * we hold page lock and mmap_sem, swp_pte cannot be inserted into the
	 * page table while we're scanning; though it could get zapped, and on
	 * some architectures (e.g. x86_32 with PAE) we might catch a glimpse
	 * of unmatched parts which look like swp_pte, so unuse_pte must
	 * recheck under pte lock.  Scanning without pte lock lets it be
	 * preemptible whenever CONFIG_PREEMPT but not CONFIG_HIGHPTE.
	 */
	pte = pte_offset_map(pmd, addr);
L
Linus Torvalds 已提交
870 871 872 873 874 875
	do {
		/*
		 * swapoff spends a _lot_ of time in this loop!
		 * Test inline before going to call unuse_pte.
		 */
		if (unlikely(pte_same(*pte, swp_pte))) {
H
Hugh Dickins 已提交
876 877 878 879 880
			pte_unmap(pte);
			ret = unuse_pte(vma, pmd, addr, entry, page);
			if (ret)
				goto out;
			pte = pte_offset_map(pmd, addr);
L
Linus Torvalds 已提交
881 882
		}
	} while (pte++, addr += PAGE_SIZE, addr != end);
H
Hugh Dickins 已提交
883 884
	pte_unmap(pte - 1);
out:
885
	return ret;
L
Linus Torvalds 已提交
886 887 888 889 890 891 892 893
}

static inline int unuse_pmd_range(struct vm_area_struct *vma, pud_t *pud,
				unsigned long addr, unsigned long end,
				swp_entry_t entry, struct page *page)
{
	pmd_t *pmd;
	unsigned long next;
894
	int ret;
L
Linus Torvalds 已提交
895 896 897 898 899 900

	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
		if (pmd_none_or_clear_bad(pmd))
			continue;
901 902 903
		ret = unuse_pte_range(vma, pmd, addr, next, entry, page);
		if (ret)
			return ret;
L
Linus Torvalds 已提交
904 905 906 907 908 909 910 911 912 913
	} while (pmd++, addr = next, addr != end);
	return 0;
}

static inline int unuse_pud_range(struct vm_area_struct *vma, pgd_t *pgd,
				unsigned long addr, unsigned long end,
				swp_entry_t entry, struct page *page)
{
	pud_t *pud;
	unsigned long next;
914
	int ret;
L
Linus Torvalds 已提交
915 916 917 918 919 920

	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
		if (pud_none_or_clear_bad(pud))
			continue;
921 922 923
		ret = unuse_pmd_range(vma, pud, addr, next, entry, page);
		if (ret)
			return ret;
L
Linus Torvalds 已提交
924 925 926 927 928 929 930 931 932
	} while (pud++, addr = next, addr != end);
	return 0;
}

static int unuse_vma(struct vm_area_struct *vma,
				swp_entry_t entry, struct page *page)
{
	pgd_t *pgd;
	unsigned long addr, end, next;
933
	int ret;
L
Linus Torvalds 已提交
934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950

	if (page->mapping) {
		addr = page_address_in_vma(page, vma);
		if (addr == -EFAULT)
			return 0;
		else
			end = addr + PAGE_SIZE;
	} else {
		addr = vma->vm_start;
		end = vma->vm_end;
	}

	pgd = pgd_offset(vma->vm_mm, addr);
	do {
		next = pgd_addr_end(addr, end);
		if (pgd_none_or_clear_bad(pgd))
			continue;
951 952 953
		ret = unuse_pud_range(vma, pgd, addr, next, entry, page);
		if (ret)
			return ret;
L
Linus Torvalds 已提交
954 955 956 957 958 959 960 961
	} while (pgd++, addr = next, addr != end);
	return 0;
}

static int unuse_mm(struct mm_struct *mm,
				swp_entry_t entry, struct page *page)
{
	struct vm_area_struct *vma;
962
	int ret = 0;
L
Linus Torvalds 已提交
963 964 965

	if (!down_read_trylock(&mm->mmap_sem)) {
		/*
966 967
		 * Activate page so shrink_inactive_list is unlikely to unmap
		 * its ptes while lock is dropped, so swapoff can make progress.
L
Linus Torvalds 已提交
968
		 */
969
		activate_page(page);
L
Linus Torvalds 已提交
970 971 972 973 974
		unlock_page(page);
		down_read(&mm->mmap_sem);
		lock_page(page);
	}
	for (vma = mm->mmap; vma; vma = vma->vm_next) {
975
		if (vma->anon_vma && (ret = unuse_vma(vma, entry, page)))
L
Linus Torvalds 已提交
976 977 978
			break;
	}
	up_read(&mm->mmap_sem);
979
	return (ret < 0)? ret: 0;
L
Linus Torvalds 已提交
980 981 982 983 984 985
}

/*
 * Scan swap_map from current position to next entry still in use.
 * Recycle to start on reaching the end, returning 0 when empty.
 */
986 987
static unsigned int find_next_to_unuse(struct swap_info_struct *si,
					unsigned int prev)
L
Linus Torvalds 已提交
988
{
989 990
	unsigned int max = si->max;
	unsigned int i = prev;
L
Linus Torvalds 已提交
991 992 993
	int count;

	/*
994
	 * No need for swap_lock here: we're just looking
L
Linus Torvalds 已提交
995 996
	 * for whether an entry is in use, not modifying it; false
	 * hits are okay, and sys_swapoff() has already prevented new
997
	 * allocations from this area (while holding swap_lock).
L
Linus Torvalds 已提交
998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013
	 */
	for (;;) {
		if (++i >= max) {
			if (!prev) {
				i = 0;
				break;
			}
			/*
			 * No entries in use at top of swap_map,
			 * loop back to start and recheck there.
			 */
			max = prev + 1;
			prev = 0;
			i = 1;
		}
		count = si->swap_map[i];
1014
		if (count && swap_count(count) != SWAP_MAP_BAD)
L
Linus Torvalds 已提交
1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032
			break;
	}
	return i;
}

/*
 * We completely avoid races by reading each swap page in advance,
 * and then search for the process using it.  All the necessary
 * page table adjustments can then be made atomically.
 */
static int try_to_unuse(unsigned int type)
{
	struct swap_info_struct * si = &swap_info[type];
	struct mm_struct *start_mm;
	unsigned short *swap_map;
	unsigned short swcount;
	struct page *page;
	swp_entry_t entry;
1033
	unsigned int i = 0;
L
Linus Torvalds 已提交
1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066
	int retval = 0;
	int reset_overflow = 0;
	int shmem;

	/*
	 * When searching mms for an entry, a good strategy is to
	 * start at the first mm we freed the previous entry from
	 * (though actually we don't notice whether we or coincidence
	 * freed the entry).  Initialize this start_mm with a hold.
	 *
	 * A simpler strategy would be to start at the last mm we
	 * freed the previous entry from; but that would take less
	 * advantage of mmlist ordering, which clusters forked mms
	 * together, child after parent.  If we race with dup_mmap(), we
	 * prefer to resolve parent before child, lest we miss entries
	 * duplicated after we scanned child: using last mm would invert
	 * that.  Though it's only a serious concern when an overflowed
	 * swap count is reset from SWAP_MAP_MAX, preventing a rescan.
	 */
	start_mm = &init_mm;
	atomic_inc(&init_mm.mm_users);

	/*
	 * Keep on scanning until all entries have gone.  Usually,
	 * one pass through swap_map is enough, but not necessarily:
	 * there are races when an instance of an entry might be missed.
	 */
	while ((i = find_next_to_unuse(si, i)) != 0) {
		if (signal_pending(current)) {
			retval = -EINTR;
			break;
		}

1067
		/*
L
Linus Torvalds 已提交
1068 1069
		 * Get a page for the entry, using the existing swap
		 * cache page if there is one.  Otherwise, get a clean
1070
		 * page and read the swap into it.
L
Linus Torvalds 已提交
1071 1072 1073
		 */
		swap_map = &si->swap_map[i];
		entry = swp_entry(type, i);
1074 1075
		page = read_swap_cache_async(entry,
					GFP_HIGHUSER_MOVABLE, NULL, 0);
L
Linus Torvalds 已提交
1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117
		if (!page) {
			/*
			 * Either swap_duplicate() failed because entry
			 * has been freed independently, and will not be
			 * reused since sys_swapoff() already disabled
			 * allocation from here, or alloc_page() failed.
			 */
			if (!*swap_map)
				continue;
			retval = -ENOMEM;
			break;
		}

		/*
		 * Don't hold on to start_mm if it looks like exiting.
		 */
		if (atomic_read(&start_mm->mm_users) == 1) {
			mmput(start_mm);
			start_mm = &init_mm;
			atomic_inc(&init_mm.mm_users);
		}

		/*
		 * Wait for and lock page.  When do_swap_page races with
		 * try_to_unuse, do_swap_page can handle the fault much
		 * faster than try_to_unuse can locate the entry.  This
		 * apparently redundant "wait_on_page_locked" lets try_to_unuse
		 * defer to do_swap_page in such a case - in some tests,
		 * do_swap_page and try_to_unuse repeatedly compete.
		 */
		wait_on_page_locked(page);
		wait_on_page_writeback(page);
		lock_page(page);
		wait_on_page_writeback(page);

		/*
		 * Remove all references to entry.
		 * Whenever we reach init_mm, there's no address space
		 * to search, but use it as a reminder to search shmem.
		 */
		shmem = 0;
		swcount = *swap_map;
1118
		if (swap_count(swcount)) {
L
Linus Torvalds 已提交
1119 1120 1121 1122 1123
			if (start_mm == &init_mm)
				shmem = shmem_unuse(entry, page);
			else
				retval = unuse_mm(start_mm, entry, page);
		}
1124
		if (swap_count(*swap_map)) {
L
Linus Torvalds 已提交
1125 1126 1127 1128 1129 1130 1131 1132 1133
			int set_start_mm = (*swap_map >= swcount);
			struct list_head *p = &start_mm->mmlist;
			struct mm_struct *new_start_mm = start_mm;
			struct mm_struct *prev_mm = start_mm;
			struct mm_struct *mm;

			atomic_inc(&new_start_mm->mm_users);
			atomic_inc(&prev_mm->mm_users);
			spin_lock(&mmlist_lock);
1134
			while (swap_count(*swap_map) && !retval && !shmem &&
L
Linus Torvalds 已提交
1135 1136
					(p = p->next) != &start_mm->mmlist) {
				mm = list_entry(p, struct mm_struct, mmlist);
1137
				if (!atomic_inc_not_zero(&mm->mm_users))
L
Linus Torvalds 已提交
1138 1139 1140 1141 1142 1143 1144 1145
					continue;
				spin_unlock(&mmlist_lock);
				mmput(prev_mm);
				prev_mm = mm;

				cond_resched();

				swcount = *swap_map;
1146
				if (!swap_count(swcount)) /* any usage ? */
L
Linus Torvalds 已提交
1147 1148 1149 1150 1151 1152
					;
				else if (mm == &init_mm) {
					set_start_mm = 1;
					shmem = shmem_unuse(entry, page);
				} else
					retval = unuse_mm(mm, entry, page);
1153 1154 1155

				if (set_start_mm &&
				    swap_count(*swap_map) < swcount) {
L
Linus Torvalds 已提交
1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167
					mmput(new_start_mm);
					atomic_inc(&mm->mm_users);
					new_start_mm = mm;
					set_start_mm = 0;
				}
				spin_lock(&mmlist_lock);
			}
			spin_unlock(&mmlist_lock);
			mmput(prev_mm);
			mmput(start_mm);
			start_mm = new_start_mm;
		}
1168 1169 1170 1171 1172 1173 1174
		if (shmem) {
			/* page has already been unlocked and released */
			if (shmem > 0)
				continue;
			retval = shmem;
			break;
		}
L
Linus Torvalds 已提交
1175 1176 1177 1178 1179 1180 1181
		if (retval) {
			unlock_page(page);
			page_cache_release(page);
			break;
		}

		/*
1182 1183 1184 1185 1186 1187
		 * How could swap count reach 0x7ffe ?
		 * There's no way to repeat a swap page within an mm
		 * (except in shmem, where it's the shared object which takes
		 * the reference count)?
		 * We believe SWAP_MAP_MAX cannot occur.(if occur, unsigned
		 * short is too small....)
L
Linus Torvalds 已提交
1188 1189 1190 1191 1192 1193
		 * If that's wrong, then we should worry more about
		 * exit_mmap() and do_munmap() cases described above:
		 * we might be resetting SWAP_MAP_MAX too early here.
		 * We know "Undead"s can happen, they're okay, so don't
		 * report them; but do report if we reset SWAP_MAP_MAX.
		 */
1194 1195 1196 1197 1198
		/* We might release the lock_page() in unuse_mm(). */
		if (!PageSwapCache(page) || page_private(page) != entry.val)
			goto retry;

		if (swap_count(*swap_map) == SWAP_MAP_MAX) {
1199
			spin_lock(&swap_lock);
1200
			*swap_map = encode_swapmap(0, true);
1201
			spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217
			reset_overflow = 1;
		}

		/*
		 * If a reference remains (rare), we would like to leave
		 * the page in the swap cache; but try_to_unmap could
		 * then re-duplicate the entry once we drop page lock,
		 * so we might loop indefinitely; also, that page could
		 * not be swapped out to other storage meanwhile.  So:
		 * delete from cache even if there's another reference,
		 * after ensuring that the data has been saved to disk -
		 * since if the reference remains (rarer), it will be
		 * read from disk into another page.  Splitting into two
		 * pages would be incorrect if swap supported "shared
		 * private" pages, but they are handled by tmpfs files.
		 */
1218 1219
		if (swap_count(*swap_map) &&
		     PageDirty(page) && PageSwapCache(page)) {
L
Linus Torvalds 已提交
1220 1221 1222 1223 1224 1225 1226 1227
			struct writeback_control wbc = {
				.sync_mode = WB_SYNC_NONE,
			};

			swap_writepage(page, &wbc);
			lock_page(page);
			wait_on_page_writeback(page);
		}
1228 1229 1230 1231 1232 1233 1234 1235 1236 1237

		/*
		 * It is conceivable that a racing task removed this page from
		 * swap cache just before we acquired the page lock at the top,
		 * or while we dropped it in unuse_mm().  The page might even
		 * be back in swap cache on another swap area: that we must not
		 * delete, since it may not have been written out to swap yet.
		 */
		if (PageSwapCache(page) &&
		    likely(page_private(page) == entry.val))
1238
			delete_from_swap_cache(page);
L
Linus Torvalds 已提交
1239 1240 1241 1242

		/*
		 * So we could skip searching mms once swap count went
		 * to 1, we did not mark any present ptes as dirty: must
1243
		 * mark page dirty so shrink_page_list will preserve it.
L
Linus Torvalds 已提交
1244 1245
		 */
		SetPageDirty(page);
1246
retry:
L
Linus Torvalds 已提交
1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265
		unlock_page(page);
		page_cache_release(page);

		/*
		 * Make sure that we aren't completely killing
		 * interactive performance.
		 */
		cond_resched();
	}

	mmput(start_mm);
	if (reset_overflow) {
		printk(KERN_WARNING "swapoff: cleared swap entry overflow\n");
		swap_overflow = 0;
	}
	return retval;
}

/*
1266 1267 1268
 * After a successful try_to_unuse, if no swap is now in use, we know
 * we can empty the mmlist.  swap_lock must be held on entry and exit.
 * Note that mmlist_lock nests inside swap_lock, and an mm must be
L
Linus Torvalds 已提交
1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300
 * added to the mmlist just after page_duplicate - before would be racy.
 */
static void drain_mmlist(void)
{
	struct list_head *p, *next;
	unsigned int i;

	for (i = 0; i < nr_swapfiles; i++)
		if (swap_info[i].inuse_pages)
			return;
	spin_lock(&mmlist_lock);
	list_for_each_safe(p, next, &init_mm.mmlist)
		list_del_init(p);
	spin_unlock(&mmlist_lock);
}

/*
 * Use this swapdev's extent info to locate the (PAGE_SIZE) block which
 * corresponds to page offset `offset'.
 */
sector_t map_swap_page(struct swap_info_struct *sis, pgoff_t offset)
{
	struct swap_extent *se = sis->curr_swap_extent;
	struct swap_extent *start_se = se;

	for ( ; ; ) {
		struct list_head *lh;

		if (se->start_page <= offset &&
				offset < (se->start_page + se->nr_pages)) {
			return se->start_block + (offset - se->start_page);
		}
1301
		lh = se->list.next;
L
Linus Torvalds 已提交
1302
		if (lh == &sis->extent_list)
1303
			lh = lh->next;
L
Linus Torvalds 已提交
1304 1305 1306 1307 1308 1309
		se = list_entry(lh, struct swap_extent, list);
		sis->curr_swap_extent = se;
		BUG_ON(se == start_se);		/* It *must* be present */
	}
}

1310
#ifdef CONFIG_HIBERNATION
1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324
/*
 * Get the (PAGE_SIZE) block corresponding to given offset on the swapdev
 * corresponding to given index in swap_info (swap type).
 */
sector_t swapdev_block(int swap_type, pgoff_t offset)
{
	struct swap_info_struct *sis;

	if (swap_type >= nr_swapfiles)
		return 0;

	sis = swap_info + swap_type;
	return (sis->flags & SWP_WRITEOK) ? map_swap_page(sis, offset) : 0;
}
1325
#endif /* CONFIG_HIBERNATION */
1326

L
Linus Torvalds 已提交
1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343
/*
 * Free all of a swapdev's extent information
 */
static void destroy_swap_extents(struct swap_info_struct *sis)
{
	while (!list_empty(&sis->extent_list)) {
		struct swap_extent *se;

		se = list_entry(sis->extent_list.next,
				struct swap_extent, list);
		list_del(&se->list);
		kfree(se);
	}
}

/*
 * Add a block range (and the corresponding page range) into this swapdev's
1344
 * extent list.  The extent list is kept sorted in page order.
L
Linus Torvalds 已提交
1345
 *
1346
 * This function rather assumes that it is called in ascending page order.
L
Linus Torvalds 已提交
1347 1348 1349 1350 1351 1352 1353 1354 1355
 */
static int
add_swap_extent(struct swap_info_struct *sis, unsigned long start_page,
		unsigned long nr_pages, sector_t start_block)
{
	struct swap_extent *se;
	struct swap_extent *new_se;
	struct list_head *lh;

1356 1357
	lh = sis->extent_list.prev;	/* The highest page extent */
	if (lh != &sis->extent_list) {
L
Linus Torvalds 已提交
1358
		se = list_entry(lh, struct swap_extent, list);
1359 1360
		BUG_ON(se->start_page + se->nr_pages != start_page);
		if (se->start_block + se->nr_pages == start_block) {
L
Linus Torvalds 已提交
1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376
			/* Merge it */
			se->nr_pages += nr_pages;
			return 0;
		}
	}

	/*
	 * No merge.  Insert a new extent, preserving ordering.
	 */
	new_se = kmalloc(sizeof(*se), GFP_KERNEL);
	if (new_se == NULL)
		return -ENOMEM;
	new_se->start_page = start_page;
	new_se->nr_pages = nr_pages;
	new_se->start_block = start_block;

1377
	list_add_tail(&new_se->list, &sis->extent_list);
1378
	return 1;
L
Linus Torvalds 已提交
1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400
}

/*
 * A `swap extent' is a simple thing which maps a contiguous range of pages
 * onto a contiguous range of disk blocks.  An ordered list of swap extents
 * is built at swapon time and is then used at swap_writepage/swap_readpage
 * time for locating where on disk a page belongs.
 *
 * If the swapfile is an S_ISBLK block device, a single extent is installed.
 * This is done so that the main operating code can treat S_ISBLK and S_ISREG
 * swap files identically.
 *
 * Whether the swapdev is an S_ISREG file or an S_ISBLK blockdev, the swap
 * extent list operates in PAGE_SIZE disk blocks.  Both S_ISREG and S_ISBLK
 * swapfiles are handled *identically* after swapon time.
 *
 * For S_ISREG swapfiles, setup_swap_extents() will walk all the file's blocks
 * and will parse them into an ordered extent list, in PAGE_SIZE chunks.  If
 * some stray blocks are found which do not fall within the PAGE_SIZE alignment
 * requirements, they are simply tossed out - we will never use those blocks
 * for swapping.
 *
1401
 * For S_ISREG swapfiles we set S_SWAPFILE across the life of the swapon.  This
L
Linus Torvalds 已提交
1402 1403 1404 1405 1406 1407 1408 1409 1410 1411
 * prevents root from shooting her foot off by ftruncating an in-use swapfile,
 * which will scribble on the fs.
 *
 * The amount of disk space which a single swap extent represents varies.
 * Typically it is in the 1-4 megabyte range.  So we can have hundreds of
 * extents in the list.  To avoid much list walking, we cache the previous
 * search location in `curr_swap_extent', and start new searches from there.
 * This is extremely effective.  The average number of iterations in
 * map_swap_page() has been measured at about 0.3 per page.  - akpm.
 */
1412
static int setup_swap_extents(struct swap_info_struct *sis, sector_t *span)
L
Linus Torvalds 已提交
1413 1414 1415 1416 1417 1418 1419
{
	struct inode *inode;
	unsigned blocks_per_page;
	unsigned long page_no;
	unsigned blkbits;
	sector_t probe_block;
	sector_t last_block;
1420 1421 1422
	sector_t lowest_block = -1;
	sector_t highest_block = 0;
	int nr_extents = 0;
L
Linus Torvalds 已提交
1423 1424 1425 1426 1427
	int ret;

	inode = sis->swap_file->f_mapping->host;
	if (S_ISBLK(inode->i_mode)) {
		ret = add_swap_extent(sis, 0, sis->max, 0);
1428
		*span = sis->pages;
L
Linus Torvalds 已提交
1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472
		goto done;
	}

	blkbits = inode->i_blkbits;
	blocks_per_page = PAGE_SIZE >> blkbits;

	/*
	 * Map all the blocks into the extent list.  This code doesn't try
	 * to be very smart.
	 */
	probe_block = 0;
	page_no = 0;
	last_block = i_size_read(inode) >> blkbits;
	while ((probe_block + blocks_per_page) <= last_block &&
			page_no < sis->max) {
		unsigned block_in_page;
		sector_t first_block;

		first_block = bmap(inode, probe_block);
		if (first_block == 0)
			goto bad_bmap;

		/*
		 * It must be PAGE_SIZE aligned on-disk
		 */
		if (first_block & (blocks_per_page - 1)) {
			probe_block++;
			goto reprobe;
		}

		for (block_in_page = 1; block_in_page < blocks_per_page;
					block_in_page++) {
			sector_t block;

			block = bmap(inode, probe_block + block_in_page);
			if (block == 0)
				goto bad_bmap;
			if (block != first_block + block_in_page) {
				/* Discontiguity */
				probe_block++;
				goto reprobe;
			}
		}

1473 1474 1475 1476 1477 1478 1479 1480
		first_block >>= (PAGE_SHIFT - blkbits);
		if (page_no) {	/* exclude the header page */
			if (first_block < lowest_block)
				lowest_block = first_block;
			if (first_block > highest_block)
				highest_block = first_block;
		}

L
Linus Torvalds 已提交
1481 1482 1483
		/*
		 * We found a PAGE_SIZE-length, PAGE_SIZE-aligned run of blocks
		 */
1484 1485
		ret = add_swap_extent(sis, page_no, 1, first_block);
		if (ret < 0)
L
Linus Torvalds 已提交
1486
			goto out;
1487
		nr_extents += ret;
L
Linus Torvalds 已提交
1488 1489 1490 1491 1492
		page_no++;
		probe_block += blocks_per_page;
reprobe:
		continue;
	}
1493 1494
	ret = nr_extents;
	*span = 1 + highest_block - lowest_block;
L
Linus Torvalds 已提交
1495
	if (page_no == 0)
1496
		page_no = 1;	/* force Empty message */
L
Linus Torvalds 已提交
1497
	sis->max = page_no;
1498
	sis->pages = page_no - 1;
L
Linus Torvalds 已提交
1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510
	sis->highest_bit = page_no - 1;
done:
	sis->curr_swap_extent = list_entry(sis->extent_list.prev,
					struct swap_extent, list);
	goto out;
bad_bmap:
	printk(KERN_ERR "swapon: swapfile has holes\n");
	ret = -EINVAL;
out:
	return ret;
}

1511
SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
L
Linus Torvalds 已提交
1512 1513 1514 1515 1516 1517 1518 1519 1520
{
	struct swap_info_struct * p = NULL;
	unsigned short *swap_map;
	struct file *swap_file, *victim;
	struct address_space *mapping;
	struct inode *inode;
	char * pathname;
	int i, type, prev;
	int err;
1521

L
Linus Torvalds 已提交
1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	pathname = getname(specialfile);
	err = PTR_ERR(pathname);
	if (IS_ERR(pathname))
		goto out;

	victim = filp_open(pathname, O_RDWR|O_LARGEFILE, 0);
	putname(pathname);
	err = PTR_ERR(victim);
	if (IS_ERR(victim))
		goto out;

	mapping = victim->f_mapping;
	prev = -1;
1538
	spin_lock(&swap_lock);
L
Linus Torvalds 已提交
1539 1540
	for (type = swap_list.head; type >= 0; type = swap_info[type].next) {
		p = swap_info + type;
H
Hugh Dickins 已提交
1541
		if (p->flags & SWP_WRITEOK) {
L
Linus Torvalds 已提交
1542 1543 1544 1545 1546 1547 1548
			if (p->swap_file->f_mapping == mapping)
				break;
		}
		prev = type;
	}
	if (type < 0) {
		err = -EINVAL;
1549
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
1550 1551 1552 1553 1554 1555
		goto out_dput;
	}
	if (!security_vm_enough_memory(p->pages))
		vm_unacct_memory(p->pages);
	else {
		err = -ENOMEM;
1556
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567
		goto out_dput;
	}
	if (prev < 0) {
		swap_list.head = p->next;
	} else {
		swap_info[prev].next = p->next;
	}
	if (type == swap_list.next) {
		/* just pick something that's safe... */
		swap_list.next = swap_list.head;
	}
1568 1569 1570 1571 1572
	if (p->prio < 0) {
		for (i = p->next; i >= 0; i = swap_info[i].next)
			swap_info[i].prio = p->prio--;
		least_priority++;
	}
L
Linus Torvalds 已提交
1573 1574 1575
	nr_swap_pages -= p->pages;
	total_swap_pages -= p->pages;
	p->flags &= ~SWP_WRITEOK;
1576
	spin_unlock(&swap_lock);
1577

L
Linus Torvalds 已提交
1578 1579 1580 1581 1582 1583
	current->flags |= PF_SWAPOFF;
	err = try_to_unuse(type);
	current->flags &= ~PF_SWAPOFF;

	if (err) {
		/* re-insert swap space back into swap_list */
1584
		spin_lock(&swap_lock);
1585 1586 1587 1588
		if (p->prio < 0)
			p->prio = --least_priority;
		prev = -1;
		for (i = swap_list.head; i >= 0; i = swap_info[i].next) {
L
Linus Torvalds 已提交
1589 1590
			if (p->prio >= swap_info[i].prio)
				break;
1591 1592
			prev = i;
		}
L
Linus Torvalds 已提交
1593 1594 1595 1596 1597 1598 1599 1600
		p->next = i;
		if (prev < 0)
			swap_list.head = swap_list.next = p - swap_info;
		else
			swap_info[prev].next = p - swap_info;
		nr_swap_pages += p->pages;
		total_swap_pages += p->pages;
		p->flags |= SWP_WRITEOK;
1601
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
1602 1603
		goto out_dput;
	}
1604 1605 1606 1607 1608

	/* wait for any unplug function to finish */
	down_write(&swap_unplug_sem);
	up_write(&swap_unplug_sem);

1609
	destroy_swap_extents(p);
I
Ingo Molnar 已提交
1610
	mutex_lock(&swapon_mutex);
1611 1612 1613
	spin_lock(&swap_lock);
	drain_mmlist();

1614 1615 1616
	/* wait for anyone still in scan_swap_map */
	p->highest_bit = 0;		/* cuts scans short */
	while (p->flags >= SWP_SCANNING) {
1617
		spin_unlock(&swap_lock);
1618
		schedule_timeout_uninterruptible(1);
1619
		spin_lock(&swap_lock);
1620 1621
	}

L
Linus Torvalds 已提交
1622 1623 1624 1625 1626 1627
	swap_file = p->swap_file;
	p->swap_file = NULL;
	p->max = 0;
	swap_map = p->swap_map;
	p->swap_map = NULL;
	p->flags = 0;
1628
	spin_unlock(&swap_lock);
I
Ingo Molnar 已提交
1629
	mutex_unlock(&swapon_mutex);
L
Linus Torvalds 已提交
1630
	vfree(swap_map);
1631 1632 1633
	/* Destroy swap account informatin */
	swap_cgroup_swapoff(type);

L
Linus Torvalds 已提交
1634 1635 1636 1637 1638 1639
	inode = mapping->host;
	if (S_ISBLK(inode->i_mode)) {
		struct block_device *bdev = I_BDEV(inode);
		set_blocksize(bdev, p->old_block_size);
		bd_release(bdev);
	} else {
1640
		mutex_lock(&inode->i_mutex);
L
Linus Torvalds 已提交
1641
		inode->i_flags &= ~S_SWAPFILE;
1642
		mutex_unlock(&inode->i_mutex);
L
Linus Torvalds 已提交
1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660
	}
	filp_close(swap_file, NULL);
	err = 0;

out_dput:
	filp_close(victim, NULL);
out:
	return err;
}

#ifdef CONFIG_PROC_FS
/* iterator */
static void *swap_start(struct seq_file *swap, loff_t *pos)
{
	struct swap_info_struct *ptr = swap_info;
	int i;
	loff_t l = *pos;

I
Ingo Molnar 已提交
1661
	mutex_lock(&swapon_mutex);
L
Linus Torvalds 已提交
1662

1663 1664 1665
	if (!l)
		return SEQ_START_TOKEN;

L
Linus Torvalds 已提交
1666 1667 1668
	for (i = 0; i < nr_swapfiles; i++, ptr++) {
		if (!(ptr->flags & SWP_USED) || !ptr->swap_map)
			continue;
1669
		if (!--l)
L
Linus Torvalds 已提交
1670 1671 1672 1673 1674 1675 1676 1677
			return ptr;
	}

	return NULL;
}

static void *swap_next(struct seq_file *swap, void *v, loff_t *pos)
{
1678
	struct swap_info_struct *ptr;
L
Linus Torvalds 已提交
1679 1680
	struct swap_info_struct *endptr = swap_info + nr_swapfiles;

1681 1682 1683 1684 1685 1686 1687 1688
	if (v == SEQ_START_TOKEN)
		ptr = swap_info;
	else {
		ptr = v;
		ptr++;
	}

	for (; ptr < endptr; ptr++) {
L
Linus Torvalds 已提交
1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699
		if (!(ptr->flags & SWP_USED) || !ptr->swap_map)
			continue;
		++*pos;
		return ptr;
	}

	return NULL;
}

static void swap_stop(struct seq_file *swap, void *v)
{
I
Ingo Molnar 已提交
1700
	mutex_unlock(&swapon_mutex);
L
Linus Torvalds 已提交
1701 1702 1703 1704 1705 1706 1707 1708
}

static int swap_show(struct seq_file *swap, void *v)
{
	struct swap_info_struct *ptr = v;
	struct file *file;
	int len;

1709 1710 1711 1712
	if (ptr == SEQ_START_TOKEN) {
		seq_puts(swap,"Filename\t\t\t\tType\t\tSize\tUsed\tPriority\n");
		return 0;
	}
L
Linus Torvalds 已提交
1713 1714

	file = ptr->swap_file;
1715
	len = seq_path(swap, &file->f_path, " \t\n\\");
1716
	seq_printf(swap, "%*s%s\t%u\t%u\t%d\n",
1717 1718
			len < 40 ? 40 - len : 1, " ",
			S_ISBLK(file->f_path.dentry->d_inode->i_mode) ?
L
Linus Torvalds 已提交
1719
				"partition" : "file\t",
1720 1721 1722
			ptr->pages << (PAGE_SHIFT - 10),
			ptr->inuse_pages << (PAGE_SHIFT - 10),
			ptr->prio);
L
Linus Torvalds 已提交
1723 1724 1725
	return 0;
}

1726
static const struct seq_operations swaps_op = {
L
Linus Torvalds 已提交
1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737
	.start =	swap_start,
	.next =		swap_next,
	.stop =		swap_stop,
	.show =		swap_show
};

static int swaps_open(struct inode *inode, struct file *file)
{
	return seq_open(file, &swaps_op);
}

1738
static const struct file_operations proc_swaps_operations = {
L
Linus Torvalds 已提交
1739 1740 1741 1742 1743 1744 1745 1746
	.open		= swaps_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
	.release	= seq_release,
};

static int __init procswaps_init(void)
{
1747
	proc_create("swaps", 0, NULL, &proc_swaps_operations);
L
Linus Torvalds 已提交
1748 1749 1750 1751 1752
	return 0;
}
__initcall(procswaps_init);
#endif /* CONFIG_PROC_FS */

J
Jan Beulich 已提交
1753 1754 1755 1756 1757 1758 1759 1760 1761
#ifdef MAX_SWAPFILES_CHECK
static int __init max_swapfiles_check(void)
{
	MAX_SWAPFILES_CHECK();
	return 0;
}
late_initcall(max_swapfiles_check);
#endif

L
Linus Torvalds 已提交
1762 1763 1764 1765 1766
/*
 * Written 01/25/92 by Simmule Turner, heavily changed by Linus.
 *
 * The swapon system call
 */
1767
SYSCALL_DEFINE2(swapon, const char __user *, specialfile, int, swap_flags)
L
Linus Torvalds 已提交
1768 1769 1770 1771 1772 1773 1774 1775 1776 1777
{
	struct swap_info_struct * p;
	char *name = NULL;
	struct block_device *bdev = NULL;
	struct file *swap_file = NULL;
	struct address_space *mapping;
	unsigned int type;
	int i, prev;
	int error;
	union swap_header *swap_header = NULL;
1778 1779
	unsigned int nr_good_pages = 0;
	int nr_extents = 0;
1780
	sector_t span;
L
Linus Torvalds 已提交
1781
	unsigned long maxpages = 1;
1782
	unsigned long swapfilepages;
1783
	unsigned short *swap_map = NULL;
L
Linus Torvalds 已提交
1784 1785 1786 1787 1788 1789
	struct page *page = NULL;
	struct inode *inode = NULL;
	int did_down = 0;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
1790
	spin_lock(&swap_lock);
L
Linus Torvalds 已提交
1791 1792 1793 1794 1795
	p = swap_info;
	for (type = 0 ; type < nr_swapfiles ; type++,p++)
		if (!(p->flags & SWP_USED))
			break;
	error = -EPERM;
1796
	if (type >= MAX_SWAPFILES) {
1797
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
1798 1799 1800 1801
		goto out;
	}
	if (type >= nr_swapfiles)
		nr_swapfiles = type+1;
1802
	memset(p, 0, sizeof(*p));
L
Linus Torvalds 已提交
1803 1804 1805
	INIT_LIST_HEAD(&p->extent_list);
	p->flags = SWP_USED;
	p->next = -1;
1806
	spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839
	name = getname(specialfile);
	error = PTR_ERR(name);
	if (IS_ERR(name)) {
		name = NULL;
		goto bad_swap_2;
	}
	swap_file = filp_open(name, O_RDWR|O_LARGEFILE, 0);
	error = PTR_ERR(swap_file);
	if (IS_ERR(swap_file)) {
		swap_file = NULL;
		goto bad_swap_2;
	}

	p->swap_file = swap_file;
	mapping = swap_file->f_mapping;
	inode = mapping->host;

	error = -EBUSY;
	for (i = 0; i < nr_swapfiles; i++) {
		struct swap_info_struct *q = &swap_info[i];

		if (i == type || !q->swap_file)
			continue;
		if (mapping == q->swap_file->f_mapping)
			goto bad_swap;
	}

	error = -EINVAL;
	if (S_ISBLK(inode->i_mode)) {
		bdev = I_BDEV(inode);
		error = bd_claim(bdev, sys_swapon);
		if (error < 0) {
			bdev = NULL;
R
Rob Landley 已提交
1840
			error = -EINVAL;
L
Linus Torvalds 已提交
1841 1842 1843 1844 1845 1846 1847 1848 1849
			goto bad_swap;
		}
		p->old_block_size = block_size(bdev);
		error = set_blocksize(bdev, PAGE_SIZE);
		if (error < 0)
			goto bad_swap;
		p->bdev = bdev;
	} else if (S_ISREG(inode->i_mode)) {
		p->bdev = inode->i_sb->s_bdev;
1850
		mutex_lock(&inode->i_mutex);
L
Linus Torvalds 已提交
1851 1852 1853 1854 1855 1856 1857 1858 1859
		did_down = 1;
		if (IS_SWAPFILE(inode)) {
			error = -EBUSY;
			goto bad_swap;
		}
	} else {
		goto bad_swap;
	}

1860
	swapfilepages = i_size_read(inode) >> PAGE_SHIFT;
L
Linus Torvalds 已提交
1861 1862 1863 1864 1865 1866 1867 1868

	/*
	 * Read the swap header.
	 */
	if (!mapping->a_ops->readpage) {
		error = -EINVAL;
		goto bad_swap;
	}
1869
	page = read_mapping_page(mapping, 0, swap_file);
L
Linus Torvalds 已提交
1870 1871 1872 1873
	if (IS_ERR(page)) {
		error = PTR_ERR(page);
		goto bad_swap;
	}
1874
	swap_header = kmap(page);
L
Linus Torvalds 已提交
1875

1876
	if (memcmp("SWAPSPACE2", swap_header->magic.magic, 10)) {
1877
		printk(KERN_ERR "Unable to find swap-space signature\n");
L
Linus Torvalds 已提交
1878 1879 1880
		error = -EINVAL;
		goto bad_swap;
	}
1881

1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894
	/* swap partition endianess hack... */
	if (swab32(swap_header->info.version) == 1) {
		swab32s(&swap_header->info.version);
		swab32s(&swap_header->info.last_page);
		swab32s(&swap_header->info.nr_badpages);
		for (i = 0; i < swap_header->info.nr_badpages; i++)
			swab32s(&swap_header->info.badpages[i]);
	}
	/* Check the swap header's sub-version */
	if (swap_header->info.version != 1) {
		printk(KERN_WARNING
		       "Unable to handle swap header version %d\n",
		       swap_header->info.version);
L
Linus Torvalds 已提交
1895 1896
		error = -EINVAL;
		goto bad_swap;
1897
	}
L
Linus Torvalds 已提交
1898

1899 1900
	p->lowest_bit  = 1;
	p->cluster_next = 1;
1901

1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920
	/*
	 * Find out how many pages are allowed for a single swap
	 * device. There are two limiting factors: 1) the number of
	 * bits for the swap offset in the swp_entry_t type and
	 * 2) the number of bits in the a swap pte as defined by
	 * the different architectures. In order to find the
	 * largest possible bit mask a swap entry with swap type 0
	 * and swap offset ~0UL is created, encoded to a swap pte,
	 * decoded to a swp_entry_t again and finally the swap
	 * offset is extracted. This will mask all the bits from
	 * the initial ~0UL mask that can't be encoded in either
	 * the swp_entry_t or the architecture definition of a
	 * swap pte.
	 */
	maxpages = swp_offset(pte_to_swp_entry(
			swp_entry_to_pte(swp_entry(0, ~0UL)))) - 1;
	if (maxpages > swap_header->info.last_page)
		maxpages = swap_header->info.last_page;
	p->highest_bit = maxpages - 1;
L
Linus Torvalds 已提交
1921

1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933
	error = -EINVAL;
	if (!maxpages)
		goto bad_swap;
	if (swapfilepages && maxpages > swapfilepages) {
		printk(KERN_WARNING
		       "Swap area shorter than signature indicates\n");
		goto bad_swap;
	}
	if (swap_header->info.nr_badpages && S_ISREG(inode->i_mode))
		goto bad_swap;
	if (swap_header->info.nr_badpages > MAX_SWAP_BADPAGES)
		goto bad_swap;
1934

1935 1936 1937 1938 1939 1940
	/* OK, set up the swap map and apply the bad block list */
	swap_map = vmalloc(maxpages * sizeof(short));
	if (!swap_map) {
		error = -ENOMEM;
		goto bad_swap;
	}
L
Linus Torvalds 已提交
1941

1942 1943 1944 1945 1946
	memset(swap_map, 0, maxpages * sizeof(short));
	for (i = 0; i < swap_header->info.nr_badpages; i++) {
		int page_nr = swap_header->info.badpages[i];
		if (page_nr <= 0 || page_nr >= swap_header->info.last_page) {
			error = -EINVAL;
L
Linus Torvalds 已提交
1947
			goto bad_swap;
1948 1949
		}
		swap_map[page_nr] = SWAP_MAP_BAD;
L
Linus Torvalds 已提交
1950
	}
1951 1952 1953 1954 1955

	error = swap_cgroup_swapon(type, maxpages);
	if (error)
		goto bad_swap;

1956 1957 1958
	nr_good_pages = swap_header->info.last_page -
			swap_header->info.nr_badpages -
			1 /* header page */;
1959 1960

	if (nr_good_pages) {
1961
		swap_map[0] = SWAP_MAP_BAD;
1962 1963
		p->max = maxpages;
		p->pages = nr_good_pages;
1964 1965 1966
		nr_extents = setup_swap_extents(p, &span);
		if (nr_extents < 0) {
			error = nr_extents;
1967
			goto bad_swap;
1968
		}
1969 1970
		nr_good_pages = p->pages;
	}
L
Linus Torvalds 已提交
1971 1972 1973 1974 1975 1976
	if (!nr_good_pages) {
		printk(KERN_WARNING "Empty swap-file\n");
		error = -EINVAL;
		goto bad_swap;
	}

1977 1978 1979 1980
	if (blk_queue_nonrot(bdev_get_queue(p->bdev))) {
		p->flags |= SWP_SOLIDSTATE;
		p->cluster_next = 1 + (random32() % p->highest_bit);
	}
1981 1982 1983
	if (discard_swap(p) == 0)
		p->flags |= SWP_DISCARDABLE;

I
Ingo Molnar 已提交
1984
	mutex_lock(&swapon_mutex);
1985
	spin_lock(&swap_lock);
1986 1987 1988 1989 1990 1991
	if (swap_flags & SWAP_FLAG_PREFER)
		p->prio =
		  (swap_flags & SWAP_FLAG_PRIO_MASK) >> SWAP_FLAG_PRIO_SHIFT;
	else
		p->prio = --least_priority;
	p->swap_map = swap_map;
H
Hugh Dickins 已提交
1992
	p->flags |= SWP_WRITEOK;
L
Linus Torvalds 已提交
1993 1994
	nr_swap_pages += nr_good_pages;
	total_swap_pages += nr_good_pages;
1995

1996
	printk(KERN_INFO "Adding %uk swap on %s.  "
1997
			"Priority:%d extents:%d across:%lluk %s%s\n",
1998
		nr_good_pages<<(PAGE_SHIFT-10), name, p->prio,
1999
		nr_extents, (unsigned long long)span<<(PAGE_SHIFT-10),
2000 2001
		(p->flags & SWP_SOLIDSTATE) ? "SS" : "",
		(p->flags & SWP_DISCARDABLE) ? "D" : "");
L
Linus Torvalds 已提交
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016

	/* insert swap space into swap_list: */
	prev = -1;
	for (i = swap_list.head; i >= 0; i = swap_info[i].next) {
		if (p->prio >= swap_info[i].prio) {
			break;
		}
		prev = i;
	}
	p->next = i;
	if (prev < 0) {
		swap_list.head = swap_list.next = p - swap_info;
	} else {
		swap_info[prev].next = p - swap_info;
	}
2017
	spin_unlock(&swap_lock);
I
Ingo Molnar 已提交
2018
	mutex_unlock(&swapon_mutex);
L
Linus Torvalds 已提交
2019 2020 2021 2022 2023 2024 2025
	error = 0;
	goto out;
bad_swap:
	if (bdev) {
		set_blocksize(bdev, p->old_block_size);
		bd_release(bdev);
	}
2026
	destroy_swap_extents(p);
2027
	swap_cgroup_swapoff(type);
L
Linus Torvalds 已提交
2028
bad_swap_2:
2029
	spin_lock(&swap_lock);
L
Linus Torvalds 已提交
2030 2031
	p->swap_file = NULL;
	p->flags = 0;
2032
	spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045
	vfree(swap_map);
	if (swap_file)
		filp_close(swap_file, NULL);
out:
	if (page && !IS_ERR(page)) {
		kunmap(page);
		page_cache_release(page);
	}
	if (name)
		putname(name);
	if (did_down) {
		if (!error)
			inode->i_flags |= S_SWAPFILE;
2046
		mutex_unlock(&inode->i_mutex);
L
Linus Torvalds 已提交
2047 2048 2049 2050 2051 2052 2053 2054 2055
	}
	return error;
}

void si_swapinfo(struct sysinfo *val)
{
	unsigned int i;
	unsigned long nr_to_be_unused = 0;

2056
	spin_lock(&swap_lock);
L
Linus Torvalds 已提交
2057 2058 2059 2060 2061 2062 2063 2064
	for (i = 0; i < nr_swapfiles; i++) {
		if (!(swap_info[i].flags & SWP_USED) ||
		     (swap_info[i].flags & SWP_WRITEOK))
			continue;
		nr_to_be_unused += swap_info[i].inuse_pages;
	}
	val->freeswap = nr_swap_pages + nr_to_be_unused;
	val->totalswap = total_swap_pages + nr_to_be_unused;
2065
	spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
2066 2067 2068 2069 2070 2071 2072
}

/*
 * Verify that a swap entry is valid and increment its swap map count.
 *
 * Note: if swap_map[] reaches SWAP_MAP_MAX the entries are treated as
 * "permanent", but will be reclaimed by the next swapoff.
2073 2074 2075 2076 2077 2078
 * Returns error code in following case.
 * - success -> 0
 * - swp_entry is invalid -> EINVAL
 * - swp_entry is migration entry -> EINVAL
 * - swap-cache reference is requested but there is already one. -> EEXIST
 * - swap-cache reference is requested but the entry is not used. -> ENOENT
L
Linus Torvalds 已提交
2079
 */
2080
static int __swap_duplicate(swp_entry_t entry, bool cache)
L
Linus Torvalds 已提交
2081 2082 2083
{
	struct swap_info_struct * p;
	unsigned long offset, type;
2084 2085 2086
	int result = -EINVAL;
	int count;
	bool has_cache;
L
Linus Torvalds 已提交
2087

2088
	if (is_migration_entry(entry))
2089
		return -EINVAL;
2090

L
Linus Torvalds 已提交
2091 2092 2093 2094 2095 2096
	type = swp_type(entry);
	if (type >= nr_swapfiles)
		goto bad_file;
	p = type + swap_info;
	offset = swp_offset(entry);

2097
	spin_lock(&swap_lock);
2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121

	if (unlikely(offset >= p->max))
		goto unlock_out;

	count = swap_count(p->swap_map[offset]);
	has_cache = swap_has_cache(p->swap_map[offset]);

	if (cache == SWAP_CACHE) { /* called for swapcache/swapin-readahead */

		/* set SWAP_HAS_CACHE if there is no cache and entry is used */
		if (!has_cache && count) {
			p->swap_map[offset] = encode_swapmap(count, true);
			result = 0;
		} else if (has_cache) /* someone added cache */
			result = -EEXIST;
		else if (!count) /* no users */
			result = -ENOENT;

	} else if (count || has_cache) {
		if (count < SWAP_MAP_MAX - 1) {
			p->swap_map[offset] = encode_swapmap(count + 1,
							     has_cache);
			result = 0;
		} else if (count <= SWAP_MAP_MAX) {
L
Linus Torvalds 已提交
2122
			if (swap_overflow++ < 5)
2123 2124 2125 2126 2127
				printk(KERN_WARNING
				       "swap_dup: swap entry overflow\n");
			p->swap_map[offset] = encode_swapmap(SWAP_MAP_MAX,
							      has_cache);
			result = 0;
L
Linus Torvalds 已提交
2128
		}
2129 2130 2131
	} else
		result = -ENOENT; /* unused swap entry */
unlock_out:
2132
	spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
2133 2134 2135 2136 2137 2138 2139
out:
	return result;

bad_file:
	printk(KERN_ERR "swap_dup: %s%08lx\n", Bad_file, entry.val);
	goto out;
}
2140 2141 2142 2143 2144 2145 2146
/*
 * increase reference count of swap entry by 1.
 */
void swap_duplicate(swp_entry_t entry)
{
	__swap_duplicate(entry, SWAP_MAP);
}
L
Linus Torvalds 已提交
2147

2148
/*
2149 2150
 * @entry: swap entry for which we allocate swap cache.
 *
2151
 * Called when allocating swap cache for exising swap entry,
2152 2153 2154
 * This can return error codes. Returns 0 at success.
 * -EBUSY means there is a swap cache.
 * Note: return code is different from swap_duplicate().
2155 2156 2157
 */
int swapcache_prepare(swp_entry_t entry)
{
2158
	return __swap_duplicate(entry, SWAP_CACHE);
2159 2160 2161
}


L
Linus Torvalds 已提交
2162 2163 2164 2165 2166 2167 2168
struct swap_info_struct *
get_swap_info_struct(unsigned type)
{
	return &swap_info[type];
}

/*
2169
 * swap_lock prevents swap_map being freed. Don't grab an extra
L
Linus Torvalds 已提交
2170 2171 2172 2173
 * reference on the swaphandle, it doesn't matter if it becomes unused.
 */
int valid_swaphandles(swp_entry_t entry, unsigned long *offset)
{
2174
	struct swap_info_struct *si;
H
Hugh Dickins 已提交
2175
	int our_page_cluster = page_cluster;
2176 2177 2178
	pgoff_t target, toff;
	pgoff_t base, end;
	int nr_pages = 0;
L
Linus Torvalds 已提交
2179

H
Hugh Dickins 已提交
2180
	if (!our_page_cluster)	/* no readahead */
L
Linus Torvalds 已提交
2181
		return 0;
2182 2183 2184 2185 2186 2187 2188

	si = &swap_info[swp_type(entry)];
	target = swp_offset(entry);
	base = (target >> our_page_cluster) << our_page_cluster;
	end = base + (1 << our_page_cluster);
	if (!base)		/* first page is swap header */
		base++;
L
Linus Torvalds 已提交
2189

2190
	spin_lock(&swap_lock);
2191 2192 2193 2194 2195 2196 2197 2198
	if (end > si->max)	/* don't go beyond end of map */
		end = si->max;

	/* Count contiguous allocated slots above our target */
	for (toff = target; ++toff < end; nr_pages++) {
		/* Don't read in free or bad pages */
		if (!si->swap_map[toff])
			break;
2199
		if (swap_count(si->swap_map[toff]) == SWAP_MAP_BAD)
L
Linus Torvalds 已提交
2200
			break;
2201 2202 2203
	}
	/* Count contiguous allocated slots below our target */
	for (toff = target; --toff >= base; nr_pages++) {
L
Linus Torvalds 已提交
2204
		/* Don't read in free or bad pages */
2205
		if (!si->swap_map[toff])
L
Linus Torvalds 已提交
2206
			break;
2207
		if (swap_count(si->swap_map[toff]) == SWAP_MAP_BAD)
L
Linus Torvalds 已提交
2208
			break;
2209
	}
2210
	spin_unlock(&swap_lock);
2211 2212 2213 2214 2215 2216 2217

	/*
	 * Indicate starting offset, and return number of pages to get:
	 * if only 1, say 0, since there's then no readahead to be done.
	 */
	*offset = ++toff;
	return nr_pages? ++nr_pages: 0;
L
Linus Torvalds 已提交
2218
}