swapfile.c 61.7 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
/*
 *  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

H
Hugh Dickins 已提交
38 39 40
static bool swap_count_continued(struct swap_info_struct *, pgoff_t,
				 unsigned char);
static void free_swap_count_continuations(struct swap_info_struct *);
41
static sector_t map_swap_entry(swp_entry_t, struct block_device**);
H
Hugh Dickins 已提交
42

A
Adrian Bunk 已提交
43 44
static DEFINE_SPINLOCK(swap_lock);
static unsigned int nr_swapfiles;
45
long nr_swap_pages;
L
Linus Torvalds 已提交
46
long total_swap_pages;
47
static int least_priority;
L
Linus Torvalds 已提交
48 49 50 51 52 53

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
54
static struct swap_list_t swap_list = {-1, -1};
L
Linus Torvalds 已提交
55

56
static struct swap_info_struct *swap_info[MAX_SWAPFILES];
L
Linus Torvalds 已提交
57

I
Ingo Molnar 已提交
58
static DEFINE_MUTEX(swapon_mutex);
L
Linus Torvalds 已提交
59

60
static inline unsigned char swap_count(unsigned char ent)
61
{
H
Hugh Dickins 已提交
62
	return ent & ~SWAP_HAS_CACHE;	/* may include SWAP_HAS_CONT flag */
63 64
}

65
/* returns 1 if swap entry is freed */
66 67 68
static int
__try_to_reclaim_swap(struct swap_info_struct *si, unsigned long offset)
{
69
	swp_entry_t entry = swp_entry(si->type, offset);
70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89
	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;
}
90

L
Linus Torvalds 已提交
91 92
/*
 * We need this because the bdev->unplug_fn can sleep and we cannot
93
 * hold swap_lock while calling the unplug_fn. And swap_lock
I
Ingo Molnar 已提交
94
 * cannot be turned into a mutex.
L
Linus Torvalds 已提交
95 96 97 98 99 100 101 102
 */
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
103
	entry.val = page_private(page);
L
Linus Torvalds 已提交
104
	if (PageSwapCache(page)) {
105
		struct block_device *bdev = swap_info[swp_type(entry)]->bdev;
L
Linus Torvalds 已提交
106 107 108 109 110
		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
111 112
		 * 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
113 114 115 116 117 118
		 * 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
119
		blk_run_backing_dev(bdi, page);
L
Linus Torvalds 已提交
120 121 122 123
	}
	up_read(&swap_unplug_sem);
}

124 125 126 127 128 129 130
/*
 * 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;
131 132
	sector_t start_block;
	sector_t nr_blocks;
133 134
	int err = 0;

135 136 137 138 139 140 141 142 143 144 145
	/* Do not discard the swap header page! */
	se = &si->first_swap_extent;
	start_block = (se->start_block + 1) << (PAGE_SHIFT - 9);
	nr_blocks = ((sector_t)se->nr_pages - 1) << (PAGE_SHIFT - 9);
	if (nr_blocks) {
		err = blkdev_issue_discard(si->bdev, start_block,
				nr_blocks, GFP_KERNEL, DISCARD_FL_BARRIER);
		if (err)
			return err;
		cond_resched();
	}
146

147 148 149
	list_for_each_entry(se, &si->first_swap_extent.list, list) {
		start_block = se->start_block << (PAGE_SHIFT - 9);
		nr_blocks = (sector_t)se->nr_pages << (PAGE_SHIFT - 9);
150 151

		err = blkdev_issue_discard(si->bdev, start_block,
152
				nr_blocks, GFP_KERNEL, DISCARD_FL_BARRIER);
153 154 155 156 157 158 159 160
		if (err)
			break;

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

161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177
/*
 * 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;
178
			sector_t nr_blocks = se->nr_pages - offset;
179 180 181 182 183 184 185 186 187 188 189 190

			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,
191
				    nr_blocks, GFP_NOIO, DISCARD_FL_BARRIER))
192 193 194 195 196 197 198 199 200 201 202 203 204 205
				break;
		}

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

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

206 207 208
#define SWAPFILE_CLUSTER	256
#define LATENCY_LIMIT		256

209
static inline unsigned long scan_swap_map(struct swap_info_struct *si,
210
					  unsigned char usage)
L
Linus Torvalds 已提交
211
{
212
	unsigned long offset;
213
	unsigned long scan_base;
214
	unsigned long last_in_cluster = 0;
215
	int latency_ration = LATENCY_LIMIT;
216
	int found_free_cluster = 0;
217

218
	/*
219 220 221 222 223 224 225
	 * 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
226
	 * And we let swap pages go all over an SSD partition.  Hugh
227 228
	 */

229
	si->flags += SWP_SCANNING;
230
	scan_base = offset = si->cluster_next;
231 232 233 234 235 236

	if (unlikely(!si->cluster_nr--)) {
		if (si->pages - si->inuse_pages < SWAPFILE_CLUSTER) {
			si->cluster_nr = SWAPFILE_CLUSTER - 1;
			goto checks;
		}
237 238 239 240 241 242 243 244 245 246 247 248 249
		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;
		}
250
		spin_unlock(&swap_lock);
251

252 253 254 255 256 257 258 259 260 261
		/*
		 * 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;
262 263 264 265
		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
266
			if (si->swap_map[offset])
267 268
				last_in_cluster = offset + SWAPFILE_CLUSTER;
			else if (offset == last_in_cluster) {
269
				spin_lock(&swap_lock);
270 271 272
				offset -= SWAPFILE_CLUSTER - 1;
				si->cluster_next = offset;
				si->cluster_nr = SWAPFILE_CLUSTER - 1;
273
				found_free_cluster = 1;
274
				goto checks;
L
Linus Torvalds 已提交
275
			}
276 277 278 279
			if (unlikely(--latency_ration < 0)) {
				cond_resched();
				latency_ration = LATENCY_LIMIT;
			}
280
		}
281 282

		offset = si->lowest_bit;
283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303
		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;
304
		spin_lock(&swap_lock);
305
		si->cluster_nr = SWAPFILE_CLUSTER - 1;
306
		si->lowest_alloc = 0;
L
Linus Torvalds 已提交
307
	}
308

309 310
checks:
	if (!(si->flags & SWP_WRITEOK))
311
		goto no_page;
312 313
	if (!si->highest_bit)
		goto no_page;
314
	if (offset > si->highest_bit)
315
		scan_base = offset = si->lowest_bit;
316 317 318 319 320 321 322 323 324 325 326 327 328

	/* 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 */
	}

329 330 331 332 333 334 335 336 337 338 339
	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
340
	}
H
Hugh Dickins 已提交
341
	si->swap_map[offset] = usage;
342 343
	si->cluster_next = offset + 1;
	si->flags -= SWP_SCANNING;
344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397

	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;
		}
	}
398
	return offset;
399

400
scan:
401
	spin_unlock(&swap_lock);
402
	while (++offset <= si->highest_bit) {
403
		if (!si->swap_map[offset]) {
404
			spin_lock(&swap_lock);
405 406
			goto checks;
		}
407 408 409 410
		if (vm_swap_full() && si->swap_map[offset] == SWAP_HAS_CACHE) {
			spin_lock(&swap_lock);
			goto checks;
		}
411 412 413 414
		if (unlikely(--latency_ration < 0)) {
			cond_resched();
			latency_ration = LATENCY_LIMIT;
		}
415
	}
416 417 418 419 420 421
	offset = si->lowest_bit;
	while (++offset < scan_base) {
		if (!si->swap_map[offset]) {
			spin_lock(&swap_lock);
			goto checks;
		}
422 423 424 425
		if (vm_swap_full() && si->swap_map[offset] == SWAP_HAS_CACHE) {
			spin_lock(&swap_lock);
			goto checks;
		}
426 427 428 429 430
		if (unlikely(--latency_ration < 0)) {
			cond_resched();
			latency_ration = LATENCY_LIMIT;
		}
	}
431
	spin_lock(&swap_lock);
432 433

no_page:
434
	si->flags -= SWP_SCANNING;
L
Linus Torvalds 已提交
435 436 437 438 439
	return 0;
}

swp_entry_t get_swap_page(void)
{
440 441 442 443
	struct swap_info_struct *si;
	pgoff_t offset;
	int type, next;
	int wrapped = 0;
L
Linus Torvalds 已提交
444

445
	spin_lock(&swap_lock);
L
Linus Torvalds 已提交
446
	if (nr_swap_pages <= 0)
447 448 449 450
		goto noswap;
	nr_swap_pages--;

	for (type = swap_list.next; type >= 0 && wrapped < 2; type = next) {
451
		si = swap_info[type];
452 453
		next = si->next;
		if (next < 0 ||
454
		    (!wrapped && si->prio != swap_info[next]->prio)) {
455 456
			next = swap_list.head;
			wrapped++;
L
Linus Torvalds 已提交
457
		}
458 459 460 461 462 463 464

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

		swap_list.next = next;
465
		/* This is called for allocating swap entry for cache */
H
Hugh Dickins 已提交
466
		offset = scan_swap_map(si, SWAP_HAS_CACHE);
467 468
		if (offset) {
			spin_unlock(&swap_lock);
469
			return swp_entry(type, offset);
470
		}
471
		next = swap_list.next;
L
Linus Torvalds 已提交
472
	}
473 474 475

	nr_swap_pages++;
noswap:
476
	spin_unlock(&swap_lock);
477
	return (swp_entry_t) {0};
L
Linus Torvalds 已提交
478 479
}

480
/* The only caller of this function is now susupend routine */
481 482 483 484 485 486
swp_entry_t get_swap_page_of_type(int type)
{
	struct swap_info_struct *si;
	pgoff_t offset;

	spin_lock(&swap_lock);
487 488
	si = swap_info[type];
	if (si && (si->flags & SWP_WRITEOK)) {
489
		nr_swap_pages--;
490
		/* This is called for allocating swap entry, not cache */
H
Hugh Dickins 已提交
491
		offset = scan_swap_map(si, 1);
492 493 494 495 496 497 498 499 500 501
		if (offset) {
			spin_unlock(&swap_lock);
			return swp_entry(type, offset);
		}
		nr_swap_pages++;
	}
	spin_unlock(&swap_lock);
	return (swp_entry_t) {0};
}

502
static struct swap_info_struct *swap_info_get(swp_entry_t entry)
L
Linus Torvalds 已提交
503
{
504
	struct swap_info_struct *p;
L
Linus Torvalds 已提交
505 506 507 508 509 510 511
	unsigned long offset, type;

	if (!entry.val)
		goto out;
	type = swp_type(entry);
	if (type >= nr_swapfiles)
		goto bad_nofile;
512
	p = swap_info[type];
L
Linus Torvalds 已提交
513 514 515 516 517 518 519
	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;
520
	spin_lock(&swap_lock);
L
Linus Torvalds 已提交
521 522 523 524 525 526 527 528 529 530 531 532 533 534 535
	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;
536
}
L
Linus Torvalds 已提交
537

538 539
static unsigned char swap_entry_free(struct swap_info_struct *p,
				     swp_entry_t entry, unsigned char usage)
L
Linus Torvalds 已提交
540
{
H
Hugh Dickins 已提交
541
	unsigned long offset = swp_offset(entry);
542 543
	unsigned char count;
	unsigned char has_cache;
544

H
Hugh Dickins 已提交
545 546 547
	count = p->swap_map[offset];
	has_cache = count & SWAP_HAS_CACHE;
	count &= ~SWAP_HAS_CACHE;
548

H
Hugh Dickins 已提交
549
	if (usage == SWAP_HAS_CACHE) {
550
		VM_BUG_ON(!has_cache);
H
Hugh Dickins 已提交
551
		has_cache = 0;
H
Hugh Dickins 已提交
552 553 554 555 556 557
	} else if (count == SWAP_MAP_SHMEM) {
		/*
		 * Or we could insist on shmem.c using a special
		 * swap_shmem_free() and free_shmem_swap_and_cache()...
		 */
		count = 0;
H
Hugh Dickins 已提交
558 559 560 561 562 563 564 565 566
	} else if ((count & ~COUNT_CONTINUED) <= SWAP_MAP_MAX) {
		if (count == COUNT_CONTINUED) {
			if (swap_count_continued(p, offset, count))
				count = SWAP_MAP_MAX | COUNT_CONTINUED;
			else
				count = SWAP_MAP_MAX;
		} else
			count--;
	}
H
Hugh Dickins 已提交
567 568 569 570 571 572

	if (!count)
		mem_cgroup_uncharge_swap(entry);

	usage = count | has_cache;
	p->swap_map[offset] = usage;
573 574

	/* free if no reference */
H
Hugh Dickins 已提交
575
	if (!usage) {
576 577 578 579
		if (offset < p->lowest_bit)
			p->lowest_bit = offset;
		if (offset > p->highest_bit)
			p->highest_bit = offset;
580 581 582
		if (swap_list.next >= 0 &&
		    p->prio > swap_info[swap_list.next]->prio)
			swap_list.next = p->type;
583 584
		nr_swap_pages++;
		p->inuse_pages--;
L
Linus Torvalds 已提交
585
	}
H
Hugh Dickins 已提交
586 587

	return usage;
L
Linus Torvalds 已提交
588 589 590 591 592 593 594 595
}

/*
 * 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)
{
596
	struct swap_info_struct *p;
L
Linus Torvalds 已提交
597 598 599

	p = swap_info_get(entry);
	if (p) {
H
Hugh Dickins 已提交
600
		swap_entry_free(p, entry, 1);
601
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
602 603 604
	}
}

605 606 607 608 609
/*
 * Called after dropping swapcache to decrease refcnt to swap entries.
 */
void swapcache_free(swp_entry_t entry, struct page *page)
{
610
	struct swap_info_struct *p;
611
	unsigned char count;
612 613 614

	p = swap_info_get(entry);
	if (p) {
H
Hugh Dickins 已提交
615 616 617
		count = swap_entry_free(p, entry, SWAP_HAS_CACHE);
		if (page)
			mem_cgroup_uncharge_swapcache(page, entry, count != 0);
618 619
		spin_unlock(&swap_lock);
	}
620 621
}

L
Linus Torvalds 已提交
622
/*
623
 * How many references to page are currently swapped out?
H
Hugh Dickins 已提交
624 625
 * This does not give an exact answer when swap count is continued,
 * but does include the high COUNT_CONTINUED flag to allow for that.
L
Linus Torvalds 已提交
626
 */
627
static inline int page_swapcount(struct page *page)
L
Linus Torvalds 已提交
628
{
629 630
	int count = 0;
	struct swap_info_struct *p;
L
Linus Torvalds 已提交
631 632
	swp_entry_t entry;

H
Hugh Dickins 已提交
633
	entry.val = page_private(page);
L
Linus Torvalds 已提交
634 635
	p = swap_info_get(entry);
	if (p) {
636
		count = swap_count(p->swap_map[swp_offset(entry)]);
637
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
638
	}
639
	return count;
L
Linus Torvalds 已提交
640 641 642
}

/*
643 644 645 646
 * 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
647
 */
648
int reuse_swap_page(struct page *page)
L
Linus Torvalds 已提交
649
{
650 651
	int count;

652
	VM_BUG_ON(!PageLocked(page));
653
	count = page_mapcount(page);
654
	if (count <= 1 && PageSwapCache(page)) {
655
		count += page_swapcount(page);
656 657 658 659 660
		if (count == 1 && !PageWriteback(page)) {
			delete_from_swap_cache(page);
			SetPageDirty(page);
		}
	}
661
	return count == 1;
L
Linus Torvalds 已提交
662 663 664
}

/*
665 666
 * 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
667
 */
668
int try_to_free_swap(struct page *page)
L
Linus Torvalds 已提交
669
{
670
	VM_BUG_ON(!PageLocked(page));
L
Linus Torvalds 已提交
671 672 673 674 675

	if (!PageSwapCache(page))
		return 0;
	if (PageWriteback(page))
		return 0;
676
	if (page_swapcount(page))
L
Linus Torvalds 已提交
677 678
		return 0;

679 680 681
	delete_from_swap_cache(page);
	SetPageDirty(page);
	return 1;
682 683
}

L
Linus Torvalds 已提交
684 685 686 687
/*
 * Free the swap entry like above, but also try to
 * free the page cache entry if it is the last user.
 */
688
int free_swap_and_cache(swp_entry_t entry)
L
Linus Torvalds 已提交
689
{
690
	struct swap_info_struct *p;
L
Linus Torvalds 已提交
691 692
	struct page *page = NULL;

693
	if (non_swap_entry(entry))
694
		return 1;
695

L
Linus Torvalds 已提交
696 697
	p = swap_info_get(entry);
	if (p) {
H
Hugh Dickins 已提交
698
		if (swap_entry_free(p, entry, 1) == SWAP_HAS_CACHE) {
699
			page = find_get_page(&swapper_space, entry.val);
N
Nick Piggin 已提交
700
			if (page && !trylock_page(page)) {
701 702 703 704
				page_cache_release(page);
				page = NULL;
			}
		}
705
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
706 707
	}
	if (page) {
708 709 710 711
		/*
		 * Not mapped elsewhere, or swap space full? Free it!
		 * Also recheck PageSwapCache now page is locked (above).
		 */
712
		if (PageSwapCache(page) && !PageWriteback(page) &&
713
				(!page_mapped(page) || vm_swap_full())) {
L
Linus Torvalds 已提交
714 715 716 717 718 719
			delete_from_swap_cache(page);
			SetPageDirty(page);
		}
		unlock_page(page);
		page_cache_release(page);
	}
720
	return p != NULL;
L
Linus Torvalds 已提交
721 722
}

723
#ifdef CONFIG_HIBERNATION
724
/*
725
 * Find the swap type that corresponds to given device (if any).
726
 *
727 728 729 730
 * @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).
731
 */
732
int swap_type_of(dev_t device, sector_t offset, struct block_device **bdev_p)
733
{
734
	struct block_device *bdev = NULL;
735
	int type;
736

737 738 739
	if (device)
		bdev = bdget(device);

740
	spin_lock(&swap_lock);
741 742
	for (type = 0; type < nr_swapfiles; type++) {
		struct swap_info_struct *sis = swap_info[type];
743

744
		if (!(sis->flags & SWP_WRITEOK))
745
			continue;
746

747
		if (!bdev) {
748
			if (bdev_p)
749
				*bdev_p = bdgrab(sis->bdev);
750

751
			spin_unlock(&swap_lock);
752
			return type;
753
		}
754
		if (bdev == sis->bdev) {
755
			struct swap_extent *se = &sis->first_swap_extent;
756 757

			if (se->start_block == offset) {
758
				if (bdev_p)
759
					*bdev_p = bdgrab(sis->bdev);
760

761 762
				spin_unlock(&swap_lock);
				bdput(bdev);
763
				return type;
764
			}
765 766 767
		}
	}
	spin_unlock(&swap_lock);
768 769 770
	if (bdev)
		bdput(bdev);

771 772 773
	return -ENODEV;
}

774 775 776 777 778 779 780 781 782 783 784 785
/*
 * 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 type, pgoff_t offset)
{
	struct block_device *bdev;

	if ((unsigned int)type >= nr_swapfiles)
		return 0;
	if (!(swap_info[type]->flags & SWP_WRITEOK))
		return 0;
786
	return map_swap_entry(swp_entry(type, offset), &bdev);
787 788
}

789 790 791 792 793 794 795 796 797 798
/*
 * 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;

799 800 801 802 803 804
	spin_lock(&swap_lock);
	if ((unsigned int)type < nr_swapfiles) {
		struct swap_info_struct *sis = swap_info[type];

		if (sis->flags & SWP_WRITEOK) {
			n = sis->pages;
805
			if (free)
806
				n -= sis->inuse_pages;
807 808
		}
	}
809
	spin_unlock(&swap_lock);
810 811
	return n;
}
812
#endif /* CONFIG_HIBERNATION */
813

L
Linus Torvalds 已提交
814
/*
815 816 817
 * 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
818
 */
H
Hugh Dickins 已提交
819
static int unuse_pte(struct vm_area_struct *vma, pmd_t *pmd,
L
Linus Torvalds 已提交
820 821
		unsigned long addr, swp_entry_t entry, struct page *page)
{
822
	struct mem_cgroup *ptr = NULL;
H
Hugh Dickins 已提交
823 824 825 826
	spinlock_t *ptl;
	pte_t *pte;
	int ret = 1;

827
	if (mem_cgroup_try_charge_swapin(vma->vm_mm, page, GFP_KERNEL, &ptr)) {
H
Hugh Dickins 已提交
828
		ret = -ENOMEM;
829 830
		goto out_nolock;
	}
H
Hugh Dickins 已提交
831 832 833 834

	pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
	if (unlikely(!pte_same(*pte, swp_entry_to_pte(entry)))) {
		if (ret > 0)
835
			mem_cgroup_cancel_charge_swapin(ptr);
H
Hugh Dickins 已提交
836 837 838
		ret = 0;
		goto out;
	}
839

840
	inc_mm_counter(vma->vm_mm, anon_rss);
L
Linus Torvalds 已提交
841 842 843 844
	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);
845
	mem_cgroup_commit_charge_swapin(page, ptr);
L
Linus Torvalds 已提交
846 847 848 849 850 851
	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
852 853
out:
	pte_unmap_unlock(pte, ptl);
854
out_nolock:
H
Hugh Dickins 已提交
855
	return ret;
L
Linus Torvalds 已提交
856 857 858 859 860 861 862
}

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);
863
	pte_t *pte;
864
	int ret = 0;
L
Linus Torvalds 已提交
865

H
Hugh Dickins 已提交
866 867 868 869 870 871 872 873 874 875
	/*
	 * 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
876 877 878 879 880 881
	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
882 883 884 885 886
			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
887 888
		}
	} while (pte++, addr += PAGE_SIZE, addr != end);
H
Hugh Dickins 已提交
889 890
	pte_unmap(pte - 1);
out:
891
	return ret;
L
Linus Torvalds 已提交
892 893 894 895 896 897 898 899
}

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;
900
	int ret;
L
Linus Torvalds 已提交
901 902 903 904 905 906

	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
		if (pmd_none_or_clear_bad(pmd))
			continue;
907 908 909
		ret = unuse_pte_range(vma, pmd, addr, next, entry, page);
		if (ret)
			return ret;
L
Linus Torvalds 已提交
910 911 912 913 914 915 916 917 918 919
	} 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;
920
	int ret;
L
Linus Torvalds 已提交
921 922 923 924 925 926

	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
		if (pud_none_or_clear_bad(pud))
			continue;
927 928 929
		ret = unuse_pmd_range(vma, pud, addr, next, entry, page);
		if (ret)
			return ret;
L
Linus Torvalds 已提交
930 931 932 933 934 935 936 937 938
	} 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;
939
	int ret;
L
Linus Torvalds 已提交
940

H
Hugh Dickins 已提交
941
	if (page_anon_vma(page)) {
L
Linus Torvalds 已提交
942 943 944 945 946 947 948 949 950 951 952 953 954 955 956
		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;
957 958 959
		ret = unuse_pud_range(vma, pgd, addr, next, entry, page);
		if (ret)
			return ret;
L
Linus Torvalds 已提交
960 961 962 963 964 965 966 967
	} 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;
968
	int ret = 0;
L
Linus Torvalds 已提交
969 970 971

	if (!down_read_trylock(&mm->mmap_sem)) {
		/*
972 973
		 * 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
974
		 */
975
		activate_page(page);
L
Linus Torvalds 已提交
976 977 978 979 980
		unlock_page(page);
		down_read(&mm->mmap_sem);
		lock_page(page);
	}
	for (vma = mm->mmap; vma; vma = vma->vm_next) {
981
		if (vma->anon_vma && (ret = unuse_vma(vma, entry, page)))
L
Linus Torvalds 已提交
982 983 984
			break;
	}
	up_read(&mm->mmap_sem);
985
	return (ret < 0)? ret: 0;
L
Linus Torvalds 已提交
986 987 988 989 990 991
}

/*
 * Scan swap_map from current position to next entry still in use.
 * Recycle to start on reaching the end, returning 0 when empty.
 */
992 993
static unsigned int find_next_to_unuse(struct swap_info_struct *si,
					unsigned int prev)
L
Linus Torvalds 已提交
994
{
995 996
	unsigned int max = si->max;
	unsigned int i = prev;
997
	unsigned char count;
L
Linus Torvalds 已提交
998 999

	/*
1000
	 * No need for swap_lock here: we're just looking
L
Linus Torvalds 已提交
1001 1002
	 * for whether an entry is in use, not modifying it; false
	 * hits are okay, and sys_swapoff() has already prevented new
1003
	 * allocations from this area (while holding swap_lock).
L
Linus Torvalds 已提交
1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019
	 */
	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];
1020
		if (count && swap_count(count) != SWAP_MAP_BAD)
L
Linus Torvalds 已提交
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)
{
1033
	struct swap_info_struct *si = swap_info[type];
L
Linus Torvalds 已提交
1034
	struct mm_struct *start_mm;
1035 1036
	unsigned char *swap_map;
	unsigned char swcount;
L
Linus Torvalds 已提交
1037 1038
	struct page *page;
	swp_entry_t entry;
1039
	unsigned int i = 0;
L
Linus Torvalds 已提交
1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053
	int retval = 0;

	/*
	 * 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
H
Hugh Dickins 已提交
1054
	 * that.
L
Linus Torvalds 已提交
1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069
	 */
	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;
		}

1070
		/*
L
Linus Torvalds 已提交
1071 1072
		 * Get a page for the entry, using the existing swap
		 * cache page if there is one.  Otherwise, get a clean
1073
		 * page and read the swap into it.
L
Linus Torvalds 已提交
1074 1075 1076
		 */
		swap_map = &si->swap_map[i];
		entry = swp_entry(type, i);
1077 1078
		page = read_swap_cache_async(entry,
					GFP_HIGHUSER_MOVABLE, NULL, 0);
L
Linus Torvalds 已提交
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.
		 */
		swcount = *swap_map;
H
Hugh Dickins 已提交
1118 1119 1120 1121 1122 1123
		if (swap_count(swcount) == SWAP_MAP_SHMEM) {
			retval = shmem_unuse(entry, page);
			/* page has already been unlocked and released */
			if (retval < 0)
				break;
			continue;
L
Linus Torvalds 已提交
1124
		}
H
Hugh Dickins 已提交
1125 1126 1127
		if (swap_count(swcount) && start_mm != &init_mm)
			retval = unuse_mm(start_mm, entry, page);

1128
		if (swap_count(*swap_map)) {
L
Linus Torvalds 已提交
1129 1130 1131 1132 1133 1134 1135 1136 1137
			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);
H
Hugh Dickins 已提交
1138
			while (swap_count(*swap_map) && !retval &&
L
Linus Torvalds 已提交
1139 1140
					(p = p->next) != &start_mm->mmlist) {
				mm = list_entry(p, struct mm_struct, mmlist);
1141
				if (!atomic_inc_not_zero(&mm->mm_users))
L
Linus Torvalds 已提交
1142 1143 1144 1145 1146 1147 1148 1149
					continue;
				spin_unlock(&mmlist_lock);
				mmput(prev_mm);
				prev_mm = mm;

				cond_resched();

				swcount = *swap_map;
1150
				if (!swap_count(swcount)) /* any usage ? */
L
Linus Torvalds 已提交
1151
					;
H
Hugh Dickins 已提交
1152
				else if (mm == &init_mm)
L
Linus Torvalds 已提交
1153
					set_start_mm = 1;
H
Hugh Dickins 已提交
1154
				else
L
Linus Torvalds 已提交
1155
					retval = unuse_mm(mm, entry, page);
1156

1157
				if (set_start_mm && *swap_map < swcount) {
L
Linus Torvalds 已提交
1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188
					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;
		}
		if (retval) {
			unlock_page(page);
			page_cache_release(page);
			break;
		}

		/*
		 * 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.
		 */
1189 1190
		if (swap_count(*swap_map) &&
		     PageDirty(page) && PageSwapCache(page)) {
L
Linus Torvalds 已提交
1191 1192 1193 1194 1195 1196 1197 1198
			struct writeback_control wbc = {
				.sync_mode = WB_SYNC_NONE,
			};

			swap_writepage(page, &wbc);
			lock_page(page);
			wait_on_page_writeback(page);
		}
1199 1200 1201 1202 1203 1204 1205 1206 1207 1208

		/*
		 * 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))
1209
			delete_from_swap_cache(page);
L
Linus Torvalds 已提交
1210 1211 1212 1213

		/*
		 * So we could skip searching mms once swap count went
		 * to 1, we did not mark any present ptes as dirty: must
1214
		 * mark page dirty so shrink_page_list will preserve it.
L
Linus Torvalds 已提交
1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231
		 */
		SetPageDirty(page);
		unlock_page(page);
		page_cache_release(page);

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

	mmput(start_mm);
	return retval;
}

/*
1232 1233 1234
 * 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
1235 1236 1237 1238 1239
 * added to the mmlist just after page_duplicate - before would be racy.
 */
static void drain_mmlist(void)
{
	struct list_head *p, *next;
1240
	unsigned int type;
L
Linus Torvalds 已提交
1241

1242 1243
	for (type = 0; type < nr_swapfiles; type++)
		if (swap_info[type]->inuse_pages)
L
Linus Torvalds 已提交
1244 1245 1246 1247 1248 1249 1250 1251 1252
			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
1253 1254 1255
 * corresponds to page offset for the specified swap entry.
 * Note that the type of this function is sector_t, but it returns page offset
 * into the bdev, not sector offset.
L
Linus Torvalds 已提交
1256
 */
1257
static sector_t map_swap_entry(swp_entry_t entry, struct block_device **bdev)
L
Linus Torvalds 已提交
1258
{
H
Hugh Dickins 已提交
1259 1260 1261 1262 1263
	struct swap_info_struct *sis;
	struct swap_extent *start_se;
	struct swap_extent *se;
	pgoff_t offset;

1264
	sis = swap_info[swp_type(entry)];
H
Hugh Dickins 已提交
1265 1266 1267 1268 1269
	*bdev = sis->bdev;

	offset = swp_offset(entry);
	start_se = sis->curr_swap_extent;
	se = start_se;
L
Linus Torvalds 已提交
1270 1271 1272 1273 1274 1275 1276 1277

	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);
		}
1278
		lh = se->list.next;
L
Linus Torvalds 已提交
1279 1280 1281 1282 1283 1284
		se = list_entry(lh, struct swap_extent, list);
		sis->curr_swap_extent = se;
		BUG_ON(se == start_se);		/* It *must* be present */
	}
}

1285 1286 1287 1288 1289 1290 1291 1292 1293 1294
/*
 * Returns the page offset into bdev for the specified page's swap entry.
 */
sector_t map_swap_page(struct page *page, struct block_device **bdev)
{
	swp_entry_t entry;
	entry.val = page_private(page);
	return map_swap_entry(entry, bdev);
}

L
Linus Torvalds 已提交
1295 1296 1297 1298 1299
/*
 * Free all of a swapdev's extent information
 */
static void destroy_swap_extents(struct swap_info_struct *sis)
{
1300
	while (!list_empty(&sis->first_swap_extent.list)) {
L
Linus Torvalds 已提交
1301 1302
		struct swap_extent *se;

1303
		se = list_entry(sis->first_swap_extent.list.next,
L
Linus Torvalds 已提交
1304 1305 1306 1307 1308 1309 1310 1311
				struct swap_extent, list);
		list_del(&se->list);
		kfree(se);
	}
}

/*
 * Add a block range (and the corresponding page range) into this swapdev's
1312
 * extent list.  The extent list is kept sorted in page order.
L
Linus Torvalds 已提交
1313
 *
1314
 * This function rather assumes that it is called in ascending page order.
L
Linus Torvalds 已提交
1315 1316 1317 1318 1319 1320 1321 1322 1323
 */
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;

1324 1325 1326 1327 1328 1329 1330 1331 1332
	if (start_page == 0) {
		se = &sis->first_swap_extent;
		sis->curr_swap_extent = se;
		se->start_page = 0;
		se->nr_pages = nr_pages;
		se->start_block = start_block;
		return 1;
	} else {
		lh = sis->first_swap_extent.list.prev;	/* Highest extent */
L
Linus Torvalds 已提交
1333
		se = list_entry(lh, struct swap_extent, list);
1334 1335
		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
1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351
			/* 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;

1352
	list_add_tail(&new_se->list, &sis->first_swap_extent.list);
1353
	return 1;
L
Linus Torvalds 已提交
1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375
}

/*
 * 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.
 *
1376
 * For S_ISREG swapfiles we set S_SWAPFILE across the life of the swapon.  This
L
Linus Torvalds 已提交
1377 1378 1379 1380 1381 1382 1383 1384 1385 1386
 * 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.
 */
1387
static int setup_swap_extents(struct swap_info_struct *sis, sector_t *span)
L
Linus Torvalds 已提交
1388 1389 1390 1391 1392 1393 1394
{
	struct inode *inode;
	unsigned blocks_per_page;
	unsigned long page_no;
	unsigned blkbits;
	sector_t probe_block;
	sector_t last_block;
1395 1396 1397
	sector_t lowest_block = -1;
	sector_t highest_block = 0;
	int nr_extents = 0;
L
Linus Torvalds 已提交
1398 1399 1400 1401 1402
	int ret;

	inode = sis->swap_file->f_mapping->host;
	if (S_ISBLK(inode->i_mode)) {
		ret = add_swap_extent(sis, 0, sis->max, 0);
1403
		*span = sis->pages;
1404
		goto out;
L
Linus Torvalds 已提交
1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447
	}

	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;
			}
		}

1448 1449 1450 1451 1452 1453 1454 1455
		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
1456 1457 1458
		/*
		 * We found a PAGE_SIZE-length, PAGE_SIZE-aligned run of blocks
		 */
1459 1460
		ret = add_swap_extent(sis, page_no, 1, first_block);
		if (ret < 0)
L
Linus Torvalds 已提交
1461
			goto out;
1462
		nr_extents += ret;
L
Linus Torvalds 已提交
1463 1464 1465 1466 1467
		page_no++;
		probe_block += blocks_per_page;
reprobe:
		continue;
	}
1468 1469
	ret = nr_extents;
	*span = 1 + highest_block - lowest_block;
L
Linus Torvalds 已提交
1470
	if (page_no == 0)
1471
		page_no = 1;	/* force Empty message */
L
Linus Torvalds 已提交
1472
	sis->max = page_no;
1473
	sis->pages = page_no - 1;
L
Linus Torvalds 已提交
1474
	sis->highest_bit = page_no - 1;
1475 1476
out:
	return ret;
L
Linus Torvalds 已提交
1477 1478 1479
bad_bmap:
	printk(KERN_ERR "swapon: swapfile has holes\n");
	ret = -EINVAL;
1480
	goto out;
L
Linus Torvalds 已提交
1481 1482
}

1483
SYSCALL_DEFINE1(swapoff, const char __user *, specialfile)
L
Linus Torvalds 已提交
1484
{
1485
	struct swap_info_struct *p = NULL;
1486
	unsigned char *swap_map;
L
Linus Torvalds 已提交
1487 1488 1489
	struct file *swap_file, *victim;
	struct address_space *mapping;
	struct inode *inode;
1490
	char *pathname;
L
Linus Torvalds 已提交
1491 1492
	int i, type, prev;
	int err;
1493

L
Linus Torvalds 已提交
1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509
	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;
1510
	spin_lock(&swap_lock);
1511 1512
	for (type = swap_list.head; type >= 0; type = swap_info[type]->next) {
		p = swap_info[type];
H
Hugh Dickins 已提交
1513
		if (p->flags & SWP_WRITEOK) {
L
Linus Torvalds 已提交
1514 1515 1516 1517 1518 1519 1520
			if (p->swap_file->f_mapping == mapping)
				break;
		}
		prev = type;
	}
	if (type < 0) {
		err = -EINVAL;
1521
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
1522 1523 1524 1525 1526 1527
		goto out_dput;
	}
	if (!security_vm_enough_memory(p->pages))
		vm_unacct_memory(p->pages);
	else {
		err = -ENOMEM;
1528
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
1529 1530
		goto out_dput;
	}
1531
	if (prev < 0)
L
Linus Torvalds 已提交
1532
		swap_list.head = p->next;
1533 1534
	else
		swap_info[prev]->next = p->next;
L
Linus Torvalds 已提交
1535 1536 1537 1538
	if (type == swap_list.next) {
		/* just pick something that's safe... */
		swap_list.next = swap_list.head;
	}
1539
	if (p->prio < 0) {
1540 1541
		for (i = p->next; i >= 0; i = swap_info[i]->next)
			swap_info[i]->prio = p->prio--;
1542 1543
		least_priority++;
	}
L
Linus Torvalds 已提交
1544 1545 1546
	nr_swap_pages -= p->pages;
	total_swap_pages -= p->pages;
	p->flags &= ~SWP_WRITEOK;
1547
	spin_unlock(&swap_lock);
1548

H
Hugh Dickins 已提交
1549
	current->flags |= PF_OOM_ORIGIN;
L
Linus Torvalds 已提交
1550
	err = try_to_unuse(type);
H
Hugh Dickins 已提交
1551
	current->flags &= ~PF_OOM_ORIGIN;
L
Linus Torvalds 已提交
1552 1553 1554

	if (err) {
		/* re-insert swap space back into swap_list */
1555
		spin_lock(&swap_lock);
1556 1557 1558
		if (p->prio < 0)
			p->prio = --least_priority;
		prev = -1;
1559 1560
		for (i = swap_list.head; i >= 0; i = swap_info[i]->next) {
			if (p->prio >= swap_info[i]->prio)
L
Linus Torvalds 已提交
1561
				break;
1562 1563
			prev = i;
		}
L
Linus Torvalds 已提交
1564 1565
		p->next = i;
		if (prev < 0)
1566
			swap_list.head = swap_list.next = type;
L
Linus Torvalds 已提交
1567
		else
1568
			swap_info[prev]->next = type;
L
Linus Torvalds 已提交
1569 1570 1571
		nr_swap_pages += p->pages;
		total_swap_pages += p->pages;
		p->flags |= SWP_WRITEOK;
1572
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
1573 1574
		goto out_dput;
	}
1575 1576 1577 1578 1579

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

1580
	destroy_swap_extents(p);
H
Hugh Dickins 已提交
1581 1582 1583
	if (p->flags & SWP_CONTINUED)
		free_swap_count_continuations(p);

I
Ingo Molnar 已提交
1584
	mutex_lock(&swapon_mutex);
1585 1586 1587
	spin_lock(&swap_lock);
	drain_mmlist();

1588 1589 1590
	/* wait for anyone still in scan_swap_map */
	p->highest_bit = 0;		/* cuts scans short */
	while (p->flags >= SWP_SCANNING) {
1591
		spin_unlock(&swap_lock);
1592
		schedule_timeout_uninterruptible(1);
1593
		spin_lock(&swap_lock);
1594 1595
	}

L
Linus Torvalds 已提交
1596 1597 1598 1599 1600 1601
	swap_file = p->swap_file;
	p->swap_file = NULL;
	p->max = 0;
	swap_map = p->swap_map;
	p->swap_map = NULL;
	p->flags = 0;
1602
	spin_unlock(&swap_lock);
I
Ingo Molnar 已提交
1603
	mutex_unlock(&swapon_mutex);
L
Linus Torvalds 已提交
1604
	vfree(swap_map);
1605 1606 1607
	/* Destroy swap account informatin */
	swap_cgroup_swapoff(type);

L
Linus Torvalds 已提交
1608 1609 1610 1611 1612 1613
	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 {
1614
		mutex_lock(&inode->i_mutex);
L
Linus Torvalds 已提交
1615
		inode->i_flags &= ~S_SWAPFILE;
1616
		mutex_unlock(&inode->i_mutex);
L
Linus Torvalds 已提交
1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630
	}
	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)
{
1631 1632
	struct swap_info_struct *si;
	int type;
L
Linus Torvalds 已提交
1633 1634
	loff_t l = *pos;

I
Ingo Molnar 已提交
1635
	mutex_lock(&swapon_mutex);
L
Linus Torvalds 已提交
1636

1637 1638 1639
	if (!l)
		return SEQ_START_TOKEN;

1640 1641 1642 1643
	for (type = 0; type < nr_swapfiles; type++) {
		smp_rmb();	/* read nr_swapfiles before swap_info[type] */
		si = swap_info[type];
		if (!(si->flags & SWP_USED) || !si->swap_map)
L
Linus Torvalds 已提交
1644
			continue;
1645
		if (!--l)
1646
			return si;
L
Linus Torvalds 已提交
1647 1648 1649 1650 1651 1652 1653
	}

	return NULL;
}

static void *swap_next(struct seq_file *swap, void *v, loff_t *pos)
{
1654 1655
	struct swap_info_struct *si = v;
	int type;
L
Linus Torvalds 已提交
1656

1657
	if (v == SEQ_START_TOKEN)
1658 1659 1660
		type = 0;
	else
		type = si->type + 1;
1661

1662 1663 1664 1665
	for (; type < nr_swapfiles; type++) {
		smp_rmb();	/* read nr_swapfiles before swap_info[type] */
		si = swap_info[type];
		if (!(si->flags & SWP_USED) || !si->swap_map)
L
Linus Torvalds 已提交
1666 1667
			continue;
		++*pos;
1668
		return si;
L
Linus Torvalds 已提交
1669 1670 1671 1672 1673 1674 1675
	}

	return NULL;
}

static void swap_stop(struct seq_file *swap, void *v)
{
I
Ingo Molnar 已提交
1676
	mutex_unlock(&swapon_mutex);
L
Linus Torvalds 已提交
1677 1678 1679 1680
}

static int swap_show(struct seq_file *swap, void *v)
{
1681
	struct swap_info_struct *si = v;
L
Linus Torvalds 已提交
1682 1683 1684
	struct file *file;
	int len;

1685
	if (si == SEQ_START_TOKEN) {
1686 1687 1688
		seq_puts(swap,"Filename\t\t\t\tType\t\tSize\tUsed\tPriority\n");
		return 0;
	}
L
Linus Torvalds 已提交
1689

1690
	file = si->swap_file;
1691
	len = seq_path(swap, &file->f_path, " \t\n\\");
1692
	seq_printf(swap, "%*s%s\t%u\t%u\t%d\n",
1693 1694
			len < 40 ? 40 - len : 1, " ",
			S_ISBLK(file->f_path.dentry->d_inode->i_mode) ?
L
Linus Torvalds 已提交
1695
				"partition" : "file\t",
1696 1697 1698
			si->pages << (PAGE_SHIFT - 10),
			si->inuse_pages << (PAGE_SHIFT - 10),
			si->prio);
L
Linus Torvalds 已提交
1699 1700 1701
	return 0;
}

1702
static const struct seq_operations swaps_op = {
L
Linus Torvalds 已提交
1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713
	.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);
}

1714
static const struct file_operations proc_swaps_operations = {
L
Linus Torvalds 已提交
1715 1716 1717 1718 1719 1720 1721 1722
	.open		= swaps_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
	.release	= seq_release,
};

static int __init procswaps_init(void)
{
1723
	proc_create("swaps", 0, NULL, &proc_swaps_operations);
L
Linus Torvalds 已提交
1724 1725 1726 1727 1728
	return 0;
}
__initcall(procswaps_init);
#endif /* CONFIG_PROC_FS */

J
Jan Beulich 已提交
1729 1730 1731 1732 1733 1734 1735 1736 1737
#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
1738 1739 1740 1741 1742
/*
 * Written 01/25/92 by Simmule Turner, heavily changed by Linus.
 *
 * The swapon system call
 */
1743
SYSCALL_DEFINE2(swapon, const char __user *, specialfile, int, swap_flags)
L
Linus Torvalds 已提交
1744
{
1745
	struct swap_info_struct *p;
L
Linus Torvalds 已提交
1746 1747 1748 1749 1750 1751 1752 1753
	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;
1754 1755
	unsigned int nr_good_pages = 0;
	int nr_extents = 0;
1756
	sector_t span;
L
Linus Torvalds 已提交
1757
	unsigned long maxpages = 1;
1758
	unsigned long swapfilepages;
1759
	unsigned char *swap_map = NULL;
L
Linus Torvalds 已提交
1760 1761 1762 1763 1764 1765
	struct page *page = NULL;
	struct inode *inode = NULL;
	int did_down = 0;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
1766 1767 1768 1769 1770

	p = kzalloc(sizeof(*p), GFP_KERNEL);
	if (!p)
		return -ENOMEM;

1771
	spin_lock(&swap_lock);
1772 1773
	for (type = 0; type < nr_swapfiles; type++) {
		if (!(swap_info[type]->flags & SWP_USED))
L
Linus Torvalds 已提交
1774
			break;
1775
	}
L
Linus Torvalds 已提交
1776
	error = -EPERM;
1777
	if (type >= MAX_SWAPFILES) {
1778
		spin_unlock(&swap_lock);
1779
		kfree(p);
L
Linus Torvalds 已提交
1780 1781
		goto out;
	}
1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799
	if (type >= nr_swapfiles) {
		p->type = type;
		swap_info[type] = p;
		/*
		 * Write swap_info[type] before nr_swapfiles, in case a
		 * racing procfs swap_start() or swap_next() is reading them.
		 * (We never shrink nr_swapfiles, we never free this entry.)
		 */
		smp_wmb();
		nr_swapfiles++;
	} else {
		kfree(p);
		p = swap_info[type];
		/*
		 * Do not memset this entry: a racing procfs swap_next()
		 * would be relying on p->type to remain valid.
		 */
	}
1800
	INIT_LIST_HEAD(&p->first_swap_extent.list);
L
Linus Torvalds 已提交
1801 1802
	p->flags = SWP_USED;
	p->next = -1;
1803
	spin_unlock(&swap_lock);
1804

L
Linus Torvalds 已提交
1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823
	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++) {
1824
		struct swap_info_struct *q = swap_info[i];
L
Linus Torvalds 已提交
1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837

		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
1838
			error = -EINVAL;
L
Linus Torvalds 已提交
1839 1840 1841 1842 1843 1844 1845 1846 1847
			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;
1848
		mutex_lock(&inode->i_mutex);
L
Linus Torvalds 已提交
1849 1850 1851 1852 1853 1854 1855 1856 1857
		did_down = 1;
		if (IS_SWAPFILE(inode)) {
			error = -EBUSY;
			goto bad_swap;
		}
	} else {
		goto bad_swap;
	}

1858
	swapfilepages = i_size_read(inode) >> PAGE_SHIFT;
L
Linus Torvalds 已提交
1859 1860 1861 1862 1863 1864 1865 1866

	/*
	 * Read the swap header.
	 */
	if (!mapping->a_ops->readpage) {
		error = -EINVAL;
		goto bad_swap;
	}
1867
	page = read_mapping_page(mapping, 0, swap_file);
L
Linus Torvalds 已提交
1868 1869 1870 1871
	if (IS_ERR(page)) {
		error = PTR_ERR(page);
		goto bad_swap;
	}
1872
	swap_header = kmap(page);
L
Linus Torvalds 已提交
1873

1874
	if (memcmp("SWAPSPACE2", swap_header->magic.magic, 10)) {
1875
		printk(KERN_ERR "Unable to find swap-space signature\n");
L
Linus Torvalds 已提交
1876 1877 1878
		error = -EINVAL;
		goto bad_swap;
	}
1879

1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892
	/* 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
1893 1894
		error = -EINVAL;
		goto bad_swap;
1895
	}
L
Linus Torvalds 已提交
1896

1897 1898
	p->lowest_bit  = 1;
	p->cluster_next = 1;
1899
	p->cluster_nr = 0;
1900

1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919
	/*
	 * 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
1920

1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932
	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;
1933

1934
	/* OK, set up the swap map and apply the bad block list */
1935
	swap_map = vmalloc(maxpages);
1936 1937 1938 1939
	if (!swap_map) {
		error = -ENOMEM;
		goto bad_swap;
	}
L
Linus Torvalds 已提交
1940

1941
	memset(swap_map, 0, maxpages);
1942 1943 1944 1945
	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
1946
			goto bad_swap;
1947 1948
		}
		swap_map[page_nr] = SWAP_MAP_BAD;
L
Linus Torvalds 已提交
1949
	}
1950 1951 1952 1953 1954

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

1955 1956 1957
	nr_good_pages = swap_header->info.last_page -
			swap_header->info.nr_badpages -
			1 /* header page */;
1958 1959

	if (nr_good_pages) {
1960
		swap_map[0] = SWAP_MAP_BAD;
1961 1962
		p->max = maxpages;
		p->pages = nr_good_pages;
1963 1964 1965
		nr_extents = setup_swap_extents(p, &span);
		if (nr_extents < 0) {
			error = nr_extents;
1966
			goto bad_swap;
1967
		}
1968 1969
		nr_good_pages = p->pages;
	}
L
Linus Torvalds 已提交
1970 1971 1972 1973 1974 1975
	if (!nr_good_pages) {
		printk(KERN_WARNING "Empty swap-file\n");
		error = -EINVAL;
		goto bad_swap;
	}

1976 1977 1978 1979 1980 1981 1982
	if (p->bdev) {
		if (blk_queue_nonrot(bdev_get_queue(p->bdev))) {
			p->flags |= SWP_SOLIDSTATE;
			p->cluster_next = 1 + (random32() % p->highest_bit);
		}
		if (discard_swap(p) == 0)
			p->flags |= SWP_DISCARDABLE;
1983
	}
1984

I
Ingo Molnar 已提交
1985
	mutex_lock(&swapon_mutex);
1986
	spin_lock(&swap_lock);
1987 1988 1989 1990 1991 1992
	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
1993
	p->flags |= SWP_WRITEOK;
L
Linus Torvalds 已提交
1994 1995
	nr_swap_pages += nr_good_pages;
	total_swap_pages += nr_good_pages;
1996

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

	/* insert swap space into swap_list: */
	prev = -1;
2006 2007
	for (i = swap_list.head; i >= 0; i = swap_info[i]->next) {
		if (p->prio >= swap_info[i]->prio)
L
Linus Torvalds 已提交
2008 2009 2010 2011
			break;
		prev = i;
	}
	p->next = i;
2012 2013 2014 2015
	if (prev < 0)
		swap_list.head = swap_list.next = type;
	else
		swap_info[prev]->next = type;
2016
	spin_unlock(&swap_lock);
I
Ingo Molnar 已提交
2017
	mutex_unlock(&swapon_mutex);
L
Linus Torvalds 已提交
2018 2019 2020 2021 2022 2023 2024
	error = 0;
	goto out;
bad_swap:
	if (bdev) {
		set_blocksize(bdev, p->old_block_size);
		bd_release(bdev);
	}
2025
	destroy_swap_extents(p);
2026
	swap_cgroup_swapoff(type);
L
Linus Torvalds 已提交
2027
bad_swap_2:
2028
	spin_lock(&swap_lock);
L
Linus Torvalds 已提交
2029 2030
	p->swap_file = NULL;
	p->flags = 0;
2031
	spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044
	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;
2045
		mutex_unlock(&inode->i_mutex);
L
Linus Torvalds 已提交
2046 2047 2048 2049 2050 2051
	}
	return error;
}

void si_swapinfo(struct sysinfo *val)
{
2052
	unsigned int type;
L
Linus Torvalds 已提交
2053 2054
	unsigned long nr_to_be_unused = 0;

2055
	spin_lock(&swap_lock);
2056 2057 2058 2059 2060
	for (type = 0; type < nr_swapfiles; type++) {
		struct swap_info_struct *si = swap_info[type];

		if ((si->flags & SWP_USED) && !(si->flags & SWP_WRITEOK))
			nr_to_be_unused += si->inuse_pages;
L
Linus Torvalds 已提交
2061 2062 2063
	}
	val->freeswap = nr_swap_pages + nr_to_be_unused;
	val->totalswap = total_swap_pages + nr_to_be_unused;
2064
	spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
2065 2066 2067 2068 2069
}

/*
 * Verify that a swap entry is valid and increment its swap map count.
 *
2070 2071 2072 2073 2074 2075
 * 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
H
Hugh Dickins 已提交
2076
 * - swap-mapped reference requested but needs continued swap count. -> ENOMEM
L
Linus Torvalds 已提交
2077
 */
2078
static int __swap_duplicate(swp_entry_t entry, unsigned char usage)
L
Linus Torvalds 已提交
2079
{
2080
	struct swap_info_struct *p;
L
Linus Torvalds 已提交
2081
	unsigned long offset, type;
2082 2083
	unsigned char count;
	unsigned char has_cache;
H
Hugh Dickins 已提交
2084
	int err = -EINVAL;
L
Linus Torvalds 已提交
2085

2086
	if (non_swap_entry(entry))
H
Hugh Dickins 已提交
2087
		goto out;
2088

L
Linus Torvalds 已提交
2089 2090 2091
	type = swp_type(entry);
	if (type >= nr_swapfiles)
		goto bad_file;
2092
	p = swap_info[type];
L
Linus Torvalds 已提交
2093 2094
	offset = swp_offset(entry);

2095
	spin_lock(&swap_lock);
2096 2097 2098
	if (unlikely(offset >= p->max))
		goto unlock_out;

H
Hugh Dickins 已提交
2099 2100 2101 2102
	count = p->swap_map[offset];
	has_cache = count & SWAP_HAS_CACHE;
	count &= ~SWAP_HAS_CACHE;
	err = 0;
2103

H
Hugh Dickins 已提交
2104
	if (usage == SWAP_HAS_CACHE) {
2105 2106

		/* set SWAP_HAS_CACHE if there is no cache and entry is used */
H
Hugh Dickins 已提交
2107 2108 2109 2110 2111 2112
		if (!has_cache && count)
			has_cache = SWAP_HAS_CACHE;
		else if (has_cache)		/* someone else added cache */
			err = -EEXIST;
		else				/* no users remaining */
			err = -ENOENT;
2113 2114

	} else if (count || has_cache) {
H
Hugh Dickins 已提交
2115

H
Hugh Dickins 已提交
2116 2117 2118
		if ((count & ~COUNT_CONTINUED) < SWAP_MAP_MAX)
			count += usage;
		else if ((count & ~COUNT_CONTINUED) > SWAP_MAP_MAX)
H
Hugh Dickins 已提交
2119
			err = -EINVAL;
H
Hugh Dickins 已提交
2120 2121 2122 2123
		else if (swap_count_continued(p, offset, count))
			count = COUNT_CONTINUED;
		else
			err = -ENOMEM;
2124
	} else
H
Hugh Dickins 已提交
2125 2126 2127 2128
		err = -ENOENT;			/* unused swap entry */

	p->swap_map[offset] = count | has_cache;

2129
unlock_out:
2130
	spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
2131
out:
H
Hugh Dickins 已提交
2132
	return err;
L
Linus Torvalds 已提交
2133 2134 2135 2136 2137

bad_file:
	printk(KERN_ERR "swap_dup: %s%08lx\n", Bad_file, entry.val);
	goto out;
}
H
Hugh Dickins 已提交
2138

H
Hugh Dickins 已提交
2139 2140 2141 2142 2143 2144 2145 2146 2147
/*
 * Help swapoff by noting that swap entry belongs to shmem/tmpfs
 * (in which case its reference count is never incremented).
 */
void swap_shmem_alloc(swp_entry_t entry)
{
	__swap_duplicate(entry, SWAP_MAP_SHMEM);
}

2148 2149 2150
/*
 * increase reference count of swap entry by 1.
 */
H
Hugh Dickins 已提交
2151
int swap_duplicate(swp_entry_t entry)
2152
{
H
Hugh Dickins 已提交
2153 2154 2155 2156 2157
	int err = 0;

	while (!err && __swap_duplicate(entry, 1) == -ENOMEM)
		err = add_swap_count_continuation(entry, GFP_ATOMIC);
	return err;
2158
}
L
Linus Torvalds 已提交
2159

2160
/*
2161 2162
 * @entry: swap entry for which we allocate swap cache.
 *
2163
 * Called when allocating swap cache for existing swap entry,
2164 2165 2166
 * This can return error codes. Returns 0 at success.
 * -EBUSY means there is a swap cache.
 * Note: return code is different from swap_duplicate().
2167 2168 2169
 */
int swapcache_prepare(swp_entry_t entry)
{
H
Hugh Dickins 已提交
2170
	return __swap_duplicate(entry, SWAP_HAS_CACHE);
2171 2172
}

L
Linus Torvalds 已提交
2173
/*
2174
 * swap_lock prevents swap_map being freed. Don't grab an extra
L
Linus Torvalds 已提交
2175 2176 2177 2178
 * reference on the swaphandle, it doesn't matter if it becomes unused.
 */
int valid_swaphandles(swp_entry_t entry, unsigned long *offset)
{
2179
	struct swap_info_struct *si;
H
Hugh Dickins 已提交
2180
	int our_page_cluster = page_cluster;
2181 2182 2183
	pgoff_t target, toff;
	pgoff_t base, end;
	int nr_pages = 0;
L
Linus Torvalds 已提交
2184

H
Hugh Dickins 已提交
2185
	if (!our_page_cluster)	/* no readahead */
L
Linus Torvalds 已提交
2186
		return 0;
2187

2188
	si = swap_info[swp_type(entry)];
2189 2190 2191 2192 2193
	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
2194

2195
	spin_lock(&swap_lock);
2196 2197 2198 2199 2200 2201 2202 2203
	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;
2204
		if (swap_count(si->swap_map[toff]) == SWAP_MAP_BAD)
L
Linus Torvalds 已提交
2205
			break;
2206 2207 2208
	}
	/* Count contiguous allocated slots below our target */
	for (toff = target; --toff >= base; nr_pages++) {
L
Linus Torvalds 已提交
2209
		/* Don't read in free or bad pages */
2210
		if (!si->swap_map[toff])
L
Linus Torvalds 已提交
2211
			break;
2212
		if (swap_count(si->swap_map[toff]) == SWAP_MAP_BAD)
L
Linus Torvalds 已提交
2213
			break;
2214
	}
2215
	spin_unlock(&swap_lock);
2216 2217 2218 2219 2220 2221 2222

	/*
	 * 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
2223
}
H
Hugh Dickins 已提交
2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 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 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439

/*
 * add_swap_count_continuation - called when a swap count is duplicated
 * beyond SWAP_MAP_MAX, it allocates a new page and links that to the entry's
 * page of the original vmalloc'ed swap_map, to hold the continuation count
 * (for that entry and for its neighbouring PAGE_SIZE swap entries).  Called
 * again when count is duplicated beyond SWAP_MAP_MAX * SWAP_CONT_MAX, etc.
 *
 * These continuation pages are seldom referenced: the common paths all work
 * on the original swap_map, only referring to a continuation page when the
 * low "digit" of a count is incremented or decremented through SWAP_MAP_MAX.
 *
 * add_swap_count_continuation(, GFP_ATOMIC) can be called while holding
 * page table locks; if it fails, add_swap_count_continuation(, GFP_KERNEL)
 * can be called after dropping locks.
 */
int add_swap_count_continuation(swp_entry_t entry, gfp_t gfp_mask)
{
	struct swap_info_struct *si;
	struct page *head;
	struct page *page;
	struct page *list_page;
	pgoff_t offset;
	unsigned char count;

	/*
	 * When debugging, it's easier to use __GFP_ZERO here; but it's better
	 * for latency not to zero a page while GFP_ATOMIC and holding locks.
	 */
	page = alloc_page(gfp_mask | __GFP_HIGHMEM);

	si = swap_info_get(entry);
	if (!si) {
		/*
		 * An acceptable race has occurred since the failing
		 * __swap_duplicate(): the swap entry has been freed,
		 * perhaps even the whole swap_map cleared for swapoff.
		 */
		goto outer;
	}

	offset = swp_offset(entry);
	count = si->swap_map[offset] & ~SWAP_HAS_CACHE;

	if ((count & ~COUNT_CONTINUED) != SWAP_MAP_MAX) {
		/*
		 * The higher the swap count, the more likely it is that tasks
		 * will race to add swap count continuation: we need to avoid
		 * over-provisioning.
		 */
		goto out;
	}

	if (!page) {
		spin_unlock(&swap_lock);
		return -ENOMEM;
	}

	/*
	 * We are fortunate that although vmalloc_to_page uses pte_offset_map,
	 * no architecture is using highmem pages for kernel pagetables: so it
	 * will not corrupt the GFP_ATOMIC caller's atomic pagetable kmaps.
	 */
	head = vmalloc_to_page(si->swap_map + offset);
	offset &= ~PAGE_MASK;

	/*
	 * Page allocation does not initialize the page's lru field,
	 * but it does always reset its private field.
	 */
	if (!page_private(head)) {
		BUG_ON(count & COUNT_CONTINUED);
		INIT_LIST_HEAD(&head->lru);
		set_page_private(head, SWP_CONTINUED);
		si->flags |= SWP_CONTINUED;
	}

	list_for_each_entry(list_page, &head->lru, lru) {
		unsigned char *map;

		/*
		 * If the previous map said no continuation, but we've found
		 * a continuation page, free our allocation and use this one.
		 */
		if (!(count & COUNT_CONTINUED))
			goto out;

		map = kmap_atomic(list_page, KM_USER0) + offset;
		count = *map;
		kunmap_atomic(map, KM_USER0);

		/*
		 * If this continuation count now has some space in it,
		 * free our allocation and use this one.
		 */
		if ((count & ~COUNT_CONTINUED) != SWAP_CONT_MAX)
			goto out;
	}

	list_add_tail(&page->lru, &head->lru);
	page = NULL;			/* now it's attached, don't free it */
out:
	spin_unlock(&swap_lock);
outer:
	if (page)
		__free_page(page);
	return 0;
}

/*
 * swap_count_continued - when the original swap_map count is incremented
 * from SWAP_MAP_MAX, check if there is already a continuation page to carry
 * into, carry if so, or else fail until a new continuation page is allocated;
 * when the original swap_map count is decremented from 0 with continuation,
 * borrow from the continuation and report whether it still holds more.
 * Called while __swap_duplicate() or swap_entry_free() holds swap_lock.
 */
static bool swap_count_continued(struct swap_info_struct *si,
				 pgoff_t offset, unsigned char count)
{
	struct page *head;
	struct page *page;
	unsigned char *map;

	head = vmalloc_to_page(si->swap_map + offset);
	if (page_private(head) != SWP_CONTINUED) {
		BUG_ON(count & COUNT_CONTINUED);
		return false;		/* need to add count continuation */
	}

	offset &= ~PAGE_MASK;
	page = list_entry(head->lru.next, struct page, lru);
	map = kmap_atomic(page, KM_USER0) + offset;

	if (count == SWAP_MAP_MAX)	/* initial increment from swap_map */
		goto init_map;		/* jump over SWAP_CONT_MAX checks */

	if (count == (SWAP_MAP_MAX | COUNT_CONTINUED)) { /* incrementing */
		/*
		 * Think of how you add 1 to 999
		 */
		while (*map == (SWAP_CONT_MAX | COUNT_CONTINUED)) {
			kunmap_atomic(map, KM_USER0);
			page = list_entry(page->lru.next, struct page, lru);
			BUG_ON(page == head);
			map = kmap_atomic(page, KM_USER0) + offset;
		}
		if (*map == SWAP_CONT_MAX) {
			kunmap_atomic(map, KM_USER0);
			page = list_entry(page->lru.next, struct page, lru);
			if (page == head)
				return false;	/* add count continuation */
			map = kmap_atomic(page, KM_USER0) + offset;
init_map:		*map = 0;		/* we didn't zero the page */
		}
		*map += 1;
		kunmap_atomic(map, KM_USER0);
		page = list_entry(page->lru.prev, struct page, lru);
		while (page != head) {
			map = kmap_atomic(page, KM_USER0) + offset;
			*map = COUNT_CONTINUED;
			kunmap_atomic(map, KM_USER0);
			page = list_entry(page->lru.prev, struct page, lru);
		}
		return true;			/* incremented */

	} else {				/* decrementing */
		/*
		 * Think of how you subtract 1 from 1000
		 */
		BUG_ON(count != COUNT_CONTINUED);
		while (*map == COUNT_CONTINUED) {
			kunmap_atomic(map, KM_USER0);
			page = list_entry(page->lru.next, struct page, lru);
			BUG_ON(page == head);
			map = kmap_atomic(page, KM_USER0) + offset;
		}
		BUG_ON(*map == 0);
		*map -= 1;
		if (*map == 0)
			count = 0;
		kunmap_atomic(map, KM_USER0);
		page = list_entry(page->lru.prev, struct page, lru);
		while (page != head) {
			map = kmap_atomic(page, KM_USER0) + offset;
			*map = SWAP_CONT_MAX | count;
			count = COUNT_CONTINUED;
			kunmap_atomic(map, KM_USER0);
			page = list_entry(page->lru.prev, struct page, lru);
		}
		return count == COUNT_CONTINUED;
	}
}

/*
 * free_swap_count_continuations - swapoff free all the continuation pages
 * appended to the swap_map, after swap_map is quiesced, before vfree'ing it.
 */
static void free_swap_count_continuations(struct swap_info_struct *si)
{
	pgoff_t offset;

	for (offset = 0; offset < si->max; offset += PAGE_SIZE) {
		struct page *head;
		head = vmalloc_to_page(si->swap_map + offset);
		if (page_private(head)) {
			struct list_head *this, *next;
			list_for_each_safe(this, next, &head->lru) {
				struct page *page;
				page = list_entry(this, struct page, lru);
				list_del(this);
				__free_page(page);
			}
		}
	}
}