swapfile.c 45.5 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
/*
 *  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>
#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
27
#include <linux/mutex.h>
28
#include <linux/capability.h>
L
Linus Torvalds 已提交
29
#include <linux/syscalls.h>
30
#include <linux/memcontrol.h>
L
Linus Torvalds 已提交
31 32 33 34 35

#include <asm/pgtable.h>
#include <asm/tlbflush.h>
#include <linux/swapops.h>

A
Adrian Bunk 已提交
36 37
static DEFINE_SPINLOCK(swap_lock);
static unsigned int nr_swapfiles;
38
long nr_swap_pages;
L
Linus Torvalds 已提交
39 40
long total_swap_pages;
static int swap_overflow;
41
static int least_priority;
L
Linus Torvalds 已提交
42 43 44 45 46 47

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
48
static struct swap_list_t swap_list = {-1, -1};
L
Linus Torvalds 已提交
49

50
static struct swap_info_struct swap_info[MAX_SWAPFILES];
L
Linus Torvalds 已提交
51

I
Ingo Molnar 已提交
52
static DEFINE_MUTEX(swapon_mutex);
L
Linus Torvalds 已提交
53 54 55

/*
 * We need this because the bdev->unplug_fn can sleep and we cannot
56
 * hold swap_lock while calling the unplug_fn. And swap_lock
I
Ingo Molnar 已提交
57
 * cannot be turned into a mutex.
L
Linus Torvalds 已提交
58 59 60 61 62 63 64 65
 */
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
66
	entry.val = page_private(page);
L
Linus Torvalds 已提交
67 68 69 70 71 72 73
	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
74 75
		 * 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
76 77 78 79 80 81
		 * 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
82
		blk_run_backing_dev(bdi, page);
L
Linus Torvalds 已提交
83 84 85 86
	}
	up_read(&swap_unplug_sem);
}

87 88 89
#define SWAPFILE_CLUSTER	256
#define LATENCY_LIMIT		256

90
static inline unsigned long scan_swap_map(struct swap_info_struct *si)
L
Linus Torvalds 已提交
91
{
92
	unsigned long offset, last_in_cluster;
93
	int latency_ration = LATENCY_LIMIT;
94

L
Linus Torvalds 已提交
95
	/* 
96 97 98 99 100 101 102 103 104
	 * 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
	 */

105
	si->flags += SWP_SCANNING;
106 107 108 109
	if (unlikely(!si->cluster_nr)) {
		si->cluster_nr = SWAPFILE_CLUSTER - 1;
		if (si->pages - si->inuse_pages < SWAPFILE_CLUSTER)
			goto lowest;
110
		spin_unlock(&swap_lock);
111 112 113 114 115 116

		offset = si->lowest_bit;
		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
117
			if (si->swap_map[offset])
118 119
				last_in_cluster = offset + SWAPFILE_CLUSTER;
			else if (offset == last_in_cluster) {
120
				spin_lock(&swap_lock);
A
Akinobu Mita 已提交
121
				si->cluster_next = offset-SWAPFILE_CLUSTER+1;
122
				goto cluster;
L
Linus Torvalds 已提交
123
			}
124 125 126 127
			if (unlikely(--latency_ration < 0)) {
				cond_resched();
				latency_ration = LATENCY_LIMIT;
			}
128
		}
129
		spin_lock(&swap_lock);
130
		goto lowest;
L
Linus Torvalds 已提交
131
	}
132 133 134 135 136 137

	si->cluster_nr--;
cluster:
	offset = si->cluster_next;
	if (offset > si->highest_bit)
lowest:		offset = si->lowest_bit;
138 139
checks:	if (!(si->flags & SWP_WRITEOK))
		goto no_page;
140 141 142
	if (!si->highest_bit)
		goto no_page;
	if (!si->swap_map[offset]) {
143
		if (offset == si->lowest_bit)
L
Linus Torvalds 已提交
144 145 146
			si->lowest_bit++;
		if (offset == si->highest_bit)
			si->highest_bit--;
147 148
		si->inuse_pages++;
		if (si->inuse_pages == si->pages) {
L
Linus Torvalds 已提交
149 150 151 152
			si->lowest_bit = si->max;
			si->highest_bit = 0;
		}
		si->swap_map[offset] = 1;
153
		si->cluster_next = offset + 1;
154
		si->flags -= SWP_SCANNING;
L
Linus Torvalds 已提交
155 156
		return offset;
	}
157

158
	spin_unlock(&swap_lock);
159
	while (++offset <= si->highest_bit) {
160
		if (!si->swap_map[offset]) {
161
			spin_lock(&swap_lock);
162 163
			goto checks;
		}
164 165 166 167
		if (unlikely(--latency_ration < 0)) {
			cond_resched();
			latency_ration = LATENCY_LIMIT;
		}
168
	}
169
	spin_lock(&swap_lock);
170 171 172
	goto lowest;

no_page:
173
	si->flags -= SWP_SCANNING;
L
Linus Torvalds 已提交
174 175 176 177 178
	return 0;
}

swp_entry_t get_swap_page(void)
{
179 180 181 182
	struct swap_info_struct *si;
	pgoff_t offset;
	int type, next;
	int wrapped = 0;
L
Linus Torvalds 已提交
183

184
	spin_lock(&swap_lock);
L
Linus Torvalds 已提交
185
	if (nr_swap_pages <= 0)
186 187 188 189 190 191 192 193 194 195
		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
196
		}
197 198 199 200 201 202 203 204

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

		swap_list.next = next;
		offset = scan_swap_map(si);
205 206
		if (offset) {
			spin_unlock(&swap_lock);
207
			return swp_entry(type, offset);
208
		}
209
		next = swap_list.next;
L
Linus Torvalds 已提交
210
	}
211 212 213

	nr_swap_pages++;
noswap:
214
	spin_unlock(&swap_lock);
215
	return (swp_entry_t) {0};
L
Linus Torvalds 已提交
216 217
}

218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237
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--;
		offset = scan_swap_map(si);
		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
238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255
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;
256
	spin_lock(&swap_lock);
L
Linus Torvalds 已提交
257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285
	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;
}	

static int swap_entry_free(struct swap_info_struct *p, unsigned long offset)
{
	int count = p->swap_map[offset];

	if (count < SWAP_MAP_MAX) {
		count--;
		p->swap_map[offset] = count;
		if (!count) {
			if (offset < p->lowest_bit)
				p->lowest_bit = offset;
			if (offset > p->highest_bit)
				p->highest_bit = offset;
286 287
			if (p->prio > swap_info[swap_list.next].prio)
				swap_list.next = p - swap_info;
L
Linus Torvalds 已提交
288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305
			nr_swap_pages++;
			p->inuse_pages--;
		}
	}
	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) {
		swap_entry_free(p, swp_offset(entry));
306
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
307 308 309 310
	}
}

/*
311
 * How many references to page are currently swapped out?
L
Linus Torvalds 已提交
312
 */
313
static inline int page_swapcount(struct page *page)
L
Linus Torvalds 已提交
314
{
315 316
	int count = 0;
	struct swap_info_struct *p;
L
Linus Torvalds 已提交
317 318
	swp_entry_t entry;

H
Hugh Dickins 已提交
319
	entry.val = page_private(page);
L
Linus Torvalds 已提交
320 321
	p = swap_info_get(entry);
	if (p) {
322 323
		/* Subtract the 1 for the swap cache itself */
		count = p->swap_map[swp_offset(entry)] - 1;
324
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
325
	}
326
	return count;
L
Linus Torvalds 已提交
327 328 329
}

/*
330 331 332 333
 * 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
334
 */
335
int reuse_swap_page(struct page *page)
L
Linus Torvalds 已提交
336
{
337 338
	int count;

339
	VM_BUG_ON(!PageLocked(page));
340
	count = page_mapcount(page);
341
	if (count <= 1 && PageSwapCache(page)) {
342
		count += page_swapcount(page);
343 344 345 346 347
		if (count == 1 && !PageWriteback(page)) {
			delete_from_swap_cache(page);
			SetPageDirty(page);
		}
	}
348
	return count == 1;
L
Linus Torvalds 已提交
349 350 351
}

/*
352 353
 * 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
354
 */
355
int try_to_free_swap(struct page *page)
L
Linus Torvalds 已提交
356
{
357
	VM_BUG_ON(!PageLocked(page));
L
Linus Torvalds 已提交
358 359 360 361 362

	if (!PageSwapCache(page))
		return 0;
	if (PageWriteback(page))
		return 0;
363
	if (page_swapcount(page))
L
Linus Torvalds 已提交
364 365
		return 0;

366 367 368
	delete_from_swap_cache(page);
	SetPageDirty(page);
	return 1;
369 370
}

L
Linus Torvalds 已提交
371 372 373 374 375 376 377 378 379
/*
 * Free the swap entry like above, but also try to
 * free the page cache entry if it is the last user.
 */
void free_swap_and_cache(swp_entry_t entry)
{
	struct swap_info_struct * p;
	struct page *page = NULL;

380 381 382
	if (is_migration_entry(entry))
		return;

L
Linus Torvalds 已提交
383 384
	p = swap_info_get(entry);
	if (p) {
385 386
		if (swap_entry_free(p, swp_offset(entry)) == 1) {
			page = find_get_page(&swapper_space, entry.val);
N
Nick Piggin 已提交
387
			if (page && !trylock_page(page)) {
388 389 390 391
				page_cache_release(page);
				page = NULL;
			}
		}
392
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
393 394
	}
	if (page) {
395 396 397 398
		/*
		 * Not mapped elsewhere, or swap space full? Free it!
		 * Also recheck PageSwapCache now page is locked (above).
		 */
399
		if (PageSwapCache(page) && !PageWriteback(page) &&
400
				(!page_mapped(page) || vm_swap_full())) {
L
Linus Torvalds 已提交
401 402 403 404 405 406 407 408
			delete_from_swap_cache(page);
			SetPageDirty(page);
		}
		unlock_page(page);
		page_cache_release(page);
	}
}

409
#ifdef CONFIG_HIBERNATION
410
/*
411
 * Find the swap type that corresponds to given device (if any).
412
 *
413 414 415 416
 * @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).
417
 */
418
int swap_type_of(dev_t device, sector_t offset, struct block_device **bdev_p)
419
{
420
	struct block_device *bdev = NULL;
421 422
	int i;

423 424 425
	if (device)
		bdev = bdget(device);

426 427
	spin_lock(&swap_lock);
	for (i = 0; i < nr_swapfiles; i++) {
428
		struct swap_info_struct *sis = swap_info + i;
429

430
		if (!(sis->flags & SWP_WRITEOK))
431
			continue;
432

433
		if (!bdev) {
434 435 436
			if (bdev_p)
				*bdev_p = sis->bdev;

437 438 439
			spin_unlock(&swap_lock);
			return i;
		}
440 441 442 443 444 445
		if (bdev == sis->bdev) {
			struct swap_extent *se;

			se = list_entry(sis->extent_list.next,
					struct swap_extent, list);
			if (se->start_block == offset) {
446 447 448
				if (bdev_p)
					*bdev_p = sis->bdev;

449 450 451 452
				spin_unlock(&swap_lock);
				bdput(bdev);
				return i;
			}
453 454 455
		}
	}
	spin_unlock(&swap_lock);
456 457 458
	if (bdev)
		bdput(bdev);

459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484
	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
485
/*
486 487 488
 * 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
489
 */
H
Hugh Dickins 已提交
490
static int unuse_pte(struct vm_area_struct *vma, pmd_t *pmd,
L
Linus Torvalds 已提交
491 492
		unsigned long addr, swp_entry_t entry, struct page *page)
{
H
Hugh Dickins 已提交
493 494 495 496
	spinlock_t *ptl;
	pte_t *pte;
	int ret = 1;

497
	if (mem_cgroup_charge(page, vma->vm_mm, GFP_KERNEL))
H
Hugh Dickins 已提交
498 499 500 501 502 503 504 505 506
		ret = -ENOMEM;

	pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
	if (unlikely(!pte_same(*pte, swp_entry_to_pte(entry)))) {
		if (ret > 0)
			mem_cgroup_uncharge_page(page);
		ret = 0;
		goto out;
	}
507

508
	inc_mm_counter(vma->vm_mm, anon_rss);
L
Linus Torvalds 已提交
509 510 511 512 513 514 515 516 517 518
	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);
	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
519 520 521
out:
	pte_unmap_unlock(pte, ptl);
	return ret;
L
Linus Torvalds 已提交
522 523 524 525 526 527 528
}

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);
529
	pte_t *pte;
530
	int ret = 0;
L
Linus Torvalds 已提交
531

H
Hugh Dickins 已提交
532 533 534 535 536 537 538 539 540 541
	/*
	 * 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
542 543 544 545 546 547
	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
548 549 550 551 552
			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
553 554
		}
	} while (pte++, addr += PAGE_SIZE, addr != end);
H
Hugh Dickins 已提交
555 556
	pte_unmap(pte - 1);
out:
557
	return ret;
L
Linus Torvalds 已提交
558 559 560 561 562 563 564 565
}

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;
566
	int ret;
L
Linus Torvalds 已提交
567 568 569 570 571 572

	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
		if (pmd_none_or_clear_bad(pmd))
			continue;
573 574 575
		ret = unuse_pte_range(vma, pmd, addr, next, entry, page);
		if (ret)
			return ret;
L
Linus Torvalds 已提交
576 577 578 579 580 581 582 583 584 585
	} 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;
586
	int ret;
L
Linus Torvalds 已提交
587 588 589 590 591 592

	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
		if (pud_none_or_clear_bad(pud))
			continue;
593 594 595
		ret = unuse_pmd_range(vma, pud, addr, next, entry, page);
		if (ret)
			return ret;
L
Linus Torvalds 已提交
596 597 598 599 600 601 602 603 604
	} 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;
605
	int ret;
L
Linus Torvalds 已提交
606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622

	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;
623 624 625
		ret = unuse_pud_range(vma, pgd, addr, next, entry, page);
		if (ret)
			return ret;
L
Linus Torvalds 已提交
626 627 628 629 630 631 632 633
	} 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;
634
	int ret = 0;
L
Linus Torvalds 已提交
635 636 637

	if (!down_read_trylock(&mm->mmap_sem)) {
		/*
638 639
		 * 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
640
		 */
641
		activate_page(page);
L
Linus Torvalds 已提交
642 643 644 645 646
		unlock_page(page);
		down_read(&mm->mmap_sem);
		lock_page(page);
	}
	for (vma = mm->mmap; vma; vma = vma->vm_next) {
647
		if (vma->anon_vma && (ret = unuse_vma(vma, entry, page)))
L
Linus Torvalds 已提交
648 649 650
			break;
	}
	up_read(&mm->mmap_sem);
651
	return (ret < 0)? ret: 0;
L
Linus Torvalds 已提交
652 653 654 655 656 657
}

/*
 * Scan swap_map from current position to next entry still in use.
 * Recycle to start on reaching the end, returning 0 when empty.
 */
658 659
static unsigned int find_next_to_unuse(struct swap_info_struct *si,
					unsigned int prev)
L
Linus Torvalds 已提交
660
{
661 662
	unsigned int max = si->max;
	unsigned int i = prev;
L
Linus Torvalds 已提交
663 664 665
	int count;

	/*
666
	 * No need for swap_lock here: we're just looking
L
Linus Torvalds 已提交
667 668
	 * for whether an entry is in use, not modifying it; false
	 * hits are okay, and sys_swapoff() has already prevented new
669
	 * allocations from this area (while holding swap_lock).
L
Linus Torvalds 已提交
670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704
	 */
	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];
		if (count && count != SWAP_MAP_BAD)
			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;
705
	unsigned int i = 0;
L
Linus Torvalds 已提交
706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745
	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;
		}

		/* 
		 * Get a page for the entry, using the existing swap
		 * cache page if there is one.  Otherwise, get a clean
		 * page and read the swap into it. 
		 */
		swap_map = &si->swap_map[i];
		entry = swp_entry(type, i);
746 747
		page = read_swap_cache_async(entry,
					GFP_HIGHUSER_MOVABLE, NULL, 0);
L
Linus Torvalds 已提交
748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805
		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;
		if (swcount > 1) {
			if (start_mm == &init_mm)
				shmem = shmem_unuse(entry, page);
			else
				retval = unuse_mm(start_mm, entry, page);
		}
		if (*swap_map > 1) {
			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);
806
			while (*swap_map > 1 && !retval && !shmem &&
L
Linus Torvalds 已提交
807 808
					(p = p->next) != &start_mm->mmlist) {
				mm = list_entry(p, struct mm_struct, mmlist);
809
				if (!atomic_inc_not_zero(&mm->mm_users))
L
Linus Torvalds 已提交
810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837
					continue;
				spin_unlock(&mmlist_lock);
				mmput(prev_mm);
				prev_mm = mm;

				cond_resched();

				swcount = *swap_map;
				if (swcount <= 1)
					;
				else if (mm == &init_mm) {
					set_start_mm = 1;
					shmem = shmem_unuse(entry, page);
				} else
					retval = unuse_mm(mm, entry, page);
				if (set_start_mm && *swap_map < swcount) {
					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;
		}
838 839 840 841 842 843 844
		if (shmem) {
			/* page has already been unlocked and released */
			if (shmem > 0)
				continue;
			retval = shmem;
			break;
		}
L
Linus Torvalds 已提交
845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864
		if (retval) {
			unlock_page(page);
			page_cache_release(page);
			break;
		}

		/*
		 * How could swap count reach 0x7fff when the maximum
		 * pid is 0x7fff, and 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 in Linux 2.4.
		 *
		 * 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.
		 */
		if (*swap_map == SWAP_MAP_MAX) {
865
			spin_lock(&swap_lock);
L
Linus Torvalds 已提交
866
			*swap_map = 1;
867
			spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892
			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.
		 */
		if ((*swap_map > 1) && PageDirty(page) && PageSwapCache(page)) {
			struct writeback_control wbc = {
				.sync_mode = WB_SYNC_NONE,
			};

			swap_writepage(page, &wbc);
			lock_page(page);
			wait_on_page_writeback(page);
		}
893 894 895 896 897 898 899 900 901 902

		/*
		 * 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))
903
			delete_from_swap_cache(page);
L
Linus Torvalds 已提交
904 905 906 907

		/*
		 * So we could skip searching mms once swap count went
		 * to 1, we did not mark any present ptes as dirty: must
908
		 * mark page dirty so shrink_page_list will preserve it.
L
Linus Torvalds 已提交
909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929
		 */
		SetPageDirty(page);
		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;
}

/*
930 931 932
 * 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
933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964
 * 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);
		}
965
		lh = se->list.next;
L
Linus Torvalds 已提交
966
		if (lh == &sis->extent_list)
967
			lh = lh->next;
L
Linus Torvalds 已提交
968 969 970 971 972 973
		se = list_entry(lh, struct swap_extent, list);
		sis->curr_swap_extent = se;
		BUG_ON(se == start_se);		/* It *must* be present */
	}
}

974
#ifdef CONFIG_HIBERNATION
975 976 977 978 979 980 981 982 983 984 985 986 987 988
/*
 * 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;
}
989
#endif /* CONFIG_HIBERNATION */
990

L
Linus Torvalds 已提交
991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007
/*
 * 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
1008
 * extent list.  The extent list is kept sorted in page order.
L
Linus Torvalds 已提交
1009
 *
1010
 * This function rather assumes that it is called in ascending page order.
L
Linus Torvalds 已提交
1011 1012 1013 1014 1015 1016 1017 1018 1019
 */
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;

1020 1021
	lh = sis->extent_list.prev;	/* The highest page extent */
	if (lh != &sis->extent_list) {
L
Linus Torvalds 已提交
1022
		se = list_entry(lh, struct swap_extent, list);
1023 1024
		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
1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040
			/* 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;

1041
	list_add_tail(&new_se->list, &sis->extent_list);
1042
	return 1;
L
Linus Torvalds 已提交
1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064
}

/*
 * 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.
 *
1065
 * For S_ISREG swapfiles we set S_SWAPFILE across the life of the swapon.  This
L
Linus Torvalds 已提交
1066 1067 1068 1069 1070 1071 1072 1073 1074 1075
 * 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.
 */
1076
static int setup_swap_extents(struct swap_info_struct *sis, sector_t *span)
L
Linus Torvalds 已提交
1077 1078 1079 1080 1081 1082 1083
{
	struct inode *inode;
	unsigned blocks_per_page;
	unsigned long page_no;
	unsigned blkbits;
	sector_t probe_block;
	sector_t last_block;
1084 1085 1086
	sector_t lowest_block = -1;
	sector_t highest_block = 0;
	int nr_extents = 0;
L
Linus Torvalds 已提交
1087 1088 1089 1090 1091
	int ret;

	inode = sis->swap_file->f_mapping->host;
	if (S_ISBLK(inode->i_mode)) {
		ret = add_swap_extent(sis, 0, sis->max, 0);
1092
		*span = sis->pages;
L
Linus Torvalds 已提交
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 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136
		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;
			}
		}

1137 1138 1139 1140 1141 1142 1143 1144
		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
1145 1146 1147
		/*
		 * We found a PAGE_SIZE-length, PAGE_SIZE-aligned run of blocks
		 */
1148 1149
		ret = add_swap_extent(sis, page_no, 1, first_block);
		if (ret < 0)
L
Linus Torvalds 已提交
1150
			goto out;
1151
		nr_extents += ret;
L
Linus Torvalds 已提交
1152 1153 1154 1155 1156
		page_no++;
		probe_block += blocks_per_page;
reprobe:
		continue;
	}
1157 1158
	ret = nr_extents;
	*span = 1 + highest_block - lowest_block;
L
Linus Torvalds 已提交
1159
	if (page_no == 0)
1160
		page_no = 1;	/* force Empty message */
L
Linus Torvalds 已提交
1161
	sis->max = page_no;
1162
	sis->pages = page_no - 1;
L
Linus Torvalds 已提交
1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180
	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;
}

#if 0	/* We don't need this yet */
#include <linux/backing-dev.h>
int page_queue_congested(struct page *page)
{
	struct backing_dev_info *bdi;

1181
	VM_BUG_ON(!PageLocked(page));	/* It pins the swap_info_struct */
L
Linus Torvalds 已提交
1182 1183

	if (PageSwapCache(page)) {
H
Hugh Dickins 已提交
1184
		swp_entry_t entry = { .val = page_private(page) };
L
Linus Torvalds 已提交
1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221
		struct swap_info_struct *sis;

		sis = get_swap_info_struct(swp_type(entry));
		bdi = sis->bdev->bd_inode->i_mapping->backing_dev_info;
	} else
		bdi = page->mapping->backing_dev_info;
	return bdi_write_congested(bdi);
}
#endif

asmlinkage long sys_swapoff(const char __user * specialfile)
{
	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;
	
	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;
1222
	spin_lock(&swap_lock);
L
Linus Torvalds 已提交
1223 1224 1225 1226 1227 1228 1229 1230 1231 1232
	for (type = swap_list.head; type >= 0; type = swap_info[type].next) {
		p = swap_info + type;
		if ((p->flags & SWP_ACTIVE) == SWP_ACTIVE) {
			if (p->swap_file->f_mapping == mapping)
				break;
		}
		prev = type;
	}
	if (type < 0) {
		err = -EINVAL;
1233
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
1234 1235 1236 1237 1238 1239
		goto out_dput;
	}
	if (!security_vm_enough_memory(p->pages))
		vm_unacct_memory(p->pages);
	else {
		err = -ENOMEM;
1240
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251
		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;
	}
1252 1253 1254 1255 1256
	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
1257 1258 1259
	nr_swap_pages -= p->pages;
	total_swap_pages -= p->pages;
	p->flags &= ~SWP_WRITEOK;
1260
	spin_unlock(&swap_lock);
1261

L
Linus Torvalds 已提交
1262 1263 1264 1265 1266 1267
	current->flags |= PF_SWAPOFF;
	err = try_to_unuse(type);
	current->flags &= ~PF_SWAPOFF;

	if (err) {
		/* re-insert swap space back into swap_list */
1268
		spin_lock(&swap_lock);
1269 1270 1271 1272
		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
1273 1274
			if (p->prio >= swap_info[i].prio)
				break;
1275 1276
			prev = i;
		}
L
Linus Torvalds 已提交
1277 1278 1279 1280 1281 1282 1283 1284
		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;
1285
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
1286 1287
		goto out_dput;
	}
1288 1289 1290 1291 1292

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

1293
	destroy_swap_extents(p);
I
Ingo Molnar 已提交
1294
	mutex_lock(&swapon_mutex);
1295 1296 1297
	spin_lock(&swap_lock);
	drain_mmlist();

1298 1299 1300
	/* wait for anyone still in scan_swap_map */
	p->highest_bit = 0;		/* cuts scans short */
	while (p->flags >= SWP_SCANNING) {
1301
		spin_unlock(&swap_lock);
1302
		schedule_timeout_uninterruptible(1);
1303
		spin_lock(&swap_lock);
1304 1305
	}

L
Linus Torvalds 已提交
1306 1307 1308 1309 1310 1311
	swap_file = p->swap_file;
	p->swap_file = NULL;
	p->max = 0;
	swap_map = p->swap_map;
	p->swap_map = NULL;
	p->flags = 0;
1312
	spin_unlock(&swap_lock);
I
Ingo Molnar 已提交
1313
	mutex_unlock(&swapon_mutex);
L
Linus Torvalds 已提交
1314 1315 1316 1317 1318 1319 1320
	vfree(swap_map);
	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 {
1321
		mutex_lock(&inode->i_mutex);
L
Linus Torvalds 已提交
1322
		inode->i_flags &= ~S_SWAPFILE;
1323
		mutex_unlock(&inode->i_mutex);
L
Linus Torvalds 已提交
1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341
	}
	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
1342
	mutex_lock(&swapon_mutex);
L
Linus Torvalds 已提交
1343

1344 1345 1346
	if (!l)
		return SEQ_START_TOKEN;

L
Linus Torvalds 已提交
1347 1348 1349
	for (i = 0; i < nr_swapfiles; i++, ptr++) {
		if (!(ptr->flags & SWP_USED) || !ptr->swap_map)
			continue;
1350
		if (!--l)
L
Linus Torvalds 已提交
1351 1352 1353 1354 1355 1356 1357 1358
			return ptr;
	}

	return NULL;
}

static void *swap_next(struct seq_file *swap, void *v, loff_t *pos)
{
1359
	struct swap_info_struct *ptr;
L
Linus Torvalds 已提交
1360 1361
	struct swap_info_struct *endptr = swap_info + nr_swapfiles;

1362 1363 1364 1365 1366 1367 1368 1369
	if (v == SEQ_START_TOKEN)
		ptr = swap_info;
	else {
		ptr = v;
		ptr++;
	}

	for (; ptr < endptr; ptr++) {
L
Linus Torvalds 已提交
1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380
		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
1381
	mutex_unlock(&swapon_mutex);
L
Linus Torvalds 已提交
1382 1383 1384 1385 1386 1387 1388 1389
}

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

1390 1391 1392 1393
	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
1394 1395

	file = ptr->swap_file;
1396
	len = seq_path(swap, &file->f_path, " \t\n\\");
1397
	seq_printf(swap, "%*s%s\t%u\t%u\t%d\n",
L
Linus Torvalds 已提交
1398
		       len < 40 ? 40 - len : 1, " ",
1399
		       S_ISBLK(file->f_path.dentry->d_inode->i_mode) ?
L
Linus Torvalds 已提交
1400 1401 1402 1403 1404 1405 1406
				"partition" : "file\t",
		       ptr->pages << (PAGE_SHIFT - 10),
		       ptr->inuse_pages << (PAGE_SHIFT - 10),
		       ptr->prio);
	return 0;
}

1407
static const struct seq_operations swaps_op = {
L
Linus Torvalds 已提交
1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418
	.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);
}

1419
static const struct file_operations proc_swaps_operations = {
L
Linus Torvalds 已提交
1420 1421 1422 1423 1424 1425 1426 1427
	.open		= swaps_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
	.release	= seq_release,
};

static int __init procswaps_init(void)
{
1428
	proc_create("swaps", 0, NULL, &proc_swaps_operations);
L
Linus Torvalds 已提交
1429 1430 1431 1432 1433
	return 0;
}
__initcall(procswaps_init);
#endif /* CONFIG_PROC_FS */

J
Jan Beulich 已提交
1434 1435 1436 1437 1438 1439 1440 1441 1442
#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
1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459
/*
 * Written 01/25/92 by Simmule Turner, heavily changed by Linus.
 *
 * The swapon system call
 */
asmlinkage long sys_swapon(const char __user * specialfile, int swap_flags)
{
	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;
	int swap_header_version;
1460 1461
	unsigned int nr_good_pages = 0;
	int nr_extents = 0;
1462
	sector_t span;
L
Linus Torvalds 已提交
1463
	unsigned long maxpages = 1;
1464
	unsigned long swapfilepages;
1465
	unsigned short *swap_map = NULL;
L
Linus Torvalds 已提交
1466 1467 1468 1469 1470 1471
	struct page *page = NULL;
	struct inode *inode = NULL;
	int did_down = 0;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
1472
	spin_lock(&swap_lock);
L
Linus Torvalds 已提交
1473 1474 1475 1476 1477
	p = swap_info;
	for (type = 0 ; type < nr_swapfiles ; type++,p++)
		if (!(p->flags & SWP_USED))
			break;
	error = -EPERM;
1478
	if (type >= MAX_SWAPFILES) {
1479
		spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
1480 1481 1482 1483
		goto out;
	}
	if (type >= nr_swapfiles)
		nr_swapfiles = type+1;
1484
	memset(p, 0, sizeof(*p));
L
Linus Torvalds 已提交
1485 1486 1487
	INIT_LIST_HEAD(&p->extent_list);
	p->flags = SWP_USED;
	p->next = -1;
1488
	spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521
	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
1522
			error = -EINVAL;
L
Linus Torvalds 已提交
1523 1524 1525 1526 1527 1528 1529 1530 1531
			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;
1532
		mutex_lock(&inode->i_mutex);
L
Linus Torvalds 已提交
1533 1534 1535 1536 1537 1538 1539 1540 1541
		did_down = 1;
		if (IS_SWAPFILE(inode)) {
			error = -EBUSY;
			goto bad_swap;
		}
	} else {
		goto bad_swap;
	}

1542
	swapfilepages = i_size_read(inode) >> PAGE_SHIFT;
L
Linus Torvalds 已提交
1543 1544 1545 1546 1547 1548 1549 1550

	/*
	 * Read the swap header.
	 */
	if (!mapping->a_ops->readpage) {
		error = -EINVAL;
		goto bad_swap;
	}
1551
	page = read_mapping_page(mapping, 0, swap_file);
L
Linus Torvalds 已提交
1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563
	if (IS_ERR(page)) {
		error = PTR_ERR(page);
		goto bad_swap;
	}
	kmap(page);
	swap_header = page_address(page);

	if (!memcmp("SWAP-SPACE",swap_header->magic.magic,10))
		swap_header_version = 1;
	else if (!memcmp("SWAPSPACE2",swap_header->magic.magic,10))
		swap_header_version = 2;
	else {
1564
		printk(KERN_ERR "Unable to find swap-space signature\n");
L
Linus Torvalds 已提交
1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575
		error = -EINVAL;
		goto bad_swap;
	}
	
	switch (swap_header_version) {
	case 1:
		printk(KERN_ERR "version 0 swap is no longer supported. "
			"Use mkswap -v1 %s\n", name);
		error = -EINVAL;
		goto bad_swap;
	case 2:
1576 1577 1578 1579 1580 1581 1582 1583
		/* 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]);
		}
L
Linus Torvalds 已提交
1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594
		/* Check the swap header's sub-version and the size of
                   the swap file and bad block lists */
		if (swap_header->info.version != 1) {
			printk(KERN_WARNING
			       "Unable to handle swap header version %d\n",
			       swap_header->info.version);
			error = -EINVAL;
			goto bad_swap;
		}

		p->lowest_bit  = 1;
1595 1596
		p->cluster_next = 1;

L
Linus Torvalds 已提交
1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616
		/*
		 * 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;

		error = -EINVAL;
1617 1618
		if (!maxpages)
			goto bad_swap;
1619
		if (swapfilepages && maxpages > swapfilepages) {
1620 1621 1622 1623
			printk(KERN_WARNING
			       "Swap area shorter than signature indicates\n");
			goto bad_swap;
		}
1624 1625
		if (swap_header->info.nr_badpages && S_ISREG(inode->i_mode))
			goto bad_swap;
L
Linus Torvalds 已提交
1626 1627
		if (swap_header->info.nr_badpages > MAX_SWAP_BADPAGES)
			goto bad_swap;
1628

L
Linus Torvalds 已提交
1629
		/* OK, set up the swap map and apply the bad block list */
1630 1631
		swap_map = vmalloc(maxpages * sizeof(short));
		if (!swap_map) {
L
Linus Torvalds 已提交
1632 1633 1634 1635 1636
			error = -ENOMEM;
			goto bad_swap;
		}

		error = 0;
1637
		memset(swap_map, 0, maxpages * sizeof(short));
1638 1639 1640
		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)
L
Linus Torvalds 已提交
1641 1642
				error = -EINVAL;
			else
1643
				swap_map[page_nr] = SWAP_MAP_BAD;
L
Linus Torvalds 已提交
1644 1645 1646 1647
		}
		nr_good_pages = swap_header->info.last_page -
				swap_header->info.nr_badpages -
				1 /* header page */;
1648
		if (error)
L
Linus Torvalds 已提交
1649 1650
			goto bad_swap;
	}
1651 1652

	if (nr_good_pages) {
1653
		swap_map[0] = SWAP_MAP_BAD;
1654 1655
		p->max = maxpages;
		p->pages = nr_good_pages;
1656 1657 1658
		nr_extents = setup_swap_extents(p, &span);
		if (nr_extents < 0) {
			error = nr_extents;
1659
			goto bad_swap;
1660
		}
1661 1662
		nr_good_pages = p->pages;
	}
L
Linus Torvalds 已提交
1663 1664 1665 1666 1667 1668
	if (!nr_good_pages) {
		printk(KERN_WARNING "Empty swap-file\n");
		error = -EINVAL;
		goto bad_swap;
	}

I
Ingo Molnar 已提交
1669
	mutex_lock(&swapon_mutex);
1670
	spin_lock(&swap_lock);
1671 1672 1673 1674 1675 1676
	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;
L
Linus Torvalds 已提交
1677 1678 1679
	p->flags = SWP_ACTIVE;
	nr_swap_pages += nr_good_pages;
	total_swap_pages += nr_good_pages;
1680

1681
	printk(KERN_INFO "Adding %uk swap on %s.  "
1682 1683 1684
			"Priority:%d extents:%d across:%lluk\n",
		nr_good_pages<<(PAGE_SHIFT-10), name, p->prio,
		nr_extents, (unsigned long long)span<<(PAGE_SHIFT-10));
L
Linus Torvalds 已提交
1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699

	/* 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;
	}
1700
	spin_unlock(&swap_lock);
I
Ingo Molnar 已提交
1701
	mutex_unlock(&swapon_mutex);
L
Linus Torvalds 已提交
1702 1703 1704 1705 1706 1707 1708
	error = 0;
	goto out;
bad_swap:
	if (bdev) {
		set_blocksize(bdev, p->old_block_size);
		bd_release(bdev);
	}
1709
	destroy_swap_extents(p);
L
Linus Torvalds 已提交
1710
bad_swap_2:
1711
	spin_lock(&swap_lock);
L
Linus Torvalds 已提交
1712 1713
	p->swap_file = NULL;
	p->flags = 0;
1714
	spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727
	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;
1728
		mutex_unlock(&inode->i_mutex);
L
Linus Torvalds 已提交
1729 1730 1731 1732 1733 1734 1735 1736 1737
	}
	return error;
}

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

1738
	spin_lock(&swap_lock);
L
Linus Torvalds 已提交
1739 1740 1741 1742 1743 1744 1745 1746
	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;
1747
	spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761
}

/*
 * 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.
 */
int swap_duplicate(swp_entry_t entry)
{
	struct swap_info_struct * p;
	unsigned long offset, type;
	int result = 0;

1762 1763 1764
	if (is_migration_entry(entry))
		return 1;

L
Linus Torvalds 已提交
1765 1766 1767 1768 1769 1770
	type = swp_type(entry);
	if (type >= nr_swapfiles)
		goto bad_file;
	p = type + swap_info;
	offset = swp_offset(entry);

1771
	spin_lock(&swap_lock);
L
Linus Torvalds 已提交
1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782
	if (offset < p->max && p->swap_map[offset]) {
		if (p->swap_map[offset] < SWAP_MAP_MAX - 1) {
			p->swap_map[offset]++;
			result = 1;
		} else if (p->swap_map[offset] <= SWAP_MAP_MAX) {
			if (swap_overflow++ < 5)
				printk(KERN_WARNING "swap_dup: swap entry overflow\n");
			p->swap_map[offset] = SWAP_MAP_MAX;
			result = 1;
		}
	}
1783
	spin_unlock(&swap_lock);
L
Linus Torvalds 已提交
1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798
out:
	return result;

bad_file:
	printk(KERN_ERR "swap_dup: %s%08lx\n", Bad_file, entry.val);
	goto out;
}

struct swap_info_struct *
get_swap_info_struct(unsigned type)
{
	return &swap_info[type];
}

/*
1799
 * swap_lock prevents swap_map being freed. Don't grab an extra
L
Linus Torvalds 已提交
1800 1801 1802 1803
 * reference on the swaphandle, it doesn't matter if it becomes unused.
 */
int valid_swaphandles(swp_entry_t entry, unsigned long *offset)
{
1804
	struct swap_info_struct *si;
H
Hugh Dickins 已提交
1805
	int our_page_cluster = page_cluster;
1806 1807 1808
	pgoff_t target, toff;
	pgoff_t base, end;
	int nr_pages = 0;
L
Linus Torvalds 已提交
1809

H
Hugh Dickins 已提交
1810
	if (!our_page_cluster)	/* no readahead */
L
Linus Torvalds 已提交
1811
		return 0;
1812 1813 1814 1815 1816 1817 1818

	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
1819

1820
	spin_lock(&swap_lock);
1821 1822 1823 1824 1825 1826 1827 1828 1829
	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;
		if (si->swap_map[toff] == SWAP_MAP_BAD)
L
Linus Torvalds 已提交
1830
			break;
1831 1832 1833
	}
	/* Count contiguous allocated slots below our target */
	for (toff = target; --toff >= base; nr_pages++) {
L
Linus Torvalds 已提交
1834
		/* Don't read in free or bad pages */
1835
		if (!si->swap_map[toff])
L
Linus Torvalds 已提交
1836
			break;
1837
		if (si->swap_map[toff] == SWAP_MAP_BAD)
L
Linus Torvalds 已提交
1838
			break;
1839
	}
1840
	spin_unlock(&swap_lock);
1841 1842 1843 1844 1845 1846 1847

	/*
	 * 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
1848
}