shmem.c 67.3 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1 2 3 4 5 6 7 8
/*
 * Resizable virtual memory filesystem for Linux.
 *
 * Copyright (C) 2000 Linus Torvalds.
 *		 2000 Transmeta Corp.
 *		 2000-2001 Christoph Rohland
 *		 2000-2001 SAP AG
 *		 2002 Red Hat Inc.
9 10
 * Copyright (C) 2002-2005 Hugh Dickins.
 * Copyright (C) 2002-2005 VERITAS Software Corporation.
Linus Torvalds's avatar
Linus Torvalds committed
11 12 13 14 15 16
 * Copyright (C) 2004 Andi Kleen, SuSE Labs
 *
 * Extended attribute support for tmpfs:
 * Copyright (c) 2004, Luke Kenneth Casson Leighton <lkcl@lkcl.net>
 * Copyright (c) 2004 Red Hat, Inc., James Morris <jmorris@redhat.com>
 *
17 18 19
 * tiny-shmem:
 * Copyright (c) 2004, 2008 Matt Mackall <mpm@selenic.com>
 *
Linus Torvalds's avatar
Linus Torvalds committed
20 21 22
 * This file is released under the GPL.
 */

23 24 25 26
#include <linux/fs.h>
#include <linux/init.h>
#include <linux/vfs.h>
#include <linux/mount.h>
Hugh Dickins's avatar
Hugh Dickins committed
27
#include <linux/pagemap.h>
28 29 30 31 32 33 34 35
#include <linux/file.h>
#include <linux/mm.h>
#include <linux/module.h>
#include <linux/swap.h>

static struct vfsmount *shm_mnt;

#ifdef CONFIG_SHMEM
Linus Torvalds's avatar
Linus Torvalds committed
36 37 38 39 40 41
/*
 * This virtual memory filesystem is heavily based on the ramfs. It
 * extends ramfs by the ability to use swap and honor resource limits
 * which makes it a completely usable filesystem.
 */

42
#include <linux/xattr.h>
43
#include <linux/exportfs.h>
44
#include <linux/posix_acl.h>
45
#include <linux/generic_acl.h>
Linus Torvalds's avatar
Linus Torvalds committed
46 47 48 49 50 51 52
#include <linux/mman.h>
#include <linux/string.h>
#include <linux/slab.h>
#include <linux/backing-dev.h>
#include <linux/shmem_fs.h>
#include <linux/writeback.h>
#include <linux/blkdev.h>
53
#include <linux/pagevec.h>
54
#include <linux/percpu_counter.h>
55
#include <linux/splice.h>
Linus Torvalds's avatar
Linus Torvalds committed
56 57 58 59
#include <linux/security.h>
#include <linux/swapops.h>
#include <linux/mempolicy.h>
#include <linux/namei.h>
60
#include <linux/ctype.h>
61
#include <linux/migrate.h>
62
#include <linux/highmem.h>
63
#include <linux/seq_file.h>
Mimi Zohar's avatar
Mimi Zohar committed
64
#include <linux/magic.h>
65

Linus Torvalds's avatar
Linus Torvalds committed
66 67 68
#include <asm/uaccess.h>
#include <asm/pgtable.h>

Hugh Dickins's avatar
Hugh Dickins committed
69
#define BLOCKS_PER_PAGE  (PAGE_CACHE_SIZE/512)
Linus Torvalds's avatar
Linus Torvalds committed
70 71 72 73 74
#define VM_ACCT(size)    (PAGE_CACHE_ALIGN(size) >> PAGE_SHIFT)

/* Pretend that each entry is of this size in directory's i_size */
#define BOGO_DIRENT_SIZE 20

75 76 77 78 79 80 81
struct shmem_xattr {
	struct list_head list;	/* anchored by shmem_inode_info->xattr_list */
	char *name;		/* xattr name */
	size_t size;
	char value[0];
};

82
/* Flag allocation requirements to shmem_getpage */
Linus Torvalds's avatar
Linus Torvalds committed
83 84 85
enum sgp_type {
	SGP_READ,	/* don't exceed i_size, don't allocate page */
	SGP_CACHE,	/* don't exceed i_size, may allocate page */
86
	SGP_DIRTY,	/* like SGP_CACHE, but set new page dirty */
Linus Torvalds's avatar
Linus Torvalds committed
87 88 89
	SGP_WRITE,	/* may exceed i_size, may allocate page */
};

Andrew Morton's avatar
Andrew Morton committed
90
#ifdef CONFIG_TMPFS
91 92 93 94 95 96 97 98 99
static unsigned long shmem_default_max_blocks(void)
{
	return totalram_pages / 2;
}

static unsigned long shmem_default_max_inodes(void)
{
	return min(totalram_pages - totalhigh_pages, totalram_pages / 2);
}
Andrew Morton's avatar
Andrew Morton committed
100
#endif
101

102 103 104 105 106 107 108 109 110
static int shmem_getpage_gfp(struct inode *inode, pgoff_t index,
	struct page **pagep, enum sgp_type sgp, gfp_t gfp, int *fault_type);

static inline int shmem_getpage(struct inode *inode, pgoff_t index,
	struct page **pagep, enum sgp_type sgp, int *fault_type)
{
	return shmem_getpage_gfp(inode, index, pagep, sgp,
			mapping_gfp_mask(inode->i_mapping), fault_type);
}
Linus Torvalds's avatar
Linus Torvalds committed
111 112 113 114 115 116 117 118 119 120 121 122 123 124

static inline struct shmem_sb_info *SHMEM_SB(struct super_block *sb)
{
	return sb->s_fs_info;
}

/*
 * shmem_file_setup pre-accounts the whole fixed size of a VM object,
 * for shared memory and for shared anonymous (/dev/zero) mappings
 * (unless MAP_NORESERVE and sysctl_overcommit_memory <= 1),
 * consistent with the pre-accounting of private mappings ...
 */
static inline int shmem_acct_size(unsigned long flags, loff_t size)
{
125 126
	return (flags & VM_NORESERVE) ?
		0 : security_vm_enough_memory_kern(VM_ACCT(size));
Linus Torvalds's avatar
Linus Torvalds committed
127 128 129 130
}

static inline void shmem_unacct_size(unsigned long flags, loff_t size)
{
131
	if (!(flags & VM_NORESERVE))
Linus Torvalds's avatar
Linus Torvalds committed
132 133 134 135 136 137 138 139 140 141 142
		vm_unacct_memory(VM_ACCT(size));
}

/*
 * ... whereas tmpfs objects are accounted incrementally as
 * pages are allocated, in order to allow huge sparse files.
 * shmem_getpage reports shmem_acct_block failure as -ENOSPC not -ENOMEM,
 * so that a failure on a sparse tmpfs mapping will give SIGBUS not OOM.
 */
static inline int shmem_acct_block(unsigned long flags)
{
143 144
	return (flags & VM_NORESERVE) ?
		security_vm_enough_memory_kern(VM_ACCT(PAGE_CACHE_SIZE)) : 0;
Linus Torvalds's avatar
Linus Torvalds committed
145 146 147 148
}

static inline void shmem_unacct_blocks(unsigned long flags, long pages)
{
149
	if (flags & VM_NORESERVE)
Linus Torvalds's avatar
Linus Torvalds committed
150 151 152
		vm_unacct_memory(pages * VM_ACCT(PAGE_CACHE_SIZE));
}

153
static const struct super_operations shmem_ops;
154
static const struct address_space_operations shmem_aops;
155
static const struct file_operations shmem_file_operations;
156 157 158
static const struct inode_operations shmem_inode_operations;
static const struct inode_operations shmem_dir_inode_operations;
static const struct inode_operations shmem_special_inode_operations;
159
static const struct vm_operations_struct shmem_vm_ops;
Linus Torvalds's avatar
Linus Torvalds committed
160

161
static struct backing_dev_info shmem_backing_dev_info  __read_mostly = {
Linus Torvalds's avatar
Linus Torvalds committed
162
	.ra_pages	= 0,	/* No readahead */
163
	.capabilities	= BDI_CAP_NO_ACCT_AND_WRITEBACK | BDI_CAP_SWAP_BACKED,
Linus Torvalds's avatar
Linus Torvalds committed
164 165 166
};

static LIST_HEAD(shmem_swaplist);
167
static DEFINE_MUTEX(shmem_swaplist_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
168 169 170 171

static void shmem_free_blocks(struct inode *inode, long pages)
{
	struct shmem_sb_info *sbinfo = SHMEM_SB(inode->i_sb);
172
	if (sbinfo->max_blocks) {
173
		percpu_counter_add(&sbinfo->used_blocks, -pages);
Linus Torvalds's avatar
Linus Torvalds committed
174 175 176 177
		inode->i_blocks -= pages*BLOCKS_PER_PAGE;
	}
}

178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202
static int shmem_reserve_inode(struct super_block *sb)
{
	struct shmem_sb_info *sbinfo = SHMEM_SB(sb);
	if (sbinfo->max_inodes) {
		spin_lock(&sbinfo->stat_lock);
		if (!sbinfo->free_inodes) {
			spin_unlock(&sbinfo->stat_lock);
			return -ENOSPC;
		}
		sbinfo->free_inodes--;
		spin_unlock(&sbinfo->stat_lock);
	}
	return 0;
}

static void shmem_free_inode(struct super_block *sb)
{
	struct shmem_sb_info *sbinfo = SHMEM_SB(sb);
	if (sbinfo->max_inodes) {
		spin_lock(&sbinfo->stat_lock);
		sbinfo->free_inodes++;
		spin_unlock(&sbinfo->stat_lock);
	}
}

203
/**
204
 * shmem_recalc_inode - recalculate the block usage of an inode
Linus Torvalds's avatar
Linus Torvalds committed
205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227
 * @inode: inode to recalc
 *
 * We have to calculate the free blocks since the mm can drop
 * undirtied hole pages behind our back.
 *
 * But normally   info->alloced == inode->i_mapping->nrpages + info->swapped
 * So mm freed is info->alloced - (inode->i_mapping->nrpages + info->swapped)
 *
 * It has to be called with the spinlock held.
 */
static void shmem_recalc_inode(struct inode *inode)
{
	struct shmem_inode_info *info = SHMEM_I(inode);
	long freed;

	freed = info->alloced - info->swapped - inode->i_mapping->nrpages;
	if (freed > 0) {
		info->alloced -= freed;
		shmem_unacct_blocks(info->flags, freed);
		shmem_free_blocks(inode, freed);
	}
}

228 229
static void shmem_put_swap(struct shmem_inode_info *info, pgoff_t index,
			   swp_entry_t swap)
Linus Torvalds's avatar
Linus Torvalds committed
230
{
231 232
	if (index < SHMEM_NR_DIRECT)
		info->i_direct[index] = swap;
Linus Torvalds's avatar
Linus Torvalds committed
233 234
}

235
static swp_entry_t shmem_get_swap(struct shmem_inode_info *info, pgoff_t index)
Linus Torvalds's avatar
Linus Torvalds committed
236
{
237 238
	return (index < SHMEM_NR_DIRECT) ?
		info->i_direct[index] : (swp_entry_t){0};
Linus Torvalds's avatar
Linus Torvalds committed
239 240
}

241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263
/*
 * Replace item expected in radix tree by a new item, while holding tree lock.
 */
static int shmem_radix_tree_replace(struct address_space *mapping,
			pgoff_t index, void *expected, void *replacement)
{
	void **pslot;
	void *item = NULL;

	VM_BUG_ON(!expected);
	pslot = radix_tree_lookup_slot(&mapping->page_tree, index);
	if (pslot)
		item = radix_tree_deref_slot_protected(pslot,
							&mapping->tree_lock);
	if (item != expected)
		return -ENOENT;
	if (replacement)
		radix_tree_replace_slot(pslot, replacement);
	else
		radix_tree_delete(&mapping->page_tree, index);
	return 0;
}

264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312
/*
 * Like add_to_page_cache_locked, but error if expected item has gone.
 */
static int shmem_add_to_page_cache(struct page *page,
				   struct address_space *mapping,
				   pgoff_t index, gfp_t gfp, void *expected)
{
	int error;

	VM_BUG_ON(!PageLocked(page));
	VM_BUG_ON(!PageSwapBacked(page));

	error = mem_cgroup_cache_charge(page, current->mm,
						gfp & GFP_RECLAIM_MASK);
	if (error)
		goto out;
	if (!expected)
		error = radix_tree_preload(gfp & GFP_RECLAIM_MASK);
	if (!error) {
		page_cache_get(page);
		page->mapping = mapping;
		page->index = index;

		spin_lock_irq(&mapping->tree_lock);
		if (!expected)
			error = radix_tree_insert(&mapping->page_tree,
							index, page);
		else
			error = shmem_radix_tree_replace(mapping, index,
							expected, page);
		if (!error) {
			mapping->nrpages++;
			__inc_zone_page_state(page, NR_FILE_PAGES);
			__inc_zone_page_state(page, NR_SHMEM);
			spin_unlock_irq(&mapping->tree_lock);
		} else {
			page->mapping = NULL;
			spin_unlock_irq(&mapping->tree_lock);
			page_cache_release(page);
		}
		if (!expected)
			radix_tree_preload_end();
	}
	if (error)
		mem_cgroup_uncharge_cache_page(page);
out:
	return error;
}

313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359
/*
 * Like find_get_pages, but collecting swap entries as well as pages.
 */
static unsigned shmem_find_get_pages_and_swap(struct address_space *mapping,
					pgoff_t start, unsigned int nr_pages,
					struct page **pages, pgoff_t *indices)
{
	unsigned int i;
	unsigned int ret;
	unsigned int nr_found;

	rcu_read_lock();
restart:
	nr_found = radix_tree_gang_lookup_slot(&mapping->page_tree,
				(void ***)pages, indices, start, nr_pages);
	ret = 0;
	for (i = 0; i < nr_found; i++) {
		struct page *page;
repeat:
		page = radix_tree_deref_slot((void **)pages[i]);
		if (unlikely(!page))
			continue;
		if (radix_tree_exception(page)) {
			if (radix_tree_exceptional_entry(page))
				goto export;
			/* radix_tree_deref_retry(page) */
			goto restart;
		}
		if (!page_cache_get_speculative(page))
			goto repeat;

		/* Has the page moved? */
		if (unlikely(page != *((void **)pages[i]))) {
			page_cache_release(page);
			goto repeat;
		}
export:
		indices[ret] = indices[i];
		pages[ret] = page;
		ret++;
	}
	if (unlikely(!ret && nr_found))
		goto restart;
	rcu_read_unlock();
	return ret;
}

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
/*
 * Lockless lookup of swap entry in radix tree, avoiding refcount on pages.
 */
static pgoff_t shmem_find_swap(struct address_space *mapping, void *radswap)
{
	void  **slots[PAGEVEC_SIZE];
	pgoff_t indices[PAGEVEC_SIZE];
	unsigned int nr_found;

restart:
	nr_found = 1;
	indices[0] = -1;
	while (nr_found) {
		pgoff_t index = indices[nr_found - 1] + 1;
		unsigned int i;

		rcu_read_lock();
		nr_found = radix_tree_gang_lookup_slot(&mapping->page_tree,
					slots, indices, index, PAGEVEC_SIZE);
		for (i = 0; i < nr_found; i++) {
			void *item = radix_tree_deref_slot(slots[i]);
			if (radix_tree_deref_retry(item)) {
				rcu_read_unlock();
				goto restart;
			}
			if (item == radswap) {
				rcu_read_unlock();
				return indices[i];
			}
		}
		rcu_read_unlock();
		cond_resched();
	}
	return -1;
}

396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430
/*
 * Remove swap entry from radix tree, free the swap and its page cache.
 */
static int shmem_free_swap(struct address_space *mapping,
			   pgoff_t index, void *radswap)
{
	int error;

	spin_lock_irq(&mapping->tree_lock);
	error = shmem_radix_tree_replace(mapping, index, radswap, NULL);
	spin_unlock_irq(&mapping->tree_lock);
	if (!error)
		free_swap_and_cache(radix_to_swp_entry(radswap));
	return error;
}

/*
 * Pagevec may contain swap entries, so shuffle up pages before releasing.
 */
static void shmem_pagevec_release(struct pagevec *pvec)
{
	int i, j;

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

/*
 * Remove range of pages and swap entries from radix tree, and free them.
 */
431
void shmem_truncate_range(struct inode *inode, loff_t lstart, loff_t lend)
Linus Torvalds's avatar
Linus Torvalds committed
432
{
433
	struct address_space *mapping = inode->i_mapping;
Linus Torvalds's avatar
Linus Torvalds committed
434
	struct shmem_inode_info *info = SHMEM_I(inode);
435
	pgoff_t start = (lstart + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
436
	unsigned partial = lstart & (PAGE_CACHE_SIZE - 1);
437
	pgoff_t end = (lend >> PAGE_CACHE_SHIFT);
438
	struct pagevec pvec;
439 440
	pgoff_t indices[PAGEVEC_SIZE];
	long nr_swaps_freed = 0;
441
	pgoff_t index;
442 443 444 445 446 447
	int i;

	BUG_ON((lend & (PAGE_CACHE_SIZE - 1)) != (PAGE_CACHE_SIZE - 1));

	pagevec_init(&pvec, 0);
	index = start;
448 449 450 451 452 453
	while (index <= end) {
		pvec.nr = shmem_find_get_pages_and_swap(mapping, index,
			min(end - index, (pgoff_t)PAGEVEC_SIZE - 1) + 1,
							pvec.pages, indices);
		if (!pvec.nr)
			break;
454 455 456 457
		mem_cgroup_uncharge_start();
		for (i = 0; i < pagevec_count(&pvec); i++) {
			struct page *page = pvec.pages[i];

458
			index = indices[i];
459 460 461
			if (index > end)
				break;

462 463 464
			if (radix_tree_exceptional_entry(page)) {
				nr_swaps_freed += !shmem_free_swap(mapping,
								index, page);
465
				continue;
466 467 468
			}

			if (!trylock_page(page))
469
				continue;
470 471 472
			if (page->mapping == mapping) {
				VM_BUG_ON(PageWriteback(page));
				truncate_inode_page(mapping, page);
473 474 475
			}
			unlock_page(page);
		}
476
		shmem_pagevec_release(&pvec);
477 478 479 480
		mem_cgroup_uncharge_end();
		cond_resched();
		index++;
	}
Linus Torvalds's avatar
Linus Torvalds committed
481

482 483 484 485 486 487 488 489 490 491 492 493 494 495
	if (partial) {
		struct page *page = NULL;
		shmem_getpage(inode, start - 1, &page, SGP_READ, NULL);
		if (page) {
			zero_user_segment(page, partial, PAGE_CACHE_SIZE);
			set_page_dirty(page);
			unlock_page(page);
			page_cache_release(page);
		}
	}

	index = start;
	for ( ; ; ) {
		cond_resched();
496 497 498 499
		pvec.nr = shmem_find_get_pages_and_swap(mapping, index,
			min(end - index, (pgoff_t)PAGEVEC_SIZE - 1) + 1,
							pvec.pages, indices);
		if (!pvec.nr) {
500 501 502 503 504
			if (index == start)
				break;
			index = start;
			continue;
		}
505 506
		if (index == start && indices[0] > end) {
			shmem_pagevec_release(&pvec);
507 508 509 510 511 512
			break;
		}
		mem_cgroup_uncharge_start();
		for (i = 0; i < pagevec_count(&pvec); i++) {
			struct page *page = pvec.pages[i];

513
			index = indices[i];
514 515 516
			if (index > end)
				break;

517 518 519 520 521 522
			if (radix_tree_exceptional_entry(page)) {
				nr_swaps_freed += !shmem_free_swap(mapping,
								index, page);
				continue;
			}

523
			lock_page(page);
524 525 526 527
			if (page->mapping == mapping) {
				VM_BUG_ON(PageWriteback(page));
				truncate_inode_page(mapping, page);
			}
528 529
			unlock_page(page);
		}
530
		shmem_pagevec_release(&pvec);
531 532 533
		mem_cgroup_uncharge_end();
		index++;
	}
534

Linus Torvalds's avatar
Linus Torvalds committed
535
	spin_lock(&info->lock);
536
	info->swapped -= nr_swaps_freed;
Linus Torvalds's avatar
Linus Torvalds committed
537 538 539
	shmem_recalc_inode(inode);
	spin_unlock(&info->lock);

540
	inode->i_ctime = inode->i_mtime = CURRENT_TIME;
Linus Torvalds's avatar
Linus Torvalds committed
541
}
542
EXPORT_SYMBOL_GPL(shmem_truncate_range);
Linus Torvalds's avatar
Linus Torvalds committed
543

544
static int shmem_setattr(struct dentry *dentry, struct iattr *attr)
Linus Torvalds's avatar
Linus Torvalds committed
545 546 547 548
{
	struct inode *inode = dentry->d_inode;
	int error;

549 550 551 552
	error = inode_change_ok(inode, attr);
	if (error)
		return error;

553 554 555
	if (S_ISREG(inode->i_mode) && (attr->ia_valid & ATTR_SIZE)) {
		loff_t oldsize = inode->i_size;
		loff_t newsize = attr->ia_size;
556

557 558 559 560 561 562 563 564 565 566 567
		if (newsize != oldsize) {
			i_size_write(inode, newsize);
			inode->i_ctime = inode->i_mtime = CURRENT_TIME;
		}
		if (newsize < oldsize) {
			loff_t holebegin = round_up(newsize, PAGE_SIZE);
			unmap_mapping_range(inode->i_mapping, holebegin, 0, 1);
			shmem_truncate_range(inode, newsize, (loff_t)-1);
			/* unmap again to remove racily COWed private pages */
			unmap_mapping_range(inode->i_mapping, holebegin, 0, 1);
		}
Linus Torvalds's avatar
Linus Torvalds committed
568 569
	}

570
	setattr_copy(inode, attr);
571
#ifdef CONFIG_TMPFS_POSIX_ACL
572
	if (attr->ia_valid & ATTR_MODE)
573
		error = generic_acl_chmod(inode);
574
#endif
Linus Torvalds's avatar
Linus Torvalds committed
575 576 577
	return error;
}

Al Viro's avatar
Al Viro committed
578
static void shmem_evict_inode(struct inode *inode)
Linus Torvalds's avatar
Linus Torvalds committed
579 580
{
	struct shmem_inode_info *info = SHMEM_I(inode);
581
	struct shmem_xattr *xattr, *nxattr;
Linus Torvalds's avatar
Linus Torvalds committed
582

583
	if (inode->i_mapping->a_ops == &shmem_aops) {
Linus Torvalds's avatar
Linus Torvalds committed
584 585
		shmem_unacct_size(info->flags, inode->i_size);
		inode->i_size = 0;
586
		shmem_truncate_range(inode, 0, (loff_t)-1);
Linus Torvalds's avatar
Linus Torvalds committed
587
		if (!list_empty(&info->swaplist)) {
588
			mutex_lock(&shmem_swaplist_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
589
			list_del_init(&info->swaplist);
590
			mutex_unlock(&shmem_swaplist_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
591 592
		}
	}
593 594 595 596 597

	list_for_each_entry_safe(xattr, nxattr, &info->xattr_list, list) {
		kfree(xattr->name);
		kfree(xattr);
	}
598
	BUG_ON(inode->i_blocks);
599
	shmem_free_inode(inode->i_sb);
Al Viro's avatar
Al Viro committed
600
	end_writeback(inode);
Linus Torvalds's avatar
Linus Torvalds committed
601 602
}

603 604 605
/*
 * If swap found in inode, free it and move page from swapcache to filecache.
 */
606 607
static int shmem_unuse_inode(struct shmem_inode_info *info,
			     swp_entry_t swap, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
608
{
609
	struct address_space *mapping = info->vfs_inode.i_mapping;
610
	void *radswap;
611
	pgoff_t index;
612
	int error;
Linus Torvalds's avatar
Linus Torvalds committed
613

614 615 616
	radswap = swp_to_radix_entry(swap);
	index = shmem_find_swap(mapping, radswap);
	if (index == -1)
617
		return 0;
618

Hugh Dickins's avatar
Hugh Dickins committed
619 620
	/*
	 * Move _head_ to start search for next from here.
Al Viro's avatar
Al Viro committed
621
	 * But be careful: shmem_evict_inode checks list_empty without taking
Hugh Dickins's avatar
Hugh Dickins committed
622
	 * mutex, and there's an instant in list_move_tail when info->swaplist
623
	 * would appear empty, if it were the only one on shmem_swaplist.
Hugh Dickins's avatar
Hugh Dickins committed
624 625 626
	 */
	if (shmem_swaplist.next != &info->swaplist)
		list_move_tail(&shmem_swaplist, &info->swaplist);
627

KAMEZAWA Hiroyuki's avatar
KAMEZAWA Hiroyuki committed
628
	/*
629 630 631
	 * We rely on shmem_swaplist_mutex, not only to protect the swaplist,
	 * but also to hold up shmem_evict_inode(): so inode cannot be freed
	 * beneath us (pagelock doesn't help until the page is in pagecache).
KAMEZAWA Hiroyuki's avatar
KAMEZAWA Hiroyuki committed
632
	 */
633 634
	error = shmem_add_to_page_cache(page, mapping, index,
						GFP_NOWAIT, radswap);
635
	/* which does mem_cgroup_uncharge_cache_page on error */
636

637
	if (error != -ENOMEM) {
638 639 640 641
		/*
		 * Truncation and eviction use free_swap_and_cache(), which
		 * only does trylock page: if we raced, best clean up here.
		 */
642 643
		delete_from_swap_cache(page);
		set_page_dirty(page);
644 645 646 647 648 649
		if (!error) {
			spin_lock(&info->lock);
			info->swapped--;
			spin_unlock(&info->lock);
			swap_free(swap);
		}
650
		error = 1;	/* not an error, but entry was found */
Linus Torvalds's avatar
Linus Torvalds committed
651
	}
652
	return error;
Linus Torvalds's avatar
Linus Torvalds committed
653 654 655
}

/*
656
 * Search through swapped inodes to find and replace swap by page.
Linus Torvalds's avatar
Linus Torvalds committed
657
 */
658
int shmem_unuse(swp_entry_t swap, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
659
{
660
	struct list_head *this, *next;
Linus Torvalds's avatar
Linus Torvalds committed
661 662
	struct shmem_inode_info *info;
	int found = 0;
663 664 665 666 667 668
	int error;

	/*
	 * Charge page using GFP_KERNEL while we can wait, before taking
	 * the shmem_swaplist_mutex which might hold up shmem_writepage().
	 * Charged back to the user (not to caller) when swap account is used.
669
	 * shmem_add_to_page_cache() will be called with GFP_NOWAIT.
670 671 672 673
	 */
	error = mem_cgroup_cache_charge(page, current->mm, GFP_KERNEL);
	if (error)
		goto out;
674
	/* No radix_tree_preload: swap entry keeps a place for page in tree */
Linus Torvalds's avatar
Linus Torvalds committed
675

676
	mutex_lock(&shmem_swaplist_mutex);
677 678
	list_for_each_safe(this, next, &shmem_swaplist) {
		info = list_entry(this, struct shmem_inode_info, swaplist);
679 680 681 682 683 684 685
		if (!info->swapped) {
			spin_lock(&info->lock);
			if (!info->swapped)
				list_del_init(&info->swaplist);
			spin_unlock(&info->lock);
		}
		if (info->swapped)
686
			found = shmem_unuse_inode(info, swap, page);
687
		cond_resched();
688
		if (found)
689
			break;
Linus Torvalds's avatar
Linus Torvalds committed
690
	}
691
	mutex_unlock(&shmem_swaplist_mutex);
692 693 694 695 696 697

	if (!found)
		mem_cgroup_uncharge_cache_page(page);
	if (found < 0)
		error = found;
out:
Hugh Dickins's avatar
Hugh Dickins committed
698 699
	unlock_page(page);
	page_cache_release(page);
700
	return error;
Linus Torvalds's avatar
Linus Torvalds committed
701 702 703 704 705 706 707 708
}

/*
 * Move the page from the page cache to the swap cache.
 */
static int shmem_writepage(struct page *page, struct writeback_control *wbc)
{
	struct shmem_inode_info *info;
709
	swp_entry_t swap, oswap;
Linus Torvalds's avatar
Linus Torvalds committed
710
	struct address_space *mapping;
711
	pgoff_t index;
Linus Torvalds's avatar
Linus Torvalds committed
712 713 714 715 716 717 718 719 720
	struct inode *inode;

	BUG_ON(!PageLocked(page));
	mapping = page->mapping;
	index = page->index;
	inode = mapping->host;
	info = SHMEM_I(inode);
	if (info->flags & VM_LOCKED)
		goto redirty;
721
	if (!total_swap_pages)
Linus Torvalds's avatar
Linus Torvalds committed
722 723
		goto redirty;

724 725 726
	/*
	 * shmem_backing_dev_info's capabilities prevent regular writeback or
	 * sync from ever calling shmem_writepage; but a stacking filesystem
727
	 * might use ->writepage of its underlying filesystem, in which case
728
	 * tmpfs should write out to swap only in response to memory pressure,
729
	 * and not for the writeback threads or sync.
730
	 */
731 732 733 734
	if (!wbc->for_reclaim) {
		WARN_ON_ONCE(1);	/* Still happens? Tell us about it! */
		goto redirty;
	}
735 736

	/*
737 738
	 * Disable even the toy swapping implementation, while we convert
	 * functions one by one to having swap entries in the radix tree.
739
	 */
740
	if (index < ULONG_MAX)
741 742
		goto redirty;

743 744 745
	swap = get_swap_page();
	if (!swap.val)
		goto redirty;
746

747 748 749 750 751 752 753 754 755
	/*
	 * Add inode to shmem_unuse()'s list of swapped-out inodes,
	 * if it's not already there.  Do it now because we cannot take
	 * mutex while holding spinlock, and must do so before the page
	 * is moved to swap cache, when its pagelock no longer protects
	 * the inode from eviction.  But don't unlock the mutex until
	 * we've taken the spinlock, because shmem_unuse_inode() will
	 * prune a !swapped inode from the swaplist under both locks.
	 */
756 757 758
	mutex_lock(&shmem_swaplist_mutex);
	if (list_empty(&info->swaplist))
		list_add_tail(&info->swaplist, &shmem_swaplist);
759

Linus Torvalds's avatar
Linus Torvalds committed
760
	spin_lock(&info->lock);
761
	mutex_unlock(&shmem_swaplist_mutex);
762

763 764
	oswap = shmem_get_swap(info, index);
	if (oswap.val) {
765
		WARN_ON_ONCE(1);	/* Still happens? Tell us about it! */
766 767 768
		free_swap_and_cache(oswap);
		shmem_put_swap(info, index, (swp_entry_t){0});
		info->swapped--;
769 770
	}
	shmem_recalc_inode(inode);
Linus Torvalds's avatar
Linus Torvalds committed
771

772
	if (add_to_swap_cache(page, swap, GFP_ATOMIC) == 0) {
773
		delete_from_page_cache(page);
774 775
		shmem_put_swap(info, index, swap);
		info->swapped++;
Hugh Dickins's avatar
Hugh Dickins committed
776
		swap_shmem_alloc(swap);
777
		spin_unlock(&info->lock);
778
		BUG_ON(page_mapped(page));
779
		swap_writepage(page, wbc);
Linus Torvalds's avatar
Linus Torvalds committed
780 781 782 783
		return 0;
	}

	spin_unlock(&info->lock);
784
	swapcache_free(swap, NULL);
Linus Torvalds's avatar
Linus Torvalds committed
785 786
redirty:
	set_page_dirty(page);
787 788 789 790
	if (wbc->for_reclaim)
		return AOP_WRITEPAGE_ACTIVATE;	/* Return with page locked */
	unlock_page(page);
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
791 792 793
}

#ifdef CONFIG_NUMA
794
#ifdef CONFIG_TMPFS
795
static void shmem_show_mpol(struct seq_file *seq, struct mempolicy *mpol)
796
{
797
	char buffer[64];
798

799
	if (!mpol || mpol->mode == MPOL_DEFAULT)
800
		return;		/* show nothing */
801

802
	mpol_to_str(buffer, sizeof(buffer), mpol, 1);
803 804

	seq_printf(seq, ",mpol=%s", buffer);
805
}
806 807 808 809 810 811 812 813 814 815 816 817

static struct mempolicy *shmem_get_sbmpol(struct shmem_sb_info *sbinfo)
{
	struct mempolicy *mpol = NULL;
	if (sbinfo->mpol) {
		spin_lock(&sbinfo->stat_lock);	/* prevent replace/use races */
		mpol = sbinfo->mpol;
		mpol_get(mpol);
		spin_unlock(&sbinfo->stat_lock);
	}
	return mpol;
}
818 819
#endif /* CONFIG_TMPFS */

820 821
static struct page *shmem_swapin(swp_entry_t swap, gfp_t gfp,
			struct shmem_inode_info *info, pgoff_t index)
Linus Torvalds's avatar
Linus Torvalds committed
822
{
823
	struct mempolicy mpol, *spol;
Linus Torvalds's avatar
Linus Torvalds committed
824 825
	struct vm_area_struct pvma;

826
	spol = mpol_cond_copy(&mpol,
827
			mpol_shared_policy_lookup(&info->policy, index));
828

Linus Torvalds's avatar
Linus Torvalds committed
829
	/* Create a pseudo vma that just contains the policy */
830
	pvma.vm_start = 0;
831
	pvma.vm_pgoff = index;
832
	pvma.vm_ops = NULL;
833
	pvma.vm_policy = spol;
834
	return swapin_readahead(swap, gfp, &pvma, 0);
Linus Torvalds's avatar
Linus Torvalds committed
835 836
}

837
static struct page *shmem_alloc_page(gfp_t gfp,
838
			struct shmem_inode_info *info, pgoff_t index)
Linus Torvalds's avatar
Linus Torvalds committed
839 840 841
{
	struct vm_area_struct pvma;

842 843
	/* Create a pseudo vma that just contains the policy */
	pvma.vm_start = 0;
844
	pvma.vm_pgoff = index;
845
	pvma.vm_ops = NULL;
846
	pvma.vm_policy = mpol_shared_policy_lookup(&info->policy, index);
847 848 849 850 851

	/*
	 * alloc_page_vma() will drop the shared policy reference
	 */
	return alloc_page_vma(gfp, &pvma, 0);
Linus Torvalds's avatar
Linus Torvalds committed
852
}
853 854
#else /* !CONFIG_NUMA */
#ifdef CONFIG_TMPFS
855
static inline void shmem_show_mpol(struct seq_file *seq, struct mempolicy *mpol)
856 857 858 859
{
}
#endif /* CONFIG_TMPFS */

860 861
static inline struct page *shmem_swapin(swp_entry_t swap, gfp_t gfp,
			struct shmem_inode_info *info, pgoff_t index)
Linus Torvalds's avatar
Linus Torvalds committed
862
{
863
	return swapin_readahead(swap, gfp, NULL, 0);
Linus Torvalds's avatar
Linus Torvalds committed
864 865
}

866
static inline struct page *shmem_alloc_page(gfp_t gfp,
867
			struct shmem_inode_info *info, pgoff_t index)
Linus Torvalds's avatar
Linus Torvalds committed
868
{
869
	return alloc_page(gfp);
Linus Torvalds's avatar
Linus Torvalds committed
870
}
871
#endif /* CONFIG_NUMA */
Linus Torvalds's avatar
Linus Torvalds committed
872

873 874 875 876 877 878 879
#if !defined(CONFIG_NUMA) || !defined(CONFIG_TMPFS)
static inline struct mempolicy *shmem_get_sbmpol(struct shmem_sb_info *sbinfo)
{
	return NULL;
}
#endif

Linus Torvalds's avatar
Linus Torvalds committed
880
/*
881
 * shmem_getpage_gfp - find page in cache, or get from swap, or allocate
Linus Torvalds's avatar
Linus Torvalds committed
882 883 884 885 886
 *
 * If we allocate a new one we do not mark it dirty. That's up to the
 * vm. If we swap it in we mark it dirty since we also free the swap
 * entry since a page cannot live in both the swap and page cache
 */
887
static int shmem_getpage_gfp(struct inode *inode, pgoff_t index,
888
	struct page **pagep, enum sgp_type sgp, gfp_t gfp, int *fault_type)
Linus Torvalds's avatar
Linus Torvalds committed
889 890 891 892
{
	struct address_space *mapping = inode->i_mapping;
	struct shmem_inode_info *info = SHMEM_I(inode);
	struct shmem_sb_info *sbinfo;
893
	struct page *page;
894
	struct page *prealloc_page = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
895 896 897
	swp_entry_t swap;
	int error;

898
	if (index > (MAX_LFS_FILESIZE >> PAGE_CACHE_SHIFT))
Linus Torvalds's avatar
Linus Torvalds committed
899 900
		return -EFBIG;
repeat:
901
	page = find_lock_page(mapping, index);
902
	if (page) {
Hugh Dickins's avatar
Hugh Dickins committed
903
		/*
904 905 906
		 * Once we can get the page lock, it must be uptodate:
		 * if there were an error in reading back from swap,
		 * the page would not be inserted into the filecache.
Hugh Dickins's avatar
Hugh Dickins committed
907
		 */
908 909 910 911 912 913 914 915 916 917 918 919 920 921
		BUG_ON(!PageUptodate(page));
		goto done;
	}

	/*
	 * Try to preload while we can wait, to not make a habit of
	 * draining atomic reserves; but don't latch on to this cpu.
	 */
	error = radix_tree_preload(gfp & GFP_RECLAIM_MASK);
	if (error)
		goto out;
	radix_tree_preload_end();

	if (sgp != SGP_READ && !prealloc_page) {
922
		prealloc_page = shmem_alloc_page(gfp, info, index);
923 924 925 926 927 928
		if (prealloc_page) {
			SetPageSwapBacked(prealloc_page);
			if (mem_cgroup_cache_charge(prealloc_page,
					current->mm, GFP_KERNEL)) {
				page_cache_release(prealloc_page);
				prealloc_page = NULL;
929 930
			}
		}
Hugh Dickins's avatar
Hugh Dickins committed
931
	}
Linus Torvalds's avatar
Linus Torvalds committed
932 933 934

	spin_lock(&info->lock);
	shmem_recalc_inode(inode);
935
	swap = shmem_get_swap(info, index);
Linus Torvalds's avatar
Linus Torvalds committed
936 937
	if (swap.val) {
		/* Look it up and read it in.. */
938 939
		page = lookup_swap_cache(swap);
		if (!page) {
940
			spin_unlock(&info->lock);
Linus Torvalds's avatar
Linus Torvalds committed
941
			/* here we actually do the io */
942 943
			if (fault_type)
				*fault_type |= VM_FAULT_MAJOR;
944
			page = shmem_swapin(swap, gfp, info, index);
945
			if (!page) {
946
				swp_entry_t nswap = shmem_get_swap(info, index);
947 948
				if (nswap.val == swap.val) {
					error = -ENOMEM;
949
					goto out;
950
				}
Linus Torvalds's avatar
Linus Torvalds committed
951 952
				goto repeat;
			}
953 954
			wait_on_page_locked(page);
			page_cache_release(page);
Linus Torvalds's avatar
Linus Torvalds committed
955 956 957 958
			goto repeat;
		}

		/* We have to do this with page locked to prevent races */
959
		if (!trylock_page(page)) {
Linus Torvalds's avatar
Linus Torvalds committed
960
			spin_unlock(&info->lock);
961 962
			wait_on_page_locked(page);
			page_cache_release(page);
Linus Torvalds's avatar
Linus Torvalds committed
963 964
			goto repeat;
		}
965
		if (PageWriteback(page)) {
Linus Torvalds's avatar
Linus Torvalds committed
966
			spin_unlock(&info->lock);
967 968 969
			wait_on_page_writeback(page);
			unlock_page(page);
			page_cache_release(page);
Linus Torvalds's avatar
Linus Torvalds committed
970 971
			goto repeat;
		}
972
		if (!PageUptodate(page)) {
Linus Torvalds's avatar
Linus Torvalds committed
973
			spin_unlock(&info->lock);
974 975
			unlock_page(page);
			page_cache_release(page);
Linus Torvalds's avatar
Linus Torvalds committed
976
			error = -EIO;
977
			goto out;
Linus Torvalds's avatar
Linus Torvalds committed
978 979
		}

980
		error = add_to_page_cache_locked(page, mapping,
981
						 index, GFP_NOWAIT);
982
		if (error) {
Linus Torvalds's avatar
Linus Torvalds committed
983
			spin_unlock(&info->lock);
984
			if (error == -ENOMEM) {
985 986 987 988 989
				/*
				 * reclaim from proper memory cgroup and
				 * call memcg's OOM if needed.
				 */
				error = mem_cgroup_shmem_charge_fallback(
990
						page, current->mm, gfp);
991
				if (error) {
992 993 994
					unlock_page(page);
					page_cache_release(page);
					goto out;
995
				}
996
			}
997 998
			unlock_page(page);
			page_cache_release(page);
Linus Torvalds's avatar
Linus Torvalds committed
999 1000
			goto repeat;
		}
1001 1002

		delete_from_swap_cache(page);
1003
		shmem_put_swap(info, index, (swp_entry_t){0});
1004
		info->swapped--;
1005 1006 1007 1008 1009
		spin_unlock(&info->lock);
		set_page_dirty(page);
		swap_free(swap);

	} else if (sgp == SGP_READ) {
1010
		page = find_get_page(mapping, index);
1011
		if (page && !trylock_page(page)) {
Linus Torvalds's avatar
Linus Torvalds committed
1012
			spin_unlock(&info->lock);
1013 1014
			wait_on_page_locked(page);
			page_cache_release(page);
Linus Torvalds's avatar
Linus Torvalds committed
1015 1016 1017
			goto repeat;
		}
		spin_unlock(&info->lock);
Hugh Dickins's avatar
Hugh Dickins committed
1018 1019

	} else if (prealloc_page) {
Linus Torvalds's avatar
Linus Torvalds committed
1020
		sbinfo = SHMEM_SB(inode->i_sb);
1021
		if (sbinfo->max_blocks) {
1022 1023
			if (percpu_counter_compare(&sbinfo->used_blocks,
						sbinfo->max_blocks) >= 0 ||
1024 1025
			    shmem_acct_block(info->flags))
				goto nospace;
1026
			percpu_counter_inc(&sbinfo->used_blocks);