shmem.c 63.5 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) 2002-2011 Hugh Dickins.
 * Copyright (C) 2011 Google Inc.
11
 * Copyright (C) 2002-2005 VERITAS Software Corporation.
Linus Torvalds's avatar
Linus Torvalds committed
12 13 14 15 16 17
 * 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>
 *
18 19 20
 * tiny-shmem:
 * Copyright (c) 2004, 2008 Matt Mackall <mpm@selenic.com>
 *
Linus Torvalds's avatar
Linus Torvalds committed
21 22 23
 * This file is released under the GPL.
 */

24 25 26 27
#include <linux/fs.h>
#include <linux/init.h>
#include <linux/vfs.h>
#include <linux/mount.h>
Hugh Dickins's avatar
Hugh Dickins committed
28
#include <linux/pagemap.h>
29 30 31 32 33 34 35 36
#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
37 38 39 40 41 42
/*
 * 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.
 */

43
#include <linux/xattr.h>
44
#include <linux/exportfs.h>
45
#include <linux/posix_acl.h>
46
#include <linux/generic_acl.h>
Linus Torvalds's avatar
Linus Torvalds committed
47 48 49 50 51 52 53
#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>
54
#include <linux/pagevec.h>
55
#include <linux/percpu_counter.h>
56
#include <linux/splice.h>
Linus Torvalds's avatar
Linus Torvalds committed
57 58 59 60
#include <linux/security.h>
#include <linux/swapops.h>
#include <linux/mempolicy.h>
#include <linux/namei.h>
61
#include <linux/ctype.h>
62
#include <linux/migrate.h>
63
#include <linux/highmem.h>
64
#include <linux/seq_file.h>
Mimi Zohar's avatar
Mimi Zohar committed
65
#include <linux/magic.h>
66

Linus Torvalds's avatar
Linus Torvalds committed
67 68 69
#include <asm/uaccess.h>
#include <asm/pgtable.h>

Hugh Dickins's avatar
Hugh Dickins committed
70
#define BLOCKS_PER_PAGE  (PAGE_CACHE_SIZE/512)
Linus Torvalds's avatar
Linus Torvalds committed
71 72 73 74 75
#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

76 77 78
/* Symlink up to this size is kmalloc'ed instead of using a swappable page */
#define SHORT_SYMLINK_LEN 128

79 80 81 82 83 84 85
struct shmem_xattr {
	struct list_head list;	/* anchored by shmem_inode_info->xattr_list */
	char *name;		/* xattr name */
	size_t size;
	char value[0];
};

86
/* Flag allocation requirements to shmem_getpage */
Linus Torvalds's avatar
Linus Torvalds committed
87 88 89
enum sgp_type {
	SGP_READ,	/* don't exceed i_size, don't allocate page */
	SGP_CACHE,	/* don't exceed i_size, may allocate page */
90
	SGP_DIRTY,	/* like SGP_CACHE, but set new page dirty */
Linus Torvalds's avatar
Linus Torvalds committed
91 92 93
	SGP_WRITE,	/* may exceed i_size, may allocate page */
};

Andrew Morton's avatar
Andrew Morton committed
94
#ifdef CONFIG_TMPFS
95 96 97 98 99 100 101 102 103
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
104
#endif
105

106 107 108 109 110 111 112 113 114
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
115 116 117 118 119 120 121 122 123 124 125 126 127 128

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)
{
129 130
	return (flags & VM_NORESERVE) ?
		0 : security_vm_enough_memory_kern(VM_ACCT(size));
Linus Torvalds's avatar
Linus Torvalds committed
131 132 133 134
}

static inline void shmem_unacct_size(unsigned long flags, loff_t size)
{
135
	if (!(flags & VM_NORESERVE))
Linus Torvalds's avatar
Linus Torvalds committed
136 137 138 139 140 141 142 143 144 145 146
		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)
{
147 148
	return (flags & VM_NORESERVE) ?
		security_vm_enough_memory_kern(VM_ACCT(PAGE_CACHE_SIZE)) : 0;
Linus Torvalds's avatar
Linus Torvalds committed
149 150 151 152
}

static inline void shmem_unacct_blocks(unsigned long flags, long pages)
{
153
	if (flags & VM_NORESERVE)
Linus Torvalds's avatar
Linus Torvalds committed
154 155 156
		vm_unacct_memory(pages * VM_ACCT(PAGE_CACHE_SIZE));
}

157
static const struct super_operations shmem_ops;
158
static const struct address_space_operations shmem_aops;
159
static const struct file_operations shmem_file_operations;
160 161 162
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;
163
static const struct vm_operations_struct shmem_vm_ops;
Linus Torvalds's avatar
Linus Torvalds committed
164

165
static struct backing_dev_info shmem_backing_dev_info  __read_mostly = {
Linus Torvalds's avatar
Linus Torvalds committed
166
	.ra_pages	= 0,	/* No readahead */
167
	.capabilities	= BDI_CAP_NO_ACCT_AND_WRITEBACK | BDI_CAP_SWAP_BACKED,
Linus Torvalds's avatar
Linus Torvalds committed
168 169 170
};

static LIST_HEAD(shmem_swaplist);
171
static DEFINE_MUTEX(shmem_swaplist_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
172

173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197
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);
	}
}

198
/**
199
 * shmem_recalc_inode - recalculate the block usage of an inode
Linus Torvalds's avatar
Linus Torvalds committed
200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216
 * @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) {
217 218 219
		struct shmem_sb_info *sbinfo = SHMEM_SB(inode->i_sb);
		if (sbinfo->max_blocks)
			percpu_counter_add(&sbinfo->used_blocks, -freed);
Linus Torvalds's avatar
Linus Torvalds committed
220
		info->alloced -= freed;
221
		inode->i_blocks -= freed * BLOCKS_PER_PAGE;
Linus Torvalds's avatar
Linus Torvalds committed
222 223 224 225
		shmem_unacct_blocks(info->flags, freed);
	}
}

226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248
/*
 * 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;
}

249 250 251 252 253 254 255
/*
 * 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)
{
256
	int error = 0;
257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292

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

	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);
	return error;
}

293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311
/*
 * Like delete_from_page_cache, but substitutes swap for page.
 */
static void shmem_delete_from_page_cache(struct page *page, void *radswap)
{
	struct address_space *mapping = page->mapping;
	int error;

	spin_lock_irq(&mapping->tree_lock);
	error = shmem_radix_tree_replace(mapping, page->index, page, radswap);
	page->mapping = NULL;
	mapping->nrpages--;
	__dec_zone_page_state(page, NR_FILE_PAGES);
	__dec_zone_page_state(page, NR_SHMEM);
	spin_unlock_irq(&mapping->tree_lock);
	page_cache_release(page);
	BUG_ON(error);
}

312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393
/*
 * 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;
}

/*
 * 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.
 */
394
void shmem_truncate_range(struct inode *inode, loff_t lstart, loff_t lend)
Linus Torvalds's avatar
Linus Torvalds committed
395
{
396
	struct address_space *mapping = inode->i_mapping;
Linus Torvalds's avatar
Linus Torvalds committed
397
	struct shmem_inode_info *info = SHMEM_I(inode);
398
	pgoff_t start = (lstart + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
399
	unsigned partial = lstart & (PAGE_CACHE_SIZE - 1);
400
	pgoff_t end = (lend >> PAGE_CACHE_SHIFT);
401
	struct pagevec pvec;
402 403
	pgoff_t indices[PAGEVEC_SIZE];
	long nr_swaps_freed = 0;
404
	pgoff_t index;
405 406 407 408 409 410
	int i;

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

	pagevec_init(&pvec, 0);
	index = start;
411 412 413 414 415 416
	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;
417 418 419 420
		mem_cgroup_uncharge_start();
		for (i = 0; i < pagevec_count(&pvec); i++) {
			struct page *page = pvec.pages[i];

421
			index = indices[i];
422 423 424
			if (index > end)
				break;

425 426 427
			if (radix_tree_exceptional_entry(page)) {
				nr_swaps_freed += !shmem_free_swap(mapping,
								index, page);
428
				continue;
429 430 431
			}

			if (!trylock_page(page))
432
				continue;
433 434 435
			if (page->mapping == mapping) {
				VM_BUG_ON(PageWriteback(page));
				truncate_inode_page(mapping, page);
436 437 438
			}
			unlock_page(page);
		}
439
		shmem_pagevec_release(&pvec);
440 441 442 443
		mem_cgroup_uncharge_end();
		cond_resched();
		index++;
	}
Linus Torvalds's avatar
Linus Torvalds committed
444

445 446 447 448 449 450 451 452 453 454 455 456 457 458
	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();
459 460 461 462
		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) {
463 464 465 466 467
			if (index == start)
				break;
			index = start;
			continue;
		}
468 469
		if (index == start && indices[0] > end) {
			shmem_pagevec_release(&pvec);
470 471 472 473 474 475
			break;
		}
		mem_cgroup_uncharge_start();
		for (i = 0; i < pagevec_count(&pvec); i++) {
			struct page *page = pvec.pages[i];

476
			index = indices[i];
477 478 479
			if (index > end)
				break;

480 481 482 483 484 485
			if (radix_tree_exceptional_entry(page)) {
				nr_swaps_freed += !shmem_free_swap(mapping,
								index, page);
				continue;
			}

486
			lock_page(page);
487 488 489 490
			if (page->mapping == mapping) {
				VM_BUG_ON(PageWriteback(page));
				truncate_inode_page(mapping, page);
			}
491 492
			unlock_page(page);
		}
493
		shmem_pagevec_release(&pvec);
494 495 496
		mem_cgroup_uncharge_end();
		index++;
	}
497

Linus Torvalds's avatar
Linus Torvalds committed
498
	spin_lock(&info->lock);
499
	info->swapped -= nr_swaps_freed;
Linus Torvalds's avatar
Linus Torvalds committed
500 501 502
	shmem_recalc_inode(inode);
	spin_unlock(&info->lock);

503
	inode->i_ctime = inode->i_mtime = CURRENT_TIME;
Linus Torvalds's avatar
Linus Torvalds committed
504
}
505
EXPORT_SYMBOL_GPL(shmem_truncate_range);
Linus Torvalds's avatar
Linus Torvalds committed
506

507
static int shmem_setattr(struct dentry *dentry, struct iattr *attr)
Linus Torvalds's avatar
Linus Torvalds committed
508 509 510 511
{
	struct inode *inode = dentry->d_inode;
	int error;

512 513 514 515
	error = inode_change_ok(inode, attr);
	if (error)
		return error;

516 517 518
	if (S_ISREG(inode->i_mode) && (attr->ia_valid & ATTR_SIZE)) {
		loff_t oldsize = inode->i_size;
		loff_t newsize = attr->ia_size;
519

520 521 522 523 524 525 526 527 528 529 530
		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
531 532
	}

533
	setattr_copy(inode, attr);
534
#ifdef CONFIG_TMPFS_POSIX_ACL
535
	if (attr->ia_valid & ATTR_MODE)
536
		error = generic_acl_chmod(inode);
537
#endif
Linus Torvalds's avatar
Linus Torvalds committed
538 539 540
	return error;
}

Al Viro's avatar
Al Viro committed
541
static void shmem_evict_inode(struct inode *inode)
Linus Torvalds's avatar
Linus Torvalds committed
542 543
{
	struct shmem_inode_info *info = SHMEM_I(inode);
544
	struct shmem_xattr *xattr, *nxattr;
Linus Torvalds's avatar
Linus Torvalds committed
545

546
	if (inode->i_mapping->a_ops == &shmem_aops) {
Linus Torvalds's avatar
Linus Torvalds committed
547 548
		shmem_unacct_size(info->flags, inode->i_size);
		inode->i_size = 0;
549
		shmem_truncate_range(inode, 0, (loff_t)-1);
Linus Torvalds's avatar
Linus Torvalds committed
550
		if (!list_empty(&info->swaplist)) {
551
			mutex_lock(&shmem_swaplist_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
552
			list_del_init(&info->swaplist);
553
			mutex_unlock(&shmem_swaplist_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
554
		}
555 556
	} else
		kfree(info->symlink);
557 558 559 560 561

	list_for_each_entry_safe(xattr, nxattr, &info->xattr_list, list) {
		kfree(xattr->name);
		kfree(xattr);
	}
562
	BUG_ON(inode->i_blocks);
563
	shmem_free_inode(inode->i_sb);
Al Viro's avatar
Al Viro committed
564
	end_writeback(inode);
Linus Torvalds's avatar
Linus Torvalds committed
565 566
}

567 568 569
/*
 * If swap found in inode, free it and move page from swapcache to filecache.
 */
570 571
static int shmem_unuse_inode(struct shmem_inode_info *info,
			     swp_entry_t swap, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
572
{
573
	struct address_space *mapping = info->vfs_inode.i_mapping;
574
	void *radswap;
575
	pgoff_t index;
576
	int error;
Linus Torvalds's avatar
Linus Torvalds committed
577

578
	radswap = swp_to_radix_entry(swap);
579
	index = radix_tree_locate_item(&mapping->page_tree, radswap);
580
	if (index == -1)
581
		return 0;
582

Hugh Dickins's avatar
Hugh Dickins committed
583 584
	/*
	 * Move _head_ to start search for next from here.
Al Viro's avatar
Al Viro committed
585
	 * But be careful: shmem_evict_inode checks list_empty without taking
Hugh Dickins's avatar
Hugh Dickins committed
586
	 * mutex, and there's an instant in list_move_tail when info->swaplist
587
	 * would appear empty, if it were the only one on shmem_swaplist.
Hugh Dickins's avatar
Hugh Dickins committed
588 589 590
	 */
	if (shmem_swaplist.next != &info->swaplist)
		list_move_tail(&shmem_swaplist, &info->swaplist);
591

KAMEZAWA Hiroyuki's avatar
KAMEZAWA Hiroyuki committed
592
	/*
593 594 595
	 * 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
596
	 */
597 598
	error = shmem_add_to_page_cache(page, mapping, index,
						GFP_NOWAIT, radswap);
599
	/* which does mem_cgroup_uncharge_cache_page on error */
600

601
	if (error != -ENOMEM) {
602 603 604 605
		/*
		 * Truncation and eviction use free_swap_and_cache(), which
		 * only does trylock page: if we raced, best clean up here.
		 */
606 607
		delete_from_swap_cache(page);
		set_page_dirty(page);
608 609 610 611 612 613
		if (!error) {
			spin_lock(&info->lock);
			info->swapped--;
			spin_unlock(&info->lock);
			swap_free(swap);
		}
614
		error = 1;	/* not an error, but entry was found */
Linus Torvalds's avatar
Linus Torvalds committed
615
	}
616
	return error;
Linus Torvalds's avatar
Linus Torvalds committed
617 618 619
}

/*
620
 * Search through swapped inodes to find and replace swap by page.
Linus Torvalds's avatar
Linus Torvalds committed
621
 */
622
int shmem_unuse(swp_entry_t swap, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
623
{
624
	struct list_head *this, *next;
Linus Torvalds's avatar
Linus Torvalds committed
625 626
	struct shmem_inode_info *info;
	int found = 0;
627 628 629 630 631 632 633 634 635 636
	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.
	 */
	error = mem_cgroup_cache_charge(page, current->mm, GFP_KERNEL);
	if (error)
		goto out;
637
	/* No radix_tree_preload: swap entry keeps a place for page in tree */
Linus Torvalds's avatar
Linus Torvalds committed
638

639
	mutex_lock(&shmem_swaplist_mutex);
640 641
	list_for_each_safe(this, next, &shmem_swaplist) {
		info = list_entry(this, struct shmem_inode_info, swaplist);
642
		if (info->swapped)
643
			found = shmem_unuse_inode(info, swap, page);
644 645
		else
			list_del_init(&info->swaplist);
646
		cond_resched();
647
		if (found)
648
			break;
Linus Torvalds's avatar
Linus Torvalds committed
649
	}
650
	mutex_unlock(&shmem_swaplist_mutex);
651 652 653 654 655 656

	if (!found)
		mem_cgroup_uncharge_cache_page(page);
	if (found < 0)
		error = found;
out:
Hugh Dickins's avatar
Hugh Dickins committed
657 658
	unlock_page(page);
	page_cache_release(page);
659
	return error;
Linus Torvalds's avatar
Linus Torvalds committed
660 661 662 663 664 665 666 667 668 669
}

/*
 * 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;
	struct address_space *mapping;
	struct inode *inode;
670 671
	swp_entry_t swap;
	pgoff_t index;
Linus Torvalds's avatar
Linus Torvalds committed
672 673 674 675 676 677 678 679

	BUG_ON(!PageLocked(page));
	mapping = page->mapping;
	index = page->index;
	inode = mapping->host;
	info = SHMEM_I(inode);
	if (info->flags & VM_LOCKED)
		goto redirty;
680
	if (!total_swap_pages)
Linus Torvalds's avatar
Linus Torvalds committed
681 682
		goto redirty;

683 684 685
	/*
	 * shmem_backing_dev_info's capabilities prevent regular writeback or
	 * sync from ever calling shmem_writepage; but a stacking filesystem
686
	 * might use ->writepage of its underlying filesystem, in which case
687
	 * tmpfs should write out to swap only in response to memory pressure,
688
	 * and not for the writeback threads or sync.
689
	 */
690 691 692 693 694 695 696
	if (!wbc->for_reclaim) {
		WARN_ON_ONCE(1);	/* Still happens? Tell us about it! */
		goto redirty;
	}
	swap = get_swap_page();
	if (!swap.val)
		goto redirty;
697

698 699
	/*
	 * Add inode to shmem_unuse()'s list of swapped-out inodes,
700 701
	 * if it's not already there.  Do it now before the page is
	 * moved to swap cache, when its pagelock no longer protects
702
	 * the inode from eviction.  But don't unlock the mutex until
703 704
	 * we've incremented swapped, because shmem_unuse_inode() will
	 * prune a !swapped inode from the swaplist under this mutex.
705
	 */
706 707 708
	mutex_lock(&shmem_swaplist_mutex);
	if (list_empty(&info->swaplist))
		list_add_tail(&info->swaplist, &shmem_swaplist);
709

710
	if (add_to_swap_cache(page, swap, GFP_ATOMIC) == 0) {
Hugh Dickins's avatar
Hugh Dickins committed
711
		swap_shmem_alloc(swap);
712 713 714 715 716
		shmem_delete_from_page_cache(page, swp_to_radix_entry(swap));

		spin_lock(&info->lock);
		info->swapped++;
		shmem_recalc_inode(inode);
717
		spin_unlock(&info->lock);
718 719

		mutex_unlock(&shmem_swaplist_mutex);
720
		BUG_ON(page_mapped(page));
721
		swap_writepage(page, wbc);
Linus Torvalds's avatar
Linus Torvalds committed
722 723 724
		return 0;
	}

725
	mutex_unlock(&shmem_swaplist_mutex);
726
	swapcache_free(swap, NULL);
Linus Torvalds's avatar
Linus Torvalds committed
727 728
redirty:
	set_page_dirty(page);
729 730 731 732
	if (wbc->for_reclaim)
		return AOP_WRITEPAGE_ACTIVATE;	/* Return with page locked */
	unlock_page(page);
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
733 734 735
}

#ifdef CONFIG_NUMA
736
#ifdef CONFIG_TMPFS
737
static void shmem_show_mpol(struct seq_file *seq, struct mempolicy *mpol)
738
{
739
	char buffer[64];
740

741
	if (!mpol || mpol->mode == MPOL_DEFAULT)
742
		return;		/* show nothing */
743

744
	mpol_to_str(buffer, sizeof(buffer), mpol, 1);
745 746

	seq_printf(seq, ",mpol=%s", buffer);
747
}
748 749 750 751 752 753 754 755 756 757 758 759

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;
}
760 761
#endif /* CONFIG_TMPFS */

762 763
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
764
{
765
	struct mempolicy mpol, *spol;
Linus Torvalds's avatar
Linus Torvalds committed
766 767
	struct vm_area_struct pvma;

768
	spol = mpol_cond_copy(&mpol,
769
			mpol_shared_policy_lookup(&info->policy, index));
770

Linus Torvalds's avatar
Linus Torvalds committed
771
	/* Create a pseudo vma that just contains the policy */
772
	pvma.vm_start = 0;
773
	pvma.vm_pgoff = index;
774
	pvma.vm_ops = NULL;
775
	pvma.vm_policy = spol;
776
	return swapin_readahead(swap, gfp, &pvma, 0);
Linus Torvalds's avatar
Linus Torvalds committed
777 778
}

779
static struct page *shmem_alloc_page(gfp_t gfp,
780
			struct shmem_inode_info *info, pgoff_t index)
Linus Torvalds's avatar
Linus Torvalds committed
781 782 783
{
	struct vm_area_struct pvma;

784 785
	/* Create a pseudo vma that just contains the policy */
	pvma.vm_start = 0;
786
	pvma.vm_pgoff = index;
787
	pvma.vm_ops = NULL;
788
	pvma.vm_policy = mpol_shared_policy_lookup(&info->policy, index);
789 790 791 792 793

	/*
	 * alloc_page_vma() will drop the shared policy reference
	 */
	return alloc_page_vma(gfp, &pvma, 0);
Linus Torvalds's avatar
Linus Torvalds committed
794
}
795 796
#else /* !CONFIG_NUMA */
#ifdef CONFIG_TMPFS
797
static inline void shmem_show_mpol(struct seq_file *seq, struct mempolicy *mpol)
798 799 800 801
{
}
#endif /* CONFIG_TMPFS */

802 803
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
804
{
805
	return swapin_readahead(swap, gfp, NULL, 0);
Linus Torvalds's avatar
Linus Torvalds committed
806 807
}

808
static inline struct page *shmem_alloc_page(gfp_t gfp,
809
			struct shmem_inode_info *info, pgoff_t index)
Linus Torvalds's avatar
Linus Torvalds committed
810
{
811
	return alloc_page(gfp);
Linus Torvalds's avatar
Linus Torvalds committed
812
}
813
#endif /* CONFIG_NUMA */
Linus Torvalds's avatar
Linus Torvalds committed
814

815 816 817 818 819 820 821
#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
822
/*
823
 * shmem_getpage_gfp - find page in cache, or get from swap, or allocate
Linus Torvalds's avatar
Linus Torvalds committed
824 825 826 827 828
 *
 * 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
 */
829
static int shmem_getpage_gfp(struct inode *inode, pgoff_t index,
830
	struct page **pagep, enum sgp_type sgp, gfp_t gfp, int *fault_type)
Linus Torvalds's avatar
Linus Torvalds committed
831 832
{
	struct address_space *mapping = inode->i_mapping;
833
	struct shmem_inode_info *info;
Linus Torvalds's avatar
Linus Torvalds committed
834
	struct shmem_sb_info *sbinfo;
835
	struct page *page;
Linus Torvalds's avatar
Linus Torvalds committed
836 837
	swp_entry_t swap;
	int error;
838
	int once = 0;
Linus Torvalds's avatar
Linus Torvalds committed
839

840
	if (index > (MAX_LFS_FILESIZE >> PAGE_CACHE_SHIFT))
Linus Torvalds's avatar
Linus Torvalds committed
841 842
		return -EFBIG;
repeat:
843
	swap.val = 0;
844
	page = find_lock_page(mapping, index);
845 846 847 848 849 850 851 852 853 854 855 856
	if (radix_tree_exceptional_entry(page)) {
		swap = radix_to_swp_entry(page);
		page = NULL;
	}

	if (sgp != SGP_WRITE &&
	    ((loff_t)index << PAGE_CACHE_SHIFT) >= i_size_read(inode)) {
		error = -EINVAL;
		goto failed;
	}

	if (page || (sgp == SGP_READ && !swap.val)) {
Hugh Dickins's avatar
Hugh Dickins committed
857
		/*
858 859 860
		 * 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
861
		 */
862 863 864
		BUG_ON(page && !PageUptodate(page));
		*pagep = page;
		return 0;
865 866 867
	}

	/*
868 869
	 * Fast cache lookup did not find it:
	 * bring it back from swap or allocate.
870
	 */
871 872
	info = SHMEM_I(inode);
	sbinfo = SHMEM_SB(inode->i_sb);
Linus Torvalds's avatar
Linus Torvalds committed
873 874 875

	if (swap.val) {
		/* Look it up and read it in.. */
876 877
		page = lookup_swap_cache(swap);
		if (!page) {
Linus Torvalds's avatar
Linus Torvalds committed
878
			/* here we actually do the io */
879 880
			if (fault_type)
				*fault_type |= VM_FAULT_MAJOR;
881
			page = shmem_swapin(swap, gfp, info, index);
882
			if (!page) {
883 884
				error = -ENOMEM;
				goto failed;
Linus Torvalds's avatar
Linus Torvalds committed
885 886 887 888
			}
		}

		/* We have to do this with page locked to prevent races */
889
		lock_page(page);
890
		if (!PageUptodate(page)) {
Linus Torvalds's avatar
Linus Torvalds committed
891
			error = -EIO;
892
			goto failed;
Linus Torvalds's avatar
Linus Torvalds committed
893
		}
894 895 896 897 898 899 900 901 902
		wait_on_page_writeback(page);

		/* Someone may have already done it for us */
		if (page->mapping) {
			if (page->mapping == mapping &&
			    page->index == index)
				goto done;
			error = -EEXIST;
			goto failed;
Linus Torvalds's avatar
Linus Torvalds committed
903
		}
904

905 906 907 908 909
		error = mem_cgroup_cache_charge(page, current->mm,
						gfp & GFP_RECLAIM_MASK);
		if (!error)
			error = shmem_add_to_page_cache(page, mapping, index,
						gfp, swp_to_radix_entry(swap));
910 911 912 913
		if (error)
			goto failed;

		spin_lock(&info->lock);
914
		info->swapped--;
915
		shmem_recalc_inode(inode);
916
		spin_unlock(&info->lock);
917 918

		delete_from_swap_cache(page);
919 920 921
		set_page_dirty(page);
		swap_free(swap);

922 923 924 925
	} else {
		if (shmem_acct_block(info->flags)) {
			error = -ENOSPC;
			goto failed;
Linus Torvalds's avatar
Linus Torvalds committed
926
		}
927
		if (sbinfo->max_blocks) {
928
			if (percpu_counter_compare(&sbinfo->used_blocks,
929 930 931 932
						sbinfo->max_blocks) >= 0) {
				error = -ENOSPC;
				goto unacct;
			}
933
			percpu_counter_inc(&sbinfo->used_blocks);
934
		}
Linus Torvalds's avatar
Linus Torvalds committed
935

936 937 938 939
		page = shmem_alloc_page(gfp, info, index);
		if (!page) {
			error = -ENOMEM;
			goto decused;
Linus Torvalds's avatar
Linus Torvalds committed
940 941
		}

942 943
		SetPageSwapBacked(page);
		__set_page_locked(page);
944 945 946 947 948
		error = mem_cgroup_cache_charge(page, current->mm,
						gfp & GFP_RECLAIM_MASK);
		if (!error)
			error = shmem_add_to_page_cache(page, mapping, index,
						gfp, NULL);
949 950 951 952 953
		if (error)
			goto decused;
		lru_cache_add_anon(page);

		spin_lock(&info->lock);
Linus Torvalds's avatar
Linus Torvalds committed
954
		info->alloced++;
955 956
		inode->i_blocks += BLOCKS_PER_PAGE;
		shmem_recalc_inode(inode);
Linus Torvalds's avatar
Linus Torvalds committed
957
		spin_unlock(&info->lock);
958

959 960 961
		clear_highpage(page);
		flush_dcache_page(page);
		SetPageUptodate(page);
962
		if (sgp == SGP_DIRTY)
963
			set_page_dirty(page);
Linus Torvalds's avatar
Linus Torvalds committed
964 965
	}
done:
966 967 968 969 970
	/* Perhaps the file has been truncated since we checked */
	if (sgp != SGP_WRITE &&
	    ((loff_t)index << PAGE_CACHE_SHIFT) >= i_size_read(inode)) {
		error = -EINVAL;
		goto trunc;
Hugh Dickins's avatar
Hugh Dickins committed
971
	}
972 973
	*pagep = page;
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
974

975
	/*
976
	 * Error recovery.
977
	 */
978 979 980 981 982 983
trunc:
	ClearPageDirty(page);
	delete_from_page_cache(page);
	spin_lock(&info->lock);
	info->alloced--;
	inode->i_blocks -= BLOCKS_PER_PAGE;
984
	spin_unlock(&info->lock);
985 986 987 988 989 990 991 992 993 994 995 996 997 998
decused:
	if (sbinfo->max_blocks)
		percpu_counter_add(&sbinfo->used_blocks, -1);
unacct:
	shmem_unacct_blocks(info->flags, 1);
failed:
	if (swap.val && error != -EINVAL) {
		struct page *test = find_get_page(mapping, index);
		if (test && !radix_tree_exceptional_entry(test))
			page_cache_release(test);
		/* Have another try if the entry has changed */
		if (test != swp_to_radix_entry(swap))
			error = -EEXIST;
	}
999
	if (page) {
1000
		unlock_page(page);
1001
		page_cache_release(page);
1002 1003 1004 1005 1006 1007
	}
	if (error == -ENOSPC && !once++) {
		info = SHMEM_I(inode);
		spin_lock(&info->lock);
		shmem_recalc_inode(inode);
		spin_unlock(&info->lock);
1008
		goto repeat;
1009
	}
1010 1011 1012
	if (error == -EEXIST)
		goto repeat;
	return error;
Linus Torvalds's avatar
Linus Torvalds committed
1013 1014
}

Nick Piggin's avatar
Nick Piggin committed
1015
static int shmem_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
Linus Torvalds's avatar
Linus Torvalds committed
1016
{
1017
	struct inode *inode = vma->vm_file->f_path.dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
1018
	int error;