shmem.c 65.7 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>
28
#include <linux/pagemap.h>
29 30
#include <linux/file.h>
#include <linux/mm.h>
31
#include <linux/export.h>
32 33 34 35 36
#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>

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
	return (flags & VM_NORESERVE) ?
130
		0 : security_vm_enough_memory_mm(current->mm, 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
	return (flags & VM_NORESERVE) ?
148
		security_vm_enough_memory_mm(current->mm, 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
/*
 * 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)) {
335 336 337 338 339 340 341 342
			if (radix_tree_deref_retry(page))
				goto restart;
			/*
			 * Otherwise, we must be storing a swap entry
			 * here as an exceptional entry: so return it
			 * without attempting to raise page count.
			 */
			goto export;
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
		}
		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.
 */
382
static void shmem_deswap_pagevec(struct pagevec *pvec)
383 384 385 386 387 388 389 390 391
{
	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;
392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421
}

/*
 * SysV IPC SHM_UNLOCK restore Unevictable pages to their evictable lists.
 */
void shmem_unlock_mapping(struct address_space *mapping)
{
	struct pagevec pvec;
	pgoff_t indices[PAGEVEC_SIZE];
	pgoff_t index = 0;

	pagevec_init(&pvec, 0);
	/*
	 * Minor point, but we might as well stop if someone else SHM_LOCKs it.
	 */
	while (!mapping_unevictable(mapping)) {
		/*
		 * Avoid pagevec_lookup(): find_get_pages() returns 0 as if it
		 * has finished, if it hits a row of PAGEVEC_SIZE swap entries.
		 */
		pvec.nr = shmem_find_get_pages_and_swap(mapping, index,
					PAGEVEC_SIZE, pvec.pages, indices);
		if (!pvec.nr)
			break;
		index = indices[pvec.nr - 1] + 1;
		shmem_deswap_pagevec(&pvec);
		check_move_unevictable_pages(pvec.pages, pvec.nr);
		pagevec_release(&pvec);
		cond_resched();
	}
422 423 424 425 426
}

/*
 * Remove range of pages and swap entries from radix tree, and free them.
 */
427
void shmem_truncate_range(struct inode *inode, loff_t lstart, loff_t lend)
Linus Torvalds's avatar
Linus Torvalds committed
428
{
429
	struct address_space *mapping = inode->i_mapping;
Linus Torvalds's avatar
Linus Torvalds committed
430
	struct shmem_inode_info *info = SHMEM_I(inode);
431
	pgoff_t start = (lstart + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
432
	unsigned partial = lstart & (PAGE_CACHE_SIZE - 1);
433
	pgoff_t end = (lend >> PAGE_CACHE_SHIFT);
434
	struct pagevec pvec;
435 436
	pgoff_t indices[PAGEVEC_SIZE];
	long nr_swaps_freed = 0;
437
	pgoff_t index;
438 439 440 441 442 443
	int i;

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

	pagevec_init(&pvec, 0);
	index = start;
444 445 446 447 448 449
	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;
450 451 452 453
		mem_cgroup_uncharge_start();
		for (i = 0; i < pagevec_count(&pvec); i++) {
			struct page *page = pvec.pages[i];

454
			index = indices[i];
455 456 457
			if (index > end)
				break;

458 459 460
			if (radix_tree_exceptional_entry(page)) {
				nr_swaps_freed += !shmem_free_swap(mapping,
								index, page);
461
				continue;
462 463 464
			}

			if (!trylock_page(page))
465
				continue;
466 467 468
			if (page->mapping == mapping) {
				VM_BUG_ON(PageWriteback(page));
				truncate_inode_page(mapping, page);
469 470 471
			}
			unlock_page(page);
		}
472 473
		shmem_deswap_pagevec(&pvec);
		pagevec_release(&pvec);
474 475 476 477
		mem_cgroup_uncharge_end();
		cond_resched();
		index++;
	}
Linus Torvalds's avatar
Linus Torvalds committed
478

479 480 481 482 483 484 485 486 487 488 489 490 491 492
	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();
493 494 495 496
		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) {
497 498 499 500 501
			if (index == start)
				break;
			index = start;
			continue;
		}
502
		if (index == start && indices[0] > end) {
503 504
			shmem_deswap_pagevec(&pvec);
			pagevec_release(&pvec);
505 506 507 508 509 510
			break;
		}
		mem_cgroup_uncharge_start();
		for (i = 0; i < pagevec_count(&pvec); i++) {
			struct page *page = pvec.pages[i];

511
			index = indices[i];
512 513 514
			if (index > end)
				break;

515 516 517 518 519 520
			if (radix_tree_exceptional_entry(page)) {
				nr_swaps_freed += !shmem_free_swap(mapping,
								index, page);
				continue;
			}

521
			lock_page(page);
522 523 524 525
			if (page->mapping == mapping) {
				VM_BUG_ON(PageWriteback(page));
				truncate_inode_page(mapping, page);
			}
526 527
			unlock_page(page);
		}
528 529
		shmem_deswap_pagevec(&pvec);
		pagevec_release(&pvec);
530 531 532
		mem_cgroup_uncharge_end();
		index++;
	}
533

Linus Torvalds's avatar
Linus Torvalds committed
534
	spin_lock(&info->lock);
535
	info->swapped -= nr_swaps_freed;
Linus Torvalds's avatar
Linus Torvalds committed
536 537 538
	shmem_recalc_inode(inode);
	spin_unlock(&info->lock);

539
	inode->i_ctime = inode->i_mtime = CURRENT_TIME;
Linus Torvalds's avatar
Linus Torvalds committed
540
}
541
EXPORT_SYMBOL_GPL(shmem_truncate_range);
Linus Torvalds's avatar
Linus Torvalds committed
542

543
static int shmem_setattr(struct dentry *dentry, struct iattr *attr)
Linus Torvalds's avatar
Linus Torvalds committed
544 545 546 547
{
	struct inode *inode = dentry->d_inode;
	int error;

548 549 550 551
	error = inode_change_ok(inode, attr);
	if (error)
		return error;

552 553 554
	if (S_ISREG(inode->i_mode) && (attr->ia_valid & ATTR_SIZE)) {
		loff_t oldsize = inode->i_size;
		loff_t newsize = attr->ia_size;
555

556 557 558 559 560 561 562 563 564 565 566
		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
567 568
	}

569
	setattr_copy(inode, attr);
570
#ifdef CONFIG_TMPFS_POSIX_ACL
571
	if (attr->ia_valid & ATTR_MODE)
572
		error = generic_acl_chmod(inode);
573
#endif
Linus Torvalds's avatar
Linus Torvalds committed
574 575 576
	return error;
}

Al Viro's avatar
Al Viro committed
577
static void shmem_evict_inode(struct inode *inode)
Linus Torvalds's avatar
Linus Torvalds committed
578 579
{
	struct shmem_inode_info *info = SHMEM_I(inode);
580
	struct shmem_xattr *xattr, *nxattr;
Linus Torvalds's avatar
Linus Torvalds committed
581

582
	if (inode->i_mapping->a_ops == &shmem_aops) {
Linus Torvalds's avatar
Linus Torvalds committed
583 584
		shmem_unacct_size(info->flags, inode->i_size);
		inode->i_size = 0;
585
		shmem_truncate_range(inode, 0, (loff_t)-1);
Linus Torvalds's avatar
Linus Torvalds committed
586
		if (!list_empty(&info->swaplist)) {
587
			mutex_lock(&shmem_swaplist_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
588
			list_del_init(&info->swaplist);
589
			mutex_unlock(&shmem_swaplist_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
590
		}
591 592
	} else
		kfree(info->symlink);
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
	radswap = swp_to_radix_entry(swap);
615
	index = radix_tree_locate_item(&mapping->page_tree, radswap);
616
	if (index == -1)
617
		return 0;
618

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
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.
624 625 626
	 */
	if (shmem_swaplist.next != &info->swaplist)
		list_move_tail(&shmem_swaplist, &info->swaplist);
627

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).
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;
663 664 665 666 667 668 669 670 671 672
	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;
673
	/* No radix_tree_preload: swap entry keeps a place for page in tree */
Linus Torvalds's avatar
Linus Torvalds committed
674

675
	mutex_lock(&shmem_swaplist_mutex);
676 677
	list_for_each_safe(this, next, &shmem_swaplist) {
		info = list_entry(this, struct shmem_inode_info, swaplist);
678
		if (info->swapped)
679
			found = shmem_unuse_inode(info, swap, page);
680 681
		else
			list_del_init(&info->swaplist);
682
		cond_resched();
683
		if (found)
684
			break;
Linus Torvalds's avatar
Linus Torvalds committed
685
	}
686
	mutex_unlock(&shmem_swaplist_mutex);
687 688 689 690 691 692

	if (!found)
		mem_cgroup_uncharge_cache_page(page);
	if (found < 0)
		error = found;
out:
693 694
	unlock_page(page);
	page_cache_release(page);
695
	return error;
Linus Torvalds's avatar
Linus Torvalds committed
696 697 698 699 700 701 702 703 704 705
}

/*
 * 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;
706 707
	swp_entry_t swap;
	pgoff_t index;
Linus Torvalds's avatar
Linus Torvalds committed
708 709 710 711 712 713 714 715

	BUG_ON(!PageLocked(page));
	mapping = page->mapping;
	index = page->index;
	inode = mapping->host;
	info = SHMEM_I(inode);
	if (info->flags & VM_LOCKED)
		goto redirty;
716
	if (!total_swap_pages)
Linus Torvalds's avatar
Linus Torvalds committed
717 718
		goto redirty;

719 720 721
	/*
	 * shmem_backing_dev_info's capabilities prevent regular writeback or
	 * sync from ever calling shmem_writepage; but a stacking filesystem
722
	 * might use ->writepage of its underlying filesystem, in which case
723
	 * tmpfs should write out to swap only in response to memory pressure,
724
	 * and not for the writeback threads or sync.
725
	 */
726 727 728 729 730 731 732
	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;
733

734 735
	/*
	 * Add inode to shmem_unuse()'s list of swapped-out inodes,
736 737
	 * if it's not already there.  Do it now before the page is
	 * moved to swap cache, when its pagelock no longer protects
738
	 * the inode from eviction.  But don't unlock the mutex until
739 740
	 * we've incremented swapped, because shmem_unuse_inode() will
	 * prune a !swapped inode from the swaplist under this mutex.
741
	 */
742 743 744
	mutex_lock(&shmem_swaplist_mutex);
	if (list_empty(&info->swaplist))
		list_add_tail(&info->swaplist, &shmem_swaplist);
745

746
	if (add_to_swap_cache(page, swap, GFP_ATOMIC) == 0) {
747
		swap_shmem_alloc(swap);
748 749 750 751 752
		shmem_delete_from_page_cache(page, swp_to_radix_entry(swap));

		spin_lock(&info->lock);
		info->swapped++;
		shmem_recalc_inode(inode);
753
		spin_unlock(&info->lock);
754 755

		mutex_unlock(&shmem_swaplist_mutex);
756
		BUG_ON(page_mapped(page));
757
		swap_writepage(page, wbc);
Linus Torvalds's avatar
Linus Torvalds committed
758 759 760
		return 0;
	}

761
	mutex_unlock(&shmem_swaplist_mutex);
762
	swapcache_free(swap, NULL);
Linus Torvalds's avatar
Linus Torvalds committed
763 764
redirty:
	set_page_dirty(page);
765 766 767 768
	if (wbc->for_reclaim)
		return AOP_WRITEPAGE_ACTIVATE;	/* Return with page locked */
	unlock_page(page);
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
769 770 771
}

#ifdef CONFIG_NUMA
772
#ifdef CONFIG_TMPFS
773
static void shmem_show_mpol(struct seq_file *seq, struct mempolicy *mpol)
774
{
775
	char buffer[64];
776

777
	if (!mpol || mpol->mode == MPOL_DEFAULT)
778
		return;		/* show nothing */
779

780
	mpol_to_str(buffer, sizeof(buffer), mpol, 1);
781 782

	seq_printf(seq, ",mpol=%s", buffer);
783
}
784 785 786 787 788 789 790 791 792 793 794 795

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;
}
796 797
#endif /* CONFIG_TMPFS */

798 799
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
800
{
801
	struct mempolicy mpol, *spol;
Linus Torvalds's avatar
Linus Torvalds committed
802 803
	struct vm_area_struct pvma;

804
	spol = mpol_cond_copy(&mpol,
805
			mpol_shared_policy_lookup(&info->policy, index));
806

Linus Torvalds's avatar
Linus Torvalds committed
807
	/* Create a pseudo vma that just contains the policy */
808
	pvma.vm_start = 0;
809
	pvma.vm_pgoff = index;
810
	pvma.vm_ops = NULL;
811
	pvma.vm_policy = spol;
812
	return swapin_readahead(swap, gfp, &pvma, 0);
Linus Torvalds's avatar
Linus Torvalds committed
813 814
}

815
static struct page *shmem_alloc_page(gfp_t gfp,
816
			struct shmem_inode_info *info, pgoff_t index)
Linus Torvalds's avatar
Linus Torvalds committed
817 818 819
{
	struct vm_area_struct pvma;

820 821
	/* Create a pseudo vma that just contains the policy */
	pvma.vm_start = 0;
822
	pvma.vm_pgoff = index;
823
	pvma.vm_ops = NULL;
824
	pvma.vm_policy = mpol_shared_policy_lookup(&info->policy, index);
825 826 827 828 829

	/*
	 * alloc_page_vma() will drop the shared policy reference
	 */
	return alloc_page_vma(gfp, &pvma, 0);
Linus Torvalds's avatar
Linus Torvalds committed
830
}
831 832
#else /* !CONFIG_NUMA */
#ifdef CONFIG_TMPFS
833
static inline void shmem_show_mpol(struct seq_file *seq, struct mempolicy *mpol)
834 835 836 837
{
}
#endif /* CONFIG_TMPFS */

838 839
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
840
{
841
	return swapin_readahead(swap, gfp, NULL, 0);
Linus Torvalds's avatar
Linus Torvalds committed
842 843
}

844
static inline struct page *shmem_alloc_page(gfp_t gfp,
845
			struct shmem_inode_info *info, pgoff_t index)
Linus Torvalds's avatar
Linus Torvalds committed
846
{
847
	return alloc_page(gfp);
Linus Torvalds's avatar
Linus Torvalds committed
848
}
849
#endif /* CONFIG_NUMA */
Linus Torvalds's avatar
Linus Torvalds committed
850

851 852 853 854 855 856 857
#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
858
/*
859
 * shmem_getpage_gfp - find page in cache, or get from swap, or allocate
Linus Torvalds's avatar
Linus Torvalds committed
860 861 862 863 864
 *
 * 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
 */
865
static int shmem_getpage_gfp(struct inode *inode, pgoff_t index,
866
	struct page **pagep, enum sgp_type sgp, gfp_t gfp, int *fault_type)
Linus Torvalds's avatar
Linus Torvalds committed
867 868
{
	struct address_space *mapping = inode->i_mapping;
869
	struct shmem_inode_info *info;
Linus Torvalds's avatar
Linus Torvalds committed
870
	struct shmem_sb_info *sbinfo;
871
	struct page *page;
Linus Torvalds's avatar
Linus Torvalds committed
872 873
	swp_entry_t swap;
	int error;
874
	int once = 0;
Linus Torvalds's avatar
Linus Torvalds committed
875

876
	if (index > (MAX_LFS_FILESIZE >> PAGE_CACHE_SHIFT))
Linus Torvalds's avatar
Linus Torvalds committed
877 878
		return -EFBIG;
repeat:
879
	swap.val = 0;
880
	page = find_lock_page(mapping, index);
881 882 883 884 885 886 887 888 889 890 891 892
	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
893
		/*
894 895 896
		 * 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
897
		 */
898 899 900
		BUG_ON(page && !PageUptodate(page));
		*pagep = page;
		return 0;
901 902 903
	}

	/*
904 905
	 * Fast cache lookup did not find it:
	 * bring it back from swap or allocate.
906
	 */
907 908
	info = SHMEM_I(inode);
	sbinfo = SHMEM_SB(inode->i_sb);
Linus Torvalds's avatar
Linus Torvalds committed
909 910 911

	if (swap.val) {
		/* Look it up and read it in.. */
912 913
		page = lookup_swap_cache(swap);
		if (!page) {
Linus Torvalds's avatar
Linus Torvalds committed
914
			/* here we actually do the io */
915 916
			if (fault_type)
				*fault_type |= VM_FAULT_MAJOR;
917
			page = shmem_swapin(swap, gfp, info, index);
918
			if (!page) {
919 920
				error = -ENOMEM;
				goto failed;
Linus Torvalds's avatar
Linus Torvalds committed
921 922 923 924
			}
		}

		/* We have to do this with page locked to prevent races */
925
		lock_page(page);
926
		if (!PageUptodate(page)) {
Linus Torvalds's avatar
Linus Torvalds committed
927
			error = -EIO;
928
			goto failed;
Linus Torvalds's avatar
Linus Torvalds committed
929
		}
930 931 932 933 934 935 936 937 938
		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
939
		}
940

941 942 943 944 945
		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));
946 947 948 949
		if (error)
			goto failed;

		spin_lock(&info->lock);
950
		info->swapped--;
951
		shmem_recalc_inode(inode);
952
		spin_unlock(&info->lock);
953 954

		delete_from_swap_cache(page);
955 956 957
		set_page_dirty(page);
		swap_free(swap);

958 959 960 961
	} else {
		if (shmem_acct_block(info->flags)) {
			error = -ENOSPC;
			goto failed;
Linus Torvalds's avatar
Linus Torvalds committed
962
		}
963
		if (sbinfo->max_blocks) {
964
			if (percpu_counter_compare(&sbinfo->used_blocks,
965 966 967 968
						sbinfo->max_blocks) >= 0) {
				error = -ENOSPC;
				goto unacct;
			}
969
			percpu_counter_inc(&sbinfo->used_blocks);
970
		}
Linus Torvalds's avatar
Linus Torvalds committed
971

972 973 974 975
		page = shmem_alloc_page(gfp, info, index);
		if (!page) {
			error = -ENOMEM;
			goto decused;
Linus Torvalds's avatar
Linus Torvalds committed
976 977
		}

978 979
		SetPageSwapBacked(page);
		__set_page_locked(page);
980 981 982 983 984
		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);
985 986 987 988 989
		if (error)
			goto decused;
		lru_cache_add_anon(page);

		spin_lock(&info->lock);
Linus Torvalds's avatar
Linus Torvalds committed
990
		info->alloced++;
991 992
		inode->i_blocks += BLOCKS_PER_PAGE;
		shmem_recalc_inode(inode);
Linus Torvalds's avatar
Linus Torvalds committed
993
		spin_unlock(&info->lock);
994

995 996 997
		clear_highpage(page);
		flush_dcache_page(page);
		SetPageUptodate(page);
998
		if (sgp == SGP_DIRTY)
999
			set_page_dirty(page);
Linus Torvalds's avatar
Linus Torvalds committed
1000 1001
	}
done:
1002 1003 1004 1005 1006
	/* 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
1007
	}
1008 1009
	*pagep = page;
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
1010

1011
	/*
1012
	 * Error recovery.
1013
	 */
1014 1015 1016 1017 1018 1019
trunc:
	ClearPageDirty(page);
	delete_from_page_cache(page);
	spin_lock(&info->lock);
	info->alloced--;
	inode->i_blocks -= BLOCKS_PER_PAGE;
1020
	spin_unlock(&info->lock);
1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034
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;
	}
1035
	if (page) {
1036
		unlock_page(page);
1037
		page_cache_release(page);
1038 1039 1040 1041 1042 1043
	}
	if (error == -ENOSPC && !once++) {
		info = SHMEM_I(inode);
		spin_lock(&info->lock);
		shmem_recalc_inode(inode);
		spin_unlock(&info->lock);
1044
		goto repeat;
1045
	}
1046 1047 1048
	if (error == -EEXIST)
		goto repeat;
	return error;
Linus Torvalds's avatar
Linus Torvalds committed
1049 1050
}

Nick Piggin's avatar
Nick Piggin committed
1051
static int shmem_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
Linus Torvalds's avatar
Linus Torvalds committed
1052
{
1053
	struct inode *inode = vma->vm_file->f_path.dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
1054
	int error;
1055
	int ret = VM_FAULT_LOCKED;
Linus Torvalds's avatar
Linus Torvalds committed
1056

1057
	error = shmem_getpage(inode, vmf->pgoff, &vmf->page, SGP_CACHE, &ret);
Nick Piggin's avatar
Nick Piggin committed
1058 1059
	if (error)
		return ((error == -ENOMEM) ? VM_FAULT_OOM : VM_FAULT_SIGBUS);
1060

1061 1062 1063 1064
	if (ret & VM_FAULT_MAJOR) {
		count_vm_event(PGMAJFAULT);
		mem_cgroup_count_vm_event(vma->vm_mm, PGMAJFAULT);
	}
1065
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
1066 1067 1068
}

#ifdef CONFIG_NUMA
1069
static int shmem_set_policy(struct vm_area_struct *vma, struct mempolicy *mpol)
Linus Torvalds's avatar
Linus Torvalds committed
1070
{
1071 1072
	struct inode *inode = vma->vm_file->f_path.dentry->d_inode;
	return mpol_set_shared_policy(&SHMEM_I(inode)->policy, vma, mpol);
Linus Torvalds's avatar
Linus Torvalds committed
1073 1074
}

1075 1076
static struct mempolicy *shmem_get_policy(struct vm_area_struct *vma,
					  unsigned long addr)
Linus Torvalds's avatar
Linus Torvalds committed
1077
{
1078 1079
	struct inode *inode = vma->vm_file->f_path.dentry->d_inode;
	pgoff_t index;
Linus Torvalds's avatar
Linus Torvalds committed
1080

1081 1082
	index = ((addr - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
	return mpol_shared_policy_lookup(&SHMEM_I(inode)->policy, index);
Linus Torvalds's avatar
Linus Torvalds committed
1083 1084 1085 1086 1087
}
#endif

int shmem_lock(struct file *file, int lock, struct user_struct *user)
{
1088
	struct inode *inode = file->f_path.dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
1089 1090 1091 1092 1093 1094 1095 1096
	struct shmem_inode_info *info = SHMEM_I(inode);
	int retval = -ENOMEM;

	spin_lock(&info->lock);
	if (lock && !(info->flags & VM_LOCKED)) {
		if (!user_shm_lock(inode->i_size, user))
			goto out_nomem;
		info->flags |= VM_LOCKED;
1097
		mapping_set_unevictable(file->f_mapping);
Linus Torvalds's avatar
Linus Torvalds committed
1098 1099 1100 1101
	}
	if (!lock && (info->flags & VM_LOCKED) && user) {
		user_shm_unlock(inode->i_size, user);
		info->flags &= ~VM_LOCKED;
1102
		mapping_clear_unevictable(file->f_mapping);
Linus Torvalds's avatar
Linus Torvalds committed
1103 1104
	}
	retval = 0;
1105

Linus Torvalds's avatar
Linus Torvalds committed
1106 1107 1108 1109 1110
out_nomem:
	spin_unlock(&info->lock);
	return retval;
}

1111
static int shmem_mmap(struct file *file, struct vm_area_struct *vma)
Linus Torvalds's avatar
Linus Torvalds committed
1112 1113 1114
{
	file_accessed(file);
	vma->vm_ops = &shmem_vm_ops;
Nick Piggin's avatar
Nick Piggin committed
1115
	vma->vm_flags |= VM_CAN_NONLINEAR;
Linus Torvalds's avatar
Linus Torvalds committed
1116 1117 1118
	return 0;
}

1119
static struct inode *shmem_get_inode(struct super_block *sb, const struct inode *dir,
1120
				     umode_t mode, dev_t dev, unsigned long flags)
Linus Torvalds's avatar
Linus Torvalds committed
1121 1122 1123 1124 1125
{
	struct inode *inode;
	struct shmem_inode_info *info;
	struct shmem_sb_info *sbinfo = SHMEM_SB(sb);

1126 1127
	if (shmem_reserve_inode(sb))
		return NULL;
Linus Torvalds's avatar
Linus Torvalds committed
1128 1129 1130

	inode = new_inode(sb);
	if (inode) {
1131
		inode->i_ino = get_next_ino();
1132
		inode_init_owner(inode, dir, mode);
Linus Torvalds's avatar
Linus Torvalds committed
1133 1134 1135
		inode->i_blocks = 0;
		inode->i_mapping->backing_dev_info = &shmem_backing_dev_info;
		inode->i_atime = inode->i_mtime = inode->i_ctime = CURRENT_TIME;
1136
		inode->i_generation = get_seconds();
Linus Torvalds's avatar
Linus Torvalds committed
1137 1138 1139
		info = SHMEM_I(inode);
		memset(info, 0, (char *)inode - (char *)info);
		spin_lock_init(&info->lock);
1140
		info->flags = flags & VM_NORESERVE;
Linus Torvalds's avatar
Linus Torvalds committed
1141
		INIT_LIST_HEAD(&info->swaplist);
1142
		INIT_LIST_HEAD(&info->xattr_list);
1143
		cache_no_acl(inode);
Linus Torvalds's avatar
Linus Torvalds committed
1144 1145 1146

		switch (mode & S_IFMT) {
		default:
1147
			inode->i_op = &shmem_special_inode_operations;
Linus Torvalds's avatar
Linus Torvalds committed
1148 1149 1150
			init_special_inode(inode, mode, dev);
			break;
		case S_IFREG:
1151
			inode->i_mapping->a_ops = &shmem_aops;
Linus Torvalds's avatar
Linus Torvalds committed
1152 1153
			inode->i_op = &shmem_inode_operations;
			inode->i_fop = &shmem_file_operations;
1154 1155
			mpol_shared_policy_init(&info->policy,
						 shmem_get_sbmpol(sbinfo));
Linus Torvalds's avatar
Linus Torvalds committed
1156 1157
			break;
		case S_IFDIR:
1158
			inc_nlink(inode);
Linus Torvalds's avatar
Linus Torvalds committed
1159 1160 1161 1162 1163 1164 1165 1166 1167 1168
			/* Some things misbehave if size == 0 on a directory */
			inode->i_size = 2 * BOGO_DIRENT_SIZE;
			inode->i_op = &shmem_dir_inode_operations;
			inode->i_fop = &simple_dir_operations;
			break;
		case S_IFLNK:
			/*
			 * Must not load anything in the rbtree,
			 * mpol_free_shared_policy will not be called.
			 */
1169
			mpol_shared_policy_init(&info->policy, NULL);
Linus Torvalds's avatar
Linus Torvalds committed
1170 1171
			break;
		}
1172 1173
	} else
		shmem_free_inode(sb);
Linus Torvalds's avatar
Linus Torvalds committed
1174 1175 1176 1177
	return inode;
}

#ifdef CONFIG_TMPFS
1178
static const struct inode_operations shmem_symlink_inode_operations;
1179
static const struct inode_operations shmem_short_symlink_operations;
Linus Torvalds's avatar
Linus Torvalds committed
1180

1181 1182 1183 1184 1185 1186
#ifdef CONFIG_TMPFS_XATTR
static int shmem_initxattrs(struct inode *, const struct xattr *, void *);
#else
#define shmem_initxattrs NULL
#endif

Linus Torvalds's avatar
Linus Torvalds committed
1187
static int
Nick Piggin's avatar
Nick Piggin committed
1188 1189 1190
shmem_write_begin(struct file *file, struct address_space *mapping,
			loff_t pos, unsigned len, unsigned flags,
			struct page **pagep, void **fsdata)
Linus Torvalds's avatar
Linus Torvalds committed
1191
{
Nick Piggin's avatar
Nick Piggin committed
1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203
	struct inode *inode = mapping->host;
	pgoff_t index = pos >> PAGE_CACHE_SHIFT;
	return shmem_getpage(inode, index, pagep, SGP_WRITE, NULL);
}

static int
shmem_write_end(struct file *file, struct address_space *mapping,
			loff_t pos, unsigned len, unsigned copied,
			struct page *page, void *fsdata)
{
	struct inode *inode = mapping->host;

1204 1205 1206
	if (pos + copied > inode->i_size)
		i_size_write(inode, pos + copied);

Nick Piggin's avatar
Nick Piggin committed
1207
	set_page_dirty(page);
1208
	unlock_page(page);
Nick Piggin's avatar
Nick Piggin committed
1209 1210 1211
	page_cache_release(page);

	return copied;
Linus Torvalds's avatar
Linus Torvalds committed
1212 1213 1214 1215
}

static void do_shmem_file_read(struct file *filp, loff_t *ppos, read_descriptor_t *desc, read_actor_t actor)
{
1216
	struct inode *inode = filp->f_path.dentry->d_inode;
Linus Torvalds's avatar
Linus Torvalds committed
1217
	struct address_space *mapping = inode->i_mapping;
1218 1219
	pgoff_t index;
	unsigned long offset;
1220 1221 1222 1223 1224 1225 1226 1227 1228
	enum sgp_type sgp = SGP_READ;

	/*
	 * Might this read be for a stacking filesystem?  Then when reading
	 * holes of a sparse file, we actually need to allocate those pages,
	 * and even mark them dirty, so it cannot exceed the max_blocks limit.
	 */
	if (segment_eq(get_fs(), KERNEL_DS))
		sgp = SGP_DIRTY;
Linus Torvalds's avatar
Linus Torvalds committed
1229 1230 1231 1232 1233 1234

	index = *ppos >> PAGE_CACHE_SHIFT;
	offset = *ppos & ~PAGE_CACHE_MASK;

	for (;;) {
		struct page *page = NULL;
1235 1236
		pgoff_t end_index;
		unsigned long nr, ret;
Linus Torvalds's avatar
Linus Torvalds committed
1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247
		loff_t i_size = i_size_read(inode);

		end_index = i_size >> PAGE_CACHE_SHIFT;
		if (index