shmem.c 64.4 KB
Newer Older
L
Linus Torvalds 已提交
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.
L
Linus Torvalds 已提交
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>
 *
L
Linus Torvalds 已提交
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>
H
Hugh Dickins 已提交
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
L
Linus Torvalds 已提交
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>
L
Linus Torvalds 已提交
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>
L
Linus Torvalds 已提交
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>
M
Mimi Zohar 已提交
65
#include <linux/magic.h>
66

L
Linus Torvalds 已提交
67 68 69
#include <asm/uaccess.h>
#include <asm/pgtable.h>

H
Hugh Dickins 已提交
70
#define BLOCKS_PER_PAGE  (PAGE_CACHE_SIZE/512)
L
Linus Torvalds 已提交
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 */
L
Linus Torvalds 已提交
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 */
L
Linus Torvalds 已提交
91 92 93
	SGP_WRITE,	/* may exceed i_size, may allocate page */
};

A
Andrew Morton 已提交
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);
}
A
Andrew Morton 已提交
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);
}
L
Linus Torvalds 已提交
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));
L
Linus Torvalds 已提交
131 132 133 134
}

static inline void shmem_unacct_size(unsigned long flags, loff_t size)
{
135
	if (!(flags & VM_NORESERVE))
L
Linus Torvalds 已提交
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;
L
Linus Torvalds 已提交
149 150 151 152
}

static inline void shmem_unacct_blocks(unsigned long flags, long pages)
{
153
	if (flags & VM_NORESERVE)
L
Linus Torvalds 已提交
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;
L
Linus Torvalds 已提交
164

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

static LIST_HEAD(shmem_swaplist);
171
static DEFINE_MUTEX(shmem_swaplist_mutex);
L
Linus Torvalds 已提交
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
L
Linus Torvalds 已提交
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);
L
Linus Torvalds 已提交
220
		info->alloced -= freed;
221
		inode->i_blocks -= freed * BLOCKS_PER_PAGE;
L
Linus Torvalds 已提交
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)
L
Linus Torvalds 已提交
428
{
429
	struct address_space *mapping = inode->i_mapping;
L
Linus Torvalds 已提交
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++;
	}
L
Linus Torvalds 已提交
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

L
Linus Torvalds 已提交
534
	spin_lock(&info->lock);
535
	info->swapped -= nr_swaps_freed;
L
Linus Torvalds 已提交
536 537 538
	shmem_recalc_inode(inode);
	spin_unlock(&info->lock);

539
	inode->i_ctime = inode->i_mtime = CURRENT_TIME;
L
Linus Torvalds 已提交
540
}
541
EXPORT_SYMBOL_GPL(shmem_truncate_range);
L
Linus Torvalds 已提交
542

543
static int shmem_setattr(struct dentry *dentry, struct iattr *attr)
L
Linus Torvalds 已提交
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);
		}
L
Linus Torvalds 已提交
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
L
Linus Torvalds 已提交
574 575 576
	return error;
}

A
Al Viro 已提交
577
static void shmem_evict_inode(struct inode *inode)
L
Linus Torvalds 已提交
578 579
{
	struct shmem_inode_info *info = SHMEM_I(inode);
580
	struct shmem_xattr *xattr, *nxattr;
L
Linus Torvalds 已提交
581

582
	if (inode->i_mapping->a_ops == &shmem_aops) {
L
Linus Torvalds 已提交
583 584
		shmem_unacct_size(info->flags, inode->i_size);
		inode->i_size = 0;
585
		shmem_truncate_range(inode, 0, (loff_t)-1);
L
Linus Torvalds 已提交
586
		if (!list_empty(&info->swaplist)) {
587
			mutex_lock(&shmem_swaplist_mutex);
L
Linus Torvalds 已提交
588
			list_del_init(&info->swaplist);
589
			mutex_unlock(&shmem_swaplist_mutex);
L
Linus Torvalds 已提交
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);
A
Al Viro 已提交
600
	end_writeback(inode);
L
Linus Torvalds 已提交
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)
L
Linus Torvalds 已提交
608
{
609
	struct address_space *mapping = info->vfs_inode.i_mapping;
610
	void *radswap;
611
	pgoff_t index;
612
	int error;
L
Linus Torvalds 已提交
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

H
Hugh Dickins 已提交
619 620
	/*
	 * Move _head_ to start search for next from here.
A
Al Viro 已提交
621
	 * But be careful: shmem_evict_inode checks list_empty without taking
H
Hugh Dickins 已提交
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.
H
Hugh Dickins 已提交
624 625 626
	 */
	if (shmem_swaplist.next != &info->swaplist)
		list_move_tail(&shmem_swaplist, &info->swaplist);
627

K
KAMEZAWA Hiroyuki 已提交
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).
K
KAMEZAWA Hiroyuki 已提交
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 */
L
Linus Torvalds 已提交
651
	}
652
	return error;
L
Linus Torvalds 已提交
653 654 655
}

/*
656
 * Search through swapped inodes to find and replace swap by page.
L
Linus Torvalds 已提交
657
 */
658
int shmem_unuse(swp_entry_t swap, struct page *page)
L
Linus Torvalds 已提交
659
{
660
	struct list_head *this, *next;
L
Linus Torvalds 已提交
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 */
L
Linus Torvalds 已提交
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;
L
Linus Torvalds 已提交
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:
H
Hugh Dickins 已提交
693 694
	unlock_page(page);
	page_cache_release(page);
695
	return error;
L
Linus Torvalds 已提交
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;
L
Linus Torvalds 已提交
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)
L
Linus Torvalds 已提交
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) {
H
Hugh Dickins 已提交
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));
H
Hugh Dickins 已提交
757
		swap_writepage(page, wbc);
L
Linus Torvalds 已提交
758 759 760
		return 0;
	}

761
	mutex_unlock(&shmem_swaplist_mutex);
762
	swapcache_free(swap, NULL);
L
Linus Torvalds 已提交
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;
L
Linus Torvalds 已提交
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)
L
Linus Torvalds 已提交
800
{
801
	struct mempolicy mpol, *spol;
L
Linus Torvalds 已提交
802 803
	struct vm_area_struct pvma;

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

L
Linus Torvalds 已提交
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);
L
Linus Torvalds 已提交
813 814
}

815
static struct page *shmem_alloc_page(gfp_t gfp,
816
			struct shmem_inode_info *info, pgoff_t index)
L
Linus Torvalds 已提交
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);
L
Linus Torvalds 已提交
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)
L
Linus Torvalds 已提交
840
{
841
	return swapin_readahead(swap, gfp, NULL, 0);
L
Linus Torvalds 已提交
842 843
}

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

L
Linus Torvalds 已提交
858
/*
859
 * shmem_getpage_gfp - find page in cache, or get from swap, or allocate
L
Linus Torvalds 已提交
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)
L
Linus Torvalds 已提交
867 868
{
	struct address_space *mapping = inode->i_mapping;
869
	struct shmem_inode_info *info;
L
Linus Torvalds 已提交
870
	struct shmem_sb_info *sbinfo;
H
Hugh Dickins 已提交
871
	struct page *page;
L
Linus Torvalds 已提交
872 873
	swp_entry_t swap;
	int error;
874
	int once = 0;
L
Linus Torvalds 已提交
875

876
	if (index > (MAX_LFS_FILESIZE >> PAGE_CACHE_SHIFT))
L
Linus Torvalds 已提交
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)) {
H
Hugh Dickins 已提交
893
		/*
H
Hugh Dickins 已提交
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.
H
Hugh Dickins 已提交
897
		 */
898 899 900
		BUG_ON(page && !PageUptodate(page));
		*pagep = page;
		return 0;
H
Hugh Dickins 已提交
901 902 903
	}

	/*
904 905
	 * Fast cache lookup did not find it:
	 * bring it back from swap or allocate.
H
Hugh Dickins 已提交
906
	 */
907 908
	info = SHMEM_I(inode);
	sbinfo = SHMEM_SB(inode->i_sb);
L
Linus Torvalds 已提交
909 910 911

	if (swap.val) {
		/* Look it up and read it in.. */
H
Hugh Dickins 已提交
912 913
		page = lookup_swap_cache(swap);
		if (!page) {
L
Linus Torvalds 已提交
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);
H
Hugh Dickins 已提交
918
			if (!page) {
919 920
				error = -ENOMEM;
				goto failed;
L
Linus Torvalds 已提交
921 922 923 924
			}
		}

		/* We have to do this with page locked to prevent races */
925
		lock_page(page);
H
Hugh Dickins 已提交
926
		if (!PageUptodate(page)) {
L
Linus Torvalds 已提交
927
			error = -EIO;
928
			goto failed;
L
Linus Torvalds 已提交
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;
L
Linus Torvalds 已提交
939
		}
H
Hugh Dickins 已提交
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);
H
Hugh Dickins 已提交
952
		spin_unlock(&info->lock);
953 954

		delete_from_swap_cache(page);
H
Hugh Dickins 已提交
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;
L
Linus Torvalds 已提交
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
		}
L
Linus Torvalds 已提交
971

972 973 974 975
		page = shmem_alloc_page(gfp, info, index);
		if (!page) {
			error = -ENOMEM;
			goto decused;
L
Linus Torvalds 已提交
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);
L
Linus Torvalds 已提交
990
		info->alloced++;
991 992
		inode->i_blocks += BLOCKS_PER_PAGE;
		shmem_recalc_inode(inode);
L
Linus Torvalds 已提交
993
		spin_unlock(&info->lock);
994

H
Hugh Dickins 已提交
995 996 997
		clear_highpage(page);
		flush_dcache_page(page);
		SetPageUptodate(page);
998
		if (sgp == SGP_DIRTY)
H
Hugh Dickins 已提交
999
			set_page_dirty(page);
L
Linus Torvalds 已提交
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;
H
Hugh Dickins 已提交
1007
	}
1008 1009
	*pagep = page;
	return 0;
L
Linus Torvalds 已提交
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;
	}
H
Hugh Dickins 已提交
1035
	if (page) {
1036
		unlock_page(page);
H
Hugh Dickins 已提交
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);
H
Hugh Dickins 已提交
1044
		goto repeat;
1045
	}
1046 1047 1048
	if (error == -EEXIST)
		goto repeat;
	return error;
L
Linus Torvalds 已提交
1049 1050
}

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

1057
	error = shmem_getpage(inode, vmf->pgoff, &vmf->page, SGP_CACHE, &ret);
N
Nick Piggin 已提交
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;
L
Linus Torvalds 已提交
1066 1067 1068
}

#ifdef CONFIG_NUMA
1069
static int shmem_set_policy(struct vm_area_struct *vma, struct mempolicy *mpol)
L
Linus Torvalds 已提交
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);
L
Linus Torvalds 已提交
1073 1074
}

A
Adrian Bunk 已提交
1075 1076
static struct mempolicy *shmem_get_policy(struct vm_area_struct *vma,
					  unsigned long addr)
L
Linus Torvalds 已提交
1077
{
1078 1079
	struct inode *inode = vma->vm_file->f_path.dentry->d_inode;
	pgoff_t index;
L
Linus Torvalds 已提交
1080

1081 1082
	index = ((addr - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
	return mpol_shared_policy_lookup(&SHMEM_I(inode)->policy, index);
L
Linus Torvalds 已提交
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;
L
Linus Torvalds 已提交
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);
L
Linus Torvalds 已提交
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);
L
Linus Torvalds 已提交
1103 1104
	}
	retval = 0;
1105

L
Linus Torvalds 已提交
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)
L
Linus Torvalds 已提交
1112 1113 1114
{
	file_accessed(file);
	vma->vm_ops = &shmem_vm_ops;
N
Nick Piggin 已提交
1115
	vma->vm_flags |= VM_CAN_NONLINEAR;
L
Linus Torvalds 已提交
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)
L
Linus Torvalds 已提交
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;
L
Linus Torvalds 已提交
1128 1129 1130

	inode = new_inode(sb);
	if (inode) {
1131
		inode->i_ino = get_next_ino();
1132
		inode_init_owner(inode, dir, mode);
L
Linus Torvalds 已提交
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();
L
Linus Torvalds 已提交
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;
L
Linus Torvalds 已提交
1141
		INIT_LIST_HEAD(&info->swaplist);
1142
		INIT_LIST_HEAD(&info->xattr_list);
1143
		cache_no_acl(inode);
L
Linus Torvalds 已提交
1144 1145 1146

		switch (mode & S_IFMT) {
		default:
1147
			inode->i_op = &shmem_special_inode_operations;
L
Linus Torvalds 已提交
1148 1149 1150
			init_special_inode(inode, mode, dev);
			break;
		case S_IFREG:
1151
			inode->i_mapping->a_ops = &shmem_aops;
L
Linus Torvalds 已提交
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));
L
Linus Torvalds 已提交
1156 1157
			break;
		case S_IFDIR:
1158
			inc_nlink(inode);
L
Linus Torvalds 已提交
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);
L
Linus Torvalds 已提交
1170 1171
			break;
		}
1172 1173
	} else
		shmem_free_inode(sb);
L
Linus Torvalds 已提交
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;
L
Linus Torvalds 已提交
1180 1181

static int
N
Nick Piggin 已提交
1182 1183 1184
shmem_write_begin(struct file *file, struct address_space *mapping,
			loff_t pos, unsigned len, unsigned flags,
			struct page **pagep, void **fsdata)
L
Linus Torvalds 已提交
1185
{
N
Nick Piggin 已提交
1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197
	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;

H
Hugh Dickins 已提交
1198 1199 1200
	if (pos + copied > inode->i_size)
		i_size_write(inode, pos + copied);

N
Nick Piggin 已提交
1201
	set_page_dirty(page);
1202
	unlock_page(page);
N
Nick Piggin 已提交
1203 1204 1205
	page_cache_release(page);

	return copied;
L
Linus Torvalds 已提交
1206 1207 1208 1209
}

static void do_shmem_file_read(struct file *filp, loff_t *ppos, read_descriptor_t *desc, read_actor_t actor)
{
1210
	struct inode *inode = filp->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
1211
	struct address_space *mapping = inode->i_mapping;
1212 1213
	pgoff_t index;
	unsigned long offset;
1214 1215 1216 1217 1218 1219 1220 1221 1222
	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;
L
Linus Torvalds 已提交
1223 1224 1225 1226 1227 1228

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

	for (;;) {
		struct page *page = NULL;
1229 1230
		pgoff_t end_index;
		unsigned long nr, ret;
L
Linus Torvalds 已提交
1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241
		loff_t i_size = i_size_read(inode);

		end_index = i_size >> PAGE_CACHE_SHIFT;
		if (index > end_index)
			break;
		if (index == end_index) {
			nr = i_size & ~PAGE_CACHE_MASK;
			if (nr <= offset)
				break;
		}

1242
		desc->error = shmem_getpage(inode, index, &page, sgp, NULL);
L
Linus Torvalds 已提交
1243 1244 1245 1246 1247
		if (desc->error) {
			if (desc->error == -EINVAL)
				desc->error = 0;
			break;
		}
H
Hugh Dickins 已提交
1248 1249
		if (page)
			unlock_page(page);
L
Linus Torvalds 已提交
1250 1251 1252

		/*
		 * We must evaluate after, since reads (unlike writes)
1253
		 * are called without i_mutex protection against truncate
L
Linus Torvalds 已提交
1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280
		 */
		nr = PAGE_CACHE_SIZE;
		i_size = i_size_read(inode);
		end_index = i_size >> PAGE_CACHE_SHIFT;
		if (index == end_index) {
			nr = i_size & ~PAGE_CACHE_MASK;
			if (nr <= offset) {
				if (page)
					page_cache_release(page);
				break;
			}
		}
		nr -= offset;

		if (page) {
			/*
			 * If users can be writing to this page using arbitrary
			 * virtual addresses, take care about potential aliasing
			 * before reading the page on the kernel side.
			 */
			if (mapping_writably_mapped(mapping))
				flush_dcache_page(page);
			/*
			 * Mark the page accessed if we read the beginning.
			 */
			if (!offset)
				mark_page_accessed(page);
N
Nick Piggin 已提交
1281
		} else {
L
Linus Torvalds 已提交
1282
			page = ZERO_PAGE(0);
N
Nick Piggin 已提交
1283 1284
			page_cache_get(page);
		}
L
Linus Torvalds 已提交
1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311

		/*
		 * Ok, we have the page, and it's up-to-date, so
		 * now we can copy it to user space...
		 *
		 * The actor routine returns how many bytes were actually used..
		 * NOTE! This may not be the same as how much of a user buffer
		 * we filled up (we may be padding etc), so we can only update
		 * "pos" here (the actor routine has to update the user buffer
		 * pointers and the remaining count).
		 */
		ret = actor(desc, page, offset, nr);
		offset += ret;
		index += offset >> PAGE_CACHE_SHIFT;
		offset &= ~PAGE_CACHE_MASK;

		page_cache_release(page);
		if (ret != nr || !desc->count)
			break;

		cond_resched();
	}

	*ppos = ((loff_t) index << PAGE_CACHE_SHIFT) + offset;
	file_accessed(filp);
}

H
Hugh Dickins 已提交
1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343
static ssize_t shmem_file_aio_read(struct kiocb *iocb,
		const struct iovec *iov, unsigned long nr_segs, loff_t pos)
{
	struct file *filp = iocb->ki_filp;
	ssize_t retval;
	unsigned long seg;
	size_t count;
	loff_t *ppos = &iocb->ki_pos;

	retval = generic_segment_checks(iov, &nr_segs, &count, VERIFY_WRITE);
	if (retval)
		return retval;

	for (seg = 0; seg < nr_segs; seg++) {
		read_descriptor_t desc;

		desc.written = 0;
		desc.arg.buf = iov[seg].iov_base;
		desc.count = iov[seg].iov_len;
		if (desc.count == 0)
			continue;
		desc.error = 0;
		do_shmem_file_read(filp, ppos, &desc, file_read_actor);
		retval += desc.written;
		if (desc.error) {
			retval = retval ?: desc.error;
			break;
		}
		if (desc.count > 0)
			break;
	}
	return retval;
L
Linus Torvalds 已提交
1344 1345
}

1346 1347 1348 1349 1350
static ssize_t shmem_file_splice_read(struct file *in, loff_t *ppos,
				struct pipe_inode_info *pipe, size_t len,
				unsigned int flags)
{
	struct address_space *mapping = in->f_mapping;
1351
	struct inode *inode = mapping->host;
1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366
	unsigned int loff, nr_pages, req_pages;
	struct page *pages[PIPE_DEF_BUFFERS];
	struct partial_page partial[PIPE_DEF_BUFFERS];
	struct page *page;
	pgoff_t index, end_index;
	loff_t isize, left;
	int error, page_nr;
	struct splice_pipe_desc spd = {
		.pages = pages,
		.partial = partial,
		.flags = flags,
		.ops = &page_cache_pipe_buf_ops,
		.spd_release = spd_release_page,
	};

1367
	isize = i_size_read(inode);
1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387
	if (unlikely(*ppos >= isize))
		return 0;

	left = isize - *ppos;
	if (unlikely(left < len))
		len = left;

	if (splice_grow_spd(pipe, &spd))
		return -ENOMEM;

	index = *ppos >> PAGE_CACHE_SHIFT;
	loff = *ppos & ~PAGE_CACHE_MASK;
	req_pages = (len + loff + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
	nr_pages = min(req_pages, pipe->buffers);

	spd.nr_pages = find_get_pages_contig(mapping, index,
						nr_pages, spd.pages);
	index += spd.nr_pages;
	error = 0;

1388 1389 1390 1391 1392
	while (spd.nr_pages < nr_pages) {
		error = shmem_getpage(inode, index, &page, SGP_CACHE, NULL);
		if (error)
			break;
		unlock_page(page);
1393 1394 1395 1396 1397 1398 1399
		spd.pages[spd.nr_pages++] = page;
		index++;
	}

	index = *ppos >> PAGE_CACHE_SHIFT;
	nr_pages = spd.nr_pages;
	spd.nr_pages = 0;
1400

1401 1402 1403 1404 1405 1406 1407 1408 1409
	for (page_nr = 0; page_nr < nr_pages; page_nr++) {
		unsigned int this_len;

		if (!len)
			break;

		this_len = min_t(unsigned long, len, PAGE_CACHE_SIZE - loff);
		page = spd.pages[page_nr];

1410 1411 1412 1413
		if (!PageUptodate(page) || page->mapping != mapping) {
			error = shmem_getpage(inode, index, &page,
							SGP_CACHE, NULL);
			if (error)
1414
				break;
1415 1416 1417
			unlock_page(page);
			page_cache_release(spd.pages[page_nr]);
			spd.pages[page_nr] = page;
1418
		}
1419 1420

		isize = i_size_read(inode);
1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458
		end_index = (isize - 1) >> PAGE_CACHE_SHIFT;
		if (unlikely(!isize || index > end_index))
			break;

		if (end_index == index) {
			unsigned int plen;

			plen = ((isize - 1) & ~PAGE_CACHE_MASK) + 1;
			if (plen <= loff)
				break;

			this_len = min(this_len, plen - loff);
			len = this_len;
		}

		spd.partial[page_nr].offset = loff;
		spd.partial[page_nr].len = this_len;
		len -= this_len;
		loff = 0;
		spd.nr_pages++;
		index++;
	}

	while (page_nr < nr_pages)
		page_cache_release(spd.pages[page_nr++]);

	if (spd.nr_pages)
		error = splice_to_pipe(pipe, &spd);

	splice_shrink_spd(pipe, &spd);

	if (error > 0) {
		*ppos += error;
		file_accessed(in);
	}
	return error;
}

1459
static int shmem_statfs(struct dentry *dentry, struct kstatfs *buf)
L
Linus Torvalds 已提交
1460
{
1461
	struct shmem_sb_info *sbinfo = SHMEM_SB(dentry->d_sb);
L
Linus Torvalds 已提交
1462 1463 1464 1465

	buf->f_type = TMPFS_MAGIC;
	buf->f_bsize = PAGE_CACHE_SIZE;
	buf->f_namelen = NAME_MAX;
1466
	if (sbinfo->max_blocks) {
L
Linus Torvalds 已提交
1467
		buf->f_blocks = sbinfo->max_blocks;
1468 1469 1470
		buf->f_bavail =
		buf->f_bfree  = sbinfo->max_blocks -
				percpu_counter_sum(&sbinfo->used_blocks);
1471 1472
	}
	if (sbinfo->max_inodes) {
L
Linus Torvalds 已提交
1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483
		buf->f_files = sbinfo->max_inodes;
		buf->f_ffree = sbinfo->free_inodes;
	}
	/* else leave those fields 0 like simple_statfs */
	return 0;
}

/*
 * File creation. Allocate an inode, and we're done..
 */
static int
A
Al Viro 已提交
1484
shmem_mknod(struct inode *dir, struct dentry *dentry, umode_t mode, dev_t dev)
L
Linus Torvalds 已提交
1485
{
1486
	struct inode *inode;
L
Linus Torvalds 已提交
1487 1488
	int error = -ENOSPC;

1489
	inode = shmem_get_inode(dir->i_sb, dir, mode, dev, VM_NORESERVE);
L
Linus Torvalds 已提交
1490
	if (inode) {
1491
		error = security_inode_init_security(inode, dir,
1492
						     &dentry->d_name,
1493
						     NULL, NULL);
1494 1495 1496 1497 1498
		if (error) {
			if (error != -EOPNOTSUPP) {
				iput(inode);
				return error;
			}
1499
		}
1500 1501
#ifdef CONFIG_TMPFS_POSIX_ACL
		error = generic_acl_init(inode, dir);
1502 1503 1504
		if (error) {
			iput(inode);
			return error;
1505
		}
A
Al Viro 已提交
1506 1507
#else
		error = 0;
1508
#endif
L
Linus Torvalds 已提交
1509 1510 1511 1512 1513 1514 1515 1516
		dir->i_size += BOGO_DIRENT_SIZE;
		dir->i_ctime = dir->i_mtime = CURRENT_TIME;
		d_instantiate(dentry, inode);
		dget(dentry); /* Extra count - pin the dentry in core */
	}
	return error;
}

1517
static int shmem_mkdir(struct inode *dir, struct dentry *dentry, umode_t mode)
L
Linus Torvalds 已提交
1518 1519 1520 1521 1522
{
	int error;

	if ((error = shmem_mknod(dir, dentry, mode | S_IFDIR, 0)))
		return error;
1523
	inc_nlink(dir);
L
Linus Torvalds 已提交
1524 1525 1526
	return 0;
}

A
Al Viro 已提交
1527
static int shmem_create(struct inode *dir, struct dentry *dentry, umode_t mode,
L
Linus Torvalds 已提交
1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538
		struct nameidata *nd)
{
	return shmem_mknod(dir, dentry, mode | S_IFREG, 0);
}

/*
 * Link a file..
 */
static int shmem_link(struct dentry *old_dentry, struct inode *dir, struct dentry *dentry)
{
	struct inode *inode = old_dentry->d_inode;
1539
	int ret;
L
Linus Torvalds 已提交
1540 1541 1542 1543 1544 1545

	/*
	 * No ordinary (disk based) filesystem counts links as inodes;
	 * but each new link needs a new dentry, pinning lowmem, and
	 * tmpfs dentries cannot be pruned until they are unlinked.
	 */
1546 1547 1548
	ret = shmem_reserve_inode(inode->i_sb);
	if (ret)
		goto out;
L
Linus Torvalds 已提交
1549 1550 1551

	dir->i_size += BOGO_DIRENT_SIZE;
	inode->i_ctime = dir->i_ctime = dir->i_mtime = CURRENT_TIME;
1552
	inc_nlink(inode);
A
Al Viro 已提交
1553
	ihold(inode);	/* New dentry reference */
L
Linus Torvalds 已提交
1554 1555
	dget(dentry);		/* Extra pinning count for the created dentry */
	d_instantiate(dentry, inode);
1556 1557
out:
	return ret;
L
Linus Torvalds 已提交
1558 1559 1560 1561 1562 1563
}

static int shmem_unlink(struct inode *dir, struct dentry *dentry)
{
	struct inode *inode = dentry->d_inode;

1564 1565
	if (inode->i_nlink > 1 && !S_ISDIR(inode->i_mode))
		shmem_free_inode(inode->i_sb);
L
Linus Torvalds 已提交
1566 1567 1568

	dir->i_size -= BOGO_DIRENT_SIZE;
	inode->i_ctime = dir->i_ctime = dir->i_mtime = CURRENT_TIME;
1569
	drop_nlink(inode);
L
Linus Torvalds 已提交
1570 1571 1572 1573 1574 1575 1576 1577 1578
	dput(dentry);	/* Undo the count from "create" - this does all the work */
	return 0;
}

static int shmem_rmdir(struct inode *dir, struct dentry *dentry)
{
	if (!simple_empty(dentry))
		return -ENOTEMPTY;

1579 1580
	drop_nlink(dentry->d_inode);
	drop_nlink(dir);
L
Linus Torvalds 已提交
1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600
	return shmem_unlink(dir, dentry);
}

/*
 * The VFS layer already does all the dentry stuff for rename,
 * we just have to decrement the usage count for the target if
 * it exists so that the VFS layer correctly free's it when it
 * gets overwritten.
 */
static int shmem_rename(struct inode *old_dir, struct dentry *old_dentry, struct inode *new_dir, struct dentry *new_dentry)
{
	struct inode *inode = old_dentry->d_inode;
	int they_are_dirs = S_ISDIR(inode->i_mode);

	if (!simple_empty(new_dentry))
		return -ENOTEMPTY;

	if (new_dentry->d_inode) {
		(void) shmem_unlink(new_dir, new_dentry);
		if (they_are_dirs)
1601
			drop_nlink(old_dir);
L
Linus Torvalds 已提交
1602
	} else if (they_are_dirs) {
1603
		drop_nlink(old_dir);
1604
		inc_nlink(new_dir);
L
Linus Torvalds 已提交
1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619
	}

	old_dir->i_size -= BOGO_DIRENT_SIZE;
	new_dir->i_size += BOGO_DIRENT_SIZE;
	old_dir->i_ctime = old_dir->i_mtime =
	new_dir->i_ctime = new_dir->i_mtime =
	inode->i_ctime = CURRENT_TIME;
	return 0;
}

static int shmem_symlink(struct inode *dir, struct dentry *dentry, const char *symname)
{
	int error;
	int len;
	struct inode *inode;
H
Hugh Dickins 已提交
1620
	struct page *page;
L
Linus Torvalds 已提交
1621 1622 1623 1624 1625 1626 1627
	char *kaddr;
	struct shmem_inode_info *info;

	len = strlen(symname) + 1;
	if (len > PAGE_CACHE_SIZE)
		return -ENAMETOOLONG;

1628
	inode = shmem_get_inode(dir->i_sb, dir, S_IFLNK|S_IRWXUGO, 0, VM_NORESERVE);
L
Linus Torvalds 已提交
1629 1630 1631
	if (!inode)
		return -ENOSPC;

1632
	error = security_inode_init_security(inode, dir, &dentry->d_name,
1633
					     NULL, NULL);
1634 1635 1636 1637 1638 1639 1640 1641
	if (error) {
		if (error != -EOPNOTSUPP) {
			iput(inode);
			return error;
		}
		error = 0;
	}

L
Linus Torvalds 已提交
1642 1643
	info = SHMEM_I(inode);
	inode->i_size = len-1;
1644 1645 1646 1647 1648 1649 1650
	if (len <= SHORT_SYMLINK_LEN) {
		info->symlink = kmemdup(symname, len, GFP_KERNEL);
		if (!info->symlink) {
			iput(inode);
			return -ENOMEM;
		}
		inode->i_op = &shmem_short_symlink_operations;
L
Linus Torvalds 已提交
1651 1652 1653 1654 1655 1656
	} else {
		error = shmem_getpage(inode, 0, &page, SGP_WRITE, NULL);
		if (error) {
			iput(inode);
			return error;
		}
1657
		inode->i_mapping->a_ops = &shmem_aops;
L
Linus Torvalds 已提交
1658 1659 1660 1661 1662
		inode->i_op = &shmem_symlink_inode_operations;
		kaddr = kmap_atomic(page, KM_USER0);
		memcpy(kaddr, symname, len);
		kunmap_atomic(kaddr, KM_USER0);
		set_page_dirty(page);
1663
		unlock_page(page);
L
Linus Torvalds 已提交
1664 1665 1666 1667 1668 1669 1670 1671 1672
		page_cache_release(page);
	}
	dir->i_size += BOGO_DIRENT_SIZE;
	dir->i_ctime = dir->i_mtime = CURRENT_TIME;
	d_instantiate(dentry, inode);
	dget(dentry);
	return 0;
}

1673
static void *shmem_follow_short_symlink(struct dentry *dentry, struct nameidata *nd)
L
Linus Torvalds 已提交
1674
{
1675
	nd_set_link(nd, SHMEM_I(dentry->d_inode)->symlink);
1676
	return NULL;
L
Linus Torvalds 已提交
1677 1678
}

1679
static void *shmem_follow_link(struct dentry *dentry, struct nameidata *nd)
L
Linus Torvalds 已提交
1680 1681
{
	struct page *page = NULL;
1682 1683
	int error = shmem_getpage(dentry->d_inode, 0, &page, SGP_READ, NULL);
	nd_set_link(nd, error ? ERR_PTR(error) : kmap(page));
H
Hugh Dickins 已提交
1684 1685
	if (page)
		unlock_page(page);
1686
	return page;
L
Linus Torvalds 已提交
1687 1688
}

1689
static void shmem_put_link(struct dentry *dentry, struct nameidata *nd, void *cookie)
L
Linus Torvalds 已提交
1690 1691
{
	if (!IS_ERR(nd_get_link(nd))) {
1692
		struct page *page = cookie;
L
Linus Torvalds 已提交
1693 1694 1695 1696 1697 1698
		kunmap(page);
		mark_page_accessed(page);
		page_cache_release(page);
	}
}

1699
#ifdef CONFIG_TMPFS_XATTR
1700
/*
1701 1702
 * Superblocks without xattr inode operations may get some security.* xattr
 * support from the LSM "for free". As soon as we have any other xattrs
1703 1704 1705 1706
 * like ACLs, we also need to implement the security.* handlers at
 * filesystem level, though.
 */

1707 1708
static int shmem_xattr_get(struct dentry *dentry, const char *name,
			   void *buffer, size_t size)
1709
{
1710 1711 1712
	struct shmem_inode_info *info;
	struct shmem_xattr *xattr;
	int ret = -ENODATA;
1713

1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731
	info = SHMEM_I(dentry->d_inode);

	spin_lock(&info->lock);
	list_for_each_entry(xattr, &info->xattr_list, list) {
		if (strcmp(name, xattr->name))
			continue;

		ret = xattr->size;
		if (buffer) {
			if (size < xattr->size)
				ret = -ERANGE;
			else
				memcpy(buffer, xattr->value, xattr->size);
		}
		break;
	}
	spin_unlock(&info->lock);
	return ret;
1732 1733
}

1734 1735
static int shmem_xattr_set(struct dentry *dentry, const char *name,
			   const void *value, size_t size, int flags)
1736
{
1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791
	struct inode *inode = dentry->d_inode;
	struct shmem_inode_info *info = SHMEM_I(inode);
	struct shmem_xattr *xattr;
	struct shmem_xattr *new_xattr = NULL;
	size_t len;
	int err = 0;

	/* value == NULL means remove */
	if (value) {
		/* wrap around? */
		len = sizeof(*new_xattr) + size;
		if (len <= sizeof(*new_xattr))
			return -ENOMEM;

		new_xattr = kmalloc(len, GFP_KERNEL);
		if (!new_xattr)
			return -ENOMEM;

		new_xattr->name = kstrdup(name, GFP_KERNEL);
		if (!new_xattr->name) {
			kfree(new_xattr);
			return -ENOMEM;
		}

		new_xattr->size = size;
		memcpy(new_xattr->value, value, size);
	}

	spin_lock(&info->lock);
	list_for_each_entry(xattr, &info->xattr_list, list) {
		if (!strcmp(name, xattr->name)) {
			if (flags & XATTR_CREATE) {
				xattr = new_xattr;
				err = -EEXIST;
			} else if (new_xattr) {
				list_replace(&xattr->list, &new_xattr->list);
			} else {
				list_del(&xattr->list);
			}
			goto out;
		}
	}
	if (flags & XATTR_REPLACE) {
		xattr = new_xattr;
		err = -ENODATA;
	} else {
		list_add(&new_xattr->list, &info->xattr_list);
		xattr = NULL;
	}
out:
	spin_unlock(&info->lock);
	if (xattr)
		kfree(xattr->name);
	kfree(xattr);
	return err;
1792 1793
}

1794
static const struct xattr_handler *shmem_xattr_handlers[] = {
1795
#ifdef CONFIG_TMPFS_POSIX_ACL
1796 1797
	&generic_acl_access_handler,
	&generic_acl_default_handler,
1798
#endif
1799 1800
	NULL
};
1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922

static int shmem_xattr_validate(const char *name)
{
	struct { const char *prefix; size_t len; } arr[] = {
		{ XATTR_SECURITY_PREFIX, XATTR_SECURITY_PREFIX_LEN },
		{ XATTR_TRUSTED_PREFIX, XATTR_TRUSTED_PREFIX_LEN }
	};
	int i;

	for (i = 0; i < ARRAY_SIZE(arr); i++) {
		size_t preflen = arr[i].len;
		if (strncmp(name, arr[i].prefix, preflen) == 0) {
			if (!name[preflen])
				return -EINVAL;
			return 0;
		}
	}
	return -EOPNOTSUPP;
}

static ssize_t shmem_getxattr(struct dentry *dentry, const char *name,
			      void *buffer, size_t size)
{
	int err;

	/*
	 * If this is a request for a synthetic attribute in the system.*
	 * namespace use the generic infrastructure to resolve a handler
	 * for it via sb->s_xattr.
	 */
	if (!strncmp(name, XATTR_SYSTEM_PREFIX, XATTR_SYSTEM_PREFIX_LEN))
		return generic_getxattr(dentry, name, buffer, size);

	err = shmem_xattr_validate(name);
	if (err)
		return err;

	return shmem_xattr_get(dentry, name, buffer, size);
}

static int shmem_setxattr(struct dentry *dentry, const char *name,
			  const void *value, size_t size, int flags)
{
	int err;

	/*
	 * If this is a request for a synthetic attribute in the system.*
	 * namespace use the generic infrastructure to resolve a handler
	 * for it via sb->s_xattr.
	 */
	if (!strncmp(name, XATTR_SYSTEM_PREFIX, XATTR_SYSTEM_PREFIX_LEN))
		return generic_setxattr(dentry, name, value, size, flags);

	err = shmem_xattr_validate(name);
	if (err)
		return err;

	if (size == 0)
		value = "";  /* empty EA, do not remove */

	return shmem_xattr_set(dentry, name, value, size, flags);

}

static int shmem_removexattr(struct dentry *dentry, const char *name)
{
	int err;

	/*
	 * If this is a request for a synthetic attribute in the system.*
	 * namespace use the generic infrastructure to resolve a handler
	 * for it via sb->s_xattr.
	 */
	if (!strncmp(name, XATTR_SYSTEM_PREFIX, XATTR_SYSTEM_PREFIX_LEN))
		return generic_removexattr(dentry, name);

	err = shmem_xattr_validate(name);
	if (err)
		return err;

	return shmem_xattr_set(dentry, name, NULL, 0, XATTR_REPLACE);
}

static bool xattr_is_trusted(const char *name)
{
	return !strncmp(name, XATTR_TRUSTED_PREFIX, XATTR_TRUSTED_PREFIX_LEN);
}

static ssize_t shmem_listxattr(struct dentry *dentry, char *buffer, size_t size)
{
	bool trusted = capable(CAP_SYS_ADMIN);
	struct shmem_xattr *xattr;
	struct shmem_inode_info *info;
	size_t used = 0;

	info = SHMEM_I(dentry->d_inode);

	spin_lock(&info->lock);
	list_for_each_entry(xattr, &info->xattr_list, list) {
		size_t len;

		/* skip "trusted." attributes for unprivileged callers */
		if (!trusted && xattr_is_trusted(xattr->name))
			continue;

		len = strlen(xattr->name) + 1;
		used += len;
		if (buffer) {
			if (size < used) {
				used = -ERANGE;
				break;
			}
			memcpy(buffer, xattr->name, len);
			buffer += len;
		}
	}
	spin_unlock(&info->lock);

	return used;
}
#endif /* CONFIG_TMPFS_XATTR */

1923
static const struct inode_operations shmem_short_symlink_operations = {
1924
	.readlink	= generic_readlink,
1925
	.follow_link	= shmem_follow_short_symlink,
1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942
#ifdef CONFIG_TMPFS_XATTR
	.setxattr	= shmem_setxattr,
	.getxattr	= shmem_getxattr,
	.listxattr	= shmem_listxattr,
	.removexattr	= shmem_removexattr,
#endif
};

static const struct inode_operations shmem_symlink_inode_operations = {
	.readlink	= generic_readlink,
	.follow_link	= shmem_follow_link,
	.put_link	= shmem_put_link,
#ifdef CONFIG_TMPFS_XATTR
	.setxattr	= shmem_setxattr,
	.getxattr	= shmem_getxattr,
	.listxattr	= shmem_listxattr,
	.removexattr	= shmem_removexattr,
1943
#endif
1944
};
1945

1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958
static struct dentry *shmem_get_parent(struct dentry *child)
{
	return ERR_PTR(-ESTALE);
}

static int shmem_match(struct inode *ino, void *vfh)
{
	__u32 *fh = vfh;
	__u64 inum = fh[2];
	inum = (inum << 32) | fh[1];
	return ino->i_ino == inum && fh[0] == ino->i_generation;
}

C
Christoph Hellwig 已提交
1959 1960
static struct dentry *shmem_fh_to_dentry(struct super_block *sb,
		struct fid *fid, int fh_len, int fh_type)
1961 1962
{
	struct inode *inode;
C
Christoph Hellwig 已提交
1963 1964 1965 1966 1967 1968
	struct dentry *dentry = NULL;
	u64 inum = fid->raw[2];
	inum = (inum << 32) | fid->raw[1];

	if (fh_len < 3)
		return NULL;
1969

C
Christoph Hellwig 已提交
1970 1971
	inode = ilookup5(sb, (unsigned long)(inum + fid->raw[0]),
			shmem_match, fid->raw);
1972
	if (inode) {
C
Christoph Hellwig 已提交
1973
		dentry = d_find_alias(inode);
1974 1975 1976
		iput(inode);
	}

C
Christoph Hellwig 已提交
1977
	return dentry;
1978 1979 1980 1981 1982 1983 1984
}

static int shmem_encode_fh(struct dentry *dentry, __u32 *fh, int *len,
				int connectable)
{
	struct inode *inode = dentry->d_inode;

1985 1986
	if (*len < 3) {
		*len = 3;
1987
		return 255;
1988
	}
1989

A
Al Viro 已提交
1990
	if (inode_unhashed(inode)) {
1991 1992 1993 1994 1995 1996 1997
		/* Unfortunately insert_inode_hash is not idempotent,
		 * so as we hash inodes here rather than at creation
		 * time, we need a lock to ensure we only try
		 * to do it once
		 */
		static DEFINE_SPINLOCK(lock);
		spin_lock(&lock);
A
Al Viro 已提交
1998
		if (inode_unhashed(inode))
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
			__insert_inode_hash(inode,
					    inode->i_ino + inode->i_generation);
		spin_unlock(&lock);
	}

	fh[0] = inode->i_generation;
	fh[1] = inode->i_ino;
	fh[2] = ((__u64)inode->i_ino) >> 32;

	*len = 3;
	return 1;
}

2012
static const struct export_operations shmem_export_ops = {
2013 2014
	.get_parent     = shmem_get_parent,
	.encode_fh      = shmem_encode_fh,
C
Christoph Hellwig 已提交
2015
	.fh_to_dentry	= shmem_fh_to_dentry,
2016 2017
};

2018 2019
static int shmem_parse_options(char *options, struct shmem_sb_info *sbinfo,
			       bool remount)
L
Linus Torvalds 已提交
2020 2021 2022
{
	char *this_char, *value, *rest;

2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039
	while (options != NULL) {
		this_char = options;
		for (;;) {
			/*
			 * NUL-terminate this option: unfortunately,
			 * mount options form a comma-separated list,
			 * but mpol's nodelist may also contain commas.
			 */
			options = strchr(options, ',');
			if (options == NULL)
				break;
			options++;
			if (!isdigit(*options)) {
				options[-1] = '\0';
				break;
			}
		}
L
Linus Torvalds 已提交
2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061
		if (!*this_char)
			continue;
		if ((value = strchr(this_char,'=')) != NULL) {
			*value++ = 0;
		} else {
			printk(KERN_ERR
			    "tmpfs: No value for mount option '%s'\n",
			    this_char);
			return 1;
		}

		if (!strcmp(this_char,"size")) {
			unsigned long long size;
			size = memparse(value,&rest);
			if (*rest == '%') {
				size <<= PAGE_SHIFT;
				size *= totalram_pages;
				do_div(size, 100);
				rest++;
			}
			if (*rest)
				goto bad_val;
2062 2063
			sbinfo->max_blocks =
				DIV_ROUND_UP(size, PAGE_CACHE_SIZE);
L
Linus Torvalds 已提交
2064
		} else if (!strcmp(this_char,"nr_blocks")) {
2065
			sbinfo->max_blocks = memparse(value, &rest);
L
Linus Torvalds 已提交
2066 2067 2068
			if (*rest)
				goto bad_val;
		} else if (!strcmp(this_char,"nr_inodes")) {
2069
			sbinfo->max_inodes = memparse(value, &rest);
L
Linus Torvalds 已提交
2070 2071 2072
			if (*rest)
				goto bad_val;
		} else if (!strcmp(this_char,"mode")) {
2073
			if (remount)
L
Linus Torvalds 已提交
2074
				continue;
2075
			sbinfo->mode = simple_strtoul(value, &rest, 8) & 07777;
L
Linus Torvalds 已提交
2076 2077 2078
			if (*rest)
				goto bad_val;
		} else if (!strcmp(this_char,"uid")) {
2079
			if (remount)
L
Linus Torvalds 已提交
2080
				continue;
2081
			sbinfo->uid = simple_strtoul(value, &rest, 0);
L
Linus Torvalds 已提交
2082 2083 2084
			if (*rest)
				goto bad_val;
		} else if (!strcmp(this_char,"gid")) {
2085
			if (remount)
L
Linus Torvalds 已提交
2086
				continue;
2087
			sbinfo->gid = simple_strtoul(value, &rest, 0);
L
Linus Torvalds 已提交
2088 2089
			if (*rest)
				goto bad_val;
2090
		} else if (!strcmp(this_char,"mpol")) {
2091
			if (mpol_parse_str(value, &sbinfo->mpol, 1))
2092
				goto bad_val;
L
Linus Torvalds 已提交
2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110
		} else {
			printk(KERN_ERR "tmpfs: Bad mount option %s\n",
			       this_char);
			return 1;
		}
	}
	return 0;

bad_val:
	printk(KERN_ERR "tmpfs: Bad value '%s' for mount option '%s'\n",
	       value, this_char);
	return 1;

}

static int shmem_remount_fs(struct super_block *sb, int *flags, char *data)
{
	struct shmem_sb_info *sbinfo = SHMEM_SB(sb);
2111
	struct shmem_sb_info config = *sbinfo;
2112 2113 2114
	unsigned long inodes;
	int error = -EINVAL;

2115
	if (shmem_parse_options(data, &config, true))
2116
		return error;
L
Linus Torvalds 已提交
2117

2118 2119
	spin_lock(&sbinfo->stat_lock);
	inodes = sbinfo->max_inodes - sbinfo->free_inodes;
2120
	if (percpu_counter_compare(&sbinfo->used_blocks, config.max_blocks) > 0)
2121
		goto out;
2122
	if (config.max_inodes < inodes)
2123 2124
		goto out;
	/*
2125
	 * Those tests disallow limited->unlimited while any are in use;
2126 2127 2128
	 * but we must separately disallow unlimited->limited, because
	 * in that case we have no record of how much is already in use.
	 */
2129
	if (config.max_blocks && !sbinfo->max_blocks)
2130
		goto out;
2131
	if (config.max_inodes && !sbinfo->max_inodes)
2132 2133 2134
		goto out;

	error = 0;
2135 2136 2137
	sbinfo->max_blocks  = config.max_blocks;
	sbinfo->max_inodes  = config.max_inodes;
	sbinfo->free_inodes = config.max_inodes - inodes;
2138 2139 2140

	mpol_put(sbinfo->mpol);
	sbinfo->mpol        = config.mpol;	/* transfers initial ref */
2141 2142 2143
out:
	spin_unlock(&sbinfo->stat_lock);
	return error;
L
Linus Torvalds 已提交
2144
}
2145

2146
static int shmem_show_options(struct seq_file *seq, struct dentry *root)
2147
{
2148
	struct shmem_sb_info *sbinfo = SHMEM_SB(root->d_sb);
2149 2150 2151 2152 2153 2154 2155

	if (sbinfo->max_blocks != shmem_default_max_blocks())
		seq_printf(seq, ",size=%luk",
			sbinfo->max_blocks << (PAGE_CACHE_SHIFT - 10));
	if (sbinfo->max_inodes != shmem_default_max_inodes())
		seq_printf(seq, ",nr_inodes=%lu", sbinfo->max_inodes);
	if (sbinfo->mode != (S_IRWXUGO | S_ISVTX))
2156
		seq_printf(seq, ",mode=%03ho", sbinfo->mode);
2157 2158 2159 2160
	if (sbinfo->uid != 0)
		seq_printf(seq, ",uid=%u", sbinfo->uid);
	if (sbinfo->gid != 0)
		seq_printf(seq, ",gid=%u", sbinfo->gid);
2161
	shmem_show_mpol(seq, sbinfo->mpol);
2162 2163 2164
	return 0;
}
#endif /* CONFIG_TMPFS */
L
Linus Torvalds 已提交
2165 2166 2167

static void shmem_put_super(struct super_block *sb)
{
2168 2169 2170 2171
	struct shmem_sb_info *sbinfo = SHMEM_SB(sb);

	percpu_counter_destroy(&sbinfo->used_blocks);
	kfree(sbinfo);
L
Linus Torvalds 已提交
2172 2173 2174
	sb->s_fs_info = NULL;
}

2175
int shmem_fill_super(struct super_block *sb, void *data, int silent)
L
Linus Torvalds 已提交
2176 2177
{
	struct inode *inode;
2178
	struct shmem_sb_info *sbinfo;
2179 2180 2181
	int err = -ENOMEM;

	/* Round up to L1_CACHE_BYTES to resist false sharing */
2182
	sbinfo = kzalloc(max((int)sizeof(struct shmem_sb_info),
2183 2184 2185 2186 2187
				L1_CACHE_BYTES), GFP_KERNEL);
	if (!sbinfo)
		return -ENOMEM;

	sbinfo->mode = S_IRWXUGO | S_ISVTX;
2188 2189
	sbinfo->uid = current_fsuid();
	sbinfo->gid = current_fsgid();
2190
	sb->s_fs_info = sbinfo;
L
Linus Torvalds 已提交
2191

2192
#ifdef CONFIG_TMPFS
L
Linus Torvalds 已提交
2193 2194 2195 2196 2197 2198
	/*
	 * Per default we only allow half of the physical ram per
	 * tmpfs instance, limiting inodes to one per page of lowmem;
	 * but the internal instance is left unlimited.
	 */
	if (!(sb->s_flags & MS_NOUSER)) {
2199 2200 2201 2202 2203 2204
		sbinfo->max_blocks = shmem_default_max_blocks();
		sbinfo->max_inodes = shmem_default_max_inodes();
		if (shmem_parse_options(data, sbinfo, false)) {
			err = -EINVAL;
			goto failed;
		}
L
Linus Torvalds 已提交
2205
	}
2206
	sb->s_export_op = &shmem_export_ops;
L
Linus Torvalds 已提交
2207 2208 2209 2210
#else
	sb->s_flags |= MS_NOUSER;
#endif

2211
	spin_lock_init(&sbinfo->stat_lock);
2212 2213
	if (percpu_counter_init(&sbinfo->used_blocks, 0))
		goto failed;
2214
	sbinfo->free_inodes = sbinfo->max_inodes;
2215

2216
	sb->s_maxbytes = MAX_LFS_FILESIZE;
L
Linus Torvalds 已提交
2217 2218 2219 2220
	sb->s_blocksize = PAGE_CACHE_SIZE;
	sb->s_blocksize_bits = PAGE_CACHE_SHIFT;
	sb->s_magic = TMPFS_MAGIC;
	sb->s_op = &shmem_ops;
2221
	sb->s_time_gran = 1;
2222
#ifdef CONFIG_TMPFS_XATTR
2223
	sb->s_xattr = shmem_xattr_handlers;
2224 2225
#endif
#ifdef CONFIG_TMPFS_POSIX_ACL
2226 2227
	sb->s_flags |= MS_POSIXACL;
#endif
2228

2229
	inode = shmem_get_inode(sb, NULL, S_IFDIR | sbinfo->mode, 0, VM_NORESERVE);
L
Linus Torvalds 已提交
2230 2231
	if (!inode)
		goto failed;
2232 2233
	inode->i_uid = sbinfo->uid;
	inode->i_gid = sbinfo->gid;
A
Al Viro 已提交
2234 2235
	sb->s_root = d_make_root(inode);
	if (!sb->s_root)
2236
		goto failed;
L
Linus Torvalds 已提交
2237 2238 2239 2240 2241 2242 2243
	return 0;

failed:
	shmem_put_super(sb);
	return err;
}

2244
static struct kmem_cache *shmem_inode_cachep;
L
Linus Torvalds 已提交
2245 2246 2247

static struct inode *shmem_alloc_inode(struct super_block *sb)
{
2248 2249 2250
	struct shmem_inode_info *info;
	info = kmem_cache_alloc(shmem_inode_cachep, GFP_KERNEL);
	if (!info)
L
Linus Torvalds 已提交
2251
		return NULL;
2252
	return &info->vfs_inode;
L
Linus Torvalds 已提交
2253 2254
}

2255
static void shmem_destroy_callback(struct rcu_head *head)
N
Nick Piggin 已提交
2256 2257 2258 2259 2260
{
	struct inode *inode = container_of(head, struct inode, i_rcu);
	kmem_cache_free(shmem_inode_cachep, SHMEM_I(inode));
}

L
Linus Torvalds 已提交
2261 2262
static void shmem_destroy_inode(struct inode *inode)
{
2263
	if (S_ISREG(inode->i_mode))
L
Linus Torvalds 已提交
2264
		mpol_free_shared_policy(&SHMEM_I(inode)->policy);
2265
	call_rcu(&inode->i_rcu, shmem_destroy_callback);
L
Linus Torvalds 已提交
2266 2267
}

2268
static void shmem_init_inode(void *foo)
L
Linus Torvalds 已提交
2269
{
2270 2271
	struct shmem_inode_info *info = foo;
	inode_init_once(&info->vfs_inode);
L
Linus Torvalds 已提交
2272 2273
}

2274
static int shmem_init_inodecache(void)
L
Linus Torvalds 已提交
2275 2276 2277
{
	shmem_inode_cachep = kmem_cache_create("shmem_inode_cache",
				sizeof(struct shmem_inode_info),
2278
				0, SLAB_PANIC, shmem_init_inode);
L
Linus Torvalds 已提交
2279 2280 2281
	return 0;
}

2282
static void shmem_destroy_inodecache(void)
L
Linus Torvalds 已提交
2283
{
2284
	kmem_cache_destroy(shmem_inode_cachep);
L
Linus Torvalds 已提交
2285 2286
}

2287
static const struct address_space_operations shmem_aops = {
L
Linus Torvalds 已提交
2288
	.writepage	= shmem_writepage,
2289
	.set_page_dirty	= __set_page_dirty_no_writeback,
L
Linus Torvalds 已提交
2290
#ifdef CONFIG_TMPFS
N
Nick Piggin 已提交
2291 2292
	.write_begin	= shmem_write_begin,
	.write_end	= shmem_write_end,
L
Linus Torvalds 已提交
2293
#endif
2294
	.migratepage	= migrate_page,
2295
	.error_remove_page = generic_error_remove_page,
L
Linus Torvalds 已提交
2296 2297
};

2298
static const struct file_operations shmem_file_operations = {
L
Linus Torvalds 已提交
2299 2300 2301
	.mmap		= shmem_mmap,
#ifdef CONFIG_TMPFS
	.llseek		= generic_file_llseek,
H
Hugh Dickins 已提交
2302
	.read		= do_sync_read,
H
Hugh Dickins 已提交
2303
	.write		= do_sync_write,
H
Hugh Dickins 已提交
2304
	.aio_read	= shmem_file_aio_read,
H
Hugh Dickins 已提交
2305
	.aio_write	= generic_file_aio_write,
2306
	.fsync		= noop_fsync,
2307
	.splice_read	= shmem_file_splice_read,
2308
	.splice_write	= generic_file_splice_write,
L
Linus Torvalds 已提交
2309 2310 2311
#endif
};

2312
static const struct inode_operations shmem_inode_operations = {
2313
	.setattr	= shmem_setattr,
2314
	.truncate_range	= shmem_truncate_range,
2315 2316 2317 2318 2319 2320
#ifdef CONFIG_TMPFS_XATTR
	.setxattr	= shmem_setxattr,
	.getxattr	= shmem_getxattr,
	.listxattr	= shmem_listxattr,
	.removexattr	= shmem_removexattr,
#endif
L
Linus Torvalds 已提交
2321 2322
};

2323
static const struct inode_operations shmem_dir_inode_operations = {
L
Linus Torvalds 已提交
2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334
#ifdef CONFIG_TMPFS
	.create		= shmem_create,
	.lookup		= simple_lookup,
	.link		= shmem_link,
	.unlink		= shmem_unlink,
	.symlink	= shmem_symlink,
	.mkdir		= shmem_mkdir,
	.rmdir		= shmem_rmdir,
	.mknod		= shmem_mknod,
	.rename		= shmem_rename,
#endif
2335 2336 2337 2338 2339 2340
#ifdef CONFIG_TMPFS_XATTR
	.setxattr	= shmem_setxattr,
	.getxattr	= shmem_getxattr,
	.listxattr	= shmem_listxattr,
	.removexattr	= shmem_removexattr,
#endif
2341
#ifdef CONFIG_TMPFS_POSIX_ACL
2342
	.setattr	= shmem_setattr,
2343 2344 2345
#endif
};

2346
static const struct inode_operations shmem_special_inode_operations = {
2347 2348 2349 2350 2351 2352
#ifdef CONFIG_TMPFS_XATTR
	.setxattr	= shmem_setxattr,
	.getxattr	= shmem_getxattr,
	.listxattr	= shmem_listxattr,
	.removexattr	= shmem_removexattr,
#endif
2353
#ifdef CONFIG_TMPFS_POSIX_ACL
2354
	.setattr	= shmem_setattr,
2355
#endif
L
Linus Torvalds 已提交
2356 2357
};

2358
static const struct super_operations shmem_ops = {
L
Linus Torvalds 已提交
2359 2360 2361 2362 2363
	.alloc_inode	= shmem_alloc_inode,
	.destroy_inode	= shmem_destroy_inode,
#ifdef CONFIG_TMPFS
	.statfs		= shmem_statfs,
	.remount_fs	= shmem_remount_fs,
2364
	.show_options	= shmem_show_options,
L
Linus Torvalds 已提交
2365
#endif
A
Al Viro 已提交
2366
	.evict_inode	= shmem_evict_inode,
L
Linus Torvalds 已提交
2367 2368 2369 2370
	.drop_inode	= generic_delete_inode,
	.put_super	= shmem_put_super,
};

2371
static const struct vm_operations_struct shmem_vm_ops = {
2372
	.fault		= shmem_fault,
L
Linus Torvalds 已提交
2373 2374 2375 2376 2377 2378
#ifdef CONFIG_NUMA
	.set_policy     = shmem_set_policy,
	.get_policy     = shmem_get_policy,
#endif
};

A
Al Viro 已提交
2379 2380
static struct dentry *shmem_mount(struct file_system_type *fs_type,
	int flags, const char *dev_name, void *data)
L
Linus Torvalds 已提交
2381
{
A
Al Viro 已提交
2382
	return mount_nodev(fs_type, flags, data, shmem_fill_super);
L
Linus Torvalds 已提交
2383 2384
}

2385
static struct file_system_type shmem_fs_type = {
L
Linus Torvalds 已提交
2386 2387
	.owner		= THIS_MODULE,
	.name		= "tmpfs",
A
Al Viro 已提交
2388
	.mount		= shmem_mount,
L
Linus Torvalds 已提交
2389 2390 2391
	.kill_sb	= kill_litter_super,
};

2392
int __init shmem_init(void)
L
Linus Torvalds 已提交
2393 2394 2395
{
	int error;

P
Peter Zijlstra 已提交
2396 2397 2398 2399
	error = bdi_init(&shmem_backing_dev_info);
	if (error)
		goto out4;

2400
	error = shmem_init_inodecache();
L
Linus Torvalds 已提交
2401 2402 2403
	if (error)
		goto out3;

2404
	error = register_filesystem(&shmem_fs_type);
L
Linus Torvalds 已提交
2405 2406 2407 2408
	if (error) {
		printk(KERN_ERR "Could not register tmpfs\n");
		goto out2;
	}
2409

2410 2411
	shm_mnt = vfs_kern_mount(&shmem_fs_type, MS_NOUSER,
				 shmem_fs_type.name, NULL);
L
Linus Torvalds 已提交
2412 2413 2414 2415 2416 2417 2418 2419
	if (IS_ERR(shm_mnt)) {
		error = PTR_ERR(shm_mnt);
		printk(KERN_ERR "Could not kern_mount tmpfs\n");
		goto out1;
	}
	return 0;

out1:
2420
	unregister_filesystem(&shmem_fs_type);
L
Linus Torvalds 已提交
2421
out2:
2422
	shmem_destroy_inodecache();
L
Linus Torvalds 已提交
2423
out3:
P
Peter Zijlstra 已提交
2424 2425
	bdi_destroy(&shmem_backing_dev_info);
out4:
L
Linus Torvalds 已提交
2426 2427 2428
	shm_mnt = ERR_PTR(error);
	return error;
}
2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442

#else /* !CONFIG_SHMEM */

/*
 * tiny-shmem: simple shmemfs and tmpfs using ramfs code
 *
 * This is intended for small system where the benefits of the full
 * shmem code (swap-backed and resource-limited) are outweighed by
 * their complexity. On systems without swap this code should be
 * effectively equivalent, but much lighter weight.
 */

#include <linux/ramfs.h>

2443
static struct file_system_type shmem_fs_type = {
2444
	.name		= "tmpfs",
A
Al Viro 已提交
2445
	.mount		= ramfs_mount,
2446 2447 2448
	.kill_sb	= kill_litter_super,
};

2449
int __init shmem_init(void)
2450
{
2451
	BUG_ON(register_filesystem(&shmem_fs_type) != 0);
2452

2453
	shm_mnt = kern_mount(&shmem_fs_type);
2454 2455 2456 2457 2458
	BUG_ON(IS_ERR(shm_mnt));

	return 0;
}

2459
int shmem_unuse(swp_entry_t swap, struct page *page)
2460 2461 2462 2463
{
	return 0;
}

H
Hugh Dickins 已提交
2464 2465 2466 2467 2468
int shmem_lock(struct file *file, int lock, struct user_struct *user)
{
	return 0;
}

2469 2470 2471 2472
void shmem_unlock_mapping(struct address_space *mapping)
{
}

2473
void shmem_truncate_range(struct inode *inode, loff_t lstart, loff_t lend)
2474
{
2475
	truncate_inode_pages_range(inode->i_mapping, lstart, lend);
2476 2477 2478
}
EXPORT_SYMBOL_GPL(shmem_truncate_range);

2479 2480
#define shmem_vm_ops				generic_file_vm_ops
#define shmem_file_operations			ramfs_file_operations
2481
#define shmem_get_inode(sb, dir, mode, dev, flags)	ramfs_get_inode(sb, dir, mode, dev)
2482 2483
#define shmem_acct_size(flags, size)		0
#define shmem_unacct_size(flags, size)		do {} while (0)
2484 2485 2486 2487

#endif /* CONFIG_SHMEM */

/* common code */
L
Linus Torvalds 已提交
2488

2489
/**
L
Linus Torvalds 已提交
2490 2491 2492
 * shmem_file_setup - get an unlinked file living in tmpfs
 * @name: name for dentry (to be seen in /proc/<pid>/maps
 * @size: size to be set for the file
2493
 * @flags: VM_NORESERVE suppresses pre-accounting of the entire object size
L
Linus Torvalds 已提交
2494
 */
2495
struct file *shmem_file_setup(const char *name, loff_t size, unsigned long flags)
L
Linus Torvalds 已提交
2496 2497 2498 2499
{
	int error;
	struct file *file;
	struct inode *inode;
2500 2501
	struct path path;
	struct dentry *root;
L
Linus Torvalds 已提交
2502 2503 2504 2505 2506
	struct qstr this;

	if (IS_ERR(shm_mnt))
		return (void *)shm_mnt;

2507
	if (size < 0 || size > MAX_LFS_FILESIZE)
L
Linus Torvalds 已提交
2508 2509 2510 2511 2512 2513 2514 2515 2516 2517
		return ERR_PTR(-EINVAL);

	if (shmem_acct_size(flags, size))
		return ERR_PTR(-ENOMEM);

	error = -ENOMEM;
	this.name = name;
	this.len = strlen(name);
	this.hash = 0; /* will go */
	root = shm_mnt->mnt_root;
2518 2519
	path.dentry = d_alloc(root, &this);
	if (!path.dentry)
L
Linus Torvalds 已提交
2520
		goto put_memory;
2521
	path.mnt = mntget(shm_mnt);
L
Linus Torvalds 已提交
2522 2523

	error = -ENOSPC;
2524
	inode = shmem_get_inode(root->d_sb, NULL, S_IFREG | S_IRWXUGO, 0, flags);
L
Linus Torvalds 已提交
2525
	if (!inode)
2526
		goto put_dentry;
L
Linus Torvalds 已提交
2527

2528
	d_instantiate(path.dentry, inode);
L
Linus Torvalds 已提交
2529
	inode->i_size = size;
2530
	clear_nlink(inode);	/* It is unlinked */
2531 2532 2533
#ifndef CONFIG_MMU
	error = ramfs_nommu_expand_for_mapping(inode, size);
	if (error)
2534
		goto put_dentry;
2535
#endif
2536 2537

	error = -ENFILE;
2538
	file = alloc_file(&path, FMODE_WRITE | FMODE_READ,
2539 2540 2541 2542
		  &shmem_file_operations);
	if (!file)
		goto put_dentry;

L
Linus Torvalds 已提交
2543 2544 2545
	return file;

put_dentry:
2546
	path_put(&path);
L
Linus Torvalds 已提交
2547 2548 2549 2550
put_memory:
	shmem_unacct_size(flags, size);
	return ERR_PTR(error);
}
2551
EXPORT_SYMBOL_GPL(shmem_file_setup);
L
Linus Torvalds 已提交
2552

2553
/**
L
Linus Torvalds 已提交
2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569
 * shmem_zero_setup - setup a shared anonymous mapping
 * @vma: the vma to be mmapped is prepared by do_mmap_pgoff
 */
int shmem_zero_setup(struct vm_area_struct *vma)
{
	struct file *file;
	loff_t size = vma->vm_end - vma->vm_start;

	file = shmem_file_setup("dev/zero", size, vma->vm_flags);
	if (IS_ERR(file))
		return PTR_ERR(file);

	if (vma->vm_file)
		fput(vma->vm_file);
	vma->vm_file = file;
	vma->vm_ops = &shmem_vm_ops;
2570
	vma->vm_flags |= VM_CAN_NONLINEAR;
L
Linus Torvalds 已提交
2571 2572
	return 0;
}
2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585

/**
 * shmem_read_mapping_page_gfp - read into page cache, using specified page allocation flags.
 * @mapping:	the page's address_space
 * @index:	the page index
 * @gfp:	the page allocator flags to use if allocating
 *
 * This behaves as a tmpfs "read_cache_page_gfp(mapping, index, gfp)",
 * with any new page allocations done using the specified allocation flags.
 * But read_cache_page_gfp() uses the ->readpage() method: which does not
 * suit tmpfs, since it may have pages in swapcache, and needs to find those
 * for itself; although drivers/gpu/drm i915 and ttm rely upon this support.
 *
2586 2587
 * i915_gem_object_get_pages_gtt() mixes __GFP_NORETRY | __GFP_NOWARN in
 * with the mapping_gfp_mask(), to avoid OOMing the machine unnecessarily.
2588 2589 2590 2591
 */
struct page *shmem_read_mapping_page_gfp(struct address_space *mapping,
					 pgoff_t index, gfp_t gfp)
{
2592 2593
#ifdef CONFIG_SHMEM
	struct inode *inode = mapping->host;
H
Hugh Dickins 已提交
2594
	struct page *page;
2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607
	int error;

	BUG_ON(mapping->a_ops != &shmem_aops);
	error = shmem_getpage_gfp(inode, index, &page, SGP_CACHE, gfp, NULL);
	if (error)
		page = ERR_PTR(error);
	else
		unlock_page(page);
	return page;
#else
	/*
	 * The tiny !SHMEM case uses ramfs without swap
	 */
2608
	return read_cache_page_gfp(mapping, index, gfp);
2609
#endif
2610 2611
}
EXPORT_SYMBOL_GPL(shmem_read_mapping_page_gfp);