shmem.c 76.8 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/falloc.h>
57
#include <linux/splice.h>
L
Linus Torvalds 已提交
58 59 60 61
#include <linux/security.h>
#include <linux/swapops.h>
#include <linux/mempolicy.h>
#include <linux/namei.h>
62
#include <linux/ctype.h>
63
#include <linux/migrate.h>
64
#include <linux/highmem.h>
65
#include <linux/seq_file.h>
M
Mimi Zohar 已提交
66
#include <linux/magic.h>
67

L
Linus Torvalds 已提交
68 69 70
#include <asm/uaccess.h>
#include <asm/pgtable.h>

H
Hugh Dickins 已提交
71
#define BLOCKS_PER_PAGE  (PAGE_CACHE_SIZE/512)
L
Linus Torvalds 已提交
72 73 74 75 76
#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

77 78 79
/* Symlink up to this size is kmalloc'ed instead of using a swappable page */
#define SHORT_SYMLINK_LEN 128

80 81 82 83 84 85 86
struct shmem_xattr {
	struct list_head list;	/* anchored by shmem_inode_info->xattr_list */
	char *name;		/* xattr name */
	size_t size;
	char value[0];
};

87 88 89 90 91 92 93 94 95 96 97 98
/*
 * shmem_fallocate and shmem_writepage communicate via inode->i_private
 * (with i_mutex making sure that it has only one user at a time):
 * we would prefer not to enlarge the shmem inode just for that.
 */
struct shmem_falloc {
	pgoff_t start;		/* start of range currently being fallocated */
	pgoff_t next;		/* the next page offset to be fallocated */
	pgoff_t nr_falloced;	/* how many new pages have been fallocated */
	pgoff_t nr_unswapped;	/* how often writepage refused to swap out */
};

99
/* Flag allocation requirements to shmem_getpage */
L
Linus Torvalds 已提交
100 101 102
enum sgp_type {
	SGP_READ,	/* don't exceed i_size, don't allocate page */
	SGP_CACHE,	/* don't exceed i_size, may allocate page */
103
	SGP_DIRTY,	/* like SGP_CACHE, but set new page dirty */
104 105
	SGP_WRITE,	/* may exceed i_size, may allocate !Uptodate page */
	SGP_FALLOC,	/* like SGP_WRITE, but make existing page Uptodate */
L
Linus Torvalds 已提交
106 107
};

A
Andrew Morton 已提交
108
#ifdef CONFIG_TMPFS
109 110 111 112 113 114 115 116 117
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
118
#endif
119

120 121 122
static bool shmem_should_replace_page(struct page *page, gfp_t gfp);
static int shmem_replace_page(struct page **pagep, gfp_t gfp,
				struct shmem_inode_info *info, pgoff_t index);
123 124 125 126 127 128 129 130 131
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
132 133 134 135 136 137 138 139 140 141 142 143 144 145

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)
{
146
	return (flags & VM_NORESERVE) ?
147
		0 : security_vm_enough_memory_mm(current->mm, VM_ACCT(size));
L
Linus Torvalds 已提交
148 149 150 151
}

static inline void shmem_unacct_size(unsigned long flags, loff_t size)
{
152
	if (!(flags & VM_NORESERVE))
L
Linus Torvalds 已提交
153 154 155 156 157 158 159 160 161 162 163
		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)
{
164
	return (flags & VM_NORESERVE) ?
165
		security_vm_enough_memory_mm(current->mm, VM_ACCT(PAGE_CACHE_SIZE)) : 0;
L
Linus Torvalds 已提交
166 167 168 169
}

static inline void shmem_unacct_blocks(unsigned long flags, long pages)
{
170
	if (flags & VM_NORESERVE)
L
Linus Torvalds 已提交
171 172 173
		vm_unacct_memory(pages * VM_ACCT(PAGE_CACHE_SIZE));
}

174
static const struct super_operations shmem_ops;
175
static const struct address_space_operations shmem_aops;
176
static const struct file_operations shmem_file_operations;
177 178 179
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;
180
static const struct vm_operations_struct shmem_vm_ops;
L
Linus Torvalds 已提交
181

182
static struct backing_dev_info shmem_backing_dev_info  __read_mostly = {
L
Linus Torvalds 已提交
183
	.ra_pages	= 0,	/* No readahead */
184
	.capabilities	= BDI_CAP_NO_ACCT_AND_WRITEBACK | BDI_CAP_SWAP_BACKED,
L
Linus Torvalds 已提交
185 186 187
};

static LIST_HEAD(shmem_swaplist);
188
static DEFINE_MUTEX(shmem_swaplist_mutex);
L
Linus Torvalds 已提交
189

190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214
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);
	}
}

215
/**
216
 * shmem_recalc_inode - recalculate the block usage of an inode
L
Linus Torvalds 已提交
217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233
 * @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) {
234 235 236
		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
237
		info->alloced -= freed;
238
		inode->i_blocks -= freed * BLOCKS_PER_PAGE;
L
Linus Torvalds 已提交
239 240 241 242
		shmem_unacct_blocks(info->flags, freed);
	}
}

243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265
/*
 * 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;
}

266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283
/*
 * Sometimes, before we decide whether to proceed or to fail, we must check
 * that an entry was not already brought back from swap by a racing thread.
 *
 * Checking page is not enough: by the time a SwapCache page is locked, it
 * might be reused, and again be SwapCache, using the same swap as before.
 */
static bool shmem_confirm_swap(struct address_space *mapping,
			       pgoff_t index, swp_entry_t swap)
{
	void *item;

	rcu_read_lock();
	item = radix_tree_lookup(&mapping->page_tree, index);
	rcu_read_unlock();
	return item == swp_to_radix_entry(swap);
}

284 285 286 287 288 289 290
/*
 * 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)
{
291
	int error;
292 293 294 295

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

296 297 298 299 300
	page_cache_get(page);
	page->mapping = mapping;
	page->index = index;

	spin_lock_irq(&mapping->tree_lock);
301
	if (!expected)
302 303 304 305
		error = radix_tree_insert(&mapping->page_tree, index, page);
	else
		error = shmem_radix_tree_replace(mapping, index, expected,
								 page);
306
	if (!error) {
307 308 309 310 311 312 313 314
		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);
315 316 317 318
	}
	return error;
}

319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337
/*
 * 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);
}

338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360
/*
 * 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)) {
361 362 363 364 365 366 367 368
			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;
369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407
		}
		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.
 */
408
static void shmem_deswap_pagevec(struct pagevec *pvec)
409 410 411 412 413 414 415 416 417
{
	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;
418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447
}

/*
 * 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();
	}
448 449 450 451
}

/*
 * Remove range of pages and swap entries from radix tree, and free them.
452
 * If !unfalloc, truncate or punch hole; if unfalloc, undo failed fallocate.
453
 */
454 455
static void shmem_undo_range(struct inode *inode, loff_t lstart, loff_t lend,
								 bool unfalloc)
L
Linus Torvalds 已提交
456
{
457
	struct address_space *mapping = inode->i_mapping;
L
Linus Torvalds 已提交
458
	struct shmem_inode_info *info = SHMEM_I(inode);
459
	pgoff_t start = (lstart + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
460 461 462
	pgoff_t end = (lend + 1) >> PAGE_CACHE_SHIFT;
	unsigned int partial_start = lstart & (PAGE_CACHE_SIZE - 1);
	unsigned int partial_end = (lend + 1) & (PAGE_CACHE_SIZE - 1);
463
	struct pagevec pvec;
464 465
	pgoff_t indices[PAGEVEC_SIZE];
	long nr_swaps_freed = 0;
466
	pgoff_t index;
467 468
	int i;

469 470
	if (lend == -1)
		end = -1;	/* unsigned, so actually very big */
471 472 473

	pagevec_init(&pvec, 0);
	index = start;
474
	while (index < end) {
475
		pvec.nr = shmem_find_get_pages_and_swap(mapping, index,
476
				min(end - index, (pgoff_t)PAGEVEC_SIZE),
477 478 479
							pvec.pages, indices);
		if (!pvec.nr)
			break;
480 481 482 483
		mem_cgroup_uncharge_start();
		for (i = 0; i < pagevec_count(&pvec); i++) {
			struct page *page = pvec.pages[i];

484
			index = indices[i];
485
			if (index >= end)
486 487
				break;

488
			if (radix_tree_exceptional_entry(page)) {
489 490
				if (unfalloc)
					continue;
491 492
				nr_swaps_freed += !shmem_free_swap(mapping,
								index, page);
493
				continue;
494 495 496
			}

			if (!trylock_page(page))
497
				continue;
498 499 500 501 502
			if (!unfalloc || !PageUptodate(page)) {
				if (page->mapping == mapping) {
					VM_BUG_ON(PageWriteback(page));
					truncate_inode_page(mapping, page);
				}
503 504 505
			}
			unlock_page(page);
		}
506 507
		shmem_deswap_pagevec(&pvec);
		pagevec_release(&pvec);
508 509 510 511
		mem_cgroup_uncharge_end();
		cond_resched();
		index++;
	}
L
Linus Torvalds 已提交
512

513
	if (partial_start) {
514 515 516
		struct page *page = NULL;
		shmem_getpage(inode, start - 1, &page, SGP_READ, NULL);
		if (page) {
517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532
			unsigned int top = PAGE_CACHE_SIZE;
			if (start > end) {
				top = partial_end;
				partial_end = 0;
			}
			zero_user_segment(page, partial_start, top);
			set_page_dirty(page);
			unlock_page(page);
			page_cache_release(page);
		}
	}
	if (partial_end) {
		struct page *page = NULL;
		shmem_getpage(inode, end, &page, SGP_READ, NULL);
		if (page) {
			zero_user_segment(page, 0, partial_end);
533 534 535 536 537
			set_page_dirty(page);
			unlock_page(page);
			page_cache_release(page);
		}
	}
538 539
	if (start >= end)
		return;
540 541 542 543

	index = start;
	for ( ; ; ) {
		cond_resched();
544
		pvec.nr = shmem_find_get_pages_and_swap(mapping, index,
545
				min(end - index, (pgoff_t)PAGEVEC_SIZE),
546 547
							pvec.pages, indices);
		if (!pvec.nr) {
548
			if (index == start || unfalloc)
549 550 551 552
				break;
			index = start;
			continue;
		}
553
		if ((index == start || unfalloc) && indices[0] >= end) {
554 555
			shmem_deswap_pagevec(&pvec);
			pagevec_release(&pvec);
556 557 558 559 560 561
			break;
		}
		mem_cgroup_uncharge_start();
		for (i = 0; i < pagevec_count(&pvec); i++) {
			struct page *page = pvec.pages[i];

562
			index = indices[i];
563
			if (index >= end)
564 565
				break;

566
			if (radix_tree_exceptional_entry(page)) {
567 568
				if (unfalloc)
					continue;
569 570 571 572 573
				nr_swaps_freed += !shmem_free_swap(mapping,
								index, page);
				continue;
			}

574
			lock_page(page);
575 576 577 578 579
			if (!unfalloc || !PageUptodate(page)) {
				if (page->mapping == mapping) {
					VM_BUG_ON(PageWriteback(page));
					truncate_inode_page(mapping, page);
				}
580
			}
581 582
			unlock_page(page);
		}
583 584
		shmem_deswap_pagevec(&pvec);
		pagevec_release(&pvec);
585 586 587
		mem_cgroup_uncharge_end();
		index++;
	}
588

L
Linus Torvalds 已提交
589
	spin_lock(&info->lock);
590
	info->swapped -= nr_swaps_freed;
L
Linus Torvalds 已提交
591 592
	shmem_recalc_inode(inode);
	spin_unlock(&info->lock);
593
}
L
Linus Torvalds 已提交
594

595 596 597
void shmem_truncate_range(struct inode *inode, loff_t lstart, loff_t lend)
{
	shmem_undo_range(inode, lstart, lend, false);
598
	inode->i_ctime = inode->i_mtime = CURRENT_TIME;
L
Linus Torvalds 已提交
599
}
600
EXPORT_SYMBOL_GPL(shmem_truncate_range);
L
Linus Torvalds 已提交
601

602
static int shmem_setattr(struct dentry *dentry, struct iattr *attr)
L
Linus Torvalds 已提交
603 604 605 606
{
	struct inode *inode = dentry->d_inode;
	int error;

607 608 609 610
	error = inode_change_ok(inode, attr);
	if (error)
		return error;

611 612 613
	if (S_ISREG(inode->i_mode) && (attr->ia_valid & ATTR_SIZE)) {
		loff_t oldsize = inode->i_size;
		loff_t newsize = attr->ia_size;
614

615 616 617 618 619 620 621 622 623 624 625
		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
626 627
	}

628
	setattr_copy(inode, attr);
629
#ifdef CONFIG_TMPFS_POSIX_ACL
630
	if (attr->ia_valid & ATTR_MODE)
631
		error = generic_acl_chmod(inode);
632
#endif
L
Linus Torvalds 已提交
633 634 635
	return error;
}

A
Al Viro 已提交
636
static void shmem_evict_inode(struct inode *inode)
L
Linus Torvalds 已提交
637 638
{
	struct shmem_inode_info *info = SHMEM_I(inode);
639
	struct shmem_xattr *xattr, *nxattr;
L
Linus Torvalds 已提交
640

641
	if (inode->i_mapping->a_ops == &shmem_aops) {
L
Linus Torvalds 已提交
642 643
		shmem_unacct_size(info->flags, inode->i_size);
		inode->i_size = 0;
644
		shmem_truncate_range(inode, 0, (loff_t)-1);
L
Linus Torvalds 已提交
645
		if (!list_empty(&info->swaplist)) {
646
			mutex_lock(&shmem_swaplist_mutex);
L
Linus Torvalds 已提交
647
			list_del_init(&info->swaplist);
648
			mutex_unlock(&shmem_swaplist_mutex);
L
Linus Torvalds 已提交
649
		}
650 651
	} else
		kfree(info->symlink);
652 653 654 655 656

	list_for_each_entry_safe(xattr, nxattr, &info->xattr_list, list) {
		kfree(xattr->name);
		kfree(xattr);
	}
657
	BUG_ON(inode->i_blocks);
658
	shmem_free_inode(inode->i_sb);
659
	clear_inode(inode);
L
Linus Torvalds 已提交
660 661
}

662 663 664
/*
 * If swap found in inode, free it and move page from swapcache to filecache.
 */
665
static int shmem_unuse_inode(struct shmem_inode_info *info,
666
			     swp_entry_t swap, struct page **pagep)
L
Linus Torvalds 已提交
667
{
668
	struct address_space *mapping = info->vfs_inode.i_mapping;
669
	void *radswap;
670
	pgoff_t index;
671 672
	gfp_t gfp;
	int error = 0;
L
Linus Torvalds 已提交
673

674
	radswap = swp_to_radix_entry(swap);
675
	index = radix_tree_locate_item(&mapping->page_tree, radswap);
676
	if (index == -1)
677
		return 0;
678

H
Hugh Dickins 已提交
679 680
	/*
	 * Move _head_ to start search for next from here.
A
Al Viro 已提交
681
	 * But be careful: shmem_evict_inode checks list_empty without taking
H
Hugh Dickins 已提交
682
	 * mutex, and there's an instant in list_move_tail when info->swaplist
683
	 * would appear empty, if it were the only one on shmem_swaplist.
H
Hugh Dickins 已提交
684 685 686
	 */
	if (shmem_swaplist.next != &info->swaplist)
		list_move_tail(&shmem_swaplist, &info->swaplist);
687

688 689 690 691 692 693 694
	gfp = mapping_gfp_mask(mapping);
	if (shmem_should_replace_page(*pagep, gfp)) {
		mutex_unlock(&shmem_swaplist_mutex);
		error = shmem_replace_page(pagep, gfp, info, index);
		mutex_lock(&shmem_swaplist_mutex);
		/*
		 * We needed to drop mutex to make that restrictive page
695 696 697 698 699 700 701 702 703 704 705 706 707 708 709
		 * allocation, but the inode might have been freed while we
		 * dropped it: although a racing shmem_evict_inode() cannot
		 * complete without emptying the radix_tree, our page lock
		 * on this swapcache page is not enough to prevent that -
		 * free_swap_and_cache() of our swap entry will only
		 * trylock_page(), removing swap from radix_tree whatever.
		 *
		 * We must not proceed to shmem_add_to_page_cache() if the
		 * inode has been freed, but of course we cannot rely on
		 * inode or mapping or info to check that.  However, we can
		 * safely check if our swap entry is still in use (and here
		 * it can't have got reused for another page): if it's still
		 * in use, then the inode cannot have been freed yet, and we
		 * can safely proceed (if it's no longer in use, that tells
		 * nothing about the inode, but we don't need to unuse swap).
710 711 712 713 714
		 */
		if (!page_swapcount(*pagep))
			error = -ENOENT;
	}

K
KAMEZAWA Hiroyuki 已提交
715
	/*
716 717 718
	 * 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
719
	 */
720 721
	if (!error)
		error = shmem_add_to_page_cache(*pagep, mapping, index,
722
						GFP_NOWAIT, radswap);
723
	if (error != -ENOMEM) {
724 725 726 727
		/*
		 * Truncation and eviction use free_swap_and_cache(), which
		 * only does trylock page: if we raced, best clean up here.
		 */
728 729
		delete_from_swap_cache(*pagep);
		set_page_dirty(*pagep);
730 731 732 733 734 735
		if (!error) {
			spin_lock(&info->lock);
			info->swapped--;
			spin_unlock(&info->lock);
			swap_free(swap);
		}
736
		error = 1;	/* not an error, but entry was found */
L
Linus Torvalds 已提交
737
	}
738
	return error;
L
Linus Torvalds 已提交
739 740 741
}

/*
742
 * Search through swapped inodes to find and replace swap by page.
L
Linus Torvalds 已提交
743
 */
744
int shmem_unuse(swp_entry_t swap, struct page *page)
L
Linus Torvalds 已提交
745
{
746
	struct list_head *this, *next;
L
Linus Torvalds 已提交
747 748
	struct shmem_inode_info *info;
	int found = 0;
749 750 751 752
	int error = 0;

	/*
	 * There's a faint possibility that swap page was replaced before
753
	 * caller locked it: caller will come back later with the right page.
754
	 */
755
	if (unlikely(!PageSwapCache(page) || page_private(page) != swap.val))
756
		goto out;
757 758 759 760 761 762 763 764 765

	/*
	 * 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;
766
	/* No radix_tree_preload: swap entry keeps a place for page in tree */
L
Linus Torvalds 已提交
767

768
	mutex_lock(&shmem_swaplist_mutex);
769 770
	list_for_each_safe(this, next, &shmem_swaplist) {
		info = list_entry(this, struct shmem_inode_info, swaplist);
771
		if (info->swapped)
772
			found = shmem_unuse_inode(info, swap, &page);
773 774
		else
			list_del_init(&info->swaplist);
775
		cond_resched();
776
		if (found)
777
			break;
L
Linus Torvalds 已提交
778
	}
779
	mutex_unlock(&shmem_swaplist_mutex);
780 781 782 783

	if (found < 0)
		error = found;
out:
H
Hugh Dickins 已提交
784 785
	unlock_page(page);
	page_cache_release(page);
786
	return error;
L
Linus Torvalds 已提交
787 788 789 790 791 792 793 794 795 796
}

/*
 * 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;
797 798
	swp_entry_t swap;
	pgoff_t index;
L
Linus Torvalds 已提交
799 800 801 802 803 804 805 806

	BUG_ON(!PageLocked(page));
	mapping = page->mapping;
	index = page->index;
	inode = mapping->host;
	info = SHMEM_I(inode);
	if (info->flags & VM_LOCKED)
		goto redirty;
807
	if (!total_swap_pages)
L
Linus Torvalds 已提交
808 809
		goto redirty;

810 811 812
	/*
	 * shmem_backing_dev_info's capabilities prevent regular writeback or
	 * sync from ever calling shmem_writepage; but a stacking filesystem
813
	 * might use ->writepage of its underlying filesystem, in which case
814
	 * tmpfs should write out to swap only in response to memory pressure,
815
	 * and not for the writeback threads or sync.
816
	 */
817 818 819 820
	if (!wbc->for_reclaim) {
		WARN_ON_ONCE(1);	/* Still happens? Tell us about it! */
		goto redirty;
	}
821 822 823 824 825

	/*
	 * This is somewhat ridiculous, but without plumbing a SWAP_MAP_FALLOC
	 * value into swapfile.c, the only way we can correctly account for a
	 * fallocated page arriving here is now to initialize it and write it.
826 827 828 829 830 831
	 *
	 * That's okay for a page already fallocated earlier, but if we have
	 * not yet completed the fallocation, then (a) we want to keep track
	 * of this page in case we have to undo it, and (b) it may not be a
	 * good idea to continue anyway, once we're pushing into swap.  So
	 * reactivate the page, and let shmem_fallocate() quit when too many.
832 833
	 */
	if (!PageUptodate(page)) {
834 835 836 837 838 839 840 841 842 843 844 845 846 847
		if (inode->i_private) {
			struct shmem_falloc *shmem_falloc;
			spin_lock(&inode->i_lock);
			shmem_falloc = inode->i_private;
			if (shmem_falloc &&
			    index >= shmem_falloc->start &&
			    index < shmem_falloc->next)
				shmem_falloc->nr_unswapped++;
			else
				shmem_falloc = NULL;
			spin_unlock(&inode->i_lock);
			if (shmem_falloc)
				goto redirty;
		}
848 849 850 851 852
		clear_highpage(page);
		flush_dcache_page(page);
		SetPageUptodate(page);
	}

853 854 855
	swap = get_swap_page();
	if (!swap.val)
		goto redirty;
856

857 858
	/*
	 * Add inode to shmem_unuse()'s list of swapped-out inodes,
859 860
	 * if it's not already there.  Do it now before the page is
	 * moved to swap cache, when its pagelock no longer protects
861
	 * the inode from eviction.  But don't unlock the mutex until
862 863
	 * we've incremented swapped, because shmem_unuse_inode() will
	 * prune a !swapped inode from the swaplist under this mutex.
864
	 */
865 866 867
	mutex_lock(&shmem_swaplist_mutex);
	if (list_empty(&info->swaplist))
		list_add_tail(&info->swaplist, &shmem_swaplist);
868

869
	if (add_to_swap_cache(page, swap, GFP_ATOMIC) == 0) {
H
Hugh Dickins 已提交
870
		swap_shmem_alloc(swap);
871 872 873 874 875
		shmem_delete_from_page_cache(page, swp_to_radix_entry(swap));

		spin_lock(&info->lock);
		info->swapped++;
		shmem_recalc_inode(inode);
876
		spin_unlock(&info->lock);
877 878

		mutex_unlock(&shmem_swaplist_mutex);
879
		BUG_ON(page_mapped(page));
H
Hugh Dickins 已提交
880
		swap_writepage(page, wbc);
L
Linus Torvalds 已提交
881 882 883
		return 0;
	}

884
	mutex_unlock(&shmem_swaplist_mutex);
885
	swapcache_free(swap, NULL);
L
Linus Torvalds 已提交
886 887
redirty:
	set_page_dirty(page);
888 889 890 891
	if (wbc->for_reclaim)
		return AOP_WRITEPAGE_ACTIVATE;	/* Return with page locked */
	unlock_page(page);
	return 0;
L
Linus Torvalds 已提交
892 893 894
}

#ifdef CONFIG_NUMA
895
#ifdef CONFIG_TMPFS
896
static void shmem_show_mpol(struct seq_file *seq, struct mempolicy *mpol)
897
{
898
	char buffer[64];
899

900
	if (!mpol || mpol->mode == MPOL_DEFAULT)
901
		return;		/* show nothing */
902

903
	mpol_to_str(buffer, sizeof(buffer), mpol, 1);
904 905

	seq_printf(seq, ",mpol=%s", buffer);
906
}
907 908 909 910 911 912 913 914 915 916 917 918

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;
}
919 920
#endif /* CONFIG_TMPFS */

921 922
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
923
{
924
	struct mempolicy mpol, *spol;
L
Linus Torvalds 已提交
925 926
	struct vm_area_struct pvma;

927
	spol = mpol_cond_copy(&mpol,
928
			mpol_shared_policy_lookup(&info->policy, index));
929

L
Linus Torvalds 已提交
930
	/* Create a pseudo vma that just contains the policy */
931
	pvma.vm_start = 0;
932 933
	/* Bias interleave by inode number to distribute better across nodes */
	pvma.vm_pgoff = index + info->vfs_inode.i_ino;
934
	pvma.vm_ops = NULL;
935
	pvma.vm_policy = spol;
936
	return swapin_readahead(swap, gfp, &pvma, 0);
L
Linus Torvalds 已提交
937 938
}

939
static struct page *shmem_alloc_page(gfp_t gfp,
940
			struct shmem_inode_info *info, pgoff_t index)
L
Linus Torvalds 已提交
941 942 943
{
	struct vm_area_struct pvma;

944 945
	/* Create a pseudo vma that just contains the policy */
	pvma.vm_start = 0;
946 947
	/* Bias interleave by inode number to distribute better across nodes */
	pvma.vm_pgoff = index + info->vfs_inode.i_ino;
948
	pvma.vm_ops = NULL;
949
	pvma.vm_policy = mpol_shared_policy_lookup(&info->policy, index);
950 951 952 953 954

	/*
	 * alloc_page_vma() will drop the shared policy reference
	 */
	return alloc_page_vma(gfp, &pvma, 0);
L
Linus Torvalds 已提交
955
}
956 957
#else /* !CONFIG_NUMA */
#ifdef CONFIG_TMPFS
958
static inline void shmem_show_mpol(struct seq_file *seq, struct mempolicy *mpol)
959 960 961 962
{
}
#endif /* CONFIG_TMPFS */

963 964
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
965
{
966
	return swapin_readahead(swap, gfp, NULL, 0);
L
Linus Torvalds 已提交
967 968
}

969
static inline struct page *shmem_alloc_page(gfp_t gfp,
970
			struct shmem_inode_info *info, pgoff_t index)
L
Linus Torvalds 已提交
971
{
972
	return alloc_page(gfp);
L
Linus Torvalds 已提交
973
}
974
#endif /* CONFIG_NUMA */
L
Linus Torvalds 已提交
975

976 977 978 979 980 981 982
#if !defined(CONFIG_NUMA) || !defined(CONFIG_TMPFS)
static inline struct mempolicy *shmem_get_sbmpol(struct shmem_sb_info *sbinfo)
{
	return NULL;
}
#endif

983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022
/*
 * When a page is moved from swapcache to shmem filecache (either by the
 * usual swapin of shmem_getpage_gfp(), or by the less common swapoff of
 * shmem_unuse_inode()), it may have been read in earlier from swap, in
 * ignorance of the mapping it belongs to.  If that mapping has special
 * constraints (like the gma500 GEM driver, which requires RAM below 4GB),
 * we may need to copy to a suitable page before moving to filecache.
 *
 * In a future release, this may well be extended to respect cpuset and
 * NUMA mempolicy, and applied also to anonymous pages in do_swap_page();
 * but for now it is a simple matter of zone.
 */
static bool shmem_should_replace_page(struct page *page, gfp_t gfp)
{
	return page_zonenum(page) > gfp_zone(gfp);
}

static int shmem_replace_page(struct page **pagep, gfp_t gfp,
				struct shmem_inode_info *info, pgoff_t index)
{
	struct page *oldpage, *newpage;
	struct address_space *swap_mapping;
	pgoff_t swap_index;
	int error;

	oldpage = *pagep;
	swap_index = page_private(oldpage);
	swap_mapping = page_mapping(oldpage);

	/*
	 * We have arrived here because our zones are constrained, so don't
	 * limit chance of success by further cpuset and node constraints.
	 */
	gfp &= ~GFP_CONSTRAINT_MASK;
	newpage = shmem_alloc_page(gfp, info, index);
	if (!newpage)
		return -ENOMEM;

	page_cache_get(newpage);
	copy_highpage(newpage, oldpage);
1023
	flush_dcache_page(newpage);
1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037

	__set_page_locked(newpage);
	SetPageUptodate(newpage);
	SetPageSwapBacked(newpage);
	set_page_private(newpage, swap_index);
	SetPageSwapCache(newpage);

	/*
	 * Our caller will very soon move newpage out of swapcache, but it's
	 * a nice clean interface for us to replace oldpage by newpage there.
	 */
	spin_lock_irq(&swap_mapping->tree_lock);
	error = shmem_radix_tree_replace(swap_mapping, swap_index, oldpage,
								   newpage);
1038 1039 1040 1041
	if (!error) {
		__inc_zone_page_state(newpage, NR_FILE_PAGES);
		__dec_zone_page_state(oldpage, NR_FILE_PAGES);
	}
1042 1043
	spin_unlock_irq(&swap_mapping->tree_lock);

1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055
	if (unlikely(error)) {
		/*
		 * Is this possible?  I think not, now that our callers check
		 * both PageSwapCache and page_private after getting page lock;
		 * but be defensive.  Reverse old to newpage for clear and free.
		 */
		oldpage = newpage;
	} else {
		mem_cgroup_replace_page_cache(oldpage, newpage);
		lru_cache_add_anon(newpage);
		*pagep = newpage;
	}
1056 1057 1058 1059 1060 1061 1062

	ClearPageSwapCache(oldpage);
	set_page_private(oldpage, 0);

	unlock_page(oldpage);
	page_cache_release(oldpage);
	page_cache_release(oldpage);
1063
	return error;
1064 1065
}

L
Linus Torvalds 已提交
1066
/*
1067
 * shmem_getpage_gfp - find page in cache, or get from swap, or allocate
L
Linus Torvalds 已提交
1068 1069 1070 1071 1072
 *
 * 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
 */
1073
static int shmem_getpage_gfp(struct inode *inode, pgoff_t index,
1074
	struct page **pagep, enum sgp_type sgp, gfp_t gfp, int *fault_type)
L
Linus Torvalds 已提交
1075 1076
{
	struct address_space *mapping = inode->i_mapping;
1077
	struct shmem_inode_info *info;
L
Linus Torvalds 已提交
1078
	struct shmem_sb_info *sbinfo;
H
Hugh Dickins 已提交
1079
	struct page *page;
L
Linus Torvalds 已提交
1080 1081
	swp_entry_t swap;
	int error;
1082
	int once = 0;
1083
	int alloced = 0;
L
Linus Torvalds 已提交
1084

1085
	if (index > (MAX_LFS_FILESIZE >> PAGE_CACHE_SHIFT))
L
Linus Torvalds 已提交
1086 1087
		return -EFBIG;
repeat:
1088
	swap.val = 0;
1089
	page = find_lock_page(mapping, index);
1090 1091 1092 1093 1094
	if (radix_tree_exceptional_entry(page)) {
		swap = radix_to_swp_entry(page);
		page = NULL;
	}

1095
	if (sgp != SGP_WRITE && sgp != SGP_FALLOC &&
1096 1097 1098 1099 1100
	    ((loff_t)index << PAGE_CACHE_SHIFT) >= i_size_read(inode)) {
		error = -EINVAL;
		goto failed;
	}

1101 1102 1103 1104 1105 1106 1107 1108
	/* fallocated page? */
	if (page && !PageUptodate(page)) {
		if (sgp != SGP_READ)
			goto clear;
		unlock_page(page);
		page_cache_release(page);
		page = NULL;
	}
1109 1110 1111
	if (page || (sgp == SGP_READ && !swap.val)) {
		*pagep = page;
		return 0;
H
Hugh Dickins 已提交
1112 1113 1114
	}

	/*
1115 1116
	 * Fast cache lookup did not find it:
	 * bring it back from swap or allocate.
H
Hugh Dickins 已提交
1117
	 */
1118 1119
	info = SHMEM_I(inode);
	sbinfo = SHMEM_SB(inode->i_sb);
L
Linus Torvalds 已提交
1120 1121 1122

	if (swap.val) {
		/* Look it up and read it in.. */
H
Hugh Dickins 已提交
1123 1124
		page = lookup_swap_cache(swap);
		if (!page) {
L
Linus Torvalds 已提交
1125
			/* here we actually do the io */
1126 1127
			if (fault_type)
				*fault_type |= VM_FAULT_MAJOR;
1128
			page = shmem_swapin(swap, gfp, info, index);
H
Hugh Dickins 已提交
1129
			if (!page) {
1130 1131
				error = -ENOMEM;
				goto failed;
L
Linus Torvalds 已提交
1132 1133 1134 1135
			}
		}

		/* We have to do this with page locked to prevent races */
1136
		lock_page(page);
1137
		if (!PageSwapCache(page) || page_private(page) != swap.val ||
1138
		    !shmem_confirm_swap(mapping, index, swap)) {
1139
			error = -EEXIST;	/* try again */
1140
			goto unlock;
1141
		}
H
Hugh Dickins 已提交
1142
		if (!PageUptodate(page)) {
L
Linus Torvalds 已提交
1143
			error = -EIO;
1144
			goto failed;
L
Linus Torvalds 已提交
1145
		}
1146 1147
		wait_on_page_writeback(page);

1148 1149 1150 1151
		if (shmem_should_replace_page(page, gfp)) {
			error = shmem_replace_page(&page, gfp, info, index);
			if (error)
				goto failed;
L
Linus Torvalds 已提交
1152
		}
H
Hugh Dickins 已提交
1153

1154 1155
		error = mem_cgroup_cache_charge(page, current->mm,
						gfp & GFP_RECLAIM_MASK);
1156
		if (!error) {
1157 1158
			error = shmem_add_to_page_cache(page, mapping, index,
						gfp, swp_to_radix_entry(swap));
1159 1160 1161
			/* We already confirmed swap, and make no allocation */
			VM_BUG_ON(error);
		}
1162 1163 1164 1165
		if (error)
			goto failed;

		spin_lock(&info->lock);
1166
		info->swapped--;
1167
		shmem_recalc_inode(inode);
H
Hugh Dickins 已提交
1168
		spin_unlock(&info->lock);
1169 1170

		delete_from_swap_cache(page);
H
Hugh Dickins 已提交
1171 1172 1173
		set_page_dirty(page);
		swap_free(swap);

1174 1175 1176 1177
	} else {
		if (shmem_acct_block(info->flags)) {
			error = -ENOSPC;
			goto failed;
L
Linus Torvalds 已提交
1178
		}
1179
		if (sbinfo->max_blocks) {
1180
			if (percpu_counter_compare(&sbinfo->used_blocks,
1181 1182 1183 1184
						sbinfo->max_blocks) >= 0) {
				error = -ENOSPC;
				goto unacct;
			}
1185
			percpu_counter_inc(&sbinfo->used_blocks);
1186
		}
L
Linus Torvalds 已提交
1187

1188 1189 1190 1191
		page = shmem_alloc_page(gfp, info, index);
		if (!page) {
			error = -ENOMEM;
			goto decused;
L
Linus Torvalds 已提交
1192 1193
		}

1194 1195
		SetPageSwapBacked(page);
		__set_page_locked(page);
1196 1197
		error = mem_cgroup_cache_charge(page, current->mm,
						gfp & GFP_RECLAIM_MASK);
1198 1199
		if (error)
			goto decused;
1200 1201 1202 1203 1204 1205 1206 1207 1208 1209
		error = radix_tree_preload(gfp & GFP_RECLAIM_MASK);
		if (!error) {
			error = shmem_add_to_page_cache(page, mapping, index,
							gfp, NULL);
			radix_tree_preload_end();
		}
		if (error) {
			mem_cgroup_uncharge_cache_page(page);
			goto decused;
		}
1210 1211 1212
		lru_cache_add_anon(page);

		spin_lock(&info->lock);
L
Linus Torvalds 已提交
1213
		info->alloced++;
1214 1215
		inode->i_blocks += BLOCKS_PER_PAGE;
		shmem_recalc_inode(inode);
L
Linus Torvalds 已提交
1216
		spin_unlock(&info->lock);
1217
		alloced = true;
1218

1219
		/*
1220 1221 1222 1223 1224 1225 1226 1227 1228
		 * Let SGP_FALLOC use the SGP_WRITE optimization on a new page.
		 */
		if (sgp == SGP_FALLOC)
			sgp = SGP_WRITE;
clear:
		/*
		 * Let SGP_WRITE caller clear ends if write does not fill page;
		 * but SGP_FALLOC on a page fallocated earlier must initialize
		 * it now, lest undo on failure cancel our earlier guarantee.
1229 1230 1231 1232 1233 1234
		 */
		if (sgp != SGP_WRITE) {
			clear_highpage(page);
			flush_dcache_page(page);
			SetPageUptodate(page);
		}
1235
		if (sgp == SGP_DIRTY)
H
Hugh Dickins 已提交
1236
			set_page_dirty(page);
L
Linus Torvalds 已提交
1237
	}
1238

1239
	/* Perhaps the file has been truncated since we checked */
1240
	if (sgp != SGP_WRITE && sgp != SGP_FALLOC &&
1241 1242
	    ((loff_t)index << PAGE_CACHE_SHIFT) >= i_size_read(inode)) {
		error = -EINVAL;
1243 1244 1245 1246
		if (alloced)
			goto trunc;
		else
			goto failed;
H
Hugh Dickins 已提交
1247
	}
1248 1249
	*pagep = page;
	return 0;
L
Linus Torvalds 已提交
1250

1251
	/*
1252
	 * Error recovery.
1253
	 */
1254
trunc:
1255
	info = SHMEM_I(inode);
1256 1257 1258 1259 1260
	ClearPageDirty(page);
	delete_from_page_cache(page);
	spin_lock(&info->lock);
	info->alloced--;
	inode->i_blocks -= BLOCKS_PER_PAGE;
1261
	spin_unlock(&info->lock);
1262
decused:
1263
	sbinfo = SHMEM_SB(inode->i_sb);
1264 1265 1266 1267 1268
	if (sbinfo->max_blocks)
		percpu_counter_add(&sbinfo->used_blocks, -1);
unacct:
	shmem_unacct_blocks(info->flags, 1);
failed:
1269 1270 1271 1272
	if (swap.val && error != -EINVAL &&
	    !shmem_confirm_swap(mapping, index, swap))
		error = -EEXIST;
unlock:
H
Hugh Dickins 已提交
1273
	if (page) {
1274
		unlock_page(page);
H
Hugh Dickins 已提交
1275
		page_cache_release(page);
1276 1277 1278 1279 1280 1281
	}
	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
1282
		goto repeat;
1283
	}
1284
	if (error == -EEXIST)	/* from above or from radix_tree_insert */
1285 1286
		goto repeat;
	return error;
L
Linus Torvalds 已提交
1287 1288
}

N
Nick Piggin 已提交
1289
static int shmem_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
L
Linus Torvalds 已提交
1290
{
1291
	struct inode *inode = vma->vm_file->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
1292
	int error;
1293
	int ret = VM_FAULT_LOCKED;
L
Linus Torvalds 已提交
1294

1295
	error = shmem_getpage(inode, vmf->pgoff, &vmf->page, SGP_CACHE, &ret);
N
Nick Piggin 已提交
1296 1297
	if (error)
		return ((error == -ENOMEM) ? VM_FAULT_OOM : VM_FAULT_SIGBUS);
1298

1299 1300 1301 1302
	if (ret & VM_FAULT_MAJOR) {
		count_vm_event(PGMAJFAULT);
		mem_cgroup_count_vm_event(vma->vm_mm, PGMAJFAULT);
	}
1303
	return ret;
L
Linus Torvalds 已提交
1304 1305 1306
}

#ifdef CONFIG_NUMA
1307
static int shmem_set_policy(struct vm_area_struct *vma, struct mempolicy *mpol)
L
Linus Torvalds 已提交
1308
{
1309 1310
	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
1311 1312
}

A
Adrian Bunk 已提交
1313 1314
static struct mempolicy *shmem_get_policy(struct vm_area_struct *vma,
					  unsigned long addr)
L
Linus Torvalds 已提交
1315
{
1316 1317
	struct inode *inode = vma->vm_file->f_path.dentry->d_inode;
	pgoff_t index;
L
Linus Torvalds 已提交
1318

1319 1320
	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
1321 1322 1323 1324 1325
}
#endif

int shmem_lock(struct file *file, int lock, struct user_struct *user)
{
1326
	struct inode *inode = file->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
1327 1328 1329 1330 1331 1332 1333 1334
	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;
1335
		mapping_set_unevictable(file->f_mapping);
L
Linus Torvalds 已提交
1336 1337 1338 1339
	}
	if (!lock && (info->flags & VM_LOCKED) && user) {
		user_shm_unlock(inode->i_size, user);
		info->flags &= ~VM_LOCKED;
1340
		mapping_clear_unevictable(file->f_mapping);
L
Linus Torvalds 已提交
1341 1342
	}
	retval = 0;
1343

L
Linus Torvalds 已提交
1344 1345 1346 1347 1348
out_nomem:
	spin_unlock(&info->lock);
	return retval;
}

1349
static int shmem_mmap(struct file *file, struct vm_area_struct *vma)
L
Linus Torvalds 已提交
1350 1351 1352
{
	file_accessed(file);
	vma->vm_ops = &shmem_vm_ops;
N
Nick Piggin 已提交
1353
	vma->vm_flags |= VM_CAN_NONLINEAR;
L
Linus Torvalds 已提交
1354 1355 1356
	return 0;
}

1357
static struct inode *shmem_get_inode(struct super_block *sb, const struct inode *dir,
1358
				     umode_t mode, dev_t dev, unsigned long flags)
L
Linus Torvalds 已提交
1359 1360 1361 1362 1363
{
	struct inode *inode;
	struct shmem_inode_info *info;
	struct shmem_sb_info *sbinfo = SHMEM_SB(sb);

1364 1365
	if (shmem_reserve_inode(sb))
		return NULL;
L
Linus Torvalds 已提交
1366 1367 1368

	inode = new_inode(sb);
	if (inode) {
1369
		inode->i_ino = get_next_ino();
1370
		inode_init_owner(inode, dir, mode);
L
Linus Torvalds 已提交
1371 1372 1373
		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;
1374
		inode->i_generation = get_seconds();
L
Linus Torvalds 已提交
1375 1376 1377
		info = SHMEM_I(inode);
		memset(info, 0, (char *)inode - (char *)info);
		spin_lock_init(&info->lock);
1378
		info->flags = flags & VM_NORESERVE;
L
Linus Torvalds 已提交
1379
		INIT_LIST_HEAD(&info->swaplist);
1380
		INIT_LIST_HEAD(&info->xattr_list);
1381
		cache_no_acl(inode);
L
Linus Torvalds 已提交
1382 1383 1384

		switch (mode & S_IFMT) {
		default:
1385
			inode->i_op = &shmem_special_inode_operations;
L
Linus Torvalds 已提交
1386 1387 1388
			init_special_inode(inode, mode, dev);
			break;
		case S_IFREG:
1389
			inode->i_mapping->a_ops = &shmem_aops;
L
Linus Torvalds 已提交
1390 1391
			inode->i_op = &shmem_inode_operations;
			inode->i_fop = &shmem_file_operations;
1392 1393
			mpol_shared_policy_init(&info->policy,
						 shmem_get_sbmpol(sbinfo));
L
Linus Torvalds 已提交
1394 1395
			break;
		case S_IFDIR:
1396
			inc_nlink(inode);
L
Linus Torvalds 已提交
1397 1398 1399 1400 1401 1402 1403 1404 1405 1406
			/* 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.
			 */
1407
			mpol_shared_policy_init(&info->policy, NULL);
L
Linus Torvalds 已提交
1408 1409
			break;
		}
1410 1411
	} else
		shmem_free_inode(sb);
L
Linus Torvalds 已提交
1412 1413 1414 1415
	return inode;
}

#ifdef CONFIG_TMPFS
1416
static const struct inode_operations shmem_symlink_inode_operations;
1417
static const struct inode_operations shmem_short_symlink_operations;
L
Linus Torvalds 已提交
1418

1419 1420 1421 1422 1423 1424
#ifdef CONFIG_TMPFS_XATTR
static int shmem_initxattrs(struct inode *, const struct xattr *, void *);
#else
#define shmem_initxattrs NULL
#endif

L
Linus Torvalds 已提交
1425
static int
N
Nick Piggin 已提交
1426 1427 1428
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
1429
{
N
Nick Piggin 已提交
1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441
	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
1442 1443 1444
	if (pos + copied > inode->i_size)
		i_size_write(inode, pos + copied);

1445 1446 1447 1448 1449 1450 1451 1452
	if (!PageUptodate(page)) {
		if (copied < PAGE_CACHE_SIZE) {
			unsigned from = pos & (PAGE_CACHE_SIZE - 1);
			zero_user_segments(page, 0, from,
					from + copied, PAGE_CACHE_SIZE);
		}
		SetPageUptodate(page);
	}
N
Nick Piggin 已提交
1453
	set_page_dirty(page);
1454
	unlock_page(page);
N
Nick Piggin 已提交
1455 1456 1457
	page_cache_release(page);

	return copied;
L
Linus Torvalds 已提交
1458 1459 1460 1461
}

static void do_shmem_file_read(struct file *filp, loff_t *ppos, read_descriptor_t *desc, read_actor_t actor)
{
1462
	struct inode *inode = filp->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
1463
	struct address_space *mapping = inode->i_mapping;
1464 1465
	pgoff_t index;
	unsigned long offset;
1466 1467 1468 1469 1470 1471 1472 1473 1474
	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
1475 1476 1477 1478 1479 1480

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

	for (;;) {
		struct page *page = NULL;
1481 1482
		pgoff_t end_index;
		unsigned long nr, ret;
L
Linus Torvalds 已提交
1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493
		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;
		}

1494
		desc->error = shmem_getpage(inode, index, &page, sgp, NULL);
L
Linus Torvalds 已提交
1495 1496 1497 1498 1499
		if (desc->error) {
			if (desc->error == -EINVAL)
				desc->error = 0;
			break;
		}
H
Hugh Dickins 已提交
1500 1501
		if (page)
			unlock_page(page);
L
Linus Torvalds 已提交
1502 1503 1504

		/*
		 * We must evaluate after, since reads (unlike writes)
1505
		 * are called without i_mutex protection against truncate
L
Linus Torvalds 已提交
1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532
		 */
		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
1533
		} else {
L
Linus Torvalds 已提交
1534
			page = ZERO_PAGE(0);
N
Nick Piggin 已提交
1535 1536
			page_cache_get(page);
		}
L
Linus Torvalds 已提交
1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563

		/*
		 * 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
1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595
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
1596 1597
}

1598 1599 1600 1601 1602
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;
1603
	struct inode *inode = mapping->host;
1604 1605 1606 1607 1608 1609 1610 1611 1612 1613
	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,
1614
		.nr_pages_max = PIPE_DEF_BUFFERS,
1615 1616 1617 1618 1619
		.flags = flags,
		.ops = &page_cache_pipe_buf_ops,
		.spd_release = spd_release_page,
	};

1620
	isize = i_size_read(inode);
1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640
	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;

1641 1642 1643 1644 1645
	while (spd.nr_pages < nr_pages) {
		error = shmem_getpage(inode, index, &page, SGP_CACHE, NULL);
		if (error)
			break;
		unlock_page(page);
1646 1647 1648 1649 1650 1651 1652
		spd.pages[spd.nr_pages++] = page;
		index++;
	}

	index = *ppos >> PAGE_CACHE_SHIFT;
	nr_pages = spd.nr_pages;
	spd.nr_pages = 0;
1653

1654 1655 1656 1657 1658 1659 1660 1661 1662
	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];

1663 1664 1665 1666
		if (!PageUptodate(page) || page->mapping != mapping) {
			error = shmem_getpage(inode, index, &page,
							SGP_CACHE, NULL);
			if (error)
1667
				break;
1668 1669 1670
			unlock_page(page);
			page_cache_release(spd.pages[page_nr]);
			spd.pages[page_nr] = page;
1671
		}
1672 1673

		isize = i_size_read(inode);
1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702
		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);

1703
	splice_shrink_spd(&spd);
1704 1705 1706 1707 1708 1709 1710 1711

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

1712 1713 1714 1715
static long shmem_fallocate(struct file *file, int mode, loff_t offset,
							 loff_t len)
{
	struct inode *inode = file->f_path.dentry->d_inode;
1716
	struct shmem_sb_info *sbinfo = SHMEM_SB(inode->i_sb);
1717
	struct shmem_falloc shmem_falloc;
1718 1719
	pgoff_t start, index, end;
	int error;
1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733

	mutex_lock(&inode->i_mutex);

	if (mode & FALLOC_FL_PUNCH_HOLE) {
		struct address_space *mapping = file->f_mapping;
		loff_t unmap_start = round_up(offset, PAGE_SIZE);
		loff_t unmap_end = round_down(offset + len, PAGE_SIZE) - 1;

		if ((u64)unmap_end > (u64)unmap_start)
			unmap_mapping_range(mapping, unmap_start,
					    1 + unmap_end - unmap_start, 0);
		shmem_truncate_range(inode, offset, offset + len - 1);
		/* No need to unmap again: hole-punching leaves COWed pages */
		error = 0;
1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747
		goto out;
	}

	/* We need to check rlimit even when FALLOC_FL_KEEP_SIZE */
	error = inode_newsize_ok(inode, offset + len);
	if (error)
		goto out;

	start = offset >> PAGE_CACHE_SHIFT;
	end = (offset + len + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
	/* Try to avoid a swapstorm if len is impossible to satisfy */
	if (sbinfo->max_blocks && end - start > sbinfo->max_blocks) {
		error = -ENOSPC;
		goto out;
1748 1749
	}

1750 1751 1752 1753 1754 1755 1756 1757
	shmem_falloc.start = start;
	shmem_falloc.next  = start;
	shmem_falloc.nr_falloced = 0;
	shmem_falloc.nr_unswapped = 0;
	spin_lock(&inode->i_lock);
	inode->i_private = &shmem_falloc;
	spin_unlock(&inode->i_lock);

1758 1759 1760 1761 1762 1763 1764 1765 1766
	for (index = start; index < end; index++) {
		struct page *page;

		/*
		 * Good, the fallocate(2) manpage permits EINTR: we may have
		 * been interrupted because we are using up too much memory.
		 */
		if (signal_pending(current))
			error = -EINTR;
1767 1768
		else if (shmem_falloc.nr_unswapped > shmem_falloc.nr_falloced)
			error = -ENOMEM;
1769
		else
1770
			error = shmem_getpage(inode, index, &page, SGP_FALLOC,
1771 1772
									NULL);
		if (error) {
1773 1774 1775 1776
			/* Remove the !PageUptodate pages we added */
			shmem_undo_range(inode,
				(loff_t)start << PAGE_CACHE_SHIFT,
				(loff_t)index << PAGE_CACHE_SHIFT, true);
1777
			goto undone;
1778 1779
		}

1780 1781 1782 1783 1784 1785 1786 1787
		/*
		 * Inform shmem_writepage() how far we have reached.
		 * No need for lock or barrier: we have the page lock.
		 */
		shmem_falloc.next++;
		if (!PageUptodate(page))
			shmem_falloc.nr_falloced++;

1788
		/*
1789 1790 1791
		 * If !PageUptodate, leave it that way so that freeable pages
		 * can be recognized if we need to rollback on error later.
		 * But set_page_dirty so that memory pressure will swap rather
1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803
		 * than free the pages we are allocating (and SGP_CACHE pages
		 * might still be clean: we now need to mark those dirty too).
		 */
		set_page_dirty(page);
		unlock_page(page);
		page_cache_release(page);
		cond_resched();
	}

	if (!(mode & FALLOC_FL_KEEP_SIZE) && offset + len > inode->i_size)
		i_size_write(inode, offset + len);
	inode->i_ctime = CURRENT_TIME;
1804 1805 1806 1807
undone:
	spin_lock(&inode->i_lock);
	inode->i_private = NULL;
	spin_unlock(&inode->i_lock);
1808
out:
1809 1810 1811 1812
	mutex_unlock(&inode->i_mutex);
	return error;
}

1813
static int shmem_statfs(struct dentry *dentry, struct kstatfs *buf)
L
Linus Torvalds 已提交
1814
{
1815
	struct shmem_sb_info *sbinfo = SHMEM_SB(dentry->d_sb);
L
Linus Torvalds 已提交
1816 1817 1818 1819

	buf->f_type = TMPFS_MAGIC;
	buf->f_bsize = PAGE_CACHE_SIZE;
	buf->f_namelen = NAME_MAX;
1820
	if (sbinfo->max_blocks) {
L
Linus Torvalds 已提交
1821
		buf->f_blocks = sbinfo->max_blocks;
1822 1823 1824
		buf->f_bavail =
		buf->f_bfree  = sbinfo->max_blocks -
				percpu_counter_sum(&sbinfo->used_blocks);
1825 1826
	}
	if (sbinfo->max_inodes) {
L
Linus Torvalds 已提交
1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837
		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
1838
shmem_mknod(struct inode *dir, struct dentry *dentry, umode_t mode, dev_t dev)
L
Linus Torvalds 已提交
1839
{
1840
	struct inode *inode;
L
Linus Torvalds 已提交
1841 1842
	int error = -ENOSPC;

1843
	inode = shmem_get_inode(dir->i_sb, dir, mode, dev, VM_NORESERVE);
L
Linus Torvalds 已提交
1844
	if (inode) {
1845
		error = security_inode_init_security(inode, dir,
1846
						     &dentry->d_name,
1847
						     shmem_initxattrs, NULL);
1848 1849 1850 1851 1852
		if (error) {
			if (error != -EOPNOTSUPP) {
				iput(inode);
				return error;
			}
1853
		}
1854 1855
#ifdef CONFIG_TMPFS_POSIX_ACL
		error = generic_acl_init(inode, dir);
1856 1857 1858
		if (error) {
			iput(inode);
			return error;
1859
		}
A
Al Viro 已提交
1860 1861
#else
		error = 0;
1862
#endif
L
Linus Torvalds 已提交
1863 1864 1865 1866 1867 1868 1869 1870
		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;
}

1871
static int shmem_mkdir(struct inode *dir, struct dentry *dentry, umode_t mode)
L
Linus Torvalds 已提交
1872 1873 1874 1875 1876
{
	int error;

	if ((error = shmem_mknod(dir, dentry, mode | S_IFDIR, 0)))
		return error;
1877
	inc_nlink(dir);
L
Linus Torvalds 已提交
1878 1879 1880
	return 0;
}

A
Al Viro 已提交
1881
static int shmem_create(struct inode *dir, struct dentry *dentry, umode_t mode,
A
Al Viro 已提交
1882
		bool excl)
L
Linus Torvalds 已提交
1883 1884 1885 1886 1887 1888 1889 1890 1891 1892
{
	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;
1893
	int ret;
L
Linus Torvalds 已提交
1894 1895 1896 1897 1898 1899

	/*
	 * 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.
	 */
1900 1901 1902
	ret = shmem_reserve_inode(inode->i_sb);
	if (ret)
		goto out;
L
Linus Torvalds 已提交
1903 1904 1905

	dir->i_size += BOGO_DIRENT_SIZE;
	inode->i_ctime = dir->i_ctime = dir->i_mtime = CURRENT_TIME;
1906
	inc_nlink(inode);
A
Al Viro 已提交
1907
	ihold(inode);	/* New dentry reference */
L
Linus Torvalds 已提交
1908 1909
	dget(dentry);		/* Extra pinning count for the created dentry */
	d_instantiate(dentry, inode);
1910 1911
out:
	return ret;
L
Linus Torvalds 已提交
1912 1913 1914 1915 1916 1917
}

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

1918 1919
	if (inode->i_nlink > 1 && !S_ISDIR(inode->i_mode))
		shmem_free_inode(inode->i_sb);
L
Linus Torvalds 已提交
1920 1921 1922

	dir->i_size -= BOGO_DIRENT_SIZE;
	inode->i_ctime = dir->i_ctime = dir->i_mtime = CURRENT_TIME;
1923
	drop_nlink(inode);
L
Linus Torvalds 已提交
1924 1925 1926 1927 1928 1929 1930 1931 1932
	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;

1933 1934
	drop_nlink(dentry->d_inode);
	drop_nlink(dir);
L
Linus Torvalds 已提交
1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954
	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)
1955
			drop_nlink(old_dir);
L
Linus Torvalds 已提交
1956
	} else if (they_are_dirs) {
1957
		drop_nlink(old_dir);
1958
		inc_nlink(new_dir);
L
Linus Torvalds 已提交
1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973
	}

	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
1974
	struct page *page;
L
Linus Torvalds 已提交
1975 1976 1977 1978 1979 1980 1981
	char *kaddr;
	struct shmem_inode_info *info;

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

1982
	inode = shmem_get_inode(dir->i_sb, dir, S_IFLNK|S_IRWXUGO, 0, VM_NORESERVE);
L
Linus Torvalds 已提交
1983 1984 1985
	if (!inode)
		return -ENOSPC;

1986
	error = security_inode_init_security(inode, dir, &dentry->d_name,
1987
					     shmem_initxattrs, NULL);
1988 1989 1990 1991 1992 1993 1994 1995
	if (error) {
		if (error != -EOPNOTSUPP) {
			iput(inode);
			return error;
		}
		error = 0;
	}

L
Linus Torvalds 已提交
1996 1997
	info = SHMEM_I(inode);
	inode->i_size = len-1;
1998 1999 2000 2001 2002 2003 2004
	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
2005 2006 2007 2008 2009 2010
	} else {
		error = shmem_getpage(inode, 0, &page, SGP_WRITE, NULL);
		if (error) {
			iput(inode);
			return error;
		}
2011
		inode->i_mapping->a_ops = &shmem_aops;
L
Linus Torvalds 已提交
2012
		inode->i_op = &shmem_symlink_inode_operations;
2013
		kaddr = kmap_atomic(page);
L
Linus Torvalds 已提交
2014
		memcpy(kaddr, symname, len);
2015
		kunmap_atomic(kaddr);
2016
		SetPageUptodate(page);
L
Linus Torvalds 已提交
2017
		set_page_dirty(page);
2018
		unlock_page(page);
L
Linus Torvalds 已提交
2019 2020 2021 2022 2023 2024 2025 2026 2027
		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;
}

2028
static void *shmem_follow_short_symlink(struct dentry *dentry, struct nameidata *nd)
L
Linus Torvalds 已提交
2029
{
2030
	nd_set_link(nd, SHMEM_I(dentry->d_inode)->symlink);
2031
	return NULL;
L
Linus Torvalds 已提交
2032 2033
}

2034
static void *shmem_follow_link(struct dentry *dentry, struct nameidata *nd)
L
Linus Torvalds 已提交
2035 2036
{
	struct page *page = NULL;
2037 2038
	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
2039 2040
	if (page)
		unlock_page(page);
2041
	return page;
L
Linus Torvalds 已提交
2042 2043
}

2044
static void shmem_put_link(struct dentry *dentry, struct nameidata *nd, void *cookie)
L
Linus Torvalds 已提交
2045 2046
{
	if (!IS_ERR(nd_get_link(nd))) {
2047
		struct page *page = cookie;
L
Linus Torvalds 已提交
2048 2049 2050 2051 2052 2053
		kunmap(page);
		mark_page_accessed(page);
		page_cache_release(page);
	}
}

2054
#ifdef CONFIG_TMPFS_XATTR
2055
/*
2056 2057
 * Superblocks without xattr inode operations may get some security.* xattr
 * support from the LSM "for free". As soon as we have any other xattrs
2058 2059 2060 2061
 * like ACLs, we also need to implement the security.* handlers at
 * filesystem level, though.
 */

2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121
/*
 * Allocate new xattr and copy in the value; but leave the name to callers.
 */
static struct shmem_xattr *shmem_xattr_alloc(const void *value, size_t size)
{
	struct shmem_xattr *new_xattr;
	size_t len;

	/* wrap around? */
	len = sizeof(*new_xattr) + size;
	if (len <= sizeof(*new_xattr))
		return NULL;

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

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

/*
 * Callback for security_inode_init_security() for acquiring xattrs.
 */
static int shmem_initxattrs(struct inode *inode,
			    const struct xattr *xattr_array,
			    void *fs_info)
{
	struct shmem_inode_info *info = SHMEM_I(inode);
	const struct xattr *xattr;
	struct shmem_xattr *new_xattr;
	size_t len;

	for (xattr = xattr_array; xattr->name != NULL; xattr++) {
		new_xattr = shmem_xattr_alloc(xattr->value, xattr->value_len);
		if (!new_xattr)
			return -ENOMEM;

		len = strlen(xattr->name) + 1;
		new_xattr->name = kmalloc(XATTR_SECURITY_PREFIX_LEN + len,
					  GFP_KERNEL);
		if (!new_xattr->name) {
			kfree(new_xattr);
			return -ENOMEM;
		}

		memcpy(new_xattr->name, XATTR_SECURITY_PREFIX,
		       XATTR_SECURITY_PREFIX_LEN);
		memcpy(new_xattr->name + XATTR_SECURITY_PREFIX_LEN,
		       xattr->name, len);

		spin_lock(&info->lock);
		list_add(&new_xattr->list, &info->xattr_list);
		spin_unlock(&info->lock);
	}

	return 0;
}

2122 2123
static int shmem_xattr_get(struct dentry *dentry, const char *name,
			   void *buffer, size_t size)
2124
{
2125 2126 2127
	struct shmem_inode_info *info;
	struct shmem_xattr *xattr;
	int ret = -ENODATA;
2128

2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146
	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;
2147 2148
}

2149
static int shmem_xattr_set(struct inode *inode, const char *name,
2150
			   const void *value, size_t size, int flags)
2151
{
2152 2153 2154 2155 2156 2157 2158
	struct shmem_inode_info *info = SHMEM_I(inode);
	struct shmem_xattr *xattr;
	struct shmem_xattr *new_xattr = NULL;
	int err = 0;

	/* value == NULL means remove */
	if (value) {
2159
		new_xattr = shmem_xattr_alloc(value, size);
2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196
		if (!new_xattr)
			return -ENOMEM;

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

	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;
2197 2198
}

2199
static const struct xattr_handler *shmem_xattr_handlers[] = {
2200
#ifdef CONFIG_TMPFS_POSIX_ACL
2201 2202
	&generic_acl_access_handler,
	&generic_acl_default_handler,
2203
#endif
2204 2205
	NULL
};
2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265

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 */

2266
	return shmem_xattr_set(dentry->d_inode, name, value, size, flags);
2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285

}

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;

2286
	return shmem_xattr_set(dentry->d_inode, name, NULL, 0, XATTR_REPLACE);
2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327
}

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 */

2328
static const struct inode_operations shmem_short_symlink_operations = {
2329
	.readlink	= generic_readlink,
2330
	.follow_link	= shmem_follow_short_symlink,
2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347
#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,
2348
#endif
2349
};
2350

2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363
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
2364 2365
static struct dentry *shmem_fh_to_dentry(struct super_block *sb,
		struct fid *fid, int fh_len, int fh_type)
2366 2367
{
	struct inode *inode;
C
Christoph Hellwig 已提交
2368 2369 2370 2371 2372 2373
	struct dentry *dentry = NULL;
	u64 inum = fid->raw[2];
	inum = (inum << 32) | fid->raw[1];

	if (fh_len < 3)
		return NULL;
2374

C
Christoph Hellwig 已提交
2375 2376
	inode = ilookup5(sb, (unsigned long)(inum + fid->raw[0]),
			shmem_match, fid->raw);
2377
	if (inode) {
C
Christoph Hellwig 已提交
2378
		dentry = d_find_alias(inode);
2379 2380 2381
		iput(inode);
	}

C
Christoph Hellwig 已提交
2382
	return dentry;
2383 2384
}

A
Al Viro 已提交
2385 2386
static int shmem_encode_fh(struct inode *inode, __u32 *fh, int *len,
				struct inode *parent)
2387
{
2388 2389
	if (*len < 3) {
		*len = 3;
2390
		return 255;
2391
	}
2392

A
Al Viro 已提交
2393
	if (inode_unhashed(inode)) {
2394 2395 2396 2397 2398 2399 2400
		/* 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
2401
		if (inode_unhashed(inode))
2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414
			__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;
}

2415
static const struct export_operations shmem_export_ops = {
2416 2417
	.get_parent     = shmem_get_parent,
	.encode_fh      = shmem_encode_fh,
C
Christoph Hellwig 已提交
2418
	.fh_to_dentry	= shmem_fh_to_dentry,
2419 2420
};

2421 2422
static int shmem_parse_options(char *options, struct shmem_sb_info *sbinfo,
			       bool remount)
L
Linus Torvalds 已提交
2423 2424
{
	char *this_char, *value, *rest;
2425 2426
	uid_t uid;
	gid_t gid;
L
Linus Torvalds 已提交
2427

2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444
	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
2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466
		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;
2467 2468
			sbinfo->max_blocks =
				DIV_ROUND_UP(size, PAGE_CACHE_SIZE);
L
Linus Torvalds 已提交
2469
		} else if (!strcmp(this_char,"nr_blocks")) {
2470
			sbinfo->max_blocks = memparse(value, &rest);
L
Linus Torvalds 已提交
2471 2472 2473
			if (*rest)
				goto bad_val;
		} else if (!strcmp(this_char,"nr_inodes")) {
2474
			sbinfo->max_inodes = memparse(value, &rest);
L
Linus Torvalds 已提交
2475 2476 2477
			if (*rest)
				goto bad_val;
		} else if (!strcmp(this_char,"mode")) {
2478
			if (remount)
L
Linus Torvalds 已提交
2479
				continue;
2480
			sbinfo->mode = simple_strtoul(value, &rest, 8) & 07777;
L
Linus Torvalds 已提交
2481 2482 2483
			if (*rest)
				goto bad_val;
		} else if (!strcmp(this_char,"uid")) {
2484
			if (remount)
L
Linus Torvalds 已提交
2485
				continue;
2486
			uid = simple_strtoul(value, &rest, 0);
L
Linus Torvalds 已提交
2487 2488
			if (*rest)
				goto bad_val;
2489 2490 2491
			sbinfo->uid = make_kuid(current_user_ns(), uid);
			if (!uid_valid(sbinfo->uid))
				goto bad_val;
L
Linus Torvalds 已提交
2492
		} else if (!strcmp(this_char,"gid")) {
2493
			if (remount)
L
Linus Torvalds 已提交
2494
				continue;
2495
			gid = simple_strtoul(value, &rest, 0);
L
Linus Torvalds 已提交
2496 2497
			if (*rest)
				goto bad_val;
2498 2499 2500
			sbinfo->gid = make_kgid(current_user_ns(), gid);
			if (!gid_valid(sbinfo->gid))
				goto bad_val;
2501
		} else if (!strcmp(this_char,"mpol")) {
2502
			if (mpol_parse_str(value, &sbinfo->mpol, 1))
2503
				goto bad_val;
L
Linus Torvalds 已提交
2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521
		} 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);
2522
	struct shmem_sb_info config = *sbinfo;
2523 2524 2525
	unsigned long inodes;
	int error = -EINVAL;

2526
	if (shmem_parse_options(data, &config, true))
2527
		return error;
L
Linus Torvalds 已提交
2528

2529 2530
	spin_lock(&sbinfo->stat_lock);
	inodes = sbinfo->max_inodes - sbinfo->free_inodes;
2531
	if (percpu_counter_compare(&sbinfo->used_blocks, config.max_blocks) > 0)
2532
		goto out;
2533
	if (config.max_inodes < inodes)
2534 2535
		goto out;
	/*
2536
	 * Those tests disallow limited->unlimited while any are in use;
2537 2538 2539
	 * but we must separately disallow unlimited->limited, because
	 * in that case we have no record of how much is already in use.
	 */
2540
	if (config.max_blocks && !sbinfo->max_blocks)
2541
		goto out;
2542
	if (config.max_inodes && !sbinfo->max_inodes)
2543 2544 2545
		goto out;

	error = 0;
2546 2547 2548
	sbinfo->max_blocks  = config.max_blocks;
	sbinfo->max_inodes  = config.max_inodes;
	sbinfo->free_inodes = config.max_inodes - inodes;
2549 2550 2551

	mpol_put(sbinfo->mpol);
	sbinfo->mpol        = config.mpol;	/* transfers initial ref */
2552 2553 2554
out:
	spin_unlock(&sbinfo->stat_lock);
	return error;
L
Linus Torvalds 已提交
2555
}
2556

2557
static int shmem_show_options(struct seq_file *seq, struct dentry *root)
2558
{
2559
	struct shmem_sb_info *sbinfo = SHMEM_SB(root->d_sb);
2560 2561 2562 2563 2564 2565 2566

	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))
2567
		seq_printf(seq, ",mode=%03ho", sbinfo->mode);
2568 2569 2570 2571 2572 2573
	if (!uid_eq(sbinfo->uid, GLOBAL_ROOT_UID))
		seq_printf(seq, ",uid=%u",
				from_kuid_munged(&init_user_ns, sbinfo->uid));
	if (!gid_eq(sbinfo->gid, GLOBAL_ROOT_GID))
		seq_printf(seq, ",gid=%u",
				from_kgid_munged(&init_user_ns, sbinfo->gid));
2574
	shmem_show_mpol(seq, sbinfo->mpol);
2575 2576 2577
	return 0;
}
#endif /* CONFIG_TMPFS */
L
Linus Torvalds 已提交
2578 2579 2580

static void shmem_put_super(struct super_block *sb)
{
2581 2582 2583 2584
	struct shmem_sb_info *sbinfo = SHMEM_SB(sb);

	percpu_counter_destroy(&sbinfo->used_blocks);
	kfree(sbinfo);
L
Linus Torvalds 已提交
2585 2586 2587
	sb->s_fs_info = NULL;
}

2588
int shmem_fill_super(struct super_block *sb, void *data, int silent)
L
Linus Torvalds 已提交
2589 2590
{
	struct inode *inode;
2591
	struct shmem_sb_info *sbinfo;
2592 2593 2594
	int err = -ENOMEM;

	/* Round up to L1_CACHE_BYTES to resist false sharing */
2595
	sbinfo = kzalloc(max((int)sizeof(struct shmem_sb_info),
2596 2597 2598 2599 2600
				L1_CACHE_BYTES), GFP_KERNEL);
	if (!sbinfo)
		return -ENOMEM;

	sbinfo->mode = S_IRWXUGO | S_ISVTX;
2601 2602
	sbinfo->uid = current_fsuid();
	sbinfo->gid = current_fsgid();
2603
	sb->s_fs_info = sbinfo;
L
Linus Torvalds 已提交
2604

2605
#ifdef CONFIG_TMPFS
L
Linus Torvalds 已提交
2606 2607 2608 2609 2610 2611
	/*
	 * 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)) {
2612 2613 2614 2615 2616 2617
		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
2618
	}
2619
	sb->s_export_op = &shmem_export_ops;
H
Hugh Dickins 已提交
2620
	sb->s_flags |= MS_NOSEC;
L
Linus Torvalds 已提交
2621 2622 2623 2624
#else
	sb->s_flags |= MS_NOUSER;
#endif

2625
	spin_lock_init(&sbinfo->stat_lock);
2626 2627
	if (percpu_counter_init(&sbinfo->used_blocks, 0))
		goto failed;
2628
	sbinfo->free_inodes = sbinfo->max_inodes;
2629

2630
	sb->s_maxbytes = MAX_LFS_FILESIZE;
L
Linus Torvalds 已提交
2631 2632 2633 2634
	sb->s_blocksize = PAGE_CACHE_SIZE;
	sb->s_blocksize_bits = PAGE_CACHE_SHIFT;
	sb->s_magic = TMPFS_MAGIC;
	sb->s_op = &shmem_ops;
2635
	sb->s_time_gran = 1;
2636
#ifdef CONFIG_TMPFS_XATTR
2637
	sb->s_xattr = shmem_xattr_handlers;
2638 2639
#endif
#ifdef CONFIG_TMPFS_POSIX_ACL
2640 2641
	sb->s_flags |= MS_POSIXACL;
#endif
2642

2643
	inode = shmem_get_inode(sb, NULL, S_IFDIR | sbinfo->mode, 0, VM_NORESERVE);
L
Linus Torvalds 已提交
2644 2645
	if (!inode)
		goto failed;
2646 2647
	inode->i_uid = sbinfo->uid;
	inode->i_gid = sbinfo->gid;
A
Al Viro 已提交
2648 2649
	sb->s_root = d_make_root(inode);
	if (!sb->s_root)
2650
		goto failed;
L
Linus Torvalds 已提交
2651 2652 2653 2654 2655 2656 2657
	return 0;

failed:
	shmem_put_super(sb);
	return err;
}

2658
static struct kmem_cache *shmem_inode_cachep;
L
Linus Torvalds 已提交
2659 2660 2661

static struct inode *shmem_alloc_inode(struct super_block *sb)
{
2662 2663 2664
	struct shmem_inode_info *info;
	info = kmem_cache_alloc(shmem_inode_cachep, GFP_KERNEL);
	if (!info)
L
Linus Torvalds 已提交
2665
		return NULL;
2666
	return &info->vfs_inode;
L
Linus Torvalds 已提交
2667 2668
}

2669
static void shmem_destroy_callback(struct rcu_head *head)
N
Nick Piggin 已提交
2670 2671 2672 2673 2674
{
	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
2675 2676
static void shmem_destroy_inode(struct inode *inode)
{
2677
	if (S_ISREG(inode->i_mode))
L
Linus Torvalds 已提交
2678
		mpol_free_shared_policy(&SHMEM_I(inode)->policy);
2679
	call_rcu(&inode->i_rcu, shmem_destroy_callback);
L
Linus Torvalds 已提交
2680 2681
}

2682
static void shmem_init_inode(void *foo)
L
Linus Torvalds 已提交
2683
{
2684 2685
	struct shmem_inode_info *info = foo;
	inode_init_once(&info->vfs_inode);
L
Linus Torvalds 已提交
2686 2687
}

2688
static int shmem_init_inodecache(void)
L
Linus Torvalds 已提交
2689 2690 2691
{
	shmem_inode_cachep = kmem_cache_create("shmem_inode_cache",
				sizeof(struct shmem_inode_info),
2692
				0, SLAB_PANIC, shmem_init_inode);
L
Linus Torvalds 已提交
2693 2694 2695
	return 0;
}

2696
static void shmem_destroy_inodecache(void)
L
Linus Torvalds 已提交
2697
{
2698
	kmem_cache_destroy(shmem_inode_cachep);
L
Linus Torvalds 已提交
2699 2700
}

2701
static const struct address_space_operations shmem_aops = {
L
Linus Torvalds 已提交
2702
	.writepage	= shmem_writepage,
2703
	.set_page_dirty	= __set_page_dirty_no_writeback,
L
Linus Torvalds 已提交
2704
#ifdef CONFIG_TMPFS
N
Nick Piggin 已提交
2705 2706
	.write_begin	= shmem_write_begin,
	.write_end	= shmem_write_end,
L
Linus Torvalds 已提交
2707
#endif
2708
	.migratepage	= migrate_page,
2709
	.error_remove_page = generic_error_remove_page,
L
Linus Torvalds 已提交
2710 2711
};

2712
static const struct file_operations shmem_file_operations = {
L
Linus Torvalds 已提交
2713 2714
	.mmap		= shmem_mmap,
#ifdef CONFIG_TMPFS
2715
	.llseek		= generic_file_llseek,
H
Hugh Dickins 已提交
2716
	.read		= do_sync_read,
H
Hugh Dickins 已提交
2717
	.write		= do_sync_write,
H
Hugh Dickins 已提交
2718
	.aio_read	= shmem_file_aio_read,
H
Hugh Dickins 已提交
2719
	.aio_write	= generic_file_aio_write,
2720
	.fsync		= noop_fsync,
2721
	.splice_read	= shmem_file_splice_read,
2722
	.splice_write	= generic_file_splice_write,
2723
	.fallocate	= shmem_fallocate,
L
Linus Torvalds 已提交
2724 2725 2726
#endif
};

2727
static const struct inode_operations shmem_inode_operations = {
2728
	.setattr	= shmem_setattr,
2729 2730 2731 2732 2733 2734
#ifdef CONFIG_TMPFS_XATTR
	.setxattr	= shmem_setxattr,
	.getxattr	= shmem_getxattr,
	.listxattr	= shmem_listxattr,
	.removexattr	= shmem_removexattr,
#endif
L
Linus Torvalds 已提交
2735 2736
};

2737
static const struct inode_operations shmem_dir_inode_operations = {
L
Linus Torvalds 已提交
2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748
#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
2749 2750 2751 2752 2753 2754
#ifdef CONFIG_TMPFS_XATTR
	.setxattr	= shmem_setxattr,
	.getxattr	= shmem_getxattr,
	.listxattr	= shmem_listxattr,
	.removexattr	= shmem_removexattr,
#endif
2755
#ifdef CONFIG_TMPFS_POSIX_ACL
2756
	.setattr	= shmem_setattr,
2757 2758 2759
#endif
};

2760
static const struct inode_operations shmem_special_inode_operations = {
2761 2762 2763 2764 2765 2766
#ifdef CONFIG_TMPFS_XATTR
	.setxattr	= shmem_setxattr,
	.getxattr	= shmem_getxattr,
	.listxattr	= shmem_listxattr,
	.removexattr	= shmem_removexattr,
#endif
2767
#ifdef CONFIG_TMPFS_POSIX_ACL
2768
	.setattr	= shmem_setattr,
2769
#endif
L
Linus Torvalds 已提交
2770 2771
};

2772
static const struct super_operations shmem_ops = {
L
Linus Torvalds 已提交
2773 2774 2775 2776 2777
	.alloc_inode	= shmem_alloc_inode,
	.destroy_inode	= shmem_destroy_inode,
#ifdef CONFIG_TMPFS
	.statfs		= shmem_statfs,
	.remount_fs	= shmem_remount_fs,
2778
	.show_options	= shmem_show_options,
L
Linus Torvalds 已提交
2779
#endif
A
Al Viro 已提交
2780
	.evict_inode	= shmem_evict_inode,
L
Linus Torvalds 已提交
2781 2782 2783 2784
	.drop_inode	= generic_delete_inode,
	.put_super	= shmem_put_super,
};

2785
static const struct vm_operations_struct shmem_vm_ops = {
2786
	.fault		= shmem_fault,
L
Linus Torvalds 已提交
2787 2788 2789 2790 2791 2792
#ifdef CONFIG_NUMA
	.set_policy     = shmem_set_policy,
	.get_policy     = shmem_get_policy,
#endif
};

A
Al Viro 已提交
2793 2794
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
2795
{
A
Al Viro 已提交
2796
	return mount_nodev(fs_type, flags, data, shmem_fill_super);
L
Linus Torvalds 已提交
2797 2798
}

2799
static struct file_system_type shmem_fs_type = {
L
Linus Torvalds 已提交
2800 2801
	.owner		= THIS_MODULE,
	.name		= "tmpfs",
A
Al Viro 已提交
2802
	.mount		= shmem_mount,
L
Linus Torvalds 已提交
2803 2804 2805
	.kill_sb	= kill_litter_super,
};

2806
int __init shmem_init(void)
L
Linus Torvalds 已提交
2807 2808 2809
{
	int error;

P
Peter Zijlstra 已提交
2810 2811 2812 2813
	error = bdi_init(&shmem_backing_dev_info);
	if (error)
		goto out4;

2814
	error = shmem_init_inodecache();
L
Linus Torvalds 已提交
2815 2816 2817
	if (error)
		goto out3;

2818
	error = register_filesystem(&shmem_fs_type);
L
Linus Torvalds 已提交
2819 2820 2821 2822
	if (error) {
		printk(KERN_ERR "Could not register tmpfs\n");
		goto out2;
	}
2823

2824 2825
	shm_mnt = vfs_kern_mount(&shmem_fs_type, MS_NOUSER,
				 shmem_fs_type.name, NULL);
L
Linus Torvalds 已提交
2826 2827 2828 2829 2830 2831 2832 2833
	if (IS_ERR(shm_mnt)) {
		error = PTR_ERR(shm_mnt);
		printk(KERN_ERR "Could not kern_mount tmpfs\n");
		goto out1;
	}
	return 0;

out1:
2834
	unregister_filesystem(&shmem_fs_type);
L
Linus Torvalds 已提交
2835
out2:
2836
	shmem_destroy_inodecache();
L
Linus Torvalds 已提交
2837
out3:
P
Peter Zijlstra 已提交
2838 2839
	bdi_destroy(&shmem_backing_dev_info);
out4:
L
Linus Torvalds 已提交
2840 2841 2842
	shm_mnt = ERR_PTR(error);
	return error;
}
2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856

#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>

2857
static struct file_system_type shmem_fs_type = {
2858
	.name		= "tmpfs",
A
Al Viro 已提交
2859
	.mount		= ramfs_mount,
2860 2861 2862
	.kill_sb	= kill_litter_super,
};

2863
int __init shmem_init(void)
2864
{
2865
	BUG_ON(register_filesystem(&shmem_fs_type) != 0);
2866

2867
	shm_mnt = kern_mount(&shmem_fs_type);
2868 2869 2870 2871 2872
	BUG_ON(IS_ERR(shm_mnt));

	return 0;
}

2873
int shmem_unuse(swp_entry_t swap, struct page *page)
2874 2875 2876 2877
{
	return 0;
}

H
Hugh Dickins 已提交
2878 2879 2880 2881 2882
int shmem_lock(struct file *file, int lock, struct user_struct *user)
{
	return 0;
}

2883 2884 2885 2886
void shmem_unlock_mapping(struct address_space *mapping)
{
}

2887
void shmem_truncate_range(struct inode *inode, loff_t lstart, loff_t lend)
2888
{
2889
	truncate_inode_pages_range(inode->i_mapping, lstart, lend);
2890 2891 2892
}
EXPORT_SYMBOL_GPL(shmem_truncate_range);

2893 2894
#define shmem_vm_ops				generic_file_vm_ops
#define shmem_file_operations			ramfs_file_operations
2895
#define shmem_get_inode(sb, dir, mode, dev, flags)	ramfs_get_inode(sb, dir, mode, dev)
2896 2897
#define shmem_acct_size(flags, size)		0
#define shmem_unacct_size(flags, size)		do {} while (0)
2898 2899 2900 2901

#endif /* CONFIG_SHMEM */

/* common code */
L
Linus Torvalds 已提交
2902

2903
/**
L
Linus Torvalds 已提交
2904 2905 2906
 * 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
2907
 * @flags: VM_NORESERVE suppresses pre-accounting of the entire object size
L
Linus Torvalds 已提交
2908
 */
2909
struct file *shmem_file_setup(const char *name, loff_t size, unsigned long flags)
L
Linus Torvalds 已提交
2910 2911 2912 2913
{
	int error;
	struct file *file;
	struct inode *inode;
2914 2915
	struct path path;
	struct dentry *root;
L
Linus Torvalds 已提交
2916 2917 2918 2919 2920
	struct qstr this;

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

2921
	if (size < 0 || size > MAX_LFS_FILESIZE)
L
Linus Torvalds 已提交
2922 2923 2924 2925 2926 2927 2928 2929 2930 2931
		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;
2932 2933
	path.dentry = d_alloc(root, &this);
	if (!path.dentry)
L
Linus Torvalds 已提交
2934
		goto put_memory;
2935
	path.mnt = mntget(shm_mnt);
L
Linus Torvalds 已提交
2936 2937

	error = -ENOSPC;
2938
	inode = shmem_get_inode(root->d_sb, NULL, S_IFREG | S_IRWXUGO, 0, flags);
L
Linus Torvalds 已提交
2939
	if (!inode)
2940
		goto put_dentry;
L
Linus Torvalds 已提交
2941

2942
	d_instantiate(path.dentry, inode);
L
Linus Torvalds 已提交
2943
	inode->i_size = size;
2944
	clear_nlink(inode);	/* It is unlinked */
2945 2946 2947
#ifndef CONFIG_MMU
	error = ramfs_nommu_expand_for_mapping(inode, size);
	if (error)
2948
		goto put_dentry;
2949
#endif
2950 2951

	error = -ENFILE;
2952
	file = alloc_file(&path, FMODE_WRITE | FMODE_READ,
2953 2954 2955 2956
		  &shmem_file_operations);
	if (!file)
		goto put_dentry;

L
Linus Torvalds 已提交
2957 2958 2959
	return file;

put_dentry:
2960
	path_put(&path);
L
Linus Torvalds 已提交
2961 2962 2963 2964
put_memory:
	shmem_unacct_size(flags, size);
	return ERR_PTR(error);
}
2965
EXPORT_SYMBOL_GPL(shmem_file_setup);
L
Linus Torvalds 已提交
2966

2967
/**
L
Linus Torvalds 已提交
2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983
 * 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;
2984
	vma->vm_flags |= VM_CAN_NONLINEAR;
L
Linus Torvalds 已提交
2985 2986
	return 0;
}
2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999

/**
 * 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.
 *
3000 3001
 * i915_gem_object_get_pages_gtt() mixes __GFP_NORETRY | __GFP_NOWARN in
 * with the mapping_gfp_mask(), to avoid OOMing the machine unnecessarily.
3002 3003 3004 3005
 */
struct page *shmem_read_mapping_page_gfp(struct address_space *mapping,
					 pgoff_t index, gfp_t gfp)
{
3006 3007
#ifdef CONFIG_SHMEM
	struct inode *inode = mapping->host;
H
Hugh Dickins 已提交
3008
	struct page *page;
3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021
	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
	 */
3022
	return read_cache_page_gfp(mapping, index, gfp);
3023
#endif
3024 3025
}
EXPORT_SYMBOL_GPL(shmem_read_mapping_page_gfp);