shmem.c 66.2 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1 2 3 4 5 6 7 8
/*
 * Resizable virtual memory filesystem for Linux.
 *
 * Copyright (C) 2000 Linus Torvalds.
 *		 2000 Transmeta Corp.
 *		 2000-2001 Christoph Rohland
 *		 2000-2001 SAP AG
 *		 2002 Red Hat Inc.
9 10
 * Copyright (C) 2002-2005 Hugh Dickins.
 * Copyright (C) 2002-2005 VERITAS Software Corporation.
Linus Torvalds's avatar
Linus Torvalds committed
11 12 13 14 15 16
 * Copyright (C) 2004 Andi Kleen, SuSE Labs
 *
 * Extended attribute support for tmpfs:
 * Copyright (c) 2004, Luke Kenneth Casson Leighton <lkcl@lkcl.net>
 * Copyright (c) 2004 Red Hat, Inc., James Morris <jmorris@redhat.com>
 *
17 18 19
 * tiny-shmem:
 * Copyright (c) 2004, 2008 Matt Mackall <mpm@selenic.com>
 *
Linus Torvalds's avatar
Linus Torvalds committed
20 21 22
 * This file is released under the GPL.
 */

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

static struct vfsmount *shm_mnt;

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193
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);
	}
}

194
/**
195
 * shmem_recalc_inode - recalculate the block usage of an inode
Linus Torvalds's avatar
Linus Torvalds committed
196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212
 * @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) {
213 214 215
		struct shmem_sb_info *sbinfo = SHMEM_SB(inode->i_sb);
		if (sbinfo->max_blocks)
			percpu_counter_add(&sbinfo->used_blocks, -freed);
Linus Torvalds's avatar
Linus Torvalds committed
216
		info->alloced -= freed;
217
		inode->i_blocks -= freed * BLOCKS_PER_PAGE;
Linus Torvalds's avatar
Linus Torvalds committed
218 219 220 221
		shmem_unacct_blocks(info->flags, freed);
	}
}

222 223
static void shmem_put_swap(struct shmem_inode_info *info, pgoff_t index,
			   swp_entry_t swap)
Linus Torvalds's avatar
Linus Torvalds committed
224
{
225 226
	if (index < SHMEM_NR_DIRECT)
		info->i_direct[index] = swap;
Linus Torvalds's avatar
Linus Torvalds committed
227 228
}

229
static swp_entry_t shmem_get_swap(struct shmem_inode_info *info, pgoff_t index)
Linus Torvalds's avatar
Linus Torvalds committed
230
{
231 232
	return (index < SHMEM_NR_DIRECT) ?
		info->i_direct[index] : (swp_entry_t){0};
Linus Torvalds's avatar
Linus Torvalds committed
233 234
}

235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257
/*
 * 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;
}

258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306
/*
 * Like add_to_page_cache_locked, but error if expected item has gone.
 */
static int shmem_add_to_page_cache(struct page *page,
				   struct address_space *mapping,
				   pgoff_t index, gfp_t gfp, void *expected)
{
	int error;

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

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

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

307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353
/*
 * Like find_get_pages, but collecting swap entries as well as pages.
 */
static unsigned shmem_find_get_pages_and_swap(struct address_space *mapping,
					pgoff_t start, unsigned int nr_pages,
					struct page **pages, pgoff_t *indices)
{
	unsigned int i;
	unsigned int ret;
	unsigned int nr_found;

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

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

354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389
/*
 * Lockless lookup of swap entry in radix tree, avoiding refcount on pages.
 */
static pgoff_t shmem_find_swap(struct address_space *mapping, void *radswap)
{
	void  **slots[PAGEVEC_SIZE];
	pgoff_t indices[PAGEVEC_SIZE];
	unsigned int nr_found;

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

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

390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424
/*
 * Remove swap entry from radix tree, free the swap and its page cache.
 */
static int shmem_free_swap(struct address_space *mapping,
			   pgoff_t index, void *radswap)
{
	int error;

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

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

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

/*
 * Remove range of pages and swap entries from radix tree, and free them.
 */
425
void shmem_truncate_range(struct inode *inode, loff_t lstart, loff_t lend)
Linus Torvalds's avatar
Linus Torvalds committed
426
{
427
	struct address_space *mapping = inode->i_mapping;
Linus Torvalds's avatar
Linus Torvalds committed
428
	struct shmem_inode_info *info = SHMEM_I(inode);
429
	pgoff_t start = (lstart + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
430
	unsigned partial = lstart & (PAGE_CACHE_SIZE - 1);
431
	pgoff_t end = (lend >> PAGE_CACHE_SHIFT);
432
	struct pagevec pvec;
433 434
	pgoff_t indices[PAGEVEC_SIZE];
	long nr_swaps_freed = 0;
435
	pgoff_t index;
436 437 438 439 440 441
	int i;

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

	pagevec_init(&pvec, 0);
	index = start;
442 443 444 445 446 447
	while (index <= end) {
		pvec.nr = shmem_find_get_pages_and_swap(mapping, index,
			min(end - index, (pgoff_t)PAGEVEC_SIZE - 1) + 1,
							pvec.pages, indices);
		if (!pvec.nr)
			break;
448 449 450 451
		mem_cgroup_uncharge_start();
		for (i = 0; i < pagevec_count(&pvec); i++) {
			struct page *page = pvec.pages[i];

452
			index = indices[i];
453 454 455
			if (index > end)
				break;

456 457 458
			if (radix_tree_exceptional_entry(page)) {
				nr_swaps_freed += !shmem_free_swap(mapping,
								index, page);
459
				continue;
460 461 462
			}

			if (!trylock_page(page))
463
				continue;
464 465 466
			if (page->mapping == mapping) {
				VM_BUG_ON(PageWriteback(page));
				truncate_inode_page(mapping, page);
467 468 469
			}
			unlock_page(page);
		}
470
		shmem_pagevec_release(&pvec);
471 472 473 474
		mem_cgroup_uncharge_end();
		cond_resched();
		index++;
	}
Linus Torvalds's avatar
Linus Torvalds committed
475

476 477 478 479 480 481 482 483 484 485 486 487 488 489
	if (partial) {
		struct page *page = NULL;
		shmem_getpage(inode, start - 1, &page, SGP_READ, NULL);
		if (page) {
			zero_user_segment(page, partial, PAGE_CACHE_SIZE);
			set_page_dirty(page);
			unlock_page(page);
			page_cache_release(page);
		}
	}

	index = start;
	for ( ; ; ) {
		cond_resched();
490 491 492 493
		pvec.nr = shmem_find_get_pages_and_swap(mapping, index,
			min(end - index, (pgoff_t)PAGEVEC_SIZE - 1) + 1,
							pvec.pages, indices);
		if (!pvec.nr) {
494 495 496 497 498
			if (index == start)
				break;
			index = start;
			continue;
		}
499 500
		if (index == start && indices[0] > end) {
			shmem_pagevec_release(&pvec);
501 502 503 504 505 506
			break;
		}
		mem_cgroup_uncharge_start();
		for (i = 0; i < pagevec_count(&pvec); i++) {
			struct page *page = pvec.pages[i];

507
			index = indices[i];
508 509 510
			if (index > end)
				break;

511 512 513 514 515 516
			if (radix_tree_exceptional_entry(page)) {
				nr_swaps_freed += !shmem_free_swap(mapping,
								index, page);
				continue;
			}

517
			lock_page(page);
518 519 520 521
			if (page->mapping == mapping) {
				VM_BUG_ON(PageWriteback(page));
				truncate_inode_page(mapping, page);
			}
522 523
			unlock_page(page);
		}
524
		shmem_pagevec_release(&pvec);
525 526 527
		mem_cgroup_uncharge_end();
		index++;
	}
528

Linus Torvalds's avatar
Linus Torvalds committed
529
	spin_lock(&info->lock);
530
	info->swapped -= nr_swaps_freed;
Linus Torvalds's avatar
Linus Torvalds committed
531 532 533
	shmem_recalc_inode(inode);
	spin_unlock(&info->lock);

534
	inode->i_ctime = inode->i_mtime = CURRENT_TIME;
Linus Torvalds's avatar
Linus Torvalds committed
535
}
536
EXPORT_SYMBOL_GPL(shmem_truncate_range);
Linus Torvalds's avatar
Linus Torvalds committed
537

538
static int shmem_setattr(struct dentry *dentry, struct iattr *attr)
Linus Torvalds's avatar
Linus Torvalds committed
539 540 541 542
{
	struct inode *inode = dentry->d_inode;
	int error;

543 544 545 546
	error = inode_change_ok(inode, attr);
	if (error)
		return error;

547 548 549
	if (S_ISREG(inode->i_mode) && (attr->ia_valid & ATTR_SIZE)) {
		loff_t oldsize = inode->i_size;
		loff_t newsize = attr->ia_size;
550

551 552 553 554 555 556 557 558 559 560 561
		if (newsize != oldsize) {
			i_size_write(inode, newsize);
			inode->i_ctime = inode->i_mtime = CURRENT_TIME;
		}
		if (newsize < oldsize) {
			loff_t holebegin = round_up(newsize, PAGE_SIZE);
			unmap_mapping_range(inode->i_mapping, holebegin, 0, 1);
			shmem_truncate_range(inode, newsize, (loff_t)-1);
			/* unmap again to remove racily COWed private pages */
			unmap_mapping_range(inode->i_mapping, holebegin, 0, 1);
		}
Linus Torvalds's avatar
Linus Torvalds committed
562 563
	}

564
	setattr_copy(inode, attr);
565
#ifdef CONFIG_TMPFS_POSIX_ACL
566
	if (attr->ia_valid & ATTR_MODE)
567
		error = generic_acl_chmod(inode);
568
#endif
Linus Torvalds's avatar
Linus Torvalds committed
569 570 571
	return error;
}

Al Viro's avatar
Al Viro committed
572
static void shmem_evict_inode(struct inode *inode)
Linus Torvalds's avatar
Linus Torvalds committed
573 574
{
	struct shmem_inode_info *info = SHMEM_I(inode);
575
	struct shmem_xattr *xattr, *nxattr;
Linus Torvalds's avatar
Linus Torvalds committed
576

577
	if (inode->i_mapping->a_ops == &shmem_aops) {
Linus Torvalds's avatar
Linus Torvalds committed
578 579
		shmem_unacct_size(info->flags, inode->i_size);
		inode->i_size = 0;
580
		shmem_truncate_range(inode, 0, (loff_t)-1);
Linus Torvalds's avatar
Linus Torvalds committed
581
		if (!list_empty(&info->swaplist)) {
582
			mutex_lock(&shmem_swaplist_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
583
			list_del_init(&info->swaplist);
584
			mutex_unlock(&shmem_swaplist_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
585 586
		}
	}
587 588 589 590 591

	list_for_each_entry_safe(xattr, nxattr, &info->xattr_list, list) {
		kfree(xattr->name);
		kfree(xattr);
	}
592
	BUG_ON(inode->i_blocks);
593
	shmem_free_inode(inode->i_sb);
Al Viro's avatar
Al Viro committed
594
	end_writeback(inode);
Linus Torvalds's avatar
Linus Torvalds committed
595 596
}

597 598 599
/*
 * If swap found in inode, free it and move page from swapcache to filecache.
 */
600 601
static int shmem_unuse_inode(struct shmem_inode_info *info,
			     swp_entry_t swap, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
602
{
603
	struct address_space *mapping = info->vfs_inode.i_mapping;
604
	void *radswap;
605
	pgoff_t index;
606
	int error;
Linus Torvalds's avatar
Linus Torvalds committed
607

608 609 610
	radswap = swp_to_radix_entry(swap);
	index = shmem_find_swap(mapping, radswap);
	if (index == -1)
611
		return 0;
612

Hugh Dickins's avatar
Hugh Dickins committed
613 614
	/*
	 * Move _head_ to start search for next from here.
Al Viro's avatar
Al Viro committed
615
	 * But be careful: shmem_evict_inode checks list_empty without taking
Hugh Dickins's avatar
Hugh Dickins committed
616
	 * mutex, and there's an instant in list_move_tail when info->swaplist
617
	 * would appear empty, if it were the only one on shmem_swaplist.
Hugh Dickins's avatar
Hugh Dickins committed
618 619 620
	 */
	if (shmem_swaplist.next != &info->swaplist)
		list_move_tail(&shmem_swaplist, &info->swaplist);
621

KAMEZAWA Hiroyuki's avatar
KAMEZAWA Hiroyuki committed
622
	/*
623 624 625
	 * We rely on shmem_swaplist_mutex, not only to protect the swaplist,
	 * but also to hold up shmem_evict_inode(): so inode cannot be freed
	 * beneath us (pagelock doesn't help until the page is in pagecache).
KAMEZAWA Hiroyuki's avatar
KAMEZAWA Hiroyuki committed
626
	 */
627 628
	error = shmem_add_to_page_cache(page, mapping, index,
						GFP_NOWAIT, radswap);
629
	/* which does mem_cgroup_uncharge_cache_page on error */
630

631
	if (error != -ENOMEM) {
632 633 634 635
		/*
		 * Truncation and eviction use free_swap_and_cache(), which
		 * only does trylock page: if we raced, best clean up here.
		 */
636 637
		delete_from_swap_cache(page);
		set_page_dirty(page);
638 639 640 641 642 643
		if (!error) {
			spin_lock(&info->lock);
			info->swapped--;
			spin_unlock(&info->lock);
			swap_free(swap);
		}
644
		error = 1;	/* not an error, but entry was found */
Linus Torvalds's avatar
Linus Torvalds committed
645
	}
646
	return error;
Linus Torvalds's avatar
Linus Torvalds committed
647 648 649
}

/*
650
 * Search through swapped inodes to find and replace swap by page.
Linus Torvalds's avatar
Linus Torvalds committed
651
 */
652
int shmem_unuse(swp_entry_t swap, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
653
{
654
	struct list_head *this, *next;
Linus Torvalds's avatar
Linus Torvalds committed
655 656
	struct shmem_inode_info *info;
	int found = 0;
657 658 659 660 661 662
	int error;

	/*
	 * Charge page using GFP_KERNEL while we can wait, before taking
	 * the shmem_swaplist_mutex which might hold up shmem_writepage().
	 * Charged back to the user (not to caller) when swap account is used.
663
	 * shmem_add_to_page_cache() will be called with GFP_NOWAIT.
664 665 666 667
	 */
	error = mem_cgroup_cache_charge(page, current->mm, GFP_KERNEL);
	if (error)
		goto out;
668
	/* No radix_tree_preload: swap entry keeps a place for page in tree */
Linus Torvalds's avatar
Linus Torvalds committed
669

670
	mutex_lock(&shmem_swaplist_mutex);
671 672
	list_for_each_safe(this, next, &shmem_swaplist) {
		info = list_entry(this, struct shmem_inode_info, swaplist);
673 674 675 676 677 678 679
		if (!info->swapped) {
			spin_lock(&info->lock);
			if (!info->swapped)
				list_del_init(&info->swaplist);
			spin_unlock(&info->lock);
		}
		if (info->swapped)
680
			found = shmem_unuse_inode(info, swap, page);
681
		cond_resched();
682
		if (found)
683
			break;
Linus Torvalds's avatar
Linus Torvalds committed
684
	}
685
	mutex_unlock(&shmem_swaplist_mutex);
686 687 688 689 690 691

	if (!found)
		mem_cgroup_uncharge_cache_page(page);
	if (found < 0)
		error = found;
out:
Hugh Dickins's avatar
Hugh Dickins committed
692 693
	unlock_page(page);
	page_cache_release(page);
694
	return error;
Linus Torvalds's avatar
Linus Torvalds committed
695 696 697 698 699 700 701 702
}

/*
 * 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;
703
	swp_entry_t swap, oswap;
Linus Torvalds's avatar
Linus Torvalds committed
704
	struct address_space *mapping;
705
	pgoff_t index;
Linus Torvalds's avatar
Linus Torvalds committed
706 707 708 709 710 711 712 713 714
	struct inode *inode;

	BUG_ON(!PageLocked(page));
	mapping = page->mapping;
	index = page->index;
	inode = mapping->host;
	info = SHMEM_I(inode);
	if (info->flags & VM_LOCKED)
		goto redirty;
715
	if (!total_swap_pages)
Linus Torvalds's avatar
Linus Torvalds committed
716 717
		goto redirty;

718 719 720
	/*
	 * shmem_backing_dev_info's capabilities prevent regular writeback or
	 * sync from ever calling shmem_writepage; but a stacking filesystem
721
	 * might use ->writepage of its underlying filesystem, in which case
722
	 * tmpfs should write out to swap only in response to memory pressure,
723
	 * and not for the writeback threads or sync.
724
	 */
725 726 727 728
	if (!wbc->for_reclaim) {
		WARN_ON_ONCE(1);	/* Still happens? Tell us about it! */
		goto redirty;
	}
729 730

	/*
731 732
	 * Disable even the toy swapping implementation, while we convert
	 * functions one by one to having swap entries in the radix tree.
733
	 */
734
	if (index < ULONG_MAX)
735 736
		goto redirty;

737 738 739
	swap = get_swap_page();
	if (!swap.val)
		goto redirty;
740

741 742 743 744 745 746 747 748 749
	/*
	 * Add inode to shmem_unuse()'s list of swapped-out inodes,
	 * if it's not already there.  Do it now because we cannot take
	 * mutex while holding spinlock, and must do so before the page
	 * is moved to swap cache, when its pagelock no longer protects
	 * the inode from eviction.  But don't unlock the mutex until
	 * we've taken the spinlock, because shmem_unuse_inode() will
	 * prune a !swapped inode from the swaplist under both locks.
	 */
750 751 752
	mutex_lock(&shmem_swaplist_mutex);
	if (list_empty(&info->swaplist))
		list_add_tail(&info->swaplist, &shmem_swaplist);
753

Linus Torvalds's avatar
Linus Torvalds committed
754
	spin_lock(&info->lock);
755
	mutex_unlock(&shmem_swaplist_mutex);
756

757 758
	oswap = shmem_get_swap(info, index);
	if (oswap.val) {
759
		WARN_ON_ONCE(1);	/* Still happens? Tell us about it! */
760 761 762
		free_swap_and_cache(oswap);
		shmem_put_swap(info, index, (swp_entry_t){0});
		info->swapped--;
763 764
	}
	shmem_recalc_inode(inode);
Linus Torvalds's avatar
Linus Torvalds committed
765

766
	if (add_to_swap_cache(page, swap, GFP_ATOMIC) == 0) {
767
		delete_from_page_cache(page);
768 769
		shmem_put_swap(info, index, swap);
		info->swapped++;
Hugh Dickins's avatar
Hugh Dickins committed
770
		swap_shmem_alloc(swap);
771
		spin_unlock(&info->lock);
772
		BUG_ON(page_mapped(page));
773
		swap_writepage(page, wbc);
Linus Torvalds's avatar
Linus Torvalds committed
774 775 776 777
		return 0;
	}

	spin_unlock(&info->lock);
778
	swapcache_free(swap, NULL);
Linus Torvalds's avatar
Linus Torvalds committed
779 780
redirty:
	set_page_dirty(page);
781 782 783 784
	if (wbc->for_reclaim)
		return AOP_WRITEPAGE_ACTIVATE;	/* Return with page locked */
	unlock_page(page);
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
785 786 787
}

#ifdef CONFIG_NUMA
788
#ifdef CONFIG_TMPFS
789
static void shmem_show_mpol(struct seq_file *seq, struct mempolicy *mpol)
790
{
791
	char buffer[64];
792

793
	if (!mpol || mpol->mode == MPOL_DEFAULT)
794
		return;		/* show nothing */
795

796
	mpol_to_str(buffer, sizeof(buffer), mpol, 1);
797 798

	seq_printf(seq, ",mpol=%s", buffer);
799
}
800 801 802 803 804 805 806 807 808 809 810 811

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;
}
812 813
#endif /* CONFIG_TMPFS */

814 815
static struct page *shmem_swapin(swp_entry_t swap, gfp_t gfp,
			struct shmem_inode_info *info, pgoff_t index)
Linus Torvalds's avatar
Linus Torvalds committed
816
{
817
	struct mempolicy mpol, *spol;
Linus Torvalds's avatar
Linus Torvalds committed
818 819
	struct vm_area_struct pvma;

820
	spol = mpol_cond_copy(&mpol,
821
			mpol_shared_policy_lookup(&info->policy, index));
822

Linus Torvalds's avatar
Linus Torvalds committed
823
	/* Create a pseudo vma that just contains the policy */
824
	pvma.vm_start = 0;
825
	pvma.vm_pgoff = index;
826
	pvma.vm_ops = NULL;
827
	pvma.vm_policy = spol;
828
	return swapin_readahead(swap, gfp, &pvma, 0);
Linus Torvalds's avatar
Linus Torvalds committed
829 830
}

831
static struct page *shmem_alloc_page(gfp_t gfp,
832
			struct shmem_inode_info *info, pgoff_t index)
Linus Torvalds's avatar
Linus Torvalds committed
833 834 835
{
	struct vm_area_struct pvma;

836 837
	/* Create a pseudo vma that just contains the policy */
	pvma.vm_start = 0;
838
	pvma.vm_pgoff = index;
839
	pvma.vm_ops = NULL;
840
	pvma.vm_policy = mpol_shared_policy_lookup(&info->policy, index);
841 842 843 844 845

	/*
	 * alloc_page_vma() will drop the shared policy reference
	 */
	return alloc_page_vma(gfp, &pvma, 0);
Linus Torvalds's avatar
Linus Torvalds committed
846
}
847 848
#else /* !CONFIG_NUMA */
#ifdef CONFIG_TMPFS
849
static inline void shmem_show_mpol(struct seq_file *seq, struct mempolicy *mpol)
850 851 852 853
{
}
#endif /* CONFIG_TMPFS */

854 855
static inline struct page *shmem_swapin(swp_entry_t swap, gfp_t gfp,
			struct shmem_inode_info *info, pgoff_t index)
Linus Torvalds's avatar
Linus Torvalds committed
856
{
857
	return swapin_readahead(swap, gfp, NULL, 0);
Linus Torvalds's avatar
Linus Torvalds committed
858 859
}

860
static inline struct page *shmem_alloc_page(gfp_t gfp,
861
			struct shmem_inode_info *info, pgoff_t index)
Linus Torvalds's avatar
Linus Torvalds committed
862
{
863
	return alloc_page(gfp);
Linus Torvalds's avatar
Linus Torvalds committed
864
}
865
#endif /* CONFIG_NUMA */
Linus Torvalds's avatar
Linus Torvalds committed
866

867 868 869 870 871 872 873
#if !defined(CONFIG_NUMA) || !defined(CONFIG_TMPFS)
static inline struct mempolicy *shmem_get_sbmpol(struct shmem_sb_info *sbinfo)
{
	return NULL;
}
#endif

Linus Torvalds's avatar
Linus Torvalds committed
874
/*
875
 * shmem_getpage_gfp - find page in cache, or get from swap, or allocate
Linus Torvalds's avatar
Linus Torvalds committed
876 877 878 879 880
 *
 * 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
 */
881
static int shmem_getpage_gfp(struct inode *inode, pgoff_t index,
882
	struct page **pagep, enum sgp_type sgp, gfp_t gfp, int *fault_type)
Linus Torvalds's avatar
Linus Torvalds committed
883 884
{
	struct address_space *mapping = inode->i_mapping;
885
	struct shmem_inode_info *info;
Linus Torvalds's avatar
Linus Torvalds committed
886
	struct shmem_sb_info *sbinfo;
887
	struct page *page;
Linus Torvalds's avatar
Linus Torvalds committed
888 889
	swp_entry_t swap;
	int error;
890
	int once = 0;
Linus Torvalds's avatar
Linus Torvalds committed
891

892
	if (index > (MAX_LFS_FILESIZE >> PAGE_CACHE_SHIFT))
Linus Torvalds's avatar
Linus Torvalds committed
893 894
		return -EFBIG;
repeat:
895
	swap.val = 0;
896
	page = find_lock_page(mapping, index);
897 898 899 900 901 902 903 904 905 906 907 908
	if (radix_tree_exceptional_entry(page)) {
		swap = radix_to_swp_entry(page);
		page = NULL;
	}

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

	if (page || (sgp == SGP_READ && !swap.val)) {
Hugh Dickins's avatar
Hugh Dickins committed
909
		/*
910 911 912
		 * Once we can get the page lock, it must be uptodate:
		 * if there were an error in reading back from swap,
		 * the page would not be inserted into the filecache.
Hugh Dickins's avatar
Hugh Dickins committed
913
		 */
914 915 916
		BUG_ON(page && !PageUptodate(page));
		*pagep = page;
		return 0;
917 918 919
	}

	/*
920 921
	 * Fast cache lookup did not find it:
	 * bring it back from swap or allocate.
922
	 */
923 924
	info = SHMEM_I(inode);
	sbinfo = SHMEM_SB(inode->i_sb);
Linus Torvalds's avatar
Linus Torvalds committed
925 926 927

	if (swap.val) {
		/* Look it up and read it in.. */
928 929
		page = lookup_swap_cache(swap);
		if (!page) {
Linus Torvalds's avatar
Linus Torvalds committed
930
			/* here we actually do the io */
931 932
			if (fault_type)
				*fault_type |= VM_FAULT_MAJOR;
933
			page = shmem_swapin(swap, gfp, info, index);
934
			if (!page) {
935 936
				error = -ENOMEM;
				goto failed;
Linus Torvalds's avatar
Linus Torvalds committed
937 938 939 940
			}
		}

		/* We have to do this with page locked to prevent races */
941
		lock_page(page);
942
		if (!PageUptodate(page)) {
Linus Torvalds's avatar
Linus Torvalds committed
943
			error = -EIO;
944
			goto failed;
Linus Torvalds's avatar
Linus Torvalds committed
945
		}
946 947 948 949 950 951 952 953 954
		wait_on_page_writeback(page);

		/* Someone may have already done it for us */
		if (page->mapping) {
			if (page->mapping == mapping &&
			    page->index == index)
				goto done;
			error = -EEXIST;
			goto failed;
Linus Torvalds's avatar
Linus Torvalds committed
955
		}
956

957 958 959 960 961 962
		error = shmem_add_to_page_cache(page, mapping, index,
					gfp, swp_to_radix_entry(swap));
		if (error)
			goto failed;

		spin_lock(&info->lock);
963
		info->swapped--;
964
		shmem_recalc_inode(inode);
965
		spin_unlock(&info->lock);
966 967

		delete_from_swap_cache(page);
968 969 970
		set_page_dirty(page);
		swap_free(swap);

971 972 973 974
	} else {
		if (shmem_acct_block(info->flags)) {
			error = -ENOSPC;
			goto failed;
Linus Torvalds's avatar
Linus Torvalds committed
975
		}
976
		if (sbinfo->max_blocks) {
977
			if (percpu_counter_compare(&sbinfo->used_blocks,
978 979 980 981
						sbinfo->max_blocks) >= 0) {
				error = -ENOSPC;
				goto unacct;
			}
982
			percpu_counter_inc(&sbinfo->used_blocks);
983
		}
Linus Torvalds's avatar
Linus Torvalds committed
984

985 986 987 988
		page = shmem_alloc_page(gfp, info, index);
		if (!page) {
			error = -ENOMEM;
			goto decused;
Linus Torvalds's avatar
Linus Torvalds committed
989 990
		}

991 992 993 994 995 996 997 998 999
		SetPageSwapBacked(page);
		__set_page_locked(page);
		error = shmem_add_to_page_cache(page, mapping, index,
								gfp, NULL);
		if (error)
			goto decused;
		lru_cache_add_anon(page);

		spin_lock(&info->lock);
Linus Torvalds's avatar
Linus Torvalds committed
1000
		info->alloced++;
1001 1002
		inode->i_blocks += BLOCKS_PER_PAGE;
		shmem_recalc_inode(inode);
Linus Torvalds's avatar
Linus Torvalds committed
1003
		spin_unlock(&info->lock);
1004

1005 1006 1007
		clear_highpage(page);
		flush_dcache_page(page);
		SetPageUptodate(page);
1008
		if (sgp == SGP_DIRTY)
1009
			set_page_dirty(page);
Linus Torvalds's avatar
Linus Torvalds committed
1010 1011
	}
done:
1012 1013 1014 1015 1016
	/* Perhaps the file has been truncated since we checked */
	if (sgp != SGP_WRITE &&
	    ((loff_t)index << PAGE_CACHE_SHIFT) >= i_size_read(inode)) {
		error = -EINVAL;
		goto trunc;
Hugh Dickins's avatar
Hugh Dickins committed
1017
	}
1018 1019
	*pagep = page;
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
1020

1021
	/*
1022
	 * Error recovery.
1023
	 */
1024 1025 1026 1027 1028 1029
trunc:
	ClearPageDirty(page);
	delete_from_page_cache(page);
	spin_lock(&info->lock);
	info->alloced--;
	inode->i_blocks -= BLOCKS_PER_PAGE;
1030
	spin_unlock(&info->lock);