dcache.c 83.6 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
/*
 * fs/dcache.c
 *
 * Complete reimplementation
 * (C) 1997 Thomas Schoebel-Theuer,
 * with heavy changes by Linus Torvalds
 */

/*
 * Notes on the allocation strategy:
 *
 * The dcache is a master of the icache - whenever a dcache entry
 * exists, the inode will always exist. "iput()" is done either when
 * the dcache entry is deleted or garbage collected.
 */

17
#include <linux/ratelimit.h>
L
Linus Torvalds 已提交
18 19 20
#include <linux/string.h>
#include <linux/mm.h>
#include <linux/fs.h>
21
#include <linux/fsnotify.h>
L
Linus Torvalds 已提交
22 23 24 25
#include <linux/slab.h>
#include <linux/init.h>
#include <linux/hash.h>
#include <linux/cache.h>
26
#include <linux/export.h>
L
Linus Torvalds 已提交
27 28 29
#include <linux/security.h>
#include <linux/seqlock.h>
#include <linux/bootmem.h>
30 31
#include <linux/bit_spinlock.h>
#include <linux/rculist_bl.h>
32
#include <linux/list_lru.h>
33
#include "internal.h"
34
#include "mount.h"
L
Linus Torvalds 已提交
35

N
Nick Piggin 已提交
36 37
/*
 * Usage:
38
 * dcache->d_inode->i_lock protects:
39
 *   - i_dentry, d_u.d_alias, d_inode of aliases
40 41
 * dcache_hash_bucket lock protects:
 *   - the dcache hash table
42 43
 * s_roots bl list spinlock protects:
 *   - the s_roots list (see __d_drop)
D
Dave Chinner 已提交
44
 * dentry->d_sb->s_dentry_lru_lock protects:
N
Nick Piggin 已提交
45 46 47 48 49
 *   - the dcache lru lists and counters
 * d_lock protects:
 *   - d_flags
 *   - d_name
 *   - d_lru
N
Nick Piggin 已提交
50
 *   - d_count
N
Nick Piggin 已提交
51
 *   - d_unhashed()
N
Nick Piggin 已提交
52 53
 *   - d_parent and d_subdirs
 *   - childrens' d_child and d_parent
54
 *   - d_u.d_alias, d_inode
N
Nick Piggin 已提交
55 56
 *
 * Ordering:
57
 * dentry->d_inode->i_lock
N
Nick Piggin 已提交
58
 *   dentry->d_lock
D
Dave Chinner 已提交
59
 *     dentry->d_sb->s_dentry_lru_lock
60
 *     dcache_hash_bucket lock
61
 *     s_roots lock
N
Nick Piggin 已提交
62
 *
N
Nick Piggin 已提交
63 64 65 66 67 68 69
 * If there is an ancestor relationship:
 * dentry->d_parent->...->d_parent->d_lock
 *   ...
 *     dentry->d_parent->d_lock
 *       dentry->d_lock
 *
 * If no ancestor relationship:
70
 * arbitrary, since it's serialized on rename_lock
N
Nick Piggin 已提交
71
 */
72
int sysctl_vfs_cache_pressure __read_mostly = 100;
L
Linus Torvalds 已提交
73 74
EXPORT_SYMBOL_GPL(sysctl_vfs_cache_pressure);

A
Al Viro 已提交
75
__cacheline_aligned_in_smp DEFINE_SEQLOCK(rename_lock);
L
Linus Torvalds 已提交
76

77
EXPORT_SYMBOL(rename_lock);
L
Linus Torvalds 已提交
78

79
static struct kmem_cache *dentry_cache __read_mostly;
L
Linus Torvalds 已提交
80

D
David Howells 已提交
81 82 83 84 85
const struct qstr empty_name = QSTR_INIT("", 0);
EXPORT_SYMBOL(empty_name);
const struct qstr slash_name = QSTR_INIT("/", 1);
EXPORT_SYMBOL(slash_name);

L
Linus Torvalds 已提交
86 87 88 89 90 91 92 93 94
/*
 * This is the single most critical data structure when it comes
 * to the dcache: the hashtable for lookups. Somebody should try
 * to make this good - I've just made it work.
 *
 * This hash-function tries to avoid losing too many bits of hash
 * information, yet avoid using a prime hash-size or similar.
 */

95
static unsigned int d_hash_shift __read_mostly;
96

97
static struct hlist_bl_head *dentry_hashtable __read_mostly;
98

99
static inline struct hlist_bl_head *d_hash(unsigned int hash)
100
{
101
	return dentry_hashtable + (hash >> d_hash_shift);
102 103
}

A
Al Viro 已提交
104 105 106 107 108 109 110 111 112 113 114
#define IN_LOOKUP_SHIFT 10
static struct hlist_bl_head in_lookup_hashtable[1 << IN_LOOKUP_SHIFT];

static inline struct hlist_bl_head *in_lookup_hash(const struct dentry *parent,
					unsigned int hash)
{
	hash += (unsigned long) parent / L1_CACHE_BYTES;
	return in_lookup_hashtable + hash_32(hash, IN_LOOKUP_SHIFT);
}


L
Linus Torvalds 已提交
115 116 117 118 119
/* Statistics gathering. */
struct dentry_stat_t dentry_stat = {
	.age_limit = 45,
};

120
static DEFINE_PER_CPU(long, nr_dentry);
121
static DEFINE_PER_CPU(long, nr_dentry_unused);
122 123

#if defined(CONFIG_SYSCTL) && defined(CONFIG_PROC_FS)
124 125 126 127 128 129 130 131 132 133 134 135 136

/*
 * Here we resort to our own counters instead of using generic per-cpu counters
 * for consistency with what the vfs inode code does. We are expected to harvest
 * better code and performance by having our own specialized counters.
 *
 * Please note that the loop is done over all possible CPUs, not over all online
 * CPUs. The reason for this is that we don't want to play games with CPUs going
 * on and off. If one of them goes off, we will just keep their counters.
 *
 * glommer: See cffbc8a for details, and if you ever intend to change this,
 * please update all vfs counters to match.
 */
137
static long get_nr_dentry(void)
138 139
{
	int i;
140
	long sum = 0;
141 142 143 144 145
	for_each_possible_cpu(i)
		sum += per_cpu(nr_dentry, i);
	return sum < 0 ? 0 : sum;
}

146 147 148 149 150 151 152 153 154
static long get_nr_dentry_unused(void)
{
	int i;
	long sum = 0;
	for_each_possible_cpu(i)
		sum += per_cpu(nr_dentry_unused, i);
	return sum < 0 ? 0 : sum;
}

155
int proc_nr_dentry(struct ctl_table *table, int write, void __user *buffer,
156 157
		   size_t *lenp, loff_t *ppos)
{
158
	dentry_stat.nr_dentry = get_nr_dentry();
159
	dentry_stat.nr_unused = get_nr_dentry_unused();
160
	return proc_doulongvec_minmax(table, write, buffer, lenp, ppos);
161 162 163
}
#endif

164 165 166 167
/*
 * Compare 2 name strings, return 0 if they match, otherwise non-zero.
 * The strings are both count bytes long, and count is non-zero.
 */
168 169 170 171 172 173 174 175 176 177 178 179
#ifdef CONFIG_DCACHE_WORD_ACCESS

#include <asm/word-at-a-time.h>
/*
 * NOTE! 'cs' and 'scount' come from a dentry, so it has a
 * aligned allocation for this particular component. We don't
 * strictly need the load_unaligned_zeropad() safety, but it
 * doesn't hurt either.
 *
 * In contrast, 'ct' and 'tcount' can be from a pathname, and do
 * need the careful unaligned handling.
 */
180
static inline int dentry_string_cmp(const unsigned char *cs, const unsigned char *ct, unsigned tcount)
181
{
182 183 184
	unsigned long a,b,mask;

	for (;;) {
185
		a = read_word_at_a_time(cs);
186
		b = load_unaligned_zeropad(ct);
187 188 189 190 191 192 193 194 195 196
		if (tcount < sizeof(unsigned long))
			break;
		if (unlikely(a != b))
			return 1;
		cs += sizeof(unsigned long);
		ct += sizeof(unsigned long);
		tcount -= sizeof(unsigned long);
		if (!tcount)
			return 0;
	}
197
	mask = bytemask_from_count(tcount);
198
	return unlikely(!!((a ^ b) & mask));
199 200
}

201
#else
202

203
static inline int dentry_string_cmp(const unsigned char *cs, const unsigned char *ct, unsigned tcount)
204
{
205 206 207 208 209 210 211 212 213 214
	do {
		if (*cs != *ct)
			return 1;
		cs++;
		ct++;
		tcount--;
	} while (tcount);
	return 0;
}

215 216
#endif

217 218 219 220
static inline int dentry_cmp(const struct dentry *dentry, const unsigned char *ct, unsigned tcount)
{
	/*
	 * Be careful about RCU walk racing with rename:
221
	 * use 'READ_ONCE' to fetch the name pointer.
222 223 224 225 226 227 228 229 230 231 232 233 234
	 *
	 * NOTE! Even if a rename will mean that the length
	 * was not loaded atomically, we don't care. The
	 * RCU walk will check the sequence count eventually,
	 * and catch it. And we won't overrun the buffer,
	 * because we're reading the name pointer atomically,
	 * and a dentry name is guaranteed to be properly
	 * terminated with a NUL byte.
	 *
	 * End result: even if 'len' is wrong, we'll exit
	 * early because the data cannot match (there can
	 * be no NUL in the ct/tcount data)
	 */
235
	const unsigned char *cs = READ_ONCE(dentry->d_name.name);
236

237
	return dentry_string_cmp(cs, ct, tcount);
238 239
}

240 241 242 243 244 245 246 247 248 249 250 251 252
struct external_name {
	union {
		atomic_t count;
		struct rcu_head head;
	} u;
	unsigned char name[];
};

static inline struct external_name *external_name(struct dentry *dentry)
{
	return container_of(dentry->d_name.name, struct external_name, name[0]);
}

C
Christoph Hellwig 已提交
253
static void __d_free(struct rcu_head *head)
L
Linus Torvalds 已提交
254
{
C
Christoph Hellwig 已提交
255 256
	struct dentry *dentry = container_of(head, struct dentry, d_u.d_rcu);

257 258 259 260 261 262 263
	kmem_cache_free(dentry_cache, dentry); 
}

static void __d_free_external(struct rcu_head *head)
{
	struct dentry *dentry = container_of(head, struct dentry, d_u.d_rcu);
	kfree(external_name(dentry));
L
Linus Torvalds 已提交
264 265 266
	kmem_cache_free(dentry_cache, dentry); 
}

A
Al Viro 已提交
267 268 269 270 271
static inline int dname_external(const struct dentry *dentry)
{
	return dentry->d_name.name != dentry->d_iname;
}

A
Al Viro 已提交
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
void take_dentry_name_snapshot(struct name_snapshot *name, struct dentry *dentry)
{
	spin_lock(&dentry->d_lock);
	if (unlikely(dname_external(dentry))) {
		struct external_name *p = external_name(dentry);
		atomic_inc(&p->u.count);
		spin_unlock(&dentry->d_lock);
		name->name = p->name;
	} else {
		memcpy(name->inline_name, dentry->d_iname, DNAME_INLINE_LEN);
		spin_unlock(&dentry->d_lock);
		name->name = name->inline_name;
	}
}
EXPORT_SYMBOL(take_dentry_name_snapshot);

void release_dentry_name_snapshot(struct name_snapshot *name)
{
	if (unlikely(name->name != name->inline_name)) {
		struct external_name *p;
		p = container_of(name->name, struct external_name, name[0]);
		if (unlikely(atomic_dec_and_test(&p->u.count)))
			kfree_rcu(p, u.head);
	}
}
EXPORT_SYMBOL(release_dentry_name_snapshot);

299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320
static inline void __d_set_inode_and_type(struct dentry *dentry,
					  struct inode *inode,
					  unsigned type_flags)
{
	unsigned flags;

	dentry->d_inode = inode;
	flags = READ_ONCE(dentry->d_flags);
	flags &= ~(DCACHE_ENTRY_TYPE | DCACHE_FALLTHRU);
	flags |= type_flags;
	WRITE_ONCE(dentry->d_flags, flags);
}

static inline void __d_clear_type_and_inode(struct dentry *dentry)
{
	unsigned flags = READ_ONCE(dentry->d_flags);

	flags &= ~(DCACHE_ENTRY_TYPE | DCACHE_FALLTHRU);
	WRITE_ONCE(dentry->d_flags, flags);
	dentry->d_inode = NULL;
}

A
Al Viro 已提交
321 322
static void dentry_free(struct dentry *dentry)
{
323
	WARN_ON(!hlist_unhashed(&dentry->d_u.d_alias));
324 325 326 327 328 329 330
	if (unlikely(dname_external(dentry))) {
		struct external_name *p = external_name(dentry);
		if (likely(atomic_dec_and_test(&p->u.count))) {
			call_rcu(&dentry->d_u.d_rcu, __d_free_external);
			return;
		}
	}
A
Al Viro 已提交
331 332 333 334 335 336 337
	/* if dentry was never visible to RCU, immediate free is OK */
	if (!(dentry->d_flags & DCACHE_RCUACCESS))
		__d_free(&dentry->d_u.d_rcu);
	else
		call_rcu(&dentry->d_u.d_rcu, __d_free);
}

L
Linus Torvalds 已提交
338 339
/*
 * Release the dentry's inode, using the filesystem
340
 * d_iput() operation if defined.
N
Nick Piggin 已提交
341 342 343
 */
static void dentry_unlink_inode(struct dentry * dentry)
	__releases(dentry->d_lock)
344
	__releases(dentry->d_inode->i_lock)
N
Nick Piggin 已提交
345 346
{
	struct inode *inode = dentry->d_inode;
347
	bool hashed = !d_unhashed(dentry);
348

349 350
	if (hashed)
		raw_write_seqcount_begin(&dentry->d_seq);
351
	__d_clear_type_and_inode(dentry);
352
	hlist_del_init(&dentry->d_u.d_alias);
353 354
	if (hashed)
		raw_write_seqcount_end(&dentry->d_seq);
N
Nick Piggin 已提交
355
	spin_unlock(&dentry->d_lock);
356
	spin_unlock(&inode->i_lock);
N
Nick Piggin 已提交
357 358 359 360 361 362 363 364
	if (!inode->i_nlink)
		fsnotify_inoderemove(inode);
	if (dentry->d_op && dentry->d_op->d_iput)
		dentry->d_op->d_iput(dentry, inode);
	else
		iput(inode);
}

365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417
/*
 * The DCACHE_LRU_LIST bit is set whenever the 'd_lru' entry
 * is in use - which includes both the "real" per-superblock
 * LRU list _and_ the DCACHE_SHRINK_LIST use.
 *
 * The DCACHE_SHRINK_LIST bit is set whenever the dentry is
 * on the shrink list (ie not on the superblock LRU list).
 *
 * The per-cpu "nr_dentry_unused" counters are updated with
 * the DCACHE_LRU_LIST bit.
 *
 * These helper functions make sure we always follow the
 * rules. d_lock must be held by the caller.
 */
#define D_FLAG_VERIFY(dentry,x) WARN_ON_ONCE(((dentry)->d_flags & (DCACHE_LRU_LIST | DCACHE_SHRINK_LIST)) != (x))
static void d_lru_add(struct dentry *dentry)
{
	D_FLAG_VERIFY(dentry, 0);
	dentry->d_flags |= DCACHE_LRU_LIST;
	this_cpu_inc(nr_dentry_unused);
	WARN_ON_ONCE(!list_lru_add(&dentry->d_sb->s_dentry_lru, &dentry->d_lru));
}

static void d_lru_del(struct dentry *dentry)
{
	D_FLAG_VERIFY(dentry, DCACHE_LRU_LIST);
	dentry->d_flags &= ~DCACHE_LRU_LIST;
	this_cpu_dec(nr_dentry_unused);
	WARN_ON_ONCE(!list_lru_del(&dentry->d_sb->s_dentry_lru, &dentry->d_lru));
}

static void d_shrink_del(struct dentry *dentry)
{
	D_FLAG_VERIFY(dentry, DCACHE_SHRINK_LIST | DCACHE_LRU_LIST);
	list_del_init(&dentry->d_lru);
	dentry->d_flags &= ~(DCACHE_SHRINK_LIST | DCACHE_LRU_LIST);
	this_cpu_dec(nr_dentry_unused);
}

static void d_shrink_add(struct dentry *dentry, struct list_head *list)
{
	D_FLAG_VERIFY(dentry, 0);
	list_add(&dentry->d_lru, list);
	dentry->d_flags |= DCACHE_SHRINK_LIST | DCACHE_LRU_LIST;
	this_cpu_inc(nr_dentry_unused);
}

/*
 * These can only be called under the global LRU lock, ie during the
 * callback for freeing the LRU list. "isolate" removes it from the
 * LRU lists entirely, while shrink_move moves it to the indicated
 * private list.
 */
418
static void d_lru_isolate(struct list_lru_one *lru, struct dentry *dentry)
419 420 421 422
{
	D_FLAG_VERIFY(dentry, DCACHE_LRU_LIST);
	dentry->d_flags &= ~DCACHE_LRU_LIST;
	this_cpu_dec(nr_dentry_unused);
423
	list_lru_isolate(lru, &dentry->d_lru);
424 425
}

426 427
static void d_lru_shrink_move(struct list_lru_one *lru, struct dentry *dentry,
			      struct list_head *list)
428 429 430
{
	D_FLAG_VERIFY(dentry, DCACHE_LRU_LIST);
	dentry->d_flags |= DCACHE_SHRINK_LIST;
431
	list_lru_isolate_move(lru, &dentry->d_lru, list);
432 433
}

N
Nick Piggin 已提交
434 435 436 437 438 439 440 441 442 443 444 445 446
/**
 * d_drop - drop a dentry
 * @dentry: dentry to drop
 *
 * d_drop() unhashes the entry from the parent dentry hashes, so that it won't
 * be found through a VFS lookup any more. Note that this is different from
 * deleting the dentry - d_delete will try to mark the dentry negative if
 * possible, giving a successful _negative_ lookup, while d_drop will
 * just make the cache lookup fail.
 *
 * d_drop() is used mainly for stuff that wants to invalidate a dentry for some
 * reason (NFS timeouts or autofs deletes).
 *
447 448 449
 * __d_drop requires dentry->d_lock
 * ___d_drop doesn't mark dentry as "unhashed"
 *   (dentry->d_hash.pprev will be LIST_POISON2, not NULL).
N
Nick Piggin 已提交
450
 */
451
static void ___d_drop(struct dentry *dentry)
N
Nick Piggin 已提交
452
{
453 454 455 456 457 458 459 460 461 462
	struct hlist_bl_head *b;
	/*
	 * Hashed dentries are normally on the dentry hashtable,
	 * with the exception of those newly allocated by
	 * d_obtain_root, which are always IS_ROOT:
	 */
	if (unlikely(IS_ROOT(dentry)))
		b = &dentry->d_sb->s_roots;
	else
		b = d_hash(dentry->d_name.hash);
463

464 465 466
	hlist_bl_lock(b);
	__hlist_bl_del(&dentry->d_hash);
	hlist_bl_unlock(b);
N
Nick Piggin 已提交
467
}
468 469 470

void __d_drop(struct dentry *dentry)
{
471 472 473 474 475
	if (!d_unhashed(dentry)) {
		___d_drop(dentry);
		dentry->d_hash.pprev = NULL;
		write_seqcount_invalidate(&dentry->d_seq);
	}
476
}
N
Nick Piggin 已提交
477 478 479 480 481 482 483 484 485 486
EXPORT_SYMBOL(__d_drop);

void d_drop(struct dentry *dentry)
{
	spin_lock(&dentry->d_lock);
	__d_drop(dentry);
	spin_unlock(&dentry->d_lock);
}
EXPORT_SYMBOL(d_drop);

A
Al Viro 已提交
487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524
static inline void dentry_unlist(struct dentry *dentry, struct dentry *parent)
{
	struct dentry *next;
	/*
	 * Inform d_walk() and shrink_dentry_list() that we are no longer
	 * attached to the dentry tree
	 */
	dentry->d_flags |= DCACHE_DENTRY_KILLED;
	if (unlikely(list_empty(&dentry->d_child)))
		return;
	__list_del_entry(&dentry->d_child);
	/*
	 * Cursors can move around the list of children.  While we'd been
	 * a normal list member, it didn't matter - ->d_child.next would've
	 * been updated.  However, from now on it won't be and for the
	 * things like d_walk() it might end up with a nasty surprise.
	 * Normally d_walk() doesn't care about cursors moving around -
	 * ->d_lock on parent prevents that and since a cursor has no children
	 * of its own, we get through it without ever unlocking the parent.
	 * There is one exception, though - if we ascend from a child that
	 * gets killed as soon as we unlock it, the next sibling is found
	 * using the value left in its ->d_child.next.  And if _that_
	 * pointed to a cursor, and cursor got moved (e.g. by lseek())
	 * before d_walk() regains parent->d_lock, we'll end up skipping
	 * everything the cursor had been moved past.
	 *
	 * Solution: make sure that the pointer left behind in ->d_child.next
	 * points to something that won't be moving around.  I.e. skip the
	 * cursors.
	 */
	while (dentry->d_child.next != &parent->d_subdirs) {
		next = list_entry(dentry->d_child.next, struct dentry, d_child);
		if (likely(!(next->d_flags & DCACHE_DENTRY_CURSOR)))
			break;
		dentry->d_child.next = next->d_child.next;
	}
}

A
Al Viro 已提交
525
static void __dentry_kill(struct dentry *dentry)
526
{
527 528 529
	struct dentry *parent = NULL;
	bool can_free = true;
	if (!IS_ROOT(dentry))
530
		parent = dentry->d_parent;
N
Nick Piggin 已提交
531

532 533 534 535 536
	/*
	 * The dentry is now unrecoverably dead to the world.
	 */
	lockref_mark_dead(&dentry->d_lockref);

S
Sage Weil 已提交
537 538 539 540
	/*
	 * inform the fs via d_prune that this dentry is about to be
	 * unhashed and destroyed.
	 */
541
	if (dentry->d_flags & DCACHE_OP_PRUNE)
Y
Yan, Zheng 已提交
542 543
		dentry->d_op->d_prune(dentry);

544 545 546 547
	if (dentry->d_flags & DCACHE_LRU_LIST) {
		if (!(dentry->d_flags & DCACHE_SHRINK_LIST))
			d_lru_del(dentry);
	}
548 549
	/* if it was on the hash then remove it */
	__d_drop(dentry);
A
Al Viro 已提交
550
	dentry_unlist(dentry, parent);
A
Al Viro 已提交
551 552
	if (parent)
		spin_unlock(&parent->d_lock);
553 554 555 556
	if (dentry->d_inode)
		dentry_unlink_inode(dentry);
	else
		spin_unlock(&dentry->d_lock);
A
Al Viro 已提交
557 558 559 560
	this_cpu_dec(nr_dentry);
	if (dentry->d_op && dentry->d_op->d_release)
		dentry->d_op->d_release(dentry);

561 562 563 564 565 566 567 568
	spin_lock(&dentry->d_lock);
	if (dentry->d_flags & DCACHE_SHRINK_LIST) {
		dentry->d_flags |= DCACHE_MAY_FREE;
		can_free = false;
	}
	spin_unlock(&dentry->d_lock);
	if (likely(can_free))
		dentry_free(dentry);
A
Al Viro 已提交
569 570
}

571
static struct dentry *__lock_parent(struct dentry *dentry)
572
{
573
	struct dentry *parent;
574
	rcu_read_lock();
575
	spin_unlock(&dentry->d_lock);
576
again:
577
	parent = READ_ONCE(dentry->d_parent);
578 579 580 581 582 583 584 585 586 587 588 589 590
	spin_lock(&parent->d_lock);
	/*
	 * We can't blindly lock dentry until we are sure
	 * that we won't violate the locking order.
	 * Any changes of dentry->d_parent must have
	 * been done with parent->d_lock held, so
	 * spin_lock() above is enough of a barrier
	 * for checking if it's still our child.
	 */
	if (unlikely(parent != dentry->d_parent)) {
		spin_unlock(&parent->d_lock);
		goto again;
	}
591 592
	rcu_read_unlock();
	if (parent != dentry)
593
		spin_lock_nested(&dentry->d_lock, DENTRY_D_LOCK_NESTED);
594
	else
595 596 597 598
		parent = NULL;
	return parent;
}

599 600 601 602 603 604 605 606 607 608
static inline struct dentry *lock_parent(struct dentry *dentry)
{
	struct dentry *parent = dentry->d_parent;
	if (IS_ROOT(dentry))
		return NULL;
	if (likely(spin_trylock(&parent->d_lock)))
		return parent;
	return __lock_parent(dentry);
}

609 610 611 612 613 614 615 616 617 618 619 620 621 622 623
static inline bool retain_dentry(struct dentry *dentry)
{
	WARN_ON(d_in_lookup(dentry));

	/* Unreachable? Get rid of it */
	if (unlikely(d_unhashed(dentry)))
		return false;

	if (unlikely(dentry->d_flags & DCACHE_DISCONNECTED))
		return false;

	if (unlikely(dentry->d_flags & DCACHE_OP_DELETE)) {
		if (dentry->d_op->d_delete(dentry))
			return false;
	}
A
Al Viro 已提交
624 625 626 627 628 629
	/* retain; LRU fodder */
	dentry->d_lockref.count--;
	if (unlikely(!(dentry->d_flags & DCACHE_LRU_LIST)))
		d_lru_add(dentry);
	else if (unlikely(!(dentry->d_flags & DCACHE_REFERENCED)))
		dentry->d_flags |= DCACHE_REFERENCED;
630 631 632
	return true;
}

633 634 635 636 637 638 639 640 641 642 643 644
/*
 * Finish off a dentry we've decided to kill.
 * dentry->d_lock must be held, returns with it unlocked.
 * Returns dentry requiring refcount drop, or NULL if we're done.
 */
static struct dentry *dentry_kill(struct dentry *dentry)
	__releases(dentry->d_lock)
{
	struct inode *inode = dentry->d_inode;
	struct dentry *parent = NULL;

	if (inode && unlikely(!spin_trylock(&inode->i_lock)))
645
		goto slow_positive;
646 647 648 649

	if (!IS_ROOT(dentry)) {
		parent = dentry->d_parent;
		if (unlikely(!spin_trylock(&parent->d_lock))) {
650 651 652 653 654 655 656 657
			parent = __lock_parent(dentry);
			if (likely(inode || !dentry->d_inode))
				goto got_locks;
			/* negative that became positive */
			if (parent)
				spin_unlock(&parent->d_lock);
			inode = dentry->d_inode;
			goto slow_positive;
658 659 660 661 662
		}
	}
	__dentry_kill(dentry);
	return parent;

663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679
slow_positive:
	spin_unlock(&dentry->d_lock);
	spin_lock(&inode->i_lock);
	spin_lock(&dentry->d_lock);
	parent = lock_parent(dentry);
got_locks:
	if (unlikely(dentry->d_lockref.count != 1)) {
		dentry->d_lockref.count--;
	} else if (likely(!retain_dentry(dentry))) {
		__dentry_kill(dentry);
		return parent;
	}
	/* we are keeping it, after all */
	if (inode)
		spin_unlock(&inode->i_lock);
	if (parent)
		spin_unlock(&parent->d_lock);
680
	spin_unlock(&dentry->d_lock);
681
	return NULL;
682 683
}

684 685 686 687 688 689 690 691 692 693 694 695 696 697 698
/*
 * Try to do a lockless dput(), and return whether that was successful.
 *
 * If unsuccessful, we return false, having already taken the dentry lock.
 *
 * The caller needs to hold the RCU read lock, so that the dentry is
 * guaranteed to stay around even if the refcount goes down to zero!
 */
static inline bool fast_dput(struct dentry *dentry)
{
	int ret;
	unsigned int d_flags;

	/*
	 * If we have a d_op->d_delete() operation, we sould not
699
	 * let the dentry count go to zero, so use "put_or_lock".
700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752
	 */
	if (unlikely(dentry->d_flags & DCACHE_OP_DELETE))
		return lockref_put_or_lock(&dentry->d_lockref);

	/*
	 * .. otherwise, we can try to just decrement the
	 * lockref optimistically.
	 */
	ret = lockref_put_return(&dentry->d_lockref);

	/*
	 * If the lockref_put_return() failed due to the lock being held
	 * by somebody else, the fast path has failed. We will need to
	 * get the lock, and then check the count again.
	 */
	if (unlikely(ret < 0)) {
		spin_lock(&dentry->d_lock);
		if (dentry->d_lockref.count > 1) {
			dentry->d_lockref.count--;
			spin_unlock(&dentry->d_lock);
			return 1;
		}
		return 0;
	}

	/*
	 * If we weren't the last ref, we're done.
	 */
	if (ret)
		return 1;

	/*
	 * Careful, careful. The reference count went down
	 * to zero, but we don't hold the dentry lock, so
	 * somebody else could get it again, and do another
	 * dput(), and we need to not race with that.
	 *
	 * However, there is a very special and common case
	 * where we don't care, because there is nothing to
	 * do: the dentry is still hashed, it does not have
	 * a 'delete' op, and it's referenced and already on
	 * the LRU list.
	 *
	 * NOTE! Since we aren't locked, these values are
	 * not "stable". However, it is sufficient that at
	 * some point after we dropped the reference the
	 * dentry was hashed and the flags had the proper
	 * value. Other dentry users may have re-gotten
	 * a reference to the dentry and change that, but
	 * our work is done - we can leave the dentry
	 * around with a zero refcount.
	 */
	smp_rmb();
753
	d_flags = READ_ONCE(dentry->d_flags);
754
	d_flags &= DCACHE_REFERENCED | DCACHE_LRU_LIST | DCACHE_DISCONNECTED;
755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787

	/* Nothing to do? Dropping the reference was all we needed? */
	if (d_flags == (DCACHE_REFERENCED | DCACHE_LRU_LIST) && !d_unhashed(dentry))
		return 1;

	/*
	 * Not the fast normal case? Get the lock. We've already decremented
	 * the refcount, but we'll need to re-check the situation after
	 * getting the lock.
	 */
	spin_lock(&dentry->d_lock);

	/*
	 * Did somebody else grab a reference to it in the meantime, and
	 * we're no longer the last user after all? Alternatively, somebody
	 * else could have killed it and marked it dead. Either way, we
	 * don't need to do anything else.
	 */
	if (dentry->d_lockref.count) {
		spin_unlock(&dentry->d_lock);
		return 1;
	}

	/*
	 * Re-get the reference we optimistically dropped. We hold the
	 * lock, and we just tested that it was zero, so we can just
	 * set it to 1.
	 */
	dentry->d_lockref.count = 1;
	return 0;
}


L
Linus Torvalds 已提交
788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815
/* 
 * This is dput
 *
 * This is complicated by the fact that we do not want to put
 * dentries that are no longer on any hash chain on the unused
 * list: we'd much rather just get rid of them immediately.
 *
 * However, that implies that we have to traverse the dentry
 * tree upwards to the parents which might _also_ now be
 * scheduled for deletion (it may have been only waiting for
 * its last child to go away).
 *
 * This tail recursion is done by hand as we don't want to depend
 * on the compiler to always get this right (gcc generally doesn't).
 * Real recursion would eat up our stack space.
 */

/*
 * dput - release a dentry
 * @dentry: dentry to release 
 *
 * Release a dentry. This will drop the usage count and if appropriate
 * call the dentry unlink method as well as removing it from the queues and
 * releasing its resources. If the parent dentries were scheduled for release
 * they too may now get deleted.
 */
void dput(struct dentry *dentry)
{
816
	if (unlikely(!dentry))
L
Linus Torvalds 已提交
817 818 819
		return;

repeat:
820 821
	might_sleep();

822 823 824
	rcu_read_lock();
	if (likely(fast_dput(dentry))) {
		rcu_read_unlock();
L
Linus Torvalds 已提交
825
		return;
826 827 828 829
	}

	/* Slow case: now with the dentry lock held */
	rcu_read_unlock();
L
Linus Torvalds 已提交
830

831 832 833
	if (likely(retain_dentry(dentry))) {
		spin_unlock(&dentry->d_lock);
		return;
L
Linus Torvalds 已提交
834
	}
N
Nick Piggin 已提交
835

836
	dentry = dentry_kill(dentry);
837 838
	if (dentry) {
		cond_resched();
839
		goto repeat;
840
	}
L
Linus Torvalds 已提交
841
}
842
EXPORT_SYMBOL(dput);
L
Linus Torvalds 已提交
843 844


N
Nick Piggin 已提交
845
/* This must be called with d_lock held */
846
static inline void __dget_dlock(struct dentry *dentry)
N
Nick Piggin 已提交
847
{
848
	dentry->d_lockref.count++;
N
Nick Piggin 已提交
849 850
}

851
static inline void __dget(struct dentry *dentry)
L
Linus Torvalds 已提交
852
{
853
	lockref_get(&dentry->d_lockref);
L
Linus Torvalds 已提交
854 855
}

N
Nick Piggin 已提交
856 857
struct dentry *dget_parent(struct dentry *dentry)
{
858
	int gotref;
N
Nick Piggin 已提交
859 860
	struct dentry *ret;

861 862 863 864 865
	/*
	 * Do optimistic parent lookup without any
	 * locking.
	 */
	rcu_read_lock();
866
	ret = READ_ONCE(dentry->d_parent);
867 868 869
	gotref = lockref_get_not_zero(&ret->d_lockref);
	rcu_read_unlock();
	if (likely(gotref)) {
870
		if (likely(ret == READ_ONCE(dentry->d_parent)))
871 872 873 874
			return ret;
		dput(ret);
	}

N
Nick Piggin 已提交
875
repeat:
N
Nick Piggin 已提交
876 877 878 879 880
	/*
	 * Don't need rcu_dereference because we re-check it was correct under
	 * the lock.
	 */
	rcu_read_lock();
N
Nick Piggin 已提交
881
	ret = dentry->d_parent;
N
Nick Piggin 已提交
882 883 884 885
	spin_lock(&ret->d_lock);
	if (unlikely(ret != dentry->d_parent)) {
		spin_unlock(&ret->d_lock);
		rcu_read_unlock();
N
Nick Piggin 已提交
886 887
		goto repeat;
	}
N
Nick Piggin 已提交
888
	rcu_read_unlock();
889 890
	BUG_ON(!ret->d_lockref.count);
	ret->d_lockref.count++;
N
Nick Piggin 已提交
891 892 893 894 895
	spin_unlock(&ret->d_lock);
	return ret;
}
EXPORT_SYMBOL(dget_parent);

L
Linus Torvalds 已提交
896 897 898 899 900 901 902 903
/**
 * d_find_alias - grab a hashed alias of inode
 * @inode: inode in question
 *
 * If inode has a hashed alias, or is a directory and has any alias,
 * acquire the reference to alias and return it. Otherwise return NULL.
 * Notice that if inode is a directory there can be only one alias and
 * it can be unhashed only if it has no children, or if it is the root
904 905
 * of a filesystem, or if the directory was renamed and d_revalidate
 * was the first vfs operation to notice.
L
Linus Torvalds 已提交
906
 *
907
 * If the inode has an IS_ROOT, DCACHE_DISCONNECTED alias, then prefer
908
 * any other hashed alias over that one.
L
Linus Torvalds 已提交
909
 */
910
static struct dentry *__d_find_alias(struct inode *inode)
L
Linus Torvalds 已提交
911
{
N
Nick Piggin 已提交
912
	struct dentry *alias, *discon_alias;
L
Linus Torvalds 已提交
913

N
Nick Piggin 已提交
914 915
again:
	discon_alias = NULL;
916
	hlist_for_each_entry(alias, &inode->i_dentry, d_u.d_alias) {
N
Nick Piggin 已提交
917
		spin_lock(&alias->d_lock);
L
Linus Torvalds 已提交
918
 		if (S_ISDIR(inode->i_mode) || !d_unhashed(alias)) {
919
			if (IS_ROOT(alias) &&
N
Nick Piggin 已提交
920
			    (alias->d_flags & DCACHE_DISCONNECTED)) {
L
Linus Torvalds 已提交
921
				discon_alias = alias;
922
			} else {
923
				__dget_dlock(alias);
N
Nick Piggin 已提交
924 925 926 927 928 929 930 931 932 933
				spin_unlock(&alias->d_lock);
				return alias;
			}
		}
		spin_unlock(&alias->d_lock);
	}
	if (discon_alias) {
		alias = discon_alias;
		spin_lock(&alias->d_lock);
		if (S_ISDIR(inode->i_mode) || !d_unhashed(alias)) {
934 935 936
			__dget_dlock(alias);
			spin_unlock(&alias->d_lock);
			return alias;
L
Linus Torvalds 已提交
937
		}
N
Nick Piggin 已提交
938 939
		spin_unlock(&alias->d_lock);
		goto again;
L
Linus Torvalds 已提交
940
	}
N
Nick Piggin 已提交
941
	return NULL;
L
Linus Torvalds 已提交
942 943
}

N
Nick Piggin 已提交
944
struct dentry *d_find_alias(struct inode *inode)
L
Linus Torvalds 已提交
945
{
D
David Howells 已提交
946 947
	struct dentry *de = NULL;

A
Al Viro 已提交
948
	if (!hlist_empty(&inode->i_dentry)) {
949
		spin_lock(&inode->i_lock);
950
		de = __d_find_alias(inode);
951
		spin_unlock(&inode->i_lock);
D
David Howells 已提交
952
	}
L
Linus Torvalds 已提交
953 954
	return de;
}
955
EXPORT_SYMBOL(d_find_alias);
L
Linus Torvalds 已提交
956 957 958 959 960 961 962

/*
 *	Try to kill dentries associated with this inode.
 * WARNING: you must own a reference to inode.
 */
void d_prune_aliases(struct inode *inode)
{
963
	struct dentry *dentry;
L
Linus Torvalds 已提交
964
restart:
965
	spin_lock(&inode->i_lock);
966
	hlist_for_each_entry(dentry, &inode->i_dentry, d_u.d_alias) {
L
Linus Torvalds 已提交
967
		spin_lock(&dentry->d_lock);
968
		if (!dentry->d_lockref.count) {
969 970 971
			struct dentry *parent = lock_parent(dentry);
			if (likely(!dentry->d_lockref.count)) {
				__dentry_kill(dentry);
972
				dput(parent);
973 974 975 976
				goto restart;
			}
			if (parent)
				spin_unlock(&parent->d_lock);
L
Linus Torvalds 已提交
977 978 979
		}
		spin_unlock(&dentry->d_lock);
	}
980
	spin_unlock(&inode->i_lock);
L
Linus Torvalds 已提交
981
}
982
EXPORT_SYMBOL(d_prune_aliases);
L
Linus Torvalds 已提交
983

984 985
/*
 * Lock a dentry from shrink list.
986 987
 * Called under rcu_read_lock() and dentry->d_lock; the former
 * guarantees that nothing we access will be freed under us.
988
 * Note that dentry is *not* protected from concurrent dentry_kill(),
989 990
 * d_delete(), etc.
 *
991 992 993 994 995
 * Return false if dentry has been disrupted or grabbed, leaving
 * the caller to kick it off-list.  Otherwise, return true and have
 * that dentry's inode and parent both locked.
 */
static bool shrink_lock_dentry(struct dentry *dentry)
L
Linus Torvalds 已提交
996
{
997 998
	struct inode *inode;
	struct dentry *parent;
999

1000 1001 1002 1003 1004 1005 1006
	if (dentry->d_lockref.count)
		return false;

	inode = dentry->d_inode;
	if (inode && unlikely(!spin_trylock(&inode->i_lock))) {
		spin_unlock(&dentry->d_lock);
		spin_lock(&inode->i_lock);
1007
		spin_lock(&dentry->d_lock);
1008 1009 1010 1011 1012 1013
		if (unlikely(dentry->d_lockref.count))
			goto out;
		/* changed inode means that somebody had grabbed it */
		if (unlikely(inode != dentry->d_inode))
			goto out;
	}
1014

1015 1016 1017
	parent = dentry->d_parent;
	if (IS_ROOT(dentry) || likely(spin_trylock(&parent->d_lock)))
		return true;
1018

1019 1020 1021 1022 1023 1024 1025 1026
	spin_unlock(&dentry->d_lock);
	spin_lock(&parent->d_lock);
	if (unlikely(parent != dentry->d_parent)) {
		spin_unlock(&parent->d_lock);
		spin_lock(&dentry->d_lock);
		goto out;
	}
	spin_lock_nested(&dentry->d_lock, DENTRY_D_LOCK_NESTED);
1027
	if (likely(!dentry->d_lockref.count))
1028 1029 1030 1031 1032 1033 1034
		return true;
	spin_unlock(&parent->d_lock);
out:
	if (inode)
		spin_unlock(&inode->i_lock);
	return false;
}
1035

1036 1037 1038 1039
static void shrink_dentry_list(struct list_head *list)
{
	while (!list_empty(list)) {
		struct dentry *dentry, *parent;
1040

1041 1042
		dentry = list_entry(list->prev, struct dentry, d_lru);
		spin_lock(&dentry->d_lock);
1043
		rcu_read_lock();
1044 1045
		if (!shrink_lock_dentry(dentry)) {
			bool can_free = false;
1046
			rcu_read_unlock();
1047 1048 1049
			d_shrink_del(dentry);
			if (dentry->d_lockref.count < 0)
				can_free = dentry->d_flags & DCACHE_MAY_FREE;
1050 1051 1052 1053 1054
			spin_unlock(&dentry->d_lock);
			if (can_free)
				dentry_free(dentry);
			continue;
		}
1055
		rcu_read_unlock();
1056 1057
		d_shrink_del(dentry);
		parent = dentry->d_parent;
1058
		__dentry_kill(dentry);
1059 1060
		if (parent == dentry)
			continue;
A
Al Viro 已提交
1061 1062 1063 1064 1065 1066 1067
		/*
		 * We need to prune ancestors too. This is necessary to prevent
		 * quadratic behavior of shrink_dcache_parent(), but is also
		 * expected to be beneficial in reducing dentry cache
		 * fragmentation.
		 */
		dentry = parent;
1068 1069
		while (dentry && !lockref_put_or_lock(&dentry->d_lockref))
			dentry = dentry_kill(dentry);
1070
	}
C
Christoph Hellwig 已提交
1071 1072
}

1073 1074
static enum lru_status dentry_lru_isolate(struct list_head *item,
		struct list_lru_one *lru, spinlock_t *lru_lock, void *arg)
1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093
{
	struct list_head *freeable = arg;
	struct dentry	*dentry = container_of(item, struct dentry, d_lru);


	/*
	 * we are inverting the lru lock/dentry->d_lock here,
	 * so use a trylock. If we fail to get the lock, just skip
	 * it
	 */
	if (!spin_trylock(&dentry->d_lock))
		return LRU_SKIP;

	/*
	 * Referenced dentries are still in use. If they have active
	 * counts, just remove them from the LRU. Otherwise give them
	 * another pass through the LRU.
	 */
	if (dentry->d_lockref.count) {
1094
		d_lru_isolate(lru, dentry);
1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124
		spin_unlock(&dentry->d_lock);
		return LRU_REMOVED;
	}

	if (dentry->d_flags & DCACHE_REFERENCED) {
		dentry->d_flags &= ~DCACHE_REFERENCED;
		spin_unlock(&dentry->d_lock);

		/*
		 * The list move itself will be made by the common LRU code. At
		 * this point, we've dropped the dentry->d_lock but keep the
		 * lru lock. This is safe to do, since every list movement is
		 * protected by the lru lock even if both locks are held.
		 *
		 * This is guaranteed by the fact that all LRU management
		 * functions are intermediated by the LRU API calls like
		 * list_lru_add and list_lru_del. List movement in this file
		 * only ever occur through this functions or through callbacks
		 * like this one, that are called from the LRU API.
		 *
		 * The only exceptions to this are functions like
		 * shrink_dentry_list, and code that first checks for the
		 * DCACHE_SHRINK_LIST flag.  Those are guaranteed to be
		 * operating only with stack provided lists after they are
		 * properly isolated from the main list.  It is thus, always a
		 * local access.
		 */
		return LRU_ROTATE;
	}

1125
	d_lru_shrink_move(lru, dentry, freeable);
1126 1127 1128 1129 1130
	spin_unlock(&dentry->d_lock);

	return LRU_REMOVED;
}

C
Christoph Hellwig 已提交
1131
/**
1132 1133
 * prune_dcache_sb - shrink the dcache
 * @sb: superblock
1134
 * @sc: shrink control, passed to list_lru_shrink_walk()
1135
 *
1136 1137
 * Attempt to shrink the superblock dcache LRU by @sc->nr_to_scan entries. This
 * is done when we need more memory and called from the superblock shrinker
1138
 * function.
C
Christoph Hellwig 已提交
1139
 *
1140 1141
 * This function may fail to free any resources if all the dentries are in
 * use.
C
Christoph Hellwig 已提交
1142
 */
1143
long prune_dcache_sb(struct super_block *sb, struct shrink_control *sc)
C
Christoph Hellwig 已提交
1144
{
1145 1146
	LIST_HEAD(dispose);
	long freed;
C
Christoph Hellwig 已提交
1147

1148 1149
	freed = list_lru_shrink_walk(&sb->s_dentry_lru, sc,
				     dentry_lru_isolate, &dispose);
1150
	shrink_dentry_list(&dispose);
1151
	return freed;
1152
}
N
Nick Piggin 已提交
1153

1154
static enum lru_status dentry_lru_isolate_shrink(struct list_head *item,
1155
		struct list_lru_one *lru, spinlock_t *lru_lock, void *arg)
1156
{
1157 1158
	struct list_head *freeable = arg;
	struct dentry	*dentry = container_of(item, struct dentry, d_lru);
1159

1160 1161 1162 1163 1164 1165 1166 1167
	/*
	 * we are inverting the lru lock/dentry->d_lock here,
	 * so use a trylock. If we fail to get the lock, just skip
	 * it
	 */
	if (!spin_trylock(&dentry->d_lock))
		return LRU_SKIP;

1168
	d_lru_shrink_move(lru, dentry, freeable);
1169
	spin_unlock(&dentry->d_lock);
1170

1171
	return LRU_REMOVED;
1172 1173
}

1174

L
Linus Torvalds 已提交
1175 1176 1177 1178
/**
 * shrink_dcache_sb - shrink dcache for a superblock
 * @sb: superblock
 *
C
Christoph Hellwig 已提交
1179 1180
 * Shrink the dcache for the specified super block. This is used to free
 * the dcache before unmounting a file system.
L
Linus Torvalds 已提交
1181
 */
C
Christoph Hellwig 已提交
1182
void shrink_dcache_sb(struct super_block *sb)
L
Linus Torvalds 已提交
1183
{
1184 1185 1186 1187 1188 1189
	long freed;

	do {
		LIST_HEAD(dispose);

		freed = list_lru_walk(&sb->s_dentry_lru,
1190
			dentry_lru_isolate_shrink, &dispose, 1024);
C
Christoph Hellwig 已提交
1191

1192 1193
		this_cpu_sub(nr_dentry_unused, freed);
		shrink_dentry_list(&dispose);
1194 1195
		cond_resched();
	} while (list_lru_count(&sb->s_dentry_lru) > 0);
L
Linus Torvalds 已提交
1196
}
1197
EXPORT_SYMBOL(shrink_dcache_sb);
L
Linus Torvalds 已提交
1198

M
Miklos Szeredi 已提交
1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211
/**
 * enum d_walk_ret - action to talke during tree walk
 * @D_WALK_CONTINUE:	contrinue walk
 * @D_WALK_QUIT:	quit walk
 * @D_WALK_NORETRY:	quit when retry is needed
 * @D_WALK_SKIP:	skip this dentry and its children
 */
enum d_walk_ret {
	D_WALK_CONTINUE,
	D_WALK_QUIT,
	D_WALK_NORETRY,
	D_WALK_SKIP,
};
1212

L
Linus Torvalds 已提交
1213
/**
M
Miklos Szeredi 已提交
1214 1215 1216 1217 1218
 * d_walk - walk the dentry tree
 * @parent:	start of walk
 * @data:	data passed to @enter() and @finish()
 * @enter:	callback when first entering the dentry
 * @finish:	callback when successfully finished the walk
L
Linus Torvalds 已提交
1219
 *
M
Miklos Szeredi 已提交
1220
 * The @enter() and @finish() callbacks are called with d_lock held.
L
Linus Torvalds 已提交
1221
 */
M
Miklos Szeredi 已提交
1222 1223 1224
static void d_walk(struct dentry *parent, void *data,
		   enum d_walk_ret (*enter)(void *, struct dentry *),
		   void (*finish)(void *))
L
Linus Torvalds 已提交
1225
{
1226
	struct dentry *this_parent;
L
Linus Torvalds 已提交
1227
	struct list_head *next;
1228
	unsigned seq = 0;
M
Miklos Szeredi 已提交
1229 1230
	enum d_walk_ret ret;
	bool retry = true;
1231

1232
again:
1233
	read_seqbegin_or_lock(&rename_lock, &seq);
1234
	this_parent = parent;
N
Nick Piggin 已提交
1235
	spin_lock(&this_parent->d_lock);
M
Miklos Szeredi 已提交
1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247

	ret = enter(data, this_parent);
	switch (ret) {
	case D_WALK_CONTINUE:
		break;
	case D_WALK_QUIT:
	case D_WALK_SKIP:
		goto out_unlock;
	case D_WALK_NORETRY:
		retry = false;
		break;
	}
L
Linus Torvalds 已提交
1248 1249 1250 1251 1252
repeat:
	next = this_parent->d_subdirs.next;
resume:
	while (next != &this_parent->d_subdirs) {
		struct list_head *tmp = next;
1253
		struct dentry *dentry = list_entry(tmp, struct dentry, d_child);
L
Linus Torvalds 已提交
1254
		next = tmp->next;
N
Nick Piggin 已提交
1255

A
Al Viro 已提交
1256 1257 1258
		if (unlikely(dentry->d_flags & DCACHE_DENTRY_CURSOR))
			continue;

N
Nick Piggin 已提交
1259
		spin_lock_nested(&dentry->d_lock, DENTRY_D_LOCK_NESTED);
M
Miklos Szeredi 已提交
1260 1261 1262 1263 1264 1265

		ret = enter(data, dentry);
		switch (ret) {
		case D_WALK_CONTINUE:
			break;
		case D_WALK_QUIT:
N
Nick Piggin 已提交
1266
			spin_unlock(&dentry->d_lock);
M
Miklos Szeredi 已提交
1267 1268 1269 1270 1271 1272 1273
			goto out_unlock;
		case D_WALK_NORETRY:
			retry = false;
			break;
		case D_WALK_SKIP:
			spin_unlock(&dentry->d_lock);
			continue;
N
Nick Piggin 已提交
1274
		}
M
Miklos Szeredi 已提交
1275

L
Linus Torvalds 已提交
1276
		if (!list_empty(&dentry->d_subdirs)) {
N
Nick Piggin 已提交
1277 1278
			spin_unlock(&this_parent->d_lock);
			spin_release(&dentry->d_lock.dep_map, 1, _RET_IP_);
L
Linus Torvalds 已提交
1279
			this_parent = dentry;
N
Nick Piggin 已提交
1280
			spin_acquire(&this_parent->d_lock.dep_map, 0, 1, _RET_IP_);
L
Linus Torvalds 已提交
1281 1282
			goto repeat;
		}
N
Nick Piggin 已提交
1283
		spin_unlock(&dentry->d_lock);
L
Linus Torvalds 已提交
1284 1285 1286 1287
	}
	/*
	 * All done at this level ... ascend and resume the search.
	 */
A
Al Viro 已提交
1288 1289
	rcu_read_lock();
ascend:
L
Linus Torvalds 已提交
1290
	if (this_parent != parent) {
1291
		struct dentry *child = this_parent;
1292 1293 1294 1295 1296
		this_parent = child->d_parent;

		spin_unlock(&child->d_lock);
		spin_lock(&this_parent->d_lock);

A
Al Viro 已提交
1297 1298
		/* might go back up the wrong parent if we have had a rename. */
		if (need_seqretry(&rename_lock, seq))
1299
			goto rename_retry;
A
Al Viro 已提交
1300 1301 1302
		/* go into the first sibling still alive */
		do {
			next = child->d_child.next;
A
Al Viro 已提交
1303 1304 1305
			if (next == &this_parent->d_subdirs)
				goto ascend;
			child = list_entry(next, struct dentry, d_child);
A
Al Viro 已提交
1306
		} while (unlikely(child->d_flags & DCACHE_DENTRY_KILLED));
1307
		rcu_read_unlock();
L
Linus Torvalds 已提交
1308 1309
		goto resume;
	}
A
Al Viro 已提交
1310
	if (need_seqretry(&rename_lock, seq))
1311
		goto rename_retry;
A
Al Viro 已提交
1312
	rcu_read_unlock();
M
Miklos Szeredi 已提交
1313 1314 1315 1316 1317
	if (finish)
		finish(data);

out_unlock:
	spin_unlock(&this_parent->d_lock);
1318
	done_seqretry(&rename_lock, seq);
M
Miklos Szeredi 已提交
1319
	return;
1320 1321

rename_retry:
A
Al Viro 已提交
1322 1323 1324
	spin_unlock(&this_parent->d_lock);
	rcu_read_unlock();
	BUG_ON(seq & 1);
M
Miklos Szeredi 已提交
1325 1326
	if (!retry)
		return;
1327
	seq = 1;
1328
	goto again;
L
Linus Torvalds 已提交
1329
}
M
Miklos Szeredi 已提交
1330

I
Ian Kent 已提交
1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369
struct check_mount {
	struct vfsmount *mnt;
	unsigned int mounted;
};

static enum d_walk_ret path_check_mount(void *data, struct dentry *dentry)
{
	struct check_mount *info = data;
	struct path path = { .mnt = info->mnt, .dentry = dentry };

	if (likely(!d_mountpoint(dentry)))
		return D_WALK_CONTINUE;
	if (__path_is_mountpoint(&path)) {
		info->mounted = 1;
		return D_WALK_QUIT;
	}
	return D_WALK_CONTINUE;
}

/**
 * path_has_submounts - check for mounts over a dentry in the
 *                      current namespace.
 * @parent: path to check.
 *
 * Return true if the parent or its subdirectories contain
 * a mount point in the current namespace.
 */
int path_has_submounts(const struct path *parent)
{
	struct check_mount data = { .mnt = parent->mnt, .mounted = 0 };

	read_seqlock_excl(&mount_lock);
	d_walk(parent->dentry, &data, path_check_mount, NULL);
	read_sequnlock_excl(&mount_lock);

	return data.mounted;
}
EXPORT_SYMBOL(path_has_submounts);

1370 1371 1372 1373 1374
/*
 * Called by mount code to set a mountpoint and check if the mountpoint is
 * reachable (e.g. NFS can unhash a directory dentry and then the complete
 * subtree can become unreachable).
 *
1375
 * Only one of d_invalidate() and d_set_mounted() must succeed.  For
1376 1377 1378 1379 1380 1381 1382 1383
 * this reason take rename_lock and d_lock on dentry and ancestors.
 */
int d_set_mounted(struct dentry *dentry)
{
	struct dentry *p;
	int ret = -ENOENT;
	write_seqlock(&rename_lock);
	for (p = dentry->d_parent; !IS_ROOT(p); p = p->d_parent) {
1384
		/* Need exclusion wrt. d_invalidate() */
1385 1386 1387 1388 1389 1390 1391 1392 1393
		spin_lock(&p->d_lock);
		if (unlikely(d_unhashed(p))) {
			spin_unlock(&p->d_lock);
			goto out;
		}
		spin_unlock(&p->d_lock);
	}
	spin_lock(&dentry->d_lock);
	if (!d_unlinked(dentry)) {
1394 1395 1396 1397 1398
		ret = -EBUSY;
		if (!d_mountpoint(dentry)) {
			dentry->d_flags |= DCACHE_MOUNTED;
			ret = 0;
		}
1399 1400 1401 1402 1403 1404 1405
	}
 	spin_unlock(&dentry->d_lock);
out:
	write_sequnlock(&rename_lock);
	return ret;
}

L
Linus Torvalds 已提交
1406
/*
1407
 * Search the dentry child list of the specified parent,
L
Linus Torvalds 已提交
1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420
 * and move any unused dentries to the end of the unused
 * list for prune_dcache(). We descend to the next level
 * whenever the d_subdirs list is non-empty and continue
 * searching.
 *
 * It returns zero iff there are no unused children,
 * otherwise  it returns the number of children moved to
 * the end of the unused list. This may not be the total
 * number of unused children, because select_parent can
 * drop the lock and return early due to latency
 * constraints.
 */

M
Miklos Szeredi 已提交
1421 1422 1423 1424 1425
struct select_data {
	struct dentry *start;
	struct list_head dispose;
	int found;
};
N
Nick Piggin 已提交
1426

M
Miklos Szeredi 已提交
1427 1428 1429 1430
static enum d_walk_ret select_collect(void *_data, struct dentry *dentry)
{
	struct select_data *data = _data;
	enum d_walk_ret ret = D_WALK_CONTINUE;
L
Linus Torvalds 已提交
1431

M
Miklos Szeredi 已提交
1432 1433
	if (data->start == dentry)
		goto out;
N
Nick Piggin 已提交
1434

1435
	if (dentry->d_flags & DCACHE_SHRINK_LIST) {
M
Miklos Szeredi 已提交
1436
		data->found++;
1437 1438 1439 1440 1441 1442 1443
	} else {
		if (dentry->d_flags & DCACHE_LRU_LIST)
			d_lru_del(dentry);
		if (!dentry->d_lockref.count) {
			d_shrink_add(dentry, &data->dispose);
			data->found++;
		}
L
Linus Torvalds 已提交
1444
	}
M
Miklos Szeredi 已提交
1445 1446 1447 1448 1449
	/*
	 * We can return to the caller if we have found some (this
	 * ensures forward progress). We'll be coming back to find
	 * the rest.
	 */
1450 1451
	if (!list_empty(&data->dispose))
		ret = need_resched() ? D_WALK_QUIT : D_WALK_NORETRY;
L
Linus Torvalds 已提交
1452
out:
M
Miklos Szeredi 已提交
1453
	return ret;
L
Linus Torvalds 已提交
1454 1455 1456 1457 1458 1459 1460 1461
}

/**
 * shrink_dcache_parent - prune dcache
 * @parent: parent of entries to prune
 *
 * Prune the dcache to remove unused children of the parent dentry.
 */
M
Miklos Szeredi 已提交
1462
void shrink_dcache_parent(struct dentry *parent)
L
Linus Torvalds 已提交
1463
{
M
Miklos Szeredi 已提交
1464 1465
	for (;;) {
		struct select_data data;
L
Linus Torvalds 已提交
1466

M
Miklos Szeredi 已提交
1467 1468 1469 1470 1471 1472 1473 1474 1475
		INIT_LIST_HEAD(&data.dispose);
		data.start = parent;
		data.found = 0;

		d_walk(parent, &data, select_collect, NULL);
		if (!data.found)
			break;

		shrink_dentry_list(&data.dispose);
1476 1477
		cond_resched();
	}
L
Linus Torvalds 已提交
1478
}
1479
EXPORT_SYMBOL(shrink_dcache_parent);
L
Linus Torvalds 已提交
1480

1481
static enum d_walk_ret umount_check(void *_data, struct dentry *dentry)
1482
{
1483 1484 1485
	/* it has busy descendents; complain about those instead */
	if (!list_empty(&dentry->d_subdirs))
		return D_WALK_CONTINUE;
1486

1487 1488 1489 1490 1491 1492
	/* root with refcount 1 is fine */
	if (dentry == _data && dentry->d_lockref.count == 1)
		return D_WALK_CONTINUE;

	printk(KERN_ERR "BUG: Dentry %p{i=%lx,n=%pd} "
			" still in use (%d) [unmount of %s %s]\n",
1493 1494 1495
		       dentry,
		       dentry->d_inode ?
		       dentry->d_inode->i_ino : 0UL,
1496
		       dentry,
1497 1498 1499
		       dentry->d_lockref.count,
		       dentry->d_sb->s_type->name,
		       dentry->d_sb->s_id);
1500 1501 1502 1503 1504 1505 1506 1507 1508 1509
	WARN_ON(1);
	return D_WALK_CONTINUE;
}

static void do_one_tree(struct dentry *dentry)
{
	shrink_dcache_parent(dentry);
	d_walk(dentry, dentry, umount_check, NULL);
	d_drop(dentry);
	dput(dentry);
1510 1511 1512 1513 1514 1515 1516 1517 1518
}

/*
 * destroy the dentries attached to a superblock on unmounting
 */
void shrink_dcache_for_umount(struct super_block *sb)
{
	struct dentry *dentry;

1519
	WARN(down_read_trylock(&sb->s_umount), "s_umount should've been locked");
1520 1521 1522

	dentry = sb->s_root;
	sb->s_root = NULL;
1523
	do_one_tree(dentry);
1524

1525 1526
	while (!hlist_bl_empty(&sb->s_roots)) {
		dentry = dget(hlist_bl_entry(hlist_bl_first(&sb->s_roots), struct dentry, d_hash));
1527
		do_one_tree(dentry);
1528 1529 1530
	}
}

1531 1532 1533 1534 1535
struct detach_data {
	struct select_data select;
	struct dentry *mountpoint;
};
static enum d_walk_ret detach_and_collect(void *_data, struct dentry *dentry)
1536
{
1537
	struct detach_data *data = _data;
1538 1539

	if (d_mountpoint(dentry)) {
1540 1541
		__dget_dlock(dentry);
		data->mountpoint = dentry;
1542 1543 1544
		return D_WALK_QUIT;
	}

1545
	return select_collect(&data->select, dentry);
1546 1547 1548 1549
}

static void check_and_drop(void *_data)
{
1550
	struct detach_data *data = _data;
1551

1552
	if (!data->mountpoint && list_empty(&data->select.dispose))
1553
		__d_drop(data->select.start);
1554 1555 1556
}

/**
1557 1558 1559 1560
 * d_invalidate - detach submounts, prune dcache, and drop
 * @dentry: dentry to invalidate (aka detach, prune and drop)
 *
 * no dcache lock.
1561
 *
1562 1563 1564
 * The final d_drop is done as an atomic operation relative to
 * rename_lock ensuring there are no races with d_set_mounted.  This
 * ensures there are no unhashed dentries on the path to a mountpoint.
1565
 */
1566
void d_invalidate(struct dentry *dentry)
1567
{
1568 1569 1570 1571 1572 1573
	/*
	 * If it's already been dropped, return OK.
	 */
	spin_lock(&dentry->d_lock);
	if (d_unhashed(dentry)) {
		spin_unlock(&dentry->d_lock);
1574
		return;
1575 1576 1577
	}
	spin_unlock(&dentry->d_lock);

1578 1579 1580
	/* Negative dentries can be dropped without further checks */
	if (!dentry->d_inode) {
		d_drop(dentry);
1581
		return;
1582 1583 1584
	}

	for (;;) {
1585
		struct detach_data data;
1586

1587 1588 1589 1590 1591 1592
		data.mountpoint = NULL;
		INIT_LIST_HEAD(&data.select.dispose);
		data.select.start = dentry;
		data.select.found = 0;

		d_walk(dentry, &data, detach_and_collect, check_and_drop);
1593

1594
		if (!list_empty(&data.select.dispose))
1595
			shrink_dentry_list(&data.select.dispose);
1596 1597
		else if (!data.mountpoint)
			return;
1598

1599 1600 1601 1602
		if (data.mountpoint) {
			detach_mounts(data.mountpoint);
			dput(data.mountpoint);
		}
1603 1604 1605
		cond_resched();
	}
}
1606
EXPORT_SYMBOL(d_invalidate);
1607

L
Linus Torvalds 已提交
1608
/**
1609 1610
 * __d_alloc	-	allocate a dcache entry
 * @sb: filesystem it will belong to
L
Linus Torvalds 已提交
1611 1612 1613 1614 1615 1616 1617
 * @name: qstr of the name
 *
 * Allocates a dentry. It returns %NULL if there is insufficient memory
 * available. On a success the dentry is returned. The name passed in is
 * copied and the copy passed in may be reused after this call.
 */
 
1618
struct dentry *__d_alloc(struct super_block *sb, const struct qstr *name)
L
Linus Torvalds 已提交
1619 1620 1621
{
	struct dentry *dentry;
	char *dname;
M
Miklos Szeredi 已提交
1622
	int err;
L
Linus Torvalds 已提交
1623

1624
	dentry = kmem_cache_alloc(dentry_cache, GFP_KERNEL);
L
Linus Torvalds 已提交
1625 1626 1627
	if (!dentry)
		return NULL;

1628 1629 1630 1631 1632 1633 1634
	/*
	 * We guarantee that the inline name is always NUL-terminated.
	 * This way the memcpy() done by the name switching in rename
	 * will still always have a NUL at the end, even if we might
	 * be overwriting an internal NUL character
	 */
	dentry->d_iname[DNAME_INLINE_LEN-1] = 0;
1635
	if (unlikely(!name)) {
D
David Howells 已提交
1636
		name = &slash_name;
1637 1638
		dname = dentry->d_iname;
	} else if (name->len > DNAME_INLINE_LEN-1) {
1639
		size_t size = offsetof(struct external_name, name[1]);
1640 1641
		struct external_name *p = kmalloc(size + name->len,
						  GFP_KERNEL_ACCOUNT);
1642
		if (!p) {
L
Linus Torvalds 已提交
1643 1644 1645
			kmem_cache_free(dentry_cache, dentry); 
			return NULL;
		}
1646 1647
		atomic_set(&p->u.count, 1);
		dname = p->name;
L
Linus Torvalds 已提交
1648 1649 1650 1651 1652 1653 1654 1655 1656
	} else  {
		dname = dentry->d_iname;
	}	

	dentry->d_name.len = name->len;
	dentry->d_name.hash = name->hash;
	memcpy(dname, name->name, name->len);
	dname[name->len] = 0;

1657
	/* Make sure we always see the terminating NUL character */
1658
	smp_store_release(&dentry->d_name.name, dname); /* ^^^ */
1659

1660
	dentry->d_lockref.count = 1;
1661
	dentry->d_flags = 0;
L
Linus Torvalds 已提交
1662
	spin_lock_init(&dentry->d_lock);
N
Nick Piggin 已提交
1663
	seqcount_init(&dentry->d_seq);
L
Linus Torvalds 已提交
1664
	dentry->d_inode = NULL;
1665 1666
	dentry->d_parent = dentry;
	dentry->d_sb = sb;
L
Linus Torvalds 已提交
1667 1668
	dentry->d_op = NULL;
	dentry->d_fsdata = NULL;
1669
	INIT_HLIST_BL_NODE(&dentry->d_hash);
L
Linus Torvalds 已提交
1670 1671
	INIT_LIST_HEAD(&dentry->d_lru);
	INIT_LIST_HEAD(&dentry->d_subdirs);
1672 1673
	INIT_HLIST_NODE(&dentry->d_u.d_alias);
	INIT_LIST_HEAD(&dentry->d_child);
1674
	d_set_d_op(dentry, dentry->d_sb->s_d_op);
L
Linus Torvalds 已提交
1675

M
Miklos Szeredi 已提交
1676 1677 1678 1679 1680 1681 1682 1683 1684 1685
	if (dentry->d_op && dentry->d_op->d_init) {
		err = dentry->d_op->d_init(dentry);
		if (err) {
			if (dname_external(dentry))
				kfree(external_name(dentry));
			kmem_cache_free(dentry_cache, dentry);
			return NULL;
		}
	}

1686
	this_cpu_inc(nr_dentry);
1687

L
Linus Torvalds 已提交
1688 1689
	return dentry;
}
1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704

/**
 * d_alloc	-	allocate a dcache entry
 * @parent: parent of entry to allocate
 * @name: qstr of the name
 *
 * Allocates a dentry. It returns %NULL if there is insufficient memory
 * available. On a success the dentry is returned. The name passed in is
 * copied and the copy passed in may be reused after this call.
 */
struct dentry *d_alloc(struct dentry * parent, const struct qstr *name)
{
	struct dentry *dentry = __d_alloc(parent->d_sb, name);
	if (!dentry)
		return NULL;
1705
	dentry->d_flags |= DCACHE_RCUACCESS;
1706 1707 1708 1709 1710 1711 1712
	spin_lock(&parent->d_lock);
	/*
	 * don't need child lock because it is not subject
	 * to concurrency here
	 */
	__dget_dlock(parent);
	dentry->d_parent = parent;
1713
	list_add(&dentry->d_child, &parent->d_subdirs);
1714 1715 1716 1717
	spin_unlock(&parent->d_lock);

	return dentry;
}
1718
EXPORT_SYMBOL(d_alloc);
L
Linus Torvalds 已提交
1719

1720 1721 1722 1723 1724 1725
struct dentry *d_alloc_anon(struct super_block *sb)
{
	return __d_alloc(sb, NULL);
}
EXPORT_SYMBOL(d_alloc_anon);

A
Al Viro 已提交
1726 1727
struct dentry *d_alloc_cursor(struct dentry * parent)
{
1728
	struct dentry *dentry = d_alloc_anon(parent->d_sb);
A
Al Viro 已提交
1729 1730 1731 1732 1733 1734 1735
	if (dentry) {
		dentry->d_flags |= DCACHE_RCUACCESS | DCACHE_DENTRY_CURSOR;
		dentry->d_parent = dget(parent);
	}
	return dentry;
}

1736 1737 1738 1739 1740 1741 1742 1743
/**
 * d_alloc_pseudo - allocate a dentry (for lookup-less filesystems)
 * @sb: the superblock
 * @name: qstr of the name
 *
 * For a filesystem that just pins its dentries in memory and never
 * performs lookups at all, return an unhashed IS_ROOT dentry.
 */
1744 1745
struct dentry *d_alloc_pseudo(struct super_block *sb, const struct qstr *name)
{
1746
	return __d_alloc(sb, name);
1747 1748 1749
}
EXPORT_SYMBOL(d_alloc_pseudo);

L
Linus Torvalds 已提交
1750 1751 1752 1753 1754
struct dentry *d_alloc_name(struct dentry *parent, const char *name)
{
	struct qstr q;

	q.name = name;
1755
	q.hash_len = hashlen_string(parent, name);
L
Linus Torvalds 已提交
1756 1757
	return d_alloc(parent, &q);
}
1758
EXPORT_SYMBOL(d_alloc_name);
L
Linus Torvalds 已提交
1759

1760 1761
void d_set_d_op(struct dentry *dentry, const struct dentry_operations *op)
{
1762 1763
	WARN_ON_ONCE(dentry->d_op);
	WARN_ON_ONCE(dentry->d_flags & (DCACHE_OP_HASH	|
1764 1765
				DCACHE_OP_COMPARE	|
				DCACHE_OP_REVALIDATE	|
1766
				DCACHE_OP_WEAK_REVALIDATE	|
1767
				DCACHE_OP_DELETE	|
M
Miklos Szeredi 已提交
1768
				DCACHE_OP_REAL));
1769 1770 1771 1772 1773 1774 1775 1776 1777
	dentry->d_op = op;
	if (!op)
		return;
	if (op->d_hash)
		dentry->d_flags |= DCACHE_OP_HASH;
	if (op->d_compare)
		dentry->d_flags |= DCACHE_OP_COMPARE;
	if (op->d_revalidate)
		dentry->d_flags |= DCACHE_OP_REVALIDATE;
1778 1779
	if (op->d_weak_revalidate)
		dentry->d_flags |= DCACHE_OP_WEAK_REVALIDATE;
1780 1781
	if (op->d_delete)
		dentry->d_flags |= DCACHE_OP_DELETE;
S
Sage Weil 已提交
1782 1783
	if (op->d_prune)
		dentry->d_flags |= DCACHE_OP_PRUNE;
M
Miklos Szeredi 已提交
1784 1785
	if (op->d_real)
		dentry->d_flags |= DCACHE_OP_REAL;
1786 1787 1788 1789

}
EXPORT_SYMBOL(d_set_d_op);

1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805

/*
 * d_set_fallthru - Mark a dentry as falling through to a lower layer
 * @dentry - The dentry to mark
 *
 * Mark a dentry as falling through to the lower layer (as set with
 * d_pin_lower()).  This flag may be recorded on the medium.
 */
void d_set_fallthru(struct dentry *dentry)
{
	spin_lock(&dentry->d_lock);
	dentry->d_flags |= DCACHE_FALLTHRU;
	spin_unlock(&dentry->d_lock);
}
EXPORT_SYMBOL(d_set_fallthru);

1806 1807
static unsigned d_flags_for_inode(struct inode *inode)
{
1808
	unsigned add_flags = DCACHE_REGULAR_TYPE;
1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820

	if (!inode)
		return DCACHE_MISS_TYPE;

	if (S_ISDIR(inode->i_mode)) {
		add_flags = DCACHE_DIRECTORY_TYPE;
		if (unlikely(!(inode->i_opflags & IOP_LOOKUP))) {
			if (unlikely(!inode->i_op->lookup))
				add_flags = DCACHE_AUTODIR_TYPE;
			else
				inode->i_opflags |= IOP_LOOKUP;
		}
1821 1822 1823 1824
		goto type_determined;
	}

	if (unlikely(!(inode->i_opflags & IOP_NOFOLLOW))) {
1825
		if (unlikely(inode->i_op->get_link)) {
1826
			add_flags = DCACHE_SYMLINK_TYPE;
1827 1828 1829
			goto type_determined;
		}
		inode->i_opflags |= IOP_NOFOLLOW;
1830 1831
	}

1832 1833 1834 1835
	if (unlikely(!S_ISREG(inode->i_mode)))
		add_flags = DCACHE_SPECIAL_TYPE;

type_determined:
1836 1837 1838 1839 1840
	if (unlikely(IS_AUTOMOUNT(inode)))
		add_flags |= DCACHE_NEED_AUTOMOUNT;
	return add_flags;
}

1841 1842
static void __d_instantiate(struct dentry *dentry, struct inode *inode)
{
1843
	unsigned add_flags = d_flags_for_inode(inode);
1844
	WARN_ON(d_in_lookup(dentry));
1845

N
Nick Piggin 已提交
1846
	spin_lock(&dentry->d_lock);
1847
	hlist_add_head(&dentry->d_u.d_alias, &inode->i_dentry);
1848
	raw_write_seqcount_begin(&dentry->d_seq);
1849
	__d_set_inode_and_type(dentry, inode, add_flags);
1850
	raw_write_seqcount_end(&dentry->d_seq);
A
Al Viro 已提交
1851
	fsnotify_update_flags(dentry);
N
Nick Piggin 已提交
1852
	spin_unlock(&dentry->d_lock);
1853 1854
}

L
Linus Torvalds 已提交
1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871
/**
 * d_instantiate - fill in inode information for a dentry
 * @entry: dentry to complete
 * @inode: inode to attach to this dentry
 *
 * Fill in inode information in the entry.
 *
 * This turns negative dentries into productive full members
 * of society.
 *
 * NOTE! This assumes that the inode count has been incremented
 * (or otherwise set) by the caller to indicate that it is now
 * in use by the dcache.
 */
 
void d_instantiate(struct dentry *entry, struct inode * inode)
{
1872
	BUG_ON(!hlist_unhashed(&entry->d_u.d_alias));
1873
	if (inode) {
1874
		security_d_instantiate(entry, inode);
1875
		spin_lock(&inode->i_lock);
1876
		__d_instantiate(entry, inode);
1877
		spin_unlock(&inode->i_lock);
1878
	}
L
Linus Torvalds 已提交
1879
}
1880
EXPORT_SYMBOL(d_instantiate);
L
Linus Torvalds 已提交
1881

1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892
/**
 * d_instantiate_no_diralias - instantiate a non-aliased dentry
 * @entry: dentry to complete
 * @inode: inode to attach to this dentry
 *
 * Fill in inode information in the entry.  If a directory alias is found, then
 * return an error (and drop inode).  Together with d_materialise_unique() this
 * guarantees that a directory inode may never have more than one alias.
 */
int d_instantiate_no_diralias(struct dentry *entry, struct inode *inode)
{
1893
	BUG_ON(!hlist_unhashed(&entry->d_u.d_alias));
1894

1895
	security_d_instantiate(entry, inode);
1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908
	spin_lock(&inode->i_lock);
	if (S_ISDIR(inode->i_mode) && !hlist_empty(&inode->i_dentry)) {
		spin_unlock(&inode->i_lock);
		iput(inode);
		return -EBUSY;
	}
	__d_instantiate(entry, inode);
	spin_unlock(&inode->i_lock);

	return 0;
}
EXPORT_SYMBOL(d_instantiate_no_diralias);

A
Al Viro 已提交
1909 1910 1911 1912 1913
struct dentry *d_make_root(struct inode *root_inode)
{
	struct dentry *res = NULL;

	if (root_inode) {
1914
		res = d_alloc_anon(root_inode->i_sb);
A
Al Viro 已提交
1915 1916 1917 1918 1919 1920 1921 1922 1923
		if (res)
			d_instantiate(res, root_inode);
		else
			iput(root_inode);
	}
	return res;
}
EXPORT_SYMBOL(d_make_root);

1924 1925 1926 1927
static struct dentry * __d_find_any_alias(struct inode *inode)
{
	struct dentry *alias;

A
Al Viro 已提交
1928
	if (hlist_empty(&inode->i_dentry))
1929
		return NULL;
1930
	alias = hlist_entry(inode->i_dentry.first, struct dentry, d_u.d_alias);
1931 1932 1933 1934
	__dget(alias);
	return alias;
}

1935 1936 1937 1938 1939 1940 1941 1942
/**
 * d_find_any_alias - find any alias for a given inode
 * @inode: inode to find an alias for
 *
 * If any aliases exist for the given inode, take and return a
 * reference for one of them.  If no aliases exist, return %NULL.
 */
struct dentry *d_find_any_alias(struct inode *inode)
1943 1944 1945 1946 1947 1948 1949 1950
{
	struct dentry *de;

	spin_lock(&inode->i_lock);
	de = __d_find_any_alias(inode);
	spin_unlock(&inode->i_lock);
	return de;
}
1951
EXPORT_SYMBOL(d_find_any_alias);
1952

1953 1954 1955
static struct dentry *__d_instantiate_anon(struct dentry *dentry,
					   struct inode *inode,
					   bool disconnected)
1956
{
C
Christoph Hellwig 已提交
1957
	struct dentry *res;
1958
	unsigned add_flags;
1959

1960
	security_d_instantiate(dentry, inode);
1961
	spin_lock(&inode->i_lock);
1962
	res = __d_find_any_alias(inode);
C
Christoph Hellwig 已提交
1963
	if (res) {
1964
		spin_unlock(&inode->i_lock);
1965
		dput(dentry);
C
Christoph Hellwig 已提交
1966 1967 1968 1969
		goto out_iput;
	}

	/* attach a disconnected dentry */
1970 1971 1972 1973
	add_flags = d_flags_for_inode(inode);

	if (disconnected)
		add_flags |= DCACHE_DISCONNECTED;
1974

1975 1976 1977
	spin_lock(&dentry->d_lock);
	__d_set_inode_and_type(dentry, inode, add_flags);
	hlist_add_head(&dentry->d_u.d_alias, &inode->i_dentry);
1978
	if (!disconnected) {
1979 1980 1981
		hlist_bl_lock(&dentry->d_sb->s_roots);
		hlist_bl_add_head(&dentry->d_hash, &dentry->d_sb->s_roots);
		hlist_bl_unlock(&dentry->d_sb->s_roots);
1982
	}
1983
	spin_unlock(&dentry->d_lock);
1984
	spin_unlock(&inode->i_lock);
C
Christoph Hellwig 已提交
1985

1986
	return dentry;
C
Christoph Hellwig 已提交
1987 1988 1989 1990

 out_iput:
	iput(inode);
	return res;
1991
}
1992

1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025
struct dentry *d_instantiate_anon(struct dentry *dentry, struct inode *inode)
{
	return __d_instantiate_anon(dentry, inode, true);
}
EXPORT_SYMBOL(d_instantiate_anon);

static struct dentry *__d_obtain_alias(struct inode *inode, bool disconnected)
{
	struct dentry *tmp;
	struct dentry *res;

	if (!inode)
		return ERR_PTR(-ESTALE);
	if (IS_ERR(inode))
		return ERR_CAST(inode);

	res = d_find_any_alias(inode);
	if (res)
		goto out_iput;

	tmp = d_alloc_anon(inode->i_sb);
	if (!tmp) {
		res = ERR_PTR(-ENOMEM);
		goto out_iput;
	}

	return __d_instantiate_anon(tmp, inode, disconnected);

out_iput:
	iput(inode);
	return res;
}

2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045
/**
 * d_obtain_alias - find or allocate a DISCONNECTED dentry for a given inode
 * @inode: inode to allocate the dentry for
 *
 * Obtain a dentry for an inode resulting from NFS filehandle conversion or
 * similar open by handle operations.  The returned dentry may be anonymous,
 * or may have a full name (if the inode was already in the cache).
 *
 * When called on a directory inode, we must ensure that the inode only ever
 * has one dentry.  If a dentry is found, that is returned instead of
 * allocating a new one.
 *
 * On successful return, the reference to the inode has been transferred
 * to the dentry.  In case of an error the reference on the inode is released.
 * To make it easier to use in export operations a %NULL or IS_ERR inode may
 * be passed in and the error will be propagated to the return value,
 * with a %NULL @inode replaced by ERR_PTR(-ESTALE).
 */
struct dentry *d_obtain_alias(struct inode *inode)
{
2046
	return __d_obtain_alias(inode, true);
2047
}
2048
EXPORT_SYMBOL(d_obtain_alias);
L
Linus Torvalds 已提交
2049

2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066
/**
 * d_obtain_root - find or allocate a dentry for a given inode
 * @inode: inode to allocate the dentry for
 *
 * Obtain an IS_ROOT dentry for the root of a filesystem.
 *
 * We must ensure that directory inodes only ever have one dentry.  If a
 * dentry is found, that is returned instead of allocating a new one.
 *
 * On successful return, the reference to the inode has been transferred
 * to the dentry.  In case of an error the reference on the inode is
 * released.  A %NULL or IS_ERR inode may be passed in and will be the
 * error will be propagate to the return value, with a %NULL @inode
 * replaced by ERR_PTR(-ESTALE).
 */
struct dentry *d_obtain_root(struct inode *inode)
{
2067
	return __d_obtain_alias(inode, false);
2068 2069 2070
}
EXPORT_SYMBOL(d_obtain_root);

2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086
/**
 * d_add_ci - lookup or allocate new dentry with case-exact name
 * @inode:  the inode case-insensitive lookup has found
 * @dentry: the negative dentry that was passed to the parent's lookup func
 * @name:   the case-exact name to be associated with the returned dentry
 *
 * This is to avoid filling the dcache with case-insensitive names to the
 * same inode, only the actual correct case is stored in the dcache for
 * case-insensitive filesystems.
 *
 * For a case-insensitive lookup match and if the the case-exact dentry
 * already exists in in the dcache, use it and return it.
 *
 * If no entry exists with the exact case name, allocate new dentry with
 * the exact case, and return the spliced entry.
 */
2087
struct dentry *d_add_ci(struct dentry *dentry, struct inode *inode,
2088 2089
			struct qstr *name)
{
2090
	struct dentry *found, *res;
2091

C
Christoph Hellwig 已提交
2092 2093 2094 2095
	/*
	 * First check if a dentry matching the name already exists,
	 * if not go ahead and create it now.
	 */
2096
	found = d_hash_and_lookup(dentry->d_parent, name);
2097 2098 2099 2100 2101 2102 2103 2104 2105 2106
	if (found) {
		iput(inode);
		return found;
	}
	if (d_in_lookup(dentry)) {
		found = d_alloc_parallel(dentry->d_parent, name,
					dentry->d_wait);
		if (IS_ERR(found) || !d_in_lookup(found)) {
			iput(inode);
			return found;
2107
		}
2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118
	} else {
		found = d_alloc(dentry->d_parent, name);
		if (!found) {
			iput(inode);
			return ERR_PTR(-ENOMEM);
		} 
	}
	res = d_splice_alias(inode, found);
	if (res) {
		dput(found);
		return res;
2119
	}
2120
	return found;
2121
}
2122
EXPORT_SYMBOL(d_add_ci);
L
Linus Torvalds 已提交
2123

2124

A
Al Viro 已提交
2125 2126 2127
static inline bool d_same_name(const struct dentry *dentry,
				const struct dentry *parent,
				const struct qstr *name)
2128
{
A
Al Viro 已提交
2129 2130 2131 2132
	if (likely(!(parent->d_flags & DCACHE_OP_COMPARE))) {
		if (dentry->d_name.len != name->len)
			return false;
		return dentry_cmp(dentry, name->name, name->len) == 0;
2133
	}
2134
	return parent->d_op->d_compare(dentry,
A
Al Viro 已提交
2135 2136
				       dentry->d_name.len, dentry->d_name.name,
				       name) == 0;
2137 2138
}

N
Nick Piggin 已提交
2139 2140 2141 2142
/**
 * __d_lookup_rcu - search for a dentry (racy, store-free)
 * @parent: parent dentry
 * @name: qstr of name we wish to find
2143
 * @seqp: returns d_seq value at the point where the dentry was found
N
Nick Piggin 已提交
2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156
 * Returns: dentry, or NULL
 *
 * __d_lookup_rcu is the dcache lookup function for rcu-walk name
 * resolution (store-free path walking) design described in
 * Documentation/filesystems/path-lookup.txt.
 *
 * This is not to be used outside core vfs.
 *
 * __d_lookup_rcu must only be used in rcu-walk mode, ie. with vfsmount lock
 * held, and rcu_read_lock held. The returned dentry must not be stored into
 * without taking d_lock and checking d_seq sequence count against @seq
 * returned here.
 *
2157
 * A refcount may be taken on the found dentry with the d_rcu_to_refcount
N
Nick Piggin 已提交
2158 2159 2160 2161 2162 2163
 * function.
 *
 * Alternatively, __d_lookup_rcu may be called again to look up the child of
 * the returned dentry, so long as its parent's seqlock is checked after the
 * child is looked up. Thus, an interlocking stepping of sequence lock checks
 * is formed, giving integrity down the path walk.
2164 2165 2166
 *
 * NOTE! The caller *has* to check the resulting dentry against the sequence
 * number we've returned before using any of the resulting dentry state!
N
Nick Piggin 已提交
2167
 */
2168 2169
struct dentry *__d_lookup_rcu(const struct dentry *parent,
				const struct qstr *name,
2170
				unsigned *seqp)
N
Nick Piggin 已提交
2171
{
2172
	u64 hashlen = name->hash_len;
N
Nick Piggin 已提交
2173
	const unsigned char *str = name->name;
2174
	struct hlist_bl_head *b = d_hash(hashlen_hash(hashlen));
2175
	struct hlist_bl_node *node;
N
Nick Piggin 已提交
2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195
	struct dentry *dentry;

	/*
	 * Note: There is significant duplication with __d_lookup_rcu which is
	 * required to prevent single threaded performance regressions
	 * especially on architectures where smp_rmb (in seqcounts) are costly.
	 * Keep the two functions in sync.
	 */

	/*
	 * The hash list is protected using RCU.
	 *
	 * Carefully use d_seq when comparing a candidate dentry, to avoid
	 * races with d_move().
	 *
	 * It is possible that concurrent renames can mess up our list
	 * walk here and result in missing our dentry, resulting in the
	 * false-negative result. d_lookup() protects against concurrent
	 * renames using rename_lock seqlock.
	 *
2196
	 * See Documentation/filesystems/path-lookup.txt for more details.
N
Nick Piggin 已提交
2197
	 */
2198
	hlist_bl_for_each_entry_rcu(dentry, node, b, d_hash) {
2199
		unsigned seq;
N
Nick Piggin 已提交
2200 2201

seqretry:
2202 2203
		/*
		 * The dentry sequence count protects us from concurrent
2204
		 * renames, and thus protects parent and name fields.
2205 2206
		 *
		 * The caller must perform a seqcount check in order
2207
		 * to do anything useful with the returned dentry.
2208 2209 2210 2211 2212 2213 2214
		 *
		 * NOTE! We do a "raw" seqcount_begin here. That means that
		 * we don't wait for the sequence count to stabilize if it
		 * is in the middle of a sequence change. If we do the slow
		 * dentry compare, we will do seqretries until it is stable,
		 * and if we end up with a successful lookup, we actually
		 * want to exit RCU lookup anyway.
A
Al Viro 已提交
2215 2216 2217
		 *
		 * Note that raw_seqcount_begin still *does* smp_rmb(), so
		 * we are still guaranteed NUL-termination of ->d_name.name.
2218 2219
		 */
		seq = raw_seqcount_begin(&dentry->d_seq);
N
Nick Piggin 已提交
2220 2221
		if (dentry->d_parent != parent)
			continue;
2222 2223
		if (d_unhashed(dentry))
			continue;
2224

2225
		if (unlikely(parent->d_flags & DCACHE_OP_COMPARE)) {
A
Al Viro 已提交
2226 2227
			int tlen;
			const char *tname;
2228 2229
			if (dentry->d_name.hash != hashlen_hash(hashlen))
				continue;
A
Al Viro 已提交
2230 2231 2232 2233 2234
			tlen = dentry->d_name.len;
			tname = dentry->d_name.name;
			/* we want a consistent (name,len) pair */
			if (read_seqcount_retry(&dentry->d_seq, seq)) {
				cpu_relax();
2235 2236
				goto seqretry;
			}
2237
			if (parent->d_op->d_compare(dentry,
A
Al Viro 已提交
2238 2239 2240 2241 2242 2243 2244
						    tlen, tname, name) != 0)
				continue;
		} else {
			if (dentry->d_name.hash_len != hashlen)
				continue;
			if (dentry_cmp(dentry, str, hashlen_len(hashlen)) != 0)
				continue;
N
Nick Piggin 已提交
2245
		}
2246
		*seqp = seq;
A
Al Viro 已提交
2247
		return dentry;
N
Nick Piggin 已提交
2248 2249 2250 2251
	}
	return NULL;
}

L
Linus Torvalds 已提交
2252 2253 2254 2255
/**
 * d_lookup - search for a dentry
 * @parent: parent dentry
 * @name: qstr of name we wish to find
2256
 * Returns: dentry, or NULL
L
Linus Torvalds 已提交
2257
 *
2258 2259 2260 2261
 * d_lookup searches the children of the parent dentry for the name in
 * question. If the dentry is found its reference count is incremented and the
 * dentry is returned. The caller must use dput to free the entry when it has
 * finished using it. %NULL is returned if the dentry does not exist.
L
Linus Torvalds 已提交
2262
 */
A
Al Viro 已提交
2263
struct dentry *d_lookup(const struct dentry *parent, const struct qstr *name)
L
Linus Torvalds 已提交
2264
{
N
Nick Piggin 已提交
2265
	struct dentry *dentry;
2266
	unsigned seq;
L
Linus Torvalds 已提交
2267

2268 2269 2270 2271
	do {
		seq = read_seqbegin(&rename_lock);
		dentry = __d_lookup(parent, name);
		if (dentry)
L
Linus Torvalds 已提交
2272 2273 2274 2275
			break;
	} while (read_seqretry(&rename_lock, seq));
	return dentry;
}
2276
EXPORT_SYMBOL(d_lookup);
L
Linus Torvalds 已提交
2277

N
Nick Piggin 已提交
2278
/**
2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292
 * __d_lookup - search for a dentry (racy)
 * @parent: parent dentry
 * @name: qstr of name we wish to find
 * Returns: dentry, or NULL
 *
 * __d_lookup is like d_lookup, however it may (rarely) return a
 * false-negative result due to unrelated rename activity.
 *
 * __d_lookup is slightly faster by avoiding rename_lock read seqlock,
 * however it must be used carefully, eg. with a following d_lookup in
 * the case of failure.
 *
 * __d_lookup callers must be commented.
 */
A
Al Viro 已提交
2293
struct dentry *__d_lookup(const struct dentry *parent, const struct qstr *name)
L
Linus Torvalds 已提交
2294 2295
{
	unsigned int hash = name->hash;
2296
	struct hlist_bl_head *b = d_hash(hash);
2297
	struct hlist_bl_node *node;
N
Nick Piggin 已提交
2298
	struct dentry *found = NULL;
2299
	struct dentry *dentry;
L
Linus Torvalds 已提交
2300

N
Nick Piggin 已提交
2301 2302 2303 2304 2305 2306 2307
	/*
	 * Note: There is significant duplication with __d_lookup_rcu which is
	 * required to prevent single threaded performance regressions
	 * especially on architectures where smp_rmb (in seqcounts) are costly.
	 * Keep the two functions in sync.
	 */

2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318
	/*
	 * The hash list is protected using RCU.
	 *
	 * Take d_lock when comparing a candidate dentry, to avoid races
	 * with d_move().
	 *
	 * It is possible that concurrent renames can mess up our list
	 * walk here and result in missing our dentry, resulting in the
	 * false-negative result. d_lookup() protects against concurrent
	 * renames using rename_lock seqlock.
	 *
2319
	 * See Documentation/filesystems/path-lookup.txt for more details.
2320
	 */
L
Linus Torvalds 已提交
2321 2322
	rcu_read_lock();
	
2323
	hlist_bl_for_each_entry_rcu(dentry, node, b, d_hash) {
L
Linus Torvalds 已提交
2324 2325 2326 2327 2328 2329 2330

		if (dentry->d_name.hash != hash)
			continue;

		spin_lock(&dentry->d_lock);
		if (dentry->d_parent != parent)
			goto next;
2331 2332 2333
		if (d_unhashed(dentry))
			goto next;

A
Al Viro 已提交
2334 2335
		if (!d_same_name(dentry, parent, name))
			goto next;
L
Linus Torvalds 已提交
2336

2337
		dentry->d_lockref.count++;
2338
		found = dentry;
L
Linus Torvalds 已提交
2339 2340 2341 2342 2343 2344 2345 2346 2347 2348
		spin_unlock(&dentry->d_lock);
		break;
next:
		spin_unlock(&dentry->d_lock);
 	}
 	rcu_read_unlock();

 	return found;
}

2349 2350 2351 2352 2353
/**
 * d_hash_and_lookup - hash the qstr then search for a dentry
 * @dir: Directory to search in
 * @name: qstr of name we wish to find
 *
2354
 * On lookup failure NULL is returned; on bad name - ERR_PTR(-error)
2355 2356 2357 2358 2359 2360 2361 2362
 */
struct dentry *d_hash_and_lookup(struct dentry *dir, struct qstr *name)
{
	/*
	 * Check for a fs-specific hash function. Note that we must
	 * calculate the standard hash first, as the d_op->d_hash()
	 * routine may choose to leave the hash value unchanged.
	 */
2363
	name->hash = full_name_hash(dir, name->name, name->len);
2364
	if (dir->d_flags & DCACHE_OP_HASH) {
2365
		int err = dir->d_op->d_hash(dir, name);
2366 2367
		if (unlikely(err < 0))
			return ERR_PTR(err);
2368
	}
2369
	return d_lookup(dir, name);
2370
}
2371
EXPORT_SYMBOL(d_hash_and_lookup);
2372

L
Linus Torvalds 已提交
2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395
/*
 * When a file is deleted, we have two options:
 * - turn this dentry into a negative dentry
 * - unhash this dentry and free it.
 *
 * Usually, we want to just turn this into
 * a negative dentry, but if anybody else is
 * currently using the dentry or the inode
 * we can't do that and we fall back on removing
 * it from the hash queues and waiting for
 * it to be deleted later when it has no users
 */
 
/**
 * d_delete - delete a dentry
 * @dentry: The dentry to delete
 *
 * Turn the dentry into a negative dentry if possible, otherwise
 * remove it from the hash queues so it can be deleted later
 */
 
void d_delete(struct dentry * dentry)
{
A
Al Viro 已提交
2396 2397 2398 2399 2400
	struct inode *inode = dentry->d_inode;
	int isdir = d_is_dir(dentry);

	spin_lock(&inode->i_lock);
	spin_lock(&dentry->d_lock);
L
Linus Torvalds 已提交
2401 2402 2403
	/*
	 * Are we the only user?
	 */
2404
	if (dentry->d_lockref.count == 1) {
A
Al Viro 已提交
2405
		dentry->d_flags &= ~DCACHE_CANT_MOUNT;
N
Nick Piggin 已提交
2406
		dentry_unlink_inode(dentry);
A
Al Viro 已提交
2407
	} else {
L
Linus Torvalds 已提交
2408
		__d_drop(dentry);
A
Al Viro 已提交
2409 2410 2411
		spin_unlock(&dentry->d_lock);
		spin_unlock(&inode->i_lock);
	}
2412
	fsnotify_nameremove(dentry, isdir);
L
Linus Torvalds 已提交
2413
}
2414
EXPORT_SYMBOL(d_delete);
L
Linus Torvalds 已提交
2415

2416
static void __d_rehash(struct dentry *entry)
L
Linus Torvalds 已提交
2417
{
2418
	struct hlist_bl_head *b = d_hash(entry->d_name.hash);
2419

2420
	hlist_bl_lock(b);
2421
	hlist_bl_add_head_rcu(&entry->d_hash, b);
2422
	hlist_bl_unlock(b);
L
Linus Torvalds 已提交
2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434
}

/**
 * d_rehash	- add an entry back to the hash
 * @entry: dentry to add to the hash
 *
 * Adds a dentry to the hash according to its name.
 */
 
void d_rehash(struct dentry * entry)
{
	spin_lock(&entry->d_lock);
2435
	__d_rehash(entry);
L
Linus Torvalds 已提交
2436 2437
	spin_unlock(&entry->d_lock);
}
2438
EXPORT_SYMBOL(d_rehash);
L
Linus Torvalds 已提交
2439

A
Al Viro 已提交
2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455
static inline unsigned start_dir_add(struct inode *dir)
{

	for (;;) {
		unsigned n = dir->i_dir_seq;
		if (!(n & 1) && cmpxchg(&dir->i_dir_seq, n, n + 1) == n)
			return n;
		cpu_relax();
	}
}

static inline void end_dir_add(struct inode *dir, unsigned n)
{
	smp_store_release(&dir->i_dir_seq, n + 2);
}

2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469
static void d_wait_lookup(struct dentry *dentry)
{
	if (d_in_lookup(dentry)) {
		DECLARE_WAITQUEUE(wait, current);
		add_wait_queue(dentry->d_wait, &wait);
		do {
			set_current_state(TASK_UNINTERRUPTIBLE);
			spin_unlock(&dentry->d_lock);
			schedule();
			spin_lock(&dentry->d_lock);
		} while (d_in_lookup(dentry));
	}
}

A
Al Viro 已提交
2470
struct dentry *d_alloc_parallel(struct dentry *parent,
2471 2472
				const struct qstr *name,
				wait_queue_head_t *wq)
A
Al Viro 已提交
2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485
{
	unsigned int hash = name->hash;
	struct hlist_bl_head *b = in_lookup_hash(parent, hash);
	struct hlist_bl_node *node;
	struct dentry *new = d_alloc(parent, name);
	struct dentry *dentry;
	unsigned seq, r_seq, d_seq;

	if (unlikely(!new))
		return ERR_PTR(-ENOMEM);

retry:
	rcu_read_lock();
2486
	seq = smp_load_acquire(&parent->d_inode->i_dir_seq);
A
Al Viro 已提交
2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506
	r_seq = read_seqbegin(&rename_lock);
	dentry = __d_lookup_rcu(parent, name, &d_seq);
	if (unlikely(dentry)) {
		if (!lockref_get_not_dead(&dentry->d_lockref)) {
			rcu_read_unlock();
			goto retry;
		}
		if (read_seqcount_retry(&dentry->d_seq, d_seq)) {
			rcu_read_unlock();
			dput(dentry);
			goto retry;
		}
		rcu_read_unlock();
		dput(new);
		return dentry;
	}
	if (unlikely(read_seqretry(&rename_lock, r_seq))) {
		rcu_read_unlock();
		goto retry;
	}
2507 2508 2509 2510 2511 2512

	if (unlikely(seq & 1)) {
		rcu_read_unlock();
		goto retry;
	}

A
Al Viro 已提交
2513
	hlist_bl_lock(b);
2514
	if (unlikely(READ_ONCE(parent->d_inode->i_dir_seq) != seq)) {
A
Al Viro 已提交
2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530
		hlist_bl_unlock(b);
		rcu_read_unlock();
		goto retry;
	}
	/*
	 * No changes for the parent since the beginning of d_lookup().
	 * Since all removals from the chain happen with hlist_bl_lock(),
	 * any potential in-lookup matches are going to stay here until
	 * we unlock the chain.  All fields are stable in everything
	 * we encounter.
	 */
	hlist_bl_for_each_entry(dentry, node, b, d_u.d_in_lookup_hash) {
		if (dentry->d_name.hash != hash)
			continue;
		if (dentry->d_parent != parent)
			continue;
A
Al Viro 已提交
2531 2532
		if (!d_same_name(dentry, parent, name))
			continue;
A
Al Viro 已提交
2533
		hlist_bl_unlock(b);
2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544
		/* now we can try to grab a reference */
		if (!lockref_get_not_dead(&dentry->d_lockref)) {
			rcu_read_unlock();
			goto retry;
		}

		rcu_read_unlock();
		/*
		 * somebody is likely to be still doing lookup for it;
		 * wait for them to finish
		 */
2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558
		spin_lock(&dentry->d_lock);
		d_wait_lookup(dentry);
		/*
		 * it's not in-lookup anymore; in principle we should repeat
		 * everything from dcache lookup, but it's likely to be what
		 * d_lookup() would've found anyway.  If it is, just return it;
		 * otherwise we really have to repeat the whole thing.
		 */
		if (unlikely(dentry->d_name.hash != hash))
			goto mismatch;
		if (unlikely(dentry->d_parent != parent))
			goto mismatch;
		if (unlikely(d_unhashed(dentry)))
			goto mismatch;
A
Al Viro 已提交
2559 2560
		if (unlikely(!d_same_name(dentry, parent, name)))
			goto mismatch;
2561 2562
		/* OK, it *is* a hashed match; return it */
		spin_unlock(&dentry->d_lock);
A
Al Viro 已提交
2563 2564 2565
		dput(new);
		return dentry;
	}
2566
	rcu_read_unlock();
A
Al Viro 已提交
2567 2568
	/* we can't take ->d_lock here; it's OK, though. */
	new->d_flags |= DCACHE_PAR_LOOKUP;
2569
	new->d_wait = wq;
A
Al Viro 已提交
2570 2571 2572
	hlist_bl_add_head_rcu(&new->d_u.d_in_lookup_hash, b);
	hlist_bl_unlock(b);
	return new;
2573 2574 2575 2576
mismatch:
	spin_unlock(&dentry->d_lock);
	dput(dentry);
	goto retry;
A
Al Viro 已提交
2577 2578 2579
}
EXPORT_SYMBOL(d_alloc_parallel);

2580 2581
void __d_lookup_done(struct dentry *dentry)
{
A
Al Viro 已提交
2582 2583 2584
	struct hlist_bl_head *b = in_lookup_hash(dentry->d_parent,
						 dentry->d_name.hash);
	hlist_bl_lock(b);
2585
	dentry->d_flags &= ~DCACHE_PAR_LOOKUP;
A
Al Viro 已提交
2586
	__hlist_bl_del(&dentry->d_u.d_in_lookup_hash);
2587 2588
	wake_up_all(dentry->d_wait);
	dentry->d_wait = NULL;
A
Al Viro 已提交
2589 2590
	hlist_bl_unlock(b);
	INIT_HLIST_NODE(&dentry->d_u.d_alias);
2591
	INIT_LIST_HEAD(&dentry->d_lru);
2592 2593
}
EXPORT_SYMBOL(__d_lookup_done);
A
Al Viro 已提交
2594 2595 2596 2597 2598

/* inode->i_lock held if inode is non-NULL */

static inline void __d_add(struct dentry *dentry, struct inode *inode)
{
A
Al Viro 已提交
2599 2600
	struct inode *dir = NULL;
	unsigned n;
A
Al Viro 已提交
2601
	spin_lock(&dentry->d_lock);
A
Al Viro 已提交
2602 2603 2604
	if (unlikely(d_in_lookup(dentry))) {
		dir = dentry->d_parent->d_inode;
		n = start_dir_add(dir);
2605
		__d_lookup_done(dentry);
A
Al Viro 已提交
2606
	}
A
Al Viro 已提交
2607
	if (inode) {
A
Al Viro 已提交
2608 2609 2610 2611 2612
		unsigned add_flags = d_flags_for_inode(inode);
		hlist_add_head(&dentry->d_u.d_alias, &inode->i_dentry);
		raw_write_seqcount_begin(&dentry->d_seq);
		__d_set_inode_and_type(dentry, inode, add_flags);
		raw_write_seqcount_end(&dentry->d_seq);
A
Al Viro 已提交
2613
		fsnotify_update_flags(dentry);
A
Al Viro 已提交
2614
	}
2615
	__d_rehash(dentry);
A
Al Viro 已提交
2616 2617
	if (dir)
		end_dir_add(dir, n);
A
Al Viro 已提交
2618 2619 2620
	spin_unlock(&dentry->d_lock);
	if (inode)
		spin_unlock(&inode->i_lock);
A
Al Viro 已提交
2621 2622
}

A
Al Viro 已提交
2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633
/**
 * d_add - add dentry to hash queues
 * @entry: dentry to add
 * @inode: The inode to attach to this dentry
 *
 * This adds the entry to the hash queues and initializes @inode.
 * The entry was actually filled in earlier during d_alloc().
 */

void d_add(struct dentry *entry, struct inode *inode)
{
2634 2635
	if (inode) {
		security_d_instantiate(entry, inode);
A
Al Viro 已提交
2636
		spin_lock(&inode->i_lock);
2637
	}
A
Al Viro 已提交
2638
	__d_add(entry, inode);
A
Al Viro 已提交
2639 2640 2641
}
EXPORT_SYMBOL(d_add);

2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668
/**
 * d_exact_alias - find and hash an exact unhashed alias
 * @entry: dentry to add
 * @inode: The inode to go with this dentry
 *
 * If an unhashed dentry with the same name/parent and desired
 * inode already exists, hash and return it.  Otherwise, return
 * NULL.
 *
 * Parent directory should be locked.
 */
struct dentry *d_exact_alias(struct dentry *entry, struct inode *inode)
{
	struct dentry *alias;
	unsigned int hash = entry->d_name.hash;

	spin_lock(&inode->i_lock);
	hlist_for_each_entry(alias, &inode->i_dentry, d_u.d_alias) {
		/*
		 * Don't need alias->d_lock here, because aliases with
		 * d_parent == entry->d_parent are not subject to name or
		 * parent changes, because the parent inode i_mutex is held.
		 */
		if (alias->d_name.hash != hash)
			continue;
		if (alias->d_parent != entry->d_parent)
			continue;
A
Al Viro 已提交
2669
		if (!d_same_name(alias, entry->d_parent, &entry->d_name))
2670 2671 2672 2673 2674 2675 2676
			continue;
		spin_lock(&alias->d_lock);
		if (!d_unhashed(alias)) {
			spin_unlock(&alias->d_lock);
			alias = NULL;
		} else {
			__dget_dlock(alias);
2677
			__d_rehash(alias);
2678 2679 2680 2681 2682 2683 2684 2685 2686 2687
			spin_unlock(&alias->d_lock);
		}
		spin_unlock(&inode->i_lock);
		return alias;
	}
	spin_unlock(&inode->i_lock);
	return NULL;
}
EXPORT_SYMBOL(d_exact_alias);

N
Nick Piggin 已提交
2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701
/**
 * dentry_update_name_case - update case insensitive dentry with a new name
 * @dentry: dentry to be updated
 * @name: new name
 *
 * Update a case insensitive dentry with new case of name.
 *
 * dentry must have been returned by d_lookup with name @name. Old and new
 * name lengths must match (ie. no d_compare which allows mismatched name
 * lengths).
 *
 * Parent inode i_mutex must be held over d_lookup and into this call (to
 * keep renames and concurrent inserts, and readdir(2) away).
 */
A
Al Viro 已提交
2702
void dentry_update_name_case(struct dentry *dentry, const struct qstr *name)
N
Nick Piggin 已提交
2703
{
A
Al Viro 已提交
2704
	BUG_ON(!inode_is_locked(dentry->d_parent->d_inode));
N
Nick Piggin 已提交
2705 2706 2707
	BUG_ON(dentry->d_name.len != name->len); /* d_lookup gives this */

	spin_lock(&dentry->d_lock);
N
Nick Piggin 已提交
2708
	write_seqcount_begin(&dentry->d_seq);
N
Nick Piggin 已提交
2709
	memcpy((unsigned char *)dentry->d_name.name, name->name, name->len);
N
Nick Piggin 已提交
2710
	write_seqcount_end(&dentry->d_seq);
N
Nick Piggin 已提交
2711 2712 2713 2714
	spin_unlock(&dentry->d_lock);
}
EXPORT_SYMBOL(dentry_update_name_case);

2715
static void swap_names(struct dentry *dentry, struct dentry *target)
L
Linus Torvalds 已提交
2716
{
2717 2718
	if (unlikely(dname_external(target))) {
		if (unlikely(dname_external(dentry))) {
L
Linus Torvalds 已提交
2719 2720 2721
			/*
			 * Both external: swap the pointers
			 */
2722
			swap(target->d_name.name, dentry->d_name.name);
L
Linus Torvalds 已提交
2723 2724 2725 2726 2727
		} else {
			/*
			 * dentry:internal, target:external.  Steal target's
			 * storage and make target internal.
			 */
2728 2729
			memcpy(target->d_iname, dentry->d_name.name,
					dentry->d_name.len + 1);
L
Linus Torvalds 已提交
2730 2731 2732 2733
			dentry->d_name.name = target->d_name.name;
			target->d_name.name = target->d_iname;
		}
	} else {
2734
		if (unlikely(dname_external(dentry))) {
L
Linus Torvalds 已提交
2735 2736 2737 2738 2739 2740 2741 2742 2743 2744
			/*
			 * dentry:external, target:internal.  Give dentry's
			 * storage to target and make dentry internal
			 */
			memcpy(dentry->d_iname, target->d_name.name,
					target->d_name.len + 1);
			target->d_name.name = dentry->d_name.name;
			dentry->d_name.name = dentry->d_iname;
		} else {
			/*
M
Miklos Szeredi 已提交
2745
			 * Both are internal.
L
Linus Torvalds 已提交
2746
			 */
M
Miklos Szeredi 已提交
2747 2748 2749 2750 2751 2752
			unsigned int i;
			BUILD_BUG_ON(!IS_ALIGNED(DNAME_INLINE_LEN, sizeof(long)));
			for (i = 0; i < DNAME_INLINE_LEN / sizeof(long); i++) {
				swap(((long *) &dentry->d_iname)[i],
				     ((long *) &target->d_iname)[i]);
			}
L
Linus Torvalds 已提交
2753 2754
		}
	}
2755
	swap(dentry->d_name.hash_len, target->d_name.hash_len);
L
Linus Torvalds 已提交
2756 2757
}

2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775
static void copy_name(struct dentry *dentry, struct dentry *target)
{
	struct external_name *old_name = NULL;
	if (unlikely(dname_external(dentry)))
		old_name = external_name(dentry);
	if (unlikely(dname_external(target))) {
		atomic_inc(&external_name(target)->u.count);
		dentry->d_name = target->d_name;
	} else {
		memcpy(dentry->d_iname, target->d_name.name,
				target->d_name.len + 1);
		dentry->d_name.name = dentry->d_iname;
		dentry->d_name.hash_len = target->d_name.hash_len;
	}
	if (old_name && likely(atomic_dec_and_test(&old_name->u.count)))
		kfree_rcu(old_name, u.head);
}

N
Nick Piggin 已提交
2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790
static void dentry_lock_for_move(struct dentry *dentry, struct dentry *target)
{
	if (IS_ROOT(dentry) || dentry->d_parent == target->d_parent)
		spin_lock(&target->d_parent->d_lock);
	else {
		if (d_ancestor(dentry->d_parent, target->d_parent)) {
			spin_lock(&dentry->d_parent->d_lock);
			spin_lock_nested(&target->d_parent->d_lock,
						DENTRY_D_LOCK_NESTED);
		} else {
			spin_lock(&target->d_parent->d_lock);
			spin_lock_nested(&dentry->d_parent->d_lock,
						DENTRY_D_LOCK_NESTED);
		}
	}
2791 2792
	spin_lock_nested(&dentry->d_lock, 2);
	spin_lock_nested(&target->d_lock, 3);
N
Nick Piggin 已提交
2793 2794
}

2795
/*
2796
 * __d_move - move a dentry
L
Linus Torvalds 已提交
2797 2798
 * @dentry: entry to move
 * @target: new dentry
M
Miklos Szeredi 已提交
2799
 * @exchange: exchange the two dentries
L
Linus Torvalds 已提交
2800 2801
 *
 * Update the dcache to reflect the move of a file name. Negative
2802 2803 2804
 * dcache entries should not be moved in this way. Caller must hold
 * rename_lock, the i_mutex of the source and target directories,
 * and the sb->s_vfs_rename_mutex if they differ. See lock_rename().
L
Linus Torvalds 已提交
2805
 */
M
Miklos Szeredi 已提交
2806 2807
static void __d_move(struct dentry *dentry, struct dentry *target,
		     bool exchange)
L
Linus Torvalds 已提交
2808
{
2809
	struct dentry *old_parent;
A
Al Viro 已提交
2810 2811
	struct inode *dir = NULL;
	unsigned n;
L
Linus Torvalds 已提交
2812 2813 2814
	if (!dentry->d_inode)
		printk(KERN_WARNING "VFS: moving negative dcache entry\n");

N
Nick Piggin 已提交
2815 2816 2817 2818
	BUG_ON(d_ancestor(dentry, target));
	BUG_ON(d_ancestor(target, dentry));

	dentry_lock_for_move(dentry, target);
A
Al Viro 已提交
2819 2820 2821
	if (unlikely(d_in_lookup(target))) {
		dir = target->d_parent->d_inode;
		n = start_dir_add(dir);
2822
		__d_lookup_done(target);
A
Al Viro 已提交
2823
	}
L
Linus Torvalds 已提交
2824

N
Nick Piggin 已提交
2825
	write_seqcount_begin(&dentry->d_seq);
2826
	write_seqcount_begin_nested(&target->d_seq, DENTRY_D_LOCK_NESTED);
N
Nick Piggin 已提交
2827

2828 2829
	old_parent = dentry->d_parent;

2830
	/* unhash both */
2831 2832 2833 2834
	if (!d_unhashed(dentry))
		___d_drop(dentry);
	if (!d_unhashed(target))
		___d_drop(target);
L
Linus Torvalds 已提交
2835

2836 2837 2838
	/* ... and switch them in the tree */
	dentry->d_parent = target->d_parent;
	if (!exchange) {
2839
		copy_name(dentry, target);
2840
		target->d_hash.pprev = NULL;
2841 2842 2843 2844 2845
		dentry->d_parent->d_lockref.count++;
		if (dentry == old_parent)
			dentry->d_flags |= DCACHE_RCUACCESS;
		else
			WARN_ON(!--old_parent->d_lockref.count);
L
Linus Torvalds 已提交
2846
	} else {
2847 2848
		target->d_parent = old_parent;
		swap_names(dentry, target);
2849
		list_move(&target->d_child, &target->d_parent->d_subdirs);
2850 2851
		__d_rehash(target);
		fsnotify_update_flags(target);
L
Linus Torvalds 已提交
2852
	}
2853 2854 2855
	list_move(&dentry->d_child, &dentry->d_parent->d_subdirs);
	__d_rehash(dentry);
	fsnotify_update_flags(dentry);
L
Linus Torvalds 已提交
2856

N
Nick Piggin 已提交
2857 2858 2859
	write_seqcount_end(&target->d_seq);
	write_seqcount_end(&dentry->d_seq);

A
Al Viro 已提交
2860 2861
	if (dir)
		end_dir_add(dir, n);
2862 2863 2864 2865 2866 2867 2868

	if (dentry->d_parent != old_parent)
		spin_unlock(&dentry->d_parent->d_lock);
	if (dentry != old_parent)
		spin_unlock(&old_parent->d_lock);
	spin_unlock(&target->d_lock);
	spin_unlock(&dentry->d_lock);
2869 2870 2871 2872 2873 2874 2875 2876
}

/*
 * d_move - move a dentry
 * @dentry: entry to move
 * @target: new dentry
 *
 * Update the dcache to reflect the move of a file name. Negative
2877 2878
 * dcache entries should not be moved in this way. See the locking
 * requirements for __d_move.
2879 2880 2881 2882
 */
void d_move(struct dentry *dentry, struct dentry *target)
{
	write_seqlock(&rename_lock);
M
Miklos Szeredi 已提交
2883
	__d_move(dentry, target, false);
L
Linus Torvalds 已提交
2884
	write_sequnlock(&rename_lock);
2885
}
2886
EXPORT_SYMBOL(d_move);
L
Linus Torvalds 已提交
2887

M
Miklos Szeredi 已提交
2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906
/*
 * d_exchange - exchange two dentries
 * @dentry1: first dentry
 * @dentry2: second dentry
 */
void d_exchange(struct dentry *dentry1, struct dentry *dentry2)
{
	write_seqlock(&rename_lock);

	WARN_ON(!dentry1->d_inode);
	WARN_ON(!dentry2->d_inode);
	WARN_ON(IS_ROOT(dentry1));
	WARN_ON(IS_ROOT(dentry2));

	__d_move(dentry1, dentry2, true);

	write_sequnlock(&rename_lock);
}

2907 2908 2909 2910 2911 2912 2913
/**
 * d_ancestor - search for an ancestor
 * @p1: ancestor dentry
 * @p2: child dentry
 *
 * Returns the ancestor dentry of p2 which is a child of p1, if p1 is
 * an ancestor of p2, else NULL.
2914
 */
2915
struct dentry *d_ancestor(struct dentry *p1, struct dentry *p2)
2916 2917 2918
{
	struct dentry *p;

2919
	for (p = p2; !IS_ROOT(p); p = p->d_parent) {
2920
		if (p->d_parent == p1)
2921
			return p;
2922
	}
2923
	return NULL;
2924 2925 2926 2927 2928 2929
}

/*
 * This helper attempts to cope with remotely renamed directories
 *
 * It assumes that the caller is already holding
2930
 * dentry->d_parent->d_inode->i_mutex, and rename_lock
2931 2932 2933 2934
 *
 * Note: If ever the locking in lock_rename() changes, then please
 * remember to update this too...
 */
2935
static int __d_unalias(struct inode *inode,
2936
		struct dentry *dentry, struct dentry *alias)
2937
{
2938 2939
	struct mutex *m1 = NULL;
	struct rw_semaphore *m2 = NULL;
2940
	int ret = -ESTALE;
2941 2942 2943 2944 2945 2946 2947 2948 2949

	/* If alias and dentry share a parent, then no extra locks required */
	if (alias->d_parent == dentry->d_parent)
		goto out_unalias;

	/* See lock_rename() */
	if (!mutex_trylock(&dentry->d_sb->s_vfs_rename_mutex))
		goto out_err;
	m1 = &dentry->d_sb->s_vfs_rename_mutex;
2950
	if (!inode_trylock_shared(alias->d_parent->d_inode))
2951
		goto out_err;
2952
	m2 = &alias->d_parent->d_inode->i_rwsem;
2953
out_unalias:
2954
	__d_move(alias, dentry, false);
2955
	ret = 0;
2956 2957
out_err:
	if (m2)
2958
		up_read(m2);
2959 2960 2961 2962 2963
	if (m1)
		mutex_unlock(m1);
	return ret;
}

J
J. Bruce Fields 已提交
2964 2965 2966 2967 2968
/**
 * d_splice_alias - splice a disconnected dentry into the tree if one exists
 * @inode:  the inode which may have a disconnected dentry
 * @dentry: a negative dentry which we want to point to the inode.
 *
2969 2970 2971
 * If inode is a directory and has an IS_ROOT alias, then d_move that in
 * place of the given dentry and return it, else simply d_add the inode
 * to the dentry and return NULL.
J
J. Bruce Fields 已提交
2972
 *
2973 2974 2975
 * If a non-IS_ROOT directory is found, the filesystem is corrupt, and
 * we should error out: directories can't have multiple aliases.
 *
J
J. Bruce Fields 已提交
2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991
 * This is needed in the lookup routine of any filesystem that is exportable
 * (via knfsd) so that we can build dcache paths to directories effectively.
 *
 * If a dentry was found and moved, then it is returned.  Otherwise NULL
 * is returned.  This matches the expected return value of ->lookup.
 *
 * Cluster filesystems may call this function with a negative, hashed dentry.
 * In that case, we know that the inode will be a regular file, and also this
 * will only occur during atomic_open. So we need to check for the dentry
 * being already hashed only in the final case.
 */
struct dentry *d_splice_alias(struct inode *inode, struct dentry *dentry)
{
	if (IS_ERR(inode))
		return ERR_CAST(inode);

2992 2993
	BUG_ON(!d_unhashed(dentry));

2994
	if (!inode)
2995
		goto out;
2996

2997
	security_d_instantiate(dentry, inode);
2998
	spin_lock(&inode->i_lock);
2999
	if (S_ISDIR(inode->i_mode)) {
3000 3001
		struct dentry *new = __d_find_any_alias(inode);
		if (unlikely(new)) {
3002 3003
			/* The reference to new ensures it remains an alias */
			spin_unlock(&inode->i_lock);
3004
			write_seqlock(&rename_lock);
3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015
			if (unlikely(d_ancestor(new, dentry))) {
				write_sequnlock(&rename_lock);
				dput(new);
				new = ERR_PTR(-ELOOP);
				pr_warn_ratelimited(
					"VFS: Lookup of '%s' in %s %s"
					" would have caused loop\n",
					dentry->d_name.name,
					inode->i_sb->s_type->name,
					inode->i_sb->s_id);
			} else if (!IS_ROOT(new)) {
3016
				struct dentry *old_parent = dget(new->d_parent);
3017
				int err = __d_unalias(inode, dentry, new);
3018
				write_sequnlock(&rename_lock);
3019 3020 3021 3022
				if (err) {
					dput(new);
					new = ERR_PTR(err);
				}
3023
				dput(old_parent);
3024
			} else {
3025 3026
				__d_move(new, dentry, false);
				write_sequnlock(&rename_lock);
3027
			}
3028 3029
			iput(inode);
			return new;
3030
		}
3031
	}
3032
out:
A
Al Viro 已提交
3033
	__d_add(dentry, inode);
3034
	return NULL;
3035
}
3036
EXPORT_SYMBOL(d_splice_alias);
3037

L
Linus Torvalds 已提交
3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048
/*
 * Test whether new_dentry is a subdirectory of old_dentry.
 *
 * Trivially implemented using the dcache structure
 */

/**
 * is_subdir - is new dentry a subdirectory of old_dentry
 * @new_dentry: new dentry
 * @old_dentry: old dentry
 *
Y
Yaowei Bai 已提交
3049 3050
 * Returns true if new_dentry is a subdirectory of the parent (at any depth).
 * Returns false otherwise.
L
Linus Torvalds 已提交
3051 3052 3053
 * Caller must ensure that "new_dentry" is pinned before calling is_subdir()
 */
  
Y
Yaowei Bai 已提交
3054
bool is_subdir(struct dentry *new_dentry, struct dentry *old_dentry)
L
Linus Torvalds 已提交
3055
{
Y
Yaowei Bai 已提交
3056
	bool result;
3057
	unsigned seq;
L
Linus Torvalds 已提交
3058

3059
	if (new_dentry == old_dentry)
Y
Yaowei Bai 已提交
3060
		return true;
3061 3062

	do {
L
Linus Torvalds 已提交
3063 3064
		/* for restarting inner loop in case of seq retry */
		seq = read_seqbegin(&rename_lock);
3065 3066 3067 3068 3069
		/*
		 * Need rcu_readlock to protect against the d_parent trashing
		 * due to d_move
		 */
		rcu_read_lock();
3070
		if (d_ancestor(old_dentry, new_dentry))
Y
Yaowei Bai 已提交
3071
			result = true;
3072
		else
Y
Yaowei Bai 已提交
3073
			result = false;
3074
		rcu_read_unlock();
L
Linus Torvalds 已提交
3075 3076 3077 3078
	} while (read_seqretry(&rename_lock, seq));

	return result;
}
3079
EXPORT_SYMBOL(is_subdir);
L
Linus Torvalds 已提交
3080

M
Miklos Szeredi 已提交
3081
static enum d_walk_ret d_genocide_kill(void *data, struct dentry *dentry)
L
Linus Torvalds 已提交
3082
{
M
Miklos Szeredi 已提交
3083 3084 3085 3086
	struct dentry *root = data;
	if (dentry != root) {
		if (d_unhashed(dentry) || !dentry->d_inode)
			return D_WALK_SKIP;
L
Linus Torvalds 已提交
3087

M
Miklos Szeredi 已提交
3088 3089 3090 3091
		if (!(dentry->d_flags & DCACHE_GENOCIDE)) {
			dentry->d_flags |= DCACHE_GENOCIDE;
			dentry->d_lockref.count--;
		}
L
Linus Torvalds 已提交
3092
	}
M
Miklos Szeredi 已提交
3093 3094
	return D_WALK_CONTINUE;
}
3095

M
Miklos Szeredi 已提交
3096 3097 3098
void d_genocide(struct dentry *parent)
{
	d_walk(parent, parent, d_genocide_kill, NULL);
L
Linus Torvalds 已提交
3099 3100
}

3101
void d_tmpfile(struct dentry *dentry, struct inode *inode)
L
Linus Torvalds 已提交
3102
{
3103 3104
	inode_dec_link_count(inode);
	BUG_ON(dentry->d_name.name != dentry->d_iname ||
3105
		!hlist_unhashed(&dentry->d_u.d_alias) ||
3106 3107 3108 3109 3110 3111 3112 3113
		!d_unlinked(dentry));
	spin_lock(&dentry->d_parent->d_lock);
	spin_lock_nested(&dentry->d_lock, DENTRY_D_LOCK_NESTED);
	dentry->d_name.len = sprintf(dentry->d_iname, "#%llu",
				(unsigned long long)inode->i_ino);
	spin_unlock(&dentry->d_lock);
	spin_unlock(&dentry->d_parent->d_lock);
	d_instantiate(dentry, inode);
L
Linus Torvalds 已提交
3114
}
3115
EXPORT_SYMBOL(d_tmpfile);
L
Linus Torvalds 已提交
3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136

static __initdata unsigned long dhash_entries;
static int __init set_dhash_entries(char *str)
{
	if (!str)
		return 0;
	dhash_entries = simple_strtoul(str, &str, 0);
	return 1;
}
__setup("dhash_entries=", set_dhash_entries);

static void __init dcache_init_early(void)
{
	/* If hashes are distributed across NUMA nodes, defer
	 * hash allocation until vmalloc space is available.
	 */
	if (hashdist)
		return;

	dentry_hashtable =
		alloc_large_system_hash("Dentry cache",
3137
					sizeof(struct hlist_bl_head),
L
Linus Torvalds 已提交
3138 3139
					dhash_entries,
					13,
3140
					HASH_EARLY | HASH_ZERO,
L
Linus Torvalds 已提交
3141
					&d_hash_shift,
3142
					NULL,
3143
					0,
L
Linus Torvalds 已提交
3144
					0);
3145
	d_hash_shift = 32 - d_hash_shift;
L
Linus Torvalds 已提交
3146 3147
}

3148
static void __init dcache_init(void)
L
Linus Torvalds 已提交
3149
{
3150
	/*
L
Linus Torvalds 已提交
3151 3152
	 * A constructor could be added for stable state like the lists,
	 * but it is probably not worth it because of the cache nature
3153
	 * of the dcache.
L
Linus Torvalds 已提交
3154
	 */
3155 3156 3157
	dentry_cache = KMEM_CACHE_USERCOPY(dentry,
		SLAB_RECLAIM_ACCOUNT|SLAB_PANIC|SLAB_MEM_SPREAD|SLAB_ACCOUNT,
		d_iname);
L
Linus Torvalds 已提交
3158 3159 3160 3161 3162 3163 3164

	/* Hash may have been set up in dcache_init_early */
	if (!hashdist)
		return;

	dentry_hashtable =
		alloc_large_system_hash("Dentry cache",
3165
					sizeof(struct hlist_bl_head),
L
Linus Torvalds 已提交
3166 3167
					dhash_entries,
					13,
3168
					HASH_ZERO,
L
Linus Torvalds 已提交
3169
					&d_hash_shift,
3170
					NULL,
3171
					0,
L
Linus Torvalds 已提交
3172
					0);
3173
	d_hash_shift = 32 - d_hash_shift;
L
Linus Torvalds 已提交
3174 3175 3176
}

/* SLAB cache for __getname() consumers */
3177
struct kmem_cache *names_cachep __read_mostly;
3178
EXPORT_SYMBOL(names_cachep);
L
Linus Torvalds 已提交
3179 3180 3181 3182 3183

EXPORT_SYMBOL(d_genocide);

void __init vfs_caches_init_early(void)
{
3184 3185 3186 3187 3188
	int i;

	for (i = 0; i < ARRAY_SIZE(in_lookup_hashtable); i++)
		INIT_HLIST_BL_HEAD(&in_lookup_hashtable[i]);

L
Linus Torvalds 已提交
3189 3190 3191 3192
	dcache_init_early();
	inode_init_early();
}

3193
void __init vfs_caches_init(void)
L
Linus Torvalds 已提交
3194
{
3195 3196
	names_cachep = kmem_cache_create_usercopy("names_cache", PATH_MAX, 0,
			SLAB_HWCACHE_ALIGN|SLAB_PANIC, 0, PATH_MAX, NULL);
L
Linus Torvalds 已提交
3197

3198 3199
	dcache_init();
	inode_init();
3200 3201
	files_init();
	files_maxfiles_init();
3202
	mnt_init();
L
Linus Torvalds 已提交
3203 3204 3205
	bdev_cache_init();
	chrdev_init();
}