namespace.c 65.6 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13
/*
 *  linux/fs/namespace.c
 *
 * (C) Copyright Al Viro 2000, 2001
 *	Released under GPL v2.
 *
 * Based on code from fs/super.c, copyright Linus Torvalds and others.
 * Heavily rewritten.
 */

#include <linux/syscalls.h>
#include <linux/slab.h>
#include <linux/sched.h>
N
Nick Piggin 已提交
14 15
#include <linux/spinlock.h>
#include <linux/percpu.h>
L
Linus Torvalds 已提交
16
#include <linux/init.h>
17
#include <linux/kernel.h>
L
Linus Torvalds 已提交
18
#include <linux/acct.h>
19
#include <linux/capability.h>
20
#include <linux/cpumask.h>
L
Linus Torvalds 已提交
21
#include <linux/module.h>
22
#include <linux/sysfs.h>
L
Linus Torvalds 已提交
23
#include <linux/seq_file.h>
24
#include <linux/mnt_namespace.h>
L
Linus Torvalds 已提交
25
#include <linux/namei.h>
A
Alexey Dobriyan 已提交
26
#include <linux/nsproxy.h>
L
Linus Torvalds 已提交
27 28
#include <linux/security.h>
#include <linux/mount.h>
29
#include <linux/ramfs.h>
E
Eric Dumazet 已提交
30
#include <linux/log2.h>
31
#include <linux/idr.h>
32
#include <linux/fs_struct.h>
33
#include <linux/fsnotify.h>
L
Linus Torvalds 已提交
34 35
#include <asm/uaccess.h>
#include <asm/unistd.h>
36
#include "pnode.h"
37
#include "internal.h"
L
Linus Torvalds 已提交
38

E
Eric Dumazet 已提交
39 40 41
#define HASH_SHIFT ilog2(PAGE_SIZE / sizeof(struct list_head))
#define HASH_SIZE (1UL << HASH_SHIFT)

A
Al Viro 已提交
42
static int event;
43
static DEFINE_IDA(mnt_id_ida);
44
static DEFINE_IDA(mnt_group_ida);
N
Nick Piggin 已提交
45
static DEFINE_SPINLOCK(mnt_id_lock);
46 47
static int mnt_id_start = 0;
static int mnt_group_start = 1;
L
Linus Torvalds 已提交
48

49
static struct list_head *mount_hashtable __read_mostly;
50
static struct kmem_cache *mnt_cache __read_mostly;
R
Ram Pai 已提交
51
static struct rw_semaphore namespace_sem;
L
Linus Torvalds 已提交
52

M
Miklos Szeredi 已提交
53
/* /sys/fs */
54 55
struct kobject *fs_kobj;
EXPORT_SYMBOL_GPL(fs_kobj);
M
Miklos Szeredi 已提交
56

N
Nick Piggin 已提交
57 58 59 60 61 62 63 64 65 66
/*
 * vfsmount lock may be taken for read to prevent changes to the
 * vfsmount hash, ie. during mountpoint lookups or walking back
 * up the tree.
 *
 * It should be taken for write in all cases where the vfsmount
 * tree or hash is modified or when a vfsmount structure is modified.
 */
DEFINE_BRLOCK(vfsmount_lock);

L
Linus Torvalds 已提交
67 68
static inline unsigned long hash(struct vfsmount *mnt, struct dentry *dentry)
{
R
Ram Pai 已提交
69 70
	unsigned long tmp = ((unsigned long)mnt / L1_CACHE_BYTES);
	tmp += ((unsigned long)dentry / L1_CACHE_BYTES);
E
Eric Dumazet 已提交
71 72
	tmp = tmp + (tmp >> HASH_SHIFT);
	return tmp & (HASH_SIZE - 1);
L
Linus Torvalds 已提交
73 74
}

75 76
#define MNT_WRITER_UNDERFLOW_LIMIT -(1<<16)

N
Nick Piggin 已提交
77 78 79 80
/*
 * allocation is serialized by namespace_sem, but we need the spinlock to
 * serialize with freeing.
 */
81 82 83 84 85 86
static int mnt_alloc_id(struct vfsmount *mnt)
{
	int res;

retry:
	ida_pre_get(&mnt_id_ida, GFP_KERNEL);
N
Nick Piggin 已提交
87
	spin_lock(&mnt_id_lock);
88 89 90
	res = ida_get_new_above(&mnt_id_ida, mnt_id_start, &mnt->mnt_id);
	if (!res)
		mnt_id_start = mnt->mnt_id + 1;
N
Nick Piggin 已提交
91
	spin_unlock(&mnt_id_lock);
92 93 94 95 96 97 98 99
	if (res == -EAGAIN)
		goto retry;

	return res;
}

static void mnt_free_id(struct vfsmount *mnt)
{
100
	int id = mnt->mnt_id;
N
Nick Piggin 已提交
101
	spin_lock(&mnt_id_lock);
102 103 104
	ida_remove(&mnt_id_ida, id);
	if (mnt_id_start > id)
		mnt_id_start = id;
N
Nick Piggin 已提交
105
	spin_unlock(&mnt_id_lock);
106 107
}

108 109 110 111 112 113 114
/*
 * Allocate a new peer group ID
 *
 * mnt_group_ida is protected by namespace_sem
 */
static int mnt_alloc_group_id(struct vfsmount *mnt)
{
115 116
	int res;

117 118 119
	if (!ida_pre_get(&mnt_group_ida, GFP_KERNEL))
		return -ENOMEM;

120 121 122 123 124 125 126
	res = ida_get_new_above(&mnt_group_ida,
				mnt_group_start,
				&mnt->mnt_group_id);
	if (!res)
		mnt_group_start = mnt->mnt_group_id + 1;

	return res;
127 128 129 130 131 132 133
}

/*
 * Release a peer group ID
 */
void mnt_release_group_id(struct vfsmount *mnt)
{
134 135 136 137
	int id = mnt->mnt_group_id;
	ida_remove(&mnt_group_ida, id);
	if (mnt_group_start > id)
		mnt_group_start = id;
138 139 140
	mnt->mnt_group_id = 0;
}

N
Nick Piggin 已提交
141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160
/*
 * vfsmount lock must be held for read
 */
static inline void mnt_add_count(struct vfsmount *mnt, int n)
{
#ifdef CONFIG_SMP
	this_cpu_add(mnt->mnt_pcp->mnt_count, n);
#else
	preempt_disable();
	mnt->mnt_count += n;
	preempt_enable();
#endif
}

/*
 * vfsmount lock must be held for write
 */
unsigned int mnt_get_count(struct vfsmount *mnt)
{
#ifdef CONFIG_SMP
A
Al Viro 已提交
161
	unsigned int count = 0;
N
Nick Piggin 已提交
162 163 164 165 166 167 168 169 170 171 172 173
	int cpu;

	for_each_possible_cpu(cpu) {
		count += per_cpu_ptr(mnt->mnt_pcp, cpu)->mnt_count;
	}

	return count;
#else
	return mnt->mnt_count;
#endif
}

174
static struct vfsmount *alloc_vfsmnt(const char *name)
L
Linus Torvalds 已提交
175
{
176
	struct vfsmount *mnt = kmem_cache_zalloc(mnt_cache, GFP_KERNEL);
L
Linus Torvalds 已提交
177
	if (mnt) {
178 179 180
		int err;

		err = mnt_alloc_id(mnt);
181 182 183 184 185 186 187
		if (err)
			goto out_free_cache;

		if (name) {
			mnt->mnt_devname = kstrdup(name, GFP_KERNEL);
			if (!mnt->mnt_devname)
				goto out_free_id;
188 189
		}

N
Nick Piggin 已提交
190 191 192 193 194
#ifdef CONFIG_SMP
		mnt->mnt_pcp = alloc_percpu(struct mnt_pcp);
		if (!mnt->mnt_pcp)
			goto out_free_devname;

A
Al Viro 已提交
195
		this_cpu_add(mnt->mnt_pcp->mnt_count, 1);
N
Nick Piggin 已提交
196 197 198 199 200
#else
		mnt->mnt_count = 1;
		mnt->mnt_writers = 0;
#endif

L
Linus Torvalds 已提交
201 202 203 204
		INIT_LIST_HEAD(&mnt->mnt_hash);
		INIT_LIST_HEAD(&mnt->mnt_child);
		INIT_LIST_HEAD(&mnt->mnt_mounts);
		INIT_LIST_HEAD(&mnt->mnt_list);
205
		INIT_LIST_HEAD(&mnt->mnt_expire);
R
Ram Pai 已提交
206
		INIT_LIST_HEAD(&mnt->mnt_share);
R
Ram Pai 已提交
207 208
		INIT_LIST_HEAD(&mnt->mnt_slave_list);
		INIT_LIST_HEAD(&mnt->mnt_slave);
209 210
#ifdef CONFIG_FSNOTIFY
		INIT_HLIST_HEAD(&mnt->mnt_fsnotify_marks);
N
npiggin@suse.de 已提交
211
#endif
L
Linus Torvalds 已提交
212 213
	}
	return mnt;
214

N
npiggin@suse.de 已提交
215 216 217 218
#ifdef CONFIG_SMP
out_free_devname:
	kfree(mnt->mnt_devname);
#endif
219 220 221 222 223
out_free_id:
	mnt_free_id(mnt);
out_free_cache:
	kmem_cache_free(mnt_cache, mnt);
	return NULL;
L
Linus Torvalds 已提交
224 225
}

226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246
/*
 * Most r/o checks on a fs are for operations that take
 * discrete amounts of time, like a write() or unlink().
 * We must keep track of when those operations start
 * (for permission checks) and when they end, so that
 * we can determine when writes are able to occur to
 * a filesystem.
 */
/*
 * __mnt_is_readonly: check whether a mount is read-only
 * @mnt: the mount to check for its write status
 *
 * This shouldn't be used directly ouside of the VFS.
 * It does not guarantee that the filesystem will stay
 * r/w, just that it is right *now*.  This can not and
 * should not be used in place of IS_RDONLY(inode).
 * mnt_want/drop_write() will _keep_ the filesystem
 * r/w.
 */
int __mnt_is_readonly(struct vfsmount *mnt)
{
247 248 249 250 251
	if (mnt->mnt_flags & MNT_READONLY)
		return 1;
	if (mnt->mnt_sb->s_flags & MS_RDONLY)
		return 1;
	return 0;
252 253 254
}
EXPORT_SYMBOL_GPL(__mnt_is_readonly);

255
static inline void mnt_inc_writers(struct vfsmount *mnt)
N
npiggin@suse.de 已提交
256 257
{
#ifdef CONFIG_SMP
N
Nick Piggin 已提交
258
	this_cpu_inc(mnt->mnt_pcp->mnt_writers);
N
npiggin@suse.de 已提交
259 260 261 262
#else
	mnt->mnt_writers++;
#endif
}
263

264
static inline void mnt_dec_writers(struct vfsmount *mnt)
265
{
N
npiggin@suse.de 已提交
266
#ifdef CONFIG_SMP
N
Nick Piggin 已提交
267
	this_cpu_dec(mnt->mnt_pcp->mnt_writers);
N
npiggin@suse.de 已提交
268 269 270
#else
	mnt->mnt_writers--;
#endif
271 272
}

273
static unsigned int mnt_get_writers(struct vfsmount *mnt)
274
{
N
npiggin@suse.de 已提交
275 276
#ifdef CONFIG_SMP
	unsigned int count = 0;
277 278 279
	int cpu;

	for_each_possible_cpu(cpu) {
N
Nick Piggin 已提交
280
		count += per_cpu_ptr(mnt->mnt_pcp, cpu)->mnt_writers;
281 282
	}

N
npiggin@suse.de 已提交
283 284 285 286
	return count;
#else
	return mnt->mnt_writers;
#endif
287 288
}

289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308
/*
 * Most r/o checks on a fs are for operations that take
 * discrete amounts of time, like a write() or unlink().
 * We must keep track of when those operations start
 * (for permission checks) and when they end, so that
 * we can determine when writes are able to occur to
 * a filesystem.
 */
/**
 * mnt_want_write - get write access to a mount
 * @mnt: the mount on which to take a write
 *
 * This tells the low-level filesystem that a write is
 * about to be performed to it, and makes sure that
 * writes are allowed before returning success.  When
 * the write operation is finished, mnt_drop_write()
 * must be called.  This is effectively a refcount.
 */
int mnt_want_write(struct vfsmount *mnt)
{
309 310
	int ret = 0;

N
npiggin@suse.de 已提交
311
	preempt_disable();
312
	mnt_inc_writers(mnt);
N
npiggin@suse.de 已提交
313
	/*
314
	 * The store to mnt_inc_writers must be visible before we pass
N
npiggin@suse.de 已提交
315 316 317 318 319 320 321 322 323 324 325 326
	 * MNT_WRITE_HOLD loop below, so that the slowpath can see our
	 * incremented count after it has set MNT_WRITE_HOLD.
	 */
	smp_mb();
	while (mnt->mnt_flags & MNT_WRITE_HOLD)
		cpu_relax();
	/*
	 * After the slowpath clears MNT_WRITE_HOLD, mnt_is_readonly will
	 * be set to match its requirements. So we must not load that until
	 * MNT_WRITE_HOLD is cleared.
	 */
	smp_rmb();
327
	if (__mnt_is_readonly(mnt)) {
328
		mnt_dec_writers(mnt);
329 330 331 332
		ret = -EROFS;
		goto out;
	}
out:
N
npiggin@suse.de 已提交
333
	preempt_enable();
334
	return ret;
335 336 337
}
EXPORT_SYMBOL_GPL(mnt_want_write);

N
npiggin@suse.de 已提交
338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355
/**
 * mnt_clone_write - get write access to a mount
 * @mnt: the mount on which to take a write
 *
 * This is effectively like mnt_want_write, except
 * it must only be used to take an extra write reference
 * on a mountpoint that we already know has a write reference
 * on it. This allows some optimisation.
 *
 * After finished, mnt_drop_write must be called as usual to
 * drop the reference.
 */
int mnt_clone_write(struct vfsmount *mnt)
{
	/* superblock may be r/o */
	if (__mnt_is_readonly(mnt))
		return -EROFS;
	preempt_disable();
356
	mnt_inc_writers(mnt);
N
npiggin@suse.de 已提交
357 358 359 360 361 362 363 364 365 366 367 368 369 370
	preempt_enable();
	return 0;
}
EXPORT_SYMBOL_GPL(mnt_clone_write);

/**
 * mnt_want_write_file - get write access to a file's mount
 * @file: the file who's mount on which to take a write
 *
 * This is like mnt_want_write, but it takes a file and can
 * do some optimisations if the file is open for write already
 */
int mnt_want_write_file(struct file *file)
{
371 372
	struct inode *inode = file->f_dentry->d_inode;
	if (!(file->f_mode & FMODE_WRITE) || special_file(inode->i_mode))
N
npiggin@suse.de 已提交
373 374 375 376 377 378
		return mnt_want_write(file->f_path.mnt);
	else
		return mnt_clone_write(file->f_path.mnt);
}
EXPORT_SYMBOL_GPL(mnt_want_write_file);

379 380 381 382 383 384 385 386 387 388
/**
 * mnt_drop_write - give up write access to a mount
 * @mnt: the mount on which to give up write access
 *
 * Tells the low-level filesystem that we are done
 * performing writes to it.  Must be matched with
 * mnt_want_write() call above.
 */
void mnt_drop_write(struct vfsmount *mnt)
{
N
npiggin@suse.de 已提交
389
	preempt_disable();
390
	mnt_dec_writers(mnt);
N
npiggin@suse.de 已提交
391
	preempt_enable();
392 393 394
}
EXPORT_SYMBOL_GPL(mnt_drop_write);

395
static int mnt_make_readonly(struct vfsmount *mnt)
396
{
397 398
	int ret = 0;

N
Nick Piggin 已提交
399
	br_write_lock(vfsmount_lock);
N
npiggin@suse.de 已提交
400
	mnt->mnt_flags |= MNT_WRITE_HOLD;
401
	/*
N
npiggin@suse.de 已提交
402 403
	 * After storing MNT_WRITE_HOLD, we'll read the counters. This store
	 * should be visible before we do.
404
	 */
N
npiggin@suse.de 已提交
405 406
	smp_mb();

407
	/*
N
npiggin@suse.de 已提交
408 409 410 411 412 413 414 415 416 417 418 419 420 421
	 * With writers on hold, if this value is zero, then there are
	 * definitely no active writers (although held writers may subsequently
	 * increment the count, they'll have to wait, and decrement it after
	 * seeing MNT_READONLY).
	 *
	 * It is OK to have counter incremented on one CPU and decremented on
	 * another: the sum will add up correctly. The danger would be when we
	 * sum up each counter, if we read a counter before it is incremented,
	 * but then read another CPU's count which it has been subsequently
	 * decremented from -- we would see more decrements than we should.
	 * MNT_WRITE_HOLD protects against this scenario, because
	 * mnt_want_write first increments count, then smp_mb, then spins on
	 * MNT_WRITE_HOLD, so it can't be decremented by another CPU while
	 * we're counting up here.
422
	 */
423
	if (mnt_get_writers(mnt) > 0)
N
npiggin@suse.de 已提交
424 425
		ret = -EBUSY;
	else
426
		mnt->mnt_flags |= MNT_READONLY;
N
npiggin@suse.de 已提交
427 428 429 430 431 432
	/*
	 * MNT_READONLY must become visible before ~MNT_WRITE_HOLD, so writers
	 * that become unheld will see MNT_READONLY.
	 */
	smp_wmb();
	mnt->mnt_flags &= ~MNT_WRITE_HOLD;
N
Nick Piggin 已提交
433
	br_write_unlock(vfsmount_lock);
434
	return ret;
435 436
}

437 438
static void __mnt_unmake_readonly(struct vfsmount *mnt)
{
N
Nick Piggin 已提交
439
	br_write_lock(vfsmount_lock);
440
	mnt->mnt_flags &= ~MNT_READONLY;
N
Nick Piggin 已提交
441
	br_write_unlock(vfsmount_lock);
442 443
}

444
static void free_vfsmnt(struct vfsmount *mnt)
L
Linus Torvalds 已提交
445 446
{
	kfree(mnt->mnt_devname);
447
	mnt_free_id(mnt);
N
npiggin@suse.de 已提交
448
#ifdef CONFIG_SMP
N
Nick Piggin 已提交
449
	free_percpu(mnt->mnt_pcp);
N
npiggin@suse.de 已提交
450
#endif
L
Linus Torvalds 已提交
451 452 453 454
	kmem_cache_free(mnt_cache, mnt);
}

/*
R
Ram Pai 已提交
455 456
 * find the first or last mount at @dentry on vfsmount @mnt depending on
 * @dir. If @dir is set return the first mount else return the last mount.
N
Nick Piggin 已提交
457
 * vfsmount_lock must be held for read or write.
L
Linus Torvalds 已提交
458
 */
R
Ram Pai 已提交
459 460
struct vfsmount *__lookup_mnt(struct vfsmount *mnt, struct dentry *dentry,
			      int dir)
L
Linus Torvalds 已提交
461
{
R
Ram Pai 已提交
462 463
	struct list_head *head = mount_hashtable + hash(mnt, dentry);
	struct list_head *tmp = head;
L
Linus Torvalds 已提交
464 465 466
	struct vfsmount *p, *found = NULL;

	for (;;) {
R
Ram Pai 已提交
467
		tmp = dir ? tmp->next : tmp->prev;
L
Linus Torvalds 已提交
468 469 470 471 472
		p = NULL;
		if (tmp == head)
			break;
		p = list_entry(tmp, struct vfsmount, mnt_hash);
		if (p->mnt_parent == mnt && p->mnt_mountpoint == dentry) {
R
Ram Pai 已提交
473
			found = p;
L
Linus Torvalds 已提交
474 475 476 477 478 479
			break;
		}
	}
	return found;
}

R
Ram Pai 已提交
480 481 482 483
/*
 * lookup_mnt increments the ref count before returning
 * the vfsmount struct.
 */
A
Al Viro 已提交
484
struct vfsmount *lookup_mnt(struct path *path)
R
Ram Pai 已提交
485 486
{
	struct vfsmount *child_mnt;
N
Nick Piggin 已提交
487 488

	br_read_lock(vfsmount_lock);
A
Al Viro 已提交
489
	if ((child_mnt = __lookup_mnt(path->mnt, path->dentry, 1)))
R
Ram Pai 已提交
490
		mntget(child_mnt);
N
Nick Piggin 已提交
491
	br_read_unlock(vfsmount_lock);
R
Ram Pai 已提交
492 493 494
	return child_mnt;
}

L
Linus Torvalds 已提交
495 496
static inline int check_mnt(struct vfsmount *mnt)
{
497
	return mnt->mnt_ns == current->nsproxy->mnt_ns;
L
Linus Torvalds 已提交
498 499
}

N
Nick Piggin 已提交
500 501 502
/*
 * vfsmount lock must be held for write
 */
503
static void touch_mnt_namespace(struct mnt_namespace *ns)
A
Al Viro 已提交
504 505 506 507 508 509 510
{
	if (ns) {
		ns->event = ++event;
		wake_up_interruptible(&ns->poll);
	}
}

N
Nick Piggin 已提交
511 512 513
/*
 * vfsmount lock must be held for write
 */
514
static void __touch_mnt_namespace(struct mnt_namespace *ns)
A
Al Viro 已提交
515 516 517 518 519 520 521
{
	if (ns && ns->event != event) {
		ns->event = event;
		wake_up_interruptible(&ns->poll);
	}
}

N
Nick Piggin 已提交
522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542
/*
 * Clear dentry's mounted state if it has no remaining mounts.
 * vfsmount_lock must be held for write.
 */
static void dentry_reset_mounted(struct vfsmount *mnt, struct dentry *dentry)
{
	unsigned u;

	for (u = 0; u < HASH_SIZE; u++) {
		struct vfsmount *p;

		list_for_each_entry(p, &mount_hashtable[u], mnt_hash) {
			if (p->mnt_mountpoint == dentry)
				return;
		}
	}
	spin_lock(&dentry->d_lock);
	dentry->d_flags &= ~DCACHE_MOUNTED;
	spin_unlock(&dentry->d_lock);
}

N
Nick Piggin 已提交
543 544 545
/*
 * vfsmount lock must be held for write
 */
546
static void detach_mnt(struct vfsmount *mnt, struct path *old_path)
L
Linus Torvalds 已提交
547
{
548 549
	old_path->dentry = mnt->mnt_mountpoint;
	old_path->mnt = mnt->mnt_parent;
L
Linus Torvalds 已提交
550 551 552 553
	mnt->mnt_parent = mnt;
	mnt->mnt_mountpoint = mnt->mnt_root;
	list_del_init(&mnt->mnt_child);
	list_del_init(&mnt->mnt_hash);
N
Nick Piggin 已提交
554
	dentry_reset_mounted(old_path->mnt, old_path->dentry);
L
Linus Torvalds 已提交
555 556
}

N
Nick Piggin 已提交
557 558 559
/*
 * vfsmount lock must be held for write
 */
560 561 562 563 564
void mnt_set_mountpoint(struct vfsmount *mnt, struct dentry *dentry,
			struct vfsmount *child_mnt)
{
	child_mnt->mnt_parent = mntget(mnt);
	child_mnt->mnt_mountpoint = dget(dentry);
N
Nick Piggin 已提交
565 566 567
	spin_lock(&dentry->d_lock);
	dentry->d_flags |= DCACHE_MOUNTED;
	spin_unlock(&dentry->d_lock);
568 569
}

N
Nick Piggin 已提交
570 571 572
/*
 * vfsmount lock must be held for write
 */
573
static void attach_mnt(struct vfsmount *mnt, struct path *path)
L
Linus Torvalds 已提交
574
{
575
	mnt_set_mountpoint(path->mnt, path->dentry, mnt);
576
	list_add_tail(&mnt->mnt_hash, mount_hashtable +
577 578
			hash(path->mnt, path->dentry));
	list_add_tail(&mnt->mnt_child, &path->mnt->mnt_mounts);
579 580
}

581 582 583 584 585 586 587 588 589 590 591 592 593 594 595
static inline void __mnt_make_longterm(struct vfsmount *mnt)
{
#ifdef CONFIG_SMP
	atomic_inc(&mnt->mnt_longterm);
#endif
}

/* needs vfsmount lock for write */
static inline void __mnt_make_shortterm(struct vfsmount *mnt)
{
#ifdef CONFIG_SMP
	atomic_dec(&mnt->mnt_longterm);
#endif
}

596
/*
N
Nick Piggin 已提交
597
 * vfsmount lock must be held for write
598 599 600 601 602 603
 */
static void commit_tree(struct vfsmount *mnt)
{
	struct vfsmount *parent = mnt->mnt_parent;
	struct vfsmount *m;
	LIST_HEAD(head);
604
	struct mnt_namespace *n = parent->mnt_ns;
605 606 607 608

	BUG_ON(parent == mnt);

	list_add_tail(&head, &mnt->mnt_list);
A
Al Viro 已提交
609
	list_for_each_entry(m, &head, mnt_list) {
610
		m->mnt_ns = n;
611
		__mnt_make_longterm(m);
A
Al Viro 已提交
612 613
	}

614 615 616 617 618
	list_splice(&head, n->list.prev);

	list_add_tail(&mnt->mnt_hash, mount_hashtable +
				hash(parent, mnt->mnt_mountpoint));
	list_add_tail(&mnt->mnt_child, &parent->mnt_mounts);
619
	touch_mnt_namespace(n);
L
Linus Torvalds 已提交
620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637
}

static struct vfsmount *next_mnt(struct vfsmount *p, struct vfsmount *root)
{
	struct list_head *next = p->mnt_mounts.next;
	if (next == &p->mnt_mounts) {
		while (1) {
			if (p == root)
				return NULL;
			next = p->mnt_child.next;
			if (next != &p->mnt_parent->mnt_mounts)
				break;
			p = p->mnt_parent;
		}
	}
	return list_entry(next, struct vfsmount, mnt_child);
}

R
Ram Pai 已提交
638 639 640 641 642 643 644 645 646 647
static struct vfsmount *skip_mnt_tree(struct vfsmount *p)
{
	struct list_head *prev = p->mnt_mounts.prev;
	while (prev != &p->mnt_mounts) {
		p = list_entry(prev, struct vfsmount, mnt_child);
		prev = p->mnt_mounts.prev;
	}
	return p;
}

648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677
struct vfsmount *
vfs_kern_mount(struct file_system_type *type, int flags, const char *name, void *data)
{
	struct vfsmount *mnt;
	struct dentry *root;

	if (!type)
		return ERR_PTR(-ENODEV);

	mnt = alloc_vfsmnt(name);
	if (!mnt)
		return ERR_PTR(-ENOMEM);

	if (flags & MS_KERNMOUNT)
		mnt->mnt_flags = MNT_INTERNAL;

	root = mount_fs(type, flags, name, data);
	if (IS_ERR(root)) {
		free_vfsmnt(mnt);
		return ERR_CAST(root);
	}

	mnt->mnt_root = root;
	mnt->mnt_sb = root->d_sb;
	mnt->mnt_mountpoint = mnt->mnt_root;
	mnt->mnt_parent = mnt;
	return mnt;
}
EXPORT_SYMBOL_GPL(vfs_kern_mount);

R
Ram Pai 已提交
678 679
static struct vfsmount *clone_mnt(struct vfsmount *old, struct dentry *root,
					int flag)
L
Linus Torvalds 已提交
680 681 682 683 684
{
	struct super_block *sb = old->mnt_sb;
	struct vfsmount *mnt = alloc_vfsmnt(old->mnt_devname);

	if (mnt) {
685 686 687 688 689 690 691 692 693 694 695
		if (flag & (CL_SLAVE | CL_PRIVATE))
			mnt->mnt_group_id = 0; /* not a peer of original */
		else
			mnt->mnt_group_id = old->mnt_group_id;

		if ((flag & CL_MAKE_SHARED) && !mnt->mnt_group_id) {
			int err = mnt_alloc_group_id(mnt);
			if (err)
				goto out_free;
		}

696
		mnt->mnt_flags = old->mnt_flags & ~MNT_WRITE_HOLD;
L
Linus Torvalds 已提交
697 698 699 700 701
		atomic_inc(&sb->s_active);
		mnt->mnt_sb = sb;
		mnt->mnt_root = dget(root);
		mnt->mnt_mountpoint = mnt->mnt_root;
		mnt->mnt_parent = mnt;
702

R
Ram Pai 已提交
703 704 705 706
		if (flag & CL_SLAVE) {
			list_add(&mnt->mnt_slave, &old->mnt_slave_list);
			mnt->mnt_master = old;
			CLEAR_MNT_SHARED(mnt);
707
		} else if (!(flag & CL_PRIVATE)) {
708
			if ((flag & CL_MAKE_SHARED) || IS_MNT_SHARED(old))
R
Ram Pai 已提交
709 710 711 712 713
				list_add(&mnt->mnt_share, &old->mnt_share);
			if (IS_MNT_SLAVE(old))
				list_add(&mnt->mnt_slave, &old->mnt_slave);
			mnt->mnt_master = old->mnt_master;
		}
714 715
		if (flag & CL_MAKE_SHARED)
			set_mnt_shared(mnt);
L
Linus Torvalds 已提交
716 717 718

		/* stick the duplicate mount on the same expiry list
		 * as the original if that was on one */
R
Ram Pai 已提交
719 720 721 722
		if (flag & CL_EXPIRE) {
			if (!list_empty(&old->mnt_expire))
				list_add(&mnt->mnt_expire, &old->mnt_expire);
		}
L
Linus Torvalds 已提交
723 724
	}
	return mnt;
725 726 727 728

 out_free:
	free_vfsmnt(mnt);
	return NULL;
L
Linus Torvalds 已提交
729 730
}

N
Nick Piggin 已提交
731
static inline void mntfree(struct vfsmount *mnt)
L
Linus Torvalds 已提交
732 733
{
	struct super_block *sb = mnt->mnt_sb;
N
Nick Piggin 已提交
734

735 736 737 738 739 740
	/*
	 * This probably indicates that somebody messed
	 * up a mnt_want/drop_write() pair.  If this
	 * happens, the filesystem was probably unable
	 * to make r/w->r/o transitions.
	 */
N
npiggin@suse.de 已提交
741
	/*
N
Nick Piggin 已提交
742 743
	 * The locking used to deal with mnt_count decrement provides barriers,
	 * so mnt_get_writers() below is safe.
N
npiggin@suse.de 已提交
744
	 */
745
	WARN_ON(mnt_get_writers(mnt));
746
	fsnotify_vfsmount_delete(mnt);
L
Linus Torvalds 已提交
747 748 749 750 751
	dput(mnt->mnt_root);
	free_vfsmnt(mnt);
	deactivate_super(sb);
}

A
Al Viro 已提交
752
static void mntput_no_expire(struct vfsmount *mnt)
N
Nick Piggin 已提交
753 754
{
put_again:
A
Al Viro 已提交
755 756 757
#ifdef CONFIG_SMP
	br_read_lock(vfsmount_lock);
	if (likely(atomic_read(&mnt->mnt_longterm))) {
758
		mnt_add_count(mnt, -1);
N
Nick Piggin 已提交
759
		br_read_unlock(vfsmount_lock);
A
Al Viro 已提交
760
		return;
N
Nick Piggin 已提交
761
	}
A
Al Viro 已提交
762
	br_read_unlock(vfsmount_lock);
N
Nick Piggin 已提交
763

N
Nick Piggin 已提交
764
	br_write_lock(vfsmount_lock);
765
	mnt_add_count(mnt, -1);
N
Nick Piggin 已提交
766
	if (mnt_get_count(mnt)) {
N
Nick Piggin 已提交
767 768 769
		br_write_unlock(vfsmount_lock);
		return;
	}
N
Nick Piggin 已提交
770
#else
771
	mnt_add_count(mnt, -1);
N
Nick Piggin 已提交
772
	if (likely(mnt_get_count(mnt)))
N
Nick Piggin 已提交
773
		return;
N
Nick Piggin 已提交
774
	br_write_lock(vfsmount_lock);
A
Al Viro 已提交
775
#endif
N
Nick Piggin 已提交
776 777 778 779 780 781
	if (unlikely(mnt->mnt_pinned)) {
		mnt_add_count(mnt, mnt->mnt_pinned + 1);
		mnt->mnt_pinned = 0;
		br_write_unlock(vfsmount_lock);
		acct_auto_close_mnt(mnt);
		goto put_again;
782
	}
N
Nick Piggin 已提交
783
	br_write_unlock(vfsmount_lock);
N
Nick Piggin 已提交
784 785 786 787 788 789 790 791 792
	mntfree(mnt);
}

void mntput(struct vfsmount *mnt)
{
	if (mnt) {
		/* avoid cacheline pingpong, hope gcc doesn't get "smart" */
		if (unlikely(mnt->mnt_expiry_mark))
			mnt->mnt_expiry_mark = 0;
A
Al Viro 已提交
793
		mntput_no_expire(mnt);
N
Nick Piggin 已提交
794 795 796 797 798 799 800
	}
}
EXPORT_SYMBOL(mntput);

struct vfsmount *mntget(struct vfsmount *mnt)
{
	if (mnt)
801
		mnt_add_count(mnt, 1);
N
Nick Piggin 已提交
802 803 804 805
	return mnt;
}
EXPORT_SYMBOL(mntget);

806 807
void mnt_pin(struct vfsmount *mnt)
{
N
Nick Piggin 已提交
808
	br_write_lock(vfsmount_lock);
809
	mnt->mnt_pinned++;
N
Nick Piggin 已提交
810
	br_write_unlock(vfsmount_lock);
811 812 813 814 815
}
EXPORT_SYMBOL(mnt_pin);

void mnt_unpin(struct vfsmount *mnt)
{
N
Nick Piggin 已提交
816
	br_write_lock(vfsmount_lock);
817
	if (mnt->mnt_pinned) {
818
		mnt_add_count(mnt, 1);
819 820
		mnt->mnt_pinned--;
	}
N
Nick Piggin 已提交
821
	br_write_unlock(vfsmount_lock);
822 823
}
EXPORT_SYMBOL(mnt_unpin);
L
Linus Torvalds 已提交
824

825 826 827 828 829 830 831 832 833 834 835 836 837
static inline void mangle(struct seq_file *m, const char *s)
{
	seq_escape(m, s, " \t\n\\");
}

/*
 * Simple .show_options callback for filesystems which don't want to
 * implement more complex mount option showing.
 *
 * See also save_mount_options().
 */
int generic_show_options(struct seq_file *m, struct vfsmount *mnt)
{
838 839 840 841
	const char *options;

	rcu_read_lock();
	options = rcu_dereference(mnt->mnt_sb->s_options);
842 843 844 845 846

	if (options != NULL && options[0]) {
		seq_putc(m, ',');
		mangle(m, options);
	}
847
	rcu_read_unlock();
848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867

	return 0;
}
EXPORT_SYMBOL(generic_show_options);

/*
 * If filesystem uses generic_show_options(), this function should be
 * called from the fill_super() callback.
 *
 * The .remount_fs callback usually needs to be handled in a special
 * way, to make sure, that previous options are not overwritten if the
 * remount fails.
 *
 * Also note, that if the filesystem's .remount_fs function doesn't
 * reset all options to their default value, but changes only newly
 * given options, then the displayed options will not reflect reality
 * any more.
 */
void save_mount_options(struct super_block *sb, char *options)
{
868 869
	BUG_ON(sb->s_options);
	rcu_assign_pointer(sb->s_options, kstrdup(options, GFP_KERNEL));
870 871 872
}
EXPORT_SYMBOL(save_mount_options);

873 874 875 876 877 878 879 880 881 882 883
void replace_mount_options(struct super_block *sb, char *options)
{
	char *old = sb->s_options;
	rcu_assign_pointer(sb->s_options, options);
	if (old) {
		synchronize_rcu();
		kfree(old);
	}
}
EXPORT_SYMBOL(replace_mount_options);

884
#ifdef CONFIG_PROC_FS
L
Linus Torvalds 已提交
885 886 887
/* iterator */
static void *m_start(struct seq_file *m, loff_t *pos)
{
888
	struct proc_mounts *p = m->private;
L
Linus Torvalds 已提交
889

R
Ram Pai 已提交
890
	down_read(&namespace_sem);
891
	return seq_list_start(&p->ns->list, *pos);
L
Linus Torvalds 已提交
892 893 894 895
}

static void *m_next(struct seq_file *m, void *v, loff_t *pos)
{
896
	struct proc_mounts *p = m->private;
897

898
	return seq_list_next(v, &p->ns->list, pos);
L
Linus Torvalds 已提交
899 900 901 902
}

static void m_stop(struct seq_file *m, void *v)
{
R
Ram Pai 已提交
903
	up_read(&namespace_sem);
L
Linus Torvalds 已提交
904 905
}

906 907 908 909 910
int mnt_had_events(struct proc_mounts *p)
{
	struct mnt_namespace *ns = p->ns;
	int res = 0;

N
Nick Piggin 已提交
911
	br_read_lock(vfsmount_lock);
912 913
	if (p->m.poll_event != ns->event) {
		p->m.poll_event = ns->event;
914 915
		res = 1;
	}
N
Nick Piggin 已提交
916
	br_read_unlock(vfsmount_lock);
917 918 919 920

	return res;
}

921 922 923 924 925
struct proc_fs_info {
	int flag;
	const char *str;
};

926
static int show_sb_opts(struct seq_file *m, struct super_block *sb)
L
Linus Torvalds 已提交
927
{
928
	static const struct proc_fs_info fs_info[] = {
L
Linus Torvalds 已提交
929 930 931 932 933
		{ MS_SYNCHRONOUS, ",sync" },
		{ MS_DIRSYNC, ",dirsync" },
		{ MS_MANDLOCK, ",mand" },
		{ 0, NULL }
	};
934 935 936 937 938 939
	const struct proc_fs_info *fs_infop;

	for (fs_infop = fs_info; fs_infop->flag; fs_infop++) {
		if (sb->s_flags & fs_infop->flag)
			seq_puts(m, fs_infop->str);
	}
940 941

	return security_sb_show_options(m, sb);
942 943 944 945 946
}

static void show_mnt_opts(struct seq_file *m, struct vfsmount *mnt)
{
	static const struct proc_fs_info mnt_info[] = {
L
Linus Torvalds 已提交
947 948 949
		{ MNT_NOSUID, ",nosuid" },
		{ MNT_NODEV, ",nodev" },
		{ MNT_NOEXEC, ",noexec" },
950 951
		{ MNT_NOATIME, ",noatime" },
		{ MNT_NODIRATIME, ",nodiratime" },
V
Valerie Henson 已提交
952
		{ MNT_RELATIME, ",relatime" },
L
Linus Torvalds 已提交
953 954
		{ 0, NULL }
	};
955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975
	const struct proc_fs_info *fs_infop;

	for (fs_infop = mnt_info; fs_infop->flag; fs_infop++) {
		if (mnt->mnt_flags & fs_infop->flag)
			seq_puts(m, fs_infop->str);
	}
}

static void show_type(struct seq_file *m, struct super_block *sb)
{
	mangle(m, sb->s_type->name);
	if (sb->s_subtype && sb->s_subtype[0]) {
		seq_putc(m, '.');
		mangle(m, sb->s_subtype);
	}
}

static int show_vfsmnt(struct seq_file *m, void *v)
{
	struct vfsmount *mnt = list_entry(v, struct vfsmount, mnt_list);
	int err = 0;
976
	struct path mnt_path = { .dentry = mnt->mnt_root, .mnt = mnt };
L
Linus Torvalds 已提交
977

978 979 980 981 982 983 984
	if (mnt->mnt_sb->s_op->show_devname) {
		err = mnt->mnt_sb->s_op->show_devname(m, mnt);
		if (err)
			goto out;
	} else {
		mangle(m, mnt->mnt_devname ? mnt->mnt_devname : "none");
	}
L
Linus Torvalds 已提交
985
	seq_putc(m, ' ');
986
	seq_path(m, &mnt_path, " \t\n\\");
L
Linus Torvalds 已提交
987
	seq_putc(m, ' ');
988
	show_type(m, mnt->mnt_sb);
989
	seq_puts(m, __mnt_is_readonly(mnt) ? " ro" : " rw");
990 991 992
	err = show_sb_opts(m, mnt->mnt_sb);
	if (err)
		goto out;
993
	show_mnt_opts(m, mnt);
L
Linus Torvalds 已提交
994 995 996
	if (mnt->mnt_sb->s_op->show_options)
		err = mnt->mnt_sb->s_op->show_options(m, mnt);
	seq_puts(m, " 0 0\n");
997
out:
L
Linus Torvalds 已提交
998 999 1000
	return err;
}

1001
const struct seq_operations mounts_op = {
L
Linus Torvalds 已提交
1002 1003 1004 1005 1006 1007
	.start	= m_start,
	.next	= m_next,
	.stop	= m_stop,
	.show	= show_vfsmnt
};

1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018
static int show_mountinfo(struct seq_file *m, void *v)
{
	struct proc_mounts *p = m->private;
	struct vfsmount *mnt = list_entry(v, struct vfsmount, mnt_list);
	struct super_block *sb = mnt->mnt_sb;
	struct path mnt_path = { .dentry = mnt->mnt_root, .mnt = mnt };
	struct path root = p->root;
	int err = 0;

	seq_printf(m, "%i %i %u:%u ", mnt->mnt_id, mnt->mnt_parent->mnt_id,
		   MAJOR(sb->s_dev), MINOR(sb->s_dev));
1019 1020 1021 1022 1023 1024
	if (sb->s_op->show_path)
		err = sb->s_op->show_path(m, mnt);
	else
		seq_dentry(m, mnt->mnt_root, " \t\n\\");
	if (err)
		goto out;
1025
	seq_putc(m, ' ');
1026 1027 1028 1029 1030 1031

	/* mountpoints outside of chroot jail will give SEQ_SKIP on this */
	err = seq_path_root(m, &mnt_path, &root, " \t\n\\");
	if (err)
		goto out;

1032 1033 1034 1035 1036 1037
	seq_puts(m, mnt->mnt_flags & MNT_READONLY ? " ro" : " rw");
	show_mnt_opts(m, mnt);

	/* Tagged fields ("foo:X" or "bar") */
	if (IS_MNT_SHARED(mnt))
		seq_printf(m, " shared:%i", mnt->mnt_group_id);
1038 1039 1040 1041 1042 1043 1044
	if (IS_MNT_SLAVE(mnt)) {
		int master = mnt->mnt_master->mnt_group_id;
		int dom = get_dominating_id(mnt, &p->root);
		seq_printf(m, " master:%i", master);
		if (dom && dom != master)
			seq_printf(m, " propagate_from:%i", dom);
	}
1045 1046 1047 1048 1049 1050 1051
	if (IS_MNT_UNBINDABLE(mnt))
		seq_puts(m, " unbindable");

	/* Filesystem specific data */
	seq_puts(m, " - ");
	show_type(m, sb);
	seq_putc(m, ' ');
1052 1053 1054 1055 1056 1057
	if (sb->s_op->show_devname)
		err = sb->s_op->show_devname(m, mnt);
	else
		mangle(m, mnt->mnt_devname ? mnt->mnt_devname : "none");
	if (err)
		goto out;
1058
	seq_puts(m, sb->s_flags & MS_RDONLY ? " ro" : " rw");
1059 1060 1061
	err = show_sb_opts(m, sb);
	if (err)
		goto out;
1062 1063 1064
	if (sb->s_op->show_options)
		err = sb->s_op->show_options(m, mnt);
	seq_putc(m, '\n');
1065
out:
1066 1067 1068 1069 1070 1071 1072 1073 1074 1075
	return err;
}

const struct seq_operations mountinfo_op = {
	.start	= m_start,
	.next	= m_next,
	.stop	= m_stop,
	.show	= show_mountinfo,
};

1076 1077
static int show_vfsstat(struct seq_file *m, void *v)
{
1078
	struct vfsmount *mnt = list_entry(v, struct vfsmount, mnt_list);
1079
	struct path mnt_path = { .dentry = mnt->mnt_root, .mnt = mnt };
1080 1081 1082
	int err = 0;

	/* device */
1083
	if (mnt->mnt_sb->s_op->show_devname) {
1084
		seq_puts(m, "device ");
1085 1086 1087 1088 1089 1090 1091 1092
		err = mnt->mnt_sb->s_op->show_devname(m, mnt);
	} else {
		if (mnt->mnt_devname) {
			seq_puts(m, "device ");
			mangle(m, mnt->mnt_devname);
		} else
			seq_puts(m, "no device");
	}
1093 1094 1095

	/* mount point */
	seq_puts(m, " mounted on ");
1096
	seq_path(m, &mnt_path, " \t\n\\");
1097 1098 1099 1100
	seq_putc(m, ' ');

	/* file system type */
	seq_puts(m, "with fstype ");
1101
	show_type(m, mnt->mnt_sb);
1102 1103 1104 1105

	/* optional statistics */
	if (mnt->mnt_sb->s_op->show_stats) {
		seq_putc(m, ' ');
1106 1107
		if (!err)
			err = mnt->mnt_sb->s_op->show_stats(m, mnt);
1108 1109 1110 1111 1112 1113
	}

	seq_putc(m, '\n');
	return err;
}

1114
const struct seq_operations mountstats_op = {
1115 1116 1117 1118 1119
	.start	= m_start,
	.next	= m_next,
	.stop	= m_stop,
	.show	= show_vfsstat,
};
1120
#endif  /* CONFIG_PROC_FS */
1121

L
Linus Torvalds 已提交
1122 1123 1124 1125 1126 1127 1128 1129 1130 1131
/**
 * may_umount_tree - check if a mount tree is busy
 * @mnt: root of mount tree
 *
 * This is called to check if a tree of mounts has any
 * open files, pwds, chroots or sub mounts that are
 * busy.
 */
int may_umount_tree(struct vfsmount *mnt)
{
R
Ram Pai 已提交
1132 1133 1134
	int actual_refs = 0;
	int minimum_refs = 0;
	struct vfsmount *p;
L
Linus Torvalds 已提交
1135

N
Nick Piggin 已提交
1136 1137
	/* write lock needed for mnt_get_count */
	br_write_lock(vfsmount_lock);
R
Ram Pai 已提交
1138
	for (p = mnt; p; p = next_mnt(p, mnt)) {
N
Nick Piggin 已提交
1139
		actual_refs += mnt_get_count(p);
L
Linus Torvalds 已提交
1140 1141
		minimum_refs += 2;
	}
N
Nick Piggin 已提交
1142
	br_write_unlock(vfsmount_lock);
L
Linus Torvalds 已提交
1143 1144

	if (actual_refs > minimum_refs)
1145
		return 0;
L
Linus Torvalds 已提交
1146

1147
	return 1;
L
Linus Torvalds 已提交
1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166
}

EXPORT_SYMBOL(may_umount_tree);

/**
 * may_umount - check if a mount point is busy
 * @mnt: root of mount
 *
 * This is called to check if a mount point has any
 * open files, pwds, chroots or sub mounts. If the
 * mount has sub mounts this will return busy
 * regardless of whether the sub mounts are busy.
 *
 * Doesn't take quota and stuff into account. IOW, in some cases it will
 * give false negatives. The main reason why it's here is that we need
 * a non-destructive way to look for easily umountable filesystems.
 */
int may_umount(struct vfsmount *mnt)
{
1167
	int ret = 1;
A
Al Viro 已提交
1168
	down_read(&namespace_sem);
N
Nick Piggin 已提交
1169
	br_write_lock(vfsmount_lock);
R
Ram Pai 已提交
1170
	if (propagate_mount_busy(mnt, 2))
1171
		ret = 0;
N
Nick Piggin 已提交
1172
	br_write_unlock(vfsmount_lock);
A
Al Viro 已提交
1173
	up_read(&namespace_sem);
R
Ram Pai 已提交
1174
	return ret;
L
Linus Torvalds 已提交
1175 1176 1177 1178
}

EXPORT_SYMBOL(may_umount);

1179
void release_mounts(struct list_head *head)
R
Ram Pai 已提交
1180 1181
{
	struct vfsmount *mnt;
M
Miklos Szeredi 已提交
1182
	while (!list_empty(head)) {
1183
		mnt = list_first_entry(head, struct vfsmount, mnt_hash);
R
Ram Pai 已提交
1184 1185 1186 1187
		list_del_init(&mnt->mnt_hash);
		if (mnt->mnt_parent != mnt) {
			struct dentry *dentry;
			struct vfsmount *m;
N
Nick Piggin 已提交
1188 1189

			br_write_lock(vfsmount_lock);
R
Ram Pai 已提交
1190 1191 1192 1193
			dentry = mnt->mnt_mountpoint;
			m = mnt->mnt_parent;
			mnt->mnt_mountpoint = mnt->mnt_root;
			mnt->mnt_parent = mnt;
1194
			m->mnt_ghosts--;
N
Nick Piggin 已提交
1195
			br_write_unlock(vfsmount_lock);
R
Ram Pai 已提交
1196 1197 1198
			dput(dentry);
			mntput(m);
		}
A
Al Viro 已提交
1199
		mntput(mnt);
R
Ram Pai 已提交
1200 1201 1202
	}
}

N
Nick Piggin 已提交
1203 1204 1205 1206
/*
 * vfsmount lock must be held for write
 * namespace_sem must be held for write
 */
R
Ram Pai 已提交
1207
void umount_tree(struct vfsmount *mnt, int propagate, struct list_head *kill)
L
Linus Torvalds 已提交
1208
{
A
Al Viro 已提交
1209
	LIST_HEAD(tmp_list);
L
Linus Torvalds 已提交
1210 1211
	struct vfsmount *p;

A
Akinobu Mita 已提交
1212
	for (p = mnt; p; p = next_mnt(p, mnt))
A
Al Viro 已提交
1213
		list_move(&p->mnt_hash, &tmp_list);
L
Linus Torvalds 已提交
1214

R
Ram Pai 已提交
1215
	if (propagate)
A
Al Viro 已提交
1216
		propagate_umount(&tmp_list);
R
Ram Pai 已提交
1217

A
Al Viro 已提交
1218
	list_for_each_entry(p, &tmp_list, mnt_hash) {
R
Ram Pai 已提交
1219 1220
		list_del_init(&p->mnt_expire);
		list_del_init(&p->mnt_list);
1221 1222
		__touch_mnt_namespace(p->mnt_ns);
		p->mnt_ns = NULL;
1223
		__mnt_make_shortterm(p);
R
Ram Pai 已提交
1224
		list_del_init(&p->mnt_child);
1225 1226
		if (p->mnt_parent != p) {
			p->mnt_parent->mnt_ghosts++;
N
Nick Piggin 已提交
1227
			dentry_reset_mounted(p->mnt_parent, p->mnt_mountpoint);
1228
		}
R
Ram Pai 已提交
1229
		change_mnt_propagation(p, MS_PRIVATE);
L
Linus Torvalds 已提交
1230
	}
A
Al Viro 已提交
1231
	list_splice(&tmp_list, kill);
L
Linus Torvalds 已提交
1232 1233
}

1234 1235
static void shrink_submounts(struct vfsmount *mnt, struct list_head *umounts);

L
Linus Torvalds 已提交
1236 1237
static int do_umount(struct vfsmount *mnt, int flags)
{
R
Ram Pai 已提交
1238
	struct super_block *sb = mnt->mnt_sb;
L
Linus Torvalds 已提交
1239
	int retval;
R
Ram Pai 已提交
1240
	LIST_HEAD(umount_list);
L
Linus Torvalds 已提交
1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252

	retval = security_sb_umount(mnt, flags);
	if (retval)
		return retval;

	/*
	 * Allow userspace to request a mountpoint be expired rather than
	 * unmounting unconditionally. Unmount only happens if:
	 *  (1) the mark is already set (the mark is cleared by mntput())
	 *  (2) the usage count == 1 [parent vfsmount] + 1 [sys_umount]
	 */
	if (flags & MNT_EXPIRE) {
J
Jan Blunck 已提交
1253
		if (mnt == current->fs->root.mnt ||
L
Linus Torvalds 已提交
1254 1255 1256
		    flags & (MNT_FORCE | MNT_DETACH))
			return -EINVAL;

N
Nick Piggin 已提交
1257 1258 1259 1260 1261 1262
		/*
		 * probably don't strictly need the lock here if we examined
		 * all race cases, but it's a slowpath.
		 */
		br_write_lock(vfsmount_lock);
		if (mnt_get_count(mnt) != 2) {
1263
			br_write_unlock(vfsmount_lock);
L
Linus Torvalds 已提交
1264
			return -EBUSY;
N
Nick Piggin 已提交
1265 1266
		}
		br_write_unlock(vfsmount_lock);
L
Linus Torvalds 已提交
1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281

		if (!xchg(&mnt->mnt_expiry_mark, 1))
			return -EAGAIN;
	}

	/*
	 * If we may have to abort operations to get out of this
	 * mount, and they will themselves hold resources we must
	 * allow the fs to do things. In the Unix tradition of
	 * 'Gee thats tricky lets do it in userspace' the umount_begin
	 * might fail to complete on the first run through as other tasks
	 * must return, and the like. Thats for the mount program to worry
	 * about for the moment.
	 */

1282 1283 1284
	if (flags & MNT_FORCE && sb->s_op->umount_begin) {
		sb->s_op->umount_begin(sb);
	}
L
Linus Torvalds 已提交
1285 1286 1287 1288 1289 1290 1291 1292 1293 1294

	/*
	 * No sense to grab the lock for this test, but test itself looks
	 * somewhat bogus. Suggestions for better replacement?
	 * Ho-hum... In principle, we might treat that as umount + switch
	 * to rootfs. GC would eventually take care of the old vfsmount.
	 * Actually it makes sense, especially if rootfs would contain a
	 * /reboot - static binary that would close all descriptors and
	 * call reboot(9). Then init(8) could umount root and exec /reboot.
	 */
J
Jan Blunck 已提交
1295
	if (mnt == current->fs->root.mnt && !(flags & MNT_DETACH)) {
L
Linus Torvalds 已提交
1296 1297 1298 1299 1300
		/*
		 * Special case for "unmounting" root ...
		 * we just try to remount it readonly.
		 */
		down_write(&sb->s_umount);
A
Al Viro 已提交
1301
		if (!(sb->s_flags & MS_RDONLY))
L
Linus Torvalds 已提交
1302 1303 1304 1305 1306
			retval = do_remount_sb(sb, MS_RDONLY, NULL, 0);
		up_write(&sb->s_umount);
		return retval;
	}

R
Ram Pai 已提交
1307
	down_write(&namespace_sem);
N
Nick Piggin 已提交
1308
	br_write_lock(vfsmount_lock);
A
Al Viro 已提交
1309
	event++;
L
Linus Torvalds 已提交
1310

1311 1312 1313
	if (!(flags & MNT_DETACH))
		shrink_submounts(mnt, &umount_list);

L
Linus Torvalds 已提交
1314
	retval = -EBUSY;
R
Ram Pai 已提交
1315
	if (flags & MNT_DETACH || !propagate_mount_busy(mnt, 2)) {
L
Linus Torvalds 已提交
1316
		if (!list_empty(&mnt->mnt_list))
R
Ram Pai 已提交
1317
			umount_tree(mnt, 1, &umount_list);
L
Linus Torvalds 已提交
1318 1319
		retval = 0;
	}
N
Nick Piggin 已提交
1320
	br_write_unlock(vfsmount_lock);
R
Ram Pai 已提交
1321
	up_write(&namespace_sem);
R
Ram Pai 已提交
1322
	release_mounts(&umount_list);
L
Linus Torvalds 已提交
1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333
	return retval;
}

/*
 * Now umount can handle mount points as well as block devices.
 * This is important for filesystems which use unnamed block devices.
 *
 * We now support a flag for forced unmount like the other 'big iron'
 * unixes. Our API is identical to OSF/1 to avoid making a mess of AMD
 */

1334
SYSCALL_DEFINE2(umount, char __user *, name, int, flags)
L
Linus Torvalds 已提交
1335
{
1336
	struct path path;
L
Linus Torvalds 已提交
1337
	int retval;
1338
	int lookup_flags = 0;
L
Linus Torvalds 已提交
1339

1340 1341 1342 1343 1344 1345 1346
	if (flags & ~(MNT_FORCE | MNT_DETACH | MNT_EXPIRE | UMOUNT_NOFOLLOW))
		return -EINVAL;

	if (!(flags & UMOUNT_NOFOLLOW))
		lookup_flags |= LOOKUP_FOLLOW;

	retval = user_path_at(AT_FDCWD, name, lookup_flags, &path);
L
Linus Torvalds 已提交
1347 1348 1349
	if (retval)
		goto out;
	retval = -EINVAL;
1350
	if (path.dentry != path.mnt->mnt_root)
L
Linus Torvalds 已提交
1351
		goto dput_and_out;
1352
	if (!check_mnt(path.mnt))
L
Linus Torvalds 已提交
1353 1354 1355 1356 1357 1358
		goto dput_and_out;

	retval = -EPERM;
	if (!capable(CAP_SYS_ADMIN))
		goto dput_and_out;

1359
	retval = do_umount(path.mnt, flags);
L
Linus Torvalds 已提交
1360
dput_and_out:
J
Jan Blunck 已提交
1361
	/* we mustn't call path_put() as that would clear mnt_expiry_mark */
1362 1363
	dput(path.dentry);
	mntput_no_expire(path.mnt);
L
Linus Torvalds 已提交
1364 1365 1366 1367 1368 1369 1370
out:
	return retval;
}

#ifdef __ARCH_WANT_SYS_OLDUMOUNT

/*
R
Ram Pai 已提交
1371
 *	The 2.0 compatible umount. No flags.
L
Linus Torvalds 已提交
1372
 */
1373
SYSCALL_DEFINE1(oldumount, char __user *, name)
L
Linus Torvalds 已提交
1374
{
R
Ram Pai 已提交
1375
	return sys_umount(name, 0);
L
Linus Torvalds 已提交
1376 1377 1378 1379
}

#endif

1380
static int mount_is_safe(struct path *path)
L
Linus Torvalds 已提交
1381 1382 1383 1384 1385
{
	if (capable(CAP_SYS_ADMIN))
		return 0;
	return -EPERM;
#ifdef notyet
1386
	if (S_ISLNK(path->dentry->d_inode->i_mode))
L
Linus Torvalds 已提交
1387
		return -EPERM;
1388
	if (path->dentry->d_inode->i_mode & S_ISVTX) {
1389
		if (current_uid() != path->dentry->d_inode->i_uid)
L
Linus Torvalds 已提交
1390 1391
			return -EPERM;
	}
1392
	if (inode_permission(path->dentry->d_inode, MAY_WRITE))
L
Linus Torvalds 已提交
1393 1394 1395 1396 1397
		return -EPERM;
	return 0;
#endif
}

1398
struct vfsmount *copy_tree(struct vfsmount *mnt, struct dentry *dentry,
R
Ram Pai 已提交
1399
					int flag)
L
Linus Torvalds 已提交
1400 1401
{
	struct vfsmount *res, *p, *q, *r, *s;
1402
	struct path path;
L
Linus Torvalds 已提交
1403

R
Ram Pai 已提交
1404 1405 1406
	if (!(flag & CL_COPY_ALL) && IS_MNT_UNBINDABLE(mnt))
		return NULL;

R
Ram Pai 已提交
1407
	res = q = clone_mnt(mnt, dentry, flag);
L
Linus Torvalds 已提交
1408 1409 1410 1411 1412
	if (!q)
		goto Enomem;
	q->mnt_mountpoint = mnt->mnt_mountpoint;

	p = mnt;
1413
	list_for_each_entry(r, &mnt->mnt_mounts, mnt_child) {
1414
		if (!is_subdir(r->mnt_mountpoint, dentry))
L
Linus Torvalds 已提交
1415 1416 1417
			continue;

		for (s = r; s; s = next_mnt(s, r)) {
R
Ram Pai 已提交
1418 1419 1420 1421
			if (!(flag & CL_COPY_ALL) && IS_MNT_UNBINDABLE(s)) {
				s = skip_mnt_tree(s);
				continue;
			}
L
Linus Torvalds 已提交
1422 1423 1424 1425 1426
			while (p != s->mnt_parent) {
				p = p->mnt_parent;
				q = q->mnt_parent;
			}
			p = s;
1427 1428
			path.mnt = q;
			path.dentry = p->mnt_mountpoint;
R
Ram Pai 已提交
1429
			q = clone_mnt(p, p->mnt_root, flag);
L
Linus Torvalds 已提交
1430 1431
			if (!q)
				goto Enomem;
N
Nick Piggin 已提交
1432
			br_write_lock(vfsmount_lock);
L
Linus Torvalds 已提交
1433
			list_add_tail(&q->mnt_list, &res->mnt_list);
1434
			attach_mnt(q, &path);
N
Nick Piggin 已提交
1435
			br_write_unlock(vfsmount_lock);
L
Linus Torvalds 已提交
1436 1437 1438
		}
	}
	return res;
R
Ram Pai 已提交
1439
Enomem:
L
Linus Torvalds 已提交
1440
	if (res) {
R
Ram Pai 已提交
1441
		LIST_HEAD(umount_list);
N
Nick Piggin 已提交
1442
		br_write_lock(vfsmount_lock);
R
Ram Pai 已提交
1443
		umount_tree(res, 0, &umount_list);
N
Nick Piggin 已提交
1444
		br_write_unlock(vfsmount_lock);
R
Ram Pai 已提交
1445
		release_mounts(&umount_list);
L
Linus Torvalds 已提交
1446 1447 1448 1449
	}
	return NULL;
}

A
Al Viro 已提交
1450
struct vfsmount *collect_mounts(struct path *path)
1451 1452
{
	struct vfsmount *tree;
1453
	down_write(&namespace_sem);
A
Al Viro 已提交
1454
	tree = copy_tree(path->mnt, path->dentry, CL_COPY_ALL | CL_PRIVATE);
1455
	up_write(&namespace_sem);
1456 1457 1458 1459 1460 1461
	return tree;
}

void drop_collected_mounts(struct vfsmount *mnt)
{
	LIST_HEAD(umount_list);
1462
	down_write(&namespace_sem);
N
Nick Piggin 已提交
1463
	br_write_lock(vfsmount_lock);
1464
	umount_tree(mnt, 0, &umount_list);
N
Nick Piggin 已提交
1465
	br_write_unlock(vfsmount_lock);
1466
	up_write(&namespace_sem);
1467 1468 1469
	release_mounts(&umount_list);
}

A
Al Viro 已提交
1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484
int iterate_mounts(int (*f)(struct vfsmount *, void *), void *arg,
		   struct vfsmount *root)
{
	struct vfsmount *mnt;
	int res = f(root, arg);
	if (res)
		return res;
	list_for_each_entry(mnt, &root->mnt_list, mnt_list) {
		res = f(mnt, arg);
		if (res)
			return res;
	}
	return 0;
}

1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511
static void cleanup_group_ids(struct vfsmount *mnt, struct vfsmount *end)
{
	struct vfsmount *p;

	for (p = mnt; p != end; p = next_mnt(p, mnt)) {
		if (p->mnt_group_id && !IS_MNT_SHARED(p))
			mnt_release_group_id(p);
	}
}

static int invent_group_ids(struct vfsmount *mnt, bool recurse)
{
	struct vfsmount *p;

	for (p = mnt; p; p = recurse ? next_mnt(p, mnt) : NULL) {
		if (!p->mnt_group_id && !IS_MNT_SHARED(p)) {
			int err = mnt_alloc_group_id(p);
			if (err) {
				cleanup_group_ids(mnt, p);
				return err;
			}
		}
	}

	return 0;
}

1512 1513
/*
 *  @source_mnt : mount tree to be attached
R
Ram Pai 已提交
1514 1515 1516 1517
 *  @nd         : place the mount tree @source_mnt is attached
 *  @parent_nd  : if non-null, detach the source_mnt from its parent and
 *  		   store the parent mount and mountpoint dentry.
 *  		   (done when source_mnt is moved)
1518 1519 1520
 *
 *  NOTE: in the table below explains the semantics when a source mount
 *  of a given type is attached to a destination mount of a given type.
R
Ram Pai 已提交
1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532
 * ---------------------------------------------------------------------------
 * |         BIND MOUNT OPERATION                                            |
 * |**************************************************************************
 * | source-->| shared        |       private  |       slave    | unbindable |
 * | dest     |               |                |                |            |
 * |   |      |               |                |                |            |
 * |   v      |               |                |                |            |
 * |**************************************************************************
 * |  shared  | shared (++)   |     shared (+) |     shared(+++)|  invalid   |
 * |          |               |                |                |            |
 * |non-shared| shared (+)    |      private   |      slave (*) |  invalid   |
 * ***************************************************************************
1533 1534 1535 1536 1537 1538 1539 1540 1541
 * A bind operation clones the source mount and mounts the clone on the
 * destination mount.
 *
 * (++)  the cloned mount is propagated to all the mounts in the propagation
 * 	 tree of the destination mount and the cloned mount is added to
 * 	 the peer group of the source mount.
 * (+)   the cloned mount is created under the destination mount and is marked
 *       as shared. The cloned mount is added to the peer group of the source
 *       mount.
R
Ram Pai 已提交
1542 1543 1544 1545 1546 1547 1548
 * (+++) the mount is propagated to all the mounts in the propagation tree
 *       of the destination mount and the cloned mount is made slave
 *       of the same master as that of the source mount. The cloned mount
 *       is marked as 'shared and slave'.
 * (*)   the cloned mount is made a slave of the same master as that of the
 * 	 source mount.
 *
R
Ram Pai 已提交
1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560
 * ---------------------------------------------------------------------------
 * |         		MOVE MOUNT OPERATION                                 |
 * |**************************************************************************
 * | source-->| shared        |       private  |       slave    | unbindable |
 * | dest     |               |                |                |            |
 * |   |      |               |                |                |            |
 * |   v      |               |                |                |            |
 * |**************************************************************************
 * |  shared  | shared (+)    |     shared (+) |    shared(+++) |  invalid   |
 * |          |               |                |                |            |
 * |non-shared| shared (+*)   |      private   |    slave (*)   | unbindable |
 * ***************************************************************************
R
Ram Pai 已提交
1561 1562 1563
 *
 * (+)  the mount is moved to the destination. And is then propagated to
 * 	all the mounts in the propagation tree of the destination mount.
R
Ram Pai 已提交
1564
 * (+*)  the mount is moved to the destination.
R
Ram Pai 已提交
1565 1566 1567 1568
 * (+++)  the mount is moved to the destination and is then propagated to
 * 	all the mounts belonging to the destination mount's propagation tree.
 * 	the mount is marked as 'shared and slave'.
 * (*)	the mount continues to be a slave at the new location.
1569 1570 1571 1572 1573 1574 1575
 *
 * if the source mount is a tree, the operations explained above is
 * applied to each mount in the tree.
 * Must be called without spinlocks held, since this function can sleep
 * in allocations.
 */
static int attach_recursive_mnt(struct vfsmount *source_mnt,
1576
			struct path *path, struct path *parent_path)
1577 1578
{
	LIST_HEAD(tree_list);
1579 1580
	struct vfsmount *dest_mnt = path->mnt;
	struct dentry *dest_dentry = path->dentry;
1581
	struct vfsmount *child, *p;
1582
	int err;
1583

1584 1585 1586 1587 1588 1589 1590 1591
	if (IS_MNT_SHARED(dest_mnt)) {
		err = invent_group_ids(source_mnt, true);
		if (err)
			goto out;
	}
	err = propagate_mnt(dest_mnt, dest_dentry, source_mnt, &tree_list);
	if (err)
		goto out_cleanup_ids;
1592

N
Nick Piggin 已提交
1593
	br_write_lock(vfsmount_lock);
1594

1595 1596 1597 1598
	if (IS_MNT_SHARED(dest_mnt)) {
		for (p = source_mnt; p; p = next_mnt(p, source_mnt))
			set_mnt_shared(p);
	}
1599 1600 1601
	if (parent_path) {
		detach_mnt(source_mnt, parent_path);
		attach_mnt(source_mnt, path);
1602
		touch_mnt_namespace(parent_path->mnt->mnt_ns);
R
Ram Pai 已提交
1603 1604 1605 1606
	} else {
		mnt_set_mountpoint(dest_mnt, dest_dentry, source_mnt);
		commit_tree(source_mnt);
	}
1607 1608 1609 1610 1611

	list_for_each_entry_safe(child, p, &tree_list, mnt_hash) {
		list_del_init(&child->mnt_hash);
		commit_tree(child);
	}
N
Nick Piggin 已提交
1612 1613
	br_write_unlock(vfsmount_lock);

1614
	return 0;
1615 1616 1617 1618 1619 1620

 out_cleanup_ids:
	if (IS_MNT_SHARED(dest_mnt))
		cleanup_group_ids(source_mnt, NULL);
 out:
	return err;
1621 1622
}

1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649
static int lock_mount(struct path *path)
{
	struct vfsmount *mnt;
retry:
	mutex_lock(&path->dentry->d_inode->i_mutex);
	if (unlikely(cant_mount(path->dentry))) {
		mutex_unlock(&path->dentry->d_inode->i_mutex);
		return -ENOENT;
	}
	down_write(&namespace_sem);
	mnt = lookup_mnt(path);
	if (likely(!mnt))
		return 0;
	up_write(&namespace_sem);
	mutex_unlock(&path->dentry->d_inode->i_mutex);
	path_put(path);
	path->mnt = mnt;
	path->dentry = dget(mnt->mnt_root);
	goto retry;
}

static void unlock_mount(struct path *path)
{
	up_write(&namespace_sem);
	mutex_unlock(&path->dentry->d_inode->i_mutex);
}

1650
static int graft_tree(struct vfsmount *mnt, struct path *path)
L
Linus Torvalds 已提交
1651 1652 1653 1654
{
	if (mnt->mnt_sb->s_flags & MS_NOUSER)
		return -EINVAL;

1655
	if (S_ISDIR(path->dentry->d_inode->i_mode) !=
L
Linus Torvalds 已提交
1656 1657 1658
	      S_ISDIR(mnt->mnt_root->d_inode->i_mode))
		return -ENOTDIR;

1659 1660
	if (d_unlinked(path->dentry))
		return -ENOENT;
L
Linus Torvalds 已提交
1661

1662
	return attach_recursive_mnt(mnt, path, NULL);
L
Linus Torvalds 已提交
1663 1664
}

1665 1666 1667 1668 1669 1670
/*
 * Sanity check the flags to change_mnt_propagation.
 */

static int flags_to_propagation_type(int flags)
{
1671
	int type = flags & ~(MS_REC | MS_SILENT);
1672 1673 1674 1675 1676 1677 1678 1679 1680 1681

	/* Fail if any non-propagation flags are set */
	if (type & ~(MS_SHARED | MS_PRIVATE | MS_SLAVE | MS_UNBINDABLE))
		return 0;
	/* Only one propagation flag should be set */
	if (!is_power_of_2(type))
		return 0;
	return type;
}

1682 1683 1684
/*
 * recursively change the type of the mountpoint.
 */
1685
static int do_change_type(struct path *path, int flag)
1686
{
1687
	struct vfsmount *m, *mnt = path->mnt;
1688
	int recurse = flag & MS_REC;
1689
	int type;
1690
	int err = 0;
1691

1692 1693 1694
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

1695
	if (path->dentry != path->mnt->mnt_root)
1696 1697
		return -EINVAL;

1698 1699 1700 1701
	type = flags_to_propagation_type(flag);
	if (!type)
		return -EINVAL;

1702
	down_write(&namespace_sem);
1703 1704 1705 1706 1707 1708
	if (type == MS_SHARED) {
		err = invent_group_ids(mnt, recurse);
		if (err)
			goto out_unlock;
	}

N
Nick Piggin 已提交
1709
	br_write_lock(vfsmount_lock);
1710 1711
	for (m = mnt; m; m = (recurse ? next_mnt(m, mnt) : NULL))
		change_mnt_propagation(m, type);
N
Nick Piggin 已提交
1712
	br_write_unlock(vfsmount_lock);
1713 1714

 out_unlock:
1715
	up_write(&namespace_sem);
1716
	return err;
1717 1718
}

L
Linus Torvalds 已提交
1719 1720 1721
/*
 * do loopback mount.
 */
1722
static int do_loopback(struct path *path, char *old_name,
1723
				int recurse)
L
Linus Torvalds 已提交
1724
{
1725
	LIST_HEAD(umount_list);
1726
	struct path old_path;
L
Linus Torvalds 已提交
1727
	struct vfsmount *mnt = NULL;
1728
	int err = mount_is_safe(path);
L
Linus Torvalds 已提交
1729 1730 1731 1732
	if (err)
		return err;
	if (!old_name || !*old_name)
		return -EINVAL;
1733
	err = kern_path(old_name, LOOKUP_FOLLOW|LOOKUP_AUTOMOUNT, &old_path);
L
Linus Torvalds 已提交
1734 1735 1736
	if (err)
		return err;

1737 1738 1739 1740
	err = lock_mount(path);
	if (err)
		goto out;

L
Linus Torvalds 已提交
1741
	err = -EINVAL;
1742
	if (IS_MNT_UNBINDABLE(old_path.mnt))
1743
		goto out2;
R
Ram Pai 已提交
1744

1745
	if (!check_mnt(path->mnt) || !check_mnt(old_path.mnt))
1746
		goto out2;
L
Linus Torvalds 已提交
1747

1748 1749
	err = -ENOMEM;
	if (recurse)
1750
		mnt = copy_tree(old_path.mnt, old_path.dentry, 0);
1751
	else
1752
		mnt = clone_mnt(old_path.mnt, old_path.dentry, 0);
1753 1754

	if (!mnt)
1755
		goto out2;
1756

1757
	err = graft_tree(mnt, path);
1758
	if (err) {
N
Nick Piggin 已提交
1759
		br_write_lock(vfsmount_lock);
R
Ram Pai 已提交
1760
		umount_tree(mnt, 0, &umount_list);
N
Nick Piggin 已提交
1761
		br_write_unlock(vfsmount_lock);
1762
	}
1763 1764 1765
out2:
	unlock_mount(path);
	release_mounts(&umount_list);
1766
out:
1767
	path_put(&old_path);
L
Linus Torvalds 已提交
1768 1769 1770
	return err;
}

1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787
static int change_mount_flags(struct vfsmount *mnt, int ms_flags)
{
	int error = 0;
	int readonly_request = 0;

	if (ms_flags & MS_RDONLY)
		readonly_request = 1;
	if (readonly_request == __mnt_is_readonly(mnt))
		return 0;

	if (readonly_request)
		error = mnt_make_readonly(mnt);
	else
		__mnt_unmake_readonly(mnt);
	return error;
}

L
Linus Torvalds 已提交
1788 1789 1790 1791 1792
/*
 * change filesystem flags. dir should be a physical root of filesystem.
 * If you've mounted a non-root directory somewhere and want to do remount
 * on it - tough luck.
 */
1793
static int do_remount(struct path *path, int flags, int mnt_flags,
L
Linus Torvalds 已提交
1794 1795 1796
		      void *data)
{
	int err;
1797
	struct super_block *sb = path->mnt->mnt_sb;
L
Linus Torvalds 已提交
1798 1799 1800 1801

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

1802
	if (!check_mnt(path->mnt))
L
Linus Torvalds 已提交
1803 1804
		return -EINVAL;

1805
	if (path->dentry != path->mnt->mnt_root)
L
Linus Torvalds 已提交
1806 1807
		return -EINVAL;

1808 1809 1810 1811
	err = security_sb_remount(sb, data);
	if (err)
		return err;

L
Linus Torvalds 已提交
1812
	down_write(&sb->s_umount);
1813
	if (flags & MS_BIND)
1814
		err = change_mount_flags(path->mnt, flags);
A
Al Viro 已提交
1815
	else
1816
		err = do_remount_sb(sb, flags, data, 0);
A
Al Viro 已提交
1817
	if (!err) {
N
Nick Piggin 已提交
1818
		br_write_lock(vfsmount_lock);
1819
		mnt_flags |= path->mnt->mnt_flags & MNT_PROPAGATION_MASK;
1820
		path->mnt->mnt_flags = mnt_flags;
N
Nick Piggin 已提交
1821
		br_write_unlock(vfsmount_lock);
A
Al Viro 已提交
1822
	}
L
Linus Torvalds 已提交
1823
	up_write(&sb->s_umount);
1824
	if (!err) {
N
Nick Piggin 已提交
1825
		br_write_lock(vfsmount_lock);
1826
		touch_mnt_namespace(path->mnt->mnt_ns);
N
Nick Piggin 已提交
1827
		br_write_unlock(vfsmount_lock);
1828
	}
L
Linus Torvalds 已提交
1829 1830 1831
	return err;
}

R
Ram Pai 已提交
1832 1833 1834 1835 1836 1837 1838 1839 1840 1841
static inline int tree_contains_unbindable(struct vfsmount *mnt)
{
	struct vfsmount *p;
	for (p = mnt; p; p = next_mnt(p, mnt)) {
		if (IS_MNT_UNBINDABLE(p))
			return 1;
	}
	return 0;
}

1842
static int do_move_mount(struct path *path, char *old_name)
L
Linus Torvalds 已提交
1843
{
1844
	struct path old_path, parent_path;
L
Linus Torvalds 已提交
1845 1846 1847 1848 1849 1850
	struct vfsmount *p;
	int err = 0;
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
	if (!old_name || !*old_name)
		return -EINVAL;
1851
	err = kern_path(old_name, LOOKUP_FOLLOW, &old_path);
L
Linus Torvalds 已提交
1852 1853 1854
	if (err)
		return err;

1855
	err = lock_mount(path);
1856 1857 1858
	if (err < 0)
		goto out;

L
Linus Torvalds 已提交
1859
	err = -EINVAL;
1860
	if (!check_mnt(path->mnt) || !check_mnt(old_path.mnt))
L
Linus Torvalds 已提交
1861 1862
		goto out1;

1863
	if (d_unlinked(path->dentry))
R
Ram Pai 已提交
1864
		goto out1;
L
Linus Torvalds 已提交
1865 1866

	err = -EINVAL;
1867
	if (old_path.dentry != old_path.mnt->mnt_root)
R
Ram Pai 已提交
1868
		goto out1;
L
Linus Torvalds 已提交
1869

1870
	if (old_path.mnt == old_path.mnt->mnt_parent)
R
Ram Pai 已提交
1871
		goto out1;
L
Linus Torvalds 已提交
1872

1873 1874
	if (S_ISDIR(path->dentry->d_inode->i_mode) !=
	      S_ISDIR(old_path.dentry->d_inode->i_mode))
R
Ram Pai 已提交
1875 1876 1877 1878
		goto out1;
	/*
	 * Don't move a mount residing in a shared parent.
	 */
1879 1880
	if (old_path.mnt->mnt_parent &&
	    IS_MNT_SHARED(old_path.mnt->mnt_parent))
R
Ram Pai 已提交
1881
		goto out1;
R
Ram Pai 已提交
1882 1883 1884 1885
	/*
	 * Don't move a mount tree containing unbindable mounts to a destination
	 * mount which is shared.
	 */
1886 1887
	if (IS_MNT_SHARED(path->mnt) &&
	    tree_contains_unbindable(old_path.mnt))
R
Ram Pai 已提交
1888
		goto out1;
L
Linus Torvalds 已提交
1889
	err = -ELOOP;
1890 1891
	for (p = path->mnt; p->mnt_parent != p; p = p->mnt_parent)
		if (p == old_path.mnt)
R
Ram Pai 已提交
1892
			goto out1;
L
Linus Torvalds 已提交
1893

1894
	err = attach_recursive_mnt(old_path.mnt, path, &parent_path);
1895
	if (err)
R
Ram Pai 已提交
1896
		goto out1;
L
Linus Torvalds 已提交
1897 1898 1899

	/* if the mount is moved, it should no longer be expire
	 * automatically */
1900
	list_del_init(&old_path.mnt->mnt_expire);
L
Linus Torvalds 已提交
1901
out1:
1902
	unlock_mount(path);
L
Linus Torvalds 已提交
1903 1904
out:
	if (!err)
1905
		path_put(&parent_path);
1906
	path_put(&old_path);
L
Linus Torvalds 已提交
1907 1908 1909
	return err;
}

1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957
static struct vfsmount *fs_set_subtype(struct vfsmount *mnt, const char *fstype)
{
	int err;
	const char *subtype = strchr(fstype, '.');
	if (subtype) {
		subtype++;
		err = -EINVAL;
		if (!subtype[0])
			goto err;
	} else
		subtype = "";

	mnt->mnt_sb->s_subtype = kstrdup(subtype, GFP_KERNEL);
	err = -ENOMEM;
	if (!mnt->mnt_sb->s_subtype)
		goto err;
	return mnt;

 err:
	mntput(mnt);
	return ERR_PTR(err);
}

struct vfsmount *
do_kern_mount(const char *fstype, int flags, const char *name, void *data)
{
	struct file_system_type *type = get_fs_type(fstype);
	struct vfsmount *mnt;
	if (!type)
		return ERR_PTR(-ENODEV);
	mnt = vfs_kern_mount(type, flags, name, data);
	if (!IS_ERR(mnt) && (type->fs_flags & FS_HAS_SUBTYPE) &&
	    !mnt->mnt_sb->s_subtype)
		mnt = fs_set_subtype(mnt, fstype);
	put_filesystem(type);
	return mnt;
}
EXPORT_SYMBOL_GPL(do_kern_mount);

/*
 * add a mount into a namespace's mount tree
 */
static int do_add_mount(struct vfsmount *newmnt, struct path *path, int mnt_flags)
{
	int err;

	mnt_flags &= ~(MNT_SHARED | MNT_WRITE_HOLD | MNT_INTERNAL);

1958 1959 1960
	err = lock_mount(path);
	if (err)
		return err;
1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979

	err = -EINVAL;
	if (!(mnt_flags & MNT_SHRINKABLE) && !check_mnt(path->mnt))
		goto unlock;

	/* Refuse the same filesystem on the same mount point */
	err = -EBUSY;
	if (path->mnt->mnt_sb == newmnt->mnt_sb &&
	    path->mnt->mnt_root == path->dentry)
		goto unlock;

	err = -EINVAL;
	if (S_ISLNK(newmnt->mnt_root->d_inode->i_mode))
		goto unlock;

	newmnt->mnt_flags = mnt_flags;
	err = graft_tree(newmnt, path);

unlock:
1980
	unlock_mount(path);
1981 1982
	return err;
}
A
Al Viro 已提交
1983

L
Linus Torvalds 已提交
1984 1985 1986 1987
/*
 * create a new mount for userspace and request it to be added into the
 * namespace's tree
 */
1988
static int do_new_mount(struct path *path, char *type, int flags,
L
Linus Torvalds 已提交
1989 1990 1991
			int mnt_flags, char *name, void *data)
{
	struct vfsmount *mnt;
1992
	int err;
L
Linus Torvalds 已提交
1993

1994
	if (!type)
L
Linus Torvalds 已提交
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
		return -EINVAL;

	/* we need capabilities... */
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	mnt = do_kern_mount(type, flags, name, data);
	if (IS_ERR(mnt))
		return PTR_ERR(mnt);

2005 2006 2007 2008
	err = do_add_mount(mnt, path, mnt_flags);
	if (err)
		mntput(mnt);
	return err;
L
Linus Torvalds 已提交
2009 2010
}

2011 2012 2013 2014 2015 2016 2017 2018 2019 2020
int finish_automount(struct vfsmount *m, struct path *path)
{
	int err;
	/* The new mount record should have at least 2 refs to prevent it being
	 * expired before we get a chance to add it
	 */
	BUG_ON(mnt_get_count(m) < 2);

	if (m->mnt_sb == path->mnt->mnt_sb &&
	    m->mnt_root == path->dentry) {
A
Al Viro 已提交
2021 2022
		err = -ELOOP;
		goto fail;
2023 2024 2025
	}

	err = do_add_mount(m, path, path->mnt->mnt_flags | MNT_SHRINKABLE);
A
Al Viro 已提交
2026 2027 2028 2029 2030 2031 2032 2033 2034 2035
	if (!err)
		return 0;
fail:
	/* remove m from any expiration list it may be on */
	if (!list_empty(&m->mnt_expire)) {
		down_write(&namespace_sem);
		br_write_lock(vfsmount_lock);
		list_del_init(&m->mnt_expire);
		br_write_unlock(vfsmount_lock);
		up_write(&namespace_sem);
2036
	}
A
Al Viro 已提交
2037 2038
	mntput(m);
	mntput(m);
2039 2040 2041
	return err;
}

2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058
/**
 * mnt_set_expiry - Put a mount on an expiration list
 * @mnt: The mount to list.
 * @expiry_list: The list to add the mount to.
 */
void mnt_set_expiry(struct vfsmount *mnt, struct list_head *expiry_list)
{
	down_write(&namespace_sem);
	br_write_lock(vfsmount_lock);

	list_add_tail(&mnt->mnt_expire, expiry_list);

	br_write_unlock(vfsmount_lock);
	up_write(&namespace_sem);
}
EXPORT_SYMBOL(mnt_set_expiry);

L
Linus Torvalds 已提交
2059 2060 2061 2062 2063 2064 2065 2066 2067
/*
 * process a list of expirable mountpoints with the intent of discarding any
 * mountpoints that aren't in use and haven't been touched since last we came
 * here
 */
void mark_mounts_for_expiry(struct list_head *mounts)
{
	struct vfsmount *mnt, *next;
	LIST_HEAD(graveyard);
2068
	LIST_HEAD(umounts);
L
Linus Torvalds 已提交
2069 2070 2071 2072

	if (list_empty(mounts))
		return;

2073
	down_write(&namespace_sem);
N
Nick Piggin 已提交
2074
	br_write_lock(vfsmount_lock);
L
Linus Torvalds 已提交
2075 2076 2077 2078 2079 2080 2081

	/* extract from the expiration list every vfsmount that matches the
	 * following criteria:
	 * - only referenced by its parent vfsmount
	 * - still marked for expiry (marked on the last call here; marks are
	 *   cleared by mntput())
	 */
2082
	list_for_each_entry_safe(mnt, next, mounts, mnt_expire) {
L
Linus Torvalds 已提交
2083
		if (!xchg(&mnt->mnt_expiry_mark, 1) ||
2084
			propagate_mount_busy(mnt, 1))
L
Linus Torvalds 已提交
2085
			continue;
2086
		list_move(&mnt->mnt_expire, &graveyard);
L
Linus Torvalds 已提交
2087
	}
2088 2089 2090 2091 2092
	while (!list_empty(&graveyard)) {
		mnt = list_first_entry(&graveyard, struct vfsmount, mnt_expire);
		touch_mnt_namespace(mnt->mnt_ns);
		umount_tree(mnt, 1, &umounts);
	}
N
Nick Piggin 已提交
2093
	br_write_unlock(vfsmount_lock);
2094 2095 2096
	up_write(&namespace_sem);

	release_mounts(&umounts);
T
Trond Myklebust 已提交
2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121
}

EXPORT_SYMBOL_GPL(mark_mounts_for_expiry);

/*
 * Ripoff of 'select_parent()'
 *
 * search the list of submounts for a given mountpoint, and move any
 * shrinkable submounts to the 'graveyard' list.
 */
static int select_submounts(struct vfsmount *parent, struct list_head *graveyard)
{
	struct vfsmount *this_parent = parent;
	struct list_head *next;
	int found = 0;

repeat:
	next = this_parent->mnt_mounts.next;
resume:
	while (next != &this_parent->mnt_mounts) {
		struct list_head *tmp = next;
		struct vfsmount *mnt = list_entry(tmp, struct vfsmount, mnt_child);

		next = tmp->next;
		if (!(mnt->mnt_flags & MNT_SHRINKABLE))
L
Linus Torvalds 已提交
2122
			continue;
T
Trond Myklebust 已提交
2123 2124 2125 2126 2127 2128 2129
		/*
		 * Descend a level if the d_mounts list is non-empty.
		 */
		if (!list_empty(&mnt->mnt_mounts)) {
			this_parent = mnt;
			goto repeat;
		}
L
Linus Torvalds 已提交
2130

T
Trond Myklebust 已提交
2131 2132 2133 2134
		if (!propagate_mount_busy(mnt, 1)) {
			list_move_tail(&mnt->mnt_expire, graveyard);
			found++;
		}
L
Linus Torvalds 已提交
2135
	}
T
Trond Myklebust 已提交
2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149
	/*
	 * All done at this level ... ascend and resume the search
	 */
	if (this_parent != parent) {
		next = this_parent->mnt_child.next;
		this_parent = this_parent->mnt_parent;
		goto resume;
	}
	return found;
}

/*
 * process a list of expirable mountpoints with the intent of discarding any
 * submounts of a specific parent mountpoint
N
Nick Piggin 已提交
2150 2151
 *
 * vfsmount_lock must be held for write
T
Trond Myklebust 已提交
2152
 */
2153
static void shrink_submounts(struct vfsmount *mnt, struct list_head *umounts)
T
Trond Myklebust 已提交
2154 2155
{
	LIST_HEAD(graveyard);
2156
	struct vfsmount *m;
T
Trond Myklebust 已提交
2157 2158

	/* extract submounts of 'mountpoint' from the expiration list */
2159
	while (select_submounts(mnt, &graveyard)) {
2160
		while (!list_empty(&graveyard)) {
2161
			m = list_first_entry(&graveyard, struct vfsmount,
2162
						mnt_expire);
E
Eric W. Biederman 已提交
2163 2164
			touch_mnt_namespace(m->mnt_ns);
			umount_tree(m, 1, umounts);
2165 2166
		}
	}
L
Linus Torvalds 已提交
2167 2168 2169 2170 2171 2172 2173 2174
}

/*
 * Some copy_from_user() implementations do not return the exact number of
 * bytes remaining to copy on a fault.  But copy_mount_options() requires that.
 * Note that this function differs from copy_from_user() in that it will oops
 * on bad values of `to', rather than returning a short copy.
 */
R
Ram Pai 已提交
2175 2176
static long exact_copy_from_user(void *to, const void __user * from,
				 unsigned long n)
L
Linus Torvalds 已提交
2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196
{
	char *t = to;
	const char __user *f = from;
	char c;

	if (!access_ok(VERIFY_READ, from, n))
		return n;

	while (n) {
		if (__get_user(c, f)) {
			memset(t, 0, n);
			break;
		}
		*t++ = c;
		f++;
		n--;
	}
	return n;
}

R
Ram Pai 已提交
2197
int copy_mount_options(const void __user * data, unsigned long *where)
L
Linus Torvalds 已提交
2198 2199 2200 2201
{
	int i;
	unsigned long page;
	unsigned long size;
R
Ram Pai 已提交
2202

L
Linus Torvalds 已提交
2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220
	*where = 0;
	if (!data)
		return 0;

	if (!(page = __get_free_page(GFP_KERNEL)))
		return -ENOMEM;

	/* We only care that *some* data at the address the user
	 * gave us is valid.  Just in case, we'll zero
	 * the remainder of the page.
	 */
	/* copy_from_user cannot cross TASK_SIZE ! */
	size = TASK_SIZE - (unsigned long)data;
	if (size > PAGE_SIZE)
		size = PAGE_SIZE;

	i = size - exact_copy_from_user((void *)page, data, size);
	if (!i) {
R
Ram Pai 已提交
2221
		free_page(page);
L
Linus Torvalds 已提交
2222 2223 2224 2225 2226 2227 2228 2229
		return -EFAULT;
	}
	if (i != PAGE_SIZE)
		memset((char *)page + i, 0, PAGE_SIZE - i);
	*where = page;
	return 0;
}

2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246
int copy_mount_string(const void __user *data, char **where)
{
	char *tmp;

	if (!data) {
		*where = NULL;
		return 0;
	}

	tmp = strndup_user(data, PAGE_SIZE);
	if (IS_ERR(tmp))
		return PTR_ERR(tmp);

	*where = tmp;
	return 0;
}

L
Linus Torvalds 已提交
2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260
/*
 * Flags is a 32-bit value that allows up to 31 non-fs dependent flags to
 * be given to the mount() call (ie: read-only, no-dev, no-suid etc).
 *
 * data is a (void *) that can point to any structure up to
 * PAGE_SIZE-1 bytes, which can contain arbitrary fs-dependent
 * information (or be NULL).
 *
 * Pre-0.97 versions of mount() didn't have a flags word.
 * When the flags word was introduced its top half was required
 * to have the magic value 0xC0ED, and this remained so until 2.4.0-test9.
 * Therefore, if this magic number is present, it carries no information
 * and must be discarded.
 */
R
Ram Pai 已提交
2261
long do_mount(char *dev_name, char *dir_name, char *type_page,
L
Linus Torvalds 已提交
2262 2263
		  unsigned long flags, void *data_page)
{
2264
	struct path path;
L
Linus Torvalds 已提交
2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279
	int retval = 0;
	int mnt_flags = 0;

	/* Discard magic */
	if ((flags & MS_MGC_MSK) == MS_MGC_VAL)
		flags &= ~MS_MGC_MSK;

	/* Basic sanity checks */

	if (!dir_name || !*dir_name || !memchr(dir_name, 0, PAGE_SIZE))
		return -EINVAL;

	if (data_page)
		((char *)data_page)[PAGE_SIZE - 1] = 0;

2280 2281 2282 2283 2284 2285 2286 2287 2288 2289
	/* ... and get the mountpoint */
	retval = kern_path(dir_name, LOOKUP_FOLLOW, &path);
	if (retval)
		return retval;

	retval = security_sb_mount(dev_name, &path,
				   type_page, flags, data_page);
	if (retval)
		goto dput_out;

2290 2291 2292
	/* Default to relatime unless overriden */
	if (!(flags & MS_NOATIME))
		mnt_flags |= MNT_RELATIME;
M
Matthew Garrett 已提交
2293

L
Linus Torvalds 已提交
2294 2295 2296 2297 2298 2299 2300
	/* Separate the per-mountpoint flags */
	if (flags & MS_NOSUID)
		mnt_flags |= MNT_NOSUID;
	if (flags & MS_NODEV)
		mnt_flags |= MNT_NODEV;
	if (flags & MS_NOEXEC)
		mnt_flags |= MNT_NOEXEC;
2301 2302 2303 2304
	if (flags & MS_NOATIME)
		mnt_flags |= MNT_NOATIME;
	if (flags & MS_NODIRATIME)
		mnt_flags |= MNT_NODIRATIME;
M
Matthew Garrett 已提交
2305 2306
	if (flags & MS_STRICTATIME)
		mnt_flags &= ~(MNT_RELATIME | MNT_NOATIME);
2307 2308
	if (flags & MS_RDONLY)
		mnt_flags |= MNT_READONLY;
2309

A
Al Viro 已提交
2310
	flags &= ~(MS_NOSUID | MS_NOEXEC | MS_NODEV | MS_ACTIVE | MS_BORN |
M
Matthew Garrett 已提交
2311 2312
		   MS_NOATIME | MS_NODIRATIME | MS_RELATIME| MS_KERNMOUNT |
		   MS_STRICTATIME);
L
Linus Torvalds 已提交
2313 2314

	if (flags & MS_REMOUNT)
2315
		retval = do_remount(&path, flags & ~MS_REMOUNT, mnt_flags,
L
Linus Torvalds 已提交
2316 2317
				    data_page);
	else if (flags & MS_BIND)
2318
		retval = do_loopback(&path, dev_name, flags & MS_REC);
R
Ram Pai 已提交
2319
	else if (flags & (MS_SHARED | MS_PRIVATE | MS_SLAVE | MS_UNBINDABLE))
2320
		retval = do_change_type(&path, flags);
L
Linus Torvalds 已提交
2321
	else if (flags & MS_MOVE)
2322
		retval = do_move_mount(&path, dev_name);
L
Linus Torvalds 已提交
2323
	else
2324
		retval = do_new_mount(&path, type_page, flags, mnt_flags,
L
Linus Torvalds 已提交
2325 2326
				      dev_name, data_page);
dput_out:
2327
	path_put(&path);
L
Linus Torvalds 已提交
2328 2329 2330
	return retval;
}

2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345
static struct mnt_namespace *alloc_mnt_ns(void)
{
	struct mnt_namespace *new_ns;

	new_ns = kmalloc(sizeof(struct mnt_namespace), GFP_KERNEL);
	if (!new_ns)
		return ERR_PTR(-ENOMEM);
	atomic_set(&new_ns->count, 1);
	new_ns->root = NULL;
	INIT_LIST_HEAD(&new_ns->list);
	init_waitqueue_head(&new_ns->poll);
	new_ns->event = 0;
	return new_ns;
}

A
Al Viro 已提交
2346 2347
void mnt_make_longterm(struct vfsmount *mnt)
{
2348
	__mnt_make_longterm(mnt);
A
Al Viro 已提交
2349 2350 2351 2352
}

void mnt_make_shortterm(struct vfsmount *mnt)
{
2353
#ifdef CONFIG_SMP
A
Al Viro 已提交
2354 2355 2356 2357 2358
	if (atomic_add_unless(&mnt->mnt_longterm, -1, 1))
		return;
	br_write_lock(vfsmount_lock);
	atomic_dec(&mnt->mnt_longterm);
	br_write_unlock(vfsmount_lock);
2359
#endif
A
Al Viro 已提交
2360 2361
}

2362 2363 2364 2365
/*
 * Allocate a new namespace structure and populate it with contents
 * copied from the namespace of the passed in task structure.
 */
2366
static struct mnt_namespace *dup_mnt_ns(struct mnt_namespace *mnt_ns,
2367
		struct fs_struct *fs)
L
Linus Torvalds 已提交
2368
{
2369
	struct mnt_namespace *new_ns;
A
Al Viro 已提交
2370
	struct vfsmount *rootmnt = NULL, *pwdmnt = NULL;
L
Linus Torvalds 已提交
2371 2372
	struct vfsmount *p, *q;

2373 2374 2375
	new_ns = alloc_mnt_ns();
	if (IS_ERR(new_ns))
		return new_ns;
L
Linus Torvalds 已提交
2376

R
Ram Pai 已提交
2377
	down_write(&namespace_sem);
L
Linus Torvalds 已提交
2378
	/* First pass: copy the tree topology */
2379
	new_ns->root = copy_tree(mnt_ns->root, mnt_ns->root->mnt_root,
R
Ram Pai 已提交
2380
					CL_COPY_ALL | CL_EXPIRE);
L
Linus Torvalds 已提交
2381
	if (!new_ns->root) {
R
Ram Pai 已提交
2382
		up_write(&namespace_sem);
L
Linus Torvalds 已提交
2383
		kfree(new_ns);
2384
		return ERR_PTR(-ENOMEM);
L
Linus Torvalds 已提交
2385
	}
N
Nick Piggin 已提交
2386
	br_write_lock(vfsmount_lock);
L
Linus Torvalds 已提交
2387
	list_add_tail(&new_ns->list, &new_ns->root->mnt_list);
N
Nick Piggin 已提交
2388
	br_write_unlock(vfsmount_lock);
L
Linus Torvalds 已提交
2389 2390 2391 2392 2393 2394

	/*
	 * Second pass: switch the tsk->fs->* elements and mark new vfsmounts
	 * as belonging to new namespace.  We have already acquired a private
	 * fs_struct, so tsk->fs->lock is not needed.
	 */
2395
	p = mnt_ns->root;
L
Linus Torvalds 已提交
2396 2397
	q = new_ns->root;
	while (p) {
2398
		q->mnt_ns = new_ns;
2399
		__mnt_make_longterm(q);
L
Linus Torvalds 已提交
2400
		if (fs) {
J
Jan Blunck 已提交
2401
			if (p == fs->root.mnt) {
A
Al Viro 已提交
2402
				fs->root.mnt = mntget(q);
2403
				__mnt_make_longterm(q);
A
Al Viro 已提交
2404
				mnt_make_shortterm(p);
L
Linus Torvalds 已提交
2405 2406
				rootmnt = p;
			}
J
Jan Blunck 已提交
2407
			if (p == fs->pwd.mnt) {
A
Al Viro 已提交
2408
				fs->pwd.mnt = mntget(q);
2409
				__mnt_make_longterm(q);
A
Al Viro 已提交
2410
				mnt_make_shortterm(p);
L
Linus Torvalds 已提交
2411 2412 2413
				pwdmnt = p;
			}
		}
2414
		p = next_mnt(p, mnt_ns->root);
L
Linus Torvalds 已提交
2415 2416
		q = next_mnt(q, new_ns->root);
	}
R
Ram Pai 已提交
2417
	up_write(&namespace_sem);
L
Linus Torvalds 已提交
2418 2419

	if (rootmnt)
A
Al Viro 已提交
2420
		mntput(rootmnt);
L
Linus Torvalds 已提交
2421
	if (pwdmnt)
A
Al Viro 已提交
2422
		mntput(pwdmnt);
L
Linus Torvalds 已提交
2423

2424 2425 2426
	return new_ns;
}

2427
struct mnt_namespace *copy_mnt_ns(unsigned long flags, struct mnt_namespace *ns,
2428
		struct fs_struct *new_fs)
2429
{
2430
	struct mnt_namespace *new_ns;
2431

2432
	BUG_ON(!ns);
2433
	get_mnt_ns(ns);
2434 2435

	if (!(flags & CLONE_NEWNS))
2436
		return ns;
2437

2438
	new_ns = dup_mnt_ns(ns, new_fs);
2439

2440
	put_mnt_ns(ns);
2441
	return new_ns;
L
Linus Torvalds 已提交
2442 2443
}

2444 2445 2446 2447
/**
 * create_mnt_ns - creates a private namespace and adds a root filesystem
 * @mnt: pointer to the new root filesystem mountpoint
 */
2448
struct mnt_namespace *create_mnt_ns(struct vfsmount *mnt)
2449 2450 2451 2452 2453 2454
{
	struct mnt_namespace *new_ns;

	new_ns = alloc_mnt_ns();
	if (!IS_ERR(new_ns)) {
		mnt->mnt_ns = new_ns;
2455
		__mnt_make_longterm(mnt);
2456 2457
		new_ns->root = mnt;
		list_add(&new_ns->list, &new_ns->root->mnt_list);
2458 2459
	} else {
		mntput(mnt);
2460 2461 2462
	}
	return new_ns;
}
2463
EXPORT_SYMBOL(create_mnt_ns);
2464

A
Al Viro 已提交
2465 2466 2467
struct dentry *mount_subtree(struct vfsmount *mnt, const char *name)
{
	struct mnt_namespace *ns;
A
Al Viro 已提交
2468
	struct super_block *s;
A
Al Viro 已提交
2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484
	struct path path;
	int err;

	ns = create_mnt_ns(mnt);
	if (IS_ERR(ns))
		return ERR_CAST(ns);

	err = vfs_path_lookup(mnt->mnt_root, mnt,
			name, LOOKUP_FOLLOW|LOOKUP_AUTOMOUNT, &path);

	put_mnt_ns(ns);

	if (err)
		return ERR_PTR(err);

	/* trade a vfsmount reference for active sb one */
A
Al Viro 已提交
2485 2486
	s = path.mnt->mnt_sb;
	atomic_inc(&s->s_active);
A
Al Viro 已提交
2487 2488
	mntput(path.mnt);
	/* lock the sucker */
A
Al Viro 已提交
2489
	down_write(&s->s_umount);
A
Al Viro 已提交
2490 2491 2492 2493 2494
	/* ... and return the root of (sub)tree on it */
	return path.dentry;
}
EXPORT_SYMBOL(mount_subtree);

2495 2496
SYSCALL_DEFINE5(mount, char __user *, dev_name, char __user *, dir_name,
		char __user *, type, unsigned long, flags, void __user *, data)
L
Linus Torvalds 已提交
2497
{
2498 2499 2500 2501
	int ret;
	char *kernel_type;
	char *kernel_dir;
	char *kernel_dev;
L
Linus Torvalds 已提交
2502 2503
	unsigned long data_page;

2504 2505 2506
	ret = copy_mount_string(type, &kernel_type);
	if (ret < 0)
		goto out_type;
L
Linus Torvalds 已提交
2507

2508 2509 2510 2511 2512
	kernel_dir = getname(dir_name);
	if (IS_ERR(kernel_dir)) {
		ret = PTR_ERR(kernel_dir);
		goto out_dir;
	}
L
Linus Torvalds 已提交
2513

2514 2515 2516
	ret = copy_mount_string(dev_name, &kernel_dev);
	if (ret < 0)
		goto out_dev;
L
Linus Torvalds 已提交
2517

2518 2519 2520
	ret = copy_mount_options(data, &data_page);
	if (ret < 0)
		goto out_data;
L
Linus Torvalds 已提交
2521

2522 2523
	ret = do_mount(kernel_dev, kernel_dir, kernel_type, flags,
		(void *) data_page);
L
Linus Torvalds 已提交
2524

2525 2526 2527 2528 2529 2530 2531 2532 2533
	free_page(data_page);
out_data:
	kfree(kernel_dev);
out_dev:
	putname(kernel_dir);
out_dir:
	kfree(kernel_type);
out_type:
	return ret;
L
Linus Torvalds 已提交
2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548
}

/*
 * pivot_root Semantics:
 * Moves the root file system of the current process to the directory put_old,
 * makes new_root as the new root file system of the current process, and sets
 * root/cwd of all processes which had them on the current root to new_root.
 *
 * Restrictions:
 * The new_root and put_old must be directories, and  must not be on the
 * same file  system as the current process root. The put_old  must  be
 * underneath new_root,  i.e. adding a non-zero number of /.. to the string
 * pointed to by put_old must yield the same directory as new_root. No other
 * file system may be mounted on put_old. After all, new_root is a mountpoint.
 *
N
Neil Brown 已提交
2549 2550 2551 2552
 * Also, the current root cannot be on the 'rootfs' (initial ramfs) filesystem.
 * See Documentation/filesystems/ramfs-rootfs-initramfs.txt for alternatives
 * in this situation.
 *
L
Linus Torvalds 已提交
2553 2554 2555 2556 2557 2558 2559 2560
 * Notes:
 *  - we don't move root/cwd if they are not at the root (reason: if something
 *    cared enough to change them, it's probably wrong to force them elsewhere)
 *  - it's okay to pick a root that isn't the root of a file system, e.g.
 *    /nfs/my_root where /nfs is the mount point. It must be a mountpoint,
 *    though, so you may need to say mount --bind /nfs/my_root /nfs/my_root
 *    first.
 */
2561 2562
SYSCALL_DEFINE2(pivot_root, const char __user *, new_root,
		const char __user *, put_old)
L
Linus Torvalds 已提交
2563 2564
{
	struct vfsmount *tmp;
2565
	struct path new, old, parent_path, root_parent, root;
L
Linus Torvalds 已提交
2566 2567 2568 2569 2570
	int error;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

2571
	error = user_path_dir(new_root, &new);
L
Linus Torvalds 已提交
2572 2573 2574
	if (error)
		goto out0;

2575
	error = user_path_dir(put_old, &old);
L
Linus Torvalds 已提交
2576 2577 2578
	if (error)
		goto out1;

2579
	error = security_sb_pivotroot(&old, &new);
2580 2581
	if (error)
		goto out2;
L
Linus Torvalds 已提交
2582

2583
	get_fs_root(current->fs, &root);
2584 2585 2586 2587
	error = lock_mount(&old);
	if (error)
		goto out3;

L
Linus Torvalds 已提交
2588
	error = -EINVAL;
2589 2590
	if (IS_MNT_SHARED(old.mnt) ||
		IS_MNT_SHARED(new.mnt->mnt_parent) ||
2591
		IS_MNT_SHARED(root.mnt->mnt_parent))
2592
		goto out4;
A
Al Viro 已提交
2593
	if (!check_mnt(root.mnt) || !check_mnt(new.mnt))
2594
		goto out4;
L
Linus Torvalds 已提交
2595
	error = -ENOENT;
2596
	if (d_unlinked(new.dentry))
2597
		goto out4;
2598
	if (d_unlinked(old.dentry))
2599
		goto out4;
L
Linus Torvalds 已提交
2600
	error = -EBUSY;
2601 2602
	if (new.mnt == root.mnt ||
	    old.mnt == root.mnt)
2603
		goto out4; /* loop, on the same file system  */
L
Linus Torvalds 已提交
2604
	error = -EINVAL;
2605
	if (root.mnt->mnt_root != root.dentry)
2606
		goto out4; /* not a mountpoint */
2607
	if (root.mnt->mnt_parent == root.mnt)
2608
		goto out4; /* not attached */
2609
	if (new.mnt->mnt_root != new.dentry)
2610
		goto out4; /* not a mountpoint */
2611
	if (new.mnt->mnt_parent == new.mnt)
2612
		goto out4; /* not attached */
2613
	/* make sure we can reach put_old from new_root */
2614 2615
	tmp = old.mnt;
	if (tmp != new.mnt) {
L
Linus Torvalds 已提交
2616 2617
		for (;;) {
			if (tmp->mnt_parent == tmp)
2618
				goto out4; /* already mounted on put_old */
2619
			if (tmp->mnt_parent == new.mnt)
L
Linus Torvalds 已提交
2620 2621 2622
				break;
			tmp = tmp->mnt_parent;
		}
2623
		if (!is_subdir(tmp->mnt_mountpoint, new.dentry))
2624
			goto out4;
2625
	} else if (!is_subdir(old.dentry, new.dentry))
2626
		goto out4;
A
Al Viro 已提交
2627
	br_write_lock(vfsmount_lock);
2628
	detach_mnt(new.mnt, &parent_path);
2629
	detach_mnt(root.mnt, &root_parent);
2630
	/* mount old root on put_old */
2631
	attach_mnt(root.mnt, &old);
2632
	/* mount new_root on / */
2633
	attach_mnt(new.mnt, &root_parent);
2634
	touch_mnt_namespace(current->nsproxy->mnt_ns);
N
Nick Piggin 已提交
2635
	br_write_unlock(vfsmount_lock);
2636
	chroot_fs_refs(&root, &new);
L
Linus Torvalds 已提交
2637
	error = 0;
2638 2639 2640 2641 2642 2643 2644
out4:
	unlock_mount(&old);
	if (!error) {
		path_put(&root_parent);
		path_put(&parent_path);
	}
out3:
2645
	path_put(&root);
2646
out2:
2647
	path_put(&old);
L
Linus Torvalds 已提交
2648
out1:
2649
	path_put(&new);
L
Linus Torvalds 已提交
2650 2651 2652 2653 2654 2655 2656
out0:
	return error;
}

static void __init init_mount_tree(void)
{
	struct vfsmount *mnt;
2657
	struct mnt_namespace *ns;
2658
	struct path root;
L
Linus Torvalds 已提交
2659 2660 2661 2662

	mnt = do_kern_mount("rootfs", 0, "rootfs", NULL);
	if (IS_ERR(mnt))
		panic("Can't create rootfs");
N
Nick Piggin 已提交
2663

2664 2665
	ns = create_mnt_ns(mnt);
	if (IS_ERR(ns))
L
Linus Torvalds 已提交
2666
		panic("Can't allocate initial namespace");
2667 2668 2669 2670

	init_task.nsproxy->mnt_ns = ns;
	get_mnt_ns(ns);

2671 2672 2673 2674 2675
	root.mnt = ns->root;
	root.dentry = ns->root->mnt_root;

	set_fs_pwd(current->fs, &root);
	set_fs_root(current->fs, &root);
L
Linus Torvalds 已提交
2676 2677
}

2678
void __init mnt_init(void)
L
Linus Torvalds 已提交
2679
{
E
Eric Dumazet 已提交
2680
	unsigned u;
2681
	int err;
L
Linus Torvalds 已提交
2682

R
Ram Pai 已提交
2683 2684
	init_rwsem(&namespace_sem);

L
Linus Torvalds 已提交
2685
	mnt_cache = kmem_cache_create("mnt_cache", sizeof(struct vfsmount),
2686
			0, SLAB_HWCACHE_ALIGN | SLAB_PANIC, NULL);
L
Linus Torvalds 已提交
2687

R
Ram Pai 已提交
2688
	mount_hashtable = (struct list_head *)__get_free_page(GFP_ATOMIC);
L
Linus Torvalds 已提交
2689 2690 2691 2692

	if (!mount_hashtable)
		panic("Failed to allocate mount hash table\n");

2693
	printk(KERN_INFO "Mount-cache hash table entries: %lu\n", HASH_SIZE);
E
Eric Dumazet 已提交
2694 2695 2696

	for (u = 0; u < HASH_SIZE; u++)
		INIT_LIST_HEAD(&mount_hashtable[u]);
L
Linus Torvalds 已提交
2697

N
Nick Piggin 已提交
2698 2699
	br_lock_init(vfsmount_lock);

2700 2701 2702
	err = sysfs_init();
	if (err)
		printk(KERN_WARNING "%s: sysfs_init error: %d\n",
2703
			__func__, err);
2704 2705
	fs_kobj = kobject_create_and_add("fs", NULL);
	if (!fs_kobj)
2706
		printk(KERN_WARNING "%s: kobj create error\n", __func__);
L
Linus Torvalds 已提交
2707 2708 2709 2710
	init_rootfs();
	init_mount_tree();
}

2711
void put_mnt_ns(struct mnt_namespace *ns)
L
Linus Torvalds 已提交
2712
{
R
Ram Pai 已提交
2713
	LIST_HEAD(umount_list);
2714

2715
	if (!atomic_dec_and_test(&ns->count))
2716
		return;
R
Ram Pai 已提交
2717
	down_write(&namespace_sem);
N
Nick Piggin 已提交
2718
	br_write_lock(vfsmount_lock);
2719
	umount_tree(ns->root, 0, &umount_list);
N
Nick Piggin 已提交
2720
	br_write_unlock(vfsmount_lock);
R
Ram Pai 已提交
2721
	up_write(&namespace_sem);
R
Ram Pai 已提交
2722
	release_mounts(&umount_list);
2723
	kfree(ns);
L
Linus Torvalds 已提交
2724
}
2725
EXPORT_SYMBOL(put_mnt_ns);
2726 2727 2728

struct vfsmount *kern_mount_data(struct file_system_type *type, void *data)
{
2729 2730 2731 2732 2733 2734 2735 2736 2737 2738
	struct vfsmount *mnt;
	mnt = vfs_kern_mount(type, MS_KERNMOUNT, type->name, data);
	if (!IS_ERR(mnt)) {
		/*
		 * it is a longterm mount, don't release mnt until
		 * we unmount before file sys is unregistered
		*/
		mnt_make_longterm(mnt);
	}
	return mnt;
2739 2740
}
EXPORT_SYMBOL_GPL(kern_mount_data);
2741 2742 2743 2744 2745 2746 2747 2748 2749 2750

void kern_unmount(struct vfsmount *mnt)
{
	/* release long term mount so mount point can be released */
	if (!IS_ERR_OR_NULL(mnt)) {
		mnt_make_shortterm(mnt);
		mntput(mnt);
	}
}
EXPORT_SYMBOL(kern_unmount);
2751 2752 2753 2754 2755

bool our_mnt(struct vfsmount *mnt)
{
	return check_mnt(mnt);
}