super.c 24.3 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 17
/*
 *  linux/fs/super.c
 *
 *  Copyright (C) 1991, 1992  Linus Torvalds
 *
 *  super.c contains code to handle: - mount structures
 *                                   - super-block tables
 *                                   - filesystem drivers list
 *                                   - mount system call
 *                                   - umount system call
 *                                   - ustat system call
 *
 * GK 2/5/95  -  Changed to support mounting the root fs via NFS
 *
 *  Added kerneld support: Jacques Gelinas and Bjorn Ekwall
 *  Added change_root: Werner Almesberger & Hans Lermen, Feb '96
 *  Added options to /proc/mounts:
18
 *    Torbjörn Lindh (torbjorn.lindh@gopta.se), April 14, 1996.
L
Linus Torvalds 已提交
19 20 21 22 23 24 25 26 27 28 29 30
 *  Added devfs support: Richard Gooch <rgooch@atnf.csiro.au>, 13-JAN-1998
 *  Heavily rewritten for 'one fs - one tree' dcache architecture. AV, Mar 2000
 */

#include <linux/module.h>
#include <linux/slab.h>
#include <linux/acct.h>
#include <linux/blkdev.h>
#include <linux/mount.h>
#include <linux/security.h>
#include <linux/writeback.h>		/* for the emergency remount stuff */
#include <linux/idr.h>
I
Ingo Molnar 已提交
31
#include <linux/mutex.h>
32
#include <linux/backing-dev.h>
33
#include <linux/rculist_bl.h>
34
#include "internal.h"
L
Linus Torvalds 已提交
35 36 37 38 39 40 41


LIST_HEAD(super_blocks);
DEFINE_SPINLOCK(sb_lock);

/**
 *	alloc_super	-	create new superblock
42
 *	@type:	filesystem type superblock should belong to
L
Linus Torvalds 已提交
43 44 45 46
 *
 *	Allocates and initializes a new &struct super_block.  alloc_super()
 *	returns a pointer new superblock or %NULL if allocation had failed.
 */
I
Ingo Molnar 已提交
47
static struct super_block *alloc_super(struct file_system_type *type)
L
Linus Torvalds 已提交
48
{
49
	struct super_block *s = kzalloc(sizeof(struct super_block),  GFP_USER);
50
	static const struct super_operations default_op;
L
Linus Torvalds 已提交
51 52 53 54 55 56 57

	if (s) {
		if (security_sb_alloc(s)) {
			kfree(s);
			s = NULL;
			goto out;
		}
N
Nick Piggin 已提交
58 59 60 61 62 63 64 65 66 67 68 69 70 71
#ifdef CONFIG_SMP
		s->s_files = alloc_percpu(struct list_head);
		if (!s->s_files) {
			security_sb_free(s);
			kfree(s);
			s = NULL;
			goto out;
		} else {
			int i;

			for_each_possible_cpu(i)
				INIT_LIST_HEAD(per_cpu_ptr(s->s_files, i));
		}
#else
L
Linus Torvalds 已提交
72
		INIT_LIST_HEAD(&s->s_files);
N
Nick Piggin 已提交
73
#endif
74
		s->s_bdi = &default_backing_dev_info;
L
Linus Torvalds 已提交
75
		INIT_LIST_HEAD(&s->s_instances);
76
		INIT_HLIST_BL_HEAD(&s->s_anon);
L
Linus Torvalds 已提交
77
		INIT_LIST_HEAD(&s->s_inodes);
78
		INIT_LIST_HEAD(&s->s_dentry_lru);
L
Linus Torvalds 已提交
79
		init_rwsem(&s->s_umount);
80
		mutex_init(&s->s_lock);
81
		lockdep_set_class(&s->s_umount, &type->s_umount_key);
I
Ingo Molnar 已提交
82 83 84 85 86 87
		/*
		 * The locking rules for s_lock are up to the
		 * filesystem. For example ext3fs has different
		 * lock ordering than usbfs:
		 */
		lockdep_set_class(&s->s_lock, &type->s_lock_key);
88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103
		/*
		 * sget() can have s_umount recursion.
		 *
		 * When it cannot find a suitable sb, it allocates a new
		 * one (this one), and tries again to find a suitable old
		 * one.
		 *
		 * In case that succeeds, it will acquire the s_umount
		 * lock of the old one. Since these are clearly distrinct
		 * locks, and this object isn't exposed yet, there's no
		 * risk of deadlocks.
		 *
		 * Annotate this by putting this lock in a different
		 * subclass.
		 */
		down_write_nested(&s->s_umount, SINGLE_DEPTH_NESTING);
A
Al Viro 已提交
104
		s->s_count = 1;
L
Linus Torvalds 已提交
105
		atomic_set(&s->s_active, 1);
106
		mutex_init(&s->s_vfs_rename_mutex);
107
		lockdep_set_class(&s->s_vfs_rename_mutex, &type->s_vfs_rename_key);
I
Ingo Molnar 已提交
108 109
		mutex_init(&s->s_dquot.dqio_mutex);
		mutex_init(&s->s_dquot.dqonoff_mutex);
L
Linus Torvalds 已提交
110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127
		init_rwsem(&s->s_dquot.dqptr_sem);
		init_waitqueue_head(&s->s_wait_unfrozen);
		s->s_maxbytes = MAX_NON_LFS;
		s->s_op = &default_op;
		s->s_time_gran = 1000000000;
	}
out:
	return s;
}

/**
 *	destroy_super	-	frees a superblock
 *	@s: superblock to free
 *
 *	Frees a superblock.
 */
static inline void destroy_super(struct super_block *s)
{
N
Nick Piggin 已提交
128 129 130
#ifdef CONFIG_SMP
	free_percpu(s->s_files);
#endif
L
Linus Torvalds 已提交
131
	security_sb_free(s);
M
Miklos Szeredi 已提交
132
	kfree(s->s_subtype);
133
	kfree(s->s_options);
L
Linus Torvalds 已提交
134 135 136 137 138 139
	kfree(s);
}

/* Superblock refcounting  */

/*
A
Al Viro 已提交
140
 * Drop a superblock's refcount.  The caller must hold sb_lock.
L
Linus Torvalds 已提交
141
 */
A
Al Viro 已提交
142
void __put_super(struct super_block *sb)
L
Linus Torvalds 已提交
143 144
{
	if (!--sb->s_count) {
145
		list_del_init(&sb->s_list);
L
Linus Torvalds 已提交
146 147 148 149 150 151 152 153 154 155 156
		destroy_super(sb);
	}
}

/**
 *	put_super	-	drop a temporary reference to superblock
 *	@sb: superblock in question
 *
 *	Drops a temporary reference, frees superblock if there's no
 *	references left.
 */
157
void put_super(struct super_block *sb)
L
Linus Torvalds 已提交
158 159 160 161 162 163 164 165
{
	spin_lock(&sb_lock);
	__put_super(sb);
	spin_unlock(&sb_lock);
}


/**
166
 *	deactivate_locked_super	-	drop an active reference to superblock
L
Linus Torvalds 已提交
167 168
 *	@s: superblock to deactivate
 *
169 170
 *	Drops an active reference to superblock, converting it into a temprory
 *	one if there is no other active references left.  In that case we
L
Linus Torvalds 已提交
171 172
 *	tell fs driver to shut it down and drop the temporary reference we
 *	had just acquired.
173 174
 *
 *	Caller holds exclusive lock on superblock; that lock is released.
L
Linus Torvalds 已提交
175
 */
176
void deactivate_locked_super(struct super_block *s)
L
Linus Torvalds 已提交
177 178
{
	struct file_system_type *fs = s->s_type;
A
Al Viro 已提交
179
	if (atomic_dec_and_test(&s->s_active)) {
L
Linus Torvalds 已提交
180
		fs->kill_sb(s);
181 182 183 184 185
		/*
		 * We need to call rcu_barrier so all the delayed rcu free
		 * inodes are flushed before we release the fs module.
		 */
		rcu_barrier();
L
Linus Torvalds 已提交
186 187
		put_filesystem(fs);
		put_super(s);
188 189
	} else {
		up_write(&s->s_umount);
L
Linus Torvalds 已提交
190 191 192
	}
}

193
EXPORT_SYMBOL(deactivate_locked_super);
L
Linus Torvalds 已提交
194

A
Al Viro 已提交
195
/**
196
 *	deactivate_super	-	drop an active reference to superblock
A
Al Viro 已提交
197 198
 *	@s: superblock to deactivate
 *
199 200 201
 *	Variant of deactivate_locked_super(), except that superblock is *not*
 *	locked by caller.  If we are going to drop the final active reference,
 *	lock will be acquired prior to that.
A
Al Viro 已提交
202
 */
203
void deactivate_super(struct super_block *s)
A
Al Viro 已提交
204
{
205 206 207
        if (!atomic_add_unless(&s->s_active, -1, 1)) {
		down_write(&s->s_umount);
		deactivate_locked_super(s);
A
Al Viro 已提交
208 209 210
	}
}

211
EXPORT_SYMBOL(deactivate_super);
A
Al Viro 已提交
212

L
Linus Torvalds 已提交
213 214 215 216 217 218 219 220 221 222 223
/**
 *	grab_super - acquire an active reference
 *	@s: reference we are trying to make active
 *
 *	Tries to acquire an active reference.  grab_super() is used when we
 * 	had just found a superblock in super_blocks or fs_type->fs_supers
 *	and want to turn it into a full-blown active reference.  grab_super()
 *	is called with sb_lock held and drops it.  Returns 1 in case of
 *	success, 0 if we had failed (superblock contents was already dead or
 *	dying when grab_super() had been called).
 */
224
static int grab_super(struct super_block *s) __releases(sb_lock)
L
Linus Torvalds 已提交
225
{
A
Al Viro 已提交
226 227 228 229 230
	if (atomic_inc_not_zero(&s->s_active)) {
		spin_unlock(&sb_lock);
		return 1;
	}
	/* it's going away */
L
Linus Torvalds 已提交
231 232
	s->s_count++;
	spin_unlock(&sb_lock);
233
	/* wait for it to die */
L
Linus Torvalds 已提交
234 235 236 237 238 239
	down_write(&s->s_umount);
	up_write(&s->s_umount);
	put_super(s);
	return 0;
}

240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257
/*
 * Superblock locking.  We really ought to get rid of these two.
 */
void lock_super(struct super_block * sb)
{
	get_fs_excl();
	mutex_lock(&sb->s_lock);
}

void unlock_super(struct super_block * sb)
{
	put_fs_excl();
	mutex_unlock(&sb->s_lock);
}

EXPORT_SYMBOL(lock_super);
EXPORT_SYMBOL(unlock_super);

L
Linus Torvalds 已提交
258 259 260 261 262 263 264 265 266
/**
 *	generic_shutdown_super	-	common helper for ->kill_sb()
 *	@sb: superblock to kill
 *
 *	generic_shutdown_super() does all fs-independent work on superblock
 *	shutdown.  Typical ->kill_sb() should pick all fs-specific objects
 *	that need destruction out of superblock, call generic_shutdown_super()
 *	and release aforementioned objects.  Note: dentries and inodes _are_
 *	taken care of and do not need specific handling.
267 268 269 270
 *
 *	Upon calling this function, the filesystem may no longer alter or
 *	rearrange the set of dentries belonging to this super_block, nor may it
 *	change the attachments of dentries to inodes.
L
Linus Torvalds 已提交
271 272 273
 */
void generic_shutdown_super(struct super_block *sb)
{
274
	const struct super_operations *sop = sb->s_op;
L
Linus Torvalds 已提交
275

276

277 278
	if (sb->s_root) {
		shrink_dcache_for_umount(sb);
279
		sync_filesystem(sb);
280
		get_fs_excl();
L
Linus Torvalds 已提交
281
		sb->s_flags &= ~MS_ACTIVE;
282

A
Al Viro 已提交
283 284 285
		fsnotify_unmount_inodes(&sb->s_inodes);

		evict_inodes(sb);
L
Linus Torvalds 已提交
286 287 288 289

		if (sop->put_super)
			sop->put_super(sb);

A
Al Viro 已提交
290
		if (!list_empty(&sb->s_inodes)) {
291 292 293
			printk("VFS: Busy inodes after unmount of %s. "
			   "Self-destruct in 5 seconds.  Have a nice day...\n",
			   sb->s_id);
L
Linus Torvalds 已提交
294
		}
295
		put_fs_excl();
L
Linus Torvalds 已提交
296 297 298
	}
	spin_lock(&sb_lock);
	/* should be initialized for __put_super_and_need_restart() */
299
	list_del_init(&sb->s_instances);
L
Linus Torvalds 已提交
300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318
	spin_unlock(&sb_lock);
	up_write(&sb->s_umount);
}

EXPORT_SYMBOL(generic_shutdown_super);

/**
 *	sget	-	find or create a superblock
 *	@type:	filesystem type superblock should belong to
 *	@test:	comparison callback
 *	@set:	setup callback
 *	@data:	argument to each of them
 */
struct super_block *sget(struct file_system_type *type,
			int (*test)(struct super_block *,void *),
			int (*set)(struct super_block *,void *),
			void *data)
{
	struct super_block *s = NULL;
319
	struct super_block *old;
L
Linus Torvalds 已提交
320 321 322 323
	int err;

retry:
	spin_lock(&sb_lock);
324 325 326 327 328 329
	if (test) {
		list_for_each_entry(old, &type->fs_supers, s_instances) {
			if (!test(old, data))
				continue;
			if (!grab_super(old))
				goto retry;
L
Li Zefan 已提交
330 331
			if (s) {
				up_write(&s->s_umount);
332
				destroy_super(s);
A
Al Viro 已提交
333
				s = NULL;
L
Li Zefan 已提交
334
			}
335
			down_write(&old->s_umount);
A
Al Viro 已提交
336 337 338 339
			if (unlikely(!(old->s_flags & MS_BORN))) {
				deactivate_locked_super(old);
				goto retry;
			}
340 341
			return old;
		}
L
Linus Torvalds 已提交
342 343 344
	}
	if (!s) {
		spin_unlock(&sb_lock);
I
Ingo Molnar 已提交
345
		s = alloc_super(type);
L
Linus Torvalds 已提交
346 347 348 349 350 351 352 353
		if (!s)
			return ERR_PTR(-ENOMEM);
		goto retry;
	}
		
	err = set(s, data);
	if (err) {
		spin_unlock(&sb_lock);
L
Li Zefan 已提交
354
		up_write(&s->s_umount);
L
Linus Torvalds 已提交
355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376
		destroy_super(s);
		return ERR_PTR(err);
	}
	s->s_type = type;
	strlcpy(s->s_id, type->name, sizeof(s->s_id));
	list_add_tail(&s->s_list, &super_blocks);
	list_add(&s->s_instances, &type->fs_supers);
	spin_unlock(&sb_lock);
	get_filesystem(type);
	return s;
}

EXPORT_SYMBOL(sget);

void drop_super(struct super_block *sb)
{
	up_read(&sb->s_umount);
	put_super(sb);
}

EXPORT_SYMBOL(drop_super);

C
Christoph Hellwig 已提交
377 378 379 380 381 382 383 384
/**
 * sync_supers - helper for periodic superblock writeback
 *
 * Call the write_super method if present on all dirty superblocks in
 * the system.  This is for the periodic writeback used by most older
 * filesystems.  For data integrity superblock writeback use
 * sync_filesystems() instead.
 *
L
Linus Torvalds 已提交
385 386 387 388 389 390
 * Note: check the dirty flag before waiting, so we don't
 * hold up the sync while mounting a device. (The newly
 * mounted device won't need syncing.)
 */
void sync_supers(void)
{
391
	struct super_block *sb, *p = NULL;
392

L
Linus Torvalds 已提交
393
	spin_lock(&sb_lock);
394
	list_for_each_entry(sb, &super_blocks, s_list) {
395 396
		if (list_empty(&sb->s_instances))
			continue;
C
Christoph Hellwig 已提交
397
		if (sb->s_op->write_super && sb->s_dirt) {
L
Linus Torvalds 已提交
398 399
			sb->s_count++;
			spin_unlock(&sb_lock);
C
Christoph Hellwig 已提交
400

L
Linus Torvalds 已提交
401
			down_read(&sb->s_umount);
C
Christoph Hellwig 已提交
402 403
			if (sb->s_root && sb->s_dirt)
				sb->s_op->write_super(sb);
404
			up_read(&sb->s_umount);
C
Christoph Hellwig 已提交
405

406
			spin_lock(&sb_lock);
407 408 409
			if (p)
				__put_super(p);
			p = sb;
410 411
		}
	}
412 413
	if (p)
		__put_super(p);
L
Linus Torvalds 已提交
414 415 416
	spin_unlock(&sb_lock);
}

A
Al Viro 已提交
417 418 419 420 421 422 423 424 425 426
/**
 *	iterate_supers - call function for all active superblocks
 *	@f: function to call
 *	@arg: argument to pass to it
 *
 *	Scans the superblock list and calls given function, passing it
 *	locked superblock and given argument.
 */
void iterate_supers(void (*f)(struct super_block *, void *), void *arg)
{
427
	struct super_block *sb, *p = NULL;
A
Al Viro 已提交
428 429

	spin_lock(&sb_lock);
430
	list_for_each_entry(sb, &super_blocks, s_list) {
A
Al Viro 已提交
431 432 433 434 435 436 437 438 439 440 441
		if (list_empty(&sb->s_instances))
			continue;
		sb->s_count++;
		spin_unlock(&sb_lock);

		down_read(&sb->s_umount);
		if (sb->s_root)
			f(sb, arg);
		up_read(&sb->s_umount);

		spin_lock(&sb_lock);
442 443 444
		if (p)
			__put_super(p);
		p = sb;
A
Al Viro 已提交
445
	}
446 447
	if (p)
		__put_super(p);
A
Al Viro 已提交
448 449 450
	spin_unlock(&sb_lock);
}

L
Linus Torvalds 已提交
451 452 453 454 455 456 457 458
/**
 *	get_super - get the superblock of a device
 *	@bdev: device to get the superblock for
 *	
 *	Scans the superblock list and finds the superblock of the file system
 *	mounted on the device given. %NULL is returned if no match is found.
 */

459
struct super_block *get_super(struct block_device *bdev)
L
Linus Torvalds 已提交
460
{
461 462
	struct super_block *sb;

L
Linus Torvalds 已提交
463 464
	if (!bdev)
		return NULL;
465

L
Linus Torvalds 已提交
466
	spin_lock(&sb_lock);
467 468
rescan:
	list_for_each_entry(sb, &super_blocks, s_list) {
469 470
		if (list_empty(&sb->s_instances))
			continue;
471 472
		if (sb->s_bdev == bdev) {
			sb->s_count++;
L
Linus Torvalds 已提交
473
			spin_unlock(&sb_lock);
474
			down_read(&sb->s_umount);
475
			/* still alive? */
476 477 478
			if (sb->s_root)
				return sb;
			up_read(&sb->s_umount);
479
			/* nope, got unmounted */
480
			spin_lock(&sb_lock);
481 482
			__put_super(sb);
			goto rescan;
L
Linus Torvalds 已提交
483 484 485 486 487 488 489
		}
	}
	spin_unlock(&sb_lock);
	return NULL;
}

EXPORT_SYMBOL(get_super);
490 491 492 493 494 495 496

/**
 * get_active_super - get an active reference to the superblock of a device
 * @bdev: device to get the superblock for
 *
 * Scans the superblock list and finds the superblock of the file system
 * mounted on the device given.  Returns the superblock with an active
497
 * reference or %NULL if none was found.
498 499 500 501 502 503 504 505
 */
struct super_block *get_active_super(struct block_device *bdev)
{
	struct super_block *sb;

	if (!bdev)
		return NULL;

A
Al Viro 已提交
506
restart:
507 508
	spin_lock(&sb_lock);
	list_for_each_entry(sb, &super_blocks, s_list) {
509 510
		if (list_empty(&sb->s_instances))
			continue;
A
Al Viro 已提交
511 512 513 514 515 516
		if (sb->s_bdev == bdev) {
			if (grab_super(sb)) /* drops sb_lock */
				return sb;
			else
				goto restart;
		}
517 518 519 520
	}
	spin_unlock(&sb_lock);
	return NULL;
}
L
Linus Torvalds 已提交
521
 
522
struct super_block *user_get_super(dev_t dev)
L
Linus Torvalds 已提交
523
{
524
	struct super_block *sb;
L
Linus Torvalds 已提交
525 526

	spin_lock(&sb_lock);
527 528
rescan:
	list_for_each_entry(sb, &super_blocks, s_list) {
529 530
		if (list_empty(&sb->s_instances))
			continue;
531 532
		if (sb->s_dev ==  dev) {
			sb->s_count++;
L
Linus Torvalds 已提交
533
			spin_unlock(&sb_lock);
534
			down_read(&sb->s_umount);
535
			/* still alive? */
536 537 538
			if (sb->s_root)
				return sb;
			up_read(&sb->s_umount);
539
			/* nope, got unmounted */
540
			spin_lock(&sb_lock);
541 542
			__put_super(sb);
			goto rescan;
L
Linus Torvalds 已提交
543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560
		}
	}
	spin_unlock(&sb_lock);
	return NULL;
}

/**
 *	do_remount_sb - asks filesystem to change mount options.
 *	@sb:	superblock in question
 *	@flags:	numeric part of options
 *	@data:	the rest of options
 *      @force: whether or not to force the change
 *
 *	Alters the mount options of a mounted file system.
 */
int do_remount_sb(struct super_block *sb, int flags, void *data, int force)
{
	int retval;
561
	int remount_ro;
562 563 564 565

	if (sb->s_frozen != SB_UNFROZEN)
		return -EBUSY;

566
#ifdef CONFIG_BLOCK
L
Linus Torvalds 已提交
567 568
	if (!(flags & MS_RDONLY) && bdev_read_only(sb->s_bdev))
		return -EACCES;
569
#endif
570

L
Linus Torvalds 已提交
571 572 573
	if (flags & MS_RDONLY)
		acct_auto_close(sb);
	shrink_dcache_sb(sb);
574
	sync_filesystem(sb);
L
Linus Torvalds 已提交
575

576 577
	remount_ro = (flags & MS_RDONLY) && !(sb->s_flags & MS_RDONLY);

L
Linus Torvalds 已提交
578 579
	/* If we are remounting RDONLY and current sb is read/write,
	   make sure there are no rw files opened */
580
	if (remount_ro) {
L
Linus Torvalds 已提交
581 582
		if (force)
			mark_files_ro(sb);
583
		else if (!fs_may_remount_ro(sb))
L
Linus Torvalds 已提交
584 585 586 587 588
			return -EBUSY;
	}

	if (sb->s_op->remount_fs) {
		retval = sb->s_op->remount_fs(sb, &flags, data);
589
		if (retval)
L
Linus Torvalds 已提交
590 591 592
			return retval;
	}
	sb->s_flags = (sb->s_flags & ~MS_RMT_MASK) | (flags & MS_RMT_MASK);
593

594 595 596 597 598 599 600 601 602 603
	/*
	 * Some filesystems modify their metadata via some other path than the
	 * bdev buffer cache (eg. use a private mapping, or directories in
	 * pagecache, etc). Also file data modifications go via their own
	 * mappings. So If we try to mount readonly then copy the filesystem
	 * from bdev, we could get stale data, so invalidate it to give a best
	 * effort at coherency.
	 */
	if (remount_ro && sb->s_bdev)
		invalidate_bdev(sb->s_bdev);
L
Linus Torvalds 已提交
604 605 606
	return 0;
}

607
static void do_emergency_remount(struct work_struct *work)
L
Linus Torvalds 已提交
608
{
609
	struct super_block *sb, *p = NULL;
L
Linus Torvalds 已提交
610 611

	spin_lock(&sb_lock);
612
	list_for_each_entry(sb, &super_blocks, s_list) {
613 614
		if (list_empty(&sb->s_instances))
			continue;
L
Linus Torvalds 已提交
615 616
		sb->s_count++;
		spin_unlock(&sb_lock);
617
		down_write(&sb->s_umount);
L
Linus Torvalds 已提交
618 619 620 621 622 623
		if (sb->s_root && sb->s_bdev && !(sb->s_flags & MS_RDONLY)) {
			/*
			 * What lock protects sb->s_flags??
			 */
			do_remount_sb(sb, MS_RDONLY, NULL, 1);
		}
624
		up_write(&sb->s_umount);
L
Linus Torvalds 已提交
625
		spin_lock(&sb_lock);
626 627 628
		if (p)
			__put_super(p);
		p = sb;
L
Linus Torvalds 已提交
629
	}
630 631
	if (p)
		__put_super(p);
L
Linus Torvalds 已提交
632
	spin_unlock(&sb_lock);
633
	kfree(work);
L
Linus Torvalds 已提交
634 635 636 637 638
	printk("Emergency Remount complete\n");
}

void emergency_remount(void)
{
639 640 641 642 643 644 645
	struct work_struct *work;

	work = kmalloc(sizeof(*work), GFP_ATOMIC);
	if (work) {
		INIT_WORK(work, do_emergency_remount);
		schedule_work(work);
	}
L
Linus Torvalds 已提交
646 647 648 649 650 651 652
}

/*
 * Unnamed block devices are dummy devices used by virtual
 * filesystems which don't use real block-devices.  -- jrs
 */

653
static DEFINE_IDA(unnamed_dev_ida);
L
Linus Torvalds 已提交
654
static DEFINE_SPINLOCK(unnamed_dev_lock);/* protects the above */
655
static int unnamed_dev_start = 0; /* don't bother trying below it */
L
Linus Torvalds 已提交
656 657 658 659 660 661 662

int set_anon_super(struct super_block *s, void *data)
{
	int dev;
	int error;

 retry:
663
	if (ida_pre_get(&unnamed_dev_ida, GFP_ATOMIC) == 0)
L
Linus Torvalds 已提交
664 665
		return -ENOMEM;
	spin_lock(&unnamed_dev_lock);
666
	error = ida_get_new_above(&unnamed_dev_ida, unnamed_dev_start, &dev);
667 668
	if (!error)
		unnamed_dev_start = dev + 1;
L
Linus Torvalds 已提交
669 670 671 672 673 674 675 676 677
	spin_unlock(&unnamed_dev_lock);
	if (error == -EAGAIN)
		/* We raced and lost with another CPU. */
		goto retry;
	else if (error)
		return -EAGAIN;

	if ((dev & MAX_ID_MASK) == (1 << MINORBITS)) {
		spin_lock(&unnamed_dev_lock);
678
		ida_remove(&unnamed_dev_ida, dev);
679 680
		if (unnamed_dev_start > dev)
			unnamed_dev_start = dev;
L
Linus Torvalds 已提交
681 682 683 684
		spin_unlock(&unnamed_dev_lock);
		return -EMFILE;
	}
	s->s_dev = MKDEV(0, dev & MINORMASK);
J
Jörn Engel 已提交
685
	s->s_bdi = &noop_backing_dev_info;
L
Linus Torvalds 已提交
686 687 688 689 690 691 692 693 694 695 696
	return 0;
}

EXPORT_SYMBOL(set_anon_super);

void kill_anon_super(struct super_block *sb)
{
	int slot = MINOR(sb->s_dev);

	generic_shutdown_super(sb);
	spin_lock(&unnamed_dev_lock);
697
	ida_remove(&unnamed_dev_ida, slot);
698 699
	if (slot < unnamed_dev_start)
		unnamed_dev_start = slot;
L
Linus Torvalds 已提交
700 701 702 703 704 705 706 707 708 709 710 711 712 713
	spin_unlock(&unnamed_dev_lock);
}

EXPORT_SYMBOL(kill_anon_super);

void kill_litter_super(struct super_block *sb)
{
	if (sb->s_root)
		d_genocide(sb->s_root);
	kill_anon_super(sb);
}

EXPORT_SYMBOL(kill_litter_super);

714 715 716 717 718 719 720 721 722 723 724
static int ns_test_super(struct super_block *sb, void *data)
{
	return sb->s_fs_info == data;
}

static int ns_set_super(struct super_block *sb, void *data)
{
	sb->s_fs_info = data;
	return set_anon_super(sb, NULL);
}

A
Al Viro 已提交
725 726
struct dentry *mount_ns(struct file_system_type *fs_type, int flags,
	void *data, int (*fill_super)(struct super_block *, void *, int))
727 728 729 730 731
{
	struct super_block *sb;

	sb = sget(fs_type, ns_test_super, ns_set_super, data);
	if (IS_ERR(sb))
A
Al Viro 已提交
732
		return ERR_CAST(sb);
733 734 735 736 737 738

	if (!sb->s_root) {
		int err;
		sb->s_flags = flags;
		err = fill_super(sb, data, flags & MS_SILENT ? 1 : 0);
		if (err) {
A
Al Viro 已提交
739
			deactivate_locked_super(sb);
A
Al Viro 已提交
740
			return ERR_PTR(err);
741 742 743 744 745
		}

		sb->s_flags |= MS_ACTIVE;
	}

A
Al Viro 已提交
746
	return dget(sb->s_root);
747 748
}

A
Al Viro 已提交
749
EXPORT_SYMBOL(mount_ns);
750

751
#ifdef CONFIG_BLOCK
L
Linus Torvalds 已提交
752 753 754 755
static int set_bdev_super(struct super_block *s, void *data)
{
	s->s_bdev = data;
	s->s_dev = s->s_bdev->bd_dev;
J
Jens Axboe 已提交
756 757 758 759 760 761

	/*
	 * We set the bdi here to the queue backing, file systems can
	 * overwrite this in ->fill_super()
	 */
	s->s_bdi = &bdev_get_queue(s->s_bdev)->backing_dev_info;
L
Linus Torvalds 已提交
762 763 764 765 766 767 768 769
	return 0;
}

static int test_bdev_super(struct super_block *s, void *data)
{
	return (void *)s->s_bdev == data;
}

A
Al Viro 已提交
770
struct dentry *mount_bdev(struct file_system_type *fs_type,
L
Linus Torvalds 已提交
771
	int flags, const char *dev_name, void *data,
A
Al Viro 已提交
772
	int (*fill_super)(struct super_block *, void *, int))
L
Linus Torvalds 已提交
773 774 775
{
	struct block_device *bdev;
	struct super_block *s;
776
	fmode_t mode = FMODE_READ | FMODE_EXCL;
L
Linus Torvalds 已提交
777 778
	int error = 0;

779 780 781
	if (!(flags & MS_RDONLY))
		mode |= FMODE_WRITE;

782
	bdev = blkdev_get_by_path(dev_name, mode, fs_type);
L
Linus Torvalds 已提交
783
	if (IS_ERR(bdev))
A
Al Viro 已提交
784
		return ERR_CAST(bdev);
L
Linus Torvalds 已提交
785 786 787 788 789 790

	/*
	 * once the super is inserted into the list by sget, s_umount
	 * will protect the lockfs code from trying to start a snapshot
	 * while we are mounting
	 */
791 792 793 794 795 796
	mutex_lock(&bdev->bd_fsfreeze_mutex);
	if (bdev->bd_fsfreeze_count > 0) {
		mutex_unlock(&bdev->bd_fsfreeze_mutex);
		error = -EBUSY;
		goto error_bdev;
	}
L
Linus Torvalds 已提交
797
	s = sget(fs_type, test_bdev_super, set_bdev_super, bdev);
798
	mutex_unlock(&bdev->bd_fsfreeze_mutex);
L
Linus Torvalds 已提交
799
	if (IS_ERR(s))
800
		goto error_s;
L
Linus Torvalds 已提交
801 802 803

	if (s->s_root) {
		if ((flags ^ s->s_flags) & MS_RDONLY) {
A
Al Viro 已提交
804
			deactivate_locked_super(s);
805 806
			error = -EBUSY;
			goto error_bdev;
L
Linus Torvalds 已提交
807
		}
808

809 810
		/*
		 * s_umount nests inside bd_mutex during
811 812 813 814
		 * __invalidate_device().  blkdev_put() acquires
		 * bd_mutex and can't be called under s_umount.  Drop
		 * s_umount temporarily.  This is safe as we're
		 * holding an active reference.
815 816
		 */
		up_write(&s->s_umount);
817
		blkdev_put(bdev, mode);
818
		down_write(&s->s_umount);
L
Linus Torvalds 已提交
819 820 821 822
	} else {
		char b[BDEVNAME_SIZE];

		s->s_flags = flags;
823
		s->s_mode = mode;
L
Linus Torvalds 已提交
824
		strlcpy(s->s_id, bdevname(bdev, b), sizeof(s->s_id));
825
		sb_set_blocksize(s, block_size(bdev));
826
		error = fill_super(s, data, flags & MS_SILENT ? 1 : 0);
L
Linus Torvalds 已提交
827
		if (error) {
A
Al Viro 已提交
828
			deactivate_locked_super(s);
829
			goto error;
830
		}
831 832

		s->s_flags |= MS_ACTIVE;
833
		bdev->bd_super = s;
L
Linus Torvalds 已提交
834 835
	}

A
Al Viro 已提交
836
	return dget(s->s_root);
L
Linus Torvalds 已提交
837

838 839 840
error_s:
	error = PTR_ERR(s);
error_bdev:
841
	blkdev_put(bdev, mode);
842
error:
A
Al Viro 已提交
843 844 845 846
	return ERR_PTR(error);
}
EXPORT_SYMBOL(mount_bdev);

L
Linus Torvalds 已提交
847 848 849
void kill_block_super(struct super_block *sb)
{
	struct block_device *bdev = sb->s_bdev;
850
	fmode_t mode = sb->s_mode;
L
Linus Torvalds 已提交
851

852
	bdev->bd_super = NULL;
L
Linus Torvalds 已提交
853 854
	generic_shutdown_super(sb);
	sync_blockdev(bdev);
855
	WARN_ON_ONCE(!(mode & FMODE_EXCL));
856
	blkdev_put(bdev, mode | FMODE_EXCL);
L
Linus Torvalds 已提交
857 858 859
}

EXPORT_SYMBOL(kill_block_super);
860
#endif
L
Linus Torvalds 已提交
861

A
Al Viro 已提交
862
struct dentry *mount_nodev(struct file_system_type *fs_type,
L
Linus Torvalds 已提交
863
	int flags, void *data,
A
Al Viro 已提交
864
	int (*fill_super)(struct super_block *, void *, int))
L
Linus Torvalds 已提交
865 866 867 868 869
{
	int error;
	struct super_block *s = sget(fs_type, NULL, set_anon_super, NULL);

	if (IS_ERR(s))
A
Al Viro 已提交
870
		return ERR_CAST(s);
L
Linus Torvalds 已提交
871 872 873

	s->s_flags = flags;

874
	error = fill_super(s, data, flags & MS_SILENT ? 1 : 0);
L
Linus Torvalds 已提交
875
	if (error) {
A
Al Viro 已提交
876
		deactivate_locked_super(s);
A
Al Viro 已提交
877
		return ERR_PTR(error);
L
Linus Torvalds 已提交
878 879
	}
	s->s_flags |= MS_ACTIVE;
A
Al Viro 已提交
880
	return dget(s->s_root);
L
Linus Torvalds 已提交
881
}
A
Al Viro 已提交
882 883
EXPORT_SYMBOL(mount_nodev);

L
Linus Torvalds 已提交
884 885 886 887 888
static int compare_single(struct super_block *s, void *p)
{
	return 1;
}

A
Al Viro 已提交
889
struct dentry *mount_single(struct file_system_type *fs_type,
L
Linus Torvalds 已提交
890
	int flags, void *data,
A
Al Viro 已提交
891
	int (*fill_super)(struct super_block *, void *, int))
L
Linus Torvalds 已提交
892 893 894 895 896 897
{
	struct super_block *s;
	int error;

	s = sget(fs_type, compare_single, set_anon_super, NULL);
	if (IS_ERR(s))
A
Al Viro 已提交
898
		return ERR_CAST(s);
L
Linus Torvalds 已提交
899 900
	if (!s->s_root) {
		s->s_flags = flags;
901
		error = fill_super(s, data, flags & MS_SILENT ? 1 : 0);
L
Linus Torvalds 已提交
902
		if (error) {
A
Al Viro 已提交
903
			deactivate_locked_super(s);
A
Al Viro 已提交
904
			return ERR_PTR(error);
L
Linus Torvalds 已提交
905 906
		}
		s->s_flags |= MS_ACTIVE;
907 908
	} else {
		do_remount_sb(s, flags, data, 0);
L
Linus Torvalds 已提交
909
	}
A
Al Viro 已提交
910 911 912 913
	return dget(s->s_root);
}
EXPORT_SYMBOL(mount_single);

914 915
struct dentry *
mount_fs(struct file_system_type *type, int flags, const char *name, void *data)
L
Linus Torvalds 已提交
916
{
A
Al Viro 已提交
917
	struct dentry *root;
918
	struct super_block *sb;
L
Linus Torvalds 已提交
919
	char *secdata = NULL;
920
	int error = -ENOMEM;
A
Al Viro 已提交
921

922
	if (data && !(type->fs_flags & FS_BINARY_MOUNTDATA)) {
L
Linus Torvalds 已提交
923
		secdata = alloc_secdata();
924
		if (!secdata)
925
			goto out;
L
Linus Torvalds 已提交
926

927
		error = security_sb_copy_data(data, secdata);
928
		if (error)
L
Linus Torvalds 已提交
929 930 931
			goto out_free_secdata;
	}

A
Al Viro 已提交
932 933 934 935
	root = type->mount(type, flags, name, data);
	if (IS_ERR(root)) {
		error = PTR_ERR(root);
		goto out_free_secdata;
A
Al Viro 已提交
936
	}
937 938 939
	sb = root->d_sb;
	BUG_ON(!sb);
	WARN_ON(!sb->s_bdi);
940
	WARN_ON(sb->s_bdi == &default_backing_dev_info);
941
	sb->s_flags |= MS_BORN;
942

943
	error = security_sb_kern_mount(sb, flags, secdata);
J
Jörn Engel 已提交
944 945
	if (error)
		goto out_sb;
946

947 948 949 950 951 952 953
	/*
	 * filesystems should never set s_maxbytes larger than MAX_LFS_FILESIZE
	 * but s_maxbytes was an unsigned long long for many releases. Throw
	 * this warning for a little while to try and catch filesystems that
	 * violate this rule. This warning should be either removed or
	 * converted to a BUG() in 2.6.34.
	 */
954 955
	WARN((sb->s_maxbytes < 0), "%s set sb->s_maxbytes to "
		"negative value (%lld)\n", type->name, sb->s_maxbytes);
956

957
	up_write(&sb->s_umount);
958
	free_secdata(secdata);
959
	return root;
L
Linus Torvalds 已提交
960
out_sb:
961 962
	dput(root);
	deactivate_locked_super(sb);
L
Linus Torvalds 已提交
963 964 965
out_free_secdata:
	free_secdata(secdata);
out:
966
	return ERR_PTR(error);
L
Linus Torvalds 已提交
967 968
}

969
/**
R
Randy Dunlap 已提交
970 971
 * freeze_super - lock the filesystem and force it into a consistent state
 * @sb: the super to lock
972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057
 *
 * Syncs the super to make sure the filesystem is consistent and calls the fs's
 * freeze_fs.  Subsequent calls to this without first thawing the fs will return
 * -EBUSY.
 */
int freeze_super(struct super_block *sb)
{
	int ret;

	atomic_inc(&sb->s_active);
	down_write(&sb->s_umount);
	if (sb->s_frozen) {
		deactivate_locked_super(sb);
		return -EBUSY;
	}

	if (sb->s_flags & MS_RDONLY) {
		sb->s_frozen = SB_FREEZE_TRANS;
		smp_wmb();
		up_write(&sb->s_umount);
		return 0;
	}

	sb->s_frozen = SB_FREEZE_WRITE;
	smp_wmb();

	sync_filesystem(sb);

	sb->s_frozen = SB_FREEZE_TRANS;
	smp_wmb();

	sync_blockdev(sb->s_bdev);
	if (sb->s_op->freeze_fs) {
		ret = sb->s_op->freeze_fs(sb);
		if (ret) {
			printk(KERN_ERR
				"VFS:Filesystem freeze failed\n");
			sb->s_frozen = SB_UNFROZEN;
			deactivate_locked_super(sb);
			return ret;
		}
	}
	up_write(&sb->s_umount);
	return 0;
}
EXPORT_SYMBOL(freeze_super);

/**
 * thaw_super -- unlock filesystem
 * @sb: the super to thaw
 *
 * Unlocks the filesystem and marks it writeable again after freeze_super().
 */
int thaw_super(struct super_block *sb)
{
	int error;

	down_write(&sb->s_umount);
	if (sb->s_frozen == SB_UNFROZEN) {
		up_write(&sb->s_umount);
		return -EINVAL;
	}

	if (sb->s_flags & MS_RDONLY)
		goto out;

	if (sb->s_op->unfreeze_fs) {
		error = sb->s_op->unfreeze_fs(sb);
		if (error) {
			printk(KERN_ERR
				"VFS:Filesystem thaw failed\n");
			sb->s_frozen = SB_FREEZE_TRANS;
			up_write(&sb->s_umount);
			return error;
		}
	}

out:
	sb->s_frozen = SB_UNFROZEN;
	smp_wmb();
	wake_up(&sb->s_wait_unfrozen);
	deactivate_locked_super(sb);

	return 0;
}
EXPORT_SYMBOL(thaw_super);