block_dev.c 41.9 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13
/*
 *  linux/fs/block_dev.c
 *
 *  Copyright (C) 1991, 1992  Linus Torvalds
 *  Copyright (C) 2001  Andrea Arcangeli <andrea@suse.de> SuSE
 */

#include <linux/init.h>
#include <linux/mm.h>
#include <linux/fcntl.h>
#include <linux/slab.h>
#include <linux/kmod.h>
#include <linux/major.h>
14
#include <linux/device_cgroup.h>
L
Linus Torvalds 已提交
15 16 17 18
#include <linux/highmem.h>
#include <linux/blkdev.h>
#include <linux/module.h>
#include <linux/blkpg.h>
19
#include <linux/magic.h>
L
Linus Torvalds 已提交
20
#include <linux/buffer_head.h>
A
Al Viro 已提交
21
#include <linux/swap.h>
N
Nick Piggin 已提交
22
#include <linux/pagevec.h>
23
#include <linux/writeback.h>
L
Linus Torvalds 已提交
24 25 26 27
#include <linux/mpage.h>
#include <linux/mount.h>
#include <linux/uio.h>
#include <linux/namei.h>
28
#include <linux/log2.h>
A
Al Viro 已提交
29
#include <linux/cleancache.h>
30
#include <linux/aio.h>
L
Linus Torvalds 已提交
31
#include <asm/uaccess.h>
32
#include "internal.h"
L
Linus Torvalds 已提交
33 34 35 36 37 38

struct bdev_inode {
	struct block_device bdev;
	struct inode vfs_inode;
};

A
Adrian Bunk 已提交
39 40
static const struct address_space_operations def_blk_aops;

L
Linus Torvalds 已提交
41 42 43 44 45 46 47 48 49 50 51
static inline struct bdev_inode *BDEV_I(struct inode *inode)
{
	return container_of(inode, struct bdev_inode, vfs_inode);
}

inline struct block_device *I_BDEV(struct inode *inode)
{
	return &BDEV_I(inode)->bdev;
}
EXPORT_SYMBOL(I_BDEV);

52
/*
53 54 55
 * Move the inode from its current bdi to a new bdi. If the inode is dirty we
 * need to move it onto the dirty list of @dst so that the inode is always on
 * the right list.
56 57 58 59
 */
static void bdev_inode_switch_bdi(struct inode *inode,
			struct backing_dev_info *dst)
{
60
	struct backing_dev_info *old = inode->i_data.backing_dev_info;
61
	bool wakeup_bdi = false;
62 63 64 65

	if (unlikely(dst == old))		/* deadlock avoidance */
		return;
	bdi_lock_two(&old->wb, &dst->wb);
66
	spin_lock(&inode->i_lock);
67
	inode->i_data.backing_dev_info = dst;
68 69 70
	if (inode->i_state & I_DIRTY) {
		if (bdi_cap_writeback_dirty(dst) && !wb_has_dirty_io(&dst->wb))
			wakeup_bdi = true;
N
Nick Piggin 已提交
71
		list_move(&inode->i_wb_list, &dst->wb.b_dirty);
72
	}
73
	spin_unlock(&inode->i_lock);
74 75
	spin_unlock(&old->wb.list_lock);
	spin_unlock(&dst->wb.list_lock);
76 77 78

	if (wakeup_bdi)
		bdi_wakeup_thread_delayed(dst);
79 80
}

P
Peter Zijlstra 已提交
81
/* Kill _all_ buffers and pagecache , dirty or not.. */
A
Al Viro 已提交
82
void kill_bdev(struct block_device *bdev)
L
Linus Torvalds 已提交
83
{
A
Al Viro 已提交
84 85
	struct address_space *mapping = bdev->bd_inode->i_mapping;

86
	if (mapping->nrpages == 0 && mapping->nrshadows == 0)
P
Peter Zijlstra 已提交
87
		return;
A
Al Viro 已提交
88

P
Peter Zijlstra 已提交
89
	invalidate_bh_lrus();
A
Al Viro 已提交
90
	truncate_inode_pages(mapping, 0);
L
Linus Torvalds 已提交
91
}	
A
Al Viro 已提交
92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107
EXPORT_SYMBOL(kill_bdev);

/* Invalidate clean unused buffers and pagecache. */
void invalidate_bdev(struct block_device *bdev)
{
	struct address_space *mapping = bdev->bd_inode->i_mapping;

	if (mapping->nrpages == 0)
		return;

	invalidate_bh_lrus();
	lru_add_drain_all();	/* make sure all lru add caches are flushed */
	invalidate_mapping_pages(mapping, 0, -1);
	/* 99% of the time, we don't need to flush the cleancache on the bdev.
	 * But, for the strange corners, lets be cautious
	 */
108
	cleancache_invalidate_inode(mapping);
A
Al Viro 已提交
109 110
}
EXPORT_SYMBOL(invalidate_bdev);
L
Linus Torvalds 已提交
111 112 113 114

int set_blocksize(struct block_device *bdev, int size)
{
	/* Size must be a power of two, and between 512 and PAGE_SIZE */
115
	if (size > PAGE_SIZE || size < 512 || !is_power_of_2(size))
L
Linus Torvalds 已提交
116 117 118
		return -EINVAL;

	/* Size cannot be smaller than the size supported by the device */
119
	if (size < bdev_logical_block_size(bdev))
L
Linus Torvalds 已提交
120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140
		return -EINVAL;

	/* Don't change the size if it is same as current */
	if (bdev->bd_block_size != size) {
		sync_blockdev(bdev);
		bdev->bd_block_size = size;
		bdev->bd_inode->i_blkbits = blksize_bits(size);
		kill_bdev(bdev);
	}
	return 0;
}

EXPORT_SYMBOL(set_blocksize);

int sb_set_blocksize(struct super_block *sb, int size)
{
	if (set_blocksize(sb->s_bdev, size))
		return 0;
	/* If we get here, we know size is power of two
	 * and it's value is between 512 and PAGE_SIZE */
	sb->s_blocksize = size;
141
	sb->s_blocksize_bits = blksize_bits(size);
L
Linus Torvalds 已提交
142 143 144 145 146 147 148
	return sb->s_blocksize;
}

EXPORT_SYMBOL(sb_set_blocksize);

int sb_min_blocksize(struct super_block *sb, int size)
{
149
	int minsize = bdev_logical_block_size(sb->s_bdev);
L
Linus Torvalds 已提交
150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166
	if (size < minsize)
		size = minsize;
	return sb_set_blocksize(sb, size);
}

EXPORT_SYMBOL(sb_min_blocksize);

static int
blkdev_get_block(struct inode *inode, sector_t iblock,
		struct buffer_head *bh, int create)
{
	bh->b_bdev = I_BDEV(inode);
	bh->b_blocknr = iblock;
	set_buffer_mapped(bh);
	return 0;
}

167
static ssize_t
A
Al Viro 已提交
168 169
blkdev_direct_IO(int rw, struct kiocb *iocb, struct iov_iter *iter,
			loff_t offset)
170 171 172 173
{
	struct file *file = iocb->ki_filp;
	struct inode *inode = file->f_mapping->host;

174 175
	return __blockdev_direct_IO(rw, iocb, inode, I_BDEV(inode), iter,
				    offset, blkdev_get_block,
A
Al Viro 已提交
176
				    NULL, NULL, 0);
177 178
}

179 180 181 182 183 184 185 186 187
int __sync_blockdev(struct block_device *bdev, int wait)
{
	if (!bdev)
		return 0;
	if (!wait)
		return filemap_flush(bdev->bd_inode->i_mapping);
	return filemap_write_and_wait(bdev->bd_inode->i_mapping);
}

N
Nick Piggin 已提交
188 189 190 191 192 193
/*
 * Write out and wait upon all the dirty data associated with a block
 * device via its mapping.  Does not take the superblock lock.
 */
int sync_blockdev(struct block_device *bdev)
{
194
	return __sync_blockdev(bdev, 1);
N
Nick Piggin 已提交
195 196 197 198 199 200 201 202 203 204 205 206
}
EXPORT_SYMBOL(sync_blockdev);

/*
 * Write out and wait upon all dirty data associated with this
 * device.   Filesystem data as well as the underlying block
 * device.  Takes the superblock lock.
 */
int fsync_bdev(struct block_device *bdev)
{
	struct super_block *sb = get_super(bdev);
	if (sb) {
207
		int res = sync_filesystem(sb);
N
Nick Piggin 已提交
208 209 210 211 212
		drop_super(sb);
		return res;
	}
	return sync_blockdev(bdev);
}
213
EXPORT_SYMBOL(fsync_bdev);
N
Nick Piggin 已提交
214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232

/**
 * freeze_bdev  --  lock a filesystem and force it into a consistent state
 * @bdev:	blockdevice to lock
 *
 * If a superblock is found on this device, we take the s_umount semaphore
 * on it to make sure nobody unmounts until the snapshot creation is done.
 * The reference counter (bd_fsfreeze_count) guarantees that only the last
 * unfreeze process can unfreeze the frozen filesystem actually when multiple
 * freeze requests arrive simultaneously. It counts up in freeze_bdev() and
 * count down in thaw_bdev(). When it becomes 0, thaw_bdev() will unfreeze
 * actually.
 */
struct super_block *freeze_bdev(struct block_device *bdev)
{
	struct super_block *sb;
	int error = 0;

	mutex_lock(&bdev->bd_fsfreeze_mutex);
233 234 235 236 237 238
	if (++bdev->bd_fsfreeze_count > 1) {
		/*
		 * We don't even need to grab a reference - the first call
		 * to freeze_bdev grab an active reference and only the last
		 * thaw_bdev drops it.
		 */
N
Nick Piggin 已提交
239
		sb = get_super(bdev);
240 241 242 243 244 245 246 247
		drop_super(sb);
		mutex_unlock(&bdev->bd_fsfreeze_mutex);
		return sb;
	}

	sb = get_active_super(bdev);
	if (!sb)
		goto out;
248 249 250 251
	error = freeze_super(sb);
	if (error) {
		deactivate_super(sb);
		bdev->bd_fsfreeze_count--;
N
Nick Piggin 已提交
252
		mutex_unlock(&bdev->bd_fsfreeze_mutex);
253
		return ERR_PTR(error);
N
Nick Piggin 已提交
254
	}
255
	deactivate_super(sb);
256
 out:
N
Nick Piggin 已提交
257 258
	sync_blockdev(bdev);
	mutex_unlock(&bdev->bd_fsfreeze_mutex);
259
	return sb;	/* thaw_bdev releases s->s_umount */
N
Nick Piggin 已提交
260 261 262 263 264 265 266 267 268 269 270 271
}
EXPORT_SYMBOL(freeze_bdev);

/**
 * thaw_bdev  -- unlock filesystem
 * @bdev:	blockdevice to unlock
 * @sb:		associated superblock
 *
 * Unlocks the filesystem and marks it writeable again after freeze_bdev().
 */
int thaw_bdev(struct block_device *bdev, struct super_block *sb)
{
272
	int error = -EINVAL;
N
Nick Piggin 已提交
273 274

	mutex_lock(&bdev->bd_fsfreeze_mutex);
275
	if (!bdev->bd_fsfreeze_count)
276
		goto out;
277 278 279

	error = 0;
	if (--bdev->bd_fsfreeze_count > 0)
280
		goto out;
281 282

	if (!sb)
283
		goto out;
284

285 286 287 288 289 290 291
	error = thaw_super(sb);
	if (error) {
		bdev->bd_fsfreeze_count++;
		mutex_unlock(&bdev->bd_fsfreeze_mutex);
		return error;
	}
out:
N
Nick Piggin 已提交
292 293 294 295 296
	mutex_unlock(&bdev->bd_fsfreeze_mutex);
	return 0;
}
EXPORT_SYMBOL(thaw_bdev);

L
Linus Torvalds 已提交
297 298 299 300 301 302 303 304 305 306
static int blkdev_writepage(struct page *page, struct writeback_control *wbc)
{
	return block_write_full_page(page, blkdev_get_block, wbc);
}

static int blkdev_readpage(struct file * file, struct page * page)
{
	return block_read_full_page(page, blkdev_get_block);
}

N
Nick Piggin 已提交
307 308 309
static int blkdev_write_begin(struct file *file, struct address_space *mapping,
			loff_t pos, unsigned len, unsigned flags,
			struct page **pagep, void **fsdata)
L
Linus Torvalds 已提交
310
{
311 312
	return block_write_begin(mapping, pos, len, flags, pagep,
				 blkdev_get_block);
L
Linus Torvalds 已提交
313 314
}

N
Nick Piggin 已提交
315 316 317
static int blkdev_write_end(struct file *file, struct address_space *mapping,
			loff_t pos, unsigned len, unsigned copied,
			struct page *page, void *fsdata)
L
Linus Torvalds 已提交
318
{
N
Nick Piggin 已提交
319 320 321 322 323 324 325
	int ret;
	ret = block_write_end(file, mapping, pos, len, copied, page, fsdata);

	unlock_page(page);
	page_cache_release(page);

	return ret;
L
Linus Torvalds 已提交
326 327 328 329
}

/*
 * private llseek:
A
Al Viro 已提交
330
 * for a block special file file_inode(file)->i_size is zero
L
Linus Torvalds 已提交
331 332
 * so we compute the size by hand (just as in block_read/write above)
 */
333
static loff_t block_llseek(struct file *file, loff_t offset, int whence)
L
Linus Torvalds 已提交
334 335 336 337
{
	struct inode *bd_inode = file->f_mapping->host;
	loff_t retval;

338
	mutex_lock(&bd_inode->i_mutex);
339
	retval = fixed_size_llseek(file, offset, whence, i_size_read(bd_inode));
340
	mutex_unlock(&bd_inode->i_mutex);
L
Linus Torvalds 已提交
341 342 343
	return retval;
}
	
344
int blkdev_fsync(struct file *filp, loff_t start, loff_t end, int datasync)
L
Linus Torvalds 已提交
345
{
346 347
	struct inode *bd_inode = filp->f_mapping->host;
	struct block_device *bdev = I_BDEV(bd_inode);
348
	int error;
349 350 351 352
	
	error = filemap_write_and_wait_range(filp->f_mapping, start, end);
	if (error)
		return error;
353

354 355 356 357 358
	/*
	 * There is no need to serialise calls to blkdev_issue_flush with
	 * i_mutex and doing so causes performance issues with concurrent
	 * O_SYNC writers to a block device.
	 */
359
	error = blkdev_issue_flush(bdev, GFP_KERNEL, NULL);
360 361
	if (error == -EOPNOTSUPP)
		error = 0;
362

363
	return error;
L
Linus Torvalds 已提交
364
}
365
EXPORT_SYMBOL(blkdev_fsync);
L
Linus Torvalds 已提交
366 367 368 369 370 371

/*
 * pseudo-fs
 */

static  __cacheline_aligned_in_smp DEFINE_SPINLOCK(bdev_lock);
372
static struct kmem_cache * bdev_cachep __read_mostly;
L
Linus Torvalds 已提交
373 374 375

static struct inode *bdev_alloc_inode(struct super_block *sb)
{
376
	struct bdev_inode *ei = kmem_cache_alloc(bdev_cachep, GFP_KERNEL);
L
Linus Torvalds 已提交
377 378 379 380 381
	if (!ei)
		return NULL;
	return &ei->vfs_inode;
}

N
Nick Piggin 已提交
382
static void bdev_i_callback(struct rcu_head *head)
L
Linus Torvalds 已提交
383
{
N
Nick Piggin 已提交
384
	struct inode *inode = container_of(head, struct inode, i_rcu);
L
Linus Torvalds 已提交
385 386 387 388 389
	struct bdev_inode *bdi = BDEV_I(inode);

	kmem_cache_free(bdev_cachep, bdi);
}

N
Nick Piggin 已提交
390 391 392 393 394
static void bdev_destroy_inode(struct inode *inode)
{
	call_rcu(&inode->i_rcu, bdev_i_callback);
}

395
static void init_once(void *foo)
L
Linus Torvalds 已提交
396 397 398 399
{
	struct bdev_inode *ei = (struct bdev_inode *) foo;
	struct block_device *bdev = &ei->bdev;

C
Christoph Lameter 已提交
400 401 402 403
	memset(bdev, 0, sizeof(*bdev));
	mutex_init(&bdev->bd_mutex);
	INIT_LIST_HEAD(&bdev->bd_inodes);
	INIT_LIST_HEAD(&bdev->bd_list);
404 405 406
#ifdef CONFIG_SYSFS
	INIT_LIST_HEAD(&bdev->bd_holder_disks);
#endif
C
Christoph Lameter 已提交
407
	inode_init_once(&ei->vfs_inode);
408 409
	/* Initialize mutex for freeze. */
	mutex_init(&bdev->bd_fsfreeze_mutex);
L
Linus Torvalds 已提交
410 411 412 413 414 415 416 417 418
}

static inline void __bd_forget(struct inode *inode)
{
	list_del_init(&inode->i_devices);
	inode->i_bdev = NULL;
	inode->i_mapping = &inode->i_data;
}

419
static void bdev_evict_inode(struct inode *inode)
L
Linus Torvalds 已提交
420 421 422
{
	struct block_device *bdev = &BDEV_I(inode)->bdev;
	struct list_head *p;
423
	truncate_inode_pages_final(&inode->i_data);
424
	invalidate_inode_buffers(inode); /* is it needed here? */
425
	clear_inode(inode);
L
Linus Torvalds 已提交
426 427 428 429 430 431 432 433
	spin_lock(&bdev_lock);
	while ( (p = bdev->bd_inodes.next) != &bdev->bd_inodes ) {
		__bd_forget(list_entry(p, struct inode, i_devices));
	}
	list_del_init(&bdev->bd_list);
	spin_unlock(&bdev_lock);
}

434
static const struct super_operations bdev_sops = {
L
Linus Torvalds 已提交
435 436 437 438
	.statfs = simple_statfs,
	.alloc_inode = bdev_alloc_inode,
	.destroy_inode = bdev_destroy_inode,
	.drop_inode = generic_delete_inode,
439
	.evict_inode = bdev_evict_inode,
L
Linus Torvalds 已提交
440 441
};

A
Al Viro 已提交
442 443
static struct dentry *bd_mount(struct file_system_type *fs_type,
	int flags, const char *dev_name, void *data)
L
Linus Torvalds 已提交
444
{
445
	return mount_pseudo(fs_type, "bdev:", &bdev_sops, NULL, BDEVFS_MAGIC);
L
Linus Torvalds 已提交
446 447 448 449
}

static struct file_system_type bd_type = {
	.name		= "bdev",
A
Al Viro 已提交
450
	.mount		= bd_mount,
L
Linus Torvalds 已提交
451 452 453
	.kill_sb	= kill_anon_super,
};

A
Al Viro 已提交
454
static struct super_block *blockdev_superblock __read_mostly;
L
Linus Torvalds 已提交
455 456 457 458

void __init bdev_cache_init(void)
{
	int err;
459
	static struct vfsmount *bd_mnt;
460

L
Linus Torvalds 已提交
461
	bdev_cachep = kmem_cache_create("bdev_cache", sizeof(struct bdev_inode),
462 463
			0, (SLAB_HWCACHE_ALIGN|SLAB_RECLAIM_ACCOUNT|
				SLAB_MEM_SPREAD|SLAB_PANIC),
464
			init_once);
L
Linus Torvalds 已提交
465 466 467 468 469 470
	err = register_filesystem(&bd_type);
	if (err)
		panic("Cannot register bdev pseudo-fs");
	bd_mnt = kern_mount(&bd_type);
	if (IS_ERR(bd_mnt))
		panic("Cannot create bdev pseudo-fs");
471
	blockdev_superblock = bd_mnt->mnt_sb;   /* For writeback */
L
Linus Torvalds 已提交
472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501
}

/*
 * Most likely _very_ bad one - but then it's hardly critical for small
 * /dev and can be fixed when somebody will need really large one.
 * Keep in mind that it will be fed through icache hash function too.
 */
static inline unsigned long hash(dev_t dev)
{
	return MAJOR(dev)+MINOR(dev);
}

static int bdev_test(struct inode *inode, void *data)
{
	return BDEV_I(inode)->bdev.bd_dev == *(dev_t *)data;
}

static int bdev_set(struct inode *inode, void *data)
{
	BDEV_I(inode)->bdev.bd_dev = *(dev_t *)data;
	return 0;
}

static LIST_HEAD(all_bdevs);

struct block_device *bdget(dev_t dev)
{
	struct block_device *bdev;
	struct inode *inode;

502
	inode = iget5_locked(blockdev_superblock, hash(dev),
L
Linus Torvalds 已提交
503 504 505 506 507 508 509 510 511
			bdev_test, bdev_set, &dev);

	if (!inode)
		return NULL;

	bdev = &BDEV_I(inode)->bdev;

	if (inode->i_state & I_NEW) {
		bdev->bd_contains = NULL;
512
		bdev->bd_super = NULL;
L
Linus Torvalds 已提交
513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532
		bdev->bd_inode = inode;
		bdev->bd_block_size = (1 << inode->i_blkbits);
		bdev->bd_part_count = 0;
		bdev->bd_invalidated = 0;
		inode->i_mode = S_IFBLK;
		inode->i_rdev = dev;
		inode->i_bdev = bdev;
		inode->i_data.a_ops = &def_blk_aops;
		mapping_set_gfp_mask(&inode->i_data, GFP_USER);
		inode->i_data.backing_dev_info = &default_backing_dev_info;
		spin_lock(&bdev_lock);
		list_add(&bdev->bd_list, &all_bdevs);
		spin_unlock(&bdev_lock);
		unlock_new_inode(inode);
	}
	return bdev;
}

EXPORT_SYMBOL(bdget);

533 534 535 536 537 538
/**
 * bdgrab -- Grab a reference to an already referenced block device
 * @bdev:	Block device to grab a reference to.
 */
struct block_device *bdgrab(struct block_device *bdev)
{
A
Al Viro 已提交
539
	ihold(bdev->bd_inode);
540 541
	return bdev;
}
542
EXPORT_SYMBOL(bdgrab);
543

L
Linus Torvalds 已提交
544 545
long nr_blockdev_pages(void)
{
546
	struct block_device *bdev;
L
Linus Torvalds 已提交
547 548
	long ret = 0;
	spin_lock(&bdev_lock);
549
	list_for_each_entry(bdev, &all_bdevs, bd_list) {
L
Linus Torvalds 已提交
550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565
		ret += bdev->bd_inode->i_mapping->nrpages;
	}
	spin_unlock(&bdev_lock);
	return ret;
}

void bdput(struct block_device *bdev)
{
	iput(bdev->bd_inode);
}

EXPORT_SYMBOL(bdput);
 
static struct block_device *bd_acquire(struct inode *inode)
{
	struct block_device *bdev;
566

L
Linus Torvalds 已提交
567 568
	spin_lock(&bdev_lock);
	bdev = inode->i_bdev;
569
	if (bdev) {
A
Al Viro 已提交
570
		ihold(bdev->bd_inode);
L
Linus Torvalds 已提交
571 572 573 574
		spin_unlock(&bdev_lock);
		return bdev;
	}
	spin_unlock(&bdev_lock);
575

L
Linus Torvalds 已提交
576 577 578
	bdev = bdget(inode->i_rdev);
	if (bdev) {
		spin_lock(&bdev_lock);
579 580
		if (!inode->i_bdev) {
			/*
A
Al Viro 已提交
581
			 * We take an additional reference to bd_inode,
582 583 584 585
			 * and it's released in clear_inode() of inode.
			 * So, we can access it via ->i_mapping always
			 * without igrab().
			 */
A
Al Viro 已提交
586
			ihold(bdev->bd_inode);
587 588 589 590
			inode->i_bdev = bdev;
			inode->i_mapping = bdev->bd_inode->i_mapping;
			list_add(&inode->i_devices, &bdev->bd_inodes);
		}
L
Linus Torvalds 已提交
591 592 593 594 595
		spin_unlock(&bdev_lock);
	}
	return bdev;
}

596
int sb_is_blkdev_sb(struct super_block *sb)
A
Al Viro 已提交
597 598 599 600
{
	return sb == blockdev_superblock;
}

L
Linus Torvalds 已提交
601 602 603 604
/* Call when you free inode */

void bd_forget(struct inode *inode)
{
605 606
	struct block_device *bdev = NULL;

L
Linus Torvalds 已提交
607
	spin_lock(&bdev_lock);
608 609 610
	if (!sb_is_blkdev_sb(inode->i_sb))
		bdev = inode->i_bdev;
	__bd_forget(inode);
L
Linus Torvalds 已提交
611
	spin_unlock(&bdev_lock);
612 613 614

	if (bdev)
		iput(bdev->bd_inode);
L
Linus Torvalds 已提交
615 616
}

T
Tejun Heo 已提交
617 618 619 620 621 622
/**
 * bd_may_claim - test whether a block device can be claimed
 * @bdev: block device of interest
 * @whole: whole block device containing @bdev, may equal @bdev
 * @holder: holder trying to claim @bdev
 *
L
Lucas De Marchi 已提交
623
 * Test whether @bdev can be claimed by @holder.
T
Tejun Heo 已提交
624 625 626 627 628 629 630 631 632
 *
 * CONTEXT:
 * spin_lock(&bdev_lock).
 *
 * RETURNS:
 * %true if @bdev can be claimed, %false otherwise.
 */
static bool bd_may_claim(struct block_device *bdev, struct block_device *whole,
			 void *holder)
L
Linus Torvalds 已提交
633 634
{
	if (bdev->bd_holder == holder)
T
Tejun Heo 已提交
635
		return true;	 /* already a holder */
L
Linus Torvalds 已提交
636
	else if (bdev->bd_holder != NULL)
T
Tejun Heo 已提交
637
		return false; 	 /* held by someone else */
L
Linus Torvalds 已提交
638
	else if (bdev->bd_contains == bdev)
T
Tejun Heo 已提交
639
		return true;  	 /* is a whole device which isn't held */
L
Linus Torvalds 已提交
640

641
	else if (whole->bd_holder == bd_may_claim)
T
Tejun Heo 已提交
642 643 644
		return true; 	 /* is a partition of a device that is being partitioned */
	else if (whole->bd_holder != NULL)
		return false;	 /* is a partition of a held device */
L
Linus Torvalds 已提交
645
	else
T
Tejun Heo 已提交
646 647 648
		return true;	 /* is a partition of an un-held device */
}

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
/**
 * bd_prepare_to_claim - prepare to claim a block device
 * @bdev: block device of interest
 * @whole: the whole device containing @bdev, may equal @bdev
 * @holder: holder trying to claim @bdev
 *
 * Prepare to claim @bdev.  This function fails if @bdev is already
 * claimed by another holder and waits if another claiming is in
 * progress.  This function doesn't actually claim.  On successful
 * return, the caller has ownership of bd_claiming and bd_holder[s].
 *
 * CONTEXT:
 * spin_lock(&bdev_lock).  Might release bdev_lock, sleep and regrab
 * it multiple times.
 *
 * RETURNS:
 * 0 if @bdev can be claimed, -EBUSY otherwise.
 */
static int bd_prepare_to_claim(struct block_device *bdev,
			       struct block_device *whole, void *holder)
{
retry:
	/* if someone else claimed, fail */
	if (!bd_may_claim(bdev, whole, holder))
		return -EBUSY;

675 676
	/* if claiming is already in progress, wait for it to finish */
	if (whole->bd_claiming) {
677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699
		wait_queue_head_t *wq = bit_waitqueue(&whole->bd_claiming, 0);
		DEFINE_WAIT(wait);

		prepare_to_wait(wq, &wait, TASK_UNINTERRUPTIBLE);
		spin_unlock(&bdev_lock);
		schedule();
		finish_wait(wq, &wait);
		spin_lock(&bdev_lock);
		goto retry;
	}

	/* yay, all mine */
	return 0;
}

/**
 * bd_start_claiming - start claiming a block device
 * @bdev: block device of interest
 * @holder: holder trying to claim @bdev
 *
 * @bdev is about to be opened exclusively.  Check @bdev can be opened
 * exclusively and mark that an exclusive open is in progress.  Each
 * successful call to this function must be matched with a call to
N
Nick Piggin 已提交
700 701 702 703 704 705 706
 * either bd_finish_claiming() or bd_abort_claiming() (which do not
 * fail).
 *
 * This function is used to gain exclusive access to the block device
 * without actually causing other exclusive open attempts to fail. It
 * should be used when the open sequence itself requires exclusive
 * access but may subsequently fail.
707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731
 *
 * CONTEXT:
 * Might sleep.
 *
 * RETURNS:
 * Pointer to the block device containing @bdev on success, ERR_PTR()
 * value on failure.
 */
static struct block_device *bd_start_claiming(struct block_device *bdev,
					      void *holder)
{
	struct gendisk *disk;
	struct block_device *whole;
	int partno, err;

	might_sleep();

	/*
	 * @bdev might not have been initialized properly yet, look up
	 * and grab the outer block device the hard way.
	 */
	disk = get_gendisk(bdev->bd_dev, &partno);
	if (!disk)
		return ERR_PTR(-ENXIO);

732 733 734 735 736 737 738 739 740 741 742 743 744
	/*
	 * Normally, @bdev should equal what's returned from bdget_disk()
	 * if partno is 0; however, some drivers (floppy) use multiple
	 * bdev's for the same physical device and @bdev may be one of the
	 * aliases.  Keep @bdev if partno is 0.  This means claimer
	 * tracking is broken for those devices but it has always been that
	 * way.
	 */
	if (partno)
		whole = bdget_disk(disk, 0);
	else
		whole = bdgrab(bdev);

745
	module_put(disk->fops->owner);
746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764
	put_disk(disk);
	if (!whole)
		return ERR_PTR(-ENOMEM);

	/* prepare to claim, if successful, mark claiming in progress */
	spin_lock(&bdev_lock);

	err = bd_prepare_to_claim(bdev, whole, holder);
	if (err == 0) {
		whole->bd_claiming = holder;
		spin_unlock(&bdev_lock);
		return whole;
	} else {
		spin_unlock(&bdev_lock);
		bdput(whole);
		return ERR_PTR(err);
	}
}

765
#ifdef CONFIG_SYSFS
766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782
struct bd_holder_disk {
	struct list_head	list;
	struct gendisk		*disk;
	int			refcnt;
};

static struct bd_holder_disk *bd_find_holder_disk(struct block_device *bdev,
						  struct gendisk *disk)
{
	struct bd_holder_disk *holder;

	list_for_each_entry(holder, &bdev->bd_holder_disks, list)
		if (holder->disk == disk)
			return holder;
	return NULL;
}

783
static int add_symlink(struct kobject *from, struct kobject *to)
784
{
785
	return sysfs_create_link(from, to, kobject_name(to));
786 787 788 789 790 791 792
}

static void del_symlink(struct kobject *from, struct kobject *to)
{
	sysfs_remove_link(from, kobject_name(to));
}

J
Jun'ichi Nomura 已提交
793
/**
794 795 796
 * bd_link_disk_holder - create symlinks between holding disk and slave bdev
 * @bdev: the claimed slave bdev
 * @disk: the holding disk
J
Jun'ichi Nomura 已提交
797
 *
798 799
 * DON'T USE THIS UNLESS YOU'RE ALREADY USING IT.
 *
800
 * This functions creates the following sysfs symlinks.
801
 *
802 803
 * - from "slaves" directory of the holder @disk to the claimed @bdev
 * - from "holders" directory of the @bdev to the holder @disk
804
 *
805 806
 * For example, if /dev/dm-0 maps to /dev/sda and disk for dm-0 is
 * passed to bd_link_disk_holder(), then:
807
 *
808 809
 *   /sys/block/dm-0/slaves/sda --> /sys/block/sda
 *   /sys/block/sda/holders/dm-0 --> /sys/block/dm-0
810
 *
811 812 813
 * The caller must have claimed @bdev before calling this function and
 * ensure that both @bdev and @disk are valid during the creation and
 * lifetime of these symlinks.
814
 *
815 816
 * CONTEXT:
 * Might sleep.
817
 *
818 819
 * RETURNS:
 * 0 on success, -errno on failure.
820
 */
821
int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk)
822
{
823
	struct bd_holder_disk *holder;
824
	int ret = 0;
825

826
	mutex_lock(&bdev->bd_mutex);
J
Jun'ichi Nomura 已提交
827

828
	WARN_ON_ONCE(!bdev->bd_holder);
829

830 831 832
	/* FIXME: remove the following once add_disk() handles errors */
	if (WARN_ON(!disk->slave_dir || !bdev->bd_part->holder_dir))
		goto out_unlock;
833

834 835 836
	holder = bd_find_holder_disk(bdev, disk);
	if (holder) {
		holder->refcnt++;
837
		goto out_unlock;
838
	}
839

840 841 842
	holder = kzalloc(sizeof(*holder), GFP_KERNEL);
	if (!holder) {
		ret = -ENOMEM;
843 844
		goto out_unlock;
	}
845

846 847 848 849 850 851 852 853 854 855 856
	INIT_LIST_HEAD(&holder->list);
	holder->disk = disk;
	holder->refcnt = 1;

	ret = add_symlink(disk->slave_dir, &part_to_dev(bdev->bd_part)->kobj);
	if (ret)
		goto out_free;

	ret = add_symlink(bdev->bd_part->holder_dir, &disk_to_dev(disk)->kobj);
	if (ret)
		goto out_del;
857 858 859 860 861
	/*
	 * bdev could be deleted beneath us which would implicitly destroy
	 * the holder directory.  Hold on to it.
	 */
	kobject_get(bdev->bd_part->holder_dir);
862 863 864 865 866 867 868 869

	list_add(&holder->list, &bdev->bd_holder_disks);
	goto out_unlock;

out_del:
	del_symlink(disk->slave_dir, &part_to_dev(bdev->bd_part)->kobj);
out_free:
	kfree(holder);
870
out_unlock:
871
	mutex_unlock(&bdev->bd_mutex);
872
	return ret;
873
}
874
EXPORT_SYMBOL_GPL(bd_link_disk_holder);
875

876 877 878 879 880 881 882 883 884 885 886
/**
 * bd_unlink_disk_holder - destroy symlinks created by bd_link_disk_holder()
 * @bdev: the calimed slave bdev
 * @disk: the holding disk
 *
 * DON'T USE THIS UNLESS YOU'RE ALREADY USING IT.
 *
 * CONTEXT:
 * Might sleep.
 */
void bd_unlink_disk_holder(struct block_device *bdev, struct gendisk *disk)
887
{
888
	struct bd_holder_disk *holder;
889

890
	mutex_lock(&bdev->bd_mutex);
891

892 893 894 895 896 897
	holder = bd_find_holder_disk(bdev, disk);

	if (!WARN_ON_ONCE(holder == NULL) && !--holder->refcnt) {
		del_symlink(disk->slave_dir, &part_to_dev(bdev->bd_part)->kobj);
		del_symlink(bdev->bd_part->holder_dir,
			    &disk_to_dev(disk)->kobj);
898
		kobject_put(bdev->bd_part->holder_dir);
899 900 901 902 903
		list_del_init(&holder->list);
		kfree(holder);
	}

	mutex_unlock(&bdev->bd_mutex);
L
Linus Torvalds 已提交
904
}
905
EXPORT_SYMBOL_GPL(bd_unlink_disk_holder);
906
#endif
L
Linus Torvalds 已提交
907

908 909 910 911
/**
 * flush_disk - invalidates all buffer-cache entries on a disk
 *
 * @bdev:      struct block device to be flushed
912
 * @kill_dirty: flag to guide handling of dirty inodes
913 914 915 916 917
 *
 * Invalidates all buffer-cache entries on a disk. It should be called
 * when a disk has been changed -- either by a media change or online
 * resize.
 */
918
static void flush_disk(struct block_device *bdev, bool kill_dirty)
919
{
920
	if (__invalidate_device(bdev, kill_dirty)) {
921 922 923 924 925 926 927 928 929 930
		char name[BDEVNAME_SIZE] = "";

		if (bdev->bd_disk)
			disk_name(bdev->bd_disk, 0, name);
		printk(KERN_WARNING "VFS: busy inodes on changed media or "
		       "resized disk %s\n", name);
	}

	if (!bdev->bd_disk)
		return;
T
Tejun Heo 已提交
931
	if (disk_part_scan_enabled(bdev->bd_disk))
932 933 934
		bdev->bd_invalidated = 1;
}

935
/**
936
 * check_disk_size_change - checks for disk size change and adjusts bdev size.
937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956
 * @disk: struct gendisk to check
 * @bdev: struct bdev to adjust.
 *
 * This routine checks to see if the bdev size does not match the disk size
 * and adjusts it if it differs.
 */
void check_disk_size_change(struct gendisk *disk, struct block_device *bdev)
{
	loff_t disk_size, bdev_size;

	disk_size = (loff_t)get_capacity(disk) << 9;
	bdev_size = i_size_read(bdev->bd_inode);
	if (disk_size != bdev_size) {
		char name[BDEVNAME_SIZE];

		disk_name(disk, 0, name);
		printk(KERN_INFO
		       "%s: detected capacity change from %lld to %lld\n",
		       name, bdev_size, disk_size);
		i_size_write(bdev->bd_inode, disk_size);
957
		flush_disk(bdev, false);
958 959 960 961
	}
}
EXPORT_SYMBOL(check_disk_size_change);

962
/**
963
 * revalidate_disk - wrapper for lower-level driver's revalidate_disk call-back
964 965 966 967 968 969 970 971
 * @disk: struct gendisk to be revalidated
 *
 * This routine is a wrapper for lower-level driver's revalidate_disk
 * call-backs.  It is used to do common pre and post operations needed
 * for all revalidate_disk operations.
 */
int revalidate_disk(struct gendisk *disk)
{
972
	struct block_device *bdev;
973 974 975 976 977
	int ret = 0;

	if (disk->fops->revalidate_disk)
		ret = disk->fops->revalidate_disk(disk);

978 979 980 981 982 983
	bdev = bdget_disk(disk, 0);
	if (!bdev)
		return ret;

	mutex_lock(&bdev->bd_mutex);
	check_disk_size_change(disk, bdev);
984
	bdev->bd_invalidated = 0;
985 986
	mutex_unlock(&bdev->bd_mutex);
	bdput(bdev);
987 988 989 990
	return ret;
}
EXPORT_SYMBOL(revalidate_disk);

L
Linus Torvalds 已提交
991 992 993 994 995 996 997 998 999 1000 1001 1002
/*
 * This routine checks whether a removable media has been changed,
 * and invalidates all buffer-cache-entries in that case. This
 * is a relatively slow routine, so we have to try to minimize using
 * it. Thus it is called only upon a 'mount' or 'open'. This
 * is the best way of combining speed and utility, I think.
 * People changing diskettes in the middle of an operation deserve
 * to lose :-)
 */
int check_disk_change(struct block_device *bdev)
{
	struct gendisk *disk = bdev->bd_disk;
1003
	const struct block_device_operations *bdops = disk->fops;
1004
	unsigned int events;
L
Linus Torvalds 已提交
1005

1006 1007 1008
	events = disk_clear_events(disk, DISK_EVENT_MEDIA_CHANGE |
				   DISK_EVENT_EJECT_REQUEST);
	if (!(events & DISK_EVENT_MEDIA_CHANGE))
L
Linus Torvalds 已提交
1009 1010
		return 0;

1011
	flush_disk(bdev, true);
L
Linus Torvalds 已提交
1012 1013 1014 1015 1016 1017 1018 1019 1020
	if (bdops->revalidate_disk)
		bdops->revalidate_disk(bdev->bd_disk);
	return 1;
}

EXPORT_SYMBOL(check_disk_change);

void bd_set_size(struct block_device *bdev, loff_t size)
{
1021
	unsigned bsize = bdev_logical_block_size(bdev);
L
Linus Torvalds 已提交
1022

1023 1024 1025
	mutex_lock(&bdev->bd_inode->i_mutex);
	i_size_write(bdev->bd_inode, size);
	mutex_unlock(&bdev->bd_inode->i_mutex);
L
Linus Torvalds 已提交
1026 1027 1028 1029 1030 1031 1032 1033 1034 1035
	while (bsize < PAGE_CACHE_SIZE) {
		if (size & bsize)
			break;
		bsize <<= 1;
	}
	bdev->bd_block_size = bsize;
	bdev->bd_inode->i_blkbits = blksize_bits(bsize);
}
EXPORT_SYMBOL(bd_set_size);

A
Al Viro 已提交
1036
static void __blkdev_put(struct block_device *bdev, fmode_t mode, int for_part);
1037

1038 1039 1040 1041 1042 1043 1044
/*
 * bd_mutex locking:
 *
 *  mutex_lock(part->bd_mutex)
 *    mutex_lock_nested(whole->bd_mutex, 1)
 */

1045
static int __blkdev_get(struct block_device *bdev, fmode_t mode, int for_part)
L
Linus Torvalds 已提交
1046 1047
{
	struct gendisk *disk;
1048
	struct module *owner;
1049
	int ret;
1050
	int partno;
1051 1052
	int perm = 0;

1053
	if (mode & FMODE_READ)
1054
		perm |= MAY_READ;
1055
	if (mode & FMODE_WRITE)
1056 1057 1058 1059
		perm |= MAY_WRITE;
	/*
	 * hooks: /n/, see "layering violations".
	 */
1060 1061 1062 1063 1064 1065
	if (!for_part) {
		ret = devcgroup_inode_permission(bdev->bd_inode, perm);
		if (ret != 0) {
			bdput(bdev);
			return ret;
		}
1066
	}
1067

1068
 restart:
T
Tejun Heo 已提交
1069

1070
	ret = -ENXIO;
1071
	disk = get_gendisk(bdev->bd_dev, &partno);
T
Tejun Heo 已提交
1072
	if (!disk)
1073
		goto out;
1074
	owner = disk->fops->owner;
L
Linus Torvalds 已提交
1075

1076
	disk_block_events(disk);
1077
	mutex_lock_nested(&bdev->bd_mutex, for_part);
L
Linus Torvalds 已提交
1078 1079
	if (!bdev->bd_openers) {
		bdev->bd_disk = disk;
1080
		bdev->bd_queue = disk->queue;
L
Linus Torvalds 已提交
1081
		bdev->bd_contains = bdev;
1082
		if (!partno) {
L
Linus Torvalds 已提交
1083
			struct backing_dev_info *bdi;
1084 1085 1086 1087 1088 1089

			ret = -ENXIO;
			bdev->bd_part = disk_get_part(disk, partno);
			if (!bdev->bd_part)
				goto out_clear;

1090
			ret = 0;
L
Linus Torvalds 已提交
1091
			if (disk->fops->open) {
1092
				ret = disk->fops->open(bdev, mode);
1093 1094 1095 1096 1097 1098 1099 1100
				if (ret == -ERESTARTSYS) {
					/* Lost a race with 'disk' being
					 * deleted, try again.
					 * See md.c
					 */
					disk_put_part(bdev->bd_part);
					bdev->bd_part = NULL;
					bdev->bd_disk = NULL;
1101
					bdev->bd_queue = NULL;
1102
					mutex_unlock(&bdev->bd_mutex);
1103 1104
					disk_unblock_events(disk);
					put_disk(disk);
1105
					module_put(owner);
1106 1107
					goto restart;
				}
L
Linus Torvalds 已提交
1108
			}
1109

1110
			if (!ret) {
1111 1112 1113 1114 1115 1116 1117
				bd_set_size(bdev,(loff_t)get_capacity(disk)<<9);
				bdi = blk_get_backing_dev_info(bdev);
				if (bdi == NULL)
					bdi = &default_backing_dev_info;
				bdev_inode_switch_bdi(bdev->bd_inode, bdi);
			}

1118 1119 1120 1121 1122 1123
			/*
			 * If the device is invalidated, rescan partition
			 * if open succeeded or failed with -ENOMEDIUM.
			 * The latter is necessary to prevent ghost
			 * partitions on a removed medium.
			 */
1124 1125 1126 1127 1128 1129
			if (bdev->bd_invalidated) {
				if (!ret)
					rescan_partitions(disk, bdev);
				else if (ret == -ENOMEDIUM)
					invalidate_partitions(disk, bdev);
			}
1130 1131
			if (ret)
				goto out_clear;
L
Linus Torvalds 已提交
1132 1133 1134 1135 1136
		} else {
			struct block_device *whole;
			whole = bdget_disk(disk, 0);
			ret = -ENOMEM;
			if (!whole)
T
Tejun Heo 已提交
1137
				goto out_clear;
1138
			BUG_ON(for_part);
1139
			ret = __blkdev_get(whole, mode, 1);
L
Linus Torvalds 已提交
1140
			if (ret)
T
Tejun Heo 已提交
1141
				goto out_clear;
L
Linus Torvalds 已提交
1142
			bdev->bd_contains = whole;
1143 1144
			bdev_inode_switch_bdi(bdev->bd_inode,
				whole->bd_inode->i_data.backing_dev_info);
1145
			bdev->bd_part = disk_get_part(disk, partno);
1146
			if (!(disk->flags & GENHD_FL_UP) ||
1147
			    !bdev->bd_part || !bdev->bd_part->nr_sects) {
L
Linus Torvalds 已提交
1148
				ret = -ENXIO;
T
Tejun Heo 已提交
1149
				goto out_clear;
L
Linus Torvalds 已提交
1150
			}
1151
			bd_set_size(bdev, (loff_t)bdev->bd_part->nr_sects << 9);
L
Linus Torvalds 已提交
1152 1153 1154
		}
	} else {
		if (bdev->bd_contains == bdev) {
1155 1156
			ret = 0;
			if (bdev->bd_disk->fops->open)
1157
				ret = bdev->bd_disk->fops->open(bdev, mode);
1158
			/* the same as first opener case, read comment there */
1159 1160 1161 1162 1163 1164
			if (bdev->bd_invalidated) {
				if (!ret)
					rescan_partitions(bdev->bd_disk, bdev);
				else if (ret == -ENOMEDIUM)
					invalidate_partitions(bdev->bd_disk, bdev);
			}
1165 1166
			if (ret)
				goto out_unlock_bdev;
L
Linus Torvalds 已提交
1167
		}
1168 1169
		/* only one opener holds refs to the module and disk */
		put_disk(disk);
1170
		module_put(owner);
L
Linus Torvalds 已提交
1171 1172
	}
	bdev->bd_openers++;
1173 1174
	if (for_part)
		bdev->bd_part_count++;
1175
	mutex_unlock(&bdev->bd_mutex);
1176
	disk_unblock_events(disk);
L
Linus Torvalds 已提交
1177 1178
	return 0;

T
Tejun Heo 已提交
1179
 out_clear:
1180
	disk_put_part(bdev->bd_part);
L
Linus Torvalds 已提交
1181
	bdev->bd_disk = NULL;
T
Tejun Heo 已提交
1182
	bdev->bd_part = NULL;
1183
	bdev->bd_queue = NULL;
1184
	bdev_inode_switch_bdi(bdev->bd_inode, &default_backing_dev_info);
L
Linus Torvalds 已提交
1185
	if (bdev != bdev->bd_contains)
1186
		__blkdev_put(bdev->bd_contains, mode, 1);
L
Linus Torvalds 已提交
1187
	bdev->bd_contains = NULL;
T
Tejun Heo 已提交
1188
 out_unlock_bdev:
1189
	mutex_unlock(&bdev->bd_mutex);
1190
	disk_unblock_events(disk);
T
Tejun Heo 已提交
1191
	put_disk(disk);
1192
	module_put(owner);
1193
 out:
T
Tejun Heo 已提交
1194 1195
	bdput(bdev);

L
Linus Torvalds 已提交
1196 1197 1198
	return ret;
}

1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217
/**
 * blkdev_get - open a block device
 * @bdev: block_device to open
 * @mode: FMODE_* mask
 * @holder: exclusive holder identifier
 *
 * Open @bdev with @mode.  If @mode includes %FMODE_EXCL, @bdev is
 * open with exclusive access.  Specifying %FMODE_EXCL with %NULL
 * @holder is invalid.  Exclusive opens may nest for the same @holder.
 *
 * On success, the reference count of @bdev is unchanged.  On failure,
 * @bdev is put.
 *
 * CONTEXT:
 * Might sleep.
 *
 * RETURNS:
 * 0 on success, -errno on failure.
 */
1218
int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder)
L
Linus Torvalds 已提交
1219
{
1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235
	struct block_device *whole = NULL;
	int res;

	WARN_ON_ONCE((mode & FMODE_EXCL) && !holder);

	if ((mode & FMODE_EXCL) && holder) {
		whole = bd_start_claiming(bdev, holder);
		if (IS_ERR(whole)) {
			bdput(bdev);
			return PTR_ERR(whole);
		}
	}

	res = __blkdev_get(bdev, mode, 0);

	if (whole) {
1236 1237
		struct gendisk *disk = whole->bd_disk;

1238
		/* finish claiming */
1239
		mutex_lock(&bdev->bd_mutex);
1240 1241
		spin_lock(&bdev_lock);

1242
		if (!res) {
1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261
			BUG_ON(!bd_may_claim(bdev, whole, holder));
			/*
			 * Note that for a whole device bd_holders
			 * will be incremented twice, and bd_holder
			 * will be set to bd_may_claim before being
			 * set to holder
			 */
			whole->bd_holders++;
			whole->bd_holder = bd_may_claim;
			bdev->bd_holders++;
			bdev->bd_holder = holder;
		}

		/* tell others that we're done */
		BUG_ON(whole->bd_claiming != holder);
		whole->bd_claiming = NULL;
		wake_up_bit(&whole->bd_claiming, 0);

		spin_unlock(&bdev_lock);
1262 1263

		/*
1264 1265 1266 1267 1268
		 * Block event polling for write claims if requested.  Any
		 * write holder makes the write_holder state stick until
		 * all are released.  This is good enough and tracking
		 * individual writeable reference is too fragile given the
		 * way @mode is used in blkdev_get/put().
1269
		 */
1270 1271
		if (!res && (mode & FMODE_WRITE) && !bdev->bd_write_holder &&
		    (disk->flags & GENHD_FL_BLOCK_EVENTS_ON_EXCL_WRITE)) {
1272
			bdev->bd_write_holder = true;
1273
			disk_block_events(disk);
1274 1275 1276
		}

		mutex_unlock(&bdev->bd_mutex);
1277
		bdput(whole);
1278 1279 1280
	}

	return res;
1281
}
L
Linus Torvalds 已提交
1282 1283
EXPORT_SYMBOL(blkdev_get);

1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314
/**
 * blkdev_get_by_path - open a block device by name
 * @path: path to the block device to open
 * @mode: FMODE_* mask
 * @holder: exclusive holder identifier
 *
 * Open the blockdevice described by the device file at @path.  @mode
 * and @holder are identical to blkdev_get().
 *
 * On success, the returned block_device has reference count of one.
 *
 * CONTEXT:
 * Might sleep.
 *
 * RETURNS:
 * Pointer to block_device on success, ERR_PTR(-errno) on failure.
 */
struct block_device *blkdev_get_by_path(const char *path, fmode_t mode,
					void *holder)
{
	struct block_device *bdev;
	int err;

	bdev = lookup_bdev(path);
	if (IS_ERR(bdev))
		return bdev;

	err = blkdev_get(bdev, mode, holder);
	if (err)
		return ERR_PTR(err);

1315 1316 1317 1318 1319
	if ((mode & FMODE_WRITE) && bdev_read_only(bdev)) {
		blkdev_put(bdev, mode);
		return ERR_PTR(-EACCES);
	}

1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362
	return bdev;
}
EXPORT_SYMBOL(blkdev_get_by_path);

/**
 * blkdev_get_by_dev - open a block device by device number
 * @dev: device number of block device to open
 * @mode: FMODE_* mask
 * @holder: exclusive holder identifier
 *
 * Open the blockdevice described by device number @dev.  @mode and
 * @holder are identical to blkdev_get().
 *
 * Use it ONLY if you really do not have anything better - i.e. when
 * you are behind a truly sucky interface and all you are given is a
 * device number.  _Never_ to be used for internal purposes.  If you
 * ever need it - reconsider your API.
 *
 * On success, the returned block_device has reference count of one.
 *
 * CONTEXT:
 * Might sleep.
 *
 * RETURNS:
 * Pointer to block_device on success, ERR_PTR(-errno) on failure.
 */
struct block_device *blkdev_get_by_dev(dev_t dev, fmode_t mode, void *holder)
{
	struct block_device *bdev;
	int err;

	bdev = bdget(dev);
	if (!bdev)
		return ERR_PTR(-ENOMEM);

	err = blkdev_get(bdev, mode, holder);
	if (err)
		return ERR_PTR(err);

	return bdev;
}
EXPORT_SYMBOL(blkdev_get_by_dev);

L
Linus Torvalds 已提交
1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374
static int blkdev_open(struct inode * inode, struct file * filp)
{
	struct block_device *bdev;

	/*
	 * Preserve backwards compatibility and allow large file access
	 * even if userspace doesn't ask for it explicitly. Some mkfs
	 * binary needs it. We might want to drop this workaround
	 * during an unstable branch.
	 */
	filp->f_flags |= O_LARGEFILE;

1375 1376 1377 1378 1379 1380 1381
	if (filp->f_flags & O_NDELAY)
		filp->f_mode |= FMODE_NDELAY;
	if (filp->f_flags & O_EXCL)
		filp->f_mode |= FMODE_EXCL;
	if ((filp->f_flags & O_ACCMODE) == 3)
		filp->f_mode |= FMODE_WRITE_IOCTL;

L
Linus Torvalds 已提交
1382
	bdev = bd_acquire(inode);
1383 1384
	if (bdev == NULL)
		return -ENOMEM;
L
Linus Torvalds 已提交
1385

1386 1387
	filp->f_mapping = bdev->bd_inode->i_mapping;

1388
	return blkdev_get(bdev, filp->f_mode, filp);
L
Linus Torvalds 已提交
1389 1390
}

A
Al Viro 已提交
1391
static void __blkdev_put(struct block_device *bdev, fmode_t mode, int for_part)
1392 1393
{
	struct gendisk *disk = bdev->bd_disk;
1394
	struct block_device *victim = NULL;
1395

1396
	mutex_lock_nested(&bdev->bd_mutex, for_part);
1397 1398 1399
	if (for_part)
		bdev->bd_part_count--;

1400
	if (!--bdev->bd_openers) {
1401
		WARN_ON_ONCE(bdev->bd_holders);
1402 1403
		sync_blockdev(bdev);
		kill_bdev(bdev);
1404 1405 1406 1407 1408
		/* ->release can cause the old bdi to disappear,
		 * so must switch it out first
		 */
		bdev_inode_switch_bdi(bdev->bd_inode,
					&default_backing_dev_info);
1409 1410 1411
	}
	if (bdev->bd_contains == bdev) {
		if (disk->fops->release)
1412
			disk->fops->release(disk, mode);
1413 1414 1415 1416
	}
	if (!bdev->bd_openers) {
		struct module *owner = disk->fops->owner;

T
Tejun Heo 已提交
1417 1418
		disk_put_part(bdev->bd_part);
		bdev->bd_part = NULL;
1419
		bdev->bd_disk = NULL;
1420 1421
		if (bdev != bdev->bd_contains)
			victim = bdev->bd_contains;
1422
		bdev->bd_contains = NULL;
1423 1424 1425

		put_disk(disk);
		module_put(owner);
1426 1427 1428
	}
	mutex_unlock(&bdev->bd_mutex);
	bdput(bdev);
1429
	if (victim)
A
Al Viro 已提交
1430
		__blkdev_put(victim, mode, 1);
1431 1432
}

A
Al Viro 已提交
1433
void blkdev_put(struct block_device *bdev, fmode_t mode)
1434
{
1435 1436
	mutex_lock(&bdev->bd_mutex);

1437
	if (mode & FMODE_EXCL) {
1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457
		bool bdev_free;

		/*
		 * Release a claim on the device.  The holder fields
		 * are protected with bdev_lock.  bd_mutex is to
		 * synchronize disk_holder unlinking.
		 */
		spin_lock(&bdev_lock);

		WARN_ON_ONCE(--bdev->bd_holders < 0);
		WARN_ON_ONCE(--bdev->bd_contains->bd_holders < 0);

		/* bd_contains might point to self, check in a separate step */
		if ((bdev_free = !bdev->bd_holders))
			bdev->bd_holder = NULL;
		if (!bdev->bd_contains->bd_holders)
			bdev->bd_contains->bd_holder = NULL;

		spin_unlock(&bdev_lock);

1458 1459 1460 1461
		/*
		 * If this was the last claim, remove holder link and
		 * unblock evpoll if it was a write holder.
		 */
1462 1463 1464
		if (bdev_free && bdev->bd_write_holder) {
			disk_unblock_events(bdev->bd_disk);
			bdev->bd_write_holder = false;
1465
		}
1466
	}
1467

1468 1469 1470 1471 1472 1473 1474 1475 1476
	/*
	 * Trigger event checking and tell drivers to flush MEDIA_CHANGE
	 * event.  This is to ensure detection of media removal commanded
	 * from userland - e.g. eject(1).
	 */
	disk_flush_events(bdev->bd_disk, DISK_EVENT_MEDIA_CHANGE);

	mutex_unlock(&bdev->bd_mutex);

A
Al Viro 已提交
1477
	__blkdev_put(bdev, mode, 0);
1478
}
1479 1480
EXPORT_SYMBOL(blkdev_put);

L
Linus Torvalds 已提交
1481 1482 1483
static int blkdev_close(struct inode * inode, struct file * filp)
{
	struct block_device *bdev = I_BDEV(filp->f_mapping->host);
A
Al Viro 已提交
1484 1485
	blkdev_put(bdev, filp->f_mode);
	return 0;
L
Linus Torvalds 已提交
1486 1487
}

1488
static long block_ioctl(struct file *file, unsigned cmd, unsigned long arg)
L
Linus Torvalds 已提交
1489
{
1490 1491
	struct block_device *bdev = I_BDEV(file->f_mapping->host);
	fmode_t mode = file->f_mode;
1492 1493 1494 1495 1496

	/*
	 * O_NDELAY can be altered using fcntl(.., F_SETFL, ..), so we have
	 * to updated it before every ioctl.
	 */
1497
	if (file->f_flags & O_NDELAY)
1498 1499 1500 1501
		mode |= FMODE_NDELAY;
	else
		mode &= ~FMODE_NDELAY;

1502
	return blkdev_ioctl(bdev, mode, cmd, arg);
L
Linus Torvalds 已提交
1503 1504
}

1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515
/*
 * Write data to the block device.  Only intended for the block device itself
 * and the raw driver which basically is a fake block device.
 *
 * Does not take i_mutex for the write and thus is not for general purpose
 * use.
 */
ssize_t blkdev_aio_write(struct kiocb *iocb, const struct iovec *iov,
			 unsigned long nr_segs, loff_t pos)
{
	struct file *file = iocb->ki_filp;
1516
	struct blk_plug plug;
1517 1518 1519 1520
	ssize_t ret;

	BUG_ON(iocb->ki_pos != pos);

1521
	blk_start_plug(&plug);
1522
	ret = __generic_file_aio_write(iocb, iov, nr_segs);
1523
	if (ret > 0) {
1524 1525 1526
		ssize_t err;

		err = generic_write_sync(file, pos, ret);
1527
		if (err < 0)
1528 1529
			ret = err;
	}
1530
	blk_finish_plug(&plug);
1531 1532 1533 1534
	return ret;
}
EXPORT_SYMBOL_GPL(blkdev_aio_write);

1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545
static ssize_t blkdev_aio_read(struct kiocb *iocb, const struct iovec *iov,
			 unsigned long nr_segs, loff_t pos)
{
	struct file *file = iocb->ki_filp;
	struct inode *bd_inode = file->f_mapping->host;
	loff_t size = i_size_read(bd_inode);

	if (pos >= size)
		return 0;

	size -= pos;
K
Kent Overstreet 已提交
1546
	if (size < iocb->ki_nbytes)
1547 1548 1549 1550
		nr_segs = iov_shorten((struct iovec *)iov, nr_segs, size);
	return generic_file_aio_read(iocb, iov, nr_segs, pos);
}

1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564
/*
 * Try to release a page associated with block device when the system
 * is under memory pressure.
 */
static int blkdev_releasepage(struct page *page, gfp_t wait)
{
	struct super_block *super = BDEV_I(page->mapping->host)->bdev.bd_super;

	if (super && super->s_op->bdev_try_to_free_page)
		return super->s_op->bdev_try_to_free_page(super, page, wait);

	return try_to_free_buffers(page);
}

A
Adrian Bunk 已提交
1565
static const struct address_space_operations def_blk_aops = {
L
Linus Torvalds 已提交
1566 1567
	.readpage	= blkdev_readpage,
	.writepage	= blkdev_writepage,
N
Nick Piggin 已提交
1568 1569
	.write_begin	= blkdev_write_begin,
	.write_end	= blkdev_write_end,
L
Linus Torvalds 已提交
1570
	.writepages	= generic_writepages,
1571
	.releasepage	= blkdev_releasepage,
L
Linus Torvalds 已提交
1572
	.direct_IO	= blkdev_direct_IO,
1573
	.is_dirty_writeback = buffer_check_dirty_writeback,
L
Linus Torvalds 已提交
1574 1575
};

1576
const struct file_operations def_blk_fops = {
L
Linus Torvalds 已提交
1577 1578 1579
	.open		= blkdev_open,
	.release	= blkdev_close,
	.llseek		= block_llseek,
1580 1581
	.read		= do_sync_read,
	.write		= do_sync_write,
1582
	.aio_read	= blkdev_aio_read,
1583
	.aio_write	= blkdev_aio_write,
1584
	.mmap		= generic_file_mmap,
1585
	.fsync		= blkdev_fsync,
1586
	.unlocked_ioctl	= block_ioctl,
L
Linus Torvalds 已提交
1587 1588 1589
#ifdef CONFIG_COMPAT
	.compat_ioctl	= compat_blkdev_ioctl,
#endif
1590 1591
	.splice_read	= generic_file_splice_read,
	.splice_write	= generic_file_splice_write,
L
Linus Torvalds 已提交
1592 1593 1594 1595 1596 1597 1598
};

int ioctl_by_bdev(struct block_device *bdev, unsigned cmd, unsigned long arg)
{
	int res;
	mm_segment_t old_fs = get_fs();
	set_fs(KERNEL_DS);
1599
	res = blkdev_ioctl(bdev, 0, cmd, arg);
L
Linus Torvalds 已提交
1600 1601 1602 1603 1604 1605 1606 1607
	set_fs(old_fs);
	return res;
}

EXPORT_SYMBOL(ioctl_by_bdev);

/**
 * lookup_bdev  - lookup a struct block_device by name
1608
 * @pathname:	special file representing the block device
L
Linus Torvalds 已提交
1609
 *
1610
 * Get a reference to the blockdevice at @pathname in the current
L
Linus Torvalds 已提交
1611 1612 1613
 * namespace if possible and return it.  Return ERR_PTR(error)
 * otherwise.
 */
1614
struct block_device *lookup_bdev(const char *pathname)
L
Linus Torvalds 已提交
1615 1616 1617
{
	struct block_device *bdev;
	struct inode *inode;
1618
	struct path path;
L
Linus Torvalds 已提交
1619 1620
	int error;

1621
	if (!pathname || !*pathname)
L
Linus Torvalds 已提交
1622 1623
		return ERR_PTR(-EINVAL);

1624
	error = kern_path(pathname, LOOKUP_FOLLOW, &path);
L
Linus Torvalds 已提交
1625 1626 1627
	if (error)
		return ERR_PTR(error);

1628
	inode = path.dentry->d_inode;
L
Linus Torvalds 已提交
1629 1630 1631 1632
	error = -ENOTBLK;
	if (!S_ISBLK(inode->i_mode))
		goto fail;
	error = -EACCES;
1633
	if (path.mnt->mnt_flags & MNT_NODEV)
L
Linus Torvalds 已提交
1634 1635 1636 1637 1638 1639
		goto fail;
	error = -ENOMEM;
	bdev = bd_acquire(inode);
	if (!bdev)
		goto fail;
out:
1640
	path_put(&path);
L
Linus Torvalds 已提交
1641 1642 1643 1644 1645
	return bdev;
fail:
	bdev = ERR_PTR(error);
	goto out;
}
1646
EXPORT_SYMBOL(lookup_bdev);
L
Linus Torvalds 已提交
1647

1648
int __invalidate_device(struct block_device *bdev, bool kill_dirty)
1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660
{
	struct super_block *sb = get_super(bdev);
	int res = 0;

	if (sb) {
		/*
		 * no need to lock the super, get_super holds the
		 * read mutex so the filesystem cannot go away
		 * under us (->put_super runs with the write lock
		 * hold).
		 */
		shrink_dcache_sb(sb);
1661
		res = invalidate_inodes(sb, kill_dirty);
1662 1663
		drop_super(sb);
	}
1664
	invalidate_bdev(bdev);
1665 1666 1667
	return res;
}
EXPORT_SYMBOL(__invalidate_device);
1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703

void iterate_bdevs(void (*func)(struct block_device *, void *), void *arg)
{
	struct inode *inode, *old_inode = NULL;

	spin_lock(&inode_sb_list_lock);
	list_for_each_entry(inode, &blockdev_superblock->s_inodes, i_sb_list) {
		struct address_space *mapping = inode->i_mapping;

		spin_lock(&inode->i_lock);
		if (inode->i_state & (I_FREEING|I_WILL_FREE|I_NEW) ||
		    mapping->nrpages == 0) {
			spin_unlock(&inode->i_lock);
			continue;
		}
		__iget(inode);
		spin_unlock(&inode->i_lock);
		spin_unlock(&inode_sb_list_lock);
		/*
		 * We hold a reference to 'inode' so it couldn't have been
		 * removed from s_inodes list while we dropped the
		 * inode_sb_list_lock.  We cannot iput the inode now as we can
		 * be holding the last reference and we cannot iput it under
		 * inode_sb_list_lock. So we keep the reference and iput it
		 * later.
		 */
		iput(old_inode);
		old_inode = inode;

		func(I_BDEV(inode), arg);

		spin_lock(&inode_sb_list_lock);
	}
	spin_unlock(&inode_sb_list_lock);
	iput(old_inode);
}