block_dev.c 40.9 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 19
#include <linux/highmem.h>
#include <linux/blkdev.h>
#include <linux/module.h>
#include <linux/blkpg.h>
#include <linux/buffer_head.h>
N
Nick Piggin 已提交
20
#include <linux/pagevec.h>
21
#include <linux/writeback.h>
L
Linus Torvalds 已提交
22 23 24 25
#include <linux/mpage.h>
#include <linux/mount.h>
#include <linux/uio.h>
#include <linux/namei.h>
26
#include <linux/log2.h>
27
#include <linux/kmemleak.h>
L
Linus Torvalds 已提交
28
#include <asm/uaccess.h>
29
#include "internal.h"
L
Linus Torvalds 已提交
30 31 32 33 34 35

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

A
Adrian Bunk 已提交
36 37
static const struct address_space_operations def_blk_aops;

L
Linus Torvalds 已提交
38 39 40 41 42 43 44 45 46 47 48
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);

49
/*
50 51 52
 * 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.
53 54 55 56
 */
static void bdev_inode_switch_bdi(struct inode *inode,
			struct backing_dev_info *dst)
{
57 58 59 60 61
	struct backing_dev_info *old = inode->i_data.backing_dev_info;

	if (unlikely(dst == old))		/* deadlock avoidance */
		return;
	bdi_lock_two(&old->wb, &dst->wb);
62
	spin_lock(&inode->i_lock);
63 64
	inode->i_data.backing_dev_info = dst;
	if (inode->i_state & I_DIRTY)
N
Nick Piggin 已提交
65
		list_move(&inode->i_wb_list, &dst->wb.b_dirty);
66
	spin_unlock(&inode->i_lock);
67 68
	spin_unlock(&old->wb.list_lock);
	spin_unlock(&dst->wb.list_lock);
69 70
}

L
Linus Torvalds 已提交
71 72 73 74 75 76 77 78 79 80 81 82 83
static sector_t max_block(struct block_device *bdev)
{
	sector_t retval = ~((sector_t)0);
	loff_t sz = i_size_read(bdev->bd_inode);

	if (sz) {
		unsigned int size = block_size(bdev);
		unsigned int sizebits = blksize_bits(size);
		retval = (sz >> sizebits);
	}
	return retval;
}

P
Peter Zijlstra 已提交
84
/* Kill _all_ buffers and pagecache , dirty or not.. */
L
Linus Torvalds 已提交
85 86
static void kill_bdev(struct block_device *bdev)
{
P
Peter Zijlstra 已提交
87 88 89
	if (bdev->bd_inode->i_mapping->nrpages == 0)
		return;
	invalidate_bh_lrus();
L
Linus Torvalds 已提交
90 91 92 93 94 95
	truncate_inode_pages(bdev->bd_inode->i_mapping, 0);
}	

int set_blocksize(struct block_device *bdev, int size)
{
	/* Size must be a power of two, and between 512 and PAGE_SIZE */
96
	if (size > PAGE_SIZE || size < 512 || !is_power_of_2(size))
L
Linus Torvalds 已提交
97 98 99
		return -EINVAL;

	/* Size cannot be smaller than the size supported by the device */
100
	if (size < bdev_logical_block_size(bdev))
L
Linus Torvalds 已提交
101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121
		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;
122
	sb->s_blocksize_bits = blksize_bits(size);
L
Linus Torvalds 已提交
123 124 125 126 127 128 129
	return sb->s_blocksize;
}

EXPORT_SYMBOL(sb_set_blocksize);

int sb_min_blocksize(struct super_block *sb, int size)
{
130
	int minsize = bdev_logical_block_size(sb->s_bdev);
L
Linus Torvalds 已提交
131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159
	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)
{
	if (iblock >= max_block(I_BDEV(inode))) {
		if (create)
			return -EIO;

		/*
		 * for reads, we're just trying to fill a partial page.
		 * return a hole, they will have to call get_block again
		 * before they can fill it, and they will get -EIO at that
		 * time
		 */
		return 0;
	}
	bh->b_bdev = I_BDEV(inode);
	bh->b_blocknr = iblock;
	set_buffer_mapped(bh);
	return 0;
}

160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194
static int
blkdev_get_blocks(struct inode *inode, sector_t iblock,
		struct buffer_head *bh, int create)
{
	sector_t end_block = max_block(I_BDEV(inode));
	unsigned long max_blocks = bh->b_size >> inode->i_blkbits;

	if ((iblock + max_blocks) > end_block) {
		max_blocks = end_block - iblock;
		if ((long)max_blocks <= 0) {
			if (create)
				return -EIO;	/* write fully beyond EOF */
			/*
			 * It is a read which is fully beyond EOF.  We return
			 * a !buffer_mapped buffer
			 */
			max_blocks = 0;
		}
	}

	bh->b_bdev = I_BDEV(inode);
	bh->b_blocknr = iblock;
	bh->b_size = max_blocks << inode->i_blkbits;
	if (max_blocks)
		set_buffer_mapped(bh);
	return 0;
}

static ssize_t
blkdev_direct_IO(int rw, struct kiocb *iocb, const struct iovec *iov,
			loff_t offset, unsigned long nr_segs)
{
	struct file *file = iocb->ki_filp;
	struct inode *inode = file->f_mapping->host;

195 196
	return __blockdev_direct_IO(rw, iocb, inode, I_BDEV(inode), iov, offset,
				    nr_segs, blkdev_get_blocks, NULL, NULL, 0);
197 198
}

199 200 201 202 203 204 205 206 207
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
208 209 210 211 212 213
/*
 * 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)
{
214
	return __sync_blockdev(bdev, 1);
N
Nick Piggin 已提交
215 216 217 218 219 220 221 222 223 224 225 226
}
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) {
227
		int res = sync_filesystem(sb);
N
Nick Piggin 已提交
228 229 230 231 232
		drop_super(sb);
		return res;
	}
	return sync_blockdev(bdev);
}
233
EXPORT_SYMBOL(fsync_bdev);
N
Nick Piggin 已提交
234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252

/**
 * 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);
253 254 255 256 257 258
	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
259
		sb = get_super(bdev);
260 261 262 263 264 265 266 267
		drop_super(sb);
		mutex_unlock(&bdev->bd_fsfreeze_mutex);
		return sb;
	}

	sb = get_active_super(bdev);
	if (!sb)
		goto out;
268 269 270 271
	error = freeze_super(sb);
	if (error) {
		deactivate_super(sb);
		bdev->bd_fsfreeze_count--;
N
Nick Piggin 已提交
272
		mutex_unlock(&bdev->bd_fsfreeze_mutex);
273
		return ERR_PTR(error);
N
Nick Piggin 已提交
274
	}
275
	deactivate_super(sb);
276
 out:
N
Nick Piggin 已提交
277 278
	sync_blockdev(bdev);
	mutex_unlock(&bdev->bd_fsfreeze_mutex);
279
	return sb;	/* thaw_bdev releases s->s_umount */
N
Nick Piggin 已提交
280 281 282 283 284 285 286 287 288 289 290 291
}
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)
{
292
	int error = -EINVAL;
N
Nick Piggin 已提交
293 294

	mutex_lock(&bdev->bd_fsfreeze_mutex);
295
	if (!bdev->bd_fsfreeze_count)
296
		goto out;
297 298 299

	error = 0;
	if (--bdev->bd_fsfreeze_count > 0)
300
		goto out;
301 302

	if (!sb)
303
		goto out;
304

305 306 307 308 309 310 311
	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
312 313 314 315 316
	mutex_unlock(&bdev->bd_fsfreeze_mutex);
	return 0;
}
EXPORT_SYMBOL(thaw_bdev);

L
Linus Torvalds 已提交
317 318 319 320 321 322 323 324 325 326
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
327 328 329
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
330
{
331 332
	return block_write_begin(mapping, pos, len, flags, pagep,
				 blkdev_get_block);
L
Linus Torvalds 已提交
333 334
}

N
Nick Piggin 已提交
335 336 337
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
338
{
N
Nick Piggin 已提交
339 340 341 342 343 344 345
	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
346 347 348 349
}

/*
 * private llseek:
350
 * for a block special file file->f_path.dentry->d_inode->i_size is zero
L
Linus Torvalds 已提交
351 352 353 354 355 356 357 358
 * so we compute the size by hand (just as in block_read/write above)
 */
static loff_t block_llseek(struct file *file, loff_t offset, int origin)
{
	struct inode *bd_inode = file->f_mapping->host;
	loff_t size;
	loff_t retval;

359
	mutex_lock(&bd_inode->i_mutex);
L
Linus Torvalds 已提交
360 361
	size = i_size_read(bd_inode);

362
	retval = -EINVAL;
L
Linus Torvalds 已提交
363
	switch (origin) {
364
		case SEEK_END:
L
Linus Torvalds 已提交
365 366
			offset += size;
			break;
367
		case SEEK_CUR:
L
Linus Torvalds 已提交
368
			offset += file->f_pos;
369 370 371 372
		case SEEK_SET:
			break;
		default:
			goto out;
L
Linus Torvalds 已提交
373 374 375 376 377 378 379
	}
	if (offset >= 0 && offset <= size) {
		if (offset != file->f_pos) {
			file->f_pos = offset;
		}
		retval = offset;
	}
380
out:
381
	mutex_unlock(&bd_inode->i_mutex);
L
Linus Torvalds 已提交
382 383 384
	return retval;
}
	
385
int blkdev_fsync(struct file *filp, loff_t start, loff_t end, int datasync)
L
Linus Torvalds 已提交
386
{
387 388
	struct inode *bd_inode = filp->f_mapping->host;
	struct block_device *bdev = I_BDEV(bd_inode);
389
	int error;
390 391 392 393
	
	error = filemap_write_and_wait_range(filp->f_mapping, start, end);
	if (error)
		return error;
394

395 396 397 398 399
	/*
	 * 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.
	 */
400
	error = blkdev_issue_flush(bdev, GFP_KERNEL, NULL);
401 402
	if (error == -EOPNOTSUPP)
		error = 0;
403

404
	return error;
L
Linus Torvalds 已提交
405
}
406
EXPORT_SYMBOL(blkdev_fsync);
L
Linus Torvalds 已提交
407 408 409 410 411 412

/*
 * pseudo-fs
 */

static  __cacheline_aligned_in_smp DEFINE_SPINLOCK(bdev_lock);
413
static struct kmem_cache * bdev_cachep __read_mostly;
L
Linus Torvalds 已提交
414 415 416

static struct inode *bdev_alloc_inode(struct super_block *sb)
{
417
	struct bdev_inode *ei = kmem_cache_alloc(bdev_cachep, GFP_KERNEL);
L
Linus Torvalds 已提交
418 419 420 421 422
	if (!ei)
		return NULL;
	return &ei->vfs_inode;
}

N
Nick Piggin 已提交
423
static void bdev_i_callback(struct rcu_head *head)
L
Linus Torvalds 已提交
424
{
N
Nick Piggin 已提交
425
	struct inode *inode = container_of(head, struct inode, i_rcu);
L
Linus Torvalds 已提交
426 427
	struct bdev_inode *bdi = BDEV_I(inode);

N
Nick Piggin 已提交
428
	INIT_LIST_HEAD(&inode->i_dentry);
L
Linus Torvalds 已提交
429 430 431
	kmem_cache_free(bdev_cachep, bdi);
}

N
Nick Piggin 已提交
432 433 434 435 436
static void bdev_destroy_inode(struct inode *inode)
{
	call_rcu(&inode->i_rcu, bdev_i_callback);
}

437
static void init_once(void *foo)
L
Linus Torvalds 已提交
438 439 440 441
{
	struct bdev_inode *ei = (struct bdev_inode *) foo;
	struct block_device *bdev = &ei->bdev;

C
Christoph Lameter 已提交
442 443 444 445
	memset(bdev, 0, sizeof(*bdev));
	mutex_init(&bdev->bd_mutex);
	INIT_LIST_HEAD(&bdev->bd_inodes);
	INIT_LIST_HEAD(&bdev->bd_list);
446 447 448
#ifdef CONFIG_SYSFS
	INIT_LIST_HEAD(&bdev->bd_holder_disks);
#endif
C
Christoph Lameter 已提交
449
	inode_init_once(&ei->vfs_inode);
450 451
	/* Initialize mutex for freeze. */
	mutex_init(&bdev->bd_fsfreeze_mutex);
L
Linus Torvalds 已提交
452 453 454 455 456 457 458 459 460
}

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

461
static void bdev_evict_inode(struct inode *inode)
L
Linus Torvalds 已提交
462 463 464
{
	struct block_device *bdev = &BDEV_I(inode)->bdev;
	struct list_head *p;
465 466 467
	truncate_inode_pages(&inode->i_data, 0);
	invalidate_inode_buffers(inode); /* is it needed here? */
	end_writeback(inode);
L
Linus Torvalds 已提交
468 469 470 471 472 473 474 475
	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);
}

476
static const struct super_operations bdev_sops = {
L
Linus Torvalds 已提交
477 478 479 480
	.statfs = simple_statfs,
	.alloc_inode = bdev_alloc_inode,
	.destroy_inode = bdev_destroy_inode,
	.drop_inode = generic_delete_inode,
481
	.evict_inode = bdev_evict_inode,
L
Linus Torvalds 已提交
482 483
};

A
Al Viro 已提交
484 485
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
486
{
487
	return mount_pseudo(fs_type, "bdev:", &bdev_sops, NULL, 0x62646576);
L
Linus Torvalds 已提交
488 489 490 491
}

static struct file_system_type bd_type = {
	.name		= "bdev",
A
Al Viro 已提交
492
	.mount		= bd_mount,
L
Linus Torvalds 已提交
493 494 495
	.kill_sb	= kill_anon_super,
};

496
struct super_block *blockdev_superblock __read_mostly;
L
Linus Torvalds 已提交
497 498 499 500

void __init bdev_cache_init(void)
{
	int err;
501 502
	struct vfsmount *bd_mnt;

L
Linus Torvalds 已提交
503
	bdev_cachep = kmem_cache_create("bdev_cache", sizeof(struct bdev_inode),
504 505
			0, (SLAB_HWCACHE_ALIGN|SLAB_RECLAIM_ACCOUNT|
				SLAB_MEM_SPREAD|SLAB_PANIC),
506
			init_once);
L
Linus Torvalds 已提交
507 508 509 510 511 512
	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");
513 514 515 516 517
	/*
	 * This vfsmount structure is only used to obtain the
	 * blockdev_superblock, so tell kmemleak not to report it.
	 */
	kmemleak_not_leak(bd_mnt);
L
Linus Torvalds 已提交
518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548
	blockdev_superblock = bd_mnt->mnt_sb;	/* For writeback */
}

/*
 * 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;

549
	inode = iget5_locked(blockdev_superblock, hash(dev),
L
Linus Torvalds 已提交
550 551 552 553 554 555 556 557 558
			bdev_test, bdev_set, &dev);

	if (!inode)
		return NULL;

	bdev = &BDEV_I(inode)->bdev;

	if (inode->i_state & I_NEW) {
		bdev->bd_contains = NULL;
559
		bdev->bd_super = NULL;
L
Linus Torvalds 已提交
560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579
		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);

580 581 582 583 584 585
/**
 * 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
586
	ihold(bdev->bd_inode);
587 588 589
	return bdev;
}

L
Linus Torvalds 已提交
590 591
long nr_blockdev_pages(void)
{
592
	struct block_device *bdev;
L
Linus Torvalds 已提交
593 594
	long ret = 0;
	spin_lock(&bdev_lock);
595
	list_for_each_entry(bdev, &all_bdevs, bd_list) {
L
Linus Torvalds 已提交
596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611
		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;
612

L
Linus Torvalds 已提交
613 614
	spin_lock(&bdev_lock);
	bdev = inode->i_bdev;
615
	if (bdev) {
A
Al Viro 已提交
616
		ihold(bdev->bd_inode);
L
Linus Torvalds 已提交
617 618 619 620
		spin_unlock(&bdev_lock);
		return bdev;
	}
	spin_unlock(&bdev_lock);
621

L
Linus Torvalds 已提交
622 623 624
	bdev = bdget(inode->i_rdev);
	if (bdev) {
		spin_lock(&bdev_lock);
625 626
		if (!inode->i_bdev) {
			/*
A
Al Viro 已提交
627
			 * We take an additional reference to bd_inode,
628 629 630 631
			 * 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
632
			ihold(bdev->bd_inode);
633 634 635 636
			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
637 638 639 640 641 642 643 644 645
		spin_unlock(&bdev_lock);
	}
	return bdev;
}

/* Call when you free inode */

void bd_forget(struct inode *inode)
{
646 647
	struct block_device *bdev = NULL;

L
Linus Torvalds 已提交
648
	spin_lock(&bdev_lock);
649
	if (inode->i_bdev) {
650
		if (!sb_is_blkdev_sb(inode->i_sb))
651
			bdev = inode->i_bdev;
L
Linus Torvalds 已提交
652
		__bd_forget(inode);
653
	}
L
Linus Torvalds 已提交
654
	spin_unlock(&bdev_lock);
655 656 657

	if (bdev)
		iput(bdev->bd_inode);
L
Linus Torvalds 已提交
658 659
}

T
Tejun Heo 已提交
660 661 662 663 664 665
/**
 * 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
666
 * Test whether @bdev can be claimed by @holder.
T
Tejun Heo 已提交
667 668 669 670 671 672 673 674 675
 *
 * 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
676 677
{
	if (bdev->bd_holder == holder)
T
Tejun Heo 已提交
678
		return true;	 /* already a holder */
L
Linus Torvalds 已提交
679
	else if (bdev->bd_holder != NULL)
T
Tejun Heo 已提交
680
		return false; 	 /* held by someone else */
L
Linus Torvalds 已提交
681
	else if (bdev->bd_contains == bdev)
T
Tejun Heo 已提交
682
		return true;  	 /* is a whole device which isn't held */
L
Linus Torvalds 已提交
683

684
	else if (whole->bd_holder == bd_may_claim)
T
Tejun Heo 已提交
685 686 687
		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
688
	else
T
Tejun Heo 已提交
689 690 691
		return true;	 /* is a partition of an un-held device */
}

692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717
/**
 * 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;

718 719
	/* if claiming is already in progress, wait for it to finish */
	if (whole->bd_claiming) {
720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742
		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
743 744 745 746 747 748 749
 * 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.
750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774
 *
 * 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);

775 776 777 778 779 780 781 782 783 784 785 786 787
	/*
	 * 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);

788
	module_put(disk->fops->owner);
789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807
	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);
	}
}

808
#ifdef CONFIG_SYSFS
809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825
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;
}

826
static int add_symlink(struct kobject *from, struct kobject *to)
827
{
828
	return sysfs_create_link(from, to, kobject_name(to));
829 830 831 832 833 834 835
}

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

J
Jun'ichi Nomura 已提交
836
/**
837 838 839
 * 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
840
 *
841 842
 * DON'T USE THIS UNLESS YOU'RE ALREADY USING IT.
 *
843
 * This functions creates the following sysfs symlinks.
844
 *
845 846
 * - from "slaves" directory of the holder @disk to the claimed @bdev
 * - from "holders" directory of the @bdev to the holder @disk
847
 *
848 849
 * For example, if /dev/dm-0 maps to /dev/sda and disk for dm-0 is
 * passed to bd_link_disk_holder(), then:
850
 *
851 852
 *   /sys/block/dm-0/slaves/sda --> /sys/block/sda
 *   /sys/block/sda/holders/dm-0 --> /sys/block/dm-0
853
 *
854 855 856
 * 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.
857
 *
858 859
 * CONTEXT:
 * Might sleep.
860
 *
861 862
 * RETURNS:
 * 0 on success, -errno on failure.
863
 */
864
int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk)
865
{
866
	struct bd_holder_disk *holder;
867
	int ret = 0;
868

869
	mutex_lock(&bdev->bd_mutex);
J
Jun'ichi Nomura 已提交
870

871
	WARN_ON_ONCE(!bdev->bd_holder);
872

873 874 875
	/* FIXME: remove the following once add_disk() handles errors */
	if (WARN_ON(!disk->slave_dir || !bdev->bd_part->holder_dir))
		goto out_unlock;
876

877 878 879
	holder = bd_find_holder_disk(bdev, disk);
	if (holder) {
		holder->refcnt++;
880
		goto out_unlock;
881
	}
882

883 884 885
	holder = kzalloc(sizeof(*holder), GFP_KERNEL);
	if (!holder) {
		ret = -ENOMEM;
886 887
		goto out_unlock;
	}
888

889 890 891 892 893 894 895 896 897 898 899
	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;
900 901 902 903 904
	/*
	 * bdev could be deleted beneath us which would implicitly destroy
	 * the holder directory.  Hold on to it.
	 */
	kobject_get(bdev->bd_part->holder_dir);
905 906 907 908 909 910 911 912

	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);
913
out_unlock:
914
	mutex_unlock(&bdev->bd_mutex);
915
	return ret;
916
}
917
EXPORT_SYMBOL_GPL(bd_link_disk_holder);
918

919 920 921 922 923 924 925 926 927 928 929
/**
 * 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)
930
{
931
	struct bd_holder_disk *holder;
932

933
	mutex_lock(&bdev->bd_mutex);
934

935 936 937 938 939 940
	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);
941
		kobject_put(bdev->bd_part->holder_dir);
942 943 944 945 946
		list_del_init(&holder->list);
		kfree(holder);
	}

	mutex_unlock(&bdev->bd_mutex);
L
Linus Torvalds 已提交
947
}
948
EXPORT_SYMBOL_GPL(bd_unlink_disk_holder);
949
#endif
L
Linus Torvalds 已提交
950

951 952 953 954
/**
 * flush_disk - invalidates all buffer-cache entries on a disk
 *
 * @bdev:      struct block device to be flushed
955
 * @kill_dirty: flag to guide handling of dirty inodes
956 957 958 959 960
 *
 * 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.
 */
961
static void flush_disk(struct block_device *bdev, bool kill_dirty)
962
{
963
	if (__invalidate_device(bdev, kill_dirty)) {
964 965 966 967 968 969 970 971 972 973
		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
974
	if (disk_part_scan_enabled(bdev->bd_disk))
975 976 977
		bdev->bd_invalidated = 1;
}

978
/**
979
 * check_disk_size_change - checks for disk size change and adjusts bdev size.
980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999
 * @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);
1000
		flush_disk(bdev, false);
1001 1002 1003 1004
	}
}
EXPORT_SYMBOL(check_disk_size_change);

1005
/**
1006
 * revalidate_disk - wrapper for lower-level driver's revalidate_disk call-back
1007 1008 1009 1010 1011 1012 1013 1014
 * @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)
{
1015
	struct block_device *bdev;
1016 1017 1018 1019 1020
	int ret = 0;

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

1021 1022 1023 1024 1025 1026 1027 1028
	bdev = bdget_disk(disk, 0);
	if (!bdev)
		return ret;

	mutex_lock(&bdev->bd_mutex);
	check_disk_size_change(disk, bdev);
	mutex_unlock(&bdev->bd_mutex);
	bdput(bdev);
1029 1030 1031 1032
	return ret;
}
EXPORT_SYMBOL(revalidate_disk);

L
Linus Torvalds 已提交
1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044
/*
 * 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;
1045
	const struct block_device_operations *bdops = disk->fops;
1046
	unsigned int events;
L
Linus Torvalds 已提交
1047

1048 1049 1050
	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
1051 1052
		return 0;

1053
	flush_disk(bdev, true);
L
Linus Torvalds 已提交
1054 1055 1056 1057 1058 1059 1060 1061 1062
	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)
{
1063
	unsigned bsize = bdev_logical_block_size(bdev);
L
Linus Torvalds 已提交
1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075

	bdev->bd_inode->i_size = size;
	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
1076
static int __blkdev_put(struct block_device *bdev, fmode_t mode, int for_part);
1077

1078 1079 1080 1081 1082 1083 1084
/*
 * bd_mutex locking:
 *
 *  mutex_lock(part->bd_mutex)
 *    mutex_lock_nested(whole->bd_mutex, 1)
 */

1085
static int __blkdev_get(struct block_device *bdev, fmode_t mode, int for_part)
L
Linus Torvalds 已提交
1086 1087
{
	struct gendisk *disk;
1088
	struct module *owner;
1089
	int ret;
1090
	int partno;
1091 1092
	int perm = 0;

1093
	if (mode & FMODE_READ)
1094
		perm |= MAY_READ;
1095
	if (mode & FMODE_WRITE)
1096 1097 1098 1099
		perm |= MAY_WRITE;
	/*
	 * hooks: /n/, see "layering violations".
	 */
1100 1101 1102 1103 1104 1105
	if (!for_part) {
		ret = devcgroup_inode_permission(bdev->bd_inode, perm);
		if (ret != 0) {
			bdput(bdev);
			return ret;
		}
1106
	}
1107

1108
 restart:
T
Tejun Heo 已提交
1109

1110
	ret = -ENXIO;
1111
	disk = get_gendisk(bdev->bd_dev, &partno);
T
Tejun Heo 已提交
1112
	if (!disk)
1113
		goto out;
1114
	owner = disk->fops->owner;
L
Linus Torvalds 已提交
1115

1116
	disk_block_events(disk);
1117
	mutex_lock_nested(&bdev->bd_mutex, for_part);
L
Linus Torvalds 已提交
1118 1119 1120
	if (!bdev->bd_openers) {
		bdev->bd_disk = disk;
		bdev->bd_contains = bdev;
1121
		if (!partno) {
L
Linus Torvalds 已提交
1122
			struct backing_dev_info *bdi;
1123 1124 1125 1126 1127 1128

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

1129
			ret = 0;
L
Linus Torvalds 已提交
1130
			if (disk->fops->open) {
1131
				ret = disk->fops->open(bdev, mode);
1132 1133 1134 1135 1136 1137 1138 1139 1140
				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;
					mutex_unlock(&bdev->bd_mutex);
1141 1142
					disk_unblock_events(disk);
					put_disk(disk);
1143
					module_put(owner);
1144 1145
					goto restart;
				}
L
Linus Torvalds 已提交
1146
			}
1147 1148 1149 1150 1151 1152 1153 1154 1155

			if (!ret && !bdev->bd_openers) {
				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);
			}

1156 1157 1158 1159 1160 1161 1162 1163 1164 1165
			/*
			 * 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.
			 */
			if (bdev->bd_invalidated && (!ret || ret == -ENOMEDIUM))
				rescan_partitions(disk, bdev);
			if (ret)
				goto out_clear;
L
Linus Torvalds 已提交
1166 1167 1168 1169 1170
		} else {
			struct block_device *whole;
			whole = bdget_disk(disk, 0);
			ret = -ENOMEM;
			if (!whole)
T
Tejun Heo 已提交
1171
				goto out_clear;
1172
			BUG_ON(for_part);
1173
			ret = __blkdev_get(whole, mode, 1);
L
Linus Torvalds 已提交
1174
			if (ret)
T
Tejun Heo 已提交
1175
				goto out_clear;
L
Linus Torvalds 已提交
1176
			bdev->bd_contains = whole;
1177 1178
			bdev_inode_switch_bdi(bdev->bd_inode,
				whole->bd_inode->i_data.backing_dev_info);
1179
			bdev->bd_part = disk_get_part(disk, partno);
1180
			if (!(disk->flags & GENHD_FL_UP) ||
1181
			    !bdev->bd_part || !bdev->bd_part->nr_sects) {
L
Linus Torvalds 已提交
1182
				ret = -ENXIO;
T
Tejun Heo 已提交
1183
				goto out_clear;
L
Linus Torvalds 已提交
1184
			}
1185
			bd_set_size(bdev, (loff_t)bdev->bd_part->nr_sects << 9);
L
Linus Torvalds 已提交
1186 1187 1188
		}
	} else {
		if (bdev->bd_contains == bdev) {
1189 1190
			ret = 0;
			if (bdev->bd_disk->fops->open)
1191
				ret = bdev->bd_disk->fops->open(bdev, mode);
1192 1193
			/* the same as first opener case, read comment there */
			if (bdev->bd_invalidated && (!ret || ret == -ENOMEDIUM))
L
Linus Torvalds 已提交
1194
				rescan_partitions(bdev->bd_disk, bdev);
1195 1196
			if (ret)
				goto out_unlock_bdev;
L
Linus Torvalds 已提交
1197
		}
1198 1199
		/* only one opener holds refs to the module and disk */
		put_disk(disk);
1200
		module_put(owner);
L
Linus Torvalds 已提交
1201 1202
	}
	bdev->bd_openers++;
1203 1204
	if (for_part)
		bdev->bd_part_count++;
1205
	mutex_unlock(&bdev->bd_mutex);
1206
	disk_unblock_events(disk);
L
Linus Torvalds 已提交
1207 1208
	return 0;

T
Tejun Heo 已提交
1209
 out_clear:
1210
	disk_put_part(bdev->bd_part);
L
Linus Torvalds 已提交
1211
	bdev->bd_disk = NULL;
T
Tejun Heo 已提交
1212
	bdev->bd_part = NULL;
1213
	bdev_inode_switch_bdi(bdev->bd_inode, &default_backing_dev_info);
L
Linus Torvalds 已提交
1214
	if (bdev != bdev->bd_contains)
1215
		__blkdev_put(bdev->bd_contains, mode, 1);
L
Linus Torvalds 已提交
1216
	bdev->bd_contains = NULL;
T
Tejun Heo 已提交
1217
 out_unlock_bdev:
1218
	mutex_unlock(&bdev->bd_mutex);
1219
	disk_unblock_events(disk);
T
Tejun Heo 已提交
1220
	put_disk(disk);
1221
	module_put(owner);
1222
 out:
T
Tejun Heo 已提交
1223 1224
	bdput(bdev);

L
Linus Torvalds 已提交
1225 1226 1227
	return ret;
}

1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246
/**
 * 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.
 */
1247
int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder)
L
Linus Torvalds 已提交
1248
{
1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264
	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) {
1265 1266
		struct gendisk *disk = whole->bd_disk;

1267
		/* finish claiming */
1268
		mutex_lock(&bdev->bd_mutex);
1269 1270
		spin_lock(&bdev_lock);

1271
		if (!res) {
1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290
			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);
1291 1292

		/*
1293 1294 1295 1296 1297
		 * 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().
1298
		 */
1299 1300
		if (!res && (mode & FMODE_WRITE) && !bdev->bd_write_holder &&
		    (disk->flags & GENHD_FL_BLOCK_EVENTS_ON_EXCL_WRITE)) {
1301
			bdev->bd_write_holder = true;
1302
			disk_block_events(disk);
1303 1304 1305
		}

		mutex_unlock(&bdev->bd_mutex);
1306
		bdput(whole);
1307 1308 1309
	}

	return res;
1310
}
L
Linus Torvalds 已提交
1311 1312
EXPORT_SYMBOL(blkdev_get);

1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343
/**
 * 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);

1344 1345 1346 1347 1348
	if ((mode & FMODE_WRITE) && bdev_read_only(bdev)) {
		blkdev_put(bdev, mode);
		return ERR_PTR(-EACCES);
	}

1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391
	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
1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403
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;

1404 1405 1406 1407 1408 1409 1410
	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
1411
	bdev = bd_acquire(inode);
1412 1413
	if (bdev == NULL)
		return -ENOMEM;
L
Linus Torvalds 已提交
1414

1415 1416
	filp->f_mapping = bdev->bd_inode->i_mapping;

1417
	return blkdev_get(bdev, filp->f_mode, filp);
L
Linus Torvalds 已提交
1418 1419
}

A
Al Viro 已提交
1420
static int __blkdev_put(struct block_device *bdev, fmode_t mode, int for_part)
1421 1422 1423
{
	int ret = 0;
	struct gendisk *disk = bdev->bd_disk;
1424
	struct block_device *victim = NULL;
1425

1426
	mutex_lock_nested(&bdev->bd_mutex, for_part);
1427 1428 1429
	if (for_part)
		bdev->bd_part_count--;

1430
	if (!--bdev->bd_openers) {
1431
		WARN_ON_ONCE(bdev->bd_holders);
1432 1433
		sync_blockdev(bdev);
		kill_bdev(bdev);
1434 1435 1436 1437 1438
		/* ->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);
1439 1440 1441
	}
	if (bdev->bd_contains == bdev) {
		if (disk->fops->release)
A
Al Viro 已提交
1442
			ret = disk->fops->release(disk, mode);
1443 1444 1445 1446
	}
	if (!bdev->bd_openers) {
		struct module *owner = disk->fops->owner;

T
Tejun Heo 已提交
1447 1448
		disk_put_part(bdev->bd_part);
		bdev->bd_part = NULL;
1449
		bdev->bd_disk = NULL;
1450 1451
		if (bdev != bdev->bd_contains)
			victim = bdev->bd_contains;
1452
		bdev->bd_contains = NULL;
1453 1454 1455

		put_disk(disk);
		module_put(owner);
1456 1457 1458
	}
	mutex_unlock(&bdev->bd_mutex);
	bdput(bdev);
1459
	if (victim)
A
Al Viro 已提交
1460
		__blkdev_put(victim, mode, 1);
1461 1462 1463
	return ret;
}

A
Al Viro 已提交
1464
int blkdev_put(struct block_device *bdev, fmode_t mode)
1465
{
1466 1467
	mutex_lock(&bdev->bd_mutex);

1468
	if (mode & FMODE_EXCL) {
1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488
		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);

1489 1490 1491 1492
		/*
		 * If this was the last claim, remove holder link and
		 * unblock evpoll if it was a write holder.
		 */
1493 1494 1495
		if (bdev_free && bdev->bd_write_holder) {
			disk_unblock_events(bdev->bd_disk);
			bdev->bd_write_holder = false;
1496
		}
1497
	}
1498

1499 1500 1501 1502 1503 1504 1505 1506 1507
	/*
	 * 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
1508
	return __blkdev_put(bdev, mode, 0);
1509
}
1510 1511
EXPORT_SYMBOL(blkdev_put);

L
Linus Torvalds 已提交
1512 1513 1514
static int blkdev_close(struct inode * inode, struct file * filp)
{
	struct block_device *bdev = I_BDEV(filp->f_mapping->host);
1515

A
Al Viro 已提交
1516
	return blkdev_put(bdev, filp->f_mode);
L
Linus Torvalds 已提交
1517 1518
}

1519
static long block_ioctl(struct file *file, unsigned cmd, unsigned long arg)
L
Linus Torvalds 已提交
1520
{
1521 1522
	struct block_device *bdev = I_BDEV(file->f_mapping->host);
	fmode_t mode = file->f_mode;
1523 1524 1525 1526 1527

	/*
	 * O_NDELAY can be altered using fcntl(.., F_SETFL, ..), so we have
	 * to updated it before every ioctl.
	 */
1528
	if (file->f_flags & O_NDELAY)
1529 1530 1531 1532
		mode |= FMODE_NDELAY;
	else
		mode &= ~FMODE_NDELAY;

1533
	return blkdev_ioctl(bdev, mode, cmd, arg);
L
Linus Torvalds 已提交
1534 1535
}

1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562
/*
 * 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;
	ssize_t ret;

	BUG_ON(iocb->ki_pos != pos);

	ret = __generic_file_aio_write(iocb, iov, nr_segs, &iocb->ki_pos);
	if (ret > 0 || ret == -EIOCBQUEUED) {
		ssize_t err;

		err = generic_write_sync(file, pos, ret);
		if (err < 0 && ret > 0)
			ret = err;
	}
	return ret;
}
EXPORT_SYMBOL_GPL(blkdev_aio_write);

1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576
/*
 * 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
1577
static const struct address_space_operations def_blk_aops = {
L
Linus Torvalds 已提交
1578 1579
	.readpage	= blkdev_readpage,
	.writepage	= blkdev_writepage,
N
Nick Piggin 已提交
1580 1581
	.write_begin	= blkdev_write_begin,
	.write_end	= blkdev_write_end,
L
Linus Torvalds 已提交
1582
	.writepages	= generic_writepages,
1583
	.releasepage	= blkdev_releasepage,
L
Linus Torvalds 已提交
1584 1585 1586
	.direct_IO	= blkdev_direct_IO,
};

1587
const struct file_operations def_blk_fops = {
L
Linus Torvalds 已提交
1588 1589 1590
	.open		= blkdev_open,
	.release	= blkdev_close,
	.llseek		= block_llseek,
1591 1592
	.read		= do_sync_read,
	.write		= do_sync_write,
L
Linus Torvalds 已提交
1593
  	.aio_read	= generic_file_aio_read,
1594
	.aio_write	= blkdev_aio_write,
L
Linus Torvalds 已提交
1595
	.mmap		= generic_file_mmap,
1596
	.fsync		= blkdev_fsync,
1597
	.unlocked_ioctl	= block_ioctl,
L
Linus Torvalds 已提交
1598 1599 1600
#ifdef CONFIG_COMPAT
	.compat_ioctl	= compat_blkdev_ioctl,
#endif
1601 1602
	.splice_read	= generic_file_splice_read,
	.splice_write	= generic_file_splice_write,
L
Linus Torvalds 已提交
1603 1604 1605 1606 1607 1608 1609
};

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);
1610
	res = blkdev_ioctl(bdev, 0, cmd, arg);
L
Linus Torvalds 已提交
1611 1612 1613 1614 1615 1616 1617 1618
	set_fs(old_fs);
	return res;
}

EXPORT_SYMBOL(ioctl_by_bdev);

/**
 * lookup_bdev  - lookup a struct block_device by name
1619
 * @pathname:	special file representing the block device
L
Linus Torvalds 已提交
1620
 *
1621
 * Get a reference to the blockdevice at @pathname in the current
L
Linus Torvalds 已提交
1622 1623 1624
 * namespace if possible and return it.  Return ERR_PTR(error)
 * otherwise.
 */
1625
struct block_device *lookup_bdev(const char *pathname)
L
Linus Torvalds 已提交
1626 1627 1628
{
	struct block_device *bdev;
	struct inode *inode;
1629
	struct path path;
L
Linus Torvalds 已提交
1630 1631
	int error;

1632
	if (!pathname || !*pathname)
L
Linus Torvalds 已提交
1633 1634
		return ERR_PTR(-EINVAL);

1635
	error = kern_path(pathname, LOOKUP_FOLLOW, &path);
L
Linus Torvalds 已提交
1636 1637 1638
	if (error)
		return ERR_PTR(error);

1639
	inode = path.dentry->d_inode;
L
Linus Torvalds 已提交
1640 1641 1642 1643
	error = -ENOTBLK;
	if (!S_ISBLK(inode->i_mode))
		goto fail;
	error = -EACCES;
1644
	if (path.mnt->mnt_flags & MNT_NODEV)
L
Linus Torvalds 已提交
1645 1646 1647 1648 1649 1650
		goto fail;
	error = -ENOMEM;
	bdev = bd_acquire(inode);
	if (!bdev)
		goto fail;
out:
1651
	path_put(&path);
L
Linus Torvalds 已提交
1652 1653 1654 1655 1656
	return bdev;
fail:
	bdev = ERR_PTR(error);
	goto out;
}
1657
EXPORT_SYMBOL(lookup_bdev);
L
Linus Torvalds 已提交
1658

1659
int __invalidate_device(struct block_device *bdev, bool kill_dirty)
1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671
{
	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);
1672
		res = invalidate_inodes(sb, kill_dirty);
1673 1674
		drop_super(sb);
	}
1675
	invalidate_bdev(bdev);
1676 1677 1678
	return res;
}
EXPORT_SYMBOL(__invalidate_device);