block_dev.c 45.8 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
#include <linux/highmem.h>
#include <linux/blkdev.h>
17
#include <linux/backing-dev.h>
L
Linus Torvalds 已提交
18 19
#include <linux/module.h>
#include <linux/blkpg.h>
20
#include <linux/magic.h>
L
Linus Torvalds 已提交
21
#include <linux/buffer_head.h>
A
Al Viro 已提交
22
#include <linux/swap.h>
N
Nick Piggin 已提交
23
#include <linux/pagevec.h>
24
#include <linux/writeback.h>
L
Linus Torvalds 已提交
25 26 27 28
#include <linux/mpage.h>
#include <linux/mount.h>
#include <linux/uio.h>
#include <linux/namei.h>
29
#include <linux/log2.h>
A
Al Viro 已提交
30
#include <linux/cleancache.h>
31
#include <linux/dax.h>
L
Linus Torvalds 已提交
32
#include <asm/uaccess.h>
33
#include "internal.h"
L
Linus Torvalds 已提交
34 35 36 37 38 39

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

A
Adrian Bunk 已提交
40 41
static const struct address_space_operations def_blk_aops;

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

47
struct block_device *I_BDEV(struct inode *inode)
L
Linus Torvalds 已提交
48 49 50 51 52
{
	return &BDEV_I(inode)->bdev;
}
EXPORT_SYMBOL(I_BDEV);

53 54 55 56 57 58 59 60 61 62 63
static void bdev_write_inode(struct inode *inode)
{
	spin_lock(&inode->i_lock);
	while (inode->i_state & I_DIRTY) {
		spin_unlock(&inode->i_lock);
		WARN_ON_ONCE(write_inode_now(inode, true));
		spin_lock(&inode->i_lock);
	}
	spin_unlock(&inode->i_lock);
}

P
Peter Zijlstra 已提交
64
/* Kill _all_ buffers and pagecache , dirty or not.. */
A
Al Viro 已提交
65
void kill_bdev(struct block_device *bdev)
L
Linus Torvalds 已提交
66
{
A
Al Viro 已提交
67 68
	struct address_space *mapping = bdev->bd_inode->i_mapping;

69
	if (mapping->nrpages == 0 && mapping->nrshadows == 0)
P
Peter Zijlstra 已提交
70
		return;
A
Al Viro 已提交
71

P
Peter Zijlstra 已提交
72
	invalidate_bh_lrus();
A
Al Viro 已提交
73
	truncate_inode_pages(mapping, 0);
L
Linus Torvalds 已提交
74
}	
A
Al Viro 已提交
75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90
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
	 */
91
	cleancache_invalidate_inode(mapping);
A
Al Viro 已提交
92 93
}
EXPORT_SYMBOL(invalidate_bdev);
L
Linus Torvalds 已提交
94 95 96 97

int set_blocksize(struct block_device *bdev, int size)
{
	/* Size must be a power of two, and between 512 and PAGE_SIZE */
98
	if (size > PAGE_SIZE || size < 512 || !is_power_of_2(size))
L
Linus Torvalds 已提交
99 100 101
		return -EINVAL;

	/* Size cannot be smaller than the size supported by the device */
102
	if (size < bdev_logical_block_size(bdev))
L
Linus Torvalds 已提交
103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123
		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;
124
	sb->s_blocksize_bits = blksize_bits(size);
L
Linus Torvalds 已提交
125 126 127 128 129 130 131
	return sb->s_blocksize;
}

EXPORT_SYMBOL(sb_set_blocksize);

int sb_min_blocksize(struct super_block *sb, int size)
{
132
	int minsize = bdev_logical_block_size(sb->s_bdev);
L
Linus Torvalds 已提交
133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149
	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;
}

150
static ssize_t
151
blkdev_direct_IO(struct kiocb *iocb, struct iov_iter *iter, loff_t offset)
152 153 154 155
{
	struct file *file = iocb->ki_filp;
	struct inode *inode = file->f_mapping->host;

156 157 158
	if (IS_DAX(inode))
		return dax_do_io(iocb, inode, iter, offset, blkdev_get_block,
				NULL, DIO_SKIP_DIO_COUNT);
159
	return __blockdev_direct_IO(iocb, inode, I_BDEV(inode), iter, offset,
160 161
				    blkdev_get_block, NULL, NULL,
				    DIO_SKIP_DIO_COUNT);
162 163
}

164 165 166 167 168 169 170 171 172
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
173 174 175 176 177 178
/*
 * 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)
{
179
	return __sync_blockdev(bdev, 1);
N
Nick Piggin 已提交
180 181 182 183 184 185 186 187 188 189 190 191
}
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) {
192
		int res = sync_filesystem(sb);
N
Nick Piggin 已提交
193 194 195 196 197
		drop_super(sb);
		return res;
	}
	return sync_blockdev(bdev);
}
198
EXPORT_SYMBOL(fsync_bdev);
N
Nick Piggin 已提交
199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217

/**
 * 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);
218 219 220 221 222 223
	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
224
		sb = get_super(bdev);
225 226 227 228 229 230 231 232
		drop_super(sb);
		mutex_unlock(&bdev->bd_fsfreeze_mutex);
		return sb;
	}

	sb = get_active_super(bdev);
	if (!sb)
		goto out;
233 234 235 236
	if (sb->s_op->freeze_super)
		error = sb->s_op->freeze_super(sb);
	else
		error = freeze_super(sb);
237 238 239
	if (error) {
		deactivate_super(sb);
		bdev->bd_fsfreeze_count--;
N
Nick Piggin 已提交
240
		mutex_unlock(&bdev->bd_fsfreeze_mutex);
241
		return ERR_PTR(error);
N
Nick Piggin 已提交
242
	}
243
	deactivate_super(sb);
244
 out:
N
Nick Piggin 已提交
245 246
	sync_blockdev(bdev);
	mutex_unlock(&bdev->bd_fsfreeze_mutex);
247
	return sb;	/* thaw_bdev releases s->s_umount */
N
Nick Piggin 已提交
248 249 250 251 252 253 254 255 256 257 258 259
}
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)
{
260
	int error = -EINVAL;
N
Nick Piggin 已提交
261 262

	mutex_lock(&bdev->bd_fsfreeze_mutex);
263
	if (!bdev->bd_fsfreeze_count)
264
		goto out;
265 266 267

	error = 0;
	if (--bdev->bd_fsfreeze_count > 0)
268
		goto out;
269 270

	if (!sb)
271
		goto out;
272

273 274 275 276
	if (sb->s_op->thaw_super)
		error = sb->s_op->thaw_super(sb);
	else
		error = thaw_super(sb);
277 278 279 280 281 282
	if (error) {
		bdev->bd_fsfreeze_count++;
		mutex_unlock(&bdev->bd_fsfreeze_mutex);
		return error;
	}
out:
N
Nick Piggin 已提交
283 284 285 286 287
	mutex_unlock(&bdev->bd_fsfreeze_mutex);
	return 0;
}
EXPORT_SYMBOL(thaw_bdev);

L
Linus Torvalds 已提交
288 289 290 291 292 293 294 295 296 297
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);
}

298 299 300 301 302 303
static int blkdev_readpages(struct file *file, struct address_space *mapping,
			struct list_head *pages, unsigned nr_pages)
{
	return mpage_readpages(mapping, pages, nr_pages, blkdev_get_block);
}

N
Nick Piggin 已提交
304 305 306
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
307
{
308 309
	return block_write_begin(mapping, pos, len, flags, pagep,
				 blkdev_get_block);
L
Linus Torvalds 已提交
310 311
}

N
Nick Piggin 已提交
312 313 314
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
315
{
N
Nick Piggin 已提交
316 317 318 319 320 321 322
	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
323 324 325 326
}

/*
 * private llseek:
A
Al Viro 已提交
327
 * for a block special file file_inode(file)->i_size is zero
L
Linus Torvalds 已提交
328 329
 * so we compute the size by hand (just as in block_read/write above)
 */
330
static loff_t block_llseek(struct file *file, loff_t offset, int whence)
L
Linus Torvalds 已提交
331 332 333 334
{
	struct inode *bd_inode = file->f_mapping->host;
	loff_t retval;

335
	mutex_lock(&bd_inode->i_mutex);
336
	retval = fixed_size_llseek(file, offset, whence, i_size_read(bd_inode));
337
	mutex_unlock(&bd_inode->i_mutex);
L
Linus Torvalds 已提交
338 339 340
	return retval;
}
	
341
int blkdev_fsync(struct file *filp, loff_t start, loff_t end, int datasync)
L
Linus Torvalds 已提交
342
{
343 344
	struct inode *bd_inode = filp->f_mapping->host;
	struct block_device *bdev = I_BDEV(bd_inode);
345
	int error;
346 347 348 349
	
	error = filemap_write_and_wait_range(filp->f_mapping, start, end);
	if (error)
		return error;
350

351 352 353 354 355
	/*
	 * 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.
	 */
356
	error = blkdev_issue_flush(bdev, GFP_KERNEL, NULL);
357 358
	if (error == -EOPNOTSUPP)
		error = 0;
359

360
	return error;
L
Linus Torvalds 已提交
361
}
362
EXPORT_SYMBOL(blkdev_fsync);
L
Linus Torvalds 已提交
363

364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383
/**
 * bdev_read_page() - Start reading a page from a block device
 * @bdev: The device to read the page from
 * @sector: The offset on the device to read the page to (need not be aligned)
 * @page: The page to read
 *
 * On entry, the page should be locked.  It will be unlocked when the page
 * has been read.  If the block driver implements rw_page synchronously,
 * that will be true on exit from this function, but it need not be.
 *
 * Errors returned by this function are usually "soft", eg out of memory, or
 * queue full; callers should try a different route to read this page rather
 * than propagate an error back up the stack.
 *
 * Return: negative errno if an error occurs, 0 if submission was successful.
 */
int bdev_read_page(struct block_device *bdev, sector_t sector,
			struct page *page)
{
	const struct block_device_operations *ops = bdev->bd_disk->fops;
384
	if (!ops->rw_page || bdev_get_integrity(bdev))
385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414
		return -EOPNOTSUPP;
	return ops->rw_page(bdev, sector + get_start_sect(bdev), page, READ);
}
EXPORT_SYMBOL_GPL(bdev_read_page);

/**
 * bdev_write_page() - Start writing a page to a block device
 * @bdev: The device to write the page to
 * @sector: The offset on the device to write the page to (need not be aligned)
 * @page: The page to write
 * @wbc: The writeback_control for the write
 *
 * On entry, the page should be locked and not currently under writeback.
 * On exit, if the write started successfully, the page will be unlocked and
 * under writeback.  If the write failed already (eg the driver failed to
 * queue the page to the device), the page will still be locked.  If the
 * caller is a ->writepage implementation, it will need to unlock the page.
 *
 * Errors returned by this function are usually "soft", eg out of memory, or
 * queue full; callers should try a different route to write this page rather
 * than propagate an error back up the stack.
 *
 * Return: negative errno if an error occurs, 0 if submission was successful.
 */
int bdev_write_page(struct block_device *bdev, sector_t sector,
			struct page *page, struct writeback_control *wbc)
{
	int result;
	int rw = (wbc->sync_mode == WB_SYNC_ALL) ? WRITE_SYNC : WRITE;
	const struct block_device_operations *ops = bdev->bd_disk->fops;
415
	if (!ops->rw_page || bdev_get_integrity(bdev))
416 417 418 419 420 421 422 423 424 425 426
		return -EOPNOTSUPP;
	set_page_writeback(page);
	result = ops->rw_page(bdev, sector + get_start_sect(bdev), page, rw);
	if (result)
		end_page_writeback(page);
	else
		unlock_page(page);
	return result;
}
EXPORT_SYMBOL_GPL(bdev_write_page);

427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444
/**
 * bdev_direct_access() - Get the address for directly-accessibly memory
 * @bdev: The device containing the memory
 * @sector: The offset within the device
 * @addr: Where to put the address of the memory
 * @pfn: The Page Frame Number for the memory
 * @size: The number of bytes requested
 *
 * If a block device is made up of directly addressable memory, this function
 * will tell the caller the PFN and the address of the memory.  The address
 * may be directly dereferenced within the kernel without the need to call
 * ioremap(), kmap() or similar.  The PFN is suitable for inserting into
 * page tables.
 *
 * Return: negative errno if an error occurs, otherwise the number of bytes
 * accessible at this address.
 */
long bdev_direct_access(struct block_device *bdev, sector_t sector,
445
			void __pmem **addr, unsigned long *pfn, long size)
446 447 448 449
{
	long avail;
	const struct block_device_operations *ops = bdev->bd_disk->fops;

450 451 452 453 454 455
	/*
	 * The device driver is allowed to sleep, in order to make the
	 * memory directly accessible.
	 */
	might_sleep();

456 457 458 459 460 461 462 463 464 465
	if (size < 0)
		return size;
	if (!ops->direct_access)
		return -EOPNOTSUPP;
	if ((sector + DIV_ROUND_UP(size, 512)) >
					part_nr_sects_read(bdev->bd_part))
		return -ERANGE;
	sector += get_start_sect(bdev);
	if (sector % (PAGE_SIZE / 512))
		return -EINVAL;
466
	avail = ops->direct_access(bdev, sector, addr, pfn);
467 468 469 470 471 472
	if (!avail)
		return -ERANGE;
	return min(avail, size);
}
EXPORT_SYMBOL_GPL(bdev_direct_access);

L
Linus Torvalds 已提交
473 474 475 476 477
/*
 * pseudo-fs
 */

static  __cacheline_aligned_in_smp DEFINE_SPINLOCK(bdev_lock);
478
static struct kmem_cache * bdev_cachep __read_mostly;
L
Linus Torvalds 已提交
479 480 481

static struct inode *bdev_alloc_inode(struct super_block *sb)
{
482
	struct bdev_inode *ei = kmem_cache_alloc(bdev_cachep, GFP_KERNEL);
L
Linus Torvalds 已提交
483 484 485 486 487
	if (!ei)
		return NULL;
	return &ei->vfs_inode;
}

N
Nick Piggin 已提交
488
static void bdev_i_callback(struct rcu_head *head)
L
Linus Torvalds 已提交
489
{
N
Nick Piggin 已提交
490
	struct inode *inode = container_of(head, struct inode, i_rcu);
L
Linus Torvalds 已提交
491 492 493 494 495
	struct bdev_inode *bdi = BDEV_I(inode);

	kmem_cache_free(bdev_cachep, bdi);
}

N
Nick Piggin 已提交
496 497 498 499 500
static void bdev_destroy_inode(struct inode *inode)
{
	call_rcu(&inode->i_rcu, bdev_i_callback);
}

501
static void init_once(void *foo)
L
Linus Torvalds 已提交
502 503 504 505
{
	struct bdev_inode *ei = (struct bdev_inode *) foo;
	struct block_device *bdev = &ei->bdev;

C
Christoph Lameter 已提交
506 507 508 509
	memset(bdev, 0, sizeof(*bdev));
	mutex_init(&bdev->bd_mutex);
	INIT_LIST_HEAD(&bdev->bd_inodes);
	INIT_LIST_HEAD(&bdev->bd_list);
510 511 512
#ifdef CONFIG_SYSFS
	INIT_LIST_HEAD(&bdev->bd_holder_disks);
#endif
C
Christoph Lameter 已提交
513
	inode_init_once(&ei->vfs_inode);
514 515
	/* Initialize mutex for freeze. */
	mutex_init(&bdev->bd_fsfreeze_mutex);
L
Linus Torvalds 已提交
516 517 518 519 520 521 522 523 524
}

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

525
static void bdev_evict_inode(struct inode *inode)
L
Linus Torvalds 已提交
526 527 528
{
	struct block_device *bdev = &BDEV_I(inode)->bdev;
	struct list_head *p;
529
	truncate_inode_pages_final(&inode->i_data);
530
	invalidate_inode_buffers(inode); /* is it needed here? */
531
	clear_inode(inode);
L
Linus Torvalds 已提交
532 533 534 535 536 537 538 539
	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);
}

540
static const struct super_operations bdev_sops = {
L
Linus Torvalds 已提交
541 542 543 544
	.statfs = simple_statfs,
	.alloc_inode = bdev_alloc_inode,
	.destroy_inode = bdev_destroy_inode,
	.drop_inode = generic_delete_inode,
545
	.evict_inode = bdev_evict_inode,
L
Linus Torvalds 已提交
546 547
};

A
Al Viro 已提交
548 549
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
550
{
551
	return mount_pseudo(fs_type, "bdev:", &bdev_sops, NULL, BDEVFS_MAGIC);
L
Linus Torvalds 已提交
552 553 554 555
}

static struct file_system_type bd_type = {
	.name		= "bdev",
A
Al Viro 已提交
556
	.mount		= bd_mount,
L
Linus Torvalds 已提交
557 558 559
	.kill_sb	= kill_anon_super,
};

T
Tejun Heo 已提交
560 561
struct super_block *blockdev_superblock __read_mostly;
EXPORT_SYMBOL_GPL(blockdev_superblock);
L
Linus Torvalds 已提交
562 563 564 565

void __init bdev_cache_init(void)
{
	int err;
566
	static struct vfsmount *bd_mnt;
567

L
Linus Torvalds 已提交
568
	bdev_cachep = kmem_cache_create("bdev_cache", sizeof(struct bdev_inode),
569 570
			0, (SLAB_HWCACHE_ALIGN|SLAB_RECLAIM_ACCOUNT|
				SLAB_MEM_SPREAD|SLAB_PANIC),
571
			init_once);
L
Linus Torvalds 已提交
572 573 574 575 576 577
	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");
578
	blockdev_superblock = bd_mnt->mnt_sb;   /* For writeback */
L
Linus Torvalds 已提交
579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608
}

/*
 * 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;

609
	inode = iget5_locked(blockdev_superblock, hash(dev),
L
Linus Torvalds 已提交
610 611 612 613 614 615 616 617 618
			bdev_test, bdev_set, &dev);

	if (!inode)
		return NULL;

	bdev = &BDEV_I(inode)->bdev;

	if (inode->i_state & I_NEW) {
		bdev->bd_contains = NULL;
619
		bdev->bd_super = NULL;
L
Linus Torvalds 已提交
620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638
		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);
		spin_lock(&bdev_lock);
		list_add(&bdev->bd_list, &all_bdevs);
		spin_unlock(&bdev_lock);
		unlock_new_inode(inode);
	}
	return bdev;
}

EXPORT_SYMBOL(bdget);

639 640 641 642 643 644
/**
 * 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
645
	ihold(bdev->bd_inode);
646 647
	return bdev;
}
648
EXPORT_SYMBOL(bdgrab);
649

L
Linus Torvalds 已提交
650 651
long nr_blockdev_pages(void)
{
652
	struct block_device *bdev;
L
Linus Torvalds 已提交
653 654
	long ret = 0;
	spin_lock(&bdev_lock);
655
	list_for_each_entry(bdev, &all_bdevs, bd_list) {
L
Linus Torvalds 已提交
656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671
		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;
672

L
Linus Torvalds 已提交
673 674
	spin_lock(&bdev_lock);
	bdev = inode->i_bdev;
675
	if (bdev) {
A
Al Viro 已提交
676
		ihold(bdev->bd_inode);
L
Linus Torvalds 已提交
677 678 679 680
		spin_unlock(&bdev_lock);
		return bdev;
	}
	spin_unlock(&bdev_lock);
681

L
Linus Torvalds 已提交
682 683 684
	bdev = bdget(inode->i_rdev);
	if (bdev) {
		spin_lock(&bdev_lock);
685 686
		if (!inode->i_bdev) {
			/*
A
Al Viro 已提交
687
			 * We take an additional reference to bd_inode,
688 689 690 691
			 * 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
692
			ihold(bdev->bd_inode);
693 694 695 696
			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
697 698 699 700 701 702 703 704 705
		spin_unlock(&bdev_lock);
	}
	return bdev;
}

/* Call when you free inode */

void bd_forget(struct inode *inode)
{
706 707
	struct block_device *bdev = NULL;

L
Linus Torvalds 已提交
708
	spin_lock(&bdev_lock);
709 710 711
	if (!sb_is_blkdev_sb(inode->i_sb))
		bdev = inode->i_bdev;
	__bd_forget(inode);
L
Linus Torvalds 已提交
712
	spin_unlock(&bdev_lock);
713 714 715

	if (bdev)
		iput(bdev->bd_inode);
L
Linus Torvalds 已提交
716 717
}

T
Tejun Heo 已提交
718 719 720 721 722 723
/**
 * 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
724
 * Test whether @bdev can be claimed by @holder.
T
Tejun Heo 已提交
725 726 727 728 729 730 731 732 733
 *
 * 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
734 735
{
	if (bdev->bd_holder == holder)
T
Tejun Heo 已提交
736
		return true;	 /* already a holder */
L
Linus Torvalds 已提交
737
	else if (bdev->bd_holder != NULL)
T
Tejun Heo 已提交
738
		return false; 	 /* held by someone else */
L
Linus Torvalds 已提交
739
	else if (bdev->bd_contains == bdev)
T
Tejun Heo 已提交
740
		return true;  	 /* is a whole device which isn't held */
L
Linus Torvalds 已提交
741

742
	else if (whole->bd_holder == bd_may_claim)
T
Tejun Heo 已提交
743 744 745
		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
746
	else
T
Tejun Heo 已提交
747 748 749
		return true;	 /* is a partition of an un-held device */
}

750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775
/**
 * 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;

776 777
	/* if claiming is already in progress, wait for it to finish */
	if (whole->bd_claiming) {
778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800
		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
801 802 803 804 805 806 807
 * 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.
808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832
 *
 * 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);

833 834 835 836 837 838 839 840 841 842 843 844 845
	/*
	 * 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);

846
	module_put(disk->fops->owner);
847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865
	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);
	}
}

866
#ifdef CONFIG_SYSFS
867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883
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;
}

884
static int add_symlink(struct kobject *from, struct kobject *to)
885
{
886
	return sysfs_create_link(from, to, kobject_name(to));
887 888 889 890 891 892 893
}

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

J
Jun'ichi Nomura 已提交
894
/**
895 896 897
 * 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
898
 *
899 900
 * DON'T USE THIS UNLESS YOU'RE ALREADY USING IT.
 *
901
 * This functions creates the following sysfs symlinks.
902
 *
903 904
 * - from "slaves" directory of the holder @disk to the claimed @bdev
 * - from "holders" directory of the @bdev to the holder @disk
905
 *
906 907
 * For example, if /dev/dm-0 maps to /dev/sda and disk for dm-0 is
 * passed to bd_link_disk_holder(), then:
908
 *
909 910
 *   /sys/block/dm-0/slaves/sda --> /sys/block/sda
 *   /sys/block/sda/holders/dm-0 --> /sys/block/dm-0
911
 *
912 913 914
 * 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.
915
 *
916 917
 * CONTEXT:
 * Might sleep.
918
 *
919 920
 * RETURNS:
 * 0 on success, -errno on failure.
921
 */
922
int bd_link_disk_holder(struct block_device *bdev, struct gendisk *disk)
923
{
924
	struct bd_holder_disk *holder;
925
	int ret = 0;
926

927
	mutex_lock(&bdev->bd_mutex);
J
Jun'ichi Nomura 已提交
928

929
	WARN_ON_ONCE(!bdev->bd_holder);
930

931 932 933
	/* FIXME: remove the following once add_disk() handles errors */
	if (WARN_ON(!disk->slave_dir || !bdev->bd_part->holder_dir))
		goto out_unlock;
934

935 936 937
	holder = bd_find_holder_disk(bdev, disk);
	if (holder) {
		holder->refcnt++;
938
		goto out_unlock;
939
	}
940

941 942 943
	holder = kzalloc(sizeof(*holder), GFP_KERNEL);
	if (!holder) {
		ret = -ENOMEM;
944 945
		goto out_unlock;
	}
946

947 948 949 950 951 952 953 954 955 956 957
	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;
958 959 960 961 962
	/*
	 * bdev could be deleted beneath us which would implicitly destroy
	 * the holder directory.  Hold on to it.
	 */
	kobject_get(bdev->bd_part->holder_dir);
963 964 965 966 967 968 969 970

	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);
971
out_unlock:
972
	mutex_unlock(&bdev->bd_mutex);
973
	return ret;
974
}
975
EXPORT_SYMBOL_GPL(bd_link_disk_holder);
976

977 978 979 980 981 982 983 984 985 986 987
/**
 * 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)
988
{
989
	struct bd_holder_disk *holder;
990

991
	mutex_lock(&bdev->bd_mutex);
992

993 994 995 996 997 998
	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);
999
		kobject_put(bdev->bd_part->holder_dir);
1000 1001 1002 1003 1004
		list_del_init(&holder->list);
		kfree(holder);
	}

	mutex_unlock(&bdev->bd_mutex);
L
Linus Torvalds 已提交
1005
}
1006
EXPORT_SYMBOL_GPL(bd_unlink_disk_holder);
1007
#endif
L
Linus Torvalds 已提交
1008

1009 1010 1011 1012
/**
 * flush_disk - invalidates all buffer-cache entries on a disk
 *
 * @bdev:      struct block device to be flushed
1013
 * @kill_dirty: flag to guide handling of dirty inodes
1014 1015 1016 1017 1018
 *
 * 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.
 */
1019
static void flush_disk(struct block_device *bdev, bool kill_dirty)
1020
{
1021
	if (__invalidate_device(bdev, kill_dirty)) {
1022 1023 1024 1025 1026 1027 1028 1029 1030 1031
		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
1032
	if (disk_part_scan_enabled(bdev->bd_disk))
1033 1034 1035
		bdev->bd_invalidated = 1;
}

1036
/**
1037
 * check_disk_size_change - checks for disk size change and adjusts bdev size.
1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057
 * @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);
1058
		flush_disk(bdev, false);
1059 1060 1061 1062
	}
}
EXPORT_SYMBOL(check_disk_size_change);

1063
/**
1064
 * revalidate_disk - wrapper for lower-level driver's revalidate_disk call-back
1065 1066 1067 1068 1069 1070 1071 1072
 * @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)
{
1073
	struct block_device *bdev;
1074 1075 1076 1077 1078
	int ret = 0;

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

1079 1080 1081 1082 1083 1084
	bdev = bdget_disk(disk, 0);
	if (!bdev)
		return ret;

	mutex_lock(&bdev->bd_mutex);
	check_disk_size_change(disk, bdev);
1085
	bdev->bd_invalidated = 0;
1086 1087
	mutex_unlock(&bdev->bd_mutex);
	bdput(bdev);
1088 1089 1090 1091
	return ret;
}
EXPORT_SYMBOL(revalidate_disk);

L
Linus Torvalds 已提交
1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103
/*
 * 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;
1104
	const struct block_device_operations *bdops = disk->fops;
1105
	unsigned int events;
L
Linus Torvalds 已提交
1106

1107 1108 1109
	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
1110 1111
		return 0;

1112
	flush_disk(bdev, true);
L
Linus Torvalds 已提交
1113 1114 1115 1116 1117 1118 1119 1120 1121
	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)
{
1122
	unsigned bsize = bdev_logical_block_size(bdev);
L
Linus Torvalds 已提交
1123

1124 1125 1126
	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
1127 1128 1129 1130 1131 1132 1133 1134 1135 1136
	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
1137
static void __blkdev_put(struct block_device *bdev, fmode_t mode, int for_part);
1138

1139 1140 1141 1142 1143 1144 1145
/*
 * bd_mutex locking:
 *
 *  mutex_lock(part->bd_mutex)
 *    mutex_lock_nested(whole->bd_mutex, 1)
 */

1146
static int __blkdev_get(struct block_device *bdev, fmode_t mode, int for_part)
L
Linus Torvalds 已提交
1147 1148
{
	struct gendisk *disk;
1149
	struct module *owner;
1150
	int ret;
1151
	int partno;
1152 1153
	int perm = 0;

1154
	if (mode & FMODE_READ)
1155
		perm |= MAY_READ;
1156
	if (mode & FMODE_WRITE)
1157 1158 1159 1160
		perm |= MAY_WRITE;
	/*
	 * hooks: /n/, see "layering violations".
	 */
1161 1162 1163 1164 1165 1166
	if (!for_part) {
		ret = devcgroup_inode_permission(bdev->bd_inode, perm);
		if (ret != 0) {
			bdput(bdev);
			return ret;
		}
1167
	}
1168

1169
 restart:
T
Tejun Heo 已提交
1170

1171
	ret = -ENXIO;
1172
	disk = get_gendisk(bdev->bd_dev, &partno);
T
Tejun Heo 已提交
1173
	if (!disk)
1174
		goto out;
1175
	owner = disk->fops->owner;
L
Linus Torvalds 已提交
1176

1177
	disk_block_events(disk);
1178
	mutex_lock_nested(&bdev->bd_mutex, for_part);
L
Linus Torvalds 已提交
1179 1180
	if (!bdev->bd_openers) {
		bdev->bd_disk = disk;
1181
		bdev->bd_queue = disk->queue;
L
Linus Torvalds 已提交
1182
		bdev->bd_contains = bdev;
1183
		bdev->bd_inode->i_flags = disk->fops->direct_access ? S_DAX : 0;
1184
		if (!partno) {
1185 1186 1187 1188 1189
			ret = -ENXIO;
			bdev->bd_part = disk_get_part(disk, partno);
			if (!bdev->bd_part)
				goto out_clear;

1190
			ret = 0;
L
Linus Torvalds 已提交
1191
			if (disk->fops->open) {
1192
				ret = disk->fops->open(bdev, mode);
1193 1194 1195 1196 1197 1198 1199 1200
				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;
1201
					bdev->bd_queue = NULL;
1202
					mutex_unlock(&bdev->bd_mutex);
1203 1204
					disk_unblock_events(disk);
					put_disk(disk);
1205
					module_put(owner);
1206 1207
					goto restart;
				}
L
Linus Torvalds 已提交
1208
			}
1209

1210
			if (!ret)
1211 1212
				bd_set_size(bdev,(loff_t)get_capacity(disk)<<9);

1213 1214 1215 1216 1217 1218
			/*
			 * 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.
			 */
1219 1220 1221 1222 1223 1224
			if (bdev->bd_invalidated) {
				if (!ret)
					rescan_partitions(disk, bdev);
				else if (ret == -ENOMEDIUM)
					invalidate_partitions(disk, bdev);
			}
1225 1226
			if (ret)
				goto out_clear;
L
Linus Torvalds 已提交
1227 1228 1229 1230 1231
		} else {
			struct block_device *whole;
			whole = bdget_disk(disk, 0);
			ret = -ENOMEM;
			if (!whole)
T
Tejun Heo 已提交
1232
				goto out_clear;
1233
			BUG_ON(for_part);
1234
			ret = __blkdev_get(whole, mode, 1);
L
Linus Torvalds 已提交
1235
			if (ret)
T
Tejun Heo 已提交
1236
				goto out_clear;
L
Linus Torvalds 已提交
1237
			bdev->bd_contains = whole;
1238
			bdev->bd_part = disk_get_part(disk, partno);
1239
			if (!(disk->flags & GENHD_FL_UP) ||
1240
			    !bdev->bd_part || !bdev->bd_part->nr_sects) {
L
Linus Torvalds 已提交
1241
				ret = -ENXIO;
T
Tejun Heo 已提交
1242
				goto out_clear;
L
Linus Torvalds 已提交
1243
			}
1244
			bd_set_size(bdev, (loff_t)bdev->bd_part->nr_sects << 9);
1245 1246 1247 1248 1249 1250 1251
			/*
			 * If the partition is not aligned on a page
			 * boundary, we can't do dax I/O to it.
			 */
			if ((bdev->bd_part->start_sect % (PAGE_SIZE / 512)) ||
			    (bdev->bd_part->nr_sects % (PAGE_SIZE / 512)))
				bdev->bd_inode->i_flags &= ~S_DAX;
L
Linus Torvalds 已提交
1252 1253 1254
		}
	} else {
		if (bdev->bd_contains == bdev) {
1255 1256
			ret = 0;
			if (bdev->bd_disk->fops->open)
1257
				ret = bdev->bd_disk->fops->open(bdev, mode);
1258
			/* the same as first opener case, read comment there */
1259 1260 1261 1262 1263 1264
			if (bdev->bd_invalidated) {
				if (!ret)
					rescan_partitions(bdev->bd_disk, bdev);
				else if (ret == -ENOMEDIUM)
					invalidate_partitions(bdev->bd_disk, bdev);
			}
1265 1266
			if (ret)
				goto out_unlock_bdev;
L
Linus Torvalds 已提交
1267
		}
1268 1269
		/* only one opener holds refs to the module and disk */
		put_disk(disk);
1270
		module_put(owner);
L
Linus Torvalds 已提交
1271 1272
	}
	bdev->bd_openers++;
1273 1274
	if (for_part)
		bdev->bd_part_count++;
1275
	mutex_unlock(&bdev->bd_mutex);
1276
	disk_unblock_events(disk);
L
Linus Torvalds 已提交
1277 1278
	return 0;

T
Tejun Heo 已提交
1279
 out_clear:
1280
	disk_put_part(bdev->bd_part);
L
Linus Torvalds 已提交
1281
	bdev->bd_disk = NULL;
T
Tejun Heo 已提交
1282
	bdev->bd_part = NULL;
1283
	bdev->bd_queue = NULL;
L
Linus Torvalds 已提交
1284
	if (bdev != bdev->bd_contains)
1285
		__blkdev_put(bdev->bd_contains, mode, 1);
L
Linus Torvalds 已提交
1286
	bdev->bd_contains = NULL;
T
Tejun Heo 已提交
1287
 out_unlock_bdev:
1288
	mutex_unlock(&bdev->bd_mutex);
1289
	disk_unblock_events(disk);
T
Tejun Heo 已提交
1290
	put_disk(disk);
1291
	module_put(owner);
1292
 out:
T
Tejun Heo 已提交
1293 1294
	bdput(bdev);

L
Linus Torvalds 已提交
1295 1296 1297
	return ret;
}

1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316
/**
 * 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.
 */
1317
int blkdev_get(struct block_device *bdev, fmode_t mode, void *holder)
L
Linus Torvalds 已提交
1318
{
1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334
	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) {
1335 1336
		struct gendisk *disk = whole->bd_disk;

1337
		/* finish claiming */
1338
		mutex_lock(&bdev->bd_mutex);
1339 1340
		spin_lock(&bdev_lock);

1341
		if (!res) {
1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360
			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);
1361 1362

		/*
1363 1364 1365 1366 1367
		 * 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().
1368
		 */
1369 1370
		if (!res && (mode & FMODE_WRITE) && !bdev->bd_write_holder &&
		    (disk->flags & GENHD_FL_BLOCK_EVENTS_ON_EXCL_WRITE)) {
1371
			bdev->bd_write_holder = true;
1372
			disk_block_events(disk);
1373 1374 1375
		}

		mutex_unlock(&bdev->bd_mutex);
1376
		bdput(whole);
1377 1378 1379
	}

	return res;
1380
}
L
Linus Torvalds 已提交
1381 1382
EXPORT_SYMBOL(blkdev_get);

1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413
/**
 * 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);

1414 1415 1416 1417 1418
	if ((mode & FMODE_WRITE) && bdev_read_only(bdev)) {
		blkdev_put(bdev, mode);
		return ERR_PTR(-EACCES);
	}

1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461
	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
1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473
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;

1474 1475 1476 1477 1478 1479 1480
	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
1481
	bdev = bd_acquire(inode);
1482 1483
	if (bdev == NULL)
		return -ENOMEM;
L
Linus Torvalds 已提交
1484

1485 1486
	filp->f_mapping = bdev->bd_inode->i_mapping;

1487
	return blkdev_get(bdev, filp->f_mode, filp);
L
Linus Torvalds 已提交
1488 1489
}

A
Al Viro 已提交
1490
static void __blkdev_put(struct block_device *bdev, fmode_t mode, int for_part)
1491 1492
{
	struct gendisk *disk = bdev->bd_disk;
1493
	struct block_device *victim = NULL;
1494

1495
	mutex_lock_nested(&bdev->bd_mutex, for_part);
1496 1497 1498
	if (for_part)
		bdev->bd_part_count--;

1499
	if (!--bdev->bd_openers) {
1500
		WARN_ON_ONCE(bdev->bd_holders);
1501 1502
		sync_blockdev(bdev);
		kill_bdev(bdev);
1503 1504 1505
		/*
		 * ->release can cause the queue to disappear, so flush all
		 * dirty data before.
1506
		 */
1507
		bdev_write_inode(bdev->bd_inode);
1508 1509 1510
	}
	if (bdev->bd_contains == bdev) {
		if (disk->fops->release)
1511
			disk->fops->release(disk, mode);
1512 1513 1514 1515
	}
	if (!bdev->bd_openers) {
		struct module *owner = disk->fops->owner;

T
Tejun Heo 已提交
1516 1517
		disk_put_part(bdev->bd_part);
		bdev->bd_part = NULL;
1518
		bdev->bd_disk = NULL;
1519 1520
		if (bdev != bdev->bd_contains)
			victim = bdev->bd_contains;
1521
		bdev->bd_contains = NULL;
1522 1523 1524

		put_disk(disk);
		module_put(owner);
1525 1526 1527
	}
	mutex_unlock(&bdev->bd_mutex);
	bdput(bdev);
1528
	if (victim)
A
Al Viro 已提交
1529
		__blkdev_put(victim, mode, 1);
1530 1531
}

A
Al Viro 已提交
1532
void blkdev_put(struct block_device *bdev, fmode_t mode)
1533
{
1534 1535
	mutex_lock(&bdev->bd_mutex);

1536
	if (mode & FMODE_EXCL) {
1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556
		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);

1557 1558 1559 1560
		/*
		 * If this was the last claim, remove holder link and
		 * unblock evpoll if it was a write holder.
		 */
1561 1562 1563
		if (bdev_free && bdev->bd_write_holder) {
			disk_unblock_events(bdev->bd_disk);
			bdev->bd_write_holder = false;
1564
		}
1565
	}
1566

1567 1568 1569 1570 1571 1572 1573 1574 1575
	/*
	 * 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
1576
	__blkdev_put(bdev, mode, 0);
1577
}
1578 1579
EXPORT_SYMBOL(blkdev_put);

L
Linus Torvalds 已提交
1580 1581 1582
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
1583 1584
	blkdev_put(bdev, filp->f_mode);
	return 0;
L
Linus Torvalds 已提交
1585 1586
}

1587
static long block_ioctl(struct file *file, unsigned cmd, unsigned long arg)
L
Linus Torvalds 已提交
1588
{
1589 1590
	struct block_device *bdev = I_BDEV(file->f_mapping->host);
	fmode_t mode = file->f_mode;
1591 1592 1593 1594 1595

	/*
	 * O_NDELAY can be altered using fcntl(.., F_SETFL, ..), so we have
	 * to updated it before every ioctl.
	 */
1596
	if (file->f_flags & O_NDELAY)
1597 1598 1599 1600
		mode |= FMODE_NDELAY;
	else
		mode &= ~FMODE_NDELAY;

1601
	return blkdev_ioctl(bdev, mode, cmd, arg);
L
Linus Torvalds 已提交
1602 1603
}

1604 1605 1606 1607 1608 1609 1610
/*
 * 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.
 */
1611
ssize_t blkdev_write_iter(struct kiocb *iocb, struct iov_iter *from)
1612 1613
{
	struct file *file = iocb->ki_filp;
1614 1615
	struct inode *bd_inode = file->f_mapping->host;
	loff_t size = i_size_read(bd_inode);
1616
	struct blk_plug plug;
1617
	ssize_t ret;
1618

1619 1620
	if (bdev_read_only(I_BDEV(bd_inode)))
		return -EPERM;
1621

1622
	if (!iov_iter_count(from))
1623 1624
		return 0;

1625 1626 1627 1628
	if (iocb->ki_pos >= size)
		return -ENOSPC;

	iov_iter_truncate(from, size - iocb->ki_pos);
1629

1630
	blk_start_plug(&plug);
1631
	ret = __generic_file_write_iter(iocb, from);
1632
	if (ret > 0) {
1633
		ssize_t err;
1634
		err = generic_write_sync(file, iocb->ki_pos - ret, ret);
1635
		if (err < 0)
1636 1637
			ret = err;
	}
1638
	blk_finish_plug(&plug);
1639 1640
	return ret;
}
1641
EXPORT_SYMBOL_GPL(blkdev_write_iter);
1642

1643
ssize_t blkdev_read_iter(struct kiocb *iocb, struct iov_iter *to)
1644 1645 1646 1647
{
	struct file *file = iocb->ki_filp;
	struct inode *bd_inode = file->f_mapping->host;
	loff_t size = i_size_read(bd_inode);
1648
	loff_t pos = iocb->ki_pos;
1649 1650 1651 1652 1653

	if (pos >= size)
		return 0;

	size -= pos;
1654 1655
	iov_iter_truncate(to, size);
	return generic_file_read_iter(iocb, to);
1656
}
1657
EXPORT_SYMBOL_GPL(blkdev_read_iter);
1658

1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672
/*
 * 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
1673
static const struct address_space_operations def_blk_aops = {
L
Linus Torvalds 已提交
1674
	.readpage	= blkdev_readpage,
1675
	.readpages	= blkdev_readpages,
L
Linus Torvalds 已提交
1676
	.writepage	= blkdev_writepage,
N
Nick Piggin 已提交
1677 1678
	.write_begin	= blkdev_write_begin,
	.write_end	= blkdev_write_end,
L
Linus Torvalds 已提交
1679
	.writepages	= generic_writepages,
1680
	.releasepage	= blkdev_releasepage,
L
Linus Torvalds 已提交
1681
	.direct_IO	= blkdev_direct_IO,
1682
	.is_dirty_writeback = buffer_check_dirty_writeback,
L
Linus Torvalds 已提交
1683 1684
};

1685
const struct file_operations def_blk_fops = {
L
Linus Torvalds 已提交
1686 1687 1688
	.open		= blkdev_open,
	.release	= blkdev_close,
	.llseek		= block_llseek,
1689
	.read_iter	= blkdev_read_iter,
1690
	.write_iter	= blkdev_write_iter,
1691
	.mmap		= generic_file_mmap,
1692
	.fsync		= blkdev_fsync,
1693
	.unlocked_ioctl	= block_ioctl,
L
Linus Torvalds 已提交
1694 1695 1696
#ifdef CONFIG_COMPAT
	.compat_ioctl	= compat_blkdev_ioctl,
#endif
1697
	.splice_read	= generic_file_splice_read,
A
Al Viro 已提交
1698
	.splice_write	= iter_file_splice_write,
L
Linus Torvalds 已提交
1699 1700 1701 1702 1703 1704 1705
};

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);
1706
	res = blkdev_ioctl(bdev, 0, cmd, arg);
L
Linus Torvalds 已提交
1707 1708 1709 1710 1711 1712 1713 1714
	set_fs(old_fs);
	return res;
}

EXPORT_SYMBOL(ioctl_by_bdev);

/**
 * lookup_bdev  - lookup a struct block_device by name
1715
 * @pathname:	special file representing the block device
L
Linus Torvalds 已提交
1716
 *
1717
 * Get a reference to the blockdevice at @pathname in the current
L
Linus Torvalds 已提交
1718 1719 1720
 * namespace if possible and return it.  Return ERR_PTR(error)
 * otherwise.
 */
1721
struct block_device *lookup_bdev(const char *pathname)
L
Linus Torvalds 已提交
1722 1723 1724
{
	struct block_device *bdev;
	struct inode *inode;
1725
	struct path path;
L
Linus Torvalds 已提交
1726 1727
	int error;

1728
	if (!pathname || !*pathname)
L
Linus Torvalds 已提交
1729 1730
		return ERR_PTR(-EINVAL);

1731
	error = kern_path(pathname, LOOKUP_FOLLOW, &path);
L
Linus Torvalds 已提交
1732 1733 1734
	if (error)
		return ERR_PTR(error);

1735
	inode = d_backing_inode(path.dentry);
L
Linus Torvalds 已提交
1736 1737 1738 1739
	error = -ENOTBLK;
	if (!S_ISBLK(inode->i_mode))
		goto fail;
	error = -EACCES;
1740
	if (path.mnt->mnt_flags & MNT_NODEV)
L
Linus Torvalds 已提交
1741 1742 1743 1744 1745 1746
		goto fail;
	error = -ENOMEM;
	bdev = bd_acquire(inode);
	if (!bdev)
		goto fail;
out:
1747
	path_put(&path);
L
Linus Torvalds 已提交
1748 1749 1750 1751 1752
	return bdev;
fail:
	bdev = ERR_PTR(error);
	goto out;
}
1753
EXPORT_SYMBOL(lookup_bdev);
L
Linus Torvalds 已提交
1754

1755
int __invalidate_device(struct block_device *bdev, bool kill_dirty)
1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767
{
	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);
1768
		res = invalidate_inodes(sb, kill_dirty);
1769 1770
		drop_super(sb);
	}
1771
	invalidate_bdev(bdev);
1772 1773 1774
	return res;
}
EXPORT_SYMBOL(__invalidate_device);
1775 1776 1777 1778 1779

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

1780
	spin_lock(&blockdev_superblock->s_inode_list_lock);
1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791
	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);
1792
		spin_unlock(&blockdev_superblock->s_inode_list_lock);
1793 1794 1795
		/*
		 * We hold a reference to 'inode' so it couldn't have been
		 * removed from s_inodes list while we dropped the
1796
		 * s_inode_list_lock  We cannot iput the inode now as we can
1797
		 * be holding the last reference and we cannot iput it under
1798
		 * s_inode_list_lock. So we keep the reference and iput it
1799 1800 1801 1802 1803 1804 1805
		 * later.
		 */
		iput(old_inode);
		old_inode = inode;

		func(I_BDEV(inode), arg);

1806
		spin_lock(&blockdev_superblock->s_inode_list_lock);
1807
	}
1808
	spin_unlock(&blockdev_superblock->s_inode_list_lock);
1809 1810
	iput(old_inode);
}