disk-io.c 126.3 KB
Newer Older
C
Chris Mason 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
/*
 * Copyright (C) 2007 Oracle.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License v2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */

C
Chris Mason 已提交
19
#include <linux/fs.h>
20
#include <linux/blkdev.h>
C
Chris Mason 已提交
21
#include <linux/scatterlist.h>
C
Chris Mason 已提交
22
#include <linux/swap.h>
23
#include <linux/radix-tree.h>
C
Chris Mason 已提交
24
#include <linux/writeback.h>
C
Chris Mason 已提交
25
#include <linux/buffer_head.h>
26
#include <linux/workqueue.h>
27
#include <linux/kthread.h>
28
#include <linux/slab.h>
29
#include <linux/migrate.h>
30
#include <linux/ratelimit.h>
31
#include <linux/uuid.h>
S
Stefan Behrens 已提交
32
#include <linux/semaphore.h>
33
#include <asm/unaligned.h>
34 35
#include "ctree.h"
#include "disk-io.h"
36
#include "hash.h"
37
#include "transaction.h"
38
#include "btrfs_inode.h"
39
#include "volumes.h"
40
#include "print-tree.h"
41
#include "locking.h"
42
#include "tree-log.h"
43
#include "free-space-cache.h"
44
#include "free-space-tree.h"
45
#include "inode-map.h"
46
#include "check-integrity.h"
47
#include "rcu-string.h"
48
#include "dev-replace.h"
D
David Woodhouse 已提交
49
#include "raid56.h"
50
#include "sysfs.h"
J
Josef Bacik 已提交
51
#include "qgroup.h"
52
#include "compression.h"
53

54 55 56 57
#ifdef CONFIG_X86
#include <asm/cpufeature.h>
#endif

58 59 60 61 62 63
#define BTRFS_SUPER_FLAG_SUPP	(BTRFS_HEADER_FLAG_WRITTEN |\
				 BTRFS_HEADER_FLAG_RELOC |\
				 BTRFS_SUPER_FLAG_ERROR |\
				 BTRFS_SUPER_FLAG_SEEDING |\
				 BTRFS_SUPER_FLAG_METADUMP)

64
static const struct extent_io_ops btree_extent_io_ops;
65
static void end_workqueue_fn(struct btrfs_work *work);
66
static void free_fs_root(struct btrfs_root *root);
67
static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info,
L
liubo 已提交
68
				    int read_only);
69
static void btrfs_destroy_ordered_extents(struct btrfs_root *root);
L
liubo 已提交
70 71
static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
				      struct btrfs_root *root);
72
static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root);
L
liubo 已提交
73 74 75 76 77
static int btrfs_destroy_marked_extents(struct btrfs_root *root,
					struct extent_io_tree *dirty_pages,
					int mark);
static int btrfs_destroy_pinned_extent(struct btrfs_root *root,
				       struct extent_io_tree *pinned_extents);
78 79
static int btrfs_cleanup_transaction(struct btrfs_root *root);
static void btrfs_error_commit_super(struct btrfs_root *root);
80

C
Chris Mason 已提交
81
/*
82 83
 * btrfs_end_io_wq structs are used to do processing in task context when an IO
 * is complete.  This is used during reads to verify checksums, and it is used
C
Chris Mason 已提交
84 85
 * by writes to insert metadata for new file extents after IO is complete.
 */
86
struct btrfs_end_io_wq {
87 88 89 90 91
	struct bio *bio;
	bio_end_io_t *end_io;
	void *private;
	struct btrfs_fs_info *info;
	int error;
92
	enum btrfs_wq_endio_type metadata;
93
	struct list_head list;
94
	struct btrfs_work work;
95
};
96

97 98 99 100 101 102 103
static struct kmem_cache *btrfs_end_io_wq_cache;

int __init btrfs_end_io_wq_init(void)
{
	btrfs_end_io_wq_cache = kmem_cache_create("btrfs_end_io_wq",
					sizeof(struct btrfs_end_io_wq),
					0,
104
					SLAB_MEM_SPREAD,
105 106 107 108 109 110 111 112
					NULL);
	if (!btrfs_end_io_wq_cache)
		return -ENOMEM;
	return 0;
}

void btrfs_end_io_wq_exit(void)
{
113
	kmem_cache_destroy(btrfs_end_io_wq_cache);
114 115
}

C
Chris Mason 已提交
116 117 118 119 120
/*
 * async submit bios are used to offload expensive checksumming
 * onto the worker threads.  They checksum file and metadata bios
 * just before they are sent down the IO stack.
 */
121 122 123 124
struct async_submit_bio {
	struct inode *inode;
	struct bio *bio;
	struct list_head list;
C
Chris Mason 已提交
125 126
	extent_submit_bio_hook_t *submit_bio_start;
	extent_submit_bio_hook_t *submit_bio_done;
127
	int mirror_num;
C
Chris Mason 已提交
128
	unsigned long bio_flags;
129 130 131 132 133
	/*
	 * bio_offset is optional, can be used if the pages in the bio
	 * can't tell us where in the file the bio should go
	 */
	u64 bio_offset;
134
	struct btrfs_work work;
135
	int error;
136 137
};

138 139 140 141 142 143 144 145 146 147
/*
 * Lockdep class keys for extent_buffer->lock's in this root.  For a given
 * eb, the lockdep key is determined by the btrfs_root it belongs to and
 * the level the eb occupies in the tree.
 *
 * Different roots are used for different purposes and may nest inside each
 * other and they require separate keysets.  As lockdep keys should be
 * static, assign keysets according to the purpose of the root as indicated
 * by btrfs_root->objectid.  This ensures that all special purpose roots
 * have separate keysets.
148
 *
149 150 151
 * Lock-nesting across peer nodes is always done with the immediate parent
 * node locked thus preventing deadlock.  As lockdep doesn't know this, use
 * subclass to avoid triggering lockdep warning in such cases.
152
 *
153 154 155
 * The key is set by the readpage_end_io_hook after the buffer has passed
 * csum validation but before the pages are unlocked.  It is also set by
 * btrfs_init_new_buffer on freshly allocated blocks.
156
 *
157 158 159
 * We also add a check to make sure the highest level of the tree is the
 * same as our lockdep setup here.  If BTRFS_MAX_LEVEL changes, this code
 * needs update as well.
160 161 162 163 164
 */
#ifdef CONFIG_DEBUG_LOCK_ALLOC
# if BTRFS_MAX_LEVEL != 8
#  error
# endif
165 166 167 168 169 170 171 172 173 174 175 176 177

static struct btrfs_lockdep_keyset {
	u64			id;		/* root objectid */
	const char		*name_stem;	/* lock name stem */
	char			names[BTRFS_MAX_LEVEL + 1][20];
	struct lock_class_key	keys[BTRFS_MAX_LEVEL + 1];
} btrfs_lockdep_keysets[] = {
	{ .id = BTRFS_ROOT_TREE_OBJECTID,	.name_stem = "root"	},
	{ .id = BTRFS_EXTENT_TREE_OBJECTID,	.name_stem = "extent"	},
	{ .id = BTRFS_CHUNK_TREE_OBJECTID,	.name_stem = "chunk"	},
	{ .id = BTRFS_DEV_TREE_OBJECTID,	.name_stem = "dev"	},
	{ .id = BTRFS_FS_TREE_OBJECTID,		.name_stem = "fs"	},
	{ .id = BTRFS_CSUM_TREE_OBJECTID,	.name_stem = "csum"	},
178
	{ .id = BTRFS_QUOTA_TREE_OBJECTID,	.name_stem = "quota"	},
179 180 181
	{ .id = BTRFS_TREE_LOG_OBJECTID,	.name_stem = "log"	},
	{ .id = BTRFS_TREE_RELOC_OBJECTID,	.name_stem = "treloc"	},
	{ .id = BTRFS_DATA_RELOC_TREE_OBJECTID,	.name_stem = "dreloc"	},
182
	{ .id = BTRFS_UUID_TREE_OBJECTID,	.name_stem = "uuid"	},
183
	{ .id = BTRFS_FREE_SPACE_TREE_OBJECTID,	.name_stem = "free-space" },
184
	{ .id = 0,				.name_stem = "tree"	},
185
};
186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216

void __init btrfs_init_lockdep(void)
{
	int i, j;

	/* initialize lockdep class names */
	for (i = 0; i < ARRAY_SIZE(btrfs_lockdep_keysets); i++) {
		struct btrfs_lockdep_keyset *ks = &btrfs_lockdep_keysets[i];

		for (j = 0; j < ARRAY_SIZE(ks->names); j++)
			snprintf(ks->names[j], sizeof(ks->names[j]),
				 "btrfs-%s-%02d", ks->name_stem, j);
	}
}

void btrfs_set_buffer_lockdep_class(u64 objectid, struct extent_buffer *eb,
				    int level)
{
	struct btrfs_lockdep_keyset *ks;

	BUG_ON(level >= ARRAY_SIZE(ks->keys));

	/* find the matching keyset, id 0 is the default entry */
	for (ks = btrfs_lockdep_keysets; ks->id; ks++)
		if (ks->id == objectid)
			break;

	lockdep_set_class_and_name(&eb->lock,
				   &ks->keys[level], ks->names[level]);
}

217 218
#endif

C
Chris Mason 已提交
219 220 221 222
/*
 * extents on the btree inode are pretty simple, there's one extent
 * that covers the entire device
 */
223
static struct extent_map *btree_get_extent(struct inode *inode,
224
		struct page *page, size_t pg_offset, u64 start, u64 len,
225
		int create)
226
{
227 228 229 230
	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
	struct extent_map *em;
	int ret;

231
	read_lock(&em_tree->lock);
232
	em = lookup_extent_mapping(em_tree, start, len);
233 234 235
	if (em) {
		em->bdev =
			BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
236
		read_unlock(&em_tree->lock);
237
		goto out;
238
	}
239
	read_unlock(&em_tree->lock);
240

241
	em = alloc_extent_map();
242 243 244 245 246
	if (!em) {
		em = ERR_PTR(-ENOMEM);
		goto out;
	}
	em->start = 0;
247
	em->len = (u64)-1;
C
Chris Mason 已提交
248
	em->block_len = (u64)-1;
249
	em->block_start = 0;
250
	em->bdev = BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
251

252
	write_lock(&em_tree->lock);
J
Josef Bacik 已提交
253
	ret = add_extent_mapping(em_tree, em, 0);
254 255
	if (ret == -EEXIST) {
		free_extent_map(em);
256
		em = lookup_extent_mapping(em_tree, start, len);
257
		if (!em)
258
			em = ERR_PTR(-EIO);
259
	} else if (ret) {
260
		free_extent_map(em);
261
		em = ERR_PTR(ret);
262
	}
263
	write_unlock(&em_tree->lock);
264

265 266
out:
	return em;
267 268
}

269
u32 btrfs_csum_data(char *data, u32 seed, size_t len)
270
{
271
	return btrfs_crc32c(seed, data, len);
272 273
}

274
void btrfs_csum_final(u32 crc, u8 *result)
275
{
276
	put_unaligned_le32(~crc, result);
277 278
}

C
Chris Mason 已提交
279 280 281 282
/*
 * compute the csum for a btree block, and either verify it or write it
 * into the csum field of the block.
 */
283 284
static int csum_tree_block(struct btrfs_fs_info *fs_info,
			   struct extent_buffer *buf,
285 286
			   int verify)
{
287
	u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
288
	char *result = NULL;
289 290 291 292 293 294 295 296
	unsigned long len;
	unsigned long cur_len;
	unsigned long offset = BTRFS_CSUM_SIZE;
	char *kaddr;
	unsigned long map_start;
	unsigned long map_len;
	int err;
	u32 crc = ~(u32)0;
297
	unsigned long inline_result;
298 299

	len = buf->len - offset;
C
Chris Mason 已提交
300
	while (len > 0) {
301
		err = map_private_extent_buffer(buf, offset, 32,
302
					&kaddr, &map_start, &map_len);
C
Chris Mason 已提交
303
		if (err)
304
			return err;
305
		cur_len = min(len, map_len - (offset - map_start));
306
		crc = btrfs_csum_data(kaddr + offset - map_start,
307 308 309 310
				      crc, cur_len);
		len -= cur_len;
		offset += cur_len;
	}
311
	if (csum_size > sizeof(inline_result)) {
312
		result = kzalloc(csum_size, GFP_NOFS);
313
		if (!result)
314
			return -ENOMEM;
315 316 317 318
	} else {
		result = (char *)&inline_result;
	}

319 320 321
	btrfs_csum_final(crc, result);

	if (verify) {
322
		if (memcmp_extent_buffer(buf, result, 0, csum_size)) {
323 324
			u32 val;
			u32 found = 0;
325
			memcpy(&found, result, csum_size);
326

327
			read_extent_buffer(buf, &val, 0, csum_size);
328
			btrfs_warn_rl(fs_info,
J
Jeff Mahoney 已提交
329
				"%s checksum verify failed on %llu wanted %X found %X level %d",
330
				fs_info->sb->s_id, buf->start,
331
				val, found, btrfs_header_level(buf));
332 333
			if (result != (char *)&inline_result)
				kfree(result);
334
			return -EUCLEAN;
335 336
		}
	} else {
337
		write_extent_buffer(buf, result, 0, csum_size);
338
	}
339 340
	if (result != (char *)&inline_result)
		kfree(result);
341 342 343
	return 0;
}

C
Chris Mason 已提交
344 345 346 347 348 349
/*
 * we can't consider a given block up to date unless the transid of the
 * block matches the transid in the parent node's pointer.  This is how we
 * detect blocks that either didn't get written at all or got written
 * in the wrong place.
 */
350
static int verify_parent_transid(struct extent_io_tree *io_tree,
351 352
				 struct extent_buffer *eb, u64 parent_transid,
				 int atomic)
353
{
354
	struct extent_state *cached_state = NULL;
355
	int ret;
356
	bool need_lock = (current->journal_info == BTRFS_SEND_TRANS_STUB);
357 358 359 360

	if (!parent_transid || btrfs_header_generation(eb) == parent_transid)
		return 0;

361 362 363
	if (atomic)
		return -EAGAIN;

364 365 366 367 368
	if (need_lock) {
		btrfs_tree_read_lock(eb);
		btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
	}

369
	lock_extent_bits(io_tree, eb->start, eb->start + eb->len - 1,
370
			 &cached_state);
371
	if (extent_buffer_uptodate(eb) &&
372 373 374 375
	    btrfs_header_generation(eb) == parent_transid) {
		ret = 0;
		goto out;
	}
376 377 378
	btrfs_err_rl(eb->fs_info,
		"parent transid verify failed on %llu wanted %llu found %llu",
			eb->start,
379
			parent_transid, btrfs_header_generation(eb));
380
	ret = 1;
381 382 383 384

	/*
	 * Things reading via commit roots that don't have normal protection,
	 * like send, can have a really old block in cache that may point at a
385
	 * block that has been freed and re-allocated.  So don't clear uptodate
386 387 388 389 390 391
	 * if we find an eb that is under IO (dirty/writeback) because we could
	 * end up reading in the stale data and then writing it back out and
	 * making everybody very sad.
	 */
	if (!extent_buffer_under_io(eb))
		clear_extent_buffer_uptodate(eb);
C
Chris Mason 已提交
392
out:
393 394
	unlock_extent_cached(io_tree, eb->start, eb->start + eb->len - 1,
			     &cached_state, GFP_NOFS);
395 396
	if (need_lock)
		btrfs_tree_read_unlock_blocking(eb);
397 398 399
	return ret;
}

D
David Sterba 已提交
400 401 402 403
/*
 * Return 0 if the superblock checksum type matches the checksum value of that
 * algorithm. Pass the raw disk superblock data.
 */
404 405
static int btrfs_check_super_csum(struct btrfs_fs_info *fs_info,
				  char *raw_disk_sb)
D
David Sterba 已提交
406 407 408 409 410 411 412 413 414 415 416 417 418 419
{
	struct btrfs_super_block *disk_sb =
		(struct btrfs_super_block *)raw_disk_sb;
	u16 csum_type = btrfs_super_csum_type(disk_sb);
	int ret = 0;

	if (csum_type == BTRFS_CSUM_TYPE_CRC32) {
		u32 crc = ~(u32)0;
		const int csum_size = sizeof(crc);
		char result[csum_size];

		/*
		 * The super_block structure does not span the whole
		 * BTRFS_SUPER_INFO_SIZE range, we expect that the unused space
420
		 * is filled with zeros and is included in the checksum.
D
David Sterba 已提交
421 422 423 424 425 426 427 428 429 430
		 */
		crc = btrfs_csum_data(raw_disk_sb + BTRFS_CSUM_SIZE,
				crc, BTRFS_SUPER_INFO_SIZE - BTRFS_CSUM_SIZE);
		btrfs_csum_final(crc, result);

		if (memcmp(raw_disk_sb, result, csum_size))
			ret = 1;
	}

	if (csum_type >= ARRAY_SIZE(btrfs_csum_sizes)) {
431
		btrfs_err(fs_info, "unsupported checksum algorithm %u",
D
David Sterba 已提交
432 433 434 435 436 437 438
				csum_type);
		ret = 1;
	}

	return ret;
}

C
Chris Mason 已提交
439 440 441 442
/*
 * helper to read a given tree block, doing retries as required when
 * the checksums don't match and we have alternate mirrors to try.
 */
443 444
static int btree_read_extent_buffer_pages(struct btrfs_root *root,
					  struct extent_buffer *eb,
445
					  u64 parent_transid)
446 447
{
	struct extent_io_tree *io_tree;
448
	int failed = 0;
449 450 451
	int ret;
	int num_copies = 0;
	int mirror_num = 0;
452
	int failed_mirror = 0;
453

454
	clear_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
455 456
	io_tree = &BTRFS_I(root->fs_info->btree_inode)->io_tree;
	while (1) {
457
		ret = read_extent_buffer_pages(io_tree, eb, WAIT_COMPLETE,
458
					       btree_get_extent, mirror_num);
459 460
		if (!ret) {
			if (!verify_parent_transid(io_tree, eb,
461
						   parent_transid, 0))
462 463 464 465
				break;
			else
				ret = -EIO;
		}
C
Chris Mason 已提交
466

467 468 469 470 471 472
		/*
		 * This buffer's crc is fine, but its contents are corrupted, so
		 * there is no reason to read the other copies, they won't be
		 * any less wrong.
		 */
		if (test_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags))
473 474
			break;

475
		num_copies = btrfs_num_copies(root->fs_info,
476
					      eb->start, eb->len);
C
Chris Mason 已提交
477
		if (num_copies == 1)
478
			break;
C
Chris Mason 已提交
479

480 481 482 483 484
		if (!failed_mirror) {
			failed = 1;
			failed_mirror = eb->read_mirror;
		}

485
		mirror_num++;
486 487 488
		if (mirror_num == failed_mirror)
			mirror_num++;

C
Chris Mason 已提交
489
		if (mirror_num > num_copies)
490
			break;
491
	}
492

493
	if (failed && !ret && failed_mirror)
494 495 496
		repair_eb_io_failure(root, eb, failed_mirror);

	return ret;
497
}
498

C
Chris Mason 已提交
499
/*
C
Chris Mason 已提交
500 501
 * checksum a dirty tree block before IO.  This has extra checks to make sure
 * we only fill in the checksum field in the first page of a multi-page block
C
Chris Mason 已提交
502
 */
C
Chris Mason 已提交
503

504
static int csum_dirty_buffer(struct btrfs_fs_info *fs_info, struct page *page)
505
{
M
Miao Xie 已提交
506
	u64 start = page_offset(page);
507 508
	u64 found_start;
	struct extent_buffer *eb;
509

J
Josef Bacik 已提交
510 511 512
	eb = (struct extent_buffer *)page->private;
	if (page != eb->pages[0])
		return 0;
513

514
	found_start = btrfs_header_bytenr(eb);
515 516 517 518 519 520 521 522 523 524 525 526
	/*
	 * Please do not consolidate these warnings into a single if.
	 * It is useful to know what went wrong.
	 */
	if (WARN_ON(found_start != start))
		return -EUCLEAN;
	if (WARN_ON(!PageUptodate(page)))
		return -EUCLEAN;

	ASSERT(memcmp_extent_buffer(eb, fs_info->fsid,
			btrfs_header_fsid(), BTRFS_FSID_SIZE) == 0);

527
	return csum_tree_block(fs_info, eb, 0);
528 529
}

530
static int check_tree_block_fsid(struct btrfs_fs_info *fs_info,
Y
Yan Zheng 已提交
531 532
				 struct extent_buffer *eb)
{
533
	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
Y
Yan Zheng 已提交
534 535 536
	u8 fsid[BTRFS_UUID_SIZE];
	int ret = 1;

537
	read_extent_buffer(eb, fsid, btrfs_header_fsid(), BTRFS_FSID_SIZE);
Y
Yan Zheng 已提交
538 539 540 541 542 543 544 545 546 547
	while (fs_devices) {
		if (!memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE)) {
			ret = 0;
			break;
		}
		fs_devices = fs_devices->seed;
	}
	return ret;
}

548
#define CORRUPT(reason, eb, root, slot)				\
549 550 551 552
	btrfs_crit(root->fs_info, "corrupt %s, %s: block=%llu,"	\
		   " root=%llu, slot=%d",			\
		   btrfs_header_level(eb) == 0 ? "leaf" : "node",\
		   reason, btrfs_header_bytenr(eb), root->objectid, slot)
553 554 555 556 557 558 559 560 561

static noinline int check_leaf(struct btrfs_root *root,
			       struct extent_buffer *leaf)
{
	struct btrfs_key key;
	struct btrfs_key leaf_key;
	u32 nritems = btrfs_header_nritems(leaf);
	int slot;

562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582
	if (nritems == 0) {
		struct btrfs_root *check_root;

		key.objectid = btrfs_header_owner(leaf);
		key.type = BTRFS_ROOT_ITEM_KEY;
		key.offset = (u64)-1;

		check_root = btrfs_get_fs_root(root->fs_info, &key, false);
		/*
		 * The only reason we also check NULL here is that during
		 * open_ctree() some roots has not yet been set up.
		 */
		if (!IS_ERR_OR_NULL(check_root)) {
			/* if leaf is the root, then it's fine */
			if (leaf->start !=
			    btrfs_root_bytenr(&check_root->root_item)) {
				CORRUPT("non-root leaf's nritems is 0",
					leaf, root, 0);
				return -EIO;
			}
		}
583
		return 0;
584
	}
585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622

	/* Check the 0 item */
	if (btrfs_item_offset_nr(leaf, 0) + btrfs_item_size_nr(leaf, 0) !=
	    BTRFS_LEAF_DATA_SIZE(root)) {
		CORRUPT("invalid item offset size pair", leaf, root, 0);
		return -EIO;
	}

	/*
	 * Check to make sure each items keys are in the correct order and their
	 * offsets make sense.  We only have to loop through nritems-1 because
	 * we check the current slot against the next slot, which verifies the
	 * next slot's offset+size makes sense and that the current's slot
	 * offset is correct.
	 */
	for (slot = 0; slot < nritems - 1; slot++) {
		btrfs_item_key_to_cpu(leaf, &leaf_key, slot);
		btrfs_item_key_to_cpu(leaf, &key, slot + 1);

		/* Make sure the keys are in the right order */
		if (btrfs_comp_cpu_keys(&leaf_key, &key) >= 0) {
			CORRUPT("bad key order", leaf, root, slot);
			return -EIO;
		}

		/*
		 * Make sure the offset and ends are right, remember that the
		 * item data starts at the end of the leaf and grows towards the
		 * front.
		 */
		if (btrfs_item_offset_nr(leaf, slot) !=
			btrfs_item_end_nr(leaf, slot + 1)) {
			CORRUPT("slot offset bad", leaf, root, slot);
			return -EIO;
		}

		/*
		 * Check to make sure that we don't point outside of the leaf,
623
		 * just in case all the items are consistent to each other, but
624 625 626 627 628 629 630 631 632 633 634 635
		 * all point outside of the leaf.
		 */
		if (btrfs_item_end_nr(leaf, slot) >
		    BTRFS_LEAF_DATA_SIZE(root)) {
			CORRUPT("slot end outside of leaf", leaf, root, slot);
			return -EIO;
		}
	}

	return 0;
}

L
Liu Bo 已提交
636 637 638
static int check_node(struct btrfs_root *root, struct extent_buffer *node)
{
	unsigned long nr = btrfs_header_nritems(node);
639 640 641 642
	struct btrfs_key key, next_key;
	int slot;
	u64 bytenr;
	int ret = 0;
L
Liu Bo 已提交
643 644 645 646 647 648 649

	if (nr == 0 || nr > BTRFS_NODEPTRS_PER_BLOCK(root)) {
		btrfs_crit(root->fs_info,
			   "corrupt node: block %llu root %llu nritems %lu",
			   node->start, root->objectid, nr);
		return -EIO;
	}
650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669

	for (slot = 0; slot < nr - 1; slot++) {
		bytenr = btrfs_node_blockptr(node, slot);
		btrfs_node_key_to_cpu(node, &key, slot);
		btrfs_node_key_to_cpu(node, &next_key, slot + 1);

		if (!bytenr) {
			CORRUPT("invalid item slot", node, root, slot);
			ret = -EIO;
			goto out;
		}

		if (btrfs_comp_cpu_keys(&key, &next_key) >= 0) {
			CORRUPT("bad key order", node, root, slot);
			ret = -EIO;
			goto out;
		}
	}
out:
	return ret;
L
Liu Bo 已提交
670 671
}

672 673 674
static int btree_readpage_end_io_hook(struct btrfs_io_bio *io_bio,
				      u64 phy_offset, struct page *page,
				      u64 start, u64 end, int mirror)
675 676 677 678 679
{
	u64 found_start;
	int found_level;
	struct extent_buffer *eb;
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
680
	struct btrfs_fs_info *fs_info = root->fs_info;
681
	int ret = 0;
682
	int reads_done;
683 684 685

	if (!page->private)
		goto out;
C
Chris Mason 已提交
686

J
Josef Bacik 已提交
687
	eb = (struct extent_buffer *)page->private;
C
Chris Mason 已提交
688

689 690 691 692 693 694
	/* the pending IO might have been the only thing that kept this buffer
	 * in memory.  Make sure we have a ref for all this other checks
	 */
	extent_buffer_get(eb);

	reads_done = atomic_dec_and_test(&eb->io_pages);
695 696
	if (!reads_done)
		goto err;
697

698
	eb->read_mirror = mirror;
699
	if (test_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags)) {
700 701 702 703
		ret = -EIO;
		goto err;
	}

704
	found_start = btrfs_header_bytenr(eb);
705
	if (found_start != eb->start) {
706 707
		btrfs_err_rl(fs_info, "bad tree block start %llu %llu",
			     found_start, eb->start);
708
		ret = -EIO;
709 710
		goto err;
	}
711 712 713
	if (check_tree_block_fsid(fs_info, eb)) {
		btrfs_err_rl(fs_info, "bad fsid on block %llu",
			     eb->start);
714 715 716
		ret = -EIO;
		goto err;
	}
717
	found_level = btrfs_header_level(eb);
718
	if (found_level >= BTRFS_MAX_LEVEL) {
719 720
		btrfs_err(fs_info, "bad tree block level %d",
			  (int)btrfs_header_level(eb));
721 722 723
		ret = -EIO;
		goto err;
	}
724

725 726
	btrfs_set_buffer_lockdep_class(btrfs_header_owner(eb),
				       eb, found_level);
727

728
	ret = csum_tree_block(fs_info, eb, 1);
729
	if (ret)
730 731 732 733 734 735 736 737 738 739 740
		goto err;

	/*
	 * If this is a leaf block and it is corrupt, set the corrupt bit so
	 * that we don't try and read the other copies of this block, just
	 * return -EIO.
	 */
	if (found_level == 0 && check_leaf(root, eb)) {
		set_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
		ret = -EIO;
	}
741

L
Liu Bo 已提交
742 743 744
	if (found_level > 0 && check_node(root, eb))
		ret = -EIO;

745 746
	if (!ret)
		set_extent_buffer_uptodate(eb);
747
err:
748 749
	if (reads_done &&
	    test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
750
		btree_readahead_hook(fs_info, eb, ret);
A
Arne Jansen 已提交
751

D
David Woodhouse 已提交
752 753 754 755 756 757 758
	if (ret) {
		/*
		 * our io error hook is going to dec the io pages
		 * again, we have to make sure it has something
		 * to decrement
		 */
		atomic_inc(&eb->io_pages);
759
		clear_extent_buffer_uptodate(eb);
D
David Woodhouse 已提交
760
	}
761
	free_extent_buffer(eb);
762
out:
763
	return ret;
764 765
}

766
static int btree_io_failed_hook(struct page *page, int failed_mirror)
A
Arne Jansen 已提交
767 768 769
{
	struct extent_buffer *eb;

J
Josef Bacik 已提交
770
	eb = (struct extent_buffer *)page->private;
771
	set_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags);
772
	eb->read_mirror = failed_mirror;
D
David Woodhouse 已提交
773
	atomic_dec(&eb->io_pages);
774
	if (test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
775
		btree_readahead_hook(eb->fs_info, eb, -EIO);
A
Arne Jansen 已提交
776 777 778
	return -EIO;	/* we fixed nothing */
}

779
static void end_workqueue_bio(struct bio *bio)
780
{
781
	struct btrfs_end_io_wq *end_io_wq = bio->bi_private;
782
	struct btrfs_fs_info *fs_info;
783 784
	struct btrfs_workqueue *wq;
	btrfs_work_func_t func;
785 786

	fs_info = end_io_wq->info;
787
	end_io_wq->error = bio->bi_error;
788

M
Mike Christie 已提交
789
	if (bio_op(bio) == REQ_OP_WRITE) {
790 791 792 793 794 795 796 797 798 799 800 801 802
		if (end_io_wq->metadata == BTRFS_WQ_ENDIO_METADATA) {
			wq = fs_info->endio_meta_write_workers;
			func = btrfs_endio_meta_write_helper;
		} else if (end_io_wq->metadata == BTRFS_WQ_ENDIO_FREE_SPACE) {
			wq = fs_info->endio_freespace_worker;
			func = btrfs_freespace_write_helper;
		} else if (end_io_wq->metadata == BTRFS_WQ_ENDIO_RAID56) {
			wq = fs_info->endio_raid56_workers;
			func = btrfs_endio_raid56_helper;
		} else {
			wq = fs_info->endio_write_workers;
			func = btrfs_endio_write_helper;
		}
803
	} else {
804 805 806 807 808
		if (unlikely(end_io_wq->metadata ==
			     BTRFS_WQ_ENDIO_DIO_REPAIR)) {
			wq = fs_info->endio_repair_workers;
			func = btrfs_endio_repair_helper;
		} else if (end_io_wq->metadata == BTRFS_WQ_ENDIO_RAID56) {
809 810 811 812 813 814 815 816 817
			wq = fs_info->endio_raid56_workers;
			func = btrfs_endio_raid56_helper;
		} else if (end_io_wq->metadata) {
			wq = fs_info->endio_meta_workers;
			func = btrfs_endio_meta_helper;
		} else {
			wq = fs_info->endio_workers;
			func = btrfs_endio_helper;
		}
818
	}
819 820 821

	btrfs_init_work(&end_io_wq->work, func, end_workqueue_fn, NULL, NULL);
	btrfs_queue_work(wq, &end_io_wq->work);
822 823
}

824
int btrfs_bio_wq_end_io(struct btrfs_fs_info *info, struct bio *bio,
825
			enum btrfs_wq_endio_type metadata)
826
{
827
	struct btrfs_end_io_wq *end_io_wq;
828

829
	end_io_wq = kmem_cache_alloc(btrfs_end_io_wq_cache, GFP_NOFS);
830 831 832 833 834
	if (!end_io_wq)
		return -ENOMEM;

	end_io_wq->private = bio->bi_private;
	end_io_wq->end_io = bio->bi_end_io;
835
	end_io_wq->info = info;
836 837
	end_io_wq->error = 0;
	end_io_wq->bio = bio;
838
	end_io_wq->metadata = metadata;
839 840 841

	bio->bi_private = end_io_wq;
	bio->bi_end_io = end_workqueue_bio;
842 843 844
	return 0;
}

845
unsigned long btrfs_async_submit_limit(struct btrfs_fs_info *info)
846
{
847
	unsigned long limit = min_t(unsigned long,
848
				    info->thread_pool_size,
849 850 851
				    info->fs_devices->open_devices);
	return 256 * limit;
}
852

C
Chris Mason 已提交
853 854 855
static void run_one_async_start(struct btrfs_work *work)
{
	struct async_submit_bio *async;
856
	int ret;
C
Chris Mason 已提交
857 858

	async = container_of(work, struct  async_submit_bio, work);
859
	ret = async->submit_bio_start(async->inode, async->bio,
860 861 862 863
				      async->mirror_num, async->bio_flags,
				      async->bio_offset);
	if (ret)
		async->error = ret;
C
Chris Mason 已提交
864 865 866
}

static void run_one_async_done(struct btrfs_work *work)
867 868 869
{
	struct btrfs_fs_info *fs_info;
	struct async_submit_bio *async;
870
	int limit;
871 872 873

	async = container_of(work, struct  async_submit_bio, work);
	fs_info = BTRFS_I(async->inode)->root->fs_info;
874

875
	limit = btrfs_async_submit_limit(fs_info);
876 877
	limit = limit * 2 / 3;

878 879 880
	/*
	 * atomic_dec_return implies a barrier for waitqueue_active
	 */
881
	if (atomic_dec_return(&fs_info->nr_async_submits) < limit &&
882
	    waitqueue_active(&fs_info->async_submit_wait))
883 884
		wake_up(&fs_info->async_submit_wait);

885
	/* If an error occurred we just want to clean up the bio and move on */
886
	if (async->error) {
887 888
		async->bio->bi_error = async->error;
		bio_endio(async->bio);
889 890 891
		return;
	}

892 893
	async->submit_bio_done(async->inode, async->bio, async->mirror_num,
			       async->bio_flags, async->bio_offset);
C
Chris Mason 已提交
894 895 896 897 898 899 900
}

static void run_one_async_free(struct btrfs_work *work)
{
	struct async_submit_bio *async;

	async = container_of(work, struct  async_submit_bio, work);
901 902 903
	kfree(async);
}

904
int btrfs_wq_submit_bio(struct btrfs_fs_info *fs_info, struct inode *inode,
905
			struct bio *bio, int mirror_num,
C
Chris Mason 已提交
906
			unsigned long bio_flags,
907
			u64 bio_offset,
C
Chris Mason 已提交
908 909
			extent_submit_bio_hook_t *submit_bio_start,
			extent_submit_bio_hook_t *submit_bio_done)
910 911 912 913 914 915 916 917 918 919
{
	struct async_submit_bio *async;

	async = kmalloc(sizeof(*async), GFP_NOFS);
	if (!async)
		return -ENOMEM;

	async->inode = inode;
	async->bio = bio;
	async->mirror_num = mirror_num;
C
Chris Mason 已提交
920 921 922
	async->submit_bio_start = submit_bio_start;
	async->submit_bio_done = submit_bio_done;

923
	btrfs_init_work(&async->work, btrfs_worker_helper, run_one_async_start,
924
			run_one_async_done, run_one_async_free);
C
Chris Mason 已提交
925

C
Chris Mason 已提交
926
	async->bio_flags = bio_flags;
927
	async->bio_offset = bio_offset;
928

929 930
	async->error = 0;

931
	atomic_inc(&fs_info->nr_async_submits);
932

J
Jens Axboe 已提交
933
	if (bio->bi_opf & REQ_SYNC)
934
		btrfs_set_work_high_priority(&async->work);
935

936
	btrfs_queue_work(fs_info->workers, &async->work);
937

C
Chris Mason 已提交
938
	while (atomic_read(&fs_info->async_submit_draining) &&
939 940 941 942 943
	      atomic_read(&fs_info->nr_async_submits)) {
		wait_event(fs_info->async_submit_wait,
			   (atomic_read(&fs_info->nr_async_submits) == 0));
	}

944 945 946
	return 0;
}

947 948
static int btree_csum_one_bio(struct bio *bio)
{
949
	struct bio_vec *bvec;
950
	struct btrfs_root *root;
951
	int i, ret = 0;
952

953
	bio_for_each_segment_all(bvec, bio, i) {
954
		root = BTRFS_I(bvec->bv_page->mapping->host)->root;
955
		ret = csum_dirty_buffer(root->fs_info, bvec->bv_page);
956 957
		if (ret)
			break;
958
	}
959

960
	return ret;
961 962
}

963 964
static int __btree_submit_bio_start(struct inode *inode, struct bio *bio,
				    int mirror_num, unsigned long bio_flags,
965
				    u64 bio_offset)
966
{
967 968
	/*
	 * when we're called for a write, we're already in the async
969
	 * submission context.  Just jump into btrfs_map_bio
970
	 */
971
	return btree_csum_one_bio(bio);
C
Chris Mason 已提交
972
}
973

974
static int __btree_submit_bio_done(struct inode *inode, struct bio *bio,
975 976
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
C
Chris Mason 已提交
977
{
978 979
	int ret;

980
	/*
C
Chris Mason 已提交
981 982
	 * when we're called for a write, we're already in the async
	 * submission context.  Just jump into btrfs_map_bio
983
	 */
984
	ret = btrfs_map_bio(BTRFS_I(inode)->root, bio, mirror_num, 1);
985 986 987 988
	if (ret) {
		bio->bi_error = ret;
		bio_endio(bio);
	}
989
	return ret;
990 991
}

992 993 994 995 996
static int check_async_write(struct inode *inode, unsigned long bio_flags)
{
	if (bio_flags & EXTENT_BIO_TREE_LOG)
		return 0;
#ifdef CONFIG_X86
997
	if (static_cpu_has(X86_FEATURE_XMM4_2))
998 999 1000 1001 1002
		return 0;
#endif
	return 1;
}

1003
static int btree_submit_bio_hook(struct inode *inode, struct bio *bio,
1004 1005
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
1006
{
1007
	int async = check_async_write(inode, bio_flags);
1008 1009
	int ret;

M
Mike Christie 已提交
1010
	if (bio_op(bio) != REQ_OP_WRITE) {
C
Chris Mason 已提交
1011 1012 1013 1014
		/*
		 * called for a read, do the setup so that checksum validation
		 * can happen in the async kernel threads
		 */
1015
		ret = btrfs_bio_wq_end_io(BTRFS_I(inode)->root->fs_info,
1016
					  bio, BTRFS_WQ_ENDIO_METADATA);
1017
		if (ret)
1018
			goto out_w_error;
1019
		ret = btrfs_map_bio(BTRFS_I(inode)->root, bio, mirror_num, 0);
1020 1021 1022
	} else if (!async) {
		ret = btree_csum_one_bio(bio);
		if (ret)
1023
			goto out_w_error;
1024
		ret = btrfs_map_bio(BTRFS_I(inode)->root, bio, mirror_num, 0);
1025 1026 1027 1028 1029 1030
	} else {
		/*
		 * kthread helpers are used to submit writes so that
		 * checksumming can happen in parallel across all CPUs
		 */
		ret = btrfs_wq_submit_bio(BTRFS_I(inode)->root->fs_info,
1031
					  inode, bio, mirror_num, 0,
1032 1033 1034
					  bio_offset,
					  __btree_submit_bio_start,
					  __btree_submit_bio_done);
1035
	}
1036

1037 1038 1039 1040
	if (ret)
		goto out_w_error;
	return 0;

1041
out_w_error:
1042 1043
	bio->bi_error = ret;
	bio_endio(bio);
1044
	return ret;
1045 1046
}

J
Jan Beulich 已提交
1047
#ifdef CONFIG_MIGRATION
1048
static int btree_migratepage(struct address_space *mapping,
1049 1050
			struct page *newpage, struct page *page,
			enum migrate_mode mode)
1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064
{
	/*
	 * we can't safely write a btree page from here,
	 * we haven't done the locking hook
	 */
	if (PageDirty(page))
		return -EAGAIN;
	/*
	 * Buffers may be managed in a filesystem specific way.
	 * We must have no buffers or drop them.
	 */
	if (page_has_private(page) &&
	    !try_to_release_page(page, GFP_KERNEL))
		return -EAGAIN;
1065
	return migrate_page(mapping, newpage, page, mode);
1066
}
J
Jan Beulich 已提交
1067
#endif
1068

1069 1070 1071 1072

static int btree_writepages(struct address_space *mapping,
			    struct writeback_control *wbc)
{
1073 1074 1075
	struct btrfs_fs_info *fs_info;
	int ret;

1076
	if (wbc->sync_mode == WB_SYNC_NONE) {
1077 1078 1079 1080

		if (wbc->for_kupdate)
			return 0;

1081
		fs_info = BTRFS_I(mapping->host)->root->fs_info;
1082
		/* this is a bit racy, but that's ok */
1083 1084 1085
		ret = percpu_counter_compare(&fs_info->dirty_metadata_bytes,
					     BTRFS_DIRTY_METADATA_THRESH);
		if (ret < 0)
1086 1087
			return 0;
	}
1088
	return btree_write_cache_pages(mapping, wbc);
1089 1090
}

1091
static int btree_readpage(struct file *file, struct page *page)
1092
{
1093 1094
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
1095
	return extent_read_full_page(tree, page, btree_get_extent, 0);
1096
}
C
Chris Mason 已提交
1097

1098
static int btree_releasepage(struct page *page, gfp_t gfp_flags)
1099
{
1100
	if (PageWriteback(page) || PageDirty(page))
C
Chris Mason 已提交
1101
		return 0;
1102

1103
	return try_release_extent_buffer(page);
1104 1105
}

1106 1107
static void btree_invalidatepage(struct page *page, unsigned int offset,
				 unsigned int length)
1108
{
1109 1110
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
1111 1112
	extent_invalidatepage(tree, page, offset);
	btree_releasepage(page, GFP_NOFS);
1113
	if (PagePrivate(page)) {
1114 1115 1116
		btrfs_warn(BTRFS_I(page->mapping->host)->root->fs_info,
			   "page private not zero on page %llu",
			   (unsigned long long)page_offset(page));
1117 1118
		ClearPagePrivate(page);
		set_page_private(page, 0);
1119
		put_page(page);
1120
	}
1121 1122
}

1123 1124
static int btree_set_page_dirty(struct page *page)
{
1125
#ifdef DEBUG
1126 1127 1128 1129 1130 1131 1132 1133
	struct extent_buffer *eb;

	BUG_ON(!PagePrivate(page));
	eb = (struct extent_buffer *)page->private;
	BUG_ON(!eb);
	BUG_ON(!test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags));
	BUG_ON(!atomic_read(&eb->refs));
	btrfs_assert_tree_locked(eb);
1134
#endif
1135 1136 1137
	return __set_page_dirty_nobuffers(page);
}

1138
static const struct address_space_operations btree_aops = {
1139
	.readpage	= btree_readpage,
1140
	.writepages	= btree_writepages,
1141 1142
	.releasepage	= btree_releasepage,
	.invalidatepage = btree_invalidatepage,
1143
#ifdef CONFIG_MIGRATION
1144
	.migratepage	= btree_migratepage,
1145
#endif
1146
	.set_page_dirty = btree_set_page_dirty,
1147 1148
};

1149
void readahead_tree_block(struct btrfs_root *root, u64 bytenr)
C
Chris Mason 已提交
1150
{
1151 1152
	struct extent_buffer *buf = NULL;
	struct inode *btree_inode = root->fs_info->btree_inode;
C
Chris Mason 已提交
1153

1154
	buf = btrfs_find_create_tree_block(root, bytenr);
1155
	if (IS_ERR(buf))
1156
		return;
1157
	read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree,
1158
				 buf, WAIT_NONE, btree_get_extent, 0);
1159
	free_extent_buffer(buf);
C
Chris Mason 已提交
1160 1161
}

1162
int reada_tree_block_flagged(struct btrfs_root *root, u64 bytenr,
1163 1164 1165 1166 1167 1168 1169
			 int mirror_num, struct extent_buffer **eb)
{
	struct extent_buffer *buf = NULL;
	struct inode *btree_inode = root->fs_info->btree_inode;
	struct extent_io_tree *io_tree = &BTRFS_I(btree_inode)->io_tree;
	int ret;

1170
	buf = btrfs_find_create_tree_block(root, bytenr);
1171
	if (IS_ERR(buf))
1172 1173 1174 1175
		return 0;

	set_bit(EXTENT_BUFFER_READAHEAD, &buf->bflags);

1176
	ret = read_extent_buffer_pages(io_tree, buf, WAIT_PAGE_LOCK,
1177 1178 1179 1180 1181 1182 1183 1184 1185
				       btree_get_extent, mirror_num);
	if (ret) {
		free_extent_buffer(buf);
		return ret;
	}

	if (test_bit(EXTENT_BUFFER_CORRUPT, &buf->bflags)) {
		free_extent_buffer(buf);
		return -EIO;
1186
	} else if (extent_buffer_uptodate(buf)) {
1187 1188 1189 1190 1191 1192 1193
		*eb = buf;
	} else {
		free_extent_buffer(buf);
	}
	return 0;
}

1194
struct extent_buffer *btrfs_find_create_tree_block(struct btrfs_root *root,
1195
						 u64 bytenr)
1196
{
1197
	if (btrfs_is_testing(root->fs_info))
1198 1199
		return alloc_test_extent_buffer(root->fs_info, bytenr,
				root->nodesize);
1200
	return alloc_extent_buffer(root->fs_info, bytenr);
1201 1202 1203
}


1204 1205
int btrfs_write_tree_block(struct extent_buffer *buf)
{
1206
	return filemap_fdatawrite_range(buf->pages[0]->mapping, buf->start,
1207
					buf->start + buf->len - 1);
1208 1209 1210 1211
}

int btrfs_wait_tree_block_writeback(struct extent_buffer *buf)
{
1212
	return filemap_fdatawait_range(buf->pages[0]->mapping,
1213
				       buf->start, buf->start + buf->len - 1);
1214 1215
}

1216
struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr,
1217
				      u64 parent_transid)
1218 1219 1220 1221
{
	struct extent_buffer *buf = NULL;
	int ret;

1222
	buf = btrfs_find_create_tree_block(root, bytenr);
1223 1224
	if (IS_ERR(buf))
		return buf;
1225

1226
	ret = btree_read_extent_buffer_pages(root, buf, parent_transid);
1227 1228
	if (ret) {
		free_extent_buffer(buf);
1229
		return ERR_PTR(ret);
1230
	}
1231
	return buf;
1232

1233 1234
}

1235 1236
void clean_tree_block(struct btrfs_trans_handle *trans,
		      struct btrfs_fs_info *fs_info,
1237
		      struct extent_buffer *buf)
1238
{
1239
	if (btrfs_header_generation(buf) ==
1240
	    fs_info->running_transaction->transid) {
1241
		btrfs_assert_tree_locked(buf);
1242

1243
		if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &buf->bflags)) {
1244 1245 1246
			__percpu_counter_add(&fs_info->dirty_metadata_bytes,
					     -buf->len,
					     fs_info->dirty_metadata_batch);
1247 1248 1249 1250
			/* ugh, clear_extent_buffer_dirty needs to lock the page */
			btrfs_set_lock_blocking(buf);
			clear_extent_buffer_dirty(buf);
		}
1251
	}
1252 1253
}

1254 1255 1256 1257 1258 1259 1260 1261 1262
static struct btrfs_subvolume_writers *btrfs_alloc_subvolume_writers(void)
{
	struct btrfs_subvolume_writers *writers;
	int ret;

	writers = kmalloc(sizeof(*writers), GFP_NOFS);
	if (!writers)
		return ERR_PTR(-ENOMEM);

1263
	ret = percpu_counter_init(&writers->counter, 0, GFP_KERNEL);
1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279
	if (ret < 0) {
		kfree(writers);
		return ERR_PTR(ret);
	}

	init_waitqueue_head(&writers->wait);
	return writers;
}

static void
btrfs_free_subvolume_writers(struct btrfs_subvolume_writers *writers)
{
	percpu_counter_destroy(&writers->counter);
	kfree(writers);
}

1280 1281
static void __setup_root(u32 nodesize, u32 sectorsize, u32 stripesize,
			 struct btrfs_root *root, struct btrfs_fs_info *fs_info,
1282
			 u64 objectid)
1283
{
1284
	bool dummy = test_bit(BTRFS_FS_STATE_DUMMY_FS_INFO, &fs_info->fs_state);
C
Chris Mason 已提交
1285
	root->node = NULL;
1286
	root->commit_root = NULL;
1287 1288
	root->sectorsize = sectorsize;
	root->nodesize = nodesize;
1289
	root->stripesize = stripesize;
1290
	root->state = 0;
1291
	root->orphan_cleanup_state = 0;
1292

1293 1294
	root->objectid = objectid;
	root->last_trans = 0;
1295
	root->highest_objectid = 0;
1296
	root->nr_delalloc_inodes = 0;
1297
	root->nr_ordered_extents = 0;
1298
	root->name = NULL;
1299
	root->inode_tree = RB_ROOT;
1300
	INIT_RADIX_TREE(&root->delayed_nodes_tree, GFP_ATOMIC);
1301
	root->block_rsv = NULL;
1302
	root->orphan_block_rsv = NULL;
1303 1304

	INIT_LIST_HEAD(&root->dirty_list);
1305
	INIT_LIST_HEAD(&root->root_list);
1306 1307
	INIT_LIST_HEAD(&root->delalloc_inodes);
	INIT_LIST_HEAD(&root->delalloc_root);
1308 1309
	INIT_LIST_HEAD(&root->ordered_extents);
	INIT_LIST_HEAD(&root->ordered_root);
1310 1311
	INIT_LIST_HEAD(&root->logged_list[0]);
	INIT_LIST_HEAD(&root->logged_list[1]);
1312
	spin_lock_init(&root->orphan_lock);
1313
	spin_lock_init(&root->inode_lock);
1314
	spin_lock_init(&root->delalloc_lock);
1315
	spin_lock_init(&root->ordered_extent_lock);
1316
	spin_lock_init(&root->accounting_lock);
1317 1318
	spin_lock_init(&root->log_extents_lock[0]);
	spin_lock_init(&root->log_extents_lock[1]);
1319
	mutex_init(&root->objectid_mutex);
1320
	mutex_init(&root->log_mutex);
1321
	mutex_init(&root->ordered_extent_mutex);
1322
	mutex_init(&root->delalloc_mutex);
Y
Yan Zheng 已提交
1323 1324 1325
	init_waitqueue_head(&root->log_writer_wait);
	init_waitqueue_head(&root->log_commit_wait[0]);
	init_waitqueue_head(&root->log_commit_wait[1]);
1326 1327
	INIT_LIST_HEAD(&root->log_ctxs[0]);
	INIT_LIST_HEAD(&root->log_ctxs[1]);
Y
Yan Zheng 已提交
1328 1329 1330
	atomic_set(&root->log_commit[0], 0);
	atomic_set(&root->log_commit[1], 0);
	atomic_set(&root->log_writers, 0);
M
Miao Xie 已提交
1331
	atomic_set(&root->log_batch, 0);
1332
	atomic_set(&root->orphan_inodes, 0);
1333
	atomic_set(&root->refs, 1);
1334
	atomic_set(&root->will_be_snapshoted, 0);
1335
	atomic_set(&root->qgroup_meta_rsv, 0);
Y
Yan Zheng 已提交
1336
	root->log_transid = 0;
1337
	root->log_transid_committed = -1;
1338
	root->last_log_commit = 0;
1339
	if (!dummy)
1340 1341
		extent_io_tree_init(&root->dirty_log_pages,
				     fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
1342

1343 1344
	memset(&root->root_key, 0, sizeof(root->root_key));
	memset(&root->root_item, 0, sizeof(root->root_item));
1345
	memset(&root->defrag_progress, 0, sizeof(root->defrag_progress));
1346
	if (!dummy)
1347 1348 1349
		root->defrag_trans_start = fs_info->generation;
	else
		root->defrag_trans_start = 0;
1350
	root->root_key.objectid = objectid;
1351
	root->anon_dev = 0;
1352

1353
	spin_lock_init(&root->root_item_lock);
1354 1355
}

1356 1357
static struct btrfs_root *btrfs_alloc_root(struct btrfs_fs_info *fs_info,
		gfp_t flags)
A
Al Viro 已提交
1358
{
1359
	struct btrfs_root *root = kzalloc(sizeof(*root), flags);
A
Al Viro 已提交
1360 1361 1362 1363 1364
	if (root)
		root->fs_info = fs_info;
	return root;
}

1365 1366
#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
/* Should only be used by the testing infrastructure */
1367 1368
struct btrfs_root *btrfs_alloc_dummy_root(struct btrfs_fs_info *fs_info,
					  u32 sectorsize, u32 nodesize)
1369 1370 1371
{
	struct btrfs_root *root;

1372 1373 1374 1375
	if (!fs_info)
		return ERR_PTR(-EINVAL);

	root = btrfs_alloc_root(fs_info, GFP_KERNEL);
1376 1377
	if (!root)
		return ERR_PTR(-ENOMEM);
1378
	/* We don't use the stripesize in selftest, set it as sectorsize */
1379
	__setup_root(nodesize, sectorsize, sectorsize, root, fs_info,
1380
			BTRFS_ROOT_TREE_OBJECTID);
1381
	root->alloc_bytenr = 0;
1382 1383 1384 1385 1386

	return root;
}
#endif

1387 1388 1389 1390 1391 1392 1393 1394 1395
struct btrfs_root *btrfs_create_tree(struct btrfs_trans_handle *trans,
				     struct btrfs_fs_info *fs_info,
				     u64 objectid)
{
	struct extent_buffer *leaf;
	struct btrfs_root *tree_root = fs_info->tree_root;
	struct btrfs_root *root;
	struct btrfs_key key;
	int ret = 0;
1396
	uuid_le uuid;
1397

1398
	root = btrfs_alloc_root(fs_info, GFP_KERNEL);
1399 1400 1401
	if (!root)
		return ERR_PTR(-ENOMEM);

1402 1403
	__setup_root(tree_root->nodesize, tree_root->sectorsize,
		tree_root->stripesize, root, fs_info, objectid);
1404 1405 1406 1407
	root->root_key.objectid = objectid;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = 0;

1408
	leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0);
1409 1410
	if (IS_ERR(leaf)) {
		ret = PTR_ERR(leaf);
1411
		leaf = NULL;
1412 1413 1414
		goto fail;
	}

1415
	memzero_extent_buffer(leaf, 0, sizeof(struct btrfs_header));
1416 1417 1418 1419 1420 1421
	btrfs_set_header_bytenr(leaf, leaf->start);
	btrfs_set_header_generation(leaf, trans->transid);
	btrfs_set_header_backref_rev(leaf, BTRFS_MIXED_BACKREF_REV);
	btrfs_set_header_owner(leaf, objectid);
	root->node = leaf;

1422 1423
	write_extent_buffer_fsid(leaf, fs_info->fsid);
	write_extent_buffer_chunk_tree_uuid(leaf, fs_info->chunk_tree_uuid);
1424 1425 1426
	btrfs_mark_buffer_dirty(leaf);

	root->commit_root = btrfs_root_node(root);
1427
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
1428 1429 1430 1431 1432 1433 1434 1435 1436 1437

	root->root_item.flags = 0;
	root->root_item.byte_limit = 0;
	btrfs_set_root_bytenr(&root->root_item, leaf->start);
	btrfs_set_root_generation(&root->root_item, trans->transid);
	btrfs_set_root_level(&root->root_item, 0);
	btrfs_set_root_refs(&root->root_item, 1);
	btrfs_set_root_used(&root->root_item, leaf->len);
	btrfs_set_root_last_snapshot(&root->root_item, 0);
	btrfs_set_root_dirid(&root->root_item, 0);
1438 1439
	uuid_le_gen(&uuid);
	memcpy(root->root_item.uuid, uuid.b, BTRFS_UUID_SIZE);
1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450
	root->root_item.drop_level = 0;

	key.objectid = objectid;
	key.type = BTRFS_ROOT_ITEM_KEY;
	key.offset = 0;
	ret = btrfs_insert_root(trans, tree_root, &key, &root->root_item);
	if (ret)
		goto fail;

	btrfs_tree_unlock(leaf);

1451 1452
	return root;

1453
fail:
1454 1455
	if (leaf) {
		btrfs_tree_unlock(leaf);
1456
		free_extent_buffer(root->commit_root);
1457 1458 1459
		free_extent_buffer(leaf);
	}
	kfree(root);
1460

1461
	return ERR_PTR(ret);
1462 1463
}

Y
Yan Zheng 已提交
1464 1465
static struct btrfs_root *alloc_log_tree(struct btrfs_trans_handle *trans,
					 struct btrfs_fs_info *fs_info)
1466 1467 1468
{
	struct btrfs_root *root;
	struct btrfs_root *tree_root = fs_info->tree_root;
Y
Yan Zheng 已提交
1469
	struct extent_buffer *leaf;
1470

1471
	root = btrfs_alloc_root(fs_info, GFP_NOFS);
1472
	if (!root)
Y
Yan Zheng 已提交
1473
		return ERR_PTR(-ENOMEM);
1474

1475 1476 1477
	__setup_root(tree_root->nodesize, tree_root->sectorsize,
		     tree_root->stripesize, root, fs_info,
		     BTRFS_TREE_LOG_OBJECTID);
1478 1479 1480 1481

	root->root_key.objectid = BTRFS_TREE_LOG_OBJECTID;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = BTRFS_TREE_LOG_OBJECTID;
1482

Y
Yan Zheng 已提交
1483
	/*
1484 1485
	 * DON'T set REF_COWS for log trees
	 *
Y
Yan Zheng 已提交
1486 1487 1488 1489 1490
	 * log trees do not get reference counted because they go away
	 * before a real commit is actually done.  They do store pointers
	 * to file data extents, and those reference counts still get
	 * updated (along with back refs to the log tree).
	 */
1491

1492 1493
	leaf = btrfs_alloc_tree_block(trans, root, 0, BTRFS_TREE_LOG_OBJECTID,
			NULL, 0, 0, 0);
Y
Yan Zheng 已提交
1494 1495 1496 1497
	if (IS_ERR(leaf)) {
		kfree(root);
		return ERR_CAST(leaf);
	}
1498

1499
	memzero_extent_buffer(leaf, 0, sizeof(struct btrfs_header));
1500 1501 1502 1503
	btrfs_set_header_bytenr(leaf, leaf->start);
	btrfs_set_header_generation(leaf, trans->transid);
	btrfs_set_header_backref_rev(leaf, BTRFS_MIXED_BACKREF_REV);
	btrfs_set_header_owner(leaf, BTRFS_TREE_LOG_OBJECTID);
Y
Yan Zheng 已提交
1504
	root->node = leaf;
1505

1506
	write_extent_buffer_fsid(root->node, root->fs_info->fsid);
1507 1508
	btrfs_mark_buffer_dirty(root->node);
	btrfs_tree_unlock(root->node);
Y
Yan Zheng 已提交
1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538
	return root;
}

int btrfs_init_log_root_tree(struct btrfs_trans_handle *trans,
			     struct btrfs_fs_info *fs_info)
{
	struct btrfs_root *log_root;

	log_root = alloc_log_tree(trans, fs_info);
	if (IS_ERR(log_root))
		return PTR_ERR(log_root);
	WARN_ON(fs_info->log_root_tree);
	fs_info->log_root_tree = log_root;
	return 0;
}

int btrfs_add_log_tree(struct btrfs_trans_handle *trans,
		       struct btrfs_root *root)
{
	struct btrfs_root *log_root;
	struct btrfs_inode_item *inode_item;

	log_root = alloc_log_tree(trans, root->fs_info);
	if (IS_ERR(log_root))
		return PTR_ERR(log_root);

	log_root->last_trans = trans->transid;
	log_root->root_key.offset = root->root_key.objectid;

	inode_item = &log_root->root_item.inode;
1539 1540 1541
	btrfs_set_stack_inode_generation(inode_item, 1);
	btrfs_set_stack_inode_size(inode_item, 3);
	btrfs_set_stack_inode_nlink(inode_item, 1);
1542
	btrfs_set_stack_inode_nbytes(inode_item, root->nodesize);
1543
	btrfs_set_stack_inode_mode(inode_item, S_IFDIR | 0755);
Y
Yan Zheng 已提交
1544

1545
	btrfs_set_root_node(&log_root->root_item, log_root->node);
Y
Yan Zheng 已提交
1546 1547 1548 1549

	WARN_ON(root->log_root);
	root->log_root = log_root;
	root->log_transid = 0;
1550
	root->log_transid_committed = -1;
1551
	root->last_log_commit = 0;
1552 1553 1554
	return 0;
}

1555 1556
static struct btrfs_root *btrfs_read_tree_root(struct btrfs_root *tree_root,
					       struct btrfs_key *key)
1557 1558 1559
{
	struct btrfs_root *root;
	struct btrfs_fs_info *fs_info = tree_root->fs_info;
1560
	struct btrfs_path *path;
1561
	u64 generation;
1562
	int ret;
1563

1564 1565
	path = btrfs_alloc_path();
	if (!path)
1566
		return ERR_PTR(-ENOMEM);
1567

1568
	root = btrfs_alloc_root(fs_info, GFP_NOFS);
1569 1570 1571
	if (!root) {
		ret = -ENOMEM;
		goto alloc_fail;
1572 1573
	}

1574 1575
	__setup_root(tree_root->nodesize, tree_root->sectorsize,
		tree_root->stripesize, root, fs_info, key->objectid);
1576

1577 1578
	ret = btrfs_find_root(tree_root, key, path,
			      &root->root_item, &root->root_key);
1579
	if (ret) {
1580 1581
		if (ret > 0)
			ret = -ENOENT;
1582
		goto find_fail;
1583
	}
1584

1585
	generation = btrfs_root_generation(&root->root_item);
1586
	root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
1587
				     generation);
1588 1589
	if (IS_ERR(root->node)) {
		ret = PTR_ERR(root->node);
1590 1591 1592
		goto find_fail;
	} else if (!btrfs_buffer_uptodate(root->node, generation, 0)) {
		ret = -EIO;
1593 1594
		free_extent_buffer(root->node);
		goto find_fail;
1595
	}
1596
	root->commit_root = btrfs_root_node(root);
1597
out:
1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617
	btrfs_free_path(path);
	return root;

find_fail:
	kfree(root);
alloc_fail:
	root = ERR_PTR(ret);
	goto out;
}

struct btrfs_root *btrfs_read_fs_root(struct btrfs_root *tree_root,
				      struct btrfs_key *location)
{
	struct btrfs_root *root;

	root = btrfs_read_tree_root(tree_root, location);
	if (IS_ERR(root))
		return root;

	if (root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID) {
1618
		set_bit(BTRFS_ROOT_REF_COWS, &root->state);
1619 1620
		btrfs_check_and_init_root_item(&root->root_item);
	}
1621

1622 1623 1624
	return root;
}

1625 1626 1627
int btrfs_init_fs_root(struct btrfs_root *root)
{
	int ret;
1628
	struct btrfs_subvolume_writers *writers;
1629 1630 1631 1632 1633 1634 1635 1636 1637

	root->free_ino_ctl = kzalloc(sizeof(*root->free_ino_ctl), GFP_NOFS);
	root->free_ino_pinned = kzalloc(sizeof(*root->free_ino_pinned),
					GFP_NOFS);
	if (!root->free_ino_pinned || !root->free_ino_ctl) {
		ret = -ENOMEM;
		goto fail;
	}

1638 1639 1640 1641 1642 1643 1644
	writers = btrfs_alloc_subvolume_writers();
	if (IS_ERR(writers)) {
		ret = PTR_ERR(writers);
		goto fail;
	}
	root->subv_writers = writers;

1645
	btrfs_init_free_ino_ctl(root);
1646 1647
	spin_lock_init(&root->ino_cache_lock);
	init_waitqueue_head(&root->ino_cache_wait);
1648 1649 1650

	ret = get_anon_bdev(&root->anon_dev);
	if (ret)
L
Liu Bo 已提交
1651
		goto fail;
1652 1653 1654 1655 1656 1657

	mutex_lock(&root->objectid_mutex);
	ret = btrfs_find_highest_objectid(root,
					&root->highest_objectid);
	if (ret) {
		mutex_unlock(&root->objectid_mutex);
L
Liu Bo 已提交
1658
		goto fail;
1659 1660 1661 1662 1663 1664
	}

	ASSERT(root->highest_objectid <= BTRFS_LAST_FREE_OBJECTID);

	mutex_unlock(&root->objectid_mutex);

1665 1666
	return 0;
fail:
L
Liu Bo 已提交
1667
	/* the caller is responsible to call free_fs_root */
1668 1669 1670
	return ret;
}

1671 1672
struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info,
					u64 root_id)
1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687
{
	struct btrfs_root *root;

	spin_lock(&fs_info->fs_roots_radix_lock);
	root = radix_tree_lookup(&fs_info->fs_roots_radix,
				 (unsigned long)root_id);
	spin_unlock(&fs_info->fs_roots_radix_lock);
	return root;
}

int btrfs_insert_fs_root(struct btrfs_fs_info *fs_info,
			 struct btrfs_root *root)
{
	int ret;

1688
	ret = radix_tree_preload(GFP_NOFS);
1689 1690 1691 1692 1693 1694 1695 1696
	if (ret)
		return ret;

	spin_lock(&fs_info->fs_roots_radix_lock);
	ret = radix_tree_insert(&fs_info->fs_roots_radix,
				(unsigned long)root->root_key.objectid,
				root);
	if (ret == 0)
1697
		set_bit(BTRFS_ROOT_IN_RADIX, &root->state);
1698 1699 1700 1701 1702 1703
	spin_unlock(&fs_info->fs_roots_radix_lock);
	radix_tree_preload_end();

	return ret;
}

1704 1705 1706
struct btrfs_root *btrfs_get_fs_root(struct btrfs_fs_info *fs_info,
				     struct btrfs_key *location,
				     bool check_ref)
1707 1708
{
	struct btrfs_root *root;
1709
	struct btrfs_path *path;
1710
	struct btrfs_key key;
1711 1712
	int ret;

1713 1714 1715 1716
	if (location->objectid == BTRFS_ROOT_TREE_OBJECTID)
		return fs_info->tree_root;
	if (location->objectid == BTRFS_EXTENT_TREE_OBJECTID)
		return fs_info->extent_root;
1717 1718 1719 1720
	if (location->objectid == BTRFS_CHUNK_TREE_OBJECTID)
		return fs_info->chunk_root;
	if (location->objectid == BTRFS_DEV_TREE_OBJECTID)
		return fs_info->dev_root;
1721 1722
	if (location->objectid == BTRFS_CSUM_TREE_OBJECTID)
		return fs_info->csum_root;
1723 1724 1725
	if (location->objectid == BTRFS_QUOTA_TREE_OBJECTID)
		return fs_info->quota_root ? fs_info->quota_root :
					     ERR_PTR(-ENOENT);
1726 1727 1728
	if (location->objectid == BTRFS_UUID_TREE_OBJECTID)
		return fs_info->uuid_root ? fs_info->uuid_root :
					    ERR_PTR(-ENOENT);
1729 1730 1731
	if (location->objectid == BTRFS_FREE_SPACE_TREE_OBJECTID)
		return fs_info->free_space_root ? fs_info->free_space_root :
						  ERR_PTR(-ENOENT);
1732
again:
1733
	root = btrfs_lookup_fs_root(fs_info, location->objectid);
1734
	if (root) {
1735
		if (check_ref && btrfs_root_refs(&root->root_item) == 0)
1736
			return ERR_PTR(-ENOENT);
1737
		return root;
1738
	}
1739

1740
	root = btrfs_read_fs_root(fs_info->tree_root, location);
1741 1742
	if (IS_ERR(root))
		return root;
1743

1744
	if (check_ref && btrfs_root_refs(&root->root_item) == 0) {
1745
		ret = -ENOENT;
1746
		goto fail;
1747
	}
1748

1749
	ret = btrfs_init_fs_root(root);
1750 1751
	if (ret)
		goto fail;
1752

1753 1754 1755 1756 1757
	path = btrfs_alloc_path();
	if (!path) {
		ret = -ENOMEM;
		goto fail;
	}
1758 1759 1760 1761 1762
	key.objectid = BTRFS_ORPHAN_OBJECTID;
	key.type = BTRFS_ORPHAN_ITEM_KEY;
	key.offset = location->objectid;

	ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0);
1763
	btrfs_free_path(path);
1764 1765 1766
	if (ret < 0)
		goto fail;
	if (ret == 0)
1767
		set_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &root->state);
1768

1769
	ret = btrfs_insert_fs_root(fs_info, root);
1770
	if (ret) {
1771 1772 1773 1774 1775
		if (ret == -EEXIST) {
			free_fs_root(root);
			goto again;
		}
		goto fail;
1776
	}
1777
	return root;
1778 1779 1780
fail:
	free_fs_root(root);
	return ERR_PTR(ret);
1781 1782
}

C
Chris Mason 已提交
1783 1784 1785 1786 1787 1788
static int btrfs_congested_fn(void *congested_data, int bdi_bits)
{
	struct btrfs_fs_info *info = (struct btrfs_fs_info *)congested_data;
	int ret = 0;
	struct btrfs_device *device;
	struct backing_dev_info *bdi;
C
Chris Mason 已提交
1789

1790 1791
	rcu_read_lock();
	list_for_each_entry_rcu(device, &info->fs_devices->devices, dev_list) {
1792 1793
		if (!device->bdev)
			continue;
C
Chris Mason 已提交
1794
		bdi = blk_get_backing_dev_info(device->bdev);
1795
		if (bdi_congested(bdi, bdi_bits)) {
C
Chris Mason 已提交
1796 1797 1798 1799
			ret = 1;
			break;
		}
	}
1800
	rcu_read_unlock();
C
Chris Mason 已提交
1801 1802 1803 1804 1805
	return ret;
}

static int setup_bdi(struct btrfs_fs_info *info, struct backing_dev_info *bdi)
{
1806 1807
	int err;

1808
	err = bdi_setup_and_register(bdi, "btrfs");
1809 1810 1811
	if (err)
		return err;

1812
	bdi->ra_pages = VM_MAX_READAHEAD * 1024 / PAGE_SIZE;
C
Chris Mason 已提交
1813 1814
	bdi->congested_fn	= btrfs_congested_fn;
	bdi->congested_data	= info;
1815
	bdi->capabilities |= BDI_CAP_CGROUP_WRITEBACK;
C
Chris Mason 已提交
1816 1817 1818
	return 0;
}

1819 1820 1821 1822 1823
/*
 * called by the kthread helper functions to finally call the bio end_io
 * functions.  This is where read checksum verification actually happens
 */
static void end_workqueue_fn(struct btrfs_work *work)
1824 1825
{
	struct bio *bio;
1826
	struct btrfs_end_io_wq *end_io_wq;
1827

1828
	end_io_wq = container_of(work, struct btrfs_end_io_wq, work);
1829
	bio = end_io_wq->bio;
1830

1831
	bio->bi_error = end_io_wq->error;
1832 1833
	bio->bi_private = end_io_wq->private;
	bio->bi_end_io = end_io_wq->end_io;
1834
	kmem_cache_free(btrfs_end_io_wq_cache, end_io_wq);
1835
	bio_endio(bio);
1836 1837
}

1838 1839 1840
static int cleaner_kthread(void *arg)
{
	struct btrfs_root *root = arg;
1841
	int again;
1842
	struct btrfs_trans_handle *trans;
1843 1844

	do {
1845
		again = 0;
1846

1847
		/* Make the cleaner go to sleep early. */
1848
		if (btrfs_need_cleaner_sleep(root))
1849 1850
			goto sleep;

1851 1852 1853 1854
		/*
		 * Do not do anything if we might cause open_ctree() to block
		 * before we have finished mounting the filesystem.
		 */
1855
		if (!test_bit(BTRFS_FS_OPEN, &root->fs_info->flags))
1856 1857
			goto sleep;

1858 1859 1860
		if (!mutex_trylock(&root->fs_info->cleaner_mutex))
			goto sleep;

1861 1862 1863 1864
		/*
		 * Avoid the problem that we change the status of the fs
		 * during the above check and trylock.
		 */
1865
		if (btrfs_need_cleaner_sleep(root)) {
1866 1867
			mutex_unlock(&root->fs_info->cleaner_mutex);
			goto sleep;
1868
		}
1869

1870
		mutex_lock(&root->fs_info->cleaner_delayed_iput_mutex);
1871
		btrfs_run_delayed_iputs(root);
1872 1873
		mutex_unlock(&root->fs_info->cleaner_delayed_iput_mutex);

1874 1875 1876 1877
		again = btrfs_clean_one_deleted_snapshot(root);
		mutex_unlock(&root->fs_info->cleaner_mutex);

		/*
1878 1879
		 * The defragger has dealt with the R/O remount and umount,
		 * needn't do anything special here.
1880 1881
		 */
		btrfs_run_defrag_inodes(root->fs_info);
1882 1883 1884 1885 1886 1887 1888 1889 1890 1891

		/*
		 * Acquires fs_info->delete_unused_bgs_mutex to avoid racing
		 * with relocation (btrfs_relocate_chunk) and relocation
		 * acquires fs_info->cleaner_mutex (btrfs_relocate_block_group)
		 * after acquiring fs_info->delete_unused_bgs_mutex. So we
		 * can't hold, nor need to, fs_info->cleaner_mutex when deleting
		 * unused block groups.
		 */
		btrfs_delete_unused_bgs(root->fs_info);
1892
sleep:
1893
		if (!again) {
1894
			set_current_state(TASK_INTERRUPTIBLE);
1895 1896
			if (!kthread_should_stop())
				schedule();
1897 1898 1899
			__set_current_state(TASK_RUNNING);
		}
	} while (!kthread_should_stop());
1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927

	/*
	 * Transaction kthread is stopped before us and wakes us up.
	 * However we might have started a new transaction and COWed some
	 * tree blocks when deleting unused block groups for example. So
	 * make sure we commit the transaction we started to have a clean
	 * shutdown when evicting the btree inode - if it has dirty pages
	 * when we do the final iput() on it, eviction will trigger a
	 * writeback for it which will fail with null pointer dereferences
	 * since work queues and other resources were already released and
	 * destroyed by the time the iput/eviction/writeback is made.
	 */
	trans = btrfs_attach_transaction(root);
	if (IS_ERR(trans)) {
		if (PTR_ERR(trans) != -ENOENT)
			btrfs_err(root->fs_info,
				  "cleaner transaction attach returned %ld",
				  PTR_ERR(trans));
	} else {
		int ret;

		ret = btrfs_commit_transaction(trans, root);
		if (ret)
			btrfs_err(root->fs_info,
				  "cleaner open transaction commit returned %d",
				  ret);
	}

1928 1929 1930 1931 1932 1933 1934 1935
	return 0;
}

static int transaction_kthread(void *arg)
{
	struct btrfs_root *root = arg;
	struct btrfs_trans_handle *trans;
	struct btrfs_transaction *cur;
1936
	u64 transid;
1937 1938
	unsigned long now;
	unsigned long delay;
1939
	bool cannot_commit;
1940 1941

	do {
1942
		cannot_commit = false;
1943
		delay = HZ * root->fs_info->commit_interval;
1944 1945
		mutex_lock(&root->fs_info->transaction_kthread_mutex);

J
Josef Bacik 已提交
1946
		spin_lock(&root->fs_info->trans_lock);
1947 1948
		cur = root->fs_info->running_transaction;
		if (!cur) {
J
Josef Bacik 已提交
1949
			spin_unlock(&root->fs_info->trans_lock);
1950 1951
			goto sleep;
		}
Y
Yan Zheng 已提交
1952

1953
		now = get_seconds();
1954
		if (cur->state < TRANS_STATE_BLOCKED &&
1955 1956
		    (now < cur->start_time ||
		     now - cur->start_time < root->fs_info->commit_interval)) {
J
Josef Bacik 已提交
1957
			spin_unlock(&root->fs_info->trans_lock);
1958 1959 1960
			delay = HZ * 5;
			goto sleep;
		}
1961
		transid = cur->transid;
J
Josef Bacik 已提交
1962
		spin_unlock(&root->fs_info->trans_lock);
1963

1964
		/* If the file system is aborted, this will always fail. */
1965
		trans = btrfs_attach_transaction(root);
1966
		if (IS_ERR(trans)) {
1967 1968
			if (PTR_ERR(trans) != -ENOENT)
				cannot_commit = true;
1969
			goto sleep;
1970
		}
1971
		if (transid == trans->transid) {
1972
			btrfs_commit_transaction(trans, root);
1973 1974 1975
		} else {
			btrfs_end_transaction(trans, root);
		}
1976 1977 1978 1979
sleep:
		wake_up_process(root->fs_info->cleaner_kthread);
		mutex_unlock(&root->fs_info->transaction_kthread_mutex);

J
Josef Bacik 已提交
1980 1981 1982
		if (unlikely(test_bit(BTRFS_FS_STATE_ERROR,
				      &root->fs_info->fs_state)))
			btrfs_cleanup_transaction(root);
1983 1984 1985 1986 1987 1988
		set_current_state(TASK_INTERRUPTIBLE);
		if (!kthread_should_stop() &&
				(!btrfs_transaction_blocked(root->fs_info) ||
				 cannot_commit))
			schedule_timeout(delay);
		__set_current_state(TASK_RUNNING);
1989 1990 1991 1992
	} while (!kthread_should_stop());
	return 0;
}

C
Chris Mason 已提交
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098
/*
 * this will find the highest generation in the array of
 * root backups.  The index of the highest array is returned,
 * or -1 if we can't find anything.
 *
 * We check to make sure the array is valid by comparing the
 * generation of the latest  root in the array with the generation
 * in the super block.  If they don't match we pitch it.
 */
static int find_newest_super_backup(struct btrfs_fs_info *info, u64 newest_gen)
{
	u64 cur;
	int newest_index = -1;
	struct btrfs_root_backup *root_backup;
	int i;

	for (i = 0; i < BTRFS_NUM_BACKUP_ROOTS; i++) {
		root_backup = info->super_copy->super_roots + i;
		cur = btrfs_backup_tree_root_gen(root_backup);
		if (cur == newest_gen)
			newest_index = i;
	}

	/* check to see if we actually wrapped around */
	if (newest_index == BTRFS_NUM_BACKUP_ROOTS - 1) {
		root_backup = info->super_copy->super_roots;
		cur = btrfs_backup_tree_root_gen(root_backup);
		if (cur == newest_gen)
			newest_index = 0;
	}
	return newest_index;
}


/*
 * find the oldest backup so we know where to store new entries
 * in the backup array.  This will set the backup_root_index
 * field in the fs_info struct
 */
static void find_oldest_super_backup(struct btrfs_fs_info *info,
				     u64 newest_gen)
{
	int newest_index = -1;

	newest_index = find_newest_super_backup(info, newest_gen);
	/* if there was garbage in there, just move along */
	if (newest_index == -1) {
		info->backup_root_index = 0;
	} else {
		info->backup_root_index = (newest_index + 1) % BTRFS_NUM_BACKUP_ROOTS;
	}
}

/*
 * copy all the root pointers into the super backup array.
 * this will bump the backup pointer by one when it is
 * done
 */
static void backup_super_roots(struct btrfs_fs_info *info)
{
	int next_backup;
	struct btrfs_root_backup *root_backup;
	int last_backup;

	next_backup = info->backup_root_index;
	last_backup = (next_backup + BTRFS_NUM_BACKUP_ROOTS - 1) %
		BTRFS_NUM_BACKUP_ROOTS;

	/*
	 * just overwrite the last backup if we're at the same generation
	 * this happens only at umount
	 */
	root_backup = info->super_for_commit->super_roots + last_backup;
	if (btrfs_backup_tree_root_gen(root_backup) ==
	    btrfs_header_generation(info->tree_root->node))
		next_backup = last_backup;

	root_backup = info->super_for_commit->super_roots + next_backup;

	/*
	 * make sure all of our padding and empty slots get zero filled
	 * regardless of which ones we use today
	 */
	memset(root_backup, 0, sizeof(*root_backup));

	info->backup_root_index = (next_backup + 1) % BTRFS_NUM_BACKUP_ROOTS;

	btrfs_set_backup_tree_root(root_backup, info->tree_root->node->start);
	btrfs_set_backup_tree_root_gen(root_backup,
			       btrfs_header_generation(info->tree_root->node));

	btrfs_set_backup_tree_root_level(root_backup,
			       btrfs_header_level(info->tree_root->node));

	btrfs_set_backup_chunk_root(root_backup, info->chunk_root->node->start);
	btrfs_set_backup_chunk_root_gen(root_backup,
			       btrfs_header_generation(info->chunk_root->node));
	btrfs_set_backup_chunk_root_level(root_backup,
			       btrfs_header_level(info->chunk_root->node));

	btrfs_set_backup_extent_root(root_backup, info->extent_root->node->start);
	btrfs_set_backup_extent_root_gen(root_backup,
			       btrfs_header_generation(info->extent_root->node));
	btrfs_set_backup_extent_root_level(root_backup,
			       btrfs_header_level(info->extent_root->node));

2099 2100 2101 2102 2103 2104 2105 2106
	/*
	 * we might commit during log recovery, which happens before we set
	 * the fs_root.  Make sure it is valid before we fill it in.
	 */
	if (info->fs_root && info->fs_root->node) {
		btrfs_set_backup_fs_root(root_backup,
					 info->fs_root->node->start);
		btrfs_set_backup_fs_root_gen(root_backup,
C
Chris Mason 已提交
2107
			       btrfs_header_generation(info->fs_root->node));
2108
		btrfs_set_backup_fs_root_level(root_backup,
C
Chris Mason 已提交
2109
			       btrfs_header_level(info->fs_root->node));
2110
	}
C
Chris Mason 已提交
2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191

	btrfs_set_backup_dev_root(root_backup, info->dev_root->node->start);
	btrfs_set_backup_dev_root_gen(root_backup,
			       btrfs_header_generation(info->dev_root->node));
	btrfs_set_backup_dev_root_level(root_backup,
				       btrfs_header_level(info->dev_root->node));

	btrfs_set_backup_csum_root(root_backup, info->csum_root->node->start);
	btrfs_set_backup_csum_root_gen(root_backup,
			       btrfs_header_generation(info->csum_root->node));
	btrfs_set_backup_csum_root_level(root_backup,
			       btrfs_header_level(info->csum_root->node));

	btrfs_set_backup_total_bytes(root_backup,
			     btrfs_super_total_bytes(info->super_copy));
	btrfs_set_backup_bytes_used(root_backup,
			     btrfs_super_bytes_used(info->super_copy));
	btrfs_set_backup_num_devices(root_backup,
			     btrfs_super_num_devices(info->super_copy));

	/*
	 * if we don't copy this out to the super_copy, it won't get remembered
	 * for the next commit
	 */
	memcpy(&info->super_copy->super_roots,
	       &info->super_for_commit->super_roots,
	       sizeof(*root_backup) * BTRFS_NUM_BACKUP_ROOTS);
}

/*
 * this copies info out of the root backup array and back into
 * the in-memory super block.  It is meant to help iterate through
 * the array, so you send it the number of backups you've already
 * tried and the last backup index you used.
 *
 * this returns -1 when it has tried all the backups
 */
static noinline int next_root_backup(struct btrfs_fs_info *info,
				     struct btrfs_super_block *super,
				     int *num_backups_tried, int *backup_index)
{
	struct btrfs_root_backup *root_backup;
	int newest = *backup_index;

	if (*num_backups_tried == 0) {
		u64 gen = btrfs_super_generation(super);

		newest = find_newest_super_backup(info, gen);
		if (newest == -1)
			return -1;

		*backup_index = newest;
		*num_backups_tried = 1;
	} else if (*num_backups_tried == BTRFS_NUM_BACKUP_ROOTS) {
		/* we've tried all the backups, all done */
		return -1;
	} else {
		/* jump to the next oldest backup */
		newest = (*backup_index + BTRFS_NUM_BACKUP_ROOTS - 1) %
			BTRFS_NUM_BACKUP_ROOTS;
		*backup_index = newest;
		*num_backups_tried += 1;
	}
	root_backup = super->super_roots + newest;

	btrfs_set_super_generation(super,
				   btrfs_backup_tree_root_gen(root_backup));
	btrfs_set_super_root(super, btrfs_backup_tree_root(root_backup));
	btrfs_set_super_root_level(super,
				   btrfs_backup_tree_root_level(root_backup));
	btrfs_set_super_bytes_used(super, btrfs_backup_bytes_used(root_backup));

	/*
	 * fixme: the total bytes and num_devices need to match or we should
	 * need a fsck
	 */
	btrfs_set_super_total_bytes(super, btrfs_backup_total_bytes(root_backup));
	btrfs_set_super_num_devices(super, btrfs_backup_num_devices(root_backup));
	return 0;
}

L
Liu Bo 已提交
2192 2193 2194
/* helper to cleanup workers */
static void btrfs_stop_all_workers(struct btrfs_fs_info *fs_info)
{
2195
	btrfs_destroy_workqueue(fs_info->fixup_workers);
2196
	btrfs_destroy_workqueue(fs_info->delalloc_workers);
2197
	btrfs_destroy_workqueue(fs_info->workers);
2198 2199 2200
	btrfs_destroy_workqueue(fs_info->endio_workers);
	btrfs_destroy_workqueue(fs_info->endio_meta_workers);
	btrfs_destroy_workqueue(fs_info->endio_raid56_workers);
2201
	btrfs_destroy_workqueue(fs_info->endio_repair_workers);
2202
	btrfs_destroy_workqueue(fs_info->rmw_workers);
2203 2204 2205
	btrfs_destroy_workqueue(fs_info->endio_meta_write_workers);
	btrfs_destroy_workqueue(fs_info->endio_write_workers);
	btrfs_destroy_workqueue(fs_info->endio_freespace_worker);
2206
	btrfs_destroy_workqueue(fs_info->submit_workers);
2207
	btrfs_destroy_workqueue(fs_info->delayed_workers);
2208
	btrfs_destroy_workqueue(fs_info->caching_workers);
2209
	btrfs_destroy_workqueue(fs_info->readahead_workers);
2210
	btrfs_destroy_workqueue(fs_info->flush_workers);
2211
	btrfs_destroy_workqueue(fs_info->qgroup_rescan_workers);
C
Chris Mason 已提交
2212
	btrfs_destroy_workqueue(fs_info->extent_workers);
L
Liu Bo 已提交
2213 2214
}

2215 2216 2217 2218 2219 2220 2221 2222 2223 2224
static void free_root_extent_buffers(struct btrfs_root *root)
{
	if (root) {
		free_extent_buffer(root->node);
		free_extent_buffer(root->commit_root);
		root->node = NULL;
		root->commit_root = NULL;
	}
}

C
Chris Mason 已提交
2225 2226 2227
/* helper to cleanup tree roots */
static void free_root_pointers(struct btrfs_fs_info *info, int chunk_root)
{
2228
	free_root_extent_buffers(info->tree_root);
2229

2230 2231 2232 2233 2234 2235 2236
	free_root_extent_buffers(info->dev_root);
	free_root_extent_buffers(info->extent_root);
	free_root_extent_buffers(info->csum_root);
	free_root_extent_buffers(info->quota_root);
	free_root_extent_buffers(info->uuid_root);
	if (chunk_root)
		free_root_extent_buffers(info->chunk_root);
2237
	free_root_extent_buffers(info->free_space_root);
C
Chris Mason 已提交
2238 2239
}

2240
void btrfs_free_fs_roots(struct btrfs_fs_info *fs_info)
2241 2242 2243 2244 2245 2246 2247 2248 2249 2250
{
	int ret;
	struct btrfs_root *gang[8];
	int i;

	while (!list_empty(&fs_info->dead_roots)) {
		gang[0] = list_entry(fs_info->dead_roots.next,
				     struct btrfs_root, root_list);
		list_del(&gang[0]->root_list);

2251
		if (test_bit(BTRFS_ROOT_IN_RADIX, &gang[0]->state)) {
2252
			btrfs_drop_and_free_fs_root(fs_info, gang[0]);
2253 2254 2255
		} else {
			free_extent_buffer(gang[0]->node);
			free_extent_buffer(gang[0]->commit_root);
2256
			btrfs_put_fs_root(gang[0]);
2257 2258 2259 2260 2261 2262 2263 2264 2265 2266
		}
	}

	while (1) {
		ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix,
					     (void **)gang, 0,
					     ARRAY_SIZE(gang));
		if (!ret)
			break;
		for (i = 0; i < ret; i++)
2267
			btrfs_drop_and_free_fs_root(fs_info, gang[i]);
2268
	}
2269 2270 2271 2272 2273 2274

	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
		btrfs_free_log_root_tree(NULL, fs_info);
		btrfs_destroy_pinned_extent(fs_info->tree_root,
					    fs_info->pinned_extents);
	}
2275
}
C
Chris Mason 已提交
2276

2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287
static void btrfs_init_scrub(struct btrfs_fs_info *fs_info)
{
	mutex_init(&fs_info->scrub_lock);
	atomic_set(&fs_info->scrubs_running, 0);
	atomic_set(&fs_info->scrub_pause_req, 0);
	atomic_set(&fs_info->scrubs_paused, 0);
	atomic_set(&fs_info->scrub_cancel_req, 0);
	init_waitqueue_head(&fs_info->scrub_pause_wait);
	fs_info->scrub_workers_refcnt = 0;
}

2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298
static void btrfs_init_balance(struct btrfs_fs_info *fs_info)
{
	spin_lock_init(&fs_info->balance_lock);
	mutex_init(&fs_info->balance_mutex);
	atomic_set(&fs_info->balance_running, 0);
	atomic_set(&fs_info->balance_pause_req, 0);
	atomic_set(&fs_info->balance_cancel_req, 0);
	fs_info->balance_ctl = NULL;
	init_waitqueue_head(&fs_info->balance_wait_q);
}

2299
static void btrfs_init_btree_inode(struct btrfs_fs_info *fs_info)
2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318
{
	fs_info->btree_inode->i_ino = BTRFS_BTREE_INODE_OBJECTID;
	set_nlink(fs_info->btree_inode, 1);
	/*
	 * we set the i_size on the btree inode to the max possible int.
	 * the real end of the address space is determined by all of
	 * the devices in the system
	 */
	fs_info->btree_inode->i_size = OFFSET_MAX;
	fs_info->btree_inode->i_mapping->a_ops = &btree_aops;

	RB_CLEAR_NODE(&BTRFS_I(fs_info->btree_inode)->rb_node);
	extent_io_tree_init(&BTRFS_I(fs_info->btree_inode)->io_tree,
			     fs_info->btree_inode->i_mapping);
	BTRFS_I(fs_info->btree_inode)->io_tree.track_uptodate = 0;
	extent_map_tree_init(&BTRFS_I(fs_info->btree_inode)->extent_tree);

	BTRFS_I(fs_info->btree_inode)->io_tree.ops = &btree_extent_io_ops;

2319
	BTRFS_I(fs_info->btree_inode)->root = fs_info->tree_root;
2320 2321 2322 2323 2324 2325 2326
	memset(&BTRFS_I(fs_info->btree_inode)->location, 0,
	       sizeof(struct btrfs_key));
	set_bit(BTRFS_INODE_DUMMY,
		&BTRFS_I(fs_info->btree_inode)->runtime_flags);
	btrfs_insert_inode_hash(fs_info->btree_inode);
}

2327 2328 2329 2330 2331
static void btrfs_init_dev_replace_locks(struct btrfs_fs_info *fs_info)
{
	fs_info->dev_replace.lock_owner = 0;
	atomic_set(&fs_info->dev_replace.nesting_level, 0);
	mutex_init(&fs_info->dev_replace.lock_finishing_cancel_unmount);
2332 2333 2334
	rwlock_init(&fs_info->dev_replace.lock);
	atomic_set(&fs_info->dev_replace.read_locks, 0);
	atomic_set(&fs_info->dev_replace.blocking_readers, 0);
2335
	init_waitqueue_head(&fs_info->replace_wait);
2336
	init_waitqueue_head(&fs_info->dev_replace.read_lock_wq);
2337 2338
}

2339 2340 2341 2342 2343 2344 2345 2346 2347
static void btrfs_init_qgroup(struct btrfs_fs_info *fs_info)
{
	spin_lock_init(&fs_info->qgroup_lock);
	mutex_init(&fs_info->qgroup_ioctl_lock);
	fs_info->qgroup_tree = RB_ROOT;
	fs_info->qgroup_op_tree = RB_ROOT;
	INIT_LIST_HEAD(&fs_info->dirty_qgroups);
	fs_info->qgroup_seq = 1;
	fs_info->qgroup_ulist = NULL;
2348
	fs_info->qgroup_rescan_running = false;
2349 2350 2351
	mutex_init(&fs_info->qgroup_rescan_lock);
}

2352 2353 2354 2355
static int btrfs_init_workqueues(struct btrfs_fs_info *fs_info,
		struct btrfs_fs_devices *fs_devices)
{
	int max_active = fs_info->thread_pool_size;
2356
	unsigned int flags = WQ_MEM_RECLAIM | WQ_FREEZABLE | WQ_UNBOUND;
2357 2358

	fs_info->workers =
2359 2360
		btrfs_alloc_workqueue(fs_info, "worker",
				      flags | WQ_HIGHPRI, max_active, 16);
2361 2362

	fs_info->delalloc_workers =
2363 2364
		btrfs_alloc_workqueue(fs_info, "delalloc",
				      flags, max_active, 2);
2365 2366

	fs_info->flush_workers =
2367 2368
		btrfs_alloc_workqueue(fs_info, "flush_delalloc",
				      flags, max_active, 0);
2369 2370

	fs_info->caching_workers =
2371
		btrfs_alloc_workqueue(fs_info, "cache", flags, max_active, 0);
2372 2373 2374 2375 2376 2377 2378

	/*
	 * a higher idle thresh on the submit workers makes it much more
	 * likely that bios will be send down in a sane order to the
	 * devices
	 */
	fs_info->submit_workers =
2379
		btrfs_alloc_workqueue(fs_info, "submit", flags,
2380 2381 2382 2383
				      min_t(u64, fs_devices->num_devices,
					    max_active), 64);

	fs_info->fixup_workers =
2384
		btrfs_alloc_workqueue(fs_info, "fixup", flags, 1, 0);
2385 2386 2387 2388 2389 2390

	/*
	 * endios are largely parallel and should have a very
	 * low idle thresh
	 */
	fs_info->endio_workers =
2391
		btrfs_alloc_workqueue(fs_info, "endio", flags, max_active, 4);
2392
	fs_info->endio_meta_workers =
2393 2394
		btrfs_alloc_workqueue(fs_info, "endio-meta", flags,
				      max_active, 4);
2395
	fs_info->endio_meta_write_workers =
2396 2397
		btrfs_alloc_workqueue(fs_info, "endio-meta-write", flags,
				      max_active, 2);
2398
	fs_info->endio_raid56_workers =
2399 2400
		btrfs_alloc_workqueue(fs_info, "endio-raid56", flags,
				      max_active, 4);
2401
	fs_info->endio_repair_workers =
2402
		btrfs_alloc_workqueue(fs_info, "endio-repair", flags, 1, 0);
2403
	fs_info->rmw_workers =
2404
		btrfs_alloc_workqueue(fs_info, "rmw", flags, max_active, 2);
2405
	fs_info->endio_write_workers =
2406 2407
		btrfs_alloc_workqueue(fs_info, "endio-write", flags,
				      max_active, 2);
2408
	fs_info->endio_freespace_worker =
2409 2410
		btrfs_alloc_workqueue(fs_info, "freespace-write", flags,
				      max_active, 0);
2411
	fs_info->delayed_workers =
2412 2413
		btrfs_alloc_workqueue(fs_info, "delayed-meta", flags,
				      max_active, 0);
2414
	fs_info->readahead_workers =
2415 2416
		btrfs_alloc_workqueue(fs_info, "readahead", flags,
				      max_active, 2);
2417
	fs_info->qgroup_rescan_workers =
2418
		btrfs_alloc_workqueue(fs_info, "qgroup-rescan", flags, 1, 0);
2419
	fs_info->extent_workers =
2420
		btrfs_alloc_workqueue(fs_info, "extent-refs", flags,
2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440
				      min_t(u64, fs_devices->num_devices,
					    max_active), 8);

	if (!(fs_info->workers && fs_info->delalloc_workers &&
	      fs_info->submit_workers && fs_info->flush_workers &&
	      fs_info->endio_workers && fs_info->endio_meta_workers &&
	      fs_info->endio_meta_write_workers &&
	      fs_info->endio_repair_workers &&
	      fs_info->endio_write_workers && fs_info->endio_raid56_workers &&
	      fs_info->endio_freespace_worker && fs_info->rmw_workers &&
	      fs_info->caching_workers && fs_info->readahead_workers &&
	      fs_info->fixup_workers && fs_info->delayed_workers &&
	      fs_info->extent_workers &&
	      fs_info->qgroup_rescan_workers)) {
		return -ENOMEM;
	}

	return 0;
}

2441 2442 2443 2444 2445 2446 2447 2448 2449 2450
static int btrfs_replay_log(struct btrfs_fs_info *fs_info,
			    struct btrfs_fs_devices *fs_devices)
{
	int ret;
	struct btrfs_root *tree_root = fs_info->tree_root;
	struct btrfs_root *log_tree_root;
	struct btrfs_super_block *disk_super = fs_info->super_copy;
	u64 bytenr = btrfs_super_log_root(disk_super);

	if (fs_devices->rw_devices == 0) {
2451
		btrfs_warn(fs_info, "log replay required on RO media");
2452 2453 2454
		return -EIO;
	}

2455
	log_tree_root = btrfs_alloc_root(fs_info, GFP_KERNEL);
2456 2457 2458 2459 2460 2461 2462 2463 2464
	if (!log_tree_root)
		return -ENOMEM;

	__setup_root(tree_root->nodesize, tree_root->sectorsize,
			tree_root->stripesize, log_tree_root, fs_info,
			BTRFS_TREE_LOG_OBJECTID);

	log_tree_root->node = read_tree_block(tree_root, bytenr,
			fs_info->generation + 1);
2465
	if (IS_ERR(log_tree_root->node)) {
2466
		btrfs_warn(fs_info, "failed to read log tree");
2467
		ret = PTR_ERR(log_tree_root->node);
2468
		kfree(log_tree_root);
2469
		return ret;
2470
	} else if (!extent_buffer_uptodate(log_tree_root->node)) {
2471
		btrfs_err(fs_info, "failed to read log tree");
2472 2473 2474 2475 2476 2477 2478
		free_extent_buffer(log_tree_root->node);
		kfree(log_tree_root);
		return -EIO;
	}
	/* returns with log_tree_root freed on success */
	ret = btrfs_recover_log_trees(log_tree_root);
	if (ret) {
2479
		btrfs_handle_fs_error(tree_root->fs_info, ret,
2480 2481 2482 2483 2484 2485 2486
			    "Failed to recover log tree");
		free_extent_buffer(log_tree_root->node);
		kfree(log_tree_root);
		return ret;
	}

	if (fs_info->sb->s_flags & MS_RDONLY) {
2487
		ret = btrfs_commit_super(fs_info);
2488 2489 2490 2491 2492 2493 2494
		if (ret)
			return ret;
	}

	return 0;
}

2495
static int btrfs_read_roots(struct btrfs_fs_info *fs_info)
2496
{
2497
	struct btrfs_root *tree_root = fs_info->tree_root;
2498
	struct btrfs_root *root;
2499 2500 2501
	struct btrfs_key location;
	int ret;

2502 2503
	BUG_ON(!fs_info->tree_root);

2504 2505 2506 2507
	location.objectid = BTRFS_EXTENT_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
	location.offset = 0;

2508 2509 2510 2511 2512
	root = btrfs_read_tree_root(tree_root, &location);
	if (IS_ERR(root))
		return PTR_ERR(root);
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
	fs_info->extent_root = root;
2513 2514

	location.objectid = BTRFS_DEV_TREE_OBJECTID;
2515 2516 2517 2518 2519
	root = btrfs_read_tree_root(tree_root, &location);
	if (IS_ERR(root))
		return PTR_ERR(root);
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
	fs_info->dev_root = root;
2520 2521 2522
	btrfs_init_devices_late(fs_info);

	location.objectid = BTRFS_CSUM_TREE_OBJECTID;
2523 2524 2525 2526 2527
	root = btrfs_read_tree_root(tree_root, &location);
	if (IS_ERR(root))
		return PTR_ERR(root);
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
	fs_info->csum_root = root;
2528 2529

	location.objectid = BTRFS_QUOTA_TREE_OBJECTID;
2530 2531 2532
	root = btrfs_read_tree_root(tree_root, &location);
	if (!IS_ERR(root)) {
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
2533
		set_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags);
2534
		fs_info->quota_root = root;
2535 2536 2537
	}

	location.objectid = BTRFS_UUID_TREE_OBJECTID;
2538 2539 2540
	root = btrfs_read_tree_root(tree_root, &location);
	if (IS_ERR(root)) {
		ret = PTR_ERR(root);
2541 2542 2543
		if (ret != -ENOENT)
			return ret;
	} else {
2544 2545
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
		fs_info->uuid_root = root;
2546 2547
	}

2548 2549 2550 2551 2552 2553 2554 2555 2556
	if (btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
		location.objectid = BTRFS_FREE_SPACE_TREE_OBJECTID;
		root = btrfs_read_tree_root(tree_root, &location);
		if (IS_ERR(root))
			return PTR_ERR(root);
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
		fs_info->free_space_root = root;
	}

2557 2558 2559
	return 0;
}

A
Al Viro 已提交
2560 2561 2562
int open_ctree(struct super_block *sb,
	       struct btrfs_fs_devices *fs_devices,
	       char *options)
2563
{
2564 2565
	u32 sectorsize;
	u32 nodesize;
2566
	u32 stripesize;
2567
	u64 generation;
2568
	u64 features;
2569
	struct btrfs_key location;
2570
	struct buffer_head *bh;
2571
	struct btrfs_super_block *disk_super;
2572
	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
2573
	struct btrfs_root *tree_root;
2574
	struct btrfs_root *chunk_root;
2575
	int ret;
2576
	int err = -EINVAL;
C
Chris Mason 已提交
2577 2578
	int num_backups_tried = 0;
	int backup_index = 0;
2579
	int max_active;
2580
	int clear_free_space_tree = 0;
2581

2582 2583
	tree_root = fs_info->tree_root = btrfs_alloc_root(fs_info, GFP_KERNEL);
	chunk_root = fs_info->chunk_root = btrfs_alloc_root(fs_info, GFP_KERNEL);
2584
	if (!tree_root || !chunk_root) {
C
Chris Mason 已提交
2585 2586 2587
		err = -ENOMEM;
		goto fail;
	}
2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600

	ret = init_srcu_struct(&fs_info->subvol_srcu);
	if (ret) {
		err = ret;
		goto fail;
	}

	ret = setup_bdi(fs_info, &fs_info->bdi);
	if (ret) {
		err = ret;
		goto fail_srcu;
	}

2601
	ret = percpu_counter_init(&fs_info->dirty_metadata_bytes, 0, GFP_KERNEL);
2602 2603 2604 2605
	if (ret) {
		err = ret;
		goto fail_bdi;
	}
2606
	fs_info->dirty_metadata_batch = PAGE_SIZE *
2607 2608
					(1 + ilog2(nr_cpu_ids));

2609
	ret = percpu_counter_init(&fs_info->delalloc_bytes, 0, GFP_KERNEL);
2610 2611 2612 2613 2614
	if (ret) {
		err = ret;
		goto fail_dirty_metadata_bytes;
	}

2615
	ret = percpu_counter_init(&fs_info->bio_counter, 0, GFP_KERNEL);
2616 2617 2618 2619 2620
	if (ret) {
		err = ret;
		goto fail_delalloc_bytes;
	}

2621 2622 2623
	fs_info->btree_inode = new_inode(sb);
	if (!fs_info->btree_inode) {
		err = -ENOMEM;
2624
		goto fail_bio_counter;
2625 2626
	}

2627
	mapping_set_gfp_mask(fs_info->btree_inode->i_mapping, GFP_NOFS);
2628

2629
	INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC);
2630
	INIT_RADIX_TREE(&fs_info->buffer_radix, GFP_ATOMIC);
C
Chris Mason 已提交
2631
	INIT_LIST_HEAD(&fs_info->trans_list);
2632
	INIT_LIST_HEAD(&fs_info->dead_roots);
Y
Yan, Zheng 已提交
2633
	INIT_LIST_HEAD(&fs_info->delayed_iputs);
2634
	INIT_LIST_HEAD(&fs_info->delalloc_roots);
2635
	INIT_LIST_HEAD(&fs_info->caching_block_groups);
2636
	spin_lock_init(&fs_info->delalloc_root_lock);
J
Josef Bacik 已提交
2637
	spin_lock_init(&fs_info->trans_lock);
2638
	spin_lock_init(&fs_info->fs_roots_radix_lock);
Y
Yan, Zheng 已提交
2639
	spin_lock_init(&fs_info->delayed_iput_lock);
C
Chris Mason 已提交
2640
	spin_lock_init(&fs_info->defrag_inodes_lock);
2641
	spin_lock_init(&fs_info->free_chunk_lock);
J
Jan Schmidt 已提交
2642
	spin_lock_init(&fs_info->tree_mod_seq_lock);
2643
	spin_lock_init(&fs_info->super_lock);
J
Josef Bacik 已提交
2644
	spin_lock_init(&fs_info->qgroup_op_lock);
2645
	spin_lock_init(&fs_info->buffer_lock);
2646
	spin_lock_init(&fs_info->unused_bgs_lock);
J
Jan Schmidt 已提交
2647
	rwlock_init(&fs_info->tree_mod_log_lock);
2648
	mutex_init(&fs_info->unused_bg_unpin_mutex);
2649
	mutex_init(&fs_info->delete_unused_bgs_mutex);
C
Chris Mason 已提交
2650
	mutex_init(&fs_info->reloc_mutex);
2651
	mutex_init(&fs_info->delalloc_root_mutex);
2652
	mutex_init(&fs_info->cleaner_delayed_iput_mutex);
2653
	seqlock_init(&fs_info->profiles_lock);
2654

2655
	INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots);
2656
	INIT_LIST_HEAD(&fs_info->space_info);
J
Jan Schmidt 已提交
2657
	INIT_LIST_HEAD(&fs_info->tree_mod_seq_list);
2658
	INIT_LIST_HEAD(&fs_info->unused_bgs);
2659
	btrfs_mapping_init(&fs_info->mapping_tree);
2660 2661 2662 2663 2664 2665 2666 2667 2668
	btrfs_init_block_rsv(&fs_info->global_block_rsv,
			     BTRFS_BLOCK_RSV_GLOBAL);
	btrfs_init_block_rsv(&fs_info->delalloc_block_rsv,
			     BTRFS_BLOCK_RSV_DELALLOC);
	btrfs_init_block_rsv(&fs_info->trans_block_rsv, BTRFS_BLOCK_RSV_TRANS);
	btrfs_init_block_rsv(&fs_info->chunk_block_rsv, BTRFS_BLOCK_RSV_CHUNK);
	btrfs_init_block_rsv(&fs_info->empty_block_rsv, BTRFS_BLOCK_RSV_EMPTY);
	btrfs_init_block_rsv(&fs_info->delayed_block_rsv,
			     BTRFS_BLOCK_RSV_DELOPS);
2669
	atomic_set(&fs_info->nr_async_submits, 0);
2670
	atomic_set(&fs_info->async_delalloc_pages, 0);
2671
	atomic_set(&fs_info->async_submit_draining, 0);
2672
	atomic_set(&fs_info->nr_async_bios, 0);
C
Chris Mason 已提交
2673
	atomic_set(&fs_info->defrag_running, 0);
J
Josef Bacik 已提交
2674
	atomic_set(&fs_info->qgroup_op_seq, 0);
Z
Zhao Lei 已提交
2675
	atomic_set(&fs_info->reada_works_cnt, 0);
2676
	atomic64_set(&fs_info->tree_mod_seq, 0);
2677
	fs_info->fs_frozen = 0;
C
Chris Mason 已提交
2678
	fs_info->sb = sb;
2679
	fs_info->max_inline = BTRFS_DEFAULT_MAX_INLINE;
J
Josef Bacik 已提交
2680
	fs_info->metadata_ratio = 0;
C
Chris Mason 已提交
2681
	fs_info->defrag_inodes = RB_ROOT;
2682
	fs_info->free_chunk_space = 0;
J
Jan Schmidt 已提交
2683
	fs_info->tree_mod_log = RB_ROOT;
2684
	fs_info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
2685
	fs_info->avg_delayed_ref_runtime = NSEC_PER_SEC >> 6; /* div by 64 */
2686
	/* readahead state */
2687
	INIT_RADIX_TREE(&fs_info->reada_tree, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
2688
	spin_lock_init(&fs_info->reada_lock);
C
Chris Mason 已提交
2689

2690 2691
	fs_info->thread_pool_size = min_t(unsigned long,
					  num_online_cpus() + 2, 8);
2692

2693 2694
	INIT_LIST_HEAD(&fs_info->ordered_roots);
	spin_lock_init(&fs_info->ordered_root_lock);
2695
	fs_info->delayed_root = kmalloc(sizeof(struct btrfs_delayed_root),
2696
					GFP_KERNEL);
2697 2698 2699 2700 2701
	if (!fs_info->delayed_root) {
		err = -ENOMEM;
		goto fail_iput;
	}
	btrfs_init_delayed_root(fs_info->delayed_root);
2702

2703
	btrfs_init_scrub(fs_info);
2704 2705 2706
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	fs_info->check_integrity_print_mask = 0;
#endif
2707
	btrfs_init_balance(fs_info);
2708
	btrfs_init_async_reclaim_work(&fs_info->async_reclaim_work);
A
Arne Jansen 已提交
2709

2710 2711
	sb->s_blocksize = 4096;
	sb->s_blocksize_bits = blksize_bits(4096);
J
Jens Axboe 已提交
2712
	sb->s_bdi = &fs_info->bdi;
2713

2714
	btrfs_init_btree_inode(fs_info);
2715

J
Josef Bacik 已提交
2716
	spin_lock_init(&fs_info->block_group_cache_lock);
2717
	fs_info->block_group_cache_tree = RB_ROOT;
2718
	fs_info->first_logical_byte = (u64)-1;
J
Josef Bacik 已提交
2719

2720
	extent_io_tree_init(&fs_info->freed_extents[0],
2721
			     fs_info->btree_inode->i_mapping);
2722
	extent_io_tree_init(&fs_info->freed_extents[1],
2723
			     fs_info->btree_inode->i_mapping);
2724
	fs_info->pinned_extents = &fs_info->freed_extents[0];
2725
	set_bit(BTRFS_FS_BARRIER, &fs_info->flags);
C
Chris Mason 已提交
2726

2727
	mutex_init(&fs_info->ordered_operations_mutex);
2728
	mutex_init(&fs_info->tree_log_mutex);
2729
	mutex_init(&fs_info->chunk_mutex);
2730 2731
	mutex_init(&fs_info->transaction_kthread_mutex);
	mutex_init(&fs_info->cleaner_mutex);
2732
	mutex_init(&fs_info->volume_mutex);
2733
	mutex_init(&fs_info->ro_block_group_mutex);
2734
	init_rwsem(&fs_info->commit_root_sem);
2735
	init_rwsem(&fs_info->cleanup_work_sem);
2736
	init_rwsem(&fs_info->subvol_sem);
S
Stefan Behrens 已提交
2737
	sema_init(&fs_info->uuid_tree_rescan_sem, 1);
2738

2739
	btrfs_init_dev_replace_locks(fs_info);
2740
	btrfs_init_qgroup(fs_info);
2741

2742 2743 2744
	btrfs_init_free_cluster(&fs_info->meta_alloc_cluster);
	btrfs_init_free_cluster(&fs_info->data_alloc_cluster);

2745
	init_waitqueue_head(&fs_info->transaction_throttle);
2746
	init_waitqueue_head(&fs_info->transaction_wait);
S
Sage Weil 已提交
2747
	init_waitqueue_head(&fs_info->transaction_blocked_wait);
2748
	init_waitqueue_head(&fs_info->async_submit_wait);
2749

2750 2751
	INIT_LIST_HEAD(&fs_info->pinned_chunks);

D
David Woodhouse 已提交
2752 2753
	ret = btrfs_alloc_stripe_hash_table(fs_info);
	if (ret) {
2754
		err = ret;
D
David Woodhouse 已提交
2755 2756 2757
		goto fail_alloc;
	}

2758
	__setup_root(4096, 4096, 4096, tree_root,
C
Chris Mason 已提交
2759
		     fs_info, BTRFS_ROOT_TREE_OBJECTID);
2760

2761
	invalidate_bdev(fs_devices->latest_bdev);
D
David Sterba 已提交
2762 2763 2764 2765

	/*
	 * Read super block and check the signature bytes only
	 */
Y
Yan Zheng 已提交
2766
	bh = btrfs_read_dev_super(fs_devices->latest_bdev);
2767 2768
	if (IS_ERR(bh)) {
		err = PTR_ERR(bh);
2769
		goto fail_alloc;
2770
	}
C
Chris Mason 已提交
2771

D
David Sterba 已提交
2772 2773 2774 2775
	/*
	 * We want to check superblock checksum, the type is stored inside.
	 * Pass the whole disk block of size BTRFS_SUPER_INFO_SIZE (4k).
	 */
2776
	if (btrfs_check_super_csum(fs_info, bh->b_data)) {
2777
		btrfs_err(fs_info, "superblock checksum mismatch");
D
David Sterba 已提交
2778
		err = -EINVAL;
2779
		brelse(bh);
D
David Sterba 已提交
2780 2781 2782 2783 2784 2785 2786 2787
		goto fail_alloc;
	}

	/*
	 * super_copy is zeroed at allocation time and we never touch the
	 * following bytes up to INFO_SIZE, the checksum is calculated from
	 * the whole block of INFO_SIZE
	 */
2788 2789 2790
	memcpy(fs_info->super_copy, bh->b_data, sizeof(*fs_info->super_copy));
	memcpy(fs_info->super_for_commit, fs_info->super_copy,
	       sizeof(*fs_info->super_for_commit));
2791
	brelse(bh);
2792

2793
	memcpy(fs_info->fsid, fs_info->super_copy->fsid, BTRFS_FSID_SIZE);
2794

D
David Sterba 已提交
2795 2796
	ret = btrfs_check_super_valid(fs_info, sb->s_flags & MS_RDONLY);
	if (ret) {
2797
		btrfs_err(fs_info, "superblock contains fatal errors");
D
David Sterba 已提交
2798 2799 2800 2801
		err = -EINVAL;
		goto fail_alloc;
	}

2802
	disk_super = fs_info->super_copy;
2803
	if (!btrfs_super_root(disk_super))
2804
		goto fail_alloc;
2805

L
liubo 已提交
2806
	/* check FS state, whether FS is broken. */
2807 2808
	if (btrfs_super_flags(disk_super) & BTRFS_SUPER_FLAG_ERROR)
		set_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state);
L
liubo 已提交
2809

C
Chris Mason 已提交
2810 2811 2812 2813 2814 2815 2816
	/*
	 * run through our array of backup supers and setup
	 * our ring pointer to the oldest one
	 */
	generation = btrfs_super_generation(disk_super);
	find_oldest_super_backup(fs_info, generation);

2817 2818 2819 2820 2821 2822
	/*
	 * In the long term, we'll store the compression type in the super
	 * block, and it'll be used for per file compression control.
	 */
	fs_info->compress_type = BTRFS_COMPRESS_ZLIB;

2823
	ret = btrfs_parse_options(tree_root, options, sb->s_flags);
Y
Yan Zheng 已提交
2824 2825
	if (ret) {
		err = ret;
2826
		goto fail_alloc;
Y
Yan Zheng 已提交
2827
	}
2828

2829 2830 2831
	features = btrfs_super_incompat_flags(disk_super) &
		~BTRFS_FEATURE_INCOMPAT_SUPP;
	if (features) {
2832 2833 2834
		btrfs_err(fs_info,
		    "cannot mount because of unsupported optional features (%llx)",
		    features);
2835
		err = -EINVAL;
2836
		goto fail_alloc;
2837 2838
	}

2839
	features = btrfs_super_incompat_flags(disk_super);
L
Li Zefan 已提交
2840
	features |= BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF;
L
Li Zefan 已提交
2841
	if (tree_root->fs_info->compress_type == BTRFS_COMPRESS_LZO)
L
Li Zefan 已提交
2842
		features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
2843

2844
	if (features & BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA)
2845
		btrfs_info(fs_info, "has skinny extents");
2846

2847 2848 2849 2850
	/*
	 * flag our filesystem as having big metadata blocks if
	 * they are bigger than the page size
	 */
2851
	if (btrfs_super_nodesize(disk_super) > PAGE_SIZE) {
2852
		if (!(features & BTRFS_FEATURE_INCOMPAT_BIG_METADATA))
2853 2854
			btrfs_info(fs_info,
				"flagging fs with big metadata feature");
2855 2856 2857
		features |= BTRFS_FEATURE_INCOMPAT_BIG_METADATA;
	}

2858 2859
	nodesize = btrfs_super_nodesize(disk_super);
	sectorsize = btrfs_super_sectorsize(disk_super);
2860
	stripesize = sectorsize;
2861
	fs_info->dirty_metadata_batch = nodesize * (1 + ilog2(nr_cpu_ids));
2862
	fs_info->delalloc_batch = sectorsize * 512 * (1 + ilog2(nr_cpu_ids));
2863 2864 2865 2866 2867 2868

	/*
	 * mixed block groups end up with duplicate but slightly offset
	 * extent buffers for the same range.  It leads to corruptions
	 */
	if ((features & BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS) &&
2869
	    (sectorsize != nodesize)) {
2870 2871 2872
		btrfs_err(fs_info,
"unequal nodesize/sectorsize (%u != %u) are not allowed for mixed block groups",
			nodesize, sectorsize);
2873 2874 2875
		goto fail_alloc;
	}

2876 2877 2878 2879
	/*
	 * Needn't use the lock because there is no other task which will
	 * update the flag.
	 */
L
Li Zefan 已提交
2880
	btrfs_set_super_incompat_flags(disk_super, features);
2881

2882 2883 2884
	features = btrfs_super_compat_ro_flags(disk_super) &
		~BTRFS_FEATURE_COMPAT_RO_SUPP;
	if (!(sb->s_flags & MS_RDONLY) && features) {
2885 2886
		btrfs_err(fs_info,
	"cannot mount read-write because of unsupported optional features (%llx)",
2887
		       features);
2888
		err = -EINVAL;
2889
		goto fail_alloc;
2890
	}
2891

2892
	max_active = fs_info->thread_pool_size;
2893

2894 2895 2896
	ret = btrfs_init_workqueues(fs_info, fs_devices);
	if (ret) {
		err = ret;
2897 2898
		goto fail_sb_buffer;
	}
2899

2900
	fs_info->bdi.ra_pages *= btrfs_super_num_devices(disk_super);
C
Chris Mason 已提交
2901
	fs_info->bdi.ra_pages = max(fs_info->bdi.ra_pages,
2902
				    SZ_4M / PAGE_SIZE);
2903

2904 2905
	tree_root->nodesize = nodesize;
	tree_root->sectorsize = sectorsize;
2906
	tree_root->stripesize = stripesize;
2907 2908 2909

	sb->s_blocksize = sectorsize;
	sb->s_blocksize_bits = blksize_bits(sectorsize);
2910

2911
	mutex_lock(&fs_info->chunk_mutex);
2912
	ret = btrfs_read_sys_array(fs_info);
2913
	mutex_unlock(&fs_info->chunk_mutex);
2914
	if (ret) {
2915
		btrfs_err(fs_info, "failed to read the system array: %d", ret);
2916
		goto fail_sb_buffer;
2917
	}
2918

2919
	generation = btrfs_super_chunk_root_generation(disk_super);
2920

2921 2922
	__setup_root(nodesize, sectorsize, stripesize, chunk_root,
		     fs_info, BTRFS_CHUNK_TREE_OBJECTID);
2923 2924 2925

	chunk_root->node = read_tree_block(chunk_root,
					   btrfs_super_chunk_root(disk_super),
2926
					   generation);
2927 2928
	if (IS_ERR(chunk_root->node) ||
	    !extent_buffer_uptodate(chunk_root->node)) {
2929
		btrfs_err(fs_info, "failed to read chunk root");
2930 2931
		if (!IS_ERR(chunk_root->node))
			free_extent_buffer(chunk_root->node);
2932
		chunk_root->node = NULL;
C
Chris Mason 已提交
2933
		goto fail_tree_roots;
2934
	}
2935 2936
	btrfs_set_root_node(&chunk_root->root_item, chunk_root->node);
	chunk_root->commit_root = btrfs_root_node(chunk_root);
2937

2938
	read_extent_buffer(chunk_root->node, fs_info->chunk_tree_uuid,
2939
	   btrfs_header_chunk_tree_uuid(chunk_root->node), BTRFS_UUID_SIZE);
2940

2941
	ret = btrfs_read_chunk_tree(fs_info);
Y
Yan Zheng 已提交
2942
	if (ret) {
2943
		btrfs_err(fs_info, "failed to read chunk tree: %d", ret);
C
Chris Mason 已提交
2944
		goto fail_tree_roots;
Y
Yan Zheng 已提交
2945
	}
2946

2947 2948 2949 2950
	/*
	 * keep the device that is marked to be the target device for the
	 * dev_replace procedure
	 */
2951
	btrfs_close_extra_devices(fs_devices, 0);
2952

2953
	if (!fs_devices->latest_bdev) {
2954
		btrfs_err(fs_info, "failed to read devices");
2955 2956 2957
		goto fail_tree_roots;
	}

C
Chris Mason 已提交
2958
retry_root_backup:
2959
	generation = btrfs_super_generation(disk_super);
2960

C
Chris Mason 已提交
2961
	tree_root->node = read_tree_block(tree_root,
2962
					  btrfs_super_root(disk_super),
2963
					  generation);
2964 2965
	if (IS_ERR(tree_root->node) ||
	    !extent_buffer_uptodate(tree_root->node)) {
2966
		btrfs_warn(fs_info, "failed to read tree root");
2967 2968
		if (!IS_ERR(tree_root->node))
			free_extent_buffer(tree_root->node);
2969
		tree_root->node = NULL;
C
Chris Mason 已提交
2970
		goto recovery_tree_root;
2971
	}
C
Chris Mason 已提交
2972

2973 2974
	btrfs_set_root_node(&tree_root->root_item, tree_root->node);
	tree_root->commit_root = btrfs_root_node(tree_root);
2975
	btrfs_set_root_refs(&tree_root->root_item, 1);
2976

2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988
	mutex_lock(&tree_root->objectid_mutex);
	ret = btrfs_find_highest_objectid(tree_root,
					&tree_root->highest_objectid);
	if (ret) {
		mutex_unlock(&tree_root->objectid_mutex);
		goto recovery_tree_root;
	}

	ASSERT(tree_root->highest_objectid <= BTRFS_LAST_FREE_OBJECTID);

	mutex_unlock(&tree_root->objectid_mutex);

2989
	ret = btrfs_read_roots(fs_info);
2990
	if (ret)
C
Chris Mason 已提交
2991
		goto recovery_tree_root;
2992

2993 2994 2995
	fs_info->generation = generation;
	fs_info->last_trans_committed = generation;

2996 2997
	ret = btrfs_recover_balance(fs_info);
	if (ret) {
2998
		btrfs_err(fs_info, "failed to recover balance: %d", ret);
2999 3000 3001
		goto fail_block_groups;
	}

3002 3003
	ret = btrfs_init_dev_stats(fs_info);
	if (ret) {
3004
		btrfs_err(fs_info, "failed to init dev_stats: %d", ret);
3005 3006 3007
		goto fail_block_groups;
	}

3008 3009
	ret = btrfs_init_dev_replace(fs_info);
	if (ret) {
3010
		btrfs_err(fs_info, "failed to init dev_replace: %d", ret);
3011 3012 3013
		goto fail_block_groups;
	}

3014
	btrfs_close_extra_devices(fs_devices, 1);
3015

3016 3017
	ret = btrfs_sysfs_add_fsid(fs_devices, NULL);
	if (ret) {
3018 3019
		btrfs_err(fs_info, "failed to init sysfs fsid interface: %d",
				ret);
3020 3021 3022 3023 3024
		goto fail_block_groups;
	}

	ret = btrfs_sysfs_add_device(fs_devices);
	if (ret) {
3025 3026
		btrfs_err(fs_info, "failed to init sysfs device interface: %d",
				ret);
3027 3028 3029
		goto fail_fsdev_sysfs;
	}

3030
	ret = btrfs_sysfs_add_mounted(fs_info);
3031
	if (ret) {
3032
		btrfs_err(fs_info, "failed to init sysfs interface: %d", ret);
3033
		goto fail_fsdev_sysfs;
3034 3035 3036 3037
	}

	ret = btrfs_init_space_info(fs_info);
	if (ret) {
3038
		btrfs_err(fs_info, "failed to initialize space info: %d", ret);
3039
		goto fail_sysfs;
3040 3041
	}

3042
	ret = btrfs_read_block_groups(fs_info);
3043
	if (ret) {
3044
		btrfs_err(fs_info, "failed to read block groups: %d", ret);
3045
		goto fail_sysfs;
3046
	}
3047 3048
	fs_info->num_tolerated_disk_barrier_failures =
		btrfs_calc_num_tolerated_disk_barrier_failures(fs_info);
3049 3050 3051
	if (fs_info->fs_devices->missing_devices >
	     fs_info->num_tolerated_disk_barrier_failures &&
	    !(sb->s_flags & MS_RDONLY)) {
3052 3053
		btrfs_warn(fs_info,
"missing devices (%llu) exceeds the limit (%d), writeable mount is not allowed",
3054 3055
			fs_info->fs_devices->missing_devices,
			fs_info->num_tolerated_disk_barrier_failures);
3056
		goto fail_sysfs;
3057
	}
C
Chris Mason 已提交
3058

3059 3060
	fs_info->cleaner_kthread = kthread_run(cleaner_kthread, tree_root,
					       "btrfs-cleaner");
3061
	if (IS_ERR(fs_info->cleaner_kthread))
3062
		goto fail_sysfs;
3063 3064 3065 3066

	fs_info->transaction_kthread = kthread_run(transaction_kthread,
						   tree_root,
						   "btrfs-transaction");
3067
	if (IS_ERR(fs_info->transaction_kthread))
3068
		goto fail_cleaner;
3069

3070 3071
	if (!btrfs_test_opt(tree_root->fs_info, SSD) &&
	    !btrfs_test_opt(tree_root->fs_info, NOSSD) &&
C
Chris Mason 已提交
3072
	    !fs_info->fs_devices->rotating) {
3073
		btrfs_info(fs_info, "detected SSD devices, enabling SSD mode");
C
Chris Mason 已提交
3074 3075 3076
		btrfs_set_opt(fs_info->mount_opt, SSD);
	}

3077
	/*
3078
	 * Mount does not set all options immediately, we can do it now and do
3079 3080 3081
	 * not have to wait for transaction commit
	 */
	btrfs_apply_pending_changes(fs_info);
3082

3083
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
3084
	if (btrfs_test_opt(tree_root->fs_info, CHECK_INTEGRITY)) {
3085
		ret = btrfsic_mount(tree_root, fs_devices,
3086
				    btrfs_test_opt(tree_root->fs_info,
3087 3088 3089 3090
					CHECK_INTEGRITY_INCLUDING_EXTENT_DATA) ?
				    1 : 0,
				    fs_info->check_integrity_print_mask);
		if (ret)
3091 3092 3093
			btrfs_warn(fs_info,
				"failed to initialize integrity check module: %d",
				ret);
3094 3095
	}
#endif
3096 3097 3098
	ret = btrfs_read_qgroup_config(fs_info);
	if (ret)
		goto fail_trans_kthread;
3099

3100 3101
	/* do not make disk changes in broken FS or nologreplay is given */
	if (btrfs_super_log_root(disk_super) != 0 &&
3102
	    !btrfs_test_opt(tree_root->fs_info, NOLOGREPLAY)) {
3103
		ret = btrfs_replay_log(fs_info, fs_devices);
3104
		if (ret) {
3105
			err = ret;
3106
			goto fail_qgroup;
3107
		}
3108
	}
Z
Zheng Yan 已提交
3109

3110
	ret = btrfs_find_orphan_roots(fs_info);
3111
	if (ret)
3112
		goto fail_qgroup;
3113

3114
	if (!(sb->s_flags & MS_RDONLY)) {
3115
		ret = btrfs_cleanup_fs_roots(fs_info);
3116
		if (ret)
3117
			goto fail_qgroup;
3118 3119

		mutex_lock(&fs_info->cleaner_mutex);
3120
		ret = btrfs_recover_relocation(tree_root);
3121
		mutex_unlock(&fs_info->cleaner_mutex);
3122
		if (ret < 0) {
3123 3124
			btrfs_warn(fs_info, "failed to recover relocation: %d",
					ret);
3125
			err = -EINVAL;
3126
			goto fail_qgroup;
3127
		}
3128
	}
Z
Zheng Yan 已提交
3129

3130 3131
	location.objectid = BTRFS_FS_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
3132
	location.offset = 0;
3133 3134

	fs_info->fs_root = btrfs_read_fs_root_no_name(fs_info, &location);
3135 3136
	if (IS_ERR(fs_info->fs_root)) {
		err = PTR_ERR(fs_info->fs_root);
3137
		goto fail_qgroup;
3138
	}
C
Chris Mason 已提交
3139

3140 3141
	if (sb->s_flags & MS_RDONLY)
		return 0;
I
Ilya Dryomov 已提交
3142

3143 3144
	if (btrfs_test_opt(fs_info, CLEAR_CACHE) &&
	    btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
3145 3146 3147 3148 3149 3150 3151 3152
		clear_free_space_tree = 1;
	} else if (btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE) &&
		   !btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE_VALID)) {
		btrfs_warn(fs_info, "free space tree is invalid");
		clear_free_space_tree = 1;
	}

	if (clear_free_space_tree) {
3153 3154 3155 3156 3157
		btrfs_info(fs_info, "clearing free space tree");
		ret = btrfs_clear_free_space_tree(fs_info);
		if (ret) {
			btrfs_warn(fs_info,
				   "failed to clear free space tree: %d", ret);
3158
			close_ctree(fs_info);
3159 3160 3161 3162
			return ret;
		}
	}

3163
	if (btrfs_test_opt(tree_root->fs_info, FREE_SPACE_TREE) &&
3164
	    !btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
3165
		btrfs_info(fs_info, "creating free space tree");
3166 3167
		ret = btrfs_create_free_space_tree(fs_info);
		if (ret) {
3168 3169
			btrfs_warn(fs_info,
				"failed to create free space tree: %d", ret);
3170
			close_ctree(fs_info);
3171 3172 3173 3174
			return ret;
		}
	}

3175 3176 3177
	down_read(&fs_info->cleanup_work_sem);
	if ((ret = btrfs_orphan_cleanup(fs_info->fs_root)) ||
	    (ret = btrfs_orphan_cleanup(fs_info->tree_root))) {
3178
		up_read(&fs_info->cleanup_work_sem);
3179
		close_ctree(fs_info);
3180 3181 3182
		return ret;
	}
	up_read(&fs_info->cleanup_work_sem);
I
Ilya Dryomov 已提交
3183

3184 3185
	ret = btrfs_resume_balance_async(fs_info);
	if (ret) {
3186
		btrfs_warn(fs_info, "failed to resume balance: %d", ret);
3187
		close_ctree(fs_info);
3188
		return ret;
3189 3190
	}

3191 3192
	ret = btrfs_resume_dev_replace_async(fs_info);
	if (ret) {
3193
		btrfs_warn(fs_info, "failed to resume device replace: %d", ret);
3194
		close_ctree(fs_info);
3195 3196 3197
		return ret;
	}

3198 3199
	btrfs_qgroup_rescan_resume(fs_info);

3200
	if (!fs_info->uuid_root) {
3201
		btrfs_info(fs_info, "creating UUID tree");
3202 3203
		ret = btrfs_create_uuid_tree(fs_info);
		if (ret) {
3204 3205
			btrfs_warn(fs_info,
				"failed to create the UUID tree: %d", ret);
3206
			close_ctree(fs_info);
3207 3208
			return ret;
		}
3209
	} else if (btrfs_test_opt(tree_root->fs_info, RESCAN_UUID_TREE) ||
3210 3211
		   fs_info->generation !=
				btrfs_super_uuid_tree_generation(disk_super)) {
3212
		btrfs_info(fs_info, "checking UUID tree");
3213 3214
		ret = btrfs_check_uuid_tree(fs_info);
		if (ret) {
3215 3216
			btrfs_warn(fs_info,
				"failed to check the UUID tree: %d", ret);
3217
			close_ctree(fs_info);
3218 3219 3220
			return ret;
		}
	} else {
3221
		set_bit(BTRFS_FS_UPDATE_UUID_TREE_GEN, &fs_info->flags);
3222
	}
3223
	set_bit(BTRFS_FS_OPEN, &fs_info->flags);
3224

3225 3226 3227 3228 3229 3230
	/*
	 * backuproot only affect mount behavior, and if open_ctree succeeded,
	 * no need to keep the flag
	 */
	btrfs_clear_opt(fs_info->mount_opt, USEBACKUPROOT);

A
Al Viro 已提交
3231
	return 0;
C
Chris Mason 已提交
3232

3233 3234
fail_qgroup:
	btrfs_free_qgroup_config(fs_info);
3235 3236
fail_trans_kthread:
	kthread_stop(fs_info->transaction_kthread);
J
Josef Bacik 已提交
3237
	btrfs_cleanup_transaction(fs_info->tree_root);
3238
	btrfs_free_fs_roots(fs_info);
3239
fail_cleaner:
3240
	kthread_stop(fs_info->cleaner_kthread);
3241 3242 3243 3244 3245 3246 3247

	/*
	 * make sure we're done with the btree inode before we stop our
	 * kthreads
	 */
	filemap_write_and_wait(fs_info->btree_inode->i_mapping);

3248
fail_sysfs:
3249
	btrfs_sysfs_remove_mounted(fs_info);
3250

3251 3252 3253
fail_fsdev_sysfs:
	btrfs_sysfs_remove_fsid(fs_info->fs_devices);

3254
fail_block_groups:
J
Josef Bacik 已提交
3255
	btrfs_put_block_group_cache(fs_info);
3256
	btrfs_free_block_groups(fs_info);
C
Chris Mason 已提交
3257 3258 3259

fail_tree_roots:
	free_root_pointers(fs_info, 1);
3260
	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
3261

C
Chris Mason 已提交
3262
fail_sb_buffer:
L
Liu Bo 已提交
3263
	btrfs_stop_all_workers(fs_info);
3264
fail_alloc:
3265
fail_iput:
3266 3267
	btrfs_mapping_tree_free(&fs_info->mapping_tree);

3268
	iput(fs_info->btree_inode);
3269 3270
fail_bio_counter:
	percpu_counter_destroy(&fs_info->bio_counter);
3271 3272
fail_delalloc_bytes:
	percpu_counter_destroy(&fs_info->delalloc_bytes);
3273 3274
fail_dirty_metadata_bytes:
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
3275
fail_bdi:
3276
	bdi_destroy(&fs_info->bdi);
3277 3278
fail_srcu:
	cleanup_srcu_struct(&fs_info->subvol_srcu);
3279
fail:
D
David Woodhouse 已提交
3280
	btrfs_free_stripe_hash_table(fs_info);
3281
	btrfs_close_devices(fs_info->fs_devices);
A
Al Viro 已提交
3282
	return err;
C
Chris Mason 已提交
3283 3284

recovery_tree_root:
3285
	if (!btrfs_test_opt(tree_root->fs_info, USEBACKUPROOT))
C
Chris Mason 已提交
3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300
		goto fail_tree_roots;

	free_root_pointers(fs_info, 0);

	/* don't use the log in recovery mode, it won't be valid */
	btrfs_set_super_log_root(disk_super, 0);

	/* we can't trust the free space cache either */
	btrfs_set_opt(fs_info->mount_opt, CLEAR_CACHE);

	ret = next_root_backup(fs_info, fs_info->super_copy,
			       &num_backups_tried, &backup_index);
	if (ret == -1)
		goto fail_block_groups;
	goto retry_root_backup;
3301 3302
}

3303 3304 3305 3306 3307
static void btrfs_end_buffer_write_sync(struct buffer_head *bh, int uptodate)
{
	if (uptodate) {
		set_buffer_uptodate(bh);
	} else {
3308 3309 3310
		struct btrfs_device *device = (struct btrfs_device *)
			bh->b_private;

3311
		btrfs_warn_rl_in_rcu(device->fs_info,
3312
				"lost page write due to IO error on %s",
3313
					  rcu_str_deref(device->name));
3314
		/* note, we don't set_buffer_write_io_error because we have
3315 3316
		 * our own ways of dealing with the IO errors
		 */
3317
		clear_buffer_uptodate(bh);
3318
		btrfs_dev_stat_inc_and_print(device, BTRFS_DEV_STAT_WRITE_ERRS);
3319 3320 3321 3322 3323
	}
	unlock_buffer(bh);
	put_bh(bh);
}

3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354
int btrfs_read_dev_one_super(struct block_device *bdev, int copy_num,
			struct buffer_head **bh_ret)
{
	struct buffer_head *bh;
	struct btrfs_super_block *super;
	u64 bytenr;

	bytenr = btrfs_sb_offset(copy_num);
	if (bytenr + BTRFS_SUPER_INFO_SIZE >= i_size_read(bdev->bd_inode))
		return -EINVAL;

	bh = __bread(bdev, bytenr / 4096, BTRFS_SUPER_INFO_SIZE);
	/*
	 * If we fail to read from the underlying devices, as of now
	 * the best option we have is to mark it EIO.
	 */
	if (!bh)
		return -EIO;

	super = (struct btrfs_super_block *)bh->b_data;
	if (btrfs_super_bytenr(super) != bytenr ||
		    btrfs_super_magic(super) != BTRFS_MAGIC) {
		brelse(bh);
		return -EINVAL;
	}

	*bh_ret = bh;
	return 0;
}


Y
Yan Zheng 已提交
3355 3356 3357 3358 3359 3360 3361
struct buffer_head *btrfs_read_dev_super(struct block_device *bdev)
{
	struct buffer_head *bh;
	struct buffer_head *latest = NULL;
	struct btrfs_super_block *super;
	int i;
	u64 transid = 0;
3362
	int ret = -EINVAL;
Y
Yan Zheng 已提交
3363 3364 3365 3366 3367 3368 3369

	/* we would like to check all the supers, but that would make
	 * a btrfs mount succeed after a mkfs from a different FS.
	 * So, we need to add a special mount option to scan for
	 * later supers, using BTRFS_SUPER_MIRROR_MAX instead
	 */
	for (i = 0; i < 1; i++) {
3370 3371
		ret = btrfs_read_dev_one_super(bdev, i, &bh);
		if (ret)
Y
Yan Zheng 已提交
3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383
			continue;

		super = (struct btrfs_super_block *)bh->b_data;

		if (!latest || btrfs_super_generation(super) > transid) {
			brelse(latest);
			latest = bh;
			transid = btrfs_super_generation(super);
		} else {
			brelse(bh);
		}
	}
3384 3385 3386 3387

	if (!latest)
		return ERR_PTR(ret);

Y
Yan Zheng 已提交
3388 3389 3390
	return latest;
}

3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401
/*
 * this should be called twice, once with wait == 0 and
 * once with wait == 1.  When wait == 0 is done, all the buffer heads
 * we write are pinned.
 *
 * They are released when wait == 1 is done.
 * max_mirrors must be the same for both runs, and it indicates how
 * many supers on this one device should be written.
 *
 * max_mirrors == 0 means to write them all.
 */
Y
Yan Zheng 已提交
3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417
static int write_dev_supers(struct btrfs_device *device,
			    struct btrfs_super_block *sb,
			    int do_barriers, int wait, int max_mirrors)
{
	struct buffer_head *bh;
	int i;
	int ret;
	int errors = 0;
	u32 crc;
	u64 bytenr;

	if (max_mirrors == 0)
		max_mirrors = BTRFS_SUPER_MIRROR_MAX;

	for (i = 0; i < max_mirrors; i++) {
		bytenr = btrfs_sb_offset(i);
3418 3419
		if (bytenr + BTRFS_SUPER_INFO_SIZE >=
		    device->commit_total_bytes)
Y
Yan Zheng 已提交
3420 3421 3422 3423 3424
			break;

		if (wait) {
			bh = __find_get_block(device->bdev, bytenr / 4096,
					      BTRFS_SUPER_INFO_SIZE);
3425 3426 3427 3428
			if (!bh) {
				errors++;
				continue;
			}
Y
Yan Zheng 已提交
3429
			wait_on_buffer(bh);
3430 3431 3432 3433 3434 3435 3436 3437 3438
			if (!buffer_uptodate(bh))
				errors++;

			/* drop our reference */
			brelse(bh);

			/* drop the reference from the wait == 0 run */
			brelse(bh);
			continue;
Y
Yan Zheng 已提交
3439 3440 3441 3442
		} else {
			btrfs_set_super_bytenr(sb, bytenr);

			crc = ~(u32)0;
3443
			crc = btrfs_csum_data((char *)sb +
Y
Yan Zheng 已提交
3444 3445 3446 3447 3448
					      BTRFS_CSUM_SIZE, crc,
					      BTRFS_SUPER_INFO_SIZE -
					      BTRFS_CSUM_SIZE);
			btrfs_csum_final(crc, sb->csum);

3449 3450 3451 3452
			/*
			 * one reference for us, and we leave it for the
			 * caller
			 */
Y
Yan Zheng 已提交
3453 3454
			bh = __getblk(device->bdev, bytenr / 4096,
				      BTRFS_SUPER_INFO_SIZE);
3455
			if (!bh) {
3456
				btrfs_err(device->fs_info,
3457 3458
				    "couldn't get super buffer head for bytenr %llu",
				    bytenr);
3459 3460 3461 3462
				errors++;
				continue;
			}

Y
Yan Zheng 已提交
3463 3464
			memcpy(bh->b_data, sb, BTRFS_SUPER_INFO_SIZE);

3465
			/* one reference for submit_bh */
Y
Yan Zheng 已提交
3466
			get_bh(bh);
3467 3468

			set_buffer_uptodate(bh);
Y
Yan Zheng 已提交
3469 3470
			lock_buffer(bh);
			bh->b_end_io = btrfs_end_buffer_write_sync;
3471
			bh->b_private = device;
Y
Yan Zheng 已提交
3472 3473
		}

C
Chris Mason 已提交
3474 3475 3476 3477
		/*
		 * we fua the first super.  The others we allow
		 * to go down lazy.
		 */
3478
		if (i == 0)
3479
			ret = btrfsic_submit_bh(REQ_OP_WRITE, WRITE_FUA, bh);
3480
		else
3481
			ret = btrfsic_submit_bh(REQ_OP_WRITE, WRITE_SYNC, bh);
3482
		if (ret)
Y
Yan Zheng 已提交
3483 3484 3485 3486 3487
			errors++;
	}
	return errors < i ? 0 : -1;
}

C
Chris Mason 已提交
3488 3489 3490 3491
/*
 * endio for the write_dev_flush, this will wake anyone waiting
 * for the barrier when it is done
 */
3492
static void btrfs_end_empty_barrier(struct bio *bio)
C
Chris Mason 已提交
3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520
{
	if (bio->bi_private)
		complete(bio->bi_private);
	bio_put(bio);
}

/*
 * trigger flushes for one the devices.  If you pass wait == 0, the flushes are
 * sent down.  With wait == 1, it waits for the previous flush.
 *
 * any device where the flush fails with eopnotsupp are flagged as not-barrier
 * capable
 */
static int write_dev_flush(struct btrfs_device *device, int wait)
{
	struct bio *bio;
	int ret = 0;

	if (device->nobarriers)
		return 0;

	if (wait) {
		bio = device->flush_bio;
		if (!bio)
			return 0;

		wait_for_completion(&device->flush_wait);

3521 3522
		if (bio->bi_error) {
			ret = bio->bi_error;
3523 3524
			btrfs_dev_stat_inc_and_print(device,
				BTRFS_DEV_STAT_FLUSH_ERRS);
C
Chris Mason 已提交
3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537
		}

		/* drop the reference from the wait == 0 run */
		bio_put(bio);
		device->flush_bio = NULL;

		return ret;
	}

	/*
	 * one reference for us, and we leave it for the
	 * caller
	 */
3538
	device->flush_bio = NULL;
3539
	bio = btrfs_io_bio_alloc(GFP_NOFS, 0);
C
Chris Mason 已提交
3540 3541 3542 3543 3544
	if (!bio)
		return -ENOMEM;

	bio->bi_end_io = btrfs_end_empty_barrier;
	bio->bi_bdev = device->bdev;
M
Mike Christie 已提交
3545
	bio_set_op_attrs(bio, REQ_OP_WRITE, WRITE_FLUSH);
C
Chris Mason 已提交
3546 3547 3548 3549 3550
	init_completion(&device->flush_wait);
	bio->bi_private = &device->flush_wait;
	device->flush_bio = bio;

	bio_get(bio);
3551
	btrfsic_submit_bio(bio);
C
Chris Mason 已提交
3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563

	return 0;
}

/*
 * send an empty flush down to each device in parallel,
 * then wait for them
 */
static int barrier_all_devices(struct btrfs_fs_info *info)
{
	struct list_head *head;
	struct btrfs_device *dev;
3564 3565
	int errors_send = 0;
	int errors_wait = 0;
C
Chris Mason 已提交
3566 3567 3568 3569 3570
	int ret;

	/* send down all the barriers */
	head = &info->fs_devices->devices;
	list_for_each_entry_rcu(dev, head, dev_list) {
3571 3572
		if (dev->missing)
			continue;
C
Chris Mason 已提交
3573
		if (!dev->bdev) {
3574
			errors_send++;
C
Chris Mason 已提交
3575 3576 3577 3578 3579 3580 3581
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

		ret = write_dev_flush(dev, 0);
		if (ret)
3582
			errors_send++;
C
Chris Mason 已提交
3583 3584 3585 3586
	}

	/* wait for all the barriers */
	list_for_each_entry_rcu(dev, head, dev_list) {
3587 3588
		if (dev->missing)
			continue;
C
Chris Mason 已提交
3589
		if (!dev->bdev) {
3590
			errors_wait++;
C
Chris Mason 已提交
3591 3592 3593 3594 3595 3596 3597
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

		ret = write_dev_flush(dev, 1);
		if (ret)
3598
			errors_wait++;
C
Chris Mason 已提交
3599
	}
3600 3601
	if (errors_send > info->num_tolerated_disk_barrier_failures ||
	    errors_wait > info->num_tolerated_disk_barrier_failures)
C
Chris Mason 已提交
3602 3603 3604 3605
		return -EIO;
	return 0;
}

3606 3607
int btrfs_get_num_tolerated_disk_barrier_failures(u64 flags)
{
3608 3609
	int raid_type;
	int min_tolerated = INT_MAX;
3610

3611 3612 3613 3614 3615
	if ((flags & BTRFS_BLOCK_GROUP_PROFILE_MASK) == 0 ||
	    (flags & BTRFS_AVAIL_ALLOC_BIT_SINGLE))
		min_tolerated = min(min_tolerated,
				    btrfs_raid_array[BTRFS_RAID_SINGLE].
				    tolerated_failures);
3616

3617 3618 3619 3620 3621 3622 3623 3624 3625
	for (raid_type = 0; raid_type < BTRFS_NR_RAID_TYPES; raid_type++) {
		if (raid_type == BTRFS_RAID_SINGLE)
			continue;
		if (!(flags & btrfs_raid_group[raid_type]))
			continue;
		min_tolerated = min(min_tolerated,
				    btrfs_raid_array[raid_type].
				    tolerated_failures);
	}
3626

3627
	if (min_tolerated == INT_MAX) {
3628
		pr_warn("BTRFS: unknown raid flag: %llu", flags);
3629 3630 3631 3632
		min_tolerated = 0;
	}

	return min_tolerated;
3633 3634
}

3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648
int btrfs_calc_num_tolerated_disk_barrier_failures(
	struct btrfs_fs_info *fs_info)
{
	struct btrfs_ioctl_space_info space;
	struct btrfs_space_info *sinfo;
	u64 types[] = {BTRFS_BLOCK_GROUP_DATA,
		       BTRFS_BLOCK_GROUP_SYSTEM,
		       BTRFS_BLOCK_GROUP_METADATA,
		       BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA};
	int i;
	int c;
	int num_tolerated_disk_barrier_failures =
		(int)fs_info->fs_devices->num_devices;

3649
	for (i = 0; i < ARRAY_SIZE(types); i++) {
3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666
		struct btrfs_space_info *tmp;

		sinfo = NULL;
		rcu_read_lock();
		list_for_each_entry_rcu(tmp, &fs_info->space_info, list) {
			if (tmp->flags == types[i]) {
				sinfo = tmp;
				break;
			}
		}
		rcu_read_unlock();

		if (!sinfo)
			continue;

		down_read(&sinfo->groups_sem);
		for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
3667 3668 3669 3670 3671 3672 3673 3674 3675 3676
			u64 flags;

			if (list_empty(&sinfo->block_groups[c]))
				continue;

			btrfs_get_block_group_info(&sinfo->block_groups[c],
						   &space);
			if (space.total_bytes == 0 || space.used_bytes == 0)
				continue;
			flags = space.flags;
3677 3678 3679 3680 3681

			num_tolerated_disk_barrier_failures = min(
				num_tolerated_disk_barrier_failures,
				btrfs_get_num_tolerated_disk_barrier_failures(
					flags));
3682 3683 3684 3685 3686 3687 3688
		}
		up_read(&sinfo->groups_sem);
	}

	return num_tolerated_disk_barrier_failures;
}

3689
static int write_all_supers(struct btrfs_root *root, int max_mirrors)
3690
{
3691
	struct list_head *head;
3692
	struct btrfs_device *dev;
3693
	struct btrfs_super_block *sb;
3694 3695 3696
	struct btrfs_dev_item *dev_item;
	int ret;
	int do_barriers;
3697 3698
	int max_errors;
	int total_errors = 0;
3699
	u64 flags;
3700

3701
	do_barriers = !btrfs_test_opt(root->fs_info, NOBARRIER);
C
Chris Mason 已提交
3702
	backup_super_roots(root->fs_info);
3703

3704
	sb = root->fs_info->super_for_commit;
3705
	dev_item = &sb->dev_item;
3706

3707
	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
3708
	head = &root->fs_info->fs_devices->devices;
3709
	max_errors = btrfs_super_num_devices(root->fs_info->super_copy) - 1;
C
Chris Mason 已提交
3710

3711 3712 3713 3714 3715
	if (do_barriers) {
		ret = barrier_all_devices(root->fs_info);
		if (ret) {
			mutex_unlock(
				&root->fs_info->fs_devices->device_list_mutex);
3716
			btrfs_handle_fs_error(root->fs_info, ret,
3717 3718 3719 3720
				    "errors while submitting device barriers.");
			return ret;
		}
	}
C
Chris Mason 已提交
3721

3722
	list_for_each_entry_rcu(dev, head, dev_list) {
3723 3724 3725 3726
		if (!dev->bdev) {
			total_errors++;
			continue;
		}
Y
Yan Zheng 已提交
3727
		if (!dev->in_fs_metadata || !dev->writeable)
3728 3729
			continue;

Y
Yan Zheng 已提交
3730
		btrfs_set_stack_device_generation(dev_item, 0);
3731 3732
		btrfs_set_stack_device_type(dev_item, dev->type);
		btrfs_set_stack_device_id(dev_item, dev->devid);
3733
		btrfs_set_stack_device_total_bytes(dev_item,
3734
						   dev->commit_total_bytes);
3735 3736
		btrfs_set_stack_device_bytes_used(dev_item,
						  dev->commit_bytes_used);
3737 3738 3739 3740
		btrfs_set_stack_device_io_align(dev_item, dev->io_align);
		btrfs_set_stack_device_io_width(dev_item, dev->io_width);
		btrfs_set_stack_device_sector_size(dev_item, dev->sector_size);
		memcpy(dev_item->uuid, dev->uuid, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3741
		memcpy(dev_item->fsid, dev->fs_devices->fsid, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3742

3743 3744 3745
		flags = btrfs_super_flags(sb);
		btrfs_set_super_flags(sb, flags | BTRFS_HEADER_FLAG_WRITTEN);

Y
Yan Zheng 已提交
3746
		ret = write_dev_supers(dev, sb, do_barriers, 0, max_mirrors);
3747 3748
		if (ret)
			total_errors++;
3749
	}
3750
	if (total_errors > max_errors) {
3751
		btrfs_err(root->fs_info, "%d errors while writing supers",
C
Chris Mason 已提交
3752
		       total_errors);
3753
		mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
3754

3755
		/* FUA is masked off if unsupported and can't be the reason */
3756
		btrfs_handle_fs_error(root->fs_info, -EIO,
3757 3758
			    "%d errors while writing supers", total_errors);
		return -EIO;
3759
	}
3760

Y
Yan Zheng 已提交
3761
	total_errors = 0;
3762
	list_for_each_entry_rcu(dev, head, dev_list) {
3763 3764
		if (!dev->bdev)
			continue;
Y
Yan Zheng 已提交
3765
		if (!dev->in_fs_metadata || !dev->writeable)
3766 3767
			continue;

Y
Yan Zheng 已提交
3768 3769 3770
		ret = write_dev_supers(dev, sb, do_barriers, 1, max_mirrors);
		if (ret)
			total_errors++;
3771
	}
3772
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
3773
	if (total_errors > max_errors) {
3774
		btrfs_handle_fs_error(root->fs_info, -EIO,
3775 3776
			    "%d errors while writing supers", total_errors);
		return -EIO;
3777
	}
3778 3779 3780
	return 0;
}

Y
Yan Zheng 已提交
3781 3782
int write_ctree_super(struct btrfs_trans_handle *trans,
		      struct btrfs_root *root, int max_mirrors)
3783
{
3784
	return write_all_supers(root, max_mirrors);
C
Chris Mason 已提交
3785 3786
}

3787 3788 3789
/* Drop a fs root from the radix tree and free it. */
void btrfs_drop_and_free_fs_root(struct btrfs_fs_info *fs_info,
				  struct btrfs_root *root)
C
Chris Mason 已提交
3790
{
3791
	spin_lock(&fs_info->fs_roots_radix_lock);
C
Chris Mason 已提交
3792 3793
	radix_tree_delete(&fs_info->fs_roots_radix,
			  (unsigned long)root->root_key.objectid);
3794
	spin_unlock(&fs_info->fs_roots_radix_lock);
3795 3796 3797 3798

	if (btrfs_root_refs(&root->root_item) == 0)
		synchronize_srcu(&fs_info->subvol_srcu);

L
Liu Bo 已提交
3799
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
L
Liu Bo 已提交
3800
		btrfs_free_log(NULL, root);
L
Liu Bo 已提交
3801 3802 3803 3804 3805 3806 3807
		if (root->reloc_root) {
			free_extent_buffer(root->reloc_root->node);
			free_extent_buffer(root->reloc_root->commit_root);
			btrfs_put_fs_root(root->reloc_root);
			root->reloc_root = NULL;
		}
	}
L
Liu Bo 已提交
3808

3809 3810 3811 3812
	if (root->free_ino_pinned)
		__btrfs_remove_free_space_cache(root->free_ino_pinned);
	if (root->free_ino_ctl)
		__btrfs_remove_free_space_cache(root->free_ino_ctl);
3813 3814 3815 3816 3817
	free_fs_root(root);
}

static void free_fs_root(struct btrfs_root *root)
{
3818
	iput(root->ino_cache_inode);
3819
	WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree));
3820 3821
	btrfs_free_block_rsv(root, root->orphan_block_rsv);
	root->orphan_block_rsv = NULL;
3822 3823
	if (root->anon_dev)
		free_anon_bdev(root->anon_dev);
3824 3825
	if (root->subv_writers)
		btrfs_free_subvolume_writers(root->subv_writers);
3826 3827
	free_extent_buffer(root->node);
	free_extent_buffer(root->commit_root);
3828 3829
	kfree(root->free_ino_ctl);
	kfree(root->free_ino_pinned);
C
Chris Mason 已提交
3830
	kfree(root->name);
3831
	btrfs_put_fs_root(root);
C
Chris Mason 已提交
3832 3833
}

3834 3835 3836
void btrfs_free_fs_root(struct btrfs_root *root)
{
	free_fs_root(root);
C
Chris Mason 已提交
3837 3838
}

Y
Yan Zheng 已提交
3839
int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
C
Chris Mason 已提交
3840
{
Y
Yan Zheng 已提交
3841 3842
	u64 root_objectid = 0;
	struct btrfs_root *gang[8];
3843 3844 3845 3846
	int i = 0;
	int err = 0;
	unsigned int ret = 0;
	int index;
3847

Y
Yan Zheng 已提交
3848
	while (1) {
3849
		index = srcu_read_lock(&fs_info->subvol_srcu);
Y
Yan Zheng 已提交
3850 3851 3852
		ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix,
					     (void **)gang, root_objectid,
					     ARRAY_SIZE(gang));
3853 3854
		if (!ret) {
			srcu_read_unlock(&fs_info->subvol_srcu, index);
Y
Yan Zheng 已提交
3855
			break;
3856
		}
3857
		root_objectid = gang[ret - 1]->root_key.objectid + 1;
3858

Y
Yan Zheng 已提交
3859
		for (i = 0; i < ret; i++) {
3860 3861 3862 3863 3864 3865 3866 3867 3868
			/* Avoid to grab roots in dead_roots */
			if (btrfs_root_refs(&gang[i]->root_item) == 0) {
				gang[i] = NULL;
				continue;
			}
			/* grab all the search result for later use */
			gang[i] = btrfs_grab_fs_root(gang[i]);
		}
		srcu_read_unlock(&fs_info->subvol_srcu, index);
3869

3870 3871 3872
		for (i = 0; i < ret; i++) {
			if (!gang[i])
				continue;
Y
Yan Zheng 已提交
3873
			root_objectid = gang[i]->root_key.objectid;
3874 3875
			err = btrfs_orphan_cleanup(gang[i]);
			if (err)
3876 3877
				break;
			btrfs_put_fs_root(gang[i]);
Y
Yan Zheng 已提交
3878 3879 3880
		}
		root_objectid++;
	}
3881 3882 3883 3884 3885 3886 3887

	/* release the uncleaned roots due to error */
	for (; i < ret; i++) {
		if (gang[i])
			btrfs_put_fs_root(gang[i]);
	}
	return err;
Y
Yan Zheng 已提交
3888
}
3889

3890
int btrfs_commit_super(struct btrfs_fs_info *fs_info)
Y
Yan Zheng 已提交
3891
{
3892
	struct btrfs_root *root = fs_info->tree_root;
Y
Yan Zheng 已提交
3893
	struct btrfs_trans_handle *trans;
3894

Y
Yan Zheng 已提交
3895
	mutex_lock(&root->fs_info->cleaner_mutex);
Y
Yan, Zheng 已提交
3896
	btrfs_run_delayed_iputs(root);
Y
Yan Zheng 已提交
3897
	mutex_unlock(&root->fs_info->cleaner_mutex);
D
David Sterba 已提交
3898
	wake_up_process(root->fs_info->cleaner_kthread);
3899 3900 3901 3902 3903

	/* wait until ongoing cleanup work done */
	down_write(&root->fs_info->cleanup_work_sem);
	up_write(&root->fs_info->cleanup_work_sem);

3904
	trans = btrfs_join_transaction(root);
3905 3906
	if (IS_ERR(trans))
		return PTR_ERR(trans);
3907
	return btrfs_commit_transaction(trans, root);
Y
Yan Zheng 已提交
3908 3909
}

3910
void close_ctree(struct btrfs_fs_info *fs_info)
Y
Yan Zheng 已提交
3911
{
3912
	struct btrfs_root *root = fs_info->tree_root;
Y
Yan Zheng 已提交
3913 3914
	int ret;

3915
	set_bit(BTRFS_FS_CLOSING_START, &fs_info->flags);
Y
Yan Zheng 已提交
3916

3917
	/* wait for the qgroup rescan worker to stop */
3918
	btrfs_qgroup_wait_for_completion(fs_info, false);
3919

S
Stefan Behrens 已提交
3920 3921 3922 3923 3924
	/* wait for the uuid_scan task to finish */
	down(&fs_info->uuid_tree_rescan_sem);
	/* avoid complains from lockdep et al., set sem back to initial state */
	up(&fs_info->uuid_tree_rescan_sem);

3925
	/* pause restriper - we want to resume on mount */
3926
	btrfs_pause_balance(fs_info);
3927

3928 3929
	btrfs_dev_replace_suspend_for_unmount(fs_info);

3930
	btrfs_scrub_cancel(fs_info);
C
Chris Mason 已提交
3931 3932 3933 3934 3935 3936

	/* wait for any defraggers to finish */
	wait_event(fs_info->transaction_wait,
		   (atomic_read(&fs_info->defrag_running) == 0));

	/* clear out the rbtree of defraggable inodes */
3937
	btrfs_cleanup_defrag_inodes(fs_info);
C
Chris Mason 已提交
3938

3939 3940
	cancel_work_sync(&fs_info->async_reclaim_work);

Y
Yan Zheng 已提交
3941
	if (!(fs_info->sb->s_flags & MS_RDONLY)) {
3942 3943 3944 3945 3946 3947 3948
		/*
		 * If the cleaner thread is stopped and there are
		 * block groups queued for removal, the deletion will be
		 * skipped when we quit the cleaner thread.
		 */
		btrfs_delete_unused_bgs(root->fs_info);

3949
		ret = btrfs_commit_super(fs_info);
L
liubo 已提交
3950
		if (ret)
3951
			btrfs_err(fs_info, "commit super ret %d", ret);
L
liubo 已提交
3952 3953
	}

3954
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state))
3955
		btrfs_error_commit_super(root);
3956

A
Al Viro 已提交
3957 3958
	kthread_stop(fs_info->transaction_kthread);
	kthread_stop(fs_info->cleaner_kthread);
3959

3960
	set_bit(BTRFS_FS_CLOSING_DONE, &fs_info->flags);
3961

3962
	btrfs_free_qgroup_config(fs_info);
3963

3964
	if (percpu_counter_sum(&fs_info->delalloc_bytes)) {
3965
		btrfs_info(fs_info, "at unmount delalloc count %lld",
3966
		       percpu_counter_sum(&fs_info->delalloc_bytes));
C
Chris Mason 已提交
3967
	}
3968

3969
	btrfs_sysfs_remove_mounted(fs_info);
3970
	btrfs_sysfs_remove_fsid(fs_info->fs_devices);
3971

3972
	btrfs_free_fs_roots(fs_info);
3973

3974 3975
	btrfs_put_block_group_cache(fs_info);

3976 3977
	btrfs_free_block_groups(fs_info);

3978 3979 3980 3981 3982
	/*
	 * we must make sure there is not any read request to
	 * submit after we stopping all workers.
	 */
	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
3983 3984
	btrfs_stop_all_workers(fs_info);

3985
	clear_bit(BTRFS_FS_OPEN, &fs_info->flags);
3986
	free_root_pointers(fs_info, 1);
3987

3988
	iput(fs_info->btree_inode);
3989

3990
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
3991
	if (btrfs_test_opt(root->fs_info, CHECK_INTEGRITY))
3992 3993 3994
		btrfsic_unmount(root, fs_info->fs_devices);
#endif

3995
	btrfs_close_devices(fs_info->fs_devices);
3996
	btrfs_mapping_tree_free(&fs_info->mapping_tree);
3997

3998
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
3999
	percpu_counter_destroy(&fs_info->delalloc_bytes);
4000
	percpu_counter_destroy(&fs_info->bio_counter);
C
Chris Mason 已提交
4001
	bdi_destroy(&fs_info->bdi);
4002
	cleanup_srcu_struct(&fs_info->subvol_srcu);
4003

D
David Woodhouse 已提交
4004 4005
	btrfs_free_stripe_hash_table(fs_info);

4006
	__btrfs_free_block_rsv(root->orphan_block_rsv);
4007
	root->orphan_block_rsv = NULL;
4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018

	lock_chunks(root);
	while (!list_empty(&fs_info->pinned_chunks)) {
		struct extent_map *em;

		em = list_first_entry(&fs_info->pinned_chunks,
				      struct extent_map, list);
		list_del_init(&em->list);
		free_extent_map(em);
	}
	unlock_chunks(root);
4019 4020
}

4021 4022
int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid,
			  int atomic)
4023
{
4024
	int ret;
4025
	struct inode *btree_inode = buf->pages[0]->mapping->host;
4026

4027
	ret = extent_buffer_uptodate(buf);
4028 4029 4030 4031
	if (!ret)
		return ret;

	ret = verify_parent_transid(&BTRFS_I(btree_inode)->io_tree, buf,
4032 4033 4034
				    parent_transid, atomic);
	if (ret == -EAGAIN)
		return ret;
4035
	return !ret;
4036 4037 4038 4039
}

void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
{
4040
	struct btrfs_root *root;
4041
	u64 transid = btrfs_header_generation(buf);
4042
	int was_dirty;
4043

4044 4045 4046 4047 4048 4049 4050 4051 4052 4053
#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
	/*
	 * This is a fast path so only do this check if we have sanity tests
	 * enabled.  Normal people shouldn't be marking dummy buffers as dirty
	 * outside of the sanity tests.
	 */
	if (unlikely(test_bit(EXTENT_BUFFER_DUMMY, &buf->bflags)))
		return;
#endif
	root = BTRFS_I(buf->pages[0]->mapping->host)->root;
4054
	btrfs_assert_tree_locked(buf);
J
Julia Lawall 已提交
4055
	if (transid != root->fs_info->generation)
J
Jeff Mahoney 已提交
4056
		WARN(1, KERN_CRIT "btrfs transid mismatch buffer %llu, found %llu running %llu\n",
4057
			buf->start, transid, root->fs_info->generation);
4058
	was_dirty = set_extent_buffer_dirty(buf);
4059 4060 4061 4062
	if (!was_dirty)
		__percpu_counter_add(&root->fs_info->dirty_metadata_bytes,
				     buf->len,
				     root->fs_info->dirty_metadata_batch);
4063 4064 4065 4066 4067 4068
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	if (btrfs_header_level(buf) == 0 && check_leaf(root, buf)) {
		btrfs_print_leaf(root, buf);
		ASSERT(0);
	}
#endif
4069 4070
}

4071 4072
static void __btrfs_btree_balance_dirty(struct btrfs_root *root,
					int flush_delayed)
4073 4074 4075 4076 4077
{
	/*
	 * looks as though older kernels can get into trouble with
	 * this code, they end up stuck in balance_dirty_pages forever
	 */
4078
	int ret;
4079 4080 4081 4082

	if (current->flags & PF_MEMALLOC)
		return;

4083 4084
	if (flush_delayed)
		btrfs_balance_delayed_items(root);
4085

4086 4087 4088
	ret = percpu_counter_compare(&root->fs_info->dirty_metadata_bytes,
				     BTRFS_DIRTY_METADATA_THRESH);
	if (ret > 0) {
4089 4090
		balance_dirty_pages_ratelimited(
				   root->fs_info->btree_inode->i_mapping);
4091 4092 4093
	}
}

4094
void btrfs_btree_balance_dirty(struct btrfs_root *root)
C
Chris Mason 已提交
4095
{
4096 4097
	__btrfs_btree_balance_dirty(root, 1);
}
4098

4099 4100 4101
void btrfs_btree_balance_dirty_nodelay(struct btrfs_root *root)
{
	__btrfs_btree_balance_dirty(root, 0);
C
Chris Mason 已提交
4102
}
4103

4104
int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid)
4105
{
4106
	struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root;
4107
	return btree_read_extent_buffer_pages(root, buf, parent_transid);
4108
}
4109

4110
static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info,
L
liubo 已提交
4111 4112
			      int read_only)
{
D
David Sterba 已提交
4113
	struct btrfs_super_block *sb = fs_info->super_copy;
4114 4115
	u64 nodesize = btrfs_super_nodesize(sb);
	u64 sectorsize = btrfs_super_sectorsize(sb);
D
David Sterba 已提交
4116 4117
	int ret = 0;

4118
	if (btrfs_super_magic(sb) != BTRFS_MAGIC) {
4119
		btrfs_err(fs_info, "no valid FS found");
4120 4121 4122
		ret = -EINVAL;
	}
	if (btrfs_super_flags(sb) & ~BTRFS_SUPER_FLAG_SUPP)
4123
		btrfs_warn(fs_info, "unrecognized super flag: %llu",
4124
				btrfs_super_flags(sb) & ~BTRFS_SUPER_FLAG_SUPP);
4125
	if (btrfs_super_root_level(sb) >= BTRFS_MAX_LEVEL) {
4126
		btrfs_err(fs_info, "tree_root level too big: %d >= %d",
4127
				btrfs_super_root_level(sb), BTRFS_MAX_LEVEL);
D
David Sterba 已提交
4128 4129
		ret = -EINVAL;
	}
4130
	if (btrfs_super_chunk_root_level(sb) >= BTRFS_MAX_LEVEL) {
4131
		btrfs_err(fs_info, "chunk_root level too big: %d >= %d",
4132
				btrfs_super_chunk_root_level(sb), BTRFS_MAX_LEVEL);
D
David Sterba 已提交
4133 4134
		ret = -EINVAL;
	}
4135
	if (btrfs_super_log_root_level(sb) >= BTRFS_MAX_LEVEL) {
4136
		btrfs_err(fs_info, "log_root level too big: %d >= %d",
4137
				btrfs_super_log_root_level(sb), BTRFS_MAX_LEVEL);
D
David Sterba 已提交
4138 4139 4140
		ret = -EINVAL;
	}

D
David Sterba 已提交
4141
	/*
4142 4143
	 * Check sectorsize and nodesize first, other check will need it.
	 * Check all possible sectorsize(4K, 8K, 16K, 32K, 64K) here.
D
David Sterba 已提交
4144
	 */
4145 4146
	if (!is_power_of_2(sectorsize) || sectorsize < 4096 ||
	    sectorsize > BTRFS_MAX_METADATA_BLOCKSIZE) {
4147
		btrfs_err(fs_info, "invalid sectorsize %llu", sectorsize);
4148 4149 4150
		ret = -EINVAL;
	}
	/* Only PAGE SIZE is supported yet */
4151
	if (sectorsize != PAGE_SIZE) {
4152 4153 4154
		btrfs_err(fs_info,
			"sectorsize %llu not supported yet, only support %lu",
			sectorsize, PAGE_SIZE);
4155 4156 4157 4158
		ret = -EINVAL;
	}
	if (!is_power_of_2(nodesize) || nodesize < sectorsize ||
	    nodesize > BTRFS_MAX_METADATA_BLOCKSIZE) {
4159
		btrfs_err(fs_info, "invalid nodesize %llu", nodesize);
4160 4161 4162
		ret = -EINVAL;
	}
	if (nodesize != le32_to_cpu(sb->__unused_leafsize)) {
4163 4164
		btrfs_err(fs_info, "invalid leafsize %u, should be %llu",
			  le32_to_cpu(sb->__unused_leafsize), nodesize);
4165 4166 4167 4168 4169
		ret = -EINVAL;
	}

	/* Root alignment check */
	if (!IS_ALIGNED(btrfs_super_root(sb), sectorsize)) {
4170 4171
		btrfs_warn(fs_info, "tree_root block unaligned: %llu",
			   btrfs_super_root(sb));
4172 4173 4174
		ret = -EINVAL;
	}
	if (!IS_ALIGNED(btrfs_super_chunk_root(sb), sectorsize)) {
4175 4176
		btrfs_warn(fs_info, "chunk_root block unaligned: %llu",
			   btrfs_super_chunk_root(sb));
4177 4178
		ret = -EINVAL;
	}
4179
	if (!IS_ALIGNED(btrfs_super_log_root(sb), sectorsize)) {
4180 4181
		btrfs_warn(fs_info, "log_root block unaligned: %llu",
			   btrfs_super_log_root(sb));
4182 4183 4184
		ret = -EINVAL;
	}

D
David Sterba 已提交
4185
	if (memcmp(fs_info->fsid, sb->dev_item.fsid, BTRFS_UUID_SIZE) != 0) {
4186 4187 4188
		btrfs_err(fs_info,
			   "dev_item UUID does not match fsid: %pU != %pU",
			   fs_info->fsid, sb->dev_item.fsid);
D
David Sterba 已提交
4189 4190 4191 4192 4193 4194 4195
		ret = -EINVAL;
	}

	/*
	 * Hint to catch really bogus numbers, bitflips or so, more exact checks are
	 * done later
	 */
4196 4197
	if (btrfs_super_bytes_used(sb) < 6 * btrfs_super_nodesize(sb)) {
		btrfs_err(fs_info, "bytes_used is too small %llu",
4198
			  btrfs_super_bytes_used(sb));
4199 4200
		ret = -EINVAL;
	}
4201
	if (!is_power_of_2(btrfs_super_stripesize(sb))) {
4202
		btrfs_err(fs_info, "invalid stripesize %u",
4203
			  btrfs_super_stripesize(sb));
4204 4205
		ret = -EINVAL;
	}
4206
	if (btrfs_super_num_devices(sb) > (1UL << 31))
4207 4208
		btrfs_warn(fs_info, "suspicious number of devices: %llu",
			   btrfs_super_num_devices(sb));
4209
	if (btrfs_super_num_devices(sb) == 0) {
4210
		btrfs_err(fs_info, "number of devices is 0");
4211 4212
		ret = -EINVAL;
	}
D
David Sterba 已提交
4213

4214
	if (btrfs_super_bytenr(sb) != BTRFS_SUPER_INFO_OFFSET) {
4215 4216
		btrfs_err(fs_info, "super offset mismatch %llu != %u",
			  btrfs_super_bytenr(sb), BTRFS_SUPER_INFO_OFFSET);
D
David Sterba 已提交
4217 4218 4219
		ret = -EINVAL;
	}

4220 4221 4222 4223 4224
	/*
	 * Obvious sys_chunk_array corruptions, it must hold at least one key
	 * and one chunk
	 */
	if (btrfs_super_sys_array_size(sb) > BTRFS_SYSTEM_CHUNK_ARRAY_SIZE) {
4225 4226 4227
		btrfs_err(fs_info, "system chunk array too big %u > %u",
			  btrfs_super_sys_array_size(sb),
			  BTRFS_SYSTEM_CHUNK_ARRAY_SIZE);
4228 4229 4230 4231
		ret = -EINVAL;
	}
	if (btrfs_super_sys_array_size(sb) < sizeof(struct btrfs_disk_key)
			+ sizeof(struct btrfs_chunk)) {
4232 4233 4234 4235
		btrfs_err(fs_info, "system chunk array too small %u < %zu",
			  btrfs_super_sys_array_size(sb),
			  sizeof(struct btrfs_disk_key)
			  + sizeof(struct btrfs_chunk));
4236 4237 4238
		ret = -EINVAL;
	}

D
David Sterba 已提交
4239 4240 4241 4242
	/*
	 * The generation is a global counter, we'll trust it more than the others
	 * but it's still possible that it's the one that's wrong.
	 */
4243
	if (btrfs_super_generation(sb) < btrfs_super_chunk_root_generation(sb))
4244 4245 4246 4247
		btrfs_warn(fs_info,
			"suspicious: generation < chunk_root_generation: %llu < %llu",
			btrfs_super_generation(sb),
			btrfs_super_chunk_root_generation(sb));
4248 4249
	if (btrfs_super_generation(sb) < btrfs_super_cache_generation(sb)
	    && btrfs_super_cache_generation(sb) != (u64)-1)
4250 4251 4252 4253
		btrfs_warn(fs_info,
			"suspicious: generation < cache_generation: %llu < %llu",
			btrfs_super_generation(sb),
			btrfs_super_cache_generation(sb));
D
David Sterba 已提交
4254 4255

	return ret;
L
liubo 已提交
4256 4257
}

4258
static void btrfs_error_commit_super(struct btrfs_root *root)
L
liubo 已提交
4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270
{
	mutex_lock(&root->fs_info->cleaner_mutex);
	btrfs_run_delayed_iputs(root);
	mutex_unlock(&root->fs_info->cleaner_mutex);

	down_write(&root->fs_info->cleanup_work_sem);
	up_write(&root->fs_info->cleanup_work_sem);

	/* cleanup FS via transaction */
	btrfs_cleanup_transaction(root);
}

4271
static void btrfs_destroy_ordered_extents(struct btrfs_root *root)
L
liubo 已提交
4272 4273 4274
{
	struct btrfs_ordered_extent *ordered;

4275
	spin_lock(&root->ordered_extent_lock);
4276 4277 4278 4279
	/*
	 * This will just short circuit the ordered completion stuff which will
	 * make sure the ordered extent gets properly cleaned up.
	 */
4280
	list_for_each_entry(ordered, &root->ordered_extents,
4281 4282
			    root_extent_list)
		set_bit(BTRFS_ORDERED_IOERR, &ordered->flags);
4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297
	spin_unlock(&root->ordered_extent_lock);
}

static void btrfs_destroy_all_ordered_extents(struct btrfs_fs_info *fs_info)
{
	struct btrfs_root *root;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

	spin_lock(&fs_info->ordered_root_lock);
	list_splice_init(&fs_info->ordered_roots, &splice);
	while (!list_empty(&splice)) {
		root = list_first_entry(&splice, struct btrfs_root,
					ordered_root);
4298 4299
		list_move_tail(&root->ordered_root,
			       &fs_info->ordered_roots);
4300

4301
		spin_unlock(&fs_info->ordered_root_lock);
4302 4303
		btrfs_destroy_ordered_extents(root);

4304 4305
		cond_resched();
		spin_lock(&fs_info->ordered_root_lock);
4306 4307
	}
	spin_unlock(&fs_info->ordered_root_lock);
L
liubo 已提交
4308 4309
}

4310 4311
static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
				      struct btrfs_root *root)
L
liubo 已提交
4312 4313 4314 4315 4316 4317 4318 4319 4320
{
	struct rb_node *node;
	struct btrfs_delayed_ref_root *delayed_refs;
	struct btrfs_delayed_ref_node *ref;
	int ret = 0;

	delayed_refs = &trans->delayed_refs;

	spin_lock(&delayed_refs->lock);
4321
	if (atomic_read(&delayed_refs->num_entries) == 0) {
4322
		spin_unlock(&delayed_refs->lock);
4323
		btrfs_info(root->fs_info, "delayed_refs has NO entry");
L
liubo 已提交
4324 4325 4326
		return ret;
	}

4327 4328
	while ((node = rb_first(&delayed_refs->href_root)) != NULL) {
		struct btrfs_delayed_ref_head *head;
4329
		struct btrfs_delayed_ref_node *tmp;
4330
		bool pin_bytes = false;
L
liubo 已提交
4331

4332 4333 4334 4335 4336
		head = rb_entry(node, struct btrfs_delayed_ref_head,
				href_node);
		if (!mutex_trylock(&head->mutex)) {
			atomic_inc(&head->node.refs);
			spin_unlock(&delayed_refs->lock);
4337

4338
			mutex_lock(&head->mutex);
4339
			mutex_unlock(&head->mutex);
4340 4341 4342 4343 4344
			btrfs_put_delayed_ref(&head->node);
			spin_lock(&delayed_refs->lock);
			continue;
		}
		spin_lock(&head->lock);
4345 4346
		list_for_each_entry_safe_reverse(ref, tmp, &head->ref_list,
						 list) {
4347
			ref->in_tree = 0;
4348
			list_del(&ref->list);
4349 4350
			if (!list_empty(&ref->add_list))
				list_del(&ref->add_list);
4351 4352
			atomic_dec(&delayed_refs->num_entries);
			btrfs_put_delayed_ref(ref);
4353
		}
4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365
		if (head->must_insert_reserved)
			pin_bytes = true;
		btrfs_free_delayed_extent_op(head->extent_op);
		delayed_refs->num_heads--;
		if (head->processing == 0)
			delayed_refs->num_heads_ready--;
		atomic_dec(&delayed_refs->num_entries);
		head->node.in_tree = 0;
		rb_erase(&head->href_node, &delayed_refs->href_root);
		spin_unlock(&head->lock);
		spin_unlock(&delayed_refs->lock);
		mutex_unlock(&head->mutex);
L
liubo 已提交
4366

4367 4368 4369 4370
		if (pin_bytes)
			btrfs_pin_extent(root, head->node.bytenr,
					 head->node.num_bytes, 1);
		btrfs_put_delayed_ref(&head->node);
L
liubo 已提交
4371 4372 4373 4374 4375 4376 4377 4378 4379
		cond_resched();
		spin_lock(&delayed_refs->lock);
	}

	spin_unlock(&delayed_refs->lock);

	return ret;
}

4380
static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root)
L
liubo 已提交
4381 4382 4383 4384 4385 4386
{
	struct btrfs_inode *btrfs_inode;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

4387 4388
	spin_lock(&root->delalloc_lock);
	list_splice_init(&root->delalloc_inodes, &splice);
L
liubo 已提交
4389 4390

	while (!list_empty(&splice)) {
4391 4392
		btrfs_inode = list_first_entry(&splice, struct btrfs_inode,
					       delalloc_inodes);
L
liubo 已提交
4393 4394

		list_del_init(&btrfs_inode->delalloc_inodes);
4395 4396
		clear_bit(BTRFS_INODE_IN_DELALLOC_LIST,
			  &btrfs_inode->runtime_flags);
4397
		spin_unlock(&root->delalloc_lock);
L
liubo 已提交
4398 4399

		btrfs_invalidate_inodes(btrfs_inode->root);
4400

4401
		spin_lock(&root->delalloc_lock);
L
liubo 已提交
4402 4403
	}

4404 4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429
	spin_unlock(&root->delalloc_lock);
}

static void btrfs_destroy_all_delalloc_inodes(struct btrfs_fs_info *fs_info)
{
	struct btrfs_root *root;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

	spin_lock(&fs_info->delalloc_root_lock);
	list_splice_init(&fs_info->delalloc_roots, &splice);
	while (!list_empty(&splice)) {
		root = list_first_entry(&splice, struct btrfs_root,
					 delalloc_root);
		list_del_init(&root->delalloc_root);
		root = btrfs_grab_fs_root(root);
		BUG_ON(!root);
		spin_unlock(&fs_info->delalloc_root_lock);

		btrfs_destroy_delalloc_inodes(root);
		btrfs_put_fs_root(root);

		spin_lock(&fs_info->delalloc_root_lock);
	}
	spin_unlock(&fs_info->delalloc_root_lock);
L
liubo 已提交
4430 4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442
}

static int btrfs_destroy_marked_extents(struct btrfs_root *root,
					struct extent_io_tree *dirty_pages,
					int mark)
{
	int ret;
	struct extent_buffer *eb;
	u64 start = 0;
	u64 end;

	while (1) {
		ret = find_first_extent_bit(dirty_pages, start, &start, &end,
4443
					    mark, NULL);
L
liubo 已提交
4444 4445 4446
		if (ret)
			break;

4447
		clear_extent_bits(dirty_pages, start, end, mark);
L
liubo 已提交
4448
		while (start <= end) {
4449
			eb = find_extent_buffer(root->fs_info, start);
4450
			start += root->nodesize;
4451
			if (!eb)
L
liubo 已提交
4452
				continue;
4453
			wait_on_extent_buffer_writeback(eb);
L
liubo 已提交
4454

4455 4456 4457 4458
			if (test_and_clear_bit(EXTENT_BUFFER_DIRTY,
					       &eb->bflags))
				clear_extent_buffer_dirty(eb);
			free_extent_buffer_stale(eb);
L
liubo 已提交
4459 4460 4461 4462 4463 4464 4465 4466 4467 4468 4469 4470 4471
		}
	}

	return ret;
}

static int btrfs_destroy_pinned_extent(struct btrfs_root *root,
				       struct extent_io_tree *pinned_extents)
{
	struct extent_io_tree *unpin;
	u64 start;
	u64 end;
	int ret;
4472
	bool loop = true;
L
liubo 已提交
4473 4474

	unpin = pinned_extents;
4475
again:
L
liubo 已提交
4476 4477
	while (1) {
		ret = find_first_extent_bit(unpin, 0, &start, &end,
4478
					    EXTENT_DIRTY, NULL);
L
liubo 已提交
4479 4480 4481
		if (ret)
			break;

4482
		clear_extent_dirty(unpin, start, end);
L
liubo 已提交
4483 4484 4485 4486
		btrfs_error_unpin_extent_range(root, start, end);
		cond_resched();
	}

4487 4488 4489 4490 4491 4492 4493 4494 4495
	if (loop) {
		if (unpin == &root->fs_info->freed_extents[0])
			unpin = &root->fs_info->freed_extents[1];
		else
			unpin = &root->fs_info->freed_extents[0];
		loop = false;
		goto again;
	}

L
liubo 已提交
4496 4497 4498
	return 0;
}

4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527 4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553 4554 4555 4556 4557 4558 4559 4560 4561 4562 4563 4564 4565
static void btrfs_cleanup_bg_io(struct btrfs_block_group_cache *cache)
{
	struct inode *inode;

	inode = cache->io_ctl.inode;
	if (inode) {
		invalidate_inode_pages2(inode->i_mapping);
		BTRFS_I(inode)->generation = 0;
		cache->io_ctl.inode = NULL;
		iput(inode);
	}
	btrfs_put_block_group(cache);
}

void btrfs_cleanup_dirty_bgs(struct btrfs_transaction *cur_trans,
			     struct btrfs_root *root)
{
	struct btrfs_block_group_cache *cache;

	spin_lock(&cur_trans->dirty_bgs_lock);
	while (!list_empty(&cur_trans->dirty_bgs)) {
		cache = list_first_entry(&cur_trans->dirty_bgs,
					 struct btrfs_block_group_cache,
					 dirty_list);
		if (!cache) {
			btrfs_err(root->fs_info,
				  "orphan block group dirty_bgs list");
			spin_unlock(&cur_trans->dirty_bgs_lock);
			return;
		}

		if (!list_empty(&cache->io_list)) {
			spin_unlock(&cur_trans->dirty_bgs_lock);
			list_del_init(&cache->io_list);
			btrfs_cleanup_bg_io(cache);
			spin_lock(&cur_trans->dirty_bgs_lock);
		}

		list_del_init(&cache->dirty_list);
		spin_lock(&cache->lock);
		cache->disk_cache_state = BTRFS_DC_ERROR;
		spin_unlock(&cache->lock);

		spin_unlock(&cur_trans->dirty_bgs_lock);
		btrfs_put_block_group(cache);
		spin_lock(&cur_trans->dirty_bgs_lock);
	}
	spin_unlock(&cur_trans->dirty_bgs_lock);

	while (!list_empty(&cur_trans->io_bgs)) {
		cache = list_first_entry(&cur_trans->io_bgs,
					 struct btrfs_block_group_cache,
					 io_list);
		if (!cache) {
			btrfs_err(root->fs_info,
				  "orphan block group on io_bgs list");
			return;
		}

		list_del_init(&cache->io_list);
		spin_lock(&cache->lock);
		cache->disk_cache_state = BTRFS_DC_ERROR;
		spin_unlock(&cache->lock);
		btrfs_cleanup_bg_io(cache);
	}
}

4566 4567 4568
void btrfs_cleanup_one_transaction(struct btrfs_transaction *cur_trans,
				   struct btrfs_root *root)
{
4569 4570 4571 4572
	btrfs_cleanup_dirty_bgs(cur_trans, root);
	ASSERT(list_empty(&cur_trans->dirty_bgs));
	ASSERT(list_empty(&cur_trans->io_bgs));

4573 4574
	btrfs_destroy_delayed_refs(cur_trans, root);

4575
	cur_trans->state = TRANS_STATE_COMMIT_START;
4576
	wake_up(&root->fs_info->transaction_blocked_wait);
4577

4578
	cur_trans->state = TRANS_STATE_UNBLOCKED;
4579
	wake_up(&root->fs_info->transaction_wait);
4580

4581 4582
	btrfs_destroy_delayed_inodes(root);
	btrfs_assert_delayed_root_empty(root);
4583 4584 4585

	btrfs_destroy_marked_extents(root, &cur_trans->dirty_pages,
				     EXTENT_DIRTY);
4586 4587
	btrfs_destroy_pinned_extent(root,
				    root->fs_info->pinned_extents);
4588

4589 4590 4591
	cur_trans->state =TRANS_STATE_COMPLETED;
	wake_up(&cur_trans->commit_wait);

4592 4593 4594 4595 4596 4597
	/*
	memset(cur_trans, 0, sizeof(*cur_trans));
	kmem_cache_free(btrfs_transaction_cachep, cur_trans);
	*/
}

4598
static int btrfs_cleanup_transaction(struct btrfs_root *root)
L
liubo 已提交
4599 4600 4601 4602 4603
{
	struct btrfs_transaction *t;

	mutex_lock(&root->fs_info->transaction_kthread_mutex);

J
Josef Bacik 已提交
4604
	spin_lock(&root->fs_info->trans_lock);
4605 4606 4607 4608 4609 4610 4611 4612 4613 4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628
	while (!list_empty(&root->fs_info->trans_list)) {
		t = list_first_entry(&root->fs_info->trans_list,
				     struct btrfs_transaction, list);
		if (t->state >= TRANS_STATE_COMMIT_START) {
			atomic_inc(&t->use_count);
			spin_unlock(&root->fs_info->trans_lock);
			btrfs_wait_for_commit(root, t->transid);
			btrfs_put_transaction(t);
			spin_lock(&root->fs_info->trans_lock);
			continue;
		}
		if (t == root->fs_info->running_transaction) {
			t->state = TRANS_STATE_COMMIT_DOING;
			spin_unlock(&root->fs_info->trans_lock);
			/*
			 * We wait for 0 num_writers since we don't hold a trans
			 * handle open currently for this transaction.
			 */
			wait_event(t->writer_wait,
				   atomic_read(&t->num_writers) == 0);
		} else {
			spin_unlock(&root->fs_info->trans_lock);
		}
		btrfs_cleanup_one_transaction(t, root);
4629

4630 4631 4632
		spin_lock(&root->fs_info->trans_lock);
		if (t == root->fs_info->running_transaction)
			root->fs_info->running_transaction = NULL;
L
liubo 已提交
4633
		list_del_init(&t->list);
4634
		spin_unlock(&root->fs_info->trans_lock);
L
liubo 已提交
4635

4636 4637 4638 4639 4640 4641 4642 4643 4644 4645
		btrfs_put_transaction(t);
		trace_btrfs_transaction_commit(root);
		spin_lock(&root->fs_info->trans_lock);
	}
	spin_unlock(&root->fs_info->trans_lock);
	btrfs_destroy_all_ordered_extents(root->fs_info);
	btrfs_destroy_delayed_inodes(root);
	btrfs_assert_delayed_root_empty(root);
	btrfs_destroy_pinned_extent(root, root->fs_info->pinned_extents);
	btrfs_destroy_all_delalloc_inodes(root->fs_info);
L
liubo 已提交
4646 4647 4648 4649 4650
	mutex_unlock(&root->fs_info->transaction_kthread_mutex);

	return 0;
}

4651
static const struct extent_io_ops btree_extent_io_ops = {
4652
	.readpage_end_io_hook = btree_readpage_end_io_hook,
A
Arne Jansen 已提交
4653
	.readpage_io_failed_hook = btree_io_failed_hook,
4654
	.submit_bio_hook = btree_submit_bio_hook,
4655 4656
	/* note we're sharing with inode.c for the merge bio hook */
	.merge_bio_hook = btrfs_merge_bio_hook,
4657
};