disk-io.c 121.2 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 <linux/error-injection.h>
34
#include <asm/unaligned.h>
35 36
#include "ctree.h"
#include "disk-io.h"
37
#include "hash.h"
38
#include "transaction.h"
39
#include "btrfs_inode.h"
40
#include "volumes.h"
41
#include "print-tree.h"
42
#include "locking.h"
43
#include "tree-log.h"
44
#include "free-space-cache.h"
45
#include "free-space-tree.h"
46
#include "inode-map.h"
47
#include "check-integrity.h"
48
#include "rcu-string.h"
49
#include "dev-replace.h"
D
David Woodhouse 已提交
50
#include "raid56.h"
51
#include "sysfs.h"
J
Josef Bacik 已提交
52
#include "qgroup.h"
53
#include "compression.h"
54
#include "tree-checker.h"
J
Josef Bacik 已提交
55
#include "ref-verify.h"
56

57 58 59 60
#ifdef CONFIG_X86
#include <asm/cpufeature.h>
#endif

61 62 63 64
#define BTRFS_SUPER_FLAG_SUPP	(BTRFS_HEADER_FLAG_WRITTEN |\
				 BTRFS_HEADER_FLAG_RELOC |\
				 BTRFS_SUPER_FLAG_ERROR |\
				 BTRFS_SUPER_FLAG_SEEDING |\
65 66
				 BTRFS_SUPER_FLAG_METADUMP |\
				 BTRFS_SUPER_FLAG_METADUMP_V2)
67

68
static const struct extent_io_ops btree_extent_io_ops;
69
static void end_workqueue_fn(struct btrfs_work *work);
70
static void free_fs_root(struct btrfs_root *root);
71
static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info);
72
static void btrfs_destroy_ordered_extents(struct btrfs_root *root);
L
liubo 已提交
73
static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
74
				      struct btrfs_fs_info *fs_info);
75
static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root);
76
static int btrfs_destroy_marked_extents(struct btrfs_fs_info *fs_info,
L
liubo 已提交
77 78
					struct extent_io_tree *dirty_pages,
					int mark);
79
static int btrfs_destroy_pinned_extent(struct btrfs_fs_info *fs_info,
L
liubo 已提交
80
				       struct extent_io_tree *pinned_extents);
81 82
static int btrfs_cleanup_transaction(struct btrfs_fs_info *fs_info);
static void btrfs_error_commit_super(struct btrfs_fs_info *fs_info);
83

C
Chris Mason 已提交
84
/*
85 86
 * 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
87 88
 * by writes to insert metadata for new file extents after IO is complete.
 */
89
struct btrfs_end_io_wq {
90 91 92 93
	struct bio *bio;
	bio_end_io_t *end_io;
	void *private;
	struct btrfs_fs_info *info;
94
	blk_status_t status;
95
	enum btrfs_wq_endio_type metadata;
96
	struct btrfs_work work;
97
};
98

99 100 101 102 103 104 105
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,
106
					SLAB_MEM_SPREAD,
107 108 109 110 111 112 113 114
					NULL);
	if (!btrfs_end_io_wq_cache)
		return -ENOMEM;
	return 0;
}

void btrfs_end_io_wq_exit(void)
{
115
	kmem_cache_destroy(btrfs_end_io_wq_cache);
116 117
}

C
Chris Mason 已提交
118 119 120 121 122
/*
 * 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.
 */
123
struct async_submit_bio {
124 125
	void *private_data;
	struct btrfs_fs_info *fs_info;
126
	struct bio *bio;
C
Chris Mason 已提交
127 128
	extent_submit_bio_hook_t *submit_bio_start;
	extent_submit_bio_hook_t *submit_bio_done;
129
	int mirror_num;
C
Chris Mason 已提交
130
	unsigned long bio_flags;
131 132 133 134 135
	/*
	 * 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;
136
	struct btrfs_work work;
137
	blk_status_t status;
138 139
};

140 141 142 143 144 145 146 147 148 149
/*
 * 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.
150
 *
151 152 153
 * 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.
154
 *
155 156 157
 * 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.
158
 *
159 160 161
 * 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.
162 163 164 165 166
 */
#ifdef CONFIG_DEBUG_LOCK_ALLOC
# if BTRFS_MAX_LEVEL != 8
#  error
# endif
167 168 169 170 171 172 173 174 175 176 177 178 179

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"	},
180
	{ .id = BTRFS_QUOTA_TREE_OBJECTID,	.name_stem = "quota"	},
181 182 183
	{ .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"	},
184
	{ .id = BTRFS_UUID_TREE_OBJECTID,	.name_stem = "uuid"	},
185
	{ .id = BTRFS_FREE_SPACE_TREE_OBJECTID,	.name_stem = "free-space" },
186
	{ .id = 0,				.name_stem = "tree"	},
187
};
188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218

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]);
}

219 220
#endif

C
Chris Mason 已提交
221 222 223 224
/*
 * extents on the btree inode are pretty simple, there's one extent
 * that covers the entire device
 */
225
struct extent_map *btree_get_extent(struct btrfs_inode *inode,
226
		struct page *page, size_t pg_offset, u64 start, u64 len,
227
		int create)
228
{
229 230
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->vfs_inode.i_sb);
	struct extent_map_tree *em_tree = &inode->extent_tree;
231 232 233
	struct extent_map *em;
	int ret;

234
	read_lock(&em_tree->lock);
235
	em = lookup_extent_mapping(em_tree, start, len);
236
	if (em) {
237
		em->bdev = fs_info->fs_devices->latest_bdev;
238
		read_unlock(&em_tree->lock);
239
		goto out;
240
	}
241
	read_unlock(&em_tree->lock);
242

243
	em = alloc_extent_map();
244 245 246 247 248
	if (!em) {
		em = ERR_PTR(-ENOMEM);
		goto out;
	}
	em->start = 0;
249
	em->len = (u64)-1;
C
Chris Mason 已提交
250
	em->block_len = (u64)-1;
251
	em->block_start = 0;
252
	em->bdev = fs_info->fs_devices->latest_bdev;
253

254
	write_lock(&em_tree->lock);
J
Josef Bacik 已提交
255
	ret = add_extent_mapping(em_tree, em, 0);
256 257
	if (ret == -EEXIST) {
		free_extent_map(em);
258
		em = lookup_extent_mapping(em_tree, start, len);
259
		if (!em)
260
			em = ERR_PTR(-EIO);
261
	} else if (ret) {
262
		free_extent_map(em);
263
		em = ERR_PTR(ret);
264
	}
265
	write_unlock(&em_tree->lock);
266

267 268
out:
	return em;
269 270
}

271
u32 btrfs_csum_data(const char *data, u32 seed, size_t len)
272
{
273
	return btrfs_crc32c(seed, data, len);
274 275
}

276
void btrfs_csum_final(u32 crc, u8 *result)
277
{
278
	put_unaligned_le32(~crc, result);
279 280
}

C
Chris Mason 已提交
281 282 283 284
/*
 * compute the csum for a btree block, and either verify it or write it
 * into the csum field of the block.
 */
285 286
static int csum_tree_block(struct btrfs_fs_info *fs_info,
			   struct extent_buffer *buf,
287 288
			   int verify)
{
289
	u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
290
	char result[BTRFS_CSUM_SIZE];
291 292 293 294 295 296 297 298 299 300
	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;

	len = buf->len - offset;
C
Chris Mason 已提交
301
	while (len > 0) {
302
		err = map_private_extent_buffer(buf, offset, 32,
303
					&kaddr, &map_start, &map_len);
C
Chris Mason 已提交
304
		if (err)
305
			return err;
306
		cur_len = min(len, map_len - (offset - map_start));
307
		crc = btrfs_csum_data(kaddr + offset - map_start,
308 309 310 311
				      crc, cur_len);
		len -= cur_len;
		offset += cur_len;
	}
312
	memset(result, 0, BTRFS_CSUM_SIZE);
313

314 315 316
	btrfs_csum_final(crc, result);

	if (verify) {
317
		if (memcmp_extent_buffer(buf, result, 0, csum_size)) {
318 319
			u32 val;
			u32 found = 0;
320
			memcpy(&found, result, csum_size);
321

322
			read_extent_buffer(buf, &val, 0, csum_size);
323
			btrfs_warn_rl(fs_info,
J
Jeff Mahoney 已提交
324
				"%s checksum verify failed on %llu wanted %X found %X level %d",
325
				fs_info->sb->s_id, buf->start,
326
				val, found, btrfs_header_level(buf));
327
			return -EUCLEAN;
328 329
		}
	} else {
330
		write_extent_buffer(buf, result, 0, csum_size);
331
	}
332

333 334 335
	return 0;
}

C
Chris Mason 已提交
336 337 338 339 340 341
/*
 * 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.
 */
342
static int verify_parent_transid(struct extent_io_tree *io_tree,
343 344
				 struct extent_buffer *eb, u64 parent_transid,
				 int atomic)
345
{
346
	struct extent_state *cached_state = NULL;
347
	int ret;
348
	bool need_lock = (current->journal_info == BTRFS_SEND_TRANS_STUB);
349 350 351 352

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

353 354 355
	if (atomic)
		return -EAGAIN;

356 357 358 359 360
	if (need_lock) {
		btrfs_tree_read_lock(eb);
		btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
	}

361
	lock_extent_bits(io_tree, eb->start, eb->start + eb->len - 1,
362
			 &cached_state);
363
	if (extent_buffer_uptodate(eb) &&
364 365 366 367
	    btrfs_header_generation(eb) == parent_transid) {
		ret = 0;
		goto out;
	}
368 369 370
	btrfs_err_rl(eb->fs_info,
		"parent transid verify failed on %llu wanted %llu found %llu",
			eb->start,
371
			parent_transid, btrfs_header_generation(eb));
372
	ret = 1;
373 374 375 376

	/*
	 * 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
377
	 * block that has been freed and re-allocated.  So don't clear uptodate
378 379 380 381 382 383
	 * 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
384
out:
385
	unlock_extent_cached(io_tree, eb->start, eb->start + eb->len - 1,
386
			     &cached_state);
387 388
	if (need_lock)
		btrfs_tree_read_unlock_blocking(eb);
389 390 391
	return ret;
}

D
David Sterba 已提交
392 393 394 395
/*
 * Return 0 if the superblock checksum type matches the checksum value of that
 * algorithm. Pass the raw disk superblock data.
 */
396 397
static int btrfs_check_super_csum(struct btrfs_fs_info *fs_info,
				  char *raw_disk_sb)
D
David Sterba 已提交
398 399 400 401 402 403 404 405 406 407 408 409 410 411
{
	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
412
		 * is filled with zeros and is included in the checksum.
D
David Sterba 已提交
413 414 415 416 417 418 419 420 421 422
		 */
		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)) {
423
		btrfs_err(fs_info, "unsupported checksum algorithm %u",
D
David Sterba 已提交
424 425 426 427 428 429 430
				csum_type);
		ret = 1;
	}

	return ret;
}

C
Chris Mason 已提交
431 432 433 434
/*
 * helper to read a given tree block, doing retries as required when
 * the checksums don't match and we have alternate mirrors to try.
 */
435
static int btree_read_extent_buffer_pages(struct btrfs_fs_info *fs_info,
436
					  struct extent_buffer *eb,
437
					  u64 parent_transid)
438 439
{
	struct extent_io_tree *io_tree;
440
	int failed = 0;
441 442 443
	int ret;
	int num_copies = 0;
	int mirror_num = 0;
444
	int failed_mirror = 0;
445

446
	clear_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
447
	io_tree = &BTRFS_I(fs_info->btree_inode)->io_tree;
448
	while (1) {
449
		ret = read_extent_buffer_pages(io_tree, eb, WAIT_COMPLETE,
450
					       mirror_num);
451 452
		if (!ret) {
			if (!verify_parent_transid(io_tree, eb,
453
						   parent_transid, 0))
454 455 456 457
				break;
			else
				ret = -EIO;
		}
C
Chris Mason 已提交
458

459 460 461 462 463 464
		/*
		 * 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))
465 466
			break;

467
		num_copies = btrfs_num_copies(fs_info,
468
					      eb->start, eb->len);
C
Chris Mason 已提交
469
		if (num_copies == 1)
470
			break;
C
Chris Mason 已提交
471

472 473 474 475 476
		if (!failed_mirror) {
			failed = 1;
			failed_mirror = eb->read_mirror;
		}

477
		mirror_num++;
478 479 480
		if (mirror_num == failed_mirror)
			mirror_num++;

C
Chris Mason 已提交
481
		if (mirror_num > num_copies)
482
			break;
483
	}
484

485
	if (failed && !ret && failed_mirror)
486
		repair_eb_io_failure(fs_info, eb, failed_mirror);
487 488

	return ret;
489
}
490

C
Chris Mason 已提交
491
/*
C
Chris Mason 已提交
492 493
 * 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
494
 */
C
Chris Mason 已提交
495

496
static int csum_dirty_buffer(struct btrfs_fs_info *fs_info, struct page *page)
497
{
M
Miao Xie 已提交
498
	u64 start = page_offset(page);
499 500
	u64 found_start;
	struct extent_buffer *eb;
501

J
Josef Bacik 已提交
502 503 504
	eb = (struct extent_buffer *)page->private;
	if (page != eb->pages[0])
		return 0;
505

506
	found_start = btrfs_header_bytenr(eb);
507 508 509 510 511 512 513 514 515 516 517 518
	/*
	 * 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);

519
	return csum_tree_block(fs_info, eb, 0);
520 521
}

522
static int check_tree_block_fsid(struct btrfs_fs_info *fs_info,
Y
Yan Zheng 已提交
523 524
				 struct extent_buffer *eb)
{
525
	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
526
	u8 fsid[BTRFS_FSID_SIZE];
Y
Yan Zheng 已提交
527 528
	int ret = 1;

529
	read_extent_buffer(eb, fsid, btrfs_header_fsid(), BTRFS_FSID_SIZE);
Y
Yan Zheng 已提交
530 531 532 533 534 535 536 537 538 539
	while (fs_devices) {
		if (!memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE)) {
			ret = 0;
			break;
		}
		fs_devices = fs_devices->seed;
	}
	return ret;
}

540 541 542
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)
543 544 545 546 547
{
	u64 found_start;
	int found_level;
	struct extent_buffer *eb;
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
548
	struct btrfs_fs_info *fs_info = root->fs_info;
549
	int ret = 0;
550
	int reads_done;
551 552 553

	if (!page->private)
		goto out;
C
Chris Mason 已提交
554

J
Josef Bacik 已提交
555
	eb = (struct extent_buffer *)page->private;
C
Chris Mason 已提交
556

557 558 559 560 561 562
	/* 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);
563 564
	if (!reads_done)
		goto err;
565

566
	eb->read_mirror = mirror;
567
	if (test_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags)) {
568 569 570 571
		ret = -EIO;
		goto err;
	}

572
	found_start = btrfs_header_bytenr(eb);
573
	if (found_start != eb->start) {
574 575
		btrfs_err_rl(fs_info, "bad tree block start %llu %llu",
			     found_start, eb->start);
576
		ret = -EIO;
577 578
		goto err;
	}
579 580 581
	if (check_tree_block_fsid(fs_info, eb)) {
		btrfs_err_rl(fs_info, "bad fsid on block %llu",
			     eb->start);
582 583 584
		ret = -EIO;
		goto err;
	}
585
	found_level = btrfs_header_level(eb);
586
	if (found_level >= BTRFS_MAX_LEVEL) {
587 588
		btrfs_err(fs_info, "bad tree block level %d",
			  (int)btrfs_header_level(eb));
589 590 591
		ret = -EIO;
		goto err;
	}
592

593 594
	btrfs_set_buffer_lockdep_class(btrfs_header_owner(eb),
				       eb, found_level);
595

596
	ret = csum_tree_block(fs_info, eb, 1);
597
	if (ret)
598 599 600 601 602 603 604
		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.
	 */
605
	if (found_level == 0 && btrfs_check_leaf_full(fs_info, eb)) {
606 607 608
		set_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
		ret = -EIO;
	}
609

610
	if (found_level > 0 && btrfs_check_node(fs_info, eb))
L
Liu Bo 已提交
611 612
		ret = -EIO;

613 614
	if (!ret)
		set_extent_buffer_uptodate(eb);
615
err:
616 617
	if (reads_done &&
	    test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
618
		btree_readahead_hook(eb, ret);
A
Arne Jansen 已提交
619

D
David Woodhouse 已提交
620 621 622 623 624 625 626
	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);
627
		clear_extent_buffer_uptodate(eb);
D
David Woodhouse 已提交
628
	}
629
	free_extent_buffer(eb);
630
out:
631
	return ret;
632 633
}

634
static int btree_io_failed_hook(struct page *page, int failed_mirror)
A
Arne Jansen 已提交
635 636 637
{
	struct extent_buffer *eb;

J
Josef Bacik 已提交
638
	eb = (struct extent_buffer *)page->private;
639
	set_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags);
640
	eb->read_mirror = failed_mirror;
D
David Woodhouse 已提交
641
	atomic_dec(&eb->io_pages);
642
	if (test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
643
		btree_readahead_hook(eb, -EIO);
A
Arne Jansen 已提交
644 645 646
	return -EIO;	/* we fixed nothing */
}

647
static void end_workqueue_bio(struct bio *bio)
648
{
649
	struct btrfs_end_io_wq *end_io_wq = bio->bi_private;
650
	struct btrfs_fs_info *fs_info;
651 652
	struct btrfs_workqueue *wq;
	btrfs_work_func_t func;
653 654

	fs_info = end_io_wq->info;
655
	end_io_wq->status = bio->bi_status;
656

M
Mike Christie 已提交
657
	if (bio_op(bio) == REQ_OP_WRITE) {
658 659 660 661 662 663 664 665 666 667 668 669 670
		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;
		}
671
	} else {
672 673 674 675 676
		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) {
677 678 679 680 681 682 683 684 685
			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;
		}
686
	}
687 688 689

	btrfs_init_work(&end_io_wq->work, func, end_workqueue_fn, NULL, NULL);
	btrfs_queue_work(wq, &end_io_wq->work);
690 691
}

692
blk_status_t btrfs_bio_wq_end_io(struct btrfs_fs_info *info, struct bio *bio,
693
			enum btrfs_wq_endio_type metadata)
694
{
695
	struct btrfs_end_io_wq *end_io_wq;
696

697
	end_io_wq = kmem_cache_alloc(btrfs_end_io_wq_cache, GFP_NOFS);
698
	if (!end_io_wq)
699
		return BLK_STS_RESOURCE;
700 701 702

	end_io_wq->private = bio->bi_private;
	end_io_wq->end_io = bio->bi_end_io;
703
	end_io_wq->info = info;
704
	end_io_wq->status = 0;
705
	end_io_wq->bio = bio;
706
	end_io_wq->metadata = metadata;
707 708 709

	bio->bi_private = end_io_wq;
	bio->bi_end_io = end_workqueue_bio;
710 711 712
	return 0;
}

C
Chris Mason 已提交
713 714 715
static void run_one_async_start(struct btrfs_work *work)
{
	struct async_submit_bio *async;
716
	blk_status_t ret;
C
Chris Mason 已提交
717 718

	async = container_of(work, struct  async_submit_bio, work);
719
	ret = async->submit_bio_start(async->private_data, async->bio,
720 721 722
				      async->mirror_num, async->bio_flags,
				      async->bio_offset);
	if (ret)
723
		async->status = ret;
C
Chris Mason 已提交
724 725 726
}

static void run_one_async_done(struct btrfs_work *work)
727 728 729 730
{
	struct async_submit_bio *async;

	async = container_of(work, struct  async_submit_bio, work);
731

732
	/* If an error occurred we just want to clean up the bio and move on */
733 734
	if (async->status) {
		async->bio->bi_status = async->status;
735
		bio_endio(async->bio);
736 737 738
		return;
	}

739
	async->submit_bio_done(async->private_data, async->bio, async->mirror_num,
740
			       async->bio_flags, async->bio_offset);
C
Chris Mason 已提交
741 742 743 744 745 746 747
}

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

	async = container_of(work, struct  async_submit_bio, work);
748 749 750
	kfree(async);
}

751 752 753 754 755
blk_status_t btrfs_wq_submit_bio(struct btrfs_fs_info *fs_info, struct bio *bio,
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset, void *private_data,
				 extent_submit_bio_hook_t *submit_bio_start,
				 extent_submit_bio_hook_t *submit_bio_done)
756 757 758 759 760
{
	struct async_submit_bio *async;

	async = kmalloc(sizeof(*async), GFP_NOFS);
	if (!async)
761
		return BLK_STS_RESOURCE;
762

763 764
	async->private_data = private_data;
	async->fs_info = fs_info;
765 766
	async->bio = bio;
	async->mirror_num = mirror_num;
C
Chris Mason 已提交
767 768 769
	async->submit_bio_start = submit_bio_start;
	async->submit_bio_done = submit_bio_done;

770
	btrfs_init_work(&async->work, btrfs_worker_helper, run_one_async_start,
771
			run_one_async_done, run_one_async_free);
C
Chris Mason 已提交
772

C
Chris Mason 已提交
773
	async->bio_flags = bio_flags;
774
	async->bio_offset = bio_offset;
775

776
	async->status = 0;
777

778
	if (op_is_sync(bio->bi_opf))
779
		btrfs_set_work_high_priority(&async->work);
780

781
	btrfs_queue_work(fs_info->workers, &async->work);
782 783 784
	return 0;
}

785
static blk_status_t btree_csum_one_bio(struct bio *bio)
786
{
787
	struct bio_vec *bvec;
788
	struct btrfs_root *root;
789
	int i, ret = 0;
790

791
	ASSERT(!bio_flagged(bio, BIO_CLONED));
792
	bio_for_each_segment_all(bvec, bio, i) {
793
		root = BTRFS_I(bvec->bv_page->mapping->host)->root;
794
		ret = csum_dirty_buffer(root->fs_info, bvec->bv_page);
795 796
		if (ret)
			break;
797
	}
798

799
	return errno_to_blk_status(ret);
800 801
}

802 803 804
static blk_status_t __btree_submit_bio_start(void *private_data, struct bio *bio,
					     int mirror_num, unsigned long bio_flags,
					     u64 bio_offset)
805
{
806 807
	/*
	 * when we're called for a write, we're already in the async
808
	 * submission context.  Just jump into btrfs_map_bio
809
	 */
810
	return btree_csum_one_bio(bio);
C
Chris Mason 已提交
811
}
812

813 814 815
static blk_status_t __btree_submit_bio_done(void *private_data, struct bio *bio,
					    int mirror_num, unsigned long bio_flags,
					    u64 bio_offset)
C
Chris Mason 已提交
816
{
817
	struct inode *inode = private_data;
818
	blk_status_t ret;
819

820
	/*
C
Chris Mason 已提交
821 822
	 * when we're called for a write, we're already in the async
	 * submission context.  Just jump into btrfs_map_bio
823
	 */
824
	ret = btrfs_map_bio(btrfs_sb(inode->i_sb), bio, mirror_num, 1);
825
	if (ret) {
826
		bio->bi_status = ret;
827 828
		bio_endio(bio);
	}
829
	return ret;
830 831
}

832
static int check_async_write(struct btrfs_inode *bi)
833
{
834 835
	if (atomic_read(&bi->sync_writers))
		return 0;
836
#ifdef CONFIG_X86
837
	if (static_cpu_has(X86_FEATURE_XMM4_2))
838 839 840 841 842
		return 0;
#endif
	return 1;
}

843 844 845
static blk_status_t btree_submit_bio_hook(void *private_data, struct bio *bio,
					  int mirror_num, unsigned long bio_flags,
					  u64 bio_offset)
846
{
847
	struct inode *inode = private_data;
848
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
849
	int async = check_async_write(BTRFS_I(inode));
850
	blk_status_t ret;
851

M
Mike Christie 已提交
852
	if (bio_op(bio) != REQ_OP_WRITE) {
C
Chris Mason 已提交
853 854 855 856
		/*
		 * called for a read, do the setup so that checksum validation
		 * can happen in the async kernel threads
		 */
857 858
		ret = btrfs_bio_wq_end_io(fs_info, bio,
					  BTRFS_WQ_ENDIO_METADATA);
859
		if (ret)
860
			goto out_w_error;
861
		ret = btrfs_map_bio(fs_info, bio, mirror_num, 0);
862 863 864
	} else if (!async) {
		ret = btree_csum_one_bio(bio);
		if (ret)
865
			goto out_w_error;
866
		ret = btrfs_map_bio(fs_info, bio, mirror_num, 0);
867 868 869 870 871
	} else {
		/*
		 * kthread helpers are used to submit writes so that
		 * checksumming can happen in parallel across all CPUs
		 */
872 873
		ret = btrfs_wq_submit_bio(fs_info, bio, mirror_num, 0,
					  bio_offset, private_data,
874 875
					  __btree_submit_bio_start,
					  __btree_submit_bio_done);
876
	}
877

878 879 880 881
	if (ret)
		goto out_w_error;
	return 0;

882
out_w_error:
883
	bio->bi_status = ret;
884
	bio_endio(bio);
885
	return ret;
886 887
}

J
Jan Beulich 已提交
888
#ifdef CONFIG_MIGRATION
889
static int btree_migratepage(struct address_space *mapping,
890 891
			struct page *newpage, struct page *page,
			enum migrate_mode mode)
892 893 894 895 896 897 898 899 900 901 902 903 904 905
{
	/*
	 * 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;
906
	return migrate_page(mapping, newpage, page, mode);
907
}
J
Jan Beulich 已提交
908
#endif
909

910 911 912 913

static int btree_writepages(struct address_space *mapping,
			    struct writeback_control *wbc)
{
914 915 916
	struct btrfs_fs_info *fs_info;
	int ret;

917
	if (wbc->sync_mode == WB_SYNC_NONE) {
918 919 920 921

		if (wbc->for_kupdate)
			return 0;

922
		fs_info = BTRFS_I(mapping->host)->root->fs_info;
923
		/* this is a bit racy, but that's ok */
924 925 926
		ret = percpu_counter_compare(&fs_info->dirty_metadata_bytes,
					     BTRFS_DIRTY_METADATA_THRESH);
		if (ret < 0)
927 928
			return 0;
	}
929
	return btree_write_cache_pages(mapping, wbc);
930 931
}

932
static int btree_readpage(struct file *file, struct page *page)
933
{
934 935
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
936
	return extent_read_full_page(tree, page, btree_get_extent, 0);
937
}
C
Chris Mason 已提交
938

939
static int btree_releasepage(struct page *page, gfp_t gfp_flags)
940
{
941
	if (PageWriteback(page) || PageDirty(page))
C
Chris Mason 已提交
942
		return 0;
943

944
	return try_release_extent_buffer(page);
945 946
}

947 948
static void btree_invalidatepage(struct page *page, unsigned int offset,
				 unsigned int length)
949
{
950 951
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
952 953
	extent_invalidatepage(tree, page, offset);
	btree_releasepage(page, GFP_NOFS);
954
	if (PagePrivate(page)) {
955 956 957
		btrfs_warn(BTRFS_I(page->mapping->host)->root->fs_info,
			   "page private not zero on page %llu",
			   (unsigned long long)page_offset(page));
958 959
		ClearPagePrivate(page);
		set_page_private(page, 0);
960
		put_page(page);
961
	}
962 963
}

964 965
static int btree_set_page_dirty(struct page *page)
{
966
#ifdef DEBUG
967 968 969 970 971 972 973 974
	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);
975
#endif
976 977 978
	return __set_page_dirty_nobuffers(page);
}

979
static const struct address_space_operations btree_aops = {
980
	.readpage	= btree_readpage,
981
	.writepages	= btree_writepages,
982 983
	.releasepage	= btree_releasepage,
	.invalidatepage = btree_invalidatepage,
984
#ifdef CONFIG_MIGRATION
985
	.migratepage	= btree_migratepage,
986
#endif
987
	.set_page_dirty = btree_set_page_dirty,
988 989
};

990
void readahead_tree_block(struct btrfs_fs_info *fs_info, u64 bytenr)
C
Chris Mason 已提交
991
{
992
	struct extent_buffer *buf = NULL;
993
	struct inode *btree_inode = fs_info->btree_inode;
C
Chris Mason 已提交
994

995
	buf = btrfs_find_create_tree_block(fs_info, bytenr);
996
	if (IS_ERR(buf))
997
		return;
998
	read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree,
999
				 buf, WAIT_NONE, 0);
1000
	free_extent_buffer(buf);
C
Chris Mason 已提交
1001 1002
}

1003
int reada_tree_block_flagged(struct btrfs_fs_info *fs_info, u64 bytenr,
1004 1005 1006
			 int mirror_num, struct extent_buffer **eb)
{
	struct extent_buffer *buf = NULL;
1007
	struct inode *btree_inode = fs_info->btree_inode;
1008 1009 1010
	struct extent_io_tree *io_tree = &BTRFS_I(btree_inode)->io_tree;
	int ret;

1011
	buf = btrfs_find_create_tree_block(fs_info, bytenr);
1012
	if (IS_ERR(buf))
1013 1014 1015 1016
		return 0;

	set_bit(EXTENT_BUFFER_READAHEAD, &buf->bflags);

1017
	ret = read_extent_buffer_pages(io_tree, buf, WAIT_PAGE_LOCK,
1018
				       mirror_num);
1019 1020 1021 1022 1023 1024 1025 1026
	if (ret) {
		free_extent_buffer(buf);
		return ret;
	}

	if (test_bit(EXTENT_BUFFER_CORRUPT, &buf->bflags)) {
		free_extent_buffer(buf);
		return -EIO;
1027
	} else if (extent_buffer_uptodate(buf)) {
1028 1029 1030 1031 1032 1033 1034
		*eb = buf;
	} else {
		free_extent_buffer(buf);
	}
	return 0;
}

1035 1036 1037
struct extent_buffer *btrfs_find_create_tree_block(
						struct btrfs_fs_info *fs_info,
						u64 bytenr)
1038
{
1039 1040 1041
	if (btrfs_is_testing(fs_info))
		return alloc_test_extent_buffer(fs_info, bytenr);
	return alloc_extent_buffer(fs_info, bytenr);
1042 1043 1044
}


1045 1046
int btrfs_write_tree_block(struct extent_buffer *buf)
{
1047
	return filemap_fdatawrite_range(buf->pages[0]->mapping, buf->start,
1048
					buf->start + buf->len - 1);
1049 1050
}

1051
void btrfs_wait_tree_block_writeback(struct extent_buffer *buf)
1052
{
1053 1054
	filemap_fdatawait_range(buf->pages[0]->mapping,
			        buf->start, buf->start + buf->len - 1);
1055 1056
}

1057
struct extent_buffer *read_tree_block(struct btrfs_fs_info *fs_info, u64 bytenr,
1058
				      u64 parent_transid)
1059 1060 1061 1062
{
	struct extent_buffer *buf = NULL;
	int ret;

1063
	buf = btrfs_find_create_tree_block(fs_info, bytenr);
1064 1065
	if (IS_ERR(buf))
		return buf;
1066

1067
	ret = btree_read_extent_buffer_pages(fs_info, buf, parent_transid);
1068 1069
	if (ret) {
		free_extent_buffer(buf);
1070
		return ERR_PTR(ret);
1071
	}
1072
	return buf;
1073

1074 1075
}

1076
void clean_tree_block(struct btrfs_fs_info *fs_info,
1077
		      struct extent_buffer *buf)
1078
{
1079
	if (btrfs_header_generation(buf) ==
1080
	    fs_info->running_transaction->transid) {
1081
		btrfs_assert_tree_locked(buf);
1082

1083
		if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &buf->bflags)) {
1084 1085 1086
			percpu_counter_add_batch(&fs_info->dirty_metadata_bytes,
						 -buf->len,
						 fs_info->dirty_metadata_batch);
1087 1088 1089 1090
			/* ugh, clear_extent_buffer_dirty needs to lock the page */
			btrfs_set_lock_blocking(buf);
			clear_extent_buffer_dirty(buf);
		}
1091
	}
1092 1093
}

1094 1095 1096 1097 1098 1099 1100 1101 1102
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);

1103
	ret = percpu_counter_init(&writers->counter, 0, GFP_KERNEL);
1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119
	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);
}

1120
static void __setup_root(struct btrfs_root *root, struct btrfs_fs_info *fs_info,
1121
			 u64 objectid)
1122
{
1123
	bool dummy = test_bit(BTRFS_FS_STATE_DUMMY_FS_INFO, &fs_info->fs_state);
C
Chris Mason 已提交
1124
	root->node = NULL;
1125
	root->commit_root = NULL;
1126
	root->state = 0;
1127
	root->orphan_cleanup_state = 0;
1128

1129 1130
	root->objectid = objectid;
	root->last_trans = 0;
1131
	root->highest_objectid = 0;
1132
	root->nr_delalloc_inodes = 0;
1133
	root->nr_ordered_extents = 0;
1134
	root->name = NULL;
1135
	root->inode_tree = RB_ROOT;
1136
	INIT_RADIX_TREE(&root->delayed_nodes_tree, GFP_ATOMIC);
1137
	root->block_rsv = NULL;
1138
	root->orphan_block_rsv = NULL;
1139 1140

	INIT_LIST_HEAD(&root->dirty_list);
1141
	INIT_LIST_HEAD(&root->root_list);
1142 1143
	INIT_LIST_HEAD(&root->delalloc_inodes);
	INIT_LIST_HEAD(&root->delalloc_root);
1144 1145
	INIT_LIST_HEAD(&root->ordered_extents);
	INIT_LIST_HEAD(&root->ordered_root);
1146 1147
	INIT_LIST_HEAD(&root->logged_list[0]);
	INIT_LIST_HEAD(&root->logged_list[1]);
1148
	spin_lock_init(&root->orphan_lock);
1149
	spin_lock_init(&root->inode_lock);
1150
	spin_lock_init(&root->delalloc_lock);
1151
	spin_lock_init(&root->ordered_extent_lock);
1152
	spin_lock_init(&root->accounting_lock);
1153 1154
	spin_lock_init(&root->log_extents_lock[0]);
	spin_lock_init(&root->log_extents_lock[1]);
1155
	mutex_init(&root->objectid_mutex);
1156
	mutex_init(&root->log_mutex);
1157
	mutex_init(&root->ordered_extent_mutex);
1158
	mutex_init(&root->delalloc_mutex);
Y
Yan Zheng 已提交
1159 1160 1161
	init_waitqueue_head(&root->log_writer_wait);
	init_waitqueue_head(&root->log_commit_wait[0]);
	init_waitqueue_head(&root->log_commit_wait[1]);
1162 1163
	INIT_LIST_HEAD(&root->log_ctxs[0]);
	INIT_LIST_HEAD(&root->log_ctxs[1]);
Y
Yan Zheng 已提交
1164 1165 1166
	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
1167
	atomic_set(&root->log_batch, 0);
1168
	atomic_set(&root->orphan_inodes, 0);
1169
	refcount_set(&root->refs, 1);
1170
	atomic_set(&root->will_be_snapshotted, 0);
1171
	atomic64_set(&root->qgroup_meta_rsv, 0);
Y
Yan Zheng 已提交
1172
	root->log_transid = 0;
1173
	root->log_transid_committed = -1;
1174
	root->last_log_commit = 0;
1175
	if (!dummy)
1176
		extent_io_tree_init(&root->dirty_log_pages, NULL);
C
Chris Mason 已提交
1177

1178 1179
	memset(&root->root_key, 0, sizeof(root->root_key));
	memset(&root->root_item, 0, sizeof(root->root_item));
1180
	memset(&root->defrag_progress, 0, sizeof(root->defrag_progress));
1181
	if (!dummy)
1182 1183 1184
		root->defrag_trans_start = fs_info->generation;
	else
		root->defrag_trans_start = 0;
1185
	root->root_key.objectid = objectid;
1186
	root->anon_dev = 0;
1187

1188
	spin_lock_init(&root->root_item_lock);
1189 1190
}

1191 1192
static struct btrfs_root *btrfs_alloc_root(struct btrfs_fs_info *fs_info,
		gfp_t flags)
A
Al Viro 已提交
1193
{
1194
	struct btrfs_root *root = kzalloc(sizeof(*root), flags);
A
Al Viro 已提交
1195 1196 1197 1198 1199
	if (root)
		root->fs_info = fs_info;
	return root;
}

1200 1201
#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
/* Should only be used by the testing infrastructure */
1202
struct btrfs_root *btrfs_alloc_dummy_root(struct btrfs_fs_info *fs_info)
1203 1204 1205
{
	struct btrfs_root *root;

1206 1207 1208 1209
	if (!fs_info)
		return ERR_PTR(-EINVAL);

	root = btrfs_alloc_root(fs_info, GFP_KERNEL);
1210 1211
	if (!root)
		return ERR_PTR(-ENOMEM);
1212

1213
	/* We don't use the stripesize in selftest, set it as sectorsize */
1214
	__setup_root(root, fs_info, BTRFS_ROOT_TREE_OBJECTID);
1215
	root->alloc_bytenr = 0;
1216 1217 1218 1219 1220

	return root;
}
#endif

1221 1222 1223 1224 1225 1226 1227 1228 1229
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;
1230
	uuid_le uuid = NULL_UUID_LE;
1231

1232
	root = btrfs_alloc_root(fs_info, GFP_KERNEL);
1233 1234 1235
	if (!root)
		return ERR_PTR(-ENOMEM);

1236
	__setup_root(root, fs_info, objectid);
1237 1238 1239 1240
	root->root_key.objectid = objectid;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = 0;

1241
	leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0);
1242 1243
	if (IS_ERR(leaf)) {
		ret = PTR_ERR(leaf);
1244
		leaf = NULL;
1245 1246 1247
		goto fail;
	}

1248
	memzero_extent_buffer(leaf, 0, sizeof(struct btrfs_header));
1249 1250 1251 1252 1253 1254
	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;

1255 1256
	write_extent_buffer_fsid(leaf, fs_info->fsid);
	write_extent_buffer_chunk_tree_uuid(leaf, fs_info->chunk_tree_uuid);
1257 1258 1259
	btrfs_mark_buffer_dirty(leaf);

	root->commit_root = btrfs_root_node(root);
1260
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
1261 1262 1263 1264 1265 1266 1267 1268 1269 1270

	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);
1271 1272
	if (is_fstree(objectid))
		uuid_le_gen(&uuid);
1273
	memcpy(root->root_item.uuid, uuid.b, BTRFS_UUID_SIZE);
1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284
	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);

1285 1286
	return root;

1287
fail:
1288 1289
	if (leaf) {
		btrfs_tree_unlock(leaf);
1290
		free_extent_buffer(root->commit_root);
1291 1292 1293
		free_extent_buffer(leaf);
	}
	kfree(root);
1294

1295
	return ERR_PTR(ret);
1296 1297
}

Y
Yan Zheng 已提交
1298 1299
static struct btrfs_root *alloc_log_tree(struct btrfs_trans_handle *trans,
					 struct btrfs_fs_info *fs_info)
1300 1301
{
	struct btrfs_root *root;
Y
Yan Zheng 已提交
1302
	struct extent_buffer *leaf;
1303

1304
	root = btrfs_alloc_root(fs_info, GFP_NOFS);
1305
	if (!root)
Y
Yan Zheng 已提交
1306
		return ERR_PTR(-ENOMEM);
1307

1308
	__setup_root(root, fs_info, BTRFS_TREE_LOG_OBJECTID);
1309 1310 1311 1312

	root->root_key.objectid = BTRFS_TREE_LOG_OBJECTID;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = BTRFS_TREE_LOG_OBJECTID;
1313

Y
Yan Zheng 已提交
1314
	/*
1315 1316
	 * DON'T set REF_COWS for log trees
	 *
Y
Yan Zheng 已提交
1317 1318 1319 1320 1321
	 * 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).
	 */
1322

1323 1324
	leaf = btrfs_alloc_tree_block(trans, root, 0, BTRFS_TREE_LOG_OBJECTID,
			NULL, 0, 0, 0);
Y
Yan Zheng 已提交
1325 1326 1327 1328
	if (IS_ERR(leaf)) {
		kfree(root);
		return ERR_CAST(leaf);
	}
1329

1330
	memzero_extent_buffer(leaf, 0, sizeof(struct btrfs_header));
1331 1332 1333 1334
	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
1335
	root->node = leaf;
1336

1337
	write_extent_buffer_fsid(root->node, fs_info->fsid);
1338 1339
	btrfs_mark_buffer_dirty(root->node);
	btrfs_tree_unlock(root->node);
Y
Yan Zheng 已提交
1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358
	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)
{
1359
	struct btrfs_fs_info *fs_info = root->fs_info;
Y
Yan Zheng 已提交
1360 1361 1362
	struct btrfs_root *log_root;
	struct btrfs_inode_item *inode_item;

1363
	log_root = alloc_log_tree(trans, fs_info);
Y
Yan Zheng 已提交
1364 1365 1366 1367 1368 1369 1370
	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;
1371 1372 1373
	btrfs_set_stack_inode_generation(inode_item, 1);
	btrfs_set_stack_inode_size(inode_item, 3);
	btrfs_set_stack_inode_nlink(inode_item, 1);
1374
	btrfs_set_stack_inode_nbytes(inode_item,
1375
				     fs_info->nodesize);
1376
	btrfs_set_stack_inode_mode(inode_item, S_IFDIR | 0755);
Y
Yan Zheng 已提交
1377

1378
	btrfs_set_root_node(&log_root->root_item, log_root->node);
Y
Yan Zheng 已提交
1379 1380 1381 1382

	WARN_ON(root->log_root);
	root->log_root = log_root;
	root->log_transid = 0;
1383
	root->log_transid_committed = -1;
1384
	root->last_log_commit = 0;
1385 1386 1387
	return 0;
}

1388 1389
static struct btrfs_root *btrfs_read_tree_root(struct btrfs_root *tree_root,
					       struct btrfs_key *key)
1390 1391 1392
{
	struct btrfs_root *root;
	struct btrfs_fs_info *fs_info = tree_root->fs_info;
1393
	struct btrfs_path *path;
1394
	u64 generation;
1395
	int ret;
1396

1397 1398
	path = btrfs_alloc_path();
	if (!path)
1399
		return ERR_PTR(-ENOMEM);
1400

1401
	root = btrfs_alloc_root(fs_info, GFP_NOFS);
1402 1403 1404
	if (!root) {
		ret = -ENOMEM;
		goto alloc_fail;
1405 1406
	}

1407
	__setup_root(root, fs_info, key->objectid);
1408

1409 1410
	ret = btrfs_find_root(tree_root, key, path,
			      &root->root_item, &root->root_key);
1411
	if (ret) {
1412 1413
		if (ret > 0)
			ret = -ENOENT;
1414
		goto find_fail;
1415
	}
1416

1417
	generation = btrfs_root_generation(&root->root_item);
1418 1419
	root->node = read_tree_block(fs_info,
				     btrfs_root_bytenr(&root->root_item),
1420
				     generation);
1421 1422
	if (IS_ERR(root->node)) {
		ret = PTR_ERR(root->node);
1423 1424 1425
		goto find_fail;
	} else if (!btrfs_buffer_uptodate(root->node, generation, 0)) {
		ret = -EIO;
1426 1427
		free_extent_buffer(root->node);
		goto find_fail;
1428
	}
1429
	root->commit_root = btrfs_root_node(root);
1430
out:
1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450
	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) {
1451
		set_bit(BTRFS_ROOT_REF_COWS, &root->state);
1452 1453
		btrfs_check_and_init_root_item(&root->root_item);
	}
1454

1455 1456 1457
	return root;
}

1458 1459 1460
int btrfs_init_fs_root(struct btrfs_root *root)
{
	int ret;
1461
	struct btrfs_subvolume_writers *writers;
1462 1463 1464 1465 1466 1467 1468 1469 1470

	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;
	}

1471 1472 1473 1474 1475 1476 1477
	writers = btrfs_alloc_subvolume_writers();
	if (IS_ERR(writers)) {
		ret = PTR_ERR(writers);
		goto fail;
	}
	root->subv_writers = writers;

1478
	btrfs_init_free_ino_ctl(root);
1479 1480
	spin_lock_init(&root->ino_cache_lock);
	init_waitqueue_head(&root->ino_cache_wait);
1481 1482 1483

	ret = get_anon_bdev(&root->anon_dev);
	if (ret)
L
Liu Bo 已提交
1484
		goto fail;
1485 1486 1487 1488 1489 1490

	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
1491
		goto fail;
1492 1493 1494 1495 1496 1497
	}

	ASSERT(root->highest_objectid <= BTRFS_LAST_FREE_OBJECTID);

	mutex_unlock(&root->objectid_mutex);

1498 1499
	return 0;
fail:
L
Liu Bo 已提交
1500
	/* the caller is responsible to call free_fs_root */
1501 1502 1503
	return ret;
}

1504 1505
struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info,
					u64 root_id)
1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520
{
	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;

1521
	ret = radix_tree_preload(GFP_NOFS);
1522 1523 1524 1525 1526 1527 1528 1529
	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)
1530
		set_bit(BTRFS_ROOT_IN_RADIX, &root->state);
1531 1532 1533 1534 1535 1536
	spin_unlock(&fs_info->fs_roots_radix_lock);
	radix_tree_preload_end();

	return ret;
}

1537 1538 1539
struct btrfs_root *btrfs_get_fs_root(struct btrfs_fs_info *fs_info,
				     struct btrfs_key *location,
				     bool check_ref)
1540 1541
{
	struct btrfs_root *root;
1542
	struct btrfs_path *path;
1543
	struct btrfs_key key;
1544 1545
	int ret;

1546 1547 1548 1549
	if (location->objectid == BTRFS_ROOT_TREE_OBJECTID)
		return fs_info->tree_root;
	if (location->objectid == BTRFS_EXTENT_TREE_OBJECTID)
		return fs_info->extent_root;
1550 1551 1552 1553
	if (location->objectid == BTRFS_CHUNK_TREE_OBJECTID)
		return fs_info->chunk_root;
	if (location->objectid == BTRFS_DEV_TREE_OBJECTID)
		return fs_info->dev_root;
1554 1555
	if (location->objectid == BTRFS_CSUM_TREE_OBJECTID)
		return fs_info->csum_root;
1556 1557 1558
	if (location->objectid == BTRFS_QUOTA_TREE_OBJECTID)
		return fs_info->quota_root ? fs_info->quota_root :
					     ERR_PTR(-ENOENT);
1559 1560 1561
	if (location->objectid == BTRFS_UUID_TREE_OBJECTID)
		return fs_info->uuid_root ? fs_info->uuid_root :
					    ERR_PTR(-ENOENT);
1562 1563 1564
	if (location->objectid == BTRFS_FREE_SPACE_TREE_OBJECTID)
		return fs_info->free_space_root ? fs_info->free_space_root :
						  ERR_PTR(-ENOENT);
1565
again:
1566
	root = btrfs_lookup_fs_root(fs_info, location->objectid);
1567
	if (root) {
1568
		if (check_ref && btrfs_root_refs(&root->root_item) == 0)
1569
			return ERR_PTR(-ENOENT);
1570
		return root;
1571
	}
1572

1573
	root = btrfs_read_fs_root(fs_info->tree_root, location);
1574 1575
	if (IS_ERR(root))
		return root;
1576

1577
	if (check_ref && btrfs_root_refs(&root->root_item) == 0) {
1578
		ret = -ENOENT;
1579
		goto fail;
1580
	}
1581

1582
	ret = btrfs_init_fs_root(root);
1583 1584
	if (ret)
		goto fail;
1585

1586 1587 1588 1589 1590
	path = btrfs_alloc_path();
	if (!path) {
		ret = -ENOMEM;
		goto fail;
	}
1591 1592 1593 1594 1595
	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);
1596
	btrfs_free_path(path);
1597 1598 1599
	if (ret < 0)
		goto fail;
	if (ret == 0)
1600
		set_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &root->state);
1601

1602
	ret = btrfs_insert_fs_root(fs_info, root);
1603
	if (ret) {
1604 1605 1606 1607 1608
		if (ret == -EEXIST) {
			free_fs_root(root);
			goto again;
		}
		goto fail;
1609
	}
1610
	return root;
1611 1612 1613
fail:
	free_fs_root(root);
	return ERR_PTR(ret);
1614 1615
}

C
Chris Mason 已提交
1616 1617 1618 1619 1620 1621
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
1622

1623 1624
	rcu_read_lock();
	list_for_each_entry_rcu(device, &info->fs_devices->devices, dev_list) {
1625 1626
		if (!device->bdev)
			continue;
1627
		bdi = device->bdev->bd_bdi;
1628
		if (bdi_congested(bdi, bdi_bits)) {
C
Chris Mason 已提交
1629 1630 1631 1632
			ret = 1;
			break;
		}
	}
1633
	rcu_read_unlock();
C
Chris Mason 已提交
1634 1635 1636
	return ret;
}

1637 1638 1639 1640 1641
/*
 * 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)
1642 1643
{
	struct bio *bio;
1644
	struct btrfs_end_io_wq *end_io_wq;
1645

1646
	end_io_wq = container_of(work, struct btrfs_end_io_wq, work);
1647
	bio = end_io_wq->bio;
1648

1649
	bio->bi_status = end_io_wq->status;
1650 1651
	bio->bi_private = end_io_wq->private;
	bio->bi_end_io = end_io_wq->end_io;
1652
	kmem_cache_free(btrfs_end_io_wq_cache, end_io_wq);
1653
	bio_endio(bio);
1654 1655
}

1656 1657 1658
static int cleaner_kthread(void *arg)
{
	struct btrfs_root *root = arg;
1659
	struct btrfs_fs_info *fs_info = root->fs_info;
1660
	int again;
1661
	struct btrfs_trans_handle *trans;
1662 1663

	do {
1664
		again = 0;
1665

1666
		/* Make the cleaner go to sleep early. */
1667
		if (btrfs_need_cleaner_sleep(fs_info))
1668 1669
			goto sleep;

1670 1671 1672 1673
		/*
		 * Do not do anything if we might cause open_ctree() to block
		 * before we have finished mounting the filesystem.
		 */
1674
		if (!test_bit(BTRFS_FS_OPEN, &fs_info->flags))
1675 1676
			goto sleep;

1677
		if (!mutex_trylock(&fs_info->cleaner_mutex))
1678 1679
			goto sleep;

1680 1681 1682 1683
		/*
		 * Avoid the problem that we change the status of the fs
		 * during the above check and trylock.
		 */
1684
		if (btrfs_need_cleaner_sleep(fs_info)) {
1685
			mutex_unlock(&fs_info->cleaner_mutex);
1686
			goto sleep;
1687
		}
1688

1689
		mutex_lock(&fs_info->cleaner_delayed_iput_mutex);
1690
		btrfs_run_delayed_iputs(fs_info);
1691
		mutex_unlock(&fs_info->cleaner_delayed_iput_mutex);
1692

1693
		again = btrfs_clean_one_deleted_snapshot(root);
1694
		mutex_unlock(&fs_info->cleaner_mutex);
1695 1696

		/*
1697 1698
		 * The defragger has dealt with the R/O remount and umount,
		 * needn't do anything special here.
1699
		 */
1700
		btrfs_run_defrag_inodes(fs_info);
1701 1702 1703 1704 1705 1706 1707 1708 1709

		/*
		 * 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.
		 */
1710
		btrfs_delete_unused_bgs(fs_info);
1711
sleep:
1712
		if (!again) {
1713
			set_current_state(TASK_INTERRUPTIBLE);
1714 1715
			if (!kthread_should_stop())
				schedule();
1716 1717 1718
			__set_current_state(TASK_RUNNING);
		}
	} while (!kthread_should_stop());
1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733

	/*
	 * 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)
1734
			btrfs_err(fs_info,
1735 1736 1737 1738 1739
				  "cleaner transaction attach returned %ld",
				  PTR_ERR(trans));
	} else {
		int ret;

1740
		ret = btrfs_commit_transaction(trans);
1741
		if (ret)
1742
			btrfs_err(fs_info,
1743 1744 1745 1746
				  "cleaner open transaction commit returned %d",
				  ret);
	}

1747 1748 1749 1750 1751 1752
	return 0;
}

static int transaction_kthread(void *arg)
{
	struct btrfs_root *root = arg;
1753
	struct btrfs_fs_info *fs_info = root->fs_info;
1754 1755
	struct btrfs_trans_handle *trans;
	struct btrfs_transaction *cur;
1756
	u64 transid;
1757 1758
	unsigned long now;
	unsigned long delay;
1759
	bool cannot_commit;
1760 1761

	do {
1762
		cannot_commit = false;
1763 1764
		delay = HZ * fs_info->commit_interval;
		mutex_lock(&fs_info->transaction_kthread_mutex);
1765

1766 1767
		spin_lock(&fs_info->trans_lock);
		cur = fs_info->running_transaction;
1768
		if (!cur) {
1769
			spin_unlock(&fs_info->trans_lock);
1770 1771
			goto sleep;
		}
Y
Yan Zheng 已提交
1772

1773
		now = get_seconds();
1774
		if (cur->state < TRANS_STATE_BLOCKED &&
1775
		    (now < cur->start_time ||
1776 1777
		     now - cur->start_time < fs_info->commit_interval)) {
			spin_unlock(&fs_info->trans_lock);
1778 1779 1780
			delay = HZ * 5;
			goto sleep;
		}
1781
		transid = cur->transid;
1782
		spin_unlock(&fs_info->trans_lock);
1783

1784
		/* If the file system is aborted, this will always fail. */
1785
		trans = btrfs_attach_transaction(root);
1786
		if (IS_ERR(trans)) {
1787 1788
			if (PTR_ERR(trans) != -ENOENT)
				cannot_commit = true;
1789
			goto sleep;
1790
		}
1791
		if (transid == trans->transid) {
1792
			btrfs_commit_transaction(trans);
1793
		} else {
1794
			btrfs_end_transaction(trans);
1795
		}
1796
sleep:
1797 1798
		wake_up_process(fs_info->cleaner_kthread);
		mutex_unlock(&fs_info->transaction_kthread_mutex);
1799

J
Josef Bacik 已提交
1800
		if (unlikely(test_bit(BTRFS_FS_STATE_ERROR,
1801
				      &fs_info->fs_state)))
1802
			btrfs_cleanup_transaction(fs_info);
1803
		if (!kthread_should_stop() &&
1804
				(!btrfs_transaction_blocked(fs_info) ||
1805
				 cannot_commit))
1806
			schedule_timeout_interruptible(delay);
1807 1808 1809 1810
	} while (!kthread_should_stop());
	return 0;
}

C
Chris Mason 已提交
1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916
/*
 * 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));

1917 1918 1919 1920 1921 1922 1923 1924
	/*
	 * 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
1925
			       btrfs_header_generation(info->fs_root->node));
1926
		btrfs_set_backup_fs_root_level(root_backup,
C
Chris Mason 已提交
1927
			       btrfs_header_level(info->fs_root->node));
1928
	}
C
Chris Mason 已提交
1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

	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
2010 2011 2012
/* helper to cleanup workers */
static void btrfs_stop_all_workers(struct btrfs_fs_info *fs_info)
{
2013
	btrfs_destroy_workqueue(fs_info->fixup_workers);
2014
	btrfs_destroy_workqueue(fs_info->delalloc_workers);
2015
	btrfs_destroy_workqueue(fs_info->workers);
2016 2017
	btrfs_destroy_workqueue(fs_info->endio_workers);
	btrfs_destroy_workqueue(fs_info->endio_raid56_workers);
2018
	btrfs_destroy_workqueue(fs_info->endio_repair_workers);
2019
	btrfs_destroy_workqueue(fs_info->rmw_workers);
2020 2021
	btrfs_destroy_workqueue(fs_info->endio_write_workers);
	btrfs_destroy_workqueue(fs_info->endio_freespace_worker);
2022
	btrfs_destroy_workqueue(fs_info->submit_workers);
2023
	btrfs_destroy_workqueue(fs_info->delayed_workers);
2024
	btrfs_destroy_workqueue(fs_info->caching_workers);
2025
	btrfs_destroy_workqueue(fs_info->readahead_workers);
2026
	btrfs_destroy_workqueue(fs_info->flush_workers);
2027
	btrfs_destroy_workqueue(fs_info->qgroup_rescan_workers);
C
Chris Mason 已提交
2028
	btrfs_destroy_workqueue(fs_info->extent_workers);
2029 2030 2031 2032 2033 2034 2035
	/*
	 * Now that all other work queues are destroyed, we can safely destroy
	 * the queues used for metadata I/O, since tasks from those other work
	 * queues can do metadata I/O operations.
	 */
	btrfs_destroy_workqueue(fs_info->endio_meta_workers);
	btrfs_destroy_workqueue(fs_info->endio_meta_write_workers);
L
Liu Bo 已提交
2036 2037
}

2038 2039 2040 2041 2042 2043 2044 2045 2046 2047
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
2048 2049 2050
/* helper to cleanup tree roots */
static void free_root_pointers(struct btrfs_fs_info *info, int chunk_root)
{
2051
	free_root_extent_buffers(info->tree_root);
2052

2053 2054 2055 2056 2057 2058 2059
	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);
2060
	free_root_extent_buffers(info->free_space_root);
C
Chris Mason 已提交
2061 2062
}

2063
void btrfs_free_fs_roots(struct btrfs_fs_info *fs_info)
2064 2065 2066 2067 2068 2069 2070 2071 2072 2073
{
	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);

2074
		if (test_bit(BTRFS_ROOT_IN_RADIX, &gang[0]->state)) {
2075
			btrfs_drop_and_free_fs_root(fs_info, gang[0]);
2076 2077 2078
		} else {
			free_extent_buffer(gang[0]->node);
			free_extent_buffer(gang[0]->commit_root);
2079
			btrfs_put_fs_root(gang[0]);
2080 2081 2082 2083 2084 2085 2086 2087 2088 2089
		}
	}

	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++)
2090
			btrfs_drop_and_free_fs_root(fs_info, gang[i]);
2091
	}
2092 2093 2094

	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
		btrfs_free_log_root_tree(NULL, fs_info);
2095
		btrfs_destroy_pinned_extent(fs_info, fs_info->pinned_extents);
2096
	}
2097
}
C
Chris Mason 已提交
2098

2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109
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;
}

2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120
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);
}

2121
static void btrfs_init_btree_inode(struct btrfs_fs_info *fs_info)
2122
{
2123 2124 2125 2126
	struct inode *inode = fs_info->btree_inode;

	inode->i_ino = BTRFS_BTREE_INODE_OBJECTID;
	set_nlink(inode, 1);
2127 2128 2129 2130 2131
	/*
	 * 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
	 */
2132 2133
	inode->i_size = OFFSET_MAX;
	inode->i_mapping->a_ops = &btree_aops;
2134

2135
	RB_CLEAR_NODE(&BTRFS_I(inode)->rb_node);
2136
	extent_io_tree_init(&BTRFS_I(inode)->io_tree, inode);
2137 2138
	BTRFS_I(inode)->io_tree.track_uptodate = 0;
	extent_map_tree_init(&BTRFS_I(inode)->extent_tree);
2139

2140
	BTRFS_I(inode)->io_tree.ops = &btree_extent_io_ops;
2141

2142 2143 2144 2145
	BTRFS_I(inode)->root = fs_info->tree_root;
	memset(&BTRFS_I(inode)->location, 0, sizeof(struct btrfs_key));
	set_bit(BTRFS_INODE_DUMMY, &BTRFS_I(inode)->runtime_flags);
	btrfs_insert_inode_hash(inode);
2146 2147
}

2148 2149 2150 2151 2152
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);
2153 2154 2155
	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);
2156
	init_waitqueue_head(&fs_info->replace_wait);
2157
	init_waitqueue_head(&fs_info->dev_replace.read_lock_wq);
2158 2159
}

2160 2161 2162 2163 2164 2165 2166 2167 2168
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;
2169
	fs_info->qgroup_rescan_running = false;
2170 2171 2172
	mutex_init(&fs_info->qgroup_rescan_lock);
}

2173 2174 2175
static int btrfs_init_workqueues(struct btrfs_fs_info *fs_info,
		struct btrfs_fs_devices *fs_devices)
{
2176
	u32 max_active = fs_info->thread_pool_size;
2177
	unsigned int flags = WQ_MEM_RECLAIM | WQ_FREEZABLE | WQ_UNBOUND;
2178 2179

	fs_info->workers =
2180 2181
		btrfs_alloc_workqueue(fs_info, "worker",
				      flags | WQ_HIGHPRI, max_active, 16);
2182 2183

	fs_info->delalloc_workers =
2184 2185
		btrfs_alloc_workqueue(fs_info, "delalloc",
				      flags, max_active, 2);
2186 2187

	fs_info->flush_workers =
2188 2189
		btrfs_alloc_workqueue(fs_info, "flush_delalloc",
				      flags, max_active, 0);
2190 2191

	fs_info->caching_workers =
2192
		btrfs_alloc_workqueue(fs_info, "cache", flags, max_active, 0);
2193 2194 2195 2196 2197 2198 2199

	/*
	 * 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 =
2200
		btrfs_alloc_workqueue(fs_info, "submit", flags,
2201 2202 2203 2204
				      min_t(u64, fs_devices->num_devices,
					    max_active), 64);

	fs_info->fixup_workers =
2205
		btrfs_alloc_workqueue(fs_info, "fixup", flags, 1, 0);
2206 2207 2208 2209 2210 2211

	/*
	 * endios are largely parallel and should have a very
	 * low idle thresh
	 */
	fs_info->endio_workers =
2212
		btrfs_alloc_workqueue(fs_info, "endio", flags, max_active, 4);
2213
	fs_info->endio_meta_workers =
2214 2215
		btrfs_alloc_workqueue(fs_info, "endio-meta", flags,
				      max_active, 4);
2216
	fs_info->endio_meta_write_workers =
2217 2218
		btrfs_alloc_workqueue(fs_info, "endio-meta-write", flags,
				      max_active, 2);
2219
	fs_info->endio_raid56_workers =
2220 2221
		btrfs_alloc_workqueue(fs_info, "endio-raid56", flags,
				      max_active, 4);
2222
	fs_info->endio_repair_workers =
2223
		btrfs_alloc_workqueue(fs_info, "endio-repair", flags, 1, 0);
2224
	fs_info->rmw_workers =
2225
		btrfs_alloc_workqueue(fs_info, "rmw", flags, max_active, 2);
2226
	fs_info->endio_write_workers =
2227 2228
		btrfs_alloc_workqueue(fs_info, "endio-write", flags,
				      max_active, 2);
2229
	fs_info->endio_freespace_worker =
2230 2231
		btrfs_alloc_workqueue(fs_info, "freespace-write", flags,
				      max_active, 0);
2232
	fs_info->delayed_workers =
2233 2234
		btrfs_alloc_workqueue(fs_info, "delayed-meta", flags,
				      max_active, 0);
2235
	fs_info->readahead_workers =
2236 2237
		btrfs_alloc_workqueue(fs_info, "readahead", flags,
				      max_active, 2);
2238
	fs_info->qgroup_rescan_workers =
2239
		btrfs_alloc_workqueue(fs_info, "qgroup-rescan", flags, 1, 0);
2240
	fs_info->extent_workers =
2241
		btrfs_alloc_workqueue(fs_info, "extent-refs", flags,
2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261
				      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;
}

2262 2263 2264 2265 2266 2267 2268 2269 2270
static int btrfs_replay_log(struct btrfs_fs_info *fs_info,
			    struct btrfs_fs_devices *fs_devices)
{
	int ret;
	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) {
2271
		btrfs_warn(fs_info, "log replay required on RO media");
2272 2273 2274
		return -EIO;
	}

2275
	log_tree_root = btrfs_alloc_root(fs_info, GFP_KERNEL);
2276 2277 2278
	if (!log_tree_root)
		return -ENOMEM;

2279
	__setup_root(log_tree_root, fs_info, BTRFS_TREE_LOG_OBJECTID);
2280

2281 2282
	log_tree_root->node = read_tree_block(fs_info, bytenr,
					      fs_info->generation + 1);
2283
	if (IS_ERR(log_tree_root->node)) {
2284
		btrfs_warn(fs_info, "failed to read log tree");
2285
		ret = PTR_ERR(log_tree_root->node);
2286
		kfree(log_tree_root);
2287
		return ret;
2288
	} else if (!extent_buffer_uptodate(log_tree_root->node)) {
2289
		btrfs_err(fs_info, "failed to read log tree");
2290 2291 2292 2293 2294 2295 2296
		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) {
2297 2298
		btrfs_handle_fs_error(fs_info, ret,
				      "Failed to recover log tree");
2299 2300 2301 2302 2303
		free_extent_buffer(log_tree_root->node);
		kfree(log_tree_root);
		return ret;
	}

2304
	if (sb_rdonly(fs_info->sb)) {
2305
		ret = btrfs_commit_super(fs_info);
2306 2307 2308 2309 2310 2311 2312
		if (ret)
			return ret;
	}

	return 0;
}

2313
static int btrfs_read_roots(struct btrfs_fs_info *fs_info)
2314
{
2315
	struct btrfs_root *tree_root = fs_info->tree_root;
2316
	struct btrfs_root *root;
2317 2318 2319
	struct btrfs_key location;
	int ret;

2320 2321
	BUG_ON(!fs_info->tree_root);

2322 2323 2324 2325
	location.objectid = BTRFS_EXTENT_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
	location.offset = 0;

2326 2327 2328 2329 2330
	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;
2331 2332

	location.objectid = BTRFS_DEV_TREE_OBJECTID;
2333 2334 2335 2336 2337
	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;
2338 2339 2340
	btrfs_init_devices_late(fs_info);

	location.objectid = BTRFS_CSUM_TREE_OBJECTID;
2341 2342 2343 2344 2345
	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;
2346 2347

	location.objectid = BTRFS_QUOTA_TREE_OBJECTID;
2348 2349 2350
	root = btrfs_read_tree_root(tree_root, &location);
	if (!IS_ERR(root)) {
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
2351
		set_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags);
2352
		fs_info->quota_root = root;
2353 2354 2355
	}

	location.objectid = BTRFS_UUID_TREE_OBJECTID;
2356 2357 2358
	root = btrfs_read_tree_root(tree_root, &location);
	if (IS_ERR(root)) {
		ret = PTR_ERR(root);
2359 2360 2361
		if (ret != -ENOENT)
			return ret;
	} else {
2362 2363
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
		fs_info->uuid_root = root;
2364 2365
	}

2366 2367 2368 2369 2370 2371 2372 2373 2374
	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;
	}

2375 2376 2377
	return 0;
}

A
Al Viro 已提交
2378 2379 2380
int open_ctree(struct super_block *sb,
	       struct btrfs_fs_devices *fs_devices,
	       char *options)
2381
{
2382 2383
	u32 sectorsize;
	u32 nodesize;
2384
	u32 stripesize;
2385
	u64 generation;
2386
	u64 features;
2387
	struct btrfs_key location;
2388
	struct buffer_head *bh;
2389
	struct btrfs_super_block *disk_super;
2390
	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
2391
	struct btrfs_root *tree_root;
2392
	struct btrfs_root *chunk_root;
2393
	int ret;
2394
	int err = -EINVAL;
C
Chris Mason 已提交
2395 2396
	int num_backups_tried = 0;
	int backup_index = 0;
2397
	u32 max_active;
2398
	int clear_free_space_tree = 0;
2399

2400 2401
	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);
2402
	if (!tree_root || !chunk_root) {
C
Chris Mason 已提交
2403 2404 2405
		err = -ENOMEM;
		goto fail;
	}
2406 2407 2408 2409 2410 2411 2412

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

2413
	ret = percpu_counter_init(&fs_info->dirty_metadata_bytes, 0, GFP_KERNEL);
2414 2415
	if (ret) {
		err = ret;
2416
		goto fail_srcu;
2417
	}
2418
	fs_info->dirty_metadata_batch = PAGE_SIZE *
2419 2420
					(1 + ilog2(nr_cpu_ids));

2421
	ret = percpu_counter_init(&fs_info->delalloc_bytes, 0, GFP_KERNEL);
2422 2423 2424 2425 2426
	if (ret) {
		err = ret;
		goto fail_dirty_metadata_bytes;
	}

2427
	ret = percpu_counter_init(&fs_info->bio_counter, 0, GFP_KERNEL);
2428 2429 2430 2431 2432
	if (ret) {
		err = ret;
		goto fail_delalloc_bytes;
	}

2433
	INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC);
2434
	INIT_RADIX_TREE(&fs_info->buffer_radix, GFP_ATOMIC);
C
Chris Mason 已提交
2435
	INIT_LIST_HEAD(&fs_info->trans_list);
2436
	INIT_LIST_HEAD(&fs_info->dead_roots);
Y
Yan, Zheng 已提交
2437
	INIT_LIST_HEAD(&fs_info->delayed_iputs);
2438
	INIT_LIST_HEAD(&fs_info->delalloc_roots);
2439
	INIT_LIST_HEAD(&fs_info->caching_block_groups);
2440
	spin_lock_init(&fs_info->delalloc_root_lock);
J
Josef Bacik 已提交
2441
	spin_lock_init(&fs_info->trans_lock);
2442
	spin_lock_init(&fs_info->fs_roots_radix_lock);
Y
Yan, Zheng 已提交
2443
	spin_lock_init(&fs_info->delayed_iput_lock);
C
Chris Mason 已提交
2444
	spin_lock_init(&fs_info->defrag_inodes_lock);
J
Jan Schmidt 已提交
2445
	spin_lock_init(&fs_info->tree_mod_seq_lock);
2446
	spin_lock_init(&fs_info->super_lock);
J
Josef Bacik 已提交
2447
	spin_lock_init(&fs_info->qgroup_op_lock);
2448
	spin_lock_init(&fs_info->buffer_lock);
2449
	spin_lock_init(&fs_info->unused_bgs_lock);
J
Jan Schmidt 已提交
2450
	rwlock_init(&fs_info->tree_mod_log_lock);
2451
	mutex_init(&fs_info->unused_bg_unpin_mutex);
2452
	mutex_init(&fs_info->delete_unused_bgs_mutex);
C
Chris Mason 已提交
2453
	mutex_init(&fs_info->reloc_mutex);
2454
	mutex_init(&fs_info->delalloc_root_mutex);
2455
	mutex_init(&fs_info->cleaner_delayed_iput_mutex);
2456
	seqlock_init(&fs_info->profiles_lock);
2457

2458
	INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots);
2459
	INIT_LIST_HEAD(&fs_info->space_info);
J
Jan Schmidt 已提交
2460
	INIT_LIST_HEAD(&fs_info->tree_mod_seq_list);
2461
	INIT_LIST_HEAD(&fs_info->unused_bgs);
2462
	btrfs_mapping_init(&fs_info->mapping_tree);
2463 2464 2465 2466 2467 2468 2469
	btrfs_init_block_rsv(&fs_info->global_block_rsv,
			     BTRFS_BLOCK_RSV_GLOBAL);
	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);
2470
	atomic_set(&fs_info->async_delalloc_pages, 0);
C
Chris Mason 已提交
2471
	atomic_set(&fs_info->defrag_running, 0);
J
Josef Bacik 已提交
2472
	atomic_set(&fs_info->qgroup_op_seq, 0);
Z
Zhao Lei 已提交
2473
	atomic_set(&fs_info->reada_works_cnt, 0);
2474
	atomic64_set(&fs_info->tree_mod_seq, 0);
C
Chris Mason 已提交
2475
	fs_info->sb = sb;
2476
	fs_info->max_inline = BTRFS_DEFAULT_MAX_INLINE;
J
Josef Bacik 已提交
2477
	fs_info->metadata_ratio = 0;
C
Chris Mason 已提交
2478
	fs_info->defrag_inodes = RB_ROOT;
2479
	atomic64_set(&fs_info->free_chunk_space, 0);
J
Jan Schmidt 已提交
2480
	fs_info->tree_mod_log = RB_ROOT;
2481
	fs_info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
2482
	fs_info->avg_delayed_ref_runtime = NSEC_PER_SEC >> 6; /* div by 64 */
2483
	/* readahead state */
2484
	INIT_RADIX_TREE(&fs_info->reada_tree, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
2485
	spin_lock_init(&fs_info->reada_lock);
J
Josef Bacik 已提交
2486
	btrfs_init_ref_verify(fs_info);
C
Chris Mason 已提交
2487

2488 2489
	fs_info->thread_pool_size = min_t(unsigned long,
					  num_online_cpus() + 2, 8);
2490

2491 2492
	INIT_LIST_HEAD(&fs_info->ordered_roots);
	spin_lock_init(&fs_info->ordered_root_lock);
2493 2494 2495 2496 2497 2498 2499 2500

	fs_info->btree_inode = new_inode(sb);
	if (!fs_info->btree_inode) {
		err = -ENOMEM;
		goto fail_bio_counter;
	}
	mapping_set_gfp_mask(fs_info->btree_inode->i_mapping, GFP_NOFS);

2501
	fs_info->delayed_root = kmalloc(sizeof(struct btrfs_delayed_root),
2502
					GFP_KERNEL);
2503 2504 2505 2506 2507
	if (!fs_info->delayed_root) {
		err = -ENOMEM;
		goto fail_iput;
	}
	btrfs_init_delayed_root(fs_info->delayed_root);
2508

2509
	btrfs_init_scrub(fs_info);
2510 2511 2512
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	fs_info->check_integrity_print_mask = 0;
#endif
2513
	btrfs_init_balance(fs_info);
2514
	btrfs_init_async_reclaim_work(&fs_info->async_reclaim_work);
A
Arne Jansen 已提交
2515

2516 2517
	sb->s_blocksize = BTRFS_BDEV_BLOCKSIZE;
	sb->s_blocksize_bits = blksize_bits(BTRFS_BDEV_BLOCKSIZE);
2518

2519
	btrfs_init_btree_inode(fs_info);
2520

J
Josef Bacik 已提交
2521
	spin_lock_init(&fs_info->block_group_cache_lock);
2522
	fs_info->block_group_cache_tree = RB_ROOT;
2523
	fs_info->first_logical_byte = (u64)-1;
J
Josef Bacik 已提交
2524

2525 2526
	extent_io_tree_init(&fs_info->freed_extents[0], NULL);
	extent_io_tree_init(&fs_info->freed_extents[1], NULL);
2527
	fs_info->pinned_extents = &fs_info->freed_extents[0];
2528
	set_bit(BTRFS_FS_BARRIER, &fs_info->flags);
C
Chris Mason 已提交
2529

2530
	mutex_init(&fs_info->ordered_operations_mutex);
2531
	mutex_init(&fs_info->tree_log_mutex);
2532
	mutex_init(&fs_info->chunk_mutex);
2533 2534
	mutex_init(&fs_info->transaction_kthread_mutex);
	mutex_init(&fs_info->cleaner_mutex);
2535
	mutex_init(&fs_info->volume_mutex);
2536
	mutex_init(&fs_info->ro_block_group_mutex);
2537
	init_rwsem(&fs_info->commit_root_sem);
2538
	init_rwsem(&fs_info->cleanup_work_sem);
2539
	init_rwsem(&fs_info->subvol_sem);
S
Stefan Behrens 已提交
2540
	sema_init(&fs_info->uuid_tree_rescan_sem, 1);
2541

2542
	btrfs_init_dev_replace_locks(fs_info);
2543
	btrfs_init_qgroup(fs_info);
2544

2545 2546 2547
	btrfs_init_free_cluster(&fs_info->meta_alloc_cluster);
	btrfs_init_free_cluster(&fs_info->data_alloc_cluster);

2548
	init_waitqueue_head(&fs_info->transaction_throttle);
2549
	init_waitqueue_head(&fs_info->transaction_wait);
S
Sage Weil 已提交
2550
	init_waitqueue_head(&fs_info->transaction_blocked_wait);
2551
	init_waitqueue_head(&fs_info->async_submit_wait);
2552

2553 2554
	INIT_LIST_HEAD(&fs_info->pinned_chunks);

2555 2556 2557 2558 2559
	/* Usable values until the real ones are cached from the superblock */
	fs_info->nodesize = 4096;
	fs_info->sectorsize = 4096;
	fs_info->stripesize = 4096;

D
David Woodhouse 已提交
2560 2561
	ret = btrfs_alloc_stripe_hash_table(fs_info);
	if (ret) {
2562
		err = ret;
D
David Woodhouse 已提交
2563 2564 2565
		goto fail_alloc;
	}

2566
	__setup_root(tree_root, fs_info, BTRFS_ROOT_TREE_OBJECTID);
2567

2568
	invalidate_bdev(fs_devices->latest_bdev);
D
David Sterba 已提交
2569 2570 2571 2572

	/*
	 * Read super block and check the signature bytes only
	 */
Y
Yan Zheng 已提交
2573
	bh = btrfs_read_dev_super(fs_devices->latest_bdev);
2574 2575
	if (IS_ERR(bh)) {
		err = PTR_ERR(bh);
2576
		goto fail_alloc;
2577
	}
C
Chris Mason 已提交
2578

D
David Sterba 已提交
2579 2580 2581 2582
	/*
	 * We want to check superblock checksum, the type is stored inside.
	 * Pass the whole disk block of size BTRFS_SUPER_INFO_SIZE (4k).
	 */
2583
	if (btrfs_check_super_csum(fs_info, bh->b_data)) {
2584
		btrfs_err(fs_info, "superblock checksum mismatch");
D
David Sterba 已提交
2585
		err = -EINVAL;
2586
		brelse(bh);
D
David Sterba 已提交
2587 2588 2589 2590 2591 2592 2593 2594
		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
	 */
2595 2596 2597
	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));
2598
	brelse(bh);
2599

2600
	memcpy(fs_info->fsid, fs_info->super_copy->fsid, BTRFS_FSID_SIZE);
2601

2602
	ret = btrfs_check_super_valid(fs_info);
D
David Sterba 已提交
2603
	if (ret) {
2604
		btrfs_err(fs_info, "superblock contains fatal errors");
D
David Sterba 已提交
2605 2606 2607 2608
		err = -EINVAL;
		goto fail_alloc;
	}

2609
	disk_super = fs_info->super_copy;
2610
	if (!btrfs_super_root(disk_super))
2611
		goto fail_alloc;
2612

L
liubo 已提交
2613
	/* check FS state, whether FS is broken. */
2614 2615
	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
2616

C
Chris Mason 已提交
2617 2618 2619 2620 2621 2622 2623
	/*
	 * 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);

2624 2625 2626 2627 2628 2629
	/*
	 * 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;

2630
	ret = btrfs_parse_options(fs_info, options, sb->s_flags);
Y
Yan Zheng 已提交
2631 2632
	if (ret) {
		err = ret;
2633
		goto fail_alloc;
Y
Yan Zheng 已提交
2634
	}
2635

2636 2637 2638
	features = btrfs_super_incompat_flags(disk_super) &
		~BTRFS_FEATURE_INCOMPAT_SUPP;
	if (features) {
2639 2640 2641
		btrfs_err(fs_info,
		    "cannot mount because of unsupported optional features (%llx)",
		    features);
2642
		err = -EINVAL;
2643
		goto fail_alloc;
2644 2645
	}

2646
	features = btrfs_super_incompat_flags(disk_super);
L
Li Zefan 已提交
2647
	features |= BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF;
2648
	if (fs_info->compress_type == BTRFS_COMPRESS_LZO)
L
Li Zefan 已提交
2649
		features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
N
Nick Terrell 已提交
2650 2651
	else if (fs_info->compress_type == BTRFS_COMPRESS_ZSTD)
		features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_ZSTD;
2652

2653
	if (features & BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA)
2654
		btrfs_info(fs_info, "has skinny extents");
2655

2656 2657 2658 2659
	/*
	 * flag our filesystem as having big metadata blocks if
	 * they are bigger than the page size
	 */
2660
	if (btrfs_super_nodesize(disk_super) > PAGE_SIZE) {
2661
		if (!(features & BTRFS_FEATURE_INCOMPAT_BIG_METADATA))
2662 2663
			btrfs_info(fs_info,
				"flagging fs with big metadata feature");
2664 2665 2666
		features |= BTRFS_FEATURE_INCOMPAT_BIG_METADATA;
	}

2667 2668
	nodesize = btrfs_super_nodesize(disk_super);
	sectorsize = btrfs_super_sectorsize(disk_super);
2669
	stripesize = sectorsize;
2670
	fs_info->dirty_metadata_batch = nodesize * (1 + ilog2(nr_cpu_ids));
2671
	fs_info->delalloc_batch = sectorsize * 512 * (1 + ilog2(nr_cpu_ids));
2672

2673 2674 2675 2676 2677
	/* Cache block sizes */
	fs_info->nodesize = nodesize;
	fs_info->sectorsize = sectorsize;
	fs_info->stripesize = stripesize;

2678 2679 2680 2681 2682
	/*
	 * 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) &&
2683
	    (sectorsize != nodesize)) {
2684 2685 2686
		btrfs_err(fs_info,
"unequal nodesize/sectorsize (%u != %u) are not allowed for mixed block groups",
			nodesize, sectorsize);
2687 2688 2689
		goto fail_alloc;
	}

2690 2691 2692 2693
	/*
	 * Needn't use the lock because there is no other task which will
	 * update the flag.
	 */
L
Li Zefan 已提交
2694
	btrfs_set_super_incompat_flags(disk_super, features);
2695

2696 2697
	features = btrfs_super_compat_ro_flags(disk_super) &
		~BTRFS_FEATURE_COMPAT_RO_SUPP;
2698
	if (!sb_rdonly(sb) && features) {
2699 2700
		btrfs_err(fs_info,
	"cannot mount read-write because of unsupported optional features (%llx)",
2701
		       features);
2702
		err = -EINVAL;
2703
		goto fail_alloc;
2704
	}
2705

2706
	max_active = fs_info->thread_pool_size;
2707

2708 2709 2710
	ret = btrfs_init_workqueues(fs_info, fs_devices);
	if (ret) {
		err = ret;
2711 2712
		goto fail_sb_buffer;
	}
2713

2714 2715 2716
	sb->s_bdi->congested_fn = btrfs_congested_fn;
	sb->s_bdi->congested_data = fs_info;
	sb->s_bdi->capabilities |= BDI_CAP_CGROUP_WRITEBACK;
2717
	sb->s_bdi->ra_pages = VM_MAX_READAHEAD * SZ_1K / PAGE_SIZE;
2718 2719
	sb->s_bdi->ra_pages *= btrfs_super_num_devices(disk_super);
	sb->s_bdi->ra_pages = max(sb->s_bdi->ra_pages, SZ_4M / PAGE_SIZE);
2720

2721 2722
	sb->s_blocksize = sectorsize;
	sb->s_blocksize_bits = blksize_bits(sectorsize);
2723
	memcpy(&sb->s_uuid, fs_info->fsid, BTRFS_FSID_SIZE);
2724

2725
	mutex_lock(&fs_info->chunk_mutex);
2726
	ret = btrfs_read_sys_array(fs_info);
2727
	mutex_unlock(&fs_info->chunk_mutex);
2728
	if (ret) {
2729
		btrfs_err(fs_info, "failed to read the system array: %d", ret);
2730
		goto fail_sb_buffer;
2731
	}
2732

2733
	generation = btrfs_super_chunk_root_generation(disk_super);
2734

2735
	__setup_root(chunk_root, fs_info, BTRFS_CHUNK_TREE_OBJECTID);
2736

2737
	chunk_root->node = read_tree_block(fs_info,
2738
					   btrfs_super_chunk_root(disk_super),
2739
					   generation);
2740 2741
	if (IS_ERR(chunk_root->node) ||
	    !extent_buffer_uptodate(chunk_root->node)) {
2742
		btrfs_err(fs_info, "failed to read chunk root");
2743 2744
		if (!IS_ERR(chunk_root->node))
			free_extent_buffer(chunk_root->node);
2745
		chunk_root->node = NULL;
C
Chris Mason 已提交
2746
		goto fail_tree_roots;
2747
	}
2748 2749
	btrfs_set_root_node(&chunk_root->root_item, chunk_root->node);
	chunk_root->commit_root = btrfs_root_node(chunk_root);
2750

2751
	read_extent_buffer(chunk_root->node, fs_info->chunk_tree_uuid,
2752
	   btrfs_header_chunk_tree_uuid(chunk_root->node), BTRFS_UUID_SIZE);
2753

2754
	ret = btrfs_read_chunk_tree(fs_info);
Y
Yan Zheng 已提交
2755
	if (ret) {
2756
		btrfs_err(fs_info, "failed to read chunk tree: %d", ret);
C
Chris Mason 已提交
2757
		goto fail_tree_roots;
Y
Yan Zheng 已提交
2758
	}
2759

2760 2761 2762 2763
	/*
	 * keep the device that is marked to be the target device for the
	 * dev_replace procedure
	 */
2764
	btrfs_close_extra_devices(fs_devices, 0);
2765

2766
	if (!fs_devices->latest_bdev) {
2767
		btrfs_err(fs_info, "failed to read devices");
2768 2769 2770
		goto fail_tree_roots;
	}

C
Chris Mason 已提交
2771
retry_root_backup:
2772
	generation = btrfs_super_generation(disk_super);
2773

2774
	tree_root->node = read_tree_block(fs_info,
2775
					  btrfs_super_root(disk_super),
2776
					  generation);
2777 2778
	if (IS_ERR(tree_root->node) ||
	    !extent_buffer_uptodate(tree_root->node)) {
2779
		btrfs_warn(fs_info, "failed to read tree root");
2780 2781
		if (!IS_ERR(tree_root->node))
			free_extent_buffer(tree_root->node);
2782
		tree_root->node = NULL;
C
Chris Mason 已提交
2783
		goto recovery_tree_root;
2784
	}
C
Chris Mason 已提交
2785

2786 2787
	btrfs_set_root_node(&tree_root->root_item, tree_root->node);
	tree_root->commit_root = btrfs_root_node(tree_root);
2788
	btrfs_set_root_refs(&tree_root->root_item, 1);
2789

2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801
	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);

2802
	ret = btrfs_read_roots(fs_info);
2803
	if (ret)
C
Chris Mason 已提交
2804
		goto recovery_tree_root;
2805

2806 2807 2808
	fs_info->generation = generation;
	fs_info->last_trans_committed = generation;

2809 2810
	ret = btrfs_recover_balance(fs_info);
	if (ret) {
2811
		btrfs_err(fs_info, "failed to recover balance: %d", ret);
2812 2813 2814
		goto fail_block_groups;
	}

2815 2816
	ret = btrfs_init_dev_stats(fs_info);
	if (ret) {
2817
		btrfs_err(fs_info, "failed to init dev_stats: %d", ret);
2818 2819 2820
		goto fail_block_groups;
	}

2821 2822
	ret = btrfs_init_dev_replace(fs_info);
	if (ret) {
2823
		btrfs_err(fs_info, "failed to init dev_replace: %d", ret);
2824 2825 2826
		goto fail_block_groups;
	}

2827
	btrfs_close_extra_devices(fs_devices, 1);
2828

2829 2830
	ret = btrfs_sysfs_add_fsid(fs_devices, NULL);
	if (ret) {
2831 2832
		btrfs_err(fs_info, "failed to init sysfs fsid interface: %d",
				ret);
2833 2834 2835 2836 2837
		goto fail_block_groups;
	}

	ret = btrfs_sysfs_add_device(fs_devices);
	if (ret) {
2838 2839
		btrfs_err(fs_info, "failed to init sysfs device interface: %d",
				ret);
2840 2841 2842
		goto fail_fsdev_sysfs;
	}

2843
	ret = btrfs_sysfs_add_mounted(fs_info);
2844
	if (ret) {
2845
		btrfs_err(fs_info, "failed to init sysfs interface: %d", ret);
2846
		goto fail_fsdev_sysfs;
2847 2848 2849 2850
	}

	ret = btrfs_init_space_info(fs_info);
	if (ret) {
2851
		btrfs_err(fs_info, "failed to initialize space info: %d", ret);
2852
		goto fail_sysfs;
2853 2854
	}

2855
	ret = btrfs_read_block_groups(fs_info);
2856
	if (ret) {
2857
		btrfs_err(fs_info, "failed to read block groups: %d", ret);
2858
		goto fail_sysfs;
2859
	}
2860

2861
	if (!sb_rdonly(sb) && !btrfs_check_rw_degradable(fs_info, NULL)) {
2862
		btrfs_warn(fs_info,
2863
		"writeable mount is not allowed due to too many missing devices");
2864
		goto fail_sysfs;
2865
	}
C
Chris Mason 已提交
2866

2867 2868
	fs_info->cleaner_kthread = kthread_run(cleaner_kthread, tree_root,
					       "btrfs-cleaner");
2869
	if (IS_ERR(fs_info->cleaner_kthread))
2870
		goto fail_sysfs;
2871 2872 2873 2874

	fs_info->transaction_kthread = kthread_run(transaction_kthread,
						   tree_root,
						   "btrfs-transaction");
2875
	if (IS_ERR(fs_info->transaction_kthread))
2876
		goto fail_cleaner;
2877

2878
	if (!btrfs_test_opt(fs_info, NOSSD) &&
C
Chris Mason 已提交
2879
	    !fs_info->fs_devices->rotating) {
2880
		btrfs_set_and_info(fs_info, SSD, "enabling ssd optimizations");
C
Chris Mason 已提交
2881 2882
	}

2883
	/*
2884
	 * Mount does not set all options immediately, we can do it now and do
2885 2886 2887
	 * not have to wait for transaction commit
	 */
	btrfs_apply_pending_changes(fs_info);
2888

2889
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
2890
	if (btrfs_test_opt(fs_info, CHECK_INTEGRITY)) {
2891
		ret = btrfsic_mount(fs_info, fs_devices,
2892
				    btrfs_test_opt(fs_info,
2893 2894 2895 2896
					CHECK_INTEGRITY_INCLUDING_EXTENT_DATA) ?
				    1 : 0,
				    fs_info->check_integrity_print_mask);
		if (ret)
2897 2898 2899
			btrfs_warn(fs_info,
				"failed to initialize integrity check module: %d",
				ret);
2900 2901
	}
#endif
2902 2903 2904
	ret = btrfs_read_qgroup_config(fs_info);
	if (ret)
		goto fail_trans_kthread;
2905

J
Josef Bacik 已提交
2906 2907 2908
	if (btrfs_build_ref_tree(fs_info))
		btrfs_err(fs_info, "couldn't build ref tree");

2909 2910
	/* do not make disk changes in broken FS or nologreplay is given */
	if (btrfs_super_log_root(disk_super) != 0 &&
2911
	    !btrfs_test_opt(fs_info, NOLOGREPLAY)) {
2912
		ret = btrfs_replay_log(fs_info, fs_devices);
2913
		if (ret) {
2914
			err = ret;
2915
			goto fail_qgroup;
2916
		}
2917
	}
Z
Zheng Yan 已提交
2918

2919
	ret = btrfs_find_orphan_roots(fs_info);
2920
	if (ret)
2921
		goto fail_qgroup;
2922

2923
	if (!sb_rdonly(sb)) {
2924
		ret = btrfs_cleanup_fs_roots(fs_info);
2925
		if (ret)
2926
			goto fail_qgroup;
2927 2928

		mutex_lock(&fs_info->cleaner_mutex);
2929
		ret = btrfs_recover_relocation(tree_root);
2930
		mutex_unlock(&fs_info->cleaner_mutex);
2931
		if (ret < 0) {
2932 2933
			btrfs_warn(fs_info, "failed to recover relocation: %d",
					ret);
2934
			err = -EINVAL;
2935
			goto fail_qgroup;
2936
		}
2937
	}
Z
Zheng Yan 已提交
2938

2939 2940
	location.objectid = BTRFS_FS_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
2941
	location.offset = 0;
2942 2943

	fs_info->fs_root = btrfs_read_fs_root_no_name(fs_info, &location);
2944 2945
	if (IS_ERR(fs_info->fs_root)) {
		err = PTR_ERR(fs_info->fs_root);
2946
		goto fail_qgroup;
2947
	}
C
Chris Mason 已提交
2948

2949
	if (sb_rdonly(sb))
2950
		return 0;
I
Ilya Dryomov 已提交
2951

2952 2953
	if (btrfs_test_opt(fs_info, CLEAR_CACHE) &&
	    btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
2954 2955 2956 2957 2958 2959 2960 2961
		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) {
2962 2963 2964 2965 2966
		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);
2967
			close_ctree(fs_info);
2968 2969 2970 2971
			return ret;
		}
	}

2972
	if (btrfs_test_opt(fs_info, FREE_SPACE_TREE) &&
2973
	    !btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
2974
		btrfs_info(fs_info, "creating free space tree");
2975 2976
		ret = btrfs_create_free_space_tree(fs_info);
		if (ret) {
2977 2978
			btrfs_warn(fs_info,
				"failed to create free space tree: %d", ret);
2979
			close_ctree(fs_info);
2980 2981 2982 2983
			return ret;
		}
	}

2984 2985 2986
	down_read(&fs_info->cleanup_work_sem);
	if ((ret = btrfs_orphan_cleanup(fs_info->fs_root)) ||
	    (ret = btrfs_orphan_cleanup(fs_info->tree_root))) {
2987
		up_read(&fs_info->cleanup_work_sem);
2988
		close_ctree(fs_info);
2989 2990 2991
		return ret;
	}
	up_read(&fs_info->cleanup_work_sem);
I
Ilya Dryomov 已提交
2992

2993 2994
	ret = btrfs_resume_balance_async(fs_info);
	if (ret) {
2995
		btrfs_warn(fs_info, "failed to resume balance: %d", ret);
2996
		close_ctree(fs_info);
2997
		return ret;
2998 2999
	}

3000 3001
	ret = btrfs_resume_dev_replace_async(fs_info);
	if (ret) {
3002
		btrfs_warn(fs_info, "failed to resume device replace: %d", ret);
3003
		close_ctree(fs_info);
3004 3005 3006
		return ret;
	}

3007 3008
	btrfs_qgroup_rescan_resume(fs_info);

3009
	if (!fs_info->uuid_root) {
3010
		btrfs_info(fs_info, "creating UUID tree");
3011 3012
		ret = btrfs_create_uuid_tree(fs_info);
		if (ret) {
3013 3014
			btrfs_warn(fs_info,
				"failed to create the UUID tree: %d", ret);
3015
			close_ctree(fs_info);
3016 3017
			return ret;
		}
3018
	} else if (btrfs_test_opt(fs_info, RESCAN_UUID_TREE) ||
3019 3020
		   fs_info->generation !=
				btrfs_super_uuid_tree_generation(disk_super)) {
3021
		btrfs_info(fs_info, "checking UUID tree");
3022 3023
		ret = btrfs_check_uuid_tree(fs_info);
		if (ret) {
3024 3025
			btrfs_warn(fs_info,
				"failed to check the UUID tree: %d", ret);
3026
			close_ctree(fs_info);
3027 3028 3029
			return ret;
		}
	} else {
3030
		set_bit(BTRFS_FS_UPDATE_UUID_TREE_GEN, &fs_info->flags);
3031
	}
3032
	set_bit(BTRFS_FS_OPEN, &fs_info->flags);
3033

3034 3035 3036 3037 3038 3039
	/*
	 * 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
3040
	return 0;
C
Chris Mason 已提交
3041

3042 3043
fail_qgroup:
	btrfs_free_qgroup_config(fs_info);
3044 3045
fail_trans_kthread:
	kthread_stop(fs_info->transaction_kthread);
3046
	btrfs_cleanup_transaction(fs_info);
3047
	btrfs_free_fs_roots(fs_info);
3048
fail_cleaner:
3049
	kthread_stop(fs_info->cleaner_kthread);
3050 3051 3052 3053 3054 3055 3056

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

3057
fail_sysfs:
3058
	btrfs_sysfs_remove_mounted(fs_info);
3059

3060 3061 3062
fail_fsdev_sysfs:
	btrfs_sysfs_remove_fsid(fs_info->fs_devices);

3063
fail_block_groups:
J
Josef Bacik 已提交
3064
	btrfs_put_block_group_cache(fs_info);
C
Chris Mason 已提交
3065 3066 3067

fail_tree_roots:
	free_root_pointers(fs_info, 1);
3068
	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
3069

C
Chris Mason 已提交
3070
fail_sb_buffer:
L
Liu Bo 已提交
3071
	btrfs_stop_all_workers(fs_info);
3072
	btrfs_free_block_groups(fs_info);
3073
fail_alloc:
3074
fail_iput:
3075 3076
	btrfs_mapping_tree_free(&fs_info->mapping_tree);

3077
	iput(fs_info->btree_inode);
3078 3079
fail_bio_counter:
	percpu_counter_destroy(&fs_info->bio_counter);
3080 3081
fail_delalloc_bytes:
	percpu_counter_destroy(&fs_info->delalloc_bytes);
3082 3083
fail_dirty_metadata_bytes:
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
3084 3085
fail_srcu:
	cleanup_srcu_struct(&fs_info->subvol_srcu);
3086
fail:
D
David Woodhouse 已提交
3087
	btrfs_free_stripe_hash_table(fs_info);
3088
	btrfs_close_devices(fs_info->fs_devices);
A
Al Viro 已提交
3089
	return err;
C
Chris Mason 已提交
3090 3091

recovery_tree_root:
3092
	if (!btrfs_test_opt(fs_info, USEBACKUPROOT))
C
Chris Mason 已提交
3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107
		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;
3108
}
3109
ALLOW_ERROR_INJECTION(open_ctree, ERRNO);
3110

3111 3112 3113 3114 3115
static void btrfs_end_buffer_write_sync(struct buffer_head *bh, int uptodate)
{
	if (uptodate) {
		set_buffer_uptodate(bh);
	} else {
3116 3117 3118
		struct btrfs_device *device = (struct btrfs_device *)
			bh->b_private;

3119
		btrfs_warn_rl_in_rcu(device->fs_info,
3120
				"lost page write due to IO error on %s",
3121
					  rcu_str_deref(device->name));
3122
		/* note, we don't set_buffer_write_io_error because we have
3123 3124
		 * our own ways of dealing with the IO errors
		 */
3125
		clear_buffer_uptodate(bh);
3126
		btrfs_dev_stat_inc_and_print(device, BTRFS_DEV_STAT_WRITE_ERRS);
3127 3128 3129 3130 3131
	}
	unlock_buffer(bh);
	put_bh(bh);
}

3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142
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;

3143
	bh = __bread(bdev, bytenr / BTRFS_BDEV_BLOCKSIZE, BTRFS_SUPER_INFO_SIZE);
3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162
	/*
	 * 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
3163 3164 3165 3166 3167 3168 3169
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;
3170
	int ret = -EINVAL;
Y
Yan Zheng 已提交
3171 3172 3173 3174 3175 3176 3177

	/* 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++) {
3178 3179
		ret = btrfs_read_dev_one_super(bdev, i, &bh);
		if (ret)
Y
Yan Zheng 已提交
3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191
			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);
		}
	}
3192 3193 3194 3195

	if (!latest)
		return ERR_PTR(ret);

Y
Yan Zheng 已提交
3196 3197 3198
	return latest;
}

3199
/*
3200 3201
 * Write superblock @sb to the @device. Do not wait for completion, all the
 * buffer heads we write are pinned.
3202
 *
3203 3204 3205
 * Write @max_mirrors copies of the superblock, where 0 means default that fit
 * the expected device size at commit time. Note that max_mirrors must be
 * same for write and wait phases.
3206
 *
3207
 * Return number of errors when buffer head is not found or submission fails.
3208
 */
Y
Yan Zheng 已提交
3209
static int write_dev_supers(struct btrfs_device *device,
3210
			    struct btrfs_super_block *sb, int max_mirrors)
Y
Yan Zheng 已提交
3211 3212 3213 3214 3215 3216 3217
{
	struct buffer_head *bh;
	int i;
	int ret;
	int errors = 0;
	u32 crc;
	u64 bytenr;
3218
	int op_flags;
Y
Yan Zheng 已提交
3219 3220 3221 3222 3223 3224

	if (max_mirrors == 0)
		max_mirrors = BTRFS_SUPER_MIRROR_MAX;

	for (i = 0; i < max_mirrors; i++) {
		bytenr = btrfs_sb_offset(i);
3225 3226
		if (bytenr + BTRFS_SUPER_INFO_SIZE >=
		    device->commit_total_bytes)
Y
Yan Zheng 已提交
3227 3228
			break;

3229
		btrfs_set_super_bytenr(sb, bytenr);
3230

3231 3232 3233 3234
		crc = ~(u32)0;
		crc = btrfs_csum_data((const char *)sb + BTRFS_CSUM_SIZE, crc,
				      BTRFS_SUPER_INFO_SIZE - BTRFS_CSUM_SIZE);
		btrfs_csum_final(crc, sb->csum);
3235

3236
		/* One reference for us, and we leave it for the caller */
3237
		bh = __getblk(device->bdev, bytenr / BTRFS_BDEV_BLOCKSIZE,
3238 3239 3240 3241 3242 3243
			      BTRFS_SUPER_INFO_SIZE);
		if (!bh) {
			btrfs_err(device->fs_info,
			    "couldn't get super buffer head for bytenr %llu",
			    bytenr);
			errors++;
3244
			continue;
3245
		}
3246

3247
		memcpy(bh->b_data, sb, BTRFS_SUPER_INFO_SIZE);
Y
Yan Zheng 已提交
3248

3249 3250
		/* one reference for submit_bh */
		get_bh(bh);
3251

3252 3253 3254 3255
		set_buffer_uptodate(bh);
		lock_buffer(bh);
		bh->b_end_io = btrfs_end_buffer_write_sync;
		bh->b_private = device;
Y
Yan Zheng 已提交
3256

C
Chris Mason 已提交
3257 3258 3259 3260
		/*
		 * we fua the first super.  The others we allow
		 * to go down lazy.
		 */
3261 3262 3263 3264
		op_flags = REQ_SYNC | REQ_META | REQ_PRIO;
		if (i == 0 && !btrfs_test_opt(device->fs_info, NOBARRIER))
			op_flags |= REQ_FUA;
		ret = btrfsic_submit_bh(REQ_OP_WRITE, op_flags, bh);
3265
		if (ret)
Y
Yan Zheng 已提交
3266 3267 3268 3269 3270
			errors++;
	}
	return errors < i ? 0 : -1;
}

3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282
/*
 * Wait for write completion of superblocks done by write_dev_supers,
 * @max_mirrors same for write and wait phases.
 *
 * Return number of errors when buffer head is not found or not marked up to
 * date.
 */
static int wait_dev_supers(struct btrfs_device *device, int max_mirrors)
{
	struct buffer_head *bh;
	int i;
	int errors = 0;
3283
	bool primary_failed = false;
3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294
	u64 bytenr;

	if (max_mirrors == 0)
		max_mirrors = BTRFS_SUPER_MIRROR_MAX;

	for (i = 0; i < max_mirrors; i++) {
		bytenr = btrfs_sb_offset(i);
		if (bytenr + BTRFS_SUPER_INFO_SIZE >=
		    device->commit_total_bytes)
			break;

3295 3296
		bh = __find_get_block(device->bdev,
				      bytenr / BTRFS_BDEV_BLOCKSIZE,
3297 3298 3299
				      BTRFS_SUPER_INFO_SIZE);
		if (!bh) {
			errors++;
3300 3301
			if (i == 0)
				primary_failed = true;
3302 3303 3304
			continue;
		}
		wait_on_buffer(bh);
3305
		if (!buffer_uptodate(bh)) {
3306
			errors++;
3307 3308 3309
			if (i == 0)
				primary_failed = true;
		}
3310 3311 3312 3313 3314 3315 3316 3317

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

		/* drop the reference from the writing run */
		brelse(bh);
	}

3318 3319 3320 3321 3322 3323 3324
	/* log error, force error return */
	if (primary_failed) {
		btrfs_err(device->fs_info, "error writing primary super block to device %llu",
			  device->devid);
		return -1;
	}

3325 3326 3327
	return errors < i ? 0 : -1;
}

C
Chris Mason 已提交
3328 3329 3330 3331
/*
 * endio for the write_dev_flush, this will wake anyone waiting
 * for the barrier when it is done
 */
3332
static void btrfs_end_empty_barrier(struct bio *bio)
C
Chris Mason 已提交
3333
{
3334
	complete(bio->bi_private);
C
Chris Mason 已提交
3335 3336 3337
}

/*
3338 3339
 * Submit a flush request to the device if it supports it. Error handling is
 * done in the waiting counterpart.
C
Chris Mason 已提交
3340
 */
3341
static void write_dev_flush(struct btrfs_device *device)
C
Chris Mason 已提交
3342
{
3343
	struct request_queue *q = bdev_get_queue(device->bdev);
3344
	struct bio *bio = device->flush_bio;
C
Chris Mason 已提交
3345

3346
	if (!test_bit(QUEUE_FLAG_WC, &q->queue_flags))
3347
		return;
C
Chris Mason 已提交
3348

3349
	bio_reset(bio);
C
Chris Mason 已提交
3350
	bio->bi_end_io = btrfs_end_empty_barrier;
3351
	bio_set_dev(bio, device->bdev);
3352
	bio->bi_opf = REQ_OP_WRITE | REQ_SYNC | REQ_PREFLUSH;
C
Chris Mason 已提交
3353 3354 3355
	init_completion(&device->flush_wait);
	bio->bi_private = &device->flush_wait;

3356
	btrfsic_submit_bio(bio);
3357
	set_bit(BTRFS_DEV_STATE_FLUSH_SENT, &device->dev_state);
3358
}
C
Chris Mason 已提交
3359

3360 3361 3362
/*
 * If the flush bio has been submitted by write_dev_flush, wait for it.
 */
3363
static blk_status_t wait_dev_flush(struct btrfs_device *device)
3364 3365
{
	struct bio *bio = device->flush_bio;
C
Chris Mason 已提交
3366

3367
	if (!test_bit(BTRFS_DEV_STATE_FLUSH_SENT, &device->dev_state))
3368
		return BLK_STS_OK;
C
Chris Mason 已提交
3369

3370
	clear_bit(BTRFS_DEV_STATE_FLUSH_SENT, &device->dev_state);
3371
	wait_for_completion_io(&device->flush_wait);
C
Chris Mason 已提交
3372

3373
	return bio->bi_status;
C
Chris Mason 已提交
3374 3375
}

3376
static int check_barrier_error(struct btrfs_fs_info *fs_info)
3377
{
3378
	if (!btrfs_check_rw_degradable(fs_info, NULL))
3379
		return -EIO;
C
Chris Mason 已提交
3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390
	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;
3391
	int errors_wait = 0;
3392
	blk_status_t ret;
C
Chris Mason 已提交
3393

3394
	lockdep_assert_held(&info->fs_devices->device_list_mutex);
C
Chris Mason 已提交
3395 3396
	/* send down all the barriers */
	head = &info->fs_devices->devices;
3397
	list_for_each_entry(dev, head, dev_list) {
3398
		if (test_bit(BTRFS_DEV_STATE_MISSING, &dev->dev_state))
3399
			continue;
3400
		if (!dev->bdev)
C
Chris Mason 已提交
3401
			continue;
3402
		if (!test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &dev->dev_state) ||
3403
		    !test_bit(BTRFS_DEV_STATE_WRITEABLE, &dev->dev_state))
C
Chris Mason 已提交
3404 3405
			continue;

3406
		write_dev_flush(dev);
3407
		dev->last_flush_error = BLK_STS_OK;
C
Chris Mason 已提交
3408 3409 3410
	}

	/* wait for all the barriers */
3411
	list_for_each_entry(dev, head, dev_list) {
3412
		if (test_bit(BTRFS_DEV_STATE_MISSING, &dev->dev_state))
3413
			continue;
C
Chris Mason 已提交
3414
		if (!dev->bdev) {
3415
			errors_wait++;
C
Chris Mason 已提交
3416 3417
			continue;
		}
3418
		if (!test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &dev->dev_state) ||
3419
		    !test_bit(BTRFS_DEV_STATE_WRITEABLE, &dev->dev_state))
C
Chris Mason 已提交
3420 3421
			continue;

3422
		ret = wait_dev_flush(dev);
3423 3424
		if (ret) {
			dev->last_flush_error = ret;
3425 3426
			btrfs_dev_stat_inc_and_print(dev,
					BTRFS_DEV_STAT_FLUSH_ERRS);
3427
			errors_wait++;
3428 3429 3430
		}
	}

3431
	if (errors_wait) {
3432 3433 3434 3435 3436
		/*
		 * At some point we need the status of all disks
		 * to arrive at the volume status. So error checking
		 * is being pushed to a separate loop.
		 */
3437
		return check_barrier_error(info);
C
Chris Mason 已提交
3438 3439 3440 3441
	}
	return 0;
}

3442 3443
int btrfs_get_num_tolerated_disk_barrier_failures(u64 flags)
{
3444 3445
	int raid_type;
	int min_tolerated = INT_MAX;
3446

3447 3448 3449 3450 3451
	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);
3452

3453 3454 3455 3456 3457 3458 3459 3460 3461
	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);
	}
3462

3463
	if (min_tolerated == INT_MAX) {
3464
		pr_warn("BTRFS: unknown raid flag: %llu", flags);
3465 3466 3467 3468
		min_tolerated = 0;
	}

	return min_tolerated;
3469 3470
}

3471
int write_all_supers(struct btrfs_fs_info *fs_info, int max_mirrors)
3472
{
3473
	struct list_head *head;
3474
	struct btrfs_device *dev;
3475
	struct btrfs_super_block *sb;
3476 3477 3478
	struct btrfs_dev_item *dev_item;
	int ret;
	int do_barriers;
3479 3480
	int max_errors;
	int total_errors = 0;
3481
	u64 flags;
3482

3483
	do_barriers = !btrfs_test_opt(fs_info, NOBARRIER);
3484 3485 3486 3487 3488 3489 3490 3491

	/*
	 * max_mirrors == 0 indicates we're from commit_transaction,
	 * not from fsync where the tree roots in fs_info have not
	 * been consistent on disk.
	 */
	if (max_mirrors == 0)
		backup_super_roots(fs_info);
3492

3493
	sb = fs_info->super_for_commit;
3494
	dev_item = &sb->dev_item;
3495

3496 3497 3498
	mutex_lock(&fs_info->fs_devices->device_list_mutex);
	head = &fs_info->fs_devices->devices;
	max_errors = btrfs_super_num_devices(fs_info->super_copy) - 1;
C
Chris Mason 已提交
3499

3500
	if (do_barriers) {
3501
		ret = barrier_all_devices(fs_info);
3502 3503
		if (ret) {
			mutex_unlock(
3504 3505 3506
				&fs_info->fs_devices->device_list_mutex);
			btrfs_handle_fs_error(fs_info, ret,
					      "errors while submitting device barriers.");
3507 3508 3509
			return ret;
		}
	}
C
Chris Mason 已提交
3510

3511
	list_for_each_entry(dev, head, dev_list) {
3512 3513 3514 3515
		if (!dev->bdev) {
			total_errors++;
			continue;
		}
3516
		if (!test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &dev->dev_state) ||
3517
		    !test_bit(BTRFS_DEV_STATE_WRITEABLE, &dev->dev_state))
3518 3519
			continue;

Y
Yan Zheng 已提交
3520
		btrfs_set_stack_device_generation(dev_item, 0);
3521 3522
		btrfs_set_stack_device_type(dev_item, dev->type);
		btrfs_set_stack_device_id(dev_item, dev->devid);
3523
		btrfs_set_stack_device_total_bytes(dev_item,
3524
						   dev->commit_total_bytes);
3525 3526
		btrfs_set_stack_device_bytes_used(dev_item,
						  dev->commit_bytes_used);
3527 3528 3529 3530
		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);
3531
		memcpy(dev_item->fsid, dev->fs_devices->fsid, BTRFS_FSID_SIZE);
Y
Yan Zheng 已提交
3532

3533 3534 3535
		flags = btrfs_super_flags(sb);
		btrfs_set_super_flags(sb, flags | BTRFS_HEADER_FLAG_WRITTEN);

3536
		ret = write_dev_supers(dev, sb, max_mirrors);
3537 3538
		if (ret)
			total_errors++;
3539
	}
3540
	if (total_errors > max_errors) {
3541 3542 3543
		btrfs_err(fs_info, "%d errors while writing supers",
			  total_errors);
		mutex_unlock(&fs_info->fs_devices->device_list_mutex);
3544

3545
		/* FUA is masked off if unsupported and can't be the reason */
3546 3547 3548
		btrfs_handle_fs_error(fs_info, -EIO,
				      "%d errors while writing supers",
				      total_errors);
3549
		return -EIO;
3550
	}
3551

Y
Yan Zheng 已提交
3552
	total_errors = 0;
3553
	list_for_each_entry(dev, head, dev_list) {
3554 3555
		if (!dev->bdev)
			continue;
3556
		if (!test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &dev->dev_state) ||
3557
		    !test_bit(BTRFS_DEV_STATE_WRITEABLE, &dev->dev_state))
3558 3559
			continue;

3560
		ret = wait_dev_supers(dev, max_mirrors);
Y
Yan Zheng 已提交
3561 3562
		if (ret)
			total_errors++;
3563
	}
3564
	mutex_unlock(&fs_info->fs_devices->device_list_mutex);
3565
	if (total_errors > max_errors) {
3566 3567 3568
		btrfs_handle_fs_error(fs_info, -EIO,
				      "%d errors while writing supers",
				      total_errors);
3569
		return -EIO;
3570
	}
3571 3572 3573
	return 0;
}

3574 3575 3576
/* 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
3577
{
3578
	spin_lock(&fs_info->fs_roots_radix_lock);
C
Chris Mason 已提交
3579 3580
	radix_tree_delete(&fs_info->fs_roots_radix,
			  (unsigned long)root->root_key.objectid);
3581
	spin_unlock(&fs_info->fs_roots_radix_lock);
3582 3583 3584 3585

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

L
Liu Bo 已提交
3586
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
L
Liu Bo 已提交
3587
		btrfs_free_log(NULL, root);
L
Liu Bo 已提交
3588 3589 3590 3591 3592 3593 3594
		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
3595

3596 3597 3598 3599
	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);
3600 3601 3602 3603 3604
	free_fs_root(root);
}

static void free_fs_root(struct btrfs_root *root)
{
3605
	iput(root->ino_cache_inode);
3606
	WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree));
3607
	btrfs_free_block_rsv(root->fs_info, root->orphan_block_rsv);
3608
	root->orphan_block_rsv = NULL;
3609 3610
	if (root->anon_dev)
		free_anon_bdev(root->anon_dev);
3611 3612
	if (root->subv_writers)
		btrfs_free_subvolume_writers(root->subv_writers);
3613 3614
	free_extent_buffer(root->node);
	free_extent_buffer(root->commit_root);
3615 3616
	kfree(root->free_ino_ctl);
	kfree(root->free_ino_pinned);
C
Chris Mason 已提交
3617
	kfree(root->name);
3618
	btrfs_put_fs_root(root);
C
Chris Mason 已提交
3619 3620
}

3621 3622 3623
void btrfs_free_fs_root(struct btrfs_root *root)
{
	free_fs_root(root);
C
Chris Mason 已提交
3624 3625
}

Y
Yan Zheng 已提交
3626
int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
C
Chris Mason 已提交
3627
{
Y
Yan Zheng 已提交
3628 3629
	u64 root_objectid = 0;
	struct btrfs_root *gang[8];
3630 3631 3632 3633
	int i = 0;
	int err = 0;
	unsigned int ret = 0;
	int index;
3634

Y
Yan Zheng 已提交
3635
	while (1) {
3636
		index = srcu_read_lock(&fs_info->subvol_srcu);
Y
Yan Zheng 已提交
3637 3638 3639
		ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix,
					     (void **)gang, root_objectid,
					     ARRAY_SIZE(gang));
3640 3641
		if (!ret) {
			srcu_read_unlock(&fs_info->subvol_srcu, index);
Y
Yan Zheng 已提交
3642
			break;
3643
		}
3644
		root_objectid = gang[ret - 1]->root_key.objectid + 1;
3645

Y
Yan Zheng 已提交
3646
		for (i = 0; i < ret; i++) {
3647 3648 3649 3650 3651 3652 3653 3654 3655
			/* 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);
3656

3657 3658 3659
		for (i = 0; i < ret; i++) {
			if (!gang[i])
				continue;
Y
Yan Zheng 已提交
3660
			root_objectid = gang[i]->root_key.objectid;
3661 3662
			err = btrfs_orphan_cleanup(gang[i]);
			if (err)
3663 3664
				break;
			btrfs_put_fs_root(gang[i]);
Y
Yan Zheng 已提交
3665 3666 3667
		}
		root_objectid++;
	}
3668 3669 3670 3671 3672 3673 3674

	/* 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
3675
}
3676

3677
int btrfs_commit_super(struct btrfs_fs_info *fs_info)
Y
Yan Zheng 已提交
3678
{
3679
	struct btrfs_root *root = fs_info->tree_root;
Y
Yan Zheng 已提交
3680
	struct btrfs_trans_handle *trans;
3681

3682
	mutex_lock(&fs_info->cleaner_mutex);
3683
	btrfs_run_delayed_iputs(fs_info);
3684 3685
	mutex_unlock(&fs_info->cleaner_mutex);
	wake_up_process(fs_info->cleaner_kthread);
3686 3687

	/* wait until ongoing cleanup work done */
3688 3689
	down_write(&fs_info->cleanup_work_sem);
	up_write(&fs_info->cleanup_work_sem);
3690

3691
	trans = btrfs_join_transaction(root);
3692 3693
	if (IS_ERR(trans))
		return PTR_ERR(trans);
3694
	return btrfs_commit_transaction(trans);
Y
Yan Zheng 已提交
3695 3696
}

3697
void close_ctree(struct btrfs_fs_info *fs_info)
Y
Yan Zheng 已提交
3698
{
3699
	struct btrfs_root *root = fs_info->tree_root;
Y
Yan Zheng 已提交
3700 3701
	int ret;

3702
	set_bit(BTRFS_FS_CLOSING_START, &fs_info->flags);
Y
Yan Zheng 已提交
3703

3704
	/* wait for the qgroup rescan worker to stop */
3705
	btrfs_qgroup_wait_for_completion(fs_info, false);
3706

S
Stefan Behrens 已提交
3707 3708 3709 3710 3711
	/* 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);

3712
	/* pause restriper - we want to resume on mount */
3713
	btrfs_pause_balance(fs_info);
3714

3715 3716
	btrfs_dev_replace_suspend_for_unmount(fs_info);

3717
	btrfs_scrub_cancel(fs_info);
C
Chris Mason 已提交
3718 3719 3720 3721 3722 3723

	/* 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 */
3724
	btrfs_cleanup_defrag_inodes(fs_info);
C
Chris Mason 已提交
3725

3726 3727
	cancel_work_sync(&fs_info->async_reclaim_work);

3728
	if (!sb_rdonly(fs_info->sb)) {
3729 3730 3731 3732 3733
		/*
		 * 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.
		 */
3734
		btrfs_delete_unused_bgs(fs_info);
3735

3736
		ret = btrfs_commit_super(fs_info);
L
liubo 已提交
3737
		if (ret)
3738
			btrfs_err(fs_info, "commit super ret %d", ret);
L
liubo 已提交
3739 3740
	}

3741
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state))
3742
		btrfs_error_commit_super(fs_info);
3743

A
Al Viro 已提交
3744 3745
	kthread_stop(fs_info->transaction_kthread);
	kthread_stop(fs_info->cleaner_kthread);
3746

3747
	set_bit(BTRFS_FS_CLOSING_DONE, &fs_info->flags);
3748

3749
	btrfs_free_qgroup_config(fs_info);
3750

3751
	if (percpu_counter_sum(&fs_info->delalloc_bytes)) {
3752
		btrfs_info(fs_info, "at unmount delalloc count %lld",
3753
		       percpu_counter_sum(&fs_info->delalloc_bytes));
C
Chris Mason 已提交
3754
	}
3755

3756
	btrfs_sysfs_remove_mounted(fs_info);
3757
	btrfs_sysfs_remove_fsid(fs_info->fs_devices);
3758

3759
	btrfs_free_fs_roots(fs_info);
3760

3761 3762
	btrfs_put_block_group_cache(fs_info);

3763 3764 3765 3766 3767
	/*
	 * 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);
3768 3769
	btrfs_stop_all_workers(fs_info);

3770 3771
	btrfs_free_block_groups(fs_info);

3772
	clear_bit(BTRFS_FS_OPEN, &fs_info->flags);
3773
	free_root_pointers(fs_info, 1);
3774

3775
	iput(fs_info->btree_inode);
3776

3777
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
3778
	if (btrfs_test_opt(fs_info, CHECK_INTEGRITY))
3779
		btrfsic_unmount(fs_info->fs_devices);
3780 3781
#endif

3782
	btrfs_close_devices(fs_info->fs_devices);
3783
	btrfs_mapping_tree_free(&fs_info->mapping_tree);
3784

3785
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
3786
	percpu_counter_destroy(&fs_info->delalloc_bytes);
3787
	percpu_counter_destroy(&fs_info->bio_counter);
3788
	cleanup_srcu_struct(&fs_info->subvol_srcu);
3789

D
David Woodhouse 已提交
3790
	btrfs_free_stripe_hash_table(fs_info);
J
Josef Bacik 已提交
3791
	btrfs_free_ref_cache(fs_info);
D
David Woodhouse 已提交
3792

3793
	__btrfs_free_block_rsv(root->orphan_block_rsv);
3794
	root->orphan_block_rsv = NULL;
3795 3796 3797 3798 3799 3800 3801 3802 3803

	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);
	}
3804 3805
}

3806 3807
int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid,
			  int atomic)
3808
{
3809
	int ret;
3810
	struct inode *btree_inode = buf->pages[0]->mapping->host;
3811

3812
	ret = extent_buffer_uptodate(buf);
3813 3814 3815 3816
	if (!ret)
		return ret;

	ret = verify_parent_transid(&BTRFS_I(btree_inode)->io_tree, buf,
3817 3818 3819
				    parent_transid, atomic);
	if (ret == -EAGAIN)
		return ret;
3820
	return !ret;
3821 3822 3823 3824
}

void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
{
3825
	struct btrfs_fs_info *fs_info;
3826
	struct btrfs_root *root;
3827
	u64 transid = btrfs_header_generation(buf);
3828
	int was_dirty;
3829

3830 3831 3832 3833 3834 3835 3836 3837 3838 3839
#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;
3840
	fs_info = root->fs_info;
3841
	btrfs_assert_tree_locked(buf);
3842
	if (transid != fs_info->generation)
J
Jeff Mahoney 已提交
3843
		WARN(1, KERN_CRIT "btrfs transid mismatch buffer %llu, found %llu running %llu\n",
3844
			buf->start, transid, fs_info->generation);
3845
	was_dirty = set_extent_buffer_dirty(buf);
3846
	if (!was_dirty)
3847 3848 3849
		percpu_counter_add_batch(&fs_info->dirty_metadata_bytes,
					 buf->len,
					 fs_info->dirty_metadata_batch);
3850
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
3851 3852 3853 3854 3855 3856
	/*
	 * Since btrfs_mark_buffer_dirty() can be called with item pointer set
	 * but item data not updated.
	 * So here we should only check item pointers, not item data.
	 */
	if (btrfs_header_level(buf) == 0 &&
3857
	    btrfs_check_leaf_relaxed(fs_info, buf)) {
3858
		btrfs_print_leaf(buf);
3859 3860 3861
		ASSERT(0);
	}
#endif
3862 3863
}

3864
static void __btrfs_btree_balance_dirty(struct btrfs_fs_info *fs_info,
3865
					int flush_delayed)
3866 3867 3868 3869 3870
{
	/*
	 * looks as though older kernels can get into trouble with
	 * this code, they end up stuck in balance_dirty_pages forever
	 */
3871
	int ret;
3872 3873 3874 3875

	if (current->flags & PF_MEMALLOC)
		return;

3876
	if (flush_delayed)
3877
		btrfs_balance_delayed_items(fs_info);
3878

3879
	ret = percpu_counter_compare(&fs_info->dirty_metadata_bytes,
3880 3881
				     BTRFS_DIRTY_METADATA_THRESH);
	if (ret > 0) {
3882
		balance_dirty_pages_ratelimited(fs_info->btree_inode->i_mapping);
3883 3884 3885
	}
}

3886
void btrfs_btree_balance_dirty(struct btrfs_fs_info *fs_info)
C
Chris Mason 已提交
3887
{
3888
	__btrfs_btree_balance_dirty(fs_info, 1);
3889
}
3890

3891
void btrfs_btree_balance_dirty_nodelay(struct btrfs_fs_info *fs_info)
3892
{
3893
	__btrfs_btree_balance_dirty(fs_info, 0);
C
Chris Mason 已提交
3894
}
3895

3896
int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid)
3897
{
3898
	struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root;
3899 3900 3901
	struct btrfs_fs_info *fs_info = root->fs_info;

	return btree_read_extent_buffer_pages(fs_info, buf, parent_transid);
3902
}
3903

3904
static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info)
L
liubo 已提交
3905
{
D
David Sterba 已提交
3906
	struct btrfs_super_block *sb = fs_info->super_copy;
3907 3908
	u64 nodesize = btrfs_super_nodesize(sb);
	u64 sectorsize = btrfs_super_sectorsize(sb);
D
David Sterba 已提交
3909 3910
	int ret = 0;

3911
	if (btrfs_super_magic(sb) != BTRFS_MAGIC) {
3912
		btrfs_err(fs_info, "no valid FS found");
3913 3914
		ret = -EINVAL;
	}
3915 3916
	if (btrfs_super_flags(sb) & ~BTRFS_SUPER_FLAG_SUPP) {
		btrfs_err(fs_info, "unrecognized or unsupported super flag: %llu",
3917
				btrfs_super_flags(sb) & ~BTRFS_SUPER_FLAG_SUPP);
3918 3919
		ret = -EINVAL;
	}
3920
	if (btrfs_super_root_level(sb) >= BTRFS_MAX_LEVEL) {
3921
		btrfs_err(fs_info, "tree_root level too big: %d >= %d",
3922
				btrfs_super_root_level(sb), BTRFS_MAX_LEVEL);
D
David Sterba 已提交
3923 3924
		ret = -EINVAL;
	}
3925
	if (btrfs_super_chunk_root_level(sb) >= BTRFS_MAX_LEVEL) {
3926
		btrfs_err(fs_info, "chunk_root level too big: %d >= %d",
3927
				btrfs_super_chunk_root_level(sb), BTRFS_MAX_LEVEL);
D
David Sterba 已提交
3928 3929
		ret = -EINVAL;
	}
3930
	if (btrfs_super_log_root_level(sb) >= BTRFS_MAX_LEVEL) {
3931
		btrfs_err(fs_info, "log_root level too big: %d >= %d",
3932
				btrfs_super_log_root_level(sb), BTRFS_MAX_LEVEL);
D
David Sterba 已提交
3933 3934 3935
		ret = -EINVAL;
	}

D
David Sterba 已提交
3936
	/*
3937 3938
	 * 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
3939
	 */
3940 3941
	if (!is_power_of_2(sectorsize) || sectorsize < 4096 ||
	    sectorsize > BTRFS_MAX_METADATA_BLOCKSIZE) {
3942
		btrfs_err(fs_info, "invalid sectorsize %llu", sectorsize);
3943 3944 3945
		ret = -EINVAL;
	}
	/* Only PAGE SIZE is supported yet */
3946
	if (sectorsize != PAGE_SIZE) {
3947 3948 3949
		btrfs_err(fs_info,
			"sectorsize %llu not supported yet, only support %lu",
			sectorsize, PAGE_SIZE);
3950 3951 3952 3953
		ret = -EINVAL;
	}
	if (!is_power_of_2(nodesize) || nodesize < sectorsize ||
	    nodesize > BTRFS_MAX_METADATA_BLOCKSIZE) {
3954
		btrfs_err(fs_info, "invalid nodesize %llu", nodesize);
3955 3956 3957
		ret = -EINVAL;
	}
	if (nodesize != le32_to_cpu(sb->__unused_leafsize)) {
3958 3959
		btrfs_err(fs_info, "invalid leafsize %u, should be %llu",
			  le32_to_cpu(sb->__unused_leafsize), nodesize);
3960 3961 3962 3963 3964
		ret = -EINVAL;
	}

	/* Root alignment check */
	if (!IS_ALIGNED(btrfs_super_root(sb), sectorsize)) {
3965 3966
		btrfs_warn(fs_info, "tree_root block unaligned: %llu",
			   btrfs_super_root(sb));
3967 3968 3969
		ret = -EINVAL;
	}
	if (!IS_ALIGNED(btrfs_super_chunk_root(sb), sectorsize)) {
3970 3971
		btrfs_warn(fs_info, "chunk_root block unaligned: %llu",
			   btrfs_super_chunk_root(sb));
3972 3973
		ret = -EINVAL;
	}
3974
	if (!IS_ALIGNED(btrfs_super_log_root(sb), sectorsize)) {
3975 3976
		btrfs_warn(fs_info, "log_root block unaligned: %llu",
			   btrfs_super_log_root(sb));
3977 3978 3979
		ret = -EINVAL;
	}

3980
	if (memcmp(fs_info->fsid, sb->dev_item.fsid, BTRFS_FSID_SIZE) != 0) {
3981 3982 3983
		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
3984 3985 3986 3987 3988 3989 3990
		ret = -EINVAL;
	}

	/*
	 * Hint to catch really bogus numbers, bitflips or so, more exact checks are
	 * done later
	 */
3991 3992
	if (btrfs_super_bytes_used(sb) < 6 * btrfs_super_nodesize(sb)) {
		btrfs_err(fs_info, "bytes_used is too small %llu",
3993
			  btrfs_super_bytes_used(sb));
3994 3995
		ret = -EINVAL;
	}
3996
	if (!is_power_of_2(btrfs_super_stripesize(sb))) {
3997
		btrfs_err(fs_info, "invalid stripesize %u",
3998
			  btrfs_super_stripesize(sb));
3999 4000
		ret = -EINVAL;
	}
4001
	if (btrfs_super_num_devices(sb) > (1UL << 31))
4002 4003
		btrfs_warn(fs_info, "suspicious number of devices: %llu",
			   btrfs_super_num_devices(sb));
4004
	if (btrfs_super_num_devices(sb) == 0) {
4005
		btrfs_err(fs_info, "number of devices is 0");
4006 4007
		ret = -EINVAL;
	}
D
David Sterba 已提交
4008

4009
	if (btrfs_super_bytenr(sb) != BTRFS_SUPER_INFO_OFFSET) {
4010 4011
		btrfs_err(fs_info, "super offset mismatch %llu != %u",
			  btrfs_super_bytenr(sb), BTRFS_SUPER_INFO_OFFSET);
D
David Sterba 已提交
4012 4013 4014
		ret = -EINVAL;
	}

4015 4016 4017 4018 4019
	/*
	 * 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) {
4020 4021 4022
		btrfs_err(fs_info, "system chunk array too big %u > %u",
			  btrfs_super_sys_array_size(sb),
			  BTRFS_SYSTEM_CHUNK_ARRAY_SIZE);
4023 4024 4025 4026
		ret = -EINVAL;
	}
	if (btrfs_super_sys_array_size(sb) < sizeof(struct btrfs_disk_key)
			+ sizeof(struct btrfs_chunk)) {
4027 4028 4029 4030
		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));
4031 4032 4033
		ret = -EINVAL;
	}

D
David Sterba 已提交
4034 4035 4036 4037
	/*
	 * 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.
	 */
4038
	if (btrfs_super_generation(sb) < btrfs_super_chunk_root_generation(sb))
4039 4040 4041 4042
		btrfs_warn(fs_info,
			"suspicious: generation < chunk_root_generation: %llu < %llu",
			btrfs_super_generation(sb),
			btrfs_super_chunk_root_generation(sb));
4043 4044
	if (btrfs_super_generation(sb) < btrfs_super_cache_generation(sb)
	    && btrfs_super_cache_generation(sb) != (u64)-1)
4045 4046 4047 4048
		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
4049 4050

	return ret;
L
liubo 已提交
4051 4052
}

4053
static void btrfs_error_commit_super(struct btrfs_fs_info *fs_info)
L
liubo 已提交
4054
{
4055
	mutex_lock(&fs_info->cleaner_mutex);
4056
	btrfs_run_delayed_iputs(fs_info);
4057
	mutex_unlock(&fs_info->cleaner_mutex);
L
liubo 已提交
4058

4059 4060
	down_write(&fs_info->cleanup_work_sem);
	up_write(&fs_info->cleanup_work_sem);
L
liubo 已提交
4061 4062

	/* cleanup FS via transaction */
4063
	btrfs_cleanup_transaction(fs_info);
L
liubo 已提交
4064 4065
}

4066
static void btrfs_destroy_ordered_extents(struct btrfs_root *root)
L
liubo 已提交
4067 4068 4069
{
	struct btrfs_ordered_extent *ordered;

4070
	spin_lock(&root->ordered_extent_lock);
4071 4072 4073 4074
	/*
	 * This will just short circuit the ordered completion stuff which will
	 * make sure the ordered extent gets properly cleaned up.
	 */
4075
	list_for_each_entry(ordered, &root->ordered_extents,
4076 4077
			    root_extent_list)
		set_bit(BTRFS_ORDERED_IOERR, &ordered->flags);
4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092
	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);
4093 4094
		list_move_tail(&root->ordered_root,
			       &fs_info->ordered_roots);
4095

4096
		spin_unlock(&fs_info->ordered_root_lock);
4097 4098
		btrfs_destroy_ordered_extents(root);

4099 4100
		cond_resched();
		spin_lock(&fs_info->ordered_root_lock);
4101 4102
	}
	spin_unlock(&fs_info->ordered_root_lock);
L
liubo 已提交
4103 4104
}

4105
static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
4106
				      struct btrfs_fs_info *fs_info)
L
liubo 已提交
4107 4108 4109 4110 4111 4112 4113 4114 4115
{
	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);
4116
	if (atomic_read(&delayed_refs->num_entries) == 0) {
4117
		spin_unlock(&delayed_refs->lock);
4118
		btrfs_info(fs_info, "delayed_refs has NO entry");
L
liubo 已提交
4119 4120 4121
		return ret;
	}

4122 4123
	while ((node = rb_first(&delayed_refs->href_root)) != NULL) {
		struct btrfs_delayed_ref_head *head;
4124
		struct rb_node *n;
4125
		bool pin_bytes = false;
L
liubo 已提交
4126

4127 4128 4129
		head = rb_entry(node, struct btrfs_delayed_ref_head,
				href_node);
		if (!mutex_trylock(&head->mutex)) {
4130
			refcount_inc(&head->refs);
4131
			spin_unlock(&delayed_refs->lock);
4132

4133
			mutex_lock(&head->mutex);
4134
			mutex_unlock(&head->mutex);
4135
			btrfs_put_delayed_ref_head(head);
4136 4137 4138 4139
			spin_lock(&delayed_refs->lock);
			continue;
		}
		spin_lock(&head->lock);
4140 4141 4142
		while ((n = rb_first(&head->ref_tree)) != NULL) {
			ref = rb_entry(n, struct btrfs_delayed_ref_node,
				       ref_node);
4143
			ref->in_tree = 0;
4144 4145
			rb_erase(&ref->ref_node, &head->ref_tree);
			RB_CLEAR_NODE(&ref->ref_node);
4146 4147
			if (!list_empty(&ref->add_list))
				list_del(&ref->add_list);
4148 4149
			atomic_dec(&delayed_refs->num_entries);
			btrfs_put_delayed_ref(ref);
4150
		}
4151 4152 4153 4154 4155 4156 4157 4158
		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);
		rb_erase(&head->href_node, &delayed_refs->href_root);
4159
		RB_CLEAR_NODE(&head->href_node);
4160 4161 4162
		spin_unlock(&head->lock);
		spin_unlock(&delayed_refs->lock);
		mutex_unlock(&head->mutex);
L
liubo 已提交
4163

4164
		if (pin_bytes)
4165 4166 4167
			btrfs_pin_extent(fs_info, head->bytenr,
					 head->num_bytes, 1);
		btrfs_put_delayed_ref_head(head);
L
liubo 已提交
4168 4169 4170 4171 4172 4173 4174 4175 4176
		cond_resched();
		spin_lock(&delayed_refs->lock);
	}

	spin_unlock(&delayed_refs->lock);

	return ret;
}

4177
static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root)
L
liubo 已提交
4178 4179 4180 4181 4182 4183
{
	struct btrfs_inode *btrfs_inode;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

4184 4185
	spin_lock(&root->delalloc_lock);
	list_splice_init(&root->delalloc_inodes, &splice);
L
liubo 已提交
4186 4187

	while (!list_empty(&splice)) {
4188 4189
		btrfs_inode = list_first_entry(&splice, struct btrfs_inode,
					       delalloc_inodes);
L
liubo 已提交
4190 4191

		list_del_init(&btrfs_inode->delalloc_inodes);
4192 4193
		clear_bit(BTRFS_INODE_IN_DELALLOC_LIST,
			  &btrfs_inode->runtime_flags);
4194
		spin_unlock(&root->delalloc_lock);
L
liubo 已提交
4195 4196

		btrfs_invalidate_inodes(btrfs_inode->root);
4197

4198
		spin_lock(&root->delalloc_lock);
L
liubo 已提交
4199 4200
	}

4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224 4225 4226
	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
4227 4228
}

4229
static int btrfs_destroy_marked_extents(struct btrfs_fs_info *fs_info,
L
liubo 已提交
4230 4231 4232 4233 4234 4235 4236 4237 4238 4239
					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,
4240
					    mark, NULL);
L
liubo 已提交
4241 4242 4243
		if (ret)
			break;

4244
		clear_extent_bits(dirty_pages, start, end, mark);
L
liubo 已提交
4245
		while (start <= end) {
4246 4247
			eb = find_extent_buffer(fs_info, start);
			start += fs_info->nodesize;
4248
			if (!eb)
L
liubo 已提交
4249
				continue;
4250
			wait_on_extent_buffer_writeback(eb);
L
liubo 已提交
4251

4252 4253 4254 4255
			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
4256 4257 4258 4259 4260 4261
		}
	}

	return ret;
}

4262
static int btrfs_destroy_pinned_extent(struct btrfs_fs_info *fs_info,
L
liubo 已提交
4263 4264 4265 4266 4267 4268
				       struct extent_io_tree *pinned_extents)
{
	struct extent_io_tree *unpin;
	u64 start;
	u64 end;
	int ret;
4269
	bool loop = true;
L
liubo 已提交
4270 4271

	unpin = pinned_extents;
4272
again:
L
liubo 已提交
4273 4274
	while (1) {
		ret = find_first_extent_bit(unpin, 0, &start, &end,
4275
					    EXTENT_DIRTY, NULL);
L
liubo 已提交
4276 4277 4278
		if (ret)
			break;

4279
		clear_extent_dirty(unpin, start, end);
4280
		btrfs_error_unpin_extent_range(fs_info, start, end);
L
liubo 已提交
4281 4282 4283
		cond_resched();
	}

4284
	if (loop) {
4285 4286
		if (unpin == &fs_info->freed_extents[0])
			unpin = &fs_info->freed_extents[1];
4287
		else
4288
			unpin = &fs_info->freed_extents[0];
4289 4290 4291 4292
		loop = false;
		goto again;
	}

L
liubo 已提交
4293 4294 4295
	return 0;
}

4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309 4310
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,
4311
			     struct btrfs_fs_info *fs_info)
4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338
{
	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 (!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);

4339 4340 4341 4342
	/*
	 * Refer to the definition of io_bgs member for details why it's safe
	 * to use it without any locking
	 */
4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355
	while (!list_empty(&cur_trans->io_bgs)) {
		cache = list_first_entry(&cur_trans->io_bgs,
					 struct btrfs_block_group_cache,
					 io_list);

		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);
	}
}

4356
void btrfs_cleanup_one_transaction(struct btrfs_transaction *cur_trans,
4357
				   struct btrfs_fs_info *fs_info)
4358
{
4359
	btrfs_cleanup_dirty_bgs(cur_trans, fs_info);
4360 4361 4362
	ASSERT(list_empty(&cur_trans->dirty_bgs));
	ASSERT(list_empty(&cur_trans->io_bgs));

4363
	btrfs_destroy_delayed_refs(cur_trans, fs_info);
4364

4365
	cur_trans->state = TRANS_STATE_COMMIT_START;
4366
	wake_up(&fs_info->transaction_blocked_wait);
4367

4368
	cur_trans->state = TRANS_STATE_UNBLOCKED;
4369
	wake_up(&fs_info->transaction_wait);
4370

4371 4372
	btrfs_destroy_delayed_inodes(fs_info);
	btrfs_assert_delayed_root_empty(fs_info);
4373

4374
	btrfs_destroy_marked_extents(fs_info, &cur_trans->dirty_pages,
4375
				     EXTENT_DIRTY);
4376
	btrfs_destroy_pinned_extent(fs_info,
4377
				    fs_info->pinned_extents);
4378

4379 4380
	cur_trans->state =TRANS_STATE_COMPLETED;
	wake_up(&cur_trans->commit_wait);
4381 4382
}

4383
static int btrfs_cleanup_transaction(struct btrfs_fs_info *fs_info)
L
liubo 已提交
4384 4385 4386
{
	struct btrfs_transaction *t;

4387
	mutex_lock(&fs_info->transaction_kthread_mutex);
L
liubo 已提交
4388

4389 4390 4391
	spin_lock(&fs_info->trans_lock);
	while (!list_empty(&fs_info->trans_list)) {
		t = list_first_entry(&fs_info->trans_list,
4392 4393
				     struct btrfs_transaction, list);
		if (t->state >= TRANS_STATE_COMMIT_START) {
4394
			refcount_inc(&t->use_count);
4395
			spin_unlock(&fs_info->trans_lock);
4396
			btrfs_wait_for_commit(fs_info, t->transid);
4397
			btrfs_put_transaction(t);
4398
			spin_lock(&fs_info->trans_lock);
4399 4400
			continue;
		}
4401
		if (t == fs_info->running_transaction) {
4402
			t->state = TRANS_STATE_COMMIT_DOING;
4403
			spin_unlock(&fs_info->trans_lock);
4404 4405 4406 4407 4408 4409 4410
			/*
			 * 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 {
4411
			spin_unlock(&fs_info->trans_lock);
4412
		}
4413
		btrfs_cleanup_one_transaction(t, fs_info);
4414

4415 4416 4417
		spin_lock(&fs_info->trans_lock);
		if (t == fs_info->running_transaction)
			fs_info->running_transaction = NULL;
L
liubo 已提交
4418
		list_del_init(&t->list);
4419
		spin_unlock(&fs_info->trans_lock);
L
liubo 已提交
4420

4421
		btrfs_put_transaction(t);
4422
		trace_btrfs_transaction_commit(fs_info->tree_root);
4423
		spin_lock(&fs_info->trans_lock);
4424
	}
4425 4426
	spin_unlock(&fs_info->trans_lock);
	btrfs_destroy_all_ordered_extents(fs_info);
4427 4428
	btrfs_destroy_delayed_inodes(fs_info);
	btrfs_assert_delayed_root_empty(fs_info);
4429
	btrfs_destroy_pinned_extent(fs_info, fs_info->pinned_extents);
4430 4431
	btrfs_destroy_all_delalloc_inodes(fs_info);
	mutex_unlock(&fs_info->transaction_kthread_mutex);
L
liubo 已提交
4432 4433 4434 4435

	return 0;
}

4436 4437 4438 4439 4440 4441
static struct btrfs_fs_info *btree_fs_info(void *private_data)
{
	struct inode *inode = private_data;
	return btrfs_sb(inode->i_sb);
}

4442
static const struct extent_io_ops btree_extent_io_ops = {
4443
	/* mandatory callbacks */
4444
	.submit_bio_hook = btree_submit_bio_hook,
4445
	.readpage_end_io_hook = btree_readpage_end_io_hook,
4446 4447
	/* note we're sharing with inode.c for the merge bio hook */
	.merge_bio_hook = btrfs_merge_bio_hook,
4448
	.readpage_io_failed_hook = btree_io_failed_hook,
4449 4450
	.set_range_writeback = btrfs_set_range_writeback,
	.tree_fs_info = btree_fs_info,
4451 4452

	/* optional callbacks */
4453
};