disk-io.c 123.2 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 <linux/crc32c.h>
35
#include <asm/unaligned.h>
36 37
#include "ctree.h"
#include "disk-io.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
					NULL);
	if (!btrfs_end_io_wq_cache)
		return -ENOMEM;
	return 0;
}

113
void __cold btrfs_end_io_wq_exit(void)
114
{
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;
127 128
	extent_submit_bio_start_t *submit_bio_start;
	extent_submit_bio_done_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 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
{
	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;
406
		char result[sizeof(crc)];
D
David Sterba 已提交
407 408 409 410

		/*
		 * The super_block structure does not span the whole
		 * BTRFS_SUPER_INFO_SIZE range, we expect that the unused space
411
		 * is filled with zeros and is included in the checksum.
D
David Sterba 已提交
412 413 414 415 416
		 */
		crc = btrfs_csum_data(raw_disk_sb + BTRFS_CSUM_SIZE,
				crc, BTRFS_SUPER_INFO_SIZE - BTRFS_CSUM_SIZE);
		btrfs_csum_final(crc, result);

417
		if (memcmp(raw_disk_sb, result, sizeof(result)))
D
David Sterba 已提交
418 419 420 421
			ret = 1;
	}

	if (csum_type >= ARRAY_SIZE(btrfs_csum_sizes)) {
422
		btrfs_err(fs_info, "unsupported checksum algorithm %u",
D
David Sterba 已提交
423 424 425 426 427 428 429
				csum_type);
		ret = 1;
	}

	return ret;
}

430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470
static int verify_level_key(struct btrfs_fs_info *fs_info,
			    struct extent_buffer *eb, int level,
			    struct btrfs_key *first_key)
{
	int found_level;
	struct btrfs_key found_key;
	int ret;

	found_level = btrfs_header_level(eb);
	if (found_level != level) {
#ifdef CONFIG_BTRFS_DEBUG
		WARN_ON(1);
		btrfs_err(fs_info,
"tree level mismatch detected, bytenr=%llu level expected=%u has=%u",
			  eb->start, level, found_level);
#endif
		return -EIO;
	}

	if (!first_key)
		return 0;

	if (found_level)
		btrfs_node_key_to_cpu(eb, &found_key, 0);
	else
		btrfs_item_key_to_cpu(eb, &found_key, 0);
	ret = btrfs_comp_cpu_keys(first_key, &found_key);

#ifdef CONFIG_BTRFS_DEBUG
	if (ret) {
		WARN_ON(1);
		btrfs_err(fs_info,
"tree first key mismatch detected, bytenr=%llu key expected=(%llu, %u, %llu) has=(%llu, %u, %llu)",
			  eb->start, first_key->objectid, first_key->type,
			  first_key->offset, found_key.objectid,
			  found_key.type, found_key.offset);
	}
#endif
	return ret;
}

C
Chris Mason 已提交
471 472 473
/*
 * helper to read a given tree block, doing retries as required when
 * the checksums don't match and we have alternate mirrors to try.
474 475 476 477
 *
 * @parent_transid:	expected transid, skip check if 0
 * @level:		expected level, mandatory check
 * @first_key:		expected key of first slot, skip check if NULL
C
Chris Mason 已提交
478
 */
479
static int btree_read_extent_buffer_pages(struct btrfs_fs_info *fs_info,
480
					  struct extent_buffer *eb,
481 482
					  u64 parent_transid, int level,
					  struct btrfs_key *first_key)
483 484
{
	struct extent_io_tree *io_tree;
485
	int failed = 0;
486 487 488
	int ret;
	int num_copies = 0;
	int mirror_num = 0;
489
	int failed_mirror = 0;
490

491
	clear_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
492
	io_tree = &BTRFS_I(fs_info->btree_inode)->io_tree;
493
	while (1) {
494
		ret = read_extent_buffer_pages(io_tree, eb, WAIT_COMPLETE,
495
					       mirror_num);
496
		if (!ret) {
497
			if (verify_parent_transid(io_tree, eb,
498
						   parent_transid, 0))
499
				ret = -EIO;
500 501 502 503 504
			else if (verify_level_key(fs_info, eb, level,
						  first_key))
				ret = -EUCLEAN;
			else
				break;
505
		}
C
Chris Mason 已提交
506

507 508 509 510 511
		/*
		 * 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.
		 */
512 513
		if (test_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags) ||
		    ret == -EUCLEAN)
514 515
			break;

516
		num_copies = btrfs_num_copies(fs_info,
517
					      eb->start, eb->len);
C
Chris Mason 已提交
518
		if (num_copies == 1)
519
			break;
C
Chris Mason 已提交
520

521 522 523 524 525
		if (!failed_mirror) {
			failed = 1;
			failed_mirror = eb->read_mirror;
		}

526
		mirror_num++;
527 528 529
		if (mirror_num == failed_mirror)
			mirror_num++;

C
Chris Mason 已提交
530
		if (mirror_num > num_copies)
531
			break;
532
	}
533

534
	if (failed && !ret && failed_mirror)
535
		repair_eb_io_failure(fs_info, eb, failed_mirror);
536 537

	return ret;
538
}
539

C
Chris Mason 已提交
540
/*
C
Chris Mason 已提交
541 542
 * 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
543
 */
C
Chris Mason 已提交
544

545
static int csum_dirty_buffer(struct btrfs_fs_info *fs_info, struct page *page)
546
{
M
Miao Xie 已提交
547
	u64 start = page_offset(page);
548 549
	u64 found_start;
	struct extent_buffer *eb;
550

J
Josef Bacik 已提交
551 552 553
	eb = (struct extent_buffer *)page->private;
	if (page != eb->pages[0])
		return 0;
554

555
	found_start = btrfs_header_bytenr(eb);
556 557 558 559 560 561 562 563 564 565 566 567
	/*
	 * 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);

568
	return csum_tree_block(fs_info, eb, 0);
569 570
}

571
static int check_tree_block_fsid(struct btrfs_fs_info *fs_info,
Y
Yan Zheng 已提交
572 573
				 struct extent_buffer *eb)
{
574
	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
575
	u8 fsid[BTRFS_FSID_SIZE];
Y
Yan Zheng 已提交
576 577
	int ret = 1;

578
	read_extent_buffer(eb, fsid, btrfs_header_fsid(), BTRFS_FSID_SIZE);
Y
Yan Zheng 已提交
579 580 581 582 583 584 585 586 587 588
	while (fs_devices) {
		if (!memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE)) {
			ret = 0;
			break;
		}
		fs_devices = fs_devices->seed;
	}
	return ret;
}

589 590 591
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)
592 593 594 595 596
{
	u64 found_start;
	int found_level;
	struct extent_buffer *eb;
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
597
	struct btrfs_fs_info *fs_info = root->fs_info;
598
	int ret = 0;
599
	int reads_done;
600 601 602

	if (!page->private)
		goto out;
C
Chris Mason 已提交
603

J
Josef Bacik 已提交
604
	eb = (struct extent_buffer *)page->private;
C
Chris Mason 已提交
605

606 607 608 609 610 611
	/* 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);
612 613
	if (!reads_done)
		goto err;
614

615
	eb->read_mirror = mirror;
616
	if (test_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags)) {
617 618 619 620
		ret = -EIO;
		goto err;
	}

621
	found_start = btrfs_header_bytenr(eb);
622
	if (found_start != eb->start) {
623 624
		btrfs_err_rl(fs_info, "bad tree block start %llu %llu",
			     found_start, eb->start);
625
		ret = -EIO;
626 627
		goto err;
	}
628 629 630
	if (check_tree_block_fsid(fs_info, eb)) {
		btrfs_err_rl(fs_info, "bad fsid on block %llu",
			     eb->start);
631 632 633
		ret = -EIO;
		goto err;
	}
634
	found_level = btrfs_header_level(eb);
635
	if (found_level >= BTRFS_MAX_LEVEL) {
636 637
		btrfs_err(fs_info, "bad tree block level %d",
			  (int)btrfs_header_level(eb));
638 639 640
		ret = -EIO;
		goto err;
	}
641

642 643
	btrfs_set_buffer_lockdep_class(btrfs_header_owner(eb),
				       eb, found_level);
644

645
	ret = csum_tree_block(fs_info, eb, 1);
646
	if (ret)
647 648 649 650 651 652 653
		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.
	 */
654
	if (found_level == 0 && btrfs_check_leaf_full(fs_info, eb)) {
655 656 657
		set_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
		ret = -EIO;
	}
658

659
	if (found_level > 0 && btrfs_check_node(fs_info, eb))
L
Liu Bo 已提交
660 661
		ret = -EIO;

662 663
	if (!ret)
		set_extent_buffer_uptodate(eb);
664
err:
665 666
	if (reads_done &&
	    test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
667
		btree_readahead_hook(eb, ret);
A
Arne Jansen 已提交
668

D
David Woodhouse 已提交
669 670 671 672 673 674 675
	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);
676
		clear_extent_buffer_uptodate(eb);
D
David Woodhouse 已提交
677
	}
678
	free_extent_buffer(eb);
679
out:
680
	return ret;
681 682
}

683
static int btree_io_failed_hook(struct page *page, int failed_mirror)
A
Arne Jansen 已提交
684 685 686
{
	struct extent_buffer *eb;

J
Josef Bacik 已提交
687
	eb = (struct extent_buffer *)page->private;
688
	set_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags);
689
	eb->read_mirror = failed_mirror;
D
David Woodhouse 已提交
690
	atomic_dec(&eb->io_pages);
691
	if (test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
692
		btree_readahead_hook(eb, -EIO);
A
Arne Jansen 已提交
693 694 695
	return -EIO;	/* we fixed nothing */
}

696
static void end_workqueue_bio(struct bio *bio)
697
{
698
	struct btrfs_end_io_wq *end_io_wq = bio->bi_private;
699
	struct btrfs_fs_info *fs_info;
700 701
	struct btrfs_workqueue *wq;
	btrfs_work_func_t func;
702 703

	fs_info = end_io_wq->info;
704
	end_io_wq->status = bio->bi_status;
705

M
Mike Christie 已提交
706
	if (bio_op(bio) == REQ_OP_WRITE) {
707 708 709 710 711 712 713 714 715 716 717 718 719
		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;
		}
720
	} else {
721 722 723 724 725
		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) {
726 727 728 729 730 731 732 733 734
			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;
		}
735
	}
736 737 738

	btrfs_init_work(&end_io_wq->work, func, end_workqueue_fn, NULL, NULL);
	btrfs_queue_work(wq, &end_io_wq->work);
739 740
}

741
blk_status_t btrfs_bio_wq_end_io(struct btrfs_fs_info *info, struct bio *bio,
742
			enum btrfs_wq_endio_type metadata)
743
{
744
	struct btrfs_end_io_wq *end_io_wq;
745

746
	end_io_wq = kmem_cache_alloc(btrfs_end_io_wq_cache, GFP_NOFS);
747
	if (!end_io_wq)
748
		return BLK_STS_RESOURCE;
749 750 751

	end_io_wq->private = bio->bi_private;
	end_io_wq->end_io = bio->bi_end_io;
752
	end_io_wq->info = info;
753
	end_io_wq->status = 0;
754
	end_io_wq->bio = bio;
755
	end_io_wq->metadata = metadata;
756 757 758

	bio->bi_private = end_io_wq;
	bio->bi_end_io = end_workqueue_bio;
759 760 761
	return 0;
}

C
Chris Mason 已提交
762 763 764
static void run_one_async_start(struct btrfs_work *work)
{
	struct async_submit_bio *async;
765
	blk_status_t ret;
C
Chris Mason 已提交
766 767

	async = container_of(work, struct  async_submit_bio, work);
768
	ret = async->submit_bio_start(async->private_data, async->bio,
769 770
				      async->bio_offset);
	if (ret)
771
		async->status = ret;
C
Chris Mason 已提交
772 773 774
}

static void run_one_async_done(struct btrfs_work *work)
775 776 777 778
{
	struct async_submit_bio *async;

	async = container_of(work, struct  async_submit_bio, work);
779

780
	/* If an error occurred we just want to clean up the bio and move on */
781 782
	if (async->status) {
		async->bio->bi_status = async->status;
783
		bio_endio(async->bio);
784 785 786
		return;
	}

787
	async->submit_bio_done(async->private_data, async->bio, async->mirror_num);
C
Chris Mason 已提交
788 789 790 791 792 793 794
}

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

	async = container_of(work, struct  async_submit_bio, work);
795 796 797
	kfree(async);
}

798 799 800
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,
801 802
				 extent_submit_bio_start_t *submit_bio_start,
				 extent_submit_bio_done_t *submit_bio_done)
803 804 805 806 807
{
	struct async_submit_bio *async;

	async = kmalloc(sizeof(*async), GFP_NOFS);
	if (!async)
808
		return BLK_STS_RESOURCE;
809

810 811
	async->private_data = private_data;
	async->fs_info = fs_info;
812 813
	async->bio = bio;
	async->mirror_num = mirror_num;
C
Chris Mason 已提交
814 815 816
	async->submit_bio_start = submit_bio_start;
	async->submit_bio_done = submit_bio_done;

817
	btrfs_init_work(&async->work, btrfs_worker_helper, run_one_async_start,
818
			run_one_async_done, run_one_async_free);
C
Chris Mason 已提交
819

C
Chris Mason 已提交
820
	async->bio_flags = bio_flags;
821
	async->bio_offset = bio_offset;
822

823
	async->status = 0;
824

825
	if (op_is_sync(bio->bi_opf))
826
		btrfs_set_work_high_priority(&async->work);
827

828
	btrfs_queue_work(fs_info->workers, &async->work);
829 830 831
	return 0;
}

832
static blk_status_t btree_csum_one_bio(struct bio *bio)
833
{
834
	struct bio_vec *bvec;
835
	struct btrfs_root *root;
836
	int i, ret = 0;
837

838
	ASSERT(!bio_flagged(bio, BIO_CLONED));
839
	bio_for_each_segment_all(bvec, bio, i) {
840
		root = BTRFS_I(bvec->bv_page->mapping->host)->root;
841
		ret = csum_dirty_buffer(root->fs_info, bvec->bv_page);
842 843
		if (ret)
			break;
844
	}
845

846
	return errno_to_blk_status(ret);
847 848
}

849
static blk_status_t btree_submit_bio_start(void *private_data, struct bio *bio,
850
					     u64 bio_offset)
851
{
852 853
	/*
	 * when we're called for a write, we're already in the async
854
	 * submission context.  Just jump into btrfs_map_bio
855
	 */
856
	return btree_csum_one_bio(bio);
C
Chris Mason 已提交
857
}
858

859
static blk_status_t btree_submit_bio_done(void *private_data, struct bio *bio,
860
					    int mirror_num)
C
Chris Mason 已提交
861
{
862
	struct inode *inode = private_data;
863
	blk_status_t ret;
864

865
	/*
C
Chris Mason 已提交
866 867
	 * when we're called for a write, we're already in the async
	 * submission context.  Just jump into btrfs_map_bio
868
	 */
869
	ret = btrfs_map_bio(btrfs_sb(inode->i_sb), bio, mirror_num, 1);
870
	if (ret) {
871
		bio->bi_status = ret;
872 873
		bio_endio(bio);
	}
874
	return ret;
875 876
}

877
static int check_async_write(struct btrfs_inode *bi)
878
{
879 880
	if (atomic_read(&bi->sync_writers))
		return 0;
881
#ifdef CONFIG_X86
882
	if (static_cpu_has(X86_FEATURE_XMM4_2))
883 884 885 886 887
		return 0;
#endif
	return 1;
}

888 889 890
static blk_status_t btree_submit_bio_hook(void *private_data, struct bio *bio,
					  int mirror_num, unsigned long bio_flags,
					  u64 bio_offset)
891
{
892
	struct inode *inode = private_data;
893
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
894
	int async = check_async_write(BTRFS_I(inode));
895
	blk_status_t ret;
896

M
Mike Christie 已提交
897
	if (bio_op(bio) != REQ_OP_WRITE) {
C
Chris Mason 已提交
898 899 900 901
		/*
		 * called for a read, do the setup so that checksum validation
		 * can happen in the async kernel threads
		 */
902 903
		ret = btrfs_bio_wq_end_io(fs_info, bio,
					  BTRFS_WQ_ENDIO_METADATA);
904
		if (ret)
905
			goto out_w_error;
906
		ret = btrfs_map_bio(fs_info, bio, mirror_num, 0);
907 908 909
	} else if (!async) {
		ret = btree_csum_one_bio(bio);
		if (ret)
910
			goto out_w_error;
911
		ret = btrfs_map_bio(fs_info, bio, mirror_num, 0);
912 913 914 915 916
	} else {
		/*
		 * kthread helpers are used to submit writes so that
		 * checksumming can happen in parallel across all CPUs
		 */
917 918
		ret = btrfs_wq_submit_bio(fs_info, bio, mirror_num, 0,
					  bio_offset, private_data,
919 920
					  btree_submit_bio_start,
					  btree_submit_bio_done);
921
	}
922

923 924 925 926
	if (ret)
		goto out_w_error;
	return 0;

927
out_w_error:
928
	bio->bi_status = ret;
929
	bio_endio(bio);
930
	return ret;
931 932
}

J
Jan Beulich 已提交
933
#ifdef CONFIG_MIGRATION
934
static int btree_migratepage(struct address_space *mapping,
935 936
			struct page *newpage, struct page *page,
			enum migrate_mode mode)
937 938 939 940 941 942 943 944 945 946 947 948 949 950
{
	/*
	 * 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;
951
	return migrate_page(mapping, newpage, page, mode);
952
}
J
Jan Beulich 已提交
953
#endif
954

955 956 957 958

static int btree_writepages(struct address_space *mapping,
			    struct writeback_control *wbc)
{
959 960 961
	struct btrfs_fs_info *fs_info;
	int ret;

962
	if (wbc->sync_mode == WB_SYNC_NONE) {
963 964 965 966

		if (wbc->for_kupdate)
			return 0;

967
		fs_info = BTRFS_I(mapping->host)->root->fs_info;
968
		/* this is a bit racy, but that's ok */
969 970 971
		ret = percpu_counter_compare(&fs_info->dirty_metadata_bytes,
					     BTRFS_DIRTY_METADATA_THRESH);
		if (ret < 0)
972 973
			return 0;
	}
974
	return btree_write_cache_pages(mapping, wbc);
975 976
}

977
static int btree_readpage(struct file *file, struct page *page)
978
{
979 980
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
981
	return extent_read_full_page(tree, page, btree_get_extent, 0);
982
}
C
Chris Mason 已提交
983

984
static int btree_releasepage(struct page *page, gfp_t gfp_flags)
985
{
986
	if (PageWriteback(page) || PageDirty(page))
C
Chris Mason 已提交
987
		return 0;
988

989
	return try_release_extent_buffer(page);
990 991
}

992 993
static void btree_invalidatepage(struct page *page, unsigned int offset,
				 unsigned int length)
994
{
995 996
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
997 998
	extent_invalidatepage(tree, page, offset);
	btree_releasepage(page, GFP_NOFS);
999
	if (PagePrivate(page)) {
1000 1001 1002
		btrfs_warn(BTRFS_I(page->mapping->host)->root->fs_info,
			   "page private not zero on page %llu",
			   (unsigned long long)page_offset(page));
1003 1004
		ClearPagePrivate(page);
		set_page_private(page, 0);
1005
		put_page(page);
1006
	}
1007 1008
}

1009 1010
static int btree_set_page_dirty(struct page *page)
{
1011
#ifdef DEBUG
1012 1013 1014 1015 1016 1017 1018 1019
	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);
1020
#endif
1021 1022 1023
	return __set_page_dirty_nobuffers(page);
}

1024
static const struct address_space_operations btree_aops = {
1025
	.readpage	= btree_readpage,
1026
	.writepages	= btree_writepages,
1027 1028
	.releasepage	= btree_releasepage,
	.invalidatepage = btree_invalidatepage,
1029
#ifdef CONFIG_MIGRATION
1030
	.migratepage	= btree_migratepage,
1031
#endif
1032
	.set_page_dirty = btree_set_page_dirty,
1033 1034
};

1035
void readahead_tree_block(struct btrfs_fs_info *fs_info, u64 bytenr)
C
Chris Mason 已提交
1036
{
1037
	struct extent_buffer *buf = NULL;
1038
	struct inode *btree_inode = fs_info->btree_inode;
C
Chris Mason 已提交
1039

1040
	buf = btrfs_find_create_tree_block(fs_info, bytenr);
1041
	if (IS_ERR(buf))
1042
		return;
1043
	read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree,
1044
				 buf, WAIT_NONE, 0);
1045
	free_extent_buffer(buf);
C
Chris Mason 已提交
1046 1047
}

1048
int reada_tree_block_flagged(struct btrfs_fs_info *fs_info, u64 bytenr,
1049 1050 1051
			 int mirror_num, struct extent_buffer **eb)
{
	struct extent_buffer *buf = NULL;
1052
	struct inode *btree_inode = fs_info->btree_inode;
1053 1054 1055
	struct extent_io_tree *io_tree = &BTRFS_I(btree_inode)->io_tree;
	int ret;

1056
	buf = btrfs_find_create_tree_block(fs_info, bytenr);
1057
	if (IS_ERR(buf))
1058 1059 1060 1061
		return 0;

	set_bit(EXTENT_BUFFER_READAHEAD, &buf->bflags);

1062
	ret = read_extent_buffer_pages(io_tree, buf, WAIT_PAGE_LOCK,
1063
				       mirror_num);
1064 1065 1066 1067 1068 1069 1070 1071
	if (ret) {
		free_extent_buffer(buf);
		return ret;
	}

	if (test_bit(EXTENT_BUFFER_CORRUPT, &buf->bflags)) {
		free_extent_buffer(buf);
		return -EIO;
1072
	} else if (extent_buffer_uptodate(buf)) {
1073 1074 1075 1076 1077 1078 1079
		*eb = buf;
	} else {
		free_extent_buffer(buf);
	}
	return 0;
}

1080 1081 1082
struct extent_buffer *btrfs_find_create_tree_block(
						struct btrfs_fs_info *fs_info,
						u64 bytenr)
1083
{
1084 1085 1086
	if (btrfs_is_testing(fs_info))
		return alloc_test_extent_buffer(fs_info, bytenr);
	return alloc_extent_buffer(fs_info, bytenr);
1087 1088 1089
}


1090 1091
int btrfs_write_tree_block(struct extent_buffer *buf)
{
1092
	return filemap_fdatawrite_range(buf->pages[0]->mapping, buf->start,
1093
					buf->start + buf->len - 1);
1094 1095
}

1096
void btrfs_wait_tree_block_writeback(struct extent_buffer *buf)
1097
{
1098 1099
	filemap_fdatawait_range(buf->pages[0]->mapping,
			        buf->start, buf->start + buf->len - 1);
1100 1101
}

1102 1103 1104 1105 1106 1107 1108 1109
/*
 * Read tree block at logical address @bytenr and do variant basic but critical
 * verification.
 *
 * @parent_transid:	expected transid of this tree block, skip check if 0
 * @level:		expected level, mandatory check
 * @first_key:		expected key in slot 0, skip check if NULL
 */
1110
struct extent_buffer *read_tree_block(struct btrfs_fs_info *fs_info, u64 bytenr,
1111 1112
				      u64 parent_transid, int level,
				      struct btrfs_key *first_key)
1113 1114 1115 1116
{
	struct extent_buffer *buf = NULL;
	int ret;

1117
	buf = btrfs_find_create_tree_block(fs_info, bytenr);
1118 1119
	if (IS_ERR(buf))
		return buf;
1120

1121 1122
	ret = btree_read_extent_buffer_pages(fs_info, buf, parent_transid,
					     level, first_key);
1123 1124
	if (ret) {
		free_extent_buffer(buf);
1125
		return ERR_PTR(ret);
1126
	}
1127
	return buf;
1128

1129 1130
}

1131
void clean_tree_block(struct btrfs_fs_info *fs_info,
1132
		      struct extent_buffer *buf)
1133
{
1134
	if (btrfs_header_generation(buf) ==
1135
	    fs_info->running_transaction->transid) {
1136
		btrfs_assert_tree_locked(buf);
1137

1138
		if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &buf->bflags)) {
1139 1140 1141
			percpu_counter_add_batch(&fs_info->dirty_metadata_bytes,
						 -buf->len,
						 fs_info->dirty_metadata_batch);
1142 1143 1144 1145
			/* ugh, clear_extent_buffer_dirty needs to lock the page */
			btrfs_set_lock_blocking(buf);
			clear_extent_buffer_dirty(buf);
		}
1146
	}
1147 1148
}

1149 1150 1151 1152 1153 1154 1155 1156 1157
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);

1158
	ret = percpu_counter_init(&writers->counter, 0, GFP_NOFS);
1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174
	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);
}

1175
static void __setup_root(struct btrfs_root *root, struct btrfs_fs_info *fs_info,
1176
			 u64 objectid)
1177
{
1178
	bool dummy = test_bit(BTRFS_FS_STATE_DUMMY_FS_INFO, &fs_info->fs_state);
C
Chris Mason 已提交
1179
	root->node = NULL;
1180
	root->commit_root = NULL;
1181
	root->state = 0;
1182
	root->orphan_cleanup_state = 0;
1183

1184 1185
	root->objectid = objectid;
	root->last_trans = 0;
1186
	root->highest_objectid = 0;
1187
	root->nr_delalloc_inodes = 0;
1188
	root->nr_ordered_extents = 0;
1189
	root->name = NULL;
1190
	root->inode_tree = RB_ROOT;
1191
	INIT_RADIX_TREE(&root->delayed_nodes_tree, GFP_ATOMIC);
1192
	root->block_rsv = NULL;
1193
	root->orphan_block_rsv = NULL;
1194 1195

	INIT_LIST_HEAD(&root->dirty_list);
1196
	INIT_LIST_HEAD(&root->root_list);
1197 1198
	INIT_LIST_HEAD(&root->delalloc_inodes);
	INIT_LIST_HEAD(&root->delalloc_root);
1199 1200
	INIT_LIST_HEAD(&root->ordered_extents);
	INIT_LIST_HEAD(&root->ordered_root);
1201 1202
	INIT_LIST_HEAD(&root->logged_list[0]);
	INIT_LIST_HEAD(&root->logged_list[1]);
1203
	spin_lock_init(&root->orphan_lock);
1204
	spin_lock_init(&root->inode_lock);
1205
	spin_lock_init(&root->delalloc_lock);
1206
	spin_lock_init(&root->ordered_extent_lock);
1207
	spin_lock_init(&root->accounting_lock);
1208 1209
	spin_lock_init(&root->log_extents_lock[0]);
	spin_lock_init(&root->log_extents_lock[1]);
1210
	spin_lock_init(&root->qgroup_meta_rsv_lock);
1211
	mutex_init(&root->objectid_mutex);
1212
	mutex_init(&root->log_mutex);
1213
	mutex_init(&root->ordered_extent_mutex);
1214
	mutex_init(&root->delalloc_mutex);
Y
Yan Zheng 已提交
1215 1216 1217
	init_waitqueue_head(&root->log_writer_wait);
	init_waitqueue_head(&root->log_commit_wait[0]);
	init_waitqueue_head(&root->log_commit_wait[1]);
1218 1219
	INIT_LIST_HEAD(&root->log_ctxs[0]);
	INIT_LIST_HEAD(&root->log_ctxs[1]);
Y
Yan Zheng 已提交
1220 1221 1222
	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
1223
	atomic_set(&root->log_batch, 0);
1224
	atomic_set(&root->orphan_inodes, 0);
1225
	refcount_set(&root->refs, 1);
1226
	atomic_set(&root->will_be_snapshotted, 0);
Y
Yan Zheng 已提交
1227
	root->log_transid = 0;
1228
	root->log_transid_committed = -1;
1229
	root->last_log_commit = 0;
1230
	if (!dummy)
1231
		extent_io_tree_init(&root->dirty_log_pages, NULL);
C
Chris Mason 已提交
1232

1233 1234
	memset(&root->root_key, 0, sizeof(root->root_key));
	memset(&root->root_item, 0, sizeof(root->root_item));
1235
	memset(&root->defrag_progress, 0, sizeof(root->defrag_progress));
1236
	if (!dummy)
1237 1238 1239
		root->defrag_trans_start = fs_info->generation;
	else
		root->defrag_trans_start = 0;
1240
	root->root_key.objectid = objectid;
1241
	root->anon_dev = 0;
1242

1243
	spin_lock_init(&root->root_item_lock);
1244 1245
}

1246 1247
static struct btrfs_root *btrfs_alloc_root(struct btrfs_fs_info *fs_info,
		gfp_t flags)
A
Al Viro 已提交
1248
{
1249
	struct btrfs_root *root = kzalloc(sizeof(*root), flags);
A
Al Viro 已提交
1250 1251 1252 1253 1254
	if (root)
		root->fs_info = fs_info;
	return root;
}

1255 1256
#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
/* Should only be used by the testing infrastructure */
1257
struct btrfs_root *btrfs_alloc_dummy_root(struct btrfs_fs_info *fs_info)
1258 1259 1260
{
	struct btrfs_root *root;

1261 1262 1263 1264
	if (!fs_info)
		return ERR_PTR(-EINVAL);

	root = btrfs_alloc_root(fs_info, GFP_KERNEL);
1265 1266
	if (!root)
		return ERR_PTR(-ENOMEM);
1267

1268
	/* We don't use the stripesize in selftest, set it as sectorsize */
1269
	__setup_root(root, fs_info, BTRFS_ROOT_TREE_OBJECTID);
1270
	root->alloc_bytenr = 0;
1271 1272 1273 1274 1275

	return root;
}
#endif

1276 1277 1278 1279 1280 1281 1282 1283 1284
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;
1285
	uuid_le uuid = NULL_UUID_LE;
1286

1287
	root = btrfs_alloc_root(fs_info, GFP_KERNEL);
1288 1289 1290
	if (!root)
		return ERR_PTR(-ENOMEM);

1291
	__setup_root(root, fs_info, objectid);
1292 1293 1294 1295
	root->root_key.objectid = objectid;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = 0;

1296
	leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0);
1297 1298
	if (IS_ERR(leaf)) {
		ret = PTR_ERR(leaf);
1299
		leaf = NULL;
1300 1301 1302
		goto fail;
	}

1303
	memzero_extent_buffer(leaf, 0, sizeof(struct btrfs_header));
1304 1305 1306 1307 1308 1309
	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;

1310 1311
	write_extent_buffer_fsid(leaf, fs_info->fsid);
	write_extent_buffer_chunk_tree_uuid(leaf, fs_info->chunk_tree_uuid);
1312 1313 1314
	btrfs_mark_buffer_dirty(leaf);

	root->commit_root = btrfs_root_node(root);
1315
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
1316 1317 1318 1319 1320 1321 1322 1323 1324 1325

	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);
1326 1327
	if (is_fstree(objectid))
		uuid_le_gen(&uuid);
1328
	memcpy(root->root_item.uuid, uuid.b, BTRFS_UUID_SIZE);
1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339
	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);

1340 1341
	return root;

1342
fail:
1343 1344
	if (leaf) {
		btrfs_tree_unlock(leaf);
1345
		free_extent_buffer(root->commit_root);
1346 1347 1348
		free_extent_buffer(leaf);
	}
	kfree(root);
1349

1350
	return ERR_PTR(ret);
1351 1352
}

Y
Yan Zheng 已提交
1353 1354
static struct btrfs_root *alloc_log_tree(struct btrfs_trans_handle *trans,
					 struct btrfs_fs_info *fs_info)
1355 1356
{
	struct btrfs_root *root;
Y
Yan Zheng 已提交
1357
	struct extent_buffer *leaf;
1358

1359
	root = btrfs_alloc_root(fs_info, GFP_NOFS);
1360
	if (!root)
Y
Yan Zheng 已提交
1361
		return ERR_PTR(-ENOMEM);
1362

1363
	__setup_root(root, fs_info, BTRFS_TREE_LOG_OBJECTID);
1364 1365 1366 1367

	root->root_key.objectid = BTRFS_TREE_LOG_OBJECTID;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = BTRFS_TREE_LOG_OBJECTID;
1368

Y
Yan Zheng 已提交
1369
	/*
1370 1371
	 * DON'T set REF_COWS for log trees
	 *
Y
Yan Zheng 已提交
1372 1373 1374 1375 1376
	 * 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).
	 */
1377

1378 1379
	leaf = btrfs_alloc_tree_block(trans, root, 0, BTRFS_TREE_LOG_OBJECTID,
			NULL, 0, 0, 0);
Y
Yan Zheng 已提交
1380 1381 1382 1383
	if (IS_ERR(leaf)) {
		kfree(root);
		return ERR_CAST(leaf);
	}
1384

1385
	memzero_extent_buffer(leaf, 0, sizeof(struct btrfs_header));
1386 1387 1388 1389
	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
1390
	root->node = leaf;
1391

1392
	write_extent_buffer_fsid(root->node, fs_info->fsid);
1393 1394
	btrfs_mark_buffer_dirty(root->node);
	btrfs_tree_unlock(root->node);
Y
Yan Zheng 已提交
1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413
	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)
{
1414
	struct btrfs_fs_info *fs_info = root->fs_info;
Y
Yan Zheng 已提交
1415 1416 1417
	struct btrfs_root *log_root;
	struct btrfs_inode_item *inode_item;

1418
	log_root = alloc_log_tree(trans, fs_info);
Y
Yan Zheng 已提交
1419 1420 1421 1422 1423 1424 1425
	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;
1426 1427 1428
	btrfs_set_stack_inode_generation(inode_item, 1);
	btrfs_set_stack_inode_size(inode_item, 3);
	btrfs_set_stack_inode_nlink(inode_item, 1);
1429
	btrfs_set_stack_inode_nbytes(inode_item,
1430
				     fs_info->nodesize);
1431
	btrfs_set_stack_inode_mode(inode_item, S_IFDIR | 0755);
Y
Yan Zheng 已提交
1432

1433
	btrfs_set_root_node(&log_root->root_item, log_root->node);
Y
Yan Zheng 已提交
1434 1435 1436 1437

	WARN_ON(root->log_root);
	root->log_root = log_root;
	root->log_transid = 0;
1438
	root->log_transid_committed = -1;
1439
	root->last_log_commit = 0;
1440 1441 1442
	return 0;
}

1443 1444
static struct btrfs_root *btrfs_read_tree_root(struct btrfs_root *tree_root,
					       struct btrfs_key *key)
1445 1446 1447
{
	struct btrfs_root *root;
	struct btrfs_fs_info *fs_info = tree_root->fs_info;
1448
	struct btrfs_path *path;
1449
	u64 generation;
1450
	int ret;
1451
	int level;
1452

1453 1454
	path = btrfs_alloc_path();
	if (!path)
1455
		return ERR_PTR(-ENOMEM);
1456

1457
	root = btrfs_alloc_root(fs_info, GFP_NOFS);
1458 1459 1460
	if (!root) {
		ret = -ENOMEM;
		goto alloc_fail;
1461 1462
	}

1463
	__setup_root(root, fs_info, key->objectid);
1464

1465 1466
	ret = btrfs_find_root(tree_root, key, path,
			      &root->root_item, &root->root_key);
1467
	if (ret) {
1468 1469
		if (ret > 0)
			ret = -ENOENT;
1470
		goto find_fail;
1471
	}
1472

1473
	generation = btrfs_root_generation(&root->root_item);
1474
	level = btrfs_root_level(&root->root_item);
1475 1476
	root->node = read_tree_block(fs_info,
				     btrfs_root_bytenr(&root->root_item),
1477
				     generation, level, NULL);
1478 1479
	if (IS_ERR(root->node)) {
		ret = PTR_ERR(root->node);
1480 1481 1482
		goto find_fail;
	} else if (!btrfs_buffer_uptodate(root->node, generation, 0)) {
		ret = -EIO;
1483 1484
		free_extent_buffer(root->node);
		goto find_fail;
1485
	}
1486
	root->commit_root = btrfs_root_node(root);
1487
out:
1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507
	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) {
1508
		set_bit(BTRFS_ROOT_REF_COWS, &root->state);
1509 1510
		btrfs_check_and_init_root_item(&root->root_item);
	}
1511

1512 1513 1514
	return root;
}

1515 1516 1517
int btrfs_init_fs_root(struct btrfs_root *root)
{
	int ret;
1518
	struct btrfs_subvolume_writers *writers;
1519 1520 1521 1522 1523 1524 1525 1526 1527

	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;
	}

1528 1529 1530 1531 1532 1533 1534
	writers = btrfs_alloc_subvolume_writers();
	if (IS_ERR(writers)) {
		ret = PTR_ERR(writers);
		goto fail;
	}
	root->subv_writers = writers;

1535
	btrfs_init_free_ino_ctl(root);
1536 1537
	spin_lock_init(&root->ino_cache_lock);
	init_waitqueue_head(&root->ino_cache_wait);
1538 1539 1540

	ret = get_anon_bdev(&root->anon_dev);
	if (ret)
L
Liu Bo 已提交
1541
		goto fail;
1542 1543 1544 1545 1546 1547

	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
1548
		goto fail;
1549 1550 1551 1552 1553 1554
	}

	ASSERT(root->highest_objectid <= BTRFS_LAST_FREE_OBJECTID);

	mutex_unlock(&root->objectid_mutex);

1555 1556
	return 0;
fail:
L
Liu Bo 已提交
1557
	/* the caller is responsible to call free_fs_root */
1558 1559 1560
	return ret;
}

1561 1562
struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info,
					u64 root_id)
1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577
{
	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;

1578
	ret = radix_tree_preload(GFP_NOFS);
1579 1580 1581 1582 1583 1584 1585 1586
	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)
1587
		set_bit(BTRFS_ROOT_IN_RADIX, &root->state);
1588 1589 1590 1591 1592 1593
	spin_unlock(&fs_info->fs_roots_radix_lock);
	radix_tree_preload_end();

	return ret;
}

1594 1595 1596
struct btrfs_root *btrfs_get_fs_root(struct btrfs_fs_info *fs_info,
				     struct btrfs_key *location,
				     bool check_ref)
1597 1598
{
	struct btrfs_root *root;
1599
	struct btrfs_path *path;
1600
	struct btrfs_key key;
1601 1602
	int ret;

1603 1604 1605 1606
	if (location->objectid == BTRFS_ROOT_TREE_OBJECTID)
		return fs_info->tree_root;
	if (location->objectid == BTRFS_EXTENT_TREE_OBJECTID)
		return fs_info->extent_root;
1607 1608 1609 1610
	if (location->objectid == BTRFS_CHUNK_TREE_OBJECTID)
		return fs_info->chunk_root;
	if (location->objectid == BTRFS_DEV_TREE_OBJECTID)
		return fs_info->dev_root;
1611 1612
	if (location->objectid == BTRFS_CSUM_TREE_OBJECTID)
		return fs_info->csum_root;
1613 1614 1615
	if (location->objectid == BTRFS_QUOTA_TREE_OBJECTID)
		return fs_info->quota_root ? fs_info->quota_root :
					     ERR_PTR(-ENOENT);
1616 1617 1618
	if (location->objectid == BTRFS_UUID_TREE_OBJECTID)
		return fs_info->uuid_root ? fs_info->uuid_root :
					    ERR_PTR(-ENOENT);
1619 1620 1621
	if (location->objectid == BTRFS_FREE_SPACE_TREE_OBJECTID)
		return fs_info->free_space_root ? fs_info->free_space_root :
						  ERR_PTR(-ENOENT);
1622
again:
1623
	root = btrfs_lookup_fs_root(fs_info, location->objectid);
1624
	if (root) {
1625
		if (check_ref && btrfs_root_refs(&root->root_item) == 0)
1626
			return ERR_PTR(-ENOENT);
1627
		return root;
1628
	}
1629

1630
	root = btrfs_read_fs_root(fs_info->tree_root, location);
1631 1632
	if (IS_ERR(root))
		return root;
1633

1634
	if (check_ref && btrfs_root_refs(&root->root_item) == 0) {
1635
		ret = -ENOENT;
1636
		goto fail;
1637
	}
1638

1639
	ret = btrfs_init_fs_root(root);
1640 1641
	if (ret)
		goto fail;
1642

1643 1644 1645 1646 1647
	path = btrfs_alloc_path();
	if (!path) {
		ret = -ENOMEM;
		goto fail;
	}
1648 1649 1650 1651 1652
	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);
1653
	btrfs_free_path(path);
1654 1655 1656
	if (ret < 0)
		goto fail;
	if (ret == 0)
1657
		set_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &root->state);
1658

1659
	ret = btrfs_insert_fs_root(fs_info, root);
1660
	if (ret) {
1661 1662 1663 1664 1665
		if (ret == -EEXIST) {
			free_fs_root(root);
			goto again;
		}
		goto fail;
1666
	}
1667
	return root;
1668 1669 1670
fail:
	free_fs_root(root);
	return ERR_PTR(ret);
1671 1672
}

C
Chris Mason 已提交
1673 1674 1675 1676 1677 1678
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
1679

1680 1681
	rcu_read_lock();
	list_for_each_entry_rcu(device, &info->fs_devices->devices, dev_list) {
1682 1683
		if (!device->bdev)
			continue;
1684
		bdi = device->bdev->bd_bdi;
1685
		if (bdi_congested(bdi, bdi_bits)) {
C
Chris Mason 已提交
1686 1687 1688 1689
			ret = 1;
			break;
		}
	}
1690
	rcu_read_unlock();
C
Chris Mason 已提交
1691 1692 1693
	return ret;
}

1694 1695 1696 1697 1698
/*
 * 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)
1699 1700
{
	struct bio *bio;
1701
	struct btrfs_end_io_wq *end_io_wq;
1702

1703
	end_io_wq = container_of(work, struct btrfs_end_io_wq, work);
1704
	bio = end_io_wq->bio;
1705

1706
	bio->bi_status = end_io_wq->status;
1707 1708
	bio->bi_private = end_io_wq->private;
	bio->bi_end_io = end_io_wq->end_io;
1709
	kmem_cache_free(btrfs_end_io_wq_cache, end_io_wq);
1710
	bio_endio(bio);
1711 1712
}

1713 1714 1715
static int cleaner_kthread(void *arg)
{
	struct btrfs_root *root = arg;
1716
	struct btrfs_fs_info *fs_info = root->fs_info;
1717
	int again;
1718
	struct btrfs_trans_handle *trans;
1719 1720

	do {
1721
		again = 0;
1722

1723
		/* Make the cleaner go to sleep early. */
1724
		if (btrfs_need_cleaner_sleep(fs_info))
1725 1726
			goto sleep;

1727 1728 1729 1730
		/*
		 * Do not do anything if we might cause open_ctree() to block
		 * before we have finished mounting the filesystem.
		 */
1731
		if (!test_bit(BTRFS_FS_OPEN, &fs_info->flags))
1732 1733
			goto sleep;

1734
		if (!mutex_trylock(&fs_info->cleaner_mutex))
1735 1736
			goto sleep;

1737 1738 1739 1740
		/*
		 * Avoid the problem that we change the status of the fs
		 * during the above check and trylock.
		 */
1741
		if (btrfs_need_cleaner_sleep(fs_info)) {
1742
			mutex_unlock(&fs_info->cleaner_mutex);
1743
			goto sleep;
1744
		}
1745

1746
		mutex_lock(&fs_info->cleaner_delayed_iput_mutex);
1747
		btrfs_run_delayed_iputs(fs_info);
1748
		mutex_unlock(&fs_info->cleaner_delayed_iput_mutex);
1749

1750
		again = btrfs_clean_one_deleted_snapshot(root);
1751
		mutex_unlock(&fs_info->cleaner_mutex);
1752 1753

		/*
1754 1755
		 * The defragger has dealt with the R/O remount and umount,
		 * needn't do anything special here.
1756
		 */
1757
		btrfs_run_defrag_inodes(fs_info);
1758 1759 1760 1761 1762 1763 1764 1765 1766

		/*
		 * 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.
		 */
1767
		btrfs_delete_unused_bgs(fs_info);
1768
sleep:
1769
		if (!again) {
1770
			set_current_state(TASK_INTERRUPTIBLE);
1771 1772
			if (!kthread_should_stop())
				schedule();
1773 1774 1775
			__set_current_state(TASK_RUNNING);
		}
	} while (!kthread_should_stop());
1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790

	/*
	 * 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)
1791
			btrfs_err(fs_info,
1792 1793 1794 1795 1796
				  "cleaner transaction attach returned %ld",
				  PTR_ERR(trans));
	} else {
		int ret;

1797
		ret = btrfs_commit_transaction(trans);
1798
		if (ret)
1799
			btrfs_err(fs_info,
1800 1801 1802 1803
				  "cleaner open transaction commit returned %d",
				  ret);
	}

1804 1805 1806 1807 1808 1809
	return 0;
}

static int transaction_kthread(void *arg)
{
	struct btrfs_root *root = arg;
1810
	struct btrfs_fs_info *fs_info = root->fs_info;
1811 1812
	struct btrfs_trans_handle *trans;
	struct btrfs_transaction *cur;
1813
	u64 transid;
1814 1815
	unsigned long now;
	unsigned long delay;
1816
	bool cannot_commit;
1817 1818

	do {
1819
		cannot_commit = false;
1820 1821
		delay = HZ * fs_info->commit_interval;
		mutex_lock(&fs_info->transaction_kthread_mutex);
1822

1823 1824
		spin_lock(&fs_info->trans_lock);
		cur = fs_info->running_transaction;
1825
		if (!cur) {
1826
			spin_unlock(&fs_info->trans_lock);
1827 1828
			goto sleep;
		}
Y
Yan Zheng 已提交
1829

1830
		now = get_seconds();
1831
		if (cur->state < TRANS_STATE_BLOCKED &&
1832
		    (now < cur->start_time ||
1833 1834
		     now - cur->start_time < fs_info->commit_interval)) {
			spin_unlock(&fs_info->trans_lock);
1835 1836 1837
			delay = HZ * 5;
			goto sleep;
		}
1838
		transid = cur->transid;
1839
		spin_unlock(&fs_info->trans_lock);
1840

1841
		/* If the file system is aborted, this will always fail. */
1842
		trans = btrfs_attach_transaction(root);
1843
		if (IS_ERR(trans)) {
1844 1845
			if (PTR_ERR(trans) != -ENOENT)
				cannot_commit = true;
1846
			goto sleep;
1847
		}
1848
		if (transid == trans->transid) {
1849
			btrfs_commit_transaction(trans);
1850
		} else {
1851
			btrfs_end_transaction(trans);
1852
		}
1853
sleep:
1854 1855
		wake_up_process(fs_info->cleaner_kthread);
		mutex_unlock(&fs_info->transaction_kthread_mutex);
1856

J
Josef Bacik 已提交
1857
		if (unlikely(test_bit(BTRFS_FS_STATE_ERROR,
1858
				      &fs_info->fs_state)))
1859
			btrfs_cleanup_transaction(fs_info);
1860
		if (!kthread_should_stop() &&
1861
				(!btrfs_transaction_blocked(fs_info) ||
1862
				 cannot_commit))
1863
			schedule_timeout_interruptible(delay);
1864 1865 1866 1867
	} while (!kthread_should_stop());
	return 0;
}

C
Chris Mason 已提交
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 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 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
/*
 * 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));

1974 1975 1976 1977 1978 1979 1980 1981
	/*
	 * 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
1982
			       btrfs_header_generation(info->fs_root->node));
1983
		btrfs_set_backup_fs_root_level(root_backup,
C
Chris Mason 已提交
1984
			       btrfs_header_level(info->fs_root->node));
1985
	}
C
Chris Mason 已提交
1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066

	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
2067 2068 2069
/* helper to cleanup workers */
static void btrfs_stop_all_workers(struct btrfs_fs_info *fs_info)
{
2070
	btrfs_destroy_workqueue(fs_info->fixup_workers);
2071
	btrfs_destroy_workqueue(fs_info->delalloc_workers);
2072
	btrfs_destroy_workqueue(fs_info->workers);
2073 2074
	btrfs_destroy_workqueue(fs_info->endio_workers);
	btrfs_destroy_workqueue(fs_info->endio_raid56_workers);
2075
	btrfs_destroy_workqueue(fs_info->endio_repair_workers);
2076
	btrfs_destroy_workqueue(fs_info->rmw_workers);
2077 2078
	btrfs_destroy_workqueue(fs_info->endio_write_workers);
	btrfs_destroy_workqueue(fs_info->endio_freespace_worker);
2079
	btrfs_destroy_workqueue(fs_info->submit_workers);
2080
	btrfs_destroy_workqueue(fs_info->delayed_workers);
2081
	btrfs_destroy_workqueue(fs_info->caching_workers);
2082
	btrfs_destroy_workqueue(fs_info->readahead_workers);
2083
	btrfs_destroy_workqueue(fs_info->flush_workers);
2084
	btrfs_destroy_workqueue(fs_info->qgroup_rescan_workers);
C
Chris Mason 已提交
2085
	btrfs_destroy_workqueue(fs_info->extent_workers);
2086 2087 2088 2089 2090 2091 2092
	/*
	 * 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
2093 2094
}

2095 2096 2097 2098 2099 2100 2101 2102 2103 2104
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
2105 2106 2107
/* helper to cleanup tree roots */
static void free_root_pointers(struct btrfs_fs_info *info, int chunk_root)
{
2108
	free_root_extent_buffers(info->tree_root);
2109

2110 2111 2112 2113 2114 2115 2116
	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);
2117
	free_root_extent_buffers(info->free_space_root);
C
Chris Mason 已提交
2118 2119
}

2120
void btrfs_free_fs_roots(struct btrfs_fs_info *fs_info)
2121 2122 2123 2124 2125 2126 2127 2128 2129 2130
{
	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);

2131
		if (test_bit(BTRFS_ROOT_IN_RADIX, &gang[0]->state)) {
2132
			btrfs_drop_and_free_fs_root(fs_info, gang[0]);
2133 2134 2135
		} else {
			free_extent_buffer(gang[0]->node);
			free_extent_buffer(gang[0]->commit_root);
2136
			btrfs_put_fs_root(gang[0]);
2137 2138 2139 2140 2141 2142 2143 2144 2145 2146
		}
	}

	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++)
2147
			btrfs_drop_and_free_fs_root(fs_info, gang[i]);
2148
	}
2149 2150 2151

	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
		btrfs_free_log_root_tree(NULL, fs_info);
2152
		btrfs_destroy_pinned_extent(fs_info, fs_info->pinned_extents);
2153
	}
2154
}
C
Chris Mason 已提交
2155

2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166
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;
}

2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177
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);
}

2178
static void btrfs_init_btree_inode(struct btrfs_fs_info *fs_info)
2179
{
2180 2181 2182 2183
	struct inode *inode = fs_info->btree_inode;

	inode->i_ino = BTRFS_BTREE_INODE_OBJECTID;
	set_nlink(inode, 1);
2184 2185 2186 2187 2188
	/*
	 * 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
	 */
2189 2190
	inode->i_size = OFFSET_MAX;
	inode->i_mapping->a_ops = &btree_aops;
2191

2192
	RB_CLEAR_NODE(&BTRFS_I(inode)->rb_node);
2193
	extent_io_tree_init(&BTRFS_I(inode)->io_tree, inode);
2194 2195
	BTRFS_I(inode)->io_tree.track_uptodate = 0;
	extent_map_tree_init(&BTRFS_I(inode)->extent_tree);
2196

2197
	BTRFS_I(inode)->io_tree.ops = &btree_extent_io_ops;
2198

2199 2200 2201 2202
	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);
2203 2204
}

2205 2206 2207 2208 2209
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);
2210 2211 2212
	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);
2213
	init_waitqueue_head(&fs_info->replace_wait);
2214
	init_waitqueue_head(&fs_info->dev_replace.read_lock_wq);
2215 2216
}

2217 2218 2219 2220 2221 2222 2223 2224 2225
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;
2226
	fs_info->qgroup_rescan_running = false;
2227 2228 2229
	mutex_init(&fs_info->qgroup_rescan_lock);
}

2230 2231 2232
static int btrfs_init_workqueues(struct btrfs_fs_info *fs_info,
		struct btrfs_fs_devices *fs_devices)
{
2233
	u32 max_active = fs_info->thread_pool_size;
2234
	unsigned int flags = WQ_MEM_RECLAIM | WQ_FREEZABLE | WQ_UNBOUND;
2235 2236

	fs_info->workers =
2237 2238
		btrfs_alloc_workqueue(fs_info, "worker",
				      flags | WQ_HIGHPRI, max_active, 16);
2239 2240

	fs_info->delalloc_workers =
2241 2242
		btrfs_alloc_workqueue(fs_info, "delalloc",
				      flags, max_active, 2);
2243 2244

	fs_info->flush_workers =
2245 2246
		btrfs_alloc_workqueue(fs_info, "flush_delalloc",
				      flags, max_active, 0);
2247 2248

	fs_info->caching_workers =
2249
		btrfs_alloc_workqueue(fs_info, "cache", flags, max_active, 0);
2250 2251 2252 2253 2254 2255 2256

	/*
	 * 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 =
2257
		btrfs_alloc_workqueue(fs_info, "submit", flags,
2258 2259 2260 2261
				      min_t(u64, fs_devices->num_devices,
					    max_active), 64);

	fs_info->fixup_workers =
2262
		btrfs_alloc_workqueue(fs_info, "fixup", flags, 1, 0);
2263 2264 2265 2266 2267 2268

	/*
	 * endios are largely parallel and should have a very
	 * low idle thresh
	 */
	fs_info->endio_workers =
2269
		btrfs_alloc_workqueue(fs_info, "endio", flags, max_active, 4);
2270
	fs_info->endio_meta_workers =
2271 2272
		btrfs_alloc_workqueue(fs_info, "endio-meta", flags,
				      max_active, 4);
2273
	fs_info->endio_meta_write_workers =
2274 2275
		btrfs_alloc_workqueue(fs_info, "endio-meta-write", flags,
				      max_active, 2);
2276
	fs_info->endio_raid56_workers =
2277 2278
		btrfs_alloc_workqueue(fs_info, "endio-raid56", flags,
				      max_active, 4);
2279
	fs_info->endio_repair_workers =
2280
		btrfs_alloc_workqueue(fs_info, "endio-repair", flags, 1, 0);
2281
	fs_info->rmw_workers =
2282
		btrfs_alloc_workqueue(fs_info, "rmw", flags, max_active, 2);
2283
	fs_info->endio_write_workers =
2284 2285
		btrfs_alloc_workqueue(fs_info, "endio-write", flags,
				      max_active, 2);
2286
	fs_info->endio_freespace_worker =
2287 2288
		btrfs_alloc_workqueue(fs_info, "freespace-write", flags,
				      max_active, 0);
2289
	fs_info->delayed_workers =
2290 2291
		btrfs_alloc_workqueue(fs_info, "delayed-meta", flags,
				      max_active, 0);
2292
	fs_info->readahead_workers =
2293 2294
		btrfs_alloc_workqueue(fs_info, "readahead", flags,
				      max_active, 2);
2295
	fs_info->qgroup_rescan_workers =
2296
		btrfs_alloc_workqueue(fs_info, "qgroup-rescan", flags, 1, 0);
2297
	fs_info->extent_workers =
2298
		btrfs_alloc_workqueue(fs_info, "extent-refs", flags,
2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318
				      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;
}

2319 2320 2321 2322 2323 2324 2325
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);
2326
	int level = btrfs_super_log_root_level(disk_super);
2327 2328

	if (fs_devices->rw_devices == 0) {
2329
		btrfs_warn(fs_info, "log replay required on RO media");
2330 2331 2332
		return -EIO;
	}

2333
	log_tree_root = btrfs_alloc_root(fs_info, GFP_KERNEL);
2334 2335 2336
	if (!log_tree_root)
		return -ENOMEM;

2337
	__setup_root(log_tree_root, fs_info, BTRFS_TREE_LOG_OBJECTID);
2338

2339
	log_tree_root->node = read_tree_block(fs_info, bytenr,
2340 2341
					      fs_info->generation + 1,
					      level, NULL);
2342
	if (IS_ERR(log_tree_root->node)) {
2343
		btrfs_warn(fs_info, "failed to read log tree");
2344
		ret = PTR_ERR(log_tree_root->node);
2345
		kfree(log_tree_root);
2346
		return ret;
2347
	} else if (!extent_buffer_uptodate(log_tree_root->node)) {
2348
		btrfs_err(fs_info, "failed to read log tree");
2349 2350 2351 2352 2353 2354 2355
		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) {
2356 2357
		btrfs_handle_fs_error(fs_info, ret,
				      "Failed to recover log tree");
2358 2359 2360 2361 2362
		free_extent_buffer(log_tree_root->node);
		kfree(log_tree_root);
		return ret;
	}

2363
	if (sb_rdonly(fs_info->sb)) {
2364
		ret = btrfs_commit_super(fs_info);
2365 2366 2367 2368 2369 2370 2371
		if (ret)
			return ret;
	}

	return 0;
}

2372
static int btrfs_read_roots(struct btrfs_fs_info *fs_info)
2373
{
2374
	struct btrfs_root *tree_root = fs_info->tree_root;
2375
	struct btrfs_root *root;
2376 2377 2378
	struct btrfs_key location;
	int ret;

2379 2380
	BUG_ON(!fs_info->tree_root);

2381 2382 2383 2384
	location.objectid = BTRFS_EXTENT_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
	location.offset = 0;

2385
	root = btrfs_read_tree_root(tree_root, &location);
2386 2387 2388 2389
	if (IS_ERR(root)) {
		ret = PTR_ERR(root);
		goto out;
	}
2390 2391
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
	fs_info->extent_root = root;
2392 2393

	location.objectid = BTRFS_DEV_TREE_OBJECTID;
2394
	root = btrfs_read_tree_root(tree_root, &location);
2395 2396 2397 2398
	if (IS_ERR(root)) {
		ret = PTR_ERR(root);
		goto out;
	}
2399 2400
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
	fs_info->dev_root = root;
2401 2402 2403
	btrfs_init_devices_late(fs_info);

	location.objectid = BTRFS_CSUM_TREE_OBJECTID;
2404
	root = btrfs_read_tree_root(tree_root, &location);
2405 2406 2407 2408
	if (IS_ERR(root)) {
		ret = PTR_ERR(root);
		goto out;
	}
2409 2410
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
	fs_info->csum_root = root;
2411 2412

	location.objectid = BTRFS_QUOTA_TREE_OBJECTID;
2413 2414 2415
	root = btrfs_read_tree_root(tree_root, &location);
	if (!IS_ERR(root)) {
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
2416
		set_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags);
2417
		fs_info->quota_root = root;
2418 2419 2420
	}

	location.objectid = BTRFS_UUID_TREE_OBJECTID;
2421 2422 2423
	root = btrfs_read_tree_root(tree_root, &location);
	if (IS_ERR(root)) {
		ret = PTR_ERR(root);
2424
		if (ret != -ENOENT)
2425
			goto out;
2426
	} else {
2427 2428
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
		fs_info->uuid_root = root;
2429 2430
	}

2431 2432 2433
	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);
2434 2435 2436 2437
		if (IS_ERR(root)) {
			ret = PTR_ERR(root);
			goto out;
		}
2438 2439 2440 2441
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
		fs_info->free_space_root = root;
	}

2442
	return 0;
2443 2444 2445 2446
out:
	btrfs_warn(fs_info, "failed to read root (objectid=%llu): %d",
		   location.objectid, ret);
	return ret;
2447 2448
}

A
Al Viro 已提交
2449 2450 2451
int open_ctree(struct super_block *sb,
	       struct btrfs_fs_devices *fs_devices,
	       char *options)
2452
{
2453 2454
	u32 sectorsize;
	u32 nodesize;
2455
	u32 stripesize;
2456
	u64 generation;
2457
	u64 features;
2458
	struct btrfs_key location;
2459
	struct buffer_head *bh;
2460
	struct btrfs_super_block *disk_super;
2461
	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
2462
	struct btrfs_root *tree_root;
2463
	struct btrfs_root *chunk_root;
2464
	int ret;
2465
	int err = -EINVAL;
C
Chris Mason 已提交
2466 2467
	int num_backups_tried = 0;
	int backup_index = 0;
2468
	int clear_free_space_tree = 0;
2469
	int level;
2470

2471 2472
	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);
2473
	if (!tree_root || !chunk_root) {
C
Chris Mason 已提交
2474 2475 2476
		err = -ENOMEM;
		goto fail;
	}
2477 2478 2479 2480 2481 2482 2483

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

2484
	ret = percpu_counter_init(&fs_info->dirty_metadata_bytes, 0, GFP_KERNEL);
2485 2486
	if (ret) {
		err = ret;
2487
		goto fail_srcu;
2488
	}
2489
	fs_info->dirty_metadata_batch = PAGE_SIZE *
2490 2491
					(1 + ilog2(nr_cpu_ids));

2492
	ret = percpu_counter_init(&fs_info->delalloc_bytes, 0, GFP_KERNEL);
2493 2494 2495 2496 2497
	if (ret) {
		err = ret;
		goto fail_dirty_metadata_bytes;
	}

2498
	ret = percpu_counter_init(&fs_info->bio_counter, 0, GFP_KERNEL);
2499 2500 2501 2502 2503
	if (ret) {
		err = ret;
		goto fail_delalloc_bytes;
	}

2504
	INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC);
2505
	INIT_RADIX_TREE(&fs_info->buffer_radix, GFP_ATOMIC);
C
Chris Mason 已提交
2506
	INIT_LIST_HEAD(&fs_info->trans_list);
2507
	INIT_LIST_HEAD(&fs_info->dead_roots);
Y
Yan, Zheng 已提交
2508
	INIT_LIST_HEAD(&fs_info->delayed_iputs);
2509
	INIT_LIST_HEAD(&fs_info->delalloc_roots);
2510
	INIT_LIST_HEAD(&fs_info->caching_block_groups);
2511 2512
	INIT_LIST_HEAD(&fs_info->pending_raid_kobjs);
	spin_lock_init(&fs_info->pending_raid_kobjs_lock);
2513
	spin_lock_init(&fs_info->delalloc_root_lock);
J
Josef Bacik 已提交
2514
	spin_lock_init(&fs_info->trans_lock);
2515
	spin_lock_init(&fs_info->fs_roots_radix_lock);
Y
Yan, Zheng 已提交
2516
	spin_lock_init(&fs_info->delayed_iput_lock);
C
Chris Mason 已提交
2517
	spin_lock_init(&fs_info->defrag_inodes_lock);
J
Jan Schmidt 已提交
2518
	spin_lock_init(&fs_info->tree_mod_seq_lock);
2519
	spin_lock_init(&fs_info->super_lock);
J
Josef Bacik 已提交
2520
	spin_lock_init(&fs_info->qgroup_op_lock);
2521
	spin_lock_init(&fs_info->buffer_lock);
2522
	spin_lock_init(&fs_info->unused_bgs_lock);
J
Jan Schmidt 已提交
2523
	rwlock_init(&fs_info->tree_mod_log_lock);
2524
	mutex_init(&fs_info->unused_bg_unpin_mutex);
2525
	mutex_init(&fs_info->delete_unused_bgs_mutex);
C
Chris Mason 已提交
2526
	mutex_init(&fs_info->reloc_mutex);
2527
	mutex_init(&fs_info->delalloc_root_mutex);
2528
	mutex_init(&fs_info->cleaner_delayed_iput_mutex);
2529
	seqlock_init(&fs_info->profiles_lock);
2530

2531
	INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots);
2532
	INIT_LIST_HEAD(&fs_info->space_info);
J
Jan Schmidt 已提交
2533
	INIT_LIST_HEAD(&fs_info->tree_mod_seq_list);
2534
	INIT_LIST_HEAD(&fs_info->unused_bgs);
2535
	btrfs_mapping_init(&fs_info->mapping_tree);
2536 2537 2538 2539 2540 2541 2542
	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);
2543
	atomic_set(&fs_info->async_delalloc_pages, 0);
C
Chris Mason 已提交
2544
	atomic_set(&fs_info->defrag_running, 0);
J
Josef Bacik 已提交
2545
	atomic_set(&fs_info->qgroup_op_seq, 0);
Z
Zhao Lei 已提交
2546
	atomic_set(&fs_info->reada_works_cnt, 0);
2547
	atomic64_set(&fs_info->tree_mod_seq, 0);
C
Chris Mason 已提交
2548
	fs_info->sb = sb;
2549
	fs_info->max_inline = BTRFS_DEFAULT_MAX_INLINE;
J
Josef Bacik 已提交
2550
	fs_info->metadata_ratio = 0;
C
Chris Mason 已提交
2551
	fs_info->defrag_inodes = RB_ROOT;
2552
	atomic64_set(&fs_info->free_chunk_space, 0);
J
Jan Schmidt 已提交
2553
	fs_info->tree_mod_log = RB_ROOT;
2554
	fs_info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
2555
	fs_info->avg_delayed_ref_runtime = NSEC_PER_SEC >> 6; /* div by 64 */
2556
	/* readahead state */
2557
	INIT_RADIX_TREE(&fs_info->reada_tree, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
2558
	spin_lock_init(&fs_info->reada_lock);
J
Josef Bacik 已提交
2559
	btrfs_init_ref_verify(fs_info);
C
Chris Mason 已提交
2560

2561 2562
	fs_info->thread_pool_size = min_t(unsigned long,
					  num_online_cpus() + 2, 8);
2563

2564 2565
	INIT_LIST_HEAD(&fs_info->ordered_roots);
	spin_lock_init(&fs_info->ordered_root_lock);
2566 2567 2568 2569 2570 2571 2572 2573

	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);

2574
	fs_info->delayed_root = kmalloc(sizeof(struct btrfs_delayed_root),
2575
					GFP_KERNEL);
2576 2577 2578 2579 2580
	if (!fs_info->delayed_root) {
		err = -ENOMEM;
		goto fail_iput;
	}
	btrfs_init_delayed_root(fs_info->delayed_root);
2581

2582
	btrfs_init_scrub(fs_info);
2583 2584 2585
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	fs_info->check_integrity_print_mask = 0;
#endif
2586
	btrfs_init_balance(fs_info);
2587
	btrfs_init_async_reclaim_work(&fs_info->async_reclaim_work);
A
Arne Jansen 已提交
2588

2589 2590
	sb->s_blocksize = BTRFS_BDEV_BLOCKSIZE;
	sb->s_blocksize_bits = blksize_bits(BTRFS_BDEV_BLOCKSIZE);
2591

2592
	btrfs_init_btree_inode(fs_info);
2593

J
Josef Bacik 已提交
2594
	spin_lock_init(&fs_info->block_group_cache_lock);
2595
	fs_info->block_group_cache_tree = RB_ROOT;
2596
	fs_info->first_logical_byte = (u64)-1;
J
Josef Bacik 已提交
2597

2598 2599
	extent_io_tree_init(&fs_info->freed_extents[0], NULL);
	extent_io_tree_init(&fs_info->freed_extents[1], NULL);
2600
	fs_info->pinned_extents = &fs_info->freed_extents[0];
2601
	set_bit(BTRFS_FS_BARRIER, &fs_info->flags);
C
Chris Mason 已提交
2602

2603
	mutex_init(&fs_info->ordered_operations_mutex);
2604
	mutex_init(&fs_info->tree_log_mutex);
2605
	mutex_init(&fs_info->chunk_mutex);
2606 2607
	mutex_init(&fs_info->transaction_kthread_mutex);
	mutex_init(&fs_info->cleaner_mutex);
2608
	mutex_init(&fs_info->volume_mutex);
2609
	mutex_init(&fs_info->ro_block_group_mutex);
2610
	init_rwsem(&fs_info->commit_root_sem);
2611
	init_rwsem(&fs_info->cleanup_work_sem);
2612
	init_rwsem(&fs_info->subvol_sem);
S
Stefan Behrens 已提交
2613
	sema_init(&fs_info->uuid_tree_rescan_sem, 1);
2614

2615
	btrfs_init_dev_replace_locks(fs_info);
2616
	btrfs_init_qgroup(fs_info);
2617

2618 2619 2620
	btrfs_init_free_cluster(&fs_info->meta_alloc_cluster);
	btrfs_init_free_cluster(&fs_info->data_alloc_cluster);

2621
	init_waitqueue_head(&fs_info->transaction_throttle);
2622
	init_waitqueue_head(&fs_info->transaction_wait);
S
Sage Weil 已提交
2623
	init_waitqueue_head(&fs_info->transaction_blocked_wait);
2624
	init_waitqueue_head(&fs_info->async_submit_wait);
2625

2626 2627
	INIT_LIST_HEAD(&fs_info->pinned_chunks);

2628 2629 2630 2631 2632
	/* 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
2633 2634
	ret = btrfs_alloc_stripe_hash_table(fs_info);
	if (ret) {
2635
		err = ret;
D
David Woodhouse 已提交
2636 2637 2638
		goto fail_alloc;
	}

2639
	__setup_root(tree_root, fs_info, BTRFS_ROOT_TREE_OBJECTID);
2640

2641
	invalidate_bdev(fs_devices->latest_bdev);
D
David Sterba 已提交
2642 2643 2644 2645

	/*
	 * Read super block and check the signature bytes only
	 */
Y
Yan Zheng 已提交
2646
	bh = btrfs_read_dev_super(fs_devices->latest_bdev);
2647 2648
	if (IS_ERR(bh)) {
		err = PTR_ERR(bh);
2649
		goto fail_alloc;
2650
	}
C
Chris Mason 已提交
2651

D
David Sterba 已提交
2652 2653 2654 2655
	/*
	 * We want to check superblock checksum, the type is stored inside.
	 * Pass the whole disk block of size BTRFS_SUPER_INFO_SIZE (4k).
	 */
2656
	if (btrfs_check_super_csum(fs_info, bh->b_data)) {
2657
		btrfs_err(fs_info, "superblock checksum mismatch");
D
David Sterba 已提交
2658
		err = -EINVAL;
2659
		brelse(bh);
D
David Sterba 已提交
2660 2661 2662 2663 2664 2665 2666 2667
		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
	 */
2668 2669 2670
	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));
2671
	brelse(bh);
2672

2673
	memcpy(fs_info->fsid, fs_info->super_copy->fsid, BTRFS_FSID_SIZE);
2674

2675
	ret = btrfs_check_super_valid(fs_info);
D
David Sterba 已提交
2676
	if (ret) {
2677
		btrfs_err(fs_info, "superblock contains fatal errors");
D
David Sterba 已提交
2678 2679 2680 2681
		err = -EINVAL;
		goto fail_alloc;
	}

2682
	disk_super = fs_info->super_copy;
2683
	if (!btrfs_super_root(disk_super))
2684
		goto fail_alloc;
2685

L
liubo 已提交
2686
	/* check FS state, whether FS is broken. */
2687 2688
	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
2689

C
Chris Mason 已提交
2690 2691 2692 2693 2694 2695 2696
	/*
	 * 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);

2697 2698 2699 2700 2701 2702
	/*
	 * 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;

2703
	ret = btrfs_parse_options(fs_info, options, sb->s_flags);
Y
Yan Zheng 已提交
2704 2705
	if (ret) {
		err = ret;
2706
		goto fail_alloc;
Y
Yan Zheng 已提交
2707
	}
2708

2709 2710 2711
	features = btrfs_super_incompat_flags(disk_super) &
		~BTRFS_FEATURE_INCOMPAT_SUPP;
	if (features) {
2712 2713 2714
		btrfs_err(fs_info,
		    "cannot mount because of unsupported optional features (%llx)",
		    features);
2715
		err = -EINVAL;
2716
		goto fail_alloc;
2717 2718
	}

2719
	features = btrfs_super_incompat_flags(disk_super);
L
Li Zefan 已提交
2720
	features |= BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF;
2721
	if (fs_info->compress_type == BTRFS_COMPRESS_LZO)
L
Li Zefan 已提交
2722
		features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
N
Nick Terrell 已提交
2723 2724
	else if (fs_info->compress_type == BTRFS_COMPRESS_ZSTD)
		features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_ZSTD;
2725

2726
	if (features & BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA)
2727
		btrfs_info(fs_info, "has skinny extents");
2728

2729 2730 2731 2732
	/*
	 * flag our filesystem as having big metadata blocks if
	 * they are bigger than the page size
	 */
2733
	if (btrfs_super_nodesize(disk_super) > PAGE_SIZE) {
2734
		if (!(features & BTRFS_FEATURE_INCOMPAT_BIG_METADATA))
2735 2736
			btrfs_info(fs_info,
				"flagging fs with big metadata feature");
2737 2738 2739
		features |= BTRFS_FEATURE_INCOMPAT_BIG_METADATA;
	}

2740 2741
	nodesize = btrfs_super_nodesize(disk_super);
	sectorsize = btrfs_super_sectorsize(disk_super);
2742
	stripesize = sectorsize;
2743
	fs_info->dirty_metadata_batch = nodesize * (1 + ilog2(nr_cpu_ids));
2744
	fs_info->delalloc_batch = sectorsize * 512 * (1 + ilog2(nr_cpu_ids));
2745

2746 2747 2748 2749 2750
	/* Cache block sizes */
	fs_info->nodesize = nodesize;
	fs_info->sectorsize = sectorsize;
	fs_info->stripesize = stripesize;

2751 2752 2753 2754 2755
	/*
	 * 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) &&
2756
	    (sectorsize != nodesize)) {
2757 2758 2759
		btrfs_err(fs_info,
"unequal nodesize/sectorsize (%u != %u) are not allowed for mixed block groups",
			nodesize, sectorsize);
2760 2761 2762
		goto fail_alloc;
	}

2763 2764 2765 2766
	/*
	 * Needn't use the lock because there is no other task which will
	 * update the flag.
	 */
L
Li Zefan 已提交
2767
	btrfs_set_super_incompat_flags(disk_super, features);
2768

2769 2770
	features = btrfs_super_compat_ro_flags(disk_super) &
		~BTRFS_FEATURE_COMPAT_RO_SUPP;
2771
	if (!sb_rdonly(sb) && features) {
2772 2773
		btrfs_err(fs_info,
	"cannot mount read-write because of unsupported optional features (%llx)",
2774
		       features);
2775
		err = -EINVAL;
2776
		goto fail_alloc;
2777
	}
2778

2779 2780 2781
	ret = btrfs_init_workqueues(fs_info, fs_devices);
	if (ret) {
		err = ret;
2782 2783
		goto fail_sb_buffer;
	}
2784

2785 2786 2787
	sb->s_bdi->congested_fn = btrfs_congested_fn;
	sb->s_bdi->congested_data = fs_info;
	sb->s_bdi->capabilities |= BDI_CAP_CGROUP_WRITEBACK;
2788
	sb->s_bdi->ra_pages = VM_MAX_READAHEAD * SZ_1K / PAGE_SIZE;
2789 2790
	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);
2791

2792 2793
	sb->s_blocksize = sectorsize;
	sb->s_blocksize_bits = blksize_bits(sectorsize);
2794
	memcpy(&sb->s_uuid, fs_info->fsid, BTRFS_FSID_SIZE);
2795

2796
	mutex_lock(&fs_info->chunk_mutex);
2797
	ret = btrfs_read_sys_array(fs_info);
2798
	mutex_unlock(&fs_info->chunk_mutex);
2799
	if (ret) {
2800
		btrfs_err(fs_info, "failed to read the system array: %d", ret);
2801
		goto fail_sb_buffer;
2802
	}
2803

2804
	generation = btrfs_super_chunk_root_generation(disk_super);
2805
	level = btrfs_super_chunk_root_level(disk_super);
2806

2807
	__setup_root(chunk_root, fs_info, BTRFS_CHUNK_TREE_OBJECTID);
2808

2809
	chunk_root->node = read_tree_block(fs_info,
2810
					   btrfs_super_chunk_root(disk_super),
2811
					   generation, level, NULL);
2812 2813
	if (IS_ERR(chunk_root->node) ||
	    !extent_buffer_uptodate(chunk_root->node)) {
2814
		btrfs_err(fs_info, "failed to read chunk root");
2815 2816
		if (!IS_ERR(chunk_root->node))
			free_extent_buffer(chunk_root->node);
2817
		chunk_root->node = NULL;
C
Chris Mason 已提交
2818
		goto fail_tree_roots;
2819
	}
2820 2821
	btrfs_set_root_node(&chunk_root->root_item, chunk_root->node);
	chunk_root->commit_root = btrfs_root_node(chunk_root);
2822

2823
	read_extent_buffer(chunk_root->node, fs_info->chunk_tree_uuid,
2824
	   btrfs_header_chunk_tree_uuid(chunk_root->node), BTRFS_UUID_SIZE);
2825

2826
	ret = btrfs_read_chunk_tree(fs_info);
Y
Yan Zheng 已提交
2827
	if (ret) {
2828
		btrfs_err(fs_info, "failed to read chunk tree: %d", ret);
C
Chris Mason 已提交
2829
		goto fail_tree_roots;
Y
Yan Zheng 已提交
2830
	}
2831

2832
	/*
2833 2834
	 * Keep the devid that is marked to be the target device for the
	 * device replace procedure
2835
	 */
2836
	btrfs_free_extra_devids(fs_devices, 0);
2837

2838
	if (!fs_devices->latest_bdev) {
2839
		btrfs_err(fs_info, "failed to read devices");
2840 2841 2842
		goto fail_tree_roots;
	}

C
Chris Mason 已提交
2843
retry_root_backup:
2844
	generation = btrfs_super_generation(disk_super);
2845
	level = btrfs_super_root_level(disk_super);
2846

2847
	tree_root->node = read_tree_block(fs_info,
2848
					  btrfs_super_root(disk_super),
2849
					  generation, level, NULL);
2850 2851
	if (IS_ERR(tree_root->node) ||
	    !extent_buffer_uptodate(tree_root->node)) {
2852
		btrfs_warn(fs_info, "failed to read tree root");
2853 2854
		if (!IS_ERR(tree_root->node))
			free_extent_buffer(tree_root->node);
2855
		tree_root->node = NULL;
C
Chris Mason 已提交
2856
		goto recovery_tree_root;
2857
	}
C
Chris Mason 已提交
2858

2859 2860
	btrfs_set_root_node(&tree_root->root_item, tree_root->node);
	tree_root->commit_root = btrfs_root_node(tree_root);
2861
	btrfs_set_root_refs(&tree_root->root_item, 1);
2862

2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874
	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);

2875
	ret = btrfs_read_roots(fs_info);
2876
	if (ret)
C
Chris Mason 已提交
2877
		goto recovery_tree_root;
2878

2879 2880 2881
	fs_info->generation = generation;
	fs_info->last_trans_committed = generation;

2882 2883
	ret = btrfs_recover_balance(fs_info);
	if (ret) {
2884
		btrfs_err(fs_info, "failed to recover balance: %d", ret);
2885 2886 2887
		goto fail_block_groups;
	}

2888 2889
	ret = btrfs_init_dev_stats(fs_info);
	if (ret) {
2890
		btrfs_err(fs_info, "failed to init dev_stats: %d", ret);
2891 2892 2893
		goto fail_block_groups;
	}

2894 2895
	ret = btrfs_init_dev_replace(fs_info);
	if (ret) {
2896
		btrfs_err(fs_info, "failed to init dev_replace: %d", ret);
2897 2898 2899
		goto fail_block_groups;
	}

2900
	btrfs_free_extra_devids(fs_devices, 1);
2901

2902 2903
	ret = btrfs_sysfs_add_fsid(fs_devices, NULL);
	if (ret) {
2904 2905
		btrfs_err(fs_info, "failed to init sysfs fsid interface: %d",
				ret);
2906 2907 2908 2909 2910
		goto fail_block_groups;
	}

	ret = btrfs_sysfs_add_device(fs_devices);
	if (ret) {
2911 2912
		btrfs_err(fs_info, "failed to init sysfs device interface: %d",
				ret);
2913 2914 2915
		goto fail_fsdev_sysfs;
	}

2916
	ret = btrfs_sysfs_add_mounted(fs_info);
2917
	if (ret) {
2918
		btrfs_err(fs_info, "failed to init sysfs interface: %d", ret);
2919
		goto fail_fsdev_sysfs;
2920 2921 2922 2923
	}

	ret = btrfs_init_space_info(fs_info);
	if (ret) {
2924
		btrfs_err(fs_info, "failed to initialize space info: %d", ret);
2925
		goto fail_sysfs;
2926 2927
	}

2928
	ret = btrfs_read_block_groups(fs_info);
2929
	if (ret) {
2930
		btrfs_err(fs_info, "failed to read block groups: %d", ret);
2931
		goto fail_sysfs;
2932
	}
2933

2934
	if (!sb_rdonly(sb) && !btrfs_check_rw_degradable(fs_info, NULL)) {
2935
		btrfs_warn(fs_info,
2936
		"writeable mount is not allowed due to too many missing devices");
2937
		goto fail_sysfs;
2938
	}
C
Chris Mason 已提交
2939

2940 2941
	fs_info->cleaner_kthread = kthread_run(cleaner_kthread, tree_root,
					       "btrfs-cleaner");
2942
	if (IS_ERR(fs_info->cleaner_kthread))
2943
		goto fail_sysfs;
2944 2945 2946 2947

	fs_info->transaction_kthread = kthread_run(transaction_kthread,
						   tree_root,
						   "btrfs-transaction");
2948
	if (IS_ERR(fs_info->transaction_kthread))
2949
		goto fail_cleaner;
2950

2951
	if (!btrfs_test_opt(fs_info, NOSSD) &&
C
Chris Mason 已提交
2952
	    !fs_info->fs_devices->rotating) {
2953
		btrfs_set_and_info(fs_info, SSD, "enabling ssd optimizations");
C
Chris Mason 已提交
2954 2955
	}

2956
	/*
2957
	 * Mount does not set all options immediately, we can do it now and do
2958 2959 2960
	 * not have to wait for transaction commit
	 */
	btrfs_apply_pending_changes(fs_info);
2961

2962
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
2963
	if (btrfs_test_opt(fs_info, CHECK_INTEGRITY)) {
2964
		ret = btrfsic_mount(fs_info, fs_devices,
2965
				    btrfs_test_opt(fs_info,
2966 2967 2968 2969
					CHECK_INTEGRITY_INCLUDING_EXTENT_DATA) ?
				    1 : 0,
				    fs_info->check_integrity_print_mask);
		if (ret)
2970 2971 2972
			btrfs_warn(fs_info,
				"failed to initialize integrity check module: %d",
				ret);
2973 2974
	}
#endif
2975 2976 2977
	ret = btrfs_read_qgroup_config(fs_info);
	if (ret)
		goto fail_trans_kthread;
2978

J
Josef Bacik 已提交
2979 2980 2981
	if (btrfs_build_ref_tree(fs_info))
		btrfs_err(fs_info, "couldn't build ref tree");

2982 2983
	/* do not make disk changes in broken FS or nologreplay is given */
	if (btrfs_super_log_root(disk_super) != 0 &&
2984
	    !btrfs_test_opt(fs_info, NOLOGREPLAY)) {
2985
		ret = btrfs_replay_log(fs_info, fs_devices);
2986
		if (ret) {
2987
			err = ret;
2988
			goto fail_qgroup;
2989
		}
2990
	}
Z
Zheng Yan 已提交
2991

2992
	ret = btrfs_find_orphan_roots(fs_info);
2993
	if (ret)
2994
		goto fail_qgroup;
2995

2996
	if (!sb_rdonly(sb)) {
2997
		ret = btrfs_cleanup_fs_roots(fs_info);
2998
		if (ret)
2999
			goto fail_qgroup;
3000 3001

		mutex_lock(&fs_info->cleaner_mutex);
3002
		ret = btrfs_recover_relocation(tree_root);
3003
		mutex_unlock(&fs_info->cleaner_mutex);
3004
		if (ret < 0) {
3005 3006
			btrfs_warn(fs_info, "failed to recover relocation: %d",
					ret);
3007
			err = -EINVAL;
3008
			goto fail_qgroup;
3009
		}
3010
	}
Z
Zheng Yan 已提交
3011

3012 3013
	location.objectid = BTRFS_FS_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
3014
	location.offset = 0;
3015 3016

	fs_info->fs_root = btrfs_read_fs_root_no_name(fs_info, &location);
3017 3018
	if (IS_ERR(fs_info->fs_root)) {
		err = PTR_ERR(fs_info->fs_root);
3019
		btrfs_warn(fs_info, "failed to read fs tree: %d", err);
3020
		goto fail_qgroup;
3021
	}
C
Chris Mason 已提交
3022

3023
	if (sb_rdonly(sb))
3024
		return 0;
I
Ilya Dryomov 已提交
3025

3026 3027
	if (btrfs_test_opt(fs_info, CLEAR_CACHE) &&
	    btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
3028 3029 3030 3031 3032 3033 3034 3035
		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) {
3036 3037 3038 3039 3040
		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);
3041
			close_ctree(fs_info);
3042 3043 3044 3045
			return ret;
		}
	}

3046
	if (btrfs_test_opt(fs_info, FREE_SPACE_TREE) &&
3047
	    !btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
3048
		btrfs_info(fs_info, "creating free space tree");
3049 3050
		ret = btrfs_create_free_space_tree(fs_info);
		if (ret) {
3051 3052
			btrfs_warn(fs_info,
				"failed to create free space tree: %d", ret);
3053
			close_ctree(fs_info);
3054 3055 3056 3057
			return ret;
		}
	}

3058 3059 3060
	down_read(&fs_info->cleanup_work_sem);
	if ((ret = btrfs_orphan_cleanup(fs_info->fs_root)) ||
	    (ret = btrfs_orphan_cleanup(fs_info->tree_root))) {
3061
		up_read(&fs_info->cleanup_work_sem);
3062
		close_ctree(fs_info);
3063 3064 3065
		return ret;
	}
	up_read(&fs_info->cleanup_work_sem);
I
Ilya Dryomov 已提交
3066

3067 3068
	ret = btrfs_resume_balance_async(fs_info);
	if (ret) {
3069
		btrfs_warn(fs_info, "failed to resume balance: %d", ret);
3070
		close_ctree(fs_info);
3071
		return ret;
3072 3073
	}

3074 3075
	ret = btrfs_resume_dev_replace_async(fs_info);
	if (ret) {
3076
		btrfs_warn(fs_info, "failed to resume device replace: %d", ret);
3077
		close_ctree(fs_info);
3078 3079 3080
		return ret;
	}

3081 3082
	btrfs_qgroup_rescan_resume(fs_info);

3083
	if (!fs_info->uuid_root) {
3084
		btrfs_info(fs_info, "creating UUID tree");
3085 3086
		ret = btrfs_create_uuid_tree(fs_info);
		if (ret) {
3087 3088
			btrfs_warn(fs_info,
				"failed to create the UUID tree: %d", ret);
3089
			close_ctree(fs_info);
3090 3091
			return ret;
		}
3092
	} else if (btrfs_test_opt(fs_info, RESCAN_UUID_TREE) ||
3093 3094
		   fs_info->generation !=
				btrfs_super_uuid_tree_generation(disk_super)) {
3095
		btrfs_info(fs_info, "checking UUID tree");
3096 3097
		ret = btrfs_check_uuid_tree(fs_info);
		if (ret) {
3098 3099
			btrfs_warn(fs_info,
				"failed to check the UUID tree: %d", ret);
3100
			close_ctree(fs_info);
3101 3102 3103
			return ret;
		}
	} else {
3104
		set_bit(BTRFS_FS_UPDATE_UUID_TREE_GEN, &fs_info->flags);
3105
	}
3106
	set_bit(BTRFS_FS_OPEN, &fs_info->flags);
3107

3108 3109 3110 3111 3112 3113
	/*
	 * 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
3114
	return 0;
C
Chris Mason 已提交
3115

3116 3117
fail_qgroup:
	btrfs_free_qgroup_config(fs_info);
3118 3119
fail_trans_kthread:
	kthread_stop(fs_info->transaction_kthread);
3120
	btrfs_cleanup_transaction(fs_info);
3121
	btrfs_free_fs_roots(fs_info);
3122
fail_cleaner:
3123
	kthread_stop(fs_info->cleaner_kthread);
3124 3125 3126 3127 3128 3129 3130

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

3131
fail_sysfs:
3132
	btrfs_sysfs_remove_mounted(fs_info);
3133

3134 3135 3136
fail_fsdev_sysfs:
	btrfs_sysfs_remove_fsid(fs_info->fs_devices);

3137
fail_block_groups:
J
Josef Bacik 已提交
3138
	btrfs_put_block_group_cache(fs_info);
C
Chris Mason 已提交
3139 3140 3141

fail_tree_roots:
	free_root_pointers(fs_info, 1);
3142
	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
3143

C
Chris Mason 已提交
3144
fail_sb_buffer:
L
Liu Bo 已提交
3145
	btrfs_stop_all_workers(fs_info);
3146
	btrfs_free_block_groups(fs_info);
3147
fail_alloc:
3148
fail_iput:
3149 3150
	btrfs_mapping_tree_free(&fs_info->mapping_tree);

3151
	iput(fs_info->btree_inode);
3152 3153
fail_bio_counter:
	percpu_counter_destroy(&fs_info->bio_counter);
3154 3155
fail_delalloc_bytes:
	percpu_counter_destroy(&fs_info->delalloc_bytes);
3156 3157
fail_dirty_metadata_bytes:
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
3158 3159
fail_srcu:
	cleanup_srcu_struct(&fs_info->subvol_srcu);
3160
fail:
D
David Woodhouse 已提交
3161
	btrfs_free_stripe_hash_table(fs_info);
3162
	btrfs_close_devices(fs_info->fs_devices);
A
Al Viro 已提交
3163
	return err;
C
Chris Mason 已提交
3164 3165

recovery_tree_root:
3166
	if (!btrfs_test_opt(fs_info, USEBACKUPROOT))
C
Chris Mason 已提交
3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181
		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;
3182
}
3183
ALLOW_ERROR_INJECTION(open_ctree, ERRNO);
3184

3185 3186 3187 3188 3189
static void btrfs_end_buffer_write_sync(struct buffer_head *bh, int uptodate)
{
	if (uptodate) {
		set_buffer_uptodate(bh);
	} else {
3190 3191 3192
		struct btrfs_device *device = (struct btrfs_device *)
			bh->b_private;

3193
		btrfs_warn_rl_in_rcu(device->fs_info,
3194
				"lost page write due to IO error on %s",
3195
					  rcu_str_deref(device->name));
3196
		/* note, we don't set_buffer_write_io_error because we have
3197 3198
		 * our own ways of dealing with the IO errors
		 */
3199
		clear_buffer_uptodate(bh);
3200
		btrfs_dev_stat_inc_and_print(device, BTRFS_DEV_STAT_WRITE_ERRS);
3201 3202 3203 3204 3205
	}
	unlock_buffer(bh);
	put_bh(bh);
}

3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216
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;

3217
	bh = __bread(bdev, bytenr / BTRFS_BDEV_BLOCKSIZE, BTRFS_SUPER_INFO_SIZE);
3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236
	/*
	 * 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
3237 3238 3239 3240 3241 3242 3243
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;
3244
	int ret = -EINVAL;
Y
Yan Zheng 已提交
3245 3246 3247 3248 3249 3250 3251

	/* 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++) {
3252 3253
		ret = btrfs_read_dev_one_super(bdev, i, &bh);
		if (ret)
Y
Yan Zheng 已提交
3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265
			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);
		}
	}
3266 3267 3268 3269

	if (!latest)
		return ERR_PTR(ret);

Y
Yan Zheng 已提交
3270 3271 3272
	return latest;
}

3273
/*
3274 3275
 * Write superblock @sb to the @device. Do not wait for completion, all the
 * buffer heads we write are pinned.
3276
 *
3277 3278 3279
 * 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.
3280
 *
3281
 * Return number of errors when buffer head is not found or submission fails.
3282
 */
Y
Yan Zheng 已提交
3283
static int write_dev_supers(struct btrfs_device *device,
3284
			    struct btrfs_super_block *sb, int max_mirrors)
Y
Yan Zheng 已提交
3285 3286 3287 3288 3289 3290 3291
{
	struct buffer_head *bh;
	int i;
	int ret;
	int errors = 0;
	u32 crc;
	u64 bytenr;
3292
	int op_flags;
Y
Yan Zheng 已提交
3293 3294 3295 3296 3297 3298

	if (max_mirrors == 0)
		max_mirrors = BTRFS_SUPER_MIRROR_MAX;

	for (i = 0; i < max_mirrors; i++) {
		bytenr = btrfs_sb_offset(i);
3299 3300
		if (bytenr + BTRFS_SUPER_INFO_SIZE >=
		    device->commit_total_bytes)
Y
Yan Zheng 已提交
3301 3302
			break;

3303
		btrfs_set_super_bytenr(sb, bytenr);
3304

3305 3306 3307 3308
		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);
3309

3310
		/* One reference for us, and we leave it for the caller */
3311
		bh = __getblk(device->bdev, bytenr / BTRFS_BDEV_BLOCKSIZE,
3312 3313 3314 3315 3316 3317
			      BTRFS_SUPER_INFO_SIZE);
		if (!bh) {
			btrfs_err(device->fs_info,
			    "couldn't get super buffer head for bytenr %llu",
			    bytenr);
			errors++;
3318
			continue;
3319
		}
3320

3321
		memcpy(bh->b_data, sb, BTRFS_SUPER_INFO_SIZE);
Y
Yan Zheng 已提交
3322

3323 3324
		/* one reference for submit_bh */
		get_bh(bh);
3325

3326 3327 3328 3329
		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
3330

C
Chris Mason 已提交
3331 3332 3333 3334
		/*
		 * we fua the first super.  The others we allow
		 * to go down lazy.
		 */
3335 3336 3337 3338
		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);
3339
		if (ret)
Y
Yan Zheng 已提交
3340 3341 3342 3343 3344
			errors++;
	}
	return errors < i ? 0 : -1;
}

3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356
/*
 * 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;
3357
	bool primary_failed = false;
3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368
	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;

3369 3370
		bh = __find_get_block(device->bdev,
				      bytenr / BTRFS_BDEV_BLOCKSIZE,
3371 3372 3373
				      BTRFS_SUPER_INFO_SIZE);
		if (!bh) {
			errors++;
3374 3375
			if (i == 0)
				primary_failed = true;
3376 3377 3378
			continue;
		}
		wait_on_buffer(bh);
3379
		if (!buffer_uptodate(bh)) {
3380
			errors++;
3381 3382 3383
			if (i == 0)
				primary_failed = true;
		}
3384 3385 3386 3387 3388 3389 3390 3391

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

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

3392 3393 3394 3395 3396 3397 3398
	/* 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;
	}

3399 3400 3401
	return errors < i ? 0 : -1;
}

C
Chris Mason 已提交
3402 3403 3404 3405
/*
 * endio for the write_dev_flush, this will wake anyone waiting
 * for the barrier when it is done
 */
3406
static void btrfs_end_empty_barrier(struct bio *bio)
C
Chris Mason 已提交
3407
{
3408
	complete(bio->bi_private);
C
Chris Mason 已提交
3409 3410 3411
}

/*
3412 3413
 * 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
3414
 */
3415
static void write_dev_flush(struct btrfs_device *device)
C
Chris Mason 已提交
3416
{
3417
	struct request_queue *q = bdev_get_queue(device->bdev);
3418
	struct bio *bio = device->flush_bio;
C
Chris Mason 已提交
3419

3420
	if (!test_bit(QUEUE_FLAG_WC, &q->queue_flags))
3421
		return;
C
Chris Mason 已提交
3422

3423
	bio_reset(bio);
C
Chris Mason 已提交
3424
	bio->bi_end_io = btrfs_end_empty_barrier;
3425
	bio_set_dev(bio, device->bdev);
3426
	bio->bi_opf = REQ_OP_WRITE | REQ_SYNC | REQ_PREFLUSH;
C
Chris Mason 已提交
3427 3428 3429
	init_completion(&device->flush_wait);
	bio->bi_private = &device->flush_wait;

3430
	btrfsic_submit_bio(bio);
3431
	set_bit(BTRFS_DEV_STATE_FLUSH_SENT, &device->dev_state);
3432
}
C
Chris Mason 已提交
3433

3434 3435 3436
/*
 * If the flush bio has been submitted by write_dev_flush, wait for it.
 */
3437
static blk_status_t wait_dev_flush(struct btrfs_device *device)
3438 3439
{
	struct bio *bio = device->flush_bio;
C
Chris Mason 已提交
3440

3441
	if (!test_bit(BTRFS_DEV_STATE_FLUSH_SENT, &device->dev_state))
3442
		return BLK_STS_OK;
C
Chris Mason 已提交
3443

3444
	clear_bit(BTRFS_DEV_STATE_FLUSH_SENT, &device->dev_state);
3445
	wait_for_completion_io(&device->flush_wait);
C
Chris Mason 已提交
3446

3447
	return bio->bi_status;
C
Chris Mason 已提交
3448 3449
}

3450
static int check_barrier_error(struct btrfs_fs_info *fs_info)
3451
{
3452
	if (!btrfs_check_rw_degradable(fs_info, NULL))
3453
		return -EIO;
C
Chris Mason 已提交
3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464
	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;
3465
	int errors_wait = 0;
3466
	blk_status_t ret;
C
Chris Mason 已提交
3467

3468
	lockdep_assert_held(&info->fs_devices->device_list_mutex);
C
Chris Mason 已提交
3469 3470
	/* send down all the barriers */
	head = &info->fs_devices->devices;
3471
	list_for_each_entry(dev, head, dev_list) {
3472
		if (test_bit(BTRFS_DEV_STATE_MISSING, &dev->dev_state))
3473
			continue;
3474
		if (!dev->bdev)
C
Chris Mason 已提交
3475
			continue;
3476
		if (!test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &dev->dev_state) ||
3477
		    !test_bit(BTRFS_DEV_STATE_WRITEABLE, &dev->dev_state))
C
Chris Mason 已提交
3478 3479
			continue;

3480
		write_dev_flush(dev);
3481
		dev->last_flush_error = BLK_STS_OK;
C
Chris Mason 已提交
3482 3483 3484
	}

	/* wait for all the barriers */
3485
	list_for_each_entry(dev, head, dev_list) {
3486
		if (test_bit(BTRFS_DEV_STATE_MISSING, &dev->dev_state))
3487
			continue;
C
Chris Mason 已提交
3488
		if (!dev->bdev) {
3489
			errors_wait++;
C
Chris Mason 已提交
3490 3491
			continue;
		}
3492
		if (!test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &dev->dev_state) ||
3493
		    !test_bit(BTRFS_DEV_STATE_WRITEABLE, &dev->dev_state))
C
Chris Mason 已提交
3494 3495
			continue;

3496
		ret = wait_dev_flush(dev);
3497 3498
		if (ret) {
			dev->last_flush_error = ret;
3499 3500
			btrfs_dev_stat_inc_and_print(dev,
					BTRFS_DEV_STAT_FLUSH_ERRS);
3501
			errors_wait++;
3502 3503 3504
		}
	}

3505
	if (errors_wait) {
3506 3507 3508 3509 3510
		/*
		 * 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.
		 */
3511
		return check_barrier_error(info);
C
Chris Mason 已提交
3512 3513 3514 3515
	}
	return 0;
}

3516 3517
int btrfs_get_num_tolerated_disk_barrier_failures(u64 flags)
{
3518 3519
	int raid_type;
	int min_tolerated = INT_MAX;
3520

3521 3522 3523 3524 3525
	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);
3526

3527 3528 3529 3530 3531 3532 3533 3534 3535
	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);
	}
3536

3537
	if (min_tolerated == INT_MAX) {
3538
		pr_warn("BTRFS: unknown raid flag: %llu", flags);
3539 3540 3541 3542
		min_tolerated = 0;
	}

	return min_tolerated;
3543 3544
}

3545
int write_all_supers(struct btrfs_fs_info *fs_info, int max_mirrors)
3546
{
3547
	struct list_head *head;
3548
	struct btrfs_device *dev;
3549
	struct btrfs_super_block *sb;
3550 3551 3552
	struct btrfs_dev_item *dev_item;
	int ret;
	int do_barriers;
3553 3554
	int max_errors;
	int total_errors = 0;
3555
	u64 flags;
3556

3557
	do_barriers = !btrfs_test_opt(fs_info, NOBARRIER);
3558 3559 3560 3561 3562 3563 3564 3565

	/*
	 * 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);
3566

3567
	sb = fs_info->super_for_commit;
3568
	dev_item = &sb->dev_item;
3569

3570 3571 3572
	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
3573

3574
	if (do_barriers) {
3575
		ret = barrier_all_devices(fs_info);
3576 3577
		if (ret) {
			mutex_unlock(
3578 3579 3580
				&fs_info->fs_devices->device_list_mutex);
			btrfs_handle_fs_error(fs_info, ret,
					      "errors while submitting device barriers.");
3581 3582 3583
			return ret;
		}
	}
C
Chris Mason 已提交
3584

3585
	list_for_each_entry(dev, head, dev_list) {
3586 3587 3588 3589
		if (!dev->bdev) {
			total_errors++;
			continue;
		}
3590
		if (!test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &dev->dev_state) ||
3591
		    !test_bit(BTRFS_DEV_STATE_WRITEABLE, &dev->dev_state))
3592 3593
			continue;

Y
Yan Zheng 已提交
3594
		btrfs_set_stack_device_generation(dev_item, 0);
3595 3596
		btrfs_set_stack_device_type(dev_item, dev->type);
		btrfs_set_stack_device_id(dev_item, dev->devid);
3597
		btrfs_set_stack_device_total_bytes(dev_item,
3598
						   dev->commit_total_bytes);
3599 3600
		btrfs_set_stack_device_bytes_used(dev_item,
						  dev->commit_bytes_used);
3601 3602 3603 3604
		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);
3605
		memcpy(dev_item->fsid, dev->fs_devices->fsid, BTRFS_FSID_SIZE);
Y
Yan Zheng 已提交
3606

3607 3608 3609
		flags = btrfs_super_flags(sb);
		btrfs_set_super_flags(sb, flags | BTRFS_HEADER_FLAG_WRITTEN);

3610
		ret = write_dev_supers(dev, sb, max_mirrors);
3611 3612
		if (ret)
			total_errors++;
3613
	}
3614
	if (total_errors > max_errors) {
3615 3616 3617
		btrfs_err(fs_info, "%d errors while writing supers",
			  total_errors);
		mutex_unlock(&fs_info->fs_devices->device_list_mutex);
3618

3619
		/* FUA is masked off if unsupported and can't be the reason */
3620 3621 3622
		btrfs_handle_fs_error(fs_info, -EIO,
				      "%d errors while writing supers",
				      total_errors);
3623
		return -EIO;
3624
	}
3625

Y
Yan Zheng 已提交
3626
	total_errors = 0;
3627
	list_for_each_entry(dev, head, dev_list) {
3628 3629
		if (!dev->bdev)
			continue;
3630
		if (!test_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &dev->dev_state) ||
3631
		    !test_bit(BTRFS_DEV_STATE_WRITEABLE, &dev->dev_state))
3632 3633
			continue;

3634
		ret = wait_dev_supers(dev, max_mirrors);
Y
Yan Zheng 已提交
3635 3636
		if (ret)
			total_errors++;
3637
	}
3638
	mutex_unlock(&fs_info->fs_devices->device_list_mutex);
3639
	if (total_errors > max_errors) {
3640 3641 3642
		btrfs_handle_fs_error(fs_info, -EIO,
				      "%d errors while writing supers",
				      total_errors);
3643
		return -EIO;
3644
	}
3645 3646 3647
	return 0;
}

3648 3649 3650
/* 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
3651
{
3652
	spin_lock(&fs_info->fs_roots_radix_lock);
C
Chris Mason 已提交
3653 3654
	radix_tree_delete(&fs_info->fs_roots_radix,
			  (unsigned long)root->root_key.objectid);
3655
	spin_unlock(&fs_info->fs_roots_radix_lock);
3656 3657 3658 3659

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

L
Liu Bo 已提交
3660
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
L
Liu Bo 已提交
3661
		btrfs_free_log(NULL, root);
L
Liu Bo 已提交
3662 3663 3664 3665 3666 3667 3668
		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
3669

3670 3671 3672 3673
	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);
3674 3675 3676 3677 3678
	free_fs_root(root);
}

static void free_fs_root(struct btrfs_root *root)
{
3679
	iput(root->ino_cache_inode);
3680
	WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree));
3681
	btrfs_free_block_rsv(root->fs_info, root->orphan_block_rsv);
3682
	root->orphan_block_rsv = NULL;
3683 3684
	if (root->anon_dev)
		free_anon_bdev(root->anon_dev);
3685 3686
	if (root->subv_writers)
		btrfs_free_subvolume_writers(root->subv_writers);
3687 3688
	free_extent_buffer(root->node);
	free_extent_buffer(root->commit_root);
3689 3690
	kfree(root->free_ino_ctl);
	kfree(root->free_ino_pinned);
C
Chris Mason 已提交
3691
	kfree(root->name);
3692
	btrfs_put_fs_root(root);
C
Chris Mason 已提交
3693 3694
}

3695 3696 3697
void btrfs_free_fs_root(struct btrfs_root *root)
{
	free_fs_root(root);
C
Chris Mason 已提交
3698 3699
}

Y
Yan Zheng 已提交
3700
int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
C
Chris Mason 已提交
3701
{
Y
Yan Zheng 已提交
3702 3703
	u64 root_objectid = 0;
	struct btrfs_root *gang[8];
3704 3705 3706 3707
	int i = 0;
	int err = 0;
	unsigned int ret = 0;
	int index;
3708

Y
Yan Zheng 已提交
3709
	while (1) {
3710
		index = srcu_read_lock(&fs_info->subvol_srcu);
Y
Yan Zheng 已提交
3711 3712 3713
		ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix,
					     (void **)gang, root_objectid,
					     ARRAY_SIZE(gang));
3714 3715
		if (!ret) {
			srcu_read_unlock(&fs_info->subvol_srcu, index);
Y
Yan Zheng 已提交
3716
			break;
3717
		}
3718
		root_objectid = gang[ret - 1]->root_key.objectid + 1;
3719

Y
Yan Zheng 已提交
3720
		for (i = 0; i < ret; i++) {
3721 3722 3723 3724 3725 3726 3727 3728 3729
			/* 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);
3730

3731 3732 3733
		for (i = 0; i < ret; i++) {
			if (!gang[i])
				continue;
Y
Yan Zheng 已提交
3734
			root_objectid = gang[i]->root_key.objectid;
3735 3736
			err = btrfs_orphan_cleanup(gang[i]);
			if (err)
3737 3738
				break;
			btrfs_put_fs_root(gang[i]);
Y
Yan Zheng 已提交
3739 3740 3741
		}
		root_objectid++;
	}
3742 3743 3744 3745 3746 3747 3748

	/* 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
3749
}
3750

3751
int btrfs_commit_super(struct btrfs_fs_info *fs_info)
Y
Yan Zheng 已提交
3752
{
3753
	struct btrfs_root *root = fs_info->tree_root;
Y
Yan Zheng 已提交
3754
	struct btrfs_trans_handle *trans;
3755

3756
	mutex_lock(&fs_info->cleaner_mutex);
3757
	btrfs_run_delayed_iputs(fs_info);
3758 3759
	mutex_unlock(&fs_info->cleaner_mutex);
	wake_up_process(fs_info->cleaner_kthread);
3760 3761

	/* wait until ongoing cleanup work done */
3762 3763
	down_write(&fs_info->cleanup_work_sem);
	up_write(&fs_info->cleanup_work_sem);
3764

3765
	trans = btrfs_join_transaction(root);
3766 3767
	if (IS_ERR(trans))
		return PTR_ERR(trans);
3768
	return btrfs_commit_transaction(trans);
Y
Yan Zheng 已提交
3769 3770
}

3771
void close_ctree(struct btrfs_fs_info *fs_info)
Y
Yan Zheng 已提交
3772
{
3773
	struct btrfs_root *root = fs_info->tree_root;
Y
Yan Zheng 已提交
3774 3775
	int ret;

3776
	set_bit(BTRFS_FS_CLOSING_START, &fs_info->flags);
Y
Yan Zheng 已提交
3777

3778
	/* wait for the qgroup rescan worker to stop */
3779
	btrfs_qgroup_wait_for_completion(fs_info, false);
3780

S
Stefan Behrens 已提交
3781 3782 3783 3784 3785
	/* 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);

3786
	/* pause restriper - we want to resume on mount */
3787
	btrfs_pause_balance(fs_info);
3788

3789 3790
	btrfs_dev_replace_suspend_for_unmount(fs_info);

3791
	btrfs_scrub_cancel(fs_info);
C
Chris Mason 已提交
3792 3793 3794 3795 3796 3797

	/* 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 */
3798
	btrfs_cleanup_defrag_inodes(fs_info);
C
Chris Mason 已提交
3799

3800 3801
	cancel_work_sync(&fs_info->async_reclaim_work);

3802
	if (!sb_rdonly(fs_info->sb)) {
3803 3804 3805 3806 3807
		/*
		 * 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.
		 */
3808
		btrfs_delete_unused_bgs(fs_info);
3809

3810
		ret = btrfs_commit_super(fs_info);
L
liubo 已提交
3811
		if (ret)
3812
			btrfs_err(fs_info, "commit super ret %d", ret);
L
liubo 已提交
3813 3814
	}

3815
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state))
3816
		btrfs_error_commit_super(fs_info);
3817

A
Al Viro 已提交
3818 3819
	kthread_stop(fs_info->transaction_kthread);
	kthread_stop(fs_info->cleaner_kthread);
3820

3821
	set_bit(BTRFS_FS_CLOSING_DONE, &fs_info->flags);
3822

3823
	btrfs_free_qgroup_config(fs_info);
3824

3825
	if (percpu_counter_sum(&fs_info->delalloc_bytes)) {
3826
		btrfs_info(fs_info, "at unmount delalloc count %lld",
3827
		       percpu_counter_sum(&fs_info->delalloc_bytes));
C
Chris Mason 已提交
3828
	}
3829

3830
	btrfs_sysfs_remove_mounted(fs_info);
3831
	btrfs_sysfs_remove_fsid(fs_info->fs_devices);
3832

3833
	btrfs_free_fs_roots(fs_info);
3834

3835 3836
	btrfs_put_block_group_cache(fs_info);

3837 3838 3839 3840 3841
	/*
	 * 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);
3842 3843
	btrfs_stop_all_workers(fs_info);

3844 3845
	btrfs_free_block_groups(fs_info);

3846
	clear_bit(BTRFS_FS_OPEN, &fs_info->flags);
3847
	free_root_pointers(fs_info, 1);
3848

3849
	iput(fs_info->btree_inode);
3850

3851
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
3852
	if (btrfs_test_opt(fs_info, CHECK_INTEGRITY))
3853
		btrfsic_unmount(fs_info->fs_devices);
3854 3855
#endif

3856
	btrfs_close_devices(fs_info->fs_devices);
3857
	btrfs_mapping_tree_free(&fs_info->mapping_tree);
3858

3859
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
3860
	percpu_counter_destroy(&fs_info->delalloc_bytes);
3861
	percpu_counter_destroy(&fs_info->bio_counter);
3862
	cleanup_srcu_struct(&fs_info->subvol_srcu);
3863

D
David Woodhouse 已提交
3864
	btrfs_free_stripe_hash_table(fs_info);
J
Josef Bacik 已提交
3865
	btrfs_free_ref_cache(fs_info);
D
David Woodhouse 已提交
3866

3867
	__btrfs_free_block_rsv(root->orphan_block_rsv);
3868
	root->orphan_block_rsv = NULL;
3869 3870 3871 3872 3873 3874 3875 3876 3877

	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);
	}
3878 3879
}

3880 3881
int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid,
			  int atomic)
3882
{
3883
	int ret;
3884
	struct inode *btree_inode = buf->pages[0]->mapping->host;
3885

3886
	ret = extent_buffer_uptodate(buf);
3887 3888 3889 3890
	if (!ret)
		return ret;

	ret = verify_parent_transid(&BTRFS_I(btree_inode)->io_tree, buf,
3891 3892 3893
				    parent_transid, atomic);
	if (ret == -EAGAIN)
		return ret;
3894
	return !ret;
3895 3896 3897 3898
}

void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
{
3899
	struct btrfs_fs_info *fs_info;
3900
	struct btrfs_root *root;
3901
	u64 transid = btrfs_header_generation(buf);
3902
	int was_dirty;
3903

3904 3905 3906 3907 3908 3909 3910 3911 3912 3913
#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;
3914
	fs_info = root->fs_info;
3915
	btrfs_assert_tree_locked(buf);
3916
	if (transid != fs_info->generation)
J
Jeff Mahoney 已提交
3917
		WARN(1, KERN_CRIT "btrfs transid mismatch buffer %llu, found %llu running %llu\n",
3918
			buf->start, transid, fs_info->generation);
3919
	was_dirty = set_extent_buffer_dirty(buf);
3920
	if (!was_dirty)
3921 3922 3923
		percpu_counter_add_batch(&fs_info->dirty_metadata_bytes,
					 buf->len,
					 fs_info->dirty_metadata_batch);
3924
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
3925 3926 3927 3928 3929 3930
	/*
	 * 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 &&
3931
	    btrfs_check_leaf_relaxed(fs_info, buf)) {
3932
		btrfs_print_leaf(buf);
3933 3934 3935
		ASSERT(0);
	}
#endif
3936 3937
}

3938
static void __btrfs_btree_balance_dirty(struct btrfs_fs_info *fs_info,
3939
					int flush_delayed)
3940 3941 3942 3943 3944
{
	/*
	 * looks as though older kernels can get into trouble with
	 * this code, they end up stuck in balance_dirty_pages forever
	 */
3945
	int ret;
3946 3947 3948 3949

	if (current->flags & PF_MEMALLOC)
		return;

3950
	if (flush_delayed)
3951
		btrfs_balance_delayed_items(fs_info);
3952

3953
	ret = percpu_counter_compare(&fs_info->dirty_metadata_bytes,
3954 3955
				     BTRFS_DIRTY_METADATA_THRESH);
	if (ret > 0) {
3956
		balance_dirty_pages_ratelimited(fs_info->btree_inode->i_mapping);
3957 3958 3959
	}
}

3960
void btrfs_btree_balance_dirty(struct btrfs_fs_info *fs_info)
C
Chris Mason 已提交
3961
{
3962
	__btrfs_btree_balance_dirty(fs_info, 1);
3963
}
3964

3965
void btrfs_btree_balance_dirty_nodelay(struct btrfs_fs_info *fs_info)
3966
{
3967
	__btrfs_btree_balance_dirty(fs_info, 0);
C
Chris Mason 已提交
3968
}
3969

3970 3971
int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid, int level,
		      struct btrfs_key *first_key)
3972
{
3973
	struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root;
3974 3975
	struct btrfs_fs_info *fs_info = root->fs_info;

3976 3977
	return btree_read_extent_buffer_pages(fs_info, buf, parent_transid,
					      level, first_key);
3978
}
3979

3980
static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info)
L
liubo 已提交
3981
{
D
David Sterba 已提交
3982
	struct btrfs_super_block *sb = fs_info->super_copy;
3983 3984
	u64 nodesize = btrfs_super_nodesize(sb);
	u64 sectorsize = btrfs_super_sectorsize(sb);
D
David Sterba 已提交
3985 3986
	int ret = 0;

3987
	if (btrfs_super_magic(sb) != BTRFS_MAGIC) {
3988
		btrfs_err(fs_info, "no valid FS found");
3989 3990
		ret = -EINVAL;
	}
3991 3992
	if (btrfs_super_flags(sb) & ~BTRFS_SUPER_FLAG_SUPP) {
		btrfs_err(fs_info, "unrecognized or unsupported super flag: %llu",
3993
				btrfs_super_flags(sb) & ~BTRFS_SUPER_FLAG_SUPP);
3994 3995
		ret = -EINVAL;
	}
3996
	if (btrfs_super_root_level(sb) >= BTRFS_MAX_LEVEL) {
3997
		btrfs_err(fs_info, "tree_root level too big: %d >= %d",
3998
				btrfs_super_root_level(sb), BTRFS_MAX_LEVEL);
D
David Sterba 已提交
3999 4000
		ret = -EINVAL;
	}
4001
	if (btrfs_super_chunk_root_level(sb) >= BTRFS_MAX_LEVEL) {
4002
		btrfs_err(fs_info, "chunk_root level too big: %d >= %d",
4003
				btrfs_super_chunk_root_level(sb), BTRFS_MAX_LEVEL);
D
David Sterba 已提交
4004 4005
		ret = -EINVAL;
	}
4006
	if (btrfs_super_log_root_level(sb) >= BTRFS_MAX_LEVEL) {
4007
		btrfs_err(fs_info, "log_root level too big: %d >= %d",
4008
				btrfs_super_log_root_level(sb), BTRFS_MAX_LEVEL);
D
David Sterba 已提交
4009 4010 4011
		ret = -EINVAL;
	}

D
David Sterba 已提交
4012
	/*
4013 4014
	 * 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
4015
	 */
4016 4017
	if (!is_power_of_2(sectorsize) || sectorsize < 4096 ||
	    sectorsize > BTRFS_MAX_METADATA_BLOCKSIZE) {
4018
		btrfs_err(fs_info, "invalid sectorsize %llu", sectorsize);
4019 4020 4021
		ret = -EINVAL;
	}
	/* Only PAGE SIZE is supported yet */
4022
	if (sectorsize != PAGE_SIZE) {
4023 4024 4025
		btrfs_err(fs_info,
			"sectorsize %llu not supported yet, only support %lu",
			sectorsize, PAGE_SIZE);
4026 4027 4028 4029
		ret = -EINVAL;
	}
	if (!is_power_of_2(nodesize) || nodesize < sectorsize ||
	    nodesize > BTRFS_MAX_METADATA_BLOCKSIZE) {
4030
		btrfs_err(fs_info, "invalid nodesize %llu", nodesize);
4031 4032 4033
		ret = -EINVAL;
	}
	if (nodesize != le32_to_cpu(sb->__unused_leafsize)) {
4034 4035
		btrfs_err(fs_info, "invalid leafsize %u, should be %llu",
			  le32_to_cpu(sb->__unused_leafsize), nodesize);
4036 4037 4038 4039 4040
		ret = -EINVAL;
	}

	/* Root alignment check */
	if (!IS_ALIGNED(btrfs_super_root(sb), sectorsize)) {
4041 4042
		btrfs_warn(fs_info, "tree_root block unaligned: %llu",
			   btrfs_super_root(sb));
4043 4044 4045
		ret = -EINVAL;
	}
	if (!IS_ALIGNED(btrfs_super_chunk_root(sb), sectorsize)) {
4046 4047
		btrfs_warn(fs_info, "chunk_root block unaligned: %llu",
			   btrfs_super_chunk_root(sb));
4048 4049
		ret = -EINVAL;
	}
4050
	if (!IS_ALIGNED(btrfs_super_log_root(sb), sectorsize)) {
4051 4052
		btrfs_warn(fs_info, "log_root block unaligned: %llu",
			   btrfs_super_log_root(sb));
4053 4054 4055
		ret = -EINVAL;
	}

4056
	if (memcmp(fs_info->fsid, sb->dev_item.fsid, BTRFS_FSID_SIZE) != 0) {
4057 4058 4059
		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
4060 4061 4062 4063 4064 4065 4066
		ret = -EINVAL;
	}

	/*
	 * Hint to catch really bogus numbers, bitflips or so, more exact checks are
	 * done later
	 */
4067 4068
	if (btrfs_super_bytes_used(sb) < 6 * btrfs_super_nodesize(sb)) {
		btrfs_err(fs_info, "bytes_used is too small %llu",
4069
			  btrfs_super_bytes_used(sb));
4070 4071
		ret = -EINVAL;
	}
4072
	if (!is_power_of_2(btrfs_super_stripesize(sb))) {
4073
		btrfs_err(fs_info, "invalid stripesize %u",
4074
			  btrfs_super_stripesize(sb));
4075 4076
		ret = -EINVAL;
	}
4077
	if (btrfs_super_num_devices(sb) > (1UL << 31))
4078 4079
		btrfs_warn(fs_info, "suspicious number of devices: %llu",
			   btrfs_super_num_devices(sb));
4080
	if (btrfs_super_num_devices(sb) == 0) {
4081
		btrfs_err(fs_info, "number of devices is 0");
4082 4083
		ret = -EINVAL;
	}
D
David Sterba 已提交
4084

4085
	if (btrfs_super_bytenr(sb) != BTRFS_SUPER_INFO_OFFSET) {
4086 4087
		btrfs_err(fs_info, "super offset mismatch %llu != %u",
			  btrfs_super_bytenr(sb), BTRFS_SUPER_INFO_OFFSET);
D
David Sterba 已提交
4088 4089 4090
		ret = -EINVAL;
	}

4091 4092 4093 4094 4095
	/*
	 * 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) {
4096 4097 4098
		btrfs_err(fs_info, "system chunk array too big %u > %u",
			  btrfs_super_sys_array_size(sb),
			  BTRFS_SYSTEM_CHUNK_ARRAY_SIZE);
4099 4100 4101 4102
		ret = -EINVAL;
	}
	if (btrfs_super_sys_array_size(sb) < sizeof(struct btrfs_disk_key)
			+ sizeof(struct btrfs_chunk)) {
4103 4104 4105 4106
		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));
4107 4108 4109
		ret = -EINVAL;
	}

D
David Sterba 已提交
4110 4111 4112 4113
	/*
	 * 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.
	 */
4114
	if (btrfs_super_generation(sb) < btrfs_super_chunk_root_generation(sb))
4115 4116 4117 4118
		btrfs_warn(fs_info,
			"suspicious: generation < chunk_root_generation: %llu < %llu",
			btrfs_super_generation(sb),
			btrfs_super_chunk_root_generation(sb));
4119 4120
	if (btrfs_super_generation(sb) < btrfs_super_cache_generation(sb)
	    && btrfs_super_cache_generation(sb) != (u64)-1)
4121 4122 4123 4124
		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
4125 4126

	return ret;
L
liubo 已提交
4127 4128
}

4129
static void btrfs_error_commit_super(struct btrfs_fs_info *fs_info)
L
liubo 已提交
4130
{
4131
	mutex_lock(&fs_info->cleaner_mutex);
4132
	btrfs_run_delayed_iputs(fs_info);
4133
	mutex_unlock(&fs_info->cleaner_mutex);
L
liubo 已提交
4134

4135 4136
	down_write(&fs_info->cleanup_work_sem);
	up_write(&fs_info->cleanup_work_sem);
L
liubo 已提交
4137 4138

	/* cleanup FS via transaction */
4139
	btrfs_cleanup_transaction(fs_info);
L
liubo 已提交
4140 4141
}

4142
static void btrfs_destroy_ordered_extents(struct btrfs_root *root)
L
liubo 已提交
4143 4144 4145
{
	struct btrfs_ordered_extent *ordered;

4146
	spin_lock(&root->ordered_extent_lock);
4147 4148 4149 4150
	/*
	 * This will just short circuit the ordered completion stuff which will
	 * make sure the ordered extent gets properly cleaned up.
	 */
4151
	list_for_each_entry(ordered, &root->ordered_extents,
4152 4153
			    root_extent_list)
		set_bit(BTRFS_ORDERED_IOERR, &ordered->flags);
4154 4155 4156 4157 4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168
	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);
4169 4170
		list_move_tail(&root->ordered_root,
			       &fs_info->ordered_roots);
4171

4172
		spin_unlock(&fs_info->ordered_root_lock);
4173 4174
		btrfs_destroy_ordered_extents(root);

4175 4176
		cond_resched();
		spin_lock(&fs_info->ordered_root_lock);
4177 4178
	}
	spin_unlock(&fs_info->ordered_root_lock);
L
liubo 已提交
4179 4180
}

4181
static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
4182
				      struct btrfs_fs_info *fs_info)
L
liubo 已提交
4183 4184 4185 4186 4187 4188 4189 4190 4191
{
	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);
4192
	if (atomic_read(&delayed_refs->num_entries) == 0) {
4193
		spin_unlock(&delayed_refs->lock);
4194
		btrfs_info(fs_info, "delayed_refs has NO entry");
L
liubo 已提交
4195 4196 4197
		return ret;
	}

4198 4199
	while ((node = rb_first(&delayed_refs->href_root)) != NULL) {
		struct btrfs_delayed_ref_head *head;
4200
		struct rb_node *n;
4201
		bool pin_bytes = false;
L
liubo 已提交
4202

4203 4204 4205
		head = rb_entry(node, struct btrfs_delayed_ref_head,
				href_node);
		if (!mutex_trylock(&head->mutex)) {
4206
			refcount_inc(&head->refs);
4207
			spin_unlock(&delayed_refs->lock);
4208

4209
			mutex_lock(&head->mutex);
4210
			mutex_unlock(&head->mutex);
4211
			btrfs_put_delayed_ref_head(head);
4212 4213 4214 4215
			spin_lock(&delayed_refs->lock);
			continue;
		}
		spin_lock(&head->lock);
4216 4217 4218
		while ((n = rb_first(&head->ref_tree)) != NULL) {
			ref = rb_entry(n, struct btrfs_delayed_ref_node,
				       ref_node);
4219
			ref->in_tree = 0;
4220 4221
			rb_erase(&ref->ref_node, &head->ref_tree);
			RB_CLEAR_NODE(&ref->ref_node);
4222 4223
			if (!list_empty(&ref->add_list))
				list_del(&ref->add_list);
4224 4225
			atomic_dec(&delayed_refs->num_entries);
			btrfs_put_delayed_ref(ref);
4226
		}
4227 4228 4229 4230 4231 4232 4233 4234
		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);
4235
		RB_CLEAR_NODE(&head->href_node);
4236 4237 4238
		spin_unlock(&head->lock);
		spin_unlock(&delayed_refs->lock);
		mutex_unlock(&head->mutex);
L
liubo 已提交
4239

4240
		if (pin_bytes)
4241 4242 4243
			btrfs_pin_extent(fs_info, head->bytenr,
					 head->num_bytes, 1);
		btrfs_put_delayed_ref_head(head);
L
liubo 已提交
4244 4245 4246 4247 4248 4249 4250 4251 4252
		cond_resched();
		spin_lock(&delayed_refs->lock);
	}

	spin_unlock(&delayed_refs->lock);

	return ret;
}

4253
static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root)
L
liubo 已提交
4254 4255 4256 4257 4258 4259
{
	struct btrfs_inode *btrfs_inode;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

4260 4261
	spin_lock(&root->delalloc_lock);
	list_splice_init(&root->delalloc_inodes, &splice);
L
liubo 已提交
4262 4263

	while (!list_empty(&splice)) {
4264 4265
		btrfs_inode = list_first_entry(&splice, struct btrfs_inode,
					       delalloc_inodes);
L
liubo 已提交
4266 4267

		list_del_init(&btrfs_inode->delalloc_inodes);
4268 4269
		clear_bit(BTRFS_INODE_IN_DELALLOC_LIST,
			  &btrfs_inode->runtime_flags);
4270
		spin_unlock(&root->delalloc_lock);
L
liubo 已提交
4271 4272

		btrfs_invalidate_inodes(btrfs_inode->root);
4273

4274
		spin_lock(&root->delalloc_lock);
L
liubo 已提交
4275 4276
	}

4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302
	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
4303 4304
}

4305
static int btrfs_destroy_marked_extents(struct btrfs_fs_info *fs_info,
L
liubo 已提交
4306 4307 4308 4309 4310 4311 4312 4313 4314 4315
					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,
4316
					    mark, NULL);
L
liubo 已提交
4317 4318 4319
		if (ret)
			break;

4320
		clear_extent_bits(dirty_pages, start, end, mark);
L
liubo 已提交
4321
		while (start <= end) {
4322 4323
			eb = find_extent_buffer(fs_info, start);
			start += fs_info->nodesize;
4324
			if (!eb)
L
liubo 已提交
4325
				continue;
4326
			wait_on_extent_buffer_writeback(eb);
L
liubo 已提交
4327

4328 4329 4330 4331
			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
4332 4333 4334 4335 4336 4337
		}
	}

	return ret;
}

4338
static int btrfs_destroy_pinned_extent(struct btrfs_fs_info *fs_info,
L
liubo 已提交
4339 4340 4341 4342 4343 4344
				       struct extent_io_tree *pinned_extents)
{
	struct extent_io_tree *unpin;
	u64 start;
	u64 end;
	int ret;
4345
	bool loop = true;
L
liubo 已提交
4346 4347

	unpin = pinned_extents;
4348
again:
L
liubo 已提交
4349 4350
	while (1) {
		ret = find_first_extent_bit(unpin, 0, &start, &end,
4351
					    EXTENT_DIRTY, NULL);
L
liubo 已提交
4352 4353 4354
		if (ret)
			break;

4355
		clear_extent_dirty(unpin, start, end);
4356
		btrfs_error_unpin_extent_range(fs_info, start, end);
L
liubo 已提交
4357 4358 4359
		cond_resched();
	}

4360
	if (loop) {
4361 4362
		if (unpin == &fs_info->freed_extents[0])
			unpin = &fs_info->freed_extents[1];
4363
		else
4364
			unpin = &fs_info->freed_extents[0];
4365 4366 4367 4368
		loop = false;
		goto again;
	}

L
liubo 已提交
4369 4370 4371
	return 0;
}

4372 4373 4374 4375 4376 4377 4378 4379 4380 4381 4382 4383 4384 4385 4386
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,
4387
			     struct btrfs_fs_info *fs_info)
4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410 4411 4412 4413 4414
{
	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);

4415 4416 4417 4418
	/*
	 * Refer to the definition of io_bgs member for details why it's safe
	 * to use it without any locking
	 */
4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431
	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);
	}
}

4432
void btrfs_cleanup_one_transaction(struct btrfs_transaction *cur_trans,
4433
				   struct btrfs_fs_info *fs_info)
4434
{
4435
	btrfs_cleanup_dirty_bgs(cur_trans, fs_info);
4436 4437 4438
	ASSERT(list_empty(&cur_trans->dirty_bgs));
	ASSERT(list_empty(&cur_trans->io_bgs));

4439
	btrfs_destroy_delayed_refs(cur_trans, fs_info);
4440

4441
	cur_trans->state = TRANS_STATE_COMMIT_START;
4442
	wake_up(&fs_info->transaction_blocked_wait);
4443

4444
	cur_trans->state = TRANS_STATE_UNBLOCKED;
4445
	wake_up(&fs_info->transaction_wait);
4446

4447 4448
	btrfs_destroy_delayed_inodes(fs_info);
	btrfs_assert_delayed_root_empty(fs_info);
4449

4450
	btrfs_destroy_marked_extents(fs_info, &cur_trans->dirty_pages,
4451
				     EXTENT_DIRTY);
4452
	btrfs_destroy_pinned_extent(fs_info,
4453
				    fs_info->pinned_extents);
4454

4455 4456
	cur_trans->state =TRANS_STATE_COMPLETED;
	wake_up(&cur_trans->commit_wait);
4457 4458
}

4459
static int btrfs_cleanup_transaction(struct btrfs_fs_info *fs_info)
L
liubo 已提交
4460 4461 4462
{
	struct btrfs_transaction *t;

4463
	mutex_lock(&fs_info->transaction_kthread_mutex);
L
liubo 已提交
4464

4465 4466 4467
	spin_lock(&fs_info->trans_lock);
	while (!list_empty(&fs_info->trans_list)) {
		t = list_first_entry(&fs_info->trans_list,
4468 4469
				     struct btrfs_transaction, list);
		if (t->state >= TRANS_STATE_COMMIT_START) {
4470
			refcount_inc(&t->use_count);
4471
			spin_unlock(&fs_info->trans_lock);
4472
			btrfs_wait_for_commit(fs_info, t->transid);
4473
			btrfs_put_transaction(t);
4474
			spin_lock(&fs_info->trans_lock);
4475 4476
			continue;
		}
4477
		if (t == fs_info->running_transaction) {
4478
			t->state = TRANS_STATE_COMMIT_DOING;
4479
			spin_unlock(&fs_info->trans_lock);
4480 4481 4482 4483 4484 4485 4486
			/*
			 * 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 {
4487
			spin_unlock(&fs_info->trans_lock);
4488
		}
4489
		btrfs_cleanup_one_transaction(t, fs_info);
4490

4491 4492 4493
		spin_lock(&fs_info->trans_lock);
		if (t == fs_info->running_transaction)
			fs_info->running_transaction = NULL;
L
liubo 已提交
4494
		list_del_init(&t->list);
4495
		spin_unlock(&fs_info->trans_lock);
L
liubo 已提交
4496

4497
		btrfs_put_transaction(t);
4498
		trace_btrfs_transaction_commit(fs_info->tree_root);
4499
		spin_lock(&fs_info->trans_lock);
4500
	}
4501 4502
	spin_unlock(&fs_info->trans_lock);
	btrfs_destroy_all_ordered_extents(fs_info);
4503 4504
	btrfs_destroy_delayed_inodes(fs_info);
	btrfs_assert_delayed_root_empty(fs_info);
4505
	btrfs_destroy_pinned_extent(fs_info, fs_info->pinned_extents);
4506 4507
	btrfs_destroy_all_delalloc_inodes(fs_info);
	mutex_unlock(&fs_info->transaction_kthread_mutex);
L
liubo 已提交
4508 4509 4510 4511

	return 0;
}

4512 4513 4514 4515 4516 4517
static struct btrfs_fs_info *btree_fs_info(void *private_data)
{
	struct inode *inode = private_data;
	return btrfs_sb(inode->i_sb);
}

4518
static const struct extent_io_ops btree_extent_io_ops = {
4519
	/* mandatory callbacks */
4520
	.submit_bio_hook = btree_submit_bio_hook,
4521
	.readpage_end_io_hook = btree_readpage_end_io_hook,
4522 4523
	/* note we're sharing with inode.c for the merge bio hook */
	.merge_bio_hook = btrfs_merge_bio_hook,
4524
	.readpage_io_failed_hook = btree_io_failed_hook,
4525 4526
	.set_range_writeback = btrfs_set_range_writeback,
	.tree_fs_info = btree_fs_info,
4527 4528

	/* optional callbacks */
4529
};