disk-io.c 125.5 KB
Newer Older
C
Chris Mason 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
/*
 * Copyright (C) 2007 Oracle.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License v2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

217 218
#endif

C
Chris Mason 已提交
219 220 221 222
/*
 * extents on the btree inode are pretty simple, there's one extent
 * that covers the entire device
 */
223
static struct extent_map *btree_get_extent(struct inode *inode,
224
		struct page *page, size_t pg_offset, u64 start, u64 len,
225
		int create)
226
{
227
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
228 229 230 231
	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
	struct extent_map *em;
	int ret;

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

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

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

265 266
out:
	return em;
267 268
}

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

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

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

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

319 320 321
	btrfs_csum_final(crc, result);

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

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

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

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

361 362 363
	if (atomic)
		return -EAGAIN;

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

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

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

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

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

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

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

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

	return ret;
}

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

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

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

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

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

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

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

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

	return ret;
497
}
498

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

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

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

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

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

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

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

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

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

static noinline int check_leaf(struct btrfs_root *root,
			       struct extent_buffer *leaf)
{
557
	struct btrfs_fs_info *fs_info = root->fs_info;
558 559 560 561 562
	struct btrfs_key key;
	struct btrfs_key leaf_key;
	u32 nritems = btrfs_header_nritems(leaf);
	int slot;

563 564 565 566 567 568 569
	if (nritems == 0) {
		struct btrfs_root *check_root;

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

570
		check_root = btrfs_get_fs_root(fs_info, &key, false);
571 572 573 574 575 576 577 578 579 580 581 582 583
		/*
		 * The only reason we also check NULL here is that during
		 * open_ctree() some roots has not yet been set up.
		 */
		if (!IS_ERR_OR_NULL(check_root)) {
			/* if leaf is the root, then it's fine */
			if (leaf->start !=
			    btrfs_root_bytenr(&check_root->root_item)) {
				CORRUPT("non-root leaf's nritems is 0",
					leaf, root, 0);
				return -EIO;
			}
		}
584
		return 0;
585
	}
586 587 588

	/* Check the 0 item */
	if (btrfs_item_offset_nr(leaf, 0) + btrfs_item_size_nr(leaf, 0) !=
589
	    BTRFS_LEAF_DATA_SIZE(fs_info)) {
590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623
		CORRUPT("invalid item offset size pair", leaf, root, 0);
		return -EIO;
	}

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

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

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

		/*
		 * Check to make sure that we don't point outside of the leaf,
624
		 * just in case all the items are consistent to each other, but
625 626 627
		 * all point outside of the leaf.
		 */
		if (btrfs_item_end_nr(leaf, slot) >
628
		    BTRFS_LEAF_DATA_SIZE(fs_info)) {
629 630 631 632 633 634 635 636
			CORRUPT("slot end outside of leaf", leaf, root, slot);
			return -EIO;
		}
	}

	return 0;
}

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

645
	if (nr == 0 || nr > BTRFS_NODEPTRS_PER_BLOCK(root->fs_info)) {
L
Liu Bo 已提交
646 647 648 649 650
		btrfs_crit(root->fs_info,
			   "corrupt node: block %llu root %llu nritems %lu",
			   node->start, root->objectid, nr);
		return -EIO;
	}
651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670

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

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

		if (btrfs_comp_cpu_keys(&key, &next_key) >= 0) {
			CORRUPT("bad key order", node, root, slot);
			ret = -EIO;
			goto out;
		}
	}
out:
	return ret;
L
Liu Bo 已提交
671 672
}

673 674 675
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)
676 677 678 679 680
{
	u64 found_start;
	int found_level;
	struct extent_buffer *eb;
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
681
	struct btrfs_fs_info *fs_info = root->fs_info;
682
	int ret = 0;
683
	int reads_done;
684 685 686

	if (!page->private)
		goto out;
C
Chris Mason 已提交
687

J
Josef Bacik 已提交
688
	eb = (struct extent_buffer *)page->private;
C
Chris Mason 已提交
689

690 691 692 693 694 695
	/* 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);
696 697
	if (!reads_done)
		goto err;
698

699
	eb->read_mirror = mirror;
700
	if (test_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags)) {
701 702 703 704
		ret = -EIO;
		goto err;
	}

705
	found_start = btrfs_header_bytenr(eb);
706
	if (found_start != eb->start) {
707 708
		btrfs_err_rl(fs_info, "bad tree block start %llu %llu",
			     found_start, eb->start);
709
		ret = -EIO;
710 711
		goto err;
	}
712 713 714
	if (check_tree_block_fsid(fs_info, eb)) {
		btrfs_err_rl(fs_info, "bad fsid on block %llu",
			     eb->start);
715 716 717
		ret = -EIO;
		goto err;
	}
718
	found_level = btrfs_header_level(eb);
719
	if (found_level >= BTRFS_MAX_LEVEL) {
720 721
		btrfs_err(fs_info, "bad tree block level %d",
			  (int)btrfs_header_level(eb));
722 723 724
		ret = -EIO;
		goto err;
	}
725

726 727
	btrfs_set_buffer_lockdep_class(btrfs_header_owner(eb),
				       eb, found_level);
728

729
	ret = csum_tree_block(fs_info, eb, 1);
730
	if (ret)
731 732 733 734 735 736 737 738 739 740 741
		goto err;

	/*
	 * If this is a leaf block and it is corrupt, set the corrupt bit so
	 * that we don't try and read the other copies of this block, just
	 * return -EIO.
	 */
	if (found_level == 0 && check_leaf(root, eb)) {
		set_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
		ret = -EIO;
	}
742

L
Liu Bo 已提交
743 744 745
	if (found_level > 0 && check_node(root, eb))
		ret = -EIO;

746 747
	if (!ret)
		set_extent_buffer_uptodate(eb);
748
err:
749 750
	if (reads_done &&
	    test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
751
		btree_readahead_hook(fs_info, eb, ret);
A
Arne Jansen 已提交
752

D
David Woodhouse 已提交
753 754 755 756 757 758 759
	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);
760
		clear_extent_buffer_uptodate(eb);
D
David Woodhouse 已提交
761
	}
762
	free_extent_buffer(eb);
763
out:
764
	return ret;
765 766
}

767
static int btree_io_failed_hook(struct page *page, int failed_mirror)
A
Arne Jansen 已提交
768 769 770
{
	struct extent_buffer *eb;

J
Josef Bacik 已提交
771
	eb = (struct extent_buffer *)page->private;
772
	set_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags);
773
	eb->read_mirror = failed_mirror;
D
David Woodhouse 已提交
774
	atomic_dec(&eb->io_pages);
775
	if (test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
776
		btree_readahead_hook(eb->fs_info, eb, -EIO);
A
Arne Jansen 已提交
777 778 779
	return -EIO;	/* we fixed nothing */
}

780
static void end_workqueue_bio(struct bio *bio)
781
{
782
	struct btrfs_end_io_wq *end_io_wq = bio->bi_private;
783
	struct btrfs_fs_info *fs_info;
784 785
	struct btrfs_workqueue *wq;
	btrfs_work_func_t func;
786 787

	fs_info = end_io_wq->info;
788
	end_io_wq->error = bio->bi_error;
789

M
Mike Christie 已提交
790
	if (bio_op(bio) == REQ_OP_WRITE) {
791 792 793 794 795 796 797 798 799 800 801 802 803
		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;
		}
804
	} else {
805 806 807 808 809
		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) {
810 811 812 813 814 815 816 817 818
			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;
		}
819
	}
820 821 822

	btrfs_init_work(&end_io_wq->work, func, end_workqueue_fn, NULL, NULL);
	btrfs_queue_work(wq, &end_io_wq->work);
823 824
}

825
int btrfs_bio_wq_end_io(struct btrfs_fs_info *info, struct bio *bio,
826
			enum btrfs_wq_endio_type metadata)
827
{
828
	struct btrfs_end_io_wq *end_io_wq;
829

830
	end_io_wq = kmem_cache_alloc(btrfs_end_io_wq_cache, GFP_NOFS);
831 832 833 834 835
	if (!end_io_wq)
		return -ENOMEM;

	end_io_wq->private = bio->bi_private;
	end_io_wq->end_io = bio->bi_end_io;
836
	end_io_wq->info = info;
837 838
	end_io_wq->error = 0;
	end_io_wq->bio = bio;
839
	end_io_wq->metadata = metadata;
840 841 842

	bio->bi_private = end_io_wq;
	bio->bi_end_io = end_workqueue_bio;
843 844 845
	return 0;
}

846
unsigned long btrfs_async_submit_limit(struct btrfs_fs_info *info)
847
{
848
	unsigned long limit = min_t(unsigned long,
849
				    info->thread_pool_size,
850 851 852
				    info->fs_devices->open_devices);
	return 256 * limit;
}
853

C
Chris Mason 已提交
854 855 856
static void run_one_async_start(struct btrfs_work *work)
{
	struct async_submit_bio *async;
857
	int ret;
C
Chris Mason 已提交
858 859

	async = container_of(work, struct  async_submit_bio, work);
860
	ret = async->submit_bio_start(async->inode, async->bio,
861 862 863 864
				      async->mirror_num, async->bio_flags,
				      async->bio_offset);
	if (ret)
		async->error = ret;
C
Chris Mason 已提交
865 866 867
}

static void run_one_async_done(struct btrfs_work *work)
868 869 870
{
	struct btrfs_fs_info *fs_info;
	struct async_submit_bio *async;
871
	int limit;
872 873 874

	async = container_of(work, struct  async_submit_bio, work);
	fs_info = BTRFS_I(async->inode)->root->fs_info;
875

876
	limit = btrfs_async_submit_limit(fs_info);
877 878
	limit = limit * 2 / 3;

879 880 881
	/*
	 * atomic_dec_return implies a barrier for waitqueue_active
	 */
882
	if (atomic_dec_return(&fs_info->nr_async_submits) < limit &&
883
	    waitqueue_active(&fs_info->async_submit_wait))
884 885
		wake_up(&fs_info->async_submit_wait);

886
	/* If an error occurred we just want to clean up the bio and move on */
887
	if (async->error) {
888 889
		async->bio->bi_error = async->error;
		bio_endio(async->bio);
890 891 892
		return;
	}

893 894
	async->submit_bio_done(async->inode, async->bio, async->mirror_num,
			       async->bio_flags, async->bio_offset);
C
Chris Mason 已提交
895 896 897 898 899 900 901
}

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

	async = container_of(work, struct  async_submit_bio, work);
902 903 904
	kfree(async);
}

905
int btrfs_wq_submit_bio(struct btrfs_fs_info *fs_info, struct inode *inode,
906
			struct bio *bio, int mirror_num,
C
Chris Mason 已提交
907
			unsigned long bio_flags,
908
			u64 bio_offset,
C
Chris Mason 已提交
909 910
			extent_submit_bio_hook_t *submit_bio_start,
			extent_submit_bio_hook_t *submit_bio_done)
911 912 913 914 915 916 917 918 919 920
{
	struct async_submit_bio *async;

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

	async->inode = inode;
	async->bio = bio;
	async->mirror_num = mirror_num;
C
Chris Mason 已提交
921 922 923
	async->submit_bio_start = submit_bio_start;
	async->submit_bio_done = submit_bio_done;

924
	btrfs_init_work(&async->work, btrfs_worker_helper, run_one_async_start,
925
			run_one_async_done, run_one_async_free);
C
Chris Mason 已提交
926

C
Chris Mason 已提交
927
	async->bio_flags = bio_flags;
928
	async->bio_offset = bio_offset;
929

930 931
	async->error = 0;

932
	atomic_inc(&fs_info->nr_async_submits);
933

J
Jens Axboe 已提交
934
	if (bio->bi_opf & REQ_SYNC)
935
		btrfs_set_work_high_priority(&async->work);
936

937
	btrfs_queue_work(fs_info->workers, &async->work);
938

C
Chris Mason 已提交
939
	while (atomic_read(&fs_info->async_submit_draining) &&
940 941 942 943 944
	      atomic_read(&fs_info->nr_async_submits)) {
		wait_event(fs_info->async_submit_wait,
			   (atomic_read(&fs_info->nr_async_submits) == 0));
	}

945 946 947
	return 0;
}

948 949
static int btree_csum_one_bio(struct bio *bio)
{
950
	struct bio_vec *bvec;
951
	struct btrfs_root *root;
952
	int i, ret = 0;
953

954
	bio_for_each_segment_all(bvec, bio, i) {
955
		root = BTRFS_I(bvec->bv_page->mapping->host)->root;
956
		ret = csum_dirty_buffer(root->fs_info, bvec->bv_page);
957 958
		if (ret)
			break;
959
	}
960

961
	return ret;
962 963
}

964 965
static int __btree_submit_bio_start(struct inode *inode, struct bio *bio,
				    int mirror_num, unsigned long bio_flags,
966
				    u64 bio_offset)
967
{
968 969
	/*
	 * when we're called for a write, we're already in the async
970
	 * submission context.  Just jump into btrfs_map_bio
971
	 */
972
	return btree_csum_one_bio(bio);
C
Chris Mason 已提交
973
}
974

975
static int __btree_submit_bio_done(struct inode *inode, struct bio *bio,
976 977
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
C
Chris Mason 已提交
978
{
979 980
	int ret;

981
	/*
C
Chris Mason 已提交
982 983
	 * when we're called for a write, we're already in the async
	 * submission context.  Just jump into btrfs_map_bio
984
	 */
985
	ret = btrfs_map_bio(btrfs_sb(inode->i_sb), bio, mirror_num, 1);
986 987 988 989
	if (ret) {
		bio->bi_error = ret;
		bio_endio(bio);
	}
990
	return ret;
991 992
}

993 994 995 996 997
static int check_async_write(struct inode *inode, unsigned long bio_flags)
{
	if (bio_flags & EXTENT_BIO_TREE_LOG)
		return 0;
#ifdef CONFIG_X86
998
	if (static_cpu_has(X86_FEATURE_XMM4_2))
999 1000 1001 1002 1003
		return 0;
#endif
	return 1;
}

1004
static int btree_submit_bio_hook(struct inode *inode, struct bio *bio,
1005 1006
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
1007
{
1008
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
1009
	int async = check_async_write(inode, bio_flags);
1010 1011
	int ret;

M
Mike Christie 已提交
1012
	if (bio_op(bio) != REQ_OP_WRITE) {
C
Chris Mason 已提交
1013 1014 1015 1016
		/*
		 * called for a read, do the setup so that checksum validation
		 * can happen in the async kernel threads
		 */
1017 1018
		ret = btrfs_bio_wq_end_io(fs_info, bio,
					  BTRFS_WQ_ENDIO_METADATA);
1019
		if (ret)
1020
			goto out_w_error;
1021
		ret = btrfs_map_bio(fs_info, bio, mirror_num, 0);
1022 1023 1024
	} else if (!async) {
		ret = btree_csum_one_bio(bio);
		if (ret)
1025
			goto out_w_error;
1026
		ret = btrfs_map_bio(fs_info, bio, mirror_num, 0);
1027 1028 1029 1030 1031
	} else {
		/*
		 * kthread helpers are used to submit writes so that
		 * checksumming can happen in parallel across all CPUs
		 */
1032
		ret = btrfs_wq_submit_bio(fs_info, inode, bio, mirror_num, 0,
1033 1034 1035
					  bio_offset,
					  __btree_submit_bio_start,
					  __btree_submit_bio_done);
1036
	}
1037

1038 1039 1040 1041
	if (ret)
		goto out_w_error;
	return 0;

1042
out_w_error:
1043 1044
	bio->bi_error = ret;
	bio_endio(bio);
1045
	return ret;
1046 1047
}

J
Jan Beulich 已提交
1048
#ifdef CONFIG_MIGRATION
1049
static int btree_migratepage(struct address_space *mapping,
1050 1051
			struct page *newpage, struct page *page,
			enum migrate_mode mode)
1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065
{
	/*
	 * 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;
1066
	return migrate_page(mapping, newpage, page, mode);
1067
}
J
Jan Beulich 已提交
1068
#endif
1069

1070 1071 1072 1073

static int btree_writepages(struct address_space *mapping,
			    struct writeback_control *wbc)
{
1074 1075 1076
	struct btrfs_fs_info *fs_info;
	int ret;

1077
	if (wbc->sync_mode == WB_SYNC_NONE) {
1078 1079 1080 1081

		if (wbc->for_kupdate)
			return 0;

1082
		fs_info = BTRFS_I(mapping->host)->root->fs_info;
1083
		/* this is a bit racy, but that's ok */
1084 1085 1086
		ret = percpu_counter_compare(&fs_info->dirty_metadata_bytes,
					     BTRFS_DIRTY_METADATA_THRESH);
		if (ret < 0)
1087 1088
			return 0;
	}
1089
	return btree_write_cache_pages(mapping, wbc);
1090 1091
}

1092
static int btree_readpage(struct file *file, struct page *page)
1093
{
1094 1095
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
1096
	return extent_read_full_page(tree, page, btree_get_extent, 0);
1097
}
C
Chris Mason 已提交
1098

1099
static int btree_releasepage(struct page *page, gfp_t gfp_flags)
1100
{
1101
	if (PageWriteback(page) || PageDirty(page))
C
Chris Mason 已提交
1102
		return 0;
1103

1104
	return try_release_extent_buffer(page);
1105 1106
}

1107 1108
static void btree_invalidatepage(struct page *page, unsigned int offset,
				 unsigned int length)
1109
{
1110 1111
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
1112 1113
	extent_invalidatepage(tree, page, offset);
	btree_releasepage(page, GFP_NOFS);
1114
	if (PagePrivate(page)) {
1115 1116 1117
		btrfs_warn(BTRFS_I(page->mapping->host)->root->fs_info,
			   "page private not zero on page %llu",
			   (unsigned long long)page_offset(page));
1118 1119
		ClearPagePrivate(page);
		set_page_private(page, 0);
1120
		put_page(page);
1121
	}
1122 1123
}

1124 1125
static int btree_set_page_dirty(struct page *page)
{
1126
#ifdef DEBUG
1127 1128 1129 1130 1131 1132 1133 1134
	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);
1135
#endif
1136 1137 1138
	return __set_page_dirty_nobuffers(page);
}

1139
static const struct address_space_operations btree_aops = {
1140
	.readpage	= btree_readpage,
1141
	.writepages	= btree_writepages,
1142 1143
	.releasepage	= btree_releasepage,
	.invalidatepage = btree_invalidatepage,
1144
#ifdef CONFIG_MIGRATION
1145
	.migratepage	= btree_migratepage,
1146
#endif
1147
	.set_page_dirty = btree_set_page_dirty,
1148 1149
};

1150
void readahead_tree_block(struct btrfs_fs_info *fs_info, u64 bytenr)
C
Chris Mason 已提交
1151
{
1152
	struct extent_buffer *buf = NULL;
1153
	struct inode *btree_inode = fs_info->btree_inode;
C
Chris Mason 已提交
1154

1155
	buf = btrfs_find_create_tree_block(fs_info, bytenr);
1156
	if (IS_ERR(buf))
1157
		return;
1158
	read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree,
1159
				 buf, WAIT_NONE, btree_get_extent, 0);
1160
	free_extent_buffer(buf);
C
Chris Mason 已提交
1161 1162
}

1163
int reada_tree_block_flagged(struct btrfs_fs_info *fs_info, u64 bytenr,
1164 1165 1166
			 int mirror_num, struct extent_buffer **eb)
{
	struct extent_buffer *buf = NULL;
1167
	struct inode *btree_inode = fs_info->btree_inode;
1168 1169 1170
	struct extent_io_tree *io_tree = &BTRFS_I(btree_inode)->io_tree;
	int ret;

1171
	buf = btrfs_find_create_tree_block(fs_info, bytenr);
1172
	if (IS_ERR(buf))
1173 1174 1175 1176
		return 0;

	set_bit(EXTENT_BUFFER_READAHEAD, &buf->bflags);

1177
	ret = read_extent_buffer_pages(io_tree, buf, WAIT_PAGE_LOCK,
1178 1179 1180 1181 1182 1183 1184 1185 1186
				       btree_get_extent, mirror_num);
	if (ret) {
		free_extent_buffer(buf);
		return ret;
	}

	if (test_bit(EXTENT_BUFFER_CORRUPT, &buf->bflags)) {
		free_extent_buffer(buf);
		return -EIO;
1187
	} else if (extent_buffer_uptodate(buf)) {
1188 1189 1190 1191 1192 1193 1194
		*eb = buf;
	} else {
		free_extent_buffer(buf);
	}
	return 0;
}

1195 1196 1197
struct extent_buffer *btrfs_find_create_tree_block(
						struct btrfs_fs_info *fs_info,
						u64 bytenr)
1198
{
1199 1200 1201
	if (btrfs_is_testing(fs_info))
		return alloc_test_extent_buffer(fs_info, bytenr);
	return alloc_extent_buffer(fs_info, bytenr);
1202 1203 1204
}


1205 1206
int btrfs_write_tree_block(struct extent_buffer *buf)
{
1207
	return filemap_fdatawrite_range(buf->pages[0]->mapping, buf->start,
1208
					buf->start + buf->len - 1);
1209 1210 1211 1212
}

int btrfs_wait_tree_block_writeback(struct extent_buffer *buf)
{
1213
	return filemap_fdatawait_range(buf->pages[0]->mapping,
1214
				       buf->start, buf->start + buf->len - 1);
1215 1216
}

1217
struct extent_buffer *read_tree_block(struct btrfs_fs_info *fs_info, u64 bytenr,
1218
				      u64 parent_transid)
1219 1220 1221 1222
{
	struct extent_buffer *buf = NULL;
	int ret;

1223
	buf = btrfs_find_create_tree_block(fs_info, bytenr);
1224 1225
	if (IS_ERR(buf))
		return buf;
1226

1227
	ret = btree_read_extent_buffer_pages(fs_info, buf, parent_transid);
1228 1229
	if (ret) {
		free_extent_buffer(buf);
1230
		return ERR_PTR(ret);
1231
	}
1232
	return buf;
1233

1234 1235
}

1236 1237
void clean_tree_block(struct btrfs_trans_handle *trans,
		      struct btrfs_fs_info *fs_info,
1238
		      struct extent_buffer *buf)
1239
{
1240
	if (btrfs_header_generation(buf) ==
1241
	    fs_info->running_transaction->transid) {
1242
		btrfs_assert_tree_locked(buf);
1243

1244
		if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &buf->bflags)) {
1245 1246 1247
			__percpu_counter_add(&fs_info->dirty_metadata_bytes,
					     -buf->len,
					     fs_info->dirty_metadata_batch);
1248 1249 1250 1251
			/* ugh, clear_extent_buffer_dirty needs to lock the page */
			btrfs_set_lock_blocking(buf);
			clear_extent_buffer_dirty(buf);
		}
1252
	}
1253 1254
}

1255 1256 1257 1258 1259 1260 1261 1262 1263
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);

1264
	ret = percpu_counter_init(&writers->counter, 0, GFP_KERNEL);
1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280
	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);
}

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

1290 1291
	root->objectid = objectid;
	root->last_trans = 0;
1292
	root->highest_objectid = 0;
1293
	root->nr_delalloc_inodes = 0;
1294
	root->nr_ordered_extents = 0;
1295
	root->name = NULL;
1296
	root->inode_tree = RB_ROOT;
1297
	INIT_RADIX_TREE(&root->delayed_nodes_tree, GFP_ATOMIC);
1298
	root->block_rsv = NULL;
1299
	root->orphan_block_rsv = NULL;
1300 1301

	INIT_LIST_HEAD(&root->dirty_list);
1302
	INIT_LIST_HEAD(&root->root_list);
1303 1304
	INIT_LIST_HEAD(&root->delalloc_inodes);
	INIT_LIST_HEAD(&root->delalloc_root);
1305 1306
	INIT_LIST_HEAD(&root->ordered_extents);
	INIT_LIST_HEAD(&root->ordered_root);
1307 1308
	INIT_LIST_HEAD(&root->logged_list[0]);
	INIT_LIST_HEAD(&root->logged_list[1]);
1309
	spin_lock_init(&root->orphan_lock);
1310
	spin_lock_init(&root->inode_lock);
1311
	spin_lock_init(&root->delalloc_lock);
1312
	spin_lock_init(&root->ordered_extent_lock);
1313
	spin_lock_init(&root->accounting_lock);
1314 1315
	spin_lock_init(&root->log_extents_lock[0]);
	spin_lock_init(&root->log_extents_lock[1]);
1316
	mutex_init(&root->objectid_mutex);
1317
	mutex_init(&root->log_mutex);
1318
	mutex_init(&root->ordered_extent_mutex);
1319
	mutex_init(&root->delalloc_mutex);
Y
Yan Zheng 已提交
1320 1321 1322
	init_waitqueue_head(&root->log_writer_wait);
	init_waitqueue_head(&root->log_commit_wait[0]);
	init_waitqueue_head(&root->log_commit_wait[1]);
1323 1324
	INIT_LIST_HEAD(&root->log_ctxs[0]);
	INIT_LIST_HEAD(&root->log_ctxs[1]);
Y
Yan Zheng 已提交
1325 1326 1327
	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
1328
	atomic_set(&root->log_batch, 0);
1329
	atomic_set(&root->orphan_inodes, 0);
1330
	atomic_set(&root->refs, 1);
1331
	atomic_set(&root->will_be_snapshoted, 0);
1332
	atomic_set(&root->qgroup_meta_rsv, 0);
Y
Yan Zheng 已提交
1333
	root->log_transid = 0;
1334
	root->log_transid_committed = -1;
1335
	root->last_log_commit = 0;
1336
	if (!dummy)
1337 1338
		extent_io_tree_init(&root->dirty_log_pages,
				     fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
1339

1340 1341
	memset(&root->root_key, 0, sizeof(root->root_key));
	memset(&root->root_item, 0, sizeof(root->root_item));
1342
	memset(&root->defrag_progress, 0, sizeof(root->defrag_progress));
1343
	if (!dummy)
1344 1345 1346
		root->defrag_trans_start = fs_info->generation;
	else
		root->defrag_trans_start = 0;
1347
	root->root_key.objectid = objectid;
1348
	root->anon_dev = 0;
1349

1350
	spin_lock_init(&root->root_item_lock);
1351 1352
}

1353 1354
static struct btrfs_root *btrfs_alloc_root(struct btrfs_fs_info *fs_info,
		gfp_t flags)
A
Al Viro 已提交
1355
{
1356
	struct btrfs_root *root = kzalloc(sizeof(*root), flags);
A
Al Viro 已提交
1357 1358 1359 1360 1361
	if (root)
		root->fs_info = fs_info;
	return root;
}

1362 1363
#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
/* Should only be used by the testing infrastructure */
1364
struct btrfs_root *btrfs_alloc_dummy_root(struct btrfs_fs_info *fs_info)
1365 1366 1367
{
	struct btrfs_root *root;

1368 1369 1370 1371
	if (!fs_info)
		return ERR_PTR(-EINVAL);

	root = btrfs_alloc_root(fs_info, GFP_KERNEL);
1372 1373
	if (!root)
		return ERR_PTR(-ENOMEM);
1374

1375
	/* We don't use the stripesize in selftest, set it as sectorsize */
1376
	__setup_root(root, fs_info, BTRFS_ROOT_TREE_OBJECTID);
1377
	root->alloc_bytenr = 0;
1378 1379 1380 1381 1382

	return root;
}
#endif

1383 1384 1385 1386 1387 1388 1389 1390 1391
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;
1392
	uuid_le uuid;
1393

1394
	root = btrfs_alloc_root(fs_info, GFP_KERNEL);
1395 1396 1397
	if (!root)
		return ERR_PTR(-ENOMEM);

1398
	__setup_root(root, fs_info, objectid);
1399 1400 1401 1402
	root->root_key.objectid = objectid;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = 0;

1403
	leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0);
1404 1405
	if (IS_ERR(leaf)) {
		ret = PTR_ERR(leaf);
1406
		leaf = NULL;
1407 1408 1409
		goto fail;
	}

1410
	memzero_extent_buffer(leaf, 0, sizeof(struct btrfs_header));
1411 1412 1413 1414 1415 1416
	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;

1417 1418
	write_extent_buffer_fsid(leaf, fs_info->fsid);
	write_extent_buffer_chunk_tree_uuid(leaf, fs_info->chunk_tree_uuid);
1419 1420 1421
	btrfs_mark_buffer_dirty(leaf);

	root->commit_root = btrfs_root_node(root);
1422
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
1423 1424 1425 1426 1427 1428 1429 1430 1431 1432

	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);
1433 1434
	uuid_le_gen(&uuid);
	memcpy(root->root_item.uuid, uuid.b, BTRFS_UUID_SIZE);
1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445
	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);

1446 1447
	return root;

1448
fail:
1449 1450
	if (leaf) {
		btrfs_tree_unlock(leaf);
1451
		free_extent_buffer(root->commit_root);
1452 1453 1454
		free_extent_buffer(leaf);
	}
	kfree(root);
1455

1456
	return ERR_PTR(ret);
1457 1458
}

Y
Yan Zheng 已提交
1459 1460
static struct btrfs_root *alloc_log_tree(struct btrfs_trans_handle *trans,
					 struct btrfs_fs_info *fs_info)
1461 1462
{
	struct btrfs_root *root;
Y
Yan Zheng 已提交
1463
	struct extent_buffer *leaf;
1464

1465
	root = btrfs_alloc_root(fs_info, GFP_NOFS);
1466
	if (!root)
Y
Yan Zheng 已提交
1467
		return ERR_PTR(-ENOMEM);
1468

1469
	__setup_root(root, fs_info, BTRFS_TREE_LOG_OBJECTID);
1470 1471 1472 1473

	root->root_key.objectid = BTRFS_TREE_LOG_OBJECTID;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = BTRFS_TREE_LOG_OBJECTID;
1474

Y
Yan Zheng 已提交
1475
	/*
1476 1477
	 * DON'T set REF_COWS for log trees
	 *
Y
Yan Zheng 已提交
1478 1479 1480 1481 1482
	 * 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).
	 */
1483

1484 1485
	leaf = btrfs_alloc_tree_block(trans, root, 0, BTRFS_TREE_LOG_OBJECTID,
			NULL, 0, 0, 0);
Y
Yan Zheng 已提交
1486 1487 1488 1489
	if (IS_ERR(leaf)) {
		kfree(root);
		return ERR_CAST(leaf);
	}
1490

1491
	memzero_extent_buffer(leaf, 0, sizeof(struct btrfs_header));
1492 1493 1494 1495
	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
1496
	root->node = leaf;
1497

1498
	write_extent_buffer_fsid(root->node, fs_info->fsid);
1499 1500
	btrfs_mark_buffer_dirty(root->node);
	btrfs_tree_unlock(root->node);
Y
Yan Zheng 已提交
1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519
	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)
{
1520
	struct btrfs_fs_info *fs_info = root->fs_info;
Y
Yan Zheng 已提交
1521 1522 1523
	struct btrfs_root *log_root;
	struct btrfs_inode_item *inode_item;

1524
	log_root = alloc_log_tree(trans, fs_info);
Y
Yan Zheng 已提交
1525 1526 1527 1528 1529 1530 1531
	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;
1532 1533 1534
	btrfs_set_stack_inode_generation(inode_item, 1);
	btrfs_set_stack_inode_size(inode_item, 3);
	btrfs_set_stack_inode_nlink(inode_item, 1);
1535
	btrfs_set_stack_inode_nbytes(inode_item,
1536
				     fs_info->nodesize);
1537
	btrfs_set_stack_inode_mode(inode_item, S_IFDIR | 0755);
Y
Yan Zheng 已提交
1538

1539
	btrfs_set_root_node(&log_root->root_item, log_root->node);
Y
Yan Zheng 已提交
1540 1541 1542 1543

	WARN_ON(root->log_root);
	root->log_root = log_root;
	root->log_transid = 0;
1544
	root->log_transid_committed = -1;
1545
	root->last_log_commit = 0;
1546 1547 1548
	return 0;
}

1549 1550
static struct btrfs_root *btrfs_read_tree_root(struct btrfs_root *tree_root,
					       struct btrfs_key *key)
1551 1552 1553
{
	struct btrfs_root *root;
	struct btrfs_fs_info *fs_info = tree_root->fs_info;
1554
	struct btrfs_path *path;
1555
	u64 generation;
1556
	int ret;
1557

1558 1559
	path = btrfs_alloc_path();
	if (!path)
1560
		return ERR_PTR(-ENOMEM);
1561

1562
	root = btrfs_alloc_root(fs_info, GFP_NOFS);
1563 1564 1565
	if (!root) {
		ret = -ENOMEM;
		goto alloc_fail;
1566 1567
	}

1568
	__setup_root(root, fs_info, key->objectid);
1569

1570 1571
	ret = btrfs_find_root(tree_root, key, path,
			      &root->root_item, &root->root_key);
1572
	if (ret) {
1573 1574
		if (ret > 0)
			ret = -ENOENT;
1575
		goto find_fail;
1576
	}
1577

1578
	generation = btrfs_root_generation(&root->root_item);
1579 1580
	root->node = read_tree_block(fs_info,
				     btrfs_root_bytenr(&root->root_item),
1581
				     generation);
1582 1583
	if (IS_ERR(root->node)) {
		ret = PTR_ERR(root->node);
1584 1585 1586
		goto find_fail;
	} else if (!btrfs_buffer_uptodate(root->node, generation, 0)) {
		ret = -EIO;
1587 1588
		free_extent_buffer(root->node);
		goto find_fail;
1589
	}
1590
	root->commit_root = btrfs_root_node(root);
1591
out:
1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611
	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) {
1612
		set_bit(BTRFS_ROOT_REF_COWS, &root->state);
1613 1614
		btrfs_check_and_init_root_item(&root->root_item);
	}
1615

1616 1617 1618
	return root;
}

1619 1620 1621
int btrfs_init_fs_root(struct btrfs_root *root)
{
	int ret;
1622
	struct btrfs_subvolume_writers *writers;
1623 1624 1625 1626 1627 1628 1629 1630 1631

	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;
	}

1632 1633 1634 1635 1636 1637 1638
	writers = btrfs_alloc_subvolume_writers();
	if (IS_ERR(writers)) {
		ret = PTR_ERR(writers);
		goto fail;
	}
	root->subv_writers = writers;

1639
	btrfs_init_free_ino_ctl(root);
1640 1641
	spin_lock_init(&root->ino_cache_lock);
	init_waitqueue_head(&root->ino_cache_wait);
1642 1643 1644

	ret = get_anon_bdev(&root->anon_dev);
	if (ret)
L
Liu Bo 已提交
1645
		goto fail;
1646 1647 1648 1649 1650 1651

	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
1652
		goto fail;
1653 1654 1655 1656 1657 1658
	}

	ASSERT(root->highest_objectid <= BTRFS_LAST_FREE_OBJECTID);

	mutex_unlock(&root->objectid_mutex);

1659 1660
	return 0;
fail:
L
Liu Bo 已提交
1661
	/* the caller is responsible to call free_fs_root */
1662 1663 1664
	return ret;
}

1665 1666
struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info,
					u64 root_id)
1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681
{
	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;

1682
	ret = radix_tree_preload(GFP_NOFS);
1683 1684 1685 1686 1687 1688 1689 1690
	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)
1691
		set_bit(BTRFS_ROOT_IN_RADIX, &root->state);
1692 1693 1694 1695 1696 1697
	spin_unlock(&fs_info->fs_roots_radix_lock);
	radix_tree_preload_end();

	return ret;
}

1698 1699 1700
struct btrfs_root *btrfs_get_fs_root(struct btrfs_fs_info *fs_info,
				     struct btrfs_key *location,
				     bool check_ref)
1701 1702
{
	struct btrfs_root *root;
1703
	struct btrfs_path *path;
1704
	struct btrfs_key key;
1705 1706
	int ret;

1707 1708 1709 1710
	if (location->objectid == BTRFS_ROOT_TREE_OBJECTID)
		return fs_info->tree_root;
	if (location->objectid == BTRFS_EXTENT_TREE_OBJECTID)
		return fs_info->extent_root;
1711 1712 1713 1714
	if (location->objectid == BTRFS_CHUNK_TREE_OBJECTID)
		return fs_info->chunk_root;
	if (location->objectid == BTRFS_DEV_TREE_OBJECTID)
		return fs_info->dev_root;
1715 1716
	if (location->objectid == BTRFS_CSUM_TREE_OBJECTID)
		return fs_info->csum_root;
1717 1718 1719
	if (location->objectid == BTRFS_QUOTA_TREE_OBJECTID)
		return fs_info->quota_root ? fs_info->quota_root :
					     ERR_PTR(-ENOENT);
1720 1721 1722
	if (location->objectid == BTRFS_UUID_TREE_OBJECTID)
		return fs_info->uuid_root ? fs_info->uuid_root :
					    ERR_PTR(-ENOENT);
1723 1724 1725
	if (location->objectid == BTRFS_FREE_SPACE_TREE_OBJECTID)
		return fs_info->free_space_root ? fs_info->free_space_root :
						  ERR_PTR(-ENOENT);
1726
again:
1727
	root = btrfs_lookup_fs_root(fs_info, location->objectid);
1728
	if (root) {
1729
		if (check_ref && btrfs_root_refs(&root->root_item) == 0)
1730
			return ERR_PTR(-ENOENT);
1731
		return root;
1732
	}
1733

1734
	root = btrfs_read_fs_root(fs_info->tree_root, location);
1735 1736
	if (IS_ERR(root))
		return root;
1737

1738
	if (check_ref && btrfs_root_refs(&root->root_item) == 0) {
1739
		ret = -ENOENT;
1740
		goto fail;
1741
	}
1742

1743
	ret = btrfs_init_fs_root(root);
1744 1745
	if (ret)
		goto fail;
1746

1747 1748 1749 1750 1751
	path = btrfs_alloc_path();
	if (!path) {
		ret = -ENOMEM;
		goto fail;
	}
1752 1753 1754 1755 1756
	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);
1757
	btrfs_free_path(path);
1758 1759 1760
	if (ret < 0)
		goto fail;
	if (ret == 0)
1761
		set_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &root->state);
1762

1763
	ret = btrfs_insert_fs_root(fs_info, root);
1764
	if (ret) {
1765 1766 1767 1768 1769
		if (ret == -EEXIST) {
			free_fs_root(root);
			goto again;
		}
		goto fail;
1770
	}
1771
	return root;
1772 1773 1774
fail:
	free_fs_root(root);
	return ERR_PTR(ret);
1775 1776
}

C
Chris Mason 已提交
1777 1778 1779 1780 1781 1782
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
1783

1784 1785
	rcu_read_lock();
	list_for_each_entry_rcu(device, &info->fs_devices->devices, dev_list) {
1786 1787
		if (!device->bdev)
			continue;
C
Chris Mason 已提交
1788
		bdi = blk_get_backing_dev_info(device->bdev);
1789
		if (bdi_congested(bdi, bdi_bits)) {
C
Chris Mason 已提交
1790 1791 1792 1793
			ret = 1;
			break;
		}
	}
1794
	rcu_read_unlock();
C
Chris Mason 已提交
1795 1796 1797 1798 1799
	return ret;
}

static int setup_bdi(struct btrfs_fs_info *info, struct backing_dev_info *bdi)
{
1800 1801
	int err;

1802
	err = bdi_setup_and_register(bdi, "btrfs");
1803 1804 1805
	if (err)
		return err;

1806
	bdi->ra_pages = VM_MAX_READAHEAD * 1024 / PAGE_SIZE;
C
Chris Mason 已提交
1807 1808
	bdi->congested_fn	= btrfs_congested_fn;
	bdi->congested_data	= info;
1809
	bdi->capabilities |= BDI_CAP_CGROUP_WRITEBACK;
C
Chris Mason 已提交
1810 1811 1812
	return 0;
}

1813 1814 1815 1816 1817
/*
 * 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)
1818 1819
{
	struct bio *bio;
1820
	struct btrfs_end_io_wq *end_io_wq;
1821

1822
	end_io_wq = container_of(work, struct btrfs_end_io_wq, work);
1823
	bio = end_io_wq->bio;
1824

1825
	bio->bi_error = end_io_wq->error;
1826 1827
	bio->bi_private = end_io_wq->private;
	bio->bi_end_io = end_io_wq->end_io;
1828
	kmem_cache_free(btrfs_end_io_wq_cache, end_io_wq);
1829
	bio_endio(bio);
1830 1831
}

1832 1833 1834
static int cleaner_kthread(void *arg)
{
	struct btrfs_root *root = arg;
1835
	struct btrfs_fs_info *fs_info = root->fs_info;
1836
	int again;
1837
	struct btrfs_trans_handle *trans;
1838 1839

	do {
1840
		again = 0;
1841

1842
		/* Make the cleaner go to sleep early. */
1843
		if (btrfs_need_cleaner_sleep(fs_info))
1844 1845
			goto sleep;

1846 1847 1848 1849
		/*
		 * Do not do anything if we might cause open_ctree() to block
		 * before we have finished mounting the filesystem.
		 */
1850
		if (!test_bit(BTRFS_FS_OPEN, &fs_info->flags))
1851 1852
			goto sleep;

1853
		if (!mutex_trylock(&fs_info->cleaner_mutex))
1854 1855
			goto sleep;

1856 1857 1858 1859
		/*
		 * Avoid the problem that we change the status of the fs
		 * during the above check and trylock.
		 */
1860
		if (btrfs_need_cleaner_sleep(fs_info)) {
1861
			mutex_unlock(&fs_info->cleaner_mutex);
1862
			goto sleep;
1863
		}
1864

1865
		mutex_lock(&fs_info->cleaner_delayed_iput_mutex);
1866
		btrfs_run_delayed_iputs(fs_info);
1867
		mutex_unlock(&fs_info->cleaner_delayed_iput_mutex);
1868

1869
		again = btrfs_clean_one_deleted_snapshot(root);
1870
		mutex_unlock(&fs_info->cleaner_mutex);
1871 1872

		/*
1873 1874
		 * The defragger has dealt with the R/O remount and umount,
		 * needn't do anything special here.
1875
		 */
1876
		btrfs_run_defrag_inodes(fs_info);
1877 1878 1879 1880 1881 1882 1883 1884 1885

		/*
		 * 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.
		 */
1886
		btrfs_delete_unused_bgs(fs_info);
1887
sleep:
1888
		if (!again) {
1889
			set_current_state(TASK_INTERRUPTIBLE);
1890 1891
			if (!kthread_should_stop())
				schedule();
1892 1893 1894
			__set_current_state(TASK_RUNNING);
		}
	} while (!kthread_should_stop());
1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909

	/*
	 * 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)
1910
			btrfs_err(fs_info,
1911 1912 1913 1914 1915
				  "cleaner transaction attach returned %ld",
				  PTR_ERR(trans));
	} else {
		int ret;

1916
		ret = btrfs_commit_transaction(trans);
1917
		if (ret)
1918
			btrfs_err(fs_info,
1919 1920 1921 1922
				  "cleaner open transaction commit returned %d",
				  ret);
	}

1923 1924 1925 1926 1927 1928
	return 0;
}

static int transaction_kthread(void *arg)
{
	struct btrfs_root *root = arg;
1929
	struct btrfs_fs_info *fs_info = root->fs_info;
1930 1931
	struct btrfs_trans_handle *trans;
	struct btrfs_transaction *cur;
1932
	u64 transid;
1933 1934
	unsigned long now;
	unsigned long delay;
1935
	bool cannot_commit;
1936 1937

	do {
1938
		cannot_commit = false;
1939 1940
		delay = HZ * fs_info->commit_interval;
		mutex_lock(&fs_info->transaction_kthread_mutex);
1941

1942 1943
		spin_lock(&fs_info->trans_lock);
		cur = fs_info->running_transaction;
1944
		if (!cur) {
1945
			spin_unlock(&fs_info->trans_lock);
1946 1947
			goto sleep;
		}
Y
Yan Zheng 已提交
1948

1949
		now = get_seconds();
1950
		if (cur->state < TRANS_STATE_BLOCKED &&
1951
		    (now < cur->start_time ||
1952 1953
		     now - cur->start_time < fs_info->commit_interval)) {
			spin_unlock(&fs_info->trans_lock);
1954 1955 1956
			delay = HZ * 5;
			goto sleep;
		}
1957
		transid = cur->transid;
1958
		spin_unlock(&fs_info->trans_lock);
1959

1960
		/* If the file system is aborted, this will always fail. */
1961
		trans = btrfs_attach_transaction(root);
1962
		if (IS_ERR(trans)) {
1963 1964
			if (PTR_ERR(trans) != -ENOENT)
				cannot_commit = true;
1965
			goto sleep;
1966
		}
1967
		if (transid == trans->transid) {
1968
			btrfs_commit_transaction(trans);
1969
		} else {
1970
			btrfs_end_transaction(trans);
1971
		}
1972
sleep:
1973 1974
		wake_up_process(fs_info->cleaner_kthread);
		mutex_unlock(&fs_info->transaction_kthread_mutex);
1975

J
Josef Bacik 已提交
1976
		if (unlikely(test_bit(BTRFS_FS_STATE_ERROR,
1977
				      &fs_info->fs_state)))
1978
			btrfs_cleanup_transaction(fs_info);
1979 1980
		set_current_state(TASK_INTERRUPTIBLE);
		if (!kthread_should_stop() &&
1981
				(!btrfs_transaction_blocked(fs_info) ||
1982 1983 1984
				 cannot_commit))
			schedule_timeout(delay);
		__set_current_state(TASK_RUNNING);
1985 1986 1987 1988
	} while (!kthread_should_stop());
	return 0;
}

C
Chris Mason 已提交
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 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094
/*
 * 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));

2095 2096 2097 2098 2099 2100 2101 2102
	/*
	 * 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
2103
			       btrfs_header_generation(info->fs_root->node));
2104
		btrfs_set_backup_fs_root_level(root_backup,
C
Chris Mason 已提交
2105
			       btrfs_header_level(info->fs_root->node));
2106
	}
C
Chris Mason 已提交
2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187

	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
2188 2189 2190
/* helper to cleanup workers */
static void btrfs_stop_all_workers(struct btrfs_fs_info *fs_info)
{
2191
	btrfs_destroy_workqueue(fs_info->fixup_workers);
2192
	btrfs_destroy_workqueue(fs_info->delalloc_workers);
2193
	btrfs_destroy_workqueue(fs_info->workers);
2194 2195 2196
	btrfs_destroy_workqueue(fs_info->endio_workers);
	btrfs_destroy_workqueue(fs_info->endio_meta_workers);
	btrfs_destroy_workqueue(fs_info->endio_raid56_workers);
2197
	btrfs_destroy_workqueue(fs_info->endio_repair_workers);
2198
	btrfs_destroy_workqueue(fs_info->rmw_workers);
2199 2200 2201
	btrfs_destroy_workqueue(fs_info->endio_meta_write_workers);
	btrfs_destroy_workqueue(fs_info->endio_write_workers);
	btrfs_destroy_workqueue(fs_info->endio_freespace_worker);
2202
	btrfs_destroy_workqueue(fs_info->submit_workers);
2203
	btrfs_destroy_workqueue(fs_info->delayed_workers);
2204
	btrfs_destroy_workqueue(fs_info->caching_workers);
2205
	btrfs_destroy_workqueue(fs_info->readahead_workers);
2206
	btrfs_destroy_workqueue(fs_info->flush_workers);
2207
	btrfs_destroy_workqueue(fs_info->qgroup_rescan_workers);
C
Chris Mason 已提交
2208
	btrfs_destroy_workqueue(fs_info->extent_workers);
L
Liu Bo 已提交
2209 2210
}

2211 2212 2213 2214 2215 2216 2217 2218 2219 2220
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
2221 2222 2223
/* helper to cleanup tree roots */
static void free_root_pointers(struct btrfs_fs_info *info, int chunk_root)
{
2224
	free_root_extent_buffers(info->tree_root);
2225

2226 2227 2228 2229 2230 2231 2232
	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);
2233
	free_root_extent_buffers(info->free_space_root);
C
Chris Mason 已提交
2234 2235
}

2236
void btrfs_free_fs_roots(struct btrfs_fs_info *fs_info)
2237 2238 2239 2240 2241 2242 2243 2244 2245 2246
{
	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);

2247
		if (test_bit(BTRFS_ROOT_IN_RADIX, &gang[0]->state)) {
2248
			btrfs_drop_and_free_fs_root(fs_info, gang[0]);
2249 2250 2251
		} else {
			free_extent_buffer(gang[0]->node);
			free_extent_buffer(gang[0]->commit_root);
2252
			btrfs_put_fs_root(gang[0]);
2253 2254 2255 2256 2257 2258 2259 2260 2261 2262
		}
	}

	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++)
2263
			btrfs_drop_and_free_fs_root(fs_info, gang[i]);
2264
	}
2265 2266 2267

	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
		btrfs_free_log_root_tree(NULL, fs_info);
2268
		btrfs_destroy_pinned_extent(fs_info, fs_info->pinned_extents);
2269
	}
2270
}
C
Chris Mason 已提交
2271

2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282
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;
}

2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293
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);
}

2294
static void btrfs_init_btree_inode(struct btrfs_fs_info *fs_info)
2295
{
2296 2297 2298 2299
	struct inode *inode = fs_info->btree_inode;

	inode->i_ino = BTRFS_BTREE_INODE_OBJECTID;
	set_nlink(inode, 1);
2300 2301 2302 2303 2304
	/*
	 * 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
	 */
2305 2306
	inode->i_size = OFFSET_MAX;
	inode->i_mapping->a_ops = &btree_aops;
2307

2308 2309 2310 2311
	RB_CLEAR_NODE(&BTRFS_I(inode)->rb_node);
	extent_io_tree_init(&BTRFS_I(inode)->io_tree, inode->i_mapping);
	BTRFS_I(inode)->io_tree.track_uptodate = 0;
	extent_map_tree_init(&BTRFS_I(inode)->extent_tree);
2312

2313
	BTRFS_I(inode)->io_tree.ops = &btree_extent_io_ops;
2314

2315 2316 2317 2318
	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);
2319 2320
}

2321 2322 2323 2324 2325
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);
2326 2327 2328
	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);
2329
	init_waitqueue_head(&fs_info->replace_wait);
2330
	init_waitqueue_head(&fs_info->dev_replace.read_lock_wq);
2331 2332
}

2333 2334 2335 2336 2337 2338 2339 2340 2341
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;
2342
	fs_info->qgroup_rescan_running = false;
2343 2344 2345
	mutex_init(&fs_info->qgroup_rescan_lock);
}

2346 2347 2348 2349
static int btrfs_init_workqueues(struct btrfs_fs_info *fs_info,
		struct btrfs_fs_devices *fs_devices)
{
	int max_active = fs_info->thread_pool_size;
2350
	unsigned int flags = WQ_MEM_RECLAIM | WQ_FREEZABLE | WQ_UNBOUND;
2351 2352

	fs_info->workers =
2353 2354
		btrfs_alloc_workqueue(fs_info, "worker",
				      flags | WQ_HIGHPRI, max_active, 16);
2355 2356

	fs_info->delalloc_workers =
2357 2358
		btrfs_alloc_workqueue(fs_info, "delalloc",
				      flags, max_active, 2);
2359 2360

	fs_info->flush_workers =
2361 2362
		btrfs_alloc_workqueue(fs_info, "flush_delalloc",
				      flags, max_active, 0);
2363 2364

	fs_info->caching_workers =
2365
		btrfs_alloc_workqueue(fs_info, "cache", flags, max_active, 0);
2366 2367 2368 2369 2370 2371 2372

	/*
	 * 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 =
2373
		btrfs_alloc_workqueue(fs_info, "submit", flags,
2374 2375 2376 2377
				      min_t(u64, fs_devices->num_devices,
					    max_active), 64);

	fs_info->fixup_workers =
2378
		btrfs_alloc_workqueue(fs_info, "fixup", flags, 1, 0);
2379 2380 2381 2382 2383 2384

	/*
	 * endios are largely parallel and should have a very
	 * low idle thresh
	 */
	fs_info->endio_workers =
2385
		btrfs_alloc_workqueue(fs_info, "endio", flags, max_active, 4);
2386
	fs_info->endio_meta_workers =
2387 2388
		btrfs_alloc_workqueue(fs_info, "endio-meta", flags,
				      max_active, 4);
2389
	fs_info->endio_meta_write_workers =
2390 2391
		btrfs_alloc_workqueue(fs_info, "endio-meta-write", flags,
				      max_active, 2);
2392
	fs_info->endio_raid56_workers =
2393 2394
		btrfs_alloc_workqueue(fs_info, "endio-raid56", flags,
				      max_active, 4);
2395
	fs_info->endio_repair_workers =
2396
		btrfs_alloc_workqueue(fs_info, "endio-repair", flags, 1, 0);
2397
	fs_info->rmw_workers =
2398
		btrfs_alloc_workqueue(fs_info, "rmw", flags, max_active, 2);
2399
	fs_info->endio_write_workers =
2400 2401
		btrfs_alloc_workqueue(fs_info, "endio-write", flags,
				      max_active, 2);
2402
	fs_info->endio_freespace_worker =
2403 2404
		btrfs_alloc_workqueue(fs_info, "freespace-write", flags,
				      max_active, 0);
2405
	fs_info->delayed_workers =
2406 2407
		btrfs_alloc_workqueue(fs_info, "delayed-meta", flags,
				      max_active, 0);
2408
	fs_info->readahead_workers =
2409 2410
		btrfs_alloc_workqueue(fs_info, "readahead", flags,
				      max_active, 2);
2411
	fs_info->qgroup_rescan_workers =
2412
		btrfs_alloc_workqueue(fs_info, "qgroup-rescan", flags, 1, 0);
2413
	fs_info->extent_workers =
2414
		btrfs_alloc_workqueue(fs_info, "extent-refs", flags,
2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434
				      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;
}

2435 2436 2437 2438 2439 2440 2441 2442 2443
static int btrfs_replay_log(struct btrfs_fs_info *fs_info,
			    struct btrfs_fs_devices *fs_devices)
{
	int ret;
	struct btrfs_root *log_tree_root;
	struct btrfs_super_block *disk_super = fs_info->super_copy;
	u64 bytenr = btrfs_super_log_root(disk_super);

	if (fs_devices->rw_devices == 0) {
2444
		btrfs_warn(fs_info, "log replay required on RO media");
2445 2446 2447
		return -EIO;
	}

2448
	log_tree_root = btrfs_alloc_root(fs_info, GFP_KERNEL);
2449 2450 2451
	if (!log_tree_root)
		return -ENOMEM;

2452
	__setup_root(log_tree_root, fs_info, BTRFS_TREE_LOG_OBJECTID);
2453

2454 2455
	log_tree_root->node = read_tree_block(fs_info, bytenr,
					      fs_info->generation + 1);
2456
	if (IS_ERR(log_tree_root->node)) {
2457
		btrfs_warn(fs_info, "failed to read log tree");
2458
		ret = PTR_ERR(log_tree_root->node);
2459
		kfree(log_tree_root);
2460
		return ret;
2461
	} else if (!extent_buffer_uptodate(log_tree_root->node)) {
2462
		btrfs_err(fs_info, "failed to read log tree");
2463 2464 2465 2466 2467 2468 2469
		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) {
2470 2471
		btrfs_handle_fs_error(fs_info, ret,
				      "Failed to recover log tree");
2472 2473 2474 2475 2476 2477
		free_extent_buffer(log_tree_root->node);
		kfree(log_tree_root);
		return ret;
	}

	if (fs_info->sb->s_flags & MS_RDONLY) {
2478
		ret = btrfs_commit_super(fs_info);
2479 2480 2481 2482 2483 2484 2485
		if (ret)
			return ret;
	}

	return 0;
}

2486
static int btrfs_read_roots(struct btrfs_fs_info *fs_info)
2487
{
2488
	struct btrfs_root *tree_root = fs_info->tree_root;
2489
	struct btrfs_root *root;
2490 2491 2492
	struct btrfs_key location;
	int ret;

2493 2494
	BUG_ON(!fs_info->tree_root);

2495 2496 2497 2498
	location.objectid = BTRFS_EXTENT_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
	location.offset = 0;

2499 2500 2501 2502 2503
	root = btrfs_read_tree_root(tree_root, &location);
	if (IS_ERR(root))
		return PTR_ERR(root);
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
	fs_info->extent_root = root;
2504 2505

	location.objectid = BTRFS_DEV_TREE_OBJECTID;
2506 2507 2508 2509 2510
	root = btrfs_read_tree_root(tree_root, &location);
	if (IS_ERR(root))
		return PTR_ERR(root);
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
	fs_info->dev_root = root;
2511 2512 2513
	btrfs_init_devices_late(fs_info);

	location.objectid = BTRFS_CSUM_TREE_OBJECTID;
2514 2515 2516 2517 2518
	root = btrfs_read_tree_root(tree_root, &location);
	if (IS_ERR(root))
		return PTR_ERR(root);
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
	fs_info->csum_root = root;
2519 2520

	location.objectid = BTRFS_QUOTA_TREE_OBJECTID;
2521 2522 2523
	root = btrfs_read_tree_root(tree_root, &location);
	if (!IS_ERR(root)) {
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
2524
		set_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags);
2525
		fs_info->quota_root = root;
2526 2527 2528
	}

	location.objectid = BTRFS_UUID_TREE_OBJECTID;
2529 2530 2531
	root = btrfs_read_tree_root(tree_root, &location);
	if (IS_ERR(root)) {
		ret = PTR_ERR(root);
2532 2533 2534
		if (ret != -ENOENT)
			return ret;
	} else {
2535 2536
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
		fs_info->uuid_root = root;
2537 2538
	}

2539 2540 2541 2542 2543 2544 2545 2546 2547
	if (btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
		location.objectid = BTRFS_FREE_SPACE_TREE_OBJECTID;
		root = btrfs_read_tree_root(tree_root, &location);
		if (IS_ERR(root))
			return PTR_ERR(root);
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
		fs_info->free_space_root = root;
	}

2548 2549 2550
	return 0;
}

A
Al Viro 已提交
2551 2552 2553
int open_ctree(struct super_block *sb,
	       struct btrfs_fs_devices *fs_devices,
	       char *options)
2554
{
2555 2556
	u32 sectorsize;
	u32 nodesize;
2557
	u32 stripesize;
2558
	u64 generation;
2559
	u64 features;
2560
	struct btrfs_key location;
2561
	struct buffer_head *bh;
2562
	struct btrfs_super_block *disk_super;
2563
	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
2564
	struct btrfs_root *tree_root;
2565
	struct btrfs_root *chunk_root;
2566
	int ret;
2567
	int err = -EINVAL;
C
Chris Mason 已提交
2568 2569
	int num_backups_tried = 0;
	int backup_index = 0;
2570
	int max_active;
2571
	int clear_free_space_tree = 0;
2572

2573 2574
	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);
2575
	if (!tree_root || !chunk_root) {
C
Chris Mason 已提交
2576 2577 2578
		err = -ENOMEM;
		goto fail;
	}
2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591

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

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

2592
	ret = percpu_counter_init(&fs_info->dirty_metadata_bytes, 0, GFP_KERNEL);
2593 2594 2595 2596
	if (ret) {
		err = ret;
		goto fail_bdi;
	}
2597
	fs_info->dirty_metadata_batch = PAGE_SIZE *
2598 2599
					(1 + ilog2(nr_cpu_ids));

2600
	ret = percpu_counter_init(&fs_info->delalloc_bytes, 0, GFP_KERNEL);
2601 2602 2603 2604 2605
	if (ret) {
		err = ret;
		goto fail_dirty_metadata_bytes;
	}

2606
	ret = percpu_counter_init(&fs_info->bio_counter, 0, GFP_KERNEL);
2607 2608 2609 2610 2611
	if (ret) {
		err = ret;
		goto fail_delalloc_bytes;
	}

2612 2613 2614
	fs_info->btree_inode = new_inode(sb);
	if (!fs_info->btree_inode) {
		err = -ENOMEM;
2615
		goto fail_bio_counter;
2616 2617
	}

2618
	mapping_set_gfp_mask(fs_info->btree_inode->i_mapping, GFP_NOFS);
2619

2620
	INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC);
2621
	INIT_RADIX_TREE(&fs_info->buffer_radix, GFP_ATOMIC);
C
Chris Mason 已提交
2622
	INIT_LIST_HEAD(&fs_info->trans_list);
2623
	INIT_LIST_HEAD(&fs_info->dead_roots);
Y
Yan, Zheng 已提交
2624
	INIT_LIST_HEAD(&fs_info->delayed_iputs);
2625
	INIT_LIST_HEAD(&fs_info->delalloc_roots);
2626
	INIT_LIST_HEAD(&fs_info->caching_block_groups);
2627
	spin_lock_init(&fs_info->delalloc_root_lock);
J
Josef Bacik 已提交
2628
	spin_lock_init(&fs_info->trans_lock);
2629
	spin_lock_init(&fs_info->fs_roots_radix_lock);
Y
Yan, Zheng 已提交
2630
	spin_lock_init(&fs_info->delayed_iput_lock);
C
Chris Mason 已提交
2631
	spin_lock_init(&fs_info->defrag_inodes_lock);
2632
	spin_lock_init(&fs_info->free_chunk_lock);
J
Jan Schmidt 已提交
2633
	spin_lock_init(&fs_info->tree_mod_seq_lock);
2634
	spin_lock_init(&fs_info->super_lock);
J
Josef Bacik 已提交
2635
	spin_lock_init(&fs_info->qgroup_op_lock);
2636
	spin_lock_init(&fs_info->buffer_lock);
2637
	spin_lock_init(&fs_info->unused_bgs_lock);
J
Jan Schmidt 已提交
2638
	rwlock_init(&fs_info->tree_mod_log_lock);
2639
	mutex_init(&fs_info->unused_bg_unpin_mutex);
2640
	mutex_init(&fs_info->delete_unused_bgs_mutex);
C
Chris Mason 已提交
2641
	mutex_init(&fs_info->reloc_mutex);
2642
	mutex_init(&fs_info->delalloc_root_mutex);
2643
	mutex_init(&fs_info->cleaner_delayed_iput_mutex);
2644
	seqlock_init(&fs_info->profiles_lock);
2645

2646
	INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots);
2647
	INIT_LIST_HEAD(&fs_info->space_info);
J
Jan Schmidt 已提交
2648
	INIT_LIST_HEAD(&fs_info->tree_mod_seq_list);
2649
	INIT_LIST_HEAD(&fs_info->unused_bgs);
2650
	btrfs_mapping_init(&fs_info->mapping_tree);
2651 2652 2653 2654 2655 2656 2657 2658 2659
	btrfs_init_block_rsv(&fs_info->global_block_rsv,
			     BTRFS_BLOCK_RSV_GLOBAL);
	btrfs_init_block_rsv(&fs_info->delalloc_block_rsv,
			     BTRFS_BLOCK_RSV_DELALLOC);
	btrfs_init_block_rsv(&fs_info->trans_block_rsv, BTRFS_BLOCK_RSV_TRANS);
	btrfs_init_block_rsv(&fs_info->chunk_block_rsv, BTRFS_BLOCK_RSV_CHUNK);
	btrfs_init_block_rsv(&fs_info->empty_block_rsv, BTRFS_BLOCK_RSV_EMPTY);
	btrfs_init_block_rsv(&fs_info->delayed_block_rsv,
			     BTRFS_BLOCK_RSV_DELOPS);
2660
	atomic_set(&fs_info->nr_async_submits, 0);
2661
	atomic_set(&fs_info->async_delalloc_pages, 0);
2662
	atomic_set(&fs_info->async_submit_draining, 0);
2663
	atomic_set(&fs_info->nr_async_bios, 0);
C
Chris Mason 已提交
2664
	atomic_set(&fs_info->defrag_running, 0);
J
Josef Bacik 已提交
2665
	atomic_set(&fs_info->qgroup_op_seq, 0);
Z
Zhao Lei 已提交
2666
	atomic_set(&fs_info->reada_works_cnt, 0);
2667
	atomic64_set(&fs_info->tree_mod_seq, 0);
2668
	fs_info->fs_frozen = 0;
C
Chris Mason 已提交
2669
	fs_info->sb = sb;
2670
	fs_info->max_inline = BTRFS_DEFAULT_MAX_INLINE;
J
Josef Bacik 已提交
2671
	fs_info->metadata_ratio = 0;
C
Chris Mason 已提交
2672
	fs_info->defrag_inodes = RB_ROOT;
2673
	fs_info->free_chunk_space = 0;
J
Jan Schmidt 已提交
2674
	fs_info->tree_mod_log = RB_ROOT;
2675
	fs_info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
2676
	fs_info->avg_delayed_ref_runtime = NSEC_PER_SEC >> 6; /* div by 64 */
2677
	/* readahead state */
2678
	INIT_RADIX_TREE(&fs_info->reada_tree, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
2679
	spin_lock_init(&fs_info->reada_lock);
C
Chris Mason 已提交
2680

2681 2682
	fs_info->thread_pool_size = min_t(unsigned long,
					  num_online_cpus() + 2, 8);
2683

2684 2685
	INIT_LIST_HEAD(&fs_info->ordered_roots);
	spin_lock_init(&fs_info->ordered_root_lock);
2686
	fs_info->delayed_root = kmalloc(sizeof(struct btrfs_delayed_root),
2687
					GFP_KERNEL);
2688 2689 2690 2691 2692
	if (!fs_info->delayed_root) {
		err = -ENOMEM;
		goto fail_iput;
	}
	btrfs_init_delayed_root(fs_info->delayed_root);
2693

2694
	btrfs_init_scrub(fs_info);
2695 2696 2697
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	fs_info->check_integrity_print_mask = 0;
#endif
2698
	btrfs_init_balance(fs_info);
2699
	btrfs_init_async_reclaim_work(&fs_info->async_reclaim_work);
A
Arne Jansen 已提交
2700

2701 2702
	sb->s_blocksize = 4096;
	sb->s_blocksize_bits = blksize_bits(4096);
J
Jens Axboe 已提交
2703
	sb->s_bdi = &fs_info->bdi;
2704

2705
	btrfs_init_btree_inode(fs_info);
2706

J
Josef Bacik 已提交
2707
	spin_lock_init(&fs_info->block_group_cache_lock);
2708
	fs_info->block_group_cache_tree = RB_ROOT;
2709
	fs_info->first_logical_byte = (u64)-1;
J
Josef Bacik 已提交
2710

2711
	extent_io_tree_init(&fs_info->freed_extents[0],
2712
			     fs_info->btree_inode->i_mapping);
2713
	extent_io_tree_init(&fs_info->freed_extents[1],
2714
			     fs_info->btree_inode->i_mapping);
2715
	fs_info->pinned_extents = &fs_info->freed_extents[0];
2716
	set_bit(BTRFS_FS_BARRIER, &fs_info->flags);
C
Chris Mason 已提交
2717

2718
	mutex_init(&fs_info->ordered_operations_mutex);
2719
	mutex_init(&fs_info->tree_log_mutex);
2720
	mutex_init(&fs_info->chunk_mutex);
2721 2722
	mutex_init(&fs_info->transaction_kthread_mutex);
	mutex_init(&fs_info->cleaner_mutex);
2723
	mutex_init(&fs_info->volume_mutex);
2724
	mutex_init(&fs_info->ro_block_group_mutex);
2725
	init_rwsem(&fs_info->commit_root_sem);
2726
	init_rwsem(&fs_info->cleanup_work_sem);
2727
	init_rwsem(&fs_info->subvol_sem);
S
Stefan Behrens 已提交
2728
	sema_init(&fs_info->uuid_tree_rescan_sem, 1);
2729

2730
	btrfs_init_dev_replace_locks(fs_info);
2731
	btrfs_init_qgroup(fs_info);
2732

2733 2734 2735
	btrfs_init_free_cluster(&fs_info->meta_alloc_cluster);
	btrfs_init_free_cluster(&fs_info->data_alloc_cluster);

2736
	init_waitqueue_head(&fs_info->transaction_throttle);
2737
	init_waitqueue_head(&fs_info->transaction_wait);
S
Sage Weil 已提交
2738
	init_waitqueue_head(&fs_info->transaction_blocked_wait);
2739
	init_waitqueue_head(&fs_info->async_submit_wait);
2740

2741 2742
	INIT_LIST_HEAD(&fs_info->pinned_chunks);

2743 2744 2745 2746 2747
	/* 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
2748 2749
	ret = btrfs_alloc_stripe_hash_table(fs_info);
	if (ret) {
2750
		err = ret;
D
David Woodhouse 已提交
2751 2752 2753
		goto fail_alloc;
	}

2754
	__setup_root(tree_root, fs_info, BTRFS_ROOT_TREE_OBJECTID);
2755

2756
	invalidate_bdev(fs_devices->latest_bdev);
D
David Sterba 已提交
2757 2758 2759 2760

	/*
	 * Read super block and check the signature bytes only
	 */
Y
Yan Zheng 已提交
2761
	bh = btrfs_read_dev_super(fs_devices->latest_bdev);
2762 2763
	if (IS_ERR(bh)) {
		err = PTR_ERR(bh);
2764
		goto fail_alloc;
2765
	}
C
Chris Mason 已提交
2766

D
David Sterba 已提交
2767 2768 2769 2770
	/*
	 * We want to check superblock checksum, the type is stored inside.
	 * Pass the whole disk block of size BTRFS_SUPER_INFO_SIZE (4k).
	 */
2771
	if (btrfs_check_super_csum(fs_info, bh->b_data)) {
2772
		btrfs_err(fs_info, "superblock checksum mismatch");
D
David Sterba 已提交
2773
		err = -EINVAL;
2774
		brelse(bh);
D
David Sterba 已提交
2775 2776 2777 2778 2779 2780 2781 2782
		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
	 */
2783 2784 2785
	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));
2786
	brelse(bh);
2787

2788
	memcpy(fs_info->fsid, fs_info->super_copy->fsid, BTRFS_FSID_SIZE);
2789

D
David Sterba 已提交
2790 2791
	ret = btrfs_check_super_valid(fs_info, sb->s_flags & MS_RDONLY);
	if (ret) {
2792
		btrfs_err(fs_info, "superblock contains fatal errors");
D
David Sterba 已提交
2793 2794 2795 2796
		err = -EINVAL;
		goto fail_alloc;
	}

2797
	disk_super = fs_info->super_copy;
2798
	if (!btrfs_super_root(disk_super))
2799
		goto fail_alloc;
2800

L
liubo 已提交
2801
	/* check FS state, whether FS is broken. */
2802 2803
	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
2804

C
Chris Mason 已提交
2805 2806 2807 2808 2809 2810 2811
	/*
	 * 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);

2812 2813 2814 2815 2816 2817
	/*
	 * 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;

2818
	ret = btrfs_parse_options(fs_info, options, sb->s_flags);
Y
Yan Zheng 已提交
2819 2820
	if (ret) {
		err = ret;
2821
		goto fail_alloc;
Y
Yan Zheng 已提交
2822
	}
2823

2824 2825 2826
	features = btrfs_super_incompat_flags(disk_super) &
		~BTRFS_FEATURE_INCOMPAT_SUPP;
	if (features) {
2827 2828 2829
		btrfs_err(fs_info,
		    "cannot mount because of unsupported optional features (%llx)",
		    features);
2830
		err = -EINVAL;
2831
		goto fail_alloc;
2832 2833
	}

2834
	features = btrfs_super_incompat_flags(disk_super);
L
Li Zefan 已提交
2835
	features |= BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF;
2836
	if (fs_info->compress_type == BTRFS_COMPRESS_LZO)
L
Li Zefan 已提交
2837
		features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
2838

2839
	if (features & BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA)
2840
		btrfs_info(fs_info, "has skinny extents");
2841

2842 2843 2844 2845
	/*
	 * flag our filesystem as having big metadata blocks if
	 * they are bigger than the page size
	 */
2846
	if (btrfs_super_nodesize(disk_super) > PAGE_SIZE) {
2847
		if (!(features & BTRFS_FEATURE_INCOMPAT_BIG_METADATA))
2848 2849
			btrfs_info(fs_info,
				"flagging fs with big metadata feature");
2850 2851 2852
		features |= BTRFS_FEATURE_INCOMPAT_BIG_METADATA;
	}

2853 2854
	nodesize = btrfs_super_nodesize(disk_super);
	sectorsize = btrfs_super_sectorsize(disk_super);
2855
	stripesize = sectorsize;
2856
	fs_info->dirty_metadata_batch = nodesize * (1 + ilog2(nr_cpu_ids));
2857
	fs_info->delalloc_batch = sectorsize * 512 * (1 + ilog2(nr_cpu_ids));
2858

2859 2860 2861 2862 2863
	/* Cache block sizes */
	fs_info->nodesize = nodesize;
	fs_info->sectorsize = sectorsize;
	fs_info->stripesize = stripesize;

2864 2865 2866 2867 2868
	/*
	 * mixed block groups end up with duplicate but slightly offset
	 * extent buffers for the same range.  It leads to corruptions
	 */
	if ((features & BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS) &&
2869
	    (sectorsize != nodesize)) {
2870 2871 2872
		btrfs_err(fs_info,
"unequal nodesize/sectorsize (%u != %u) are not allowed for mixed block groups",
			nodesize, sectorsize);
2873 2874 2875
		goto fail_alloc;
	}

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

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

2892
	max_active = fs_info->thread_pool_size;
2893

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

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

2904 2905
	sb->s_blocksize = sectorsize;
	sb->s_blocksize_bits = blksize_bits(sectorsize);
2906

2907
	mutex_lock(&fs_info->chunk_mutex);
2908
	ret = btrfs_read_sys_array(fs_info);
2909
	mutex_unlock(&fs_info->chunk_mutex);
2910
	if (ret) {
2911
		btrfs_err(fs_info, "failed to read the system array: %d", ret);
2912
		goto fail_sb_buffer;
2913
	}
2914

2915
	generation = btrfs_super_chunk_root_generation(disk_super);
2916

2917
	__setup_root(chunk_root, fs_info, BTRFS_CHUNK_TREE_OBJECTID);
2918

2919
	chunk_root->node = read_tree_block(fs_info,
2920
					   btrfs_super_chunk_root(disk_super),
2921
					   generation);
2922 2923
	if (IS_ERR(chunk_root->node) ||
	    !extent_buffer_uptodate(chunk_root->node)) {
2924
		btrfs_err(fs_info, "failed to read chunk root");
2925 2926
		if (!IS_ERR(chunk_root->node))
			free_extent_buffer(chunk_root->node);
2927
		chunk_root->node = NULL;
C
Chris Mason 已提交
2928
		goto fail_tree_roots;
2929
	}
2930 2931
	btrfs_set_root_node(&chunk_root->root_item, chunk_root->node);
	chunk_root->commit_root = btrfs_root_node(chunk_root);
2932

2933
	read_extent_buffer(chunk_root->node, fs_info->chunk_tree_uuid,
2934
	   btrfs_header_chunk_tree_uuid(chunk_root->node), BTRFS_UUID_SIZE);
2935

2936
	ret = btrfs_read_chunk_tree(fs_info);
Y
Yan Zheng 已提交
2937
	if (ret) {
2938
		btrfs_err(fs_info, "failed to read chunk tree: %d", ret);
C
Chris Mason 已提交
2939
		goto fail_tree_roots;
Y
Yan Zheng 已提交
2940
	}
2941

2942 2943 2944 2945
	/*
	 * keep the device that is marked to be the target device for the
	 * dev_replace procedure
	 */
2946
	btrfs_close_extra_devices(fs_devices, 0);
2947

2948
	if (!fs_devices->latest_bdev) {
2949
		btrfs_err(fs_info, "failed to read devices");
2950 2951 2952
		goto fail_tree_roots;
	}

C
Chris Mason 已提交
2953
retry_root_backup:
2954
	generation = btrfs_super_generation(disk_super);
2955

2956
	tree_root->node = read_tree_block(fs_info,
2957
					  btrfs_super_root(disk_super),
2958
					  generation);
2959 2960
	if (IS_ERR(tree_root->node) ||
	    !extent_buffer_uptodate(tree_root->node)) {
2961
		btrfs_warn(fs_info, "failed to read tree root");
2962 2963
		if (!IS_ERR(tree_root->node))
			free_extent_buffer(tree_root->node);
2964
		tree_root->node = NULL;
C
Chris Mason 已提交
2965
		goto recovery_tree_root;
2966
	}
C
Chris Mason 已提交
2967

2968 2969
	btrfs_set_root_node(&tree_root->root_item, tree_root->node);
	tree_root->commit_root = btrfs_root_node(tree_root);
2970
	btrfs_set_root_refs(&tree_root->root_item, 1);
2971

2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983
	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);

2984
	ret = btrfs_read_roots(fs_info);
2985
	if (ret)
C
Chris Mason 已提交
2986
		goto recovery_tree_root;
2987

2988 2989 2990
	fs_info->generation = generation;
	fs_info->last_trans_committed = generation;

2991 2992
	ret = btrfs_recover_balance(fs_info);
	if (ret) {
2993
		btrfs_err(fs_info, "failed to recover balance: %d", ret);
2994 2995 2996
		goto fail_block_groups;
	}

2997 2998
	ret = btrfs_init_dev_stats(fs_info);
	if (ret) {
2999
		btrfs_err(fs_info, "failed to init dev_stats: %d", ret);
3000 3001 3002
		goto fail_block_groups;
	}

3003 3004
	ret = btrfs_init_dev_replace(fs_info);
	if (ret) {
3005
		btrfs_err(fs_info, "failed to init dev_replace: %d", ret);
3006 3007 3008
		goto fail_block_groups;
	}

3009
	btrfs_close_extra_devices(fs_devices, 1);
3010

3011 3012
	ret = btrfs_sysfs_add_fsid(fs_devices, NULL);
	if (ret) {
3013 3014
		btrfs_err(fs_info, "failed to init sysfs fsid interface: %d",
				ret);
3015 3016 3017 3018 3019
		goto fail_block_groups;
	}

	ret = btrfs_sysfs_add_device(fs_devices);
	if (ret) {
3020 3021
		btrfs_err(fs_info, "failed to init sysfs device interface: %d",
				ret);
3022 3023 3024
		goto fail_fsdev_sysfs;
	}

3025
	ret = btrfs_sysfs_add_mounted(fs_info);
3026
	if (ret) {
3027
		btrfs_err(fs_info, "failed to init sysfs interface: %d", ret);
3028
		goto fail_fsdev_sysfs;
3029 3030 3031 3032
	}

	ret = btrfs_init_space_info(fs_info);
	if (ret) {
3033
		btrfs_err(fs_info, "failed to initialize space info: %d", ret);
3034
		goto fail_sysfs;
3035 3036
	}

3037
	ret = btrfs_read_block_groups(fs_info);
3038
	if (ret) {
3039
		btrfs_err(fs_info, "failed to read block groups: %d", ret);
3040
		goto fail_sysfs;
3041
	}
3042 3043
	fs_info->num_tolerated_disk_barrier_failures =
		btrfs_calc_num_tolerated_disk_barrier_failures(fs_info);
3044 3045 3046
	if (fs_info->fs_devices->missing_devices >
	     fs_info->num_tolerated_disk_barrier_failures &&
	    !(sb->s_flags & MS_RDONLY)) {
3047 3048
		btrfs_warn(fs_info,
"missing devices (%llu) exceeds the limit (%d), writeable mount is not allowed",
3049 3050
			fs_info->fs_devices->missing_devices,
			fs_info->num_tolerated_disk_barrier_failures);
3051
		goto fail_sysfs;
3052
	}
C
Chris Mason 已提交
3053

3054 3055
	fs_info->cleaner_kthread = kthread_run(cleaner_kthread, tree_root,
					       "btrfs-cleaner");
3056
	if (IS_ERR(fs_info->cleaner_kthread))
3057
		goto fail_sysfs;
3058 3059 3060 3061

	fs_info->transaction_kthread = kthread_run(transaction_kthread,
						   tree_root,
						   "btrfs-transaction");
3062
	if (IS_ERR(fs_info->transaction_kthread))
3063
		goto fail_cleaner;
3064

3065 3066
	if (!btrfs_test_opt(fs_info, SSD) &&
	    !btrfs_test_opt(fs_info, NOSSD) &&
C
Chris Mason 已提交
3067
	    !fs_info->fs_devices->rotating) {
3068
		btrfs_info(fs_info, "detected SSD devices, enabling SSD mode");
C
Chris Mason 已提交
3069 3070 3071
		btrfs_set_opt(fs_info->mount_opt, SSD);
	}

3072
	/*
3073
	 * Mount does not set all options immediately, we can do it now and do
3074 3075 3076
	 * not have to wait for transaction commit
	 */
	btrfs_apply_pending_changes(fs_info);
3077

3078
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
3079
	if (btrfs_test_opt(fs_info, CHECK_INTEGRITY)) {
3080
		ret = btrfsic_mount(fs_info, fs_devices,
3081
				    btrfs_test_opt(fs_info,
3082 3083 3084 3085
					CHECK_INTEGRITY_INCLUDING_EXTENT_DATA) ?
				    1 : 0,
				    fs_info->check_integrity_print_mask);
		if (ret)
3086 3087 3088
			btrfs_warn(fs_info,
				"failed to initialize integrity check module: %d",
				ret);
3089 3090
	}
#endif
3091 3092 3093
	ret = btrfs_read_qgroup_config(fs_info);
	if (ret)
		goto fail_trans_kthread;
3094

3095 3096
	/* do not make disk changes in broken FS or nologreplay is given */
	if (btrfs_super_log_root(disk_super) != 0 &&
3097
	    !btrfs_test_opt(fs_info, NOLOGREPLAY)) {
3098
		ret = btrfs_replay_log(fs_info, fs_devices);
3099
		if (ret) {
3100
			err = ret;
3101
			goto fail_qgroup;
3102
		}
3103
	}
Z
Zheng Yan 已提交
3104

3105
	ret = btrfs_find_orphan_roots(fs_info);
3106
	if (ret)
3107
		goto fail_qgroup;
3108

3109
	if (!(sb->s_flags & MS_RDONLY)) {
3110
		ret = btrfs_cleanup_fs_roots(fs_info);
3111
		if (ret)
3112
			goto fail_qgroup;
3113 3114

		mutex_lock(&fs_info->cleaner_mutex);
3115
		ret = btrfs_recover_relocation(tree_root);
3116
		mutex_unlock(&fs_info->cleaner_mutex);
3117
		if (ret < 0) {
3118 3119
			btrfs_warn(fs_info, "failed to recover relocation: %d",
					ret);
3120
			err = -EINVAL;
3121
			goto fail_qgroup;
3122
		}
3123
	}
Z
Zheng Yan 已提交
3124

3125 3126
	location.objectid = BTRFS_FS_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
3127
	location.offset = 0;
3128 3129

	fs_info->fs_root = btrfs_read_fs_root_no_name(fs_info, &location);
3130 3131
	if (IS_ERR(fs_info->fs_root)) {
		err = PTR_ERR(fs_info->fs_root);
3132
		goto fail_qgroup;
3133
	}
C
Chris Mason 已提交
3134

3135 3136
	if (sb->s_flags & MS_RDONLY)
		return 0;
I
Ilya Dryomov 已提交
3137

3138 3139
	if (btrfs_test_opt(fs_info, CLEAR_CACHE) &&
	    btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
3140 3141 3142 3143 3144 3145 3146 3147
		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) {
3148 3149 3150 3151 3152
		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);
3153
			close_ctree(fs_info);
3154 3155 3156 3157
			return ret;
		}
	}

3158
	if (btrfs_test_opt(fs_info, FREE_SPACE_TREE) &&
3159
	    !btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
3160
		btrfs_info(fs_info, "creating free space tree");
3161 3162
		ret = btrfs_create_free_space_tree(fs_info);
		if (ret) {
3163 3164
			btrfs_warn(fs_info,
				"failed to create free space tree: %d", ret);
3165
			close_ctree(fs_info);
3166 3167 3168 3169
			return ret;
		}
	}

3170 3171 3172
	down_read(&fs_info->cleanup_work_sem);
	if ((ret = btrfs_orphan_cleanup(fs_info->fs_root)) ||
	    (ret = btrfs_orphan_cleanup(fs_info->tree_root))) {
3173
		up_read(&fs_info->cleanup_work_sem);
3174
		close_ctree(fs_info);
3175 3176 3177
		return ret;
	}
	up_read(&fs_info->cleanup_work_sem);
I
Ilya Dryomov 已提交
3178

3179 3180
	ret = btrfs_resume_balance_async(fs_info);
	if (ret) {
3181
		btrfs_warn(fs_info, "failed to resume balance: %d", ret);
3182
		close_ctree(fs_info);
3183
		return ret;
3184 3185
	}

3186 3187
	ret = btrfs_resume_dev_replace_async(fs_info);
	if (ret) {
3188
		btrfs_warn(fs_info, "failed to resume device replace: %d", ret);
3189
		close_ctree(fs_info);
3190 3191 3192
		return ret;
	}

3193 3194
	btrfs_qgroup_rescan_resume(fs_info);

3195
	if (!fs_info->uuid_root) {
3196
		btrfs_info(fs_info, "creating UUID tree");
3197 3198
		ret = btrfs_create_uuid_tree(fs_info);
		if (ret) {
3199 3200
			btrfs_warn(fs_info,
				"failed to create the UUID tree: %d", ret);
3201
			close_ctree(fs_info);
3202 3203
			return ret;
		}
3204
	} else if (btrfs_test_opt(fs_info, RESCAN_UUID_TREE) ||
3205 3206
		   fs_info->generation !=
				btrfs_super_uuid_tree_generation(disk_super)) {
3207
		btrfs_info(fs_info, "checking UUID tree");
3208 3209
		ret = btrfs_check_uuid_tree(fs_info);
		if (ret) {
3210 3211
			btrfs_warn(fs_info,
				"failed to check the UUID tree: %d", ret);
3212
			close_ctree(fs_info);
3213 3214 3215
			return ret;
		}
	} else {
3216
		set_bit(BTRFS_FS_UPDATE_UUID_TREE_GEN, &fs_info->flags);
3217
	}
3218
	set_bit(BTRFS_FS_OPEN, &fs_info->flags);
3219

3220 3221 3222 3223 3224 3225
	/*
	 * 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
3226
	return 0;
C
Chris Mason 已提交
3227

3228 3229
fail_qgroup:
	btrfs_free_qgroup_config(fs_info);
3230 3231
fail_trans_kthread:
	kthread_stop(fs_info->transaction_kthread);
3232
	btrfs_cleanup_transaction(fs_info);
3233
	btrfs_free_fs_roots(fs_info);
3234
fail_cleaner:
3235
	kthread_stop(fs_info->cleaner_kthread);
3236 3237 3238 3239 3240 3241 3242

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

3243
fail_sysfs:
3244
	btrfs_sysfs_remove_mounted(fs_info);
3245

3246 3247 3248
fail_fsdev_sysfs:
	btrfs_sysfs_remove_fsid(fs_info->fs_devices);

3249
fail_block_groups:
J
Josef Bacik 已提交
3250
	btrfs_put_block_group_cache(fs_info);
3251
	btrfs_free_block_groups(fs_info);
C
Chris Mason 已提交
3252 3253 3254

fail_tree_roots:
	free_root_pointers(fs_info, 1);
3255
	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
3256

C
Chris Mason 已提交
3257
fail_sb_buffer:
L
Liu Bo 已提交
3258
	btrfs_stop_all_workers(fs_info);
3259
fail_alloc:
3260
fail_iput:
3261 3262
	btrfs_mapping_tree_free(&fs_info->mapping_tree);

3263
	iput(fs_info->btree_inode);
3264 3265
fail_bio_counter:
	percpu_counter_destroy(&fs_info->bio_counter);
3266 3267
fail_delalloc_bytes:
	percpu_counter_destroy(&fs_info->delalloc_bytes);
3268 3269
fail_dirty_metadata_bytes:
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
3270
fail_bdi:
3271
	bdi_destroy(&fs_info->bdi);
3272 3273
fail_srcu:
	cleanup_srcu_struct(&fs_info->subvol_srcu);
3274
fail:
D
David Woodhouse 已提交
3275
	btrfs_free_stripe_hash_table(fs_info);
3276
	btrfs_close_devices(fs_info->fs_devices);
A
Al Viro 已提交
3277
	return err;
C
Chris Mason 已提交
3278 3279

recovery_tree_root:
3280
	if (!btrfs_test_opt(fs_info, USEBACKUPROOT))
C
Chris Mason 已提交
3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295
		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;
3296 3297
}

3298 3299 3300 3301 3302
static void btrfs_end_buffer_write_sync(struct buffer_head *bh, int uptodate)
{
	if (uptodate) {
		set_buffer_uptodate(bh);
	} else {
3303 3304 3305
		struct btrfs_device *device = (struct btrfs_device *)
			bh->b_private;

3306
		btrfs_warn_rl_in_rcu(device->fs_info,
3307
				"lost page write due to IO error on %s",
3308
					  rcu_str_deref(device->name));
3309
		/* note, we don't set_buffer_write_io_error because we have
3310 3311
		 * our own ways of dealing with the IO errors
		 */
3312
		clear_buffer_uptodate(bh);
3313
		btrfs_dev_stat_inc_and_print(device, BTRFS_DEV_STAT_WRITE_ERRS);
3314 3315 3316 3317 3318
	}
	unlock_buffer(bh);
	put_bh(bh);
}

3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349
int btrfs_read_dev_one_super(struct block_device *bdev, int copy_num,
			struct buffer_head **bh_ret)
{
	struct buffer_head *bh;
	struct btrfs_super_block *super;
	u64 bytenr;

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

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

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

	*bh_ret = bh;
	return 0;
}


Y
Yan Zheng 已提交
3350 3351 3352 3353 3354 3355 3356
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;
3357
	int ret = -EINVAL;
Y
Yan Zheng 已提交
3358 3359 3360 3361 3362 3363 3364

	/* 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++) {
3365 3366
		ret = btrfs_read_dev_one_super(bdev, i, &bh);
		if (ret)
Y
Yan Zheng 已提交
3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378
			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);
		}
	}
3379 3380 3381 3382

	if (!latest)
		return ERR_PTR(ret);

Y
Yan Zheng 已提交
3383 3384 3385
	return latest;
}

3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396
/*
 * this should be called twice, once with wait == 0 and
 * once with wait == 1.  When wait == 0 is done, all the buffer heads
 * we write are pinned.
 *
 * They are released when wait == 1 is done.
 * max_mirrors must be the same for both runs, and it indicates how
 * many supers on this one device should be written.
 *
 * max_mirrors == 0 means to write them all.
 */
Y
Yan Zheng 已提交
3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412
static int write_dev_supers(struct btrfs_device *device,
			    struct btrfs_super_block *sb,
			    int do_barriers, int wait, int max_mirrors)
{
	struct buffer_head *bh;
	int i;
	int ret;
	int errors = 0;
	u32 crc;
	u64 bytenr;

	if (max_mirrors == 0)
		max_mirrors = BTRFS_SUPER_MIRROR_MAX;

	for (i = 0; i < max_mirrors; i++) {
		bytenr = btrfs_sb_offset(i);
3413 3414
		if (bytenr + BTRFS_SUPER_INFO_SIZE >=
		    device->commit_total_bytes)
Y
Yan Zheng 已提交
3415 3416 3417 3418 3419
			break;

		if (wait) {
			bh = __find_get_block(device->bdev, bytenr / 4096,
					      BTRFS_SUPER_INFO_SIZE);
3420 3421 3422 3423
			if (!bh) {
				errors++;
				continue;
			}
Y
Yan Zheng 已提交
3424
			wait_on_buffer(bh);
3425 3426 3427 3428 3429 3430 3431 3432 3433
			if (!buffer_uptodate(bh))
				errors++;

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

			/* drop the reference from the wait == 0 run */
			brelse(bh);
			continue;
Y
Yan Zheng 已提交
3434 3435 3436 3437
		} else {
			btrfs_set_super_bytenr(sb, bytenr);

			crc = ~(u32)0;
3438
			crc = btrfs_csum_data((char *)sb +
Y
Yan Zheng 已提交
3439 3440 3441 3442 3443
					      BTRFS_CSUM_SIZE, crc,
					      BTRFS_SUPER_INFO_SIZE -
					      BTRFS_CSUM_SIZE);
			btrfs_csum_final(crc, sb->csum);

3444 3445 3446 3447
			/*
			 * one reference for us, and we leave it for the
			 * caller
			 */
Y
Yan Zheng 已提交
3448 3449
			bh = __getblk(device->bdev, bytenr / 4096,
				      BTRFS_SUPER_INFO_SIZE);
3450
			if (!bh) {
3451
				btrfs_err(device->fs_info,
3452 3453
				    "couldn't get super buffer head for bytenr %llu",
				    bytenr);
3454 3455 3456 3457
				errors++;
				continue;
			}

Y
Yan Zheng 已提交
3458 3459
			memcpy(bh->b_data, sb, BTRFS_SUPER_INFO_SIZE);

3460
			/* one reference for submit_bh */
Y
Yan Zheng 已提交
3461
			get_bh(bh);
3462 3463

			set_buffer_uptodate(bh);
Y
Yan Zheng 已提交
3464 3465
			lock_buffer(bh);
			bh->b_end_io = btrfs_end_buffer_write_sync;
3466
			bh->b_private = device;
Y
Yan Zheng 已提交
3467 3468
		}

C
Chris Mason 已提交
3469 3470 3471 3472
		/*
		 * we fua the first super.  The others we allow
		 * to go down lazy.
		 */
3473
		if (i == 0)
3474
			ret = btrfsic_submit_bh(REQ_OP_WRITE, WRITE_FUA, bh);
3475
		else
3476
			ret = btrfsic_submit_bh(REQ_OP_WRITE, WRITE_SYNC, bh);
3477
		if (ret)
Y
Yan Zheng 已提交
3478 3479 3480 3481 3482
			errors++;
	}
	return errors < i ? 0 : -1;
}

C
Chris Mason 已提交
3483 3484 3485 3486
/*
 * endio for the write_dev_flush, this will wake anyone waiting
 * for the barrier when it is done
 */
3487
static void btrfs_end_empty_barrier(struct bio *bio)
C
Chris Mason 已提交
3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515
{
	if (bio->bi_private)
		complete(bio->bi_private);
	bio_put(bio);
}

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

	if (device->nobarriers)
		return 0;

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

		wait_for_completion(&device->flush_wait);

3516 3517
		if (bio->bi_error) {
			ret = bio->bi_error;
3518 3519
			btrfs_dev_stat_inc_and_print(device,
				BTRFS_DEV_STAT_FLUSH_ERRS);
C
Chris Mason 已提交
3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532
		}

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

		return ret;
	}

	/*
	 * one reference for us, and we leave it for the
	 * caller
	 */
3533
	device->flush_bio = NULL;
3534
	bio = btrfs_io_bio_alloc(GFP_NOFS, 0);
C
Chris Mason 已提交
3535 3536 3537 3538 3539
	if (!bio)
		return -ENOMEM;

	bio->bi_end_io = btrfs_end_empty_barrier;
	bio->bi_bdev = device->bdev;
M
Mike Christie 已提交
3540
	bio_set_op_attrs(bio, REQ_OP_WRITE, WRITE_FLUSH);
C
Chris Mason 已提交
3541 3542 3543 3544 3545
	init_completion(&device->flush_wait);
	bio->bi_private = &device->flush_wait;
	device->flush_bio = bio;

	bio_get(bio);
3546
	btrfsic_submit_bio(bio);
C
Chris Mason 已提交
3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558

	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;
3559 3560
	int errors_send = 0;
	int errors_wait = 0;
C
Chris Mason 已提交
3561 3562 3563 3564 3565
	int ret;

	/* send down all the barriers */
	head = &info->fs_devices->devices;
	list_for_each_entry_rcu(dev, head, dev_list) {
3566 3567
		if (dev->missing)
			continue;
C
Chris Mason 已提交
3568
		if (!dev->bdev) {
3569
			errors_send++;
C
Chris Mason 已提交
3570 3571 3572 3573 3574 3575 3576
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

		ret = write_dev_flush(dev, 0);
		if (ret)
3577
			errors_send++;
C
Chris Mason 已提交
3578 3579 3580 3581
	}

	/* wait for all the barriers */
	list_for_each_entry_rcu(dev, head, dev_list) {
3582 3583
		if (dev->missing)
			continue;
C
Chris Mason 已提交
3584
		if (!dev->bdev) {
3585
			errors_wait++;
C
Chris Mason 已提交
3586 3587 3588 3589 3590 3591 3592
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

		ret = write_dev_flush(dev, 1);
		if (ret)
3593
			errors_wait++;
C
Chris Mason 已提交
3594
	}
3595 3596
	if (errors_send > info->num_tolerated_disk_barrier_failures ||
	    errors_wait > info->num_tolerated_disk_barrier_failures)
C
Chris Mason 已提交
3597 3598 3599 3600
		return -EIO;
	return 0;
}

3601 3602
int btrfs_get_num_tolerated_disk_barrier_failures(u64 flags)
{
3603 3604
	int raid_type;
	int min_tolerated = INT_MAX;
3605

3606 3607 3608 3609 3610
	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);
3611

3612 3613 3614 3615 3616 3617 3618 3619 3620
	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);
	}
3621

3622
	if (min_tolerated == INT_MAX) {
3623
		pr_warn("BTRFS: unknown raid flag: %llu", flags);
3624 3625 3626 3627
		min_tolerated = 0;
	}

	return min_tolerated;
3628 3629
}

3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643
int btrfs_calc_num_tolerated_disk_barrier_failures(
	struct btrfs_fs_info *fs_info)
{
	struct btrfs_ioctl_space_info space;
	struct btrfs_space_info *sinfo;
	u64 types[] = {BTRFS_BLOCK_GROUP_DATA,
		       BTRFS_BLOCK_GROUP_SYSTEM,
		       BTRFS_BLOCK_GROUP_METADATA,
		       BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA};
	int i;
	int c;
	int num_tolerated_disk_barrier_failures =
		(int)fs_info->fs_devices->num_devices;

3644
	for (i = 0; i < ARRAY_SIZE(types); i++) {
3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661
		struct btrfs_space_info *tmp;

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

		if (!sinfo)
			continue;

		down_read(&sinfo->groups_sem);
		for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
3662 3663 3664 3665 3666 3667 3668 3669 3670 3671
			u64 flags;

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

			btrfs_get_block_group_info(&sinfo->block_groups[c],
						   &space);
			if (space.total_bytes == 0 || space.used_bytes == 0)
				continue;
			flags = space.flags;
3672 3673 3674 3675 3676

			num_tolerated_disk_barrier_failures = min(
				num_tolerated_disk_barrier_failures,
				btrfs_get_num_tolerated_disk_barrier_failures(
					flags));
3677 3678 3679 3680 3681 3682 3683
		}
		up_read(&sinfo->groups_sem);
	}

	return num_tolerated_disk_barrier_failures;
}

3684
static int write_all_supers(struct btrfs_fs_info *fs_info, int max_mirrors)
3685
{
3686
	struct list_head *head;
3687
	struct btrfs_device *dev;
3688
	struct btrfs_super_block *sb;
3689 3690 3691
	struct btrfs_dev_item *dev_item;
	int ret;
	int do_barriers;
3692 3693
	int max_errors;
	int total_errors = 0;
3694
	u64 flags;
3695

3696 3697
	do_barriers = !btrfs_test_opt(fs_info, NOBARRIER);
	backup_super_roots(fs_info);
3698

3699
	sb = fs_info->super_for_commit;
3700
	dev_item = &sb->dev_item;
3701

3702 3703 3704
	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
3705

3706
	if (do_barriers) {
3707
		ret = barrier_all_devices(fs_info);
3708 3709
		if (ret) {
			mutex_unlock(
3710 3711 3712
				&fs_info->fs_devices->device_list_mutex);
			btrfs_handle_fs_error(fs_info, ret,
					      "errors while submitting device barriers.");
3713 3714 3715
			return ret;
		}
	}
C
Chris Mason 已提交
3716

3717
	list_for_each_entry_rcu(dev, head, dev_list) {
3718 3719 3720 3721
		if (!dev->bdev) {
			total_errors++;
			continue;
		}
Y
Yan Zheng 已提交
3722
		if (!dev->in_fs_metadata || !dev->writeable)
3723 3724
			continue;

Y
Yan Zheng 已提交
3725
		btrfs_set_stack_device_generation(dev_item, 0);
3726 3727
		btrfs_set_stack_device_type(dev_item, dev->type);
		btrfs_set_stack_device_id(dev_item, dev->devid);
3728
		btrfs_set_stack_device_total_bytes(dev_item,
3729
						   dev->commit_total_bytes);
3730 3731
		btrfs_set_stack_device_bytes_used(dev_item,
						  dev->commit_bytes_used);
3732 3733 3734 3735
		btrfs_set_stack_device_io_align(dev_item, dev->io_align);
		btrfs_set_stack_device_io_width(dev_item, dev->io_width);
		btrfs_set_stack_device_sector_size(dev_item, dev->sector_size);
		memcpy(dev_item->uuid, dev->uuid, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3736
		memcpy(dev_item->fsid, dev->fs_devices->fsid, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3737

3738 3739 3740
		flags = btrfs_super_flags(sb);
		btrfs_set_super_flags(sb, flags | BTRFS_HEADER_FLAG_WRITTEN);

Y
Yan Zheng 已提交
3741
		ret = write_dev_supers(dev, sb, do_barriers, 0, max_mirrors);
3742 3743
		if (ret)
			total_errors++;
3744
	}
3745
	if (total_errors > max_errors) {
3746 3747 3748
		btrfs_err(fs_info, "%d errors while writing supers",
			  total_errors);
		mutex_unlock(&fs_info->fs_devices->device_list_mutex);
3749

3750
		/* FUA is masked off if unsupported and can't be the reason */
3751 3752 3753
		btrfs_handle_fs_error(fs_info, -EIO,
				      "%d errors while writing supers",
				      total_errors);
3754
		return -EIO;
3755
	}
3756

Y
Yan Zheng 已提交
3757
	total_errors = 0;
3758
	list_for_each_entry_rcu(dev, head, dev_list) {
3759 3760
		if (!dev->bdev)
			continue;
Y
Yan Zheng 已提交
3761
		if (!dev->in_fs_metadata || !dev->writeable)
3762 3763
			continue;

Y
Yan Zheng 已提交
3764 3765 3766
		ret = write_dev_supers(dev, sb, do_barriers, 1, max_mirrors);
		if (ret)
			total_errors++;
3767
	}
3768
	mutex_unlock(&fs_info->fs_devices->device_list_mutex);
3769
	if (total_errors > max_errors) {
3770 3771 3772
		btrfs_handle_fs_error(fs_info, -EIO,
				      "%d errors while writing supers",
				      total_errors);
3773
		return -EIO;
3774
	}
3775 3776 3777
	return 0;
}

Y
Yan Zheng 已提交
3778
int write_ctree_super(struct btrfs_trans_handle *trans,
3779
		      struct btrfs_fs_info *fs_info, int max_mirrors)
3780
{
3781
	return write_all_supers(fs_info, max_mirrors);
C
Chris Mason 已提交
3782 3783
}

3784 3785 3786
/* 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
3787
{
3788
	spin_lock(&fs_info->fs_roots_radix_lock);
C
Chris Mason 已提交
3789 3790
	radix_tree_delete(&fs_info->fs_roots_radix,
			  (unsigned long)root->root_key.objectid);
3791
	spin_unlock(&fs_info->fs_roots_radix_lock);
3792 3793 3794 3795

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

L
Liu Bo 已提交
3796
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
L
Liu Bo 已提交
3797
		btrfs_free_log(NULL, root);
L
Liu Bo 已提交
3798 3799 3800 3801 3802 3803 3804
		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
3805

3806 3807 3808 3809
	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);
3810 3811 3812 3813 3814
	free_fs_root(root);
}

static void free_fs_root(struct btrfs_root *root)
{
3815
	iput(root->ino_cache_inode);
3816
	WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree));
3817
	btrfs_free_block_rsv(root->fs_info, root->orphan_block_rsv);
3818
	root->orphan_block_rsv = NULL;
3819 3820
	if (root->anon_dev)
		free_anon_bdev(root->anon_dev);
3821 3822
	if (root->subv_writers)
		btrfs_free_subvolume_writers(root->subv_writers);
3823 3824
	free_extent_buffer(root->node);
	free_extent_buffer(root->commit_root);
3825 3826
	kfree(root->free_ino_ctl);
	kfree(root->free_ino_pinned);
C
Chris Mason 已提交
3827
	kfree(root->name);
3828
	btrfs_put_fs_root(root);
C
Chris Mason 已提交
3829 3830
}

3831 3832 3833
void btrfs_free_fs_root(struct btrfs_root *root)
{
	free_fs_root(root);
C
Chris Mason 已提交
3834 3835
}

Y
Yan Zheng 已提交
3836
int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
C
Chris Mason 已提交
3837
{
Y
Yan Zheng 已提交
3838 3839
	u64 root_objectid = 0;
	struct btrfs_root *gang[8];
3840 3841 3842 3843
	int i = 0;
	int err = 0;
	unsigned int ret = 0;
	int index;
3844

Y
Yan Zheng 已提交
3845
	while (1) {
3846
		index = srcu_read_lock(&fs_info->subvol_srcu);
Y
Yan Zheng 已提交
3847 3848 3849
		ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix,
					     (void **)gang, root_objectid,
					     ARRAY_SIZE(gang));
3850 3851
		if (!ret) {
			srcu_read_unlock(&fs_info->subvol_srcu, index);
Y
Yan Zheng 已提交
3852
			break;
3853
		}
3854
		root_objectid = gang[ret - 1]->root_key.objectid + 1;
3855

Y
Yan Zheng 已提交
3856
		for (i = 0; i < ret; i++) {
3857 3858 3859 3860 3861 3862 3863 3864 3865
			/* 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);
3866

3867 3868 3869
		for (i = 0; i < ret; i++) {
			if (!gang[i])
				continue;
Y
Yan Zheng 已提交
3870
			root_objectid = gang[i]->root_key.objectid;
3871 3872
			err = btrfs_orphan_cleanup(gang[i]);
			if (err)
3873 3874
				break;
			btrfs_put_fs_root(gang[i]);
Y
Yan Zheng 已提交
3875 3876 3877
		}
		root_objectid++;
	}
3878 3879 3880 3881 3882 3883 3884

	/* 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
3885
}
3886

3887
int btrfs_commit_super(struct btrfs_fs_info *fs_info)
Y
Yan Zheng 已提交
3888
{
3889
	struct btrfs_root *root = fs_info->tree_root;
Y
Yan Zheng 已提交
3890
	struct btrfs_trans_handle *trans;
3891

3892
	mutex_lock(&fs_info->cleaner_mutex);
3893
	btrfs_run_delayed_iputs(fs_info);
3894 3895
	mutex_unlock(&fs_info->cleaner_mutex);
	wake_up_process(fs_info->cleaner_kthread);
3896 3897

	/* wait until ongoing cleanup work done */
3898 3899
	down_write(&fs_info->cleanup_work_sem);
	up_write(&fs_info->cleanup_work_sem);
3900

3901
	trans = btrfs_join_transaction(root);
3902 3903
	if (IS_ERR(trans))
		return PTR_ERR(trans);
3904
	return btrfs_commit_transaction(trans);
Y
Yan Zheng 已提交
3905 3906
}

3907
void close_ctree(struct btrfs_fs_info *fs_info)
Y
Yan Zheng 已提交
3908
{
3909
	struct btrfs_root *root = fs_info->tree_root;
Y
Yan Zheng 已提交
3910 3911
	int ret;

3912
	set_bit(BTRFS_FS_CLOSING_START, &fs_info->flags);
Y
Yan Zheng 已提交
3913

3914
	/* wait for the qgroup rescan worker to stop */
3915
	btrfs_qgroup_wait_for_completion(fs_info, false);
3916

S
Stefan Behrens 已提交
3917 3918 3919 3920 3921
	/* 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);

3922
	/* pause restriper - we want to resume on mount */
3923
	btrfs_pause_balance(fs_info);
3924

3925 3926
	btrfs_dev_replace_suspend_for_unmount(fs_info);

3927
	btrfs_scrub_cancel(fs_info);
C
Chris Mason 已提交
3928 3929 3930 3931 3932 3933

	/* 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 */
3934
	btrfs_cleanup_defrag_inodes(fs_info);
C
Chris Mason 已提交
3935

3936 3937
	cancel_work_sync(&fs_info->async_reclaim_work);

Y
Yan Zheng 已提交
3938
	if (!(fs_info->sb->s_flags & MS_RDONLY)) {
3939 3940 3941 3942 3943
		/*
		 * 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.
		 */
3944
		btrfs_delete_unused_bgs(fs_info);
3945

3946
		ret = btrfs_commit_super(fs_info);
L
liubo 已提交
3947
		if (ret)
3948
			btrfs_err(fs_info, "commit super ret %d", ret);
L
liubo 已提交
3949 3950
	}

3951
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state))
3952
		btrfs_error_commit_super(fs_info);
3953

A
Al Viro 已提交
3954 3955
	kthread_stop(fs_info->transaction_kthread);
	kthread_stop(fs_info->cleaner_kthread);
3956

3957
	set_bit(BTRFS_FS_CLOSING_DONE, &fs_info->flags);
3958

3959
	btrfs_free_qgroup_config(fs_info);
3960

3961
	if (percpu_counter_sum(&fs_info->delalloc_bytes)) {
3962
		btrfs_info(fs_info, "at unmount delalloc count %lld",
3963
		       percpu_counter_sum(&fs_info->delalloc_bytes));
C
Chris Mason 已提交
3964
	}
3965

3966
	btrfs_sysfs_remove_mounted(fs_info);
3967
	btrfs_sysfs_remove_fsid(fs_info->fs_devices);
3968

3969
	btrfs_free_fs_roots(fs_info);
3970

3971 3972
	btrfs_put_block_group_cache(fs_info);

3973 3974
	btrfs_free_block_groups(fs_info);

3975 3976 3977 3978 3979
	/*
	 * 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);
3980 3981
	btrfs_stop_all_workers(fs_info);

3982
	clear_bit(BTRFS_FS_OPEN, &fs_info->flags);
3983
	free_root_pointers(fs_info, 1);
3984

3985
	iput(fs_info->btree_inode);
3986

3987
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
3988
	if (btrfs_test_opt(fs_info, CHECK_INTEGRITY))
3989
		btrfsic_unmount(fs_info->fs_devices);
3990 3991
#endif

3992
	btrfs_close_devices(fs_info->fs_devices);
3993
	btrfs_mapping_tree_free(&fs_info->mapping_tree);
3994

3995
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
3996
	percpu_counter_destroy(&fs_info->delalloc_bytes);
3997
	percpu_counter_destroy(&fs_info->bio_counter);
C
Chris Mason 已提交
3998
	bdi_destroy(&fs_info->bdi);
3999
	cleanup_srcu_struct(&fs_info->subvol_srcu);
4000

D
David Woodhouse 已提交
4001 4002
	btrfs_free_stripe_hash_table(fs_info);

4003
	__btrfs_free_block_rsv(root->orphan_block_rsv);
4004
	root->orphan_block_rsv = NULL;
4005

4006
	lock_chunks(fs_info);
4007 4008 4009 4010 4011 4012 4013 4014
	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);
	}
4015
	unlock_chunks(fs_info);
4016 4017
}

4018 4019
int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid,
			  int atomic)
4020
{
4021
	int ret;
4022
	struct inode *btree_inode = buf->pages[0]->mapping->host;
4023

4024
	ret = extent_buffer_uptodate(buf);
4025 4026 4027 4028
	if (!ret)
		return ret;

	ret = verify_parent_transid(&BTRFS_I(btree_inode)->io_tree, buf,
4029 4030 4031
				    parent_transid, atomic);
	if (ret == -EAGAIN)
		return ret;
4032
	return !ret;
4033 4034 4035 4036
}

void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
{
4037
	struct btrfs_fs_info *fs_info;
4038
	struct btrfs_root *root;
4039
	u64 transid = btrfs_header_generation(buf);
4040
	int was_dirty;
4041

4042 4043 4044 4045 4046 4047 4048 4049 4050 4051
#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;
4052
	fs_info = root->fs_info;
4053
	btrfs_assert_tree_locked(buf);
4054
	if (transid != fs_info->generation)
J
Jeff Mahoney 已提交
4055
		WARN(1, KERN_CRIT "btrfs transid mismatch buffer %llu, found %llu running %llu\n",
4056
			buf->start, transid, fs_info->generation);
4057
	was_dirty = set_extent_buffer_dirty(buf);
4058
	if (!was_dirty)
4059
		__percpu_counter_add(&fs_info->dirty_metadata_bytes,
4060
				     buf->len,
4061
				     fs_info->dirty_metadata_batch);
4062 4063
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	if (btrfs_header_level(buf) == 0 && check_leaf(root, buf)) {
4064
		btrfs_print_leaf(fs_info, buf);
4065 4066 4067
		ASSERT(0);
	}
#endif
4068 4069
}

4070
static void __btrfs_btree_balance_dirty(struct btrfs_fs_info *fs_info,
4071
					int flush_delayed)
4072 4073 4074 4075 4076
{
	/*
	 * looks as though older kernels can get into trouble with
	 * this code, they end up stuck in balance_dirty_pages forever
	 */
4077
	int ret;
4078 4079 4080 4081

	if (current->flags & PF_MEMALLOC)
		return;

4082
	if (flush_delayed)
4083
		btrfs_balance_delayed_items(fs_info);
4084

4085
	ret = percpu_counter_compare(&fs_info->dirty_metadata_bytes,
4086 4087
				     BTRFS_DIRTY_METADATA_THRESH);
	if (ret > 0) {
4088
		balance_dirty_pages_ratelimited(fs_info->btree_inode->i_mapping);
4089 4090 4091
	}
}

4092
void btrfs_btree_balance_dirty(struct btrfs_fs_info *fs_info)
C
Chris Mason 已提交
4093
{
4094
	__btrfs_btree_balance_dirty(fs_info, 1);
4095
}
4096

4097
void btrfs_btree_balance_dirty_nodelay(struct btrfs_fs_info *fs_info)
4098
{
4099
	__btrfs_btree_balance_dirty(fs_info, 0);
C
Chris Mason 已提交
4100
}
4101

4102
int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid)
4103
{
4104
	struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root;
4105 4106 4107
	struct btrfs_fs_info *fs_info = root->fs_info;

	return btree_read_extent_buffer_pages(fs_info, buf, parent_transid);
4108
}
4109

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

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

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

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

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

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

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

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

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

	return ret;
L
liubo 已提交
4256 4257
}

4258
static void btrfs_error_commit_super(struct btrfs_fs_info *fs_info)
L
liubo 已提交
4259
{
4260
	mutex_lock(&fs_info->cleaner_mutex);
4261
	btrfs_run_delayed_iputs(fs_info);
4262
	mutex_unlock(&fs_info->cleaner_mutex);
L
liubo 已提交
4263

4264 4265
	down_write(&fs_info->cleanup_work_sem);
	up_write(&fs_info->cleanup_work_sem);
L
liubo 已提交
4266 4267

	/* cleanup FS via transaction */
4268
	btrfs_cleanup_transaction(fs_info);
L
liubo 已提交
4269 4270
}

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

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

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

	INIT_LIST_HEAD(&splice);

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

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

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

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

	delayed_refs = &trans->delayed_refs;

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

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

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

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

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

	spin_unlock(&delayed_refs->lock);

	return ret;
}

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

	INIT_LIST_HEAD(&splice);

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

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

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

		btrfs_invalidate_inodes(btrfs_inode->root);
4400

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

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

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

	INIT_LIST_HEAD(&splice);

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

		btrfs_destroy_delalloc_inodes(root);
		btrfs_put_fs_root(root);

		spin_lock(&fs_info->delalloc_root_lock);
	}
	spin_unlock(&fs_info->delalloc_root_lock);
L
liubo 已提交
4430 4431
}

4432
static int btrfs_destroy_marked_extents(struct btrfs_fs_info *fs_info,
L
liubo 已提交
4433 4434 4435 4436 4437 4438 4439 4440 4441 4442
					struct extent_io_tree *dirty_pages,
					int mark)
{
	int ret;
	struct extent_buffer *eb;
	u64 start = 0;
	u64 end;

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

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

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

	return ret;
}

4465
static int btrfs_destroy_pinned_extent(struct btrfs_fs_info *fs_info,
L
liubo 已提交
4466 4467 4468 4469 4470 4471
				       struct extent_io_tree *pinned_extents)
{
	struct extent_io_tree *unpin;
	u64 start;
	u64 end;
	int ret;
4472
	bool loop = true;
L
liubo 已提交
4473 4474

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

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

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

L
liubo 已提交
4496 4497 4498
	return 0;
}

4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512 4513
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,
4514
			     struct btrfs_fs_info *fs_info)
4515 4516 4517 4518 4519 4520 4521 4522 4523
{
	struct btrfs_block_group_cache *cache;

	spin_lock(&cur_trans->dirty_bgs_lock);
	while (!list_empty(&cur_trans->dirty_bgs)) {
		cache = list_first_entry(&cur_trans->dirty_bgs,
					 struct btrfs_block_group_cache,
					 dirty_list);
		if (!cache) {
4524
			btrfs_err(fs_info, "orphan block group dirty_bgs list");
4525 4526 4527 4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551
			spin_unlock(&cur_trans->dirty_bgs_lock);
			return;
		}

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

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

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

	while (!list_empty(&cur_trans->io_bgs)) {
		cache = list_first_entry(&cur_trans->io_bgs,
					 struct btrfs_block_group_cache,
					 io_list);
		if (!cache) {
4552
			btrfs_err(fs_info, "orphan block group on io_bgs list");
4553 4554 4555 4556 4557 4558 4559 4560 4561 4562 4563
			return;
		}

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

4564
void btrfs_cleanup_one_transaction(struct btrfs_transaction *cur_trans,
4565
				   struct btrfs_fs_info *fs_info)
4566
{
4567
	btrfs_cleanup_dirty_bgs(cur_trans, fs_info);
4568 4569 4570
	ASSERT(list_empty(&cur_trans->dirty_bgs));
	ASSERT(list_empty(&cur_trans->io_bgs));

4571
	btrfs_destroy_delayed_refs(cur_trans, fs_info);
4572

4573
	cur_trans->state = TRANS_STATE_COMMIT_START;
4574
	wake_up(&fs_info->transaction_blocked_wait);
4575

4576
	cur_trans->state = TRANS_STATE_UNBLOCKED;
4577
	wake_up(&fs_info->transaction_wait);
4578

4579 4580
	btrfs_destroy_delayed_inodes(fs_info);
	btrfs_assert_delayed_root_empty(fs_info);
4581

4582
	btrfs_destroy_marked_extents(fs_info, &cur_trans->dirty_pages,
4583
				     EXTENT_DIRTY);
4584
	btrfs_destroy_pinned_extent(fs_info,
4585
				    fs_info->pinned_extents);
4586

4587 4588 4589
	cur_trans->state =TRANS_STATE_COMPLETED;
	wake_up(&cur_trans->commit_wait);

4590 4591 4592 4593 4594 4595
	/*
	memset(cur_trans, 0, sizeof(*cur_trans));
	kmem_cache_free(btrfs_transaction_cachep, cur_trans);
	*/
}

4596
static int btrfs_cleanup_transaction(struct btrfs_fs_info *fs_info)
L
liubo 已提交
4597 4598 4599
{
	struct btrfs_transaction *t;

4600
	mutex_lock(&fs_info->transaction_kthread_mutex);
L
liubo 已提交
4601

4602 4603 4604
	spin_lock(&fs_info->trans_lock);
	while (!list_empty(&fs_info->trans_list)) {
		t = list_first_entry(&fs_info->trans_list,
4605 4606 4607
				     struct btrfs_transaction, list);
		if (t->state >= TRANS_STATE_COMMIT_START) {
			atomic_inc(&t->use_count);
4608
			spin_unlock(&fs_info->trans_lock);
4609
			btrfs_wait_for_commit(fs_info, t->transid);
4610
			btrfs_put_transaction(t);
4611
			spin_lock(&fs_info->trans_lock);
4612 4613
			continue;
		}
4614
		if (t == fs_info->running_transaction) {
4615
			t->state = TRANS_STATE_COMMIT_DOING;
4616
			spin_unlock(&fs_info->trans_lock);
4617 4618 4619 4620 4621 4622 4623
			/*
			 * 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 {
4624
			spin_unlock(&fs_info->trans_lock);
4625
		}
4626
		btrfs_cleanup_one_transaction(t, fs_info);
4627

4628 4629 4630
		spin_lock(&fs_info->trans_lock);
		if (t == fs_info->running_transaction)
			fs_info->running_transaction = NULL;
L
liubo 已提交
4631
		list_del_init(&t->list);
4632
		spin_unlock(&fs_info->trans_lock);
L
liubo 已提交
4633

4634
		btrfs_put_transaction(t);
4635
		trace_btrfs_transaction_commit(fs_info->tree_root);
4636
		spin_lock(&fs_info->trans_lock);
4637
	}
4638 4639
	spin_unlock(&fs_info->trans_lock);
	btrfs_destroy_all_ordered_extents(fs_info);
4640 4641
	btrfs_destroy_delayed_inodes(fs_info);
	btrfs_assert_delayed_root_empty(fs_info);
4642
	btrfs_destroy_pinned_extent(fs_info, fs_info->pinned_extents);
4643 4644
	btrfs_destroy_all_delalloc_inodes(fs_info);
	mutex_unlock(&fs_info->transaction_kthread_mutex);
L
liubo 已提交
4645 4646 4647 4648

	return 0;
}

4649
static const struct extent_io_ops btree_extent_io_ops = {
4650
	.readpage_end_io_hook = btree_readpage_end_io_hook,
A
Arne Jansen 已提交
4651
	.readpage_io_failed_hook = btree_io_failed_hook,
4652
	.submit_bio_hook = btree_submit_bio_hook,
4653 4654
	/* note we're sharing with inode.c for the merge bio hook */
	.merge_bio_hook = btrfs_merge_bio_hook,
4655
};