disk-io.c 125.2 KB
Newer Older
C
Chris Mason 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
/*
 * Copyright (C) 2007 Oracle.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License v2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */

C
Chris Mason 已提交
19
#include <linux/fs.h>
20
#include <linux/blkdev.h>
C
Chris Mason 已提交
21
#include <linux/scatterlist.h>
C
Chris Mason 已提交
22
#include <linux/swap.h>
23
#include <linux/radix-tree.h>
C
Chris Mason 已提交
24
#include <linux/writeback.h>
C
Chris Mason 已提交
25
#include <linux/buffer_head.h>
26
#include <linux/workqueue.h>
27
#include <linux/kthread.h>
28
#include <linux/slab.h>
29
#include <linux/migrate.h>
30
#include <linux/ratelimit.h>
31
#include <linux/uuid.h>
S
Stefan Behrens 已提交
32
#include <linux/semaphore.h>
33
#include <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);
68
static void btrfs_destroy_ordered_extents(struct btrfs_root *root);
L
liubo 已提交
69
static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
70
				      struct btrfs_fs_info *fs_info);
71
static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root);
72
static int btrfs_destroy_marked_extents(struct btrfs_fs_info *fs_info,
L
liubo 已提交
73 74
					struct extent_io_tree *dirty_pages,
					int mark);
75
static int btrfs_destroy_pinned_extent(struct btrfs_fs_info *fs_info,
L
liubo 已提交
76
				       struct extent_io_tree *pinned_extents);
77 78
static int btrfs_cleanup_transaction(struct btrfs_fs_info *fs_info);
static void btrfs_error_commit_super(struct btrfs_fs_info *fs_info);
79

C
Chris Mason 已提交
80
/*
81 82
 * 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
83 84
 * by writes to insert metadata for new file extents after IO is complete.
 */
85
struct btrfs_end_io_wq {
86 87 88 89
	struct bio *bio;
	bio_end_io_t *end_io;
	void *private;
	struct btrfs_fs_info *info;
90
	blk_status_t status;
91
	enum btrfs_wq_endio_type metadata;
92
	struct btrfs_work work;
93
};
94

95 96 97 98 99 100 101
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,
102
					SLAB_MEM_SPREAD,
103 104 105 106 107 108 109 110
					NULL);
	if (!btrfs_end_io_wq_cache)
		return -ENOMEM;
	return 0;
}

void btrfs_end_io_wq_exit(void)
{
111
	kmem_cache_destroy(btrfs_end_io_wq_cache);
112 113
}

C
Chris Mason 已提交
114 115 116 117 118
/*
 * 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.
 */
119
struct async_submit_bio {
120 121
	void *private_data;
	struct btrfs_fs_info *fs_info;
122
	struct bio *bio;
C
Chris Mason 已提交
123 124
	extent_submit_bio_hook_t *submit_bio_start;
	extent_submit_bio_hook_t *submit_bio_done;
125
	int mirror_num;
C
Chris Mason 已提交
126
	unsigned long bio_flags;
127 128 129 130 131
	/*
	 * 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;
132
	struct btrfs_work work;
133
	blk_status_t status;
134 135
};

136 137 138 139 140 141 142 143 144 145
/*
 * 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.
146
 *
147 148 149
 * 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.
150
 *
151 152 153
 * 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.
154
 *
155 156 157
 * 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.
158 159 160 161 162
 */
#ifdef CONFIG_DEBUG_LOCK_ALLOC
# if BTRFS_MAX_LEVEL != 8
#  error
# endif
163 164 165 166 167 168 169 170 171 172 173 174 175

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"	},
176
	{ .id = BTRFS_QUOTA_TREE_OBJECTID,	.name_stem = "quota"	},
177 178 179
	{ .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"	},
180
	{ .id = BTRFS_UUID_TREE_OBJECTID,	.name_stem = "uuid"	},
181
	{ .id = BTRFS_FREE_SPACE_TREE_OBJECTID,	.name_stem = "free-space" },
182
	{ .id = 0,				.name_stem = "tree"	},
183
};
184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214

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]);
}

215 216
#endif

C
Chris Mason 已提交
217 218 219 220
/*
 * extents on the btree inode are pretty simple, there's one extent
 * that covers the entire device
 */
221
static struct extent_map *btree_get_extent(struct btrfs_inode *inode,
222
		struct page *page, size_t pg_offset, u64 start, u64 len,
223
		int create)
224
{
225 226
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->vfs_inode.i_sb);
	struct extent_map_tree *em_tree = &inode->extent_tree;
227 228 229
	struct extent_map *em;
	int ret;

230
	read_lock(&em_tree->lock);
231
	em = lookup_extent_mapping(em_tree, start, len);
232
	if (em) {
233
		em->bdev = fs_info->fs_devices->latest_bdev;
234
		read_unlock(&em_tree->lock);
235
		goto out;
236
	}
237
	read_unlock(&em_tree->lock);
238

239
	em = alloc_extent_map();
240 241 242 243 244
	if (!em) {
		em = ERR_PTR(-ENOMEM);
		goto out;
	}
	em->start = 0;
245
	em->len = (u64)-1;
C
Chris Mason 已提交
246
	em->block_len = (u64)-1;
247
	em->block_start = 0;
248
	em->bdev = fs_info->fs_devices->latest_bdev;
249

250
	write_lock(&em_tree->lock);
J
Josef Bacik 已提交
251
	ret = add_extent_mapping(em_tree, em, 0);
252 253
	if (ret == -EEXIST) {
		free_extent_map(em);
254
		em = lookup_extent_mapping(em_tree, start, len);
255
		if (!em)
256
			em = ERR_PTR(-EIO);
257
	} else if (ret) {
258
		free_extent_map(em);
259
		em = ERR_PTR(ret);
260
	}
261
	write_unlock(&em_tree->lock);
262

263 264
out:
	return em;
265 266
}

267
u32 btrfs_csum_data(const char *data, u32 seed, size_t len)
268
{
269
	return btrfs_crc32c(seed, data, len);
270 271
}

272
void btrfs_csum_final(u32 crc, u8 *result)
273
{
274
	put_unaligned_le32(~crc, result);
275 276
}

C
Chris Mason 已提交
277 278 279 280
/*
 * compute the csum for a btree block, and either verify it or write it
 * into the csum field of the block.
 */
281 282
static int csum_tree_block(struct btrfs_fs_info *fs_info,
			   struct extent_buffer *buf,
283 284
			   int verify)
{
285
	u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
286
	char *result = NULL;
287 288 289 290 291 292 293 294
	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;
295
	unsigned long inline_result;
296 297

	len = buf->len - offset;
C
Chris Mason 已提交
298
	while (len > 0) {
299
		err = map_private_extent_buffer(buf, offset, 32,
300
					&kaddr, &map_start, &map_len);
C
Chris Mason 已提交
301
		if (err)
302
			return err;
303
		cur_len = min(len, map_len - (offset - map_start));
304
		crc = btrfs_csum_data(kaddr + offset - map_start,
305 306 307 308
				      crc, cur_len);
		len -= cur_len;
		offset += cur_len;
	}
309
	if (csum_size > sizeof(inline_result)) {
310
		result = kzalloc(csum_size, GFP_NOFS);
311
		if (!result)
312
			return -ENOMEM;
313 314 315 316
	} else {
		result = (char *)&inline_result;
	}

317 318 319
	btrfs_csum_final(crc, result);

	if (verify) {
320
		if (memcmp_extent_buffer(buf, result, 0, csum_size)) {
321 322
			u32 val;
			u32 found = 0;
323
			memcpy(&found, result, csum_size);
324

325
			read_extent_buffer(buf, &val, 0, csum_size);
326
			btrfs_warn_rl(fs_info,
J
Jeff Mahoney 已提交
327
				"%s checksum verify failed on %llu wanted %X found %X level %d",
328
				fs_info->sb->s_id, buf->start,
329
				val, found, btrfs_header_level(buf));
330 331
			if (result != (char *)&inline_result)
				kfree(result);
332
			return -EUCLEAN;
333 334
		}
	} else {
335
		write_extent_buffer(buf, result, 0, csum_size);
336
	}
337 338
	if (result != (char *)&inline_result)
		kfree(result);
339 340 341
	return 0;
}

C
Chris Mason 已提交
342 343 344 345 346 347
/*
 * 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.
 */
348
static int verify_parent_transid(struct extent_io_tree *io_tree,
349 350
				 struct extent_buffer *eb, u64 parent_transid,
				 int atomic)
351
{
352
	struct extent_state *cached_state = NULL;
353
	int ret;
354
	bool need_lock = (current->journal_info == BTRFS_SEND_TRANS_STUB);
355 356 357 358

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

359 360 361
	if (atomic)
		return -EAGAIN;

362 363 364 365 366
	if (need_lock) {
		btrfs_tree_read_lock(eb);
		btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
	}

367
	lock_extent_bits(io_tree, eb->start, eb->start + eb->len - 1,
368
			 &cached_state);
369
	if (extent_buffer_uptodate(eb) &&
370 371 372 373
	    btrfs_header_generation(eb) == parent_transid) {
		ret = 0;
		goto out;
	}
374 375 376
	btrfs_err_rl(eb->fs_info,
		"parent transid verify failed on %llu wanted %llu found %llu",
			eb->start,
377
			parent_transid, btrfs_header_generation(eb));
378
	ret = 1;
379 380 381 382

	/*
	 * 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
383
	 * block that has been freed and re-allocated.  So don't clear uptodate
384 385 386 387 388 389
	 * 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
390
out:
391 392
	unlock_extent_cached(io_tree, eb->start, eb->start + eb->len - 1,
			     &cached_state, GFP_NOFS);
393 394
	if (need_lock)
		btrfs_tree_read_unlock_blocking(eb);
395 396 397
	return ret;
}

D
David Sterba 已提交
398 399 400 401
/*
 * Return 0 if the superblock checksum type matches the checksum value of that
 * algorithm. Pass the raw disk superblock data.
 */
402 403
static int btrfs_check_super_csum(struct btrfs_fs_info *fs_info,
				  char *raw_disk_sb)
D
David Sterba 已提交
404 405 406 407 408 409 410 411 412 413 414 415 416 417
{
	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
418
		 * is filled with zeros and is included in the checksum.
D
David Sterba 已提交
419 420 421 422 423 424 425 426 427 428
		 */
		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)) {
429
		btrfs_err(fs_info, "unsupported checksum algorithm %u",
D
David Sterba 已提交
430 431 432 433 434 435 436
				csum_type);
		ret = 1;
	}

	return ret;
}

C
Chris Mason 已提交
437 438 439 440
/*
 * helper to read a given tree block, doing retries as required when
 * the checksums don't match and we have alternate mirrors to try.
 */
441
static int btree_read_extent_buffer_pages(struct btrfs_fs_info *fs_info,
442
					  struct extent_buffer *eb,
443
					  u64 parent_transid)
444 445
{
	struct extent_io_tree *io_tree;
446
	int failed = 0;
447 448 449
	int ret;
	int num_copies = 0;
	int mirror_num = 0;
450
	int failed_mirror = 0;
451

452
	clear_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
453
	io_tree = &BTRFS_I(fs_info->btree_inode)->io_tree;
454
	while (1) {
455
		ret = read_extent_buffer_pages(io_tree, eb, WAIT_COMPLETE,
456
					       btree_get_extent, mirror_num);
457 458
		if (!ret) {
			if (!verify_parent_transid(io_tree, eb,
459
						   parent_transid, 0))
460 461 462 463
				break;
			else
				ret = -EIO;
		}
C
Chris Mason 已提交
464

465 466 467 468 469 470
		/*
		 * 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))
471 472
			break;

473
		num_copies = btrfs_num_copies(fs_info,
474
					      eb->start, eb->len);
C
Chris Mason 已提交
475
		if (num_copies == 1)
476
			break;
C
Chris Mason 已提交
477

478 479 480 481 482
		if (!failed_mirror) {
			failed = 1;
			failed_mirror = eb->read_mirror;
		}

483
		mirror_num++;
484 485 486
		if (mirror_num == failed_mirror)
			mirror_num++;

C
Chris Mason 已提交
487
		if (mirror_num > num_copies)
488
			break;
489
	}
490

491
	if (failed && !ret && failed_mirror)
492
		repair_eb_io_failure(fs_info, eb, failed_mirror);
493 494

	return ret;
495
}
496

C
Chris Mason 已提交
497
/*
C
Chris Mason 已提交
498 499
 * 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
500
 */
C
Chris Mason 已提交
501

502
static int csum_dirty_buffer(struct btrfs_fs_info *fs_info, struct page *page)
503
{
M
Miao Xie 已提交
504
	u64 start = page_offset(page);
505 506
	u64 found_start;
	struct extent_buffer *eb;
507

J
Josef Bacik 已提交
508 509 510
	eb = (struct extent_buffer *)page->private;
	if (page != eb->pages[0])
		return 0;
511

512
	found_start = btrfs_header_bytenr(eb);
513 514 515 516 517 518 519 520 521 522 523 524
	/*
	 * 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);

525
	return csum_tree_block(fs_info, eb, 0);
526 527
}

528
static int check_tree_block_fsid(struct btrfs_fs_info *fs_info,
Y
Yan Zheng 已提交
529 530
				 struct extent_buffer *eb)
{
531
	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
532
	u8 fsid[BTRFS_FSID_SIZE];
Y
Yan Zheng 已提交
533 534
	int ret = 1;

535
	read_extent_buffer(eb, fsid, btrfs_header_fsid(), BTRFS_FSID_SIZE);
Y
Yan Zheng 已提交
536 537 538 539 540 541 542 543 544 545
	while (fs_devices) {
		if (!memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE)) {
			ret = 0;
			break;
		}
		fs_devices = fs_devices->seed;
	}
	return ret;
}

546 547 548 549
#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",	\
550
		   reason, btrfs_header_bytenr(eb), root->objectid, slot)
551 552 553 554

static noinline int check_leaf(struct btrfs_root *root,
			       struct extent_buffer *leaf)
{
555
	struct btrfs_fs_info *fs_info = root->fs_info;
556 557 558 559 560
	struct btrfs_key key;
	struct btrfs_key leaf_key;
	u32 nritems = btrfs_header_nritems(leaf);
	int slot;

561 562 563 564 565 566 567 568 569
	/*
	 * Extent buffers from a relocation tree have a owner field that
	 * corresponds to the subvolume tree they are based on. So just from an
	 * extent buffer alone we can not find out what is the id of the
	 * corresponding subvolume tree, so we can not figure out if the extent
	 * buffer corresponds to the root of the relocation tree or not. So skip
	 * this check for relocation trees.
	 */
	if (nritems == 0 && !btrfs_header_flag(leaf, BTRFS_HEADER_FLAG_RELOC)) {
570 571 572 573 574 575
		struct btrfs_root *check_root;

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

576
		check_root = btrfs_get_fs_root(fs_info, &key, false);
577 578 579 580 581
		/*
		 * 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)) {
582 583 584
			struct extent_buffer *eb;

			eb = btrfs_root_node(check_root);
585
			/* if leaf is the root, then it's fine */
586
			if (leaf != eb) {
587
				CORRUPT("non-root leaf's nritems is 0",
588 589
					leaf, check_root, 0);
				free_extent_buffer(eb);
590 591
				return -EIO;
			}
592
			free_extent_buffer(eb);
593
		}
594
		return 0;
595
	}
596

597 598 599
	if (nritems == 0)
		return 0;

600 601
	/* Check the 0 item */
	if (btrfs_item_offset_nr(leaf, 0) + btrfs_item_size_nr(leaf, 0) !=
602
	    BTRFS_LEAF_DATA_SIZE(fs_info)) {
603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636
		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,
637
		 * just in case all the items are consistent to each other, but
638 639 640
		 * all point outside of the leaf.
		 */
		if (btrfs_item_end_nr(leaf, slot) >
641
		    BTRFS_LEAF_DATA_SIZE(fs_info)) {
642 643 644 645 646 647 648 649
			CORRUPT("slot end outside of leaf", leaf, root, slot);
			return -EIO;
		}
	}

	return 0;
}

L
Liu Bo 已提交
650 651 652
static int check_node(struct btrfs_root *root, struct extent_buffer *node)
{
	unsigned long nr = btrfs_header_nritems(node);
653 654 655 656
	struct btrfs_key key, next_key;
	int slot;
	u64 bytenr;
	int ret = 0;
L
Liu Bo 已提交
657

658
	if (nr == 0 || nr > BTRFS_NODEPTRS_PER_BLOCK(root->fs_info)) {
L
Liu Bo 已提交
659 660 661 662 663
		btrfs_crit(root->fs_info,
			   "corrupt node: block %llu root %llu nritems %lu",
			   node->start, root->objectid, nr);
		return -EIO;
	}
664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683

	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
684 685
}

686 687 688
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)
689 690 691 692 693
{
	u64 found_start;
	int found_level;
	struct extent_buffer *eb;
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
694
	struct btrfs_fs_info *fs_info = root->fs_info;
695
	int ret = 0;
696
	int reads_done;
697 698 699

	if (!page->private)
		goto out;
C
Chris Mason 已提交
700

J
Josef Bacik 已提交
701
	eb = (struct extent_buffer *)page->private;
C
Chris Mason 已提交
702

703 704 705 706 707 708
	/* 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);
709 710
	if (!reads_done)
		goto err;
711

712
	eb->read_mirror = mirror;
713
	if (test_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags)) {
714 715 716 717
		ret = -EIO;
		goto err;
	}

718
	found_start = btrfs_header_bytenr(eb);
719
	if (found_start != eb->start) {
720 721
		btrfs_err_rl(fs_info, "bad tree block start %llu %llu",
			     found_start, eb->start);
722
		ret = -EIO;
723 724
		goto err;
	}
725 726 727
	if (check_tree_block_fsid(fs_info, eb)) {
		btrfs_err_rl(fs_info, "bad fsid on block %llu",
			     eb->start);
728 729 730
		ret = -EIO;
		goto err;
	}
731
	found_level = btrfs_header_level(eb);
732
	if (found_level >= BTRFS_MAX_LEVEL) {
733 734
		btrfs_err(fs_info, "bad tree block level %d",
			  (int)btrfs_header_level(eb));
735 736 737
		ret = -EIO;
		goto err;
	}
738

739 740
	btrfs_set_buffer_lockdep_class(btrfs_header_owner(eb),
				       eb, found_level);
741

742
	ret = csum_tree_block(fs_info, eb, 1);
743
	if (ret)
744 745 746 747 748 749 750 751 752 753 754
		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;
	}
755

L
Liu Bo 已提交
756 757 758
	if (found_level > 0 && check_node(root, eb))
		ret = -EIO;

759 760
	if (!ret)
		set_extent_buffer_uptodate(eb);
761
err:
762 763
	if (reads_done &&
	    test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
764
		btree_readahead_hook(eb, ret);
A
Arne Jansen 已提交
765

D
David Woodhouse 已提交
766 767 768 769 770 771 772
	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);
773
		clear_extent_buffer_uptodate(eb);
D
David Woodhouse 已提交
774
	}
775
	free_extent_buffer(eb);
776
out:
777
	return ret;
778 779
}

780
static int btree_io_failed_hook(struct page *page, int failed_mirror)
A
Arne Jansen 已提交
781 782 783
{
	struct extent_buffer *eb;

J
Josef Bacik 已提交
784
	eb = (struct extent_buffer *)page->private;
785
	set_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags);
786
	eb->read_mirror = failed_mirror;
D
David Woodhouse 已提交
787
	atomic_dec(&eb->io_pages);
788
	if (test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
789
		btree_readahead_hook(eb, -EIO);
A
Arne Jansen 已提交
790 791 792
	return -EIO;	/* we fixed nothing */
}

793
static void end_workqueue_bio(struct bio *bio)
794
{
795
	struct btrfs_end_io_wq *end_io_wq = bio->bi_private;
796
	struct btrfs_fs_info *fs_info;
797 798
	struct btrfs_workqueue *wq;
	btrfs_work_func_t func;
799 800

	fs_info = end_io_wq->info;
801
	end_io_wq->status = bio->bi_status;
802

M
Mike Christie 已提交
803
	if (bio_op(bio) == REQ_OP_WRITE) {
804 805 806 807 808 809 810 811 812 813 814 815 816
		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;
		}
817
	} else {
818 819 820 821 822
		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) {
823 824 825 826 827 828 829 830 831
			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;
		}
832
	}
833 834 835

	btrfs_init_work(&end_io_wq->work, func, end_workqueue_fn, NULL, NULL);
	btrfs_queue_work(wq, &end_io_wq->work);
836 837
}

838
blk_status_t btrfs_bio_wq_end_io(struct btrfs_fs_info *info, struct bio *bio,
839
			enum btrfs_wq_endio_type metadata)
840
{
841
	struct btrfs_end_io_wq *end_io_wq;
842

843
	end_io_wq = kmem_cache_alloc(btrfs_end_io_wq_cache, GFP_NOFS);
844
	if (!end_io_wq)
845
		return BLK_STS_RESOURCE;
846 847 848

	end_io_wq->private = bio->bi_private;
	end_io_wq->end_io = bio->bi_end_io;
849
	end_io_wq->info = info;
850
	end_io_wq->status = 0;
851
	end_io_wq->bio = bio;
852
	end_io_wq->metadata = metadata;
853 854 855

	bio->bi_private = end_io_wq;
	bio->bi_end_io = end_workqueue_bio;
856 857 858
	return 0;
}

859
unsigned long btrfs_async_submit_limit(struct btrfs_fs_info *info)
860
{
861
	unsigned long limit = min_t(unsigned long,
862
				    info->thread_pool_size,
863 864 865
				    info->fs_devices->open_devices);
	return 256 * limit;
}
866

C
Chris Mason 已提交
867 868 869
static void run_one_async_start(struct btrfs_work *work)
{
	struct async_submit_bio *async;
870
	blk_status_t ret;
C
Chris Mason 已提交
871 872

	async = container_of(work, struct  async_submit_bio, work);
873
	ret = async->submit_bio_start(async->private_data, async->bio,
874 875 876
				      async->mirror_num, async->bio_flags,
				      async->bio_offset);
	if (ret)
877
		async->status = ret;
C
Chris Mason 已提交
878 879 880
}

static void run_one_async_done(struct btrfs_work *work)
881 882 883
{
	struct btrfs_fs_info *fs_info;
	struct async_submit_bio *async;
884
	int limit;
885 886

	async = container_of(work, struct  async_submit_bio, work);
887
	fs_info = async->fs_info;
888

889
	limit = btrfs_async_submit_limit(fs_info);
890 891
	limit = limit * 2 / 3;

892 893 894
	/*
	 * atomic_dec_return implies a barrier for waitqueue_active
	 */
895
	if (atomic_dec_return(&fs_info->nr_async_submits) < limit &&
896
	    waitqueue_active(&fs_info->async_submit_wait))
897 898
		wake_up(&fs_info->async_submit_wait);

899
	/* If an error occurred we just want to clean up the bio and move on */
900 901
	if (async->status) {
		async->bio->bi_status = async->status;
902
		bio_endio(async->bio);
903 904 905
		return;
	}

906
	async->submit_bio_done(async->private_data, async->bio, async->mirror_num,
907
			       async->bio_flags, async->bio_offset);
C
Chris Mason 已提交
908 909 910 911 912 913 914
}

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

	async = container_of(work, struct  async_submit_bio, work);
915 916 917
	kfree(async);
}

918 919 920 921 922
blk_status_t btrfs_wq_submit_bio(struct btrfs_fs_info *fs_info, struct bio *bio,
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset, void *private_data,
				 extent_submit_bio_hook_t *submit_bio_start,
				 extent_submit_bio_hook_t *submit_bio_done)
923 924 925 926 927
{
	struct async_submit_bio *async;

	async = kmalloc(sizeof(*async), GFP_NOFS);
	if (!async)
928
		return BLK_STS_RESOURCE;
929

930 931
	async->private_data = private_data;
	async->fs_info = fs_info;
932 933
	async->bio = bio;
	async->mirror_num = mirror_num;
C
Chris Mason 已提交
934 935 936
	async->submit_bio_start = submit_bio_start;
	async->submit_bio_done = submit_bio_done;

937
	btrfs_init_work(&async->work, btrfs_worker_helper, run_one_async_start,
938
			run_one_async_done, run_one_async_free);
C
Chris Mason 已提交
939

C
Chris Mason 已提交
940
	async->bio_flags = bio_flags;
941
	async->bio_offset = bio_offset;
942

943
	async->status = 0;
944

945
	atomic_inc(&fs_info->nr_async_submits);
946

947
	if (op_is_sync(bio->bi_opf))
948
		btrfs_set_work_high_priority(&async->work);
949

950
	btrfs_queue_work(fs_info->workers, &async->work);
951

C
Chris Mason 已提交
952
	while (atomic_read(&fs_info->async_submit_draining) &&
953 954 955 956 957
	      atomic_read(&fs_info->nr_async_submits)) {
		wait_event(fs_info->async_submit_wait,
			   (atomic_read(&fs_info->nr_async_submits) == 0));
	}

958 959 960
	return 0;
}

961
static blk_status_t btree_csum_one_bio(struct bio *bio)
962
{
963
	struct bio_vec *bvec;
964
	struct btrfs_root *root;
965
	int i, ret = 0;
966

967
	ASSERT(!bio_flagged(bio, BIO_CLONED));
968
	bio_for_each_segment_all(bvec, bio, i) {
969
		root = BTRFS_I(bvec->bv_page->mapping->host)->root;
970
		ret = csum_dirty_buffer(root->fs_info, bvec->bv_page);
971 972
		if (ret)
			break;
973
	}
974

975
	return errno_to_blk_status(ret);
976 977
}

978 979 980
static blk_status_t __btree_submit_bio_start(void *private_data, struct bio *bio,
					     int mirror_num, unsigned long bio_flags,
					     u64 bio_offset)
981
{
982 983
	/*
	 * when we're called for a write, we're already in the async
984
	 * submission context.  Just jump into btrfs_map_bio
985
	 */
986
	return btree_csum_one_bio(bio);
C
Chris Mason 已提交
987
}
988

989 990 991
static blk_status_t __btree_submit_bio_done(void *private_data, struct bio *bio,
					    int mirror_num, unsigned long bio_flags,
					    u64 bio_offset)
C
Chris Mason 已提交
992
{
993
	struct inode *inode = private_data;
994
	blk_status_t ret;
995

996
	/*
C
Chris Mason 已提交
997 998
	 * when we're called for a write, we're already in the async
	 * submission context.  Just jump into btrfs_map_bio
999
	 */
1000
	ret = btrfs_map_bio(btrfs_sb(inode->i_sb), bio, mirror_num, 1);
1001
	if (ret) {
1002
		bio->bi_status = ret;
1003 1004
		bio_endio(bio);
	}
1005
	return ret;
1006 1007
}

1008
static int check_async_write(struct btrfs_inode *bi, unsigned long bio_flags)
1009
{
1010 1011
	if (atomic_read(&bi->sync_writers))
		return 0;
1012 1013 1014
	if (bio_flags & EXTENT_BIO_TREE_LOG)
		return 0;
#ifdef CONFIG_X86
1015
	if (static_cpu_has(X86_FEATURE_XMM4_2))
1016 1017 1018 1019 1020
		return 0;
#endif
	return 1;
}

1021 1022 1023
static blk_status_t btree_submit_bio_hook(void *private_data, struct bio *bio,
					  int mirror_num, unsigned long bio_flags,
					  u64 bio_offset)
1024
{
1025
	struct inode *inode = private_data;
1026
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
1027
	int async = check_async_write(BTRFS_I(inode), bio_flags);
1028
	blk_status_t ret;
1029

M
Mike Christie 已提交
1030
	if (bio_op(bio) != REQ_OP_WRITE) {
C
Chris Mason 已提交
1031 1032 1033 1034
		/*
		 * called for a read, do the setup so that checksum validation
		 * can happen in the async kernel threads
		 */
1035 1036
		ret = btrfs_bio_wq_end_io(fs_info, bio,
					  BTRFS_WQ_ENDIO_METADATA);
1037
		if (ret)
1038
			goto out_w_error;
1039
		ret = btrfs_map_bio(fs_info, bio, mirror_num, 0);
1040 1041 1042
	} else if (!async) {
		ret = btree_csum_one_bio(bio);
		if (ret)
1043
			goto out_w_error;
1044
		ret = btrfs_map_bio(fs_info, bio, mirror_num, 0);
1045 1046 1047 1048 1049
	} else {
		/*
		 * kthread helpers are used to submit writes so that
		 * checksumming can happen in parallel across all CPUs
		 */
1050 1051
		ret = btrfs_wq_submit_bio(fs_info, bio, mirror_num, 0,
					  bio_offset, private_data,
1052 1053
					  __btree_submit_bio_start,
					  __btree_submit_bio_done);
1054
	}
1055

1056 1057 1058 1059
	if (ret)
		goto out_w_error;
	return 0;

1060
out_w_error:
1061
	bio->bi_status = ret;
1062
	bio_endio(bio);
1063
	return ret;
1064 1065
}

J
Jan Beulich 已提交
1066
#ifdef CONFIG_MIGRATION
1067
static int btree_migratepage(struct address_space *mapping,
1068 1069
			struct page *newpage, struct page *page,
			enum migrate_mode mode)
1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083
{
	/*
	 * 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;
1084
	return migrate_page(mapping, newpage, page, mode);
1085
}
J
Jan Beulich 已提交
1086
#endif
1087

1088 1089 1090 1091

static int btree_writepages(struct address_space *mapping,
			    struct writeback_control *wbc)
{
1092 1093 1094
	struct btrfs_fs_info *fs_info;
	int ret;

1095
	if (wbc->sync_mode == WB_SYNC_NONE) {
1096 1097 1098 1099

		if (wbc->for_kupdate)
			return 0;

1100
		fs_info = BTRFS_I(mapping->host)->root->fs_info;
1101
		/* this is a bit racy, but that's ok */
1102 1103 1104
		ret = percpu_counter_compare(&fs_info->dirty_metadata_bytes,
					     BTRFS_DIRTY_METADATA_THRESH);
		if (ret < 0)
1105 1106
			return 0;
	}
1107
	return btree_write_cache_pages(mapping, wbc);
1108 1109
}

1110
static int btree_readpage(struct file *file, struct page *page)
1111
{
1112 1113
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
1114
	return extent_read_full_page(tree, page, btree_get_extent, 0);
1115
}
C
Chris Mason 已提交
1116

1117
static int btree_releasepage(struct page *page, gfp_t gfp_flags)
1118
{
1119
	if (PageWriteback(page) || PageDirty(page))
C
Chris Mason 已提交
1120
		return 0;
1121

1122
	return try_release_extent_buffer(page);
1123 1124
}

1125 1126
static void btree_invalidatepage(struct page *page, unsigned int offset,
				 unsigned int length)
1127
{
1128 1129
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
1130 1131
	extent_invalidatepage(tree, page, offset);
	btree_releasepage(page, GFP_NOFS);
1132
	if (PagePrivate(page)) {
1133 1134 1135
		btrfs_warn(BTRFS_I(page->mapping->host)->root->fs_info,
			   "page private not zero on page %llu",
			   (unsigned long long)page_offset(page));
1136 1137
		ClearPagePrivate(page);
		set_page_private(page, 0);
1138
		put_page(page);
1139
	}
1140 1141
}

1142 1143
static int btree_set_page_dirty(struct page *page)
{
1144
#ifdef DEBUG
1145 1146 1147 1148 1149 1150 1151 1152
	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);
1153
#endif
1154 1155 1156
	return __set_page_dirty_nobuffers(page);
}

1157
static const struct address_space_operations btree_aops = {
1158
	.readpage	= btree_readpage,
1159
	.writepages	= btree_writepages,
1160 1161
	.releasepage	= btree_releasepage,
	.invalidatepage = btree_invalidatepage,
1162
#ifdef CONFIG_MIGRATION
1163
	.migratepage	= btree_migratepage,
1164
#endif
1165
	.set_page_dirty = btree_set_page_dirty,
1166 1167
};

1168
void readahead_tree_block(struct btrfs_fs_info *fs_info, u64 bytenr)
C
Chris Mason 已提交
1169
{
1170
	struct extent_buffer *buf = NULL;
1171
	struct inode *btree_inode = fs_info->btree_inode;
C
Chris Mason 已提交
1172

1173
	buf = btrfs_find_create_tree_block(fs_info, bytenr);
1174
	if (IS_ERR(buf))
1175
		return;
1176
	read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree,
1177
				 buf, WAIT_NONE, btree_get_extent, 0);
1178
	free_extent_buffer(buf);
C
Chris Mason 已提交
1179 1180
}

1181
int reada_tree_block_flagged(struct btrfs_fs_info *fs_info, u64 bytenr,
1182 1183 1184
			 int mirror_num, struct extent_buffer **eb)
{
	struct extent_buffer *buf = NULL;
1185
	struct inode *btree_inode = fs_info->btree_inode;
1186 1187 1188
	struct extent_io_tree *io_tree = &BTRFS_I(btree_inode)->io_tree;
	int ret;

1189
	buf = btrfs_find_create_tree_block(fs_info, bytenr);
1190
	if (IS_ERR(buf))
1191 1192 1193 1194
		return 0;

	set_bit(EXTENT_BUFFER_READAHEAD, &buf->bflags);

1195
	ret = read_extent_buffer_pages(io_tree, buf, WAIT_PAGE_LOCK,
1196 1197 1198 1199 1200 1201 1202 1203 1204
				       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;
1205
	} else if (extent_buffer_uptodate(buf)) {
1206 1207 1208 1209 1210 1211 1212
		*eb = buf;
	} else {
		free_extent_buffer(buf);
	}
	return 0;
}

1213 1214 1215
struct extent_buffer *btrfs_find_create_tree_block(
						struct btrfs_fs_info *fs_info,
						u64 bytenr)
1216
{
1217 1218 1219
	if (btrfs_is_testing(fs_info))
		return alloc_test_extent_buffer(fs_info, bytenr);
	return alloc_extent_buffer(fs_info, bytenr);
1220 1221 1222
}


1223 1224
int btrfs_write_tree_block(struct extent_buffer *buf)
{
1225
	return filemap_fdatawrite_range(buf->pages[0]->mapping, buf->start,
1226
					buf->start + buf->len - 1);
1227 1228
}

1229
void btrfs_wait_tree_block_writeback(struct extent_buffer *buf)
1230
{
1231 1232
	filemap_fdatawait_range(buf->pages[0]->mapping,
			        buf->start, buf->start + buf->len - 1);
1233 1234
}

1235
struct extent_buffer *read_tree_block(struct btrfs_fs_info *fs_info, u64 bytenr,
1236
				      u64 parent_transid)
1237 1238 1239 1240
{
	struct extent_buffer *buf = NULL;
	int ret;

1241
	buf = btrfs_find_create_tree_block(fs_info, bytenr);
1242 1243
	if (IS_ERR(buf))
		return buf;
1244

1245
	ret = btree_read_extent_buffer_pages(fs_info, buf, parent_transid);
1246 1247
	if (ret) {
		free_extent_buffer(buf);
1248
		return ERR_PTR(ret);
1249
	}
1250
	return buf;
1251

1252 1253
}

1254
void clean_tree_block(struct btrfs_fs_info *fs_info,
1255
		      struct extent_buffer *buf)
1256
{
1257
	if (btrfs_header_generation(buf) ==
1258
	    fs_info->running_transaction->transid) {
1259
		btrfs_assert_tree_locked(buf);
1260

1261
		if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &buf->bflags)) {
1262 1263 1264
			percpu_counter_add_batch(&fs_info->dirty_metadata_bytes,
						 -buf->len,
						 fs_info->dirty_metadata_batch);
1265 1266 1267 1268
			/* ugh, clear_extent_buffer_dirty needs to lock the page */
			btrfs_set_lock_blocking(buf);
			clear_extent_buffer_dirty(buf);
		}
1269
	}
1270 1271
}

1272 1273 1274 1275 1276 1277 1278 1279 1280
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);

1281
	ret = percpu_counter_init(&writers->counter, 0, GFP_KERNEL);
1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297
	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);
}

1298
static void __setup_root(struct btrfs_root *root, struct btrfs_fs_info *fs_info,
1299
			 u64 objectid)
1300
{
1301
	bool dummy = test_bit(BTRFS_FS_STATE_DUMMY_FS_INFO, &fs_info->fs_state);
C
Chris Mason 已提交
1302
	root->node = NULL;
1303
	root->commit_root = NULL;
1304
	root->state = 0;
1305
	root->orphan_cleanup_state = 0;
1306

1307 1308
	root->objectid = objectid;
	root->last_trans = 0;
1309
	root->highest_objectid = 0;
1310
	root->nr_delalloc_inodes = 0;
1311
	root->nr_ordered_extents = 0;
1312
	root->name = NULL;
1313
	root->inode_tree = RB_ROOT;
1314
	INIT_RADIX_TREE(&root->delayed_nodes_tree, GFP_ATOMIC);
1315
	root->block_rsv = NULL;
1316
	root->orphan_block_rsv = NULL;
1317 1318

	INIT_LIST_HEAD(&root->dirty_list);
1319
	INIT_LIST_HEAD(&root->root_list);
1320 1321
	INIT_LIST_HEAD(&root->delalloc_inodes);
	INIT_LIST_HEAD(&root->delalloc_root);
1322 1323
	INIT_LIST_HEAD(&root->ordered_extents);
	INIT_LIST_HEAD(&root->ordered_root);
1324 1325
	INIT_LIST_HEAD(&root->logged_list[0]);
	INIT_LIST_HEAD(&root->logged_list[1]);
1326
	spin_lock_init(&root->orphan_lock);
1327
	spin_lock_init(&root->inode_lock);
1328
	spin_lock_init(&root->delalloc_lock);
1329
	spin_lock_init(&root->ordered_extent_lock);
1330
	spin_lock_init(&root->accounting_lock);
1331 1332
	spin_lock_init(&root->log_extents_lock[0]);
	spin_lock_init(&root->log_extents_lock[1]);
1333
	mutex_init(&root->objectid_mutex);
1334
	mutex_init(&root->log_mutex);
1335
	mutex_init(&root->ordered_extent_mutex);
1336
	mutex_init(&root->delalloc_mutex);
Y
Yan Zheng 已提交
1337 1338 1339
	init_waitqueue_head(&root->log_writer_wait);
	init_waitqueue_head(&root->log_commit_wait[0]);
	init_waitqueue_head(&root->log_commit_wait[1]);
1340 1341
	INIT_LIST_HEAD(&root->log_ctxs[0]);
	INIT_LIST_HEAD(&root->log_ctxs[1]);
Y
Yan Zheng 已提交
1342 1343 1344
	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
1345
	atomic_set(&root->log_batch, 0);
1346
	atomic_set(&root->orphan_inodes, 0);
1347
	refcount_set(&root->refs, 1);
1348
	atomic_set(&root->will_be_snapshotted, 0);
1349
	atomic64_set(&root->qgroup_meta_rsv, 0);
Y
Yan Zheng 已提交
1350
	root->log_transid = 0;
1351
	root->log_transid_committed = -1;
1352
	root->last_log_commit = 0;
1353
	if (!dummy)
1354
		extent_io_tree_init(&root->dirty_log_pages, NULL);
C
Chris Mason 已提交
1355

1356 1357
	memset(&root->root_key, 0, sizeof(root->root_key));
	memset(&root->root_item, 0, sizeof(root->root_item));
1358
	memset(&root->defrag_progress, 0, sizeof(root->defrag_progress));
1359
	if (!dummy)
1360 1361 1362
		root->defrag_trans_start = fs_info->generation;
	else
		root->defrag_trans_start = 0;
1363
	root->root_key.objectid = objectid;
1364
	root->anon_dev = 0;
1365

1366
	spin_lock_init(&root->root_item_lock);
1367 1368
}

1369 1370
static struct btrfs_root *btrfs_alloc_root(struct btrfs_fs_info *fs_info,
		gfp_t flags)
A
Al Viro 已提交
1371
{
1372
	struct btrfs_root *root = kzalloc(sizeof(*root), flags);
A
Al Viro 已提交
1373 1374 1375 1376 1377
	if (root)
		root->fs_info = fs_info;
	return root;
}

1378 1379
#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
/* Should only be used by the testing infrastructure */
1380
struct btrfs_root *btrfs_alloc_dummy_root(struct btrfs_fs_info *fs_info)
1381 1382 1383
{
	struct btrfs_root *root;

1384 1385 1386 1387
	if (!fs_info)
		return ERR_PTR(-EINVAL);

	root = btrfs_alloc_root(fs_info, GFP_KERNEL);
1388 1389
	if (!root)
		return ERR_PTR(-ENOMEM);
1390

1391
	/* We don't use the stripesize in selftest, set it as sectorsize */
1392
	__setup_root(root, fs_info, BTRFS_ROOT_TREE_OBJECTID);
1393
	root->alloc_bytenr = 0;
1394 1395 1396 1397 1398

	return root;
}
#endif

1399 1400 1401 1402 1403 1404 1405 1406 1407
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;
1408
	uuid_le uuid;
1409

1410
	root = btrfs_alloc_root(fs_info, GFP_KERNEL);
1411 1412 1413
	if (!root)
		return ERR_PTR(-ENOMEM);

1414
	__setup_root(root, fs_info, objectid);
1415 1416 1417 1418
	root->root_key.objectid = objectid;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = 0;

1419
	leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0);
1420 1421
	if (IS_ERR(leaf)) {
		ret = PTR_ERR(leaf);
1422
		leaf = NULL;
1423 1424 1425
		goto fail;
	}

1426
	memzero_extent_buffer(leaf, 0, sizeof(struct btrfs_header));
1427 1428 1429 1430 1431 1432
	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;

1433 1434
	write_extent_buffer_fsid(leaf, fs_info->fsid);
	write_extent_buffer_chunk_tree_uuid(leaf, fs_info->chunk_tree_uuid);
1435 1436 1437
	btrfs_mark_buffer_dirty(leaf);

	root->commit_root = btrfs_root_node(root);
1438
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
1439 1440 1441 1442 1443 1444 1445 1446 1447 1448

	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);
1449 1450
	uuid_le_gen(&uuid);
	memcpy(root->root_item.uuid, uuid.b, BTRFS_UUID_SIZE);
1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461
	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);

1462 1463
	return root;

1464
fail:
1465 1466
	if (leaf) {
		btrfs_tree_unlock(leaf);
1467
		free_extent_buffer(root->commit_root);
1468 1469 1470
		free_extent_buffer(leaf);
	}
	kfree(root);
1471

1472
	return ERR_PTR(ret);
1473 1474
}

Y
Yan Zheng 已提交
1475 1476
static struct btrfs_root *alloc_log_tree(struct btrfs_trans_handle *trans,
					 struct btrfs_fs_info *fs_info)
1477 1478
{
	struct btrfs_root *root;
Y
Yan Zheng 已提交
1479
	struct extent_buffer *leaf;
1480

1481
	root = btrfs_alloc_root(fs_info, GFP_NOFS);
1482
	if (!root)
Y
Yan Zheng 已提交
1483
		return ERR_PTR(-ENOMEM);
1484

1485
	__setup_root(root, fs_info, BTRFS_TREE_LOG_OBJECTID);
1486 1487 1488 1489

	root->root_key.objectid = BTRFS_TREE_LOG_OBJECTID;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = BTRFS_TREE_LOG_OBJECTID;
1490

Y
Yan Zheng 已提交
1491
	/*
1492 1493
	 * DON'T set REF_COWS for log trees
	 *
Y
Yan Zheng 已提交
1494 1495 1496 1497 1498
	 * 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).
	 */
1499

1500 1501
	leaf = btrfs_alloc_tree_block(trans, root, 0, BTRFS_TREE_LOG_OBJECTID,
			NULL, 0, 0, 0);
Y
Yan Zheng 已提交
1502 1503 1504 1505
	if (IS_ERR(leaf)) {
		kfree(root);
		return ERR_CAST(leaf);
	}
1506

1507
	memzero_extent_buffer(leaf, 0, sizeof(struct btrfs_header));
1508 1509 1510 1511
	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
1512
	root->node = leaf;
1513

1514
	write_extent_buffer_fsid(root->node, fs_info->fsid);
1515 1516
	btrfs_mark_buffer_dirty(root->node);
	btrfs_tree_unlock(root->node);
Y
Yan Zheng 已提交
1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535
	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)
{
1536
	struct btrfs_fs_info *fs_info = root->fs_info;
Y
Yan Zheng 已提交
1537 1538 1539
	struct btrfs_root *log_root;
	struct btrfs_inode_item *inode_item;

1540
	log_root = alloc_log_tree(trans, fs_info);
Y
Yan Zheng 已提交
1541 1542 1543 1544 1545 1546 1547
	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;
1548 1549 1550
	btrfs_set_stack_inode_generation(inode_item, 1);
	btrfs_set_stack_inode_size(inode_item, 3);
	btrfs_set_stack_inode_nlink(inode_item, 1);
1551
	btrfs_set_stack_inode_nbytes(inode_item,
1552
				     fs_info->nodesize);
1553
	btrfs_set_stack_inode_mode(inode_item, S_IFDIR | 0755);
Y
Yan Zheng 已提交
1554

1555
	btrfs_set_root_node(&log_root->root_item, log_root->node);
Y
Yan Zheng 已提交
1556 1557 1558 1559

	WARN_ON(root->log_root);
	root->log_root = log_root;
	root->log_transid = 0;
1560
	root->log_transid_committed = -1;
1561
	root->last_log_commit = 0;
1562 1563 1564
	return 0;
}

1565 1566
static struct btrfs_root *btrfs_read_tree_root(struct btrfs_root *tree_root,
					       struct btrfs_key *key)
1567 1568 1569
{
	struct btrfs_root *root;
	struct btrfs_fs_info *fs_info = tree_root->fs_info;
1570
	struct btrfs_path *path;
1571
	u64 generation;
1572
	int ret;
1573

1574 1575
	path = btrfs_alloc_path();
	if (!path)
1576
		return ERR_PTR(-ENOMEM);
1577

1578
	root = btrfs_alloc_root(fs_info, GFP_NOFS);
1579 1580 1581
	if (!root) {
		ret = -ENOMEM;
		goto alloc_fail;
1582 1583
	}

1584
	__setup_root(root, fs_info, key->objectid);
1585

1586 1587
	ret = btrfs_find_root(tree_root, key, path,
			      &root->root_item, &root->root_key);
1588
	if (ret) {
1589 1590
		if (ret > 0)
			ret = -ENOENT;
1591
		goto find_fail;
1592
	}
1593

1594
	generation = btrfs_root_generation(&root->root_item);
1595 1596
	root->node = read_tree_block(fs_info,
				     btrfs_root_bytenr(&root->root_item),
1597
				     generation);
1598 1599
	if (IS_ERR(root->node)) {
		ret = PTR_ERR(root->node);
1600 1601 1602
		goto find_fail;
	} else if (!btrfs_buffer_uptodate(root->node, generation, 0)) {
		ret = -EIO;
1603 1604
		free_extent_buffer(root->node);
		goto find_fail;
1605
	}
1606
	root->commit_root = btrfs_root_node(root);
1607
out:
1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627
	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) {
1628
		set_bit(BTRFS_ROOT_REF_COWS, &root->state);
1629 1630
		btrfs_check_and_init_root_item(&root->root_item);
	}
1631

1632 1633 1634
	return root;
}

1635 1636 1637
int btrfs_init_fs_root(struct btrfs_root *root)
{
	int ret;
1638
	struct btrfs_subvolume_writers *writers;
1639 1640 1641 1642 1643 1644 1645 1646 1647

	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;
	}

1648 1649 1650 1651 1652 1653 1654
	writers = btrfs_alloc_subvolume_writers();
	if (IS_ERR(writers)) {
		ret = PTR_ERR(writers);
		goto fail;
	}
	root->subv_writers = writers;

1655
	btrfs_init_free_ino_ctl(root);
1656 1657
	spin_lock_init(&root->ino_cache_lock);
	init_waitqueue_head(&root->ino_cache_wait);
1658 1659 1660

	ret = get_anon_bdev(&root->anon_dev);
	if (ret)
L
Liu Bo 已提交
1661
		goto fail;
1662 1663 1664 1665 1666 1667

	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
1668
		goto fail;
1669 1670 1671 1672 1673 1674
	}

	ASSERT(root->highest_objectid <= BTRFS_LAST_FREE_OBJECTID);

	mutex_unlock(&root->objectid_mutex);

1675 1676
	return 0;
fail:
L
Liu Bo 已提交
1677
	/* the caller is responsible to call free_fs_root */
1678 1679 1680
	return ret;
}

1681 1682
struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info,
					u64 root_id)
1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697
{
	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;

1698
	ret = radix_tree_preload(GFP_NOFS);
1699 1700 1701 1702 1703 1704 1705 1706
	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)
1707
		set_bit(BTRFS_ROOT_IN_RADIX, &root->state);
1708 1709 1710 1711 1712 1713
	spin_unlock(&fs_info->fs_roots_radix_lock);
	radix_tree_preload_end();

	return ret;
}

1714 1715 1716
struct btrfs_root *btrfs_get_fs_root(struct btrfs_fs_info *fs_info,
				     struct btrfs_key *location,
				     bool check_ref)
1717 1718
{
	struct btrfs_root *root;
1719
	struct btrfs_path *path;
1720
	struct btrfs_key key;
1721 1722
	int ret;

1723 1724 1725 1726
	if (location->objectid == BTRFS_ROOT_TREE_OBJECTID)
		return fs_info->tree_root;
	if (location->objectid == BTRFS_EXTENT_TREE_OBJECTID)
		return fs_info->extent_root;
1727 1728 1729 1730
	if (location->objectid == BTRFS_CHUNK_TREE_OBJECTID)
		return fs_info->chunk_root;
	if (location->objectid == BTRFS_DEV_TREE_OBJECTID)
		return fs_info->dev_root;
1731 1732
	if (location->objectid == BTRFS_CSUM_TREE_OBJECTID)
		return fs_info->csum_root;
1733 1734 1735
	if (location->objectid == BTRFS_QUOTA_TREE_OBJECTID)
		return fs_info->quota_root ? fs_info->quota_root :
					     ERR_PTR(-ENOENT);
1736 1737 1738
	if (location->objectid == BTRFS_UUID_TREE_OBJECTID)
		return fs_info->uuid_root ? fs_info->uuid_root :
					    ERR_PTR(-ENOENT);
1739 1740 1741
	if (location->objectid == BTRFS_FREE_SPACE_TREE_OBJECTID)
		return fs_info->free_space_root ? fs_info->free_space_root :
						  ERR_PTR(-ENOENT);
1742
again:
1743
	root = btrfs_lookup_fs_root(fs_info, location->objectid);
1744
	if (root) {
1745
		if (check_ref && btrfs_root_refs(&root->root_item) == 0)
1746
			return ERR_PTR(-ENOENT);
1747
		return root;
1748
	}
1749

1750
	root = btrfs_read_fs_root(fs_info->tree_root, location);
1751 1752
	if (IS_ERR(root))
		return root;
1753

1754
	if (check_ref && btrfs_root_refs(&root->root_item) == 0) {
1755
		ret = -ENOENT;
1756
		goto fail;
1757
	}
1758

1759
	ret = btrfs_init_fs_root(root);
1760 1761
	if (ret)
		goto fail;
1762

1763 1764 1765 1766 1767
	path = btrfs_alloc_path();
	if (!path) {
		ret = -ENOMEM;
		goto fail;
	}
1768 1769 1770 1771 1772
	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);
1773
	btrfs_free_path(path);
1774 1775 1776
	if (ret < 0)
		goto fail;
	if (ret == 0)
1777
		set_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &root->state);
1778

1779
	ret = btrfs_insert_fs_root(fs_info, root);
1780
	if (ret) {
1781 1782 1783 1784 1785
		if (ret == -EEXIST) {
			free_fs_root(root);
			goto again;
		}
		goto fail;
1786
	}
1787
	return root;
1788 1789 1790
fail:
	free_fs_root(root);
	return ERR_PTR(ret);
1791 1792
}

C
Chris Mason 已提交
1793 1794 1795 1796 1797 1798
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
1799

1800 1801
	rcu_read_lock();
	list_for_each_entry_rcu(device, &info->fs_devices->devices, dev_list) {
1802 1803
		if (!device->bdev)
			continue;
1804
		bdi = device->bdev->bd_bdi;
1805
		if (bdi_congested(bdi, bdi_bits)) {
C
Chris Mason 已提交
1806 1807 1808 1809
			ret = 1;
			break;
		}
	}
1810
	rcu_read_unlock();
C
Chris Mason 已提交
1811 1812 1813
	return ret;
}

1814 1815 1816 1817 1818
/*
 * 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)
1819 1820
{
	struct bio *bio;
1821
	struct btrfs_end_io_wq *end_io_wq;
1822

1823
	end_io_wq = container_of(work, struct btrfs_end_io_wq, work);
1824
	bio = end_io_wq->bio;
1825

1826
	bio->bi_status = end_io_wq->status;
1827 1828
	bio->bi_private = end_io_wq->private;
	bio->bi_end_io = end_io_wq->end_io;
1829
	kmem_cache_free(btrfs_end_io_wq_cache, end_io_wq);
1830
	bio_endio(bio);
1831 1832
}

1833 1834 1835
static int cleaner_kthread(void *arg)
{
	struct btrfs_root *root = arg;
1836
	struct btrfs_fs_info *fs_info = root->fs_info;
1837
	int again;
1838
	struct btrfs_trans_handle *trans;
1839 1840

	do {
1841
		again = 0;
1842

1843
		/* Make the cleaner go to sleep early. */
1844
		if (btrfs_need_cleaner_sleep(fs_info))
1845 1846
			goto sleep;

1847 1848 1849 1850
		/*
		 * Do not do anything if we might cause open_ctree() to block
		 * before we have finished mounting the filesystem.
		 */
1851
		if (!test_bit(BTRFS_FS_OPEN, &fs_info->flags))
1852 1853
			goto sleep;

1854
		if (!mutex_trylock(&fs_info->cleaner_mutex))
1855 1856
			goto sleep;

1857 1858 1859 1860
		/*
		 * Avoid the problem that we change the status of the fs
		 * during the above check and trylock.
		 */
1861
		if (btrfs_need_cleaner_sleep(fs_info)) {
1862
			mutex_unlock(&fs_info->cleaner_mutex);
1863
			goto sleep;
1864
		}
1865

1866
		mutex_lock(&fs_info->cleaner_delayed_iput_mutex);
1867
		btrfs_run_delayed_iputs(fs_info);
1868
		mutex_unlock(&fs_info->cleaner_delayed_iput_mutex);
1869

1870
		again = btrfs_clean_one_deleted_snapshot(root);
1871
		mutex_unlock(&fs_info->cleaner_mutex);
1872 1873

		/*
1874 1875
		 * The defragger has dealt with the R/O remount and umount,
		 * needn't do anything special here.
1876
		 */
1877
		btrfs_run_defrag_inodes(fs_info);
1878 1879 1880 1881 1882 1883 1884 1885 1886

		/*
		 * 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.
		 */
1887
		btrfs_delete_unused_bgs(fs_info);
1888
sleep:
1889
		if (!again) {
1890
			set_current_state(TASK_INTERRUPTIBLE);
1891 1892
			if (!kthread_should_stop())
				schedule();
1893 1894 1895
			__set_current_state(TASK_RUNNING);
		}
	} while (!kthread_should_stop());
1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910

	/*
	 * 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)
1911
			btrfs_err(fs_info,
1912 1913 1914 1915 1916
				  "cleaner transaction attach returned %ld",
				  PTR_ERR(trans));
	} else {
		int ret;

1917
		ret = btrfs_commit_transaction(trans);
1918
		if (ret)
1919
			btrfs_err(fs_info,
1920 1921 1922 1923
				  "cleaner open transaction commit returned %d",
				  ret);
	}

1924 1925 1926 1927 1928 1929
	return 0;
}

static int transaction_kthread(void *arg)
{
	struct btrfs_root *root = arg;
1930
	struct btrfs_fs_info *fs_info = root->fs_info;
1931 1932
	struct btrfs_trans_handle *trans;
	struct btrfs_transaction *cur;
1933
	u64 transid;
1934 1935
	unsigned long now;
	unsigned long delay;
1936
	bool cannot_commit;
1937 1938

	do {
1939
		cannot_commit = false;
1940 1941
		delay = HZ * fs_info->commit_interval;
		mutex_lock(&fs_info->transaction_kthread_mutex);
1942

1943 1944
		spin_lock(&fs_info->trans_lock);
		cur = fs_info->running_transaction;
1945
		if (!cur) {
1946
			spin_unlock(&fs_info->trans_lock);
1947 1948
			goto sleep;
		}
Y
Yan Zheng 已提交
1949

1950
		now = get_seconds();
1951
		if (cur->state < TRANS_STATE_BLOCKED &&
1952
		    (now < cur->start_time ||
1953 1954
		     now - cur->start_time < fs_info->commit_interval)) {
			spin_unlock(&fs_info->trans_lock);
1955 1956 1957
			delay = HZ * 5;
			goto sleep;
		}
1958
		transid = cur->transid;
1959
		spin_unlock(&fs_info->trans_lock);
1960

1961
		/* If the file system is aborted, this will always fail. */
1962
		trans = btrfs_attach_transaction(root);
1963
		if (IS_ERR(trans)) {
1964 1965
			if (PTR_ERR(trans) != -ENOENT)
				cannot_commit = true;
1966
			goto sleep;
1967
		}
1968
		if (transid == trans->transid) {
1969
			btrfs_commit_transaction(trans);
1970
		} else {
1971
			btrfs_end_transaction(trans);
1972
		}
1973
sleep:
1974 1975
		wake_up_process(fs_info->cleaner_kthread);
		mutex_unlock(&fs_info->transaction_kthread_mutex);
1976

J
Josef Bacik 已提交
1977
		if (unlikely(test_bit(BTRFS_FS_STATE_ERROR,
1978
				      &fs_info->fs_state)))
1979
			btrfs_cleanup_transaction(fs_info);
1980 1981
		set_current_state(TASK_INTERRUPTIBLE);
		if (!kthread_should_stop() &&
1982
				(!btrfs_transaction_blocked(fs_info) ||
1983 1984 1985
				 cannot_commit))
			schedule_timeout(delay);
		__set_current_state(TASK_RUNNING);
1986 1987 1988 1989
	} while (!kthread_should_stop());
	return 0;
}

C
Chris Mason 已提交
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 2095
/*
 * 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));

2096 2097 2098 2099 2100 2101 2102 2103
	/*
	 * 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
2104
			       btrfs_header_generation(info->fs_root->node));
2105
		btrfs_set_backup_fs_root_level(root_backup,
C
Chris Mason 已提交
2106
			       btrfs_header_level(info->fs_root->node));
2107
	}
C
Chris Mason 已提交
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 2188

	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
2189 2190 2191
/* helper to cleanup workers */
static void btrfs_stop_all_workers(struct btrfs_fs_info *fs_info)
{
2192
	btrfs_destroy_workqueue(fs_info->fixup_workers);
2193
	btrfs_destroy_workqueue(fs_info->delalloc_workers);
2194
	btrfs_destroy_workqueue(fs_info->workers);
2195 2196
	btrfs_destroy_workqueue(fs_info->endio_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
	btrfs_destroy_workqueue(fs_info->endio_write_workers);
	btrfs_destroy_workqueue(fs_info->endio_freespace_worker);
2201
	btrfs_destroy_workqueue(fs_info->submit_workers);
2202
	btrfs_destroy_workqueue(fs_info->delayed_workers);
2203
	btrfs_destroy_workqueue(fs_info->caching_workers);
2204
	btrfs_destroy_workqueue(fs_info->readahead_workers);
2205
	btrfs_destroy_workqueue(fs_info->flush_workers);
2206
	btrfs_destroy_workqueue(fs_info->qgroup_rescan_workers);
C
Chris Mason 已提交
2207
	btrfs_destroy_workqueue(fs_info->extent_workers);
2208 2209 2210 2211 2212 2213 2214
	/*
	 * Now that all other work queues are destroyed, we can safely destroy
	 * the queues used for metadata I/O, since tasks from those other work
	 * queues can do metadata I/O operations.
	 */
	btrfs_destroy_workqueue(fs_info->endio_meta_workers);
	btrfs_destroy_workqueue(fs_info->endio_meta_write_workers);
L
Liu Bo 已提交
2215 2216
}

2217 2218 2219 2220 2221 2222 2223 2224 2225 2226
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
2227 2228 2229
/* helper to cleanup tree roots */
static void free_root_pointers(struct btrfs_fs_info *info, int chunk_root)
{
2230
	free_root_extent_buffers(info->tree_root);
2231

2232 2233 2234 2235 2236 2237 2238
	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);
2239
	free_root_extent_buffers(info->free_space_root);
C
Chris Mason 已提交
2240 2241
}

2242
void btrfs_free_fs_roots(struct btrfs_fs_info *fs_info)
2243 2244 2245 2246 2247 2248 2249 2250 2251 2252
{
	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);

2253
		if (test_bit(BTRFS_ROOT_IN_RADIX, &gang[0]->state)) {
2254
			btrfs_drop_and_free_fs_root(fs_info, gang[0]);
2255 2256 2257
		} else {
			free_extent_buffer(gang[0]->node);
			free_extent_buffer(gang[0]->commit_root);
2258
			btrfs_put_fs_root(gang[0]);
2259 2260 2261 2262 2263 2264 2265 2266 2267 2268
		}
	}

	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++)
2269
			btrfs_drop_and_free_fs_root(fs_info, gang[i]);
2270
	}
2271 2272 2273

	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
		btrfs_free_log_root_tree(NULL, fs_info);
2274
		btrfs_destroy_pinned_extent(fs_info, fs_info->pinned_extents);
2275
	}
2276
}
C
Chris Mason 已提交
2277

2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288
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;
}

2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299
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);
}

2300
static void btrfs_init_btree_inode(struct btrfs_fs_info *fs_info)
2301
{
2302 2303 2304 2305
	struct inode *inode = fs_info->btree_inode;

	inode->i_ino = BTRFS_BTREE_INODE_OBJECTID;
	set_nlink(inode, 1);
2306 2307 2308 2309 2310
	/*
	 * 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
	 */
2311 2312
	inode->i_size = OFFSET_MAX;
	inode->i_mapping->a_ops = &btree_aops;
2313

2314
	RB_CLEAR_NODE(&BTRFS_I(inode)->rb_node);
2315
	extent_io_tree_init(&BTRFS_I(inode)->io_tree, inode);
2316 2317
	BTRFS_I(inode)->io_tree.track_uptodate = 0;
	extent_map_tree_init(&BTRFS_I(inode)->extent_tree);
2318

2319
	BTRFS_I(inode)->io_tree.ops = &btree_extent_io_ops;
2320

2321 2322 2323 2324
	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);
2325 2326
}

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

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

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

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

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

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

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

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

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

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

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

	return 0;
}

2441 2442 2443 2444 2445 2446 2447 2448 2449
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) {
2450
		btrfs_warn(fs_info, "log replay required on RO media");
2451 2452 2453
		return -EIO;
	}

2454
	log_tree_root = btrfs_alloc_root(fs_info, GFP_KERNEL);
2455 2456 2457
	if (!log_tree_root)
		return -ENOMEM;

2458
	__setup_root(log_tree_root, fs_info, BTRFS_TREE_LOG_OBJECTID);
2459

2460 2461
	log_tree_root->node = read_tree_block(fs_info, bytenr,
					      fs_info->generation + 1);
2462
	if (IS_ERR(log_tree_root->node)) {
2463
		btrfs_warn(fs_info, "failed to read log tree");
2464
		ret = PTR_ERR(log_tree_root->node);
2465
		kfree(log_tree_root);
2466
		return ret;
2467
	} else if (!extent_buffer_uptodate(log_tree_root->node)) {
2468
		btrfs_err(fs_info, "failed to read log tree");
2469 2470 2471 2472 2473 2474 2475
		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) {
2476 2477
		btrfs_handle_fs_error(fs_info, ret,
				      "Failed to recover log tree");
2478 2479 2480 2481 2482
		free_extent_buffer(log_tree_root->node);
		kfree(log_tree_root);
		return ret;
	}

2483
	if (sb_rdonly(fs_info->sb)) {
2484
		ret = btrfs_commit_super(fs_info);
2485 2486 2487 2488 2489 2490 2491
		if (ret)
			return ret;
	}

	return 0;
}

2492
static int btrfs_read_roots(struct btrfs_fs_info *fs_info)
2493
{
2494
	struct btrfs_root *tree_root = fs_info->tree_root;
2495
	struct btrfs_root *root;
2496 2497 2498
	struct btrfs_key location;
	int ret;

2499 2500
	BUG_ON(!fs_info->tree_root);

2501 2502 2503 2504
	location.objectid = BTRFS_EXTENT_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
	location.offset = 0;

2505 2506 2507 2508 2509
	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;
2510 2511

	location.objectid = BTRFS_DEV_TREE_OBJECTID;
2512 2513 2514 2515 2516
	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;
2517 2518 2519
	btrfs_init_devices_late(fs_info);

	location.objectid = BTRFS_CSUM_TREE_OBJECTID;
2520 2521 2522 2523 2524
	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;
2525 2526

	location.objectid = BTRFS_QUOTA_TREE_OBJECTID;
2527 2528 2529
	root = btrfs_read_tree_root(tree_root, &location);
	if (!IS_ERR(root)) {
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
2530
		set_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags);
2531
		fs_info->quota_root = root;
2532 2533 2534
	}

	location.objectid = BTRFS_UUID_TREE_OBJECTID;
2535 2536 2537
	root = btrfs_read_tree_root(tree_root, &location);
	if (IS_ERR(root)) {
		ret = PTR_ERR(root);
2538 2539 2540
		if (ret != -ENOENT)
			return ret;
	} else {
2541 2542
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
		fs_info->uuid_root = root;
2543 2544
	}

2545 2546 2547 2548 2549 2550 2551 2552 2553
	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;
	}

2554 2555 2556
	return 0;
}

A
Al Viro 已提交
2557 2558 2559
int open_ctree(struct super_block *sb,
	       struct btrfs_fs_devices *fs_devices,
	       char *options)
2560
{
2561 2562
	u32 sectorsize;
	u32 nodesize;
2563
	u32 stripesize;
2564
	u64 generation;
2565
	u64 features;
2566
	struct btrfs_key location;
2567
	struct buffer_head *bh;
2568
	struct btrfs_super_block *disk_super;
2569
	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
2570
	struct btrfs_root *tree_root;
2571
	struct btrfs_root *chunk_root;
2572
	int ret;
2573
	int err = -EINVAL;
C
Chris Mason 已提交
2574 2575
	int num_backups_tried = 0;
	int backup_index = 0;
2576
	int max_active;
2577
	int clear_free_space_tree = 0;
2578

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

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

2592
	ret = percpu_counter_init(&fs_info->dirty_metadata_bytes, 0, GFP_KERNEL);
2593 2594
	if (ret) {
		err = ret;
2595
		goto fail_srcu;
2596
	}
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);
J
Jan Schmidt 已提交
2632
	spin_lock_init(&fs_info->tree_mod_seq_lock);
2633
	spin_lock_init(&fs_info->super_lock);
J
Josef Bacik 已提交
2634
	spin_lock_init(&fs_info->qgroup_op_lock);
2635
	spin_lock_init(&fs_info->buffer_lock);
2636
	spin_lock_init(&fs_info->unused_bgs_lock);
J
Jan Schmidt 已提交
2637
	rwlock_init(&fs_info->tree_mod_log_lock);
2638
	mutex_init(&fs_info->unused_bg_unpin_mutex);
2639
	mutex_init(&fs_info->delete_unused_bgs_mutex);
C
Chris Mason 已提交
2640
	mutex_init(&fs_info->reloc_mutex);
2641
	mutex_init(&fs_info->delalloc_root_mutex);
2642
	mutex_init(&fs_info->cleaner_delayed_iput_mutex);
2643
	seqlock_init(&fs_info->profiles_lock);
2644

2645
	INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots);
2646
	INIT_LIST_HEAD(&fs_info->space_info);
J
Jan Schmidt 已提交
2647
	INIT_LIST_HEAD(&fs_info->tree_mod_seq_list);
2648
	INIT_LIST_HEAD(&fs_info->unused_bgs);
2649
	btrfs_mapping_init(&fs_info->mapping_tree);
2650 2651 2652 2653 2654 2655 2656 2657 2658
	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);
2659
	atomic_set(&fs_info->nr_async_submits, 0);
2660
	atomic_set(&fs_info->async_delalloc_pages, 0);
2661
	atomic_set(&fs_info->async_submit_draining, 0);
2662
	atomic_set(&fs_info->nr_async_bios, 0);
C
Chris Mason 已提交
2663
	atomic_set(&fs_info->defrag_running, 0);
J
Josef Bacik 已提交
2664
	atomic_set(&fs_info->qgroup_op_seq, 0);
Z
Zhao Lei 已提交
2665
	atomic_set(&fs_info->reada_works_cnt, 0);
2666
	atomic64_set(&fs_info->tree_mod_seq, 0);
C
Chris Mason 已提交
2667
	fs_info->sb = sb;
2668
	fs_info->max_inline = BTRFS_DEFAULT_MAX_INLINE;
J
Josef Bacik 已提交
2669
	fs_info->metadata_ratio = 0;
C
Chris Mason 已提交
2670
	fs_info->defrag_inodes = RB_ROOT;
2671
	atomic64_set(&fs_info->free_chunk_space, 0);
J
Jan Schmidt 已提交
2672
	fs_info->tree_mod_log = RB_ROOT;
2673
	fs_info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
2674
	fs_info->avg_delayed_ref_runtime = NSEC_PER_SEC >> 6; /* div by 64 */
2675
	/* readahead state */
2676
	INIT_RADIX_TREE(&fs_info->reada_tree, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
2677
	spin_lock_init(&fs_info->reada_lock);
C
Chris Mason 已提交
2678

2679 2680
	fs_info->thread_pool_size = min_t(unsigned long,
					  num_online_cpus() + 2, 8);
2681

2682 2683
	INIT_LIST_HEAD(&fs_info->ordered_roots);
	spin_lock_init(&fs_info->ordered_root_lock);
2684
	fs_info->delayed_root = kmalloc(sizeof(struct btrfs_delayed_root),
2685
					GFP_KERNEL);
2686 2687 2688 2689 2690
	if (!fs_info->delayed_root) {
		err = -ENOMEM;
		goto fail_iput;
	}
	btrfs_init_delayed_root(fs_info->delayed_root);
2691

2692
	btrfs_init_scrub(fs_info);
2693 2694 2695
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	fs_info->check_integrity_print_mask = 0;
#endif
2696
	btrfs_init_balance(fs_info);
2697
	btrfs_init_async_reclaim_work(&fs_info->async_reclaim_work);
A
Arne Jansen 已提交
2698

2699 2700
	sb->s_blocksize = BTRFS_BDEV_BLOCKSIZE;
	sb->s_blocksize_bits = blksize_bits(BTRFS_BDEV_BLOCKSIZE);
2701

2702
	btrfs_init_btree_inode(fs_info);
2703

J
Josef Bacik 已提交
2704
	spin_lock_init(&fs_info->block_group_cache_lock);
2705
	fs_info->block_group_cache_tree = RB_ROOT;
2706
	fs_info->first_logical_byte = (u64)-1;
J
Josef Bacik 已提交
2707

2708 2709
	extent_io_tree_init(&fs_info->freed_extents[0], NULL);
	extent_io_tree_init(&fs_info->freed_extents[1], NULL);
2710
	fs_info->pinned_extents = &fs_info->freed_extents[0];
2711
	set_bit(BTRFS_FS_BARRIER, &fs_info->flags);
C
Chris Mason 已提交
2712

2713
	mutex_init(&fs_info->ordered_operations_mutex);
2714
	mutex_init(&fs_info->tree_log_mutex);
2715
	mutex_init(&fs_info->chunk_mutex);
2716 2717
	mutex_init(&fs_info->transaction_kthread_mutex);
	mutex_init(&fs_info->cleaner_mutex);
2718
	mutex_init(&fs_info->volume_mutex);
2719
	mutex_init(&fs_info->ro_block_group_mutex);
2720
	init_rwsem(&fs_info->commit_root_sem);
2721
	init_rwsem(&fs_info->cleanup_work_sem);
2722
	init_rwsem(&fs_info->subvol_sem);
S
Stefan Behrens 已提交
2723
	sema_init(&fs_info->uuid_tree_rescan_sem, 1);
2724

2725
	btrfs_init_dev_replace_locks(fs_info);
2726
	btrfs_init_qgroup(fs_info);
2727

2728 2729 2730
	btrfs_init_free_cluster(&fs_info->meta_alloc_cluster);
	btrfs_init_free_cluster(&fs_info->data_alloc_cluster);

2731
	init_waitqueue_head(&fs_info->transaction_throttle);
2732
	init_waitqueue_head(&fs_info->transaction_wait);
S
Sage Weil 已提交
2733
	init_waitqueue_head(&fs_info->transaction_blocked_wait);
2734
	init_waitqueue_head(&fs_info->async_submit_wait);
2735

2736 2737
	INIT_LIST_HEAD(&fs_info->pinned_chunks);

2738 2739 2740 2741 2742
	/* 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
2743 2744
	ret = btrfs_alloc_stripe_hash_table(fs_info);
	if (ret) {
2745
		err = ret;
D
David Woodhouse 已提交
2746 2747 2748
		goto fail_alloc;
	}

2749
	__setup_root(tree_root, fs_info, BTRFS_ROOT_TREE_OBJECTID);
2750

2751
	invalidate_bdev(fs_devices->latest_bdev);
D
David Sterba 已提交
2752 2753 2754 2755

	/*
	 * Read super block and check the signature bytes only
	 */
Y
Yan Zheng 已提交
2756
	bh = btrfs_read_dev_super(fs_devices->latest_bdev);
2757 2758
	if (IS_ERR(bh)) {
		err = PTR_ERR(bh);
2759
		goto fail_alloc;
2760
	}
C
Chris Mason 已提交
2761

D
David Sterba 已提交
2762 2763 2764 2765
	/*
	 * We want to check superblock checksum, the type is stored inside.
	 * Pass the whole disk block of size BTRFS_SUPER_INFO_SIZE (4k).
	 */
2766
	if (btrfs_check_super_csum(fs_info, bh->b_data)) {
2767
		btrfs_err(fs_info, "superblock checksum mismatch");
D
David Sterba 已提交
2768
		err = -EINVAL;
2769
		brelse(bh);
D
David Sterba 已提交
2770 2771 2772 2773 2774 2775 2776 2777
		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
	 */
2778 2779 2780
	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));
2781
	brelse(bh);
2782

2783
	memcpy(fs_info->fsid, fs_info->super_copy->fsid, BTRFS_FSID_SIZE);
2784

2785
	ret = btrfs_check_super_valid(fs_info);
D
David Sterba 已提交
2786
	if (ret) {
2787
		btrfs_err(fs_info, "superblock contains fatal errors");
D
David Sterba 已提交
2788 2789 2790 2791
		err = -EINVAL;
		goto fail_alloc;
	}

2792
	disk_super = fs_info->super_copy;
2793
	if (!btrfs_super_root(disk_super))
2794
		goto fail_alloc;
2795

L
liubo 已提交
2796
	/* check FS state, whether FS is broken. */
2797 2798
	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
2799

C
Chris Mason 已提交
2800 2801 2802 2803 2804 2805 2806
	/*
	 * 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);

2807 2808 2809 2810 2811 2812
	/*
	 * 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;

2813
	ret = btrfs_parse_options(fs_info, options, sb->s_flags);
Y
Yan Zheng 已提交
2814 2815
	if (ret) {
		err = ret;
2816
		goto fail_alloc;
Y
Yan Zheng 已提交
2817
	}
2818

2819 2820 2821
	features = btrfs_super_incompat_flags(disk_super) &
		~BTRFS_FEATURE_INCOMPAT_SUPP;
	if (features) {
2822 2823 2824
		btrfs_err(fs_info,
		    "cannot mount because of unsupported optional features (%llx)",
		    features);
2825
		err = -EINVAL;
2826
		goto fail_alloc;
2827 2828
	}

2829
	features = btrfs_super_incompat_flags(disk_super);
L
Li Zefan 已提交
2830
	features |= BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF;
2831
	if (fs_info->compress_type == BTRFS_COMPRESS_LZO)
L
Li Zefan 已提交
2832
		features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
N
Nick Terrell 已提交
2833 2834
	else if (fs_info->compress_type == BTRFS_COMPRESS_ZSTD)
		features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_ZSTD;
2835

2836
	if (features & BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA)
2837
		btrfs_info(fs_info, "has skinny extents");
2838

2839 2840 2841 2842
	/*
	 * flag our filesystem as having big metadata blocks if
	 * they are bigger than the page size
	 */
2843
	if (btrfs_super_nodesize(disk_super) > PAGE_SIZE) {
2844
		if (!(features & BTRFS_FEATURE_INCOMPAT_BIG_METADATA))
2845 2846
			btrfs_info(fs_info,
				"flagging fs with big metadata feature");
2847 2848 2849
		features |= BTRFS_FEATURE_INCOMPAT_BIG_METADATA;
	}

2850 2851
	nodesize = btrfs_super_nodesize(disk_super);
	sectorsize = btrfs_super_sectorsize(disk_super);
2852
	stripesize = sectorsize;
2853
	fs_info->dirty_metadata_batch = nodesize * (1 + ilog2(nr_cpu_ids));
2854
	fs_info->delalloc_batch = sectorsize * 512 * (1 + ilog2(nr_cpu_ids));
2855

2856 2857 2858 2859 2860
	/* Cache block sizes */
	fs_info->nodesize = nodesize;
	fs_info->sectorsize = sectorsize;
	fs_info->stripesize = stripesize;

2861 2862 2863 2864 2865
	/*
	 * 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) &&
2866
	    (sectorsize != nodesize)) {
2867 2868 2869
		btrfs_err(fs_info,
"unequal nodesize/sectorsize (%u != %u) are not allowed for mixed block groups",
			nodesize, sectorsize);
2870 2871 2872
		goto fail_alloc;
	}

2873 2874 2875 2876
	/*
	 * Needn't use the lock because there is no other task which will
	 * update the flag.
	 */
L
Li Zefan 已提交
2877
	btrfs_set_super_incompat_flags(disk_super, features);
2878

2879 2880
	features = btrfs_super_compat_ro_flags(disk_super) &
		~BTRFS_FEATURE_COMPAT_RO_SUPP;
2881
	if (!sb_rdonly(sb) && features) {
2882 2883
		btrfs_err(fs_info,
	"cannot mount read-write because of unsupported optional features (%llx)",
2884
		       features);
2885
		err = -EINVAL;
2886
		goto fail_alloc;
2887
	}
2888

2889
	max_active = fs_info->thread_pool_size;
2890

2891 2892 2893
	ret = btrfs_init_workqueues(fs_info, fs_devices);
	if (ret) {
		err = ret;
2894 2895
		goto fail_sb_buffer;
	}
2896

2897 2898 2899 2900 2901 2902
	sb->s_bdi->congested_fn = btrfs_congested_fn;
	sb->s_bdi->congested_data = fs_info;
	sb->s_bdi->capabilities |= BDI_CAP_CGROUP_WRITEBACK;
	sb->s_bdi->ra_pages = VM_MAX_READAHEAD * 1024 / PAGE_SIZE;
	sb->s_bdi->ra_pages *= btrfs_super_num_devices(disk_super);
	sb->s_bdi->ra_pages = max(sb->s_bdi->ra_pages, SZ_4M / PAGE_SIZE);
2903

2904 2905
	sb->s_blocksize = sectorsize;
	sb->s_blocksize_bits = blksize_bits(sectorsize);
2906
	memcpy(&sb->s_uuid, fs_info->fsid, BTRFS_FSID_SIZE);
2907

2908
	mutex_lock(&fs_info->chunk_mutex);
2909
	ret = btrfs_read_sys_array(fs_info);
2910
	mutex_unlock(&fs_info->chunk_mutex);
2911
	if (ret) {
2912
		btrfs_err(fs_info, "failed to read the system array: %d", ret);
2913
		goto fail_sb_buffer;
2914
	}
2915

2916
	generation = btrfs_super_chunk_root_generation(disk_super);
2917

2918
	__setup_root(chunk_root, fs_info, BTRFS_CHUNK_TREE_OBJECTID);
2919

2920
	chunk_root->node = read_tree_block(fs_info,
2921
					   btrfs_super_chunk_root(disk_super),
2922
					   generation);
2923 2924
	if (IS_ERR(chunk_root->node) ||
	    !extent_buffer_uptodate(chunk_root->node)) {
2925
		btrfs_err(fs_info, "failed to read chunk root");
2926 2927
		if (!IS_ERR(chunk_root->node))
			free_extent_buffer(chunk_root->node);
2928
		chunk_root->node = NULL;
C
Chris Mason 已提交
2929
		goto fail_tree_roots;
2930
	}
2931 2932
	btrfs_set_root_node(&chunk_root->root_item, chunk_root->node);
	chunk_root->commit_root = btrfs_root_node(chunk_root);
2933

2934
	read_extent_buffer(chunk_root->node, fs_info->chunk_tree_uuid,
2935
	   btrfs_header_chunk_tree_uuid(chunk_root->node), BTRFS_UUID_SIZE);
2936

2937
	ret = btrfs_read_chunk_tree(fs_info);
Y
Yan Zheng 已提交
2938
	if (ret) {
2939
		btrfs_err(fs_info, "failed to read chunk tree: %d", ret);
C
Chris Mason 已提交
2940
		goto fail_tree_roots;
Y
Yan Zheng 已提交
2941
	}
2942

2943 2944 2945 2946
	/*
	 * keep the device that is marked to be the target device for the
	 * dev_replace procedure
	 */
2947
	btrfs_close_extra_devices(fs_devices, 0);
2948

2949
	if (!fs_devices->latest_bdev) {
2950
		btrfs_err(fs_info, "failed to read devices");
2951 2952 2953
		goto fail_tree_roots;
	}

C
Chris Mason 已提交
2954
retry_root_backup:
2955
	generation = btrfs_super_generation(disk_super);
2956

2957
	tree_root->node = read_tree_block(fs_info,
2958
					  btrfs_super_root(disk_super),
2959
					  generation);
2960 2961
	if (IS_ERR(tree_root->node) ||
	    !extent_buffer_uptodate(tree_root->node)) {
2962
		btrfs_warn(fs_info, "failed to read tree root");
2963 2964
		if (!IS_ERR(tree_root->node))
			free_extent_buffer(tree_root->node);
2965
		tree_root->node = NULL;
C
Chris Mason 已提交
2966
		goto recovery_tree_root;
2967
	}
C
Chris Mason 已提交
2968

2969 2970
	btrfs_set_root_node(&tree_root->root_item, tree_root->node);
	tree_root->commit_root = btrfs_root_node(tree_root);
2971
	btrfs_set_root_refs(&tree_root->root_item, 1);
2972

2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984
	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);

2985
	ret = btrfs_read_roots(fs_info);
2986
	if (ret)
C
Chris Mason 已提交
2987
		goto recovery_tree_root;
2988

2989 2990 2991
	fs_info->generation = generation;
	fs_info->last_trans_committed = generation;

2992 2993
	ret = btrfs_recover_balance(fs_info);
	if (ret) {
2994
		btrfs_err(fs_info, "failed to recover balance: %d", ret);
2995 2996 2997
		goto fail_block_groups;
	}

2998 2999
	ret = btrfs_init_dev_stats(fs_info);
	if (ret) {
3000
		btrfs_err(fs_info, "failed to init dev_stats: %d", ret);
3001 3002 3003
		goto fail_block_groups;
	}

3004 3005
	ret = btrfs_init_dev_replace(fs_info);
	if (ret) {
3006
		btrfs_err(fs_info, "failed to init dev_replace: %d", ret);
3007 3008 3009
		goto fail_block_groups;
	}

3010
	btrfs_close_extra_devices(fs_devices, 1);
3011

3012 3013
	ret = btrfs_sysfs_add_fsid(fs_devices, NULL);
	if (ret) {
3014 3015
		btrfs_err(fs_info, "failed to init sysfs fsid interface: %d",
				ret);
3016 3017 3018 3019 3020
		goto fail_block_groups;
	}

	ret = btrfs_sysfs_add_device(fs_devices);
	if (ret) {
3021 3022
		btrfs_err(fs_info, "failed to init sysfs device interface: %d",
				ret);
3023 3024 3025
		goto fail_fsdev_sysfs;
	}

3026
	ret = btrfs_sysfs_add_mounted(fs_info);
3027
	if (ret) {
3028
		btrfs_err(fs_info, "failed to init sysfs interface: %d", ret);
3029
		goto fail_fsdev_sysfs;
3030 3031 3032 3033
	}

	ret = btrfs_init_space_info(fs_info);
	if (ret) {
3034
		btrfs_err(fs_info, "failed to initialize space info: %d", ret);
3035
		goto fail_sysfs;
3036 3037
	}

3038
	ret = btrfs_read_block_groups(fs_info);
3039
	if (ret) {
3040
		btrfs_err(fs_info, "failed to read block groups: %d", ret);
3041
		goto fail_sysfs;
3042
	}
3043

3044
	if (!sb_rdonly(sb) && !btrfs_check_rw_degradable(fs_info)) {
3045
		btrfs_warn(fs_info,
3046
		"writeable mount is not allowed due to too many missing devices");
3047
		goto fail_sysfs;
3048
	}
C
Chris Mason 已提交
3049

3050 3051
	fs_info->cleaner_kthread = kthread_run(cleaner_kthread, tree_root,
					       "btrfs-cleaner");
3052
	if (IS_ERR(fs_info->cleaner_kthread))
3053
		goto fail_sysfs;
3054 3055 3056 3057

	fs_info->transaction_kthread = kthread_run(transaction_kthread,
						   tree_root,
						   "btrfs-transaction");
3058
	if (IS_ERR(fs_info->transaction_kthread))
3059
		goto fail_cleaner;
3060

3061
	if (!btrfs_test_opt(fs_info, NOSSD) &&
C
Chris Mason 已提交
3062
	    !fs_info->fs_devices->rotating) {
3063
		btrfs_set_and_info(fs_info, SSD, "enabling ssd optimizations");
C
Chris Mason 已提交
3064 3065
	}

3066
	/*
3067
	 * Mount does not set all options immediately, we can do it now and do
3068 3069 3070
	 * not have to wait for transaction commit
	 */
	btrfs_apply_pending_changes(fs_info);
3071

3072
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
3073
	if (btrfs_test_opt(fs_info, CHECK_INTEGRITY)) {
3074
		ret = btrfsic_mount(fs_info, fs_devices,
3075
				    btrfs_test_opt(fs_info,
3076 3077 3078 3079
					CHECK_INTEGRITY_INCLUDING_EXTENT_DATA) ?
				    1 : 0,
				    fs_info->check_integrity_print_mask);
		if (ret)
3080 3081 3082
			btrfs_warn(fs_info,
				"failed to initialize integrity check module: %d",
				ret);
3083 3084
	}
#endif
3085 3086 3087
	ret = btrfs_read_qgroup_config(fs_info);
	if (ret)
		goto fail_trans_kthread;
3088

3089 3090
	/* do not make disk changes in broken FS or nologreplay is given */
	if (btrfs_super_log_root(disk_super) != 0 &&
3091
	    !btrfs_test_opt(fs_info, NOLOGREPLAY)) {
3092
		ret = btrfs_replay_log(fs_info, fs_devices);
3093
		if (ret) {
3094
			err = ret;
3095
			goto fail_qgroup;
3096
		}
3097
	}
Z
Zheng Yan 已提交
3098

3099
	ret = btrfs_find_orphan_roots(fs_info);
3100
	if (ret)
3101
		goto fail_qgroup;
3102

3103
	if (!sb_rdonly(sb)) {
3104
		ret = btrfs_cleanup_fs_roots(fs_info);
3105
		if (ret)
3106
			goto fail_qgroup;
3107 3108

		mutex_lock(&fs_info->cleaner_mutex);
3109
		ret = btrfs_recover_relocation(tree_root);
3110
		mutex_unlock(&fs_info->cleaner_mutex);
3111
		if (ret < 0) {
3112 3113
			btrfs_warn(fs_info, "failed to recover relocation: %d",
					ret);
3114
			err = -EINVAL;
3115
			goto fail_qgroup;
3116
		}
3117
	}
Z
Zheng Yan 已提交
3118

3119 3120
	location.objectid = BTRFS_FS_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
3121
	location.offset = 0;
3122 3123

	fs_info->fs_root = btrfs_read_fs_root_no_name(fs_info, &location);
3124 3125
	if (IS_ERR(fs_info->fs_root)) {
		err = PTR_ERR(fs_info->fs_root);
3126
		goto fail_qgroup;
3127
	}
C
Chris Mason 已提交
3128

3129
	if (sb_rdonly(sb))
3130
		return 0;
I
Ilya Dryomov 已提交
3131

3132 3133
	if (btrfs_test_opt(fs_info, CLEAR_CACHE) &&
	    btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
3134 3135 3136 3137 3138 3139 3140 3141
		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) {
3142 3143 3144 3145 3146
		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);
3147
			close_ctree(fs_info);
3148 3149 3150 3151
			return ret;
		}
	}

3152
	if (btrfs_test_opt(fs_info, FREE_SPACE_TREE) &&
3153
	    !btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
3154
		btrfs_info(fs_info, "creating free space tree");
3155 3156
		ret = btrfs_create_free_space_tree(fs_info);
		if (ret) {
3157 3158
			btrfs_warn(fs_info,
				"failed to create free space tree: %d", ret);
3159
			close_ctree(fs_info);
3160 3161 3162 3163
			return ret;
		}
	}

3164 3165 3166
	down_read(&fs_info->cleanup_work_sem);
	if ((ret = btrfs_orphan_cleanup(fs_info->fs_root)) ||
	    (ret = btrfs_orphan_cleanup(fs_info->tree_root))) {
3167
		up_read(&fs_info->cleanup_work_sem);
3168
		close_ctree(fs_info);
3169 3170 3171
		return ret;
	}
	up_read(&fs_info->cleanup_work_sem);
I
Ilya Dryomov 已提交
3172

3173 3174
	ret = btrfs_resume_balance_async(fs_info);
	if (ret) {
3175
		btrfs_warn(fs_info, "failed to resume balance: %d", ret);
3176
		close_ctree(fs_info);
3177
		return ret;
3178 3179
	}

3180 3181
	ret = btrfs_resume_dev_replace_async(fs_info);
	if (ret) {
3182
		btrfs_warn(fs_info, "failed to resume device replace: %d", ret);
3183
		close_ctree(fs_info);
3184 3185 3186
		return ret;
	}

3187 3188
	btrfs_qgroup_rescan_resume(fs_info);

3189
	if (!fs_info->uuid_root) {
3190
		btrfs_info(fs_info, "creating UUID tree");
3191 3192
		ret = btrfs_create_uuid_tree(fs_info);
		if (ret) {
3193 3194
			btrfs_warn(fs_info,
				"failed to create the UUID tree: %d", ret);
3195
			close_ctree(fs_info);
3196 3197
			return ret;
		}
3198
	} else if (btrfs_test_opt(fs_info, RESCAN_UUID_TREE) ||
3199 3200
		   fs_info->generation !=
				btrfs_super_uuid_tree_generation(disk_super)) {
3201
		btrfs_info(fs_info, "checking UUID tree");
3202 3203
		ret = btrfs_check_uuid_tree(fs_info);
		if (ret) {
3204 3205
			btrfs_warn(fs_info,
				"failed to check the UUID tree: %d", ret);
3206
			close_ctree(fs_info);
3207 3208 3209
			return ret;
		}
	} else {
3210
		set_bit(BTRFS_FS_UPDATE_UUID_TREE_GEN, &fs_info->flags);
3211
	}
3212
	set_bit(BTRFS_FS_OPEN, &fs_info->flags);
3213

3214 3215 3216 3217 3218 3219
	/*
	 * 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
3220
	return 0;
C
Chris Mason 已提交
3221

3222 3223
fail_qgroup:
	btrfs_free_qgroup_config(fs_info);
3224 3225
fail_trans_kthread:
	kthread_stop(fs_info->transaction_kthread);
3226
	btrfs_cleanup_transaction(fs_info);
3227
	btrfs_free_fs_roots(fs_info);
3228
fail_cleaner:
3229
	kthread_stop(fs_info->cleaner_kthread);
3230 3231 3232 3233 3234 3235 3236

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

3237
fail_sysfs:
3238
	btrfs_sysfs_remove_mounted(fs_info);
3239

3240 3241 3242
fail_fsdev_sysfs:
	btrfs_sysfs_remove_fsid(fs_info->fs_devices);

3243
fail_block_groups:
J
Josef Bacik 已提交
3244
	btrfs_put_block_group_cache(fs_info);
C
Chris Mason 已提交
3245 3246 3247

fail_tree_roots:
	free_root_pointers(fs_info, 1);
3248
	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
3249

C
Chris Mason 已提交
3250
fail_sb_buffer:
L
Liu Bo 已提交
3251
	btrfs_stop_all_workers(fs_info);
3252
	btrfs_free_block_groups(fs_info);
3253
fail_alloc:
3254
fail_iput:
3255 3256
	btrfs_mapping_tree_free(&fs_info->mapping_tree);

3257
	iput(fs_info->btree_inode);
3258 3259
fail_bio_counter:
	percpu_counter_destroy(&fs_info->bio_counter);
3260 3261
fail_delalloc_bytes:
	percpu_counter_destroy(&fs_info->delalloc_bytes);
3262 3263
fail_dirty_metadata_bytes:
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
3264 3265
fail_srcu:
	cleanup_srcu_struct(&fs_info->subvol_srcu);
3266
fail:
D
David Woodhouse 已提交
3267
	btrfs_free_stripe_hash_table(fs_info);
3268
	btrfs_close_devices(fs_info->fs_devices);
A
Al Viro 已提交
3269
	return err;
C
Chris Mason 已提交
3270 3271

recovery_tree_root:
3272
	if (!btrfs_test_opt(fs_info, USEBACKUPROOT))
C
Chris Mason 已提交
3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287
		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;
3288 3289
}

3290 3291 3292 3293 3294
static void btrfs_end_buffer_write_sync(struct buffer_head *bh, int uptodate)
{
	if (uptodate) {
		set_buffer_uptodate(bh);
	} else {
3295 3296 3297
		struct btrfs_device *device = (struct btrfs_device *)
			bh->b_private;

3298
		btrfs_warn_rl_in_rcu(device->fs_info,
3299
				"lost page write due to IO error on %s",
3300
					  rcu_str_deref(device->name));
3301
		/* note, we don't set_buffer_write_io_error because we have
3302 3303
		 * our own ways of dealing with the IO errors
		 */
3304
		clear_buffer_uptodate(bh);
3305
		btrfs_dev_stat_inc_and_print(device, BTRFS_DEV_STAT_WRITE_ERRS);
3306 3307 3308 3309 3310
	}
	unlock_buffer(bh);
	put_bh(bh);
}

3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321
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;

3322
	bh = __bread(bdev, bytenr / BTRFS_BDEV_BLOCKSIZE, BTRFS_SUPER_INFO_SIZE);
3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341
	/*
	 * 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
3342 3343 3344 3345 3346 3347 3348
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;
3349
	int ret = -EINVAL;
Y
Yan Zheng 已提交
3350 3351 3352 3353 3354 3355 3356

	/* 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++) {
3357 3358
		ret = btrfs_read_dev_one_super(bdev, i, &bh);
		if (ret)
Y
Yan Zheng 已提交
3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370
			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);
		}
	}
3371 3372 3373 3374

	if (!latest)
		return ERR_PTR(ret);

Y
Yan Zheng 已提交
3375 3376 3377
	return latest;
}

3378
/*
3379 3380
 * Write superblock @sb to the @device. Do not wait for completion, all the
 * buffer heads we write are pinned.
3381
 *
3382 3383 3384
 * Write @max_mirrors copies of the superblock, where 0 means default that fit
 * the expected device size at commit time. Note that max_mirrors must be
 * same for write and wait phases.
3385
 *
3386
 * Return number of errors when buffer head is not found or submission fails.
3387
 */
Y
Yan Zheng 已提交
3388
static int write_dev_supers(struct btrfs_device *device,
3389
			    struct btrfs_super_block *sb, int max_mirrors)
Y
Yan Zheng 已提交
3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402
{
	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);
3403 3404
		if (bytenr + BTRFS_SUPER_INFO_SIZE >=
		    device->commit_total_bytes)
Y
Yan Zheng 已提交
3405 3406
			break;

3407
		btrfs_set_super_bytenr(sb, bytenr);
3408

3409 3410 3411 3412
		crc = ~(u32)0;
		crc = btrfs_csum_data((const char *)sb + BTRFS_CSUM_SIZE, crc,
				      BTRFS_SUPER_INFO_SIZE - BTRFS_CSUM_SIZE);
		btrfs_csum_final(crc, sb->csum);
3413

3414
		/* One reference for us, and we leave it for the caller */
3415
		bh = __getblk(device->bdev, bytenr / BTRFS_BDEV_BLOCKSIZE,
3416 3417 3418 3419 3420 3421
			      BTRFS_SUPER_INFO_SIZE);
		if (!bh) {
			btrfs_err(device->fs_info,
			    "couldn't get super buffer head for bytenr %llu",
			    bytenr);
			errors++;
3422
			continue;
3423
		}
3424

3425
		memcpy(bh->b_data, sb, BTRFS_SUPER_INFO_SIZE);
Y
Yan Zheng 已提交
3426

3427 3428
		/* one reference for submit_bh */
		get_bh(bh);
3429

3430 3431 3432 3433
		set_buffer_uptodate(bh);
		lock_buffer(bh);
		bh->b_end_io = btrfs_end_buffer_write_sync;
		bh->b_private = device;
Y
Yan Zheng 已提交
3434

C
Chris Mason 已提交
3435 3436 3437 3438
		/*
		 * we fua the first super.  The others we allow
		 * to go down lazy.
		 */
3439 3440
		if (i == 0) {
			ret = btrfsic_submit_bh(REQ_OP_WRITE,
3441
				REQ_SYNC | REQ_FUA | REQ_META | REQ_PRIO, bh);
3442
		} else {
3443 3444
			ret = btrfsic_submit_bh(REQ_OP_WRITE,
				REQ_SYNC | REQ_META | REQ_PRIO, bh);
3445
		}
3446
		if (ret)
Y
Yan Zheng 已提交
3447 3448 3449 3450 3451
			errors++;
	}
	return errors < i ? 0 : -1;
}

3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474
/*
 * Wait for write completion of superblocks done by write_dev_supers,
 * @max_mirrors same for write and wait phases.
 *
 * Return number of errors when buffer head is not found or not marked up to
 * date.
 */
static int wait_dev_supers(struct btrfs_device *device, int max_mirrors)
{
	struct buffer_head *bh;
	int i;
	int errors = 0;
	u64 bytenr;

	if (max_mirrors == 0)
		max_mirrors = BTRFS_SUPER_MIRROR_MAX;

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

3475 3476
		bh = __find_get_block(device->bdev,
				      bytenr / BTRFS_BDEV_BLOCKSIZE,
3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495
				      BTRFS_SUPER_INFO_SIZE);
		if (!bh) {
			errors++;
			continue;
		}
		wait_on_buffer(bh);
		if (!buffer_uptodate(bh))
			errors++;

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

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

	return errors < i ? 0 : -1;
}

C
Chris Mason 已提交
3496 3497 3498 3499
/*
 * endio for the write_dev_flush, this will wake anyone waiting
 * for the barrier when it is done
 */
3500
static void btrfs_end_empty_barrier(struct bio *bio)
C
Chris Mason 已提交
3501
{
3502
	complete(bio->bi_private);
C
Chris Mason 已提交
3503 3504 3505
}

/*
3506 3507
 * Submit a flush request to the device if it supports it. Error handling is
 * done in the waiting counterpart.
C
Chris Mason 已提交
3508
 */
3509
static void write_dev_flush(struct btrfs_device *device)
C
Chris Mason 已提交
3510
{
3511
	struct request_queue *q = bdev_get_queue(device->bdev);
3512
	struct bio *bio = device->flush_bio;
C
Chris Mason 已提交
3513

3514
	if (!test_bit(QUEUE_FLAG_WC, &q->queue_flags))
3515
		return;
C
Chris Mason 已提交
3516

3517
	bio_reset(bio);
C
Chris Mason 已提交
3518
	bio->bi_end_io = btrfs_end_empty_barrier;
3519
	bio_set_dev(bio, device->bdev);
3520
	bio->bi_opf = REQ_OP_WRITE | REQ_SYNC | REQ_PREFLUSH;
C
Chris Mason 已提交
3521 3522 3523
	init_completion(&device->flush_wait);
	bio->bi_private = &device->flush_wait;

3524
	btrfsic_submit_bio(bio);
3525
	device->flush_bio_sent = 1;
3526
}
C
Chris Mason 已提交
3527

3528 3529 3530
/*
 * If the flush bio has been submitted by write_dev_flush, wait for it.
 */
3531
static blk_status_t wait_dev_flush(struct btrfs_device *device)
3532 3533
{
	struct bio *bio = device->flush_bio;
C
Chris Mason 已提交
3534

3535
	if (!device->flush_bio_sent)
3536
		return BLK_STS_OK;
C
Chris Mason 已提交
3537

3538
	device->flush_bio_sent = 0;
3539
	wait_for_completion_io(&device->flush_wait);
C
Chris Mason 已提交
3540

3541
	return bio->bi_status;
C
Chris Mason 已提交
3542 3543
}

3544
static int check_barrier_error(struct btrfs_fs_info *fs_info)
3545
{
3546
	if (!btrfs_check_rw_degradable(fs_info))
3547
		return -EIO;
C
Chris Mason 已提交
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
	int errors_wait = 0;
3560
	blk_status_t ret;
C
Chris Mason 已提交
3561 3562 3563 3564

	/* send down all the barriers */
	head = &info->fs_devices->devices;
	list_for_each_entry_rcu(dev, head, dev_list) {
3565 3566
		if (dev->missing)
			continue;
3567
		if (!dev->bdev)
C
Chris Mason 已提交
3568 3569 3570 3571
			continue;
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

3572
		write_dev_flush(dev);
3573
		dev->last_flush_error = BLK_STS_OK;
C
Chris Mason 已提交
3574 3575 3576 3577
	}

	/* wait for all the barriers */
	list_for_each_entry_rcu(dev, head, dev_list) {
3578 3579
		if (dev->missing)
			continue;
C
Chris Mason 已提交
3580
		if (!dev->bdev) {
3581
			errors_wait++;
C
Chris Mason 已提交
3582 3583 3584 3585 3586
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

3587
		ret = wait_dev_flush(dev);
3588 3589
		if (ret) {
			dev->last_flush_error = ret;
3590 3591
			btrfs_dev_stat_inc_and_print(dev,
					BTRFS_DEV_STAT_FLUSH_ERRS);
3592
			errors_wait++;
3593 3594 3595
		}
	}

3596
	if (errors_wait) {
3597 3598 3599 3600 3601
		/*
		 * At some point we need the status of all disks
		 * to arrive at the volume status. So error checking
		 * is being pushed to a separate loop.
		 */
3602
		return check_barrier_error(info);
C
Chris Mason 已提交
3603 3604 3605 3606
	}
	return 0;
}

3607 3608
int btrfs_get_num_tolerated_disk_barrier_failures(u64 flags)
{
3609 3610
	int raid_type;
	int min_tolerated = INT_MAX;
3611

3612 3613 3614 3615 3616
	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);
3617

3618 3619 3620 3621 3622 3623 3624 3625 3626
	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);
	}
3627

3628
	if (min_tolerated == INT_MAX) {
3629
		pr_warn("BTRFS: unknown raid flag: %llu", flags);
3630 3631 3632 3633
		min_tolerated = 0;
	}

	return min_tolerated;
3634 3635
}

3636
int write_all_supers(struct btrfs_fs_info *fs_info, int max_mirrors)
3637
{
3638
	struct list_head *head;
3639
	struct btrfs_device *dev;
3640
	struct btrfs_super_block *sb;
3641 3642 3643
	struct btrfs_dev_item *dev_item;
	int ret;
	int do_barriers;
3644 3645
	int max_errors;
	int total_errors = 0;
3646
	u64 flags;
3647

3648
	do_barriers = !btrfs_test_opt(fs_info, NOBARRIER);
3649 3650 3651 3652 3653 3654 3655 3656

	/*
	 * max_mirrors == 0 indicates we're from commit_transaction,
	 * not from fsync where the tree roots in fs_info have not
	 * been consistent on disk.
	 */
	if (max_mirrors == 0)
		backup_super_roots(fs_info);
3657

3658
	sb = fs_info->super_for_commit;
3659
	dev_item = &sb->dev_item;
3660

3661 3662 3663
	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
3664

3665
	if (do_barriers) {
3666
		ret = barrier_all_devices(fs_info);
3667 3668
		if (ret) {
			mutex_unlock(
3669 3670 3671
				&fs_info->fs_devices->device_list_mutex);
			btrfs_handle_fs_error(fs_info, ret,
					      "errors while submitting device barriers.");
3672 3673 3674
			return ret;
		}
	}
C
Chris Mason 已提交
3675

3676
	list_for_each_entry_rcu(dev, head, dev_list) {
3677 3678 3679 3680
		if (!dev->bdev) {
			total_errors++;
			continue;
		}
Y
Yan Zheng 已提交
3681
		if (!dev->in_fs_metadata || !dev->writeable)
3682 3683
			continue;

Y
Yan Zheng 已提交
3684
		btrfs_set_stack_device_generation(dev_item, 0);
3685 3686
		btrfs_set_stack_device_type(dev_item, dev->type);
		btrfs_set_stack_device_id(dev_item, dev->devid);
3687
		btrfs_set_stack_device_total_bytes(dev_item,
3688
						   dev->commit_total_bytes);
3689 3690
		btrfs_set_stack_device_bytes_used(dev_item,
						  dev->commit_bytes_used);
3691 3692 3693 3694
		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);
3695
		memcpy(dev_item->fsid, dev->fs_devices->fsid, BTRFS_FSID_SIZE);
Y
Yan Zheng 已提交
3696

3697 3698 3699
		flags = btrfs_super_flags(sb);
		btrfs_set_super_flags(sb, flags | BTRFS_HEADER_FLAG_WRITTEN);

3700
		ret = write_dev_supers(dev, sb, max_mirrors);
3701 3702
		if (ret)
			total_errors++;
3703
	}
3704
	if (total_errors > max_errors) {
3705 3706 3707
		btrfs_err(fs_info, "%d errors while writing supers",
			  total_errors);
		mutex_unlock(&fs_info->fs_devices->device_list_mutex);
3708

3709
		/* FUA is masked off if unsupported and can't be the reason */
3710 3711 3712
		btrfs_handle_fs_error(fs_info, -EIO,
				      "%d errors while writing supers",
				      total_errors);
3713
		return -EIO;
3714
	}
3715

Y
Yan Zheng 已提交
3716
	total_errors = 0;
3717
	list_for_each_entry_rcu(dev, head, dev_list) {
3718 3719
		if (!dev->bdev)
			continue;
Y
Yan Zheng 已提交
3720
		if (!dev->in_fs_metadata || !dev->writeable)
3721 3722
			continue;

3723
		ret = wait_dev_supers(dev, max_mirrors);
Y
Yan Zheng 已提交
3724 3725
		if (ret)
			total_errors++;
3726
	}
3727
	mutex_unlock(&fs_info->fs_devices->device_list_mutex);
3728
	if (total_errors > max_errors) {
3729 3730 3731
		btrfs_handle_fs_error(fs_info, -EIO,
				      "%d errors while writing supers",
				      total_errors);
3732
		return -EIO;
3733
	}
3734 3735 3736
	return 0;
}

3737 3738 3739
/* 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
3740
{
3741
	spin_lock(&fs_info->fs_roots_radix_lock);
C
Chris Mason 已提交
3742 3743
	radix_tree_delete(&fs_info->fs_roots_radix,
			  (unsigned long)root->root_key.objectid);
3744
	spin_unlock(&fs_info->fs_roots_radix_lock);
3745 3746 3747 3748

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

L
Liu Bo 已提交
3749
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
L
Liu Bo 已提交
3750
		btrfs_free_log(NULL, root);
L
Liu Bo 已提交
3751 3752 3753 3754 3755 3756 3757
		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
3758

3759 3760 3761 3762
	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);
3763 3764 3765 3766 3767
	free_fs_root(root);
}

static void free_fs_root(struct btrfs_root *root)
{
3768
	iput(root->ino_cache_inode);
3769
	WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree));
3770
	btrfs_free_block_rsv(root->fs_info, root->orphan_block_rsv);
3771
	root->orphan_block_rsv = NULL;
3772 3773
	if (root->anon_dev)
		free_anon_bdev(root->anon_dev);
3774 3775
	if (root->subv_writers)
		btrfs_free_subvolume_writers(root->subv_writers);
3776 3777
	free_extent_buffer(root->node);
	free_extent_buffer(root->commit_root);
3778 3779
	kfree(root->free_ino_ctl);
	kfree(root->free_ino_pinned);
C
Chris Mason 已提交
3780
	kfree(root->name);
3781
	btrfs_put_fs_root(root);
C
Chris Mason 已提交
3782 3783
}

3784 3785 3786
void btrfs_free_fs_root(struct btrfs_root *root)
{
	free_fs_root(root);
C
Chris Mason 已提交
3787 3788
}

Y
Yan Zheng 已提交
3789
int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
C
Chris Mason 已提交
3790
{
Y
Yan Zheng 已提交
3791 3792
	u64 root_objectid = 0;
	struct btrfs_root *gang[8];
3793 3794 3795 3796
	int i = 0;
	int err = 0;
	unsigned int ret = 0;
	int index;
3797

Y
Yan Zheng 已提交
3798
	while (1) {
3799
		index = srcu_read_lock(&fs_info->subvol_srcu);
Y
Yan Zheng 已提交
3800 3801 3802
		ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix,
					     (void **)gang, root_objectid,
					     ARRAY_SIZE(gang));
3803 3804
		if (!ret) {
			srcu_read_unlock(&fs_info->subvol_srcu, index);
Y
Yan Zheng 已提交
3805
			break;
3806
		}
3807
		root_objectid = gang[ret - 1]->root_key.objectid + 1;
3808

Y
Yan Zheng 已提交
3809
		for (i = 0; i < ret; i++) {
3810 3811 3812 3813 3814 3815 3816 3817 3818
			/* 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);
3819

3820 3821 3822
		for (i = 0; i < ret; i++) {
			if (!gang[i])
				continue;
Y
Yan Zheng 已提交
3823
			root_objectid = gang[i]->root_key.objectid;
3824 3825
			err = btrfs_orphan_cleanup(gang[i]);
			if (err)
3826 3827
				break;
			btrfs_put_fs_root(gang[i]);
Y
Yan Zheng 已提交
3828 3829 3830
		}
		root_objectid++;
	}
3831 3832 3833 3834 3835 3836 3837

	/* 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
3838
}
3839

3840
int btrfs_commit_super(struct btrfs_fs_info *fs_info)
Y
Yan Zheng 已提交
3841
{
3842
	struct btrfs_root *root = fs_info->tree_root;
Y
Yan Zheng 已提交
3843
	struct btrfs_trans_handle *trans;
3844

3845
	mutex_lock(&fs_info->cleaner_mutex);
3846
	btrfs_run_delayed_iputs(fs_info);
3847 3848
	mutex_unlock(&fs_info->cleaner_mutex);
	wake_up_process(fs_info->cleaner_kthread);
3849 3850

	/* wait until ongoing cleanup work done */
3851 3852
	down_write(&fs_info->cleanup_work_sem);
	up_write(&fs_info->cleanup_work_sem);
3853

3854
	trans = btrfs_join_transaction(root);
3855 3856
	if (IS_ERR(trans))
		return PTR_ERR(trans);
3857
	return btrfs_commit_transaction(trans);
Y
Yan Zheng 已提交
3858 3859
}

3860
void close_ctree(struct btrfs_fs_info *fs_info)
Y
Yan Zheng 已提交
3861
{
3862
	struct btrfs_root *root = fs_info->tree_root;
Y
Yan Zheng 已提交
3863 3864
	int ret;

3865
	set_bit(BTRFS_FS_CLOSING_START, &fs_info->flags);
Y
Yan Zheng 已提交
3866

3867
	/* wait for the qgroup rescan worker to stop */
3868
	btrfs_qgroup_wait_for_completion(fs_info, false);
3869

S
Stefan Behrens 已提交
3870 3871 3872 3873 3874
	/* 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);

3875
	/* pause restriper - we want to resume on mount */
3876
	btrfs_pause_balance(fs_info);
3877

3878 3879
	btrfs_dev_replace_suspend_for_unmount(fs_info);

3880
	btrfs_scrub_cancel(fs_info);
C
Chris Mason 已提交
3881 3882 3883 3884 3885 3886

	/* 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 */
3887
	btrfs_cleanup_defrag_inodes(fs_info);
C
Chris Mason 已提交
3888

3889 3890
	cancel_work_sync(&fs_info->async_reclaim_work);

3891
	if (!sb_rdonly(fs_info->sb)) {
3892 3893 3894 3895 3896
		/*
		 * 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.
		 */
3897
		btrfs_delete_unused_bgs(fs_info);
3898

3899
		ret = btrfs_commit_super(fs_info);
L
liubo 已提交
3900
		if (ret)
3901
			btrfs_err(fs_info, "commit super ret %d", ret);
L
liubo 已提交
3902 3903
	}

3904
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state))
3905
		btrfs_error_commit_super(fs_info);
3906

A
Al Viro 已提交
3907 3908
	kthread_stop(fs_info->transaction_kthread);
	kthread_stop(fs_info->cleaner_kthread);
3909

3910
	set_bit(BTRFS_FS_CLOSING_DONE, &fs_info->flags);
3911

3912
	btrfs_free_qgroup_config(fs_info);
3913

3914
	if (percpu_counter_sum(&fs_info->delalloc_bytes)) {
3915
		btrfs_info(fs_info, "at unmount delalloc count %lld",
3916
		       percpu_counter_sum(&fs_info->delalloc_bytes));
C
Chris Mason 已提交
3917
	}
3918

3919
	btrfs_sysfs_remove_mounted(fs_info);
3920
	btrfs_sysfs_remove_fsid(fs_info->fs_devices);
3921

3922
	btrfs_free_fs_roots(fs_info);
3923

3924 3925
	btrfs_put_block_group_cache(fs_info);

3926 3927 3928 3929 3930
	/*
	 * 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);
3931 3932
	btrfs_stop_all_workers(fs_info);

3933 3934
	btrfs_free_block_groups(fs_info);

3935
	clear_bit(BTRFS_FS_OPEN, &fs_info->flags);
3936
	free_root_pointers(fs_info, 1);
3937

3938
	iput(fs_info->btree_inode);
3939

3940
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
3941
	if (btrfs_test_opt(fs_info, CHECK_INTEGRITY))
3942
		btrfsic_unmount(fs_info->fs_devices);
3943 3944
#endif

3945
	btrfs_close_devices(fs_info->fs_devices);
3946
	btrfs_mapping_tree_free(&fs_info->mapping_tree);
3947

3948
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
3949
	percpu_counter_destroy(&fs_info->delalloc_bytes);
3950
	percpu_counter_destroy(&fs_info->bio_counter);
3951
	cleanup_srcu_struct(&fs_info->subvol_srcu);
3952

D
David Woodhouse 已提交
3953 3954
	btrfs_free_stripe_hash_table(fs_info);

3955
	__btrfs_free_block_rsv(root->orphan_block_rsv);
3956
	root->orphan_block_rsv = NULL;
3957 3958 3959 3960 3961 3962 3963 3964 3965

	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);
	}
3966 3967
}

3968 3969
int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid,
			  int atomic)
3970
{
3971
	int ret;
3972
	struct inode *btree_inode = buf->pages[0]->mapping->host;
3973

3974
	ret = extent_buffer_uptodate(buf);
3975 3976 3977 3978
	if (!ret)
		return ret;

	ret = verify_parent_transid(&BTRFS_I(btree_inode)->io_tree, buf,
3979 3980 3981
				    parent_transid, atomic);
	if (ret == -EAGAIN)
		return ret;
3982
	return !ret;
3983 3984 3985 3986
}

void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
{
3987
	struct btrfs_fs_info *fs_info;
3988
	struct btrfs_root *root;
3989
	u64 transid = btrfs_header_generation(buf);
3990
	int was_dirty;
3991

3992 3993 3994 3995 3996 3997 3998 3999 4000 4001
#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;
4002
	fs_info = root->fs_info;
4003
	btrfs_assert_tree_locked(buf);
4004
	if (transid != fs_info->generation)
J
Jeff Mahoney 已提交
4005
		WARN(1, KERN_CRIT "btrfs transid mismatch buffer %llu, found %llu running %llu\n",
4006
			buf->start, transid, fs_info->generation);
4007
	was_dirty = set_extent_buffer_dirty(buf);
4008
	if (!was_dirty)
4009 4010 4011
		percpu_counter_add_batch(&fs_info->dirty_metadata_bytes,
					 buf->len,
					 fs_info->dirty_metadata_batch);
4012 4013
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	if (btrfs_header_level(buf) == 0 && check_leaf(root, buf)) {
4014
		btrfs_print_leaf(buf);
4015 4016 4017
		ASSERT(0);
	}
#endif
4018 4019
}

4020
static void __btrfs_btree_balance_dirty(struct btrfs_fs_info *fs_info,
4021
					int flush_delayed)
4022 4023 4024 4025 4026
{
	/*
	 * looks as though older kernels can get into trouble with
	 * this code, they end up stuck in balance_dirty_pages forever
	 */
4027
	int ret;
4028 4029 4030 4031

	if (current->flags & PF_MEMALLOC)
		return;

4032
	if (flush_delayed)
4033
		btrfs_balance_delayed_items(fs_info);
4034

4035
	ret = percpu_counter_compare(&fs_info->dirty_metadata_bytes,
4036 4037
				     BTRFS_DIRTY_METADATA_THRESH);
	if (ret > 0) {
4038
		balance_dirty_pages_ratelimited(fs_info->btree_inode->i_mapping);
4039 4040 4041
	}
}

4042
void btrfs_btree_balance_dirty(struct btrfs_fs_info *fs_info)
C
Chris Mason 已提交
4043
{
4044
	__btrfs_btree_balance_dirty(fs_info, 1);
4045
}
4046

4047
void btrfs_btree_balance_dirty_nodelay(struct btrfs_fs_info *fs_info)
4048
{
4049
	__btrfs_btree_balance_dirty(fs_info, 0);
C
Chris Mason 已提交
4050
}
4051

4052
int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid)
4053
{
4054
	struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root;
4055 4056 4057
	struct btrfs_fs_info *fs_info = root->fs_info;

	return btree_read_extent_buffer_pages(fs_info, buf, parent_transid);
4058
}
4059

4060
static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info)
L
liubo 已提交
4061
{
D
David Sterba 已提交
4062
	struct btrfs_super_block *sb = fs_info->super_copy;
4063 4064
	u64 nodesize = btrfs_super_nodesize(sb);
	u64 sectorsize = btrfs_super_sectorsize(sb);
D
David Sterba 已提交
4065 4066
	int ret = 0;

4067
	if (btrfs_super_magic(sb) != BTRFS_MAGIC) {
4068
		btrfs_err(fs_info, "no valid FS found");
4069 4070 4071
		ret = -EINVAL;
	}
	if (btrfs_super_flags(sb) & ~BTRFS_SUPER_FLAG_SUPP)
4072
		btrfs_warn(fs_info, "unrecognized super flag: %llu",
4073
				btrfs_super_flags(sb) & ~BTRFS_SUPER_FLAG_SUPP);
4074
	if (btrfs_super_root_level(sb) >= BTRFS_MAX_LEVEL) {
4075
		btrfs_err(fs_info, "tree_root level too big: %d >= %d",
4076
				btrfs_super_root_level(sb), BTRFS_MAX_LEVEL);
D
David Sterba 已提交
4077 4078
		ret = -EINVAL;
	}
4079
	if (btrfs_super_chunk_root_level(sb) >= BTRFS_MAX_LEVEL) {
4080
		btrfs_err(fs_info, "chunk_root level too big: %d >= %d",
4081
				btrfs_super_chunk_root_level(sb), BTRFS_MAX_LEVEL);
D
David Sterba 已提交
4082 4083
		ret = -EINVAL;
	}
4084
	if (btrfs_super_log_root_level(sb) >= BTRFS_MAX_LEVEL) {
4085
		btrfs_err(fs_info, "log_root level too big: %d >= %d",
4086
				btrfs_super_log_root_level(sb), BTRFS_MAX_LEVEL);
D
David Sterba 已提交
4087 4088 4089
		ret = -EINVAL;
	}

D
David Sterba 已提交
4090
	/*
4091 4092
	 * 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
4093
	 */
4094 4095
	if (!is_power_of_2(sectorsize) || sectorsize < 4096 ||
	    sectorsize > BTRFS_MAX_METADATA_BLOCKSIZE) {
4096
		btrfs_err(fs_info, "invalid sectorsize %llu", sectorsize);
4097 4098 4099
		ret = -EINVAL;
	}
	/* Only PAGE SIZE is supported yet */
4100
	if (sectorsize != PAGE_SIZE) {
4101 4102 4103
		btrfs_err(fs_info,
			"sectorsize %llu not supported yet, only support %lu",
			sectorsize, PAGE_SIZE);
4104 4105 4106 4107
		ret = -EINVAL;
	}
	if (!is_power_of_2(nodesize) || nodesize < sectorsize ||
	    nodesize > BTRFS_MAX_METADATA_BLOCKSIZE) {
4108
		btrfs_err(fs_info, "invalid nodesize %llu", nodesize);
4109 4110 4111
		ret = -EINVAL;
	}
	if (nodesize != le32_to_cpu(sb->__unused_leafsize)) {
4112 4113
		btrfs_err(fs_info, "invalid leafsize %u, should be %llu",
			  le32_to_cpu(sb->__unused_leafsize), nodesize);
4114 4115 4116 4117 4118
		ret = -EINVAL;
	}

	/* Root alignment check */
	if (!IS_ALIGNED(btrfs_super_root(sb), sectorsize)) {
4119 4120
		btrfs_warn(fs_info, "tree_root block unaligned: %llu",
			   btrfs_super_root(sb));
4121 4122 4123
		ret = -EINVAL;
	}
	if (!IS_ALIGNED(btrfs_super_chunk_root(sb), sectorsize)) {
4124 4125
		btrfs_warn(fs_info, "chunk_root block unaligned: %llu",
			   btrfs_super_chunk_root(sb));
4126 4127
		ret = -EINVAL;
	}
4128
	if (!IS_ALIGNED(btrfs_super_log_root(sb), sectorsize)) {
4129 4130
		btrfs_warn(fs_info, "log_root block unaligned: %llu",
			   btrfs_super_log_root(sb));
4131 4132 4133
		ret = -EINVAL;
	}

4134
	if (memcmp(fs_info->fsid, sb->dev_item.fsid, BTRFS_FSID_SIZE) != 0) {
4135 4136 4137
		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
4138 4139 4140 4141 4142 4143 4144
		ret = -EINVAL;
	}

	/*
	 * Hint to catch really bogus numbers, bitflips or so, more exact checks are
	 * done later
	 */
4145 4146
	if (btrfs_super_bytes_used(sb) < 6 * btrfs_super_nodesize(sb)) {
		btrfs_err(fs_info, "bytes_used is too small %llu",
4147
			  btrfs_super_bytes_used(sb));
4148 4149
		ret = -EINVAL;
	}
4150
	if (!is_power_of_2(btrfs_super_stripesize(sb))) {
4151
		btrfs_err(fs_info, "invalid stripesize %u",
4152
			  btrfs_super_stripesize(sb));
4153 4154
		ret = -EINVAL;
	}
4155
	if (btrfs_super_num_devices(sb) > (1UL << 31))
4156 4157
		btrfs_warn(fs_info, "suspicious number of devices: %llu",
			   btrfs_super_num_devices(sb));
4158
	if (btrfs_super_num_devices(sb) == 0) {
4159
		btrfs_err(fs_info, "number of devices is 0");
4160 4161
		ret = -EINVAL;
	}
D
David Sterba 已提交
4162

4163
	if (btrfs_super_bytenr(sb) != BTRFS_SUPER_INFO_OFFSET) {
4164 4165
		btrfs_err(fs_info, "super offset mismatch %llu != %u",
			  btrfs_super_bytenr(sb), BTRFS_SUPER_INFO_OFFSET);
D
David Sterba 已提交
4166 4167 4168
		ret = -EINVAL;
	}

4169 4170 4171 4172 4173
	/*
	 * 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) {
4174 4175 4176
		btrfs_err(fs_info, "system chunk array too big %u > %u",
			  btrfs_super_sys_array_size(sb),
			  BTRFS_SYSTEM_CHUNK_ARRAY_SIZE);
4177 4178 4179 4180
		ret = -EINVAL;
	}
	if (btrfs_super_sys_array_size(sb) < sizeof(struct btrfs_disk_key)
			+ sizeof(struct btrfs_chunk)) {
4181 4182 4183 4184
		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));
4185 4186 4187
		ret = -EINVAL;
	}

D
David Sterba 已提交
4188 4189 4190 4191
	/*
	 * 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.
	 */
4192
	if (btrfs_super_generation(sb) < btrfs_super_chunk_root_generation(sb))
4193 4194 4195 4196
		btrfs_warn(fs_info,
			"suspicious: generation < chunk_root_generation: %llu < %llu",
			btrfs_super_generation(sb),
			btrfs_super_chunk_root_generation(sb));
4197 4198
	if (btrfs_super_generation(sb) < btrfs_super_cache_generation(sb)
	    && btrfs_super_cache_generation(sb) != (u64)-1)
4199 4200 4201 4202
		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
4203 4204

	return ret;
L
liubo 已提交
4205 4206
}

4207
static void btrfs_error_commit_super(struct btrfs_fs_info *fs_info)
L
liubo 已提交
4208
{
4209
	mutex_lock(&fs_info->cleaner_mutex);
4210
	btrfs_run_delayed_iputs(fs_info);
4211
	mutex_unlock(&fs_info->cleaner_mutex);
L
liubo 已提交
4212

4213 4214
	down_write(&fs_info->cleanup_work_sem);
	up_write(&fs_info->cleanup_work_sem);
L
liubo 已提交
4215 4216

	/* cleanup FS via transaction */
4217
	btrfs_cleanup_transaction(fs_info);
L
liubo 已提交
4218 4219
}

4220
static void btrfs_destroy_ordered_extents(struct btrfs_root *root)
L
liubo 已提交
4221 4222 4223
{
	struct btrfs_ordered_extent *ordered;

4224
	spin_lock(&root->ordered_extent_lock);
4225 4226 4227 4228
	/*
	 * This will just short circuit the ordered completion stuff which will
	 * make sure the ordered extent gets properly cleaned up.
	 */
4229
	list_for_each_entry(ordered, &root->ordered_extents,
4230 4231
			    root_extent_list)
		set_bit(BTRFS_ORDERED_IOERR, &ordered->flags);
4232 4233 4234 4235 4236 4237 4238 4239 4240 4241 4242 4243 4244 4245 4246
	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);
4247 4248
		list_move_tail(&root->ordered_root,
			       &fs_info->ordered_roots);
4249

4250
		spin_unlock(&fs_info->ordered_root_lock);
4251 4252
		btrfs_destroy_ordered_extents(root);

4253 4254
		cond_resched();
		spin_lock(&fs_info->ordered_root_lock);
4255 4256
	}
	spin_unlock(&fs_info->ordered_root_lock);
L
liubo 已提交
4257 4258
}

4259
static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
4260
				      struct btrfs_fs_info *fs_info)
L
liubo 已提交
4261 4262 4263 4264 4265 4266 4267 4268 4269
{
	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);
4270
	if (atomic_read(&delayed_refs->num_entries) == 0) {
4271
		spin_unlock(&delayed_refs->lock);
4272
		btrfs_info(fs_info, "delayed_refs has NO entry");
L
liubo 已提交
4273 4274 4275
		return ret;
	}

4276 4277
	while ((node = rb_first(&delayed_refs->href_root)) != NULL) {
		struct btrfs_delayed_ref_head *head;
4278
		struct btrfs_delayed_ref_node *tmp;
4279
		bool pin_bytes = false;
L
liubo 已提交
4280

4281 4282 4283
		head = rb_entry(node, struct btrfs_delayed_ref_head,
				href_node);
		if (!mutex_trylock(&head->mutex)) {
4284
			refcount_inc(&head->node.refs);
4285
			spin_unlock(&delayed_refs->lock);
4286

4287
			mutex_lock(&head->mutex);
4288
			mutex_unlock(&head->mutex);
4289 4290 4291 4292 4293
			btrfs_put_delayed_ref(&head->node);
			spin_lock(&delayed_refs->lock);
			continue;
		}
		spin_lock(&head->lock);
4294 4295
		list_for_each_entry_safe_reverse(ref, tmp, &head->ref_list,
						 list) {
4296
			ref->in_tree = 0;
4297
			list_del(&ref->list);
4298 4299
			if (!list_empty(&ref->add_list))
				list_del(&ref->add_list);
4300 4301
			atomic_dec(&delayed_refs->num_entries);
			btrfs_put_delayed_ref(ref);
4302
		}
4303 4304 4305 4306 4307 4308 4309 4310 4311 4312 4313 4314
		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
4315

4316
		if (pin_bytes)
4317
			btrfs_pin_extent(fs_info, head->node.bytenr,
4318 4319
					 head->node.num_bytes, 1);
		btrfs_put_delayed_ref(&head->node);
L
liubo 已提交
4320 4321 4322 4323 4324 4325 4326 4327 4328
		cond_resched();
		spin_lock(&delayed_refs->lock);
	}

	spin_unlock(&delayed_refs->lock);

	return ret;
}

4329
static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root)
L
liubo 已提交
4330 4331 4332 4333 4334 4335
{
	struct btrfs_inode *btrfs_inode;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

4336 4337
	spin_lock(&root->delalloc_lock);
	list_splice_init(&root->delalloc_inodes, &splice);
L
liubo 已提交
4338 4339

	while (!list_empty(&splice)) {
4340 4341
		btrfs_inode = list_first_entry(&splice, struct btrfs_inode,
					       delalloc_inodes);
L
liubo 已提交
4342 4343

		list_del_init(&btrfs_inode->delalloc_inodes);
4344 4345
		clear_bit(BTRFS_INODE_IN_DELALLOC_LIST,
			  &btrfs_inode->runtime_flags);
4346
		spin_unlock(&root->delalloc_lock);
L
liubo 已提交
4347 4348

		btrfs_invalidate_inodes(btrfs_inode->root);
4349

4350
		spin_lock(&root->delalloc_lock);
L
liubo 已提交
4351 4352
	}

4353 4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369 4370 4371 4372 4373 4374 4375 4376 4377 4378
	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
4379 4380
}

4381
static int btrfs_destroy_marked_extents(struct btrfs_fs_info *fs_info,
L
liubo 已提交
4382 4383 4384 4385 4386 4387 4388 4389 4390 4391
					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,
4392
					    mark, NULL);
L
liubo 已提交
4393 4394 4395
		if (ret)
			break;

4396
		clear_extent_bits(dirty_pages, start, end, mark);
L
liubo 已提交
4397
		while (start <= end) {
4398 4399
			eb = find_extent_buffer(fs_info, start);
			start += fs_info->nodesize;
4400
			if (!eb)
L
liubo 已提交
4401
				continue;
4402
			wait_on_extent_buffer_writeback(eb);
L
liubo 已提交
4403

4404 4405 4406 4407
			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
4408 4409 4410 4411 4412 4413
		}
	}

	return ret;
}

4414
static int btrfs_destroy_pinned_extent(struct btrfs_fs_info *fs_info,
L
liubo 已提交
4415 4416 4417 4418 4419 4420
				       struct extent_io_tree *pinned_extents)
{
	struct extent_io_tree *unpin;
	u64 start;
	u64 end;
	int ret;
4421
	bool loop = true;
L
liubo 已提交
4422 4423

	unpin = pinned_extents;
4424
again:
L
liubo 已提交
4425 4426
	while (1) {
		ret = find_first_extent_bit(unpin, 0, &start, &end,
4427
					    EXTENT_DIRTY, NULL);
L
liubo 已提交
4428 4429 4430
		if (ret)
			break;

4431
		clear_extent_dirty(unpin, start, end);
4432
		btrfs_error_unpin_extent_range(fs_info, start, end);
L
liubo 已提交
4433 4434 4435
		cond_resched();
	}

4436
	if (loop) {
4437 4438
		if (unpin == &fs_info->freed_extents[0])
			unpin = &fs_info->freed_extents[1];
4439
		else
4440
			unpin = &fs_info->freed_extents[0];
4441 4442 4443 4444
		loop = false;
		goto again;
	}

L
liubo 已提交
4445 4446 4447
	return 0;
}

4448 4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462
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,
4463
			     struct btrfs_fs_info *fs_info)
4464 4465 4466 4467 4468 4469 4470 4471 4472
{
	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) {
4473
			btrfs_err(fs_info, "orphan block group dirty_bgs list");
4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500
			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) {
4501
			btrfs_err(fs_info, "orphan block group on io_bgs list");
4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512
			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);
	}
}

4513
void btrfs_cleanup_one_transaction(struct btrfs_transaction *cur_trans,
4514
				   struct btrfs_fs_info *fs_info)
4515
{
4516
	btrfs_cleanup_dirty_bgs(cur_trans, fs_info);
4517 4518 4519
	ASSERT(list_empty(&cur_trans->dirty_bgs));
	ASSERT(list_empty(&cur_trans->io_bgs));

4520
	btrfs_destroy_delayed_refs(cur_trans, fs_info);
4521

4522
	cur_trans->state = TRANS_STATE_COMMIT_START;
4523
	wake_up(&fs_info->transaction_blocked_wait);
4524

4525
	cur_trans->state = TRANS_STATE_UNBLOCKED;
4526
	wake_up(&fs_info->transaction_wait);
4527

4528 4529
	btrfs_destroy_delayed_inodes(fs_info);
	btrfs_assert_delayed_root_empty(fs_info);
4530

4531
	btrfs_destroy_marked_extents(fs_info, &cur_trans->dirty_pages,
4532
				     EXTENT_DIRTY);
4533
	btrfs_destroy_pinned_extent(fs_info,
4534
				    fs_info->pinned_extents);
4535

4536 4537
	cur_trans->state =TRANS_STATE_COMPLETED;
	wake_up(&cur_trans->commit_wait);
4538 4539
}

4540
static int btrfs_cleanup_transaction(struct btrfs_fs_info *fs_info)
L
liubo 已提交
4541 4542 4543
{
	struct btrfs_transaction *t;

4544
	mutex_lock(&fs_info->transaction_kthread_mutex);
L
liubo 已提交
4545

4546 4547 4548
	spin_lock(&fs_info->trans_lock);
	while (!list_empty(&fs_info->trans_list)) {
		t = list_first_entry(&fs_info->trans_list,
4549 4550
				     struct btrfs_transaction, list);
		if (t->state >= TRANS_STATE_COMMIT_START) {
4551
			refcount_inc(&t->use_count);
4552
			spin_unlock(&fs_info->trans_lock);
4553
			btrfs_wait_for_commit(fs_info, t->transid);
4554
			btrfs_put_transaction(t);
4555
			spin_lock(&fs_info->trans_lock);
4556 4557
			continue;
		}
4558
		if (t == fs_info->running_transaction) {
4559
			t->state = TRANS_STATE_COMMIT_DOING;
4560
			spin_unlock(&fs_info->trans_lock);
4561 4562 4563 4564 4565 4566 4567
			/*
			 * 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 {
4568
			spin_unlock(&fs_info->trans_lock);
4569
		}
4570
		btrfs_cleanup_one_transaction(t, fs_info);
4571

4572 4573 4574
		spin_lock(&fs_info->trans_lock);
		if (t == fs_info->running_transaction)
			fs_info->running_transaction = NULL;
L
liubo 已提交
4575
		list_del_init(&t->list);
4576
		spin_unlock(&fs_info->trans_lock);
L
liubo 已提交
4577

4578
		btrfs_put_transaction(t);
4579
		trace_btrfs_transaction_commit(fs_info->tree_root);
4580
		spin_lock(&fs_info->trans_lock);
4581
	}
4582 4583
	spin_unlock(&fs_info->trans_lock);
	btrfs_destroy_all_ordered_extents(fs_info);
4584 4585
	btrfs_destroy_delayed_inodes(fs_info);
	btrfs_assert_delayed_root_empty(fs_info);
4586
	btrfs_destroy_pinned_extent(fs_info, fs_info->pinned_extents);
4587 4588
	btrfs_destroy_all_delalloc_inodes(fs_info);
	mutex_unlock(&fs_info->transaction_kthread_mutex);
L
liubo 已提交
4589 4590 4591 4592

	return 0;
}

4593 4594 4595 4596 4597 4598
static struct btrfs_fs_info *btree_fs_info(void *private_data)
{
	struct inode *inode = private_data;
	return btrfs_sb(inode->i_sb);
}

4599
static const struct extent_io_ops btree_extent_io_ops = {
4600
	/* mandatory callbacks */
4601
	.submit_bio_hook = btree_submit_bio_hook,
4602
	.readpage_end_io_hook = btree_readpage_end_io_hook,
4603 4604
	/* note we're sharing with inode.c for the merge bio hook */
	.merge_bio_hook = btrfs_merge_bio_hook,
4605
	.readpage_io_failed_hook = btree_io_failed_hook,
4606 4607
	.set_range_writeback = btrfs_set_range_writeback,
	.tree_fs_info = btree_fs_info,
4608 4609

	/* optional callbacks */
4610
};