disk-io.c 123.1 KB
Newer Older
C
Chris Mason 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
/*
 * Copyright (C) 2007 Oracle.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License v2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */

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

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

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

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

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

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

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

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

C
Chris Mason 已提交
116 117 118 119 120
/*
 * async submit bios are used to offload expensive checksumming
 * onto the worker threads.  They checksum file and metadata bios
 * just before they are sent down the IO stack.
 */
121 122 123 124
struct async_submit_bio {
	struct inode *inode;
	struct bio *bio;
	struct list_head list;
C
Chris Mason 已提交
125 126
	extent_submit_bio_hook_t *submit_bio_start;
	extent_submit_bio_hook_t *submit_bio_done;
127 128
	int rw;
	int mirror_num;
C
Chris Mason 已提交
129
	unsigned long bio_flags;
130 131 132 133 134
	/*
	 * 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;
135
	struct btrfs_work work;
136
	int error;
137 138
};

139 140 141 142 143 144 145 146 147 148
/*
 * 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.
149
 *
150 151 152
 * 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.
153
 *
154 155 156
 * 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.
157
 *
158 159 160
 * 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.
161 162 163 164 165
 */
#ifdef CONFIG_DEBUG_LOCK_ALLOC
# if BTRFS_MAX_LEVEL != 8
#  error
# endif
166 167 168 169 170 171 172 173 174 175 176 177 178

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"	},
179
	{ .id = BTRFS_QUOTA_TREE_OBJECTID,	.name_stem = "quota"	},
180 181 182
	{ .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"	},
183
	{ .id = BTRFS_UUID_TREE_OBJECTID,	.name_stem = "uuid"	},
184
	{ .id = BTRFS_FREE_SPACE_TREE_OBJECTID,	.name_stem = "free-space" },
185
	{ .id = 0,				.name_stem = "tree"	},
186
};
187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217

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]);
}

218 219
#endif

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

232
	read_lock(&em_tree->lock);
233
	em = lookup_extent_mapping(em_tree, start, len);
234 235 236
	if (em) {
		em->bdev =
			BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
237
		read_unlock(&em_tree->lock);
238
		goto out;
239
	}
240
	read_unlock(&em_tree->lock);
241

242
	em = alloc_extent_map();
243 244 245 246 247
	if (!em) {
		em = ERR_PTR(-ENOMEM);
		goto out;
	}
	em->start = 0;
248
	em->len = (u64)-1;
C
Chris Mason 已提交
249
	em->block_len = (u64)-1;
250
	em->block_start = 0;
251
	em->bdev = BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
252

253
	write_lock(&em_tree->lock);
J
Josef Bacik 已提交
254
	ret = add_extent_mapping(em_tree, em, 0);
255 256
	if (ret == -EEXIST) {
		free_extent_map(em);
257
		em = lookup_extent_mapping(em_tree, start, len);
258
		if (!em)
259
			em = ERR_PTR(-EIO);
260
	} else if (ret) {
261
		free_extent_map(em);
262
		em = ERR_PTR(ret);
263
	}
264
	write_unlock(&em_tree->lock);
265

266 267
out:
	return em;
268 269
}

270
u32 btrfs_csum_data(char *data, u32 seed, size_t len)
271
{
272
	return btrfs_crc32c(seed, data, len);
273 274 275 276
}

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

C
Chris Mason 已提交
280 281 282 283
/*
 * compute the csum for a btree block, and either verify it or write it
 * into the csum field of the block.
 */
284 285
static int csum_tree_block(struct btrfs_fs_info *fs_info,
			   struct extent_buffer *buf,
286 287
			   int verify)
{
288
	u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
289
	char *result = NULL;
290 291 292 293 294 295 296 297
	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;
298
	unsigned long inline_result;
299 300

	len = buf->len - offset;
C
Chris Mason 已提交
301
	while (len > 0) {
302
		err = map_private_extent_buffer(buf, offset, 32,
303
					&kaddr, &map_start, &map_len);
C
Chris Mason 已提交
304
		if (err)
305
			return err;
306
		cur_len = min(len, map_len - (offset - map_start));
307
		crc = btrfs_csum_data(kaddr + offset - map_start,
308 309 310 311
				      crc, cur_len);
		len -= cur_len;
		offset += cur_len;
	}
312
	if (csum_size > sizeof(inline_result)) {
313
		result = kzalloc(csum_size, GFP_NOFS);
314
		if (!result)
315
			return -ENOMEM;
316 317 318 319
	} else {
		result = (char *)&inline_result;
	}

320 321 322
	btrfs_csum_final(crc, result);

	if (verify) {
323
		if (memcmp_extent_buffer(buf, result, 0, csum_size)) {
324 325
			u32 val;
			u32 found = 0;
326
			memcpy(&found, result, csum_size);
327

328
			read_extent_buffer(buf, &val, 0, csum_size);
329 330 331
			btrfs_warn_rl(fs_info,
				"%s checksum verify failed on %llu wanted %X found %X "
				"level %d",
332
				fs_info->sb->s_id, buf->start,
333
				val, found, btrfs_header_level(buf));
334 335
			if (result != (char *)&inline_result)
				kfree(result);
336
			return -EUCLEAN;
337 338
		}
	} else {
339
		write_extent_buffer(buf, result, 0, csum_size);
340
	}
341 342
	if (result != (char *)&inline_result)
		kfree(result);
343 344 345
	return 0;
}

C
Chris Mason 已提交
346 347 348 349 350 351
/*
 * 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.
 */
352
static int verify_parent_transid(struct extent_io_tree *io_tree,
353 354
				 struct extent_buffer *eb, u64 parent_transid,
				 int atomic)
355
{
356
	struct extent_state *cached_state = NULL;
357
	int ret;
358
	bool need_lock = (current->journal_info == BTRFS_SEND_TRANS_STUB);
359 360 361 362

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

363 364 365
	if (atomic)
		return -EAGAIN;

366 367 368 369 370
	if (need_lock) {
		btrfs_tree_read_lock(eb);
		btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
	}

371
	lock_extent_bits(io_tree, eb->start, eb->start + eb->len - 1,
372
			 &cached_state);
373
	if (extent_buffer_uptodate(eb) &&
374 375 376 377
	    btrfs_header_generation(eb) == parent_transid) {
		ret = 0;
		goto out;
	}
378 379 380
	btrfs_err_rl(eb->fs_info,
		"parent transid verify failed on %llu wanted %llu found %llu",
			eb->start,
381
			parent_transid, btrfs_header_generation(eb));
382
	ret = 1;
383 384 385 386

	/*
	 * 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
387
	 * block that has been freed and re-allocated.  So don't clear uptodate
388 389 390 391 392 393
	 * 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
394
out:
395 396
	unlock_extent_cached(io_tree, eb->start, eb->start + eb->len - 1,
			     &cached_state, GFP_NOFS);
397 398
	if (need_lock)
		btrfs_tree_read_unlock_blocking(eb);
399 400 401
	return ret;
}

D
David Sterba 已提交
402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420
/*
 * Return 0 if the superblock checksum type matches the checksum value of that
 * algorithm. Pass the raw disk superblock data.
 */
static int btrfs_check_super_csum(char *raw_disk_sb)
{
	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
421
		 * is filled with zeros and is included in the checksum.
D
David Sterba 已提交
422 423 424 425 426 427 428 429 430 431
		 */
		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)) {
432
		printk(KERN_ERR "BTRFS: unsupported checksum algorithm %u\n",
D
David Sterba 已提交
433 434 435 436 437 438 439
				csum_type);
		ret = 1;
	}

	return ret;
}

C
Chris Mason 已提交
440 441 442 443
/*
 * helper to read a given tree block, doing retries as required when
 * the checksums don't match and we have alternate mirrors to try.
 */
444 445
static int btree_read_extent_buffer_pages(struct btrfs_root *root,
					  struct extent_buffer *eb,
446
					  u64 start, u64 parent_transid)
447 448
{
	struct extent_io_tree *io_tree;
449
	int failed = 0;
450 451 452
	int ret;
	int num_copies = 0;
	int mirror_num = 0;
453
	int failed_mirror = 0;
454

455
	clear_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
456 457
	io_tree = &BTRFS_I(root->fs_info->btree_inode)->io_tree;
	while (1) {
458 459
		ret = read_extent_buffer_pages(io_tree, eb, start,
					       WAIT_COMPLETE,
460
					       btree_get_extent, mirror_num);
461 462
		if (!ret) {
			if (!verify_parent_transid(io_tree, eb,
463
						   parent_transid, 0))
464 465 466 467
				break;
			else
				ret = -EIO;
		}
C
Chris Mason 已提交
468

469 470 471 472 473 474
		/*
		 * 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))
475 476
			break;

477
		num_copies = btrfs_num_copies(root->fs_info,
478
					      eb->start, eb->len);
C
Chris Mason 已提交
479
		if (num_copies == 1)
480
			break;
C
Chris Mason 已提交
481

482 483 484 485 486
		if (!failed_mirror) {
			failed = 1;
			failed_mirror = eb->read_mirror;
		}

487
		mirror_num++;
488 489 490
		if (mirror_num == failed_mirror)
			mirror_num++;

C
Chris Mason 已提交
491
		if (mirror_num > num_copies)
492
			break;
493
	}
494

495
	if (failed && !ret && failed_mirror)
496 497 498
		repair_eb_io_failure(root, eb, failed_mirror);

	return ret;
499
}
500

C
Chris Mason 已提交
501
/*
C
Chris Mason 已提交
502 503
 * 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
504
 */
C
Chris Mason 已提交
505

506
static int csum_dirty_buffer(struct btrfs_fs_info *fs_info, struct page *page)
507
{
M
Miao Xie 已提交
508
	u64 start = page_offset(page);
509 510
	u64 found_start;
	struct extent_buffer *eb;
511

J
Josef Bacik 已提交
512 513 514
	eb = (struct extent_buffer *)page->private;
	if (page != eb->pages[0])
		return 0;
515

516
	found_start = btrfs_header_bytenr(eb);
517 518 519 520 521 522 523 524 525 526 527 528
	/*
	 * 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);

529
	return csum_tree_block(fs_info, eb, 0);
530 531
}

532
static int check_tree_block_fsid(struct btrfs_fs_info *fs_info,
Y
Yan Zheng 已提交
533 534
				 struct extent_buffer *eb)
{
535
	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
Y
Yan Zheng 已提交
536 537 538
	u8 fsid[BTRFS_UUID_SIZE];
	int ret = 1;

539
	read_extent_buffer(eb, fsid, btrfs_header_fsid(), BTRFS_FSID_SIZE);
Y
Yan Zheng 已提交
540 541 542 543 544 545 546 547 548 549
	while (fs_devices) {
		if (!memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE)) {
			ret = 0;
			break;
		}
		fs_devices = fs_devices->seed;
	}
	return ret;
}

550
#define CORRUPT(reason, eb, root, slot)				\
551 552
	btrfs_crit(root->fs_info, "corrupt leaf, %s: block=%llu,"	\
		   "root=%llu, slot=%d", reason,			\
553
	       btrfs_header_bytenr(eb),	root->objectid, slot)
554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602

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

	if (nritems == 0)
		return 0;

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

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

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

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

		/*
		 * Check to make sure that we don't point outside of the leaf,
603
		 * just in case all the items are consistent to each other, but
604 605 606 607 608 609 610 611 612 613 614 615
		 * all point outside of the leaf.
		 */
		if (btrfs_item_end_nr(leaf, slot) >
		    BTRFS_LEAF_DATA_SIZE(root)) {
			CORRUPT("slot end outside of leaf", leaf, root, slot);
			return -EIO;
		}
	}

	return 0;
}

616 617 618
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)
619 620 621 622 623
{
	u64 found_start;
	int found_level;
	struct extent_buffer *eb;
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
624
	struct btrfs_fs_info *fs_info = root->fs_info;
625
	int ret = 0;
626
	int reads_done;
627 628 629

	if (!page->private)
		goto out;
C
Chris Mason 已提交
630

J
Josef Bacik 已提交
631
	eb = (struct extent_buffer *)page->private;
C
Chris Mason 已提交
632

633 634 635 636 637 638
	/* 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);
639 640
	if (!reads_done)
		goto err;
641

642
	eb->read_mirror = mirror;
643
	if (test_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags)) {
644 645 646 647
		ret = -EIO;
		goto err;
	}

648
	found_start = btrfs_header_bytenr(eb);
649
	if (found_start != eb->start) {
650 651
		btrfs_err_rl(fs_info, "bad tree block start %llu %llu",
			     found_start, eb->start);
652
		ret = -EIO;
653 654
		goto err;
	}
655 656 657
	if (check_tree_block_fsid(fs_info, eb)) {
		btrfs_err_rl(fs_info, "bad fsid on block %llu",
			     eb->start);
658 659 660
		ret = -EIO;
		goto err;
	}
661
	found_level = btrfs_header_level(eb);
662
	if (found_level >= BTRFS_MAX_LEVEL) {
663 664
		btrfs_err(fs_info, "bad tree block level %d",
			  (int)btrfs_header_level(eb));
665 666 667
		ret = -EIO;
		goto err;
	}
668

669 670
	btrfs_set_buffer_lockdep_class(btrfs_header_owner(eb),
				       eb, found_level);
671

672
	ret = csum_tree_block(fs_info, eb, 1);
673
	if (ret)
674 675 676 677 678 679 680 681 682 683 684
		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;
	}
685

686 687
	if (!ret)
		set_extent_buffer_uptodate(eb);
688
err:
689 690
	if (reads_done &&
	    test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
691
		btree_readahead_hook(fs_info, eb, eb->start, ret);
A
Arne Jansen 已提交
692

D
David Woodhouse 已提交
693 694 695 696 697 698 699
	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);
700
		clear_extent_buffer_uptodate(eb);
D
David Woodhouse 已提交
701
	}
702
	free_extent_buffer(eb);
703
out:
704
	return ret;
705 706
}

707
static int btree_io_failed_hook(struct page *page, int failed_mirror)
A
Arne Jansen 已提交
708 709 710
{
	struct extent_buffer *eb;

J
Josef Bacik 已提交
711
	eb = (struct extent_buffer *)page->private;
712
	set_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags);
713
	eb->read_mirror = failed_mirror;
D
David Woodhouse 已提交
714
	atomic_dec(&eb->io_pages);
715
	if (test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
716
		btree_readahead_hook(eb->fs_info, eb, eb->start, -EIO);
A
Arne Jansen 已提交
717 718 719
	return -EIO;	/* we fixed nothing */
}

720
static void end_workqueue_bio(struct bio *bio)
721
{
722
	struct btrfs_end_io_wq *end_io_wq = bio->bi_private;
723
	struct btrfs_fs_info *fs_info;
724 725
	struct btrfs_workqueue *wq;
	btrfs_work_func_t func;
726 727

	fs_info = end_io_wq->info;
728
	end_io_wq->error = bio->bi_error;
729

730
	if (bio->bi_rw & REQ_WRITE) {
731 732 733 734 735 736 737 738 739 740 741 742 743
		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;
		}
744
	} else {
745 746 747 748 749
		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) {
750 751 752 753 754 755 756 757 758
			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;
		}
759
	}
760 761 762

	btrfs_init_work(&end_io_wq->work, func, end_workqueue_fn, NULL, NULL);
	btrfs_queue_work(wq, &end_io_wq->work);
763 764
}

765
int btrfs_bio_wq_end_io(struct btrfs_fs_info *info, struct bio *bio,
766
			enum btrfs_wq_endio_type metadata)
767
{
768
	struct btrfs_end_io_wq *end_io_wq;
769

770
	end_io_wq = kmem_cache_alloc(btrfs_end_io_wq_cache, GFP_NOFS);
771 772 773 774 775
	if (!end_io_wq)
		return -ENOMEM;

	end_io_wq->private = bio->bi_private;
	end_io_wq->end_io = bio->bi_end_io;
776
	end_io_wq->info = info;
777 778
	end_io_wq->error = 0;
	end_io_wq->bio = bio;
779
	end_io_wq->metadata = metadata;
780 781 782

	bio->bi_private = end_io_wq;
	bio->bi_end_io = end_workqueue_bio;
783 784 785
	return 0;
}

786
unsigned long btrfs_async_submit_limit(struct btrfs_fs_info *info)
787
{
788
	unsigned long limit = min_t(unsigned long,
789
				    info->thread_pool_size,
790 791 792
				    info->fs_devices->open_devices);
	return 256 * limit;
}
793

C
Chris Mason 已提交
794 795 796
static void run_one_async_start(struct btrfs_work *work)
{
	struct async_submit_bio *async;
797
	int ret;
C
Chris Mason 已提交
798 799

	async = container_of(work, struct  async_submit_bio, work);
800 801 802 803 804
	ret = async->submit_bio_start(async->inode, async->rw, async->bio,
				      async->mirror_num, async->bio_flags,
				      async->bio_offset);
	if (ret)
		async->error = ret;
C
Chris Mason 已提交
805 806 807
}

static void run_one_async_done(struct btrfs_work *work)
808 809 810
{
	struct btrfs_fs_info *fs_info;
	struct async_submit_bio *async;
811
	int limit;
812 813 814

	async = container_of(work, struct  async_submit_bio, work);
	fs_info = BTRFS_I(async->inode)->root->fs_info;
815

816
	limit = btrfs_async_submit_limit(fs_info);
817 818
	limit = limit * 2 / 3;

819 820 821
	/*
	 * atomic_dec_return implies a barrier for waitqueue_active
	 */
822
	if (atomic_dec_return(&fs_info->nr_async_submits) < limit &&
823
	    waitqueue_active(&fs_info->async_submit_wait))
824 825
		wake_up(&fs_info->async_submit_wait);

826
	/* If an error occurred we just want to clean up the bio and move on */
827
	if (async->error) {
828 829
		async->bio->bi_error = async->error;
		bio_endio(async->bio);
830 831 832
		return;
	}

C
Chris Mason 已提交
833
	async->submit_bio_done(async->inode, async->rw, async->bio,
834 835
			       async->mirror_num, async->bio_flags,
			       async->bio_offset);
C
Chris Mason 已提交
836 837 838 839 840 841 842
}

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

	async = container_of(work, struct  async_submit_bio, work);
843 844 845
	kfree(async);
}

846 847
int btrfs_wq_submit_bio(struct btrfs_fs_info *fs_info, struct inode *inode,
			int rw, struct bio *bio, int mirror_num,
C
Chris Mason 已提交
848
			unsigned long bio_flags,
849
			u64 bio_offset,
C
Chris Mason 已提交
850 851
			extent_submit_bio_hook_t *submit_bio_start,
			extent_submit_bio_hook_t *submit_bio_done)
852 853 854 855 856 857 858 859 860 861 862
{
	struct async_submit_bio *async;

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

	async->inode = inode;
	async->rw = rw;
	async->bio = bio;
	async->mirror_num = mirror_num;
C
Chris Mason 已提交
863 864 865
	async->submit_bio_start = submit_bio_start;
	async->submit_bio_done = submit_bio_done;

866
	btrfs_init_work(&async->work, btrfs_worker_helper, run_one_async_start,
867
			run_one_async_done, run_one_async_free);
C
Chris Mason 已提交
868

C
Chris Mason 已提交
869
	async->bio_flags = bio_flags;
870
	async->bio_offset = bio_offset;
871

872 873
	async->error = 0;

874
	atomic_inc(&fs_info->nr_async_submits);
875

876
	if (rw & REQ_SYNC)
877
		btrfs_set_work_high_priority(&async->work);
878

879
	btrfs_queue_work(fs_info->workers, &async->work);
880

C
Chris Mason 已提交
881
	while (atomic_read(&fs_info->async_submit_draining) &&
882 883 884 885 886
	      atomic_read(&fs_info->nr_async_submits)) {
		wait_event(fs_info->async_submit_wait,
			   (atomic_read(&fs_info->nr_async_submits) == 0));
	}

887 888 889
	return 0;
}

890 891
static int btree_csum_one_bio(struct bio *bio)
{
892
	struct bio_vec *bvec;
893
	struct btrfs_root *root;
894
	int i, ret = 0;
895

896
	bio_for_each_segment_all(bvec, bio, i) {
897
		root = BTRFS_I(bvec->bv_page->mapping->host)->root;
898
		ret = csum_dirty_buffer(root->fs_info, bvec->bv_page);
899 900
		if (ret)
			break;
901
	}
902

903
	return ret;
904 905
}

C
Chris Mason 已提交
906 907
static int __btree_submit_bio_start(struct inode *inode, int rw,
				    struct bio *bio, int mirror_num,
908 909
				    unsigned long bio_flags,
				    u64 bio_offset)
910
{
911 912
	/*
	 * when we're called for a write, we're already in the async
913
	 * submission context.  Just jump into btrfs_map_bio
914
	 */
915
	return btree_csum_one_bio(bio);
C
Chris Mason 已提交
916
}
917

C
Chris Mason 已提交
918
static int __btree_submit_bio_done(struct inode *inode, int rw, struct bio *bio,
919 920
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
C
Chris Mason 已提交
921
{
922 923
	int ret;

924
	/*
C
Chris Mason 已提交
925 926
	 * when we're called for a write, we're already in the async
	 * submission context.  Just jump into btrfs_map_bio
927
	 */
928
	ret = btrfs_map_bio(BTRFS_I(inode)->root, rw, bio, mirror_num, 1);
929 930 931 932
	if (ret) {
		bio->bi_error = ret;
		bio_endio(bio);
	}
933
	return ret;
934 935
}

936 937 938 939 940
static int check_async_write(struct inode *inode, unsigned long bio_flags)
{
	if (bio_flags & EXTENT_BIO_TREE_LOG)
		return 0;
#ifdef CONFIG_X86
941
	if (static_cpu_has(X86_FEATURE_XMM4_2))
942 943 944 945 946
		return 0;
#endif
	return 1;
}

947
static int btree_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
948 949
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
950
{
951
	int async = check_async_write(inode, bio_flags);
952 953
	int ret;

954
	if (!(rw & REQ_WRITE)) {
C
Chris Mason 已提交
955 956 957 958
		/*
		 * called for a read, do the setup so that checksum validation
		 * can happen in the async kernel threads
		 */
959
		ret = btrfs_bio_wq_end_io(BTRFS_I(inode)->root->fs_info,
960
					  bio, BTRFS_WQ_ENDIO_METADATA);
961
		if (ret)
962 963 964
			goto out_w_error;
		ret = btrfs_map_bio(BTRFS_I(inode)->root, rw, bio,
				    mirror_num, 0);
965 966 967
	} else if (!async) {
		ret = btree_csum_one_bio(bio);
		if (ret)
968 969 970 971 972 973 974 975 976 977 978 979 980
			goto out_w_error;
		ret = btrfs_map_bio(BTRFS_I(inode)->root, rw, bio,
				    mirror_num, 0);
	} else {
		/*
		 * kthread helpers are used to submit writes so that
		 * checksumming can happen in parallel across all CPUs
		 */
		ret = btrfs_wq_submit_bio(BTRFS_I(inode)->root->fs_info,
					  inode, rw, bio, mirror_num, 0,
					  bio_offset,
					  __btree_submit_bio_start,
					  __btree_submit_bio_done);
981
	}
982

983 984 985 986
	if (ret)
		goto out_w_error;
	return 0;

987
out_w_error:
988 989
	bio->bi_error = ret;
	bio_endio(bio);
990
	return ret;
991 992
}

J
Jan Beulich 已提交
993
#ifdef CONFIG_MIGRATION
994
static int btree_migratepage(struct address_space *mapping,
995 996
			struct page *newpage, struct page *page,
			enum migrate_mode mode)
997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010
{
	/*
	 * 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;
1011
	return migrate_page(mapping, newpage, page, mode);
1012
}
J
Jan Beulich 已提交
1013
#endif
1014

1015 1016 1017 1018

static int btree_writepages(struct address_space *mapping,
			    struct writeback_control *wbc)
{
1019 1020 1021
	struct btrfs_fs_info *fs_info;
	int ret;

1022
	if (wbc->sync_mode == WB_SYNC_NONE) {
1023 1024 1025 1026

		if (wbc->for_kupdate)
			return 0;

1027
		fs_info = BTRFS_I(mapping->host)->root->fs_info;
1028
		/* this is a bit racy, but that's ok */
1029 1030 1031
		ret = percpu_counter_compare(&fs_info->dirty_metadata_bytes,
					     BTRFS_DIRTY_METADATA_THRESH);
		if (ret < 0)
1032 1033
			return 0;
	}
1034
	return btree_write_cache_pages(mapping, wbc);
1035 1036
}

1037
static int btree_readpage(struct file *file, struct page *page)
1038
{
1039 1040
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
1041
	return extent_read_full_page(tree, page, btree_get_extent, 0);
1042
}
C
Chris Mason 已提交
1043

1044
static int btree_releasepage(struct page *page, gfp_t gfp_flags)
1045
{
1046
	if (PageWriteback(page) || PageDirty(page))
C
Chris Mason 已提交
1047
		return 0;
1048

1049
	return try_release_extent_buffer(page);
1050 1051
}

1052 1053
static void btree_invalidatepage(struct page *page, unsigned int offset,
				 unsigned int length)
1054
{
1055 1056
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
1057 1058
	extent_invalidatepage(tree, page, offset);
	btree_releasepage(page, GFP_NOFS);
1059
	if (PagePrivate(page)) {
1060 1061 1062
		btrfs_warn(BTRFS_I(page->mapping->host)->root->fs_info,
			   "page private not zero on page %llu",
			   (unsigned long long)page_offset(page));
1063 1064
		ClearPagePrivate(page);
		set_page_private(page, 0);
1065
		put_page(page);
1066
	}
1067 1068
}

1069 1070
static int btree_set_page_dirty(struct page *page)
{
1071
#ifdef DEBUG
1072 1073 1074 1075 1076 1077 1078 1079
	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);
1080
#endif
1081 1082 1083
	return __set_page_dirty_nobuffers(page);
}

1084
static const struct address_space_operations btree_aops = {
1085
	.readpage	= btree_readpage,
1086
	.writepages	= btree_writepages,
1087 1088
	.releasepage	= btree_releasepage,
	.invalidatepage = btree_invalidatepage,
1089
#ifdef CONFIG_MIGRATION
1090
	.migratepage	= btree_migratepage,
1091
#endif
1092
	.set_page_dirty = btree_set_page_dirty,
1093 1094
};

1095
void readahead_tree_block(struct btrfs_root *root, u64 bytenr)
C
Chris Mason 已提交
1096
{
1097 1098
	struct extent_buffer *buf = NULL;
	struct inode *btree_inode = root->fs_info->btree_inode;
C
Chris Mason 已提交
1099

1100
	buf = btrfs_find_create_tree_block(root, bytenr);
1101
	if (IS_ERR(buf))
1102
		return;
1103
	read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree,
1104
				 buf, 0, WAIT_NONE, btree_get_extent, 0);
1105
	free_extent_buffer(buf);
C
Chris Mason 已提交
1106 1107
}

1108
int reada_tree_block_flagged(struct btrfs_root *root, u64 bytenr,
1109 1110 1111 1112 1113 1114 1115
			 int mirror_num, struct extent_buffer **eb)
{
	struct extent_buffer *buf = NULL;
	struct inode *btree_inode = root->fs_info->btree_inode;
	struct extent_io_tree *io_tree = &BTRFS_I(btree_inode)->io_tree;
	int ret;

1116
	buf = btrfs_find_create_tree_block(root, bytenr);
1117
	if (IS_ERR(buf))
1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131
		return 0;

	set_bit(EXTENT_BUFFER_READAHEAD, &buf->bflags);

	ret = read_extent_buffer_pages(io_tree, buf, 0, WAIT_PAGE_LOCK,
				       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;
1132
	} else if (extent_buffer_uptodate(buf)) {
1133 1134 1135 1136 1137 1138 1139
		*eb = buf;
	} else {
		free_extent_buffer(buf);
	}
	return 0;
}

1140
struct extent_buffer *btrfs_find_tree_block(struct btrfs_fs_info *fs_info,
1141
					    u64 bytenr)
1142
{
1143
	return find_extent_buffer(fs_info, bytenr);
1144 1145 1146
}

struct extent_buffer *btrfs_find_create_tree_block(struct btrfs_root *root,
1147
						 u64 bytenr)
1148
{
1149
	if (btrfs_is_testing(root->fs_info))
1150 1151
		return alloc_test_extent_buffer(root->fs_info, bytenr,
				root->nodesize);
1152
	return alloc_extent_buffer(root->fs_info, bytenr);
1153 1154 1155
}


1156 1157
int btrfs_write_tree_block(struct extent_buffer *buf)
{
1158
	return filemap_fdatawrite_range(buf->pages[0]->mapping, buf->start,
1159
					buf->start + buf->len - 1);
1160 1161 1162 1163
}

int btrfs_wait_tree_block_writeback(struct extent_buffer *buf)
{
1164
	return filemap_fdatawait_range(buf->pages[0]->mapping,
1165
				       buf->start, buf->start + buf->len - 1);
1166 1167
}

1168
struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr,
1169
				      u64 parent_transid)
1170 1171 1172 1173
{
	struct extent_buffer *buf = NULL;
	int ret;

1174
	buf = btrfs_find_create_tree_block(root, bytenr);
1175 1176
	if (IS_ERR(buf))
		return buf;
1177

1178
	ret = btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
1179 1180
	if (ret) {
		free_extent_buffer(buf);
1181
		return ERR_PTR(ret);
1182
	}
1183
	return buf;
1184

1185 1186
}

1187 1188
void clean_tree_block(struct btrfs_trans_handle *trans,
		      struct btrfs_fs_info *fs_info,
1189
		      struct extent_buffer *buf)
1190
{
1191
	if (btrfs_header_generation(buf) ==
1192
	    fs_info->running_transaction->transid) {
1193
		btrfs_assert_tree_locked(buf);
1194

1195
		if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &buf->bflags)) {
1196 1197 1198
			__percpu_counter_add(&fs_info->dirty_metadata_bytes,
					     -buf->len,
					     fs_info->dirty_metadata_batch);
1199 1200 1201 1202
			/* ugh, clear_extent_buffer_dirty needs to lock the page */
			btrfs_set_lock_blocking(buf);
			clear_extent_buffer_dirty(buf);
		}
1203
	}
1204 1205
}

1206 1207 1208 1209 1210 1211 1212 1213 1214
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);

1215
	ret = percpu_counter_init(&writers->counter, 0, GFP_KERNEL);
1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231
	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);
}

1232 1233
static void __setup_root(u32 nodesize, u32 sectorsize, u32 stripesize,
			 struct btrfs_root *root, struct btrfs_fs_info *fs_info,
1234
			 u64 objectid)
1235
{
1236
	bool dummy = test_bit(BTRFS_FS_STATE_DUMMY_FS_INFO, &fs_info->fs_state);
C
Chris Mason 已提交
1237
	root->node = NULL;
1238
	root->commit_root = NULL;
1239 1240
	root->sectorsize = sectorsize;
	root->nodesize = nodesize;
1241
	root->stripesize = stripesize;
1242
	root->state = 0;
1243
	root->orphan_cleanup_state = 0;
1244

1245 1246
	root->objectid = objectid;
	root->last_trans = 0;
1247
	root->highest_objectid = 0;
1248
	root->nr_delalloc_inodes = 0;
1249
	root->nr_ordered_extents = 0;
1250
	root->name = NULL;
1251
	root->inode_tree = RB_ROOT;
1252
	INIT_RADIX_TREE(&root->delayed_nodes_tree, GFP_ATOMIC);
1253
	root->block_rsv = NULL;
1254
	root->orphan_block_rsv = NULL;
1255 1256

	INIT_LIST_HEAD(&root->dirty_list);
1257
	INIT_LIST_HEAD(&root->root_list);
1258 1259
	INIT_LIST_HEAD(&root->delalloc_inodes);
	INIT_LIST_HEAD(&root->delalloc_root);
1260 1261
	INIT_LIST_HEAD(&root->ordered_extents);
	INIT_LIST_HEAD(&root->ordered_root);
1262 1263
	INIT_LIST_HEAD(&root->logged_list[0]);
	INIT_LIST_HEAD(&root->logged_list[1]);
1264
	spin_lock_init(&root->orphan_lock);
1265
	spin_lock_init(&root->inode_lock);
1266
	spin_lock_init(&root->delalloc_lock);
1267
	spin_lock_init(&root->ordered_extent_lock);
1268
	spin_lock_init(&root->accounting_lock);
1269 1270
	spin_lock_init(&root->log_extents_lock[0]);
	spin_lock_init(&root->log_extents_lock[1]);
1271
	mutex_init(&root->objectid_mutex);
1272
	mutex_init(&root->log_mutex);
1273
	mutex_init(&root->ordered_extent_mutex);
1274
	mutex_init(&root->delalloc_mutex);
Y
Yan Zheng 已提交
1275 1276 1277
	init_waitqueue_head(&root->log_writer_wait);
	init_waitqueue_head(&root->log_commit_wait[0]);
	init_waitqueue_head(&root->log_commit_wait[1]);
1278 1279
	INIT_LIST_HEAD(&root->log_ctxs[0]);
	INIT_LIST_HEAD(&root->log_ctxs[1]);
Y
Yan Zheng 已提交
1280 1281 1282
	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
1283
	atomic_set(&root->log_batch, 0);
1284
	atomic_set(&root->orphan_inodes, 0);
1285
	atomic_set(&root->refs, 1);
1286
	atomic_set(&root->will_be_snapshoted, 0);
1287
	atomic_set(&root->qgroup_meta_rsv, 0);
Y
Yan Zheng 已提交
1288
	root->log_transid = 0;
1289
	root->log_transid_committed = -1;
1290
	root->last_log_commit = 0;
1291
	if (!dummy)
1292 1293
		extent_io_tree_init(&root->dirty_log_pages,
				     fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
1294

1295 1296
	memset(&root->root_key, 0, sizeof(root->root_key));
	memset(&root->root_item, 0, sizeof(root->root_item));
1297
	memset(&root->defrag_progress, 0, sizeof(root->defrag_progress));
1298
	if (!dummy)
1299 1300 1301
		root->defrag_trans_start = fs_info->generation;
	else
		root->defrag_trans_start = 0;
1302
	root->root_key.objectid = objectid;
1303
	root->anon_dev = 0;
1304

1305
	spin_lock_init(&root->root_item_lock);
1306 1307
}

1308 1309
static struct btrfs_root *btrfs_alloc_root(struct btrfs_fs_info *fs_info,
		gfp_t flags)
A
Al Viro 已提交
1310
{
1311
	struct btrfs_root *root = kzalloc(sizeof(*root), flags);
A
Al Viro 已提交
1312 1313 1314 1315 1316
	if (root)
		root->fs_info = fs_info;
	return root;
}

1317 1318
#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
/* Should only be used by the testing infrastructure */
1319 1320
struct btrfs_root *btrfs_alloc_dummy_root(struct btrfs_fs_info *fs_info,
					  u32 sectorsize, u32 nodesize)
1321 1322 1323
{
	struct btrfs_root *root;

1324 1325 1326 1327
	if (!fs_info)
		return ERR_PTR(-EINVAL);

	root = btrfs_alloc_root(fs_info, GFP_KERNEL);
1328 1329
	if (!root)
		return ERR_PTR(-ENOMEM);
1330
	/* We don't use the stripesize in selftest, set it as sectorsize */
1331
	__setup_root(nodesize, sectorsize, sectorsize, root, fs_info,
1332
			BTRFS_ROOT_TREE_OBJECTID);
1333
	root->alloc_bytenr = 0;
1334 1335 1336 1337 1338

	return root;
}
#endif

1339 1340 1341 1342 1343 1344 1345 1346 1347
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;
1348
	uuid_le uuid;
1349

1350
	root = btrfs_alloc_root(fs_info, GFP_KERNEL);
1351 1352 1353
	if (!root)
		return ERR_PTR(-ENOMEM);

1354 1355
	__setup_root(tree_root->nodesize, tree_root->sectorsize,
		tree_root->stripesize, root, fs_info, objectid);
1356 1357 1358 1359
	root->root_key.objectid = objectid;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = 0;

1360
	leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0);
1361 1362
	if (IS_ERR(leaf)) {
		ret = PTR_ERR(leaf);
1363
		leaf = NULL;
1364 1365 1366 1367 1368 1369 1370 1371 1372 1373
		goto fail;
	}

	memset_extent_buffer(leaf, 0, 0, sizeof(struct btrfs_header));
	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;

1374
	write_extent_buffer(leaf, fs_info->fsid, btrfs_header_fsid(),
1375 1376
			    BTRFS_FSID_SIZE);
	write_extent_buffer(leaf, fs_info->chunk_tree_uuid,
1377
			    btrfs_header_chunk_tree_uuid(leaf),
1378 1379 1380 1381
			    BTRFS_UUID_SIZE);
	btrfs_mark_buffer_dirty(leaf);

	root->commit_root = btrfs_root_node(root);
1382
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
1383 1384 1385 1386 1387 1388 1389 1390 1391 1392

	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);
1393 1394
	uuid_le_gen(&uuid);
	memcpy(root->root_item.uuid, uuid.b, BTRFS_UUID_SIZE);
1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405
	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);

1406 1407
	return root;

1408
fail:
1409 1410
	if (leaf) {
		btrfs_tree_unlock(leaf);
1411
		free_extent_buffer(root->commit_root);
1412 1413 1414
		free_extent_buffer(leaf);
	}
	kfree(root);
1415

1416
	return ERR_PTR(ret);
1417 1418
}

Y
Yan Zheng 已提交
1419 1420
static struct btrfs_root *alloc_log_tree(struct btrfs_trans_handle *trans,
					 struct btrfs_fs_info *fs_info)
1421 1422 1423
{
	struct btrfs_root *root;
	struct btrfs_root *tree_root = fs_info->tree_root;
Y
Yan Zheng 已提交
1424
	struct extent_buffer *leaf;
1425

1426
	root = btrfs_alloc_root(fs_info, GFP_NOFS);
1427
	if (!root)
Y
Yan Zheng 已提交
1428
		return ERR_PTR(-ENOMEM);
1429

1430 1431 1432
	__setup_root(tree_root->nodesize, tree_root->sectorsize,
		     tree_root->stripesize, root, fs_info,
		     BTRFS_TREE_LOG_OBJECTID);
1433 1434 1435 1436

	root->root_key.objectid = BTRFS_TREE_LOG_OBJECTID;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = BTRFS_TREE_LOG_OBJECTID;
1437

Y
Yan Zheng 已提交
1438
	/*
1439 1440
	 * DON'T set REF_COWS for log trees
	 *
Y
Yan Zheng 已提交
1441 1442 1443 1444 1445
	 * 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).
	 */
1446

1447 1448
	leaf = btrfs_alloc_tree_block(trans, root, 0, BTRFS_TREE_LOG_OBJECTID,
			NULL, 0, 0, 0);
Y
Yan Zheng 已提交
1449 1450 1451 1452
	if (IS_ERR(leaf)) {
		kfree(root);
		return ERR_CAST(leaf);
	}
1453

1454 1455 1456 1457 1458
	memset_extent_buffer(leaf, 0, 0, sizeof(struct btrfs_header));
	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
1459
	root->node = leaf;
1460 1461

	write_extent_buffer(root->node, root->fs_info->fsid,
1462
			    btrfs_header_fsid(), BTRFS_FSID_SIZE);
1463 1464
	btrfs_mark_buffer_dirty(root->node);
	btrfs_tree_unlock(root->node);
Y
Yan Zheng 已提交
1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494
	return root;
}

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

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

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

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

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

	inode_item = &log_root->root_item.inode;
1495 1496 1497
	btrfs_set_stack_inode_generation(inode_item, 1);
	btrfs_set_stack_inode_size(inode_item, 3);
	btrfs_set_stack_inode_nlink(inode_item, 1);
1498
	btrfs_set_stack_inode_nbytes(inode_item, root->nodesize);
1499
	btrfs_set_stack_inode_mode(inode_item, S_IFDIR | 0755);
Y
Yan Zheng 已提交
1500

1501
	btrfs_set_root_node(&log_root->root_item, log_root->node);
Y
Yan Zheng 已提交
1502 1503 1504 1505

	WARN_ON(root->log_root);
	root->log_root = log_root;
	root->log_transid = 0;
1506
	root->log_transid_committed = -1;
1507
	root->last_log_commit = 0;
1508 1509 1510
	return 0;
}

1511 1512
static struct btrfs_root *btrfs_read_tree_root(struct btrfs_root *tree_root,
					       struct btrfs_key *key)
1513 1514 1515
{
	struct btrfs_root *root;
	struct btrfs_fs_info *fs_info = tree_root->fs_info;
1516
	struct btrfs_path *path;
1517
	u64 generation;
1518
	int ret;
1519

1520 1521
	path = btrfs_alloc_path();
	if (!path)
1522
		return ERR_PTR(-ENOMEM);
1523

1524
	root = btrfs_alloc_root(fs_info, GFP_NOFS);
1525 1526 1527
	if (!root) {
		ret = -ENOMEM;
		goto alloc_fail;
1528 1529
	}

1530 1531
	__setup_root(tree_root->nodesize, tree_root->sectorsize,
		tree_root->stripesize, root, fs_info, key->objectid);
1532

1533 1534
	ret = btrfs_find_root(tree_root, key, path,
			      &root->root_item, &root->root_key);
1535
	if (ret) {
1536 1537
		if (ret > 0)
			ret = -ENOENT;
1538
		goto find_fail;
1539
	}
1540

1541
	generation = btrfs_root_generation(&root->root_item);
1542
	root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
1543
				     generation);
1544 1545
	if (IS_ERR(root->node)) {
		ret = PTR_ERR(root->node);
1546 1547 1548
		goto find_fail;
	} else if (!btrfs_buffer_uptodate(root->node, generation, 0)) {
		ret = -EIO;
1549 1550
		free_extent_buffer(root->node);
		goto find_fail;
1551
	}
1552
	root->commit_root = btrfs_root_node(root);
1553
out:
1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573
	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) {
1574
		set_bit(BTRFS_ROOT_REF_COWS, &root->state);
1575 1576
		btrfs_check_and_init_root_item(&root->root_item);
	}
1577

1578 1579 1580
	return root;
}

1581 1582 1583
int btrfs_init_fs_root(struct btrfs_root *root)
{
	int ret;
1584
	struct btrfs_subvolume_writers *writers;
1585 1586 1587 1588 1589 1590 1591 1592 1593

	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;
	}

1594 1595 1596 1597 1598 1599 1600
	writers = btrfs_alloc_subvolume_writers();
	if (IS_ERR(writers)) {
		ret = PTR_ERR(writers);
		goto fail;
	}
	root->subv_writers = writers;

1601
	btrfs_init_free_ino_ctl(root);
1602 1603
	spin_lock_init(&root->ino_cache_lock);
	init_waitqueue_head(&root->ino_cache_wait);
1604 1605 1606

	ret = get_anon_bdev(&root->anon_dev);
	if (ret)
L
Liu Bo 已提交
1607
		goto fail;
1608 1609 1610 1611 1612 1613

	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
1614
		goto fail;
1615 1616 1617 1618 1619 1620
	}

	ASSERT(root->highest_objectid <= BTRFS_LAST_FREE_OBJECTID);

	mutex_unlock(&root->objectid_mutex);

1621 1622
	return 0;
fail:
L
Liu Bo 已提交
1623
	/* the caller is responsible to call free_fs_root */
1624 1625 1626
	return ret;
}

1627 1628
static struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info,
					       u64 root_id)
1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643
{
	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;

1644
	ret = radix_tree_preload(GFP_NOFS);
1645 1646 1647 1648 1649 1650 1651 1652
	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)
1653
		set_bit(BTRFS_ROOT_IN_RADIX, &root->state);
1654 1655 1656 1657 1658 1659
	spin_unlock(&fs_info->fs_roots_radix_lock);
	radix_tree_preload_end();

	return ret;
}

1660 1661 1662
struct btrfs_root *btrfs_get_fs_root(struct btrfs_fs_info *fs_info,
				     struct btrfs_key *location,
				     bool check_ref)
1663 1664
{
	struct btrfs_root *root;
1665
	struct btrfs_path *path;
1666
	struct btrfs_key key;
1667 1668
	int ret;

1669 1670 1671 1672
	if (location->objectid == BTRFS_ROOT_TREE_OBJECTID)
		return fs_info->tree_root;
	if (location->objectid == BTRFS_EXTENT_TREE_OBJECTID)
		return fs_info->extent_root;
1673 1674 1675 1676
	if (location->objectid == BTRFS_CHUNK_TREE_OBJECTID)
		return fs_info->chunk_root;
	if (location->objectid == BTRFS_DEV_TREE_OBJECTID)
		return fs_info->dev_root;
1677 1678
	if (location->objectid == BTRFS_CSUM_TREE_OBJECTID)
		return fs_info->csum_root;
1679 1680 1681
	if (location->objectid == BTRFS_QUOTA_TREE_OBJECTID)
		return fs_info->quota_root ? fs_info->quota_root :
					     ERR_PTR(-ENOENT);
1682 1683 1684
	if (location->objectid == BTRFS_UUID_TREE_OBJECTID)
		return fs_info->uuid_root ? fs_info->uuid_root :
					    ERR_PTR(-ENOENT);
1685 1686 1687
	if (location->objectid == BTRFS_FREE_SPACE_TREE_OBJECTID)
		return fs_info->free_space_root ? fs_info->free_space_root :
						  ERR_PTR(-ENOENT);
1688
again:
1689
	root = btrfs_lookup_fs_root(fs_info, location->objectid);
1690
	if (root) {
1691
		if (check_ref && btrfs_root_refs(&root->root_item) == 0)
1692
			return ERR_PTR(-ENOENT);
1693
		return root;
1694
	}
1695

1696
	root = btrfs_read_fs_root(fs_info->tree_root, location);
1697 1698
	if (IS_ERR(root))
		return root;
1699

1700
	if (check_ref && btrfs_root_refs(&root->root_item) == 0) {
1701
		ret = -ENOENT;
1702
		goto fail;
1703
	}
1704

1705
	ret = btrfs_init_fs_root(root);
1706 1707
	if (ret)
		goto fail;
1708

1709 1710 1711 1712 1713
	path = btrfs_alloc_path();
	if (!path) {
		ret = -ENOMEM;
		goto fail;
	}
1714 1715 1716 1717 1718
	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);
1719
	btrfs_free_path(path);
1720 1721 1722
	if (ret < 0)
		goto fail;
	if (ret == 0)
1723
		set_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &root->state);
1724

1725
	ret = btrfs_insert_fs_root(fs_info, root);
1726
	if (ret) {
1727 1728 1729 1730 1731
		if (ret == -EEXIST) {
			free_fs_root(root);
			goto again;
		}
		goto fail;
1732
	}
1733
	return root;
1734 1735 1736
fail:
	free_fs_root(root);
	return ERR_PTR(ret);
1737 1738
}

C
Chris Mason 已提交
1739 1740 1741 1742 1743 1744
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
1745

1746 1747
	rcu_read_lock();
	list_for_each_entry_rcu(device, &info->fs_devices->devices, dev_list) {
1748 1749
		if (!device->bdev)
			continue;
C
Chris Mason 已提交
1750
		bdi = blk_get_backing_dev_info(device->bdev);
1751
		if (bdi_congested(bdi, bdi_bits)) {
C
Chris Mason 已提交
1752 1753 1754 1755
			ret = 1;
			break;
		}
	}
1756
	rcu_read_unlock();
C
Chris Mason 已提交
1757 1758 1759 1760 1761
	return ret;
}

static int setup_bdi(struct btrfs_fs_info *info, struct backing_dev_info *bdi)
{
1762 1763
	int err;

1764
	err = bdi_setup_and_register(bdi, "btrfs");
1765 1766 1767
	if (err)
		return err;

1768
	bdi->ra_pages = VM_MAX_READAHEAD * 1024 / PAGE_SIZE;
C
Chris Mason 已提交
1769 1770
	bdi->congested_fn	= btrfs_congested_fn;
	bdi->congested_data	= info;
1771
	bdi->capabilities |= BDI_CAP_CGROUP_WRITEBACK;
C
Chris Mason 已提交
1772 1773 1774
	return 0;
}

1775 1776 1777 1778 1779
/*
 * 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)
1780 1781
{
	struct bio *bio;
1782
	struct btrfs_end_io_wq *end_io_wq;
1783

1784
	end_io_wq = container_of(work, struct btrfs_end_io_wq, work);
1785
	bio = end_io_wq->bio;
1786

1787
	bio->bi_error = end_io_wq->error;
1788 1789
	bio->bi_private = end_io_wq->private;
	bio->bi_end_io = end_io_wq->end_io;
1790
	kmem_cache_free(btrfs_end_io_wq_cache, end_io_wq);
1791
	bio_endio(bio);
1792 1793
}

1794 1795 1796
static int cleaner_kthread(void *arg)
{
	struct btrfs_root *root = arg;
1797
	int again;
1798
	struct btrfs_trans_handle *trans;
1799 1800

	do {
1801
		again = 0;
1802

1803
		/* Make the cleaner go to sleep early. */
1804
		if (btrfs_need_cleaner_sleep(root))
1805 1806
			goto sleep;

1807 1808 1809 1810 1811 1812 1813
		/*
		 * Do not do anything if we might cause open_ctree() to block
		 * before we have finished mounting the filesystem.
		 */
		if (!root->fs_info->open)
			goto sleep;

1814 1815 1816
		if (!mutex_trylock(&root->fs_info->cleaner_mutex))
			goto sleep;

1817 1818 1819 1820
		/*
		 * Avoid the problem that we change the status of the fs
		 * during the above check and trylock.
		 */
1821
		if (btrfs_need_cleaner_sleep(root)) {
1822 1823
			mutex_unlock(&root->fs_info->cleaner_mutex);
			goto sleep;
1824
		}
1825

1826
		mutex_lock(&root->fs_info->cleaner_delayed_iput_mutex);
1827
		btrfs_run_delayed_iputs(root);
1828 1829
		mutex_unlock(&root->fs_info->cleaner_delayed_iput_mutex);

1830 1831 1832 1833
		again = btrfs_clean_one_deleted_snapshot(root);
		mutex_unlock(&root->fs_info->cleaner_mutex);

		/*
1834 1835
		 * The defragger has dealt with the R/O remount and umount,
		 * needn't do anything special here.
1836 1837
		 */
		btrfs_run_defrag_inodes(root->fs_info);
1838 1839 1840 1841 1842 1843 1844 1845 1846 1847

		/*
		 * Acquires fs_info->delete_unused_bgs_mutex to avoid racing
		 * with relocation (btrfs_relocate_chunk) and relocation
		 * acquires fs_info->cleaner_mutex (btrfs_relocate_block_group)
		 * after acquiring fs_info->delete_unused_bgs_mutex. So we
		 * can't hold, nor need to, fs_info->cleaner_mutex when deleting
		 * unused block groups.
		 */
		btrfs_delete_unused_bgs(root->fs_info);
1848
sleep:
1849
		if (!again) {
1850
			set_current_state(TASK_INTERRUPTIBLE);
1851 1852
			if (!kthread_should_stop())
				schedule();
1853 1854 1855
			__set_current_state(TASK_RUNNING);
		}
	} while (!kthread_should_stop());
1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883

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

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

1884 1885 1886 1887 1888 1889 1890 1891
	return 0;
}

static int transaction_kthread(void *arg)
{
	struct btrfs_root *root = arg;
	struct btrfs_trans_handle *trans;
	struct btrfs_transaction *cur;
1892
	u64 transid;
1893 1894
	unsigned long now;
	unsigned long delay;
1895
	bool cannot_commit;
1896 1897

	do {
1898
		cannot_commit = false;
1899
		delay = HZ * root->fs_info->commit_interval;
1900 1901
		mutex_lock(&root->fs_info->transaction_kthread_mutex);

J
Josef Bacik 已提交
1902
		spin_lock(&root->fs_info->trans_lock);
1903 1904
		cur = root->fs_info->running_transaction;
		if (!cur) {
J
Josef Bacik 已提交
1905
			spin_unlock(&root->fs_info->trans_lock);
1906 1907
			goto sleep;
		}
Y
Yan Zheng 已提交
1908

1909
		now = get_seconds();
1910
		if (cur->state < TRANS_STATE_BLOCKED &&
1911 1912
		    (now < cur->start_time ||
		     now - cur->start_time < root->fs_info->commit_interval)) {
J
Josef Bacik 已提交
1913
			spin_unlock(&root->fs_info->trans_lock);
1914 1915 1916
			delay = HZ * 5;
			goto sleep;
		}
1917
		transid = cur->transid;
J
Josef Bacik 已提交
1918
		spin_unlock(&root->fs_info->trans_lock);
1919

1920
		/* If the file system is aborted, this will always fail. */
1921
		trans = btrfs_attach_transaction(root);
1922
		if (IS_ERR(trans)) {
1923 1924
			if (PTR_ERR(trans) != -ENOENT)
				cannot_commit = true;
1925
			goto sleep;
1926
		}
1927
		if (transid == trans->transid) {
1928
			btrfs_commit_transaction(trans, root);
1929 1930 1931
		} else {
			btrfs_end_transaction(trans, root);
		}
1932 1933 1934 1935
sleep:
		wake_up_process(root->fs_info->cleaner_kthread);
		mutex_unlock(&root->fs_info->transaction_kthread_mutex);

J
Josef Bacik 已提交
1936 1937 1938
		if (unlikely(test_bit(BTRFS_FS_STATE_ERROR,
				      &root->fs_info->fs_state)))
			btrfs_cleanup_transaction(root);
1939 1940 1941 1942 1943 1944
		set_current_state(TASK_INTERRUPTIBLE);
		if (!kthread_should_stop() &&
				(!btrfs_transaction_blocked(root->fs_info) ||
				 cannot_commit))
			schedule_timeout(delay);
		__set_current_state(TASK_RUNNING);
1945 1946 1947 1948
	} while (!kthread_should_stop());
	return 0;
}

C
Chris Mason 已提交
1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 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
/*
 * 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));

2055 2056 2057 2058 2059 2060 2061 2062
	/*
	 * 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
2063
			       btrfs_header_generation(info->fs_root->node));
2064
		btrfs_set_backup_fs_root_level(root_backup,
C
Chris Mason 已提交
2065
			       btrfs_header_level(info->fs_root->node));
2066
	}
C
Chris Mason 已提交
2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147

	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
2148 2149 2150
/* helper to cleanup workers */
static void btrfs_stop_all_workers(struct btrfs_fs_info *fs_info)
{
2151
	btrfs_destroy_workqueue(fs_info->fixup_workers);
2152
	btrfs_destroy_workqueue(fs_info->delalloc_workers);
2153
	btrfs_destroy_workqueue(fs_info->workers);
2154 2155 2156
	btrfs_destroy_workqueue(fs_info->endio_workers);
	btrfs_destroy_workqueue(fs_info->endio_meta_workers);
	btrfs_destroy_workqueue(fs_info->endio_raid56_workers);
2157
	btrfs_destroy_workqueue(fs_info->endio_repair_workers);
2158
	btrfs_destroy_workqueue(fs_info->rmw_workers);
2159 2160 2161
	btrfs_destroy_workqueue(fs_info->endio_meta_write_workers);
	btrfs_destroy_workqueue(fs_info->endio_write_workers);
	btrfs_destroy_workqueue(fs_info->endio_freespace_worker);
2162
	btrfs_destroy_workqueue(fs_info->submit_workers);
2163
	btrfs_destroy_workqueue(fs_info->delayed_workers);
2164
	btrfs_destroy_workqueue(fs_info->caching_workers);
2165
	btrfs_destroy_workqueue(fs_info->readahead_workers);
2166
	btrfs_destroy_workqueue(fs_info->flush_workers);
2167
	btrfs_destroy_workqueue(fs_info->qgroup_rescan_workers);
C
Chris Mason 已提交
2168
	btrfs_destroy_workqueue(fs_info->extent_workers);
L
Liu Bo 已提交
2169 2170
}

2171 2172 2173 2174 2175 2176 2177 2178 2179 2180
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
2181 2182 2183
/* helper to cleanup tree roots */
static void free_root_pointers(struct btrfs_fs_info *info, int chunk_root)
{
2184
	free_root_extent_buffers(info->tree_root);
2185

2186 2187 2188 2189 2190 2191 2192
	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);
2193
	free_root_extent_buffers(info->free_space_root);
C
Chris Mason 已提交
2194 2195
}

2196
void btrfs_free_fs_roots(struct btrfs_fs_info *fs_info)
2197 2198 2199 2200 2201 2202 2203 2204 2205 2206
{
	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);

2207
		if (test_bit(BTRFS_ROOT_IN_RADIX, &gang[0]->state)) {
2208
			btrfs_drop_and_free_fs_root(fs_info, gang[0]);
2209 2210 2211
		} else {
			free_extent_buffer(gang[0]->node);
			free_extent_buffer(gang[0]->commit_root);
2212
			btrfs_put_fs_root(gang[0]);
2213 2214 2215 2216 2217 2218 2219 2220 2221 2222
		}
	}

	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++)
2223
			btrfs_drop_and_free_fs_root(fs_info, gang[i]);
2224
	}
2225 2226 2227 2228 2229 2230

	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
		btrfs_free_log_root_tree(NULL, fs_info);
		btrfs_destroy_pinned_extent(fs_info->tree_root,
					    fs_info->pinned_extents);
	}
2231
}
C
Chris Mason 已提交
2232

2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243
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;
}

2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254
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);
}

2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283
static void btrfs_init_btree_inode(struct btrfs_fs_info *fs_info,
				   struct btrfs_root *tree_root)
{
	fs_info->btree_inode->i_ino = BTRFS_BTREE_INODE_OBJECTID;
	set_nlink(fs_info->btree_inode, 1);
	/*
	 * we set the i_size on the btree inode to the max possible int.
	 * the real end of the address space is determined by all of
	 * the devices in the system
	 */
	fs_info->btree_inode->i_size = OFFSET_MAX;
	fs_info->btree_inode->i_mapping->a_ops = &btree_aops;

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

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

	BTRFS_I(fs_info->btree_inode)->root = tree_root;
	memset(&BTRFS_I(fs_info->btree_inode)->location, 0,
	       sizeof(struct btrfs_key));
	set_bit(BTRFS_INODE_DUMMY,
		&BTRFS_I(fs_info->btree_inode)->runtime_flags);
	btrfs_insert_inode_hash(fs_info->btree_inode);
}

2284 2285 2286 2287 2288
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);
2289 2290 2291
	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);
2292
	init_waitqueue_head(&fs_info->replace_wait);
2293
	init_waitqueue_head(&fs_info->dev_replace.read_lock_wq);
2294 2295
}

2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306
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->quota_enabled = 0;
	fs_info->pending_quota_state = 0;
	fs_info->qgroup_ulist = NULL;
2307
	fs_info->qgroup_rescan_running = false;
2308 2309 2310
	mutex_init(&fs_info->qgroup_rescan_lock);
}

2311 2312 2313 2314
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;
2315
	unsigned int flags = WQ_MEM_RECLAIM | WQ_FREEZABLE | WQ_UNBOUND;
2316 2317

	fs_info->workers =
2318 2319
		btrfs_alloc_workqueue(fs_info, "worker",
				      flags | WQ_HIGHPRI, max_active, 16);
2320 2321

	fs_info->delalloc_workers =
2322 2323
		btrfs_alloc_workqueue(fs_info, "delalloc",
				      flags, max_active, 2);
2324 2325

	fs_info->flush_workers =
2326 2327
		btrfs_alloc_workqueue(fs_info, "flush_delalloc",
				      flags, max_active, 0);
2328 2329

	fs_info->caching_workers =
2330
		btrfs_alloc_workqueue(fs_info, "cache", flags, max_active, 0);
2331 2332 2333 2334 2335 2336 2337

	/*
	 * 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 =
2338
		btrfs_alloc_workqueue(fs_info, "submit", flags,
2339 2340 2341 2342
				      min_t(u64, fs_devices->num_devices,
					    max_active), 64);

	fs_info->fixup_workers =
2343
		btrfs_alloc_workqueue(fs_info, "fixup", flags, 1, 0);
2344 2345 2346 2347 2348 2349

	/*
	 * endios are largely parallel and should have a very
	 * low idle thresh
	 */
	fs_info->endio_workers =
2350
		btrfs_alloc_workqueue(fs_info, "endio", flags, max_active, 4);
2351
	fs_info->endio_meta_workers =
2352 2353
		btrfs_alloc_workqueue(fs_info, "endio-meta", flags,
				      max_active, 4);
2354
	fs_info->endio_meta_write_workers =
2355 2356
		btrfs_alloc_workqueue(fs_info, "endio-meta-write", flags,
				      max_active, 2);
2357
	fs_info->endio_raid56_workers =
2358 2359
		btrfs_alloc_workqueue(fs_info, "endio-raid56", flags,
				      max_active, 4);
2360
	fs_info->endio_repair_workers =
2361
		btrfs_alloc_workqueue(fs_info, "endio-repair", flags, 1, 0);
2362
	fs_info->rmw_workers =
2363
		btrfs_alloc_workqueue(fs_info, "rmw", flags, max_active, 2);
2364
	fs_info->endio_write_workers =
2365 2366
		btrfs_alloc_workqueue(fs_info, "endio-write", flags,
				      max_active, 2);
2367
	fs_info->endio_freespace_worker =
2368 2369
		btrfs_alloc_workqueue(fs_info, "freespace-write", flags,
				      max_active, 0);
2370
	fs_info->delayed_workers =
2371 2372
		btrfs_alloc_workqueue(fs_info, "delayed-meta", flags,
				      max_active, 0);
2373
	fs_info->readahead_workers =
2374 2375
		btrfs_alloc_workqueue(fs_info, "readahead", flags,
				      max_active, 2);
2376
	fs_info->qgroup_rescan_workers =
2377
		btrfs_alloc_workqueue(fs_info, "qgroup-rescan", flags, 1, 0);
2378
	fs_info->extent_workers =
2379
		btrfs_alloc_workqueue(fs_info, "extent-refs", flags,
2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399
				      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;
}

2400 2401 2402 2403 2404 2405 2406 2407 2408 2409
static int btrfs_replay_log(struct btrfs_fs_info *fs_info,
			    struct btrfs_fs_devices *fs_devices)
{
	int ret;
	struct btrfs_root *tree_root = fs_info->tree_root;
	struct btrfs_root *log_tree_root;
	struct btrfs_super_block *disk_super = fs_info->super_copy;
	u64 bytenr = btrfs_super_log_root(disk_super);

	if (fs_devices->rw_devices == 0) {
2410
		btrfs_warn(fs_info, "log replay required on RO media");
2411 2412 2413
		return -EIO;
	}

2414
	log_tree_root = btrfs_alloc_root(fs_info, GFP_KERNEL);
2415 2416 2417 2418 2419 2420 2421 2422 2423
	if (!log_tree_root)
		return -ENOMEM;

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

	log_tree_root->node = read_tree_block(tree_root, bytenr,
			fs_info->generation + 1);
2424
	if (IS_ERR(log_tree_root->node)) {
2425
		btrfs_warn(fs_info, "failed to read log tree");
2426
		ret = PTR_ERR(log_tree_root->node);
2427
		kfree(log_tree_root);
2428
		return ret;
2429
	} else if (!extent_buffer_uptodate(log_tree_root->node)) {
2430
		btrfs_err(fs_info, "failed to read log tree");
2431 2432 2433 2434 2435 2436 2437
		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) {
2438
		btrfs_handle_fs_error(tree_root->fs_info, ret,
2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453
			    "Failed to recover log tree");
		free_extent_buffer(log_tree_root->node);
		kfree(log_tree_root);
		return ret;
	}

	if (fs_info->sb->s_flags & MS_RDONLY) {
		ret = btrfs_commit_super(tree_root);
		if (ret)
			return ret;
	}

	return 0;
}

2454 2455 2456
static int btrfs_read_roots(struct btrfs_fs_info *fs_info,
			    struct btrfs_root *tree_root)
{
2457
	struct btrfs_root *root;
2458 2459 2460 2461 2462 2463 2464
	struct btrfs_key location;
	int ret;

	location.objectid = BTRFS_EXTENT_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
	location.offset = 0;

2465 2466 2467 2468 2469
	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;
2470 2471

	location.objectid = BTRFS_DEV_TREE_OBJECTID;
2472 2473 2474 2475 2476
	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;
2477 2478 2479
	btrfs_init_devices_late(fs_info);

	location.objectid = BTRFS_CSUM_TREE_OBJECTID;
2480 2481 2482 2483 2484
	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;
2485 2486

	location.objectid = BTRFS_QUOTA_TREE_OBJECTID;
2487 2488 2489
	root = btrfs_read_tree_root(tree_root, &location);
	if (!IS_ERR(root)) {
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
2490 2491
		fs_info->quota_enabled = 1;
		fs_info->pending_quota_state = 1;
2492
		fs_info->quota_root = root;
2493 2494 2495
	}

	location.objectid = BTRFS_UUID_TREE_OBJECTID;
2496 2497 2498
	root = btrfs_read_tree_root(tree_root, &location);
	if (IS_ERR(root)) {
		ret = PTR_ERR(root);
2499 2500 2501
		if (ret != -ENOENT)
			return ret;
	} else {
2502 2503
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
		fs_info->uuid_root = root;
2504 2505
	}

2506 2507 2508 2509 2510 2511 2512 2513 2514
	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;
	}

2515 2516 2517
	return 0;
}

A
Al Viro 已提交
2518 2519 2520
int open_ctree(struct super_block *sb,
	       struct btrfs_fs_devices *fs_devices,
	       char *options)
2521
{
2522 2523
	u32 sectorsize;
	u32 nodesize;
2524
	u32 stripesize;
2525
	u64 generation;
2526
	u64 features;
2527
	struct btrfs_key location;
2528
	struct buffer_head *bh;
2529
	struct btrfs_super_block *disk_super;
2530
	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
2531
	struct btrfs_root *tree_root;
2532
	struct btrfs_root *chunk_root;
2533
	int ret;
2534
	int err = -EINVAL;
C
Chris Mason 已提交
2535 2536
	int num_backups_tried = 0;
	int backup_index = 0;
2537
	int max_active;
2538

2539 2540
	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);
2541
	if (!tree_root || !chunk_root) {
C
Chris Mason 已提交
2542 2543 2544
		err = -ENOMEM;
		goto fail;
	}
2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557

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

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

2558
	ret = percpu_counter_init(&fs_info->dirty_metadata_bytes, 0, GFP_KERNEL);
2559 2560 2561 2562
	if (ret) {
		err = ret;
		goto fail_bdi;
	}
2563
	fs_info->dirty_metadata_batch = PAGE_SIZE *
2564 2565
					(1 + ilog2(nr_cpu_ids));

2566
	ret = percpu_counter_init(&fs_info->delalloc_bytes, 0, GFP_KERNEL);
2567 2568 2569 2570 2571
	if (ret) {
		err = ret;
		goto fail_dirty_metadata_bytes;
	}

2572
	ret = percpu_counter_init(&fs_info->bio_counter, 0, GFP_KERNEL);
2573 2574 2575 2576 2577
	if (ret) {
		err = ret;
		goto fail_delalloc_bytes;
	}

2578 2579 2580
	fs_info->btree_inode = new_inode(sb);
	if (!fs_info->btree_inode) {
		err = -ENOMEM;
2581
		goto fail_bio_counter;
2582 2583
	}

2584
	mapping_set_gfp_mask(fs_info->btree_inode->i_mapping, GFP_NOFS);
2585

2586
	INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC);
2587
	INIT_RADIX_TREE(&fs_info->buffer_radix, GFP_ATOMIC);
C
Chris Mason 已提交
2588
	INIT_LIST_HEAD(&fs_info->trans_list);
2589
	INIT_LIST_HEAD(&fs_info->dead_roots);
Y
Yan, Zheng 已提交
2590
	INIT_LIST_HEAD(&fs_info->delayed_iputs);
2591
	INIT_LIST_HEAD(&fs_info->delalloc_roots);
2592
	INIT_LIST_HEAD(&fs_info->caching_block_groups);
2593
	spin_lock_init(&fs_info->delalloc_root_lock);
J
Josef Bacik 已提交
2594
	spin_lock_init(&fs_info->trans_lock);
2595
	spin_lock_init(&fs_info->fs_roots_radix_lock);
Y
Yan, Zheng 已提交
2596
	spin_lock_init(&fs_info->delayed_iput_lock);
C
Chris Mason 已提交
2597
	spin_lock_init(&fs_info->defrag_inodes_lock);
2598
	spin_lock_init(&fs_info->free_chunk_lock);
J
Jan Schmidt 已提交
2599
	spin_lock_init(&fs_info->tree_mod_seq_lock);
2600
	spin_lock_init(&fs_info->super_lock);
J
Josef Bacik 已提交
2601
	spin_lock_init(&fs_info->qgroup_op_lock);
2602
	spin_lock_init(&fs_info->buffer_lock);
2603
	spin_lock_init(&fs_info->unused_bgs_lock);
J
Jan Schmidt 已提交
2604
	rwlock_init(&fs_info->tree_mod_log_lock);
2605
	mutex_init(&fs_info->unused_bg_unpin_mutex);
2606
	mutex_init(&fs_info->delete_unused_bgs_mutex);
C
Chris Mason 已提交
2607
	mutex_init(&fs_info->reloc_mutex);
2608
	mutex_init(&fs_info->delalloc_root_mutex);
2609
	mutex_init(&fs_info->cleaner_delayed_iput_mutex);
2610
	seqlock_init(&fs_info->profiles_lock);
2611

2612
	INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots);
2613
	INIT_LIST_HEAD(&fs_info->space_info);
J
Jan Schmidt 已提交
2614
	INIT_LIST_HEAD(&fs_info->tree_mod_seq_list);
2615
	INIT_LIST_HEAD(&fs_info->unused_bgs);
2616
	btrfs_mapping_init(&fs_info->mapping_tree);
2617 2618 2619 2620 2621 2622 2623 2624 2625
	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);
2626
	atomic_set(&fs_info->nr_async_submits, 0);
2627
	atomic_set(&fs_info->async_delalloc_pages, 0);
2628
	atomic_set(&fs_info->async_submit_draining, 0);
2629
	atomic_set(&fs_info->nr_async_bios, 0);
C
Chris Mason 已提交
2630
	atomic_set(&fs_info->defrag_running, 0);
J
Josef Bacik 已提交
2631
	atomic_set(&fs_info->qgroup_op_seq, 0);
Z
Zhao Lei 已提交
2632
	atomic_set(&fs_info->reada_works_cnt, 0);
2633
	atomic64_set(&fs_info->tree_mod_seq, 0);
C
Chris Mason 已提交
2634
	fs_info->sb = sb;
2635
	fs_info->max_inline = BTRFS_DEFAULT_MAX_INLINE;
J
Josef Bacik 已提交
2636
	fs_info->metadata_ratio = 0;
C
Chris Mason 已提交
2637
	fs_info->defrag_inodes = RB_ROOT;
2638
	fs_info->free_chunk_space = 0;
J
Jan Schmidt 已提交
2639
	fs_info->tree_mod_log = RB_ROOT;
2640
	fs_info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
2641
	fs_info->avg_delayed_ref_runtime = NSEC_PER_SEC >> 6; /* div by 64 */
2642
	/* readahead state */
2643
	INIT_RADIX_TREE(&fs_info->reada_tree, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
2644
	spin_lock_init(&fs_info->reada_lock);
C
Chris Mason 已提交
2645

2646 2647
	fs_info->thread_pool_size = min_t(unsigned long,
					  num_online_cpus() + 2, 8);
2648

2649 2650
	INIT_LIST_HEAD(&fs_info->ordered_roots);
	spin_lock_init(&fs_info->ordered_root_lock);
2651
	fs_info->delayed_root = kmalloc(sizeof(struct btrfs_delayed_root),
2652
					GFP_KERNEL);
2653 2654 2655 2656 2657
	if (!fs_info->delayed_root) {
		err = -ENOMEM;
		goto fail_iput;
	}
	btrfs_init_delayed_root(fs_info->delayed_root);
2658

2659
	btrfs_init_scrub(fs_info);
2660 2661 2662
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	fs_info->check_integrity_print_mask = 0;
#endif
2663
	btrfs_init_balance(fs_info);
2664
	btrfs_init_async_reclaim_work(&fs_info->async_reclaim_work);
A
Arne Jansen 已提交
2665

2666 2667
	sb->s_blocksize = 4096;
	sb->s_blocksize_bits = blksize_bits(4096);
J
Jens Axboe 已提交
2668
	sb->s_bdi = &fs_info->bdi;
2669

2670
	btrfs_init_btree_inode(fs_info, tree_root);
2671

J
Josef Bacik 已提交
2672
	spin_lock_init(&fs_info->block_group_cache_lock);
2673
	fs_info->block_group_cache_tree = RB_ROOT;
2674
	fs_info->first_logical_byte = (u64)-1;
J
Josef Bacik 已提交
2675

2676
	extent_io_tree_init(&fs_info->freed_extents[0],
2677
			     fs_info->btree_inode->i_mapping);
2678
	extent_io_tree_init(&fs_info->freed_extents[1],
2679
			     fs_info->btree_inode->i_mapping);
2680
	fs_info->pinned_extents = &fs_info->freed_extents[0];
2681
	fs_info->do_barriers = 1;
2682

C
Chris Mason 已提交
2683

2684
	mutex_init(&fs_info->ordered_operations_mutex);
2685
	mutex_init(&fs_info->tree_log_mutex);
2686
	mutex_init(&fs_info->chunk_mutex);
2687 2688
	mutex_init(&fs_info->transaction_kthread_mutex);
	mutex_init(&fs_info->cleaner_mutex);
2689
	mutex_init(&fs_info->volume_mutex);
2690
	mutex_init(&fs_info->ro_block_group_mutex);
2691
	init_rwsem(&fs_info->commit_root_sem);
2692
	init_rwsem(&fs_info->cleanup_work_sem);
2693
	init_rwsem(&fs_info->subvol_sem);
S
Stefan Behrens 已提交
2694
	sema_init(&fs_info->uuid_tree_rescan_sem, 1);
2695

2696
	btrfs_init_dev_replace_locks(fs_info);
2697
	btrfs_init_qgroup(fs_info);
2698

2699 2700 2701
	btrfs_init_free_cluster(&fs_info->meta_alloc_cluster);
	btrfs_init_free_cluster(&fs_info->data_alloc_cluster);

2702
	init_waitqueue_head(&fs_info->transaction_throttle);
2703
	init_waitqueue_head(&fs_info->transaction_wait);
S
Sage Weil 已提交
2704
	init_waitqueue_head(&fs_info->transaction_blocked_wait);
2705
	init_waitqueue_head(&fs_info->async_submit_wait);
2706

2707 2708
	INIT_LIST_HEAD(&fs_info->pinned_chunks);

D
David Woodhouse 已提交
2709 2710
	ret = btrfs_alloc_stripe_hash_table(fs_info);
	if (ret) {
2711
		err = ret;
D
David Woodhouse 已提交
2712 2713 2714
		goto fail_alloc;
	}

2715
	__setup_root(4096, 4096, 4096, tree_root,
C
Chris Mason 已提交
2716
		     fs_info, BTRFS_ROOT_TREE_OBJECTID);
2717

2718
	invalidate_bdev(fs_devices->latest_bdev);
D
David Sterba 已提交
2719 2720 2721 2722

	/*
	 * Read super block and check the signature bytes only
	 */
Y
Yan Zheng 已提交
2723
	bh = btrfs_read_dev_super(fs_devices->latest_bdev);
2724 2725
	if (IS_ERR(bh)) {
		err = PTR_ERR(bh);
2726
		goto fail_alloc;
2727
	}
C
Chris Mason 已提交
2728

D
David Sterba 已提交
2729 2730 2731 2732 2733
	/*
	 * We want to check superblock checksum, the type is stored inside.
	 * Pass the whole disk block of size BTRFS_SUPER_INFO_SIZE (4k).
	 */
	if (btrfs_check_super_csum(bh->b_data)) {
2734
		btrfs_err(fs_info, "superblock checksum mismatch");
D
David Sterba 已提交
2735
		err = -EINVAL;
2736
		brelse(bh);
D
David Sterba 已提交
2737 2738 2739 2740 2741 2742 2743 2744
		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
	 */
2745 2746 2747
	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));
2748
	brelse(bh);
2749

2750
	memcpy(fs_info->fsid, fs_info->super_copy->fsid, BTRFS_FSID_SIZE);
2751

D
David Sterba 已提交
2752 2753
	ret = btrfs_check_super_valid(fs_info, sb->s_flags & MS_RDONLY);
	if (ret) {
2754
		btrfs_err(fs_info, "superblock contains fatal errors");
D
David Sterba 已提交
2755 2756 2757 2758
		err = -EINVAL;
		goto fail_alloc;
	}

2759
	disk_super = fs_info->super_copy;
2760
	if (!btrfs_super_root(disk_super))
2761
		goto fail_alloc;
2762

L
liubo 已提交
2763
	/* check FS state, whether FS is broken. */
2764 2765
	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
2766

C
Chris Mason 已提交
2767 2768 2769 2770 2771 2772 2773
	/*
	 * 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);

2774 2775 2776 2777 2778 2779
	/*
	 * 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;

2780
	ret = btrfs_parse_options(tree_root, options, sb->s_flags);
Y
Yan Zheng 已提交
2781 2782
	if (ret) {
		err = ret;
2783
		goto fail_alloc;
Y
Yan Zheng 已提交
2784
	}
2785

2786 2787 2788
	features = btrfs_super_incompat_flags(disk_super) &
		~BTRFS_FEATURE_INCOMPAT_SUPP;
	if (features) {
2789 2790 2791
		btrfs_err(fs_info,
		    "cannot mount because of unsupported optional features (%llx)",
		    features);
2792
		err = -EINVAL;
2793
		goto fail_alloc;
2794 2795
	}

2796
	features = btrfs_super_incompat_flags(disk_super);
L
Li Zefan 已提交
2797
	features |= BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF;
L
Li Zefan 已提交
2798
	if (tree_root->fs_info->compress_type == BTRFS_COMPRESS_LZO)
L
Li Zefan 已提交
2799
		features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
2800

2801
	if (features & BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA)
2802
		btrfs_info(fs_info, "has skinny extents");
2803

2804 2805 2806 2807
	/*
	 * flag our filesystem as having big metadata blocks if
	 * they are bigger than the page size
	 */
2808
	if (btrfs_super_nodesize(disk_super) > PAGE_SIZE) {
2809
		if (!(features & BTRFS_FEATURE_INCOMPAT_BIG_METADATA))
2810 2811
			btrfs_info(fs_info,
				"flagging fs with big metadata feature");
2812 2813 2814
		features |= BTRFS_FEATURE_INCOMPAT_BIG_METADATA;
	}

2815 2816
	nodesize = btrfs_super_nodesize(disk_super);
	sectorsize = btrfs_super_sectorsize(disk_super);
2817
	stripesize = sectorsize;
2818
	fs_info->dirty_metadata_batch = nodesize * (1 + ilog2(nr_cpu_ids));
2819
	fs_info->delalloc_batch = sectorsize * 512 * (1 + ilog2(nr_cpu_ids));
2820 2821 2822 2823 2824 2825

	/*
	 * 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) &&
2826
	    (sectorsize != nodesize)) {
2827 2828 2829
		btrfs_err(fs_info,
"unequal nodesize/sectorsize (%u != %u) are not allowed for mixed block groups",
			nodesize, sectorsize);
2830 2831 2832
		goto fail_alloc;
	}

2833 2834 2835 2836
	/*
	 * Needn't use the lock because there is no other task which will
	 * update the flag.
	 */
L
Li Zefan 已提交
2837
	btrfs_set_super_incompat_flags(disk_super, features);
2838

2839 2840 2841
	features = btrfs_super_compat_ro_flags(disk_super) &
		~BTRFS_FEATURE_COMPAT_RO_SUPP;
	if (!(sb->s_flags & MS_RDONLY) && features) {
2842 2843
		btrfs_err(fs_info,
	"cannot mount read-write because of unsupported optional features (%llx)",
2844
		       features);
2845
		err = -EINVAL;
2846
		goto fail_alloc;
2847
	}
2848

2849
	max_active = fs_info->thread_pool_size;
2850

2851 2852 2853
	ret = btrfs_init_workqueues(fs_info, fs_devices);
	if (ret) {
		err = ret;
2854 2855
		goto fail_sb_buffer;
	}
2856

2857
	fs_info->bdi.ra_pages *= btrfs_super_num_devices(disk_super);
C
Chris Mason 已提交
2858
	fs_info->bdi.ra_pages = max(fs_info->bdi.ra_pages,
2859
				    SZ_4M / PAGE_SIZE);
2860

2861 2862
	tree_root->nodesize = nodesize;
	tree_root->sectorsize = sectorsize;
2863
	tree_root->stripesize = stripesize;
2864 2865 2866

	sb->s_blocksize = sectorsize;
	sb->s_blocksize_bits = blksize_bits(sectorsize);
2867

2868
	mutex_lock(&fs_info->chunk_mutex);
Y
Yan Zheng 已提交
2869
	ret = btrfs_read_sys_array(tree_root);
2870
	mutex_unlock(&fs_info->chunk_mutex);
2871
	if (ret) {
2872
		btrfs_err(fs_info, "failed to read the system array: %d", ret);
2873
		goto fail_sb_buffer;
2874
	}
2875

2876
	generation = btrfs_super_chunk_root_generation(disk_super);
2877

2878 2879
	__setup_root(nodesize, sectorsize, stripesize, chunk_root,
		     fs_info, BTRFS_CHUNK_TREE_OBJECTID);
2880 2881 2882

	chunk_root->node = read_tree_block(chunk_root,
					   btrfs_super_chunk_root(disk_super),
2883
					   generation);
2884 2885
	if (IS_ERR(chunk_root->node) ||
	    !extent_buffer_uptodate(chunk_root->node)) {
2886
		btrfs_err(fs_info, "failed to read chunk root");
2887 2888
		if (!IS_ERR(chunk_root->node))
			free_extent_buffer(chunk_root->node);
2889
		chunk_root->node = NULL;
C
Chris Mason 已提交
2890
		goto fail_tree_roots;
2891
	}
2892 2893
	btrfs_set_root_node(&chunk_root->root_item, chunk_root->node);
	chunk_root->commit_root = btrfs_root_node(chunk_root);
2894

2895
	read_extent_buffer(chunk_root->node, fs_info->chunk_tree_uuid,
2896
	   btrfs_header_chunk_tree_uuid(chunk_root->node), BTRFS_UUID_SIZE);
2897

2898
	ret = btrfs_read_chunk_tree(chunk_root);
Y
Yan Zheng 已提交
2899
	if (ret) {
2900
		btrfs_err(fs_info, "failed to read chunk tree: %d", ret);
C
Chris Mason 已提交
2901
		goto fail_tree_roots;
Y
Yan Zheng 已提交
2902
	}
2903

2904 2905 2906 2907
	/*
	 * keep the device that is marked to be the target device for the
	 * dev_replace procedure
	 */
2908
	btrfs_close_extra_devices(fs_devices, 0);
2909

2910
	if (!fs_devices->latest_bdev) {
2911
		btrfs_err(fs_info, "failed to read devices");
2912 2913 2914
		goto fail_tree_roots;
	}

C
Chris Mason 已提交
2915
retry_root_backup:
2916
	generation = btrfs_super_generation(disk_super);
2917

C
Chris Mason 已提交
2918
	tree_root->node = read_tree_block(tree_root,
2919
					  btrfs_super_root(disk_super),
2920
					  generation);
2921 2922
	if (IS_ERR(tree_root->node) ||
	    !extent_buffer_uptodate(tree_root->node)) {
2923
		btrfs_warn(fs_info, "failed to read tree root");
2924 2925
		if (!IS_ERR(tree_root->node))
			free_extent_buffer(tree_root->node);
2926
		tree_root->node = NULL;
C
Chris Mason 已提交
2927
		goto recovery_tree_root;
2928
	}
C
Chris Mason 已提交
2929

2930 2931
	btrfs_set_root_node(&tree_root->root_item, tree_root->node);
	tree_root->commit_root = btrfs_root_node(tree_root);
2932
	btrfs_set_root_refs(&tree_root->root_item, 1);
2933

2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945
	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);

2946 2947
	ret = btrfs_read_roots(fs_info, tree_root);
	if (ret)
C
Chris Mason 已提交
2948
		goto recovery_tree_root;
2949

2950 2951 2952
	fs_info->generation = generation;
	fs_info->last_trans_committed = generation;

2953 2954
	ret = btrfs_recover_balance(fs_info);
	if (ret) {
2955
		btrfs_err(fs_info, "failed to recover balance: %d", ret);
2956 2957 2958
		goto fail_block_groups;
	}

2959 2960
	ret = btrfs_init_dev_stats(fs_info);
	if (ret) {
2961
		btrfs_err(fs_info, "failed to init dev_stats: %d", ret);
2962 2963 2964
		goto fail_block_groups;
	}

2965 2966
	ret = btrfs_init_dev_replace(fs_info);
	if (ret) {
2967
		btrfs_err(fs_info, "failed to init dev_replace: %d", ret);
2968 2969 2970
		goto fail_block_groups;
	}

2971
	btrfs_close_extra_devices(fs_devices, 1);
2972

2973 2974
	ret = btrfs_sysfs_add_fsid(fs_devices, NULL);
	if (ret) {
2975 2976
		btrfs_err(fs_info, "failed to init sysfs fsid interface: %d",
				ret);
2977 2978 2979 2980 2981
		goto fail_block_groups;
	}

	ret = btrfs_sysfs_add_device(fs_devices);
	if (ret) {
2982 2983
		btrfs_err(fs_info, "failed to init sysfs device interface: %d",
				ret);
2984 2985 2986
		goto fail_fsdev_sysfs;
	}

2987
	ret = btrfs_sysfs_add_mounted(fs_info);
2988
	if (ret) {
2989
		btrfs_err(fs_info, "failed to init sysfs interface: %d", ret);
2990
		goto fail_fsdev_sysfs;
2991 2992 2993 2994
	}

	ret = btrfs_init_space_info(fs_info);
	if (ret) {
2995
		btrfs_err(fs_info, "failed to initialize space info: %d", ret);
2996
		goto fail_sysfs;
2997 2998
	}

2999
	ret = btrfs_read_block_groups(fs_info->extent_root);
3000
	if (ret) {
3001
		btrfs_err(fs_info, "failed to read block groups: %d", ret);
3002
		goto fail_sysfs;
3003
	}
3004 3005
	fs_info->num_tolerated_disk_barrier_failures =
		btrfs_calc_num_tolerated_disk_barrier_failures(fs_info);
3006 3007 3008
	if (fs_info->fs_devices->missing_devices >
	     fs_info->num_tolerated_disk_barrier_failures &&
	    !(sb->s_flags & MS_RDONLY)) {
3009 3010
		btrfs_warn(fs_info,
"missing devices (%llu) exceeds the limit (%d), writeable mount is not allowed",
3011 3012
			fs_info->fs_devices->missing_devices,
			fs_info->num_tolerated_disk_barrier_failures);
3013
		goto fail_sysfs;
3014
	}
C
Chris Mason 已提交
3015

3016 3017
	fs_info->cleaner_kthread = kthread_run(cleaner_kthread, tree_root,
					       "btrfs-cleaner");
3018
	if (IS_ERR(fs_info->cleaner_kthread))
3019
		goto fail_sysfs;
3020 3021 3022 3023

	fs_info->transaction_kthread = kthread_run(transaction_kthread,
						   tree_root,
						   "btrfs-transaction");
3024
	if (IS_ERR(fs_info->transaction_kthread))
3025
		goto fail_cleaner;
3026

3027 3028
	if (!btrfs_test_opt(tree_root->fs_info, SSD) &&
	    !btrfs_test_opt(tree_root->fs_info, NOSSD) &&
C
Chris Mason 已提交
3029
	    !fs_info->fs_devices->rotating) {
3030
		btrfs_info(fs_info, "detected SSD devices, enabling SSD mode");
C
Chris Mason 已提交
3031 3032 3033
		btrfs_set_opt(fs_info->mount_opt, SSD);
	}

3034
	/*
3035
	 * Mount does not set all options immediately, we can do it now and do
3036 3037 3038
	 * not have to wait for transaction commit
	 */
	btrfs_apply_pending_changes(fs_info);
3039

3040
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
3041
	if (btrfs_test_opt(tree_root->fs_info, CHECK_INTEGRITY)) {
3042
		ret = btrfsic_mount(tree_root, fs_devices,
3043
				    btrfs_test_opt(tree_root->fs_info,
3044 3045 3046 3047
					CHECK_INTEGRITY_INCLUDING_EXTENT_DATA) ?
				    1 : 0,
				    fs_info->check_integrity_print_mask);
		if (ret)
3048 3049 3050
			btrfs_warn(fs_info,
				"failed to initialize integrity check module: %d",
				ret);
3051 3052
	}
#endif
3053 3054 3055
	ret = btrfs_read_qgroup_config(fs_info);
	if (ret)
		goto fail_trans_kthread;
3056

3057 3058
	/* do not make disk changes in broken FS or nologreplay is given */
	if (btrfs_super_log_root(disk_super) != 0 &&
3059
	    !btrfs_test_opt(tree_root->fs_info, NOLOGREPLAY)) {
3060
		ret = btrfs_replay_log(fs_info, fs_devices);
3061
		if (ret) {
3062
			err = ret;
3063
			goto fail_qgroup;
3064
		}
3065
	}
Z
Zheng Yan 已提交
3066

3067
	ret = btrfs_find_orphan_roots(tree_root);
3068
	if (ret)
3069
		goto fail_qgroup;
3070

3071
	if (!(sb->s_flags & MS_RDONLY)) {
3072
		ret = btrfs_cleanup_fs_roots(fs_info);
3073
		if (ret)
3074
			goto fail_qgroup;
3075 3076

		mutex_lock(&fs_info->cleaner_mutex);
3077
		ret = btrfs_recover_relocation(tree_root);
3078
		mutex_unlock(&fs_info->cleaner_mutex);
3079
		if (ret < 0) {
3080 3081
			btrfs_warn(fs_info, "failed to recover relocation: %d",
					ret);
3082
			err = -EINVAL;
3083
			goto fail_qgroup;
3084
		}
3085
	}
Z
Zheng Yan 已提交
3086

3087 3088
	location.objectid = BTRFS_FS_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
3089
	location.offset = 0;
3090 3091

	fs_info->fs_root = btrfs_read_fs_root_no_name(fs_info, &location);
3092 3093
	if (IS_ERR(fs_info->fs_root)) {
		err = PTR_ERR(fs_info->fs_root);
3094
		goto fail_qgroup;
3095
	}
C
Chris Mason 已提交
3096

3097 3098
	if (sb->s_flags & MS_RDONLY)
		return 0;
I
Ilya Dryomov 已提交
3099

3100
	if (btrfs_test_opt(tree_root->fs_info, FREE_SPACE_TREE) &&
3101
	    !btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
3102
		btrfs_info(fs_info, "creating free space tree");
3103 3104
		ret = btrfs_create_free_space_tree(fs_info);
		if (ret) {
3105 3106
			btrfs_warn(fs_info,
				"failed to create free space tree: %d", ret);
3107 3108 3109 3110 3111
			close_ctree(tree_root);
			return ret;
		}
	}

3112 3113 3114
	down_read(&fs_info->cleanup_work_sem);
	if ((ret = btrfs_orphan_cleanup(fs_info->fs_root)) ||
	    (ret = btrfs_orphan_cleanup(fs_info->tree_root))) {
3115
		up_read(&fs_info->cleanup_work_sem);
3116 3117 3118 3119
		close_ctree(tree_root);
		return ret;
	}
	up_read(&fs_info->cleanup_work_sem);
I
Ilya Dryomov 已提交
3120

3121 3122
	ret = btrfs_resume_balance_async(fs_info);
	if (ret) {
3123
		btrfs_warn(fs_info, "failed to resume balance: %d", ret);
3124 3125
		close_ctree(tree_root);
		return ret;
3126 3127
	}

3128 3129
	ret = btrfs_resume_dev_replace_async(fs_info);
	if (ret) {
3130
		btrfs_warn(fs_info, "failed to resume device replace: %d", ret);
3131 3132 3133 3134
		close_ctree(tree_root);
		return ret;
	}

3135 3136
	btrfs_qgroup_rescan_resume(fs_info);

3137
	if (btrfs_test_opt(tree_root->fs_info, CLEAR_CACHE) &&
3138
	    btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
3139
		btrfs_info(fs_info, "clearing free space tree");
3140 3141
		ret = btrfs_clear_free_space_tree(fs_info);
		if (ret) {
3142 3143
			btrfs_warn(fs_info,
				"failed to clear free space tree: %d", ret);
3144 3145 3146 3147 3148
			close_ctree(tree_root);
			return ret;
		}
	}

3149
	if (!fs_info->uuid_root) {
3150
		btrfs_info(fs_info, "creating UUID tree");
3151 3152
		ret = btrfs_create_uuid_tree(fs_info);
		if (ret) {
3153 3154
			btrfs_warn(fs_info,
				"failed to create the UUID tree: %d", ret);
3155 3156 3157
			close_ctree(tree_root);
			return ret;
		}
3158
	} else if (btrfs_test_opt(tree_root->fs_info, RESCAN_UUID_TREE) ||
3159 3160
		   fs_info->generation !=
				btrfs_super_uuid_tree_generation(disk_super)) {
3161
		btrfs_info(fs_info, "checking UUID tree");
3162 3163
		ret = btrfs_check_uuid_tree(fs_info);
		if (ret) {
3164 3165
			btrfs_warn(fs_info,
				"failed to check the UUID tree: %d", ret);
3166 3167 3168 3169 3170
			close_ctree(tree_root);
			return ret;
		}
	} else {
		fs_info->update_uuid_tree_gen = 1;
3171 3172
	}

3173 3174
	fs_info->open = 1;

3175 3176 3177 3178 3179 3180
	/*
	 * 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
3181
	return 0;
C
Chris Mason 已提交
3182

3183 3184
fail_qgroup:
	btrfs_free_qgroup_config(fs_info);
3185 3186
fail_trans_kthread:
	kthread_stop(fs_info->transaction_kthread);
J
Josef Bacik 已提交
3187
	btrfs_cleanup_transaction(fs_info->tree_root);
3188
	btrfs_free_fs_roots(fs_info);
3189
fail_cleaner:
3190
	kthread_stop(fs_info->cleaner_kthread);
3191 3192 3193 3194 3195 3196 3197

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

3198
fail_sysfs:
3199
	btrfs_sysfs_remove_mounted(fs_info);
3200

3201 3202 3203
fail_fsdev_sysfs:
	btrfs_sysfs_remove_fsid(fs_info->fs_devices);

3204
fail_block_groups:
J
Josef Bacik 已提交
3205
	btrfs_put_block_group_cache(fs_info);
3206
	btrfs_free_block_groups(fs_info);
C
Chris Mason 已提交
3207 3208 3209

fail_tree_roots:
	free_root_pointers(fs_info, 1);
3210
	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
3211

C
Chris Mason 已提交
3212
fail_sb_buffer:
L
Liu Bo 已提交
3213
	btrfs_stop_all_workers(fs_info);
3214
fail_alloc:
3215
fail_iput:
3216 3217
	btrfs_mapping_tree_free(&fs_info->mapping_tree);

3218
	iput(fs_info->btree_inode);
3219 3220
fail_bio_counter:
	percpu_counter_destroy(&fs_info->bio_counter);
3221 3222
fail_delalloc_bytes:
	percpu_counter_destroy(&fs_info->delalloc_bytes);
3223 3224
fail_dirty_metadata_bytes:
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
3225
fail_bdi:
3226
	bdi_destroy(&fs_info->bdi);
3227 3228
fail_srcu:
	cleanup_srcu_struct(&fs_info->subvol_srcu);
3229
fail:
D
David Woodhouse 已提交
3230
	btrfs_free_stripe_hash_table(fs_info);
3231
	btrfs_close_devices(fs_info->fs_devices);
A
Al Viro 已提交
3232
	return err;
C
Chris Mason 已提交
3233 3234

recovery_tree_root:
3235
	if (!btrfs_test_opt(tree_root->fs_info, USEBACKUPROOT))
C
Chris Mason 已提交
3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250
		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;
3251 3252
}

3253 3254 3255 3256 3257
static void btrfs_end_buffer_write_sync(struct buffer_head *bh, int uptodate)
{
	if (uptodate) {
		set_buffer_uptodate(bh);
	} else {
3258 3259 3260
		struct btrfs_device *device = (struct btrfs_device *)
			bh->b_private;

3261 3262
		btrfs_warn_rl_in_rcu(device->dev_root->fs_info,
				"lost page write due to IO error on %s",
3263
					  rcu_str_deref(device->name));
3264
		/* note, we don't set_buffer_write_io_error because we have
3265 3266
		 * our own ways of dealing with the IO errors
		 */
3267
		clear_buffer_uptodate(bh);
3268
		btrfs_dev_stat_inc_and_print(device, BTRFS_DEV_STAT_WRITE_ERRS);
3269 3270 3271 3272 3273
	}
	unlock_buffer(bh);
	put_bh(bh);
}

3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304
int btrfs_read_dev_one_super(struct block_device *bdev, int copy_num,
			struct buffer_head **bh_ret)
{
	struct buffer_head *bh;
	struct btrfs_super_block *super;
	u64 bytenr;

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

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

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

	*bh_ret = bh;
	return 0;
}


Y
Yan Zheng 已提交
3305 3306 3307 3308 3309 3310 3311
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;
3312
	int ret = -EINVAL;
Y
Yan Zheng 已提交
3313 3314 3315 3316 3317 3318 3319

	/* 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++) {
3320 3321
		ret = btrfs_read_dev_one_super(bdev, i, &bh);
		if (ret)
Y
Yan Zheng 已提交
3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333
			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);
		}
	}
3334 3335 3336 3337

	if (!latest)
		return ERR_PTR(ret);

Y
Yan Zheng 已提交
3338 3339 3340
	return latest;
}

3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351
/*
 * this should be called twice, once with wait == 0 and
 * once with wait == 1.  When wait == 0 is done, all the buffer heads
 * we write are pinned.
 *
 * They are released when wait == 1 is done.
 * max_mirrors must be the same for both runs, and it indicates how
 * many supers on this one device should be written.
 *
 * max_mirrors == 0 means to write them all.
 */
Y
Yan Zheng 已提交
3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367
static int write_dev_supers(struct btrfs_device *device,
			    struct btrfs_super_block *sb,
			    int do_barriers, int wait, int max_mirrors)
{
	struct buffer_head *bh;
	int i;
	int ret;
	int errors = 0;
	u32 crc;
	u64 bytenr;

	if (max_mirrors == 0)
		max_mirrors = BTRFS_SUPER_MIRROR_MAX;

	for (i = 0; i < max_mirrors; i++) {
		bytenr = btrfs_sb_offset(i);
3368 3369
		if (bytenr + BTRFS_SUPER_INFO_SIZE >=
		    device->commit_total_bytes)
Y
Yan Zheng 已提交
3370 3371 3372 3373 3374
			break;

		if (wait) {
			bh = __find_get_block(device->bdev, bytenr / 4096,
					      BTRFS_SUPER_INFO_SIZE);
3375 3376 3377 3378
			if (!bh) {
				errors++;
				continue;
			}
Y
Yan Zheng 已提交
3379
			wait_on_buffer(bh);
3380 3381 3382 3383 3384 3385 3386 3387 3388
			if (!buffer_uptodate(bh))
				errors++;

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

			/* drop the reference from the wait == 0 run */
			brelse(bh);
			continue;
Y
Yan Zheng 已提交
3389 3390 3391 3392
		} else {
			btrfs_set_super_bytenr(sb, bytenr);

			crc = ~(u32)0;
3393
			crc = btrfs_csum_data((char *)sb +
Y
Yan Zheng 已提交
3394 3395 3396 3397 3398
					      BTRFS_CSUM_SIZE, crc,
					      BTRFS_SUPER_INFO_SIZE -
					      BTRFS_CSUM_SIZE);
			btrfs_csum_final(crc, sb->csum);

3399 3400 3401 3402
			/*
			 * one reference for us, and we leave it for the
			 * caller
			 */
Y
Yan Zheng 已提交
3403 3404
			bh = __getblk(device->bdev, bytenr / 4096,
				      BTRFS_SUPER_INFO_SIZE);
3405
			if (!bh) {
3406 3407 3408
				btrfs_err(device->dev_root->fs_info,
				    "couldn't get super buffer head for bytenr %llu",
				    bytenr);
3409 3410 3411 3412
				errors++;
				continue;
			}

Y
Yan Zheng 已提交
3413 3414
			memcpy(bh->b_data, sb, BTRFS_SUPER_INFO_SIZE);

3415
			/* one reference for submit_bh */
Y
Yan Zheng 已提交
3416
			get_bh(bh);
3417 3418

			set_buffer_uptodate(bh);
Y
Yan Zheng 已提交
3419 3420
			lock_buffer(bh);
			bh->b_end_io = btrfs_end_buffer_write_sync;
3421
			bh->b_private = device;
Y
Yan Zheng 已提交
3422 3423
		}

C
Chris Mason 已提交
3424 3425 3426 3427
		/*
		 * we fua the first super.  The others we allow
		 * to go down lazy.
		 */
3428 3429 3430 3431
		if (i == 0)
			ret = btrfsic_submit_bh(WRITE_FUA, bh);
		else
			ret = btrfsic_submit_bh(WRITE_SYNC, bh);
3432
		if (ret)
Y
Yan Zheng 已提交
3433 3434 3435 3436 3437
			errors++;
	}
	return errors < i ? 0 : -1;
}

C
Chris Mason 已提交
3438 3439 3440 3441
/*
 * endio for the write_dev_flush, this will wake anyone waiting
 * for the barrier when it is done
 */
3442
static void btrfs_end_empty_barrier(struct bio *bio)
C
Chris Mason 已提交
3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470
{
	if (bio->bi_private)
		complete(bio->bi_private);
	bio_put(bio);
}

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

	if (device->nobarriers)
		return 0;

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

		wait_for_completion(&device->flush_wait);

3471 3472
		if (bio->bi_error) {
			ret = bio->bi_error;
3473 3474
			btrfs_dev_stat_inc_and_print(device,
				BTRFS_DEV_STAT_FLUSH_ERRS);
C
Chris Mason 已提交
3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487
		}

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

		return ret;
	}

	/*
	 * one reference for us, and we leave it for the
	 * caller
	 */
3488
	device->flush_bio = NULL;
3489
	bio = btrfs_io_bio_alloc(GFP_NOFS, 0);
C
Chris Mason 已提交
3490 3491 3492 3493 3494 3495 3496 3497 3498 3499
	if (!bio)
		return -ENOMEM;

	bio->bi_end_io = btrfs_end_empty_barrier;
	bio->bi_bdev = device->bdev;
	init_completion(&device->flush_wait);
	bio->bi_private = &device->flush_wait;
	device->flush_bio = bio;

	bio_get(bio);
3500
	btrfsic_submit_bio(WRITE_FLUSH, bio);
C
Chris Mason 已提交
3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512

	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;
3513 3514
	int errors_send = 0;
	int errors_wait = 0;
C
Chris Mason 已提交
3515 3516 3517 3518 3519
	int ret;

	/* send down all the barriers */
	head = &info->fs_devices->devices;
	list_for_each_entry_rcu(dev, head, dev_list) {
3520 3521
		if (dev->missing)
			continue;
C
Chris Mason 已提交
3522
		if (!dev->bdev) {
3523
			errors_send++;
C
Chris Mason 已提交
3524 3525 3526 3527 3528 3529 3530
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

		ret = write_dev_flush(dev, 0);
		if (ret)
3531
			errors_send++;
C
Chris Mason 已提交
3532 3533 3534 3535
	}

	/* wait for all the barriers */
	list_for_each_entry_rcu(dev, head, dev_list) {
3536 3537
		if (dev->missing)
			continue;
C
Chris Mason 已提交
3538
		if (!dev->bdev) {
3539
			errors_wait++;
C
Chris Mason 已提交
3540 3541 3542 3543 3544 3545 3546
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

		ret = write_dev_flush(dev, 1);
		if (ret)
3547
			errors_wait++;
C
Chris Mason 已提交
3548
	}
3549 3550
	if (errors_send > info->num_tolerated_disk_barrier_failures ||
	    errors_wait > info->num_tolerated_disk_barrier_failures)
C
Chris Mason 已提交
3551 3552 3553 3554
		return -EIO;
	return 0;
}

3555 3556
int btrfs_get_num_tolerated_disk_barrier_failures(u64 flags)
{
3557 3558
	int raid_type;
	int min_tolerated = INT_MAX;
3559

3560 3561 3562 3563 3564
	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);
3565

3566 3567 3568 3569 3570 3571 3572 3573 3574
	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);
	}
3575

3576 3577 3578 3579 3580 3581
	if (min_tolerated == INT_MAX) {
		pr_warn("BTRFS: unknown raid flag: %llu\n", flags);
		min_tolerated = 0;
	}

	return min_tolerated;
3582 3583
}

3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597
int btrfs_calc_num_tolerated_disk_barrier_failures(
	struct btrfs_fs_info *fs_info)
{
	struct btrfs_ioctl_space_info space;
	struct btrfs_space_info *sinfo;
	u64 types[] = {BTRFS_BLOCK_GROUP_DATA,
		       BTRFS_BLOCK_GROUP_SYSTEM,
		       BTRFS_BLOCK_GROUP_METADATA,
		       BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA};
	int i;
	int c;
	int num_tolerated_disk_barrier_failures =
		(int)fs_info->fs_devices->num_devices;

3598
	for (i = 0; i < ARRAY_SIZE(types); i++) {
3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615
		struct btrfs_space_info *tmp;

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

		if (!sinfo)
			continue;

		down_read(&sinfo->groups_sem);
		for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
3616 3617 3618 3619 3620 3621 3622 3623 3624 3625
			u64 flags;

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

			btrfs_get_block_group_info(&sinfo->block_groups[c],
						   &space);
			if (space.total_bytes == 0 || space.used_bytes == 0)
				continue;
			flags = space.flags;
3626 3627 3628 3629 3630

			num_tolerated_disk_barrier_failures = min(
				num_tolerated_disk_barrier_failures,
				btrfs_get_num_tolerated_disk_barrier_failures(
					flags));
3631 3632 3633 3634 3635 3636 3637
		}
		up_read(&sinfo->groups_sem);
	}

	return num_tolerated_disk_barrier_failures;
}

3638
static int write_all_supers(struct btrfs_root *root, int max_mirrors)
3639
{
3640
	struct list_head *head;
3641
	struct btrfs_device *dev;
3642
	struct btrfs_super_block *sb;
3643 3644 3645
	struct btrfs_dev_item *dev_item;
	int ret;
	int do_barriers;
3646 3647
	int max_errors;
	int total_errors = 0;
3648
	u64 flags;
3649

3650
	do_barriers = !btrfs_test_opt(root->fs_info, NOBARRIER);
C
Chris Mason 已提交
3651
	backup_super_roots(root->fs_info);
3652

3653
	sb = root->fs_info->super_for_commit;
3654
	dev_item = &sb->dev_item;
3655

3656
	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
3657
	head = &root->fs_info->fs_devices->devices;
3658
	max_errors = btrfs_super_num_devices(root->fs_info->super_copy) - 1;
C
Chris Mason 已提交
3659

3660 3661 3662 3663 3664
	if (do_barriers) {
		ret = barrier_all_devices(root->fs_info);
		if (ret) {
			mutex_unlock(
				&root->fs_info->fs_devices->device_list_mutex);
3665
			btrfs_handle_fs_error(root->fs_info, ret,
3666 3667 3668 3669
				    "errors while submitting device barriers.");
			return ret;
		}
	}
C
Chris Mason 已提交
3670

3671
	list_for_each_entry_rcu(dev, head, dev_list) {
3672 3673 3674 3675
		if (!dev->bdev) {
			total_errors++;
			continue;
		}
Y
Yan Zheng 已提交
3676
		if (!dev->in_fs_metadata || !dev->writeable)
3677 3678
			continue;

Y
Yan Zheng 已提交
3679
		btrfs_set_stack_device_generation(dev_item, 0);
3680 3681
		btrfs_set_stack_device_type(dev_item, dev->type);
		btrfs_set_stack_device_id(dev_item, dev->devid);
3682
		btrfs_set_stack_device_total_bytes(dev_item,
3683
						   dev->commit_total_bytes);
3684 3685
		btrfs_set_stack_device_bytes_used(dev_item,
						  dev->commit_bytes_used);
3686 3687 3688 3689
		btrfs_set_stack_device_io_align(dev_item, dev->io_align);
		btrfs_set_stack_device_io_width(dev_item, dev->io_width);
		btrfs_set_stack_device_sector_size(dev_item, dev->sector_size);
		memcpy(dev_item->uuid, dev->uuid, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3690
		memcpy(dev_item->fsid, dev->fs_devices->fsid, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3691

3692 3693 3694
		flags = btrfs_super_flags(sb);
		btrfs_set_super_flags(sb, flags | BTRFS_HEADER_FLAG_WRITTEN);

Y
Yan Zheng 已提交
3695
		ret = write_dev_supers(dev, sb, do_barriers, 0, max_mirrors);
3696 3697
		if (ret)
			total_errors++;
3698
	}
3699
	if (total_errors > max_errors) {
3700
		btrfs_err(root->fs_info, "%d errors while writing supers",
C
Chris Mason 已提交
3701
		       total_errors);
3702
		mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
3703

3704
		/* FUA is masked off if unsupported and can't be the reason */
3705
		btrfs_handle_fs_error(root->fs_info, -EIO,
3706 3707
			    "%d errors while writing supers", total_errors);
		return -EIO;
3708
	}
3709

Y
Yan Zheng 已提交
3710
	total_errors = 0;
3711
	list_for_each_entry_rcu(dev, head, dev_list) {
3712 3713
		if (!dev->bdev)
			continue;
Y
Yan Zheng 已提交
3714
		if (!dev->in_fs_metadata || !dev->writeable)
3715 3716
			continue;

Y
Yan Zheng 已提交
3717 3718 3719
		ret = write_dev_supers(dev, sb, do_barriers, 1, max_mirrors);
		if (ret)
			total_errors++;
3720
	}
3721
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
3722
	if (total_errors > max_errors) {
3723
		btrfs_handle_fs_error(root->fs_info, -EIO,
3724 3725
			    "%d errors while writing supers", total_errors);
		return -EIO;
3726
	}
3727 3728 3729
	return 0;
}

Y
Yan Zheng 已提交
3730 3731
int write_ctree_super(struct btrfs_trans_handle *trans,
		      struct btrfs_root *root, int max_mirrors)
3732
{
3733
	return write_all_supers(root, max_mirrors);
C
Chris Mason 已提交
3734 3735
}

3736 3737 3738
/* 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
3739
{
3740
	spin_lock(&fs_info->fs_roots_radix_lock);
C
Chris Mason 已提交
3741 3742
	radix_tree_delete(&fs_info->fs_roots_radix,
			  (unsigned long)root->root_key.objectid);
3743
	spin_unlock(&fs_info->fs_roots_radix_lock);
3744 3745 3746 3747

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

L
Liu Bo 已提交
3748
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
L
Liu Bo 已提交
3749
		btrfs_free_log(NULL, root);
L
Liu Bo 已提交
3750 3751 3752 3753 3754 3755 3756
		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
3757

3758 3759 3760 3761
	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);
3762 3763 3764 3765 3766
	free_fs_root(root);
}

static void free_fs_root(struct btrfs_root *root)
{
3767
	iput(root->ino_cache_inode);
3768
	WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree));
3769 3770
	btrfs_free_block_rsv(root, root->orphan_block_rsv);
	root->orphan_block_rsv = NULL;
3771 3772
	if (root->anon_dev)
		free_anon_bdev(root->anon_dev);
3773 3774
	if (root->subv_writers)
		btrfs_free_subvolume_writers(root->subv_writers);
3775 3776
	free_extent_buffer(root->node);
	free_extent_buffer(root->commit_root);
3777 3778
	kfree(root->free_ino_ctl);
	kfree(root->free_ino_pinned);
C
Chris Mason 已提交
3779
	kfree(root->name);
3780
	btrfs_put_fs_root(root);
C
Chris Mason 已提交
3781 3782
}

3783 3784 3785
void btrfs_free_fs_root(struct btrfs_root *root)
{
	free_fs_root(root);
C
Chris Mason 已提交
3786 3787
}

Y
Yan Zheng 已提交
3788
int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
C
Chris Mason 已提交
3789
{
Y
Yan Zheng 已提交
3790 3791
	u64 root_objectid = 0;
	struct btrfs_root *gang[8];
3792 3793 3794 3795
	int i = 0;
	int err = 0;
	unsigned int ret = 0;
	int index;
3796

Y
Yan Zheng 已提交
3797
	while (1) {
3798
		index = srcu_read_lock(&fs_info->subvol_srcu);
Y
Yan Zheng 已提交
3799 3800 3801
		ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix,
					     (void **)gang, root_objectid,
					     ARRAY_SIZE(gang));
3802 3803
		if (!ret) {
			srcu_read_unlock(&fs_info->subvol_srcu, index);
Y
Yan Zheng 已提交
3804
			break;
3805
		}
3806
		root_objectid = gang[ret - 1]->root_key.objectid + 1;
3807

Y
Yan Zheng 已提交
3808
		for (i = 0; i < ret; i++) {
3809 3810 3811 3812 3813 3814 3815 3816 3817
			/* 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);
3818

3819 3820 3821
		for (i = 0; i < ret; i++) {
			if (!gang[i])
				continue;
Y
Yan Zheng 已提交
3822
			root_objectid = gang[i]->root_key.objectid;
3823 3824
			err = btrfs_orphan_cleanup(gang[i]);
			if (err)
3825 3826
				break;
			btrfs_put_fs_root(gang[i]);
Y
Yan Zheng 已提交
3827 3828 3829
		}
		root_objectid++;
	}
3830 3831 3832 3833 3834 3835 3836

	/* 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
3837
}
3838

Y
Yan Zheng 已提交
3839 3840 3841
int btrfs_commit_super(struct btrfs_root *root)
{
	struct btrfs_trans_handle *trans;
3842

Y
Yan Zheng 已提交
3843
	mutex_lock(&root->fs_info->cleaner_mutex);
Y
Yan, Zheng 已提交
3844
	btrfs_run_delayed_iputs(root);
Y
Yan Zheng 已提交
3845
	mutex_unlock(&root->fs_info->cleaner_mutex);
D
David Sterba 已提交
3846
	wake_up_process(root->fs_info->cleaner_kthread);
3847 3848 3849 3850 3851

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

3852
	trans = btrfs_join_transaction(root);
3853 3854
	if (IS_ERR(trans))
		return PTR_ERR(trans);
3855
	return btrfs_commit_transaction(trans, root);
Y
Yan Zheng 已提交
3856 3857
}

3858
void close_ctree(struct btrfs_root *root)
Y
Yan Zheng 已提交
3859 3860 3861 3862 3863 3864 3865
{
	struct btrfs_fs_info *fs_info = root->fs_info;
	int ret;

	fs_info->closing = 1;
	smp_mb();

3866
	/* wait for the qgroup rescan worker to stop */
3867
	btrfs_qgroup_wait_for_completion(fs_info, false);
3868

S
Stefan Behrens 已提交
3869 3870 3871 3872 3873
	/* 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);

3874
	/* pause restriper - we want to resume on mount */
3875
	btrfs_pause_balance(fs_info);
3876

3877 3878
	btrfs_dev_replace_suspend_for_unmount(fs_info);

3879
	btrfs_scrub_cancel(fs_info);
C
Chris Mason 已提交
3880 3881 3882 3883 3884 3885

	/* 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 */
3886
	btrfs_cleanup_defrag_inodes(fs_info);
C
Chris Mason 已提交
3887

3888 3889
	cancel_work_sync(&fs_info->async_reclaim_work);

Y
Yan Zheng 已提交
3890
	if (!(fs_info->sb->s_flags & MS_RDONLY)) {
3891 3892 3893 3894 3895 3896 3897
		/*
		 * If the cleaner thread is stopped and there are
		 * block groups queued for removal, the deletion will be
		 * skipped when we quit the cleaner thread.
		 */
		btrfs_delete_unused_bgs(root->fs_info);

L
liubo 已提交
3898 3899
		ret = btrfs_commit_super(root);
		if (ret)
3900
			btrfs_err(fs_info, "commit super ret %d", ret);
L
liubo 已提交
3901 3902
	}

3903
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state))
3904
		btrfs_error_commit_super(root);
3905

A
Al Viro 已提交
3906 3907
	kthread_stop(fs_info->transaction_kthread);
	kthread_stop(fs_info->cleaner_kthread);
3908

3909 3910 3911
	fs_info->closing = 2;
	smp_mb();

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
	btrfs_free_block_groups(fs_info);

3928 3929 3930 3931 3932
	/*
	 * 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);
3933 3934
	btrfs_stop_all_workers(fs_info);

3935
	fs_info->open = 0;
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(root->fs_info, CHECK_INTEGRITY))
3942 3943 3944
		btrfsic_unmount(root, fs_info->fs_devices);
#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);
C
Chris Mason 已提交
3951
	bdi_destroy(&fs_info->bdi);
3952
	cleanup_srcu_struct(&fs_info->subvol_srcu);
3953

D
David Woodhouse 已提交
3954 3955
	btrfs_free_stripe_hash_table(fs_info);

3956
	__btrfs_free_block_rsv(root->orphan_block_rsv);
3957
	root->orphan_block_rsv = NULL;
3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968

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

		em = list_first_entry(&fs_info->pinned_chunks,
				      struct extent_map, list);
		list_del_init(&em->list);
		free_extent_map(em);
	}
	unlock_chunks(root);
3969 3970
}

3971 3972
int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid,
			  int atomic)
3973
{
3974
	int ret;
3975
	struct inode *btree_inode = buf->pages[0]->mapping->host;
3976

3977
	ret = extent_buffer_uptodate(buf);
3978 3979 3980 3981
	if (!ret)
		return ret;

	ret = verify_parent_transid(&BTRFS_I(btree_inode)->io_tree, buf,
3982 3983 3984
				    parent_transid, atomic);
	if (ret == -EAGAIN)
		return ret;
3985
	return !ret;
3986 3987 3988 3989
}

void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
{
3990
	struct btrfs_root *root;
3991
	u64 transid = btrfs_header_generation(buf);
3992
	int was_dirty;
3993

3994 3995 3996 3997 3998 3999 4000 4001 4002 4003
#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;
4004
	btrfs_assert_tree_locked(buf);
J
Julia Lawall 已提交
4005 4006
	if (transid != root->fs_info->generation)
		WARN(1, KERN_CRIT "btrfs transid mismatch buffer %llu, "
C
Chris Mason 已提交
4007
		       "found %llu running %llu\n",
4008
			buf->start, transid, root->fs_info->generation);
4009
	was_dirty = set_extent_buffer_dirty(buf);
4010 4011 4012 4013
	if (!was_dirty)
		__percpu_counter_add(&root->fs_info->dirty_metadata_bytes,
				     buf->len,
				     root->fs_info->dirty_metadata_batch);
4014 4015 4016 4017 4018 4019
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	if (btrfs_header_level(buf) == 0 && check_leaf(root, buf)) {
		btrfs_print_leaf(root, buf);
		ASSERT(0);
	}
#endif
4020 4021
}

4022 4023
static void __btrfs_btree_balance_dirty(struct btrfs_root *root,
					int flush_delayed)
4024 4025 4026 4027 4028
{
	/*
	 * looks as though older kernels can get into trouble with
	 * this code, they end up stuck in balance_dirty_pages forever
	 */
4029
	int ret;
4030 4031 4032 4033

	if (current->flags & PF_MEMALLOC)
		return;

4034 4035
	if (flush_delayed)
		btrfs_balance_delayed_items(root);
4036

4037 4038 4039
	ret = percpu_counter_compare(&root->fs_info->dirty_metadata_bytes,
				     BTRFS_DIRTY_METADATA_THRESH);
	if (ret > 0) {
4040 4041
		balance_dirty_pages_ratelimited(
				   root->fs_info->btree_inode->i_mapping);
4042 4043 4044
	}
}

4045
void btrfs_btree_balance_dirty(struct btrfs_root *root)
C
Chris Mason 已提交
4046
{
4047 4048
	__btrfs_btree_balance_dirty(root, 1);
}
4049

4050 4051 4052
void btrfs_btree_balance_dirty_nodelay(struct btrfs_root *root)
{
	__btrfs_btree_balance_dirty(root, 0);
C
Chris Mason 已提交
4053
}
4054

4055
int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid)
4056
{
4057
	struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root;
4058
	return btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
4059
}
4060

4061
static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info,
L
liubo 已提交
4062 4063
			      int read_only)
{
D
David Sterba 已提交
4064
	struct btrfs_super_block *sb = fs_info->super_copy;
4065 4066
	u64 nodesize = btrfs_super_nodesize(sb);
	u64 sectorsize = btrfs_super_sectorsize(sb);
D
David Sterba 已提交
4067 4068
	int ret = 0;

4069 4070 4071 4072 4073 4074 4075
	if (btrfs_super_magic(sb) != BTRFS_MAGIC) {
		printk(KERN_ERR "BTRFS: no valid FS found\n");
		ret = -EINVAL;
	}
	if (btrfs_super_flags(sb) & ~BTRFS_SUPER_FLAG_SUPP)
		printk(KERN_WARNING "BTRFS: unrecognized super flag: %llu\n",
				btrfs_super_flags(sb) & ~BTRFS_SUPER_FLAG_SUPP);
4076 4077 4078
	if (btrfs_super_root_level(sb) >= BTRFS_MAX_LEVEL) {
		printk(KERN_ERR "BTRFS: tree_root level too big: %d >= %d\n",
				btrfs_super_root_level(sb), BTRFS_MAX_LEVEL);
D
David Sterba 已提交
4079 4080
		ret = -EINVAL;
	}
4081 4082 4083
	if (btrfs_super_chunk_root_level(sb) >= BTRFS_MAX_LEVEL) {
		printk(KERN_ERR "BTRFS: chunk_root level too big: %d >= %d\n",
				btrfs_super_chunk_root_level(sb), BTRFS_MAX_LEVEL);
D
David Sterba 已提交
4084 4085
		ret = -EINVAL;
	}
4086 4087 4088
	if (btrfs_super_log_root_level(sb) >= BTRFS_MAX_LEVEL) {
		printk(KERN_ERR "BTRFS: log_root level too big: %d >= %d\n",
				btrfs_super_log_root_level(sb), BTRFS_MAX_LEVEL);
D
David Sterba 已提交
4089 4090 4091
		ret = -EINVAL;
	}

D
David Sterba 已提交
4092
	/*
4093 4094
	 * 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
4095
	 */
4096 4097 4098 4099 4100 4101
	if (!is_power_of_2(sectorsize) || sectorsize < 4096 ||
	    sectorsize > BTRFS_MAX_METADATA_BLOCKSIZE) {
		printk(KERN_ERR "BTRFS: invalid sectorsize %llu\n", sectorsize);
		ret = -EINVAL;
	}
	/* Only PAGE SIZE is supported yet */
4102
	if (sectorsize != PAGE_SIZE) {
4103
		printk(KERN_ERR "BTRFS: sectorsize %llu not supported yet, only support %lu\n",
4104
				sectorsize, PAGE_SIZE);
4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120
		ret = -EINVAL;
	}
	if (!is_power_of_2(nodesize) || nodesize < sectorsize ||
	    nodesize > BTRFS_MAX_METADATA_BLOCKSIZE) {
		printk(KERN_ERR "BTRFS: invalid nodesize %llu\n", nodesize);
		ret = -EINVAL;
	}
	if (nodesize != le32_to_cpu(sb->__unused_leafsize)) {
		printk(KERN_ERR "BTRFS: invalid leafsize %u, should be %llu\n",
				le32_to_cpu(sb->__unused_leafsize),
				nodesize);
		ret = -EINVAL;
	}

	/* Root alignment check */
	if (!IS_ALIGNED(btrfs_super_root(sb), sectorsize)) {
D
David Sterba 已提交
4121
		printk(KERN_WARNING "BTRFS: tree_root block unaligned: %llu\n",
4122
				btrfs_super_root(sb));
4123 4124 4125
		ret = -EINVAL;
	}
	if (!IS_ALIGNED(btrfs_super_chunk_root(sb), sectorsize)) {
4126 4127
		printk(KERN_WARNING "BTRFS: chunk_root block unaligned: %llu\n",
				btrfs_super_chunk_root(sb));
4128 4129
		ret = -EINVAL;
	}
4130 4131 4132
	if (!IS_ALIGNED(btrfs_super_log_root(sb), sectorsize)) {
		printk(KERN_WARNING "BTRFS: log_root block unaligned: %llu\n",
				btrfs_super_log_root(sb));
4133 4134 4135
		ret = -EINVAL;
	}

D
David Sterba 已提交
4136 4137 4138 4139 4140 4141 4142 4143 4144 4145
	if (memcmp(fs_info->fsid, sb->dev_item.fsid, BTRFS_UUID_SIZE) != 0) {
		printk(KERN_ERR "BTRFS: dev_item UUID does not match fsid: %pU != %pU\n",
				fs_info->fsid, sb->dev_item.fsid);
		ret = -EINVAL;
	}

	/*
	 * Hint to catch really bogus numbers, bitflips or so, more exact checks are
	 * done later
	 */
4146 4147 4148 4149 4150
	if (btrfs_super_bytes_used(sb) < 6 * btrfs_super_nodesize(sb)) {
		btrfs_err(fs_info, "bytes_used is too small %llu",
		       btrfs_super_bytes_used(sb));
		ret = -EINVAL;
	}
4151
	if (!is_power_of_2(btrfs_super_stripesize(sb))) {
4152 4153 4154 4155
		btrfs_err(fs_info, "invalid stripesize %u",
		       btrfs_super_stripesize(sb));
		ret = -EINVAL;
	}
4156
	if (btrfs_super_num_devices(sb) > (1UL << 31))
D
David Sterba 已提交
4157
		printk(KERN_WARNING "BTRFS: suspicious number of devices: %llu\n",
4158
				btrfs_super_num_devices(sb));
4159 4160 4161 4162
	if (btrfs_super_num_devices(sb) == 0) {
		printk(KERN_ERR "BTRFS: number of devices is 0\n");
		ret = -EINVAL;
	}
D
David Sterba 已提交
4163

4164
	if (btrfs_super_bytenr(sb) != BTRFS_SUPER_INFO_OFFSET) {
D
David Sterba 已提交
4165
		printk(KERN_ERR "BTRFS: super offset mismatch %llu != %u\n",
4166
				btrfs_super_bytenr(sb), BTRFS_SUPER_INFO_OFFSET);
D
David Sterba 已提交
4167 4168 4169
		ret = -EINVAL;
	}

4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181
	/*
	 * 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) {
		printk(KERN_ERR "BTRFS: system chunk array too big %u > %u\n",
				btrfs_super_sys_array_size(sb),
				BTRFS_SYSTEM_CHUNK_ARRAY_SIZE);
		ret = -EINVAL;
	}
	if (btrfs_super_sys_array_size(sb) < sizeof(struct btrfs_disk_key)
			+ sizeof(struct btrfs_chunk)) {
4182
		printk(KERN_ERR "BTRFS: system chunk array too small %u < %zu\n",
4183 4184 4185 4186 4187 4188
				btrfs_super_sys_array_size(sb),
				sizeof(struct btrfs_disk_key)
				+ sizeof(struct btrfs_chunk));
		ret = -EINVAL;
	}

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

	return ret;
L
liubo 已提交
4204 4205
}

4206
static void btrfs_error_commit_super(struct btrfs_root *root)
L
liubo 已提交
4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218
{
	mutex_lock(&root->fs_info->cleaner_mutex);
	btrfs_run_delayed_iputs(root);
	mutex_unlock(&root->fs_info->cleaner_mutex);

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

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

4219
static void btrfs_destroy_ordered_extents(struct btrfs_root *root)
L
liubo 已提交
4220 4221 4222
{
	struct btrfs_ordered_extent *ordered;

4223
	spin_lock(&root->ordered_extent_lock);
4224 4225 4226 4227
	/*
	 * This will just short circuit the ordered completion stuff which will
	 * make sure the ordered extent gets properly cleaned up.
	 */
4228
	list_for_each_entry(ordered, &root->ordered_extents,
4229 4230
			    root_extent_list)
		set_bit(BTRFS_ORDERED_IOERR, &ordered->flags);
4231 4232 4233 4234 4235 4236 4237 4238 4239 4240 4241 4242 4243 4244 4245
	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);
4246 4247
		list_move_tail(&root->ordered_root,
			       &fs_info->ordered_roots);
4248

4249
		spin_unlock(&fs_info->ordered_root_lock);
4250 4251
		btrfs_destroy_ordered_extents(root);

4252 4253
		cond_resched();
		spin_lock(&fs_info->ordered_root_lock);
4254 4255
	}
	spin_unlock(&fs_info->ordered_root_lock);
L
liubo 已提交
4256 4257
}

4258 4259
static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
				      struct btrfs_root *root)
L
liubo 已提交
4260 4261 4262 4263 4264 4265 4266 4267 4268
{
	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);
4269
	if (atomic_read(&delayed_refs->num_entries) == 0) {
4270
		spin_unlock(&delayed_refs->lock);
4271
		btrfs_info(root->fs_info, "delayed_refs has NO entry");
L
liubo 已提交
4272 4273 4274
		return ret;
	}

4275 4276
	while ((node = rb_first(&delayed_refs->href_root)) != NULL) {
		struct btrfs_delayed_ref_head *head;
4277
		struct btrfs_delayed_ref_node *tmp;
4278
		bool pin_bytes = false;
L
liubo 已提交
4279

4280 4281 4282 4283 4284
		head = rb_entry(node, struct btrfs_delayed_ref_head,
				href_node);
		if (!mutex_trylock(&head->mutex)) {
			atomic_inc(&head->node.refs);
			spin_unlock(&delayed_refs->lock);
4285

4286
			mutex_lock(&head->mutex);
4287
			mutex_unlock(&head->mutex);
4288 4289 4290 4291 4292
			btrfs_put_delayed_ref(&head->node);
			spin_lock(&delayed_refs->lock);
			continue;
		}
		spin_lock(&head->lock);
4293 4294
		list_for_each_entry_safe_reverse(ref, tmp, &head->ref_list,
						 list) {
4295
			ref->in_tree = 0;
4296
			list_del(&ref->list);
4297 4298
			atomic_dec(&delayed_refs->num_entries);
			btrfs_put_delayed_ref(ref);
4299
		}
4300 4301 4302 4303 4304 4305 4306 4307 4308 4309 4310 4311
		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
4312

4313 4314 4315 4316
		if (pin_bytes)
			btrfs_pin_extent(root, head->node.bytenr,
					 head->node.num_bytes, 1);
		btrfs_put_delayed_ref(&head->node);
L
liubo 已提交
4317 4318 4319 4320 4321 4322 4323 4324 4325
		cond_resched();
		spin_lock(&delayed_refs->lock);
	}

	spin_unlock(&delayed_refs->lock);

	return ret;
}

4326
static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root)
L
liubo 已提交
4327 4328 4329 4330 4331 4332
{
	struct btrfs_inode *btrfs_inode;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

4333 4334
	spin_lock(&root->delalloc_lock);
	list_splice_init(&root->delalloc_inodes, &splice);
L
liubo 已提交
4335 4336

	while (!list_empty(&splice)) {
4337 4338
		btrfs_inode = list_first_entry(&splice, struct btrfs_inode,
					       delalloc_inodes);
L
liubo 已提交
4339 4340

		list_del_init(&btrfs_inode->delalloc_inodes);
4341 4342
		clear_bit(BTRFS_INODE_IN_DELALLOC_LIST,
			  &btrfs_inode->runtime_flags);
4343
		spin_unlock(&root->delalloc_lock);
L
liubo 已提交
4344 4345

		btrfs_invalidate_inodes(btrfs_inode->root);
4346

4347
		spin_lock(&root->delalloc_lock);
L
liubo 已提交
4348 4349
	}

4350 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
	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
4376 4377 4378 4379 4380 4381 4382 4383 4384 4385 4386 4387 4388
}

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

	while (1) {
		ret = find_first_extent_bit(dirty_pages, start, &start, &end,
4389
					    mark, NULL);
L
liubo 已提交
4390 4391 4392
		if (ret)
			break;

4393
		clear_extent_bits(dirty_pages, start, end, mark);
L
liubo 已提交
4394
		while (start <= end) {
4395
			eb = btrfs_find_tree_block(root->fs_info, start);
4396
			start += root->nodesize;
4397
			if (!eb)
L
liubo 已提交
4398
				continue;
4399
			wait_on_extent_buffer_writeback(eb);
L
liubo 已提交
4400

4401 4402 4403 4404
			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
4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417
		}
	}

	return ret;
}

static int btrfs_destroy_pinned_extent(struct btrfs_root *root,
				       struct extent_io_tree *pinned_extents)
{
	struct extent_io_tree *unpin;
	u64 start;
	u64 end;
	int ret;
4418
	bool loop = true;
L
liubo 已提交
4419 4420

	unpin = pinned_extents;
4421
again:
L
liubo 已提交
4422 4423
	while (1) {
		ret = find_first_extent_bit(unpin, 0, &start, &end,
4424
					    EXTENT_DIRTY, NULL);
L
liubo 已提交
4425 4426 4427
		if (ret)
			break;

4428
		clear_extent_dirty(unpin, start, end);
L
liubo 已提交
4429 4430 4431 4432
		btrfs_error_unpin_extent_range(root, start, end);
		cond_resched();
	}

4433 4434 4435 4436 4437 4438 4439 4440 4441
	if (loop) {
		if (unpin == &root->fs_info->freed_extents[0])
			unpin = &root->fs_info->freed_extents[1];
		else
			unpin = &root->fs_info->freed_extents[0];
		loop = false;
		goto again;
	}

L
liubo 已提交
4442 4443 4444
	return 0;
}

4445 4446 4447 4448 4449
void btrfs_cleanup_one_transaction(struct btrfs_transaction *cur_trans,
				   struct btrfs_root *root)
{
	btrfs_destroy_delayed_refs(cur_trans, root);

4450
	cur_trans->state = TRANS_STATE_COMMIT_START;
4451
	wake_up(&root->fs_info->transaction_blocked_wait);
4452

4453
	cur_trans->state = TRANS_STATE_UNBLOCKED;
4454
	wake_up(&root->fs_info->transaction_wait);
4455

4456 4457
	btrfs_destroy_delayed_inodes(root);
	btrfs_assert_delayed_root_empty(root);
4458 4459 4460

	btrfs_destroy_marked_extents(root, &cur_trans->dirty_pages,
				     EXTENT_DIRTY);
4461 4462
	btrfs_destroy_pinned_extent(root,
				    root->fs_info->pinned_extents);
4463

4464 4465 4466
	cur_trans->state =TRANS_STATE_COMPLETED;
	wake_up(&cur_trans->commit_wait);

4467 4468 4469 4470 4471 4472
	/*
	memset(cur_trans, 0, sizeof(*cur_trans));
	kmem_cache_free(btrfs_transaction_cachep, cur_trans);
	*/
}

4473
static int btrfs_cleanup_transaction(struct btrfs_root *root)
L
liubo 已提交
4474 4475 4476 4477 4478
{
	struct btrfs_transaction *t;

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

J
Josef Bacik 已提交
4479
	spin_lock(&root->fs_info->trans_lock);
4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503
	while (!list_empty(&root->fs_info->trans_list)) {
		t = list_first_entry(&root->fs_info->trans_list,
				     struct btrfs_transaction, list);
		if (t->state >= TRANS_STATE_COMMIT_START) {
			atomic_inc(&t->use_count);
			spin_unlock(&root->fs_info->trans_lock);
			btrfs_wait_for_commit(root, t->transid);
			btrfs_put_transaction(t);
			spin_lock(&root->fs_info->trans_lock);
			continue;
		}
		if (t == root->fs_info->running_transaction) {
			t->state = TRANS_STATE_COMMIT_DOING;
			spin_unlock(&root->fs_info->trans_lock);
			/*
			 * We wait for 0 num_writers since we don't hold a trans
			 * handle open currently for this transaction.
			 */
			wait_event(t->writer_wait,
				   atomic_read(&t->num_writers) == 0);
		} else {
			spin_unlock(&root->fs_info->trans_lock);
		}
		btrfs_cleanup_one_transaction(t, root);
4504

4505 4506 4507
		spin_lock(&root->fs_info->trans_lock);
		if (t == root->fs_info->running_transaction)
			root->fs_info->running_transaction = NULL;
L
liubo 已提交
4508
		list_del_init(&t->list);
4509
		spin_unlock(&root->fs_info->trans_lock);
L
liubo 已提交
4510

4511 4512 4513 4514 4515 4516 4517 4518 4519 4520
		btrfs_put_transaction(t);
		trace_btrfs_transaction_commit(root);
		spin_lock(&root->fs_info->trans_lock);
	}
	spin_unlock(&root->fs_info->trans_lock);
	btrfs_destroy_all_ordered_extents(root->fs_info);
	btrfs_destroy_delayed_inodes(root);
	btrfs_assert_delayed_root_empty(root);
	btrfs_destroy_pinned_extent(root, root->fs_info->pinned_extents);
	btrfs_destroy_all_delalloc_inodes(root->fs_info);
L
liubo 已提交
4521 4522 4523 4524 4525
	mutex_unlock(&root->fs_info->transaction_kthread_mutex);

	return 0;
}

4526
static const struct extent_io_ops btree_extent_io_ops = {
4527
	.readpage_end_io_hook = btree_readpage_end_io_hook,
A
Arne Jansen 已提交
4528
	.readpage_io_failed_hook = btree_io_failed_hook,
4529
	.submit_bio_hook = btree_submit_bio_hook,
4530 4531
	/* note we're sharing with inode.c for the merge bio hook */
	.merge_bio_hook = btrfs_merge_bio_hook,
4532
};