disk-io.c 124.0 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
	int mirror_num;
C
Chris Mason 已提交
128
	unsigned long bio_flags;
129 130 131 132 133
	/*
	 * bio_offset is optional, can be used if the pages in the bio
	 * can't tell us where in the file the bio should go
	 */
	u64 bio_offset;
134
	struct btrfs_work work;
135
	int error;
136 137
};

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

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

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

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

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

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

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

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

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

217 218
#endif

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

231
	read_lock(&em_tree->lock);
232
	em = lookup_extent_mapping(em_tree, start, len);
233 234 235
	if (em) {
		em->bdev =
			BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
236
		read_unlock(&em_tree->lock);
237
		goto out;
238
	}
239
	read_unlock(&em_tree->lock);
240

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

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

265 266
out:
	return em;
267 268
}

269
u32 btrfs_csum_data(char *data, u32 seed, size_t len)
270
{
271
	return btrfs_crc32c(seed, data, len);
272 273 274 275
}

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

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

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

319 320 321
	btrfs_csum_final(crc, result);

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

327
			read_extent_buffer(buf, &val, 0, csum_size);
328 329 330
			btrfs_warn_rl(fs_info,
				"%s checksum verify failed on %llu wanted %X found %X "
				"level %d",
331
				fs_info->sb->s_id, buf->start,
332
				val, found, btrfs_header_level(buf));
333 334
			if (result != (char *)&inline_result)
				kfree(result);
335
			return -EUCLEAN;
336 337
		}
	} else {
338
		write_extent_buffer(buf, result, 0, csum_size);
339
	}
340 341
	if (result != (char *)&inline_result)
		kfree(result);
342 343 344
	return 0;
}

C
Chris Mason 已提交
345 346 347 348 349 350
/*
 * 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.
 */
351
static int verify_parent_transid(struct extent_io_tree *io_tree,
352 353
				 struct extent_buffer *eb, u64 parent_transid,
				 int atomic)
354
{
355
	struct extent_state *cached_state = NULL;
356
	int ret;
357
	bool need_lock = (current->journal_info == BTRFS_SEND_TRANS_STUB);
358 359 360 361

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

362 363 364
	if (atomic)
		return -EAGAIN;

365 366 367 368 369
	if (need_lock) {
		btrfs_tree_read_lock(eb);
		btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
	}

370
	lock_extent_bits(io_tree, eb->start, eb->start + eb->len - 1,
371
			 &cached_state);
372
	if (extent_buffer_uptodate(eb) &&
373 374 375 376
	    btrfs_header_generation(eb) == parent_transid) {
		ret = 0;
		goto out;
	}
377 378 379
	btrfs_err_rl(eb->fs_info,
		"parent transid verify failed on %llu wanted %llu found %llu",
			eb->start,
380
			parent_transid, btrfs_header_generation(eb));
381
	ret = 1;
382 383 384 385

	/*
	 * 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
386
	 * block that has been freed and re-allocated.  So don't clear uptodate
387 388 389 390 391 392
	 * 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
393
out:
394 395
	unlock_extent_cached(io_tree, eb->start, eb->start + eb->len - 1,
			     &cached_state, GFP_NOFS);
396 397
	if (need_lock)
		btrfs_tree_read_unlock_blocking(eb);
398 399 400
	return ret;
}

D
David Sterba 已提交
401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419
/*
 * 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
420
		 * is filled with zeros and is included in the checksum.
D
David Sterba 已提交
421 422 423 424 425 426 427 428 429 430
		 */
		crc = btrfs_csum_data(raw_disk_sb + BTRFS_CSUM_SIZE,
				crc, BTRFS_SUPER_INFO_SIZE - BTRFS_CSUM_SIZE);
		btrfs_csum_final(crc, result);

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

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

	return ret;
}

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

454
	clear_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
455 456
	io_tree = &BTRFS_I(root->fs_info->btree_inode)->io_tree;
	while (1) {
457 458
		ret = read_extent_buffer_pages(io_tree, eb, start,
					       WAIT_COMPLETE,
459
					       btree_get_extent, mirror_num);
460 461
		if (!ret) {
			if (!verify_parent_transid(io_tree, eb,
462
						   parent_transid, 0))
463 464 465 466
				break;
			else
				ret = -EIO;
		}
C
Chris Mason 已提交
467

468 469 470 471 472 473
		/*
		 * 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))
474 475
			break;

476
		num_copies = btrfs_num_copies(root->fs_info,
477
					      eb->start, eb->len);
C
Chris Mason 已提交
478
		if (num_copies == 1)
479
			break;
C
Chris Mason 已提交
480

481 482 483 484 485
		if (!failed_mirror) {
			failed = 1;
			failed_mirror = eb->read_mirror;
		}

486
		mirror_num++;
487 488 489
		if (mirror_num == failed_mirror)
			mirror_num++;

C
Chris Mason 已提交
490
		if (mirror_num > num_copies)
491
			break;
492
	}
493

494
	if (failed && !ret && failed_mirror)
495 496 497
		repair_eb_io_failure(root, eb, failed_mirror);

	return ret;
498
}
499

C
Chris Mason 已提交
500
/*
C
Chris Mason 已提交
501 502
 * 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
503
 */
C
Chris Mason 已提交
504

505
static int csum_dirty_buffer(struct btrfs_fs_info *fs_info, struct page *page)
506
{
M
Miao Xie 已提交
507
	u64 start = page_offset(page);
508 509
	u64 found_start;
	struct extent_buffer *eb;
510

J
Josef Bacik 已提交
511 512 513
	eb = (struct extent_buffer *)page->private;
	if (page != eb->pages[0])
		return 0;
514

515
	found_start = btrfs_header_bytenr(eb);
516 517 518 519 520 521 522 523 524 525 526 527
	/*
	 * 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);

528
	return csum_tree_block(fs_info, eb, 0);
529 530
}

531
static int check_tree_block_fsid(struct btrfs_fs_info *fs_info,
Y
Yan Zheng 已提交
532 533
				 struct extent_buffer *eb)
{
534
	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
Y
Yan Zheng 已提交
535 536 537
	u8 fsid[BTRFS_UUID_SIZE];
	int ret = 1;

538
	read_extent_buffer(eb, fsid, btrfs_header_fsid(), BTRFS_FSID_SIZE);
Y
Yan Zheng 已提交
539 540 541 542 543 544 545 546 547 548
	while (fs_devices) {
		if (!memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE)) {
			ret = 0;
			break;
		}
		fs_devices = fs_devices->seed;
	}
	return ret;
}

549
#define CORRUPT(reason, eb, root, slot)				\
550 551
	btrfs_crit(root->fs_info, "corrupt leaf, %s: block=%llu,"	\
		   "root=%llu, slot=%d", reason,			\
552
	       btrfs_header_bytenr(eb),	root->objectid, slot)
553 554 555 556 557 558 559 560 561

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;

562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582
	if (nritems == 0) {
		struct btrfs_root *check_root;

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

		check_root = btrfs_get_fs_root(root->fs_info, &key, false);
		/*
		 * The only reason we also check NULL here is that during
		 * open_ctree() some roots has not yet been set up.
		 */
		if (!IS_ERR_OR_NULL(check_root)) {
			/* if leaf is the root, then it's fine */
			if (leaf->start !=
			    btrfs_root_bytenr(&check_root->root_item)) {
				CORRUPT("non-root leaf's nritems is 0",
					leaf, root, 0);
				return -EIO;
			}
		}
583
		return 0;
584
	}
585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622

	/* 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,
623
		 * just in case all the items are consistent to each other, but
624 625 626 627 628 629 630 631 632 633 634 635
		 * 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;
}

L
Liu Bo 已提交
636 637 638 639 640 641 642 643 644 645 646 647 648
static int check_node(struct btrfs_root *root, struct extent_buffer *node)
{
	unsigned long nr = btrfs_header_nritems(node);

	if (nr == 0 || nr > BTRFS_NODEPTRS_PER_BLOCK(root)) {
		btrfs_crit(root->fs_info,
			   "corrupt node: block %llu root %llu nritems %lu",
			   node->start, root->objectid, nr);
		return -EIO;
	}
	return 0;
}

649 650 651
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)
652 653 654 655 656
{
	u64 found_start;
	int found_level;
	struct extent_buffer *eb;
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
657
	struct btrfs_fs_info *fs_info = root->fs_info;
658
	int ret = 0;
659
	int reads_done;
660 661 662

	if (!page->private)
		goto out;
C
Chris Mason 已提交
663

J
Josef Bacik 已提交
664
	eb = (struct extent_buffer *)page->private;
C
Chris Mason 已提交
665

666 667 668 669 670 671
	/* 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);
672 673
	if (!reads_done)
		goto err;
674

675
	eb->read_mirror = mirror;
676
	if (test_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags)) {
677 678 679 680
		ret = -EIO;
		goto err;
	}

681
	found_start = btrfs_header_bytenr(eb);
682
	if (found_start != eb->start) {
683 684
		btrfs_err_rl(fs_info, "bad tree block start %llu %llu",
			     found_start, eb->start);
685
		ret = -EIO;
686 687
		goto err;
	}
688 689 690
	if (check_tree_block_fsid(fs_info, eb)) {
		btrfs_err_rl(fs_info, "bad fsid on block %llu",
			     eb->start);
691 692 693
		ret = -EIO;
		goto err;
	}
694
	found_level = btrfs_header_level(eb);
695
	if (found_level >= BTRFS_MAX_LEVEL) {
696 697
		btrfs_err(fs_info, "bad tree block level %d",
			  (int)btrfs_header_level(eb));
698 699 700
		ret = -EIO;
		goto err;
	}
701

702 703
	btrfs_set_buffer_lockdep_class(btrfs_header_owner(eb),
				       eb, found_level);
704

705
	ret = csum_tree_block(fs_info, eb, 1);
706
	if (ret)
707 708 709 710 711 712 713 714 715 716 717
		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;
	}
718

L
Liu Bo 已提交
719 720 721
	if (found_level > 0 && check_node(root, eb))
		ret = -EIO;

722 723
	if (!ret)
		set_extent_buffer_uptodate(eb);
724
err:
725 726
	if (reads_done &&
	    test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
727
		btree_readahead_hook(fs_info, eb, eb->start, ret);
A
Arne Jansen 已提交
728

D
David Woodhouse 已提交
729 730 731 732 733 734 735
	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);
736
		clear_extent_buffer_uptodate(eb);
D
David Woodhouse 已提交
737
	}
738
	free_extent_buffer(eb);
739
out:
740
	return ret;
741 742
}

743
static int btree_io_failed_hook(struct page *page, int failed_mirror)
A
Arne Jansen 已提交
744 745 746
{
	struct extent_buffer *eb;

J
Josef Bacik 已提交
747
	eb = (struct extent_buffer *)page->private;
748
	set_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags);
749
	eb->read_mirror = failed_mirror;
D
David Woodhouse 已提交
750
	atomic_dec(&eb->io_pages);
751
	if (test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
752
		btree_readahead_hook(eb->fs_info, eb, eb->start, -EIO);
A
Arne Jansen 已提交
753 754 755
	return -EIO;	/* we fixed nothing */
}

756
static void end_workqueue_bio(struct bio *bio)
757
{
758
	struct btrfs_end_io_wq *end_io_wq = bio->bi_private;
759
	struct btrfs_fs_info *fs_info;
760 761
	struct btrfs_workqueue *wq;
	btrfs_work_func_t func;
762 763

	fs_info = end_io_wq->info;
764
	end_io_wq->error = bio->bi_error;
765

M
Mike Christie 已提交
766
	if (bio_op(bio) == REQ_OP_WRITE) {
767 768 769 770 771 772 773 774 775 776 777 778 779
		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;
		}
780
	} else {
781 782 783 784 785
		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) {
786 787 788 789 790 791 792 793 794
			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;
		}
795
	}
796 797 798

	btrfs_init_work(&end_io_wq->work, func, end_workqueue_fn, NULL, NULL);
	btrfs_queue_work(wq, &end_io_wq->work);
799 800
}

801
int btrfs_bio_wq_end_io(struct btrfs_fs_info *info, struct bio *bio,
802
			enum btrfs_wq_endio_type metadata)
803
{
804
	struct btrfs_end_io_wq *end_io_wq;
805

806
	end_io_wq = kmem_cache_alloc(btrfs_end_io_wq_cache, GFP_NOFS);
807 808 809 810 811
	if (!end_io_wq)
		return -ENOMEM;

	end_io_wq->private = bio->bi_private;
	end_io_wq->end_io = bio->bi_end_io;
812
	end_io_wq->info = info;
813 814
	end_io_wq->error = 0;
	end_io_wq->bio = bio;
815
	end_io_wq->metadata = metadata;
816 817 818

	bio->bi_private = end_io_wq;
	bio->bi_end_io = end_workqueue_bio;
819 820 821
	return 0;
}

822
unsigned long btrfs_async_submit_limit(struct btrfs_fs_info *info)
823
{
824
	unsigned long limit = min_t(unsigned long,
825
				    info->thread_pool_size,
826 827 828
				    info->fs_devices->open_devices);
	return 256 * limit;
}
829

C
Chris Mason 已提交
830 831 832
static void run_one_async_start(struct btrfs_work *work)
{
	struct async_submit_bio *async;
833
	int ret;
C
Chris Mason 已提交
834 835

	async = container_of(work, struct  async_submit_bio, work);
836
	ret = async->submit_bio_start(async->inode, async->bio,
837 838 839 840
				      async->mirror_num, async->bio_flags,
				      async->bio_offset);
	if (ret)
		async->error = ret;
C
Chris Mason 已提交
841 842 843
}

static void run_one_async_done(struct btrfs_work *work)
844 845 846
{
	struct btrfs_fs_info *fs_info;
	struct async_submit_bio *async;
847
	int limit;
848 849 850

	async = container_of(work, struct  async_submit_bio, work);
	fs_info = BTRFS_I(async->inode)->root->fs_info;
851

852
	limit = btrfs_async_submit_limit(fs_info);
853 854
	limit = limit * 2 / 3;

855 856 857
	/*
	 * atomic_dec_return implies a barrier for waitqueue_active
	 */
858
	if (atomic_dec_return(&fs_info->nr_async_submits) < limit &&
859
	    waitqueue_active(&fs_info->async_submit_wait))
860 861
		wake_up(&fs_info->async_submit_wait);

862
	/* If an error occurred we just want to clean up the bio and move on */
863
	if (async->error) {
864 865
		async->bio->bi_error = async->error;
		bio_endio(async->bio);
866 867 868
		return;
	}

869 870
	async->submit_bio_done(async->inode, async->bio, async->mirror_num,
			       async->bio_flags, async->bio_offset);
C
Chris Mason 已提交
871 872 873 874 875 876 877
}

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

	async = container_of(work, struct  async_submit_bio, work);
878 879 880
	kfree(async);
}

881
int btrfs_wq_submit_bio(struct btrfs_fs_info *fs_info, struct inode *inode,
882
			struct bio *bio, int mirror_num,
C
Chris Mason 已提交
883
			unsigned long bio_flags,
884
			u64 bio_offset,
C
Chris Mason 已提交
885 886
			extent_submit_bio_hook_t *submit_bio_start,
			extent_submit_bio_hook_t *submit_bio_done)
887 888 889 890 891 892 893 894 895 896
{
	struct async_submit_bio *async;

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

	async->inode = inode;
	async->bio = bio;
	async->mirror_num = mirror_num;
C
Chris Mason 已提交
897 898 899
	async->submit_bio_start = submit_bio_start;
	async->submit_bio_done = submit_bio_done;

900
	btrfs_init_work(&async->work, btrfs_worker_helper, run_one_async_start,
901
			run_one_async_done, run_one_async_free);
C
Chris Mason 已提交
902

C
Chris Mason 已提交
903
	async->bio_flags = bio_flags;
904
	async->bio_offset = bio_offset;
905

906 907
	async->error = 0;

908
	atomic_inc(&fs_info->nr_async_submits);
909

J
Jens Axboe 已提交
910
	if (bio->bi_opf & REQ_SYNC)
911
		btrfs_set_work_high_priority(&async->work);
912

913
	btrfs_queue_work(fs_info->workers, &async->work);
914

C
Chris Mason 已提交
915
	while (atomic_read(&fs_info->async_submit_draining) &&
916 917 918 919 920
	      atomic_read(&fs_info->nr_async_submits)) {
		wait_event(fs_info->async_submit_wait,
			   (atomic_read(&fs_info->nr_async_submits) == 0));
	}

921 922 923
	return 0;
}

924 925
static int btree_csum_one_bio(struct bio *bio)
{
926
	struct bio_vec *bvec;
927
	struct btrfs_root *root;
928
	int i, ret = 0;
929

930
	bio_for_each_segment_all(bvec, bio, i) {
931
		root = BTRFS_I(bvec->bv_page->mapping->host)->root;
932
		ret = csum_dirty_buffer(root->fs_info, bvec->bv_page);
933 934
		if (ret)
			break;
935
	}
936

937
	return ret;
938 939
}

940 941
static int __btree_submit_bio_start(struct inode *inode, struct bio *bio,
				    int mirror_num, unsigned long bio_flags,
942
				    u64 bio_offset)
943
{
944 945
	/*
	 * when we're called for a write, we're already in the async
946
	 * submission context.  Just jump into btrfs_map_bio
947
	 */
948
	return btree_csum_one_bio(bio);
C
Chris Mason 已提交
949
}
950

951
static int __btree_submit_bio_done(struct inode *inode, struct bio *bio,
952 953
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
C
Chris Mason 已提交
954
{
955 956
	int ret;

957
	/*
C
Chris Mason 已提交
958 959
	 * when we're called for a write, we're already in the async
	 * submission context.  Just jump into btrfs_map_bio
960
	 */
961
	ret = btrfs_map_bio(BTRFS_I(inode)->root, bio, mirror_num, 1);
962 963 964 965
	if (ret) {
		bio->bi_error = ret;
		bio_endio(bio);
	}
966
	return ret;
967 968
}

969 970 971 972 973
static int check_async_write(struct inode *inode, unsigned long bio_flags)
{
	if (bio_flags & EXTENT_BIO_TREE_LOG)
		return 0;
#ifdef CONFIG_X86
974
	if (static_cpu_has(X86_FEATURE_XMM4_2))
975 976 977 978 979
		return 0;
#endif
	return 1;
}

980
static int btree_submit_bio_hook(struct inode *inode, struct bio *bio,
981 982
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
983
{
984
	int async = check_async_write(inode, bio_flags);
985 986
	int ret;

M
Mike Christie 已提交
987
	if (bio_op(bio) != REQ_OP_WRITE) {
C
Chris Mason 已提交
988 989 990 991
		/*
		 * called for a read, do the setup so that checksum validation
		 * can happen in the async kernel threads
		 */
992
		ret = btrfs_bio_wq_end_io(BTRFS_I(inode)->root->fs_info,
993
					  bio, BTRFS_WQ_ENDIO_METADATA);
994
		if (ret)
995
			goto out_w_error;
996
		ret = btrfs_map_bio(BTRFS_I(inode)->root, bio, mirror_num, 0);
997 998 999
	} else if (!async) {
		ret = btree_csum_one_bio(bio);
		if (ret)
1000
			goto out_w_error;
1001
		ret = btrfs_map_bio(BTRFS_I(inode)->root, bio, mirror_num, 0);
1002 1003 1004 1005 1006 1007
	} 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,
1008
					  inode, bio, mirror_num, 0,
1009 1010 1011
					  bio_offset,
					  __btree_submit_bio_start,
					  __btree_submit_bio_done);
1012
	}
1013

1014 1015 1016 1017
	if (ret)
		goto out_w_error;
	return 0;

1018
out_w_error:
1019 1020
	bio->bi_error = ret;
	bio_endio(bio);
1021
	return ret;
1022 1023
}

J
Jan Beulich 已提交
1024
#ifdef CONFIG_MIGRATION
1025
static int btree_migratepage(struct address_space *mapping,
1026 1027
			struct page *newpage, struct page *page,
			enum migrate_mode mode)
1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041
{
	/*
	 * 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;
1042
	return migrate_page(mapping, newpage, page, mode);
1043
}
J
Jan Beulich 已提交
1044
#endif
1045

1046 1047 1048 1049

static int btree_writepages(struct address_space *mapping,
			    struct writeback_control *wbc)
{
1050 1051 1052
	struct btrfs_fs_info *fs_info;
	int ret;

1053
	if (wbc->sync_mode == WB_SYNC_NONE) {
1054 1055 1056 1057

		if (wbc->for_kupdate)
			return 0;

1058
		fs_info = BTRFS_I(mapping->host)->root->fs_info;
1059
		/* this is a bit racy, but that's ok */
1060 1061 1062
		ret = percpu_counter_compare(&fs_info->dirty_metadata_bytes,
					     BTRFS_DIRTY_METADATA_THRESH);
		if (ret < 0)
1063 1064
			return 0;
	}
1065
	return btree_write_cache_pages(mapping, wbc);
1066 1067
}

1068
static int btree_readpage(struct file *file, struct page *page)
1069
{
1070 1071
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
1072
	return extent_read_full_page(tree, page, btree_get_extent, 0);
1073
}
C
Chris Mason 已提交
1074

1075
static int btree_releasepage(struct page *page, gfp_t gfp_flags)
1076
{
1077
	if (PageWriteback(page) || PageDirty(page))
C
Chris Mason 已提交
1078
		return 0;
1079

1080
	return try_release_extent_buffer(page);
1081 1082
}

1083 1084
static void btree_invalidatepage(struct page *page, unsigned int offset,
				 unsigned int length)
1085
{
1086 1087
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
1088 1089
	extent_invalidatepage(tree, page, offset);
	btree_releasepage(page, GFP_NOFS);
1090
	if (PagePrivate(page)) {
1091 1092 1093
		btrfs_warn(BTRFS_I(page->mapping->host)->root->fs_info,
			   "page private not zero on page %llu",
			   (unsigned long long)page_offset(page));
1094 1095
		ClearPagePrivate(page);
		set_page_private(page, 0);
1096
		put_page(page);
1097
	}
1098 1099
}

1100 1101
static int btree_set_page_dirty(struct page *page)
{
1102
#ifdef DEBUG
1103 1104 1105 1106 1107 1108 1109 1110
	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);
1111
#endif
1112 1113 1114
	return __set_page_dirty_nobuffers(page);
}

1115
static const struct address_space_operations btree_aops = {
1116
	.readpage	= btree_readpage,
1117
	.writepages	= btree_writepages,
1118 1119
	.releasepage	= btree_releasepage,
	.invalidatepage = btree_invalidatepage,
1120
#ifdef CONFIG_MIGRATION
1121
	.migratepage	= btree_migratepage,
1122
#endif
1123
	.set_page_dirty = btree_set_page_dirty,
1124 1125
};

1126
void readahead_tree_block(struct btrfs_root *root, u64 bytenr)
C
Chris Mason 已提交
1127
{
1128 1129
	struct extent_buffer *buf = NULL;
	struct inode *btree_inode = root->fs_info->btree_inode;
C
Chris Mason 已提交
1130

1131
	buf = btrfs_find_create_tree_block(root, bytenr);
1132
	if (IS_ERR(buf))
1133
		return;
1134
	read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree,
1135
				 buf, 0, WAIT_NONE, btree_get_extent, 0);
1136
	free_extent_buffer(buf);
C
Chris Mason 已提交
1137 1138
}

1139
int reada_tree_block_flagged(struct btrfs_root *root, u64 bytenr,
1140 1141 1142 1143 1144 1145 1146
			 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;

1147
	buf = btrfs_find_create_tree_block(root, bytenr);
1148
	if (IS_ERR(buf))
1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162
		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;
1163
	} else if (extent_buffer_uptodate(buf)) {
1164 1165 1166 1167 1168 1169 1170
		*eb = buf;
	} else {
		free_extent_buffer(buf);
	}
	return 0;
}

1171
struct extent_buffer *btrfs_find_tree_block(struct btrfs_fs_info *fs_info,
1172
					    u64 bytenr)
1173
{
1174
	return find_extent_buffer(fs_info, bytenr);
1175 1176 1177
}

struct extent_buffer *btrfs_find_create_tree_block(struct btrfs_root *root,
1178
						 u64 bytenr)
1179
{
1180
	if (btrfs_is_testing(root->fs_info))
1181 1182
		return alloc_test_extent_buffer(root->fs_info, bytenr,
				root->nodesize);
1183
	return alloc_extent_buffer(root->fs_info, bytenr);
1184 1185 1186
}


1187 1188
int btrfs_write_tree_block(struct extent_buffer *buf)
{
1189
	return filemap_fdatawrite_range(buf->pages[0]->mapping, buf->start,
1190
					buf->start + buf->len - 1);
1191 1192 1193 1194
}

int btrfs_wait_tree_block_writeback(struct extent_buffer *buf)
{
1195
	return filemap_fdatawait_range(buf->pages[0]->mapping,
1196
				       buf->start, buf->start + buf->len - 1);
1197 1198
}

1199
struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr,
1200
				      u64 parent_transid)
1201 1202 1203 1204
{
	struct extent_buffer *buf = NULL;
	int ret;

1205
	buf = btrfs_find_create_tree_block(root, bytenr);
1206 1207
	if (IS_ERR(buf))
		return buf;
1208

1209
	ret = btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
1210 1211
	if (ret) {
		free_extent_buffer(buf);
1212
		return ERR_PTR(ret);
1213
	}
1214
	return buf;
1215

1216 1217
}

1218 1219
void clean_tree_block(struct btrfs_trans_handle *trans,
		      struct btrfs_fs_info *fs_info,
1220
		      struct extent_buffer *buf)
1221
{
1222
	if (btrfs_header_generation(buf) ==
1223
	    fs_info->running_transaction->transid) {
1224
		btrfs_assert_tree_locked(buf);
1225

1226
		if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &buf->bflags)) {
1227 1228 1229
			__percpu_counter_add(&fs_info->dirty_metadata_bytes,
					     -buf->len,
					     fs_info->dirty_metadata_batch);
1230 1231 1232 1233
			/* ugh, clear_extent_buffer_dirty needs to lock the page */
			btrfs_set_lock_blocking(buf);
			clear_extent_buffer_dirty(buf);
		}
1234
	}
1235 1236
}

1237 1238 1239 1240 1241 1242 1243 1244 1245
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);

1246
	ret = percpu_counter_init(&writers->counter, 0, GFP_KERNEL);
1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262
	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);
}

1263 1264
static void __setup_root(u32 nodesize, u32 sectorsize, u32 stripesize,
			 struct btrfs_root *root, struct btrfs_fs_info *fs_info,
1265
			 u64 objectid)
1266
{
1267
	bool dummy = test_bit(BTRFS_FS_STATE_DUMMY_FS_INFO, &fs_info->fs_state);
C
Chris Mason 已提交
1268
	root->node = NULL;
1269
	root->commit_root = NULL;
1270 1271
	root->sectorsize = sectorsize;
	root->nodesize = nodesize;
1272
	root->stripesize = stripesize;
1273
	root->state = 0;
1274
	root->orphan_cleanup_state = 0;
1275

1276 1277
	root->objectid = objectid;
	root->last_trans = 0;
1278
	root->highest_objectid = 0;
1279
	root->nr_delalloc_inodes = 0;
1280
	root->nr_ordered_extents = 0;
1281
	root->name = NULL;
1282
	root->inode_tree = RB_ROOT;
1283
	INIT_RADIX_TREE(&root->delayed_nodes_tree, GFP_ATOMIC);
1284
	root->block_rsv = NULL;
1285
	root->orphan_block_rsv = NULL;
1286 1287

	INIT_LIST_HEAD(&root->dirty_list);
1288
	INIT_LIST_HEAD(&root->root_list);
1289 1290
	INIT_LIST_HEAD(&root->delalloc_inodes);
	INIT_LIST_HEAD(&root->delalloc_root);
1291 1292
	INIT_LIST_HEAD(&root->ordered_extents);
	INIT_LIST_HEAD(&root->ordered_root);
1293 1294
	INIT_LIST_HEAD(&root->logged_list[0]);
	INIT_LIST_HEAD(&root->logged_list[1]);
1295
	spin_lock_init(&root->orphan_lock);
1296
	spin_lock_init(&root->inode_lock);
1297
	spin_lock_init(&root->delalloc_lock);
1298
	spin_lock_init(&root->ordered_extent_lock);
1299
	spin_lock_init(&root->accounting_lock);
1300 1301
	spin_lock_init(&root->log_extents_lock[0]);
	spin_lock_init(&root->log_extents_lock[1]);
1302
	mutex_init(&root->objectid_mutex);
1303
	mutex_init(&root->log_mutex);
1304
	mutex_init(&root->ordered_extent_mutex);
1305
	mutex_init(&root->delalloc_mutex);
Y
Yan Zheng 已提交
1306 1307 1308
	init_waitqueue_head(&root->log_writer_wait);
	init_waitqueue_head(&root->log_commit_wait[0]);
	init_waitqueue_head(&root->log_commit_wait[1]);
1309 1310
	INIT_LIST_HEAD(&root->log_ctxs[0]);
	INIT_LIST_HEAD(&root->log_ctxs[1]);
Y
Yan Zheng 已提交
1311 1312 1313
	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
1314
	atomic_set(&root->log_batch, 0);
1315
	atomic_set(&root->orphan_inodes, 0);
1316
	atomic_set(&root->refs, 1);
1317
	atomic_set(&root->will_be_snapshoted, 0);
1318
	atomic_set(&root->qgroup_meta_rsv, 0);
Y
Yan Zheng 已提交
1319
	root->log_transid = 0;
1320
	root->log_transid_committed = -1;
1321
	root->last_log_commit = 0;
1322
	if (!dummy)
1323 1324
		extent_io_tree_init(&root->dirty_log_pages,
				     fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
1325

1326 1327
	memset(&root->root_key, 0, sizeof(root->root_key));
	memset(&root->root_item, 0, sizeof(root->root_item));
1328
	memset(&root->defrag_progress, 0, sizeof(root->defrag_progress));
1329
	if (!dummy)
1330 1331 1332
		root->defrag_trans_start = fs_info->generation;
	else
		root->defrag_trans_start = 0;
1333
	root->root_key.objectid = objectid;
1334
	root->anon_dev = 0;
1335

1336
	spin_lock_init(&root->root_item_lock);
1337 1338
}

1339 1340
static struct btrfs_root *btrfs_alloc_root(struct btrfs_fs_info *fs_info,
		gfp_t flags)
A
Al Viro 已提交
1341
{
1342
	struct btrfs_root *root = kzalloc(sizeof(*root), flags);
A
Al Viro 已提交
1343 1344 1345 1346 1347
	if (root)
		root->fs_info = fs_info;
	return root;
}

1348 1349
#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
/* Should only be used by the testing infrastructure */
1350 1351
struct btrfs_root *btrfs_alloc_dummy_root(struct btrfs_fs_info *fs_info,
					  u32 sectorsize, u32 nodesize)
1352 1353 1354
{
	struct btrfs_root *root;

1355 1356 1357 1358
	if (!fs_info)
		return ERR_PTR(-EINVAL);

	root = btrfs_alloc_root(fs_info, GFP_KERNEL);
1359 1360
	if (!root)
		return ERR_PTR(-ENOMEM);
1361
	/* We don't use the stripesize in selftest, set it as sectorsize */
1362
	__setup_root(nodesize, sectorsize, sectorsize, root, fs_info,
1363
			BTRFS_ROOT_TREE_OBJECTID);
1364
	root->alloc_bytenr = 0;
1365 1366 1367 1368 1369

	return root;
}
#endif

1370 1371 1372 1373 1374 1375 1376 1377 1378
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;
1379
	uuid_le uuid;
1380

1381
	root = btrfs_alloc_root(fs_info, GFP_KERNEL);
1382 1383 1384
	if (!root)
		return ERR_PTR(-ENOMEM);

1385 1386
	__setup_root(tree_root->nodesize, tree_root->sectorsize,
		tree_root->stripesize, root, fs_info, objectid);
1387 1388 1389 1390
	root->root_key.objectid = objectid;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = 0;

1391
	leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0);
1392 1393
	if (IS_ERR(leaf)) {
		ret = PTR_ERR(leaf);
1394
		leaf = NULL;
1395 1396 1397 1398 1399 1400 1401 1402 1403 1404
		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;

1405
	write_extent_buffer(leaf, fs_info->fsid, btrfs_header_fsid(),
1406 1407
			    BTRFS_FSID_SIZE);
	write_extent_buffer(leaf, fs_info->chunk_tree_uuid,
1408
			    btrfs_header_chunk_tree_uuid(leaf),
1409 1410 1411 1412
			    BTRFS_UUID_SIZE);
	btrfs_mark_buffer_dirty(leaf);

	root->commit_root = btrfs_root_node(root);
1413
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
1414 1415 1416 1417 1418 1419 1420 1421 1422 1423

	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);
1424 1425
	uuid_le_gen(&uuid);
	memcpy(root->root_item.uuid, uuid.b, BTRFS_UUID_SIZE);
1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436
	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);

1437 1438
	return root;

1439
fail:
1440 1441
	if (leaf) {
		btrfs_tree_unlock(leaf);
1442
		free_extent_buffer(root->commit_root);
1443 1444 1445
		free_extent_buffer(leaf);
	}
	kfree(root);
1446

1447
	return ERR_PTR(ret);
1448 1449
}

Y
Yan Zheng 已提交
1450 1451
static struct btrfs_root *alloc_log_tree(struct btrfs_trans_handle *trans,
					 struct btrfs_fs_info *fs_info)
1452 1453 1454
{
	struct btrfs_root *root;
	struct btrfs_root *tree_root = fs_info->tree_root;
Y
Yan Zheng 已提交
1455
	struct extent_buffer *leaf;
1456

1457
	root = btrfs_alloc_root(fs_info, GFP_NOFS);
1458
	if (!root)
Y
Yan Zheng 已提交
1459
		return ERR_PTR(-ENOMEM);
1460

1461 1462 1463
	__setup_root(tree_root->nodesize, tree_root->sectorsize,
		     tree_root->stripesize, root, fs_info,
		     BTRFS_TREE_LOG_OBJECTID);
1464 1465 1466 1467

	root->root_key.objectid = BTRFS_TREE_LOG_OBJECTID;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = BTRFS_TREE_LOG_OBJECTID;
1468

Y
Yan Zheng 已提交
1469
	/*
1470 1471
	 * DON'T set REF_COWS for log trees
	 *
Y
Yan Zheng 已提交
1472 1473 1474 1475 1476
	 * 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).
	 */
1477

1478 1479
	leaf = btrfs_alloc_tree_block(trans, root, 0, BTRFS_TREE_LOG_OBJECTID,
			NULL, 0, 0, 0);
Y
Yan Zheng 已提交
1480 1481 1482 1483
	if (IS_ERR(leaf)) {
		kfree(root);
		return ERR_CAST(leaf);
	}
1484

1485 1486 1487 1488 1489
	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
1490
	root->node = leaf;
1491 1492

	write_extent_buffer(root->node, root->fs_info->fsid,
1493
			    btrfs_header_fsid(), BTRFS_FSID_SIZE);
1494 1495
	btrfs_mark_buffer_dirty(root->node);
	btrfs_tree_unlock(root->node);
Y
Yan Zheng 已提交
1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525
	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;
1526 1527 1528
	btrfs_set_stack_inode_generation(inode_item, 1);
	btrfs_set_stack_inode_size(inode_item, 3);
	btrfs_set_stack_inode_nlink(inode_item, 1);
1529
	btrfs_set_stack_inode_nbytes(inode_item, root->nodesize);
1530
	btrfs_set_stack_inode_mode(inode_item, S_IFDIR | 0755);
Y
Yan Zheng 已提交
1531

1532
	btrfs_set_root_node(&log_root->root_item, log_root->node);
Y
Yan Zheng 已提交
1533 1534 1535 1536

	WARN_ON(root->log_root);
	root->log_root = log_root;
	root->log_transid = 0;
1537
	root->log_transid_committed = -1;
1538
	root->last_log_commit = 0;
1539 1540 1541
	return 0;
}

1542 1543
static struct btrfs_root *btrfs_read_tree_root(struct btrfs_root *tree_root,
					       struct btrfs_key *key)
1544 1545 1546
{
	struct btrfs_root *root;
	struct btrfs_fs_info *fs_info = tree_root->fs_info;
1547
	struct btrfs_path *path;
1548
	u64 generation;
1549
	int ret;
1550

1551 1552
	path = btrfs_alloc_path();
	if (!path)
1553
		return ERR_PTR(-ENOMEM);
1554

1555
	root = btrfs_alloc_root(fs_info, GFP_NOFS);
1556 1557 1558
	if (!root) {
		ret = -ENOMEM;
		goto alloc_fail;
1559 1560
	}

1561 1562
	__setup_root(tree_root->nodesize, tree_root->sectorsize,
		tree_root->stripesize, root, fs_info, key->objectid);
1563

1564 1565
	ret = btrfs_find_root(tree_root, key, path,
			      &root->root_item, &root->root_key);
1566
	if (ret) {
1567 1568
		if (ret > 0)
			ret = -ENOENT;
1569
		goto find_fail;
1570
	}
1571

1572
	generation = btrfs_root_generation(&root->root_item);
1573
	root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
1574
				     generation);
1575 1576
	if (IS_ERR(root->node)) {
		ret = PTR_ERR(root->node);
1577 1578 1579
		goto find_fail;
	} else if (!btrfs_buffer_uptodate(root->node, generation, 0)) {
		ret = -EIO;
1580 1581
		free_extent_buffer(root->node);
		goto find_fail;
1582
	}
1583
	root->commit_root = btrfs_root_node(root);
1584
out:
1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604
	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) {
1605
		set_bit(BTRFS_ROOT_REF_COWS, &root->state);
1606 1607
		btrfs_check_and_init_root_item(&root->root_item);
	}
1608

1609 1610 1611
	return root;
}

1612 1613 1614
int btrfs_init_fs_root(struct btrfs_root *root)
{
	int ret;
1615
	struct btrfs_subvolume_writers *writers;
1616 1617 1618 1619 1620 1621 1622 1623 1624

	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;
	}

1625 1626 1627 1628 1629 1630 1631
	writers = btrfs_alloc_subvolume_writers();
	if (IS_ERR(writers)) {
		ret = PTR_ERR(writers);
		goto fail;
	}
	root->subv_writers = writers;

1632
	btrfs_init_free_ino_ctl(root);
1633 1634
	spin_lock_init(&root->ino_cache_lock);
	init_waitqueue_head(&root->ino_cache_wait);
1635 1636 1637

	ret = get_anon_bdev(&root->anon_dev);
	if (ret)
L
Liu Bo 已提交
1638
		goto fail;
1639 1640 1641 1642 1643 1644

	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
1645
		goto fail;
1646 1647 1648 1649 1650 1651
	}

	ASSERT(root->highest_objectid <= BTRFS_LAST_FREE_OBJECTID);

	mutex_unlock(&root->objectid_mutex);

1652 1653
	return 0;
fail:
L
Liu Bo 已提交
1654
	/* the caller is responsible to call free_fs_root */
1655 1656 1657
	return ret;
}

1658 1659
struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info,
					u64 root_id)
1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674
{
	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;

1675
	ret = radix_tree_preload(GFP_NOFS);
1676 1677 1678 1679 1680 1681 1682 1683
	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)
1684
		set_bit(BTRFS_ROOT_IN_RADIX, &root->state);
1685 1686 1687 1688 1689 1690
	spin_unlock(&fs_info->fs_roots_radix_lock);
	radix_tree_preload_end();

	return ret;
}

1691 1692 1693
struct btrfs_root *btrfs_get_fs_root(struct btrfs_fs_info *fs_info,
				     struct btrfs_key *location,
				     bool check_ref)
1694 1695
{
	struct btrfs_root *root;
1696
	struct btrfs_path *path;
1697
	struct btrfs_key key;
1698 1699
	int ret;

1700 1701 1702 1703
	if (location->objectid == BTRFS_ROOT_TREE_OBJECTID)
		return fs_info->tree_root;
	if (location->objectid == BTRFS_EXTENT_TREE_OBJECTID)
		return fs_info->extent_root;
1704 1705 1706 1707
	if (location->objectid == BTRFS_CHUNK_TREE_OBJECTID)
		return fs_info->chunk_root;
	if (location->objectid == BTRFS_DEV_TREE_OBJECTID)
		return fs_info->dev_root;
1708 1709
	if (location->objectid == BTRFS_CSUM_TREE_OBJECTID)
		return fs_info->csum_root;
1710 1711 1712
	if (location->objectid == BTRFS_QUOTA_TREE_OBJECTID)
		return fs_info->quota_root ? fs_info->quota_root :
					     ERR_PTR(-ENOENT);
1713 1714 1715
	if (location->objectid == BTRFS_UUID_TREE_OBJECTID)
		return fs_info->uuid_root ? fs_info->uuid_root :
					    ERR_PTR(-ENOENT);
1716 1717 1718
	if (location->objectid == BTRFS_FREE_SPACE_TREE_OBJECTID)
		return fs_info->free_space_root ? fs_info->free_space_root :
						  ERR_PTR(-ENOENT);
1719
again:
1720
	root = btrfs_lookup_fs_root(fs_info, location->objectid);
1721
	if (root) {
1722
		if (check_ref && btrfs_root_refs(&root->root_item) == 0)
1723
			return ERR_PTR(-ENOENT);
1724
		return root;
1725
	}
1726

1727
	root = btrfs_read_fs_root(fs_info->tree_root, location);
1728 1729
	if (IS_ERR(root))
		return root;
1730

1731
	if (check_ref && btrfs_root_refs(&root->root_item) == 0) {
1732
		ret = -ENOENT;
1733
		goto fail;
1734
	}
1735

1736
	ret = btrfs_init_fs_root(root);
1737 1738
	if (ret)
		goto fail;
1739

1740 1741 1742 1743 1744
	path = btrfs_alloc_path();
	if (!path) {
		ret = -ENOMEM;
		goto fail;
	}
1745 1746 1747 1748 1749
	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);
1750
	btrfs_free_path(path);
1751 1752 1753
	if (ret < 0)
		goto fail;
	if (ret == 0)
1754
		set_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &root->state);
1755

1756
	ret = btrfs_insert_fs_root(fs_info, root);
1757
	if (ret) {
1758 1759 1760 1761 1762
		if (ret == -EEXIST) {
			free_fs_root(root);
			goto again;
		}
		goto fail;
1763
	}
1764
	return root;
1765 1766 1767
fail:
	free_fs_root(root);
	return ERR_PTR(ret);
1768 1769
}

C
Chris Mason 已提交
1770 1771 1772 1773 1774 1775
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
1776

1777 1778
	rcu_read_lock();
	list_for_each_entry_rcu(device, &info->fs_devices->devices, dev_list) {
1779 1780
		if (!device->bdev)
			continue;
C
Chris Mason 已提交
1781
		bdi = blk_get_backing_dev_info(device->bdev);
1782
		if (bdi_congested(bdi, bdi_bits)) {
C
Chris Mason 已提交
1783 1784 1785 1786
			ret = 1;
			break;
		}
	}
1787
	rcu_read_unlock();
C
Chris Mason 已提交
1788 1789 1790 1791 1792
	return ret;
}

static int setup_bdi(struct btrfs_fs_info *info, struct backing_dev_info *bdi)
{
1793 1794
	int err;

1795
	err = bdi_setup_and_register(bdi, "btrfs");
1796 1797 1798
	if (err)
		return err;

1799
	bdi->ra_pages = VM_MAX_READAHEAD * 1024 / PAGE_SIZE;
C
Chris Mason 已提交
1800 1801
	bdi->congested_fn	= btrfs_congested_fn;
	bdi->congested_data	= info;
1802
	bdi->capabilities |= BDI_CAP_CGROUP_WRITEBACK;
C
Chris Mason 已提交
1803 1804 1805
	return 0;
}

1806 1807 1808 1809 1810
/*
 * 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)
1811 1812
{
	struct bio *bio;
1813
	struct btrfs_end_io_wq *end_io_wq;
1814

1815
	end_io_wq = container_of(work, struct btrfs_end_io_wq, work);
1816
	bio = end_io_wq->bio;
1817

1818
	bio->bi_error = end_io_wq->error;
1819 1820
	bio->bi_private = end_io_wq->private;
	bio->bi_end_io = end_io_wq->end_io;
1821
	kmem_cache_free(btrfs_end_io_wq_cache, end_io_wq);
1822
	bio_endio(bio);
1823 1824
}

1825 1826 1827
static int cleaner_kthread(void *arg)
{
	struct btrfs_root *root = arg;
1828
	int again;
1829
	struct btrfs_trans_handle *trans;
1830 1831

	do {
1832
		again = 0;
1833

1834
		/* Make the cleaner go to sleep early. */
1835
		if (btrfs_need_cleaner_sleep(root))
1836 1837
			goto sleep;

1838 1839 1840 1841 1842 1843 1844
		/*
		 * 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;

1845 1846 1847
		if (!mutex_trylock(&root->fs_info->cleaner_mutex))
			goto sleep;

1848 1849 1850 1851
		/*
		 * Avoid the problem that we change the status of the fs
		 * during the above check and trylock.
		 */
1852
		if (btrfs_need_cleaner_sleep(root)) {
1853 1854
			mutex_unlock(&root->fs_info->cleaner_mutex);
			goto sleep;
1855
		}
1856

1857
		mutex_lock(&root->fs_info->cleaner_delayed_iput_mutex);
1858
		btrfs_run_delayed_iputs(root);
1859 1860
		mutex_unlock(&root->fs_info->cleaner_delayed_iput_mutex);

1861 1862 1863 1864
		again = btrfs_clean_one_deleted_snapshot(root);
		mutex_unlock(&root->fs_info->cleaner_mutex);

		/*
1865 1866
		 * The defragger has dealt with the R/O remount and umount,
		 * needn't do anything special here.
1867 1868
		 */
		btrfs_run_defrag_inodes(root->fs_info);
1869 1870 1871 1872 1873 1874 1875 1876 1877 1878

		/*
		 * 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);
1879
sleep:
1880
		if (!again) {
1881
			set_current_state(TASK_INTERRUPTIBLE);
1882 1883
			if (!kthread_should_stop())
				schedule();
1884 1885 1886
			__set_current_state(TASK_RUNNING);
		}
	} while (!kthread_should_stop());
1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914

	/*
	 * 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);
	}

1915 1916 1917 1918 1919 1920 1921 1922
	return 0;
}

static int transaction_kthread(void *arg)
{
	struct btrfs_root *root = arg;
	struct btrfs_trans_handle *trans;
	struct btrfs_transaction *cur;
1923
	u64 transid;
1924 1925
	unsigned long now;
	unsigned long delay;
1926
	bool cannot_commit;
1927 1928

	do {
1929
		cannot_commit = false;
1930
		delay = HZ * root->fs_info->commit_interval;
1931 1932
		mutex_lock(&root->fs_info->transaction_kthread_mutex);

J
Josef Bacik 已提交
1933
		spin_lock(&root->fs_info->trans_lock);
1934 1935
		cur = root->fs_info->running_transaction;
		if (!cur) {
J
Josef Bacik 已提交
1936
			spin_unlock(&root->fs_info->trans_lock);
1937 1938
			goto sleep;
		}
Y
Yan Zheng 已提交
1939

1940
		now = get_seconds();
1941
		if (cur->state < TRANS_STATE_BLOCKED &&
1942 1943
		    (now < cur->start_time ||
		     now - cur->start_time < root->fs_info->commit_interval)) {
J
Josef Bacik 已提交
1944
			spin_unlock(&root->fs_info->trans_lock);
1945 1946 1947
			delay = HZ * 5;
			goto sleep;
		}
1948
		transid = cur->transid;
J
Josef Bacik 已提交
1949
		spin_unlock(&root->fs_info->trans_lock);
1950

1951
		/* If the file system is aborted, this will always fail. */
1952
		trans = btrfs_attach_transaction(root);
1953
		if (IS_ERR(trans)) {
1954 1955
			if (PTR_ERR(trans) != -ENOENT)
				cannot_commit = true;
1956
			goto sleep;
1957
		}
1958
		if (transid == trans->transid) {
1959
			btrfs_commit_transaction(trans, root);
1960 1961 1962
		} else {
			btrfs_end_transaction(trans, root);
		}
1963 1964 1965 1966
sleep:
		wake_up_process(root->fs_info->cleaner_kthread);
		mutex_unlock(&root->fs_info->transaction_kthread_mutex);

J
Josef Bacik 已提交
1967 1968 1969
		if (unlikely(test_bit(BTRFS_FS_STATE_ERROR,
				      &root->fs_info->fs_state)))
			btrfs_cleanup_transaction(root);
1970 1971 1972 1973 1974 1975
		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);
1976 1977 1978 1979
	} while (!kthread_should_stop());
	return 0;
}

C
Chris Mason 已提交
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 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085
/*
 * 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));

2086 2087 2088 2089 2090 2091 2092 2093
	/*
	 * 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
2094
			       btrfs_header_generation(info->fs_root->node));
2095
		btrfs_set_backup_fs_root_level(root_backup,
C
Chris Mason 已提交
2096
			       btrfs_header_level(info->fs_root->node));
2097
	}
C
Chris Mason 已提交
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 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178

	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
2179 2180 2181
/* helper to cleanup workers */
static void btrfs_stop_all_workers(struct btrfs_fs_info *fs_info)
{
2182
	btrfs_destroy_workqueue(fs_info->fixup_workers);
2183
	btrfs_destroy_workqueue(fs_info->delalloc_workers);
2184
	btrfs_destroy_workqueue(fs_info->workers);
2185 2186 2187
	btrfs_destroy_workqueue(fs_info->endio_workers);
	btrfs_destroy_workqueue(fs_info->endio_meta_workers);
	btrfs_destroy_workqueue(fs_info->endio_raid56_workers);
2188
	btrfs_destroy_workqueue(fs_info->endio_repair_workers);
2189
	btrfs_destroy_workqueue(fs_info->rmw_workers);
2190 2191 2192
	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);
2193
	btrfs_destroy_workqueue(fs_info->submit_workers);
2194
	btrfs_destroy_workqueue(fs_info->delayed_workers);
2195
	btrfs_destroy_workqueue(fs_info->caching_workers);
2196
	btrfs_destroy_workqueue(fs_info->readahead_workers);
2197
	btrfs_destroy_workqueue(fs_info->flush_workers);
2198
	btrfs_destroy_workqueue(fs_info->qgroup_rescan_workers);
C
Chris Mason 已提交
2199
	btrfs_destroy_workqueue(fs_info->extent_workers);
L
Liu Bo 已提交
2200 2201
}

2202 2203 2204 2205 2206 2207 2208 2209 2210 2211
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
2212 2213 2214
/* helper to cleanup tree roots */
static void free_root_pointers(struct btrfs_fs_info *info, int chunk_root)
{
2215
	free_root_extent_buffers(info->tree_root);
2216

2217 2218 2219 2220 2221 2222 2223
	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);
2224
	free_root_extent_buffers(info->free_space_root);
C
Chris Mason 已提交
2225 2226
}

2227
void btrfs_free_fs_roots(struct btrfs_fs_info *fs_info)
2228 2229 2230 2231 2232 2233 2234 2235 2236 2237
{
	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);

2238
		if (test_bit(BTRFS_ROOT_IN_RADIX, &gang[0]->state)) {
2239
			btrfs_drop_and_free_fs_root(fs_info, gang[0]);
2240 2241 2242
		} else {
			free_extent_buffer(gang[0]->node);
			free_extent_buffer(gang[0]->commit_root);
2243
			btrfs_put_fs_root(gang[0]);
2244 2245 2246 2247 2248 2249 2250 2251 2252 2253
		}
	}

	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++)
2254
			btrfs_drop_and_free_fs_root(fs_info, gang[i]);
2255
	}
2256 2257 2258 2259 2260 2261

	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);
	}
2262
}
C
Chris Mason 已提交
2263

2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274
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;
}

2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285
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);
}

2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314
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);
}

2315 2316 2317 2318 2319
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);
2320 2321 2322
	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);
2323
	init_waitqueue_head(&fs_info->replace_wait);
2324
	init_waitqueue_head(&fs_info->dev_replace.read_lock_wq);
2325 2326
}

2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337
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;
2338
	fs_info->qgroup_rescan_running = false;
2339 2340 2341
	mutex_init(&fs_info->qgroup_rescan_lock);
}

2342 2343 2344 2345
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;
2346
	unsigned int flags = WQ_MEM_RECLAIM | WQ_FREEZABLE | WQ_UNBOUND;
2347 2348

	fs_info->workers =
2349 2350
		btrfs_alloc_workqueue(fs_info, "worker",
				      flags | WQ_HIGHPRI, max_active, 16);
2351 2352

	fs_info->delalloc_workers =
2353 2354
		btrfs_alloc_workqueue(fs_info, "delalloc",
				      flags, max_active, 2);
2355 2356

	fs_info->flush_workers =
2357 2358
		btrfs_alloc_workqueue(fs_info, "flush_delalloc",
				      flags, max_active, 0);
2359 2360

	fs_info->caching_workers =
2361
		btrfs_alloc_workqueue(fs_info, "cache", flags, max_active, 0);
2362 2363 2364 2365 2366 2367 2368

	/*
	 * 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 =
2369
		btrfs_alloc_workqueue(fs_info, "submit", flags,
2370 2371 2372 2373
				      min_t(u64, fs_devices->num_devices,
					    max_active), 64);

	fs_info->fixup_workers =
2374
		btrfs_alloc_workqueue(fs_info, "fixup", flags, 1, 0);
2375 2376 2377 2378 2379 2380

	/*
	 * endios are largely parallel and should have a very
	 * low idle thresh
	 */
	fs_info->endio_workers =
2381
		btrfs_alloc_workqueue(fs_info, "endio", flags, max_active, 4);
2382
	fs_info->endio_meta_workers =
2383 2384
		btrfs_alloc_workqueue(fs_info, "endio-meta", flags,
				      max_active, 4);
2385
	fs_info->endio_meta_write_workers =
2386 2387
		btrfs_alloc_workqueue(fs_info, "endio-meta-write", flags,
				      max_active, 2);
2388
	fs_info->endio_raid56_workers =
2389 2390
		btrfs_alloc_workqueue(fs_info, "endio-raid56", flags,
				      max_active, 4);
2391
	fs_info->endio_repair_workers =
2392
		btrfs_alloc_workqueue(fs_info, "endio-repair", flags, 1, 0);
2393
	fs_info->rmw_workers =
2394
		btrfs_alloc_workqueue(fs_info, "rmw", flags, max_active, 2);
2395
	fs_info->endio_write_workers =
2396 2397
		btrfs_alloc_workqueue(fs_info, "endio-write", flags,
				      max_active, 2);
2398
	fs_info->endio_freespace_worker =
2399 2400
		btrfs_alloc_workqueue(fs_info, "freespace-write", flags,
				      max_active, 0);
2401
	fs_info->delayed_workers =
2402 2403
		btrfs_alloc_workqueue(fs_info, "delayed-meta", flags,
				      max_active, 0);
2404
	fs_info->readahead_workers =
2405 2406
		btrfs_alloc_workqueue(fs_info, "readahead", flags,
				      max_active, 2);
2407
	fs_info->qgroup_rescan_workers =
2408
		btrfs_alloc_workqueue(fs_info, "qgroup-rescan", flags, 1, 0);
2409
	fs_info->extent_workers =
2410
		btrfs_alloc_workqueue(fs_info, "extent-refs", flags,
2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430
				      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;
}

2431 2432 2433 2434 2435 2436 2437 2438 2439 2440
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) {
2441
		btrfs_warn(fs_info, "log replay required on RO media");
2442 2443 2444
		return -EIO;
	}

2445
	log_tree_root = btrfs_alloc_root(fs_info, GFP_KERNEL);
2446 2447 2448 2449 2450 2451 2452 2453 2454
	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);
2455
	if (IS_ERR(log_tree_root->node)) {
2456
		btrfs_warn(fs_info, "failed to read log tree");
2457
		ret = PTR_ERR(log_tree_root->node);
2458
		kfree(log_tree_root);
2459
		return ret;
2460
	} else if (!extent_buffer_uptodate(log_tree_root->node)) {
2461
		btrfs_err(fs_info, "failed to read log tree");
2462 2463 2464 2465 2466 2467 2468
		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) {
2469
		btrfs_handle_fs_error(tree_root->fs_info, ret,
2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484
			    "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;
}

2485 2486 2487
static int btrfs_read_roots(struct btrfs_fs_info *fs_info,
			    struct btrfs_root *tree_root)
{
2488
	struct btrfs_root *root;
2489 2490 2491 2492 2493 2494 2495
	struct btrfs_key location;
	int ret;

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

2496 2497 2498 2499 2500
	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;
2501 2502

	location.objectid = BTRFS_DEV_TREE_OBJECTID;
2503 2504 2505 2506 2507
	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;
2508 2509 2510
	btrfs_init_devices_late(fs_info);

	location.objectid = BTRFS_CSUM_TREE_OBJECTID;
2511 2512 2513 2514 2515
	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;
2516 2517

	location.objectid = BTRFS_QUOTA_TREE_OBJECTID;
2518 2519 2520
	root = btrfs_read_tree_root(tree_root, &location);
	if (!IS_ERR(root)) {
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
2521 2522
		fs_info->quota_enabled = 1;
		fs_info->pending_quota_state = 1;
2523
		fs_info->quota_root = root;
2524 2525 2526
	}

	location.objectid = BTRFS_UUID_TREE_OBJECTID;
2527 2528 2529
	root = btrfs_read_tree_root(tree_root, &location);
	if (IS_ERR(root)) {
		ret = PTR_ERR(root);
2530 2531 2532
		if (ret != -ENOENT)
			return ret;
	} else {
2533 2534
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
		fs_info->uuid_root = root;
2535 2536
	}

2537 2538 2539 2540 2541 2542 2543 2544 2545
	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;
	}

2546 2547 2548
	return 0;
}

A
Al Viro 已提交
2549 2550 2551
int open_ctree(struct super_block *sb,
	       struct btrfs_fs_devices *fs_devices,
	       char *options)
2552
{
2553 2554
	u32 sectorsize;
	u32 nodesize;
2555
	u32 stripesize;
2556
	u64 generation;
2557
	u64 features;
2558
	struct btrfs_key location;
2559
	struct buffer_head *bh;
2560
	struct btrfs_super_block *disk_super;
2561
	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
2562
	struct btrfs_root *tree_root;
2563
	struct btrfs_root *chunk_root;
2564
	int ret;
2565
	int err = -EINVAL;
C
Chris Mason 已提交
2566 2567
	int num_backups_tried = 0;
	int backup_index = 0;
2568
	int max_active;
2569

2570 2571
	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);
2572
	if (!tree_root || !chunk_root) {
C
Chris Mason 已提交
2573 2574 2575
		err = -ENOMEM;
		goto fail;
	}
2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588

	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;
	}

2589
	ret = percpu_counter_init(&fs_info->dirty_metadata_bytes, 0, GFP_KERNEL);
2590 2591 2592 2593
	if (ret) {
		err = ret;
		goto fail_bdi;
	}
2594
	fs_info->dirty_metadata_batch = PAGE_SIZE *
2595 2596
					(1 + ilog2(nr_cpu_ids));

2597
	ret = percpu_counter_init(&fs_info->delalloc_bytes, 0, GFP_KERNEL);
2598 2599 2600 2601 2602
	if (ret) {
		err = ret;
		goto fail_dirty_metadata_bytes;
	}

2603
	ret = percpu_counter_init(&fs_info->bio_counter, 0, GFP_KERNEL);
2604 2605 2606 2607 2608
	if (ret) {
		err = ret;
		goto fail_delalloc_bytes;
	}

2609 2610 2611
	fs_info->btree_inode = new_inode(sb);
	if (!fs_info->btree_inode) {
		err = -ENOMEM;
2612
		goto fail_bio_counter;
2613 2614
	}

2615
	mapping_set_gfp_mask(fs_info->btree_inode->i_mapping, GFP_NOFS);
2616

2617
	INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC);
2618
	INIT_RADIX_TREE(&fs_info->buffer_radix, GFP_ATOMIC);
C
Chris Mason 已提交
2619
	INIT_LIST_HEAD(&fs_info->trans_list);
2620
	INIT_LIST_HEAD(&fs_info->dead_roots);
Y
Yan, Zheng 已提交
2621
	INIT_LIST_HEAD(&fs_info->delayed_iputs);
2622
	INIT_LIST_HEAD(&fs_info->delalloc_roots);
2623
	INIT_LIST_HEAD(&fs_info->caching_block_groups);
2624
	spin_lock_init(&fs_info->delalloc_root_lock);
J
Josef Bacik 已提交
2625
	spin_lock_init(&fs_info->trans_lock);
2626
	spin_lock_init(&fs_info->fs_roots_radix_lock);
Y
Yan, Zheng 已提交
2627
	spin_lock_init(&fs_info->delayed_iput_lock);
C
Chris Mason 已提交
2628
	spin_lock_init(&fs_info->defrag_inodes_lock);
2629
	spin_lock_init(&fs_info->free_chunk_lock);
J
Jan Schmidt 已提交
2630
	spin_lock_init(&fs_info->tree_mod_seq_lock);
2631
	spin_lock_init(&fs_info->super_lock);
J
Josef Bacik 已提交
2632
	spin_lock_init(&fs_info->qgroup_op_lock);
2633
	spin_lock_init(&fs_info->buffer_lock);
2634
	spin_lock_init(&fs_info->unused_bgs_lock);
J
Jan Schmidt 已提交
2635
	rwlock_init(&fs_info->tree_mod_log_lock);
2636
	mutex_init(&fs_info->unused_bg_unpin_mutex);
2637
	mutex_init(&fs_info->delete_unused_bgs_mutex);
C
Chris Mason 已提交
2638
	mutex_init(&fs_info->reloc_mutex);
2639
	mutex_init(&fs_info->delalloc_root_mutex);
2640
	mutex_init(&fs_info->cleaner_delayed_iput_mutex);
2641
	seqlock_init(&fs_info->profiles_lock);
2642

2643
	INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots);
2644
	INIT_LIST_HEAD(&fs_info->space_info);
J
Jan Schmidt 已提交
2645
	INIT_LIST_HEAD(&fs_info->tree_mod_seq_list);
2646
	INIT_LIST_HEAD(&fs_info->unused_bgs);
2647
	btrfs_mapping_init(&fs_info->mapping_tree);
2648 2649 2650 2651 2652 2653 2654 2655 2656
	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);
2657
	atomic_set(&fs_info->nr_async_submits, 0);
2658
	atomic_set(&fs_info->async_delalloc_pages, 0);
2659
	atomic_set(&fs_info->async_submit_draining, 0);
2660
	atomic_set(&fs_info->nr_async_bios, 0);
C
Chris Mason 已提交
2661
	atomic_set(&fs_info->defrag_running, 0);
J
Josef Bacik 已提交
2662
	atomic_set(&fs_info->qgroup_op_seq, 0);
Z
Zhao Lei 已提交
2663
	atomic_set(&fs_info->reada_works_cnt, 0);
2664
	atomic64_set(&fs_info->tree_mod_seq, 0);
2665
	fs_info->fs_frozen = 0;
C
Chris Mason 已提交
2666
	fs_info->sb = sb;
2667
	fs_info->max_inline = BTRFS_DEFAULT_MAX_INLINE;
J
Josef Bacik 已提交
2668
	fs_info->metadata_ratio = 0;
C
Chris Mason 已提交
2669
	fs_info->defrag_inodes = RB_ROOT;
2670
	fs_info->free_chunk_space = 0;
J
Jan Schmidt 已提交
2671
	fs_info->tree_mod_log = RB_ROOT;
2672
	fs_info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
2673
	fs_info->avg_delayed_ref_runtime = NSEC_PER_SEC >> 6; /* div by 64 */
2674
	/* readahead state */
2675
	INIT_RADIX_TREE(&fs_info->reada_tree, GFP_NOFS & ~__GFP_DIRECT_RECLAIM);
2676
	spin_lock_init(&fs_info->reada_lock);
C
Chris Mason 已提交
2677

2678 2679
	fs_info->thread_pool_size = min_t(unsigned long,
					  num_online_cpus() + 2, 8);
2680

2681 2682
	INIT_LIST_HEAD(&fs_info->ordered_roots);
	spin_lock_init(&fs_info->ordered_root_lock);
2683
	fs_info->delayed_root = kmalloc(sizeof(struct btrfs_delayed_root),
2684
					GFP_KERNEL);
2685 2686 2687 2688 2689
	if (!fs_info->delayed_root) {
		err = -ENOMEM;
		goto fail_iput;
	}
	btrfs_init_delayed_root(fs_info->delayed_root);
2690

2691
	btrfs_init_scrub(fs_info);
2692 2693 2694
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	fs_info->check_integrity_print_mask = 0;
#endif
2695
	btrfs_init_balance(fs_info);
2696
	btrfs_init_async_reclaim_work(&fs_info->async_reclaim_work);
A
Arne Jansen 已提交
2697

2698 2699
	sb->s_blocksize = 4096;
	sb->s_blocksize_bits = blksize_bits(4096);
J
Jens Axboe 已提交
2700
	sb->s_bdi = &fs_info->bdi;
2701

2702
	btrfs_init_btree_inode(fs_info, tree_root);
2703

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

2708
	extent_io_tree_init(&fs_info->freed_extents[0],
2709
			     fs_info->btree_inode->i_mapping);
2710
	extent_io_tree_init(&fs_info->freed_extents[1],
2711
			     fs_info->btree_inode->i_mapping);
2712
	fs_info->pinned_extents = &fs_info->freed_extents[0];
2713
	fs_info->do_barriers = 1;
2714

C
Chris Mason 已提交
2715

2716
	mutex_init(&fs_info->ordered_operations_mutex);
2717
	mutex_init(&fs_info->tree_log_mutex);
2718
	mutex_init(&fs_info->chunk_mutex);
2719 2720
	mutex_init(&fs_info->transaction_kthread_mutex);
	mutex_init(&fs_info->cleaner_mutex);
2721
	mutex_init(&fs_info->volume_mutex);
2722
	mutex_init(&fs_info->ro_block_group_mutex);
2723
	init_rwsem(&fs_info->commit_root_sem);
2724
	init_rwsem(&fs_info->cleanup_work_sem);
2725
	init_rwsem(&fs_info->subvol_sem);
S
Stefan Behrens 已提交
2726
	sema_init(&fs_info->uuid_tree_rescan_sem, 1);
2727

2728
	btrfs_init_dev_replace_locks(fs_info);
2729
	btrfs_init_qgroup(fs_info);
2730

2731 2732 2733
	btrfs_init_free_cluster(&fs_info->meta_alloc_cluster);
	btrfs_init_free_cluster(&fs_info->data_alloc_cluster);

2734
	init_waitqueue_head(&fs_info->transaction_throttle);
2735
	init_waitqueue_head(&fs_info->transaction_wait);
S
Sage Weil 已提交
2736
	init_waitqueue_head(&fs_info->transaction_blocked_wait);
2737
	init_waitqueue_head(&fs_info->async_submit_wait);
2738

2739 2740
	INIT_LIST_HEAD(&fs_info->pinned_chunks);

D
David Woodhouse 已提交
2741 2742
	ret = btrfs_alloc_stripe_hash_table(fs_info);
	if (ret) {
2743
		err = ret;
D
David Woodhouse 已提交
2744 2745 2746
		goto fail_alloc;
	}

2747
	__setup_root(4096, 4096, 4096, tree_root,
C
Chris Mason 已提交
2748
		     fs_info, BTRFS_ROOT_TREE_OBJECTID);
2749

2750
	invalidate_bdev(fs_devices->latest_bdev);
D
David Sterba 已提交
2751 2752 2753 2754

	/*
	 * Read super block and check the signature bytes only
	 */
Y
Yan Zheng 已提交
2755
	bh = btrfs_read_dev_super(fs_devices->latest_bdev);
2756 2757
	if (IS_ERR(bh)) {
		err = PTR_ERR(bh);
2758
		goto fail_alloc;
2759
	}
C
Chris Mason 已提交
2760

D
David Sterba 已提交
2761 2762 2763 2764 2765
	/*
	 * We want to check superblock checksum, the type is stored inside.
	 * Pass the whole disk block of size BTRFS_SUPER_INFO_SIZE (4k).
	 */
	if (btrfs_check_super_csum(bh->b_data)) {
2766
		btrfs_err(fs_info, "superblock checksum mismatch");
D
David Sterba 已提交
2767
		err = -EINVAL;
2768
		brelse(bh);
D
David Sterba 已提交
2769 2770 2771 2772 2773 2774 2775 2776
		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
	 */
2777 2778 2779
	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));
2780
	brelse(bh);
2781

2782
	memcpy(fs_info->fsid, fs_info->super_copy->fsid, BTRFS_FSID_SIZE);
2783

D
David Sterba 已提交
2784 2785
	ret = btrfs_check_super_valid(fs_info, sb->s_flags & MS_RDONLY);
	if (ret) {
2786
		btrfs_err(fs_info, "superblock contains fatal errors");
D
David Sterba 已提交
2787 2788 2789 2790
		err = -EINVAL;
		goto fail_alloc;
	}

2791
	disk_super = fs_info->super_copy;
2792
	if (!btrfs_super_root(disk_super))
2793
		goto fail_alloc;
2794

L
liubo 已提交
2795
	/* check FS state, whether FS is broken. */
2796 2797
	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
2798

C
Chris Mason 已提交
2799 2800 2801 2802 2803 2804 2805
	/*
	 * 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);

2806 2807 2808 2809 2810 2811
	/*
	 * 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;

2812
	ret = btrfs_parse_options(tree_root, options, sb->s_flags);
Y
Yan Zheng 已提交
2813 2814
	if (ret) {
		err = ret;
2815
		goto fail_alloc;
Y
Yan Zheng 已提交
2816
	}
2817

2818 2819 2820
	features = btrfs_super_incompat_flags(disk_super) &
		~BTRFS_FEATURE_INCOMPAT_SUPP;
	if (features) {
2821 2822 2823
		btrfs_err(fs_info,
		    "cannot mount because of unsupported optional features (%llx)",
		    features);
2824
		err = -EINVAL;
2825
		goto fail_alloc;
2826 2827
	}

2828
	features = btrfs_super_incompat_flags(disk_super);
L
Li Zefan 已提交
2829
	features |= BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF;
L
Li Zefan 已提交
2830
	if (tree_root->fs_info->compress_type == BTRFS_COMPRESS_LZO)
L
Li Zefan 已提交
2831
		features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
2832

2833
	if (features & BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA)
2834
		btrfs_info(fs_info, "has skinny extents");
2835

2836 2837 2838 2839
	/*
	 * flag our filesystem as having big metadata blocks if
	 * they are bigger than the page size
	 */
2840
	if (btrfs_super_nodesize(disk_super) > PAGE_SIZE) {
2841
		if (!(features & BTRFS_FEATURE_INCOMPAT_BIG_METADATA))
2842 2843
			btrfs_info(fs_info,
				"flagging fs with big metadata feature");
2844 2845 2846
		features |= BTRFS_FEATURE_INCOMPAT_BIG_METADATA;
	}

2847 2848
	nodesize = btrfs_super_nodesize(disk_super);
	sectorsize = btrfs_super_sectorsize(disk_super);
2849
	stripesize = sectorsize;
2850
	fs_info->dirty_metadata_batch = nodesize * (1 + ilog2(nr_cpu_ids));
2851
	fs_info->delalloc_batch = sectorsize * 512 * (1 + ilog2(nr_cpu_ids));
2852 2853 2854 2855 2856 2857

	/*
	 * 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) &&
2858
	    (sectorsize != nodesize)) {
2859 2860 2861
		btrfs_err(fs_info,
"unequal nodesize/sectorsize (%u != %u) are not allowed for mixed block groups",
			nodesize, sectorsize);
2862 2863 2864
		goto fail_alloc;
	}

2865 2866 2867 2868
	/*
	 * Needn't use the lock because there is no other task which will
	 * update the flag.
	 */
L
Li Zefan 已提交
2869
	btrfs_set_super_incompat_flags(disk_super, features);
2870

2871 2872 2873
	features = btrfs_super_compat_ro_flags(disk_super) &
		~BTRFS_FEATURE_COMPAT_RO_SUPP;
	if (!(sb->s_flags & MS_RDONLY) && features) {
2874 2875
		btrfs_err(fs_info,
	"cannot mount read-write because of unsupported optional features (%llx)",
2876
		       features);
2877
		err = -EINVAL;
2878
		goto fail_alloc;
2879
	}
2880

2881
	max_active = fs_info->thread_pool_size;
2882

2883 2884 2885
	ret = btrfs_init_workqueues(fs_info, fs_devices);
	if (ret) {
		err = ret;
2886 2887
		goto fail_sb_buffer;
	}
2888

2889
	fs_info->bdi.ra_pages *= btrfs_super_num_devices(disk_super);
C
Chris Mason 已提交
2890
	fs_info->bdi.ra_pages = max(fs_info->bdi.ra_pages,
2891
				    SZ_4M / PAGE_SIZE);
2892

2893 2894
	tree_root->nodesize = nodesize;
	tree_root->sectorsize = sectorsize;
2895
	tree_root->stripesize = stripesize;
2896 2897 2898

	sb->s_blocksize = sectorsize;
	sb->s_blocksize_bits = blksize_bits(sectorsize);
2899

2900
	mutex_lock(&fs_info->chunk_mutex);
Y
Yan Zheng 已提交
2901
	ret = btrfs_read_sys_array(tree_root);
2902
	mutex_unlock(&fs_info->chunk_mutex);
2903
	if (ret) {
2904
		btrfs_err(fs_info, "failed to read the system array: %d", ret);
2905
		goto fail_sb_buffer;
2906
	}
2907

2908
	generation = btrfs_super_chunk_root_generation(disk_super);
2909

2910 2911
	__setup_root(nodesize, sectorsize, stripesize, chunk_root,
		     fs_info, BTRFS_CHUNK_TREE_OBJECTID);
2912 2913 2914

	chunk_root->node = read_tree_block(chunk_root,
					   btrfs_super_chunk_root(disk_super),
2915
					   generation);
2916 2917
	if (IS_ERR(chunk_root->node) ||
	    !extent_buffer_uptodate(chunk_root->node)) {
2918
		btrfs_err(fs_info, "failed to read chunk root");
2919 2920
		if (!IS_ERR(chunk_root->node))
			free_extent_buffer(chunk_root->node);
2921
		chunk_root->node = NULL;
C
Chris Mason 已提交
2922
		goto fail_tree_roots;
2923
	}
2924 2925
	btrfs_set_root_node(&chunk_root->root_item, chunk_root->node);
	chunk_root->commit_root = btrfs_root_node(chunk_root);
2926

2927
	read_extent_buffer(chunk_root->node, fs_info->chunk_tree_uuid,
2928
	   btrfs_header_chunk_tree_uuid(chunk_root->node), BTRFS_UUID_SIZE);
2929

2930
	ret = btrfs_read_chunk_tree(chunk_root);
Y
Yan Zheng 已提交
2931
	if (ret) {
2932
		btrfs_err(fs_info, "failed to read chunk tree: %d", ret);
C
Chris Mason 已提交
2933
		goto fail_tree_roots;
Y
Yan Zheng 已提交
2934
	}
2935

2936 2937 2938 2939
	/*
	 * keep the device that is marked to be the target device for the
	 * dev_replace procedure
	 */
2940
	btrfs_close_extra_devices(fs_devices, 0);
2941

2942
	if (!fs_devices->latest_bdev) {
2943
		btrfs_err(fs_info, "failed to read devices");
2944 2945 2946
		goto fail_tree_roots;
	}

C
Chris Mason 已提交
2947
retry_root_backup:
2948
	generation = btrfs_super_generation(disk_super);
2949

C
Chris Mason 已提交
2950
	tree_root->node = read_tree_block(tree_root,
2951
					  btrfs_super_root(disk_super),
2952
					  generation);
2953 2954
	if (IS_ERR(tree_root->node) ||
	    !extent_buffer_uptodate(tree_root->node)) {
2955
		btrfs_warn(fs_info, "failed to read tree root");
2956 2957
		if (!IS_ERR(tree_root->node))
			free_extent_buffer(tree_root->node);
2958
		tree_root->node = NULL;
C
Chris Mason 已提交
2959
		goto recovery_tree_root;
2960
	}
C
Chris Mason 已提交
2961

2962 2963
	btrfs_set_root_node(&tree_root->root_item, tree_root->node);
	tree_root->commit_root = btrfs_root_node(tree_root);
2964
	btrfs_set_root_refs(&tree_root->root_item, 1);
2965

2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977
	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);

2978 2979
	ret = btrfs_read_roots(fs_info, tree_root);
	if (ret)
C
Chris Mason 已提交
2980
		goto recovery_tree_root;
2981

2982 2983 2984
	fs_info->generation = generation;
	fs_info->last_trans_committed = generation;

2985 2986
	ret = btrfs_recover_balance(fs_info);
	if (ret) {
2987
		btrfs_err(fs_info, "failed to recover balance: %d", ret);
2988 2989 2990
		goto fail_block_groups;
	}

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

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

3003
	btrfs_close_extra_devices(fs_devices, 1);
3004

3005 3006
	ret = btrfs_sysfs_add_fsid(fs_devices, NULL);
	if (ret) {
3007 3008
		btrfs_err(fs_info, "failed to init sysfs fsid interface: %d",
				ret);
3009 3010 3011 3012 3013
		goto fail_block_groups;
	}

	ret = btrfs_sysfs_add_device(fs_devices);
	if (ret) {
3014 3015
		btrfs_err(fs_info, "failed to init sysfs device interface: %d",
				ret);
3016 3017 3018
		goto fail_fsdev_sysfs;
	}

3019
	ret = btrfs_sysfs_add_mounted(fs_info);
3020
	if (ret) {
3021
		btrfs_err(fs_info, "failed to init sysfs interface: %d", ret);
3022
		goto fail_fsdev_sysfs;
3023 3024 3025 3026
	}

	ret = btrfs_init_space_info(fs_info);
	if (ret) {
3027
		btrfs_err(fs_info, "failed to initialize space info: %d", ret);
3028
		goto fail_sysfs;
3029 3030
	}

3031
	ret = btrfs_read_block_groups(fs_info->extent_root);
3032
	if (ret) {
3033
		btrfs_err(fs_info, "failed to read block groups: %d", ret);
3034
		goto fail_sysfs;
3035
	}
3036 3037
	fs_info->num_tolerated_disk_barrier_failures =
		btrfs_calc_num_tolerated_disk_barrier_failures(fs_info);
3038 3039 3040
	if (fs_info->fs_devices->missing_devices >
	     fs_info->num_tolerated_disk_barrier_failures &&
	    !(sb->s_flags & MS_RDONLY)) {
3041 3042
		btrfs_warn(fs_info,
"missing devices (%llu) exceeds the limit (%d), writeable mount is not allowed",
3043 3044
			fs_info->fs_devices->missing_devices,
			fs_info->num_tolerated_disk_barrier_failures);
3045
		goto fail_sysfs;
3046
	}
C
Chris Mason 已提交
3047

3048 3049
	fs_info->cleaner_kthread = kthread_run(cleaner_kthread, tree_root,
					       "btrfs-cleaner");
3050
	if (IS_ERR(fs_info->cleaner_kthread))
3051
		goto fail_sysfs;
3052 3053 3054 3055

	fs_info->transaction_kthread = kthread_run(transaction_kthread,
						   tree_root,
						   "btrfs-transaction");
3056
	if (IS_ERR(fs_info->transaction_kthread))
3057
		goto fail_cleaner;
3058

3059 3060
	if (!btrfs_test_opt(tree_root->fs_info, SSD) &&
	    !btrfs_test_opt(tree_root->fs_info, NOSSD) &&
C
Chris Mason 已提交
3061
	    !fs_info->fs_devices->rotating) {
3062
		btrfs_info(fs_info, "detected SSD devices, enabling SSD mode");
C
Chris Mason 已提交
3063 3064 3065
		btrfs_set_opt(fs_info->mount_opt, SSD);
	}

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

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

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

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

3103
	if (!(sb->s_flags & MS_RDONLY)) {
3104
		ret = btrfs_cleanup_fs_roots(fs_info);
3105
		if (ret)
3106
			goto fail_qgroup;
3107 3108

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

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

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

3129 3130
	if (sb->s_flags & MS_RDONLY)
		return 0;
I
Ilya Dryomov 已提交
3131

3132
	if (btrfs_test_opt(tree_root->fs_info, FREE_SPACE_TREE) &&
3133
	    !btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
3134
		btrfs_info(fs_info, "creating free space tree");
3135 3136
		ret = btrfs_create_free_space_tree(fs_info);
		if (ret) {
3137 3138
			btrfs_warn(fs_info,
				"failed to create free space tree: %d", ret);
3139 3140 3141 3142 3143
			close_ctree(tree_root);
			return ret;
		}
	}

3144 3145 3146
	down_read(&fs_info->cleanup_work_sem);
	if ((ret = btrfs_orphan_cleanup(fs_info->fs_root)) ||
	    (ret = btrfs_orphan_cleanup(fs_info->tree_root))) {
3147
		up_read(&fs_info->cleanup_work_sem);
3148 3149 3150 3151
		close_ctree(tree_root);
		return ret;
	}
	up_read(&fs_info->cleanup_work_sem);
I
Ilya Dryomov 已提交
3152

3153 3154
	ret = btrfs_resume_balance_async(fs_info);
	if (ret) {
3155
		btrfs_warn(fs_info, "failed to resume balance: %d", ret);
3156 3157
		close_ctree(tree_root);
		return ret;
3158 3159
	}

3160 3161
	ret = btrfs_resume_dev_replace_async(fs_info);
	if (ret) {
3162
		btrfs_warn(fs_info, "failed to resume device replace: %d", ret);
3163 3164 3165 3166
		close_ctree(tree_root);
		return ret;
	}

3167 3168
	btrfs_qgroup_rescan_resume(fs_info);

3169
	if (btrfs_test_opt(tree_root->fs_info, CLEAR_CACHE) &&
3170
	    btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
3171
		btrfs_info(fs_info, "clearing free space tree");
3172 3173
		ret = btrfs_clear_free_space_tree(fs_info);
		if (ret) {
3174 3175
			btrfs_warn(fs_info,
				"failed to clear free space tree: %d", ret);
3176 3177 3178 3179 3180
			close_ctree(tree_root);
			return ret;
		}
	}

3181
	if (!fs_info->uuid_root) {
3182
		btrfs_info(fs_info, "creating UUID tree");
3183 3184
		ret = btrfs_create_uuid_tree(fs_info);
		if (ret) {
3185 3186
			btrfs_warn(fs_info,
				"failed to create the UUID tree: %d", ret);
3187 3188 3189
			close_ctree(tree_root);
			return ret;
		}
3190
	} else if (btrfs_test_opt(tree_root->fs_info, RESCAN_UUID_TREE) ||
3191 3192
		   fs_info->generation !=
				btrfs_super_uuid_tree_generation(disk_super)) {
3193
		btrfs_info(fs_info, "checking UUID tree");
3194 3195
		ret = btrfs_check_uuid_tree(fs_info);
		if (ret) {
3196 3197
			btrfs_warn(fs_info,
				"failed to check the UUID tree: %d", ret);
3198 3199 3200 3201 3202
			close_ctree(tree_root);
			return ret;
		}
	} else {
		fs_info->update_uuid_tree_gen = 1;
3203 3204
	}

3205 3206
	fs_info->open = 1;

3207 3208 3209 3210 3211 3212
	/*
	 * 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
3213
	return 0;
C
Chris Mason 已提交
3214

3215 3216
fail_qgroup:
	btrfs_free_qgroup_config(fs_info);
3217 3218
fail_trans_kthread:
	kthread_stop(fs_info->transaction_kthread);
J
Josef Bacik 已提交
3219
	btrfs_cleanup_transaction(fs_info->tree_root);
3220
	btrfs_free_fs_roots(fs_info);
3221
fail_cleaner:
3222
	kthread_stop(fs_info->cleaner_kthread);
3223 3224 3225 3226 3227 3228 3229

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

3230
fail_sysfs:
3231
	btrfs_sysfs_remove_mounted(fs_info);
3232

3233 3234 3235
fail_fsdev_sysfs:
	btrfs_sysfs_remove_fsid(fs_info->fs_devices);

3236
fail_block_groups:
J
Josef Bacik 已提交
3237
	btrfs_put_block_group_cache(fs_info);
3238
	btrfs_free_block_groups(fs_info);
C
Chris Mason 已提交
3239 3240 3241

fail_tree_roots:
	free_root_pointers(fs_info, 1);
3242
	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
3243

C
Chris Mason 已提交
3244
fail_sb_buffer:
L
Liu Bo 已提交
3245
	btrfs_stop_all_workers(fs_info);
3246
fail_alloc:
3247
fail_iput:
3248 3249
	btrfs_mapping_tree_free(&fs_info->mapping_tree);

3250
	iput(fs_info->btree_inode);
3251 3252
fail_bio_counter:
	percpu_counter_destroy(&fs_info->bio_counter);
3253 3254
fail_delalloc_bytes:
	percpu_counter_destroy(&fs_info->delalloc_bytes);
3255 3256
fail_dirty_metadata_bytes:
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
3257
fail_bdi:
3258
	bdi_destroy(&fs_info->bdi);
3259 3260
fail_srcu:
	cleanup_srcu_struct(&fs_info->subvol_srcu);
3261
fail:
D
David Woodhouse 已提交
3262
	btrfs_free_stripe_hash_table(fs_info);
3263
	btrfs_close_devices(fs_info->fs_devices);
A
Al Viro 已提交
3264
	return err;
C
Chris Mason 已提交
3265 3266

recovery_tree_root:
3267
	if (!btrfs_test_opt(tree_root->fs_info, USEBACKUPROOT))
C
Chris Mason 已提交
3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282
		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;
3283 3284
}

3285 3286 3287 3288 3289
static void btrfs_end_buffer_write_sync(struct buffer_head *bh, int uptodate)
{
	if (uptodate) {
		set_buffer_uptodate(bh);
	} else {
3290 3291 3292
		struct btrfs_device *device = (struct btrfs_device *)
			bh->b_private;

3293 3294
		btrfs_warn_rl_in_rcu(device->dev_root->fs_info,
				"lost page write due to IO error on %s",
3295
					  rcu_str_deref(device->name));
3296
		/* note, we don't set_buffer_write_io_error because we have
3297 3298
		 * our own ways of dealing with the IO errors
		 */
3299
		clear_buffer_uptodate(bh);
3300
		btrfs_dev_stat_inc_and_print(device, BTRFS_DEV_STAT_WRITE_ERRS);
3301 3302 3303 3304 3305
	}
	unlock_buffer(bh);
	put_bh(bh);
}

3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336
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
3337 3338 3339 3340 3341 3342 3343
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;
3344
	int ret = -EINVAL;
Y
Yan Zheng 已提交
3345 3346 3347 3348 3349 3350 3351

	/* 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++) {
3352 3353
		ret = btrfs_read_dev_one_super(bdev, i, &bh);
		if (ret)
Y
Yan Zheng 已提交
3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365
			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);
		}
	}
3366 3367 3368 3369

	if (!latest)
		return ERR_PTR(ret);

Y
Yan Zheng 已提交
3370 3371 3372
	return latest;
}

3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383
/*
 * 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
3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399
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);
3400 3401
		if (bytenr + BTRFS_SUPER_INFO_SIZE >=
		    device->commit_total_bytes)
Y
Yan Zheng 已提交
3402 3403 3404 3405 3406
			break;

		if (wait) {
			bh = __find_get_block(device->bdev, bytenr / 4096,
					      BTRFS_SUPER_INFO_SIZE);
3407 3408 3409 3410
			if (!bh) {
				errors++;
				continue;
			}
Y
Yan Zheng 已提交
3411
			wait_on_buffer(bh);
3412 3413 3414 3415 3416 3417 3418 3419 3420
			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
3421 3422 3423 3424
		} else {
			btrfs_set_super_bytenr(sb, bytenr);

			crc = ~(u32)0;
3425
			crc = btrfs_csum_data((char *)sb +
Y
Yan Zheng 已提交
3426 3427 3428 3429 3430
					      BTRFS_CSUM_SIZE, crc,
					      BTRFS_SUPER_INFO_SIZE -
					      BTRFS_CSUM_SIZE);
			btrfs_csum_final(crc, sb->csum);

3431 3432 3433 3434
			/*
			 * one reference for us, and we leave it for the
			 * caller
			 */
Y
Yan Zheng 已提交
3435 3436
			bh = __getblk(device->bdev, bytenr / 4096,
				      BTRFS_SUPER_INFO_SIZE);
3437
			if (!bh) {
3438 3439 3440
				btrfs_err(device->dev_root->fs_info,
				    "couldn't get super buffer head for bytenr %llu",
				    bytenr);
3441 3442 3443 3444
				errors++;
				continue;
			}

Y
Yan Zheng 已提交
3445 3446
			memcpy(bh->b_data, sb, BTRFS_SUPER_INFO_SIZE);

3447
			/* one reference for submit_bh */
Y
Yan Zheng 已提交
3448
			get_bh(bh);
3449 3450

			set_buffer_uptodate(bh);
Y
Yan Zheng 已提交
3451 3452
			lock_buffer(bh);
			bh->b_end_io = btrfs_end_buffer_write_sync;
3453
			bh->b_private = device;
Y
Yan Zheng 已提交
3454 3455
		}

C
Chris Mason 已提交
3456 3457 3458 3459
		/*
		 * we fua the first super.  The others we allow
		 * to go down lazy.
		 */
3460
		if (i == 0)
3461
			ret = btrfsic_submit_bh(REQ_OP_WRITE, WRITE_FUA, bh);
3462
		else
3463
			ret = btrfsic_submit_bh(REQ_OP_WRITE, WRITE_SYNC, bh);
3464
		if (ret)
Y
Yan Zheng 已提交
3465 3466 3467 3468 3469
			errors++;
	}
	return errors < i ? 0 : -1;
}

C
Chris Mason 已提交
3470 3471 3472 3473
/*
 * endio for the write_dev_flush, this will wake anyone waiting
 * for the barrier when it is done
 */
3474
static void btrfs_end_empty_barrier(struct bio *bio)
C
Chris Mason 已提交
3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502
{
	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);

3503 3504
		if (bio->bi_error) {
			ret = bio->bi_error;
3505 3506
			btrfs_dev_stat_inc_and_print(device,
				BTRFS_DEV_STAT_FLUSH_ERRS);
C
Chris Mason 已提交
3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519
		}

		/* 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
	 */
3520
	device->flush_bio = NULL;
3521
	bio = btrfs_io_bio_alloc(GFP_NOFS, 0);
C
Chris Mason 已提交
3522 3523 3524 3525 3526
	if (!bio)
		return -ENOMEM;

	bio->bi_end_io = btrfs_end_empty_barrier;
	bio->bi_bdev = device->bdev;
M
Mike Christie 已提交
3527
	bio_set_op_attrs(bio, REQ_OP_WRITE, WRITE_FLUSH);
C
Chris Mason 已提交
3528 3529 3530 3531 3532
	init_completion(&device->flush_wait);
	bio->bi_private = &device->flush_wait;
	device->flush_bio = bio;

	bio_get(bio);
3533
	btrfsic_submit_bio(bio);
C
Chris Mason 已提交
3534 3535 3536 3537 3538 3539 3540 3541 3542 3543 3544 3545

	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;
3546 3547
	int errors_send = 0;
	int errors_wait = 0;
C
Chris Mason 已提交
3548 3549 3550 3551 3552
	int ret;

	/* send down all the barriers */
	head = &info->fs_devices->devices;
	list_for_each_entry_rcu(dev, head, dev_list) {
3553 3554
		if (dev->missing)
			continue;
C
Chris Mason 已提交
3555
		if (!dev->bdev) {
3556
			errors_send++;
C
Chris Mason 已提交
3557 3558 3559 3560 3561 3562 3563
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

		ret = write_dev_flush(dev, 0);
		if (ret)
3564
			errors_send++;
C
Chris Mason 已提交
3565 3566 3567 3568
	}

	/* wait for all the barriers */
	list_for_each_entry_rcu(dev, head, dev_list) {
3569 3570
		if (dev->missing)
			continue;
C
Chris Mason 已提交
3571
		if (!dev->bdev) {
3572
			errors_wait++;
C
Chris Mason 已提交
3573 3574 3575 3576 3577 3578 3579
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

		ret = write_dev_flush(dev, 1);
		if (ret)
3580
			errors_wait++;
C
Chris Mason 已提交
3581
	}
3582 3583
	if (errors_send > info->num_tolerated_disk_barrier_failures ||
	    errors_wait > info->num_tolerated_disk_barrier_failures)
C
Chris Mason 已提交
3584 3585 3586 3587
		return -EIO;
	return 0;
}

3588 3589
int btrfs_get_num_tolerated_disk_barrier_failures(u64 flags)
{
3590 3591
	int raid_type;
	int min_tolerated = INT_MAX;
3592

3593 3594 3595 3596 3597
	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);
3598

3599 3600 3601 3602 3603 3604 3605 3606 3607
	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);
	}
3608

3609 3610 3611 3612 3613 3614
	if (min_tolerated == INT_MAX) {
		pr_warn("BTRFS: unknown raid flag: %llu\n", flags);
		min_tolerated = 0;
	}

	return min_tolerated;
3615 3616
}

3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630
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;

3631
	for (i = 0; i < ARRAY_SIZE(types); i++) {
3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648
		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++) {
3649 3650 3651 3652 3653 3654 3655 3656 3657 3658
			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;
3659 3660 3661 3662 3663

			num_tolerated_disk_barrier_failures = min(
				num_tolerated_disk_barrier_failures,
				btrfs_get_num_tolerated_disk_barrier_failures(
					flags));
3664 3665 3666 3667 3668 3669 3670
		}
		up_read(&sinfo->groups_sem);
	}

	return num_tolerated_disk_barrier_failures;
}

3671
static int write_all_supers(struct btrfs_root *root, int max_mirrors)
3672
{
3673
	struct list_head *head;
3674
	struct btrfs_device *dev;
3675
	struct btrfs_super_block *sb;
3676 3677 3678
	struct btrfs_dev_item *dev_item;
	int ret;
	int do_barriers;
3679 3680
	int max_errors;
	int total_errors = 0;
3681
	u64 flags;
3682

3683
	do_barriers = !btrfs_test_opt(root->fs_info, NOBARRIER);
C
Chris Mason 已提交
3684
	backup_super_roots(root->fs_info);
3685

3686
	sb = root->fs_info->super_for_commit;
3687
	dev_item = &sb->dev_item;
3688

3689
	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
3690
	head = &root->fs_info->fs_devices->devices;
3691
	max_errors = btrfs_super_num_devices(root->fs_info->super_copy) - 1;
C
Chris Mason 已提交
3692

3693 3694 3695 3696 3697
	if (do_barriers) {
		ret = barrier_all_devices(root->fs_info);
		if (ret) {
			mutex_unlock(
				&root->fs_info->fs_devices->device_list_mutex);
3698
			btrfs_handle_fs_error(root->fs_info, ret,
3699 3700 3701 3702
				    "errors while submitting device barriers.");
			return ret;
		}
	}
C
Chris Mason 已提交
3703

3704
	list_for_each_entry_rcu(dev, head, dev_list) {
3705 3706 3707 3708
		if (!dev->bdev) {
			total_errors++;
			continue;
		}
Y
Yan Zheng 已提交
3709
		if (!dev->in_fs_metadata || !dev->writeable)
3710 3711
			continue;

Y
Yan Zheng 已提交
3712
		btrfs_set_stack_device_generation(dev_item, 0);
3713 3714
		btrfs_set_stack_device_type(dev_item, dev->type);
		btrfs_set_stack_device_id(dev_item, dev->devid);
3715
		btrfs_set_stack_device_total_bytes(dev_item,
3716
						   dev->commit_total_bytes);
3717 3718
		btrfs_set_stack_device_bytes_used(dev_item,
						  dev->commit_bytes_used);
3719 3720 3721 3722
		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
3723
		memcpy(dev_item->fsid, dev->fs_devices->fsid, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3724

3725 3726 3727
		flags = btrfs_super_flags(sb);
		btrfs_set_super_flags(sb, flags | BTRFS_HEADER_FLAG_WRITTEN);

Y
Yan Zheng 已提交
3728
		ret = write_dev_supers(dev, sb, do_barriers, 0, max_mirrors);
3729 3730
		if (ret)
			total_errors++;
3731
	}
3732
	if (total_errors > max_errors) {
3733
		btrfs_err(root->fs_info, "%d errors while writing supers",
C
Chris Mason 已提交
3734
		       total_errors);
3735
		mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
3736

3737
		/* FUA is masked off if unsupported and can't be the reason */
3738
		btrfs_handle_fs_error(root->fs_info, -EIO,
3739 3740
			    "%d errors while writing supers", total_errors);
		return -EIO;
3741
	}
3742

Y
Yan Zheng 已提交
3743
	total_errors = 0;
3744
	list_for_each_entry_rcu(dev, head, dev_list) {
3745 3746
		if (!dev->bdev)
			continue;
Y
Yan Zheng 已提交
3747
		if (!dev->in_fs_metadata || !dev->writeable)
3748 3749
			continue;

Y
Yan Zheng 已提交
3750 3751 3752
		ret = write_dev_supers(dev, sb, do_barriers, 1, max_mirrors);
		if (ret)
			total_errors++;
3753
	}
3754
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
3755
	if (total_errors > max_errors) {
3756
		btrfs_handle_fs_error(root->fs_info, -EIO,
3757 3758
			    "%d errors while writing supers", total_errors);
		return -EIO;
3759
	}
3760 3761 3762
	return 0;
}

Y
Yan Zheng 已提交
3763 3764
int write_ctree_super(struct btrfs_trans_handle *trans,
		      struct btrfs_root *root, int max_mirrors)
3765
{
3766
	return write_all_supers(root, max_mirrors);
C
Chris Mason 已提交
3767 3768
}

3769 3770 3771
/* 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
3772
{
3773
	spin_lock(&fs_info->fs_roots_radix_lock);
C
Chris Mason 已提交
3774 3775
	radix_tree_delete(&fs_info->fs_roots_radix,
			  (unsigned long)root->root_key.objectid);
3776
	spin_unlock(&fs_info->fs_roots_radix_lock);
3777 3778 3779 3780

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

L
Liu Bo 已提交
3781
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
L
Liu Bo 已提交
3782
		btrfs_free_log(NULL, root);
L
Liu Bo 已提交
3783 3784 3785 3786 3787 3788 3789
		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
3790

3791 3792 3793 3794
	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);
3795 3796 3797 3798 3799
	free_fs_root(root);
}

static void free_fs_root(struct btrfs_root *root)
{
3800
	iput(root->ino_cache_inode);
3801
	WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree));
3802 3803
	btrfs_free_block_rsv(root, root->orphan_block_rsv);
	root->orphan_block_rsv = NULL;
3804 3805
	if (root->anon_dev)
		free_anon_bdev(root->anon_dev);
3806 3807
	if (root->subv_writers)
		btrfs_free_subvolume_writers(root->subv_writers);
3808 3809
	free_extent_buffer(root->node);
	free_extent_buffer(root->commit_root);
3810 3811
	kfree(root->free_ino_ctl);
	kfree(root->free_ino_pinned);
C
Chris Mason 已提交
3812
	kfree(root->name);
3813
	btrfs_put_fs_root(root);
C
Chris Mason 已提交
3814 3815
}

3816 3817 3818
void btrfs_free_fs_root(struct btrfs_root *root)
{
	free_fs_root(root);
C
Chris Mason 已提交
3819 3820
}

Y
Yan Zheng 已提交
3821
int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
C
Chris Mason 已提交
3822
{
Y
Yan Zheng 已提交
3823 3824
	u64 root_objectid = 0;
	struct btrfs_root *gang[8];
3825 3826 3827 3828
	int i = 0;
	int err = 0;
	unsigned int ret = 0;
	int index;
3829

Y
Yan Zheng 已提交
3830
	while (1) {
3831
		index = srcu_read_lock(&fs_info->subvol_srcu);
Y
Yan Zheng 已提交
3832 3833 3834
		ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix,
					     (void **)gang, root_objectid,
					     ARRAY_SIZE(gang));
3835 3836
		if (!ret) {
			srcu_read_unlock(&fs_info->subvol_srcu, index);
Y
Yan Zheng 已提交
3837
			break;
3838
		}
3839
		root_objectid = gang[ret - 1]->root_key.objectid + 1;
3840

Y
Yan Zheng 已提交
3841
		for (i = 0; i < ret; i++) {
3842 3843 3844 3845 3846 3847 3848 3849 3850
			/* 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);
3851

3852 3853 3854
		for (i = 0; i < ret; i++) {
			if (!gang[i])
				continue;
Y
Yan Zheng 已提交
3855
			root_objectid = gang[i]->root_key.objectid;
3856 3857
			err = btrfs_orphan_cleanup(gang[i]);
			if (err)
3858 3859
				break;
			btrfs_put_fs_root(gang[i]);
Y
Yan Zheng 已提交
3860 3861 3862
		}
		root_objectid++;
	}
3863 3864 3865 3866 3867 3868 3869

	/* 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
3870
}
3871

Y
Yan Zheng 已提交
3872 3873 3874
int btrfs_commit_super(struct btrfs_root *root)
{
	struct btrfs_trans_handle *trans;
3875

Y
Yan Zheng 已提交
3876
	mutex_lock(&root->fs_info->cleaner_mutex);
Y
Yan, Zheng 已提交
3877
	btrfs_run_delayed_iputs(root);
Y
Yan Zheng 已提交
3878
	mutex_unlock(&root->fs_info->cleaner_mutex);
D
David Sterba 已提交
3879
	wake_up_process(root->fs_info->cleaner_kthread);
3880 3881 3882 3883 3884

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

3885
	trans = btrfs_join_transaction(root);
3886 3887
	if (IS_ERR(trans))
		return PTR_ERR(trans);
3888
	return btrfs_commit_transaction(trans, root);
Y
Yan Zheng 已提交
3889 3890
}

3891
void close_ctree(struct btrfs_root *root)
Y
Yan Zheng 已提交
3892 3893 3894 3895 3896 3897 3898
{
	struct btrfs_fs_info *fs_info = root->fs_info;
	int ret;

	fs_info->closing = 1;
	smp_mb();

3899
	/* wait for the qgroup rescan worker to stop */
3900
	btrfs_qgroup_wait_for_completion(fs_info, false);
3901

S
Stefan Behrens 已提交
3902 3903 3904 3905 3906
	/* 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);

3907
	/* pause restriper - we want to resume on mount */
3908
	btrfs_pause_balance(fs_info);
3909

3910 3911
	btrfs_dev_replace_suspend_for_unmount(fs_info);

3912
	btrfs_scrub_cancel(fs_info);
C
Chris Mason 已提交
3913 3914 3915 3916 3917 3918

	/* 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 */
3919
	btrfs_cleanup_defrag_inodes(fs_info);
C
Chris Mason 已提交
3920

3921 3922
	cancel_work_sync(&fs_info->async_reclaim_work);

Y
Yan Zheng 已提交
3923
	if (!(fs_info->sb->s_flags & MS_RDONLY)) {
3924 3925 3926 3927 3928 3929 3930
		/*
		 * 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
3931 3932
		ret = btrfs_commit_super(root);
		if (ret)
3933
			btrfs_err(fs_info, "commit super ret %d", ret);
L
liubo 已提交
3934 3935
	}

3936
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state))
3937
		btrfs_error_commit_super(root);
3938

A
Al Viro 已提交
3939 3940
	kthread_stop(fs_info->transaction_kthread);
	kthread_stop(fs_info->cleaner_kthread);
3941

3942 3943 3944
	fs_info->closing = 2;
	smp_mb();

3945
	btrfs_free_qgroup_config(fs_info);
3946

3947
	if (percpu_counter_sum(&fs_info->delalloc_bytes)) {
3948
		btrfs_info(fs_info, "at unmount delalloc count %lld",
3949
		       percpu_counter_sum(&fs_info->delalloc_bytes));
C
Chris Mason 已提交
3950
	}
3951

3952
	btrfs_sysfs_remove_mounted(fs_info);
3953
	btrfs_sysfs_remove_fsid(fs_info->fs_devices);
3954

3955
	btrfs_free_fs_roots(fs_info);
3956

3957 3958
	btrfs_put_block_group_cache(fs_info);

3959 3960
	btrfs_free_block_groups(fs_info);

3961 3962 3963 3964 3965
	/*
	 * 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);
3966 3967
	btrfs_stop_all_workers(fs_info);

3968
	fs_info->open = 0;
3969
	free_root_pointers(fs_info, 1);
3970

3971
	iput(fs_info->btree_inode);
3972

3973
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
3974
	if (btrfs_test_opt(root->fs_info, CHECK_INTEGRITY))
3975 3976 3977
		btrfsic_unmount(root, fs_info->fs_devices);
#endif

3978
	btrfs_close_devices(fs_info->fs_devices);
3979
	btrfs_mapping_tree_free(&fs_info->mapping_tree);
3980

3981
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
3982
	percpu_counter_destroy(&fs_info->delalloc_bytes);
3983
	percpu_counter_destroy(&fs_info->bio_counter);
C
Chris Mason 已提交
3984
	bdi_destroy(&fs_info->bdi);
3985
	cleanup_srcu_struct(&fs_info->subvol_srcu);
3986

D
David Woodhouse 已提交
3987 3988
	btrfs_free_stripe_hash_table(fs_info);

3989
	__btrfs_free_block_rsv(root->orphan_block_rsv);
3990
	root->orphan_block_rsv = NULL;
3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001

	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);
4002 4003
}

4004 4005
int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid,
			  int atomic)
4006
{
4007
	int ret;
4008
	struct inode *btree_inode = buf->pages[0]->mapping->host;
4009

4010
	ret = extent_buffer_uptodate(buf);
4011 4012 4013 4014
	if (!ret)
		return ret;

	ret = verify_parent_transid(&BTRFS_I(btree_inode)->io_tree, buf,
4015 4016 4017
				    parent_transid, atomic);
	if (ret == -EAGAIN)
		return ret;
4018
	return !ret;
4019 4020 4021 4022
}

void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
{
4023
	struct btrfs_root *root;
4024
	u64 transid = btrfs_header_generation(buf);
4025
	int was_dirty;
4026

4027 4028 4029 4030 4031 4032 4033 4034 4035 4036
#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;
4037
	btrfs_assert_tree_locked(buf);
J
Julia Lawall 已提交
4038 4039
	if (transid != root->fs_info->generation)
		WARN(1, KERN_CRIT "btrfs transid mismatch buffer %llu, "
C
Chris Mason 已提交
4040
		       "found %llu running %llu\n",
4041
			buf->start, transid, root->fs_info->generation);
4042
	was_dirty = set_extent_buffer_dirty(buf);
4043 4044 4045 4046
	if (!was_dirty)
		__percpu_counter_add(&root->fs_info->dirty_metadata_bytes,
				     buf->len,
				     root->fs_info->dirty_metadata_batch);
4047 4048 4049 4050 4051 4052
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	if (btrfs_header_level(buf) == 0 && check_leaf(root, buf)) {
		btrfs_print_leaf(root, buf);
		ASSERT(0);
	}
#endif
4053 4054
}

4055 4056
static void __btrfs_btree_balance_dirty(struct btrfs_root *root,
					int flush_delayed)
4057 4058 4059 4060 4061
{
	/*
	 * looks as though older kernels can get into trouble with
	 * this code, they end up stuck in balance_dirty_pages forever
	 */
4062
	int ret;
4063 4064 4065 4066

	if (current->flags & PF_MEMALLOC)
		return;

4067 4068
	if (flush_delayed)
		btrfs_balance_delayed_items(root);
4069

4070 4071 4072
	ret = percpu_counter_compare(&root->fs_info->dirty_metadata_bytes,
				     BTRFS_DIRTY_METADATA_THRESH);
	if (ret > 0) {
4073 4074
		balance_dirty_pages_ratelimited(
				   root->fs_info->btree_inode->i_mapping);
4075 4076 4077
	}
}

4078
void btrfs_btree_balance_dirty(struct btrfs_root *root)
C
Chris Mason 已提交
4079
{
4080 4081
	__btrfs_btree_balance_dirty(root, 1);
}
4082

4083 4084 4085
void btrfs_btree_balance_dirty_nodelay(struct btrfs_root *root)
{
	__btrfs_btree_balance_dirty(root, 0);
C
Chris Mason 已提交
4086
}
4087

4088
int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid)
4089
{
4090
	struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root;
4091
	return btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
4092
}
4093

4094
static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info,
L
liubo 已提交
4095 4096
			      int read_only)
{
D
David Sterba 已提交
4097
	struct btrfs_super_block *sb = fs_info->super_copy;
4098 4099
	u64 nodesize = btrfs_super_nodesize(sb);
	u64 sectorsize = btrfs_super_sectorsize(sb);
D
David Sterba 已提交
4100 4101
	int ret = 0;

4102 4103 4104 4105 4106 4107 4108
	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);
4109 4110 4111
	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
4112 4113
		ret = -EINVAL;
	}
4114 4115 4116
	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
4117 4118
		ret = -EINVAL;
	}
4119 4120 4121
	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
4122 4123 4124
		ret = -EINVAL;
	}

D
David Sterba 已提交
4125
	/*
4126 4127
	 * 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
4128
	 */
4129 4130 4131 4132 4133 4134
	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 */
4135
	if (sectorsize != PAGE_SIZE) {
4136
		printk(KERN_ERR "BTRFS: sectorsize %llu not supported yet, only support %lu\n",
4137
				sectorsize, PAGE_SIZE);
4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153
		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
4154
		printk(KERN_WARNING "BTRFS: tree_root block unaligned: %llu\n",
4155
				btrfs_super_root(sb));
4156 4157 4158
		ret = -EINVAL;
	}
	if (!IS_ALIGNED(btrfs_super_chunk_root(sb), sectorsize)) {
4159 4160
		printk(KERN_WARNING "BTRFS: chunk_root block unaligned: %llu\n",
				btrfs_super_chunk_root(sb));
4161 4162
		ret = -EINVAL;
	}
4163 4164 4165
	if (!IS_ALIGNED(btrfs_super_log_root(sb), sectorsize)) {
		printk(KERN_WARNING "BTRFS: log_root block unaligned: %llu\n",
				btrfs_super_log_root(sb));
4166 4167 4168
		ret = -EINVAL;
	}

D
David Sterba 已提交
4169 4170 4171 4172 4173 4174 4175 4176 4177 4178
	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
	 */
4179 4180 4181 4182 4183
	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;
	}
4184
	if (!is_power_of_2(btrfs_super_stripesize(sb))) {
4185 4186 4187 4188
		btrfs_err(fs_info, "invalid stripesize %u",
		       btrfs_super_stripesize(sb));
		ret = -EINVAL;
	}
4189
	if (btrfs_super_num_devices(sb) > (1UL << 31))
D
David Sterba 已提交
4190
		printk(KERN_WARNING "BTRFS: suspicious number of devices: %llu\n",
4191
				btrfs_super_num_devices(sb));
4192 4193 4194 4195
	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
4196

4197
	if (btrfs_super_bytenr(sb) != BTRFS_SUPER_INFO_OFFSET) {
D
David Sterba 已提交
4198
		printk(KERN_ERR "BTRFS: super offset mismatch %llu != %u\n",
4199
				btrfs_super_bytenr(sb), BTRFS_SUPER_INFO_OFFSET);
D
David Sterba 已提交
4200 4201 4202
		ret = -EINVAL;
	}

4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214
	/*
	 * 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)) {
4215
		printk(KERN_ERR "BTRFS: system chunk array too small %u < %zu\n",
4216 4217 4218 4219 4220 4221
				btrfs_super_sys_array_size(sb),
				sizeof(struct btrfs_disk_key)
				+ sizeof(struct btrfs_chunk));
		ret = -EINVAL;
	}

D
David Sterba 已提交
4222 4223 4224 4225
	/*
	 * 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.
	 */
4226
	if (btrfs_super_generation(sb) < btrfs_super_chunk_root_generation(sb))
D
David Sterba 已提交
4227 4228
		printk(KERN_WARNING
			"BTRFS: suspicious: generation < chunk_root_generation: %llu < %llu\n",
4229 4230 4231
			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
4232 4233
		printk(KERN_WARNING
			"BTRFS: suspicious: generation < cache_generation: %llu < %llu\n",
4234
			btrfs_super_generation(sb), btrfs_super_cache_generation(sb));
D
David Sterba 已提交
4235 4236

	return ret;
L
liubo 已提交
4237 4238
}

4239
static void btrfs_error_commit_super(struct btrfs_root *root)
L
liubo 已提交
4240 4241 4242 4243 4244 4245 4246 4247 4248 4249 4250 4251
{
	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);
}

4252
static void btrfs_destroy_ordered_extents(struct btrfs_root *root)
L
liubo 已提交
4253 4254 4255
{
	struct btrfs_ordered_extent *ordered;

4256
	spin_lock(&root->ordered_extent_lock);
4257 4258 4259 4260
	/*
	 * This will just short circuit the ordered completion stuff which will
	 * make sure the ordered extent gets properly cleaned up.
	 */
4261
	list_for_each_entry(ordered, &root->ordered_extents,
4262 4263
			    root_extent_list)
		set_bit(BTRFS_ORDERED_IOERR, &ordered->flags);
4264 4265 4266 4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278
	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);
4279 4280
		list_move_tail(&root->ordered_root,
			       &fs_info->ordered_roots);
4281

4282
		spin_unlock(&fs_info->ordered_root_lock);
4283 4284
		btrfs_destroy_ordered_extents(root);

4285 4286
		cond_resched();
		spin_lock(&fs_info->ordered_root_lock);
4287 4288
	}
	spin_unlock(&fs_info->ordered_root_lock);
L
liubo 已提交
4289 4290
}

4291 4292
static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
				      struct btrfs_root *root)
L
liubo 已提交
4293 4294 4295 4296 4297 4298 4299 4300 4301
{
	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);
4302
	if (atomic_read(&delayed_refs->num_entries) == 0) {
4303
		spin_unlock(&delayed_refs->lock);
4304
		btrfs_info(root->fs_info, "delayed_refs has NO entry");
L
liubo 已提交
4305 4306 4307
		return ret;
	}

4308 4309
	while ((node = rb_first(&delayed_refs->href_root)) != NULL) {
		struct btrfs_delayed_ref_head *head;
4310
		struct btrfs_delayed_ref_node *tmp;
4311
		bool pin_bytes = false;
L
liubo 已提交
4312

4313 4314 4315 4316 4317
		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);
4318

4319
			mutex_lock(&head->mutex);
4320
			mutex_unlock(&head->mutex);
4321 4322 4323 4324 4325
			btrfs_put_delayed_ref(&head->node);
			spin_lock(&delayed_refs->lock);
			continue;
		}
		spin_lock(&head->lock);
4326 4327
		list_for_each_entry_safe_reverse(ref, tmp, &head->ref_list,
						 list) {
4328
			ref->in_tree = 0;
4329
			list_del(&ref->list);
4330 4331
			atomic_dec(&delayed_refs->num_entries);
			btrfs_put_delayed_ref(ref);
4332
		}
4333 4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344
		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
4345

4346 4347 4348 4349
		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
4350 4351 4352 4353 4354 4355 4356 4357 4358
		cond_resched();
		spin_lock(&delayed_refs->lock);
	}

	spin_unlock(&delayed_refs->lock);

	return ret;
}

4359
static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root)
L
liubo 已提交
4360 4361 4362 4363 4364 4365
{
	struct btrfs_inode *btrfs_inode;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

4366 4367
	spin_lock(&root->delalloc_lock);
	list_splice_init(&root->delalloc_inodes, &splice);
L
liubo 已提交
4368 4369

	while (!list_empty(&splice)) {
4370 4371
		btrfs_inode = list_first_entry(&splice, struct btrfs_inode,
					       delalloc_inodes);
L
liubo 已提交
4372 4373

		list_del_init(&btrfs_inode->delalloc_inodes);
4374 4375
		clear_bit(BTRFS_INODE_IN_DELALLOC_LIST,
			  &btrfs_inode->runtime_flags);
4376
		spin_unlock(&root->delalloc_lock);
L
liubo 已提交
4377 4378

		btrfs_invalidate_inodes(btrfs_inode->root);
4379

4380
		spin_lock(&root->delalloc_lock);
L
liubo 已提交
4381 4382
	}

4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408
	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
4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421
}

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,
4422
					    mark, NULL);
L
liubo 已提交
4423 4424 4425
		if (ret)
			break;

4426
		clear_extent_bits(dirty_pages, start, end, mark);
L
liubo 已提交
4427
		while (start <= end) {
4428
			eb = btrfs_find_tree_block(root->fs_info, start);
4429
			start += root->nodesize;
4430
			if (!eb)
L
liubo 已提交
4431
				continue;
4432
			wait_on_extent_buffer_writeback(eb);
L
liubo 已提交
4433

4434 4435 4436 4437
			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
4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450
		}
	}

	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;
4451
	bool loop = true;
L
liubo 已提交
4452 4453

	unpin = pinned_extents;
4454
again:
L
liubo 已提交
4455 4456
	while (1) {
		ret = find_first_extent_bit(unpin, 0, &start, &end,
4457
					    EXTENT_DIRTY, NULL);
L
liubo 已提交
4458 4459 4460
		if (ret)
			break;

4461
		clear_extent_dirty(unpin, start, end);
L
liubo 已提交
4462 4463 4464 4465
		btrfs_error_unpin_extent_range(root, start, end);
		cond_resched();
	}

4466 4467 4468 4469 4470 4471 4472 4473 4474
	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
4475 4476 4477
	return 0;
}

4478 4479 4480 4481 4482
void btrfs_cleanup_one_transaction(struct btrfs_transaction *cur_trans,
				   struct btrfs_root *root)
{
	btrfs_destroy_delayed_refs(cur_trans, root);

4483
	cur_trans->state = TRANS_STATE_COMMIT_START;
4484
	wake_up(&root->fs_info->transaction_blocked_wait);
4485

4486
	cur_trans->state = TRANS_STATE_UNBLOCKED;
4487
	wake_up(&root->fs_info->transaction_wait);
4488

4489 4490
	btrfs_destroy_delayed_inodes(root);
	btrfs_assert_delayed_root_empty(root);
4491 4492 4493

	btrfs_destroy_marked_extents(root, &cur_trans->dirty_pages,
				     EXTENT_DIRTY);
4494 4495
	btrfs_destroy_pinned_extent(root,
				    root->fs_info->pinned_extents);
4496

4497 4498 4499
	cur_trans->state =TRANS_STATE_COMPLETED;
	wake_up(&cur_trans->commit_wait);

4500 4501 4502 4503 4504 4505
	/*
	memset(cur_trans, 0, sizeof(*cur_trans));
	kmem_cache_free(btrfs_transaction_cachep, cur_trans);
	*/
}

4506
static int btrfs_cleanup_transaction(struct btrfs_root *root)
L
liubo 已提交
4507 4508 4509 4510 4511
{
	struct btrfs_transaction *t;

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

J
Josef Bacik 已提交
4512
	spin_lock(&root->fs_info->trans_lock);
4513 4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527 4528 4529 4530 4531 4532 4533 4534 4535 4536
	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);
4537

4538 4539 4540
		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
4541
		list_del_init(&t->list);
4542
		spin_unlock(&root->fs_info->trans_lock);
L
liubo 已提交
4543

4544 4545 4546 4547 4548 4549 4550 4551 4552 4553
		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
4554 4555 4556 4557 4558
	mutex_unlock(&root->fs_info->transaction_kthread_mutex);

	return 0;
}

4559
static const struct extent_io_ops btree_extent_io_ops = {
4560
	.readpage_end_io_hook = btree_readpage_end_io_hook,
A
Arne Jansen 已提交
4561
	.readpage_io_failed_hook = btree_io_failed_hook,
4562
	.submit_bio_hook = btree_submit_bio_hook,
4563 4564
	/* note we're sharing with inode.c for the merge bio hook */
	.merge_bio_hook = btrfs_merge_bio_hook,
4565
};