disk-io.c 126.9 KB
Newer Older
C
Chris Mason 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
/*
 * Copyright (C) 2007 Oracle.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License v2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */

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

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

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

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

C
Chris Mason 已提交
80
/*
81 82
 * btrfs_end_io_wq structs are used to do processing in task context when an IO
 * is complete.  This is used during reads to verify checksums, and it is used
C
Chris Mason 已提交
83 84
 * by writes to insert metadata for new file extents after IO is complete.
 */
85
struct btrfs_end_io_wq {
86 87 88 89 90
	struct bio *bio;
	bio_end_io_t *end_io;
	void *private;
	struct btrfs_fs_info *info;
	int error;
91
	enum btrfs_wq_endio_type metadata;
92
	struct list_head list;
93
	struct btrfs_work work;
94
};
95

96 97 98 99 100 101 102
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,
103
					SLAB_MEM_SPREAD,
104 105 106 107 108 109 110 111
					NULL);
	if (!btrfs_end_io_wq_cache)
		return -ENOMEM;
	return 0;
}

void btrfs_end_io_wq_exit(void)
{
112
	kmem_cache_destroy(btrfs_end_io_wq_cache);
113 114
}

C
Chris Mason 已提交
115 116 117 118 119
/*
 * 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.
 */
120 121 122 123
struct async_submit_bio {
	struct inode *inode;
	struct bio *bio;
	struct list_head list;
C
Chris Mason 已提交
124 125
	extent_submit_bio_hook_t *submit_bio_start;
	extent_submit_bio_hook_t *submit_bio_done;
126
	int mirror_num;
C
Chris Mason 已提交
127
	unsigned long bio_flags;
128 129 130 131 132
	/*
	 * 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;
133
	struct btrfs_work work;
134
	int error;
135 136
};

137 138 139 140 141 142 143 144 145 146
/*
 * 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.
147
 *
148 149 150
 * 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.
151
 *
152 153 154
 * 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.
155
 *
156 157 158
 * 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.
159 160 161 162 163
 */
#ifdef CONFIG_DEBUG_LOCK_ALLOC
# if BTRFS_MAX_LEVEL != 8
#  error
# endif
164 165 166 167 168 169 170 171 172 173 174 175 176

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"	},
177
	{ .id = BTRFS_QUOTA_TREE_OBJECTID,	.name_stem = "quota"	},
178 179 180
	{ .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"	},
181
	{ .id = BTRFS_UUID_TREE_OBJECTID,	.name_stem = "uuid"	},
182
	{ .id = BTRFS_FREE_SPACE_TREE_OBJECTID,	.name_stem = "free-space" },
183
	{ .id = 0,				.name_stem = "tree"	},
184
};
185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215

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]);
}

216 217
#endif

C
Chris Mason 已提交
218 219 220 221
/*
 * extents on the btree inode are pretty simple, there's one extent
 * that covers the entire device
 */
222
static struct extent_map *btree_get_extent(struct btrfs_inode *inode,
223
		struct page *page, size_t pg_offset, u64 start, u64 len,
224
		int create)
225
{
226 227
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->vfs_inode.i_sb);
	struct extent_map_tree *em_tree = &inode->extent_tree;
228 229 230
	struct extent_map *em;
	int ret;

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

240
	em = alloc_extent_map();
241 242 243 244 245
	if (!em) {
		em = ERR_PTR(-ENOMEM);
		goto out;
	}
	em->start = 0;
246
	em->len = (u64)-1;
C
Chris Mason 已提交
247
	em->block_len = (u64)-1;
248
	em->block_start = 0;
249
	em->bdev = fs_info->fs_devices->latest_bdev;
250

251
	write_lock(&em_tree->lock);
J
Josef Bacik 已提交
252
	ret = add_extent_mapping(em_tree, em, 0);
253 254
	if (ret == -EEXIST) {
		free_extent_map(em);
255
		em = lookup_extent_mapping(em_tree, start, len);
256
		if (!em)
257
			em = ERR_PTR(-EIO);
258
	} else if (ret) {
259
		free_extent_map(em);
260
		em = ERR_PTR(ret);
261
	}
262
	write_unlock(&em_tree->lock);
263

264 265
out:
	return em;
266 267
}

268
u32 btrfs_csum_data(const char *data, u32 seed, size_t len)
269
{
270
	return btrfs_crc32c(seed, data, len);
271 272
}

273
void btrfs_csum_final(u32 crc, u8 *result)
274
{
275
	put_unaligned_le32(~crc, result);
276 277
}

C
Chris Mason 已提交
278 279 280 281
/*
 * compute the csum for a btree block, and either verify it or write it
 * into the csum field of the block.
 */
282 283
static int csum_tree_block(struct btrfs_fs_info *fs_info,
			   struct extent_buffer *buf,
284 285
			   int verify)
{
286
	u16 csum_size = btrfs_super_csum_size(fs_info->super_copy);
287
	char *result = NULL;
288 289 290 291 292 293 294 295
	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;
296
	unsigned long inline_result;
297 298

	len = buf->len - offset;
C
Chris Mason 已提交
299
	while (len > 0) {
300
		err = map_private_extent_buffer(buf, offset, 32,
301
					&kaddr, &map_start, &map_len);
C
Chris Mason 已提交
302
		if (err)
303
			return err;
304
		cur_len = min(len, map_len - (offset - map_start));
305
		crc = btrfs_csum_data(kaddr + offset - map_start,
306 307 308 309
				      crc, cur_len);
		len -= cur_len;
		offset += cur_len;
	}
310
	if (csum_size > sizeof(inline_result)) {
311
		result = kzalloc(csum_size, GFP_NOFS);
312
		if (!result)
313
			return -ENOMEM;
314 315 316 317
	} else {
		result = (char *)&inline_result;
	}

318 319 320
	btrfs_csum_final(crc, result);

	if (verify) {
321
		if (memcmp_extent_buffer(buf, result, 0, csum_size)) {
322 323
			u32 val;
			u32 found = 0;
324
			memcpy(&found, result, csum_size);
325

326
			read_extent_buffer(buf, &val, 0, csum_size);
327
			btrfs_warn_rl(fs_info,
J
Jeff Mahoney 已提交
328
				"%s checksum verify failed on %llu wanted %X found %X level %d",
329
				fs_info->sb->s_id, buf->start,
330
				val, found, btrfs_header_level(buf));
331 332
			if (result != (char *)&inline_result)
				kfree(result);
333
			return -EUCLEAN;
334 335
		}
	} else {
336
		write_extent_buffer(buf, result, 0, csum_size);
337
	}
338 339
	if (result != (char *)&inline_result)
		kfree(result);
340 341 342
	return 0;
}

C
Chris Mason 已提交
343 344 345 346 347 348
/*
 * 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.
 */
349
static int verify_parent_transid(struct extent_io_tree *io_tree,
350 351
				 struct extent_buffer *eb, u64 parent_transid,
				 int atomic)
352
{
353
	struct extent_state *cached_state = NULL;
354
	int ret;
355
	bool need_lock = (current->journal_info == BTRFS_SEND_TRANS_STUB);
356 357 358 359

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

360 361 362
	if (atomic)
		return -EAGAIN;

363 364 365 366 367
	if (need_lock) {
		btrfs_tree_read_lock(eb);
		btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
	}

368
	lock_extent_bits(io_tree, eb->start, eb->start + eb->len - 1,
369
			 &cached_state);
370
	if (extent_buffer_uptodate(eb) &&
371 372 373 374
	    btrfs_header_generation(eb) == parent_transid) {
		ret = 0;
		goto out;
	}
375 376 377
	btrfs_err_rl(eb->fs_info,
		"parent transid verify failed on %llu wanted %llu found %llu",
			eb->start,
378
			parent_transid, btrfs_header_generation(eb));
379
	ret = 1;
380 381 382 383

	/*
	 * 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
384
	 * block that has been freed and re-allocated.  So don't clear uptodate
385 386 387 388 389 390
	 * 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
391
out:
392 393
	unlock_extent_cached(io_tree, eb->start, eb->start + eb->len - 1,
			     &cached_state, GFP_NOFS);
394 395
	if (need_lock)
		btrfs_tree_read_unlock_blocking(eb);
396 397 398
	return ret;
}

D
David Sterba 已提交
399 400 401 402
/*
 * Return 0 if the superblock checksum type matches the checksum value of that
 * algorithm. Pass the raw disk superblock data.
 */
403 404
static int btrfs_check_super_csum(struct btrfs_fs_info *fs_info,
				  char *raw_disk_sb)
D
David Sterba 已提交
405 406 407 408 409 410 411 412 413 414 415 416 417 418
{
	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
419
		 * is filled with zeros and is included in the checksum.
D
David Sterba 已提交
420 421 422 423 424 425 426 427 428 429
		 */
		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)) {
430
		btrfs_err(fs_info, "unsupported checksum algorithm %u",
D
David Sterba 已提交
431 432 433 434 435 436 437
				csum_type);
		ret = 1;
	}

	return ret;
}

C
Chris Mason 已提交
438 439 440 441
/*
 * helper to read a given tree block, doing retries as required when
 * the checksums don't match and we have alternate mirrors to try.
 */
442
static int btree_read_extent_buffer_pages(struct btrfs_fs_info *fs_info,
443
					  struct extent_buffer *eb,
444
					  u64 parent_transid)
445 446
{
	struct extent_io_tree *io_tree;
447
	int failed = 0;
448 449 450
	int ret;
	int num_copies = 0;
	int mirror_num = 0;
451
	int failed_mirror = 0;
452

453
	clear_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
454
	io_tree = &BTRFS_I(fs_info->btree_inode)->io_tree;
455
	while (1) {
456
		ret = read_extent_buffer_pages(io_tree, eb, WAIT_COMPLETE,
457
					       btree_get_extent, mirror_num);
458 459
		if (!ret) {
			if (!verify_parent_transid(io_tree, eb,
460
						   parent_transid, 0))
461 462 463 464
				break;
			else
				ret = -EIO;
		}
C
Chris Mason 已提交
465

466 467 468 469 470 471
		/*
		 * 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))
472 473
			break;

474
		num_copies = btrfs_num_copies(fs_info,
475
					      eb->start, eb->len);
C
Chris Mason 已提交
476
		if (num_copies == 1)
477
			break;
C
Chris Mason 已提交
478

479 480 481 482 483
		if (!failed_mirror) {
			failed = 1;
			failed_mirror = eb->read_mirror;
		}

484
		mirror_num++;
485 486 487
		if (mirror_num == failed_mirror)
			mirror_num++;

C
Chris Mason 已提交
488
		if (mirror_num > num_copies)
489
			break;
490
	}
491

492
	if (failed && !ret && failed_mirror)
493
		repair_eb_io_failure(fs_info, eb, failed_mirror);
494 495

	return ret;
496
}
497

C
Chris Mason 已提交
498
/*
C
Chris Mason 已提交
499 500
 * 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
501
 */
C
Chris Mason 已提交
502

503
static int csum_dirty_buffer(struct btrfs_fs_info *fs_info, struct page *page)
504
{
M
Miao Xie 已提交
505
	u64 start = page_offset(page);
506 507
	u64 found_start;
	struct extent_buffer *eb;
508

J
Josef Bacik 已提交
509 510 511
	eb = (struct extent_buffer *)page->private;
	if (page != eb->pages[0])
		return 0;
512

513
	found_start = btrfs_header_bytenr(eb);
514 515 516 517 518 519 520 521 522 523 524 525
	/*
	 * 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);

526
	return csum_tree_block(fs_info, eb, 0);
527 528
}

529
static int check_tree_block_fsid(struct btrfs_fs_info *fs_info,
Y
Yan Zheng 已提交
530 531
				 struct extent_buffer *eb)
{
532
	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
Y
Yan Zheng 已提交
533 534 535
	u8 fsid[BTRFS_UUID_SIZE];
	int ret = 1;

536
	read_extent_buffer(eb, fsid, btrfs_header_fsid(), BTRFS_FSID_SIZE);
Y
Yan Zheng 已提交
537 538 539 540 541 542 543 544 545 546
	while (fs_devices) {
		if (!memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE)) {
			ret = 0;
			break;
		}
		fs_devices = fs_devices->seed;
	}
	return ret;
}

547 548 549 550
#define CORRUPT(reason, eb, root, slot)					\
	btrfs_crit(root->fs_info,					\
		   "corrupt %s, %s: block=%llu, root=%llu, slot=%d",	\
		   btrfs_header_level(eb) == 0 ? "leaf" : "node",	\
551
		   reason, btrfs_header_bytenr(eb), root->objectid, slot)
552 553 554 555

static noinline int check_leaf(struct btrfs_root *root,
			       struct extent_buffer *leaf)
{
556
	struct btrfs_fs_info *fs_info = root->fs_info;
557 558 559 560 561
	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
	/*
	 * Extent buffers from a relocation tree have a owner field that
	 * corresponds to the subvolume tree they are based on. So just from an
	 * extent buffer alone we can not find out what is the id of the
	 * corresponding subvolume tree, so we can not figure out if the extent
	 * buffer corresponds to the root of the relocation tree or not. So skip
	 * this check for relocation trees.
	 */
	if (nritems == 0 && !btrfs_header_flag(leaf, BTRFS_HEADER_FLAG_RELOC)) {
571 572 573 574 575 576
		struct btrfs_root *check_root;

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

577
		check_root = btrfs_get_fs_root(fs_info, &key, false);
578 579 580 581 582
		/*
		 * 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)) {
583 584 585
			struct extent_buffer *eb;

			eb = btrfs_root_node(check_root);
586
			/* if leaf is the root, then it's fine */
587
			if (leaf != eb) {
588
				CORRUPT("non-root leaf's nritems is 0",
589 590
					leaf, check_root, 0);
				free_extent_buffer(eb);
591 592
				return -EIO;
			}
593
			free_extent_buffer(eb);
594
		}
595
		return 0;
596
	}
597

598 599 600
	if (nritems == 0)
		return 0;

601 602
	/* Check the 0 item */
	if (btrfs_item_offset_nr(leaf, 0) + btrfs_item_size_nr(leaf, 0) !=
603
	    BTRFS_LEAF_DATA_SIZE(fs_info)) {
604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637
		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,
638
		 * just in case all the items are consistent to each other, but
639 640 641
		 * all point outside of the leaf.
		 */
		if (btrfs_item_end_nr(leaf, slot) >
642
		    BTRFS_LEAF_DATA_SIZE(fs_info)) {
643 644 645 646 647 648 649 650
			CORRUPT("slot end outside of leaf", leaf, root, slot);
			return -EIO;
		}
	}

	return 0;
}

L
Liu Bo 已提交
651 652 653
static int check_node(struct btrfs_root *root, struct extent_buffer *node)
{
	unsigned long nr = btrfs_header_nritems(node);
654 655 656 657
	struct btrfs_key key, next_key;
	int slot;
	u64 bytenr;
	int ret = 0;
L
Liu Bo 已提交
658

659
	if (nr == 0 || nr > BTRFS_NODEPTRS_PER_BLOCK(root->fs_info)) {
L
Liu Bo 已提交
660 661 662 663 664
		btrfs_crit(root->fs_info,
			   "corrupt node: block %llu root %llu nritems %lu",
			   node->start, root->objectid, nr);
		return -EIO;
	}
665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684

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

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

		if (btrfs_comp_cpu_keys(&key, &next_key) >= 0) {
			CORRUPT("bad key order", node, root, slot);
			ret = -EIO;
			goto out;
		}
	}
out:
	return ret;
L
Liu Bo 已提交
685 686
}

687 688 689
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)
690 691 692 693 694
{
	u64 found_start;
	int found_level;
	struct extent_buffer *eb;
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
695
	struct btrfs_fs_info *fs_info = root->fs_info;
696
	int ret = 0;
697
	int reads_done;
698 699 700

	if (!page->private)
		goto out;
C
Chris Mason 已提交
701

J
Josef Bacik 已提交
702
	eb = (struct extent_buffer *)page->private;
C
Chris Mason 已提交
703

704 705 706 707 708 709
	/* 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);
710 711
	if (!reads_done)
		goto err;
712

713
	eb->read_mirror = mirror;
714
	if (test_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags)) {
715 716 717 718
		ret = -EIO;
		goto err;
	}

719
	found_start = btrfs_header_bytenr(eb);
720
	if (found_start != eb->start) {
721 722
		btrfs_err_rl(fs_info, "bad tree block start %llu %llu",
			     found_start, eb->start);
723
		ret = -EIO;
724 725
		goto err;
	}
726 727 728
	if (check_tree_block_fsid(fs_info, eb)) {
		btrfs_err_rl(fs_info, "bad fsid on block %llu",
			     eb->start);
729 730 731
		ret = -EIO;
		goto err;
	}
732
	found_level = btrfs_header_level(eb);
733
	if (found_level >= BTRFS_MAX_LEVEL) {
734 735
		btrfs_err(fs_info, "bad tree block level %d",
			  (int)btrfs_header_level(eb));
736 737 738
		ret = -EIO;
		goto err;
	}
739

740 741
	btrfs_set_buffer_lockdep_class(btrfs_header_owner(eb),
				       eb, found_level);
742

743
	ret = csum_tree_block(fs_info, eb, 1);
744
	if (ret)
745 746 747 748 749 750 751 752 753 754 755
		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;
	}
756

L
Liu Bo 已提交
757 758 759
	if (found_level > 0 && check_node(root, eb))
		ret = -EIO;

760 761
	if (!ret)
		set_extent_buffer_uptodate(eb);
762
err:
763 764
	if (reads_done &&
	    test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
765
		btree_readahead_hook(eb, ret);
A
Arne Jansen 已提交
766

D
David Woodhouse 已提交
767 768 769 770 771 772 773
	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);
774
		clear_extent_buffer_uptodate(eb);
D
David Woodhouse 已提交
775
	}
776
	free_extent_buffer(eb);
777
out:
778
	return ret;
779 780
}

781
static int btree_io_failed_hook(struct page *page, int failed_mirror)
A
Arne Jansen 已提交
782 783 784
{
	struct extent_buffer *eb;

J
Josef Bacik 已提交
785
	eb = (struct extent_buffer *)page->private;
786
	set_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags);
787
	eb->read_mirror = failed_mirror;
D
David Woodhouse 已提交
788
	atomic_dec(&eb->io_pages);
789
	if (test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
790
		btree_readahead_hook(eb, -EIO);
A
Arne Jansen 已提交
791 792 793
	return -EIO;	/* we fixed nothing */
}

794
static void end_workqueue_bio(struct bio *bio)
795
{
796
	struct btrfs_end_io_wq *end_io_wq = bio->bi_private;
797
	struct btrfs_fs_info *fs_info;
798 799
	struct btrfs_workqueue *wq;
	btrfs_work_func_t func;
800 801

	fs_info = end_io_wq->info;
802
	end_io_wq->error = bio->bi_error;
803

M
Mike Christie 已提交
804
	if (bio_op(bio) == REQ_OP_WRITE) {
805 806 807 808 809 810 811 812 813 814 815 816 817
		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;
		}
818
	} else {
819 820 821 822 823
		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) {
824 825 826 827 828 829 830 831 832
			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;
		}
833
	}
834 835 836

	btrfs_init_work(&end_io_wq->work, func, end_workqueue_fn, NULL, NULL);
	btrfs_queue_work(wq, &end_io_wq->work);
837 838
}

839
int btrfs_bio_wq_end_io(struct btrfs_fs_info *info, struct bio *bio,
840
			enum btrfs_wq_endio_type metadata)
841
{
842
	struct btrfs_end_io_wq *end_io_wq;
843

844
	end_io_wq = kmem_cache_alloc(btrfs_end_io_wq_cache, GFP_NOFS);
845 846 847 848 849
	if (!end_io_wq)
		return -ENOMEM;

	end_io_wq->private = bio->bi_private;
	end_io_wq->end_io = bio->bi_end_io;
850
	end_io_wq->info = info;
851 852
	end_io_wq->error = 0;
	end_io_wq->bio = bio;
853
	end_io_wq->metadata = metadata;
854 855 856

	bio->bi_private = end_io_wq;
	bio->bi_end_io = end_workqueue_bio;
857 858 859
	return 0;
}

860
unsigned long btrfs_async_submit_limit(struct btrfs_fs_info *info)
861
{
862
	unsigned long limit = min_t(unsigned long,
863
				    info->thread_pool_size,
864 865 866
				    info->fs_devices->open_devices);
	return 256 * limit;
}
867

C
Chris Mason 已提交
868 869 870
static void run_one_async_start(struct btrfs_work *work)
{
	struct async_submit_bio *async;
871
	int ret;
C
Chris Mason 已提交
872 873

	async = container_of(work, struct  async_submit_bio, work);
874
	ret = async->submit_bio_start(async->inode, async->bio,
875 876 877 878
				      async->mirror_num, async->bio_flags,
				      async->bio_offset);
	if (ret)
		async->error = ret;
C
Chris Mason 已提交
879 880 881
}

static void run_one_async_done(struct btrfs_work *work)
882 883 884
{
	struct btrfs_fs_info *fs_info;
	struct async_submit_bio *async;
885
	int limit;
886 887 888

	async = container_of(work, struct  async_submit_bio, work);
	fs_info = BTRFS_I(async->inode)->root->fs_info;
889

890
	limit = btrfs_async_submit_limit(fs_info);
891 892
	limit = limit * 2 / 3;

893 894 895
	/*
	 * atomic_dec_return implies a barrier for waitqueue_active
	 */
896
	if (atomic_dec_return(&fs_info->nr_async_submits) < limit &&
897
	    waitqueue_active(&fs_info->async_submit_wait))
898 899
		wake_up(&fs_info->async_submit_wait);

900
	/* If an error occurred we just want to clean up the bio and move on */
901
	if (async->error) {
902 903
		async->bio->bi_error = async->error;
		bio_endio(async->bio);
904 905 906
		return;
	}

907 908
	async->submit_bio_done(async->inode, async->bio, async->mirror_num,
			       async->bio_flags, async->bio_offset);
C
Chris Mason 已提交
909 910 911 912 913 914 915
}

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

	async = container_of(work, struct  async_submit_bio, work);
916 917 918
	kfree(async);
}

919
int btrfs_wq_submit_bio(struct btrfs_fs_info *fs_info, struct inode *inode,
920
			struct bio *bio, int mirror_num,
C
Chris Mason 已提交
921
			unsigned long bio_flags,
922
			u64 bio_offset,
C
Chris Mason 已提交
923 924
			extent_submit_bio_hook_t *submit_bio_start,
			extent_submit_bio_hook_t *submit_bio_done)
925 926 927 928 929 930 931 932 933 934
{
	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
935 936 937
	async->submit_bio_start = submit_bio_start;
	async->submit_bio_done = submit_bio_done;

938
	btrfs_init_work(&async->work, btrfs_worker_helper, run_one_async_start,
939
			run_one_async_done, run_one_async_free);
C
Chris Mason 已提交
940

C
Chris Mason 已提交
941
	async->bio_flags = bio_flags;
942
	async->bio_offset = bio_offset;
943

944 945
	async->error = 0;

946
	atomic_inc(&fs_info->nr_async_submits);
947

948
	if (op_is_sync(bio->bi_opf))
949
		btrfs_set_work_high_priority(&async->work);
950

951
	btrfs_queue_work(fs_info->workers, &async->work);
952

C
Chris Mason 已提交
953
	while (atomic_read(&fs_info->async_submit_draining) &&
954 955 956 957 958
	      atomic_read(&fs_info->nr_async_submits)) {
		wait_event(fs_info->async_submit_wait,
			   (atomic_read(&fs_info->nr_async_submits) == 0));
	}

959 960 961
	return 0;
}

962 963
static int btree_csum_one_bio(struct bio *bio)
{
964
	struct bio_vec *bvec;
965
	struct btrfs_root *root;
966
	int i, ret = 0;
967

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

975
	return ret;
976 977
}

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

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

995
	/*
C
Chris Mason 已提交
996 997
	 * when we're called for a write, we're already in the async
	 * submission context.  Just jump into btrfs_map_bio
998
	 */
999
	ret = btrfs_map_bio(btrfs_sb(inode->i_sb), bio, mirror_num, 1);
1000 1001 1002 1003
	if (ret) {
		bio->bi_error = ret;
		bio_endio(bio);
	}
1004
	return ret;
1005 1006
}

1007
static int check_async_write(unsigned long bio_flags)
1008 1009 1010 1011
{
	if (bio_flags & EXTENT_BIO_TREE_LOG)
		return 0;
#ifdef CONFIG_X86
1012
	if (static_cpu_has(X86_FEATURE_XMM4_2))
1013 1014 1015 1016 1017
		return 0;
#endif
	return 1;
}

1018
static int btree_submit_bio_hook(struct inode *inode, struct bio *bio,
1019 1020
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
1021
{
1022
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
1023
	int async = check_async_write(bio_flags);
1024 1025
	int ret;

M
Mike Christie 已提交
1026
	if (bio_op(bio) != REQ_OP_WRITE) {
C
Chris Mason 已提交
1027 1028 1029 1030
		/*
		 * called for a read, do the setup so that checksum validation
		 * can happen in the async kernel threads
		 */
1031 1032
		ret = btrfs_bio_wq_end_io(fs_info, bio,
					  BTRFS_WQ_ENDIO_METADATA);
1033
		if (ret)
1034
			goto out_w_error;
1035
		ret = btrfs_map_bio(fs_info, bio, mirror_num, 0);
1036 1037 1038
	} else if (!async) {
		ret = btree_csum_one_bio(bio);
		if (ret)
1039
			goto out_w_error;
1040
		ret = btrfs_map_bio(fs_info, bio, mirror_num, 0);
1041 1042 1043 1044 1045
	} else {
		/*
		 * kthread helpers are used to submit writes so that
		 * checksumming can happen in parallel across all CPUs
		 */
1046
		ret = btrfs_wq_submit_bio(fs_info, inode, bio, mirror_num, 0,
1047 1048 1049
					  bio_offset,
					  __btree_submit_bio_start,
					  __btree_submit_bio_done);
1050
	}
1051

1052 1053 1054 1055
	if (ret)
		goto out_w_error;
	return 0;

1056
out_w_error:
1057 1058
	bio->bi_error = ret;
	bio_endio(bio);
1059
	return ret;
1060 1061
}

J
Jan Beulich 已提交
1062
#ifdef CONFIG_MIGRATION
1063
static int btree_migratepage(struct address_space *mapping,
1064 1065
			struct page *newpage, struct page *page,
			enum migrate_mode mode)
1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079
{
	/*
	 * 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;
1080
	return migrate_page(mapping, newpage, page, mode);
1081
}
J
Jan Beulich 已提交
1082
#endif
1083

1084 1085 1086 1087

static int btree_writepages(struct address_space *mapping,
			    struct writeback_control *wbc)
{
1088 1089 1090
	struct btrfs_fs_info *fs_info;
	int ret;

1091
	if (wbc->sync_mode == WB_SYNC_NONE) {
1092 1093 1094 1095

		if (wbc->for_kupdate)
			return 0;

1096
		fs_info = BTRFS_I(mapping->host)->root->fs_info;
1097
		/* this is a bit racy, but that's ok */
1098 1099 1100
		ret = percpu_counter_compare(&fs_info->dirty_metadata_bytes,
					     BTRFS_DIRTY_METADATA_THRESH);
		if (ret < 0)
1101 1102
			return 0;
	}
1103
	return btree_write_cache_pages(mapping, wbc);
1104 1105
}

1106
static int btree_readpage(struct file *file, struct page *page)
1107
{
1108 1109
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
1110
	return extent_read_full_page(tree, page, btree_get_extent, 0);
1111
}
C
Chris Mason 已提交
1112

1113
static int btree_releasepage(struct page *page, gfp_t gfp_flags)
1114
{
1115
	if (PageWriteback(page) || PageDirty(page))
C
Chris Mason 已提交
1116
		return 0;
1117

1118
	return try_release_extent_buffer(page);
1119 1120
}

1121 1122
static void btree_invalidatepage(struct page *page, unsigned int offset,
				 unsigned int length)
1123
{
1124 1125
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
1126 1127
	extent_invalidatepage(tree, page, offset);
	btree_releasepage(page, GFP_NOFS);
1128
	if (PagePrivate(page)) {
1129 1130 1131
		btrfs_warn(BTRFS_I(page->mapping->host)->root->fs_info,
			   "page private not zero on page %llu",
			   (unsigned long long)page_offset(page));
1132 1133
		ClearPagePrivate(page);
		set_page_private(page, 0);
1134
		put_page(page);
1135
	}
1136 1137
}

1138 1139
static int btree_set_page_dirty(struct page *page)
{
1140
#ifdef DEBUG
1141 1142 1143 1144 1145 1146 1147 1148
	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);
1149
#endif
1150 1151 1152
	return __set_page_dirty_nobuffers(page);
}

1153
static const struct address_space_operations btree_aops = {
1154
	.readpage	= btree_readpage,
1155
	.writepages	= btree_writepages,
1156 1157
	.releasepage	= btree_releasepage,
	.invalidatepage = btree_invalidatepage,
1158
#ifdef CONFIG_MIGRATION
1159
	.migratepage	= btree_migratepage,
1160
#endif
1161
	.set_page_dirty = btree_set_page_dirty,
1162 1163
};

1164
void readahead_tree_block(struct btrfs_fs_info *fs_info, u64 bytenr)
C
Chris Mason 已提交
1165
{
1166
	struct extent_buffer *buf = NULL;
1167
	struct inode *btree_inode = fs_info->btree_inode;
C
Chris Mason 已提交
1168

1169
	buf = btrfs_find_create_tree_block(fs_info, bytenr);
1170
	if (IS_ERR(buf))
1171
		return;
1172
	read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree,
1173
				 buf, WAIT_NONE, btree_get_extent, 0);
1174
	free_extent_buffer(buf);
C
Chris Mason 已提交
1175 1176
}

1177
int reada_tree_block_flagged(struct btrfs_fs_info *fs_info, u64 bytenr,
1178 1179 1180
			 int mirror_num, struct extent_buffer **eb)
{
	struct extent_buffer *buf = NULL;
1181
	struct inode *btree_inode = fs_info->btree_inode;
1182 1183 1184
	struct extent_io_tree *io_tree = &BTRFS_I(btree_inode)->io_tree;
	int ret;

1185
	buf = btrfs_find_create_tree_block(fs_info, bytenr);
1186
	if (IS_ERR(buf))
1187 1188 1189 1190
		return 0;

	set_bit(EXTENT_BUFFER_READAHEAD, &buf->bflags);

1191
	ret = read_extent_buffer_pages(io_tree, buf, WAIT_PAGE_LOCK,
1192 1193 1194 1195 1196 1197 1198 1199 1200
				       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;
1201
	} else if (extent_buffer_uptodate(buf)) {
1202 1203 1204 1205 1206 1207 1208
		*eb = buf;
	} else {
		free_extent_buffer(buf);
	}
	return 0;
}

1209 1210 1211
struct extent_buffer *btrfs_find_create_tree_block(
						struct btrfs_fs_info *fs_info,
						u64 bytenr)
1212
{
1213 1214 1215
	if (btrfs_is_testing(fs_info))
		return alloc_test_extent_buffer(fs_info, bytenr);
	return alloc_extent_buffer(fs_info, bytenr);
1216 1217 1218
}


1219 1220
int btrfs_write_tree_block(struct extent_buffer *buf)
{
1221
	return filemap_fdatawrite_range(buf->pages[0]->mapping, buf->start,
1222
					buf->start + buf->len - 1);
1223 1224 1225 1226
}

int btrfs_wait_tree_block_writeback(struct extent_buffer *buf)
{
1227
	return filemap_fdatawait_range(buf->pages[0]->mapping,
1228
				       buf->start, buf->start + buf->len - 1);
1229 1230
}

1231
struct extent_buffer *read_tree_block(struct btrfs_fs_info *fs_info, u64 bytenr,
1232
				      u64 parent_transid)
1233 1234 1235 1236
{
	struct extent_buffer *buf = NULL;
	int ret;

1237
	buf = btrfs_find_create_tree_block(fs_info, bytenr);
1238 1239
	if (IS_ERR(buf))
		return buf;
1240

1241
	ret = btree_read_extent_buffer_pages(fs_info, buf, parent_transid);
1242 1243
	if (ret) {
		free_extent_buffer(buf);
1244
		return ERR_PTR(ret);
1245
	}
1246
	return buf;
1247

1248 1249
}

1250
void clean_tree_block(struct btrfs_fs_info *fs_info,
1251
		      struct extent_buffer *buf)
1252
{
1253
	if (btrfs_header_generation(buf) ==
1254
	    fs_info->running_transaction->transid) {
1255
		btrfs_assert_tree_locked(buf);
1256

1257
		if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &buf->bflags)) {
1258 1259 1260
			__percpu_counter_add(&fs_info->dirty_metadata_bytes,
					     -buf->len,
					     fs_info->dirty_metadata_batch);
1261 1262 1263 1264
			/* ugh, clear_extent_buffer_dirty needs to lock the page */
			btrfs_set_lock_blocking(buf);
			clear_extent_buffer_dirty(buf);
		}
1265
	}
1266 1267
}

1268 1269 1270 1271 1272 1273 1274 1275 1276
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);

1277
	ret = percpu_counter_init(&writers->counter, 0, GFP_KERNEL);
1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293
	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);
}

1294
static void __setup_root(struct btrfs_root *root, struct btrfs_fs_info *fs_info,
1295
			 u64 objectid)
1296
{
1297
	bool dummy = test_bit(BTRFS_FS_STATE_DUMMY_FS_INFO, &fs_info->fs_state);
C
Chris Mason 已提交
1298
	root->node = NULL;
1299
	root->commit_root = NULL;
1300
	root->state = 0;
1301
	root->orphan_cleanup_state = 0;
1302

1303 1304
	root->objectid = objectid;
	root->last_trans = 0;
1305
	root->highest_objectid = 0;
1306
	root->nr_delalloc_inodes = 0;
1307
	root->nr_ordered_extents = 0;
1308
	root->name = NULL;
1309
	root->inode_tree = RB_ROOT;
1310
	INIT_RADIX_TREE(&root->delayed_nodes_tree, GFP_ATOMIC);
1311
	root->block_rsv = NULL;
1312
	root->orphan_block_rsv = NULL;
1313 1314

	INIT_LIST_HEAD(&root->dirty_list);
1315
	INIT_LIST_HEAD(&root->root_list);
1316 1317
	INIT_LIST_HEAD(&root->delalloc_inodes);
	INIT_LIST_HEAD(&root->delalloc_root);
1318 1319
	INIT_LIST_HEAD(&root->ordered_extents);
	INIT_LIST_HEAD(&root->ordered_root);
1320 1321
	INIT_LIST_HEAD(&root->logged_list[0]);
	INIT_LIST_HEAD(&root->logged_list[1]);
1322
	spin_lock_init(&root->orphan_lock);
1323
	spin_lock_init(&root->inode_lock);
1324
	spin_lock_init(&root->delalloc_lock);
1325
	spin_lock_init(&root->ordered_extent_lock);
1326
	spin_lock_init(&root->accounting_lock);
1327 1328
	spin_lock_init(&root->log_extents_lock[0]);
	spin_lock_init(&root->log_extents_lock[1]);
1329
	mutex_init(&root->objectid_mutex);
1330
	mutex_init(&root->log_mutex);
1331
	mutex_init(&root->ordered_extent_mutex);
1332
	mutex_init(&root->delalloc_mutex);
Y
Yan Zheng 已提交
1333 1334 1335
	init_waitqueue_head(&root->log_writer_wait);
	init_waitqueue_head(&root->log_commit_wait[0]);
	init_waitqueue_head(&root->log_commit_wait[1]);
1336 1337
	INIT_LIST_HEAD(&root->log_ctxs[0]);
	INIT_LIST_HEAD(&root->log_ctxs[1]);
Y
Yan Zheng 已提交
1338 1339 1340
	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
1341
	atomic_set(&root->log_batch, 0);
1342
	atomic_set(&root->orphan_inodes, 0);
1343
	refcount_set(&root->refs, 1);
1344
	atomic_set(&root->will_be_snapshoted, 0);
1345
	atomic64_set(&root->qgroup_meta_rsv, 0);
Y
Yan Zheng 已提交
1346
	root->log_transid = 0;
1347
	root->log_transid_committed = -1;
1348
	root->last_log_commit = 0;
1349
	if (!dummy)
1350 1351
		extent_io_tree_init(&root->dirty_log_pages,
				     fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
1352

1353 1354
	memset(&root->root_key, 0, sizeof(root->root_key));
	memset(&root->root_item, 0, sizeof(root->root_item));
1355
	memset(&root->defrag_progress, 0, sizeof(root->defrag_progress));
1356
	if (!dummy)
1357 1358 1359
		root->defrag_trans_start = fs_info->generation;
	else
		root->defrag_trans_start = 0;
1360
	root->root_key.objectid = objectid;
1361
	root->anon_dev = 0;
1362

1363
	spin_lock_init(&root->root_item_lock);
1364 1365
}

1366 1367
static struct btrfs_root *btrfs_alloc_root(struct btrfs_fs_info *fs_info,
		gfp_t flags)
A
Al Viro 已提交
1368
{
1369
	struct btrfs_root *root = kzalloc(sizeof(*root), flags);
A
Al Viro 已提交
1370 1371 1372 1373 1374
	if (root)
		root->fs_info = fs_info;
	return root;
}

1375 1376
#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
/* Should only be used by the testing infrastructure */
1377
struct btrfs_root *btrfs_alloc_dummy_root(struct btrfs_fs_info *fs_info)
1378 1379 1380
{
	struct btrfs_root *root;

1381 1382 1383 1384
	if (!fs_info)
		return ERR_PTR(-EINVAL);

	root = btrfs_alloc_root(fs_info, GFP_KERNEL);
1385 1386
	if (!root)
		return ERR_PTR(-ENOMEM);
1387

1388
	/* We don't use the stripesize in selftest, set it as sectorsize */
1389
	__setup_root(root, fs_info, BTRFS_ROOT_TREE_OBJECTID);
1390
	root->alloc_bytenr = 0;
1391 1392 1393 1394 1395

	return root;
}
#endif

1396 1397 1398 1399 1400 1401 1402 1403 1404
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;
1405
	uuid_le uuid;
1406

1407
	root = btrfs_alloc_root(fs_info, GFP_KERNEL);
1408 1409 1410
	if (!root)
		return ERR_PTR(-ENOMEM);

1411
	__setup_root(root, fs_info, objectid);
1412 1413 1414 1415
	root->root_key.objectid = objectid;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = 0;

1416
	leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0);
1417 1418
	if (IS_ERR(leaf)) {
		ret = PTR_ERR(leaf);
1419
		leaf = NULL;
1420 1421 1422
		goto fail;
	}

1423
	memzero_extent_buffer(leaf, 0, sizeof(struct btrfs_header));
1424 1425 1426 1427 1428 1429
	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;

1430 1431
	write_extent_buffer_fsid(leaf, fs_info->fsid);
	write_extent_buffer_chunk_tree_uuid(leaf, fs_info->chunk_tree_uuid);
1432 1433 1434
	btrfs_mark_buffer_dirty(leaf);

	root->commit_root = btrfs_root_node(root);
1435
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
1436 1437 1438 1439 1440 1441 1442 1443 1444 1445

	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);
1446 1447
	uuid_le_gen(&uuid);
	memcpy(root->root_item.uuid, uuid.b, BTRFS_UUID_SIZE);
1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458
	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);

1459 1460
	return root;

1461
fail:
1462 1463
	if (leaf) {
		btrfs_tree_unlock(leaf);
1464
		free_extent_buffer(root->commit_root);
1465 1466 1467
		free_extent_buffer(leaf);
	}
	kfree(root);
1468

1469
	return ERR_PTR(ret);
1470 1471
}

Y
Yan Zheng 已提交
1472 1473
static struct btrfs_root *alloc_log_tree(struct btrfs_trans_handle *trans,
					 struct btrfs_fs_info *fs_info)
1474 1475
{
	struct btrfs_root *root;
Y
Yan Zheng 已提交
1476
	struct extent_buffer *leaf;
1477

1478
	root = btrfs_alloc_root(fs_info, GFP_NOFS);
1479
	if (!root)
Y
Yan Zheng 已提交
1480
		return ERR_PTR(-ENOMEM);
1481

1482
	__setup_root(root, fs_info, BTRFS_TREE_LOG_OBJECTID);
1483 1484 1485 1486

	root->root_key.objectid = BTRFS_TREE_LOG_OBJECTID;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = BTRFS_TREE_LOG_OBJECTID;
1487

Y
Yan Zheng 已提交
1488
	/*
1489 1490
	 * DON'T set REF_COWS for log trees
	 *
Y
Yan Zheng 已提交
1491 1492 1493 1494 1495
	 * 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).
	 */
1496

1497 1498
	leaf = btrfs_alloc_tree_block(trans, root, 0, BTRFS_TREE_LOG_OBJECTID,
			NULL, 0, 0, 0);
Y
Yan Zheng 已提交
1499 1500 1501 1502
	if (IS_ERR(leaf)) {
		kfree(root);
		return ERR_CAST(leaf);
	}
1503

1504
	memzero_extent_buffer(leaf, 0, sizeof(struct btrfs_header));
1505 1506 1507 1508
	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
1509
	root->node = leaf;
1510

1511
	write_extent_buffer_fsid(root->node, fs_info->fsid);
1512 1513
	btrfs_mark_buffer_dirty(root->node);
	btrfs_tree_unlock(root->node);
Y
Yan Zheng 已提交
1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532
	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)
{
1533
	struct btrfs_fs_info *fs_info = root->fs_info;
Y
Yan Zheng 已提交
1534 1535 1536
	struct btrfs_root *log_root;
	struct btrfs_inode_item *inode_item;

1537
	log_root = alloc_log_tree(trans, fs_info);
Y
Yan Zheng 已提交
1538 1539 1540 1541 1542 1543 1544
	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;
1545 1546 1547
	btrfs_set_stack_inode_generation(inode_item, 1);
	btrfs_set_stack_inode_size(inode_item, 3);
	btrfs_set_stack_inode_nlink(inode_item, 1);
1548
	btrfs_set_stack_inode_nbytes(inode_item,
1549
				     fs_info->nodesize);
1550
	btrfs_set_stack_inode_mode(inode_item, S_IFDIR | 0755);
Y
Yan Zheng 已提交
1551

1552
	btrfs_set_root_node(&log_root->root_item, log_root->node);
Y
Yan Zheng 已提交
1553 1554 1555 1556

	WARN_ON(root->log_root);
	root->log_root = log_root;
	root->log_transid = 0;
1557
	root->log_transid_committed = -1;
1558
	root->last_log_commit = 0;
1559 1560 1561
	return 0;
}

1562 1563
static struct btrfs_root *btrfs_read_tree_root(struct btrfs_root *tree_root,
					       struct btrfs_key *key)
1564 1565 1566
{
	struct btrfs_root *root;
	struct btrfs_fs_info *fs_info = tree_root->fs_info;
1567
	struct btrfs_path *path;
1568
	u64 generation;
1569
	int ret;
1570

1571 1572
	path = btrfs_alloc_path();
	if (!path)
1573
		return ERR_PTR(-ENOMEM);
1574

1575
	root = btrfs_alloc_root(fs_info, GFP_NOFS);
1576 1577 1578
	if (!root) {
		ret = -ENOMEM;
		goto alloc_fail;
1579 1580
	}

1581
	__setup_root(root, fs_info, key->objectid);
1582

1583 1584
	ret = btrfs_find_root(tree_root, key, path,
			      &root->root_item, &root->root_key);
1585
	if (ret) {
1586 1587
		if (ret > 0)
			ret = -ENOENT;
1588
		goto find_fail;
1589
	}
1590

1591
	generation = btrfs_root_generation(&root->root_item);
1592 1593
	root->node = read_tree_block(fs_info,
				     btrfs_root_bytenr(&root->root_item),
1594
				     generation);
1595 1596
	if (IS_ERR(root->node)) {
		ret = PTR_ERR(root->node);
1597 1598 1599
		goto find_fail;
	} else if (!btrfs_buffer_uptodate(root->node, generation, 0)) {
		ret = -EIO;
1600 1601
		free_extent_buffer(root->node);
		goto find_fail;
1602
	}
1603
	root->commit_root = btrfs_root_node(root);
1604
out:
1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624
	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) {
1625
		set_bit(BTRFS_ROOT_REF_COWS, &root->state);
1626 1627
		btrfs_check_and_init_root_item(&root->root_item);
	}
1628

1629 1630 1631
	return root;
}

1632 1633 1634
int btrfs_init_fs_root(struct btrfs_root *root)
{
	int ret;
1635
	struct btrfs_subvolume_writers *writers;
1636 1637 1638 1639 1640 1641 1642 1643 1644

	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;
	}

1645 1646 1647 1648 1649 1650 1651
	writers = btrfs_alloc_subvolume_writers();
	if (IS_ERR(writers)) {
		ret = PTR_ERR(writers);
		goto fail;
	}
	root->subv_writers = writers;

1652
	btrfs_init_free_ino_ctl(root);
1653 1654
	spin_lock_init(&root->ino_cache_lock);
	init_waitqueue_head(&root->ino_cache_wait);
1655 1656 1657

	ret = get_anon_bdev(&root->anon_dev);
	if (ret)
L
Liu Bo 已提交
1658
		goto fail;
1659 1660 1661 1662 1663 1664

	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
1665
		goto fail;
1666 1667 1668 1669 1670 1671
	}

	ASSERT(root->highest_objectid <= BTRFS_LAST_FREE_OBJECTID);

	mutex_unlock(&root->objectid_mutex);

1672 1673
	return 0;
fail:
L
Liu Bo 已提交
1674
	/* the caller is responsible to call free_fs_root */
1675 1676 1677
	return ret;
}

1678 1679
struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info,
					u64 root_id)
1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694
{
	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;

1695
	ret = radix_tree_preload(GFP_NOFS);
1696 1697 1698 1699 1700 1701 1702 1703
	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)
1704
		set_bit(BTRFS_ROOT_IN_RADIX, &root->state);
1705 1706 1707 1708 1709 1710
	spin_unlock(&fs_info->fs_roots_radix_lock);
	radix_tree_preload_end();

	return ret;
}

1711 1712 1713
struct btrfs_root *btrfs_get_fs_root(struct btrfs_fs_info *fs_info,
				     struct btrfs_key *location,
				     bool check_ref)
1714 1715
{
	struct btrfs_root *root;
1716
	struct btrfs_path *path;
1717
	struct btrfs_key key;
1718 1719
	int ret;

1720 1721 1722 1723
	if (location->objectid == BTRFS_ROOT_TREE_OBJECTID)
		return fs_info->tree_root;
	if (location->objectid == BTRFS_EXTENT_TREE_OBJECTID)
		return fs_info->extent_root;
1724 1725 1726 1727
	if (location->objectid == BTRFS_CHUNK_TREE_OBJECTID)
		return fs_info->chunk_root;
	if (location->objectid == BTRFS_DEV_TREE_OBJECTID)
		return fs_info->dev_root;
1728 1729
	if (location->objectid == BTRFS_CSUM_TREE_OBJECTID)
		return fs_info->csum_root;
1730 1731 1732
	if (location->objectid == BTRFS_QUOTA_TREE_OBJECTID)
		return fs_info->quota_root ? fs_info->quota_root :
					     ERR_PTR(-ENOENT);
1733 1734 1735
	if (location->objectid == BTRFS_UUID_TREE_OBJECTID)
		return fs_info->uuid_root ? fs_info->uuid_root :
					    ERR_PTR(-ENOENT);
1736 1737 1738
	if (location->objectid == BTRFS_FREE_SPACE_TREE_OBJECTID)
		return fs_info->free_space_root ? fs_info->free_space_root :
						  ERR_PTR(-ENOENT);
1739
again:
1740
	root = btrfs_lookup_fs_root(fs_info, location->objectid);
1741
	if (root) {
1742
		if (check_ref && btrfs_root_refs(&root->root_item) == 0)
1743
			return ERR_PTR(-ENOENT);
1744
		return root;
1745
	}
1746

1747
	root = btrfs_read_fs_root(fs_info->tree_root, location);
1748 1749
	if (IS_ERR(root))
		return root;
1750

1751
	if (check_ref && btrfs_root_refs(&root->root_item) == 0) {
1752
		ret = -ENOENT;
1753
		goto fail;
1754
	}
1755

1756
	ret = btrfs_init_fs_root(root);
1757 1758
	if (ret)
		goto fail;
1759

1760 1761 1762 1763 1764
	path = btrfs_alloc_path();
	if (!path) {
		ret = -ENOMEM;
		goto fail;
	}
1765 1766 1767 1768 1769
	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);
1770
	btrfs_free_path(path);
1771 1772 1773
	if (ret < 0)
		goto fail;
	if (ret == 0)
1774
		set_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &root->state);
1775

1776
	ret = btrfs_insert_fs_root(fs_info, root);
1777
	if (ret) {
1778 1779 1780 1781 1782
		if (ret == -EEXIST) {
			free_fs_root(root);
			goto again;
		}
		goto fail;
1783
	}
1784
	return root;
1785 1786 1787
fail:
	free_fs_root(root);
	return ERR_PTR(ret);
1788 1789
}

C
Chris Mason 已提交
1790 1791 1792 1793 1794 1795
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
1796

1797 1798
	rcu_read_lock();
	list_for_each_entry_rcu(device, &info->fs_devices->devices, dev_list) {
1799 1800
		if (!device->bdev)
			continue;
1801
		bdi = device->bdev->bd_bdi;
1802
		if (bdi_congested(bdi, bdi_bits)) {
C
Chris Mason 已提交
1803 1804 1805 1806
			ret = 1;
			break;
		}
	}
1807
	rcu_read_unlock();
C
Chris Mason 已提交
1808 1809 1810
	return ret;
}

1811 1812 1813 1814 1815
/*
 * 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)
1816 1817
{
	struct bio *bio;
1818
	struct btrfs_end_io_wq *end_io_wq;
1819

1820
	end_io_wq = container_of(work, struct btrfs_end_io_wq, work);
1821
	bio = end_io_wq->bio;
1822

1823
	bio->bi_error = end_io_wq->error;
1824 1825
	bio->bi_private = end_io_wq->private;
	bio->bi_end_io = end_io_wq->end_io;
1826
	kmem_cache_free(btrfs_end_io_wq_cache, end_io_wq);
1827
	bio_endio(bio);
1828 1829
}

1830 1831 1832
static int cleaner_kthread(void *arg)
{
	struct btrfs_root *root = arg;
1833
	struct btrfs_fs_info *fs_info = root->fs_info;
1834
	int again;
1835
	struct btrfs_trans_handle *trans;
1836 1837

	do {
1838
		again = 0;
1839

1840
		/* Make the cleaner go to sleep early. */
1841
		if (btrfs_need_cleaner_sleep(fs_info))
1842 1843
			goto sleep;

1844 1845 1846 1847
		/*
		 * Do not do anything if we might cause open_ctree() to block
		 * before we have finished mounting the filesystem.
		 */
1848
		if (!test_bit(BTRFS_FS_OPEN, &fs_info->flags))
1849 1850
			goto sleep;

1851
		if (!mutex_trylock(&fs_info->cleaner_mutex))
1852 1853
			goto sleep;

1854 1855 1856 1857
		/*
		 * Avoid the problem that we change the status of the fs
		 * during the above check and trylock.
		 */
1858
		if (btrfs_need_cleaner_sleep(fs_info)) {
1859
			mutex_unlock(&fs_info->cleaner_mutex);
1860
			goto sleep;
1861
		}
1862

1863
		mutex_lock(&fs_info->cleaner_delayed_iput_mutex);
1864
		btrfs_run_delayed_iputs(fs_info);
1865
		mutex_unlock(&fs_info->cleaner_delayed_iput_mutex);
1866

1867
		again = btrfs_clean_one_deleted_snapshot(root);
1868
		mutex_unlock(&fs_info->cleaner_mutex);
1869 1870

		/*
1871 1872
		 * The defragger has dealt with the R/O remount and umount,
		 * needn't do anything special here.
1873
		 */
1874
		btrfs_run_defrag_inodes(fs_info);
1875 1876 1877 1878 1879 1880 1881 1882 1883

		/*
		 * 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.
		 */
1884
		btrfs_delete_unused_bgs(fs_info);
1885
sleep:
1886
		if (!again) {
1887
			set_current_state(TASK_INTERRUPTIBLE);
1888 1889
			if (!kthread_should_stop())
				schedule();
1890 1891 1892
			__set_current_state(TASK_RUNNING);
		}
	} while (!kthread_should_stop());
1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907

	/*
	 * 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)
1908
			btrfs_err(fs_info,
1909 1910 1911 1912 1913
				  "cleaner transaction attach returned %ld",
				  PTR_ERR(trans));
	} else {
		int ret;

1914
		ret = btrfs_commit_transaction(trans);
1915
		if (ret)
1916
			btrfs_err(fs_info,
1917 1918 1919 1920
				  "cleaner open transaction commit returned %d",
				  ret);
	}

1921 1922 1923 1924 1925 1926
	return 0;
}

static int transaction_kthread(void *arg)
{
	struct btrfs_root *root = arg;
1927
	struct btrfs_fs_info *fs_info = root->fs_info;
1928 1929
	struct btrfs_trans_handle *trans;
	struct btrfs_transaction *cur;
1930
	u64 transid;
1931 1932
	unsigned long now;
	unsigned long delay;
1933
	bool cannot_commit;
1934 1935

	do {
1936
		cannot_commit = false;
1937 1938
		delay = HZ * fs_info->commit_interval;
		mutex_lock(&fs_info->transaction_kthread_mutex);
1939

1940 1941
		spin_lock(&fs_info->trans_lock);
		cur = fs_info->running_transaction;
1942
		if (!cur) {
1943
			spin_unlock(&fs_info->trans_lock);
1944 1945
			goto sleep;
		}
Y
Yan Zheng 已提交
1946

1947
		now = get_seconds();
1948
		if (cur->state < TRANS_STATE_BLOCKED &&
1949
		    (now < cur->start_time ||
1950 1951
		     now - cur->start_time < fs_info->commit_interval)) {
			spin_unlock(&fs_info->trans_lock);
1952 1953 1954
			delay = HZ * 5;
			goto sleep;
		}
1955
		transid = cur->transid;
1956
		spin_unlock(&fs_info->trans_lock);
1957

1958
		/* If the file system is aborted, this will always fail. */
1959
		trans = btrfs_attach_transaction(root);
1960
		if (IS_ERR(trans)) {
1961 1962
			if (PTR_ERR(trans) != -ENOENT)
				cannot_commit = true;
1963
			goto sleep;
1964
		}
1965
		if (transid == trans->transid) {
1966
			btrfs_commit_transaction(trans);
1967
		} else {
1968
			btrfs_end_transaction(trans);
1969
		}
1970
sleep:
1971 1972
		wake_up_process(fs_info->cleaner_kthread);
		mutex_unlock(&fs_info->transaction_kthread_mutex);
1973

J
Josef Bacik 已提交
1974
		if (unlikely(test_bit(BTRFS_FS_STATE_ERROR,
1975
				      &fs_info->fs_state)))
1976
			btrfs_cleanup_transaction(fs_info);
1977 1978
		set_current_state(TASK_INTERRUPTIBLE);
		if (!kthread_should_stop() &&
1979
				(!btrfs_transaction_blocked(fs_info) ||
1980 1981 1982
				 cannot_commit))
			schedule_timeout(delay);
		__set_current_state(TASK_RUNNING);
1983 1984 1985 1986
	} while (!kthread_should_stop());
	return 0;
}

C
Chris Mason 已提交
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 2086 2087 2088 2089 2090 2091 2092
/*
 * 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));

2093 2094 2095 2096 2097 2098 2099 2100
	/*
	 * 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
2101
			       btrfs_header_generation(info->fs_root->node));
2102
		btrfs_set_backup_fs_root_level(root_backup,
C
Chris Mason 已提交
2103
			       btrfs_header_level(info->fs_root->node));
2104
	}
C
Chris Mason 已提交
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 2179 2180 2181 2182 2183 2184 2185

	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
2186 2187 2188
/* helper to cleanup workers */
static void btrfs_stop_all_workers(struct btrfs_fs_info *fs_info)
{
2189
	btrfs_destroy_workqueue(fs_info->fixup_workers);
2190
	btrfs_destroy_workqueue(fs_info->delalloc_workers);
2191
	btrfs_destroy_workqueue(fs_info->workers);
2192 2193
	btrfs_destroy_workqueue(fs_info->endio_workers);
	btrfs_destroy_workqueue(fs_info->endio_raid56_workers);
2194
	btrfs_destroy_workqueue(fs_info->endio_repair_workers);
2195
	btrfs_destroy_workqueue(fs_info->rmw_workers);
2196 2197
	btrfs_destroy_workqueue(fs_info->endio_write_workers);
	btrfs_destroy_workqueue(fs_info->endio_freespace_worker);
2198
	btrfs_destroy_workqueue(fs_info->submit_workers);
2199
	btrfs_destroy_workqueue(fs_info->delayed_workers);
2200
	btrfs_destroy_workqueue(fs_info->caching_workers);
2201
	btrfs_destroy_workqueue(fs_info->readahead_workers);
2202
	btrfs_destroy_workqueue(fs_info->flush_workers);
2203
	btrfs_destroy_workqueue(fs_info->qgroup_rescan_workers);
C
Chris Mason 已提交
2204
	btrfs_destroy_workqueue(fs_info->extent_workers);
2205 2206 2207 2208 2209 2210 2211
	/*
	 * Now that all other work queues are destroyed, we can safely destroy
	 * the queues used for metadata I/O, since tasks from those other work
	 * queues can do metadata I/O operations.
	 */
	btrfs_destroy_workqueue(fs_info->endio_meta_workers);
	btrfs_destroy_workqueue(fs_info->endio_meta_write_workers);
L
Liu Bo 已提交
2212 2213
}

2214 2215 2216 2217 2218 2219 2220 2221 2222 2223
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
2224 2225 2226
/* helper to cleanup tree roots */
static void free_root_pointers(struct btrfs_fs_info *info, int chunk_root)
{
2227
	free_root_extent_buffers(info->tree_root);
2228

2229 2230 2231 2232 2233 2234 2235
	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);
2236
	free_root_extent_buffers(info->free_space_root);
C
Chris Mason 已提交
2237 2238
}

2239
void btrfs_free_fs_roots(struct btrfs_fs_info *fs_info)
2240 2241 2242 2243 2244 2245 2246 2247 2248 2249
{
	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);

2250
		if (test_bit(BTRFS_ROOT_IN_RADIX, &gang[0]->state)) {
2251
			btrfs_drop_and_free_fs_root(fs_info, gang[0]);
2252 2253 2254
		} else {
			free_extent_buffer(gang[0]->node);
			free_extent_buffer(gang[0]->commit_root);
2255
			btrfs_put_fs_root(gang[0]);
2256 2257 2258 2259 2260 2261 2262 2263 2264 2265
		}
	}

	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++)
2266
			btrfs_drop_and_free_fs_root(fs_info, gang[i]);
2267
	}
2268 2269 2270

	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
		btrfs_free_log_root_tree(NULL, fs_info);
2271
		btrfs_destroy_pinned_extent(fs_info, fs_info->pinned_extents);
2272
	}
2273
}
C
Chris Mason 已提交
2274

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

2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296
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);
}

2297
static void btrfs_init_btree_inode(struct btrfs_fs_info *fs_info)
2298
{
2299 2300 2301 2302
	struct inode *inode = fs_info->btree_inode;

	inode->i_ino = BTRFS_BTREE_INODE_OBJECTID;
	set_nlink(inode, 1);
2303 2304 2305 2306 2307
	/*
	 * 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
	 */
2308 2309
	inode->i_size = OFFSET_MAX;
	inode->i_mapping->a_ops = &btree_aops;
2310

2311 2312 2313 2314
	RB_CLEAR_NODE(&BTRFS_I(inode)->rb_node);
	extent_io_tree_init(&BTRFS_I(inode)->io_tree, inode->i_mapping);
	BTRFS_I(inode)->io_tree.track_uptodate = 0;
	extent_map_tree_init(&BTRFS_I(inode)->extent_tree);
2315

2316
	BTRFS_I(inode)->io_tree.ops = &btree_extent_io_ops;
2317

2318 2319 2320 2321
	BTRFS_I(inode)->root = fs_info->tree_root;
	memset(&BTRFS_I(inode)->location, 0, sizeof(struct btrfs_key));
	set_bit(BTRFS_INODE_DUMMY, &BTRFS_I(inode)->runtime_flags);
	btrfs_insert_inode_hash(inode);
2322 2323
}

2324 2325 2326 2327 2328
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);
2329 2330 2331
	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);
2332
	init_waitqueue_head(&fs_info->replace_wait);
2333
	init_waitqueue_head(&fs_info->dev_replace.read_lock_wq);
2334 2335
}

2336 2337 2338 2339 2340 2341 2342 2343 2344
static void btrfs_init_qgroup(struct btrfs_fs_info *fs_info)
{
	spin_lock_init(&fs_info->qgroup_lock);
	mutex_init(&fs_info->qgroup_ioctl_lock);
	fs_info->qgroup_tree = RB_ROOT;
	fs_info->qgroup_op_tree = RB_ROOT;
	INIT_LIST_HEAD(&fs_info->dirty_qgroups);
	fs_info->qgroup_seq = 1;
	fs_info->qgroup_ulist = NULL;
2345
	fs_info->qgroup_rescan_running = false;
2346 2347 2348
	mutex_init(&fs_info->qgroup_rescan_lock);
}

2349 2350 2351 2352
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;
2353
	unsigned int flags = WQ_MEM_RECLAIM | WQ_FREEZABLE | WQ_UNBOUND;
2354 2355

	fs_info->workers =
2356 2357
		btrfs_alloc_workqueue(fs_info, "worker",
				      flags | WQ_HIGHPRI, max_active, 16);
2358 2359

	fs_info->delalloc_workers =
2360 2361
		btrfs_alloc_workqueue(fs_info, "delalloc",
				      flags, max_active, 2);
2362 2363

	fs_info->flush_workers =
2364 2365
		btrfs_alloc_workqueue(fs_info, "flush_delalloc",
				      flags, max_active, 0);
2366 2367

	fs_info->caching_workers =
2368
		btrfs_alloc_workqueue(fs_info, "cache", flags, max_active, 0);
2369 2370 2371 2372 2373 2374 2375

	/*
	 * 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 =
2376
		btrfs_alloc_workqueue(fs_info, "submit", flags,
2377 2378 2379 2380
				      min_t(u64, fs_devices->num_devices,
					    max_active), 64);

	fs_info->fixup_workers =
2381
		btrfs_alloc_workqueue(fs_info, "fixup", flags, 1, 0);
2382 2383 2384 2385 2386 2387

	/*
	 * endios are largely parallel and should have a very
	 * low idle thresh
	 */
	fs_info->endio_workers =
2388
		btrfs_alloc_workqueue(fs_info, "endio", flags, max_active, 4);
2389
	fs_info->endio_meta_workers =
2390 2391
		btrfs_alloc_workqueue(fs_info, "endio-meta", flags,
				      max_active, 4);
2392
	fs_info->endio_meta_write_workers =
2393 2394
		btrfs_alloc_workqueue(fs_info, "endio-meta-write", flags,
				      max_active, 2);
2395
	fs_info->endio_raid56_workers =
2396 2397
		btrfs_alloc_workqueue(fs_info, "endio-raid56", flags,
				      max_active, 4);
2398
	fs_info->endio_repair_workers =
2399
		btrfs_alloc_workqueue(fs_info, "endio-repair", flags, 1, 0);
2400
	fs_info->rmw_workers =
2401
		btrfs_alloc_workqueue(fs_info, "rmw", flags, max_active, 2);
2402
	fs_info->endio_write_workers =
2403 2404
		btrfs_alloc_workqueue(fs_info, "endio-write", flags,
				      max_active, 2);
2405
	fs_info->endio_freespace_worker =
2406 2407
		btrfs_alloc_workqueue(fs_info, "freespace-write", flags,
				      max_active, 0);
2408
	fs_info->delayed_workers =
2409 2410
		btrfs_alloc_workqueue(fs_info, "delayed-meta", flags,
				      max_active, 0);
2411
	fs_info->readahead_workers =
2412 2413
		btrfs_alloc_workqueue(fs_info, "readahead", flags,
				      max_active, 2);
2414
	fs_info->qgroup_rescan_workers =
2415
		btrfs_alloc_workqueue(fs_info, "qgroup-rescan", flags, 1, 0);
2416
	fs_info->extent_workers =
2417
		btrfs_alloc_workqueue(fs_info, "extent-refs", flags,
2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437
				      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;
}

2438 2439 2440 2441 2442 2443 2444 2445 2446
static int btrfs_replay_log(struct btrfs_fs_info *fs_info,
			    struct btrfs_fs_devices *fs_devices)
{
	int ret;
	struct btrfs_root *log_tree_root;
	struct btrfs_super_block *disk_super = fs_info->super_copy;
	u64 bytenr = btrfs_super_log_root(disk_super);

	if (fs_devices->rw_devices == 0) {
2447
		btrfs_warn(fs_info, "log replay required on RO media");
2448 2449 2450
		return -EIO;
	}

2451
	log_tree_root = btrfs_alloc_root(fs_info, GFP_KERNEL);
2452 2453 2454
	if (!log_tree_root)
		return -ENOMEM;

2455
	__setup_root(log_tree_root, fs_info, BTRFS_TREE_LOG_OBJECTID);
2456

2457 2458
	log_tree_root->node = read_tree_block(fs_info, bytenr,
					      fs_info->generation + 1);
2459
	if (IS_ERR(log_tree_root->node)) {
2460
		btrfs_warn(fs_info, "failed to read log tree");
2461
		ret = PTR_ERR(log_tree_root->node);
2462
		kfree(log_tree_root);
2463
		return ret;
2464
	} else if (!extent_buffer_uptodate(log_tree_root->node)) {
2465
		btrfs_err(fs_info, "failed to read log tree");
2466 2467 2468 2469 2470 2471 2472
		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) {
2473 2474
		btrfs_handle_fs_error(fs_info, ret,
				      "Failed to recover log tree");
2475 2476 2477 2478 2479 2480
		free_extent_buffer(log_tree_root->node);
		kfree(log_tree_root);
		return ret;
	}

	if (fs_info->sb->s_flags & MS_RDONLY) {
2481
		ret = btrfs_commit_super(fs_info);
2482 2483 2484 2485 2486 2487 2488
		if (ret)
			return ret;
	}

	return 0;
}

2489
static int btrfs_read_roots(struct btrfs_fs_info *fs_info)
2490
{
2491
	struct btrfs_root *tree_root = fs_info->tree_root;
2492
	struct btrfs_root *root;
2493 2494 2495
	struct btrfs_key location;
	int ret;

2496 2497
	BUG_ON(!fs_info->tree_root);

2498 2499 2500 2501
	location.objectid = BTRFS_EXTENT_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
	location.offset = 0;

2502 2503 2504 2505 2506
	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;
2507 2508

	location.objectid = BTRFS_DEV_TREE_OBJECTID;
2509 2510 2511 2512 2513
	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;
2514 2515 2516
	btrfs_init_devices_late(fs_info);

	location.objectid = BTRFS_CSUM_TREE_OBJECTID;
2517 2518 2519 2520 2521
	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;
2522 2523

	location.objectid = BTRFS_QUOTA_TREE_OBJECTID;
2524 2525 2526
	root = btrfs_read_tree_root(tree_root, &location);
	if (!IS_ERR(root)) {
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
2527
		set_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags);
2528
		fs_info->quota_root = root;
2529 2530 2531
	}

	location.objectid = BTRFS_UUID_TREE_OBJECTID;
2532 2533 2534
	root = btrfs_read_tree_root(tree_root, &location);
	if (IS_ERR(root)) {
		ret = PTR_ERR(root);
2535 2536 2537
		if (ret != -ENOENT)
			return ret;
	} else {
2538 2539
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
		fs_info->uuid_root = root;
2540 2541
	}

2542 2543 2544 2545 2546 2547 2548 2549 2550
	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;
	}

2551 2552 2553
	return 0;
}

A
Al Viro 已提交
2554 2555 2556
int open_ctree(struct super_block *sb,
	       struct btrfs_fs_devices *fs_devices,
	       char *options)
2557
{
2558 2559
	u32 sectorsize;
	u32 nodesize;
2560
	u32 stripesize;
2561
	u64 generation;
2562
	u64 features;
2563
	struct btrfs_key location;
2564
	struct buffer_head *bh;
2565
	struct btrfs_super_block *disk_super;
2566
	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
2567
	struct btrfs_root *tree_root;
2568
	struct btrfs_root *chunk_root;
2569
	int ret;
2570
	int err = -EINVAL;
C
Chris Mason 已提交
2571 2572
	int num_backups_tried = 0;
	int backup_index = 0;
2573
	int max_active;
2574
	int clear_free_space_tree = 0;
2575

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

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

2589
	ret = percpu_counter_init(&fs_info->dirty_metadata_bytes, 0, GFP_KERNEL);
2590 2591
	if (ret) {
		err = ret;
2592
		goto fail_srcu;
2593
	}
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 2700
	sb->s_blocksize = 4096;
	sb->s_blocksize_bits = blksize_bits(4096);

2701
	btrfs_init_btree_inode(fs_info);
2702

J
Josef Bacik 已提交
2703
	spin_lock_init(&fs_info->block_group_cache_lock);
2704
	fs_info->block_group_cache_tree = RB_ROOT;
2705
	fs_info->first_logical_byte = (u64)-1;
J
Josef Bacik 已提交
2706

2707
	extent_io_tree_init(&fs_info->freed_extents[0],
2708
			     fs_info->btree_inode->i_mapping);
2709
	extent_io_tree_init(&fs_info->freed_extents[1],
2710
			     fs_info->btree_inode->i_mapping);
2711
	fs_info->pinned_extents = &fs_info->freed_extents[0];
2712
	set_bit(BTRFS_FS_BARRIER, &fs_info->flags);
C
Chris Mason 已提交
2713

2714
	mutex_init(&fs_info->ordered_operations_mutex);
2715
	mutex_init(&fs_info->tree_log_mutex);
2716
	mutex_init(&fs_info->chunk_mutex);
2717 2718
	mutex_init(&fs_info->transaction_kthread_mutex);
	mutex_init(&fs_info->cleaner_mutex);
2719
	mutex_init(&fs_info->volume_mutex);
2720
	mutex_init(&fs_info->ro_block_group_mutex);
2721
	init_rwsem(&fs_info->commit_root_sem);
2722
	init_rwsem(&fs_info->cleanup_work_sem);
2723
	init_rwsem(&fs_info->subvol_sem);
S
Stefan Behrens 已提交
2724
	sema_init(&fs_info->uuid_tree_rescan_sem, 1);
2725

2726
	btrfs_init_dev_replace_locks(fs_info);
2727
	btrfs_init_qgroup(fs_info);
2728

2729 2730 2731
	btrfs_init_free_cluster(&fs_info->meta_alloc_cluster);
	btrfs_init_free_cluster(&fs_info->data_alloc_cluster);

2732
	init_waitqueue_head(&fs_info->transaction_throttle);
2733
	init_waitqueue_head(&fs_info->transaction_wait);
S
Sage Weil 已提交
2734
	init_waitqueue_head(&fs_info->transaction_blocked_wait);
2735
	init_waitqueue_head(&fs_info->async_submit_wait);
2736

2737 2738
	INIT_LIST_HEAD(&fs_info->pinned_chunks);

2739 2740 2741 2742 2743
	/* Usable values until the real ones are cached from the superblock */
	fs_info->nodesize = 4096;
	fs_info->sectorsize = 4096;
	fs_info->stripesize = 4096;

D
David Woodhouse 已提交
2744 2745
	ret = btrfs_alloc_stripe_hash_table(fs_info);
	if (ret) {
2746
		err = ret;
D
David Woodhouse 已提交
2747 2748 2749
		goto fail_alloc;
	}

2750
	__setup_root(tree_root, fs_info, BTRFS_ROOT_TREE_OBJECTID);
2751

2752
	invalidate_bdev(fs_devices->latest_bdev);
D
David Sterba 已提交
2753 2754 2755 2756

	/*
	 * Read super block and check the signature bytes only
	 */
Y
Yan Zheng 已提交
2757
	bh = btrfs_read_dev_super(fs_devices->latest_bdev);
2758 2759
	if (IS_ERR(bh)) {
		err = PTR_ERR(bh);
2760
		goto fail_alloc;
2761
	}
C
Chris Mason 已提交
2762

D
David Sterba 已提交
2763 2764 2765 2766
	/*
	 * We want to check superblock checksum, the type is stored inside.
	 * Pass the whole disk block of size BTRFS_SUPER_INFO_SIZE (4k).
	 */
2767
	if (btrfs_check_super_csum(fs_info, bh->b_data)) {
2768
		btrfs_err(fs_info, "superblock checksum mismatch");
D
David Sterba 已提交
2769
		err = -EINVAL;
2770
		brelse(bh);
D
David Sterba 已提交
2771 2772 2773 2774 2775 2776 2777 2778
		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
	 */
2779 2780 2781
	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));
2782
	brelse(bh);
2783

2784
	memcpy(fs_info->fsid, fs_info->super_copy->fsid, BTRFS_FSID_SIZE);
2785

2786
	ret = btrfs_check_super_valid(fs_info);
D
David Sterba 已提交
2787
	if (ret) {
2788
		btrfs_err(fs_info, "superblock contains fatal errors");
D
David Sterba 已提交
2789 2790 2791 2792
		err = -EINVAL;
		goto fail_alloc;
	}

2793
	disk_super = fs_info->super_copy;
2794
	if (!btrfs_super_root(disk_super))
2795
		goto fail_alloc;
2796

L
liubo 已提交
2797
	/* check FS state, whether FS is broken. */
2798 2799
	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
2800

C
Chris Mason 已提交
2801 2802 2803 2804 2805 2806 2807
	/*
	 * 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);

2808 2809 2810 2811 2812 2813
	/*
	 * 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;

2814
	ret = btrfs_parse_options(fs_info, options, sb->s_flags);
Y
Yan Zheng 已提交
2815 2816
	if (ret) {
		err = ret;
2817
		goto fail_alloc;
Y
Yan Zheng 已提交
2818
	}
2819

2820 2821 2822
	features = btrfs_super_incompat_flags(disk_super) &
		~BTRFS_FEATURE_INCOMPAT_SUPP;
	if (features) {
2823 2824 2825
		btrfs_err(fs_info,
		    "cannot mount because of unsupported optional features (%llx)",
		    features);
2826
		err = -EINVAL;
2827
		goto fail_alloc;
2828 2829
	}

2830
	features = btrfs_super_incompat_flags(disk_super);
L
Li Zefan 已提交
2831
	features |= BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF;
2832
	if (fs_info->compress_type == BTRFS_COMPRESS_LZO)
L
Li Zefan 已提交
2833
		features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
2834

2835
	if (features & BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA)
2836
		btrfs_info(fs_info, "has skinny extents");
2837

2838 2839 2840 2841
	/*
	 * flag our filesystem as having big metadata blocks if
	 * they are bigger than the page size
	 */
2842
	if (btrfs_super_nodesize(disk_super) > PAGE_SIZE) {
2843
		if (!(features & BTRFS_FEATURE_INCOMPAT_BIG_METADATA))
2844 2845
			btrfs_info(fs_info,
				"flagging fs with big metadata feature");
2846 2847 2848
		features |= BTRFS_FEATURE_INCOMPAT_BIG_METADATA;
	}

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

2855 2856 2857 2858 2859
	/* Cache block sizes */
	fs_info->nodesize = nodesize;
	fs_info->sectorsize = sectorsize;
	fs_info->stripesize = stripesize;

2860 2861 2862 2863 2864
	/*
	 * 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) &&
2865
	    (sectorsize != nodesize)) {
2866 2867 2868
		btrfs_err(fs_info,
"unequal nodesize/sectorsize (%u != %u) are not allowed for mixed block groups",
			nodesize, sectorsize);
2869 2870 2871
		goto fail_alloc;
	}

2872 2873 2874 2875
	/*
	 * Needn't use the lock because there is no other task which will
	 * update the flag.
	 */
L
Li Zefan 已提交
2876
	btrfs_set_super_incompat_flags(disk_super, features);
2877

2878 2879 2880
	features = btrfs_super_compat_ro_flags(disk_super) &
		~BTRFS_FEATURE_COMPAT_RO_SUPP;
	if (!(sb->s_flags & MS_RDONLY) && features) {
2881 2882
		btrfs_err(fs_info,
	"cannot mount read-write because of unsupported optional features (%llx)",
2883
		       features);
2884
		err = -EINVAL;
2885
		goto fail_alloc;
2886
	}
2887

2888
	max_active = fs_info->thread_pool_size;
2889

2890 2891 2892
	ret = btrfs_init_workqueues(fs_info, fs_devices);
	if (ret) {
		err = ret;
2893 2894
		goto fail_sb_buffer;
	}
2895

2896 2897 2898 2899 2900 2901
	sb->s_bdi->congested_fn = btrfs_congested_fn;
	sb->s_bdi->congested_data = fs_info;
	sb->s_bdi->capabilities |= BDI_CAP_CGROUP_WRITEBACK;
	sb->s_bdi->ra_pages = VM_MAX_READAHEAD * 1024 / PAGE_SIZE;
	sb->s_bdi->ra_pages *= btrfs_super_num_devices(disk_super);
	sb->s_bdi->ra_pages = max(sb->s_bdi->ra_pages, SZ_4M / PAGE_SIZE);
2902

2903 2904
	sb->s_blocksize = sectorsize;
	sb->s_blocksize_bits = blksize_bits(sectorsize);
2905

2906
	mutex_lock(&fs_info->chunk_mutex);
2907
	ret = btrfs_read_sys_array(fs_info);
2908
	mutex_unlock(&fs_info->chunk_mutex);
2909
	if (ret) {
2910
		btrfs_err(fs_info, "failed to read the system array: %d", ret);
2911
		goto fail_sb_buffer;
2912
	}
2913

2914
	generation = btrfs_super_chunk_root_generation(disk_super);
2915

2916
	__setup_root(chunk_root, fs_info, BTRFS_CHUNK_TREE_OBJECTID);
2917

2918
	chunk_root->node = read_tree_block(fs_info,
2919
					   btrfs_super_chunk_root(disk_super),
2920
					   generation);
2921 2922
	if (IS_ERR(chunk_root->node) ||
	    !extent_buffer_uptodate(chunk_root->node)) {
2923
		btrfs_err(fs_info, "failed to read chunk root");
2924 2925
		if (!IS_ERR(chunk_root->node))
			free_extent_buffer(chunk_root->node);
2926
		chunk_root->node = NULL;
C
Chris Mason 已提交
2927
		goto fail_tree_roots;
2928
	}
2929 2930
	btrfs_set_root_node(&chunk_root->root_item, chunk_root->node);
	chunk_root->commit_root = btrfs_root_node(chunk_root);
2931

2932
	read_extent_buffer(chunk_root->node, fs_info->chunk_tree_uuid,
2933
	   btrfs_header_chunk_tree_uuid(chunk_root->node), BTRFS_UUID_SIZE);
2934

2935
	ret = btrfs_read_chunk_tree(fs_info);
Y
Yan Zheng 已提交
2936
	if (ret) {
2937
		btrfs_err(fs_info, "failed to read chunk tree: %d", ret);
C
Chris Mason 已提交
2938
		goto fail_tree_roots;
Y
Yan Zheng 已提交
2939
	}
2940

2941 2942 2943 2944
	/*
	 * keep the device that is marked to be the target device for the
	 * dev_replace procedure
	 */
2945
	btrfs_close_extra_devices(fs_devices, 0);
2946

2947
	if (!fs_devices->latest_bdev) {
2948
		btrfs_err(fs_info, "failed to read devices");
2949 2950 2951
		goto fail_tree_roots;
	}

C
Chris Mason 已提交
2952
retry_root_backup:
2953
	generation = btrfs_super_generation(disk_super);
2954

2955
	tree_root->node = read_tree_block(fs_info,
2956
					  btrfs_super_root(disk_super),
2957
					  generation);
2958 2959
	if (IS_ERR(tree_root->node) ||
	    !extent_buffer_uptodate(tree_root->node)) {
2960
		btrfs_warn(fs_info, "failed to read tree root");
2961 2962
		if (!IS_ERR(tree_root->node))
			free_extent_buffer(tree_root->node);
2963
		tree_root->node = NULL;
C
Chris Mason 已提交
2964
		goto recovery_tree_root;
2965
	}
C
Chris Mason 已提交
2966

2967 2968
	btrfs_set_root_node(&tree_root->root_item, tree_root->node);
	tree_root->commit_root = btrfs_root_node(tree_root);
2969
	btrfs_set_root_refs(&tree_root->root_item, 1);
2970

2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982
	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);

2983
	ret = btrfs_read_roots(fs_info);
2984
	if (ret)
C
Chris Mason 已提交
2985
		goto recovery_tree_root;
2986

2987 2988 2989
	fs_info->generation = generation;
	fs_info->last_trans_committed = generation;

2990 2991
	ret = btrfs_recover_balance(fs_info);
	if (ret) {
2992
		btrfs_err(fs_info, "failed to recover balance: %d", ret);
2993 2994 2995
		goto fail_block_groups;
	}

2996 2997
	ret = btrfs_init_dev_stats(fs_info);
	if (ret) {
2998
		btrfs_err(fs_info, "failed to init dev_stats: %d", ret);
2999 3000 3001
		goto fail_block_groups;
	}

3002 3003
	ret = btrfs_init_dev_replace(fs_info);
	if (ret) {
3004
		btrfs_err(fs_info, "failed to init dev_replace: %d", ret);
3005 3006 3007
		goto fail_block_groups;
	}

3008
	btrfs_close_extra_devices(fs_devices, 1);
3009

3010 3011
	ret = btrfs_sysfs_add_fsid(fs_devices, NULL);
	if (ret) {
3012 3013
		btrfs_err(fs_info, "failed to init sysfs fsid interface: %d",
				ret);
3014 3015 3016 3017 3018
		goto fail_block_groups;
	}

	ret = btrfs_sysfs_add_device(fs_devices);
	if (ret) {
3019 3020
		btrfs_err(fs_info, "failed to init sysfs device interface: %d",
				ret);
3021 3022 3023
		goto fail_fsdev_sysfs;
	}

3024
	ret = btrfs_sysfs_add_mounted(fs_info);
3025
	if (ret) {
3026
		btrfs_err(fs_info, "failed to init sysfs interface: %d", ret);
3027
		goto fail_fsdev_sysfs;
3028 3029 3030 3031
	}

	ret = btrfs_init_space_info(fs_info);
	if (ret) {
3032
		btrfs_err(fs_info, "failed to initialize space info: %d", ret);
3033
		goto fail_sysfs;
3034 3035
	}

3036
	ret = btrfs_read_block_groups(fs_info);
3037
	if (ret) {
3038
		btrfs_err(fs_info, "failed to read block groups: %d", ret);
3039
		goto fail_sysfs;
3040
	}
3041 3042
	fs_info->num_tolerated_disk_barrier_failures =
		btrfs_calc_num_tolerated_disk_barrier_failures(fs_info);
3043 3044 3045
	if (fs_info->fs_devices->missing_devices >
	     fs_info->num_tolerated_disk_barrier_failures &&
	    !(sb->s_flags & MS_RDONLY)) {
3046 3047
		btrfs_warn(fs_info,
"missing devices (%llu) exceeds the limit (%d), writeable mount is not allowed",
3048 3049
			fs_info->fs_devices->missing_devices,
			fs_info->num_tolerated_disk_barrier_failures);
3050
		goto fail_sysfs;
3051
	}
C
Chris Mason 已提交
3052

3053 3054
	fs_info->cleaner_kthread = kthread_run(cleaner_kthread, tree_root,
					       "btrfs-cleaner");
3055
	if (IS_ERR(fs_info->cleaner_kthread))
3056
		goto fail_sysfs;
3057 3058 3059 3060

	fs_info->transaction_kthread = kthread_run(transaction_kthread,
						   tree_root,
						   "btrfs-transaction");
3061
	if (IS_ERR(fs_info->transaction_kthread))
3062
		goto fail_cleaner;
3063

3064 3065
	if (!btrfs_test_opt(fs_info, SSD) &&
	    !btrfs_test_opt(fs_info, NOSSD) &&
C
Chris Mason 已提交
3066
	    !fs_info->fs_devices->rotating) {
3067
		btrfs_info(fs_info, "detected SSD devices, enabling SSD mode");
C
Chris Mason 已提交
3068 3069 3070
		btrfs_set_opt(fs_info->mount_opt, SSD);
	}

3071
	/*
3072
	 * Mount does not set all options immediately, we can do it now and do
3073 3074 3075
	 * not have to wait for transaction commit
	 */
	btrfs_apply_pending_changes(fs_info);
3076

3077
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
3078
	if (btrfs_test_opt(fs_info, CHECK_INTEGRITY)) {
3079
		ret = btrfsic_mount(fs_info, fs_devices,
3080
				    btrfs_test_opt(fs_info,
3081 3082 3083 3084
					CHECK_INTEGRITY_INCLUDING_EXTENT_DATA) ?
				    1 : 0,
				    fs_info->check_integrity_print_mask);
		if (ret)
3085 3086 3087
			btrfs_warn(fs_info,
				"failed to initialize integrity check module: %d",
				ret);
3088 3089
	}
#endif
3090 3091 3092
	ret = btrfs_read_qgroup_config(fs_info);
	if (ret)
		goto fail_trans_kthread;
3093

3094 3095
	/* do not make disk changes in broken FS or nologreplay is given */
	if (btrfs_super_log_root(disk_super) != 0 &&
3096
	    !btrfs_test_opt(fs_info, NOLOGREPLAY)) {
3097
		ret = btrfs_replay_log(fs_info, fs_devices);
3098
		if (ret) {
3099
			err = ret;
3100
			goto fail_qgroup;
3101
		}
3102
	}
Z
Zheng Yan 已提交
3103

3104
	ret = btrfs_find_orphan_roots(fs_info);
3105
	if (ret)
3106
		goto fail_qgroup;
3107

3108
	if (!(sb->s_flags & MS_RDONLY)) {
3109
		ret = btrfs_cleanup_fs_roots(fs_info);
3110
		if (ret)
3111
			goto fail_qgroup;
3112 3113

		mutex_lock(&fs_info->cleaner_mutex);
3114
		ret = btrfs_recover_relocation(tree_root);
3115
		mutex_unlock(&fs_info->cleaner_mutex);
3116
		if (ret < 0) {
3117 3118
			btrfs_warn(fs_info, "failed to recover relocation: %d",
					ret);
3119
			err = -EINVAL;
3120
			goto fail_qgroup;
3121
		}
3122
	}
Z
Zheng Yan 已提交
3123

3124 3125
	location.objectid = BTRFS_FS_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
3126
	location.offset = 0;
3127 3128

	fs_info->fs_root = btrfs_read_fs_root_no_name(fs_info, &location);
3129 3130
	if (IS_ERR(fs_info->fs_root)) {
		err = PTR_ERR(fs_info->fs_root);
3131
		goto fail_qgroup;
3132
	}
C
Chris Mason 已提交
3133

3134 3135
	if (sb->s_flags & MS_RDONLY)
		return 0;
I
Ilya Dryomov 已提交
3136

3137 3138
	if (btrfs_test_opt(fs_info, CLEAR_CACHE) &&
	    btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
3139 3140 3141 3142 3143 3144 3145 3146
		clear_free_space_tree = 1;
	} else if (btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE) &&
		   !btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE_VALID)) {
		btrfs_warn(fs_info, "free space tree is invalid");
		clear_free_space_tree = 1;
	}

	if (clear_free_space_tree) {
3147 3148 3149 3150 3151
		btrfs_info(fs_info, "clearing free space tree");
		ret = btrfs_clear_free_space_tree(fs_info);
		if (ret) {
			btrfs_warn(fs_info,
				   "failed to clear free space tree: %d", ret);
3152
			close_ctree(fs_info);
3153 3154 3155 3156
			return ret;
		}
	}

3157
	if (btrfs_test_opt(fs_info, FREE_SPACE_TREE) &&
3158
	    !btrfs_fs_compat_ro(fs_info, FREE_SPACE_TREE)) {
3159
		btrfs_info(fs_info, "creating free space tree");
3160 3161
		ret = btrfs_create_free_space_tree(fs_info);
		if (ret) {
3162 3163
			btrfs_warn(fs_info,
				"failed to create free space tree: %d", ret);
3164
			close_ctree(fs_info);
3165 3166 3167 3168
			return ret;
		}
	}

3169 3170 3171
	down_read(&fs_info->cleanup_work_sem);
	if ((ret = btrfs_orphan_cleanup(fs_info->fs_root)) ||
	    (ret = btrfs_orphan_cleanup(fs_info->tree_root))) {
3172
		up_read(&fs_info->cleanup_work_sem);
3173
		close_ctree(fs_info);
3174 3175 3176
		return ret;
	}
	up_read(&fs_info->cleanup_work_sem);
I
Ilya Dryomov 已提交
3177

3178 3179
	ret = btrfs_resume_balance_async(fs_info);
	if (ret) {
3180
		btrfs_warn(fs_info, "failed to resume balance: %d", ret);
3181
		close_ctree(fs_info);
3182
		return ret;
3183 3184
	}

3185 3186
	ret = btrfs_resume_dev_replace_async(fs_info);
	if (ret) {
3187
		btrfs_warn(fs_info, "failed to resume device replace: %d", ret);
3188
		close_ctree(fs_info);
3189 3190 3191
		return ret;
	}

3192 3193
	btrfs_qgroup_rescan_resume(fs_info);

3194
	if (!fs_info->uuid_root) {
3195
		btrfs_info(fs_info, "creating UUID tree");
3196 3197
		ret = btrfs_create_uuid_tree(fs_info);
		if (ret) {
3198 3199
			btrfs_warn(fs_info,
				"failed to create the UUID tree: %d", ret);
3200
			close_ctree(fs_info);
3201 3202
			return ret;
		}
3203
	} else if (btrfs_test_opt(fs_info, RESCAN_UUID_TREE) ||
3204 3205
		   fs_info->generation !=
				btrfs_super_uuid_tree_generation(disk_super)) {
3206
		btrfs_info(fs_info, "checking UUID tree");
3207 3208
		ret = btrfs_check_uuid_tree(fs_info);
		if (ret) {
3209 3210
			btrfs_warn(fs_info,
				"failed to check the UUID tree: %d", ret);
3211
			close_ctree(fs_info);
3212 3213 3214
			return ret;
		}
	} else {
3215
		set_bit(BTRFS_FS_UPDATE_UUID_TREE_GEN, &fs_info->flags);
3216
	}
3217
	set_bit(BTRFS_FS_OPEN, &fs_info->flags);
3218

3219 3220 3221 3222 3223 3224
	/*
	 * 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
3225
	return 0;
C
Chris Mason 已提交
3226

3227 3228
fail_qgroup:
	btrfs_free_qgroup_config(fs_info);
3229 3230
fail_trans_kthread:
	kthread_stop(fs_info->transaction_kthread);
3231
	btrfs_cleanup_transaction(fs_info);
3232
	btrfs_free_fs_roots(fs_info);
3233
fail_cleaner:
3234
	kthread_stop(fs_info->cleaner_kthread);
3235 3236 3237 3238 3239 3240 3241

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

3242
fail_sysfs:
3243
	btrfs_sysfs_remove_mounted(fs_info);
3244

3245 3246 3247
fail_fsdev_sysfs:
	btrfs_sysfs_remove_fsid(fs_info->fs_devices);

3248
fail_block_groups:
J
Josef Bacik 已提交
3249
	btrfs_put_block_group_cache(fs_info);
C
Chris Mason 已提交
3250 3251 3252

fail_tree_roots:
	free_root_pointers(fs_info, 1);
3253
	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
3254

C
Chris Mason 已提交
3255
fail_sb_buffer:
L
Liu Bo 已提交
3256
	btrfs_stop_all_workers(fs_info);
3257
	btrfs_free_block_groups(fs_info);
3258
fail_alloc:
3259
fail_iput:
3260 3261
	btrfs_mapping_tree_free(&fs_info->mapping_tree);

3262
	iput(fs_info->btree_inode);
3263 3264
fail_bio_counter:
	percpu_counter_destroy(&fs_info->bio_counter);
3265 3266
fail_delalloc_bytes:
	percpu_counter_destroy(&fs_info->delalloc_bytes);
3267 3268
fail_dirty_metadata_bytes:
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
3269 3270
fail_srcu:
	cleanup_srcu_struct(&fs_info->subvol_srcu);
3271
fail:
D
David Woodhouse 已提交
3272
	btrfs_free_stripe_hash_table(fs_info);
3273
	btrfs_close_devices(fs_info->fs_devices);
A
Al Viro 已提交
3274
	return err;
C
Chris Mason 已提交
3275 3276

recovery_tree_root:
3277
	if (!btrfs_test_opt(fs_info, USEBACKUPROOT))
C
Chris Mason 已提交
3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292
		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;
3293 3294
}

3295 3296 3297 3298 3299
static void btrfs_end_buffer_write_sync(struct buffer_head *bh, int uptodate)
{
	if (uptodate) {
		set_buffer_uptodate(bh);
	} else {
3300 3301 3302
		struct btrfs_device *device = (struct btrfs_device *)
			bh->b_private;

3303
		btrfs_warn_rl_in_rcu(device->fs_info,
3304
				"lost page write due to IO error on %s",
3305
					  rcu_str_deref(device->name));
3306
		/* note, we don't set_buffer_write_io_error because we have
3307 3308
		 * our own ways of dealing with the IO errors
		 */
3309
		clear_buffer_uptodate(bh);
3310
		btrfs_dev_stat_inc_and_print(device, BTRFS_DEV_STAT_WRITE_ERRS);
3311 3312 3313 3314 3315
	}
	unlock_buffer(bh);
	put_bh(bh);
}

3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346
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
3347 3348 3349 3350 3351 3352 3353
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;
3354
	int ret = -EINVAL;
Y
Yan Zheng 已提交
3355 3356 3357 3358 3359 3360 3361

	/* 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++) {
3362 3363
		ret = btrfs_read_dev_one_super(bdev, i, &bh);
		if (ret)
Y
Yan Zheng 已提交
3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375
			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);
		}
	}
3376 3377 3378 3379

	if (!latest)
		return ERR_PTR(ret);

Y
Yan Zheng 已提交
3380 3381 3382
	return latest;
}

3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393
/*
 * 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
3394 3395
static int write_dev_supers(struct btrfs_device *device,
			    struct btrfs_super_block *sb,
3396
			    int wait, int max_mirrors)
Y
Yan Zheng 已提交
3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409
{
	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);
3410 3411
		if (bytenr + BTRFS_SUPER_INFO_SIZE >=
		    device->commit_total_bytes)
Y
Yan Zheng 已提交
3412 3413 3414 3415 3416
			break;

		if (wait) {
			bh = __find_get_block(device->bdev, bytenr / 4096,
					      BTRFS_SUPER_INFO_SIZE);
3417 3418 3419 3420
			if (!bh) {
				errors++;
				continue;
			}
Y
Yan Zheng 已提交
3421
			wait_on_buffer(bh);
3422 3423 3424 3425 3426 3427 3428 3429 3430
			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
3431 3432 3433 3434
		} else {
			btrfs_set_super_bytenr(sb, bytenr);

			crc = ~(u32)0;
3435
			crc = btrfs_csum_data((const char *)sb +
Y
Yan Zheng 已提交
3436 3437 3438 3439 3440
					      BTRFS_CSUM_SIZE, crc,
					      BTRFS_SUPER_INFO_SIZE -
					      BTRFS_CSUM_SIZE);
			btrfs_csum_final(crc, sb->csum);

3441 3442 3443 3444
			/*
			 * one reference for us, and we leave it for the
			 * caller
			 */
Y
Yan Zheng 已提交
3445 3446
			bh = __getblk(device->bdev, bytenr / 4096,
				      BTRFS_SUPER_INFO_SIZE);
3447
			if (!bh) {
3448
				btrfs_err(device->fs_info,
3449 3450
				    "couldn't get super buffer head for bytenr %llu",
				    bytenr);
3451 3452 3453 3454
				errors++;
				continue;
			}

Y
Yan Zheng 已提交
3455 3456
			memcpy(bh->b_data, sb, BTRFS_SUPER_INFO_SIZE);

3457
			/* one reference for submit_bh */
Y
Yan Zheng 已提交
3458
			get_bh(bh);
3459 3460

			set_buffer_uptodate(bh);
Y
Yan Zheng 已提交
3461 3462
			lock_buffer(bh);
			bh->b_end_io = btrfs_end_buffer_write_sync;
3463
			bh->b_private = device;
Y
Yan Zheng 已提交
3464 3465
		}

C
Chris Mason 已提交
3466 3467 3468 3469
		/*
		 * we fua the first super.  The others we allow
		 * to go down lazy.
		 */
3470 3471 3472 3473
		if (i == 0) {
			ret = btrfsic_submit_bh(REQ_OP_WRITE,
						REQ_SYNC | REQ_FUA, bh);
		} else {
3474
			ret = btrfsic_submit_bh(REQ_OP_WRITE, REQ_SYNC, bh);
3475
		}
3476
		if (ret)
Y
Yan Zheng 已提交
3477 3478 3479 3480 3481
			errors++;
	}
	return errors < i ? 0 : -1;
}

C
Chris Mason 已提交
3482 3483 3484 3485
/*
 * endio for the write_dev_flush, this will wake anyone waiting
 * for the barrier when it is done
 */
3486
static void btrfs_end_empty_barrier(struct bio *bio)
C
Chris Mason 已提交
3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501
{
	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)
{
3502
	struct request_queue *q = bdev_get_queue(device->bdev);
C
Chris Mason 已提交
3503 3504 3505
	struct bio *bio;
	int ret = 0;

3506
	if (!test_bit(QUEUE_FLAG_WC, &q->queue_flags))
C
Chris Mason 已提交
3507 3508 3509 3510 3511
		return 0;

	if (wait) {
		bio = device->flush_bio;
		if (!bio)
3512 3513 3514 3515
			/*
			 * This means the alloc has failed with ENOMEM, however
			 * here we return 0, as its not a device error.
			 */
C
Chris Mason 已提交
3516 3517 3518 3519
			return 0;

		wait_for_completion(&device->flush_wait);

3520 3521
		if (bio->bi_error) {
			ret = bio->bi_error;
3522 3523
			btrfs_dev_stat_inc_and_print(device,
				BTRFS_DEV_STAT_FLUSH_ERRS);
C
Chris Mason 已提交
3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536
		}

		/* 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
	 */
3537
	device->flush_bio = NULL;
3538
	bio = btrfs_io_bio_alloc(GFP_NOFS, 0);
C
Chris Mason 已提交
3539 3540 3541 3542 3543
	if (!bio)
		return -ENOMEM;

	bio->bi_end_io = btrfs_end_empty_barrier;
	bio->bi_bdev = device->bdev;
3544
	bio->bi_opf = REQ_OP_WRITE | REQ_SYNC | REQ_PREFLUSH;
C
Chris Mason 已提交
3545 3546 3547 3548 3549
	init_completion(&device->flush_wait);
	bio->bi_private = &device->flush_wait;
	device->flush_bio = bio;

	bio_get(bio);
3550
	btrfsic_submit_bio(bio);
C
Chris Mason 已提交
3551 3552 3553 3554

	return 0;
}

3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580
static int check_barrier_error(struct btrfs_fs_devices *fsdevs)
{
	int submit_flush_error = 0;
	int dev_flush_error = 0;
	struct btrfs_device *dev;
	int tolerance;

	list_for_each_entry_rcu(dev, &fsdevs->devices, dev_list) {
		if (!dev->bdev) {
			submit_flush_error++;
			dev_flush_error++;
			continue;
		}
		if (dev->last_flush_error == -ENOMEM)
			submit_flush_error++;
		if (dev->last_flush_error && dev->last_flush_error != -ENOMEM)
			dev_flush_error++;
	}

	tolerance = fsdevs->fs_info->num_tolerated_disk_barrier_failures;
	if (submit_flush_error > tolerance || dev_flush_error > tolerance)
		return -EIO;

	return 0;
}

C
Chris Mason 已提交
3581 3582 3583 3584 3585 3586 3587 3588
/*
 * 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;
3589 3590
	int errors_send = 0;
	int errors_wait = 0;
C
Chris Mason 已提交
3591 3592 3593 3594 3595
	int ret;

	/* send down all the barriers */
	head = &info->fs_devices->devices;
	list_for_each_entry_rcu(dev, head, dev_list) {
3596 3597
		if (dev->missing)
			continue;
C
Chris Mason 已提交
3598
		if (!dev->bdev) {
3599
			errors_send++;
C
Chris Mason 已提交
3600 3601 3602 3603 3604 3605 3606
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

		ret = write_dev_flush(dev, 0);
		if (ret)
3607
			errors_send++;
3608
		dev->last_flush_error = ret;
C
Chris Mason 已提交
3609 3610 3611 3612
	}

	/* wait for all the barriers */
	list_for_each_entry_rcu(dev, head, dev_list) {
3613 3614
		if (dev->missing)
			continue;
C
Chris Mason 已提交
3615
		if (!dev->bdev) {
3616
			errors_wait++;
C
Chris Mason 已提交
3617 3618 3619 3620 3621 3622
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

		ret = write_dev_flush(dev, 1);
3623 3624
		if (ret) {
			dev->last_flush_error = ret;
3625
			errors_wait++;
3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645
		}
	}

	/*
	 * Try hard in case of flush. Lets say, in RAID1 we have
	 * the following situation
	 *  dev1: EIO dev2: ENOMEM
	 * this is not a fatal error as we hope to recover from
	 * ENOMEM in the next attempt to flush.
	 * But the following is considered as fatal
	 *  dev1: ENOMEM dev2: ENOMEM
	 *  dev1: bdev == NULL dev2: ENOMEM
	 */
	if (errors_send || errors_wait) {
		/*
		 * At some point we need the status of all disks
		 * to arrive at the volume status. So error checking
		 * is being pushed to a separate loop.
		 */
		return check_barrier_error(info->fs_devices);
C
Chris Mason 已提交
3646 3647 3648 3649
	}
	return 0;
}

3650 3651
int btrfs_get_num_tolerated_disk_barrier_failures(u64 flags)
{
3652 3653
	int raid_type;
	int min_tolerated = INT_MAX;
3654

3655 3656 3657 3658 3659
	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);
3660

3661 3662 3663 3664 3665 3666 3667 3668 3669
	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);
	}
3670

3671
	if (min_tolerated == INT_MAX) {
3672
		pr_warn("BTRFS: unknown raid flag: %llu", flags);
3673 3674 3675 3676
		min_tolerated = 0;
	}

	return min_tolerated;
3677 3678
}

3679 3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691 3692
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;

3693
	for (i = 0; i < ARRAY_SIZE(types); i++) {
3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710
		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++) {
3711 3712 3713 3714 3715 3716 3717 3718 3719 3720
			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;
3721 3722 3723 3724 3725

			num_tolerated_disk_barrier_failures = min(
				num_tolerated_disk_barrier_failures,
				btrfs_get_num_tolerated_disk_barrier_failures(
					flags));
3726 3727 3728 3729 3730 3731 3732
		}
		up_read(&sinfo->groups_sem);
	}

	return num_tolerated_disk_barrier_failures;
}

3733
int write_all_supers(struct btrfs_fs_info *fs_info, int max_mirrors)
3734
{
3735
	struct list_head *head;
3736
	struct btrfs_device *dev;
3737
	struct btrfs_super_block *sb;
3738 3739 3740
	struct btrfs_dev_item *dev_item;
	int ret;
	int do_barriers;
3741 3742
	int max_errors;
	int total_errors = 0;
3743
	u64 flags;
3744

3745 3746
	do_barriers = !btrfs_test_opt(fs_info, NOBARRIER);
	backup_super_roots(fs_info);
3747

3748
	sb = fs_info->super_for_commit;
3749
	dev_item = &sb->dev_item;
3750

3751 3752 3753
	mutex_lock(&fs_info->fs_devices->device_list_mutex);
	head = &fs_info->fs_devices->devices;
	max_errors = btrfs_super_num_devices(fs_info->super_copy) - 1;
C
Chris Mason 已提交
3754

3755
	if (do_barriers) {
3756
		ret = barrier_all_devices(fs_info);
3757 3758
		if (ret) {
			mutex_unlock(
3759 3760 3761
				&fs_info->fs_devices->device_list_mutex);
			btrfs_handle_fs_error(fs_info, ret,
					      "errors while submitting device barriers.");
3762 3763 3764
			return ret;
		}
	}
C
Chris Mason 已提交
3765

3766
	list_for_each_entry_rcu(dev, head, dev_list) {
3767 3768 3769 3770
		if (!dev->bdev) {
			total_errors++;
			continue;
		}
Y
Yan Zheng 已提交
3771
		if (!dev->in_fs_metadata || !dev->writeable)
3772 3773
			continue;

Y
Yan Zheng 已提交
3774
		btrfs_set_stack_device_generation(dev_item, 0);
3775 3776
		btrfs_set_stack_device_type(dev_item, dev->type);
		btrfs_set_stack_device_id(dev_item, dev->devid);
3777
		btrfs_set_stack_device_total_bytes(dev_item,
3778
						   dev->commit_total_bytes);
3779 3780
		btrfs_set_stack_device_bytes_used(dev_item,
						  dev->commit_bytes_used);
3781 3782 3783 3784
		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
3785
		memcpy(dev_item->fsid, dev->fs_devices->fsid, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3786

3787 3788 3789
		flags = btrfs_super_flags(sb);
		btrfs_set_super_flags(sb, flags | BTRFS_HEADER_FLAG_WRITTEN);

3790
		ret = write_dev_supers(dev, sb, 0, max_mirrors);
3791 3792
		if (ret)
			total_errors++;
3793
	}
3794
	if (total_errors > max_errors) {
3795 3796 3797
		btrfs_err(fs_info, "%d errors while writing supers",
			  total_errors);
		mutex_unlock(&fs_info->fs_devices->device_list_mutex);
3798

3799
		/* FUA is masked off if unsupported and can't be the reason */
3800 3801 3802
		btrfs_handle_fs_error(fs_info, -EIO,
				      "%d errors while writing supers",
				      total_errors);
3803
		return -EIO;
3804
	}
3805

Y
Yan Zheng 已提交
3806
	total_errors = 0;
3807
	list_for_each_entry_rcu(dev, head, dev_list) {
3808 3809
		if (!dev->bdev)
			continue;
Y
Yan Zheng 已提交
3810
		if (!dev->in_fs_metadata || !dev->writeable)
3811 3812
			continue;

3813
		ret = write_dev_supers(dev, sb, 1, max_mirrors);
Y
Yan Zheng 已提交
3814 3815
		if (ret)
			total_errors++;
3816
	}
3817
	mutex_unlock(&fs_info->fs_devices->device_list_mutex);
3818
	if (total_errors > max_errors) {
3819 3820 3821
		btrfs_handle_fs_error(fs_info, -EIO,
				      "%d errors while writing supers",
				      total_errors);
3822
		return -EIO;
3823
	}
3824 3825 3826
	return 0;
}

3827 3828 3829
/* 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
3830
{
3831
	spin_lock(&fs_info->fs_roots_radix_lock);
C
Chris Mason 已提交
3832 3833
	radix_tree_delete(&fs_info->fs_roots_radix,
			  (unsigned long)root->root_key.objectid);
3834
	spin_unlock(&fs_info->fs_roots_radix_lock);
3835 3836 3837 3838

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

L
Liu Bo 已提交
3839
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
L
Liu Bo 已提交
3840
		btrfs_free_log(NULL, root);
L
Liu Bo 已提交
3841 3842 3843 3844 3845 3846 3847
		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
3848

3849 3850 3851 3852
	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);
3853 3854 3855 3856 3857
	free_fs_root(root);
}

static void free_fs_root(struct btrfs_root *root)
{
3858
	iput(root->ino_cache_inode);
3859
	WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree));
3860
	btrfs_free_block_rsv(root->fs_info, root->orphan_block_rsv);
3861
	root->orphan_block_rsv = NULL;
3862 3863
	if (root->anon_dev)
		free_anon_bdev(root->anon_dev);
3864 3865
	if (root->subv_writers)
		btrfs_free_subvolume_writers(root->subv_writers);
3866 3867
	free_extent_buffer(root->node);
	free_extent_buffer(root->commit_root);
3868 3869
	kfree(root->free_ino_ctl);
	kfree(root->free_ino_pinned);
C
Chris Mason 已提交
3870
	kfree(root->name);
3871
	btrfs_put_fs_root(root);
C
Chris Mason 已提交
3872 3873
}

3874 3875 3876
void btrfs_free_fs_root(struct btrfs_root *root)
{
	free_fs_root(root);
C
Chris Mason 已提交
3877 3878
}

Y
Yan Zheng 已提交
3879
int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
C
Chris Mason 已提交
3880
{
Y
Yan Zheng 已提交
3881 3882
	u64 root_objectid = 0;
	struct btrfs_root *gang[8];
3883 3884 3885 3886
	int i = 0;
	int err = 0;
	unsigned int ret = 0;
	int index;
3887

Y
Yan Zheng 已提交
3888
	while (1) {
3889
		index = srcu_read_lock(&fs_info->subvol_srcu);
Y
Yan Zheng 已提交
3890 3891 3892
		ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix,
					     (void **)gang, root_objectid,
					     ARRAY_SIZE(gang));
3893 3894
		if (!ret) {
			srcu_read_unlock(&fs_info->subvol_srcu, index);
Y
Yan Zheng 已提交
3895
			break;
3896
		}
3897
		root_objectid = gang[ret - 1]->root_key.objectid + 1;
3898

Y
Yan Zheng 已提交
3899
		for (i = 0; i < ret; i++) {
3900 3901 3902 3903 3904 3905 3906 3907 3908
			/* 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);
3909

3910 3911 3912
		for (i = 0; i < ret; i++) {
			if (!gang[i])
				continue;
Y
Yan Zheng 已提交
3913
			root_objectid = gang[i]->root_key.objectid;
3914 3915
			err = btrfs_orphan_cleanup(gang[i]);
			if (err)
3916 3917
				break;
			btrfs_put_fs_root(gang[i]);
Y
Yan Zheng 已提交
3918 3919 3920
		}
		root_objectid++;
	}
3921 3922 3923 3924 3925 3926 3927

	/* 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
3928
}
3929

3930
int btrfs_commit_super(struct btrfs_fs_info *fs_info)
Y
Yan Zheng 已提交
3931
{
3932
	struct btrfs_root *root = fs_info->tree_root;
Y
Yan Zheng 已提交
3933
	struct btrfs_trans_handle *trans;
3934

3935
	mutex_lock(&fs_info->cleaner_mutex);
3936
	btrfs_run_delayed_iputs(fs_info);
3937 3938
	mutex_unlock(&fs_info->cleaner_mutex);
	wake_up_process(fs_info->cleaner_kthread);
3939 3940

	/* wait until ongoing cleanup work done */
3941 3942
	down_write(&fs_info->cleanup_work_sem);
	up_write(&fs_info->cleanup_work_sem);
3943

3944
	trans = btrfs_join_transaction(root);
3945 3946
	if (IS_ERR(trans))
		return PTR_ERR(trans);
3947
	return btrfs_commit_transaction(trans);
Y
Yan Zheng 已提交
3948 3949
}

3950
void close_ctree(struct btrfs_fs_info *fs_info)
Y
Yan Zheng 已提交
3951
{
3952
	struct btrfs_root *root = fs_info->tree_root;
Y
Yan Zheng 已提交
3953 3954
	int ret;

3955
	set_bit(BTRFS_FS_CLOSING_START, &fs_info->flags);
Y
Yan Zheng 已提交
3956

3957
	/* wait for the qgroup rescan worker to stop */
3958
	btrfs_qgroup_wait_for_completion(fs_info, false);
3959

S
Stefan Behrens 已提交
3960 3961 3962 3963 3964
	/* 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);

3965
	/* pause restriper - we want to resume on mount */
3966
	btrfs_pause_balance(fs_info);
3967

3968 3969
	btrfs_dev_replace_suspend_for_unmount(fs_info);

3970
	btrfs_scrub_cancel(fs_info);
C
Chris Mason 已提交
3971 3972 3973 3974 3975 3976

	/* 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 */
3977
	btrfs_cleanup_defrag_inodes(fs_info);
C
Chris Mason 已提交
3978

3979 3980
	cancel_work_sync(&fs_info->async_reclaim_work);

Y
Yan Zheng 已提交
3981
	if (!(fs_info->sb->s_flags & MS_RDONLY)) {
3982 3983 3984 3985 3986
		/*
		 * 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.
		 */
3987
		btrfs_delete_unused_bgs(fs_info);
3988

3989
		ret = btrfs_commit_super(fs_info);
L
liubo 已提交
3990
		if (ret)
3991
			btrfs_err(fs_info, "commit super ret %d", ret);
L
liubo 已提交
3992 3993
	}

3994
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state))
3995
		btrfs_error_commit_super(fs_info);
3996

A
Al Viro 已提交
3997 3998
	kthread_stop(fs_info->transaction_kthread);
	kthread_stop(fs_info->cleaner_kthread);
3999

4000
	set_bit(BTRFS_FS_CLOSING_DONE, &fs_info->flags);
4001

4002
	btrfs_free_qgroup_config(fs_info);
4003

4004
	if (percpu_counter_sum(&fs_info->delalloc_bytes)) {
4005
		btrfs_info(fs_info, "at unmount delalloc count %lld",
4006
		       percpu_counter_sum(&fs_info->delalloc_bytes));
C
Chris Mason 已提交
4007
	}
4008

4009
	btrfs_sysfs_remove_mounted(fs_info);
4010
	btrfs_sysfs_remove_fsid(fs_info->fs_devices);
4011

4012
	btrfs_free_fs_roots(fs_info);
4013

4014 4015
	btrfs_put_block_group_cache(fs_info);

4016 4017 4018 4019 4020
	/*
	 * 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);
4021 4022
	btrfs_stop_all_workers(fs_info);

4023 4024
	btrfs_free_block_groups(fs_info);

4025
	clear_bit(BTRFS_FS_OPEN, &fs_info->flags);
4026
	free_root_pointers(fs_info, 1);
4027

4028
	iput(fs_info->btree_inode);
4029

4030
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
4031
	if (btrfs_test_opt(fs_info, CHECK_INTEGRITY))
4032
		btrfsic_unmount(fs_info->fs_devices);
4033 4034
#endif

4035
	btrfs_close_devices(fs_info->fs_devices);
4036
	btrfs_mapping_tree_free(&fs_info->mapping_tree);
4037

4038
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
4039
	percpu_counter_destroy(&fs_info->delalloc_bytes);
4040
	percpu_counter_destroy(&fs_info->bio_counter);
4041
	cleanup_srcu_struct(&fs_info->subvol_srcu);
4042

D
David Woodhouse 已提交
4043 4044
	btrfs_free_stripe_hash_table(fs_info);

4045
	__btrfs_free_block_rsv(root->orphan_block_rsv);
4046
	root->orphan_block_rsv = NULL;
4047

4048
	mutex_lock(&fs_info->chunk_mutex);
4049 4050 4051 4052 4053 4054 4055 4056
	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);
	}
4057
	mutex_unlock(&fs_info->chunk_mutex);
4058 4059
}

4060 4061
int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid,
			  int atomic)
4062
{
4063
	int ret;
4064
	struct inode *btree_inode = buf->pages[0]->mapping->host;
4065

4066
	ret = extent_buffer_uptodate(buf);
4067 4068 4069 4070
	if (!ret)
		return ret;

	ret = verify_parent_transid(&BTRFS_I(btree_inode)->io_tree, buf,
4071 4072 4073
				    parent_transid, atomic);
	if (ret == -EAGAIN)
		return ret;
4074
	return !ret;
4075 4076 4077 4078
}

void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
{
4079
	struct btrfs_fs_info *fs_info;
4080
	struct btrfs_root *root;
4081
	u64 transid = btrfs_header_generation(buf);
4082
	int was_dirty;
4083

4084 4085 4086 4087 4088 4089 4090 4091 4092 4093
#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;
4094
	fs_info = root->fs_info;
4095
	btrfs_assert_tree_locked(buf);
4096
	if (transid != fs_info->generation)
J
Jeff Mahoney 已提交
4097
		WARN(1, KERN_CRIT "btrfs transid mismatch buffer %llu, found %llu running %llu\n",
4098
			buf->start, transid, fs_info->generation);
4099
	was_dirty = set_extent_buffer_dirty(buf);
4100
	if (!was_dirty)
4101
		__percpu_counter_add(&fs_info->dirty_metadata_bytes,
4102
				     buf->len,
4103
				     fs_info->dirty_metadata_batch);
4104 4105
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	if (btrfs_header_level(buf) == 0 && check_leaf(root, buf)) {
4106
		btrfs_print_leaf(fs_info, buf);
4107 4108 4109
		ASSERT(0);
	}
#endif
4110 4111
}

4112
static void __btrfs_btree_balance_dirty(struct btrfs_fs_info *fs_info,
4113
					int flush_delayed)
4114 4115 4116 4117 4118
{
	/*
	 * looks as though older kernels can get into trouble with
	 * this code, they end up stuck in balance_dirty_pages forever
	 */
4119
	int ret;
4120 4121 4122 4123

	if (current->flags & PF_MEMALLOC)
		return;

4124
	if (flush_delayed)
4125
		btrfs_balance_delayed_items(fs_info);
4126

4127
	ret = percpu_counter_compare(&fs_info->dirty_metadata_bytes,
4128 4129
				     BTRFS_DIRTY_METADATA_THRESH);
	if (ret > 0) {
4130
		balance_dirty_pages_ratelimited(fs_info->btree_inode->i_mapping);
4131 4132 4133
	}
}

4134
void btrfs_btree_balance_dirty(struct btrfs_fs_info *fs_info)
C
Chris Mason 已提交
4135
{
4136
	__btrfs_btree_balance_dirty(fs_info, 1);
4137
}
4138

4139
void btrfs_btree_balance_dirty_nodelay(struct btrfs_fs_info *fs_info)
4140
{
4141
	__btrfs_btree_balance_dirty(fs_info, 0);
C
Chris Mason 已提交
4142
}
4143

4144
int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid)
4145
{
4146
	struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root;
4147 4148 4149
	struct btrfs_fs_info *fs_info = root->fs_info;

	return btree_read_extent_buffer_pages(fs_info, buf, parent_transid);
4150
}
4151

4152
static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info)
L
liubo 已提交
4153
{
D
David Sterba 已提交
4154
	struct btrfs_super_block *sb = fs_info->super_copy;
4155 4156
	u64 nodesize = btrfs_super_nodesize(sb);
	u64 sectorsize = btrfs_super_sectorsize(sb);
D
David Sterba 已提交
4157 4158
	int ret = 0;

4159
	if (btrfs_super_magic(sb) != BTRFS_MAGIC) {
4160
		btrfs_err(fs_info, "no valid FS found");
4161 4162 4163
		ret = -EINVAL;
	}
	if (btrfs_super_flags(sb) & ~BTRFS_SUPER_FLAG_SUPP)
4164
		btrfs_warn(fs_info, "unrecognized super flag: %llu",
4165
				btrfs_super_flags(sb) & ~BTRFS_SUPER_FLAG_SUPP);
4166
	if (btrfs_super_root_level(sb) >= BTRFS_MAX_LEVEL) {
4167
		btrfs_err(fs_info, "tree_root level too big: %d >= %d",
4168
				btrfs_super_root_level(sb), BTRFS_MAX_LEVEL);
D
David Sterba 已提交
4169 4170
		ret = -EINVAL;
	}
4171
	if (btrfs_super_chunk_root_level(sb) >= BTRFS_MAX_LEVEL) {
4172
		btrfs_err(fs_info, "chunk_root level too big: %d >= %d",
4173
				btrfs_super_chunk_root_level(sb), BTRFS_MAX_LEVEL);
D
David Sterba 已提交
4174 4175
		ret = -EINVAL;
	}
4176
	if (btrfs_super_log_root_level(sb) >= BTRFS_MAX_LEVEL) {
4177
		btrfs_err(fs_info, "log_root level too big: %d >= %d",
4178
				btrfs_super_log_root_level(sb), BTRFS_MAX_LEVEL);
D
David Sterba 已提交
4179 4180 4181
		ret = -EINVAL;
	}

D
David Sterba 已提交
4182
	/*
4183 4184
	 * 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
4185
	 */
4186 4187
	if (!is_power_of_2(sectorsize) || sectorsize < 4096 ||
	    sectorsize > BTRFS_MAX_METADATA_BLOCKSIZE) {
4188
		btrfs_err(fs_info, "invalid sectorsize %llu", sectorsize);
4189 4190 4191
		ret = -EINVAL;
	}
	/* Only PAGE SIZE is supported yet */
4192
	if (sectorsize != PAGE_SIZE) {
4193 4194 4195
		btrfs_err(fs_info,
			"sectorsize %llu not supported yet, only support %lu",
			sectorsize, PAGE_SIZE);
4196 4197 4198 4199
		ret = -EINVAL;
	}
	if (!is_power_of_2(nodesize) || nodesize < sectorsize ||
	    nodesize > BTRFS_MAX_METADATA_BLOCKSIZE) {
4200
		btrfs_err(fs_info, "invalid nodesize %llu", nodesize);
4201 4202 4203
		ret = -EINVAL;
	}
	if (nodesize != le32_to_cpu(sb->__unused_leafsize)) {
4204 4205
		btrfs_err(fs_info, "invalid leafsize %u, should be %llu",
			  le32_to_cpu(sb->__unused_leafsize), nodesize);
4206 4207 4208 4209 4210
		ret = -EINVAL;
	}

	/* Root alignment check */
	if (!IS_ALIGNED(btrfs_super_root(sb), sectorsize)) {
4211 4212
		btrfs_warn(fs_info, "tree_root block unaligned: %llu",
			   btrfs_super_root(sb));
4213 4214 4215
		ret = -EINVAL;
	}
	if (!IS_ALIGNED(btrfs_super_chunk_root(sb), sectorsize)) {
4216 4217
		btrfs_warn(fs_info, "chunk_root block unaligned: %llu",
			   btrfs_super_chunk_root(sb));
4218 4219
		ret = -EINVAL;
	}
4220
	if (!IS_ALIGNED(btrfs_super_log_root(sb), sectorsize)) {
4221 4222
		btrfs_warn(fs_info, "log_root block unaligned: %llu",
			   btrfs_super_log_root(sb));
4223 4224 4225
		ret = -EINVAL;
	}

D
David Sterba 已提交
4226
	if (memcmp(fs_info->fsid, sb->dev_item.fsid, BTRFS_UUID_SIZE) != 0) {
4227 4228 4229
		btrfs_err(fs_info,
			   "dev_item UUID does not match fsid: %pU != %pU",
			   fs_info->fsid, sb->dev_item.fsid);
D
David Sterba 已提交
4230 4231 4232 4233 4234 4235 4236
		ret = -EINVAL;
	}

	/*
	 * Hint to catch really bogus numbers, bitflips or so, more exact checks are
	 * done later
	 */
4237 4238
	if (btrfs_super_bytes_used(sb) < 6 * btrfs_super_nodesize(sb)) {
		btrfs_err(fs_info, "bytes_used is too small %llu",
4239
			  btrfs_super_bytes_used(sb));
4240 4241
		ret = -EINVAL;
	}
4242
	if (!is_power_of_2(btrfs_super_stripesize(sb))) {
4243
		btrfs_err(fs_info, "invalid stripesize %u",
4244
			  btrfs_super_stripesize(sb));
4245 4246
		ret = -EINVAL;
	}
4247
	if (btrfs_super_num_devices(sb) > (1UL << 31))
4248 4249
		btrfs_warn(fs_info, "suspicious number of devices: %llu",
			   btrfs_super_num_devices(sb));
4250
	if (btrfs_super_num_devices(sb) == 0) {
4251
		btrfs_err(fs_info, "number of devices is 0");
4252 4253
		ret = -EINVAL;
	}
D
David Sterba 已提交
4254

4255
	if (btrfs_super_bytenr(sb) != BTRFS_SUPER_INFO_OFFSET) {
4256 4257
		btrfs_err(fs_info, "super offset mismatch %llu != %u",
			  btrfs_super_bytenr(sb), BTRFS_SUPER_INFO_OFFSET);
D
David Sterba 已提交
4258 4259 4260
		ret = -EINVAL;
	}

4261 4262 4263 4264 4265
	/*
	 * 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) {
4266 4267 4268
		btrfs_err(fs_info, "system chunk array too big %u > %u",
			  btrfs_super_sys_array_size(sb),
			  BTRFS_SYSTEM_CHUNK_ARRAY_SIZE);
4269 4270 4271 4272
		ret = -EINVAL;
	}
	if (btrfs_super_sys_array_size(sb) < sizeof(struct btrfs_disk_key)
			+ sizeof(struct btrfs_chunk)) {
4273 4274 4275 4276
		btrfs_err(fs_info, "system chunk array too small %u < %zu",
			  btrfs_super_sys_array_size(sb),
			  sizeof(struct btrfs_disk_key)
			  + sizeof(struct btrfs_chunk));
4277 4278 4279
		ret = -EINVAL;
	}

D
David Sterba 已提交
4280 4281 4282 4283
	/*
	 * 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.
	 */
4284
	if (btrfs_super_generation(sb) < btrfs_super_chunk_root_generation(sb))
4285 4286 4287 4288
		btrfs_warn(fs_info,
			"suspicious: generation < chunk_root_generation: %llu < %llu",
			btrfs_super_generation(sb),
			btrfs_super_chunk_root_generation(sb));
4289 4290
	if (btrfs_super_generation(sb) < btrfs_super_cache_generation(sb)
	    && btrfs_super_cache_generation(sb) != (u64)-1)
4291 4292 4293 4294
		btrfs_warn(fs_info,
			"suspicious: generation < cache_generation: %llu < %llu",
			btrfs_super_generation(sb),
			btrfs_super_cache_generation(sb));
D
David Sterba 已提交
4295 4296

	return ret;
L
liubo 已提交
4297 4298
}

4299
static void btrfs_error_commit_super(struct btrfs_fs_info *fs_info)
L
liubo 已提交
4300
{
4301
	mutex_lock(&fs_info->cleaner_mutex);
4302
	btrfs_run_delayed_iputs(fs_info);
4303
	mutex_unlock(&fs_info->cleaner_mutex);
L
liubo 已提交
4304

4305 4306
	down_write(&fs_info->cleanup_work_sem);
	up_write(&fs_info->cleanup_work_sem);
L
liubo 已提交
4307 4308

	/* cleanup FS via transaction */
4309
	btrfs_cleanup_transaction(fs_info);
L
liubo 已提交
4310 4311
}

4312
static void btrfs_destroy_ordered_extents(struct btrfs_root *root)
L
liubo 已提交
4313 4314 4315
{
	struct btrfs_ordered_extent *ordered;

4316
	spin_lock(&root->ordered_extent_lock);
4317 4318 4319 4320
	/*
	 * This will just short circuit the ordered completion stuff which will
	 * make sure the ordered extent gets properly cleaned up.
	 */
4321
	list_for_each_entry(ordered, &root->ordered_extents,
4322 4323
			    root_extent_list)
		set_bit(BTRFS_ORDERED_IOERR, &ordered->flags);
4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338
	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);
4339 4340
		list_move_tail(&root->ordered_root,
			       &fs_info->ordered_roots);
4341

4342
		spin_unlock(&fs_info->ordered_root_lock);
4343 4344
		btrfs_destroy_ordered_extents(root);

4345 4346
		cond_resched();
		spin_lock(&fs_info->ordered_root_lock);
4347 4348
	}
	spin_unlock(&fs_info->ordered_root_lock);
L
liubo 已提交
4349 4350
}

4351
static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
4352
				      struct btrfs_fs_info *fs_info)
L
liubo 已提交
4353 4354 4355 4356 4357 4358 4359 4360 4361
{
	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);
4362
	if (atomic_read(&delayed_refs->num_entries) == 0) {
4363
		spin_unlock(&delayed_refs->lock);
4364
		btrfs_info(fs_info, "delayed_refs has NO entry");
L
liubo 已提交
4365 4366 4367
		return ret;
	}

4368 4369
	while ((node = rb_first(&delayed_refs->href_root)) != NULL) {
		struct btrfs_delayed_ref_head *head;
4370
		struct btrfs_delayed_ref_node *tmp;
4371
		bool pin_bytes = false;
L
liubo 已提交
4372

4373 4374 4375
		head = rb_entry(node, struct btrfs_delayed_ref_head,
				href_node);
		if (!mutex_trylock(&head->mutex)) {
4376
			refcount_inc(&head->node.refs);
4377
			spin_unlock(&delayed_refs->lock);
4378

4379
			mutex_lock(&head->mutex);
4380
			mutex_unlock(&head->mutex);
4381 4382 4383 4384 4385
			btrfs_put_delayed_ref(&head->node);
			spin_lock(&delayed_refs->lock);
			continue;
		}
		spin_lock(&head->lock);
4386 4387
		list_for_each_entry_safe_reverse(ref, tmp, &head->ref_list,
						 list) {
4388
			ref->in_tree = 0;
4389
			list_del(&ref->list);
4390 4391
			if (!list_empty(&ref->add_list))
				list_del(&ref->add_list);
4392 4393
			atomic_dec(&delayed_refs->num_entries);
			btrfs_put_delayed_ref(ref);
4394
		}
4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406
		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
4407

4408
		if (pin_bytes)
4409
			btrfs_pin_extent(fs_info, head->node.bytenr,
4410 4411
					 head->node.num_bytes, 1);
		btrfs_put_delayed_ref(&head->node);
L
liubo 已提交
4412 4413 4414 4415 4416 4417 4418 4419 4420
		cond_resched();
		spin_lock(&delayed_refs->lock);
	}

	spin_unlock(&delayed_refs->lock);

	return ret;
}

4421
static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root)
L
liubo 已提交
4422 4423 4424 4425 4426 4427
{
	struct btrfs_inode *btrfs_inode;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

4428 4429
	spin_lock(&root->delalloc_lock);
	list_splice_init(&root->delalloc_inodes, &splice);
L
liubo 已提交
4430 4431

	while (!list_empty(&splice)) {
4432 4433
		btrfs_inode = list_first_entry(&splice, struct btrfs_inode,
					       delalloc_inodes);
L
liubo 已提交
4434 4435

		list_del_init(&btrfs_inode->delalloc_inodes);
4436 4437
		clear_bit(BTRFS_INODE_IN_DELALLOC_LIST,
			  &btrfs_inode->runtime_flags);
4438
		spin_unlock(&root->delalloc_lock);
L
liubo 已提交
4439 4440

		btrfs_invalidate_inodes(btrfs_inode->root);
4441

4442
		spin_lock(&root->delalloc_lock);
L
liubo 已提交
4443 4444
	}

4445 4446 4447 4448 4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463 4464 4465 4466 4467 4468 4469 4470
	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
4471 4472
}

4473
static int btrfs_destroy_marked_extents(struct btrfs_fs_info *fs_info,
L
liubo 已提交
4474 4475 4476 4477 4478 4479 4480 4481 4482 4483
					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,
4484
					    mark, NULL);
L
liubo 已提交
4485 4486 4487
		if (ret)
			break;

4488
		clear_extent_bits(dirty_pages, start, end, mark);
L
liubo 已提交
4489
		while (start <= end) {
4490 4491
			eb = find_extent_buffer(fs_info, start);
			start += fs_info->nodesize;
4492
			if (!eb)
L
liubo 已提交
4493
				continue;
4494
			wait_on_extent_buffer_writeback(eb);
L
liubo 已提交
4495

4496 4497 4498 4499
			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
4500 4501 4502 4503 4504 4505
		}
	}

	return ret;
}

4506
static int btrfs_destroy_pinned_extent(struct btrfs_fs_info *fs_info,
L
liubo 已提交
4507 4508 4509 4510 4511 4512
				       struct extent_io_tree *pinned_extents)
{
	struct extent_io_tree *unpin;
	u64 start;
	u64 end;
	int ret;
4513
	bool loop = true;
L
liubo 已提交
4514 4515

	unpin = pinned_extents;
4516
again:
L
liubo 已提交
4517 4518
	while (1) {
		ret = find_first_extent_bit(unpin, 0, &start, &end,
4519
					    EXTENT_DIRTY, NULL);
L
liubo 已提交
4520 4521 4522
		if (ret)
			break;

4523
		clear_extent_dirty(unpin, start, end);
4524
		btrfs_error_unpin_extent_range(fs_info, start, end);
L
liubo 已提交
4525 4526 4527
		cond_resched();
	}

4528
	if (loop) {
4529 4530
		if (unpin == &fs_info->freed_extents[0])
			unpin = &fs_info->freed_extents[1];
4531
		else
4532
			unpin = &fs_info->freed_extents[0];
4533 4534 4535 4536
		loop = false;
		goto again;
	}

L
liubo 已提交
4537 4538 4539
	return 0;
}

4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553 4554
static void btrfs_cleanup_bg_io(struct btrfs_block_group_cache *cache)
{
	struct inode *inode;

	inode = cache->io_ctl.inode;
	if (inode) {
		invalidate_inode_pages2(inode->i_mapping);
		BTRFS_I(inode)->generation = 0;
		cache->io_ctl.inode = NULL;
		iput(inode);
	}
	btrfs_put_block_group(cache);
}

void btrfs_cleanup_dirty_bgs(struct btrfs_transaction *cur_trans,
4555
			     struct btrfs_fs_info *fs_info)
4556 4557 4558 4559 4560 4561 4562 4563 4564
{
	struct btrfs_block_group_cache *cache;

	spin_lock(&cur_trans->dirty_bgs_lock);
	while (!list_empty(&cur_trans->dirty_bgs)) {
		cache = list_first_entry(&cur_trans->dirty_bgs,
					 struct btrfs_block_group_cache,
					 dirty_list);
		if (!cache) {
4565
			btrfs_err(fs_info, "orphan block group dirty_bgs list");
4566 4567 4568 4569 4570 4571 4572 4573 4574 4575 4576 4577 4578 4579 4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592
			spin_unlock(&cur_trans->dirty_bgs_lock);
			return;
		}

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

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

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

	while (!list_empty(&cur_trans->io_bgs)) {
		cache = list_first_entry(&cur_trans->io_bgs,
					 struct btrfs_block_group_cache,
					 io_list);
		if (!cache) {
4593
			btrfs_err(fs_info, "orphan block group on io_bgs list");
4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604
			return;
		}

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

4605
void btrfs_cleanup_one_transaction(struct btrfs_transaction *cur_trans,
4606
				   struct btrfs_fs_info *fs_info)
4607
{
4608
	btrfs_cleanup_dirty_bgs(cur_trans, fs_info);
4609 4610 4611
	ASSERT(list_empty(&cur_trans->dirty_bgs));
	ASSERT(list_empty(&cur_trans->io_bgs));

4612
	btrfs_destroy_delayed_refs(cur_trans, fs_info);
4613

4614
	cur_trans->state = TRANS_STATE_COMMIT_START;
4615
	wake_up(&fs_info->transaction_blocked_wait);
4616

4617
	cur_trans->state = TRANS_STATE_UNBLOCKED;
4618
	wake_up(&fs_info->transaction_wait);
4619

4620 4621
	btrfs_destroy_delayed_inodes(fs_info);
	btrfs_assert_delayed_root_empty(fs_info);
4622

4623
	btrfs_destroy_marked_extents(fs_info, &cur_trans->dirty_pages,
4624
				     EXTENT_DIRTY);
4625
	btrfs_destroy_pinned_extent(fs_info,
4626
				    fs_info->pinned_extents);
4627

4628 4629 4630
	cur_trans->state =TRANS_STATE_COMPLETED;
	wake_up(&cur_trans->commit_wait);

4631 4632 4633 4634 4635 4636
	/*
	memset(cur_trans, 0, sizeof(*cur_trans));
	kmem_cache_free(btrfs_transaction_cachep, cur_trans);
	*/
}

4637
static int btrfs_cleanup_transaction(struct btrfs_fs_info *fs_info)
L
liubo 已提交
4638 4639 4640
{
	struct btrfs_transaction *t;

4641
	mutex_lock(&fs_info->transaction_kthread_mutex);
L
liubo 已提交
4642

4643 4644 4645
	spin_lock(&fs_info->trans_lock);
	while (!list_empty(&fs_info->trans_list)) {
		t = list_first_entry(&fs_info->trans_list,
4646 4647
				     struct btrfs_transaction, list);
		if (t->state >= TRANS_STATE_COMMIT_START) {
4648
			refcount_inc(&t->use_count);
4649
			spin_unlock(&fs_info->trans_lock);
4650
			btrfs_wait_for_commit(fs_info, t->transid);
4651
			btrfs_put_transaction(t);
4652
			spin_lock(&fs_info->trans_lock);
4653 4654
			continue;
		}
4655
		if (t == fs_info->running_transaction) {
4656
			t->state = TRANS_STATE_COMMIT_DOING;
4657
			spin_unlock(&fs_info->trans_lock);
4658 4659 4660 4661 4662 4663 4664
			/*
			 * 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 {
4665
			spin_unlock(&fs_info->trans_lock);
4666
		}
4667
		btrfs_cleanup_one_transaction(t, fs_info);
4668

4669 4670 4671
		spin_lock(&fs_info->trans_lock);
		if (t == fs_info->running_transaction)
			fs_info->running_transaction = NULL;
L
liubo 已提交
4672
		list_del_init(&t->list);
4673
		spin_unlock(&fs_info->trans_lock);
L
liubo 已提交
4674

4675
		btrfs_put_transaction(t);
4676
		trace_btrfs_transaction_commit(fs_info->tree_root);
4677
		spin_lock(&fs_info->trans_lock);
4678
	}
4679 4680
	spin_unlock(&fs_info->trans_lock);
	btrfs_destroy_all_ordered_extents(fs_info);
4681 4682
	btrfs_destroy_delayed_inodes(fs_info);
	btrfs_assert_delayed_root_empty(fs_info);
4683
	btrfs_destroy_pinned_extent(fs_info, fs_info->pinned_extents);
4684 4685
	btrfs_destroy_all_delalloc_inodes(fs_info);
	mutex_unlock(&fs_info->transaction_kthread_mutex);
L
liubo 已提交
4686 4687 4688 4689

	return 0;
}

4690
static const struct extent_io_ops btree_extent_io_ops = {
4691
	/* mandatory callbacks */
4692
	.submit_bio_hook = btree_submit_bio_hook,
4693
	.readpage_end_io_hook = btree_readpage_end_io_hook,
4694 4695
	/* note we're sharing with inode.c for the merge bio hook */
	.merge_bio_hook = btrfs_merge_bio_hook,
4696
	.readpage_io_failed_hook = btree_io_failed_hook,
4697 4698

	/* optional callbacks */
4699
};