disk-io.c 108.1 KB
Newer Older
C
Chris Mason 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
/*
 * Copyright (C) 2007 Oracle.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License v2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */

C
Chris Mason 已提交
19
#include <linux/fs.h>
20
#include <linux/blkdev.h>
C
Chris Mason 已提交
21
#include <linux/scatterlist.h>
C
Chris Mason 已提交
22
#include <linux/swap.h>
23
#include <linux/radix-tree.h>
C
Chris Mason 已提交
24
#include <linux/writeback.h>
C
Chris Mason 已提交
25
#include <linux/buffer_head.h>
26
#include <linux/workqueue.h>
27
#include <linux/kthread.h>
C
Chris Mason 已提交
28
#include <linux/freezer.h>
29
#include <linux/crc32c.h>
30
#include <linux/slab.h>
31
#include <linux/migrate.h>
32
#include <linux/ratelimit.h>
33
#include <linux/uuid.h>
34
#include <asm/unaligned.h>
C
Chris Mason 已提交
35
#include "compat.h"
36 37
#include "ctree.h"
#include "disk-io.h"
38
#include "transaction.h"
39
#include "btrfs_inode.h"
40
#include "volumes.h"
41
#include "print-tree.h"
42
#include "async-thread.h"
43
#include "locking.h"
44
#include "tree-log.h"
45
#include "free-space-cache.h"
46
#include "inode-map.h"
47
#include "check-integrity.h"
48
#include "rcu-string.h"
49
#include "dev-replace.h"
D
David Woodhouse 已提交
50
#include "raid56.h"
51

52 53 54 55
#ifdef CONFIG_X86
#include <asm/cpufeature.h>
#endif

56
static struct extent_io_ops btree_extent_io_ops;
57
static void end_workqueue_fn(struct btrfs_work *work);
58
static void free_fs_root(struct btrfs_root *root);
59
static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info,
L
liubo 已提交
60
				    int read_only);
61 62
static void btrfs_destroy_ordered_operations(struct btrfs_transaction *t,
					     struct btrfs_root *root);
63
static void btrfs_destroy_ordered_extents(struct btrfs_root *root);
L
liubo 已提交
64 65
static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
				      struct btrfs_root *root);
66
static void btrfs_evict_pending_snapshots(struct btrfs_transaction *t);
67
static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root);
L
liubo 已提交
68 69 70 71 72
static int btrfs_destroy_marked_extents(struct btrfs_root *root,
					struct extent_io_tree *dirty_pages,
					int mark);
static int btrfs_destroy_pinned_extent(struct btrfs_root *root,
				       struct extent_io_tree *pinned_extents);
73 74
static int btrfs_cleanup_transaction(struct btrfs_root *root);
static void btrfs_error_commit_super(struct btrfs_root *root);
75

C
Chris Mason 已提交
76 77 78 79 80
/*
 * 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
 * by writes to insert metadata for new file extents after IO is complete.
 */
81 82 83 84 85 86
struct end_io_wq {
	struct bio *bio;
	bio_end_io_t *end_io;
	void *private;
	struct btrfs_fs_info *info;
	int error;
87
	int metadata;
88
	struct list_head list;
89
	struct btrfs_work work;
90
};
91

C
Chris Mason 已提交
92 93 94 95 96
/*
 * 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.
 */
97 98 99 100
struct async_submit_bio {
	struct inode *inode;
	struct bio *bio;
	struct list_head list;
C
Chris Mason 已提交
101 102
	extent_submit_bio_hook_t *submit_bio_start;
	extent_submit_bio_hook_t *submit_bio_done;
103 104
	int rw;
	int mirror_num;
C
Chris Mason 已提交
105
	unsigned long bio_flags;
106 107 108 109 110
	/*
	 * 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;
111
	struct btrfs_work work;
112
	int error;
113 114
};

115 116 117 118 119 120 121 122 123 124
/*
 * 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.
125
 *
126 127 128
 * 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.
129
 *
130 131 132
 * 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.
133
 *
134 135 136
 * 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.
137 138 139 140 141
 */
#ifdef CONFIG_DEBUG_LOCK_ALLOC
# if BTRFS_MAX_LEVEL != 8
#  error
# endif
142 143 144 145 146 147 148 149 150 151 152 153 154

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"	},
155
	{ .id = BTRFS_QUOTA_TREE_OBJECTID,	.name_stem = "quota"	},
156 157 158 159
	{ .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"	},
	{ .id = 0,				.name_stem = "tree"	},
160
};
161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191

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]);
}

192 193
#endif

C
Chris Mason 已提交
194 195 196 197
/*
 * extents on the btree inode are pretty simple, there's one extent
 * that covers the entire device
 */
198
static struct extent_map *btree_get_extent(struct inode *inode,
199
		struct page *page, size_t pg_offset, u64 start, u64 len,
200
		int create)
201
{
202 203 204 205
	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
	struct extent_map *em;
	int ret;

206
	read_lock(&em_tree->lock);
207
	em = lookup_extent_mapping(em_tree, start, len);
208 209 210
	if (em) {
		em->bdev =
			BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
211
		read_unlock(&em_tree->lock);
212
		goto out;
213
	}
214
	read_unlock(&em_tree->lock);
215

216
	em = alloc_extent_map();
217 218 219 220 221
	if (!em) {
		em = ERR_PTR(-ENOMEM);
		goto out;
	}
	em->start = 0;
222
	em->len = (u64)-1;
C
Chris Mason 已提交
223
	em->block_len = (u64)-1;
224
	em->block_start = 0;
225
	em->bdev = BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
226

227
	write_lock(&em_tree->lock);
J
Josef Bacik 已提交
228
	ret = add_extent_mapping(em_tree, em, 0);
229 230
	if (ret == -EEXIST) {
		free_extent_map(em);
231
		em = lookup_extent_mapping(em_tree, start, len);
232
		if (!em)
233
			em = ERR_PTR(-EIO);
234
	} else if (ret) {
235
		free_extent_map(em);
236
		em = ERR_PTR(ret);
237
	}
238
	write_unlock(&em_tree->lock);
239

240 241
out:
	return em;
242 243
}

244
u32 btrfs_csum_data(char *data, u32 seed, size_t len)
245
{
246
	return crc32c(seed, data, len);
247 248 249 250
}

void btrfs_csum_final(u32 crc, char *result)
{
251
	put_unaligned_le32(~crc, result);
252 253
}

C
Chris Mason 已提交
254 255 256 257
/*
 * compute the csum for a btree block, and either verify it or write it
 * into the csum field of the block.
 */
258 259 260
static int csum_tree_block(struct btrfs_root *root, struct extent_buffer *buf,
			   int verify)
{
261
	u16 csum_size = btrfs_super_csum_size(root->fs_info->super_copy);
262
	char *result = NULL;
263 264 265 266 267 268 269 270
	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;
271
	unsigned long inline_result;
272 273

	len = buf->len - offset;
C
Chris Mason 已提交
274
	while (len > 0) {
275
		err = map_private_extent_buffer(buf, offset, 32,
276
					&kaddr, &map_start, &map_len);
C
Chris Mason 已提交
277
		if (err)
278 279
			return 1;
		cur_len = min(len, map_len - (offset - map_start));
280
		crc = btrfs_csum_data(kaddr + offset - map_start,
281 282 283 284
				      crc, cur_len);
		len -= cur_len;
		offset += cur_len;
	}
285 286 287 288 289 290 291 292
	if (csum_size > sizeof(inline_result)) {
		result = kzalloc(csum_size * sizeof(char), GFP_NOFS);
		if (!result)
			return 1;
	} else {
		result = (char *)&inline_result;
	}

293 294 295
	btrfs_csum_final(crc, result);

	if (verify) {
296
		if (memcmp_extent_buffer(buf, result, 0, csum_size)) {
297 298
			u32 val;
			u32 found = 0;
299
			memcpy(&found, result, csum_size);
300

301
			read_extent_buffer(buf, &val, 0, csum_size);
302
			printk_ratelimited(KERN_INFO "btrfs: %s checksum verify "
C
Chris Mason 已提交
303 304 305 306 307
				       "failed on %llu wanted %X found %X "
				       "level %d\n",
				       root->fs_info->sb->s_id,
				       (unsigned long long)buf->start, val, found,
				       btrfs_header_level(buf));
308 309
			if (result != (char *)&inline_result)
				kfree(result);
310 311 312
			return 1;
		}
	} else {
313
		write_extent_buffer(buf, result, 0, csum_size);
314
	}
315 316
	if (result != (char *)&inline_result)
		kfree(result);
317 318 319
	return 0;
}

C
Chris Mason 已提交
320 321 322 323 324 325
/*
 * 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.
 */
326
static int verify_parent_transid(struct extent_io_tree *io_tree,
327 328
				 struct extent_buffer *eb, u64 parent_transid,
				 int atomic)
329
{
330
	struct extent_state *cached_state = NULL;
331 332 333 334 335
	int ret;

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

336 337 338
	if (atomic)
		return -EAGAIN;

339
	lock_extent_bits(io_tree, eb->start, eb->start + eb->len - 1,
340
			 0, &cached_state);
341
	if (extent_buffer_uptodate(eb) &&
342 343 344 345
	    btrfs_header_generation(eb) == parent_transid) {
		ret = 0;
		goto out;
	}
346
	printk_ratelimited("parent transid verify failed on %llu wanted %llu "
C
Chris Mason 已提交
347 348 349 350
		       "found %llu\n",
		       (unsigned long long)eb->start,
		       (unsigned long long)parent_transid,
		       (unsigned long long)btrfs_header_generation(eb));
351
	ret = 1;
352
	clear_extent_buffer_uptodate(eb);
C
Chris Mason 已提交
353
out:
354 355
	unlock_extent_cached(io_tree, eb->start, eb->start + eb->len - 1,
			     &cached_state, GFP_NOFS);
356 357 358
	return ret;
}

D
David Sterba 已提交
359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385
/*
 * Return 0 if the superblock checksum type matches the checksum value of that
 * algorithm. Pass the raw disk superblock data.
 */
static int btrfs_check_super_csum(char *raw_disk_sb)
{
	struct btrfs_super_block *disk_sb =
		(struct btrfs_super_block *)raw_disk_sb;
	u16 csum_type = btrfs_super_csum_type(disk_sb);
	int ret = 0;

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

		/*
		 * The super_block structure does not span the whole
		 * BTRFS_SUPER_INFO_SIZE range, we expect that the unused space
		 * is filled with zeros and is included in the checkum.
		 */
		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;
386 387 388 389 390

		if (ret && btrfs_super_generation(disk_sb) < 10) {
			printk(KERN_WARNING "btrfs: super block crcs don't match, older mkfs detected\n");
			ret = 0;
		}
D
David Sterba 已提交
391 392 393 394 395 396 397 398 399 400 401
	}

	if (csum_type >= ARRAY_SIZE(btrfs_csum_sizes)) {
		printk(KERN_ERR "btrfs: unsupported checksum algorithm %u\n",
				csum_type);
		ret = 1;
	}

	return ret;
}

C
Chris Mason 已提交
402 403 404 405
/*
 * helper to read a given tree block, doing retries as required when
 * the checksums don't match and we have alternate mirrors to try.
 */
406 407
static int btree_read_extent_buffer_pages(struct btrfs_root *root,
					  struct extent_buffer *eb,
408
					  u64 start, u64 parent_transid)
409 410
{
	struct extent_io_tree *io_tree;
411
	int failed = 0;
412 413 414
	int ret;
	int num_copies = 0;
	int mirror_num = 0;
415
	int failed_mirror = 0;
416

417
	clear_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
418 419
	io_tree = &BTRFS_I(root->fs_info->btree_inode)->io_tree;
	while (1) {
420 421
		ret = read_extent_buffer_pages(io_tree, eb, start,
					       WAIT_COMPLETE,
422
					       btree_get_extent, mirror_num);
423 424
		if (!ret) {
			if (!verify_parent_transid(io_tree, eb,
425
						   parent_transid, 0))
426 427 428 429
				break;
			else
				ret = -EIO;
		}
C
Chris Mason 已提交
430

431 432 433 434 435 436
		/*
		 * 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))
437 438
			break;

439
		num_copies = btrfs_num_copies(root->fs_info,
440
					      eb->start, eb->len);
C
Chris Mason 已提交
441
		if (num_copies == 1)
442
			break;
C
Chris Mason 已提交
443

444 445 446 447 448
		if (!failed_mirror) {
			failed = 1;
			failed_mirror = eb->read_mirror;
		}

449
		mirror_num++;
450 451 452
		if (mirror_num == failed_mirror)
			mirror_num++;

C
Chris Mason 已提交
453
		if (mirror_num > num_copies)
454
			break;
455
	}
456

457
	if (failed && !ret && failed_mirror)
458 459 460
		repair_eb_io_failure(root, eb, failed_mirror);

	return ret;
461
}
462

C
Chris Mason 已提交
463
/*
C
Chris Mason 已提交
464 465
 * 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
466
 */
C
Chris Mason 已提交
467

468
static int csum_dirty_buffer(struct btrfs_root *root, struct page *page)
469
{
470
	struct extent_io_tree *tree;
M
Miao Xie 已提交
471
	u64 start = page_offset(page);
472 473
	u64 found_start;
	struct extent_buffer *eb;
474

475
	tree = &BTRFS_I(page->mapping->host)->io_tree;
476

J
Josef Bacik 已提交
477 478 479
	eb = (struct extent_buffer *)page->private;
	if (page != eb->pages[0])
		return 0;
480 481
	found_start = btrfs_header_bytenr(eb);
	if (found_start != start) {
482
		WARN_ON(1);
J
Josef Bacik 已提交
483
		return 0;
484 485 486
	}
	if (!PageUptodate(page)) {
		WARN_ON(1);
J
Josef Bacik 已提交
487
		return 0;
488 489 490 491 492
	}
	csum_tree_block(root, eb, 0);
	return 0;
}

Y
Yan Zheng 已提交
493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511
static int check_tree_block_fsid(struct btrfs_root *root,
				 struct extent_buffer *eb)
{
	struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
	u8 fsid[BTRFS_UUID_SIZE];
	int ret = 1;

	read_extent_buffer(eb, fsid, (unsigned long)btrfs_header_fsid(eb),
			   BTRFS_FSID_SIZE);
	while (fs_devices) {
		if (!memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE)) {
			ret = 0;
			break;
		}
		fs_devices = fs_devices->seed;
	}
	return ret;
}

512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578
#define CORRUPT(reason, eb, root, slot)				\
	printk(KERN_CRIT "btrfs: corrupt leaf, %s: block=%llu,"	\
	       "root=%llu, slot=%d\n", reason,			\
	       (unsigned long long)btrfs_header_bytenr(eb),	\
	       (unsigned long long)root->objectid, slot)

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

	if (nritems == 0)
		return 0;

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

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

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

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

		/*
		 * Check to make sure that we don't point outside of the leaf,
		 * just incase all the items are consistent to eachother, but
		 * all point outside of the leaf.
		 */
		if (btrfs_item_end_nr(leaf, slot) >
		    BTRFS_LEAF_DATA_SIZE(root)) {
			CORRUPT("slot end outside of leaf", leaf, root, slot);
			return -EIO;
		}
	}

	return 0;
}

579
static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end,
580
			       struct extent_state *state, int mirror)
581 582 583 584 585 586
{
	struct extent_io_tree *tree;
	u64 found_start;
	int found_level;
	struct extent_buffer *eb;
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
587
	int ret = 0;
588
	int reads_done;
589 590 591

	if (!page->private)
		goto out;
C
Chris Mason 已提交
592

593
	tree = &BTRFS_I(page->mapping->host)->io_tree;
J
Josef Bacik 已提交
594
	eb = (struct extent_buffer *)page->private;
C
Chris Mason 已提交
595

596 597 598 599 600 601
	/* 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);
602 603
	if (!reads_done)
		goto err;
604

605
	eb->read_mirror = mirror;
606 607 608 609 610
	if (test_bit(EXTENT_BUFFER_IOERR, &eb->bflags)) {
		ret = -EIO;
		goto err;
	}

611
	found_start = btrfs_header_bytenr(eb);
612
	if (found_start != eb->start) {
613
		printk_ratelimited(KERN_INFO "btrfs bad tree block start "
C
Chris Mason 已提交
614 615 616
			       "%llu %llu\n",
			       (unsigned long long)found_start,
			       (unsigned long long)eb->start);
617
		ret = -EIO;
618 619
		goto err;
	}
Y
Yan Zheng 已提交
620
	if (check_tree_block_fsid(root, eb)) {
621
		printk_ratelimited(KERN_INFO "btrfs bad fsid on block %llu\n",
C
Chris Mason 已提交
622
			       (unsigned long long)eb->start);
623 624 625
		ret = -EIO;
		goto err;
	}
626
	found_level = btrfs_header_level(eb);
627 628 629 630 631 632
	if (found_level >= BTRFS_MAX_LEVEL) {
		btrfs_info(root->fs_info, "bad tree block level %d\n",
			   (int)btrfs_header_level(eb));
		ret = -EIO;
		goto err;
	}
633

634 635
	btrfs_set_buffer_lockdep_class(btrfs_header_owner(eb),
				       eb, found_level);
636

637
	ret = csum_tree_block(root, eb, 1);
638
	if (ret) {
639
		ret = -EIO;
640 641 642 643 644 645 646 647 648 649 650 651
		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;
	}
652

653 654
	if (!ret)
		set_extent_buffer_uptodate(eb);
655
err:
656 657
	if (reads_done &&
	    test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
A
Arne Jansen 已提交
658 659
		btree_readahead_hook(root, eb, eb->start, ret);

D
David Woodhouse 已提交
660 661 662 663 664 665 666
	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);
667
		clear_extent_buffer_uptodate(eb);
D
David Woodhouse 已提交
668
	}
669
	free_extent_buffer(eb);
670
out:
671
	return ret;
672 673
}

674
static int btree_io_failed_hook(struct page *page, int failed_mirror)
A
Arne Jansen 已提交
675 676 677 678
{
	struct extent_buffer *eb;
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;

J
Josef Bacik 已提交
679
	eb = (struct extent_buffer *)page->private;
680
	set_bit(EXTENT_BUFFER_IOERR, &eb->bflags);
681
	eb->read_mirror = failed_mirror;
D
David Woodhouse 已提交
682
	atomic_dec(&eb->io_pages);
683
	if (test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
A
Arne Jansen 已提交
684 685 686 687
		btree_readahead_hook(root, eb, eb->start, -EIO);
	return -EIO;	/* we fixed nothing */
}

688 689 690 691 692 693 694
static void end_workqueue_bio(struct bio *bio, int err)
{
	struct end_io_wq *end_io_wq = bio->bi_private;
	struct btrfs_fs_info *fs_info;

	fs_info = end_io_wq->info;
	end_io_wq->error = err;
695 696
	end_io_wq->work.func = end_workqueue_fn;
	end_io_wq->work.flags = 0;
697

698
	if (bio->bi_rw & REQ_WRITE) {
D
David Woodhouse 已提交
699
		if (end_io_wq->metadata == BTRFS_WQ_ENDIO_METADATA)
700 701
			btrfs_queue_worker(&fs_info->endio_meta_write_workers,
					   &end_io_wq->work);
D
David Woodhouse 已提交
702
		else if (end_io_wq->metadata == BTRFS_WQ_ENDIO_FREE_SPACE)
J
Josef Bacik 已提交
703 704
			btrfs_queue_worker(&fs_info->endio_freespace_worker,
					   &end_io_wq->work);
D
David Woodhouse 已提交
705 706 707
		else if (end_io_wq->metadata == BTRFS_WQ_ENDIO_RAID56)
			btrfs_queue_worker(&fs_info->endio_raid56_workers,
					   &end_io_wq->work);
708 709 710
		else
			btrfs_queue_worker(&fs_info->endio_write_workers,
					   &end_io_wq->work);
711
	} else {
D
David Woodhouse 已提交
712 713 714 715
		if (end_io_wq->metadata == BTRFS_WQ_ENDIO_RAID56)
			btrfs_queue_worker(&fs_info->endio_raid56_workers,
					   &end_io_wq->work);
		else if (end_io_wq->metadata)
716 717 718 719 720 721
			btrfs_queue_worker(&fs_info->endio_meta_workers,
					   &end_io_wq->work);
		else
			btrfs_queue_worker(&fs_info->endio_workers,
					   &end_io_wq->work);
	}
722 723
}

J
Josef Bacik 已提交
724 725 726 727 728 729
/*
 * For the metadata arg you want
 *
 * 0 - if data
 * 1 - if normal metadta
 * 2 - if writing to the free space cache area
D
David Woodhouse 已提交
730
 * 3 - raid parity work
J
Josef Bacik 已提交
731
 */
732 733
int btrfs_bio_wq_end_io(struct btrfs_fs_info *info, struct bio *bio,
			int metadata)
734
{
735 736 737 738 739 740 741
	struct end_io_wq *end_io_wq;
	end_io_wq = kmalloc(sizeof(*end_io_wq), GFP_NOFS);
	if (!end_io_wq)
		return -ENOMEM;

	end_io_wq->private = bio->bi_private;
	end_io_wq->end_io = bio->bi_end_io;
742
	end_io_wq->info = info;
743 744
	end_io_wq->error = 0;
	end_io_wq->bio = bio;
745
	end_io_wq->metadata = metadata;
746 747 748

	bio->bi_private = end_io_wq;
	bio->bi_end_io = end_workqueue_bio;
749 750 751
	return 0;
}

752
unsigned long btrfs_async_submit_limit(struct btrfs_fs_info *info)
753
{
754 755 756 757 758
	unsigned long limit = min_t(unsigned long,
				    info->workers.max_workers,
				    info->fs_devices->open_devices);
	return 256 * limit;
}
759

C
Chris Mason 已提交
760 761 762
static void run_one_async_start(struct btrfs_work *work)
{
	struct async_submit_bio *async;
763
	int ret;
C
Chris Mason 已提交
764 765

	async = container_of(work, struct  async_submit_bio, work);
766 767 768 769 770
	ret = async->submit_bio_start(async->inode, async->rw, async->bio,
				      async->mirror_num, async->bio_flags,
				      async->bio_offset);
	if (ret)
		async->error = ret;
C
Chris Mason 已提交
771 772 773
}

static void run_one_async_done(struct btrfs_work *work)
774 775 776
{
	struct btrfs_fs_info *fs_info;
	struct async_submit_bio *async;
777
	int limit;
778 779 780

	async = container_of(work, struct  async_submit_bio, work);
	fs_info = BTRFS_I(async->inode)->root->fs_info;
781

782
	limit = btrfs_async_submit_limit(fs_info);
783 784
	limit = limit * 2 / 3;

785
	if (atomic_dec_return(&fs_info->nr_async_submits) < limit &&
786
	    waitqueue_active(&fs_info->async_submit_wait))
787 788
		wake_up(&fs_info->async_submit_wait);

789 790 791 792 793 794
	/* If an error occured we just want to clean up the bio and move on */
	if (async->error) {
		bio_endio(async->bio, async->error);
		return;
	}

C
Chris Mason 已提交
795
	async->submit_bio_done(async->inode, async->rw, async->bio,
796 797
			       async->mirror_num, async->bio_flags,
			       async->bio_offset);
C
Chris Mason 已提交
798 799 800 801 802 803 804
}

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

	async = container_of(work, struct  async_submit_bio, work);
805 806 807
	kfree(async);
}

808 809
int btrfs_wq_submit_bio(struct btrfs_fs_info *fs_info, struct inode *inode,
			int rw, struct bio *bio, int mirror_num,
C
Chris Mason 已提交
810
			unsigned long bio_flags,
811
			u64 bio_offset,
C
Chris Mason 已提交
812 813
			extent_submit_bio_hook_t *submit_bio_start,
			extent_submit_bio_hook_t *submit_bio_done)
814 815 816 817 818 819 820 821 822 823 824
{
	struct async_submit_bio *async;

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

	async->inode = inode;
	async->rw = rw;
	async->bio = bio;
	async->mirror_num = mirror_num;
C
Chris Mason 已提交
825 826 827 828 829 830 831
	async->submit_bio_start = submit_bio_start;
	async->submit_bio_done = submit_bio_done;

	async->work.func = run_one_async_start;
	async->work.ordered_func = run_one_async_done;
	async->work.ordered_free = run_one_async_free;

832
	async->work.flags = 0;
C
Chris Mason 已提交
833
	async->bio_flags = bio_flags;
834
	async->bio_offset = bio_offset;
835

836 837
	async->error = 0;

838
	atomic_inc(&fs_info->nr_async_submits);
839

840
	if (rw & REQ_SYNC)
841 842
		btrfs_set_work_high_prio(&async->work);

843
	btrfs_queue_worker(&fs_info->workers, &async->work);
844

C
Chris Mason 已提交
845
	while (atomic_read(&fs_info->async_submit_draining) &&
846 847 848 849 850
	      atomic_read(&fs_info->nr_async_submits)) {
		wait_event(fs_info->async_submit_wait,
			   (atomic_read(&fs_info->nr_async_submits) == 0));
	}

851 852 853
	return 0;
}

854 855 856 857 858
static int btree_csum_one_bio(struct bio *bio)
{
	struct bio_vec *bvec = bio->bi_io_vec;
	int bio_index = 0;
	struct btrfs_root *root;
859
	int ret = 0;
860 861

	WARN_ON(bio->bi_vcnt <= 0);
C
Chris Mason 已提交
862
	while (bio_index < bio->bi_vcnt) {
863
		root = BTRFS_I(bvec->bv_page->mapping->host)->root;
864 865 866
		ret = csum_dirty_buffer(root, bvec->bv_page);
		if (ret)
			break;
867 868 869
		bio_index++;
		bvec++;
	}
870
	return ret;
871 872
}

C
Chris Mason 已提交
873 874
static int __btree_submit_bio_start(struct inode *inode, int rw,
				    struct bio *bio, int mirror_num,
875 876
				    unsigned long bio_flags,
				    u64 bio_offset)
877
{
878 879
	/*
	 * when we're called for a write, we're already in the async
880
	 * submission context.  Just jump into btrfs_map_bio
881
	 */
882
	return btree_csum_one_bio(bio);
C
Chris Mason 已提交
883
}
884

C
Chris Mason 已提交
885
static int __btree_submit_bio_done(struct inode *inode, int rw, struct bio *bio,
886 887
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
C
Chris Mason 已提交
888
{
889 890
	int ret;

891
	/*
C
Chris Mason 已提交
892 893
	 * when we're called for a write, we're already in the async
	 * submission context.  Just jump into btrfs_map_bio
894
	 */
895 896 897 898
	ret = btrfs_map_bio(BTRFS_I(inode)->root, rw, bio, mirror_num, 1);
	if (ret)
		bio_endio(bio, ret);
	return ret;
899 900
}

901 902 903 904 905 906 907 908 909 910 911
static int check_async_write(struct inode *inode, unsigned long bio_flags)
{
	if (bio_flags & EXTENT_BIO_TREE_LOG)
		return 0;
#ifdef CONFIG_X86
	if (cpu_has_xmm4_2)
		return 0;
#endif
	return 1;
}

912
static int btree_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
913 914
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
915
{
916
	int async = check_async_write(inode, bio_flags);
917 918
	int ret;

919
	if (!(rw & REQ_WRITE)) {
C
Chris Mason 已提交
920 921 922 923
		/*
		 * called for a read, do the setup so that checksum validation
		 * can happen in the async kernel threads
		 */
924 925
		ret = btrfs_bio_wq_end_io(BTRFS_I(inode)->root->fs_info,
					  bio, 1);
926
		if (ret)
927 928 929
			goto out_w_error;
		ret = btrfs_map_bio(BTRFS_I(inode)->root, rw, bio,
				    mirror_num, 0);
930 931 932
	} else if (!async) {
		ret = btree_csum_one_bio(bio);
		if (ret)
933 934 935 936 937 938 939 940 941 942 943 944 945
			goto out_w_error;
		ret = btrfs_map_bio(BTRFS_I(inode)->root, rw, bio,
				    mirror_num, 0);
	} else {
		/*
		 * kthread helpers are used to submit writes so that
		 * checksumming can happen in parallel across all CPUs
		 */
		ret = btrfs_wq_submit_bio(BTRFS_I(inode)->root->fs_info,
					  inode, rw, bio, mirror_num, 0,
					  bio_offset,
					  __btree_submit_bio_start,
					  __btree_submit_bio_done);
946
	}
947

948 949 950 951 952
	if (ret) {
out_w_error:
		bio_endio(bio, ret);
	}
	return ret;
953 954
}

J
Jan Beulich 已提交
955
#ifdef CONFIG_MIGRATION
956
static int btree_migratepage(struct address_space *mapping,
957 958
			struct page *newpage, struct page *page,
			enum migrate_mode mode)
959 960 961 962 963 964 965 966 967 968 969 970 971 972
{
	/*
	 * 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;
973
	return migrate_page(mapping, newpage, page, mode);
974
}
J
Jan Beulich 已提交
975
#endif
976

977 978 979 980

static int btree_writepages(struct address_space *mapping,
			    struct writeback_control *wbc)
{
981
	struct extent_io_tree *tree;
982 983 984
	struct btrfs_fs_info *fs_info;
	int ret;

985
	tree = &BTRFS_I(mapping->host)->io_tree;
986
	if (wbc->sync_mode == WB_SYNC_NONE) {
987 988 989 990

		if (wbc->for_kupdate)
			return 0;

991
		fs_info = BTRFS_I(mapping->host)->root->fs_info;
992
		/* this is a bit racy, but that's ok */
993 994 995
		ret = percpu_counter_compare(&fs_info->dirty_metadata_bytes,
					     BTRFS_DIRTY_METADATA_THRESH);
		if (ret < 0)
996 997
			return 0;
	}
998
	return btree_write_cache_pages(mapping, wbc);
999 1000
}

1001
static int btree_readpage(struct file *file, struct page *page)
1002
{
1003 1004
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
1005
	return extent_read_full_page(tree, page, btree_get_extent, 0);
1006
}
C
Chris Mason 已提交
1007

1008
static int btree_releasepage(struct page *page, gfp_t gfp_flags)
1009
{
1010
	if (PageWriteback(page) || PageDirty(page))
C
Chris Mason 已提交
1011
		return 0;
1012

1013
	return try_release_extent_buffer(page);
1014 1015
}

1016
static void btree_invalidatepage(struct page *page, unsigned long offset)
1017
{
1018 1019
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
1020 1021
	extent_invalidatepage(tree, page, offset);
	btree_releasepage(page, GFP_NOFS);
1022
	if (PagePrivate(page)) {
C
Chris Mason 已提交
1023 1024
		printk(KERN_WARNING "btrfs warning page private not zero "
		       "on page %llu\n", (unsigned long long)page_offset(page));
1025 1026 1027 1028
		ClearPagePrivate(page);
		set_page_private(page, 0);
		page_cache_release(page);
	}
1029 1030
}

1031 1032
static int btree_set_page_dirty(struct page *page)
{
1033
#ifdef DEBUG
1034 1035 1036 1037 1038 1039 1040 1041
	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);
1042
#endif
1043 1044 1045
	return __set_page_dirty_nobuffers(page);
}

1046
static const struct address_space_operations btree_aops = {
1047
	.readpage	= btree_readpage,
1048
	.writepages	= btree_writepages,
1049 1050
	.releasepage	= btree_releasepage,
	.invalidatepage = btree_invalidatepage,
1051
#ifdef CONFIG_MIGRATION
1052
	.migratepage	= btree_migratepage,
1053
#endif
1054
	.set_page_dirty = btree_set_page_dirty,
1055 1056
};

1057 1058
int readahead_tree_block(struct btrfs_root *root, u64 bytenr, u32 blocksize,
			 u64 parent_transid)
C
Chris Mason 已提交
1059
{
1060 1061
	struct extent_buffer *buf = NULL;
	struct inode *btree_inode = root->fs_info->btree_inode;
1062
	int ret = 0;
C
Chris Mason 已提交
1063

1064
	buf = btrfs_find_create_tree_block(root, bytenr, blocksize);
1065
	if (!buf)
C
Chris Mason 已提交
1066
		return 0;
1067
	read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree,
1068
				 buf, 0, WAIT_NONE, btree_get_extent, 0);
1069
	free_extent_buffer(buf);
1070
	return ret;
C
Chris Mason 已提交
1071 1072
}

1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096
int reada_tree_block_flagged(struct btrfs_root *root, u64 bytenr, u32 blocksize,
			 int mirror_num, struct extent_buffer **eb)
{
	struct extent_buffer *buf = NULL;
	struct inode *btree_inode = root->fs_info->btree_inode;
	struct extent_io_tree *io_tree = &BTRFS_I(btree_inode)->io_tree;
	int ret;

	buf = btrfs_find_create_tree_block(root, bytenr, blocksize);
	if (!buf)
		return 0;

	set_bit(EXTENT_BUFFER_READAHEAD, &buf->bflags);

	ret = read_extent_buffer_pages(io_tree, buf, 0, WAIT_PAGE_LOCK,
				       btree_get_extent, mirror_num);
	if (ret) {
		free_extent_buffer(buf);
		return ret;
	}

	if (test_bit(EXTENT_BUFFER_CORRUPT, &buf->bflags)) {
		free_extent_buffer(buf);
		return -EIO;
1097
	} else if (extent_buffer_uptodate(buf)) {
1098 1099 1100 1101 1102 1103 1104
		*eb = buf;
	} else {
		free_extent_buffer(buf);
	}
	return 0;
}

1105 1106 1107 1108 1109 1110
struct extent_buffer *btrfs_find_tree_block(struct btrfs_root *root,
					    u64 bytenr, u32 blocksize)
{
	struct inode *btree_inode = root->fs_info->btree_inode;
	struct extent_buffer *eb;
	eb = find_extent_buffer(&BTRFS_I(btree_inode)->io_tree,
1111
				bytenr, blocksize);
1112 1113 1114 1115 1116 1117 1118 1119 1120 1121
	return eb;
}

struct extent_buffer *btrfs_find_create_tree_block(struct btrfs_root *root,
						 u64 bytenr, u32 blocksize)
{
	struct inode *btree_inode = root->fs_info->btree_inode;
	struct extent_buffer *eb;

	eb = alloc_extent_buffer(&BTRFS_I(btree_inode)->io_tree,
1122
				 bytenr, blocksize);
1123 1124 1125 1126
	return eb;
}


1127 1128
int btrfs_write_tree_block(struct extent_buffer *buf)
{
1129
	return filemap_fdatawrite_range(buf->pages[0]->mapping, buf->start,
1130
					buf->start + buf->len - 1);
1131 1132 1133 1134
}

int btrfs_wait_tree_block_writeback(struct extent_buffer *buf)
{
1135
	return filemap_fdatawait_range(buf->pages[0]->mapping,
1136
				       buf->start, buf->start + buf->len - 1);
1137 1138
}

1139
struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr,
1140
				      u32 blocksize, u64 parent_transid)
1141 1142 1143 1144 1145 1146 1147 1148
{
	struct extent_buffer *buf = NULL;
	int ret;

	buf = btrfs_find_create_tree_block(root, bytenr, blocksize);
	if (!buf)
		return NULL;

1149
	ret = btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
1150
	return buf;
1151

1152 1153
}

1154 1155
void clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root,
		      struct extent_buffer *buf)
1156
{
1157 1158
	struct btrfs_fs_info *fs_info = root->fs_info;

1159
	if (btrfs_header_generation(buf) ==
1160
	    fs_info->running_transaction->transid) {
1161
		btrfs_assert_tree_locked(buf);
1162

1163
		if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &buf->bflags)) {
1164 1165 1166
			__percpu_counter_add(&fs_info->dirty_metadata_bytes,
					     -buf->len,
					     fs_info->dirty_metadata_batch);
1167 1168 1169 1170
			/* ugh, clear_extent_buffer_dirty needs to lock the page */
			btrfs_set_lock_blocking(buf);
			clear_extent_buffer_dirty(buf);
		}
1171
	}
1172 1173
}

1174 1175 1176 1177
static void __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
			 u32 stripesize, struct btrfs_root *root,
			 struct btrfs_fs_info *fs_info,
			 u64 objectid)
1178
{
C
Chris Mason 已提交
1179
	root->node = NULL;
1180
	root->commit_root = NULL;
1181 1182 1183
	root->sectorsize = sectorsize;
	root->nodesize = nodesize;
	root->leafsize = leafsize;
1184
	root->stripesize = stripesize;
C
Chris Mason 已提交
1185
	root->ref_cows = 0;
1186
	root->track_dirty = 0;
1187
	root->in_radix = 0;
1188 1189
	root->orphan_item_inserted = 0;
	root->orphan_cleanup_state = 0;
1190

1191 1192
	root->objectid = objectid;
	root->last_trans = 0;
1193
	root->highest_objectid = 0;
1194
	root->name = NULL;
1195
	root->inode_tree = RB_ROOT;
1196
	INIT_RADIX_TREE(&root->delayed_nodes_tree, GFP_ATOMIC);
1197
	root->block_rsv = NULL;
1198
	root->orphan_block_rsv = NULL;
1199 1200

	INIT_LIST_HEAD(&root->dirty_list);
1201
	INIT_LIST_HEAD(&root->root_list);
1202 1203
	INIT_LIST_HEAD(&root->logged_list[0]);
	INIT_LIST_HEAD(&root->logged_list[1]);
1204
	spin_lock_init(&root->orphan_lock);
1205
	spin_lock_init(&root->inode_lock);
1206
	spin_lock_init(&root->accounting_lock);
1207 1208
	spin_lock_init(&root->log_extents_lock[0]);
	spin_lock_init(&root->log_extents_lock[1]);
1209
	mutex_init(&root->objectid_mutex);
1210
	mutex_init(&root->log_mutex);
Y
Yan Zheng 已提交
1211 1212 1213 1214 1215 1216
	init_waitqueue_head(&root->log_writer_wait);
	init_waitqueue_head(&root->log_commit_wait[0]);
	init_waitqueue_head(&root->log_commit_wait[1]);
	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
1217
	atomic_set(&root->log_batch, 0);
1218
	atomic_set(&root->orphan_inodes, 0);
Y
Yan Zheng 已提交
1219
	root->log_transid = 0;
1220
	root->last_log_commit = 0;
1221
	extent_io_tree_init(&root->dirty_log_pages,
1222
			     fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
1223

1224 1225
	memset(&root->root_key, 0, sizeof(root->root_key));
	memset(&root->root_item, 0, sizeof(root->root_item));
1226
	memset(&root->defrag_progress, 0, sizeof(root->defrag_progress));
1227
	memset(&root->root_kobj, 0, sizeof(root->root_kobj));
1228
	root->defrag_trans_start = fs_info->generation;
1229
	init_completion(&root->kobj_unregister);
1230
	root->defrag_running = 0;
1231
	root->root_key.objectid = objectid;
1232
	root->anon_dev = 0;
1233

1234
	spin_lock_init(&root->root_item_lock);
1235 1236
}

1237 1238 1239 1240
static int __must_check find_and_setup_root(struct btrfs_root *tree_root,
					    struct btrfs_fs_info *fs_info,
					    u64 objectid,
					    struct btrfs_root *root)
1241 1242
{
	int ret;
1243
	u32 blocksize;
1244
	u64 generation;
1245

1246
	__setup_root(tree_root->nodesize, tree_root->leafsize,
1247 1248
		     tree_root->sectorsize, tree_root->stripesize,
		     root, fs_info, objectid);
1249 1250
	ret = btrfs_find_last_root(tree_root, objectid,
				   &root->root_item, &root->root_key);
1251 1252
	if (ret > 0)
		return -ENOENT;
1253 1254
	else if (ret < 0)
		return ret;
1255

1256
	generation = btrfs_root_generation(&root->root_item);
1257
	blocksize = btrfs_level_size(root, btrfs_root_level(&root->root_item));
C
Chris Mason 已提交
1258
	root->commit_root = NULL;
1259
	root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
1260
				     blocksize, generation);
1261
	if (!root->node || !btrfs_buffer_uptodate(root->node, generation, 0)) {
1262
		free_extent_buffer(root->node);
C
Chris Mason 已提交
1263
		root->node = NULL;
1264 1265
		return -EIO;
	}
1266
	root->commit_root = btrfs_root_node(root);
1267 1268 1269
	return 0;
}

1270
static struct btrfs_root *btrfs_alloc_root(struct btrfs_fs_info *fs_info)
A
Al Viro 已提交
1271 1272 1273 1274 1275 1276 1277
{
	struct btrfs_root *root = kzalloc(sizeof(*root), GFP_NOFS);
	if (root)
		root->fs_info = fs_info;
	return root;
}

1278 1279 1280 1281 1282 1283 1284 1285 1286 1287
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;
	u64 bytenr;
1288
	uuid_le uuid;
1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304

	root = btrfs_alloc_root(fs_info);
	if (!root)
		return ERR_PTR(-ENOMEM);

	__setup_root(tree_root->nodesize, tree_root->leafsize,
		     tree_root->sectorsize, tree_root->stripesize,
		     root, fs_info, objectid);
	root->root_key.objectid = objectid;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = 0;

	leaf = btrfs_alloc_free_block(trans, root, root->leafsize,
				      0, objectid, NULL, 0, 0, 0);
	if (IS_ERR(leaf)) {
		ret = PTR_ERR(leaf);
1305
		leaf = NULL;
1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337
		goto fail;
	}

	bytenr = leaf->start;
	memset_extent_buffer(leaf, 0, 0, sizeof(struct btrfs_header));
	btrfs_set_header_bytenr(leaf, leaf->start);
	btrfs_set_header_generation(leaf, trans->transid);
	btrfs_set_header_backref_rev(leaf, BTRFS_MIXED_BACKREF_REV);
	btrfs_set_header_owner(leaf, objectid);
	root->node = leaf;

	write_extent_buffer(leaf, fs_info->fsid,
			    (unsigned long)btrfs_header_fsid(leaf),
			    BTRFS_FSID_SIZE);
	write_extent_buffer(leaf, fs_info->chunk_tree_uuid,
			    (unsigned long)btrfs_header_chunk_tree_uuid(leaf),
			    BTRFS_UUID_SIZE);
	btrfs_mark_buffer_dirty(leaf);

	root->commit_root = btrfs_root_node(root);
	root->track_dirty = 1;


	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);
1338 1339
	uuid_le_gen(&uuid);
	memcpy(root->root_item.uuid, uuid.b, BTRFS_UUID_SIZE);
1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350
	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);

1351 1352
	return root;

1353
fail:
1354 1355 1356 1357 1358
	if (leaf) {
		btrfs_tree_unlock(leaf);
		free_extent_buffer(leaf);
	}
	kfree(root);
1359

1360
	return ERR_PTR(ret);
1361 1362
}

Y
Yan Zheng 已提交
1363 1364
static struct btrfs_root *alloc_log_tree(struct btrfs_trans_handle *trans,
					 struct btrfs_fs_info *fs_info)
1365 1366 1367
{
	struct btrfs_root *root;
	struct btrfs_root *tree_root = fs_info->tree_root;
Y
Yan Zheng 已提交
1368
	struct extent_buffer *leaf;
1369

A
Al Viro 已提交
1370
	root = btrfs_alloc_root(fs_info);
1371
	if (!root)
Y
Yan Zheng 已提交
1372
		return ERR_PTR(-ENOMEM);
1373 1374 1375 1376 1377 1378 1379 1380

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

	root->root_key.objectid = BTRFS_TREE_LOG_OBJECTID;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = BTRFS_TREE_LOG_OBJECTID;
Y
Yan Zheng 已提交
1381 1382 1383 1384 1385 1386
	/*
	 * 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).
	 */
1387 1388
	root->ref_cows = 0;

1389
	leaf = btrfs_alloc_free_block(trans, root, root->leafsize, 0,
A
Arne Jansen 已提交
1390
				      BTRFS_TREE_LOG_OBJECTID, NULL,
1391
				      0, 0, 0);
Y
Yan Zheng 已提交
1392 1393 1394 1395
	if (IS_ERR(leaf)) {
		kfree(root);
		return ERR_CAST(leaf);
	}
1396

1397 1398 1399 1400 1401
	memset_extent_buffer(leaf, 0, 0, sizeof(struct btrfs_header));
	btrfs_set_header_bytenr(leaf, leaf->start);
	btrfs_set_header_generation(leaf, trans->transid);
	btrfs_set_header_backref_rev(leaf, BTRFS_MIXED_BACKREF_REV);
	btrfs_set_header_owner(leaf, BTRFS_TREE_LOG_OBJECTID);
Y
Yan Zheng 已提交
1402
	root->node = leaf;
1403 1404 1405 1406 1407 1408

	write_extent_buffer(root->node, root->fs_info->fsid,
			    (unsigned long)btrfs_header_fsid(root->node),
			    BTRFS_FSID_SIZE);
	btrfs_mark_buffer_dirty(root->node);
	btrfs_tree_unlock(root->node);
Y
Yan Zheng 已提交
1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444
	return root;
}

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

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

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

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

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

	inode_item = &log_root->root_item.inode;
	inode_item->generation = cpu_to_le64(1);
	inode_item->size = cpu_to_le64(3);
	inode_item->nlink = cpu_to_le32(1);
	inode_item->nbytes = cpu_to_le64(root->leafsize);
	inode_item->mode = cpu_to_le32(S_IFDIR | 0755);

1445
	btrfs_set_root_node(&log_root->root_item, log_root->node);
Y
Yan Zheng 已提交
1446 1447 1448 1449

	WARN_ON(root->log_root);
	root->log_root = log_root;
	root->log_transid = 0;
1450
	root->last_log_commit = 0;
1451 1452 1453 1454 1455 1456 1457 1458
	return 0;
}

struct btrfs_root *btrfs_read_fs_root_no_radix(struct btrfs_root *tree_root,
					       struct btrfs_key *location)
{
	struct btrfs_root *root;
	struct btrfs_fs_info *fs_info = tree_root->fs_info;
1459
	struct btrfs_path *path;
1460
	struct extent_buffer *l;
1461
	u64 generation;
1462
	u32 blocksize;
1463
	int ret = 0;
1464
	int slot;
1465

A
Al Viro 已提交
1466
	root = btrfs_alloc_root(fs_info);
C
Chris Mason 已提交
1467
	if (!root)
1468 1469
		return ERR_PTR(-ENOMEM);
	if (location->offset == (u64)-1) {
1470
		ret = find_and_setup_root(tree_root, fs_info,
1471 1472 1473 1474 1475
					  location->objectid, root);
		if (ret) {
			kfree(root);
			return ERR_PTR(ret);
		}
1476
		goto out;
1477 1478
	}

1479
	__setup_root(tree_root->nodesize, tree_root->leafsize,
1480 1481
		     tree_root->sectorsize, tree_root->stripesize,
		     root, fs_info, location->objectid);
1482 1483

	path = btrfs_alloc_path();
T
Tsutomu Itoh 已提交
1484 1485 1486 1487
	if (!path) {
		kfree(root);
		return ERR_PTR(-ENOMEM);
	}
1488
	ret = btrfs_search_slot(NULL, tree_root, location, path, 0, 0);
1489 1490
	if (ret == 0) {
		l = path->nodes[0];
1491
		slot = path->slots[0];
1492
		btrfs_read_root_item(l, slot, &root->root_item);
1493
		memcpy(&root->root_key, location, sizeof(*location));
1494 1495 1496
	}
	btrfs_free_path(path);
	if (ret) {
1497
		kfree(root);
1498 1499
		if (ret > 0)
			ret = -ENOENT;
1500 1501
		return ERR_PTR(ret);
	}
1502

1503
	generation = btrfs_root_generation(&root->root_item);
1504 1505
	blocksize = btrfs_level_size(root, btrfs_root_level(&root->root_item));
	root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
1506
				     blocksize, generation);
1507 1508 1509 1510 1511 1512 1513 1514
	if (!root->node || !extent_buffer_uptodate(root->node)) {
		ret = (!root->node) ? -ENOMEM : -EIO;

		free_extent_buffer(root->node);
		kfree(root);
		return ERR_PTR(ret);
	}

1515
	root->commit_root = btrfs_root_node(root);
1516
out:
1517
	if (location->objectid != BTRFS_TREE_LOG_OBJECTID) {
1518
		root->ref_cows = 1;
1519 1520
		btrfs_check_and_init_root_item(&root->root_item);
	}
1521

1522 1523 1524
	return root;
}

1525 1526
struct btrfs_root *btrfs_read_fs_root_no_name(struct btrfs_fs_info *fs_info,
					      struct btrfs_key *location)
1527 1528 1529 1530
{
	struct btrfs_root *root;
	int ret;

1531 1532 1533 1534
	if (location->objectid == BTRFS_ROOT_TREE_OBJECTID)
		return fs_info->tree_root;
	if (location->objectid == BTRFS_EXTENT_TREE_OBJECTID)
		return fs_info->extent_root;
1535 1536 1537 1538
	if (location->objectid == BTRFS_CHUNK_TREE_OBJECTID)
		return fs_info->chunk_root;
	if (location->objectid == BTRFS_DEV_TREE_OBJECTID)
		return fs_info->dev_root;
1539 1540
	if (location->objectid == BTRFS_CSUM_TREE_OBJECTID)
		return fs_info->csum_root;
1541 1542 1543
	if (location->objectid == BTRFS_QUOTA_TREE_OBJECTID)
		return fs_info->quota_root ? fs_info->quota_root :
					     ERR_PTR(-ENOENT);
1544 1545
again:
	spin_lock(&fs_info->fs_roots_radix_lock);
1546 1547
	root = radix_tree_lookup(&fs_info->fs_roots_radix,
				 (unsigned long)location->objectid);
1548
	spin_unlock(&fs_info->fs_roots_radix_lock);
1549 1550 1551
	if (root)
		return root;

1552
	root = btrfs_read_fs_root_no_radix(fs_info->tree_root, location);
1553 1554
	if (IS_ERR(root))
		return root;
1555

1556 1557 1558
	root->free_ino_ctl = kzalloc(sizeof(*root->free_ino_ctl), GFP_NOFS);
	root->free_ino_pinned = kzalloc(sizeof(*root->free_ino_pinned),
					GFP_NOFS);
1559 1560
	if (!root->free_ino_pinned || !root->free_ino_ctl) {
		ret = -ENOMEM;
1561
		goto fail;
1562
	}
1563 1564 1565 1566 1567 1568

	btrfs_init_free_ino_ctl(root);
	mutex_init(&root->fs_commit_mutex);
	spin_lock_init(&root->cache_lock);
	init_waitqueue_head(&root->cache_wait);

1569
	ret = get_anon_bdev(&root->anon_dev);
1570 1571
	if (ret)
		goto fail;
1572

1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583
	if (btrfs_root_refs(&root->root_item) == 0) {
		ret = -ENOENT;
		goto fail;
	}

	ret = btrfs_find_orphan_item(fs_info->tree_root, location->objectid);
	if (ret < 0)
		goto fail;
	if (ret == 0)
		root->orphan_item_inserted = 1;

1584 1585 1586 1587 1588
	ret = radix_tree_preload(GFP_NOFS & ~__GFP_HIGHMEM);
	if (ret)
		goto fail;

	spin_lock(&fs_info->fs_roots_radix_lock);
C
Chris Mason 已提交
1589 1590
	ret = radix_tree_insert(&fs_info->fs_roots_radix,
				(unsigned long)root->root_key.objectid,
1591
				root);
1592
	if (ret == 0)
1593
		root->in_radix = 1;
1594

1595 1596
	spin_unlock(&fs_info->fs_roots_radix_lock);
	radix_tree_preload_end();
1597
	if (ret) {
1598 1599 1600 1601 1602
		if (ret == -EEXIST) {
			free_fs_root(root);
			goto again;
		}
		goto fail;
1603
	}
1604 1605 1606 1607

	ret = btrfs_find_dead_roots(fs_info->tree_root,
				    root->root_key.objectid);
	WARN_ON(ret);
1608
	return root;
1609 1610 1611
fail:
	free_fs_root(root);
	return ERR_PTR(ret);
1612 1613
}

C
Chris Mason 已提交
1614 1615 1616 1617 1618 1619
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
1620

1621 1622
	rcu_read_lock();
	list_for_each_entry_rcu(device, &info->fs_devices->devices, dev_list) {
1623 1624
		if (!device->bdev)
			continue;
C
Chris Mason 已提交
1625 1626 1627 1628 1629 1630
		bdi = blk_get_backing_dev_info(device->bdev);
		if (bdi && bdi_congested(bdi, bdi_bits)) {
			ret = 1;
			break;
		}
	}
1631
	rcu_read_unlock();
C
Chris Mason 已提交
1632 1633 1634
	return ret;
}

1635 1636 1637 1638
/*
 * If this fails, caller must call bdi_destroy() to get rid of the
 * bdi again.
 */
C
Chris Mason 已提交
1639 1640
static int setup_bdi(struct btrfs_fs_info *info, struct backing_dev_info *bdi)
{
1641 1642 1643
	int err;

	bdi->capabilities = BDI_CAP_MAP_COPY;
1644
	err = bdi_setup_and_register(bdi, "btrfs", BDI_CAP_MAP_COPY);
1645 1646 1647
	if (err)
		return err;

1648
	bdi->ra_pages	= default_backing_dev_info.ra_pages;
C
Chris Mason 已提交
1649 1650 1651 1652 1653
	bdi->congested_fn	= btrfs_congested_fn;
	bdi->congested_data	= info;
	return 0;
}

1654 1655 1656 1657 1658
/*
 * 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)
1659 1660
{
	struct bio *bio;
1661 1662
	struct end_io_wq *end_io_wq;
	struct btrfs_fs_info *fs_info;
1663 1664
	int error;

1665 1666 1667
	end_io_wq = container_of(work, struct end_io_wq, work);
	bio = end_io_wq->bio;
	fs_info = end_io_wq->info;
1668

1669 1670 1671 1672 1673
	error = end_io_wq->error;
	bio->bi_private = end_io_wq->private;
	bio->bi_end_io = end_io_wq->end_io;
	kfree(end_io_wq);
	bio_endio(bio, error);
1674 1675
}

1676 1677 1678 1679 1680
static int cleaner_kthread(void *arg)
{
	struct btrfs_root *root = arg;

	do {
D
David Sterba 已提交
1681 1682
		int again = 0;

1683
		if (!(root->fs_info->sb->s_flags & MS_RDONLY) &&
D
David Sterba 已提交
1684 1685 1686 1687 1688 1689
		    down_read_trylock(&root->fs_info->sb->s_umount)) {
			if (mutex_trylock(&root->fs_info->cleaner_mutex)) {
				btrfs_run_delayed_iputs(root);
				again = btrfs_clean_one_deleted_snapshot(root);
				mutex_unlock(&root->fs_info->cleaner_mutex);
			}
C
Chris Mason 已提交
1690
			btrfs_run_defrag_inodes(root->fs_info);
D
David Sterba 已提交
1691
			up_read(&root->fs_info->sb->s_umount);
1692
		}
1693

D
David Sterba 已提交
1694
		if (!try_to_freeze() && !again) {
1695
			set_current_state(TASK_INTERRUPTIBLE);
1696 1697
			if (!kthread_should_stop())
				schedule();
1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708
			__set_current_state(TASK_RUNNING);
		}
	} while (!kthread_should_stop());
	return 0;
}

static int transaction_kthread(void *arg)
{
	struct btrfs_root *root = arg;
	struct btrfs_trans_handle *trans;
	struct btrfs_transaction *cur;
1709
	u64 transid;
1710 1711
	unsigned long now;
	unsigned long delay;
1712
	bool cannot_commit;
1713 1714

	do {
1715
		cannot_commit = false;
1716 1717 1718
		delay = HZ * 30;
		mutex_lock(&root->fs_info->transaction_kthread_mutex);

J
Josef Bacik 已提交
1719
		spin_lock(&root->fs_info->trans_lock);
1720 1721
		cur = root->fs_info->running_transaction;
		if (!cur) {
J
Josef Bacik 已提交
1722
			spin_unlock(&root->fs_info->trans_lock);
1723 1724
			goto sleep;
		}
Y
Yan Zheng 已提交
1725

1726
		now = get_seconds();
1727 1728
		if (!cur->blocked &&
		    (now < cur->start_time || now - cur->start_time < 30)) {
J
Josef Bacik 已提交
1729
			spin_unlock(&root->fs_info->trans_lock);
1730 1731 1732
			delay = HZ * 5;
			goto sleep;
		}
1733
		transid = cur->transid;
J
Josef Bacik 已提交
1734
		spin_unlock(&root->fs_info->trans_lock);
1735

1736
		/* If the file system is aborted, this will always fail. */
1737
		trans = btrfs_attach_transaction(root);
1738
		if (IS_ERR(trans)) {
1739 1740
			if (PTR_ERR(trans) != -ENOENT)
				cannot_commit = true;
1741
			goto sleep;
1742
		}
1743
		if (transid == trans->transid) {
1744
			btrfs_commit_transaction(trans, root);
1745 1746 1747
		} else {
			btrfs_end_transaction(trans, root);
		}
1748 1749 1750 1751
sleep:
		wake_up_process(root->fs_info->cleaner_kthread);
		mutex_unlock(&root->fs_info->transaction_kthread_mutex);

1752
		if (!try_to_freeze()) {
1753
			set_current_state(TASK_INTERRUPTIBLE);
1754
			if (!kthread_should_stop() &&
1755 1756
			    (!btrfs_transaction_blocked(root->fs_info) ||
			     cannot_commit))
1757
				schedule_timeout(delay);
1758 1759 1760 1761 1762 1763
			__set_current_state(TASK_RUNNING);
		}
	} while (!kthread_should_stop());
	return 0;
}

C
Chris Mason 已提交
1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869
/*
 * 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));

1870 1871 1872 1873 1874 1875 1876 1877
	/*
	 * 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
1878
			       btrfs_header_generation(info->fs_root->node));
1879
		btrfs_set_backup_fs_root_level(root_backup,
C
Chris Mason 已提交
1880
			       btrfs_header_level(info->fs_root->node));
1881
	}
C
Chris Mason 已提交
1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962

	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
1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981
/* helper to cleanup workers */
static void btrfs_stop_all_workers(struct btrfs_fs_info *fs_info)
{
	btrfs_stop_workers(&fs_info->generic_worker);
	btrfs_stop_workers(&fs_info->fixup_workers);
	btrfs_stop_workers(&fs_info->delalloc_workers);
	btrfs_stop_workers(&fs_info->workers);
	btrfs_stop_workers(&fs_info->endio_workers);
	btrfs_stop_workers(&fs_info->endio_meta_workers);
	btrfs_stop_workers(&fs_info->endio_raid56_workers);
	btrfs_stop_workers(&fs_info->rmw_workers);
	btrfs_stop_workers(&fs_info->endio_meta_write_workers);
	btrfs_stop_workers(&fs_info->endio_write_workers);
	btrfs_stop_workers(&fs_info->endio_freespace_worker);
	btrfs_stop_workers(&fs_info->submit_workers);
	btrfs_stop_workers(&fs_info->delayed_workers);
	btrfs_stop_workers(&fs_info->caching_workers);
	btrfs_stop_workers(&fs_info->readahead_workers);
	btrfs_stop_workers(&fs_info->flush_workers);
J
Jan Schmidt 已提交
1982
	btrfs_stop_workers(&fs_info->qgroup_rescan_workers);
L
Liu Bo 已提交
1983 1984
}

C
Chris Mason 已提交
1985 1986 1987 1988 1989 1990 1991
/* helper to cleanup tree roots */
static void free_root_pointers(struct btrfs_fs_info *info, int chunk_root)
{
	free_extent_buffer(info->tree_root->node);
	free_extent_buffer(info->tree_root->commit_root);
	info->tree_root->node = NULL;
	info->tree_root->commit_root = NULL;
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

	if (info->dev_root) {
		free_extent_buffer(info->dev_root->node);
		free_extent_buffer(info->dev_root->commit_root);
		info->dev_root->node = NULL;
		info->dev_root->commit_root = NULL;
	}
	if (info->extent_root) {
		free_extent_buffer(info->extent_root->node);
		free_extent_buffer(info->extent_root->commit_root);
		info->extent_root->node = NULL;
		info->extent_root->commit_root = NULL;
	}
	if (info->csum_root) {
		free_extent_buffer(info->csum_root->node);
		free_extent_buffer(info->csum_root->commit_root);
		info->csum_root->node = NULL;
		info->csum_root->commit_root = NULL;
	}
2011
	if (info->quota_root) {
2012 2013
		free_extent_buffer(info->quota_root->node);
		free_extent_buffer(info->quota_root->commit_root);
2014 2015 2016
		info->quota_root->node = NULL;
		info->quota_root->commit_root = NULL;
	}
C
Chris Mason 已提交
2017 2018 2019 2020 2021 2022 2023 2024
	if (chunk_root) {
		free_extent_buffer(info->chunk_root->node);
		free_extent_buffer(info->chunk_root->commit_root);
		info->chunk_root->node = NULL;
		info->chunk_root->commit_root = NULL;
	}
}

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
static void del_fs_roots(struct btrfs_fs_info *fs_info)
{
	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);

		if (gang[0]->in_radix) {
			btrfs_free_fs_root(fs_info, gang[0]);
		} else {
			free_extent_buffer(gang[0]->node);
			free_extent_buffer(gang[0]->commit_root);
			kfree(gang[0]);
		}
	}

	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++)
			btrfs_free_fs_root(fs_info, gang[i]);
	}
}
C
Chris Mason 已提交
2055

A
Al Viro 已提交
2056 2057 2058
int open_ctree(struct super_block *sb,
	       struct btrfs_fs_devices *fs_devices,
	       char *options)
2059
{
2060 2061 2062 2063
	u32 sectorsize;
	u32 nodesize;
	u32 leafsize;
	u32 blocksize;
2064
	u32 stripesize;
2065
	u64 generation;
2066
	u64 features;
2067
	struct btrfs_key location;
2068
	struct buffer_head *bh;
2069
	struct btrfs_super_block *disk_super;
2070
	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
2071
	struct btrfs_root *tree_root;
2072 2073 2074 2075
	struct btrfs_root *extent_root;
	struct btrfs_root *csum_root;
	struct btrfs_root *chunk_root;
	struct btrfs_root *dev_root;
2076
	struct btrfs_root *quota_root;
2077
	struct btrfs_root *log_tree_root;
2078
	int ret;
2079
	int err = -EINVAL;
C
Chris Mason 已提交
2080 2081
	int num_backups_tried = 0;
	int backup_index = 0;
2082

2083
	tree_root = fs_info->tree_root = btrfs_alloc_root(fs_info);
A
Al Viro 已提交
2084 2085 2086 2087
	extent_root = fs_info->extent_root = btrfs_alloc_root(fs_info);
	csum_root = fs_info->csum_root = btrfs_alloc_root(fs_info);
	chunk_root = fs_info->chunk_root = btrfs_alloc_root(fs_info);
	dev_root = fs_info->dev_root = btrfs_alloc_root(fs_info);
2088
	quota_root = fs_info->quota_root = btrfs_alloc_root(fs_info);
2089

2090
	if (!tree_root || !extent_root || !csum_root ||
2091
	    !chunk_root || !dev_root || !quota_root) {
C
Chris Mason 已提交
2092 2093 2094
		err = -ENOMEM;
		goto fail;
	}
2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107

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

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

2108 2109 2110 2111 2112 2113 2114 2115
	ret = percpu_counter_init(&fs_info->dirty_metadata_bytes, 0);
	if (ret) {
		err = ret;
		goto fail_bdi;
	}
	fs_info->dirty_metadata_batch = PAGE_CACHE_SIZE *
					(1 + ilog2(nr_cpu_ids));

2116 2117 2118 2119 2120 2121
	ret = percpu_counter_init(&fs_info->delalloc_bytes, 0);
	if (ret) {
		err = ret;
		goto fail_dirty_metadata_bytes;
	}

2122 2123 2124
	fs_info->btree_inode = new_inode(sb);
	if (!fs_info->btree_inode) {
		err = -ENOMEM;
2125
		goto fail_delalloc_bytes;
2126 2127
	}

2128
	mapping_set_gfp_mask(fs_info->btree_inode->i_mapping, GFP_NOFS);
2129

2130
	INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC);
C
Chris Mason 已提交
2131
	INIT_LIST_HEAD(&fs_info->trans_list);
2132
	INIT_LIST_HEAD(&fs_info->dead_roots);
Y
Yan, Zheng 已提交
2133
	INIT_LIST_HEAD(&fs_info->delayed_iputs);
2134
	INIT_LIST_HEAD(&fs_info->delalloc_inodes);
2135
	INIT_LIST_HEAD(&fs_info->caching_block_groups);
2136
	spin_lock_init(&fs_info->delalloc_lock);
J
Josef Bacik 已提交
2137
	spin_lock_init(&fs_info->trans_lock);
2138
	spin_lock_init(&fs_info->fs_roots_radix_lock);
Y
Yan, Zheng 已提交
2139
	spin_lock_init(&fs_info->delayed_iput_lock);
C
Chris Mason 已提交
2140
	spin_lock_init(&fs_info->defrag_inodes_lock);
2141
	spin_lock_init(&fs_info->free_chunk_lock);
J
Jan Schmidt 已提交
2142
	spin_lock_init(&fs_info->tree_mod_seq_lock);
2143
	spin_lock_init(&fs_info->super_lock);
J
Jan Schmidt 已提交
2144
	rwlock_init(&fs_info->tree_mod_log_lock);
C
Chris Mason 已提交
2145
	mutex_init(&fs_info->reloc_mutex);
2146
	seqlock_init(&fs_info->profiles_lock);
2147

2148
	init_completion(&fs_info->kobj_unregister);
2149
	INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots);
2150
	INIT_LIST_HEAD(&fs_info->space_info);
J
Jan Schmidt 已提交
2151
	INIT_LIST_HEAD(&fs_info->tree_mod_seq_list);
2152
	btrfs_mapping_init(&fs_info->mapping_tree);
2153 2154 2155 2156 2157 2158 2159 2160 2161
	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);
2162
	atomic_set(&fs_info->nr_async_submits, 0);
2163
	atomic_set(&fs_info->async_delalloc_pages, 0);
2164
	atomic_set(&fs_info->async_submit_draining, 0);
2165
	atomic_set(&fs_info->nr_async_bios, 0);
C
Chris Mason 已提交
2166
	atomic_set(&fs_info->defrag_running, 0);
2167
	atomic64_set(&fs_info->tree_mod_seq, 0);
C
Chris Mason 已提交
2168
	fs_info->sb = sb;
2169
	fs_info->max_inline = 8192 * 1024;
J
Josef Bacik 已提交
2170
	fs_info->metadata_ratio = 0;
C
Chris Mason 已提交
2171
	fs_info->defrag_inodes = RB_ROOT;
J
Josef Bacik 已提交
2172
	fs_info->trans_no_join = 0;
2173
	fs_info->free_chunk_space = 0;
J
Jan Schmidt 已提交
2174
	fs_info->tree_mod_log = RB_ROOT;
C
Chris Mason 已提交
2175

2176 2177 2178
	/* readahead state */
	INIT_RADIX_TREE(&fs_info->reada_tree, GFP_NOFS & ~__GFP_WAIT);
	spin_lock_init(&fs_info->reada_lock);
C
Chris Mason 已提交
2179

2180 2181
	fs_info->thread_pool_size = min_t(unsigned long,
					  num_online_cpus() + 2, 8);
2182

2183 2184
	INIT_LIST_HEAD(&fs_info->ordered_extents);
	spin_lock_init(&fs_info->ordered_extent_lock);
2185 2186 2187 2188 2189 2190 2191
	fs_info->delayed_root = kmalloc(sizeof(struct btrfs_delayed_root),
					GFP_NOFS);
	if (!fs_info->delayed_root) {
		err = -ENOMEM;
		goto fail_iput;
	}
	btrfs_init_delayed_root(fs_info->delayed_root);
2192

A
Arne Jansen 已提交
2193 2194 2195 2196 2197 2198 2199 2200
	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);
	init_rwsem(&fs_info->scrub_super_lock);
	fs_info->scrub_workers_refcnt = 0;
2201 2202 2203
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	fs_info->check_integrity_print_mask = 0;
#endif
A
Arne Jansen 已提交
2204

2205 2206
	spin_lock_init(&fs_info->balance_lock);
	mutex_init(&fs_info->balance_mutex);
2207 2208
	atomic_set(&fs_info->balance_running, 0);
	atomic_set(&fs_info->balance_pause_req, 0);
2209
	atomic_set(&fs_info->balance_cancel_req, 0);
2210
	fs_info->balance_ctl = NULL;
2211
	init_waitqueue_head(&fs_info->balance_wait_q);
A
Arne Jansen 已提交
2212

2213 2214
	sb->s_blocksize = 4096;
	sb->s_blocksize_bits = blksize_bits(4096);
J
Jens Axboe 已提交
2215
	sb->s_bdi = &fs_info->bdi;
2216

2217
	fs_info->btree_inode->i_ino = BTRFS_BTREE_INODE_OBJECTID;
M
Miklos Szeredi 已提交
2218
	set_nlink(fs_info->btree_inode, 1);
2219 2220 2221 2222 2223 2224
	/*
	 * we set the i_size on the btree inode to the max possible int.
	 * the real end of the address space is determined by all of
	 * the devices in the system
	 */
	fs_info->btree_inode->i_size = OFFSET_MAX;
2225
	fs_info->btree_inode->i_mapping->a_ops = &btree_aops;
C
Chris Mason 已提交
2226 2227
	fs_info->btree_inode->i_mapping->backing_dev_info = &fs_info->bdi;

2228
	RB_CLEAR_NODE(&BTRFS_I(fs_info->btree_inode)->rb_node);
2229
	extent_io_tree_init(&BTRFS_I(fs_info->btree_inode)->io_tree,
2230
			     fs_info->btree_inode->i_mapping);
2231
	BTRFS_I(fs_info->btree_inode)->io_tree.track_uptodate = 0;
2232
	extent_map_tree_init(&BTRFS_I(fs_info->btree_inode)->extent_tree);
2233 2234

	BTRFS_I(fs_info->btree_inode)->io_tree.ops = &btree_extent_io_ops;
2235

2236 2237 2238
	BTRFS_I(fs_info->btree_inode)->root = tree_root;
	memset(&BTRFS_I(fs_info->btree_inode)->location, 0,
	       sizeof(struct btrfs_key));
2239 2240
	set_bit(BTRFS_INODE_DUMMY,
		&BTRFS_I(fs_info->btree_inode)->runtime_flags);
2241
	insert_inode_hash(fs_info->btree_inode);
2242

J
Josef Bacik 已提交
2243
	spin_lock_init(&fs_info->block_group_cache_lock);
2244
	fs_info->block_group_cache_tree = RB_ROOT;
2245
	fs_info->first_logical_byte = (u64)-1;
J
Josef Bacik 已提交
2246

2247
	extent_io_tree_init(&fs_info->freed_extents[0],
2248
			     fs_info->btree_inode->i_mapping);
2249
	extent_io_tree_init(&fs_info->freed_extents[1],
2250
			     fs_info->btree_inode->i_mapping);
2251
	fs_info->pinned_extents = &fs_info->freed_extents[0];
2252
	fs_info->do_barriers = 1;
2253

C
Chris Mason 已提交
2254

2255
	mutex_init(&fs_info->ordered_operations_mutex);
2256
	mutex_init(&fs_info->tree_log_mutex);
2257
	mutex_init(&fs_info->chunk_mutex);
2258 2259
	mutex_init(&fs_info->transaction_kthread_mutex);
	mutex_init(&fs_info->cleaner_mutex);
2260
	mutex_init(&fs_info->volume_mutex);
2261
	init_rwsem(&fs_info->extent_commit_sem);
2262
	init_rwsem(&fs_info->cleanup_work_sem);
2263
	init_rwsem(&fs_info->subvol_sem);
2264 2265 2266 2267 2268
	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);
	mutex_init(&fs_info->dev_replace.lock_management_lock);
	mutex_init(&fs_info->dev_replace.lock);
2269

2270
	spin_lock_init(&fs_info->qgroup_lock);
2271
	mutex_init(&fs_info->qgroup_ioctl_lock);
2272 2273 2274 2275 2276
	fs_info->qgroup_tree = RB_ROOT;
	INIT_LIST_HEAD(&fs_info->dirty_qgroups);
	fs_info->qgroup_seq = 1;
	fs_info->quota_enabled = 0;
	fs_info->pending_quota_state = 0;
J
Jan Schmidt 已提交
2277
	mutex_init(&fs_info->qgroup_rescan_lock);
2278

2279 2280 2281
	btrfs_init_free_cluster(&fs_info->meta_alloc_cluster);
	btrfs_init_free_cluster(&fs_info->data_alloc_cluster);

2282
	init_waitqueue_head(&fs_info->transaction_throttle);
2283
	init_waitqueue_head(&fs_info->transaction_wait);
S
Sage Weil 已提交
2284
	init_waitqueue_head(&fs_info->transaction_blocked_wait);
2285
	init_waitqueue_head(&fs_info->async_submit_wait);
2286

D
David Woodhouse 已提交
2287 2288
	ret = btrfs_alloc_stripe_hash_table(fs_info);
	if (ret) {
2289
		err = ret;
D
David Woodhouse 已提交
2290 2291 2292
		goto fail_alloc;
	}

2293
	__setup_root(4096, 4096, 4096, 4096, tree_root,
C
Chris Mason 已提交
2294
		     fs_info, BTRFS_ROOT_TREE_OBJECTID);
2295

2296
	invalidate_bdev(fs_devices->latest_bdev);
D
David Sterba 已提交
2297 2298 2299 2300

	/*
	 * Read super block and check the signature bytes only
	 */
Y
Yan Zheng 已提交
2301
	bh = btrfs_read_dev_super(fs_devices->latest_bdev);
2302 2303
	if (!bh) {
		err = -EINVAL;
2304
		goto fail_alloc;
2305
	}
C
Chris Mason 已提交
2306

D
David Sterba 已提交
2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321
	/*
	 * We want to check superblock checksum, the type is stored inside.
	 * Pass the whole disk block of size BTRFS_SUPER_INFO_SIZE (4k).
	 */
	if (btrfs_check_super_csum(bh->b_data)) {
		printk(KERN_ERR "btrfs: superblock checksum mismatch\n");
		err = -EINVAL;
		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
	 */
2322 2323 2324
	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));
2325
	brelse(bh);
2326

2327
	memcpy(fs_info->fsid, fs_info->super_copy->fsid, BTRFS_FSID_SIZE);
2328

D
David Sterba 已提交
2329 2330 2331 2332 2333 2334 2335
	ret = btrfs_check_super_valid(fs_info, sb->s_flags & MS_RDONLY);
	if (ret) {
		printk(KERN_ERR "btrfs: superblock contains fatal errors\n");
		err = -EINVAL;
		goto fail_alloc;
	}

2336
	disk_super = fs_info->super_copy;
2337
	if (!btrfs_super_root(disk_super))
2338
		goto fail_alloc;
2339

L
liubo 已提交
2340
	/* check FS state, whether FS is broken. */
2341 2342
	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
2343

C
Chris Mason 已提交
2344 2345 2346 2347 2348 2349 2350
	/*
	 * 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);

2351 2352 2353 2354 2355 2356
	/*
	 * 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;

Y
Yan Zheng 已提交
2357 2358 2359
	ret = btrfs_parse_options(tree_root, options);
	if (ret) {
		err = ret;
2360
		goto fail_alloc;
Y
Yan Zheng 已提交
2361
	}
2362

2363 2364 2365 2366 2367
	features = btrfs_super_incompat_flags(disk_super) &
		~BTRFS_FEATURE_INCOMPAT_SUPP;
	if (features) {
		printk(KERN_ERR "BTRFS: couldn't mount because of "
		       "unsupported optional features (%Lx).\n",
2368
		       (unsigned long long)features);
2369
		err = -EINVAL;
2370
		goto fail_alloc;
2371 2372
	}

2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389
	if (btrfs_super_leafsize(disk_super) !=
	    btrfs_super_nodesize(disk_super)) {
		printk(KERN_ERR "BTRFS: couldn't mount because metadata "
		       "blocksizes don't match.  node %d leaf %d\n",
		       btrfs_super_nodesize(disk_super),
		       btrfs_super_leafsize(disk_super));
		err = -EINVAL;
		goto fail_alloc;
	}
	if (btrfs_super_leafsize(disk_super) > BTRFS_MAX_METADATA_BLOCKSIZE) {
		printk(KERN_ERR "BTRFS: couldn't mount because metadata "
		       "blocksize (%d) was too large\n",
		       btrfs_super_leafsize(disk_super));
		err = -EINVAL;
		goto fail_alloc;
	}

2390
	features = btrfs_super_incompat_flags(disk_super);
L
Li Zefan 已提交
2391
	features |= BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF;
L
Li Zefan 已提交
2392
	if (tree_root->fs_info->compress_type == BTRFS_COMPRESS_LZO)
L
Li Zefan 已提交
2393
		features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
2394

2395 2396 2397
	if (features & BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA)
		printk(KERN_ERR "btrfs: has skinny extents\n");

2398 2399 2400 2401 2402 2403 2404 2405 2406 2407
	/*
	 * flag our filesystem as having big metadata blocks if
	 * they are bigger than the page size
	 */
	if (btrfs_super_leafsize(disk_super) > PAGE_CACHE_SIZE) {
		if (!(features & BTRFS_FEATURE_INCOMPAT_BIG_METADATA))
			printk(KERN_INFO "btrfs flagging fs with big metadata feature\n");
		features |= BTRFS_FEATURE_INCOMPAT_BIG_METADATA;
	}

2408 2409 2410 2411
	nodesize = btrfs_super_nodesize(disk_super);
	leafsize = btrfs_super_leafsize(disk_super);
	sectorsize = btrfs_super_sectorsize(disk_super);
	stripesize = btrfs_super_stripesize(disk_super);
2412
	fs_info->dirty_metadata_batch = leafsize * (1 + ilog2(nr_cpu_ids));
2413
	fs_info->delalloc_batch = sectorsize * 512 * (1 + ilog2(nr_cpu_ids));
2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426

	/*
	 * 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) &&
	    (sectorsize != leafsize)) {
		printk(KERN_WARNING "btrfs: unequal leaf/node/sector sizes "
				"are not allowed for mixed block groups on %s\n",
				sb->s_id);
		goto fail_alloc;
	}

2427 2428 2429 2430
	/*
	 * Needn't use the lock because there is no other task which will
	 * update the flag.
	 */
L
Li Zefan 已提交
2431
	btrfs_set_super_incompat_flags(disk_super, features);
2432

2433 2434 2435 2436 2437
	features = btrfs_super_compat_ro_flags(disk_super) &
		~BTRFS_FEATURE_COMPAT_RO_SUPP;
	if (!(sb->s_flags & MS_RDONLY) && features) {
		printk(KERN_ERR "BTRFS: couldn't mount RDWR because of "
		       "unsupported option features (%Lx).\n",
2438
		       (unsigned long long)features);
2439
		err = -EINVAL;
2440
		goto fail_alloc;
2441
	}
2442 2443 2444 2445

	btrfs_init_workers(&fs_info->generic_worker,
			   "genwork", 1, NULL);

2446
	btrfs_init_workers(&fs_info->workers, "worker",
2447 2448
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
C
Chris Mason 已提交
2449

2450
	btrfs_init_workers(&fs_info->delalloc_workers, "delalloc",
2451 2452
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2453

2454 2455 2456 2457
	btrfs_init_workers(&fs_info->flush_workers, "flush_delalloc",
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);

2458
	btrfs_init_workers(&fs_info->submit_workers, "submit",
2459
			   min_t(u64, fs_devices->num_devices,
2460 2461
			   fs_info->thread_pool_size),
			   &fs_info->generic_worker);
2462

2463 2464 2465
	btrfs_init_workers(&fs_info->caching_workers, "cache",
			   2, &fs_info->generic_worker);

2466 2467 2468 2469 2470
	/* 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.idle_thresh = 64;
2471

2472
	fs_info->workers.idle_thresh = 16;
C
Chris Mason 已提交
2473
	fs_info->workers.ordered = 1;
2474

2475 2476 2477
	fs_info->delalloc_workers.idle_thresh = 2;
	fs_info->delalloc_workers.ordered = 1;

2478 2479
	btrfs_init_workers(&fs_info->fixup_workers, "fixup", 1,
			   &fs_info->generic_worker);
2480
	btrfs_init_workers(&fs_info->endio_workers, "endio",
2481 2482
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2483
	btrfs_init_workers(&fs_info->endio_meta_workers, "endio-meta",
2484 2485
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2486
	btrfs_init_workers(&fs_info->endio_meta_write_workers,
2487 2488
			   "endio-meta-write", fs_info->thread_pool_size,
			   &fs_info->generic_worker);
D
David Woodhouse 已提交
2489 2490 2491 2492 2493 2494
	btrfs_init_workers(&fs_info->endio_raid56_workers,
			   "endio-raid56", fs_info->thread_pool_size,
			   &fs_info->generic_worker);
	btrfs_init_workers(&fs_info->rmw_workers,
			   "rmw", fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2495
	btrfs_init_workers(&fs_info->endio_write_workers, "endio-write",
2496 2497
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
J
Josef Bacik 已提交
2498 2499
	btrfs_init_workers(&fs_info->endio_freespace_worker, "freespace-write",
			   1, &fs_info->generic_worker);
2500 2501 2502
	btrfs_init_workers(&fs_info->delayed_workers, "delayed-meta",
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2503 2504 2505
	btrfs_init_workers(&fs_info->readahead_workers, "readahead",
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
J
Jan Schmidt 已提交
2506 2507
	btrfs_init_workers(&fs_info->qgroup_rescan_workers, "qgroup-rescan", 1,
			   &fs_info->generic_worker);
2508 2509 2510 2511 2512 2513

	/*
	 * endios are largely parallel and should have a very
	 * low idle thresh
	 */
	fs_info->endio_workers.idle_thresh = 4;
2514
	fs_info->endio_meta_workers.idle_thresh = 4;
D
David Woodhouse 已提交
2515 2516
	fs_info->endio_raid56_workers.idle_thresh = 4;
	fs_info->rmw_workers.idle_thresh = 2;
2517

2518 2519
	fs_info->endio_write_workers.idle_thresh = 2;
	fs_info->endio_meta_write_workers.idle_thresh = 2;
2520
	fs_info->readahead_workers.idle_thresh = 2;
2521

2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532
	/*
	 * btrfs_start_workers can really only fail because of ENOMEM so just
	 * return -ENOMEM if any of these fail.
	 */
	ret = btrfs_start_workers(&fs_info->workers);
	ret |= btrfs_start_workers(&fs_info->generic_worker);
	ret |= btrfs_start_workers(&fs_info->submit_workers);
	ret |= btrfs_start_workers(&fs_info->delalloc_workers);
	ret |= btrfs_start_workers(&fs_info->fixup_workers);
	ret |= btrfs_start_workers(&fs_info->endio_workers);
	ret |= btrfs_start_workers(&fs_info->endio_meta_workers);
D
David Woodhouse 已提交
2533 2534
	ret |= btrfs_start_workers(&fs_info->rmw_workers);
	ret |= btrfs_start_workers(&fs_info->endio_raid56_workers);
2535 2536 2537 2538 2539 2540
	ret |= btrfs_start_workers(&fs_info->endio_meta_write_workers);
	ret |= btrfs_start_workers(&fs_info->endio_write_workers);
	ret |= btrfs_start_workers(&fs_info->endio_freespace_worker);
	ret |= btrfs_start_workers(&fs_info->delayed_workers);
	ret |= btrfs_start_workers(&fs_info->caching_workers);
	ret |= btrfs_start_workers(&fs_info->readahead_workers);
2541
	ret |= btrfs_start_workers(&fs_info->flush_workers);
J
Jan Schmidt 已提交
2542
	ret |= btrfs_start_workers(&fs_info->qgroup_rescan_workers);
2543
	if (ret) {
2544
		err = -ENOMEM;
2545 2546
		goto fail_sb_buffer;
	}
2547

2548
	fs_info->bdi.ra_pages *= btrfs_super_num_devices(disk_super);
C
Chris Mason 已提交
2549 2550
	fs_info->bdi.ra_pages = max(fs_info->bdi.ra_pages,
				    4 * 1024 * 1024 / PAGE_CACHE_SIZE);
2551

2552 2553 2554
	tree_root->nodesize = nodesize;
	tree_root->leafsize = leafsize;
	tree_root->sectorsize = sectorsize;
2555
	tree_root->stripesize = stripesize;
2556 2557 2558

	sb->s_blocksize = sectorsize;
	sb->s_blocksize_bits = blksize_bits(sectorsize);
2559

2560
	if (disk_super->magic != cpu_to_le64(BTRFS_MAGIC)) {
C
Chris Mason 已提交
2561
		printk(KERN_INFO "btrfs: valid FS not found on %s\n", sb->s_id);
C
Chris Mason 已提交
2562 2563
		goto fail_sb_buffer;
	}
2564

2565 2566 2567
	if (sectorsize != PAGE_SIZE) {
		printk(KERN_WARNING "btrfs: Incompatible sector size(%lu) "
		       "found on %s\n", (unsigned long)sectorsize, sb->s_id);
2568 2569 2570
		goto fail_sb_buffer;
	}

2571
	mutex_lock(&fs_info->chunk_mutex);
Y
Yan Zheng 已提交
2572
	ret = btrfs_read_sys_array(tree_root);
2573
	mutex_unlock(&fs_info->chunk_mutex);
2574
	if (ret) {
C
Chris Mason 已提交
2575 2576
		printk(KERN_WARNING "btrfs: failed to read the system "
		       "array on %s\n", sb->s_id);
2577
		goto fail_sb_buffer;
2578
	}
2579 2580 2581

	blocksize = btrfs_level_size(tree_root,
				     btrfs_super_chunk_root_level(disk_super));
2582
	generation = btrfs_super_chunk_root_generation(disk_super);
2583 2584 2585 2586 2587 2588

	__setup_root(nodesize, leafsize, sectorsize, stripesize,
		     chunk_root, fs_info, BTRFS_CHUNK_TREE_OBJECTID);

	chunk_root->node = read_tree_block(chunk_root,
					   btrfs_super_chunk_root(disk_super),
2589
					   blocksize, generation);
2590 2591
	if (!chunk_root->node ||
	    !test_bit(EXTENT_BUFFER_UPTODATE, &chunk_root->node->bflags)) {
2592 2593
		printk(KERN_WARNING "btrfs: failed to read chunk root on %s\n",
		       sb->s_id);
C
Chris Mason 已提交
2594
		goto fail_tree_roots;
2595
	}
2596 2597
	btrfs_set_root_node(&chunk_root->root_item, chunk_root->node);
	chunk_root->commit_root = btrfs_root_node(chunk_root);
2598

2599
	read_extent_buffer(chunk_root->node, fs_info->chunk_tree_uuid,
C
Chris Mason 已提交
2600 2601
	   (unsigned long)btrfs_header_chunk_tree_uuid(chunk_root->node),
	   BTRFS_UUID_SIZE);
2602

2603
	ret = btrfs_read_chunk_tree(chunk_root);
Y
Yan Zheng 已提交
2604
	if (ret) {
C
Chris Mason 已提交
2605 2606
		printk(KERN_WARNING "btrfs: failed to read chunk tree on %s\n",
		       sb->s_id);
C
Chris Mason 已提交
2607
		goto fail_tree_roots;
Y
Yan Zheng 已提交
2608
	}
2609

2610 2611 2612 2613 2614
	/*
	 * keep the device that is marked to be the target device for the
	 * dev_replace procedure
	 */
	btrfs_close_extra_devices(fs_info, fs_devices, 0);
2615

2616 2617 2618 2619 2620 2621
	if (!fs_devices->latest_bdev) {
		printk(KERN_CRIT "btrfs: failed to read devices on %s\n",
		       sb->s_id);
		goto fail_tree_roots;
	}

C
Chris Mason 已提交
2622
retry_root_backup:
2623 2624
	blocksize = btrfs_level_size(tree_root,
				     btrfs_super_root_level(disk_super));
2625
	generation = btrfs_super_generation(disk_super);
2626

C
Chris Mason 已提交
2627
	tree_root->node = read_tree_block(tree_root,
2628
					  btrfs_super_root(disk_super),
2629
					  blocksize, generation);
C
Chris Mason 已提交
2630 2631
	if (!tree_root->node ||
	    !test_bit(EXTENT_BUFFER_UPTODATE, &tree_root->node->bflags)) {
2632 2633
		printk(KERN_WARNING "btrfs: failed to read tree root on %s\n",
		       sb->s_id);
C
Chris Mason 已提交
2634 2635

		goto recovery_tree_root;
2636
	}
C
Chris Mason 已提交
2637

2638 2639
	btrfs_set_root_node(&tree_root->root_item, tree_root->node);
	tree_root->commit_root = btrfs_root_node(tree_root);
2640 2641

	ret = find_and_setup_root(tree_root, fs_info,
C
Chris Mason 已提交
2642
				  BTRFS_EXTENT_TREE_OBJECTID, extent_root);
2643
	if (ret)
C
Chris Mason 已提交
2644
		goto recovery_tree_root;
2645 2646 2647 2648 2649
	extent_root->track_dirty = 1;

	ret = find_and_setup_root(tree_root, fs_info,
				  BTRFS_DEV_TREE_OBJECTID, dev_root);
	if (ret)
C
Chris Mason 已提交
2650
		goto recovery_tree_root;
2651
	dev_root->track_dirty = 1;
2652

2653 2654 2655
	ret = find_and_setup_root(tree_root, fs_info,
				  BTRFS_CSUM_TREE_OBJECTID, csum_root);
	if (ret)
C
Chris Mason 已提交
2656
		goto recovery_tree_root;
2657 2658
	csum_root->track_dirty = 1;

2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669
	ret = find_and_setup_root(tree_root, fs_info,
				  BTRFS_QUOTA_TREE_OBJECTID, quota_root);
	if (ret) {
		kfree(quota_root);
		quota_root = fs_info->quota_root = NULL;
	} else {
		quota_root->track_dirty = 1;
		fs_info->quota_enabled = 1;
		fs_info->pending_quota_state = 1;
	}

2670 2671 2672
	fs_info->generation = generation;
	fs_info->last_trans_committed = generation;

2673 2674 2675 2676 2677 2678
	ret = btrfs_recover_balance(fs_info);
	if (ret) {
		printk(KERN_WARNING "btrfs: failed to recover balance\n");
		goto fail_block_groups;
	}

2679 2680 2681 2682 2683 2684 2685
	ret = btrfs_init_dev_stats(fs_info);
	if (ret) {
		printk(KERN_ERR "btrfs: failed to init dev_stats: %d\n",
		       ret);
		goto fail_block_groups;
	}

2686 2687 2688 2689 2690 2691 2692 2693
	ret = btrfs_init_dev_replace(fs_info);
	if (ret) {
		pr_err("btrfs: failed to init dev_replace: %d\n", ret);
		goto fail_block_groups;
	}

	btrfs_close_extra_devices(fs_info, fs_devices, 1);

2694 2695 2696 2697 2698 2699
	ret = btrfs_init_space_info(fs_info);
	if (ret) {
		printk(KERN_ERR "Failed to initial space info: %d\n", ret);
		goto fail_block_groups;
	}

2700 2701 2702 2703 2704
	ret = btrfs_read_block_groups(extent_root);
	if (ret) {
		printk(KERN_ERR "Failed to read block groups: %d\n", ret);
		goto fail_block_groups;
	}
2705 2706
	fs_info->num_tolerated_disk_barrier_failures =
		btrfs_calc_num_tolerated_disk_barrier_failures(fs_info);
2707 2708 2709 2710 2711 2712 2713
	if (fs_info->fs_devices->missing_devices >
	     fs_info->num_tolerated_disk_barrier_failures &&
	    !(sb->s_flags & MS_RDONLY)) {
		printk(KERN_WARNING
		       "Btrfs: too many missing devices, writeable mount is not allowed\n");
		goto fail_block_groups;
	}
C
Chris Mason 已提交
2714

2715 2716
	fs_info->cleaner_kthread = kthread_run(cleaner_kthread, tree_root,
					       "btrfs-cleaner");
2717
	if (IS_ERR(fs_info->cleaner_kthread))
2718
		goto fail_block_groups;
2719 2720 2721 2722

	fs_info->transaction_kthread = kthread_run(transaction_kthread,
						   tree_root,
						   "btrfs-transaction");
2723
	if (IS_ERR(fs_info->transaction_kthread))
2724
		goto fail_cleaner;
2725

C
Chris Mason 已提交
2726 2727 2728 2729 2730 2731 2732 2733
	if (!btrfs_test_opt(tree_root, SSD) &&
	    !btrfs_test_opt(tree_root, NOSSD) &&
	    !fs_info->fs_devices->rotating) {
		printk(KERN_INFO "Btrfs detected SSD devices, enabling SSD "
		       "mode\n");
		btrfs_set_opt(fs_info->mount_opt, SSD);
	}

2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	if (btrfs_test_opt(tree_root, CHECK_INTEGRITY)) {
		ret = btrfsic_mount(tree_root, fs_devices,
				    btrfs_test_opt(tree_root,
					CHECK_INTEGRITY_INCLUDING_EXTENT_DATA) ?
				    1 : 0,
				    fs_info->check_integrity_print_mask);
		if (ret)
			printk(KERN_WARNING "btrfs: failed to initialize"
			       " integrity check module %s\n", sb->s_id);
	}
#endif
2746 2747 2748
	ret = btrfs_read_qgroup_config(fs_info);
	if (ret)
		goto fail_trans_kthread;
2749

L
liubo 已提交
2750
	/* do not make disk changes in broken FS */
2751
	if (btrfs_super_log_root(disk_super) != 0) {
2752 2753
		u64 bytenr = btrfs_super_log_root(disk_super);

2754
		if (fs_devices->rw_devices == 0) {
C
Chris Mason 已提交
2755 2756
			printk(KERN_WARNING "Btrfs log replay required "
			       "on RO media\n");
2757
			err = -EIO;
2758
			goto fail_qgroup;
2759
		}
2760 2761 2762
		blocksize =
		     btrfs_level_size(tree_root,
				      btrfs_super_log_root_level(disk_super));
C
Chris Mason 已提交
2763

A
Al Viro 已提交
2764
		log_tree_root = btrfs_alloc_root(fs_info);
2765 2766
		if (!log_tree_root) {
			err = -ENOMEM;
2767
			goto fail_qgroup;
2768
		}
2769 2770 2771 2772 2773

		__setup_root(nodesize, leafsize, sectorsize, stripesize,
			     log_tree_root, fs_info, BTRFS_TREE_LOG_OBJECTID);

		log_tree_root->node = read_tree_block(tree_root, bytenr,
2774 2775
						      blocksize,
						      generation + 1);
2776 2777 2778 2779 2780 2781 2782
		if (!log_tree_root->node ||
		    !extent_buffer_uptodate(log_tree_root->node)) {
			printk(KERN_ERR "btrfs: failed to read log tree\n");
			free_extent_buffer(log_tree_root->node);
			kfree(log_tree_root);
			goto fail_trans_kthread;
		}
2783
		/* returns with log_tree_root freed on success */
2784
		ret = btrfs_recover_log_trees(log_tree_root);
2785 2786 2787 2788 2789 2790 2791
		if (ret) {
			btrfs_error(tree_root->fs_info, ret,
				    "Failed to recover log tree");
			free_extent_buffer(log_tree_root->node);
			kfree(log_tree_root);
			goto fail_trans_kthread;
		}
2792 2793

		if (sb->s_flags & MS_RDONLY) {
2794 2795 2796
			ret = btrfs_commit_super(tree_root);
			if (ret)
				goto fail_trans_kthread;
2797
		}
2798
	}
Z
Zheng Yan 已提交
2799

2800
	ret = btrfs_find_orphan_roots(tree_root);
2801 2802
	if (ret)
		goto fail_trans_kthread;
2803

2804
	if (!(sb->s_flags & MS_RDONLY)) {
2805
		ret = btrfs_cleanup_fs_roots(fs_info);
2806 2807
		if (ret)
			goto fail_trans_kthread;
2808

2809
		ret = btrfs_recover_relocation(tree_root);
2810 2811 2812 2813
		if (ret < 0) {
			printk(KERN_WARNING
			       "btrfs: failed to recover relocation\n");
			err = -EINVAL;
2814
			goto fail_qgroup;
2815
		}
2816
	}
Z
Zheng Yan 已提交
2817

2818 2819 2820 2821 2822 2823
	location.objectid = BTRFS_FS_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
	location.offset = (u64)-1;

	fs_info->fs_root = btrfs_read_fs_root_no_name(fs_info, &location);
	if (!fs_info->fs_root)
2824
		goto fail_qgroup;
2825 2826
	if (IS_ERR(fs_info->fs_root)) {
		err = PTR_ERR(fs_info->fs_root);
2827
		goto fail_qgroup;
2828
	}
C
Chris Mason 已提交
2829

2830 2831
	if (sb->s_flags & MS_RDONLY)
		return 0;
I
Ilya Dryomov 已提交
2832

2833 2834 2835
	down_read(&fs_info->cleanup_work_sem);
	if ((ret = btrfs_orphan_cleanup(fs_info->fs_root)) ||
	    (ret = btrfs_orphan_cleanup(fs_info->tree_root))) {
2836
		up_read(&fs_info->cleanup_work_sem);
2837 2838 2839 2840
		close_ctree(tree_root);
		return ret;
	}
	up_read(&fs_info->cleanup_work_sem);
I
Ilya Dryomov 已提交
2841

2842 2843 2844 2845 2846
	ret = btrfs_resume_balance_async(fs_info);
	if (ret) {
		printk(KERN_WARNING "btrfs: failed to resume balance\n");
		close_ctree(tree_root);
		return ret;
2847 2848
	}

2849 2850 2851 2852 2853 2854 2855
	ret = btrfs_resume_dev_replace_async(fs_info);
	if (ret) {
		pr_warn("btrfs: failed to resume dev_replace\n");
		close_ctree(tree_root);
		return ret;
	}

A
Al Viro 已提交
2856
	return 0;
C
Chris Mason 已提交
2857

2858 2859
fail_qgroup:
	btrfs_free_qgroup_config(fs_info);
2860 2861
fail_trans_kthread:
	kthread_stop(fs_info->transaction_kthread);
2862
	del_fs_roots(fs_info);
J
Josef Bacik 已提交
2863
	btrfs_cleanup_transaction(fs_info->tree_root);
2864
fail_cleaner:
2865
	kthread_stop(fs_info->cleaner_kthread);
2866 2867 2868 2869 2870 2871 2872

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

2873
fail_block_groups:
J
Josef Bacik 已提交
2874
	btrfs_put_block_group_cache(fs_info);
2875
	btrfs_free_block_groups(fs_info);
C
Chris Mason 已提交
2876 2877 2878

fail_tree_roots:
	free_root_pointers(fs_info, 1);
2879
	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
2880

C
Chris Mason 已提交
2881
fail_sb_buffer:
L
Liu Bo 已提交
2882
	btrfs_stop_all_workers(fs_info);
2883
fail_alloc:
2884
fail_iput:
2885 2886
	btrfs_mapping_tree_free(&fs_info->mapping_tree);

2887
	iput(fs_info->btree_inode);
2888 2889
fail_delalloc_bytes:
	percpu_counter_destroy(&fs_info->delalloc_bytes);
2890 2891
fail_dirty_metadata_bytes:
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
2892
fail_bdi:
2893
	bdi_destroy(&fs_info->bdi);
2894 2895
fail_srcu:
	cleanup_srcu_struct(&fs_info->subvol_srcu);
2896
fail:
D
David Woodhouse 已提交
2897
	btrfs_free_stripe_hash_table(fs_info);
2898
	btrfs_close_devices(fs_info->fs_devices);
A
Al Viro 已提交
2899
	return err;
C
Chris Mason 已提交
2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917

recovery_tree_root:
	if (!btrfs_test_opt(tree_root, RECOVERY))
		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;
2918 2919
}

2920 2921 2922 2923 2924
static void btrfs_end_buffer_write_sync(struct buffer_head *bh, int uptodate)
{
	if (uptodate) {
		set_buffer_uptodate(bh);
	} else {
2925 2926 2927
		struct btrfs_device *device = (struct btrfs_device *)
			bh->b_private;

2928 2929 2930
		printk_ratelimited_in_rcu(KERN_WARNING "lost page write due to "
					  "I/O error on %s\n",
					  rcu_str_deref(device->name));
2931 2932 2933
		/* note, we dont' set_buffer_write_io_error because we have
		 * our own ways of dealing with the IO errors
		 */
2934
		clear_buffer_uptodate(bh);
2935
		btrfs_dev_stat_inc_and_print(device, BTRFS_DEV_STAT_WRITE_ERRS);
2936 2937 2938 2939 2940
	}
	unlock_buffer(bh);
	put_bh(bh);
}

Y
Yan Zheng 已提交
2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964
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;
	u64 bytenr;

	/* 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++) {
		bytenr = btrfs_sb_offset(i);
		if (bytenr + 4096 >= i_size_read(bdev->bd_inode))
			break;
		bh = __bread(bdev, bytenr / 4096, 4096);
		if (!bh)
			continue;

		super = (struct btrfs_super_block *)bh->b_data;
		if (btrfs_super_bytenr(super) != bytenr ||
2965
		    super->magic != cpu_to_le64(BTRFS_MAGIC)) {
Y
Yan Zheng 已提交
2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980
			brelse(bh);
			continue;
		}

		if (!latest || btrfs_super_generation(super) > transid) {
			brelse(latest);
			latest = bh;
			transid = btrfs_super_generation(super);
		} else {
			brelse(bh);
		}
	}
	return latest;
}

2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991
/*
 * 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
2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013
static int write_dev_supers(struct btrfs_device *device,
			    struct btrfs_super_block *sb,
			    int do_barriers, int wait, int max_mirrors)
{
	struct buffer_head *bh;
	int i;
	int ret;
	int errors = 0;
	u32 crc;
	u64 bytenr;

	if (max_mirrors == 0)
		max_mirrors = BTRFS_SUPER_MIRROR_MAX;

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

		if (wait) {
			bh = __find_get_block(device->bdev, bytenr / 4096,
					      BTRFS_SUPER_INFO_SIZE);
3014 3015 3016 3017
			if (!bh) {
				errors++;
				continue;
			}
Y
Yan Zheng 已提交
3018
			wait_on_buffer(bh);
3019 3020 3021 3022 3023 3024 3025 3026 3027
			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
3028 3029 3030 3031
		} else {
			btrfs_set_super_bytenr(sb, bytenr);

			crc = ~(u32)0;
3032
			crc = btrfs_csum_data((char *)sb +
Y
Yan Zheng 已提交
3033 3034 3035 3036 3037
					      BTRFS_CSUM_SIZE, crc,
					      BTRFS_SUPER_INFO_SIZE -
					      BTRFS_CSUM_SIZE);
			btrfs_csum_final(crc, sb->csum);

3038 3039 3040 3041
			/*
			 * one reference for us, and we leave it for the
			 * caller
			 */
Y
Yan Zheng 已提交
3042 3043
			bh = __getblk(device->bdev, bytenr / 4096,
				      BTRFS_SUPER_INFO_SIZE);
3044 3045 3046 3047 3048 3049 3050
			if (!bh) {
				printk(KERN_ERR "btrfs: couldn't get super "
				       "buffer head for bytenr %Lu\n", bytenr);
				errors++;
				continue;
			}

Y
Yan Zheng 已提交
3051 3052
			memcpy(bh->b_data, sb, BTRFS_SUPER_INFO_SIZE);

3053
			/* one reference for submit_bh */
Y
Yan Zheng 已提交
3054
			get_bh(bh);
3055 3056

			set_buffer_uptodate(bh);
Y
Yan Zheng 已提交
3057 3058
			lock_buffer(bh);
			bh->b_end_io = btrfs_end_buffer_write_sync;
3059
			bh->b_private = device;
Y
Yan Zheng 已提交
3060 3061
		}

C
Chris Mason 已提交
3062 3063 3064 3065
		/*
		 * we fua the first super.  The others we allow
		 * to go down lazy.
		 */
3066
		ret = btrfsic_submit_bh(WRITE_FUA, bh);
3067
		if (ret)
Y
Yan Zheng 已提交
3068 3069 3070 3071 3072
			errors++;
	}
	return errors < i ? 0 : -1;
}

C
Chris Mason 已提交
3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111
/*
 * endio for the write_dev_flush, this will wake anyone waiting
 * for the barrier when it is done
 */
static void btrfs_end_empty_barrier(struct bio *bio, int err)
{
	if (err) {
		if (err == -EOPNOTSUPP)
			set_bit(BIO_EOPNOTSUPP, &bio->bi_flags);
		clear_bit(BIO_UPTODATE, &bio->bi_flags);
	}
	if (bio->bi_private)
		complete(bio->bi_private);
	bio_put(bio);
}

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

	if (device->nobarriers)
		return 0;

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

		wait_for_completion(&device->flush_wait);

		if (bio_flagged(bio, BIO_EOPNOTSUPP)) {
3112 3113
			printk_in_rcu("btrfs: disabling barriers on dev %s\n",
				      rcu_str_deref(device->name));
C
Chris Mason 已提交
3114
			device->nobarriers = 1;
3115
		} else if (!bio_flagged(bio, BIO_UPTODATE)) {
C
Chris Mason 已提交
3116
			ret = -EIO;
3117 3118
			btrfs_dev_stat_inc_and_print(device,
				BTRFS_DEV_STAT_FLUSH_ERRS);
C
Chris Mason 已提交
3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131
		}

		/* 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
	 */
3132
	device->flush_bio = NULL;
3133
	bio = btrfs_io_bio_alloc(GFP_NOFS, 0);
C
Chris Mason 已提交
3134 3135 3136 3137 3138 3139 3140 3141 3142 3143
	if (!bio)
		return -ENOMEM;

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

	bio_get(bio);
3144
	btrfsic_submit_bio(WRITE_FLUSH, bio);
C
Chris Mason 已提交
3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156

	return 0;
}

/*
 * send an empty flush down to each device in parallel,
 * then wait for them
 */
static int barrier_all_devices(struct btrfs_fs_info *info)
{
	struct list_head *head;
	struct btrfs_device *dev;
3157 3158
	int errors_send = 0;
	int errors_wait = 0;
C
Chris Mason 已提交
3159 3160 3161 3162 3163 3164
	int ret;

	/* send down all the barriers */
	head = &info->fs_devices->devices;
	list_for_each_entry_rcu(dev, head, dev_list) {
		if (!dev->bdev) {
3165
			errors_send++;
C
Chris Mason 已提交
3166 3167 3168 3169 3170 3171 3172
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

		ret = write_dev_flush(dev, 0);
		if (ret)
3173
			errors_send++;
C
Chris Mason 已提交
3174 3175 3176 3177 3178
	}

	/* wait for all the barriers */
	list_for_each_entry_rcu(dev, head, dev_list) {
		if (!dev->bdev) {
3179
			errors_wait++;
C
Chris Mason 已提交
3180 3181 3182 3183 3184 3185 3186
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

		ret = write_dev_flush(dev, 1);
		if (ret)
3187
			errors_wait++;
C
Chris Mason 已提交
3188
	}
3189 3190
	if (errors_send > info->num_tolerated_disk_barrier_failures ||
	    errors_wait > info->num_tolerated_disk_barrier_failures)
C
Chris Mason 已提交
3191 3192 3193 3194
		return -EIO;
	return 0;
}

3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254
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 num_types = 4;
	int i;
	int c;
	int num_tolerated_disk_barrier_failures =
		(int)fs_info->fs_devices->num_devices;

	for (i = 0; i < num_types; i++) {
		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++) {
			if (!list_empty(&sinfo->block_groups[c])) {
				u64 flags;

				btrfs_get_block_group_info(
					&sinfo->block_groups[c], &space);
				if (space.total_bytes == 0 ||
				    space.used_bytes == 0)
					continue;
				flags = space.flags;
				/*
				 * return
				 * 0: if dup, single or RAID0 is configured for
				 *    any of metadata, system or data, else
				 * 1: if RAID5 is configured, or if RAID1 or
				 *    RAID10 is configured and only two mirrors
				 *    are used, else
				 * 2: if RAID6 is configured, else
				 * num_mirrors - 1: if RAID1 or RAID10 is
				 *                  configured and more than
				 *                  2 mirrors are used.
				 */
				if (num_tolerated_disk_barrier_failures > 0 &&
				    ((flags & (BTRFS_BLOCK_GROUP_DUP |
					       BTRFS_BLOCK_GROUP_RAID0)) ||
				     ((flags & BTRFS_BLOCK_GROUP_PROFILE_MASK)
				      == 0)))
					num_tolerated_disk_barrier_failures = 0;
D
David Woodhouse 已提交
3255 3256 3257 3258 3259 3260 3261 3262 3263 3264
				else if (num_tolerated_disk_barrier_failures > 1) {
					if (flags & (BTRFS_BLOCK_GROUP_RAID1 |
					    BTRFS_BLOCK_GROUP_RAID5 |
					    BTRFS_BLOCK_GROUP_RAID10)) {
						num_tolerated_disk_barrier_failures = 1;
					} else if (flags &
						   BTRFS_BLOCK_GROUP_RAID5) {
						num_tolerated_disk_barrier_failures = 2;
					}
				}
3265 3266 3267 3268 3269 3270 3271 3272
			}
		}
		up_read(&sinfo->groups_sem);
	}

	return num_tolerated_disk_barrier_failures;
}

3273
static int write_all_supers(struct btrfs_root *root, int max_mirrors)
3274
{
3275
	struct list_head *head;
3276
	struct btrfs_device *dev;
3277
	struct btrfs_super_block *sb;
3278 3279 3280
	struct btrfs_dev_item *dev_item;
	int ret;
	int do_barriers;
3281 3282
	int max_errors;
	int total_errors = 0;
3283
	u64 flags;
3284

3285
	max_errors = btrfs_super_num_devices(root->fs_info->super_copy) - 1;
3286
	do_barriers = !btrfs_test_opt(root, NOBARRIER);
C
Chris Mason 已提交
3287
	backup_super_roots(root->fs_info);
3288

3289
	sb = root->fs_info->super_for_commit;
3290
	dev_item = &sb->dev_item;
3291

3292
	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
3293
	head = &root->fs_info->fs_devices->devices;
C
Chris Mason 已提交
3294

3295 3296 3297 3298 3299 3300 3301 3302 3303 3304
	if (do_barriers) {
		ret = barrier_all_devices(root->fs_info);
		if (ret) {
			mutex_unlock(
				&root->fs_info->fs_devices->device_list_mutex);
			btrfs_error(root->fs_info, ret,
				    "errors while submitting device barriers.");
			return ret;
		}
	}
C
Chris Mason 已提交
3305

3306
	list_for_each_entry_rcu(dev, head, dev_list) {
3307 3308 3309 3310
		if (!dev->bdev) {
			total_errors++;
			continue;
		}
Y
Yan Zheng 已提交
3311
		if (!dev->in_fs_metadata || !dev->writeable)
3312 3313
			continue;

Y
Yan Zheng 已提交
3314
		btrfs_set_stack_device_generation(dev_item, 0);
3315 3316 3317 3318 3319 3320 3321 3322
		btrfs_set_stack_device_type(dev_item, dev->type);
		btrfs_set_stack_device_id(dev_item, dev->devid);
		btrfs_set_stack_device_total_bytes(dev_item, dev->total_bytes);
		btrfs_set_stack_device_bytes_used(dev_item, dev->bytes_used);
		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
3323
		memcpy(dev_item->fsid, dev->fs_devices->fsid, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3324

3325 3326 3327
		flags = btrfs_super_flags(sb);
		btrfs_set_super_flags(sb, flags | BTRFS_HEADER_FLAG_WRITTEN);

Y
Yan Zheng 已提交
3328
		ret = write_dev_supers(dev, sb, do_barriers, 0, max_mirrors);
3329 3330
		if (ret)
			total_errors++;
3331
	}
3332
	if (total_errors > max_errors) {
C
Chris Mason 已提交
3333 3334
		printk(KERN_ERR "btrfs: %d errors while writing supers\n",
		       total_errors);
3335 3336

		/* This shouldn't happen. FUA is masked off if unsupported */
3337 3338
		BUG();
	}
3339

Y
Yan Zheng 已提交
3340
	total_errors = 0;
3341
	list_for_each_entry_rcu(dev, head, dev_list) {
3342 3343
		if (!dev->bdev)
			continue;
Y
Yan Zheng 已提交
3344
		if (!dev->in_fs_metadata || !dev->writeable)
3345 3346
			continue;

Y
Yan Zheng 已提交
3347 3348 3349
		ret = write_dev_supers(dev, sb, do_barriers, 1, max_mirrors);
		if (ret)
			total_errors++;
3350
	}
3351
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
3352
	if (total_errors > max_errors) {
3353 3354 3355
		btrfs_error(root->fs_info, -EIO,
			    "%d errors while writing supers", total_errors);
		return -EIO;
3356
	}
3357 3358 3359
	return 0;
}

Y
Yan Zheng 已提交
3360 3361
int write_ctree_super(struct btrfs_trans_handle *trans,
		      struct btrfs_root *root, int max_mirrors)
3362
{
3363
	int ret;
3364

Y
Yan Zheng 已提交
3365
	ret = write_all_supers(root, max_mirrors);
3366
	return ret;
C
Chris Mason 已提交
3367 3368
}

3369
void btrfs_free_fs_root(struct btrfs_fs_info *fs_info, struct btrfs_root *root)
C
Chris Mason 已提交
3370
{
3371
	spin_lock(&fs_info->fs_roots_radix_lock);
C
Chris Mason 已提交
3372 3373
	radix_tree_delete(&fs_info->fs_roots_radix,
			  (unsigned long)root->root_key.objectid);
3374
	spin_unlock(&fs_info->fs_roots_radix_lock);
3375 3376 3377 3378

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

L
Liu Bo 已提交
3379
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
L
Liu Bo 已提交
3380 3381 3382 3383
		btrfs_free_log(NULL, root);
		btrfs_free_log_root_tree(NULL, fs_info);
	}

3384 3385
	__btrfs_remove_free_space_cache(root->free_ino_pinned);
	__btrfs_remove_free_space_cache(root->free_ino_ctl);
3386 3387 3388 3389 3390
	free_fs_root(root);
}

static void free_fs_root(struct btrfs_root *root)
{
3391
	iput(root->cache_inode);
3392
	WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree));
3393 3394
	if (root->anon_dev)
		free_anon_bdev(root->anon_dev);
3395 3396
	free_extent_buffer(root->node);
	free_extent_buffer(root->commit_root);
3397 3398
	kfree(root->free_ino_ctl);
	kfree(root->free_ino_pinned);
C
Chris Mason 已提交
3399
	kfree(root->name);
C
Chris Mason 已提交
3400 3401 3402
	kfree(root);
}

Y
Yan Zheng 已提交
3403
int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
C
Chris Mason 已提交
3404
{
Y
Yan Zheng 已提交
3405 3406 3407
	u64 root_objectid = 0;
	struct btrfs_root *gang[8];
	int i;
3408
	int ret;
3409

Y
Yan Zheng 已提交
3410 3411 3412 3413 3414 3415
	while (1) {
		ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix,
					     (void **)gang, root_objectid,
					     ARRAY_SIZE(gang));
		if (!ret)
			break;
3416 3417

		root_objectid = gang[ret - 1]->root_key.objectid + 1;
Y
Yan Zheng 已提交
3418
		for (i = 0; i < ret; i++) {
3419 3420
			int err;

Y
Yan Zheng 已提交
3421
			root_objectid = gang[i]->root_key.objectid;
3422 3423 3424
			err = btrfs_orphan_cleanup(gang[i]);
			if (err)
				return err;
Y
Yan Zheng 已提交
3425 3426 3427 3428 3429
		}
		root_objectid++;
	}
	return 0;
}
3430

Y
Yan Zheng 已提交
3431 3432 3433 3434
int btrfs_commit_super(struct btrfs_root *root)
{
	struct btrfs_trans_handle *trans;
	int ret;
3435

Y
Yan Zheng 已提交
3436
	mutex_lock(&root->fs_info->cleaner_mutex);
Y
Yan, Zheng 已提交
3437
	btrfs_run_delayed_iputs(root);
Y
Yan Zheng 已提交
3438
	mutex_unlock(&root->fs_info->cleaner_mutex);
D
David Sterba 已提交
3439
	wake_up_process(root->fs_info->cleaner_kthread);
3440 3441 3442 3443 3444

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

3445
	trans = btrfs_join_transaction(root);
3446 3447
	if (IS_ERR(trans))
		return PTR_ERR(trans);
3448
	ret = btrfs_commit_transaction(trans, root);
3449 3450
	if (ret)
		return ret;
Y
Yan Zheng 已提交
3451
	/* run commit again to drop the original snapshot */
3452
	trans = btrfs_join_transaction(root);
3453 3454
	if (IS_ERR(trans))
		return PTR_ERR(trans);
3455 3456 3457
	ret = btrfs_commit_transaction(trans, root);
	if (ret)
		return ret;
C
Chris Mason 已提交
3458
	ret = btrfs_write_and_wait_transaction(NULL, root);
3459 3460 3461 3462 3463
	if (ret) {
		btrfs_error(root->fs_info, ret,
			    "Failed to sync btree inode to disk.");
		return ret;
	}
3464

Y
Yan Zheng 已提交
3465
	ret = write_ctree_super(NULL, root, 0);
Y
Yan Zheng 已提交
3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476
	return ret;
}

int close_ctree(struct btrfs_root *root)
{
	struct btrfs_fs_info *fs_info = root->fs_info;
	int ret;

	fs_info->closing = 1;
	smp_mb();

3477
	/* pause restriper - we want to resume on mount */
3478
	btrfs_pause_balance(fs_info);
3479

3480 3481
	btrfs_dev_replace_suspend_for_unmount(fs_info);

3482
	btrfs_scrub_cancel(fs_info);
C
Chris Mason 已提交
3483 3484 3485 3486 3487 3488

	/* 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 */
3489
	btrfs_cleanup_defrag_inodes(fs_info);
C
Chris Mason 已提交
3490

Y
Yan Zheng 已提交
3491
	if (!(fs_info->sb->s_flags & MS_RDONLY)) {
L
liubo 已提交
3492 3493 3494 3495 3496
		ret = btrfs_commit_super(root);
		if (ret)
			printk(KERN_ERR "btrfs: commit super ret %d\n", ret);
	}

3497
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state))
3498
		btrfs_error_commit_super(root);
3499

3500 3501
	btrfs_put_block_group_cache(fs_info);

A
Al Viro 已提交
3502 3503
	kthread_stop(fs_info->transaction_kthread);
	kthread_stop(fs_info->cleaner_kthread);
3504

3505 3506 3507
	fs_info->closing = 2;
	smp_mb();

3508 3509
	btrfs_free_qgroup_config(root->fs_info);

3510 3511 3512
	if (percpu_counter_sum(&fs_info->delalloc_bytes)) {
		printk(KERN_INFO "btrfs: at unmount delalloc count %lld\n",
		       percpu_counter_sum(&fs_info->delalloc_bytes));
C
Chris Mason 已提交
3513
	}
3514

L
Liu Bo 已提交
3515
	free_root_pointers(fs_info, 1);
3516

A
Al Viro 已提交
3517
	btrfs_free_block_groups(fs_info);
3518

Y
Yan Zheng 已提交
3519
	del_fs_roots(fs_info);
3520

Y
Yan Zheng 已提交
3521
	iput(fs_info->btree_inode);
3522

L
Liu Bo 已提交
3523
	btrfs_stop_all_workers(fs_info);
3524

3525 3526 3527 3528 3529
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	if (btrfs_test_opt(root, CHECK_INTEGRITY))
		btrfsic_unmount(root, fs_info->fs_devices);
#endif

3530
	btrfs_close_devices(fs_info->fs_devices);
3531
	btrfs_mapping_tree_free(&fs_info->mapping_tree);
3532

3533
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
3534
	percpu_counter_destroy(&fs_info->delalloc_bytes);
C
Chris Mason 已提交
3535
	bdi_destroy(&fs_info->bdi);
3536
	cleanup_srcu_struct(&fs_info->subvol_srcu);
3537

D
David Woodhouse 已提交
3538 3539
	btrfs_free_stripe_hash_table(fs_info);

3540 3541 3542
	return 0;
}

3543 3544
int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid,
			  int atomic)
3545
{
3546
	int ret;
3547
	struct inode *btree_inode = buf->pages[0]->mapping->host;
3548

3549
	ret = extent_buffer_uptodate(buf);
3550 3551 3552 3553
	if (!ret)
		return ret;

	ret = verify_parent_transid(&BTRFS_I(btree_inode)->io_tree, buf,
3554 3555 3556
				    parent_transid, atomic);
	if (ret == -EAGAIN)
		return ret;
3557
	return !ret;
3558 3559 3560
}

int btrfs_set_buffer_uptodate(struct extent_buffer *buf)
C
Chris Mason 已提交
3561
{
3562
	return set_extent_buffer_uptodate(buf);
3563
}
3564

3565 3566
void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
{
3567
	struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root;
3568
	u64 transid = btrfs_header_generation(buf);
3569
	int was_dirty;
3570

3571
	btrfs_assert_tree_locked(buf);
J
Julia Lawall 已提交
3572 3573
	if (transid != root->fs_info->generation)
		WARN(1, KERN_CRIT "btrfs transid mismatch buffer %llu, "
C
Chris Mason 已提交
3574
		       "found %llu running %llu\n",
3575
			(unsigned long long)buf->start,
C
Chris Mason 已提交
3576 3577
			(unsigned long long)transid,
			(unsigned long long)root->fs_info->generation);
3578
	was_dirty = set_extent_buffer_dirty(buf);
3579 3580 3581 3582
	if (!was_dirty)
		__percpu_counter_add(&root->fs_info->dirty_metadata_bytes,
				     buf->len,
				     root->fs_info->dirty_metadata_batch);
3583 3584
}

3585 3586
static void __btrfs_btree_balance_dirty(struct btrfs_root *root,
					int flush_delayed)
3587 3588 3589 3590 3591
{
	/*
	 * looks as though older kernels can get into trouble with
	 * this code, they end up stuck in balance_dirty_pages forever
	 */
3592
	int ret;
3593 3594 3595 3596

	if (current->flags & PF_MEMALLOC)
		return;

3597 3598
	if (flush_delayed)
		btrfs_balance_delayed_items(root);
3599

3600 3601 3602
	ret = percpu_counter_compare(&root->fs_info->dirty_metadata_bytes,
				     BTRFS_DIRTY_METADATA_THRESH);
	if (ret > 0) {
3603 3604
		balance_dirty_pages_ratelimited(
				   root->fs_info->btree_inode->i_mapping);
3605 3606 3607 3608
	}
	return;
}

3609
void btrfs_btree_balance_dirty(struct btrfs_root *root)
C
Chris Mason 已提交
3610
{
3611 3612
	__btrfs_btree_balance_dirty(root, 1);
}
3613

3614 3615 3616
void btrfs_btree_balance_dirty_nodelay(struct btrfs_root *root)
{
	__btrfs_btree_balance_dirty(root, 0);
C
Chris Mason 已提交
3617
}
3618

3619
int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid)
3620
{
3621
	struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root;
3622
	return btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
3623
}
3624

3625
static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info,
L
liubo 已提交
3626 3627
			      int read_only)
{
D
David Sterba 已提交
3628 3629 3630
	/*
	 * Placeholder for checks
	 */
3631
	return 0;
L
liubo 已提交
3632 3633
}

3634
static void btrfs_error_commit_super(struct btrfs_root *root)
L
liubo 已提交
3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646
{
	mutex_lock(&root->fs_info->cleaner_mutex);
	btrfs_run_delayed_iputs(root);
	mutex_unlock(&root->fs_info->cleaner_mutex);

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

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

3647 3648
static void btrfs_destroy_ordered_operations(struct btrfs_transaction *t,
					     struct btrfs_root *root)
L
liubo 已提交
3649 3650 3651 3652 3653 3654 3655 3656 3657
{
	struct btrfs_inode *btrfs_inode;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

	mutex_lock(&root->fs_info->ordered_operations_mutex);
	spin_lock(&root->fs_info->ordered_extent_lock);

3658
	list_splice_init(&t->ordered_operations, &splice);
L
liubo 已提交
3659 3660 3661 3662 3663
	while (!list_empty(&splice)) {
		btrfs_inode = list_entry(splice.next, struct btrfs_inode,
					 ordered_operations);

		list_del_init(&btrfs_inode->ordered_operations);
3664
		spin_unlock(&root->fs_info->ordered_extent_lock);
L
liubo 已提交
3665 3666

		btrfs_invalidate_inodes(btrfs_inode->root);
3667 3668

		spin_lock(&root->fs_info->ordered_extent_lock);
L
liubo 已提交
3669 3670 3671 3672 3673 3674
	}

	spin_unlock(&root->fs_info->ordered_extent_lock);
	mutex_unlock(&root->fs_info->ordered_operations_mutex);
}

3675
static void btrfs_destroy_ordered_extents(struct btrfs_root *root)
L
liubo 已提交
3676 3677 3678 3679
{
	struct btrfs_ordered_extent *ordered;

	spin_lock(&root->fs_info->ordered_extent_lock);
3680 3681 3682 3683 3684 3685 3686
	/*
	 * This will just short circuit the ordered completion stuff which will
	 * make sure the ordered extent gets properly cleaned up.
	 */
	list_for_each_entry(ordered, &root->fs_info->ordered_extents,
			    root_extent_list)
		set_bit(BTRFS_ORDERED_IOERR, &ordered->flags);
L
liubo 已提交
3687 3688 3689
	spin_unlock(&root->fs_info->ordered_extent_lock);
}

3690 3691
int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
			       struct btrfs_root *root)
L
liubo 已提交
3692 3693 3694 3695 3696 3697 3698 3699 3700 3701
{
	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);
	if (delayed_refs->num_entries == 0) {
3702
		spin_unlock(&delayed_refs->lock);
L
liubo 已提交
3703 3704 3705 3706
		printk(KERN_INFO "delayed_refs has NO entry\n");
		return ret;
	}

3707
	while ((node = rb_first(&delayed_refs->root)) != NULL) {
3708
		struct btrfs_delayed_ref_head *head = NULL;
L
liubo 已提交
3709

3710
		ref = rb_entry(node, struct btrfs_delayed_ref_node, rb_node);
L
liubo 已提交
3711 3712 3713 3714
		atomic_set(&ref->refs, 1);
		if (btrfs_delayed_ref_is_head(ref)) {

			head = btrfs_delayed_node_to_head(ref);
3715 3716 3717 3718 3719 3720 3721 3722 3723
			if (!mutex_trylock(&head->mutex)) {
				atomic_inc(&ref->refs);
				spin_unlock(&delayed_refs->lock);

				/* Need to wait for the delayed ref to run */
				mutex_lock(&head->mutex);
				mutex_unlock(&head->mutex);
				btrfs_put_delayed_ref(ref);

J
Josef Bacik 已提交
3724
				spin_lock(&delayed_refs->lock);
3725 3726 3727
				continue;
			}

J
Josef Bacik 已提交
3728 3729 3730
			if (head->must_insert_reserved)
				btrfs_pin_extent(root, ref->bytenr,
						 ref->num_bytes, 1);
3731
			btrfs_free_delayed_extent_op(head->extent_op);
L
liubo 已提交
3732 3733 3734 3735 3736
			delayed_refs->num_heads--;
			if (list_empty(&head->cluster))
				delayed_refs->num_heads_ready--;
			list_del_init(&head->cluster);
		}
3737

3738 3739 3740
		ref->in_tree = 0;
		rb_erase(&ref->rb_node, &delayed_refs->root);
		delayed_refs->num_entries--;
3741 3742
		if (head)
			mutex_unlock(&head->mutex);
L
liubo 已提交
3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754
		spin_unlock(&delayed_refs->lock);
		btrfs_put_delayed_ref(ref);

		cond_resched();
		spin_lock(&delayed_refs->lock);
	}

	spin_unlock(&delayed_refs->lock);

	return ret;
}

3755
static void btrfs_evict_pending_snapshots(struct btrfs_transaction *t)
L
liubo 已提交
3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767
{
	struct btrfs_pending_snapshot *snapshot;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

	list_splice_init(&t->pending_snapshots, &splice);

	while (!list_empty(&splice)) {
		snapshot = list_entry(splice.next,
				      struct btrfs_pending_snapshot,
				      list);
3768
		snapshot->error = -ECANCELED;
L
liubo 已提交
3769 3770 3771 3772
		list_del_init(&snapshot->list);
	}
}

3773
static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root)
L
liubo 已提交
3774 3775 3776 3777 3778 3779 3780
{
	struct btrfs_inode *btrfs_inode;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

	spin_lock(&root->fs_info->delalloc_lock);
3781
	list_splice_init(&root->fs_info->delalloc_inodes, &splice);
L
liubo 已提交
3782 3783 3784 3785 3786 3787

	while (!list_empty(&splice)) {
		btrfs_inode = list_entry(splice.next, struct btrfs_inode,
				    delalloc_inodes);

		list_del_init(&btrfs_inode->delalloc_inodes);
3788 3789
		clear_bit(BTRFS_INODE_IN_DELALLOC_LIST,
			  &btrfs_inode->runtime_flags);
3790
		spin_unlock(&root->fs_info->delalloc_lock);
L
liubo 已提交
3791 3792

		btrfs_invalidate_inodes(btrfs_inode->root);
3793 3794

		spin_lock(&root->fs_info->delalloc_lock);
L
liubo 已提交
3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810
	}

	spin_unlock(&root->fs_info->delalloc_lock);
}

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

	while (1) {
		ret = find_first_extent_bit(dirty_pages, start, &start, &end,
3811
					    mark, NULL);
L
liubo 已提交
3812 3813 3814 3815 3816
		if (ret)
			break;

		clear_extent_bits(dirty_pages, start, end, mark, GFP_NOFS);
		while (start <= end) {
3817 3818
			eb = btrfs_find_tree_block(root, start,
						   root->leafsize);
3819
			start += root->leafsize;
3820
			if (!eb)
L
liubo 已提交
3821
				continue;
3822
			wait_on_extent_buffer_writeback(eb);
L
liubo 已提交
3823

3824 3825 3826 3827
			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
3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839 3840
		}
	}

	return ret;
}

static int btrfs_destroy_pinned_extent(struct btrfs_root *root,
				       struct extent_io_tree *pinned_extents)
{
	struct extent_io_tree *unpin;
	u64 start;
	u64 end;
	int ret;
3841
	bool loop = true;
L
liubo 已提交
3842 3843

	unpin = pinned_extents;
3844
again:
L
liubo 已提交
3845 3846
	while (1) {
		ret = find_first_extent_bit(unpin, 0, &start, &end,
3847
					    EXTENT_DIRTY, NULL);
L
liubo 已提交
3848 3849 3850 3851
		if (ret)
			break;

		/* opt_discard */
3852 3853 3854 3855
		if (btrfs_test_opt(root, DISCARD))
			ret = btrfs_error_discard_extent(root, start,
							 end + 1 - start,
							 NULL);
L
liubo 已提交
3856 3857 3858 3859 3860 3861

		clear_extent_dirty(unpin, start, end, GFP_NOFS);
		btrfs_error_unpin_extent_range(root, start, end);
		cond_resched();
	}

3862 3863 3864 3865 3866 3867 3868 3869 3870
	if (loop) {
		if (unpin == &root->fs_info->freed_extents[0])
			unpin = &root->fs_info->freed_extents[1];
		else
			unpin = &root->fs_info->freed_extents[0];
		loop = false;
		goto again;
	}

L
liubo 已提交
3871 3872 3873
	return 0;
}

3874 3875 3876 3877 3878 3879 3880 3881 3882 3883
void btrfs_cleanup_one_transaction(struct btrfs_transaction *cur_trans,
				   struct btrfs_root *root)
{
	btrfs_destroy_delayed_refs(cur_trans, root);
	btrfs_block_rsv_release(root, &root->fs_info->trans_block_rsv,
				cur_trans->dirty_pages.dirty_bytes);

	/* FIXME: cleanup wait for commit */
	cur_trans->in_commit = 1;
	cur_trans->blocked = 1;
3884
	wake_up(&root->fs_info->transaction_blocked_wait);
3885

3886 3887
	btrfs_evict_pending_snapshots(cur_trans);

3888
	cur_trans->blocked = 0;
3889
	wake_up(&root->fs_info->transaction_wait);
3890 3891

	cur_trans->commit_done = 1;
3892
	wake_up(&cur_trans->commit_wait);
3893

3894 3895
	btrfs_destroy_delayed_inodes(root);
	btrfs_assert_delayed_root_empty(root);
3896 3897 3898

	btrfs_destroy_marked_extents(root, &cur_trans->dirty_pages,
				     EXTENT_DIRTY);
3899 3900
	btrfs_destroy_pinned_extent(root,
				    root->fs_info->pinned_extents);
3901 3902 3903 3904 3905 3906 3907

	/*
	memset(cur_trans, 0, sizeof(*cur_trans));
	kmem_cache_free(btrfs_transaction_cachep, cur_trans);
	*/
}

3908
static int btrfs_cleanup_transaction(struct btrfs_root *root)
L
liubo 已提交
3909 3910 3911 3912 3913 3914
{
	struct btrfs_transaction *t;
	LIST_HEAD(list);

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

J
Josef Bacik 已提交
3915
	spin_lock(&root->fs_info->trans_lock);
L
liubo 已提交
3916
	list_splice_init(&root->fs_info->trans_list, &list);
J
Josef Bacik 已提交
3917 3918 3919
	root->fs_info->trans_no_join = 1;
	spin_unlock(&root->fs_info->trans_lock);

L
liubo 已提交
3920 3921 3922
	while (!list_empty(&list)) {
		t = list_entry(list.next, struct btrfs_transaction, list);

3923
		btrfs_destroy_ordered_operations(t, root);
L
liubo 已提交
3924 3925 3926 3927 3928 3929 3930 3931

		btrfs_destroy_ordered_extents(root);

		btrfs_destroy_delayed_refs(t, root);

		/* FIXME: cleanup wait for commit */
		t->in_commit = 1;
		t->blocked = 1;
3932
		smp_mb();
L
liubo 已提交
3933 3934 3935
		if (waitqueue_active(&root->fs_info->transaction_blocked_wait))
			wake_up(&root->fs_info->transaction_blocked_wait);

3936 3937
		btrfs_evict_pending_snapshots(t);

L
liubo 已提交
3938
		t->blocked = 0;
3939
		smp_mb();
L
liubo 已提交
3940 3941 3942 3943
		if (waitqueue_active(&root->fs_info->transaction_wait))
			wake_up(&root->fs_info->transaction_wait);

		t->commit_done = 1;
3944
		smp_mb();
L
liubo 已提交
3945 3946 3947
		if (waitqueue_active(&t->commit_wait))
			wake_up(&t->commit_wait);

3948 3949 3950
		btrfs_destroy_delayed_inodes(root);
		btrfs_assert_delayed_root_empty(root);

L
liubo 已提交
3951 3952
		btrfs_destroy_delalloc_inodes(root);

J
Josef Bacik 已提交
3953
		spin_lock(&root->fs_info->trans_lock);
L
liubo 已提交
3954
		root->fs_info->running_transaction = NULL;
J
Josef Bacik 已提交
3955
		spin_unlock(&root->fs_info->trans_lock);
L
liubo 已提交
3956 3957 3958 3959 3960 3961 3962

		btrfs_destroy_marked_extents(root, &t->dirty_pages,
					     EXTENT_DIRTY);

		btrfs_destroy_pinned_extent(root,
					    root->fs_info->pinned_extents);

3963
		atomic_set(&t->use_count, 0);
L
liubo 已提交
3964 3965 3966 3967 3968
		list_del_init(&t->list);
		memset(t, 0, sizeof(*t));
		kmem_cache_free(btrfs_transaction_cachep, t);
	}

J
Josef Bacik 已提交
3969 3970 3971
	spin_lock(&root->fs_info->trans_lock);
	root->fs_info->trans_no_join = 0;
	spin_unlock(&root->fs_info->trans_lock);
L
liubo 已提交
3972 3973 3974 3975 3976
	mutex_unlock(&root->fs_info->transaction_kthread_mutex);

	return 0;
}

3977
static struct extent_io_ops btree_extent_io_ops = {
3978
	.readpage_end_io_hook = btree_readpage_end_io_hook,
A
Arne Jansen 已提交
3979
	.readpage_io_failed_hook = btree_io_failed_hook,
3980
	.submit_bio_hook = btree_submit_bio_hook,
3981 3982
	/* note we're sharing with inode.c for the merge bio hook */
	.merge_bio_hook = btrfs_merge_bio_hook,
3983
};