disk-io.c 110.9 KB
Newer Older
C
Chris Mason 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
/*
 * Copyright (C) 2007 Oracle.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License v2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */

C
Chris Mason 已提交
19
#include <linux/fs.h>
20
#include <linux/blkdev.h>
C
Chris Mason 已提交
21
#include <linux/scatterlist.h>
C
Chris Mason 已提交
22
#include <linux/swap.h>
23
#include <linux/radix-tree.h>
C
Chris Mason 已提交
24
#include <linux/writeback.h>
C
Chris Mason 已提交
25
#include <linux/buffer_head.h>
26
#include <linux/workqueue.h>
27
#include <linux/kthread.h>
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>
S
Stefan Behrens 已提交
34
#include <linux/semaphore.h>
35
#include <asm/unaligned.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_destroy_delalloc_inodes(struct btrfs_root *root);
L
liubo 已提交
67 68 69 70 71
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);
72 73
static int btrfs_cleanup_transaction(struct btrfs_root *root);
static void btrfs_error_commit_super(struct btrfs_root *root);
74

C
Chris Mason 已提交
75 76 77 78 79
/*
 * 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.
 */
80 81 82 83 84 85
struct end_io_wq {
	struct bio *bio;
	bio_end_io_t *end_io;
	void *private;
	struct btrfs_fs_info *info;
	int error;
86
	int metadata;
87
	struct list_head list;
88
	struct btrfs_work work;
89
};
90

C
Chris Mason 已提交
91 92 93 94 95
/*
 * 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.
 */
96 97 98 99
struct async_submit_bio {
	struct inode *inode;
	struct bio *bio;
	struct list_head list;
C
Chris Mason 已提交
100 101
	extent_submit_bio_hook_t *submit_bio_start;
	extent_submit_bio_hook_t *submit_bio_done;
102 103
	int rw;
	int mirror_num;
C
Chris Mason 已提交
104
	unsigned long bio_flags;
105 106 107 108 109
	/*
	 * 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;
110
	struct btrfs_work work;
111
	int error;
112 113
};

114 115 116 117 118 119 120 121 122 123
/*
 * 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.
124
 *
125 126 127
 * 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.
128
 *
129 130 131
 * 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.
132
 *
133 134 135
 * 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.
136 137 138 139 140
 */
#ifdef CONFIG_DEBUG_LOCK_ALLOC
# if BTRFS_MAX_LEVEL != 8
#  error
# endif
141 142 143 144 145 146 147 148 149 150 151 152 153

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"	},
154
	{ .id = BTRFS_QUOTA_TREE_OBJECTID,	.name_stem = "quota"	},
155 156 157
	{ .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"	},
158
	{ .id = BTRFS_UUID_TREE_OBJECTID,	.name_stem = "uuid"	},
159
	{ .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
				       "failed on %llu wanted %X found %X "
				       "level %d\n",
305 306
				       root->fs_info->sb->s_id, buf->start,
				       val, found, btrfs_header_level(buf));
307 308
			if (result != (char *)&inline_result)
				kfree(result);
309 310 311
			return 1;
		}
	} else {
312
		write_extent_buffer(buf, result, 0, csum_size);
313
	}
314 315
	if (result != (char *)&inline_result)
		kfree(result);
316 317 318
	return 0;
}

C
Chris Mason 已提交
319 320 321 322 323 324
/*
 * 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.
 */
325
static int verify_parent_transid(struct extent_io_tree *io_tree,
326 327
				 struct extent_buffer *eb, u64 parent_transid,
				 int atomic)
328
{
329
	struct extent_state *cached_state = NULL;
330 331 332 333 334
	int ret;

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

335 336 337
	if (atomic)
		return -EAGAIN;

338
	lock_extent_bits(io_tree, eb->start, eb->start + eb->len - 1,
339
			 0, &cached_state);
340
	if (extent_buffer_uptodate(eb) &&
341 342 343 344
	    btrfs_header_generation(eb) == parent_transid) {
		ret = 0;
		goto out;
	}
345
	printk_ratelimited("parent transid verify failed on %llu wanted %llu "
C
Chris Mason 已提交
346
		       "found %llu\n",
347
		       eb->start, parent_transid, btrfs_header_generation(eb));
348
	ret = 1;
349
	clear_extent_buffer_uptodate(eb);
C
Chris Mason 已提交
350
out:
351 352
	unlock_extent_cached(io_tree, eb->start, eb->start + eb->len - 1,
			     &cached_state, GFP_NOFS);
353 354 355
	return ret;
}

D
David Sterba 已提交
356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382
/*
 * 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;
383 384 385 386 387

		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
388 389 390 391 392 393 394 395 396 397 398
	}

	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
399 400 401 402
/*
 * helper to read a given tree block, doing retries as required when
 * the checksums don't match and we have alternate mirrors to try.
 */
403 404
static int btree_read_extent_buffer_pages(struct btrfs_root *root,
					  struct extent_buffer *eb,
405
					  u64 start, u64 parent_transid)
406 407
{
	struct extent_io_tree *io_tree;
408
	int failed = 0;
409 410 411
	int ret;
	int num_copies = 0;
	int mirror_num = 0;
412
	int failed_mirror = 0;
413

414
	clear_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
415 416
	io_tree = &BTRFS_I(root->fs_info->btree_inode)->io_tree;
	while (1) {
417 418
		ret = read_extent_buffer_pages(io_tree, eb, start,
					       WAIT_COMPLETE,
419
					       btree_get_extent, mirror_num);
420 421
		if (!ret) {
			if (!verify_parent_transid(io_tree, eb,
422
						   parent_transid, 0))
423 424 425 426
				break;
			else
				ret = -EIO;
		}
C
Chris Mason 已提交
427

428 429 430 431 432 433
		/*
		 * 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))
434 435
			break;

436
		num_copies = btrfs_num_copies(root->fs_info,
437
					      eb->start, eb->len);
C
Chris Mason 已提交
438
		if (num_copies == 1)
439
			break;
C
Chris Mason 已提交
440

441 442 443 444 445
		if (!failed_mirror) {
			failed = 1;
			failed_mirror = eb->read_mirror;
		}

446
		mirror_num++;
447 448 449
		if (mirror_num == failed_mirror)
			mirror_num++;

C
Chris Mason 已提交
450
		if (mirror_num > num_copies)
451
			break;
452
	}
453

454
	if (failed && !ret && failed_mirror)
455 456 457
		repair_eb_io_failure(root, eb, failed_mirror);

	return ret;
458
}
459

C
Chris Mason 已提交
460
/*
C
Chris Mason 已提交
461 462
 * 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
463
 */
C
Chris Mason 已提交
464

465
static int csum_dirty_buffer(struct btrfs_root *root, struct page *page)
466
{
467
	struct extent_io_tree *tree;
M
Miao Xie 已提交
468
	u64 start = page_offset(page);
469 470
	u64 found_start;
	struct extent_buffer *eb;
471

472
	tree = &BTRFS_I(page->mapping->host)->io_tree;
473

J
Josef Bacik 已提交
474 475 476
	eb = (struct extent_buffer *)page->private;
	if (page != eb->pages[0])
		return 0;
477
	found_start = btrfs_header_bytenr(eb);
478
	if (WARN_ON(found_start != start || !PageUptodate(page)))
J
Josef Bacik 已提交
479
		return 0;
480 481 482 483
	csum_tree_block(root, eb, 0);
	return 0;
}

Y
Yan Zheng 已提交
484 485 486 487 488 489 490
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;

491
	read_extent_buffer(eb, fsid, btrfs_header_fsid(), BTRFS_FSID_SIZE);
Y
Yan Zheng 已提交
492 493 494 495 496 497 498 499 500 501
	while (fs_devices) {
		if (!memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE)) {
			ret = 0;
			break;
		}
		fs_devices = fs_devices->seed;
	}
	return ret;
}

502 503 504
#define CORRUPT(reason, eb, root, slot)				\
	printk(KERN_CRIT "btrfs: corrupt leaf, %s: block=%llu,"	\
	       "root=%llu, slot=%d\n", reason,			\
505
	       btrfs_header_bytenr(eb),	root->objectid, slot)
506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567

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;
}

568 569 570
static int btree_readpage_end_io_hook(struct btrfs_io_bio *io_bio,
				      u64 phy_offset, struct page *page,
				      u64 start, u64 end, int mirror)
571 572 573 574 575 576
{
	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;
577
	int ret = 0;
578
	int reads_done;
579 580 581

	if (!page->private)
		goto out;
C
Chris Mason 已提交
582

583
	tree = &BTRFS_I(page->mapping->host)->io_tree;
J
Josef Bacik 已提交
584
	eb = (struct extent_buffer *)page->private;
C
Chris Mason 已提交
585

586 587 588 589 590 591
	/* 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);
592 593
	if (!reads_done)
		goto err;
594

595
	eb->read_mirror = mirror;
596 597 598 599 600
	if (test_bit(EXTENT_BUFFER_IOERR, &eb->bflags)) {
		ret = -EIO;
		goto err;
	}

601
	found_start = btrfs_header_bytenr(eb);
602
	if (found_start != eb->start) {
603
		printk_ratelimited(KERN_INFO "btrfs bad tree block start "
C
Chris Mason 已提交
604
			       "%llu %llu\n",
605
			       found_start, eb->start);
606
		ret = -EIO;
607 608
		goto err;
	}
Y
Yan Zheng 已提交
609
	if (check_tree_block_fsid(root, eb)) {
610
		printk_ratelimited(KERN_INFO "btrfs bad fsid on block %llu\n",
611
			       eb->start);
612 613 614
		ret = -EIO;
		goto err;
	}
615
	found_level = btrfs_header_level(eb);
616 617 618 619 620 621
	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;
	}
622

623 624
	btrfs_set_buffer_lockdep_class(btrfs_header_owner(eb),
				       eb, found_level);
625

626
	ret = csum_tree_block(root, eb, 1);
627
	if (ret) {
628
		ret = -EIO;
629 630 631 632 633 634 635 636 637 638 639 640
		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;
	}
641

642 643
	if (!ret)
		set_extent_buffer_uptodate(eb);
644
err:
645 646
	if (reads_done &&
	    test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
A
Arne Jansen 已提交
647 648
		btree_readahead_hook(root, eb, eb->start, ret);

D
David Woodhouse 已提交
649 650 651 652 653 654 655
	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);
656
		clear_extent_buffer_uptodate(eb);
D
David Woodhouse 已提交
657
	}
658
	free_extent_buffer(eb);
659
out:
660
	return ret;
661 662
}

663
static int btree_io_failed_hook(struct page *page, int failed_mirror)
A
Arne Jansen 已提交
664 665 666 667
{
	struct extent_buffer *eb;
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;

J
Josef Bacik 已提交
668
	eb = (struct extent_buffer *)page->private;
669
	set_bit(EXTENT_BUFFER_IOERR, &eb->bflags);
670
	eb->read_mirror = failed_mirror;
D
David Woodhouse 已提交
671
	atomic_dec(&eb->io_pages);
672
	if (test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
A
Arne Jansen 已提交
673 674 675 676
		btree_readahead_hook(root, eb, eb->start, -EIO);
	return -EIO;	/* we fixed nothing */
}

677 678 679 680 681 682 683
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;
684 685
	end_io_wq->work.func = end_workqueue_fn;
	end_io_wq->work.flags = 0;
686

687
	if (bio->bi_rw & REQ_WRITE) {
D
David Woodhouse 已提交
688
		if (end_io_wq->metadata == BTRFS_WQ_ENDIO_METADATA)
689 690
			btrfs_queue_worker(&fs_info->endio_meta_write_workers,
					   &end_io_wq->work);
D
David Woodhouse 已提交
691
		else if (end_io_wq->metadata == BTRFS_WQ_ENDIO_FREE_SPACE)
J
Josef Bacik 已提交
692 693
			btrfs_queue_worker(&fs_info->endio_freespace_worker,
					   &end_io_wq->work);
D
David Woodhouse 已提交
694 695 696
		else if (end_io_wq->metadata == BTRFS_WQ_ENDIO_RAID56)
			btrfs_queue_worker(&fs_info->endio_raid56_workers,
					   &end_io_wq->work);
697 698 699
		else
			btrfs_queue_worker(&fs_info->endio_write_workers,
					   &end_io_wq->work);
700
	} else {
D
David Woodhouse 已提交
701 702 703 704
		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)
705 706 707 708 709 710
			btrfs_queue_worker(&fs_info->endio_meta_workers,
					   &end_io_wq->work);
		else
			btrfs_queue_worker(&fs_info->endio_workers,
					   &end_io_wq->work);
	}
711 712
}

J
Josef Bacik 已提交
713 714 715 716 717 718
/*
 * 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
719
 * 3 - raid parity work
J
Josef Bacik 已提交
720
 */
721 722
int btrfs_bio_wq_end_io(struct btrfs_fs_info *info, struct bio *bio,
			int metadata)
723
{
724 725 726 727 728 729 730
	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;
731
	end_io_wq->info = info;
732 733
	end_io_wq->error = 0;
	end_io_wq->bio = bio;
734
	end_io_wq->metadata = metadata;
735 736 737

	bio->bi_private = end_io_wq;
	bio->bi_end_io = end_workqueue_bio;
738 739 740
	return 0;
}

741
unsigned long btrfs_async_submit_limit(struct btrfs_fs_info *info)
742
{
743 744 745 746 747
	unsigned long limit = min_t(unsigned long,
				    info->workers.max_workers,
				    info->fs_devices->open_devices);
	return 256 * limit;
}
748

C
Chris Mason 已提交
749 750 751
static void run_one_async_start(struct btrfs_work *work)
{
	struct async_submit_bio *async;
752
	int ret;
C
Chris Mason 已提交
753 754

	async = container_of(work, struct  async_submit_bio, work);
755 756 757 758 759
	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
760 761 762
}

static void run_one_async_done(struct btrfs_work *work)
763 764 765
{
	struct btrfs_fs_info *fs_info;
	struct async_submit_bio *async;
766
	int limit;
767 768 769

	async = container_of(work, struct  async_submit_bio, work);
	fs_info = BTRFS_I(async->inode)->root->fs_info;
770

771
	limit = btrfs_async_submit_limit(fs_info);
772 773
	limit = limit * 2 / 3;

774
	if (atomic_dec_return(&fs_info->nr_async_submits) < limit &&
775
	    waitqueue_active(&fs_info->async_submit_wait))
776 777
		wake_up(&fs_info->async_submit_wait);

778 779 780 781 782 783
	/* 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
784
	async->submit_bio_done(async->inode, async->rw, async->bio,
785 786
			       async->mirror_num, async->bio_flags,
			       async->bio_offset);
C
Chris Mason 已提交
787 788 789 790 791 792 793
}

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

	async = container_of(work, struct  async_submit_bio, work);
794 795 796
	kfree(async);
}

797 798
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
799
			unsigned long bio_flags,
800
			u64 bio_offset,
C
Chris Mason 已提交
801 802
			extent_submit_bio_hook_t *submit_bio_start,
			extent_submit_bio_hook_t *submit_bio_done)
803 804 805 806 807 808 809 810 811 812 813
{
	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
814 815 816 817 818 819 820
	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;

821
	async->work.flags = 0;
C
Chris Mason 已提交
822
	async->bio_flags = bio_flags;
823
	async->bio_offset = bio_offset;
824

825 826
	async->error = 0;

827
	atomic_inc(&fs_info->nr_async_submits);
828

829
	if (rw & REQ_SYNC)
830 831
		btrfs_set_work_high_prio(&async->work);

832
	btrfs_queue_worker(&fs_info->workers, &async->work);
833

C
Chris Mason 已提交
834
	while (atomic_read(&fs_info->async_submit_draining) &&
835 836 837 838 839
	      atomic_read(&fs_info->nr_async_submits)) {
		wait_event(fs_info->async_submit_wait,
			   (atomic_read(&fs_info->nr_async_submits) == 0));
	}

840 841 842
	return 0;
}

843 844 845 846 847
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;
848
	int ret = 0;
849 850

	WARN_ON(bio->bi_vcnt <= 0);
C
Chris Mason 已提交
851
	while (bio_index < bio->bi_vcnt) {
852
		root = BTRFS_I(bvec->bv_page->mapping->host)->root;
853 854 855
		ret = csum_dirty_buffer(root, bvec->bv_page);
		if (ret)
			break;
856 857 858
		bio_index++;
		bvec++;
	}
859
	return ret;
860 861
}

C
Chris Mason 已提交
862 863
static int __btree_submit_bio_start(struct inode *inode, int rw,
				    struct bio *bio, int mirror_num,
864 865
				    unsigned long bio_flags,
				    u64 bio_offset)
866
{
867 868
	/*
	 * when we're called for a write, we're already in the async
869
	 * submission context.  Just jump into btrfs_map_bio
870
	 */
871
	return btree_csum_one_bio(bio);
C
Chris Mason 已提交
872
}
873

C
Chris Mason 已提交
874
static int __btree_submit_bio_done(struct inode *inode, int rw, struct bio *bio,
875 876
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
C
Chris Mason 已提交
877
{
878 879
	int ret;

880
	/*
C
Chris Mason 已提交
881 882
	 * when we're called for a write, we're already in the async
	 * submission context.  Just jump into btrfs_map_bio
883
	 */
884 885 886 887
	ret = btrfs_map_bio(BTRFS_I(inode)->root, rw, bio, mirror_num, 1);
	if (ret)
		bio_endio(bio, ret);
	return ret;
888 889
}

890 891 892 893 894 895 896 897 898 899 900
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;
}

901
static int btree_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
902 903
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
904
{
905
	int async = check_async_write(inode, bio_flags);
906 907
	int ret;

908
	if (!(rw & REQ_WRITE)) {
C
Chris Mason 已提交
909 910 911 912
		/*
		 * called for a read, do the setup so that checksum validation
		 * can happen in the async kernel threads
		 */
913 914
		ret = btrfs_bio_wq_end_io(BTRFS_I(inode)->root->fs_info,
					  bio, 1);
915
		if (ret)
916 917 918
			goto out_w_error;
		ret = btrfs_map_bio(BTRFS_I(inode)->root, rw, bio,
				    mirror_num, 0);
919 920 921
	} else if (!async) {
		ret = btree_csum_one_bio(bio);
		if (ret)
922 923 924 925 926 927 928 929 930 931 932 933 934
			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);
935
	}
936

937 938 939 940 941
	if (ret) {
out_w_error:
		bio_endio(bio, ret);
	}
	return ret;
942 943
}

J
Jan Beulich 已提交
944
#ifdef CONFIG_MIGRATION
945
static int btree_migratepage(struct address_space *mapping,
946 947
			struct page *newpage, struct page *page,
			enum migrate_mode mode)
948 949 950 951 952 953 954 955 956 957 958 959 960 961
{
	/*
	 * 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;
962
	return migrate_page(mapping, newpage, page, mode);
963
}
J
Jan Beulich 已提交
964
#endif
965

966 967 968 969

static int btree_writepages(struct address_space *mapping,
			    struct writeback_control *wbc)
{
970
	struct extent_io_tree *tree;
971 972 973
	struct btrfs_fs_info *fs_info;
	int ret;

974
	tree = &BTRFS_I(mapping->host)->io_tree;
975
	if (wbc->sync_mode == WB_SYNC_NONE) {
976 977 978 979

		if (wbc->for_kupdate)
			return 0;

980
		fs_info = BTRFS_I(mapping->host)->root->fs_info;
981
		/* this is a bit racy, but that's ok */
982 983 984
		ret = percpu_counter_compare(&fs_info->dirty_metadata_bytes,
					     BTRFS_DIRTY_METADATA_THRESH);
		if (ret < 0)
985 986
			return 0;
	}
987
	return btree_write_cache_pages(mapping, wbc);
988 989
}

990
static int btree_readpage(struct file *file, struct page *page)
991
{
992 993
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
994
	return extent_read_full_page(tree, page, btree_get_extent, 0);
995
}
C
Chris Mason 已提交
996

997
static int btree_releasepage(struct page *page, gfp_t gfp_flags)
998
{
999
	if (PageWriteback(page) || PageDirty(page))
C
Chris Mason 已提交
1000
		return 0;
1001

1002
	return try_release_extent_buffer(page);
1003 1004
}

1005 1006
static void btree_invalidatepage(struct page *page, unsigned int offset,
				 unsigned int length)
1007
{
1008 1009
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
1010 1011
	extent_invalidatepage(tree, page, offset);
	btree_releasepage(page, GFP_NOFS);
1012
	if (PagePrivate(page)) {
C
Chris Mason 已提交
1013 1014
		printk(KERN_WARNING "btrfs warning page private not zero "
		       "on page %llu\n", (unsigned long long)page_offset(page));
1015 1016 1017 1018
		ClearPagePrivate(page);
		set_page_private(page, 0);
		page_cache_release(page);
	}
1019 1020
}

1021 1022
static int btree_set_page_dirty(struct page *page)
{
1023
#ifdef DEBUG
1024 1025 1026 1027 1028 1029 1030 1031
	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);
1032
#endif
1033 1034 1035
	return __set_page_dirty_nobuffers(page);
}

1036
static const struct address_space_operations btree_aops = {
1037
	.readpage	= btree_readpage,
1038
	.writepages	= btree_writepages,
1039 1040
	.releasepage	= btree_releasepage,
	.invalidatepage = btree_invalidatepage,
1041
#ifdef CONFIG_MIGRATION
1042
	.migratepage	= btree_migratepage,
1043
#endif
1044
	.set_page_dirty = btree_set_page_dirty,
1045 1046
};

1047 1048
int readahead_tree_block(struct btrfs_root *root, u64 bytenr, u32 blocksize,
			 u64 parent_transid)
C
Chris Mason 已提交
1049
{
1050 1051
	struct extent_buffer *buf = NULL;
	struct inode *btree_inode = root->fs_info->btree_inode;
1052
	int ret = 0;
C
Chris Mason 已提交
1053

1054
	buf = btrfs_find_create_tree_block(root, bytenr, blocksize);
1055
	if (!buf)
C
Chris Mason 已提交
1056
		return 0;
1057
	read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree,
1058
				 buf, 0, WAIT_NONE, btree_get_extent, 0);
1059
	free_extent_buffer(buf);
1060
	return ret;
C
Chris Mason 已提交
1061 1062
}

1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086
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;
1087
	} else if (extent_buffer_uptodate(buf)) {
1088 1089 1090 1091 1092 1093 1094
		*eb = buf;
	} else {
		free_extent_buffer(buf);
	}
	return 0;
}

1095 1096 1097 1098 1099
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;
1100
	eb = find_extent_buffer(&BTRFS_I(btree_inode)->io_tree, bytenr);
1101 1102 1103 1104 1105 1106 1107 1108 1109 1110
	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,
1111
				 bytenr, blocksize);
1112 1113 1114 1115
	return eb;
}


1116 1117
int btrfs_write_tree_block(struct extent_buffer *buf)
{
1118
	return filemap_fdatawrite_range(buf->pages[0]->mapping, buf->start,
1119
					buf->start + buf->len - 1);
1120 1121 1122 1123
}

int btrfs_wait_tree_block_writeback(struct extent_buffer *buf)
{
1124
	return filemap_fdatawait_range(buf->pages[0]->mapping,
1125
				       buf->start, buf->start + buf->len - 1);
1126 1127
}

1128
struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr,
1129
				      u32 blocksize, u64 parent_transid)
1130 1131 1132 1133 1134 1135 1136 1137
{
	struct extent_buffer *buf = NULL;
	int ret;

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

1138
	ret = btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
1139 1140 1141 1142
	if (ret) {
		free_extent_buffer(buf);
		return NULL;
	}
1143
	return buf;
1144

1145 1146
}

1147 1148
void clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root,
		      struct extent_buffer *buf)
1149
{
1150 1151
	struct btrfs_fs_info *fs_info = root->fs_info;

1152
	if (btrfs_header_generation(buf) ==
1153
	    fs_info->running_transaction->transid) {
1154
		btrfs_assert_tree_locked(buf);
1155

1156
		if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &buf->bflags)) {
1157 1158 1159
			__percpu_counter_add(&fs_info->dirty_metadata_bytes,
					     -buf->len,
					     fs_info->dirty_metadata_batch);
1160 1161 1162 1163
			/* ugh, clear_extent_buffer_dirty needs to lock the page */
			btrfs_set_lock_blocking(buf);
			clear_extent_buffer_dirty(buf);
		}
1164
	}
1165 1166
}

1167 1168 1169 1170
static void __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
			 u32 stripesize, struct btrfs_root *root,
			 struct btrfs_fs_info *fs_info,
			 u64 objectid)
1171
{
C
Chris Mason 已提交
1172
	root->node = NULL;
1173
	root->commit_root = NULL;
1174 1175 1176
	root->sectorsize = sectorsize;
	root->nodesize = nodesize;
	root->leafsize = leafsize;
1177
	root->stripesize = stripesize;
C
Chris Mason 已提交
1178
	root->ref_cows = 0;
1179
	root->track_dirty = 0;
1180
	root->in_radix = 0;
1181 1182
	root->orphan_item_inserted = 0;
	root->orphan_cleanup_state = 0;
1183

1184 1185
	root->objectid = objectid;
	root->last_trans = 0;
1186
	root->highest_objectid = 0;
1187
	root->nr_delalloc_inodes = 0;
1188
	root->nr_ordered_extents = 0;
1189
	root->name = NULL;
1190
	root->inode_tree = RB_ROOT;
1191
	INIT_RADIX_TREE(&root->delayed_nodes_tree, GFP_ATOMIC);
1192
	root->block_rsv = NULL;
1193
	root->orphan_block_rsv = NULL;
1194 1195

	INIT_LIST_HEAD(&root->dirty_list);
1196
	INIT_LIST_HEAD(&root->root_list);
1197 1198
	INIT_LIST_HEAD(&root->delalloc_inodes);
	INIT_LIST_HEAD(&root->delalloc_root);
1199 1200
	INIT_LIST_HEAD(&root->ordered_extents);
	INIT_LIST_HEAD(&root->ordered_root);
1201 1202
	INIT_LIST_HEAD(&root->logged_list[0]);
	INIT_LIST_HEAD(&root->logged_list[1]);
1203
	spin_lock_init(&root->orphan_lock);
1204
	spin_lock_init(&root->inode_lock);
1205
	spin_lock_init(&root->delalloc_lock);
1206
	spin_lock_init(&root->ordered_extent_lock);
1207
	spin_lock_init(&root->accounting_lock);
1208 1209
	spin_lock_init(&root->log_extents_lock[0]);
	spin_lock_init(&root->log_extents_lock[1]);
1210
	mutex_init(&root->objectid_mutex);
1211
	mutex_init(&root->log_mutex);
Y
Yan Zheng 已提交
1212 1213 1214 1215 1216 1217
	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
1218
	atomic_set(&root->log_batch, 0);
1219
	atomic_set(&root->orphan_inodes, 0);
1220
	atomic_set(&root->refs, 1);
Y
Yan Zheng 已提交
1221
	root->log_transid = 0;
1222
	root->last_log_commit = 0;
1223 1224 1225
	if (fs_info)
		extent_io_tree_init(&root->dirty_log_pages,
				     fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
1226

1227 1228
	memset(&root->root_key, 0, sizeof(root->root_key));
	memset(&root->root_item, 0, sizeof(root->root_item));
1229
	memset(&root->defrag_progress, 0, sizeof(root->defrag_progress));
1230
	memset(&root->root_kobj, 0, sizeof(root->root_kobj));
1231 1232 1233 1234
	if (fs_info)
		root->defrag_trans_start = fs_info->generation;
	else
		root->defrag_trans_start = 0;
1235
	init_completion(&root->kobj_unregister);
1236
	root->defrag_running = 0;
1237
	root->root_key.objectid = objectid;
1238
	root->anon_dev = 0;
1239

1240
	spin_lock_init(&root->root_item_lock);
1241 1242
}

1243
static struct btrfs_root *btrfs_alloc_root(struct btrfs_fs_info *fs_info)
A
Al Viro 已提交
1244 1245 1246 1247 1248 1249 1250
{
	struct btrfs_root *root = kzalloc(sizeof(*root), GFP_NOFS);
	if (root)
		root->fs_info = fs_info;
	return root;
}

1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266
#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
/* Should only be used by the testing infrastructure */
struct btrfs_root *btrfs_alloc_dummy_root(void)
{
	struct btrfs_root *root;

	root = btrfs_alloc_root(NULL);
	if (!root)
		return ERR_PTR(-ENOMEM);
	__setup_root(4096, 4096, 4096, 4096, root, NULL, 1);
	root->dummy_root = 1;

	return root;
}
#endif

1267 1268 1269 1270 1271 1272 1273 1274 1275 1276
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;
1277
	uuid_le uuid;
1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293

	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);
1294
		leaf = NULL;
1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305
		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;

1306
	write_extent_buffer(leaf, fs_info->fsid, btrfs_header_fsid(),
1307 1308
			    BTRFS_FSID_SIZE);
	write_extent_buffer(leaf, fs_info->chunk_tree_uuid,
1309
			    btrfs_header_chunk_tree_uuid(leaf),
1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325
			    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);
1326 1327
	uuid_le_gen(&uuid);
	memcpy(root->root_item.uuid, uuid.b, BTRFS_UUID_SIZE);
1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338
	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);

1339 1340
	return root;

1341
fail:
1342 1343 1344 1345 1346
	if (leaf) {
		btrfs_tree_unlock(leaf);
		free_extent_buffer(leaf);
	}
	kfree(root);
1347

1348
	return ERR_PTR(ret);
1349 1350
}

Y
Yan Zheng 已提交
1351 1352
static struct btrfs_root *alloc_log_tree(struct btrfs_trans_handle *trans,
					 struct btrfs_fs_info *fs_info)
1353 1354 1355
{
	struct btrfs_root *root;
	struct btrfs_root *tree_root = fs_info->tree_root;
Y
Yan Zheng 已提交
1356
	struct extent_buffer *leaf;
1357

A
Al Viro 已提交
1358
	root = btrfs_alloc_root(fs_info);
1359
	if (!root)
Y
Yan Zheng 已提交
1360
		return ERR_PTR(-ENOMEM);
1361 1362 1363 1364 1365 1366 1367 1368

	__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
1369 1370 1371 1372 1373 1374
	/*
	 * 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).
	 */
1375 1376
	root->ref_cows = 0;

1377
	leaf = btrfs_alloc_free_block(trans, root, root->leafsize, 0,
A
Arne Jansen 已提交
1378
				      BTRFS_TREE_LOG_OBJECTID, NULL,
1379
				      0, 0, 0);
Y
Yan Zheng 已提交
1380 1381 1382 1383
	if (IS_ERR(leaf)) {
		kfree(root);
		return ERR_CAST(leaf);
	}
1384

1385 1386 1387 1388 1389
	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
1390
	root->node = leaf;
1391 1392

	write_extent_buffer(root->node, root->fs_info->fsid,
1393
			    btrfs_header_fsid(), BTRFS_FSID_SIZE);
1394 1395
	btrfs_mark_buffer_dirty(root->node);
	btrfs_tree_unlock(root->node);
Y
Yan Zheng 已提交
1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425
	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;
1426 1427 1428 1429 1430
	btrfs_set_stack_inode_generation(inode_item, 1);
	btrfs_set_stack_inode_size(inode_item, 3);
	btrfs_set_stack_inode_nlink(inode_item, 1);
	btrfs_set_stack_inode_nbytes(inode_item, root->leafsize);
	btrfs_set_stack_inode_mode(inode_item, S_IFDIR | 0755);
Y
Yan Zheng 已提交
1431

1432
	btrfs_set_root_node(&log_root->root_item, log_root->node);
Y
Yan Zheng 已提交
1433 1434 1435 1436

	WARN_ON(root->log_root);
	root->log_root = log_root;
	root->log_transid = 0;
1437
	root->last_log_commit = 0;
1438 1439 1440
	return 0;
}

1441 1442
static struct btrfs_root *btrfs_read_tree_root(struct btrfs_root *tree_root,
					       struct btrfs_key *key)
1443 1444 1445
{
	struct btrfs_root *root;
	struct btrfs_fs_info *fs_info = tree_root->fs_info;
1446
	struct btrfs_path *path;
1447
	u64 generation;
1448
	u32 blocksize;
1449
	int ret;
1450

1451 1452
	path = btrfs_alloc_path();
	if (!path)
1453
		return ERR_PTR(-ENOMEM);
1454 1455 1456 1457 1458

	root = btrfs_alloc_root(fs_info);
	if (!root) {
		ret = -ENOMEM;
		goto alloc_fail;
1459 1460
	}

1461
	__setup_root(tree_root->nodesize, tree_root->leafsize,
1462
		     tree_root->sectorsize, tree_root->stripesize,
1463
		     root, fs_info, key->objectid);
1464

1465 1466
	ret = btrfs_find_root(tree_root, key, path,
			      &root->root_item, &root->root_key);
1467
	if (ret) {
1468 1469
		if (ret > 0)
			ret = -ENOENT;
1470
		goto find_fail;
1471
	}
1472

1473
	generation = btrfs_root_generation(&root->root_item);
1474 1475
	blocksize = btrfs_level_size(root, btrfs_root_level(&root->root_item));
	root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
1476
				     blocksize, generation);
1477 1478 1479 1480 1481 1482
	if (!root->node) {
		ret = -ENOMEM;
		goto find_fail;
	} else if (!btrfs_buffer_uptodate(root->node, generation, 0)) {
		ret = -EIO;
		goto read_fail;
1483
	}
1484
	root->commit_root = btrfs_root_node(root);
1485
out:
1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507
	btrfs_free_path(path);
	return root;

read_fail:
	free_extent_buffer(root->node);
find_fail:
	kfree(root);
alloc_fail:
	root = ERR_PTR(ret);
	goto out;
}

struct btrfs_root *btrfs_read_fs_root(struct btrfs_root *tree_root,
				      struct btrfs_key *location)
{
	struct btrfs_root *root;

	root = btrfs_read_tree_root(tree_root, location);
	if (IS_ERR(root))
		return root;

	if (root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID) {
1508
		root->ref_cows = 1;
1509 1510
		btrfs_check_and_init_root_item(&root->root_item);
	}
1511

1512 1513 1514
	return root;
}

1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541
int btrfs_init_fs_root(struct btrfs_root *root)
{
	int ret;

	root->free_ino_ctl = kzalloc(sizeof(*root->free_ino_ctl), GFP_NOFS);
	root->free_ino_pinned = kzalloc(sizeof(*root->free_ino_pinned),
					GFP_NOFS);
	if (!root->free_ino_pinned || !root->free_ino_ctl) {
		ret = -ENOMEM;
		goto fail;
	}

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

	ret = get_anon_bdev(&root->anon_dev);
	if (ret)
		goto fail;
	return 0;
fail:
	kfree(root->free_ino_ctl);
	kfree(root->free_ino_pinned);
	return ret;
}

1542 1543
static struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info,
					       u64 root_id)
1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574
{
	struct btrfs_root *root;

	spin_lock(&fs_info->fs_roots_radix_lock);
	root = radix_tree_lookup(&fs_info->fs_roots_radix,
				 (unsigned long)root_id);
	spin_unlock(&fs_info->fs_roots_radix_lock);
	return root;
}

int btrfs_insert_fs_root(struct btrfs_fs_info *fs_info,
			 struct btrfs_root *root)
{
	int ret;

	ret = radix_tree_preload(GFP_NOFS & ~__GFP_HIGHMEM);
	if (ret)
		return ret;

	spin_lock(&fs_info->fs_roots_radix_lock);
	ret = radix_tree_insert(&fs_info->fs_roots_radix,
				(unsigned long)root->root_key.objectid,
				root);
	if (ret == 0)
		root->in_radix = 1;
	spin_unlock(&fs_info->fs_roots_radix_lock);
	radix_tree_preload_end();

	return ret;
}

1575 1576 1577
struct btrfs_root *btrfs_get_fs_root(struct btrfs_fs_info *fs_info,
				     struct btrfs_key *location,
				     bool check_ref)
1578 1579 1580 1581
{
	struct btrfs_root *root;
	int ret;

1582 1583 1584 1585
	if (location->objectid == BTRFS_ROOT_TREE_OBJECTID)
		return fs_info->tree_root;
	if (location->objectid == BTRFS_EXTENT_TREE_OBJECTID)
		return fs_info->extent_root;
1586 1587 1588 1589
	if (location->objectid == BTRFS_CHUNK_TREE_OBJECTID)
		return fs_info->chunk_root;
	if (location->objectid == BTRFS_DEV_TREE_OBJECTID)
		return fs_info->dev_root;
1590 1591
	if (location->objectid == BTRFS_CSUM_TREE_OBJECTID)
		return fs_info->csum_root;
1592 1593 1594
	if (location->objectid == BTRFS_QUOTA_TREE_OBJECTID)
		return fs_info->quota_root ? fs_info->quota_root :
					     ERR_PTR(-ENOENT);
1595 1596 1597
	if (location->objectid == BTRFS_UUID_TREE_OBJECTID)
		return fs_info->uuid_root ? fs_info->uuid_root :
					    ERR_PTR(-ENOENT);
1598
again:
1599
	root = btrfs_lookup_fs_root(fs_info, location->objectid);
1600
	if (root) {
1601
		if (check_ref && btrfs_root_refs(&root->root_item) == 0)
1602
			return ERR_PTR(-ENOENT);
1603
		return root;
1604
	}
1605

1606
	root = btrfs_read_fs_root(fs_info->tree_root, location);
1607 1608
	if (IS_ERR(root))
		return root;
1609

1610
	if (check_ref && btrfs_root_refs(&root->root_item) == 0) {
1611
		ret = -ENOENT;
1612
		goto fail;
1613
	}
1614

1615
	ret = btrfs_init_fs_root(root);
1616 1617
	if (ret)
		goto fail;
1618

1619 1620 1621 1622 1623 1624
	ret = btrfs_find_orphan_item(fs_info->tree_root, location->objectid);
	if (ret < 0)
		goto fail;
	if (ret == 0)
		root->orphan_item_inserted = 1;

1625
	ret = btrfs_insert_fs_root(fs_info, root);
1626
	if (ret) {
1627 1628 1629 1630 1631
		if (ret == -EEXIST) {
			free_fs_root(root);
			goto again;
		}
		goto fail;
1632
	}
1633
	return root;
1634 1635 1636
fail:
	free_fs_root(root);
	return ERR_PTR(ret);
1637 1638
}

C
Chris Mason 已提交
1639 1640 1641 1642 1643 1644
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
1645

1646 1647
	rcu_read_lock();
	list_for_each_entry_rcu(device, &info->fs_devices->devices, dev_list) {
1648 1649
		if (!device->bdev)
			continue;
C
Chris Mason 已提交
1650 1651 1652 1653 1654 1655
		bdi = blk_get_backing_dev_info(device->bdev);
		if (bdi && bdi_congested(bdi, bdi_bits)) {
			ret = 1;
			break;
		}
	}
1656
	rcu_read_unlock();
C
Chris Mason 已提交
1657 1658 1659
	return ret;
}

1660 1661 1662 1663
/*
 * If this fails, caller must call bdi_destroy() to get rid of the
 * bdi again.
 */
C
Chris Mason 已提交
1664 1665
static int setup_bdi(struct btrfs_fs_info *info, struct backing_dev_info *bdi)
{
1666 1667 1668
	int err;

	bdi->capabilities = BDI_CAP_MAP_COPY;
1669
	err = bdi_setup_and_register(bdi, "btrfs", BDI_CAP_MAP_COPY);
1670 1671 1672
	if (err)
		return err;

1673
	bdi->ra_pages	= default_backing_dev_info.ra_pages;
C
Chris Mason 已提交
1674 1675 1676 1677 1678
	bdi->congested_fn	= btrfs_congested_fn;
	bdi->congested_data	= info;
	return 0;
}

1679 1680 1681 1682 1683
/*
 * 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)
1684 1685
{
	struct bio *bio;
1686 1687
	struct end_io_wq *end_io_wq;
	struct btrfs_fs_info *fs_info;
1688 1689
	int error;

1690 1691 1692
	end_io_wq = container_of(work, struct end_io_wq, work);
	bio = end_io_wq->bio;
	fs_info = end_io_wq->info;
1693

1694 1695 1696 1697 1698
	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);
1699 1700
}

1701 1702 1703
static int cleaner_kthread(void *arg)
{
	struct btrfs_root *root = arg;
1704
	int again;
1705 1706

	do {
1707
		again = 0;
1708

1709
		/* Make the cleaner go to sleep early. */
1710
		if (btrfs_need_cleaner_sleep(root))
1711 1712 1713 1714 1715
			goto sleep;

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

1716 1717 1718 1719
		/*
		 * Avoid the problem that we change the status of the fs
		 * during the above check and trylock.
		 */
1720
		if (btrfs_need_cleaner_sleep(root)) {
1721 1722
			mutex_unlock(&root->fs_info->cleaner_mutex);
			goto sleep;
1723
		}
1724

1725 1726 1727 1728 1729
		btrfs_run_delayed_iputs(root);
		again = btrfs_clean_one_deleted_snapshot(root);
		mutex_unlock(&root->fs_info->cleaner_mutex);

		/*
1730 1731
		 * The defragger has dealt with the R/O remount and umount,
		 * needn't do anything special here.
1732 1733 1734
		 */
		btrfs_run_defrag_inodes(root->fs_info);
sleep:
D
David Sterba 已提交
1735
		if (!try_to_freeze() && !again) {
1736
			set_current_state(TASK_INTERRUPTIBLE);
1737 1738
			if (!kthread_should_stop())
				schedule();
1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749
			__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;
1750
	u64 transid;
1751 1752
	unsigned long now;
	unsigned long delay;
1753
	bool cannot_commit;
1754 1755

	do {
1756
		cannot_commit = false;
1757
		delay = HZ * root->fs_info->commit_interval;
1758 1759
		mutex_lock(&root->fs_info->transaction_kthread_mutex);

J
Josef Bacik 已提交
1760
		spin_lock(&root->fs_info->trans_lock);
1761 1762
		cur = root->fs_info->running_transaction;
		if (!cur) {
J
Josef Bacik 已提交
1763
			spin_unlock(&root->fs_info->trans_lock);
1764 1765
			goto sleep;
		}
Y
Yan Zheng 已提交
1766

1767
		now = get_seconds();
1768
		if (cur->state < TRANS_STATE_BLOCKED &&
1769 1770
		    (now < cur->start_time ||
		     now - cur->start_time < root->fs_info->commit_interval)) {
J
Josef Bacik 已提交
1771
			spin_unlock(&root->fs_info->trans_lock);
1772 1773 1774
			delay = HZ * 5;
			goto sleep;
		}
1775
		transid = cur->transid;
J
Josef Bacik 已提交
1776
		spin_unlock(&root->fs_info->trans_lock);
1777

1778
		/* If the file system is aborted, this will always fail. */
1779
		trans = btrfs_attach_transaction(root);
1780
		if (IS_ERR(trans)) {
1781 1782
			if (PTR_ERR(trans) != -ENOENT)
				cannot_commit = true;
1783
			goto sleep;
1784
		}
1785
		if (transid == trans->transid) {
1786
			btrfs_commit_transaction(trans, root);
1787 1788 1789
		} else {
			btrfs_end_transaction(trans, root);
		}
1790 1791 1792 1793
sleep:
		wake_up_process(root->fs_info->cleaner_kthread);
		mutex_unlock(&root->fs_info->transaction_kthread_mutex);

J
Josef Bacik 已提交
1794 1795 1796
		if (unlikely(test_bit(BTRFS_FS_STATE_ERROR,
				      &root->fs_info->fs_state)))
			btrfs_cleanup_transaction(root);
1797
		if (!try_to_freeze()) {
1798
			set_current_state(TASK_INTERRUPTIBLE);
1799
			if (!kthread_should_stop() &&
1800 1801
			    (!btrfs_transaction_blocked(root->fs_info) ||
			     cannot_commit))
1802
				schedule_timeout(delay);
1803 1804 1805 1806 1807 1808
			__set_current_state(TASK_RUNNING);
		}
	} while (!kthread_should_stop());
	return 0;
}

C
Chris Mason 已提交
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 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 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
/*
 * 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));

1915 1916 1917 1918 1919 1920 1921 1922
	/*
	 * 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
1923
			       btrfs_header_generation(info->fs_root->node));
1924
		btrfs_set_backup_fs_root_level(root_backup,
C
Chris Mason 已提交
1925
			       btrfs_header_level(info->fs_root->node));
1926
	}
C
Chris Mason 已提交
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 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

	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
2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026
/* 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
2027
	btrfs_stop_workers(&fs_info->qgroup_rescan_workers);
L
Liu Bo 已提交
2028 2029
}

2030 2031 2032 2033 2034 2035 2036 2037 2038 2039
static void free_root_extent_buffers(struct btrfs_root *root)
{
	if (root) {
		free_extent_buffer(root->node);
		free_extent_buffer(root->commit_root);
		root->node = NULL;
		root->commit_root = NULL;
	}
}

C
Chris Mason 已提交
2040 2041 2042
/* helper to cleanup tree roots */
static void free_root_pointers(struct btrfs_fs_info *info, int chunk_root)
{
2043
	free_root_extent_buffers(info->tree_root);
2044

2045 2046 2047 2048 2049 2050 2051
	free_root_extent_buffers(info->dev_root);
	free_root_extent_buffers(info->extent_root);
	free_root_extent_buffers(info->csum_root);
	free_root_extent_buffers(info->quota_root);
	free_root_extent_buffers(info->uuid_root);
	if (chunk_root)
		free_root_extent_buffers(info->chunk_root);
C
Chris Mason 已提交
2052 2053
}

2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065
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) {
2066
			btrfs_drop_and_free_fs_root(fs_info, gang[0]);
2067 2068 2069
		} else {
			free_extent_buffer(gang[0]->node);
			free_extent_buffer(gang[0]->commit_root);
2070
			btrfs_put_fs_root(gang[0]);
2071 2072 2073 2074 2075 2076 2077 2078 2079 2080
		}
	}

	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++)
2081
			btrfs_drop_and_free_fs_root(fs_info, gang[i]);
2082 2083
	}
}
C
Chris Mason 已提交
2084

A
Al Viro 已提交
2085 2086 2087
int open_ctree(struct super_block *sb,
	       struct btrfs_fs_devices *fs_devices,
	       char *options)
2088
{
2089 2090 2091 2092
	u32 sectorsize;
	u32 nodesize;
	u32 leafsize;
	u32 blocksize;
2093
	u32 stripesize;
2094
	u64 generation;
2095
	u64 features;
2096
	struct btrfs_key location;
2097
	struct buffer_head *bh;
2098
	struct btrfs_super_block *disk_super;
2099
	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
2100
	struct btrfs_root *tree_root;
2101 2102 2103 2104
	struct btrfs_root *extent_root;
	struct btrfs_root *csum_root;
	struct btrfs_root *chunk_root;
	struct btrfs_root *dev_root;
2105
	struct btrfs_root *quota_root;
2106
	struct btrfs_root *uuid_root;
2107
	struct btrfs_root *log_tree_root;
2108
	int ret;
2109
	int err = -EINVAL;
C
Chris Mason 已提交
2110 2111
	int num_backups_tried = 0;
	int backup_index = 0;
2112 2113
	bool create_uuid_tree;
	bool check_uuid_tree;
2114

2115
	tree_root = fs_info->tree_root = btrfs_alloc_root(fs_info);
A
Al Viro 已提交
2116
	chunk_root = fs_info->chunk_root = btrfs_alloc_root(fs_info);
2117
	if (!tree_root || !chunk_root) {
C
Chris Mason 已提交
2118 2119 2120
		err = -ENOMEM;
		goto fail;
	}
2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133

	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;
	}

2134 2135 2136 2137 2138 2139 2140 2141
	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));

2142 2143 2144 2145 2146 2147
	ret = percpu_counter_init(&fs_info->delalloc_bytes, 0);
	if (ret) {
		err = ret;
		goto fail_dirty_metadata_bytes;
	}

2148 2149 2150
	fs_info->btree_inode = new_inode(sb);
	if (!fs_info->btree_inode) {
		err = -ENOMEM;
2151
		goto fail_delalloc_bytes;
2152 2153
	}

2154
	mapping_set_gfp_mask(fs_info->btree_inode->i_mapping, GFP_NOFS);
2155

2156
	INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC);
C
Chris Mason 已提交
2157
	INIT_LIST_HEAD(&fs_info->trans_list);
2158
	INIT_LIST_HEAD(&fs_info->dead_roots);
Y
Yan, Zheng 已提交
2159
	INIT_LIST_HEAD(&fs_info->delayed_iputs);
2160
	INIT_LIST_HEAD(&fs_info->delalloc_roots);
2161
	INIT_LIST_HEAD(&fs_info->caching_block_groups);
2162
	spin_lock_init(&fs_info->delalloc_root_lock);
J
Josef Bacik 已提交
2163
	spin_lock_init(&fs_info->trans_lock);
2164
	spin_lock_init(&fs_info->fs_roots_radix_lock);
Y
Yan, Zheng 已提交
2165
	spin_lock_init(&fs_info->delayed_iput_lock);
C
Chris Mason 已提交
2166
	spin_lock_init(&fs_info->defrag_inodes_lock);
2167
	spin_lock_init(&fs_info->free_chunk_lock);
J
Jan Schmidt 已提交
2168
	spin_lock_init(&fs_info->tree_mod_seq_lock);
2169
	spin_lock_init(&fs_info->super_lock);
J
Jan Schmidt 已提交
2170
	rwlock_init(&fs_info->tree_mod_log_lock);
C
Chris Mason 已提交
2171
	mutex_init(&fs_info->reloc_mutex);
2172
	seqlock_init(&fs_info->profiles_lock);
2173

2174
	init_completion(&fs_info->kobj_unregister);
2175
	INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots);
2176
	INIT_LIST_HEAD(&fs_info->space_info);
J
Jan Schmidt 已提交
2177
	INIT_LIST_HEAD(&fs_info->tree_mod_seq_list);
2178
	btrfs_mapping_init(&fs_info->mapping_tree);
2179 2180 2181 2182 2183 2184 2185 2186 2187
	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);
2188
	atomic_set(&fs_info->nr_async_submits, 0);
2189
	atomic_set(&fs_info->async_delalloc_pages, 0);
2190
	atomic_set(&fs_info->async_submit_draining, 0);
2191
	atomic_set(&fs_info->nr_async_bios, 0);
C
Chris Mason 已提交
2192
	atomic_set(&fs_info->defrag_running, 0);
2193
	atomic64_set(&fs_info->tree_mod_seq, 0);
C
Chris Mason 已提交
2194
	fs_info->sb = sb;
2195
	fs_info->max_inline = 8192 * 1024;
J
Josef Bacik 已提交
2196
	fs_info->metadata_ratio = 0;
C
Chris Mason 已提交
2197
	fs_info->defrag_inodes = RB_ROOT;
2198
	fs_info->free_chunk_space = 0;
J
Jan Schmidt 已提交
2199
	fs_info->tree_mod_log = RB_ROOT;
2200
	fs_info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
C
Chris Mason 已提交
2201

2202 2203 2204
	/* 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
2205

2206 2207
	fs_info->thread_pool_size = min_t(unsigned long,
					  num_online_cpus() + 2, 8);
2208

2209 2210
	INIT_LIST_HEAD(&fs_info->ordered_roots);
	spin_lock_init(&fs_info->ordered_root_lock);
2211 2212 2213 2214 2215 2216 2217
	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);
2218

A
Arne Jansen 已提交
2219 2220 2221 2222 2223 2224 2225
	mutex_init(&fs_info->scrub_lock);
	atomic_set(&fs_info->scrubs_running, 0);
	atomic_set(&fs_info->scrub_pause_req, 0);
	atomic_set(&fs_info->scrubs_paused, 0);
	atomic_set(&fs_info->scrub_cancel_req, 0);
	init_waitqueue_head(&fs_info->scrub_pause_wait);
	fs_info->scrub_workers_refcnt = 0;
2226 2227 2228
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	fs_info->check_integrity_print_mask = 0;
#endif
A
Arne Jansen 已提交
2229

2230 2231
	spin_lock_init(&fs_info->balance_lock);
	mutex_init(&fs_info->balance_mutex);
2232 2233
	atomic_set(&fs_info->balance_running, 0);
	atomic_set(&fs_info->balance_pause_req, 0);
2234
	atomic_set(&fs_info->balance_cancel_req, 0);
2235
	fs_info->balance_ctl = NULL;
2236
	init_waitqueue_head(&fs_info->balance_wait_q);
A
Arne Jansen 已提交
2237

2238 2239
	sb->s_blocksize = 4096;
	sb->s_blocksize_bits = blksize_bits(4096);
J
Jens Axboe 已提交
2240
	sb->s_bdi = &fs_info->bdi;
2241

2242
	fs_info->btree_inode->i_ino = BTRFS_BTREE_INODE_OBJECTID;
M
Miklos Szeredi 已提交
2243
	set_nlink(fs_info->btree_inode, 1);
2244 2245 2246 2247 2248 2249
	/*
	 * 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;
2250
	fs_info->btree_inode->i_mapping->a_ops = &btree_aops;
C
Chris Mason 已提交
2251 2252
	fs_info->btree_inode->i_mapping->backing_dev_info = &fs_info->bdi;

2253
	RB_CLEAR_NODE(&BTRFS_I(fs_info->btree_inode)->rb_node);
2254
	extent_io_tree_init(&BTRFS_I(fs_info->btree_inode)->io_tree,
2255
			     fs_info->btree_inode->i_mapping);
2256
	BTRFS_I(fs_info->btree_inode)->io_tree.track_uptodate = 0;
2257
	extent_map_tree_init(&BTRFS_I(fs_info->btree_inode)->extent_tree);
2258 2259

	BTRFS_I(fs_info->btree_inode)->io_tree.ops = &btree_extent_io_ops;
2260

2261 2262 2263
	BTRFS_I(fs_info->btree_inode)->root = tree_root;
	memset(&BTRFS_I(fs_info->btree_inode)->location, 0,
	       sizeof(struct btrfs_key));
2264 2265
	set_bit(BTRFS_INODE_DUMMY,
		&BTRFS_I(fs_info->btree_inode)->runtime_flags);
2266
	btrfs_insert_inode_hash(fs_info->btree_inode);
2267

J
Josef Bacik 已提交
2268
	spin_lock_init(&fs_info->block_group_cache_lock);
2269
	fs_info->block_group_cache_tree = RB_ROOT;
2270
	fs_info->first_logical_byte = (u64)-1;
J
Josef Bacik 已提交
2271

2272
	extent_io_tree_init(&fs_info->freed_extents[0],
2273
			     fs_info->btree_inode->i_mapping);
2274
	extent_io_tree_init(&fs_info->freed_extents[1],
2275
			     fs_info->btree_inode->i_mapping);
2276
	fs_info->pinned_extents = &fs_info->freed_extents[0];
2277
	fs_info->do_barriers = 1;
2278

C
Chris Mason 已提交
2279

2280
	mutex_init(&fs_info->ordered_operations_mutex);
2281
	mutex_init(&fs_info->ordered_extent_flush_mutex);
2282
	mutex_init(&fs_info->tree_log_mutex);
2283
	mutex_init(&fs_info->chunk_mutex);
2284 2285
	mutex_init(&fs_info->transaction_kthread_mutex);
	mutex_init(&fs_info->cleaner_mutex);
2286
	mutex_init(&fs_info->volume_mutex);
2287
	init_rwsem(&fs_info->extent_commit_sem);
2288
	init_rwsem(&fs_info->cleanup_work_sem);
2289
	init_rwsem(&fs_info->subvol_sem);
S
Stefan Behrens 已提交
2290
	sema_init(&fs_info->uuid_tree_rescan_sem, 1);
2291 2292 2293 2294 2295
	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);
2296

2297
	spin_lock_init(&fs_info->qgroup_lock);
2298
	mutex_init(&fs_info->qgroup_ioctl_lock);
2299 2300 2301 2302 2303
	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;
2304
	fs_info->qgroup_ulist = NULL;
J
Jan Schmidt 已提交
2305
	mutex_init(&fs_info->qgroup_rescan_lock);
2306

2307 2308 2309
	btrfs_init_free_cluster(&fs_info->meta_alloc_cluster);
	btrfs_init_free_cluster(&fs_info->data_alloc_cluster);

2310
	init_waitqueue_head(&fs_info->transaction_throttle);
2311
	init_waitqueue_head(&fs_info->transaction_wait);
S
Sage Weil 已提交
2312
	init_waitqueue_head(&fs_info->transaction_blocked_wait);
2313
	init_waitqueue_head(&fs_info->async_submit_wait);
2314

D
David Woodhouse 已提交
2315 2316
	ret = btrfs_alloc_stripe_hash_table(fs_info);
	if (ret) {
2317
		err = ret;
D
David Woodhouse 已提交
2318 2319 2320
		goto fail_alloc;
	}

2321
	__setup_root(4096, 4096, 4096, 4096, tree_root,
C
Chris Mason 已提交
2322
		     fs_info, BTRFS_ROOT_TREE_OBJECTID);
2323

2324
	invalidate_bdev(fs_devices->latest_bdev);
D
David Sterba 已提交
2325 2326 2327 2328

	/*
	 * Read super block and check the signature bytes only
	 */
Y
Yan Zheng 已提交
2329
	bh = btrfs_read_dev_super(fs_devices->latest_bdev);
2330 2331
	if (!bh) {
		err = -EINVAL;
2332
		goto fail_alloc;
2333
	}
C
Chris Mason 已提交
2334

D
David Sterba 已提交
2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349
	/*
	 * 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
	 */
2350 2351 2352
	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));
2353
	brelse(bh);
2354

2355
	memcpy(fs_info->fsid, fs_info->super_copy->fsid, BTRFS_FSID_SIZE);
2356

D
David Sterba 已提交
2357 2358 2359 2360 2361 2362 2363
	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;
	}

2364
	disk_super = fs_info->super_copy;
2365
	if (!btrfs_super_root(disk_super))
2366
		goto fail_alloc;
2367

L
liubo 已提交
2368
	/* check FS state, whether FS is broken. */
2369 2370
	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
2371

C
Chris Mason 已提交
2372 2373 2374 2375 2376 2377 2378
	/*
	 * 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);

2379 2380 2381 2382 2383 2384
	/*
	 * 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
2385 2386 2387
	ret = btrfs_parse_options(tree_root, options);
	if (ret) {
		err = ret;
2388
		goto fail_alloc;
Y
Yan Zheng 已提交
2389
	}
2390

2391 2392 2393 2394 2395
	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",
2396
		       features);
2397
		err = -EINVAL;
2398
		goto fail_alloc;
2399 2400
	}

2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417
	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;
	}

2418
	features = btrfs_super_incompat_flags(disk_super);
L
Li Zefan 已提交
2419
	features |= BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF;
L
Li Zefan 已提交
2420
	if (tree_root->fs_info->compress_type == BTRFS_COMPRESS_LZO)
L
Li Zefan 已提交
2421
		features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
2422

2423 2424 2425
	if (features & BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA)
		printk(KERN_ERR "btrfs: has skinny extents\n");

2426 2427 2428 2429 2430 2431 2432 2433 2434 2435
	/*
	 * 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;
	}

2436 2437 2438 2439
	nodesize = btrfs_super_nodesize(disk_super);
	leafsize = btrfs_super_leafsize(disk_super);
	sectorsize = btrfs_super_sectorsize(disk_super);
	stripesize = btrfs_super_stripesize(disk_super);
2440
	fs_info->dirty_metadata_batch = leafsize * (1 + ilog2(nr_cpu_ids));
2441
	fs_info->delalloc_batch = sectorsize * 512 * (1 + ilog2(nr_cpu_ids));
2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454

	/*
	 * 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;
	}

2455 2456 2457 2458
	/*
	 * Needn't use the lock because there is no other task which will
	 * update the flag.
	 */
L
Li Zefan 已提交
2459
	btrfs_set_super_incompat_flags(disk_super, features);
2460

2461 2462 2463 2464 2465
	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",
2466
		       features);
2467
		err = -EINVAL;
2468
		goto fail_alloc;
2469
	}
2470 2471 2472 2473

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

2474
	btrfs_init_workers(&fs_info->workers, "worker",
2475 2476
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
C
Chris Mason 已提交
2477

2478
	btrfs_init_workers(&fs_info->delalloc_workers, "delalloc",
2479
			   fs_info->thread_pool_size, NULL);
2480

2481
	btrfs_init_workers(&fs_info->flush_workers, "flush_delalloc",
2482
			   fs_info->thread_pool_size, NULL);
2483

2484
	btrfs_init_workers(&fs_info->submit_workers, "submit",
2485
			   min_t(u64, fs_devices->num_devices,
2486
			   fs_info->thread_pool_size), NULL);
2487

2488
	btrfs_init_workers(&fs_info->caching_workers, "cache",
2489
			   fs_info->thread_pool_size, NULL);
2490

2491 2492 2493 2494 2495
	/* 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;
2496

2497
	fs_info->workers.idle_thresh = 16;
C
Chris Mason 已提交
2498
	fs_info->workers.ordered = 1;
2499

2500 2501 2502
	fs_info->delalloc_workers.idle_thresh = 2;
	fs_info->delalloc_workers.ordered = 1;

2503 2504
	btrfs_init_workers(&fs_info->fixup_workers, "fixup", 1,
			   &fs_info->generic_worker);
2505
	btrfs_init_workers(&fs_info->endio_workers, "endio",
2506 2507
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2508
	btrfs_init_workers(&fs_info->endio_meta_workers, "endio-meta",
2509 2510
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2511
	btrfs_init_workers(&fs_info->endio_meta_write_workers,
2512 2513
			   "endio-meta-write", fs_info->thread_pool_size,
			   &fs_info->generic_worker);
D
David Woodhouse 已提交
2514 2515 2516 2517 2518 2519
	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);
2520
	btrfs_init_workers(&fs_info->endio_write_workers, "endio-write",
2521 2522
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
J
Josef Bacik 已提交
2523 2524
	btrfs_init_workers(&fs_info->endio_freespace_worker, "freespace-write",
			   1, &fs_info->generic_worker);
2525 2526 2527
	btrfs_init_workers(&fs_info->delayed_workers, "delayed-meta",
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2528 2529 2530
	btrfs_init_workers(&fs_info->readahead_workers, "readahead",
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
J
Jan Schmidt 已提交
2531 2532
	btrfs_init_workers(&fs_info->qgroup_rescan_workers, "qgroup-rescan", 1,
			   &fs_info->generic_worker);
2533 2534 2535 2536 2537 2538

	/*
	 * endios are largely parallel and should have a very
	 * low idle thresh
	 */
	fs_info->endio_workers.idle_thresh = 4;
2539
	fs_info->endio_meta_workers.idle_thresh = 4;
D
David Woodhouse 已提交
2540 2541
	fs_info->endio_raid56_workers.idle_thresh = 4;
	fs_info->rmw_workers.idle_thresh = 2;
2542

2543 2544
	fs_info->endio_write_workers.idle_thresh = 2;
	fs_info->endio_meta_write_workers.idle_thresh = 2;
2545
	fs_info->readahead_workers.idle_thresh = 2;
2546

2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557
	/*
	 * 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
2558 2559
	ret |= btrfs_start_workers(&fs_info->rmw_workers);
	ret |= btrfs_start_workers(&fs_info->endio_raid56_workers);
2560 2561 2562 2563 2564 2565
	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);
2566
	ret |= btrfs_start_workers(&fs_info->flush_workers);
J
Jan Schmidt 已提交
2567
	ret |= btrfs_start_workers(&fs_info->qgroup_rescan_workers);
2568
	if (ret) {
2569
		err = -ENOMEM;
2570 2571
		goto fail_sb_buffer;
	}
2572

2573
	fs_info->bdi.ra_pages *= btrfs_super_num_devices(disk_super);
C
Chris Mason 已提交
2574 2575
	fs_info->bdi.ra_pages = max(fs_info->bdi.ra_pages,
				    4 * 1024 * 1024 / PAGE_CACHE_SIZE);
2576

2577 2578 2579
	tree_root->nodesize = nodesize;
	tree_root->leafsize = leafsize;
	tree_root->sectorsize = sectorsize;
2580
	tree_root->stripesize = stripesize;
2581 2582 2583

	sb->s_blocksize = sectorsize;
	sb->s_blocksize_bits = blksize_bits(sectorsize);
2584

2585
	if (btrfs_super_magic(disk_super) != BTRFS_MAGIC) {
C
Chris Mason 已提交
2586
		printk(KERN_INFO "btrfs: valid FS not found on %s\n", sb->s_id);
C
Chris Mason 已提交
2587 2588
		goto fail_sb_buffer;
	}
2589

2590 2591 2592
	if (sectorsize != PAGE_SIZE) {
		printk(KERN_WARNING "btrfs: Incompatible sector size(%lu) "
		       "found on %s\n", (unsigned long)sectorsize, sb->s_id);
2593 2594 2595
		goto fail_sb_buffer;
	}

2596
	mutex_lock(&fs_info->chunk_mutex);
Y
Yan Zheng 已提交
2597
	ret = btrfs_read_sys_array(tree_root);
2598
	mutex_unlock(&fs_info->chunk_mutex);
2599
	if (ret) {
C
Chris Mason 已提交
2600 2601
		printk(KERN_WARNING "btrfs: failed to read the system "
		       "array on %s\n", sb->s_id);
2602
		goto fail_sb_buffer;
2603
	}
2604 2605 2606

	blocksize = btrfs_level_size(tree_root,
				     btrfs_super_chunk_root_level(disk_super));
2607
	generation = btrfs_super_chunk_root_generation(disk_super);
2608 2609 2610 2611 2612 2613

	__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),
2614
					   blocksize, generation);
2615 2616
	if (!chunk_root->node ||
	    !test_bit(EXTENT_BUFFER_UPTODATE, &chunk_root->node->bflags)) {
2617 2618
		printk(KERN_WARNING "btrfs: failed to read chunk root on %s\n",
		       sb->s_id);
C
Chris Mason 已提交
2619
		goto fail_tree_roots;
2620
	}
2621 2622
	btrfs_set_root_node(&chunk_root->root_item, chunk_root->node);
	chunk_root->commit_root = btrfs_root_node(chunk_root);
2623

2624
	read_extent_buffer(chunk_root->node, fs_info->chunk_tree_uuid,
2625
	   btrfs_header_chunk_tree_uuid(chunk_root->node), BTRFS_UUID_SIZE);
2626

2627
	ret = btrfs_read_chunk_tree(chunk_root);
Y
Yan Zheng 已提交
2628
	if (ret) {
C
Chris Mason 已提交
2629 2630
		printk(KERN_WARNING "btrfs: failed to read chunk tree on %s\n",
		       sb->s_id);
C
Chris Mason 已提交
2631
		goto fail_tree_roots;
Y
Yan Zheng 已提交
2632
	}
2633

2634 2635 2636 2637 2638
	/*
	 * 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);
2639

2640 2641 2642 2643 2644 2645
	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
2646
retry_root_backup:
2647 2648
	blocksize = btrfs_level_size(tree_root,
				     btrfs_super_root_level(disk_super));
2649
	generation = btrfs_super_generation(disk_super);
2650

C
Chris Mason 已提交
2651
	tree_root->node = read_tree_block(tree_root,
2652
					  btrfs_super_root(disk_super),
2653
					  blocksize, generation);
C
Chris Mason 已提交
2654 2655
	if (!tree_root->node ||
	    !test_bit(EXTENT_BUFFER_UPTODATE, &tree_root->node->bflags)) {
2656 2657
		printk(KERN_WARNING "btrfs: failed to read tree root on %s\n",
		       sb->s_id);
C
Chris Mason 已提交
2658 2659

		goto recovery_tree_root;
2660
	}
C
Chris Mason 已提交
2661

2662 2663
	btrfs_set_root_node(&tree_root->root_item, tree_root->node);
	tree_root->commit_root = btrfs_root_node(tree_root);
2664
	btrfs_set_root_refs(&tree_root->root_item, 1);
2665

2666 2667 2668 2669 2670 2671 2672
	location.objectid = BTRFS_EXTENT_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
	location.offset = 0;

	extent_root = btrfs_read_tree_root(tree_root, &location);
	if (IS_ERR(extent_root)) {
		ret = PTR_ERR(extent_root);
C
Chris Mason 已提交
2673
		goto recovery_tree_root;
2674
	}
2675
	extent_root->track_dirty = 1;
2676
	fs_info->extent_root = extent_root;
2677

2678 2679 2680 2681
	location.objectid = BTRFS_DEV_TREE_OBJECTID;
	dev_root = btrfs_read_tree_root(tree_root, &location);
	if (IS_ERR(dev_root)) {
		ret = PTR_ERR(dev_root);
C
Chris Mason 已提交
2682
		goto recovery_tree_root;
2683
	}
2684
	dev_root->track_dirty = 1;
2685 2686
	fs_info->dev_root = dev_root;
	btrfs_init_devices_late(fs_info);
2687

2688 2689 2690 2691
	location.objectid = BTRFS_CSUM_TREE_OBJECTID;
	csum_root = btrfs_read_tree_root(tree_root, &location);
	if (IS_ERR(csum_root)) {
		ret = PTR_ERR(csum_root);
C
Chris Mason 已提交
2692
		goto recovery_tree_root;
2693
	}
2694
	csum_root->track_dirty = 1;
2695
	fs_info->csum_root = csum_root;
2696

2697 2698 2699
	location.objectid = BTRFS_QUOTA_TREE_OBJECTID;
	quota_root = btrfs_read_tree_root(tree_root, &location);
	if (!IS_ERR(quota_root)) {
2700 2701 2702
		quota_root->track_dirty = 1;
		fs_info->quota_enabled = 1;
		fs_info->pending_quota_state = 1;
2703
		fs_info->quota_root = quota_root;
2704 2705
	}

2706 2707 2708 2709 2710 2711 2712
	location.objectid = BTRFS_UUID_TREE_OBJECTID;
	uuid_root = btrfs_read_tree_root(tree_root, &location);
	if (IS_ERR(uuid_root)) {
		ret = PTR_ERR(uuid_root);
		if (ret != -ENOENT)
			goto recovery_tree_root;
		create_uuid_tree = true;
2713
		check_uuid_tree = false;
2714 2715 2716
	} else {
		uuid_root->track_dirty = 1;
		fs_info->uuid_root = uuid_root;
2717 2718 2719
		create_uuid_tree = false;
		check_uuid_tree =
		    generation != btrfs_super_uuid_tree_generation(disk_super);
2720 2721
	}

2722 2723 2724
	fs_info->generation = generation;
	fs_info->last_trans_committed = generation;

2725 2726 2727 2728 2729 2730
	ret = btrfs_recover_balance(fs_info);
	if (ret) {
		printk(KERN_WARNING "btrfs: failed to recover balance\n");
		goto fail_block_groups;
	}

2731 2732 2733 2734 2735 2736 2737
	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;
	}

2738 2739 2740 2741 2742 2743 2744 2745
	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);

2746 2747 2748 2749 2750 2751
	ret = btrfs_init_space_info(fs_info);
	if (ret) {
		printk(KERN_ERR "Failed to initial space info: %d\n", ret);
		goto fail_block_groups;
	}

2752 2753 2754 2755 2756
	ret = btrfs_read_block_groups(extent_root);
	if (ret) {
		printk(KERN_ERR "Failed to read block groups: %d\n", ret);
		goto fail_block_groups;
	}
2757 2758
	fs_info->num_tolerated_disk_barrier_failures =
		btrfs_calc_num_tolerated_disk_barrier_failures(fs_info);
2759 2760 2761 2762 2763 2764 2765
	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
2766

2767 2768
	fs_info->cleaner_kthread = kthread_run(cleaner_kthread, tree_root,
					       "btrfs-cleaner");
2769
	if (IS_ERR(fs_info->cleaner_kthread))
2770
		goto fail_block_groups;
2771 2772 2773 2774

	fs_info->transaction_kthread = kthread_run(transaction_kthread,
						   tree_root,
						   "btrfs-transaction");
2775
	if (IS_ERR(fs_info->transaction_kthread))
2776
		goto fail_cleaner;
2777

C
Chris Mason 已提交
2778 2779 2780 2781 2782 2783 2784 2785
	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);
	}

2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797
#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
2798 2799 2800
	ret = btrfs_read_qgroup_config(fs_info);
	if (ret)
		goto fail_trans_kthread;
2801

L
liubo 已提交
2802
	/* do not make disk changes in broken FS */
2803
	if (btrfs_super_log_root(disk_super) != 0) {
2804 2805
		u64 bytenr = btrfs_super_log_root(disk_super);

2806
		if (fs_devices->rw_devices == 0) {
C
Chris Mason 已提交
2807 2808
			printk(KERN_WARNING "Btrfs log replay required "
			       "on RO media\n");
2809
			err = -EIO;
2810
			goto fail_qgroup;
2811
		}
2812 2813 2814
		blocksize =
		     btrfs_level_size(tree_root,
				      btrfs_super_log_root_level(disk_super));
C
Chris Mason 已提交
2815

A
Al Viro 已提交
2816
		log_tree_root = btrfs_alloc_root(fs_info);
2817 2818
		if (!log_tree_root) {
			err = -ENOMEM;
2819
			goto fail_qgroup;
2820
		}
2821 2822 2823 2824 2825

		__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,
2826 2827
						      blocksize,
						      generation + 1);
2828 2829 2830 2831 2832 2833 2834
		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;
		}
2835
		/* returns with log_tree_root freed on success */
2836
		ret = btrfs_recover_log_trees(log_tree_root);
2837 2838 2839 2840 2841 2842 2843
		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;
		}
2844 2845

		if (sb->s_flags & MS_RDONLY) {
2846 2847 2848
			ret = btrfs_commit_super(tree_root);
			if (ret)
				goto fail_trans_kthread;
2849
		}
2850
	}
Z
Zheng Yan 已提交
2851

2852
	ret = btrfs_find_orphan_roots(tree_root);
2853 2854
	if (ret)
		goto fail_trans_kthread;
2855

2856
	if (!(sb->s_flags & MS_RDONLY)) {
2857
		ret = btrfs_cleanup_fs_roots(fs_info);
2858 2859
		if (ret)
			goto fail_trans_kthread;
2860

2861
		ret = btrfs_recover_relocation(tree_root);
2862 2863 2864 2865
		if (ret < 0) {
			printk(KERN_WARNING
			       "btrfs: failed to recover relocation\n");
			err = -EINVAL;
2866
			goto fail_qgroup;
2867
		}
2868
	}
Z
Zheng Yan 已提交
2869

2870 2871
	location.objectid = BTRFS_FS_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
2872
	location.offset = 0;
2873 2874

	fs_info->fs_root = btrfs_read_fs_root_no_name(fs_info, &location);
2875 2876
	if (IS_ERR(fs_info->fs_root)) {
		err = PTR_ERR(fs_info->fs_root);
2877
		goto fail_qgroup;
2878
	}
C
Chris Mason 已提交
2879

2880 2881
	if (sb->s_flags & MS_RDONLY)
		return 0;
I
Ilya Dryomov 已提交
2882

2883 2884 2885
	down_read(&fs_info->cleanup_work_sem);
	if ((ret = btrfs_orphan_cleanup(fs_info->fs_root)) ||
	    (ret = btrfs_orphan_cleanup(fs_info->tree_root))) {
2886
		up_read(&fs_info->cleanup_work_sem);
2887 2888 2889 2890
		close_ctree(tree_root);
		return ret;
	}
	up_read(&fs_info->cleanup_work_sem);
I
Ilya Dryomov 已提交
2891

2892 2893 2894 2895 2896
	ret = btrfs_resume_balance_async(fs_info);
	if (ret) {
		printk(KERN_WARNING "btrfs: failed to resume balance\n");
		close_ctree(tree_root);
		return ret;
2897 2898
	}

2899 2900 2901 2902 2903 2904 2905
	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;
	}

2906 2907
	btrfs_qgroup_rescan_resume(fs_info);

2908 2909 2910 2911 2912 2913 2914 2915 2916
	if (create_uuid_tree) {
		pr_info("btrfs: creating UUID tree\n");
		ret = btrfs_create_uuid_tree(fs_info);
		if (ret) {
			pr_warn("btrfs: failed to create the UUID tree %d\n",
				ret);
			close_ctree(tree_root);
			return ret;
		}
2917 2918
	} else if (check_uuid_tree ||
		   btrfs_test_opt(tree_root, RESCAN_UUID_TREE)) {
2919 2920 2921 2922 2923 2924 2925 2926 2927 2928
		pr_info("btrfs: checking UUID tree\n");
		ret = btrfs_check_uuid_tree(fs_info);
		if (ret) {
			pr_warn("btrfs: failed to check the UUID tree %d\n",
				ret);
			close_ctree(tree_root);
			return ret;
		}
	} else {
		fs_info->update_uuid_tree_gen = 1;
2929 2930
	}

A
Al Viro 已提交
2931
	return 0;
C
Chris Mason 已提交
2932

2933 2934
fail_qgroup:
	btrfs_free_qgroup_config(fs_info);
2935 2936
fail_trans_kthread:
	kthread_stop(fs_info->transaction_kthread);
J
Josef Bacik 已提交
2937
	btrfs_cleanup_transaction(fs_info->tree_root);
2938
	del_fs_roots(fs_info);
2939
fail_cleaner:
2940
	kthread_stop(fs_info->cleaner_kthread);
2941 2942 2943 2944 2945 2946 2947

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

2948
fail_block_groups:
J
Josef Bacik 已提交
2949
	btrfs_put_block_group_cache(fs_info);
2950
	btrfs_free_block_groups(fs_info);
C
Chris Mason 已提交
2951 2952 2953

fail_tree_roots:
	free_root_pointers(fs_info, 1);
2954
	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
2955

C
Chris Mason 已提交
2956
fail_sb_buffer:
L
Liu Bo 已提交
2957
	btrfs_stop_all_workers(fs_info);
2958
fail_alloc:
2959
fail_iput:
2960 2961
	btrfs_mapping_tree_free(&fs_info->mapping_tree);

2962
	iput(fs_info->btree_inode);
2963 2964
fail_delalloc_bytes:
	percpu_counter_destroy(&fs_info->delalloc_bytes);
2965 2966
fail_dirty_metadata_bytes:
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
2967
fail_bdi:
2968
	bdi_destroy(&fs_info->bdi);
2969 2970
fail_srcu:
	cleanup_srcu_struct(&fs_info->subvol_srcu);
2971
fail:
D
David Woodhouse 已提交
2972
	btrfs_free_stripe_hash_table(fs_info);
2973
	btrfs_close_devices(fs_info->fs_devices);
A
Al Viro 已提交
2974
	return err;
C
Chris Mason 已提交
2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992

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;
2993 2994
}

2995 2996 2997 2998 2999
static void btrfs_end_buffer_write_sync(struct buffer_head *bh, int uptodate)
{
	if (uptodate) {
		set_buffer_uptodate(bh);
	} else {
3000 3001 3002
		struct btrfs_device *device = (struct btrfs_device *)
			bh->b_private;

3003 3004 3005
		printk_ratelimited_in_rcu(KERN_WARNING "lost page write due to "
					  "I/O error on %s\n",
					  rcu_str_deref(device->name));
3006 3007 3008
		/* note, we dont' set_buffer_write_io_error because we have
		 * our own ways of dealing with the IO errors
		 */
3009
		clear_buffer_uptodate(bh);
3010
		btrfs_dev_stat_inc_and_print(device, BTRFS_DEV_STAT_WRITE_ERRS);
3011 3012 3013 3014 3015
	}
	unlock_buffer(bh);
	put_bh(bh);
}

Y
Yan Zheng 已提交
3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031
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);
3032 3033
		if (bytenr + BTRFS_SUPER_INFO_SIZE >=
					i_size_read(bdev->bd_inode))
Y
Yan Zheng 已提交
3034
			break;
3035 3036
		bh = __bread(bdev, bytenr / 4096,
					BTRFS_SUPER_INFO_SIZE);
Y
Yan Zheng 已提交
3037 3038 3039 3040 3041
		if (!bh)
			continue;

		super = (struct btrfs_super_block *)bh->b_data;
		if (btrfs_super_bytenr(super) != bytenr ||
3042
		    btrfs_super_magic(super) != BTRFS_MAGIC) {
Y
Yan Zheng 已提交
3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057
			brelse(bh);
			continue;
		}

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

3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068
/*
 * 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
3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090
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);
3091 3092 3093 3094
			if (!bh) {
				errors++;
				continue;
			}
Y
Yan Zheng 已提交
3095
			wait_on_buffer(bh);
3096 3097 3098 3099 3100 3101 3102 3103 3104
			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
3105 3106 3107 3108
		} else {
			btrfs_set_super_bytenr(sb, bytenr);

			crc = ~(u32)0;
3109
			crc = btrfs_csum_data((char *)sb +
Y
Yan Zheng 已提交
3110 3111 3112 3113 3114
					      BTRFS_CSUM_SIZE, crc,
					      BTRFS_SUPER_INFO_SIZE -
					      BTRFS_CSUM_SIZE);
			btrfs_csum_final(crc, sb->csum);

3115 3116 3117 3118
			/*
			 * one reference for us, and we leave it for the
			 * caller
			 */
Y
Yan Zheng 已提交
3119 3120
			bh = __getblk(device->bdev, bytenr / 4096,
				      BTRFS_SUPER_INFO_SIZE);
3121 3122 3123 3124 3125 3126 3127
			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
3128 3129
			memcpy(bh->b_data, sb, BTRFS_SUPER_INFO_SIZE);

3130
			/* one reference for submit_bh */
Y
Yan Zheng 已提交
3131
			get_bh(bh);
3132 3133

			set_buffer_uptodate(bh);
Y
Yan Zheng 已提交
3134 3135
			lock_buffer(bh);
			bh->b_end_io = btrfs_end_buffer_write_sync;
3136
			bh->b_private = device;
Y
Yan Zheng 已提交
3137 3138
		}

C
Chris Mason 已提交
3139 3140 3141 3142
		/*
		 * we fua the first super.  The others we allow
		 * to go down lazy.
		 */
3143
		ret = btrfsic_submit_bh(WRITE_FUA, bh);
3144
		if (ret)
Y
Yan Zheng 已提交
3145 3146 3147 3148 3149
			errors++;
	}
	return errors < i ? 0 : -1;
}

C
Chris Mason 已提交
3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188
/*
 * 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)) {
3189 3190
			printk_in_rcu("btrfs: disabling barriers on dev %s\n",
				      rcu_str_deref(device->name));
C
Chris Mason 已提交
3191
			device->nobarriers = 1;
3192
		} else if (!bio_flagged(bio, BIO_UPTODATE)) {
C
Chris Mason 已提交
3193
			ret = -EIO;
3194 3195
			btrfs_dev_stat_inc_and_print(device,
				BTRFS_DEV_STAT_FLUSH_ERRS);
C
Chris Mason 已提交
3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208
		}

		/* 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
	 */
3209
	device->flush_bio = NULL;
3210
	bio = btrfs_io_bio_alloc(GFP_NOFS, 0);
C
Chris Mason 已提交
3211 3212 3213 3214 3215 3216 3217 3218 3219 3220
	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);
3221
	btrfsic_submit_bio(WRITE_FLUSH, bio);
C
Chris Mason 已提交
3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233

	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;
3234 3235
	int errors_send = 0;
	int errors_wait = 0;
C
Chris Mason 已提交
3236 3237 3238 3239 3240 3241
	int ret;

	/* send down all the barriers */
	head = &info->fs_devices->devices;
	list_for_each_entry_rcu(dev, head, dev_list) {
		if (!dev->bdev) {
3242
			errors_send++;
C
Chris Mason 已提交
3243 3244 3245 3246 3247 3248 3249
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

		ret = write_dev_flush(dev, 0);
		if (ret)
3250
			errors_send++;
C
Chris Mason 已提交
3251 3252 3253 3254 3255
	}

	/* wait for all the barriers */
	list_for_each_entry_rcu(dev, head, dev_list) {
		if (!dev->bdev) {
3256
			errors_wait++;
C
Chris Mason 已提交
3257 3258 3259 3260 3261 3262 3263
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

		ret = write_dev_flush(dev, 1);
		if (ret)
3264
			errors_wait++;
C
Chris Mason 已提交
3265
	}
3266 3267
	if (errors_send > info->num_tolerated_disk_barrier_failures ||
	    errors_wait > info->num_tolerated_disk_barrier_failures)
C
Chris Mason 已提交
3268 3269 3270 3271
		return -EIO;
	return 0;
}

3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331
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
3332 3333 3334 3335 3336 3337
				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 &
3338
						   BTRFS_BLOCK_GROUP_RAID6) {
D
David Woodhouse 已提交
3339 3340 3341
						num_tolerated_disk_barrier_failures = 2;
					}
				}
3342 3343 3344 3345 3346 3347 3348 3349
			}
		}
		up_read(&sinfo->groups_sem);
	}

	return num_tolerated_disk_barrier_failures;
}

3350
static int write_all_supers(struct btrfs_root *root, int max_mirrors)
3351
{
3352
	struct list_head *head;
3353
	struct btrfs_device *dev;
3354
	struct btrfs_super_block *sb;
3355 3356 3357
	struct btrfs_dev_item *dev_item;
	int ret;
	int do_barriers;
3358 3359
	int max_errors;
	int total_errors = 0;
3360
	u64 flags;
3361 3362

	do_barriers = !btrfs_test_opt(root, NOBARRIER);
C
Chris Mason 已提交
3363
	backup_super_roots(root->fs_info);
3364

3365
	sb = root->fs_info->super_for_commit;
3366
	dev_item = &sb->dev_item;
3367

3368
	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
3369
	head = &root->fs_info->fs_devices->devices;
3370
	max_errors = btrfs_super_num_devices(root->fs_info->super_copy) - 1;
C
Chris Mason 已提交
3371

3372 3373 3374 3375 3376 3377 3378 3379 3380 3381
	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
3382

3383
	list_for_each_entry_rcu(dev, head, dev_list) {
3384 3385 3386 3387
		if (!dev->bdev) {
			total_errors++;
			continue;
		}
Y
Yan Zheng 已提交
3388
		if (!dev->in_fs_metadata || !dev->writeable)
3389 3390
			continue;

Y
Yan Zheng 已提交
3391
		btrfs_set_stack_device_generation(dev_item, 0);
3392 3393 3394 3395 3396 3397 3398 3399
		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
3400
		memcpy(dev_item->fsid, dev->fs_devices->fsid, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3401

3402 3403 3404
		flags = btrfs_super_flags(sb);
		btrfs_set_super_flags(sb, flags | BTRFS_HEADER_FLAG_WRITTEN);

Y
Yan Zheng 已提交
3405
		ret = write_dev_supers(dev, sb, do_barriers, 0, max_mirrors);
3406 3407
		if (ret)
			total_errors++;
3408
	}
3409
	if (total_errors > max_errors) {
C
Chris Mason 已提交
3410 3411
		printk(KERN_ERR "btrfs: %d errors while writing supers\n",
		       total_errors);
3412
		mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
3413

3414 3415 3416 3417
		/* FUA is masked off if unsupported and can't be the reason */
		btrfs_error(root->fs_info, -EIO,
			    "%d errors while writing supers", total_errors);
		return -EIO;
3418
	}
3419

Y
Yan Zheng 已提交
3420
	total_errors = 0;
3421
	list_for_each_entry_rcu(dev, head, dev_list) {
3422 3423
		if (!dev->bdev)
			continue;
Y
Yan Zheng 已提交
3424
		if (!dev->in_fs_metadata || !dev->writeable)
3425 3426
			continue;

Y
Yan Zheng 已提交
3427 3428 3429
		ret = write_dev_supers(dev, sb, do_barriers, 1, max_mirrors);
		if (ret)
			total_errors++;
3430
	}
3431
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
3432
	if (total_errors > max_errors) {
3433 3434 3435
		btrfs_error(root->fs_info, -EIO,
			    "%d errors while writing supers", total_errors);
		return -EIO;
3436
	}
3437 3438 3439
	return 0;
}

Y
Yan Zheng 已提交
3440 3441
int write_ctree_super(struct btrfs_trans_handle *trans,
		      struct btrfs_root *root, int max_mirrors)
3442
{
3443
	return write_all_supers(root, max_mirrors);
C
Chris Mason 已提交
3444 3445
}

3446 3447 3448
/* Drop a fs root from the radix tree and free it. */
void btrfs_drop_and_free_fs_root(struct btrfs_fs_info *fs_info,
				  struct btrfs_root *root)
C
Chris Mason 已提交
3449
{
3450
	spin_lock(&fs_info->fs_roots_radix_lock);
C
Chris Mason 已提交
3451 3452
	radix_tree_delete(&fs_info->fs_roots_radix,
			  (unsigned long)root->root_key.objectid);
3453
	spin_unlock(&fs_info->fs_roots_radix_lock);
3454 3455 3456 3457

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

L
Liu Bo 已提交
3458
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
L
Liu Bo 已提交
3459 3460 3461 3462
		btrfs_free_log(NULL, root);
		btrfs_free_log_root_tree(NULL, fs_info);
	}

3463 3464
	__btrfs_remove_free_space_cache(root->free_ino_pinned);
	__btrfs_remove_free_space_cache(root->free_ino_ctl);
3465 3466 3467 3468 3469
	free_fs_root(root);
}

static void free_fs_root(struct btrfs_root *root)
{
3470
	iput(root->cache_inode);
3471
	WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree));
3472 3473
	btrfs_free_block_rsv(root, root->orphan_block_rsv);
	root->orphan_block_rsv = NULL;
3474 3475
	if (root->anon_dev)
		free_anon_bdev(root->anon_dev);
3476 3477
	free_extent_buffer(root->node);
	free_extent_buffer(root->commit_root);
3478 3479
	kfree(root->free_ino_ctl);
	kfree(root->free_ino_pinned);
C
Chris Mason 已提交
3480
	kfree(root->name);
3481
	btrfs_put_fs_root(root);
C
Chris Mason 已提交
3482 3483
}

3484 3485 3486
void btrfs_free_fs_root(struct btrfs_root *root)
{
	free_fs_root(root);
C
Chris Mason 已提交
3487 3488
}

Y
Yan Zheng 已提交
3489
int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
C
Chris Mason 已提交
3490
{
Y
Yan Zheng 已提交
3491 3492 3493
	u64 root_objectid = 0;
	struct btrfs_root *gang[8];
	int i;
3494
	int ret;
3495

Y
Yan Zheng 已提交
3496 3497 3498 3499 3500 3501
	while (1) {
		ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix,
					     (void **)gang, root_objectid,
					     ARRAY_SIZE(gang));
		if (!ret)
			break;
3502 3503

		root_objectid = gang[ret - 1]->root_key.objectid + 1;
Y
Yan Zheng 已提交
3504
		for (i = 0; i < ret; i++) {
3505 3506
			int err;

Y
Yan Zheng 已提交
3507
			root_objectid = gang[i]->root_key.objectid;
3508 3509 3510
			err = btrfs_orphan_cleanup(gang[i]);
			if (err)
				return err;
Y
Yan Zheng 已提交
3511 3512 3513 3514 3515
		}
		root_objectid++;
	}
	return 0;
}
3516

Y
Yan Zheng 已提交
3517 3518 3519 3520
int btrfs_commit_super(struct btrfs_root *root)
{
	struct btrfs_trans_handle *trans;
	int ret;
3521

Y
Yan Zheng 已提交
3522
	mutex_lock(&root->fs_info->cleaner_mutex);
Y
Yan, Zheng 已提交
3523
	btrfs_run_delayed_iputs(root);
Y
Yan Zheng 已提交
3524
	mutex_unlock(&root->fs_info->cleaner_mutex);
D
David Sterba 已提交
3525
	wake_up_process(root->fs_info->cleaner_kthread);
3526 3527 3528 3529 3530

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

3531
	trans = btrfs_join_transaction(root);
3532 3533
	if (IS_ERR(trans))
		return PTR_ERR(trans);
3534
	ret = btrfs_commit_transaction(trans, root);
3535 3536
	if (ret)
		return ret;
Y
Yan Zheng 已提交
3537
	/* run commit again to drop the original snapshot */
3538
	trans = btrfs_join_transaction(root);
3539 3540
	if (IS_ERR(trans))
		return PTR_ERR(trans);
3541 3542 3543
	ret = btrfs_commit_transaction(trans, root);
	if (ret)
		return ret;
C
Chris Mason 已提交
3544
	ret = btrfs_write_and_wait_transaction(NULL, root);
3545 3546 3547 3548 3549
	if (ret) {
		btrfs_error(root->fs_info, ret,
			    "Failed to sync btree inode to disk.");
		return ret;
	}
3550

Y
Yan Zheng 已提交
3551
	ret = write_ctree_super(NULL, root, 0);
Y
Yan Zheng 已提交
3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562
	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();

S
Stefan Behrens 已提交
3563 3564 3565 3566 3567
	/* wait for the uuid_scan task to finish */
	down(&fs_info->uuid_tree_rescan_sem);
	/* avoid complains from lockdep et al., set sem back to initial state */
	up(&fs_info->uuid_tree_rescan_sem);

3568
	/* pause restriper - we want to resume on mount */
3569
	btrfs_pause_balance(fs_info);
3570

3571 3572
	btrfs_dev_replace_suspend_for_unmount(fs_info);

3573
	btrfs_scrub_cancel(fs_info);
C
Chris Mason 已提交
3574 3575 3576 3577 3578 3579

	/* 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 */
3580
	btrfs_cleanup_defrag_inodes(fs_info);
C
Chris Mason 已提交
3581

Y
Yan Zheng 已提交
3582
	if (!(fs_info->sb->s_flags & MS_RDONLY)) {
L
liubo 已提交
3583 3584 3585 3586 3587
		ret = btrfs_commit_super(root);
		if (ret)
			printk(KERN_ERR "btrfs: commit super ret %d\n", ret);
	}

3588
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state))
3589
		btrfs_error_commit_super(root);
3590

3591 3592
	btrfs_put_block_group_cache(fs_info);

A
Al Viro 已提交
3593 3594
	kthread_stop(fs_info->transaction_kthread);
	kthread_stop(fs_info->cleaner_kthread);
3595

3596 3597 3598
	fs_info->closing = 2;
	smp_mb();

3599 3600
	btrfs_free_qgroup_config(root->fs_info);

3601 3602 3603
	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
3604
	}
3605

Y
Yan Zheng 已提交
3606
	del_fs_roots(fs_info);
3607

3608 3609
	btrfs_free_block_groups(fs_info);

3610 3611
	btrfs_stop_all_workers(fs_info);

3612
	free_root_pointers(fs_info, 1);
3613

3614
	iput(fs_info->btree_inode);
3615

3616 3617 3618 3619 3620
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	if (btrfs_test_opt(root, CHECK_INTEGRITY))
		btrfsic_unmount(root, fs_info->fs_devices);
#endif

3621
	btrfs_close_devices(fs_info->fs_devices);
3622
	btrfs_mapping_tree_free(&fs_info->mapping_tree);
3623

3624
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
3625
	percpu_counter_destroy(&fs_info->delalloc_bytes);
C
Chris Mason 已提交
3626
	bdi_destroy(&fs_info->bdi);
3627
	cleanup_srcu_struct(&fs_info->subvol_srcu);
3628

D
David Woodhouse 已提交
3629 3630
	btrfs_free_stripe_hash_table(fs_info);

3631 3632 3633
	btrfs_free_block_rsv(root, root->orphan_block_rsv);
	root->orphan_block_rsv = NULL;

3634 3635 3636
	return 0;
}

3637 3638
int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid,
			  int atomic)
3639
{
3640
	int ret;
3641
	struct inode *btree_inode = buf->pages[0]->mapping->host;
3642

3643
	ret = extent_buffer_uptodate(buf);
3644 3645 3646 3647
	if (!ret)
		return ret;

	ret = verify_parent_transid(&BTRFS_I(btree_inode)->io_tree, buf,
3648 3649 3650
				    parent_transid, atomic);
	if (ret == -EAGAIN)
		return ret;
3651
	return !ret;
3652 3653 3654
}

int btrfs_set_buffer_uptodate(struct extent_buffer *buf)
C
Chris Mason 已提交
3655
{
3656
	return set_extent_buffer_uptodate(buf);
3657
}
3658

3659 3660
void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
{
3661
	struct btrfs_root *root;
3662
	u64 transid = btrfs_header_generation(buf);
3663
	int was_dirty;
3664

3665 3666 3667 3668 3669 3670 3671 3672 3673 3674
#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
	/*
	 * This is a fast path so only do this check if we have sanity tests
	 * enabled.  Normal people shouldn't be marking dummy buffers as dirty
	 * outside of the sanity tests.
	 */
	if (unlikely(test_bit(EXTENT_BUFFER_DUMMY, &buf->bflags)))
		return;
#endif
	root = BTRFS_I(buf->pages[0]->mapping->host)->root;
3675
	btrfs_assert_tree_locked(buf);
J
Julia Lawall 已提交
3676 3677
	if (transid != root->fs_info->generation)
		WARN(1, KERN_CRIT "btrfs transid mismatch buffer %llu, "
C
Chris Mason 已提交
3678
		       "found %llu running %llu\n",
3679
			buf->start, transid, root->fs_info->generation);
3680
	was_dirty = set_extent_buffer_dirty(buf);
3681 3682 3683 3684
	if (!was_dirty)
		__percpu_counter_add(&root->fs_info->dirty_metadata_bytes,
				     buf->len,
				     root->fs_info->dirty_metadata_batch);
3685 3686
}

3687 3688
static void __btrfs_btree_balance_dirty(struct btrfs_root *root,
					int flush_delayed)
3689 3690 3691 3692 3693
{
	/*
	 * looks as though older kernels can get into trouble with
	 * this code, they end up stuck in balance_dirty_pages forever
	 */
3694
	int ret;
3695 3696 3697 3698

	if (current->flags & PF_MEMALLOC)
		return;

3699 3700
	if (flush_delayed)
		btrfs_balance_delayed_items(root);
3701

3702 3703 3704
	ret = percpu_counter_compare(&root->fs_info->dirty_metadata_bytes,
				     BTRFS_DIRTY_METADATA_THRESH);
	if (ret > 0) {
3705 3706
		balance_dirty_pages_ratelimited(
				   root->fs_info->btree_inode->i_mapping);
3707 3708 3709 3710
	}
	return;
}

3711
void btrfs_btree_balance_dirty(struct btrfs_root *root)
C
Chris Mason 已提交
3712
{
3713 3714
	__btrfs_btree_balance_dirty(root, 1);
}
3715

3716 3717 3718
void btrfs_btree_balance_dirty_nodelay(struct btrfs_root *root)
{
	__btrfs_btree_balance_dirty(root, 0);
C
Chris Mason 已提交
3719
}
3720

3721
int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid)
3722
{
3723
	struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root;
3724
	return btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
3725
}
3726

3727
static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info,
L
liubo 已提交
3728 3729
			      int read_only)
{
D
David Sterba 已提交
3730 3731 3732
	/*
	 * Placeholder for checks
	 */
3733
	return 0;
L
liubo 已提交
3734 3735
}

3736
static void btrfs_error_commit_super(struct btrfs_root *root)
L
liubo 已提交
3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748
{
	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);
}

3749 3750
static void btrfs_destroy_ordered_operations(struct btrfs_transaction *t,
					     struct btrfs_root *root)
L
liubo 已提交
3751 3752 3753 3754 3755 3756 3757
{
	struct btrfs_inode *btrfs_inode;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

	mutex_lock(&root->fs_info->ordered_operations_mutex);
3758
	spin_lock(&root->fs_info->ordered_root_lock);
L
liubo 已提交
3759

3760
	list_splice_init(&t->ordered_operations, &splice);
L
liubo 已提交
3761 3762 3763 3764 3765
	while (!list_empty(&splice)) {
		btrfs_inode = list_entry(splice.next, struct btrfs_inode,
					 ordered_operations);

		list_del_init(&btrfs_inode->ordered_operations);
3766
		spin_unlock(&root->fs_info->ordered_root_lock);
L
liubo 已提交
3767 3768

		btrfs_invalidate_inodes(btrfs_inode->root);
3769

3770
		spin_lock(&root->fs_info->ordered_root_lock);
L
liubo 已提交
3771 3772
	}

3773
	spin_unlock(&root->fs_info->ordered_root_lock);
L
liubo 已提交
3774 3775 3776
	mutex_unlock(&root->fs_info->ordered_operations_mutex);
}

3777
static void btrfs_destroy_ordered_extents(struct btrfs_root *root)
L
liubo 已提交
3778 3779 3780
{
	struct btrfs_ordered_extent *ordered;

3781
	spin_lock(&root->ordered_extent_lock);
3782 3783 3784 3785
	/*
	 * This will just short circuit the ordered completion stuff which will
	 * make sure the ordered extent gets properly cleaned up.
	 */
3786
	list_for_each_entry(ordered, &root->ordered_extents,
3787 3788
			    root_extent_list)
		set_bit(BTRFS_ORDERED_IOERR, &ordered->flags);
3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803
	spin_unlock(&root->ordered_extent_lock);
}

static void btrfs_destroy_all_ordered_extents(struct btrfs_fs_info *fs_info)
{
	struct btrfs_root *root;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

	spin_lock(&fs_info->ordered_root_lock);
	list_splice_init(&fs_info->ordered_roots, &splice);
	while (!list_empty(&splice)) {
		root = list_first_entry(&splice, struct btrfs_root,
					ordered_root);
3804 3805
		list_move_tail(&root->ordered_root,
			       &fs_info->ordered_roots);
3806 3807 3808 3809 3810 3811

		btrfs_destroy_ordered_extents(root);

		cond_resched_lock(&fs_info->ordered_root_lock);
	}
	spin_unlock(&fs_info->ordered_root_lock);
L
liubo 已提交
3812 3813
}

3814 3815
static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
				      struct btrfs_root *root)
L
liubo 已提交
3816 3817 3818 3819 3820 3821 3822 3823 3824 3825
{
	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) {
3826
		spin_unlock(&delayed_refs->lock);
L
liubo 已提交
3827 3828 3829 3830
		printk(KERN_INFO "delayed_refs has NO entry\n");
		return ret;
	}

3831
	while ((node = rb_first(&delayed_refs->root)) != NULL) {
3832
		struct btrfs_delayed_ref_head *head = NULL;
3833
		bool pin_bytes = false;
L
liubo 已提交
3834

3835
		ref = rb_entry(node, struct btrfs_delayed_ref_node, rb_node);
L
liubo 已提交
3836 3837 3838 3839
		atomic_set(&ref->refs, 1);
		if (btrfs_delayed_ref_is_head(ref)) {

			head = btrfs_delayed_node_to_head(ref);
3840 3841 3842 3843 3844 3845 3846 3847 3848
			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
3849
				spin_lock(&delayed_refs->lock);
3850 3851 3852
				continue;
			}

J
Josef Bacik 已提交
3853
			if (head->must_insert_reserved)
3854
				pin_bytes = true;
3855
			btrfs_free_delayed_extent_op(head->extent_op);
L
liubo 已提交
3856 3857 3858 3859 3860
			delayed_refs->num_heads--;
			if (list_empty(&head->cluster))
				delayed_refs->num_heads_ready--;
			list_del_init(&head->cluster);
		}
3861

3862 3863 3864
		ref->in_tree = 0;
		rb_erase(&ref->rb_node, &delayed_refs->root);
		delayed_refs->num_entries--;
L
liubo 已提交
3865
		spin_unlock(&delayed_refs->lock);
3866 3867 3868 3869 3870 3871
		if (head) {
			if (pin_bytes)
				btrfs_pin_extent(root, ref->bytenr,
						 ref->num_bytes, 1);
			mutex_unlock(&head->mutex);
		}
L
liubo 已提交
3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882
		btrfs_put_delayed_ref(ref);

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

	spin_unlock(&delayed_refs->lock);

	return ret;
}

3883
static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root)
L
liubo 已提交
3884 3885 3886 3887 3888 3889
{
	struct btrfs_inode *btrfs_inode;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

3890 3891
	spin_lock(&root->delalloc_lock);
	list_splice_init(&root->delalloc_inodes, &splice);
L
liubo 已提交
3892 3893

	while (!list_empty(&splice)) {
3894 3895
		btrfs_inode = list_first_entry(&splice, struct btrfs_inode,
					       delalloc_inodes);
L
liubo 已提交
3896 3897

		list_del_init(&btrfs_inode->delalloc_inodes);
3898 3899
		clear_bit(BTRFS_INODE_IN_DELALLOC_LIST,
			  &btrfs_inode->runtime_flags);
3900
		spin_unlock(&root->delalloc_lock);
L
liubo 已提交
3901 3902

		btrfs_invalidate_inodes(btrfs_inode->root);
3903

3904
		spin_lock(&root->delalloc_lock);
L
liubo 已提交
3905 3906
	}

3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932
	spin_unlock(&root->delalloc_lock);
}

static void btrfs_destroy_all_delalloc_inodes(struct btrfs_fs_info *fs_info)
{
	struct btrfs_root *root;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

	spin_lock(&fs_info->delalloc_root_lock);
	list_splice_init(&fs_info->delalloc_roots, &splice);
	while (!list_empty(&splice)) {
		root = list_first_entry(&splice, struct btrfs_root,
					 delalloc_root);
		list_del_init(&root->delalloc_root);
		root = btrfs_grab_fs_root(root);
		BUG_ON(!root);
		spin_unlock(&fs_info->delalloc_root_lock);

		btrfs_destroy_delalloc_inodes(root);
		btrfs_put_fs_root(root);

		spin_lock(&fs_info->delalloc_root_lock);
	}
	spin_unlock(&fs_info->delalloc_root_lock);
L
liubo 已提交
3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945
}

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,
3946
					    mark, NULL);
L
liubo 已提交
3947 3948 3949 3950 3951
		if (ret)
			break;

		clear_extent_bits(dirty_pages, start, end, mark, GFP_NOFS);
		while (start <= end) {
3952 3953
			eb = btrfs_find_tree_block(root, start,
						   root->leafsize);
3954
			start += root->leafsize;
3955
			if (!eb)
L
liubo 已提交
3956
				continue;
3957
			wait_on_extent_buffer_writeback(eb);
L
liubo 已提交
3958

3959 3960 3961 3962
			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
3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975
		}
	}

	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;
3976
	bool loop = true;
L
liubo 已提交
3977 3978

	unpin = pinned_extents;
3979
again:
L
liubo 已提交
3980 3981
	while (1) {
		ret = find_first_extent_bit(unpin, 0, &start, &end,
3982
					    EXTENT_DIRTY, NULL);
L
liubo 已提交
3983 3984 3985 3986
		if (ret)
			break;

		/* opt_discard */
3987 3988 3989 3990
		if (btrfs_test_opt(root, DISCARD))
			ret = btrfs_error_discard_extent(root, start,
							 end + 1 - start,
							 NULL);
L
liubo 已提交
3991 3992 3993 3994 3995 3996

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

3997 3998 3999 4000 4001 4002 4003 4004 4005
	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
4006 4007 4008
	return 0;
}

4009 4010 4011
void btrfs_cleanup_one_transaction(struct btrfs_transaction *cur_trans,
				   struct btrfs_root *root)
{
4012 4013
	btrfs_destroy_ordered_operations(cur_trans, root);

4014 4015
	btrfs_destroy_delayed_refs(cur_trans, root);

4016
	cur_trans->state = TRANS_STATE_COMMIT_START;
4017
	wake_up(&root->fs_info->transaction_blocked_wait);
4018

4019
	cur_trans->state = TRANS_STATE_UNBLOCKED;
4020
	wake_up(&root->fs_info->transaction_wait);
4021

4022 4023
	btrfs_destroy_delayed_inodes(root);
	btrfs_assert_delayed_root_empty(root);
4024 4025 4026

	btrfs_destroy_marked_extents(root, &cur_trans->dirty_pages,
				     EXTENT_DIRTY);
4027 4028
	btrfs_destroy_pinned_extent(root,
				    root->fs_info->pinned_extents);
4029

4030 4031 4032
	cur_trans->state =TRANS_STATE_COMPLETED;
	wake_up(&cur_trans->commit_wait);

4033 4034 4035 4036 4037 4038
	/*
	memset(cur_trans, 0, sizeof(*cur_trans));
	kmem_cache_free(btrfs_transaction_cachep, cur_trans);
	*/
}

4039
static int btrfs_cleanup_transaction(struct btrfs_root *root)
L
liubo 已提交
4040 4041 4042 4043 4044
{
	struct btrfs_transaction *t;

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

J
Josef Bacik 已提交
4045
	spin_lock(&root->fs_info->trans_lock);
4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069
	while (!list_empty(&root->fs_info->trans_list)) {
		t = list_first_entry(&root->fs_info->trans_list,
				     struct btrfs_transaction, list);
		if (t->state >= TRANS_STATE_COMMIT_START) {
			atomic_inc(&t->use_count);
			spin_unlock(&root->fs_info->trans_lock);
			btrfs_wait_for_commit(root, t->transid);
			btrfs_put_transaction(t);
			spin_lock(&root->fs_info->trans_lock);
			continue;
		}
		if (t == root->fs_info->running_transaction) {
			t->state = TRANS_STATE_COMMIT_DOING;
			spin_unlock(&root->fs_info->trans_lock);
			/*
			 * We wait for 0 num_writers since we don't hold a trans
			 * handle open currently for this transaction.
			 */
			wait_event(t->writer_wait,
				   atomic_read(&t->num_writers) == 0);
		} else {
			spin_unlock(&root->fs_info->trans_lock);
		}
		btrfs_cleanup_one_transaction(t, root);
4070

4071 4072 4073
		spin_lock(&root->fs_info->trans_lock);
		if (t == root->fs_info->running_transaction)
			root->fs_info->running_transaction = NULL;
L
liubo 已提交
4074
		list_del_init(&t->list);
4075
		spin_unlock(&root->fs_info->trans_lock);
L
liubo 已提交
4076

4077 4078 4079 4080 4081 4082 4083 4084 4085 4086
		btrfs_put_transaction(t);
		trace_btrfs_transaction_commit(root);
		spin_lock(&root->fs_info->trans_lock);
	}
	spin_unlock(&root->fs_info->trans_lock);
	btrfs_destroy_all_ordered_extents(root->fs_info);
	btrfs_destroy_delayed_inodes(root);
	btrfs_assert_delayed_root_empty(root);
	btrfs_destroy_pinned_extent(root, root->fs_info->pinned_extents);
	btrfs_destroy_all_delalloc_inodes(root->fs_info);
L
liubo 已提交
4087 4088 4089 4090 4091
	mutex_unlock(&root->fs_info->transaction_kthread_mutex);

	return 0;
}

4092
static struct extent_io_ops btree_extent_io_ops = {
4093
	.readpage_end_io_hook = btree_readpage_end_io_hook,
A
Arne Jansen 已提交
4094
	.readpage_io_failed_hook = btree_io_failed_hook,
4095
	.submit_bio_hook = btree_submit_bio_hook,
4096 4097
	/* note we're sharing with inode.c for the merge bio hook */
	.merge_bio_hook = btrfs_merge_bio_hook,
4098
};