disk-io.c 114.9 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/slab.h>
30
#include <linux/migrate.h>
31
#include <linux/ratelimit.h>
32
#include <linux/uuid.h>
S
Stefan Behrens 已提交
33
#include <linux/semaphore.h>
34
#include <asm/unaligned.h>
35 36
#include "ctree.h"
#include "disk-io.h"
37
#include "hash.h"
38
#include "transaction.h"
39
#include "btrfs_inode.h"
40
#include "volumes.h"
41
#include "print-tree.h"
42
#include "locking.h"
43
#include "tree-log.h"
44
#include "free-space-cache.h"
45
#include "inode-map.h"
46
#include "check-integrity.h"
47
#include "rcu-string.h"
48
#include "dev-replace.h"
D
David Woodhouse 已提交
49
#include "raid56.h"
50
#include "sysfs.h"
J
Josef Bacik 已提交
51
#include "qgroup.h"
52

53 54 55 56
#ifdef CONFIG_X86
#include <asm/cpufeature.h>
#endif

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

C
Chris Mason 已提交
74
/*
75 76
 * btrfs_end_io_wq structs are used to do processing in task context when an IO
 * is complete.  This is used during reads to verify checksums, and it is used
C
Chris Mason 已提交
77 78
 * by writes to insert metadata for new file extents after IO is complete.
 */
79
struct btrfs_end_io_wq {
80 81 82 83 84
	struct bio *bio;
	bio_end_io_t *end_io;
	void *private;
	struct btrfs_fs_info *info;
	int error;
85
	enum btrfs_wq_endio_type metadata;
86
	struct list_head list;
87
	struct btrfs_work work;
88
};
89

90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109
static struct kmem_cache *btrfs_end_io_wq_cache;

int __init btrfs_end_io_wq_init(void)
{
	btrfs_end_io_wq_cache = kmem_cache_create("btrfs_end_io_wq",
					sizeof(struct btrfs_end_io_wq),
					0,
					SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD,
					NULL);
	if (!btrfs_end_io_wq_cache)
		return -ENOMEM;
	return 0;
}

void btrfs_end_io_wq_exit(void)
{
	if (btrfs_end_io_wq_cache)
		kmem_cache_destroy(btrfs_end_io_wq_cache);
}

C
Chris Mason 已提交
110 111 112 113 114
/*
 * 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.
 */
115 116 117 118
struct async_submit_bio {
	struct inode *inode;
	struct bio *bio;
	struct list_head list;
C
Chris Mason 已提交
119 120
	extent_submit_bio_hook_t *submit_bio_start;
	extent_submit_bio_hook_t *submit_bio_done;
121 122
	int rw;
	int mirror_num;
C
Chris Mason 已提交
123
	unsigned long bio_flags;
124 125 126 127 128
	/*
	 * 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;
129
	struct btrfs_work work;
130
	int error;
131 132
};

133 134 135 136 137 138 139 140 141 142
/*
 * 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.
143
 *
144 145 146
 * 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.
147
 *
148 149 150
 * 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.
151
 *
152 153 154
 * 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.
155 156 157 158 159
 */
#ifdef CONFIG_DEBUG_LOCK_ALLOC
# if BTRFS_MAX_LEVEL != 8
#  error
# endif
160 161 162 163 164 165 166 167 168 169 170 171 172

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"	},
173
	{ .id = BTRFS_QUOTA_TREE_OBJECTID,	.name_stem = "quota"	},
174 175 176
	{ .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"	},
177
	{ .id = BTRFS_UUID_TREE_OBJECTID,	.name_stem = "uuid"	},
178
	{ .id = 0,				.name_stem = "tree"	},
179
};
180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210

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]);
}

211 212
#endif

C
Chris Mason 已提交
213 214 215 216
/*
 * extents on the btree inode are pretty simple, there's one extent
 * that covers the entire device
 */
217
static struct extent_map *btree_get_extent(struct inode *inode,
218
		struct page *page, size_t pg_offset, u64 start, u64 len,
219
		int create)
220
{
221 222 223 224
	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
	struct extent_map *em;
	int ret;

225
	read_lock(&em_tree->lock);
226
	em = lookup_extent_mapping(em_tree, start, len);
227 228 229
	if (em) {
		em->bdev =
			BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
230
		read_unlock(&em_tree->lock);
231
		goto out;
232
	}
233
	read_unlock(&em_tree->lock);
234

235
	em = alloc_extent_map();
236 237 238 239 240
	if (!em) {
		em = ERR_PTR(-ENOMEM);
		goto out;
	}
	em->start = 0;
241
	em->len = (u64)-1;
C
Chris Mason 已提交
242
	em->block_len = (u64)-1;
243
	em->block_start = 0;
244
	em->bdev = BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
245

246
	write_lock(&em_tree->lock);
J
Josef Bacik 已提交
247
	ret = add_extent_mapping(em_tree, em, 0);
248 249
	if (ret == -EEXIST) {
		free_extent_map(em);
250
		em = lookup_extent_mapping(em_tree, start, len);
251
		if (!em)
252
			em = ERR_PTR(-EIO);
253
	} else if (ret) {
254
		free_extent_map(em);
255
		em = ERR_PTR(ret);
256
	}
257
	write_unlock(&em_tree->lock);
258

259 260
out:
	return em;
261 262
}

263
u32 btrfs_csum_data(char *data, u32 seed, size_t len)
264
{
265
	return btrfs_crc32c(seed, data, len);
266 267 268 269
}

void btrfs_csum_final(u32 crc, char *result)
{
270
	put_unaligned_le32(~crc, result);
271 272
}

C
Chris Mason 已提交
273 274 275 276
/*
 * compute the csum for a btree block, and either verify it or write it
 * into the csum field of the block.
 */
277 278 279
static int csum_tree_block(struct btrfs_root *root, struct extent_buffer *buf,
			   int verify)
{
280
	u16 csum_size = btrfs_super_csum_size(root->fs_info->super_copy);
281
	char *result = NULL;
282 283 284 285 286 287 288 289
	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;
290
	unsigned long inline_result;
291 292

	len = buf->len - offset;
C
Chris Mason 已提交
293
	while (len > 0) {
294
		err = map_private_extent_buffer(buf, offset, 32,
295
					&kaddr, &map_start, &map_len);
C
Chris Mason 已提交
296
		if (err)
297 298
			return 1;
		cur_len = min(len, map_len - (offset - map_start));
299
		crc = btrfs_csum_data(kaddr + offset - map_start,
300 301 302 303
				      crc, cur_len);
		len -= cur_len;
		offset += cur_len;
	}
304 305 306 307 308 309 310 311
	if (csum_size > sizeof(inline_result)) {
		result = kzalloc(csum_size * sizeof(char), GFP_NOFS);
		if (!result)
			return 1;
	} else {
		result = (char *)&inline_result;
	}

312 313 314
	btrfs_csum_final(crc, result);

	if (verify) {
315
		if (memcmp_extent_buffer(buf, result, 0, csum_size)) {
316 317
			u32 val;
			u32 found = 0;
318
			memcpy(&found, result, csum_size);
319

320
			read_extent_buffer(buf, &val, 0, csum_size);
321 322 323 324 325
			printk_ratelimited(KERN_INFO
				"BTRFS: %s checksum verify failed on %llu wanted %X found %X "
				"level %d\n",
				root->fs_info->sb->s_id, buf->start,
				val, found, btrfs_header_level(buf));
326 327
			if (result != (char *)&inline_result)
				kfree(result);
328 329 330
			return 1;
		}
	} else {
331
		write_extent_buffer(buf, result, 0, csum_size);
332
	}
333 334
	if (result != (char *)&inline_result)
		kfree(result);
335 336 337
	return 0;
}

C
Chris Mason 已提交
338 339 340 341 342 343
/*
 * 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.
 */
344
static int verify_parent_transid(struct extent_io_tree *io_tree,
345 346
				 struct extent_buffer *eb, u64 parent_transid,
				 int atomic)
347
{
348
	struct extent_state *cached_state = NULL;
349
	int ret;
350
	bool need_lock = (current->journal_info == BTRFS_SEND_TRANS_STUB);
351 352 353 354

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

355 356 357
	if (atomic)
		return -EAGAIN;

358 359 360 361 362
	if (need_lock) {
		btrfs_tree_read_lock(eb);
		btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
	}

363
	lock_extent_bits(io_tree, eb->start, eb->start + eb->len - 1,
364
			 0, &cached_state);
365
	if (extent_buffer_uptodate(eb) &&
366 367 368 369
	    btrfs_header_generation(eb) == parent_transid) {
		ret = 0;
		goto out;
	}
370 371 372
	printk_ratelimited(KERN_INFO "BTRFS (device %s): parent transid verify failed on %llu wanted %llu found %llu\n",
			eb->fs_info->sb->s_id, eb->start,
			parent_transid, btrfs_header_generation(eb));
373
	ret = 1;
374 375 376 377 378 379 380 381 382 383 384

	/*
	 * Things reading via commit roots that don't have normal protection,
	 * like send, can have a really old block in cache that may point at a
	 * block that has been free'd and re-allocated.  So don't clear uptodate
	 * if we find an eb that is under IO (dirty/writeback) because we could
	 * end up reading in the stale data and then writing it back out and
	 * making everybody very sad.
	 */
	if (!extent_buffer_under_io(eb))
		clear_extent_buffer_uptodate(eb);
C
Chris Mason 已提交
385
out:
386 387
	unlock_extent_cached(io_tree, eb->start, eb->start + eb->len - 1,
			     &cached_state, GFP_NOFS);
388 389
	if (need_lock)
		btrfs_tree_read_unlock_blocking(eb);
390 391 392
	return ret;
}

D
David Sterba 已提交
393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419
/*
 * Return 0 if the superblock checksum type matches the checksum value of that
 * algorithm. Pass the raw disk superblock data.
 */
static int btrfs_check_super_csum(char *raw_disk_sb)
{
	struct btrfs_super_block *disk_sb =
		(struct btrfs_super_block *)raw_disk_sb;
	u16 csum_type = btrfs_super_csum_type(disk_sb);
	int ret = 0;

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

		/*
		 * The super_block structure does not span the whole
		 * BTRFS_SUPER_INFO_SIZE range, we expect that the unused space
		 * 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;
420 421

		if (ret && btrfs_super_generation(disk_sb) < 10) {
422 423
			printk(KERN_WARNING
				"BTRFS: super block crcs don't match, older mkfs detected\n");
424 425
			ret = 0;
		}
D
David Sterba 已提交
426 427 428
	}

	if (csum_type >= ARRAY_SIZE(btrfs_csum_sizes)) {
429
		printk(KERN_ERR "BTRFS: unsupported checksum algorithm %u\n",
D
David Sterba 已提交
430 431 432 433 434 435 436
				csum_type);
		ret = 1;
	}

	return ret;
}

C
Chris Mason 已提交
437 438 439 440
/*
 * helper to read a given tree block, doing retries as required when
 * the checksums don't match and we have alternate mirrors to try.
 */
441 442
static int btree_read_extent_buffer_pages(struct btrfs_root *root,
					  struct extent_buffer *eb,
443
					  u64 start, u64 parent_transid)
444 445
{
	struct extent_io_tree *io_tree;
446
	int failed = 0;
447 448 449
	int ret;
	int num_copies = 0;
	int mirror_num = 0;
450
	int failed_mirror = 0;
451

452
	clear_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
453 454
	io_tree = &BTRFS_I(root->fs_info->btree_inode)->io_tree;
	while (1) {
455 456
		ret = read_extent_buffer_pages(io_tree, eb, start,
					       WAIT_COMPLETE,
457
					       btree_get_extent, mirror_num);
458 459
		if (!ret) {
			if (!verify_parent_transid(io_tree, eb,
460
						   parent_transid, 0))
461 462 463 464
				break;
			else
				ret = -EIO;
		}
C
Chris Mason 已提交
465

466 467 468 469 470 471
		/*
		 * This buffer's crc is fine, but its contents are corrupted, so
		 * there is no reason to read the other copies, they won't be
		 * any less wrong.
		 */
		if (test_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags))
472 473
			break;

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

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

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

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

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

	return ret;
496
}
497

C
Chris Mason 已提交
498
/*
C
Chris Mason 已提交
499 500
 * checksum a dirty tree block before IO.  This has extra checks to make sure
 * we only fill in the checksum field in the first page of a multi-page block
C
Chris Mason 已提交
501
 */
C
Chris Mason 已提交
502

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

J
Josef Bacik 已提交
509 510 511
	eb = (struct extent_buffer *)page->private;
	if (page != eb->pages[0])
		return 0;
512
	found_start = btrfs_header_bytenr(eb);
513
	if (WARN_ON(found_start != start || !PageUptodate(page)))
J
Josef Bacik 已提交
514
		return 0;
515 516 517 518
	csum_tree_block(root, eb, 0);
	return 0;
}

Y
Yan Zheng 已提交
519 520 521 522 523 524 525
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;

526
	read_extent_buffer(eb, fsid, btrfs_header_fsid(), BTRFS_FSID_SIZE);
Y
Yan Zheng 已提交
527 528 529 530 531 532 533 534 535 536
	while (fs_devices) {
		if (!memcmp(fsid, fs_devices->fsid, BTRFS_FSID_SIZE)) {
			ret = 0;
			break;
		}
		fs_devices = fs_devices->seed;
	}
	return ret;
}

537
#define CORRUPT(reason, eb, root, slot)				\
538 539
	btrfs_crit(root->fs_info, "corrupt leaf, %s: block=%llu,"	\
		   "root=%llu, slot=%d", reason,			\
540
	       btrfs_header_bytenr(eb),	root->objectid, slot)
541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602

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

	if (nritems == 0)
		return 0;

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

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

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

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

		/*
		 * Check to make sure that we don't point outside of the leaf,
		 * 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;
}

603 604 605
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)
606 607 608 609 610
{
	u64 found_start;
	int found_level;
	struct extent_buffer *eb;
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
611
	int ret = 0;
612
	int reads_done;
613 614 615

	if (!page->private)
		goto out;
C
Chris Mason 已提交
616

J
Josef Bacik 已提交
617
	eb = (struct extent_buffer *)page->private;
C
Chris Mason 已提交
618

619 620 621 622 623 624
	/* 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);
625 626
	if (!reads_done)
		goto err;
627

628
	eb->read_mirror = mirror;
629
	if (test_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags)) {
630 631 632 633
		ret = -EIO;
		goto err;
	}

634
	found_start = btrfs_header_bytenr(eb);
635
	if (found_start != eb->start) {
636
		printk_ratelimited(KERN_INFO "BTRFS (device %s): bad tree block start "
C
Chris Mason 已提交
637
			       "%llu %llu\n",
638
			       eb->fs_info->sb->s_id, found_start, eb->start);
639
		ret = -EIO;
640 641
		goto err;
	}
Y
Yan Zheng 已提交
642
	if (check_tree_block_fsid(root, eb)) {
643 644
		printk_ratelimited(KERN_INFO "BTRFS (device %s): bad fsid on block %llu\n",
			       eb->fs_info->sb->s_id, eb->start);
645 646 647
		ret = -EIO;
		goto err;
	}
648
	found_level = btrfs_header_level(eb);
649
	if (found_level >= BTRFS_MAX_LEVEL) {
650
		btrfs_info(root->fs_info, "bad tree block level %d",
651 652 653 654
			   (int)btrfs_header_level(eb));
		ret = -EIO;
		goto err;
	}
655

656 657
	btrfs_set_buffer_lockdep_class(btrfs_header_owner(eb),
				       eb, found_level);
658

659
	ret = csum_tree_block(root, eb, 1);
660
	if (ret) {
661
		ret = -EIO;
662 663 664 665 666 667 668 669 670 671 672 673
		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;
	}
674

675 676
	if (!ret)
		set_extent_buffer_uptodate(eb);
677
err:
678 679
	if (reads_done &&
	    test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
A
Arne Jansen 已提交
680 681
		btree_readahead_hook(root, eb, eb->start, ret);

D
David Woodhouse 已提交
682 683 684 685 686 687 688
	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);
689
		clear_extent_buffer_uptodate(eb);
D
David Woodhouse 已提交
690
	}
691
	free_extent_buffer(eb);
692
out:
693
	return ret;
694 695
}

696
static int btree_io_failed_hook(struct page *page, int failed_mirror)
A
Arne Jansen 已提交
697 698 699 700
{
	struct extent_buffer *eb;
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;

J
Josef Bacik 已提交
701
	eb = (struct extent_buffer *)page->private;
702
	set_bit(EXTENT_BUFFER_READ_ERR, &eb->bflags);
703
	eb->read_mirror = failed_mirror;
D
David Woodhouse 已提交
704
	atomic_dec(&eb->io_pages);
705
	if (test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
A
Arne Jansen 已提交
706 707 708 709
		btree_readahead_hook(root, eb, eb->start, -EIO);
	return -EIO;	/* we fixed nothing */
}

710 711
static void end_workqueue_bio(struct bio *bio, int err)
{
712
	struct btrfs_end_io_wq *end_io_wq = bio->bi_private;
713
	struct btrfs_fs_info *fs_info;
714 715
	struct btrfs_workqueue *wq;
	btrfs_work_func_t func;
716 717 718

	fs_info = end_io_wq->info;
	end_io_wq->error = err;
719

720
	if (bio->bi_rw & REQ_WRITE) {
721 722 723 724 725 726 727 728 729 730 731 732 733
		if (end_io_wq->metadata == BTRFS_WQ_ENDIO_METADATA) {
			wq = fs_info->endio_meta_write_workers;
			func = btrfs_endio_meta_write_helper;
		} else if (end_io_wq->metadata == BTRFS_WQ_ENDIO_FREE_SPACE) {
			wq = fs_info->endio_freespace_worker;
			func = btrfs_freespace_write_helper;
		} else if (end_io_wq->metadata == BTRFS_WQ_ENDIO_RAID56) {
			wq = fs_info->endio_raid56_workers;
			func = btrfs_endio_raid56_helper;
		} else {
			wq = fs_info->endio_write_workers;
			func = btrfs_endio_write_helper;
		}
734
	} else {
735 736 737 738 739
		if (unlikely(end_io_wq->metadata ==
			     BTRFS_WQ_ENDIO_DIO_REPAIR)) {
			wq = fs_info->endio_repair_workers;
			func = btrfs_endio_repair_helper;
		} else if (end_io_wq->metadata == BTRFS_WQ_ENDIO_RAID56) {
740 741 742 743 744 745 746 747 748
			wq = fs_info->endio_raid56_workers;
			func = btrfs_endio_raid56_helper;
		} else if (end_io_wq->metadata) {
			wq = fs_info->endio_meta_workers;
			func = btrfs_endio_meta_helper;
		} else {
			wq = fs_info->endio_workers;
			func = btrfs_endio_helper;
		}
749
	}
750 751 752

	btrfs_init_work(&end_io_wq->work, func, end_workqueue_fn, NULL, NULL);
	btrfs_queue_work(wq, &end_io_wq->work);
753 754
}

755
int btrfs_bio_wq_end_io(struct btrfs_fs_info *info, struct bio *bio,
756
			enum btrfs_wq_endio_type metadata)
757
{
758
	struct btrfs_end_io_wq *end_io_wq;
759

760
	end_io_wq = kmem_cache_alloc(btrfs_end_io_wq_cache, GFP_NOFS);
761 762 763 764 765
	if (!end_io_wq)
		return -ENOMEM;

	end_io_wq->private = bio->bi_private;
	end_io_wq->end_io = bio->bi_end_io;
766
	end_io_wq->info = info;
767 768
	end_io_wq->error = 0;
	end_io_wq->bio = bio;
769
	end_io_wq->metadata = metadata;
770 771 772

	bio->bi_private = end_io_wq;
	bio->bi_end_io = end_workqueue_bio;
773 774 775
	return 0;
}

776
unsigned long btrfs_async_submit_limit(struct btrfs_fs_info *info)
777
{
778
	unsigned long limit = min_t(unsigned long,
779
				    info->thread_pool_size,
780 781 782
				    info->fs_devices->open_devices);
	return 256 * limit;
}
783

C
Chris Mason 已提交
784 785 786
static void run_one_async_start(struct btrfs_work *work)
{
	struct async_submit_bio *async;
787
	int ret;
C
Chris Mason 已提交
788 789

	async = container_of(work, struct  async_submit_bio, work);
790 791 792 793 794
	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
795 796 797
}

static void run_one_async_done(struct btrfs_work *work)
798 799 800
{
	struct btrfs_fs_info *fs_info;
	struct async_submit_bio *async;
801
	int limit;
802 803 804

	async = container_of(work, struct  async_submit_bio, work);
	fs_info = BTRFS_I(async->inode)->root->fs_info;
805

806
	limit = btrfs_async_submit_limit(fs_info);
807 808
	limit = limit * 2 / 3;

809
	if (atomic_dec_return(&fs_info->nr_async_submits) < limit &&
810
	    waitqueue_active(&fs_info->async_submit_wait))
811 812
		wake_up(&fs_info->async_submit_wait);

813 814 815 816 817 818
	/* 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
819
	async->submit_bio_done(async->inode, async->rw, async->bio,
820 821
			       async->mirror_num, async->bio_flags,
			       async->bio_offset);
C
Chris Mason 已提交
822 823 824 825 826 827 828
}

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

	async = container_of(work, struct  async_submit_bio, work);
829 830 831
	kfree(async);
}

832 833
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
834
			unsigned long bio_flags,
835
			u64 bio_offset,
C
Chris Mason 已提交
836 837
			extent_submit_bio_hook_t *submit_bio_start,
			extent_submit_bio_hook_t *submit_bio_done)
838 839 840 841 842 843 844 845 846 847 848
{
	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
849 850 851
	async->submit_bio_start = submit_bio_start;
	async->submit_bio_done = submit_bio_done;

852
	btrfs_init_work(&async->work, btrfs_worker_helper, run_one_async_start,
853
			run_one_async_done, run_one_async_free);
C
Chris Mason 已提交
854

C
Chris Mason 已提交
855
	async->bio_flags = bio_flags;
856
	async->bio_offset = bio_offset;
857

858 859
	async->error = 0;

860
	atomic_inc(&fs_info->nr_async_submits);
861

862
	if (rw & REQ_SYNC)
863
		btrfs_set_work_high_priority(&async->work);
864

865
	btrfs_queue_work(fs_info->workers, &async->work);
866

C
Chris Mason 已提交
867
	while (atomic_read(&fs_info->async_submit_draining) &&
868 869 870 871 872
	      atomic_read(&fs_info->nr_async_submits)) {
		wait_event(fs_info->async_submit_wait,
			   (atomic_read(&fs_info->nr_async_submits) == 0));
	}

873 874 875
	return 0;
}

876 877
static int btree_csum_one_bio(struct bio *bio)
{
878
	struct bio_vec *bvec;
879
	struct btrfs_root *root;
880
	int i, ret = 0;
881

882
	bio_for_each_segment_all(bvec, bio, i) {
883
		root = BTRFS_I(bvec->bv_page->mapping->host)->root;
884 885 886
		ret = csum_dirty_buffer(root, bvec->bv_page);
		if (ret)
			break;
887
	}
888

889
	return ret;
890 891
}

C
Chris Mason 已提交
892 893
static int __btree_submit_bio_start(struct inode *inode, int rw,
				    struct bio *bio, int mirror_num,
894 895
				    unsigned long bio_flags,
				    u64 bio_offset)
896
{
897 898
	/*
	 * when we're called for a write, we're already in the async
899
	 * submission context.  Just jump into btrfs_map_bio
900
	 */
901
	return btree_csum_one_bio(bio);
C
Chris Mason 已提交
902
}
903

C
Chris Mason 已提交
904
static int __btree_submit_bio_done(struct inode *inode, int rw, struct bio *bio,
905 906
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
C
Chris Mason 已提交
907
{
908 909
	int ret;

910
	/*
C
Chris Mason 已提交
911 912
	 * when we're called for a write, we're already in the async
	 * submission context.  Just jump into btrfs_map_bio
913
	 */
914 915 916 917
	ret = btrfs_map_bio(BTRFS_I(inode)->root, rw, bio, mirror_num, 1);
	if (ret)
		bio_endio(bio, ret);
	return ret;
918 919
}

920 921 922 923 924 925 926 927 928 929 930
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;
}

931
static int btree_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
932 933
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
934
{
935
	int async = check_async_write(inode, bio_flags);
936 937
	int ret;

938
	if (!(rw & REQ_WRITE)) {
C
Chris Mason 已提交
939 940 941 942
		/*
		 * called for a read, do the setup so that checksum validation
		 * can happen in the async kernel threads
		 */
943
		ret = btrfs_bio_wq_end_io(BTRFS_I(inode)->root->fs_info,
944
					  bio, BTRFS_WQ_ENDIO_METADATA);
945
		if (ret)
946 947 948
			goto out_w_error;
		ret = btrfs_map_bio(BTRFS_I(inode)->root, rw, bio,
				    mirror_num, 0);
949 950 951
	} else if (!async) {
		ret = btree_csum_one_bio(bio);
		if (ret)
952 953 954 955 956 957 958 959 960 961 962 963 964
			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);
965
	}
966

967 968 969 970 971
	if (ret) {
out_w_error:
		bio_endio(bio, ret);
	}
	return ret;
972 973
}

J
Jan Beulich 已提交
974
#ifdef CONFIG_MIGRATION
975
static int btree_migratepage(struct address_space *mapping,
976 977
			struct page *newpage, struct page *page,
			enum migrate_mode mode)
978 979 980 981 982 983 984 985 986 987 988 989 990 991
{
	/*
	 * 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;
992
	return migrate_page(mapping, newpage, page, mode);
993
}
J
Jan Beulich 已提交
994
#endif
995

996 997 998 999

static int btree_writepages(struct address_space *mapping,
			    struct writeback_control *wbc)
{
1000 1001 1002
	struct btrfs_fs_info *fs_info;
	int ret;

1003
	if (wbc->sync_mode == WB_SYNC_NONE) {
1004 1005 1006 1007

		if (wbc->for_kupdate)
			return 0;

1008
		fs_info = BTRFS_I(mapping->host)->root->fs_info;
1009
		/* this is a bit racy, but that's ok */
1010 1011 1012
		ret = percpu_counter_compare(&fs_info->dirty_metadata_bytes,
					     BTRFS_DIRTY_METADATA_THRESH);
		if (ret < 0)
1013 1014
			return 0;
	}
1015
	return btree_write_cache_pages(mapping, wbc);
1016 1017
}

1018
static int btree_readpage(struct file *file, struct page *page)
1019
{
1020 1021
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
1022
	return extent_read_full_page(tree, page, btree_get_extent, 0);
1023
}
C
Chris Mason 已提交
1024

1025
static int btree_releasepage(struct page *page, gfp_t gfp_flags)
1026
{
1027
	if (PageWriteback(page) || PageDirty(page))
C
Chris Mason 已提交
1028
		return 0;
1029

1030
	return try_release_extent_buffer(page);
1031 1032
}

1033 1034
static void btree_invalidatepage(struct page *page, unsigned int offset,
				 unsigned int length)
1035
{
1036 1037
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
1038 1039
	extent_invalidatepage(tree, page, offset);
	btree_releasepage(page, GFP_NOFS);
1040
	if (PagePrivate(page)) {
1041 1042 1043
		btrfs_warn(BTRFS_I(page->mapping->host)->root->fs_info,
			   "page private not zero on page %llu",
			   (unsigned long long)page_offset(page));
1044 1045 1046 1047
		ClearPagePrivate(page);
		set_page_private(page, 0);
		page_cache_release(page);
	}
1048 1049
}

1050 1051
static int btree_set_page_dirty(struct page *page)
{
1052
#ifdef DEBUG
1053 1054 1055 1056 1057 1058 1059 1060
	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);
1061
#endif
1062 1063 1064
	return __set_page_dirty_nobuffers(page);
}

1065
static const struct address_space_operations btree_aops = {
1066
	.readpage	= btree_readpage,
1067
	.writepages	= btree_writepages,
1068 1069
	.releasepage	= btree_releasepage,
	.invalidatepage = btree_invalidatepage,
1070
#ifdef CONFIG_MIGRATION
1071
	.migratepage	= btree_migratepage,
1072
#endif
1073
	.set_page_dirty = btree_set_page_dirty,
1074 1075
};

1076
void readahead_tree_block(struct btrfs_root *root, u64 bytenr, u32 blocksize)
C
Chris Mason 已提交
1077
{
1078 1079
	struct extent_buffer *buf = NULL;
	struct inode *btree_inode = root->fs_info->btree_inode;
C
Chris Mason 已提交
1080

1081
	buf = btrfs_find_create_tree_block(root, bytenr, blocksize);
1082
	if (!buf)
1083
		return;
1084
	read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree,
1085
				 buf, 0, WAIT_NONE, btree_get_extent, 0);
1086
	free_extent_buffer(buf);
C
Chris Mason 已提交
1087 1088
}

1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112
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;
1113
	} else if (extent_buffer_uptodate(buf)) {
1114 1115 1116 1117 1118 1119 1120
		*eb = buf;
	} else {
		free_extent_buffer(buf);
	}
	return 0;
}

1121
struct extent_buffer *btrfs_find_tree_block(struct btrfs_root *root,
1122
					    u64 bytenr)
1123
{
1124
	return find_extent_buffer(root->fs_info, bytenr);
1125 1126 1127 1128 1129
}

struct extent_buffer *btrfs_find_create_tree_block(struct btrfs_root *root,
						 u64 bytenr, u32 blocksize)
{
1130
	if (btrfs_test_is_dummy_root(root))
1131 1132
		return alloc_test_extent_buffer(root->fs_info, bytenr,
						blocksize);
1133
	return alloc_extent_buffer(root->fs_info, bytenr, blocksize);
1134 1135 1136
}


1137 1138
int btrfs_write_tree_block(struct extent_buffer *buf)
{
1139
	return filemap_fdatawrite_range(buf->pages[0]->mapping, buf->start,
1140
					buf->start + buf->len - 1);
1141 1142 1143 1144
}

int btrfs_wait_tree_block_writeback(struct extent_buffer *buf)
{
1145
	return filemap_fdatawait_range(buf->pages[0]->mapping,
1146
				       buf->start, buf->start + buf->len - 1);
1147 1148
}

1149
struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr,
1150
				      u64 parent_transid)
1151 1152 1153 1154
{
	struct extent_buffer *buf = NULL;
	int ret;

1155
	buf = btrfs_find_create_tree_block(root, bytenr, root->nodesize);
1156 1157 1158
	if (!buf)
		return NULL;

1159
	ret = btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
1160 1161 1162 1163
	if (ret) {
		free_extent_buffer(buf);
		return NULL;
	}
1164
	return buf;
1165

1166 1167
}

1168 1169
void clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root,
		      struct extent_buffer *buf)
1170
{
1171 1172
	struct btrfs_fs_info *fs_info = root->fs_info;

1173
	if (btrfs_header_generation(buf) ==
1174
	    fs_info->running_transaction->transid) {
1175
		btrfs_assert_tree_locked(buf);
1176

1177
		if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &buf->bflags)) {
1178 1179 1180
			__percpu_counter_add(&fs_info->dirty_metadata_bytes,
					     -buf->len,
					     fs_info->dirty_metadata_batch);
1181 1182 1183 1184
			/* ugh, clear_extent_buffer_dirty needs to lock the page */
			btrfs_set_lock_blocking(buf);
			clear_extent_buffer_dirty(buf);
		}
1185
	}
1186 1187
}

1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213
static struct btrfs_subvolume_writers *btrfs_alloc_subvolume_writers(void)
{
	struct btrfs_subvolume_writers *writers;
	int ret;

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

	ret = percpu_counter_init(&writers->counter, 0);
	if (ret < 0) {
		kfree(writers);
		return ERR_PTR(ret);
	}

	init_waitqueue_head(&writers->wait);
	return writers;
}

static void
btrfs_free_subvolume_writers(struct btrfs_subvolume_writers *writers)
{
	percpu_counter_destroy(&writers->counter);
	kfree(writers);
}

1214 1215
static void __setup_root(u32 nodesize, u32 sectorsize, u32 stripesize,
			 struct btrfs_root *root, struct btrfs_fs_info *fs_info,
1216
			 u64 objectid)
1217
{
C
Chris Mason 已提交
1218
	root->node = NULL;
1219
	root->commit_root = NULL;
1220 1221
	root->sectorsize = sectorsize;
	root->nodesize = nodesize;
1222
	root->stripesize = stripesize;
1223
	root->state = 0;
1224
	root->orphan_cleanup_state = 0;
1225

1226 1227
	root->objectid = objectid;
	root->last_trans = 0;
1228
	root->highest_objectid = 0;
1229
	root->nr_delalloc_inodes = 0;
1230
	root->nr_ordered_extents = 0;
1231
	root->name = NULL;
1232
	root->inode_tree = RB_ROOT;
1233
	INIT_RADIX_TREE(&root->delayed_nodes_tree, GFP_ATOMIC);
1234
	root->block_rsv = NULL;
1235
	root->orphan_block_rsv = NULL;
1236 1237

	INIT_LIST_HEAD(&root->dirty_list);
1238
	INIT_LIST_HEAD(&root->root_list);
1239 1240
	INIT_LIST_HEAD(&root->delalloc_inodes);
	INIT_LIST_HEAD(&root->delalloc_root);
1241 1242
	INIT_LIST_HEAD(&root->ordered_extents);
	INIT_LIST_HEAD(&root->ordered_root);
1243 1244
	INIT_LIST_HEAD(&root->logged_list[0]);
	INIT_LIST_HEAD(&root->logged_list[1]);
1245
	spin_lock_init(&root->orphan_lock);
1246
	spin_lock_init(&root->inode_lock);
1247
	spin_lock_init(&root->delalloc_lock);
1248
	spin_lock_init(&root->ordered_extent_lock);
1249
	spin_lock_init(&root->accounting_lock);
1250 1251
	spin_lock_init(&root->log_extents_lock[0]);
	spin_lock_init(&root->log_extents_lock[1]);
1252
	mutex_init(&root->objectid_mutex);
1253
	mutex_init(&root->log_mutex);
1254
	mutex_init(&root->ordered_extent_mutex);
1255
	mutex_init(&root->delalloc_mutex);
Y
Yan Zheng 已提交
1256 1257 1258
	init_waitqueue_head(&root->log_writer_wait);
	init_waitqueue_head(&root->log_commit_wait[0]);
	init_waitqueue_head(&root->log_commit_wait[1]);
1259 1260
	INIT_LIST_HEAD(&root->log_ctxs[0]);
	INIT_LIST_HEAD(&root->log_ctxs[1]);
Y
Yan Zheng 已提交
1261 1262 1263
	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
1264
	atomic_set(&root->log_batch, 0);
1265
	atomic_set(&root->orphan_inodes, 0);
1266
	atomic_set(&root->refs, 1);
1267
	atomic_set(&root->will_be_snapshoted, 0);
Y
Yan Zheng 已提交
1268
	root->log_transid = 0;
1269
	root->log_transid_committed = -1;
1270
	root->last_log_commit = 0;
1271 1272 1273
	if (fs_info)
		extent_io_tree_init(&root->dirty_log_pages,
				     fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
1274

1275 1276
	memset(&root->root_key, 0, sizeof(root->root_key));
	memset(&root->root_item, 0, sizeof(root->root_item));
1277
	memset(&root->defrag_progress, 0, sizeof(root->defrag_progress));
1278
	memset(&root->root_kobj, 0, sizeof(root->root_kobj));
1279 1280 1281 1282
	if (fs_info)
		root->defrag_trans_start = fs_info->generation;
	else
		root->defrag_trans_start = 0;
1283
	init_completion(&root->kobj_unregister);
1284
	root->root_key.objectid = objectid;
1285
	root->anon_dev = 0;
1286

1287
	spin_lock_init(&root->root_item_lock);
1288 1289
}

1290
static struct btrfs_root *btrfs_alloc_root(struct btrfs_fs_info *fs_info)
A
Al Viro 已提交
1291 1292 1293 1294 1295 1296 1297
{
	struct btrfs_root *root = kzalloc(sizeof(*root), GFP_NOFS);
	if (root)
		root->fs_info = fs_info;
	return root;
}

1298 1299 1300 1301 1302 1303 1304 1305 1306
#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);
1307
	__setup_root(4096, 4096, 4096, root, NULL, 1);
1308
	set_bit(BTRFS_ROOT_DUMMY_ROOT, &root->state);
1309
	root->alloc_bytenr = 0;
1310 1311 1312 1313 1314

	return root;
}
#endif

1315 1316 1317 1318 1319 1320 1321 1322 1323
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;
1324
	uuid_le uuid;
1325 1326 1327 1328 1329

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

1330 1331
	__setup_root(tree_root->nodesize, tree_root->sectorsize,
		tree_root->stripesize, root, fs_info, objectid);
1332 1333 1334 1335
	root->root_key.objectid = objectid;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = 0;

1336
	leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0);
1337 1338
	if (IS_ERR(leaf)) {
		ret = PTR_ERR(leaf);
1339
		leaf = NULL;
1340 1341 1342 1343 1344 1345 1346 1347 1348 1349
		goto fail;
	}

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

1350
	write_extent_buffer(leaf, fs_info->fsid, btrfs_header_fsid(),
1351 1352
			    BTRFS_FSID_SIZE);
	write_extent_buffer(leaf, fs_info->chunk_tree_uuid,
1353
			    btrfs_header_chunk_tree_uuid(leaf),
1354 1355 1356 1357
			    BTRFS_UUID_SIZE);
	btrfs_mark_buffer_dirty(leaf);

	root->commit_root = btrfs_root_node(root);
1358
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &root->state);
1359 1360 1361 1362 1363 1364 1365 1366 1367 1368

	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);
1369 1370
	uuid_le_gen(&uuid);
	memcpy(root->root_item.uuid, uuid.b, BTRFS_UUID_SIZE);
1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381
	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);

1382 1383
	return root;

1384
fail:
1385 1386
	if (leaf) {
		btrfs_tree_unlock(leaf);
1387
		free_extent_buffer(root->commit_root);
1388 1389 1390
		free_extent_buffer(leaf);
	}
	kfree(root);
1391

1392
	return ERR_PTR(ret);
1393 1394
}

Y
Yan Zheng 已提交
1395 1396
static struct btrfs_root *alloc_log_tree(struct btrfs_trans_handle *trans,
					 struct btrfs_fs_info *fs_info)
1397 1398 1399
{
	struct btrfs_root *root;
	struct btrfs_root *tree_root = fs_info->tree_root;
Y
Yan Zheng 已提交
1400
	struct extent_buffer *leaf;
1401

A
Al Viro 已提交
1402
	root = btrfs_alloc_root(fs_info);
1403
	if (!root)
Y
Yan Zheng 已提交
1404
		return ERR_PTR(-ENOMEM);
1405

1406 1407 1408
	__setup_root(tree_root->nodesize, tree_root->sectorsize,
		     tree_root->stripesize, root, fs_info,
		     BTRFS_TREE_LOG_OBJECTID);
1409 1410 1411 1412

	root->root_key.objectid = BTRFS_TREE_LOG_OBJECTID;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = BTRFS_TREE_LOG_OBJECTID;
1413

Y
Yan Zheng 已提交
1414
	/*
1415 1416
	 * DON'T set REF_COWS for log trees
	 *
Y
Yan Zheng 已提交
1417 1418 1419 1420 1421
	 * 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).
	 */
1422

1423 1424
	leaf = btrfs_alloc_tree_block(trans, root, 0, BTRFS_TREE_LOG_OBJECTID,
			NULL, 0, 0, 0);
Y
Yan Zheng 已提交
1425 1426 1427 1428
	if (IS_ERR(leaf)) {
		kfree(root);
		return ERR_CAST(leaf);
	}
1429

1430 1431 1432 1433 1434
	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
1435
	root->node = leaf;
1436 1437

	write_extent_buffer(root->node, root->fs_info->fsid,
1438
			    btrfs_header_fsid(), BTRFS_FSID_SIZE);
1439 1440
	btrfs_mark_buffer_dirty(root->node);
	btrfs_tree_unlock(root->node);
Y
Yan Zheng 已提交
1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470
	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;
1471 1472 1473
	btrfs_set_stack_inode_generation(inode_item, 1);
	btrfs_set_stack_inode_size(inode_item, 3);
	btrfs_set_stack_inode_nlink(inode_item, 1);
1474
	btrfs_set_stack_inode_nbytes(inode_item, root->nodesize);
1475
	btrfs_set_stack_inode_mode(inode_item, S_IFDIR | 0755);
Y
Yan Zheng 已提交
1476

1477
	btrfs_set_root_node(&log_root->root_item, log_root->node);
Y
Yan Zheng 已提交
1478 1479 1480 1481

	WARN_ON(root->log_root);
	root->log_root = log_root;
	root->log_transid = 0;
1482
	root->log_transid_committed = -1;
1483
	root->last_log_commit = 0;
1484 1485 1486
	return 0;
}

1487 1488
static struct btrfs_root *btrfs_read_tree_root(struct btrfs_root *tree_root,
					       struct btrfs_key *key)
1489 1490 1491
{
	struct btrfs_root *root;
	struct btrfs_fs_info *fs_info = tree_root->fs_info;
1492
	struct btrfs_path *path;
1493
	u64 generation;
1494
	int ret;
1495

1496 1497
	path = btrfs_alloc_path();
	if (!path)
1498
		return ERR_PTR(-ENOMEM);
1499 1500 1501 1502 1503

	root = btrfs_alloc_root(fs_info);
	if (!root) {
		ret = -ENOMEM;
		goto alloc_fail;
1504 1505
	}

1506 1507
	__setup_root(tree_root->nodesize, tree_root->sectorsize,
		tree_root->stripesize, root, fs_info, key->objectid);
1508

1509 1510
	ret = btrfs_find_root(tree_root, key, path,
			      &root->root_item, &root->root_key);
1511
	if (ret) {
1512 1513
		if (ret > 0)
			ret = -ENOENT;
1514
		goto find_fail;
1515
	}
1516

1517
	generation = btrfs_root_generation(&root->root_item);
1518
	root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
1519
				     generation);
1520 1521 1522 1523 1524 1525
	if (!root->node) {
		ret = -ENOMEM;
		goto find_fail;
	} else if (!btrfs_buffer_uptodate(root->node, generation, 0)) {
		ret = -EIO;
		goto read_fail;
1526
	}
1527
	root->commit_root = btrfs_root_node(root);
1528
out:
1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550
	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) {
1551
		set_bit(BTRFS_ROOT_REF_COWS, &root->state);
1552 1553
		btrfs_check_and_init_root_item(&root->root_item);
	}
1554

1555 1556 1557
	return root;
}

1558 1559 1560
int btrfs_init_fs_root(struct btrfs_root *root)
{
	int ret;
1561
	struct btrfs_subvolume_writers *writers;
1562 1563 1564 1565 1566 1567 1568 1569 1570

	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;
	}

1571 1572 1573 1574 1575 1576 1577
	writers = btrfs_alloc_subvolume_writers();
	if (IS_ERR(writers)) {
		ret = PTR_ERR(writers);
		goto fail;
	}
	root->subv_writers = writers;

1578
	btrfs_init_free_ino_ctl(root);
1579 1580
	spin_lock_init(&root->ino_cache_lock);
	init_waitqueue_head(&root->ino_cache_wait);
1581 1582 1583

	ret = get_anon_bdev(&root->anon_dev);
	if (ret)
1584
		goto free_writers;
1585
	return 0;
1586 1587 1588

free_writers:
	btrfs_free_subvolume_writers(root->subv_writers);
1589 1590 1591 1592 1593 1594
fail:
	kfree(root->free_ino_ctl);
	kfree(root->free_ino_pinned);
	return ret;
}

1595 1596
static struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info,
					       u64 root_id)
1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620
{
	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)
1621
		set_bit(BTRFS_ROOT_IN_RADIX, &root->state);
1622 1623 1624 1625 1626 1627
	spin_unlock(&fs_info->fs_roots_radix_lock);
	radix_tree_preload_end();

	return ret;
}

1628 1629 1630
struct btrfs_root *btrfs_get_fs_root(struct btrfs_fs_info *fs_info,
				     struct btrfs_key *location,
				     bool check_ref)
1631 1632 1633 1634
{
	struct btrfs_root *root;
	int ret;

1635 1636 1637 1638
	if (location->objectid == BTRFS_ROOT_TREE_OBJECTID)
		return fs_info->tree_root;
	if (location->objectid == BTRFS_EXTENT_TREE_OBJECTID)
		return fs_info->extent_root;
1639 1640 1641 1642
	if (location->objectid == BTRFS_CHUNK_TREE_OBJECTID)
		return fs_info->chunk_root;
	if (location->objectid == BTRFS_DEV_TREE_OBJECTID)
		return fs_info->dev_root;
1643 1644
	if (location->objectid == BTRFS_CSUM_TREE_OBJECTID)
		return fs_info->csum_root;
1645 1646 1647
	if (location->objectid == BTRFS_QUOTA_TREE_OBJECTID)
		return fs_info->quota_root ? fs_info->quota_root :
					     ERR_PTR(-ENOENT);
1648 1649 1650
	if (location->objectid == BTRFS_UUID_TREE_OBJECTID)
		return fs_info->uuid_root ? fs_info->uuid_root :
					    ERR_PTR(-ENOENT);
1651
again:
1652
	root = btrfs_lookup_fs_root(fs_info, location->objectid);
1653
	if (root) {
1654
		if (check_ref && btrfs_root_refs(&root->root_item) == 0)
1655
			return ERR_PTR(-ENOENT);
1656
		return root;
1657
	}
1658

1659
	root = btrfs_read_fs_root(fs_info->tree_root, location);
1660 1661
	if (IS_ERR(root))
		return root;
1662

1663
	if (check_ref && btrfs_root_refs(&root->root_item) == 0) {
1664
		ret = -ENOENT;
1665
		goto fail;
1666
	}
1667

1668
	ret = btrfs_init_fs_root(root);
1669 1670
	if (ret)
		goto fail;
1671

1672 1673
	ret = btrfs_find_item(fs_info->tree_root, NULL, BTRFS_ORPHAN_OBJECTID,
			location->objectid, BTRFS_ORPHAN_ITEM_KEY, NULL);
1674 1675 1676
	if (ret < 0)
		goto fail;
	if (ret == 0)
1677
		set_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &root->state);
1678

1679
	ret = btrfs_insert_fs_root(fs_info, root);
1680
	if (ret) {
1681 1682 1683 1684 1685
		if (ret == -EEXIST) {
			free_fs_root(root);
			goto again;
		}
		goto fail;
1686
	}
1687
	return root;
1688 1689 1690
fail:
	free_fs_root(root);
	return ERR_PTR(ret);
1691 1692
}

C
Chris Mason 已提交
1693 1694 1695 1696 1697 1698
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
1699

1700 1701
	rcu_read_lock();
	list_for_each_entry_rcu(device, &info->fs_devices->devices, dev_list) {
1702 1703
		if (!device->bdev)
			continue;
C
Chris Mason 已提交
1704 1705 1706 1707 1708 1709
		bdi = blk_get_backing_dev_info(device->bdev);
		if (bdi && bdi_congested(bdi, bdi_bits)) {
			ret = 1;
			break;
		}
	}
1710
	rcu_read_unlock();
C
Chris Mason 已提交
1711 1712 1713 1714 1715
	return ret;
}

static int setup_bdi(struct btrfs_fs_info *info, struct backing_dev_info *bdi)
{
1716 1717 1718
	int err;

	bdi->capabilities = BDI_CAP_MAP_COPY;
1719
	err = bdi_setup_and_register(bdi, "btrfs", BDI_CAP_MAP_COPY);
1720 1721 1722
	if (err)
		return err;

1723
	bdi->ra_pages	= default_backing_dev_info.ra_pages;
C
Chris Mason 已提交
1724 1725 1726 1727 1728
	bdi->congested_fn	= btrfs_congested_fn;
	bdi->congested_data	= info;
	return 0;
}

1729 1730 1731 1732 1733
/*
 * 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)
1734 1735
{
	struct bio *bio;
1736
	struct btrfs_end_io_wq *end_io_wq;
1737 1738
	int error;

1739
	end_io_wq = container_of(work, struct btrfs_end_io_wq, work);
1740
	bio = end_io_wq->bio;
1741

1742 1743 1744
	error = end_io_wq->error;
	bio->bi_private = end_io_wq->private;
	bio->bi_end_io = end_io_wq->end_io;
1745
	kmem_cache_free(btrfs_end_io_wq_cache, end_io_wq);
1746
	bio_endio_nodec(bio, error);
1747 1748
}

1749 1750 1751
static int cleaner_kthread(void *arg)
{
	struct btrfs_root *root = arg;
1752
	int again;
1753 1754

	do {
1755
		again = 0;
1756

1757
		/* Make the cleaner go to sleep early. */
1758
		if (btrfs_need_cleaner_sleep(root))
1759 1760 1761 1762 1763
			goto sleep;

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

1764 1765 1766 1767
		/*
		 * Avoid the problem that we change the status of the fs
		 * during the above check and trylock.
		 */
1768
		if (btrfs_need_cleaner_sleep(root)) {
1769 1770
			mutex_unlock(&root->fs_info->cleaner_mutex);
			goto sleep;
1771
		}
1772

1773
		btrfs_run_delayed_iputs(root);
1774
		btrfs_delete_unused_bgs(root->fs_info);
1775 1776 1777 1778
		again = btrfs_clean_one_deleted_snapshot(root);
		mutex_unlock(&root->fs_info->cleaner_mutex);

		/*
1779 1780
		 * The defragger has dealt with the R/O remount and umount,
		 * needn't do anything special here.
1781 1782 1783
		 */
		btrfs_run_defrag_inodes(root->fs_info);
sleep:
D
David Sterba 已提交
1784
		if (!try_to_freeze() && !again) {
1785
			set_current_state(TASK_INTERRUPTIBLE);
1786 1787
			if (!kthread_should_stop())
				schedule();
1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798
			__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;
1799
	u64 transid;
1800 1801
	unsigned long now;
	unsigned long delay;
1802
	bool cannot_commit;
1803 1804

	do {
1805
		cannot_commit = false;
1806
		delay = HZ * root->fs_info->commit_interval;
1807 1808
		mutex_lock(&root->fs_info->transaction_kthread_mutex);

J
Josef Bacik 已提交
1809
		spin_lock(&root->fs_info->trans_lock);
1810 1811
		cur = root->fs_info->running_transaction;
		if (!cur) {
J
Josef Bacik 已提交
1812
			spin_unlock(&root->fs_info->trans_lock);
1813 1814
			goto sleep;
		}
Y
Yan Zheng 已提交
1815

1816
		now = get_seconds();
1817
		if (cur->state < TRANS_STATE_BLOCKED &&
1818 1819
		    (now < cur->start_time ||
		     now - cur->start_time < root->fs_info->commit_interval)) {
J
Josef Bacik 已提交
1820
			spin_unlock(&root->fs_info->trans_lock);
1821 1822 1823
			delay = HZ * 5;
			goto sleep;
		}
1824
		transid = cur->transid;
J
Josef Bacik 已提交
1825
		spin_unlock(&root->fs_info->trans_lock);
1826

1827
		/* If the file system is aborted, this will always fail. */
1828
		trans = btrfs_attach_transaction(root);
1829
		if (IS_ERR(trans)) {
1830 1831
			if (PTR_ERR(trans) != -ENOENT)
				cannot_commit = true;
1832
			goto sleep;
1833
		}
1834
		if (transid == trans->transid) {
1835
			btrfs_commit_transaction(trans, root);
1836 1837 1838
		} else {
			btrfs_end_transaction(trans, root);
		}
1839 1840 1841 1842
sleep:
		wake_up_process(root->fs_info->cleaner_kthread);
		mutex_unlock(&root->fs_info->transaction_kthread_mutex);

J
Josef Bacik 已提交
1843 1844 1845
		if (unlikely(test_bit(BTRFS_FS_STATE_ERROR,
				      &root->fs_info->fs_state)))
			btrfs_cleanup_transaction(root);
1846
		if (!try_to_freeze()) {
1847
			set_current_state(TASK_INTERRUPTIBLE);
1848
			if (!kthread_should_stop() &&
1849 1850
			    (!btrfs_transaction_blocked(root->fs_info) ||
			     cannot_commit))
1851
				schedule_timeout(delay);
1852 1853 1854 1855 1856 1857
			__set_current_state(TASK_RUNNING);
		}
	} while (!kthread_should_stop());
	return 0;
}

C
Chris Mason 已提交
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 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963
/*
 * 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));

1964 1965 1966 1967 1968 1969 1970 1971
	/*
	 * 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
1972
			       btrfs_header_generation(info->fs_root->node));
1973
		btrfs_set_backup_fs_root_level(root_backup,
C
Chris Mason 已提交
1974
			       btrfs_header_level(info->fs_root->node));
1975
	}
C
Chris Mason 已提交
1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056

	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
2057 2058 2059
/* helper to cleanup workers */
static void btrfs_stop_all_workers(struct btrfs_fs_info *fs_info)
{
2060
	btrfs_destroy_workqueue(fs_info->fixup_workers);
2061
	btrfs_destroy_workqueue(fs_info->delalloc_workers);
2062
	btrfs_destroy_workqueue(fs_info->workers);
2063 2064 2065
	btrfs_destroy_workqueue(fs_info->endio_workers);
	btrfs_destroy_workqueue(fs_info->endio_meta_workers);
	btrfs_destroy_workqueue(fs_info->endio_raid56_workers);
2066
	btrfs_destroy_workqueue(fs_info->endio_repair_workers);
2067
	btrfs_destroy_workqueue(fs_info->rmw_workers);
2068 2069 2070
	btrfs_destroy_workqueue(fs_info->endio_meta_write_workers);
	btrfs_destroy_workqueue(fs_info->endio_write_workers);
	btrfs_destroy_workqueue(fs_info->endio_freespace_worker);
2071
	btrfs_destroy_workqueue(fs_info->submit_workers);
2072
	btrfs_destroy_workqueue(fs_info->delayed_workers);
2073
	btrfs_destroy_workqueue(fs_info->caching_workers);
2074
	btrfs_destroy_workqueue(fs_info->readahead_workers);
2075
	btrfs_destroy_workqueue(fs_info->flush_workers);
2076
	btrfs_destroy_workqueue(fs_info->qgroup_rescan_workers);
C
Chris Mason 已提交
2077
	btrfs_destroy_workqueue(fs_info->extent_workers);
L
Liu Bo 已提交
2078 2079
}

2080 2081 2082 2083 2084 2085 2086 2087 2088 2089
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
2090 2091 2092
/* helper to cleanup tree roots */
static void free_root_pointers(struct btrfs_fs_info *info, int chunk_root)
{
2093
	free_root_extent_buffers(info->tree_root);
2094

2095 2096 2097 2098 2099 2100 2101
	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
2102 2103
}

2104
void btrfs_free_fs_roots(struct btrfs_fs_info *fs_info)
2105 2106 2107 2108 2109 2110 2111 2112 2113 2114
{
	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);

2115
		if (test_bit(BTRFS_ROOT_IN_RADIX, &gang[0]->state)) {
2116
			btrfs_drop_and_free_fs_root(fs_info, gang[0]);
2117 2118 2119
		} else {
			free_extent_buffer(gang[0]->node);
			free_extent_buffer(gang[0]->commit_root);
2120
			btrfs_put_fs_root(gang[0]);
2121 2122 2123 2124 2125 2126 2127 2128 2129 2130
		}
	}

	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++)
2131
			btrfs_drop_and_free_fs_root(fs_info, gang[i]);
2132
	}
2133 2134 2135 2136 2137 2138

	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
		btrfs_free_log_root_tree(NULL, fs_info);
		btrfs_destroy_pinned_extent(fs_info->tree_root,
					    fs_info->pinned_extents);
	}
2139
}
C
Chris Mason 已提交
2140

A
Al Viro 已提交
2141 2142 2143
int open_ctree(struct super_block *sb,
	       struct btrfs_fs_devices *fs_devices,
	       char *options)
2144
{
2145 2146
	u32 sectorsize;
	u32 nodesize;
2147
	u32 stripesize;
2148
	u64 generation;
2149
	u64 features;
2150
	struct btrfs_key location;
2151
	struct buffer_head *bh;
2152
	struct btrfs_super_block *disk_super;
2153
	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
2154
	struct btrfs_root *tree_root;
2155 2156 2157 2158
	struct btrfs_root *extent_root;
	struct btrfs_root *csum_root;
	struct btrfs_root *chunk_root;
	struct btrfs_root *dev_root;
2159
	struct btrfs_root *quota_root;
2160
	struct btrfs_root *uuid_root;
2161
	struct btrfs_root *log_tree_root;
2162
	int ret;
2163
	int err = -EINVAL;
C
Chris Mason 已提交
2164 2165
	int num_backups_tried = 0;
	int backup_index = 0;
2166 2167
	int max_active;
	int flags = WQ_MEM_RECLAIM | WQ_FREEZABLE | WQ_UNBOUND;
2168 2169
	bool create_uuid_tree;
	bool check_uuid_tree;
2170

2171
	tree_root = fs_info->tree_root = btrfs_alloc_root(fs_info);
A
Al Viro 已提交
2172
	chunk_root = fs_info->chunk_root = btrfs_alloc_root(fs_info);
2173
	if (!tree_root || !chunk_root) {
C
Chris Mason 已提交
2174 2175 2176
		err = -ENOMEM;
		goto fail;
	}
2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189

	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;
	}

2190 2191 2192 2193 2194 2195 2196 2197
	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));

2198 2199 2200 2201 2202 2203
	ret = percpu_counter_init(&fs_info->delalloc_bytes, 0);
	if (ret) {
		err = ret;
		goto fail_dirty_metadata_bytes;
	}

2204 2205 2206 2207 2208 2209
	ret = percpu_counter_init(&fs_info->bio_counter, 0);
	if (ret) {
		err = ret;
		goto fail_delalloc_bytes;
	}

2210 2211 2212
	fs_info->btree_inode = new_inode(sb);
	if (!fs_info->btree_inode) {
		err = -ENOMEM;
2213
		goto fail_bio_counter;
2214 2215
	}

2216
	mapping_set_gfp_mask(fs_info->btree_inode->i_mapping, GFP_NOFS);
2217

2218
	INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC);
2219
	INIT_RADIX_TREE(&fs_info->buffer_radix, GFP_ATOMIC);
C
Chris Mason 已提交
2220
	INIT_LIST_HEAD(&fs_info->trans_list);
2221
	INIT_LIST_HEAD(&fs_info->dead_roots);
Y
Yan, Zheng 已提交
2222
	INIT_LIST_HEAD(&fs_info->delayed_iputs);
2223
	INIT_LIST_HEAD(&fs_info->delalloc_roots);
2224
	INIT_LIST_HEAD(&fs_info->caching_block_groups);
2225
	spin_lock_init(&fs_info->delalloc_root_lock);
J
Josef Bacik 已提交
2226
	spin_lock_init(&fs_info->trans_lock);
2227
	spin_lock_init(&fs_info->fs_roots_radix_lock);
Y
Yan, Zheng 已提交
2228
	spin_lock_init(&fs_info->delayed_iput_lock);
C
Chris Mason 已提交
2229
	spin_lock_init(&fs_info->defrag_inodes_lock);
2230
	spin_lock_init(&fs_info->free_chunk_lock);
J
Jan Schmidt 已提交
2231
	spin_lock_init(&fs_info->tree_mod_seq_lock);
2232
	spin_lock_init(&fs_info->super_lock);
J
Josef Bacik 已提交
2233
	spin_lock_init(&fs_info->qgroup_op_lock);
2234
	spin_lock_init(&fs_info->buffer_lock);
2235
	spin_lock_init(&fs_info->unused_bgs_lock);
J
Jan Schmidt 已提交
2236
	rwlock_init(&fs_info->tree_mod_log_lock);
C
Chris Mason 已提交
2237
	mutex_init(&fs_info->reloc_mutex);
2238
	mutex_init(&fs_info->delalloc_root_mutex);
2239
	seqlock_init(&fs_info->profiles_lock);
2240

2241
	init_completion(&fs_info->kobj_unregister);
2242
	INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots);
2243
	INIT_LIST_HEAD(&fs_info->space_info);
J
Jan Schmidt 已提交
2244
	INIT_LIST_HEAD(&fs_info->tree_mod_seq_list);
2245
	INIT_LIST_HEAD(&fs_info->unused_bgs);
2246
	btrfs_mapping_init(&fs_info->mapping_tree);
2247 2248 2249 2250 2251 2252 2253 2254 2255
	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);
2256
	atomic_set(&fs_info->nr_async_submits, 0);
2257
	atomic_set(&fs_info->async_delalloc_pages, 0);
2258
	atomic_set(&fs_info->async_submit_draining, 0);
2259
	atomic_set(&fs_info->nr_async_bios, 0);
C
Chris Mason 已提交
2260
	atomic_set(&fs_info->defrag_running, 0);
J
Josef Bacik 已提交
2261
	atomic_set(&fs_info->qgroup_op_seq, 0);
2262
	atomic64_set(&fs_info->tree_mod_seq, 0);
C
Chris Mason 已提交
2263
	fs_info->sb = sb;
2264
	fs_info->max_inline = BTRFS_DEFAULT_MAX_INLINE;
J
Josef Bacik 已提交
2265
	fs_info->metadata_ratio = 0;
C
Chris Mason 已提交
2266
	fs_info->defrag_inodes = RB_ROOT;
2267
	fs_info->free_chunk_space = 0;
J
Jan Schmidt 已提交
2268
	fs_info->tree_mod_log = RB_ROOT;
2269
	fs_info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
2270
	fs_info->avg_delayed_ref_runtime = div64_u64(NSEC_PER_SEC, 64);
2271 2272 2273
	/* 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
2274

2275 2276
	fs_info->thread_pool_size = min_t(unsigned long,
					  num_online_cpus() + 2, 8);
2277

2278 2279
	INIT_LIST_HEAD(&fs_info->ordered_roots);
	spin_lock_init(&fs_info->ordered_root_lock);
2280 2281 2282 2283 2284 2285 2286
	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);
2287

A
Arne Jansen 已提交
2288 2289 2290 2291 2292
	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);
2293
	init_waitqueue_head(&fs_info->replace_wait);
A
Arne Jansen 已提交
2294 2295
	init_waitqueue_head(&fs_info->scrub_pause_wait);
	fs_info->scrub_workers_refcnt = 0;
2296 2297 2298
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	fs_info->check_integrity_print_mask = 0;
#endif
A
Arne Jansen 已提交
2299

2300 2301
	spin_lock_init(&fs_info->balance_lock);
	mutex_init(&fs_info->balance_mutex);
2302 2303
	atomic_set(&fs_info->balance_running, 0);
	atomic_set(&fs_info->balance_pause_req, 0);
2304
	atomic_set(&fs_info->balance_cancel_req, 0);
2305
	fs_info->balance_ctl = NULL;
2306
	init_waitqueue_head(&fs_info->balance_wait_q);
2307
	btrfs_init_async_reclaim_work(&fs_info->async_reclaim_work);
A
Arne Jansen 已提交
2308

2309 2310
	sb->s_blocksize = 4096;
	sb->s_blocksize_bits = blksize_bits(4096);
J
Jens Axboe 已提交
2311
	sb->s_bdi = &fs_info->bdi;
2312

2313
	fs_info->btree_inode->i_ino = BTRFS_BTREE_INODE_OBJECTID;
M
Miklos Szeredi 已提交
2314
	set_nlink(fs_info->btree_inode, 1);
2315 2316 2317 2318 2319 2320
	/*
	 * 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;
2321
	fs_info->btree_inode->i_mapping->a_ops = &btree_aops;
C
Chris Mason 已提交
2322 2323
	fs_info->btree_inode->i_mapping->backing_dev_info = &fs_info->bdi;

2324
	RB_CLEAR_NODE(&BTRFS_I(fs_info->btree_inode)->rb_node);
2325
	extent_io_tree_init(&BTRFS_I(fs_info->btree_inode)->io_tree,
2326
			     fs_info->btree_inode->i_mapping);
2327
	BTRFS_I(fs_info->btree_inode)->io_tree.track_uptodate = 0;
2328
	extent_map_tree_init(&BTRFS_I(fs_info->btree_inode)->extent_tree);
2329 2330

	BTRFS_I(fs_info->btree_inode)->io_tree.ops = &btree_extent_io_ops;
2331

2332 2333 2334
	BTRFS_I(fs_info->btree_inode)->root = tree_root;
	memset(&BTRFS_I(fs_info->btree_inode)->location, 0,
	       sizeof(struct btrfs_key));
2335 2336
	set_bit(BTRFS_INODE_DUMMY,
		&BTRFS_I(fs_info->btree_inode)->runtime_flags);
2337
	btrfs_insert_inode_hash(fs_info->btree_inode);
2338

J
Josef Bacik 已提交
2339
	spin_lock_init(&fs_info->block_group_cache_lock);
2340
	fs_info->block_group_cache_tree = RB_ROOT;
2341
	fs_info->first_logical_byte = (u64)-1;
J
Josef Bacik 已提交
2342

2343
	extent_io_tree_init(&fs_info->freed_extents[0],
2344
			     fs_info->btree_inode->i_mapping);
2345
	extent_io_tree_init(&fs_info->freed_extents[1],
2346
			     fs_info->btree_inode->i_mapping);
2347
	fs_info->pinned_extents = &fs_info->freed_extents[0];
2348
	fs_info->do_barriers = 1;
2349

C
Chris Mason 已提交
2350

2351
	mutex_init(&fs_info->ordered_operations_mutex);
2352
	mutex_init(&fs_info->ordered_extent_flush_mutex);
2353
	mutex_init(&fs_info->tree_log_mutex);
2354
	mutex_init(&fs_info->chunk_mutex);
2355 2356
	mutex_init(&fs_info->transaction_kthread_mutex);
	mutex_init(&fs_info->cleaner_mutex);
2357
	mutex_init(&fs_info->volume_mutex);
2358
	init_rwsem(&fs_info->commit_root_sem);
2359
	init_rwsem(&fs_info->cleanup_work_sem);
2360
	init_rwsem(&fs_info->subvol_sem);
S
Stefan Behrens 已提交
2361
	sema_init(&fs_info->uuid_tree_rescan_sem, 1);
2362 2363 2364 2365 2366
	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);
2367

2368
	spin_lock_init(&fs_info->qgroup_lock);
2369
	mutex_init(&fs_info->qgroup_ioctl_lock);
2370
	fs_info->qgroup_tree = RB_ROOT;
J
Josef Bacik 已提交
2371
	fs_info->qgroup_op_tree = RB_ROOT;
2372 2373 2374 2375
	INIT_LIST_HEAD(&fs_info->dirty_qgroups);
	fs_info->qgroup_seq = 1;
	fs_info->quota_enabled = 0;
	fs_info->pending_quota_state = 0;
2376
	fs_info->qgroup_ulist = NULL;
J
Jan Schmidt 已提交
2377
	mutex_init(&fs_info->qgroup_rescan_lock);
2378

2379 2380 2381
	btrfs_init_free_cluster(&fs_info->meta_alloc_cluster);
	btrfs_init_free_cluster(&fs_info->data_alloc_cluster);

2382
	init_waitqueue_head(&fs_info->transaction_throttle);
2383
	init_waitqueue_head(&fs_info->transaction_wait);
S
Sage Weil 已提交
2384
	init_waitqueue_head(&fs_info->transaction_blocked_wait);
2385
	init_waitqueue_head(&fs_info->async_submit_wait);
2386

D
David Woodhouse 已提交
2387 2388
	ret = btrfs_alloc_stripe_hash_table(fs_info);
	if (ret) {
2389
		err = ret;
D
David Woodhouse 已提交
2390 2391 2392
		goto fail_alloc;
	}

2393
	__setup_root(4096, 4096, 4096, tree_root,
C
Chris Mason 已提交
2394
		     fs_info, BTRFS_ROOT_TREE_OBJECTID);
2395

2396
	invalidate_bdev(fs_devices->latest_bdev);
D
David Sterba 已提交
2397 2398 2399 2400

	/*
	 * Read super block and check the signature bytes only
	 */
Y
Yan Zheng 已提交
2401
	bh = btrfs_read_dev_super(fs_devices->latest_bdev);
2402 2403
	if (!bh) {
		err = -EINVAL;
2404
		goto fail_alloc;
2405
	}
C
Chris Mason 已提交
2406

D
David Sterba 已提交
2407 2408 2409 2410 2411
	/*
	 * 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)) {
2412
		printk(KERN_ERR "BTRFS: superblock checksum mismatch\n");
D
David Sterba 已提交
2413 2414 2415 2416 2417 2418 2419 2420 2421
		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
	 */
2422 2423 2424
	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));
2425
	brelse(bh);
2426

2427
	memcpy(fs_info->fsid, fs_info->super_copy->fsid, BTRFS_FSID_SIZE);
2428

D
David Sterba 已提交
2429 2430
	ret = btrfs_check_super_valid(fs_info, sb->s_flags & MS_RDONLY);
	if (ret) {
2431
		printk(KERN_ERR "BTRFS: superblock contains fatal errors\n");
D
David Sterba 已提交
2432 2433 2434 2435
		err = -EINVAL;
		goto fail_alloc;
	}

2436
	disk_super = fs_info->super_copy;
2437
	if (!btrfs_super_root(disk_super))
2438
		goto fail_alloc;
2439

L
liubo 已提交
2440
	/* check FS state, whether FS is broken. */
2441 2442
	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
2443

C
Chris Mason 已提交
2444 2445 2446 2447 2448 2449 2450
	/*
	 * 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);

2451 2452 2453 2454 2455 2456
	/*
	 * 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
2457 2458 2459
	ret = btrfs_parse_options(tree_root, options);
	if (ret) {
		err = ret;
2460
		goto fail_alloc;
Y
Yan Zheng 已提交
2461
	}
2462

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

2473 2474 2475 2476
	/*
	 * Leafsize and nodesize were always equal, this is only a sanity check.
	 */
	if (le32_to_cpu(disk_super->__unused_leafsize) !=
2477 2478 2479 2480
	    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),
2481
		       le32_to_cpu(disk_super->__unused_leafsize));
2482 2483 2484
		err = -EINVAL;
		goto fail_alloc;
	}
2485
	if (btrfs_super_nodesize(disk_super) > BTRFS_MAX_METADATA_BLOCKSIZE) {
2486 2487
		printk(KERN_ERR "BTRFS: couldn't mount because metadata "
		       "blocksize (%d) was too large\n",
2488
		       btrfs_super_nodesize(disk_super));
2489 2490 2491 2492
		err = -EINVAL;
		goto fail_alloc;
	}

2493
	features = btrfs_super_incompat_flags(disk_super);
L
Li Zefan 已提交
2494
	features |= BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF;
L
Li Zefan 已提交
2495
	if (tree_root->fs_info->compress_type == BTRFS_COMPRESS_LZO)
L
Li Zefan 已提交
2496
		features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
2497

2498
	if (features & BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA)
2499
		printk(KERN_ERR "BTRFS: has skinny extents\n");
2500

2501 2502 2503 2504
	/*
	 * flag our filesystem as having big metadata blocks if
	 * they are bigger than the page size
	 */
2505
	if (btrfs_super_nodesize(disk_super) > PAGE_CACHE_SIZE) {
2506
		if (!(features & BTRFS_FEATURE_INCOMPAT_BIG_METADATA))
2507
			printk(KERN_INFO "BTRFS: flagging fs with big metadata feature\n");
2508 2509 2510
		features |= BTRFS_FEATURE_INCOMPAT_BIG_METADATA;
	}

2511 2512 2513
	nodesize = btrfs_super_nodesize(disk_super);
	sectorsize = btrfs_super_sectorsize(disk_super);
	stripesize = btrfs_super_stripesize(disk_super);
2514
	fs_info->dirty_metadata_batch = nodesize * (1 + ilog2(nr_cpu_ids));
2515
	fs_info->delalloc_batch = sectorsize * 512 * (1 + ilog2(nr_cpu_ids));
2516 2517 2518 2519 2520 2521

	/*
	 * 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) &&
2522
	    (sectorsize != nodesize)) {
2523
		printk(KERN_WARNING "BTRFS: unequal leaf/node/sector sizes "
2524 2525 2526 2527 2528
				"are not allowed for mixed block groups on %s\n",
				sb->s_id);
		goto fail_alloc;
	}

2529 2530 2531 2532
	/*
	 * Needn't use the lock because there is no other task which will
	 * update the flag.
	 */
L
Li Zefan 已提交
2533
	btrfs_set_super_incompat_flags(disk_super, features);
2534

2535 2536 2537 2538 2539
	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",
2540
		       features);
2541
		err = -EINVAL;
2542
		goto fail_alloc;
2543
	}
2544

2545
	max_active = fs_info->thread_pool_size;
2546

2547 2548 2549
	fs_info->workers =
		btrfs_alloc_workqueue("worker", flags | WQ_HIGHPRI,
				      max_active, 16);
C
Chris Mason 已提交
2550

2551 2552
	fs_info->delalloc_workers =
		btrfs_alloc_workqueue("delalloc", flags, max_active, 2);
2553

2554 2555
	fs_info->flush_workers =
		btrfs_alloc_workqueue("flush_delalloc", flags, max_active, 0);
2556

2557 2558
	fs_info->caching_workers =
		btrfs_alloc_workqueue("cache", flags, max_active, 0);
2559

2560 2561
	/*
	 * a higher idle thresh on the submit workers makes it much more
2562 2563 2564
	 * likely that bios will be send down in a sane order to the
	 * devices
	 */
2565 2566 2567 2568
	fs_info->submit_workers =
		btrfs_alloc_workqueue("submit", flags,
				      min_t(u64, fs_devices->num_devices,
					    max_active), 64);
2569

2570 2571
	fs_info->fixup_workers =
		btrfs_alloc_workqueue("fixup", flags, 1, 0);
2572 2573 2574 2575 2576

	/*
	 * endios are largely parallel and should have a very
	 * low idle thresh
	 */
2577 2578 2579 2580 2581 2582 2583 2584
	fs_info->endio_workers =
		btrfs_alloc_workqueue("endio", flags, max_active, 4);
	fs_info->endio_meta_workers =
		btrfs_alloc_workqueue("endio-meta", flags, max_active, 4);
	fs_info->endio_meta_write_workers =
		btrfs_alloc_workqueue("endio-meta-write", flags, max_active, 2);
	fs_info->endio_raid56_workers =
		btrfs_alloc_workqueue("endio-raid56", flags, max_active, 4);
2585 2586
	fs_info->endio_repair_workers =
		btrfs_alloc_workqueue("endio-repair", flags, 1, 0);
2587 2588
	fs_info->rmw_workers =
		btrfs_alloc_workqueue("rmw", flags, max_active, 2);
2589 2590 2591 2592
	fs_info->endio_write_workers =
		btrfs_alloc_workqueue("endio-write", flags, max_active, 2);
	fs_info->endio_freespace_worker =
		btrfs_alloc_workqueue("freespace-write", flags, max_active, 0);
2593 2594
	fs_info->delayed_workers =
		btrfs_alloc_workqueue("delayed-meta", flags, max_active, 0);
2595 2596
	fs_info->readahead_workers =
		btrfs_alloc_workqueue("readahead", flags, max_active, 2);
2597 2598
	fs_info->qgroup_rescan_workers =
		btrfs_alloc_workqueue("qgroup-rescan", flags, 1, 0);
C
Chris Mason 已提交
2599 2600 2601 2602
	fs_info->extent_workers =
		btrfs_alloc_workqueue("extent-refs", flags,
				      min_t(u64, fs_devices->num_devices,
					    max_active), 8);
2603

2604
	if (!(fs_info->workers && fs_info->delalloc_workers &&
2605 2606 2607
	      fs_info->submit_workers && fs_info->flush_workers &&
	      fs_info->endio_workers && fs_info->endio_meta_workers &&
	      fs_info->endio_meta_write_workers &&
2608
	      fs_info->endio_repair_workers &&
2609
	      fs_info->endio_write_workers && fs_info->endio_raid56_workers &&
2610
	      fs_info->endio_freespace_worker && fs_info->rmw_workers &&
2611
	      fs_info->caching_workers && fs_info->readahead_workers &&
2612
	      fs_info->fixup_workers && fs_info->delayed_workers &&
2613
	      fs_info->extent_workers &&
2614
	      fs_info->qgroup_rescan_workers)) {
2615
		err = -ENOMEM;
2616 2617
		goto fail_sb_buffer;
	}
2618

2619
	fs_info->bdi.ra_pages *= btrfs_super_num_devices(disk_super);
C
Chris Mason 已提交
2620 2621
	fs_info->bdi.ra_pages = max(fs_info->bdi.ra_pages,
				    4 * 1024 * 1024 / PAGE_CACHE_SIZE);
2622

2623 2624
	tree_root->nodesize = nodesize;
	tree_root->sectorsize = sectorsize;
2625
	tree_root->stripesize = stripesize;
2626 2627 2628

	sb->s_blocksize = sectorsize;
	sb->s_blocksize_bits = blksize_bits(sectorsize);
2629

2630
	if (btrfs_super_magic(disk_super) != BTRFS_MAGIC) {
2631
		printk(KERN_INFO "BTRFS: valid FS not found on %s\n", sb->s_id);
C
Chris Mason 已提交
2632 2633
		goto fail_sb_buffer;
	}
2634

2635
	if (sectorsize != PAGE_SIZE) {
2636
		printk(KERN_WARNING "BTRFS: Incompatible sector size(%lu) "
2637
		       "found on %s\n", (unsigned long)sectorsize, sb->s_id);
2638 2639 2640
		goto fail_sb_buffer;
	}

2641
	mutex_lock(&fs_info->chunk_mutex);
Y
Yan Zheng 已提交
2642
	ret = btrfs_read_sys_array(tree_root);
2643
	mutex_unlock(&fs_info->chunk_mutex);
2644
	if (ret) {
2645
		printk(KERN_WARNING "BTRFS: failed to read the system "
C
Chris Mason 已提交
2646
		       "array on %s\n", sb->s_id);
2647
		goto fail_sb_buffer;
2648
	}
2649

2650
	generation = btrfs_super_chunk_root_generation(disk_super);
2651

2652 2653
	__setup_root(nodesize, sectorsize, stripesize, chunk_root,
		     fs_info, BTRFS_CHUNK_TREE_OBJECTID);
2654 2655 2656

	chunk_root->node = read_tree_block(chunk_root,
					   btrfs_super_chunk_root(disk_super),
2657
					   generation);
2658 2659
	if (!chunk_root->node ||
	    !test_bit(EXTENT_BUFFER_UPTODATE, &chunk_root->node->bflags)) {
2660
		printk(KERN_WARNING "BTRFS: failed to read chunk root on %s\n",
2661
		       sb->s_id);
C
Chris Mason 已提交
2662
		goto fail_tree_roots;
2663
	}
2664 2665
	btrfs_set_root_node(&chunk_root->root_item, chunk_root->node);
	chunk_root->commit_root = btrfs_root_node(chunk_root);
2666

2667
	read_extent_buffer(chunk_root->node, fs_info->chunk_tree_uuid,
2668
	   btrfs_header_chunk_tree_uuid(chunk_root->node), BTRFS_UUID_SIZE);
2669

2670
	ret = btrfs_read_chunk_tree(chunk_root);
Y
Yan Zheng 已提交
2671
	if (ret) {
2672
		printk(KERN_WARNING "BTRFS: failed to read chunk tree on %s\n",
C
Chris Mason 已提交
2673
		       sb->s_id);
C
Chris Mason 已提交
2674
		goto fail_tree_roots;
Y
Yan Zheng 已提交
2675
	}
2676

2677 2678 2679 2680 2681
	/*
	 * 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);
2682

2683
	if (!fs_devices->latest_bdev) {
2684
		printk(KERN_CRIT "BTRFS: failed to read devices on %s\n",
2685 2686 2687 2688
		       sb->s_id);
		goto fail_tree_roots;
	}

C
Chris Mason 已提交
2689
retry_root_backup:
2690
	generation = btrfs_super_generation(disk_super);
2691

C
Chris Mason 已提交
2692
	tree_root->node = read_tree_block(tree_root,
2693
					  btrfs_super_root(disk_super),
2694
					  generation);
C
Chris Mason 已提交
2695 2696
	if (!tree_root->node ||
	    !test_bit(EXTENT_BUFFER_UPTODATE, &tree_root->node->bflags)) {
2697
		printk(KERN_WARNING "BTRFS: failed to read tree root on %s\n",
2698
		       sb->s_id);
C
Chris Mason 已提交
2699 2700

		goto recovery_tree_root;
2701
	}
C
Chris Mason 已提交
2702

2703 2704
	btrfs_set_root_node(&tree_root->root_item, tree_root->node);
	tree_root->commit_root = btrfs_root_node(tree_root);
2705
	btrfs_set_root_refs(&tree_root->root_item, 1);
2706

2707 2708 2709 2710 2711 2712 2713
	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
2714
		goto recovery_tree_root;
2715
	}
2716
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &extent_root->state);
2717
	fs_info->extent_root = extent_root;
2718

2719 2720 2721 2722
	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
2723
		goto recovery_tree_root;
2724
	}
2725
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &dev_root->state);
2726 2727
	fs_info->dev_root = dev_root;
	btrfs_init_devices_late(fs_info);
2728

2729 2730 2731 2732
	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
2733
		goto recovery_tree_root;
2734
	}
2735
	set_bit(BTRFS_ROOT_TRACK_DIRTY, &csum_root->state);
2736
	fs_info->csum_root = csum_root;
2737

2738 2739 2740
	location.objectid = BTRFS_QUOTA_TREE_OBJECTID;
	quota_root = btrfs_read_tree_root(tree_root, &location);
	if (!IS_ERR(quota_root)) {
2741
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &quota_root->state);
2742 2743
		fs_info->quota_enabled = 1;
		fs_info->pending_quota_state = 1;
2744
		fs_info->quota_root = quota_root;
2745 2746
	}

2747 2748 2749 2750 2751 2752 2753
	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;
2754
		check_uuid_tree = false;
2755
	} else {
2756
		set_bit(BTRFS_ROOT_TRACK_DIRTY, &uuid_root->state);
2757
		fs_info->uuid_root = uuid_root;
2758 2759 2760
		create_uuid_tree = false;
		check_uuid_tree =
		    generation != btrfs_super_uuid_tree_generation(disk_super);
2761 2762
	}

2763 2764 2765
	fs_info->generation = generation;
	fs_info->last_trans_committed = generation;

2766 2767
	ret = btrfs_recover_balance(fs_info);
	if (ret) {
2768
		printk(KERN_WARNING "BTRFS: failed to recover balance\n");
2769 2770 2771
		goto fail_block_groups;
	}

2772 2773
	ret = btrfs_init_dev_stats(fs_info);
	if (ret) {
2774
		printk(KERN_ERR "BTRFS: failed to init dev_stats: %d\n",
2775 2776 2777 2778
		       ret);
		goto fail_block_groups;
	}

2779 2780
	ret = btrfs_init_dev_replace(fs_info);
	if (ret) {
2781
		pr_err("BTRFS: failed to init dev_replace: %d\n", ret);
2782 2783 2784 2785 2786
		goto fail_block_groups;
	}

	btrfs_close_extra_devices(fs_info, fs_devices, 1);

2787
	ret = btrfs_sysfs_add_one(fs_info);
2788
	if (ret) {
2789
		pr_err("BTRFS: failed to init sysfs interface: %d\n", ret);
2790 2791 2792 2793 2794
		goto fail_block_groups;
	}

	ret = btrfs_init_space_info(fs_info);
	if (ret) {
2795
		printk(KERN_ERR "BTRFS: Failed to initial space info: %d\n", ret);
2796
		goto fail_sysfs;
2797 2798
	}

2799 2800
	ret = btrfs_read_block_groups(extent_root);
	if (ret) {
2801
		printk(KERN_ERR "BTRFS: Failed to read block groups: %d\n", ret);
2802
		goto fail_sysfs;
2803
	}
2804 2805
	fs_info->num_tolerated_disk_barrier_failures =
		btrfs_calc_num_tolerated_disk_barrier_failures(fs_info);
2806 2807 2808
	if (fs_info->fs_devices->missing_devices >
	     fs_info->num_tolerated_disk_barrier_failures &&
	    !(sb->s_flags & MS_RDONLY)) {
2809 2810
		printk(KERN_WARNING "BTRFS: "
			"too many missing devices, writeable mount is not allowed\n");
2811
		goto fail_sysfs;
2812
	}
C
Chris Mason 已提交
2813

2814 2815
	fs_info->cleaner_kthread = kthread_run(cleaner_kthread, tree_root,
					       "btrfs-cleaner");
2816
	if (IS_ERR(fs_info->cleaner_kthread))
2817
		goto fail_sysfs;
2818 2819 2820 2821

	fs_info->transaction_kthread = kthread_run(transaction_kthread,
						   tree_root,
						   "btrfs-transaction");
2822
	if (IS_ERR(fs_info->transaction_kthread))
2823
		goto fail_cleaner;
2824

C
Chris Mason 已提交
2825 2826 2827
	if (!btrfs_test_opt(tree_root, SSD) &&
	    !btrfs_test_opt(tree_root, NOSSD) &&
	    !fs_info->fs_devices->rotating) {
2828
		printk(KERN_INFO "BTRFS: detected SSD devices, enabling SSD "
C
Chris Mason 已提交
2829 2830 2831 2832
		       "mode\n");
		btrfs_set_opt(fs_info->mount_opt, SSD);
	}

2833 2834 2835 2836
	/* Set the real inode map cache flag */
	if (btrfs_test_opt(tree_root, CHANGE_INODE_CACHE))
		btrfs_set_opt(tree_root->fs_info->mount_opt, INODE_MAP_CACHE);

2837 2838 2839 2840 2841 2842 2843 2844
#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)
2845
			printk(KERN_WARNING "BTRFS: failed to initialize"
2846 2847 2848
			       " integrity check module %s\n", sb->s_id);
	}
#endif
2849 2850 2851
	ret = btrfs_read_qgroup_config(fs_info);
	if (ret)
		goto fail_trans_kthread;
2852

L
liubo 已提交
2853
	/* do not make disk changes in broken FS */
2854
	if (btrfs_super_log_root(disk_super) != 0) {
2855 2856
		u64 bytenr = btrfs_super_log_root(disk_super);

2857
		if (fs_devices->rw_devices == 0) {
2858
			printk(KERN_WARNING "BTRFS: log replay required "
C
Chris Mason 已提交
2859
			       "on RO media\n");
2860
			err = -EIO;
2861
			goto fail_qgroup;
2862
		}
C
Chris Mason 已提交
2863

A
Al Viro 已提交
2864
		log_tree_root = btrfs_alloc_root(fs_info);
2865 2866
		if (!log_tree_root) {
			err = -ENOMEM;
2867
			goto fail_qgroup;
2868
		}
2869

2870
		__setup_root(nodesize, sectorsize, stripesize,
2871 2872 2873
			     log_tree_root, fs_info, BTRFS_TREE_LOG_OBJECTID);

		log_tree_root->node = read_tree_block(tree_root, bytenr,
2874
						      generation + 1);
2875 2876
		if (!log_tree_root->node ||
		    !extent_buffer_uptodate(log_tree_root->node)) {
2877
			printk(KERN_ERR "BTRFS: failed to read log tree\n");
2878 2879
			free_extent_buffer(log_tree_root->node);
			kfree(log_tree_root);
2880
			goto fail_qgroup;
2881
		}
2882
		/* returns with log_tree_root freed on success */
2883
		ret = btrfs_recover_log_trees(log_tree_root);
2884 2885 2886 2887 2888
		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);
2889
			goto fail_qgroup;
2890
		}
2891 2892

		if (sb->s_flags & MS_RDONLY) {
2893 2894
			ret = btrfs_commit_super(tree_root);
			if (ret)
2895
				goto fail_qgroup;
2896
		}
2897
	}
Z
Zheng Yan 已提交
2898

2899
	ret = btrfs_find_orphan_roots(tree_root);
2900
	if (ret)
2901
		goto fail_qgroup;
2902

2903
	if (!(sb->s_flags & MS_RDONLY)) {
2904
		ret = btrfs_cleanup_fs_roots(fs_info);
2905
		if (ret)
2906
			goto fail_qgroup;
2907

2908
		mutex_lock(&fs_info->cleaner_mutex);
2909
		ret = btrfs_recover_relocation(tree_root);
2910
		mutex_unlock(&fs_info->cleaner_mutex);
2911 2912
		if (ret < 0) {
			printk(KERN_WARNING
2913
			       "BTRFS: failed to recover relocation\n");
2914
			err = -EINVAL;
2915
			goto fail_qgroup;
2916
		}
2917
	}
Z
Zheng Yan 已提交
2918

2919 2920
	location.objectid = BTRFS_FS_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
2921
	location.offset = 0;
2922 2923

	fs_info->fs_root = btrfs_read_fs_root_no_name(fs_info, &location);
2924 2925
	if (IS_ERR(fs_info->fs_root)) {
		err = PTR_ERR(fs_info->fs_root);
2926
		goto fail_qgroup;
2927
	}
C
Chris Mason 已提交
2928

2929 2930
	if (sb->s_flags & MS_RDONLY)
		return 0;
I
Ilya Dryomov 已提交
2931

2932 2933 2934
	down_read(&fs_info->cleanup_work_sem);
	if ((ret = btrfs_orphan_cleanup(fs_info->fs_root)) ||
	    (ret = btrfs_orphan_cleanup(fs_info->tree_root))) {
2935
		up_read(&fs_info->cleanup_work_sem);
2936 2937 2938 2939
		close_ctree(tree_root);
		return ret;
	}
	up_read(&fs_info->cleanup_work_sem);
I
Ilya Dryomov 已提交
2940

2941 2942
	ret = btrfs_resume_balance_async(fs_info);
	if (ret) {
2943
		printk(KERN_WARNING "BTRFS: failed to resume balance\n");
2944 2945
		close_ctree(tree_root);
		return ret;
2946 2947
	}

2948 2949
	ret = btrfs_resume_dev_replace_async(fs_info);
	if (ret) {
2950
		pr_warn("BTRFS: failed to resume dev_replace\n");
2951 2952 2953 2954
		close_ctree(tree_root);
		return ret;
	}

2955 2956
	btrfs_qgroup_rescan_resume(fs_info);

2957
	if (create_uuid_tree) {
2958
		pr_info("BTRFS: creating UUID tree\n");
2959 2960
		ret = btrfs_create_uuid_tree(fs_info);
		if (ret) {
2961
			pr_warn("BTRFS: failed to create the UUID tree %d\n",
2962 2963 2964 2965
				ret);
			close_ctree(tree_root);
			return ret;
		}
2966 2967
	} else if (check_uuid_tree ||
		   btrfs_test_opt(tree_root, RESCAN_UUID_TREE)) {
2968
		pr_info("BTRFS: checking UUID tree\n");
2969 2970
		ret = btrfs_check_uuid_tree(fs_info);
		if (ret) {
2971
			pr_warn("BTRFS: failed to check the UUID tree %d\n",
2972 2973 2974 2975 2976 2977
				ret);
			close_ctree(tree_root);
			return ret;
		}
	} else {
		fs_info->update_uuid_tree_gen = 1;
2978 2979
	}

2980 2981
	fs_info->open = 1;

A
Al Viro 已提交
2982
	return 0;
C
Chris Mason 已提交
2983

2984 2985
fail_qgroup:
	btrfs_free_qgroup_config(fs_info);
2986 2987
fail_trans_kthread:
	kthread_stop(fs_info->transaction_kthread);
J
Josef Bacik 已提交
2988
	btrfs_cleanup_transaction(fs_info->tree_root);
2989
	btrfs_free_fs_roots(fs_info);
2990
fail_cleaner:
2991
	kthread_stop(fs_info->cleaner_kthread);
2992 2993 2994 2995 2996 2997 2998

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

2999 3000 3001
fail_sysfs:
	btrfs_sysfs_remove_one(fs_info);

3002
fail_block_groups:
J
Josef Bacik 已提交
3003
	btrfs_put_block_group_cache(fs_info);
3004
	btrfs_free_block_groups(fs_info);
C
Chris Mason 已提交
3005 3006 3007

fail_tree_roots:
	free_root_pointers(fs_info, 1);
3008
	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
3009

C
Chris Mason 已提交
3010
fail_sb_buffer:
L
Liu Bo 已提交
3011
	btrfs_stop_all_workers(fs_info);
3012
fail_alloc:
3013
fail_iput:
3014 3015
	btrfs_mapping_tree_free(&fs_info->mapping_tree);

3016
	iput(fs_info->btree_inode);
3017 3018
fail_bio_counter:
	percpu_counter_destroy(&fs_info->bio_counter);
3019 3020
fail_delalloc_bytes:
	percpu_counter_destroy(&fs_info->delalloc_bytes);
3021 3022
fail_dirty_metadata_bytes:
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
3023
fail_bdi:
3024
	bdi_destroy(&fs_info->bdi);
3025 3026
fail_srcu:
	cleanup_srcu_struct(&fs_info->subvol_srcu);
3027
fail:
D
David Woodhouse 已提交
3028
	btrfs_free_stripe_hash_table(fs_info);
3029
	btrfs_close_devices(fs_info->fs_devices);
A
Al Viro 已提交
3030
	return err;
C
Chris Mason 已提交
3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048

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;
3049 3050
}

3051 3052 3053 3054 3055
static void btrfs_end_buffer_write_sync(struct buffer_head *bh, int uptodate)
{
	if (uptodate) {
		set_buffer_uptodate(bh);
	} else {
3056 3057 3058
		struct btrfs_device *device = (struct btrfs_device *)
			bh->b_private;

3059
		printk_ratelimited_in_rcu(KERN_WARNING "BTRFS: lost page write due to "
3060 3061
					  "I/O error on %s\n",
					  rcu_str_deref(device->name));
3062 3063 3064
		/* note, we dont' set_buffer_write_io_error because we have
		 * our own ways of dealing with the IO errors
		 */
3065
		clear_buffer_uptodate(bh);
3066
		btrfs_dev_stat_inc_and_print(device, BTRFS_DEV_STAT_WRITE_ERRS);
3067 3068 3069 3070 3071
	}
	unlock_buffer(bh);
	put_bh(bh);
}

Y
Yan Zheng 已提交
3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087
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);
3088 3089
		if (bytenr + BTRFS_SUPER_INFO_SIZE >=
					i_size_read(bdev->bd_inode))
Y
Yan Zheng 已提交
3090
			break;
3091 3092
		bh = __bread(bdev, bytenr / 4096,
					BTRFS_SUPER_INFO_SIZE);
Y
Yan Zheng 已提交
3093 3094 3095 3096 3097
		if (!bh)
			continue;

		super = (struct btrfs_super_block *)bh->b_data;
		if (btrfs_super_bytenr(super) != bytenr ||
3098
		    btrfs_super_magic(super) != BTRFS_MAGIC) {
Y
Yan Zheng 已提交
3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113
			brelse(bh);
			continue;
		}

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

3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124
/*
 * 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
3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140
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);
3141 3142
		if (bytenr + BTRFS_SUPER_INFO_SIZE >=
		    device->commit_total_bytes)
Y
Yan Zheng 已提交
3143 3144 3145 3146 3147
			break;

		if (wait) {
			bh = __find_get_block(device->bdev, bytenr / 4096,
					      BTRFS_SUPER_INFO_SIZE);
3148 3149 3150 3151
			if (!bh) {
				errors++;
				continue;
			}
Y
Yan Zheng 已提交
3152
			wait_on_buffer(bh);
3153 3154 3155 3156 3157 3158 3159 3160 3161
			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
3162 3163 3164 3165
		} else {
			btrfs_set_super_bytenr(sb, bytenr);

			crc = ~(u32)0;
3166
			crc = btrfs_csum_data((char *)sb +
Y
Yan Zheng 已提交
3167 3168 3169 3170 3171
					      BTRFS_CSUM_SIZE, crc,
					      BTRFS_SUPER_INFO_SIZE -
					      BTRFS_CSUM_SIZE);
			btrfs_csum_final(crc, sb->csum);

3172 3173 3174 3175
			/*
			 * one reference for us, and we leave it for the
			 * caller
			 */
Y
Yan Zheng 已提交
3176 3177
			bh = __getblk(device->bdev, bytenr / 4096,
				      BTRFS_SUPER_INFO_SIZE);
3178
			if (!bh) {
3179
				printk(KERN_ERR "BTRFS: couldn't get super "
3180 3181 3182 3183 3184
				       "buffer head for bytenr %Lu\n", bytenr);
				errors++;
				continue;
			}

Y
Yan Zheng 已提交
3185 3186
			memcpy(bh->b_data, sb, BTRFS_SUPER_INFO_SIZE);

3187
			/* one reference for submit_bh */
Y
Yan Zheng 已提交
3188
			get_bh(bh);
3189 3190

			set_buffer_uptodate(bh);
Y
Yan Zheng 已提交
3191 3192
			lock_buffer(bh);
			bh->b_end_io = btrfs_end_buffer_write_sync;
3193
			bh->b_private = device;
Y
Yan Zheng 已提交
3194 3195
		}

C
Chris Mason 已提交
3196 3197 3198 3199
		/*
		 * we fua the first super.  The others we allow
		 * to go down lazy.
		 */
3200 3201 3202 3203
		if (i == 0)
			ret = btrfsic_submit_bh(WRITE_FUA, bh);
		else
			ret = btrfsic_submit_bh(WRITE_SYNC, bh);
3204
		if (ret)
Y
Yan Zheng 已提交
3205 3206 3207 3208 3209
			errors++;
	}
	return errors < i ? 0 : -1;
}

C
Chris Mason 已提交
3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248
/*
 * 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)) {
3249
			printk_in_rcu("BTRFS: disabling barriers on dev %s\n",
3250
				      rcu_str_deref(device->name));
C
Chris Mason 已提交
3251
			device->nobarriers = 1;
3252
		} else if (!bio_flagged(bio, BIO_UPTODATE)) {
C
Chris Mason 已提交
3253
			ret = -EIO;
3254 3255
			btrfs_dev_stat_inc_and_print(device,
				BTRFS_DEV_STAT_FLUSH_ERRS);
C
Chris Mason 已提交
3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268
		}

		/* 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
	 */
3269
	device->flush_bio = NULL;
3270
	bio = btrfs_io_bio_alloc(GFP_NOFS, 0);
C
Chris Mason 已提交
3271 3272 3273 3274 3275 3276 3277 3278 3279 3280
	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);
3281
	btrfsic_submit_bio(WRITE_FLUSH, bio);
C
Chris Mason 已提交
3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293

	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;
3294 3295
	int errors_send = 0;
	int errors_wait = 0;
C
Chris Mason 已提交
3296 3297 3298 3299 3300
	int ret;

	/* send down all the barriers */
	head = &info->fs_devices->devices;
	list_for_each_entry_rcu(dev, head, dev_list) {
3301 3302
		if (dev->missing)
			continue;
C
Chris Mason 已提交
3303
		if (!dev->bdev) {
3304
			errors_send++;
C
Chris Mason 已提交
3305 3306 3307 3308 3309 3310 3311
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

		ret = write_dev_flush(dev, 0);
		if (ret)
3312
			errors_send++;
C
Chris Mason 已提交
3313 3314 3315 3316
	}

	/* wait for all the barriers */
	list_for_each_entry_rcu(dev, head, dev_list) {
3317 3318
		if (dev->missing)
			continue;
C
Chris Mason 已提交
3319
		if (!dev->bdev) {
3320
			errors_wait++;
C
Chris Mason 已提交
3321 3322 3323 3324 3325 3326 3327
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

		ret = write_dev_flush(dev, 1);
		if (ret)
3328
			errors_wait++;
C
Chris Mason 已提交
3329
	}
3330 3331
	if (errors_send > info->num_tolerated_disk_barrier_failures ||
	    errors_wait > info->num_tolerated_disk_barrier_failures)
C
Chris Mason 已提交
3332 3333 3334 3335
		return -EIO;
	return 0;
}

3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395
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
3396 3397 3398 3399 3400 3401
				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 &
3402
						   BTRFS_BLOCK_GROUP_RAID6) {
D
David Woodhouse 已提交
3403 3404 3405
						num_tolerated_disk_barrier_failures = 2;
					}
				}
3406 3407 3408 3409 3410 3411 3412 3413
			}
		}
		up_read(&sinfo->groups_sem);
	}

	return num_tolerated_disk_barrier_failures;
}

3414
static int write_all_supers(struct btrfs_root *root, int max_mirrors)
3415
{
3416
	struct list_head *head;
3417
	struct btrfs_device *dev;
3418
	struct btrfs_super_block *sb;
3419 3420 3421
	struct btrfs_dev_item *dev_item;
	int ret;
	int do_barriers;
3422 3423
	int max_errors;
	int total_errors = 0;
3424
	u64 flags;
3425 3426

	do_barriers = !btrfs_test_opt(root, NOBARRIER);
C
Chris Mason 已提交
3427
	backup_super_roots(root->fs_info);
3428

3429
	sb = root->fs_info->super_for_commit;
3430
	dev_item = &sb->dev_item;
3431

3432
	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
3433
	head = &root->fs_info->fs_devices->devices;
3434
	max_errors = btrfs_super_num_devices(root->fs_info->super_copy) - 1;
C
Chris Mason 已提交
3435

3436 3437 3438 3439 3440 3441 3442 3443 3444 3445
	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
3446

3447
	list_for_each_entry_rcu(dev, head, dev_list) {
3448 3449 3450 3451
		if (!dev->bdev) {
			total_errors++;
			continue;
		}
Y
Yan Zheng 已提交
3452
		if (!dev->in_fs_metadata || !dev->writeable)
3453 3454
			continue;

Y
Yan Zheng 已提交
3455
		btrfs_set_stack_device_generation(dev_item, 0);
3456 3457
		btrfs_set_stack_device_type(dev_item, dev->type);
		btrfs_set_stack_device_id(dev_item, dev->devid);
3458
		btrfs_set_stack_device_total_bytes(dev_item,
3459
						   dev->commit_total_bytes);
3460 3461
		btrfs_set_stack_device_bytes_used(dev_item,
						  dev->commit_bytes_used);
3462 3463 3464 3465
		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
3466
		memcpy(dev_item->fsid, dev->fs_devices->fsid, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3467

3468 3469 3470
		flags = btrfs_super_flags(sb);
		btrfs_set_super_flags(sb, flags | BTRFS_HEADER_FLAG_WRITTEN);

Y
Yan Zheng 已提交
3471
		ret = write_dev_supers(dev, sb, do_barriers, 0, max_mirrors);
3472 3473
		if (ret)
			total_errors++;
3474
	}
3475
	if (total_errors > max_errors) {
3476
		btrfs_err(root->fs_info, "%d errors while writing supers",
C
Chris Mason 已提交
3477
		       total_errors);
3478
		mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
3479

3480 3481 3482 3483
		/* 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;
3484
	}
3485

Y
Yan Zheng 已提交
3486
	total_errors = 0;
3487
	list_for_each_entry_rcu(dev, head, dev_list) {
3488 3489
		if (!dev->bdev)
			continue;
Y
Yan Zheng 已提交
3490
		if (!dev->in_fs_metadata || !dev->writeable)
3491 3492
			continue;

Y
Yan Zheng 已提交
3493 3494 3495
		ret = write_dev_supers(dev, sb, do_barriers, 1, max_mirrors);
		if (ret)
			total_errors++;
3496
	}
3497
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
3498
	if (total_errors > max_errors) {
3499 3500 3501
		btrfs_error(root->fs_info, -EIO,
			    "%d errors while writing supers", total_errors);
		return -EIO;
3502
	}
3503 3504 3505
	return 0;
}

Y
Yan Zheng 已提交
3506 3507
int write_ctree_super(struct btrfs_trans_handle *trans,
		      struct btrfs_root *root, int max_mirrors)
3508
{
3509
	return write_all_supers(root, max_mirrors);
C
Chris Mason 已提交
3510 3511
}

3512 3513 3514
/* 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
3515
{
3516
	spin_lock(&fs_info->fs_roots_radix_lock);
C
Chris Mason 已提交
3517 3518
	radix_tree_delete(&fs_info->fs_roots_radix,
			  (unsigned long)root->root_key.objectid);
3519
	spin_unlock(&fs_info->fs_roots_radix_lock);
3520 3521 3522 3523

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

3524
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state))
L
Liu Bo 已提交
3525 3526
		btrfs_free_log(NULL, root);

3527 3528 3529 3530
	if (root->free_ino_pinned)
		__btrfs_remove_free_space_cache(root->free_ino_pinned);
	if (root->free_ino_ctl)
		__btrfs_remove_free_space_cache(root->free_ino_ctl);
3531 3532 3533 3534 3535
	free_fs_root(root);
}

static void free_fs_root(struct btrfs_root *root)
{
3536
	iput(root->ino_cache_inode);
3537
	WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree));
3538 3539
	btrfs_free_block_rsv(root, root->orphan_block_rsv);
	root->orphan_block_rsv = NULL;
3540 3541
	if (root->anon_dev)
		free_anon_bdev(root->anon_dev);
3542 3543
	if (root->subv_writers)
		btrfs_free_subvolume_writers(root->subv_writers);
3544 3545
	free_extent_buffer(root->node);
	free_extent_buffer(root->commit_root);
3546 3547
	kfree(root->free_ino_ctl);
	kfree(root->free_ino_pinned);
C
Chris Mason 已提交
3548
	kfree(root->name);
3549
	btrfs_put_fs_root(root);
C
Chris Mason 已提交
3550 3551
}

3552 3553 3554
void btrfs_free_fs_root(struct btrfs_root *root)
{
	free_fs_root(root);
C
Chris Mason 已提交
3555 3556
}

Y
Yan Zheng 已提交
3557
int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
C
Chris Mason 已提交
3558
{
Y
Yan Zheng 已提交
3559 3560
	u64 root_objectid = 0;
	struct btrfs_root *gang[8];
3561 3562 3563 3564
	int i = 0;
	int err = 0;
	unsigned int ret = 0;
	int index;
3565

Y
Yan Zheng 已提交
3566
	while (1) {
3567
		index = srcu_read_lock(&fs_info->subvol_srcu);
Y
Yan Zheng 已提交
3568 3569 3570
		ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix,
					     (void **)gang, root_objectid,
					     ARRAY_SIZE(gang));
3571 3572
		if (!ret) {
			srcu_read_unlock(&fs_info->subvol_srcu, index);
Y
Yan Zheng 已提交
3573
			break;
3574
		}
3575
		root_objectid = gang[ret - 1]->root_key.objectid + 1;
3576

Y
Yan Zheng 已提交
3577
		for (i = 0; i < ret; i++) {
3578 3579 3580 3581 3582 3583 3584 3585 3586
			/* Avoid to grab roots in dead_roots */
			if (btrfs_root_refs(&gang[i]->root_item) == 0) {
				gang[i] = NULL;
				continue;
			}
			/* grab all the search result for later use */
			gang[i] = btrfs_grab_fs_root(gang[i]);
		}
		srcu_read_unlock(&fs_info->subvol_srcu, index);
3587

3588 3589 3590
		for (i = 0; i < ret; i++) {
			if (!gang[i])
				continue;
Y
Yan Zheng 已提交
3591
			root_objectid = gang[i]->root_key.objectid;
3592 3593
			err = btrfs_orphan_cleanup(gang[i]);
			if (err)
3594 3595
				break;
			btrfs_put_fs_root(gang[i]);
Y
Yan Zheng 已提交
3596 3597 3598
		}
		root_objectid++;
	}
3599 3600 3601 3602 3603 3604 3605

	/* release the uncleaned roots due to error */
	for (; i < ret; i++) {
		if (gang[i])
			btrfs_put_fs_root(gang[i]);
	}
	return err;
Y
Yan Zheng 已提交
3606
}
3607

Y
Yan Zheng 已提交
3608 3609 3610
int btrfs_commit_super(struct btrfs_root *root)
{
	struct btrfs_trans_handle *trans;
3611

Y
Yan Zheng 已提交
3612
	mutex_lock(&root->fs_info->cleaner_mutex);
Y
Yan, Zheng 已提交
3613
	btrfs_run_delayed_iputs(root);
Y
Yan Zheng 已提交
3614
	mutex_unlock(&root->fs_info->cleaner_mutex);
D
David Sterba 已提交
3615
	wake_up_process(root->fs_info->cleaner_kthread);
3616 3617 3618 3619 3620

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

3621
	trans = btrfs_join_transaction(root);
3622 3623
	if (IS_ERR(trans))
		return PTR_ERR(trans);
3624
	return btrfs_commit_transaction(trans, root);
Y
Yan Zheng 已提交
3625 3626
}

3627
void close_ctree(struct btrfs_root *root)
Y
Yan Zheng 已提交
3628 3629 3630 3631 3632 3633 3634
{
	struct btrfs_fs_info *fs_info = root->fs_info;
	int ret;

	fs_info->closing = 1;
	smp_mb();

S
Stefan Behrens 已提交
3635 3636 3637 3638 3639
	/* 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);

3640
	/* pause restriper - we want to resume on mount */
3641
	btrfs_pause_balance(fs_info);
3642

3643 3644
	btrfs_dev_replace_suspend_for_unmount(fs_info);

3645
	btrfs_scrub_cancel(fs_info);
C
Chris Mason 已提交
3646 3647 3648 3649 3650 3651

	/* 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 */
3652
	btrfs_cleanup_defrag_inodes(fs_info);
C
Chris Mason 已提交
3653

3654 3655
	cancel_work_sync(&fs_info->async_reclaim_work);

Y
Yan Zheng 已提交
3656
	if (!(fs_info->sb->s_flags & MS_RDONLY)) {
L
liubo 已提交
3657 3658
		ret = btrfs_commit_super(root);
		if (ret)
3659
			btrfs_err(root->fs_info, "commit super ret %d", ret);
L
liubo 已提交
3660 3661
	}

3662
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state))
3663
		btrfs_error_commit_super(root);
3664

A
Al Viro 已提交
3665 3666
	kthread_stop(fs_info->transaction_kthread);
	kthread_stop(fs_info->cleaner_kthread);
3667

3668 3669 3670
	fs_info->closing = 2;
	smp_mb();

3671 3672
	btrfs_free_qgroup_config(root->fs_info);

3673
	if (percpu_counter_sum(&fs_info->delalloc_bytes)) {
3674
		btrfs_info(root->fs_info, "at unmount delalloc count %lld",
3675
		       percpu_counter_sum(&fs_info->delalloc_bytes));
C
Chris Mason 已提交
3676
	}
3677

3678 3679
	btrfs_sysfs_remove_one(fs_info);

3680
	btrfs_free_fs_roots(fs_info);
3681

3682 3683
	btrfs_put_block_group_cache(fs_info);

3684 3685
	btrfs_free_block_groups(fs_info);

3686 3687 3688 3689 3690
	/*
	 * we must make sure there is not any read request to
	 * submit after we stopping all workers.
	 */
	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
3691 3692
	btrfs_stop_all_workers(fs_info);

3693
	fs_info->open = 0;
3694
	free_root_pointers(fs_info, 1);
3695

3696
	iput(fs_info->btree_inode);
3697

3698 3699 3700 3701 3702
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	if (btrfs_test_opt(root, CHECK_INTEGRITY))
		btrfsic_unmount(root, fs_info->fs_devices);
#endif

3703
	btrfs_close_devices(fs_info->fs_devices);
3704
	btrfs_mapping_tree_free(&fs_info->mapping_tree);
3705

3706
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
3707
	percpu_counter_destroy(&fs_info->delalloc_bytes);
3708
	percpu_counter_destroy(&fs_info->bio_counter);
C
Chris Mason 已提交
3709
	bdi_destroy(&fs_info->bdi);
3710
	cleanup_srcu_struct(&fs_info->subvol_srcu);
3711

D
David Woodhouse 已提交
3712 3713
	btrfs_free_stripe_hash_table(fs_info);

3714 3715
	btrfs_free_block_rsv(root, root->orphan_block_rsv);
	root->orphan_block_rsv = NULL;
3716 3717
}

3718 3719
int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid,
			  int atomic)
3720
{
3721
	int ret;
3722
	struct inode *btree_inode = buf->pages[0]->mapping->host;
3723

3724
	ret = extent_buffer_uptodate(buf);
3725 3726 3727 3728
	if (!ret)
		return ret;

	ret = verify_parent_transid(&BTRFS_I(btree_inode)->io_tree, buf,
3729 3730 3731
				    parent_transid, atomic);
	if (ret == -EAGAIN)
		return ret;
3732
	return !ret;
3733 3734 3735
}

int btrfs_set_buffer_uptodate(struct extent_buffer *buf)
C
Chris Mason 已提交
3736
{
3737
	return set_extent_buffer_uptodate(buf);
3738
}
3739

3740 3741
void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
{
3742
	struct btrfs_root *root;
3743
	u64 transid = btrfs_header_generation(buf);
3744
	int was_dirty;
3745

3746 3747 3748 3749 3750 3751 3752 3753 3754 3755
#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;
3756
	btrfs_assert_tree_locked(buf);
J
Julia Lawall 已提交
3757 3758
	if (transid != root->fs_info->generation)
		WARN(1, KERN_CRIT "btrfs transid mismatch buffer %llu, "
C
Chris Mason 已提交
3759
		       "found %llu running %llu\n",
3760
			buf->start, transid, root->fs_info->generation);
3761
	was_dirty = set_extent_buffer_dirty(buf);
3762 3763 3764 3765
	if (!was_dirty)
		__percpu_counter_add(&root->fs_info->dirty_metadata_bytes,
				     buf->len,
				     root->fs_info->dirty_metadata_batch);
3766 3767 3768 3769 3770 3771
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	if (btrfs_header_level(buf) == 0 && check_leaf(root, buf)) {
		btrfs_print_leaf(root, buf);
		ASSERT(0);
	}
#endif
3772 3773
}

3774 3775
static void __btrfs_btree_balance_dirty(struct btrfs_root *root,
					int flush_delayed)
3776 3777 3778 3779 3780
{
	/*
	 * looks as though older kernels can get into trouble with
	 * this code, they end up stuck in balance_dirty_pages forever
	 */
3781
	int ret;
3782 3783 3784 3785

	if (current->flags & PF_MEMALLOC)
		return;

3786 3787
	if (flush_delayed)
		btrfs_balance_delayed_items(root);
3788

3789 3790 3791
	ret = percpu_counter_compare(&root->fs_info->dirty_metadata_bytes,
				     BTRFS_DIRTY_METADATA_THRESH);
	if (ret > 0) {
3792 3793
		balance_dirty_pages_ratelimited(
				   root->fs_info->btree_inode->i_mapping);
3794 3795 3796 3797
	}
	return;
}

3798
void btrfs_btree_balance_dirty(struct btrfs_root *root)
C
Chris Mason 已提交
3799
{
3800 3801
	__btrfs_btree_balance_dirty(root, 1);
}
3802

3803 3804 3805
void btrfs_btree_balance_dirty_nodelay(struct btrfs_root *root)
{
	__btrfs_btree_balance_dirty(root, 0);
C
Chris Mason 已提交
3806
}
3807

3808
int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid)
3809
{
3810
	struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root;
3811
	return btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
3812
}
3813

3814
static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info,
L
liubo 已提交
3815 3816
			      int read_only)
{
D
David Sterba 已提交
3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835
	struct btrfs_super_block *sb = fs_info->super_copy;
	int ret = 0;

	if (sb->root_level > BTRFS_MAX_LEVEL) {
		printk(KERN_ERR "BTRFS: tree_root level too big: %d > %d\n",
				sb->root_level, BTRFS_MAX_LEVEL);
		ret = -EINVAL;
	}
	if (sb->chunk_root_level > BTRFS_MAX_LEVEL) {
		printk(KERN_ERR "BTRFS: chunk_root level too big: %d > %d\n",
				sb->chunk_root_level, BTRFS_MAX_LEVEL);
		ret = -EINVAL;
	}
	if (sb->log_root_level > BTRFS_MAX_LEVEL) {
		printk(KERN_ERR "BTRFS: log_root level too big: %d > %d\n",
				sb->log_root_level, BTRFS_MAX_LEVEL);
		ret = -EINVAL;
	}

D
David Sterba 已提交
3836
	/*
D
David Sterba 已提交
3837 3838
	 * The common minimum, we don't know if we can trust the nodesize/sectorsize
	 * items yet, they'll be verified later. Issue just a warning.
D
David Sterba 已提交
3839
	 */
D
David Sterba 已提交
3840 3841 3842 3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883
	if (!IS_ALIGNED(sb->root, 4096))
		printk(KERN_WARNING "BTRFS: tree_root block unaligned: %llu\n",
				sb->root);
	if (!IS_ALIGNED(sb->chunk_root, 4096))
		printk(KERN_WARNING "BTRFS: tree_root block unaligned: %llu\n",
				sb->chunk_root);
	if (!IS_ALIGNED(sb->log_root, 4096))
		printk(KERN_WARNING "BTRFS: tree_root block unaligned: %llu\n",
				sb->log_root);

	if (memcmp(fs_info->fsid, sb->dev_item.fsid, BTRFS_UUID_SIZE) != 0) {
		printk(KERN_ERR "BTRFS: dev_item UUID does not match fsid: %pU != %pU\n",
				fs_info->fsid, sb->dev_item.fsid);
		ret = -EINVAL;
	}

	/*
	 * Hint to catch really bogus numbers, bitflips or so, more exact checks are
	 * done later
	 */
	if (sb->num_devices > (1UL << 31))
		printk(KERN_WARNING "BTRFS: suspicious number of devices: %llu\n",
				sb->num_devices);

	if (sb->bytenr != BTRFS_SUPER_INFO_OFFSET) {
		printk(KERN_ERR "BTRFS: super offset mismatch %llu != %u\n",
				sb->bytenr, BTRFS_SUPER_INFO_OFFSET);
		ret = -EINVAL;
	}

	/*
	 * The generation is a global counter, we'll trust it more than the others
	 * but it's still possible that it's the one that's wrong.
	 */
	if (sb->generation < sb->chunk_root_generation)
		printk(KERN_WARNING
			"BTRFS: suspicious: generation < chunk_root_generation: %llu < %llu\n",
			sb->generation, sb->chunk_root_generation);
	if (sb->generation < sb->cache_generation && sb->cache_generation != (u64)-1)
		printk(KERN_WARNING
			"BTRFS: suspicious: generation < cache_generation: %llu < %llu\n",
			sb->generation, sb->cache_generation);

	return ret;
L
liubo 已提交
3884 3885
}

3886
static void btrfs_error_commit_super(struct btrfs_root *root)
L
liubo 已提交
3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898
{
	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);
}

3899
static void btrfs_destroy_ordered_extents(struct btrfs_root *root)
L
liubo 已提交
3900 3901 3902
{
	struct btrfs_ordered_extent *ordered;

3903
	spin_lock(&root->ordered_extent_lock);
3904 3905 3906 3907
	/*
	 * This will just short circuit the ordered completion stuff which will
	 * make sure the ordered extent gets properly cleaned up.
	 */
3908
	list_for_each_entry(ordered, &root->ordered_extents,
3909 3910
			    root_extent_list)
		set_bit(BTRFS_ORDERED_IOERR, &ordered->flags);
3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925
	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);
3926 3927
		list_move_tail(&root->ordered_root,
			       &fs_info->ordered_roots);
3928

3929
		spin_unlock(&fs_info->ordered_root_lock);
3930 3931
		btrfs_destroy_ordered_extents(root);

3932 3933
		cond_resched();
		spin_lock(&fs_info->ordered_root_lock);
3934 3935
	}
	spin_unlock(&fs_info->ordered_root_lock);
L
liubo 已提交
3936 3937
}

3938 3939
static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
				      struct btrfs_root *root)
L
liubo 已提交
3940 3941 3942 3943 3944 3945 3946 3947 3948
{
	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);
3949
	if (atomic_read(&delayed_refs->num_entries) == 0) {
3950
		spin_unlock(&delayed_refs->lock);
3951
		btrfs_info(root->fs_info, "delayed_refs has NO entry");
L
liubo 已提交
3952 3953 3954
		return ret;
	}

3955 3956
	while ((node = rb_first(&delayed_refs->href_root)) != NULL) {
		struct btrfs_delayed_ref_head *head;
3957
		bool pin_bytes = false;
L
liubo 已提交
3958

3959 3960 3961 3962 3963
		head = rb_entry(node, struct btrfs_delayed_ref_head,
				href_node);
		if (!mutex_trylock(&head->mutex)) {
			atomic_inc(&head->node.refs);
			spin_unlock(&delayed_refs->lock);
3964

3965
			mutex_lock(&head->mutex);
3966
			mutex_unlock(&head->mutex);
3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978
			btrfs_put_delayed_ref(&head->node);
			spin_lock(&delayed_refs->lock);
			continue;
		}
		spin_lock(&head->lock);
		while ((node = rb_first(&head->ref_root)) != NULL) {
			ref = rb_entry(node, struct btrfs_delayed_ref_node,
				       rb_node);
			ref->in_tree = 0;
			rb_erase(&ref->rb_node, &head->ref_root);
			atomic_dec(&delayed_refs->num_entries);
			btrfs_put_delayed_ref(ref);
3979
		}
3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991
		if (head->must_insert_reserved)
			pin_bytes = true;
		btrfs_free_delayed_extent_op(head->extent_op);
		delayed_refs->num_heads--;
		if (head->processing == 0)
			delayed_refs->num_heads_ready--;
		atomic_dec(&delayed_refs->num_entries);
		head->node.in_tree = 0;
		rb_erase(&head->href_node, &delayed_refs->href_root);
		spin_unlock(&head->lock);
		spin_unlock(&delayed_refs->lock);
		mutex_unlock(&head->mutex);
L
liubo 已提交
3992

3993 3994 3995 3996
		if (pin_bytes)
			btrfs_pin_extent(root, head->node.bytenr,
					 head->node.num_bytes, 1);
		btrfs_put_delayed_ref(&head->node);
L
liubo 已提交
3997 3998 3999 4000 4001 4002 4003 4004 4005
		cond_resched();
		spin_lock(&delayed_refs->lock);
	}

	spin_unlock(&delayed_refs->lock);

	return ret;
}

4006
static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root)
L
liubo 已提交
4007 4008 4009 4010 4011 4012
{
	struct btrfs_inode *btrfs_inode;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

4013 4014
	spin_lock(&root->delalloc_lock);
	list_splice_init(&root->delalloc_inodes, &splice);
L
liubo 已提交
4015 4016

	while (!list_empty(&splice)) {
4017 4018
		btrfs_inode = list_first_entry(&splice, struct btrfs_inode,
					       delalloc_inodes);
L
liubo 已提交
4019 4020

		list_del_init(&btrfs_inode->delalloc_inodes);
4021 4022
		clear_bit(BTRFS_INODE_IN_DELALLOC_LIST,
			  &btrfs_inode->runtime_flags);
4023
		spin_unlock(&root->delalloc_lock);
L
liubo 已提交
4024 4025

		btrfs_invalidate_inodes(btrfs_inode->root);
4026

4027
		spin_lock(&root->delalloc_lock);
L
liubo 已提交
4028 4029
	}

4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055
	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
4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068
}

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,
4069
					    mark, NULL);
L
liubo 已提交
4070 4071 4072 4073 4074
		if (ret)
			break;

		clear_extent_bits(dirty_pages, start, end, mark, GFP_NOFS);
		while (start <= end) {
4075
			eb = btrfs_find_tree_block(root, start);
4076
			start += root->nodesize;
4077
			if (!eb)
L
liubo 已提交
4078
				continue;
4079
			wait_on_extent_buffer_writeback(eb);
L
liubo 已提交
4080

4081 4082 4083 4084
			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
4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097
		}
	}

	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;
4098
	bool loop = true;
L
liubo 已提交
4099 4100

	unpin = pinned_extents;
4101
again:
L
liubo 已提交
4102 4103
	while (1) {
		ret = find_first_extent_bit(unpin, 0, &start, &end,
4104
					    EXTENT_DIRTY, NULL);
L
liubo 已提交
4105 4106 4107 4108
		if (ret)
			break;

		/* opt_discard */
4109 4110 4111 4112
		if (btrfs_test_opt(root, DISCARD))
			ret = btrfs_error_discard_extent(root, start,
							 end + 1 - start,
							 NULL);
L
liubo 已提交
4113 4114 4115 4116 4117 4118

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

4119 4120 4121 4122 4123 4124 4125 4126 4127
	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
4128 4129 4130
	return 0;
}

4131 4132 4133 4134 4135
void btrfs_cleanup_one_transaction(struct btrfs_transaction *cur_trans,
				   struct btrfs_root *root)
{
	btrfs_destroy_delayed_refs(cur_trans, root);

4136
	cur_trans->state = TRANS_STATE_COMMIT_START;
4137
	wake_up(&root->fs_info->transaction_blocked_wait);
4138

4139
	cur_trans->state = TRANS_STATE_UNBLOCKED;
4140
	wake_up(&root->fs_info->transaction_wait);
4141

4142 4143
	btrfs_destroy_delayed_inodes(root);
	btrfs_assert_delayed_root_empty(root);
4144 4145 4146

	btrfs_destroy_marked_extents(root, &cur_trans->dirty_pages,
				     EXTENT_DIRTY);
4147 4148
	btrfs_destroy_pinned_extent(root,
				    root->fs_info->pinned_extents);
4149

4150 4151 4152
	cur_trans->state =TRANS_STATE_COMPLETED;
	wake_up(&cur_trans->commit_wait);

4153 4154 4155 4156 4157 4158
	/*
	memset(cur_trans, 0, sizeof(*cur_trans));
	kmem_cache_free(btrfs_transaction_cachep, cur_trans);
	*/
}

4159
static int btrfs_cleanup_transaction(struct btrfs_root *root)
L
liubo 已提交
4160 4161 4162 4163 4164
{
	struct btrfs_transaction *t;

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

J
Josef Bacik 已提交
4165
	spin_lock(&root->fs_info->trans_lock);
4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184 4185 4186 4187 4188 4189
	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);
4190

4191 4192 4193
		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
4194
		list_del_init(&t->list);
4195
		spin_unlock(&root->fs_info->trans_lock);
L
liubo 已提交
4196

4197 4198 4199 4200 4201 4202 4203 4204 4205 4206
		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
4207 4208 4209 4210 4211
	mutex_unlock(&root->fs_info->transaction_kthread_mutex);

	return 0;
}

4212
static struct extent_io_ops btree_extent_io_ops = {
4213
	.readpage_end_io_hook = btree_readpage_end_io_hook,
A
Arne Jansen 已提交
4214
	.readpage_io_failed_hook = btree_io_failed_hook,
4215
	.submit_bio_hook = btree_submit_bio_hook,
4216 4217
	/* note we're sharing with inode.c for the merge bio hook */
	.merge_bio_hook = btrfs_merge_bio_hook,
4218
};