disk-io.c 111.5 KB
Newer Older
C
Chris Mason 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
/*
 * Copyright (C) 2007 Oracle.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License v2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */

C
Chris Mason 已提交
19
#include <linux/fs.h>
20
#include <linux/blkdev.h>
C
Chris Mason 已提交
21
#include <linux/scatterlist.h>
C
Chris Mason 已提交
22
#include <linux/swap.h>
23
#include <linux/radix-tree.h>
C
Chris Mason 已提交
24
#include <linux/writeback.h>
C
Chris Mason 已提交
25
#include <linux/buffer_head.h>
26
#include <linux/workqueue.h>
27
#include <linux/kthread.h>
C
Chris Mason 已提交
28
#include <linux/freezer.h>
29
#include <linux/crc32c.h>
30
#include <linux/slab.h>
31
#include <linux/migrate.h>
32
#include <linux/ratelimit.h>
33
#include <linux/uuid.h>
S
Stefan Behrens 已提交
34
#include <linux/semaphore.h>
35
#include <asm/unaligned.h>
C
Chris Mason 已提交
36
#include "compat.h"
37 38
#include "ctree.h"
#include "disk-io.h"
39
#include "transaction.h"
40
#include "btrfs_inode.h"
41
#include "volumes.h"
42
#include "print-tree.h"
43
#include "async-thread.h"
44
#include "locking.h"
45
#include "tree-log.h"
46
#include "free-space-cache.h"
47
#include "inode-map.h"
48
#include "check-integrity.h"
49
#include "rcu-string.h"
50
#include "dev-replace.h"
D
David Woodhouse 已提交
51
#include "raid56.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 63
static void btrfs_destroy_ordered_operations(struct btrfs_transaction *t,
					     struct btrfs_root *root);
64
static void btrfs_destroy_ordered_extents(struct btrfs_root *root);
L
liubo 已提交
65 66
static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
				      struct btrfs_root *root);
67
static void btrfs_evict_pending_snapshots(struct btrfs_transaction *t);
68
static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root);
L
liubo 已提交
69 70 71 72 73
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);
74 75
static int btrfs_cleanup_transaction(struct btrfs_root *root);
static void btrfs_error_commit_super(struct btrfs_root *root);
76

C
Chris Mason 已提交
77 78 79 80 81
/*
 * end_io_wq structs are used to do processing in task context when an IO is
 * complete.  This is used during reads to verify checksums, and it is used
 * by writes to insert metadata for new file extents after IO is complete.
 */
82 83 84 85 86 87
struct end_io_wq {
	struct bio *bio;
	bio_end_io_t *end_io;
	void *private;
	struct btrfs_fs_info *info;
	int error;
88
	int metadata;
89
	struct list_head list;
90
	struct btrfs_work work;
91
};
92

C
Chris Mason 已提交
93 94 95 96 97
/*
 * 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.
 */
98 99 100 101
struct async_submit_bio {
	struct inode *inode;
	struct bio *bio;
	struct list_head list;
C
Chris Mason 已提交
102 103
	extent_submit_bio_hook_t *submit_bio_start;
	extent_submit_bio_hook_t *submit_bio_done;
104 105
	int rw;
	int mirror_num;
C
Chris Mason 已提交
106
	unsigned long bio_flags;
107 108 109 110 111
	/*
	 * 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;
112
	struct btrfs_work work;
113
	int error;
114 115
};

116 117 118 119 120 121 122 123 124 125
/*
 * 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.
126
 *
127 128 129
 * 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.
130
 *
131 132 133
 * 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.
134
 *
135 136 137
 * 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.
138 139 140 141 142
 */
#ifdef CONFIG_DEBUG_LOCK_ALLOC
# if BTRFS_MAX_LEVEL != 8
#  error
# endif
143 144 145 146 147 148 149 150 151 152 153 154 155

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"	},
156
	{ .id = BTRFS_QUOTA_TREE_OBJECTID,	.name_stem = "quota"	},
157 158 159 160
	{ .id = BTRFS_TREE_LOG_OBJECTID,	.name_stem = "log"	},
	{ .id = BTRFS_TREE_RELOC_OBJECTID,	.name_stem = "treloc"	},
	{ .id = BTRFS_DATA_RELOC_TREE_OBJECTID,	.name_stem = "dreloc"	},
	{ .id = 0,				.name_stem = "tree"	},
161
};
162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192

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]);
}

193 194
#endif

C
Chris Mason 已提交
195 196 197 198
/*
 * extents on the btree inode are pretty simple, there's one extent
 * that covers the entire device
 */
199
static struct extent_map *btree_get_extent(struct inode *inode,
200
		struct page *page, size_t pg_offset, u64 start, u64 len,
201
		int create)
202
{
203 204 205 206
	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
	struct extent_map *em;
	int ret;

207
	read_lock(&em_tree->lock);
208
	em = lookup_extent_mapping(em_tree, start, len);
209 210 211
	if (em) {
		em->bdev =
			BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
212
		read_unlock(&em_tree->lock);
213
		goto out;
214
	}
215
	read_unlock(&em_tree->lock);
216

217
	em = alloc_extent_map();
218 219 220 221 222
	if (!em) {
		em = ERR_PTR(-ENOMEM);
		goto out;
	}
	em->start = 0;
223
	em->len = (u64)-1;
C
Chris Mason 已提交
224
	em->block_len = (u64)-1;
225
	em->block_start = 0;
226
	em->bdev = BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
227

228
	write_lock(&em_tree->lock);
J
Josef Bacik 已提交
229
	ret = add_extent_mapping(em_tree, em, 0);
230 231
	if (ret == -EEXIST) {
		free_extent_map(em);
232
		em = lookup_extent_mapping(em_tree, start, len);
233
		if (!em)
234
			em = ERR_PTR(-EIO);
235
	} else if (ret) {
236
		free_extent_map(em);
237
		em = ERR_PTR(ret);
238
	}
239
	write_unlock(&em_tree->lock);
240

241 242
out:
	return em;
243 244
}

245
u32 btrfs_csum_data(char *data, u32 seed, size_t len)
246
{
247
	return crc32c(seed, data, len);
248 249 250 251
}

void btrfs_csum_final(u32 crc, char *result)
{
252
	put_unaligned_le32(~crc, result);
253 254
}

C
Chris Mason 已提交
255 256 257 258
/*
 * compute the csum for a btree block, and either verify it or write it
 * into the csum field of the block.
 */
259 260 261
static int csum_tree_block(struct btrfs_root *root, struct extent_buffer *buf,
			   int verify)
{
262
	u16 csum_size = btrfs_super_csum_size(root->fs_info->super_copy);
263
	char *result = NULL;
264 265 266 267 268 269 270 271
	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;
272
	unsigned long inline_result;
273 274

	len = buf->len - offset;
C
Chris Mason 已提交
275
	while (len > 0) {
276
		err = map_private_extent_buffer(buf, offset, 32,
277
					&kaddr, &map_start, &map_len);
C
Chris Mason 已提交
278
		if (err)
279 280
			return 1;
		cur_len = min(len, map_len - (offset - map_start));
281
		crc = btrfs_csum_data(kaddr + offset - map_start,
282 283 284 285
				      crc, cur_len);
		len -= cur_len;
		offset += cur_len;
	}
286 287 288 289 290 291 292 293
	if (csum_size > sizeof(inline_result)) {
		result = kzalloc(csum_size * sizeof(char), GFP_NOFS);
		if (!result)
			return 1;
	} else {
		result = (char *)&inline_result;
	}

294 295 296
	btrfs_csum_final(crc, result);

	if (verify) {
297
		if (memcmp_extent_buffer(buf, result, 0, csum_size)) {
298 299
			u32 val;
			u32 found = 0;
300
			memcpy(&found, result, csum_size);
301

302
			read_extent_buffer(buf, &val, 0, csum_size);
303
			printk_ratelimited(KERN_INFO "btrfs: %s checksum verify "
C
Chris Mason 已提交
304 305 306 307 308
				       "failed on %llu wanted %X found %X "
				       "level %d\n",
				       root->fs_info->sb->s_id,
				       (unsigned long long)buf->start, val, found,
				       btrfs_header_level(buf));
309 310
			if (result != (char *)&inline_result)
				kfree(result);
311 312 313
			return 1;
		}
	} else {
314
		write_extent_buffer(buf, result, 0, csum_size);
315
	}
316 317
	if (result != (char *)&inline_result)
		kfree(result);
318 319 320
	return 0;
}

C
Chris Mason 已提交
321 322 323 324 325 326
/*
 * 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.
 */
327
static int verify_parent_transid(struct extent_io_tree *io_tree,
328 329
				 struct extent_buffer *eb, u64 parent_transid,
				 int atomic)
330
{
331
	struct extent_state *cached_state = NULL;
332 333 334 335 336
	int ret;

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

337 338 339
	if (atomic)
		return -EAGAIN;

340
	lock_extent_bits(io_tree, eb->start, eb->start + eb->len - 1,
341
			 0, &cached_state);
342
	if (extent_buffer_uptodate(eb) &&
343 344 345 346
	    btrfs_header_generation(eb) == parent_transid) {
		ret = 0;
		goto out;
	}
347
	printk_ratelimited("parent transid verify failed on %llu wanted %llu "
C
Chris Mason 已提交
348 349 350 351
		       "found %llu\n",
		       (unsigned long long)eb->start,
		       (unsigned long long)parent_transid,
		       (unsigned long long)btrfs_header_generation(eb));
352
	ret = 1;
353
	clear_extent_buffer_uptodate(eb);
C
Chris Mason 已提交
354
out:
355 356
	unlock_extent_cached(io_tree, eb->start, eb->start + eb->len - 1,
			     &cached_state, GFP_NOFS);
357 358 359
	return ret;
}

D
David Sterba 已提交
360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386
/*
 * 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;
387 388 389 390 391

		if (ret && btrfs_super_generation(disk_sb) < 10) {
			printk(KERN_WARNING "btrfs: super block crcs don't match, older mkfs detected\n");
			ret = 0;
		}
D
David Sterba 已提交
392 393 394 395 396 397 398 399 400 401 402
	}

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

	return ret;
}

C
Chris Mason 已提交
403 404 405 406
/*
 * helper to read a given tree block, doing retries as required when
 * the checksums don't match and we have alternate mirrors to try.
 */
407 408
static int btree_read_extent_buffer_pages(struct btrfs_root *root,
					  struct extent_buffer *eb,
409
					  u64 start, u64 parent_transid)
410 411
{
	struct extent_io_tree *io_tree;
412
	int failed = 0;
413 414 415
	int ret;
	int num_copies = 0;
	int mirror_num = 0;
416
	int failed_mirror = 0;
417

418
	clear_bit(EXTENT_BUFFER_CORRUPT, &eb->bflags);
419 420
	io_tree = &BTRFS_I(root->fs_info->btree_inode)->io_tree;
	while (1) {
421 422
		ret = read_extent_buffer_pages(io_tree, eb, start,
					       WAIT_COMPLETE,
423
					       btree_get_extent, mirror_num);
424 425
		if (!ret) {
			if (!verify_parent_transid(io_tree, eb,
426
						   parent_transid, 0))
427 428 429 430
				break;
			else
				ret = -EIO;
		}
C
Chris Mason 已提交
431

432 433 434 435 436 437
		/*
		 * 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))
438 439
			break;

440
		num_copies = btrfs_num_copies(root->fs_info,
441
					      eb->start, eb->len);
C
Chris Mason 已提交
442
		if (num_copies == 1)
443
			break;
C
Chris Mason 已提交
444

445 446 447 448 449
		if (!failed_mirror) {
			failed = 1;
			failed_mirror = eb->read_mirror;
		}

450
		mirror_num++;
451 452 453
		if (mirror_num == failed_mirror)
			mirror_num++;

C
Chris Mason 已提交
454
		if (mirror_num > num_copies)
455
			break;
456
	}
457

458
	if (failed && !ret && failed_mirror)
459 460 461
		repair_eb_io_failure(root, eb, failed_mirror);

	return ret;
462
}
463

C
Chris Mason 已提交
464
/*
C
Chris Mason 已提交
465 466
 * 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
467
 */
C
Chris Mason 已提交
468

469
static int csum_dirty_buffer(struct btrfs_root *root, struct page *page)
470
{
471
	struct extent_io_tree *tree;
M
Miao Xie 已提交
472
	u64 start = page_offset(page);
473 474
	u64 found_start;
	struct extent_buffer *eb;
475

476
	tree = &BTRFS_I(page->mapping->host)->io_tree;
477

J
Josef Bacik 已提交
478 479 480
	eb = (struct extent_buffer *)page->private;
	if (page != eb->pages[0])
		return 0;
481 482
	found_start = btrfs_header_bytenr(eb);
	if (found_start != start) {
483
		WARN_ON(1);
J
Josef Bacik 已提交
484
		return 0;
485 486 487
	}
	if (!PageUptodate(page)) {
		WARN_ON(1);
J
Josef Bacik 已提交
488
		return 0;
489 490 491 492 493
	}
	csum_tree_block(root, eb, 0);
	return 0;
}

Y
Yan Zheng 已提交
494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512
static int check_tree_block_fsid(struct btrfs_root *root,
				 struct extent_buffer *eb)
{
	struct btrfs_fs_devices *fs_devices = root->fs_info->fs_devices;
	u8 fsid[BTRFS_UUID_SIZE];
	int ret = 1;

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

513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579
#define CORRUPT(reason, eb, root, slot)				\
	printk(KERN_CRIT "btrfs: corrupt leaf, %s: block=%llu,"	\
	       "root=%llu, slot=%d\n", reason,			\
	       (unsigned long long)btrfs_header_bytenr(eb),	\
	       (unsigned long long)root->objectid, slot)

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

	if (nritems == 0)
		return 0;

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

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

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

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

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

	return 0;
}

580 581 582
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)
583 584 585 586 587 588
{
	struct extent_io_tree *tree;
	u64 found_start;
	int found_level;
	struct extent_buffer *eb;
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;
589
	int ret = 0;
590
	int reads_done;
591 592 593

	if (!page->private)
		goto out;
C
Chris Mason 已提交
594

595
	tree = &BTRFS_I(page->mapping->host)->io_tree;
J
Josef Bacik 已提交
596
	eb = (struct extent_buffer *)page->private;
C
Chris Mason 已提交
597

598 599 600 601 602 603
	/* 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);
604 605
	if (!reads_done)
		goto err;
606

607
	eb->read_mirror = mirror;
608 609 610 611 612
	if (test_bit(EXTENT_BUFFER_IOERR, &eb->bflags)) {
		ret = -EIO;
		goto err;
	}

613
	found_start = btrfs_header_bytenr(eb);
614
	if (found_start != eb->start) {
615
		printk_ratelimited(KERN_INFO "btrfs bad tree block start "
C
Chris Mason 已提交
616 617 618
			       "%llu %llu\n",
			       (unsigned long long)found_start,
			       (unsigned long long)eb->start);
619
		ret = -EIO;
620 621
		goto err;
	}
Y
Yan Zheng 已提交
622
	if (check_tree_block_fsid(root, eb)) {
623
		printk_ratelimited(KERN_INFO "btrfs bad fsid on block %llu\n",
C
Chris Mason 已提交
624
			       (unsigned long long)eb->start);
625 626 627
		ret = -EIO;
		goto err;
	}
628
	found_level = btrfs_header_level(eb);
629 630 631 632 633 634
	if (found_level >= BTRFS_MAX_LEVEL) {
		btrfs_info(root->fs_info, "bad tree block level %d\n",
			   (int)btrfs_header_level(eb));
		ret = -EIO;
		goto err;
	}
635

636 637
	btrfs_set_buffer_lockdep_class(btrfs_header_owner(eb),
				       eb, found_level);
638

639
	ret = csum_tree_block(root, eb, 1);
640
	if (ret) {
641
		ret = -EIO;
642 643 644 645 646 647 648 649 650 651 652 653
		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;
	}
654

655 656
	if (!ret)
		set_extent_buffer_uptodate(eb);
657
err:
658 659
	if (reads_done &&
	    test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
A
Arne Jansen 已提交
660 661
		btree_readahead_hook(root, eb, eb->start, ret);

D
David Woodhouse 已提交
662 663 664 665 666 667 668
	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);
669
		clear_extent_buffer_uptodate(eb);
D
David Woodhouse 已提交
670
	}
671
	free_extent_buffer(eb);
672
out:
673
	return ret;
674 675
}

676
static int btree_io_failed_hook(struct page *page, int failed_mirror)
A
Arne Jansen 已提交
677 678 679 680
{
	struct extent_buffer *eb;
	struct btrfs_root *root = BTRFS_I(page->mapping->host)->root;

J
Josef Bacik 已提交
681
	eb = (struct extent_buffer *)page->private;
682
	set_bit(EXTENT_BUFFER_IOERR, &eb->bflags);
683
	eb->read_mirror = failed_mirror;
D
David Woodhouse 已提交
684
	atomic_dec(&eb->io_pages);
685
	if (test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
A
Arne Jansen 已提交
686 687 688 689
		btree_readahead_hook(root, eb, eb->start, -EIO);
	return -EIO;	/* we fixed nothing */
}

690 691 692 693 694 695 696
static void end_workqueue_bio(struct bio *bio, int err)
{
	struct end_io_wq *end_io_wq = bio->bi_private;
	struct btrfs_fs_info *fs_info;

	fs_info = end_io_wq->info;
	end_io_wq->error = err;
697 698
	end_io_wq->work.func = end_workqueue_fn;
	end_io_wq->work.flags = 0;
699

700
	if (bio->bi_rw & REQ_WRITE) {
D
David Woodhouse 已提交
701
		if (end_io_wq->metadata == BTRFS_WQ_ENDIO_METADATA)
702 703
			btrfs_queue_worker(&fs_info->endio_meta_write_workers,
					   &end_io_wq->work);
D
David Woodhouse 已提交
704
		else if (end_io_wq->metadata == BTRFS_WQ_ENDIO_FREE_SPACE)
J
Josef Bacik 已提交
705 706
			btrfs_queue_worker(&fs_info->endio_freespace_worker,
					   &end_io_wq->work);
D
David Woodhouse 已提交
707 708 709
		else if (end_io_wq->metadata == BTRFS_WQ_ENDIO_RAID56)
			btrfs_queue_worker(&fs_info->endio_raid56_workers,
					   &end_io_wq->work);
710 711 712
		else
			btrfs_queue_worker(&fs_info->endio_write_workers,
					   &end_io_wq->work);
713
	} else {
D
David Woodhouse 已提交
714 715 716 717
		if (end_io_wq->metadata == BTRFS_WQ_ENDIO_RAID56)
			btrfs_queue_worker(&fs_info->endio_raid56_workers,
					   &end_io_wq->work);
		else if (end_io_wq->metadata)
718 719 720 721 722 723
			btrfs_queue_worker(&fs_info->endio_meta_workers,
					   &end_io_wq->work);
		else
			btrfs_queue_worker(&fs_info->endio_workers,
					   &end_io_wq->work);
	}
724 725
}

J
Josef Bacik 已提交
726 727 728 729 730 731
/*
 * For the metadata arg you want
 *
 * 0 - if data
 * 1 - if normal metadta
 * 2 - if writing to the free space cache area
D
David Woodhouse 已提交
732
 * 3 - raid parity work
J
Josef Bacik 已提交
733
 */
734 735
int btrfs_bio_wq_end_io(struct btrfs_fs_info *info, struct bio *bio,
			int metadata)
736
{
737 738 739 740 741 742 743
	struct end_io_wq *end_io_wq;
	end_io_wq = kmalloc(sizeof(*end_io_wq), GFP_NOFS);
	if (!end_io_wq)
		return -ENOMEM;

	end_io_wq->private = bio->bi_private;
	end_io_wq->end_io = bio->bi_end_io;
744
	end_io_wq->info = info;
745 746
	end_io_wq->error = 0;
	end_io_wq->bio = bio;
747
	end_io_wq->metadata = metadata;
748 749 750

	bio->bi_private = end_io_wq;
	bio->bi_end_io = end_workqueue_bio;
751 752 753
	return 0;
}

754
unsigned long btrfs_async_submit_limit(struct btrfs_fs_info *info)
755
{
756 757 758 759 760
	unsigned long limit = min_t(unsigned long,
				    info->workers.max_workers,
				    info->fs_devices->open_devices);
	return 256 * limit;
}
761

C
Chris Mason 已提交
762 763 764
static void run_one_async_start(struct btrfs_work *work)
{
	struct async_submit_bio *async;
765
	int ret;
C
Chris Mason 已提交
766 767

	async = container_of(work, struct  async_submit_bio, work);
768 769 770 771 772
	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
773 774 775
}

static void run_one_async_done(struct btrfs_work *work)
776 777 778
{
	struct btrfs_fs_info *fs_info;
	struct async_submit_bio *async;
779
	int limit;
780 781 782

	async = container_of(work, struct  async_submit_bio, work);
	fs_info = BTRFS_I(async->inode)->root->fs_info;
783

784
	limit = btrfs_async_submit_limit(fs_info);
785 786
	limit = limit * 2 / 3;

787
	if (atomic_dec_return(&fs_info->nr_async_submits) < limit &&
788
	    waitqueue_active(&fs_info->async_submit_wait))
789 790
		wake_up(&fs_info->async_submit_wait);

791 792 793 794 795 796
	/* 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
797
	async->submit_bio_done(async->inode, async->rw, async->bio,
798 799
			       async->mirror_num, async->bio_flags,
			       async->bio_offset);
C
Chris Mason 已提交
800 801 802 803 804 805 806
}

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

	async = container_of(work, struct  async_submit_bio, work);
807 808 809
	kfree(async);
}

810 811
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
812
			unsigned long bio_flags,
813
			u64 bio_offset,
C
Chris Mason 已提交
814 815
			extent_submit_bio_hook_t *submit_bio_start,
			extent_submit_bio_hook_t *submit_bio_done)
816 817 818 819 820 821 822 823 824 825 826
{
	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
827 828 829 830 831 832 833
	async->submit_bio_start = submit_bio_start;
	async->submit_bio_done = submit_bio_done;

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

834
	async->work.flags = 0;
C
Chris Mason 已提交
835
	async->bio_flags = bio_flags;
836
	async->bio_offset = bio_offset;
837

838 839
	async->error = 0;

840
	atomic_inc(&fs_info->nr_async_submits);
841

842
	if (rw & REQ_SYNC)
843 844
		btrfs_set_work_high_prio(&async->work);

845
	btrfs_queue_worker(&fs_info->workers, &async->work);
846

C
Chris Mason 已提交
847
	while (atomic_read(&fs_info->async_submit_draining) &&
848 849 850 851 852
	      atomic_read(&fs_info->nr_async_submits)) {
		wait_event(fs_info->async_submit_wait,
			   (atomic_read(&fs_info->nr_async_submits) == 0));
	}

853 854 855
	return 0;
}

856 857 858 859 860
static int btree_csum_one_bio(struct bio *bio)
{
	struct bio_vec *bvec = bio->bi_io_vec;
	int bio_index = 0;
	struct btrfs_root *root;
861
	int ret = 0;
862 863

	WARN_ON(bio->bi_vcnt <= 0);
C
Chris Mason 已提交
864
	while (bio_index < bio->bi_vcnt) {
865
		root = BTRFS_I(bvec->bv_page->mapping->host)->root;
866 867 868
		ret = csum_dirty_buffer(root, bvec->bv_page);
		if (ret)
			break;
869 870 871
		bio_index++;
		bvec++;
	}
872
	return ret;
873 874
}

C
Chris Mason 已提交
875 876
static int __btree_submit_bio_start(struct inode *inode, int rw,
				    struct bio *bio, int mirror_num,
877 878
				    unsigned long bio_flags,
				    u64 bio_offset)
879
{
880 881
	/*
	 * when we're called for a write, we're already in the async
882
	 * submission context.  Just jump into btrfs_map_bio
883
	 */
884
	return btree_csum_one_bio(bio);
C
Chris Mason 已提交
885
}
886

C
Chris Mason 已提交
887
static int __btree_submit_bio_done(struct inode *inode, int rw, struct bio *bio,
888 889
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
C
Chris Mason 已提交
890
{
891 892
	int ret;

893
	/*
C
Chris Mason 已提交
894 895
	 * when we're called for a write, we're already in the async
	 * submission context.  Just jump into btrfs_map_bio
896
	 */
897 898 899 900
	ret = btrfs_map_bio(BTRFS_I(inode)->root, rw, bio, mirror_num, 1);
	if (ret)
		bio_endio(bio, ret);
	return ret;
901 902
}

903 904 905 906 907 908 909 910 911 912 913
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;
}

914
static int btree_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
915 916
				 int mirror_num, unsigned long bio_flags,
				 u64 bio_offset)
917
{
918
	int async = check_async_write(inode, bio_flags);
919 920
	int ret;

921
	if (!(rw & REQ_WRITE)) {
C
Chris Mason 已提交
922 923 924 925
		/*
		 * called for a read, do the setup so that checksum validation
		 * can happen in the async kernel threads
		 */
926 927
		ret = btrfs_bio_wq_end_io(BTRFS_I(inode)->root->fs_info,
					  bio, 1);
928
		if (ret)
929 930 931
			goto out_w_error;
		ret = btrfs_map_bio(BTRFS_I(inode)->root, rw, bio,
				    mirror_num, 0);
932 933 934
	} else if (!async) {
		ret = btree_csum_one_bio(bio);
		if (ret)
935 936 937 938 939 940 941 942 943 944 945 946 947
			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);
948
	}
949

950 951 952 953 954
	if (ret) {
out_w_error:
		bio_endio(bio, ret);
	}
	return ret;
955 956
}

J
Jan Beulich 已提交
957
#ifdef CONFIG_MIGRATION
958
static int btree_migratepage(struct address_space *mapping,
959 960
			struct page *newpage, struct page *page,
			enum migrate_mode mode)
961 962 963 964 965 966 967 968 969 970 971 972 973 974
{
	/*
	 * 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;
975
	return migrate_page(mapping, newpage, page, mode);
976
}
J
Jan Beulich 已提交
977
#endif
978

979 980 981 982

static int btree_writepages(struct address_space *mapping,
			    struct writeback_control *wbc)
{
983
	struct extent_io_tree *tree;
984 985 986
	struct btrfs_fs_info *fs_info;
	int ret;

987
	tree = &BTRFS_I(mapping->host)->io_tree;
988
	if (wbc->sync_mode == WB_SYNC_NONE) {
989 990 991 992

		if (wbc->for_kupdate)
			return 0;

993
		fs_info = BTRFS_I(mapping->host)->root->fs_info;
994
		/* this is a bit racy, but that's ok */
995 996 997
		ret = percpu_counter_compare(&fs_info->dirty_metadata_bytes,
					     BTRFS_DIRTY_METADATA_THRESH);
		if (ret < 0)
998 999
			return 0;
	}
1000
	return btree_write_cache_pages(mapping, wbc);
1001 1002
}

1003
static int btree_readpage(struct file *file, struct page *page)
1004
{
1005 1006
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
1007
	return extent_read_full_page(tree, page, btree_get_extent, 0);
1008
}
C
Chris Mason 已提交
1009

1010
static int btree_releasepage(struct page *page, gfp_t gfp_flags)
1011
{
1012
	if (PageWriteback(page) || PageDirty(page))
C
Chris Mason 已提交
1013
		return 0;
1014

1015
	return try_release_extent_buffer(page);
1016 1017
}

1018 1019
static void btree_invalidatepage(struct page *page, unsigned int offset,
				 unsigned int length)
1020
{
1021 1022
	struct extent_io_tree *tree;
	tree = &BTRFS_I(page->mapping->host)->io_tree;
1023 1024
	extent_invalidatepage(tree, page, offset);
	btree_releasepage(page, GFP_NOFS);
1025
	if (PagePrivate(page)) {
C
Chris Mason 已提交
1026 1027
		printk(KERN_WARNING "btrfs warning page private not zero "
		       "on page %llu\n", (unsigned long long)page_offset(page));
1028 1029 1030 1031
		ClearPagePrivate(page);
		set_page_private(page, 0);
		page_cache_release(page);
	}
1032 1033
}

1034 1035
static int btree_set_page_dirty(struct page *page)
{
1036
#ifdef DEBUG
1037 1038 1039 1040 1041 1042 1043 1044
	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);
1045
#endif
1046 1047 1048
	return __set_page_dirty_nobuffers(page);
}

1049
static const struct address_space_operations btree_aops = {
1050
	.readpage	= btree_readpage,
1051
	.writepages	= btree_writepages,
1052 1053
	.releasepage	= btree_releasepage,
	.invalidatepage = btree_invalidatepage,
1054
#ifdef CONFIG_MIGRATION
1055
	.migratepage	= btree_migratepage,
1056
#endif
1057
	.set_page_dirty = btree_set_page_dirty,
1058 1059
};

1060 1061
int readahead_tree_block(struct btrfs_root *root, u64 bytenr, u32 blocksize,
			 u64 parent_transid)
C
Chris Mason 已提交
1062
{
1063 1064
	struct extent_buffer *buf = NULL;
	struct inode *btree_inode = root->fs_info->btree_inode;
1065
	int ret = 0;
C
Chris Mason 已提交
1066

1067
	buf = btrfs_find_create_tree_block(root, bytenr, blocksize);
1068
	if (!buf)
C
Chris Mason 已提交
1069
		return 0;
1070
	read_extent_buffer_pages(&BTRFS_I(btree_inode)->io_tree,
1071
				 buf, 0, WAIT_NONE, btree_get_extent, 0);
1072
	free_extent_buffer(buf);
1073
	return ret;
C
Chris Mason 已提交
1074 1075
}

1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099
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;
1100
	} else if (extent_buffer_uptodate(buf)) {
1101 1102 1103 1104 1105 1106 1107
		*eb = buf;
	} else {
		free_extent_buffer(buf);
	}
	return 0;
}

1108 1109 1110 1111 1112 1113
struct extent_buffer *btrfs_find_tree_block(struct btrfs_root *root,
					    u64 bytenr, u32 blocksize)
{
	struct inode *btree_inode = root->fs_info->btree_inode;
	struct extent_buffer *eb;
	eb = find_extent_buffer(&BTRFS_I(btree_inode)->io_tree,
1114
				bytenr, blocksize);
1115 1116 1117 1118 1119 1120 1121 1122 1123 1124
	return eb;
}

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

	eb = alloc_extent_buffer(&BTRFS_I(btree_inode)->io_tree,
1125
				 bytenr, blocksize);
1126 1127 1128 1129
	return eb;
}


1130 1131
int btrfs_write_tree_block(struct extent_buffer *buf)
{
1132
	return filemap_fdatawrite_range(buf->pages[0]->mapping, buf->start,
1133
					buf->start + buf->len - 1);
1134 1135 1136 1137
}

int btrfs_wait_tree_block_writeback(struct extent_buffer *buf)
{
1138
	return filemap_fdatawait_range(buf->pages[0]->mapping,
1139
				       buf->start, buf->start + buf->len - 1);
1140 1141
}

1142
struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr,
1143
				      u32 blocksize, u64 parent_transid)
1144 1145 1146 1147 1148 1149 1150 1151
{
	struct extent_buffer *buf = NULL;
	int ret;

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

1152
	ret = btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
1153 1154 1155 1156
	if (ret) {
		free_extent_buffer(buf);
		return NULL;
	}
1157
	return buf;
1158

1159 1160
}

1161 1162
void clean_tree_block(struct btrfs_trans_handle *trans, struct btrfs_root *root,
		      struct extent_buffer *buf)
1163
{
1164 1165
	struct btrfs_fs_info *fs_info = root->fs_info;

1166
	if (btrfs_header_generation(buf) ==
1167
	    fs_info->running_transaction->transid) {
1168
		btrfs_assert_tree_locked(buf);
1169

1170
		if (test_and_clear_bit(EXTENT_BUFFER_DIRTY, &buf->bflags)) {
1171 1172 1173
			__percpu_counter_add(&fs_info->dirty_metadata_bytes,
					     -buf->len,
					     fs_info->dirty_metadata_batch);
1174 1175 1176 1177
			/* ugh, clear_extent_buffer_dirty needs to lock the page */
			btrfs_set_lock_blocking(buf);
			clear_extent_buffer_dirty(buf);
		}
1178
	}
1179 1180
}

1181 1182 1183 1184
static void __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
			 u32 stripesize, struct btrfs_root *root,
			 struct btrfs_fs_info *fs_info,
			 u64 objectid)
1185
{
C
Chris Mason 已提交
1186
	root->node = NULL;
1187
	root->commit_root = NULL;
1188 1189 1190
	root->sectorsize = sectorsize;
	root->nodesize = nodesize;
	root->leafsize = leafsize;
1191
	root->stripesize = stripesize;
C
Chris Mason 已提交
1192
	root->ref_cows = 0;
1193
	root->track_dirty = 0;
1194
	root->in_radix = 0;
1195 1196
	root->orphan_item_inserted = 0;
	root->orphan_cleanup_state = 0;
1197

1198 1199
	root->objectid = objectid;
	root->last_trans = 0;
1200
	root->highest_objectid = 0;
1201
	root->nr_delalloc_inodes = 0;
1202
	root->nr_ordered_extents = 0;
1203
	root->name = NULL;
1204
	root->inode_tree = RB_ROOT;
1205
	INIT_RADIX_TREE(&root->delayed_nodes_tree, GFP_ATOMIC);
1206
	root->block_rsv = NULL;
1207
	root->orphan_block_rsv = NULL;
1208 1209

	INIT_LIST_HEAD(&root->dirty_list);
1210
	INIT_LIST_HEAD(&root->root_list);
1211 1212
	INIT_LIST_HEAD(&root->delalloc_inodes);
	INIT_LIST_HEAD(&root->delalloc_root);
1213 1214
	INIT_LIST_HEAD(&root->ordered_extents);
	INIT_LIST_HEAD(&root->ordered_root);
1215 1216
	INIT_LIST_HEAD(&root->logged_list[0]);
	INIT_LIST_HEAD(&root->logged_list[1]);
1217
	spin_lock_init(&root->orphan_lock);
1218
	spin_lock_init(&root->inode_lock);
1219
	spin_lock_init(&root->delalloc_lock);
1220
	spin_lock_init(&root->ordered_extent_lock);
1221
	spin_lock_init(&root->accounting_lock);
1222 1223
	spin_lock_init(&root->log_extents_lock[0]);
	spin_lock_init(&root->log_extents_lock[1]);
1224
	mutex_init(&root->objectid_mutex);
1225
	mutex_init(&root->log_mutex);
Y
Yan Zheng 已提交
1226 1227 1228 1229 1230 1231
	init_waitqueue_head(&root->log_writer_wait);
	init_waitqueue_head(&root->log_commit_wait[0]);
	init_waitqueue_head(&root->log_commit_wait[1]);
	atomic_set(&root->log_commit[0], 0);
	atomic_set(&root->log_commit[1], 0);
	atomic_set(&root->log_writers, 0);
M
Miao Xie 已提交
1232
	atomic_set(&root->log_batch, 0);
1233
	atomic_set(&root->orphan_inodes, 0);
1234
	atomic_set(&root->refs, 1);
Y
Yan Zheng 已提交
1235
	root->log_transid = 0;
1236
	root->last_log_commit = 0;
1237
	extent_io_tree_init(&root->dirty_log_pages,
1238
			     fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
1239

1240 1241
	memset(&root->root_key, 0, sizeof(root->root_key));
	memset(&root->root_item, 0, sizeof(root->root_item));
1242
	memset(&root->defrag_progress, 0, sizeof(root->defrag_progress));
1243
	memset(&root->root_kobj, 0, sizeof(root->root_kobj));
1244
	root->defrag_trans_start = fs_info->generation;
1245
	init_completion(&root->kobj_unregister);
1246
	root->defrag_running = 0;
1247
	root->root_key.objectid = objectid;
1248
	root->anon_dev = 0;
1249

1250
	spin_lock_init(&root->root_item_lock);
1251 1252
}

1253
static struct btrfs_root *btrfs_alloc_root(struct btrfs_fs_info *fs_info)
A
Al Viro 已提交
1254 1255 1256 1257 1258 1259 1260
{
	struct btrfs_root *root = kzalloc(sizeof(*root), GFP_NOFS);
	if (root)
		root->fs_info = fs_info;
	return root;
}

1261 1262 1263 1264 1265 1266 1267 1268 1269 1270
struct btrfs_root *btrfs_create_tree(struct btrfs_trans_handle *trans,
				     struct btrfs_fs_info *fs_info,
				     u64 objectid)
{
	struct extent_buffer *leaf;
	struct btrfs_root *tree_root = fs_info->tree_root;
	struct btrfs_root *root;
	struct btrfs_key key;
	int ret = 0;
	u64 bytenr;
1271
	uuid_le uuid;
1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287

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

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

	leaf = btrfs_alloc_free_block(trans, root, root->leafsize,
				      0, objectid, NULL, 0, 0, 0);
	if (IS_ERR(leaf)) {
		ret = PTR_ERR(leaf);
1288
		leaf = NULL;
1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320
		goto fail;
	}

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

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

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


	root->root_item.flags = 0;
	root->root_item.byte_limit = 0;
	btrfs_set_root_bytenr(&root->root_item, leaf->start);
	btrfs_set_root_generation(&root->root_item, trans->transid);
	btrfs_set_root_level(&root->root_item, 0);
	btrfs_set_root_refs(&root->root_item, 1);
	btrfs_set_root_used(&root->root_item, leaf->len);
	btrfs_set_root_last_snapshot(&root->root_item, 0);
	btrfs_set_root_dirid(&root->root_item, 0);
1321 1322
	uuid_le_gen(&uuid);
	memcpy(root->root_item.uuid, uuid.b, BTRFS_UUID_SIZE);
1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333
	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);

1334 1335
	return root;

1336
fail:
1337 1338 1339 1340 1341
	if (leaf) {
		btrfs_tree_unlock(leaf);
		free_extent_buffer(leaf);
	}
	kfree(root);
1342

1343
	return ERR_PTR(ret);
1344 1345
}

Y
Yan Zheng 已提交
1346 1347
static struct btrfs_root *alloc_log_tree(struct btrfs_trans_handle *trans,
					 struct btrfs_fs_info *fs_info)
1348 1349 1350
{
	struct btrfs_root *root;
	struct btrfs_root *tree_root = fs_info->tree_root;
Y
Yan Zheng 已提交
1351
	struct extent_buffer *leaf;
1352

A
Al Viro 已提交
1353
	root = btrfs_alloc_root(fs_info);
1354
	if (!root)
Y
Yan Zheng 已提交
1355
		return ERR_PTR(-ENOMEM);
1356 1357 1358 1359 1360 1361 1362 1363

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

	root->root_key.objectid = BTRFS_TREE_LOG_OBJECTID;
	root->root_key.type = BTRFS_ROOT_ITEM_KEY;
	root->root_key.offset = BTRFS_TREE_LOG_OBJECTID;
Y
Yan Zheng 已提交
1364 1365 1366 1367 1368 1369
	/*
	 * 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).
	 */
1370 1371
	root->ref_cows = 0;

1372
	leaf = btrfs_alloc_free_block(trans, root, root->leafsize, 0,
A
Arne Jansen 已提交
1373
				      BTRFS_TREE_LOG_OBJECTID, NULL,
1374
				      0, 0, 0);
Y
Yan Zheng 已提交
1375 1376 1377 1378
	if (IS_ERR(leaf)) {
		kfree(root);
		return ERR_CAST(leaf);
	}
1379

1380 1381 1382 1383 1384
	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
1385
	root->node = leaf;
1386 1387 1388 1389 1390 1391

	write_extent_buffer(root->node, root->fs_info->fsid,
			    (unsigned long)btrfs_header_fsid(root->node),
			    BTRFS_FSID_SIZE);
	btrfs_mark_buffer_dirty(root->node);
	btrfs_tree_unlock(root->node);
Y
Yan Zheng 已提交
1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421
	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;
1422 1423 1424 1425 1426
	btrfs_set_stack_inode_generation(inode_item, 1);
	btrfs_set_stack_inode_size(inode_item, 3);
	btrfs_set_stack_inode_nlink(inode_item, 1);
	btrfs_set_stack_inode_nbytes(inode_item, root->leafsize);
	btrfs_set_stack_inode_mode(inode_item, S_IFDIR | 0755);
Y
Yan Zheng 已提交
1427

1428
	btrfs_set_root_node(&log_root->root_item, log_root->node);
Y
Yan Zheng 已提交
1429 1430 1431 1432

	WARN_ON(root->log_root);
	root->log_root = log_root;
	root->log_transid = 0;
1433
	root->last_log_commit = 0;
1434 1435 1436
	return 0;
}

1437 1438
static struct btrfs_root *btrfs_read_tree_root(struct btrfs_root *tree_root,
					       struct btrfs_key *key)
1439 1440 1441
{
	struct btrfs_root *root;
	struct btrfs_fs_info *fs_info = tree_root->fs_info;
1442
	struct btrfs_path *path;
1443
	u64 generation;
1444
	u32 blocksize;
1445
	int ret;
1446

1447 1448
	path = btrfs_alloc_path();
	if (!path)
1449
		return ERR_PTR(-ENOMEM);
1450 1451 1452 1453 1454

	root = btrfs_alloc_root(fs_info);
	if (!root) {
		ret = -ENOMEM;
		goto alloc_fail;
1455 1456
	}

1457
	__setup_root(tree_root->nodesize, tree_root->leafsize,
1458
		     tree_root->sectorsize, tree_root->stripesize,
1459
		     root, fs_info, key->objectid);
1460

1461 1462
	ret = btrfs_find_root(tree_root, key, path,
			      &root->root_item, &root->root_key);
1463
	if (ret) {
1464 1465
		if (ret > 0)
			ret = -ENOENT;
1466
		goto find_fail;
1467
	}
1468

1469
	generation = btrfs_root_generation(&root->root_item);
1470 1471
	blocksize = btrfs_level_size(root, btrfs_root_level(&root->root_item));
	root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
1472
				     blocksize, generation);
1473 1474 1475 1476 1477 1478
	if (!root->node) {
		ret = -ENOMEM;
		goto find_fail;
	} else if (!btrfs_buffer_uptodate(root->node, generation, 0)) {
		ret = -EIO;
		goto read_fail;
1479
	}
1480
	root->commit_root = btrfs_root_node(root);
1481
out:
1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503
	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) {
1504
		root->ref_cows = 1;
1505 1506
		btrfs_check_and_init_root_item(&root->root_item);
	}
1507

1508 1509 1510
	return root;
}

1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537
int btrfs_init_fs_root(struct btrfs_root *root)
{
	int ret;

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

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

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

1538 1539
static struct btrfs_root *btrfs_lookup_fs_root(struct btrfs_fs_info *fs_info,
					       u64 root_id)
1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570
{
	struct btrfs_root *root;

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

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

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

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

	return ret;
}

1571 1572
struct btrfs_root *btrfs_read_fs_root_no_name(struct btrfs_fs_info *fs_info,
					      struct btrfs_key *location)
1573 1574 1575 1576
{
	struct btrfs_root *root;
	int ret;

1577 1578 1579 1580
	if (location->objectid == BTRFS_ROOT_TREE_OBJECTID)
		return fs_info->tree_root;
	if (location->objectid == BTRFS_EXTENT_TREE_OBJECTID)
		return fs_info->extent_root;
1581 1582 1583 1584
	if (location->objectid == BTRFS_CHUNK_TREE_OBJECTID)
		return fs_info->chunk_root;
	if (location->objectid == BTRFS_DEV_TREE_OBJECTID)
		return fs_info->dev_root;
1585 1586
	if (location->objectid == BTRFS_CSUM_TREE_OBJECTID)
		return fs_info->csum_root;
1587 1588 1589
	if (location->objectid == BTRFS_QUOTA_TREE_OBJECTID)
		return fs_info->quota_root ? fs_info->quota_root :
					     ERR_PTR(-ENOENT);
1590 1591 1592
	if (location->objectid == BTRFS_UUID_TREE_OBJECTID)
		return fs_info->uuid_root ? fs_info->uuid_root :
					    ERR_PTR(-ENOENT);
1593
again:
1594
	root = btrfs_lookup_fs_root(fs_info, location->objectid);
1595 1596 1597
	if (root)
		return root;

1598
	root = btrfs_read_fs_root(fs_info->tree_root, location);
1599 1600
	if (IS_ERR(root))
		return root;
1601

1602 1603
	if (btrfs_root_refs(&root->root_item) == 0) {
		ret = -ENOENT;
1604
		goto fail;
1605
	}
1606

1607
	ret = btrfs_init_fs_root(root);
1608 1609
	if (ret)
		goto fail;
1610

1611 1612 1613 1614 1615 1616
	ret = btrfs_find_orphan_item(fs_info->tree_root, location->objectid);
	if (ret < 0)
		goto fail;
	if (ret == 0)
		root->orphan_item_inserted = 1;

1617
	ret = btrfs_insert_fs_root(fs_info, root);
1618
	if (ret) {
1619 1620 1621 1622 1623
		if (ret == -EEXIST) {
			free_fs_root(root);
			goto again;
		}
		goto fail;
1624
	}
1625
	return root;
1626 1627 1628
fail:
	free_fs_root(root);
	return ERR_PTR(ret);
1629 1630
}

C
Chris Mason 已提交
1631 1632 1633 1634 1635 1636
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
1637

1638 1639
	rcu_read_lock();
	list_for_each_entry_rcu(device, &info->fs_devices->devices, dev_list) {
1640 1641
		if (!device->bdev)
			continue;
C
Chris Mason 已提交
1642 1643 1644 1645 1646 1647
		bdi = blk_get_backing_dev_info(device->bdev);
		if (bdi && bdi_congested(bdi, bdi_bits)) {
			ret = 1;
			break;
		}
	}
1648
	rcu_read_unlock();
C
Chris Mason 已提交
1649 1650 1651
	return ret;
}

1652 1653 1654 1655
/*
 * If this fails, caller must call bdi_destroy() to get rid of the
 * bdi again.
 */
C
Chris Mason 已提交
1656 1657
static int setup_bdi(struct btrfs_fs_info *info, struct backing_dev_info *bdi)
{
1658 1659 1660
	int err;

	bdi->capabilities = BDI_CAP_MAP_COPY;
1661
	err = bdi_setup_and_register(bdi, "btrfs", BDI_CAP_MAP_COPY);
1662 1663 1664
	if (err)
		return err;

1665
	bdi->ra_pages	= default_backing_dev_info.ra_pages;
C
Chris Mason 已提交
1666 1667 1668 1669 1670
	bdi->congested_fn	= btrfs_congested_fn;
	bdi->congested_data	= info;
	return 0;
}

1671 1672 1673 1674 1675
/*
 * 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)
1676 1677
{
	struct bio *bio;
1678 1679
	struct end_io_wq *end_io_wq;
	struct btrfs_fs_info *fs_info;
1680 1681
	int error;

1682 1683 1684
	end_io_wq = container_of(work, struct end_io_wq, work);
	bio = end_io_wq->bio;
	fs_info = end_io_wq->info;
1685

1686 1687 1688 1689 1690
	error = end_io_wq->error;
	bio->bi_private = end_io_wq->private;
	bio->bi_end_io = end_io_wq->end_io;
	kfree(end_io_wq);
	bio_endio(bio, error);
1691 1692
}

1693 1694 1695
static int cleaner_kthread(void *arg)
{
	struct btrfs_root *root = arg;
1696
	int again;
1697 1698

	do {
1699
		again = 0;
1700

1701
		/* Make the cleaner go to sleep early. */
1702
		if (btrfs_need_cleaner_sleep(root))
1703 1704 1705 1706 1707
			goto sleep;

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

1708 1709 1710 1711
		/*
		 * Avoid the problem that we change the status of the fs
		 * during the above check and trylock.
		 */
1712
		if (btrfs_need_cleaner_sleep(root)) {
1713 1714
			mutex_unlock(&root->fs_info->cleaner_mutex);
			goto sleep;
1715
		}
1716

1717 1718 1719 1720 1721
		btrfs_run_delayed_iputs(root);
		again = btrfs_clean_one_deleted_snapshot(root);
		mutex_unlock(&root->fs_info->cleaner_mutex);

		/*
1722 1723
		 * The defragger has dealt with the R/O remount and umount,
		 * needn't do anything special here.
1724 1725 1726
		 */
		btrfs_run_defrag_inodes(root->fs_info);
sleep:
D
David Sterba 已提交
1727
		if (!try_to_freeze() && !again) {
1728
			set_current_state(TASK_INTERRUPTIBLE);
1729 1730
			if (!kthread_should_stop())
				schedule();
1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741
			__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;
1742
	u64 transid;
1743 1744
	unsigned long now;
	unsigned long delay;
1745
	bool cannot_commit;
1746 1747

	do {
1748
		cannot_commit = false;
1749
		delay = HZ * root->fs_info->commit_interval;
1750 1751
		mutex_lock(&root->fs_info->transaction_kthread_mutex);

J
Josef Bacik 已提交
1752
		spin_lock(&root->fs_info->trans_lock);
1753 1754
		cur = root->fs_info->running_transaction;
		if (!cur) {
J
Josef Bacik 已提交
1755
			spin_unlock(&root->fs_info->trans_lock);
1756 1757
			goto sleep;
		}
Y
Yan Zheng 已提交
1758

1759
		now = get_seconds();
1760
		if (cur->state < TRANS_STATE_BLOCKED &&
1761 1762
		    (now < cur->start_time ||
		     now - cur->start_time < root->fs_info->commit_interval)) {
J
Josef Bacik 已提交
1763
			spin_unlock(&root->fs_info->trans_lock);
1764 1765 1766
			delay = HZ * 5;
			goto sleep;
		}
1767
		transid = cur->transid;
J
Josef Bacik 已提交
1768
		spin_unlock(&root->fs_info->trans_lock);
1769

1770
		/* If the file system is aborted, this will always fail. */
1771
		trans = btrfs_attach_transaction(root);
1772
		if (IS_ERR(trans)) {
1773 1774
			if (PTR_ERR(trans) != -ENOENT)
				cannot_commit = true;
1775
			goto sleep;
1776
		}
1777
		if (transid == trans->transid) {
1778
			btrfs_commit_transaction(trans, root);
1779 1780 1781
		} else {
			btrfs_end_transaction(trans, root);
		}
1782 1783 1784 1785
sleep:
		wake_up_process(root->fs_info->cleaner_kthread);
		mutex_unlock(&root->fs_info->transaction_kthread_mutex);

1786
		if (!try_to_freeze()) {
1787
			set_current_state(TASK_INTERRUPTIBLE);
1788
			if (!kthread_should_stop() &&
1789 1790
			    (!btrfs_transaction_blocked(root->fs_info) ||
			     cannot_commit))
1791
				schedule_timeout(delay);
1792 1793 1794 1795 1796 1797
			__set_current_state(TASK_RUNNING);
		}
	} while (!kthread_should_stop());
	return 0;
}

C
Chris Mason 已提交
1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 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
/*
 * 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));

1904 1905 1906 1907 1908 1909 1910 1911
	/*
	 * 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
1912
			       btrfs_header_generation(info->fs_root->node));
1913
		btrfs_set_backup_fs_root_level(root_backup,
C
Chris Mason 已提交
1914
			       btrfs_header_level(info->fs_root->node));
1915
	}
C
Chris Mason 已提交
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 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996

	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
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
/* helper to cleanup workers */
static void btrfs_stop_all_workers(struct btrfs_fs_info *fs_info)
{
	btrfs_stop_workers(&fs_info->generic_worker);
	btrfs_stop_workers(&fs_info->fixup_workers);
	btrfs_stop_workers(&fs_info->delalloc_workers);
	btrfs_stop_workers(&fs_info->workers);
	btrfs_stop_workers(&fs_info->endio_workers);
	btrfs_stop_workers(&fs_info->endio_meta_workers);
	btrfs_stop_workers(&fs_info->endio_raid56_workers);
	btrfs_stop_workers(&fs_info->rmw_workers);
	btrfs_stop_workers(&fs_info->endio_meta_write_workers);
	btrfs_stop_workers(&fs_info->endio_write_workers);
	btrfs_stop_workers(&fs_info->endio_freespace_worker);
	btrfs_stop_workers(&fs_info->submit_workers);
	btrfs_stop_workers(&fs_info->delayed_workers);
	btrfs_stop_workers(&fs_info->caching_workers);
	btrfs_stop_workers(&fs_info->readahead_workers);
	btrfs_stop_workers(&fs_info->flush_workers);
J
Jan Schmidt 已提交
2016
	btrfs_stop_workers(&fs_info->qgroup_rescan_workers);
L
Liu Bo 已提交
2017 2018
}

C
Chris Mason 已提交
2019 2020 2021 2022 2023 2024 2025
/* helper to cleanup tree roots */
static void free_root_pointers(struct btrfs_fs_info *info, int chunk_root)
{
	free_extent_buffer(info->tree_root->node);
	free_extent_buffer(info->tree_root->commit_root);
	info->tree_root->node = NULL;
	info->tree_root->commit_root = NULL;
2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044

	if (info->dev_root) {
		free_extent_buffer(info->dev_root->node);
		free_extent_buffer(info->dev_root->commit_root);
		info->dev_root->node = NULL;
		info->dev_root->commit_root = NULL;
	}
	if (info->extent_root) {
		free_extent_buffer(info->extent_root->node);
		free_extent_buffer(info->extent_root->commit_root);
		info->extent_root->node = NULL;
		info->extent_root->commit_root = NULL;
	}
	if (info->csum_root) {
		free_extent_buffer(info->csum_root->node);
		free_extent_buffer(info->csum_root->commit_root);
		info->csum_root->node = NULL;
		info->csum_root->commit_root = NULL;
	}
2045
	if (info->quota_root) {
2046 2047
		free_extent_buffer(info->quota_root->node);
		free_extent_buffer(info->quota_root->commit_root);
2048 2049 2050
		info->quota_root->node = NULL;
		info->quota_root->commit_root = NULL;
	}
2051 2052 2053 2054 2055 2056
	if (info->uuid_root) {
		free_extent_buffer(info->uuid_root->node);
		free_extent_buffer(info->uuid_root->commit_root);
		info->uuid_root->node = NULL;
		info->uuid_root->commit_root = NULL;
	}
C
Chris Mason 已提交
2057 2058 2059 2060 2061 2062 2063 2064
	if (chunk_root) {
		free_extent_buffer(info->chunk_root->node);
		free_extent_buffer(info->chunk_root->commit_root);
		info->chunk_root->node = NULL;
		info->chunk_root->commit_root = NULL;
	}
}

2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076
static void del_fs_roots(struct btrfs_fs_info *fs_info)
{
	int ret;
	struct btrfs_root *gang[8];
	int i;

	while (!list_empty(&fs_info->dead_roots)) {
		gang[0] = list_entry(fs_info->dead_roots.next,
				     struct btrfs_root, root_list);
		list_del(&gang[0]->root_list);

		if (gang[0]->in_radix) {
2077
			btrfs_drop_and_free_fs_root(fs_info, gang[0]);
2078 2079 2080
		} else {
			free_extent_buffer(gang[0]->node);
			free_extent_buffer(gang[0]->commit_root);
2081
			btrfs_put_fs_root(gang[0]);
2082 2083 2084 2085 2086 2087 2088 2089 2090 2091
		}
	}

	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++)
2092
			btrfs_drop_and_free_fs_root(fs_info, gang[i]);
2093 2094
	}
}
C
Chris Mason 已提交
2095

A
Al Viro 已提交
2096 2097 2098
int open_ctree(struct super_block *sb,
	       struct btrfs_fs_devices *fs_devices,
	       char *options)
2099
{
2100 2101 2102 2103
	u32 sectorsize;
	u32 nodesize;
	u32 leafsize;
	u32 blocksize;
2104
	u32 stripesize;
2105
	u64 generation;
2106
	u64 features;
2107
	struct btrfs_key location;
2108
	struct buffer_head *bh;
2109
	struct btrfs_super_block *disk_super;
2110
	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
2111
	struct btrfs_root *tree_root;
2112 2113 2114 2115
	struct btrfs_root *extent_root;
	struct btrfs_root *csum_root;
	struct btrfs_root *chunk_root;
	struct btrfs_root *dev_root;
2116
	struct btrfs_root *quota_root;
2117
	struct btrfs_root *uuid_root;
2118
	struct btrfs_root *log_tree_root;
2119
	int ret;
2120
	int err = -EINVAL;
C
Chris Mason 已提交
2121 2122
	int num_backups_tried = 0;
	int backup_index = 0;
2123 2124
	bool create_uuid_tree;
	bool check_uuid_tree;
2125

2126
	tree_root = fs_info->tree_root = btrfs_alloc_root(fs_info);
A
Al Viro 已提交
2127
	chunk_root = fs_info->chunk_root = btrfs_alloc_root(fs_info);
2128
	if (!tree_root || !chunk_root) {
C
Chris Mason 已提交
2129 2130 2131
		err = -ENOMEM;
		goto fail;
	}
2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144

	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;
	}

2145 2146 2147 2148 2149 2150 2151 2152
	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));

2153 2154 2155 2156 2157 2158
	ret = percpu_counter_init(&fs_info->delalloc_bytes, 0);
	if (ret) {
		err = ret;
		goto fail_dirty_metadata_bytes;
	}

2159 2160 2161
	fs_info->btree_inode = new_inode(sb);
	if (!fs_info->btree_inode) {
		err = -ENOMEM;
2162
		goto fail_delalloc_bytes;
2163 2164
	}

2165
	mapping_set_gfp_mask(fs_info->btree_inode->i_mapping, GFP_NOFS);
2166

2167
	INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC);
C
Chris Mason 已提交
2168
	INIT_LIST_HEAD(&fs_info->trans_list);
2169
	INIT_LIST_HEAD(&fs_info->dead_roots);
Y
Yan, Zheng 已提交
2170
	INIT_LIST_HEAD(&fs_info->delayed_iputs);
2171
	INIT_LIST_HEAD(&fs_info->delalloc_roots);
2172
	INIT_LIST_HEAD(&fs_info->caching_block_groups);
2173
	spin_lock_init(&fs_info->delalloc_root_lock);
J
Josef Bacik 已提交
2174
	spin_lock_init(&fs_info->trans_lock);
2175
	spin_lock_init(&fs_info->fs_roots_radix_lock);
Y
Yan, Zheng 已提交
2176
	spin_lock_init(&fs_info->delayed_iput_lock);
C
Chris Mason 已提交
2177
	spin_lock_init(&fs_info->defrag_inodes_lock);
2178
	spin_lock_init(&fs_info->free_chunk_lock);
J
Jan Schmidt 已提交
2179
	spin_lock_init(&fs_info->tree_mod_seq_lock);
2180
	spin_lock_init(&fs_info->super_lock);
J
Jan Schmidt 已提交
2181
	rwlock_init(&fs_info->tree_mod_log_lock);
C
Chris Mason 已提交
2182
	mutex_init(&fs_info->reloc_mutex);
2183
	seqlock_init(&fs_info->profiles_lock);
2184

2185
	init_completion(&fs_info->kobj_unregister);
2186
	INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots);
2187
	INIT_LIST_HEAD(&fs_info->space_info);
J
Jan Schmidt 已提交
2188
	INIT_LIST_HEAD(&fs_info->tree_mod_seq_list);
2189
	btrfs_mapping_init(&fs_info->mapping_tree);
2190 2191 2192 2193 2194 2195 2196 2197 2198
	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);
2199
	atomic_set(&fs_info->nr_async_submits, 0);
2200
	atomic_set(&fs_info->async_delalloc_pages, 0);
2201
	atomic_set(&fs_info->async_submit_draining, 0);
2202
	atomic_set(&fs_info->nr_async_bios, 0);
C
Chris Mason 已提交
2203
	atomic_set(&fs_info->defrag_running, 0);
2204
	atomic64_set(&fs_info->tree_mod_seq, 0);
C
Chris Mason 已提交
2205
	fs_info->sb = sb;
2206
	fs_info->max_inline = 8192 * 1024;
J
Josef Bacik 已提交
2207
	fs_info->metadata_ratio = 0;
C
Chris Mason 已提交
2208
	fs_info->defrag_inodes = RB_ROOT;
2209
	fs_info->free_chunk_space = 0;
J
Jan Schmidt 已提交
2210
	fs_info->tree_mod_log = RB_ROOT;
2211
	fs_info->commit_interval = BTRFS_DEFAULT_COMMIT_INTERVAL;
C
Chris Mason 已提交
2212

2213 2214 2215
	/* 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
2216

2217 2218
	fs_info->thread_pool_size = min_t(unsigned long,
					  num_online_cpus() + 2, 8);
2219

2220 2221
	INIT_LIST_HEAD(&fs_info->ordered_roots);
	spin_lock_init(&fs_info->ordered_root_lock);
2222 2223 2224 2225 2226 2227 2228
	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);
2229

A
Arne Jansen 已提交
2230 2231 2232 2233 2234 2235 2236 2237
	mutex_init(&fs_info->scrub_lock);
	atomic_set(&fs_info->scrubs_running, 0);
	atomic_set(&fs_info->scrub_pause_req, 0);
	atomic_set(&fs_info->scrubs_paused, 0);
	atomic_set(&fs_info->scrub_cancel_req, 0);
	init_waitqueue_head(&fs_info->scrub_pause_wait);
	init_rwsem(&fs_info->scrub_super_lock);
	fs_info->scrub_workers_refcnt = 0;
2238 2239 2240
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	fs_info->check_integrity_print_mask = 0;
#endif
A
Arne Jansen 已提交
2241

2242 2243
	spin_lock_init(&fs_info->balance_lock);
	mutex_init(&fs_info->balance_mutex);
2244 2245
	atomic_set(&fs_info->balance_running, 0);
	atomic_set(&fs_info->balance_pause_req, 0);
2246
	atomic_set(&fs_info->balance_cancel_req, 0);
2247
	fs_info->balance_ctl = NULL;
2248
	init_waitqueue_head(&fs_info->balance_wait_q);
A
Arne Jansen 已提交
2249

2250 2251
	sb->s_blocksize = 4096;
	sb->s_blocksize_bits = blksize_bits(4096);
J
Jens Axboe 已提交
2252
	sb->s_bdi = &fs_info->bdi;
2253

2254
	fs_info->btree_inode->i_ino = BTRFS_BTREE_INODE_OBJECTID;
M
Miklos Szeredi 已提交
2255
	set_nlink(fs_info->btree_inode, 1);
2256 2257 2258 2259 2260 2261
	/*
	 * 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;
2262
	fs_info->btree_inode->i_mapping->a_ops = &btree_aops;
C
Chris Mason 已提交
2263 2264
	fs_info->btree_inode->i_mapping->backing_dev_info = &fs_info->bdi;

2265
	RB_CLEAR_NODE(&BTRFS_I(fs_info->btree_inode)->rb_node);
2266
	extent_io_tree_init(&BTRFS_I(fs_info->btree_inode)->io_tree,
2267
			     fs_info->btree_inode->i_mapping);
2268
	BTRFS_I(fs_info->btree_inode)->io_tree.track_uptodate = 0;
2269
	extent_map_tree_init(&BTRFS_I(fs_info->btree_inode)->extent_tree);
2270 2271

	BTRFS_I(fs_info->btree_inode)->io_tree.ops = &btree_extent_io_ops;
2272

2273 2274 2275
	BTRFS_I(fs_info->btree_inode)->root = tree_root;
	memset(&BTRFS_I(fs_info->btree_inode)->location, 0,
	       sizeof(struct btrfs_key));
2276 2277
	set_bit(BTRFS_INODE_DUMMY,
		&BTRFS_I(fs_info->btree_inode)->runtime_flags);
2278
	insert_inode_hash(fs_info->btree_inode);
2279

J
Josef Bacik 已提交
2280
	spin_lock_init(&fs_info->block_group_cache_lock);
2281
	fs_info->block_group_cache_tree = RB_ROOT;
2282
	fs_info->first_logical_byte = (u64)-1;
J
Josef Bacik 已提交
2283

2284
	extent_io_tree_init(&fs_info->freed_extents[0],
2285
			     fs_info->btree_inode->i_mapping);
2286
	extent_io_tree_init(&fs_info->freed_extents[1],
2287
			     fs_info->btree_inode->i_mapping);
2288
	fs_info->pinned_extents = &fs_info->freed_extents[0];
2289
	fs_info->do_barriers = 1;
2290

C
Chris Mason 已提交
2291

2292
	mutex_init(&fs_info->ordered_operations_mutex);
2293
	mutex_init(&fs_info->ordered_extent_flush_mutex);
2294
	mutex_init(&fs_info->tree_log_mutex);
2295
	mutex_init(&fs_info->chunk_mutex);
2296 2297
	mutex_init(&fs_info->transaction_kthread_mutex);
	mutex_init(&fs_info->cleaner_mutex);
2298
	mutex_init(&fs_info->volume_mutex);
2299
	init_rwsem(&fs_info->extent_commit_sem);
2300
	init_rwsem(&fs_info->cleanup_work_sem);
2301
	init_rwsem(&fs_info->subvol_sem);
S
Stefan Behrens 已提交
2302
	sema_init(&fs_info->uuid_tree_rescan_sem, 1);
2303 2304 2305 2306 2307
	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);
2308

2309
	spin_lock_init(&fs_info->qgroup_lock);
2310
	mutex_init(&fs_info->qgroup_ioctl_lock);
2311 2312 2313 2314 2315
	fs_info->qgroup_tree = RB_ROOT;
	INIT_LIST_HEAD(&fs_info->dirty_qgroups);
	fs_info->qgroup_seq = 1;
	fs_info->quota_enabled = 0;
	fs_info->pending_quota_state = 0;
2316
	fs_info->qgroup_ulist = NULL;
J
Jan Schmidt 已提交
2317
	mutex_init(&fs_info->qgroup_rescan_lock);
2318

2319 2320 2321
	btrfs_init_free_cluster(&fs_info->meta_alloc_cluster);
	btrfs_init_free_cluster(&fs_info->data_alloc_cluster);

2322
	init_waitqueue_head(&fs_info->transaction_throttle);
2323
	init_waitqueue_head(&fs_info->transaction_wait);
S
Sage Weil 已提交
2324
	init_waitqueue_head(&fs_info->transaction_blocked_wait);
2325
	init_waitqueue_head(&fs_info->async_submit_wait);
2326

D
David Woodhouse 已提交
2327 2328
	ret = btrfs_alloc_stripe_hash_table(fs_info);
	if (ret) {
2329
		err = ret;
D
David Woodhouse 已提交
2330 2331 2332
		goto fail_alloc;
	}

2333
	__setup_root(4096, 4096, 4096, 4096, tree_root,
C
Chris Mason 已提交
2334
		     fs_info, BTRFS_ROOT_TREE_OBJECTID);
2335

2336
	invalidate_bdev(fs_devices->latest_bdev);
D
David Sterba 已提交
2337 2338 2339 2340

	/*
	 * Read super block and check the signature bytes only
	 */
Y
Yan Zheng 已提交
2341
	bh = btrfs_read_dev_super(fs_devices->latest_bdev);
2342 2343
	if (!bh) {
		err = -EINVAL;
2344
		goto fail_alloc;
2345
	}
C
Chris Mason 已提交
2346

D
David Sterba 已提交
2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361
	/*
	 * We want to check superblock checksum, the type is stored inside.
	 * Pass the whole disk block of size BTRFS_SUPER_INFO_SIZE (4k).
	 */
	if (btrfs_check_super_csum(bh->b_data)) {
		printk(KERN_ERR "btrfs: superblock checksum mismatch\n");
		err = -EINVAL;
		goto fail_alloc;
	}

	/*
	 * super_copy is zeroed at allocation time and we never touch the
	 * following bytes up to INFO_SIZE, the checksum is calculated from
	 * the whole block of INFO_SIZE
	 */
2362 2363 2364
	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));
2365
	brelse(bh);
2366

2367
	memcpy(fs_info->fsid, fs_info->super_copy->fsid, BTRFS_FSID_SIZE);
2368

D
David Sterba 已提交
2369 2370 2371 2372 2373 2374 2375
	ret = btrfs_check_super_valid(fs_info, sb->s_flags & MS_RDONLY);
	if (ret) {
		printk(KERN_ERR "btrfs: superblock contains fatal errors\n");
		err = -EINVAL;
		goto fail_alloc;
	}

2376
	disk_super = fs_info->super_copy;
2377
	if (!btrfs_super_root(disk_super))
2378
		goto fail_alloc;
2379

L
liubo 已提交
2380
	/* check FS state, whether FS is broken. */
2381 2382
	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
2383

C
Chris Mason 已提交
2384 2385 2386 2387 2388 2389 2390
	/*
	 * 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);

2391 2392 2393 2394 2395 2396
	/*
	 * 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
2397 2398 2399
	ret = btrfs_parse_options(tree_root, options);
	if (ret) {
		err = ret;
2400
		goto fail_alloc;
Y
Yan Zheng 已提交
2401
	}
2402

2403 2404 2405 2406 2407
	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",
2408
		       (unsigned long long)features);
2409
		err = -EINVAL;
2410
		goto fail_alloc;
2411 2412
	}

2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429
	if (btrfs_super_leafsize(disk_super) !=
	    btrfs_super_nodesize(disk_super)) {
		printk(KERN_ERR "BTRFS: couldn't mount because metadata "
		       "blocksizes don't match.  node %d leaf %d\n",
		       btrfs_super_nodesize(disk_super),
		       btrfs_super_leafsize(disk_super));
		err = -EINVAL;
		goto fail_alloc;
	}
	if (btrfs_super_leafsize(disk_super) > BTRFS_MAX_METADATA_BLOCKSIZE) {
		printk(KERN_ERR "BTRFS: couldn't mount because metadata "
		       "blocksize (%d) was too large\n",
		       btrfs_super_leafsize(disk_super));
		err = -EINVAL;
		goto fail_alloc;
	}

2430
	features = btrfs_super_incompat_flags(disk_super);
L
Li Zefan 已提交
2431
	features |= BTRFS_FEATURE_INCOMPAT_MIXED_BACKREF;
L
Li Zefan 已提交
2432
	if (tree_root->fs_info->compress_type == BTRFS_COMPRESS_LZO)
L
Li Zefan 已提交
2433
		features |= BTRFS_FEATURE_INCOMPAT_COMPRESS_LZO;
2434

2435 2436 2437
	if (features & BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA)
		printk(KERN_ERR "btrfs: has skinny extents\n");

2438 2439 2440 2441 2442 2443 2444 2445 2446 2447
	/*
	 * flag our filesystem as having big metadata blocks if
	 * they are bigger than the page size
	 */
	if (btrfs_super_leafsize(disk_super) > PAGE_CACHE_SIZE) {
		if (!(features & BTRFS_FEATURE_INCOMPAT_BIG_METADATA))
			printk(KERN_INFO "btrfs flagging fs with big metadata feature\n");
		features |= BTRFS_FEATURE_INCOMPAT_BIG_METADATA;
	}

2448 2449 2450 2451
	nodesize = btrfs_super_nodesize(disk_super);
	leafsize = btrfs_super_leafsize(disk_super);
	sectorsize = btrfs_super_sectorsize(disk_super);
	stripesize = btrfs_super_stripesize(disk_super);
2452
	fs_info->dirty_metadata_batch = leafsize * (1 + ilog2(nr_cpu_ids));
2453
	fs_info->delalloc_batch = sectorsize * 512 * (1 + ilog2(nr_cpu_ids));
2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466

	/*
	 * mixed block groups end up with duplicate but slightly offset
	 * extent buffers for the same range.  It leads to corruptions
	 */
	if ((features & BTRFS_FEATURE_INCOMPAT_MIXED_GROUPS) &&
	    (sectorsize != leafsize)) {
		printk(KERN_WARNING "btrfs: unequal leaf/node/sector sizes "
				"are not allowed for mixed block groups on %s\n",
				sb->s_id);
		goto fail_alloc;
	}

2467 2468 2469 2470
	/*
	 * Needn't use the lock because there is no other task which will
	 * update the flag.
	 */
L
Li Zefan 已提交
2471
	btrfs_set_super_incompat_flags(disk_super, features);
2472

2473 2474 2475 2476 2477
	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",
2478
		       (unsigned long long)features);
2479
		err = -EINVAL;
2480
		goto fail_alloc;
2481
	}
2482 2483 2484 2485

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

2486
	btrfs_init_workers(&fs_info->workers, "worker",
2487 2488
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
C
Chris Mason 已提交
2489

2490
	btrfs_init_workers(&fs_info->delalloc_workers, "delalloc",
2491 2492
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2493

2494 2495 2496 2497
	btrfs_init_workers(&fs_info->flush_workers, "flush_delalloc",
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);

2498
	btrfs_init_workers(&fs_info->submit_workers, "submit",
2499
			   min_t(u64, fs_devices->num_devices,
2500 2501
			   fs_info->thread_pool_size),
			   &fs_info->generic_worker);
2502

2503 2504 2505
	btrfs_init_workers(&fs_info->caching_workers, "cache",
			   2, &fs_info->generic_worker);

2506 2507 2508 2509 2510
	/* a higher idle thresh on the submit workers makes it much more
	 * likely that bios will be send down in a sane order to the
	 * devices
	 */
	fs_info->submit_workers.idle_thresh = 64;
2511

2512
	fs_info->workers.idle_thresh = 16;
C
Chris Mason 已提交
2513
	fs_info->workers.ordered = 1;
2514

2515 2516 2517
	fs_info->delalloc_workers.idle_thresh = 2;
	fs_info->delalloc_workers.ordered = 1;

2518 2519
	btrfs_init_workers(&fs_info->fixup_workers, "fixup", 1,
			   &fs_info->generic_worker);
2520
	btrfs_init_workers(&fs_info->endio_workers, "endio",
2521 2522
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2523
	btrfs_init_workers(&fs_info->endio_meta_workers, "endio-meta",
2524 2525
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2526
	btrfs_init_workers(&fs_info->endio_meta_write_workers,
2527 2528
			   "endio-meta-write", fs_info->thread_pool_size,
			   &fs_info->generic_worker);
D
David Woodhouse 已提交
2529 2530 2531 2532 2533 2534
	btrfs_init_workers(&fs_info->endio_raid56_workers,
			   "endio-raid56", fs_info->thread_pool_size,
			   &fs_info->generic_worker);
	btrfs_init_workers(&fs_info->rmw_workers,
			   "rmw", fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2535
	btrfs_init_workers(&fs_info->endio_write_workers, "endio-write",
2536 2537
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
J
Josef Bacik 已提交
2538 2539
	btrfs_init_workers(&fs_info->endio_freespace_worker, "freespace-write",
			   1, &fs_info->generic_worker);
2540 2541 2542
	btrfs_init_workers(&fs_info->delayed_workers, "delayed-meta",
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
2543 2544 2545
	btrfs_init_workers(&fs_info->readahead_workers, "readahead",
			   fs_info->thread_pool_size,
			   &fs_info->generic_worker);
J
Jan Schmidt 已提交
2546 2547
	btrfs_init_workers(&fs_info->qgroup_rescan_workers, "qgroup-rescan", 1,
			   &fs_info->generic_worker);
2548 2549 2550 2551 2552 2553

	/*
	 * endios are largely parallel and should have a very
	 * low idle thresh
	 */
	fs_info->endio_workers.idle_thresh = 4;
2554
	fs_info->endio_meta_workers.idle_thresh = 4;
D
David Woodhouse 已提交
2555 2556
	fs_info->endio_raid56_workers.idle_thresh = 4;
	fs_info->rmw_workers.idle_thresh = 2;
2557

2558 2559
	fs_info->endio_write_workers.idle_thresh = 2;
	fs_info->endio_meta_write_workers.idle_thresh = 2;
2560
	fs_info->readahead_workers.idle_thresh = 2;
2561

2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572
	/*
	 * btrfs_start_workers can really only fail because of ENOMEM so just
	 * return -ENOMEM if any of these fail.
	 */
	ret = btrfs_start_workers(&fs_info->workers);
	ret |= btrfs_start_workers(&fs_info->generic_worker);
	ret |= btrfs_start_workers(&fs_info->submit_workers);
	ret |= btrfs_start_workers(&fs_info->delalloc_workers);
	ret |= btrfs_start_workers(&fs_info->fixup_workers);
	ret |= btrfs_start_workers(&fs_info->endio_workers);
	ret |= btrfs_start_workers(&fs_info->endio_meta_workers);
D
David Woodhouse 已提交
2573 2574
	ret |= btrfs_start_workers(&fs_info->rmw_workers);
	ret |= btrfs_start_workers(&fs_info->endio_raid56_workers);
2575 2576 2577 2578 2579 2580
	ret |= btrfs_start_workers(&fs_info->endio_meta_write_workers);
	ret |= btrfs_start_workers(&fs_info->endio_write_workers);
	ret |= btrfs_start_workers(&fs_info->endio_freespace_worker);
	ret |= btrfs_start_workers(&fs_info->delayed_workers);
	ret |= btrfs_start_workers(&fs_info->caching_workers);
	ret |= btrfs_start_workers(&fs_info->readahead_workers);
2581
	ret |= btrfs_start_workers(&fs_info->flush_workers);
J
Jan Schmidt 已提交
2582
	ret |= btrfs_start_workers(&fs_info->qgroup_rescan_workers);
2583
	if (ret) {
2584
		err = -ENOMEM;
2585 2586
		goto fail_sb_buffer;
	}
2587

2588
	fs_info->bdi.ra_pages *= btrfs_super_num_devices(disk_super);
C
Chris Mason 已提交
2589 2590
	fs_info->bdi.ra_pages = max(fs_info->bdi.ra_pages,
				    4 * 1024 * 1024 / PAGE_CACHE_SIZE);
2591

2592 2593 2594
	tree_root->nodesize = nodesize;
	tree_root->leafsize = leafsize;
	tree_root->sectorsize = sectorsize;
2595
	tree_root->stripesize = stripesize;
2596 2597 2598

	sb->s_blocksize = sectorsize;
	sb->s_blocksize_bits = blksize_bits(sectorsize);
2599

2600
	if (btrfs_super_magic(disk_super) != BTRFS_MAGIC) {
C
Chris Mason 已提交
2601
		printk(KERN_INFO "btrfs: valid FS not found on %s\n", sb->s_id);
C
Chris Mason 已提交
2602 2603
		goto fail_sb_buffer;
	}
2604

2605 2606 2607
	if (sectorsize != PAGE_SIZE) {
		printk(KERN_WARNING "btrfs: Incompatible sector size(%lu) "
		       "found on %s\n", (unsigned long)sectorsize, sb->s_id);
2608 2609 2610
		goto fail_sb_buffer;
	}

2611
	mutex_lock(&fs_info->chunk_mutex);
Y
Yan Zheng 已提交
2612
	ret = btrfs_read_sys_array(tree_root);
2613
	mutex_unlock(&fs_info->chunk_mutex);
2614
	if (ret) {
C
Chris Mason 已提交
2615 2616
		printk(KERN_WARNING "btrfs: failed to read the system "
		       "array on %s\n", sb->s_id);
2617
		goto fail_sb_buffer;
2618
	}
2619 2620 2621

	blocksize = btrfs_level_size(tree_root,
				     btrfs_super_chunk_root_level(disk_super));
2622
	generation = btrfs_super_chunk_root_generation(disk_super);
2623 2624 2625 2626 2627 2628

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

	chunk_root->node = read_tree_block(chunk_root,
					   btrfs_super_chunk_root(disk_super),
2629
					   blocksize, generation);
2630 2631
	if (!chunk_root->node ||
	    !test_bit(EXTENT_BUFFER_UPTODATE, &chunk_root->node->bflags)) {
2632 2633
		printk(KERN_WARNING "btrfs: failed to read chunk root on %s\n",
		       sb->s_id);
C
Chris Mason 已提交
2634
		goto fail_tree_roots;
2635
	}
2636 2637
	btrfs_set_root_node(&chunk_root->root_item, chunk_root->node);
	chunk_root->commit_root = btrfs_root_node(chunk_root);
2638

2639
	read_extent_buffer(chunk_root->node, fs_info->chunk_tree_uuid,
C
Chris Mason 已提交
2640 2641
	   (unsigned long)btrfs_header_chunk_tree_uuid(chunk_root->node),
	   BTRFS_UUID_SIZE);
2642

2643
	ret = btrfs_read_chunk_tree(chunk_root);
Y
Yan Zheng 已提交
2644
	if (ret) {
C
Chris Mason 已提交
2645 2646
		printk(KERN_WARNING "btrfs: failed to read chunk tree on %s\n",
		       sb->s_id);
C
Chris Mason 已提交
2647
		goto fail_tree_roots;
Y
Yan Zheng 已提交
2648
	}
2649

2650 2651 2652 2653 2654
	/*
	 * 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);
2655

2656 2657 2658 2659 2660 2661
	if (!fs_devices->latest_bdev) {
		printk(KERN_CRIT "btrfs: failed to read devices on %s\n",
		       sb->s_id);
		goto fail_tree_roots;
	}

C
Chris Mason 已提交
2662
retry_root_backup:
2663 2664
	blocksize = btrfs_level_size(tree_root,
				     btrfs_super_root_level(disk_super));
2665
	generation = btrfs_super_generation(disk_super);
2666

C
Chris Mason 已提交
2667
	tree_root->node = read_tree_block(tree_root,
2668
					  btrfs_super_root(disk_super),
2669
					  blocksize, generation);
C
Chris Mason 已提交
2670 2671
	if (!tree_root->node ||
	    !test_bit(EXTENT_BUFFER_UPTODATE, &tree_root->node->bflags)) {
2672 2673
		printk(KERN_WARNING "btrfs: failed to read tree root on %s\n",
		       sb->s_id);
C
Chris Mason 已提交
2674 2675

		goto recovery_tree_root;
2676
	}
C
Chris Mason 已提交
2677

2678 2679
	btrfs_set_root_node(&tree_root->root_item, tree_root->node);
	tree_root->commit_root = btrfs_root_node(tree_root);
2680

2681 2682 2683 2684 2685 2686 2687
	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
2688
		goto recovery_tree_root;
2689
	}
2690
	extent_root->track_dirty = 1;
2691
	fs_info->extent_root = extent_root;
2692

2693 2694 2695 2696
	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
2697
		goto recovery_tree_root;
2698
	}
2699
	dev_root->track_dirty = 1;
2700 2701
	fs_info->dev_root = dev_root;
	btrfs_init_devices_late(fs_info);
2702

2703 2704 2705 2706
	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
2707
		goto recovery_tree_root;
2708
	}
2709
	csum_root->track_dirty = 1;
2710
	fs_info->csum_root = csum_root;
2711

2712 2713 2714
	location.objectid = BTRFS_QUOTA_TREE_OBJECTID;
	quota_root = btrfs_read_tree_root(tree_root, &location);
	if (!IS_ERR(quota_root)) {
2715 2716 2717
		quota_root->track_dirty = 1;
		fs_info->quota_enabled = 1;
		fs_info->pending_quota_state = 1;
2718
		fs_info->quota_root = quota_root;
2719 2720
	}

2721 2722 2723 2724 2725 2726 2727
	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;
2728
		check_uuid_tree = false;
2729 2730 2731
	} else {
		uuid_root->track_dirty = 1;
		fs_info->uuid_root = uuid_root;
2732 2733 2734
		create_uuid_tree = false;
		check_uuid_tree =
		    generation != btrfs_super_uuid_tree_generation(disk_super);
2735 2736
	}

2737 2738 2739
	fs_info->generation = generation;
	fs_info->last_trans_committed = generation;

2740 2741 2742 2743 2744 2745
	ret = btrfs_recover_balance(fs_info);
	if (ret) {
		printk(KERN_WARNING "btrfs: failed to recover balance\n");
		goto fail_block_groups;
	}

2746 2747 2748 2749 2750 2751 2752
	ret = btrfs_init_dev_stats(fs_info);
	if (ret) {
		printk(KERN_ERR "btrfs: failed to init dev_stats: %d\n",
		       ret);
		goto fail_block_groups;
	}

2753 2754 2755 2756 2757 2758 2759 2760
	ret = btrfs_init_dev_replace(fs_info);
	if (ret) {
		pr_err("btrfs: failed to init dev_replace: %d\n", ret);
		goto fail_block_groups;
	}

	btrfs_close_extra_devices(fs_info, fs_devices, 1);

2761 2762 2763 2764 2765 2766
	ret = btrfs_init_space_info(fs_info);
	if (ret) {
		printk(KERN_ERR "Failed to initial space info: %d\n", ret);
		goto fail_block_groups;
	}

2767 2768 2769 2770 2771
	ret = btrfs_read_block_groups(extent_root);
	if (ret) {
		printk(KERN_ERR "Failed to read block groups: %d\n", ret);
		goto fail_block_groups;
	}
2772 2773
	fs_info->num_tolerated_disk_barrier_failures =
		btrfs_calc_num_tolerated_disk_barrier_failures(fs_info);
2774 2775 2776 2777 2778 2779 2780
	if (fs_info->fs_devices->missing_devices >
	     fs_info->num_tolerated_disk_barrier_failures &&
	    !(sb->s_flags & MS_RDONLY)) {
		printk(KERN_WARNING
		       "Btrfs: too many missing devices, writeable mount is not allowed\n");
		goto fail_block_groups;
	}
C
Chris Mason 已提交
2781

2782 2783
	fs_info->cleaner_kthread = kthread_run(cleaner_kthread, tree_root,
					       "btrfs-cleaner");
2784
	if (IS_ERR(fs_info->cleaner_kthread))
2785
		goto fail_block_groups;
2786 2787 2788 2789

	fs_info->transaction_kthread = kthread_run(transaction_kthread,
						   tree_root,
						   "btrfs-transaction");
2790
	if (IS_ERR(fs_info->transaction_kthread))
2791
		goto fail_cleaner;
2792

C
Chris Mason 已提交
2793 2794 2795 2796 2797 2798 2799 2800
	if (!btrfs_test_opt(tree_root, SSD) &&
	    !btrfs_test_opt(tree_root, NOSSD) &&
	    !fs_info->fs_devices->rotating) {
		printk(KERN_INFO "Btrfs detected SSD devices, enabling SSD "
		       "mode\n");
		btrfs_set_opt(fs_info->mount_opt, SSD);
	}

2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	if (btrfs_test_opt(tree_root, CHECK_INTEGRITY)) {
		ret = btrfsic_mount(tree_root, fs_devices,
				    btrfs_test_opt(tree_root,
					CHECK_INTEGRITY_INCLUDING_EXTENT_DATA) ?
				    1 : 0,
				    fs_info->check_integrity_print_mask);
		if (ret)
			printk(KERN_WARNING "btrfs: failed to initialize"
			       " integrity check module %s\n", sb->s_id);
	}
#endif
2813 2814 2815
	ret = btrfs_read_qgroup_config(fs_info);
	if (ret)
		goto fail_trans_kthread;
2816

L
liubo 已提交
2817
	/* do not make disk changes in broken FS */
2818
	if (btrfs_super_log_root(disk_super) != 0) {
2819 2820
		u64 bytenr = btrfs_super_log_root(disk_super);

2821
		if (fs_devices->rw_devices == 0) {
C
Chris Mason 已提交
2822 2823
			printk(KERN_WARNING "Btrfs log replay required "
			       "on RO media\n");
2824
			err = -EIO;
2825
			goto fail_qgroup;
2826
		}
2827 2828 2829
		blocksize =
		     btrfs_level_size(tree_root,
				      btrfs_super_log_root_level(disk_super));
C
Chris Mason 已提交
2830

A
Al Viro 已提交
2831
		log_tree_root = btrfs_alloc_root(fs_info);
2832 2833
		if (!log_tree_root) {
			err = -ENOMEM;
2834
			goto fail_qgroup;
2835
		}
2836 2837 2838 2839 2840

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

		log_tree_root->node = read_tree_block(tree_root, bytenr,
2841 2842
						      blocksize,
						      generation + 1);
2843 2844 2845 2846 2847 2848 2849
		if (!log_tree_root->node ||
		    !extent_buffer_uptodate(log_tree_root->node)) {
			printk(KERN_ERR "btrfs: failed to read log tree\n");
			free_extent_buffer(log_tree_root->node);
			kfree(log_tree_root);
			goto fail_trans_kthread;
		}
2850
		/* returns with log_tree_root freed on success */
2851
		ret = btrfs_recover_log_trees(log_tree_root);
2852 2853 2854 2855 2856 2857 2858
		if (ret) {
			btrfs_error(tree_root->fs_info, ret,
				    "Failed to recover log tree");
			free_extent_buffer(log_tree_root->node);
			kfree(log_tree_root);
			goto fail_trans_kthread;
		}
2859 2860

		if (sb->s_flags & MS_RDONLY) {
2861 2862 2863
			ret = btrfs_commit_super(tree_root);
			if (ret)
				goto fail_trans_kthread;
2864
		}
2865
	}
Z
Zheng Yan 已提交
2866

2867
	ret = btrfs_find_orphan_roots(tree_root);
2868 2869
	if (ret)
		goto fail_trans_kthread;
2870

2871
	if (!(sb->s_flags & MS_RDONLY)) {
2872
		ret = btrfs_cleanup_fs_roots(fs_info);
2873 2874
		if (ret)
			goto fail_trans_kthread;
2875

2876
		ret = btrfs_recover_relocation(tree_root);
2877 2878 2879 2880
		if (ret < 0) {
			printk(KERN_WARNING
			       "btrfs: failed to recover relocation\n");
			err = -EINVAL;
2881
			goto fail_qgroup;
2882
		}
2883
	}
Z
Zheng Yan 已提交
2884

2885 2886
	location.objectid = BTRFS_FS_TREE_OBJECTID;
	location.type = BTRFS_ROOT_ITEM_KEY;
2887
	location.offset = 0;
2888 2889

	fs_info->fs_root = btrfs_read_fs_root_no_name(fs_info, &location);
2890 2891
	if (IS_ERR(fs_info->fs_root)) {
		err = PTR_ERR(fs_info->fs_root);
2892
		goto fail_qgroup;
2893
	}
C
Chris Mason 已提交
2894

2895 2896
	if (sb->s_flags & MS_RDONLY)
		return 0;
I
Ilya Dryomov 已提交
2897

2898 2899 2900
	down_read(&fs_info->cleanup_work_sem);
	if ((ret = btrfs_orphan_cleanup(fs_info->fs_root)) ||
	    (ret = btrfs_orphan_cleanup(fs_info->tree_root))) {
2901
		up_read(&fs_info->cleanup_work_sem);
2902 2903 2904 2905
		close_ctree(tree_root);
		return ret;
	}
	up_read(&fs_info->cleanup_work_sem);
I
Ilya Dryomov 已提交
2906

2907 2908 2909 2910 2911
	ret = btrfs_resume_balance_async(fs_info);
	if (ret) {
		printk(KERN_WARNING "btrfs: failed to resume balance\n");
		close_ctree(tree_root);
		return ret;
2912 2913
	}

2914 2915 2916 2917 2918 2919 2920
	ret = btrfs_resume_dev_replace_async(fs_info);
	if (ret) {
		pr_warn("btrfs: failed to resume dev_replace\n");
		close_ctree(tree_root);
		return ret;
	}

2921 2922
	btrfs_qgroup_rescan_resume(fs_info);

2923 2924 2925 2926 2927 2928 2929 2930 2931
	if (create_uuid_tree) {
		pr_info("btrfs: creating UUID tree\n");
		ret = btrfs_create_uuid_tree(fs_info);
		if (ret) {
			pr_warn("btrfs: failed to create the UUID tree %d\n",
				ret);
			close_ctree(tree_root);
			return ret;
		}
2932 2933
	} else if (check_uuid_tree ||
		   btrfs_test_opt(tree_root, RESCAN_UUID_TREE)) {
2934 2935 2936 2937 2938 2939 2940 2941 2942 2943
		pr_info("btrfs: checking UUID tree\n");
		ret = btrfs_check_uuid_tree(fs_info);
		if (ret) {
			pr_warn("btrfs: failed to check the UUID tree %d\n",
				ret);
			close_ctree(tree_root);
			return ret;
		}
	} else {
		fs_info->update_uuid_tree_gen = 1;
2944 2945
	}

A
Al Viro 已提交
2946
	return 0;
C
Chris Mason 已提交
2947

2948 2949
fail_qgroup:
	btrfs_free_qgroup_config(fs_info);
2950 2951
fail_trans_kthread:
	kthread_stop(fs_info->transaction_kthread);
J
Josef Bacik 已提交
2952
	btrfs_cleanup_transaction(fs_info->tree_root);
2953
	del_fs_roots(fs_info);
2954
fail_cleaner:
2955
	kthread_stop(fs_info->cleaner_kthread);
2956 2957 2958 2959 2960 2961 2962

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

2963
fail_block_groups:
J
Josef Bacik 已提交
2964
	btrfs_put_block_group_cache(fs_info);
2965
	btrfs_free_block_groups(fs_info);
C
Chris Mason 已提交
2966 2967 2968

fail_tree_roots:
	free_root_pointers(fs_info, 1);
2969
	invalidate_inode_pages2(fs_info->btree_inode->i_mapping);
C
Chris Mason 已提交
2970

C
Chris Mason 已提交
2971
fail_sb_buffer:
L
Liu Bo 已提交
2972
	btrfs_stop_all_workers(fs_info);
2973
fail_alloc:
2974
fail_iput:
2975 2976
	btrfs_mapping_tree_free(&fs_info->mapping_tree);

2977
	iput(fs_info->btree_inode);
2978 2979
fail_delalloc_bytes:
	percpu_counter_destroy(&fs_info->delalloc_bytes);
2980 2981
fail_dirty_metadata_bytes:
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
2982
fail_bdi:
2983
	bdi_destroy(&fs_info->bdi);
2984 2985
fail_srcu:
	cleanup_srcu_struct(&fs_info->subvol_srcu);
2986
fail:
D
David Woodhouse 已提交
2987
	btrfs_free_stripe_hash_table(fs_info);
2988
	btrfs_close_devices(fs_info->fs_devices);
A
Al Viro 已提交
2989
	return err;
C
Chris Mason 已提交
2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007

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;
3008 3009
}

3010 3011 3012 3013 3014
static void btrfs_end_buffer_write_sync(struct buffer_head *bh, int uptodate)
{
	if (uptodate) {
		set_buffer_uptodate(bh);
	} else {
3015 3016 3017
		struct btrfs_device *device = (struct btrfs_device *)
			bh->b_private;

3018 3019 3020
		printk_ratelimited_in_rcu(KERN_WARNING "lost page write due to "
					  "I/O error on %s\n",
					  rcu_str_deref(device->name));
3021 3022 3023
		/* note, we dont' set_buffer_write_io_error because we have
		 * our own ways of dealing with the IO errors
		 */
3024
		clear_buffer_uptodate(bh);
3025
		btrfs_dev_stat_inc_and_print(device, BTRFS_DEV_STAT_WRITE_ERRS);
3026 3027 3028 3029 3030
	}
	unlock_buffer(bh);
	put_bh(bh);
}

Y
Yan Zheng 已提交
3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046
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);
3047 3048
		if (bytenr + BTRFS_SUPER_INFO_SIZE >=
					i_size_read(bdev->bd_inode))
Y
Yan Zheng 已提交
3049
			break;
3050 3051
		bh = __bread(bdev, bytenr / 4096,
					BTRFS_SUPER_INFO_SIZE);
Y
Yan Zheng 已提交
3052 3053 3054 3055 3056
		if (!bh)
			continue;

		super = (struct btrfs_super_block *)bh->b_data;
		if (btrfs_super_bytenr(super) != bytenr ||
3057
		    btrfs_super_magic(super) != BTRFS_MAGIC) {
Y
Yan Zheng 已提交
3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072
			brelse(bh);
			continue;
		}

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

3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083
/*
 * 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
3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105
static int write_dev_supers(struct btrfs_device *device,
			    struct btrfs_super_block *sb,
			    int do_barriers, int wait, int max_mirrors)
{
	struct buffer_head *bh;
	int i;
	int ret;
	int errors = 0;
	u32 crc;
	u64 bytenr;

	if (max_mirrors == 0)
		max_mirrors = BTRFS_SUPER_MIRROR_MAX;

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

		if (wait) {
			bh = __find_get_block(device->bdev, bytenr / 4096,
					      BTRFS_SUPER_INFO_SIZE);
3106 3107 3108 3109
			if (!bh) {
				errors++;
				continue;
			}
Y
Yan Zheng 已提交
3110
			wait_on_buffer(bh);
3111 3112 3113 3114 3115 3116 3117 3118 3119
			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
3120 3121 3122 3123
		} else {
			btrfs_set_super_bytenr(sb, bytenr);

			crc = ~(u32)0;
3124
			crc = btrfs_csum_data((char *)sb +
Y
Yan Zheng 已提交
3125 3126 3127 3128 3129
					      BTRFS_CSUM_SIZE, crc,
					      BTRFS_SUPER_INFO_SIZE -
					      BTRFS_CSUM_SIZE);
			btrfs_csum_final(crc, sb->csum);

3130 3131 3132 3133
			/*
			 * one reference for us, and we leave it for the
			 * caller
			 */
Y
Yan Zheng 已提交
3134 3135
			bh = __getblk(device->bdev, bytenr / 4096,
				      BTRFS_SUPER_INFO_SIZE);
3136 3137 3138 3139 3140 3141 3142
			if (!bh) {
				printk(KERN_ERR "btrfs: couldn't get super "
				       "buffer head for bytenr %Lu\n", bytenr);
				errors++;
				continue;
			}

Y
Yan Zheng 已提交
3143 3144
			memcpy(bh->b_data, sb, BTRFS_SUPER_INFO_SIZE);

3145
			/* one reference for submit_bh */
Y
Yan Zheng 已提交
3146
			get_bh(bh);
3147 3148

			set_buffer_uptodate(bh);
Y
Yan Zheng 已提交
3149 3150
			lock_buffer(bh);
			bh->b_end_io = btrfs_end_buffer_write_sync;
3151
			bh->b_private = device;
Y
Yan Zheng 已提交
3152 3153
		}

C
Chris Mason 已提交
3154 3155 3156 3157
		/*
		 * we fua the first super.  The others we allow
		 * to go down lazy.
		 */
3158
		ret = btrfsic_submit_bh(WRITE_FUA, bh);
3159
		if (ret)
Y
Yan Zheng 已提交
3160 3161 3162 3163 3164
			errors++;
	}
	return errors < i ? 0 : -1;
}

C
Chris Mason 已提交
3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203
/*
 * 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)) {
3204 3205
			printk_in_rcu("btrfs: disabling barriers on dev %s\n",
				      rcu_str_deref(device->name));
C
Chris Mason 已提交
3206
			device->nobarriers = 1;
3207
		} else if (!bio_flagged(bio, BIO_UPTODATE)) {
C
Chris Mason 已提交
3208
			ret = -EIO;
3209 3210
			btrfs_dev_stat_inc_and_print(device,
				BTRFS_DEV_STAT_FLUSH_ERRS);
C
Chris Mason 已提交
3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223
		}

		/* 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
	 */
3224
	device->flush_bio = NULL;
3225
	bio = btrfs_io_bio_alloc(GFP_NOFS, 0);
C
Chris Mason 已提交
3226 3227 3228 3229 3230 3231 3232 3233 3234 3235
	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);
3236
	btrfsic_submit_bio(WRITE_FLUSH, bio);
C
Chris Mason 已提交
3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248

	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;
3249 3250
	int errors_send = 0;
	int errors_wait = 0;
C
Chris Mason 已提交
3251 3252 3253 3254 3255 3256
	int ret;

	/* send down all the barriers */
	head = &info->fs_devices->devices;
	list_for_each_entry_rcu(dev, head, dev_list) {
		if (!dev->bdev) {
3257
			errors_send++;
C
Chris Mason 已提交
3258 3259 3260 3261 3262 3263 3264
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

		ret = write_dev_flush(dev, 0);
		if (ret)
3265
			errors_send++;
C
Chris Mason 已提交
3266 3267 3268 3269 3270
	}

	/* wait for all the barriers */
	list_for_each_entry_rcu(dev, head, dev_list) {
		if (!dev->bdev) {
3271
			errors_wait++;
C
Chris Mason 已提交
3272 3273 3274 3275 3276 3277 3278
			continue;
		}
		if (!dev->in_fs_metadata || !dev->writeable)
			continue;

		ret = write_dev_flush(dev, 1);
		if (ret)
3279
			errors_wait++;
C
Chris Mason 已提交
3280
	}
3281 3282
	if (errors_send > info->num_tolerated_disk_barrier_failures ||
	    errors_wait > info->num_tolerated_disk_barrier_failures)
C
Chris Mason 已提交
3283 3284 3285 3286
		return -EIO;
	return 0;
}

3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346
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
3347 3348 3349 3350 3351 3352
				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 &
3353
						   BTRFS_BLOCK_GROUP_RAID6) {
D
David Woodhouse 已提交
3354 3355 3356
						num_tolerated_disk_barrier_failures = 2;
					}
				}
3357 3358 3359 3360 3361 3362 3363 3364
			}
		}
		up_read(&sinfo->groups_sem);
	}

	return num_tolerated_disk_barrier_failures;
}

3365
static int write_all_supers(struct btrfs_root *root, int max_mirrors)
3366
{
3367
	struct list_head *head;
3368
	struct btrfs_device *dev;
3369
	struct btrfs_super_block *sb;
3370 3371 3372
	struct btrfs_dev_item *dev_item;
	int ret;
	int do_barriers;
3373 3374
	int max_errors;
	int total_errors = 0;
3375
	u64 flags;
3376

3377
	max_errors = btrfs_super_num_devices(root->fs_info->super_copy) - 1;
3378
	do_barriers = !btrfs_test_opt(root, NOBARRIER);
C
Chris Mason 已提交
3379
	backup_super_roots(root->fs_info);
3380

3381
	sb = root->fs_info->super_for_commit;
3382
	dev_item = &sb->dev_item;
3383

3384
	mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
3385
	head = &root->fs_info->fs_devices->devices;
C
Chris Mason 已提交
3386

3387 3388 3389 3390 3391 3392 3393 3394 3395 3396
	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
3397

3398
	list_for_each_entry_rcu(dev, head, dev_list) {
3399 3400 3401 3402
		if (!dev->bdev) {
			total_errors++;
			continue;
		}
Y
Yan Zheng 已提交
3403
		if (!dev->in_fs_metadata || !dev->writeable)
3404 3405
			continue;

Y
Yan Zheng 已提交
3406
		btrfs_set_stack_device_generation(dev_item, 0);
3407 3408 3409 3410 3411 3412 3413 3414
		btrfs_set_stack_device_type(dev_item, dev->type);
		btrfs_set_stack_device_id(dev_item, dev->devid);
		btrfs_set_stack_device_total_bytes(dev_item, dev->total_bytes);
		btrfs_set_stack_device_bytes_used(dev_item, dev->bytes_used);
		btrfs_set_stack_device_io_align(dev_item, dev->io_align);
		btrfs_set_stack_device_io_width(dev_item, dev->io_width);
		btrfs_set_stack_device_sector_size(dev_item, dev->sector_size);
		memcpy(dev_item->uuid, dev->uuid, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3415
		memcpy(dev_item->fsid, dev->fs_devices->fsid, BTRFS_UUID_SIZE);
Y
Yan Zheng 已提交
3416

3417 3418 3419
		flags = btrfs_super_flags(sb);
		btrfs_set_super_flags(sb, flags | BTRFS_HEADER_FLAG_WRITTEN);

Y
Yan Zheng 已提交
3420
		ret = write_dev_supers(dev, sb, do_barriers, 0, max_mirrors);
3421 3422
		if (ret)
			total_errors++;
3423
	}
3424
	if (total_errors > max_errors) {
C
Chris Mason 已提交
3425 3426
		printk(KERN_ERR "btrfs: %d errors while writing supers\n",
		       total_errors);
3427 3428

		/* This shouldn't happen. FUA is masked off if unsupported */
3429 3430
		BUG();
	}
3431

Y
Yan Zheng 已提交
3432
	total_errors = 0;
3433
	list_for_each_entry_rcu(dev, head, dev_list) {
3434 3435
		if (!dev->bdev)
			continue;
Y
Yan Zheng 已提交
3436
		if (!dev->in_fs_metadata || !dev->writeable)
3437 3438
			continue;

Y
Yan Zheng 已提交
3439 3440 3441
		ret = write_dev_supers(dev, sb, do_barriers, 1, max_mirrors);
		if (ret)
			total_errors++;
3442
	}
3443
	mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
3444
	if (total_errors > max_errors) {
3445 3446 3447
		btrfs_error(root->fs_info, -EIO,
			    "%d errors while writing supers", total_errors);
		return -EIO;
3448
	}
3449 3450 3451
	return 0;
}

Y
Yan Zheng 已提交
3452 3453
int write_ctree_super(struct btrfs_trans_handle *trans,
		      struct btrfs_root *root, int max_mirrors)
3454
{
3455
	int ret;
3456

Y
Yan Zheng 已提交
3457
	ret = write_all_supers(root, max_mirrors);
3458
	return ret;
C
Chris Mason 已提交
3459 3460
}

3461 3462 3463
/* 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
3464
{
3465
	spin_lock(&fs_info->fs_roots_radix_lock);
C
Chris Mason 已提交
3466 3467
	radix_tree_delete(&fs_info->fs_roots_radix,
			  (unsigned long)root->root_key.objectid);
3468
	spin_unlock(&fs_info->fs_roots_radix_lock);
3469 3470 3471 3472

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

L
Liu Bo 已提交
3473
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state)) {
L
Liu Bo 已提交
3474 3475 3476 3477
		btrfs_free_log(NULL, root);
		btrfs_free_log_root_tree(NULL, fs_info);
	}

3478 3479
	__btrfs_remove_free_space_cache(root->free_ino_pinned);
	__btrfs_remove_free_space_cache(root->free_ino_ctl);
3480 3481 3482 3483 3484
	free_fs_root(root);
}

static void free_fs_root(struct btrfs_root *root)
{
3485
	iput(root->cache_inode);
3486
	WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree));
3487 3488
	if (root->anon_dev)
		free_anon_bdev(root->anon_dev);
3489 3490
	free_extent_buffer(root->node);
	free_extent_buffer(root->commit_root);
3491 3492
	kfree(root->free_ino_ctl);
	kfree(root->free_ino_pinned);
C
Chris Mason 已提交
3493
	kfree(root->name);
3494
	btrfs_put_fs_root(root);
C
Chris Mason 已提交
3495 3496
}

3497 3498 3499
void btrfs_free_fs_root(struct btrfs_root *root)
{
	free_fs_root(root);
C
Chris Mason 已提交
3500 3501
}

Y
Yan Zheng 已提交
3502
int btrfs_cleanup_fs_roots(struct btrfs_fs_info *fs_info)
C
Chris Mason 已提交
3503
{
Y
Yan Zheng 已提交
3504 3505 3506
	u64 root_objectid = 0;
	struct btrfs_root *gang[8];
	int i;
3507
	int ret;
3508

Y
Yan Zheng 已提交
3509 3510 3511 3512 3513 3514
	while (1) {
		ret = radix_tree_gang_lookup(&fs_info->fs_roots_radix,
					     (void **)gang, root_objectid,
					     ARRAY_SIZE(gang));
		if (!ret)
			break;
3515 3516

		root_objectid = gang[ret - 1]->root_key.objectid + 1;
Y
Yan Zheng 已提交
3517
		for (i = 0; i < ret; i++) {
3518 3519
			int err;

Y
Yan Zheng 已提交
3520
			root_objectid = gang[i]->root_key.objectid;
3521 3522 3523
			err = btrfs_orphan_cleanup(gang[i]);
			if (err)
				return err;
Y
Yan Zheng 已提交
3524 3525 3526 3527 3528
		}
		root_objectid++;
	}
	return 0;
}
3529

Y
Yan Zheng 已提交
3530 3531 3532 3533
int btrfs_commit_super(struct btrfs_root *root)
{
	struct btrfs_trans_handle *trans;
	int ret;
3534

Y
Yan Zheng 已提交
3535
	mutex_lock(&root->fs_info->cleaner_mutex);
Y
Yan, Zheng 已提交
3536
	btrfs_run_delayed_iputs(root);
Y
Yan Zheng 已提交
3537
	mutex_unlock(&root->fs_info->cleaner_mutex);
D
David Sterba 已提交
3538
	wake_up_process(root->fs_info->cleaner_kthread);
3539 3540 3541 3542 3543

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

3544
	trans = btrfs_join_transaction(root);
3545 3546
	if (IS_ERR(trans))
		return PTR_ERR(trans);
3547
	ret = btrfs_commit_transaction(trans, root);
3548 3549
	if (ret)
		return ret;
Y
Yan Zheng 已提交
3550
	/* run commit again to drop the original snapshot */
3551
	trans = btrfs_join_transaction(root);
3552 3553
	if (IS_ERR(trans))
		return PTR_ERR(trans);
3554 3555 3556
	ret = btrfs_commit_transaction(trans, root);
	if (ret)
		return ret;
C
Chris Mason 已提交
3557
	ret = btrfs_write_and_wait_transaction(NULL, root);
3558 3559 3560 3561 3562
	if (ret) {
		btrfs_error(root->fs_info, ret,
			    "Failed to sync btree inode to disk.");
		return ret;
	}
3563

Y
Yan Zheng 已提交
3564
	ret = write_ctree_super(NULL, root, 0);
Y
Yan Zheng 已提交
3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575
	return ret;
}

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

	fs_info->closing = 1;
	smp_mb();

S
Stefan Behrens 已提交
3576 3577 3578 3579 3580
	/* 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);

3581
	/* pause restriper - we want to resume on mount */
3582
	btrfs_pause_balance(fs_info);
3583

3584 3585
	btrfs_dev_replace_suspend_for_unmount(fs_info);

3586
	btrfs_scrub_cancel(fs_info);
C
Chris Mason 已提交
3587 3588 3589 3590 3591 3592

	/* 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 */
3593
	btrfs_cleanup_defrag_inodes(fs_info);
C
Chris Mason 已提交
3594

Y
Yan Zheng 已提交
3595
	if (!(fs_info->sb->s_flags & MS_RDONLY)) {
L
liubo 已提交
3596 3597 3598 3599 3600
		ret = btrfs_commit_super(root);
		if (ret)
			printk(KERN_ERR "btrfs: commit super ret %d\n", ret);
	}

3601
	if (test_bit(BTRFS_FS_STATE_ERROR, &fs_info->fs_state))
3602
		btrfs_error_commit_super(root);
3603

3604 3605
	btrfs_put_block_group_cache(fs_info);

A
Al Viro 已提交
3606 3607
	kthread_stop(fs_info->transaction_kthread);
	kthread_stop(fs_info->cleaner_kthread);
3608

3609 3610 3611
	fs_info->closing = 2;
	smp_mb();

3612 3613
	btrfs_free_qgroup_config(root->fs_info);

3614 3615 3616
	if (percpu_counter_sum(&fs_info->delalloc_bytes)) {
		printk(KERN_INFO "btrfs: at unmount delalloc count %lld\n",
		       percpu_counter_sum(&fs_info->delalloc_bytes));
C
Chris Mason 已提交
3617
	}
3618

A
Al Viro 已提交
3619
	btrfs_free_block_groups(fs_info);
3620

3621
	btrfs_stop_all_workers(fs_info);
3622

Y
Yan Zheng 已提交
3623
	del_fs_roots(fs_info);
3624

3625
	free_root_pointers(fs_info, 1);
3626

3627
	iput(fs_info->btree_inode);
3628

3629 3630 3631 3632 3633
#ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
	if (btrfs_test_opt(root, CHECK_INTEGRITY))
		btrfsic_unmount(root, fs_info->fs_devices);
#endif

3634
	btrfs_close_devices(fs_info->fs_devices);
3635
	btrfs_mapping_tree_free(&fs_info->mapping_tree);
3636

3637
	percpu_counter_destroy(&fs_info->dirty_metadata_bytes);
3638
	percpu_counter_destroy(&fs_info->delalloc_bytes);
C
Chris Mason 已提交
3639
	bdi_destroy(&fs_info->bdi);
3640
	cleanup_srcu_struct(&fs_info->subvol_srcu);
3641

D
David Woodhouse 已提交
3642 3643
	btrfs_free_stripe_hash_table(fs_info);

3644 3645 3646
	return 0;
}

3647 3648
int btrfs_buffer_uptodate(struct extent_buffer *buf, u64 parent_transid,
			  int atomic)
3649
{
3650
	int ret;
3651
	struct inode *btree_inode = buf->pages[0]->mapping->host;
3652

3653
	ret = extent_buffer_uptodate(buf);
3654 3655 3656 3657
	if (!ret)
		return ret;

	ret = verify_parent_transid(&BTRFS_I(btree_inode)->io_tree, buf,
3658 3659 3660
				    parent_transid, atomic);
	if (ret == -EAGAIN)
		return ret;
3661
	return !ret;
3662 3663 3664
}

int btrfs_set_buffer_uptodate(struct extent_buffer *buf)
C
Chris Mason 已提交
3665
{
3666
	return set_extent_buffer_uptodate(buf);
3667
}
3668

3669 3670
void btrfs_mark_buffer_dirty(struct extent_buffer *buf)
{
3671
	struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root;
3672
	u64 transid = btrfs_header_generation(buf);
3673
	int was_dirty;
3674

3675
	btrfs_assert_tree_locked(buf);
J
Julia Lawall 已提交
3676 3677
	if (transid != root->fs_info->generation)
		WARN(1, KERN_CRIT "btrfs transid mismatch buffer %llu, "
C
Chris Mason 已提交
3678
		       "found %llu running %llu\n",
3679
			(unsigned long long)buf->start,
C
Chris Mason 已提交
3680 3681
			(unsigned long long)transid,
			(unsigned long long)root->fs_info->generation);
3682
	was_dirty = set_extent_buffer_dirty(buf);
3683 3684 3685 3686
	if (!was_dirty)
		__percpu_counter_add(&root->fs_info->dirty_metadata_bytes,
				     buf->len,
				     root->fs_info->dirty_metadata_batch);
3687 3688
}

3689 3690
static void __btrfs_btree_balance_dirty(struct btrfs_root *root,
					int flush_delayed)
3691 3692 3693 3694 3695
{
	/*
	 * looks as though older kernels can get into trouble with
	 * this code, they end up stuck in balance_dirty_pages forever
	 */
3696
	int ret;
3697 3698 3699 3700

	if (current->flags & PF_MEMALLOC)
		return;

3701 3702
	if (flush_delayed)
		btrfs_balance_delayed_items(root);
3703

3704 3705 3706
	ret = percpu_counter_compare(&root->fs_info->dirty_metadata_bytes,
				     BTRFS_DIRTY_METADATA_THRESH);
	if (ret > 0) {
3707 3708
		balance_dirty_pages_ratelimited(
				   root->fs_info->btree_inode->i_mapping);
3709 3710 3711 3712
	}
	return;
}

3713
void btrfs_btree_balance_dirty(struct btrfs_root *root)
C
Chris Mason 已提交
3714
{
3715 3716
	__btrfs_btree_balance_dirty(root, 1);
}
3717

3718 3719 3720
void btrfs_btree_balance_dirty_nodelay(struct btrfs_root *root)
{
	__btrfs_btree_balance_dirty(root, 0);
C
Chris Mason 已提交
3721
}
3722

3723
int btrfs_read_buffer(struct extent_buffer *buf, u64 parent_transid)
3724
{
3725
	struct btrfs_root *root = BTRFS_I(buf->pages[0]->mapping->host)->root;
3726
	return btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
3727
}
3728

3729
static int btrfs_check_super_valid(struct btrfs_fs_info *fs_info,
L
liubo 已提交
3730 3731
			      int read_only)
{
D
David Sterba 已提交
3732 3733 3734
	/*
	 * Placeholder for checks
	 */
3735
	return 0;
L
liubo 已提交
3736 3737
}

3738
static void btrfs_error_commit_super(struct btrfs_root *root)
L
liubo 已提交
3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750
{
	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);
}

3751 3752
static void btrfs_destroy_ordered_operations(struct btrfs_transaction *t,
					     struct btrfs_root *root)
L
liubo 已提交
3753 3754 3755 3756 3757 3758 3759
{
	struct btrfs_inode *btrfs_inode;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

	mutex_lock(&root->fs_info->ordered_operations_mutex);
3760
	spin_lock(&root->fs_info->ordered_root_lock);
L
liubo 已提交
3761

3762
	list_splice_init(&t->ordered_operations, &splice);
L
liubo 已提交
3763 3764 3765 3766 3767
	while (!list_empty(&splice)) {
		btrfs_inode = list_entry(splice.next, struct btrfs_inode,
					 ordered_operations);

		list_del_init(&btrfs_inode->ordered_operations);
3768
		spin_unlock(&root->fs_info->ordered_root_lock);
L
liubo 已提交
3769 3770

		btrfs_invalidate_inodes(btrfs_inode->root);
3771

3772
		spin_lock(&root->fs_info->ordered_root_lock);
L
liubo 已提交
3773 3774
	}

3775
	spin_unlock(&root->fs_info->ordered_root_lock);
L
liubo 已提交
3776 3777 3778
	mutex_unlock(&root->fs_info->ordered_operations_mutex);
}

3779
static void btrfs_destroy_ordered_extents(struct btrfs_root *root)
L
liubo 已提交
3780 3781 3782
{
	struct btrfs_ordered_extent *ordered;

3783
	spin_lock(&root->ordered_extent_lock);
3784 3785 3786 3787
	/*
	 * This will just short circuit the ordered completion stuff which will
	 * make sure the ordered extent gets properly cleaned up.
	 */
3788
	list_for_each_entry(ordered, &root->ordered_extents,
3789 3790
			    root_extent_list)
		set_bit(BTRFS_ORDERED_IOERR, &ordered->flags);
3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812
	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);
		list_del_init(&root->ordered_root);

		btrfs_destroy_ordered_extents(root);

		cond_resched_lock(&fs_info->ordered_root_lock);
	}
	spin_unlock(&fs_info->ordered_root_lock);
L
liubo 已提交
3813 3814
}

3815 3816
static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
				      struct btrfs_root *root)
L
liubo 已提交
3817 3818 3819 3820 3821 3822 3823 3824 3825 3826
{
	struct rb_node *node;
	struct btrfs_delayed_ref_root *delayed_refs;
	struct btrfs_delayed_ref_node *ref;
	int ret = 0;

	delayed_refs = &trans->delayed_refs;

	spin_lock(&delayed_refs->lock);
	if (delayed_refs->num_entries == 0) {
3827
		spin_unlock(&delayed_refs->lock);
L
liubo 已提交
3828 3829 3830 3831
		printk(KERN_INFO "delayed_refs has NO entry\n");
		return ret;
	}

3832
	while ((node = rb_first(&delayed_refs->root)) != NULL) {
3833
		struct btrfs_delayed_ref_head *head = NULL;
3834
		bool pin_bytes = false;
L
liubo 已提交
3835

3836
		ref = rb_entry(node, struct btrfs_delayed_ref_node, rb_node);
L
liubo 已提交
3837 3838 3839 3840
		atomic_set(&ref->refs, 1);
		if (btrfs_delayed_ref_is_head(ref)) {

			head = btrfs_delayed_node_to_head(ref);
3841 3842 3843 3844 3845 3846 3847 3848 3849
			if (!mutex_trylock(&head->mutex)) {
				atomic_inc(&ref->refs);
				spin_unlock(&delayed_refs->lock);

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

J
Josef Bacik 已提交
3850
				spin_lock(&delayed_refs->lock);
3851 3852 3853
				continue;
			}

J
Josef Bacik 已提交
3854
			if (head->must_insert_reserved)
3855
				pin_bytes = true;
3856
			btrfs_free_delayed_extent_op(head->extent_op);
L
liubo 已提交
3857 3858 3859 3860 3861
			delayed_refs->num_heads--;
			if (list_empty(&head->cluster))
				delayed_refs->num_heads_ready--;
			list_del_init(&head->cluster);
		}
3862

3863 3864 3865
		ref->in_tree = 0;
		rb_erase(&ref->rb_node, &delayed_refs->root);
		delayed_refs->num_entries--;
L
liubo 已提交
3866
		spin_unlock(&delayed_refs->lock);
3867 3868 3869 3870 3871 3872
		if (head) {
			if (pin_bytes)
				btrfs_pin_extent(root, ref->bytenr,
						 ref->num_bytes, 1);
			mutex_unlock(&head->mutex);
		}
L
liubo 已提交
3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883
		btrfs_put_delayed_ref(ref);

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

	spin_unlock(&delayed_refs->lock);

	return ret;
}

3884
static void btrfs_evict_pending_snapshots(struct btrfs_transaction *t)
L
liubo 已提交
3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896
{
	struct btrfs_pending_snapshot *snapshot;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

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

	while (!list_empty(&splice)) {
		snapshot = list_entry(splice.next,
				      struct btrfs_pending_snapshot,
				      list);
3897
		snapshot->error = -ECANCELED;
L
liubo 已提交
3898 3899 3900 3901
		list_del_init(&snapshot->list);
	}
}

3902
static void btrfs_destroy_delalloc_inodes(struct btrfs_root *root)
L
liubo 已提交
3903 3904 3905 3906 3907 3908
{
	struct btrfs_inode *btrfs_inode;
	struct list_head splice;

	INIT_LIST_HEAD(&splice);

3909 3910
	spin_lock(&root->delalloc_lock);
	list_splice_init(&root->delalloc_inodes, &splice);
L
liubo 已提交
3911 3912

	while (!list_empty(&splice)) {
3913 3914
		btrfs_inode = list_first_entry(&splice, struct btrfs_inode,
					       delalloc_inodes);
L
liubo 已提交
3915 3916

		list_del_init(&btrfs_inode->delalloc_inodes);
3917 3918
		clear_bit(BTRFS_INODE_IN_DELALLOC_LIST,
			  &btrfs_inode->runtime_flags);
3919
		spin_unlock(&root->delalloc_lock);
L
liubo 已提交
3920 3921

		btrfs_invalidate_inodes(btrfs_inode->root);
3922

3923
		spin_lock(&root->delalloc_lock);
L
liubo 已提交
3924 3925
	}

3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951
	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
3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964
}

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,
3965
					    mark, NULL);
L
liubo 已提交
3966 3967 3968 3969 3970
		if (ret)
			break;

		clear_extent_bits(dirty_pages, start, end, mark, GFP_NOFS);
		while (start <= end) {
3971 3972
			eb = btrfs_find_tree_block(root, start,
						   root->leafsize);
3973
			start += root->leafsize;
3974
			if (!eb)
L
liubo 已提交
3975
				continue;
3976
			wait_on_extent_buffer_writeback(eb);
L
liubo 已提交
3977

3978 3979 3980 3981
			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
3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994
		}
	}

	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;
3995
	bool loop = true;
L
liubo 已提交
3996 3997

	unpin = pinned_extents;
3998
again:
L
liubo 已提交
3999 4000
	while (1) {
		ret = find_first_extent_bit(unpin, 0, &start, &end,
4001
					    EXTENT_DIRTY, NULL);
L
liubo 已提交
4002 4003 4004 4005
		if (ret)
			break;

		/* opt_discard */
4006 4007 4008 4009
		if (btrfs_test_opt(root, DISCARD))
			ret = btrfs_error_discard_extent(root, start,
							 end + 1 - start,
							 NULL);
L
liubo 已提交
4010 4011 4012 4013 4014 4015

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

4016 4017 4018 4019 4020 4021 4022 4023 4024
	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
4025 4026 4027
	return 0;
}

4028 4029 4030 4031 4032 4033 4034
void btrfs_cleanup_one_transaction(struct btrfs_transaction *cur_trans,
				   struct btrfs_root *root)
{
	btrfs_destroy_delayed_refs(cur_trans, root);
	btrfs_block_rsv_release(root, &root->fs_info->trans_block_rsv,
				cur_trans->dirty_pages.dirty_bytes);

4035
	cur_trans->state = TRANS_STATE_COMMIT_START;
4036
	wake_up(&root->fs_info->transaction_blocked_wait);
4037

4038 4039
	btrfs_evict_pending_snapshots(cur_trans);

4040
	cur_trans->state = TRANS_STATE_UNBLOCKED;
4041
	wake_up(&root->fs_info->transaction_wait);
4042

4043 4044
	btrfs_destroy_delayed_inodes(root);
	btrfs_assert_delayed_root_empty(root);
4045 4046 4047

	btrfs_destroy_marked_extents(root, &cur_trans->dirty_pages,
				     EXTENT_DIRTY);
4048 4049
	btrfs_destroy_pinned_extent(root,
				    root->fs_info->pinned_extents);
4050

4051 4052 4053
	cur_trans->state =TRANS_STATE_COMPLETED;
	wake_up(&cur_trans->commit_wait);

4054 4055 4056 4057 4058 4059
	/*
	memset(cur_trans, 0, sizeof(*cur_trans));
	kmem_cache_free(btrfs_transaction_cachep, cur_trans);
	*/
}

4060
static int btrfs_cleanup_transaction(struct btrfs_root *root)
L
liubo 已提交
4061 4062 4063 4064 4065 4066
{
	struct btrfs_transaction *t;
	LIST_HEAD(list);

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

J
Josef Bacik 已提交
4067
	spin_lock(&root->fs_info->trans_lock);
L
liubo 已提交
4068
	list_splice_init(&root->fs_info->trans_list, &list);
4069
	root->fs_info->running_transaction = NULL;
J
Josef Bacik 已提交
4070 4071
	spin_unlock(&root->fs_info->trans_lock);

L
liubo 已提交
4072 4073 4074
	while (!list_empty(&list)) {
		t = list_entry(list.next, struct btrfs_transaction, list);

4075
		btrfs_destroy_ordered_operations(t, root);
L
liubo 已提交
4076

4077
		btrfs_destroy_all_ordered_extents(root->fs_info);
L
liubo 已提交
4078 4079 4080

		btrfs_destroy_delayed_refs(t, root);

4081 4082 4083 4084 4085 4086
		/*
		 *  FIXME: cleanup wait for commit
		 *  We needn't acquire the lock here, because we are during
		 *  the umount, there is no other task which will change it.
		 */
		t->state = TRANS_STATE_COMMIT_START;
4087
		smp_mb();
L
liubo 已提交
4088 4089 4090
		if (waitqueue_active(&root->fs_info->transaction_blocked_wait))
			wake_up(&root->fs_info->transaction_blocked_wait);

4091 4092
		btrfs_evict_pending_snapshots(t);

4093
		t->state = TRANS_STATE_UNBLOCKED;
4094
		smp_mb();
L
liubo 已提交
4095 4096 4097
		if (waitqueue_active(&root->fs_info->transaction_wait))
			wake_up(&root->fs_info->transaction_wait);

4098 4099 4100
		btrfs_destroy_delayed_inodes(root);
		btrfs_assert_delayed_root_empty(root);

4101
		btrfs_destroy_all_delalloc_inodes(root->fs_info);
L
liubo 已提交
4102 4103 4104 4105 4106 4107 4108

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

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

4109 4110 4111 4112 4113
		t->state = TRANS_STATE_COMPLETED;
		smp_mb();
		if (waitqueue_active(&t->commit_wait))
			wake_up(&t->commit_wait);

4114
		atomic_set(&t->use_count, 0);
L
liubo 已提交
4115 4116 4117 4118 4119 4120 4121 4122 4123 4124
		list_del_init(&t->list);
		memset(t, 0, sizeof(*t));
		kmem_cache_free(btrfs_transaction_cachep, t);
	}

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

	return 0;
}

4125
static struct extent_io_ops btree_extent_io_ops = {
4126
	.readpage_end_io_hook = btree_readpage_end_io_hook,
A
Arne Jansen 已提交
4127
	.readpage_io_failed_hook = btree_io_failed_hook,
4128
	.submit_bio_hook = btree_submit_bio_hook,
4129 4130
	/* note we're sharing with inode.c for the merge bio hook */
	.merge_bio_hook = btrfs_merge_bio_hook,
4131
};